From e0a330756368f66a8b7986f6d4ea7fcbbf76b145 Mon Sep 17 00:00:00 2001
From: takatost <takatost@users.noreply.github.com>
Date: Fri, 20 Sep 2024 19:47:25 +0800
Subject: [PATCH 001/235] fix(workflow): "Max submit count reached" error
 occurred when executing workflow as tool in iteration (#8595)

---
 api/core/workflow/graph_engine/graph_engine.py      | 10 +++++++---
 api/core/workflow/nodes/iteration/iteration_node.py |  1 +
 2 files changed, 8 insertions(+), 3 deletions(-)

diff --git a/api/core/workflow/graph_engine/graph_engine.py b/api/core/workflow/graph_engine/graph_engine.py
index 57e4f716fdda4..8342dbd13d2a5 100644
--- a/api/core/workflow/graph_engine/graph_engine.py
+++ b/api/core/workflow/graph_engine/graph_engine.py
@@ -180,16 +180,20 @@ def run(self) -> Generator[GraphEngineEvent, None, None]:
 
             # trigger graph run success event
             yield GraphRunSucceededEvent(outputs=self.graph_runtime_state.outputs)
+            self._release_thread()
         except GraphRunFailedError as e:
             yield GraphRunFailedEvent(error=e.error)
+            self._release_thread()
             return
         except Exception as e:
             logger.exception("Unknown Error when graph running")
             yield GraphRunFailedEvent(error=str(e))
+            self._release_thread()
             raise e
-        finally:
-            if self.is_main_thread_pool and self.thread_pool_id in GraphEngine.workflow_thread_pool_mapping:
-                del GraphEngine.workflow_thread_pool_mapping[self.thread_pool_id]
+
+    def _release_thread(self):
+        if self.is_main_thread_pool and self.thread_pool_id in GraphEngine.workflow_thread_pool_mapping:
+            del GraphEngine.workflow_thread_pool_mapping[self.thread_pool_id]
 
     def _run(
         self,
diff --git a/api/core/workflow/nodes/iteration/iteration_node.py b/api/core/workflow/nodes/iteration/iteration_node.py
index 6f20745dafc8d..01bb4e9076e5a 100644
--- a/api/core/workflow/nodes/iteration/iteration_node.py
+++ b/api/core/workflow/nodes/iteration/iteration_node.py
@@ -89,6 +89,7 @@ def _run(self) -> Generator[RunEvent | InNodeEvent, None, None]:
             variable_pool=variable_pool,
             max_execution_steps=dify_config.WORKFLOW_MAX_EXECUTION_STEPS,
             max_execution_time=dify_config.WORKFLOW_MAX_EXECUTION_TIME,
+            thread_pool_id=self.thread_pool_id,
         )
 
         start_at = datetime.now(timezone.utc).replace(tzinfo=None)

From d63a5a1c3ce526df43984899b6c2234a4d2aaf3e Mon Sep 17 00:00:00 2001
From: =?UTF-8?q?=E9=9D=9E=E6=B3=95=E6=93=8D=E4=BD=9C?= <hjlarry@163.com>
Date: Sat, 21 Sep 2024 17:30:30 +0800
Subject: [PATCH 002/235] fix: a helper link error (#8508)

---
 .../app/configuration/dataset-config/settings-modal/index.tsx   | 2 +-
 web/app/components/datasets/create/step-two/index.tsx           | 2 +-
 .../components/datasets/hit-testing/modify-retrieval-modal.tsx  | 2 +-
 web/app/components/datasets/settings/form/index.tsx             | 2 +-
 4 files changed, 4 insertions(+), 4 deletions(-)

diff --git a/web/app/components/app/configuration/dataset-config/settings-modal/index.tsx b/web/app/components/app/configuration/dataset-config/settings-modal/index.tsx
index 65858ce8cfd36..e538c347d95c9 100644
--- a/web/app/components/app/configuration/dataset-config/settings-modal/index.tsx
+++ b/web/app/components/app/configuration/dataset-config/settings-modal/index.tsx
@@ -263,7 +263,7 @@ const SettingsModal: FC<SettingsModalProps> = ({
             <div>
               <div>{t('datasetSettings.form.retrievalSetting.title')}</div>
               <div className='leading-[18px] text-xs font-normal text-gray-500'>
-                <a target='_blank' rel='noopener noreferrer' href='https://docs.dify.ai/guides/knowledge-base/create-knowledge-and-upload-documents#id-6-retrieval-settings' className='text-[#155eef]'>{t('datasetSettings.form.retrievalSetting.learnMore')}</a>
+                <a target='_blank' rel='noopener noreferrer' href='https://docs.dify.ai/guides/knowledge-base/create-knowledge-and-upload-documents#id-4-retrieval-settings' className='text-[#155eef]'>{t('datasetSettings.form.retrievalSetting.learnMore')}</a>
                 {t('datasetSettings.form.retrievalSetting.description')}
               </div>
             </div>
diff --git a/web/app/components/datasets/create/step-two/index.tsx b/web/app/components/datasets/create/step-two/index.tsx
index 94614918dbe40..f4fc58ee2a2d8 100644
--- a/web/app/components/datasets/create/step-two/index.tsx
+++ b/web/app/components/datasets/create/step-two/index.tsx
@@ -820,7 +820,7 @@ const StepTwo = ({
                   <div className={s.label}>
                     <div className='shrink-0 mr-4'>{t('datasetSettings.form.retrievalSetting.title')}</div>
                     <div className='leading-[18px] text-xs font-normal text-gray-500'>
-                      <a target='_blank' rel='noopener noreferrer' href='https://docs.dify.ai/guides/knowledge-base/create-knowledge-and-upload-documents#id-6-retrieval-settings' className='text-[#155eef]'>{t('datasetSettings.form.retrievalSetting.learnMore')}</a>
+                      <a target='_blank' rel='noopener noreferrer' href='https://docs.dify.ai/guides/knowledge-base/create-knowledge-and-upload-documents#id-4-retrieval-settings' className='text-[#155eef]'>{t('datasetSettings.form.retrievalSetting.learnMore')}</a>
                       {t('datasetSettings.form.retrievalSetting.longDescription')}
                     </div>
                   </div>
diff --git a/web/app/components/datasets/hit-testing/modify-retrieval-modal.tsx b/web/app/components/datasets/hit-testing/modify-retrieval-modal.tsx
index 999f1cdf0d2b0..1fc5b68d671eb 100644
--- a/web/app/components/datasets/hit-testing/modify-retrieval-modal.tsx
+++ b/web/app/components/datasets/hit-testing/modify-retrieval-modal.tsx
@@ -77,7 +77,7 @@ const ModifyRetrievalModal: FC<Props> = ({
         <div className='text-base font-semibold text-gray-900'>
           <div>{t('datasetSettings.form.retrievalSetting.title')}</div>
           <div className='leading-[18px] text-xs font-normal text-gray-500'>
-            <a target='_blank' rel='noopener noreferrer' href='https://docs.dify.ai/guides/knowledge-base/create-knowledge-and-upload-documents#id-6-retrieval-settings' className='text-[#155eef]'>{t('datasetSettings.form.retrievalSetting.learnMore')}</a>
+            <a target='_blank' rel='noopener noreferrer' href='https://docs.dify.ai/guides/knowledge-base/create-knowledge-and-upload-documents#id-4-retrieval-settings' className='text-[#155eef]'>{t('datasetSettings.form.retrievalSetting.learnMore')}</a>
             {t('datasetSettings.form.retrievalSetting.description')}
           </div>
         </div>
diff --git a/web/app/components/datasets/settings/form/index.tsx b/web/app/components/datasets/settings/form/index.tsx
index 0f6bdd0a59e0d..15b8abc242e1a 100644
--- a/web/app/components/datasets/settings/form/index.tsx
+++ b/web/app/components/datasets/settings/form/index.tsx
@@ -245,7 +245,7 @@ const Form = () => {
           <div>
             <div>{t('datasetSettings.form.retrievalSetting.title')}</div>
             <div className='leading-[18px] text-xs font-normal text-gray-500'>
-              <a target='_blank' rel='noopener noreferrer' href='https://docs.dify.ai/guides/knowledge-base/create-knowledge-and-upload-documents#id-6-retrieval-settings' className='text-[#155eef]'>{t('datasetSettings.form.retrievalSetting.learnMore')}</a>
+              <a target='_blank' rel='noopener noreferrer' href='https://docs.dify.ai/guides/knowledge-base/create-knowledge-and-upload-documents#id-4-retrieval-settings' className='text-[#155eef]'>{t('datasetSettings.form.retrievalSetting.learnMore')}</a>
               {t('datasetSettings.form.retrievalSetting.description')}
             </div>
           </div>

From 483ead55d5d0e337f2b82a8dab2cdf501f390caa Mon Sep 17 00:00:00 2001
From: "github-actions[bot]"
 <41898282+github-actions[bot]@users.noreply.github.com>
Date: Sat, 21 Sep 2024 17:30:43 +0800
Subject: [PATCH 003/235] chore: translate i18n files (#8557)

Co-authored-by: iamjoel <2120155+iamjoel@users.noreply.github.com>
Co-authored-by: crazywoola <100913391+crazywoola@users.noreply.github.com>
---
 web/i18n/de-DE/dataset-creation.ts   | 1 +
 web/i18n/es-ES/dataset-creation.ts   | 1 +
 web/i18n/fa-IR/dataset-creation.ts   | 1 +
 web/i18n/fr-FR/dataset-creation.ts   | 1 +
 web/i18n/hi-IN/dataset-creation.ts   | 1 +
 web/i18n/it-IT/dataset-creation.ts   | 1 +
 web/i18n/ja-JP/dataset-creation.ts   | 1 +
 web/i18n/ko-KR/dataset-creation.ts   | 1 +
 web/i18n/pl-PL/dataset-creation.ts   | 1 +
 web/i18n/pt-BR/dataset-creation.ts   | 1 +
 web/i18n/ro-RO/dataset-creation.ts   | 1 +
 web/i18n/ru-RU/dataset-creation.ts   | 1 +
 web/i18n/tr-TR/dataset-creation.ts   | 1 +
 web/i18n/uk-UA/dataset-creation.ts   | 1 +
 web/i18n/vi-VN/dataset-creation.ts   | 1 +
 web/i18n/zh-Hant/dataset-creation.ts | 1 +
 16 files changed, 16 insertions(+)

diff --git a/web/i18n/de-DE/dataset-creation.ts b/web/i18n/de-DE/dataset-creation.ts
index 8b27395049e12..5251cc94e846a 100644
--- a/web/i18n/de-DE/dataset-creation.ts
+++ b/web/i18n/de-DE/dataset-creation.ts
@@ -133,6 +133,7 @@ const translation = {
     datasetSettingLink: 'Wissenseinstellungen.',
     websiteSource: 'Preprocess-Website',
     webpageUnit: 'Seiten',
+    separatorTip: 'Ein Trennzeichen ist das Zeichen, das zum Trennen von Text verwendet wird. \\n\\n und \\n sind häufig verwendete Trennzeichen zum Trennen von Absätzen und Zeilen. In Kombination mit Kommas (\\n\\n,\\n) werden Absätze nach Zeilen segmentiert, wenn die maximale Blocklänge überschritten wird. Sie können auch spezielle, von Ihnen selbst definierte Trennzeichen verwenden (z. B. ***).',
   },
   stepThree: {
     creationTitle: '🎉 Wissen erstellt',
diff --git a/web/i18n/es-ES/dataset-creation.ts b/web/i18n/es-ES/dataset-creation.ts
index 132c9cbb9b849..e093632bc4fc2 100644
--- a/web/i18n/es-ES/dataset-creation.ts
+++ b/web/i18n/es-ES/dataset-creation.ts
@@ -138,6 +138,7 @@ const translation = {
     indexSettingTip: 'Para cambiar el método de índice, por favor ve a la ',
     retrievalSettingTip: 'Para cambiar el método de índice, por favor ve a la ',
     datasetSettingLink: 'configuración del conocimiento.',
+    separatorTip: 'Un delimitador es el carácter que se utiliza para separar el texto. \\n\\n y \\n son delimitadores comúnmente utilizados para separar párrafos y líneas. Combinado con comas (\\n\\n,\\n), los párrafos se segmentarán por líneas cuando excedan la longitud máxima del fragmento. También puede utilizar delimitadores especiales definidos por usted mismo (por ejemplo, ***).',
   },
   stepThree: {
     creationTitle: '🎉 Conocimiento creado',
diff --git a/web/i18n/fa-IR/dataset-creation.ts b/web/i18n/fa-IR/dataset-creation.ts
index e6e6ad5bfb4d4..40dae4b02d826 100644
--- a/web/i18n/fa-IR/dataset-creation.ts
+++ b/web/i18n/fa-IR/dataset-creation.ts
@@ -138,6 +138,7 @@ const translation = {
     indexSettingTip: 'برای تغییر روش شاخص، لطفاً به',
     retrievalSettingTip: 'برای تغییر روش شاخص، لطفاً به',
     datasetSettingLink: 'تنظیمات دانش بروید.',
+    separatorTip: 'جداکننده نویسه ای است که برای جداسازی متن استفاده می شود. \\n\\n و \\n معمولا برای جداسازی پاراگراف ها و خطوط استفاده می شوند. همراه با کاما (\\n\\n,\\n)، پاراگراف ها زمانی که از حداکثر طول تکه فراتر می روند، با خطوط تقسیم بندی می شوند. همچنین می توانید از جداکننده های خاصی که توسط خودتان تعریف شده اند استفاده کنید (مثلا ***).',
   },
   stepThree: {
     creationTitle: ' دانش ایجاد شد',
diff --git a/web/i18n/fr-FR/dataset-creation.ts b/web/i18n/fr-FR/dataset-creation.ts
index c08a3e5731bb3..38066885d0c6e 100644
--- a/web/i18n/fr-FR/dataset-creation.ts
+++ b/web/i18n/fr-FR/dataset-creation.ts
@@ -133,6 +133,7 @@ const translation = {
     datasetSettingLink: 'Paramètres de connaissance.',
     webpageUnit: 'Pages',
     websiteSource: 'Site web de prétraitement',
+    separatorTip: 'Un délimiteur est le caractère utilisé pour séparer le texte. \\n\\n et \\n sont des délimiteurs couramment utilisés pour séparer les paragraphes et les lignes. Combiné à des virgules (\\n\\n,\\n), les paragraphes seront segmentés par des lignes lorsqu’ils dépasseront la longueur maximale des morceaux. Vous pouvez également utiliser des délimiteurs spéciaux définis par vous-même (par exemple ***).',
   },
   stepThree: {
     creationTitle: '🎉 Connaissance créée',
diff --git a/web/i18n/hi-IN/dataset-creation.ts b/web/i18n/hi-IN/dataset-creation.ts
index 0fa71acf4aa32..19c396081a37e 100644
--- a/web/i18n/hi-IN/dataset-creation.ts
+++ b/web/i18n/hi-IN/dataset-creation.ts
@@ -155,6 +155,7 @@ const translation = {
     indexSettingTip: 'इंडेक्स विधि बदलने के लिए, कृपया जाएं ',
     retrievalSettingTip: 'इंडेक्स विधि बदलने के लिए, कृपया जाएं ',
     datasetSettingLink: 'ज्ञान सेटिंग्स।',
+    separatorTip: 'एक सीमांकक पाठ को अलग करने के लिए उपयोग किया जाने वाला वर्ण है। \\n\\n और \\n आमतौर पर पैराग्राफ और लाइनों को अलग करने के लिए उपयोग किए जाने वाले सीमांकक हैं। अल्पविराम (\\n\\n,\\n) के साथ संयुक्त, अधिकतम खंड लंबाई से अधिक होने पर अनुच्छेदों को पंक्तियों द्वारा खंडित किया जाएगा। आप स्वयं द्वारा परिभाषित विशेष सीमांकक का भी उपयोग कर सकते हैं (उदा. ***).',
   },
   stepThree: {
     creationTitle: '🎉 ज्ञान बनाया गया',
diff --git a/web/i18n/it-IT/dataset-creation.ts b/web/i18n/it-IT/dataset-creation.ts
index 1629776bf3c09..46889b80a60c8 100644
--- a/web/i18n/it-IT/dataset-creation.ts
+++ b/web/i18n/it-IT/dataset-creation.ts
@@ -158,6 +158,7 @@ const translation = {
     indexSettingTip: 'Per cambiare il metodo di indicizzazione, vai alle ',
     retrievalSettingTip: 'Per cambiare il metodo di indicizzazione, vai alle ',
     datasetSettingLink: 'impostazioni della Conoscenza.',
+    separatorTip: 'Un delimitatore è il carattere utilizzato per separare il testo. \\n\\n e \\n sono delimitatori comunemente usati per separare paragrafi e righe. In combinazione con le virgole (\\n\\n,\\n), i paragrafi verranno segmentati per righe quando superano la lunghezza massima del blocco. È inoltre possibile utilizzare delimitatori speciali definiti dall\'utente (ad es. ***).',
   },
   stepThree: {
     creationTitle: '🎉 Conoscenza creata',
diff --git a/web/i18n/ja-JP/dataset-creation.ts b/web/i18n/ja-JP/dataset-creation.ts
index e6d204840a50e..d11a0c94e9bba 100644
--- a/web/i18n/ja-JP/dataset-creation.ts
+++ b/web/i18n/ja-JP/dataset-creation.ts
@@ -138,6 +138,7 @@ const translation = {
     indexSettingTip: 'インデックス方法を変更するには、',
     retrievalSettingTip: '検索方法を変更するには、',
     datasetSettingLink: 'ナレッジ設定',
+    separatorTip: '区切り文字は、テキストを区切るために使用される文字です。\\n\\n と \\n は、段落と行を区切るために一般的に使用される区切り記号です。カンマ (\\n\\n,\\n) と組み合わせると、最大チャンク長を超えると、段落は行で区切られます。自分で定義した特別な区切り文字を使用することもできます(例:***)。',
   },
   stepThree: {
     creationTitle: '🎉 ナレッジが作成されました',
diff --git a/web/i18n/ko-KR/dataset-creation.ts b/web/i18n/ko-KR/dataset-creation.ts
index e8851acd2fd51..6a4126d8dbe52 100644
--- a/web/i18n/ko-KR/dataset-creation.ts
+++ b/web/i18n/ko-KR/dataset-creation.ts
@@ -133,6 +133,7 @@ const translation = {
     datasetSettingLink: '지식 설정',
     webpageUnit: '페이지',
     websiteSource: '웹 사이트 전처리',
+    separatorTip: '구분 기호는 텍스트를 구분하는 데 사용되는 문자입니다. \\n\\n 및 \\n은 단락과 줄을 구분하는 데 일반적으로 사용되는 구분 기호입니다. 쉼표(\\n\\n,\\n)와 함께 사용하면 최대 청크 길이를 초과할 경우 단락이 줄로 분할됩니다. 직접 정의한 특수 구분 기호(예: ***)를 사용할 수도 있습니다.',
   },
   stepThree: {
     creationTitle: '🎉 지식이 생성되었습니다',
diff --git a/web/i18n/pl-PL/dataset-creation.ts b/web/i18n/pl-PL/dataset-creation.ts
index 64e50c6b33af0..f5b36e62eec26 100644
--- a/web/i18n/pl-PL/dataset-creation.ts
+++ b/web/i18n/pl-PL/dataset-creation.ts
@@ -146,6 +146,7 @@ const translation = {
     datasetSettingLink: 'ustawień Wiedzy.',
     webpageUnit: 'Stron',
     websiteSource: 'Witryna internetowa przetwarzania wstępnego',
+    separatorTip: 'Ogranicznik to znak używany do oddzielania tekstu. \\n\\n i \\n są powszechnie używanymi ogranicznikami do oddzielania akapitów i wierszy. W połączeniu z przecinkami (\\n\\n,\\n), akapity będą segmentowane wierszami po przekroczeniu maksymalnej długości fragmentu. Możesz również skorzystać ze zdefiniowanych przez siebie specjalnych ograniczników (np. ***).',
   },
   stepThree: {
     creationTitle: '🎉 Utworzono Wiedzę',
diff --git a/web/i18n/pt-BR/dataset-creation.ts b/web/i18n/pt-BR/dataset-creation.ts
index 4ab78a50c7706..511f0d5bcbd37 100644
--- a/web/i18n/pt-BR/dataset-creation.ts
+++ b/web/i18n/pt-BR/dataset-creation.ts
@@ -133,6 +133,7 @@ const translation = {
     datasetSettingLink: 'configurações do Conhecimento.',
     websiteSource: 'Site de pré-processamento',
     webpageUnit: 'Páginas',
+    separatorTip: 'Um delimitador é o caractere usado para separar o texto. \\n\\n e \\n são delimitadores comumente usados para separar parágrafos e linhas. Combinado com vírgulas (\\n\\n,\\n), os parágrafos serão segmentados por linhas ao exceder o comprimento máximo do bloco. Você também pode usar delimitadores especiais definidos por você (por exemplo, ***).',
   },
   stepThree: {
     creationTitle: '🎉 Conhecimento criado',
diff --git a/web/i18n/ro-RO/dataset-creation.ts b/web/i18n/ro-RO/dataset-creation.ts
index efe3bb246cc58..4ea0b0475827d 100644
--- a/web/i18n/ro-RO/dataset-creation.ts
+++ b/web/i18n/ro-RO/dataset-creation.ts
@@ -133,6 +133,7 @@ const translation = {
     datasetSettingLink: 'setările Cunoștinței.',
     webpageUnit: 'Pagini',
     websiteSource: 'Site-ul web de preprocesare',
+    separatorTip: 'Un delimitator este caracterul folosit pentru a separa textul. \\n\\n și \\n sunt delimitatori utilizați în mod obișnuit pentru separarea paragrafelor și liniilor. Combinate cu virgule (\\n\\n,\\n), paragrafele vor fi segmentate pe linii atunci când depășesc lungimea maximă a bucății. De asemenea, puteți utiliza delimitatori speciali definiți de dumneavoastră (de exemplu, ***).',
   },
   stepThree: {
     creationTitle: '🎉 Cunoștință creată',
diff --git a/web/i18n/ru-RU/dataset-creation.ts b/web/i18n/ru-RU/dataset-creation.ts
index c4dce774d818d..c97daeeeceac7 100644
--- a/web/i18n/ru-RU/dataset-creation.ts
+++ b/web/i18n/ru-RU/dataset-creation.ts
@@ -138,6 +138,7 @@ const translation = {
     indexSettingTip: 'Чтобы изменить метод индексации, пожалуйста, перейдите в ',
     retrievalSettingTip: 'Чтобы изменить метод индексации, пожалуйста, перейдите в ',
     datasetSettingLink: 'настройки базы знаний.',
+    separatorTip: 'Разделитель — это символ, используемый для разделения текста. \\n\\n и \\n — это часто используемые разделители для разделения абзацев и строк. В сочетании с запятыми (\\n\\n,\\n) абзацы будут сегментированы по строкам, если максимальная длина блока превышает их. Вы также можете использовать специальные разделители, определенные вами (например, ***).',
   },
   stepThree: {
     creationTitle: '🎉 База знаний создана',
diff --git a/web/i18n/tr-TR/dataset-creation.ts b/web/i18n/tr-TR/dataset-creation.ts
index b26608c39ffeb..c29e3045b8cf1 100644
--- a/web/i18n/tr-TR/dataset-creation.ts
+++ b/web/i18n/tr-TR/dataset-creation.ts
@@ -138,6 +138,7 @@ const translation = {
     indexSettingTip: 'Dizin yöntemini değiştirmek için, lütfen',
     retrievalSettingTip: 'Dizin yöntemini değiştirmek için, lütfen',
     datasetSettingLink: 'Bilgi ayarlarına gidin.',
+    separatorTip: 'Sınırlayıcı, metni ayırmak için kullanılan karakterdir. \\n\\n ve \\n, paragrafları ve satırları ayırmak için yaygın olarak kullanılan sınırlayıcılardır. Virgüllerle (\\n\\n,\\n) birleştirildiğinde, paragraflar maksimum öbek uzunluğunu aştığında satırlarla bölünür. Kendiniz tarafından tanımlanan özel sınırlayıcıları da kullanabilirsiniz (örn.',
   },
   stepThree: {
     creationTitle: '🎉 Bilgi oluşturuldu',
diff --git a/web/i18n/uk-UA/dataset-creation.ts b/web/i18n/uk-UA/dataset-creation.ts
index e4a38f41f4664..5b2c9503cf1c4 100644
--- a/web/i18n/uk-UA/dataset-creation.ts
+++ b/web/i18n/uk-UA/dataset-creation.ts
@@ -133,6 +133,7 @@ const translation = {
     datasetSettingLink: 'Налаштування знань.',
     webpageUnit: 'Сторінок',
     websiteSource: 'Веб-сайт попередньої обробки',
+    separatorTip: 'Роздільник – це символ, який використовується для поділу тексту. \\n\\n та \\n є часто використовуваними роздільниками для відокремлення абзаців та рядків. У поєднанні з комами (\\n\\n,\\n) абзаци будуть розділені лініями, якщо вони перевищують максимальну довжину фрагмента. Ви також можете використовувати спеціальні роздільники, визначені вами (наприклад, ***).',
   },
   stepThree: {
     creationTitle: '🎉 Знання створено',
diff --git a/web/i18n/vi-VN/dataset-creation.ts b/web/i18n/vi-VN/dataset-creation.ts
index da69020287334..af49575b906c4 100644
--- a/web/i18n/vi-VN/dataset-creation.ts
+++ b/web/i18n/vi-VN/dataset-creation.ts
@@ -133,6 +133,7 @@ const translation = {
     datasetSettingLink: 'cài đặt Kiến thức.',
     websiteSource: 'Trang web tiền xử lý',
     webpageUnit: 'Trang',
+    separatorTip: 'Dấu phân cách là ký tự được sử dụng để phân tách văn bản. \\n\\n và \\n là dấu phân cách thường được sử dụng để tách các đoạn văn và dòng. Kết hợp với dấu phẩy (\\n\\n,\\n), các đoạn văn sẽ được phân đoạn theo các dòng khi vượt quá độ dài đoạn tối đa. Bạn cũng có thể sử dụng dấu phân cách đặc biệt do chính bạn xác định (ví dụ: ***).',
   },
   stepThree: {
     creationTitle: '🎉 Kiến thức đã được tạo',
diff --git a/web/i18n/zh-Hant/dataset-creation.ts b/web/i18n/zh-Hant/dataset-creation.ts
index fd810d41c16bb..73a57db6a06c4 100644
--- a/web/i18n/zh-Hant/dataset-creation.ts
+++ b/web/i18n/zh-Hant/dataset-creation.ts
@@ -133,6 +133,7 @@ const translation = {
     datasetSettingLink: '知識庫設定。',
     websiteSource: '預處理網站',
     webpageUnit: '頁面',
+    separatorTip: '分隔符是用於分隔文字的字元。\\n\\n 和 \\n 是分隔段落和行的常用分隔符。與逗號 （\\n\\n，\\n） 組合使用時，當超過最大區塊長度時，段落將按行分段。您也可以使用自定義的特殊分隔符（例如 ***）。',
   },
   stepThree: {
     creationTitle: '🎉 知識庫已建立',

From e75c33a56166c0825faa1d07d955086a4358d125 Mon Sep 17 00:00:00 2001
From: =?UTF-8?q?=E6=96=B9=E7=A8=8B?= <1787003204@qq.com>
Date: Sat, 21 Sep 2024 17:30:58 +0800
Subject: [PATCH 004/235] Enhance Readme Documentation to Clarify the
 Importance of Celery Service (#8558)

---
 api/README.md | 4 +---
 1 file changed, 1 insertion(+), 3 deletions(-)

diff --git a/api/README.md b/api/README.md
index 70ca2e86a85a7..bab33f9293a31 100644
--- a/api/README.md
+++ b/api/README.md
@@ -65,14 +65,12 @@
 
 8. Start Dify [web](../web) service.
 9. Setup your application by visiting `http://localhost:3000`...
-10. If you need to debug local async processing, please start the worker service.
+10. If you need to handle and debug the async tasks (e.g. dataset importing and documents indexing), please start the worker service.
 
    ```bash
    poetry run python -m celery -A app.celery worker -P gevent -c 1 --loglevel INFO -Q dataset,generation,mail,ops_trace,app_deletion
    ```
 
-   The started celery app handles the async tasks, e.g. dataset importing and documents indexing.
-
 ## Testing
 
 1. Install dependencies for both the backend and the test environment

From b3cb97f0adff4b670d95bc5146d15c12382e1ead Mon Sep 17 00:00:00 2001
From: WalterMitty <waltermitty121906@gmail.com>
Date: Sat, 21 Sep 2024 17:31:49 +0800
Subject: [PATCH 005/235] docs: Update ssrf_proxy related doc link in
 docker-compose file (#8516)

---
 docker-legacy/docker-compose.middleware.yaml | 2 +-
 docker-legacy/docker-compose.yaml            | 2 +-
 docker/docker-compose.middleware.yaml        | 2 +-
 3 files changed, 3 insertions(+), 3 deletions(-)

diff --git a/docker-legacy/docker-compose.middleware.yaml b/docker-legacy/docker-compose.middleware.yaml
index fadbb3e6084c2..da54fe33fdd1a 100644
--- a/docker-legacy/docker-compose.middleware.yaml
+++ b/docker-legacy/docker-compose.middleware.yaml
@@ -73,7 +73,7 @@ services:
 
   # ssrf_proxy server
   # for more information, please refer to
-  # https://docs.dify.ai/learn-more/faq/self-host-faq#id-18.-why-is-ssrf_proxy-needed
+  # https://docs.dify.ai/learn-more/faq/install-faq#id-18.-why-is-ssrf_proxy-needed
   ssrf_proxy:
     image: ubuntu/squid:latest
     restart: always
diff --git a/docker-legacy/docker-compose.yaml b/docker-legacy/docker-compose.yaml
index 513915a1f1591..1636bb6a214bd 100644
--- a/docker-legacy/docker-compose.yaml
+++ b/docker-legacy/docker-compose.yaml
@@ -500,7 +500,7 @@ services:
 
   # ssrf_proxy server
   # for more information, please refer to
-  # https://docs.dify.ai/learn-more/faq/self-host-faq#id-18.-why-is-ssrf_proxy-needed
+  # https://docs.dify.ai/learn-more/faq/install-faq#id-18.-why-is-ssrf_proxy-needed
   ssrf_proxy:
     image: ubuntu/squid:latest
     restart: always
diff --git a/docker/docker-compose.middleware.yaml b/docker/docker-compose.middleware.yaml
index 251c62fee103c..d7900def73f1e 100644
--- a/docker/docker-compose.middleware.yaml
+++ b/docker/docker-compose.middleware.yaml
@@ -63,7 +63,7 @@ services:
 
   # ssrf_proxy server
   # for more information, please refer to
-  # https://docs.dify.ai/learn-more/faq/self-host-faq#id-18.-why-is-ssrf_proxy-needed
+  # https://docs.dify.ai/learn-more/faq/install-faq#id-18.-why-is-ssrf_proxy-needed
   ssrf_proxy:
     image: ubuntu/squid:latest
     restart: always

From 5541248264fdf69392917a25924786a1b834e91e Mon Sep 17 00:00:00 2001
From: Hongbin <61153998+BingGeX@users.noreply.github.com>
Date: Sat, 21 Sep 2024 17:33:15 +0800
Subject: [PATCH 006/235] =?UTF-8?q?Update=20the=20PerfXCloud=20provider=20?=
 =?UTF-8?q?model=20list=EF=BC=8CUpdate=20PerfXCloudProvider=20validate=5Fp?=
 =?UTF-8?q?rovider=5Fcredentials=20method.=20(#8587)?=
MIME-Version: 1.0
Content-Type: text/plain; charset=UTF-8
Content-Transfer-Encoding: 8bit

Co-authored-by: xhb <466010723@qq.com>
---
 .../perfxcloud/llm/Llama3-Chinese_v2.yaml     |  1 +
 .../Meta-Llama-3-70B-Instruct-GPTQ-Int4.yaml  |  1 +
 .../llm/Meta-Llama-3-8B-Instruct.yaml         |  1 +
 ...Meta-Llama-3.1-405B-Instruct-AWQ-INT4.yaml |  1 +
 .../llm/Qwen1.5-72B-Chat-GPTQ-Int4.yaml       |  1 +
 .../perfxcloud/llm/Qwen1.5-7B.yaml            |  1 +
 .../llm/Qwen2-72B-Instruct-AWQ-int4.yaml      | 61 ++++++++++++++++++
 .../llm/Qwen2-72B-Instruct-GPTQ-Int4.yaml     |  1 +
 .../perfxcloud/llm/Qwen2-7B-Instruct.yaml     | 63 +++++++++++++++++++
 .../perfxcloud/llm/Qwen2-7B.yaml              |  1 +
 .../perfxcloud/llm/Qwen2.5-72B-Instruct.yaml  | 61 ++++++++++++++++++
 .../perfxcloud/llm/Qwen2.5-7B-Instruct.yaml   | 61 ++++++++++++++++++
 .../llm/Reflection-Llama-3.1-70B.yaml         | 61 ++++++++++++++++++
 .../perfxcloud/llm/Yi-1_5-9B-Chat-16K.yaml    | 61 ++++++++++++++++++
 .../perfxcloud/llm/Yi-Coder-1.5B-Chat.yaml    | 61 ++++++++++++++++++
 .../perfxcloud/llm/Yi-Coder-9B-Chat.yaml      | 61 ++++++++++++++++++
 .../perfxcloud/llm/_position.yaml             | 19 ++++--
 .../perfxcloud/llm/chatglm3-6b.yaml           |  1 +
 .../perfxcloud/llm/deepseek-v2-chat.yaml      |  1 +
 .../perfxcloud/llm/deepseek-v2-lite-chat.yaml |  1 +
 .../model_providers/perfxcloud/perfxcloud.py  | 20 +-----
 .../text_embedding/gte-Qwen2-7B-instruct.yaml |  4 ++
 22 files changed, 520 insertions(+), 24 deletions(-)
 create mode 100644 api/core/model_runtime/model_providers/perfxcloud/llm/Qwen2-72B-Instruct-AWQ-int4.yaml
 create mode 100644 api/core/model_runtime/model_providers/perfxcloud/llm/Qwen2-7B-Instruct.yaml
 create mode 100644 api/core/model_runtime/model_providers/perfxcloud/llm/Qwen2.5-72B-Instruct.yaml
 create mode 100644 api/core/model_runtime/model_providers/perfxcloud/llm/Qwen2.5-7B-Instruct.yaml
 create mode 100644 api/core/model_runtime/model_providers/perfxcloud/llm/Reflection-Llama-3.1-70B.yaml
 create mode 100644 api/core/model_runtime/model_providers/perfxcloud/llm/Yi-1_5-9B-Chat-16K.yaml
 create mode 100644 api/core/model_runtime/model_providers/perfxcloud/llm/Yi-Coder-1.5B-Chat.yaml
 create mode 100644 api/core/model_runtime/model_providers/perfxcloud/llm/Yi-Coder-9B-Chat.yaml
 create mode 100644 api/core/model_runtime/model_providers/perfxcloud/text_embedding/gte-Qwen2-7B-instruct.yaml

diff --git a/api/core/model_runtime/model_providers/perfxcloud/llm/Llama3-Chinese_v2.yaml b/api/core/model_runtime/model_providers/perfxcloud/llm/Llama3-Chinese_v2.yaml
index 87712874b9f5c..bf91468fcf56b 100644
--- a/api/core/model_runtime/model_providers/perfxcloud/llm/Llama3-Chinese_v2.yaml
+++ b/api/core/model_runtime/model_providers/perfxcloud/llm/Llama3-Chinese_v2.yaml
@@ -59,3 +59,4 @@ pricing:
   output: "0.000"
   unit: "0.000"
   currency: RMB
+deprecated: true
diff --git a/api/core/model_runtime/model_providers/perfxcloud/llm/Meta-Llama-3-70B-Instruct-GPTQ-Int4.yaml b/api/core/model_runtime/model_providers/perfxcloud/llm/Meta-Llama-3-70B-Instruct-GPTQ-Int4.yaml
index f16f3de60b8d3..781b837e8ecee 100644
--- a/api/core/model_runtime/model_providers/perfxcloud/llm/Meta-Llama-3-70B-Instruct-GPTQ-Int4.yaml
+++ b/api/core/model_runtime/model_providers/perfxcloud/llm/Meta-Llama-3-70B-Instruct-GPTQ-Int4.yaml
@@ -59,3 +59,4 @@ pricing:
   output: "0.000"
   unit: "0.000"
   currency: RMB
+deprecated: true
diff --git a/api/core/model_runtime/model_providers/perfxcloud/llm/Meta-Llama-3-8B-Instruct.yaml b/api/core/model_runtime/model_providers/perfxcloud/llm/Meta-Llama-3-8B-Instruct.yaml
index 21267c240bc0e..67210e9020fe6 100644
--- a/api/core/model_runtime/model_providers/perfxcloud/llm/Meta-Llama-3-8B-Instruct.yaml
+++ b/api/core/model_runtime/model_providers/perfxcloud/llm/Meta-Llama-3-8B-Instruct.yaml
@@ -59,3 +59,4 @@ pricing:
   output: "0.000"
   unit: "0.000"
   currency: RMB
+deprecated: true
diff --git a/api/core/model_runtime/model_providers/perfxcloud/llm/Meta-Llama-3.1-405B-Instruct-AWQ-INT4.yaml b/api/core/model_runtime/model_providers/perfxcloud/llm/Meta-Llama-3.1-405B-Instruct-AWQ-INT4.yaml
index 80c7ec40f22cb..482632ff0673d 100644
--- a/api/core/model_runtime/model_providers/perfxcloud/llm/Meta-Llama-3.1-405B-Instruct-AWQ-INT4.yaml
+++ b/api/core/model_runtime/model_providers/perfxcloud/llm/Meta-Llama-3.1-405B-Instruct-AWQ-INT4.yaml
@@ -59,3 +59,4 @@ pricing:
   output: "0.000"
   unit: "0.000"
   currency: RMB
+deprecated: true
diff --git a/api/core/model_runtime/model_providers/perfxcloud/llm/Qwen1.5-72B-Chat-GPTQ-Int4.yaml b/api/core/model_runtime/model_providers/perfxcloud/llm/Qwen1.5-72B-Chat-GPTQ-Int4.yaml
index 841dd97f353b0..ddb6fd977c5a5 100644
--- a/api/core/model_runtime/model_providers/perfxcloud/llm/Qwen1.5-72B-Chat-GPTQ-Int4.yaml
+++ b/api/core/model_runtime/model_providers/perfxcloud/llm/Qwen1.5-72B-Chat-GPTQ-Int4.yaml
@@ -59,3 +59,4 @@ pricing:
   output: "0.000"
   unit: "0.000"
   currency: RMB
+deprecated: true
diff --git a/api/core/model_runtime/model_providers/perfxcloud/llm/Qwen1.5-7B.yaml b/api/core/model_runtime/model_providers/perfxcloud/llm/Qwen1.5-7B.yaml
index 33d5d12b223d2..024c79dbcfd76 100644
--- a/api/core/model_runtime/model_providers/perfxcloud/llm/Qwen1.5-7B.yaml
+++ b/api/core/model_runtime/model_providers/perfxcloud/llm/Qwen1.5-7B.yaml
@@ -59,3 +59,4 @@ pricing:
   output: "0.000"
   unit: "0.000"
   currency: RMB
+deprecated: true
diff --git a/api/core/model_runtime/model_providers/perfxcloud/llm/Qwen2-72B-Instruct-AWQ-int4.yaml b/api/core/model_runtime/model_providers/perfxcloud/llm/Qwen2-72B-Instruct-AWQ-int4.yaml
new file mode 100644
index 0000000000000..94f661f40d5c9
--- /dev/null
+++ b/api/core/model_runtime/model_providers/perfxcloud/llm/Qwen2-72B-Instruct-AWQ-int4.yaml
@@ -0,0 +1,61 @@
+model: Qwen2-72B-Instruct-AWQ-int4
+label:
+  en_US: Qwen2-72B-Instruct-AWQ-int4
+model_type: llm
+features:
+  - agent-thought
+model_properties:
+  mode: chat
+  context_size: 32768
+parameter_rules:
+  - name: temperature
+    use_template: temperature
+    type: float
+    default: 0.5
+    min: 0.0
+    max: 2.0
+    help:
+      zh_Hans: 用于控制随机性和多样性的程度。具体来说，temperature值控制了生成文本时对每个候选词的概率分布进行平滑的程度。较高的temperature值会降低概率分布的峰值，使得更多的低概率词被选择，生成结果更加多样化；而较低的temperature值则会增强概率分布的峰值，使得高概率词更容易被选择，生成结果更加确定。
+      en_US: Used to control the degree of randomness and diversity. Specifically, the temperature value controls the degree to which the probability distribution of each candidate word is smoothed when generating text. A higher temperature value will reduce the peak value of the probability distribution, allowing more low-probability words to be selected, and the generated results will be more diverse; while a lower temperature value will enhance the peak value of the probability distribution, making it easier for high-probability words to be selected. , the generated results are more certain.
+  - name: max_tokens
+    use_template: max_tokens
+    type: int
+    default: 600
+    min: 1
+    max: 1248
+    help:
+      zh_Hans: 用于指定模型在生成内容时token的最大数量，它定义了生成的上限，但不保证每次都会生成到这个数量。
+      en_US: It is used to specify the maximum number of tokens when the model generates content. It defines the upper limit of generation, but does not guarantee that this number will be generated every time.
+  - name: top_p
+    use_template: top_p
+    type: float
+    default: 0.8
+    min: 0.1
+    max: 0.9
+    help:
+      zh_Hans: 生成过程中核采样方法概率阈值，例如，取值为0.8时，仅保留概率加起来大于等于0.8的最可能token的最小集合作为候选集。取值范围为（0,1.0)，取值越大，生成的随机性越高；取值越低，生成的确定性越高。
+      en_US: The probability threshold of the kernel sampling method during the generation process. For example, when the value is 0.8, only the smallest set of the most likely tokens with a sum of probabilities greater than or equal to 0.8 is retained as the candidate set. The value range is (0,1.0). The larger the value, the higher the randomness generated; the lower the value, the higher the certainty generated.
+  - name: top_k
+    type: int
+    min: 0
+    max: 99
+    label:
+      zh_Hans: 取样数量
+      en_US: Top k
+    help:
+      zh_Hans: 生成时，采样候选集的大小。例如，取值为50时，仅将单次生成中得分最高的50个token组成随机采样的候选集。取值越大，生成的随机性越高；取值越小，生成的确定性越高。
+      en_US: The size of the sample candidate set when generated. For example, when the value is 50, only the 50 highest-scoring tokens in a single generation form a randomly sampled candidate set. The larger the value, the higher the randomness generated; the smaller the value, the higher the certainty generated.
+  - name: repetition_penalty
+    required: false
+    type: float
+    default: 1.1
+    label:
+      en_US: Repetition penalty
+    help:
+      zh_Hans: 用于控制模型生成时的重复度。提高repetition_penalty时可以降低模型生成的重复度。1.0表示不做惩罚。
+      en_US: Used to control the repeatability when generating models. Increasing repetition_penalty can reduce the duplication of model generation. 1.0 means no punishment.
+pricing:
+  input: "0.000"
+  output: "0.000"
+  unit: "0.000"
+  currency: RMB
diff --git a/api/core/model_runtime/model_providers/perfxcloud/llm/Qwen2-72B-Instruct-GPTQ-Int4.yaml b/api/core/model_runtime/model_providers/perfxcloud/llm/Qwen2-72B-Instruct-GPTQ-Int4.yaml
index 62255cc7d2e56..a06f8d5ab18f4 100644
--- a/api/core/model_runtime/model_providers/perfxcloud/llm/Qwen2-72B-Instruct-GPTQ-Int4.yaml
+++ b/api/core/model_runtime/model_providers/perfxcloud/llm/Qwen2-72B-Instruct-GPTQ-Int4.yaml
@@ -61,3 +61,4 @@ pricing:
   output: "0.000"
   unit: "0.000"
   currency: RMB
+deprecated: true
diff --git a/api/core/model_runtime/model_providers/perfxcloud/llm/Qwen2-7B-Instruct.yaml b/api/core/model_runtime/model_providers/perfxcloud/llm/Qwen2-7B-Instruct.yaml
new file mode 100644
index 0000000000000..4369411399e72
--- /dev/null
+++ b/api/core/model_runtime/model_providers/perfxcloud/llm/Qwen2-7B-Instruct.yaml
@@ -0,0 +1,63 @@
+model: Qwen2-7B-Instruct
+label:
+  en_US: Qwen2-7B-Instruct
+model_type: llm
+features:
+  - multi-tool-call
+  - agent-thought
+  - stream-tool-call
+model_properties:
+  mode: completion
+  context_size: 32768
+parameter_rules:
+  - name: temperature
+    use_template: temperature
+    type: float
+    default: 0.3
+    min: 0.0
+    max: 2.0
+    help:
+      zh_Hans: 用于控制随机性和多样性的程度。具体来说，temperature值控制了生成文本时对每个候选词的概率分布进行平滑的程度。较高的temperature值会降低概率分布的峰值，使得更多的低概率词被选择，生成结果更加多样化；而较低的temperature值则会增强概率分布的峰值，使得高概率词更容易被选择，生成结果更加确定。
+      en_US: Used to control the degree of randomness and diversity. Specifically, the temperature value controls the degree to which the probability distribution of each candidate word is smoothed when generating text. A higher temperature value will reduce the peak value of the probability distribution, allowing more low-probability words to be selected, and the generated results will be more diverse; while a lower temperature value will enhance the peak value of the probability distribution, making it easier for high-probability words to be selected. , the generated results are more certain.
+  - name: max_tokens
+    use_template: max_tokens
+    type: int
+    default: 600
+    min: 1
+    max: 2000
+    help:
+      zh_Hans: 用于指定模型在生成内容时token的最大数量，它定义了生成的上限，但不保证每次都会生成到这个数量。
+      en_US: It is used to specify the maximum number of tokens when the model generates content. It defines the upper limit of generation, but does not guarantee that this number will be generated every time.
+  - name: top_p
+    use_template: top_p
+    type: float
+    default: 0.8
+    min: 0.1
+    max: 0.9
+    help:
+      zh_Hans: 生成过程中核采样方法概率阈值，例如，取值为0.8时，仅保留概率加起来大于等于0.8的最可能token的最小集合作为候选集。取值范围为（0,1.0)，取值越大，生成的随机性越高；取值越低，生成的确定性越高。
+      en_US: The probability threshold of the kernel sampling method during the generation process. For example, when the value is 0.8, only the smallest set of the most likely tokens with a sum of probabilities greater than or equal to 0.8 is retained as the candidate set. The value range is (0,1.0). The larger the value, the higher the randomness generated; the lower the value, the higher the certainty generated.
+  - name: top_k
+    type: int
+    min: 0
+    max: 99
+    label:
+      zh_Hans: 取样数量
+      en_US: Top k
+    help:
+      zh_Hans: 生成时，采样候选集的大小。例如，取值为50时，仅将单次生成中得分最高的50个token组成随机采样的候选集。取值越大，生成的随机性越高；取值越小，生成的确定性越高。
+      en_US: The size of the sample candidate set when generated. For example, when the value is 50, only the 50 highest-scoring tokens in a single generation form a randomly sampled candidate set. The larger the value, the higher the randomness generated; the smaller the value, the higher the certainty generated.
+  - name: repetition_penalty
+    required: false
+    type: float
+    default: 1.1
+    label:
+      en_US: Repetition penalty
+    help:
+      zh_Hans: 用于控制模型生成时的重复度。提高repetition_penalty时可以降低模型生成的重复度。1.0表示不做惩罚。
+      en_US: Used to control the repeatability when generating models. Increasing repetition_penalty can reduce the duplication of model generation. 1.0 means no punishment.
+pricing:
+  input: "0.000"
+  output: "0.000"
+  unit: "0.000"
+  currency: RMB
diff --git a/api/core/model_runtime/model_providers/perfxcloud/llm/Qwen2-7B.yaml b/api/core/model_runtime/model_providers/perfxcloud/llm/Qwen2-7B.yaml
index 2f3f1f0225d71..d549ecd227dfb 100644
--- a/api/core/model_runtime/model_providers/perfxcloud/llm/Qwen2-7B.yaml
+++ b/api/core/model_runtime/model_providers/perfxcloud/llm/Qwen2-7B.yaml
@@ -61,3 +61,4 @@ pricing:
   output: "0.000"
   unit: "0.000"
   currency: RMB
+deprecated: true
diff --git a/api/core/model_runtime/model_providers/perfxcloud/llm/Qwen2.5-72B-Instruct.yaml b/api/core/model_runtime/model_providers/perfxcloud/llm/Qwen2.5-72B-Instruct.yaml
new file mode 100644
index 0000000000000..15cbf01f1f66d
--- /dev/null
+++ b/api/core/model_runtime/model_providers/perfxcloud/llm/Qwen2.5-72B-Instruct.yaml
@@ -0,0 +1,61 @@
+model: Qwen2.5-72B-Instruct
+label:
+  en_US: Qwen2.5-72B-Instruct
+model_type: llm
+features:
+  - agent-thought
+model_properties:
+  mode: chat
+  context_size: 30720
+parameter_rules:
+  - name: temperature
+    use_template: temperature
+    type: float
+    default: 0.5
+    min: 0.0
+    max: 2.0
+    help:
+      zh_Hans: 用于控制随机性和多样性的程度。具体来说，temperature值控制了生成文本时对每个候选词的概率分布进行平滑的程度。较高的temperature值会降低概率分布的峰值，使得更多的低概率词被选择，生成结果更加多样化；而较低的temperature值则会增强概率分布的峰值，使得高概率词更容易被选择，生成结果更加确定。
+      en_US: Used to control the degree of randomness and diversity. Specifically, the temperature value controls the degree to which the probability distribution of each candidate word is smoothed when generating text. A higher temperature value will reduce the peak value of the probability distribution, allowing more low-probability words to be selected, and the generated results will be more diverse; while a lower temperature value will enhance the peak value of the probability distribution, making it easier for high-probability words to be selected. , the generated results are more certain.
+  - name: max_tokens
+    use_template: max_tokens
+    type: int
+    default: 600
+    min: 1
+    max: 1248
+    help:
+      zh_Hans: 用于指定模型在生成内容时token的最大数量，它定义了生成的上限，但不保证每次都会生成到这个数量。
+      en_US: It is used to specify the maximum number of tokens when the model generates content. It defines the upper limit of generation, but does not guarantee that this number will be generated every time.
+  - name: top_p
+    use_template: top_p
+    type: float
+    default: 0.8
+    min: 0.1
+    max: 0.9
+    help:
+      zh_Hans: 生成过程中核采样方法概率阈值，例如，取值为0.8时，仅保留概率加起来大于等于0.8的最可能token的最小集合作为候选集。取值范围为（0,1.0)，取值越大，生成的随机性越高；取值越低，生成的确定性越高。
+      en_US: The probability threshold of the kernel sampling method during the generation process. For example, when the value is 0.8, only the smallest set of the most likely tokens with a sum of probabilities greater than or equal to 0.8 is retained as the candidate set. The value range is (0,1.0). The larger the value, the higher the randomness generated; the lower the value, the higher the certainty generated.
+  - name: top_k
+    type: int
+    min: 0
+    max: 99
+    label:
+      zh_Hans: 取样数量
+      en_US: Top k
+    help:
+      zh_Hans: 生成时，采样候选集的大小。例如，取值为50时，仅将单次生成中得分最高的50个token组成随机采样的候选集。取值越大，生成的随机性越高；取值越小，生成的确定性越高。
+      en_US: The size of the sample candidate set when generated. For example, when the value is 50, only the 50 highest-scoring tokens in a single generation form a randomly sampled candidate set. The larger the value, the higher the randomness generated; the smaller the value, the higher the certainty generated.
+  - name: repetition_penalty
+    required: false
+    type: float
+    default: 1.1
+    label:
+      en_US: Repetition penalty
+    help:
+      zh_Hans: 用于控制模型生成时的重复度。提高repetition_penalty时可以降低模型生成的重复度。1.0表示不做惩罚。
+      en_US: Used to control the repeatability when generating models. Increasing repetition_penalty can reduce the duplication of model generation. 1.0 means no punishment.
+pricing:
+  input: "0.000"
+  output: "0.000"
+  unit: "0.000"
+  currency: RMB
diff --git a/api/core/model_runtime/model_providers/perfxcloud/llm/Qwen2.5-7B-Instruct.yaml b/api/core/model_runtime/model_providers/perfxcloud/llm/Qwen2.5-7B-Instruct.yaml
new file mode 100644
index 0000000000000..dadc8f8f3275e
--- /dev/null
+++ b/api/core/model_runtime/model_providers/perfxcloud/llm/Qwen2.5-7B-Instruct.yaml
@@ -0,0 +1,61 @@
+model: Qwen2.5-7B-Instruct
+label:
+  en_US: Qwen2.5-7B-Instruct
+model_type: llm
+features:
+  - agent-thought
+model_properties:
+  mode: chat
+  context_size: 8192
+parameter_rules:
+  - name: temperature
+    use_template: temperature
+    type: float
+    default: 0.5
+    min: 0.0
+    max: 2.0
+    help:
+      zh_Hans: 用于控制随机性和多样性的程度。具体来说，temperature值控制了生成文本时对每个候选词的概率分布进行平滑的程度。较高的temperature值会降低概率分布的峰值，使得更多的低概率词被选择，生成结果更加多样化；而较低的temperature值则会增强概率分布的峰值，使得高概率词更容易被选择，生成结果更加确定。
+      en_US: Used to control the degree of randomness and diversity. Specifically, the temperature value controls the degree to which the probability distribution of each candidate word is smoothed when generating text. A higher temperature value will reduce the peak value of the probability distribution, allowing more low-probability words to be selected, and the generated results will be more diverse; while a lower temperature value will enhance the peak value of the probability distribution, making it easier for high-probability words to be selected. , the generated results are more certain.
+  - name: max_tokens
+    use_template: max_tokens
+    type: int
+    default: 600
+    min: 1
+    max: 1248
+    help:
+      zh_Hans: 用于指定模型在生成内容时token的最大数量，它定义了生成的上限，但不保证每次都会生成到这个数量。
+      en_US: It is used to specify the maximum number of tokens when the model generates content. It defines the upper limit of generation, but does not guarantee that this number will be generated every time.
+  - name: top_p
+    use_template: top_p
+    type: float
+    default: 0.8
+    min: 0.1
+    max: 0.9
+    help:
+      zh_Hans: 生成过程中核采样方法概率阈值，例如，取值为0.8时，仅保留概率加起来大于等于0.8的最可能token的最小集合作为候选集。取值范围为（0,1.0)，取值越大，生成的随机性越高；取值越低，生成的确定性越高。
+      en_US: The probability threshold of the kernel sampling method during the generation process. For example, when the value is 0.8, only the smallest set of the most likely tokens with a sum of probabilities greater than or equal to 0.8 is retained as the candidate set. The value range is (0,1.0). The larger the value, the higher the randomness generated; the lower the value, the higher the certainty generated.
+  - name: top_k
+    type: int
+    min: 0
+    max: 99
+    label:
+      zh_Hans: 取样数量
+      en_US: Top k
+    help:
+      zh_Hans: 生成时，采样候选集的大小。例如，取值为50时，仅将单次生成中得分最高的50个token组成随机采样的候选集。取值越大，生成的随机性越高；取值越小，生成的确定性越高。
+      en_US: The size of the sample candidate set when generated. For example, when the value is 50, only the 50 highest-scoring tokens in a single generation form a randomly sampled candidate set. The larger the value, the higher the randomness generated; the smaller the value, the higher the certainty generated.
+  - name: repetition_penalty
+    required: false
+    type: float
+    default: 1.1
+    label:
+      en_US: Repetition penalty
+    help:
+      zh_Hans: 用于控制模型生成时的重复度。提高repetition_penalty时可以降低模型生成的重复度。1.0表示不做惩罚。
+      en_US: Used to control the repeatability when generating models. Increasing repetition_penalty can reduce the duplication of model generation. 1.0 means no punishment.
+pricing:
+  input: "0.000"
+  output: "0.000"
+  unit: "0.000"
+  currency: RMB
diff --git a/api/core/model_runtime/model_providers/perfxcloud/llm/Reflection-Llama-3.1-70B.yaml b/api/core/model_runtime/model_providers/perfxcloud/llm/Reflection-Llama-3.1-70B.yaml
new file mode 100644
index 0000000000000..649be20b48abe
--- /dev/null
+++ b/api/core/model_runtime/model_providers/perfxcloud/llm/Reflection-Llama-3.1-70B.yaml
@@ -0,0 +1,61 @@
+model: Reflection-Llama-3.1-70B
+label:
+  en_US: Reflection-Llama-3.1-70B
+model_type: llm
+features:
+  - agent-thought
+model_properties:
+  mode: chat
+  context_size: 10240
+parameter_rules:
+  - name: temperature
+    use_template: temperature
+    type: float
+    default: 0.5
+    min: 0.0
+    max: 2.0
+    help:
+      zh_Hans: 用于控制随机性和多样性的程度。具体来说，temperature值控制了生成文本时对每个候选词的概率分布进行平滑的程度。较高的temperature值会降低概率分布的峰值，使得更多的低概率词被选择，生成结果更加多样化；而较低的temperature值则会增强概率分布的峰值，使得高概率词更容易被选择，生成结果更加确定。
+      en_US: Used to control the degree of randomness and diversity. Specifically, the temperature value controls the degree to which the probability distribution of each candidate word is smoothed when generating text. A higher temperature value will reduce the peak value of the probability distribution, allowing more low-probability words to be selected, and the generated results will be more diverse; while a lower temperature value will enhance the peak value of the probability distribution, making it easier for high-probability words to be selected. , the generated results are more certain.
+  - name: max_tokens
+    use_template: max_tokens
+    type: int
+    default: 600
+    min: 1
+    max: 1248
+    help:
+      zh_Hans: 用于指定模型在生成内容时token的最大数量，它定义了生成的上限，但不保证每次都会生成到这个数量。
+      en_US: It is used to specify the maximum number of tokens when the model generates content. It defines the upper limit of generation, but does not guarantee that this number will be generated every time.
+  - name: top_p
+    use_template: top_p
+    type: float
+    default: 0.8
+    min: 0.1
+    max: 0.9
+    help:
+      zh_Hans: 生成过程中核采样方法概率阈值，例如，取值为0.8时，仅保留概率加起来大于等于0.8的最可能token的最小集合作为候选集。取值范围为（0,1.0)，取值越大，生成的随机性越高；取值越低，生成的确定性越高。
+      en_US: The probability threshold of the kernel sampling method during the generation process. For example, when the value is 0.8, only the smallest set of the most likely tokens with a sum of probabilities greater than or equal to 0.8 is retained as the candidate set. The value range is (0,1.0). The larger the value, the higher the randomness generated; the lower the value, the higher the certainty generated.
+  - name: top_k
+    type: int
+    min: 0
+    max: 99
+    label:
+      zh_Hans: 取样数量
+      en_US: Top k
+    help:
+      zh_Hans: 生成时，采样候选集的大小。例如，取值为50时，仅将单次生成中得分最高的50个token组成随机采样的候选集。取值越大，生成的随机性越高；取值越小，生成的确定性越高。
+      en_US: The size of the sample candidate set when generated. For example, when the value is 50, only the 50 highest-scoring tokens in a single generation form a randomly sampled candidate set. The larger the value, the higher the randomness generated; the smaller the value, the higher the certainty generated.
+  - name: repetition_penalty
+    required: false
+    type: float
+    default: 1.1
+    label:
+      en_US: Repetition penalty
+    help:
+      zh_Hans: 用于控制模型生成时的重复度。提高repetition_penalty时可以降低模型生成的重复度。1.0表示不做惩罚。
+      en_US: Used to control the repeatability when generating models. Increasing repetition_penalty can reduce the duplication of model generation. 1.0 means no punishment.
+pricing:
+  input: "0.000"
+  output: "0.000"
+  unit: "0.000"
+  currency: RMB
diff --git a/api/core/model_runtime/model_providers/perfxcloud/llm/Yi-1_5-9B-Chat-16K.yaml b/api/core/model_runtime/model_providers/perfxcloud/llm/Yi-1_5-9B-Chat-16K.yaml
new file mode 100644
index 0000000000000..92eae6804f61f
--- /dev/null
+++ b/api/core/model_runtime/model_providers/perfxcloud/llm/Yi-1_5-9B-Chat-16K.yaml
@@ -0,0 +1,61 @@
+model: Yi-1_5-9B-Chat-16K
+label:
+  en_US: Yi-1_5-9B-Chat-16K
+model_type: llm
+features:
+  - agent-thought
+model_properties:
+  mode: chat
+  context_size: 16384
+parameter_rules:
+  - name: temperature
+    use_template: temperature
+    type: float
+    default: 0.5
+    min: 0.0
+    max: 2.0
+    help:
+      zh_Hans: 用于控制随机性和多样性的程度。具体来说，temperature值控制了生成文本时对每个候选词的概率分布进行平滑的程度。较高的temperature值会降低概率分布的峰值，使得更多的低概率词被选择，生成结果更加多样化；而较低的temperature值则会增强概率分布的峰值，使得高概率词更容易被选择，生成结果更加确定。
+      en_US: Used to control the degree of randomness and diversity. Specifically, the temperature value controls the degree to which the probability distribution of each candidate word is smoothed when generating text. A higher temperature value will reduce the peak value of the probability distribution, allowing more low-probability words to be selected, and the generated results will be more diverse; while a lower temperature value will enhance the peak value of the probability distribution, making it easier for high-probability words to be selected. , the generated results are more certain.
+  - name: max_tokens
+    use_template: max_tokens
+    type: int
+    default: 600
+    min: 1
+    max: 1248
+    help:
+      zh_Hans: 用于指定模型在生成内容时token的最大数量，它定义了生成的上限，但不保证每次都会生成到这个数量。
+      en_US: It is used to specify the maximum number of tokens when the model generates content. It defines the upper limit of generation, but does not guarantee that this number will be generated every time.
+  - name: top_p
+    use_template: top_p
+    type: float
+    default: 0.8
+    min: 0.1
+    max: 0.9
+    help:
+      zh_Hans: 生成过程中核采样方法概率阈值，例如，取值为0.8时，仅保留概率加起来大于等于0.8的最可能token的最小集合作为候选集。取值范围为（0,1.0)，取值越大，生成的随机性越高；取值越低，生成的确定性越高。
+      en_US: The probability threshold of the kernel sampling method during the generation process. For example, when the value is 0.8, only the smallest set of the most likely tokens with a sum of probabilities greater than or equal to 0.8 is retained as the candidate set. The value range is (0,1.0). The larger the value, the higher the randomness generated; the lower the value, the higher the certainty generated.
+  - name: top_k
+    type: int
+    min: 0
+    max: 99
+    label:
+      zh_Hans: 取样数量
+      en_US: Top k
+    help:
+      zh_Hans: 生成时，采样候选集的大小。例如，取值为50时，仅将单次生成中得分最高的50个token组成随机采样的候选集。取值越大，生成的随机性越高；取值越小，生成的确定性越高。
+      en_US: The size of the sample candidate set when generated. For example, when the value is 50, only the 50 highest-scoring tokens in a single generation form a randomly sampled candidate set. The larger the value, the higher the randomness generated; the smaller the value, the higher the certainty generated.
+  - name: repetition_penalty
+    required: false
+    type: float
+    default: 1.1
+    label:
+      en_US: Repetition penalty
+    help:
+      zh_Hans: 用于控制模型生成时的重复度。提高repetition_penalty时可以降低模型生成的重复度。1.0表示不做惩罚。
+      en_US: Used to control the repeatability when generating models. Increasing repetition_penalty can reduce the duplication of model generation. 1.0 means no punishment.
+pricing:
+  input: "0.000"
+  output: "0.000"
+  unit: "0.000"
+  currency: RMB
diff --git a/api/core/model_runtime/model_providers/perfxcloud/llm/Yi-Coder-1.5B-Chat.yaml b/api/core/model_runtime/model_providers/perfxcloud/llm/Yi-Coder-1.5B-Chat.yaml
new file mode 100644
index 0000000000000..0e21ce148c39b
--- /dev/null
+++ b/api/core/model_runtime/model_providers/perfxcloud/llm/Yi-Coder-1.5B-Chat.yaml
@@ -0,0 +1,61 @@
+model: Yi-Coder-1.5B-Chat
+label:
+  en_US: Yi-Coder-1.5B-Chat
+model_type: llm
+features:
+  - agent-thought
+model_properties:
+  mode: chat
+  context_size: 20480
+parameter_rules:
+  - name: temperature
+    use_template: temperature
+    type: float
+    default: 0.5
+    min: 0.0
+    max: 2.0
+    help:
+      zh_Hans: 用于控制随机性和多样性的程度。具体来说，temperature值控制了生成文本时对每个候选词的概率分布进行平滑的程度。较高的temperature值会降低概率分布的峰值，使得更多的低概率词被选择，生成结果更加多样化；而较低的temperature值则会增强概率分布的峰值，使得高概率词更容易被选择，生成结果更加确定。
+      en_US: Used to control the degree of randomness and diversity. Specifically, the temperature value controls the degree to which the probability distribution of each candidate word is smoothed when generating text. A higher temperature value will reduce the peak value of the probability distribution, allowing more low-probability words to be selected, and the generated results will be more diverse; while a lower temperature value will enhance the peak value of the probability distribution, making it easier for high-probability words to be selected. , the generated results are more certain.
+  - name: max_tokens
+    use_template: max_tokens
+    type: int
+    default: 600
+    min: 1
+    max: 1248
+    help:
+      zh_Hans: 用于指定模型在生成内容时token的最大数量，它定义了生成的上限，但不保证每次都会生成到这个数量。
+      en_US: It is used to specify the maximum number of tokens when the model generates content. It defines the upper limit of generation, but does not guarantee that this number will be generated every time.
+  - name: top_p
+    use_template: top_p
+    type: float
+    default: 0.8
+    min: 0.1
+    max: 0.9
+    help:
+      zh_Hans: 生成过程中核采样方法概率阈值，例如，取值为0.8时，仅保留概率加起来大于等于0.8的最可能token的最小集合作为候选集。取值范围为（0,1.0)，取值越大，生成的随机性越高；取值越低，生成的确定性越高。
+      en_US: The probability threshold of the kernel sampling method during the generation process. For example, when the value is 0.8, only the smallest set of the most likely tokens with a sum of probabilities greater than or equal to 0.8 is retained as the candidate set. The value range is (0,1.0). The larger the value, the higher the randomness generated; the lower the value, the higher the certainty generated.
+  - name: top_k
+    type: int
+    min: 0
+    max: 99
+    label:
+      zh_Hans: 取样数量
+      en_US: Top k
+    help:
+      zh_Hans: 生成时，采样候选集的大小。例如，取值为50时，仅将单次生成中得分最高的50个token组成随机采样的候选集。取值越大，生成的随机性越高；取值越小，生成的确定性越高。
+      en_US: The size of the sample candidate set when generated. For example, when the value is 50, only the 50 highest-scoring tokens in a single generation form a randomly sampled candidate set. The larger the value, the higher the randomness generated; the smaller the value, the higher the certainty generated.
+  - name: repetition_penalty
+    required: false
+    type: float
+    default: 1.1
+    label:
+      en_US: Repetition penalty
+    help:
+      zh_Hans: 用于控制模型生成时的重复度。提高repetition_penalty时可以降低模型生成的重复度。1.0表示不做惩罚。
+      en_US: Used to control the repeatability when generating models. Increasing repetition_penalty can reduce the duplication of model generation. 1.0 means no punishment.
+pricing:
+  input: "0.000"
+  output: "0.000"
+  unit: "0.000"
+  currency: RMB
diff --git a/api/core/model_runtime/model_providers/perfxcloud/llm/Yi-Coder-9B-Chat.yaml b/api/core/model_runtime/model_providers/perfxcloud/llm/Yi-Coder-9B-Chat.yaml
new file mode 100644
index 0000000000000..23b0841ce4ed6
--- /dev/null
+++ b/api/core/model_runtime/model_providers/perfxcloud/llm/Yi-Coder-9B-Chat.yaml
@@ -0,0 +1,61 @@
+model: Yi-Coder-9B-Chat
+label:
+  en_US: Yi-Coder-9B-Chat
+model_type: llm
+features:
+  - agent-thought
+model_properties:
+  mode: chat
+  context_size: 20480
+parameter_rules:
+  - name: temperature
+    use_template: temperature
+    type: float
+    default: 0.5
+    min: 0.0
+    max: 2.0
+    help:
+      zh_Hans: 用于控制随机性和多样性的程度。具体来说，temperature值控制了生成文本时对每个候选词的概率分布进行平滑的程度。较高的temperature值会降低概率分布的峰值，使得更多的低概率词被选择，生成结果更加多样化；而较低的temperature值则会增强概率分布的峰值，使得高概率词更容易被选择，生成结果更加确定。
+      en_US: Used to control the degree of randomness and diversity. Specifically, the temperature value controls the degree to which the probability distribution of each candidate word is smoothed when generating text. A higher temperature value will reduce the peak value of the probability distribution, allowing more low-probability words to be selected, and the generated results will be more diverse; while a lower temperature value will enhance the peak value of the probability distribution, making it easier for high-probability words to be selected. , the generated results are more certain.
+  - name: max_tokens
+    use_template: max_tokens
+    type: int
+    default: 600
+    min: 1
+    max: 1248
+    help:
+      zh_Hans: 用于指定模型在生成内容时token的最大数量，它定义了生成的上限，但不保证每次都会生成到这个数量。
+      en_US: It is used to specify the maximum number of tokens when the model generates content. It defines the upper limit of generation, but does not guarantee that this number will be generated every time.
+  - name: top_p
+    use_template: top_p
+    type: float
+    default: 0.8
+    min: 0.1
+    max: 0.9
+    help:
+      zh_Hans: 生成过程中核采样方法概率阈值，例如，取值为0.8时，仅保留概率加起来大于等于0.8的最可能token的最小集合作为候选集。取值范围为（0,1.0)，取值越大，生成的随机性越高；取值越低，生成的确定性越高。
+      en_US: The probability threshold of the kernel sampling method during the generation process. For example, when the value is 0.8, only the smallest set of the most likely tokens with a sum of probabilities greater than or equal to 0.8 is retained as the candidate set. The value range is (0,1.0). The larger the value, the higher the randomness generated; the lower the value, the higher the certainty generated.
+  - name: top_k
+    type: int
+    min: 0
+    max: 99
+    label:
+      zh_Hans: 取样数量
+      en_US: Top k
+    help:
+      zh_Hans: 生成时，采样候选集的大小。例如，取值为50时，仅将单次生成中得分最高的50个token组成随机采样的候选集。取值越大，生成的随机性越高；取值越小，生成的确定性越高。
+      en_US: The size of the sample candidate set when generated. For example, when the value is 50, only the 50 highest-scoring tokens in a single generation form a randomly sampled candidate set. The larger the value, the higher the randomness generated; the smaller the value, the higher the certainty generated.
+  - name: repetition_penalty
+    required: false
+    type: float
+    default: 1.1
+    label:
+      en_US: Repetition penalty
+    help:
+      zh_Hans: 用于控制模型生成时的重复度。提高repetition_penalty时可以降低模型生成的重复度。1.0表示不做惩罚。
+      en_US: Used to control the repeatability when generating models. Increasing repetition_penalty can reduce the duplication of model generation. 1.0 means no punishment.
+pricing:
+  input: "0.000"
+  output: "0.000"
+  unit: "0.000"
+  currency: RMB
diff --git a/api/core/model_runtime/model_providers/perfxcloud/llm/_position.yaml b/api/core/model_runtime/model_providers/perfxcloud/llm/_position.yaml
index 2c9eac0e49a4d..37bf400f1e347 100644
--- a/api/core/model_runtime/model_providers/perfxcloud/llm/_position.yaml
+++ b/api/core/model_runtime/model_providers/perfxcloud/llm/_position.yaml
@@ -1,15 +1,24 @@
-- Meta-Llama-3.1-405B-Instruct-AWQ-INT4
+- Qwen2.5-72B-Instruct
+- Qwen2.5-7B-Instruct
+- Yi-Coder-1.5B-Chat
+- Yi-Coder-9B-Chat
+- Qwen2-72B-Instruct-AWQ-int4
+- Yi-1_5-9B-Chat-16K
+- Qwen2-7B-Instruct
+- Reflection-Llama-3.1-70B
+- Qwen2-72B-Instruct
 - Meta-Llama-3.1-8B-Instruct
+
+- Meta-Llama-3.1-405B-Instruct-AWQ-INT4
 - Meta-Llama-3-70B-Instruct-GPTQ-Int4
+- chatglm3-6b
 - Meta-Llama-3-8B-Instruct
+- Llama3-Chinese_v2
+- deepseek-v2-lite-chat
 - Qwen2-72B-Instruct-GPTQ-Int4
-- Qwen2-72B-Instruct
 - Qwen2-7B
 - Qwen-14B-Chat-Int4
 - Qwen1.5-72B-Chat-GPTQ-Int4
 - Qwen1.5-7B
 - Qwen1.5-110B-Chat-GPTQ-Int4
 - deepseek-v2-chat
-- deepseek-v2-lite-chat
-- Llama3-Chinese_v2
-- chatglm3-6b
diff --git a/api/core/model_runtime/model_providers/perfxcloud/llm/chatglm3-6b.yaml b/api/core/model_runtime/model_providers/perfxcloud/llm/chatglm3-6b.yaml
index f9c26b7f9002c..75d80f784a71f 100644
--- a/api/core/model_runtime/model_providers/perfxcloud/llm/chatglm3-6b.yaml
+++ b/api/core/model_runtime/model_providers/perfxcloud/llm/chatglm3-6b.yaml
@@ -59,3 +59,4 @@ pricing:
   output: "0.000"
   unit: "0.000"
   currency: RMB
+deprecated: true
diff --git a/api/core/model_runtime/model_providers/perfxcloud/llm/deepseek-v2-chat.yaml b/api/core/model_runtime/model_providers/perfxcloud/llm/deepseek-v2-chat.yaml
index 078922ef95b08..fa9a7b7175e9d 100644
--- a/api/core/model_runtime/model_providers/perfxcloud/llm/deepseek-v2-chat.yaml
+++ b/api/core/model_runtime/model_providers/perfxcloud/llm/deepseek-v2-chat.yaml
@@ -59,3 +59,4 @@ pricing:
   output: "0.000"
   unit: "0.000"
   currency: RMB
+deprecated: true
diff --git a/api/core/model_runtime/model_providers/perfxcloud/llm/deepseek-v2-lite-chat.yaml b/api/core/model_runtime/model_providers/perfxcloud/llm/deepseek-v2-lite-chat.yaml
index 4ff3af7b51776..75a26d25051ae 100644
--- a/api/core/model_runtime/model_providers/perfxcloud/llm/deepseek-v2-lite-chat.yaml
+++ b/api/core/model_runtime/model_providers/perfxcloud/llm/deepseek-v2-lite-chat.yaml
@@ -59,3 +59,4 @@ pricing:
   output: "0.000"
   unit: "0.000"
   currency: RMB
+deprecated: true
diff --git a/api/core/model_runtime/model_providers/perfxcloud/perfxcloud.py b/api/core/model_runtime/model_providers/perfxcloud/perfxcloud.py
index 450d22fb75943..9a4ead031d018 100644
--- a/api/core/model_runtime/model_providers/perfxcloud/perfxcloud.py
+++ b/api/core/model_runtime/model_providers/perfxcloud/perfxcloud.py
@@ -1,7 +1,5 @@
 import logging
 
-from core.model_runtime.entities.model_entities import ModelType
-from core.model_runtime.errors.validate import CredentialsValidateFailedError
 from core.model_runtime.model_providers.__base.model_provider import ModelProvider
 
 logger = logging.getLogger(__name__)
@@ -9,20 +7,4 @@
 
 class PerfXCloudProvider(ModelProvider):
     def validate_provider_credentials(self, credentials: dict) -> None:
-        """
-        Validate provider credentials
-        if validate failed, raise exception
-
-        :param credentials: provider credentials, credentials form defined in `provider_credential_schema`.
-        """
-        try:
-            model_instance = self.get_model_instance(ModelType.LLM)
-
-            # Use `Qwen2_72B_Chat_GPTQ_Int4` model for validate,
-            # no matter what model you pass in, text completion model or chat model
-            model_instance.validate_credentials(model="Qwen2-72B-Instruct-GPTQ-Int4", credentials=credentials)
-        except CredentialsValidateFailedError as ex:
-            raise ex
-        except Exception as ex:
-            logger.exception(f"{self.get_provider_schema().provider} credentials validate failed")
-            raise ex
+        pass
diff --git a/api/core/model_runtime/model_providers/perfxcloud/text_embedding/gte-Qwen2-7B-instruct.yaml b/api/core/model_runtime/model_providers/perfxcloud/text_embedding/gte-Qwen2-7B-instruct.yaml
new file mode 100644
index 0000000000000..03db0d8bce850
--- /dev/null
+++ b/api/core/model_runtime/model_providers/perfxcloud/text_embedding/gte-Qwen2-7B-instruct.yaml
@@ -0,0 +1,4 @@
+model: gte-Qwen2-7B-instruct
+model_type: text-embedding
+model_properties:
+  context_size: 2048

From 5ddb601e4345a5099668618d5e0e7e6c3dd56c79 Mon Sep 17 00:00:00 2001
From: AAEE86 <33052466+AAEE86@users.noreply.github.com>
Date: Sat, 21 Sep 2024 18:08:07 +0800
Subject: [PATCH 007/235] add MixtralAI Model (#8517)

---
 .../mistralai/llm/_position.yaml              |  5 ++
 .../mistralai/llm/codestral-latest.yaml       | 51 +++++++++++++++++++
 .../mistralai/llm/mistral-embed.yaml          | 51 +++++++++++++++++++
 .../mistralai/llm/open-codestral-mamba.yaml   | 51 +++++++++++++++++++
 .../mistralai/llm/open-mistral-nemo.yaml      | 51 +++++++++++++++++++
 .../mistralai/llm/pixtral-12b-2409.yaml       | 51 +++++++++++++++++++
 6 files changed, 260 insertions(+)
 create mode 100644 api/core/model_runtime/model_providers/mistralai/llm/codestral-latest.yaml
 create mode 100644 api/core/model_runtime/model_providers/mistralai/llm/mistral-embed.yaml
 create mode 100644 api/core/model_runtime/model_providers/mistralai/llm/open-codestral-mamba.yaml
 create mode 100644 api/core/model_runtime/model_providers/mistralai/llm/open-mistral-nemo.yaml
 create mode 100644 api/core/model_runtime/model_providers/mistralai/llm/pixtral-12b-2409.yaml

diff --git a/api/core/model_runtime/model_providers/mistralai/llm/_position.yaml b/api/core/model_runtime/model_providers/mistralai/llm/_position.yaml
index 751003d71e888..bdb06b7fff637 100644
--- a/api/core/model_runtime/model_providers/mistralai/llm/_position.yaml
+++ b/api/core/model_runtime/model_providers/mistralai/llm/_position.yaml
@@ -1,3 +1,8 @@
+- pixtral-12b-2409
+- codestral-latest
+- mistral-embed
+- open-mistral-nemo
+- open-codestral-mamba
 - open-mistral-7b
 - open-mixtral-8x7b
 - open-mixtral-8x22b
diff --git a/api/core/model_runtime/model_providers/mistralai/llm/codestral-latest.yaml b/api/core/model_runtime/model_providers/mistralai/llm/codestral-latest.yaml
new file mode 100644
index 0000000000000..5f1260233fe97
--- /dev/null
+++ b/api/core/model_runtime/model_providers/mistralai/llm/codestral-latest.yaml
@@ -0,0 +1,51 @@
+model: codestral-latest
+label:
+  zh_Hans: codestral-latest
+  en_US: codestral-latest
+model_type: llm
+features:
+  - agent-thought
+model_properties:
+  mode: chat
+  context_size: 32000
+parameter_rules:
+  - name: temperature
+    use_template: temperature
+    default: 0.7
+    min: 0
+    max: 1
+  - name: top_p
+    use_template: top_p
+    default: 1
+    min: 0
+    max: 1
+  - name: max_tokens
+    use_template: max_tokens
+    default: 1024
+    min: 1
+    max: 4096
+  - name: safe_prompt
+    default: false
+    type: boolean
+    help:
+      en_US: Whether to inject a safety prompt before all conversations.
+      zh_Hans: 是否开启提示词审查
+    label:
+      en_US: SafePrompt
+      zh_Hans: 提示词审查
+  - name: random_seed
+    type: int
+    help:
+      en_US: The seed to use for random sampling. If set, different calls will generate deterministic results.
+      zh_Hans: 当开启随机数种子以后，你可以通过指定一个固定的种子来使得回答结果更加稳定
+    label:
+      en_US: RandomSeed
+      zh_Hans: 随机数种子
+    default: 0
+    min: 0
+    max: 2147483647
+pricing:
+  input: '0.008'
+  output: '0.024'
+  unit: '0.001'
+  currency: USD
diff --git a/api/core/model_runtime/model_providers/mistralai/llm/mistral-embed.yaml b/api/core/model_runtime/model_providers/mistralai/llm/mistral-embed.yaml
new file mode 100644
index 0000000000000..d759103d08a94
--- /dev/null
+++ b/api/core/model_runtime/model_providers/mistralai/llm/mistral-embed.yaml
@@ -0,0 +1,51 @@
+model: mistral-embed
+label:
+  zh_Hans: mistral-embed
+  en_US: mistral-embed
+model_type: llm
+features:
+  - agent-thought
+model_properties:
+  mode: chat
+  context_size: 8192
+parameter_rules:
+  - name: temperature
+    use_template: temperature
+    default: 0.7
+    min: 0
+    max: 1
+  - name: top_p
+    use_template: top_p
+    default: 1
+    min: 0
+    max: 1
+  - name: max_tokens
+    use_template: max_tokens
+    default: 1024
+    min: 1
+    max: 1024
+  - name: safe_prompt
+    default: false
+    type: boolean
+    help:
+      en_US: Whether to inject a safety prompt before all conversations.
+      zh_Hans: 是否开启提示词审查
+    label:
+      en_US: SafePrompt
+      zh_Hans: 提示词审查
+  - name: random_seed
+    type: int
+    help:
+      en_US: The seed to use for random sampling. If set, different calls will generate deterministic results.
+      zh_Hans: 当开启随机数种子以后，你可以通过指定一个固定的种子来使得回答结果更加稳定
+    label:
+      en_US: RandomSeed
+      zh_Hans: 随机数种子
+    default: 0
+    min: 0
+    max: 2147483647
+pricing:
+  input: '0.008'
+  output: '0.024'
+  unit: '0.001'
+  currency: USD
diff --git a/api/core/model_runtime/model_providers/mistralai/llm/open-codestral-mamba.yaml b/api/core/model_runtime/model_providers/mistralai/llm/open-codestral-mamba.yaml
new file mode 100644
index 0000000000000..d7ffb9ea02084
--- /dev/null
+++ b/api/core/model_runtime/model_providers/mistralai/llm/open-codestral-mamba.yaml
@@ -0,0 +1,51 @@
+model: open-codestral-mamba
+label:
+  zh_Hans: open-codestral-mamba
+  en_US: open-codestral-mamba
+model_type: llm
+features:
+  - agent-thought
+model_properties:
+  mode: chat
+  context_size: 256000
+parameter_rules:
+  - name: temperature
+    use_template: temperature
+    default: 0.7
+    min: 0
+    max: 1
+  - name: top_p
+    use_template: top_p
+    default: 1
+    min: 0
+    max: 1
+  - name: max_tokens
+    use_template: max_tokens
+    default: 1024
+    min: 1
+    max: 16384
+  - name: safe_prompt
+    default: false
+    type: boolean
+    help:
+      en_US: Whether to inject a safety prompt before all conversations.
+      zh_Hans: 是否开启提示词审查
+    label:
+      en_US: SafePrompt
+      zh_Hans: 提示词审查
+  - name: random_seed
+    type: int
+    help:
+      en_US: The seed to use for random sampling. If set, different calls will generate deterministic results.
+      zh_Hans: 当开启随机数种子以后，你可以通过指定一个固定的种子来使得回答结果更加稳定
+    label:
+      en_US: RandomSeed
+      zh_Hans: 随机数种子
+    default: 0
+    min: 0
+    max: 2147483647
+pricing:
+  input: '0.008'
+  output: '0.024'
+  unit: '0.001'
+  currency: USD
diff --git a/api/core/model_runtime/model_providers/mistralai/llm/open-mistral-nemo.yaml b/api/core/model_runtime/model_providers/mistralai/llm/open-mistral-nemo.yaml
new file mode 100644
index 0000000000000..dcda4fbce7e82
--- /dev/null
+++ b/api/core/model_runtime/model_providers/mistralai/llm/open-mistral-nemo.yaml
@@ -0,0 +1,51 @@
+model: open-mistral-nemo
+label:
+  zh_Hans: open-mistral-nemo
+  en_US: open-mistral-nemo
+model_type: llm
+features:
+  - agent-thought
+model_properties:
+  mode: chat
+  context_size: 128000
+parameter_rules:
+  - name: temperature
+    use_template: temperature
+    default: 0.7
+    min: 0
+    max: 1
+  - name: top_p
+    use_template: top_p
+    default: 1
+    min: 0
+    max: 1
+  - name: max_tokens
+    use_template: max_tokens
+    default: 1024
+    min: 1
+    max: 8192
+  - name: safe_prompt
+    default: false
+    type: boolean
+    help:
+      en_US: Whether to inject a safety prompt before all conversations.
+      zh_Hans: 是否开启提示词审查
+    label:
+      en_US: SafePrompt
+      zh_Hans: 提示词审查
+  - name: random_seed
+    type: int
+    help:
+      en_US: The seed to use for random sampling. If set, different calls will generate deterministic results.
+      zh_Hans: 当开启随机数种子以后，你可以通过指定一个固定的种子来使得回答结果更加稳定
+    label:
+      en_US: RandomSeed
+      zh_Hans: 随机数种子
+    default: 0
+    min: 0
+    max: 2147483647
+pricing:
+  input: '0.008'
+  output: '0.024'
+  unit: '0.001'
+  currency: USD
diff --git a/api/core/model_runtime/model_providers/mistralai/llm/pixtral-12b-2409.yaml b/api/core/model_runtime/model_providers/mistralai/llm/pixtral-12b-2409.yaml
new file mode 100644
index 0000000000000..0b002b49cac8e
--- /dev/null
+++ b/api/core/model_runtime/model_providers/mistralai/llm/pixtral-12b-2409.yaml
@@ -0,0 +1,51 @@
+model: pixtral-12b-2409
+label:
+  zh_Hans: pixtral-12b-2409
+  en_US: pixtral-12b-2409
+model_type: llm
+features:
+  - agent-thought
+model_properties:
+  mode: chat
+  context_size: 128000
+parameter_rules:
+  - name: temperature
+    use_template: temperature
+    default: 0.7
+    min: 0
+    max: 1
+  - name: top_p
+    use_template: top_p
+    default: 1
+    min: 0
+    max: 1
+  - name: max_tokens
+    use_template: max_tokens
+    default: 1024
+    min: 1
+    max: 8192
+  - name: safe_prompt
+    default: false
+    type: boolean
+    help:
+      en_US: Whether to inject a safety prompt before all conversations.
+      zh_Hans: 是否开启提示词审查
+    label:
+      en_US: SafePrompt
+      zh_Hans: 提示词审查
+  - name: random_seed
+    type: int
+    help:
+      en_US: The seed to use for random sampling. If set, different calls will generate deterministic results.
+      zh_Hans: 当开启随机数种子以后，你可以通过指定一个固定的种子来使得回答结果更加稳定
+    label:
+      en_US: RandomSeed
+      zh_Hans: 随机数种子
+    default: 0
+    min: 0
+    max: 2147483647
+pricing:
+  input: '0.008'
+  output: '0.024'
+  unit: '0.001'
+  currency: USD

From 8219f9e0909a1b7ef6b3f887133fc328f62bcff6 Mon Sep 17 00:00:00 2001
From: Nam Vu <zuzoovn@gmail.com>
Date: Sat, 21 Sep 2024 19:49:01 +0700
Subject: [PATCH 008/235] fix: api/core/ops/ops_trace_manager.py (#8501)

---
 api/core/ops/ops_trace_manager.py | 13 ++++++++++---
 1 file changed, 10 insertions(+), 3 deletions(-)

diff --git a/api/core/ops/ops_trace_manager.py b/api/core/ops/ops_trace_manager.py
index 6f17bade9748e..876790f5b8889 100644
--- a/api/core/ops/ops_trace_manager.py
+++ b/api/core/ops/ops_trace_manager.py
@@ -176,11 +176,18 @@ def get_ops_trace_instance(
             return None
 
         app: App = db.session.query(App).filter(App.id == app_id).first()
+
+        if app is None:
+            return None
+
         app_ops_trace_config = json.loads(app.tracing) if app.tracing else None
 
-        if app_ops_trace_config is not None:
-            tracing_provider = app_ops_trace_config.get("tracing_provider")
-        else:
+        if app_ops_trace_config is None:
+            return None
+
+        tracing_provider = app_ops_trace_config.get("tracing_provider")
+
+        if tracing_provider is None or tracing_provider not in provider_config_map:
             return None
 
         # decrypt_token

From 1a8dcae10e4e0f7411cce3e48171dde1a79db951 Mon Sep 17 00:00:00 2001
From: AAEE86 <33052466+AAEE86@users.noreply.github.com>
Date: Sat, 21 Sep 2024 22:52:10 +0800
Subject: [PATCH 009/235] add Qwen custom add model interface (#8565)

---
 .../model_providers/tongyi/llm/llm.py         | 71 ++++++++++++++++++-
 .../model_providers/tongyi/tongyi.yaml        | 18 +++++
 2 files changed, 88 insertions(+), 1 deletion(-)

diff --git a/api/core/model_runtime/model_providers/tongyi/llm/llm.py b/api/core/model_runtime/model_providers/tongyi/llm/llm.py
index 1d4eba6668c0c..f90c7f075fe86 100644
--- a/api/core/model_runtime/model_providers/tongyi/llm/llm.py
+++ b/api/core/model_runtime/model_providers/tongyi/llm/llm.py
@@ -30,7 +30,15 @@
     ToolPromptMessage,
     UserPromptMessage,
 )
-from core.model_runtime.entities.model_entities import ModelFeature
+from core.model_runtime.entities.model_entities import (
+    AIModelEntity,
+    FetchFrom,
+    I18nObject,
+    ModelFeature,
+    ModelType,
+    ParameterRule,
+    ParameterType,
+)
 from core.model_runtime.errors.invoke import (
     InvokeAuthorizationError,
     InvokeBadRequestError,
@@ -520,3 +528,64 @@ def _invoke_error_mapping(self) -> dict[type[InvokeError], list[type[Exception]]
                 UnsupportedHTTPMethod,
             ],
         }
+
+    def get_customizable_model_schema(self, model: str, credentials: dict) -> AIModelEntity | None:
+        """
+        Architecture for defining customizable models
+
+        :param model: model name
+        :param credentials: model credentials
+        :return: AIModelEntity or None
+        """
+        rules = [
+            ParameterRule(
+                name="temperature",
+                type=ParameterType.FLOAT,
+                use_template="temperature",
+                label=I18nObject(zh_Hans="温度", en_US="Temperature"),
+            ),
+            ParameterRule(
+                name="top_p",
+                type=ParameterType.FLOAT,
+                use_template="top_p",
+                label=I18nObject(zh_Hans="Top P", en_US="Top P"),
+            ),
+            ParameterRule(
+                name="top_k",
+                type=ParameterType.INT,
+                min=0,
+                max=99,
+                label=I18nObject(zh_Hans="top_k", en_US="top_k"),
+            ),
+            ParameterRule(
+                name="max_tokens",
+                type=ParameterType.INT,
+                min=1,
+                max=128000,
+                default=1024,
+                label=I18nObject(zh_Hans="最大生成长度", en_US="Max Tokens"),
+            ),
+            ParameterRule(
+                name="seed",
+                type=ParameterType.INT,
+                default=1234,
+                label=I18nObject(zh_Hans="随机种子", en_US="Random Seed"),
+            ),
+            ParameterRule(
+                name="repetition_penalty",
+                type=ParameterType.FLOAT,
+                default=1.1,
+                label=I18nObject(zh_Hans="重复惩罚", en_US="Repetition Penalty"),
+            ),
+        ]
+
+        entity = AIModelEntity(
+            model=model,
+            label=I18nObject(en_US=model),
+            fetch_from=FetchFrom.CUSTOMIZABLE_MODEL,
+            model_type=ModelType.LLM,
+            model_properties={},
+            parameter_rules=rules,
+        )
+
+        return entity
diff --git a/api/core/model_runtime/model_providers/tongyi/tongyi.yaml b/api/core/model_runtime/model_providers/tongyi/tongyi.yaml
index de2c289c947c3..fabe6d90e6b20 100644
--- a/api/core/model_runtime/model_providers/tongyi/tongyi.yaml
+++ b/api/core/model_runtime/model_providers/tongyi/tongyi.yaml
@@ -20,6 +20,7 @@ supported_model_types:
   - text-embedding
 configurate_methods:
   - predefined-model
+  - customizable-model
 provider_credential_schema:
   credential_form_schemas:
     - variable: dashscope_api_key
@@ -30,3 +31,20 @@ provider_credential_schema:
       placeholder:
         zh_Hans: 在此输入您的 API Key
         en_US: Enter your API Key
+model_credential_schema:
+  model:
+    label:
+      en_US: Model Name
+      zh_Hans: 模型名称
+    placeholder:
+      en_US: Enter full model name
+      zh_Hans: 输入模型全称
+  credential_form_schemas:
+    - variable: dashscope_api_key
+      required: true
+      label:
+        en_US: API Key
+      type: secret-input
+      placeholder:
+        zh_Hans: 在此输入您的 API Key
+        en_US: Enter your API Key

From 831c5a93af83058da71f56c2df1d634ee768e001 Mon Sep 17 00:00:00 2001
From: zhuhao <37029601+hwzhuhao@users.noreply.github.com>
Date: Sat, 21 Sep 2024 22:56:37 +0800
Subject: [PATCH 010/235] refactor(ops): Optimize the iteration for
 filter_none_values and use logging.error to record logs when an exception
 occurs (#8461)

---
 api/core/ops/entities/trace_entity.py | 3 +--
 api/core/ops/ops_trace_manager.py     | 4 ++--
 api/core/ops/utils.py                 | 7 +++++--
 3 files changed, 8 insertions(+), 6 deletions(-)

diff --git a/api/core/ops/entities/trace_entity.py b/api/core/ops/entities/trace_entity.py
index f27a0af6e0de1..db6ce9d9c38ad 100644
--- a/api/core/ops/entities/trace_entity.py
+++ b/api/core/ops/entities/trace_entity.py
@@ -21,8 +21,7 @@ def ensure_type(cls, v):
             return None
         if isinstance(v, str | dict | list):
             return v
-        else:
-            return ""
+        return ""
 
 
 class WorkflowTraceInfo(BaseTraceInfo):
diff --git a/api/core/ops/ops_trace_manager.py b/api/core/ops/ops_trace_manager.py
index 876790f5b8889..0200f4a32df16 100644
--- a/api/core/ops/ops_trace_manager.py
+++ b/api/core/ops/ops_trace_manager.py
@@ -708,7 +708,7 @@ def add_trace_task(self, trace_task: TraceTask):
                 trace_task.app_id = self.app_id
                 trace_manager_queue.put(trace_task)
         except Exception as e:
-            logging.debug(f"Error adding trace task: {e}")
+            logging.error(f"Error adding trace task: {e}")
         finally:
             self.start_timer()
 
@@ -727,7 +727,7 @@ def run(self):
             if tasks:
                 self.send_to_celery(tasks)
         except Exception as e:
-            logging.debug(f"Error processing trace tasks: {e}")
+            logging.error(f"Error processing trace tasks: {e}")
 
     def start_timer(self):
         global trace_manager_timer
diff --git a/api/core/ops/utils.py b/api/core/ops/utils.py
index 498685b3426d3..3cd3fb57565a5 100644
--- a/api/core/ops/utils.py
+++ b/api/core/ops/utils.py
@@ -6,12 +6,15 @@
 
 
 def filter_none_values(data: dict):
+    new_data = {}
     for key, value in data.items():
         if value is None:
             continue
         if isinstance(value, datetime):
-            data[key] = value.isoformat()
-    return {key: value for key, value in data.items() if value is not None}
+            new_data[key] = value.isoformat()
+        else:
+            new_data[key] = value
+    return new_data
 
 
 def get_message_data(message_id):

From b32a7713e024d63d327303b3749d9380fbaa592e Mon Sep 17 00:00:00 2001
From: Joe <79627742+ZhouhaoJiang@users.noreply.github.com>
Date: Sat, 21 Sep 2024 23:59:50 +0800
Subject: [PATCH 011/235] feat: update pyproject.toml (#8368)

---
 api/poetry.lock    | 100 ++++++++++++++++++++++++---------------------
 api/pyproject.toml |   2 +-
 2 files changed, 54 insertions(+), 48 deletions(-)

diff --git a/api/poetry.lock b/api/poetry.lock
index 28c688cc9cde4..8d4d680031a7c 100644
--- a/api/poetry.lock
+++ b/api/poetry.lock
@@ -616,13 +616,13 @@ files = [
 
 [[package]]
 name = "azure-core"
-version = "1.30.2"
+version = "1.31.0"
 description = "Microsoft Azure Core Library for Python"
 optional = false
 python-versions = ">=3.8"
 files = [
-    {file = "azure-core-1.30.2.tar.gz", hash = "sha256:a14dc210efcd608821aa472d9fb8e8d035d29b68993819147bc290a8ac224472"},
-    {file = "azure_core-1.30.2-py3-none-any.whl", hash = "sha256:cf019c1ca832e96274ae85abd3d9f752397194d9fea3b41487290562ac8abe4a"},
+    {file = "azure_core-1.31.0-py3-none-any.whl", hash = "sha256:22954de3777e0250029360ef31d80448ef1be13b80a459bff80ba7073379e2cd"},
+    {file = "azure_core-1.31.0.tar.gz", hash = "sha256:656a0dd61e1869b1506b7c6a3b31d62f15984b1a573d6326f6aa2f3e4123284b"},
 ]
 
 [package.dependencies]
@@ -828,13 +828,13 @@ crt = ["botocore[crt] (>=1.21.0,<2.0a0)"]
 
 [[package]]
 name = "botocore"
-version = "1.35.17"
+version = "1.35.19"
 description = "Low-level, data-driven core of boto 3."
 optional = false
 python-versions = ">=3.8"
 files = [
-    {file = "botocore-1.35.17-py3-none-any.whl", hash = "sha256:a93f773ca93139529b5d36730b382dbee63ab4c7f26129aa5c84835255ca999d"},
-    {file = "botocore-1.35.17.tar.gz", hash = "sha256:0d35d03ea647b5d464c7f77bdab6fb23ae5d49752b13cf97ab84444518c7b1bd"},
+    {file = "botocore-1.35.19-py3-none-any.whl", hash = "sha256:c83f7f0cacfe7c19b109b363ebfa8736e570d24922f16ed371681f58ebab44a9"},
+    {file = "botocore-1.35.19.tar.gz", hash = "sha256:42d6d8db7250cbd7899f786f9861e02cab17dc238f64d6acb976098ed9809625"},
 ]
 
 [package.dependencies]
@@ -2429,13 +2429,13 @@ test = ["pytest (>=6)"]
 
 [[package]]
 name = "fastapi"
-version = "0.114.1"
+version = "0.114.2"
 description = "FastAPI framework, high performance, easy to learn, fast to code, ready for production"
 optional = false
 python-versions = ">=3.8"
 files = [
-    {file = "fastapi-0.114.1-py3-none-any.whl", hash = "sha256:5d4746f6e4b7dff0b4f6b6c6d5445645285f662fe75886e99af7ee2d6b58bb3e"},
-    {file = "fastapi-0.114.1.tar.gz", hash = "sha256:1d7bbbeabbaae0acb0c22f0ab0b040f642d3093ca3645f8c876b6f91391861d8"},
+    {file = "fastapi-0.114.2-py3-none-any.whl", hash = "sha256:44474a22913057b1acb973ab90f4b671ba5200482e7622816d79105dcece1ac5"},
+    {file = "fastapi-0.114.2.tar.gz", hash = "sha256:0adb148b62edb09e8c6eeefa3ea934e8f276dabc038c5a82989ea6346050c3da"},
 ]
 
 [package.dependencies]
@@ -3990,15 +3990,18 @@ files = [
 
 [[package]]
 name = "idna"
-version = "3.8"
+version = "3.9"
 description = "Internationalized Domain Names in Applications (IDNA)"
 optional = false
 python-versions = ">=3.6"
 files = [
-    {file = "idna-3.8-py3-none-any.whl", hash = "sha256:050b4e5baadcd44d760cedbd2b8e639f2ff89bbc7a5730fcc662954303377aac"},
-    {file = "idna-3.8.tar.gz", hash = "sha256:d838c2c0ed6fced7693d5e8ab8e734d5f8fda53a039c0164afb0b82e771e3603"},
+    {file = "idna-3.9-py3-none-any.whl", hash = "sha256:69297d5da0cc9281c77efffb4e730254dd45943f45bbfb461de5991713989b1e"},
+    {file = "idna-3.9.tar.gz", hash = "sha256:e5c5dafde284f26e9e0f28f6ea2d6400abd5ca099864a67f576f3981c6476124"},
 ]
 
+[package.extras]
+all = ["flake8 (>=7.1.1)", "mypy (>=1.11.2)", "pytest (>=8.3.2)", "ruff (>=0.6.2)"]
+
 [[package]]
 name = "importlib-metadata"
 version = "6.11.0"
@@ -4393,13 +4396,13 @@ six = "*"
 
 [[package]]
 name = "langfuse"
-version = "2.48.0"
+version = "2.48.1"
 description = "A client library for accessing langfuse"
 optional = false
 python-versions = "<4.0,>=3.8.1"
 files = [
-    {file = "langfuse-2.48.0-py3-none-any.whl", hash = "sha256:475b047e461f8a45e3c7d81b6a87e0b9e389c489d465b838aa69cbdd16eeacce"},
-    {file = "langfuse-2.48.0.tar.gz", hash = "sha256:46e7e6e6e97fe03115a9f95d7f29b3fcd1848a9d1bb34608ebb42a3931919e45"},
+    {file = "langfuse-2.48.1-py3-none-any.whl", hash = "sha256:8661070b6d94ba1d7da92c054f3110b6ecf4489d6e8204a4080f934f3f49ebf2"},
+    {file = "langfuse-2.48.1.tar.gz", hash = "sha256:b8117d90babec6be1bc3303b42e0b71848531eae44118e6e0123d03e7961d0fc"},
 ]
 
 [package.dependencies]
@@ -4418,13 +4421,13 @@ openai = ["openai (>=0.27.8)"]
 
 [[package]]
 name = "langsmith"
-version = "0.1.118"
+version = "0.1.120"
 description = "Client library to connect to the LangSmith LLM Tracing and Evaluation Platform."
 optional = false
 python-versions = "<4.0,>=3.8.1"
 files = [
-    {file = "langsmith-0.1.118-py3-none-any.whl", hash = "sha256:f017127b3efb037da5e46ff4f8583e8192e7955191737240c327f3eadc144d7c"},
-    {file = "langsmith-0.1.118.tar.gz", hash = "sha256:ff1ca06c92c6081250244ebbce5d0bb347b9d898d2e9b60a13b11f0f0720f09f"},
+    {file = "langsmith-0.1.120-py3-none-any.whl", hash = "sha256:54d2785e301646c0988e0a69ebe4d976488c87b41928b358cb153b6ddd8db62b"},
+    {file = "langsmith-0.1.120.tar.gz", hash = "sha256:25499ca187b41bd89d784b272b97a8d76f60e0e21bdf20336e8a2aa6a9b23ac9"},
 ]
 
 [package.dependencies]
@@ -6232,13 +6235,13 @@ xmp = ["defusedxml"]
 
 [[package]]
 name = "platformdirs"
-version = "4.3.2"
+version = "4.3.3"
 description = "A small Python package for determining appropriate platform-specific dirs, e.g. a `user data dir`."
 optional = false
 python-versions = ">=3.8"
 files = [
-    {file = "platformdirs-4.3.2-py3-none-any.whl", hash = "sha256:eb1c8582560b34ed4ba105009a4badf7f6f85768b30126f351328507b2beb617"},
-    {file = "platformdirs-4.3.2.tar.gz", hash = "sha256:9e5e27a08aa095dd127b9f2e764d74254f482fef22b0970773bfba79d091ab8c"},
+    {file = "platformdirs-4.3.3-py3-none-any.whl", hash = "sha256:50a5450e2e84f44539718293cbb1da0a0885c9d14adf21b77bae4e66fc99d9b5"},
+    {file = "platformdirs-4.3.3.tar.gz", hash = "sha256:d4e0b7d8ec176b341fb03cb11ca12d0276faa8c485f9cd218f613840463fc2c0"},
 ]
 
 [package.extras]
@@ -6248,13 +6251,13 @@ type = ["mypy (>=1.11.2)"]
 
 [[package]]
 name = "plotly"
-version = "5.24.0"
+version = "5.24.1"
 description = "An open-source, interactive data visualization library for Python"
 optional = false
 python-versions = ">=3.8"
 files = [
-    {file = "plotly-5.24.0-py3-none-any.whl", hash = "sha256:0e54efe52c8cef899f7daa41be9ed97dfb6be622613a2a8f56a86a0634b2b67e"},
-    {file = "plotly-5.24.0.tar.gz", hash = "sha256:eae9f4f54448682442c92c1e97148e3ad0c52f0cf86306e1b76daba24add554a"},
+    {file = "plotly-5.24.1-py3-none-any.whl", hash = "sha256:f67073a1e637eb0dc3e46324d9d51e2fe76e9727c892dde64ddf1e1b51f29089"},
+    {file = "plotly-5.24.1.tar.gz", hash = "sha256:dbc8ac8339d248a4bcc36e08a5659bacfe1b079390b8953533f4eb22169b4bae"},
 ]
 
 [package.dependencies]
@@ -7025,15 +7028,18 @@ files = [
 
 [[package]]
 name = "pyreadline3"
-version = "3.4.3"
+version = "3.5.2"
 description = "A python implementation of GNU readline."
 optional = false
-python-versions = "*"
+python-versions = ">=3.8"
 files = [
-    {file = "pyreadline3-3.4.3-py3-none-any.whl", hash = "sha256:f832c5898f4f9a0f81d48a8c499b39d0179de1a465ea3def1a7e7231840b4ed6"},
-    {file = "pyreadline3-3.4.3.tar.gz", hash = "sha256:ebab0baca37f50e2faa1dd99a6da1c75de60e0d68a3b229c134bbd12786250e2"},
+    {file = "pyreadline3-3.5.2-py3-none-any.whl", hash = "sha256:a87d56791e2965b2b187e2ea33dcf664600842c997c0623c95cf8ef07db83de9"},
+    {file = "pyreadline3-3.5.2.tar.gz", hash = "sha256:ba82292e52c5a3bb256b291af0c40b457c1e8699cac9a873abbcaac8aef3a1bb"},
 ]
 
+[package.extras]
+dev = ["build", "flake8", "pytest", "twine"]
+
 [[package]]
 name = "pytest"
 version = "8.3.3"
@@ -8778,13 +8784,13 @@ test = ["pytest", "tornado (>=4.5)", "typeguard"]
 
 [[package]]
 name = "tencentcloud-sdk-python-common"
-version = "3.0.1230"
+version = "3.0.1231"
 description = "Tencent Cloud Common SDK for Python"
 optional = false
 python-versions = "*"
 files = [
-    {file = "tencentcloud-sdk-python-common-3.0.1230.tar.gz", hash = "sha256:1e0f3bab80026fcb0083820869239b3f8cf30beb8e00e12c213bdecc75eb7577"},
-    {file = "tencentcloud_sdk_python_common-3.0.1230-py2.py3-none-any.whl", hash = "sha256:03616c79685c154c689536a9c823d52b855cf49eada70679826a92aff5afd596"},
+    {file = "tencentcloud-sdk-python-common-3.0.1231.tar.gz", hash = "sha256:22aa281ca2eac511e1615b2953da7c4a0bec87cf93a05a7a15dbb61b23a215ee"},
+    {file = "tencentcloud_sdk_python_common-3.0.1231-py2.py3-none-any.whl", hash = "sha256:bd0f7c4df4b156ec35c8731afa1f498043c7e1cd5d2feb595ee441fdb45a061e"},
 ]
 
 [package.dependencies]
@@ -8792,17 +8798,17 @@ requests = ">=2.16.0"
 
 [[package]]
 name = "tencentcloud-sdk-python-hunyuan"
-version = "3.0.1230"
+version = "3.0.1231"
 description = "Tencent Cloud Hunyuan SDK for Python"
 optional = false
 python-versions = "*"
 files = [
-    {file = "tencentcloud-sdk-python-hunyuan-3.0.1230.tar.gz", hash = "sha256:900d15cb9dc2217b1282d985898ec7ecf97859351c86c6f7efc74685f08a5f85"},
-    {file = "tencentcloud_sdk_python_hunyuan-3.0.1230-py2.py3-none-any.whl", hash = "sha256:604dab0d4d66ea942f23d7980c76b5f0f6af3d68a8374e619331a4dd2910991e"},
+    {file = "tencentcloud-sdk-python-hunyuan-3.0.1231.tar.gz", hash = "sha256:6da12f418f14305b3a6b7bb29b6d95bf4038a6b66b81c0e03b8dafc4f6df99ca"},
+    {file = "tencentcloud_sdk_python_hunyuan-3.0.1231-py2.py3-none-any.whl", hash = "sha256:21ba28f69c34c15e20900be3f2c06376fcaf7e58265f939833c55631f2348792"},
 ]
 
 [package.dependencies]
-tencentcloud-sdk-python-common = "3.0.1230"
+tencentcloud-sdk-python-common = "3.0.1231"
 
 [[package]]
 name = "threadpoolctl"
@@ -9205,13 +9211,13 @@ typing-extensions = ">=3.7.4.3"
 
 [[package]]
 name = "types-requests"
-version = "2.32.0.20240907"
+version = "2.32.0.20240914"
 description = "Typing stubs for requests"
 optional = false
 python-versions = ">=3.8"
 files = [
-    {file = "types-requests-2.32.0.20240907.tar.gz", hash = "sha256:ff33935f061b5e81ec87997e91050f7b4af4f82027a7a7a9d9aaea04a963fdf8"},
-    {file = "types_requests-2.32.0.20240907-py3-none-any.whl", hash = "sha256:1d1e79faeaf9d42def77f3c304893dea17a97cae98168ac69f3cb465516ee8da"},
+    {file = "types-requests-2.32.0.20240914.tar.gz", hash = "sha256:2850e178db3919d9bf809e434eef65ba49d0e7e33ac92d588f4a5e295fffd405"},
+    {file = "types_requests-2.32.0.20240914-py3-none-any.whl", hash = "sha256:59c2f673eb55f32a99b2894faf6020e1a9f4a402ad0f192bfee0b64469054310"},
 ]
 
 [package.dependencies]
@@ -9454,13 +9460,13 @@ files = [
 
 [[package]]
 name = "urllib3"
-version = "2.2.2"
+version = "2.2.3"
 description = "HTTP library with thread-safe connection pooling, file post, and more."
 optional = false
 python-versions = ">=3.8"
 files = [
-    {file = "urllib3-2.2.2-py3-none-any.whl", hash = "sha256:a448b2f64d686155468037e1ace9f2d2199776e17f0a46610480d311f73e3472"},
-    {file = "urllib3-2.2.2.tar.gz", hash = "sha256:dd505485549a7a552833da5e6063639d0d177c04f23bc3864e41e5dc5f612168"},
+    {file = "urllib3-2.2.3-py3-none-any.whl", hash = "sha256:ca899ca043dcb1bafa3e262d73aa25c465bfb49e0bd9dd5d59f1d0acba2f8fac"},
+    {file = "urllib3-2.2.3.tar.gz", hash = "sha256:e7d814a81dad81e6caf2ec9fdedb284ecc9c73076b62654547cc64ccdcae26e9"},
 ]
 
 [package.extras]
@@ -9614,12 +9620,12 @@ files = [
 
 [[package]]
 name = "volcengine-python-sdk"
-version = "1.0.100"
+version = "1.0.101"
 description = "Volcengine SDK for Python"
 optional = false
 python-versions = "*"
 files = [
-    {file = "volcengine-python-sdk-1.0.100.tar.gz", hash = "sha256:cdc194fe3ce51adda6892d2ca1c43edba3300699321dc6c69119c59fc3b28932"},
+    {file = "volcengine-python-sdk-1.0.101.tar.gz", hash = "sha256:1b76e71a6dcf3d5be1b2c058e7d281359e6cca2cc920ffe2567d3115beea1d02"},
 ]
 
 [package.dependencies]
@@ -10215,13 +10221,13 @@ requests = "*"
 
 [[package]]
 name = "zipp"
-version = "3.20.1"
+version = "3.20.2"
 description = "Backport of pathlib-compatible object wrapper for zip files"
 optional = false
 python-versions = ">=3.8"
 files = [
-    {file = "zipp-3.20.1-py3-none-any.whl", hash = "sha256:9960cd8967c8f85a56f920d5d507274e74f9ff813a0ab8889a5b5be2daf44064"},
-    {file = "zipp-3.20.1.tar.gz", hash = "sha256:c22b14cc4763c5a5b04134207736c107db42e9d3ef2d9779d465f5f1bcba572b"},
+    {file = "zipp-3.20.2-py3-none-any.whl", hash = "sha256:a817ac80d6cf4b23bf7f2828b7cabf326f15a001bea8b1f9b49631780ba28350"},
+    {file = "zipp-3.20.2.tar.gz", hash = "sha256:bc9eb26f4506fda01b81bcde0ca78103b6e62f991b381fec825435c836edbc29"},
 ]
 
 [package.extras]
@@ -10416,4 +10422,4 @@ cffi = ["cffi (>=1.11)"]
 [metadata]
 lock-version = "2.0"
 python-versions = ">=3.10,<3.13"
-content-hash = "9173a56b2efea12804c980511e1465fba43c7a3d83b1ad284ee149851ed67fc5"
+content-hash = "18924ae12a00bde4438a46168bc167ed69613ab1ab0c387f193cd47ac24379b2"
diff --git a/api/pyproject.toml b/api/pyproject.toml
index 8c10f1dad9795..829c39a12bf91 100644
--- a/api/pyproject.toml
+++ b/api/pyproject.toml
@@ -168,7 +168,7 @@ gunicorn = "~22.0.0"
 httpx = { version = "~0.27.0", extras = ["socks"] }
 huggingface-hub = "~0.16.4"
 jieba = "0.42.1"
-langfuse = "^2.36.1"
+langfuse = "^2.48.0"
 langsmith = "^0.1.77"
 mailchimp-transactional = "~1.0.50"
 markdown = "~3.5.1"

From 8c51d06222d5ada18bab11c4777b7e7feb644fbd Mon Sep 17 00:00:00 2001
From: Hash Brown <hi@xzd.me>
Date: Sun, 22 Sep 2024 03:15:11 +0800
Subject: [PATCH 012/235] feat: regenerate in `Chat`, `agent` and `Chatflow`
 app (#7661)

---
 api/constants/__init__.py                     |   1 +
 api/controllers/console/app/completion.py     |   1 +
 api/controllers/console/app/message.py        |   2 -
 api/controllers/console/app/workflow.py       |   2 +
 api/controllers/console/explore/completion.py |   1 +
 api/controllers/console/explore/message.py    |   2 +-
 api/controllers/service_api/app/message.py    |   1 +
 api/controllers/web/completion.py             |   1 +
 api/controllers/web/message.py                |   3 +-
 api/core/agent/base_agent_runner.py           |   5 +-
 .../app/apps/advanced_chat/app_generator.py   |   1 +
 api/core/app/apps/agent_chat/app_generator.py |   1 +
 api/core/app/apps/chat/app_generator.py       |   1 +
 .../app/apps/message_based_app_generator.py   |   1 +
 api/core/app/entities/app_invoke_entities.py  |   3 +
 api/core/memory/token_buffer_memory.py        |  21 ++-
 .../prompt/utils/extract_thread_messages.py   |  22 +++
 api/fields/conversation_fields.py             |   1 +
 api/fields/message_fields.py                  |   1 +
 ...bb251_add_parent_message_id_to_messages.py |  36 +++++
 api/models/model.py                           |   1 +
 api/services/message_service.py               |   4 +-
 .../prompt/test_extract_thread_messages.py    |  91 +++++++++++
 .../debug-with-multiple-model/chat-item.tsx   |   4 +-
 .../debug/debug-with-single-model/index.tsx   |  26 ++-
 web/app/components/app/log/list.tsx           | 149 ++++++++++--------
 .../chat/chat-with-history/chat-wrapper.tsx   |  25 ++-
 .../base/chat/chat-with-history/hooks.tsx     |  35 +---
 .../base/chat/chat/answer/index.tsx           |   3 +
 .../base/chat/chat/answer/operation.tsx       |  13 +-
 web/app/components/base/chat/chat/context.tsx |   3 +
 web/app/components/base/chat/chat/hooks.ts    |   3 +-
 web/app/components/base/chat/chat/index.tsx   |   5 +
 web/app/components/base/chat/chat/type.ts     |   1 +
 web/app/components/base/chat/constants.ts     |   1 +
 .../chat/embedded-chatbot/chat-wrapper.tsx    |  25 ++-
 .../base/chat/embedded-chatbot/hooks.tsx      |  38 +----
 web/app/components/base/chat/types.ts         |   4 +-
 web/app/components/base/chat/utils.ts         |  57 ++++++-
 .../assets/vender/line/general/refresh.svg    |   1 +
 .../src/vender/line/general/Refresh.json      |  23 +++
 .../icons/src/vender/line/general/Refresh.tsx |  16 ++
 .../icons/src/vender/line/general/index.ts    |   1 +
 .../components/base/regenerate-btn/index.tsx  |  31 ++++
 .../workflow/panel/chat-record/index.tsx      |  79 ++++++----
 .../panel/debug-and-preview/chat-wrapper.tsx  |  26 ++-
 .../workflow/panel/debug-and-preview/hooks.ts |   2 +
 web/hooks/use-app-favicon.ts                  |   6 +-
 web/i18n/en-US/app-api.ts                     |   1 +
 web/i18n/zh-Hans/app-api.ts                   |   1 +
 web/models/log.ts                             |   1 +
 51 files changed, 604 insertions(+), 179 deletions(-)
 create mode 100644 api/core/prompt/utils/extract_thread_messages.py
 create mode 100644 api/migrations/versions/2024_09_11_1012-d57ba9ebb251_add_parent_message_id_to_messages.py
 create mode 100644 api/tests/unit_tests/core/prompt/test_extract_thread_messages.py
 create mode 100644 web/app/components/base/icons/assets/vender/line/general/refresh.svg
 create mode 100644 web/app/components/base/icons/src/vender/line/general/Refresh.json
 create mode 100644 web/app/components/base/icons/src/vender/line/general/Refresh.tsx
 create mode 100644 web/app/components/base/regenerate-btn/index.tsx

diff --git a/api/constants/__init__.py b/api/constants/__init__.py
index e22c3268ef428..75eaf81638cdf 100644
--- a/api/constants/__init__.py
+++ b/api/constants/__init__.py
@@ -1 +1,2 @@
 HIDDEN_VALUE = "[__HIDDEN__]"
+UUID_NIL = "00000000-0000-0000-0000-000000000000"
diff --git a/api/controllers/console/app/completion.py b/api/controllers/console/app/completion.py
index 53de51c24d798..d3296d3dff44a 100644
--- a/api/controllers/console/app/completion.py
+++ b/api/controllers/console/app/completion.py
@@ -109,6 +109,7 @@ def post(self, app_model):
         parser.add_argument("files", type=list, required=False, location="json")
         parser.add_argument("model_config", type=dict, required=True, location="json")
         parser.add_argument("conversation_id", type=uuid_value, location="json")
+        parser.add_argument("parent_message_id", type=uuid_value, required=False, location="json")
         parser.add_argument("response_mode", type=str, choices=["blocking", "streaming"], location="json")
         parser.add_argument("retriever_from", type=str, required=False, default="dev", location="json")
         args = parser.parse_args()
diff --git a/api/controllers/console/app/message.py b/api/controllers/console/app/message.py
index fe06201982374..2fba3e0af02be 100644
--- a/api/controllers/console/app/message.py
+++ b/api/controllers/console/app/message.py
@@ -105,8 +105,6 @@ def get(self, app_model):
             if rest_count > 0:
                 has_more = True
 
-        history_messages = list(reversed(history_messages))
-
         return InfiniteScrollPagination(data=history_messages, limit=args["limit"], has_more=has_more)
 
 
diff --git a/api/controllers/console/app/workflow.py b/api/controllers/console/app/workflow.py
index b488deb89d90c..0a693b84e27ec 100644
--- a/api/controllers/console/app/workflow.py
+++ b/api/controllers/console/app/workflow.py
@@ -166,6 +166,8 @@ def post(self, app_model: App):
         parser.add_argument("query", type=str, required=True, location="json", default="")
         parser.add_argument("files", type=list, location="json")
         parser.add_argument("conversation_id", type=uuid_value, location="json")
+        parser.add_argument("parent_message_id", type=uuid_value, required=False, location="json")
+
         args = parser.parse_args()
 
         try:
diff --git a/api/controllers/console/explore/completion.py b/api/controllers/console/explore/completion.py
index f4646920982bd..125bc1af8c41e 100644
--- a/api/controllers/console/explore/completion.py
+++ b/api/controllers/console/explore/completion.py
@@ -100,6 +100,7 @@ def post(self, installed_app):
         parser.add_argument("query", type=str, required=True, location="json")
         parser.add_argument("files", type=list, required=False, location="json")
         parser.add_argument("conversation_id", type=uuid_value, location="json")
+        parser.add_argument("parent_message_id", type=uuid_value, required=False, location="json")
         parser.add_argument("retriever_from", type=str, required=False, default="explore_app", location="json")
         args = parser.parse_args()
 
diff --git a/api/controllers/console/explore/message.py b/api/controllers/console/explore/message.py
index 0e0238556cf9a..3d221ff30a659 100644
--- a/api/controllers/console/explore/message.py
+++ b/api/controllers/console/explore/message.py
@@ -51,7 +51,7 @@ def get(self, installed_app):
 
         try:
             return MessageService.pagination_by_first_id(
-                app_model, current_user, args["conversation_id"], args["first_id"], args["limit"]
+                app_model, current_user, args["conversation_id"], args["first_id"], args["limit"], "desc"
             )
         except services.errors.conversation.ConversationNotExistsError:
             raise NotFound("Conversation Not Exists.")
diff --git a/api/controllers/service_api/app/message.py b/api/controllers/service_api/app/message.py
index e54e6f4903d57..a70ee89b5e2bf 100644
--- a/api/controllers/service_api/app/message.py
+++ b/api/controllers/service_api/app/message.py
@@ -54,6 +54,7 @@ class MessageListApi(Resource):
     message_fields = {
         "id": fields.String,
         "conversation_id": fields.String,
+        "parent_message_id": fields.String,
         "inputs": fields.Raw,
         "query": fields.String,
         "answer": fields.String(attribute="re_sign_file_url_answer"),
diff --git a/api/controllers/web/completion.py b/api/controllers/web/completion.py
index 115492b7966c0..45b890dfc4899 100644
--- a/api/controllers/web/completion.py
+++ b/api/controllers/web/completion.py
@@ -96,6 +96,7 @@ def post(self, app_model, end_user):
         parser.add_argument("files", type=list, required=False, location="json")
         parser.add_argument("response_mode", type=str, choices=["blocking", "streaming"], location="json")
         parser.add_argument("conversation_id", type=uuid_value, location="json")
+        parser.add_argument("parent_message_id", type=uuid_value, required=False, location="json")
         parser.add_argument("retriever_from", type=str, required=False, default="web_app", location="json")
 
         args = parser.parse_args()
diff --git a/api/controllers/web/message.py b/api/controllers/web/message.py
index 0d4047f4efbaf..2d2a5866c8038 100644
--- a/api/controllers/web/message.py
+++ b/api/controllers/web/message.py
@@ -57,6 +57,7 @@ class MessageListApi(WebApiResource):
     message_fields = {
         "id": fields.String,
         "conversation_id": fields.String,
+        "parent_message_id": fields.String,
         "inputs": fields.Raw,
         "query": fields.String,
         "answer": fields.String(attribute="re_sign_file_url_answer"),
@@ -89,7 +90,7 @@ def get(self, app_model, end_user):
 
         try:
             return MessageService.pagination_by_first_id(
-                app_model, end_user, args["conversation_id"], args["first_id"], args["limit"]
+                app_model, end_user, args["conversation_id"], args["first_id"], args["limit"], "desc"
             )
         except services.errors.conversation.ConversationNotExistsError:
             raise NotFound("Conversation Not Exists.")
diff --git a/api/core/agent/base_agent_runner.py b/api/core/agent/base_agent_runner.py
index d09a9956a4a59..5295f97bdbb86 100644
--- a/api/core/agent/base_agent_runner.py
+++ b/api/core/agent/base_agent_runner.py
@@ -32,6 +32,7 @@
 from core.model_runtime.entities.model_entities import ModelFeature
 from core.model_runtime.model_providers.__base.large_language_model import LargeLanguageModel
 from core.model_runtime.utils.encoders import jsonable_encoder
+from core.prompt.utils.extract_thread_messages import extract_thread_messages
 from core.tools.entities.tool_entities import (
     ToolParameter,
     ToolRuntimeVariablePool,
@@ -441,10 +442,12 @@ def organize_agent_history(self, prompt_messages: list[PromptMessage]) -> list[P
             .filter(
                 Message.conversation_id == self.message.conversation_id,
             )
-            .order_by(Message.created_at.asc())
+            .order_by(Message.created_at.desc())
             .all()
         )
 
+        messages = list(reversed(extract_thread_messages(messages)))
+
         for message in messages:
             if message.id == self.message.id:
                 continue
diff --git a/api/core/app/apps/advanced_chat/app_generator.py b/api/core/app/apps/advanced_chat/app_generator.py
index 88e1256ed54e4..445ef6d0ab19f 100644
--- a/api/core/app/apps/advanced_chat/app_generator.py
+++ b/api/core/app/apps/advanced_chat/app_generator.py
@@ -121,6 +121,7 @@ def generate(
             inputs=conversation.inputs if conversation else self._get_cleaned_inputs(inputs, app_config),
             query=query,
             files=file_objs,
+            parent_message_id=args.get("parent_message_id"),
             user_id=user.id,
             stream=stream,
             invoke_from=invoke_from,
diff --git a/api/core/app/apps/agent_chat/app_generator.py b/api/core/app/apps/agent_chat/app_generator.py
index abf8a332ab27d..99abccf4f9840 100644
--- a/api/core/app/apps/agent_chat/app_generator.py
+++ b/api/core/app/apps/agent_chat/app_generator.py
@@ -127,6 +127,7 @@ def generate(
             inputs=conversation.inputs if conversation else self._get_cleaned_inputs(inputs, app_config),
             query=query,
             files=file_objs,
+            parent_message_id=args.get("parent_message_id"),
             user_id=user.id,
             stream=stream,
             invoke_from=invoke_from,
diff --git a/api/core/app/apps/chat/app_generator.py b/api/core/app/apps/chat/app_generator.py
index 032556ec4c1ce..9ef1366a0f668 100644
--- a/api/core/app/apps/chat/app_generator.py
+++ b/api/core/app/apps/chat/app_generator.py
@@ -128,6 +128,7 @@ def generate(
             inputs=conversation.inputs if conversation else self._get_cleaned_inputs(inputs, app_config),
             query=query,
             files=file_objs,
+            parent_message_id=args.get("parent_message_id"),
             user_id=user.id,
             stream=stream,
             invoke_from=invoke_from,
diff --git a/api/core/app/apps/message_based_app_generator.py b/api/core/app/apps/message_based_app_generator.py
index c4db95cbd0c4a..65b759acf5376 100644
--- a/api/core/app/apps/message_based_app_generator.py
+++ b/api/core/app/apps/message_based_app_generator.py
@@ -218,6 +218,7 @@ def _init_generate_records(
             answer_tokens=0,
             answer_unit_price=0,
             answer_price_unit=0,
+            parent_message_id=getattr(application_generate_entity, "parent_message_id", None),
             provider_response_latency=0,
             total_price=0,
             currency="USD",
diff --git a/api/core/app/entities/app_invoke_entities.py b/api/core/app/entities/app_invoke_entities.py
index ab8d4e374e26f..87ca51ef1b1db 100644
--- a/api/core/app/entities/app_invoke_entities.py
+++ b/api/core/app/entities/app_invoke_entities.py
@@ -122,6 +122,7 @@ class ChatAppGenerateEntity(EasyUIBasedAppGenerateEntity):
     """
 
     conversation_id: Optional[str] = None
+    parent_message_id: Optional[str] = None
 
 
 class CompletionAppGenerateEntity(EasyUIBasedAppGenerateEntity):
@@ -138,6 +139,7 @@ class AgentChatAppGenerateEntity(EasyUIBasedAppGenerateEntity):
     """
 
     conversation_id: Optional[str] = None
+    parent_message_id: Optional[str] = None
 
 
 class AdvancedChatAppGenerateEntity(AppGenerateEntity):
@@ -149,6 +151,7 @@ class AdvancedChatAppGenerateEntity(AppGenerateEntity):
     app_config: WorkflowUIBasedAppConfig
 
     conversation_id: Optional[str] = None
+    parent_message_id: Optional[str] = None
     query: str
 
     class SingleIterationRunEntity(BaseModel):
diff --git a/api/core/memory/token_buffer_memory.py b/api/core/memory/token_buffer_memory.py
index d3185c3b11aec..60b36c50f00d3 100644
--- a/api/core/memory/token_buffer_memory.py
+++ b/api/core/memory/token_buffer_memory.py
@@ -11,6 +11,7 @@
     TextPromptMessageContent,
     UserPromptMessage,
 )
+from core.prompt.utils.extract_thread_messages import extract_thread_messages
 from extensions.ext_database import db
 from models.model import AppMode, Conversation, Message, MessageFile
 from models.workflow import WorkflowRun
@@ -33,8 +34,17 @@ def get_history_prompt_messages(
 
         # fetch limited messages, and return reversed
         query = (
-            db.session.query(Message.id, Message.query, Message.answer, Message.created_at, Message.workflow_run_id)
-            .filter(Message.conversation_id == self.conversation.id, Message.answer != "")
+            db.session.query(
+                Message.id,
+                Message.query,
+                Message.answer,
+                Message.created_at,
+                Message.workflow_run_id,
+                Message.parent_message_id,
+            )
+            .filter(
+                Message.conversation_id == self.conversation.id,
+            )
             .order_by(Message.created_at.desc())
         )
 
@@ -45,7 +55,12 @@ def get_history_prompt_messages(
 
         messages = query.limit(message_limit).all()
 
-        messages = list(reversed(messages))
+        # instead of all messages from the conversation, we only need to extract messages
+        # that belong to the thread of last message
+        thread_messages = extract_thread_messages(messages)
+        thread_messages.pop(0)
+        messages = list(reversed(thread_messages))
+
         message_file_parser = MessageFileParser(tenant_id=app_record.tenant_id, app_id=app_record.id)
         prompt_messages = []
         for message in messages:
diff --git a/api/core/prompt/utils/extract_thread_messages.py b/api/core/prompt/utils/extract_thread_messages.py
new file mode 100644
index 0000000000000..e8b626499fce1
--- /dev/null
+++ b/api/core/prompt/utils/extract_thread_messages.py
@@ -0,0 +1,22 @@
+from constants import UUID_NIL
+
+
+def extract_thread_messages(messages: list[dict]) -> list[dict]:
+    thread_messages = []
+    next_message = None
+
+    for message in messages:
+        if not message.parent_message_id:
+            # If the message is regenerated and does not have a parent message, it is the start of a new thread
+            thread_messages.append(message)
+            break
+
+        if not next_message:
+            thread_messages.append(message)
+            next_message = message.parent_message_id
+        else:
+            if next_message in {message.id, UUID_NIL}:
+                thread_messages.append(message)
+                next_message = message.parent_message_id
+
+    return thread_messages
diff --git a/api/fields/conversation_fields.py b/api/fields/conversation_fields.py
index 9207314fc2132..3dcd88d1de31d 100644
--- a/api/fields/conversation_fields.py
+++ b/api/fields/conversation_fields.py
@@ -75,6 +75,7 @@ def format(self, value):
     "metadata": fields.Raw(attribute="message_metadata_dict"),
     "status": fields.String,
     "error": fields.String,
+    "parent_message_id": fields.String,
 }
 
 feedback_stat_fields = {"like": fields.Integer, "dislike": fields.Integer}
diff --git a/api/fields/message_fields.py b/api/fields/message_fields.py
index 3d2df87afb9b1..c938097131f7a 100644
--- a/api/fields/message_fields.py
+++ b/api/fields/message_fields.py
@@ -62,6 +62,7 @@
 message_fields = {
     "id": fields.String,
     "conversation_id": fields.String,
+    "parent_message_id": fields.String,
     "inputs": fields.Raw,
     "query": fields.String,
     "answer": fields.String(attribute="re_sign_file_url_answer"),
diff --git a/api/migrations/versions/2024_09_11_1012-d57ba9ebb251_add_parent_message_id_to_messages.py b/api/migrations/versions/2024_09_11_1012-d57ba9ebb251_add_parent_message_id_to_messages.py
new file mode 100644
index 0000000000000..fd957eeafb2b6
--- /dev/null
+++ b/api/migrations/versions/2024_09_11_1012-d57ba9ebb251_add_parent_message_id_to_messages.py
@@ -0,0 +1,36 @@
+"""add parent_message_id to messages
+
+Revision ID: d57ba9ebb251
+Revises: 675b5321501b
+Create Date: 2024-09-11 10:12:45.826265
+
+"""
+import sqlalchemy as sa
+from alembic import op
+
+import models as models
+
+# revision identifiers, used by Alembic.
+revision = 'd57ba9ebb251'
+down_revision = '675b5321501b'
+branch_labels = None
+depends_on = None
+
+
+def upgrade():
+    # ### commands auto generated by Alembic - please adjust! ###
+    with op.batch_alter_table('messages', schema=None) as batch_op:
+        batch_op.add_column(sa.Column('parent_message_id', models.types.StringUUID(), nullable=True))
+
+    # Set parent_message_id for existing messages to uuid_nil() to distinguish them from new messages with actual parent IDs or NULLs
+    op.execute('UPDATE messages SET parent_message_id = uuid_nil() WHERE parent_message_id IS NULL')
+
+    # ### end Alembic commands ###
+
+
+def downgrade():
+    # ### commands auto generated by Alembic - please adjust! ###
+    with op.batch_alter_table('messages', schema=None) as batch_op:
+        batch_op.drop_column('parent_message_id')
+
+    # ### end Alembic commands ###
diff --git a/api/models/model.py b/api/models/model.py
index ae0bc3210b646..53940a5a16cc8 100644
--- a/api/models/model.py
+++ b/api/models/model.py
@@ -710,6 +710,7 @@ class Message(db.Model):
     answer_tokens = db.Column(db.Integer, nullable=False, server_default=db.text("0"))
     answer_unit_price = db.Column(db.Numeric(10, 4), nullable=False)
     answer_price_unit = db.Column(db.Numeric(10, 7), nullable=False, server_default=db.text("0.001"))
+    parent_message_id = db.Column(StringUUID, nullable=True)
     provider_response_latency = db.Column(db.Float, nullable=False, server_default=db.text("0"))
     total_price = db.Column(db.Numeric(10, 7))
     currency = db.Column(db.String(255), nullable=False)
diff --git a/api/services/message_service.py b/api/services/message_service.py
index ecb121c36e4cc..f432a77c80e51 100644
--- a/api/services/message_service.py
+++ b/api/services/message_service.py
@@ -34,6 +34,7 @@ def pagination_by_first_id(
         conversation_id: str,
         first_id: Optional[str],
         limit: int,
+        order: str = "asc",
     ) -> InfiniteScrollPagination:
         if not user:
             return InfiniteScrollPagination(data=[], limit=limit, has_more=False)
@@ -91,7 +92,8 @@ def pagination_by_first_id(
             if rest_count > 0:
                 has_more = True
 
-        history_messages = list(reversed(history_messages))
+        if order == "asc":
+            history_messages = list(reversed(history_messages))
 
         return InfiniteScrollPagination(data=history_messages, limit=limit, has_more=has_more)
 
diff --git a/api/tests/unit_tests/core/prompt/test_extract_thread_messages.py b/api/tests/unit_tests/core/prompt/test_extract_thread_messages.py
new file mode 100644
index 0000000000000..ba3c1eb5e032a
--- /dev/null
+++ b/api/tests/unit_tests/core/prompt/test_extract_thread_messages.py
@@ -0,0 +1,91 @@
+from uuid import uuid4
+
+from constants import UUID_NIL
+from core.prompt.utils.extract_thread_messages import extract_thread_messages
+
+
+class TestMessage:
+    def __init__(self, id, parent_message_id):
+        self.id = id
+        self.parent_message_id = parent_message_id
+
+    def __getitem__(self, item):
+        return getattr(self, item)
+
+
+def test_extract_thread_messages_single_message():
+    messages = [TestMessage(str(uuid4()), UUID_NIL)]
+    result = extract_thread_messages(messages)
+    assert len(result) == 1
+    assert result[0] == messages[0]
+
+
+def test_extract_thread_messages_linear_thread():
+    id1, id2, id3, id4, id5 = str(uuid4()), str(uuid4()), str(uuid4()), str(uuid4()), str(uuid4())
+    messages = [
+        TestMessage(id5, id4),
+        TestMessage(id4, id3),
+        TestMessage(id3, id2),
+        TestMessage(id2, id1),
+        TestMessage(id1, UUID_NIL),
+    ]
+    result = extract_thread_messages(messages)
+    assert len(result) == 5
+    assert [msg["id"] for msg in result] == [id5, id4, id3, id2, id1]
+
+
+def test_extract_thread_messages_branched_thread():
+    id1, id2, id3, id4 = str(uuid4()), str(uuid4()), str(uuid4()), str(uuid4())
+    messages = [
+        TestMessage(id4, id2),
+        TestMessage(id3, id2),
+        TestMessage(id2, id1),
+        TestMessage(id1, UUID_NIL),
+    ]
+    result = extract_thread_messages(messages)
+    assert len(result) == 3
+    assert [msg["id"] for msg in result] == [id4, id2, id1]
+
+
+def test_extract_thread_messages_empty_list():
+    messages = []
+    result = extract_thread_messages(messages)
+    assert len(result) == 0
+
+
+def test_extract_thread_messages_partially_loaded():
+    id0, id1, id2, id3 = str(uuid4()), str(uuid4()), str(uuid4()), str(uuid4())
+    messages = [
+        TestMessage(id3, id2),
+        TestMessage(id2, id1),
+        TestMessage(id1, id0),
+    ]
+    result = extract_thread_messages(messages)
+    assert len(result) == 3
+    assert [msg["id"] for msg in result] == [id3, id2, id1]
+
+
+def test_extract_thread_messages_legacy_messages():
+    id1, id2, id3 = str(uuid4()), str(uuid4()), str(uuid4())
+    messages = [
+        TestMessage(id3, UUID_NIL),
+        TestMessage(id2, UUID_NIL),
+        TestMessage(id1, UUID_NIL),
+    ]
+    result = extract_thread_messages(messages)
+    assert len(result) == 3
+    assert [msg["id"] for msg in result] == [id3, id2, id1]
+
+
+def test_extract_thread_messages_mixed_with_legacy_messages():
+    id1, id2, id3, id4, id5 = str(uuid4()), str(uuid4()), str(uuid4()), str(uuid4()), str(uuid4())
+    messages = [
+        TestMessage(id5, id4),
+        TestMessage(id4, id2),
+        TestMessage(id3, id2),
+        TestMessage(id2, UUID_NIL),
+        TestMessage(id1, UUID_NIL),
+    ]
+    result = extract_thread_messages(messages)
+    assert len(result) == 4
+    assert [msg["id"] for msg in result] == [id5, id4, id2, id1]
diff --git a/web/app/components/app/configuration/debug/debug-with-multiple-model/chat-item.tsx b/web/app/components/app/configuration/debug/debug-with-multiple-model/chat-item.tsx
index 80dfb5c534ea5..1c70f4fe77a9d 100644
--- a/web/app/components/app/configuration/debug/debug-with-multiple-model/chat-item.tsx
+++ b/web/app/components/app/configuration/debug/debug-with-multiple-model/chat-item.tsx
@@ -46,6 +46,7 @@ const ChatItem: FC<ChatItemProps> = ({
   const config = useConfigFromDebugContext()
   const {
     chatList,
+    chatListRef,
     isResponding,
     handleSend,
     suggestedQuestions,
@@ -80,6 +81,7 @@ const ChatItem: FC<ChatItemProps> = ({
       query: message,
       inputs,
       model_config: configData,
+      parent_message_id: chatListRef.current.at(-1)?.id || null,
     }
 
     if (visionConfig.enabled && files?.length && supportVision)
@@ -93,7 +95,7 @@ const ChatItem: FC<ChatItemProps> = ({
         onGetSuggestedQuestions: (responseItemId, getAbortController) => fetchSuggestedQuestions(appId, responseItemId, getAbortController),
       },
     )
-  }, [appId, config, handleSend, inputs, modelAndParameter, textGenerationModelList, visionConfig.enabled])
+  }, [appId, config, handleSend, inputs, modelAndParameter, textGenerationModelList, visionConfig.enabled, chatListRef])
 
   const { eventEmitter } = useEventEmitterContextContext()
   eventEmitter?.useSubscription((v: any) => {
diff --git a/web/app/components/app/configuration/debug/debug-with-single-model/index.tsx b/web/app/components/app/configuration/debug/debug-with-single-model/index.tsx
index d93ad00659e47..80e7c98a8f0c9 100644
--- a/web/app/components/app/configuration/debug/debug-with-single-model/index.tsx
+++ b/web/app/components/app/configuration/debug/debug-with-single-model/index.tsx
@@ -12,7 +12,7 @@ import {
 import Chat from '@/app/components/base/chat/chat'
 import { useChat } from '@/app/components/base/chat/chat/hooks'
 import { useDebugConfigurationContext } from '@/context/debug-configuration'
-import type { OnSend } from '@/app/components/base/chat/types'
+import type { ChatItem, OnSend } from '@/app/components/base/chat/types'
 import { useProviderContext } from '@/context/provider-context'
 import {
   fetchConversationMessages,
@@ -45,10 +45,12 @@ const DebugWithSingleModel = forwardRef<DebugWithSingleModelRefType, DebugWithSi
   const config = useConfigFromDebugContext()
   const {
     chatList,
+    chatListRef,
     isResponding,
     handleSend,
     suggestedQuestions,
     handleStop,
+    handleUpdateChatList,
     handleRestart,
     handleAnnotationAdded,
     handleAnnotationEdited,
@@ -64,7 +66,7 @@ const DebugWithSingleModel = forwardRef<DebugWithSingleModelRefType, DebugWithSi
   )
   useFormattingChangedSubscription(chatList)
 
-  const doSend: OnSend = useCallback((message, files) => {
+  const doSend: OnSend = useCallback((message, files, last_answer) => {
     if (checkCanSend && !checkCanSend())
       return
     const currentProvider = textGenerationModelList.find(item => item.provider === modelConfig.provider)
@@ -85,6 +87,7 @@ const DebugWithSingleModel = forwardRef<DebugWithSingleModelRefType, DebugWithSi
       query: message,
       inputs,
       model_config: configData,
+      parent_message_id: last_answer?.id || chatListRef.current.at(-1)?.id || null,
     }
 
     if (visionConfig.enabled && files?.length && supportVision)
@@ -98,7 +101,23 @@ const DebugWithSingleModel = forwardRef<DebugWithSingleModelRefType, DebugWithSi
         onGetSuggestedQuestions: (responseItemId, getAbortController) => fetchSuggestedQuestions(appId, responseItemId, getAbortController),
       },
     )
-  }, [appId, checkCanSend, completionParams, config, handleSend, inputs, modelConfig, textGenerationModelList, visionConfig.enabled])
+  }, [chatListRef, appId, checkCanSend, completionParams, config, handleSend, inputs, modelConfig, textGenerationModelList, visionConfig.enabled])
+
+  const doRegenerate = useCallback((chatItem: ChatItem) => {
+    const index = chatList.findIndex(item => item.id === chatItem.id)
+    if (index === -1)
+      return
+
+    const prevMessages = chatList.slice(0, index)
+    const question = prevMessages.pop()
+    const lastAnswer = prevMessages.at(-1)
+
+    if (!question)
+      return
+
+    handleUpdateChatList(prevMessages)
+    doSend(question.content, question.message_files, (!lastAnswer || lastAnswer.isOpeningStatement) ? undefined : lastAnswer)
+  }, [chatList, handleUpdateChatList, doSend])
 
   const allToolIcons = useMemo(() => {
     const icons: Record<string, any> = {}
@@ -123,6 +142,7 @@ const DebugWithSingleModel = forwardRef<DebugWithSingleModelRefType, DebugWithSi
       chatFooterClassName='px-6 pt-10 pb-4'
       suggestedQuestions={suggestedQuestions}
       onSend={doSend}
+      onRegenerate={doRegenerate}
       onStopResponding={handleStop}
       showPromptLog
       questionIcon={<Avatar name={userProfile.name} size={40} />}
diff --git a/web/app/components/app/log/list.tsx b/web/app/components/app/log/list.tsx
index caec10c4f714b..149e877fa4ac2 100644
--- a/web/app/components/app/log/list.tsx
+++ b/web/app/components/app/log/list.tsx
@@ -16,6 +16,7 @@ import timezone from 'dayjs/plugin/timezone'
 import { createContext, useContext } from 'use-context-selector'
 import { useShallow } from 'zustand/react/shallow'
 import { useTranslation } from 'react-i18next'
+import { UUID_NIL } from '../../base/chat/constants'
 import s from './style.module.css'
 import VarPanel from './var-panel'
 import cn from '@/utils/classnames'
@@ -81,72 +82,92 @@ const PARAM_MAP = {
   frequency_penalty: 'Frequency Penalty',
 }
 
-// Format interface data for easy display
-const getFormattedChatList = (messages: ChatMessage[], conversationId: string, timezone: string, format: string) => {
-  const newChatList: IChatItem[] = []
-  messages.forEach((item: ChatMessage) => {
-    newChatList.push({
-      id: `question-${item.id}`,
-      content: item.inputs.query || item.inputs.default_input || item.query, // text generation: item.inputs.query; chat: item.query
-      isAnswer: false,
-      message_files: item.message_files?.filter((file: any) => file.belongs_to === 'user') || [],
-    })
-    newChatList.push({
-      id: item.id,
-      content: item.answer,
-      agent_thoughts: addFileInfos(item.agent_thoughts ? sortAgentSorts(item.agent_thoughts) : item.agent_thoughts, item.message_files),
-      feedback: item.feedbacks.find(item => item.from_source === 'user'), // user feedback
-      adminFeedback: item.feedbacks.find(item => item.from_source === 'admin'), // admin feedback
-      feedbackDisabled: false,
-      isAnswer: true,
-      message_files: item.message_files?.filter((file: any) => file.belongs_to === 'assistant') || [],
-      log: [
-        ...item.message,
-        ...(item.message[item.message.length - 1]?.role !== 'assistant'
-          ? [
-            {
-              role: 'assistant',
-              text: item.answer,
-              files: item.message_files?.filter((file: any) => file.belongs_to === 'assistant') || [],
-            },
-          ]
-          : []),
-      ],
-      workflow_run_id: item.workflow_run_id,
-      conversationId,
-      input: {
-        inputs: item.inputs,
-        query: item.query,
-      },
-      more: {
-        time: dayjs.unix(item.created_at).tz(timezone).format(format),
-        tokens: item.answer_tokens + item.message_tokens,
-        latency: item.provider_response_latency.toFixed(2),
-      },
-      citation: item.metadata?.retriever_resources,
-      annotation: (() => {
-        if (item.annotation_hit_history) {
-          return {
-            id: item.annotation_hit_history.annotation_id,
-            authorName: item.annotation_hit_history.annotation_create_account?.name || 'N/A',
-            created_at: item.annotation_hit_history.created_at,
-          }
+function appendQAToChatList(newChatList: IChatItem[], item: any, conversationId: string, timezone: string, format: string) {
+  newChatList.push({
+    id: item.id,
+    content: item.answer,
+    agent_thoughts: addFileInfos(item.agent_thoughts ? sortAgentSorts(item.agent_thoughts) : item.agent_thoughts, item.message_files),
+    feedback: item.feedbacks.find((item: any) => item.from_source === 'user'), // user feedback
+    adminFeedback: item.feedbacks.find((item: any) => item.from_source === 'admin'), // admin feedback
+    feedbackDisabled: false,
+    isAnswer: true,
+    message_files: item.message_files?.filter((file: any) => file.belongs_to === 'assistant') || [],
+    log: [
+      ...item.message,
+      ...(item.message[item.message.length - 1]?.role !== 'assistant'
+        ? [
+          {
+            role: 'assistant',
+            text: item.answer,
+            files: item.message_files?.filter((file: any) => file.belongs_to === 'assistant') || [],
+          },
+        ]
+        : []),
+    ],
+    workflow_run_id: item.workflow_run_id,
+    conversationId,
+    input: {
+      inputs: item.inputs,
+      query: item.query,
+    },
+    more: {
+      time: dayjs.unix(item.created_at).tz(timezone).format(format),
+      tokens: item.answer_tokens + item.message_tokens,
+      latency: item.provider_response_latency.toFixed(2),
+    },
+    citation: item.metadata?.retriever_resources,
+    annotation: (() => {
+      if (item.annotation_hit_history) {
+        return {
+          id: item.annotation_hit_history.annotation_id,
+          authorName: item.annotation_hit_history.annotation_create_account?.name || 'N/A',
+          created_at: item.annotation_hit_history.created_at,
         }
+      }
 
-        if (item.annotation) {
-          return {
-            id: item.annotation.id,
-            authorName: item.annotation.account.name,
-            logAnnotation: item.annotation,
-            created_at: 0,
-          }
+      if (item.annotation) {
+        return {
+          id: item.annotation.id,
+          authorName: item.annotation.account.name,
+          logAnnotation: item.annotation,
+          created_at: 0,
         }
+      }
 
-        return undefined
-      })(),
-    })
+      return undefined
+    })(),
+    parentMessageId: `question-${item.id}`,
   })
-  return newChatList
+  newChatList.push({
+    id: `question-${item.id}`,
+    content: item.inputs.query || item.inputs.default_input || item.query, // text generation: item.inputs.query; chat: item.query
+    isAnswer: false,
+    message_files: item.message_files?.filter((file: any) => file.belongs_to === 'user') || [],
+    parentMessageId: item.parent_message_id || undefined,
+  })
+}
+
+const getFormattedChatList = (messages: ChatMessage[], conversationId: string, timezone: string, format: string) => {
+  const newChatList: IChatItem[] = []
+  let nextMessageId = null
+  for (const item of messages) {
+    if (!item.parent_message_id) {
+      appendQAToChatList(newChatList, item, conversationId, timezone, format)
+      break
+    }
+
+    if (!nextMessageId) {
+      appendQAToChatList(newChatList, item, conversationId, timezone, format)
+      nextMessageId = item.parent_message_id
+    }
+    else {
+      if (item.id === nextMessageId || nextMessageId === UUID_NIL) {
+        appendQAToChatList(newChatList, item, conversationId, timezone, format)
+        nextMessageId = item.parent_message_id
+      }
+    }
+  }
+  return newChatList.reverse()
 }
 
 // const displayedParams = CompletionParams.slice(0, -2)
@@ -171,6 +192,7 @@ function DetailPanel<T extends ChatConversationFullDetailResponse | CompletionCo
   })))
   const { t } = useTranslation()
   const [items, setItems] = React.useState<IChatItem[]>([])
+  const fetchedMessages = useRef<ChatMessage[]>([])
   const [hasMore, setHasMore] = useState(true)
   const [varValues, setVarValues] = useState<Record<string, string>>({})
   const fetchData = async () => {
@@ -192,7 +214,8 @@ function DetailPanel<T extends ChatConversationFullDetailResponse | CompletionCo
         const varValues = messageRes.data[0].inputs
         setVarValues(varValues)
       }
-      const newItems = [...getFormattedChatList(messageRes.data, detail.id, timezone!, t('appLog.dateTimeFormat') as string), ...items]
+      fetchedMessages.current = [...fetchedMessages.current, ...messageRes.data]
+      const newItems = getFormattedChatList(fetchedMessages.current, detail.id, timezone!, t('appLog.dateTimeFormat') as string)
       if (messageRes.has_more === false && detail?.model_config?.configs?.introduction) {
         newItems.unshift({
           id: 'introduction',
@@ -435,7 +458,7 @@ function DetailPanel<T extends ChatConversationFullDetailResponse | CompletionCo
             siteInfo={null}
           />
         </div>
-        : items.length < 8
+        : (items.length < 8 && !hasMore)
           ? <div className="pt-4 mb-4">
             <Chat
               config={{
diff --git a/web/app/components/base/chat/chat-with-history/chat-wrapper.tsx b/web/app/components/base/chat/chat-with-history/chat-wrapper.tsx
index 5a7bf1f17e592..79f6008d13671 100644
--- a/web/app/components/base/chat/chat-with-history/chat-wrapper.tsx
+++ b/web/app/components/base/chat/chat-with-history/chat-wrapper.tsx
@@ -2,6 +2,7 @@ import { useCallback, useEffect, useMemo } from 'react'
 import Chat from '../chat'
 import type {
   ChatConfig,
+  ChatItem,
   OnSend,
 } from '../types'
 import { useChat } from '../chat/hooks'
@@ -44,6 +45,8 @@ const ChatWrapper = () => {
   }, [appParams, currentConversationItem?.introduction, currentConversationId])
   const {
     chatList,
+    chatListRef,
+    handleUpdateChatList,
     handleSend,
     handleStop,
     isResponding,
@@ -63,11 +66,12 @@ const ChatWrapper = () => {
       currentChatInstanceRef.current.handleStop = handleStop
   }, [])
 
-  const doSend: OnSend = useCallback((message, files) => {
+  const doSend: OnSend = useCallback((message, files, last_answer) => {
     const data: any = {
       query: message,
       inputs: currentConversationId ? currentConversationItem?.inputs : newConversationInputs,
       conversation_id: currentConversationId,
+      parent_message_id: last_answer?.id || chatListRef.current.at(-1)?.id || null,
     }
 
     if (appConfig?.file_upload?.image.enabled && files?.length)
@@ -83,6 +87,7 @@ const ChatWrapper = () => {
       },
     )
   }, [
+    chatListRef,
     appConfig,
     currentConversationId,
     currentConversationItem,
@@ -92,6 +97,23 @@ const ChatWrapper = () => {
     isInstalledApp,
     appId,
   ])
+
+  const doRegenerate = useCallback((chatItem: ChatItem) => {
+    const index = chatList.findIndex(item => item.id === chatItem.id)
+    if (index === -1)
+      return
+
+    const prevMessages = chatList.slice(0, index)
+    const question = prevMessages.pop()
+    const lastAnswer = prevMessages.at(-1)
+
+    if (!question)
+      return
+
+    handleUpdateChatList(prevMessages)
+    doSend(question.content, question.message_files, (!lastAnswer || lastAnswer.isOpeningStatement) ? undefined : lastAnswer)
+  }, [chatList, handleUpdateChatList, doSend])
+
   const chatNode = useMemo(() => {
     if (inputsForms.length) {
       return (
@@ -148,6 +170,7 @@ const ChatWrapper = () => {
       chatFooterClassName='pb-4'
       chatFooterInnerClassName={`mx-auto w-full max-w-full ${isMobile && 'px-4'}`}
       onSend={doSend}
+      onRegenerate={doRegenerate}
       onStopResponding={handleStop}
       chatNode={chatNode}
       allToolIcons={appMeta?.tool_icons || {}}
diff --git a/web/app/components/base/chat/chat-with-history/hooks.tsx b/web/app/components/base/chat/chat-with-history/hooks.tsx
index 1e05cc39ef5dc..b9ebc42ec8889 100644
--- a/web/app/components/base/chat/chat-with-history/hooks.tsx
+++ b/web/app/components/base/chat/chat-with-history/hooks.tsx
@@ -12,10 +12,10 @@ import produce from 'immer'
 import type {
   Callback,
   ChatConfig,
-  ChatItem,
   Feedback,
 } from '../types'
 import { CONVERSATION_ID_INFO } from '../constants'
+import { getPrevChatList } from '../utils'
 import {
   delConversation,
   fetchAppInfo,
@@ -34,7 +34,6 @@ import type {
   AppData,
   ConversationItem,
 } from '@/models/share'
-import { addFileInfos, sortAgentSorts } from '@/app/components/tools/utils'
 import { useToastContext } from '@/app/components/base/toast'
 import { changeLanguage } from '@/i18n/i18next-config'
 import { useAppFavicon } from '@/hooks/use-app-favicon'
@@ -108,32 +107,12 @@ export const useChatWithHistory = (installedAppInfo?: InstalledApp) => {
   const { data: appConversationData, isLoading: appConversationDataLoading, mutate: mutateAppConversationData } = useSWR(['appConversationData', isInstalledApp, appId, false], () => fetchConversations(isInstalledApp, appId, undefined, false, 100))
   const { data: appChatListData, isLoading: appChatListDataLoading } = useSWR(chatShouldReloadKey ? ['appChatList', chatShouldReloadKey, isInstalledApp, appId] : null, () => fetchChatList(chatShouldReloadKey, isInstalledApp, appId))
 
-  const appPrevChatList = useMemo(() => {
-    const data = appChatListData?.data || []
-    const chatList: ChatItem[] = []
-
-    if (currentConversationId && data.length) {
-      data.forEach((item: any) => {
-        chatList.push({
-          id: `question-${item.id}`,
-          content: item.query,
-          isAnswer: false,
-          message_files: item.message_files?.filter((file: any) => file.belongs_to === 'user') || [],
-        })
-        chatList.push({
-          id: item.id,
-          content: item.answer,
-          agent_thoughts: addFileInfos(item.agent_thoughts ? sortAgentSorts(item.agent_thoughts) : item.agent_thoughts, item.message_files),
-          feedback: item.feedback,
-          isAnswer: true,
-          citation: item.retriever_resources,
-          message_files: item.message_files?.filter((file: any) => file.belongs_to === 'assistant') || [],
-        })
-      })
-    }
-
-    return chatList
-  }, [appChatListData, currentConversationId])
+  const appPrevChatList = useMemo(
+    () => (currentConversationId && appChatListData?.data.length)
+      ? getPrevChatList(appChatListData.data)
+      : [],
+    [appChatListData, currentConversationId],
+  )
 
   const [showNewConversationItemInList, setShowNewConversationItemInList] = useState(false)
 
diff --git a/web/app/components/base/chat/chat/answer/index.tsx b/web/app/components/base/chat/chat/answer/index.tsx
index 5fe2a7bad5107..705cd73ddf18d 100644
--- a/web/app/components/base/chat/chat/answer/index.tsx
+++ b/web/app/components/base/chat/chat/answer/index.tsx
@@ -35,6 +35,7 @@ type AnswerProps = {
   chatAnswerContainerInner?: string
   hideProcessDetail?: boolean
   appData?: AppData
+  noChatInput?: boolean
 }
 const Answer: FC<AnswerProps> = ({
   item,
@@ -48,6 +49,7 @@ const Answer: FC<AnswerProps> = ({
   chatAnswerContainerInner,
   hideProcessDetail,
   appData,
+  noChatInput,
 }) => {
   const { t } = useTranslation()
   const {
@@ -110,6 +112,7 @@ const Answer: FC<AnswerProps> = ({
                   question={question}
                   index={index}
                   showPromptLog={showPromptLog}
+                  noChatInput={noChatInput}
                 />
               )
             }
diff --git a/web/app/components/base/chat/chat/answer/operation.tsx b/web/app/components/base/chat/chat/answer/operation.tsx
index 08267bb09cd5c..5e5fc3b204641 100644
--- a/web/app/components/base/chat/chat/answer/operation.tsx
+++ b/web/app/components/base/chat/chat/answer/operation.tsx
@@ -7,6 +7,7 @@ import {
 import { useTranslation } from 'react-i18next'
 import type { ChatItem } from '../../types'
 import { useChatContext } from '../context'
+import RegenerateBtn from '@/app/components/base/regenerate-btn'
 import cn from '@/utils/classnames'
 import CopyBtn from '@/app/components/base/copy-btn'
 import { MessageFast } from '@/app/components/base/icons/src/vender/solid/communication'
@@ -28,6 +29,7 @@ type OperationProps = {
   maxSize: number
   contentWidth: number
   hasWorkflowProcess: boolean
+  noChatInput?: boolean
 }
 const Operation: FC<OperationProps> = ({
   item,
@@ -37,6 +39,7 @@ const Operation: FC<OperationProps> = ({
   maxSize,
   contentWidth,
   hasWorkflowProcess,
+  noChatInput,
 }) => {
   const { t } = useTranslation()
   const {
@@ -45,6 +48,7 @@ const Operation: FC<OperationProps> = ({
     onAnnotationEdited,
     onAnnotationRemoved,
     onFeedback,
+    onRegenerate,
   } = useChatContext()
   const [isShowReplyModal, setIsShowReplyModal] = useState(false)
   const {
@@ -159,12 +163,13 @@ const Operation: FC<OperationProps> = ({
             </div>
           )
         }
+        {
+          !isOpeningStatement && !noChatInput && <RegenerateBtn className='hidden group-hover:block mr-1' onClick={() => onRegenerate?.(item)} />
+        }
         {
           config?.supportFeedback && !localFeedback?.rating && onFeedback && !isOpeningStatement && (
-            <div className='hidden group-hover:flex ml-1 shrink-0 items-center px-0.5 bg-white border-[0.5px] border-gray-100 shadow-md text-gray-500 rounded-lg'>
-              <Tooltip
-                popupContent={t('appDebug.operation.agree')}
-              >
+            <div className='hidden group-hover:flex shrink-0 items-center px-0.5 bg-white border-[0.5px] border-gray-100 shadow-md text-gray-500 rounded-lg'>
+              <Tooltip popupContent={t('appDebug.operation.agree')}>
                 <div
                   className='flex items-center justify-center mr-0.5 w-6 h-6 rounded-md hover:bg-black/5 hover:text-gray-800 cursor-pointer'
                   onClick={() => handleFeedback('like')}
diff --git a/web/app/components/base/chat/chat/context.tsx b/web/app/components/base/chat/chat/context.tsx
index ba6f67189e1cc..c47b7501762e1 100644
--- a/web/app/components/base/chat/chat/context.tsx
+++ b/web/app/components/base/chat/chat/context.tsx
@@ -12,6 +12,7 @@ export type ChatContextValue = Pick<ChatProps, 'config'
   | 'answerIcon'
   | 'allToolIcons'
   | 'onSend'
+  | 'onRegenerate'
   | 'onAnnotationEdited'
   | 'onAnnotationAdded'
   | 'onAnnotationRemoved'
@@ -36,6 +37,7 @@ export const ChatContextProvider = ({
   answerIcon,
   allToolIcons,
   onSend,
+  onRegenerate,
   onAnnotationEdited,
   onAnnotationAdded,
   onAnnotationRemoved,
@@ -51,6 +53,7 @@ export const ChatContextProvider = ({
       answerIcon,
       allToolIcons,
       onSend,
+      onRegenerate,
       onAnnotationEdited,
       onAnnotationAdded,
       onAnnotationRemoved,
diff --git a/web/app/components/base/chat/chat/hooks.ts b/web/app/components/base/chat/chat/hooks.ts
index 892f88c4ad72b..dfb5a1b6855d5 100644
--- a/web/app/components/base/chat/chat/hooks.ts
+++ b/web/app/components/base/chat/chat/hooks.ts
@@ -647,7 +647,8 @@ export const useChat = (
 
   return {
     chatList,
-    setChatList,
+    chatListRef,
+    handleUpdateChatList,
     conversationId: conversationId.current,
     isResponding,
     setIsResponding,
diff --git a/web/app/components/base/chat/chat/index.tsx b/web/app/components/base/chat/chat/index.tsx
index 68194193c49dc..2847ae20c385a 100644
--- a/web/app/components/base/chat/chat/index.tsx
+++ b/web/app/components/base/chat/chat/index.tsx
@@ -16,6 +16,7 @@ import type {
   ChatConfig,
   ChatItem,
   Feedback,
+  OnRegenerate,
   OnSend,
 } from '../types'
 import type { ThemeBuilder } from '../embedded-chatbot/theme/theme-context'
@@ -42,6 +43,7 @@ export type ChatProps = {
   onStopResponding?: () => void
   noChatInput?: boolean
   onSend?: OnSend
+  onRegenerate?: OnRegenerate
   chatContainerClassName?: string
   chatContainerInnerClassName?: string
   chatFooterClassName?: string
@@ -67,6 +69,7 @@ const Chat: FC<ChatProps> = ({
   appData,
   config,
   onSend,
+  onRegenerate,
   chatList,
   isResponding,
   noStopResponding,
@@ -186,6 +189,7 @@ const Chat: FC<ChatProps> = ({
       answerIcon={answerIcon}
       allToolIcons={allToolIcons}
       onSend={onSend}
+      onRegenerate={onRegenerate}
       onAnnotationAdded={onAnnotationAdded}
       onAnnotationEdited={onAnnotationEdited}
       onAnnotationRemoved={onAnnotationRemoved}
@@ -219,6 +223,7 @@ const Chat: FC<ChatProps> = ({
                       showPromptLog={showPromptLog}
                       chatAnswerContainerInner={chatAnswerContainerInner}
                       hideProcessDetail={hideProcessDetail}
+                      noChatInput={noChatInput}
                     />
                   )
                 }
diff --git a/web/app/components/base/chat/chat/type.ts b/web/app/components/base/chat/chat/type.ts
index b2cb18011c692..dd26a4179d4fb 100644
--- a/web/app/components/base/chat/chat/type.ts
+++ b/web/app/components/base/chat/chat/type.ts
@@ -95,6 +95,7 @@ export type IChatItem = {
   // for agent log
   conversationId?: string
   input?: any
+  parentMessageId?: string
 }
 
 export type Metadata = {
diff --git a/web/app/components/base/chat/constants.ts b/web/app/components/base/chat/constants.ts
index 8249be7375d2e..309f0f04a716b 100644
--- a/web/app/components/base/chat/constants.ts
+++ b/web/app/components/base/chat/constants.ts
@@ -1 +1,2 @@
 export const CONVERSATION_ID_INFO = 'conversationIdInfo'
+export const UUID_NIL = '00000000-0000-0000-0000-000000000000'
diff --git a/web/app/components/base/chat/embedded-chatbot/chat-wrapper.tsx b/web/app/components/base/chat/embedded-chatbot/chat-wrapper.tsx
index 48ee4110584ae..8cb546fd52a0e 100644
--- a/web/app/components/base/chat/embedded-chatbot/chat-wrapper.tsx
+++ b/web/app/components/base/chat/embedded-chatbot/chat-wrapper.tsx
@@ -2,6 +2,7 @@ import { useCallback, useEffect, useMemo } from 'react'
 import Chat from '../chat'
 import type {
   ChatConfig,
+  ChatItem,
   OnSend,
 } from '../types'
 import { useChat } from '../chat/hooks'
@@ -45,11 +46,13 @@ const ChatWrapper = () => {
     } as ChatConfig
   }, [appParams, currentConversationItem?.introduction, currentConversationId])
   const {
+    chatListRef,
     chatList,
     handleSend,
     handleStop,
     isResponding,
     suggestedQuestions,
+    handleUpdateChatList,
   } = useChat(
     appConfig,
     {
@@ -65,11 +68,12 @@ const ChatWrapper = () => {
       currentChatInstanceRef.current.handleStop = handleStop
   }, [])
 
-  const doSend: OnSend = useCallback((message, files) => {
+  const doSend: OnSend = useCallback((message, files, last_answer) => {
     const data: any = {
       query: message,
       inputs: currentConversationId ? currentConversationItem?.inputs : newConversationInputs,
       conversation_id: currentConversationId,
+      parent_message_id: last_answer?.id || chatListRef.current.at(-1)?.id || null,
     }
 
     if (appConfig?.file_upload?.image.enabled && files?.length)
@@ -85,6 +89,7 @@ const ChatWrapper = () => {
       },
     )
   }, [
+    chatListRef,
     appConfig,
     currentConversationId,
     currentConversationItem,
@@ -94,6 +99,23 @@ const ChatWrapper = () => {
     isInstalledApp,
     appId,
   ])
+
+  const doRegenerate = useCallback((chatItem: ChatItem) => {
+    const index = chatList.findIndex(item => item.id === chatItem.id)
+    if (index === -1)
+      return
+
+    const prevMessages = chatList.slice(0, index)
+    const question = prevMessages.pop()
+    const lastAnswer = prevMessages.at(-1)
+
+    if (!question)
+      return
+
+    handleUpdateChatList(prevMessages)
+    doSend(question.content, question.message_files, (!lastAnswer || lastAnswer.isOpeningStatement) ? undefined : lastAnswer)
+  }, [chatList, handleUpdateChatList, doSend])
+
   const chatNode = useMemo(() => {
     if (inputsForms.length) {
       return (
@@ -136,6 +158,7 @@ const ChatWrapper = () => {
       chatFooterClassName='pb-4'
       chatFooterInnerClassName={cn('mx-auto w-full max-w-full tablet:px-4', isMobile && 'px-4')}
       onSend={doSend}
+      onRegenerate={doRegenerate}
       onStopResponding={handleStop}
       chatNode={chatNode}
       allToolIcons={appMeta?.tool_icons || {}}
diff --git a/web/app/components/base/chat/embedded-chatbot/hooks.tsx b/web/app/components/base/chat/embedded-chatbot/hooks.tsx
index 39d25f57d194e..fd89efcbff3ab 100644
--- a/web/app/components/base/chat/embedded-chatbot/hooks.tsx
+++ b/web/app/components/base/chat/embedded-chatbot/hooks.tsx
@@ -11,10 +11,10 @@ import { useLocalStorageState } from 'ahooks'
 import produce from 'immer'
 import type {
   ChatConfig,
-  ChatItem,
   Feedback,
 } from '../types'
 import { CONVERSATION_ID_INFO } from '../constants'
+import { getPrevChatList, getProcessedInputsFromUrlParams } from '../utils'
 import {
   fetchAppInfo,
   fetchAppMeta,
@@ -28,10 +28,8 @@ import type {
   // AppData,
   ConversationItem,
 } from '@/models/share'
-import { addFileInfos, sortAgentSorts } from '@/app/components/tools/utils'
 import { useToastContext } from '@/app/components/base/toast'
 import { changeLanguage } from '@/i18n/i18next-config'
-import { getProcessedInputsFromUrlParams } from '@/app/components/base/chat/utils'
 
 export const useEmbeddedChatbot = () => {
   const isInstalledApp = false
@@ -75,32 +73,12 @@ export const useEmbeddedChatbot = () => {
   const { data: appConversationData, isLoading: appConversationDataLoading, mutate: mutateAppConversationData } = useSWR(['appConversationData', isInstalledApp, appId, false], () => fetchConversations(isInstalledApp, appId, undefined, false, 100))
   const { data: appChatListData, isLoading: appChatListDataLoading } = useSWR(chatShouldReloadKey ? ['appChatList', chatShouldReloadKey, isInstalledApp, appId] : null, () => fetchChatList(chatShouldReloadKey, isInstalledApp, appId))
 
-  const appPrevChatList = useMemo(() => {
-    const data = appChatListData?.data || []
-    const chatList: ChatItem[] = []
-
-    if (currentConversationId && data.length) {
-      data.forEach((item: any) => {
-        chatList.push({
-          id: `question-${item.id}`,
-          content: item.query,
-          isAnswer: false,
-          message_files: item.message_files?.filter((file: any) => file.belongs_to === 'user') || [],
-        })
-        chatList.push({
-          id: item.id,
-          content: item.answer,
-          agent_thoughts: addFileInfos(item.agent_thoughts ? sortAgentSorts(item.agent_thoughts) : item.agent_thoughts, item.message_files),
-          feedback: item.feedback,
-          isAnswer: true,
-          citation: item.retriever_resources,
-          message_files: item.message_files?.filter((file: any) => file.belongs_to === 'assistant') || [],
-        })
-      })
-    }
-
-    return chatList
-  }, [appChatListData, currentConversationId])
+  const appPrevChatList = useMemo(
+    () => (currentConversationId && appChatListData?.data.length)
+      ? getPrevChatList(appChatListData.data)
+      : [],
+    [appChatListData, currentConversationId],
+  )
 
   const [showNewConversationItemInList, setShowNewConversationItemInList] = useState(false)
 
@@ -155,7 +133,7 @@ export const useEmbeddedChatbot = () => {
         type: 'text-input',
       }
     })
-  }, [appParams])
+  }, [initInputs, appParams])
 
   useEffect(() => {
     // init inputs from url params
diff --git a/web/app/components/base/chat/types.ts b/web/app/components/base/chat/types.ts
index 21277fec57099..489dbb44cf67e 100644
--- a/web/app/components/base/chat/types.ts
+++ b/web/app/components/base/chat/types.ts
@@ -63,7 +63,9 @@ export type ChatItem = IChatItem & {
   conversationId?: string
 }
 
-export type OnSend = (message: string, files?: VisionFile[]) => void
+export type OnSend = (message: string, files?: VisionFile[], last_answer?: ChatItem) => void
+
+export type OnRegenerate = (chatItem: ChatItem) => void
 
 export type Callback = {
   onSuccess: () => void
diff --git a/web/app/components/base/chat/utils.ts b/web/app/components/base/chat/utils.ts
index 3fe5050cc7b34..e851c4c4633fb 100644
--- a/web/app/components/base/chat/utils.ts
+++ b/web/app/components/base/chat/utils.ts
@@ -1,7 +1,11 @@
+import { addFileInfos, sortAgentSorts } from '../../tools/utils'
+import { UUID_NIL } from './constants'
+import type { ChatItem } from './types'
+
 async function decodeBase64AndDecompress(base64String: string) {
   const binaryString = atob(base64String)
   const compressedUint8Array = Uint8Array.from(binaryString, char => char.charCodeAt(0))
-  const decompressedStream = new Response(compressedUint8Array).body.pipeThrough(new DecompressionStream('gzip'))
+  const decompressedStream = new Response(compressedUint8Array).body?.pipeThrough(new DecompressionStream('gzip'))
   const decompressedArrayBuffer = await new Response(decompressedStream).arrayBuffer()
   return new TextDecoder().decode(decompressedArrayBuffer)
 }
@@ -15,6 +19,57 @@ function getProcessedInputsFromUrlParams(): Record<string, any> {
   return inputs
 }
 
+function appendQAToChatList(chatList: ChatItem[], item: any) {
+  // we append answer first and then question since will reverse the whole chatList later
+  chatList.push({
+    id: item.id,
+    content: item.answer,
+    agent_thoughts: addFileInfos(item.agent_thoughts ? sortAgentSorts(item.agent_thoughts) : item.agent_thoughts, item.message_files),
+    feedback: item.feedback,
+    isAnswer: true,
+    citation: item.retriever_resources,
+    message_files: item.message_files?.filter((file: any) => file.belongs_to === 'assistant') || [],
+  })
+  chatList.push({
+    id: `question-${item.id}`,
+    content: item.query,
+    isAnswer: false,
+    message_files: item.message_files?.filter((file: any) => file.belongs_to === 'user') || [],
+  })
+}
+
+/**
+ * Computes the latest thread messages from all messages of the conversation.
+ * Same logic as backend codebase `api/core/prompt/utils/extract_thread_messages.py`
+ *
+ * @param fetchedMessages - The history chat list data from the backend, sorted by created_at in descending order. This includes all flattened history messages of the conversation.
+ * @returns An array of ChatItems representing the latest thread.
+ */
+function getPrevChatList(fetchedMessages: any[]) {
+  const ret: ChatItem[] = []
+  let nextMessageId = null
+
+  for (const item of fetchedMessages) {
+    if (!item.parent_message_id) {
+      appendQAToChatList(ret, item)
+      break
+    }
+
+    if (!nextMessageId) {
+      appendQAToChatList(ret, item)
+      nextMessageId = item.parent_message_id
+    }
+    else {
+      if (item.id === nextMessageId || nextMessageId === UUID_NIL) {
+        appendQAToChatList(ret, item)
+        nextMessageId = item.parent_message_id
+      }
+    }
+  }
+  return ret.reverse()
+}
+
 export {
   getProcessedInputsFromUrlParams,
+  getPrevChatList,
 }
diff --git a/web/app/components/base/icons/assets/vender/line/general/refresh.svg b/web/app/components/base/icons/assets/vender/line/general/refresh.svg
new file mode 100644
index 0000000000000..05cf98682734a
--- /dev/null
+++ b/web/app/components/base/icons/assets/vender/line/general/refresh.svg
@@ -0,0 +1 @@
+<svg xmlns="http://www.w3.org/2000/svg" viewBox="0 0 24 24" fill="currentColor"><path d="M5.46257 4.43262C7.21556 2.91688 9.5007 2 12 2C17.5228 2 22 6.47715 22 12C22 14.1361 21.3302 16.1158 20.1892 17.7406L17 12H20C20 7.58172 16.4183 4 12 4C9.84982 4 7.89777 4.84827 6.46023 6.22842L5.46257 4.43262ZM18.5374 19.5674C16.7844 21.0831 14.4993 22 12 22C6.47715 22 2 17.5228 2 12C2 9.86386 2.66979 7.88416 3.8108 6.25944L7 12H4C4 16.4183 7.58172 20 12 20C14.1502 20 16.1022 19.1517 17.5398 17.7716L18.5374 19.5674Z"></path></svg>
\ No newline at end of file
diff --git a/web/app/components/base/icons/src/vender/line/general/Refresh.json b/web/app/components/base/icons/src/vender/line/general/Refresh.json
new file mode 100644
index 0000000000000..128dcb7d4d713
--- /dev/null
+++ b/web/app/components/base/icons/src/vender/line/general/Refresh.json
@@ -0,0 +1,23 @@
+{
+	"icon": {
+		"type": "element",
+		"isRootNode": true,
+		"name": "svg",
+		"attributes": {
+			"xmlns": "http://www.w3.org/2000/svg",
+			"viewBox": "0 0 24 24",
+			"fill": "currentColor"
+		},
+		"children": [
+			{
+				"type": "element",
+				"name": "path",
+				"attributes": {
+					"d": "M5.46257 4.43262C7.21556 2.91688 9.5007 2 12 2C17.5228 2 22 6.47715 22 12C22 14.1361 21.3302 16.1158 20.1892 17.7406L17 12H20C20 7.58172 16.4183 4 12 4C9.84982 4 7.89777 4.84827 6.46023 6.22842L5.46257 4.43262ZM18.5374 19.5674C16.7844 21.0831 14.4993 22 12 22C6.47715 22 2 17.5228 2 12C2 9.86386 2.66979 7.88416 3.8108 6.25944L7 12H4C4 16.4183 7.58172 20 12 20C14.1502 20 16.1022 19.1517 17.5398 17.7716L18.5374 19.5674Z"
+				},
+				"children": []
+			}
+		]
+	},
+	"name": "Refresh"
+}
\ No newline at end of file
diff --git a/web/app/components/base/icons/src/vender/line/general/Refresh.tsx b/web/app/components/base/icons/src/vender/line/general/Refresh.tsx
new file mode 100644
index 0000000000000..96641f1c4243b
--- /dev/null
+++ b/web/app/components/base/icons/src/vender/line/general/Refresh.tsx
@@ -0,0 +1,16 @@
+// GENERATE BY script
+// DON NOT EDIT IT MANUALLY
+
+import * as React from 'react'
+import data from './Refresh.json'
+import IconBase from '@/app/components/base/icons/IconBase'
+import type { IconBaseProps, IconData } from '@/app/components/base/icons/IconBase'
+
+const Icon = React.forwardRef<React.MutableRefObject<SVGElement>, Omit<IconBaseProps, 'data'>>((
+  props,
+  ref,
+) => <IconBase {...props} ref={ref} data={data as IconData} />)
+
+Icon.displayName = 'Refresh'
+
+export default Icon
diff --git a/web/app/components/base/icons/src/vender/line/general/index.ts b/web/app/components/base/icons/src/vender/line/general/index.ts
index c1af2e49948a4..b5c7a7bbc1d9a 100644
--- a/web/app/components/base/icons/src/vender/line/general/index.ts
+++ b/web/app/components/base/icons/src/vender/line/general/index.ts
@@ -18,6 +18,7 @@ export { default as Menu01 } from './Menu01'
 export { default as Pin01 } from './Pin01'
 export { default as Pin02 } from './Pin02'
 export { default as Plus02 } from './Plus02'
+export { default as Refresh } from './Refresh'
 export { default as Settings01 } from './Settings01'
 export { default as Settings04 } from './Settings04'
 export { default as Target04 } from './Target04'
diff --git a/web/app/components/base/regenerate-btn/index.tsx b/web/app/components/base/regenerate-btn/index.tsx
new file mode 100644
index 0000000000000..aaf0206df609d
--- /dev/null
+++ b/web/app/components/base/regenerate-btn/index.tsx
@@ -0,0 +1,31 @@
+'use client'
+import { t } from 'i18next'
+import { Refresh } from '../icons/src/vender/line/general'
+import Tooltip from '@/app/components/base/tooltip'
+
+type Props = {
+  className?: string
+  onClick?: () => void
+}
+
+const RegenerateBtn = ({ className, onClick }: Props) => {
+  return (
+    <div className={`${className}`}>
+      <Tooltip
+        popupContent={t('appApi.regenerate') as string}
+      >
+        <div
+          className={'box-border p-0.5 flex items-center justify-center rounded-md bg-white cursor-pointer'}
+          onClick={() => onClick?.()}
+          style={{
+            boxShadow: '0px 4px 8px -2px rgba(16, 24, 40, 0.1), 0px 2px 4px -2px rgba(16, 24, 40, 0.06)',
+          }}
+        >
+          <Refresh className="p-[3.5px] w-6 h-6 text-[#667085] hover:bg-gray-50" />
+        </div>
+      </Tooltip>
+    </div>
+  )
+}
+
+export default RegenerateBtn
diff --git a/web/app/components/workflow/panel/chat-record/index.tsx b/web/app/components/workflow/panel/chat-record/index.tsx
index afd20b7358670..1bcfd6474d7c8 100644
--- a/web/app/components/workflow/panel/chat-record/index.tsx
+++ b/web/app/components/workflow/panel/chat-record/index.tsx
@@ -2,7 +2,6 @@ import {
   memo,
   useCallback,
   useEffect,
-  useMemo,
   useState,
 } from 'react'
 import { RiCloseLine } from '@remixicon/react'
@@ -17,50 +16,70 @@ import type { ChatItem } from '@/app/components/base/chat/types'
 import { fetchConversationMessages } from '@/service/debug'
 import { useStore as useAppStore } from '@/app/components/app/store'
 import Loading from '@/app/components/base/loading'
+import { UUID_NIL } from '@/app/components/base/chat/constants'
+
+function appendQAToChatList(newChatList: ChatItem[], item: any) {
+  newChatList.push({
+    id: item.id,
+    content: item.answer,
+    feedback: item.feedback,
+    isAnswer: true,
+    citation: item.metadata?.retriever_resources,
+    message_files: item.message_files?.filter((file: any) => file.belongs_to === 'assistant') || [],
+    workflow_run_id: item.workflow_run_id,
+  })
+  newChatList.push({
+    id: `question-${item.id}`,
+    content: item.query,
+    isAnswer: false,
+    message_files: item.message_files?.filter((file: any) => file.belongs_to === 'user') || [],
+  })
+}
+
+function getFormattedChatList(messages: any[]) {
+  const newChatList: ChatItem[] = []
+  let nextMessageId = null
+  for (const item of messages) {
+    if (!item.parent_message_id) {
+      appendQAToChatList(newChatList, item)
+      break
+    }
+
+    if (!nextMessageId) {
+      appendQAToChatList(newChatList, item)
+      nextMessageId = item.parent_message_id
+    }
+    else {
+      if (item.id === nextMessageId || nextMessageId === UUID_NIL) {
+        appendQAToChatList(newChatList, item)
+        nextMessageId = item.parent_message_id
+      }
+    }
+  }
+  return newChatList.reverse()
+}
 
 const ChatRecord = () => {
   const [fetched, setFetched] = useState(false)
-  const [chatList, setChatList] = useState([])
+  const [chatList, setChatList] = useState<ChatItem[]>([])
   const appDetail = useAppStore(s => s.appDetail)
   const workflowStore = useWorkflowStore()
   const { handleLoadBackupDraft } = useWorkflowRun()
   const historyWorkflowData = useStore(s => s.historyWorkflowData)
   const currentConversationID = historyWorkflowData?.conversation_id
 
-  const chatMessageList = useMemo(() => {
-    const res: ChatItem[] = []
-    if (chatList.length) {
-      chatList.forEach((item: any) => {
-        res.push({
-          id: `question-${item.id}`,
-          content: item.query,
-          isAnswer: false,
-          message_files: item.message_files?.filter((file: any) => file.belongs_to === 'user') || [],
-        })
-        res.push({
-          id: item.id,
-          content: item.answer,
-          feedback: item.feedback,
-          isAnswer: true,
-          citation: item.metadata?.retriever_resources,
-          message_files: item.message_files?.filter((file: any) => file.belongs_to === 'assistant') || [],
-          workflow_run_id: item.workflow_run_id,
-        })
-      })
-    }
-    return res
-  }, [chatList])
-
   const handleFetchConversationMessages = useCallback(async () => {
     if (appDetail && currentConversationID) {
       try {
         setFetched(false)
         const res = await fetchConversationMessages(appDetail.id, currentConversationID)
-        setFetched(true)
-        setChatList((res as any).data)
+        setChatList(getFormattedChatList((res as any).data))
       }
       catch (e) {
-
+        console.error(e)
+      }
+      finally {
+        setFetched(true)
       }
     }
   }, [appDetail, currentConversationID])
@@ -101,7 +120,7 @@ const ChatRecord = () => {
               config={{
                 supportCitationHitInfo: true,
               } as any}
-              chatList={chatMessageList}
+              chatList={chatList}
               chatContainerClassName='px-4'
               chatContainerInnerClassName='pt-6 w-full max-w-full mx-auto'
               chatFooterClassName='px-4 rounded-b-2xl'
diff --git a/web/app/components/workflow/panel/debug-and-preview/chat-wrapper.tsx b/web/app/components/workflow/panel/debug-and-preview/chat-wrapper.tsx
index a7dd607e221ea..107a5dc698f74 100644
--- a/web/app/components/workflow/panel/debug-and-preview/chat-wrapper.tsx
+++ b/web/app/components/workflow/panel/debug-and-preview/chat-wrapper.tsx
@@ -18,7 +18,7 @@ import ConversationVariableModal from './conversation-variable-modal'
 import { useChat } from './hooks'
 import type { ChatWrapperRefType } from './index'
 import Chat from '@/app/components/base/chat/chat'
-import type { OnSend } from '@/app/components/base/chat/types'
+import type { ChatItem, OnSend } from '@/app/components/base/chat/types'
 import { useFeaturesStore } from '@/app/components/base/features/hooks'
 import {
   fetchSuggestedQuestions,
@@ -58,6 +58,8 @@ const ChatWrapper = forwardRef<ChatWrapperRefType, ChatWrapperProps>(({ showConv
   const {
     conversationId,
     chatList,
+    chatListRef,
+    handleUpdateChatList,
     handleStop,
     isResponding,
     suggestedQuestions,
@@ -73,19 +75,36 @@ const ChatWrapper = forwardRef<ChatWrapperRefType, ChatWrapperProps>(({ showConv
     taskId => stopChatMessageResponding(appDetail!.id, taskId),
   )
 
-  const doSend = useCallback<OnSend>((query, files) => {
+  const doSend = useCallback<OnSend>((query, files, last_answer) => {
     handleSend(
       {
         query,
         files,
         inputs: workflowStore.getState().inputs,
         conversation_id: conversationId,
+        parent_message_id: last_answer?.id || chatListRef.current.at(-1)?.id || null,
       },
       {
         onGetSuggestedQuestions: (messageId, getAbortController) => fetchSuggestedQuestions(appDetail!.id, messageId, getAbortController),
       },
     )
-  }, [conversationId, handleSend, workflowStore, appDetail])
+  }, [chatListRef, conversationId, handleSend, workflowStore, appDetail])
+
+  const doRegenerate = useCallback((chatItem: ChatItem) => {
+    const index = chatList.findIndex(item => item.id === chatItem.id)
+    if (index === -1)
+      return
+
+    const prevMessages = chatList.slice(0, index)
+    const question = prevMessages.pop()
+    const lastAnswer = prevMessages.at(-1)
+
+    if (!question)
+      return
+
+    handleUpdateChatList(prevMessages)
+    doSend(question.content, question.message_files, (!lastAnswer || lastAnswer.isOpeningStatement) ? undefined : lastAnswer)
+  }, [chatList, handleUpdateChatList, doSend])
 
   useImperativeHandle(ref, () => {
     return {
@@ -107,6 +126,7 @@ const ChatWrapper = forwardRef<ChatWrapperRefType, ChatWrapperProps>(({ showConv
         chatFooterClassName='px-4 rounded-bl-2xl'
         chatFooterInnerClassName='pb-4 w-full max-w-full mx-auto'
         onSend={doSend}
+        onRegenerate={doRegenerate}
         onStopResponding={handleStop}
         chatNode={(
           <>
diff --git a/web/app/components/workflow/panel/debug-and-preview/hooks.ts b/web/app/components/workflow/panel/debug-and-preview/hooks.ts
index 51a018bcb15b4..cad76a4490c85 100644
--- a/web/app/components/workflow/panel/debug-and-preview/hooks.ts
+++ b/web/app/components/workflow/panel/debug-and-preview/hooks.ts
@@ -387,6 +387,8 @@ export const useChat = (
   return {
     conversationId: conversationId.current,
     chatList,
+    chatListRef,
+    handleUpdateChatList,
     handleSend,
     handleStop,
     handleRestart,
diff --git a/web/hooks/use-app-favicon.ts b/web/hooks/use-app-favicon.ts
index 86eadc1b3d086..1ff743928faae 100644
--- a/web/hooks/use-app-favicon.ts
+++ b/web/hooks/use-app-favicon.ts
@@ -5,10 +5,10 @@ import type { AppIconType } from '@/types/app'
 
 type UseAppFaviconOptions = {
   enable?: boolean
-  icon_type?: AppIconType
+  icon_type?: AppIconType | null
   icon?: string
-  icon_background?: string
-  icon_url?: string
+  icon_background?: string | null
+  icon_url?: string | null
 }
 
 export function useAppFavicon(options: UseAppFaviconOptions) {
diff --git a/web/i18n/en-US/app-api.ts b/web/i18n/en-US/app-api.ts
index 631faeee9a5e7..355ff30602736 100644
--- a/web/i18n/en-US/app-api.ts
+++ b/web/i18n/en-US/app-api.ts
@@ -6,6 +6,7 @@ const translation = {
   ok: 'In Service',
   copy: 'Copy',
   copied: 'Copied',
+  regenerate: 'Regenerate',
   play: 'Play',
   pause: 'Pause',
   playing: 'Playing',
diff --git a/web/i18n/zh-Hans/app-api.ts b/web/i18n/zh-Hans/app-api.ts
index 6b9048b66efb9..a0defdab62517 100644
--- a/web/i18n/zh-Hans/app-api.ts
+++ b/web/i18n/zh-Hans/app-api.ts
@@ -6,6 +6,7 @@ const translation = {
   ok: '运行中',
   copy: '复制',
   copied: '已复制',
+  regenerate: '重新生成',
   play: '播放',
   pause: '暂停',
   playing: '播放中',
diff --git a/web/models/log.ts b/web/models/log.ts
index 8da1c4cf4e826..dc557bfe2142b 100644
--- a/web/models/log.ts
+++ b/web/models/log.ts
@@ -106,6 +106,7 @@ export type MessageContent = {
   metadata: Metadata
   agent_thoughts: any[] // TODO
   workflow_run_id: string
+  parent_message_id: string | null
 }
 
 export type CompletionConversationGeneralDetail = {

From a587f0d3f14cde1c36ae5206c8eac5b87ec84738 Mon Sep 17 00:00:00 2001
From: Shota Totsuka <153569547+totsukash@users.noreply.github.com>
Date: Sun, 22 Sep 2024 10:04:00 +0900
Subject: [PATCH 013/235] docs: Add Japanese documentation for tools (#8469)

---
 api/core/tools/README.md                      |   4 +-
 api/core/tools/README_CN.md                   |   4 +-
 api/core/tools/README_JP.md                   |  31 ++
 api/core/tools/docs/en_US/tool_scale_out.md   |   2 +-
 .../{zh_Hans => }/images/index/image-1.png    | Bin
 .../{zh_Hans => }/images/index/image-2.png    | Bin
 .../docs/{zh_Hans => }/images/index/image.png | Bin
 .../tools/docs/ja_JP/advanced_scale_out.md    | 283 ++++++++++++++++++
 api/core/tools/docs/ja_JP/tool_scale_out.md   | 240 +++++++++++++++
 api/core/tools/docs/zh_Hans/tool_scale_out.md |   2 +-
 10 files changed, 560 insertions(+), 6 deletions(-)
 create mode 100644 api/core/tools/README_JP.md
 rename api/core/tools/docs/{zh_Hans => }/images/index/image-1.png (100%)
 rename api/core/tools/docs/{zh_Hans => }/images/index/image-2.png (100%)
 rename api/core/tools/docs/{zh_Hans => }/images/index/image.png (100%)
 create mode 100644 api/core/tools/docs/ja_JP/advanced_scale_out.md
 create mode 100644 api/core/tools/docs/ja_JP/tool_scale_out.md

diff --git a/api/core/tools/README.md b/api/core/tools/README.md
index c7ee81422efbd..b5d0a30d348a9 100644
--- a/api/core/tools/README.md
+++ b/api/core/tools/README.md
@@ -9,10 +9,10 @@ The tools provided for Agents and Workflows are currently divided into two categ
 - `Api-Based Tools`  leverage third-party APIs for implementation. You don't need to code to integrate these -- simply provide interface definitions in formats like `OpenAPI` , `Swagger`, or the `OpenAI-plugin` on the front-end.
 
 ### Built-in Tool Providers
-![Alt text](docs/zh_Hans/images/index/image.png)
+![Alt text](docs/images/index/image.png)
 
 ### API Tool Providers
-![Alt text](docs/zh_Hans/images/index/image-1.png)
+![Alt text](docs/images/index/image-1.png)
 
 ## Tool Integration
 
diff --git a/api/core/tools/README_CN.md b/api/core/tools/README_CN.md
index fda5d0630ca8c..7e18441131fd7 100644
--- a/api/core/tools/README_CN.md
+++ b/api/core/tools/README_CN.md
@@ -12,10 +12,10 @@
 - `Api-Based Tools` 基于API的工具，即通过调用第三方API实现的工具，`Api-Based Tool`不需要再额外定义，只需提供`OpenAPI` `Swagger` `OpenAI plugin`等接口文档即可。
 
 ### 内置工具供应商
-![Alt text](docs/zh_Hans/images/index/image.png)
+![Alt text](docs/images/index/image.png)
 
 ### API工具供应商
-![Alt text](docs/zh_Hans/images/index/image-1.png)
+![Alt text](docs/images/index/image-1.png)
 
 ## 工具接入
 为了实现更灵活更强大的功能，Tools提供了一系列的接口，帮助开发者快速构建想要的工具，本文作为开发者的入门指南，将会以[快速接入](./docs/zh_Hans/tool_scale_out.md)和[高级接入](./docs/zh_Hans/advanced_scale_out.md)两部分介绍如何接入工具。
diff --git a/api/core/tools/README_JP.md b/api/core/tools/README_JP.md
new file mode 100644
index 0000000000000..39d0bf1762ad0
--- /dev/null
+++ b/api/core/tools/README_JP.md
@@ -0,0 +1,31 @@
+# Tools
+
+このモジュールは、Difyのエージェントアシスタントやワークフローで使用される組み込みツールを実装しています。このモジュールでは、フロントエンドのロジックを変更することなく、独自のツールを定義し表示することができます。この分離により、Difyの機能を容易に水平方向にスケールアウトできます。
+
+## 機能紹介
+
+エージェントとワークフロー向けに提供されるツールは、現在2つのカテゴリーに分類されています。
+
+- `Built-in Tools`はDify内部で実装され、エージェントとワークフローで使用するためにハードコードされています。
+- `Api-Based Tools`はサードパーティのAPIを利用して実装されています。これらを統合するためのコーディングは不要で、フロントエンドで
+  `OpenAPI`, `Swagger`または`OpenAI-plugin`などの形式でインターフェース定義を提供するだけです。
+
+### 組み込みツールプロバイダー
+
+![Alt text](docs/images/index/image.png)
+
+### APIツールプロバイダー
+
+![Alt text](docs/images/index/image-1.png)
+
+## ツールの統合
+
+開発者が柔軟で強力なツールを構築できるよう、2つのガイドを提供しています。
+
+### [クイック統合 👈🏻](./docs/ja_JP/tool_scale_out.md)
+
+クイック統合は、Google検索ツールの例を通じて、ツール統合の基本をすばやく理解できるようにすることを目的としています。
+
+### [高度な統合 👈🏻](./docs/ja_JP/advanced_scale_out.md)
+
+高度な統合では、モジュールインターフェースについてより深く掘り下げ、画像生成、複数ツールの組み合わせ、異なるツール間でのパラメーター、画像、ファイルのフロー管理など、より複雑な機能の実装方法を説明します。
\ No newline at end of file
diff --git a/api/core/tools/docs/en_US/tool_scale_out.md b/api/core/tools/docs/en_US/tool_scale_out.md
index 121b7a5a76d22..1deaf04a47539 100644
--- a/api/core/tools/docs/en_US/tool_scale_out.md
+++ b/api/core/tools/docs/en_US/tool_scale_out.md
@@ -245,4 +245,4 @@ After the above steps are completed, we can see this tool on the frontend, and i
 
 Of course, because google_search needs a credential, before using it, you also need to input your credentials on the frontend.
 
-![Alt text](../zh_Hans/images/index/image-2.png)
+![Alt text](../images/index/image-2.png)
diff --git a/api/core/tools/docs/zh_Hans/images/index/image-1.png b/api/core/tools/docs/images/index/image-1.png
similarity index 100%
rename from api/core/tools/docs/zh_Hans/images/index/image-1.png
rename to api/core/tools/docs/images/index/image-1.png
diff --git a/api/core/tools/docs/zh_Hans/images/index/image-2.png b/api/core/tools/docs/images/index/image-2.png
similarity index 100%
rename from api/core/tools/docs/zh_Hans/images/index/image-2.png
rename to api/core/tools/docs/images/index/image-2.png
diff --git a/api/core/tools/docs/zh_Hans/images/index/image.png b/api/core/tools/docs/images/index/image.png
similarity index 100%
rename from api/core/tools/docs/zh_Hans/images/index/image.png
rename to api/core/tools/docs/images/index/image.png
diff --git a/api/core/tools/docs/ja_JP/advanced_scale_out.md b/api/core/tools/docs/ja_JP/advanced_scale_out.md
new file mode 100644
index 0000000000000..96f843354f91b
--- /dev/null
+++ b/api/core/tools/docs/ja_JP/advanced_scale_out.md
@@ -0,0 +1,283 @@
+# 高度なツール統合
+
+このガイドを始める前に、Difyのツール統合プロセスの基本を理解していることを確認してください。簡単な概要については[クイック統合](./tool_scale_out.md)をご覧ください。
+
+## ツールインターフェース
+
+より複雑なツールを迅速に構築するのを支援するため、`Tool`クラスに一連のヘルパーメソッドを定義しています。
+
+### メッセージの返却
+
+Difyは`テキスト`、`リンク`、`画像`、`ファイルBLOB`、`JSON`などの様々なメッセージタイプをサポートしています。以下のインターフェースを通じて、異なるタイプのメッセージをLLMとユーザーに返すことができます。
+
+注意：以下のインターフェースの一部のパラメータについては、後のセクションで説明します。
+
+#### 画像URL
+画像のURLを渡すだけで、Difyが自動的に画像をダウンロードしてユーザーに返します。
+
+```python
+    def create_image_message(self, image: str, save_as: str = '') -> ToolInvokeMessage:
+    """
+        create an image message
+
+        :param image: the url of the image
+        :param save_as: save as
+        :return: the image message
+    """
+```
+
+#### リンク
+リンクを返す必要がある場合は、以下のインターフェースを使用できます。
+
+```python
+    def create_link_message(self, link: str, save_as: str = '') -> ToolInvokeMessage:
+    """
+        create a link message
+
+        :param link: the url of the link
+        :param save_as: save as
+        :return: the link message
+    """
+```
+
+#### テキスト
+テキストメッセージを返す必要がある場合は、以下のインターフェースを使用できます。
+
+```python
+    def create_text_message(self, text: str, save_as: str = '') -> ToolInvokeMessage:
+    """
+        create a text message
+
+        :param text: the text of the message
+        :param save_as: save as
+        :return: the text message
+    """
+```
+
+#### ファイルBLOB
+画像、音声、動画、PPT、Word、Excelなどのファイルの生データを返す必要がある場合は、以下のインターフェースを使用できます。
+
+- `blob` ファイルの生データ（bytes型）
+- `meta` ファイルのメタデータ。ファイルの種類が分かっている場合は、`mime_type`を渡すことをお勧めします。そうでない場合、Difyはデフォルトタイプとして`octet/stream`を使用します。
+
+```python
+    def create_blob_message(self, blob: bytes, meta: dict = None, save_as: str = '') -> ToolInvokeMessage:
+    """
+        create a blob message
+
+        :param blob: the blob
+        :param meta: meta
+        :param save_as: save as
+        :return: the blob message
+    """
+```
+
+#### JSON
+フォーマットされたJSONを返す必要がある場合は、以下のインターフェースを使用できます。これは通常、ワークフロー内のノード間のデータ伝送に使用されますが、エージェントモードでは、ほとんどの大規模言語モデルもJSONを読み取り、理解することができます。
+
+- `object` Pythonの辞書オブジェクトで、自動的にJSONにシリアライズされます。
+
+```python
+    def create_json_message(self, object: dict) -> ToolInvokeMessage:
+    """
+        create a json message
+    """
+```
+
+### ショートカットツール
+
+大規模モデルアプリケーションでは、以下の2つの一般的なニーズがあります：
+- まず長いテキストを事前に要約し、その要約内容をLLMに渡すことで、元のテキストが長すぎてLLMが処理できない問題を防ぐ
+- ツールが取得したコンテンツがリンクである場合、Webページ情報をクロールしてからLLMに返す必要がある
+
+開発者がこれら2つのニーズを迅速に実装できるよう、以下の2つのショートカットツールを提供しています。
+
+#### テキスト要約ツール
+
+このツールはuser_idと要約するテキストを入力として受け取り、要約されたテキストを返します。Difyは現在のワークスペースのデフォルトモデルを使用して長文を要約します。
+
+```python
+    def summary(self, user_id: str, content: str) -> str:
+    """
+        summary the content
+
+        :param user_id: the user id
+        :param content: the content
+        :return: the summary
+    """
+```
+
+#### Webページクローリングツール
+
+このツールはクロールするWebページのリンクとユーザーエージェント（空でも可）を入力として受け取り、そのWebページの情報を含む文字列を返します。`user_agent`はオプションのパラメータで、ツールを識別するために使用できます。渡さない場合、Difyはデフォルトの`user_agent`を使用します。
+
+```python
+    def get_url(self, url: str, user_agent: str = None) -> str:
+    """
+        get url from the crawled result
+    """ 
+```
+
+### 変数プール
+
+`Tool`内に変数プールを導入し、ツールの実行中に生成された変数やファイルなどを保存します。これらの変数は、ツールの実行中に他のツールが使用することができます。
+
+次に、`DallE3`と`Vectorizer.AI`を例に、変数プールの使用方法を紹介します。
+
+- `DallE3`は画像生成ツールで、テキストに基づいて画像を生成できます。ここでは、`DallE3`にカフェのロゴを生成させます。
+- `Vectorizer.AI`はベクター画像変換ツールで、画像をベクター画像に変換できるため、画像を無限に拡大しても品質が損なわれません。ここでは、`DallE3`が生成したPNGアイコンをベクター画像に変換し、デザイナーが実際に使用できるようにします。
+
+#### DallE3
+まず、DallE3を使用します。画像を作成した後、その画像を変数プールに保存します。コードは以下の通りです：
+
+```python
+from typing import Any, Dict, List, Union
+from core.tools.entities.tool_entities import ToolInvokeMessage
+from core.tools.tool.builtin_tool import BuiltinTool
+
+from base64 import b64decode
+
+from openai import OpenAI
+
+class DallE3Tool(BuiltinTool):
+    def _invoke(self,
+                user_id: str,
+                tool_parameters: Dict[str, Any],
+                ) -> Union[ToolInvokeMessage, List[ToolInvokeMessage]]:
+        """
+            invoke tools
+        """
+        client = OpenAI(
+            api_key=self.runtime.credentials['openai_api_key'],
+        )
+
+        # prompt
+        prompt = tool_parameters.get('prompt', '')
+        if not prompt:
+            return self.create_text_message('Please input prompt')
+
+        # call openapi dalle3
+        response = client.images.generate(
+            prompt=prompt, model='dall-e-3',
+            size='1024x1024', n=1, style='vivid', quality='standard',
+            response_format='b64_json'
+        )
+
+        result = []
+        for image in response.data:
+            # Save all images to the variable pool through the save_as parameter. The variable name is self.VARIABLE_KEY.IMAGE.value. If new images are generated later, they will overwrite the previous images.
+            result.append(self.create_blob_message(blob=b64decode(image.b64_json),
+                                                   meta={ 'mime_type': 'image/png' },
+                                                   save_as=self.VARIABLE_KEY.IMAGE.value))
+
+        return result
+```
+
+ここでは画像の変数名として`self.VARIABLE_KEY.IMAGE.value`を使用していることに注意してください。開発者のツールが互いに連携できるよう、この`KEY`を定義しました。自由に使用することも、この`KEY`を使用しないこともできます。カスタムのKEYを渡すこともできます。
+
+#### Vectorizer.AI
+次に、Vectorizer.AIを使用して、DallE3が生成したPNGアイコンをベクター画像に変換します。ここで定義した関数を見てみましょう。コードは以下の通りです：
+
+```python
+from core.tools.tool.builtin_tool import BuiltinTool
+from core.tools.entities.tool_entities import ToolInvokeMessage, ToolParameter
+from core.tools.errors import ToolProviderCredentialValidationError
+
+from typing import Any, Dict, List, Union
+from httpx import post
+from base64 import b64decode
+
+class VectorizerTool(BuiltinTool):
+    def _invoke(self, user_id: str, tool_parameters: Dict[str, Any])
+        -> Union[ToolInvokeMessage, List[ToolInvokeMessage]]:
+        """
+        Tool invocation, the image variable name needs to be passed in from here, so that we can get the image from the variable pool
+        """
+
+
+    def get_runtime_parameters(self) -> List[ToolParameter]:
+        """
+        Override the tool parameter list, we can dynamically generate the parameter list based on the actual situation in the current variable pool, so that the LLM can generate the form based on the parameter list
+        """
+
+
+    def is_tool_available(self) -> bool:
+        """
+        Whether the current tool is available, if there is no image in the current variable pool, then we don't need to display this tool, just return False here
+        """     
+```
+
+次に、これら3つの関数を実装します：
+
+```python
+from core.tools.tool.builtin_tool import BuiltinTool
+from core.tools.entities.tool_entities import ToolInvokeMessage, ToolParameter
+from core.tools.errors import ToolProviderCredentialValidationError
+
+from typing import Any, Dict, List, Union
+from httpx import post
+from base64 import b64decode
+
+class VectorizerTool(BuiltinTool):
+    def _invoke(self, user_id: str, tool_parameters: Dict[str, Any])
+        -> Union[ToolInvokeMessage, List[ToolInvokeMessage]]:
+        """
+            invoke tools
+        """
+        api_key_name = self.runtime.credentials.get('api_key_name', None)
+        api_key_value = self.runtime.credentials.get('api_key_value', None)
+
+        if not api_key_name or not api_key_value:
+            raise ToolProviderCredentialValidationError('Please input api key name and value')
+
+        # Get image_id, the definition of image_id can be found in get_runtime_parameters
+        image_id = tool_parameters.get('image_id', '')
+        if not image_id:
+            return self.create_text_message('Please input image id')
+
+        # Get the image generated by DallE from the variable pool
+        image_binary = self.get_variable_file(self.VARIABLE_KEY.IMAGE)
+        if not image_binary:
+            return self.create_text_message('Image not found, please request user to generate image firstly.')
+
+        # Generate vector image
+        response = post(
+            'https://vectorizer.ai/api/v1/vectorize',
+            files={ 'image': image_binary },
+            data={ 'mode': 'test' },
+            auth=(api_key_name, api_key_value),
+            timeout=30
+        )
+
+        if response.status_code != 200:
+            raise Exception(response.text)
+
+        return [
+            self.create_text_message('the vectorized svg is saved as an image.'),
+            self.create_blob_message(blob=response.content,
+                                     meta={'mime_type': 'image/svg+xml'})
+        ]
+
+    def get_runtime_parameters(self) -> List[ToolParameter]:
+        """
+        override the runtime parameters
+        """
+        # Here, we override the tool parameter list, define the image_id, and set its option list to all images in the current variable pool. The configuration here is consistent with the configuration in yaml.
+        return [
+            ToolParameter.get_simple_instance(
+                name='image_id',
+                llm_description=f'the image id that you want to vectorize, \
+                    and the image id should be specified in \
+                        {[i.name for i in self.list_default_image_variables()]}',
+                type=ToolParameter.ToolParameterType.SELECT,
+                required=True,
+                options=[i.name for i in self.list_default_image_variables()]
+            )
+        ]
+
+    def is_tool_available(self) -> bool:
+        # Only when there are images in the variable pool, the LLM needs to use this tool
+        return len(self.list_default_image_variables()) > 0
+```
+
+ここで注目すべきは、実際には`image_id`を使用していないことです。このツールを呼び出す際には、デフォルトの変数プールに必ず画像があると仮定し、直接`image_binary = self.get_variable_file(self.VARIABLE_KEY.IMAGE)`を使用して画像を取得しています。モデルの能力が弱い場合、開発者にもこの方法を推奨します。これにより、エラー許容度を効果的に向上させ、モデルが誤ったパラメータを渡すのを防ぐことができます。
\ No newline at end of file
diff --git a/api/core/tools/docs/ja_JP/tool_scale_out.md b/api/core/tools/docs/ja_JP/tool_scale_out.md
new file mode 100644
index 0000000000000..a721023d00bdd
--- /dev/null
+++ b/api/core/tools/docs/ja_JP/tool_scale_out.md
@@ -0,0 +1,240 @@
+# ツールの迅速な統合
+
+ここでは、GoogleSearchを例にツールを迅速に統合する方法を紹介します。
+
+## 1. ツールプロバイダーのyamlを準備する
+
+### 概要
+
+このyamlファイルには、プロバイダー名、アイコン、作者などの詳細情報が含まれ、フロントエンドでの柔軟な表示を可能にします。
+
+### 例
+
+`core/tools/provider/builtin`の下に`google`モジュール（フォルダ）を作成し、`google.yaml`を作成します。名前はモジュール名と一致している必要があります。
+
+以降、このツールに関するすべての操作はこのモジュール内で行います。
+
+```yaml
+identity: # ツールプロバイダーの基本情報
+  author: Dify # 作者
+  name: google # 名前（一意、他のプロバイダーと重複不可）
+  label: # フロントエンド表示用のラベル
+    en_US: Google # 英語ラベル
+    zh_Hans: Google # 中国語ラベル
+  description: # フロントエンド表示用の説明
+    en_US: Google # 英語説明
+    zh_Hans: Google # 中国語説明
+  icon: icon.svg # アイコン（現在のモジュールの_assetsフォルダに配置）
+  tags: # タグ（フロントエンド表示用）
+    - search
+```
+
+- `identity`フィールドは必須で、ツールプロバイダーの基本情報（作者、名前、ラベル、説明、アイコンなど）が含まれます。
+  - アイコンは現在のモジュールの`_assets`フォルダに配置する必要があります。[こちら](../../provider/builtin/google/_assets/icon.svg)を参照してください。
+  - タグはフロントエンドでの表示に使用され、ユーザーがこのツールプロバイダーを素早く見つけるのに役立ちます。現在サポートされているすべてのタグは以下の通りです：
+    ```python
+    class ToolLabelEnum(Enum):
+      SEARCH = 'search'
+      IMAGE = 'image'
+      VIDEOS = 'videos'
+      WEATHER = 'weather'
+      FINANCE = 'finance'
+      DESIGN = 'design'
+      TRAVEL = 'travel'
+      SOCIAL = 'social'
+      NEWS = 'news'
+      MEDICAL = 'medical'
+      PRODUCTIVITY = 'productivity'
+      EDUCATION = 'education'
+      BUSINESS = 'business'
+      ENTERTAINMENT = 'entertainment'
+      UTILITIES = 'utilities'
+      OTHER = 'other'
+    ```
+
+## 2. プロバイダーの認証情報を準備する
+
+GoogleはSerpApiが提供するAPIを使用するサードパーティツールであり、SerpApiを使用するにはAPI Keyが必要です。つまり、このツールを使用するには認証情報が必要です。一方、`wikipedia`のようなツールでは認証情報フィールドを記入する必要はありません。[こちら](../../provider/builtin/wikipedia/wikipedia.yaml)を参照してください。
+
+認証情報フィールドを設定すると、以下のようになります：
+
+```yaml
+identity:
+  author: Dify
+  name: google
+  label:
+    en_US: Google
+    zh_Hans: Google
+  description:
+    en_US: Google
+    zh_Hans: Google
+  icon: icon.svg
+credentials_for_provider: # 認証情報フィールド
+  serpapi_api_key: # 認証情報フィールド名
+    type: secret-input # 認証情報フィールドタイプ
+    required: true # 必須かどうか
+    label: # 認証情報フィールドラベル
+      en_US: SerpApi API key # 英語ラベル
+      zh_Hans: SerpApi API key # 中国語ラベル
+    placeholder: # 認証情報フィールドプレースホルダー
+      en_US: Please input your SerpApi API key # 英語プレースホルダー
+      zh_Hans: 请输入你的 SerpApi API key # 中国語プレースホルダー
+    help: # 認証情報フィールドヘルプテキスト
+      en_US: Get your SerpApi API key from SerpApi # 英語ヘルプテキスト
+      zh_Hans: 从 SerpApi 获取您的 SerpApi API key # 中国語ヘルプテキスト
+    url: https://serpapi.com/manage-api-key # 認証情報フィールドヘルプリンク
+```
+
+- `type`：認証情報フィールドタイプ。現在、`secret-input`、`text-input`、`select`の3種類をサポートしており、それぞれパスワード入力ボックス、テキスト入力ボックス、ドロップダウンボックスに対応します。`secret-input`の場合、フロントエンドで入力内容が隠され、バックエンドで入力内容が暗号化されます。
+
+## 3. ツールのyamlを準備する
+
+1つのプロバイダーの下に複数のツールを持つことができ、各ツールにはyamlファイルが必要です。このファイルにはツールの基本情報、パラメータ、出力などが含まれます。
+
+引き続きGoogleSearchを例に、`google`モジュールの下に`tools`モジュールを作成し、`tools/google_search.yaml`を作成します。内容は以下の通りです：
+
+```yaml
+identity: # ツールの基本情報
+  name: google_search # ツール名（一意、他のツールと重複不可）
+  author: Dify # 作者
+  label: # フロントエンド表示用のラベル
+    en_US: GoogleSearch # 英語ラベル
+    zh_Hans: 谷歌搜索 # 中国語ラベル
+description: # フロントエンド表示用の説明
+  human: # フロントエンド表示用の紹介（多言語対応）
+    en_US: A tool for performing a Google SERP search and extracting snippets and webpages. Input should be a search query.
+    zh_Hans: 一个用于执行 Google SERP 搜索并提取片段和网页的工具。输入应该是一个搜索查询。
+  llm: A tool for performing a Google SERP search and extracting snippets and webpages. Input should be a search query. # LLMに渡す紹介文。LLMがこのツールをより理解できるよう、できるだけ詳細な情報を記述することをお勧めします。
+parameters: # パラメータリスト
+  - name: query # パラメータ名
+    type: string # パラメータタイプ
+    required: true # 必須かどうか
+    label: # パラメータラベル
+      en_US: Query string # 英語ラベル
+      zh_Hans: 查询语句 # 中国語ラベル
+    human_description: # フロントエンド表示用の紹介（多言語対応）
+      en_US: used for searching
+      zh_Hans: 用于搜索网页内容
+    llm_description: key words for searching # LLMに渡す紹介文。LLMがこのパラメータをより理解できるよう、できるだけ詳細な情報を記述することをお勧めします。
+    form: llm # フォームタイプ。llmはこのパラメータがAgentによって推論される必要があることを意味し、フロントエンドではこのパラメータは表示されません。
+  - name: result_type
+    type: select # パラメータタイプ
+    required: true
+    options: # ドロップダウンボックスのオプション
+      - value: text
+        label:
+          en_US: text
+          zh_Hans: 文本
+      - value: link
+        label:
+          en_US: link
+          zh_Hans: 链接
+    default: link
+    label:
+      en_US: Result type
+      zh_Hans: 结果类型
+    human_description:
+      en_US: used for selecting the result type, text or link
+      zh_Hans: 用于选择结果类型，使用文本还是链接进行展示
+    form: form # フォームタイプ。formはこのパラメータが対話開始前にフロントエンドでユーザーによって入力される必要があることを意味します。
+```
+
+- `identity`フィールドは必須で、ツールの基本情報（名前、作者、ラベル、説明など）が含まれます。
+- `parameters` パラメータリスト
+  - `name`（必須）パラメータ名。一意で、他のパラメータと重複しないようにしてください。
+  - `type`（必須）パラメータタイプ。現在、`string`、`number`、`boolean`、`select`、`secret-input`の5種類をサポートしており、それぞれ文字列、数値、ブール値、ドロップダウンボックス、暗号化入力ボックスに対応します。機密情報には`secret-input`タイプの使用をお勧めします。
+  - `label`（必須）パラメータラベル。フロントエンド表示用です。
+  - `form`（必須）フォームタイプ。現在、`llm`と`form`の2種類をサポートしています。
+    - エージェントアプリケーションでは、`llm`はこのパラメータがLLM自身によって推論されることを示し、`form`はこのツールを使用するために事前に設定できるパラメータであることを示します。
+    - ワークフローアプリケーションでは、`llm`と`form`の両方がフロントエンドで入力する必要がありますが、`llm`のパラメータはツールノードの入力変数として使用されます。
+  - `required` パラメータが必須かどうかを示します。
+    - `llm`モードでは、パラメータが必須の場合、Agentはこのパラメータを推論する必要があります。
+    - `form`モードでは、パラメータが必須の場合、ユーザーは対話開始前にフロントエンドでこのパラメータを入力する必要があります。
+  - `options` パラメータオプション
+    - `llm`モードでは、DifyはすべてのオプションをLLMに渡し、LLMはこれらのオプションに基づいて推論できます。
+    - `form`モードで、`type`が`select`の場合、フロントエンドはこれらのオプションを表示します。
+  - `default` デフォルト値
+  - `min` 最小値。パラメータタイプが`number`の場合に設定できます。
+  - `max` 最大値。パラメータタイプが`number`の場合に設定できます。
+  - `human_description` フロントエンド表示用の紹介。多言語対応です。
+  - `placeholder` 入力ボックスのプロンプトテキスト。フォームタイプが`form`で、パラメータタイプが`string`、`number`、`secret-input`の場合に設定できます。多言語対応です。
+  - `llm_description` LLMに渡す紹介文。LLMがこのパラメータをより理解できるよう、できるだけ詳細な情報を記述することをお勧めします。
+
+## 4. ツールコードを準備する
+
+ツールの設定が完了したら、ツールのロジックを実装するコードを作成します。
+
+`google/tools`モジュールの下に`google_search.py`を作成し、内容は以下の通りです：
+
+```python
+from core.tools.tool.builtin_tool import BuiltinTool
+from core.tools.entities.tool_entities import ToolInvokeMessage
+
+from typing import Any, Dict, List, Union
+
+class GoogleSearchTool(BuiltinTool):
+    def _invoke(self, 
+                user_id: str,
+               tool_parameters: Dict[str, Any], 
+        ) -> Union[ToolInvokeMessage, List[ToolInvokeMessage]]:
+        """
+            ツールを呼び出す
+        """
+        query = tool_parameters['query']
+        result_type = tool_parameters['result_type']
+        api_key = self.runtime.credentials['serpapi_api_key']
+        result = SerpAPI(api_key).run(query, result_type=result_type)
+
+        if result_type == 'text':
+            return self.create_text_message(text=result)
+        return self.create_link_message(link=result)
+```
+
+### パラメータ
+ツールの全体的なロジックは`_invoke`メソッドにあります。このメソッドは2つのパラメータ（`user_id`とtool_parameters`）を受け取り、それぞれユーザーIDとツールパラメータを表します。
+
+### 戻り値
+ツールの戻り値として、1つのメッセージまたは複数のメッセージを選択できます。ここでは1つのメッセージを返しています。`create_text_message`と`create_link_message`を使用して、テキストメッセージまたはリンクメッセージを作成できます。複数のメッセージを返す場合は、リストを構築できます（例：`[self.create_text_message('msg1'), self.create_text_message('msg2')]`）。
+
+## 5. プロバイダーコードを準備する
+
+最後に、プロバイダーモジュールの下にプロバイダークラスを作成し、プロバイダーの認証情報検証ロジックを実装する必要があります。認証情報の検証が失敗した場合、`ToolProviderCredentialValidationError`例外が発生します。
+
+`google`モジュールの下に`google.py`を作成し、内容は以下の通りです：
+
+```python
+from core.tools.provider.builtin_tool_provider import BuiltinToolProviderController
+from core.tools.errors import ToolProviderCredentialValidationError
+
+from core.tools.provider.builtin.google.tools.google_search import GoogleSearchTool
+
+from typing import Any, Dict
+
+class GoogleProvider(BuiltinToolProviderController):
+    def _validate_credentials(self, credentials: Dict[str, Any]) -> None:
+        try:
+            # 1. ここでGoogleSearchTool()を使ってGoogleSearchToolをインスタンス化する必要があります。これによりGoogleSearchToolのyaml設定が自動的に読み込まれますが、この時点では認証情報は含まれていません
+            # 2. 次に、fork_tool_runtimeメソッドを使用して、現在の認証情報をGoogleSearchToolに渡す必要があります
+            # 3. 最後に、invokeを呼び出します。パラメータはGoogleSearchToolのyamlで設定されたパラメータルールに従って渡す必要があります
+            GoogleSearchTool().fork_tool_runtime(
+                meta={
+                    "credentials": credentials,
+                }
+            ).invoke(
+                user_id='',
+                tool_parameters={
+                    "query": "test",
+                    "result_type": "link"
+                },
+            )
+        except Exception as e:
+            raise ToolProviderCredentialValidationError(str(e))
+```
+
+## 完了
+
+以上のステップが完了すると、このツールをフロントエンドで確認し、Agentで使用することができるようになります。
+
+もちろん、google_searchには認証情報が必要なため、使用する前にフロントエンドで認証情報を入力する必要があります。
+
+![Alt text](../images/index/image-2.png)
\ No newline at end of file
diff --git a/api/core/tools/docs/zh_Hans/tool_scale_out.md b/api/core/tools/docs/zh_Hans/tool_scale_out.md
index 06a8d9a4f9a9d..ec61e4677bae7 100644
--- a/api/core/tools/docs/zh_Hans/tool_scale_out.md
+++ b/api/core/tools/docs/zh_Hans/tool_scale_out.md
@@ -234,4 +234,4 @@ class GoogleProvider(BuiltinToolProviderController):
 
 当然，因为google_search需要一个凭据，在使用之前，还需要在前端配置它的凭据。
 
-![Alt text](images/index/image-2.png)
+![Alt text](../images/index/image-2.png)

From c8b9bdebfe88e26a13abf36330420b77c3106b16 Mon Sep 17 00:00:00 2001
From: =?UTF-8?q?=E5=91=86=E8=90=8C=E9=97=B7=E6=B2=B9=E7=93=B6?=
 <253605712@qq.com>
Date: Sun, 22 Sep 2024 10:08:35 +0800
Subject: [PATCH 014/235] feat:use xinference tts stream mode (#8616)

---
 .../model_providers/xinference/llm/llm.py            |  3 +--
 .../model_providers/xinference/tts/tts.py            | 12 ++++++------
 api/poetry.lock                                      |  8 ++++----
 api/pyproject.toml                                   |  2 +-
 .../model_runtime/__mock/xinference.py               |  5 +----
 5 files changed, 13 insertions(+), 17 deletions(-)

diff --git a/api/core/model_runtime/model_providers/xinference/llm/llm.py b/api/core/model_runtime/model_providers/xinference/llm/llm.py
index 4fadda5df5d85..286640079b02a 100644
--- a/api/core/model_runtime/model_providers/xinference/llm/llm.py
+++ b/api/core/model_runtime/model_providers/xinference/llm/llm.py
@@ -19,7 +19,6 @@
 from openai.types.completion import Completion
 from xinference_client.client.restful.restful_client import (
     Client,
-    RESTfulChatglmCppChatModelHandle,
     RESTfulChatModelHandle,
     RESTfulGenerateModelHandle,
 )
@@ -491,7 +490,7 @@ def _generate(
         if tools and len(tools) > 0:
             generate_config["tools"] = [{"type": "function", "function": helper.dump_model(tool)} for tool in tools]
         vision = credentials.get("support_vision", False)
-        if isinstance(xinference_model, RESTfulChatModelHandle | RESTfulChatglmCppChatModelHandle):
+        if isinstance(xinference_model, RESTfulChatModelHandle):
             resp = client.chat.completions.create(
                 model=credentials["model_uid"],
                 messages=[self._convert_prompt_message_to_dict(message) for message in prompt_messages],
diff --git a/api/core/model_runtime/model_providers/xinference/tts/tts.py b/api/core/model_runtime/model_providers/xinference/tts/tts.py
index 10538b5788a3c..81dbe397d2f10 100644
--- a/api/core/model_runtime/model_providers/xinference/tts/tts.py
+++ b/api/core/model_runtime/model_providers/xinference/tts/tts.py
@@ -208,21 +208,21 @@ def _tts_invoke_streaming(self, model: str, credentials: dict, content_text: str
                 executor = concurrent.futures.ThreadPoolExecutor(max_workers=min(3, len(sentences)))
                 futures = [
                     executor.submit(
-                        handle.speech, input=sentences[i], voice=voice, response_format="mp3", speed=1.0, stream=False
+                        handle.speech, input=sentences[i], voice=voice, response_format="mp3", speed=1.0, stream=True
                     )
                     for i in range(len(sentences))
                 ]
 
                 for future in futures:
                     response = future.result()
-                    for i in range(0, len(response), 1024):
-                        yield response[i : i + 1024]
+                    for chunk in response:
+                        yield chunk
             else:
                 response = handle.speech(
-                    input=content_text.strip(), voice=voice, response_format="mp3", speed=1.0, stream=False
+                    input=content_text.strip(), voice=voice, response_format="mp3", speed=1.0, stream=True
                 )
 
-                for i in range(0, len(response), 1024):
-                    yield response[i : i + 1024]
+                for chunk in response:
+                    yield chunk
         except Exception as ex:
             raise InvokeBadRequestError(str(ex))
diff --git a/api/poetry.lock b/api/poetry.lock
index 8d4d680031a7c..e532394a24e74 100644
--- a/api/poetry.lock
+++ b/api/poetry.lock
@@ -10014,13 +10014,13 @@ h11 = ">=0.9.0,<1"
 
 [[package]]
 name = "xinference-client"
-version = "0.13.3"
+version = "0.15.2"
 description = "Client for Xinference"
 optional = false
 python-versions = "*"
 files = [
-    {file = "xinference-client-0.13.3.tar.gz", hash = "sha256:822b722100affdff049c27760be7d62ac92de58c87a40d3361066df446ba648f"},
-    {file = "xinference_client-0.13.3-py3-none-any.whl", hash = "sha256:f0eff3858b1ebcef2129726f82b09259c177e11db466a7ca23def3d4849c419f"},
+    {file = "xinference-client-0.15.2.tar.gz", hash = "sha256:5c2259bb133148d1cc9bd2b8ec6eb8b5bbeba7f11d6252959f4e6cd79baa53ed"},
+    {file = "xinference_client-0.15.2-py3-none-any.whl", hash = "sha256:b6275adab695e75e75a33e21e0ad212488fc2d5a4d0f693d544c0e78469abbe3"},
 ]
 
 [package.dependencies]
@@ -10422,4 +10422,4 @@ cffi = ["cffi (>=1.11)"]
 [metadata]
 lock-version = "2.0"
 python-versions = ">=3.10,<3.13"
-content-hash = "18924ae12a00bde4438a46168bc167ed69613ab1ab0c387f193cd47ac24379b2"
+content-hash = "85aa4be7defee8fe6622cf95ba03e81895121502ebf6d666d6ce376ff019fac7"
diff --git a/api/pyproject.toml b/api/pyproject.toml
index 829c39a12bf91..1f483fc49f032 100644
--- a/api/pyproject.toml
+++ b/api/pyproject.toml
@@ -203,7 +203,7 @@ transformers = "~4.35.0"
 unstructured = { version = "~0.10.27", extras = ["docx", "epub", "md", "msg", "ppt", "pptx"] }
 websocket-client = "~1.7.0"
 werkzeug = "~3.0.1"
-xinference-client = "0.13.3"
+xinference-client = "0.15.2"
 yarl = "~1.9.4"
 zhipuai = "1.0.7"
 # Before adding new dependency, consider place it in alphabet order (a-z) and suitable group.
diff --git a/api/tests/integration_tests/model_runtime/__mock/xinference.py b/api/tests/integration_tests/model_runtime/__mock/xinference.py
index 299523f4f5b7c..8deb50635f3d5 100644
--- a/api/tests/integration_tests/model_runtime/__mock/xinference.py
+++ b/api/tests/integration_tests/model_runtime/__mock/xinference.py
@@ -9,7 +9,6 @@
 from requests.sessions import Session
 from xinference_client.client.restful.restful_client import (
     Client,
-    RESTfulChatglmCppChatModelHandle,
     RESTfulChatModelHandle,
     RESTfulEmbeddingModelHandle,
     RESTfulGenerateModelHandle,
@@ -19,9 +18,7 @@
 
 
 class MockXinferenceClass:
-    def get_chat_model(
-        self: Client, model_uid: str
-    ) -> Union[RESTfulChatglmCppChatModelHandle, RESTfulGenerateModelHandle, RESTfulChatModelHandle]:
+    def get_chat_model(self: Client, model_uid: str) -> Union[RESTfulGenerateModelHandle, RESTfulChatModelHandle]:
         if not re.match(r"https?:\/\/[^\s\/$.?#].[^\s]*$", self.base_url):
             raise RuntimeError("404 Not Found")
 

From 0665268578599ca340ff134eec041f17364ef937 Mon Sep 17 00:00:00 2001
From: ice yao <yao3690093@gmail.com>
Date: Sun, 22 Sep 2024 10:13:00 +0800
Subject: [PATCH 015/235] Add Fireworks AI as new model provider (#8428)

---
 .../model_providers/_position.yaml            |   1 +
 .../model_providers/fireworks/__init__.py     |   0
 .../fireworks/_assets/icon_l_en.svg           |   3 +
 .../fireworks/_assets/icon_s_en.svg           |   5 +
 .../model_providers/fireworks/_common.py      |  52 ++
 .../model_providers/fireworks/fireworks.py    |  27 +
 .../model_providers/fireworks/fireworks.yaml  |  29 +
 .../model_providers/fireworks/llm/__init__.py |   0
 .../fireworks/llm/_position.yaml              |  16 +
 .../fireworks/llm/firefunction-v1.yaml        |  46 ++
 .../fireworks/llm/firefunction-v2.yaml        |  46 ++
 .../fireworks/llm/gemma2-9b-it.yaml           |  45 ++
 .../llm/llama-v3-70b-instruct-hf.yaml         |  46 ++
 .../fireworks/llm/llama-v3-70b-instruct.yaml  |  46 ++
 .../llm/llama-v3-8b-instruct-hf.yaml          |  46 ++
 .../fireworks/llm/llama-v3-8b-instruct.yaml   |  46 ++
 .../llm/llama-v3p1-405b-instruct.yaml         |  46 ++
 .../llm/llama-v3p1-70b-instruct.yaml          |  46 ++
 .../fireworks/llm/llama-v3p1-8b-instruct.yaml |  46 ++
 .../model_providers/fireworks/llm/llm.py      | 610 ++++++++++++++++++
 .../fireworks/llm/mixtral-8x22b-instruct.yaml |  46 ++
 .../llm/mixtral-8x7b-instruct-hf.yaml         |  46 ++
 .../fireworks/llm/mixtral-8x7b-instruct.yaml  |  46 ++
 .../fireworks/llm/mythomax-l2-13b.yaml        |  46 ++
 .../llm/phi-3-vision-128k-instruct.yaml       |  46 ++
 .../fireworks/llm/yi-large.yaml               |  45 ++
 api/pyproject.toml                            |   1 +
 .../model_runtime/fireworks/__init__.py       |   0
 .../model_runtime/fireworks/test_llm.py       | 186 ++++++
 .../model_runtime/fireworks/test_provider.py  |  17 +
 dev/pytest/pytest_model_runtime.sh            |   4 +-
 31 files changed, 1683 insertions(+), 2 deletions(-)
 create mode 100644 api/core/model_runtime/model_providers/fireworks/__init__.py
 create mode 100644 api/core/model_runtime/model_providers/fireworks/_assets/icon_l_en.svg
 create mode 100644 api/core/model_runtime/model_providers/fireworks/_assets/icon_s_en.svg
 create mode 100644 api/core/model_runtime/model_providers/fireworks/_common.py
 create mode 100644 api/core/model_runtime/model_providers/fireworks/fireworks.py
 create mode 100644 api/core/model_runtime/model_providers/fireworks/fireworks.yaml
 create mode 100644 api/core/model_runtime/model_providers/fireworks/llm/__init__.py
 create mode 100644 api/core/model_runtime/model_providers/fireworks/llm/_position.yaml
 create mode 100644 api/core/model_runtime/model_providers/fireworks/llm/firefunction-v1.yaml
 create mode 100644 api/core/model_runtime/model_providers/fireworks/llm/firefunction-v2.yaml
 create mode 100644 api/core/model_runtime/model_providers/fireworks/llm/gemma2-9b-it.yaml
 create mode 100644 api/core/model_runtime/model_providers/fireworks/llm/llama-v3-70b-instruct-hf.yaml
 create mode 100644 api/core/model_runtime/model_providers/fireworks/llm/llama-v3-70b-instruct.yaml
 create mode 100644 api/core/model_runtime/model_providers/fireworks/llm/llama-v3-8b-instruct-hf.yaml
 create mode 100644 api/core/model_runtime/model_providers/fireworks/llm/llama-v3-8b-instruct.yaml
 create mode 100644 api/core/model_runtime/model_providers/fireworks/llm/llama-v3p1-405b-instruct.yaml
 create mode 100644 api/core/model_runtime/model_providers/fireworks/llm/llama-v3p1-70b-instruct.yaml
 create mode 100644 api/core/model_runtime/model_providers/fireworks/llm/llama-v3p1-8b-instruct.yaml
 create mode 100644 api/core/model_runtime/model_providers/fireworks/llm/llm.py
 create mode 100644 api/core/model_runtime/model_providers/fireworks/llm/mixtral-8x22b-instruct.yaml
 create mode 100644 api/core/model_runtime/model_providers/fireworks/llm/mixtral-8x7b-instruct-hf.yaml
 create mode 100644 api/core/model_runtime/model_providers/fireworks/llm/mixtral-8x7b-instruct.yaml
 create mode 100644 api/core/model_runtime/model_providers/fireworks/llm/mythomax-l2-13b.yaml
 create mode 100644 api/core/model_runtime/model_providers/fireworks/llm/phi-3-vision-128k-instruct.yaml
 create mode 100644 api/core/model_runtime/model_providers/fireworks/llm/yi-large.yaml
 create mode 100644 api/tests/integration_tests/model_runtime/fireworks/__init__.py
 create mode 100644 api/tests/integration_tests/model_runtime/fireworks/test_llm.py
 create mode 100644 api/tests/integration_tests/model_runtime/fireworks/test_provider.py

diff --git a/api/core/model_runtime/model_providers/_position.yaml b/api/core/model_runtime/model_providers/_position.yaml
index d10314ba039e6..1f5f64019a166 100644
--- a/api/core/model_runtime/model_providers/_position.yaml
+++ b/api/core/model_runtime/model_providers/_position.yaml
@@ -37,3 +37,4 @@
 - siliconflow
 - perfxcloud
 - zhinao
+- fireworks
diff --git a/api/core/model_runtime/model_providers/fireworks/__init__.py b/api/core/model_runtime/model_providers/fireworks/__init__.py
new file mode 100644
index 0000000000000..e69de29bb2d1d
diff --git a/api/core/model_runtime/model_providers/fireworks/_assets/icon_l_en.svg b/api/core/model_runtime/model_providers/fireworks/_assets/icon_l_en.svg
new file mode 100644
index 0000000000000..582605cc422cc
--- /dev/null
+++ b/api/core/model_runtime/model_providers/fireworks/_assets/icon_l_en.svg
@@ -0,0 +1,3 @@
+<svg width="130" role="graphics-symbol" aria-label="Fireworks AI Home" viewBox="0 0 835 130" xmlns="http://www.w3.org/2000/svg">
+<path fill-rule="evenodd" clip-rule="evenodd" d="M112.65 0L91.33 51.09L69.99 0H56.3L79.69 55.85C81.63 60.51 86.18 63.52 91.25 63.52C96.32 63.52 100.86 60.51 102.81 55.87L126.34 0H112.65ZM121.76 77.84L160.76 38.41L155.44 25.86L112.84 69.01C109.28 72.62 108.26 77.94 110.23 82.6C112.19 87.22 116.72 90.21 121.77 90.21L121.79 90.23L182.68 90.08L177.36 77.53L121.77 77.84H121.76ZM21.92 38.38L27.24 25.83L69.84 68.98C73.4 72.58 74.43 77.92 72.45 82.57C70.49 87.2 65.94 90.18 60.91 90.18L0.02 90.04L0 90.06L5.32 77.51L60.91 77.82L21.92 38.38Z" fill="#6720FF"></path>
+<path d="M231.32 85.2198L231.33 85.2298H241.8V49.1698H275.62V39.8198H241.8V16.3598H279V7.00977H231.32V85.2198Z" class="fill-black dark:fill-white"></path><path d="M299.68 28.73H289.86V85.22H299.68V28.73Z" class="fill-black dark:fill-white"></path><path d="M324.58 36.2198H324.59C324.37 36.7598 324.16 37.0898 323.5 37.0898C322.95 37.0898 322.74 36.8798 322.74 36.3398V28.7298H312.92V85.2198H322.72V53.1598C322.72 42.3098 327.75 38.0698 337.24 38.0698H345.1V28.5098H338.77C331.03 28.5098 327.1 30.7898 324.58 36.2198Z" class="fill-black dark:fill-white"></path><path d="M377.76 78.3996C367.23 78.3996 359.37 72.4196 358.71 59.7196H404.6V54.2796C404.6 38.5296 395 27.1196 377.53 27.1196C360.06 27.1196 348.93 38.5296 348.93 56.9896C348.93 75.4496 359.73 86.8596 377.74 86.8596C395.75 86.8596 403.15 75.8996 404.81 67.3196H394.57C392.98 73.7396 388.29 78.3996 377.76 78.3996ZM377.53 35.5696C387.91 35.5696 394.33 41.1196 394.78 51.5496H358.98C360.61 40.8896 368.14 35.5696 377.53 35.5696Z" class="fill-black dark:fill-white"></path><path d="M474.29 74.68C474.05 75.66 473.75 75.99 472.97 75.99C472.19 75.99 471.86 75.66 471.65 74.68L460.73 28.73H443.81L432.89 74.68C432.65 75.66 432.35 75.99 431.57 75.99C430.79 75.99 430.46 75.66 430.25 74.68L419.33 28.73H409.73V30.91H409.79L423.11 85.22H439.97L451.22 37.85C451.43 37.08 451.64 36.87 452.3 36.87C452.84 36.87 453.17 37.1 453.38 37.85L464.63 85.22H481.49L494.81 30.91V28.73H485.21L474.29 74.68Z" class="fill-black dark:fill-white"></path><path d="M529.05 27.1099C512.56 27.1099 499.47 37.4199 499.47 56.9799C499.47 76.5399 512.55 86.8499 529.05 86.8499C545.55 86.8499 558.64 76.5399 558.64 56.9799C558.64 37.4199 545.54 27.1099 529.05 27.1099ZM529.07 78.1599C517.61 78.1599 509.42 70.5699 509.42 56.9799C509.42 43.3899 517.61 35.7999 529.07 35.7999C540.53 35.7999 548.72 43.4099 548.72 56.9799C548.72 70.5499 540.53 78.1599 529.07 78.1599Z" class="fill-black dark:fill-white"></path><path d="M580.68 36.2198C580.47 36.7598 580.26 37.0898 579.6 37.0898C579.05 37.0898 578.841 36.8798 578.841 36.3398V28.7298H569.021V85.2098H578.82V53.1598C578.82 42.3098 583.851 38.0698 593.341 38.0698H601.201V28.5098H594.87C587.13 28.5098 583.2 30.7898 580.68 36.2198Z" class="fill-black dark:fill-white"></path><path d="M618.591 55.0198V7.00977H608.771V85.2698H618.591V67.2298L629.24 58.1498L650.42 85.2498H661.16V83.0698L636.49 51.9398L661.16 30.9098V28.7298H648.54L618.591 55.0198Z" class="fill-black dark:fill-white"></path><path d="M695.19 52.8899L687.12 51.3699C679.38 49.8999 675.99 48.2799 675.99 43.5999C675.99 38.9199 679.82 35.4499 688.98 35.4499C698.14 35.4499 703.38 38.9399 704.14 46.6499H714.14C713.03 32.8799 702.34 27.1299 688.94 27.1299C675.54 27.1299 666.13 32.8899 666.13 43.7399C666.13 54.5899 673.83 58.3499 684.91 60.4099L692.98 61.9299C700.84 63.3999 704.77 65.0899 704.77 69.9699C704.77 74.8499 700.83 78.4899 691.35 78.4899C681.87 78.4899 675.58 74.5799 674.82 67.0799H664.83C665.76 80.5499 676.73 86.8499 691.36 86.8499C705.99 86.8499 714.61 80.6099 714.61 69.4099C714.61 58.2099 705.55 54.8399 695.19 52.8899Z" class="fill-black dark:fill-white"></path><path d="M834.64 7.00977H823.63V85.2698H834.64V7.00977Z" class="fill-black dark:fill-white"></path><path d="M770.23 7.77L739.71 83.8398V85.2698H750.61L758.34 64.8398H795.08L802.81 85.2698H814.04V83.8598L783.3 7.00977H770.23ZM761.97 55.3798L775.09 21.0098H775.08C775.3 20.4198 775.87 20.0298 776.5 20.0298H777.04C777.67 20.0298 778.24 20.4198 778.46 21.0098L791.48 55.3798H761.97Z" class="fill-black dark:fill-white"></path><path d="M299.68 7.00977H289.86V18.5298H299.68V7.00977Z" class="fill-black dark:fill-white"></path></svg>
\ No newline at end of file
diff --git a/api/core/model_runtime/model_providers/fireworks/_assets/icon_s_en.svg b/api/core/model_runtime/model_providers/fireworks/_assets/icon_s_en.svg
new file mode 100644
index 0000000000000..86eeba66f9290
--- /dev/null
+++ b/api/core/model_runtime/model_providers/fireworks/_assets/icon_s_en.svg
@@ -0,0 +1,5 @@
+<svg width="638" height="315" viewBox="0 0 638 315" fill="none" xmlns="http://www.w3.org/2000/svg">
+<path d="M318.563 221.755C300.863 221.755 284.979 211.247 278.206 194.978L196.549 0H244.342L318.842 178.361L393.273 0H441.066L358.92 195.048C352.112 211.247 336.263 221.755 318.563 221.755Z" fill="#6720FF"/>
+<path d="M425.111 314.933C407.481 314.933 391.667 304.494 384.824 288.366C377.947 272.097 381.507 253.524 393.936 240.921L542.657 90.2803L561.229 134.094L425.076 271.748L619.147 270.666L637.72 314.479L425.146 315.003L425.076 314.933H425.111Z" fill="#6720FF"/>
+<path d="M0 314.408L18.5727 270.595L212.643 271.677L76.525 133.988L95.0977 90.1748L243.819 240.816C256.247 253.384 259.843 272.026 252.93 288.26C246.088 304.424 230.203 314.827 212.643 314.827L0.0698221 314.339L0 314.408Z" fill="#6720FF"/>
+</svg>
diff --git a/api/core/model_runtime/model_providers/fireworks/_common.py b/api/core/model_runtime/model_providers/fireworks/_common.py
new file mode 100644
index 0000000000000..378ced3a4019b
--- /dev/null
+++ b/api/core/model_runtime/model_providers/fireworks/_common.py
@@ -0,0 +1,52 @@
+from collections.abc import Mapping
+
+import openai
+
+from core.model_runtime.errors.invoke import (
+    InvokeAuthorizationError,
+    InvokeBadRequestError,
+    InvokeConnectionError,
+    InvokeError,
+    InvokeRateLimitError,
+    InvokeServerUnavailableError,
+)
+
+
+class _CommonFireworks:
+    def _to_credential_kwargs(self, credentials: Mapping) -> dict:
+        """
+        Transform credentials to kwargs for model instance
+
+        :param credentials:
+        :return:
+        """
+        credentials_kwargs = {
+            "api_key": credentials["fireworks_api_key"],
+            "base_url": "https://api.fireworks.ai/inference/v1",
+            "max_retries": 1,
+        }
+
+        return credentials_kwargs
+
+    @property
+    def _invoke_error_mapping(self) -> dict[type[InvokeError], list[type[Exception]]]:
+        """
+        Map model invoke error to unified error
+        The key is the error type thrown to the caller
+        The value is the error type thrown by the model,
+        which needs to be converted into a unified error type for the caller.
+
+        :return: Invoke error mapping
+        """
+        return {
+            InvokeConnectionError: [openai.APIConnectionError, openai.APITimeoutError],
+            InvokeServerUnavailableError: [openai.InternalServerError],
+            InvokeRateLimitError: [openai.RateLimitError],
+            InvokeAuthorizationError: [openai.AuthenticationError, openai.PermissionDeniedError],
+            InvokeBadRequestError: [
+                openai.BadRequestError,
+                openai.NotFoundError,
+                openai.UnprocessableEntityError,
+                openai.APIError,
+            ],
+        }
diff --git a/api/core/model_runtime/model_providers/fireworks/fireworks.py b/api/core/model_runtime/model_providers/fireworks/fireworks.py
new file mode 100644
index 0000000000000..15f25badab994
--- /dev/null
+++ b/api/core/model_runtime/model_providers/fireworks/fireworks.py
@@ -0,0 +1,27 @@
+import logging
+
+from core.model_runtime.entities.model_entities import ModelType
+from core.model_runtime.errors.validate import CredentialsValidateFailedError
+from core.model_runtime.model_providers.__base.model_provider import ModelProvider
+
+logger = logging.getLogger(__name__)
+
+
+class FireworksProvider(ModelProvider):
+    def validate_provider_credentials(self, credentials: dict) -> None:
+        """
+        Validate provider credentials
+        if validate failed, raise exception
+
+        :param credentials: provider credentials, credentials form defined in `provider_credential_schema`.
+        """
+        try:
+            model_instance = self.get_model_instance(ModelType.LLM)
+            model_instance.validate_credentials(
+                model="accounts/fireworks/models/llama-v3p1-8b-instruct", credentials=credentials
+            )
+        except CredentialsValidateFailedError as ex:
+            raise ex
+        except Exception as ex:
+            logger.exception(f"{self.get_provider_schema().provider} credentials validate failed")
+            raise ex
diff --git a/api/core/model_runtime/model_providers/fireworks/fireworks.yaml b/api/core/model_runtime/model_providers/fireworks/fireworks.yaml
new file mode 100644
index 0000000000000..f886fa23b5bd8
--- /dev/null
+++ b/api/core/model_runtime/model_providers/fireworks/fireworks.yaml
@@ -0,0 +1,29 @@
+provider: fireworks
+label:
+  zh_Hans: Fireworks AI
+  en_US: Fireworks AI
+icon_small:
+  en_US: icon_s_en.svg
+icon_large:
+  en_US: icon_l_en.svg
+background: "#FCFDFF"
+help:
+  title:
+    en_US: Get your API Key from Fireworks AI
+    zh_Hans: 从 Fireworks AI 获取 API Key
+  url:
+    en_US: https://fireworks.ai/account/api-keys
+supported_model_types:
+  - llm
+configurate_methods:
+  - predefined-model
+provider_credential_schema:
+  credential_form_schemas:
+    - variable: fireworks_api_key
+      label:
+        en_US: API Key
+      type: secret-input
+      required: true
+      placeholder:
+        zh_Hans: 在此输入您的 API Key
+        en_US: Enter your API Key
diff --git a/api/core/model_runtime/model_providers/fireworks/llm/__init__.py b/api/core/model_runtime/model_providers/fireworks/llm/__init__.py
new file mode 100644
index 0000000000000..e69de29bb2d1d
diff --git a/api/core/model_runtime/model_providers/fireworks/llm/_position.yaml b/api/core/model_runtime/model_providers/fireworks/llm/_position.yaml
new file mode 100644
index 0000000000000..9f7c1af68cef7
--- /dev/null
+++ b/api/core/model_runtime/model_providers/fireworks/llm/_position.yaml
@@ -0,0 +1,16 @@
+- llama-v3p1-405b-instruct
+- llama-v3p1-70b-instruct
+- llama-v3p1-8b-instruct
+- llama-v3-70b-instruct
+- mixtral-8x22b-instruct
+- mixtral-8x7b-instruct
+- firefunction-v2
+- firefunction-v1
+- gemma2-9b-it
+- llama-v3-70b-instruct-hf
+- llama-v3-8b-instruct
+- llama-v3-8b-instruct-hf
+- mixtral-8x7b-instruct-hf
+- mythomax-l2-13b
+- phi-3-vision-128k-instruct
+- yi-large
diff --git a/api/core/model_runtime/model_providers/fireworks/llm/firefunction-v1.yaml b/api/core/model_runtime/model_providers/fireworks/llm/firefunction-v1.yaml
new file mode 100644
index 0000000000000..f6bac12832d64
--- /dev/null
+++ b/api/core/model_runtime/model_providers/fireworks/llm/firefunction-v1.yaml
@@ -0,0 +1,46 @@
+model: accounts/fireworks/models/firefunction-v1
+label:
+  zh_Hans: Firefunction V1
+  en_US: Firefunction V1
+model_type: llm
+features:
+  - agent-thought
+  - tool-call
+model_properties:
+  mode: chat
+  context_size: 32768
+parameter_rules:
+  - name: temperature
+    use_template: temperature
+  - name: top_p
+    use_template: top_p
+  - name: top_k
+    label:
+      zh_Hans: 取样数量
+      en_US: Top k
+    type: int
+    help:
+      zh_Hans: 仅从每个后续标记的前 K 个选项中采样。
+      en_US: Only sample from the top K options for each subsequent token.
+  - name: max_tokens
+    use_template: max_tokens
+  - name: context_length_exceeded_behavior
+    default: None
+    label:
+      zh_Hans: 上下文长度超出行为
+      en_US: Context Length Exceeded Behavior
+    help:
+      zh_Hans: 上下文长度超出行为
+      en_US: Context Length Exceeded Behavior
+    type: string
+    options:
+      - None
+      - truncate
+      - error
+  - name: response_format
+    use_template: response_format
+pricing:
+  input: '0.5'
+  output: '0.5'
+  unit: '0.000001'
+  currency: USD
diff --git a/api/core/model_runtime/model_providers/fireworks/llm/firefunction-v2.yaml b/api/core/model_runtime/model_providers/fireworks/llm/firefunction-v2.yaml
new file mode 100644
index 0000000000000..2979cb46d572a
--- /dev/null
+++ b/api/core/model_runtime/model_providers/fireworks/llm/firefunction-v2.yaml
@@ -0,0 +1,46 @@
+model: accounts/fireworks/models/firefunction-v2
+label:
+  zh_Hans: Firefunction V2
+  en_US: Firefunction V2
+model_type: llm
+features:
+  - agent-thought
+  - tool-call
+model_properties:
+  mode: chat
+  context_size: 8192
+parameter_rules:
+  - name: temperature
+    use_template: temperature
+  - name: top_p
+    use_template: top_p
+  - name: top_k
+    label:
+      zh_Hans: 取样数量
+      en_US: Top k
+    type: int
+    help:
+      zh_Hans: 仅从每个后续标记的前 K 个选项中采样。
+      en_US: Only sample from the top K options for each subsequent token.
+  - name: max_tokens
+    use_template: max_tokens
+  - name: context_length_exceeded_behavior
+    default: None
+    label:
+      zh_Hans: 上下文长度超出行为
+      en_US: Context Length Exceeded Behavior
+    help:
+      zh_Hans: 上下文长度超出行为
+      en_US: Context Length Exceeded Behavior
+    type: string
+    options:
+      - None
+      - truncate
+      - error
+  - name: response_format
+    use_template: response_format
+pricing:
+  input: '0.9'
+  output: '0.9'
+  unit: '0.000001'
+  currency: USD
diff --git a/api/core/model_runtime/model_providers/fireworks/llm/gemma2-9b-it.yaml b/api/core/model_runtime/model_providers/fireworks/llm/gemma2-9b-it.yaml
new file mode 100644
index 0000000000000..ee41a7e2fdc3d
--- /dev/null
+++ b/api/core/model_runtime/model_providers/fireworks/llm/gemma2-9b-it.yaml
@@ -0,0 +1,45 @@
+model: accounts/fireworks/models/gemma2-9b-it
+label:
+  zh_Hans: Gemma2 9B Instruct
+  en_US: Gemma2 9B Instruct
+model_type: llm
+features:
+  - agent-thought
+model_properties:
+  mode: chat
+  context_size: 8192
+parameter_rules:
+  - name: temperature
+    use_template: temperature
+  - name: top_p
+    use_template: top_p
+  - name: top_k
+    label:
+      zh_Hans: 取样数量
+      en_US: Top k
+    type: int
+    help:
+      zh_Hans: 仅从每个后续标记的前 K 个选项中采样。
+      en_US: Only sample from the top K options for each subsequent token.
+  - name: max_tokens
+    use_template: max_tokens
+  - name: context_length_exceeded_behavior
+    default: None
+    label:
+      zh_Hans: 上下文长度超出行为
+      en_US: Context Length Exceeded Behavior
+    help:
+      zh_Hans: 上下文长度超出行为
+      en_US: Context Length Exceeded Behavior
+    type: string
+    options:
+      - None
+      - truncate
+      - error
+  - name: response_format
+    use_template: response_format
+pricing:
+  input: '0.2'
+  output: '0.2'
+  unit: '0.000001'
+  currency: USD
diff --git a/api/core/model_runtime/model_providers/fireworks/llm/llama-v3-70b-instruct-hf.yaml b/api/core/model_runtime/model_providers/fireworks/llm/llama-v3-70b-instruct-hf.yaml
new file mode 100644
index 0000000000000..2ae89b88165d1
--- /dev/null
+++ b/api/core/model_runtime/model_providers/fireworks/llm/llama-v3-70b-instruct-hf.yaml
@@ -0,0 +1,46 @@
+model: accounts/fireworks/models/llama-v3-70b-instruct-hf
+label:
+  zh_Hans: Llama3 70B Instruct(HF version)
+  en_US: Llama3 70B Instruct(HF version)
+model_type: llm
+features:
+  - agent-thought
+  - tool-call
+model_properties:
+  mode: chat
+  context_size: 8192
+parameter_rules:
+  - name: temperature
+    use_template: temperature
+  - name: top_p
+    use_template: top_p
+  - name: top_k
+    label:
+      zh_Hans: 取样数量
+      en_US: Top k
+    type: int
+    help:
+      zh_Hans: 仅从每个后续标记的前 K 个选项中采样。
+      en_US: Only sample from the top K options for each subsequent token.
+  - name: max_tokens
+    use_template: max_tokens
+  - name: context_length_exceeded_behavior
+    default: None
+    label:
+      zh_Hans: 上下文长度超出行为
+      en_US: Context Length Exceeded Behavior
+    help:
+      zh_Hans: 上下文长度超出行为
+      en_US: Context Length Exceeded Behavior
+    type: string
+    options:
+      - None
+      - truncate
+      - error
+  - name: response_format
+    use_template: response_format
+pricing:
+  input: '0.9'
+  output: '0.9'
+  unit: '0.000001'
+  currency: USD
diff --git a/api/core/model_runtime/model_providers/fireworks/llm/llama-v3-70b-instruct.yaml b/api/core/model_runtime/model_providers/fireworks/llm/llama-v3-70b-instruct.yaml
new file mode 100644
index 0000000000000..7c24b08ca5cca
--- /dev/null
+++ b/api/core/model_runtime/model_providers/fireworks/llm/llama-v3-70b-instruct.yaml
@@ -0,0 +1,46 @@
+model: accounts/fireworks/models/llama-v3-70b-instruct
+label:
+  zh_Hans: Llama3 70B Instruct
+  en_US: Llama3 70B Instruct
+model_type: llm
+features:
+  - agent-thought
+  - tool-call
+model_properties:
+  mode: chat
+  context_size: 8192
+parameter_rules:
+  - name: temperature
+    use_template: temperature
+  - name: top_p
+    use_template: top_p
+  - name: top_k
+    label:
+      zh_Hans: 取样数量
+      en_US: Top k
+    type: int
+    help:
+      zh_Hans: 仅从每个后续标记的前 K 个选项中采样。
+      en_US: Only sample from the top K options for each subsequent token.
+  - name: max_tokens
+    use_template: max_tokens
+  - name: context_length_exceeded_behavior
+    default: None
+    label:
+      zh_Hans: 上下文长度超出行为
+      en_US: Context Length Exceeded Behavior
+    help:
+      zh_Hans: 上下文长度超出行为
+      en_US: Context Length Exceeded Behavior
+    type: string
+    options:
+      - None
+      - truncate
+      - error
+  - name: response_format
+    use_template: response_format
+pricing:
+  input: '0.9'
+  output: '0.9'
+  unit: '0.000001'
+  currency: USD
diff --git a/api/core/model_runtime/model_providers/fireworks/llm/llama-v3-8b-instruct-hf.yaml b/api/core/model_runtime/model_providers/fireworks/llm/llama-v3-8b-instruct-hf.yaml
new file mode 100644
index 0000000000000..83507ef3e5276
--- /dev/null
+++ b/api/core/model_runtime/model_providers/fireworks/llm/llama-v3-8b-instruct-hf.yaml
@@ -0,0 +1,46 @@
+model: accounts/fireworks/models/llama-v3-8b-instruct-hf
+label:
+  zh_Hans: Llama3 8B Instruct(HF version)
+  en_US: Llama3 8B Instruct(HF version)
+model_type: llm
+features:
+  - agent-thought
+  - tool-call
+model_properties:
+  mode: chat
+  context_size: 8192
+parameter_rules:
+  - name: temperature
+    use_template: temperature
+  - name: top_p
+    use_template: top_p
+  - name: top_k
+    label:
+      zh_Hans: 取样数量
+      en_US: Top k
+    type: int
+    help:
+      zh_Hans: 仅从每个后续标记的前 K 个选项中采样。
+      en_US: Only sample from the top K options for each subsequent token.
+  - name: max_tokens
+    use_template: max_tokens
+  - name: context_length_exceeded_behavior
+    default: None
+    label:
+      zh_Hans: 上下文长度超出行为
+      en_US: Context Length Exceeded Behavior
+    help:
+      zh_Hans: 上下文长度超出行为
+      en_US: Context Length Exceeded Behavior
+    type: string
+    options:
+      - None
+      - truncate
+      - error
+  - name: response_format
+    use_template: response_format
+pricing:
+  input: '0.2'
+  output: '0.2'
+  unit: '0.000001'
+  currency: USD
diff --git a/api/core/model_runtime/model_providers/fireworks/llm/llama-v3-8b-instruct.yaml b/api/core/model_runtime/model_providers/fireworks/llm/llama-v3-8b-instruct.yaml
new file mode 100644
index 0000000000000..d8ac9537b80e7
--- /dev/null
+++ b/api/core/model_runtime/model_providers/fireworks/llm/llama-v3-8b-instruct.yaml
@@ -0,0 +1,46 @@
+model: accounts/fireworks/models/llama-v3-8b-instruct
+label:
+  zh_Hans: Llama3 8B Instruct
+  en_US: Llama3 8B Instruct
+model_type: llm
+features:
+  - agent-thought
+  - tool-call
+model_properties:
+  mode: chat
+  context_size: 8192
+parameter_rules:
+  - name: temperature
+    use_template: temperature
+  - name: top_p
+    use_template: top_p
+  - name: top_k
+    label:
+      zh_Hans: 取样数量
+      en_US: Top k
+    type: int
+    help:
+      zh_Hans: 仅从每个后续标记的前 K 个选项中采样。
+      en_US: Only sample from the top K options for each subsequent token.
+  - name: max_tokens
+    use_template: max_tokens
+  - name: context_length_exceeded_behavior
+    default: None
+    label:
+      zh_Hans: 上下文长度超出行为
+      en_US: Context Length Exceeded Behavior
+    help:
+      zh_Hans: 上下文长度超出行为
+      en_US: Context Length Exceeded Behavior
+    type: string
+    options:
+      - None
+      - truncate
+      - error
+  - name: response_format
+    use_template: response_format
+pricing:
+  input: '0.2'
+  output: '0.2'
+  unit: '0.000001'
+  currency: USD
diff --git a/api/core/model_runtime/model_providers/fireworks/llm/llama-v3p1-405b-instruct.yaml b/api/core/model_runtime/model_providers/fireworks/llm/llama-v3p1-405b-instruct.yaml
new file mode 100644
index 0000000000000..c4ddb3e9246d4
--- /dev/null
+++ b/api/core/model_runtime/model_providers/fireworks/llm/llama-v3p1-405b-instruct.yaml
@@ -0,0 +1,46 @@
+model: accounts/fireworks/models/llama-v3p1-405b-instruct
+label:
+  zh_Hans: Llama3.1 405B Instruct
+  en_US: Llama3.1 405B Instruct
+model_type: llm
+features:
+  - agent-thought
+  - tool-call
+model_properties:
+  mode: chat
+  context_size: 131072
+parameter_rules:
+  - name: temperature
+    use_template: temperature
+  - name: top_p
+    use_template: top_p
+  - name: top_k
+    label:
+      zh_Hans: 取样数量
+      en_US: Top k
+    type: int
+    help:
+      zh_Hans: 仅从每个后续标记的前 K 个选项中采样。
+      en_US: Only sample from the top K options for each subsequent token.
+  - name: max_tokens
+    use_template: max_tokens
+  - name: context_length_exceeded_behavior
+    default: None
+    label:
+      zh_Hans: 上下文长度超出行为
+      en_US: Context Length Exceeded Behavior
+    help:
+      zh_Hans: 上下文长度超出行为
+      en_US: Context Length Exceeded Behavior
+    type: string
+    options:
+      - None
+      - truncate
+      - error
+  - name: response_format
+    use_template: response_format
+pricing:
+  input: '3'
+  output: '3'
+  unit: '0.000001'
+  currency: USD
diff --git a/api/core/model_runtime/model_providers/fireworks/llm/llama-v3p1-70b-instruct.yaml b/api/core/model_runtime/model_providers/fireworks/llm/llama-v3p1-70b-instruct.yaml
new file mode 100644
index 0000000000000..62f84f87fa560
--- /dev/null
+++ b/api/core/model_runtime/model_providers/fireworks/llm/llama-v3p1-70b-instruct.yaml
@@ -0,0 +1,46 @@
+model: accounts/fireworks/models/llama-v3p1-70b-instruct
+label:
+  zh_Hans: Llama3.1 70B Instruct
+  en_US: Llama3.1 70B Instruct
+model_type: llm
+features:
+  - agent-thought
+  - tool-call
+model_properties:
+  mode: chat
+  context_size: 131072
+parameter_rules:
+  - name: temperature
+    use_template: temperature
+  - name: top_p
+    use_template: top_p
+  - name: top_k
+    label:
+      zh_Hans: 取样数量
+      en_US: Top k
+    type: int
+    help:
+      zh_Hans: 仅从每个后续标记的前 K 个选项中采样。
+      en_US: Only sample from the top K options for each subsequent token.
+  - name: max_tokens
+    use_template: max_tokens
+  - name: context_length_exceeded_behavior
+    default: None
+    label:
+      zh_Hans: 上下文长度超出行为
+      en_US: Context Length Exceeded Behavior
+    help:
+      zh_Hans: 上下文长度超出行为
+      en_US: Context Length Exceeded Behavior
+    type: string
+    options:
+      - None
+      - truncate
+      - error
+  - name: response_format
+    use_template: response_format
+pricing:
+  input: '0.2'
+  output: '0.2'
+  unit: '0.000001'
+  currency: USD
diff --git a/api/core/model_runtime/model_providers/fireworks/llm/llama-v3p1-8b-instruct.yaml b/api/core/model_runtime/model_providers/fireworks/llm/llama-v3p1-8b-instruct.yaml
new file mode 100644
index 0000000000000..9bb99c91b65b0
--- /dev/null
+++ b/api/core/model_runtime/model_providers/fireworks/llm/llama-v3p1-8b-instruct.yaml
@@ -0,0 +1,46 @@
+model: accounts/fireworks/models/llama-v3p1-8b-instruct
+label:
+  zh_Hans: Llama3.1 8B Instruct
+  en_US: Llama3.1 8B Instruct
+model_type: llm
+features:
+  - agent-thought
+  - tool-call
+model_properties:
+  mode: chat
+  context_size: 131072
+parameter_rules:
+  - name: temperature
+    use_template: temperature
+  - name: top_p
+    use_template: top_p
+  - name: top_k
+    label:
+      zh_Hans: 取样数量
+      en_US: Top k
+    type: int
+    help:
+      zh_Hans: 仅从每个后续标记的前 K 个选项中采样。
+      en_US: Only sample from the top K options for each subsequent token.
+  - name: max_tokens
+    use_template: max_tokens
+  - name: context_length_exceeded_behavior
+    default: None
+    label:
+      zh_Hans: 上下文长度超出行为
+      en_US: Context Length Exceeded Behavior
+    help:
+      zh_Hans: 上下文长度超出行为
+      en_US: Context Length Exceeded Behavior
+    type: string
+    options:
+      - None
+      - truncate
+      - error
+  - name: response_format
+    use_template: response_format
+pricing:
+  input: '0.2'
+  output: '0.2'
+  unit: '0.000001'
+  currency: USD
diff --git a/api/core/model_runtime/model_providers/fireworks/llm/llm.py b/api/core/model_runtime/model_providers/fireworks/llm/llm.py
new file mode 100644
index 0000000000000..2dcf1adba6451
--- /dev/null
+++ b/api/core/model_runtime/model_providers/fireworks/llm/llm.py
@@ -0,0 +1,610 @@
+import logging
+from collections.abc import Generator
+from typing import Optional, Union, cast
+
+from openai import OpenAI, Stream
+from openai.types.chat import ChatCompletion, ChatCompletionChunk, ChatCompletionMessageToolCall
+from openai.types.chat.chat_completion_chunk import ChoiceDeltaFunctionCall, ChoiceDeltaToolCall
+from openai.types.chat.chat_completion_message import FunctionCall
+
+from core.model_runtime.callbacks.base_callback import Callback
+from core.model_runtime.entities.llm_entities import LLMResult, LLMResultChunk, LLMResultChunkDelta
+from core.model_runtime.entities.message_entities import (
+    AssistantPromptMessage,
+    ImagePromptMessageContent,
+    PromptMessage,
+    PromptMessageContentType,
+    PromptMessageTool,
+    SystemPromptMessage,
+    TextPromptMessageContent,
+    ToolPromptMessage,
+    UserPromptMessage,
+)
+from core.model_runtime.errors.validate import CredentialsValidateFailedError
+from core.model_runtime.model_providers.__base.large_language_model import LargeLanguageModel
+from core.model_runtime.model_providers.fireworks._common import _CommonFireworks
+
+logger = logging.getLogger(__name__)
+
+FIREWORKS_BLOCK_MODE_PROMPT = """You should always follow the instructions and output a valid {{block}} object.
+The structure of the {{block}} object you can found in the instructions, use {"answer": "$your_answer"} as the default structure
+if you are not sure about the structure.
+
+<instructions>
+{{instructions}}
+</instructions>
+"""  # noqa: E501
+
+
+class FireworksLargeLanguageModel(_CommonFireworks, LargeLanguageModel):
+    """
+    Model class for Fireworks large language model.
+    """
+
+    def _invoke(
+        self,
+        model: str,
+        credentials: dict,
+        prompt_messages: list[PromptMessage],
+        model_parameters: dict,
+        tools: Optional[list[PromptMessageTool]] = None,
+        stop: Optional[list[str]] = None,
+        stream: bool = True,
+        user: Optional[str] = None,
+    ) -> Union[LLMResult, Generator]:
+        """
+        Invoke large language model
+
+        :param model: model name
+        :param credentials: model credentials
+        :param prompt_messages: prompt messages
+        :param model_parameters: model parameters
+        :param tools: tools for tool calling
+        :param stop: stop words
+        :param stream: is stream response
+        :param user: unique user id
+        :return: full response or stream response chunk generator result
+        """
+
+        return self._chat_generate(
+            model=model,
+            credentials=credentials,
+            prompt_messages=prompt_messages,
+            model_parameters=model_parameters,
+            tools=tools,
+            stop=stop,
+            stream=stream,
+            user=user,
+        )
+
+    def _code_block_mode_wrapper(
+        self,
+        model: str,
+        credentials: dict,
+        prompt_messages: list[PromptMessage],
+        model_parameters: dict,
+        tools: Optional[list[PromptMessageTool]] = None,
+        stop: Optional[list[str]] = None,
+        stream: bool = True,
+        user: Optional[str] = None,
+        callbacks: Optional[list[Callback]] = None,
+    ) -> Union[LLMResult, Generator]:
+        """
+        Code block mode wrapper for invoking large language model
+        """
+        if "response_format" in model_parameters and model_parameters["response_format"] in {"JSON", "XML"}:
+            stop = stop or []
+            self._transform_chat_json_prompts(
+                model=model,
+                credentials=credentials,
+                prompt_messages=prompt_messages,
+                model_parameters=model_parameters,
+                tools=tools,
+                stop=stop,
+                stream=stream,
+                user=user,
+                response_format=model_parameters["response_format"],
+            )
+            model_parameters.pop("response_format")
+
+            return self._invoke(
+                model=model,
+                credentials=credentials,
+                prompt_messages=prompt_messages,
+                model_parameters=model_parameters,
+                tools=tools,
+                stop=stop,
+                stream=stream,
+                user=user,
+            )
+
+    def _transform_chat_json_prompts(
+        self,
+        model: str,
+        credentials: dict,
+        prompt_messages: list[PromptMessage],
+        model_parameters: dict,
+        tools: list[PromptMessageTool] | None = None,
+        stop: list[str] | None = None,
+        stream: bool = True,
+        user: str | None = None,
+        response_format: str = "JSON",
+    ) -> None:
+        """
+        Transform json prompts
+        """
+        if stop is None:
+            stop = []
+        if "```\n" not in stop:
+            stop.append("```\n")
+        if "\n```" not in stop:
+            stop.append("\n```")
+
+        if len(prompt_messages) > 0 and isinstance(prompt_messages[0], SystemPromptMessage):
+            prompt_messages[0] = SystemPromptMessage(
+                content=FIREWORKS_BLOCK_MODE_PROMPT.replace("{{instructions}}", prompt_messages[0].content).replace(
+                    "{{block}}", response_format
+                )
+            )
+            prompt_messages.append(AssistantPromptMessage(content=f"\n```{response_format}\n"))
+        else:
+            prompt_messages.insert(
+                0,
+                SystemPromptMessage(
+                    content=FIREWORKS_BLOCK_MODE_PROMPT.replace(
+                        "{{instructions}}", f"Please output a valid {response_format} object."
+                    ).replace("{{block}}", response_format)
+                ),
+            )
+            prompt_messages.append(AssistantPromptMessage(content=f"\n```{response_format}"))
+
+    def get_num_tokens(
+        self,
+        model: str,
+        credentials: dict,
+        prompt_messages: list[PromptMessage],
+        tools: Optional[list[PromptMessageTool]] = None,
+    ) -> int:
+        """
+        Get number of tokens for given prompt messages
+
+        :param model: model name
+        :param credentials: model credentials
+        :param prompt_messages: prompt messages
+        :param tools: tools for tool calling
+        :return:
+        """
+        return self._num_tokens_from_messages(model, prompt_messages, tools)
+
+    def validate_credentials(self, model: str, credentials: dict) -> None:
+        """
+        Validate model credentials
+
+        :param model: model name
+        :param credentials: model credentials
+        :return:
+        """
+        try:
+            credentials_kwargs = self._to_credential_kwargs(credentials)
+            client = OpenAI(**credentials_kwargs)
+
+            client.chat.completions.create(
+                messages=[{"role": "user", "content": "ping"}], model=model, temperature=0, max_tokens=10, stream=False
+            )
+        except Exception as e:
+            raise CredentialsValidateFailedError(str(e))
+
+    def _chat_generate(
+        self,
+        model: str,
+        credentials: dict,
+        prompt_messages: list[PromptMessage],
+        model_parameters: dict,
+        tools: Optional[list[PromptMessageTool]] = None,
+        stop: Optional[list[str]] = None,
+        stream: bool = True,
+        user: Optional[str] = None,
+    ) -> Union[LLMResult, Generator]:
+        credentials_kwargs = self._to_credential_kwargs(credentials)
+        client = OpenAI(**credentials_kwargs)
+
+        extra_model_kwargs = {}
+
+        if tools:
+            extra_model_kwargs["functions"] = [
+                {"name": tool.name, "description": tool.description, "parameters": tool.parameters} for tool in tools
+            ]
+
+        if stop:
+            extra_model_kwargs["stop"] = stop
+
+        if user:
+            extra_model_kwargs["user"] = user
+
+        # chat model
+        response = client.chat.completions.create(
+            messages=[self._convert_prompt_message_to_dict(m) for m in prompt_messages],
+            model=model,
+            stream=stream,
+            **model_parameters,
+            **extra_model_kwargs,
+        )
+
+        if stream:
+            return self._handle_chat_generate_stream_response(model, credentials, response, prompt_messages, tools)
+        return self._handle_chat_generate_response(model, credentials, response, prompt_messages, tools)
+
+    def _handle_chat_generate_response(
+        self,
+        model: str,
+        credentials: dict,
+        response: ChatCompletion,
+        prompt_messages: list[PromptMessage],
+        tools: Optional[list[PromptMessageTool]] = None,
+    ) -> LLMResult:
+        """
+        Handle llm chat response
+
+        :param model: model name
+        :param credentials: credentials
+        :param response: response
+        :param prompt_messages: prompt messages
+        :param tools: tools for tool calling
+        :return: llm response
+        """
+        assistant_message = response.choices[0].message
+        # assistant_message_tool_calls = assistant_message.tool_calls
+        assistant_message_function_call = assistant_message.function_call
+
+        # extract tool calls from response
+        # tool_calls = self._extract_response_tool_calls(assistant_message_tool_calls)
+        function_call = self._extract_response_function_call(assistant_message_function_call)
+        tool_calls = [function_call] if function_call else []
+
+        # transform assistant message to prompt message
+        assistant_prompt_message = AssistantPromptMessage(content=assistant_message.content, tool_calls=tool_calls)
+
+        # calculate num tokens
+        if response.usage:
+            # transform usage
+            prompt_tokens = response.usage.prompt_tokens
+            completion_tokens = response.usage.completion_tokens
+        else:
+            # calculate num tokens
+            prompt_tokens = self._num_tokens_from_messages(model, prompt_messages, tools)
+            completion_tokens = self._num_tokens_from_messages(model, [assistant_prompt_message])
+
+        # transform usage
+        usage = self._calc_response_usage(model, credentials, prompt_tokens, completion_tokens)
+
+        # transform response
+        response = LLMResult(
+            model=response.model,
+            prompt_messages=prompt_messages,
+            message=assistant_prompt_message,
+            usage=usage,
+            system_fingerprint=response.system_fingerprint,
+        )
+
+        return response
+
+    def _handle_chat_generate_stream_response(
+        self,
+        model: str,
+        credentials: dict,
+        response: Stream[ChatCompletionChunk],
+        prompt_messages: list[PromptMessage],
+        tools: Optional[list[PromptMessageTool]] = None,
+    ) -> Generator:
+        """
+        Handle llm chat stream response
+
+        :param model: model name
+        :param response: response
+        :param prompt_messages: prompt messages
+        :param tools: tools for tool calling
+        :return: llm response chunk generator
+        """
+        full_assistant_content = ""
+        delta_assistant_message_function_call_storage: Optional[ChoiceDeltaFunctionCall] = None
+        prompt_tokens = 0
+        completion_tokens = 0
+        final_tool_calls = []
+        final_chunk = LLMResultChunk(
+            model=model,
+            prompt_messages=prompt_messages,
+            delta=LLMResultChunkDelta(
+                index=0,
+                message=AssistantPromptMessage(content=""),
+            ),
+        )
+
+        for chunk in response:
+            if len(chunk.choices) == 0:
+                if chunk.usage:
+                    # calculate num tokens
+                    prompt_tokens = chunk.usage.prompt_tokens
+                    completion_tokens = chunk.usage.completion_tokens
+                continue
+
+            delta = chunk.choices[0]
+            has_finish_reason = delta.finish_reason is not None
+
+            if (
+                not has_finish_reason
+                and (delta.delta.content is None or delta.delta.content == "")
+                and delta.delta.function_call is None
+            ):
+                continue
+
+            # assistant_message_tool_calls = delta.delta.tool_calls
+            assistant_message_function_call = delta.delta.function_call
+
+            # extract tool calls from response
+            if delta_assistant_message_function_call_storage is not None:
+                # handle process of stream function call
+                if assistant_message_function_call:
+                    # message has not ended ever
+                    delta_assistant_message_function_call_storage.arguments += assistant_message_function_call.arguments
+                    continue
+                else:
+                    # message has ended
+                    assistant_message_function_call = delta_assistant_message_function_call_storage
+                    delta_assistant_message_function_call_storage = None
+            else:
+                if assistant_message_function_call:
+                    # start of stream function call
+                    delta_assistant_message_function_call_storage = assistant_message_function_call
+                    if delta_assistant_message_function_call_storage.arguments is None:
+                        delta_assistant_message_function_call_storage.arguments = ""
+                    if not has_finish_reason:
+                        continue
+
+            # tool_calls = self._extract_response_tool_calls(assistant_message_tool_calls)
+            function_call = self._extract_response_function_call(assistant_message_function_call)
+            tool_calls = [function_call] if function_call else []
+            if tool_calls:
+                final_tool_calls.extend(tool_calls)
+
+            # transform assistant message to prompt message
+            assistant_prompt_message = AssistantPromptMessage(content=delta.delta.content or "", tool_calls=tool_calls)
+
+            full_assistant_content += delta.delta.content or ""
+
+            if has_finish_reason:
+                final_chunk = LLMResultChunk(
+                    model=chunk.model,
+                    prompt_messages=prompt_messages,
+                    system_fingerprint=chunk.system_fingerprint,
+                    delta=LLMResultChunkDelta(
+                        index=delta.index,
+                        message=assistant_prompt_message,
+                        finish_reason=delta.finish_reason,
+                    ),
+                )
+            else:
+                yield LLMResultChunk(
+                    model=chunk.model,
+                    prompt_messages=prompt_messages,
+                    system_fingerprint=chunk.system_fingerprint,
+                    delta=LLMResultChunkDelta(
+                        index=delta.index,
+                        message=assistant_prompt_message,
+                    ),
+                )
+
+        if not prompt_tokens:
+            prompt_tokens = self._num_tokens_from_messages(model, prompt_messages, tools)
+
+        if not completion_tokens:
+            full_assistant_prompt_message = AssistantPromptMessage(
+                content=full_assistant_content, tool_calls=final_tool_calls
+            )
+            completion_tokens = self._num_tokens_from_messages(model, [full_assistant_prompt_message])
+
+        # transform usage
+        usage = self._calc_response_usage(model, credentials, prompt_tokens, completion_tokens)
+        final_chunk.delta.usage = usage
+
+        yield final_chunk
+
+    def _extract_response_tool_calls(
+        self, response_tool_calls: list[ChatCompletionMessageToolCall | ChoiceDeltaToolCall]
+    ) -> list[AssistantPromptMessage.ToolCall]:
+        """
+        Extract tool calls from response
+
+        :param response_tool_calls: response tool calls
+        :return: list of tool calls
+        """
+        tool_calls = []
+        if response_tool_calls:
+            for response_tool_call in response_tool_calls:
+                function = AssistantPromptMessage.ToolCall.ToolCallFunction(
+                    name=response_tool_call.function.name, arguments=response_tool_call.function.arguments
+                )
+
+                tool_call = AssistantPromptMessage.ToolCall(
+                    id=response_tool_call.id, type=response_tool_call.type, function=function
+                )
+                tool_calls.append(tool_call)
+
+        return tool_calls
+
+    def _extract_response_function_call(
+        self, response_function_call: FunctionCall | ChoiceDeltaFunctionCall
+    ) -> AssistantPromptMessage.ToolCall:
+        """
+        Extract function call from response
+
+        :param response_function_call: response function call
+        :return: tool call
+        """
+        tool_call = None
+        if response_function_call:
+            function = AssistantPromptMessage.ToolCall.ToolCallFunction(
+                name=response_function_call.name, arguments=response_function_call.arguments
+            )
+
+            tool_call = AssistantPromptMessage.ToolCall(
+                id=response_function_call.name, type="function", function=function
+            )
+
+        return tool_call
+
+    def _convert_prompt_message_to_dict(self, message: PromptMessage) -> dict:
+        """
+        Convert PromptMessage to dict for Fireworks API
+        """
+        if isinstance(message, UserPromptMessage):
+            message = cast(UserPromptMessage, message)
+            if isinstance(message.content, str):
+                message_dict = {"role": "user", "content": message.content}
+            else:
+                sub_messages = []
+                for message_content in message.content:
+                    if message_content.type == PromptMessageContentType.TEXT:
+                        message_content = cast(TextPromptMessageContent, message_content)
+                        sub_message_dict = {"type": "text", "text": message_content.data}
+                        sub_messages.append(sub_message_dict)
+                    elif message_content.type == PromptMessageContentType.IMAGE:
+                        message_content = cast(ImagePromptMessageContent, message_content)
+                        sub_message_dict = {
+                            "type": "image_url",
+                            "image_url": {"url": message_content.data, "detail": message_content.detail.value},
+                        }
+                        sub_messages.append(sub_message_dict)
+
+                message_dict = {"role": "user", "content": sub_messages}
+        elif isinstance(message, AssistantPromptMessage):
+            message = cast(AssistantPromptMessage, message)
+            message_dict = {"role": "assistant", "content": message.content}
+            if message.tool_calls:
+                # message_dict["tool_calls"] = [tool_call.dict() for tool_call in
+                #                               message.tool_calls]
+                function_call = message.tool_calls[0]
+                message_dict["function_call"] = {
+                    "name": function_call.function.name,
+                    "arguments": function_call.function.arguments,
+                }
+        elif isinstance(message, SystemPromptMessage):
+            message = cast(SystemPromptMessage, message)
+            message_dict = {"role": "system", "content": message.content}
+        elif isinstance(message, ToolPromptMessage):
+            message = cast(ToolPromptMessage, message)
+            # message_dict = {
+            #     "role": "tool",
+            #     "content": message.content,
+            #     "tool_call_id": message.tool_call_id
+            # }
+            message_dict = {"role": "function", "content": message.content, "name": message.tool_call_id}
+        else:
+            raise ValueError(f"Got unknown type {message}")
+
+        if message.name:
+            message_dict["name"] = message.name
+
+        return message_dict
+
+    def _num_tokens_from_messages(
+        self,
+        model: str,
+        messages: list[PromptMessage],
+        tools: Optional[list[PromptMessageTool]] = None,
+        credentials: dict = None,
+    ) -> int:
+        """
+        Approximate num tokens with GPT2 tokenizer.
+        """
+
+        tokens_per_message = 3
+        tokens_per_name = 1
+
+        num_tokens = 0
+        messages_dict = [self._convert_prompt_message_to_dict(m) for m in messages]
+        for message in messages_dict:
+            num_tokens += tokens_per_message
+            for key, value in message.items():
+                # Cast str(value) in case the message value is not a string
+                # This occurs with function messages
+                # TODO: The current token calculation method for the image type is not implemented,
+                #  which need to download the image and then get the resolution for calculation,
+                #  and will increase the request delay
+                if isinstance(value, list):
+                    text = ""
+                    for item in value:
+                        if isinstance(item, dict) and item["type"] == "text":
+                            text += item["text"]
+
+                    value = text
+
+                if key == "tool_calls":
+                    for tool_call in value:
+                        for t_key, t_value in tool_call.items():
+                            num_tokens += self._get_num_tokens_by_gpt2(t_key)
+                            if t_key == "function":
+                                for f_key, f_value in t_value.items():
+                                    num_tokens += self._get_num_tokens_by_gpt2(f_key)
+                                    num_tokens += self._get_num_tokens_by_gpt2(f_value)
+                            else:
+                                num_tokens += self._get_num_tokens_by_gpt2(t_key)
+                                num_tokens += self._get_num_tokens_by_gpt2(t_value)
+                else:
+                    num_tokens += self._get_num_tokens_by_gpt2(str(value))
+
+                if key == "name":
+                    num_tokens += tokens_per_name
+
+        # every reply is primed with <im_start>assistant
+        num_tokens += 3
+
+        if tools:
+            num_tokens += self._num_tokens_for_tools(tools)
+
+        return num_tokens
+
+    def _num_tokens_for_tools(self, tools: list[PromptMessageTool]) -> int:
+        """
+        Calculate num tokens for tool calling with tiktoken package.
+
+        :param tools: tools for tool calling
+        :return: number of tokens
+        """
+        num_tokens = 0
+        for tool in tools:
+            num_tokens += self._get_num_tokens_by_gpt2("type")
+            num_tokens += self._get_num_tokens_by_gpt2("function")
+            num_tokens += self._get_num_tokens_by_gpt2("function")
+
+            # calculate num tokens for function object
+            num_tokens += self._get_num_tokens_by_gpt2("name")
+            num_tokens += self._get_num_tokens_by_gpt2(tool.name)
+            num_tokens += self._get_num_tokens_by_gpt2("description")
+            num_tokens += self._get_num_tokens_by_gpt2(tool.description)
+            parameters = tool.parameters
+            num_tokens += self._get_num_tokens_by_gpt2("parameters")
+            if "title" in parameters:
+                num_tokens += self._get_num_tokens_by_gpt2("title")
+                num_tokens += self._get_num_tokens_by_gpt2(parameters.get("title"))
+            num_tokens += self._get_num_tokens_by_gpt2("type")
+            num_tokens += self._get_num_tokens_by_gpt2(parameters.get("type"))
+            if "properties" in parameters:
+                num_tokens += self._get_num_tokens_by_gpt2("properties")
+                for key, value in parameters.get("properties").items():
+                    num_tokens += self._get_num_tokens_by_gpt2(key)
+                    for field_key, field_value in value.items():
+                        num_tokens += self._get_num_tokens_by_gpt2(field_key)
+                        if field_key == "enum":
+                            for enum_field in field_value:
+                                num_tokens += 3
+                                num_tokens += self._get_num_tokens_by_gpt2(enum_field)
+                        else:
+                            num_tokens += self._get_num_tokens_by_gpt2(field_key)
+                            num_tokens += self._get_num_tokens_by_gpt2(str(field_value))
+            if "required" in parameters:
+                num_tokens += self._get_num_tokens_by_gpt2("required")
+                for required_field in parameters["required"]:
+                    num_tokens += 3
+                    num_tokens += self._get_num_tokens_by_gpt2(required_field)
+
+        return num_tokens
diff --git a/api/core/model_runtime/model_providers/fireworks/llm/mixtral-8x22b-instruct.yaml b/api/core/model_runtime/model_providers/fireworks/llm/mixtral-8x22b-instruct.yaml
new file mode 100644
index 0000000000000..87d977e26cf1b
--- /dev/null
+++ b/api/core/model_runtime/model_providers/fireworks/llm/mixtral-8x22b-instruct.yaml
@@ -0,0 +1,46 @@
+model: accounts/fireworks/models/mixtral-8x22b-instruct
+label:
+  zh_Hans: Mixtral MoE 8x22B Instruct
+  en_US: Mixtral MoE 8x22B Instruct
+model_type: llm
+features:
+  - agent-thought
+  - tool-call
+model_properties:
+  mode: chat
+  context_size: 65536
+parameter_rules:
+  - name: temperature
+    use_template: temperature
+  - name: top_p
+    use_template: top_p
+  - name: top_k
+    label:
+      zh_Hans: 取样数量
+      en_US: Top k
+    type: int
+    help:
+      zh_Hans: 仅从每个后续标记的前 K 个选项中采样。
+      en_US: Only sample from the top K options for each subsequent token.
+  - name: max_tokens
+    use_template: max_tokens
+  - name: context_length_exceeded_behavior
+    default: None
+    label:
+      zh_Hans: 上下文长度超出行为
+      en_US: Context Length Exceeded Behavior
+    help:
+      zh_Hans: 上下文长度超出行为
+      en_US: Context Length Exceeded Behavior
+    type: string
+    options:
+      - None
+      - truncate
+      - error
+  - name: response_format
+    use_template: response_format
+pricing:
+  input: '1.2'
+  output: '1.2'
+  unit: '0.000001'
+  currency: USD
diff --git a/api/core/model_runtime/model_providers/fireworks/llm/mixtral-8x7b-instruct-hf.yaml b/api/core/model_runtime/model_providers/fireworks/llm/mixtral-8x7b-instruct-hf.yaml
new file mode 100644
index 0000000000000..e3d5a90858c5a
--- /dev/null
+++ b/api/core/model_runtime/model_providers/fireworks/llm/mixtral-8x7b-instruct-hf.yaml
@@ -0,0 +1,46 @@
+model: accounts/fireworks/models/mixtral-8x7b-instruct-hf
+label:
+  zh_Hans: Mixtral MoE 8x7B Instruct(HF version)
+  en_US: Mixtral MoE 8x7B Instruct(HF version)
+model_type: llm
+features:
+  - agent-thought
+  - tool-call
+model_properties:
+  mode: chat
+  context_size: 32768
+parameter_rules:
+  - name: temperature
+    use_template: temperature
+  - name: top_p
+    use_template: top_p
+  - name: top_k
+    label:
+      zh_Hans: 取样数量
+      en_US: Top k
+    type: int
+    help:
+      zh_Hans: 仅从每个后续标记的前 K 个选项中采样。
+      en_US: Only sample from the top K options for each subsequent token.
+  - name: max_tokens
+    use_template: max_tokens
+  - name: context_length_exceeded_behavior
+    default: None
+    label:
+      zh_Hans: 上下文长度超出行为
+      en_US: Context Length Exceeded Behavior
+    help:
+      zh_Hans: 上下文长度超出行为
+      en_US: Context Length Exceeded Behavior
+    type: string
+    options:
+      - None
+      - truncate
+      - error
+  - name: response_format
+    use_template: response_format
+pricing:
+  input: '0.5'
+  output: '0.5'
+  unit: '0.000001'
+  currency: USD
diff --git a/api/core/model_runtime/model_providers/fireworks/llm/mixtral-8x7b-instruct.yaml b/api/core/model_runtime/model_providers/fireworks/llm/mixtral-8x7b-instruct.yaml
new file mode 100644
index 0000000000000..45f632ceff2cf
--- /dev/null
+++ b/api/core/model_runtime/model_providers/fireworks/llm/mixtral-8x7b-instruct.yaml
@@ -0,0 +1,46 @@
+model: accounts/fireworks/models/mixtral-8x7b-instruct
+label:
+  zh_Hans: Mixtral MoE 8x7B Instruct
+  en_US: Mixtral MoE 8x7B Instruct
+model_type: llm
+features:
+  - agent-thought
+  - tool-call
+model_properties:
+  mode: chat
+  context_size: 32768
+parameter_rules:
+  - name: temperature
+    use_template: temperature
+  - name: top_p
+    use_template: top_p
+  - name: top_k
+    label:
+      zh_Hans: 取样数量
+      en_US: Top k
+    type: int
+    help:
+      zh_Hans: 仅从每个后续标记的前 K 个选项中采样。
+      en_US: Only sample from the top K options for each subsequent token.
+  - name: max_tokens
+    use_template: max_tokens
+  - name: context_length_exceeded_behavior
+    default: None
+    label:
+      zh_Hans: 上下文长度超出行为
+      en_US: Context Length Exceeded Behavior
+    help:
+      zh_Hans: 上下文长度超出行为
+      en_US: Context Length Exceeded Behavior
+    type: string
+    options:
+      - None
+      - truncate
+      - error
+  - name: response_format
+    use_template: response_format
+pricing:
+  input: '0.5'
+  output: '0.5'
+  unit: '0.000001'
+  currency: USD
diff --git a/api/core/model_runtime/model_providers/fireworks/llm/mythomax-l2-13b.yaml b/api/core/model_runtime/model_providers/fireworks/llm/mythomax-l2-13b.yaml
new file mode 100644
index 0000000000000..9c3486ba10751
--- /dev/null
+++ b/api/core/model_runtime/model_providers/fireworks/llm/mythomax-l2-13b.yaml
@@ -0,0 +1,46 @@
+model: accounts/fireworks/models/mythomax-l2-13b
+label:
+  zh_Hans: MythoMax L2 13b
+  en_US: MythoMax L2 13b
+model_type: llm
+features:
+  - agent-thought
+  - tool-call
+model_properties:
+  mode: chat
+  context_size: 4096
+parameter_rules:
+  - name: temperature
+    use_template: temperature
+  - name: top_p
+    use_template: top_p
+  - name: top_k
+    label:
+      zh_Hans: 取样数量
+      en_US: Top k
+    type: int
+    help:
+      zh_Hans: 仅从每个后续标记的前 K 个选项中采样。
+      en_US: Only sample from the top K options for each subsequent token.
+  - name: max_tokens
+    use_template: max_tokens
+  - name: context_length_exceeded_behavior
+    default: None
+    label:
+      zh_Hans: 上下文长度超出行为
+      en_US: Context Length Exceeded Behavior
+    help:
+      zh_Hans: 上下文长度超出行为
+      en_US: Context Length Exceeded Behavior
+    type: string
+    options:
+      - None
+      - truncate
+      - error
+  - name: response_format
+    use_template: response_format
+pricing:
+  input: '0.2'
+  output: '0.2'
+  unit: '0.000001'
+  currency: USD
diff --git a/api/core/model_runtime/model_providers/fireworks/llm/phi-3-vision-128k-instruct.yaml b/api/core/model_runtime/model_providers/fireworks/llm/phi-3-vision-128k-instruct.yaml
new file mode 100644
index 0000000000000..e399f2edb1b1b
--- /dev/null
+++ b/api/core/model_runtime/model_providers/fireworks/llm/phi-3-vision-128k-instruct.yaml
@@ -0,0 +1,46 @@
+model: accounts/fireworks/models/phi-3-vision-128k-instruct
+label:
+  zh_Hans: Phi3.5 Vision Instruct
+  en_US: Phi3.5 Vision Instruct
+model_type: llm
+features:
+  - agent-thought
+  - vision
+model_properties:
+  mode: chat
+  context_size: 8192
+parameter_rules:
+  - name: temperature
+    use_template: temperature
+  - name: top_p
+    use_template: top_p
+  - name: top_k
+    label:
+      zh_Hans: 取样数量
+      en_US: Top k
+    type: int
+    help:
+      zh_Hans: 仅从每个后续标记的前 K 个选项中采样。
+      en_US: Only sample from the top K options for each subsequent token.
+  - name: max_tokens
+    use_template: max_tokens
+  - name: context_length_exceeded_behavior
+    default: None
+    label:
+      zh_Hans: 上下文长度超出行为
+      en_US: Context Length Exceeded Behavior
+    help:
+      zh_Hans: 上下文长度超出行为
+      en_US: Context Length Exceeded Behavior
+    type: string
+    options:
+      - None
+      - truncate
+      - error
+  - name: response_format
+    use_template: response_format
+pricing:
+  input: '0.2'
+  output: '0.2'
+  unit: '0.000001'
+  currency: USD
diff --git a/api/core/model_runtime/model_providers/fireworks/llm/yi-large.yaml b/api/core/model_runtime/model_providers/fireworks/llm/yi-large.yaml
new file mode 100644
index 0000000000000..bb4b6f994ec12
--- /dev/null
+++ b/api/core/model_runtime/model_providers/fireworks/llm/yi-large.yaml
@@ -0,0 +1,45 @@
+model: accounts/yi-01-ai/models/yi-large
+label:
+  zh_Hans: Yi-Large
+  en_US: Yi-Large
+model_type: llm
+features:
+  - agent-thought
+model_properties:
+  mode: chat
+  context_size: 32768
+parameter_rules:
+  - name: temperature
+    use_template: temperature
+  - name: top_p
+    use_template: top_p
+  - name: top_k
+    label:
+      zh_Hans: 取样数量
+      en_US: Top k
+    type: int
+    help:
+      zh_Hans: 仅从每个后续标记的前 K 个选项中采样。
+      en_US: Only sample from the top K options for each subsequent token.
+  - name: max_tokens
+    use_template: max_tokens
+  - name: context_length_exceeded_behavior
+    default: None
+    label:
+      zh_Hans: 上下文长度超出行为
+      en_US: Context Length Exceeded Behavior
+    help:
+      zh_Hans: 上下文长度超出行为
+      en_US: Context Length Exceeded Behavior
+    type: string
+    options:
+      - None
+      - truncate
+      - error
+  - name: response_format
+    use_template: response_format
+pricing:
+  input: '3'
+  output: '3'
+  unit: '0.000001'
+  currency: USD
diff --git a/api/pyproject.toml b/api/pyproject.toml
index 1f483fc49f032..93482b032d536 100644
--- a/api/pyproject.toml
+++ b/api/pyproject.toml
@@ -100,6 +100,7 @@ exclude = [
 [tool.pytest_env]
 OPENAI_API_KEY = "sk-IamNotARealKeyJustForMockTestKawaiiiiiiiiii"
 UPSTAGE_API_KEY = "up-aaaaaaaaaaaaaaaaaaaa"
+FIREWORKS_API_KEY = "fw_aaaaaaaaaaaaaaaaaaaa"
 AZURE_OPENAI_API_BASE = "https://difyai-openai.openai.azure.com"
 AZURE_OPENAI_API_KEY = "xxxxb1707exxxxxxxxxxaaxxxxxf94"
 ANTHROPIC_API_KEY = "sk-ant-api11-IamNotARealKeyJustForMockTestKawaiiiiiiiiii-NotBaka-ASkksz"
diff --git a/api/tests/integration_tests/model_runtime/fireworks/__init__.py b/api/tests/integration_tests/model_runtime/fireworks/__init__.py
new file mode 100644
index 0000000000000..e69de29bb2d1d
diff --git a/api/tests/integration_tests/model_runtime/fireworks/test_llm.py b/api/tests/integration_tests/model_runtime/fireworks/test_llm.py
new file mode 100644
index 0000000000000..699ca293a2fca
--- /dev/null
+++ b/api/tests/integration_tests/model_runtime/fireworks/test_llm.py
@@ -0,0 +1,186 @@
+import os
+from collections.abc import Generator
+
+import pytest
+
+from core.model_runtime.entities.llm_entities import LLMResult, LLMResultChunk, LLMResultChunkDelta
+from core.model_runtime.entities.message_entities import (
+    AssistantPromptMessage,
+    PromptMessageTool,
+    SystemPromptMessage,
+    UserPromptMessage,
+)
+from core.model_runtime.entities.model_entities import AIModelEntity
+from core.model_runtime.errors.validate import CredentialsValidateFailedError
+from core.model_runtime.model_providers.fireworks.llm.llm import FireworksLargeLanguageModel
+
+"""FOR MOCK FIXTURES, DO NOT REMOVE"""
+from tests.integration_tests.model_runtime.__mock.openai import setup_openai_mock
+
+
+def test_predefined_models():
+    model = FireworksLargeLanguageModel()
+    model_schemas = model.predefined_models()
+
+    assert len(model_schemas) >= 1
+    assert isinstance(model_schemas[0], AIModelEntity)
+
+
+@pytest.mark.parametrize("setup_openai_mock", [["chat"]], indirect=True)
+def test_validate_credentials_for_chat_model(setup_openai_mock):
+    model = FireworksLargeLanguageModel()
+
+    with pytest.raises(CredentialsValidateFailedError):
+        # model name to gpt-3.5-turbo because of mocking
+        model.validate_credentials(model="gpt-3.5-turbo", credentials={"fireworks_api_key": "invalid_key"})
+
+    model.validate_credentials(
+        model="accounts/fireworks/models/llama-v3p1-8b-instruct",
+        credentials={"fireworks_api_key": os.environ.get("FIREWORKS_API_KEY")},
+    )
+
+
+@pytest.mark.parametrize("setup_openai_mock", [["chat"]], indirect=True)
+def test_invoke_chat_model(setup_openai_mock):
+    model = FireworksLargeLanguageModel()
+
+    result = model.invoke(
+        model="accounts/fireworks/models/llama-v3p1-8b-instruct",
+        credentials={"fireworks_api_key": os.environ.get("FIREWORKS_API_KEY")},
+        prompt_messages=[
+            SystemPromptMessage(
+                content="You are a helpful AI assistant.",
+            ),
+            UserPromptMessage(content="Hello World!"),
+        ],
+        model_parameters={
+            "temperature": 0.0,
+            "top_p": 1.0,
+            "presence_penalty": 0.0,
+            "frequency_penalty": 0.0,
+            "max_tokens": 10,
+        },
+        stop=["How"],
+        stream=False,
+        user="foo",
+    )
+
+    assert isinstance(result, LLMResult)
+    assert len(result.message.content) > 0
+
+
+@pytest.mark.parametrize("setup_openai_mock", [["chat"]], indirect=True)
+def test_invoke_chat_model_with_tools(setup_openai_mock):
+    model = FireworksLargeLanguageModel()
+
+    result = model.invoke(
+        model="accounts/fireworks/models/llama-v3p1-8b-instruct",
+        credentials={"fireworks_api_key": os.environ.get("FIREWORKS_API_KEY")},
+        prompt_messages=[
+            SystemPromptMessage(
+                content="You are a helpful AI assistant.",
+            ),
+            UserPromptMessage(
+                content="what's the weather today in London?",
+            ),
+        ],
+        model_parameters={"temperature": 0.0, "max_tokens": 100},
+        tools=[
+            PromptMessageTool(
+                name="get_weather",
+                description="Determine weather in my location",
+                parameters={
+                    "type": "object",
+                    "properties": {
+                        "location": {"type": "string", "description": "The city and state e.g. San Francisco, CA"},
+                        "unit": {"type": "string", "enum": ["c", "f"]},
+                    },
+                    "required": ["location"],
+                },
+            ),
+            PromptMessageTool(
+                name="get_stock_price",
+                description="Get the current stock price",
+                parameters={
+                    "type": "object",
+                    "properties": {"symbol": {"type": "string", "description": "The stock symbol"}},
+                    "required": ["symbol"],
+                },
+            ),
+        ],
+        stream=False,
+        user="foo",
+    )
+
+    assert isinstance(result, LLMResult)
+    assert isinstance(result.message, AssistantPromptMessage)
+    assert len(result.message.tool_calls) > 0
+
+
+@pytest.mark.parametrize("setup_openai_mock", [["chat"]], indirect=True)
+def test_invoke_stream_chat_model(setup_openai_mock):
+    model = FireworksLargeLanguageModel()
+
+    result = model.invoke(
+        model="accounts/fireworks/models/llama-v3p1-8b-instruct",
+        credentials={"fireworks_api_key": os.environ.get("FIREWORKS_API_KEY")},
+        prompt_messages=[
+            SystemPromptMessage(
+                content="You are a helpful AI assistant.",
+            ),
+            UserPromptMessage(content="Hello World!"),
+        ],
+        model_parameters={"temperature": 0.0, "max_tokens": 100},
+        stream=True,
+        user="foo",
+    )
+
+    assert isinstance(result, Generator)
+
+    for chunk in result:
+        assert isinstance(chunk, LLMResultChunk)
+        assert isinstance(chunk.delta, LLMResultChunkDelta)
+        assert isinstance(chunk.delta.message, AssistantPromptMessage)
+        assert len(chunk.delta.message.content) > 0 if chunk.delta.finish_reason is None else True
+        if chunk.delta.finish_reason is not None:
+            assert chunk.delta.usage is not None
+            assert chunk.delta.usage.completion_tokens > 0
+
+
+def test_get_num_tokens():
+    model = FireworksLargeLanguageModel()
+
+    num_tokens = model.get_num_tokens(
+        model="accounts/fireworks/models/llama-v3p1-8b-instruct",
+        credentials={"fireworks_api_key": os.environ.get("FIREWORKS_API_KEY")},
+        prompt_messages=[UserPromptMessage(content="Hello World!")],
+    )
+
+    assert num_tokens == 10
+
+    num_tokens = model.get_num_tokens(
+        model="accounts/fireworks/models/llama-v3p1-8b-instruct",
+        credentials={"fireworks_api_key": os.environ.get("FIREWORKS_API_KEY")},
+        prompt_messages=[
+            SystemPromptMessage(
+                content="You are a helpful AI assistant.",
+            ),
+            UserPromptMessage(content="Hello World!"),
+        ],
+        tools=[
+            PromptMessageTool(
+                name="get_weather",
+                description="Determine weather in my location",
+                parameters={
+                    "type": "object",
+                    "properties": {
+                        "location": {"type": "string", "description": "The city and state e.g. San Francisco, CA"},
+                        "unit": {"type": "string", "enum": ["c", "f"]},
+                    },
+                    "required": ["location"],
+                },
+            ),
+        ],
+    )
+
+    assert num_tokens == 77
diff --git a/api/tests/integration_tests/model_runtime/fireworks/test_provider.py b/api/tests/integration_tests/model_runtime/fireworks/test_provider.py
new file mode 100644
index 0000000000000..a68cf1a1a8fbd
--- /dev/null
+++ b/api/tests/integration_tests/model_runtime/fireworks/test_provider.py
@@ -0,0 +1,17 @@
+import os
+
+import pytest
+
+from core.model_runtime.errors.validate import CredentialsValidateFailedError
+from core.model_runtime.model_providers.fireworks.fireworks import FireworksProvider
+from tests.integration_tests.model_runtime.__mock.openai import setup_openai_mock
+
+
+@pytest.mark.parametrize("setup_openai_mock", [["chat"]], indirect=True)
+def test_validate_provider_credentials(setup_openai_mock):
+    provider = FireworksProvider()
+
+    with pytest.raises(CredentialsValidateFailedError):
+        provider.validate_provider_credentials(credentials={})
+
+    provider.validate_provider_credentials(credentials={"fireworks_api_key": os.environ.get("FIREWORKS_API_KEY")})
diff --git a/dev/pytest/pytest_model_runtime.sh b/dev/pytest/pytest_model_runtime.sh
index aba13292ab831..4c1c6bf4f3ab1 100755
--- a/dev/pytest/pytest_model_runtime.sh
+++ b/dev/pytest/pytest_model_runtime.sh
@@ -6,5 +6,5 @@ pytest api/tests/integration_tests/model_runtime/anthropic \
   api/tests/integration_tests/model_runtime/openai api/tests/integration_tests/model_runtime/chatglm \
   api/tests/integration_tests/model_runtime/google api/tests/integration_tests/model_runtime/xinference \
   api/tests/integration_tests/model_runtime/huggingface_hub/test_llm.py \
-  api/tests/integration_tests/model_runtime/upstage
-
+  api/tests/integration_tests/model_runtime/upstage \
+  api/tests/integration_tests/model_runtime/fireworks

From 740fad06c1fe9a5a24a6afdd7174807b82bbb464 Mon Sep 17 00:00:00 2001
From: Waffle <52460705+ox01024@users.noreply.github.com>
Date: Sun, 22 Sep 2024 10:14:14 +0800
Subject: [PATCH 016/235] feat(tools/cogview):  Updated cogview tool to support
 cogview-3 and the latest cogview-3-plus (#8382)

---
 .../zhipuai/zhipuai_sdk/__init__.py           |   3 +-
 .../zhipuai/zhipuai_sdk/__version__.py        |   2 +-
 .../zhipuai/zhipuai_sdk/_client.py            |  29 +-
 .../zhipuai_sdk/api_resource/__init__.py      |  35 +-
 .../api_resource/assistant/__init__.py        |   3 +
 .../api_resource/assistant/assistant.py       | 122 +++
 .../zhipuai_sdk/api_resource/batches.py       | 146 ++++
 .../zhipuai_sdk/api_resource/chat/__init__.py |   5 +
 .../api_resource/chat/async_completions.py    |  91 ++-
 .../zhipuai_sdk/api_resource/chat/chat.py     |  15 +-
 .../api_resource/chat/completions.py          |  80 +-
 .../zhipuai_sdk/api_resource/embeddings.py    |  13 +-
 .../zhipuai/zhipuai_sdk/api_resource/files.py | 167 +++-
 .../api_resource/fine_tuning/__init__.py      |   5 +
 .../api_resource/fine_tuning/fine_tuning.py   |  15 +-
 .../api_resource/fine_tuning/jobs/__init__.py |   3 +
 .../fine_tuning/{ => jobs}/jobs.py            |  68 +-
 .../fine_tuning/models/__init__.py            |   3 +
 .../fine_tuning/models/fine_tuned_models.py   |  41 +
 .../zhipuai_sdk/api_resource/images.py        |  13 +-
 .../api_resource/knowledge/__init__.py        |   3 +
 .../knowledge/document/__init__.py            |   3 +
 .../knowledge/document/document.py            | 217 +++++
 .../api_resource/knowledge/knowledge.py       | 173 ++++
 .../api_resource/tools/__init__.py            |   3 +
 .../zhipuai_sdk/api_resource/tools/tools.py   |  65 ++
 .../api_resource/videos/__init__.py           |   7 +
 .../zhipuai_sdk/api_resource/videos/videos.py |  77 ++
 .../zhipuai/zhipuai_sdk/core/__init__.py      | 108 +++
 .../zhipuai/zhipuai_sdk/core/_base_api.py     |   1 +
 .../zhipuai/zhipuai_sdk/core/_base_compat.py  | 209 +++++
 .../zhipuai/zhipuai_sdk/core/_base_models.py  | 671 ++++++++++++++++
 .../zhipuai/zhipuai_sdk/core/_base_type.py    |  99 ++-
 .../zhipuai/zhipuai_sdk/core/_constants.py    |  12 +
 .../zhipuai/zhipuai_sdk/core/_errors.py       |  17 +-
 .../zhipuai/zhipuai_sdk/core/_files.py        |  78 +-
 .../zhipuai/zhipuai_sdk/core/_http_client.py  | 756 +++++++++++++++---
 .../zhipuai/zhipuai_sdk/core/_jwt_token.py    |   6 +-
 .../core/_legacy_binary_response.py           | 207 +++++
 .../zhipuai_sdk/core/_legacy_response.py      | 341 ++++++++
 .../zhipuai/zhipuai_sdk/core/_request_opt.py  |  89 ++-
 .../zhipuai/zhipuai_sdk/core/_response.py     | 361 ++++++++-
 .../zhipuai/zhipuai_sdk/core/_sse_client.py   |  68 +-
 .../zhipuai/zhipuai_sdk/core/_utils.py        |  19 -
 .../zhipuai_sdk/core/_utils/__init__.py       |  52 ++
 .../zhipuai_sdk/core/_utils/_transform.py     | 383 +++++++++
 .../zhipuai_sdk/core/_utils/_typing.py        | 122 +++
 .../zhipuai/zhipuai_sdk/core/_utils/_utils.py | 409 ++++++++++
 .../zhipuai/zhipuai_sdk/core/logs.py          |  78 ++
 .../zhipuai/zhipuai_sdk/core/pagination.py    |  62 ++
 .../zhipuai_sdk/types/assistant/__init__.py   |   5 +
 .../types/assistant/assistant_completion.py   |  40 +
 .../assistant_conversation_params.py          |   7 +
 .../assistant/assistant_conversation_resp.py  |  29 +
 .../assistant/assistant_create_params.py      |  32 +
 .../types/assistant/assistant_support_resp.py |  21 +
 .../types/assistant/message/__init__.py       |   3 +
 .../assistant/message/message_content.py      |  13 +
 .../assistant/message/text_content_block.py   |  14 +
 .../tools/code_interpreter_delta_block.py     |  27 +
 .../message/tools/drawing_tool_delta_block.py |  21 +
 .../message/tools/function_delta_block.py     |  22 +
 .../message/tools/retrieval_delta_black.py    |  41 +
 .../assistant/message/tools/tools_type.py     |  16 +
 .../message/tools/web_browser_delta_block.py  |  48 ++
 .../assistant/message/tools_delta_block.py    |  16 +
 .../zhipuai/zhipuai_sdk/types/batch.py        |  82 ++
 .../zhipuai_sdk/types/batch_create_params.py  |  37 +
 .../zhipuai/zhipuai_sdk/types/batch_error.py  |  21 +
 .../zhipuai_sdk/types/batch_list_params.py    |  20 +
 .../zhipuai_sdk/types/batch_request_counts.py |  14 +
 .../types/chat/async_chat_completion.py       |   5 +-
 .../zhipuai_sdk/types/chat/chat_completion.py |   2 +-
 .../types/chat/chat_completion_chunk.py       |   6 +-
 .../types/chat/code_geex/code_geex_params.py  | 146 ++++
 .../zhipuai/zhipuai_sdk/types/embeddings.py   |   3 +-
 .../zhipuai_sdk/types/files/__init__.py       |   5 +
 .../types/files/file_create_params.py         |  38 +
 .../zhipuai_sdk/types/files/file_deleted.py   |  13 +
 .../types/{ => files}/file_object.py          |   4 +-
 .../zhipuai_sdk/types/files/upload_detail.py  |  13 +
 .../types/fine_tuning/fine_tuning_job.py      |   2 +-
 .../fine_tuning/fine_tuning_job_event.py      |   2 +-
 .../types/fine_tuning/models/__init__.py      |   1 +
 .../fine_tuning/models/fine_tuned_models.py   |  13 +
 .../zhipuai/zhipuai_sdk/types/image.py        |   2 +-
 .../zhipuai_sdk/types/knowledge/__init__.py   |   8 +
 .../types/knowledge/document/__init__.py      |   8 +
 .../types/knowledge/document/document.py      |  51 ++
 .../document/document_edit_params.py          |  29 +
 .../document/document_list_params.py          |  26 +
 .../knowledge/document/document_list_resp.py  |  11 +
 .../zhipuai_sdk/types/knowledge/knowledge.py  |  21 +
 .../knowledge/knowledge_create_params.py      |  30 +
 .../types/knowledge/knowledge_list_params.py  |  15 +
 .../types/knowledge/knowledge_list_resp.py    |  11 +
 .../types/knowledge/knowledge_used.py         |  21 +
 .../types/sensitive_word_check/__init__.py    |   3 +
 .../sensitive_word_check.py                   |  14 +
 .../zhipuai_sdk/types/tools/__init__.py       |   9 +
 .../types/tools/tools_web_search_params.py    |  35 +
 .../zhipuai_sdk/types/tools/web_search.py     |  71 ++
 .../types/tools/web_search_chunk.py           |  33 +
 .../zhipuai_sdk/types/video/__init__.py       |   3 +
 .../types/video/video_create_params.py        |  27 +
 .../zhipuai_sdk/types/video/video_object.py   |  30 +
 .../builtin/cogview/tools/cogview3.py         |  48 +-
 .../builtin/cogview/tools/cogview3.yaml       |  45 +-
 108 files changed, 6498 insertions(+), 390 deletions(-)
 create mode 100644 api/core/model_runtime/model_providers/zhipuai/zhipuai_sdk/api_resource/assistant/__init__.py
 create mode 100644 api/core/model_runtime/model_providers/zhipuai/zhipuai_sdk/api_resource/assistant/assistant.py
 create mode 100644 api/core/model_runtime/model_providers/zhipuai/zhipuai_sdk/api_resource/batches.py
 create mode 100644 api/core/model_runtime/model_providers/zhipuai/zhipuai_sdk/api_resource/fine_tuning/jobs/__init__.py
 rename api/core/model_runtime/model_providers/zhipuai/zhipuai_sdk/api_resource/fine_tuning/{ => jobs}/jobs.py (53%)
 create mode 100644 api/core/model_runtime/model_providers/zhipuai/zhipuai_sdk/api_resource/fine_tuning/models/__init__.py
 create mode 100644 api/core/model_runtime/model_providers/zhipuai/zhipuai_sdk/api_resource/fine_tuning/models/fine_tuned_models.py
 create mode 100644 api/core/model_runtime/model_providers/zhipuai/zhipuai_sdk/api_resource/knowledge/__init__.py
 create mode 100644 api/core/model_runtime/model_providers/zhipuai/zhipuai_sdk/api_resource/knowledge/document/__init__.py
 create mode 100644 api/core/model_runtime/model_providers/zhipuai/zhipuai_sdk/api_resource/knowledge/document/document.py
 create mode 100644 api/core/model_runtime/model_providers/zhipuai/zhipuai_sdk/api_resource/knowledge/knowledge.py
 create mode 100644 api/core/model_runtime/model_providers/zhipuai/zhipuai_sdk/api_resource/tools/__init__.py
 create mode 100644 api/core/model_runtime/model_providers/zhipuai/zhipuai_sdk/api_resource/tools/tools.py
 create mode 100644 api/core/model_runtime/model_providers/zhipuai/zhipuai_sdk/api_resource/videos/__init__.py
 create mode 100644 api/core/model_runtime/model_providers/zhipuai/zhipuai_sdk/api_resource/videos/videos.py
 create mode 100644 api/core/model_runtime/model_providers/zhipuai/zhipuai_sdk/core/_base_compat.py
 create mode 100644 api/core/model_runtime/model_providers/zhipuai/zhipuai_sdk/core/_base_models.py
 create mode 100644 api/core/model_runtime/model_providers/zhipuai/zhipuai_sdk/core/_constants.py
 create mode 100644 api/core/model_runtime/model_providers/zhipuai/zhipuai_sdk/core/_legacy_binary_response.py
 create mode 100644 api/core/model_runtime/model_providers/zhipuai/zhipuai_sdk/core/_legacy_response.py
 delete mode 100644 api/core/model_runtime/model_providers/zhipuai/zhipuai_sdk/core/_utils.py
 create mode 100644 api/core/model_runtime/model_providers/zhipuai/zhipuai_sdk/core/_utils/__init__.py
 create mode 100644 api/core/model_runtime/model_providers/zhipuai/zhipuai_sdk/core/_utils/_transform.py
 create mode 100644 api/core/model_runtime/model_providers/zhipuai/zhipuai_sdk/core/_utils/_typing.py
 create mode 100644 api/core/model_runtime/model_providers/zhipuai/zhipuai_sdk/core/_utils/_utils.py
 create mode 100644 api/core/model_runtime/model_providers/zhipuai/zhipuai_sdk/core/logs.py
 create mode 100644 api/core/model_runtime/model_providers/zhipuai/zhipuai_sdk/core/pagination.py
 create mode 100644 api/core/model_runtime/model_providers/zhipuai/zhipuai_sdk/types/assistant/__init__.py
 create mode 100644 api/core/model_runtime/model_providers/zhipuai/zhipuai_sdk/types/assistant/assistant_completion.py
 create mode 100644 api/core/model_runtime/model_providers/zhipuai/zhipuai_sdk/types/assistant/assistant_conversation_params.py
 create mode 100644 api/core/model_runtime/model_providers/zhipuai/zhipuai_sdk/types/assistant/assistant_conversation_resp.py
 create mode 100644 api/core/model_runtime/model_providers/zhipuai/zhipuai_sdk/types/assistant/assistant_create_params.py
 create mode 100644 api/core/model_runtime/model_providers/zhipuai/zhipuai_sdk/types/assistant/assistant_support_resp.py
 create mode 100644 api/core/model_runtime/model_providers/zhipuai/zhipuai_sdk/types/assistant/message/__init__.py
 create mode 100644 api/core/model_runtime/model_providers/zhipuai/zhipuai_sdk/types/assistant/message/message_content.py
 create mode 100644 api/core/model_runtime/model_providers/zhipuai/zhipuai_sdk/types/assistant/message/text_content_block.py
 create mode 100644 api/core/model_runtime/model_providers/zhipuai/zhipuai_sdk/types/assistant/message/tools/code_interpreter_delta_block.py
 create mode 100644 api/core/model_runtime/model_providers/zhipuai/zhipuai_sdk/types/assistant/message/tools/drawing_tool_delta_block.py
 create mode 100644 api/core/model_runtime/model_providers/zhipuai/zhipuai_sdk/types/assistant/message/tools/function_delta_block.py
 create mode 100644 api/core/model_runtime/model_providers/zhipuai/zhipuai_sdk/types/assistant/message/tools/retrieval_delta_black.py
 create mode 100644 api/core/model_runtime/model_providers/zhipuai/zhipuai_sdk/types/assistant/message/tools/tools_type.py
 create mode 100644 api/core/model_runtime/model_providers/zhipuai/zhipuai_sdk/types/assistant/message/tools/web_browser_delta_block.py
 create mode 100644 api/core/model_runtime/model_providers/zhipuai/zhipuai_sdk/types/assistant/message/tools_delta_block.py
 create mode 100644 api/core/model_runtime/model_providers/zhipuai/zhipuai_sdk/types/batch.py
 create mode 100644 api/core/model_runtime/model_providers/zhipuai/zhipuai_sdk/types/batch_create_params.py
 create mode 100644 api/core/model_runtime/model_providers/zhipuai/zhipuai_sdk/types/batch_error.py
 create mode 100644 api/core/model_runtime/model_providers/zhipuai/zhipuai_sdk/types/batch_list_params.py
 create mode 100644 api/core/model_runtime/model_providers/zhipuai/zhipuai_sdk/types/batch_request_counts.py
 create mode 100644 api/core/model_runtime/model_providers/zhipuai/zhipuai_sdk/types/chat/code_geex/code_geex_params.py
 create mode 100644 api/core/model_runtime/model_providers/zhipuai/zhipuai_sdk/types/files/__init__.py
 create mode 100644 api/core/model_runtime/model_providers/zhipuai/zhipuai_sdk/types/files/file_create_params.py
 create mode 100644 api/core/model_runtime/model_providers/zhipuai/zhipuai_sdk/types/files/file_deleted.py
 rename api/core/model_runtime/model_providers/zhipuai/zhipuai_sdk/types/{ => files}/file_object.py (86%)
 create mode 100644 api/core/model_runtime/model_providers/zhipuai/zhipuai_sdk/types/files/upload_detail.py
 create mode 100644 api/core/model_runtime/model_providers/zhipuai/zhipuai_sdk/types/fine_tuning/models/__init__.py
 create mode 100644 api/core/model_runtime/model_providers/zhipuai/zhipuai_sdk/types/fine_tuning/models/fine_tuned_models.py
 create mode 100644 api/core/model_runtime/model_providers/zhipuai/zhipuai_sdk/types/knowledge/__init__.py
 create mode 100644 api/core/model_runtime/model_providers/zhipuai/zhipuai_sdk/types/knowledge/document/__init__.py
 create mode 100644 api/core/model_runtime/model_providers/zhipuai/zhipuai_sdk/types/knowledge/document/document.py
 create mode 100644 api/core/model_runtime/model_providers/zhipuai/zhipuai_sdk/types/knowledge/document/document_edit_params.py
 create mode 100644 api/core/model_runtime/model_providers/zhipuai/zhipuai_sdk/types/knowledge/document/document_list_params.py
 create mode 100644 api/core/model_runtime/model_providers/zhipuai/zhipuai_sdk/types/knowledge/document/document_list_resp.py
 create mode 100644 api/core/model_runtime/model_providers/zhipuai/zhipuai_sdk/types/knowledge/knowledge.py
 create mode 100644 api/core/model_runtime/model_providers/zhipuai/zhipuai_sdk/types/knowledge/knowledge_create_params.py
 create mode 100644 api/core/model_runtime/model_providers/zhipuai/zhipuai_sdk/types/knowledge/knowledge_list_params.py
 create mode 100644 api/core/model_runtime/model_providers/zhipuai/zhipuai_sdk/types/knowledge/knowledge_list_resp.py
 create mode 100644 api/core/model_runtime/model_providers/zhipuai/zhipuai_sdk/types/knowledge/knowledge_used.py
 create mode 100644 api/core/model_runtime/model_providers/zhipuai/zhipuai_sdk/types/sensitive_word_check/__init__.py
 create mode 100644 api/core/model_runtime/model_providers/zhipuai/zhipuai_sdk/types/sensitive_word_check/sensitive_word_check.py
 create mode 100644 api/core/model_runtime/model_providers/zhipuai/zhipuai_sdk/types/tools/__init__.py
 create mode 100644 api/core/model_runtime/model_providers/zhipuai/zhipuai_sdk/types/tools/tools_web_search_params.py
 create mode 100644 api/core/model_runtime/model_providers/zhipuai/zhipuai_sdk/types/tools/web_search.py
 create mode 100644 api/core/model_runtime/model_providers/zhipuai/zhipuai_sdk/types/tools/web_search_chunk.py
 create mode 100644 api/core/model_runtime/model_providers/zhipuai/zhipuai_sdk/types/video/__init__.py
 create mode 100644 api/core/model_runtime/model_providers/zhipuai/zhipuai_sdk/types/video/video_create_params.py
 create mode 100644 api/core/model_runtime/model_providers/zhipuai/zhipuai_sdk/types/video/video_object.py

diff --git a/api/core/model_runtime/model_providers/zhipuai/zhipuai_sdk/__init__.py b/api/core/model_runtime/model_providers/zhipuai/zhipuai_sdk/__init__.py
index bf9b093cb3f02..fc71d64714bd9 100644
--- a/api/core/model_runtime/model_providers/zhipuai/zhipuai_sdk/__init__.py
+++ b/api/core/model_runtime/model_providers/zhipuai/zhipuai_sdk/__init__.py
@@ -1,7 +1,8 @@
 from .__version__ import __version__
 from ._client import ZhipuAI
-from .core._errors import (
+from .core import (
     APIAuthenticationError,
+    APIConnectionError,
     APIInternalError,
     APIReachLimitError,
     APIRequestFailedError,
diff --git a/api/core/model_runtime/model_providers/zhipuai/zhipuai_sdk/__version__.py b/api/core/model_runtime/model_providers/zhipuai/zhipuai_sdk/__version__.py
index 659f38d7ff32d..51f8c49ecb827 100644
--- a/api/core/model_runtime/model_providers/zhipuai/zhipuai_sdk/__version__.py
+++ b/api/core/model_runtime/model_providers/zhipuai/zhipuai_sdk/__version__.py
@@ -1 +1 @@
-__version__ = "v2.0.1"
+__version__ = "v2.1.0"
diff --git a/api/core/model_runtime/model_providers/zhipuai/zhipuai_sdk/_client.py b/api/core/model_runtime/model_providers/zhipuai/zhipuai_sdk/_client.py
index df9e506095fab..705d371e628f0 100644
--- a/api/core/model_runtime/model_providers/zhipuai/zhipuai_sdk/_client.py
+++ b/api/core/model_runtime/model_providers/zhipuai/zhipuai_sdk/_client.py
@@ -9,15 +9,13 @@
 from typing_extensions import override
 
 from . import api_resource
-from .core import _jwt_token
-from .core._base_type import NOT_GIVEN, NotGiven
-from .core._errors import ZhipuAIError
-from .core._http_client import ZHIPUAI_DEFAULT_MAX_RETRIES, HttpClient
+from .core import NOT_GIVEN, ZHIPUAI_DEFAULT_MAX_RETRIES, HttpClient, NotGiven, ZhipuAIError, _jwt_token
 
 
 class ZhipuAI(HttpClient):
-    chat: api_resource.chat
+    chat: api_resource.chat.Chat
     api_key: str
+    _disable_token_cache: bool = True
 
     def __init__(
         self,
@@ -28,10 +26,15 @@ def __init__(
         max_retries: int = ZHIPUAI_DEFAULT_MAX_RETRIES,
         http_client: httpx.Client | None = None,
         custom_headers: Mapping[str, str] | None = None,
+        disable_token_cache: bool = True,
+        _strict_response_validation: bool = False,
     ) -> None:
         if api_key is None:
-            raise ZhipuAIError("No api_key provided, please provide it through parameters or environment variables")
+            api_key = os.environ.get("ZHIPUAI_API_KEY")
+        if api_key is None:
+            raise ZhipuAIError("未提供api_key，请通过参数或环境变量提供")
         self.api_key = api_key
+        self._disable_token_cache = disable_token_cache
 
         if base_url is None:
             base_url = os.environ.get("ZHIPUAI_BASE_URL")
@@ -42,21 +45,31 @@ def __init__(
         super().__init__(
             version=__version__,
             base_url=base_url,
+            max_retries=max_retries,
             timeout=timeout,
             custom_httpx_client=http_client,
             custom_headers=custom_headers,
+            _strict_response_validation=_strict_response_validation,
         )
         self.chat = api_resource.chat.Chat(self)
         self.images = api_resource.images.Images(self)
         self.embeddings = api_resource.embeddings.Embeddings(self)
         self.files = api_resource.files.Files(self)
         self.fine_tuning = api_resource.fine_tuning.FineTuning(self)
+        self.batches = api_resource.Batches(self)
+        self.knowledge = api_resource.Knowledge(self)
+        self.tools = api_resource.Tools(self)
+        self.videos = api_resource.Videos(self)
+        self.assistant = api_resource.Assistant(self)
 
     @property
     @override
-    def _auth_headers(self) -> dict[str, str]:
+    def auth_headers(self) -> dict[str, str]:
         api_key = self.api_key
-        return {"Authorization": f"{_jwt_token.generate_token(api_key)}"}
+        if self._disable_token_cache:
+            return {"Authorization": f"Bearer {api_key}"}
+        else:
+            return {"Authorization": f"Bearer {_jwt_token.generate_token(api_key)}"}
 
     def __del__(self) -> None:
         if not hasattr(self, "_has_custom_http_client") or not hasattr(self, "close") or not hasattr(self, "_client"):
diff --git a/api/core/model_runtime/model_providers/zhipuai/zhipuai_sdk/api_resource/__init__.py b/api/core/model_runtime/model_providers/zhipuai/zhipuai_sdk/api_resource/__init__.py
index 0a90e21e48bcc..4fe0719dde3e0 100644
--- a/api/core/model_runtime/model_providers/zhipuai/zhipuai_sdk/api_resource/__init__.py
+++ b/api/core/model_runtime/model_providers/zhipuai/zhipuai_sdk/api_resource/__init__.py
@@ -1,5 +1,34 @@
-from .chat import chat
+from .assistant import (
+    Assistant,
+)
+from .batches import Batches
+from .chat import (
+    AsyncCompletions,
+    Chat,
+    Completions,
+)
 from .embeddings import Embeddings
-from .files import Files
-from .fine_tuning import fine_tuning
+from .files import Files, FilesWithRawResponse
+from .fine_tuning import FineTuning
 from .images import Images
+from .knowledge import Knowledge
+from .tools import Tools
+from .videos import (
+    Videos,
+)
+
+__all__ = [
+    "Videos",
+    "AsyncCompletions",
+    "Chat",
+    "Completions",
+    "Images",
+    "Embeddings",
+    "Files",
+    "FilesWithRawResponse",
+    "FineTuning",
+    "Batches",
+    "Knowledge",
+    "Tools",
+    "Assistant",
+]
diff --git a/api/core/model_runtime/model_providers/zhipuai/zhipuai_sdk/api_resource/assistant/__init__.py b/api/core/model_runtime/model_providers/zhipuai/zhipuai_sdk/api_resource/assistant/__init__.py
new file mode 100644
index 0000000000000..ce619aa7f0922
--- /dev/null
+++ b/api/core/model_runtime/model_providers/zhipuai/zhipuai_sdk/api_resource/assistant/__init__.py
@@ -0,0 +1,3 @@
+from .assistant import Assistant
+
+__all__ = ["Assistant"]
diff --git a/api/core/model_runtime/model_providers/zhipuai/zhipuai_sdk/api_resource/assistant/assistant.py b/api/core/model_runtime/model_providers/zhipuai/zhipuai_sdk/api_resource/assistant/assistant.py
new file mode 100644
index 0000000000000..f772340a82c4b
--- /dev/null
+++ b/api/core/model_runtime/model_providers/zhipuai/zhipuai_sdk/api_resource/assistant/assistant.py
@@ -0,0 +1,122 @@
+from __future__ import annotations
+
+from typing import TYPE_CHECKING, Optional
+
+import httpx
+
+from ...core import (
+    NOT_GIVEN,
+    BaseAPI,
+    Body,
+    Headers,
+    NotGiven,
+    StreamResponse,
+    deepcopy_minimal,
+    make_request_options,
+    maybe_transform,
+)
+from ...types.assistant import AssistantCompletion
+from ...types.assistant.assistant_conversation_resp import ConversationUsageListResp
+from ...types.assistant.assistant_support_resp import AssistantSupportResp
+
+if TYPE_CHECKING:
+    from ..._client import ZhipuAI
+
+from ...types.assistant import assistant_conversation_params, assistant_create_params
+
+__all__ = ["Assistant"]
+
+
+class Assistant(BaseAPI):
+    def __init__(self, client: ZhipuAI) -> None:
+        super().__init__(client)
+
+    def conversation(
+        self,
+        assistant_id: str,
+        model: str,
+        messages: list[assistant_create_params.ConversationMessage],
+        *,
+        stream: bool = True,
+        conversation_id: Optional[str] = None,
+        attachments: Optional[list[assistant_create_params.AssistantAttachments]] = None,
+        metadata: dict | None = None,
+        request_id: str = None,
+        user_id: str = None,
+        extra_headers: Headers | None = None,
+        extra_body: Body | None = None,
+        timeout: float | httpx.Timeout | None | NotGiven = NOT_GIVEN,
+    ) -> StreamResponse[AssistantCompletion]:
+        body = deepcopy_minimal(
+            {
+                "assistant_id": assistant_id,
+                "model": model,
+                "messages": messages,
+                "stream": stream,
+                "conversation_id": conversation_id,
+                "attachments": attachments,
+                "metadata": metadata,
+                "request_id": request_id,
+                "user_id": user_id,
+            }
+        )
+        return self._post(
+            "/assistant",
+            body=maybe_transform(body, assistant_create_params.AssistantParameters),
+            options=make_request_options(extra_headers=extra_headers, extra_body=extra_body, timeout=timeout),
+            cast_type=AssistantCompletion,
+            stream=stream or True,
+            stream_cls=StreamResponse[AssistantCompletion],
+        )
+
+    def query_support(
+        self,
+        *,
+        assistant_id_list: list[str] = None,
+        request_id: str = None,
+        user_id: str = None,
+        extra_headers: Headers | None = None,
+        extra_body: Body | None = None,
+        timeout: float | httpx.Timeout | None | NotGiven = NOT_GIVEN,
+    ) -> AssistantSupportResp:
+        body = deepcopy_minimal(
+            {
+                "assistant_id_list": assistant_id_list,
+                "request_id": request_id,
+                "user_id": user_id,
+            }
+        )
+        return self._post(
+            "/assistant/list",
+            body=body,
+            options=make_request_options(extra_headers=extra_headers, extra_body=extra_body, timeout=timeout),
+            cast_type=AssistantSupportResp,
+        )
+
+    def query_conversation_usage(
+        self,
+        assistant_id: str,
+        page: int = 1,
+        page_size: int = 10,
+        *,
+        request_id: str = None,
+        user_id: str = None,
+        extra_headers: Headers | None = None,
+        extra_body: Body | None = None,
+        timeout: float | httpx.Timeout | None | NotGiven = NOT_GIVEN,
+    ) -> ConversationUsageListResp:
+        body = deepcopy_minimal(
+            {
+                "assistant_id": assistant_id,
+                "page": page,
+                "page_size": page_size,
+                "request_id": request_id,
+                "user_id": user_id,
+            }
+        )
+        return self._post(
+            "/assistant/conversation/list",
+            body=maybe_transform(body, assistant_conversation_params.ConversationParameters),
+            options=make_request_options(extra_headers=extra_headers, extra_body=extra_body, timeout=timeout),
+            cast_type=ConversationUsageListResp,
+        )
diff --git a/api/core/model_runtime/model_providers/zhipuai/zhipuai_sdk/api_resource/batches.py b/api/core/model_runtime/model_providers/zhipuai/zhipuai_sdk/api_resource/batches.py
new file mode 100644
index 0000000000000..ae2f2be85eb9b
--- /dev/null
+++ b/api/core/model_runtime/model_providers/zhipuai/zhipuai_sdk/api_resource/batches.py
@@ -0,0 +1,146 @@
+from __future__ import annotations
+
+from typing import TYPE_CHECKING, Literal, Optional
+
+import httpx
+
+from ..core import NOT_GIVEN, BaseAPI, Body, Headers, NotGiven, make_request_options, maybe_transform
+from ..core.pagination import SyncCursorPage
+from ..types import batch_create_params, batch_list_params
+from ..types.batch import Batch
+
+if TYPE_CHECKING:
+    from .._client import ZhipuAI
+
+
+class Batches(BaseAPI):
+    def __init__(self, client: ZhipuAI) -> None:
+        super().__init__(client)
+
+    def create(
+        self,
+        *,
+        completion_window: str | None = None,
+        endpoint: Literal["/v1/chat/completions", "/v1/embeddings"],
+        input_file_id: str,
+        metadata: Optional[dict[str, str]] | NotGiven = NOT_GIVEN,
+        auto_delete_input_file: bool = True,
+        extra_headers: Headers | None = None,
+        extra_body: Body | None = None,
+        timeout: float | httpx.Timeout | None | NotGiven = NOT_GIVEN,
+    ) -> Batch:
+        return self._post(
+            "/batches",
+            body=maybe_transform(
+                {
+                    "completion_window": completion_window,
+                    "endpoint": endpoint,
+                    "input_file_id": input_file_id,
+                    "metadata": metadata,
+                    "auto_delete_input_file": auto_delete_input_file,
+                },
+                batch_create_params.BatchCreateParams,
+            ),
+            options=make_request_options(extra_headers=extra_headers, extra_body=extra_body, timeout=timeout),
+            cast_type=Batch,
+        )
+
+    def retrieve(
+        self,
+        batch_id: str,
+        *,
+        extra_headers: Headers | None = None,
+        extra_body: Body | None = None,
+        timeout: float | httpx.Timeout | None | NotGiven = NOT_GIVEN,
+    ) -> Batch:
+        """
+        Retrieves a batch.
+
+        Args:
+          extra_headers: Send extra headers
+
+          extra_body: Add additional JSON properties to the request
+
+          timeout: Override the client-level default timeout for this request, in seconds
+        """
+        if not batch_id:
+            raise ValueError(f"Expected a non-empty value for `batch_id` but received {batch_id!r}")
+        return self._get(
+            f"/batches/{batch_id}",
+            options=make_request_options(extra_headers=extra_headers, extra_body=extra_body, timeout=timeout),
+            cast_type=Batch,
+        )
+
+    def list(
+        self,
+        *,
+        after: str | NotGiven = NOT_GIVEN,
+        limit: int | NotGiven = NOT_GIVEN,
+        extra_headers: Headers | None = None,
+        extra_body: Body | None = None,
+        timeout: float | httpx.Timeout | None | NotGiven = NOT_GIVEN,
+    ) -> SyncCursorPage[Batch]:
+        """List your organization's batches.
+
+        Args:
+          after: A cursor for use in pagination.
+
+            `after` is an object ID that defines your place
+              in the list. For instance, if you make a list request and receive 100 objects,
+              ending with obj_foo, your subsequent call can include after=obj_foo in order to
+              fetch the next page of the list.
+
+          limit: A limit on the number of objects to be returned. Limit can range between 1 and
+              100, and the default is 20.
+
+          extra_headers: Send extra headers
+
+          extra_body: Add additional JSON properties to the request
+
+          timeout: Override the client-level default timeout for this request, in seconds
+        """
+        return self._get_api_list(
+            "/batches",
+            page=SyncCursorPage[Batch],
+            options=make_request_options(
+                extra_headers=extra_headers,
+                extra_body=extra_body,
+                timeout=timeout,
+                query=maybe_transform(
+                    {
+                        "after": after,
+                        "limit": limit,
+                    },
+                    batch_list_params.BatchListParams,
+                ),
+            ),
+            model=Batch,
+        )
+
+    def cancel(
+        self,
+        batch_id: str,
+        *,
+        extra_headers: Headers | None = None,
+        extra_body: Body | None = None,
+        timeout: float | httpx.Timeout | None | NotGiven = NOT_GIVEN,
+    ) -> Batch:
+        """
+        Cancels an in-progress batch.
+
+        Args:
+          batch_id: The ID of the batch to cancel.
+          extra_headers: Send extra headers
+
+          extra_body: Add additional JSON properties to the request
+
+          timeout: Override the client-level default timeout for this request, in seconds
+
+        """
+        if not batch_id:
+            raise ValueError(f"Expected a non-empty value for `batch_id` but received {batch_id!r}")
+        return self._post(
+            f"/batches/{batch_id}/cancel",
+            options=make_request_options(extra_headers=extra_headers, extra_body=extra_body, timeout=timeout),
+            cast_type=Batch,
+        )
diff --git a/api/core/model_runtime/model_providers/zhipuai/zhipuai_sdk/api_resource/chat/__init__.py b/api/core/model_runtime/model_providers/zhipuai/zhipuai_sdk/api_resource/chat/__init__.py
index e69de29bb2d1d..5cd8dc6f339a6 100644
--- a/api/core/model_runtime/model_providers/zhipuai/zhipuai_sdk/api_resource/chat/__init__.py
+++ b/api/core/model_runtime/model_providers/zhipuai/zhipuai_sdk/api_resource/chat/__init__.py
@@ -0,0 +1,5 @@
+from .async_completions import AsyncCompletions
+from .chat import Chat
+from .completions import Completions
+
+__all__ = ["AsyncCompletions", "Chat", "Completions"]
diff --git a/api/core/model_runtime/model_providers/zhipuai/zhipuai_sdk/api_resource/chat/async_completions.py b/api/core/model_runtime/model_providers/zhipuai/zhipuai_sdk/api_resource/chat/async_completions.py
index 1f8011973951b..d8ecc310644d1 100644
--- a/api/core/model_runtime/model_providers/zhipuai/zhipuai_sdk/api_resource/chat/async_completions.py
+++ b/api/core/model_runtime/model_providers/zhipuai/zhipuai_sdk/api_resource/chat/async_completions.py
@@ -1,13 +1,25 @@
 from __future__ import annotations
 
+import logging
 from typing import TYPE_CHECKING, Literal, Optional, Union
 
 import httpx
 
-from ...core._base_api import BaseAPI
-from ...core._base_type import NOT_GIVEN, Headers, NotGiven
-from ...core._http_client import make_user_request_input
+from ...core import (
+    NOT_GIVEN,
+    BaseAPI,
+    Body,
+    Headers,
+    NotGiven,
+    drop_prefix_image_data,
+    make_request_options,
+    maybe_transform,
+)
 from ...types.chat.async_chat_completion import AsyncCompletion, AsyncTaskStatus
+from ...types.chat.code_geex import code_geex_params
+from ...types.sensitive_word_check import SensitiveWordCheckRequest
+
+logger = logging.getLogger(__name__)
 
 if TYPE_CHECKING:
     from ..._client import ZhipuAI
@@ -22,6 +34,7 @@ def create(
         *,
         model: str,
         request_id: Optional[str] | NotGiven = NOT_GIVEN,
+        user_id: Optional[str] | NotGiven = NOT_GIVEN,
         do_sample: Optional[Literal[False]] | Literal[True] | NotGiven = NOT_GIVEN,
         temperature: Optional[float] | NotGiven = NOT_GIVEN,
         top_p: Optional[float] | NotGiven = NOT_GIVEN,
@@ -29,50 +42,74 @@ def create(
         seed: int | NotGiven = NOT_GIVEN,
         messages: Union[str, list[str], list[int], list[list[int]], None],
         stop: Optional[Union[str, list[str], None]] | NotGiven = NOT_GIVEN,
-        sensitive_word_check: Optional[object] | NotGiven = NOT_GIVEN,
+        sensitive_word_check: Optional[SensitiveWordCheckRequest] | NotGiven = NOT_GIVEN,
         tools: Optional[object] | NotGiven = NOT_GIVEN,
         tool_choice: str | NotGiven = NOT_GIVEN,
+        meta: Optional[dict[str, str]] | NotGiven = NOT_GIVEN,
+        extra: Optional[code_geex_params.CodeGeexExtra] | NotGiven = NOT_GIVEN,
         extra_headers: Headers | None = None,
-        disable_strict_validation: Optional[bool] | None = None,
+        extra_body: Body | None = None,
         timeout: float | httpx.Timeout | None | NotGiven = NOT_GIVEN,
     ) -> AsyncTaskStatus:
         _cast_type = AsyncTaskStatus
+        logger.debug(f"temperature:{temperature}, top_p:{top_p}")
+        if temperature is not None and temperature != NOT_GIVEN:
+            if temperature <= 0:
+                do_sample = False
+                temperature = 0.01
+                # logger.warning("temperature:取值范围是：(0.0, 1.0) 开区间，do_sample重写为:false（参数top_p temperture不生效）")  # noqa: E501
+            if temperature >= 1:
+                temperature = 0.99
+                # logger.warning("temperature:取值范围是：(0.0, 1.0) 开区间")
+        if top_p is not None and top_p != NOT_GIVEN:
+            if top_p >= 1:
+                top_p = 0.99
+                # logger.warning("top_p:取值范围是：(0.0, 1.0) 开区间，不能等于 0 或 1")
+            if top_p <= 0:
+                top_p = 0.01
+                # logger.warning("top_p:取值范围是：(0.0, 1.0) 开区间，不能等于 0 或 1")
+
+        logger.debug(f"temperature:{temperature}, top_p:{top_p}")
+        if isinstance(messages, list):
+            for item in messages:
+                if item.get("content"):
+                    item["content"] = drop_prefix_image_data(item["content"])
 
-        if disable_strict_validation:
-            _cast_type = object
+        body = {
+            "model": model,
+            "request_id": request_id,
+            "user_id": user_id,
+            "temperature": temperature,
+            "top_p": top_p,
+            "do_sample": do_sample,
+            "max_tokens": max_tokens,
+            "seed": seed,
+            "messages": messages,
+            "stop": stop,
+            "sensitive_word_check": sensitive_word_check,
+            "tools": tools,
+            "tool_choice": tool_choice,
+            "meta": meta,
+            "extra": maybe_transform(extra, code_geex_params.CodeGeexExtra),
+        }
         return self._post(
             "/async/chat/completions",
-            body={
-                "model": model,
-                "request_id": request_id,
-                "temperature": temperature,
-                "top_p": top_p,
-                "do_sample": do_sample,
-                "max_tokens": max_tokens,
-                "seed": seed,
-                "messages": messages,
-                "stop": stop,
-                "sensitive_word_check": sensitive_word_check,
-                "tools": tools,
-                "tool_choice": tool_choice,
-            },
-            options=make_user_request_input(extra_headers=extra_headers, timeout=timeout),
+            body=body,
+            options=make_request_options(extra_headers=extra_headers, extra_body=extra_body, timeout=timeout),
             cast_type=_cast_type,
-            enable_stream=False,
+            stream=False,
         )
 
     def retrieve_completion_result(
         self,
         id: str,
         extra_headers: Headers | None = None,
-        disable_strict_validation: Optional[bool] | None = None,
+        extra_body: Body | None = None,
         timeout: float | httpx.Timeout | None | NotGiven = NOT_GIVEN,
     ) -> Union[AsyncCompletion, AsyncTaskStatus]:
         _cast_type = Union[AsyncCompletion, AsyncTaskStatus]
-        if disable_strict_validation:
-            _cast_type = object
         return self._get(
             path=f"/async-result/{id}",
             cast_type=_cast_type,
-            options=make_user_request_input(extra_headers=extra_headers, timeout=timeout),
+            options=make_request_options(extra_headers=extra_headers, extra_body=extra_body, timeout=timeout),
         )
diff --git a/api/core/model_runtime/model_providers/zhipuai/zhipuai_sdk/api_resource/chat/chat.py b/api/core/model_runtime/model_providers/zhipuai/zhipuai_sdk/api_resource/chat/chat.py
index 92362fc50a725..b3cc46566c7bf 100644
--- a/api/core/model_runtime/model_providers/zhipuai/zhipuai_sdk/api_resource/chat/chat.py
+++ b/api/core/model_runtime/model_providers/zhipuai/zhipuai_sdk/api_resource/chat/chat.py
@@ -1,17 +1,18 @@
 from typing import TYPE_CHECKING
 
-from ...core._base_api import BaseAPI
+from ...core import BaseAPI, cached_property
 from .async_completions import AsyncCompletions
 from .completions import Completions
 
 if TYPE_CHECKING:
-    from ..._client import ZhipuAI
+    pass
 
 
 class Chat(BaseAPI):
-    completions: Completions
+    @cached_property
+    def completions(self) -> Completions:
+        return Completions(self._client)
 
-    def __init__(self, client: "ZhipuAI") -> None:
-        super().__init__(client)
-        self.completions = Completions(client)
-        self.asyncCompletions = AsyncCompletions(client)
+    @cached_property
+    def asyncCompletions(self) -> AsyncCompletions:  # noqa: N802
+        return AsyncCompletions(self._client)
diff --git a/api/core/model_runtime/model_providers/zhipuai/zhipuai_sdk/api_resource/chat/completions.py b/api/core/model_runtime/model_providers/zhipuai/zhipuai_sdk/api_resource/chat/completions.py
index ec29f33864203..1c23473a03ae3 100644
--- a/api/core/model_runtime/model_providers/zhipuai/zhipuai_sdk/api_resource/chat/completions.py
+++ b/api/core/model_runtime/model_providers/zhipuai/zhipuai_sdk/api_resource/chat/completions.py
@@ -1,15 +1,28 @@
 from __future__ import annotations
 
+import logging
 from typing import TYPE_CHECKING, Literal, Optional, Union
 
 import httpx
 
-from ...core._base_api import BaseAPI
-from ...core._base_type import NOT_GIVEN, Headers, NotGiven
-from ...core._http_client import make_user_request_input
-from ...core._sse_client import StreamResponse
+from ...core import (
+    NOT_GIVEN,
+    BaseAPI,
+    Body,
+    Headers,
+    NotGiven,
+    StreamResponse,
+    deepcopy_minimal,
+    drop_prefix_image_data,
+    make_request_options,
+    maybe_transform,
+)
 from ...types.chat.chat_completion import Completion
 from ...types.chat.chat_completion_chunk import ChatCompletionChunk
+from ...types.chat.code_geex import code_geex_params
+from ...types.sensitive_word_check import SensitiveWordCheckRequest
+
+logger = logging.getLogger(__name__)
 
 if TYPE_CHECKING:
     from ..._client import ZhipuAI
@@ -24,6 +37,7 @@ def create(
         *,
         model: str,
         request_id: Optional[str] | NotGiven = NOT_GIVEN,
+        user_id: Optional[str] | NotGiven = NOT_GIVEN,
         do_sample: Optional[Literal[False]] | Literal[True] | NotGiven = NOT_GIVEN,
         stream: Optional[Literal[False]] | Literal[True] | NotGiven = NOT_GIVEN,
         temperature: Optional[float] | NotGiven = NOT_GIVEN,
@@ -32,23 +46,43 @@ def create(
         seed: int | NotGiven = NOT_GIVEN,
         messages: Union[str, list[str], list[int], object, None],
         stop: Optional[Union[str, list[str], None]] | NotGiven = NOT_GIVEN,
-        sensitive_word_check: Optional[object] | NotGiven = NOT_GIVEN,
+        sensitive_word_check: Optional[SensitiveWordCheckRequest] | NotGiven = NOT_GIVEN,
         tools: Optional[object] | NotGiven = NOT_GIVEN,
         tool_choice: str | NotGiven = NOT_GIVEN,
+        meta: Optional[dict[str, str]] | NotGiven = NOT_GIVEN,
+        extra: Optional[code_geex_params.CodeGeexExtra] | NotGiven = NOT_GIVEN,
         extra_headers: Headers | None = None,
-        disable_strict_validation: Optional[bool] | None = None,
+        extra_body: Body | None = None,
         timeout: float | httpx.Timeout | None | NotGiven = NOT_GIVEN,
     ) -> Completion | StreamResponse[ChatCompletionChunk]:
-        _cast_type = Completion
-        _stream_cls = StreamResponse[ChatCompletionChunk]
-        if disable_strict_validation:
-            _cast_type = object
-            _stream_cls = StreamResponse[object]
-        return self._post(
-            "/chat/completions",
-            body={
+        logger.debug(f"temperature:{temperature}, top_p:{top_p}")
+        if temperature is not None and temperature != NOT_GIVEN:
+            if temperature <= 0:
+                do_sample = False
+                temperature = 0.01
+                # logger.warning("temperature:取值范围是：(0.0, 1.0) 开区间，do_sample重写为:false（参数top_p temperture不生效）")  # noqa: E501
+            if temperature >= 1:
+                temperature = 0.99
+                # logger.warning("temperature:取值范围是：(0.0, 1.0) 开区间")
+        if top_p is not None and top_p != NOT_GIVEN:
+            if top_p >= 1:
+                top_p = 0.99
+                # logger.warning("top_p:取值范围是：(0.0, 1.0) 开区间，不能等于 0 或 1")
+            if top_p <= 0:
+                top_p = 0.01
+                # logger.warning("top_p:取值范围是：(0.0, 1.0) 开区间，不能等于 0 或 1")
+
+        logger.debug(f"temperature:{temperature}, top_p:{top_p}")
+        if isinstance(messages, list):
+            for item in messages:
+                if item.get("content"):
+                    item["content"] = drop_prefix_image_data(item["content"])
+
+        body = deepcopy_minimal(
+            {
                 "model": model,
                 "request_id": request_id,
+                "user_id": user_id,
                 "temperature": temperature,
                 "top_p": top_p,
                 "do_sample": do_sample,
@@ -60,11 +94,15 @@ def create(
                 "stream": stream,
                 "tools": tools,
                 "tool_choice": tool_choice,
-            },
-            options=make_user_request_input(
-                extra_headers=extra_headers,
-            ),
-            cast_type=_cast_type,
-            enable_stream=stream or False,
-            stream_cls=_stream_cls,
+                "meta": meta,
+                "extra": maybe_transform(extra, code_geex_params.CodeGeexExtra),
+            }
+        )
+        return self._post(
+            "/chat/completions",
+            body=body,
+            options=make_request_options(extra_headers=extra_headers, extra_body=extra_body, timeout=timeout),
+            cast_type=Completion,
+            stream=stream or False,
+            stream_cls=StreamResponse[ChatCompletionChunk],
         )
diff --git a/api/core/model_runtime/model_providers/zhipuai/zhipuai_sdk/api_resource/embeddings.py b/api/core/model_runtime/model_providers/zhipuai/zhipuai_sdk/api_resource/embeddings.py
index 2308a204514e1..4b4baef9421ba 100644
--- a/api/core/model_runtime/model_providers/zhipuai/zhipuai_sdk/api_resource/embeddings.py
+++ b/api/core/model_runtime/model_providers/zhipuai/zhipuai_sdk/api_resource/embeddings.py
@@ -4,9 +4,7 @@
 
 import httpx
 
-from ..core._base_api import BaseAPI
-from ..core._base_type import NOT_GIVEN, Headers, NotGiven
-from ..core._http_client import make_user_request_input
+from ..core import NOT_GIVEN, BaseAPI, Body, Headers, NotGiven, make_request_options
 from ..types.embeddings import EmbeddingsResponded
 
 if TYPE_CHECKING:
@@ -22,10 +20,13 @@ def create(
         *,
         input: Union[str, list[str], list[int], list[list[int]]],
         model: Union[str],
+        dimensions: Union[int] | NotGiven = NOT_GIVEN,
         encoding_format: str | NotGiven = NOT_GIVEN,
         user: str | NotGiven = NOT_GIVEN,
+        request_id: Optional[str] | NotGiven = NOT_GIVEN,
         sensitive_word_check: Optional[object] | NotGiven = NOT_GIVEN,
         extra_headers: Headers | None = None,
+        extra_body: Body | None = None,
         disable_strict_validation: Optional[bool] | None = None,
         timeout: float | httpx.Timeout | None | NotGiven = NOT_GIVEN,
     ) -> EmbeddingsResponded:
@@ -37,11 +38,13 @@ def create(
             body={
                 "input": input,
                 "model": model,
+                "dimensions": dimensions,
                 "encoding_format": encoding_format,
                 "user": user,
+                "request_id": request_id,
                 "sensitive_word_check": sensitive_word_check,
             },
-            options=make_user_request_input(extra_headers=extra_headers, timeout=timeout),
+            options=make_request_options(extra_headers=extra_headers, extra_body=extra_body, timeout=timeout),
             cast_type=_cast_type,
-            enable_stream=False,
+            stream=False,
         )
diff --git a/api/core/model_runtime/model_providers/zhipuai/zhipuai_sdk/api_resource/files.py b/api/core/model_runtime/model_providers/zhipuai/zhipuai_sdk/api_resource/files.py
index f2ac74bffa843..ba9de75b7ef09 100644
--- a/api/core/model_runtime/model_providers/zhipuai/zhipuai_sdk/api_resource/files.py
+++ b/api/core/model_runtime/model_providers/zhipuai/zhipuai_sdk/api_resource/files.py
@@ -1,19 +1,30 @@
 from __future__ import annotations
 
-from typing import TYPE_CHECKING
+from collections.abc import Mapping
+from typing import TYPE_CHECKING, Literal, cast
 
 import httpx
 
-from ..core._base_api import BaseAPI
-from ..core._base_type import NOT_GIVEN, FileTypes, Headers, NotGiven
-from ..core._files import is_file_content
-from ..core._http_client import make_user_request_input
-from ..types.file_object import FileObject, ListOfFileObject
+from ..core import (
+    NOT_GIVEN,
+    BaseAPI,
+    Body,
+    FileTypes,
+    Headers,
+    NotGiven,
+    _legacy_binary_response,
+    _legacy_response,
+    deepcopy_minimal,
+    extract_files,
+    make_request_options,
+    maybe_transform,
+)
+from ..types.files import FileDeleted, FileObject, ListOfFileObject, UploadDetail, file_create_params
 
 if TYPE_CHECKING:
     from .._client import ZhipuAI
 
-__all__ = ["Files"]
+__all__ = ["Files", "FilesWithRawResponse"]
 
 
 class Files(BaseAPI):
@@ -23,30 +34,69 @@ def __init__(self, client: ZhipuAI) -> None:
     def create(
         self,
         *,
-        file: FileTypes,
-        purpose: str,
+        file: FileTypes = None,
+        upload_detail: list[UploadDetail] = None,
+        purpose: Literal["fine-tune", "retrieval", "batch"],
+        knowledge_id: str = None,
+        sentence_size: int = None,
         extra_headers: Headers | None = None,
+        extra_body: Body | None = None,
         timeout: float | httpx.Timeout | None | NotGiven = NOT_GIVEN,
     ) -> FileObject:
-        if not is_file_content(file):
-            prefix = f"Expected file input `{file!r}`"
-            raise RuntimeError(
-                f"{prefix} to be bytes, an io.IOBase instance, PathLike or a tuple but received {type(file)} instead."
-            ) from None
-        files = [("file", file)]
-
-        extra_headers = {"Content-Type": "multipart/form-data", **(extra_headers or {})}
-
+        if not file and not upload_detail:
+            raise ValueError("At least one of `file` and `upload_detail` must be provided.")
+        body = deepcopy_minimal(
+            {
+                "file": file,
+                "upload_detail": upload_detail,
+                "purpose": purpose,
+                "knowledge_id": knowledge_id,
+                "sentence_size": sentence_size,
+            }
+        )
+        files = extract_files(cast(Mapping[str, object], body), paths=[["file"]])
+        if files:
+            # It should be noted that the actual Content-Type header that will be
+            # sent to the server will contain a `boundary` parameter, e.g.
+            # multipart/form-data; boundary=---abc--
+            extra_headers = {"Content-Type": "multipart/form-data", **(extra_headers or {})}
         return self._post(
             "/files",
-            body={
-                "purpose": purpose,
-            },
+            body=maybe_transform(body, file_create_params.FileCreateParams),
             files=files,
-            options=make_user_request_input(extra_headers=extra_headers, timeout=timeout),
+            options=make_request_options(extra_headers=extra_headers, extra_body=extra_body, timeout=timeout),
             cast_type=FileObject,
         )
 
+    # def retrieve(
+    #         self,
+    #         file_id: str,
+    #         *,
+    #         extra_headers: Headers | None = None,
+    #         extra_body: Body | None = None,
+    #         timeout: float | httpx.Timeout | None | NotGiven = NOT_GIVEN,
+    # ) -> FileObject:
+    #     """
+    #     Returns information about a specific file.
+    #
+    #     Args:
+    #       file_id: The ID of the file to retrieve information about
+    #       extra_headers: Send extra headers
+    #
+    #       extra_body: Add additional JSON properties to the request
+    #
+    #       timeout: Override the client-level default timeout for this request, in seconds
+    #     """
+    #     if not file_id:
+    #         raise ValueError(f"Expected a non-empty value for `file_id` but received {file_id!r}")
+    #     return self._get(
+    #         f"/files/{file_id}",
+    #         options=make_request_options(
+    #             extra_headers=extra_headers, extra_body=extra_body, timeout=timeout
+    #         ),
+    #         cast_type=FileObject,
+    #     )
+
     def list(
         self,
         *,
@@ -55,13 +105,15 @@ def list(
         after: str | NotGiven = NOT_GIVEN,
         order: str | NotGiven = NOT_GIVEN,
         extra_headers: Headers | None = None,
+        extra_body: Body | None = None,
         timeout: float | httpx.Timeout | None | NotGiven = NOT_GIVEN,
     ) -> ListOfFileObject:
         return self._get(
             "/files",
             cast_type=ListOfFileObject,
-            options=make_user_request_input(
+            options=make_request_options(
                 extra_headers=extra_headers,
+                extra_body=extra_body,
                 timeout=timeout,
                 query={
                     "purpose": purpose,
@@ -71,3 +123,72 @@ def list(
                 },
             ),
         )
+
+    def delete(
+        self,
+        file_id: str,
+        *,
+        extra_headers: Headers | None = None,
+        extra_body: Body | None = None,
+        timeout: float | httpx.Timeout | None | NotGiven = NOT_GIVEN,
+    ) -> FileDeleted:
+        """
+        Delete a file.
+
+        Args:
+          file_id: The ID of the file to delete
+          extra_headers: Send extra headers
+
+          extra_body: Add additional JSON properties to the request
+
+          timeout: Override the client-level default timeout for this request, in seconds
+        """
+        if not file_id:
+            raise ValueError(f"Expected a non-empty value for `file_id` but received {file_id!r}")
+        return self._delete(
+            f"/files/{file_id}",
+            options=make_request_options(extra_headers=extra_headers, extra_body=extra_body, timeout=timeout),
+            cast_type=FileDeleted,
+        )
+
+    def content(
+        self,
+        file_id: str,
+        *,
+        extra_headers: Headers | None = None,
+        extra_body: Body | None = None,
+        timeout: float | httpx.Timeout | None | NotGiven = NOT_GIVEN,
+    ) -> _legacy_response.HttpxBinaryResponseContent:
+        """
+        Returns the contents of the specified file.
+
+        Args:
+          extra_headers: Send extra headers
+
+          extra_body: Add additional JSON properties to the request
+
+          timeout: Override the client-level default timeout for this request, in seconds
+        """
+        if not file_id:
+            raise ValueError(f"Expected a non-empty value for `file_id` but received {file_id!r}")
+        extra_headers = {"Accept": "application/binary", **(extra_headers or {})}
+        return self._get(
+            f"/files/{file_id}/content",
+            options=make_request_options(extra_headers=extra_headers, extra_body=extra_body, timeout=timeout),
+            cast_type=_legacy_binary_response.HttpxBinaryResponseContent,
+        )
+
+
+class FilesWithRawResponse:
+    def __init__(self, files: Files) -> None:
+        self._files = files
+
+        self.create = _legacy_response.to_raw_response_wrapper(
+            files.create,
+        )
+        self.list = _legacy_response.to_raw_response_wrapper(
+            files.list,
+        )
+        self.content = _legacy_response.to_raw_response_wrapper(
+            files.content,
+        )
diff --git a/api/core/model_runtime/model_providers/zhipuai/zhipuai_sdk/api_resource/fine_tuning/__init__.py b/api/core/model_runtime/model_providers/zhipuai/zhipuai_sdk/api_resource/fine_tuning/__init__.py
index e69de29bb2d1d..7c309b8341680 100644
--- a/api/core/model_runtime/model_providers/zhipuai/zhipuai_sdk/api_resource/fine_tuning/__init__.py
+++ b/api/core/model_runtime/model_providers/zhipuai/zhipuai_sdk/api_resource/fine_tuning/__init__.py
@@ -0,0 +1,5 @@
+from .fine_tuning import FineTuning
+from .jobs import Jobs
+from .models import FineTunedModels
+
+__all__ = ["Jobs", "FineTunedModels", "FineTuning"]
diff --git a/api/core/model_runtime/model_providers/zhipuai/zhipuai_sdk/api_resource/fine_tuning/fine_tuning.py b/api/core/model_runtime/model_providers/zhipuai/zhipuai_sdk/api_resource/fine_tuning/fine_tuning.py
index dc30bd33edfbb..8670f7de00df8 100644
--- a/api/core/model_runtime/model_providers/zhipuai/zhipuai_sdk/api_resource/fine_tuning/fine_tuning.py
+++ b/api/core/model_runtime/model_providers/zhipuai/zhipuai_sdk/api_resource/fine_tuning/fine_tuning.py
@@ -1,15 +1,18 @@
 from typing import TYPE_CHECKING
 
-from ...core._base_api import BaseAPI
+from ...core import BaseAPI, cached_property
 from .jobs import Jobs
+from .models import FineTunedModels
 
 if TYPE_CHECKING:
-    from ..._client import ZhipuAI
+    pass
 
 
 class FineTuning(BaseAPI):
-    jobs: Jobs
+    @cached_property
+    def jobs(self) -> Jobs:
+        return Jobs(self._client)
 
-    def __init__(self, client: "ZhipuAI") -> None:
-        super().__init__(client)
-        self.jobs = Jobs(client)
+    @cached_property
+    def models(self) -> FineTunedModels:
+        return FineTunedModels(self._client)
diff --git a/api/core/model_runtime/model_providers/zhipuai/zhipuai_sdk/api_resource/fine_tuning/jobs/__init__.py b/api/core/model_runtime/model_providers/zhipuai/zhipuai_sdk/api_resource/fine_tuning/jobs/__init__.py
new file mode 100644
index 0000000000000..40777a153f272
--- /dev/null
+++ b/api/core/model_runtime/model_providers/zhipuai/zhipuai_sdk/api_resource/fine_tuning/jobs/__init__.py
@@ -0,0 +1,3 @@
+from .jobs import Jobs
+
+__all__ = ["Jobs"]
diff --git a/api/core/model_runtime/model_providers/zhipuai/zhipuai_sdk/api_resource/fine_tuning/jobs.py b/api/core/model_runtime/model_providers/zhipuai/zhipuai_sdk/api_resource/fine_tuning/jobs/jobs.py
similarity index 53%
rename from api/core/model_runtime/model_providers/zhipuai/zhipuai_sdk/api_resource/fine_tuning/jobs.py
rename to api/core/model_runtime/model_providers/zhipuai/zhipuai_sdk/api_resource/fine_tuning/jobs/jobs.py
index 3d2e9208a11f1..8b038cadc0640 100644
--- a/api/core/model_runtime/model_providers/zhipuai/zhipuai_sdk/api_resource/fine_tuning/jobs.py
+++ b/api/core/model_runtime/model_providers/zhipuai/zhipuai_sdk/api_resource/fine_tuning/jobs/jobs.py
@@ -4,13 +4,23 @@
 
 import httpx
 
-from ...core._base_api import BaseAPI
-from ...core._base_type import NOT_GIVEN, Headers, NotGiven
-from ...core._http_client import make_user_request_input
-from ...types.fine_tuning import FineTuningJob, FineTuningJobEvent, ListOfFineTuningJob, job_create_params
+from ....core import (
+    NOT_GIVEN,
+    BaseAPI,
+    Body,
+    Headers,
+    NotGiven,
+    make_request_options,
+)
+from ....types.fine_tuning import (
+    FineTuningJob,
+    FineTuningJobEvent,
+    ListOfFineTuningJob,
+    job_create_params,
+)
 
 if TYPE_CHECKING:
-    from ..._client import ZhipuAI
+    from ...._client import ZhipuAI
 
 __all__ = ["Jobs"]
 
@@ -29,6 +39,7 @@ def create(
         request_id: Optional[str] | NotGiven = NOT_GIVEN,
         validation_file: Optional[str] | NotGiven = NOT_GIVEN,
         extra_headers: Headers | None = None,
+        extra_body: Body | None = None,
         timeout: float | httpx.Timeout | None | NotGiven = NOT_GIVEN,
     ) -> FineTuningJob:
         return self._post(
@@ -41,7 +52,7 @@ def create(
                 "validation_file": validation_file,
                 "request_id": request_id,
             },
-            options=make_user_request_input(extra_headers=extra_headers, timeout=timeout),
+            options=make_request_options(extra_headers=extra_headers, extra_body=extra_body, timeout=timeout),
             cast_type=FineTuningJob,
         )
 
@@ -50,11 +61,12 @@ def retrieve(
         fine_tuning_job_id: str,
         *,
         extra_headers: Headers | None = None,
+        extra_body: Body | None = None,
         timeout: float | httpx.Timeout | None | NotGiven = NOT_GIVEN,
     ) -> FineTuningJob:
         return self._get(
             f"/fine_tuning/jobs/{fine_tuning_job_id}",
-            options=make_user_request_input(extra_headers=extra_headers, timeout=timeout),
+            options=make_request_options(extra_headers=extra_headers, extra_body=extra_body, timeout=timeout),
             cast_type=FineTuningJob,
         )
 
@@ -64,13 +76,15 @@ def list(
         after: str | NotGiven = NOT_GIVEN,
         limit: int | NotGiven = NOT_GIVEN,
         extra_headers: Headers | None = None,
+        extra_body: Body | None = None,
         timeout: float | httpx.Timeout | None | NotGiven = NOT_GIVEN,
     ) -> ListOfFineTuningJob:
         return self._get(
             "/fine_tuning/jobs",
             cast_type=ListOfFineTuningJob,
-            options=make_user_request_input(
+            options=make_request_options(
                 extra_headers=extra_headers,
+                extra_body=extra_body,
                 timeout=timeout,
                 query={
                     "after": after,
@@ -79,6 +93,24 @@ def list(
             ),
         )
 
+    def cancel(
+        self,
+        fine_tuning_job_id: str,
+        *,
+        # Use the following arguments if you need to pass additional parameters to the API that aren't available via kwargs.  # noqa: E501
+        # The extra values given here take precedence over values defined on the client or passed to this method.
+        extra_headers: Headers | None = None,
+        extra_body: Body | None = None,
+        timeout: float | httpx.Timeout | None | NotGiven = NOT_GIVEN,
+    ) -> FineTuningJob:
+        if not fine_tuning_job_id:
+            raise ValueError(f"Expected a non-empty value for `fine_tuning_job_id` but received {fine_tuning_job_id!r}")
+        return self._post(
+            f"/fine_tuning/jobs/{fine_tuning_job_id}/cancel",
+            options=make_request_options(extra_headers=extra_headers, extra_body=extra_body, timeout=timeout),
+            cast_type=FineTuningJob,
+        )
+
     def list_events(
         self,
         fine_tuning_job_id: str,
@@ -86,13 +118,15 @@ def list_events(
         after: str | NotGiven = NOT_GIVEN,
         limit: int | NotGiven = NOT_GIVEN,
         extra_headers: Headers | None = None,
+        extra_body: Body | None = None,
         timeout: float | httpx.Timeout | None | NotGiven = NOT_GIVEN,
     ) -> FineTuningJobEvent:
         return self._get(
             f"/fine_tuning/jobs/{fine_tuning_job_id}/events",
             cast_type=FineTuningJobEvent,
-            options=make_user_request_input(
+            options=make_request_options(
                 extra_headers=extra_headers,
+                extra_body=extra_body,
                 timeout=timeout,
                 query={
                     "after": after,
@@ -100,3 +134,19 @@ def list_events(
                 },
             ),
         )
+
+    def delete(
+        self,
+        fine_tuning_job_id: str,
+        *,
+        extra_headers: Headers | None = None,
+        extra_body: Body | None = None,
+        timeout: float | httpx.Timeout | None | NotGiven = NOT_GIVEN,
+    ) -> FineTuningJob:
+        if not fine_tuning_job_id:
+            raise ValueError(f"Expected a non-empty value for `fine_tuning_job_id` but received {fine_tuning_job_id!r}")
+        return self._delete(
+            f"/fine_tuning/jobs/{fine_tuning_job_id}",
+            options=make_request_options(extra_headers=extra_headers, extra_body=extra_body, timeout=timeout),
+            cast_type=FineTuningJob,
+        )
diff --git a/api/core/model_runtime/model_providers/zhipuai/zhipuai_sdk/api_resource/fine_tuning/models/__init__.py b/api/core/model_runtime/model_providers/zhipuai/zhipuai_sdk/api_resource/fine_tuning/models/__init__.py
new file mode 100644
index 0000000000000..d832635bafbc6
--- /dev/null
+++ b/api/core/model_runtime/model_providers/zhipuai/zhipuai_sdk/api_resource/fine_tuning/models/__init__.py
@@ -0,0 +1,3 @@
+from .fine_tuned_models import FineTunedModels
+
+__all__ = ["FineTunedModels"]
diff --git a/api/core/model_runtime/model_providers/zhipuai/zhipuai_sdk/api_resource/fine_tuning/models/fine_tuned_models.py b/api/core/model_runtime/model_providers/zhipuai/zhipuai_sdk/api_resource/fine_tuning/models/fine_tuned_models.py
new file mode 100644
index 0000000000000..29c023e3b1cd5
--- /dev/null
+++ b/api/core/model_runtime/model_providers/zhipuai/zhipuai_sdk/api_resource/fine_tuning/models/fine_tuned_models.py
@@ -0,0 +1,41 @@
+from __future__ import annotations
+
+from typing import TYPE_CHECKING
+
+import httpx
+
+from ....core import (
+    NOT_GIVEN,
+    BaseAPI,
+    Body,
+    Headers,
+    NotGiven,
+    make_request_options,
+)
+from ....types.fine_tuning.models import FineTunedModelsStatus
+
+if TYPE_CHECKING:
+    from ...._client import ZhipuAI
+
+__all__ = ["FineTunedModels"]
+
+
+class FineTunedModels(BaseAPI):
+    def __init__(self, client: ZhipuAI) -> None:
+        super().__init__(client)
+
+    def delete(
+        self,
+        fine_tuned_model: str,
+        *,
+        extra_headers: Headers | None = None,
+        extra_body: Body | None = None,
+        timeout: float | httpx.Timeout | None | NotGiven = NOT_GIVEN,
+    ) -> FineTunedModelsStatus:
+        if not fine_tuned_model:
+            raise ValueError(f"Expected a non-empty value for `fine_tuned_model` but received {fine_tuned_model!r}")
+        return self._delete(
+            f"fine_tuning/fine_tuned_models/{fine_tuned_model}",
+            options=make_request_options(extra_headers=extra_headers, extra_body=extra_body, timeout=timeout),
+            cast_type=FineTunedModelsStatus,
+        )
diff --git a/api/core/model_runtime/model_providers/zhipuai/zhipuai_sdk/api_resource/images.py b/api/core/model_runtime/model_providers/zhipuai/zhipuai_sdk/api_resource/images.py
index 2692b093af8b4..8ad411913fa11 100644
--- a/api/core/model_runtime/model_providers/zhipuai/zhipuai_sdk/api_resource/images.py
+++ b/api/core/model_runtime/model_providers/zhipuai/zhipuai_sdk/api_resource/images.py
@@ -4,10 +4,9 @@
 
 import httpx
 
-from ..core._base_api import BaseAPI
-from ..core._base_type import NOT_GIVEN, Body, Headers, NotGiven
-from ..core._http_client import make_user_request_input
+from ..core import NOT_GIVEN, BaseAPI, Body, Headers, NotGiven, make_request_options
 from ..types.image import ImagesResponded
+from ..types.sensitive_word_check import SensitiveWordCheckRequest
 
 if TYPE_CHECKING:
     from .._client import ZhipuAI
@@ -27,8 +26,10 @@ def generations(
         response_format: Optional[str] | NotGiven = NOT_GIVEN,
         size: Optional[str] | NotGiven = NOT_GIVEN,
         style: Optional[str] | NotGiven = NOT_GIVEN,
+        sensitive_word_check: Optional[SensitiveWordCheckRequest] | NotGiven = NOT_GIVEN,
         user: str | NotGiven = NOT_GIVEN,
         request_id: Optional[str] | NotGiven = NOT_GIVEN,
+        user_id: Optional[str] | NotGiven = NOT_GIVEN,
         extra_headers: Headers | None = None,
         extra_body: Body | None = None,
         disable_strict_validation: Optional[bool] | None = None,
@@ -45,12 +46,14 @@ def generations(
                 "n": n,
                 "quality": quality,
                 "response_format": response_format,
+                "sensitive_word_check": sensitive_word_check,
                 "size": size,
                 "style": style,
                 "user": user,
+                "user_id": user_id,
                 "request_id": request_id,
             },
-            options=make_user_request_input(extra_headers=extra_headers, extra_body=extra_body, timeout=timeout),
+            options=make_request_options(extra_headers=extra_headers, extra_body=extra_body, timeout=timeout),
             cast_type=_cast_type,
-            enable_stream=False,
+            stream=False,
         )
diff --git a/api/core/model_runtime/model_providers/zhipuai/zhipuai_sdk/api_resource/knowledge/__init__.py b/api/core/model_runtime/model_providers/zhipuai/zhipuai_sdk/api_resource/knowledge/__init__.py
new file mode 100644
index 0000000000000..5a67d743c35b9
--- /dev/null
+++ b/api/core/model_runtime/model_providers/zhipuai/zhipuai_sdk/api_resource/knowledge/__init__.py
@@ -0,0 +1,3 @@
+from .knowledge import Knowledge
+
+__all__ = ["Knowledge"]
diff --git a/api/core/model_runtime/model_providers/zhipuai/zhipuai_sdk/api_resource/knowledge/document/__init__.py b/api/core/model_runtime/model_providers/zhipuai/zhipuai_sdk/api_resource/knowledge/document/__init__.py
new file mode 100644
index 0000000000000..fd289e2232b95
--- /dev/null
+++ b/api/core/model_runtime/model_providers/zhipuai/zhipuai_sdk/api_resource/knowledge/document/__init__.py
@@ -0,0 +1,3 @@
+from .document import Document
+
+__all__ = ["Document"]
diff --git a/api/core/model_runtime/model_providers/zhipuai/zhipuai_sdk/api_resource/knowledge/document/document.py b/api/core/model_runtime/model_providers/zhipuai/zhipuai_sdk/api_resource/knowledge/document/document.py
new file mode 100644
index 0000000000000..2c4066d893034
--- /dev/null
+++ b/api/core/model_runtime/model_providers/zhipuai/zhipuai_sdk/api_resource/knowledge/document/document.py
@@ -0,0 +1,217 @@
+from __future__ import annotations
+
+from collections.abc import Mapping
+from typing import TYPE_CHECKING, Literal, Optional, cast
+
+import httpx
+
+from ....core import (
+    NOT_GIVEN,
+    BaseAPI,
+    Body,
+    FileTypes,
+    Headers,
+    NotGiven,
+    deepcopy_minimal,
+    extract_files,
+    make_request_options,
+    maybe_transform,
+)
+from ....types.files import UploadDetail, file_create_params
+from ....types.knowledge.document import DocumentData, DocumentObject, document_edit_params, document_list_params
+from ....types.knowledge.document.document_list_resp import DocumentPage
+
+if TYPE_CHECKING:
+    from ...._client import ZhipuAI
+
+__all__ = ["Document"]
+
+
+class Document(BaseAPI):
+    def __init__(self, client: ZhipuAI) -> None:
+        super().__init__(client)
+
+    def create(
+        self,
+        *,
+        file: FileTypes = None,
+        custom_separator: Optional[list[str]] = None,
+        upload_detail: list[UploadDetail] = None,
+        purpose: Literal["retrieval"],
+        knowledge_id: str = None,
+        sentence_size: int = None,
+        extra_headers: Headers | None = None,
+        extra_body: Body | None = None,
+        timeout: float | httpx.Timeout | None | NotGiven = NOT_GIVEN,
+    ) -> DocumentObject:
+        if not file and not upload_detail:
+            raise ValueError("At least one of `file` and `upload_detail` must be provided.")
+        body = deepcopy_minimal(
+            {
+                "file": file,
+                "upload_detail": upload_detail,
+                "purpose": purpose,
+                "custom_separator": custom_separator,
+                "knowledge_id": knowledge_id,
+                "sentence_size": sentence_size,
+            }
+        )
+        files = extract_files(cast(Mapping[str, object], body), paths=[["file"]])
+        if files:
+            # It should be noted that the actual Content-Type header that will be
+            # sent to the server will contain a `boundary` parameter, e.g.
+            # multipart/form-data; boundary=---abc--
+            extra_headers = {"Content-Type": "multipart/form-data", **(extra_headers or {})}
+        return self._post(
+            "/files",
+            body=maybe_transform(body, file_create_params.FileCreateParams),
+            files=files,
+            options=make_request_options(extra_headers=extra_headers, extra_body=extra_body, timeout=timeout),
+            cast_type=DocumentObject,
+        )
+
+    def edit(
+        self,
+        document_id: str,
+        knowledge_type: str,
+        *,
+        custom_separator: Optional[list[str]] = None,
+        sentence_size: Optional[int] = None,
+        callback_url: Optional[str] = None,
+        callback_header: Optional[dict[str, str]] = None,
+        extra_headers: Headers | None = None,
+        extra_body: Body | None = None,
+        timeout: float | httpx.Timeout | None | NotGiven = NOT_GIVEN,
+    ) -> httpx.Response:
+        """
+
+        Args:
+          document_id: 知识id
+          knowledge_type: 知识类型:
+                        1:文章知识: 支持pdf,url,docx
+                        2.问答知识-文档:  支持pdf,url,docx
+                        3.问答知识-表格:  支持xlsx
+                        4.商品库-表格:  支持xlsx
+                        5.自定义:  支持pdf,url,docx
+          extra_headers: Send extra headers
+
+          extra_body: Add additional JSON properties to the request
+
+          timeout: Override the client-level default timeout for this request, in seconds
+          :param knowledge_type:
+          :param document_id:
+          :param timeout:
+          :param extra_body:
+          :param callback_header:
+          :param sentence_size:
+          :param extra_headers:
+          :param callback_url:
+          :param custom_separator:
+        """
+        if not document_id:
+            raise ValueError(f"Expected a non-empty value for `document_id` but received {document_id!r}")
+
+        body = deepcopy_minimal(
+            {
+                "id": document_id,
+                "knowledge_type": knowledge_type,
+                "custom_separator": custom_separator,
+                "sentence_size": sentence_size,
+                "callback_url": callback_url,
+                "callback_header": callback_header,
+            }
+        )
+
+        return self._put(
+            f"/document/{document_id}",
+            body=maybe_transform(body, document_edit_params.DocumentEditParams),
+            options=make_request_options(extra_headers=extra_headers, extra_body=extra_body, timeout=timeout),
+            cast_type=httpx.Response,
+        )
+
+    def list(
+        self,
+        knowledge_id: str,
+        *,
+        purpose: str | NotGiven = NOT_GIVEN,
+        page: str | NotGiven = NOT_GIVEN,
+        limit: str | NotGiven = NOT_GIVEN,
+        order: Literal["desc", "asc"] | NotGiven = NOT_GIVEN,
+        extra_headers: Headers | None = None,
+        extra_body: Body | None = None,
+        timeout: float | httpx.Timeout | None | NotGiven = NOT_GIVEN,
+    ) -> DocumentPage:
+        return self._get(
+            "/files",
+            options=make_request_options(
+                extra_headers=extra_headers,
+                extra_body=extra_body,
+                timeout=timeout,
+                query=maybe_transform(
+                    {
+                        "knowledge_id": knowledge_id,
+                        "purpose": purpose,
+                        "page": page,
+                        "limit": limit,
+                        "order": order,
+                    },
+                    document_list_params.DocumentListParams,
+                ),
+            ),
+            cast_type=DocumentPage,
+        )
+
+    def delete(
+        self,
+        document_id: str,
+        *,
+        extra_headers: Headers | None = None,
+        extra_body: Body | None = None,
+        timeout: float | httpx.Timeout | None | NotGiven = NOT_GIVEN,
+    ) -> httpx.Response:
+        """
+        Delete a file.
+
+        Args:
+
+          document_id: 知识id
+          extra_headers: Send extra headers
+
+          extra_body: Add additional JSON properties to the request
+
+          timeout: Override the client-level default timeout for this request, in seconds
+        """
+        if not document_id:
+            raise ValueError(f"Expected a non-empty value for `document_id` but received {document_id!r}")
+
+        return self._delete(
+            f"/document/{document_id}",
+            options=make_request_options(extra_headers=extra_headers, extra_body=extra_body, timeout=timeout),
+            cast_type=httpx.Response,
+        )
+
+    def retrieve(
+        self,
+        document_id: str,
+        *,
+        extra_headers: Headers | None = None,
+        extra_body: Body | None = None,
+        timeout: float | httpx.Timeout | None | NotGiven = NOT_GIVEN,
+    ) -> DocumentData:
+        """
+
+        Args:
+          extra_headers: Send extra headers
+
+          extra_body: Add additional JSON properties to the request
+
+          timeout: Override the client-level default timeout for this request, in seconds
+        """
+        if not document_id:
+            raise ValueError(f"Expected a non-empty value for `document_id` but received {document_id!r}")
+
+        return self._get(
+            f"/document/{document_id}",
+            options=make_request_options(extra_headers=extra_headers, extra_body=extra_body, timeout=timeout),
+            cast_type=DocumentData,
+        )
diff --git a/api/core/model_runtime/model_providers/zhipuai/zhipuai_sdk/api_resource/knowledge/knowledge.py b/api/core/model_runtime/model_providers/zhipuai/zhipuai_sdk/api_resource/knowledge/knowledge.py
new file mode 100644
index 0000000000000..fea4c73ac997c
--- /dev/null
+++ b/api/core/model_runtime/model_providers/zhipuai/zhipuai_sdk/api_resource/knowledge/knowledge.py
@@ -0,0 +1,173 @@
+from __future__ import annotations
+
+from typing import TYPE_CHECKING, Literal, Optional
+
+import httpx
+
+from ...core import (
+    NOT_GIVEN,
+    BaseAPI,
+    Body,
+    Headers,
+    NotGiven,
+    cached_property,
+    deepcopy_minimal,
+    make_request_options,
+    maybe_transform,
+)
+from ...types.knowledge import KnowledgeInfo, KnowledgeUsed, knowledge_create_params, knowledge_list_params
+from ...types.knowledge.knowledge_list_resp import KnowledgePage
+from .document import Document
+
+if TYPE_CHECKING:
+    from ..._client import ZhipuAI
+
+__all__ = ["Knowledge"]
+
+
+class Knowledge(BaseAPI):
+    def __init__(self, client: ZhipuAI) -> None:
+        super().__init__(client)
+
+    @cached_property
+    def document(self) -> Document:
+        return Document(self._client)
+
+    def create(
+        self,
+        embedding_id: int,
+        name: str,
+        *,
+        customer_identifier: Optional[str] = None,
+        description: Optional[str] = None,
+        background: Optional[Literal["blue", "red", "orange", "purple", "sky"]] = None,
+        icon: Optional[Literal["question", "book", "seal", "wrench", "tag", "horn", "house"]] = None,
+        bucket_id: Optional[str] = None,
+        extra_headers: Headers | None = None,
+        extra_body: Body | None = None,
+        timeout: float | httpx.Timeout | None | NotGiven = NOT_GIVEN,
+    ) -> KnowledgeInfo:
+        body = deepcopy_minimal(
+            {
+                "embedding_id": embedding_id,
+                "name": name,
+                "customer_identifier": customer_identifier,
+                "description": description,
+                "background": background,
+                "icon": icon,
+                "bucket_id": bucket_id,
+            }
+        )
+        return self._post(
+            "/knowledge",
+            body=maybe_transform(body, knowledge_create_params.KnowledgeBaseParams),
+            options=make_request_options(extra_headers=extra_headers, extra_body=extra_body, timeout=timeout),
+            cast_type=KnowledgeInfo,
+        )
+
+    def modify(
+        self,
+        knowledge_id: str,
+        embedding_id: int,
+        *,
+        name: str,
+        description: Optional[str] = None,
+        background: Optional[Literal["blue", "red", "orange", "purple", "sky"]] = None,
+        icon: Optional[Literal["question", "book", "seal", "wrench", "tag", "horn", "house"]] = None,
+        extra_headers: Headers | None = None,
+        extra_body: Body | None = None,
+        timeout: float | httpx.Timeout | None | NotGiven = NOT_GIVEN,
+    ) -> httpx.Response:
+        body = deepcopy_minimal(
+            {
+                "id": knowledge_id,
+                "embedding_id": embedding_id,
+                "name": name,
+                "description": description,
+                "background": background,
+                "icon": icon,
+            }
+        )
+        return self._put(
+            f"/knowledge/{knowledge_id}",
+            body=maybe_transform(body, knowledge_create_params.KnowledgeBaseParams),
+            options=make_request_options(extra_headers=extra_headers, extra_body=extra_body, timeout=timeout),
+            cast_type=httpx.Response,
+        )
+
+    def query(
+        self,
+        *,
+        page: int | NotGiven = 1,
+        size: int | NotGiven = 10,
+        extra_headers: Headers | None = None,
+        extra_body: Body | None = None,
+        timeout: float | httpx.Timeout | None | NotGiven = NOT_GIVEN,
+    ) -> KnowledgePage:
+        return self._get(
+            "/knowledge",
+            options=make_request_options(
+                extra_headers=extra_headers,
+                extra_body=extra_body,
+                timeout=timeout,
+                query=maybe_transform(
+                    {
+                        "page": page,
+                        "size": size,
+                    },
+                    knowledge_list_params.KnowledgeListParams,
+                ),
+            ),
+            cast_type=KnowledgePage,
+        )
+
+    def delete(
+        self,
+        knowledge_id: str,
+        *,
+        extra_headers: Headers | None = None,
+        extra_body: Body | None = None,
+        timeout: float | httpx.Timeout | None | NotGiven = NOT_GIVEN,
+    ) -> httpx.Response:
+        """
+        Delete a file.
+
+        Args:
+          knowledge_id: 知识库ID
+          extra_headers: Send extra headers
+
+          extra_body: Add additional JSON properties to the request
+
+          timeout: Override the client-level default timeout for this request, in seconds
+        """
+        if not knowledge_id:
+            raise ValueError("Expected a non-empty value for `knowledge_id`")
+
+        return self._delete(
+            f"/knowledge/{knowledge_id}",
+            options=make_request_options(extra_headers=extra_headers, extra_body=extra_body, timeout=timeout),
+            cast_type=httpx.Response,
+        )
+
+    def used(
+        self,
+        *,
+        extra_headers: Headers | None = None,
+        extra_body: Body | None = None,
+        timeout: float | httpx.Timeout | None | NotGiven = NOT_GIVEN,
+    ) -> KnowledgeUsed:
+        """
+        Returns the contents of the specified file.
+
+        Args:
+          extra_headers: Send extra headers
+
+          extra_body: Add additional JSON properties to the request
+
+          timeout: Override the client-level default timeout for this request, in seconds
+        """
+        return self._get(
+            "/knowledge/capacity",
+            options=make_request_options(extra_headers=extra_headers, extra_body=extra_body, timeout=timeout),
+            cast_type=KnowledgeUsed,
+        )
diff --git a/api/core/model_runtime/model_providers/zhipuai/zhipuai_sdk/api_resource/tools/__init__.py b/api/core/model_runtime/model_providers/zhipuai/zhipuai_sdk/api_resource/tools/__init__.py
new file mode 100644
index 0000000000000..43e4e37da1779
--- /dev/null
+++ b/api/core/model_runtime/model_providers/zhipuai/zhipuai_sdk/api_resource/tools/__init__.py
@@ -0,0 +1,3 @@
+from .tools import Tools
+
+__all__ = ["Tools"]
diff --git a/api/core/model_runtime/model_providers/zhipuai/zhipuai_sdk/api_resource/tools/tools.py b/api/core/model_runtime/model_providers/zhipuai/zhipuai_sdk/api_resource/tools/tools.py
new file mode 100644
index 0000000000000..3c3a630aff47d
--- /dev/null
+++ b/api/core/model_runtime/model_providers/zhipuai/zhipuai_sdk/api_resource/tools/tools.py
@@ -0,0 +1,65 @@
+from __future__ import annotations
+
+import logging
+from typing import TYPE_CHECKING, Literal, Optional, Union
+
+import httpx
+
+from ...core import (
+    NOT_GIVEN,
+    BaseAPI,
+    Body,
+    Headers,
+    NotGiven,
+    StreamResponse,
+    deepcopy_minimal,
+    make_request_options,
+    maybe_transform,
+)
+from ...types.tools import WebSearch, WebSearchChunk, tools_web_search_params
+
+logger = logging.getLogger(__name__)
+
+if TYPE_CHECKING:
+    from ..._client import ZhipuAI
+
+__all__ = ["Tools"]
+
+
+class Tools(BaseAPI):
+    def __init__(self, client: ZhipuAI) -> None:
+        super().__init__(client)
+
+    def web_search(
+        self,
+        *,
+        model: str,
+        request_id: Optional[str] | NotGiven = NOT_GIVEN,
+        stream: Optional[Literal[False]] | Literal[True] | NotGiven = NOT_GIVEN,
+        messages: Union[str, list[str], list[int], object, None],
+        scope: Optional[str] | NotGiven = NOT_GIVEN,
+        location: Optional[str] | NotGiven = NOT_GIVEN,
+        recent_days: Optional[int] | NotGiven = NOT_GIVEN,
+        extra_headers: Headers | None = None,
+        extra_body: Body | None = None,
+        timeout: float | httpx.Timeout | None | NotGiven = NOT_GIVEN,
+    ) -> WebSearch | StreamResponse[WebSearchChunk]:
+        body = deepcopy_minimal(
+            {
+                "model": model,
+                "request_id": request_id,
+                "messages": messages,
+                "stream": stream,
+                "scope": scope,
+                "location": location,
+                "recent_days": recent_days,
+            }
+        )
+        return self._post(
+            "/tools",
+            body=maybe_transform(body, tools_web_search_params.WebSearchParams),
+            options=make_request_options(extra_headers=extra_headers, extra_body=extra_body, timeout=timeout),
+            cast_type=WebSearch,
+            stream=stream or False,
+            stream_cls=StreamResponse[WebSearchChunk],
+        )
diff --git a/api/core/model_runtime/model_providers/zhipuai/zhipuai_sdk/api_resource/videos/__init__.py b/api/core/model_runtime/model_providers/zhipuai/zhipuai_sdk/api_resource/videos/__init__.py
new file mode 100644
index 0000000000000..6b0f99ed09efe
--- /dev/null
+++ b/api/core/model_runtime/model_providers/zhipuai/zhipuai_sdk/api_resource/videos/__init__.py
@@ -0,0 +1,7 @@
+from .videos import (
+    Videos,
+)
+
+__all__ = [
+    "Videos",
+]
diff --git a/api/core/model_runtime/model_providers/zhipuai/zhipuai_sdk/api_resource/videos/videos.py b/api/core/model_runtime/model_providers/zhipuai/zhipuai_sdk/api_resource/videos/videos.py
new file mode 100644
index 0000000000000..f1f1c08036a66
--- /dev/null
+++ b/api/core/model_runtime/model_providers/zhipuai/zhipuai_sdk/api_resource/videos/videos.py
@@ -0,0 +1,77 @@
+from __future__ import annotations
+
+from typing import TYPE_CHECKING, Optional
+
+import httpx
+
+from ...core import (
+    NOT_GIVEN,
+    BaseAPI,
+    Body,
+    Headers,
+    NotGiven,
+    deepcopy_minimal,
+    make_request_options,
+    maybe_transform,
+)
+from ...types.sensitive_word_check import SensitiveWordCheckRequest
+from ...types.video import VideoObject, video_create_params
+
+if TYPE_CHECKING:
+    from ..._client import ZhipuAI
+
+__all__ = ["Videos"]
+
+
+class Videos(BaseAPI):
+    def __init__(self, client: ZhipuAI) -> None:
+        super().__init__(client)
+
+    def generations(
+        self,
+        model: str,
+        *,
+        prompt: str = None,
+        image_url: str = None,
+        sensitive_word_check: Optional[SensitiveWordCheckRequest] | NotGiven = NOT_GIVEN,
+        request_id: str = None,
+        user_id: str = None,
+        extra_headers: Headers | None = None,
+        extra_body: Body | None = None,
+        timeout: float | httpx.Timeout | None | NotGiven = NOT_GIVEN,
+    ) -> VideoObject:
+        if not model and not model:
+            raise ValueError("At least one of `model` and `prompt` must be provided.")
+        body = deepcopy_minimal(
+            {
+                "model": model,
+                "prompt": prompt,
+                "image_url": image_url,
+                "sensitive_word_check": sensitive_word_check,
+                "request_id": request_id,
+                "user_id": user_id,
+            }
+        )
+        return self._post(
+            "/videos/generations",
+            body=maybe_transform(body, video_create_params.VideoCreateParams),
+            options=make_request_options(extra_headers=extra_headers, extra_body=extra_body, timeout=timeout),
+            cast_type=VideoObject,
+        )
+
+    def retrieve_videos_result(
+        self,
+        id: str,
+        *,
+        extra_headers: Headers | None = None,
+        extra_body: Body | None = None,
+        timeout: float | httpx.Timeout | None | NotGiven = NOT_GIVEN,
+    ) -> VideoObject:
+        if not id:
+            raise ValueError("At least one of `id` must be provided.")
+
+        return self._get(
+            f"/async-result/{id}",
+            options=make_request_options(extra_headers=extra_headers, extra_body=extra_body, timeout=timeout),
+            cast_type=VideoObject,
+        )
diff --git a/api/core/model_runtime/model_providers/zhipuai/zhipuai_sdk/core/__init__.py b/api/core/model_runtime/model_providers/zhipuai/zhipuai_sdk/core/__init__.py
index e69de29bb2d1d..3d6466d279861 100644
--- a/api/core/model_runtime/model_providers/zhipuai/zhipuai_sdk/core/__init__.py
+++ b/api/core/model_runtime/model_providers/zhipuai/zhipuai_sdk/core/__init__.py
@@ -0,0 +1,108 @@
+from ._base_api import BaseAPI
+from ._base_compat import (
+    PYDANTIC_V2,
+    ConfigDict,
+    GenericModel,
+    cached_property,
+    field_get_default,
+    get_args,
+    get_model_config,
+    get_model_fields,
+    get_origin,
+    is_literal_type,
+    is_union,
+    parse_obj,
+)
+from ._base_models import BaseModel, construct_type
+from ._base_type import (
+    NOT_GIVEN,
+    Body,
+    FileTypes,
+    Headers,
+    IncEx,
+    ModelT,
+    NotGiven,
+    Query,
+)
+from ._constants import (
+    ZHIPUAI_DEFAULT_LIMITS,
+    ZHIPUAI_DEFAULT_MAX_RETRIES,
+    ZHIPUAI_DEFAULT_TIMEOUT,
+)
+from ._errors import (
+    APIAuthenticationError,
+    APIConnectionError,
+    APIInternalError,
+    APIReachLimitError,
+    APIRequestFailedError,
+    APIResponseError,
+    APIResponseValidationError,
+    APIServerFlowExceedError,
+    APIStatusError,
+    APITimeoutError,
+    ZhipuAIError,
+)
+from ._files import is_file_content
+from ._http_client import HttpClient, make_request_options
+from ._sse_client import StreamResponse
+from ._utils import (
+    deepcopy_minimal,
+    drop_prefix_image_data,
+    extract_files,
+    is_given,
+    is_list,
+    is_mapping,
+    maybe_transform,
+    parse_date,
+    parse_datetime,
+)
+
+__all__ = [
+    "BaseModel",
+    "construct_type",
+    "BaseAPI",
+    "NOT_GIVEN",
+    "Headers",
+    "NotGiven",
+    "Body",
+    "IncEx",
+    "ModelT",
+    "Query",
+    "FileTypes",
+    "PYDANTIC_V2",
+    "ConfigDict",
+    "GenericModel",
+    "get_args",
+    "is_union",
+    "parse_obj",
+    "get_origin",
+    "is_literal_type",
+    "get_model_config",
+    "get_model_fields",
+    "field_get_default",
+    "is_file_content",
+    "ZhipuAIError",
+    "APIStatusError",
+    "APIRequestFailedError",
+    "APIAuthenticationError",
+    "APIReachLimitError",
+    "APIInternalError",
+    "APIServerFlowExceedError",
+    "APIResponseError",
+    "APIResponseValidationError",
+    "APITimeoutError",
+    "make_request_options",
+    "HttpClient",
+    "ZHIPUAI_DEFAULT_TIMEOUT",
+    "ZHIPUAI_DEFAULT_MAX_RETRIES",
+    "ZHIPUAI_DEFAULT_LIMITS",
+    "is_list",
+    "is_mapping",
+    "parse_date",
+    "parse_datetime",
+    "is_given",
+    "maybe_transform",
+    "deepcopy_minimal",
+    "extract_files",
+    "StreamResponse",
+]
diff --git a/api/core/model_runtime/model_providers/zhipuai/zhipuai_sdk/core/_base_api.py b/api/core/model_runtime/model_providers/zhipuai/zhipuai_sdk/core/_base_api.py
index 10b46ff8e381a..3592ea6bacd17 100644
--- a/api/core/model_runtime/model_providers/zhipuai/zhipuai_sdk/core/_base_api.py
+++ b/api/core/model_runtime/model_providers/zhipuai/zhipuai_sdk/core/_base_api.py
@@ -16,3 +16,4 @@ def __init__(self, client: ZhipuAI) -> None:
         self._post = client.post
         self._put = client.put
         self._patch = client.patch
+        self._get_api_list = client.get_api_list
diff --git a/api/core/model_runtime/model_providers/zhipuai/zhipuai_sdk/core/_base_compat.py b/api/core/model_runtime/model_providers/zhipuai/zhipuai_sdk/core/_base_compat.py
new file mode 100644
index 0000000000000..92a5d683be673
--- /dev/null
+++ b/api/core/model_runtime/model_providers/zhipuai/zhipuai_sdk/core/_base_compat.py
@@ -0,0 +1,209 @@
+from __future__ import annotations
+
+from collections.abc import Callable
+from datetime import date, datetime
+from typing import TYPE_CHECKING, Any, Generic, TypeVar, Union, cast, overload
+
+import pydantic
+from pydantic.fields import FieldInfo
+from typing_extensions import Self
+
+from ._base_type import StrBytesIntFloat
+
+_T = TypeVar("_T")
+_ModelT = TypeVar("_ModelT", bound=pydantic.BaseModel)
+
+# --------------- Pydantic v2 compatibility ---------------
+
+# Pyright incorrectly reports some of our functions as overriding a method when they don't
+# pyright: reportIncompatibleMethodOverride=false
+
+PYDANTIC_V2 = pydantic.VERSION.startswith("2.")
+
+# v1 re-exports
+if TYPE_CHECKING:
+
+    def parse_date(value: date | StrBytesIntFloat) -> date: ...
+
+    def parse_datetime(value: Union[datetime, StrBytesIntFloat]) -> datetime: ...
+
+    def get_args(t: type[Any]) -> tuple[Any, ...]: ...
+
+    def is_union(tp: type[Any] | None) -> bool: ...
+
+    def get_origin(t: type[Any]) -> type[Any] | None: ...
+
+    def is_literal_type(type_: type[Any]) -> bool: ...
+
+    def is_typeddict(type_: type[Any]) -> bool: ...
+
+else:
+    if PYDANTIC_V2:
+        from pydantic.v1.typing import (  # noqa: I001
+            get_args as get_args,  # noqa: PLC0414
+            is_union as is_union,  # noqa: PLC0414
+            get_origin as get_origin,  # noqa: PLC0414
+            is_typeddict as is_typeddict,  # noqa: PLC0414
+            is_literal_type as is_literal_type,  # noqa: PLC0414
+        )
+        from pydantic.v1.datetime_parse import parse_date as parse_date, parse_datetime as parse_datetime  # noqa: PLC0414
+    else:
+        from pydantic.typing import (  # noqa: I001
+            get_args as get_args,  # noqa: PLC0414
+            is_union as is_union,  # noqa: PLC0414
+            get_origin as get_origin,  # noqa: PLC0414
+            is_typeddict as is_typeddict,  # noqa: PLC0414
+            is_literal_type as is_literal_type,  # noqa: PLC0414
+        )
+        from pydantic.datetime_parse import parse_date as parse_date, parse_datetime as parse_datetime  # noqa: PLC0414
+
+
+# refactored config
+if TYPE_CHECKING:
+    from pydantic import ConfigDict
+else:
+    if PYDANTIC_V2:
+        from pydantic import ConfigDict
+    else:
+        # TODO: provide an error message here?
+        ConfigDict = None
+
+
+# renamed methods / properties
+def parse_obj(model: type[_ModelT], value: object) -> _ModelT:
+    if PYDANTIC_V2:
+        return model.model_validate(value)
+    else:
+        # pyright: ignore[reportDeprecated, reportUnnecessaryCast]
+        return cast(_ModelT, model.parse_obj(value))
+
+
+def field_is_required(field: FieldInfo) -> bool:
+    if PYDANTIC_V2:
+        return field.is_required()
+    return field.required  # type: ignore
+
+
+def field_get_default(field: FieldInfo) -> Any:
+    value = field.get_default()
+    if PYDANTIC_V2:
+        from pydantic_core import PydanticUndefined
+
+        if value == PydanticUndefined:
+            return None
+        return value
+    return value
+
+
+def field_outer_type(field: FieldInfo) -> Any:
+    if PYDANTIC_V2:
+        return field.annotation
+    return field.outer_type_  # type: ignore
+
+
+def get_model_config(model: type[pydantic.BaseModel]) -> Any:
+    if PYDANTIC_V2:
+        return model.model_config
+    return model.__config__  # type: ignore
+
+
+def get_model_fields(model: type[pydantic.BaseModel]) -> dict[str, FieldInfo]:
+    if PYDANTIC_V2:
+        return model.model_fields
+    return model.__fields__  # type: ignore
+
+
+def model_copy(model: _ModelT) -> _ModelT:
+    if PYDANTIC_V2:
+        return model.model_copy()
+    return model.copy()  # type: ignore
+
+
+def model_json(model: pydantic.BaseModel, *, indent: int | None = None) -> str:
+    if PYDANTIC_V2:
+        return model.model_dump_json(indent=indent)
+    return model.json(indent=indent)  # type: ignore
+
+
+def model_dump(
+    model: pydantic.BaseModel,
+    *,
+    exclude_unset: bool = False,
+    exclude_defaults: bool = False,
+) -> dict[str, Any]:
+    if PYDANTIC_V2:
+        return model.model_dump(
+            exclude_unset=exclude_unset,
+            exclude_defaults=exclude_defaults,
+        )
+    return cast(
+        "dict[str, Any]",
+        model.dict(  # pyright: ignore[reportDeprecated, reportUnnecessaryCast]
+            exclude_unset=exclude_unset,
+            exclude_defaults=exclude_defaults,
+        ),
+    )
+
+
+def model_parse(model: type[_ModelT], data: Any) -> _ModelT:
+    if PYDANTIC_V2:
+        return model.model_validate(data)
+    return model.parse_obj(data)  # pyright: ignore[reportDeprecated]
+
+
+# generic models
+if TYPE_CHECKING:
+
+    class GenericModel(pydantic.BaseModel): ...
+
+else:
+    if PYDANTIC_V2:
+        # there no longer needs to be a distinction in v2 but
+        # we still have to create our own subclass to avoid
+        # inconsistent MRO ordering errors
+        class GenericModel(pydantic.BaseModel): ...
+
+    else:
+        import pydantic.generics
+
+        class GenericModel(pydantic.generics.GenericModel, pydantic.BaseModel): ...
+
+
+# cached properties
+if TYPE_CHECKING:
+    cached_property = property
+
+    # we define a separate type (copied from typeshed)
+    # that represents that `cached_property` is `set`able
+    # at runtime, which differs from `@property`.
+    #
+    # this is a separate type as editors likely special case
+    # `@property` and we don't want to cause issues just to have
+    # more helpful internal types.
+
+    class typed_cached_property(Generic[_T]):  # noqa: N801
+        func: Callable[[Any], _T]
+        attrname: str | None
+
+        def __init__(self, func: Callable[[Any], _T]) -> None: ...
+
+        @overload
+        def __get__(self, instance: None, owner: type[Any] | None = None) -> Self: ...
+
+        @overload
+        def __get__(self, instance: object, owner: type[Any] | None = None) -> _T: ...
+
+        def __get__(self, instance: object, owner: type[Any] | None = None) -> _T | Self:
+            raise NotImplementedError()
+
+        def __set_name__(self, owner: type[Any], name: str) -> None: ...
+
+        # __set__ is not defined at runtime, but @cached_property is designed to be settable
+        def __set__(self, instance: object, value: _T) -> None: ...
+else:
+    try:
+        from functools import cached_property
+    except ImportError:
+        from cached_property import cached_property
+
+    typed_cached_property = cached_property
diff --git a/api/core/model_runtime/model_providers/zhipuai/zhipuai_sdk/core/_base_models.py b/api/core/model_runtime/model_providers/zhipuai/zhipuai_sdk/core/_base_models.py
new file mode 100644
index 0000000000000..5e9a7e0a987e2
--- /dev/null
+++ b/api/core/model_runtime/model_providers/zhipuai/zhipuai_sdk/core/_base_models.py
@@ -0,0 +1,671 @@
+from __future__ import annotations
+
+import inspect
+import os
+from collections.abc import Callable
+from datetime import date, datetime
+from typing import TYPE_CHECKING, Any, ClassVar, Generic, Literal, TypeGuard, TypeVar, cast
+
+import pydantic
+import pydantic.generics
+from pydantic.fields import FieldInfo
+from typing_extensions import (
+    ParamSpec,
+    Protocol,
+    override,
+    runtime_checkable,
+)
+
+from ._base_compat import (
+    PYDANTIC_V2,
+    ConfigDict,
+    field_get_default,
+    get_args,
+    get_model_config,
+    get_model_fields,
+    get_origin,
+    is_literal_type,
+    is_union,
+    parse_obj,
+)
+from ._base_compat import (
+    GenericModel as BaseGenericModel,
+)
+from ._base_type import (
+    IncEx,
+    ModelT,
+)
+from ._utils import (
+    PropertyInfo,
+    coerce_boolean,
+    extract_type_arg,
+    is_annotated_type,
+    is_list,
+    is_mapping,
+    parse_date,
+    parse_datetime,
+    strip_annotated_type,
+)
+
+if TYPE_CHECKING:
+    from pydantic_core.core_schema import LiteralSchema, ModelField, ModelFieldsSchema
+
+__all__ = ["BaseModel", "GenericModel"]
+_BaseModelT = TypeVar("_BaseModelT", bound="BaseModel")
+
+_T = TypeVar("_T")
+P = ParamSpec("P")
+
+
+@runtime_checkable
+class _ConfigProtocol(Protocol):
+    allow_population_by_field_name: bool
+
+
+class BaseModel(pydantic.BaseModel):
+    if PYDANTIC_V2:
+        model_config: ClassVar[ConfigDict] = ConfigDict(
+            extra="allow", defer_build=coerce_boolean(os.environ.get("DEFER_PYDANTIC_BUILD", "true"))
+        )
+    else:
+
+        @property
+        @override
+        def model_fields_set(self) -> set[str]:
+            # a forwards-compat shim for pydantic v2
+            return self.__fields_set__  # type: ignore
+
+        class Config(pydantic.BaseConfig):  # pyright: ignore[reportDeprecated]
+            extra: Any = pydantic.Extra.allow  # type: ignore
+
+    def to_dict(
+        self,
+        *,
+        mode: Literal["json", "python"] = "python",
+        use_api_names: bool = True,
+        exclude_unset: bool = True,
+        exclude_defaults: bool = False,
+        exclude_none: bool = False,
+        warnings: bool = True,
+    ) -> dict[str, object]:
+        """Recursively generate a dictionary representation of the model, optionally specifying which fields to include or exclude.
+
+        By default, fields that were not set by the API will not be included,
+        and keys will match the API response, *not* the property names from the model.
+
+        For example, if the API responds with `"fooBar": true` but we've defined a `foo_bar: bool` property,
+        the output will use the `"fooBar"` key (unless `use_api_names=False` is passed).
+
+        Args:
+            mode:
+                If mode is 'json', the dictionary will only contain JSON serializable types. e.g. `datetime` will be turned into a string, `"2024-3-22T18:11:19.117000Z"`.
+                If mode is 'python', the dictionary may contain any Python objects. e.g. `datetime(2024, 3, 22)`
+
+            use_api_names: Whether to use the key that the API responded with or the property name. Defaults to `True`.
+            exclude_unset: Whether to exclude fields that have not been explicitly set.
+            exclude_defaults: Whether to exclude fields that are set to their default value from the output.
+            exclude_none: Whether to exclude fields that have a value of `None` from the output.
+            warnings: Whether to log warnings when invalid fields are encountered. This is only supported in Pydantic v2.
+        """  # noqa: E501
+        return self.model_dump(
+            mode=mode,
+            by_alias=use_api_names,
+            exclude_unset=exclude_unset,
+            exclude_defaults=exclude_defaults,
+            exclude_none=exclude_none,
+            warnings=warnings,
+        )
+
+    def to_json(
+        self,
+        *,
+        indent: int | None = 2,
+        use_api_names: bool = True,
+        exclude_unset: bool = True,
+        exclude_defaults: bool = False,
+        exclude_none: bool = False,
+        warnings: bool = True,
+    ) -> str:
+        """Generates a JSON string representing this model as it would be received from or sent to the API (but with indentation).
+
+        By default, fields that were not set by the API will not be included,
+        and keys will match the API response, *not* the property names from the model.
+
+        For example, if the API responds with `"fooBar": true` but we've defined a `foo_bar: bool` property,
+        the output will use the `"fooBar"` key (unless `use_api_names=False` is passed).
+
+        Args:
+            indent: Indentation to use in the JSON output. If `None` is passed, the output will be compact. Defaults to `2`
+            use_api_names: Whether to use the key that the API responded with or the property name. Defaults to `True`.
+            exclude_unset: Whether to exclude fields that have not been explicitly set.
+            exclude_defaults: Whether to exclude fields that have the default value.
+            exclude_none: Whether to exclude fields that have a value of `None`.
+            warnings: Whether to show any warnings that occurred during serialization. This is only supported in Pydantic v2.
+        """  # noqa: E501
+        return self.model_dump_json(
+            indent=indent,
+            by_alias=use_api_names,
+            exclude_unset=exclude_unset,
+            exclude_defaults=exclude_defaults,
+            exclude_none=exclude_none,
+            warnings=warnings,
+        )
+
+    @override
+    def __str__(self) -> str:
+        # mypy complains about an invalid self arg
+        return f'{self.__repr_name__()}({self.__repr_str__(", ")})'  # type: ignore[misc]
+
+    # Override the 'construct' method in a way that supports recursive parsing without validation.
+    # Based on https://github.com/samuelcolvin/pydantic/issues/1168#issuecomment-817742836.
+    @classmethod
+    @override
+    def construct(
+        cls: type[ModelT],
+        _fields_set: set[str] | None = None,
+        **values: object,
+    ) -> ModelT:
+        m = cls.__new__(cls)
+        fields_values: dict[str, object] = {}
+
+        config = get_model_config(cls)
+        populate_by_name = (
+            config.allow_population_by_field_name
+            if isinstance(config, _ConfigProtocol)
+            else config.get("populate_by_name")
+        )
+
+        if _fields_set is None:
+            _fields_set = set()
+
+        model_fields = get_model_fields(cls)
+        for name, field in model_fields.items():
+            key = field.alias
+            if key is None or (key not in values and populate_by_name):
+                key = name
+
+            if key in values:
+                fields_values[name] = _construct_field(value=values[key], field=field, key=key)
+                _fields_set.add(name)
+            else:
+                fields_values[name] = field_get_default(field)
+
+        _extra = {}
+        for key, value in values.items():
+            if key not in model_fields:
+                if PYDANTIC_V2:
+                    _extra[key] = value
+                else:
+                    _fields_set.add(key)
+                    fields_values[key] = value
+
+        object.__setattr__(m, "__dict__", fields_values)  # noqa: PLC2801
+
+        if PYDANTIC_V2:
+            # these properties are copied from Pydantic's `model_construct()` method
+            object.__setattr__(m, "__pydantic_private__", None)  # noqa: PLC2801
+            object.__setattr__(m, "__pydantic_extra__", _extra)  # noqa: PLC2801
+            object.__setattr__(m, "__pydantic_fields_set__", _fields_set)  # noqa: PLC2801
+        else:
+            # init_private_attributes() does not exist in v2
+            m._init_private_attributes()  # type: ignore
+
+            # copied from Pydantic v1's `construct()` method
+            object.__setattr__(m, "__fields_set__", _fields_set)  # noqa: PLC2801
+
+        return m
+
+    if not TYPE_CHECKING:
+        # type checkers incorrectly complain about this assignment
+        # because the type signatures are technically different
+        # although not in practice
+        model_construct = construct
+
+    if not PYDANTIC_V2:
+        # we define aliases for some of the new pydantic v2 methods so
+        # that we can just document these methods without having to specify
+        # a specific pydantic version as some users may not know which
+        # pydantic version they are currently using
+
+        @override
+        def model_dump(
+            self,
+            *,
+            mode: Literal["json", "python"] | str = "python",
+            include: IncEx = None,
+            exclude: IncEx = None,
+            by_alias: bool = False,
+            exclude_unset: bool = False,
+            exclude_defaults: bool = False,
+            exclude_none: bool = False,
+            round_trip: bool = False,
+            warnings: bool | Literal["none", "warn", "error"] = True,
+            context: dict[str, Any] | None = None,
+            serialize_as_any: bool = False,
+        ) -> dict[str, Any]:
+            """Usage docs: https://docs.pydantic.dev/2.4/concepts/serialization/#modelmodel_dump
+
+            Generate a dictionary representation of the model, optionally specifying which fields to include or exclude.
+
+            Args:
+                mode: The mode in which `to_python` should run.
+                    If mode is 'json', the dictionary will only contain JSON serializable types.
+                    If mode is 'python', the dictionary may contain any Python objects.
+                include: A list of fields to include in the output.
+                exclude: A list of fields to exclude from the output.
+                by_alias: Whether to use the field's alias in the dictionary key if defined.
+                exclude_unset: Whether to exclude fields that are unset or None from the output.
+                exclude_defaults: Whether to exclude fields that are set to their default value from the output.
+                exclude_none: Whether to exclude fields that have a value of `None` from the output.
+                round_trip: Whether to enable serialization and deserialization round-trip support.
+                warnings: Whether to log warnings when invalid fields are encountered.
+
+            Returns:
+                A dictionary representation of the model.
+            """
+            if mode != "python":
+                raise ValueError("mode is only supported in Pydantic v2")
+            if round_trip != False:
+                raise ValueError("round_trip is only supported in Pydantic v2")
+            if warnings != True:
+                raise ValueError("warnings is only supported in Pydantic v2")
+            if context is not None:
+                raise ValueError("context is only supported in Pydantic v2")
+            if serialize_as_any != False:
+                raise ValueError("serialize_as_any is only supported in Pydantic v2")
+            return super().dict(  # pyright: ignore[reportDeprecated]
+                include=include,
+                exclude=exclude,
+                by_alias=by_alias,
+                exclude_unset=exclude_unset,
+                exclude_defaults=exclude_defaults,
+                exclude_none=exclude_none,
+            )
+
+        @override
+        def model_dump_json(
+            self,
+            *,
+            indent: int | None = None,
+            include: IncEx = None,
+            exclude: IncEx = None,
+            by_alias: bool = False,
+            exclude_unset: bool = False,
+            exclude_defaults: bool = False,
+            exclude_none: bool = False,
+            round_trip: bool = False,
+            warnings: bool | Literal["none", "warn", "error"] = True,
+            context: dict[str, Any] | None = None,
+            serialize_as_any: bool = False,
+        ) -> str:
+            """Usage docs: https://docs.pydantic.dev/2.4/concepts/serialization/#modelmodel_dump_json
+
+            Generates a JSON representation of the model using Pydantic's `to_json` method.
+
+            Args:
+                indent: Indentation to use in the JSON output. If None is passed, the output will be compact.
+                include: Field(s) to include in the JSON output. Can take either a string or set of strings.
+                exclude: Field(s) to exclude from the JSON output. Can take either a string or set of strings.
+                by_alias: Whether to serialize using field aliases.
+                exclude_unset: Whether to exclude fields that have not been explicitly set.
+                exclude_defaults: Whether to exclude fields that have the default value.
+                exclude_none: Whether to exclude fields that have a value of `None`.
+                round_trip: Whether to use serialization/deserialization between JSON and class instance.
+                warnings: Whether to show any warnings that occurred during serialization.
+
+            Returns:
+                A JSON string representation of the model.
+            """
+            if round_trip != False:
+                raise ValueError("round_trip is only supported in Pydantic v2")
+            if warnings != True:
+                raise ValueError("warnings is only supported in Pydantic v2")
+            if context is not None:
+                raise ValueError("context is only supported in Pydantic v2")
+            if serialize_as_any != False:
+                raise ValueError("serialize_as_any is only supported in Pydantic v2")
+            return super().json(  # type: ignore[reportDeprecated]
+                indent=indent,
+                include=include,
+                exclude=exclude,
+                by_alias=by_alias,
+                exclude_unset=exclude_unset,
+                exclude_defaults=exclude_defaults,
+                exclude_none=exclude_none,
+            )
+
+
+def _construct_field(value: object, field: FieldInfo, key: str) -> object:
+    if value is None:
+        return field_get_default(field)
+
+    if PYDANTIC_V2:
+        type_ = field.annotation
+    else:
+        type_ = cast(type, field.outer_type_)  # type: ignore
+
+    if type_ is None:
+        raise RuntimeError(f"Unexpected field type is None for {key}")
+
+    return construct_type(value=value, type_=type_)
+
+
+def is_basemodel(type_: type) -> bool:
+    """Returns whether or not the given type is either a `BaseModel` or a union of `BaseModel`"""
+    if is_union(type_):
+        return any(is_basemodel(variant) for variant in get_args(type_))
+
+    return is_basemodel_type(type_)
+
+
+def is_basemodel_type(type_: type) -> TypeGuard[type[BaseModel] | type[GenericModel]]:
+    origin = get_origin(type_) or type_
+    return issubclass(origin, BaseModel) or issubclass(origin, GenericModel)
+
+
+def build(
+    base_model_cls: Callable[P, _BaseModelT],
+    *args: P.args,
+    **kwargs: P.kwargs,
+) -> _BaseModelT:
+    """Construct a BaseModel class without validation.
+
+    This is useful for cases where you need to instantiate a `BaseModel`
+    from an API response as this provides type-safe params which isn't supported
+    by helpers like `construct_type()`.
+
+    ```py
+    build(MyModel, my_field_a="foo", my_field_b=123)
+    ```
+    """
+    if args:
+        raise TypeError(
+            "Received positional arguments which are not supported; Keyword arguments must be used instead",
+        )
+
+    return cast(_BaseModelT, construct_type(type_=base_model_cls, value=kwargs))
+
+
+def construct_type_unchecked(*, value: object, type_: type[_T]) -> _T:
+    """Loose coercion to the expected type with construction of nested values.
+
+    Note: the returned value from this function is not guaranteed to match the
+    given type.
+    """
+    return cast(_T, construct_type(value=value, type_=type_))
+
+
+def construct_type(*, value: object, type_: type) -> object:
+    """Loose coercion to the expected type with construction of nested values.
+
+    If the given value does not match the expected type then it is returned as-is.
+    """
+    # we allow `object` as the input type because otherwise, passing things like
+    # `Literal['value']` will be reported as a type error by type checkers
+    type_ = cast("type[object]", type_)
+
+    # unwrap `Annotated[T, ...]` -> `T`
+    if is_annotated_type(type_):
+        meta: tuple[Any, ...] = get_args(type_)[1:]
+        type_ = extract_type_arg(type_, 0)
+    else:
+        meta = ()
+    # we need to use the origin class for any types that are subscripted generics
+    # e.g. Dict[str, object]
+    origin = get_origin(type_) or type_
+    args = get_args(type_)
+
+    if is_union(origin):
+        try:
+            return validate_type(type_=cast("type[object]", type_), value=value)
+        except Exception:
+            pass
+
+        # if the type is a discriminated union then we want to construct the right variant
+        # in the union, even if the data doesn't match exactly, otherwise we'd break code
+        # that relies on the constructed class types, e.g.
+        #
+        # class FooType:
+        #   kind: Literal['foo']
+        #   value: str
+        #
+        # class BarType:
+        #   kind: Literal['bar']
+        #   value: int
+        #
+        # without this block, if the data we get is something like `{'kind': 'bar', 'value': 'foo'}` then
+        # we'd end up constructing `FooType` when it should be `BarType`.
+        discriminator = _build_discriminated_union_meta(union=type_, meta_annotations=meta)
+        if discriminator and is_mapping(value):
+            variant_value = value.get(discriminator.field_alias_from or discriminator.field_name)
+            if variant_value and isinstance(variant_value, str):
+                variant_type = discriminator.mapping.get(variant_value)
+                if variant_type:
+                    return construct_type(type_=variant_type, value=value)
+
+        # if the data is not valid, use the first variant that doesn't fail while deserializing
+        for variant in args:
+            try:
+                return construct_type(value=value, type_=variant)
+            except Exception:
+                continue
+
+        raise RuntimeError(f"Could not convert data into a valid instance of {type_}")
+    if origin == dict:
+        if not is_mapping(value):
+            return value
+
+        _, items_type = get_args(type_)  # Dict[_, items_type]
+        return {key: construct_type(value=item, type_=items_type) for key, item in value.items()}
+
+    if not is_literal_type(type_) and (issubclass(origin, BaseModel) or issubclass(origin, GenericModel)):
+        if is_list(value):
+            return [cast(Any, type_).construct(**entry) if is_mapping(entry) else entry for entry in value]
+
+        if is_mapping(value):
+            if issubclass(type_, BaseModel):
+                return type_.construct(**value)  # type: ignore[arg-type]
+
+            return cast(Any, type_).construct(**value)
+
+    if origin == list:
+        if not is_list(value):
+            return value
+
+        inner_type = args[0]  # List[inner_type]
+        return [construct_type(value=entry, type_=inner_type) for entry in value]
+
+    if origin == float:
+        if isinstance(value, int):
+            coerced = float(value)
+            if coerced != value:
+                return value
+            return coerced
+
+        return value
+
+    if type_ == datetime:
+        try:
+            return parse_datetime(value)  # type: ignore
+        except Exception:
+            return value
+
+    if type_ == date:
+        try:
+            return parse_date(value)  # type: ignore
+        except Exception:
+            return value
+
+    return value
+
+
+@runtime_checkable
+class CachedDiscriminatorType(Protocol):
+    __discriminator__: DiscriminatorDetails
+
+
+class DiscriminatorDetails:
+    field_name: str
+    """The name of the discriminator field in the variant class, e.g.
+
+    ```py
+    class Foo(BaseModel):
+        type: Literal['foo']
+    ```
+
+    Will result in field_name='type'
+    """
+
+    field_alias_from: str | None
+    """The name of the discriminator field in the API response, e.g.
+
+    ```py
+    class Foo(BaseModel):
+        type: Literal['foo'] = Field(alias='type_from_api')
+    ```
+
+    Will result in field_alias_from='type_from_api'
+    """
+
+    mapping: dict[str, type]
+    """Mapping of discriminator value to variant type, e.g.
+
+    {'foo': FooVariant, 'bar': BarVariant}
+    """
+
+    def __init__(
+        self,
+        *,
+        mapping: dict[str, type],
+        discriminator_field: str,
+        discriminator_alias: str | None,
+    ) -> None:
+        self.mapping = mapping
+        self.field_name = discriminator_field
+        self.field_alias_from = discriminator_alias
+
+
+def _build_discriminated_union_meta(*, union: type, meta_annotations: tuple[Any, ...]) -> DiscriminatorDetails | None:
+    if isinstance(union, CachedDiscriminatorType):
+        return union.__discriminator__
+
+    discriminator_field_name: str | None = None
+
+    for annotation in meta_annotations:
+        if isinstance(annotation, PropertyInfo) and annotation.discriminator is not None:
+            discriminator_field_name = annotation.discriminator
+            break
+
+    if not discriminator_field_name:
+        return None
+
+    mapping: dict[str, type] = {}
+    discriminator_alias: str | None = None
+
+    for variant in get_args(union):
+        variant = strip_annotated_type(variant)
+        if is_basemodel_type(variant):
+            if PYDANTIC_V2:
+                field = _extract_field_schema_pv2(variant, discriminator_field_name)
+                if not field:
+                    continue
+
+                # Note: if one variant defines an alias then they all should
+                discriminator_alias = field.get("serialization_alias")
+
+                field_schema = field["schema"]
+
+                if field_schema["type"] == "literal":
+                    for entry in cast("LiteralSchema", field_schema)["expected"]:
+                        if isinstance(entry, str):
+                            mapping[entry] = variant
+            else:
+                field_info = cast("dict[str, FieldInfo]", variant.__fields__).get(discriminator_field_name)  # pyright: ignore[reportDeprecated, reportUnnecessaryCast]
+                if not field_info:
+                    continue
+
+                # Note: if one variant defines an alias then they all should
+                discriminator_alias = field_info.alias
+
+                if field_info.annotation and is_literal_type(field_info.annotation):
+                    for entry in get_args(field_info.annotation):
+                        if isinstance(entry, str):
+                            mapping[entry] = variant
+
+    if not mapping:
+        return None
+
+    details = DiscriminatorDetails(
+        mapping=mapping,
+        discriminator_field=discriminator_field_name,
+        discriminator_alias=discriminator_alias,
+    )
+    cast(CachedDiscriminatorType, union).__discriminator__ = details
+    return details
+
+
+def _extract_field_schema_pv2(model: type[BaseModel], field_name: str) -> ModelField | None:
+    schema = model.__pydantic_core_schema__
+    if schema["type"] != "model":
+        return None
+
+    fields_schema = schema["schema"]
+    if fields_schema["type"] != "model-fields":
+        return None
+
+    fields_schema = cast("ModelFieldsSchema", fields_schema)
+
+    field = fields_schema["fields"].get(field_name)
+    if not field:
+        return None
+
+    return cast("ModelField", field)  # pyright: ignore[reportUnnecessaryCast]
+
+
+def validate_type(*, type_: type[_T], value: object) -> _T:
+    """Strict validation that the given value matches the expected type"""
+    if inspect.isclass(type_) and issubclass(type_, pydantic.BaseModel):
+        return cast(_T, parse_obj(type_, value))
+
+    return cast(_T, _validate_non_model_type(type_=type_, value=value))
+
+
+# our use of subclasssing here causes weirdness for type checkers,
+# so we just pretend that we don't subclass
+if TYPE_CHECKING:
+    GenericModel = BaseModel
+else:
+
+    class GenericModel(BaseGenericModel, BaseModel):
+        pass
+
+
+if PYDANTIC_V2:
+    from pydantic import TypeAdapter
+
+    def _validate_non_model_type(*, type_: type[_T], value: object) -> _T:
+        return TypeAdapter(type_).validate_python(value)
+
+elif not TYPE_CHECKING:
+
+    class TypeAdapter(Generic[_T]):
+        """Used as a placeholder to easily convert runtime types to a Pydantic format
+        to provide validation.
+
+        For example:
+        ```py
+        validated = RootModel[int](__root__="5").__root__
+        # validated: 5
+        ```
+        """
+
+        def __init__(self, type_: type[_T]):
+            self.type_ = type_
+
+        def validate_python(self, value: Any) -> _T:
+            if not isinstance(value, self.type_):
+                raise ValueError(f"Invalid type: {value} is not of type {self.type_}")
+            return value
+
+    def _validate_non_model_type(*, type_: type[_T], value: object) -> _T:
+        return TypeAdapter(type_).validate_python(value)
diff --git a/api/core/model_runtime/model_providers/zhipuai/zhipuai_sdk/core/_base_type.py b/api/core/model_runtime/model_providers/zhipuai/zhipuai_sdk/core/_base_type.py
index 7a91f9b79627c..ea1d3f09dc42e 100644
--- a/api/core/model_runtime/model_providers/zhipuai/zhipuai_sdk/core/_base_type.py
+++ b/api/core/model_runtime/model_providers/zhipuai/zhipuai_sdk/core/_base_type.py
@@ -1,11 +1,21 @@
 from __future__ import annotations
 
-from collections.abc import Mapping, Sequence
+from collections.abc import Callable, Mapping, Sequence
 from os import PathLike
-from typing import IO, TYPE_CHECKING, Any, Literal, TypeVar, Union
+from typing import (
+    IO,
+    TYPE_CHECKING,
+    Any,
+    Literal,
+    Optional,
+    TypeAlias,
+    TypeVar,
+    Union,
+)
 
 import pydantic
-from typing_extensions import override
+from httpx import Response
+from typing_extensions import Protocol, TypedDict, override, runtime_checkable
 
 Query = Mapping[str, object]
 Body = object
@@ -22,7 +32,7 @@
 
 
 # Sentinel class used until PEP 0661 is accepted
-class NotGiven(pydantic.BaseModel):
+class NotGiven:
     """
     A sentinel singleton class used to distinguish omitted keyword arguments
     from those passed in with the value None (which may have different behavior).
@@ -50,7 +60,7 @@ def __repr__(self) -> str:
 NOT_GIVEN = NotGiven()
 
 
-class Omit(pydantic.BaseModel):
+class Omit:
     """In certain situations you need to be able to represent a case where a default value has
     to be explicitly removed and `None` is not an appropriate substitute, for example:
 
@@ -71,37 +81,90 @@ def __bool__(self) -> Literal[False]:
         return False
 
 
+@runtime_checkable
+class ModelBuilderProtocol(Protocol):
+    @classmethod
+    def build(
+        cls: type[_T],
+        *,
+        response: Response,
+        data: object,
+    ) -> _T: ...
+
+
 Headers = Mapping[str, Union[str, Omit]]
 
+
+class HeadersLikeProtocol(Protocol):
+    def get(self, __key: str) -> str | None: ...
+
+
+HeadersLike = Union[Headers, HeadersLikeProtocol]
+
 ResponseT = TypeVar(
     "ResponseT",
-    bound="Union[str, None, BaseModel, list[Any], Dict[str, Any], Response, UnknownResponse, ModelBuilderProtocol,"
-    " BinaryResponseContent]",
+    bound="Union[str, None, BaseModel, list[Any], dict[str, Any], Response, UnknownResponse, ModelBuilderProtocol, BinaryResponseContent]",  # noqa: E501
 )
 
+StrBytesIntFloat = Union[str, bytes, int, float]
+
+# Note: copied from Pydantic
+# https://github.com/pydantic/pydantic/blob/32ea570bf96e84234d2992e1ddf40ab8a565925a/pydantic/main.py#L49
+IncEx: TypeAlias = "set[int] | set[str] | dict[int, Any] | dict[str, Any] | None"
+
+PostParser = Callable[[Any], Any]
+
+
+@runtime_checkable
+class InheritsGeneric(Protocol):
+    """Represents a type that has inherited from `Generic`
+
+    The `__orig_bases__` property can be used to determine the resolved
+    type variable for a given base class.
+    """
+
+    __orig_bases__: tuple[_GenericAlias]
+
+
+class _GenericAlias(Protocol):
+    __origin__: type[object]
+
+
+class HttpxSendArgs(TypedDict, total=False):
+    auth: httpx.Auth
+
+
 # for user input files
 if TYPE_CHECKING:
+    Base64FileInput = Union[IO[bytes], PathLike[str]]
     FileContent = Union[IO[bytes], bytes, PathLike[str]]
 else:
+    Base64FileInput = Union[IO[bytes], PathLike]
     FileContent = Union[IO[bytes], bytes, PathLike]
 
 FileTypes = Union[
-    FileContent,  # file content
-    tuple[str, FileContent],  # (filename, file)
-    tuple[str, FileContent, str],  # (filename, file , content_type)
-    tuple[str, FileContent, str, Mapping[str, str]],  # (filename, file , content_type, headers)
+    # file (or bytes)
+    FileContent,
+    # (filename, file (or bytes))
+    tuple[Optional[str], FileContent],
+    # (filename, file (or bytes), content_type)
+    tuple[Optional[str], FileContent, Optional[str]],
+    # (filename, file (or bytes), content_type, headers)
+    tuple[Optional[str], FileContent, Optional[str], Mapping[str, str]],
 ]
-
 RequestFiles = Union[Mapping[str, FileTypes], Sequence[tuple[str, FileTypes]]]
 
-# for httpx client supported files
-
+# duplicate of the above but without our custom file support
 HttpxFileContent = Union[bytes, IO[bytes]]
 HttpxFileTypes = Union[
-    FileContent,  # file content
-    tuple[str, HttpxFileContent],  # (filename, file)
-    tuple[str, HttpxFileContent, str],  # (filename, file , content_type)
-    tuple[str, HttpxFileContent, str, Mapping[str, str]],  # (filename, file , content_type, headers)
+    # file (or bytes)
+    HttpxFileContent,
+    # (filename, file (or bytes))
+    tuple[Optional[str], HttpxFileContent],
+    # (filename, file (or bytes), content_type)
+    tuple[Optional[str], HttpxFileContent, Optional[str]],
+    # (filename, file (or bytes), content_type, headers)
+    tuple[Optional[str], HttpxFileContent, Optional[str], Mapping[str, str]],
 ]
 
 HttpxRequestFiles = Union[Mapping[str, HttpxFileTypes], Sequence[tuple[str, HttpxFileTypes]]]
diff --git a/api/core/model_runtime/model_providers/zhipuai/zhipuai_sdk/core/_constants.py b/api/core/model_runtime/model_providers/zhipuai/zhipuai_sdk/core/_constants.py
new file mode 100644
index 0000000000000..8e43bdebecb61
--- /dev/null
+++ b/api/core/model_runtime/model_providers/zhipuai/zhipuai_sdk/core/_constants.py
@@ -0,0 +1,12 @@
+import httpx
+
+RAW_RESPONSE_HEADER = "X-Stainless-Raw-Response"
+# 通过 `Timeout` 控制接口`connect` 和 `read` 超时时间，默认为`timeout=300.0, connect=8.0`
+ZHIPUAI_DEFAULT_TIMEOUT = httpx.Timeout(timeout=300.0, connect=8.0)
+# 通过 `retry` 参数控制重试次数，默认为3次
+ZHIPUAI_DEFAULT_MAX_RETRIES = 3
+# 通过 `Limits` 控制最大连接数和保持连接数，默认为`max_connections=50, max_keepalive_connections=10`
+ZHIPUAI_DEFAULT_LIMITS = httpx.Limits(max_connections=50, max_keepalive_connections=10)
+
+INITIAL_RETRY_DELAY = 0.5
+MAX_RETRY_DELAY = 8.0
diff --git a/api/core/model_runtime/model_providers/zhipuai/zhipuai_sdk/core/_errors.py b/api/core/model_runtime/model_providers/zhipuai/zhipuai_sdk/core/_errors.py
index 1027c1bc5b1e5..e2c9d24c6c0d2 100644
--- a/api/core/model_runtime/model_providers/zhipuai/zhipuai_sdk/core/_errors.py
+++ b/api/core/model_runtime/model_providers/zhipuai/zhipuai_sdk/core/_errors.py
@@ -13,6 +13,7 @@
     "APIResponseError",
     "APIResponseValidationError",
     "APITimeoutError",
+    "APIConnectionError",
 ]
 
 
@@ -24,7 +25,7 @@ def __init__(
         super().__init__(message)
 
 
-class APIStatusError(Exception):
+class APIStatusError(ZhipuAIError):
     response: httpx.Response
     status_code: int
 
@@ -49,7 +50,7 @@ class APIInternalError(APIStatusError): ...
 class APIServerFlowExceedError(APIStatusError): ...
 
 
-class APIResponseError(Exception):
+class APIResponseError(ZhipuAIError):
     message: str
     request: httpx.Request
     json_data: object
@@ -75,9 +76,11 @@ def __init__(self, response: httpx.Response, json_data: object | None, *, messag
         self.status_code = response.status_code
 
 
-class APITimeoutError(Exception):
-    request: httpx.Request
+class APIConnectionError(APIResponseError):
+    def __init__(self, *, message: str = "Connection error.", request: httpx.Request) -> None:
+        super().__init__(message, request, json_data=None)
 
-    def __init__(self, request: httpx.Request):
-        self.request = request
-        super().__init__("Request Timeout")
+
+class APITimeoutError(APIConnectionError):
+    def __init__(self, request: httpx.Request) -> None:
+        super().__init__(message="Request timed out.", request=request)
diff --git a/api/core/model_runtime/model_providers/zhipuai/zhipuai_sdk/core/_files.py b/api/core/model_runtime/model_providers/zhipuai/zhipuai_sdk/core/_files.py
index 0796bfe11cc65..f9d2e14d9ecb9 100644
--- a/api/core/model_runtime/model_providers/zhipuai/zhipuai_sdk/core/_files.py
+++ b/api/core/model_runtime/model_providers/zhipuai/zhipuai_sdk/core/_files.py
@@ -2,40 +2,74 @@
 
 import io
 import os
-from collections.abc import Mapping, Sequence
-from pathlib import Path
+import pathlib
+from typing import TypeGuard, overload
 
-from ._base_type import FileTypes, HttpxFileTypes, HttpxRequestFiles, RequestFiles
+from ._base_type import (
+    Base64FileInput,
+    FileContent,
+    FileTypes,
+    HttpxFileContent,
+    HttpxFileTypes,
+    HttpxRequestFiles,
+    RequestFiles,
+)
+from ._utils import is_mapping_t, is_sequence_t, is_tuple_t
 
 
-def is_file_content(obj: object) -> bool:
+def is_base64_file_input(obj: object) -> TypeGuard[Base64FileInput]:
+    return isinstance(obj, io.IOBase | os.PathLike)
+
+
+def is_file_content(obj: object) -> TypeGuard[FileContent]:
     return isinstance(obj, bytes | tuple | io.IOBase | os.PathLike)
 
 
-def _transform_file(file: FileTypes) -> HttpxFileTypes:
-    if is_file_content(file):
-        if isinstance(file, os.PathLike):
-            path = Path(file)
-            return path.name, path.read_bytes()
-        else:
-            return file
-    if isinstance(file, tuple):
-        if isinstance(file[1], os.PathLike):
-            return (file[0], Path(file[1]).read_bytes(), *file[2:])
-        else:
-            return (file[0], file[1], *file[2:])
-    else:
-        raise TypeError(f"Unexpected input file with type {type(file)},Expected FileContent type or tuple type")
+def assert_is_file_content(obj: object, *, key: str | None = None) -> None:
+    if not is_file_content(obj):
+        prefix = f"Expected entry at `{key}`" if key is not None else f"Expected file input `{obj!r}`"
+        raise RuntimeError(
+            f"{prefix} to be bytes, an io.IOBase instance, PathLike or a tuple but received {type(obj)} instead. See https://github.com/openai/openai-python/tree/main#file-uploads"
+        ) from None
 
 
-def make_httpx_files(files: RequestFiles | None) -> HttpxRequestFiles | None:
+@overload
+def to_httpx_files(files: None) -> None: ...
+
+
+@overload
+def to_httpx_files(files: RequestFiles) -> HttpxRequestFiles: ...
+
+
+def to_httpx_files(files: RequestFiles | None) -> HttpxRequestFiles | None:
     if files is None:
         return None
 
-    if isinstance(files, Mapping):
+    if is_mapping_t(files):
         files = {key: _transform_file(file) for key, file in files.items()}
-    elif isinstance(files, Sequence):
+    elif is_sequence_t(files):
         files = [(key, _transform_file(file)) for key, file in files]
     else:
-        raise TypeError(f"Unexpected input file with type {type(files)}, excepted Mapping or Sequence")
+        raise TypeError(f"Unexpected file type input {type(files)}, expected mapping or sequence")
+
     return files
+
+
+def _transform_file(file: FileTypes) -> HttpxFileTypes:
+    if is_file_content(file):
+        if isinstance(file, os.PathLike):
+            path = pathlib.Path(file)
+            return (path.name, path.read_bytes())
+
+        return file
+
+    if is_tuple_t(file):
+        return (file[0], _read_file_content(file[1]), *file[2:])
+
+    raise TypeError("Expected file types input to be a FileContent type or to be a tuple")
+
+
+def _read_file_content(file: FileContent) -> HttpxFileContent:
+    if isinstance(file, os.PathLike):
+        return pathlib.Path(file).read_bytes()
+    return file
diff --git a/api/core/model_runtime/model_providers/zhipuai/zhipuai_sdk/core/_http_client.py b/api/core/model_runtime/model_providers/zhipuai/zhipuai_sdk/core/_http_client.py
index 5f7f6d04f20d9..d0f933d814138 100644
--- a/api/core/model_runtime/model_providers/zhipuai/zhipuai_sdk/core/_http_client.py
+++ b/api/core/model_runtime/model_providers/zhipuai/zhipuai_sdk/core/_http_client.py
@@ -1,23 +1,70 @@
 from __future__ import annotations
 
 import inspect
-from collections.abc import Mapping
-from typing import Any, Union, cast
+import logging
+import time
+import warnings
+from collections.abc import Iterator, Mapping
+from itertools import starmap
+from random import random
+from typing import TYPE_CHECKING, Any, Generic, Literal, Optional, TypeVar, Union, cast, overload
 
 import httpx
 import pydantic
 from httpx import URL, Timeout
-from tenacity import retry
-from tenacity.stop import stop_after_attempt
-
-from . import _errors
-from ._base_type import NOT_GIVEN, AnyMapping, Body, Data, Headers, NotGiven, Query, RequestFiles, ResponseT
-from ._errors import APIResponseValidationError, APIStatusError, APITimeoutError
-from ._files import make_httpx_files
-from ._request_opt import ClientRequestParam, UserRequestInput
-from ._response import HttpResponse
+
+from . import _errors, get_origin
+from ._base_compat import model_copy
+from ._base_models import GenericModel, construct_type, validate_type
+from ._base_type import (
+    NOT_GIVEN,
+    AnyMapping,
+    Body,
+    Data,
+    Headers,
+    HttpxSendArgs,
+    ModelBuilderProtocol,
+    NotGiven,
+    Omit,
+    PostParser,
+    Query,
+    RequestFiles,
+    ResponseT,
+)
+from ._constants import (
+    INITIAL_RETRY_DELAY,
+    MAX_RETRY_DELAY,
+    RAW_RESPONSE_HEADER,
+    ZHIPUAI_DEFAULT_LIMITS,
+    ZHIPUAI_DEFAULT_MAX_RETRIES,
+    ZHIPUAI_DEFAULT_TIMEOUT,
+)
+from ._errors import APIConnectionError, APIResponseValidationError, APIStatusError, APITimeoutError
+from ._files import to_httpx_files
+from ._legacy_response import LegacyAPIResponse
+from ._request_opt import FinalRequestOptions, UserRequestInput
+from ._response import APIResponse, BaseAPIResponse, extract_response_type
 from ._sse_client import StreamResponse
-from ._utils import flatten
+from ._utils import flatten, is_given, is_mapping
+
+log: logging.Logger = logging.getLogger(__name__)
+
+# TODO: make base page type vars covariant
+SyncPageT = TypeVar("SyncPageT", bound="BaseSyncPage[Any]")
+# AsyncPageT = TypeVar("AsyncPageT", bound="BaseAsyncPage[Any]")
+
+_T = TypeVar("_T")
+_T_co = TypeVar("_T_co", covariant=True)
+
+if TYPE_CHECKING:
+    from httpx._config import DEFAULT_TIMEOUT_CONFIG as HTTPX_DEFAULT_TIMEOUT
+else:
+    try:
+        from httpx._config import DEFAULT_TIMEOUT_CONFIG as HTTPX_DEFAULT_TIMEOUT
+    except ImportError:
+        # taken from https://github.com/encode/httpx/blob/3ba5fe0d7ac70222590e759c31442b1cab263791/httpx/_config.py#L366
+        HTTPX_DEFAULT_TIMEOUT = Timeout(5.0)
+
 
 headers = {
     "Accept": "application/json",
@@ -25,50 +72,180 @@
 }
 
 
-def _merge_map(map1: Mapping, map2: Mapping) -> Mapping:
-    merged = {**map1, **map2}
-    return {key: val for key, val in merged.items() if val is not None}
+class PageInfo:
+    """Stores the necessary information to build the request to retrieve the next page.
 
+    Either `url` or `params` must be set.
+    """
 
-from itertools import starmap
+    url: URL | NotGiven
+    params: Query | NotGiven
+
+    @overload
+    def __init__(
+        self,
+        *,
+        url: URL,
+    ) -> None: ...
+
+    @overload
+    def __init__(
+        self,
+        *,
+        params: Query,
+    ) -> None: ...
+
+    def __init__(
+        self,
+        *,
+        url: URL | NotGiven = NOT_GIVEN,
+        params: Query | NotGiven = NOT_GIVEN,
+    ) -> None:
+        self.url = url
+        self.params = params
+
+
+class BasePage(GenericModel, Generic[_T]):
+    """
+    Defines the core interface for pagination.
+
+    Type Args:
+        ModelT: The pydantic model that represents an item in the response.
+
+    Methods:
+        has_next_page(): Check if there is another page available
+        next_page_info(): Get the necessary information to make a request for the next page
+    """
+
+    _options: FinalRequestOptions = pydantic.PrivateAttr()
+    _model: type[_T] = pydantic.PrivateAttr()
 
-from httpx._config import DEFAULT_TIMEOUT_CONFIG as HTTPX_DEFAULT_TIMEOUT
+    def has_next_page(self) -> bool:
+        items = self._get_page_items()
+        if not items:
+            return False
+        return self.next_page_info() is not None
 
-ZHIPUAI_DEFAULT_TIMEOUT = httpx.Timeout(timeout=300.0, connect=8.0)
-ZHIPUAI_DEFAULT_MAX_RETRIES = 3
-ZHIPUAI_DEFAULT_LIMITS = httpx.Limits(max_connections=5, max_keepalive_connections=5)
+    def next_page_info(self) -> Optional[PageInfo]: ...
+
+    def _get_page_items(self) -> Iterable[_T]:  # type: ignore[empty-body]
+        ...
+
+    def _params_from_url(self, url: URL) -> httpx.QueryParams:
+        # TODO: do we have to preprocess params here?
+        return httpx.QueryParams(cast(Any, self._options.params)).merge(url.params)
+
+    def _info_to_options(self, info: PageInfo) -> FinalRequestOptions:
+        options = model_copy(self._options)
+        options._strip_raw_response_header()
+
+        if not isinstance(info.params, NotGiven):
+            options.params = {**options.params, **info.params}
+            return options
+
+        if not isinstance(info.url, NotGiven):
+            params = self._params_from_url(info.url)
+            url = info.url.copy_with(params=params)
+            options.params = dict(url.params)
+            options.url = str(url)
+            return options
+
+        raise ValueError("Unexpected PageInfo state")
+
+
+class BaseSyncPage(BasePage[_T], Generic[_T]):
+    _client: HttpClient = pydantic.PrivateAttr()
+
+    def _set_private_attributes(
+        self,
+        client: HttpClient,
+        model: type[_T],
+        options: FinalRequestOptions,
+    ) -> None:
+        self._model = model
+        self._client = client
+        self._options = options
+
+    # Pydantic uses a custom `__iter__` method to support casting BaseModels
+    # to dictionaries. e.g. dict(model).
+    # As we want to support `for item in page`, this is inherently incompatible
+    # with the default pydantic behaviour. It is not possible to support both
+    # use cases at once. Fortunately, this is not a big deal as all other pydantic
+    # methods should continue to work as expected as there is an alternative method
+    # to cast a model to a dictionary, model.dict(), which is used internally
+    # by pydantic.
+    def __iter__(self) -> Iterator[_T]:  # type: ignore
+        for page in self.iter_pages():
+            yield from page._get_page_items()
+
+    def iter_pages(self: SyncPageT) -> Iterator[SyncPageT]:
+        page = self
+        while True:
+            yield page
+            if page.has_next_page():
+                page = page.get_next_page()
+            else:
+                return
+
+    def get_next_page(self: SyncPageT) -> SyncPageT:
+        info = self.next_page_info()
+        if not info:
+            raise RuntimeError(
+                "No next page expected; please check `.has_next_page()` before calling `.get_next_page()`."
+            )
+
+        options = self._info_to_options(info)
+        return self._client._request_api_list(self._model, page=self.__class__, options=options)
 
 
 class HttpClient:
     _client: httpx.Client
     _version: str
     _base_url: URL
-
+    max_retries: int
     timeout: Union[float, Timeout, None]
     _limits: httpx.Limits
     _has_custom_http_client: bool
     _default_stream_cls: type[StreamResponse[Any]] | None = None
 
+    _strict_response_validation: bool
+
     def __init__(
         self,
         *,
         version: str,
         base_url: URL,
+        _strict_response_validation: bool,
+        max_retries: int = ZHIPUAI_DEFAULT_MAX_RETRIES,
         timeout: Union[float, Timeout, None],
+        limits: httpx.Limits | None = None,
         custom_httpx_client: httpx.Client | None = None,
         custom_headers: Mapping[str, str] | None = None,
     ) -> None:
-        if timeout is None or isinstance(timeout, NotGiven):
+        if limits is not None:
+            warnings.warn(
+                "The `connection_pool_limits` argument is deprecated. The `http_client` argument should be passed instead",  # noqa: E501
+                category=DeprecationWarning,
+                stacklevel=3,
+            )
+            if custom_httpx_client is not None:
+                raise ValueError("The `http_client` argument is mutually exclusive with `connection_pool_limits`")
+        else:
+            limits = ZHIPUAI_DEFAULT_LIMITS
+
+        if not is_given(timeout):
             if custom_httpx_client and custom_httpx_client.timeout != HTTPX_DEFAULT_TIMEOUT:
                 timeout = custom_httpx_client.timeout
             else:
                 timeout = ZHIPUAI_DEFAULT_TIMEOUT
-        self.timeout = cast(Timeout, timeout)
+        self.max_retries = max_retries
+        self.timeout = timeout
+        self._limits = limits
         self._has_custom_http_client = bool(custom_httpx_client)
         self._client = custom_httpx_client or httpx.Client(
             base_url=base_url,
             timeout=self.timeout,
-            limits=ZHIPUAI_DEFAULT_LIMITS,
+            limits=limits,
         )
         self._version = version
         url = URL(url=base_url)
@@ -76,6 +253,7 @@ def __init__(
             url = url.copy_with(raw_path=url.raw_path + b"/")
         self._base_url = url
         self._custom_headers = custom_headers or {}
+        self._strict_response_validation = _strict_response_validation
 
     def _prepare_url(self, url: str) -> URL:
         sub_url = URL(url)
@@ -93,55 +271,101 @@ def _default_headers(self):
             "ZhipuAI-SDK-Ver": self._version,
             "source_type": "zhipu-sdk-python",
             "x-request-sdk": "zhipu-sdk-python",
-            **self._auth_headers,
+            **self.auth_headers,
             **self._custom_headers,
         }
 
     @property
-    def _auth_headers(self):
+    def custom_auth(self) -> httpx.Auth | None:
+        return None
+
+    @property
+    def auth_headers(self):
         return {}
 
-    def _prepare_headers(self, request_param: ClientRequestParam) -> httpx.Headers:
-        custom_headers = request_param.headers or {}
-        headers_dict = _merge_map(self._default_headers, custom_headers)
+    def _prepare_headers(self, options: FinalRequestOptions) -> httpx.Headers:
+        custom_headers = options.headers or {}
+        headers_dict = _merge_mappings(self._default_headers, custom_headers)
 
         httpx_headers = httpx.Headers(headers_dict)
 
         return httpx_headers
 
-    def _prepare_request(self, request_param: ClientRequestParam) -> httpx.Request:
+    def _remaining_retries(
+        self,
+        remaining_retries: Optional[int],
+        options: FinalRequestOptions,
+    ) -> int:
+        return remaining_retries if remaining_retries is not None else options.get_max_retries(self.max_retries)
+
+    def _calculate_retry_timeout(
+        self,
+        remaining_retries: int,
+        options: FinalRequestOptions,
+        response_headers: Optional[httpx.Headers] = None,
+    ) -> float:
+        max_retries = options.get_max_retries(self.max_retries)
+
+        # If the API asks us to wait a certain amount of time (and it's a reasonable amount), just do what it says.
+        # retry_after = self._parse_retry_after_header(response_headers)
+        # if retry_after is not None and 0 < retry_after <= 60:
+        #     return retry_after
+
+        nb_retries = max_retries - remaining_retries
+
+        # Apply exponential backoff, but not more than the max.
+        sleep_seconds = min(INITIAL_RETRY_DELAY * pow(2.0, nb_retries), MAX_RETRY_DELAY)
+
+        # Apply some jitter, plus-or-minus half a second.
+        jitter = 1 - 0.25 * random()
+        timeout = sleep_seconds * jitter
+        return max(timeout, 0)
+
+    def _build_request(self, options: FinalRequestOptions) -> httpx.Request:
         kwargs: dict[str, Any] = {}
-        json_data = request_param.json_data
-        headers = self._prepare_headers(request_param)
-        url = self._prepare_url(request_param.url)
-        json_data = request_param.json_data
+        headers = self._prepare_headers(options)
+        url = self._prepare_url(options.url)
+        json_data = options.json_data
+        if options.extra_json is not None:
+            if json_data is None:
+                json_data = cast(Body, options.extra_json)
+            elif is_mapping(json_data):
+                json_data = _merge_mappings(json_data, options.extra_json)
+            else:
+                raise RuntimeError(f"Unexpected JSON data type, {type(json_data)}, cannot merge with `extra_body`")
+
+        content_type = headers.get("Content-Type")
+        # multipart/form-data; boundary=---abc--
         if headers.get("Content-Type") == "multipart/form-data":
-            headers.pop("Content-Type")
+            if "boundary" not in content_type:
+                # only remove the header if the boundary hasn't been explicitly set
+                # as the caller doesn't want httpx to come up with their own boundary
+                headers.pop("Content-Type")
 
             if json_data:
                 kwargs["data"] = self._make_multipartform(json_data)
 
         return self._client.build_request(
             headers=headers,
-            timeout=self.timeout if isinstance(request_param.timeout, NotGiven) else request_param.timeout,
-            method=request_param.method,
+            timeout=self.timeout if isinstance(options.timeout, NotGiven) else options.timeout,
+            method=options.method,
             url=url,
             json=json_data,
-            files=request_param.files,
-            params=request_param.params,
+            files=options.files,
+            params=options.params,
             **kwargs,
         )
 
-    def _object_to_formdata(self, key: str, value: Data | Mapping[object, object]) -> list[tuple[str, str]]:
+    def _object_to_formfata(self, key: str, value: Data | Mapping[object, object]) -> list[tuple[str, str]]:
         items = []
 
         if isinstance(value, Mapping):
             for k, v in value.items():
-                items.extend(self._object_to_formdata(f"{key}[{k}]", v))
+                items.extend(self._object_to_formfata(f"{key}[{k}]", v))
             return items
         if isinstance(value, list | tuple):
             for v in value:
-                items.extend(self._object_to_formdata(key + "[]", v))
+                items.extend(self._object_to_formfata(key + "[]", v))
             return items
 
         def _primitive_value_to_str(val) -> str:
@@ -161,7 +385,7 @@ def _primitive_value_to_str(val) -> str:
         return [(key, str_data)]
 
     def _make_multipartform(self, data: Mapping[object, object]) -> dict[str, object]:
-        items = flatten(list(starmap(self._object_to_formdata, data.items())))
+        items = flatten(list(starmap(self._object_to_formfata, data.items())))
 
         serialized: dict[str, object] = {}
         for key, value in items:
@@ -170,20 +394,6 @@ def _make_multipartform(self, data: Mapping[object, object]) -> dict[str, object
             serialized[key] = value
         return serialized
 
-    def _parse_response(
-        self,
-        *,
-        cast_type: type[ResponseT],
-        response: httpx.Response,
-        enable_stream: bool,
-        request_param: ClientRequestParam,
-        stream_cls: type[StreamResponse[Any]] | None = None,
-    ) -> HttpResponse:
-        http_response = HttpResponse(
-            raw_response=response, cast_type=cast_type, client=self, enable_stream=enable_stream, stream_cls=stream_cls
-        )
-        return http_response.parse()
-
     def _process_response_data(
         self,
         *,
@@ -194,14 +404,58 @@ def _process_response_data(
         if data is None:
             return cast(ResponseT, None)
 
+        if cast_type is object:
+            return cast(ResponseT, data)
+
         try:
-            if inspect.isclass(cast_type) and issubclass(cast_type, pydantic.BaseModel):
-                return cast(ResponseT, cast_type.validate(data))
+            if inspect.isclass(cast_type) and issubclass(cast_type, ModelBuilderProtocol):
+                return cast(ResponseT, cast_type.build(response=response, data=data))
+
+            if self._strict_response_validation:
+                return cast(ResponseT, validate_type(type_=cast_type, value=data))
 
-            return cast(ResponseT, pydantic.TypeAdapter(cast_type).validate_python(data))
+            return cast(ResponseT, construct_type(type_=cast_type, value=data))
         except pydantic.ValidationError as err:
             raise APIResponseValidationError(response=response, json_data=data) from err
 
+    def _should_stream_response_body(self, request: httpx.Request) -> bool:
+        return request.headers.get(RAW_RESPONSE_HEADER) == "stream"  # type: ignore[no-any-return]
+
+    def _should_retry(self, response: httpx.Response) -> bool:
+        # Note: this is not a standard header
+        should_retry_header = response.headers.get("x-should-retry")
+
+        # If the server explicitly says whether or not to retry, obey.
+        if should_retry_header == "true":
+            log.debug("Retrying as header `x-should-retry` is set to `true`")
+            return True
+        if should_retry_header == "false":
+            log.debug("Not retrying as header `x-should-retry` is set to `false`")
+            return False
+
+        # Retry on request timeouts.
+        if response.status_code == 408:
+            log.debug("Retrying due to status code %i", response.status_code)
+            return True
+
+        # Retry on lock timeouts.
+        if response.status_code == 409:
+            log.debug("Retrying due to status code %i", response.status_code)
+            return True
+
+        # Retry on rate limits.
+        if response.status_code == 429:
+            log.debug("Retrying due to status code %i", response.status_code)
+            return True
+
+        # Retry internal errors.
+        if response.status_code >= 500:
+            log.debug("Retrying due to status code %i", response.status_code)
+            return True
+
+        log.debug("Not retrying")
+        return False
+
     def is_closed(self) -> bool:
         return self._client.is_closed
 
@@ -214,117 +468,385 @@ def __enter__(self):
     def __exit__(self, exc_type, exc_val, exc_tb):
         self.close()
 
-    @retry(stop=stop_after_attempt(ZHIPUAI_DEFAULT_MAX_RETRIES))
     def request(
+        self,
+        cast_type: type[ResponseT],
+        options: FinalRequestOptions,
+        remaining_retries: Optional[int] = None,
+        *,
+        stream: bool = False,
+        stream_cls: type[StreamResponse] | None = None,
+    ) -> ResponseT | StreamResponse:
+        return self._request(
+            cast_type=cast_type,
+            options=options,
+            stream=stream,
+            stream_cls=stream_cls,
+            remaining_retries=remaining_retries,
+        )
+
+    def _request(
         self,
         *,
         cast_type: type[ResponseT],
-        params: ClientRequestParam,
-        enable_stream: bool = False,
-        stream_cls: type[StreamResponse[Any]] | None = None,
+        options: FinalRequestOptions,
+        remaining_retries: int | None,
+        stream: bool,
+        stream_cls: type[StreamResponse] | None,
     ) -> ResponseT | StreamResponse:
-        request = self._prepare_request(params)
+        retries = self._remaining_retries(remaining_retries, options)
+        request = self._build_request(options)
 
+        kwargs: HttpxSendArgs = {}
+        if self.custom_auth is not None:
+            kwargs["auth"] = self.custom_auth
         try:
             response = self._client.send(
                 request,
-                stream=enable_stream,
+                stream=stream or self._should_stream_response_body(request=request),
+                **kwargs,
             )
-            response.raise_for_status()
         except httpx.TimeoutException as err:
+            log.debug("Encountered httpx.TimeoutException", exc_info=True)
+
+            if retries > 0:
+                return self._retry_request(
+                    options,
+                    cast_type,
+                    retries,
+                    stream=stream,
+                    stream_cls=stream_cls,
+                    response_headers=None,
+                )
+
+            log.debug("Raising timeout error")
             raise APITimeoutError(request=request) from err
-        except httpx.HTTPStatusError as err:
-            err.response.read()
-            # raise err
-            raise self._make_status_error(err.response) from None
-
         except Exception as err:
-            raise err
+            log.debug("Encountered Exception", exc_info=True)
+
+            if retries > 0:
+                return self._retry_request(
+                    options,
+                    cast_type,
+                    retries,
+                    stream=stream,
+                    stream_cls=stream_cls,
+                    response_headers=None,
+                )
+
+            log.debug("Raising connection error")
+            raise APIConnectionError(request=request) from err
+
+        log.debug(
+            'HTTP Request: %s %s "%i %s"', request.method, request.url, response.status_code, response.reason_phrase
+        )
 
-        return self._parse_response(
+        try:
+            response.raise_for_status()
+        except httpx.HTTPStatusError as err:  # thrown on 4xx and 5xx status code
+            log.debug("Encountered httpx.HTTPStatusError", exc_info=True)
+
+            if retries > 0 and self._should_retry(err.response):
+                err.response.close()
+                return self._retry_request(
+                    options,
+                    cast_type,
+                    retries,
+                    err.response.headers,
+                    stream=stream,
+                    stream_cls=stream_cls,
+                )
+
+            # If the response is streamed then we need to explicitly read the response
+            # to completion before attempting to access the response text.
+            if not err.response.is_closed:
+                err.response.read()
+
+            log.debug("Re-raising status error")
+            raise self._make_status_error(err.response) from None
+
+        # return self._parse_response(
+        #     cast_type=cast_type,
+        #     options=options,
+        #     response=response,
+        #     stream=stream,
+        #     stream_cls=stream_cls,
+        # )
+        return self._process_response(
             cast_type=cast_type,
-            request_param=params,
+            options=options,
             response=response,
-            enable_stream=enable_stream,
+            stream=stream,
+            stream_cls=stream_cls,
+        )
+
+    def _retry_request(
+        self,
+        options: FinalRequestOptions,
+        cast_type: type[ResponseT],
+        remaining_retries: int,
+        response_headers: httpx.Headers | None,
+        *,
+        stream: bool,
+        stream_cls: type[StreamResponse] | None,
+    ) -> ResponseT | StreamResponse:
+        remaining = remaining_retries - 1
+        if remaining == 1:
+            log.debug("1 retry left")
+        else:
+            log.debug("%i retries left", remaining)
+
+        timeout = self._calculate_retry_timeout(remaining, options, response_headers)
+        log.info("Retrying request to %s in %f seconds", options.url, timeout)
+
+        # In a synchronous context we are blocking the entire thread. Up to the library user to run the client in a
+        # different thread if necessary.
+        time.sleep(timeout)
+
+        return self._request(
+            options=options,
+            cast_type=cast_type,
+            remaining_retries=remaining,
+            stream=stream,
+            stream_cls=stream_cls,
+        )
+
+    def _process_response(
+        self,
+        *,
+        cast_type: type[ResponseT],
+        options: FinalRequestOptions,
+        response: httpx.Response,
+        stream: bool,
+        stream_cls: type[StreamResponse] | None,
+    ) -> ResponseT:
+        # _legacy_response with raw_response_header to paser method
+        if response.request.headers.get(RAW_RESPONSE_HEADER) == "true":
+            return cast(
+                ResponseT,
+                LegacyAPIResponse(
+                    raw=response,
+                    client=self,
+                    cast_type=cast_type,
+                    stream=stream,
+                    stream_cls=stream_cls,
+                    options=options,
+                ),
+            )
+
+        origin = get_origin(cast_type) or cast_type
+
+        if inspect.isclass(origin) and issubclass(origin, BaseAPIResponse):
+            if not issubclass(origin, APIResponse):
+                raise TypeError(f"API Response types must subclass {APIResponse}; Received {origin}")
+
+            response_cls = cast("type[BaseAPIResponse[Any]]", cast_type)
+            return cast(
+                ResponseT,
+                response_cls(
+                    raw=response,
+                    client=self,
+                    cast_type=extract_response_type(response_cls),
+                    stream=stream,
+                    stream_cls=stream_cls,
+                    options=options,
+                ),
+            )
+
+        if cast_type == httpx.Response:
+            return cast(ResponseT, response)
+
+        api_response = APIResponse(
+            raw=response,
+            client=self,
+            cast_type=cast("type[ResponseT]", cast_type),  # pyright: ignore[reportUnnecessaryCast]
+            stream=stream,
             stream_cls=stream_cls,
+            options=options,
         )
+        if bool(response.request.headers.get(RAW_RESPONSE_HEADER)):
+            return cast(ResponseT, api_response)
+
+        return api_response.parse()
+
+    def _request_api_list(
+        self,
+        model: type[object],
+        page: type[SyncPageT],
+        options: FinalRequestOptions,
+    ) -> SyncPageT:
+        def _parser(resp: SyncPageT) -> SyncPageT:
+            resp._set_private_attributes(
+                client=self,
+                model=model,
+                options=options,
+            )
+            return resp
+
+        options.post_parser = _parser
+
+        return self.request(page, options, stream=False)
 
+    @overload
     def get(
         self,
         path: str,
         *,
         cast_type: type[ResponseT],
         options: UserRequestInput = {},
-        enable_stream: bool = False,
-    ) -> ResponseT | StreamResponse:
-        opts = ClientRequestParam.construct(method="get", url=path, **options)
-        return self.request(cast_type=cast_type, params=opts, enable_stream=enable_stream)
+        stream: Literal[False] = False,
+    ) -> ResponseT: ...
 
+    @overload
+    def get(
+        self,
+        path: str,
+        *,
+        cast_type: type[ResponseT],
+        options: UserRequestInput = {},
+        stream: Literal[True],
+        stream_cls: type[StreamResponse],
+    ) -> StreamResponse: ...
+
+    @overload
+    def get(
+        self,
+        path: str,
+        *,
+        cast_type: type[ResponseT],
+        options: UserRequestInput = {},
+        stream: bool,
+        stream_cls: type[StreamResponse] | None = None,
+    ) -> ResponseT | StreamResponse: ...
+
+    def get(
+        self,
+        path: str,
+        *,
+        cast_type: type[ResponseT],
+        options: UserRequestInput = {},
+        stream: bool = False,
+        stream_cls: type[StreamResponse] | None = None,
+    ) -> ResponseT:
+        opts = FinalRequestOptions.construct(method="get", url=path, **options)
+        return cast(ResponseT, self.request(cast_type, opts, stream=stream, stream_cls=stream_cls))
+
+    @overload
     def post(
         self,
         path: str,
         *,
+        cast_type: type[ResponseT],
         body: Body | None = None,
+        options: UserRequestInput = {},
+        files: RequestFiles | None = None,
+        stream: Literal[False] = False,
+    ) -> ResponseT: ...
+
+    @overload
+    def post(
+        self,
+        path: str,
+        *,
         cast_type: type[ResponseT],
+        body: Body | None = None,
         options: UserRequestInput = {},
         files: RequestFiles | None = None,
-        enable_stream: bool = False,
+        stream: Literal[True],
+        stream_cls: type[StreamResponse],
+    ) -> StreamResponse: ...
+
+    @overload
+    def post(
+        self,
+        path: str,
+        *,
+        cast_type: type[ResponseT],
+        body: Body | None = None,
+        options: UserRequestInput = {},
+        files: RequestFiles | None = None,
+        stream: bool,
+        stream_cls: type[StreamResponse] | None = None,
+    ) -> ResponseT | StreamResponse: ...
+
+    def post(
+        self,
+        path: str,
+        *,
+        cast_type: type[ResponseT],
+        body: Body | None = None,
+        options: UserRequestInput = {},
+        files: RequestFiles | None = None,
+        stream: bool = False,
         stream_cls: type[StreamResponse[Any]] | None = None,
     ) -> ResponseT | StreamResponse:
-        opts = ClientRequestParam.construct(
-            method="post", json_data=body, files=make_httpx_files(files), url=path, **options
+        opts = FinalRequestOptions.construct(
+            method="post", url=path, json_data=body, files=to_httpx_files(files), **options
         )
 
-        return self.request(cast_type=cast_type, params=opts, enable_stream=enable_stream, stream_cls=stream_cls)
+        return cast(ResponseT, self.request(cast_type, opts, stream=stream, stream_cls=stream_cls))
 
     def patch(
         self,
         path: str,
         *,
-        body: Body | None = None,
         cast_type: type[ResponseT],
+        body: Body | None = None,
         options: UserRequestInput = {},
     ) -> ResponseT:
-        opts = ClientRequestParam.construct(method="patch", url=path, json_data=body, **options)
+        opts = FinalRequestOptions.construct(method="patch", url=path, json_data=body, **options)
 
         return self.request(
             cast_type=cast_type,
-            params=opts,
+            options=opts,
         )
 
     def put(
         self,
         path: str,
         *,
-        body: Body | None = None,
         cast_type: type[ResponseT],
+        body: Body | None = None,
         options: UserRequestInput = {},
         files: RequestFiles | None = None,
     ) -> ResponseT | StreamResponse:
-        opts = ClientRequestParam.construct(
-            method="put", url=path, json_data=body, files=make_httpx_files(files), **options
+        opts = FinalRequestOptions.construct(
+            method="put", url=path, json_data=body, files=to_httpx_files(files), **options
         )
 
         return self.request(
             cast_type=cast_type,
-            params=opts,
+            options=opts,
         )
 
     def delete(
         self,
         path: str,
         *,
-        body: Body | None = None,
         cast_type: type[ResponseT],
+        body: Body | None = None,
         options: UserRequestInput = {},
     ) -> ResponseT | StreamResponse:
-        opts = ClientRequestParam.construct(method="delete", url=path, json_data=body, **options)
+        opts = FinalRequestOptions.construct(method="delete", url=path, json_data=body, **options)
 
         return self.request(
             cast_type=cast_type,
-            params=opts,
+            options=opts,
         )
 
+    def get_api_list(
+        self,
+        path: str,
+        *,
+        model: type[object],
+        page: type[SyncPageT],
+        body: Body | None = None,
+        options: UserRequestInput = {},
+        method: str = "get",
+    ) -> SyncPageT:
+        opts = FinalRequestOptions.construct(method=method, url=path, json_data=body, **options)
+        return self._request_api_list(model, page, opts)
+
     def _make_status_error(self, response) -> APIStatusError:
         response_text = response.text.strip()
         status_code = response.status_code
@@ -343,24 +865,46 @@ def _make_status_error(self, response) -> APIStatusError:
         return APIStatusError(message=error_msg, response=response)
 
 
-def make_user_request_input(
-    max_retries: int | None = None,
-    timeout: float | Timeout | None | NotGiven = NOT_GIVEN,
-    extra_headers: Headers = None,
-    extra_body: Body | None = None,
+def make_request_options(
+    *,
     query: Query | None = None,
+    extra_headers: Headers | None = None,
+    extra_query: Query | None = None,
+    extra_body: Body | None = None,
+    timeout: float | httpx.Timeout | None | NotGiven = NOT_GIVEN,
+    post_parser: PostParser | NotGiven = NOT_GIVEN,
 ) -> UserRequestInput:
+    """Create a dict of type RequestOptions without keys of NotGiven values."""
     options: UserRequestInput = {}
-
     if extra_headers is not None:
         options["headers"] = extra_headers
-    if max_retries is not None:
-        options["max_retries"] = max_retries
-    if not isinstance(timeout, NotGiven):
-        options["timeout"] = timeout
-    if query is not None:
-        options["params"] = query
+
     if extra_body is not None:
         options["extra_json"] = cast(AnyMapping, extra_body)
 
+    if query is not None:
+        options["params"] = query
+
+    if extra_query is not None:
+        options["params"] = {**options.get("params", {}), **extra_query}
+
+    if not isinstance(timeout, NotGiven):
+        options["timeout"] = timeout
+
+    if is_given(post_parser):
+        # internal
+        options["post_parser"] = post_parser  # type: ignore
+
     return options
+
+
+def _merge_mappings(
+    obj1: Mapping[_T_co, Union[_T, Omit]],
+    obj2: Mapping[_T_co, Union[_T, Omit]],
+) -> dict[_T_co, _T]:
+    """Merge two mappings of the same type, removing any values that are instances of `Omit`.
+
+    In cases with duplicate keys the second mapping takes precedence.
+    """
+    merged = {**obj1, **obj2}
+    return {key: value for key, value in merged.items() if not isinstance(value, Omit)}
diff --git a/api/core/model_runtime/model_providers/zhipuai/zhipuai_sdk/core/_jwt_token.py b/api/core/model_runtime/model_providers/zhipuai/zhipuai_sdk/core/_jwt_token.py
index b0a91d04a9944..21f158a5f4525 100644
--- a/api/core/model_runtime/model_providers/zhipuai/zhipuai_sdk/core/_jwt_token.py
+++ b/api/core/model_runtime/model_providers/zhipuai/zhipuai_sdk/core/_jwt_token.py
@@ -3,9 +3,11 @@
 import cachetools.func
 import jwt
 
-API_TOKEN_TTL_SECONDS = 3 * 60
+# 缓存时间 3分钟
+CACHE_TTL_SECONDS = 3 * 60
 
-CACHE_TTL_SECONDS = API_TOKEN_TTL_SECONDS - 30
+# token 有效期比缓存时间 多30秒
+API_TOKEN_TTL_SECONDS = CACHE_TTL_SECONDS + 30
 
 
 @cachetools.func.ttl_cache(maxsize=10, ttl=CACHE_TTL_SECONDS)
diff --git a/api/core/model_runtime/model_providers/zhipuai/zhipuai_sdk/core/_legacy_binary_response.py b/api/core/model_runtime/model_providers/zhipuai/zhipuai_sdk/core/_legacy_binary_response.py
new file mode 100644
index 0000000000000..51623bd860951
--- /dev/null
+++ b/api/core/model_runtime/model_providers/zhipuai/zhipuai_sdk/core/_legacy_binary_response.py
@@ -0,0 +1,207 @@
+from __future__ import annotations
+
+import os
+from collections.abc import AsyncIterator, Iterator
+from typing import Any
+
+import httpx
+
+
+class HttpxResponseContent:
+    @property
+    def content(self) -> bytes:
+        raise NotImplementedError("This method is not implemented for this class.")
+
+    @property
+    def text(self) -> str:
+        raise NotImplementedError("This method is not implemented for this class.")
+
+    @property
+    def encoding(self) -> str | None:
+        raise NotImplementedError("This method is not implemented for this class.")
+
+    @property
+    def charset_encoding(self) -> str | None:
+        raise NotImplementedError("This method is not implemented for this class.")
+
+    def json(self, **kwargs: Any) -> Any:
+        raise NotImplementedError("This method is not implemented for this class.")
+
+    def read(self) -> bytes:
+        raise NotImplementedError("This method is not implemented for this class.")
+
+    def iter_bytes(self, chunk_size: int | None = None) -> Iterator[bytes]:
+        raise NotImplementedError("This method is not implemented for this class.")
+
+    def iter_text(self, chunk_size: int | None = None) -> Iterator[str]:
+        raise NotImplementedError("This method is not implemented for this class.")
+
+    def iter_lines(self) -> Iterator[str]:
+        raise NotImplementedError("This method is not implemented for this class.")
+
+    def iter_raw(self, chunk_size: int | None = None) -> Iterator[bytes]:
+        raise NotImplementedError("This method is not implemented for this class.")
+
+    def write_to_file(
+        self,
+        file: str | os.PathLike[str],
+    ) -> None:
+        raise NotImplementedError("This method is not implemented for this class.")
+
+    def stream_to_file(
+        self,
+        file: str | os.PathLike[str],
+        *,
+        chunk_size: int | None = None,
+    ) -> None:
+        raise NotImplementedError("This method is not implemented for this class.")
+
+    def close(self) -> None:
+        raise NotImplementedError("This method is not implemented for this class.")
+
+    async def aread(self) -> bytes:
+        raise NotImplementedError("This method is not implemented for this class.")
+
+    async def aiter_bytes(self, chunk_size: int | None = None) -> AsyncIterator[bytes]:
+        raise NotImplementedError("This method is not implemented for this class.")
+
+    async def aiter_text(self, chunk_size: int | None = None) -> AsyncIterator[str]:
+        raise NotImplementedError("This method is not implemented for this class.")
+
+    async def aiter_lines(self) -> AsyncIterator[str]:
+        raise NotImplementedError("This method is not implemented for this class.")
+
+    async def aiter_raw(self, chunk_size: int | None = None) -> AsyncIterator[bytes]:
+        raise NotImplementedError("This method is not implemented for this class.")
+
+    async def astream_to_file(
+        self,
+        file: str | os.PathLike[str],
+        *,
+        chunk_size: int | None = None,
+    ) -> None:
+        raise NotImplementedError("This method is not implemented for this class.")
+
+    async def aclose(self) -> None:
+        raise NotImplementedError("This method is not implemented for this class.")
+
+
+class HttpxBinaryResponseContent(HttpxResponseContent):
+    response: httpx.Response
+
+    def __init__(self, response: httpx.Response) -> None:
+        self.response = response
+
+    @property
+    def content(self) -> bytes:
+        return self.response.content
+
+    @property
+    def encoding(self) -> str | None:
+        return self.response.encoding
+
+    @property
+    def charset_encoding(self) -> str | None:
+        return self.response.charset_encoding
+
+    def read(self) -> bytes:
+        return self.response.read()
+
+    def text(self) -> str:
+        raise NotImplementedError("Not implemented for binary response content")
+
+    def json(self, **kwargs: Any) -> Any:
+        raise NotImplementedError("Not implemented for binary response content")
+
+    def iter_text(self, chunk_size: int | None = None) -> Iterator[str]:
+        raise NotImplementedError("Not implemented for binary response content")
+
+    def iter_lines(self) -> Iterator[str]:
+        raise NotImplementedError("Not implemented for binary response content")
+
+    async def aiter_text(self, chunk_size: int | None = None) -> AsyncIterator[str]:
+        raise NotImplementedError("Not implemented for binary response content")
+
+    async def aiter_lines(self) -> AsyncIterator[str]:
+        raise NotImplementedError("Not implemented for binary response content")
+
+    def iter_bytes(self, chunk_size: int | None = None) -> Iterator[bytes]:
+        return self.response.iter_bytes(chunk_size)
+
+    def iter_raw(self, chunk_size: int | None = None) -> Iterator[bytes]:
+        return self.response.iter_raw(chunk_size)
+
+    def write_to_file(
+        self,
+        file: str | os.PathLike[str],
+    ) -> None:
+        """Write the output to the given file.
+
+        Accepts a filename or any path-like object, e.g. pathlib.Path
+
+        Note: if you want to stream the data to the file instead of writing
+        all at once then you should use `.with_streaming_response` when making
+        the API request, e.g. `client.with_streaming_response.foo().stream_to_file('my_filename.txt')`
+        """
+        with open(file, mode="wb") as f:
+            for data in self.response.iter_bytes():
+                f.write(data)
+
+    def stream_to_file(
+        self,
+        file: str | os.PathLike[str],
+        *,
+        chunk_size: int | None = None,
+    ) -> None:
+        with open(file, mode="wb") as f:
+            for data in self.response.iter_bytes(chunk_size):
+                f.write(data)
+
+    def close(self) -> None:
+        return self.response.close()
+
+    async def aread(self) -> bytes:
+        return await self.response.aread()
+
+    async def aiter_bytes(self, chunk_size: int | None = None) -> AsyncIterator[bytes]:
+        return self.response.aiter_bytes(chunk_size)
+
+    async def aiter_raw(self, chunk_size: int | None = None) -> AsyncIterator[bytes]:
+        return self.response.aiter_raw(chunk_size)
+
+    async def astream_to_file(
+        self,
+        file: str | os.PathLike[str],
+        *,
+        chunk_size: int | None = None,
+    ) -> None:
+        path = anyio.Path(file)
+        async with await path.open(mode="wb") as f:
+            async for data in self.response.aiter_bytes(chunk_size):
+                await f.write(data)
+
+    async def aclose(self) -> None:
+        return await self.response.aclose()
+
+
+class HttpxTextBinaryResponseContent(HttpxBinaryResponseContent):
+    response: httpx.Response
+
+    @property
+    def text(self) -> str:
+        return self.response.text
+
+    def json(self, **kwargs: Any) -> Any:
+        return self.response.json(**kwargs)
+
+    def iter_text(self, chunk_size: int | None = None) -> Iterator[str]:
+        return self.response.iter_text(chunk_size)
+
+    def iter_lines(self) -> Iterator[str]:
+        return self.response.iter_lines()
+
+    async def aiter_text(self, chunk_size: int | None = None) -> AsyncIterator[str]:
+        return self.response.aiter_text(chunk_size)
+
+    async def aiter_lines(self) -> AsyncIterator[str]:
+        return self.response.aiter_lines()
diff --git a/api/core/model_runtime/model_providers/zhipuai/zhipuai_sdk/core/_legacy_response.py b/api/core/model_runtime/model_providers/zhipuai/zhipuai_sdk/core/_legacy_response.py
new file mode 100644
index 0000000000000..47183b9eee9c0
--- /dev/null
+++ b/api/core/model_runtime/model_providers/zhipuai/zhipuai_sdk/core/_legacy_response.py
@@ -0,0 +1,341 @@
+from __future__ import annotations
+
+import datetime
+import functools
+import inspect
+import logging
+from collections.abc import Callable
+from typing import TYPE_CHECKING, Any, Generic, TypeVar, Union, cast, get_origin, overload
+
+import httpx
+import pydantic
+from typing_extensions import ParamSpec, override
+
+from ._base_models import BaseModel, is_basemodel
+from ._base_type import NoneType
+from ._constants import RAW_RESPONSE_HEADER
+from ._errors import APIResponseValidationError
+from ._legacy_binary_response import HttpxResponseContent, HttpxTextBinaryResponseContent
+from ._sse_client import StreamResponse, extract_stream_chunk_type, is_stream_class_type
+from ._utils import extract_type_arg, is_annotated_type, is_given
+
+if TYPE_CHECKING:
+    from ._http_client import HttpClient
+    from ._request_opt import FinalRequestOptions
+
+P = ParamSpec("P")
+R = TypeVar("R")
+_T = TypeVar("_T")
+
+log: logging.Logger = logging.getLogger(__name__)
+
+
+class LegacyAPIResponse(Generic[R]):
+    """This is a legacy class as it will be replaced by `APIResponse`
+    and `AsyncAPIResponse` in the `_response.py` file in the next major
+    release.
+
+    For the sync client this will mostly be the same with the exception
+    of `content` & `text` will be methods instead of properties. In the
+    async client, all methods will be async.
+
+    A migration script will be provided & the migration in general should
+    be smooth.
+    """
+
+    _cast_type: type[R]
+    _client: HttpClient
+    _parsed_by_type: dict[type[Any], Any]
+    _stream: bool
+    _stream_cls: type[StreamResponse[Any]] | None
+    _options: FinalRequestOptions
+
+    http_response: httpx.Response
+
+    def __init__(
+        self,
+        *,
+        raw: httpx.Response,
+        cast_type: type[R],
+        client: HttpClient,
+        stream: bool,
+        stream_cls: type[StreamResponse[Any]] | None,
+        options: FinalRequestOptions,
+    ) -> None:
+        self._cast_type = cast_type
+        self._client = client
+        self._parsed_by_type = {}
+        self._stream = stream
+        self._stream_cls = stream_cls
+        self._options = options
+        self.http_response = raw
+
+    @property
+    def request_id(self) -> str | None:
+        return self.http_response.headers.get("x-request-id")  # type: ignore[no-any-return]
+
+    @overload
+    def parse(self, *, to: type[_T]) -> _T: ...
+
+    @overload
+    def parse(self) -> R: ...
+
+    def parse(self, *, to: type[_T] | None = None) -> R | _T:
+        """Returns the rich python representation of this response's data.
+
+        NOTE: For the async client: this will become a coroutine in the next major version.
+
+        For lower-level control, see `.read()`, `.json()`, `.iter_bytes()`.
+
+        You can customise the type that the response is parsed into through
+        the `to` argument, e.g.
+
+        ```py
+        from zhipuai import BaseModel
+
+
+        class MyModel(BaseModel):
+            foo: str
+
+
+        obj = response.parse(to=MyModel)
+        print(obj.foo)
+        ```
+
+        We support parsing:
+          - `BaseModel`
+          - `dict`
+          - `list`
+          - `Union`
+          - `str`
+          - `int`
+          - `float`
+          - `httpx.Response`
+        """
+        cache_key = to if to is not None else self._cast_type
+        cached = self._parsed_by_type.get(cache_key)
+        if cached is not None:
+            return cached  # type: ignore[no-any-return]
+
+        parsed = self._parse(to=to)
+        if is_given(self._options.post_parser):
+            parsed = self._options.post_parser(parsed)
+
+        self._parsed_by_type[cache_key] = parsed
+        return parsed
+
+    @property
+    def headers(self) -> httpx.Headers:
+        return self.http_response.headers
+
+    @property
+    def http_request(self) -> httpx.Request:
+        return self.http_response.request
+
+    @property
+    def status_code(self) -> int:
+        return self.http_response.status_code
+
+    @property
+    def url(self) -> httpx.URL:
+        return self.http_response.url
+
+    @property
+    def method(self) -> str:
+        return self.http_request.method
+
+    @property
+    def content(self) -> bytes:
+        """Return the binary response content.
+
+        NOTE: this will be removed in favour of `.read()` in the
+        next major version.
+        """
+        return self.http_response.content
+
+    @property
+    def text(self) -> str:
+        """Return the decoded response content.
+
+        NOTE: this will be turned into a method in the next major version.
+        """
+        return self.http_response.text
+
+    @property
+    def http_version(self) -> str:
+        return self.http_response.http_version
+
+    @property
+    def is_closed(self) -> bool:
+        return self.http_response.is_closed
+
+    @property
+    def elapsed(self) -> datetime.timedelta:
+        """The time taken for the complete request/response cycle to complete."""
+        return self.http_response.elapsed
+
+    def _parse(self, *, to: type[_T] | None = None) -> R | _T:
+        # unwrap `Annotated[T, ...]` -> `T`
+        if to and is_annotated_type(to):
+            to = extract_type_arg(to, 0)
+
+        if self._stream:
+            if to:
+                if not is_stream_class_type(to):
+                    raise TypeError(f"Expected custom parse type to be a subclass of {StreamResponse}")
+
+                return cast(
+                    _T,
+                    to(
+                        cast_type=extract_stream_chunk_type(
+                            to,
+                            failure_message="Expected custom stream type to be passed with a type argument, e.g. StreamResponse[ChunkType]",  # noqa: E501
+                        ),
+                        response=self.http_response,
+                        client=cast(Any, self._client),
+                    ),
+                )
+
+            if self._stream_cls:
+                return cast(
+                    R,
+                    self._stream_cls(
+                        cast_type=extract_stream_chunk_type(self._stream_cls),
+                        response=self.http_response,
+                        client=cast(Any, self._client),
+                    ),
+                )
+
+            stream_cls = cast("type[StreamResponse[Any]] | None", self._client._default_stream_cls)
+            if stream_cls is None:
+                raise MissingStreamClassError()
+
+            return cast(
+                R,
+                stream_cls(
+                    cast_type=self._cast_type,
+                    response=self.http_response,
+                    client=cast(Any, self._client),
+                ),
+            )
+
+        cast_type = to if to is not None else self._cast_type
+
+        # unwrap `Annotated[T, ...]` -> `T`
+        if is_annotated_type(cast_type):
+            cast_type = extract_type_arg(cast_type, 0)
+
+        if cast_type is NoneType:
+            return cast(R, None)
+
+        response = self.http_response
+        if cast_type == str:
+            return cast(R, response.text)
+
+        if cast_type == int:
+            return cast(R, int(response.text))
+
+        if cast_type == float:
+            return cast(R, float(response.text))
+
+        origin = get_origin(cast_type) or cast_type
+
+        if inspect.isclass(origin) and issubclass(origin, HttpxResponseContent):
+            # in the response, e.g. mime file
+            *_, filename = response.headers.get("content-disposition", "").split("filename=")
+            # 判断文件类型是jsonl类型的使用HttpxTextBinaryResponseContent
+            if filename and filename.endswith(".jsonl") or filename and filename.endswith(".xlsx"):
+                return cast(R, HttpxTextBinaryResponseContent(response))
+            else:
+                return cast(R, cast_type(response))  # type: ignore
+
+        if origin == LegacyAPIResponse:
+            raise RuntimeError("Unexpected state - cast_type is `APIResponse`")
+
+        if inspect.isclass(origin) and issubclass(origin, httpx.Response):
+            # Because of the invariance of our ResponseT TypeVar, users can subclass httpx.Response
+            # and pass that class to our request functions. We cannot change the variance to be either
+            # covariant or contravariant as that makes our usage of ResponseT illegal. We could construct
+            # the response class ourselves but that is something that should be supported directly in httpx
+            # as it would be easy to incorrectly construct the Response object due to the multitude of arguments.
+            if cast_type != httpx.Response:
+                raise ValueError("Subclasses of httpx.Response cannot be passed to `cast_type`")
+            return cast(R, response)
+
+        if inspect.isclass(origin) and not issubclass(origin, BaseModel) and issubclass(origin, pydantic.BaseModel):
+            raise TypeError("Pydantic models must subclass our base model type, e.g. `from openai import BaseModel`")
+
+        if (
+            cast_type is not object
+            and origin is not list
+            and origin is not dict
+            and origin is not Union
+            and not issubclass(origin, BaseModel)
+        ):
+            raise RuntimeError(
+                f"Unsupported type, expected {cast_type} to be a subclass of {BaseModel}, {dict}, {list}, {Union}, {NoneType}, {str} or {httpx.Response}."  # noqa: E501
+            )
+
+        # split is required to handle cases where additional information is included
+        # in the response, e.g. application/json; charset=utf-8
+        content_type, *_ = response.headers.get("content-type", "*").split(";")
+        if content_type != "application/json":
+            if is_basemodel(cast_type):
+                try:
+                    data = response.json()
+                except Exception as exc:
+                    log.debug("Could not read JSON from response data due to %s - %s", type(exc), exc)
+                else:
+                    return self._client._process_response_data(
+                        data=data,
+                        cast_type=cast_type,  # type: ignore
+                        response=response,
+                    )
+
+            if self._client._strict_response_validation:
+                raise APIResponseValidationError(
+                    response=response,
+                    message=f"Expected Content-Type response header to be `application/json` but received `{content_type}` instead.",  # noqa: E501
+                    json_data=response.text,
+                )
+
+            # If the API responds with content that isn't JSON then we just return
+            # the (decoded) text without performing any parsing so that you can still
+            # handle the response however you need to.
+            return response.text  # type: ignore
+
+        data = response.json()
+
+        return self._client._process_response_data(
+            data=data,
+            cast_type=cast_type,  # type: ignore
+            response=response,
+        )
+
+    @override
+    def __repr__(self) -> str:
+        return f"<APIResponse [{self.status_code} {self.http_response.reason_phrase}] type={self._cast_type}>"
+
+
+class MissingStreamClassError(TypeError):
+    def __init__(self) -> None:
+        super().__init__(
+            "The `stream` argument was set to `True` but the `stream_cls` argument was not given. See `openai._streaming` for reference",  # noqa: E501
+        )
+
+
+def to_raw_response_wrapper(func: Callable[P, R]) -> Callable[P, LegacyAPIResponse[R]]:
+    """Higher order function that takes one of our bound API methods and wraps it
+    to support returning the raw `APIResponse` object directly.
+    """
+
+    @functools.wraps(func)
+    def wrapped(*args: P.args, **kwargs: P.kwargs) -> LegacyAPIResponse[R]:
+        extra_headers: dict[str, str] = {**(cast(Any, kwargs.get("extra_headers")) or {})}
+        extra_headers[RAW_RESPONSE_HEADER] = "true"
+
+        kwargs["extra_headers"] = extra_headers
+
+        return cast(LegacyAPIResponse[R], func(*args, **kwargs))
+
+    return wrapped
diff --git a/api/core/model_runtime/model_providers/zhipuai/zhipuai_sdk/core/_request_opt.py b/api/core/model_runtime/model_providers/zhipuai/zhipuai_sdk/core/_request_opt.py
index ac459151fc3a4..c3b894b3a3d88 100644
--- a/api/core/model_runtime/model_providers/zhipuai/zhipuai_sdk/core/_request_opt.py
+++ b/api/core/model_runtime/model_providers/zhipuai/zhipuai_sdk/core/_request_opt.py
@@ -1,48 +1,97 @@
 from __future__ import annotations
 
-from typing import Any, ClassVar, Union
+from collections.abc import Callable
+from typing import TYPE_CHECKING, Any, ClassVar, Union, cast
 
+import pydantic.generics
 from httpx import Timeout
-from pydantic import ConfigDict
-from typing_extensions import TypedDict, Unpack
+from typing_extensions import Required, TypedDict, Unpack, final
 
-from ._base_type import Body, Headers, HttpxRequestFiles, NotGiven, Query
-from ._utils import remove_notgiven_indict
+from ._base_compat import PYDANTIC_V2, ConfigDict
+from ._base_type import AnyMapping, Body, Headers, HttpxRequestFiles, NotGiven, Query
+from ._constants import RAW_RESPONSE_HEADER
+from ._utils import is_given, strip_not_given
 
 
 class UserRequestInput(TypedDict, total=False):
+    headers: Headers
     max_retries: int
     timeout: float | Timeout | None
+    params: Query
+    extra_json: AnyMapping
+
+
+class FinalRequestOptionsInput(TypedDict, total=False):
+    method: Required[str]
+    url: Required[str]
+    params: Query
     headers: Headers
-    params: Query | None
+    max_retries: int
+    timeout: float | Timeout | None
+    files: HttpxRequestFiles | None
+    json_data: Body
+    extra_json: AnyMapping
 
 
-class ClientRequestParam:
+@final
+class FinalRequestOptions(pydantic.BaseModel):
     method: str
     url: str
-    max_retries: Union[int, NotGiven] = NotGiven()
-    timeout: Union[float, NotGiven] = NotGiven()
+    params: Query = {}
     headers: Union[Headers, NotGiven] = NotGiven()
-    json_data: Union[Body, None] = None
+    max_retries: Union[int, NotGiven] = NotGiven()
+    timeout: Union[float, Timeout, None, NotGiven] = NotGiven()
     files: Union[HttpxRequestFiles, None] = None
-    params: Query = {}
-    model_config: ClassVar[ConfigDict] = ConfigDict(arbitrary_types_allowed=True)
+    idempotency_key: Union[str, None] = None
+    post_parser: Union[Callable[[Any], Any], NotGiven] = NotGiven()
+
+    # It should be noted that we cannot use `json` here as that would override
+    # a BaseModel method in an incompatible fashion.
+    json_data: Union[Body, None] = None
+    extra_json: Union[AnyMapping, None] = None
+
+    if PYDANTIC_V2:
+        model_config: ClassVar[ConfigDict] = ConfigDict(arbitrary_types_allowed=True)
+    else:
 
-    def get_max_retries(self, max_retries) -> int:
+        class Config(pydantic.BaseConfig):  # pyright: ignore[reportDeprecated]
+            arbitrary_types_allowed: bool = True
+
+    def get_max_retries(self, max_retries: int) -> int:
         if isinstance(self.max_retries, NotGiven):
             return max_retries
         return self.max_retries
 
+    def _strip_raw_response_header(self) -> None:
+        if not is_given(self.headers):
+            return
+
+        if self.headers.get(RAW_RESPONSE_HEADER):
+            self.headers = {**self.headers}
+            self.headers.pop(RAW_RESPONSE_HEADER)
+
+    # override the `construct` method so that we can run custom transformations.
+    # this is necessary as we don't want to do any actual runtime type checking
+    # (which means we can't use validators) but we do want to ensure that `NotGiven`
+    # values are not present
+    #
+    # type ignore required because we're adding explicit types to `**values`
     @classmethod
     def construct(  # type: ignore
         cls,
         _fields_set: set[str] | None = None,
         **values: Unpack[UserRequestInput],
-    ) -> ClientRequestParam:
-        kwargs: dict[str, Any] = {key: remove_notgiven_indict(value) for key, value in values.items()}
-        client = cls()
-        client.__dict__.update(kwargs)
-
-        return client
+    ) -> FinalRequestOptions:
+        kwargs: dict[str, Any] = {
+            # we unconditionally call `strip_not_given` on any value
+            # as it will just ignore any non-mapping types
+            key: strip_not_given(value)
+            for key, value in values.items()
+        }
+        if PYDANTIC_V2:
+            return super().model_construct(_fields_set, **kwargs)
+        return cast(FinalRequestOptions, super().construct(_fields_set, **kwargs))  # pyright: ignore[reportDeprecated]
 
-    model_construct = construct
+    if not TYPE_CHECKING:
+        # type checkers incorrectly complain about this assignment
+        model_construct = construct
diff --git a/api/core/model_runtime/model_providers/zhipuai/zhipuai_sdk/core/_response.py b/api/core/model_runtime/model_providers/zhipuai/zhipuai_sdk/core/_response.py
index 56e60a793407c..45443da662d57 100644
--- a/api/core/model_runtime/model_providers/zhipuai/zhipuai_sdk/core/_response.py
+++ b/api/core/model_runtime/model_providers/zhipuai/zhipuai_sdk/core/_response.py
@@ -1,87 +1,193 @@
 from __future__ import annotations
 
 import datetime
-from typing import TYPE_CHECKING, Any, Generic, TypeVar, cast, get_args, get_origin
+import inspect
+import logging
+from collections.abc import Iterator
+from typing import TYPE_CHECKING, Any, Generic, TypeVar, Union, cast, get_origin, overload
 
 import httpx
 import pydantic
-from typing_extensions import ParamSpec
+from typing_extensions import ParamSpec, override
 
+from ._base_models import BaseModel, is_basemodel
 from ._base_type import NoneType
-from ._sse_client import StreamResponse
+from ._errors import APIResponseValidationError, ZhipuAIError
+from ._sse_client import StreamResponse, extract_stream_chunk_type, is_stream_class_type
+from ._utils import extract_type_arg, extract_type_var_from_base, is_annotated_type, is_given
 
 if TYPE_CHECKING:
     from ._http_client import HttpClient
+    from ._request_opt import FinalRequestOptions
 
 P = ParamSpec("P")
 R = TypeVar("R")
+_T = TypeVar("_T")
+_APIResponseT = TypeVar("_APIResponseT", bound="APIResponse[Any]")
+log: logging.Logger = logging.getLogger(__name__)
 
 
-class HttpResponse(Generic[R]):
+class BaseAPIResponse(Generic[R]):
     _cast_type: type[R]
     _client: HttpClient
-    _parsed: R | None
-    _enable_stream: bool
+    _parsed_by_type: dict[type[Any], Any]
+    _is_sse_stream: bool
     _stream_cls: type[StreamResponse[Any]]
+    _options: FinalRequestOptions
     http_response: httpx.Response
 
     def __init__(
         self,
         *,
-        raw_response: httpx.Response,
+        raw: httpx.Response,
         cast_type: type[R],
         client: HttpClient,
-        enable_stream: bool = False,
+        stream: bool,
         stream_cls: type[StreamResponse[Any]] | None = None,
+        options: FinalRequestOptions,
     ) -> None:
         self._cast_type = cast_type
         self._client = client
-        self._parsed = None
+        self._parsed_by_type = {}
+        self._is_sse_stream = stream
         self._stream_cls = stream_cls
-        self._enable_stream = enable_stream
-        self.http_response = raw_response
+        self._options = options
+        self.http_response = raw
 
-    def parse(self) -> R:
-        self._parsed = self._parse()
-        return self._parsed
+    def _parse(self, *, to: type[_T] | None = None) -> R | _T:
+        # unwrap `Annotated[T, ...]` -> `T`
+        if to and is_annotated_type(to):
+            to = extract_type_arg(to, 0)
 
-    def _parse(self) -> R:
-        if self._enable_stream:
-            self._parsed = cast(
+        if self._is_sse_stream:
+            if to:
+                if not is_stream_class_type(to):
+                    raise TypeError(f"Expected custom parse type to be a subclass of {StreamResponse}")
+
+                return cast(
+                    _T,
+                    to(
+                        cast_type=extract_stream_chunk_type(
+                            to,
+                            failure_message="Expected custom stream type to be passed with a type argument, e.g. StreamResponse[ChunkType]",  # noqa: E501
+                        ),
+                        response=self.http_response,
+                        client=cast(Any, self._client),
+                    ),
+                )
+
+            if self._stream_cls:
+                return cast(
+                    R,
+                    self._stream_cls(
+                        cast_type=extract_stream_chunk_type(self._stream_cls),
+                        response=self.http_response,
+                        client=cast(Any, self._client),
+                    ),
+                )
+
+            stream_cls = cast("type[Stream[Any]] | None", self._client._default_stream_cls)
+            if stream_cls is None:
+                raise MissingStreamClassError()
+
+            return cast(
                 R,
-                self._stream_cls(
-                    cast_type=cast(type, get_args(self._stream_cls)[0]),
+                stream_cls(
+                    cast_type=self._cast_type,
                     response=self.http_response,
-                    client=self._client,
+                    client=cast(Any, self._client),
                 ),
             )
-            return self._parsed
-        cast_type = self._cast_type
+
+        cast_type = to if to is not None else self._cast_type
+
+        # unwrap `Annotated[T, ...]` -> `T`
+        if is_annotated_type(cast_type):
+            cast_type = extract_type_arg(cast_type, 0)
+
         if cast_type is NoneType:
             return cast(R, None)
-        http_response = self.http_response
+
+        response = self.http_response
         if cast_type == str:
-            return cast(R, http_response.text)
+            return cast(R, response.text)
+
+        if cast_type == bytes:
+            return cast(R, response.content)
+
+        if cast_type == int:
+            return cast(R, int(response.text))
+
+        if cast_type == float:
+            return cast(R, float(response.text))
 
-        content_type, *_ = http_response.headers.get("content-type", "application/json").split(";")
         origin = get_origin(cast_type) or cast_type
+
+        # handle the legacy binary response case
+        if inspect.isclass(cast_type) and cast_type.__name__ == "HttpxBinaryResponseContent":
+            return cast(R, cast_type(response))  # type: ignore
+
+        if origin == APIResponse:
+            raise RuntimeError("Unexpected state - cast_type is `APIResponse`")
+
+        if inspect.isclass(origin) and issubclass(origin, httpx.Response):
+            # Because of the invariance of our ResponseT TypeVar, users can subclass httpx.Response
+            # and pass that class to our request functions. We cannot change the variance to be either
+            # covariant or contravariant as that makes our usage of ResponseT illegal. We could construct
+            # the response class ourselves but that is something that should be supported directly in httpx
+            # as it would be easy to incorrectly construct the Response object due to the multitude of arguments.
+            if cast_type != httpx.Response:
+                raise ValueError("Subclasses of httpx.Response cannot be passed to `cast_type`")
+            return cast(R, response)
+
+        if inspect.isclass(origin) and not issubclass(origin, BaseModel) and issubclass(origin, pydantic.BaseModel):
+            raise TypeError("Pydantic models must subclass our base model type, e.g. `from openai import BaseModel`")
+
+        if (
+            cast_type is not object
+            and origin is not list
+            and origin is not dict
+            and origin is not Union
+            and not issubclass(origin, BaseModel)
+        ):
+            raise RuntimeError(
+                f"Unsupported type, expected {cast_type} to be a subclass of {BaseModel}, {dict}, {list}, {Union}, {NoneType}, {str} or {httpx.Response}."  # noqa: E501
+            )
+
+        # split is required to handle cases where additional information is included
+        # in the response, e.g. application/json; charset=utf-8
+        content_type, *_ = response.headers.get("content-type", "*").split(";")
         if content_type != "application/json":
-            if issubclass(origin, pydantic.BaseModel):
-                data = http_response.json()
-                return self._client._process_response_data(
-                    data=data,
-                    cast_type=cast_type,  # type: ignore
-                    response=http_response,
+            if is_basemodel(cast_type):
+                try:
+                    data = response.json()
+                except Exception as exc:
+                    log.debug("Could not read JSON from response data due to %s - %s", type(exc), exc)
+                else:
+                    return self._client._process_response_data(
+                        data=data,
+                        cast_type=cast_type,  # type: ignore
+                        response=response,
+                    )
+
+            if self._client._strict_response_validation:
+                raise APIResponseValidationError(
+                    response=response,
+                    message=f"Expected Content-Type response header to be `application/json` but received `{content_type}` instead.",  # noqa: E501
+                    json_data=response.text,
                 )
 
-            return http_response.text
+            # If the API responds with content that isn't JSON then we just return
+            # the (decoded) text without performing any parsing so that you can still
+            # handle the response however you need to.
+            return response.text  # type: ignore
 
-        data = http_response.json()
+        data = response.json()
 
         return self._client._process_response_data(
             data=data,
             cast_type=cast_type,  # type: ignore
-            response=http_response,
+            response=response,
         )
 
     @property
@@ -90,6 +196,7 @@ def headers(self) -> httpx.Headers:
 
     @property
     def http_request(self) -> httpx.Request:
+        """Returns the httpx Request instance associated with the current response."""
         return self.http_response.request
 
     @property
@@ -98,24 +205,194 @@ def status_code(self) -> int:
 
     @property
     def url(self) -> httpx.URL:
+        """Returns the URL for which the request was made."""
         return self.http_response.url
 
     @property
     def method(self) -> str:
         return self.http_request.method
 
-    @property
-    def content(self) -> bytes:
-        return self.http_response.content
-
-    @property
-    def text(self) -> str:
-        return self.http_response.text
-
     @property
     def http_version(self) -> str:
         return self.http_response.http_version
 
     @property
     def elapsed(self) -> datetime.timedelta:
+        """The time taken for the complete request/response cycle to complete."""
         return self.http_response.elapsed
+
+    @property
+    def is_closed(self) -> bool:
+        """Whether or not the response body has been closed.
+
+        If this is False then there is response data that has not been read yet.
+        You must either fully consume the response body or call `.close()`
+        before discarding the response to prevent resource leaks.
+        """
+        return self.http_response.is_closed
+
+    @override
+    def __repr__(self) -> str:
+        return f"<{self.__class__.__name__} [{self.status_code} {self.http_response.reason_phrase}] type={self._cast_type}>"  # noqa: E501
+
+
+class APIResponse(BaseAPIResponse[R]):
+    @property
+    def request_id(self) -> str | None:
+        return self.http_response.headers.get("x-request-id")  # type: ignore[no-any-return]
+
+    @overload
+    def parse(self, *, to: type[_T]) -> _T: ...
+
+    @overload
+    def parse(self) -> R: ...
+
+    def parse(self, *, to: type[_T] | None = None) -> R | _T:
+        """Returns the rich python representation of this response's data.
+
+        For lower-level control, see `.read()`, `.json()`, `.iter_bytes()`.
+
+        You can customise the type that the response is parsed into through
+        the `to` argument, e.g.
+
+        ```py
+        from openai import BaseModel
+
+
+        class MyModel(BaseModel):
+            foo: str
+
+
+        obj = response.parse(to=MyModel)
+        print(obj.foo)
+        ```
+
+        We support parsing:
+          - `BaseModel`
+          - `dict`
+          - `list`
+          - `Union`
+          - `str`
+          - `int`
+          - `float`
+          - `httpx.Response`
+        """
+        cache_key = to if to is not None else self._cast_type
+        cached = self._parsed_by_type.get(cache_key)
+        if cached is not None:
+            return cached  # type: ignore[no-any-return]
+
+        if not self._is_sse_stream:
+            self.read()
+
+        parsed = self._parse(to=to)
+        if is_given(self._options.post_parser):
+            parsed = self._options.post_parser(parsed)
+
+        self._parsed_by_type[cache_key] = parsed
+        return parsed
+
+    def read(self) -> bytes:
+        """Read and return the binary response content."""
+        try:
+            return self.http_response.read()
+        except httpx.StreamConsumed as exc:
+            # The default error raised by httpx isn't very
+            # helpful in our case so we re-raise it with
+            # a different error message.
+            raise StreamAlreadyConsumed() from exc
+
+    def text(self) -> str:
+        """Read and decode the response content into a string."""
+        self.read()
+        return self.http_response.text
+
+    def json(self) -> object:
+        """Read and decode the JSON response content."""
+        self.read()
+        return self.http_response.json()
+
+    def close(self) -> None:
+        """Close the response and release the connection.
+
+        Automatically called if the response body is read to completion.
+        """
+        self.http_response.close()
+
+    def iter_bytes(self, chunk_size: int | None = None) -> Iterator[bytes]:
+        """
+        A byte-iterator over the decoded response content.
+
+        This automatically handles gzip, deflate and brotli encoded responses.
+        """
+        yield from self.http_response.iter_bytes(chunk_size)
+
+    def iter_text(self, chunk_size: int | None = None) -> Iterator[str]:
+        """A str-iterator over the decoded response content
+        that handles both gzip, deflate, etc but also detects the content's
+        string encoding.
+        """
+        yield from self.http_response.iter_text(chunk_size)
+
+    def iter_lines(self) -> Iterator[str]:
+        """Like `iter_text()` but will only yield chunks for each line"""
+        yield from self.http_response.iter_lines()
+
+
+class MissingStreamClassError(TypeError):
+    def __init__(self) -> None:
+        super().__init__(
+            "The `stream` argument was set to `True` but the `stream_cls` argument was not given. See `openai._streaming` for reference",  # noqa: E501
+        )
+
+
+class StreamAlreadyConsumed(ZhipuAIError):  # noqa: N818
+    """
+    Attempted to read or stream content, but the content has already
+    been streamed.
+
+    This can happen if you use a method like `.iter_lines()` and then attempt
+    to read th entire response body afterwards, e.g.
+
+    ```py
+    response = await client.post(...)
+    async for line in response.iter_lines():
+        ...  # do something with `line`
+
+    content = await response.read()
+    # ^ error
+    ```
+
+    If you want this behaviour you'll need to either manually accumulate the response
+    content or call `await response.read()` before iterating over the stream.
+    """
+
+    def __init__(self) -> None:
+        message = (
+            "Attempted to read or stream some content, but the content has "
+            "already been streamed. "
+            "This could be due to attempting to stream the response "
+            "content more than once."
+            "\n\n"
+            "You can fix this by manually accumulating the response content while streaming "
+            "or by calling `.read()` before starting to stream."
+        )
+        super().__init__(message)
+
+
+def extract_response_type(typ: type[BaseAPIResponse[Any]]) -> type:
+    """Given a type like `APIResponse[T]`, returns the generic type variable `T`.
+
+    This also handles the case where a concrete subclass is given, e.g.
+    ```py
+    class MyResponse(APIResponse[bytes]):
+        ...
+
+    extract_response_type(MyResponse) -> bytes
+    ```
+    """
+    return extract_type_var_from_base(
+        typ,
+        generic_bases=cast("tuple[type, ...]", (BaseAPIResponse, APIResponse)),
+        index=0,
+    )
diff --git a/api/core/model_runtime/model_providers/zhipuai/zhipuai_sdk/core/_sse_client.py b/api/core/model_runtime/model_providers/zhipuai/zhipuai_sdk/core/_sse_client.py
index ec2745d05912d..cbc449d24421d 100644
--- a/api/core/model_runtime/model_providers/zhipuai/zhipuai_sdk/core/_sse_client.py
+++ b/api/core/model_runtime/model_providers/zhipuai/zhipuai_sdk/core/_sse_client.py
@@ -1,13 +1,16 @@
 from __future__ import annotations
 
+import inspect
 import json
 from collections.abc import Iterator, Mapping
-from typing import TYPE_CHECKING, Generic
+from typing import TYPE_CHECKING, Generic, TypeGuard, cast
 
 import httpx
 
+from . import get_origin
 from ._base_type import ResponseT
 from ._errors import APIResponseError
+from ._utils import extract_type_var_from_base, is_mapping
 
 _FIELD_SEPARATOR = ":"
 
@@ -53,8 +56,41 @@ def __stream__(self) -> Iterator[ResponseT]:
                         request=self.response.request,
                         json_data=data["error"],
                     )
+            if sse.event is None:
+                data = sse.json_data()
+                if is_mapping(data) and data.get("error"):
+                    message = None
+                    error = data.get("error")
+                    if is_mapping(error):
+                        message = error.get("message")
+                    if not message or not isinstance(message, str):
+                        message = "An error occurred during streaming"
 
+                    raise APIResponseError(
+                        message=message,
+                        request=self.response.request,
+                        json_data=data["error"],
+                    )
                 yield self._data_process_func(data=data, cast_type=self._cast_type, response=self.response)
+
+            else:
+                data = sse.json_data()
+
+                if sse.event == "error" and is_mapping(data) and data.get("error"):
+                    message = None
+                    error = data.get("error")
+                    if is_mapping(error):
+                        message = error.get("message")
+                    if not message or not isinstance(message, str):
+                        message = "An error occurred during streaming"
+
+                    raise APIResponseError(
+                        message=message,
+                        request=self.response.request,
+                        json_data=data["error"],
+                    )
+                yield self._data_process_func(data=data, cast_type=self._cast_type, response=self.response)
+
         for sse in iterator:
             pass
 
@@ -138,3 +174,33 @@ def decode_line(self, line: str):
             except (TypeError, ValueError):
                 pass
         return
+
+
+def is_stream_class_type(typ: type) -> TypeGuard[type[StreamResponse[object]]]:
+    """TypeGuard for determining whether or not the given type is a subclass of `Stream` / `AsyncStream`"""
+    origin = get_origin(typ) or typ
+    return inspect.isclass(origin) and issubclass(origin, StreamResponse)
+
+
+def extract_stream_chunk_type(
+    stream_cls: type,
+    *,
+    failure_message: str | None = None,
+) -> type:
+    """Given a type like `StreamResponse[T]`, returns the generic type variable `T`.
+
+    This also handles the case where a concrete subclass is given, e.g.
+    ```py
+    class MyStream(StreamResponse[bytes]):
+        ...
+
+    extract_stream_chunk_type(MyStream) -> bytes
+    ```
+    """
+
+    return extract_type_var_from_base(
+        stream_cls,
+        index=0,
+        generic_bases=cast("tuple[type, ...]", (StreamResponse,)),
+        failure_message=failure_message,
+    )
diff --git a/api/core/model_runtime/model_providers/zhipuai/zhipuai_sdk/core/_utils.py b/api/core/model_runtime/model_providers/zhipuai/zhipuai_sdk/core/_utils.py
deleted file mode 100644
index 6b610567daa09..0000000000000
--- a/api/core/model_runtime/model_providers/zhipuai/zhipuai_sdk/core/_utils.py
+++ /dev/null
@@ -1,19 +0,0 @@
-from __future__ import annotations
-
-from collections.abc import Iterable, Mapping
-from typing import TypeVar
-
-from ._base_type import NotGiven
-
-
-def remove_notgiven_indict(obj):
-    if obj is None or (not isinstance(obj, Mapping)):
-        return obj
-    return {key: value for key, value in obj.items() if not isinstance(value, NotGiven)}
-
-
-_T = TypeVar("_T")
-
-
-def flatten(t: Iterable[Iterable[_T]]) -> list[_T]:
-    return [item for sublist in t for item in sublist]
diff --git a/api/core/model_runtime/model_providers/zhipuai/zhipuai_sdk/core/_utils/__init__.py b/api/core/model_runtime/model_providers/zhipuai/zhipuai_sdk/core/_utils/__init__.py
new file mode 100644
index 0000000000000..a66b095816b8b
--- /dev/null
+++ b/api/core/model_runtime/model_providers/zhipuai/zhipuai_sdk/core/_utils/__init__.py
@@ -0,0 +1,52 @@
+from ._utils import (  # noqa: I001
+    remove_notgiven_indict as remove_notgiven_indict,  # noqa: PLC0414
+    flatten as flatten,  # noqa: PLC0414
+    is_dict as is_dict,  # noqa: PLC0414
+    is_list as is_list,  # noqa: PLC0414
+    is_given as is_given,  # noqa: PLC0414
+    is_tuple as is_tuple,  # noqa: PLC0414
+    is_mapping as is_mapping,  # noqa: PLC0414
+    is_tuple_t as is_tuple_t,  # noqa: PLC0414
+    parse_date as parse_date,  # noqa: PLC0414
+    is_iterable as is_iterable,  # noqa: PLC0414
+    is_sequence as is_sequence,  # noqa: PLC0414
+    coerce_float as coerce_float,  # noqa: PLC0414
+    is_mapping_t as is_mapping_t,  # noqa: PLC0414
+    removeprefix as removeprefix,  # noqa: PLC0414
+    removesuffix as removesuffix,  # noqa: PLC0414
+    extract_files as extract_files,  # noqa: PLC0414
+    is_sequence_t as is_sequence_t,  # noqa: PLC0414
+    required_args as required_args,  # noqa: PLC0414
+    coerce_boolean as coerce_boolean,  # noqa: PLC0414
+    coerce_integer as coerce_integer,  # noqa: PLC0414
+    file_from_path as file_from_path,  # noqa: PLC0414
+    parse_datetime as parse_datetime,  # noqa: PLC0414
+    strip_not_given as strip_not_given,  # noqa: PLC0414
+    deepcopy_minimal as deepcopy_minimal,  # noqa: PLC0414
+    get_async_library as get_async_library,  # noqa: PLC0414
+    maybe_coerce_float as maybe_coerce_float,  # noqa: PLC0414
+    get_required_header as get_required_header,  # noqa: PLC0414
+    maybe_coerce_boolean as maybe_coerce_boolean,  # noqa: PLC0414
+    maybe_coerce_integer as maybe_coerce_integer,  # noqa: PLC0414
+    drop_prefix_image_data as drop_prefix_image_data,  # noqa: PLC0414
+)
+
+
+from ._typing import (
+    is_list_type as is_list_type,  # noqa: PLC0414
+    is_union_type as is_union_type,  # noqa: PLC0414
+    extract_type_arg as extract_type_arg,  # noqa: PLC0414
+    is_iterable_type as is_iterable_type,  # noqa: PLC0414
+    is_required_type as is_required_type,  # noqa: PLC0414
+    is_annotated_type as is_annotated_type,  # noqa: PLC0414
+    strip_annotated_type as strip_annotated_type,  # noqa: PLC0414
+    extract_type_var_from_base as extract_type_var_from_base,  # noqa: PLC0414
+)
+
+from ._transform import (
+    PropertyInfo as PropertyInfo,  # noqa: PLC0414
+    transform as transform,  # noqa: PLC0414
+    async_transform as async_transform,  # noqa: PLC0414
+    maybe_transform as maybe_transform,  # noqa: PLC0414
+    async_maybe_transform as async_maybe_transform,  # noqa: PLC0414
+)
diff --git a/api/core/model_runtime/model_providers/zhipuai/zhipuai_sdk/core/_utils/_transform.py b/api/core/model_runtime/model_providers/zhipuai/zhipuai_sdk/core/_utils/_transform.py
new file mode 100644
index 0000000000000..e8ef1f79358a9
--- /dev/null
+++ b/api/core/model_runtime/model_providers/zhipuai/zhipuai_sdk/core/_utils/_transform.py
@@ -0,0 +1,383 @@
+from __future__ import annotations
+
+import base64
+import io
+import pathlib
+from collections.abc import Mapping
+from datetime import date, datetime
+from typing import Any, Literal, TypeVar, cast, get_args, get_type_hints
+
+import anyio
+import pydantic
+from typing_extensions import override
+
+from .._base_compat import is_typeddict, model_dump
+from .._files import is_base64_file_input
+from ._typing import (
+    extract_type_arg,
+    is_annotated_type,
+    is_iterable_type,
+    is_list_type,
+    is_required_type,
+    is_union_type,
+    strip_annotated_type,
+)
+from ._utils import (
+    is_iterable,
+    is_list,
+    is_mapping,
+)
+
+_T = TypeVar("_T")
+
+
+# TODO: support for drilling globals() and locals()
+# TODO: ensure works correctly with forward references in all cases
+
+
+PropertyFormat = Literal["iso8601", "base64", "custom"]
+
+
+class PropertyInfo:
+    """Metadata class to be used in Annotated types to provide information about a given type.
+
+    For example:
+
+    class MyParams(TypedDict):
+        account_holder_name: Annotated[str, PropertyInfo(alias='accountHolderName')]
+
+    This means that {'account_holder_name': 'Robert'} will be transformed to {'accountHolderName': 'Robert'} before being sent to the API.
+    """  # noqa: E501
+
+    alias: str | None
+    format: PropertyFormat | None
+    format_template: str | None
+    discriminator: str | None
+
+    def __init__(
+        self,
+        *,
+        alias: str | None = None,
+        format: PropertyFormat | None = None,
+        format_template: str | None = None,
+        discriminator: str | None = None,
+    ) -> None:
+        self.alias = alias
+        self.format = format
+        self.format_template = format_template
+        self.discriminator = discriminator
+
+    @override
+    def __repr__(self) -> str:
+        return f"{self.__class__.__name__}(alias='{self.alias}', format={self.format}, format_template='{self.format_template}', discriminator='{self.discriminator}')"  # noqa: E501
+
+
+def maybe_transform(
+    data: object,
+    expected_type: object,
+) -> Any | None:
+    """Wrapper over `transform()` that allows `None` to be passed.
+
+    See `transform()` for more details.
+    """
+    if data is None:
+        return None
+    return transform(data, expected_type)
+
+
+# Wrapper over _transform_recursive providing fake types
+def transform(
+    data: _T,
+    expected_type: object,
+) -> _T:
+    """Transform dictionaries based off of type information from the given type, for example:
+
+    ```py
+    class Params(TypedDict, total=False):
+        card_id: Required[Annotated[str, PropertyInfo(alias="cardID")]]
+
+
+    transformed = transform({"card_id": "<my card ID>"}, Params)
+    # {'cardID': '<my card ID>'}
+    ```
+
+    Any keys / data that does not have type information given will be included as is.
+
+    It should be noted that the transformations that this function does are not represented in the type system.
+    """
+    transformed = _transform_recursive(data, annotation=cast(type, expected_type))
+    return cast(_T, transformed)
+
+
+def _get_annotated_type(type_: type) -> type | None:
+    """If the given type is an `Annotated` type then it is returned, if not `None` is returned.
+
+    This also unwraps the type when applicable, e.g. `Required[Annotated[T, ...]]`
+    """
+    if is_required_type(type_):
+        # Unwrap `Required[Annotated[T, ...]]` to `Annotated[T, ...]`
+        type_ = get_args(type_)[0]
+
+    if is_annotated_type(type_):
+        return type_
+
+    return None
+
+
+def _maybe_transform_key(key: str, type_: type) -> str:
+    """Transform the given `data` based on the annotations provided in `type_`.
+
+    Note: this function only looks at `Annotated` types that contain `PropertInfo` metadata.
+    """
+    annotated_type = _get_annotated_type(type_)
+    if annotated_type is None:
+        # no `Annotated` definition for this type, no transformation needed
+        return key
+
+    # ignore the first argument as it is the actual type
+    annotations = get_args(annotated_type)[1:]
+    for annotation in annotations:
+        if isinstance(annotation, PropertyInfo) and annotation.alias is not None:
+            return annotation.alias
+
+    return key
+
+
+def _transform_recursive(
+    data: object,
+    *,
+    annotation: type,
+    inner_type: type | None = None,
+) -> object:
+    """Transform the given data against the expected type.
+
+    Args:
+        annotation: The direct type annotation given to the particular piece of data.
+            This may or may not be wrapped in metadata types, e.g. `Required[T]`, `Annotated[T, ...]` etc
+
+        inner_type: If applicable, this is the "inside" type. This is useful in certain cases where the outside type
+            is a container type such as `List[T]`. In that case `inner_type` should be set to `T` so that each entry in
+            the list can be transformed using the metadata from the container type.
+
+            Defaults to the same value as the `annotation` argument.
+    """
+    if inner_type is None:
+        inner_type = annotation
+
+    stripped_type = strip_annotated_type(inner_type)
+    if is_typeddict(stripped_type) and is_mapping(data):
+        return _transform_typeddict(data, stripped_type)
+
+    if (
+        # List[T]
+        (is_list_type(stripped_type) and is_list(data))
+        # Iterable[T]
+        or (is_iterable_type(stripped_type) and is_iterable(data) and not isinstance(data, str))
+    ):
+        inner_type = extract_type_arg(stripped_type, 0)
+        return [_transform_recursive(d, annotation=annotation, inner_type=inner_type) for d in data]
+
+    if is_union_type(stripped_type):
+        # For union types we run the transformation against all subtypes to ensure that everything is transformed.
+        #
+        # TODO: there may be edge cases where the same normalized field name will transform to two different names
+        # in different subtypes.
+        for subtype in get_args(stripped_type):
+            data = _transform_recursive(data, annotation=annotation, inner_type=subtype)
+        return data
+
+    if isinstance(data, pydantic.BaseModel):
+        return model_dump(data, exclude_unset=True)
+
+    annotated_type = _get_annotated_type(annotation)
+    if annotated_type is None:
+        return data
+
+    # ignore the first argument as it is the actual type
+    annotations = get_args(annotated_type)[1:]
+    for annotation in annotations:
+        if isinstance(annotation, PropertyInfo) and annotation.format is not None:
+            return _format_data(data, annotation.format, annotation.format_template)
+
+    return data
+
+
+def _format_data(data: object, format_: PropertyFormat, format_template: str | None) -> object:
+    if isinstance(data, date | datetime):
+        if format_ == "iso8601":
+            return data.isoformat()
+
+        if format_ == "custom" and format_template is not None:
+            return data.strftime(format_template)
+
+    if format_ == "base64" and is_base64_file_input(data):
+        binary: str | bytes | None = None
+
+        if isinstance(data, pathlib.Path):
+            binary = data.read_bytes()
+        elif isinstance(data, io.IOBase):
+            binary = data.read()
+
+            if isinstance(binary, str):  # type: ignore[unreachable]
+                binary = binary.encode()
+
+        if not isinstance(binary, bytes):
+            raise RuntimeError(f"Could not read bytes from {data}; Received {type(binary)}")
+
+        return base64.b64encode(binary).decode("ascii")
+
+    return data
+
+
+def _transform_typeddict(
+    data: Mapping[str, object],
+    expected_type: type,
+) -> Mapping[str, object]:
+    result: dict[str, object] = {}
+    annotations = get_type_hints(expected_type, include_extras=True)
+    for key, value in data.items():
+        type_ = annotations.get(key)
+        if type_ is None:
+            # we do not have a type annotation for this field, leave it as is
+            result[key] = value
+        else:
+            result[_maybe_transform_key(key, type_)] = _transform_recursive(value, annotation=type_)
+    return result
+
+
+async def async_maybe_transform(
+    data: object,
+    expected_type: object,
+) -> Any | None:
+    """Wrapper over `async_transform()` that allows `None` to be passed.
+
+    See `async_transform()` for more details.
+    """
+    if data is None:
+        return None
+    return await async_transform(data, expected_type)
+
+
+async def async_transform(
+    data: _T,
+    expected_type: object,
+) -> _T:
+    """Transform dictionaries based off of type information from the given type, for example:
+
+    ```py
+    class Params(TypedDict, total=False):
+        card_id: Required[Annotated[str, PropertyInfo(alias="cardID")]]
+
+
+    transformed = transform({"card_id": "<my card ID>"}, Params)
+    # {'cardID': '<my card ID>'}
+    ```
+
+    Any keys / data that does not have type information given will be included as is.
+
+    It should be noted that the transformations that this function does are not represented in the type system.
+    """
+    transformed = await _async_transform_recursive(data, annotation=cast(type, expected_type))
+    return cast(_T, transformed)
+
+
+async def _async_transform_recursive(
+    data: object,
+    *,
+    annotation: type,
+    inner_type: type | None = None,
+) -> object:
+    """Transform the given data against the expected type.
+
+    Args:
+        annotation: The direct type annotation given to the particular piece of data.
+            This may or may not be wrapped in metadata types, e.g. `Required[T]`, `Annotated[T, ...]` etc
+
+        inner_type: If applicable, this is the "inside" type. This is useful in certain cases where the outside type
+            is a container type such as `List[T]`. In that case `inner_type` should be set to `T` so that each entry in
+            the list can be transformed using the metadata from the container type.
+
+            Defaults to the same value as the `annotation` argument.
+    """
+    if inner_type is None:
+        inner_type = annotation
+
+    stripped_type = strip_annotated_type(inner_type)
+    if is_typeddict(stripped_type) and is_mapping(data):
+        return await _async_transform_typeddict(data, stripped_type)
+
+    if (
+        # List[T]
+        (is_list_type(stripped_type) and is_list(data))
+        # Iterable[T]
+        or (is_iterable_type(stripped_type) and is_iterable(data) and not isinstance(data, str))
+    ):
+        inner_type = extract_type_arg(stripped_type, 0)
+        return [await _async_transform_recursive(d, annotation=annotation, inner_type=inner_type) for d in data]
+
+    if is_union_type(stripped_type):
+        # For union types we run the transformation against all subtypes to ensure that everything is transformed.
+        #
+        # TODO: there may be edge cases where the same normalized field name will transform to two different names
+        # in different subtypes.
+        for subtype in get_args(stripped_type):
+            data = await _async_transform_recursive(data, annotation=annotation, inner_type=subtype)
+        return data
+
+    if isinstance(data, pydantic.BaseModel):
+        return model_dump(data, exclude_unset=True)
+
+    annotated_type = _get_annotated_type(annotation)
+    if annotated_type is None:
+        return data
+
+    # ignore the first argument as it is the actual type
+    annotations = get_args(annotated_type)[1:]
+    for annotation in annotations:
+        if isinstance(annotation, PropertyInfo) and annotation.format is not None:
+            return await _async_format_data(data, annotation.format, annotation.format_template)
+
+    return data
+
+
+async def _async_format_data(data: object, format_: PropertyFormat, format_template: str | None) -> object:
+    if isinstance(data, date | datetime):
+        if format_ == "iso8601":
+            return data.isoformat()
+
+        if format_ == "custom" and format_template is not None:
+            return data.strftime(format_template)
+
+    if format_ == "base64" and is_base64_file_input(data):
+        binary: str | bytes | None = None
+
+        if isinstance(data, pathlib.Path):
+            binary = await anyio.Path(data).read_bytes()
+        elif isinstance(data, io.IOBase):
+            binary = data.read()
+
+            if isinstance(binary, str):  # type: ignore[unreachable]
+                binary = binary.encode()
+
+        if not isinstance(binary, bytes):
+            raise RuntimeError(f"Could not read bytes from {data}; Received {type(binary)}")
+
+        return base64.b64encode(binary).decode("ascii")
+
+    return data
+
+
+async def _async_transform_typeddict(
+    data: Mapping[str, object],
+    expected_type: type,
+) -> Mapping[str, object]:
+    result: dict[str, object] = {}
+    annotations = get_type_hints(expected_type, include_extras=True)
+    for key, value in data.items():
+        type_ = annotations.get(key)
+        if type_ is None:
+            # we do not have a type annotation for this field, leave it as is
+            result[key] = value
+        else:
+            result[_maybe_transform_key(key, type_)] = await _async_transform_recursive(value, annotation=type_)
+    return result
diff --git a/api/core/model_runtime/model_providers/zhipuai/zhipuai_sdk/core/_utils/_typing.py b/api/core/model_runtime/model_providers/zhipuai/zhipuai_sdk/core/_utils/_typing.py
new file mode 100644
index 0000000000000..c7c54dcc37458
--- /dev/null
+++ b/api/core/model_runtime/model_providers/zhipuai/zhipuai_sdk/core/_utils/_typing.py
@@ -0,0 +1,122 @@
+from __future__ import annotations
+
+from collections import abc as _c_abc
+from collections.abc import Iterable
+from typing import Annotated, Any, TypeVar, cast, get_args, get_origin
+
+from typing_extensions import Required
+
+from .._base_compat import is_union as _is_union
+from .._base_type import InheritsGeneric
+
+
+def is_annotated_type(typ: type) -> bool:
+    return get_origin(typ) == Annotated
+
+
+def is_list_type(typ: type) -> bool:
+    return (get_origin(typ) or typ) == list
+
+
+def is_iterable_type(typ: type) -> bool:
+    """If the given type is `typing.Iterable[T]`"""
+    origin = get_origin(typ) or typ
+    return origin in {Iterable, _c_abc.Iterable}
+
+
+def is_union_type(typ: type) -> bool:
+    return _is_union(get_origin(typ))
+
+
+def is_required_type(typ: type) -> bool:
+    return get_origin(typ) == Required
+
+
+def is_typevar(typ: type) -> bool:
+    # type ignore is required because type checkers
+    # think this expression will always return False
+    return type(typ) == TypeVar  # type: ignore
+
+
+# Extracts T from Annotated[T, ...] or from Required[Annotated[T, ...]]
+def strip_annotated_type(typ: type) -> type:
+    if is_required_type(typ) or is_annotated_type(typ):
+        return strip_annotated_type(cast(type, get_args(typ)[0]))
+
+    return typ
+
+
+def extract_type_arg(typ: type, index: int) -> type:
+    args = get_args(typ)
+    try:
+        return cast(type, args[index])
+    except IndexError as err:
+        raise RuntimeError(f"Expected type {typ} to have a type argument at index {index} but it did not") from err
+
+
+def extract_type_var_from_base(
+    typ: type,
+    *,
+    generic_bases: tuple[type, ...],
+    index: int,
+    failure_message: str | None = None,
+) -> type:
+    """Given a type like `Foo[T]`, returns the generic type variable `T`.
+
+    This also handles the case where a concrete subclass is given, e.g.
+    ```py
+    class MyResponse(Foo[bytes]):
+        ...
+
+    extract_type_var(MyResponse, bases=(Foo,), index=0) -> bytes
+    ```
+
+    And where a generic subclass is given:
+    ```py
+    _T = TypeVar('_T')
+    class MyResponse(Foo[_T]):
+        ...
+
+    extract_type_var(MyResponse[bytes], bases=(Foo,), index=0) -> bytes
+    ```
+    """
+    cls = cast(object, get_origin(typ) or typ)
+    if cls in generic_bases:
+        # we're given the class directly
+        return extract_type_arg(typ, index)
+
+    # if a subclass is given
+    # ---
+    # this is needed as __orig_bases__ is not present in the typeshed stubs
+    # because it is intended to be for internal use only, however there does
+    # not seem to be a way to resolve generic TypeVars for inherited subclasses
+    # without using it.
+    if isinstance(cls, InheritsGeneric):
+        target_base_class: Any | None = None
+        for base in cls.__orig_bases__:
+            if base.__origin__ in generic_bases:
+                target_base_class = base
+                break
+
+        if target_base_class is None:
+            raise RuntimeError(
+                "Could not find the generic base class;\n"
+                "This should never happen;\n"
+                f"Does {cls} inherit from one of {generic_bases} ?"
+            )
+
+        extracted = extract_type_arg(target_base_class, index)
+        if is_typevar(extracted):
+            # If the extracted type argument is itself a type variable
+            # then that means the subclass itself is generic, so we have
+            # to resolve the type argument from the class itself, not
+            # the base class.
+            #
+            # Note: if there is more than 1 type argument, the subclass could
+            # change the ordering of the type arguments, this is not currently
+            # supported.
+            return extract_type_arg(typ, index)
+
+        return extracted
+
+    raise RuntimeError(failure_message or f"Could not resolve inner type variable at index {index} for {typ}")
diff --git a/api/core/model_runtime/model_providers/zhipuai/zhipuai_sdk/core/_utils/_utils.py b/api/core/model_runtime/model_providers/zhipuai/zhipuai_sdk/core/_utils/_utils.py
new file mode 100644
index 0000000000000..ce5e7786aa293
--- /dev/null
+++ b/api/core/model_runtime/model_providers/zhipuai/zhipuai_sdk/core/_utils/_utils.py
@@ -0,0 +1,409 @@
+from __future__ import annotations
+
+import functools
+import inspect
+import os
+import re
+from collections.abc import Callable, Iterable, Mapping, Sequence
+from pathlib import Path
+from typing import (
+    Any,
+    TypeGuard,
+    TypeVar,
+    Union,
+    cast,
+    overload,
+)
+
+import sniffio
+
+from .._base_compat import parse_date as parse_date  # noqa: PLC0414
+from .._base_compat import parse_datetime as parse_datetime  # noqa: PLC0414
+from .._base_type import FileTypes, Headers, HeadersLike, NotGiven, NotGivenOr
+
+
+def remove_notgiven_indict(obj):
+    if obj is None or (not isinstance(obj, Mapping)):
+        return obj
+    return {key: value for key, value in obj.items() if not isinstance(value, NotGiven)}
+
+
+_T = TypeVar("_T")
+_TupleT = TypeVar("_TupleT", bound=tuple[object, ...])
+_MappingT = TypeVar("_MappingT", bound=Mapping[str, object])
+_SequenceT = TypeVar("_SequenceT", bound=Sequence[object])
+CallableT = TypeVar("CallableT", bound=Callable[..., Any])
+
+
+def flatten(t: Iterable[Iterable[_T]]) -> list[_T]:
+    return [item for sublist in t for item in sublist]
+
+
+def extract_files(
+    # TODO: this needs to take Dict but variance issues.....
+    # create protocol type ?
+    query: Mapping[str, object],
+    *,
+    paths: Sequence[Sequence[str]],
+) -> list[tuple[str, FileTypes]]:
+    """Recursively extract files from the given dictionary based on specified paths.
+
+    A path may look like this ['foo', 'files', '<array>', 'data'].
+
+    Note: this mutates the given dictionary.
+    """
+    files: list[tuple[str, FileTypes]] = []
+    for path in paths:
+        files.extend(_extract_items(query, path, index=0, flattened_key=None))
+    return files
+
+
+def _extract_items(
+    obj: object,
+    path: Sequence[str],
+    *,
+    index: int,
+    flattened_key: str | None,
+) -> list[tuple[str, FileTypes]]:
+    try:
+        key = path[index]
+    except IndexError:
+        if isinstance(obj, NotGiven):
+            # no value was provided - we can safely ignore
+            return []
+
+        # cyclical import
+        from .._files import assert_is_file_content
+
+        # We have exhausted the path, return the entry we found.
+        assert_is_file_content(obj, key=flattened_key)
+        assert flattened_key is not None
+        return [(flattened_key, cast(FileTypes, obj))]
+
+    index += 1
+    if is_dict(obj):
+        try:
+            # We are at the last entry in the path so we must remove the field
+            if (len(path)) == index:
+                item = obj.pop(key)
+            else:
+                item = obj[key]
+        except KeyError:
+            # Key was not present in the dictionary, this is not indicative of an error
+            # as the given path may not point to a required field. We also do not want
+            # to enforce required fields as the API may differ from the spec in some cases.
+            return []
+        if flattened_key is None:
+            flattened_key = key
+        else:
+            flattened_key += f"[{key}]"
+        return _extract_items(
+            item,
+            path,
+            index=index,
+            flattened_key=flattened_key,
+        )
+    elif is_list(obj):
+        if key != "<array>":
+            return []
+
+        return flatten(
+            [
+                _extract_items(
+                    item,
+                    path,
+                    index=index,
+                    flattened_key=flattened_key + "[]" if flattened_key is not None else "[]",
+                )
+                for item in obj
+            ]
+        )
+
+    # Something unexpected was passed, just ignore it.
+    return []
+
+
+def is_given(obj: NotGivenOr[_T]) -> TypeGuard[_T]:
+    return not isinstance(obj, NotGiven)
+
+
+# Type safe methods for narrowing types with TypeVars.
+# The default narrowing for isinstance(obj, dict) is dict[unknown, unknown],
+# however this cause Pyright to rightfully report errors. As we know we don't
+# care about the contained types we can safely use `object` in it's place.
+#
+# There are two separate functions defined, `is_*` and `is_*_t` for different use cases.
+# `is_*` is for when you're dealing with an unknown input
+# `is_*_t` is for when you're narrowing a known union type to a specific subset
+
+
+def is_tuple(obj: object) -> TypeGuard[tuple[object, ...]]:
+    return isinstance(obj, tuple)
+
+
+def is_tuple_t(obj: _TupleT | object) -> TypeGuard[_TupleT]:
+    return isinstance(obj, tuple)
+
+
+def is_sequence(obj: object) -> TypeGuard[Sequence[object]]:
+    return isinstance(obj, Sequence)
+
+
+def is_sequence_t(obj: _SequenceT | object) -> TypeGuard[_SequenceT]:
+    return isinstance(obj, Sequence)
+
+
+def is_mapping(obj: object) -> TypeGuard[Mapping[str, object]]:
+    return isinstance(obj, Mapping)
+
+
+def is_mapping_t(obj: _MappingT | object) -> TypeGuard[_MappingT]:
+    return isinstance(obj, Mapping)
+
+
+def is_dict(obj: object) -> TypeGuard[dict[object, object]]:
+    return isinstance(obj, dict)
+
+
+def is_list(obj: object) -> TypeGuard[list[object]]:
+    return isinstance(obj, list)
+
+
+def is_iterable(obj: object) -> TypeGuard[Iterable[object]]:
+    return isinstance(obj, Iterable)
+
+
+def deepcopy_minimal(item: _T) -> _T:
+    """Minimal reimplementation of copy.deepcopy() that will only copy certain object types:
+
+    - mappings, e.g. `dict`
+    - list
+
+    This is done for performance reasons.
+    """
+    if is_mapping(item):
+        return cast(_T, {k: deepcopy_minimal(v) for k, v in item.items()})
+    if is_list(item):
+        return cast(_T, [deepcopy_minimal(entry) for entry in item])
+    return item
+
+
+# copied from https://github.com/Rapptz/RoboDanny
+def human_join(seq: Sequence[str], *, delim: str = ", ", final: str = "or") -> str:
+    size = len(seq)
+    if size == 0:
+        return ""
+
+    if size == 1:
+        return seq[0]
+
+    if size == 2:
+        return f"{seq[0]} {final} {seq[1]}"
+
+    return delim.join(seq[:-1]) + f" {final} {seq[-1]}"
+
+
+def quote(string: str) -> str:
+    """Add single quotation marks around the given string. Does *not* do any escaping."""
+    return f"'{string}'"
+
+
+def required_args(*variants: Sequence[str]) -> Callable[[CallableT], CallableT]:
+    """Decorator to enforce a given set of arguments or variants of arguments are passed to the decorated function.
+
+    Useful for enforcing runtime validation of overloaded functions.
+
+    Example usage:
+    ```py
+    @overload
+    def foo(*, a: str) -> str:
+        ...
+
+
+    @overload
+    def foo(*, b: bool) -> str:
+        ...
+
+
+    # This enforces the same constraints that a static type checker would
+    # i.e. that either a or b must be passed to the function
+    @required_args(["a"], ["b"])
+    def foo(*, a: str | None = None, b: bool | None = None) -> str:
+        ...
+    ```
+    """
+
+    def inner(func: CallableT) -> CallableT:
+        params = inspect.signature(func).parameters
+        positional = [
+            name
+            for name, param in params.items()
+            if param.kind
+            in {
+                param.POSITIONAL_ONLY,
+                param.POSITIONAL_OR_KEYWORD,
+            }
+        ]
+
+        @functools.wraps(func)
+        def wrapper(*args: object, **kwargs: object) -> object:
+            given_params: set[str] = set()
+            for i, _ in enumerate(args):
+                try:
+                    given_params.add(positional[i])
+                except IndexError:
+                    raise TypeError(
+                        f"{func.__name__}() takes {len(positional)} argument(s) but {len(args)} were given"
+                    ) from None
+
+            given_params.update(kwargs.keys())
+
+            for variant in variants:
+                matches = all(param in given_params for param in variant)
+                if matches:
+                    break
+            else:  # no break
+                if len(variants) > 1:
+                    variations = human_join(
+                        ["(" + human_join([quote(arg) for arg in variant], final="and") + ")" for variant in variants]
+                    )
+                    msg = f"Missing required arguments; Expected either {variations} arguments to be given"
+                else:
+                    # TODO: this error message is not deterministic
+                    missing = list(set(variants[0]) - given_params)
+                    if len(missing) > 1:
+                        msg = f"Missing required arguments: {human_join([quote(arg) for arg in missing])}"
+                    else:
+                        msg = f"Missing required argument: {quote(missing[0])}"
+                raise TypeError(msg)
+            return func(*args, **kwargs)
+
+        return wrapper  # type: ignore
+
+    return inner
+
+
+_K = TypeVar("_K")
+_V = TypeVar("_V")
+
+
+@overload
+def strip_not_given(obj: None) -> None: ...
+
+
+@overload
+def strip_not_given(obj: Mapping[_K, _V | NotGiven]) -> dict[_K, _V]: ...
+
+
+@overload
+def strip_not_given(obj: object) -> object: ...
+
+
+def strip_not_given(obj: object | None) -> object:
+    """Remove all top-level keys where their values are instances of `NotGiven`"""
+    if obj is None:
+        return None
+
+    if not is_mapping(obj):
+        return obj
+
+    return {key: value for key, value in obj.items() if not isinstance(value, NotGiven)}
+
+
+def coerce_integer(val: str) -> int:
+    return int(val, base=10)
+
+
+def coerce_float(val: str) -> float:
+    return float(val)
+
+
+def coerce_boolean(val: str) -> bool:
+    return val in {"true", "1", "on"}
+
+
+def maybe_coerce_integer(val: str | None) -> int | None:
+    if val is None:
+        return None
+    return coerce_integer(val)
+
+
+def maybe_coerce_float(val: str | None) -> float | None:
+    if val is None:
+        return None
+    return coerce_float(val)
+
+
+def maybe_coerce_boolean(val: str | None) -> bool | None:
+    if val is None:
+        return None
+    return coerce_boolean(val)
+
+
+def removeprefix(string: str, prefix: str) -> str:
+    """Remove a prefix from a string.
+
+    Backport of `str.removeprefix` for Python < 3.9
+    """
+    if string.startswith(prefix):
+        return string[len(prefix) :]
+    return string
+
+
+def removesuffix(string: str, suffix: str) -> str:
+    """Remove a suffix from a string.
+
+    Backport of `str.removesuffix` for Python < 3.9
+    """
+    if string.endswith(suffix):
+        return string[: -len(suffix)]
+    return string
+
+
+def file_from_path(path: str) -> FileTypes:
+    contents = Path(path).read_bytes()
+    file_name = os.path.basename(path)
+    return (file_name, contents)
+
+
+def get_required_header(headers: HeadersLike, header: str) -> str:
+    lower_header = header.lower()
+    if isinstance(headers, Mapping):
+        headers = cast(Headers, headers)
+        for k, v in headers.items():
+            if k.lower() == lower_header and isinstance(v, str):
+                return v
+
+    """ to deal with the case where the header looks like Stainless-Event-Id """
+    intercaps_header = re.sub(r"([^\w])(\w)", lambda pat: pat.group(1) + pat.group(2).upper(), header.capitalize())
+
+    for normalized_header in [header, lower_header, header.upper(), intercaps_header]:
+        value = headers.get(normalized_header)
+        if value:
+            return value
+
+    raise ValueError(f"Could not find {header} header")
+
+
+def get_async_library() -> str:
+    try:
+        return sniffio.current_async_library()
+    except Exception:
+        return "false"
+
+
+def drop_prefix_image_data(content: Union[str, list[dict]]) -> Union[str, list[dict]]:
+    """
+    删除 ;base64, 前缀
+    :param image_data:
+    :return:
+    """
+    if isinstance(content, list):
+        for data in content:
+            if data.get("type") == "image_url":
+                image_data = data.get("image_url").get("url")
+                if image_data.startswith("data:image/"):
+                    image_data = image_data.split("base64,")[-1]
+                    data["image_url"]["url"] = image_data
+
+    return content
diff --git a/api/core/model_runtime/model_providers/zhipuai/zhipuai_sdk/core/logs.py b/api/core/model_runtime/model_providers/zhipuai/zhipuai_sdk/core/logs.py
new file mode 100644
index 0000000000000..e5fce94c00e9e
--- /dev/null
+++ b/api/core/model_runtime/model_providers/zhipuai/zhipuai_sdk/core/logs.py
@@ -0,0 +1,78 @@
+import logging
+import os
+import time
+
+logger = logging.getLogger(__name__)
+
+
+class LoggerNameFilter(logging.Filter):
+    def filter(self, record):
+        # return record.name.startswith("loom_core") or record.name in "ERROR" or (
+        #         record.name.startswith("uvicorn.error")
+        #         and record.getMessage().startswith("Uvicorn running on")
+        # )
+        return True
+
+
+def get_log_file(log_path: str, sub_dir: str):
+    """
+    sub_dir should contain a timestamp.
+    """
+    log_dir = os.path.join(log_path, sub_dir)
+    # Here should be creating a new directory each time, so `exist_ok=False`
+    os.makedirs(log_dir, exist_ok=False)
+    return os.path.join(log_dir, "zhipuai.log")
+
+
+def get_config_dict(log_level: str, log_file_path: str, log_backup_count: int, log_max_bytes: int) -> dict:
+    # for windows, the path should be a raw string.
+    log_file_path = log_file_path.encode("unicode-escape").decode() if os.name == "nt" else log_file_path
+    log_level = log_level.upper()
+    config_dict = {
+        "version": 1,
+        "disable_existing_loggers": False,
+        "formatters": {
+            "formatter": {"format": ("%(asctime)s %(name)-12s %(process)d %(levelname)-8s %(message)s")},
+        },
+        "filters": {
+            "logger_name_filter": {
+                "()": __name__ + ".LoggerNameFilter",
+            },
+        },
+        "handlers": {
+            "stream_handler": {
+                "class": "logging.StreamHandler",
+                "formatter": "formatter",
+                "level": log_level,
+                # "stream": "ext://sys.stdout",
+                # "filters": ["logger_name_filter"],
+            },
+            "file_handler": {
+                "class": "logging.handlers.RotatingFileHandler",
+                "formatter": "formatter",
+                "level": log_level,
+                "filename": log_file_path,
+                "mode": "a",
+                "maxBytes": log_max_bytes,
+                "backupCount": log_backup_count,
+                "encoding": "utf8",
+            },
+        },
+        "loggers": {
+            "loom_core": {
+                "handlers": ["stream_handler", "file_handler"],
+                "level": log_level,
+                "propagate": False,
+            }
+        },
+        "root": {
+            "level": log_level,
+            "handlers": ["stream_handler", "file_handler"],
+        },
+    }
+    return config_dict
+
+
+def get_timestamp_ms():
+    t = time.time()
+    return int(round(t * 1000))
diff --git a/api/core/model_runtime/model_providers/zhipuai/zhipuai_sdk/core/pagination.py b/api/core/model_runtime/model_providers/zhipuai/zhipuai_sdk/core/pagination.py
new file mode 100644
index 0000000000000..7f0b1b91d9855
--- /dev/null
+++ b/api/core/model_runtime/model_providers/zhipuai/zhipuai_sdk/core/pagination.py
@@ -0,0 +1,62 @@
+# File generated from our OpenAPI spec by Stainless. See CONTRIBUTING.md for details.
+
+from typing import Any, Generic, Optional, TypeVar, cast
+
+from typing_extensions import Protocol, override, runtime_checkable
+
+from ._http_client import BasePage, BaseSyncPage, PageInfo
+
+__all__ = ["SyncPage", "SyncCursorPage"]
+
+_T = TypeVar("_T")
+
+
+@runtime_checkable
+class CursorPageItem(Protocol):
+    id: Optional[str]
+
+
+class SyncPage(BaseSyncPage[_T], BasePage[_T], Generic[_T]):
+    """Note: no pagination actually occurs yet, this is for forwards-compatibility."""
+
+    data: list[_T]
+    object: str
+
+    @override
+    def _get_page_items(self) -> list[_T]:
+        data = self.data
+        if not data:
+            return []
+        return data
+
+    @override
+    def next_page_info(self) -> None:
+        """
+        This page represents a response that isn't actually paginated at the API level
+        so there will never be a next page.
+        """
+        return None
+
+
+class SyncCursorPage(BaseSyncPage[_T], BasePage[_T], Generic[_T]):
+    data: list[_T]
+
+    @override
+    def _get_page_items(self) -> list[_T]:
+        data = self.data
+        if not data:
+            return []
+        return data
+
+    @override
+    def next_page_info(self) -> Optional[PageInfo]:
+        data = self.data
+        if not data:
+            return None
+
+        item = cast(Any, data[-1])
+        if not isinstance(item, CursorPageItem) or item.id is None:
+            # TODO emit warning log
+            return None
+
+        return PageInfo(params={"after": item.id})
diff --git a/api/core/model_runtime/model_providers/zhipuai/zhipuai_sdk/types/assistant/__init__.py b/api/core/model_runtime/model_providers/zhipuai/zhipuai_sdk/types/assistant/__init__.py
new file mode 100644
index 0000000000000..9f941fb91c877
--- /dev/null
+++ b/api/core/model_runtime/model_providers/zhipuai/zhipuai_sdk/types/assistant/__init__.py
@@ -0,0 +1,5 @@
+from .assistant_completion import AssistantCompletion
+
+__all__ = [
+    "AssistantCompletion",
+]
diff --git a/api/core/model_runtime/model_providers/zhipuai/zhipuai_sdk/types/assistant/assistant_completion.py b/api/core/model_runtime/model_providers/zhipuai/zhipuai_sdk/types/assistant/assistant_completion.py
new file mode 100644
index 0000000000000..cbfb6edaeb1f1
--- /dev/null
+++ b/api/core/model_runtime/model_providers/zhipuai/zhipuai_sdk/types/assistant/assistant_completion.py
@@ -0,0 +1,40 @@
+from typing import Any, Optional
+
+from ...core import BaseModel
+from .message import MessageContent
+
+__all__ = ["AssistantCompletion", "CompletionUsage"]
+
+
+class ErrorInfo(BaseModel):
+    code: str  # 错误码
+    message: str  # 错误信息
+
+
+class AssistantChoice(BaseModel):
+    index: int  # 结果下标
+    delta: MessageContent  # 当前会话输出消息体
+    finish_reason: str
+    """
+    # 推理结束原因 stop代表推理自然结束或触发停止词。  sensitive 代表模型推理内容被安全审核接口拦截。请注意，针对此类内容，请用户自行判断并决定是否撤回已公开的内容。 
+    # network_error 代表模型推理服务异常。
+    """  # noqa: E501
+    metadata: dict  # 元信息，拓展字段
+
+
+class CompletionUsage(BaseModel):
+    prompt_tokens: int  # 输入的 tokens 数量
+    completion_tokens: int  # 输出的 tokens 数量
+    total_tokens: int  # 总 tokens 数量
+
+
+class AssistantCompletion(BaseModel):
+    id: str  # 请求 ID
+    conversation_id: str  # 会话 ID
+    assistant_id: str  # 智能体 ID
+    created: int  # 请求创建时间，Unix 时间戳
+    status: str  # 返回状态，包括：`completed` 表示生成结束`in_progress`表示生成中 `failed` 表示生成异常
+    last_error: Optional[ErrorInfo]  # 异常信息
+    choices: list[AssistantChoice]  # 增量返回的信息
+    metadata: Optional[dict[str, Any]]  # 元信息，拓展字段
+    usage: Optional[CompletionUsage]  # tokens 数量统计
diff --git a/api/core/model_runtime/model_providers/zhipuai/zhipuai_sdk/types/assistant/assistant_conversation_params.py b/api/core/model_runtime/model_providers/zhipuai/zhipuai_sdk/types/assistant/assistant_conversation_params.py
new file mode 100644
index 0000000000000..03f14f4238f37
--- /dev/null
+++ b/api/core/model_runtime/model_providers/zhipuai/zhipuai_sdk/types/assistant/assistant_conversation_params.py
@@ -0,0 +1,7 @@
+from typing import TypedDict
+
+
+class ConversationParameters(TypedDict, total=False):
+    assistant_id: str  # 智能体 ID
+    page: int  # 当前分页
+    page_size: int  # 分页数量
diff --git a/api/core/model_runtime/model_providers/zhipuai/zhipuai_sdk/types/assistant/assistant_conversation_resp.py b/api/core/model_runtime/model_providers/zhipuai/zhipuai_sdk/types/assistant/assistant_conversation_resp.py
new file mode 100644
index 0000000000000..d1833d220a2e3
--- /dev/null
+++ b/api/core/model_runtime/model_providers/zhipuai/zhipuai_sdk/types/assistant/assistant_conversation_resp.py
@@ -0,0 +1,29 @@
+from ...core import BaseModel
+
+__all__ = ["ConversationUsageListResp"]
+
+
+class Usage(BaseModel):
+    prompt_tokens: int  # 用户输入的 tokens 数量
+    completion_tokens: int  # 模型输入的 tokens 数量
+    total_tokens: int  # 总 tokens 数量
+
+
+class ConversationUsage(BaseModel):
+    id: str  # 会话 id
+    assistant_id: str  # 智能体Assistant id
+    create_time: int  # 创建时间
+    update_time: int  # 更新时间
+    usage: Usage  # 会话中 tokens 数量统计
+
+
+class ConversationUsageList(BaseModel):
+    assistant_id: str  # 智能体id
+    has_more: bool  # 是否还有更多页
+    conversation_list: list[ConversationUsage]  # 返回的
+
+
+class ConversationUsageListResp(BaseModel):
+    code: int
+    msg: str
+    data: ConversationUsageList
diff --git a/api/core/model_runtime/model_providers/zhipuai/zhipuai_sdk/types/assistant/assistant_create_params.py b/api/core/model_runtime/model_providers/zhipuai/zhipuai_sdk/types/assistant/assistant_create_params.py
new file mode 100644
index 0000000000000..2def1025cd2b3
--- /dev/null
+++ b/api/core/model_runtime/model_providers/zhipuai/zhipuai_sdk/types/assistant/assistant_create_params.py
@@ -0,0 +1,32 @@
+from typing import Optional, TypedDict, Union
+
+
+class AssistantAttachments:
+    file_id: str
+
+
+class MessageTextContent:
+    type: str  # 目前支持 type = text
+    text: str
+
+
+MessageContent = Union[MessageTextContent]
+
+
+class ConversationMessage(TypedDict):
+    """会话消息体"""
+
+    role: str  # 用户的输入角色，例如 'user'
+    content: list[MessageContent]  # 会话消息体的内容
+
+
+class AssistantParameters(TypedDict, total=False):
+    """智能体参数类"""
+
+    assistant_id: str  # 智能体 ID
+    conversation_id: Optional[str]  # 会话 ID，不传则创建新会话
+    model: str  # 模型名称，默认为 'GLM-4-Assistant'
+    stream: bool  # 是否支持流式 SSE，需要传入 True
+    messages: list[ConversationMessage]  # 会话消息体
+    attachments: Optional[list[AssistantAttachments]]  # 会话指定的文件，非必填
+    metadata: Optional[dict]  # 元信息，拓展字段，非必填
diff --git a/api/core/model_runtime/model_providers/zhipuai/zhipuai_sdk/types/assistant/assistant_support_resp.py b/api/core/model_runtime/model_providers/zhipuai/zhipuai_sdk/types/assistant/assistant_support_resp.py
new file mode 100644
index 0000000000000..0709cdbcad25e
--- /dev/null
+++ b/api/core/model_runtime/model_providers/zhipuai/zhipuai_sdk/types/assistant/assistant_support_resp.py
@@ -0,0 +1,21 @@
+from ...core import BaseModel
+
+__all__ = ["AssistantSupportResp"]
+
+
+class AssistantSupport(BaseModel):
+    assistant_id: str  # 智能体的 Assistant id，用于智能体会话
+    created_at: int  # 创建时间
+    updated_at: int  # 更新时间
+    name: str  # 智能体名称
+    avatar: str  # 智能体头像
+    description: str  # 智能体描述
+    status: str  # 智能体状态，目前只有 publish
+    tools: list[str]  # 智能体支持的工具名
+    starter_prompts: list[str]  # 智能体启动推荐的 prompt
+
+
+class AssistantSupportResp(BaseModel):
+    code: int
+    msg: str
+    data: list[AssistantSupport]  # 智能体列表
diff --git a/api/core/model_runtime/model_providers/zhipuai/zhipuai_sdk/types/assistant/message/__init__.py b/api/core/model_runtime/model_providers/zhipuai/zhipuai_sdk/types/assistant/message/__init__.py
new file mode 100644
index 0000000000000..562e0151e53b4
--- /dev/null
+++ b/api/core/model_runtime/model_providers/zhipuai/zhipuai_sdk/types/assistant/message/__init__.py
@@ -0,0 +1,3 @@
+from .message_content import MessageContent
+
+__all__ = ["MessageContent"]
diff --git a/api/core/model_runtime/model_providers/zhipuai/zhipuai_sdk/types/assistant/message/message_content.py b/api/core/model_runtime/model_providers/zhipuai/zhipuai_sdk/types/assistant/message/message_content.py
new file mode 100644
index 0000000000000..6a1a438a6fe03
--- /dev/null
+++ b/api/core/model_runtime/model_providers/zhipuai/zhipuai_sdk/types/assistant/message/message_content.py
@@ -0,0 +1,13 @@
+from typing import Annotated, TypeAlias, Union
+
+from ....core._utils import PropertyInfo
+from .text_content_block import TextContentBlock
+from .tools_delta_block import ToolsDeltaBlock
+
+__all__ = ["MessageContent"]
+
+
+MessageContent: TypeAlias = Annotated[
+    Union[ToolsDeltaBlock, TextContentBlock],
+    PropertyInfo(discriminator="type"),
+]
diff --git a/api/core/model_runtime/model_providers/zhipuai/zhipuai_sdk/types/assistant/message/text_content_block.py b/api/core/model_runtime/model_providers/zhipuai/zhipuai_sdk/types/assistant/message/text_content_block.py
new file mode 100644
index 0000000000000..865fb1139e2f7
--- /dev/null
+++ b/api/core/model_runtime/model_providers/zhipuai/zhipuai_sdk/types/assistant/message/text_content_block.py
@@ -0,0 +1,14 @@
+from typing import Literal
+
+from ....core import BaseModel
+
+__all__ = ["TextContentBlock"]
+
+
+class TextContentBlock(BaseModel):
+    content: str
+
+    role: str = "assistant"
+
+    type: Literal["content"] = "content"
+    """Always `content`."""
diff --git a/api/core/model_runtime/model_providers/zhipuai/zhipuai_sdk/types/assistant/message/tools/code_interpreter_delta_block.py b/api/core/model_runtime/model_providers/zhipuai/zhipuai_sdk/types/assistant/message/tools/code_interpreter_delta_block.py
new file mode 100644
index 0000000000000..9d569b282ef9f
--- /dev/null
+++ b/api/core/model_runtime/model_providers/zhipuai/zhipuai_sdk/types/assistant/message/tools/code_interpreter_delta_block.py
@@ -0,0 +1,27 @@
+from typing import Literal
+
+__all__ = ["CodeInterpreterToolBlock"]
+
+from .....core import BaseModel
+
+
+class CodeInterpreterToolOutput(BaseModel):
+    """代码工具输出结果"""
+
+    type: str  # 代码执行日志，目前只有 logs
+    logs: str  # 代码执行的日志结果
+    error_msg: str  # 错误信息
+
+
+class CodeInterpreter(BaseModel):
+    """代码解释器"""
+
+    input: str  # 生成的代码片段，输入给代码沙盒
+    outputs: list[CodeInterpreterToolOutput]  # 代码执行后的输出结果
+
+
+class CodeInterpreterToolBlock(BaseModel):
+    """代码工具块"""
+
+    code_interpreter: CodeInterpreter  # 代码解释器对象
+    type: Literal["code_interpreter"]  # 调用工具的类型，始终为 `code_interpreter`
diff --git a/api/core/model_runtime/model_providers/zhipuai/zhipuai_sdk/types/assistant/message/tools/drawing_tool_delta_block.py b/api/core/model_runtime/model_providers/zhipuai/zhipuai_sdk/types/assistant/message/tools/drawing_tool_delta_block.py
new file mode 100644
index 0000000000000..0b6895556b616
--- /dev/null
+++ b/api/core/model_runtime/model_providers/zhipuai/zhipuai_sdk/types/assistant/message/tools/drawing_tool_delta_block.py
@@ -0,0 +1,21 @@
+from typing import Literal
+
+from .....core import BaseModel
+
+__all__ = ["DrawingToolBlock"]
+
+
+class DrawingToolOutput(BaseModel):
+    image: str
+
+
+class DrawingTool(BaseModel):
+    input: str
+    outputs: list[DrawingToolOutput]
+
+
+class DrawingToolBlock(BaseModel):
+    drawing_tool: DrawingTool
+
+    type: Literal["drawing_tool"]
+    """Always `drawing_tool`."""
diff --git a/api/core/model_runtime/model_providers/zhipuai/zhipuai_sdk/types/assistant/message/tools/function_delta_block.py b/api/core/model_runtime/model_providers/zhipuai/zhipuai_sdk/types/assistant/message/tools/function_delta_block.py
new file mode 100644
index 0000000000000..c439bc4b3fbbb
--- /dev/null
+++ b/api/core/model_runtime/model_providers/zhipuai/zhipuai_sdk/types/assistant/message/tools/function_delta_block.py
@@ -0,0 +1,22 @@
+from typing import Literal, Union
+
+__all__ = ["FunctionToolBlock"]
+
+from .....core import BaseModel
+
+
+class FunctionToolOutput(BaseModel):
+    content: str
+
+
+class FunctionTool(BaseModel):
+    name: str
+    arguments: Union[str, dict]
+    outputs: list[FunctionToolOutput]
+
+
+class FunctionToolBlock(BaseModel):
+    function: FunctionTool
+
+    type: Literal["function"]
+    """Always `drawing_tool`."""
diff --git a/api/core/model_runtime/model_providers/zhipuai/zhipuai_sdk/types/assistant/message/tools/retrieval_delta_black.py b/api/core/model_runtime/model_providers/zhipuai/zhipuai_sdk/types/assistant/message/tools/retrieval_delta_black.py
new file mode 100644
index 0000000000000..4789e9378a8a3
--- /dev/null
+++ b/api/core/model_runtime/model_providers/zhipuai/zhipuai_sdk/types/assistant/message/tools/retrieval_delta_black.py
@@ -0,0 +1,41 @@
+from typing import Literal
+
+from .....core import BaseModel
+
+
+class RetrievalToolOutput(BaseModel):
+    """
+    This class represents the output of a retrieval tool.
+
+    Attributes:
+    - text (str): The text snippet retrieved from the knowledge base.
+    - document (str): The name of the document from which the text snippet was retrieved, returned only in intelligent configuration.
+    """  # noqa: E501
+
+    text: str
+    document: str
+
+
+class RetrievalTool(BaseModel):
+    """
+    This class represents the outputs of a retrieval tool.
+
+    Attributes:
+    - outputs (List[RetrievalToolOutput]): A list of text snippets and their respective document names retrieved from the knowledge base.
+    """  # noqa: E501
+
+    outputs: list[RetrievalToolOutput]
+
+
+class RetrievalToolBlock(BaseModel):
+    """
+    This class represents a block for invoking the retrieval tool.
+
+    Attributes:
+    - retrieval (RetrievalTool): An instance of the RetrievalTool class containing the retrieval outputs.
+    - type (Literal["retrieval"]): The type of tool being used, always set to "retrieval".
+    """
+
+    retrieval: RetrievalTool
+    type: Literal["retrieval"]
+    """Always `retrieval`."""
diff --git a/api/core/model_runtime/model_providers/zhipuai/zhipuai_sdk/types/assistant/message/tools/tools_type.py b/api/core/model_runtime/model_providers/zhipuai/zhipuai_sdk/types/assistant/message/tools/tools_type.py
new file mode 100644
index 0000000000000..98544053d4c83
--- /dev/null
+++ b/api/core/model_runtime/model_providers/zhipuai/zhipuai_sdk/types/assistant/message/tools/tools_type.py
@@ -0,0 +1,16 @@
+from typing import Annotated, TypeAlias, Union
+
+from .....core._utils import PropertyInfo
+from .code_interpreter_delta_block import CodeInterpreterToolBlock
+from .drawing_tool_delta_block import DrawingToolBlock
+from .function_delta_block import FunctionToolBlock
+from .retrieval_delta_black import RetrievalToolBlock
+from .web_browser_delta_block import WebBrowserToolBlock
+
+__all__ = ["ToolsType"]
+
+
+ToolsType: TypeAlias = Annotated[
+    Union[DrawingToolBlock, CodeInterpreterToolBlock, WebBrowserToolBlock, RetrievalToolBlock, FunctionToolBlock],
+    PropertyInfo(discriminator="type"),
+]
diff --git a/api/core/model_runtime/model_providers/zhipuai/zhipuai_sdk/types/assistant/message/tools/web_browser_delta_block.py b/api/core/model_runtime/model_providers/zhipuai/zhipuai_sdk/types/assistant/message/tools/web_browser_delta_block.py
new file mode 100644
index 0000000000000..966e6fe0c84fe
--- /dev/null
+++ b/api/core/model_runtime/model_providers/zhipuai/zhipuai_sdk/types/assistant/message/tools/web_browser_delta_block.py
@@ -0,0 +1,48 @@
+from typing import Literal
+
+from .....core import BaseModel
+
+__all__ = ["WebBrowserToolBlock"]
+
+
+class WebBrowserOutput(BaseModel):
+    """
+    This class represents the output of a web browser search result.
+
+    Attributes:
+    - title (str): The title of the search result.
+    - link (str): The URL link to the search result's webpage.
+    - content (str): The textual content extracted from the search result.
+    - error_msg (str): Any error message encountered during the search or retrieval process.
+    """
+
+    title: str
+    link: str
+    content: str
+    error_msg: str
+
+
+class WebBrowser(BaseModel):
+    """
+    This class represents the input and outputs of a web browser search.
+
+    Attributes:
+    - input (str): The input query for the web browser search.
+    - outputs (List[WebBrowserOutput]): A list of search results returned by the web browser.
+    """
+
+    input: str
+    outputs: list[WebBrowserOutput]
+
+
+class WebBrowserToolBlock(BaseModel):
+    """
+    This class represents a block for invoking the web browser tool.
+
+    Attributes:
+    - web_browser (WebBrowser): An instance of the WebBrowser class containing the search input and outputs.
+    - type (Literal["web_browser"]): The type of tool being used, always set to "web_browser".
+    """
+
+    web_browser: WebBrowser
+    type: Literal["web_browser"]
diff --git a/api/core/model_runtime/model_providers/zhipuai/zhipuai_sdk/types/assistant/message/tools_delta_block.py b/api/core/model_runtime/model_providers/zhipuai/zhipuai_sdk/types/assistant/message/tools_delta_block.py
new file mode 100644
index 0000000000000..781a1ab819c28
--- /dev/null
+++ b/api/core/model_runtime/model_providers/zhipuai/zhipuai_sdk/types/assistant/message/tools_delta_block.py
@@ -0,0 +1,16 @@
+from typing import Literal
+
+from ....core import BaseModel
+from .tools.tools_type import ToolsType
+
+__all__ = ["ToolsDeltaBlock"]
+
+
+class ToolsDeltaBlock(BaseModel):
+    tool_calls: list[ToolsType]
+    """The index of the content part in the message."""
+
+    role: str = "tool"
+
+    type: Literal["tool_calls"] = "tool_calls"
+    """Always `tool_calls`."""
diff --git a/api/core/model_runtime/model_providers/zhipuai/zhipuai_sdk/types/batch.py b/api/core/model_runtime/model_providers/zhipuai/zhipuai_sdk/types/batch.py
new file mode 100644
index 0000000000000..560562915c9d3
--- /dev/null
+++ b/api/core/model_runtime/model_providers/zhipuai/zhipuai_sdk/types/batch.py
@@ -0,0 +1,82 @@
+# File generated from our OpenAPI spec by Stainless. See CONTRIBUTING.md for details.
+
+import builtins
+from typing import Literal, Optional
+
+from ..core import BaseModel
+from .batch_error import BatchError
+from .batch_request_counts import BatchRequestCounts
+
+__all__ = ["Batch", "Errors"]
+
+
+class Errors(BaseModel):
+    data: Optional[list[BatchError]] = None
+
+    object: Optional[str] = None
+    """这个类型，一直是`list`。"""
+
+
+class Batch(BaseModel):
+    id: str
+
+    completion_window: str
+    """用于执行请求的地址信息。"""
+
+    created_at: int
+    """这是 Unix timestamp (in seconds) 表示的创建时间。"""
+
+    endpoint: str
+    """这是ZhipuAI endpoint的地址。"""
+
+    input_file_id: str
+    """标记为batch的输入文件的ID。"""
+
+    object: Literal["batch"]
+    """这个类型，一直是`batch`."""
+
+    status: Literal[
+        "validating", "failed", "in_progress", "finalizing", "completed", "expired", "cancelling", "cancelled"
+    ]
+    """batch 的状态。"""
+
+    cancelled_at: Optional[int] = None
+    """Unix timestamp (in seconds) 表示的取消时间。"""
+
+    cancelling_at: Optional[int] = None
+    """Unix timestamp (in seconds) 表示发起取消的请求时间 """
+
+    completed_at: Optional[int] = None
+    """Unix timestamp (in seconds) 表示的完成时间。"""
+
+    error_file_id: Optional[str] = None
+    """这个文件id包含了执行请求失败的请求的输出。"""
+
+    errors: Optional[Errors] = None
+
+    expired_at: Optional[int] = None
+    """Unix timestamp (in seconds) 表示的将在过期时间。"""
+
+    expires_at: Optional[int] = None
+    """Unix timestamp (in seconds) 触发过期"""
+
+    failed_at: Optional[int] = None
+    """Unix timestamp (in seconds) 表示的失败时间。"""
+
+    finalizing_at: Optional[int] = None
+    """Unix timestamp (in seconds) 表示的最终时间。"""
+
+    in_progress_at: Optional[int] = None
+    """Unix timestamp (in seconds) 表示的开始处理时间。"""
+
+    metadata: Optional[builtins.object] = None
+    """ 
+    key:value形式的元数据，以便将信息存储
+        结构化格式。键的长度是64个字符，值最长512个字符
+    """
+
+    output_file_id: Optional[str] = None
+    """完成请求的输出文件的ID。"""
+
+    request_counts: Optional[BatchRequestCounts] = None
+    """批次中不同状态的请求计数"""
diff --git a/api/core/model_runtime/model_providers/zhipuai/zhipuai_sdk/types/batch_create_params.py b/api/core/model_runtime/model_providers/zhipuai/zhipuai_sdk/types/batch_create_params.py
new file mode 100644
index 0000000000000..3dae65ea46fcb
--- /dev/null
+++ b/api/core/model_runtime/model_providers/zhipuai/zhipuai_sdk/types/batch_create_params.py
@@ -0,0 +1,37 @@
+from __future__ import annotations
+
+from typing import Literal, Optional
+
+from typing_extensions import Required, TypedDict
+
+__all__ = ["BatchCreateParams"]
+
+
+class BatchCreateParams(TypedDict, total=False):
+    completion_window: Required[str]
+    """The time frame within which the batch should be processed.
+
+    Currently only `24h` is supported.
+    """
+
+    endpoint: Required[Literal["/v1/chat/completions", "/v1/embeddings"]]
+    """The endpoint to be used for all requests in the batch.
+
+    Currently `/v1/chat/completions` and `/v1/embeddings` are supported.
+    """
+
+    input_file_id: Required[str]
+    """The ID of an uploaded file that contains requests for the new batch.
+
+    See [upload file](https://platform.openai.com/docs/api-reference/files/create)
+    for how to upload a file.
+
+    Your input file must be formatted as a
+    [JSONL file](https://platform.openai.com/docs/api-reference/batch/requestInput),
+    and must be uploaded with the purpose `batch`.
+    """
+
+    metadata: Optional[dict[str, str]]
+    """Optional custom metadata for the batch."""
+
+    auto_delete_input_file: Optional[bool]
diff --git a/api/core/model_runtime/model_providers/zhipuai/zhipuai_sdk/types/batch_error.py b/api/core/model_runtime/model_providers/zhipuai/zhipuai_sdk/types/batch_error.py
new file mode 100644
index 0000000000000..f934db19781e4
--- /dev/null
+++ b/api/core/model_runtime/model_providers/zhipuai/zhipuai_sdk/types/batch_error.py
@@ -0,0 +1,21 @@
+# File generated from our OpenAPI spec by Stainless. See CONTRIBUTING.md for details.
+
+from typing import Optional
+
+from ..core import BaseModel
+
+__all__ = ["BatchError"]
+
+
+class BatchError(BaseModel):
+    code: Optional[str] = None
+    """定义的业务错误码"""
+
+    line: Optional[int] = None
+    """文件中的行号"""
+
+    message: Optional[str] = None
+    """关于对话文件中的错误的描述"""
+
+    param: Optional[str] = None
+    """参数名称，如果有的话"""
diff --git a/api/core/model_runtime/model_providers/zhipuai/zhipuai_sdk/types/batch_list_params.py b/api/core/model_runtime/model_providers/zhipuai/zhipuai_sdk/types/batch_list_params.py
new file mode 100644
index 0000000000000..1a681671320ec
--- /dev/null
+++ b/api/core/model_runtime/model_providers/zhipuai/zhipuai_sdk/types/batch_list_params.py
@@ -0,0 +1,20 @@
+from __future__ import annotations
+
+from typing_extensions import TypedDict
+
+__all__ = ["BatchListParams"]
+
+
+class BatchListParams(TypedDict, total=False):
+    after: str
+    """分页的游标，用于获取下一页的数据。
+
+    `after` 是一个指向当前页面的游标，用于获取下一页的数据。如果没有提供 `after`，则返回第一页的数据。
+    list.
+    """
+
+    limit: int
+    """这个参数用于限制返回的结果数量。
+
+    Limit 用于限制返回的结果数量。默认值为 10
+    """
diff --git a/api/core/model_runtime/model_providers/zhipuai/zhipuai_sdk/types/batch_request_counts.py b/api/core/model_runtime/model_providers/zhipuai/zhipuai_sdk/types/batch_request_counts.py
new file mode 100644
index 0000000000000..ca3ccae625052
--- /dev/null
+++ b/api/core/model_runtime/model_providers/zhipuai/zhipuai_sdk/types/batch_request_counts.py
@@ -0,0 +1,14 @@
+from ..core import BaseModel
+
+__all__ = ["BatchRequestCounts"]
+
+
+class BatchRequestCounts(BaseModel):
+    completed: int
+    """这个数字表示已经完成的请求。"""
+
+    failed: int
+    """这个数字表示失败的请求。"""
+
+    total: int
+    """这个数字表示总的请求。"""
diff --git a/api/core/model_runtime/model_providers/zhipuai/zhipuai_sdk/types/chat/async_chat_completion.py b/api/core/model_runtime/model_providers/zhipuai/zhipuai_sdk/types/chat/async_chat_completion.py
index a0645b0916882..c1eed070f32d9 100644
--- a/api/core/model_runtime/model_providers/zhipuai/zhipuai_sdk/types/chat/async_chat_completion.py
+++ b/api/core/model_runtime/model_providers/zhipuai/zhipuai_sdk/types/chat/async_chat_completion.py
@@ -1,10 +1,9 @@
 from typing import Optional
 
-from pydantic import BaseModel
-
+from ...core import BaseModel
 from .chat_completion import CompletionChoice, CompletionUsage
 
-__all__ = ["AsyncTaskStatus"]
+__all__ = ["AsyncTaskStatus", "AsyncCompletion"]
 
 
 class AsyncTaskStatus(BaseModel):
diff --git a/api/core/model_runtime/model_providers/zhipuai/zhipuai_sdk/types/chat/chat_completion.py b/api/core/model_runtime/model_providers/zhipuai/zhipuai_sdk/types/chat/chat_completion.py
index 4b3a929a2b816..1945a826cda2d 100644
--- a/api/core/model_runtime/model_providers/zhipuai/zhipuai_sdk/types/chat/chat_completion.py
+++ b/api/core/model_runtime/model_providers/zhipuai/zhipuai_sdk/types/chat/chat_completion.py
@@ -1,6 +1,6 @@
 from typing import Optional
 
-from pydantic import BaseModel
+from ...core import BaseModel
 
 __all__ = ["Completion", "CompletionUsage"]
 
diff --git a/api/core/model_runtime/model_providers/zhipuai/zhipuai_sdk/types/chat/chat_completion_chunk.py b/api/core/model_runtime/model_providers/zhipuai/zhipuai_sdk/types/chat/chat_completion_chunk.py
index c250699741981..27fad0008a1dd 100644
--- a/api/core/model_runtime/model_providers/zhipuai/zhipuai_sdk/types/chat/chat_completion_chunk.py
+++ b/api/core/model_runtime/model_providers/zhipuai/zhipuai_sdk/types/chat/chat_completion_chunk.py
@@ -1,8 +1,9 @@
-from typing import Optional
+from typing import Any, Optional
 
-from pydantic import BaseModel
+from ...core import BaseModel
 
 __all__ = [
+    "CompletionUsage",
     "ChatCompletionChunk",
     "Choice",
     "ChoiceDelta",
@@ -53,3 +54,4 @@ class ChatCompletionChunk(BaseModel):
     created: Optional[int] = None
     model: Optional[str] = None
     usage: Optional[CompletionUsage] = None
+    extra_json: dict[str, Any]
diff --git a/api/core/model_runtime/model_providers/zhipuai/zhipuai_sdk/types/chat/code_geex/code_geex_params.py b/api/core/model_runtime/model_providers/zhipuai/zhipuai_sdk/types/chat/code_geex/code_geex_params.py
new file mode 100644
index 0000000000000..666b38855cd63
--- /dev/null
+++ b/api/core/model_runtime/model_providers/zhipuai/zhipuai_sdk/types/chat/code_geex/code_geex_params.py
@@ -0,0 +1,146 @@
+from typing import Literal, Optional
+
+from typing_extensions import Required, TypedDict
+
+__all__ = [
+    "CodeGeexTarget",
+    "CodeGeexContext",
+    "CodeGeexExtra",
+]
+
+
+class CodeGeexTarget(TypedDict, total=False):
+    """补全的内容参数"""
+
+    path: Optional[str]
+    """文件路径"""
+    language: Required[
+        Literal[
+            "c",
+            "c++",
+            "cpp",
+            "c#",
+            "csharp",
+            "c-sharp",
+            "css",
+            "cuda",
+            "dart",
+            "lua",
+            "objectivec",
+            "objective-c",
+            "objective-c++",
+            "python",
+            "perl",
+            "prolog",
+            "swift",
+            "lisp",
+            "java",
+            "scala",
+            "tex",
+            "jsx",
+            "tsx",
+            "vue",
+            "markdown",
+            "html",
+            "php",
+            "js",
+            "javascript",
+            "typescript",
+            "go",
+            "shell",
+            "rust",
+            "sql",
+            "kotlin",
+            "vb",
+            "ruby",
+            "pascal",
+            "r",
+            "fortran",
+            "lean",
+            "matlab",
+            "delphi",
+            "scheme",
+            "basic",
+            "assembly",
+            "groovy",
+            "abap",
+            "gdscript",
+            "haskell",
+            "julia",
+            "elixir",
+            "excel",
+            "clojure",
+            "actionscript",
+            "solidity",
+            "powershell",
+            "erlang",
+            "cobol",
+            "alloy",
+            "awk",
+            "thrift",
+            "sparql",
+            "augeas",
+            "cmake",
+            "f-sharp",
+            "stan",
+            "isabelle",
+            "dockerfile",
+            "rmarkdown",
+            "literate-agda",
+            "tcl",
+            "glsl",
+            "antlr",
+            "verilog",
+            "racket",
+            "standard-ml",
+            "elm",
+            "yaml",
+            "smalltalk",
+            "ocaml",
+            "idris",
+            "visual-basic",
+            "protocol-buffer",
+            "bluespec",
+            "applescript",
+            "makefile",
+            "tcsh",
+            "maple",
+            "systemverilog",
+            "literate-coffeescript",
+            "vhdl",
+            "restructuredtext",
+            "sas",
+            "literate-haskell",
+            "java-server-pages",
+            "coffeescript",
+            "emacs-lisp",
+            "mathematica",
+            "xslt",
+            "zig",
+            "common-lisp",
+            "stata",
+            "agda",
+            "ada",
+        ]
+    ]
+    """代码语言类型，如python"""
+    code_prefix: Required[str]
+    """补全位置的前文"""
+    code_suffix: Required[str]
+    """补全位置的后文"""
+
+
+class CodeGeexContext(TypedDict, total=False):
+    """附加代码"""
+
+    path: Required[str]
+    """附加代码文件的路径"""
+    code: Required[str]
+    """附加的代码内容"""
+
+
+class CodeGeexExtra(TypedDict, total=False):
+    target: Required[CodeGeexTarget]
+    """补全的内容参数"""
+    contexts: Optional[list[CodeGeexContext]]
+    """附加代码"""
diff --git a/api/core/model_runtime/model_providers/zhipuai/zhipuai_sdk/types/embeddings.py b/api/core/model_runtime/model_providers/zhipuai/zhipuai_sdk/types/embeddings.py
index e01f2c815fb38..8425b5c86688d 100644
--- a/api/core/model_runtime/model_providers/zhipuai/zhipuai_sdk/types/embeddings.py
+++ b/api/core/model_runtime/model_providers/zhipuai/zhipuai_sdk/types/embeddings.py
@@ -2,8 +2,7 @@
 
 from typing import Optional
 
-from pydantic import BaseModel
-
+from ..core import BaseModel
 from .chat.chat_completion import CompletionUsage
 
 __all__ = ["Embedding", "EmbeddingsResponded"]
diff --git a/api/core/model_runtime/model_providers/zhipuai/zhipuai_sdk/types/files/__init__.py b/api/core/model_runtime/model_providers/zhipuai/zhipuai_sdk/types/files/__init__.py
new file mode 100644
index 0000000000000..bbaf59e4d7d17
--- /dev/null
+++ b/api/core/model_runtime/model_providers/zhipuai/zhipuai_sdk/types/files/__init__.py
@@ -0,0 +1,5 @@
+from .file_deleted import FileDeleted
+from .file_object import FileObject, ListOfFileObject
+from .upload_detail import UploadDetail
+
+__all__ = ["FileObject", "ListOfFileObject", "UploadDetail", "FileDeleted"]
diff --git a/api/core/model_runtime/model_providers/zhipuai/zhipuai_sdk/types/files/file_create_params.py b/api/core/model_runtime/model_providers/zhipuai/zhipuai_sdk/types/files/file_create_params.py
new file mode 100644
index 0000000000000..4ef93b1c05aca
--- /dev/null
+++ b/api/core/model_runtime/model_providers/zhipuai/zhipuai_sdk/types/files/file_create_params.py
@@ -0,0 +1,38 @@
+from __future__ import annotations
+
+from typing import Literal, Optional
+
+from typing_extensions import Required, TypedDict
+
+__all__ = ["FileCreateParams"]
+
+from ...core import FileTypes
+from . import UploadDetail
+
+
+class FileCreateParams(TypedDict, total=False):
+    file: FileTypes
+    """file和 upload_detail二选一必填"""
+
+    upload_detail: list[UploadDetail]
+    """file和 upload_detail二选一必填"""
+
+    purpose: Required[Literal["fine-tune", "retrieval", "batch"]]
+    """ 
+    上传文件的用途，支持 "fine-tune和 "retrieval"
+    retrieval支持上传Doc、Docx、PDF、Xlsx、URL类型文件，且单个文件的大小不超过 5MB。
+    fine-tune支持上传.jsonl文件且当前单个文件的大小最大可为 100 MB ，文件中语料格式需满足微调指南中所描述的格式。
+    """
+    custom_separator: Optional[list[str]]
+    """ 
+    当 purpose 为 retrieval 且文件类型为 pdf, url, docx 时上传，切片规则默认为 `\n`。
+    """
+    knowledge_id: str
+    """ 
+        当文件上传目的为  retrieval 时，需要指定知识库ID进行上传。
+    """
+
+    sentence_size: int
+    """ 
+        当文件上传目的为  retrieval 时，需要指定知识库ID进行上传。
+    """
diff --git a/api/core/model_runtime/model_providers/zhipuai/zhipuai_sdk/types/files/file_deleted.py b/api/core/model_runtime/model_providers/zhipuai/zhipuai_sdk/types/files/file_deleted.py
new file mode 100644
index 0000000000000..a384b1a69a573
--- /dev/null
+++ b/api/core/model_runtime/model_providers/zhipuai/zhipuai_sdk/types/files/file_deleted.py
@@ -0,0 +1,13 @@
+from typing import Literal
+
+from ...core import BaseModel
+
+__all__ = ["FileDeleted"]
+
+
+class FileDeleted(BaseModel):
+    id: str
+
+    deleted: bool
+
+    object: Literal["file"]
diff --git a/api/core/model_runtime/model_providers/zhipuai/zhipuai_sdk/types/file_object.py b/api/core/model_runtime/model_providers/zhipuai/zhipuai_sdk/types/files/file_object.py
similarity index 86%
rename from api/core/model_runtime/model_providers/zhipuai/zhipuai_sdk/types/file_object.py
rename to api/core/model_runtime/model_providers/zhipuai/zhipuai_sdk/types/files/file_object.py
index 75f76fe969faf..8f9d0fbb8e6ce 100644
--- a/api/core/model_runtime/model_providers/zhipuai/zhipuai_sdk/types/file_object.py
+++ b/api/core/model_runtime/model_providers/zhipuai/zhipuai_sdk/types/files/file_object.py
@@ -1,8 +1,8 @@
 from typing import Optional
 
-from pydantic import BaseModel
+from ...core import BaseModel
 
-__all__ = ["FileObject"]
+__all__ = ["FileObject", "ListOfFileObject"]
 
 
 class FileObject(BaseModel):
diff --git a/api/core/model_runtime/model_providers/zhipuai/zhipuai_sdk/types/files/upload_detail.py b/api/core/model_runtime/model_providers/zhipuai/zhipuai_sdk/types/files/upload_detail.py
new file mode 100644
index 0000000000000..8f1ca5ce5756a
--- /dev/null
+++ b/api/core/model_runtime/model_providers/zhipuai/zhipuai_sdk/types/files/upload_detail.py
@@ -0,0 +1,13 @@
+from typing import Optional
+
+from ...core import BaseModel
+
+
+class UploadDetail(BaseModel):
+    url: str
+    knowledge_type: int
+    file_name: Optional[str] = None
+    sentence_size: Optional[int] = None
+    custom_separator: Optional[list[str]] = None
+    callback_url: Optional[str] = None
+    callback_header: Optional[dict[str, str]] = None
diff --git a/api/core/model_runtime/model_providers/zhipuai/zhipuai_sdk/types/fine_tuning/fine_tuning_job.py b/api/core/model_runtime/model_providers/zhipuai/zhipuai_sdk/types/fine_tuning/fine_tuning_job.py
index 1d3930286b89d..75c7553dbe35c 100644
--- a/api/core/model_runtime/model_providers/zhipuai/zhipuai_sdk/types/fine_tuning/fine_tuning_job.py
+++ b/api/core/model_runtime/model_providers/zhipuai/zhipuai_sdk/types/fine_tuning/fine_tuning_job.py
@@ -1,6 +1,6 @@
 from typing import Optional, Union
 
-from pydantic import BaseModel
+from ...core import BaseModel
 
 __all__ = ["FineTuningJob", "Error", "Hyperparameters", "ListOfFineTuningJob"]
 
diff --git a/api/core/model_runtime/model_providers/zhipuai/zhipuai_sdk/types/fine_tuning/fine_tuning_job_event.py b/api/core/model_runtime/model_providers/zhipuai/zhipuai_sdk/types/fine_tuning/fine_tuning_job_event.py
index e26b448534246..f996cff11430b 100644
--- a/api/core/model_runtime/model_providers/zhipuai/zhipuai_sdk/types/fine_tuning/fine_tuning_job_event.py
+++ b/api/core/model_runtime/model_providers/zhipuai/zhipuai_sdk/types/fine_tuning/fine_tuning_job_event.py
@@ -1,6 +1,6 @@
 from typing import Optional, Union
 
-from pydantic import BaseModel
+from ...core import BaseModel
 
 __all__ = ["FineTuningJobEvent", "Metric", "JobEvent"]
 
diff --git a/api/core/model_runtime/model_providers/zhipuai/zhipuai_sdk/types/fine_tuning/models/__init__.py b/api/core/model_runtime/model_providers/zhipuai/zhipuai_sdk/types/fine_tuning/models/__init__.py
new file mode 100644
index 0000000000000..57d0d2511dbc1
--- /dev/null
+++ b/api/core/model_runtime/model_providers/zhipuai/zhipuai_sdk/types/fine_tuning/models/__init__.py
@@ -0,0 +1 @@
+from .fine_tuned_models import FineTunedModelsStatus
diff --git a/api/core/model_runtime/model_providers/zhipuai/zhipuai_sdk/types/fine_tuning/models/fine_tuned_models.py b/api/core/model_runtime/model_providers/zhipuai/zhipuai_sdk/types/fine_tuning/models/fine_tuned_models.py
new file mode 100644
index 0000000000000..b286a5b5774d3
--- /dev/null
+++ b/api/core/model_runtime/model_providers/zhipuai/zhipuai_sdk/types/fine_tuning/models/fine_tuned_models.py
@@ -0,0 +1,13 @@
+from typing import ClassVar
+
+from ....core import PYDANTIC_V2, BaseModel, ConfigDict
+
+__all__ = ["FineTunedModelsStatus"]
+
+
+class FineTunedModelsStatus(BaseModel):
+    if PYDANTIC_V2:
+        model_config: ClassVar[ConfigDict] = ConfigDict(extra="allow", protected_namespaces=())
+    request_id: str  # 请求id
+    model_name: str  # 模型名称
+    delete_status: str  # 删除状态 deleting（删除中）, deleted （已删除）
diff --git a/api/core/model_runtime/model_providers/zhipuai/zhipuai_sdk/types/image.py b/api/core/model_runtime/model_providers/zhipuai/zhipuai_sdk/types/image.py
index b352ce0954ad5..3bcad0acabd21 100644
--- a/api/core/model_runtime/model_providers/zhipuai/zhipuai_sdk/types/image.py
+++ b/api/core/model_runtime/model_providers/zhipuai/zhipuai_sdk/types/image.py
@@ -2,7 +2,7 @@
 
 from typing import Optional
 
-from pydantic import BaseModel
+from ..core import BaseModel
 
 __all__ = ["GeneratedImage", "ImagesResponded"]
 
diff --git a/api/core/model_runtime/model_providers/zhipuai/zhipuai_sdk/types/knowledge/__init__.py b/api/core/model_runtime/model_providers/zhipuai/zhipuai_sdk/types/knowledge/__init__.py
new file mode 100644
index 0000000000000..8c81d703e214a
--- /dev/null
+++ b/api/core/model_runtime/model_providers/zhipuai/zhipuai_sdk/types/knowledge/__init__.py
@@ -0,0 +1,8 @@
+from .knowledge import KnowledgeInfo
+from .knowledge_used import KnowledgeStatistics, KnowledgeUsed
+
+__all__ = [
+    "KnowledgeInfo",
+    "KnowledgeStatistics",
+    "KnowledgeUsed",
+]
diff --git a/api/core/model_runtime/model_providers/zhipuai/zhipuai_sdk/types/knowledge/document/__init__.py b/api/core/model_runtime/model_providers/zhipuai/zhipuai_sdk/types/knowledge/document/__init__.py
new file mode 100644
index 0000000000000..32e23e6dab307
--- /dev/null
+++ b/api/core/model_runtime/model_providers/zhipuai/zhipuai_sdk/types/knowledge/document/__init__.py
@@ -0,0 +1,8 @@
+from .document import DocumentData, DocumentFailedInfo, DocumentObject, DocumentSuccessinfo
+
+__all__ = [
+    "DocumentData",
+    "DocumentObject",
+    "DocumentSuccessinfo",
+    "DocumentFailedInfo",
+]
diff --git a/api/core/model_runtime/model_providers/zhipuai/zhipuai_sdk/types/knowledge/document/document.py b/api/core/model_runtime/model_providers/zhipuai/zhipuai_sdk/types/knowledge/document/document.py
new file mode 100644
index 0000000000000..b9a1646391ece
--- /dev/null
+++ b/api/core/model_runtime/model_providers/zhipuai/zhipuai_sdk/types/knowledge/document/document.py
@@ -0,0 +1,51 @@
+from typing import Optional
+
+from ....core import BaseModel
+
+__all__ = ["DocumentData", "DocumentObject", "DocumentSuccessinfo", "DocumentFailedInfo"]
+
+
+class DocumentSuccessinfo(BaseModel):
+    documentId: Optional[str] = None
+    """文件id"""
+    filename: Optional[str] = None
+    """文件名称"""
+
+
+class DocumentFailedInfo(BaseModel):
+    failReason: Optional[str] = None
+    """上传失败的原因，包括：文件格式不支持、文件大小超出限制、知识库容量已满、容量上限为 50 万字。"""
+    filename: Optional[str] = None
+    """文件名称"""
+    documentId: Optional[str] = None
+    """知识库id"""
+
+
+class DocumentObject(BaseModel):
+    """文档信息"""
+
+    successInfos: Optional[list[DocumentSuccessinfo]] = None
+    """上传成功的文件信息"""
+    failedInfos: Optional[list[DocumentFailedInfo]] = None
+    """上传失败的文件信息"""
+
+
+class DocumentDataFailInfo(BaseModel):
+    """失败原因"""
+
+    embedding_code: Optional[int] = (
+        None  # 失败码 10001：知识不可用，知识库空间已达上限 10002：知识不可用，知识库空间已达上限(字数超出限制)
+    )
+    embedding_msg: Optional[str] = None  # 失败原因
+
+
+class DocumentData(BaseModel):
+    id: str = None  # 知识唯一id
+    custom_separator: list[str] = None  # 切片规则
+    sentence_size: str = None  # 切片大小
+    length: int = None  # 文件大小（字节）
+    word_num: int = None  # 文件字数
+    name: str = None  # 文件名
+    url: str = None  # 文件下载链接
+    embedding_stat: int = None  # 0:向量化中 1:向量化完成 2:向量化失败
+    failInfo: Optional[DocumentDataFailInfo] = None  # 失败原因 向量化失败embedding_stat=2的时候 会有此值
diff --git a/api/core/model_runtime/model_providers/zhipuai/zhipuai_sdk/types/knowledge/document/document_edit_params.py b/api/core/model_runtime/model_providers/zhipuai/zhipuai_sdk/types/knowledge/document/document_edit_params.py
new file mode 100644
index 0000000000000..509cb3a451af5
--- /dev/null
+++ b/api/core/model_runtime/model_providers/zhipuai/zhipuai_sdk/types/knowledge/document/document_edit_params.py
@@ -0,0 +1,29 @@
+from typing import Optional, TypedDict
+
+__all__ = ["DocumentEditParams"]
+
+
+class DocumentEditParams(TypedDict):
+    """
+    知识参数类型定义
+
+    Attributes:
+        id (str): 知识ID
+        knowledge_type (int): 知识类型:
+                        1:文章知识: 支持pdf,url,docx
+                        2.问答知识-文档:  支持pdf,url,docx
+                        3.问答知识-表格:  支持xlsx
+                        4.商品库-表格:  支持xlsx
+                        5.自定义:  支持pdf,url,docx
+        custom_separator (Optional[List[str]]): 当前知识类型为自定义(knowledge_type=5)时的切片规则，默认\n
+        sentence_size (Optional[int]): 当前知识类型为自定义(knowledge_type=5)时的切片字数，取值范围: 20-2000，默认300
+        callback_url (Optional[str]): 回调地址
+        callback_header (Optional[dict]): 回调时携带的header
+    """
+
+    id: str
+    knowledge_type: int
+    custom_separator: Optional[list[str]]
+    sentence_size: Optional[int]
+    callback_url: Optional[str]
+    callback_header: Optional[dict[str, str]]
diff --git a/api/core/model_runtime/model_providers/zhipuai/zhipuai_sdk/types/knowledge/document/document_list_params.py b/api/core/model_runtime/model_providers/zhipuai/zhipuai_sdk/types/knowledge/document/document_list_params.py
new file mode 100644
index 0000000000000..910c8c045e1b9
--- /dev/null
+++ b/api/core/model_runtime/model_providers/zhipuai/zhipuai_sdk/types/knowledge/document/document_list_params.py
@@ -0,0 +1,26 @@
+from __future__ import annotations
+
+from typing import Optional
+
+from typing_extensions import TypedDict
+
+
+class DocumentListParams(TypedDict, total=False):
+    """
+    文件查询参数类型定义
+
+    Attributes:
+        purpose (Optional[str]): 文件用途
+        knowledge_id (Optional[str]): 当文件用途为 retrieval 时，需要提供查询的知识库ID
+        page (Optional[int]): 页，默认1
+        limit (Optional[int]): 查询文件列表数，默认10
+        after (Optional[str]): 查询指定fileID之后的文件列表（当文件用途为 fine-tune 时需要）
+        order (Optional[str]): 排序规则，可选值['desc', 'asc']，默认desc（当文件用途为 fine-tune 时需要）
+    """
+
+    purpose: Optional[str]
+    knowledge_id: Optional[str]
+    page: Optional[int]
+    limit: Optional[int]
+    after: Optional[str]
+    order: Optional[str]
diff --git a/api/core/model_runtime/model_providers/zhipuai/zhipuai_sdk/types/knowledge/document/document_list_resp.py b/api/core/model_runtime/model_providers/zhipuai/zhipuai_sdk/types/knowledge/document/document_list_resp.py
new file mode 100644
index 0000000000000..acae4fad9ff36
--- /dev/null
+++ b/api/core/model_runtime/model_providers/zhipuai/zhipuai_sdk/types/knowledge/document/document_list_resp.py
@@ -0,0 +1,11 @@
+from __future__ import annotations
+
+from ....core import BaseModel
+from . import DocumentData
+
+__all__ = ["DocumentPage"]
+
+
+class DocumentPage(BaseModel):
+    list: list[DocumentData]
+    object: str
diff --git a/api/core/model_runtime/model_providers/zhipuai/zhipuai_sdk/types/knowledge/knowledge.py b/api/core/model_runtime/model_providers/zhipuai/zhipuai_sdk/types/knowledge/knowledge.py
new file mode 100644
index 0000000000000..bc6f159eb211e
--- /dev/null
+++ b/api/core/model_runtime/model_providers/zhipuai/zhipuai_sdk/types/knowledge/knowledge.py
@@ -0,0 +1,21 @@
+from typing import Optional
+
+from ...core import BaseModel
+
+__all__ = ["KnowledgeInfo"]
+
+
+class KnowledgeInfo(BaseModel):
+    id: Optional[str] = None
+    """知识库唯一 id"""
+    embedding_id: Optional[str] = (
+        None  # 知识库绑定的向量化模型 见模型列表 [内部服务开放接口文档](https://lslfd0slxc.feishu.cn/docx/YauWdbBiMopV0FxB7KncPWCEn8f#H15NduiQZo3ugmxnWQFcfAHpnQ4)
+    )
+    name: Optional[str] = None  # 知识库名称 100字限制
+    customer_identifier: Optional[str] = None  # 用户标识 长度32位以内
+    description: Optional[str] = None  # 知识库描述 500字限制
+    background: Optional[str] = None  # 背景颜色（给枚举）'blue', 'red', 'orange', 'purple', 'sky'
+    icon: Optional[str] = (
+        None  # 知识库图标（给枚举） question: 问号、book: 书籍、seal: 印章、wrench: 扳手、tag: 标签、horn: 喇叭、house: 房子  # noqa: E501
+    )
+    bucket_id: Optional[str] = None  # 桶id 限制32位
diff --git a/api/core/model_runtime/model_providers/zhipuai/zhipuai_sdk/types/knowledge/knowledge_create_params.py b/api/core/model_runtime/model_providers/zhipuai/zhipuai_sdk/types/knowledge/knowledge_create_params.py
new file mode 100644
index 0000000000000..c3da201727c34
--- /dev/null
+++ b/api/core/model_runtime/model_providers/zhipuai/zhipuai_sdk/types/knowledge/knowledge_create_params.py
@@ -0,0 +1,30 @@
+from __future__ import annotations
+
+from typing import Literal, Optional
+
+from typing_extensions import TypedDict
+
+__all__ = ["KnowledgeBaseParams"]
+
+
+class KnowledgeBaseParams(TypedDict):
+    """
+    知识库参数类型定义
+
+    Attributes:
+        embedding_id (int): 知识库绑定的向量化模型ID
+        name (str): 知识库名称，限制100字
+        customer_identifier (Optional[str]): 用户标识，长度32位以内
+        description (Optional[str]): 知识库描述，限制500字
+        background (Optional[Literal['blue', 'red', 'orange', 'purple', 'sky']]): 背景颜色
+        icon (Optional[Literal['question', 'book', 'seal', 'wrench', 'tag', 'horn', 'house']]): 知识库图标
+        bucket_id (Optional[str]): 桶ID，限制32位
+    """
+
+    embedding_id: int
+    name: str
+    customer_identifier: Optional[str]
+    description: Optional[str]
+    background: Optional[Literal["blue", "red", "orange", "purple", "sky"]] = None
+    icon: Optional[Literal["question", "book", "seal", "wrench", "tag", "horn", "house"]] = None
+    bucket_id: Optional[str]
diff --git a/api/core/model_runtime/model_providers/zhipuai/zhipuai_sdk/types/knowledge/knowledge_list_params.py b/api/core/model_runtime/model_providers/zhipuai/zhipuai_sdk/types/knowledge/knowledge_list_params.py
new file mode 100644
index 0000000000000..a221b28e4603b
--- /dev/null
+++ b/api/core/model_runtime/model_providers/zhipuai/zhipuai_sdk/types/knowledge/knowledge_list_params.py
@@ -0,0 +1,15 @@
+from __future__ import annotations
+
+from typing_extensions import TypedDict
+
+__all__ = ["KnowledgeListParams"]
+
+
+class KnowledgeListParams(TypedDict, total=False):
+    page: int = 1
+    """ 页码，默认 1，第一页
+    """
+
+    size: int = 10
+    """每页数量 默认10
+    """
diff --git a/api/core/model_runtime/model_providers/zhipuai/zhipuai_sdk/types/knowledge/knowledge_list_resp.py b/api/core/model_runtime/model_providers/zhipuai/zhipuai_sdk/types/knowledge/knowledge_list_resp.py
new file mode 100644
index 0000000000000..e462eddc550d6
--- /dev/null
+++ b/api/core/model_runtime/model_providers/zhipuai/zhipuai_sdk/types/knowledge/knowledge_list_resp.py
@@ -0,0 +1,11 @@
+from __future__ import annotations
+
+from ...core import BaseModel
+from . import KnowledgeInfo
+
+__all__ = ["KnowledgePage"]
+
+
+class KnowledgePage(BaseModel):
+    list: list[KnowledgeInfo]
+    object: str
diff --git a/api/core/model_runtime/model_providers/zhipuai/zhipuai_sdk/types/knowledge/knowledge_used.py b/api/core/model_runtime/model_providers/zhipuai/zhipuai_sdk/types/knowledge/knowledge_used.py
new file mode 100644
index 0000000000000..cfda7097026c5
--- /dev/null
+++ b/api/core/model_runtime/model_providers/zhipuai/zhipuai_sdk/types/knowledge/knowledge_used.py
@@ -0,0 +1,21 @@
+from typing import Optional
+
+from ...core import BaseModel
+
+__all__ = ["KnowledgeStatistics", "KnowledgeUsed"]
+
+
+class KnowledgeStatistics(BaseModel):
+    """
+    使用量统计
+    """
+
+    word_num: Optional[int] = None
+    length: Optional[int] = None
+
+
+class KnowledgeUsed(BaseModel):
+    used: Optional[KnowledgeStatistics] = None
+    """已使用量"""
+    total: Optional[KnowledgeStatistics] = None
+    """知识库总量"""
diff --git a/api/core/model_runtime/model_providers/zhipuai/zhipuai_sdk/types/sensitive_word_check/__init__.py b/api/core/model_runtime/model_providers/zhipuai/zhipuai_sdk/types/sensitive_word_check/__init__.py
new file mode 100644
index 0000000000000..c9bd60419ce60
--- /dev/null
+++ b/api/core/model_runtime/model_providers/zhipuai/zhipuai_sdk/types/sensitive_word_check/__init__.py
@@ -0,0 +1,3 @@
+from .sensitive_word_check import SensitiveWordCheckRequest
+
+__all__ = ["SensitiveWordCheckRequest"]
diff --git a/api/core/model_runtime/model_providers/zhipuai/zhipuai_sdk/types/sensitive_word_check/sensitive_word_check.py b/api/core/model_runtime/model_providers/zhipuai/zhipuai_sdk/types/sensitive_word_check/sensitive_word_check.py
new file mode 100644
index 0000000000000..0c37d99e65329
--- /dev/null
+++ b/api/core/model_runtime/model_providers/zhipuai/zhipuai_sdk/types/sensitive_word_check/sensitive_word_check.py
@@ -0,0 +1,14 @@
+from typing import Optional
+
+from typing_extensions import TypedDict
+
+
+class SensitiveWordCheckRequest(TypedDict, total=False):
+    type: Optional[str]
+    """敏感词类型，当前仅支持ALL"""
+    status: Optional[str]
+    """敏感词启用禁用状态
+        启用：ENABLE
+        禁用：DISABLE
+       备注：默认开启敏感词校验，如果要关闭敏感词校验，需联系商务获取对应权限，否则敏感词禁用不生效。
+    """
diff --git a/api/core/model_runtime/model_providers/zhipuai/zhipuai_sdk/types/tools/__init__.py b/api/core/model_runtime/model_providers/zhipuai/zhipuai_sdk/types/tools/__init__.py
new file mode 100644
index 0000000000000..62f77344eee56
--- /dev/null
+++ b/api/core/model_runtime/model_providers/zhipuai/zhipuai_sdk/types/tools/__init__.py
@@ -0,0 +1,9 @@
+from .web_search import (
+    SearchIntent,
+    SearchRecommend,
+    SearchResult,
+    WebSearch,
+)
+from .web_search_chunk import WebSearchChunk
+
+__all__ = ["WebSearch", "SearchIntent", "SearchResult", "SearchRecommend", "WebSearchChunk"]
diff --git a/api/core/model_runtime/model_providers/zhipuai/zhipuai_sdk/types/tools/tools_web_search_params.py b/api/core/model_runtime/model_providers/zhipuai/zhipuai_sdk/types/tools/tools_web_search_params.py
new file mode 100644
index 0000000000000..b3a3b26f07ee5
--- /dev/null
+++ b/api/core/model_runtime/model_providers/zhipuai/zhipuai_sdk/types/tools/tools_web_search_params.py
@@ -0,0 +1,35 @@
+from __future__ import annotations
+
+from typing import Optional, Union
+
+from typing_extensions import TypedDict
+
+__all__ = ["WebSearchParams"]
+
+
+class WebSearchParams(TypedDict):
+    """
+    工具名：web-search-pro参数类型定义
+
+    Attributes:
+        :param model: str, 模型名称
+        :param request_id: Optional[str], 请求ID
+        :param stream: Optional[bool], 是否流式
+        :param messages: Union[str, List[str], List[int], object, None],
+                        包含历史对话上下文的内容，按照 {"role": "user", "content": "你好"} 的json 数组形式进行传参
+                        当前版本仅支持 User Message 单轮对话，工具会理解User Message并进行搜索，
+                        请尽可能传入不带指令格式的用户原始提问，以提高搜索准确率。
+        :param scope: Optional[str], 指定搜索范围，全网、学术等，默认全网
+        :param location: Optional[str], 指定搜索用户地区 location 提高相关性
+        :param recent_days: Optional[int],支持指定返回 N 天（1-30）更新的搜索结果
+
+
+    """
+
+    model: str
+    request_id: Optional[str]
+    stream: Optional[bool]
+    messages: Union[str, list[str], list[int], object, None]
+    scope: Optional[str] = None
+    location: Optional[str] = None
+    recent_days: Optional[int] = None
diff --git a/api/core/model_runtime/model_providers/zhipuai/zhipuai_sdk/types/tools/web_search.py b/api/core/model_runtime/model_providers/zhipuai/zhipuai_sdk/types/tools/web_search.py
new file mode 100644
index 0000000000000..ac9fa3821e979
--- /dev/null
+++ b/api/core/model_runtime/model_providers/zhipuai/zhipuai_sdk/types/tools/web_search.py
@@ -0,0 +1,71 @@
+from typing import Optional
+
+from ...core import BaseModel
+
+__all__ = [
+    "WebSearch",
+    "SearchIntent",
+    "SearchResult",
+    "SearchRecommend",
+]
+
+
+class SearchIntent(BaseModel):
+    index: int
+    # 搜索轮次，默认为 0
+    query: str
+    # 搜索优化 query
+    intent: str
+    # 判断的意图类型
+    keywords: str
+    # 搜索关键词
+
+
+class SearchResult(BaseModel):
+    index: int
+    # 搜索轮次，默认为 0
+    title: str
+    # 标题
+    link: str
+    # 链接
+    content: str
+    # 内容
+    icon: str
+    # 图标
+    media: str
+    # 来源媒体
+    refer: str
+    # 角标序号 [ref_1]
+
+
+class SearchRecommend(BaseModel):
+    index: int
+    # 搜索轮次，默认为 0
+    query: str
+    # 推荐query
+
+
+class WebSearchMessageToolCall(BaseModel):
+    id: str
+    search_intent: Optional[SearchIntent]
+    search_result: Optional[SearchResult]
+    search_recommend: Optional[SearchRecommend]
+    type: str
+
+
+class WebSearchMessage(BaseModel):
+    role: str
+    tool_calls: Optional[list[WebSearchMessageToolCall]] = None
+
+
+class WebSearchChoice(BaseModel):
+    index: int
+    finish_reason: str
+    message: WebSearchMessage
+
+
+class WebSearch(BaseModel):
+    created: Optional[int] = None
+    choices: list[WebSearchChoice]
+    request_id: Optional[str] = None
+    id: Optional[str] = None
diff --git a/api/core/model_runtime/model_providers/zhipuai/zhipuai_sdk/types/tools/web_search_chunk.py b/api/core/model_runtime/model_providers/zhipuai/zhipuai_sdk/types/tools/web_search_chunk.py
new file mode 100644
index 0000000000000..7fb0e02bb5871
--- /dev/null
+++ b/api/core/model_runtime/model_providers/zhipuai/zhipuai_sdk/types/tools/web_search_chunk.py
@@ -0,0 +1,33 @@
+from typing import Optional
+
+from ...core import BaseModel
+from .web_search import SearchIntent, SearchRecommend, SearchResult
+
+__all__ = ["WebSearchChunk"]
+
+
+class ChoiceDeltaToolCall(BaseModel):
+    index: int
+    id: Optional[str] = None
+
+    search_intent: Optional[SearchIntent] = None
+    search_result: Optional[SearchResult] = None
+    search_recommend: Optional[SearchRecommend] = None
+    type: Optional[str] = None
+
+
+class ChoiceDelta(BaseModel):
+    role: Optional[str] = None
+    tool_calls: Optional[list[ChoiceDeltaToolCall]] = None
+
+
+class Choice(BaseModel):
+    delta: ChoiceDelta
+    finish_reason: Optional[str] = None
+    index: int
+
+
+class WebSearchChunk(BaseModel):
+    id: Optional[str] = None
+    choices: list[Choice]
+    created: Optional[int] = None
diff --git a/api/core/model_runtime/model_providers/zhipuai/zhipuai_sdk/types/video/__init__.py b/api/core/model_runtime/model_providers/zhipuai/zhipuai_sdk/types/video/__init__.py
new file mode 100644
index 0000000000000..b14072b1a771a
--- /dev/null
+++ b/api/core/model_runtime/model_providers/zhipuai/zhipuai_sdk/types/video/__init__.py
@@ -0,0 +1,3 @@
+from .video_object import VideoObject, VideoResult
+
+__all__ = ["VideoObject", "VideoResult"]
diff --git a/api/core/model_runtime/model_providers/zhipuai/zhipuai_sdk/types/video/video_create_params.py b/api/core/model_runtime/model_providers/zhipuai/zhipuai_sdk/types/video/video_create_params.py
new file mode 100644
index 0000000000000..f5489d708e722
--- /dev/null
+++ b/api/core/model_runtime/model_providers/zhipuai/zhipuai_sdk/types/video/video_create_params.py
@@ -0,0 +1,27 @@
+from __future__ import annotations
+
+from typing import Optional
+
+from typing_extensions import TypedDict
+
+__all__ = ["VideoCreateParams"]
+
+from ..sensitive_word_check import SensitiveWordCheckRequest
+
+
+class VideoCreateParams(TypedDict, total=False):
+    model: str
+    """模型编码"""
+    prompt: str
+    """所需视频的文本描述"""
+    image_url: str
+    """所需视频的文本描述"""
+    sensitive_word_check: Optional[SensitiveWordCheckRequest]
+    """支持 URL 或者 Base64、传入 image 奖进行图生视频
+     * 图片格式：
+     *   图片大小："""
+    request_id: str
+    """由用户端传参，需保证唯一性；用于区分每次请求的唯一标识，用户端不传时平台会默认生成。"""
+
+    user_id: str
+    """用户端。"""
diff --git a/api/core/model_runtime/model_providers/zhipuai/zhipuai_sdk/types/video/video_object.py b/api/core/model_runtime/model_providers/zhipuai/zhipuai_sdk/types/video/video_object.py
new file mode 100644
index 0000000000000..85c3844d8a791
--- /dev/null
+++ b/api/core/model_runtime/model_providers/zhipuai/zhipuai_sdk/types/video/video_object.py
@@ -0,0 +1,30 @@
+from typing import Optional
+
+from ...core import BaseModel
+
+__all__ = ["VideoObject", "VideoResult"]
+
+
+class VideoResult(BaseModel):
+    url: str
+    """视频url"""
+    cover_image_url: str
+    """预览图"""
+
+
+class VideoObject(BaseModel):
+    id: Optional[str] = None
+    """智谱 AI 开放平台生成的任务订单号，调用请求结果接口时请使用此订单号"""
+
+    model: str
+    """模型名称"""
+
+    video_result: list[VideoResult]
+    """视频生成结果"""
+
+    task_status: str
+    """处理状态，PROCESSING（处理中），SUCCESS（成功），FAIL（失败）
+    注：处理中状态需通过查询获取结果"""
+
+    request_id: str
+    """用户在客户端请求时提交的任务编号或者平台生成的任务编号"""
diff --git a/api/core/tools/provider/builtin/cogview/tools/cogview3.py b/api/core/tools/provider/builtin/cogview/tools/cogview3.py
index 9039708588df1..085084ca38355 100644
--- a/api/core/tools/provider/builtin/cogview/tools/cogview3.py
+++ b/api/core/tools/provider/builtin/cogview/tools/cogview3.py
@@ -21,15 +21,22 @@ def _invoke(
         )
         size_mapping = {
             "square": "1024x1024",
-            "vertical": "1024x1792",
-            "horizontal": "1792x1024",
+            "vertical_768": "768x1344",
+            "vertical_864": "864x1152",
+            "horizontal_1344": "1344x768",
+            "horizontal_1152": "1152x864",
+            "widescreen_1440": "1440x720",
+            "tallscreen_720": "720x1440",
         }
         # prompt
         prompt = tool_parameters.get("prompt", "")
         if not prompt:
             return self.create_text_message("Please input prompt")
-        # get size
-        size = size_mapping[tool_parameters.get("size", "square")]
+        # get size key
+        size_key = tool_parameters.get("size", "square")
+        # cogview-3-plus get size
+        if size_key != "cogview_3":
+            size = size_mapping[size_key]
         # get n
         n = tool_parameters.get("n", 1)
         # get quality
@@ -43,16 +50,29 @@ def _invoke(
         # set extra body
         seed_id = tool_parameters.get("seed_id", self._generate_random_id(8))
         extra_body = {"seed": seed_id}
-        response = client.images.generations(
-            prompt=prompt,
-            model="cogview-3",
-            size=size,
-            n=n,
-            extra_body=extra_body,
-            style=style,
-            quality=quality,
-            response_format="b64_json",
-        )
+        # cogview-3-plus
+        if size_key != "cogview_3":
+            response = client.images.generations(
+                prompt=prompt,
+                model="cogview-3-plus",
+                size=size,
+                n=n,
+                extra_body=extra_body,
+                style=style,
+                quality=quality,
+                response_format="b64_json",
+            )
+        # cogview-3
+        else:
+            response = client.images.generations(
+                prompt=prompt,
+                model="cogview-3",
+                n=n,
+                extra_body=extra_body,
+                style=style,
+                quality=quality,
+                response_format="b64_json",
+            )
         result = []
         for image in response.data:
             result.append(self.create_image_message(image=image.url))
diff --git a/api/core/tools/provider/builtin/cogview/tools/cogview3.yaml b/api/core/tools/provider/builtin/cogview/tools/cogview3.yaml
index 1de3f599b6ac0..9ab5c2729bf7a 100644
--- a/api/core/tools/provider/builtin/cogview/tools/cogview3.yaml
+++ b/api/core/tools/provider/builtin/cogview/tools/cogview3.yaml
@@ -42,21 +42,46 @@ parameters:
       pt_BR: Image size
     form: form
     options:
+      - value: cogview_3
+        label:
+          en_US: Square_cogview_3(1024x1024)
+          zh_Hans: 方_cogview_3(1024x1024)
+          pt_BR: Square_cogview_3(1024x1024)
       - value: square
         label:
-          en_US: Squre(1024x1024)
+          en_US: Square(1024x1024)
           zh_Hans: 方(1024x1024)
-          pt_BR: Squre(1024x1024)
-      - value: vertical
+          pt_BR: Square(1024x1024)
+      - value: vertical_768
+        label:
+          en_US: Vertical(768x1344)
+          zh_Hans: 竖屏(768x1344)
+          pt_BR: Vertical(768x1344)
+      - value: vertical_864
+        label:
+          en_US: Vertical(864x1152)
+          zh_Hans: 竖屏(864x1152)
+          pt_BR: Vertical(864x1152)
+      - value: horizontal_1344
+        label:
+          en_US: Horizontal(1344x768)
+          zh_Hans: 横屏(1344x768)
+          pt_BR: Horizontal(1344x768)
+      - value: horizontal_1152
+        label:
+          en_US: Horizontal(1152x864)
+          zh_Hans: 横屏(1152x864)
+          pt_BR: Horizontal(1152x864)
+      - value: widescreen_1440
         label:
-          en_US: Vertical(1024x1792)
-          zh_Hans: 竖屏(1024x1792)
-          pt_BR: Vertical(1024x1792)
-      - value: horizontal
+          en_US: Widescreen(1440x720)
+          zh_Hans: 宽屏(1440x720)
+          pt_BR: Widescreen(1440x720)
+      - value: tallscreen_720
         label:
-          en_US: Horizontal(1792x1024)
-          zh_Hans: 横屏(1792x1024)
-          pt_BR: Horizontal(1792x1024)
+          en_US: Tallscreen(720x1440)
+          zh_Hans: 高屏(720x1440)
+          pt_BR: Tallscreen(720x1440)
     default: square
   - name: n
     type: number

From c9f1e18df1e400ce5c6599d7a528ce4d50e1bdea Mon Sep 17 00:00:00 2001
From: AAEE86 <33052466+AAEE86@users.noreply.github.com>
Date: Sun, 22 Sep 2024 10:14:33 +0800
Subject: [PATCH 017/235] Add model parameter translation (#8509)
MIME-Version: 1.0
Content-Type: text/plain; charset=UTF-8
Content-Transfer-Encoding: 8bit

Co-authored-by: swingchen01 <swings@126.com>
Co-authored-by: 陈长君 <chenchangjun@shuwen.com>
---
 .../model_providers/ollama/llm/llm.py         | 83 ++++++++++++-------
 .../openai_api_compatible/llm/llm.py          | 35 ++++++--
 2 files changed, 82 insertions(+), 36 deletions(-)

diff --git a/api/core/model_runtime/model_providers/ollama/llm/llm.py b/api/core/model_runtime/model_providers/ollama/llm/llm.py
index 1ed77a2ee8ed4..ff732e69259e2 100644
--- a/api/core/model_runtime/model_providers/ollama/llm/llm.py
+++ b/api/core/model_runtime/model_providers/ollama/llm/llm.py
@@ -472,12 +472,13 @@ def get_customizable_model_schema(self, model: str, credentials: dict) -> AIMode
                 ParameterRule(
                     name=DefaultParameterName.TEMPERATURE.value,
                     use_template=DefaultParameterName.TEMPERATURE.value,
-                    label=I18nObject(en_US="Temperature"),
+                    label=I18nObject(en_US="Temperature", zh_Hans="温度"),
                     type=ParameterType.FLOAT,
                     help=I18nObject(
                         en_US="The temperature of the model. "
                         "Increasing the temperature will make the model answer "
-                        "more creatively. (Default: 0.8)"
+                        "more creatively. (Default: 0.8)",
+                        zh_Hans="模型的温度。增加温度将使模型的回答更具创造性。（默认值：0.8）",
                     ),
                     default=0.1,
                     min=0,
@@ -486,12 +487,13 @@ def get_customizable_model_schema(self, model: str, credentials: dict) -> AIMode
                 ParameterRule(
                     name=DefaultParameterName.TOP_P.value,
                     use_template=DefaultParameterName.TOP_P.value,
-                    label=I18nObject(en_US="Top P"),
+                    label=I18nObject(en_US="Top P", zh_Hans="Top P"),
                     type=ParameterType.FLOAT,
                     help=I18nObject(
                         en_US="Works together with top-k. A higher value (e.g., 0.95) will lead to "
                         "more diverse text, while a lower value (e.g., 0.5) will generate more "
-                        "focused and conservative text. (Default: 0.9)"
+                        "focused and conservative text. (Default: 0.9)",
+                        zh_Hans="与top-k一起工作。较高的值（例如，0.95）会导致生成更多样化的文本，而较低的值（例如，0.5）会生成更专注和保守的文本。（默认值：0.9）",
                     ),
                     default=0.9,
                     min=0,
@@ -499,12 +501,13 @@ def get_customizable_model_schema(self, model: str, credentials: dict) -> AIMode
                 ),
                 ParameterRule(
                     name="top_k",
-                    label=I18nObject(en_US="Top K"),
+                    label=I18nObject(en_US="Top K", zh_Hans="Top K"),
                     type=ParameterType.INT,
                     help=I18nObject(
                         en_US="Reduces the probability of generating nonsense. "
                         "A higher value (e.g. 100) will give more diverse answers, "
-                        "while a lower value (e.g. 10) will be more conservative. (Default: 40)"
+                        "while a lower value (e.g. 10) will be more conservative. (Default: 40)",
+                        zh_Hans="减少生成无意义内容的可能性。较高的值（例如100）将提供更多样化的答案，而较低的值（例如10）将更为保守。（默认值：40）",
                     ),
                     min=1,
                     max=100,
@@ -516,7 +519,8 @@ def get_customizable_model_schema(self, model: str, credentials: dict) -> AIMode
                     help=I18nObject(
                         en_US="Sets how strongly to penalize repetitions. "
                         "A higher value (e.g., 1.5) will penalize repetitions more strongly, "
-                        "while a lower value (e.g., 0.9) will be more lenient. (Default: 1.1)"
+                        "while a lower value (e.g., 0.9) will be more lenient. (Default: 1.1)",
+                        zh_Hans="设置对重复内容的惩罚强度。一个较高的值（例如，1.5）会更强地惩罚重复内容，而一个较低的值（例如，0.9）则会相对宽容。（默认值：1.1）",
                     ),
                     min=-2,
                     max=2,
@@ -524,11 +528,12 @@ def get_customizable_model_schema(self, model: str, credentials: dict) -> AIMode
                 ParameterRule(
                     name="num_predict",
                     use_template="max_tokens",
-                    label=I18nObject(en_US="Num Predict"),
+                    label=I18nObject(en_US="Num Predict", zh_Hans="最大令牌数预测"),
                     type=ParameterType.INT,
                     help=I18nObject(
                         en_US="Maximum number of tokens to predict when generating text. "
-                        "(Default: 128, -1 = infinite generation, -2 = fill context)"
+                        "(Default: 128, -1 = infinite generation, -2 = fill context)",
+                        zh_Hans="生成文本时预测的最大令牌数。（默认值：128，-1 = 无限生成，-2 = 填充上下文）",
                     ),
                     default=(512 if int(credentials.get("max_tokens", 4096)) >= 768 else 128),
                     min=-2,
@@ -536,121 +541,137 @@ def get_customizable_model_schema(self, model: str, credentials: dict) -> AIMode
                 ),
                 ParameterRule(
                     name="mirostat",
-                    label=I18nObject(en_US="Mirostat sampling"),
+                    label=I18nObject(en_US="Mirostat sampling", zh_Hans="Mirostat 采样"),
                     type=ParameterType.INT,
                     help=I18nObject(
                         en_US="Enable Mirostat sampling for controlling perplexity. "
-                        "(default: 0, 0 = disabled, 1 = Mirostat, 2 = Mirostat 2.0)"
+                        "(default: 0, 0 = disabled, 1 = Mirostat, 2 = Mirostat 2.0)",
+                        zh_Hans="启用 Mirostat 采样以控制困惑度。"
+                        "（默认值：0，0 = 禁用，1 = Mirostat，2 = Mirostat 2.0）",
                     ),
                     min=0,
                     max=2,
                 ),
                 ParameterRule(
                     name="mirostat_eta",
-                    label=I18nObject(en_US="Mirostat Eta"),
+                    label=I18nObject(en_US="Mirostat Eta", zh_Hans="学习率"),
                     type=ParameterType.FLOAT,
                     help=I18nObject(
                         en_US="Influences how quickly the algorithm responds to feedback from "
                         "the generated text. A lower learning rate will result in slower adjustments, "
                         "while a higher learning rate will make the algorithm more responsive. "
-                        "(Default: 0.1)"
+                        "(Default: 0.1)",
+                        zh_Hans="影响算法对生成文本反馈响应的速度。较低的学习率会导致调整速度变慢，而较高的学习率会使得算法更加灵敏。（默认值：0.1）",
                     ),
                     precision=1,
                 ),
                 ParameterRule(
                     name="mirostat_tau",
-                    label=I18nObject(en_US="Mirostat Tau"),
+                    label=I18nObject(en_US="Mirostat Tau", zh_Hans="文本连贯度"),
                     type=ParameterType.FLOAT,
                     help=I18nObject(
                         en_US="Controls the balance between coherence and diversity of the output. "
-                        "A lower value will result in more focused and coherent text. (Default: 5.0)"
+                        "A lower value will result in more focused and coherent text. (Default: 5.0)",
+                        zh_Hans="控制输出的连贯性和多样性之间的平衡。较低的值会导致更专注和连贯的文本。（默认值：5.0）",
                     ),
                     precision=1,
                 ),
                 ParameterRule(
                     name="num_ctx",
-                    label=I18nObject(en_US="Size of context window"),
+                    label=I18nObject(en_US="Size of context window", zh_Hans="上下文窗口大小"),
                     type=ParameterType.INT,
                     help=I18nObject(
-                        en_US="Sets the size of the context window used to generate the next token. (Default: 2048)"
+                        en_US="Sets the size of the context window used to generate the next token. (Default: 2048)",
+                        zh_Hans="设置用于生成下一个标记的上下文窗口大小。（默认值：2048）",
                     ),
                     default=2048,
                     min=1,
                 ),
                 ParameterRule(
                     name="num_gpu",
-                    label=I18nObject(en_US="GPU Layers"),
+                    label=I18nObject(en_US="GPU Layers", zh_Hans="GPU 层数"),
                     type=ParameterType.INT,
                     help=I18nObject(
                         en_US="The number of layers to offload to the GPU(s). "
                         "On macOS it defaults to 1 to enable metal support, 0 to disable."
                         "As long as a model fits into one gpu it stays in one. "
-                        "It does not set the number of GPU(s). "
+                        "It does not set the number of GPU(s). ",
+                        zh_Hans="加载到 GPU 的层数。在 macOS 上，默认为 1 以启用 Metal 支持，设置为 0 则禁用。"
+                        "只要模型适合一个 GPU，它就保留在其中。它不设置 GPU 的数量。",
                     ),
                     min=-1,
                     default=1,
                 ),
                 ParameterRule(
                     name="num_thread",
-                    label=I18nObject(en_US="Num Thread"),
+                    label=I18nObject(en_US="Num Thread", zh_Hans="线程数"),
                     type=ParameterType.INT,
                     help=I18nObject(
                         en_US="Sets the number of threads to use during computation. "
                         "By default, Ollama will detect this for optimal performance. "
                         "It is recommended to set this value to the number of physical CPU cores "
-                        "your system has (as opposed to the logical number of cores)."
+                        "your system has (as opposed to the logical number of cores).",
+                        zh_Hans="设置计算过程中使用的线程数。默认情况下，Ollama会检测以获得最佳性能。建议将此值设置为系统拥有的物理CPU核心数（而不是逻辑核心数）。",
                     ),
                     min=1,
                 ),
                 ParameterRule(
                     name="repeat_last_n",
-                    label=I18nObject(en_US="Repeat last N"),
+                    label=I18nObject(en_US="Repeat last N", zh_Hans="回溯内容"),
                     type=ParameterType.INT,
                     help=I18nObject(
                         en_US="Sets how far back for the model to look back to prevent repetition. "
-                        "(Default: 64, 0 = disabled, -1 = num_ctx)"
+                        "(Default: 64, 0 = disabled, -1 = num_ctx)",
+                        zh_Hans="设置模型回溯多远的内容以防止重复。（默认值：64，0 = 禁用，-1 = num_ctx）",
                     ),
                     min=-1,
                 ),
                 ParameterRule(
                     name="tfs_z",
-                    label=I18nObject(en_US="TFS Z"),
+                    label=I18nObject(en_US="TFS Z", zh_Hans="减少标记影响"),
                     type=ParameterType.FLOAT,
                     help=I18nObject(
                         en_US="Tail free sampling is used to reduce the impact of less probable tokens "
                         "from the output. A higher value (e.g., 2.0) will reduce the impact more, "
-                        "while a value of 1.0 disables this setting. (default: 1)"
+                        "while a value of 1.0 disables this setting. (default: 1)",
+                        zh_Hans="用于减少输出中不太可能的标记的影响。较高的值（例如，2.0）会更多地减少这种影响，而1.0的值则会禁用此设置。（默认值：1）",
                     ),
                     precision=1,
                 ),
                 ParameterRule(
                     name="seed",
-                    label=I18nObject(en_US="Seed"),
+                    label=I18nObject(en_US="Seed", zh_Hans="随机数种子"),
                     type=ParameterType.INT,
                     help=I18nObject(
                         en_US="Sets the random number seed to use for generation. Setting this to "
                         "a specific number will make the model generate the same text for "
-                        "the same prompt. (Default: 0)"
+                        "the same prompt. (Default: 0)",
+                        zh_Hans="设置用于生成的随机数种子。将此设置为特定数字将使模型对相同的提示生成相同的文本。（默认值：0）",
                     ),
                 ),
                 ParameterRule(
                     name="keep_alive",
-                    label=I18nObject(en_US="Keep Alive"),
+                    label=I18nObject(en_US="Keep Alive", zh_Hans="模型存活时间"),
                     type=ParameterType.STRING,
                     help=I18nObject(
                         en_US="Sets how long the model is kept in memory after generating a response. "
                         "This must be a duration string with a unit (e.g., '10m' for 10 minutes or '24h' for 24 hours)."
                         " A negative number keeps the model loaded indefinitely, and '0' unloads the model"
                         " immediately after generating a response."
-                        " Valid time units are 's','m','h'. (Default: 5m)"
+                        " Valid time units are 's','m','h'. (Default: 5m)",
+                        zh_Hans="设置模型在生成响应后在内存中保留的时间。"
+                        "这必须是一个带有单位的持续时间字符串（例如，'10m' 表示10分钟，'24h' 表示24小时）。"
+                        "负数表示无限期地保留模型，'0'表示在生成响应后立即卸载模型。"
+                        "有效的时间单位有 's'（秒）、'm'（分钟）、'h'（小时）。（默认值：5m）",
                     ),
                 ),
                 ParameterRule(
                     name="format",
-                    label=I18nObject(en_US="Format"),
+                    label=I18nObject(en_US="Format", zh_Hans="返回格式"),
                     type=ParameterType.STRING,
                     help=I18nObject(
-                        en_US="the format to return a response in. Currently the only accepted value is json."
+                        en_US="the format to return a response in. Currently the only accepted value is json.",
+                        zh_Hans="返回响应的格式。目前唯一接受的值是json。",
                     ),
                     options=["json"],
                 ),
diff --git a/api/core/model_runtime/model_providers/openai_api_compatible/llm/llm.py b/api/core/model_runtime/model_providers/openai_api_compatible/llm/llm.py
index 5a8a754f72422..c2ffe653c8b70 100644
--- a/api/core/model_runtime/model_providers/openai_api_compatible/llm/llm.py
+++ b/api/core/model_runtime/model_providers/openai_api_compatible/llm/llm.py
@@ -205,7 +205,13 @@ def get_customizable_model_schema(self, model: str, credentials: dict) -> AIMode
             parameter_rules=[
                 ParameterRule(
                     name=DefaultParameterName.TEMPERATURE.value,
-                    label=I18nObject(en_US="Temperature"),
+                    label=I18nObject(en_US="Temperature", zh_Hans="温度"),
+                    help=I18nObject(
+                        en_US="Kernel sampling threshold. Used to determine the randomness of the results."
+                        "The higher the value, the stronger the randomness."
+                        "The higher the possibility of getting different answers to the same question.",
+                        zh_Hans="核采样阈值。用于决定结果随机性，取值越高随机性越强即相同的问题得到的不同答案的可能性越高。",
+                    ),
                     type=ParameterType.FLOAT,
                     default=float(credentials.get("temperature", 0.7)),
                     min=0,
@@ -214,7 +220,13 @@ def get_customizable_model_schema(self, model: str, credentials: dict) -> AIMode
                 ),
                 ParameterRule(
                     name=DefaultParameterName.TOP_P.value,
-                    label=I18nObject(en_US="Top P"),
+                    label=I18nObject(en_US="Top P", zh_Hans="Top P"),
+                    help=I18nObject(
+                        en_US="The probability threshold of the nucleus sampling method during the generation process."
+                        "The larger the value is, the higher the randomness of generation will be."
+                        "The smaller the value is, the higher the certainty of generation will be.",
+                        zh_Hans="生成过程中核采样方法概率阈值。取值越大，生成的随机性越高；取值越小，生成的确定性越高。",
+                    ),
                     type=ParameterType.FLOAT,
                     default=float(credentials.get("top_p", 1)),
                     min=0,
@@ -223,7 +235,12 @@ def get_customizable_model_schema(self, model: str, credentials: dict) -> AIMode
                 ),
                 ParameterRule(
                     name=DefaultParameterName.FREQUENCY_PENALTY.value,
-                    label=I18nObject(en_US="Frequency Penalty"),
+                    label=I18nObject(en_US="Frequency Penalty", zh_Hans="频率惩罚"),
+                    help=I18nObject(
+                        en_US="For controlling the repetition rate of words used by the model."
+                        "Increasing this can reduce the repetition of the same words in the model's output.",
+                        zh_Hans="用于控制模型已使用字词的重复率。 提高此项可以降低模型在输出中重复相同字词的重复度。",
+                    ),
                     type=ParameterType.FLOAT,
                     default=float(credentials.get("frequency_penalty", 0)),
                     min=-2,
@@ -231,7 +248,12 @@ def get_customizable_model_schema(self, model: str, credentials: dict) -> AIMode
                 ),
                 ParameterRule(
                     name=DefaultParameterName.PRESENCE_PENALTY.value,
-                    label=I18nObject(en_US="Presence Penalty"),
+                    label=I18nObject(en_US="Presence Penalty", zh_Hans="存在惩罚"),
+                    help=I18nObject(
+                        en_US="Used to control the repetition rate when generating models."
+                        "Increasing this can reduce the repetition rate of model generation.",
+                        zh_Hans="用于控制模型生成时的重复度。提高此项可以降低模型生成的重复度。",
+                    ),
                     type=ParameterType.FLOAT,
                     default=float(credentials.get("presence_penalty", 0)),
                     min=-2,
@@ -239,7 +261,10 @@ def get_customizable_model_schema(self, model: str, credentials: dict) -> AIMode
                 ),
                 ParameterRule(
                     name=DefaultParameterName.MAX_TOKENS.value,
-                    label=I18nObject(en_US="Max Tokens"),
+                    label=I18nObject(en_US="Max Tokens", zh_Hans="最大标记"),
+                    help=I18nObject(
+                        en_US="Maximum length of tokens for the model response.", zh_Hans="模型回答的tokens的最大长度。"
+                    ),
                     type=ParameterType.INT,
                     default=512,
                     min=1,

From 6c2fa8defc8a2c6a66be45ade7b13e2388fb4b69 Mon Sep 17 00:00:00 2001
From: HJY <1398145450@qq.com>
Date: Sun, 22 Sep 2024 10:14:43 +0800
Subject: [PATCH 018/235] fix: form input add tabIndex (#8478)

---
 web/app/activate/activateForm.tsx | 2 ++
 web/app/signin/normalForm.tsx     | 2 ++
 2 files changed, 4 insertions(+)

diff --git a/web/app/activate/activateForm.tsx b/web/app/activate/activateForm.tsx
index 3b1eed6f09a5d..8e9691b35476d 100644
--- a/web/app/activate/activateForm.tsx
+++ b/web/app/activate/activateForm.tsx
@@ -143,6 +143,7 @@ const ActivateForm = () => {
                     onChange={e => setName(e.target.value)}
                     placeholder={t('login.namePlaceholder') || ''}
                     className={'appearance-none block w-full rounded-lg pl-[14px] px-3 py-2 border border-gray-200 hover:border-gray-300 hover:shadow-sm focus:outline-none focus:ring-primary-500 focus:border-primary-500 placeholder-gray-400 caret-primary-600 sm:text-sm pr-10'}
+                    tabIndex={1}
                   />
                 </div>
               </div>
@@ -159,6 +160,7 @@ const ActivateForm = () => {
                     onChange={e => setPassword(e.target.value)}
                     placeholder={t('login.passwordPlaceholder') || ''}
                     className={'appearance-none block w-full rounded-lg pl-[14px] px-3 py-2 border border-gray-200 hover:border-gray-300 hover:shadow-sm focus:outline-none focus:ring-primary-500 focus:border-primary-500 placeholder-gray-400 caret-primary-600 sm:text-sm pr-10'}
+                    tabIndex={2}
                   />
                 </div>
                 <div className='mt-1 text-xs text-gray-500'>{t('login.error.passwordInvalid')}</div>
diff --git a/web/app/signin/normalForm.tsx b/web/app/signin/normalForm.tsx
index 7f23c7d22e5ff..816df8007d9e7 100644
--- a/web/app/signin/normalForm.tsx
+++ b/web/app/signin/normalForm.tsx
@@ -217,6 +217,7 @@ const NormalForm = () => {
                       autoComplete="email"
                       placeholder={t('login.emailPlaceholder') || ''}
                       className={'appearance-none block w-full rounded-lg pl-[14px] px-3 py-2 border border-gray-200 hover:border-gray-300 hover:shadow-sm focus:outline-none focus:ring-primary-500 focus:border-primary-500 placeholder-gray-400 caret-primary-600 sm:text-sm'}
+                      tabIndex={1}
                     />
                   </div>
                 </div>
@@ -241,6 +242,7 @@ const NormalForm = () => {
                       autoComplete="current-password"
                       placeholder={t('login.passwordPlaceholder') || ''}
                       className={'appearance-none block w-full rounded-lg pl-[14px] px-3 py-2 border border-gray-200 hover:border-gray-300 hover:shadow-sm focus:outline-none focus:ring-primary-500 focus:border-primary-500 placeholder-gray-400 caret-primary-600 sm:text-sm pr-10'}
+                      tabIndex={2}
                     />
                     <div className="absolute inset-y-0 right-0 flex items-center pr-3">
                       <button

From 6d56d5c1f63825ccbba659a6d3f5da03de97dede Mon Sep 17 00:00:00 2001
From: sino <sino2322@gmail.com>
Date: Sun, 22 Sep 2024 10:23:50 +0800
Subject: [PATCH 019/235] feat: support o1 series models for openrouter (#8358)

---
 .../openrouter/llm/_position.yaml             |  2 +
 .../model_providers/openrouter/llm/llm.py     | 47 +++++++++++++++++--
 .../openrouter/llm/o1-mini.yaml               | 40 ++++++++++++++++
 .../openrouter/llm/o1-preview.yaml            | 40 ++++++++++++++++
 4 files changed, 126 insertions(+), 3 deletions(-)
 create mode 100644 api/core/model_runtime/model_providers/openrouter/llm/o1-mini.yaml
 create mode 100644 api/core/model_runtime/model_providers/openrouter/llm/o1-preview.yaml

diff --git a/api/core/model_runtime/model_providers/openrouter/llm/_position.yaml b/api/core/model_runtime/model_providers/openrouter/llm/_position.yaml
index 7e00dd3f4bc3c..d9497b76b894c 100644
--- a/api/core/model_runtime/model_providers/openrouter/llm/_position.yaml
+++ b/api/core/model_runtime/model_providers/openrouter/llm/_position.yaml
@@ -1,3 +1,5 @@
+- openai/o1-preview
+- openai/o1-mini
 - openai/gpt-4o
 - openai/gpt-4o-mini
 - openai/gpt-4
diff --git a/api/core/model_runtime/model_providers/openrouter/llm/llm.py b/api/core/model_runtime/model_providers/openrouter/llm/llm.py
index b6bb249a0407b..736ab8e7a8f6c 100644
--- a/api/core/model_runtime/model_providers/openrouter/llm/llm.py
+++ b/api/core/model_runtime/model_providers/openrouter/llm/llm.py
@@ -1,7 +1,7 @@
 from collections.abc import Generator
 from typing import Optional, Union
 
-from core.model_runtime.entities.llm_entities import LLMResult
+from core.model_runtime.entities.llm_entities import LLMResult, LLMResultChunk, LLMResultChunkDelta
 from core.model_runtime.entities.message_entities import PromptMessage, PromptMessageTool
 from core.model_runtime.entities.model_entities import AIModelEntity
 from core.model_runtime.model_providers.openai_api_compatible.llm.llm import OAIAPICompatLargeLanguageModel
@@ -26,7 +26,7 @@ def _invoke(
     ) -> Union[LLMResult, Generator]:
         self._update_credential(model, credentials)
 
-        return super()._invoke(model, credentials, prompt_messages, model_parameters, tools, stop, stream, user)
+        return self._generate(model, credentials, prompt_messages, model_parameters, tools, stop, stream, user)
 
     def validate_credentials(self, model: str, credentials: dict) -> None:
         self._update_credential(model, credentials)
@@ -46,7 +46,48 @@ def _generate(
     ) -> Union[LLMResult, Generator]:
         self._update_credential(model, credentials)
 
-        return super()._generate(model, credentials, prompt_messages, model_parameters, tools, stop, stream, user)
+        block_as_stream = False
+        if model.startswith("openai/o1"):
+            block_as_stream = True
+            stop = None
+
+        # invoke block as stream
+        if stream and block_as_stream:
+            return self._generate_block_as_stream(
+                model, credentials, prompt_messages, model_parameters, tools, stop, user
+            )
+        else:
+            return super()._generate(model, credentials, prompt_messages, model_parameters, tools, stop, stream, user)
+
+    def _generate_block_as_stream(
+        self,
+        model: str,
+        credentials: dict,
+        prompt_messages: list[PromptMessage],
+        model_parameters: dict,
+        tools: Optional[list[PromptMessageTool]] = None,
+        stop: Optional[list[str]] = None,
+        user: Optional[str] = None,
+    ) -> Generator:
+        resp: LLMResult = super()._generate(
+            model, credentials, prompt_messages, model_parameters, tools, stop, False, user
+        )
+
+        yield LLMResultChunk(
+            model=model,
+            prompt_messages=prompt_messages,
+            delta=LLMResultChunkDelta(
+                index=0,
+                message=resp.message,
+                usage=self._calc_response_usage(
+                    model=model,
+                    credentials=credentials,
+                    prompt_tokens=resp.usage.prompt_tokens,
+                    completion_tokens=resp.usage.completion_tokens,
+                ),
+                finish_reason="stop",
+            ),
+        )
 
     def get_customizable_model_schema(self, model: str, credentials: dict) -> AIModelEntity:
         self._update_credential(model, credentials)
diff --git a/api/core/model_runtime/model_providers/openrouter/llm/o1-mini.yaml b/api/core/model_runtime/model_providers/openrouter/llm/o1-mini.yaml
new file mode 100644
index 0000000000000..85a918ff5e0ed
--- /dev/null
+++ b/api/core/model_runtime/model_providers/openrouter/llm/o1-mini.yaml
@@ -0,0 +1,40 @@
+model: openai/o1-mini
+label:
+  en_US: o1-mini
+model_type: llm
+features:
+  - agent-thought
+model_properties:
+  mode: chat
+  context_size: 128000
+parameter_rules:
+  - name: temperature
+    use_template: temperature
+  - name: top_p
+    use_template: top_p
+  - name: presence_penalty
+    use_template: presence_penalty
+  - name: frequency_penalty
+    use_template: frequency_penalty
+  - name: max_tokens
+    use_template: max_tokens
+    default: 512
+    min: 1
+    max: 65536
+  - name: response_format
+    label:
+      zh_Hans: 回复格式
+      en_US: response_format
+    type: string
+    help:
+      zh_Hans: 指定模型必须输出的格式
+      en_US: specifying the format that the model must output
+    required: false
+    options:
+      - text
+      - json_object
+pricing:
+  input: "3.00"
+  output: "12.00"
+  unit: "0.000001"
+  currency: USD
diff --git a/api/core/model_runtime/model_providers/openrouter/llm/o1-preview.yaml b/api/core/model_runtime/model_providers/openrouter/llm/o1-preview.yaml
new file mode 100644
index 0000000000000..74b0a511bed27
--- /dev/null
+++ b/api/core/model_runtime/model_providers/openrouter/llm/o1-preview.yaml
@@ -0,0 +1,40 @@
+model: openai/o1-preview
+label:
+  en_US: o1-preview
+model_type: llm
+features:
+  - agent-thought
+model_properties:
+  mode: chat
+  context_size: 128000
+parameter_rules:
+  - name: temperature
+    use_template: temperature
+  - name: top_p
+    use_template: top_p
+  - name: presence_penalty
+    use_template: presence_penalty
+  - name: frequency_penalty
+    use_template: frequency_penalty
+  - name: max_tokens
+    use_template: max_tokens
+    default: 512
+    min: 1
+    max: 32768
+  - name: response_format
+    label:
+      zh_Hans: 回复格式
+      en_US: response_format
+    type: string
+    help:
+      zh_Hans: 指定模型必须输出的格式
+      en_US: specifying the format that the model must output
+    required: false
+    options:
+      - text
+      - json_object
+pricing:
+  input: "15.00"
+  output: "60.00"
+  unit: "0.000001"
+  currency: USD

From 97895ec41a715e476710c13407f4b5134ef36796 Mon Sep 17 00:00:00 2001
From: CXwudi <cx2010017@gmail.com>
Date: Sun, 22 Sep 2024 01:38:08 -0400
Subject: [PATCH 020/235] chore: add Gemini newest experimental models (close
 #7121) (#8621)

---
 .../llm/gemini-1.5-flash-8b-exp-0827.yaml     | 39 +++++++++++++++++++
 .../google/llm/gemini-1.5-flash-exp-0827.yaml | 39 +++++++++++++++++++
 .../google/llm/gemini-1.5-pro-exp-0801.yaml   | 39 +++++++++++++++++++
 .../google/llm/gemini-1.5-pro-exp-0827.yaml   | 39 +++++++++++++++++++
 .../google/llm/gemini-1.5-pro-latest.yaml     |  2 +-
 api/poetry.lock                               | 16 ++++----
 api/pyproject.toml                            |  4 +-
 7 files changed, 167 insertions(+), 11 deletions(-)
 create mode 100644 api/core/model_runtime/model_providers/google/llm/gemini-1.5-flash-8b-exp-0827.yaml
 create mode 100644 api/core/model_runtime/model_providers/google/llm/gemini-1.5-flash-exp-0827.yaml
 create mode 100644 api/core/model_runtime/model_providers/google/llm/gemini-1.5-pro-exp-0801.yaml
 create mode 100644 api/core/model_runtime/model_providers/google/llm/gemini-1.5-pro-exp-0827.yaml

diff --git a/api/core/model_runtime/model_providers/google/llm/gemini-1.5-flash-8b-exp-0827.yaml b/api/core/model_runtime/model_providers/google/llm/gemini-1.5-flash-8b-exp-0827.yaml
new file mode 100644
index 0000000000000..bbc697e934e05
--- /dev/null
+++ b/api/core/model_runtime/model_providers/google/llm/gemini-1.5-flash-8b-exp-0827.yaml
@@ -0,0 +1,39 @@
+model: gemini-1.5-flash-8b-exp-0827
+label:
+  en_US: Gemini 1.5 Flash 8B 0827
+model_type: llm
+features:
+  - agent-thought
+  - vision
+  - tool-call
+  - stream-tool-call
+model_properties:
+  mode: chat
+  context_size: 1048576
+parameter_rules:
+  - name: temperature
+    use_template: temperature
+  - name: top_p
+    use_template: top_p
+  - name: top_k
+    label:
+      zh_Hans: 取样数量
+      en_US: Top k
+    type: int
+    help:
+      zh_Hans: 仅从每个后续标记的前 K 个选项中采样。
+      en_US: Only sample from the top K options for each subsequent token.
+    required: false
+  - name: max_tokens_to_sample
+    use_template: max_tokens
+    required: true
+    default: 8192
+    min: 1
+    max: 8192
+  - name: response_format
+    use_template: response_format
+pricing:
+  input: '0.00'
+  output: '0.00'
+  unit: '0.000001'
+  currency: USD
diff --git a/api/core/model_runtime/model_providers/google/llm/gemini-1.5-flash-exp-0827.yaml b/api/core/model_runtime/model_providers/google/llm/gemini-1.5-flash-exp-0827.yaml
new file mode 100644
index 0000000000000..c5695e5dda8eb
--- /dev/null
+++ b/api/core/model_runtime/model_providers/google/llm/gemini-1.5-flash-exp-0827.yaml
@@ -0,0 +1,39 @@
+model: gemini-1.5-flash-exp-0827
+label:
+  en_US: Gemini 1.5 Flash 0827
+model_type: llm
+features:
+  - agent-thought
+  - vision
+  - tool-call
+  - stream-tool-call
+model_properties:
+  mode: chat
+  context_size: 1048576
+parameter_rules:
+  - name: temperature
+    use_template: temperature
+  - name: top_p
+    use_template: top_p
+  - name: top_k
+    label:
+      zh_Hans: 取样数量
+      en_US: Top k
+    type: int
+    help:
+      zh_Hans: 仅从每个后续标记的前 K 个选项中采样。
+      en_US: Only sample from the top K options for each subsequent token.
+    required: false
+  - name: max_tokens_to_sample
+    use_template: max_tokens
+    required: true
+    default: 8192
+    min: 1
+    max: 8192
+  - name: response_format
+    use_template: response_format
+pricing:
+  input: '0.00'
+  output: '0.00'
+  unit: '0.000001'
+  currency: USD
diff --git a/api/core/model_runtime/model_providers/google/llm/gemini-1.5-pro-exp-0801.yaml b/api/core/model_runtime/model_providers/google/llm/gemini-1.5-pro-exp-0801.yaml
new file mode 100644
index 0000000000000..0a918e0d7b1ac
--- /dev/null
+++ b/api/core/model_runtime/model_providers/google/llm/gemini-1.5-pro-exp-0801.yaml
@@ -0,0 +1,39 @@
+model: gemini-1.5-pro-exp-0801
+label:
+  en_US: Gemini 1.5 Pro 0801
+model_type: llm
+features:
+  - agent-thought
+  - vision
+  - tool-call
+  - stream-tool-call
+model_properties:
+  mode: chat
+  context_size: 2097152
+parameter_rules:
+  - name: temperature
+    use_template: temperature
+  - name: top_p
+    use_template: top_p
+  - name: top_k
+    label:
+      zh_Hans: 取样数量
+      en_US: Top k
+    type: int
+    help:
+      zh_Hans: 仅从每个后续标记的前 K 个选项中采样。
+      en_US: Only sample from the top K options for each subsequent token.
+    required: false
+  - name: max_tokens_to_sample
+    use_template: max_tokens
+    required: true
+    default: 8192
+    min: 1
+    max: 8192
+  - name: response_format
+    use_template: response_format
+pricing:
+  input: '0.00'
+  output: '0.00'
+  unit: '0.000001'
+  currency: USD
diff --git a/api/core/model_runtime/model_providers/google/llm/gemini-1.5-pro-exp-0827.yaml b/api/core/model_runtime/model_providers/google/llm/gemini-1.5-pro-exp-0827.yaml
new file mode 100644
index 0000000000000..7452ce46e7dcb
--- /dev/null
+++ b/api/core/model_runtime/model_providers/google/llm/gemini-1.5-pro-exp-0827.yaml
@@ -0,0 +1,39 @@
+model: gemini-1.5-pro-exp-0827
+label:
+  en_US: Gemini 1.5 Pro 0827
+model_type: llm
+features:
+  - agent-thought
+  - vision
+  - tool-call
+  - stream-tool-call
+model_properties:
+  mode: chat
+  context_size: 2097152
+parameter_rules:
+  - name: temperature
+    use_template: temperature
+  - name: top_p
+    use_template: top_p
+  - name: top_k
+    label:
+      zh_Hans: 取样数量
+      en_US: Top k
+    type: int
+    help:
+      zh_Hans: 仅从每个后续标记的前 K 个选项中采样。
+      en_US: Only sample from the top K options for each subsequent token.
+    required: false
+  - name: max_tokens_to_sample
+    use_template: max_tokens
+    required: true
+    default: 8192
+    min: 1
+    max: 8192
+  - name: response_format
+    use_template: response_format
+pricing:
+  input: '0.00'
+  output: '0.00'
+  unit: '0.000001'
+  currency: USD
diff --git a/api/core/model_runtime/model_providers/google/llm/gemini-1.5-pro-latest.yaml b/api/core/model_runtime/model_providers/google/llm/gemini-1.5-pro-latest.yaml
index d65dc02674979..b3e1ecf3aff37 100644
--- a/api/core/model_runtime/model_providers/google/llm/gemini-1.5-pro-latest.yaml
+++ b/api/core/model_runtime/model_providers/google/llm/gemini-1.5-pro-latest.yaml
@@ -9,7 +9,7 @@ features:
   - stream-tool-call
 model_properties:
   mode: chat
-  context_size: 1048576
+  context_size: 2097152
 parameter_rules:
   - name: temperature
     use_template: temperature
diff --git a/api/poetry.lock b/api/poetry.lock
index e532394a24e74..78816683d8d24 100644
--- a/api/poetry.lock
+++ b/api/poetry.lock
@@ -3057,20 +3057,20 @@ tests = ["cython", "hypothesis", "mpmath", "pytest", "setuptools"]
 
 [[package]]
 name = "google-ai-generativelanguage"
-version = "0.6.1"
+version = "0.6.9"
 description = "Google Ai Generativelanguage API client library"
 optional = false
 python-versions = ">=3.7"
 files = [
-    {file = "google-ai-generativelanguage-0.6.1.tar.gz", hash = "sha256:4abf37000718b20c43f4b90672b3ab8850738b02457efffd11f3184e03272ed2"},
-    {file = "google_ai_generativelanguage-0.6.1-py3-none-any.whl", hash = "sha256:d2afc991c47663bdf65bd4aabcd89723550b81ad0a6d0be8bfb0160755da4cf0"},
+    {file = "google_ai_generativelanguage-0.6.9-py3-none-any.whl", hash = "sha256:50360cd80015d1a8cc70952e98560f32fa06ddee2e8e9f4b4b98e431dc561e0b"},
+    {file = "google_ai_generativelanguage-0.6.9.tar.gz", hash = "sha256:899f1d3a06efa9739f1cd9d2788070178db33c89d4a76f2e8f4da76f649155fa"},
 ]
 
 [package.dependencies]
 google-api-core = {version = ">=1.34.1,<2.0.dev0 || >=2.11.dev0,<3.0.0dev", extras = ["grpc"]}
 google-auth = ">=2.14.1,<2.24.0 || >2.24.0,<2.25.0 || >2.25.0,<3.0.0dev"
 proto-plus = ">=1.22.3,<2.0.0dev"
-protobuf = ">=3.19.5,<3.20.0 || >3.20.0,<3.20.1 || >3.20.1,<4.21.0 || >4.21.0,<4.21.1 || >4.21.1,<4.21.2 || >4.21.2,<4.21.3 || >4.21.3,<4.21.4 || >4.21.4,<4.21.5 || >4.21.5,<5.0.0dev"
+protobuf = ">=3.20.2,<4.21.0 || >4.21.0,<4.21.1 || >4.21.1,<4.21.2 || >4.21.2,<4.21.3 || >4.21.3,<4.21.4 || >4.21.4,<4.21.5 || >4.21.5,<6.0.0dev"
 
 [[package]]
 name = "google-api-core"
@@ -3336,16 +3336,16 @@ testing = ["pytest"]
 
 [[package]]
 name = "google-generativeai"
-version = "0.5.0"
+version = "0.8.1"
 description = "Google Generative AI High level API client library and tools."
 optional = false
 python-versions = ">=3.9"
 files = [
-    {file = "google_generativeai-0.5.0-py3-none-any.whl", hash = "sha256:207ed12c6a2eeab549a45abbf5373c82077f62b16030bdb502556c78f6d1b5d2"},
+    {file = "google_generativeai-0.8.1-py3-none-any.whl", hash = "sha256:b031877f24d51af0945207657c085896a0a886eceec7a1cb7029327b0aa6e2f6"},
 ]
 
 [package.dependencies]
-google-ai-generativelanguage = "0.6.1"
+google-ai-generativelanguage = "0.6.9"
 google-api-core = "*"
 google-api-python-client = "*"
 google-auth = ">=2.15.0"
@@ -10422,4 +10422,4 @@ cffi = ["cffi (>=1.11)"]
 [metadata]
 lock-version = "2.0"
 python-versions = ">=3.10,<3.13"
-content-hash = "85aa4be7defee8fe6622cf95ba03e81895121502ebf6d666d6ce376ff019fac7"
+content-hash = "eb7ef7be5c7790e214f37f17f92b69407ad557cb80055ef7e49e36eb51b3fca6"
diff --git a/api/pyproject.toml b/api/pyproject.toml
index 93482b032d536..506f379aaf116 100644
--- a/api/pyproject.toml
+++ b/api/pyproject.toml
@@ -156,14 +156,14 @@ flask-restful = "~0.3.10"
 Flask-SQLAlchemy = "~3.1.1"
 gevent = "~23.9.1"
 gmpy2 = "~2.2.1"
-google-ai-generativelanguage = "0.6.1"
+google-ai-generativelanguage = "0.6.9"
 google-api-core = "2.18.0"
 google-api-python-client = "2.90.0"
 google-auth = "2.29.0"
 google-auth-httplib2 = "0.2.0"
 google-cloud-aiplatform = "1.49.0"
 google-cloud-storage = "2.16.0"
-google-generativeai = "0.5.0"
+google-generativeai = "0.8.1"
 googleapis-common-protos = "1.63.0"
 gunicorn = "~22.0.0"
 httpx = { version = "~0.27.0", extras = ["socks"] }

From ddf6569dc5ceaaf0350d9e59456e7b0a7842116b Mon Sep 17 00:00:00 2001
From: Nam Vu <zuzoovn@gmail.com>
Date: Sun, 22 Sep 2024 12:38:41 +0700
Subject: [PATCH 021/235] chore: enhance configuration descriptions (#8624)

---
 api/configs/deploy/__init__.py                |  12 +-
 api/configs/enterprise/__init__.py            |   6 +-
 api/configs/extra/notion_config.py            |  13 +-
 api/configs/extra/sentry_config.py            |  11 +-
 api/configs/feature/__init__.py               | 217 +++++++++---------
 .../feature/hosted_service/__init__.py        |  66 +++---
 api/configs/middleware/__init__.py            |  49 ++--
 api/configs/middleware/cache/redis_config.py  |  26 +--
 .../storage/aliyun_oss_storage_config.py      |  16 +-
 .../storage/amazon_s3_storage_config.py       |  16 +-
 .../storage/azure_blob_storage_config.py      |  10 +-
 .../storage/google_cloud_storage_config.py    |   6 +-
 .../storage/huawei_obs_storage_config.py      |  10 +-
 .../middleware/storage/oci_storage_config.py  |  12 +-
 .../storage/tencent_cos_storage_config.py     |  12 +-
 .../storage/volcengine_tos_storage_config.py  |  12 +-
 .../middleware/vdb/analyticdb_config.py       |  23 +-
 api/configs/middleware/vdb/chroma_config.py   |  14 +-
 .../middleware/vdb/elasticsearch_config.py    |  10 +-
 api/configs/middleware/vdb/milvus_config.py   |  12 +-
 api/configs/middleware/vdb/myscale_config.py  |  14 +-
 .../middleware/vdb/opensearch_config.py       |  12 +-
 api/configs/middleware/vdb/oracle_config.py   |  12 +-
 api/configs/middleware/vdb/pgvector_config.py |  12 +-
 .../middleware/vdb/pgvectors_config.py        |  12 +-
 api/configs/middleware/vdb/qdrant_config.py   |  12 +-
 api/configs/middleware/vdb/relyt_config.py    |  12 +-
 .../middleware/vdb/tencent_vector_config.py   |  18 +-
 .../middleware/vdb/tidb_vector_config.py      |  12 +-
 api/configs/middleware/vdb/weaviate_config.py |  10 +-
 30 files changed, 343 insertions(+), 336 deletions(-)

diff --git a/api/configs/deploy/__init__.py b/api/configs/deploy/__init__.py
index 10271483c4669..66d6a55b4c7ea 100644
--- a/api/configs/deploy/__init__.py
+++ b/api/configs/deploy/__init__.py
@@ -4,30 +4,30 @@
 
 class DeploymentConfig(BaseSettings):
     """
-    Deployment configs
+    Configuration settings for application deployment
     """
 
     APPLICATION_NAME: str = Field(
-        description="application name",
+        description="Name of the application, used for identification and logging purposes",
         default="langgenius/dify",
     )
 
     DEBUG: bool = Field(
-        description="whether to enable debug mode.",
+        description="Enable debug mode for additional logging and development features",
         default=False,
     )
 
     TESTING: bool = Field(
-        description="",
+        description="Enable testing mode for running automated tests",
         default=False,
     )
 
     EDITION: str = Field(
-        description="deployment edition",
+        description="Deployment edition of the application (e.g., 'SELF_HOSTED', 'CLOUD')",
         default="SELF_HOSTED",
     )
 
     DEPLOY_ENV: str = Field(
-        description="deployment environment, default to PRODUCTION.",
+        description="Deployment environment (e.g., 'PRODUCTION', 'DEVELOPMENT'), default to PRODUCTION",
         default="PRODUCTION",
     )
diff --git a/api/configs/enterprise/__init__.py b/api/configs/enterprise/__init__.py
index c661593a44264..eda6345e145a9 100644
--- a/api/configs/enterprise/__init__.py
+++ b/api/configs/enterprise/__init__.py
@@ -4,17 +4,17 @@
 
 class EnterpriseFeatureConfig(BaseSettings):
     """
-    Enterprise feature configs.
+    Configuration for enterprise-level features.
     **Before using, please contact business@dify.ai by email to inquire about licensing matters.**
     """
 
     ENTERPRISE_ENABLED: bool = Field(
-        description="whether to enable enterprise features."
+        description="Enable or disable enterprise-level features."
         "Before using, please contact business@dify.ai by email to inquire about licensing matters.",
         default=False,
     )
 
     CAN_REPLACE_LOGO: bool = Field(
-        description="whether to allow replacing enterprise logo.",
+        description="Allow customization of the enterprise logo.",
         default=False,
     )
diff --git a/api/configs/extra/notion_config.py b/api/configs/extra/notion_config.py
index bd1268fa4565f..f9c4d7346399a 100644
--- a/api/configs/extra/notion_config.py
+++ b/api/configs/extra/notion_config.py
@@ -6,30 +6,31 @@
 
 class NotionConfig(BaseSettings):
     """
-    Notion integration configs
+    Configuration settings for Notion integration
     """
 
     NOTION_CLIENT_ID: Optional[str] = Field(
-        description="Notion client ID",
+        description="Client ID for Notion API authentication. Required for OAuth 2.0 flow.",
         default=None,
     )
 
     NOTION_CLIENT_SECRET: Optional[str] = Field(
-        description="Notion client secret key",
+        description="Client secret for Notion API authentication. Required for OAuth 2.0 flow.",
         default=None,
     )
 
     NOTION_INTEGRATION_TYPE: Optional[str] = Field(
-        description="Notion integration type, default to None, available values: internal.",
+        description="Type of Notion integration."
+        " Set to 'internal' for internal integrations, or None for public integrations.",
         default=None,
     )
 
     NOTION_INTERNAL_SECRET: Optional[str] = Field(
-        description="Notion internal secret key",
+        description="Secret key for internal Notion integrations. Required when NOTION_INTEGRATION_TYPE is 'internal'.",
         default=None,
     )
 
     NOTION_INTEGRATION_TOKEN: Optional[str] = Field(
-        description="Notion integration token",
+        description="Integration token for Notion API access. Used for direct API calls without OAuth flow.",
         default=None,
     )
diff --git a/api/configs/extra/sentry_config.py b/api/configs/extra/sentry_config.py
index ea9ea60ffbd6c..f76a6bdb95ca5 100644
--- a/api/configs/extra/sentry_config.py
+++ b/api/configs/extra/sentry_config.py
@@ -6,20 +6,23 @@
 
 class SentryConfig(BaseSettings):
     """
-    Sentry configs
+    Configuration settings for Sentry error tracking and performance monitoring
     """
 
     SENTRY_DSN: Optional[str] = Field(
-        description="Sentry DSN",
+        description="Sentry Data Source Name (DSN)."
+        " This is the unique identifier of your Sentry project, used to send events to the correct project.",
         default=None,
     )
 
     SENTRY_TRACES_SAMPLE_RATE: NonNegativeFloat = Field(
-        description="Sentry trace sample rate",
+        description="Sample rate for Sentry performance monitoring traces."
+        " Value between 0.0 and 1.0, where 1.0 means 100% of traces are sent to Sentry.",
         default=1.0,
     )
 
     SENTRY_PROFILES_SAMPLE_RATE: NonNegativeFloat = Field(
-        description="Sentry profiles sample rate",
+        description="Sample rate for Sentry profiling."
+        " Value between 0.0 and 1.0, where 1.0 means 100% of profiles are sent to Sentry.",
         default=1.0,
     )
diff --git a/api/configs/feature/__init__.py b/api/configs/feature/__init__.py
index f794552c36d26..9218d529cc09e 100644
--- a/api/configs/feature/__init__.py
+++ b/api/configs/feature/__init__.py
@@ -8,145 +8,143 @@
 
 class SecurityConfig(BaseSettings):
     """
-    Secret Key configs
+    Security-related configurations for the application
     """
 
     SECRET_KEY: Optional[str] = Field(
-        description="Your App secret key will be used for securely signing the session cookie"
+        description="Secret key for secure session cookie signing."
         "Make sure you are changing this key for your deployment with a strong key."
-        "You can generate a strong key using `openssl rand -base64 42`."
-        "Alternatively you can set it with `SECRET_KEY` environment variable.",
+        "Generate a strong key using `openssl rand -base64 42` or set via the `SECRET_KEY` environment variable.",
         default=None,
     )
 
     RESET_PASSWORD_TOKEN_EXPIRY_HOURS: PositiveInt = Field(
-        description="Expiry time in hours for reset token",
+        description="Duration in hours for which a password reset token remains valid",
         default=24,
     )
 
 
 class AppExecutionConfig(BaseSettings):
     """
-    App Execution configs
+    Configuration parameters for application execution
     """
 
     APP_MAX_EXECUTION_TIME: PositiveInt = Field(
-        description="execution timeout in seconds for app execution",
+        description="Maximum allowed execution time for the application in seconds",
         default=1200,
     )
     APP_MAX_ACTIVE_REQUESTS: NonNegativeInt = Field(
-        description="max active request per app, 0 means unlimited",
+        description="Maximum number of concurrent active requests per app (0 for unlimited)",
         default=0,
     )
 
 
 class CodeExecutionSandboxConfig(BaseSettings):
     """
-    Code Execution Sandbox configs
+    Configuration for the code execution sandbox environment
     """
 
     CODE_EXECUTION_ENDPOINT: HttpUrl = Field(
-        description="endpoint URL of code execution service",
+        description="URL endpoint for the code execution service",
         default="http://sandbox:8194",
     )
 
     CODE_EXECUTION_API_KEY: str = Field(
-        description="API key for code execution service",
+        description="API key for accessing the code execution service",
         default="dify-sandbox",
     )
 
     CODE_EXECUTION_CONNECT_TIMEOUT: Optional[float] = Field(
-        description="connect timeout in seconds for code execution request",
+        description="Connection timeout in seconds for code execution requests",
         default=10.0,
     )
 
     CODE_EXECUTION_READ_TIMEOUT: Optional[float] = Field(
-        description="read timeout in seconds for code execution request",
+        description="Read timeout in seconds for code execution requests",
         default=60.0,
     )
 
     CODE_EXECUTION_WRITE_TIMEOUT: Optional[float] = Field(
-        description="write timeout in seconds for code execution request",
+        description="Write timeout in seconds for code execution request",
         default=10.0,
     )
 
     CODE_MAX_NUMBER: PositiveInt = Field(
-        description="max depth for code execution",
+        description="Maximum allowed numeric value in code execution",
         default=9223372036854775807,
     )
 
     CODE_MIN_NUMBER: NegativeInt = Field(
-        description="",
+        description="Minimum allowed numeric value in code execution",
         default=-9223372036854775807,
     )
 
     CODE_MAX_DEPTH: PositiveInt = Field(
-        description="max depth for code execution",
+        description="Maximum allowed depth for nested structures in code execution",
         default=5,
     )
 
     CODE_MAX_PRECISION: PositiveInt = Field(
-        description="max precision digits for float type in code execution",
+        description="mMaximum number of decimal places for floating-point numbers in code execution",
         default=20,
     )
 
     CODE_MAX_STRING_LENGTH: PositiveInt = Field(
-        description="max string length for code execution",
+        description="Maximum allowed length for strings in code execution",
         default=80000,
     )
 
     CODE_MAX_STRING_ARRAY_LENGTH: PositiveInt = Field(
-        description="",
+        description="Maximum allowed length for string arrays in code execution",
         default=30,
     )
 
     CODE_MAX_OBJECT_ARRAY_LENGTH: PositiveInt = Field(
-        description="",
+        description="Maximum allowed length for object arrays in code execution",
         default=30,
     )
 
     CODE_MAX_NUMBER_ARRAY_LENGTH: PositiveInt = Field(
-        description="",
+        description="Maximum allowed length for numeric arrays in code execution",
         default=1000,
     )
 
 
 class EndpointConfig(BaseSettings):
     """
-    Module URL configs
+    Configuration for various application endpoints and URLs
     """
 
     CONSOLE_API_URL: str = Field(
-        description="The backend URL prefix of the console API."
-        "used to concatenate the login authorization callback or notion integration callback.",
+        description="Base URL for the console API,"
+        "used for login authentication callback or notion integration callbacks",
         default="",
     )
 
     CONSOLE_WEB_URL: str = Field(
-        description="The front-end URL prefix of the console web."
-        "used to concatenate some front-end addresses and for CORS configuration use.",
+        description="Base URL for the console web interface," "used for frontend references and CORS configuration",
         default="",
     )
 
     SERVICE_API_URL: str = Field(
-        description="Service API Url prefix. used to display Service API Base Url to the front-end.",
+        description="Base URL for the service API, displayed to users for API access",
         default="",
     )
 
     APP_WEB_URL: str = Field(
-        description="WebApp Url prefix. used to display WebAPP API Base Url to the front-end.",
+        description="Base URL for the web application, used for frontend references",
         default="",
     )
 
 
 class FileAccessConfig(BaseSettings):
     """
-    File Access configs
+    Configuration for file access and handling
     """
 
     FILES_URL: str = Field(
-        description="File preview or download Url prefix."
-        " used to display File preview or download Url to the front-end or as Multi-model inputs;"
+        description="Base URL for file preview or download,"
+        " used for frontend display and multi-model inputs"
         "Url is signed and has expiration time.",
         validation_alias=AliasChoices("FILES_URL", "CONSOLE_API_URL"),
         alias_priority=1,
@@ -154,49 +152,49 @@ class FileAccessConfig(BaseSettings):
     )
 
     FILES_ACCESS_TIMEOUT: int = Field(
-        description="timeout in seconds for file accessing",
+        description="Expiration time in seconds for file access URLs",
         default=300,
     )
 
 
 class FileUploadConfig(BaseSettings):
     """
-    File Uploading configs
+    Configuration for file upload limitations
     """
 
     UPLOAD_FILE_SIZE_LIMIT: NonNegativeInt = Field(
-        description="size limit in Megabytes for uploading files",
+        description="Maximum allowed file size for uploads in megabytes",
         default=15,
     )
 
     UPLOAD_FILE_BATCH_LIMIT: NonNegativeInt = Field(
-        description="batch size limit for uploading files",
+        description="Maximum number of files allowed in a single upload batch",
         default=5,
     )
 
     UPLOAD_IMAGE_FILE_SIZE_LIMIT: NonNegativeInt = Field(
-        description="image file size limit in Megabytes for uploading files",
+        description="Maximum allowed image file size for uploads in megabytes",
         default=10,
     )
 
     BATCH_UPLOAD_LIMIT: NonNegativeInt = Field(
-        description="",  # todo: to be clarified
+        description="Maximum number of files allowed in a batch upload operation",
         default=20,
     )
 
 
 class HttpConfig(BaseSettings):
     """
-    HTTP configs
+    HTTP-related configurations for the application
     """
 
     API_COMPRESSION_ENABLED: bool = Field(
-        description="whether to enable HTTP response compression of gzip",
+        description="Enable or disable gzip compression for HTTP responses",
         default=False,
     )
 
     inner_CONSOLE_CORS_ALLOW_ORIGINS: str = Field(
-        description="",
+        description="Comma-separated list of allowed origins for CORS in the console",
         validation_alias=AliasChoices("CONSOLE_CORS_ALLOW_ORIGINS", "CONSOLE_WEB_URL"),
         default="",
     )
@@ -218,359 +216,360 @@ def WEB_API_CORS_ALLOW_ORIGINS(self) -> list[str]:
         return self.inner_WEB_API_CORS_ALLOW_ORIGINS.split(",")
 
     HTTP_REQUEST_MAX_CONNECT_TIMEOUT: Annotated[
-        PositiveInt, Field(ge=10, description="connect timeout in seconds for HTTP request")
+        PositiveInt, Field(ge=10, description="Maximum connection timeout in seconds for HTTP requests")
     ] = 10
 
     HTTP_REQUEST_MAX_READ_TIMEOUT: Annotated[
-        PositiveInt, Field(ge=60, description="read timeout in seconds for HTTP request")
+        PositiveInt, Field(ge=60, description="Maximum read timeout in seconds for HTTP requests")
     ] = 60
 
     HTTP_REQUEST_MAX_WRITE_TIMEOUT: Annotated[
-        PositiveInt, Field(ge=10, description="read timeout in seconds for HTTP request")
+        PositiveInt, Field(ge=10, description="Maximum write timeout in seconds for HTTP requests")
     ] = 20
 
     HTTP_REQUEST_NODE_MAX_BINARY_SIZE: PositiveInt = Field(
-        description="",
+        description="Maximum allowed size in bytes for binary data in HTTP requests",
         default=10 * 1024 * 1024,
     )
 
     HTTP_REQUEST_NODE_MAX_TEXT_SIZE: PositiveInt = Field(
-        description="",
+        description="Maximum allowed size in bytes for text data in HTTP requests",
         default=1 * 1024 * 1024,
     )
 
     SSRF_PROXY_HTTP_URL: Optional[str] = Field(
-        description="HTTP URL for SSRF proxy",
+        description="Proxy URL for HTTP requests to prevent Server-Side Request Forgery (SSRF)",
         default=None,
     )
 
     SSRF_PROXY_HTTPS_URL: Optional[str] = Field(
-        description="HTTPS URL for SSRF proxy",
+        description="Proxy URL for HTTPS requests to prevent Server-Side Request Forgery (SSRF)",
         default=None,
     )
 
 
 class InnerAPIConfig(BaseSettings):
     """
-    Inner API configs
+    Configuration for internal API functionality
     """
 
     INNER_API: bool = Field(
-        description="whether to enable the inner API",
+        description="Enable or disable the internal API",
         default=False,
     )
 
     INNER_API_KEY: Optional[str] = Field(
-        description="The inner API key is used to authenticate the inner API",
+        description="API key for accessing the internal API",
         default=None,
     )
 
 
 class LoggingConfig(BaseSettings):
     """
-    Logging configs
+    Configuration for application logging
     """
 
     LOG_LEVEL: str = Field(
-        description="Log output level, default to INFO. It is recommended to set it to ERROR for production.",
+        description="Logging level, default to INFO. Set to ERROR for production environments.",
         default="INFO",
     )
 
     LOG_FILE: Optional[str] = Field(
-        description="logging output file path",
+        description="File path for log output.",
         default=None,
     )
 
     LOG_FORMAT: str = Field(
-        description="log format",
+        description="Format string for log messages",
         default="%(asctime)s.%(msecs)03d %(levelname)s [%(threadName)s] [%(filename)s:%(lineno)d] - %(message)s",
     )
 
     LOG_DATEFORMAT: Optional[str] = Field(
-        description="log date format",
+        description="Date format string for log timestamps",
         default=None,
     )
 
     LOG_TZ: Optional[str] = Field(
-        description="specify log timezone, eg: America/New_York",
+        description="Timezone for log timestamps (e.g., 'America/New_York')",
         default=None,
     )
 
 
 class ModelLoadBalanceConfig(BaseSettings):
     """
-    Model load balance configs
+    Configuration for model load balancing
     """
 
     MODEL_LB_ENABLED: bool = Field(
-        description="whether to enable model load balancing",
+        description="Enable or disable load balancing for models",
         default=False,
     )
 
 
 class BillingConfig(BaseSettings):
     """
-    Platform Billing Configurations
+    Configuration for platform billing features
     """
 
     BILLING_ENABLED: bool = Field(
-        description="whether to enable billing",
+        description="Enable or disable billing functionality",
         default=False,
     )
 
 
 class UpdateConfig(BaseSettings):
     """
-    Update configs
+    Configuration for application update checks
     """
 
     CHECK_UPDATE_URL: str = Field(
-        description="url for checking updates",
+        description="URL to check for application updates",
         default="https://updates.dify.ai",
     )
 
 
 class WorkflowConfig(BaseSettings):
     """
-    Workflow feature configs
+    Configuration for workflow execution
     """
 
     WORKFLOW_MAX_EXECUTION_STEPS: PositiveInt = Field(
-        description="max execution steps in single workflow execution",
+        description="Maximum number of steps allowed in a single workflow execution",
         default=500,
     )
 
     WORKFLOW_MAX_EXECUTION_TIME: PositiveInt = Field(
-        description="max execution time in seconds in single workflow execution",
+        description="Maximum execution time in seconds for a single workflow",
         default=1200,
     )
 
     WORKFLOW_CALL_MAX_DEPTH: PositiveInt = Field(
-        description="max depth of calling in single workflow execution",
+        description="Maximum allowed depth for nested workflow calls",
         default=5,
     )
 
     MAX_VARIABLE_SIZE: PositiveInt = Field(
-        description="The maximum size in bytes of a variable. default to 5KB.",
+        description="Maximum size in bytes for a single variable in workflows. Default to 5KB.",
         default=5 * 1024,
     )
 
 
 class OAuthConfig(BaseSettings):
     """
-    oauth configs
+    Configuration for OAuth authentication
     """
 
     OAUTH_REDIRECT_PATH: str = Field(
-        description="redirect path for OAuth",
+        description="Redirect path for OAuth authentication callbacks",
         default="/console/api/oauth/authorize",
     )
 
     GITHUB_CLIENT_ID: Optional[str] = Field(
-        description="GitHub client id for OAuth",
+        description="GitHub OAuth client secret",
         default=None,
     )
 
     GITHUB_CLIENT_SECRET: Optional[str] = Field(
-        description="GitHub client secret key for OAuth",
+        description="GitHub OAuth client secret",
         default=None,
     )
 
     GOOGLE_CLIENT_ID: Optional[str] = Field(
-        description="Google client id for OAuth",
+        description="Google OAuth client ID",
         default=None,
     )
 
     GOOGLE_CLIENT_SECRET: Optional[str] = Field(
-        description="Google client secret key for OAuth",
+        description="Google OAuth client secret",
         default=None,
     )
 
 
 class ModerationConfig(BaseSettings):
     """
-    Moderation in app configs.
+    Configuration for content moderation
     """
 
     MODERATION_BUFFER_SIZE: PositiveInt = Field(
-        description="buffer size for moderation",
+        description="Size of the buffer for content moderation processing",
         default=300,
     )
 
 
 class ToolConfig(BaseSettings):
     """
-    Tool configs
+    Configuration for tool management
     """
 
     TOOL_ICON_CACHE_MAX_AGE: PositiveInt = Field(
-        description="max age in seconds for tool icon caching",
+        description="Maximum age in seconds for caching tool icons",
         default=3600,
     )
 
 
 class MailConfig(BaseSettings):
     """
-    Mail Configurations
+    Configuration for email services
     """
 
     MAIL_TYPE: Optional[str] = Field(
-        description="Mail provider type name, default to None, available values are `smtp` and `resend`.",
+        description="Email service provider type ('smtp' or 'resend'), default to None.",
         default=None,
     )
 
     MAIL_DEFAULT_SEND_FROM: Optional[str] = Field(
-        description="default email address for sending from ",
+        description="Default email address to use as the sender",
         default=None,
     )
 
     RESEND_API_KEY: Optional[str] = Field(
-        description="API key for Resend",
+        description="API key for Resend email service",
         default=None,
     )
 
     RESEND_API_URL: Optional[str] = Field(
-        description="API URL for Resend",
+        description="API URL for Resend email service",
         default=None,
     )
 
     SMTP_SERVER: Optional[str] = Field(
-        description="smtp server host",
+        description="SMTP server hostname",
         default=None,
     )
 
     SMTP_PORT: Optional[int] = Field(
-        description="smtp server port",
+        description="SMTP server port number",
         default=465,
     )
 
     SMTP_USERNAME: Optional[str] = Field(
-        description="smtp server username",
+        description="Username for SMTP authentication",
         default=None,
     )
 
     SMTP_PASSWORD: Optional[str] = Field(
-        description="smtp server password",
+        description="Password for SMTP authentication",
         default=None,
     )
 
     SMTP_USE_TLS: bool = Field(
-        description="whether to use TLS connection to smtp server",
+        description="Enable TLS encryption for SMTP connections",
         default=False,
     )
 
     SMTP_OPPORTUNISTIC_TLS: bool = Field(
-        description="whether to use opportunistic TLS connection to smtp server",
+        description="Enable opportunistic TLS for SMTP connections",
         default=False,
     )
 
 
 class RagEtlConfig(BaseSettings):
     """
-    RAG ETL Configurations.
+    Configuration for RAG ETL processes
     """
 
     ETL_TYPE: str = Field(
-        description="RAG ETL type name, default to `dify`, available values are `dify` and `Unstructured`. ",
+        description="RAG ETL type ('dify' or 'Unstructured'), default to 'dify'",
         default="dify",
     )
 
     KEYWORD_DATA_SOURCE_TYPE: str = Field(
-        description="source type for keyword data, default to `database`, available values are `database` .",
+        description="Data source type for keyword extraction"
+        " ('database' or other supported types), default to 'database'",
         default="database",
     )
 
     UNSTRUCTURED_API_URL: Optional[str] = Field(
-        description="API URL for Unstructured",
+        description="API URL for Unstructured.io service",
         default=None,
     )
 
     UNSTRUCTURED_API_KEY: Optional[str] = Field(
-        description="API key for Unstructured",
+        description="API key for Unstructured.io service",
         default=None,
     )
 
 
 class DataSetConfig(BaseSettings):
     """
-    Dataset configs
+    Configuration for dataset management
     """
 
     CLEAN_DAY_SETTING: PositiveInt = Field(
-        description="interval in days for cleaning up dataset",
+        description="Interval in days for dataset cleanup operations",
         default=30,
     )
 
     DATASET_OPERATOR_ENABLED: bool = Field(
-        description="whether to enable dataset operator",
+        description="Enable or disable dataset operator functionality",
         default=False,
     )
 
 
 class WorkspaceConfig(BaseSettings):
     """
-    Workspace configs
+    Configuration for workspace management
     """
 
     INVITE_EXPIRY_HOURS: PositiveInt = Field(
-        description="workspaces invitation expiration in hours",
+        description="Expiration time in hours for workspace invitation links",
         default=72,
     )
 
 
 class IndexingConfig(BaseSettings):
     """
-    Indexing configs.
+    Configuration for indexing operations
     """
 
     INDEXING_MAX_SEGMENTATION_TOKENS_LENGTH: PositiveInt = Field(
-        description="max segmentation token length for indexing",
+        description="Maximum token length for text segmentation during indexing",
         default=1000,
     )
 
 
 class ImageFormatConfig(BaseSettings):
     MULTIMODAL_SEND_IMAGE_FORMAT: str = Field(
-        description="multi model send image format, support base64, url, default is base64",
+        description="Format for sending images in multimodal contexts ('base64' or 'url'), default is base64",
         default="base64",
     )
 
 
 class CeleryBeatConfig(BaseSettings):
     CELERY_BEAT_SCHEDULER_TIME: int = Field(
-        description="the time of the celery scheduler, default to 1 day",
+        description="Interval in days for Celery Beat scheduler execution, default to 1 day",
         default=1,
     )
 
 
 class PositionConfig(BaseSettings):
     POSITION_PROVIDER_PINS: str = Field(
-        description="The heads of model providers",
+        description="Comma-separated list of pinned model providers",
         default="",
     )
 
     POSITION_PROVIDER_INCLUDES: str = Field(
-        description="The included model providers",
+        description="Comma-separated list of included model providers",
         default="",
     )
 
     POSITION_PROVIDER_EXCLUDES: str = Field(
-        description="The excluded model providers",
+        description="Comma-separated list of excluded model providers",
         default="",
     )
 
     POSITION_TOOL_PINS: str = Field(
-        description="The heads of tools",
+        description="Comma-separated list of pinned tools",
         default="",
     )
 
     POSITION_TOOL_INCLUDES: str = Field(
-        description="The included tools",
+        description="Comma-separated list of included tools",
         default="",
     )
 
     POSITION_TOOL_EXCLUDES: str = Field(
-        description="The excluded tools",
+        description="Comma-separated list of excluded tools",
         default="",
     )
 
diff --git a/api/configs/feature/hosted_service/__init__.py b/api/configs/feature/hosted_service/__init__.py
index f269d0ab9c89c..7f103be8f4f90 100644
--- a/api/configs/feature/hosted_service/__init__.py
+++ b/api/configs/feature/hosted_service/__init__.py
@@ -6,31 +6,31 @@
 
 class HostedOpenAiConfig(BaseSettings):
     """
-    Hosted OpenAI service config
+    Configuration for hosted OpenAI service
     """
 
     HOSTED_OPENAI_API_KEY: Optional[str] = Field(
-        description="",
+        description="API key for hosted OpenAI service",
         default=None,
     )
 
     HOSTED_OPENAI_API_BASE: Optional[str] = Field(
-        description="",
+        description="Base URL for hosted OpenAI API",
         default=None,
     )
 
     HOSTED_OPENAI_API_ORGANIZATION: Optional[str] = Field(
-        description="",
+        description="Organization ID for hosted OpenAI service",
         default=None,
     )
 
     HOSTED_OPENAI_TRIAL_ENABLED: bool = Field(
-        description="",
+        description="Enable trial access to hosted OpenAI service",
         default=False,
     )
 
     HOSTED_OPENAI_TRIAL_MODELS: str = Field(
-        description="",
+        description="Comma-separated list of available models for trial access",
         default="gpt-3.5-turbo,"
         "gpt-3.5-turbo-1106,"
         "gpt-3.5-turbo-instruct,"
@@ -42,17 +42,17 @@ class HostedOpenAiConfig(BaseSettings):
     )
 
     HOSTED_OPENAI_QUOTA_LIMIT: NonNegativeInt = Field(
-        description="",
+        description="Quota limit for hosted OpenAI service usage",
         default=200,
     )
 
     HOSTED_OPENAI_PAID_ENABLED: bool = Field(
-        description="",
+        description="Enable paid access to hosted OpenAI service",
         default=False,
     )
 
     HOSTED_OPENAI_PAID_MODELS: str = Field(
-        description="",
+        description="Comma-separated list of available models for paid access",
         default="gpt-4,"
         "gpt-4-turbo-preview,"
         "gpt-4-turbo-2024-04-09,"
@@ -71,124 +71,122 @@ class HostedOpenAiConfig(BaseSettings):
 
 class HostedAzureOpenAiConfig(BaseSettings):
     """
-    Hosted OpenAI service config
+    Configuration for hosted Azure OpenAI service
     """
 
     HOSTED_AZURE_OPENAI_ENABLED: bool = Field(
-        description="",
+        description="Enable hosted Azure OpenAI service",
         default=False,
     )
 
     HOSTED_AZURE_OPENAI_API_KEY: Optional[str] = Field(
-        description="",
+        description="API key for hosted Azure OpenAI service",
         default=None,
     )
 
     HOSTED_AZURE_OPENAI_API_BASE: Optional[str] = Field(
-        description="",
+        description="Base URL for hosted Azure OpenAI API",
         default=None,
     )
 
     HOSTED_AZURE_OPENAI_QUOTA_LIMIT: NonNegativeInt = Field(
-        description="",
+        description="Quota limit for hosted Azure OpenAI service usage",
         default=200,
     )
 
 
 class HostedAnthropicConfig(BaseSettings):
     """
-    Hosted Azure OpenAI service config
+    Configuration for hosted Anthropic service
     """
 
     HOSTED_ANTHROPIC_API_BASE: Optional[str] = Field(
-        description="",
+        description="Base URL for hosted Anthropic API",
         default=None,
     )
 
     HOSTED_ANTHROPIC_API_KEY: Optional[str] = Field(
-        description="",
+        description="API key for hosted Anthropic service",
         default=None,
     )
 
     HOSTED_ANTHROPIC_TRIAL_ENABLED: bool = Field(
-        description="",
+        description="Enable trial access to hosted Anthropic service",
         default=False,
     )
 
     HOSTED_ANTHROPIC_QUOTA_LIMIT: NonNegativeInt = Field(
-        description="",
+        description="Quota limit for hosted Anthropic service usage",
         default=600000,
     )
 
     HOSTED_ANTHROPIC_PAID_ENABLED: bool = Field(
-        description="",
+        description="Enable paid access to hosted Anthropic service",
         default=False,
     )
 
 
 class HostedMinmaxConfig(BaseSettings):
     """
-    Hosted Minmax service config
+    Configuration for hosted Minmax service
     """
 
     HOSTED_MINIMAX_ENABLED: bool = Field(
-        description="",
+        description="Enable hosted Minmax service",
         default=False,
     )
 
 
 class HostedSparkConfig(BaseSettings):
     """
-    Hosted Spark service config
+    Configuration for hosted Spark service
     """
 
     HOSTED_SPARK_ENABLED: bool = Field(
-        description="",
+        description="Enable hosted Spark service",
         default=False,
     )
 
 
 class HostedZhipuAIConfig(BaseSettings):
     """
-    Hosted Minmax service config
+    Configuration for hosted ZhipuAI service
     """
 
     HOSTED_ZHIPUAI_ENABLED: bool = Field(
-        description="",
+        description="Enable hosted ZhipuAI service",
         default=False,
     )
 
 
 class HostedModerationConfig(BaseSettings):
     """
-    Hosted Moderation service config
+    Configuration for hosted Moderation service
     """
 
     HOSTED_MODERATION_ENABLED: bool = Field(
-        description="",
+        description="Enable hosted Moderation service",
         default=False,
     )
 
     HOSTED_MODERATION_PROVIDERS: str = Field(
-        description="",
+        description="Comma-separated list of moderation providers",
         default="",
     )
 
 
 class HostedFetchAppTemplateConfig(BaseSettings):
     """
-    Hosted Moderation service config
+    Configuration for fetching app templates
     """
 
     HOSTED_FETCH_APP_TEMPLATES_MODE: str = Field(
-        description="the mode for fetching app templates,"
-        " default to remote,"
-        " available values: remote, db, builtin",
+        description="Mode for fetching app templates: remote, db, or builtin" " default to remote,",
         default="remote",
     )
 
     HOSTED_FETCH_APP_TEMPLATES_REMOTE_DOMAIN: str = Field(
-        description="the domain for fetching remote app templates",
+        description="Domain for fetching remote app templates",
         default="https://tmpl.dify.ai",
     )
 
diff --git a/api/configs/middleware/__init__.py b/api/configs/middleware/__init__.py
index e017c2c5b828e..6ad216c1913f9 100644
--- a/api/configs/middleware/__init__.py
+++ b/api/configs/middleware/__init__.py
@@ -31,70 +31,71 @@
 
 class StorageConfig(BaseSettings):
     STORAGE_TYPE: str = Field(
-        description="storage type,"
-        " default to `local`,"
-        " available values are `local`, `s3`, `azure-blob`, `aliyun-oss`, `google-storage`.",
+        description="Type of storage to use."
+        " Options: 'local', 's3', 'azure-blob', 'aliyun-oss', 'google-storage'. Default is 'local'.",
         default="local",
     )
 
     STORAGE_LOCAL_PATH: str = Field(
-        description="local storage path",
+        description="Path for local storage when STORAGE_TYPE is set to 'local'.",
         default="storage",
     )
 
 
 class VectorStoreConfig(BaseSettings):
     VECTOR_STORE: Optional[str] = Field(
-        description="vector store type",
+        description="Type of vector store to use for efficient similarity search."
+        " Set to None if not using a vector store.",
         default=None,
     )
 
 
 class KeywordStoreConfig(BaseSettings):
     KEYWORD_STORE: str = Field(
-        description="keyword store type",
+        description="Method for keyword extraction and storage."
+        " Default is 'jieba', a Chinese text segmentation library.",
         default="jieba",
     )
 
 
 class DatabaseConfig:
     DB_HOST: str = Field(
-        description="db host",
+        description="Hostname or IP address of the database server.",
         default="localhost",
     )
 
     DB_PORT: PositiveInt = Field(
-        description="db port",
+        description="Port number for database connection.",
         default=5432,
     )
 
     DB_USERNAME: str = Field(
-        description="db username",
+        description="Username for database authentication.",
         default="postgres",
     )
 
     DB_PASSWORD: str = Field(
-        description="db password",
+        description="Password for database authentication.",
         default="",
     )
 
     DB_DATABASE: str = Field(
-        description="db database",
+        description="Name of the database to connect to.",
         default="dify",
     )
 
     DB_CHARSET: str = Field(
-        description="db charset",
+        description="Character set for database connection.",
         default="",
     )
 
     DB_EXTRAS: str = Field(
-        description="db extras options. Example: keepalives_idle=60&keepalives=1",
+        description="Additional database connection parameters. Example: 'keepalives_idle=60&keepalives=1'",
         default="",
     )
 
     SQLALCHEMY_DATABASE_URI_SCHEME: str = Field(
-        description="db uri scheme",
+        description="Database URI scheme for SQLAlchemy connection.",
         default="postgresql",
     )
 
@@ -112,27 +113,27 @@ def SQLALCHEMY_DATABASE_URI(self) -> str:
         )
 
     SQLALCHEMY_POOL_SIZE: NonNegativeInt = Field(
-        description="pool size of SqlAlchemy",
+        description="Maximum number of database connections in the pool.",
         default=30,
     )
 
     SQLALCHEMY_MAX_OVERFLOW: NonNegativeInt = Field(
-        description="max overflows for SqlAlchemy",
+        description="Maximum number of connections that can be created beyond the pool_size.",
         default=10,
     )
 
     SQLALCHEMY_POOL_RECYCLE: NonNegativeInt = Field(
-        description="SqlAlchemy pool recycle",
+        description="Number of seconds after which a connection is automatically recycled.",
         default=3600,
     )
 
     SQLALCHEMY_POOL_PRE_PING: bool = Field(
-        description="whether to enable pool pre-ping in SqlAlchemy",
+        description="If True, enables connection pool pre-ping feature to check connections.",
         default=False,
     )
 
     SQLALCHEMY_ECHO: bool | str = Field(
-        description="whether to enable SqlAlchemy echo",
+        description="If True, SQLAlchemy will log all SQL statements.",
         default=False,
     )
 
@@ -150,27 +151,27 @@ def SQLALCHEMY_ENGINE_OPTIONS(self) -> dict[str, Any]:
 
 class CeleryConfig(DatabaseConfig):
     CELERY_BACKEND: str = Field(
-        description="Celery backend, available values are `database`, `redis`",
+        description="Backend for Celery task results. Options: 'database', 'redis'.",
         default="database",
     )
 
     CELERY_BROKER_URL: Optional[str] = Field(
-        description="CELERY_BROKER_URL",
+        description="URL of the message broker for Celery tasks.",
         default=None,
     )
 
     CELERY_USE_SENTINEL: Optional[bool] = Field(
-        description="Whether to use Redis Sentinel mode",
+        description="Whether to use Redis Sentinel for high availability.",
         default=False,
     )
 
     CELERY_SENTINEL_MASTER_NAME: Optional[str] = Field(
-        description="Redis Sentinel master name",
+        description="Name of the Redis Sentinel master.",
         default=None,
     )
 
     CELERY_SENTINEL_SOCKET_TIMEOUT: Optional[PositiveFloat] = Field(
-        description="Redis Sentinel socket timeout",
+        description="Timeout for Redis Sentinel socket operations in seconds.",
         default=0.1,
     )
 
diff --git a/api/configs/middleware/cache/redis_config.py b/api/configs/middleware/cache/redis_config.py
index 4fcd52ddc9cbc..26b9b1347c61b 100644
--- a/api/configs/middleware/cache/redis_config.py
+++ b/api/configs/middleware/cache/redis_config.py
@@ -6,65 +6,65 @@
 
 class RedisConfig(BaseSettings):
     """
-    Redis configs
+    Configuration settings for Redis connection
     """
 
     REDIS_HOST: str = Field(
-        description="Redis host",
+        description="Hostname or IP address of the Redis server",
         default="localhost",
     )
 
     REDIS_PORT: PositiveInt = Field(
-        description="Redis port",
+        description="Port number on which the Redis server is listening",
         default=6379,
     )
 
     REDIS_USERNAME: Optional[str] = Field(
-        description="Redis username",
+        description="Username for Redis authentication (if required)",
         default=None,
     )
 
     REDIS_PASSWORD: Optional[str] = Field(
-        description="Redis password",
+        description="Password for Redis authentication (if required)",
         default=None,
     )
 
     REDIS_DB: NonNegativeInt = Field(
-        description="Redis database id, default to 0",
+        description="Redis database number to use (0-15)",
         default=0,
     )
 
     REDIS_USE_SSL: bool = Field(
-        description="whether to use SSL for Redis connection",
+        description="Enable SSL/TLS for the Redis connection",
         default=False,
     )
 
     REDIS_USE_SENTINEL: Optional[bool] = Field(
-        description="Whether to use Redis Sentinel mode",
+        description="Enable Redis Sentinel mode for high availability",
         default=False,
     )
 
     REDIS_SENTINELS: Optional[str] = Field(
-        description="Redis Sentinel nodes",
+        description="Comma-separated list of Redis Sentinel nodes (host:port)",
         default=None,
     )
 
     REDIS_SENTINEL_SERVICE_NAME: Optional[str] = Field(
-        description="Redis Sentinel service name",
+        description="Name of the Redis Sentinel service to monitor",
         default=None,
     )
 
     REDIS_SENTINEL_USERNAME: Optional[str] = Field(
-        description="Redis Sentinel username",
+        description="Username for Redis Sentinel authentication (if required)",
         default=None,
     )
 
     REDIS_SENTINEL_PASSWORD: Optional[str] = Field(
-        description="Redis Sentinel password",
+        description="Password for Redis Sentinel authentication (if required)",
         default=None,
     )
 
     REDIS_SENTINEL_SOCKET_TIMEOUT: Optional[PositiveFloat] = Field(
-        description="Redis Sentinel socket timeout",
+        description="Socket timeout in seconds for Redis Sentinel connections",
         default=0.1,
     )
diff --git a/api/configs/middleware/storage/aliyun_oss_storage_config.py b/api/configs/middleware/storage/aliyun_oss_storage_config.py
index c1843dc26cee0..07eb527170b2e 100644
--- a/api/configs/middleware/storage/aliyun_oss_storage_config.py
+++ b/api/configs/middleware/storage/aliyun_oss_storage_config.py
@@ -6,40 +6,40 @@
 
 class AliyunOSSStorageConfig(BaseSettings):
     """
-    Aliyun storage configs
+    Configuration settings for Aliyun Object Storage Service (OSS)
     """
 
     ALIYUN_OSS_BUCKET_NAME: Optional[str] = Field(
-        description="Aliyun OSS bucket name",
+        description="Name of the Aliyun OSS bucket to store and retrieve objects",
         default=None,
     )
 
     ALIYUN_OSS_ACCESS_KEY: Optional[str] = Field(
-        description="Aliyun OSS access key",
+        description="Access key ID for authenticating with Aliyun OSS",
         default=None,
     )
 
     ALIYUN_OSS_SECRET_KEY: Optional[str] = Field(
-        description="Aliyun OSS secret key",
+        description="Secret access key for authenticating with Aliyun OSS",
         default=None,
     )
 
     ALIYUN_OSS_ENDPOINT: Optional[str] = Field(
-        description="Aliyun OSS endpoint URL",
+        description="URL of the Aliyun OSS endpoint for your chosen region",
         default=None,
     )
 
     ALIYUN_OSS_REGION: Optional[str] = Field(
-        description="Aliyun OSS region",
+        description="Aliyun OSS region where your bucket is located (e.g., 'oss-cn-hangzhou')",
         default=None,
     )
 
     ALIYUN_OSS_AUTH_VERSION: Optional[str] = Field(
-        description="Aliyun OSS authentication version",
+        description="Version of the authentication protocol to use with Aliyun OSS (e.g., 'v4')",
         default=None,
     )
 
     ALIYUN_OSS_PATH: Optional[str] = Field(
-        description="Aliyun OSS path",
+        description="Base path within the bucket to store objects (e.g., 'my-app-data/')",
         default=None,
     )
diff --git a/api/configs/middleware/storage/amazon_s3_storage_config.py b/api/configs/middleware/storage/amazon_s3_storage_config.py
index bef9326108709..f2d94b12ffa97 100644
--- a/api/configs/middleware/storage/amazon_s3_storage_config.py
+++ b/api/configs/middleware/storage/amazon_s3_storage_config.py
@@ -6,40 +6,40 @@
 
 class S3StorageConfig(BaseSettings):
     """
-    S3 storage configs
+    Configuration settings for S3-compatible object storage
     """
 
     S3_ENDPOINT: Optional[str] = Field(
-        description="S3 storage endpoint",
+        description="URL of the S3-compatible storage endpoint (e.g., 'https://s3.amazonaws.com')",
         default=None,
     )
 
     S3_REGION: Optional[str] = Field(
-        description="S3 storage region",
+        description="Region where the S3 bucket is located (e.g., 'us-east-1')",
         default=None,
     )
 
     S3_BUCKET_NAME: Optional[str] = Field(
-        description="S3 storage bucket name",
+        description="Name of the S3 bucket to store and retrieve objects",
         default=None,
     )
 
     S3_ACCESS_KEY: Optional[str] = Field(
-        description="S3 storage access key",
+        description="Access key ID for authenticating with the S3 service",
         default=None,
     )
 
     S3_SECRET_KEY: Optional[str] = Field(
-        description="S3 storage secret key",
+        description="Secret access key for authenticating with the S3 service",
         default=None,
     )
 
     S3_ADDRESS_STYLE: str = Field(
-        description="S3 storage address style",
+        description="S3 addressing style: 'auto', 'path', or 'virtual'",
         default="auto",
     )
 
     S3_USE_AWS_MANAGED_IAM: bool = Field(
-        description="whether to use aws managed IAM for S3",
+        description="Use AWS managed IAM roles for authentication instead of access/secret keys",
         default=False,
     )
diff --git a/api/configs/middleware/storage/azure_blob_storage_config.py b/api/configs/middleware/storage/azure_blob_storage_config.py
index 10944b58eddc6..b7ab5247a9d4d 100644
--- a/api/configs/middleware/storage/azure_blob_storage_config.py
+++ b/api/configs/middleware/storage/azure_blob_storage_config.py
@@ -6,25 +6,25 @@
 
 class AzureBlobStorageConfig(BaseSettings):
     """
-    Azure Blob storage configs
+    Configuration settings for Azure Blob Storage
     """
 
     AZURE_BLOB_ACCOUNT_NAME: Optional[str] = Field(
-        description="Azure Blob account name",
+        description="Name of the Azure Storage account (e.g., 'mystorageaccount')",
         default=None,
     )
 
     AZURE_BLOB_ACCOUNT_KEY: Optional[str] = Field(
-        description="Azure Blob account key",
+        description="Access key for authenticating with the Azure Storage account",
         default=None,
     )
 
     AZURE_BLOB_CONTAINER_NAME: Optional[str] = Field(
-        description="Azure Blob container name",
+        description="Name of the Azure Blob container to store and retrieve objects",
         default=None,
     )
 
     AZURE_BLOB_ACCOUNT_URL: Optional[str] = Field(
-        description="Azure Blob account URL",
+        description="URL of the Azure Blob storage endpoint (e.g., 'https://mystorageaccount.blob.core.windows.net')",
         default=None,
     )
diff --git a/api/configs/middleware/storage/google_cloud_storage_config.py b/api/configs/middleware/storage/google_cloud_storage_config.py
index 10a2d97e8dbcd..e5d763d7f5c61 100644
--- a/api/configs/middleware/storage/google_cloud_storage_config.py
+++ b/api/configs/middleware/storage/google_cloud_storage_config.py
@@ -6,15 +6,15 @@
 
 class GoogleCloudStorageConfig(BaseSettings):
     """
-    Google Cloud storage configs
+    Configuration settings for Google Cloud Storage
     """
 
     GOOGLE_STORAGE_BUCKET_NAME: Optional[str] = Field(
-        description="Google Cloud storage bucket name",
+        description="Name of the Google Cloud Storage bucket to store and retrieve objects (e.g., 'my-gcs-bucket')",
         default=None,
     )
 
     GOOGLE_STORAGE_SERVICE_ACCOUNT_JSON_BASE64: Optional[str] = Field(
-        description="Google Cloud storage service account json base64",
+        description="Base64-encoded JSON key file for Google Cloud service account authentication",
         default=None,
     )
diff --git a/api/configs/middleware/storage/huawei_obs_storage_config.py b/api/configs/middleware/storage/huawei_obs_storage_config.py
index c5cb379caee3c..3e9e7543ab2ba 100644
--- a/api/configs/middleware/storage/huawei_obs_storage_config.py
+++ b/api/configs/middleware/storage/huawei_obs_storage_config.py
@@ -5,25 +5,25 @@
 
 class HuaweiCloudOBSStorageConfig(BaseModel):
     """
-    Huawei Cloud OBS storage configs
+    Configuration settings for Huawei Cloud Object Storage Service (OBS)
     """
 
     HUAWEI_OBS_BUCKET_NAME: Optional[str] = Field(
-        description="Huawei Cloud OBS bucket name",
+        description="Name of the Huawei Cloud OBS bucket to store and retrieve objects (e.g., 'my-obs-bucket')",
         default=None,
     )
 
     HUAWEI_OBS_ACCESS_KEY: Optional[str] = Field(
-        description="Huawei Cloud OBS Access key",
+        description="Access Key ID for authenticating with Huawei Cloud OBS",
         default=None,
     )
 
     HUAWEI_OBS_SECRET_KEY: Optional[str] = Field(
-        description="Huawei Cloud OBS Secret key",
+        description="Secret Access Key for authenticating with Huawei Cloud OBS",
         default=None,
     )
 
     HUAWEI_OBS_SERVER: Optional[str] = Field(
-        description="Huawei Cloud OBS server URL",
+        description="Endpoint URL for Huawei Cloud OBS (e.g., 'https://obs.cn-north-4.myhuaweicloud.com')",
         default=None,
     )
diff --git a/api/configs/middleware/storage/oci_storage_config.py b/api/configs/middleware/storage/oci_storage_config.py
index f8993496c952b..edc245bcac59b 100644
--- a/api/configs/middleware/storage/oci_storage_config.py
+++ b/api/configs/middleware/storage/oci_storage_config.py
@@ -6,30 +6,30 @@
 
 class OCIStorageConfig(BaseSettings):
     """
-    OCI storage configs
+    Configuration settings for Oracle Cloud Infrastructure (OCI) Object Storage
     """
 
     OCI_ENDPOINT: Optional[str] = Field(
-        description="OCI storage endpoint",
+        description="URL of the OCI Object Storage endpoint (e.g., 'https://objectstorage.us-phoenix-1.oraclecloud.com')",
         default=None,
     )
 
     OCI_REGION: Optional[str] = Field(
-        description="OCI storage region",
+        description="OCI region where the bucket is located (e.g., 'us-phoenix-1')",
         default=None,
     )
 
     OCI_BUCKET_NAME: Optional[str] = Field(
-        description="OCI storage bucket name",
+        description="Name of the OCI Object Storage bucket to store and retrieve objects (e.g., 'my-oci-bucket')",
         default=None,
     )
 
     OCI_ACCESS_KEY: Optional[str] = Field(
-        description="OCI storage access key",
+        description="Access key (also known as API key) for authenticating with OCI Object Storage",
         default=None,
     )
 
     OCI_SECRET_KEY: Optional[str] = Field(
-        description="OCI storage secret key",
+        description="Secret key associated with the access key for authenticating with OCI Object Storage",
         default=None,
     )
diff --git a/api/configs/middleware/storage/tencent_cos_storage_config.py b/api/configs/middleware/storage/tencent_cos_storage_config.py
index 765ac08f3e404..255c4e8938e0f 100644
--- a/api/configs/middleware/storage/tencent_cos_storage_config.py
+++ b/api/configs/middleware/storage/tencent_cos_storage_config.py
@@ -6,30 +6,30 @@
 
 class TencentCloudCOSStorageConfig(BaseSettings):
     """
-    Tencent Cloud COS storage configs
+    Configuration settings for Tencent Cloud Object Storage (COS)
     """
 
     TENCENT_COS_BUCKET_NAME: Optional[str] = Field(
-        description="Tencent Cloud COS bucket name",
+        description="Name of the Tencent Cloud COS bucket to store and retrieve objects",
         default=None,
     )
 
     TENCENT_COS_REGION: Optional[str] = Field(
-        description="Tencent Cloud COS region",
+        description="Tencent Cloud region where the COS bucket is located (e.g., 'ap-guangzhou')",
         default=None,
     )
 
     TENCENT_COS_SECRET_ID: Optional[str] = Field(
-        description="Tencent Cloud COS secret id",
+        description="SecretId for authenticating with Tencent Cloud COS (part of API credentials)",
         default=None,
     )
 
     TENCENT_COS_SECRET_KEY: Optional[str] = Field(
-        description="Tencent Cloud COS secret key",
+        description="SecretKey for authenticating with Tencent Cloud COS (part of API credentials)",
         default=None,
     )
 
     TENCENT_COS_SCHEME: Optional[str] = Field(
-        description="Tencent Cloud COS scheme",
+        description="Protocol scheme for COS requests: 'https' (recommended) or 'http'",
         default=None,
     )
diff --git a/api/configs/middleware/storage/volcengine_tos_storage_config.py b/api/configs/middleware/storage/volcengine_tos_storage_config.py
index a0e09a3cc7671..89ea885002300 100644
--- a/api/configs/middleware/storage/volcengine_tos_storage_config.py
+++ b/api/configs/middleware/storage/volcengine_tos_storage_config.py
@@ -5,30 +5,30 @@
 
 class VolcengineTOSStorageConfig(BaseModel):
     """
-    Volcengine tos storage configs
+    Configuration settings for Volcengine Tinder Object Storage (TOS)
     """
 
     VOLCENGINE_TOS_BUCKET_NAME: Optional[str] = Field(
-        description="Volcengine TOS Bucket Name",
+        description="Name of the Volcengine TOS bucket to store and retrieve objects (e.g., 'my-tos-bucket')",
         default=None,
     )
 
     VOLCENGINE_TOS_ACCESS_KEY: Optional[str] = Field(
-        description="Volcengine TOS Access Key",
+        description="Access Key ID for authenticating with Volcengine TOS",
         default=None,
     )
 
     VOLCENGINE_TOS_SECRET_KEY: Optional[str] = Field(
-        description="Volcengine TOS Secret Key",
+        description="Secret Access Key for authenticating with Volcengine TOS",
         default=None,
     )
 
     VOLCENGINE_TOS_ENDPOINT: Optional[str] = Field(
-        description="Volcengine TOS Endpoint URL",
+        description="URL of the Volcengine TOS endpoint (e.g., 'https://tos-cn-beijing.volces.com')",
         default=None,
     )
 
     VOLCENGINE_TOS_REGION: Optional[str] = Field(
-        description="Volcengine TOS Region",
+        description="Volcengine region where the TOS bucket is located (e.g., 'cn-beijing')",
         default=None,
     )
diff --git a/api/configs/middleware/vdb/analyticdb_config.py b/api/configs/middleware/vdb/analyticdb_config.py
index 04f5b0e5bf769..247a8ea555948 100644
--- a/api/configs/middleware/vdb/analyticdb_config.py
+++ b/api/configs/middleware/vdb/analyticdb_config.py
@@ -5,33 +5,38 @@
 
 class AnalyticdbConfig(BaseModel):
     """
-    Configuration for connecting to AnalyticDB.
+    Configuration for connecting to Alibaba Cloud AnalyticDB for PostgreSQL.
     Refer to the following documentation for details on obtaining credentials:
     https://www.alibabacloud.com/help/en/analyticdb-for-postgresql/getting-started/create-an-instance-instances-with-vector-engine-optimization-enabled
     """
 
     ANALYTICDB_KEY_ID: Optional[str] = Field(
-        default=None, description="The Access Key ID provided by Alibaba Cloud for authentication."
+        default=None, description="The Access Key ID provided by Alibaba Cloud for API authentication."
     )
     ANALYTICDB_KEY_SECRET: Optional[str] = Field(
-        default=None, description="The Secret Access Key corresponding to the Access Key ID for secure access."
+        default=None, description="The Secret Access Key corresponding to the Access Key ID for secure API access."
     )
     ANALYTICDB_REGION_ID: Optional[str] = Field(
-        default=None, description="The region where the AnalyticDB instance is deployed (e.g., 'cn-hangzhou')."
+        default=None,
+        description="The region where the AnalyticDB instance is deployed (e.g., 'cn-hangzhou', 'ap-southeast-1').",
     )
     ANALYTICDB_INSTANCE_ID: Optional[str] = Field(
         default=None,
-        description="The unique identifier of the AnalyticDB instance you want to connect to (e.g., 'gp-ab123456')..",
+        description="The unique identifier of the AnalyticDB instance you want to connect to.",
     )
     ANALYTICDB_ACCOUNT: Optional[str] = Field(
-        default=None, description="The account name used to log in to the AnalyticDB instance."
+        default=None,
+        description="The account name used to log in to the AnalyticDB instance"
+        " (usually the initial account created with the instance).",
     )
     ANALYTICDB_PASSWORD: Optional[str] = Field(
-        default=None, description="The password associated with the AnalyticDB account for authentication."
+        default=None, description="The password associated with the AnalyticDB account for database authentication."
     )
     ANALYTICDB_NAMESPACE: Optional[str] = Field(
-        default=None, description="The namespace within AnalyticDB for schema isolation."
+        default=None, description="The namespace within AnalyticDB for schema isolation (if using namespace feature)."
     )
     ANALYTICDB_NAMESPACE_PASSWORD: Optional[str] = Field(
-        default=None, description="The password for accessing the specified namespace within the AnalyticDB instance."
+        default=None,
+        description="The password for accessing the specified namespace within the AnalyticDB instance"
+        " (if namespace feature is enabled).",
     )
diff --git a/api/configs/middleware/vdb/chroma_config.py b/api/configs/middleware/vdb/chroma_config.py
index d386623a569d3..e83a9902dee90 100644
--- a/api/configs/middleware/vdb/chroma_config.py
+++ b/api/configs/middleware/vdb/chroma_config.py
@@ -6,35 +6,35 @@
 
 class ChromaConfig(BaseSettings):
     """
-    Chroma configs
+    Configuration settings for Chroma vector database
     """
 
     CHROMA_HOST: Optional[str] = Field(
-        description="Chroma host",
+        description="Hostname or IP address of the Chroma server (e.g., 'localhost' or '192.168.1.100')",
         default=None,
     )
 
     CHROMA_PORT: PositiveInt = Field(
-        description="Chroma port",
+        description="Port number on which the Chroma server is listening (default is 8000)",
         default=8000,
     )
 
     CHROMA_TENANT: Optional[str] = Field(
-        description="Chroma database",
+        description="Tenant identifier for multi-tenancy support in Chroma",
         default=None,
     )
 
     CHROMA_DATABASE: Optional[str] = Field(
-        description="Chroma database",
+        description="Name of the Chroma database to connect to",
         default=None,
     )
 
     CHROMA_AUTH_PROVIDER: Optional[str] = Field(
-        description="Chroma authentication provider",
+        description="Authentication provider for Chroma (e.g., 'basic', 'token', or a custom provider)",
         default=None,
     )
 
     CHROMA_AUTH_CREDENTIALS: Optional[str] = Field(
-        description="Chroma authentication credentials",
+        description="Authentication credentials for Chroma (format depends on the auth provider)",
         default=None,
     )
diff --git a/api/configs/middleware/vdb/elasticsearch_config.py b/api/configs/middleware/vdb/elasticsearch_config.py
index 5b6a8fd939c29..df8182985dc19 100644
--- a/api/configs/middleware/vdb/elasticsearch_config.py
+++ b/api/configs/middleware/vdb/elasticsearch_config.py
@@ -6,25 +6,25 @@
 
 class ElasticsearchConfig(BaseSettings):
     """
-    Elasticsearch configs
+    Configuration settings for Elasticsearch
     """
 
     ELASTICSEARCH_HOST: Optional[str] = Field(
-        description="Elasticsearch host",
+        description="Hostname or IP address of the Elasticsearch server (e.g., 'localhost' or '192.168.1.100')",
         default="127.0.0.1",
     )
 
     ELASTICSEARCH_PORT: PositiveInt = Field(
-        description="Elasticsearch port",
+        description="Port number on which the Elasticsearch server is listening (default is 9200)",
         default=9200,
     )
 
     ELASTICSEARCH_USERNAME: Optional[str] = Field(
-        description="Elasticsearch username",
+        description="Username for authenticating with Elasticsearch (default is 'elastic')",
         default="elastic",
     )
 
     ELASTICSEARCH_PASSWORD: Optional[str] = Field(
-        description="Elasticsearch password",
+        description="Password for authenticating with Elasticsearch (default is 'elastic')",
         default="elastic",
     )
diff --git a/api/configs/middleware/vdb/milvus_config.py b/api/configs/middleware/vdb/milvus_config.py
index 98d375966a194..231cbbbe8ffc9 100644
--- a/api/configs/middleware/vdb/milvus_config.py
+++ b/api/configs/middleware/vdb/milvus_config.py
@@ -6,30 +6,30 @@
 
 class MilvusConfig(BaseSettings):
     """
-    Milvus configs
+    Configuration settings for Milvus vector database
     """
 
     MILVUS_URI: Optional[str] = Field(
-        description="Milvus uri",
+        description="URI for connecting to the Milvus server (e.g., 'http://localhost:19530' or 'https://milvus-instance.example.com:19530')",
         default="http://127.0.0.1:19530",
     )
 
     MILVUS_TOKEN: Optional[str] = Field(
-        description="Milvus token",
+        description="Authentication token for Milvus, if token-based authentication is enabled",
         default=None,
     )
 
     MILVUS_USER: Optional[str] = Field(
-        description="Milvus user",
+        description="Username for authenticating with Milvus, if username/password authentication is enabled",
         default=None,
     )
 
     MILVUS_PASSWORD: Optional[str] = Field(
-        description="Milvus password",
+        description="Password for authenticating with Milvus, if username/password authentication is enabled",
         default=None,
     )
 
     MILVUS_DATABASE: str = Field(
-        description="Milvus database, default to `default`",
+        description="Name of the Milvus database to connect to (default is 'default')",
         default="default",
     )
diff --git a/api/configs/middleware/vdb/myscale_config.py b/api/configs/middleware/vdb/myscale_config.py
index 6451d26e1c65d..5896c19d27d11 100644
--- a/api/configs/middleware/vdb/myscale_config.py
+++ b/api/configs/middleware/vdb/myscale_config.py
@@ -3,35 +3,35 @@
 
 class MyScaleConfig(BaseModel):
     """
-    MyScale configs
+    Configuration settings for MyScale vector database
     """
 
     MYSCALE_HOST: str = Field(
-        description="MyScale host",
+        description="Hostname or IP address of the MyScale server (e.g., 'localhost' or 'myscale.example.com')",
         default="localhost",
     )
 
     MYSCALE_PORT: PositiveInt = Field(
-        description="MyScale port",
+        description="Port number on which the MyScale server is listening (default is 8123)",
         default=8123,
     )
 
     MYSCALE_USER: str = Field(
-        description="MyScale user",
+        description="Username for authenticating with MyScale (default is 'default')",
         default="default",
     )
 
     MYSCALE_PASSWORD: str = Field(
-        description="MyScale password",
+        description="Password for authenticating with MyScale (default is an empty string)",
         default="",
     )
 
     MYSCALE_DATABASE: str = Field(
-        description="MyScale database name",
+        description="Name of the MyScale database to connect to (default is 'default')",
         default="default",
     )
 
     MYSCALE_FTS_PARAMS: str = Field(
-        description="MyScale fts index parameters",
+        description="Additional parameters for MyScale Full Text Search index)",
         default="",
     )
diff --git a/api/configs/middleware/vdb/opensearch_config.py b/api/configs/middleware/vdb/opensearch_config.py
index 5823dc1433d83..81dde4c04d472 100644
--- a/api/configs/middleware/vdb/opensearch_config.py
+++ b/api/configs/middleware/vdb/opensearch_config.py
@@ -6,30 +6,30 @@
 
 class OpenSearchConfig(BaseSettings):
     """
-    OpenSearch configs
+    Configuration settings for OpenSearch
     """
 
     OPENSEARCH_HOST: Optional[str] = Field(
-        description="OpenSearch host",
+        description="Hostname or IP address of the OpenSearch server (e.g., 'localhost' or 'opensearch.example.com')",
         default=None,
     )
 
     OPENSEARCH_PORT: PositiveInt = Field(
-        description="OpenSearch port",
+        description="Port number on which the OpenSearch server is listening (default is 9200)",
         default=9200,
     )
 
     OPENSEARCH_USER: Optional[str] = Field(
-        description="OpenSearch user",
+        description="Username for authenticating with OpenSearch",
         default=None,
     )
 
     OPENSEARCH_PASSWORD: Optional[str] = Field(
-        description="OpenSearch password",
+        description="Password for authenticating with OpenSearch",
         default=None,
     )
 
     OPENSEARCH_SECURE: bool = Field(
-        description="whether to use SSL connection for OpenSearch",
+        description="Whether to use SSL/TLS encrypted connection for OpenSearch (True for HTTPS, False for HTTP)",
         default=False,
     )
diff --git a/api/configs/middleware/vdb/oracle_config.py b/api/configs/middleware/vdb/oracle_config.py
index 62614ae870cc8..44e2f133458f1 100644
--- a/api/configs/middleware/vdb/oracle_config.py
+++ b/api/configs/middleware/vdb/oracle_config.py
@@ -6,30 +6,30 @@
 
 class OracleConfig(BaseSettings):
     """
-    ORACLE configs
+    Configuration settings for Oracle database
     """
 
     ORACLE_HOST: Optional[str] = Field(
-        description="ORACLE host",
+        description="Hostname or IP address of the Oracle database server (e.g., 'localhost' or 'oracle.example.com')",
         default=None,
     )
 
     ORACLE_PORT: Optional[PositiveInt] = Field(
-        description="ORACLE port",
+        description="Port number on which the Oracle database server is listening (default is 1521)",
         default=1521,
     )
 
     ORACLE_USER: Optional[str] = Field(
-        description="ORACLE user",
+        description="Username for authenticating with the Oracle database",
         default=None,
     )
 
     ORACLE_PASSWORD: Optional[str] = Field(
-        description="ORACLE password",
+        description="Password for authenticating with the Oracle database",
         default=None,
     )
 
     ORACLE_DATABASE: Optional[str] = Field(
-        description="ORACLE database",
+        description="Name of the Oracle database or service to connect to (e.g., 'ORCL' or 'pdborcl')",
         default=None,
     )
diff --git a/api/configs/middleware/vdb/pgvector_config.py b/api/configs/middleware/vdb/pgvector_config.py
index 39a7c1d8d5e4a..395dcaa4208a9 100644
--- a/api/configs/middleware/vdb/pgvector_config.py
+++ b/api/configs/middleware/vdb/pgvector_config.py
@@ -6,30 +6,30 @@
 
 class PGVectorConfig(BaseSettings):
     """
-    PGVector configs
+    Configuration settings for PGVector (PostgreSQL with vector extension)
     """
 
     PGVECTOR_HOST: Optional[str] = Field(
-        description="PGVector host",
+        description="Hostname or IP address of the PostgreSQL server with PGVector extension (e.g., 'localhost')",
         default=None,
     )
 
     PGVECTOR_PORT: Optional[PositiveInt] = Field(
-        description="PGVector port",
+        description="Port number on which the PostgreSQL server is listening (default is 5433)",
         default=5433,
     )
 
     PGVECTOR_USER: Optional[str] = Field(
-        description="PGVector user",
+        description="Username for authenticating with the PostgreSQL database",
         default=None,
     )
 
     PGVECTOR_PASSWORD: Optional[str] = Field(
-        description="PGVector password",
+        description="Password for authenticating with the PostgreSQL database",
         default=None,
     )
 
     PGVECTOR_DATABASE: Optional[str] = Field(
-        description="PGVector database",
+        description="Name of the PostgreSQL database to connect to",
         default=None,
     )
diff --git a/api/configs/middleware/vdb/pgvectors_config.py b/api/configs/middleware/vdb/pgvectors_config.py
index c40e5ff92103f..8d7a4b8d25511 100644
--- a/api/configs/middleware/vdb/pgvectors_config.py
+++ b/api/configs/middleware/vdb/pgvectors_config.py
@@ -6,30 +6,30 @@
 
 class PGVectoRSConfig(BaseSettings):
     """
-    PGVectoRS configs
+    Configuration settings for PGVecto.RS (Rust-based vector extension for PostgreSQL)
     """
 
     PGVECTO_RS_HOST: Optional[str] = Field(
-        description="PGVectoRS host",
+        description="Hostname or IP address of the PostgreSQL server with PGVecto.RS extension (e.g., 'localhost')",
         default=None,
     )
 
     PGVECTO_RS_PORT: Optional[PositiveInt] = Field(
-        description="PGVectoRS port",
+        description="Port number on which the PostgreSQL server with PGVecto.RS is listening (default is 5431)",
         default=5431,
     )
 
     PGVECTO_RS_USER: Optional[str] = Field(
-        description="PGVectoRS user",
+        description="Username for authenticating with the PostgreSQL database using PGVecto.RS",
         default=None,
     )
 
     PGVECTO_RS_PASSWORD: Optional[str] = Field(
-        description="PGVectoRS password",
+        description="Password for authenticating with the PostgreSQL database using PGVecto.RS",
         default=None,
     )
 
     PGVECTO_RS_DATABASE: Optional[str] = Field(
-        description="PGVectoRS database",
+        description="Name of the PostgreSQL database with PGVecto.RS extension to connect to",
         default=None,
     )
diff --git a/api/configs/middleware/vdb/qdrant_config.py b/api/configs/middleware/vdb/qdrant_config.py
index 27f75491c9876..b70f6246523c5 100644
--- a/api/configs/middleware/vdb/qdrant_config.py
+++ b/api/configs/middleware/vdb/qdrant_config.py
@@ -6,30 +6,30 @@
 
 class QdrantConfig(BaseSettings):
     """
-    Qdrant configs
+    Configuration settings for Qdrant vector database
     """
 
     QDRANT_URL: Optional[str] = Field(
-        description="Qdrant url",
+        description="URL of the Qdrant server (e.g., 'http://localhost:6333' or 'https://qdrant.example.com')",
         default=None,
     )
 
     QDRANT_API_KEY: Optional[str] = Field(
-        description="Qdrant api key",
+        description="API key for authenticating with the Qdrant server",
         default=None,
     )
 
     QDRANT_CLIENT_TIMEOUT: NonNegativeInt = Field(
-        description="Qdrant client timeout in seconds",
+        description="Timeout in seconds for Qdrant client operations (default is 20 seconds)",
         default=20,
     )
 
     QDRANT_GRPC_ENABLED: bool = Field(
-        description="whether enable grpc support for Qdrant connection",
+        description="Whether to enable gRPC support for Qdrant connection (True for gRPC, False for HTTP)",
         default=False,
     )
 
     QDRANT_GRPC_PORT: PositiveInt = Field(
-        description="Qdrant grpc port",
+        description="Port number for gRPC connection to Qdrant server (default is 6334)",
         default=6334,
     )
diff --git a/api/configs/middleware/vdb/relyt_config.py b/api/configs/middleware/vdb/relyt_config.py
index 66b9ecc03f6aa..5ffbea7b19bb8 100644
--- a/api/configs/middleware/vdb/relyt_config.py
+++ b/api/configs/middleware/vdb/relyt_config.py
@@ -6,30 +6,30 @@
 
 class RelytConfig(BaseSettings):
     """
-    Relyt configs
+    Configuration settings for Relyt database
     """
 
     RELYT_HOST: Optional[str] = Field(
-        description="Relyt host",
+        description="Hostname or IP address of the Relyt server (e.g., 'localhost' or 'relyt.example.com')",
         default=None,
     )
 
     RELYT_PORT: PositiveInt = Field(
-        description="Relyt port",
+        description="Port number on which the Relyt server is listening (default is 9200)",
         default=9200,
     )
 
     RELYT_USER: Optional[str] = Field(
-        description="Relyt user",
+        description="Username for authenticating with the Relyt database",
         default=None,
     )
 
     RELYT_PASSWORD: Optional[str] = Field(
-        description="Relyt password",
+        description="Password for authenticating with the Relyt database",
         default=None,
     )
 
     RELYT_DATABASE: Optional[str] = Field(
-        description="Relyt database",
+        description="Name of the Relyt database to connect to (default is 'default')",
         default="default",
     )
diff --git a/api/configs/middleware/vdb/tencent_vector_config.py b/api/configs/middleware/vdb/tencent_vector_config.py
index 46b4cb6a24ff8..9cf4d07f6fe66 100644
--- a/api/configs/middleware/vdb/tencent_vector_config.py
+++ b/api/configs/middleware/vdb/tencent_vector_config.py
@@ -6,45 +6,45 @@
 
 class TencentVectorDBConfig(BaseSettings):
     """
-    Tencent Vector configs
+    Configuration settings for Tencent Vector Database
     """
 
     TENCENT_VECTOR_DB_URL: Optional[str] = Field(
-        description="Tencent Vector URL",
+        description="URL of the Tencent Vector Database service (e.g., 'https://vectordb.tencentcloudapi.com')",
         default=None,
     )
 
     TENCENT_VECTOR_DB_API_KEY: Optional[str] = Field(
-        description="Tencent Vector API key",
+        description="API key for authenticating with the Tencent Vector Database service",
         default=None,
     )
 
     TENCENT_VECTOR_DB_TIMEOUT: PositiveInt = Field(
-        description="Tencent Vector timeout in seconds",
+        description="Timeout in seconds for Tencent Vector Database operations (default is 30 seconds)",
         default=30,
     )
 
     TENCENT_VECTOR_DB_USERNAME: Optional[str] = Field(
-        description="Tencent Vector username",
+        description="Username for authenticating with the Tencent Vector Database (if required)",
         default=None,
     )
 
     TENCENT_VECTOR_DB_PASSWORD: Optional[str] = Field(
-        description="Tencent Vector password",
+        description="Password for authenticating with the Tencent Vector Database (if required)",
         default=None,
     )
 
     TENCENT_VECTOR_DB_SHARD: PositiveInt = Field(
-        description="Tencent Vector sharding number",
+        description="Number of shards for the Tencent Vector Database (default is 1)",
         default=1,
     )
 
     TENCENT_VECTOR_DB_REPLICAS: NonNegativeInt = Field(
-        description="Tencent Vector replicas",
+        description="Number of replicas for the Tencent Vector Database (default is 2)",
         default=2,
     )
 
     TENCENT_VECTOR_DB_DATABASE: Optional[str] = Field(
-        description="Tencent Vector Database",
+        description="Name of the specific Tencent Vector Database to connect to",
         default=None,
     )
diff --git a/api/configs/middleware/vdb/tidb_vector_config.py b/api/configs/middleware/vdb/tidb_vector_config.py
index dbcb276c01a1f..bc68be69d86ad 100644
--- a/api/configs/middleware/vdb/tidb_vector_config.py
+++ b/api/configs/middleware/vdb/tidb_vector_config.py
@@ -6,30 +6,30 @@
 
 class TiDBVectorConfig(BaseSettings):
     """
-    TiDB Vector configs
+    Configuration settings for TiDB Vector database
     """
 
     TIDB_VECTOR_HOST: Optional[str] = Field(
-        description="TiDB Vector host",
+        description="Hostname or IP address of the TiDB Vector server (e.g., 'localhost' or 'tidb.example.com')",
         default=None,
     )
 
     TIDB_VECTOR_PORT: Optional[PositiveInt] = Field(
-        description="TiDB Vector port",
+        description="Port number on which the TiDB Vector server is listening (default is 4000)",
         default=4000,
     )
 
     TIDB_VECTOR_USER: Optional[str] = Field(
-        description="TiDB Vector user",
+        description="Username for authenticating with the TiDB Vector database",
         default=None,
     )
 
     TIDB_VECTOR_PASSWORD: Optional[str] = Field(
-        description="TiDB Vector password",
+        description="Password for authenticating with the TiDB Vector database",
         default=None,
     )
 
     TIDB_VECTOR_DATABASE: Optional[str] = Field(
-        description="TiDB Vector database",
+        description="Name of the TiDB Vector database to connect to",
         default=None,
     )
diff --git a/api/configs/middleware/vdb/weaviate_config.py b/api/configs/middleware/vdb/weaviate_config.py
index 63d1022f6a451..25000e8bde290 100644
--- a/api/configs/middleware/vdb/weaviate_config.py
+++ b/api/configs/middleware/vdb/weaviate_config.py
@@ -6,25 +6,25 @@
 
 class WeaviateConfig(BaseSettings):
     """
-    Weaviate configs
+    Configuration settings for Weaviate vector database
     """
 
     WEAVIATE_ENDPOINT: Optional[str] = Field(
-        description="Weaviate endpoint URL",
+        description="URL of the Weaviate server (e.g., 'http://localhost:8080' or 'https://weaviate.example.com')",
         default=None,
     )
 
     WEAVIATE_API_KEY: Optional[str] = Field(
-        description="Weaviate API key",
+        description="API key for authenticating with the Weaviate server",
         default=None,
     )
 
     WEAVIATE_GRPC_ENABLED: bool = Field(
-        description="whether to enable gRPC for Weaviate connection",
+        description="Whether to enable gRPC for Weaviate connection (True for gRPC, False for HTTP)",
         default=True,
     )
 
     WEAVIATE_BATCH_SIZE: PositiveInt = Field(
-        description="Weaviate batch size",
+        description="Number of objects to be processed in a single batch operation (default is 100)",
         default=100,
     )

From 8fd297f8b406820fe5673a5d27338cbc30f47834 Mon Sep 17 00:00:00 2001
From: omr <145922434+y-omr@users.noreply.github.com>
Date: Sun, 22 Sep 2024 14:39:41 +0900
Subject: [PATCH 022/235] fix: redundant check for available_document_count
 (#8491)

---
 api/core/rag/retrieval/dataset_retrieval.py | 4 ++--
 1 file changed, 2 insertions(+), 2 deletions(-)

diff --git a/api/core/rag/retrieval/dataset_retrieval.py b/api/core/rag/retrieval/dataset_retrieval.py
index 286ecd4c03684..4603957d68599 100644
--- a/api/core/rag/retrieval/dataset_retrieval.py
+++ b/api/core/rag/retrieval/dataset_retrieval.py
@@ -110,7 +110,7 @@ def retrieve(
                 continue
 
             # pass if dataset is not available
-            if dataset and dataset.available_document_count == 0 and dataset.available_document_count == 0:
+            if dataset and dataset.available_document_count == 0:
                 continue
 
             available_datasets.append(dataset)
@@ -468,7 +468,7 @@ def to_dataset_retriever_tool(
                 continue
 
             # pass if dataset is not available
-            if dataset and dataset.available_document_count == 0 and dataset.available_document_count == 0:
+            if dataset and dataset.available_document_count == 0:
                 continue
 
             available_datasets.append(dataset)

From 6e37750fbd2b466765e9adacdfe23fcff4936f10 Mon Sep 17 00:00:00 2001
From: Nam Vu <zuzoovn@gmail.com>
Date: Sun, 22 Sep 2024 12:41:09 +0700
Subject: [PATCH 023/235] fix: commands.py (#8483)

---
 api/commands.py | 130 +++++++++++++++++++++++-------------------------
 1 file changed, 62 insertions(+), 68 deletions(-)

diff --git a/api/commands.py b/api/commands.py
index 3a6b4963cfa68..b8fc81af673af 100644
--- a/api/commands.py
+++ b/api/commands.py
@@ -28,28 +28,28 @@
 
 
 @click.command("reset-password", help="Reset the account password.")
-@click.option("--email", prompt=True, help="The email address of the account whose password you need to reset")
-@click.option("--new-password", prompt=True, help="the new password.")
-@click.option("--password-confirm", prompt=True, help="the new password confirm.")
+@click.option("--email", prompt=True, help="Account email to reset password for")
+@click.option("--new-password", prompt=True, help="New password")
+@click.option("--password-confirm", prompt=True, help="Confirm new password")
 def reset_password(email, new_password, password_confirm):
     """
     Reset password of owner account
     Only available in SELF_HOSTED mode
     """
     if str(new_password).strip() != str(password_confirm).strip():
-        click.echo(click.style("sorry. The two passwords do not match.", fg="red"))
+        click.echo(click.style("Passwords do not match.", fg="red"))
         return
 
     account = db.session.query(Account).filter(Account.email == email).one_or_none()
 
     if not account:
-        click.echo(click.style("sorry. the account: [{}] not exist .".format(email), fg="red"))
+        click.echo(click.style("Account not found for email: {}".format(email), fg="red"))
         return
 
     try:
         valid_password(new_password)
     except:
-        click.echo(click.style("sorry. The passwords must match {} ".format(password_pattern), fg="red"))
+        click.echo(click.style("Invalid password. Must match {}".format(password_pattern), fg="red"))
         return
 
     # generate password salt
@@ -62,37 +62,37 @@ def reset_password(email, new_password, password_confirm):
     account.password = base64_password_hashed
     account.password_salt = base64_salt
     db.session.commit()
-    click.echo(click.style("Congratulations! Password has been reset.", fg="green"))
+    click.echo(click.style("Password reset successfully.", fg="green"))
 
 
 @click.command("reset-email", help="Reset the account email.")
-@click.option("--email", prompt=True, help="The old email address of the account whose email you need to reset")
-@click.option("--new-email", prompt=True, help="the new email.")
-@click.option("--email-confirm", prompt=True, help="the new email confirm.")
+@click.option("--email", prompt=True, help="Current account email")
+@click.option("--new-email", prompt=True, help="New email")
+@click.option("--email-confirm", prompt=True, help="Confirm new email")
 def reset_email(email, new_email, email_confirm):
     """
     Replace account email
     :return:
     """
     if str(new_email).strip() != str(email_confirm).strip():
-        click.echo(click.style("Sorry, new email and confirm email do not match.", fg="red"))
+        click.echo(click.style("New emails do not match.", fg="red"))
         return
 
     account = db.session.query(Account).filter(Account.email == email).one_or_none()
 
     if not account:
-        click.echo(click.style("sorry. the account: [{}] not exist .".format(email), fg="red"))
+        click.echo(click.style("Account not found for email: {}".format(email), fg="red"))
         return
 
     try:
         email_validate(new_email)
     except:
-        click.echo(click.style("sorry. {} is not a valid email. ".format(email), fg="red"))
+        click.echo(click.style("Invalid email: {}".format(new_email), fg="red"))
         return
 
     account.email = new_email
     db.session.commit()
-    click.echo(click.style("Congratulations!, email has been reset.", fg="green"))
+    click.echo(click.style("Email updated successfully.", fg="green"))
 
 
 @click.command(
@@ -104,7 +104,7 @@ def reset_email(email, new_email, email_confirm):
 )
 @click.confirmation_option(
     prompt=click.style(
-        "Are you sure you want to reset encrypt key pair? this operation cannot be rolled back!", fg="red"
+        "Are you sure you want to reset encrypt key pair? This operation cannot be rolled back!", fg="red"
     )
 )
 def reset_encrypt_key_pair():
@@ -114,13 +114,13 @@ def reset_encrypt_key_pair():
     Only support SELF_HOSTED mode.
     """
     if dify_config.EDITION != "SELF_HOSTED":
-        click.echo(click.style("Sorry, only support SELF_HOSTED mode.", fg="red"))
+        click.echo(click.style("This command is only for SELF_HOSTED installations.", fg="red"))
         return
 
     tenants = db.session.query(Tenant).all()
     for tenant in tenants:
         if not tenant:
-            click.echo(click.style("Sorry, no workspace found. Please enter /install to initialize.", fg="red"))
+            click.echo(click.style("No workspaces found. Run /install first.", fg="red"))
             return
 
         tenant.encrypt_public_key = generate_key_pair(tenant.id)
@@ -137,7 +137,7 @@ def reset_encrypt_key_pair():
         )
 
 
-@click.command("vdb-migrate", help="migrate vector db.")
+@click.command("vdb-migrate", help="Migrate vector db.")
 @click.option("--scope", default="all", prompt=False, help="The scope of vector database to migrate, Default is All.")
 def vdb_migrate(scope: str):
     if scope in {"knowledge", "all"}:
@@ -150,7 +150,7 @@ def migrate_annotation_vector_database():
     """
     Migrate annotation datas to target vector database .
     """
-    click.echo(click.style("Start migrate annotation data.", fg="green"))
+    click.echo(click.style("Starting annotation data migration.", fg="green"))
     create_count = 0
     skipped_count = 0
     total_count = 0
@@ -174,14 +174,14 @@ def migrate_annotation_vector_database():
                 f"Processing the {total_count} app {app.id}. " + f"{create_count} created, {skipped_count} skipped."
             )
             try:
-                click.echo("Create app annotation index: {}".format(app.id))
+                click.echo("Creating app annotation index: {}".format(app.id))
                 app_annotation_setting = (
                     db.session.query(AppAnnotationSetting).filter(AppAnnotationSetting.app_id == app.id).first()
                 )
 
                 if not app_annotation_setting:
                     skipped_count = skipped_count + 1
-                    click.echo("App annotation setting is disabled: {}".format(app.id))
+                    click.echo("App annotation setting disabled: {}".format(app.id))
                     continue
                 # get dataset_collection_binding info
                 dataset_collection_binding = (
@@ -190,7 +190,7 @@ def migrate_annotation_vector_database():
                     .first()
                 )
                 if not dataset_collection_binding:
-                    click.echo("App annotation collection binding is not exist: {}".format(app.id))
+                    click.echo("App annotation collection binding not found: {}".format(app.id))
                     continue
                 annotations = db.session.query(MessageAnnotation).filter(MessageAnnotation.app_id == app.id).all()
                 dataset = Dataset(
@@ -211,11 +211,11 @@ def migrate_annotation_vector_database():
                         documents.append(document)
 
                 vector = Vector(dataset, attributes=["doc_id", "annotation_id", "app_id"])
-                click.echo(f"Start to migrate annotation, app_id: {app.id}.")
+                click.echo(f"Migrating annotations for app: {app.id}.")
 
                 try:
                     vector.delete()
-                    click.echo(click.style(f"Successfully delete vector index for app: {app.id}.", fg="green"))
+                    click.echo(click.style(f"Deleted vector index for app {app.id}.", fg="green"))
                 except Exception as e:
                     click.echo(click.style(f"Failed to delete vector index for app {app.id}.", fg="red"))
                     raise e
@@ -223,12 +223,12 @@ def migrate_annotation_vector_database():
                     try:
                         click.echo(
                             click.style(
-                                f"Start to created vector index with {len(documents)} annotations for app {app.id}.",
+                                f"Creating vector index with {len(documents)} annotations for app {app.id}.",
                                 fg="green",
                             )
                         )
                         vector.create(documents)
-                        click.echo(click.style(f"Successfully created vector index for app {app.id}.", fg="green"))
+                        click.echo(click.style(f"Created vector index for app {app.id}.", fg="green"))
                     except Exception as e:
                         click.echo(click.style(f"Failed to created vector index for app {app.id}.", fg="red"))
                         raise e
@@ -237,14 +237,14 @@ def migrate_annotation_vector_database():
             except Exception as e:
                 click.echo(
                     click.style(
-                        "Create app annotation index error: {} {}".format(e.__class__.__name__, str(e)), fg="red"
+                        "Error creating app annotation index: {} {}".format(e.__class__.__name__, str(e)), fg="red"
                     )
                 )
                 continue
 
     click.echo(
         click.style(
-            f"Congratulations! Create {create_count} app annotation indexes, and skipped {skipped_count} apps.",
+            f"Migration complete. Created {create_count} app annotation indexes. Skipped {skipped_count} apps.",
             fg="green",
         )
     )
@@ -254,7 +254,7 @@ def migrate_knowledge_vector_database():
     """
     Migrate vector database datas to target vector database .
     """
-    click.echo(click.style("Start migrate vector db.", fg="green"))
+    click.echo(click.style("Starting vector database migration.", fg="green"))
     create_count = 0
     skipped_count = 0
     total_count = 0
@@ -278,7 +278,7 @@ def migrate_knowledge_vector_database():
                 f"Processing the {total_count} dataset {dataset.id}. {create_count} created, {skipped_count} skipped."
             )
             try:
-                click.echo("Create dataset vdb index: {}".format(dataset.id))
+                click.echo("Creating dataset vector database index: {}".format(dataset.id))
                 if dataset.index_struct_dict:
                     if dataset.index_struct_dict["type"] == vector_type:
                         skipped_count = skipped_count + 1
@@ -299,7 +299,7 @@ def migrate_knowledge_vector_database():
                         if dataset_collection_binding:
                             collection_name = dataset_collection_binding.collection_name
                         else:
-                            raise ValueError("Dataset Collection Bindings is not exist!")
+                            raise ValueError("Dataset Collection Binding not found")
                     else:
                         dataset_id = dataset.id
                         collection_name = Dataset.gen_collection_name_by_id(dataset_id)
@@ -351,14 +351,12 @@ def migrate_knowledge_vector_database():
                     raise ValueError(f"Vector store {vector_type} is not supported.")
 
                 vector = Vector(dataset)
-                click.echo(f"Start to migrate dataset {dataset.id}.")
+                click.echo(f"Migrating dataset {dataset.id}.")
 
                 try:
                     vector.delete()
                     click.echo(
-                        click.style(
-                            f"Successfully delete vector index {collection_name} for dataset {dataset.id}.", fg="green"
-                        )
+                        click.style(f"Deleted vector index {collection_name} for dataset {dataset.id}.", fg="green")
                     )
                 except Exception as e:
                     click.echo(
@@ -410,15 +408,13 @@ def migrate_knowledge_vector_database():
                     try:
                         click.echo(
                             click.style(
-                                f"Start to created vector index with {len(documents)} documents of {segments_count}"
+                                f"Creating vector index with {len(documents)} documents of {segments_count}"
                                 f" segments for dataset {dataset.id}.",
                                 fg="green",
                             )
                         )
                         vector.create(documents)
-                        click.echo(
-                            click.style(f"Successfully created vector index for dataset {dataset.id}.", fg="green")
-                        )
+                        click.echo(click.style(f"Created vector index for dataset {dataset.id}.", fg="green"))
                     except Exception as e:
                         click.echo(click.style(f"Failed to created vector index for dataset {dataset.id}.", fg="red"))
                         raise e
@@ -429,13 +425,13 @@ def migrate_knowledge_vector_database():
             except Exception as e:
                 db.session.rollback()
                 click.echo(
-                    click.style("Create dataset index error: {} {}".format(e.__class__.__name__, str(e)), fg="red")
+                    click.style("Error creating dataset index: {} {}".format(e.__class__.__name__, str(e)), fg="red")
                 )
                 continue
 
     click.echo(
         click.style(
-            f"Congratulations! Create {create_count} dataset indexes, and skipped {skipped_count} datasets.", fg="green"
+            f"Migration complete. Created {create_count} dataset indexes. Skipped {skipped_count} datasets.", fg="green"
         )
     )
 
@@ -445,7 +441,7 @@ def convert_to_agent_apps():
     """
     Convert Agent Assistant to Agent App.
     """
-    click.echo(click.style("Start convert to agent apps.", fg="green"))
+    click.echo(click.style("Starting convert to agent apps.", fg="green"))
 
     proceeded_app_ids = []
 
@@ -496,23 +492,23 @@ def convert_to_agent_apps():
             except Exception as e:
                 click.echo(click.style("Convert app error: {} {}".format(e.__class__.__name__, str(e)), fg="red"))
 
-    click.echo(click.style("Congratulations! Converted {} agent apps.".format(len(proceeded_app_ids)), fg="green"))
+    click.echo(click.style("Conversion complete. Converted {} agent apps.".format(len(proceeded_app_ids)), fg="green"))
 
 
-@click.command("add-qdrant-doc-id-index", help="add qdrant doc_id index.")
-@click.option("--field", default="metadata.doc_id", prompt=False, help="index field , default is metadata.doc_id.")
+@click.command("add-qdrant-doc-id-index", help="Add Qdrant doc_id index.")
+@click.option("--field", default="metadata.doc_id", prompt=False, help="Index field , default is metadata.doc_id.")
 def add_qdrant_doc_id_index(field: str):
-    click.echo(click.style("Start add qdrant doc_id index.", fg="green"))
+    click.echo(click.style("Starting Qdrant doc_id index creation.", fg="green"))
     vector_type = dify_config.VECTOR_STORE
     if vector_type != "qdrant":
-        click.echo(click.style("Sorry, only support qdrant vector store.", fg="red"))
+        click.echo(click.style("This command only supports Qdrant vector store.", fg="red"))
         return
     create_count = 0
 
     try:
         bindings = db.session.query(DatasetCollectionBinding).all()
         if not bindings:
-            click.echo(click.style("Sorry, no dataset collection bindings found.", fg="red"))
+            click.echo(click.style("No dataset collection bindings found.", fg="red"))
             return
         import qdrant_client
         from qdrant_client.http.exceptions import UnexpectedResponse
@@ -522,7 +518,7 @@ def add_qdrant_doc_id_index(field: str):
 
         for binding in bindings:
             if dify_config.QDRANT_URL is None:
-                raise ValueError("Qdrant url is required.")
+                raise ValueError("Qdrant URL is required.")
             qdrant_config = QdrantConfig(
                 endpoint=dify_config.QDRANT_URL,
                 api_key=dify_config.QDRANT_API_KEY,
@@ -539,41 +535,39 @@ def add_qdrant_doc_id_index(field: str):
             except UnexpectedResponse as e:
                 # Collection does not exist, so return
                 if e.status_code == 404:
-                    click.echo(
-                        click.style(f"Collection not found, collection_name:{binding.collection_name}.", fg="red")
-                    )
+                    click.echo(click.style(f"Collection not found: {binding.collection_name}.", fg="red"))
                     continue
                 # Some other error occurred, so re-raise the exception
                 else:
                     click.echo(
                         click.style(
-                            f"Failed to create qdrant index, collection_name:{binding.collection_name}.", fg="red"
+                            f"Failed to create Qdrant index for collection: {binding.collection_name}.", fg="red"
                         )
                     )
 
     except Exception as e:
-        click.echo(click.style("Failed to create qdrant client.", fg="red"))
+        click.echo(click.style("Failed to create Qdrant client.", fg="red"))
 
-    click.echo(click.style(f"Congratulations! Create {create_count} collection indexes.", fg="green"))
+    click.echo(click.style(f"Index creation complete. Created {create_count} collection indexes.", fg="green"))
 
 
 @click.command("create-tenant", help="Create account and tenant.")
-@click.option("--email", prompt=True, help="The email address of the tenant account.")
-@click.option("--name", prompt=True, help="The workspace name of the tenant account.")
+@click.option("--email", prompt=True, help="Tenant account email.")
+@click.option("--name", prompt=True, help="Workspace name.")
 @click.option("--language", prompt=True, help="Account language, default: en-US.")
 def create_tenant(email: str, language: Optional[str] = None, name: Optional[str] = None):
     """
     Create tenant account
     """
     if not email:
-        click.echo(click.style("Sorry, email is required.", fg="red"))
+        click.echo(click.style("Email is required.", fg="red"))
         return
 
     # Create account
     email = email.strip()
 
     if "@" not in email:
-        click.echo(click.style("Sorry, invalid email address.", fg="red"))
+        click.echo(click.style("Invalid email address.", fg="red"))
         return
 
     account_name = email.split("@")[0]
@@ -593,19 +587,19 @@ def create_tenant(email: str, language: Optional[str] = None, name: Optional[str
 
     click.echo(
         click.style(
-            "Congratulations! Account and tenant created.\nAccount: {}\nPassword: {}".format(email, new_password),
+            "Account and tenant created.\nAccount: {}\nPassword: {}".format(email, new_password),
             fg="green",
         )
     )
 
 
-@click.command("upgrade-db", help="upgrade the database")
+@click.command("upgrade-db", help="Upgrade the database")
 def upgrade_db():
     click.echo("Preparing database migration...")
     lock = redis_client.lock(name="db_upgrade_lock", timeout=60)
     if lock.acquire(blocking=False):
         try:
-            click.echo(click.style("Start database migration.", fg="green"))
+            click.echo(click.style("Starting database migration.", fg="green"))
 
             # run db migration
             import flask_migrate
@@ -615,7 +609,7 @@ def upgrade_db():
             click.echo(click.style("Database migration successful!", fg="green"))
 
         except Exception as e:
-            logging.exception(f"Database migration failed, error: {e}")
+            logging.exception(f"Database migration failed: {e}")
         finally:
             lock.release()
     else:
@@ -627,7 +621,7 @@ def fix_app_site_missing():
     """
     Fix app related site missing issue.
     """
-    click.echo(click.style("Start fix app related site missing issue.", fg="green"))
+    click.echo(click.style("Starting fix for missing app-related sites.", fg="green"))
 
     failed_app_ids = []
     while True:
@@ -650,22 +644,22 @@ def fix_app_site_missing():
                     if tenant:
                         accounts = tenant.get_accounts()
                         if not accounts:
-                            print("Fix app {} failed.".format(app.id))
+                            print("Fix failed for app {}".format(app.id))
                             continue
 
                         account = accounts[0]
-                        print("Fix app {} related site missing issue.".format(app.id))
+                        print("Fixing missing site for app {}".format(app.id))
                         app_was_created.send(app, account=account)
                 except Exception as e:
                     failed_app_ids.append(app_id)
-                    click.echo(click.style("Fix app {} related site missing issue failed!".format(app_id), fg="red"))
+                    click.echo(click.style("FFailed to fix missing site for app {}".format(app_id), fg="red"))
                     logging.exception(f"Fix app related site missing issue failed, error: {e}")
                     continue
 
             if not processed_count:
                 break
 
-    click.echo(click.style("Congratulations! Fix app related site missing issue successful!", fg="green"))
+    click.echo(click.style("Fix for missing app-related sites completed successfully!", fg="green"))
 
 
 def register_commands(app):

From eaa7e9b1f095ce537e592e22526b03dba816f105 Mon Sep 17 00:00:00 2001
From: Nam Vu <zuzoovn@gmail.com>
Date: Sun, 22 Sep 2024 13:02:12 +0700
Subject: [PATCH 024/235] fix: llm_generator.py JSONDecodeError (#8504)

---
 api/core/llm_generator/llm_generator.py                     | 2 ++
 .../workflow/nodes/question_classifier/template_prompts.py  | 6 +++---
 2 files changed, 5 insertions(+), 3 deletions(-)

diff --git a/api/core/llm_generator/llm_generator.py b/api/core/llm_generator/llm_generator.py
index 78a6d6e6836d3..39bd6fee69cf1 100644
--- a/api/core/llm_generator/llm_generator.py
+++ b/api/core/llm_generator/llm_generator.py
@@ -47,6 +47,8 @@ def generate_conversation_name(
             )
         answer = response.message.content
         cleaned_answer = re.sub(r"^.*(\{.*\}).*$", r"\1", answer, flags=re.DOTALL)
+        if cleaned_answer is None:
+            return ""
         result_dict = json.loads(cleaned_answer)
         answer = result_dict["Your Output"]
         name = answer.strip()
diff --git a/api/core/workflow/nodes/question_classifier/template_prompts.py b/api/core/workflow/nodes/question_classifier/template_prompts.py
index ce32b01aa4189..4bca2d9dd4edc 100644
--- a/api/core/workflow/nodes/question_classifier/template_prompts.py
+++ b/api/core/workflow/nodes/question_classifier/template_prompts.py
@@ -2,9 +2,9 @@
     ### Job Description',
     You are a text classification engine that analyzes text data and assigns categories based on user input or automatically determined categories.
     ### Task
-    Your task is to assign one categories ONLY to the input text and only one category may be assigned returned in the output.Additionally, you need to extract the key words from the text that are related to the classification.
+    Your task is to assign one categories ONLY to the input text and only one category may be assigned returned in the output. Additionally, you need to extract the key words from the text that are related to the classification.
     ### Format
-    The input text is in the variable input_text.Categories are specified as a category list with two filed category_id and category_name in the variable categories .Classification instructions may be included to improve the classification accuracy.
+    The input text is in the variable input_text. Categories are specified as a category list with two filed category_id and category_name in the variable categories. Classification instructions may be included to improve the classification accuracy.
     ### Constraint
     DO NOT include anything other than the JSON array in your response.
     ### Memory
@@ -52,7 +52,7 @@
 ### Job Description
 You are a text classification engine that analyzes text data and assigns categories based on user input or automatically determined categories.
 ### Task
-Your task is to assign one categories ONLY to the input text and only one category may be assigned returned in the output.  Additionally, you need to extract the key words from the text that are related to the classification.
+Your task is to assign one categories ONLY to the input text and only one category may be assigned returned in the output. Additionally, you need to extract the key words from the text that are related to the classification.
 ### Format
 The input text is in the variable input_text. Categories are specified as a category list  with two filed category_id and category_name in the variable categories. Classification instructions may be included to improve the classification accuracy.
 ### Constraint 

From 2d869d68313dfd709b603628a338f82f33e619c5 Mon Sep 17 00:00:00 2001
From: Hash Brown <hi@xzd.me>
Date: Sun, 22 Sep 2024 16:41:40 +0800
Subject: [PATCH 025/235] fix: send message error when chatting with opening
 statement (#8627)

---
 .../configuration/debug/debug-with-single-model/index.tsx | 8 +++++++-
 .../base/chat/chat-with-history/chat-wrapper.tsx          | 8 +++++++-
 .../base/chat/embedded-chatbot/chat-wrapper.tsx           | 8 +++++++-
 .../workflow/panel/debug-and-preview/chat-wrapper.tsx     | 8 +++++++-
 4 files changed, 28 insertions(+), 4 deletions(-)

diff --git a/web/app/components/app/configuration/debug/debug-with-single-model/index.tsx b/web/app/components/app/configuration/debug/debug-with-single-model/index.tsx
index 80e7c98a8f0c9..5faef46d988e0 100644
--- a/web/app/components/app/configuration/debug/debug-with-single-model/index.tsx
+++ b/web/app/components/app/configuration/debug/debug-with-single-model/index.tsx
@@ -83,11 +83,17 @@ const DebugWithSingleModel = forwardRef<DebugWithSingleModelRefType, DebugWithSi
       },
     }
 
+    const lastAnswer = chatListRef.current.at(-1)
+
     const data: any = {
       query: message,
       inputs,
       model_config: configData,
-      parent_message_id: last_answer?.id || chatListRef.current.at(-1)?.id || null,
+      parent_message_id: last_answer?.id || (lastAnswer
+        ? lastAnswer.isOpeningStatement
+          ? null
+          : lastAnswer.id
+        : null),
     }
 
     if (visionConfig.enabled && files?.length && supportVision)
diff --git a/web/app/components/base/chat/chat-with-history/chat-wrapper.tsx b/web/app/components/base/chat/chat-with-history/chat-wrapper.tsx
index 79f6008d13671..225bbac714084 100644
--- a/web/app/components/base/chat/chat-with-history/chat-wrapper.tsx
+++ b/web/app/components/base/chat/chat-with-history/chat-wrapper.tsx
@@ -67,11 +67,17 @@ const ChatWrapper = () => {
   }, [])
 
   const doSend: OnSend = useCallback((message, files, last_answer) => {
+    const lastAnswer = chatListRef.current.at(-1)
+
     const data: any = {
       query: message,
       inputs: currentConversationId ? currentConversationItem?.inputs : newConversationInputs,
       conversation_id: currentConversationId,
-      parent_message_id: last_answer?.id || chatListRef.current.at(-1)?.id || null,
+      parent_message_id: last_answer?.id || (lastAnswer
+        ? lastAnswer.isOpeningStatement
+          ? null
+          : lastAnswer.id
+        : null),
     }
 
     if (appConfig?.file_upload?.image.enabled && files?.length)
diff --git a/web/app/components/base/chat/embedded-chatbot/chat-wrapper.tsx b/web/app/components/base/chat/embedded-chatbot/chat-wrapper.tsx
index 8cb546fd52a0e..ed2f24274dcd7 100644
--- a/web/app/components/base/chat/embedded-chatbot/chat-wrapper.tsx
+++ b/web/app/components/base/chat/embedded-chatbot/chat-wrapper.tsx
@@ -69,11 +69,17 @@ const ChatWrapper = () => {
   }, [])
 
   const doSend: OnSend = useCallback((message, files, last_answer) => {
+    const lastAnswer = chatListRef.current.at(-1)
+
     const data: any = {
       query: message,
       inputs: currentConversationId ? currentConversationItem?.inputs : newConversationInputs,
       conversation_id: currentConversationId,
-      parent_message_id: last_answer?.id || chatListRef.current.at(-1)?.id || null,
+      parent_message_id: last_answer?.id || (lastAnswer
+        ? lastAnswer.isOpeningStatement
+          ? null
+          : lastAnswer.id
+        : null),
     }
 
     if (appConfig?.file_upload?.image.enabled && files?.length)
diff --git a/web/app/components/workflow/panel/debug-and-preview/chat-wrapper.tsx b/web/app/components/workflow/panel/debug-and-preview/chat-wrapper.tsx
index 107a5dc698f74..86519af60357d 100644
--- a/web/app/components/workflow/panel/debug-and-preview/chat-wrapper.tsx
+++ b/web/app/components/workflow/panel/debug-and-preview/chat-wrapper.tsx
@@ -76,13 +76,19 @@ const ChatWrapper = forwardRef<ChatWrapperRefType, ChatWrapperProps>(({ showConv
   )
 
   const doSend = useCallback<OnSend>((query, files, last_answer) => {
+    const lastAnswer = chatListRef.current.at(-1)
+
     handleSend(
       {
         query,
         files,
         inputs: workflowStore.getState().inputs,
         conversation_id: conversationId,
-        parent_message_id: last_answer?.id || chatListRef.current.at(-1)?.id || null,
+        parent_message_id: last_answer?.id || (lastAnswer
+          ? lastAnswer.isOpeningStatement
+            ? null
+            : lastAnswer.id
+          : null),
       },
       {
         onGetSuggestedQuestions: (messageId, getAbortController) => fetchSuggestedQuestions(appDetail!.id, messageId, getAbortController),

From 45c0a44411c1554ca9233dc9b355ad905ed6f5fc Mon Sep 17 00:00:00 2001
From: zhuhao <37029601+hwzhuhao@users.noreply.github.com>
Date: Sun, 22 Sep 2024 16:42:34 +0800
Subject: [PATCH 026/235] feat: add qwen2.5 for model provider siliconflow
 (#8630)

---
 .../siliconflow/llm/_position.yaml            |  4 +++
 .../siliconflow/llm/qwen2.5-14b-instruct.yaml | 30 +++++++++++++++++++
 .../siliconflow/llm/qwen2.5-32b-instruct.yaml | 30 +++++++++++++++++++
 .../siliconflow/llm/qwen2.5-72b-instruct.yaml | 30 +++++++++++++++++++
 .../siliconflow/llm/qwen2.5-7b-instruct.yaml  | 30 +++++++++++++++++++
 5 files changed, 124 insertions(+)
 create mode 100644 api/core/model_runtime/model_providers/siliconflow/llm/qwen2.5-14b-instruct.yaml
 create mode 100644 api/core/model_runtime/model_providers/siliconflow/llm/qwen2.5-32b-instruct.yaml
 create mode 100644 api/core/model_runtime/model_providers/siliconflow/llm/qwen2.5-72b-instruct.yaml
 create mode 100644 api/core/model_runtime/model_providers/siliconflow/llm/qwen2.5-7b-instruct.yaml

diff --git a/api/core/model_runtime/model_providers/siliconflow/llm/_position.yaml b/api/core/model_runtime/model_providers/siliconflow/llm/_position.yaml
index c2f0eb0536032..dbf4e95f1fe3c 100644
--- a/api/core/model_runtime/model_providers/siliconflow/llm/_position.yaml
+++ b/api/core/model_runtime/model_providers/siliconflow/llm/_position.yaml
@@ -1,3 +1,7 @@
+- Qwen/Qwen2.5-7B-Instruct
+- Qwen/Qwen2.5-14B-Instruct
+- Qwen/Qwen2.5-32B-Instruct
+- Qwen/Qwen2.5-72B-Instruct
 - Qwen/Qwen2-72B-Instruct
 - Qwen/Qwen2-57B-A14B-Instruct
 - Qwen/Qwen2-7B-Instruct
diff --git a/api/core/model_runtime/model_providers/siliconflow/llm/qwen2.5-14b-instruct.yaml b/api/core/model_runtime/model_providers/siliconflow/llm/qwen2.5-14b-instruct.yaml
new file mode 100644
index 0000000000000..02a401464bb8c
--- /dev/null
+++ b/api/core/model_runtime/model_providers/siliconflow/llm/qwen2.5-14b-instruct.yaml
@@ -0,0 +1,30 @@
+model: Qwen/Qwen2.5-14B-Instruct
+label:
+  en_US: Qwen/Qwen2.5-14B-Instruct
+model_type: llm
+features:
+  - agent-thought
+model_properties:
+  mode: chat
+  context_size: 32768
+parameter_rules:
+  - name: temperature
+    use_template: temperature
+  - name: max_tokens
+    use_template: max_tokens
+    type: int
+    default: 512
+    min: 1
+    max: 8192
+    help:
+      zh_Hans: 指定生成结果长度的上限。如果生成结果截断，可以调大该参数。
+      en_US: Specifies the upper limit on the length of generated results. If the generated results are truncated, you can increase this parameter.
+  - name: top_p
+    use_template: top_p
+  - name: frequency_penalty
+    use_template: frequency_penalty
+pricing:
+  input: '0.7'
+  output: '0.7'
+  unit: '0.000001'
+  currency: RMB
diff --git a/api/core/model_runtime/model_providers/siliconflow/llm/qwen2.5-32b-instruct.yaml b/api/core/model_runtime/model_providers/siliconflow/llm/qwen2.5-32b-instruct.yaml
new file mode 100644
index 0000000000000..d084617e7d736
--- /dev/null
+++ b/api/core/model_runtime/model_providers/siliconflow/llm/qwen2.5-32b-instruct.yaml
@@ -0,0 +1,30 @@
+model: Qwen/Qwen2.5-32B-Instruct
+label:
+  en_US: Qwen/Qwen2.5-32B-Instruct
+model_type: llm
+features:
+  - agent-thought
+model_properties:
+  mode: chat
+  context_size: 32768
+parameter_rules:
+  - name: temperature
+    use_template: temperature
+  - name: max_tokens
+    use_template: max_tokens
+    type: int
+    default: 512
+    min: 1
+    max: 8192
+    help:
+      zh_Hans: 指定生成结果长度的上限。如果生成结果截断，可以调大该参数。
+      en_US: Specifies the upper limit on the length of generated results. If the generated results are truncated, you can increase this parameter.
+  - name: top_p
+    use_template: top_p
+  - name: frequency_penalty
+    use_template: frequency_penalty
+pricing:
+  input: '1.26'
+  output: '1.26'
+  unit: '0.000001'
+  currency: RMB
diff --git a/api/core/model_runtime/model_providers/siliconflow/llm/qwen2.5-72b-instruct.yaml b/api/core/model_runtime/model_providers/siliconflow/llm/qwen2.5-72b-instruct.yaml
new file mode 100644
index 0000000000000..dfbad2494cdf7
--- /dev/null
+++ b/api/core/model_runtime/model_providers/siliconflow/llm/qwen2.5-72b-instruct.yaml
@@ -0,0 +1,30 @@
+model: Qwen/Qwen2.5-72B-Instruct
+label:
+  en_US: Qwen/Qwen2.5-72B-Instruct
+model_type: llm
+features:
+  - agent-thought
+model_properties:
+  mode: chat
+  context_size: 32768
+parameter_rules:
+  - name: temperature
+    use_template: temperature
+  - name: max_tokens
+    use_template: max_tokens
+    type: int
+    default: 512
+    min: 1
+    max: 8192
+    help:
+      zh_Hans: 指定生成结果长度的上限。如果生成结果截断，可以调大该参数。
+      en_US: Specifies the upper limit on the length of generated results. If the generated results are truncated, you can increase this parameter.
+  - name: top_p
+    use_template: top_p
+  - name: frequency_penalty
+    use_template: frequency_penalty
+pricing:
+  input: '4.13'
+  output: '4.13'
+  unit: '0.000001'
+  currency: RMB
diff --git a/api/core/model_runtime/model_providers/siliconflow/llm/qwen2.5-7b-instruct.yaml b/api/core/model_runtime/model_providers/siliconflow/llm/qwen2.5-7b-instruct.yaml
new file mode 100644
index 0000000000000..cdc8ffc4d2104
--- /dev/null
+++ b/api/core/model_runtime/model_providers/siliconflow/llm/qwen2.5-7b-instruct.yaml
@@ -0,0 +1,30 @@
+model: Qwen/Qwen2.5-7B-Instruct
+label:
+  en_US: Qwen/Qwen2.5-7B-Instruct
+model_type: llm
+features:
+  - agent-thought
+model_properties:
+  mode: chat
+  context_size: 32768
+parameter_rules:
+  - name: temperature
+    use_template: temperature
+  - name: max_tokens
+    use_template: max_tokens
+    type: int
+    default: 512
+    min: 1
+    max: 8192
+    help:
+      zh_Hans: 指定生成结果长度的上限。如果生成结果截断，可以调大该参数。
+      en_US: Specifies the upper limit on the length of generated results. If the generated results are truncated, you can increase this parameter.
+  - name: top_p
+    use_template: top_p
+  - name: frequency_penalty
+    use_template: frequency_penalty
+pricing:
+  input: '0'
+  output: '0'
+  unit: '0.000001'
+  currency: RMB

From 6df77038a2a368d5efd4ed931acad13df49df156 Mon Sep 17 00:00:00 2001
From: zhuhao <37029601+hwzhuhao@users.noreply.github.com>
Date: Sun, 22 Sep 2024 16:45:45 +0800
Subject: [PATCH 027/235] docs: fix predefined_model_scale_out.md redirect
 error (#8633)

---
 .../model_runtime/docs/zh_Hans/predefined_model_scale_out.md    | 2 +-
 1 file changed, 1 insertion(+), 1 deletion(-)

diff --git a/api/core/model_runtime/docs/zh_Hans/predefined_model_scale_out.md b/api/core/model_runtime/docs/zh_Hans/predefined_model_scale_out.md
index 56f379a92fdd3..17fc088a63a92 100644
--- a/api/core/model_runtime/docs/zh_Hans/predefined_model_scale_out.md
+++ b/api/core/model_runtime/docs/zh_Hans/predefined_model_scale_out.md
@@ -62,7 +62,7 @@ pricing:  # 价格信息
 
 建议将所有模型配置都准备完毕后再开始模型代码的实现。
 
-同样，也可以参考  `model_providers` 目录下其他供应商对应模型类型目录下的 YAML 配置信息，完整的 YAML 规则见：[Schema](schema.md#AIModel)。
+同样，也可以参考  `model_providers` 目录下其他供应商对应模型类型目录下的 YAML 配置信息，完整的 YAML 规则见：[Schema](schema.md#aimodelentity)。
 
 ### 实现模型调用代码
 

From e34f04380db89087ee82505a9eb1100bb6b0b004 Mon Sep 17 00:00:00 2001
From: zhuhao <37029601+hwzhuhao@users.noreply.github.com>
Date: Sun, 22 Sep 2024 21:44:06 +0800
Subject: [PATCH 028/235] feat: add deepseek-v2.5 for model provider
 siliconflow (#8639)

---
 .../siliconflow/llm/_position.yaml            |  1 +
 .../siliconflow/llm/deepseek-v2.5.yaml        | 30 +++++++++++++++++++
 2 files changed, 31 insertions(+)
 create mode 100644 api/core/model_runtime/model_providers/siliconflow/llm/deepseek-v2.5.yaml

diff --git a/api/core/model_runtime/model_providers/siliconflow/llm/_position.yaml b/api/core/model_runtime/model_providers/siliconflow/llm/_position.yaml
index dbf4e95f1fe3c..43db4aed11916 100644
--- a/api/core/model_runtime/model_providers/siliconflow/llm/_position.yaml
+++ b/api/core/model_runtime/model_providers/siliconflow/llm/_position.yaml
@@ -10,6 +10,7 @@
 - 01-ai/Yi-1.5-9B-Chat-16K
 - 01-ai/Yi-1.5-6B-Chat
 - THUDM/glm-4-9b-chat
+- deepseek-ai/DeepSeek-V2.5
 - deepseek-ai/DeepSeek-V2-Chat
 - deepseek-ai/DeepSeek-Coder-V2-Instruct
 - internlm/internlm2_5-7b-chat
diff --git a/api/core/model_runtime/model_providers/siliconflow/llm/deepseek-v2.5.yaml b/api/core/model_runtime/model_providers/siliconflow/llm/deepseek-v2.5.yaml
new file mode 100644
index 0000000000000..1c8e15ae52afa
--- /dev/null
+++ b/api/core/model_runtime/model_providers/siliconflow/llm/deepseek-v2.5.yaml
@@ -0,0 +1,30 @@
+model: deepseek-ai/DeepSeek-V2.5
+label:
+  en_US: deepseek-ai/DeepSeek-V2.5
+model_type: llm
+features:
+  - agent-thought
+model_properties:
+  mode: chat
+  context_size: 32768
+parameter_rules:
+  - name: temperature
+    use_template: temperature
+  - name: max_tokens
+    use_template: max_tokens
+    type: int
+    default: 512
+    min: 1
+    max: 4096
+    help:
+      zh_Hans: 指定生成结果长度的上限。如果生成结果截断，可以调大该参数。
+      en_US: Specifies the upper limit on the length of generated results. If the generated results are truncated, you can increase this parameter.
+  - name: top_p
+    use_template: top_p
+  - name: frequency_penalty
+    use_template: frequency_penalty
+pricing:
+  input: '1.33'
+  output: '1.33'
+  unit: '0.000001'
+  currency: RMB

From cae73b9a329cb8edc9dae8ef53593fa29f5c2d9f Mon Sep 17 00:00:00 2001
From: Hiroshi Fujita <fujita-h@users.noreply.github.com>
Date: Mon, 23 Sep 2024 10:05:02 +0900
Subject: [PATCH 029/235] Make WORKFLOW_* configurable as environment
 variables. (#8644)

---
 docker/.env.example        | 5 +++++
 docker/docker-compose.yaml | 3 +++
 2 files changed, 8 insertions(+)

diff --git a/docker/.env.example b/docker/.env.example
index c892c15636e69..7eaaceb92834d 100644
--- a/docker/.env.example
+++ b/docker/.env.example
@@ -563,6 +563,11 @@ CODE_MAX_STRING_ARRAY_LENGTH=30
 CODE_MAX_OBJECT_ARRAY_LENGTH=30
 CODE_MAX_NUMBER_ARRAY_LENGTH=1000
 
+# Workflow runtime configuration
+WORKFLOW_MAX_EXECUTION_STEPS=500
+WORKFLOW_MAX_EXECUTION_TIME=1200
+WORKFLOW_CALL_MAX_DEPTH=5
+
 # SSRF Proxy server HTTP URL
 SSRF_PROXY_HTTP_URL=http://ssrf_proxy:3128
 # SSRF Proxy server HTTPS URL
diff --git a/docker/docker-compose.yaml b/docker/docker-compose.yaml
index e72c3724f9ca8..16bef279bcc68 100644
--- a/docker/docker-compose.yaml
+++ b/docker/docker-compose.yaml
@@ -202,6 +202,9 @@ x-shared-env: &shared-api-worker-env
   CODE_MAX_STRING_ARRAY_LENGTH: ${CODE_MAX_STRING_ARRAY_LENGTH:-30}
   CODE_MAX_OBJECT_ARRAY_LENGTH: ${CODE_MAX_OBJECT_ARRAY_LENGTH:-30}
   CODE_MAX_NUMBER_ARRAY_LENGTH: ${CODE_MAX_NUMBER_ARRAY_LENGTH:-1000}
+  WORKFLOW_MAX_EXECUTION_STEPS: ${WORKFLOW_MAX_EXECUTION_STEPS:-500}
+  WORKFLOW_MAX_EXECUTION_TIME: ${WORKFLOW_MAX_EXECUTION_TIME:-1200}
+  WORKFLOW_CALL_MAX_DEPTH: ${WORKFLOW_MAX_EXECUTION_TIME:-5}
   SSRF_PROXY_HTTP_URL: ${SSRF_PROXY_HTTP_URL:-http://ssrf_proxy:3128}
   SSRF_PROXY_HTTPS_URL: ${SSRF_PROXY_HTTPS_URL:-http://ssrf_proxy:3128}
 

From 03fdf5e7f8e44950cce683fc8c6ef4b89b90cf02 Mon Sep 17 00:00:00 2001
From: Shota Totsuka <153569547+totsukash@users.noreply.github.com>
Date: Mon, 23 Sep 2024 10:06:01 +0900
Subject: [PATCH 030/235] chore: Enable Japanese descriptions for Tools (#8646)

---
 api/core/tools/entities/common_entities.py                  | 5 ++++-
 api/core/tools/provider/builtin/arxiv/arxiv.yaml            | 2 ++
 .../tools/provider/builtin/arxiv/tools/arxiv_search.yaml    | 4 ++++
 api/services/tools/tools_transform_service.py               | 2 ++
 web/i18n/language.ts                                        | 6 ++----
 5 files changed, 14 insertions(+), 5 deletions(-)

diff --git a/api/core/tools/entities/common_entities.py b/api/core/tools/entities/common_entities.py
index 37a926697bb43..b52119fdc4759 100644
--- a/api/core/tools/entities/common_entities.py
+++ b/api/core/tools/entities/common_entities.py
@@ -10,6 +10,7 @@ class I18nObject(BaseModel):
 
     zh_Hans: Optional[str] = None
     pt_BR: Optional[str] = None
+    ja_JP: Optional[str] = None
     en_US: str
 
     def __init__(self, **data):
@@ -18,6 +19,8 @@ def __init__(self, **data):
             self.zh_Hans = self.en_US
         if not self.pt_BR:
             self.pt_BR = self.en_US
+        if not self.ja_JP:
+            self.ja_JP = self.en_US
 
     def to_dict(self) -> dict:
-        return {"zh_Hans": self.zh_Hans, "en_US": self.en_US, "pt_BR": self.pt_BR}
+        return {"zh_Hans": self.zh_Hans, "en_US": self.en_US, "pt_BR": self.pt_BR, "ja_JP": self.ja_JP}
diff --git a/api/core/tools/provider/builtin/arxiv/arxiv.yaml b/api/core/tools/provider/builtin/arxiv/arxiv.yaml
index d26993b3364ea..25aec97bb795e 100644
--- a/api/core/tools/provider/builtin/arxiv/arxiv.yaml
+++ b/api/core/tools/provider/builtin/arxiv/arxiv.yaml
@@ -4,9 +4,11 @@ identity:
   label:
     en_US: ArXiv
     zh_Hans: ArXiv
+    ja_JP: ArXiv
   description:
     en_US: Access to a vast repository of scientific papers and articles in various fields of research.
     zh_Hans: 访问各个研究领域大量科学论文和文章的存储库。
+    ja_JP: 多様な研究分野の科学論文や記事の膨大なリポジトリへのアクセス。
   icon: icon.svg
   tags:
     - search
diff --git a/api/core/tools/provider/builtin/arxiv/tools/arxiv_search.yaml b/api/core/tools/provider/builtin/arxiv/tools/arxiv_search.yaml
index 7439a48658c6a..afc1925df3b45 100644
--- a/api/core/tools/provider/builtin/arxiv/tools/arxiv_search.yaml
+++ b/api/core/tools/provider/builtin/arxiv/tools/arxiv_search.yaml
@@ -4,10 +4,12 @@ identity:
   label:
     en_US: Arxiv Search
     zh_Hans: Arxiv 搜索
+    ja_JP: Arxiv 検索
 description:
   human:
     en_US: A tool for searching scientific papers and articles from the Arxiv repository. Input can be an Arxiv ID or an author's name.
     zh_Hans: 一个用于从Arxiv存储库搜索科学论文和文章的工具。 输入可以是Arxiv ID或作者姓名。
+    ja_JP: Arxivリポジトリから科学論文や記事を検索するためのツールです。入力はArxiv IDまたは著者名にすることができます。
   llm: A tool for searching scientific papers and articles from the Arxiv repository. Input can be an Arxiv ID or an author's name.
 parameters:
   - name: query
@@ -16,8 +18,10 @@ parameters:
     label:
       en_US: Query string
       zh_Hans: 查询字符串
+      ja_JP: クエリ文字列
     human_description:
       en_US: The Arxiv ID or author's name used for searching.
       zh_Hans: 用于搜索的Arxiv ID或作者姓名。
+      ja_JP: 検索に使用されるArxiv IDまたは著者名。
     llm_description: The Arxiv ID or author's name used for searching.
     form: llm
diff --git a/api/services/tools/tools_transform_service.py b/api/services/tools/tools_transform_service.py
index 7ae1b9f23173e..2bc48c41850de 100644
--- a/api/services/tools/tools_transform_service.py
+++ b/api/services/tools/tools_transform_service.py
@@ -74,12 +74,14 @@ def builtin_provider_to_user_provider(
                 en_US=provider_controller.identity.description.en_US,
                 zh_Hans=provider_controller.identity.description.zh_Hans,
                 pt_BR=provider_controller.identity.description.pt_BR,
+                ja_JP=provider_controller.identity.description.ja_JP,
             ),
             icon=provider_controller.identity.icon,
             label=I18nObject(
                 en_US=provider_controller.identity.label.en_US,
                 zh_Hans=provider_controller.identity.label.zh_Hans,
                 pt_BR=provider_controller.identity.label.pt_BR,
+                ja_JP=provider_controller.identity.label.ja_JP,
             ),
             type=ToolProviderType.BUILT_IN,
             masked_credentials={},
diff --git a/web/i18n/language.ts b/web/i18n/language.ts
index fde69328bd922..c2e23cc19a638 100644
--- a/web/i18n/language.ts
+++ b/web/i18n/language.ts
@@ -31,10 +31,8 @@ export const languages = data.languages
 export const LanguagesSupported = languages.filter(item => item.supported).map(item => item.value)
 
 export const getLanguage = (locale: string) => {
-  if (locale === 'zh-Hans')
-    return locale.replace('-', '_')
-
-  return LanguagesSupported[0].replace('-', '_')
+  const supportedLocale = LanguagesSupported.find(lang => lang.startsWith(locale.split('-')[0]))
+  return (supportedLocale || LanguagesSupported[0]).replace('-', '_')
 }
 
 export const NOTICE_I18N = {

From 3618a97c20e386aa49a39c1f6af55ce7414ffc5c Mon Sep 17 00:00:00 2001
From: Aaron Ji <127167174+DresAaron@users.noreply.github.com>
Date: Mon, 23 Sep 2024 13:45:09 +0800
Subject: [PATCH 031/235] feat: extend api params for Jina Embeddings V3
 (#8657)

---
 .../model_providers/jina/jina.yaml            | 43 +++++++++++++++++++
 .../jina/text_embedding/text_embedding.py     | 42 ++++++++++++++----
 2 files changed, 76 insertions(+), 9 deletions(-)

diff --git a/api/core/model_runtime/model_providers/jina/jina.yaml b/api/core/model_runtime/model_providers/jina/jina.yaml
index 23e18ad75f688..9c70d6ff33ba4 100644
--- a/api/core/model_runtime/model_providers/jina/jina.yaml
+++ b/api/core/model_runtime/model_providers/jina/jina.yaml
@@ -67,3 +67,46 @@ model_credential_schema:
       required: false
       type: text-input
       default: '8192'
+    - variable: task
+      label:
+        zh_Hans: 下游任务
+        en_US: Downstream task
+      placeholder:
+        zh_Hans: 选择将使用向量模型的下游任务。模型将返回针对该任务优化的向量。
+        en_US: Select the downstream task for which the embeddings will be used. The model will return the optimized embeddings for that task.
+      required: false
+      type: select
+      options:
+        - value: retrieval.query
+          label:
+            en_US: retrieval.query
+        - value: retrieval.passage
+          label:
+            en_US: retrieval.passage
+        - value: separation
+          label:
+            en_US: separation
+        - value: classification
+          label:
+            en_US: classification
+        - value: text-matching
+          label:
+            en_US: text-matching
+    - variable: dimensions
+      label:
+        zh_Hans: 输出维度
+        en_US: Output dimensions
+      placeholder:
+        zh_Hans: 输入您的输出维度
+        en_US: Enter output dimensions
+      required: false
+      type: text-input
+    - variable: late_chunking
+      label:
+        zh_Hans: 后期分块
+        en_US: Late chunking
+      placeholder:
+        zh_Hans: 应用后期分块技术来利用模型的长上下文功能来生成上下文块向量化。
+        en_US: Apply the late chunking technique to leverage the model's long-context capabilities for generating contextual chunk embeddings.
+      required: false
+      type: switch
diff --git a/api/core/model_runtime/model_providers/jina/text_embedding/text_embedding.py b/api/core/model_runtime/model_providers/jina/text_embedding/text_embedding.py
index ceb79567d54ca..6c96699ea2e3f 100644
--- a/api/core/model_runtime/model_providers/jina/text_embedding/text_embedding.py
+++ b/api/core/model_runtime/model_providers/jina/text_embedding/text_embedding.py
@@ -27,6 +27,38 @@ class JinaTextEmbeddingModel(TextEmbeddingModel):
 
     api_base: str = "https://api.jina.ai/v1"
 
+    def _to_payload(self, model: str, texts: list[str], credentials: dict) -> dict:
+        """
+        Parse model credentials
+
+        :param model: model name
+        :param credentials: model credentials
+        :param texts: texts to embed
+        :return: parsed credentials
+        """
+
+        def transform_jina_input_text(model, text):
+            if model == "jina-clip-v1":
+                return {"text": text}
+            return text
+
+        data = {"model": model, "input": [transform_jina_input_text(model, text) for text in texts]}
+
+        task = credentials.get("task")
+        dimensions = credentials.get("dimensions")
+        late_chunking = credentials.get("late_chunking")
+
+        if task is not None:
+            data["task"] = task
+
+        if dimensions is not None:
+            data["dimensions"] = int(dimensions)
+
+        if late_chunking is not None:
+            data["late_chunking"] = late_chunking
+
+        return data
+
     def _invoke(
         self, model: str, credentials: dict, texts: list[str], user: Optional[str] = None
     ) -> TextEmbeddingResult:
@@ -49,15 +81,7 @@ def _invoke(
         url = base_url + "/embeddings"
         headers = {"Authorization": "Bearer " + api_key, "Content-Type": "application/json"}
 
-        def transform_jina_input_text(model, text):
-            if model == "jina-clip-v1":
-                return {"text": text}
-            return text
-
-        data = {"model": model, "input": [transform_jina_input_text(model, text) for text in texts]}
-
-        if model == "jina-embeddings-v3":
-            data["task"] = "text-matching"
+        data = self._to_payload(model=model, texts=texts, credentials=credentials)
 
         try:
             response = post(url, headers=headers, data=dumps(data))

From 4c7beb9d7b5e566209600ea0975677c7d46a6926 Mon Sep 17 00:00:00 2001
From: haike-1213 <2100797950@qq.com>
Date: Mon, 23 Sep 2024 07:23:52 +0000
Subject: [PATCH 032/235] fix: Assignment exception (#8663)

Co-authored-by: fum <fum@investoday.com.cn>
---
 api/core/app/apps/base_app_generate_response_converter.py | 5 +++--
 1 file changed, 3 insertions(+), 2 deletions(-)

diff --git a/api/core/app/apps/base_app_generate_response_converter.py b/api/core/app/apps/base_app_generate_response_converter.py
index c6855ac85494d..62e79ec444a48 100644
--- a/api/core/app/apps/base_app_generate_response_converter.py
+++ b/api/core/app/apps/base_app_generate_response_converter.py
@@ -75,10 +75,10 @@ def _get_simple_metadata(cls, metadata: dict[str, Any]):
         :return:
         """
         # show_retrieve_source
+        updated_resources = []
         if "retriever_resources" in metadata:
-            metadata["retriever_resources"] = []
             for resource in metadata["retriever_resources"]:
-                metadata["retriever_resources"].append(
+                updated_resources.append(
                     {
                         "segment_id": resource["segment_id"],
                         "position": resource["position"],
@@ -87,6 +87,7 @@ def _get_simple_metadata(cls, metadata: dict[str, Any]):
                         "content": resource["content"],
                     }
                 )
+            metadata["retriever_resources"] = updated_resources
 
         # show annotation reply
         if "annotation_reply" in metadata:

From 86f90fd9ff4905e5c1b85cd445466cfb3f2c457c Mon Sep 17 00:00:00 2001
From: Bowen Liang <liangbowen@gf.com.cn>
Date: Mon, 23 Sep 2024 15:28:57 +0800
Subject: [PATCH 033/235] chore: skip PLR6201 linter rule (#8666)

---
 api/core/tools/provider/builtin/firecrawl/tools/crawl.py  | 4 ++--
 api/core/tools/provider/builtin/firecrawl/tools/scrape.py | 4 ++--
 api/pyproject.toml                                        | 1 -
 3 files changed, 4 insertions(+), 5 deletions(-)

diff --git a/api/core/tools/provider/builtin/firecrawl/tools/crawl.py b/api/core/tools/provider/builtin/firecrawl/tools/crawl.py
index 9675b8eb91335..15ab510c6c889 100644
--- a/api/core/tools/provider/builtin/firecrawl/tools/crawl.py
+++ b/api/core/tools/provider/builtin/firecrawl/tools/crawl.py
@@ -35,10 +35,10 @@ def _invoke(self, user_id: str, tool_parameters: dict[str, Any]) -> ToolInvokeMe
         scrapeOptions["excludeTags"] = get_array_params(tool_parameters, "excludeTags")
         scrapeOptions["onlyMainContent"] = tool_parameters.get("onlyMainContent", False)
         scrapeOptions["waitFor"] = tool_parameters.get("waitFor", 0)
-        scrapeOptions = {k: v for k, v in scrapeOptions.items() if v not in {None, ""}}
+        scrapeOptions = {k: v for k, v in scrapeOptions.items() if v not in (None, "")}
         payload["scrapeOptions"] = scrapeOptions or None
 
-        payload = {k: v for k, v in payload.items() if v not in {None, ""}}
+        payload = {k: v for k, v in payload.items() if v not in (None, "")}
 
         crawl_result = app.crawl_url(url=tool_parameters["url"], wait=wait_for_results, **payload)
 
diff --git a/api/core/tools/provider/builtin/firecrawl/tools/scrape.py b/api/core/tools/provider/builtin/firecrawl/tools/scrape.py
index 538b4a1fcbf05..f00a9b31ce8c2 100644
--- a/api/core/tools/provider/builtin/firecrawl/tools/scrape.py
+++ b/api/core/tools/provider/builtin/firecrawl/tools/scrape.py
@@ -29,10 +29,10 @@ def _invoke(self, user_id: str, tool_parameters: dict[str, Any]) -> list[ToolInv
         extract["schema"] = get_json_params(tool_parameters, "schema")
         extract["systemPrompt"] = tool_parameters.get("systemPrompt")
         extract["prompt"] = tool_parameters.get("prompt")
-        extract = {k: v for k, v in extract.items() if v not in {None, ""}}
+        extract = {k: v for k, v in extract.items() if v not in (None, "")}
         payload["extract"] = extract or None
 
-        payload = {k: v for k, v in payload.items() if v not in {None, ""}}
+        payload = {k: v for k, v in payload.items() if v not in (None, "")}
 
         crawl_result = app.scrape_url(url=tool_parameters["url"], **payload)
         markdown_result = crawl_result.get("data", {}).get("markdown", "")
diff --git a/api/pyproject.toml b/api/pyproject.toml
index 506f379aaf116..066b4772a9918 100644
--- a/api/pyproject.toml
+++ b/api/pyproject.toml
@@ -28,7 +28,6 @@ select = [
     "PLR0402", # manual-from-import
     "PLR1711", # useless-return
     "PLR1714", # repeated-equality-comparison
-    "PLR6201", # literal-membership
     "RUF019", # unnecessary-key-check
     "RUF100", # unused-noqa
     "RUF101", # redirected-noqa

From b37954b9661067977f45524dce0bdb8863309b9b Mon Sep 17 00:00:00 2001
From: =?UTF-8?q?=E9=9D=9E=E6=B3=95=E6=93=8D=E4=BD=9C?= <hjlarry@163.com>
Date: Mon, 23 Sep 2024 15:33:06 +0800
Subject: [PATCH 034/235] fix: png avatar upload as jpeg (#8665)

---
 .../components/base/app-icon-picker/index.tsx |  2 +-
 .../components/base/app-icon-picker/utils.ts  | 21 ++++++++++++++++++-
 2 files changed, 21 insertions(+), 2 deletions(-)

diff --git a/web/app/components/base/app-icon-picker/index.tsx b/web/app/components/base/app-icon-picker/index.tsx
index 825481547501a..ba375abdd9c04 100644
--- a/web/app/components/base/app-icon-picker/index.tsx
+++ b/web/app/components/base/app-icon-picker/index.tsx
@@ -88,7 +88,7 @@ const AppIconPicker: FC<AppIconPickerProps> = ({
       if (!imageCropInfo)
         return
       setUploading(true)
-      const blob = await getCroppedImg(imageCropInfo.tempUrl, imageCropInfo.croppedAreaPixels)
+      const blob = await getCroppedImg(imageCropInfo.tempUrl, imageCropInfo.croppedAreaPixels, imageCropInfo.fileName)
       const file = new File([blob], imageCropInfo.fileName, { type: blob.type })
       handleLocalFileUpload(file)
     }
diff --git a/web/app/components/base/app-icon-picker/utils.ts b/web/app/components/base/app-icon-picker/utils.ts
index 0c90e96febda8..14c9ae3f289d7 100644
--- a/web/app/components/base/app-icon-picker/utils.ts
+++ b/web/app/components/base/app-icon-picker/utils.ts
@@ -11,6 +11,23 @@ export function getRadianAngle(degreeValue: number) {
   return (degreeValue * Math.PI) / 180
 }
 
+export function getMimeType(fileName: string): string {
+  const extension = fileName.split('.').pop()?.toLowerCase()
+  switch (extension) {
+    case 'png':
+      return 'image/png'
+    case 'jpg':
+    case 'jpeg':
+      return 'image/jpeg'
+    case 'gif':
+      return 'image/gif'
+    case 'webp':
+      return 'image/webp'
+    default:
+      return 'image/jpeg'
+  }
+}
+
 /**
  * Returns the new bounding area of a rotated rectangle.
  */
@@ -31,12 +48,14 @@ export function rotateSize(width: number, height: number, rotation: number) {
 export default async function getCroppedImg(
   imageSrc: string,
   pixelCrop: { x: number; y: number; width: number; height: number },
+  fileName: string,
   rotation = 0,
   flip = { horizontal: false, vertical: false },
 ): Promise<Blob> {
   const image = await createImage(imageSrc)
   const canvas = document.createElement('canvas')
   const ctx = canvas.getContext('2d')
+  const mimeType = getMimeType(fileName)
 
   if (!ctx)
     throw new Error('Could not create a canvas context')
@@ -93,6 +112,6 @@ export default async function getCroppedImg(
         resolve(file)
       else
         reject(new Error('Could not create a blob'))
-    }, 'image/jpeg')
+    }, mimeType)
   })
 }

From c66cecaa555722b7f82e069f58c3874bf39c771f Mon Sep 17 00:00:00 2001
From: AAEE86 <33052466+AAEE86@users.noreply.github.com>
Date: Mon, 23 Sep 2024 16:18:55 +0800
Subject: [PATCH 035/235] add Qwen model translate (#8674)

---
 .../tongyi/llm/farui-plus.yaml                |  4 +++
 .../tongyi/llm/qwen-coder-turbo-0919.yaml     |  4 +++
 .../tongyi/llm/qwen-coder-turbo-latest.yaml   |  4 +++
 .../tongyi/llm/qwen-coder-turbo.yaml          |  4 +++
 .../model_providers/tongyi/llm/qwen-long.yaml |  4 +++
 .../tongyi/llm/qwen-math-plus-0816.yaml       |  4 +++
 .../tongyi/llm/qwen-math-plus-0919.yaml       |  4 +++
 .../tongyi/llm/qwen-math-plus-latest.yaml     |  4 +++
 .../tongyi/llm/qwen-math-plus.yaml            |  4 +++
 .../tongyi/llm/qwen-math-turbo-0919.yaml      |  4 +++
 .../tongyi/llm/qwen-math-turbo-latest.yaml    |  4 +++
 .../tongyi/llm/qwen-math-turbo.yaml           |  4 +++
 .../tongyi/llm/qwen-max-0107.yaml             |  4 +++
 .../tongyi/llm/qwen-max-0403.yaml             |  4 +++
 .../tongyi/llm/qwen-max-0428.yaml             |  4 +++
 .../tongyi/llm/qwen-max-0919.yaml             |  4 +++
 .../tongyi/llm/qwen-max-latest.yaml           |  4 +++
 .../tongyi/llm/qwen-max-longcontext.yaml      |  4 +++
 .../model_providers/tongyi/llm/qwen-max.yaml  |  4 +++
 .../tongyi/llm/qwen-plus-0206.yaml            |  4 +++
 .../tongyi/llm/qwen-plus-0624.yaml            |  4 +++
 .../tongyi/llm/qwen-plus-0723.yaml            |  4 +++
 .../tongyi/llm/qwen-plus-0806.yaml            |  4 +++
 .../tongyi/llm/qwen-plus-0919.yaml            |  4 +++
 .../tongyi/llm/qwen-plus-chat.yaml            |  4 +++
 .../tongyi/llm/qwen-plus-latest.yaml          |  4 +++
 .../model_providers/tongyi/llm/qwen-plus.yaml |  4 +++
 .../tongyi/llm/qwen-turbo-0206.yaml           |  4 +++
 .../tongyi/llm/qwen-turbo-0624.yaml           |  4 +++
 .../tongyi/llm/qwen-turbo-0919.yaml           |  4 +++
 .../tongyi/llm/qwen-turbo-chat.yaml           |  4 +++
 .../tongyi/llm/qwen-turbo-latest.yaml         |  4 +++
 .../tongyi/llm/qwen-turbo.yaml                |  4 +++
 .../tongyi/llm/qwen-vl-max-0809.yaml          | 30 +++++++++++++++++++
 .../tongyi/llm/qwen-vl-max.yaml               | 30 +++++++++++++++++++
 .../tongyi/llm/qwen-vl-plus-0201.yaml         | 30 +++++++++++++++++++
 .../tongyi/llm/qwen-vl-plus-0809.yaml         | 30 +++++++++++++++++++
 .../tongyi/llm/qwen-vl-plus.yaml              | 30 +++++++++++++++++++
 .../tongyi/llm/qwen2-math-1.5b-instruct.yaml  |  4 +++
 .../tongyi/llm/qwen2-math-72b-instruct.yaml   |  4 +++
 .../tongyi/llm/qwen2-math-7b-instruct.yaml    |  4 +++
 .../tongyi/llm/qwen2.5-0.5b-instruct.yaml     |  4 +++
 .../tongyi/llm/qwen2.5-1.5b-instruct.yaml     |  4 +++
 .../tongyi/llm/qwen2.5-14b-instruct.yaml      |  4 +++
 .../tongyi/llm/qwen2.5-32b-instruct.yaml      |  4 +++
 .../tongyi/llm/qwen2.5-3b-instruct.yaml       |  4 +++
 .../tongyi/llm/qwen2.5-72b-instruct.yaml      |  4 +++
 .../tongyi/llm/qwen2.5-7b-instruct.yaml       |  4 +++
 .../tongyi/llm/qwen2.5-coder-7b-instruct.yaml |  4 +++
 49 files changed, 326 insertions(+)

diff --git a/api/core/model_runtime/model_providers/tongyi/llm/farui-plus.yaml b/api/core/model_runtime/model_providers/tongyi/llm/farui-plus.yaml
index aad07f56736e5..e5de586c1caf1 100644
--- a/api/core/model_runtime/model_providers/tongyi/llm/farui-plus.yaml
+++ b/api/core/model_runtime/model_providers/tongyi/llm/farui-plus.yaml
@@ -62,6 +62,7 @@ parameter_rules:
     type: float
     default: 1.1
     label:
+      zh_Hans: 重复惩罚
       en_US: Repetition penalty
     help:
       zh_Hans: 用于控制模型生成时的重复度。提高repetition_penalty时可以降低模型生成的重复度。1.0表示不做惩罚。
@@ -69,6 +70,9 @@ parameter_rules:
   - name: enable_search
     type: boolean
     default: false
+    label:
+      zh_Hans: 联网搜索
+      en_US: Web Search
     help:
       zh_Hans: 模型内置了互联网搜索服务，该参数控制模型在生成文本时是否参考使用互联网搜索结果。启用互联网搜索，模型会将搜索结果作为文本生成过程中的参考信息，但模型会基于其内部逻辑“自行判断”是否使用互联网搜索结果。
       en_US: The model has a built-in Internet search service. This parameter controls whether the model refers to Internet search results when generating text. When Internet search is enabled, the model will use the search results as reference information in the text generation process, but the model will "judge" whether to use Internet search results based on its internal logic.
diff --git a/api/core/model_runtime/model_providers/tongyi/llm/qwen-coder-turbo-0919.yaml b/api/core/model_runtime/model_providers/tongyi/llm/qwen-coder-turbo-0919.yaml
index ebba565d572ae..6ab39cde2d052 100644
--- a/api/core/model_runtime/model_providers/tongyi/llm/qwen-coder-turbo-0919.yaml
+++ b/api/core/model_runtime/model_providers/tongyi/llm/qwen-coder-turbo-0919.yaml
@@ -60,6 +60,7 @@ parameter_rules:
     type: float
     default: 1.1
     label:
+      zh_Hans: 重复惩罚
       en_US: Repetition penalty
     help:
       zh_Hans: 用于控制模型生成时的重复度。提高repetition_penalty时可以降低模型生成的重复度。1.0表示不做惩罚。
@@ -67,6 +68,9 @@ parameter_rules:
   - name: enable_search
     type: boolean
     default: false
+    label:
+      zh_Hans: 联网搜索
+      en_US: Web Search
     help:
       zh_Hans: 模型内置了互联网搜索服务，该参数控制模型在生成文本时是否参考使用互联网搜索结果。启用互联网搜索，模型会将搜索结果作为文本生成过程中的参考信息，但模型会基于其内部逻辑“自行判断”是否使用互联网搜索结果。
       en_US: The model has a built-in Internet search service. This parameter controls whether the model refers to Internet search results when generating text. When Internet search is enabled, the model will use the search results as reference information in the text generation process, but the model will "judge" whether to use Internet search results based on its internal logic.
diff --git a/api/core/model_runtime/model_providers/tongyi/llm/qwen-coder-turbo-latest.yaml b/api/core/model_runtime/model_providers/tongyi/llm/qwen-coder-turbo-latest.yaml
index 361e2c2373d65..be6d9a0e07272 100644
--- a/api/core/model_runtime/model_providers/tongyi/llm/qwen-coder-turbo-latest.yaml
+++ b/api/core/model_runtime/model_providers/tongyi/llm/qwen-coder-turbo-latest.yaml
@@ -60,6 +60,7 @@ parameter_rules:
     type: float
     default: 1.1
     label:
+      zh_Hans: 重复惩罚
       en_US: Repetition penalty
     help:
       zh_Hans: 用于控制模型生成时的重复度。提高repetition_penalty时可以降低模型生成的重复度。1.0表示不做惩罚。
@@ -67,6 +68,9 @@ parameter_rules:
   - name: enable_search
     type: boolean
     default: false
+    label:
+      zh_Hans: 联网搜索
+      en_US: Web Search
     help:
       zh_Hans: 模型内置了互联网搜索服务，该参数控制模型在生成文本时是否参考使用互联网搜索结果。启用互联网搜索，模型会将搜索结果作为文本生成过程中的参考信息，但模型会基于其内部逻辑“自行判断”是否使用互联网搜索结果。
       en_US: The model has a built-in Internet search service. This parameter controls whether the model refers to Internet search results when generating text. When Internet search is enabled, the model will use the search results as reference information in the text generation process, but the model will "judge" whether to use Internet search results based on its internal logic.
diff --git a/api/core/model_runtime/model_providers/tongyi/llm/qwen-coder-turbo.yaml b/api/core/model_runtime/model_providers/tongyi/llm/qwen-coder-turbo.yaml
index f4032a4dd316d..d2aca4f514ae0 100644
--- a/api/core/model_runtime/model_providers/tongyi/llm/qwen-coder-turbo.yaml
+++ b/api/core/model_runtime/model_providers/tongyi/llm/qwen-coder-turbo.yaml
@@ -60,6 +60,7 @@ parameter_rules:
     type: float
     default: 1.1
     label:
+      zh_Hans: 重复惩罚
       en_US: Repetition penalty
     help:
       zh_Hans: 用于控制模型生成时的重复度。提高repetition_penalty时可以降低模型生成的重复度。1.0表示不做惩罚。
@@ -67,6 +68,9 @@ parameter_rules:
   - name: enable_search
     type: boolean
     default: false
+    label:
+      zh_Hans: 联网搜索
+      en_US: Web Search
     help:
       zh_Hans: 模型内置了互联网搜索服务，该参数控制模型在生成文本时是否参考使用互联网搜索结果。启用互联网搜索，模型会将搜索结果作为文本生成过程中的参考信息，但模型会基于其内部逻辑“自行判断”是否使用互联网搜索结果。
       en_US: The model has a built-in Internet search service. This parameter controls whether the model refers to Internet search results when generating text. When Internet search is enabled, the model will use the search results as reference information in the text generation process, but the model will "judge" whether to use Internet search results based on its internal logic.
diff --git a/api/core/model_runtime/model_providers/tongyi/llm/qwen-long.yaml b/api/core/model_runtime/model_providers/tongyi/llm/qwen-long.yaml
index dbe7d024a50f1..a59a3350f62dd 100644
--- a/api/core/model_runtime/model_providers/tongyi/llm/qwen-long.yaml
+++ b/api/core/model_runtime/model_providers/tongyi/llm/qwen-long.yaml
@@ -63,6 +63,7 @@ parameter_rules:
     type: float
     default: 1.1
     label:
+      zh_Hans: 重复惩罚
       en_US: Repetition penalty
     help:
       zh_Hans: 用于控制模型生成时的重复度。提高repetition_penalty时可以降低模型生成的重复度。1.0表示不做惩罚。
@@ -70,6 +71,9 @@ parameter_rules:
   - name: enable_search
     type: boolean
     default: false
+    label:
+      zh_Hans: 联网搜索
+      en_US: Web Search
     help:
       zh_Hans: 模型内置了互联网搜索服务，该参数控制模型在生成文本时是否参考使用互联网搜索结果。启用互联网搜索，模型会将搜索结果作为文本生成过程中的参考信息，但模型会基于其内部逻辑“自行判断”是否使用互联网搜索结果。
       en_US: The model has a built-in Internet search service. This parameter controls whether the model refers to Internet search results when generating text. When Internet search is enabled, the model will use the search results as reference information in the text generation process, but the model will "judge" whether to use Internet search results based on its internal logic.
diff --git a/api/core/model_runtime/model_providers/tongyi/llm/qwen-math-plus-0816.yaml b/api/core/model_runtime/model_providers/tongyi/llm/qwen-math-plus-0816.yaml
index 89d1302abef70..cab7233c98d69 100644
--- a/api/core/model_runtime/model_providers/tongyi/llm/qwen-math-plus-0816.yaml
+++ b/api/core/model_runtime/model_providers/tongyi/llm/qwen-math-plus-0816.yaml
@@ -60,6 +60,7 @@ parameter_rules:
     type: float
     default: 1.1
     label:
+      zh_Hans: 重复惩罚
       en_US: Repetition penalty
     help:
       zh_Hans: 用于控制模型生成时的重复度。提高repetition_penalty时可以降低模型生成的重复度。1.0表示不做惩罚。
@@ -67,6 +68,9 @@ parameter_rules:
   - name: enable_search
     type: boolean
     default: false
+    label:
+      zh_Hans: 联网搜索
+      en_US: Web Search
     help:
       zh_Hans: 模型内置了互联网搜索服务，该参数控制模型在生成文本时是否参考使用互联网搜索结果。启用互联网搜索，模型会将搜索结果作为文本生成过程中的参考信息，但模型会基于其内部逻辑“自行判断”是否使用互联网搜索结果。
       en_US: The model has a built-in Internet search service. This parameter controls whether the model refers to Internet search results when generating text. When Internet search is enabled, the model will use the search results as reference information in the text generation process, but the model will "judge" whether to use Internet search results based on its internal logic.
diff --git a/api/core/model_runtime/model_providers/tongyi/llm/qwen-math-plus-0919.yaml b/api/core/model_runtime/model_providers/tongyi/llm/qwen-math-plus-0919.yaml
index 032b3c970d7f4..f82fba0c012aa 100644
--- a/api/core/model_runtime/model_providers/tongyi/llm/qwen-math-plus-0919.yaml
+++ b/api/core/model_runtime/model_providers/tongyi/llm/qwen-math-plus-0919.yaml
@@ -60,6 +60,7 @@ parameter_rules:
     type: float
     default: 1.1
     label:
+      zh_Hans: 重复惩罚
       en_US: Repetition penalty
     help:
       zh_Hans: 用于控制模型生成时的重复度。提高repetition_penalty时可以降低模型生成的重复度。1.0表示不做惩罚。
@@ -67,6 +68,9 @@ parameter_rules:
   - name: enable_search
     type: boolean
     default: false
+    label:
+      zh_Hans: 联网搜索
+      en_US: Web Search
     help:
       zh_Hans: 模型内置了互联网搜索服务，该参数控制模型在生成文本时是否参考使用互联网搜索结果。启用互联网搜索，模型会将搜索结果作为文本生成过程中的参考信息，但模型会基于其内部逻辑“自行判断”是否使用互联网搜索结果。
       en_US: The model has a built-in Internet search service. This parameter controls whether the model refers to Internet search results when generating text. When Internet search is enabled, the model will use the search results as reference information in the text generation process, but the model will "judge" whether to use Internet search results based on its internal logic.
diff --git a/api/core/model_runtime/model_providers/tongyi/llm/qwen-math-plus-latest.yaml b/api/core/model_runtime/model_providers/tongyi/llm/qwen-math-plus-latest.yaml
index 31dd9f6972576..e2fb6e0e554d4 100644
--- a/api/core/model_runtime/model_providers/tongyi/llm/qwen-math-plus-latest.yaml
+++ b/api/core/model_runtime/model_providers/tongyi/llm/qwen-math-plus-latest.yaml
@@ -60,6 +60,7 @@ parameter_rules:
     type: float
     default: 1.1
     label:
+      zh_Hans: 重复惩罚
       en_US: Repetition penalty
     help:
       zh_Hans: 用于控制模型生成时的重复度。提高repetition_penalty时可以降低模型生成的重复度。1.0表示不做惩罚。
@@ -67,6 +68,9 @@ parameter_rules:
   - name: enable_search
     type: boolean
     default: false
+    label:
+      zh_Hans: 联网搜索
+      en_US: Web Search
     help:
       zh_Hans: 模型内置了互联网搜索服务，该参数控制模型在生成文本时是否参考使用互联网搜索结果。启用互联网搜索，模型会将搜索结果作为文本生成过程中的参考信息，但模型会基于其内部逻辑“自行判断”是否使用互联网搜索结果。
       en_US: The model has a built-in Internet search service. This parameter controls whether the model refers to Internet search results when generating text. When Internet search is enabled, the model will use the search results as reference information in the text generation process, but the model will "judge" whether to use Internet search results based on its internal logic.
diff --git a/api/core/model_runtime/model_providers/tongyi/llm/qwen-math-plus.yaml b/api/core/model_runtime/model_providers/tongyi/llm/qwen-math-plus.yaml
index 1a51d57f7814b..8803e747e5eb7 100644
--- a/api/core/model_runtime/model_providers/tongyi/llm/qwen-math-plus.yaml
+++ b/api/core/model_runtime/model_providers/tongyi/llm/qwen-math-plus.yaml
@@ -60,6 +60,7 @@ parameter_rules:
     type: float
     default: 1.1
     label:
+      zh_Hans: 重复惩罚
       en_US: Repetition penalty
     help:
       zh_Hans: 用于控制模型生成时的重复度。提高repetition_penalty时可以降低模型生成的重复度。1.0表示不做惩罚。
@@ -67,6 +68,9 @@ parameter_rules:
   - name: enable_search
     type: boolean
     default: false
+    label:
+      zh_Hans: 联网搜索
+      en_US: Web Search
     help:
       zh_Hans: 模型内置了互联网搜索服务，该参数控制模型在生成文本时是否参考使用互联网搜索结果。启用互联网搜索，模型会将搜索结果作为文本生成过程中的参考信息，但模型会基于其内部逻辑“自行判断”是否使用互联网搜索结果。
       en_US: The model has a built-in Internet search service. This parameter controls whether the model refers to Internet search results when generating text. When Internet search is enabled, the model will use the search results as reference information in the text generation process, but the model will "judge" whether to use Internet search results based on its internal logic.
diff --git a/api/core/model_runtime/model_providers/tongyi/llm/qwen-math-turbo-0919.yaml b/api/core/model_runtime/model_providers/tongyi/llm/qwen-math-turbo-0919.yaml
index 1894eea417a43..0dc5a066f0300 100644
--- a/api/core/model_runtime/model_providers/tongyi/llm/qwen-math-turbo-0919.yaml
+++ b/api/core/model_runtime/model_providers/tongyi/llm/qwen-math-turbo-0919.yaml
@@ -60,6 +60,7 @@ parameter_rules:
     type: float
     default: 1.1
     label:
+      zh_Hans: 重复惩罚
       en_US: Repetition penalty
     help:
       zh_Hans: 用于控制模型生成时的重复度。提高repetition_penalty时可以降低模型生成的重复度。1.0表示不做惩罚。
@@ -67,6 +68,9 @@ parameter_rules:
   - name: enable_search
     type: boolean
     default: false
+    label:
+      zh_Hans: 联网搜索
+      en_US: Web Search
     help:
       zh_Hans: 模型内置了互联网搜索服务，该参数控制模型在生成文本时是否参考使用互联网搜索结果。启用互联网搜索，模型会将搜索结果作为文本生成过程中的参考信息，但模型会基于其内部逻辑“自行判断”是否使用互联网搜索结果。
       en_US: The model has a built-in Internet search service. This parameter controls whether the model refers to Internet search results when generating text. When Internet search is enabled, the model will use the search results as reference information in the text generation process, but the model will "judge" whether to use Internet search results based on its internal logic.
diff --git a/api/core/model_runtime/model_providers/tongyi/llm/qwen-math-turbo-latest.yaml b/api/core/model_runtime/model_providers/tongyi/llm/qwen-math-turbo-latest.yaml
index b8365618b0499..2ac0e4692a701 100644
--- a/api/core/model_runtime/model_providers/tongyi/llm/qwen-math-turbo-latest.yaml
+++ b/api/core/model_runtime/model_providers/tongyi/llm/qwen-math-turbo-latest.yaml
@@ -60,6 +60,7 @@ parameter_rules:
     type: float
     default: 1.1
     label:
+      zh_Hans: 重复惩罚
       en_US: Repetition penalty
     help:
       zh_Hans: 用于控制模型生成时的重复度。提高repetition_penalty时可以降低模型生成的重复度。1.0表示不做惩罚。
@@ -67,6 +68,9 @@ parameter_rules:
   - name: enable_search
     type: boolean
     default: false
+    label:
+      zh_Hans: 联网搜索
+      en_US: Web Search
     help:
       zh_Hans: 模型内置了互联网搜索服务，该参数控制模型在生成文本时是否参考使用互联网搜索结果。启用互联网搜索，模型会将搜索结果作为文本生成过程中的参考信息，但模型会基于其内部逻辑“自行判断”是否使用互联网搜索结果。
       en_US: The model has a built-in Internet search service. This parameter controls whether the model refers to Internet search results when generating text. When Internet search is enabled, the model will use the search results as reference information in the text generation process, but the model will "judge" whether to use Internet search results based on its internal logic.
diff --git a/api/core/model_runtime/model_providers/tongyi/llm/qwen-math-turbo.yaml b/api/core/model_runtime/model_providers/tongyi/llm/qwen-math-turbo.yaml
index 8d346d691e6b2..9a7f1312e9724 100644
--- a/api/core/model_runtime/model_providers/tongyi/llm/qwen-math-turbo.yaml
+++ b/api/core/model_runtime/model_providers/tongyi/llm/qwen-math-turbo.yaml
@@ -60,6 +60,7 @@ parameter_rules:
     type: float
     default: 1.1
     label:
+      zh_Hans: 重复惩罚
       en_US: Repetition penalty
     help:
       zh_Hans: 用于控制模型生成时的重复度。提高repetition_penalty时可以降低模型生成的重复度。1.0表示不做惩罚。
@@ -67,6 +68,9 @@ parameter_rules:
   - name: enable_search
     type: boolean
     default: false
+    label:
+      zh_Hans: 联网搜索
+      en_US: Web Search
     help:
       zh_Hans: 模型内置了互联网搜索服务，该参数控制模型在生成文本时是否参考使用互联网搜索结果。启用互联网搜索，模型会将搜索结果作为文本生成过程中的参考信息，但模型会基于其内部逻辑“自行判断”是否使用互联网搜索结果。
       en_US: The model has a built-in Internet search service. This parameter controls whether the model refers to Internet search results when generating text. When Internet search is enabled, the model will use the search results as reference information in the text generation process, but the model will "judge" whether to use Internet search results based on its internal logic.
diff --git a/api/core/model_runtime/model_providers/tongyi/llm/qwen-max-0107.yaml b/api/core/model_runtime/model_providers/tongyi/llm/qwen-max-0107.yaml
index c0ad12b85e546..c0eef375570d4 100644
--- a/api/core/model_runtime/model_providers/tongyi/llm/qwen-max-0107.yaml
+++ b/api/core/model_runtime/model_providers/tongyi/llm/qwen-max-0107.yaml
@@ -62,6 +62,7 @@ parameter_rules:
     type: float
     default: 1.1
     label:
+      zh_Hans: 重复惩罚
       en_US: Repetition penalty
     help:
       zh_Hans: 用于控制模型生成时的重复度。提高repetition_penalty时可以降低模型生成的重复度。1.0表示不做惩罚。
@@ -69,6 +70,9 @@ parameter_rules:
   - name: enable_search
     type: boolean
     default: false
+    label:
+      zh_Hans: 联网搜索
+      en_US: Web Search
     help:
       zh_Hans: 模型内置了互联网搜索服务，该参数控制模型在生成文本时是否参考使用互联网搜索结果。启用互联网搜索，模型会将搜索结果作为文本生成过程中的参考信息，但模型会基于其内部逻辑“自行判断”是否使用互联网搜索结果。
       en_US: The model has a built-in Internet search service. This parameter controls whether the model refers to Internet search results when generating text. When Internet search is enabled, the model will use the search results as reference information in the text generation process, but the model will "judge" whether to use Internet search results based on its internal logic.
diff --git a/api/core/model_runtime/model_providers/tongyi/llm/qwen-max-0403.yaml b/api/core/model_runtime/model_providers/tongyi/llm/qwen-max-0403.yaml
index b00fb44d29fa7..c12444bd7b57b 100644
--- a/api/core/model_runtime/model_providers/tongyi/llm/qwen-max-0403.yaml
+++ b/api/core/model_runtime/model_providers/tongyi/llm/qwen-max-0403.yaml
@@ -62,6 +62,7 @@ parameter_rules:
     type: float
     default: 1.1
     label:
+      zh_Hans: 重复惩罚
       en_US: Repetition penalty
     help:
       zh_Hans: 用于控制模型生成时的重复度。提高repetition_penalty时可以降低模型生成的重复度。1.0表示不做惩罚。
@@ -69,6 +70,9 @@ parameter_rules:
   - name: enable_search
     type: boolean
     default: false
+    label:
+      zh_Hans: 联网搜索
+      en_US: Web Search
     help:
       zh_Hans: 模型内置了互联网搜索服务，该参数控制模型在生成文本时是否参考使用互联网搜索结果。启用互联网搜索，模型会将搜索结果作为文本生成过程中的参考信息，但模型会基于其内部逻辑“自行判断”是否使用互联网搜索结果。
       en_US: The model has a built-in Internet search service. This parameter controls whether the model refers to Internet search results when generating text. When Internet search is enabled, the model will use the search results as reference information in the text generation process, but the model will "judge" whether to use Internet search results based on its internal logic.
diff --git a/api/core/model_runtime/model_providers/tongyi/llm/qwen-max-0428.yaml b/api/core/model_runtime/model_providers/tongyi/llm/qwen-max-0428.yaml
index 1848dcc07d185..173c55b6b9796 100644
--- a/api/core/model_runtime/model_providers/tongyi/llm/qwen-max-0428.yaml
+++ b/api/core/model_runtime/model_providers/tongyi/llm/qwen-max-0428.yaml
@@ -62,6 +62,7 @@ parameter_rules:
     type: float
     default: 1.1
     label:
+      zh_Hans: 重复惩罚
       en_US: Repetition penalty
     help:
       zh_Hans: 用于控制模型生成时的重复度。提高repetition_penalty时可以降低模型生成的重复度。1.0表示不做惩罚。
@@ -69,6 +70,9 @@ parameter_rules:
   - name: enable_search
     type: boolean
     default: false
+    label:
+      zh_Hans: 联网搜索
+      en_US: Web Search
     help:
       zh_Hans: 模型内置了互联网搜索服务，该参数控制模型在生成文本时是否参考使用互联网搜索结果。启用互联网搜索，模型会将搜索结果作为文本生成过程中的参考信息，但模型会基于其内部逻辑“自行判断”是否使用互联网搜索结果。
       en_US: The model has a built-in Internet search service. This parameter controls whether the model refers to Internet search results when generating text. When Internet search is enabled, the model will use the search results as reference information in the text generation process, but the model will "judge" whether to use Internet search results based on its internal logic.
diff --git a/api/core/model_runtime/model_providers/tongyi/llm/qwen-max-0919.yaml b/api/core/model_runtime/model_providers/tongyi/llm/qwen-max-0919.yaml
index 238882bb12189..692a38140dbed 100644
--- a/api/core/model_runtime/model_providers/tongyi/llm/qwen-max-0919.yaml
+++ b/api/core/model_runtime/model_providers/tongyi/llm/qwen-max-0919.yaml
@@ -62,6 +62,7 @@ parameter_rules:
     type: float
     default: 1.1
     label:
+      zh_Hans: 重复惩罚
       en_US: Repetition penalty
     help:
       zh_Hans: 用于控制模型生成时的重复度。提高repetition_penalty时可以降低模型生成的重复度。1.0表示不做惩罚。
@@ -69,6 +70,9 @@ parameter_rules:
   - name: enable_search
     type: boolean
     default: false
+    label:
+      zh_Hans: 联网搜索
+      en_US: Web Search
     help:
       zh_Hans: 模型内置了互联网搜索服务，该参数控制模型在生成文本时是否参考使用互联网搜索结果。启用互联网搜索，模型会将搜索结果作为文本生成过程中的参考信息，但模型会基于其内部逻辑“自行判断”是否使用互联网搜索结果。
       en_US: The model has a built-in Internet search service. This parameter controls whether the model refers to Internet search results when generating text. When Internet search is enabled, the model will use the search results as reference information in the text generation process, but the model will "judge" whether to use Internet search results based on its internal logic.
diff --git a/api/core/model_runtime/model_providers/tongyi/llm/qwen-max-latest.yaml b/api/core/model_runtime/model_providers/tongyi/llm/qwen-max-latest.yaml
index 9d7d3c2fcbf67..afd7fb4b77696 100644
--- a/api/core/model_runtime/model_providers/tongyi/llm/qwen-max-latest.yaml
+++ b/api/core/model_runtime/model_providers/tongyi/llm/qwen-max-latest.yaml
@@ -62,6 +62,7 @@ parameter_rules:
     type: float
     default: 1.1
     label:
+      zh_Hans: 重复惩罚
       en_US: Repetition penalty
     help:
       zh_Hans: 用于控制模型生成时的重复度。提高repetition_penalty时可以降低模型生成的重复度。1.0表示不做惩罚。
@@ -69,6 +70,9 @@ parameter_rules:
   - name: enable_search
     type: boolean
     default: false
+    label:
+      zh_Hans: 联网搜索
+      en_US: Web Search
     help:
       zh_Hans: 模型内置了互联网搜索服务，该参数控制模型在生成文本时是否参考使用互联网搜索结果。启用互联网搜索，模型会将搜索结果作为文本生成过程中的参考信息，但模型会基于其内部逻辑“自行判断”是否使用互联网搜索结果。
       en_US: The model has a built-in Internet search service. This parameter controls whether the model refers to Internet search results when generating text. When Internet search is enabled, the model will use the search results as reference information in the text generation process, but the model will "judge" whether to use Internet search results based on its internal logic.
diff --git a/api/core/model_runtime/model_providers/tongyi/llm/qwen-max-longcontext.yaml b/api/core/model_runtime/model_providers/tongyi/llm/qwen-max-longcontext.yaml
index a7bdc42f7347e..d02ba7af18318 100644
--- a/api/core/model_runtime/model_providers/tongyi/llm/qwen-max-longcontext.yaml
+++ b/api/core/model_runtime/model_providers/tongyi/llm/qwen-max-longcontext.yaml
@@ -62,6 +62,7 @@ parameter_rules:
     type: float
     default: 1.1
     label:
+      zh_Hans: 重复惩罚
       en_US: Repetition penalty
     help:
       zh_Hans: 用于控制模型生成时的重复度。提高repetition_penalty时可以降低模型生成的重复度。1.0表示不做惩罚。
@@ -69,6 +70,9 @@ parameter_rules:
   - name: enable_search
     type: boolean
     default: false
+    label:
+      zh_Hans: 联网搜索
+      en_US: Web Search
     help:
       zh_Hans: 模型内置了互联网搜索服务，该参数控制模型在生成文本时是否参考使用互联网搜索结果。启用互联网搜索，模型会将搜索结果作为文本生成过程中的参考信息，但模型会基于其内部逻辑“自行判断”是否使用互联网搜索结果。
       en_US: The model has a built-in Internet search service. This parameter controls whether the model refers to Internet search results when generating text. When Internet search is enabled, the model will use the search results as reference information in the text generation process, but the model will "judge" whether to use Internet search results based on its internal logic.
diff --git a/api/core/model_runtime/model_providers/tongyi/llm/qwen-max.yaml b/api/core/model_runtime/model_providers/tongyi/llm/qwen-max.yaml
index 57888406afa0a..c6a64dc507497 100644
--- a/api/core/model_runtime/model_providers/tongyi/llm/qwen-max.yaml
+++ b/api/core/model_runtime/model_providers/tongyi/llm/qwen-max.yaml
@@ -62,6 +62,7 @@ parameter_rules:
     type: float
     default: 1.1
     label:
+      zh_Hans: 重复惩罚
       en_US: Repetition penalty
     help:
       zh_Hans: 用于控制模型生成时的重复度。提高repetition_penalty时可以降低模型生成的重复度。1.0表示不做惩罚。
@@ -69,6 +70,9 @@ parameter_rules:
   - name: enable_search
     type: boolean
     default: false
+    label:
+      zh_Hans: 联网搜索
+      en_US: Web Search
     help:
       zh_Hans: 模型内置了互联网搜索服务，该参数控制模型在生成文本时是否参考使用互联网搜索结果。启用互联网搜索，模型会将搜索结果作为文本生成过程中的参考信息，但模型会基于其内部逻辑“自行判断”是否使用互联网搜索结果。
       en_US: The model has a built-in Internet search service. This parameter controls whether the model refers to Internet search results when generating text. When Internet search is enabled, the model will use the search results as reference information in the text generation process, but the model will "judge" whether to use Internet search results based on its internal logic.
diff --git a/api/core/model_runtime/model_providers/tongyi/llm/qwen-plus-0206.yaml b/api/core/model_runtime/model_providers/tongyi/llm/qwen-plus-0206.yaml
index 1e0b81661731a..1111298c3712a 100644
--- a/api/core/model_runtime/model_providers/tongyi/llm/qwen-plus-0206.yaml
+++ b/api/core/model_runtime/model_providers/tongyi/llm/qwen-plus-0206.yaml
@@ -60,6 +60,7 @@ parameter_rules:
     type: float
     default: 1.1
     label:
+      zh_Hans: 重复惩罚
       en_US: Repetition penalty
     help:
       zh_Hans: 用于控制模型生成时的重复度。提高repetition_penalty时可以降低模型生成的重复度。1.0表示不做惩罚。
@@ -67,6 +68,9 @@ parameter_rules:
   - name: enable_search
     type: boolean
     default: false
+    label:
+      zh_Hans: 联网搜索
+      en_US: Web Search
     help:
       zh_Hans: 模型内置了互联网搜索服务，该参数控制模型在生成文本时是否参考使用互联网搜索结果。启用互联网搜索，模型会将搜索结果作为文本生成过程中的参考信息，但模型会基于其内部逻辑“自行判断”是否使用互联网搜索结果。
       en_US: The model has a built-in Internet search service. This parameter controls whether the model refers to Internet search results when generating text. When Internet search is enabled, the model will use the search results as reference information in the text generation process, but the model will "judge" whether to use Internet search results based on its internal logic.
diff --git a/api/core/model_runtime/model_providers/tongyi/llm/qwen-plus-0624.yaml b/api/core/model_runtime/model_providers/tongyi/llm/qwen-plus-0624.yaml
index f70c373922fb3..ef8dd083ade5d 100644
--- a/api/core/model_runtime/model_providers/tongyi/llm/qwen-plus-0624.yaml
+++ b/api/core/model_runtime/model_providers/tongyi/llm/qwen-plus-0624.yaml
@@ -60,6 +60,7 @@ parameter_rules:
     type: float
     default: 1.1
     label:
+      zh_Hans: 重复惩罚
       en_US: Repetition penalty
     help:
       zh_Hans: 用于控制模型生成时的重复度。提高repetition_penalty时可以降低模型生成的重复度。1.0表示不做惩罚。
@@ -67,6 +68,9 @@ parameter_rules:
   - name: enable_search
     type: boolean
     default: false
+    label:
+      zh_Hans: 联网搜索
+      en_US: Web Search
     help:
       zh_Hans: 模型内置了互联网搜索服务，该参数控制模型在生成文本时是否参考使用互联网搜索结果。启用互联网搜索，模型会将搜索结果作为文本生成过程中的参考信息，但模型会基于其内部逻辑“自行判断”是否使用互联网搜索结果。
       en_US: The model has a built-in Internet search service. This parameter controls whether the model refers to Internet search results when generating text. When Internet search is enabled, the model will use the search results as reference information in the text generation process, but the model will "judge" whether to use Internet search results based on its internal logic.
diff --git a/api/core/model_runtime/model_providers/tongyi/llm/qwen-plus-0723.yaml b/api/core/model_runtime/model_providers/tongyi/llm/qwen-plus-0723.yaml
index c6007e9164ca2..87a4417df59b8 100644
--- a/api/core/model_runtime/model_providers/tongyi/llm/qwen-plus-0723.yaml
+++ b/api/core/model_runtime/model_providers/tongyi/llm/qwen-plus-0723.yaml
@@ -60,6 +60,7 @@ parameter_rules:
     type: float
     default: 1.1
     label:
+      zh_Hans: 重复惩罚
       en_US: Repetition penalty
     help:
       zh_Hans: 用于控制模型生成时的重复度。提高repetition_penalty时可以降低模型生成的重复度。1.0表示不做惩罚。
@@ -67,6 +68,9 @@ parameter_rules:
   - name: enable_search
     type: boolean
     default: false
+    label:
+      zh_Hans: 联网搜索
+      en_US: Web Search
     help:
       zh_Hans: 模型内置了互联网搜索服务，该参数控制模型在生成文本时是否参考使用互联网搜索结果。启用互联网搜索，模型会将搜索结果作为文本生成过程中的参考信息，但模型会基于其内部逻辑“自行判断”是否使用互联网搜索结果。
       en_US: The model has a built-in Internet search service. This parameter controls whether the model refers to Internet search results when generating text. When Internet search is enabled, the model will use the search results as reference information in the text generation process, but the model will "judge" whether to use Internet search results based on its internal logic.
diff --git a/api/core/model_runtime/model_providers/tongyi/llm/qwen-plus-0806.yaml b/api/core/model_runtime/model_providers/tongyi/llm/qwen-plus-0806.yaml
index 2f53c43336698..967f258fa9162 100644
--- a/api/core/model_runtime/model_providers/tongyi/llm/qwen-plus-0806.yaml
+++ b/api/core/model_runtime/model_providers/tongyi/llm/qwen-plus-0806.yaml
@@ -60,6 +60,7 @@ parameter_rules:
     type: float
     default: 1.1
     label:
+      zh_Hans: 重复惩罚
       en_US: Repetition penalty
     help:
       zh_Hans: 用于控制模型生成时的重复度。提高repetition_penalty时可以降低模型生成的重复度。1.0表示不做惩罚。
@@ -67,6 +68,9 @@ parameter_rules:
   - name: enable_search
     type: boolean
     default: false
+    label:
+      zh_Hans: 联网搜索
+      en_US: Web Search
     help:
       zh_Hans: 模型内置了互联网搜索服务，该参数控制模型在生成文本时是否参考使用互联网搜索结果。启用互联网搜索，模型会将搜索结果作为文本生成过程中的参考信息，但模型会基于其内部逻辑“自行判断”是否使用互联网搜索结果。
       en_US: The model has a built-in Internet search service. This parameter controls whether the model refers to Internet search results when generating text. When Internet search is enabled, the model will use the search results as reference information in the text generation process, but the model will "judge" whether to use Internet search results based on its internal logic.
diff --git a/api/core/model_runtime/model_providers/tongyi/llm/qwen-plus-0919.yaml b/api/core/model_runtime/model_providers/tongyi/llm/qwen-plus-0919.yaml
index 90b54ca52e2d4..9d44852ac9cd5 100644
--- a/api/core/model_runtime/model_providers/tongyi/llm/qwen-plus-0919.yaml
+++ b/api/core/model_runtime/model_providers/tongyi/llm/qwen-plus-0919.yaml
@@ -60,6 +60,7 @@ parameter_rules:
     type: float
     default: 1.1
     label:
+      zh_Hans: 重复惩罚
       en_US: Repetition penalty
     help:
       zh_Hans: 用于控制模型生成时的重复度。提高repetition_penalty时可以降低模型生成的重复度。1.0表示不做惩罚。
@@ -67,6 +68,9 @@ parameter_rules:
   - name: enable_search
     type: boolean
     default: false
+    label:
+      zh_Hans: 联网搜索
+      en_US: Web Search
     help:
       zh_Hans: 模型内置了互联网搜索服务，该参数控制模型在生成文本时是否参考使用互联网搜索结果。启用互联网搜索，模型会将搜索结果作为文本生成过程中的参考信息，但模型会基于其内部逻辑“自行判断”是否使用互联网搜索结果。
       en_US: The model has a built-in Internet search service. This parameter controls whether the model refers to Internet search results when generating text. When Internet search is enabled, the model will use the search results as reference information in the text generation process, but the model will "judge" whether to use Internet search results based on its internal logic.
diff --git a/api/core/model_runtime/model_providers/tongyi/llm/qwen-plus-chat.yaml b/api/core/model_runtime/model_providers/tongyi/llm/qwen-plus-chat.yaml
index 59e8851240b0a..df9448ae04ddf 100644
--- a/api/core/model_runtime/model_providers/tongyi/llm/qwen-plus-chat.yaml
+++ b/api/core/model_runtime/model_providers/tongyi/llm/qwen-plus-chat.yaml
@@ -62,6 +62,7 @@ parameter_rules:
     type: float
     default: 1.1
     label:
+      zh_Hans: 重复惩罚
       en_US: Repetition penalty
     help:
       zh_Hans: 用于控制模型生成时的重复度。提高repetition_penalty时可以降低模型生成的重复度。1.0表示不做惩罚。
@@ -69,6 +70,9 @@ parameter_rules:
   - name: enable_search
     type: boolean
     default: false
+    label:
+      zh_Hans: 联网搜索
+      en_US: Web Search
     help:
       zh_Hans: 模型内置了互联网搜索服务，该参数控制模型在生成文本时是否参考使用互联网搜索结果。启用互联网搜索，模型会将搜索结果作为文本生成过程中的参考信息，但模型会基于其内部逻辑“自行判断”是否使用互联网搜索结果。
       en_US: The model has a built-in Internet search service. This parameter controls whether the model refers to Internet search results when generating text. When Internet search is enabled, the model will use the search results as reference information in the text generation process, but the model will "judge" whether to use Internet search results based on its internal logic.
diff --git a/api/core/model_runtime/model_providers/tongyi/llm/qwen-plus-latest.yaml b/api/core/model_runtime/model_providers/tongyi/llm/qwen-plus-latest.yaml
index 2a821dbcfe001..32ccb8d615903 100644
--- a/api/core/model_runtime/model_providers/tongyi/llm/qwen-plus-latest.yaml
+++ b/api/core/model_runtime/model_providers/tongyi/llm/qwen-plus-latest.yaml
@@ -60,6 +60,7 @@ parameter_rules:
     type: float
     default: 1.1
     label:
+      zh_Hans: 重复惩罚
       en_US: Repetition penalty
     help:
       zh_Hans: 用于控制模型生成时的重复度。提高repetition_penalty时可以降低模型生成的重复度。1.0表示不做惩罚。
@@ -67,6 +68,9 @@ parameter_rules:
   - name: enable_search
     type: boolean
     default: false
+    label:
+      zh_Hans: 联网搜索
+      en_US: Web Search
     help:
       zh_Hans: 模型内置了互联网搜索服务，该参数控制模型在生成文本时是否参考使用互联网搜索结果。启用互联网搜索，模型会将搜索结果作为文本生成过程中的参考信息，但模型会基于其内部逻辑“自行判断”是否使用互联网搜索结果。
       en_US: The model has a built-in Internet search service. This parameter controls whether the model refers to Internet search results when generating text. When Internet search is enabled, the model will use the search results as reference information in the text generation process, but the model will "judge" whether to use Internet search results based on its internal logic.
diff --git a/api/core/model_runtime/model_providers/tongyi/llm/qwen-plus.yaml b/api/core/model_runtime/model_providers/tongyi/llm/qwen-plus.yaml
index 626884f4b2954..f3fce30209e48 100644
--- a/api/core/model_runtime/model_providers/tongyi/llm/qwen-plus.yaml
+++ b/api/core/model_runtime/model_providers/tongyi/llm/qwen-plus.yaml
@@ -62,6 +62,7 @@ parameter_rules:
     type: float
     default: 1.1
     label:
+      zh_Hans: 重复惩罚
       en_US: Repetition penalty
     help:
       zh_Hans: 用于控制模型生成时的重复度。提高repetition_penalty时可以降低模型生成的重复度。1.0表示不做惩罚。
@@ -69,6 +70,9 @@ parameter_rules:
   - name: enable_search
     type: boolean
     default: false
+    label:
+      zh_Hans: 联网搜索
+      en_US: Web Search
     help:
       zh_Hans: 模型内置了互联网搜索服务，该参数控制模型在生成文本时是否参考使用互联网搜索结果。启用互联网搜索，模型会将搜索结果作为文本生成过程中的参考信息，但模型会基于其内部逻辑“自行判断”是否使用互联网搜索结果。
       en_US: The model has a built-in Internet search service. This parameter controls whether the model refers to Internet search results when generating text. When Internet search is enabled, the model will use the search results as reference information in the text generation process, but the model will "judge" whether to use Internet search results based on its internal logic.
diff --git a/api/core/model_runtime/model_providers/tongyi/llm/qwen-turbo-0206.yaml b/api/core/model_runtime/model_providers/tongyi/llm/qwen-turbo-0206.yaml
index 844fced77a684..bf976b518a911 100644
--- a/api/core/model_runtime/model_providers/tongyi/llm/qwen-turbo-0206.yaml
+++ b/api/core/model_runtime/model_providers/tongyi/llm/qwen-turbo-0206.yaml
@@ -60,6 +60,7 @@ parameter_rules:
     type: float
     default: 1.1
     label:
+      zh_Hans: 重复惩罚
       en_US: Repetition penalty
     help:
       zh_Hans: 用于控制模型生成时的重复度。提高repetition_penalty时可以降低模型生成的重复度。1.0表示不做惩罚。
@@ -67,6 +68,9 @@ parameter_rules:
   - name: enable_search
     type: boolean
     default: false
+    label:
+      zh_Hans: 联网搜索
+      en_US: Web Search
     help:
       zh_Hans: 模型内置了互联网搜索服务，该参数控制模型在生成文本时是否参考使用互联网搜索结果。启用互联网搜索，模型会将搜索结果作为文本生成过程中的参考信息，但模型会基于其内部逻辑“自行判断”是否使用互联网搜索结果。
       en_US: The model has a built-in Internet search service. This parameter controls whether the model refers to Internet search results when generating text. When Internet search is enabled, the model will use the search results as reference information in the text generation process, but the model will "judge" whether to use Internet search results based on its internal logic.
diff --git a/api/core/model_runtime/model_providers/tongyi/llm/qwen-turbo-0624.yaml b/api/core/model_runtime/model_providers/tongyi/llm/qwen-turbo-0624.yaml
index 0152f75579e5c..060e7fb4c9d09 100644
--- a/api/core/model_runtime/model_providers/tongyi/llm/qwen-turbo-0624.yaml
+++ b/api/core/model_runtime/model_providers/tongyi/llm/qwen-turbo-0624.yaml
@@ -60,6 +60,7 @@ parameter_rules:
     type: float
     default: 1.1
     label:
+      zh_Hans: 重复惩罚
       en_US: Repetition penalty
     help:
       zh_Hans: 用于控制模型生成时的重复度。提高repetition_penalty时可以降低模型生成的重复度。1.0表示不做惩罚。
@@ -67,6 +68,9 @@ parameter_rules:
   - name: enable_search
     type: boolean
     default: false
+    label:
+      zh_Hans: 联网搜索
+      en_US: Web Search
     help:
       zh_Hans: 模型内置了互联网搜索服务，该参数控制模型在生成文本时是否参考使用互联网搜索结果。启用互联网搜索，模型会将搜索结果作为文本生成过程中的参考信息，但模型会基于其内部逻辑“自行判断”是否使用互联网搜索结果。
       en_US: The model has a built-in Internet search service. This parameter controls whether the model refers to Internet search results when generating text. When Internet search is enabled, the model will use the search results as reference information in the text generation process, but the model will "judge" whether to use Internet search results based on its internal logic.
diff --git a/api/core/model_runtime/model_providers/tongyi/llm/qwen-turbo-0919.yaml b/api/core/model_runtime/model_providers/tongyi/llm/qwen-turbo-0919.yaml
index 19c6c8d293f4e..97cd34929b4f4 100644
--- a/api/core/model_runtime/model_providers/tongyi/llm/qwen-turbo-0919.yaml
+++ b/api/core/model_runtime/model_providers/tongyi/llm/qwen-turbo-0919.yaml
@@ -60,6 +60,7 @@ parameter_rules:
     type: float
     default: 1.1
     label:
+      zh_Hans: 重复惩罚
       en_US: Repetition penalty
     help:
       zh_Hans: 用于控制模型生成时的重复度。提高repetition_penalty时可以降低模型生成的重复度。1.0表示不做惩罚。
@@ -67,6 +68,9 @@ parameter_rules:
   - name: enable_search
     type: boolean
     default: false
+    label:
+      zh_Hans: 联网搜索
+      en_US: Web Search
     help:
       zh_Hans: 模型内置了互联网搜索服务，该参数控制模型在生成文本时是否参考使用互联网搜索结果。启用互联网搜索，模型会将搜索结果作为文本生成过程中的参考信息，但模型会基于其内部逻辑“自行判断”是否使用互联网搜索结果。
       en_US: The model has a built-in Internet search service. This parameter controls whether the model refers to Internet search results when generating text. When Internet search is enabled, the model will use the search results as reference information in the text generation process, but the model will "judge" whether to use Internet search results based on its internal logic.
diff --git a/api/core/model_runtime/model_providers/tongyi/llm/qwen-turbo-chat.yaml b/api/core/model_runtime/model_providers/tongyi/llm/qwen-turbo-chat.yaml
index f557f311ef9c6..8d77ba7a2a698 100644
--- a/api/core/model_runtime/model_providers/tongyi/llm/qwen-turbo-chat.yaml
+++ b/api/core/model_runtime/model_providers/tongyi/llm/qwen-turbo-chat.yaml
@@ -62,6 +62,7 @@ parameter_rules:
     type: float
     default: 1.1
     label:
+      zh_Hans: 重复惩罚
       en_US: Repetition penalty
     help:
       zh_Hans: 用于控制模型生成时的重复度。提高repetition_penalty时可以降低模型生成的重复度。1.0表示不做惩罚。
@@ -69,6 +70,9 @@ parameter_rules:
   - name: enable_search
     type: boolean
     default: false
+    label:
+      zh_Hans: 联网搜索
+      en_US: Web Search
     help:
       zh_Hans: 模型内置了互联网搜索服务，该参数控制模型在生成文本时是否参考使用互联网搜索结果。启用互联网搜索，模型会将搜索结果作为文本生成过程中的参考信息，但模型会基于其内部逻辑“自行判断”是否使用互联网搜索结果。
       en_US: The model has a built-in Internet search service. This parameter controls whether the model refers to Internet search results when generating text. When Internet search is enabled, the model will use the search results as reference information in the text generation process, but the model will "judge" whether to use Internet search results based on its internal logic.
diff --git a/api/core/model_runtime/model_providers/tongyi/llm/qwen-turbo-latest.yaml b/api/core/model_runtime/model_providers/tongyi/llm/qwen-turbo-latest.yaml
index be2475847ef22..4458c706aad40 100644
--- a/api/core/model_runtime/model_providers/tongyi/llm/qwen-turbo-latest.yaml
+++ b/api/core/model_runtime/model_providers/tongyi/llm/qwen-turbo-latest.yaml
@@ -60,6 +60,7 @@ parameter_rules:
     type: float
     default: 1.1
     label:
+      zh_Hans: 重复惩罚
       en_US: Repetition penalty
     help:
       zh_Hans: 用于控制模型生成时的重复度。提高repetition_penalty时可以降低模型生成的重复度。1.0表示不做惩罚。
@@ -67,6 +68,9 @@ parameter_rules:
   - name: enable_search
     type: boolean
     default: false
+    label:
+      zh_Hans: 联网搜索
+      en_US: Web Search
     help:
       zh_Hans: 模型内置了互联网搜索服务，该参数控制模型在生成文本时是否参考使用互联网搜索结果。启用互联网搜索，模型会将搜索结果作为文本生成过程中的参考信息，但模型会基于其内部逻辑“自行判断”是否使用互联网搜索结果。
       en_US: The model has a built-in Internet search service. This parameter controls whether the model refers to Internet search results when generating text. When Internet search is enabled, the model will use the search results as reference information in the text generation process, but the model will "judge" whether to use Internet search results based on its internal logic.
diff --git a/api/core/model_runtime/model_providers/tongyi/llm/qwen-turbo.yaml b/api/core/model_runtime/model_providers/tongyi/llm/qwen-turbo.yaml
index 90f13dc19f5ae..33f05967c2a46 100644
--- a/api/core/model_runtime/model_providers/tongyi/llm/qwen-turbo.yaml
+++ b/api/core/model_runtime/model_providers/tongyi/llm/qwen-turbo.yaml
@@ -62,6 +62,7 @@ parameter_rules:
     type: float
     default: 1.1
     label:
+      zh_Hans: 重复惩罚
       en_US: Repetition penalty
     help:
       zh_Hans: 用于控制模型生成时的重复度。提高repetition_penalty时可以降低模型生成的重复度。1.0表示不做惩罚。
@@ -69,6 +70,9 @@ parameter_rules:
   - name: enable_search
     type: boolean
     default: false
+    label:
+      zh_Hans: 联网搜索
+      en_US: Web Search
     help:
       zh_Hans: 模型内置了互联网搜索服务，该参数控制模型在生成文本时是否参考使用互联网搜索结果。启用互联网搜索，模型会将搜索结果作为文本生成过程中的参考信息，但模型会基于其内部逻辑“自行判断”是否使用互联网搜索结果。
       en_US: The model has a built-in Internet search service. This parameter controls whether the model refers to Internet search results when generating text. When Internet search is enabled, the model will use the search results as reference information in the text generation process, but the model will "judge" whether to use Internet search results based on its internal logic.
diff --git a/api/core/model_runtime/model_providers/tongyi/llm/qwen-vl-max-0809.yaml b/api/core/model_runtime/model_providers/tongyi/llm/qwen-vl-max-0809.yaml
index 41d45966e9d62..12e9e0dd569f9 100644
--- a/api/core/model_runtime/model_providers/tongyi/llm/qwen-vl-max-0809.yaml
+++ b/api/core/model_runtime/model_providers/tongyi/llm/qwen-vl-max-0809.yaml
@@ -9,6 +9,15 @@ model_properties:
   mode: chat
   context_size: 32000
 parameter_rules:
+  - name: temperature
+    use_template: temperature
+    type: float
+    default: 0.3
+    min: 0.0
+    max: 2.0
+    help:
+      zh_Hans: 用于控制随机性和多样性的程度。具体来说，temperature值控制了生成文本时对每个候选词的概率分布进行平滑的程度。较高的temperature值会降低概率分布的峰值，使得更多的低概率词被选择，生成结果更加多样化；而较低的temperature值则会增强概率分布的峰值，使得高概率词更容易被选择，生成结果更加确定。
+      en_US: Used to control the degree of randomness and diversity. Specifically, the temperature value controls the degree to which the probability distribution of each candidate word is smoothed when generating text. A higher temperature value will reduce the peak value of the probability distribution, allowing more low-probability words to be selected, and the generated results will be more diverse; while a lower temperature value will enhance the peak value of the probability distribution, making it easier for high-probability words to be selected. , the generated results are more certain.
   - name: top_p
     use_template: top_p
     type: float
@@ -50,6 +59,27 @@ parameter_rules:
       en_US: The random number seed used when generating, the user controls the randomness of the content generated by the model. Supports unsigned 64-bit integers, default value is 1234. When using seed, the model will try its best to generate the same or similar results, but there is currently no guarantee that the results will be exactly the same every time.
   - name: response_format
     use_template: response_format
+  - name: repetition_penalty
+    required: false
+    type: float
+    default: 1.1
+    label:
+      zh_Hans: 重复惩罚
+      en_US: Repetition penalty
+    help:
+      zh_Hans: 用于控制模型生成时的重复度。提高repetition_penalty时可以降低模型生成的重复度。1.0表示不做惩罚。
+      en_US: Used to control the repeatability when generating models. Increasing repetition_penalty can reduce the duplication of model generation. 1.0 means no punishment.
+  - name: enable_search
+    type: boolean
+    default: false
+    label:
+      zh_Hans: 联网搜索
+      en_US: Web Search
+    help:
+      zh_Hans: 模型内置了互联网搜索服务，该参数控制模型在生成文本时是否参考使用互联网搜索结果。启用互联网搜索，模型会将搜索结果作为文本生成过程中的参考信息，但模型会基于其内部逻辑“自行判断”是否使用互联网搜索结果。
+      en_US: The model has a built-in Internet search service. This parameter controls whether the model refers to Internet search results when generating text. When Internet search is enabled, the model will use the search results as reference information in the text generation process, but the model will "judge" whether to use Internet search results based on its internal logic.
+  - name: response_format
+    use_template: response_format
 pricing:
   input: '0.02'
   output: '0.02'
diff --git a/api/core/model_runtime/model_providers/tongyi/llm/qwen-vl-max.yaml b/api/core/model_runtime/model_providers/tongyi/llm/qwen-vl-max.yaml
index 78d0509374131..b811fdece48f5 100644
--- a/api/core/model_runtime/model_providers/tongyi/llm/qwen-vl-max.yaml
+++ b/api/core/model_runtime/model_providers/tongyi/llm/qwen-vl-max.yaml
@@ -9,6 +9,15 @@ model_properties:
   mode: chat
   context_size: 32000
 parameter_rules:
+  - name: temperature
+    use_template: temperature
+    type: float
+    default: 0.3
+    min: 0.0
+    max: 2.0
+    help:
+      zh_Hans: 用于控制随机性和多样性的程度。具体来说，temperature值控制了生成文本时对每个候选词的概率分布进行平滑的程度。较高的temperature值会降低概率分布的峰值，使得更多的低概率词被选择，生成结果更加多样化；而较低的temperature值则会增强概率分布的峰值，使得高概率词更容易被选择，生成结果更加确定。
+      en_US: Used to control the degree of randomness and diversity. Specifically, the temperature value controls the degree to which the probability distribution of each candidate word is smoothed when generating text. A higher temperature value will reduce the peak value of the probability distribution, allowing more low-probability words to be selected, and the generated results will be more diverse; while a lower temperature value will enhance the peak value of the probability distribution, making it easier for high-probability words to be selected. , the generated results are more certain.
   - name: top_p
     use_template: top_p
     type: float
@@ -50,6 +59,27 @@ parameter_rules:
       en_US: The random number seed used when generating, the user controls the randomness of the content generated by the model. Supports unsigned 64-bit integers, default value is 1234. When using seed, the model will try its best to generate the same or similar results, but there is currently no guarantee that the results will be exactly the same every time.
   - name: response_format
     use_template: response_format
+  - name: repetition_penalty
+    required: false
+    type: float
+    default: 1.1
+    label:
+      zh_Hans: 重复惩罚
+      en_US: Repetition penalty
+    help:
+      zh_Hans: 用于控制模型生成时的重复度。提高repetition_penalty时可以降低模型生成的重复度。1.0表示不做惩罚。
+      en_US: Used to control the repeatability when generating models. Increasing repetition_penalty can reduce the duplication of model generation. 1.0 means no punishment.
+  - name: enable_search
+    type: boolean
+    default: false
+    label:
+      zh_Hans: 联网搜索
+      en_US: Web Search
+    help:
+      zh_Hans: 模型内置了互联网搜索服务，该参数控制模型在生成文本时是否参考使用互联网搜索结果。启用互联网搜索，模型会将搜索结果作为文本生成过程中的参考信息，但模型会基于其内部逻辑“自行判断”是否使用互联网搜索结果。
+      en_US: The model has a built-in Internet search service. This parameter controls whether the model refers to Internet search results when generating text. When Internet search is enabled, the model will use the search results as reference information in the text generation process, but the model will "judge" whether to use Internet search results based on its internal logic.
+  - name: response_format
+    use_template: response_format
 pricing:
   input: '0.02'
   output: '0.02'
diff --git a/api/core/model_runtime/model_providers/tongyi/llm/qwen-vl-plus-0201.yaml b/api/core/model_runtime/model_providers/tongyi/llm/qwen-vl-plus-0201.yaml
index 8944388b1ee5a..188dea389a7e2 100644
--- a/api/core/model_runtime/model_providers/tongyi/llm/qwen-vl-plus-0201.yaml
+++ b/api/core/model_runtime/model_providers/tongyi/llm/qwen-vl-plus-0201.yaml
@@ -9,6 +9,15 @@ model_properties:
   mode: chat
   context_size: 8000
 parameter_rules:
+  - name: temperature
+    use_template: temperature
+    type: float
+    default: 0.3
+    min: 0.0
+    max: 2.0
+    help:
+      zh_Hans: 用于控制随机性和多样性的程度。具体来说，temperature值控制了生成文本时对每个候选词的概率分布进行平滑的程度。较高的temperature值会降低概率分布的峰值，使得更多的低概率词被选择，生成结果更加多样化；而较低的temperature值则会增强概率分布的峰值，使得高概率词更容易被选择，生成结果更加确定。
+      en_US: Used to control the degree of randomness and diversity. Specifically, the temperature value controls the degree to which the probability distribution of each candidate word is smoothed when generating text. A higher temperature value will reduce the peak value of the probability distribution, allowing more low-probability words to be selected, and the generated results will be more diverse; while a lower temperature value will enhance the peak value of the probability distribution, making it easier for high-probability words to be selected. , the generated results are more certain.
   - name: top_p
     use_template: top_p
     type: float
@@ -50,6 +59,27 @@ parameter_rules:
       en_US: The random number seed used when generating, the user controls the randomness of the content generated by the model. Supports unsigned 64-bit integers, default value is 1234. When using seed, the model will try its best to generate the same or similar results, but there is currently no guarantee that the results will be exactly the same every time.
   - name: response_format
     use_template: response_format
+  - name: repetition_penalty
+    required: false
+    type: float
+    default: 1.1
+    label:
+      zh_Hans: 重复惩罚
+      en_US: Repetition penalty
+    help:
+      zh_Hans: 用于控制模型生成时的重复度。提高repetition_penalty时可以降低模型生成的重复度。1.0表示不做惩罚。
+      en_US: Used to control the repeatability when generating models. Increasing repetition_penalty can reduce the duplication of model generation. 1.0 means no punishment.
+  - name: enable_search
+    type: boolean
+    default: false
+    label:
+      zh_Hans: 联网搜索
+      en_US: Web Search
+    help:
+      zh_Hans: 模型内置了互联网搜索服务，该参数控制模型在生成文本时是否参考使用互联网搜索结果。启用互联网搜索，模型会将搜索结果作为文本生成过程中的参考信息，但模型会基于其内部逻辑“自行判断”是否使用互联网搜索结果。
+      en_US: The model has a built-in Internet search service. This parameter controls whether the model refers to Internet search results when generating text. When Internet search is enabled, the model will use the search results as reference information in the text generation process, but the model will "judge" whether to use Internet search results based on its internal logic.
+  - name: response_format
+    use_template: response_format
 pricing:
   input: '0.02'
   output: '0.02'
diff --git a/api/core/model_runtime/model_providers/tongyi/llm/qwen-vl-plus-0809.yaml b/api/core/model_runtime/model_providers/tongyi/llm/qwen-vl-plus-0809.yaml
index 869e0ea71c01b..bc623e2f03211 100644
--- a/api/core/model_runtime/model_providers/tongyi/llm/qwen-vl-plus-0809.yaml
+++ b/api/core/model_runtime/model_providers/tongyi/llm/qwen-vl-plus-0809.yaml
@@ -9,6 +9,15 @@ model_properties:
   mode: chat
   context_size: 32768
 parameter_rules:
+  - name: temperature
+    use_template: temperature
+    type: float
+    default: 0.3
+    min: 0.0
+    max: 2.0
+    help:
+      zh_Hans: 用于控制随机性和多样性的程度。具体来说，temperature值控制了生成文本时对每个候选词的概率分布进行平滑的程度。较高的temperature值会降低概率分布的峰值，使得更多的低概率词被选择，生成结果更加多样化；而较低的temperature值则会增强概率分布的峰值，使得高概率词更容易被选择，生成结果更加确定。
+      en_US: Used to control the degree of randomness and diversity. Specifically, the temperature value controls the degree to which the probability distribution of each candidate word is smoothed when generating text. A higher temperature value will reduce the peak value of the probability distribution, allowing more low-probability words to be selected, and the generated results will be more diverse; while a lower temperature value will enhance the peak value of the probability distribution, making it easier for high-probability words to be selected. , the generated results are more certain.
   - name: top_p
     use_template: top_p
     type: float
@@ -50,6 +59,27 @@ parameter_rules:
       en_US: The random number seed used when generating, the user controls the randomness of the content generated by the model. Supports unsigned 64-bit integers, default value is 1234. When using seed, the model will try its best to generate the same or similar results, but there is currently no guarantee that the results will be exactly the same every time.
   - name: response_format
     use_template: response_format
+  - name: repetition_penalty
+    required: false
+    type: float
+    default: 1.1
+    label:
+      zh_Hans: 重复惩罚
+      en_US: Repetition penalty
+    help:
+      zh_Hans: 用于控制模型生成时的重复度。提高repetition_penalty时可以降低模型生成的重复度。1.0表示不做惩罚。
+      en_US: Used to control the repeatability when generating models. Increasing repetition_penalty can reduce the duplication of model generation. 1.0 means no punishment.
+  - name: enable_search
+    type: boolean
+    default: false
+    label:
+      zh_Hans: 联网搜索
+      en_US: Web Search
+    help:
+      zh_Hans: 模型内置了互联网搜索服务，该参数控制模型在生成文本时是否参考使用互联网搜索结果。启用互联网搜索，模型会将搜索结果作为文本生成过程中的参考信息，但模型会基于其内部逻辑“自行判断”是否使用互联网搜索结果。
+      en_US: The model has a built-in Internet search service. This parameter controls whether the model refers to Internet search results when generating text. When Internet search is enabled, the model will use the search results as reference information in the text generation process, but the model will "judge" whether to use Internet search results based on its internal logic.
+  - name: response_format
+    use_template: response_format
 pricing:
   input: '0.008'
   output: '0.008'
diff --git a/api/core/model_runtime/model_providers/tongyi/llm/qwen-vl-plus.yaml b/api/core/model_runtime/model_providers/tongyi/llm/qwen-vl-plus.yaml
index da11bacc646c8..8977e12e4f2d9 100644
--- a/api/core/model_runtime/model_providers/tongyi/llm/qwen-vl-plus.yaml
+++ b/api/core/model_runtime/model_providers/tongyi/llm/qwen-vl-plus.yaml
@@ -9,6 +9,15 @@ model_properties:
   mode: chat
   context_size: 8000
 parameter_rules:
+  - name: temperature
+    use_template: temperature
+    type: float
+    default: 0.3
+    min: 0.0
+    max: 2.0
+    help:
+      zh_Hans: 用于控制随机性和多样性的程度。具体来说，temperature值控制了生成文本时对每个候选词的概率分布进行平滑的程度。较高的temperature值会降低概率分布的峰值，使得更多的低概率词被选择，生成结果更加多样化；而较低的temperature值则会增强概率分布的峰值，使得高概率词更容易被选择，生成结果更加确定。
+      en_US: Used to control the degree of randomness and diversity. Specifically, the temperature value controls the degree to which the probability distribution of each candidate word is smoothed when generating text. A higher temperature value will reduce the peak value of the probability distribution, allowing more low-probability words to be selected, and the generated results will be more diverse; while a lower temperature value will enhance the peak value of the probability distribution, making it easier for high-probability words to be selected. , the generated results are more certain.
   - name: top_p
     use_template: top_p
     type: float
@@ -50,6 +59,27 @@ parameter_rules:
       en_US: The random number seed used when generating, the user controls the randomness of the content generated by the model. Supports unsigned 64-bit integers, default value is 1234. When using seed, the model will try its best to generate the same or similar results, but there is currently no guarantee that the results will be exactly the same every time.
   - name: response_format
     use_template: response_format
+  - name: repetition_penalty
+    required: false
+    type: float
+    default: 1.1
+    label:
+      zh_Hans: 重复惩罚
+      en_US: Repetition penalty
+    help:
+      zh_Hans: 用于控制模型生成时的重复度。提高repetition_penalty时可以降低模型生成的重复度。1.0表示不做惩罚。
+      en_US: Used to control the repeatability when generating models. Increasing repetition_penalty can reduce the duplication of model generation. 1.0 means no punishment.
+  - name: enable_search
+    type: boolean
+    default: false
+    label:
+      zh_Hans: 联网搜索
+      en_US: Web Search
+    help:
+      zh_Hans: 模型内置了互联网搜索服务，该参数控制模型在生成文本时是否参考使用互联网搜索结果。启用互联网搜索，模型会将搜索结果作为文本生成过程中的参考信息，但模型会基于其内部逻辑“自行判断”是否使用互联网搜索结果。
+      en_US: The model has a built-in Internet search service. This parameter controls whether the model refers to Internet search results when generating text. When Internet search is enabled, the model will use the search results as reference information in the text generation process, but the model will "judge" whether to use Internet search results based on its internal logic.
+  - name: response_format
+    use_template: response_format
 pricing:
   input: '0.008'
   output: '0.008'
diff --git a/api/core/model_runtime/model_providers/tongyi/llm/qwen2-math-1.5b-instruct.yaml b/api/core/model_runtime/model_providers/tongyi/llm/qwen2-math-1.5b-instruct.yaml
index cfe4b5a6662a3..de237842af7d8 100644
--- a/api/core/model_runtime/model_providers/tongyi/llm/qwen2-math-1.5b-instruct.yaml
+++ b/api/core/model_runtime/model_providers/tongyi/llm/qwen2-math-1.5b-instruct.yaml
@@ -60,6 +60,7 @@ parameter_rules:
     type: float
     default: 1.1
     label:
+      zh_Hans: 重复惩罚
       en_US: Repetition penalty
     help:
       zh_Hans: 用于控制模型生成时的重复度。提高repetition_penalty时可以降低模型生成的重复度。1.0表示不做惩罚。
@@ -67,6 +68,9 @@ parameter_rules:
   - name: enable_search
     type: boolean
     default: false
+    label:
+      zh_Hans: 联网搜索
+      en_US: Web Search
     help:
       zh_Hans: 模型内置了互联网搜索服务，该参数控制模型在生成文本时是否参考使用互联网搜索结果。启用互联网搜索，模型会将搜索结果作为文本生成过程中的参考信息，但模型会基于其内部逻辑“自行判断”是否使用互联网搜索结果。
       en_US: The model has a built-in Internet search service. This parameter controls whether the model refers to Internet search results when generating text. When Internet search is enabled, the model will use the search results as reference information in the text generation process, but the model will "judge" whether to use Internet search results based on its internal logic.
diff --git a/api/core/model_runtime/model_providers/tongyi/llm/qwen2-math-72b-instruct.yaml b/api/core/model_runtime/model_providers/tongyi/llm/qwen2-math-72b-instruct.yaml
index e541c197b0ff4..1fda35abaf1fe 100644
--- a/api/core/model_runtime/model_providers/tongyi/llm/qwen2-math-72b-instruct.yaml
+++ b/api/core/model_runtime/model_providers/tongyi/llm/qwen2-math-72b-instruct.yaml
@@ -60,6 +60,7 @@ parameter_rules:
     type: float
     default: 1.1
     label:
+      zh_Hans: 重复惩罚
       en_US: Repetition penalty
     help:
       zh_Hans: 用于控制模型生成时的重复度。提高repetition_penalty时可以降低模型生成的重复度。1.0表示不做惩罚。
@@ -67,6 +68,9 @@ parameter_rules:
   - name: enable_search
     type: boolean
     default: false
+    label:
+      zh_Hans: 联网搜索
+      en_US: Web Search
     help:
       zh_Hans: 模型内置了互联网搜索服务，该参数控制模型在生成文本时是否参考使用互联网搜索结果。启用互联网搜索，模型会将搜索结果作为文本生成过程中的参考信息，但模型会基于其内部逻辑“自行判断”是否使用互联网搜索结果。
       en_US: The model has a built-in Internet search service. This parameter controls whether the model refers to Internet search results when generating text. When Internet search is enabled, the model will use the search results as reference information in the text generation process, but the model will "judge" whether to use Internet search results based on its internal logic.
diff --git a/api/core/model_runtime/model_providers/tongyi/llm/qwen2-math-7b-instruct.yaml b/api/core/model_runtime/model_providers/tongyi/llm/qwen2-math-7b-instruct.yaml
index ba4514e3d6ead..06fd33c5f4740 100644
--- a/api/core/model_runtime/model_providers/tongyi/llm/qwen2-math-7b-instruct.yaml
+++ b/api/core/model_runtime/model_providers/tongyi/llm/qwen2-math-7b-instruct.yaml
@@ -60,6 +60,7 @@ parameter_rules:
     type: float
     default: 1.1
     label:
+      zh_Hans: 重复惩罚
       en_US: Repetition penalty
     help:
       zh_Hans: 用于控制模型生成时的重复度。提高repetition_penalty时可以降低模型生成的重复度。1.0表示不做惩罚。
@@ -67,6 +68,9 @@ parameter_rules:
   - name: enable_search
     type: boolean
     default: false
+    label:
+      zh_Hans: 联网搜索
+      en_US: Web Search
     help:
       zh_Hans: 模型内置了互联网搜索服务，该参数控制模型在生成文本时是否参考使用互联网搜索结果。启用互联网搜索，模型会将搜索结果作为文本生成过程中的参考信息，但模型会基于其内部逻辑“自行判断”是否使用互联网搜索结果。
       en_US: The model has a built-in Internet search service. This parameter controls whether the model refers to Internet search results when generating text. When Internet search is enabled, the model will use the search results as reference information in the text generation process, but the model will "judge" whether to use Internet search results based on its internal logic.
diff --git a/api/core/model_runtime/model_providers/tongyi/llm/qwen2.5-0.5b-instruct.yaml b/api/core/model_runtime/model_providers/tongyi/llm/qwen2.5-0.5b-instruct.yaml
index e5596041af6f2..ebf809955337b 100644
--- a/api/core/model_runtime/model_providers/tongyi/llm/qwen2.5-0.5b-instruct.yaml
+++ b/api/core/model_runtime/model_providers/tongyi/llm/qwen2.5-0.5b-instruct.yaml
@@ -60,6 +60,7 @@ parameter_rules:
     type: float
     default: 1.1
     label:
+      zh_Hans: 重复惩罚
       en_US: Repetition penalty
     help:
       zh_Hans: 用于控制模型生成时的重复度。提高repetition_penalty时可以降低模型生成的重复度。1.0表示不做惩罚。
@@ -67,6 +68,9 @@ parameter_rules:
   - name: enable_search
     type: boolean
     default: false
+    label:
+      zh_Hans: 联网搜索
+      en_US: Web Search
     help:
       zh_Hans: 模型内置了互联网搜索服务，该参数控制模型在生成文本时是否参考使用互联网搜索结果。启用互联网搜索，模型会将搜索结果作为文本生成过程中的参考信息，但模型会基于其内部逻辑“自行判断”是否使用互联网搜索结果。
       en_US: The model has a built-in Internet search service. This parameter controls whether the model refers to Internet search results when generating text. When Internet search is enabled, the model will use the search results as reference information in the text generation process, but the model will "judge" whether to use Internet search results based on its internal logic.
diff --git a/api/core/model_runtime/model_providers/tongyi/llm/qwen2.5-1.5b-instruct.yaml b/api/core/model_runtime/model_providers/tongyi/llm/qwen2.5-1.5b-instruct.yaml
index 4004c59417107..e9bc99339de0b 100644
--- a/api/core/model_runtime/model_providers/tongyi/llm/qwen2.5-1.5b-instruct.yaml
+++ b/api/core/model_runtime/model_providers/tongyi/llm/qwen2.5-1.5b-instruct.yaml
@@ -60,6 +60,7 @@ parameter_rules:
     type: float
     default: 1.1
     label:
+      zh_Hans: 重复惩罚
       en_US: Repetition penalty
     help:
       zh_Hans: 用于控制模型生成时的重复度。提高repetition_penalty时可以降低模型生成的重复度。1.0表示不做惩罚。
@@ -67,6 +68,9 @@ parameter_rules:
   - name: enable_search
     type: boolean
     default: false
+    label:
+      zh_Hans: 联网搜索
+      en_US: Web Search
     help:
       zh_Hans: 模型内置了互联网搜索服务，该参数控制模型在生成文本时是否参考使用互联网搜索结果。启用互联网搜索，模型会将搜索结果作为文本生成过程中的参考信息，但模型会基于其内部逻辑“自行判断”是否使用互联网搜索结果。
       en_US: The model has a built-in Internet search service. This parameter controls whether the model refers to Internet search results when generating text. When Internet search is enabled, the model will use the search results as reference information in the text generation process, but the model will "judge" whether to use Internet search results based on its internal logic.
diff --git a/api/core/model_runtime/model_providers/tongyi/llm/qwen2.5-14b-instruct.yaml b/api/core/model_runtime/model_providers/tongyi/llm/qwen2.5-14b-instruct.yaml
index d8f53666ced41..3ed85dade8281 100644
--- a/api/core/model_runtime/model_providers/tongyi/llm/qwen2.5-14b-instruct.yaml
+++ b/api/core/model_runtime/model_providers/tongyi/llm/qwen2.5-14b-instruct.yaml
@@ -60,6 +60,7 @@ parameter_rules:
     type: float
     default: 1.1
     label:
+      zh_Hans: 重复惩罚
       en_US: Repetition penalty
     help:
       zh_Hans: 用于控制模型生成时的重复度。提高repetition_penalty时可以降低模型生成的重复度。1.0表示不做惩罚。
@@ -67,6 +68,9 @@ parameter_rules:
   - name: enable_search
     type: boolean
     default: false
+    label:
+      zh_Hans: 联网搜索
+      en_US: Web Search
     help:
       zh_Hans: 模型内置了互联网搜索服务，该参数控制模型在生成文本时是否参考使用互联网搜索结果。启用互联网搜索，模型会将搜索结果作为文本生成过程中的参考信息，但模型会基于其内部逻辑“自行判断”是否使用互联网搜索结果。
       en_US: The model has a built-in Internet search service. This parameter controls whether the model refers to Internet search results when generating text. When Internet search is enabled, the model will use the search results as reference information in the text generation process, but the model will "judge" whether to use Internet search results based on its internal logic.
diff --git a/api/core/model_runtime/model_providers/tongyi/llm/qwen2.5-32b-instruct.yaml b/api/core/model_runtime/model_providers/tongyi/llm/qwen2.5-32b-instruct.yaml
index 890f7e6e4e5d9..328519c16822f 100644
--- a/api/core/model_runtime/model_providers/tongyi/llm/qwen2.5-32b-instruct.yaml
+++ b/api/core/model_runtime/model_providers/tongyi/llm/qwen2.5-32b-instruct.yaml
@@ -60,6 +60,7 @@ parameter_rules:
     type: float
     default: 1.1
     label:
+      zh_Hans: 重复惩罚
       en_US: Repetition penalty
     help:
       zh_Hans: 用于控制模型生成时的重复度。提高repetition_penalty时可以降低模型生成的重复度。1.0表示不做惩罚。
@@ -67,6 +68,9 @@ parameter_rules:
   - name: enable_search
     type: boolean
     default: false
+    label:
+      zh_Hans: 联网搜索
+      en_US: Web Search
     help:
       zh_Hans: 模型内置了互联网搜索服务，该参数控制模型在生成文本时是否参考使用互联网搜索结果。启用互联网搜索，模型会将搜索结果作为文本生成过程中的参考信息，但模型会基于其内部逻辑“自行判断”是否使用互联网搜索结果。
       en_US: The model has a built-in Internet search service. This parameter controls whether the model refers to Internet search results when generating text. When Internet search is enabled, the model will use the search results as reference information in the text generation process, but the model will "judge" whether to use Internet search results based on its internal logic.
diff --git a/api/core/model_runtime/model_providers/tongyi/llm/qwen2.5-3b-instruct.yaml b/api/core/model_runtime/model_providers/tongyi/llm/qwen2.5-3b-instruct.yaml
index 6d3d2dd5bb5e6..d1ed3c2a73edf 100644
--- a/api/core/model_runtime/model_providers/tongyi/llm/qwen2.5-3b-instruct.yaml
+++ b/api/core/model_runtime/model_providers/tongyi/llm/qwen2.5-3b-instruct.yaml
@@ -60,6 +60,7 @@ parameter_rules:
     type: float
     default: 1.1
     label:
+      zh_Hans: 重复惩罚
       en_US: Repetition penalty
     help:
       zh_Hans: 用于控制模型生成时的重复度。提高repetition_penalty时可以降低模型生成的重复度。1.0表示不做惩罚。
@@ -67,6 +68,9 @@ parameter_rules:
   - name: enable_search
     type: boolean
     default: false
+    label:
+      zh_Hans: 联网搜索
+      en_US: Web Search
     help:
       zh_Hans: 模型内置了互联网搜索服务，该参数控制模型在生成文本时是否参考使用互联网搜索结果。启用互联网搜索，模型会将搜索结果作为文本生成过程中的参考信息，但模型会基于其内部逻辑“自行判断”是否使用互联网搜索结果。
       en_US: The model has a built-in Internet search service. This parameter controls whether the model refers to Internet search results when generating text. When Internet search is enabled, the model will use the search results as reference information in the text generation process, but the model will "judge" whether to use Internet search results based on its internal logic.
diff --git a/api/core/model_runtime/model_providers/tongyi/llm/qwen2.5-72b-instruct.yaml b/api/core/model_runtime/model_providers/tongyi/llm/qwen2.5-72b-instruct.yaml
index 17d0eb5b351d8..0e88c24aa8347 100644
--- a/api/core/model_runtime/model_providers/tongyi/llm/qwen2.5-72b-instruct.yaml
+++ b/api/core/model_runtime/model_providers/tongyi/llm/qwen2.5-72b-instruct.yaml
@@ -60,6 +60,7 @@ parameter_rules:
     type: float
     default: 1.1
     label:
+      zh_Hans: 重复惩罚
       en_US: Repetition penalty
     help:
       zh_Hans: 用于控制模型生成时的重复度。提高repetition_penalty时可以降低模型生成的重复度。1.0表示不做惩罚。
@@ -67,6 +68,9 @@ parameter_rules:
   - name: enable_search
     type: boolean
     default: false
+    label:
+      zh_Hans: 联网搜索
+      en_US: Web Search
     help:
       zh_Hans: 模型内置了互联网搜索服务，该参数控制模型在生成文本时是否参考使用互联网搜索结果。启用互联网搜索，模型会将搜索结果作为文本生成过程中的参考信息，但模型会基于其内部逻辑“自行判断”是否使用互联网搜索结果。
       en_US: The model has a built-in Internet search service. This parameter controls whether the model refers to Internet search results when generating text. When Internet search is enabled, the model will use the search results as reference information in the text generation process, but the model will "judge" whether to use Internet search results based on its internal logic.
diff --git a/api/core/model_runtime/model_providers/tongyi/llm/qwen2.5-7b-instruct.yaml b/api/core/model_runtime/model_providers/tongyi/llm/qwen2.5-7b-instruct.yaml
index 435b3f90a2be5..35313cd1f7094 100644
--- a/api/core/model_runtime/model_providers/tongyi/llm/qwen2.5-7b-instruct.yaml
+++ b/api/core/model_runtime/model_providers/tongyi/llm/qwen2.5-7b-instruct.yaml
@@ -60,6 +60,7 @@ parameter_rules:
     type: float
     default: 1.1
     label:
+      zh_Hans: 重复惩罚
       en_US: Repetition penalty
     help:
       zh_Hans: 用于控制模型生成时的重复度。提高repetition_penalty时可以降低模型生成的重复度。1.0表示不做惩罚。
@@ -67,6 +68,9 @@ parameter_rules:
   - name: enable_search
     type: boolean
     default: false
+    label:
+      zh_Hans: 联网搜索
+      en_US: Web Search
     help:
       zh_Hans: 模型内置了互联网搜索服务，该参数控制模型在生成文本时是否参考使用互联网搜索结果。启用互联网搜索，模型会将搜索结果作为文本生成过程中的参考信息，但模型会基于其内部逻辑“自行判断”是否使用互联网搜索结果。
       en_US: The model has a built-in Internet search service. This parameter controls whether the model refers to Internet search results when generating text. When Internet search is enabled, the model will use the search results as reference information in the text generation process, but the model will "judge" whether to use Internet search results based on its internal logic.
diff --git a/api/core/model_runtime/model_providers/tongyi/llm/qwen2.5-coder-7b-instruct.yaml b/api/core/model_runtime/model_providers/tongyi/llm/qwen2.5-coder-7b-instruct.yaml
index 435b3f90a2be5..35313cd1f7094 100644
--- a/api/core/model_runtime/model_providers/tongyi/llm/qwen2.5-coder-7b-instruct.yaml
+++ b/api/core/model_runtime/model_providers/tongyi/llm/qwen2.5-coder-7b-instruct.yaml
@@ -60,6 +60,7 @@ parameter_rules:
     type: float
     default: 1.1
     label:
+      zh_Hans: 重复惩罚
       en_US: Repetition penalty
     help:
       zh_Hans: 用于控制模型生成时的重复度。提高repetition_penalty时可以降低模型生成的重复度。1.0表示不做惩罚。
@@ -67,6 +68,9 @@ parameter_rules:
   - name: enable_search
     type: boolean
     default: false
+    label:
+      zh_Hans: 联网搜索
+      en_US: Web Search
     help:
       zh_Hans: 模型内置了互联网搜索服务，该参数控制模型在生成文本时是否参考使用互联网搜索结果。启用互联网搜索，模型会将搜索结果作为文本生成过程中的参考信息，但模型会基于其内部逻辑“自行判断”是否使用互联网搜索结果。
       en_US: The model has a built-in Internet search service. This parameter controls whether the model refers to Internet search results when generating text. When Internet search is enabled, the model will use the search results as reference information in the text generation process, but the model will "judge" whether to use Internet search results based on its internal logic.

From 3554a803e70152e1e5064e2163d5dc13d389264d Mon Sep 17 00:00:00 2001
From: AAEE86 <33052466+AAEE86@users.noreply.github.com>
Date: Mon, 23 Sep 2024 16:19:42 +0800
Subject: [PATCH 036/235] add zhipuai web search (#8668)

---
 .../model_providers/zhipuai/llm/glm-4-0520.yaml          | 9 +++++++++
 .../model_providers/zhipuai/llm/glm-4-air.yaml           | 9 +++++++++
 .../model_providers/zhipuai/llm/glm-4-airx.yaml          | 9 +++++++++
 .../model_providers/zhipuai/llm/glm-4-flash.yaml         | 9 +++++++++
 .../model_providers/zhipuai/llm/glm_3_turbo.yaml         | 9 +++++++++
 .../model_runtime/model_providers/zhipuai/llm/glm_4.yaml | 9 +++++++++
 .../model_providers/zhipuai/llm/glm_4_long.yaml          | 9 +++++++++
 .../model_providers/zhipuai/llm/glm_4_plus.yaml          | 9 +++++++++
 .../model_providers/zhipuai/llm/glm_4v.yaml              | 9 +++++++++
 .../model_providers/zhipuai/llm/glm_4v_plus.yaml         | 9 +++++++++
 10 files changed, 90 insertions(+)

diff --git a/api/core/model_runtime/model_providers/zhipuai/llm/glm-4-0520.yaml b/api/core/model_runtime/model_providers/zhipuai/llm/glm-4-0520.yaml
index b1f9b7485cd9d..7fcf6922023fb 100644
--- a/api/core/model_runtime/model_providers/zhipuai/llm/glm-4-0520.yaml
+++ b/api/core/model_runtime/model_providers/zhipuai/llm/glm-4-0520.yaml
@@ -46,6 +46,15 @@ parameter_rules:
     default: 1024
     min: 1
     max: 4095
+  - name: web_search
+    type: boolean
+    label:
+      zh_Hans: 联网搜索
+      en_US: Web Search
+    default: false
+    help:
+      zh_Hans: 模型内置了互联网搜索服务，该参数控制模型在生成文本时是否参考使用互联网搜索结果。启用互联网搜索，模型会将搜索结果作为文本生成过程中的参考信息，但模型会基于其内部逻辑“自行判断”是否使用互联网搜索结果。
+      en_US: The model has a built-in Internet search service. This parameter controls whether the model refers to Internet search results when generating text. When Internet search is enabled, the model will use the search results as reference information in the text generation process, but the model will "judge" whether to use Internet search results based on its internal logic.
 pricing:
   input: '0.1'
   output: '0.1'
diff --git a/api/core/model_runtime/model_providers/zhipuai/llm/glm-4-air.yaml b/api/core/model_runtime/model_providers/zhipuai/llm/glm-4-air.yaml
index 4e7d5fd3cc977..fcd7c7768c717 100644
--- a/api/core/model_runtime/model_providers/zhipuai/llm/glm-4-air.yaml
+++ b/api/core/model_runtime/model_providers/zhipuai/llm/glm-4-air.yaml
@@ -46,6 +46,15 @@ parameter_rules:
     default: 1024
     min: 1
     max: 4095
+  - name: web_search
+    type: boolean
+    label:
+      zh_Hans: 联网搜索
+      en_US: Web Search
+    default: false
+    help:
+      zh_Hans: 模型内置了互联网搜索服务，该参数控制模型在生成文本时是否参考使用互联网搜索结果。启用互联网搜索，模型会将搜索结果作为文本生成过程中的参考信息，但模型会基于其内部逻辑“自行判断”是否使用互联网搜索结果。
+      en_US: The model has a built-in Internet search service. This parameter controls whether the model refers to Internet search results when generating text. When Internet search is enabled, the model will use the search results as reference information in the text generation process, but the model will "judge" whether to use Internet search results based on its internal logic.
 pricing:
   input: '0.001'
   output: '0.001'
diff --git a/api/core/model_runtime/model_providers/zhipuai/llm/glm-4-airx.yaml b/api/core/model_runtime/model_providers/zhipuai/llm/glm-4-airx.yaml
index 14f17db5d6c06..c9ae5abf19636 100644
--- a/api/core/model_runtime/model_providers/zhipuai/llm/glm-4-airx.yaml
+++ b/api/core/model_runtime/model_providers/zhipuai/llm/glm-4-airx.yaml
@@ -46,6 +46,15 @@ parameter_rules:
     default: 1024
     min: 1
     max: 4095
+  - name: web_search
+    type: boolean
+    label:
+      zh_Hans: 联网搜索
+      en_US: Web Search
+    default: false
+    help:
+      zh_Hans: 模型内置了互联网搜索服务，该参数控制模型在生成文本时是否参考使用互联网搜索结果。启用互联网搜索，模型会将搜索结果作为文本生成过程中的参考信息，但模型会基于其内部逻辑“自行判断”是否使用互联网搜索结果。
+      en_US: The model has a built-in Internet search service. This parameter controls whether the model refers to Internet search results when generating text. When Internet search is enabled, the model will use the search results as reference information in the text generation process, but the model will "judge" whether to use Internet search results based on its internal logic.
 pricing:
   input: '0.01'
   output: '0.01'
diff --git a/api/core/model_runtime/model_providers/zhipuai/llm/glm-4-flash.yaml b/api/core/model_runtime/model_providers/zhipuai/llm/glm-4-flash.yaml
index 3361474d737df..98c4f72c7237f 100644
--- a/api/core/model_runtime/model_providers/zhipuai/llm/glm-4-flash.yaml
+++ b/api/core/model_runtime/model_providers/zhipuai/llm/glm-4-flash.yaml
@@ -46,6 +46,15 @@ parameter_rules:
     default: 1024
     min: 1
     max: 4095
+  - name: web_search
+    type: boolean
+    label:
+      zh_Hans: 联网搜索
+      en_US: Web Search
+    default: false
+    help:
+      zh_Hans: 模型内置了互联网搜索服务，该参数控制模型在生成文本时是否参考使用互联网搜索结果。启用互联网搜索，模型会将搜索结果作为文本生成过程中的参考信息，但模型会基于其内部逻辑“自行判断”是否使用互联网搜索结果。
+      en_US: The model has a built-in Internet search service. This parameter controls whether the model refers to Internet search results when generating text. When Internet search is enabled, the model will use the search results as reference information in the text generation process, but the model will "judge" whether to use Internet search results based on its internal logic.
 pricing:
   input: '0'
   output: '0'
diff --git a/api/core/model_runtime/model_providers/zhipuai/llm/glm_3_turbo.yaml b/api/core/model_runtime/model_providers/zhipuai/llm/glm_3_turbo.yaml
index bf0135d198548..0b5391ce2f83f 100644
--- a/api/core/model_runtime/model_providers/zhipuai/llm/glm_3_turbo.yaml
+++ b/api/core/model_runtime/model_providers/zhipuai/llm/glm_3_turbo.yaml
@@ -46,6 +46,15 @@ parameter_rules:
     default: 1024
     min: 1
     max: 8192
+  - name: web_search
+    type: boolean
+    label:
+      zh_Hans: 联网搜索
+      en_US: Web Search
+    default: false
+    help:
+      zh_Hans: 模型内置了互联网搜索服务，该参数控制模型在生成文本时是否参考使用互联网搜索结果。启用互联网搜索，模型会将搜索结果作为文本生成过程中的参考信息，但模型会基于其内部逻辑“自行判断”是否使用互联网搜索结果。
+      en_US: The model has a built-in Internet search service. This parameter controls whether the model refers to Internet search results when generating text. When Internet search is enabled, the model will use the search results as reference information in the text generation process, but the model will "judge" whether to use Internet search results based on its internal logic.
 pricing:
   input: '0.001'
   output: '0.001'
diff --git a/api/core/model_runtime/model_providers/zhipuai/llm/glm_4.yaml b/api/core/model_runtime/model_providers/zhipuai/llm/glm_4.yaml
index ab4b32dd826b0..62f453fb775b9 100644
--- a/api/core/model_runtime/model_providers/zhipuai/llm/glm_4.yaml
+++ b/api/core/model_runtime/model_providers/zhipuai/llm/glm_4.yaml
@@ -46,6 +46,15 @@ parameter_rules:
     default: 1024
     min: 1
     max: 4095
+  - name: web_search
+    type: boolean
+    label:
+      zh_Hans: 联网搜索
+      en_US: Web Search
+    default: false
+    help:
+      zh_Hans: 模型内置了互联网搜索服务，该参数控制模型在生成文本时是否参考使用互联网搜索结果。启用互联网搜索，模型会将搜索结果作为文本生成过程中的参考信息，但模型会基于其内部逻辑“自行判断”是否使用互联网搜索结果。
+      en_US: The model has a built-in Internet search service. This parameter controls whether the model refers to Internet search results when generating text. When Internet search is enabled, the model will use the search results as reference information in the text generation process, but the model will "judge" whether to use Internet search results based on its internal logic.
 pricing:
   input: '0.1'
   output: '0.1'
diff --git a/api/core/model_runtime/model_providers/zhipuai/llm/glm_4_long.yaml b/api/core/model_runtime/model_providers/zhipuai/llm/glm_4_long.yaml
index d1b01731f5463..350b080c3fc11 100644
--- a/api/core/model_runtime/model_providers/zhipuai/llm/glm_4_long.yaml
+++ b/api/core/model_runtime/model_providers/zhipuai/llm/glm_4_long.yaml
@@ -49,6 +49,15 @@ parameter_rules:
     default: 1024
     min: 1
     max: 4095
+  - name: web_search
+    type: boolean
+    label:
+      zh_Hans: 联网搜索
+      en_US: Web Search
+    default: false
+    help:
+      zh_Hans: 模型内置了互联网搜索服务，该参数控制模型在生成文本时是否参考使用互联网搜索结果。启用互联网搜索，模型会将搜索结果作为文本生成过程中的参考信息，但模型会基于其内部逻辑“自行判断”是否使用互联网搜索结果。
+      en_US: The model has a built-in Internet search service. This parameter controls whether the model refers to Internet search results when generating text. When Internet search is enabled, the model will use the search results as reference information in the text generation process, but the model will "judge" whether to use Internet search results based on its internal logic.
 pricing:
   input: '0.001'
   output: '0.001'
diff --git a/api/core/model_runtime/model_providers/zhipuai/llm/glm_4_plus.yaml b/api/core/model_runtime/model_providers/zhipuai/llm/glm_4_plus.yaml
index 9ede308f18d55..2d7ebd71cf26e 100644
--- a/api/core/model_runtime/model_providers/zhipuai/llm/glm_4_plus.yaml
+++ b/api/core/model_runtime/model_providers/zhipuai/llm/glm_4_plus.yaml
@@ -46,6 +46,15 @@ parameter_rules:
     default: 1024
     min: 1
     max: 4095
+  - name: web_search
+    type: boolean
+    label:
+      zh_Hans: 联网搜索
+      en_US: Web Search
+    default: false
+    help:
+      zh_Hans: 模型内置了互联网搜索服务，该参数控制模型在生成文本时是否参考使用互联网搜索结果。启用互联网搜索，模型会将搜索结果作为文本生成过程中的参考信息，但模型会基于其内部逻辑“自行判断”是否使用互联网搜索结果。
+      en_US: The model has a built-in Internet search service. This parameter controls whether the model refers to Internet search results when generating text. When Internet search is enabled, the model will use the search results as reference information in the text generation process, but the model will "judge" whether to use Internet search results based on its internal logic.
 pricing:
   input: '0.05'
   output: '0.05'
diff --git a/api/core/model_runtime/model_providers/zhipuai/llm/glm_4v.yaml b/api/core/model_runtime/model_providers/zhipuai/llm/glm_4v.yaml
index 28286580a7ec1..3a1120ff375c1 100644
--- a/api/core/model_runtime/model_providers/zhipuai/llm/glm_4v.yaml
+++ b/api/core/model_runtime/model_providers/zhipuai/llm/glm_4v.yaml
@@ -44,6 +44,15 @@ parameter_rules:
     default: 1024
     min: 1
     max: 1024
+  - name: web_search
+    type: boolean
+    label:
+      zh_Hans: 联网搜索
+      en_US: Web Search
+    default: false
+    help:
+      zh_Hans: 模型内置了互联网搜索服务，该参数控制模型在生成文本时是否参考使用互联网搜索结果。启用互联网搜索，模型会将搜索结果作为文本生成过程中的参考信息，但模型会基于其内部逻辑“自行判断”是否使用互联网搜索结果。
+      en_US: The model has a built-in Internet search service. This parameter controls whether the model refers to Internet search results when generating text. When Internet search is enabled, the model will use the search results as reference information in the text generation process, but the model will "judge" whether to use Internet search results based on its internal logic.
 pricing:
   input: '0.05'
   output: '0.05'
diff --git a/api/core/model_runtime/model_providers/zhipuai/llm/glm_4v_plus.yaml b/api/core/model_runtime/model_providers/zhipuai/llm/glm_4v_plus.yaml
index 4c5fa2403413e..14b9623e5a8c4 100644
--- a/api/core/model_runtime/model_providers/zhipuai/llm/glm_4v_plus.yaml
+++ b/api/core/model_runtime/model_providers/zhipuai/llm/glm_4v_plus.yaml
@@ -44,6 +44,15 @@ parameter_rules:
     default: 1024
     min: 1
     max: 1024
+  - name: web_search
+    type: boolean
+    label:
+      zh_Hans: 联网搜索
+      en_US: Web Search
+    default: false
+    help:
+      zh_Hans: 模型内置了互联网搜索服务，该参数控制模型在生成文本时是否参考使用互联网搜索结果。启用互联网搜索，模型会将搜索结果作为文本生成过程中的参考信息，但模型会基于其内部逻辑“自行判断”是否使用互联网搜索结果。
+      en_US: The model has a built-in Internet search service. This parameter controls whether the model refers to Internet search results when generating text. When Internet search is enabled, the model will use the search results as reference information in the text generation process, but the model will "judge" whether to use Internet search results based on its internal logic.
 pricing:
   input: '0.01'
   output: '0.01'

From a126d535cf146ce5473974bfbe500d209a5b1693 Mon Sep 17 00:00:00 2001
From: AAEE86 <33052466+AAEE86@users.noreply.github.com>
Date: Mon, 23 Sep 2024 16:39:46 +0800
Subject: [PATCH 037/235] add Spark Max-32K (#8676)

---
 .../model_providers/spark/llm/_client.py      |  3 +-
 .../model_providers/spark/llm/_position.yaml  |  1 +
 .../spark/llm/spark-max-32k.yaml              | 33 +++++++++++++++++++
 3 files changed, 36 insertions(+), 1 deletion(-)
 create mode 100644 api/core/model_runtime/model_providers/spark/llm/spark-max-32k.yaml

diff --git a/api/core/model_runtime/model_providers/spark/llm/_client.py b/api/core/model_runtime/model_providers/spark/llm/_client.py
index b99a657e715ed..48911f657a52e 100644
--- a/api/core/model_runtime/model_providers/spark/llm/_client.py
+++ b/api/core/model_runtime/model_providers/spark/llm/_client.py
@@ -25,6 +25,7 @@ def __init__(self, model: str, app_id: str, api_key: str, api_secret: str, api_d
             "spark-pro": {"version": "v3.1", "chat_domain": "generalv3"},
             "spark-pro-128k": {"version": "pro-128k", "chat_domain": "pro-128k"},
             "spark-max": {"version": "v3.5", "chat_domain": "generalv3.5"},
+            "spark-max-32k": {"version": "max-32k", "chat_domain": "max-32k"},
             "spark-4.0-ultra": {"version": "v4.0", "chat_domain": "4.0Ultra"},
         }
 
@@ -32,7 +33,7 @@ def __init__(self, model: str, app_id: str, api_key: str, api_secret: str, api_d
 
         self.chat_domain = model_api_configs[model]["chat_domain"]
 
-        if model == "spark-pro-128k":
+        if model in ["spark-pro-128k", "spark-max-32k"]:
             self.api_base = f"wss://{domain}/{endpoint}/{api_version}"
         else:
             self.api_base = f"wss://{domain}/{api_version}/{endpoint}"
diff --git a/api/core/model_runtime/model_providers/spark/llm/_position.yaml b/api/core/model_runtime/model_providers/spark/llm/_position.yaml
index 458397f2aaf1c..73f39cb1197b4 100644
--- a/api/core/model_runtime/model_providers/spark/llm/_position.yaml
+++ b/api/core/model_runtime/model_providers/spark/llm/_position.yaml
@@ -1,3 +1,4 @@
+- spark-max-32k
 - spark-4.0-ultra
 - spark-max
 - spark-pro-128k
diff --git a/api/core/model_runtime/model_providers/spark/llm/spark-max-32k.yaml b/api/core/model_runtime/model_providers/spark/llm/spark-max-32k.yaml
new file mode 100644
index 0000000000000..1a1ab6844c69c
--- /dev/null
+++ b/api/core/model_runtime/model_providers/spark/llm/spark-max-32k.yaml
@@ -0,0 +1,33 @@
+model: spark-max-32k
+label:
+  en_US: Spark Max-32K
+model_type: llm
+model_properties:
+  mode: chat
+parameter_rules:
+  - name: temperature
+    use_template: temperature
+    default: 0.5
+    help:
+      zh_Hans: 核采样阈值。用于决定结果随机性，取值越高随机性越强即相同的问题得到的不同答案的可能性越高。
+      en_US: Kernel sampling threshold. Used to determine the randomness of the results. The higher the value, the stronger the randomness, that is, the higher the possibility of getting different answers to the same question.
+  - name: max_tokens
+    use_template: max_tokens
+    default: 4096
+    min: 1
+    max: 8192
+    help:
+      zh_Hans: 模型回答的tokens的最大长度。
+      en_US: Maximum length of tokens for the model response.
+  - name: top_k
+    label:
+      zh_Hans: 取样数量
+      en_US: Top k
+    type: int
+    default: 4
+    min: 1
+    max: 6
+    help:
+      zh_Hans: 从 k 个候选中随机选择一个（非等概率）。
+      en_US: Randomly select one from k candidates (non-equal probability).
+    required: false

From c7eacd1aacf7c67e9ec10c885d6f46bee1333fc2 Mon Sep 17 00:00:00 2001
From: Nam Vu <zuzoovn@gmail.com>
Date: Mon, 23 Sep 2024 17:40:40 +0700
Subject: [PATCH 038/235] chore: Optimize I18nObject class for better
 performance and readability (#8681)

---
 api/core/tools/entities/common_entities.py | 17 +++++++----------
 1 file changed, 7 insertions(+), 10 deletions(-)

diff --git a/api/core/tools/entities/common_entities.py b/api/core/tools/entities/common_entities.py
index b52119fdc4759..924e6fc0cf9f1 100644
--- a/api/core/tools/entities/common_entities.py
+++ b/api/core/tools/entities/common_entities.py
@@ -1,6 +1,6 @@
 from typing import Optional
 
-from pydantic import BaseModel
+from pydantic import BaseModel, Field
 
 
 class I18nObject(BaseModel):
@@ -8,19 +8,16 @@ class I18nObject(BaseModel):
     Model class for i18n object.
     """
 
-    zh_Hans: Optional[str] = None
-    pt_BR: Optional[str] = None
-    ja_JP: Optional[str] = None
     en_US: str
+    zh_Hans: Optional[str] = Field(default=None)
+    pt_BR: Optional[str] = Field(default=None)
+    ja_JP: Optional[str] = Field(default=None)
 
     def __init__(self, **data):
         super().__init__(**data)
-        if not self.zh_Hans:
-            self.zh_Hans = self.en_US
-        if not self.pt_BR:
-            self.pt_BR = self.en_US
-        if not self.ja_JP:
-            self.ja_JP = self.en_US
+        self.zh_Hans = self.zh_Hans or self.en_US
+        self.pt_BR = self.pt_BR or self.en_US
+        self.ja_JP = self.ja_JP or self.en_US
 
     def to_dict(self) -> dict:
         return {"zh_Hans": self.zh_Hans, "en_US": self.en_US, "pt_BR": self.pt_BR, "ja_JP": self.ja_JP}

From 11d09a92d0cd954aa81e6cbcfad961e6d8ddb815 Mon Sep 17 00:00:00 2001
From: Hash Brown <hi@xzd.me>
Date: Mon, 23 Sep 2024 18:44:09 +0800
Subject: [PATCH 039/235] fix: send message error when last sent message not
 succeeded (#8682)

---
 .../debug/debug-with-single-model/index.tsx         | 13 ++++---------
 .../base/chat/chat-with-history/chat-wrapper.tsx    | 13 ++++---------
 .../base/chat/embedded-chatbot/chat-wrapper.tsx     | 13 ++++---------
 web/app/components/base/chat/types.ts               |  2 +-
 web/app/components/base/chat/utils.ts               | 10 ++++++++++
 .../panel/debug-and-preview/chat-wrapper.tsx        | 13 ++++---------
 6 files changed, 27 insertions(+), 37 deletions(-)

diff --git a/web/app/components/app/configuration/debug/debug-with-single-model/index.tsx b/web/app/components/app/configuration/debug/debug-with-single-model/index.tsx
index 5faef46d988e0..2de500a3a6211 100644
--- a/web/app/components/app/configuration/debug/debug-with-single-model/index.tsx
+++ b/web/app/components/app/configuration/debug/debug-with-single-model/index.tsx
@@ -22,6 +22,7 @@ import {
 import Avatar from '@/app/components/base/avatar'
 import { useAppContext } from '@/context/app-context'
 import { ModelFeatureEnum } from '@/app/components/header/account-setting/model-provider-page/declarations'
+import { getLastAnswer } from '@/app/components/base/chat/utils'
 
 type DebugWithSingleModelProps = {
   checkCanSend?: () => boolean
@@ -83,17 +84,11 @@ const DebugWithSingleModel = forwardRef<DebugWithSingleModelRefType, DebugWithSi
       },
     }
 
-    const lastAnswer = chatListRef.current.at(-1)
-
     const data: any = {
       query: message,
       inputs,
       model_config: configData,
-      parent_message_id: last_answer?.id || (lastAnswer
-        ? lastAnswer.isOpeningStatement
-          ? null
-          : lastAnswer.id
-        : null),
+      parent_message_id: last_answer?.id || getLastAnswer(chatListRef.current)?.id || null,
     }
 
     if (visionConfig.enabled && files?.length && supportVision)
@@ -116,13 +111,13 @@ const DebugWithSingleModel = forwardRef<DebugWithSingleModelRefType, DebugWithSi
 
     const prevMessages = chatList.slice(0, index)
     const question = prevMessages.pop()
-    const lastAnswer = prevMessages.at(-1)
+    const lastAnswer = getLastAnswer(prevMessages)
 
     if (!question)
       return
 
     handleUpdateChatList(prevMessages)
-    doSend(question.content, question.message_files, (!lastAnswer || lastAnswer.isOpeningStatement) ? undefined : lastAnswer)
+    doSend(question.content, question.message_files, lastAnswer)
   }, [chatList, handleUpdateChatList, doSend])
 
   const allToolIcons = useMemo(() => {
diff --git a/web/app/components/base/chat/chat-with-history/chat-wrapper.tsx b/web/app/components/base/chat/chat-with-history/chat-wrapper.tsx
index 225bbac714084..9d7b360f38d06 100644
--- a/web/app/components/base/chat/chat-with-history/chat-wrapper.tsx
+++ b/web/app/components/base/chat/chat-with-history/chat-wrapper.tsx
@@ -6,6 +6,7 @@ import type {
   OnSend,
 } from '../types'
 import { useChat } from '../chat/hooks'
+import { getLastAnswer } from '../utils'
 import { useChatWithHistoryContext } from './context'
 import Header from './header'
 import ConfigPanel from './config-panel'
@@ -67,17 +68,11 @@ const ChatWrapper = () => {
   }, [])
 
   const doSend: OnSend = useCallback((message, files, last_answer) => {
-    const lastAnswer = chatListRef.current.at(-1)
-
     const data: any = {
       query: message,
       inputs: currentConversationId ? currentConversationItem?.inputs : newConversationInputs,
       conversation_id: currentConversationId,
-      parent_message_id: last_answer?.id || (lastAnswer
-        ? lastAnswer.isOpeningStatement
-          ? null
-          : lastAnswer.id
-        : null),
+      parent_message_id: last_answer?.id || getLastAnswer(chatListRef.current)?.id || null,
     }
 
     if (appConfig?.file_upload?.image.enabled && files?.length)
@@ -111,13 +106,13 @@ const ChatWrapper = () => {
 
     const prevMessages = chatList.slice(0, index)
     const question = prevMessages.pop()
-    const lastAnswer = prevMessages.at(-1)
+    const lastAnswer = getLastAnswer(prevMessages)
 
     if (!question)
       return
 
     handleUpdateChatList(prevMessages)
-    doSend(question.content, question.message_files, (!lastAnswer || lastAnswer.isOpeningStatement) ? undefined : lastAnswer)
+    doSend(question.content, question.message_files, lastAnswer)
   }, [chatList, handleUpdateChatList, doSend])
 
   const chatNode = useMemo(() => {
diff --git a/web/app/components/base/chat/embedded-chatbot/chat-wrapper.tsx b/web/app/components/base/chat/embedded-chatbot/chat-wrapper.tsx
index ed2f24274dcd7..b97c940eece7d 100644
--- a/web/app/components/base/chat/embedded-chatbot/chat-wrapper.tsx
+++ b/web/app/components/base/chat/embedded-chatbot/chat-wrapper.tsx
@@ -6,6 +6,7 @@ import type {
   OnSend,
 } from '../types'
 import { useChat } from '../chat/hooks'
+import { getLastAnswer } from '../utils'
 import { useEmbeddedChatbotContext } from './context'
 import ConfigPanel from './config-panel'
 import { isDify } from './utils'
@@ -69,17 +70,11 @@ const ChatWrapper = () => {
   }, [])
 
   const doSend: OnSend = useCallback((message, files, last_answer) => {
-    const lastAnswer = chatListRef.current.at(-1)
-
     const data: any = {
       query: message,
       inputs: currentConversationId ? currentConversationItem?.inputs : newConversationInputs,
       conversation_id: currentConversationId,
-      parent_message_id: last_answer?.id || (lastAnswer
-        ? lastAnswer.isOpeningStatement
-          ? null
-          : lastAnswer.id
-        : null),
+      parent_message_id: last_answer?.id || getLastAnswer(chatListRef.current)?.id || null,
     }
 
     if (appConfig?.file_upload?.image.enabled && files?.length)
@@ -113,13 +108,13 @@ const ChatWrapper = () => {
 
     const prevMessages = chatList.slice(0, index)
     const question = prevMessages.pop()
-    const lastAnswer = prevMessages.at(-1)
+    const lastAnswer = getLastAnswer(prevMessages)
 
     if (!question)
       return
 
     handleUpdateChatList(prevMessages)
-    doSend(question.content, question.message_files, (!lastAnswer || lastAnswer.isOpeningStatement) ? undefined : lastAnswer)
+    doSend(question.content, question.message_files, lastAnswer)
   }, [chatList, handleUpdateChatList, doSend])
 
   const chatNode = useMemo(() => {
diff --git a/web/app/components/base/chat/types.ts b/web/app/components/base/chat/types.ts
index 489dbb44cf67e..0bc50518ebb81 100644
--- a/web/app/components/base/chat/types.ts
+++ b/web/app/components/base/chat/types.ts
@@ -63,7 +63,7 @@ export type ChatItem = IChatItem & {
   conversationId?: string
 }
 
-export type OnSend = (message: string, files?: VisionFile[], last_answer?: ChatItem) => void
+export type OnSend = (message: string, files?: VisionFile[], last_answer?: ChatItem | null) => void
 
 export type OnRegenerate = (chatItem: ChatItem) => void
 
diff --git a/web/app/components/base/chat/utils.ts b/web/app/components/base/chat/utils.ts
index e851c4c4633fb..305df5995d128 100644
--- a/web/app/components/base/chat/utils.ts
+++ b/web/app/components/base/chat/utils.ts
@@ -19,6 +19,15 @@ function getProcessedInputsFromUrlParams(): Record<string, any> {
   return inputs
 }
 
+function getLastAnswer(chatList: ChatItem[]) {
+  for (let i = chatList.length - 1; i >= 0; i--) {
+    const item = chatList[i]
+    if (item.isAnswer && !item.isOpeningStatement)
+      return item
+  }
+  return null
+}
+
 function appendQAToChatList(chatList: ChatItem[], item: any) {
   // we append answer first and then question since will reverse the whole chatList later
   chatList.push({
@@ -71,5 +80,6 @@ function getPrevChatList(fetchedMessages: any[]) {
 
 export {
   getProcessedInputsFromUrlParams,
+  getLastAnswer,
   getPrevChatList,
 }
diff --git a/web/app/components/workflow/panel/debug-and-preview/chat-wrapper.tsx b/web/app/components/workflow/panel/debug-and-preview/chat-wrapper.tsx
index 86519af60357d..230b2d7fa0c65 100644
--- a/web/app/components/workflow/panel/debug-and-preview/chat-wrapper.tsx
+++ b/web/app/components/workflow/panel/debug-and-preview/chat-wrapper.tsx
@@ -25,6 +25,7 @@ import {
   stopChatMessageResponding,
 } from '@/service/debug'
 import { useStore as useAppStore } from '@/app/components/app/store'
+import { getLastAnswer } from '@/app/components/base/chat/utils'
 
 type ChatWrapperProps = {
   showConversationVariableModal: boolean
@@ -76,19 +77,13 @@ const ChatWrapper = forwardRef<ChatWrapperRefType, ChatWrapperProps>(({ showConv
   )
 
   const doSend = useCallback<OnSend>((query, files, last_answer) => {
-    const lastAnswer = chatListRef.current.at(-1)
-
     handleSend(
       {
         query,
         files,
         inputs: workflowStore.getState().inputs,
         conversation_id: conversationId,
-        parent_message_id: last_answer?.id || (lastAnswer
-          ? lastAnswer.isOpeningStatement
-            ? null
-            : lastAnswer.id
-          : null),
+        parent_message_id: last_answer?.id || getLastAnswer(chatListRef.current)?.id || null,
       },
       {
         onGetSuggestedQuestions: (messageId, getAbortController) => fetchSuggestedQuestions(appDetail!.id, messageId, getAbortController),
@@ -103,13 +98,13 @@ const ChatWrapper = forwardRef<ChatWrapperRefType, ChatWrapperProps>(({ showConv
 
     const prevMessages = chatList.slice(0, index)
     const question = prevMessages.pop()
-    const lastAnswer = prevMessages.at(-1)
+    const lastAnswer = getLastAnswer(prevMessages)
 
     if (!question)
       return
 
     handleUpdateChatList(prevMessages)
-    doSend(question.content, question.message_files, (!lastAnswer || lastAnswer.isOpeningStatement) ? undefined : lastAnswer)
+    doSend(question.content, question.message_files, lastAnswer)
   }, [chatList, handleUpdateChatList, doSend])
 
   useImperativeHandle(ref, () => {

From 52da5b16e7002fc15343d2df73838b3f8220cccd Mon Sep 17 00:00:00 2001
From: Likename Haojie <likenamehaojie@hotmail.com>
Date: Mon, 23 Sep 2024 18:44:24 +0800
Subject: [PATCH 040/235] fixbug tts(stream) not work on ios safari(17.1+)
 (#8645)

Co-authored-by: crazywoola <427733928@qq.com>
---
 web/app/components/base/audio-btn/audio.ts | 8 ++++++--
 1 file changed, 6 insertions(+), 2 deletions(-)

diff --git a/web/app/components/base/audio-btn/audio.ts b/web/app/components/base/audio-btn/audio.ts
index a61fd085d44c4..baf675d0be92d 100644
--- a/web/app/components/base/audio-btn/audio.ts
+++ b/web/app/components/base/audio-btn/audio.ts
@@ -12,7 +12,7 @@ export default class AudioPlayer {
   mediaSource: MediaSource | null
   audio: HTMLAudioElement
   audioContext: AudioContext
-  sourceBuffer?: SourceBuffer
+  sourceBuffer?: any
   cacheBuffers: ArrayBuffer[] = []
   pauseTimer: number | null = null
   msgId: string | undefined
@@ -33,7 +33,7 @@ export default class AudioPlayer {
     this.callback = callback
 
     // Compatible with iphone ios17 ManagedMediaSource
-    const MediaSource = window.MediaSource || window.ManagedMediaSource
+    const MediaSource = window.ManagedMediaSource || window.MediaSource
     if (!MediaSource) {
       Toast.notify({
         message: 'Your browser does not support audio streaming, if you are using an iPhone, please update to iOS 17.1 or later.',
@@ -43,6 +43,10 @@ export default class AudioPlayer {
     this.mediaSource = MediaSource ? new MediaSource() : null
     this.audio = new Audio()
     this.setCallback(callback)
+    if (!window.MediaSource) { // if use  ManagedMediaSource
+      this.audio.disableRemotePlayback = true
+      this.audio.controls = true
+    }
     this.audio.src = this.mediaSource ? URL.createObjectURL(this.mediaSource) : ''
     this.audio.autoplay = true
 

From 4f69adc8ab0b67d45ff5f578502c521086f0ba13 Mon Sep 17 00:00:00 2001
From: Vikey Chen <vikeytk@gmail.com>
Date: Mon, 23 Sep 2024 18:45:10 +0800
Subject: [PATCH 041/235] fix: document_create_args_validate (#8569)

---
 api/services/dataset_service.py | 4 ++--
 1 file changed, 2 insertions(+), 2 deletions(-)

diff --git a/api/services/dataset_service.py b/api/services/dataset_service.py
index 30c010ef29f62..e96f06ed40fa4 100644
--- a/api/services/dataset_service.py
+++ b/api/services/dataset_service.py
@@ -1100,8 +1100,8 @@ def document_create_args_validate(cls, args: dict):
             DocumentService.data_source_args_validate(args)
             DocumentService.process_rule_args_validate(args)
         else:
-            if ("data_source" not in args and not args["data_source"]) and (
-                "process_rule" not in args and not args["process_rule"]
+            if ("data_source" not in args or not args["data_source"]) and (
+                "process_rule" not in args or not args["process_rule"]
             ):
                 raise ValueError("Data source or Process rule is required")
             else:

From d7aada38a111028ade07b597ff2f9e201d57e35b Mon Sep 17 00:00:00 2001
From: ice yao <yao3690093@gmail.com>
Date: Mon, 23 Sep 2024 19:57:21 +0800
Subject: [PATCH 042/235] Add nomic embedding model provider (#8640)

---
 .../model_providers/nomic/__init__.py         |   0
 .../nomic/_assets/icon_l_en.svg               |  13 ++
 .../nomic/_assets/icon_s_en.png               | Bin 0 -> 25814 bytes
 .../model_providers/nomic/_common.py          |  28 +++
 .../model_providers/nomic/nomic.py            |  26 +++
 .../model_providers/nomic/nomic.yaml          |  29 +++
 .../nomic/text_embedding/__init__.py          |   0
 .../text_embedding/nomic-embed-text-v1.5.yaml |   8 +
 .../text_embedding/nomic-embed-text-v1.yaml   |   8 +
 .../nomic/text_embedding/text_embedding.py    | 170 ++++++++++++++++++
 api/poetry.lock                               |  78 +++++++-
 api/pyproject.toml                            |   2 +
 .../model_runtime/__mock/nomic_embeddings.py  |  59 ++++++
 .../model_runtime/nomic/__init__.py           |   0
 .../model_runtime/nomic/test_embeddings.py    |  62 +++++++
 .../model_runtime/nomic/test_provider.py      |  22 +++
 dev/pytest/pytest_model_runtime.sh            |   3 +-
 17 files changed, 506 insertions(+), 2 deletions(-)
 create mode 100644 api/core/model_runtime/model_providers/nomic/__init__.py
 create mode 100644 api/core/model_runtime/model_providers/nomic/_assets/icon_l_en.svg
 create mode 100644 api/core/model_runtime/model_providers/nomic/_assets/icon_s_en.png
 create mode 100644 api/core/model_runtime/model_providers/nomic/_common.py
 create mode 100644 api/core/model_runtime/model_providers/nomic/nomic.py
 create mode 100644 api/core/model_runtime/model_providers/nomic/nomic.yaml
 create mode 100644 api/core/model_runtime/model_providers/nomic/text_embedding/__init__.py
 create mode 100644 api/core/model_runtime/model_providers/nomic/text_embedding/nomic-embed-text-v1.5.yaml
 create mode 100644 api/core/model_runtime/model_providers/nomic/text_embedding/nomic-embed-text-v1.yaml
 create mode 100644 api/core/model_runtime/model_providers/nomic/text_embedding/text_embedding.py
 create mode 100644 api/tests/integration_tests/model_runtime/__mock/nomic_embeddings.py
 create mode 100644 api/tests/integration_tests/model_runtime/nomic/__init__.py
 create mode 100644 api/tests/integration_tests/model_runtime/nomic/test_embeddings.py
 create mode 100644 api/tests/integration_tests/model_runtime/nomic/test_provider.py

diff --git a/api/core/model_runtime/model_providers/nomic/__init__.py b/api/core/model_runtime/model_providers/nomic/__init__.py
new file mode 100644
index 0000000000000..e69de29bb2d1d
diff --git a/api/core/model_runtime/model_providers/nomic/_assets/icon_l_en.svg b/api/core/model_runtime/model_providers/nomic/_assets/icon_l_en.svg
new file mode 100644
index 0000000000000..6c4a1058ab9c7
--- /dev/null
+++ b/api/core/model_runtime/model_providers/nomic/_assets/icon_l_en.svg
@@ -0,0 +1,13 @@
+<svg width="93" height="31" viewBox="0 0 93 31" fill="none" xmlns="http://www.w3.org/2000/svg">
+<path d="M89.6037 29.888C88.9464 29.888 88.3667 29.7302 87.8647 29.4145C87.3626 29.0936 86.9719 28.6407 86.6924 28.0559L87.2979 27.4037C87.5464 27.9109 87.8698 28.3069 88.2684 28.5915C88.6669 28.871 89.1094 29.0108 89.5959 29.0108C89.922 29.0108 90.2196 28.9435 90.4887 28.8089C90.763 28.6744 90.9804 28.4829 91.1408 28.2344C91.3064 27.9808 91.3892 27.6806 91.3892 27.3339C91.3892 27.0182 91.3116 26.7697 91.1563 26.5886C91.0062 26.4074 90.7837 26.2522 90.4887 26.1228C90.1988 25.9882 89.8366 25.8381 89.4018 25.6725C89.0654 25.5379 88.7393 25.3853 88.4236 25.2145C88.1079 25.0437 87.8465 24.8289 87.6395 24.5701C87.4377 24.3061 87.3367 23.9723 87.3367 23.5686C87.3367 23.1598 87.4454 22.7975 87.6628 22.4817C87.8802 22.1609 88.1804 21.9098 88.5634 21.7287C88.9464 21.5424 89.3811 21.4492 89.8676 21.4492C90.3127 21.4492 90.7293 21.545 91.1175 21.7365C91.5109 21.928 91.8628 22.1997 92.1733 22.5516L91.6532 23.2115C91.177 22.5853 90.5844 22.2721 89.8754 22.2721C89.4406 22.2721 89.0861 22.386 88.8118 22.6137C88.5427 22.8415 88.4081 23.1391 88.4081 23.5065C88.4081 23.7705 88.4935 23.9904 88.6643 24.1664C88.8351 24.3424 89.0576 24.4925 89.3319 24.6167C89.6114 24.7409 89.9116 24.8651 90.2325 24.9893C90.6983 25.1653 91.102 25.3413 91.4436 25.5172C91.7903 25.6932 92.0595 25.9183 92.251 26.1927C92.4425 26.4618 92.5382 26.8293 92.5382 27.2951C92.5382 27.8281 92.414 28.2888 92.1656 28.6769C91.9171 29.0651 91.5704 29.3653 91.1253 29.5775C90.6854 29.7845 90.1781 29.888 89.6037 29.888Z" fill="#3C593D"/>
+<path d="M79.8324 29.8841C79.0871 29.8841 78.4143 29.7029 77.8139 29.3406C77.2187 28.9732 76.7451 28.4711 76.3932 27.8345C76.0464 27.1979 75.873 26.4708 75.873 25.653C75.873 24.8456 76.0438 24.1262 76.3854 23.4948C76.7322 22.8582 77.2032 22.3562 77.7984 21.9887C78.3987 21.6212 79.0767 21.4375 79.8324 21.4375C80.5518 21.4375 81.2039 21.6057 81.7888 21.9421C82.3736 22.2785 82.8187 22.7443 83.1241 23.3395V21.6859H84.2575V29.6356H83.1241V27.9587C82.7825 28.5591 82.3244 29.0301 81.7499 29.3717C81.1754 29.7133 80.5363 29.8841 79.8324 29.8841ZM80.1119 28.8981C80.7071 28.8981 81.2324 28.761 81.6878 28.4867C82.1485 28.2072 82.5107 27.8242 82.7747 27.3377C83.0387 26.846 83.1706 26.287 83.1706 25.6608C83.1706 25.0294 83.0387 24.4704 82.7747 23.9839C82.5159 23.4974 82.1562 23.117 81.6956 22.8427C81.235 22.5632 80.7071 22.4235 80.1119 22.4235C79.5167 22.4235 78.9888 22.5632 78.5281 22.8427C78.0675 23.117 77.7052 23.4974 77.4413 23.9839C77.1773 24.4704 77.0453 25.0294 77.0453 25.6608C77.0453 26.287 77.1773 26.846 77.4413 27.3377C77.7052 27.8242 78.0675 28.2072 78.5281 28.4867C78.9888 28.761 79.5167 28.8981 80.1119 28.8981Z" fill="#3C593D"/>
+<path d="M71.9658 29.6382V16.2852H73.0993V29.6382H71.9658Z" fill="#3C593D"/>
+<path d="M68.1539 29.8864C67.5587 29.8864 67.0955 29.6871 66.7643 29.2886C66.4382 28.8849 66.2752 28.3182 66.2752 27.5884V22.5422H65.4678V21.6882H66.2752V18.7148H67.4086V21.6882H69.3883V22.5422H67.4086V27.5263C67.4086 27.9662 67.494 28.3026 67.6648 28.5355C67.8356 28.7684 68.0789 28.8849 68.3946 28.8849C68.6999 28.8849 68.9691 28.7995 69.202 28.6287L69.4892 29.5292C69.3132 29.6379 69.1062 29.7233 68.8681 29.7854C68.6301 29.8527 68.392 29.8864 68.1539 29.8864Z" fill="#3C593D"/>
+<path d="M58.513 29.8841C57.7678 29.8841 57.0949 29.7029 56.4946 29.3406C55.8994 28.9732 55.4258 28.4711 55.0739 27.8345C54.7271 27.1979 54.5537 26.4708 54.5537 25.653C54.5537 24.8456 54.7245 24.1262 55.0661 23.4948C55.4129 22.8582 55.8838 22.3562 56.479 21.9887C57.0794 21.6212 57.7574 21.4375 58.513 21.4375C59.2324 21.4375 59.8846 21.6057 60.4694 21.9421C61.0543 22.2785 61.4994 22.7443 61.8047 23.3395V21.6859H62.9382V29.6356H61.8047V27.9587C61.4631 28.5591 61.0051 29.0301 60.4306 29.3717C59.8561 29.7133 59.2169 29.8841 58.513 29.8841ZM58.7925 28.8981C59.3877 28.8981 59.913 28.761 60.3685 28.4867C60.8291 28.2072 61.1914 27.8242 61.4554 27.3377C61.7193 26.846 61.8513 26.287 61.8513 25.6608C61.8513 25.0294 61.7193 24.4704 61.4554 23.9839C61.1966 23.4974 60.8369 23.117 60.3763 22.8427C59.9156 22.5632 59.3877 22.4235 58.7925 22.4235C58.1973 22.4235 57.6694 22.5632 57.2088 22.8427C56.7482 23.117 56.3859 23.4974 56.1219 23.9839C55.858 24.4704 55.726 25.0294 55.726 25.6608C55.726 26.287 55.858 26.846 56.1219 27.3377C56.3859 27.8242 56.7482 28.2072 57.2088 28.4867C57.6694 28.761 58.1973 28.8981 58.7925 28.8981Z" fill="#3C593D"/>
+<path d="M5.41228 22.6607V0H6.76535V30.2143H5.41228L1.35307 7.55357V30.2143H0V0H1.35307L5.41228 22.6607Z" fill="#3C593D"/>
+<path d="M13.6575 28.9006C14.024 28.9006 14.3341 28.7775 14.5878 28.5312C14.8697 28.2848 15.0106 27.9701 15.0106 27.587V2.62733C15.0106 2.27154 14.8697 1.9705 14.5878 1.72418C14.3341 1.4505 14.024 1.31366 13.6575 1.31366C13.2629 1.31366 12.9387 1.4505 12.685 1.72418C12.4313 1.9705 12.3045 2.27154 12.3045 2.62733V27.587C12.3045 27.9701 12.4313 28.2848 12.685 28.5312C12.9387 28.7775 13.2629 28.9006 13.6575 28.9006ZM13.6575 30.2143C12.8964 30.2143 12.2481 29.968 11.7125 29.4753C11.2051 28.9554 10.9514 28.3259 10.9514 27.587V2.62733C10.9514 1.91576 11.2051 1.29998 11.7125 0.779988C12.2481 0.259996 12.8964 0 13.6575 0C14.3905 0 15.0247 0.259996 15.5603 0.779988C16.0959 1.29998 16.3637 1.91576 16.3637 2.62733V27.587C16.3637 28.3259 16.0959 28.9554 15.5603 29.4753C15.0247 29.968 14.3905 30.2143 13.6575 30.2143Z" fill="#3C593D"/>
+<path d="M28.3299 0H29.683V30.2143H28.3299V5.25466L24.9472 18.3913L21.5645 5.25466V30.2143H20.2115V0H21.5645L24.9472 13.1366L28.3299 0Z" fill="#3C593D"/>
+<path d="M33.6999 30.2143V0H35.0529V30.2143H33.6999Z" fill="#3C593D"/>
+<path d="M41.776 30.2143C41.0149 30.2143 40.3666 29.968 39.831 29.4753C39.3236 28.9554 39.0699 28.3259 39.0699 27.587V2.62733C39.0699 1.91576 39.3236 1.29998 39.831 0.779988C40.3666 0.259996 41.0149 0 41.776 0C42.5089 0 43.1432 0.259996 43.6788 0.779988C44.2143 1.29998 44.4821 1.91576 44.4821 2.62733V5.25466H43.1291V2.62733C43.1291 2.27154 42.9881 1.9705 42.7062 1.72418C42.4525 1.4505 42.1425 1.31366 41.776 1.31366C41.3814 1.31366 41.0572 1.4505 40.8035 1.72418C40.5498 1.9705 40.4229 2.27154 40.4229 2.62733V27.587C40.4229 27.9701 40.5498 28.2848 40.8035 28.5312C41.0572 28.7775 41.3814 28.9006 41.776 28.9006C42.1425 28.9006 42.4525 28.7775 42.7062 28.5312C42.9881 28.2848 43.1291 27.9701 43.1291 27.587V24.9596H44.4821V27.587C44.4821 28.3259 44.2143 28.9554 43.6788 29.4753C43.1432 29.968 42.5089 30.2143 41.776 30.2143Z" fill="#3C593D"/>
+<path d="M56 1H91" stroke="#3C593D" stroke-linecap="round" stroke-dasharray="0.1 2"/>
+</svg>
diff --git a/api/core/model_runtime/model_providers/nomic/_assets/icon_s_en.png b/api/core/model_runtime/model_providers/nomic/_assets/icon_s_en.png
new file mode 100644
index 0000000000000000000000000000000000000000..3eba3b82bc1e3fcb6b27cd9038786e19b7c2adcf
GIT binary patch
literal 25814
zcmce8c{o*H`?q<XnvDldR3Z_I!zm%j&_pFeUzM?r6w)T7QpuE12`94<aSozgN`q3F
z$;OE$itUtYpCt9Jz4uAJ-}iT2@AdoVeXi?y9__W)TK9Z^?t86$&N|pxNlPe7@bU3U
zTW?sug^y1N<Kq)NAtnf()Y}!W1pmi>c#G9K{_ZDQ;ou+Q{u_24=Hr{Y2>&m?^_GR7
z`1k~x9X4&V0E2h$-VF~A9~c;DY;3HotYk14=gysriHW&(?OIh;)$`}i-@JKq`SRtr
zZ{IdGHPzJA<mTqa$H$*Kb?WNXs}&U$Po6yK@9zieEL^y-uC9(oqrHFszPPyf*s)`K
z_wKc~w_m+_^@0TpWMpJ!%$T8}p+Timw{G3)?d=^A5pnzWZChL0ef#z~I5-#^8!uY4
zNKQ`9#KdIFmMvXfUC*99b8~YuG&DSX_^^VaVtIM_jF~f+EnDW{;gOn}dj9-*M@Pry
z%a@-%efrg_SEZ$;J9g~Q($b2Git6d<*}s2(cXzk6w6v$Er@p?vxR}`5wQKL)yLa>E
zO=o9k5n*8~E32fWq>mpzDk>?l+3YiC&gkjsMMg#<2$GhTW^QgiciucYo&M<2qoJXp
zg9i^vN=T@xsumU&E?&I2y1M$ri4!MJo@{S#-??+=yLayn95}#Yv0PnU@87?F<;s=$
z^XD5G8Ch6ZJbwIm=B!yGBO^zT9t{l*6%rDvudmnD)g2rh%*@P`m6c^OnGgg;M@Rel
z`Ptdog@lBxSh2#!#>U;<-O|!hO-=32ojV^sd=M28xpe8$g9i^98XEfg`lO_!UcP)e
zIy$PUsX1%*>_dkRJ$(4^#*G{Q{PRyjLPAPPijR-a_3PJ%hlkg!S<}|mrlhP~P*8C8
z?AdMGw#CNAmXwrqc6R#u`l_p|8yFarm6Zhq1UNZ4xwyEjTenV9Qu683r!6fl$;ruK
zVPQHtI@`Bz-?eL3U|?WkVxpj+U`|fX#fum7^Yb@u+}P34v1-+-O`A5=*4DPRwq|5x
zWM^lanwoCdut8ZxB`z+GLZJi)2XEfIIVdRT`0?YrckfP5PtVHAl2=ezzkdDfIdiP7
zttG_8U%YrBCMxRh?>|po{`KqEMMXs@ist3ztz5bC!i5XX&CR!N-I_CZE)2uHy}f((
z>{+^W>5?T&jvP56ARyr7<;Bm>udS_ZW@aWVuD=o3;-@29?6&g#{@wqZ_n*4oWq<zs
zg#VH7=U@NZ|96-+!T;~yjK6=bApVT_`vCvTo%c_<zmI8u|NP(o)%;uS|C;o)t|5ev
z&pq3E{n~9uucGPP?7iEQD`T!6UYeLwm$2Si=FAD9Xw4HFy@YSBeX`a;NNRnc1MS2L
zWaeH6p=}OA$}8k76V@&~lT|5UbYjn=6|e4o3Vb!%6K38~7|ek?YrC*JO--Jq-$u6v
zX5N7hc67rTa_hv@PyT=VpEGqcZ3Le6n^<bGzOHQ4uKFpl>;^LcUAUq_vZa67j00{(
zd|LNnJnF6&mE3uB#kT~0Q8V0OxAN+uAK4+ox=lZ58BR05T)cVjw9<jwPl{aO4-dYk
z9r$JM?02i9la(F2cUgb(%e<flsY-s#qMf}g(#Y0}&W0ve_1wV%du*ldd>F2LrT!LD
zX>kp=a?J==Mp5R|VYe1mOgKBFzV5g3KS<BMC}c@exBEPIw^moh=Wv%g3P+|6s;+~S
ziYqz{YUdkbv2^6X2V_rlq*v)(JM`j{;wV2GB&ypzeUVaVsX5Y<4qm50N_P})<;T4A
zXo7t-C}KHqiBJ9}5hR5vTF!M&=(%+!yxa2?a&{h{H{HxNqmxzj+Z0*|&v<{hE^@dl
zj>^{{i%N47S~8fTH^;{476;!ilsPtej^)uL+RE-Z8X_a|`auU<ogQvArnKjLu(q@D
z4}CRsm0@p`CDcl(n<;wGaLu@yD>Sr+RpEgcUro*5gR1BbyM%C}bPd9}i7dkv3LBv3
z(p(1E!tg>iTd=R;p7JESJNb<{GF)oe*j9|it;WKH=_kjPkBsQD40UI0f=Z;hWiHlG
zpVsf=Lk2-*PinRsMtF0Kc4auSs_uW@M~~8tbGwEaERGP*VhvFAktA%*YW^oC=c4S2
zZwYULG+M@(g>aJn5=d6-^~EG#`o0;LFh!-&J=s+6(x|QWQ3nUST%lvItvfQox4cP#
zcEum*QTfq7^TCY=vXX&O-laDOC<b8F(}&-S-gG|WK))X4%`s=1Nc+&wRy=k6x#a4c
zx3)4qr6mKD>LuQEF;%uuZKPU9-kM9;3)AT2d@7pm{Pn)<>t4iJQBZ%qb3N6Msewul
zEKq$os%n>8D3b^O@^j4SWO8rD&x1ckjemdaLoX8b<$U_$BWl}!yZ)xFH57FT{p|Yc
z+fPMjewNk5KBg{Oa%gRfujzTo&H`9>2Q(?^3*KGlZ{N!1GcZxQ4V@j?xkrbAY5((X
zTwZQ*9Hy;(46d!N#yZ_mxHhO+h%G&8jikM})(_hZ&ieU00)h0F*>6R9%7;2e_ctPG
z<}B_LWbLf)3#&D}>57>GbQc-t=3qe|j@!<VU9Idz7Ny@bwF_?>Z-b<zxy`sfW9_4Y
zwd0*`R)LqoiY|=l=fGdBN^)*m^`X}gU$5%uI4lP)3>AS7U*5fuEn5Kp+LvSU+=~;C
zX|{jU+Y^nx2jx0ljq9mLfOKr{O}KqO{=1?#+Aa&JJ#lexqneEny4+<<vA#W2d3iA$
zbQ&7+vPaU+w`KZk9G-L7ssIaG{LXtlv@G>@o%n7SdQ2I1MPc+FP%?dHra9eNp-Zmn
zq)+^}JwGj(PGe(>H&1_KeouW>NrG<(w2sB4=a?Mu;%r}3dT{1rr=A5jV8_$YCw!|q
zjs9B4r-LDB>sVbH>60xdu{#c*ki2yh58iI`M_DU-qpYA-_CGI9(s!*c%|#_T*(<M_
z9FEXrv8p!yl1y7$@>usZ;=8vxIv&WuB6P+b&$+uFBXf)4;ngAU`8F^0l?~mWreN6K
zz4!)fc^di-_EZubKX-dkvCh5wNNr-0r;J?g4J_zs_-4KFEy-bvy-H&TD5k(N=a#=2
z{0!}{54ak84>67*Uc8POXs)I#Pg+W!5JC@ZS>U!(+`i!=_Ne;9tIy8+A1=L^eh-1n
z7TJZOk+RqI-$fZZz4SVP-BJG7P^GT?WnZXyD_j2is|cxa2@lUQ?8jAAxSMw23R1i1
z_cyKnqB%iV81V6j_3rL=_Ald^hMCT@MlY$9x+`xy<;yWxcjXgO+c#RqSlIcVzRCjH
zF8f<>weB9XR<=oJ&!T}ig|Sul+5YJB`8ndHaM$47D;A+)`+hiXWmbt(UDhm#ezgJm
zQnp5ZrNaF^q1BjCX5rZqxM$3*aou-&726h9!Fu>2Ft{#&Sg}ZdJD?(Tvi{kddfJ6!
z$b{uD-<?lYM&LZJ(#QeVr$C6q+kZfh1b))ZZot`n70)!782042x~$Q!<ouD)K|?p_
zgK@yJ&?vZjA^K2tH1|Bvjw}7t#fepCx$v=x(hRJS-p3lRI=t-mn|gL_J*(?(k8xw@
zd|yt6V1R3Vy}k24zr`)$c7AYun{JxUNeE8Mfo&TGTCLY~?Lvkpt)5OQda|Lg);HO9
zzgojq@B5j_y~v<GRX?4`3J(7<%st1sVa8~7ebu<)6}R;5sDU*!5Di%ayVXHc-v~<`
zFXbLj@VFy-wJH_Ng4D|}%BJI=4NUiV-Au0Bc>L%a^9L|ClQr*nbt)eEC~WNy$DhBk
zt$#IaqjD_Nhvjlr_4`EzJQ0?p)xT72!y<Ij^R*4P-fB%F^~%pZmsArCY9jPe3&8j!
zw(<=t*rOHw$mM+e<;f{apEVp}CZM(!qJOh3kElkRb<sd**ki#m%9xw?qF2^0-~H1H
zRJO`_jdeDxPzrBtGfn>et>w3d%j?fM=O$h31N+=hV5W~Ry#iU|(UMXmH8tAD_4k{L
zS(AEKtBlU-qrza@SZx1oEMr5Bv$46GMlDyv+9J~*pQmXSxQA7BZ@+l($2V0`+v&fL
zuG{{!D`;+F9<|fMOK<1m1A2Nc(2`91zJv6#p*YozA>yHCWyx98NAt|<eK^f0mA-Kd
z@9j&<Ko1*y(S39l>4}zKp8mjo_X~d%Rv&P5xgw|i!K{Y<smJXKmhn}pJRvYrc{?-p
zQ({u>1&LS5pCpPuFmJ70=tFn@hk2yZw&$5t7kl!1`j4)h`*iE0(z}13xRM3un?z|X
zypGMxxB*-L^WmiTgc;;_K<q3sG4}3(U*YZy10-z#shyno6uWV9^UvA86hl>2X1`0w
zg2N9z)3w`%!t?If*lTnhR}2>ylg*dc*bH5ZhRy<U-he&(Czyu=Q0dR}1{9)8Vw1Aq
zU#@X$=g!;?eJXTT@|+z$^SJPFpF20N<@o4~`RjaZY}a#!(z2_T48q4nf6n(ic<1KB
zh2HeQ$7+(|*I~~V=dW%%-`zc(@_Q)mk&=?#b!@&Yccv8tCYhOYwgA8)@qPRE%by=D
zPROBh=ue-&IjCkSj8cWz*D^gjqwn6lnf#%$W<v*i;7~M_`bOiew!5V-XXdBH@9uP^
zu6H|Sv#Dy;*@dXeaH_{u24a3UtNP%<sQfySrk$hLleVG4l+oNcY=0$ooS&UD!d>%&
z{xc}0_R6LU*zpfw#wDay>h{_}&*170+jm5}DU4oE+lxL_8C_(iacHrx?6~+<`(F?o
zeSFE6m9;Lm>ef(cP|VFUmyonmeVG?H_D#a7Pd7kM1In@2GTbeIsa;V22wHby?hw?+
z`JS;5{WdB3QI{1xIq&NitCX*NnKQ;*im0*kcS3ujp(jAxC9vQ%<FCWe#$yqjz`L>%
z`Q9A4D&;p?(mDOUvJ<yf7Z!_S?nU8--Ey>Y?7j^50<p(S?xFClokCEl>cNU^{v2<5
zFqm_)o&BWwd3Sc1<acYk7-{t!U!9NsI+J1-FBik{^9_eiWc_|w$N83Ypi)0wA4Lb}
zc+5~|1us5w_*|yI5AOo$A9XQXT3L_3kNqPv<o0BqUKA^M(UB<A?T83en(2Krdu@()
zlbew`OBA+zC(GaW*)PX_AY8}2l@<JbPmY(ja*7F(_5^XBb<6B-*d3-c=K7oYFir%x
z&E4OA@w2u-&71j}Qn=kSKO|w#VnSP4U2ci9m&ZPM{7;joh%X0inA9~Y{-vMO^!Z(w
zND{y%yGxbdj0l}{_vI`e>KrkvY1VX&P)z1VHZ;2qygG8#Xi3-ty&X{VyZ|>%PSj+&
zc)0&Q#j}X<izScd&-dZXc>Chg&zY4=S_>!k&e!Yb$Sq&|-d`t-Y4uaW-3V2$^4uC@
zSxQ?ILKRlq42^~t#}vRi>cNlRMsk`5eeUjU@zz&BBv<z~%A?QSQI*rZlqA_a^WTjs
zVKPn>;ARn9&s(f-XD@9?*B!?mARF?xYpsc0Y!0bMLkd8x#qeZek4Hd_P{91k9*y@i
zSzPse`IsHzV)r`PT7a@6$7XzZ%S}$Ve(R<4+F!?mX&A+dzPhh1yd`JO*xlZ*p4dMN
zeK>~)jo!&?WMK*U5sLfP9T$_gy7G05Iqy()sxFdt4>3P(XmG4_0lloNhk3;^3;R)#
z9&movW@v){GW%eg#Q67!-=2>iH?**|Kc9x8(F-@@FBieD_zwk@c?ZsJ{&*<Ou?@IY
zx!vkVXA9toh)tGvV;4O3>Gf>$nY|j3H0o_sz<KYJeDEYB<tJZ4eA&+`+am8AAI|)I
z<u`!&!?e6tJ<K1}=5t?jTC~uIN~7=2WBbdngr0q0?SE2f%St&`IX+DjPO-OWB9$0h
zcyG<~yp993qTp_(qD2b!TACYR2>~4%3m!Un)BDuV$<b_slDx$0R8FbobK#X`Ebdyh
zw(SNb31&HA&3PS~<;-~ptApFLSNB}Hh1ANNIec+*sT)7+_sh#9b{qQ5345aZVz-(A
zh(`RE9a-M2rOiHiX^z<qt**a??2InWDi6{<%!>XQ;P=%`X=s*~o##mn-zLsD=J8B;
zei<veH)Spqn3NB*j0W60pCS*Rbx`!|9+qhMawIa9-+*^EKvH8&&#wa_dO;1yDdyud
zZ5Oa0d{ZBqkazsr4D9&mw`b=~4*m$~)Xt$=UQJ<E#oEUm%AvZJ9t3*xt4=ipsZ#S0
zz_IU5S3gR`j>0uh+FgqWT-g9GWIyn>h9<+pPH3%1-hBJGcq?@CG!z07a{SEvC20ES
zti9^3lu{?d%ZTKv-bQ8A8oFp?>&-cqsU{gBzCjJuZD@Bb7;t@0F0~Od{2{dS4D<1D
zhwTk)zBG3wK#7z^c3Y7zlZ$Wv5058q+9#aa0Nr-&d+Vdq?5`6H;PB?x`nmA0=cd+k
z)zHt!0+WhhT0PtgFp$;RwiK-R&RG+=x9YoFDN%mtLxoZFa;>m;{EYnNsLM#wUVehu
zZlvc#lB|17@A3`Md7}f~oHdyXB@+n7JK1zN{_?ujgH^Yj6EOKhNGltg+NlKo)p#{x
zLC9)Iw3V{J&Ka4wYJWkPB+r*_Z&#PN7r<HS#;cV~fQvMq%Et>#=2At*k(iaejTAyQ
zvEFp)Oid5o9GwP#op6$+Ja|!9)_HY;qf%I{9{x#kbP2Qje0Jq&LN<}uKT_N=Qe++o
zRAhfqmr$bwbDl?asu`h1#WvU9CYf4aZYg-_pt_pPu4e{Z?~&5Cg*?CSy&TV5YMm5U
z0<TmT+4&;QJsORn#8MlX^Y&CbtspI;jh$q=^Zf#xrmb19SUr4#qzOQTu_r4tlOi!7
z<(a68OL#T+b&K$LE-)`pxmkkk2{T#(iu^#kM>38&X8|p*nIL|;n(jH)N~v%%v_KAr
z87Vw?xac!J>pgRjo4Uq3zN;v0J{p#sbTg_n<v}Wt|CKhj;NL06KP>DTuFZ}mg#4Zv
zTx@?S#6;={cyYFO=@0)>FH)BI@Z+%iUvsGYOAkI}IziP=I2hCfYh}E;mPlB@KBS1+
zyG)T#$#d`0yV>@N-tpZ<X)Dp~l+k2zeT*4==6`%#b=&@mD<_C?i}{%000l-v`$&JF
zg8Y?aiFjZi^Zxz#)@3W)dBD@~g4MmyXW4xNIMWz&Fvu&1P{%dwx(t`H<e4@urhJfQ
zGL2d(H=PV~k)qYTY1=QkARZ6A9iYeK4?dt`UsEvoi#{AvyJwNTPJf0bzCTFB&cHQG
zI@qc=-n~_>W1S+YT8oA?YQ!<EMmlT@u^;jN0Wu7{s03bA0LzZ^CV>QC_k#b~(pa1T
zF!0OPG=#8+b1Wx%wbL5X9@^P{nI>I)CBUX1B1Hzh%SfAE?nB?1X+p^L-;X)-d5_;A
zMXLZCS9jxVY=-QTln7M^pdE^%b4Zt&%c_dDr|=$6X1(y|Aa1iYa1+?o?i#)?)nr02
z9qbEM((hfSO2!!<PP{frrWDi3v)Zu8q#BQZIoJ*VCxq2{FyJ-y@EB>eIoRltsmE^q
z7woQul5!}759}TzJW?L?lU8YbTv%z)upAdeAvM!{LA=l!8%Rov87v7#4(jqo;9qwr
z<lb6Lr#H0N`yTp_h=9?~{xCJb4Y6ytmIp-?hzSO)syr)#K_y9ayZq?RKaYd6tun$Q
zAGJ~hXQ7{?p}C~mY*Ff9B4IX<)<ZuBG9HpXycN=pw%mvZ#7kcBX1x=)E&TB4W{ZAP
z4QwEJ&Z7SCZ`w<;b{+0fCRuyw6;JKeQy_zoPtzN|Av69kAW#FuLzY{)!kXasI&5bA
zmT=vhEflEQb0ry4I@p*D$*UjJiZa?wQf|o_{QN}&-F^!dT+vDiuJ+U;t)zq9hmmie
zXI2SopzHj|FjIs~M22M)MOFe%7oh3St!oo14y>~Q5xB&MZaqO7kR@_>O{n{daQ&xR
zxGP+Tw~l-vRHKHN-`E*9EBO4o^DFT=HUc2oQ+ON6QPmDzuBYy0M(<6*PCWRWwE>#^
z@{F8FLnbm?x>iI(bzYqC9QgF>qnq>KsOMjl`Hv)H@-)O;8>IY%7WRc!SMx)+HQU(8
zdy;xJG+k=<1Eo>*Y&a>GH!uJBpGD2bM(ds!$Ey_2w}I9YYv%JdbjMR6-UKAq%b4WF
zoJ)AZD+9LJ<)Ezdl-c#FaZQseZL4=Ye;&Mn?1QbE{WxYF3j3uKiWg~f%^|yOis#-~
zT*G8oF9;2TmQuKt8*7<{KfSAE^RFrmlts)_@#W2HXY1ANOWiEsCbBRj@6Rap;MC~z
zx_ymX1$1N=-n#W?^s=hrElTq?#DmXA=Y_v4p2(#ie$H}$CVOH@mKlhoUaufXJ_K8!
z^ZAGrlLKm@FnSFbZRs^5*NZ+)$NUH)qkMFp`pc^D*4YM?>XO7q`N*F2`@QJV2a|o5
zoDj*Gk7Kr=lc|R?aw9T;UF?8<8~j-F$X`|-o$Pses_sq>tOk_Qk@2mG!L+*j&OCy}
zRq0>%dw{YMnRxLsx$ab}Yl4yMY+w3~1Eg>5fZ_$_{gA~qn(g5vFmEHFonpB3<$J=g
zBUvC<isD@-+!Bi6Fh7#Lenjvzly>WJpWl%)!aQzf`*NHsLo!5nZG@!!Q49IefvEwy
zY!7L$)!2L~?kApB&w+l(6MZ7XO5XqplZr<eT&V%*bKU2JT#^{@b1SzG)LubLC{5?~
z?3by`c?;v_y+fQk$%(e0OCYfO<3z=Y+p%WlpLX?c@z&V@42D&e=$(-q83$6X5ZWiW
zyQbFnWnZKaSL^oN>TQG|1JFLtD}?1NUJGsbc)+Fpps$i6Z>XE1xP$-~S@6et_~aim
z1O-p`N)i&B!+d19TX=T6qB-NkhbAJXFTr!A4IVjwMck)BwL9PMoa3d5>TYOpMaiy`
zuV&ml5u~<_x6%bHPBjb-irZPG)c5PV#XP}@3%FN;xn|5~q!e=Cn24Ck<GT_JKU>7y
z5YZ;w7QB$hi*5_OIg9%Lxr^%=@^0y<dr0UR#OLswJ!F<xh<1RA7^G@kk$bMeZPH(^
zN_<Vi<a4}BL5b;2N?uYDE%ClY3dI&$6%7pw!t;3A2a}ddliJVmmQ@(-`*`uIBLg5A
z(nYnHo*}&Sd<9GBPcvH7_rRo((AHd};^*pKJ}(U*gm%}51C$jA)OWHmGRzvcwmn&}
z6-q|kg+5^263)!o3aD-98b$wG#Pr2TPqQ{D^eoI#P@AiO2bqq=yJdH4qrzw3oH~nX
zS2mJ@Du9pG!w+~QZDMjgYe~B+1Un8j>iCNvyEpV`7SAnLw6G~bAuj|GYe;Gft7?Wn
z{McV7kSB>*sKPI`jnK2&K8x#^k2KyHAbh#3g<Z3s^yMuqC*SH+Jllhab1D`)?w&%5
zZYPv{8Y<*j|4T29LIp|16QoG5*9M0HxNTTM`Ip|U78zO$#L~&$?H)PlFAK`Gbf&-#
zHG;2LY(ysZA`Y~zJ1G|?liur^eVDNow@^tX^jlNKQ<CQ!u+w?hvciyOwpNB=58uw=
zg@F~V?0b$Rp{*?L68BTxO8eAB1*vAMLd46~5vwdfRrV#5B)dR|Bj?!5=Z6Xmtq$Rd
zU;#=!kUSW`xXgU4<^ruc4SnFTB=6lc(x*lE$2C}@9{$P`_<5%1ABc7u`(6*9pL*V9
zc3+cJ;c4u>fQ_Hot>{o6dhQb6h(&~{?(pX9%O-`SfS9+^Jro0!G86s`Ib@qe^TRn*
zk<|$FZD>(tJ=LAb{aEN5L0V9xm+Uyd(R1Dgao7vt&K)v%NV~Ol9=|UU6C;@1mX@v^
zDm<^whLa;!rPRY$c?7DVVItKHQwzMDKJr{ZyQXh3Nqh<xF~H_OtspDkcMt&(1NY<v
zY*a+MrhajgEo~X7AXC8hbGxOvFZgO~l}Jqsv{J_GcCx3|sIG_mA`cqdIty!&p1uhx
zy-8xNq+IH#zm6|Y0h^$^@?k9$15%4)Ozypwn32<XYL04i;wQs|KpQ(PvdE8gpts%}
zYxfQXOYy`Oc2Zj^e-Xia9Wz+izk}IiweOHjvo+$8kPoLvY_C;X^z09TZS2Il?QXg}
z8uF=o1<mv~emAc=Noc?wF=sE;xI#jmC+!r_yugcK-00J>JNqt>7PJ+*O;)bBP#9Rm
zEY1O*^*(85x7Mvp!=sb)vAQ+xmkGCKU@}75wNK_Z#n2W2x#eL;>)~fSeb7)9(cK1*
z=aW*uf!#4ZcKP`Xp-iQ|q2Z{`Kb*Hy#!g0jS<YkKhvV&@{t$N|X_IEz9U8a`2^Ub`
z!sUq;WDMMmsyCICeHY-4(K2gAmkV{0@uz^AWWFF5FYp6cj}=GX;b%=CNJjc_l(Kd7
zF2CF@2{@PE>l?4%+sMn%0v(jGL&byA<kowALD&i_Hex@H(EgM1X)Nwa|31~JH|Q?z
zd&Y6%N{sy%l{paZw+(7Y+h-U>nteB7j`>?3x@}|fh9s*H){*Clyo;TiJugL;8_~kf
zjdBv;DZZ03+`K^zh)B&1dL`_bLYRE8sE)8Bk;VCa;{&}GHiHT)p%gBPNY2F(UvNi%
zZrw&AMY^NFvQ1jm<nUJ_yg0~+?L`^9UT%=LAjsPRHM|<s))y0E0`_Nn$;yw0)voBP
zvZ|8DfiBR#a543*Xn<`LjRp#owUOA*Xy^|Xxa32(nR;$2rR8!%vp0YiEi3D~6wmOI
zyP+KO1(AUBC;@a_mZ;KQZ4TY8V_B`ypE)!il;SNJ6F9Dc!yC2`Sg9W|mx;ZdZz9)h
zL6G}U-S@XZF<h?JlT4euE44Gm<c$?U_HL-*TB1^ls6Y+q^=xDsD(!Gy*=ueIRYVsj
z6I9=zzSRh58l%zX0jkep2?tEJC7{<%)Gj0_b4KAi<R%U|FgM;6OkO4r{CN4p&-2$p
zS9w&=N5^Hj8}aN9sNT>_3Gmt*qQ*GOe9YIvUR^)8iGS`O0jio0|4c}_l#9bM;;py+
zqQ-fA`nW*7n}VyIa7pe0aP<UsP(T})kX-Y6#KR$v>R!#gGvdWja`~xYKF<bE7hv;-
z__ulwBO?|AsjWm&sp1zZcp^-XzL7%s;!W(SY^=}~!1YsRbis#|XM3yoSkcS26H#gj
zGSSqMnqsD<4C)?T8Z&rni)BDGWRXqva$JDBip4oB!F++CrnJykP?*j3;b>$#4a#uO
z{o%0%@c8y#Ze#=nY`+d2m*zhGJ2a*jvdvqK5lxi%4N<f&r=bQf&iR2xq7v9s0^bvD
z(cmK&mhFC`Q*mOld}J<>2IpHt=XpZig5Ht_HtyP$x{#Qcm_vPZEs-V(Fn<<Yzozq*
zc(Lja>2jI(e1V*i9pUn2uDHTu7UG6%nzcf%c^#15bvP>W+iKj_yS_!QB1{LMql5jg
zXGsY1F9$g!7x@CpXxzzDULOTpyiE`F+M~t*y+Fne$czI2a#FZ;d{vF#;Fl&6EZu<H
zcvBxDUlc~4^Is3baTGo~k9r>v2%-xhu(4L6Sa$ZMWM$vE4uO+51X2ZsiRinbjWW=@
zfe}TlZ(0wx;xTTEvQ8^AT3Im!+Nnki?S{1fJ7hIAM5%{Y)1K=baZJD!Z2|CA=F<yb
z>~G2dRRo~woRandwiaVQH$IoBmPS}kM}g<B1=6@d-K!U`k$3F&_g@<bE7}bGinbI0
z2}R@#7WYB>^ShQgTZm$>v2RWsKemW4g)HnRup~ubNm5L%Fyp~UjzdM{xjsDm=Nh>J
zl(t$wAUBIf6ZDRcjKs7BH|9ul_w&!u8P--O@cVjHJt|Q&Mu1-$>?49ysE#J%AQG1t
z2*|Oa|7=!cL^2=Ot%K_PIgA_!(O7Mxe?ZOyg_)!%nyjKIt`BTb4Kb##mmVgHm*-~`
z2l-c#^6bc%AcY=z3i~D2UTvh__K*gf#TAg7Z$s6iErU!`J!aFw0DY&h1oXC^dJ^8%
z^IZ}*uMm$T&qB6+pC9pqXd<YX)>wktOCD7*y?&aI-xknd-ksq#Q>*~=;Q&aaN6fDn
z*noE$aQ2US(X%hm@U?-`r7h=lD{7t~e7!UKoFCupL>#+=!yz_L%!`o<l~K6u^?OqO
z<iI<kl>kbpyQddS9y@X~@sDVI>2u5a$4}5`A%OH)Oe-2%JT-(%{cUKjAOp01>nx#o
zzU|Ef;NQD<4r%7#tm|HHiJF{)E8^jvO>G>=c*0{^8-eI+)$oZRJ{4v+j;ySaScBJ8
zwdZdHEG^R^5&}Sp<1LbyrkD_(@8rEXFzs@;gW6uE=+CHSgkF!Jt5p@nw}Cd3JUAR=
zJj-xkP$M@M#cMC%4FkzjE~|<afW8=La+GH`(IWA71`#B?@BngZJTl@V5c@I|U0EY#
z&|85QN_c2>fg0@J&*F`)hl&IOi%&zA*)aP$O@*K?l<DN}wa@{y_S|!+Z!c3D@MH>0
zwI>oTaT*H+77Nm`R58<YRd2-)#~ebHE1823m<2j%?s;%#{EaU=8P7beQpZ=%v?9VH
z0GlAIXc)`!Uh#-rmz}xEShOHO>+c)CW>@YBQe%K_8Ia!74cBHxtn-a6K5-tiqrEPc
znoVH1`Q{)zFUI)W*YZ6C%H5AtD2!gFe%eIrat6!0BObA69GC<?7=-94jPCpsL$aw|
zWc(8}Mt4ea-}9})uQ=fmB7lex2M($+DuKrBpk)HdmeSl#0iR;AT$1fLEJ2oWBTY%H
z05qh^+9>6P6_yu`8b|Js#%2ZS2>4POrG?r-4^#~iw3q#k4sFaP5M<SkOM|=$3|QY=
z&o7sIX|w~t>aJmSf*S*60U>kj>hF^a*+3aQb3TnYi)eOG@r2swzJ~~qiMiA!z(Ffe
zt2vr1%N^lqyrGq?<vzo|<{2#lAG1OE>*3%dbB{hA66ojI<xX7uyUq0E9CqPJoxY@O
z?#kZUa*xU6x?}##E@z1XG9aa3392u$R|@jy0`a9F#wrDYTOR*NiujfvwUeM>AP?^4
z(O`on%W$ReY7M03hPI41ykJp?#Zg>}e5*yzV{w+*BwzMW9N;S%ESiT*2s!wD$BimX
zAa*`s&n9>zp(M@|D*)X!8ncSBYh)ec<tN<fYUN_fDGr;6{9}T;2o|5n7&lU3lmk5h
zRCgysspk9YM$nwcyBS~^&g)l$!U9V{W7Lo3q~BYCH;jPQ-?xV@^Y*Vq%uhqAO68q+
zw;9C2m`&(7h=aH%zVzWp(^z}(l*Sb?I!oF$YcLqt^$uXyMev1zj1N44Ud9C)YP3ua
z9|8i^x!tR3=`=BvkXt7h{wqSXTawp70Wbb~a(e;!qC05#@V^t0<(i5V|5T(1wXoYO
z1XRS|Uv>u)1StE>t=*1lj55I74k()M9i~tnpz8ik{iqxfcZ_^=I3U51!{hbTo<HWx
z0uB3x3062EI`7hqy73`bju?JX5p<#6`fzr;yMT3XfOU64x9X{HrY3^Scx#K@pP^vB
zZz6cp(e<etnUWKW;Ks;6+4n*MA3?9Fm=*1ZQUz1s$KT^fwK@S{Z8Y3wPjdnGd<$C{
z4L#yT(Y?M+JkVUX8!{9SZn{+YQOdm|c8M-&$|clNU`FM@T-QTR`LlbKG>yWlqt9cQ
z-PuE{90=GR;|K7@TuRCeG0kNl%yqf4{hN;Ql;%g|o^*aADMm5i$`#aaE}tga=7xoT
z&2n3qOQ!EiV9d>5^aS|#0lUMHMi3}?*>T`Hi~~%3dk77iQE>G2#Z?*tzIQHNTt~!_
zanSD7pGg8ZZhp?WE`d4u1nOs5FZm_3?IMwSKrtZXxL-)%IuOkWD^;&o5AXluG1d)7
z#!dO6#E>$xYLP?07;f+fF3<h5zOllNTY<;a7}o*yYoRLsMH)nEJK(KO%BK`y=dY)~
z(rDX(WTP-wp^fZS-Va^s&dqCEN67xZ6Qlr4dOKk442Ibji~m#@DNUODGw|z)CXM|s
z=kd%>82PfA<m6K`Wo57N(fi*-1-mqOsBAL>KS_wp0&<W=4~Vs2HNxYK(&GK303&gv
z%C(mfnhjdgN4+@8jQ0<4r!>~kjUv25f*Jknw9ckKFX#kT58`KZIM4=jRa-PJ2=ZG1
z3bm05C5I!QaT!c5Uv+|*dzF=X9C#M!?Sh)9p*>57&*Bw4^f6e_>28AJzy?MEQM@if
zCLRste&7W~ZKPIQv*svnC<iyjf@`h7e`liUlEt<%LhTnZ`D(AyU~DKWh5%Y4Z-vOF
zFoCG4g^>|g0H3O8t`v7SFLG*vBwdHRZIN`?ig@&A!#e!iA4lVe+UAKkp@+>`)KnlK
zeByfOgKDFRRKbtRQ(bqYs|_ZrQzj#?6CMUij5lP3SO=;)e-1}dXL4f5ri~XHHCr<t
zgV<;dG?|1Yl<MCw3wrmXE?G2BloWF!wx2(yE#v!gbOsdo1_0${v62L?ul~wMa?r}&
zUiZ3arXc?!AWS=?R#~&w9T$SOy(}q&`KWq)Y^av5oWmAEo>$J-)W&Ipmp~8>dOj2`
z&4V82n1hFa(7{Xnxzr=ogZK#~uC2>hM0OdrL5e^mN4+`AXno_H>XL}zI~Q8q2$$H0
z95&LQSpfR0pIX_vV7PR}nBdV(mf=Kby6BCArYI(jW&yZN#hM2)+IW$29|DQ*fRFD|
zV`Kqaw}oEfR)D6qx$>4ZKfbYPlO*m`AanNU=6{N!4U{li*ioy|U?B<O6l0(iegL#^
z4rsv;O&6Vge9_l0!LD`?vGr`9re8ea+d}a31gb!nUrL7yvHB^b!c2drCQfu4`#5di
z+px>4cpP8D66iZz(`SqEMmZ5rf!>36_K{~Q06sVUqp?s}KmvHT9NL$$T_MC|`PiFa
zA4hIa<Wo}ZKC<Jwu;~&x5U)J?^Qm6dhF5O$LaP{JV@&Ss-_G_;aEW`nD&#r2^mXbN
zzzitK1+CDM$dQ1BzHe8h65y&3%Mj|+5El3d`lQONs-@+PiN#mCb`z}Rk_l&#oP#~s
ztN>;QH5gcxx$LKt^7o-Hai3%_iHrcp*uYd}Jv-U*-S5Q=UW`nxaBwEl=N1&4riizL
zR`gLHjwqP@De4)Uw6D&LZ@lE4k2zXau0D~h2NA_^)()r>_})8EUSuG8@?_5jlo#NH
zVka%~B8|2hsJjSbpB$-3GJID_)W;S#u^U1XBNyewG(jU8RKNiTKR-?g3mU8F?bHx_
zNYv<>os@u4@279nA_D-uPAD}VEHHT3`M;aqK!#l+{FqUa<;LXDZs-v&7Tf{1TDrYN
zyY>Z>yWYai17vy~h%S`H)!Un`Qk>AYlMvlITy!Q<vO?Pc?`lZx>e@A9Xohym)6q~p
zfOGNEHULG1wpGt0U-|A+>QCNVz+K}OC<5ERe5TP>0ah!pU*|`TXokO_YvI91G7g(5
zLvDJNRi#zlc1<ho)ed4)P>oT^!0X@1KGt9tUG93^gV2?{R}%)(_$A;CJ0aLMu|#H1
z+lf1bB39rHj)lLX4e?QcfwP6bnZ^1Aie+jKO=vo};iO6mqUx$A5k(MlaZUPiLiZM#
zZ}>i|8(ihaPaGK39$;Pn0IAYZ^c0v2&G|~UW2~X?K*_p*iz-awG_;79sdsoaedYO&
z3sh=_tw}i9WNu5M_PK~2U;<!6T#feDQL5^YN<;icge?f_M*Qex9l*+~Fe^cR7O;>}
z#8{wvm=>q$O;j=9jKf8`eTOJO#TP(m_AL$OEAQr4c4#LXC4blc)WsG}crMc~XyHP(
zF|w$QJP!#+sWE;*d`)u$2_C=+gceZpMxbOl6yDa(uMj8o`eW7&9ex#}WCb;(7DG%?
zOcPN2smt}fe<^;Ap9K^(3stvh&Xi+ZWIm2yz;yN6^hfrMS2hyo1M8uD0Si|?0D1uv
zLTE|sNRXm$>Z<?lD27bTD(~_BSy)_1BrrdwC#e2$*hOo0<s5DZ%8v+dI<$gNWLZnb
z01);fqIT_&fPT*I9-Ym0eR+5l=-uE4YX-PY-kuBM9+1rnX7KFtz9q`rUv4FA#{j97
zC<vG{SFB(ms_X7WciUUL^<`CAK5yIL<qpCoBNF7Gjt6-U0+=c|8Ic0gSqh}1iPWlg
za~6*399Tlw=>oKy2XfVrG-u@ProDHQj4GJ#c7s+LPL3r~frl%w9m9*W0Us=%K8X9z
z90?&%fyp~CyCw|E2gE~btKaNq#PDXDn+n#Lf>;U`==b8x-djJ*D0lk!pVj-l=ySl}
zj%kDI(LKPUPXW<`k8a}qjtYFbtYv;#@(5Ai-`@!B=lS~?%yB6_3Eb{bMZY-;ARCAy
ztz2pau#`2x+Am=26C+1w(3j9P367VcC4%jd6cHIvxXLFmqeIgnU8jqN7l{sDGM1q5
z{M-gLMjSAQ9N0HOMd$UEvD2jLS~I$V>QsR0tRa}I?(EtyA@zeiN!kQCPEixIawy8A
z=+ITV;@2MWA)>W+>M~Xg!U4GDDzyXYk?tN_7o;~toU}^n0W2$_9vHC_keY%8t*^YD
zpdvlBYCPX5ipWR4eLx6W=preu77w4SXk%}oU4tXn&=deUwh-MCGkKJ@9Gs(pM&*5y
z8hp|jETz)xh+z4k?!hvzD*fgSM5HiA6?jT-W(FJMgH_;Y5D#1A$TlFEW}wbF<sU+R
zKFvuZ%xeUQs!>c+5Fh-+3|0kWy`HVMgk1(9wKEHZ&eugQ1zcocqZS6#c*wSfq{a?P
z6-<r*+8_rhmIi%~O@Hev5;5<kcheAvdFP0z)@M*sHjwcgU^T<$jRd9UXty$1gUPYD
z+xK=&mLmG(xyLok4@mSJkmwl%;_D9l&7ZcP43+KdC|Y5ol_NhnSgQaA_fHK{#!51m
zO$|ojEe^7)B=6Jo9#8@<`+@jcV+mLC9Xqh6-PeUd9(eCXV@-i0fX2AH566Mgb<Gkc
z<7y#(ctK?CJd7fIjOx>FL&zir(E|V*w8X%K(%|L>N70?E3xtv7Vwi}}@>SQx9Lzz8
z=wQn%a7NubHu7ef-`V@Mp<q5ung|g^#aAYe?=5=e%&L*#A@{!vGkiE1i6x^qeHIbQ
z+lubw3Cn?H7*OqrfA@rYe<7v?M70r!sv3cO*5~IxTSIDQ8&nbTPqq)=B48=I(J=b@
z{FcnaRzzb1OmY-Dhni)HIDnpRT?gA{!5VmPr!_%j9#X5Q8Psq_ktR+MDci%E2=vT$
z>sj?<1rJYdVtL|t63nG_vMbYLD_OpOy88+|A`@6u`#rN)p^79ij*tj&p*et*y=X~i
zx7Vs#OVaH+GVTGl1D7KCcS7m))OsF}@?%w*IvjZL>TQGsFuL_nh4jRM))voM^`;en
z&b5o+t&u(p0Pjly?=KN3E4yjd=`d1lMbs#uHhgL&_CDzAW#5-%W_yGz1IQ4s@Z5b%
zx!u<enEG~G5?qrbn&6Ho!<Um}-gDmcoEkrwxP3XgjNX{^IT7HXQLqL(sDiwAY&0Ov
zv6C`(>hY*ffPgUJkMDOF?DqCbOV;Xn#%uPT0<cK<X`tCAL2_;Skbe`e2O;T!CIe_T
z8}wp7v8tXYX!l$Sx-2+iL->X-ItPs5{RuEOk)U0h_TRB$&?f|AZ3)`Om#2@xeC@`p
zB0kW)$4jh@Jt-Xdmv2=nZEpyX?dMZ0%u+RS)$2<BVNh1gVFo{`Tb%mRfHVvib`Io{
zPS8nN>O~)qv`wixKy+Co52MKxlAdkN>{h`vtph#l3BkQbmI1J^1GFRy7;xm@F2ggH
zp`_+Z76l)nF}b>LGGwKPJ|`TCkLR5)-ro%k^9h%Z?vT|2oU}Ky0|bw0dG?y~R_#&M
zd<hQN%0O??!5j*f9XeEXNS08@CErp2UEJ{tkEbz_pKC|5JC~DcFNN=u-Ws(OnRv&X
zdzX2TXj)4`D6CNM@}}|*(8nvtf}QhN)emb<k&%xTEa-64V|5wPBI8I6=MCHW-n<bQ
zqsKe`-ct&fM4r=-b(1LuZN6<SZ0115IIqgV5YR(=348&c?O+AP+kf~}9U@8wKi|?%
zfJ_!36MS7^`(Z`nZL)k;W;*d2H(<tAS&t8<M;DP(jJ!%mz?7Rrl)8r<xNBB;kBcfP
zOYw|C6it`CsC-9p2WXfT<iNppKA-F+L<mdtZ>Ef$i5X}T7oCeM&ztkwT3Wl3PSlB@
z>8~PkQ5n?TWuR^4U=68ewD;Fz(uB|k7;rfd+Cy;Gu@Onr?>)>rjk|9SSxjZU&fp>A
z&#PIPhYIeR@?wAq)9DCU&=D+X13`8Lck|7|2);qbU@hbQyZX6;zQDd%OhahhWVg&W
z0^<!_12+S%nCGvn+XW5>m!sc))fD`6qdp~2K5s{SRtGb<Gi%H<<RVcaor=LW0zb#w
z&)2=_B^K|x#x!+^pg0dr2JuD<^eNPk!{hvyDhhro$B#7_6DL=a^Fa1JC*v%Mk8f*Z
zXANXD^7N*S^eAc8v?s9Ywg5V)D7C8W$6@UblEy4JIx?#sY?uOyz>nUXoKqe40fpNC
z*|5Eh{k=l##u46tDmsT3`9P`hEdSf9$LD_c<$vGDyDjpm#TCRka5Nmh)&hk+i@!sm
z=SZflpu3{iWp+vwgGzW?JKNObA@}nVax+yloo^ew|I-%tTwp|5*j(k{J>d?I<q3s<
zYGHTKIzR_iU>WYlEJK6d!}t|qyzglR3G<$>Vg+~IMuJy&kY|`l*cfncT#$0;WJ@9U
z>=&QRp=}fKJSzQLU4H>8*8wVR(dXfxY99*@4-$tDk{YOr@I{r|Gk1Whx}Xrwc%w}U
z3-m2MnCeLH7T3b*?_l@*pnqAqoRF0UdVueuN+>WAP$)ZWhd85mdy1HZ(_LkzhLg`h
z6V<o${qTgzy(6a1f39ZPK;1mk+s^9x(nFp#+-9bRfjM?yEhp%B=}lsx+f1&!1NY%t
z$;d!pz%@u?NYfs&QDy@T3j~TDUnvQ$D(DszQSYiQC(YBFvx_lWXE+aE2?WR)cC)uq
z{sik(tehvG9IC`UlUZ7E^9~}7czgr5h!QYOp)Z@tj9ja|>Gq6IjKI2;psHe8sbBIu
z0fcS~>F}MhS1FVE&$}E7co%rsK9D`U<JYZd5{h|qDh&$&8>J8xP05S1|2+Np$57f_
z--Cn*&OHCFOt!I$K!$0hoN~_)pZi$S45jjKqm)VQ042{9UEjSq>lkC1c{+lQVBska
zjw3y)1t$F<up|?KrSYm}-t9jj#SGmC;(9TF_#2?vCqfqQP}Y${_uUa6<2l^aDsf07
zs1<eqs6Vxzkq_OF@%8KWy7|+o-3&E+QC3Z_l)&Ew1*Cw+xg{EWyi0eRp(erHe303|
z)tu=&Qj>NnX;LQvjfu?1G$dQ@ncltEM+XYwVltrrm0Uhn4oK?+COk!224bvQKs_*{
z4BWI(LpO^NjrC9MY-I+<1k+a$2x3`4)}`|Kyx=8=hG|q_duXUPeT94M|6|7@EC3G%
z(|5dq7|+l&4U{ee*9oR}1X(Vz_qD9RL!HcWksSa_f-?;%z<lE$=C?!3_#fs2achCJ
zJ|*tmtklE6uBWf`slydNng5p&IDx+Nb^DzH#Q-i$&6Pu4F1m!RX8gFgs7_+K5}cr5
zzK*W6+d?}4L<LG%1L(;A<6cgXtYAkz2v`4FE0mRrCo%x)ra1ENbp7!kIs{tSmgDc*
zfbq_qE|3?9@oY`A5aBW!;Z*B@Q)&Of8x>fBRkt0$*WljWG~@B*h%tV?1^N-5<`}|R
zu>G8N(Wl`n1HlRQUqdK?@i+1A*^~=aVRt0kGnstV1kj<YfGs=!_;D2$C-BjJ7vFAh
z!aQZJpljmJ_=R7L!GX_|E3JnH=eED%D=_<DEApTBb;ijS$zM8d!0lju<MjE#;BZ9$
z(orh7PKg&6N=BPEg$VNOte3@=Z|;96-X)j{vhtL{p20G-yvwwk-Cu)f`))c@wE!(D
zZZgrISq*T{Up;iWe*BLf?z=%}L5y?)%gUiDD{D3x$^2pdE`|xM*;K7u4KT^n62Lf*
zY>t+Wlxfnr^Y0PRhoRD<#d(WO(fA9o!O57uh8p;YhZG3lQw2nkUVa`hIf1nT7XkNX
zr<-P2Rq?|sw3vrZg5ZkHU*?6tA<jH4U=Gy)j7-s|=0y)@JbVp=8v}&v%x|3o|3ii&
zvZso_3~~h>Xc-eBfvm7ORCg^=RI5fF2lP{FRzoIi_QdP(QRliy?FL6kQv)lQRZ7hs
zEf4Eh(<BZSaTcV$kiRQm4WP^vIqqJZ5XNK2H}eF$00@{8${8$3`@jXztRs)zI59v7
zp)+`;+iWx!_|)#D)3$VAR?TSM`xC!5{WqSU$b!E_E`HM`kUD+o+30irE9cE>q{02^
zzjOfC_4H1vk68Q$=m&ovG^HPbba8-}E*M^*13&hAuB!xIDon)%Bg6)@?W;k<e~P1i
zFV1>K<9(Y*94<^5e;js4;K<ISUJLyJ#F#SkNH31$Ll;MZn&PtlE&TwrOy!~z+3-oA
zBYcfBn8Dliy27h(kYMWq_6ZD3_fTQ*DKI~xRXwvB<o_u;)V%4d8P88Fi<F#B$9m}Q
z5_*)f6APrNet<vzZkPl2L>frq)<ql@K#{W^UYy5+=4)Fi0Mz<ln@0M*Rti8IxO%2`
zbob`yGn&BteE<=r!UqU6+&FE$3+9og9`gjcrsqp70K2}fKKhHCupV-Iy<Yd74Nu@;
zF)^?h@Y{c_W&?ftkAt4bqk<}>e>y|}vGeBK*bEWH=XYhQH%+6E;my%xwEP^7Xq=wo
zuV6CZRZy?u?*gP^1Z+P}kZ1e0kP=_)d)XP0|C$1NnwO6V$i(($Yv89!3;ra-es5Xa
znKw?`v)`Mez}RY!EWkcm{>iuZ+Snen{J~u6Kcxpjiaa&z;q14c$Q%Dh5$AbB%wqh@
zt=K=12k@iew~IgBRp(!usrkqPr0_4F1pOUPM)JQVpD2K-k&eW7WoIx7u%vfuc_-I$
z;5*-TZUO!&+Vriu{B}_#m^H5W$7@p*uWhG*TFQSqXt>u-@%MT=NPI9h<+z*B1L9vV
z;CH<f|Dd=Ptg6M$Hu!2HEo=^m+T$uy=%CN5eRv&X-kAQs{I(EY5b0_o&^0~rZNiVF
z@y}kTwjwS_3H<v>pSZptaplT63YhPzsf*Qgo1*=-{LlHOtNl8-bw#{A<;?{MMDY@t
zo;)2`;0RxQ`V)Uvuxj%GePH41A+vrUO;Mh~g1~QBaZriB&vJT@1h1Qtum8!X!cNfP
zh=bYFqewM4cO5bS-y;6?#v<z5=7yH(%$>nxn7-WQjJ^#-2ZZtl?5)ewWCAOAe)GYF
z)7@?-)X@EG!Sr1kkTglA61dwrE&R~U>W|ZE6Q}Rvhw?{T#sQQBJi!x$-Jeptu<M7y
zB14AL(a(tz_+8}j?CC)1B~mN+@0Wn_`btgo$$V8hHErvAof!}DFQkAJm-z$0D8Ms`
zAw<tee|k6tU7{U~0kP?dtHET172^N^nu<IC0N&hJpBj03!E{=Nz=4UEk-&7Z-oX1+
zL`+S2w_bPOPM-WgduMs`O>q17uL4Tp{K&Yz>4G_fvG*c}4lq0nr_0$55*1Xboi69^
zh_UR@O8|C3@*W2QRI_R)j#=+@xsDSo|FHxMVKiNUIWirPf;5zVpwTlTC4KtNbHD{E
zn^)5{w|OlTEV$fw`nF5GO20R6AUcVvhq5k>E#UFi-t79}U)zF+Y^KEhsn=WqZlq2z
zoOK%07C9CKQ0!D52d$#hv?lQR_*5E;QAD8GL)}1;fVT?3n~5X|Ouzu>PEFSva0l<<
zye58I<LN57?nSp@c!Te#r}Bfv9JEASADymE&^T&~%u|^zf$hxTh0RPb3EWj=t=_z8
zt;{OogbWzh@blUf{CkLLbhof&BX87C7tTj7jvZsze>y8`kXnJEHOkXP5C<Z{5fyM~
z0l3=?xKkG9!LId?>IzQH%H}}ueePe%Rq!jG(>LDM$_}ITT$@gKG7e0FG^bBngf5q4
zs*t_<%V3wQ;#0f4DyJJOc^S|CL)4{1DMJ3pu|;T~5HSOI?usau)1l#HyeR}eA_foC
z(2keKmZM>kL&j3XytPQt&2PJq9?|I!0ztG29GaNJ;H)6RhvNV+|2e=!C<|UlwX=7V
z&s!<N-ldN^eZ5QH;-&<QgILT5-03#~KP&L}-;i|kQ89w46HF(MVc;60Ek{EbpXG?f
zZ-BTc9P*$9e3q*>ixn*PuXjPq$@o7HocyQpvL*|ja9Lv}bA)pKFMm{2MUr>`G;<Kh
zY(xae%;Ul3L~v>`WFr5Uv||XQ@Sg`glG7gmLF9jDzA)t(YXauu2WMd!0=r~MKm7AR
zgm?gApa?(%2l)UR(0%=E6Ivq8q=3VXKXSJ~)A^@A0K`{)XPU}-QRV*9bPgGY=L}KB
z{yw>a7z+~*fH_0JoNwXxiGZ-oSyc<0A-TVTUB%)GGPr7l-4d#qfrho^2XYSn>!GiY
z4tKh~t|L7ZFH^q1`&~yS^qNEYz`Ee>0tmM!Oa?dO;wwaI<)^<0ghR)?`-b~O^i?DN
za=9I)O8n;mXl9U>zX4iR>FNYE@%QDE7xFbh1ssik+aZ5we=&AYsK%beh`3-?TG(Nx
zC4+~k&0+YVqIW!DJ%Oy?HUD`~rSqQ$slm930PS;txW=g<Tg1=MsK)MmY&&M!!-wXR
zaf={kCloC3p9fC=eIu3sKW{7(oc<uri!Pbj>G4yvIUGN3TMxa$vn>QNJ@|kh{G!T4
z#ILPnc*%hkol8r=XN~`PurKmI54vf;3(3IH>Ka?J0O`4x|4n{yuQD!15cR1*Y7JY%
zdF!*IBlxzGag*5K1}w|RpGU$Ff=3r<;=d0b^G|;x&Wp1ovs3btq|l`kgxszGuPMeO
z4N&_Nr14LZCCvQ}mhjx3?CJ3SjS0@&^sfg&qSGG$Jyx2Q%nJ}paK}$vd3yZDs#-uk
zSP$6)D;pCc)Z}Rij(~jB?l1gn#e!a(RN6`Fzw!`mX5akR10bLW`QMCW`DQHst6B;T
zDiS0f0EsOF6073bStW)E%-=+IRB#Tc5!f-h(8^7JU7C6V`1te(fCY`PT{^F=)#}&3
zM0#Wzn31+c(h-h5;0Iw&;lLZK6n-7rQqc98RvI09%vSvb)+w@b)5#LpCUTQ;Hxm)}
z3M{lO+KhuoZ%!NSs%Pvm6~3=)o)vV>p6|txB9@(SQfpgdP}0f2YqAKu3oc-#<50h4
zo}LqQSN+aM@VhjTc0#C15-1Fvp$6l_bJ84(9(3O!_D>Ody?bw)PdKAP$8VRdU<=#8
z4xabfQSE`yOT_%Wilv!mL+y)k8@)FK39=!k&tnZr<i6c*$=mKo4@UcBPhR2!i{ADy
zOLJ_{b22*seoP0iM?n+s%;9skMX&fvZXa(q2Ok}(MEmQUj(lc(eze7aR`lxS9|>S0
zuT8+O_k^gAR(P+9l|!5b#dP^AF}Y>oeR&=6k&Zc;*6X3h!vMa8UUfFf1mD+%e)33j
zr1S4mm<&IrsLl*Fl>0{jTqgYz_v_f)bUVv6i4<=wP%GuZ9T7&8Nw&KtufLWAMQ}9W
zED6*XIJwz(Pg`hgC>wM>qDWLF7(=Ha?Pq<hRO<)V7(2l;*I+c!xl9b)J`J(@O^dN>
zp5?@?90CX7pyi9MRn42owa8XVv~w#T(vOe_mzZt|$GH2--m;qft?PB<g<GOsjw*nJ
z>tc|J;ngCsL2K89pe6wOSRTAL|0NQ|mY7s&oVXqFbsP#&6$9KQmhUK6E-M92|66G6
zEBJjU!AaFFv`|Tx8LV_f_R%$8PG;gZkyqP#XMXsSF=h&4_bD%ALNOu#?!J)Bwgl7P
zyV;&A&S2V~eeRq|ON7eAYOvAuTasl>;!gf5KYZcdX^=7{T9{Q@V1~_S!RRvXkMz05
zitDA&Z|wNqlZS?o9=UVde^&JhUNaDN9udRAyDh}!3;8iwva!dV^G$krkp3C)jfrH9
zq~tYUS!iUkNTT+(RNy%4NgSwmz}j6C%5nwYyQm{D`&Hw9n>O(*>QHI-n`TxQq;hdM
zvV3u7<dsXINBP!*rR^ZPpun8pHPLXFn<w}|FK~zYTl0Y(V&Kkgz1WY*@MRwl30ID@
z&R6QsTn|;1-IouLEGq@S>UEbda$|U(;H%-|vV8#hB{*{}+{0hEf_c&Z+}rpI@t4Px
z_J#Xa*v8@v*g<B30^b^s_nE&Mj=plhXX#pIl}?cA+`2BdpVmuH`v!Vss4heieR;q~
z2nf_4ssEL?`K#;iC7oGjJ$_raif6&>_x<UYWZfXR;^tnn5YsAt?3GtV^e;Z(4EMcD
zgY^>%+{VCfEp;-&E8pE%Jj()Vh<g91kO4>6dELq^{i#$F%$CgoewKl0^Ur*)!G(X#
zD7!QGYIc?v-SeyK*CH>@PW_(6JKvd|O8H%I&#9j#hd$5Bz3DL**g^>vTwt4UwW3_V
z9g*2R>O21oCZBnk?{}ph__eriGFJ8I<;}{+$~)xER=s-Z0Cez~87+G?fIZxBZ>{9Z
z-cXO2v*0_c{@Tq}5a(1|?(f1t-_T*p!7ZS?oh{(QQBZ3Ypjhk3bd8#M+^w)!-@?8Q
zx`v(2uJ?D|iC>g*TuKMUI!5@(mobju<+%wz<ePH`o=$RJjh5dMUf)7FBEntg<0X3x
z(q4F=IsW|1$Ldu3;{YD6qoKK1_g)Ih=#V%%@N3d8O=-b-Y_y)fVGbIZxr4IcU4PQW
z(s$`$89i7YXlk4(fDQRBKf1MW;INkc;&$}zJ$m>hOfz#%t=L`l?M(cq33a-S<AsVS
z<1bELpDevK>MsFoeZR;`@M#t*)#b`jq7@u+jXr)b<kdyl!f$s6YVNH-Y`#xcox_eP
z<=^%)yE`Ptcq?KvdSMGQ{Pe_P4ZNf)gf$`@V!G)XjJGiw=!I=x#k-_Ww@|)MmX5c#
zu+LeS?h3C+w~F1nHMc8VV_zm7|JM0%+Fc^I&5~b!$Js62@le-R7ejeVXfWc{?~h%r
zl%S_|LNY&)-Z-Bpt;5{#^*tZK?;c60vr;j&>hU8noYUdfC4+(5|EITW|A#6K<7A_<
zMN8STjYJ}3Q#x5O)+LqLO-!G48J9_oq*+~D<}hw;YE?t`*v6tr%0x4U#v!6wng|(&
z&QOf4=`_TgL&MCz{)zqheLm+rpXc*F?{oXU-*?)hna(B5r>}uTes#IS+xpwSROg31
zOoLCkjvGU@i0iH~Y|p*etrRW8SjVTR9C1RYlp|3-B~K^z8e;J*)Gb{av4``)a<(dY
z^L1+Ciz|)-u2SLrM<0Xb1v{H{b48pd4S?&BDb<ehqkV<vN`pv0S(jZRHt^rk8D`U*
zo-Xw9VzY8i%OSHA@d*MWM!HQE<82z3*#kGd)gI^ruw(Dy)E%T3PG?08>Iq>&Quiy9
zxEaxyTGi#ZC1=MNf^3(#ENigg^TaODWDsvC&Hmy+|9RIAgb~y9&||e^U+5xE$ztqr
z!fsvu66j8&L@+2|JqWf5_pO%ce=hLk5vvGe+7kok@=^rdw~!mXiczWv0x}dZ?r2Kc
zu0BqPz%(J@;Qq!^e8}g=;7P2qqj@@=&Q9Runa9op4`lI6a$hcN*5fY#M9Wh(fjy+-
z!X16tjR0`(5sE#?A)jr%V!sI1zm2?otqRx&BQ$+#M~YiuWEpVyT8S_7UELZzd%wQm
zwb+_W^hUk}R<BsdrEg0N>?;RGc!*PUJ5KTLj1R?s>jfX)oH@mGDTYc*@h<-<r)DCy
z-*di<?WHR6Iu;C{?`Ky(a+k=9q?F2g2&(X<_Sm5z(S_hoFC_8@OHV(FHhuMMU5w)T
z(1Y>{1dUnm5UV@T+iz8jT>j`~`_2;7NvKD4LkZ=HaF^=O+pbeLIBk+F<<Y0lZLt7g
zmJoQc^ok(Dr1wA`_=@(ptf!aCo;iFWEH-X_ITS`TdLvujNA^}4Vs&Yolgul!pfZ4P
z*_a$-lw8|ubq5^Ne6i4Y5xpgv=jYvpQiqKE?*T|>^y`X(X2&<UHE;_qIKDGZTkUu;
zupf{)J~{lb7x-6yvDoCWfEnV&J22dC(Yn@P`(@nfcgx()OvQ4-cn|zbB(VV`LxPn^
zv$cRSSOiShfHO~|pB=U97!0t+GREjTJ4vIOI!{iQ2<(zhAUrnWwrOjI&s`S}9zmjf
zKi+uJ9Nb}23cMXA=`7)qH*>pT<KRvzmk4)n`_An#Igwn%OVmwZN71?Bf!qIHCAC^z
zO$1z1bWivddniesGlHE4pmq~?$OgE*nGIa!GP4U4nM+)%0A;xU{P@aGhvgQaYucS?
zv<F569*H*RaVBvM8C)|M0>^fdvME4;m5cJ9mrWO(yMw5T28?}RWT3{D*_2)NHjWQj
zVR`hOFw(8j@)G_1?D-1Nz4!KgWO=f1K@1mCHMUbHGiHfTk^unywv@c4GQ!xzR|gh_
zW?RJ*=$<TF7sC~Y#EPrMsjG$0lK#5P^C>dNl*~sj$;h}Mwc@DdHT>F{V?lP9vNFR!
zJEKpLFEI^H2U0ee689*)bb7+~bD46bkuQvt<Q1&g3nOTIp-sw+$>h!Jv5Z4Mc)^Ow
zHI7+{^Mk=Njw+FR#CgqcLdhnOITq#)tndf&H@E3!GgybP3wEIA+L@0tk);B?b!=Ix
zw7gailx6Q*>z36;YJ->^Uy`*%synCt*|dtx2fY}dB3GbmHp_64Yo<Yc;0e7Z9;9DK
zTh15DMXtnjG0}8KI%=kk&P#e@%1z;FPxnv5N11OwoPrGZ!UA*1z7&^DG~4tyesMj?
z!p1>qW4*N;Nxsh-8p{~bnJJUYSp)SZ{xD=Au)hy9vq(sbd!`!dQlMcj(<gH~ehqlF
z)k*iYb9^}9bV9i)(BQ&@g>4<=<&jVmPYjcDH$22;RV`KF{;)-h%9}ACh=h`Y=U2n0
z7aOK6V4Iza!(P(XN@z!9_al<4lZHPCQkTJ|^<UL;9d$8u1}GkbDT7nFKHaD|bM&7>
z=W+^HPlQM{akHN={{93IH0WXQB)I8G@`J^C?fY8_Ny2y)L{ekrGf=7~t^Rn4K5e#!
z47xqd<V;+)MT4jv!Z~=3RGYkbh8iBVTPunaojJOB_{*F)b;kC<%luQV@fA|ULH^B1
zK+RX5l=ZTVqdB$QIj!JMbUTCzrz{+P;Lq+R)W=#YaPs|x)&C2b|Hes;_V~fZow!cS
SNAOgG)^2ZquS$=I%>MwX+v|J)

literal 0
HcmV?d00001

diff --git a/api/core/model_runtime/model_providers/nomic/_common.py b/api/core/model_runtime/model_providers/nomic/_common.py
new file mode 100644
index 0000000000000..406577dcd7e70
--- /dev/null
+++ b/api/core/model_runtime/model_providers/nomic/_common.py
@@ -0,0 +1,28 @@
+from core.model_runtime.errors.invoke import (
+    InvokeAuthorizationError,
+    InvokeBadRequestError,
+    InvokeConnectionError,
+    InvokeError,
+    InvokeRateLimitError,
+    InvokeServerUnavailableError,
+)
+
+
+class _CommonNomic:
+    @property
+    def _invoke_error_mapping(self) -> dict[type[InvokeError], list[type[Exception]]]:
+        """
+        Map model invoke error to unified error
+        The key is the error type thrown to the caller
+        The value is the error type thrown by the model,
+        which needs to be converted into a unified error type for the caller.
+
+        :return: Invoke error mapping
+        """
+        return {
+            InvokeConnectionError: [InvokeConnectionError],
+            InvokeServerUnavailableError: [InvokeServerUnavailableError],
+            InvokeRateLimitError: [InvokeRateLimitError],
+            InvokeAuthorizationError: [InvokeAuthorizationError],
+            InvokeBadRequestError: [KeyError, InvokeBadRequestError],
+        }
diff --git a/api/core/model_runtime/model_providers/nomic/nomic.py b/api/core/model_runtime/model_providers/nomic/nomic.py
new file mode 100644
index 0000000000000..d4e5da2e98ec9
--- /dev/null
+++ b/api/core/model_runtime/model_providers/nomic/nomic.py
@@ -0,0 +1,26 @@
+import logging
+
+from core.model_runtime.entities.model_entities import ModelType
+from core.model_runtime.errors.validate import CredentialsValidateFailedError
+from core.model_runtime.model_providers.__base.model_provider import ModelProvider
+
+logger = logging.getLogger(__name__)
+
+
+class NomicAtlasProvider(ModelProvider):
+    def validate_provider_credentials(self, credentials: dict) -> None:
+        """
+        Validate provider credentials
+
+        if validate failed, raise exception
+
+        :param credentials: provider credentials, credentials form defined in `provider_credential_schema`.
+        """
+        try:
+            model_instance = self.get_model_instance(ModelType.TEXT_EMBEDDING)
+            model_instance.validate_credentials(model="nomic-embed-text-v1.5", credentials=credentials)
+        except CredentialsValidateFailedError as ex:
+            raise ex
+        except Exception as ex:
+            logger.exception(f"{self.get_provider_schema().provider} credentials validate failed")
+            raise ex
diff --git a/api/core/model_runtime/model_providers/nomic/nomic.yaml b/api/core/model_runtime/model_providers/nomic/nomic.yaml
new file mode 100644
index 0000000000000..60dcf1facb475
--- /dev/null
+++ b/api/core/model_runtime/model_providers/nomic/nomic.yaml
@@ -0,0 +1,29 @@
+provider: nomic
+label:
+  zh_Hans: Nomic Atlas
+  en_US: Nomic Atlas
+icon_small:
+  en_US: icon_s_en.png
+icon_large:
+  en_US: icon_l_en.svg
+background: "#EFF1FE"
+help:
+  title:
+    en_US: Get your API key from Nomic Atlas
+    zh_Hans: 从Nomic Atlas获取 API Key
+  url:
+    en_US: https://atlas.nomic.ai/data
+supported_model_types:
+  - text-embedding
+configurate_methods:
+  - predefined-model
+provider_credential_schema:
+  credential_form_schemas:
+    - variable: nomic_api_key
+      label:
+        en_US: API Key
+      type: secret-input
+      required: true
+      placeholder:
+        zh_Hans: 在此输入您的 API Key
+        en_US: Enter your API Key
diff --git a/api/core/model_runtime/model_providers/nomic/text_embedding/__init__.py b/api/core/model_runtime/model_providers/nomic/text_embedding/__init__.py
new file mode 100644
index 0000000000000..e69de29bb2d1d
diff --git a/api/core/model_runtime/model_providers/nomic/text_embedding/nomic-embed-text-v1.5.yaml b/api/core/model_runtime/model_providers/nomic/text_embedding/nomic-embed-text-v1.5.yaml
new file mode 100644
index 0000000000000..111452df579f8
--- /dev/null
+++ b/api/core/model_runtime/model_providers/nomic/text_embedding/nomic-embed-text-v1.5.yaml
@@ -0,0 +1,8 @@
+model: nomic-embed-text-v1.5
+model_type: text-embedding
+model_properties:
+  context_size: 8192
+pricing:
+  input: "0.1"
+  unit: "0.000001"
+  currency: USD
diff --git a/api/core/model_runtime/model_providers/nomic/text_embedding/nomic-embed-text-v1.yaml b/api/core/model_runtime/model_providers/nomic/text_embedding/nomic-embed-text-v1.yaml
new file mode 100644
index 0000000000000..ac59f106ed292
--- /dev/null
+++ b/api/core/model_runtime/model_providers/nomic/text_embedding/nomic-embed-text-v1.yaml
@@ -0,0 +1,8 @@
+model: nomic-embed-text-v1
+model_type: text-embedding
+model_properties:
+  context_size: 8192
+pricing:
+  input: "0.1"
+  unit: "0.000001"
+  currency: USD
diff --git a/api/core/model_runtime/model_providers/nomic/text_embedding/text_embedding.py b/api/core/model_runtime/model_providers/nomic/text_embedding/text_embedding.py
new file mode 100644
index 0000000000000..6cccff6d4601c
--- /dev/null
+++ b/api/core/model_runtime/model_providers/nomic/text_embedding/text_embedding.py
@@ -0,0 +1,170 @@
+import time
+from functools import wraps
+from typing import Optional
+
+from nomic import embed
+from nomic import login as nomic_login
+
+from core.model_runtime.entities.model_entities import PriceType
+from core.model_runtime.entities.text_embedding_entities import (
+    EmbeddingUsage,
+    TextEmbeddingResult,
+)
+from core.model_runtime.errors.validate import CredentialsValidateFailedError
+from core.model_runtime.model_providers.__base.text_embedding_model import (
+    TextEmbeddingModel,
+)
+from core.model_runtime.model_providers.nomic._common import _CommonNomic
+
+
+def nomic_login_required(func):
+    @wraps(func)
+    def wrapper(*args, **kwargs):
+        try:
+            if not kwargs.get("credentials"):
+                raise ValueError("missing credentials parameters")
+            credentials = kwargs.get("credentials")
+            if "nomic_api_key" not in credentials:
+                raise ValueError("missing nomic_api_key in credentials parameters")
+            # nomic login
+            nomic_login(credentials["nomic_api_key"])
+        except Exception as ex:
+            raise CredentialsValidateFailedError(str(ex))
+        return func(*args, **kwargs)
+
+    return wrapper
+
+
+class NomicTextEmbeddingModel(_CommonNomic, TextEmbeddingModel):
+    """
+    Model class for nomic text embedding model.
+    """
+
+    def _invoke(
+        self,
+        model: str,
+        credentials: dict,
+        texts: list[str],
+        user: Optional[str] = None,
+    ) -> TextEmbeddingResult:
+        """
+        Invoke text embedding model
+
+        :param model: model name
+        :param credentials: model credentials
+        :param texts: texts to embed
+        :param user: unique user id
+        :return: embeddings result
+        """
+        embeddings, prompt_tokens, total_tokens = self.embed_text(
+            model=model,
+            credentials=credentials,
+            texts=texts,
+        )
+
+        # calc usage
+        usage = self._calc_response_usage(
+            model=model, credentials=credentials, tokens=prompt_tokens, total_tokens=total_tokens
+        )
+        return TextEmbeddingResult(embeddings=embeddings, usage=usage, model=model)
+
+    def get_num_tokens(self, model: str, credentials: dict, texts: list[str]) -> int:
+        """
+        Get number of tokens for given prompt messages
+
+        :param model: model name
+        :param credentials: model credentials
+        :param texts: texts to embed
+        :return:
+        """
+        if len(texts) == 0:
+            return 0
+
+        _, prompt_tokens, _ = self.embed_text(
+            model=model,
+            credentials=credentials,
+            texts=texts,
+        )
+        return prompt_tokens
+
+    def validate_credentials(self, model: str, credentials: dict) -> None:
+        """
+        Validate model credentials
+
+        :param model: model name
+        :param credentials: model credentials
+        :return:
+        """
+        try:
+            # call embedding model
+            self.embed_text(model=model, credentials=credentials, texts=["ping"])
+        except Exception as ex:
+            raise CredentialsValidateFailedError(str(ex))
+
+    @nomic_login_required
+    def embed_text(self, model: str, credentials: dict, texts: list[str]) -> tuple[list[list[float]], int, int]:
+        """Call out to Nomic's embedding endpoint.
+
+        Args:
+            model: The model to use for embedding.
+            texts: The list of texts to embed.
+
+        Returns:
+            List of embeddings, one for each text, and tokens usage.
+        """
+        embeddings: list[list[float]] = []
+        prompt_tokens = 0
+        total_tokens = 0
+
+        response = embed.text(
+            model=model,
+            texts=texts,
+        )
+
+        if not (response and "embeddings" in response):
+            raise ValueError("Embedding data is missing in the response.")
+
+        if not (response and "usage" in response):
+            raise ValueError("Response usage is missing.")
+
+        if "prompt_tokens" not in response["usage"]:
+            raise ValueError("Response usage does not contain prompt tokens.")
+
+        if "total_tokens" not in response["usage"]:
+            raise ValueError("Response usage does not contain total tokens.")
+
+        embeddings = [list(map(float, e)) for e in response["embeddings"]]
+        total_tokens = response["usage"]["total_tokens"]
+        prompt_tokens = response["usage"]["prompt_tokens"]
+        return embeddings, prompt_tokens, total_tokens
+
+    def _calc_response_usage(self, model: str, credentials: dict, tokens: int, total_tokens: int) -> EmbeddingUsage:
+        """
+        Calculate response usage
+
+        :param model: model name
+        :param credentials: model credentials
+        :param tokens: prompt tokens
+        :param total_tokens: total tokens
+        :return: usage
+        """
+        # get input price info
+        input_price_info = self.get_price(
+            model=model,
+            credentials=credentials,
+            price_type=PriceType.INPUT,
+            tokens=tokens,
+        )
+
+        # transform usage
+        usage = EmbeddingUsage(
+            tokens=tokens,
+            total_tokens=total_tokens,
+            unit_price=input_price_info.unit_price,
+            price_unit=input_price_info.unit,
+            total_price=input_price_info.total_amount,
+            currency=input_price_info.currency,
+            latency=time.perf_counter() - self.started_at,
+        )
+
+        return usage
diff --git a/api/poetry.lock b/api/poetry.lock
index 78816683d8d24..184cdb9e81e57 100644
--- a/api/poetry.lock
+++ b/api/poetry.lock
@@ -4135,6 +4135,20 @@ files = [
     {file = "joblib-1.4.2.tar.gz", hash = "sha256:2382c5816b2636fbd20a09e0f4e9dad4736765fdfb7dca582943b9c1366b3f0e"},
 ]
 
+[[package]]
+name = "jsonlines"
+version = "4.0.0"
+description = "Library with helpers for the jsonlines file format"
+optional = false
+python-versions = ">=3.8"
+files = [
+    {file = "jsonlines-4.0.0-py3-none-any.whl", hash = "sha256:185b334ff2ca5a91362993f42e83588a360cf95ce4b71a73548502bda52a7c55"},
+    {file = "jsonlines-4.0.0.tar.gz", hash = "sha256:0c6d2c09117550c089995247f605ae4cf77dd1533041d366351f6f298822ea74"},
+]
+
+[package.dependencies]
+attrs = ">=19.2.0"
+
 [[package]]
 name = "jsonpath-ng"
 version = "1.6.1"
@@ -4469,6 +4483,24 @@ files = [
     {file = "llvmlite-0.43.0.tar.gz", hash = "sha256:ae2b5b5c3ef67354824fb75517c8db5fbe93bc02cd9671f3c62271626bc041d5"},
 ]
 
+[[package]]
+name = "loguru"
+version = "0.7.2"
+description = "Python logging made (stupidly) simple"
+optional = false
+python-versions = ">=3.5"
+files = [
+    {file = "loguru-0.7.2-py3-none-any.whl", hash = "sha256:003d71e3d3ed35f0f8984898359d65b79e5b21943f78af86aa5491210429b8eb"},
+    {file = "loguru-0.7.2.tar.gz", hash = "sha256:e671a53522515f34fd406340ee968cb9ecafbc4b36c679da03c18fd8d0bd51ac"},
+]
+
+[package.dependencies]
+colorama = {version = ">=0.3.4", markers = "sys_platform == \"win32\""}
+win32-setctime = {version = ">=1.0.0", markers = "sys_platform == \"win32\""}
+
+[package.extras]
+dev = ["Sphinx (==7.2.5)", "colorama (==0.4.5)", "colorama (==0.4.6)", "exceptiongroup (==1.1.3)", "freezegun (==1.1.0)", "freezegun (==1.2.2)", "mypy (==v0.910)", "mypy (==v0.971)", "mypy (==v1.4.1)", "mypy (==v1.5.1)", "pre-commit (==3.4.0)", "pytest (==6.1.2)", "pytest (==7.4.0)", "pytest-cov (==2.12.1)", "pytest-cov (==4.1.0)", "pytest-mypy-plugins (==1.9.3)", "pytest-mypy-plugins (==3.0.0)", "sphinx-autobuild (==2021.3.14)", "sphinx-rtd-theme (==1.3.0)", "tox (==3.27.1)", "tox (==4.11.0)"]
+
 [[package]]
 name = "lxml"
 version = "5.3.0"
@@ -5320,6 +5352,36 @@ plot = ["matplotlib"]
 tgrep = ["pyparsing"]
 twitter = ["twython"]
 
+[[package]]
+name = "nomic"
+version = "3.1.2"
+description = "The official Nomic python client."
+optional = false
+python-versions = "*"
+files = [
+    {file = "nomic-3.1.2.tar.gz", hash = "sha256:2de1ab1dcf2429011c92987bb2f1eafe1a3a4901c3185b18f994bf89616f606d"},
+]
+
+[package.dependencies]
+click = "*"
+jsonlines = "*"
+loguru = "*"
+numpy = "*"
+pandas = "*"
+pillow = "*"
+pyarrow = "*"
+pydantic = "*"
+pyjwt = "*"
+requests = "*"
+rich = "*"
+tqdm = "*"
+
+[package.extras]
+all = ["nomic[aws,local]"]
+aws = ["boto3", "sagemaker"]
+dev = ["black (==24.3.0)", "cairosvg", "coverage", "isort", "mkautodoc", "mkdocs-jupyter", "mkdocs-material", "mkdocstrings[python]", "myst-parser", "nomic[all]", "pandas", "pillow", "pylint", "pyright", "pytest", "pytorch-lightning", "twine"]
+local = ["gpt4all (>=2.5.0,<3)"]
+
 [[package]]
 name = "novita-client"
 version = "0.5.7"
@@ -9919,6 +9981,20 @@ files = [
 beautifulsoup4 = "*"
 requests = ">=2.0.0,<3.0.0"
 
+[[package]]
+name = "win32-setctime"
+version = "1.1.0"
+description = "A small Python utility to set file creation time on Windows"
+optional = false
+python-versions = ">=3.5"
+files = [
+    {file = "win32_setctime-1.1.0-py3-none-any.whl", hash = "sha256:231db239e959c2fe7eb1d7dc129f11172354f98361c4fa2d6d2d7e278baa8aad"},
+    {file = "win32_setctime-1.1.0.tar.gz", hash = "sha256:15cf5750465118d6929ae4de4eb46e8edae9a5634350c01ba582df868e932cb2"},
+]
+
+[package.extras]
+dev = ["black (>=19.3b0)", "pytest (>=4.6.2)"]
+
 [[package]]
 name = "wrapt"
 version = "1.16.0"
@@ -10422,4 +10498,4 @@ cffi = ["cffi (>=1.11)"]
 [metadata]
 lock-version = "2.0"
 python-versions = ">=3.10,<3.13"
-content-hash = "eb7ef7be5c7790e214f37f17f92b69407ad557cb80055ef7e49e36eb51b3fca6"
+content-hash = "17c4108d92c415d987f8b437ea3e0484c5601a05bfe175339a8546c93c159bc5"
diff --git a/api/pyproject.toml b/api/pyproject.toml
index 066b4772a9918..41244f516c95c 100644
--- a/api/pyproject.toml
+++ b/api/pyproject.toml
@@ -100,6 +100,7 @@ exclude = [
 OPENAI_API_KEY = "sk-IamNotARealKeyJustForMockTestKawaiiiiiiiiii"
 UPSTAGE_API_KEY = "up-aaaaaaaaaaaaaaaaaaaa"
 FIREWORKS_API_KEY = "fw_aaaaaaaaaaaaaaaaaaaa"
+NOMIC_API_KEY = "nk-aaaaaaaaaaaaaaaaaaaa"
 AZURE_OPENAI_API_BASE = "https://difyai-openai.openai.azure.com"
 AZURE_OPENAI_API_KEY = "xxxxb1707exxxxxxxxxxaaxxxxxf94"
 ANTHROPIC_API_KEY = "sk-ant-api11-IamNotARealKeyJustForMockTestKawaiiiiiiiiii-NotBaka-ASkksz"
@@ -217,6 +218,7 @@ azure-ai-inference = "^1.0.0b3"
 volcengine-python-sdk = {extras = ["ark"], version = "^1.0.98"}
 oci = "^2.133.0"
 tos = "^2.7.1"
+nomic = "^3.1.2"
 [tool.poetry.group.indriect.dependencies]
 kaleido = "0.2.1"
 rank-bm25 = "~0.2.2"
diff --git a/api/tests/integration_tests/model_runtime/__mock/nomic_embeddings.py b/api/tests/integration_tests/model_runtime/__mock/nomic_embeddings.py
new file mode 100644
index 0000000000000..281e866e45c2e
--- /dev/null
+++ b/api/tests/integration_tests/model_runtime/__mock/nomic_embeddings.py
@@ -0,0 +1,59 @@
+import os
+from collections.abc import Callable
+from typing import Any, Literal, Union
+
+import pytest
+
+# import monkeypatch
+from _pytest.monkeypatch import MonkeyPatch
+from nomic import embed
+
+
+def create_embedding(texts: list[str], model: str, **kwargs: Any) -> dict:
+    texts_len = len(texts)
+
+    foo_embedding_sample = 0.123456
+
+    combined = {
+        "embeddings": [[foo_embedding_sample for _ in range(768)] for _ in range(texts_len)],
+        "usage": {"prompt_tokens": texts_len, "total_tokens": texts_len},
+        "model": model,
+        "inference_mode": "remote",
+    }
+
+    return combined
+
+
+def mock_nomic(
+    monkeypatch: MonkeyPatch,
+    methods: list[Literal["text_embedding"]],
+) -> Callable[[], None]:
+    """
+    mock nomic module
+
+    :param monkeypatch: pytest monkeypatch fixture
+    :return: unpatch function
+    """
+
+    def unpatch() -> None:
+        monkeypatch.undo()
+
+    if "text_embedding" in methods:
+        monkeypatch.setattr(embed, "text", create_embedding)
+
+    return unpatch
+
+
+MOCK = os.getenv("MOCK_SWITCH", "false").lower() == "true"
+
+
+@pytest.fixture
+def setup_nomic_mock(request, monkeypatch):
+    methods = request.param if hasattr(request, "param") else []
+    if MOCK:
+        unpatch = mock_nomic(monkeypatch, methods=methods)
+
+    yield
+
+    if MOCK:
+        unpatch()
diff --git a/api/tests/integration_tests/model_runtime/nomic/__init__.py b/api/tests/integration_tests/model_runtime/nomic/__init__.py
new file mode 100644
index 0000000000000..e69de29bb2d1d
diff --git a/api/tests/integration_tests/model_runtime/nomic/test_embeddings.py b/api/tests/integration_tests/model_runtime/nomic/test_embeddings.py
new file mode 100644
index 0000000000000..52dc96ee95c1b
--- /dev/null
+++ b/api/tests/integration_tests/model_runtime/nomic/test_embeddings.py
@@ -0,0 +1,62 @@
+import os
+
+import pytest
+
+from core.model_runtime.entities.text_embedding_entities import TextEmbeddingResult
+from core.model_runtime.errors.validate import CredentialsValidateFailedError
+from core.model_runtime.model_providers.nomic.text_embedding.text_embedding import NomicTextEmbeddingModel
+from tests.integration_tests.model_runtime.__mock.nomic_embeddings import setup_nomic_mock
+
+
+@pytest.mark.parametrize("setup_nomic_mock", [["text_embedding"]], indirect=True)
+def test_validate_credentials(setup_nomic_mock):
+    model = NomicTextEmbeddingModel()
+
+    with pytest.raises(CredentialsValidateFailedError):
+        model.validate_credentials(
+            model="nomic-embed-text-v1.5",
+            credentials={
+                "nomic_api_key": "invalid_key",
+            },
+        )
+
+    model.validate_credentials(
+        model="nomic-embed-text-v1.5",
+        credentials={
+            "nomic_api_key": os.environ.get("NOMIC_API_KEY"),
+        },
+    )
+
+
+@pytest.mark.parametrize("setup_nomic_mock", [["text_embedding"]], indirect=True)
+def test_invoke_model(setup_nomic_mock):
+    model = NomicTextEmbeddingModel()
+
+    result = model.invoke(
+        model="nomic-embed-text-v1.5",
+        credentials={
+            "nomic_api_key": os.environ.get("NOMIC_API_KEY"),
+        },
+        texts=["hello", "world"],
+        user="foo",
+    )
+
+    assert isinstance(result, TextEmbeddingResult)
+    assert result.model == "nomic-embed-text-v1.5"
+    assert len(result.embeddings) == 2
+    assert result.usage.total_tokens == 2
+
+
+@pytest.mark.parametrize("setup_nomic_mock", [["text_embedding"]], indirect=True)
+def test_get_num_tokens(setup_nomic_mock):
+    model = NomicTextEmbeddingModel()
+
+    num_tokens = model.get_num_tokens(
+        model="nomic-embed-text-v1.5",
+        credentials={
+            "nomic_api_key": os.environ.get("NOMIC_API_KEY"),
+        },
+        texts=["hello", "world"],
+    )
+
+    assert num_tokens == 2
diff --git a/api/tests/integration_tests/model_runtime/nomic/test_provider.py b/api/tests/integration_tests/model_runtime/nomic/test_provider.py
new file mode 100644
index 0000000000000..6cad400c06955
--- /dev/null
+++ b/api/tests/integration_tests/model_runtime/nomic/test_provider.py
@@ -0,0 +1,22 @@
+import os
+
+import pytest
+
+from core.model_runtime.errors.validate import CredentialsValidateFailedError
+from core.model_runtime.model_providers.nomic.nomic import NomicAtlasProvider
+from core.model_runtime.model_providers.nomic.text_embedding.text_embedding import NomicTextEmbeddingModel
+from tests.integration_tests.model_runtime.__mock.nomic_embeddings import setup_nomic_mock
+
+
+@pytest.mark.parametrize("setup_nomic_mock", [["text_embedding"]], indirect=True)
+def test_validate_provider_credentials(setup_nomic_mock):
+    provider = NomicAtlasProvider()
+
+    with pytest.raises(CredentialsValidateFailedError):
+        provider.validate_provider_credentials(credentials={})
+
+    provider.validate_provider_credentials(
+        credentials={
+            "nomic_api_key": os.environ.get("NOMIC_API_KEY"),
+        },
+    )
diff --git a/dev/pytest/pytest_model_runtime.sh b/dev/pytest/pytest_model_runtime.sh
index 4c1c6bf4f3ab1..4c0083a2de343 100755
--- a/dev/pytest/pytest_model_runtime.sh
+++ b/dev/pytest/pytest_model_runtime.sh
@@ -7,4 +7,5 @@ pytest api/tests/integration_tests/model_runtime/anthropic \
   api/tests/integration_tests/model_runtime/google api/tests/integration_tests/model_runtime/xinference \
   api/tests/integration_tests/model_runtime/huggingface_hub/test_llm.py \
   api/tests/integration_tests/model_runtime/upstage \
-  api/tests/integration_tests/model_runtime/fireworks
+  api/tests/integration_tests/model_runtime/fireworks \
+  api/tests/integration_tests/model_runtime/nomic

From 8cc9e683631e155e9c7c1c282e91f9933f2858ba Mon Sep 17 00:00:00 2001
From: crazywoola <100913391+crazywoola@users.noreply.github.com>
Date: Mon, 23 Sep 2024 21:00:34 +0900
Subject: [PATCH 043/235] fix: prompt for the follow-up suggestions (#8685)

---
 api/core/llm_generator/prompts.py | 1 -
 1 file changed, 1 deletion(-)

diff --git a/api/core/llm_generator/prompts.py b/api/core/llm_generator/prompts.py
index c40b6d180804c..e5b678451600a 100644
--- a/api/core/llm_generator/prompts.py
+++ b/api/core/llm_generator/prompts.py
@@ -65,7 +65,6 @@
     "Please help me predict the three most likely questions that human would ask, "
     "and keeping each question under 20 characters.\n"
     "MAKE SURE your output is the SAME language as the Assistant's latest response"
-    "(if the main response is written in Chinese, then the language of your output must be using Chinese.)!\n"
     "The output must be an array in JSON format following the specified schema:\n"
     '["question1","question2","question3"]\n'
 )

From bef83a4d2ecae95b5b07be68b10fe5181cfa0ac1 Mon Sep 17 00:00:00 2001
From: Nam Vu <zuzoovn@gmail.com>
Date: Mon, 23 Sep 2024 20:32:58 +0700
Subject: [PATCH 044/235] fix: typos and improve naming conventions: (#8687)

---
 api/commands.py                               |  2 +-
 .../api_resource/chat/async_completions.py    |  2 +-
 .../api_resource/chat/completions.py          |  2 +-
 .../zhipuai/zhipuai_sdk/core/_base_models.py  |  3 +--
 .../zhipuai/zhipuai_sdk/core/_http_client.py  | 12 +++++-----
 .../zhipuai_sdk/core/_legacy_response.py      |  2 +-
 .../zhipuai/zhipuai_sdk/core/_response.py     |  4 ++--
 .../types/knowledge/document/__init__.py      |  4 ++--
 .../types/knowledge/document/document.py      |  6 ++---
 .../nodes/end/end_stream_processor.py         | 22 +++++++++----------
 web/app/components/base/chat/chat/hooks.ts    |  4 ++--
 .../base/image-uploader/image-preview.tsx     |  4 ++--
 .../develop/template/template_workflow.en.mdx |  2 +-
 .../workflow/hooks/use-workflow-run.ts        |  6 ++---
 14 files changed, 37 insertions(+), 38 deletions(-)

diff --git a/api/commands.py b/api/commands.py
index b8fc81af673af..7ef4aed7f7766 100644
--- a/api/commands.py
+++ b/api/commands.py
@@ -652,7 +652,7 @@ def fix_app_site_missing():
                         app_was_created.send(app, account=account)
                 except Exception as e:
                     failed_app_ids.append(app_id)
-                    click.echo(click.style("FFailed to fix missing site for app {}".format(app_id), fg="red"))
+                    click.echo(click.style("Failed to fix missing site for app {}".format(app_id), fg="red"))
                     logging.exception(f"Fix app related site missing issue failed, error: {e}")
                     continue
 
diff --git a/api/core/model_runtime/model_providers/zhipuai/zhipuai_sdk/api_resource/chat/async_completions.py b/api/core/model_runtime/model_providers/zhipuai/zhipuai_sdk/api_resource/chat/async_completions.py
index d8ecc310644d1..05510a3ec421d 100644
--- a/api/core/model_runtime/model_providers/zhipuai/zhipuai_sdk/api_resource/chat/async_completions.py
+++ b/api/core/model_runtime/model_providers/zhipuai/zhipuai_sdk/api_resource/chat/async_completions.py
@@ -57,7 +57,7 @@ def create(
             if temperature <= 0:
                 do_sample = False
                 temperature = 0.01
-                # logger.warning("temperature:取值范围是：(0.0, 1.0) 开区间，do_sample重写为:false（参数top_p temperture不生效）")  # noqa: E501
+                # logger.warning("temperature:取值范围是：(0.0, 1.0) 开区间，do_sample重写为:false（参数top_p temperature不生效）")  # noqa: E501
             if temperature >= 1:
                 temperature = 0.99
                 # logger.warning("temperature:取值范围是：(0.0, 1.0) 开区间")
diff --git a/api/core/model_runtime/model_providers/zhipuai/zhipuai_sdk/api_resource/chat/completions.py b/api/core/model_runtime/model_providers/zhipuai/zhipuai_sdk/api_resource/chat/completions.py
index 1c23473a03ae3..8e5bb454e6ce7 100644
--- a/api/core/model_runtime/model_providers/zhipuai/zhipuai_sdk/api_resource/chat/completions.py
+++ b/api/core/model_runtime/model_providers/zhipuai/zhipuai_sdk/api_resource/chat/completions.py
@@ -60,7 +60,7 @@ def create(
             if temperature <= 0:
                 do_sample = False
                 temperature = 0.01
-                # logger.warning("temperature:取值范围是：(0.0, 1.0) 开区间，do_sample重写为:false（参数top_p temperture不生效）")  # noqa: E501
+                # logger.warning("temperature:取值范围是：(0.0, 1.0) 开区间，do_sample重写为:false（参数top_p temperature不生效）")  # noqa: E501
             if temperature >= 1:
                 temperature = 0.99
                 # logger.warning("temperature:取值范围是：(0.0, 1.0) 开区间")
diff --git a/api/core/model_runtime/model_providers/zhipuai/zhipuai_sdk/core/_base_models.py b/api/core/model_runtime/model_providers/zhipuai/zhipuai_sdk/core/_base_models.py
index 5e9a7e0a987e2..6d8ba700b7b1d 100644
--- a/api/core/model_runtime/model_providers/zhipuai/zhipuai_sdk/core/_base_models.py
+++ b/api/core/model_runtime/model_providers/zhipuai/zhipuai_sdk/core/_base_models.py
@@ -630,8 +630,7 @@ def validate_type(*, type_: type[_T], value: object) -> _T:
     return cast(_T, _validate_non_model_type(type_=type_, value=value))
 
 
-# our use of subclasssing here causes weirdness for type checkers,
-# so we just pretend that we don't subclass
+# Subclassing here confuses type checkers, so we treat this class as non-inheriting.
 if TYPE_CHECKING:
     GenericModel = BaseModel
 else:
diff --git a/api/core/model_runtime/model_providers/zhipuai/zhipuai_sdk/core/_http_client.py b/api/core/model_runtime/model_providers/zhipuai/zhipuai_sdk/core/_http_client.py
index d0f933d814138..ffdafb85d581f 100644
--- a/api/core/model_runtime/model_providers/zhipuai/zhipuai_sdk/core/_http_client.py
+++ b/api/core/model_runtime/model_providers/zhipuai/zhipuai_sdk/core/_http_client.py
@@ -169,7 +169,7 @@ def _set_private_attributes(
     # Pydantic uses a custom `__iter__` method to support casting BaseModels
     # to dictionaries. e.g. dict(model).
     # As we want to support `for item in page`, this is inherently incompatible
-    # with the default pydantic behaviour. It is not possible to support both
+    # with the default pydantic behavior. It is not possible to support both
     # use cases at once. Fortunately, this is not a big deal as all other pydantic
     # methods should continue to work as expected as there is an alternative method
     # to cast a model to a dictionary, model.dict(), which is used internally
@@ -356,16 +356,16 @@ def _build_request(self, options: FinalRequestOptions) -> httpx.Request:
             **kwargs,
         )
 
-    def _object_to_formfata(self, key: str, value: Data | Mapping[object, object]) -> list[tuple[str, str]]:
+    def _object_to_formdata(self, key: str, value: Data | Mapping[object, object]) -> list[tuple[str, str]]:
         items = []
 
         if isinstance(value, Mapping):
             for k, v in value.items():
-                items.extend(self._object_to_formfata(f"{key}[{k}]", v))
+                items.extend(self._object_to_formdata(f"{key}[{k}]", v))
             return items
         if isinstance(value, list | tuple):
             for v in value:
-                items.extend(self._object_to_formfata(key + "[]", v))
+                items.extend(self._object_to_formdata(key + "[]", v))
             return items
 
         def _primitive_value_to_str(val) -> str:
@@ -385,7 +385,7 @@ def _primitive_value_to_str(val) -> str:
         return [(key, str_data)]
 
     def _make_multipartform(self, data: Mapping[object, object]) -> dict[str, object]:
-        items = flatten(list(starmap(self._object_to_formfata, data.items())))
+        items = flatten(list(starmap(self._object_to_formdata, data.items())))
 
         serialized: dict[str, object] = {}
         for key, value in items:
@@ -620,7 +620,7 @@ def _process_response(
         stream: bool,
         stream_cls: type[StreamResponse] | None,
     ) -> ResponseT:
-        # _legacy_response with raw_response_header to paser method
+        # _legacy_response with raw_response_header to parser method
         if response.request.headers.get(RAW_RESPONSE_HEADER) == "true":
             return cast(
                 ResponseT,
diff --git a/api/core/model_runtime/model_providers/zhipuai/zhipuai_sdk/core/_legacy_response.py b/api/core/model_runtime/model_providers/zhipuai/zhipuai_sdk/core/_legacy_response.py
index 47183b9eee9c0..51bf21bcdc17a 100644
--- a/api/core/model_runtime/model_providers/zhipuai/zhipuai_sdk/core/_legacy_response.py
+++ b/api/core/model_runtime/model_providers/zhipuai/zhipuai_sdk/core/_legacy_response.py
@@ -87,7 +87,7 @@ def parse(self, *, to: type[_T] | None = None) -> R | _T:
 
         For lower-level control, see `.read()`, `.json()`, `.iter_bytes()`.
 
-        You can customise the type that the response is parsed into through
+        You can customize the type that the response is parsed into through
         the `to` argument, e.g.
 
         ```py
diff --git a/api/core/model_runtime/model_providers/zhipuai/zhipuai_sdk/core/_response.py b/api/core/model_runtime/model_providers/zhipuai/zhipuai_sdk/core/_response.py
index 45443da662d57..92e601805569f 100644
--- a/api/core/model_runtime/model_providers/zhipuai/zhipuai_sdk/core/_response.py
+++ b/api/core/model_runtime/model_providers/zhipuai/zhipuai_sdk/core/_response.py
@@ -252,7 +252,7 @@ def parse(self, *, to: type[_T] | None = None) -> R | _T:
 
         For lower-level control, see `.read()`, `.json()`, `.iter_bytes()`.
 
-        You can customise the type that the response is parsed into through
+        You can customize the type that the response is parsed into through
         the `to` argument, e.g.
 
         ```py
@@ -363,7 +363,7 @@ class StreamAlreadyConsumed(ZhipuAIError):  # noqa: N818
     # ^ error
     ```
 
-    If you want this behaviour you'll need to either manually accumulate the response
+    If you want this behavior you'll need to either manually accumulate the response
     content or call `await response.read()` before iterating over the stream.
     """
 
diff --git a/api/core/model_runtime/model_providers/zhipuai/zhipuai_sdk/types/knowledge/document/__init__.py b/api/core/model_runtime/model_providers/zhipuai/zhipuai_sdk/types/knowledge/document/__init__.py
index 32e23e6dab307..59cb41d7124a7 100644
--- a/api/core/model_runtime/model_providers/zhipuai/zhipuai_sdk/types/knowledge/document/__init__.py
+++ b/api/core/model_runtime/model_providers/zhipuai/zhipuai_sdk/types/knowledge/document/__init__.py
@@ -1,8 +1,8 @@
-from .document import DocumentData, DocumentFailedInfo, DocumentObject, DocumentSuccessinfo
+from .document import DocumentData, DocumentFailedInfo, DocumentObject, DocumentSuccessInfo
 
 __all__ = [
     "DocumentData",
     "DocumentObject",
-    "DocumentSuccessinfo",
+    "DocumentSuccessInfo",
     "DocumentFailedInfo",
 ]
diff --git a/api/core/model_runtime/model_providers/zhipuai/zhipuai_sdk/types/knowledge/document/document.py b/api/core/model_runtime/model_providers/zhipuai/zhipuai_sdk/types/knowledge/document/document.py
index b9a1646391ece..980bc6f4a7c40 100644
--- a/api/core/model_runtime/model_providers/zhipuai/zhipuai_sdk/types/knowledge/document/document.py
+++ b/api/core/model_runtime/model_providers/zhipuai/zhipuai_sdk/types/knowledge/document/document.py
@@ -2,10 +2,10 @@
 
 from ....core import BaseModel
 
-__all__ = ["DocumentData", "DocumentObject", "DocumentSuccessinfo", "DocumentFailedInfo"]
+__all__ = ["DocumentData", "DocumentObject", "DocumentSuccessInfo", "DocumentFailedInfo"]
 
 
-class DocumentSuccessinfo(BaseModel):
+class DocumentSuccessInfo(BaseModel):
     documentId: Optional[str] = None
     """文件id"""
     filename: Optional[str] = None
@@ -24,7 +24,7 @@ class DocumentFailedInfo(BaseModel):
 class DocumentObject(BaseModel):
     """文档信息"""
 
-    successInfos: Optional[list[DocumentSuccessinfo]] = None
+    successInfos: Optional[list[DocumentSuccessInfo]] = None
     """上传成功的文件信息"""
     failedInfos: Optional[list[DocumentFailedInfo]] = None
     """上传失败的文件信息"""
diff --git a/api/core/workflow/nodes/end/end_stream_processor.py b/api/core/workflow/nodes/end/end_stream_processor.py
index 0366d7965d7c1..1aecf863ac5fb 100644
--- a/api/core/workflow/nodes/end/end_stream_processor.py
+++ b/api/core/workflow/nodes/end/end_stream_processor.py
@@ -22,8 +22,8 @@ def __init__(self, graph: Graph, variable_pool: VariablePool) -> None:
         for end_node_id, _ in self.end_stream_param.end_stream_variable_selector_mapping.items():
             self.route_position[end_node_id] = 0
         self.current_stream_chunk_generating_node_ids: dict[str, list[str]] = {}
-        self.has_outputed = False
-        self.outputed_node_ids = set()
+        self.has_output = False
+        self.output_node_ids = set()
 
     def process(self, generator: Generator[GraphEngineEvent, None, None]) -> Generator[GraphEngineEvent, None, None]:
         for event in generator:
@@ -34,11 +34,11 @@ def process(self, generator: Generator[GraphEngineEvent, None, None]) -> Generat
                 yield event
             elif isinstance(event, NodeRunStreamChunkEvent):
                 if event.in_iteration_id:
-                    if self.has_outputed and event.node_id not in self.outputed_node_ids:
+                    if self.has_output and event.node_id not in self.output_node_ids:
                         event.chunk_content = "\n" + event.chunk_content
 
-                    self.outputed_node_ids.add(event.node_id)
-                    self.has_outputed = True
+                    self.output_node_ids.add(event.node_id)
+                    self.has_output = True
                     yield event
                     continue
 
@@ -53,11 +53,11 @@ def process(self, generator: Generator[GraphEngineEvent, None, None]) -> Generat
                     )
 
                 if stream_out_end_node_ids:
-                    if self.has_outputed and event.node_id not in self.outputed_node_ids:
+                    if self.has_output and event.node_id not in self.output_node_ids:
                         event.chunk_content = "\n" + event.chunk_content
 
-                    self.outputed_node_ids.add(event.node_id)
-                    self.has_outputed = True
+                    self.output_node_ids.add(event.node_id)
+                    self.has_output = True
                     yield event
             elif isinstance(event, NodeRunSucceededEvent):
                 yield event
@@ -124,11 +124,11 @@ def _generate_stream_outputs_when_node_finished(
 
                 if text:
                     current_node_id = value_selector[0]
-                    if self.has_outputed and current_node_id not in self.outputed_node_ids:
+                    if self.has_output and current_node_id not in self.output_node_ids:
                         text = "\n" + text
 
-                    self.outputed_node_ids.add(current_node_id)
-                    self.has_outputed = True
+                    self.output_node_ids.add(current_node_id)
+                    self.has_output = True
                     yield NodeRunStreamChunkEvent(
                         id=event.id,
                         node_id=event.node_id,
diff --git a/web/app/components/base/chat/chat/hooks.ts b/web/app/components/base/chat/chat/hooks.ts
index dfb5a1b6855d5..64c238f9d1e56 100644
--- a/web/app/components/base/chat/chat/hooks.ts
+++ b/web/app/components/base/chat/chat/hooks.ts
@@ -334,9 +334,9 @@ export const useChat = (
             const newChatList = produce(chatListRef.current, (draft) => {
               const index = draft.findIndex(item => item.id === responseItem.id)
               if (index !== -1) {
-                const requestion = draft[index - 1]
+                const question = draft[index - 1]
                 draft[index - 1] = {
-                  ...requestion,
+                  ...question,
                 }
                 draft[index] = {
                   ...draft[index],
diff --git a/web/app/components/base/image-uploader/image-preview.tsx b/web/app/components/base/image-uploader/image-preview.tsx
index e5bd4c1bbc0dc..096facabfd2a8 100644
--- a/web/app/components/base/image-uploader/image-preview.tsx
+++ b/web/app/components/base/image-uploader/image-preview.tsx
@@ -88,7 +88,7 @@ const ImagePreview: FC<ImagePreviewProps> = ({
     })
   }
 
-  const imageTobase64ToBlob = (base64: string, type = 'image/png'): Blob => {
+  const imageBase64ToBlob = (base64: string, type = 'image/png'): Blob => {
     const byteCharacters = atob(base64)
     const byteArrays = []
 
@@ -109,7 +109,7 @@ const ImagePreview: FC<ImagePreviewProps> = ({
     const shareImage = async () => {
       try {
         const base64Data = url.split(',')[1]
-        const blob = imageTobase64ToBlob(base64Data, 'image/png')
+        const blob = imageBase64ToBlob(base64Data, 'image/png')
 
         await navigator.clipboard.write([
           new ClipboardItem({
diff --git a/web/app/components/develop/template/template_workflow.en.mdx b/web/app/components/develop/template/template_workflow.en.mdx
index 2bd0fe9daf436..5c712c2c2979e 100644
--- a/web/app/components/develop/template/template_workflow.en.mdx
+++ b/web/app/components/develop/template/template_workflow.en.mdx
@@ -424,7 +424,7 @@ Workflow applications offers non-session support and is ideal for translation, a
 />
 <Row>
   <Col>
-    Returns worklfow logs, with the first page returning the latest `{limit}` messages, i.e., in reverse order.
+    Returns workflow logs, with the first page returning the latest `{limit}` messages, i.e., in reverse order.
 
     ### Query
 
diff --git a/web/app/components/workflow/hooks/use-workflow-run.ts b/web/app/components/workflow/hooks/use-workflow-run.ts
index e1da503f3825f..68c3ff0a4b458 100644
--- a/web/app/components/workflow/hooks/use-workflow-run.ts
+++ b/web/app/components/workflow/hooks/use-workflow-run.ts
@@ -185,7 +185,7 @@ export const useWorkflowRun = () => {
             draft.forEach((edge) => {
               edge.data = {
                 ...edge.data,
-                _runned: false,
+                _run: false,
               }
             })
           })
@@ -292,7 +292,7 @@ export const useWorkflowRun = () => {
             const newEdges = produce(edges, (draft) => {
               draft.forEach((edge) => {
                 if (edge.target === data.node_id && incomeNodesId.includes(edge.source))
-                  edge.data = { ...edge.data, _runned: true } as any
+                  edge.data = { ...edge.data, _run: true } as any
               })
             })
             setEdges(newEdges)
@@ -416,7 +416,7 @@ export const useWorkflowRun = () => {
             const edge = draft.find(edge => edge.target === data.node_id && edge.source === prevNodeId)
 
             if (edge)
-              edge.data = { ...edge.data, _runned: true } as any
+              edge.data = { ...edge.data, _run: true } as any
           })
           setEdges(newEdges)
 

From 7f1b0288409626fcb87bee2fe12643ebfa0f3059 Mon Sep 17 00:00:00 2001
From: Sa Zhang <55871322+Nick17t@users.noreply.github.com>
Date: Mon, 23 Sep 2024 21:39:26 +0800
Subject: [PATCH 045/235] fix: change the brand name to Jina AI (#8691)

Co-authored-by: sa zhang <sa.zhang@jina.ai>
---
 .../model_providers/jina/jina.yaml            |  6 +++---
 .../tools/provider/builtin/jina/jina.yaml     | 20 +++++++++----------
 2 files changed, 13 insertions(+), 13 deletions(-)

diff --git a/api/core/model_runtime/model_providers/jina/jina.yaml b/api/core/model_runtime/model_providers/jina/jina.yaml
index 9c70d6ff33ba4..2f2d6e6daad73 100644
--- a/api/core/model_runtime/model_providers/jina/jina.yaml
+++ b/api/core/model_runtime/model_providers/jina/jina.yaml
@@ -1,6 +1,6 @@
-provider: jina
+provider: Jina AI
 label:
-  en_US: Jina
+  en_US: Jina AI
 description:
   en_US: Embedding and Rerank Model Supported
 icon_small:
@@ -11,7 +11,7 @@ background: "#EFFDFD"
 help:
   title:
     en_US: Get your API key from Jina AI
-    zh_Hans: 从 Jina 获取 API Key
+    zh_Hans: 从 Jina AI 获取 API Key
   url:
     en_US: https://jina.ai/
 supported_model_types:
diff --git a/api/core/tools/provider/builtin/jina/jina.yaml b/api/core/tools/provider/builtin/jina/jina.yaml
index 06f23382d92a3..9ce5cbd6d1afa 100644
--- a/api/core/tools/provider/builtin/jina/jina.yaml
+++ b/api/core/tools/provider/builtin/jina/jina.yaml
@@ -1,10 +1,10 @@
 identity:
   author: Dify
-  name: jina
+  name: Jina AI
   label:
-    en_US: Jina
-    zh_Hans: Jina
-    pt_BR: Jina
+    en_US: Jina AI
+    zh_Hans: Jina AI
+    pt_BR: Jina AI
   description:
     en_US: Convert any URL to an LLM-friendly input or perform searches on the web for grounding information. Experience improved output for your agent and RAG systems at no cost.
     zh_Hans: 将任何URL转换为LLM易读的输入或在网页上搜索引擎上搜索引擎。
@@ -22,11 +22,11 @@ credentials_for_provider:
       zh_Hans: API 密钥（可留空）
       pt_BR: Chave API (deixe vazio se você não tiver uma)
     placeholder:
-      en_US: Please enter your Jina API key
-      zh_Hans: 请输入你的 Jina API 密钥
-      pt_BR: Por favor, insira sua chave de API do Jina
+      en_US: Please enter your Jina AI API key
+      zh_Hans: 请输入你的 Jina AI API 密钥
+      pt_BR: Por favor, insira sua chave de API do Jina AI
     help:
-      en_US: Get your Jina API key from Jina (optional, but you can get a higher rate)
-      zh_Hans: 从 Jina 获取您的 Jina API 密钥（非必须，能得到更高的速率）
-      pt_BR: Obtenha sua chave de API do Jina na Jina (opcional, mas você pode obter uma taxa mais alta)
+      en_US: Get your Jina AI API key from Jina AI (optional, but you can get a higher rate)
+      zh_Hans: 从 Jina AI 获取您的 Jina AI API 密钥（非必须，能得到更高的速率）
+      pt_BR: Obtenha sua chave de API do Jina AI na Jina AI (opcional, mas você pode obter uma taxa mais alta)
     url: https://jina.ai

From 21e9608b239036c98367193e32ade356c886de99 Mon Sep 17 00:00:00 2001
From: themanforfree <themanforfree@gmail.com>
Date: Tue, 24 Sep 2024 10:20:06 +0800
Subject: [PATCH 046/235] feat: add xinference sd web ui api tool (#8385)

Signed-off-by: themanforfree <themanforfree@gmail.com>
---
 .../builtin/xinference/_assets/icon.png       | Bin 0 -> 57667 bytes
 .../xinference/tools/stable_diffusion.py      | 412 ++++++++++++++++++
 .../xinference/tools/stable_diffusion.yaml    |  87 ++++
 .../provider/builtin/xinference/xinference.py |  18 +
 .../builtin/xinference/xinference.yaml        |  40 ++
 5 files changed, 557 insertions(+)
 create mode 100644 api/core/tools/provider/builtin/xinference/_assets/icon.png
 create mode 100644 api/core/tools/provider/builtin/xinference/tools/stable_diffusion.py
 create mode 100644 api/core/tools/provider/builtin/xinference/tools/stable_diffusion.yaml
 create mode 100644 api/core/tools/provider/builtin/xinference/xinference.py
 create mode 100644 api/core/tools/provider/builtin/xinference/xinference.yaml

diff --git a/api/core/tools/provider/builtin/xinference/_assets/icon.png b/api/core/tools/provider/builtin/xinference/_assets/icon.png
new file mode 100644
index 0000000000000000000000000000000000000000..e58cacbd123b5887b34fc8414d8b57aa801bb690
GIT binary patch
literal 57667
zcmeFZg;!N=&_2BB=Fp+Qp;5X7Bo84eDIwh=NOyB60Z~d2kZ$Qlx=Xr2Iz+m=f7?gj
z^}YYXx7N#A2R7$y_8l|#%r)1{5T>FegM;}96951lIaw(+06?Pt^Fl`j-<h*VeGC4;
zu$R?#0ss<*KQ9D1HTpyFO#~-383~|dh++%;1=&JeQ5*ouBe8CcQ2=0hMovmx-5p`~
z&LfsyE_Es7ZeXofsF1sIsA#)jzofZzE3x^F>qfUpS;Gt#o@*O}lzgVthoC1P{6Ynx
zQlXK`k%$=049MFe{_XPgaD5}vh8aSyfgzu>m(X#y>7V*qni_A`RXr>y_N8wUWz1ek
zOV=jVC3#7k`HG!o($WpI57KgvEi|83FAWE&<AL-0^YsfZ^Y;RXBE+NqdkLS10|<XF
zf1x6y|GlE-r-uE#0_qSD5dU8G;-LR~5X=?^|M#&Y5+d^7%c#fy_sIX-OW^;M2>8F#
z1N{HRY^c)zXS_!Ed)IMd+`h<BJfd}2<wO{<`)A-zk>Uh@Sbch%CE=*B0D(kN1J{th
zRuC=<V=q#<yI7p@Wlc?BJn)x8OkaXw|7Sp9934YF+l-@qU+R~}l*7u!3CsuR0`HNR
z^!|(nr^?;(?$uSeZoX9DP}vfDAVy3#^#5ywH@SYvZ#wtth6^qyccC><uTCI`<#QpX
zr~S|849qT=?DJy*S@+7#3fcRCC*P|tpWLrv37#XteNq3~;&hpjAsG#fsFMu9_=twT
z(0BzF-Fb@aA+rpJ-~X?BdTyB~?-!x_%xgJXU51{~z^2<jvwxnjeEYHaqQy)6IaORr
zu&HFmiGx)DFa!CwExTOz(_$kRXNC{IOSrj2WxF`Ua$cYVHwgdsMugV=q`+0!eJd`v
zuDY(#P^Jv&xU;Gm%}5OR>rKjJChof~Tn1zgXJr;xK9wUK^CFZP{_{k=7$lD?)UuWr
z9!qT-_zHh-YPiCk!FWpg!)qVgBa&{9ea7(LJbnd{P!}fsdf3v{aYu#@zoQw$;8MwK
z4~Nj5Bkc<PTMV}}R;1OZTf0V?wqgd%l0}^0kb@B0kln*fgmrl1DW{28Yx6%ZMs4~_
zqT}ka`-Vubc0rA{zU7-<cDup`I{faRYlKPLyA~QrKc`w!Z#VUQfe~F@^9W~0GV}%#
zccOavqvy0c5ZIOZpMhx;wCVFHAP?0n^mc6IP13Onf`=3erLDdwe+AXbb$`L3VsB6f
zJusN&f>zZ`icCq4J4;O1l)+Gx8g#okadwpt2mjy@z#cms@^C{cXOJ?Eu&HT(t)IP!
zazplSzhgI9wXh0}2YR~Zh#Gwp0}{QYlHyW`=S+v@D56(68f90}!Bzf`?5P=&P2wx{
zwul<!gL<!?(VoqaES?~hLycOH{z@kb50~}Au!W%8^V{Qayr#-4>GkG08amt&&AiKS
zj&0W0z^=t14%(k-$%c|pc$&67-}BB!X-EwW1pYz@IK%O_{&9<&L66r&%N3RC<Px8t
zs}0-CwG=6T_T_g`8GwT5-!8F6$9*-hs&4wl%wS?21qeL(C)>{S;P;Lq0lMM8{7!sg
zocf9MnzK?OZN4TY_ESRK#u(VMFU5k95OHz;!4A~29r3t&rSB{UHUbW!Ut19n&@ldO
zkFEO7uNE&wv;?h(d+67u+(8UW536ayF29pX*#<0J3aWnq2&M%7)t^<p4Kx+qZaLET
zF}xW(a4jMFJ}eWyH_FByRB7lT+k}Nn_HRtKvd#22u7%abD$hD?7@wHp&Q`F7`$-L;
z3H1J3MU~3K;KAD?)te2rp%5(mF?b=d-~9{98w46eT-v`+5g6%}7r8aI;qe~8m0jDQ
z6mXWH7Ht*(cTmy3F0Zm>T79y5SkO8%eK=#WdDb}ur4*nEHgv}?hc34w{wFgyQrP^r
zC)M~-jeMlGyfZ(}HiN>@f7)gjZ}yU@*2!jmq#+ZiU7mx0i-RM{*_&$$K_~q#OV54>
zELdvbuimG~4an#=tttyEt!6ulqH-c5p^<vu{MiiqboAd^r>KF)eUvdteJDSR_KQYR
zS+r`zWN9QAWvT)Ebpix?mB03>)t}b=jP?g;zWU>E1->)-qg$$6hB19RT<t*q|5g;R
zyjLi=Pj|VD`lI1OmYK9uz-h8X(1teqY)dJ`{qRYce@wmcCJJ!>2a*H!)ZhH}C^6X=
z#<WX|QX&QL3AB}QYEy+Ccl*Blgb$!m{uT1B@)X}074-N#bMZCj@=cdU%9RzGB2M%a
zM%cGJ%Baeos5?v`;6E$dD$0;xIQg^q)~+dS{^~7L2KrH|h(a(a9=P7WCjjmf>?&x>
z7X5JELvP0()#w@@5FB@z5ZSw{Wpjr8B2>UDwlSY;*Ao|p^;d>7T5pezeiIk?-H61H
z+Pz|QCZ;JS4#k8}(jAOiW{%Pz0zrQj!pN{}&-=;I%EKk*XuHnF2R<&S>KSdNVQ@sK
zE6l*13>O_>`=2{Um_-Bl#WMA_RzuzU5S})=M*zUZ-;IVDb|Cz|hp?;Iuj5`p7gRxa
zbJzZaw1e?CT|uLy$U(W~aV}H)L<0b){QI=@($T~cgC;!ow)!TiCLkO_LSuqm4h5?I
zEx1|2R#SF}{d00+-ZlT?BqDt+pNwPH?A?o8p;W@>4ic}uiJC8YDewTnzZM*R-wru_
zAJ8_4Qt)uJv*P5CNo~=2noU5HhYplI{wtDGv`N9qmTM9er;FZ)ZlOL5O9$bXtIsQX
zeJ3@Qg`kb6+q<fWfb%~^_wg^9dgMSAE_mquonXdGV1b>ZfdGl&?+tJysbvL`SELh@
z#SW#=n{ASyix|5C`x^(m#jj#m&5MP)<wg6LQtRU4?XM!yIx3D;1gMuc0H76P*Za0h
zmwuWr|DLIItA|E$CnFH=XXUO|-nUC94g$qbgg`nEC_@T*Gs%~{FJ6c=Ug+)z)?=f{
zolLP7K0`*w!g48R!^y|8q4C~zXf-dVQS!8=<O9+le!7ar0$XvcTKeRa6Gu=0z7Ov7
z2-@4W^v>4bB65>#nbjx^1O?l^s#eBje9#Z5KBdSVeb(O)HGjAF6mENvK;m%7l_s7^
z1?*Cjd>#JL?U(5ipgJ~`C`m%4n3v_S<gI5VK+CHBIy*Kbn1nS4frMQ`grh2)o3ncu
z<IyA1dRq^AXH`_`2^(JFq4a`RDlXEv`O{x!Z16AUP=H;9fQyoyn{7Kk-Mc*G-Xtwo
z&6yRYs<Z*Acu##<{FBj6uWM}L)aMF|JQnBx9u8=>@8H<M;v@EK-VCDK#0=7WIrUoY
zW(-8sI|?{$4*ZRR>`S<5hINFPQN=EGUKtVt1XjkcOSKn^Tn*iWE|BOlOx36`7YLdy
z98^o+i#a{?Rp0D?d|S)xSr<<LZJZBZ!dmJuM=*NmC<$raP(&$oi7C{b=nwTbA$#zj
zfdD3;VjDqL+AiJ`;KaFP<|M_`#75P+B^|`Mt7ID6(4@PpyIQldM3)a>X46Pbz`S*1
z4}DCZSNWzsDyN(Xy-K%NNh4EODZg8+aa}Csc<*!chy=TV`r2jjPIp{&RH5$CMe(`;
z-F<f5?K?ACB1XE;84Zr786zamJ%_Ra9wEI<A-?uq_2*Q|;hph9C<Ndso6C+20?d9F
zY44RgF(M=Or9x)#NN?a$6DP4Zs<)HF`H4iJp09$S#V8VEjg@{fHc3_c^X#CJY6TyZ
zJwr$1`zNnky%172@>{pUc&0gpy05EDwSGMdNAK9Q_610Di<2n{NvQ5M-G>ohVnv4?
zCJ?WmmK}c6>U70xf}aEF98nZTCG(Iy?^D8Agtn3mAb~nN_VoY&L*+$Vg!-~oF|hkE
zjT}94V@IC7TJ|ELN@BS%G;I>@8%f!`<lHNti=Z><6Po&Qd{<3OgcjNzfk_$8?IJz0
zOp__egp5N;;Cz`aL-$H5OQ!Q22Z-XR_B%@9?4Ps?zR>5Qtls^`guPAwWP?sjP_c_c
zSTR^RC955415_4ByTCt(S%-jz43A}9ME_PV?!5O2#*tEgYz3=k%PbVbJaA)uX2n2j
z<1<zpzJRuU$eERXvX9ocJoEh2e@#X@SIS-N>E?_n6;1@E;Bop$Vm%}Du$|fifH5;a
zEMK_HhdkW7G@V-ADDy;*8|$#|M&PTKUTPU;`#;Pr(F5@3wcdzEp+lk?knu&ni=Jg4
z9)WQpEWC*EWmffM#_!F<(Hk}Wt<!JpQ?C$d@D1PaY(^H@xse4*psiyr#jBTM7TCzn
zG7}aL7cTh(BmK-#!q<C(Di06J4@#7v+pj9S6R-{nTtfiZE)nEhvbwohBCg)PX&j48
z65}0y#)z=O_|!FQ-dbShTUMoh-1BPado^Gtpyd&OtzAnsaw`*KYlwa-`q+(KR!8H)
ztrWScdP;57`P}zHAsIXQ)i}}$B+<3)aJiqy8*B6YX3*l6a-B}5eL#0y*-IxV*^^Ii
zRPatc!GC9lE|ncBGK&O#0*FZ2JEi_w1u^@tXJD*Sbwi`Y|B8}FGiaCA1kGi<W{6_4
z9J-h_jtrE6a2C#IMmNbOMA_)g+{l#jN!}t`jlE~eDxtc-;GKuD#8FfZ&$95Q5Q5ST
z-o?H3O-!~))=ShF#4{(G!t$Ozd4cp5MOtk*_D5MZm;eg*(upHmGsj_*qte&@?@=jO
z6ht+Zx@@IOOLJLU-cso3TcB~DqH$ae3YDH>X+{HP{K@cu+2z4IcW6ZLj#`h){I$wp
zfFIt-F`L|l?Tgzwda~P%RATvT0oOS$?l3XAn|oT!ijOv1uHJh=$tR~u`#n%nUf*O>
zIZR&PmkVN#_EpF93Q$^o^X^kFs^(N5Y~5SJ6GlU2@$>Kvdqf$fdVFswJAHvD<X%_%
z<(1bFeW(ZQLyzx|3@D8Cdd5CxH$YK5Nso#(7I{Zl6NHcKxibPyCeX%PPXvPQTIaRv
z<TS7x@HAAVs5&>dY%UwRmM9ErNGF1?6ZzwbJWqc3<45-OJzb$(Q9zQ&!(_*Ouh4)e
zgm{jXtfsYO%b{v-q5AWO{ukY@T1ds&0T27{jyOVzhMldM$RDZ%pG}bgSulPOI6MT7
zrfv|57vr3_gnT!>8>v*9XkDyj*cHquaEJ8yZ49{(L{7;Yb}=~?bQ`H`(VaGg{o>ox
z!UI`FC&+$(lzFw?EmW<XilHA}Zo0BazfL1SXXKa8qjz^H@3ujQ!RWuC-4*oYsBIt3
zL=2F{1&53}wC1hMmFnrn935!zcNLqtTHbK3DfDcIId&jgC<N%0p$)-Ze6Y}Z>*q^!
zHtPBM`vcye)R1(4+|*y#$&hqB(5%H@T~4U~6jV;v=#e>8n-dfsvXBHzYxV!_ja4!n
zJcMS^WOXMbOBO<tr14Tpb|;*msyfS{QmIaM@rT@TZZ`%Hg^iA9zq~BIgl2f=vORxs
zKQX=KRN6$oTM*xW>(mk+nOq;$!$CKGG3|Qj`K-}~9gWv5tcLWv6TJQqnbL1!T&eeT
zJmazB_RoNNrmvo|MmMiV_$U0_Q@8!D>q(%Ew|xxVwAo?hbhJW^_8VNU-9i2TyL>!W
zJsb{ZH+ED(ugN$ls;9?*$h!OC8-IN{p32k^o+!R^x<FO8FaBrR-8J;=ZkOav@^}V0
zT{l1M&sj5qezbl|ul5*Af*!vK?IIZMrSAzve<X%qfbWaln%bdVEHGv25FF6YL)7S{
zAw*rJx~I3qi%ZuIo&-Y67qU_VJ+5)(G?{CJ^h!#205v5Z^)l=b#7b1SOZPG^1zMrD
zjg$7mYS^q7b49lX)Sk;c`@u_2gj*-JbZQoQy3IMrCm}*N_S@tPFjVA}yV$&VQ<3Yd
zLn2Xb3|-`15mr(6y10i9EcLrjrO#4-Jybl<#pKd<-uHg;nrz)Psy*nDd*Ta>f#Jb9
zL!px~aCln{vnZ{<ZBns6e92wfH0-RoNEA}JuPubcR}LNOd!_D1Vo3KfRX(K4qOp8{
zizem`2b!S57i<aRrt&8)MZLYTBmK3~5kX0GxU_PRUW7f*dF=^>;#)`w%S9H$h+YHk
z99FKN>2{Ecvj`nN!L+872$!JLVpd)(A-605cZR{?K0i{h{0(%@w(TZPbN!*d)ok5#
z+d4Cl%3}`-L0kJ&1ttQG?}FbxUD_|9Sx!!jeLD1NymYv{yq%Un;J$Mf&pY-xWZ-D@
ztNYf~v%F?64LdIbM!mfCa4#-rx@yoZeQ2|q%{GI|4vo-|fha0;yfThuqWZ=vt8<fD
zg{u%lNZ^2-AK9kSl8xSTB-OJu6S1+Oz6NSV=;~rx7W=PqVfN_4v9qw6)fv(H#|*`b
zI8{iFXJ;cA+Kn`pMp7fuP7((k+^)$yRv7Ca(U~D9F~!0Y#4`t*3|jFPu{Y&X_j6k9
z3xf@#*Ks?)LuYCU@m;FWfBND%eEfDi<(Yi8z4a&FgiBJxTASf0=yoLT_m7(wi*Fai
zj`AyfANowV=N(1iN)o029f^>evOG!qo>x_cPAf+Ou0dA`h$;2)7bF;!Z#{T@B}R^V
zFnK-Isozox_@jhAOc7(ghJW)iefh<loHEO++~rHrx7<DkX76JP&=}aGBcpFBo4G3*
zhXh{i*Pf=pH8*k`VF8p4Uvv!@#|DZH{CSqWHANBjQ~X0iB20A&;)L!$5_YALou*w7
zYwIpojjuHG?ZW5psnb_FB<#F<+k@<G6JFgq-l8`!0?u8KC$(P677fQ@wYfZc<%dQz
z6`!`?Vg;$j10jU?CSy~uXv{RhJX06|*n)t`C(Qb`2dQ9340m&I>CDnnIyY%biRX!n
z6o&wkc8t8Yz>ZX(W5`S>L^t5k=(V@N*Lr>@KlQH16vtF_pDltH!&3BP@+%EY-CWV}
z@cap=8|USFyl=POePVb`>B`6NO3CU9J&>kSl6_7@4FGPS<~5r*X>U{Z+ciG5ytjgB
z8Q{FVUxV{$nVVQpe;0T7pnWzYT5FFwhH5TEF~j8g30u^3?qa&w*GcAh1tWtWje;uj
zU3%Q%%_HsVb$g28UuZE_kCU%HoEq?{bvIYhRkO(zx?Ls2kk}#f9)3x0Q(ga%eE#-r
zz~*69Ui{mmH=zvAR)mDNwgz;TjXJhm)@MmH{7&A@B4F4KSc;Dke<lmx6XcSh-+#@;
z{(jPzrr4(Z@bYwxo-tYp7nokIK)n7?;CMYv8dh>YllNHci}Qvztlv#d5DL&|9Hp$K
z+v46*i9FI*cJ{33?>Nun*P_3ACO#Ig1E2W*L=7&rGCgVY>|-M496E&8Rd{5-_msz<
zgho<c<*0&nC>I3*5i}$Qj%Qg|&MM0@kQ7%^g;KyCmz^b@<FyJ;wE{oAz86pWJ3C~-
zxyf1m_4o6ix2ASJKM%7Q8XxFupp~&`J?2CW#_hu?Ae#JHrb7Ej1<8)e+euH#BX$A7
zMT|`JO-7pop+T4_(*&2uSli-mW*izJpY_N=JpoSRx{L}|N~GtfQ*5>#H_o`~?zOYB
zWPRMlgZDfgM<RLI=Vxn&S6@9BOXZuiaKj&y(B00?GTMzjwbaD8NjgAQ*IX<4tW5IO
z=p(;4>*nyrC850(F2P_`eYMza^L2&u%&2gVQ27ZN_jMOziTq2EpT5oE+ZV9t4{3fn
zcMkFBz%q%RG|aFpLoc$fUoAZH<P%YYaBaW|f*uW~JUlk8m5hm!&f)D>XZZPyYB?o;
zo0oOaMZs;*NProKwjs8R@(xx8PkTV}fk!?qRzz`}3f#I_mQA!~D!93{B*Yecj~YGR
zG}vUlnMZ=OT|{~+zJ1s~-m-i~&{K&SAwaA#5HatkId05J`6mMET{r;D(`CQ3`h^rl
zIcB$O=Br`V=CIcT`fa{R;GL=5_RnP{ZHkmPX7cX4dTYvKjV(2%34+v3hvWy$5&1n0
zA@L!|l%jj1t>~kpAtFuAAG%(6WYciRj&r4>{$!PK&rR!kpZ6kc<cR84ScvMyy6@%<
z!mQJ4v(20LbnS=%2R>>0OK3P3^yzyHde<A>pDunJO@4Z)VSdd0(y|!2zt1NBMc>rd
z#)Hf!)4w*KEBmoW-kOPj_t~#Zy%&~dEZ<u3&Ad)BWU8x`xWAA<3wpI@wJMFBONVLs
zN`_uc`el=?lI)R%!9+Q9z2!~|16ZaKpiXBpc4!OCe)g<V$c;f|KcrqC6CHRb^(3H?
zV9OoVK>nG7&PdePNiku{+zb(K=8rB$d0A<1*@Np_=$Wr{o*9hQR!a+Vro0ekA>E68
zqN?@^abw@y5R2GgzYFnKuVluu(MBf=-i!sZmaQd11)DG~+3GKP&>;5Wfn(<_FV9Y1
zadxlmiDAAy@`Ujmpn8(H7P@?1XiAQYe9D;;7da3R8^U9><^7s+?$;&x>B=cnyZBML
zOuP1V79^xUTS=6B5fTx6gthBgmZi#CK8Jrla0WSseeHTN(p%OQfzcCPMyS1jN%`Y<
zck1NX2>wq!48tDVu!vyQebrY9Z{xQpD58C76B4JnsLl?W9_RkN>(&L&S%iqFj>C%c
zYp5jIR3(R<o5cJ3^@kT|8R1XOS`n$SKq5btRxh&Tjj;v&U?b_~I`xIEBpnjZ%%Blt
zY?i8e=4vNlssXdhd6#`6o(XAvR(vY{*5#L|p|r{4J&?0XB{TZXF2}YR9;oK`n=bha
zS)2A<2m#d75aKF2A(A(rWAzD_dFG4)ZVg7|A_yF7uXSUccLL$f_mJ%>VHV5#5A;iR
zgwP*a31m(4-)yOdU=ofFqSiMi`ecTpXXylFk^xknT=RtiD5yQ|FU2T_>Q7(^Pk9d%
z$nY9WcHdr-M^a&kjXQmM9hF<yn<ml|S2U=w$B9GLZn4x~Hr}RCY~^<Zl7szHJG#c=
z$(Qy~(S7%1xEEMk6<V@e@nSLneD}}9kcJD6ZCpgL1Bx}4z2QTl0qAJo*Q{F8gL*R7
zB$oKmx~PS|Cbl%bPg38mdbzm9svCn?)V}Q#J2KK=q&X$WUOXl)^O0sB2`&u!BrnX;
zXB!h4JO{|97Je-yNCqB|fL6}MdCywQW$c+R7F*ae-M(hVg}dfcEC!<MPs<&)e{ve#
z?qZ=(vife5D-Rd(Uo09OA$X#>ic_O{J${hcIJ5m-E#~#<*KQ^O?fnquUQ4rpR$DNk
z$^4g(?p%py`3fmxr6A)>4FLf`Y6n6|GOn|W()$}Pih1*%bR0@WY6%){P6FF|es1#p
zA6!8>A|g24re}N=Ti!+dt0HhzkSx3c>9})b5{l%XR#!FIewPZw1zH{!C!wUL5>tf&
zW3_}O6$V|#(H{LCb>fpNlG)>HsorDb-+4IDzY6Y*_Nv$7Q-RR5u_ffebJN#;H{|_8
z)hf5*&rYp`358{)U1mGYn1)ndAteRMh&$M<K5&f0;{3mY|6jNV{AJFs#{7<g2SJ!L
zXsb!3U5b7EUr1@9>hA`HKBwL6TnnJOF^N_Z#;=Dq5_*(l_aMQmkljdJKFaJ38)ntl
zThj8V<qYte>u}K0#T{|EPVm@E&_-xnuMdBAa|{6fTco4K(fbL+pi(DqC()7W{A9`H
zy1mwUt1As@Q|Ba6^#0)}=(Zw&=Fabkx%Qk<m~yRc_~<C8ck>g1JuGU{GhwA5dO_Y+
zW1n>Ywd3UbK{e*jQo)R8dh7>dh|JqT)hA?sk`KTYNQXLZh<VY>YkyDN>f16}WQa&c
zNa8E%l#w?38S*<rAa^Y>jeV1{FYPv;HV)Mz-Z2yx(4(;%P`mRwh<+F{KVE9CXpku4
z40VwB7-6hEj4}uE+F|=de%B2DvHq}T5RJ^&@>iZ<#^&%unQ4mRTh52^9z<e?>E4_F
zlqsC9QGKK`6yhi#->7?TR{QR`9aXPTEFlF1NCv6nBAe)6*jl%>k4qN9YK;!6Y!a#7
z$)wc_&<;A$g+q|jkfMa}sHyqy@3q?1BgN?!8vE~P$D+b6Wxc)$cduD?hA<>cU-&D#
zJ7318E8kaV4mrF0{@m-20Do6?a&&j`q?SADfRduz!OmWS6CbrC8IR>6=cb})b+}n-
z5#vvQ%@01_>9-i}w|XUYyM60aRlWB=6h7_ru(fdGDQjDzm){xc<q#bVS3yyU8I`d1
zd|!&~Z4GX{_O?HK4zwcp!W=6zFV50)o`IRT3?tX*fGPb!{Vs2*xTLB{&t&1*Syj6{
z^rPJF&^{}4_3ro2fL)Y7Bi^4LOj?>NjrCctVPI{Tl6q}olZ8cobQr5bIx_AXsZA%n
zXL{lDQ0E89dkpAxtOhRNrfh8&mVVaL$ymOg-~W)mfcA3*u5Z8P+!?JB<jx^hk{l^V
z0Ol@$V>CL%DDwXAhy^zIU~FDhPERO=q<O~Us8wN)T0b?+#ZD&&U2T8%b-%llJQB4h
zj1ju5-w(Pkvc(_CJbG84XIaTG73C{jFdVVtg7WQuH0NwgpqNGs0yA7Qgw!zDo7?3y
zlqv`1v2`d&1kn=WL^wTr?C6YIPvr%npxzPfv_FkL1pL;;o^I`GLYe{?Z~=iloxJ5N
zayC)shqNWEv>k4A(mhCWEg^UEr(Txdbhv9Z?ebk(RC(<uAR|}C4^qz?tOoI(m2E<#
zh&sCJV*Dmz70#&VUQ;ZACYAN%&!q+?0s}snXZAGXVMZLS1z-wvfY^tGYTDwgnd)qM
z=pC9_lJG*<n~X4Y%ly<vwm?r*8xM=f_a!pqOiaKz!7U2n#=IzWq^~F&a^IeJhSITB
z8XI_`*E@+>&g?2F@8GyUj`E0%@U*LiGF8*vR2+co3d}U#2VR^W7j7T;G<bE9M2~Vs
zYNRTKp74GCzAkc>KIF;3*aOZU#OHv8HEZxwSG#7y&6J^cx}O5DEfy3FY}NS{mp$5N
zNM0qMwv8TFD4pe{oqZFQI&D4K_EpzkcF;tiCcK><ez(!Fk$$2ha*yJG<ZH1XM2By(
z7DR8|FXW`Y5YdJJ^0RG*)b`!S)-c^dDX8~kUg8D)BtPX#m-Q9wY<wvBXLG1tDCGB*
z;ggZx?DP%a5?~q$j-7u$HGSG0aDVmgH^&DjiR9=?!WfkxqEufzFmG}S;8C}dvkHQU
zze*6gf0Z$O)xH4ZAH^c5XHn9NV<*05+lgb}SGn(u7kMN^V*5mZv+oFs2L)>B6I${v
zt`3*#C`42BCnCRdIdFbG`!!UcKa7yit@hkif_9FU1%Fk<UfrU(9VC5EQU8p@zjU@W
zbrUo7n_-eU_O1P{x;H)jw6XJ}qm_am$g6=Ks|qMzeY>;9hdtCO;Fmz_vlx_6M`0^N
zA4M_TH}UmaeA>3eOcKe|y0d*NjYdx-8Y_E#2=OcO3WAY&6B;6N_U}cu{PDdZ{TSy9
zdD!IOZl%GN36=%*k@-l5kn!DpFU8zUB=36KWfvl4Dv7f^MkGifd%OsR0D7xwEd{|c
zcMWzSk8y_BzBX}!SwFVfqrlIU!;iW1&?*eWczRC9SOlp5TMG3vRv1YM3U48D@BF$M
zmRHILh}N#ZP;^Je;NvZ|`&$l@tu|7R&*IrXQK+#3Af;Q1DDMpYB$sqk@L6BTGJlLP
zFK6vX(Z5zVLOA{*gAR;5Ho64Vc)V!RSxT?fwQX#<9-km$O}AFeG&X~2v$6;6P-F|A
zUh10Qlt?r~){#*s3A7}vNC3cHc?}_<IMZ6nf(Q4w<`&jC74izJ>Ik4VAWrDl5e`a>
zeq*JJ7*!MT&=uP8h<v$2IgE1Pi3@C!gD2W*f8h(Sm+IJ*j=?KEBMzhKH4g@fWp5w~
z$I(?4?duM2!|WAcX?)a1D@6@fW~3r^phqRlK~byhCC6kyu=)EA&b$-rGgXLf@+kR&
zUvXDv_QK+ryMNtg_^wd~(xR6x(}MQ_QVPb)A?gpCLPYaA)DL%&p&)9pweB_xvpQiJ
z6F{|bsTZV?+L{Bk^JS#@DAPeckNWK~lM5;<06PGI0YPQ+w~{zFdK*@`S-o8ADSCgA
zjb6#rA=hWwB}?|AfITEUa+z-3b3S!iC>hDkUEbaOwJ!>6*nb7NOd54%uUiN~|MI;u
z)^z*EtLI2xDvvdL_`1p)OxC>3$mdM-7B$ZgErbOx5(5?w);6Cn?onV80;Kky`uWKb
zvu;a_4}p<!KqAv@!RALTp->m%N7}i10;mp3vS11-l5lb=LcJ-)z-$KOi70Oz_G5>9
zQGiVmY8d>P+WYTk9{D^*N4bGU4_tx_o0q0kpQ!M81yGL=;4Hu1Xb4fZ<vgBC!bbdV
zU$1}Ai5$lL{yA_;d)bdXHPA>p^65KL{VkGVk3VuFmz}f&4oGp)9={MW^iruX5V?O5
zNp&L;tt7^@IZ;VPEdS?B!Y?ikK_qhh?%<WZ&6iNhNL=!KSNtvHDtF?i6oLBYgUl5Q
z<>NJ9YZN@0q$otBR(p+^z3-nh->s8|_JJwNQ!4O`K5`WnShsbNywl6Emi}!t`~(N%
zQ6jc}<T0mj9Z)Nwxg*#>%+V5~z=F)3zHB)yiH&14E)IBtH5^7S^%bWkC-!EOVtg-$
zeRRG?3o$Od{JB%4#mG@vkVJb<cXq?$qaE8n@gzx*WivzB>$Ec-3w(hZxBMaZWTnMR
zd!b}ge@G~XxmluSl5oq7XGxS~{WF6CnIH|mjYr?zVXClXs-pw|G+`m5b4vI4BFhI^
z%gyS%*zNJnQeoWlJY)>~qY*tdC!@9g*rT|za!W{nCn-3&fogtlkT!&~d(Y)HXl189
zkEF=W{ec-LGgVki`X=G5FMX!S_7Pf+DxahJNO=f~$ayfL&ym;@;}9o6t-Zf6Aj#>M
zA1~p)mFV|x{Nxk#y|R;WwKI65IzMc5^DE7A93%uO!Yd7k{f^0*y;muSQ~iR#Y2v(p
z1pM~}7N{a2nnZi<X@HbwVav+SC$!2B@9asG3BCu-q(5l0NjQ$7buf&nx?5dm@XDT!
zgr~ZSO6MXIM{%OUJ2xllp~F!R-5v$EL(`9xyBnV{<95h-2OEDW3bM%ojqOg=v(?p3
zG++}0Jl3q@pL$E>D1)dbCfTBFzd#s##1W6yUd}jwGio?xcB%WJu&T-a?RXU!T^8u^
z`j*x&A4P-?L;(>;={hz-4H_zzKUFy8jjE<-(9USl&SG2#tj%UMbw%QBs;61vWn-}f
z*W$>?vDd@^Km>w}9%{bM4)XFbw#|=6+Rv!xe&k}KZj>}cSwfH9{XX$^M_^o^E-{aV
zF(E681x_Xa^e+~5GiD8MLx0WFPBx#<?qh;Nn|8LYeD>;#{WKGPxuNWIGrjTmrB%fU
zwIddeXgAxhAJTlj0KgprI9it$uB^=~s|&iqbSM2}zRfx0p8cTEC*dBJtTp|mCjF1r
zC7x5FxUfKiwM##&Q>a6L8zNWXs*^VIlKPx!^N(LBx?-8~r>i4)32BsOX)g}M?8MOk
zd(1xzaZwrzwtIT9X60&eJ6~a-T~*HH6Dm{_D6JsxP9em!8S5xwa6f5sYXJc^L-llW
z1|PQ5-X7t5^P1kb3!{xV&~~mT7UdL|l+)Op#TY+H1vdy()XfONIyCP2D{mJAHFLU5
z+Yj0vOhd4*=_g-~S-V$<uBkt!ez3_Dpayxi^wE526=pDv-EE&<*PN!Cy}W2uTiWw`
z>JBqE@SRrMhDQythV9swe$lw_Uy2{21w`c*k1&oIdoX|zF%XiM*q1%$Xwj*q*;@VX
zQ()(kpRfk-DQuW3_pD%a%BXe69=hL?%Im<X4=Dg4fqvrfh;68^!ogcAZ|BOK3EJMd
z+|+PW=VH<=4f>7=3_1UZyUTtK%Olm>t$1(*uwV@w;k@wdg`6%+Mf6?YuFDkZqy8La
z#`W%!K+9^Pa%g_PqJasSA&;ZfLauChWSb;1z?oeaShs{neeF;^Wf0DdPvfv3DChk`
ztj1gog3Av0V1PS>Y0*-DZz|`;p-@mXx){~Ayl{=zUeP?>z0kJhI_#4R>w%tZuB5@H
zSO9CIjllfdH9z84JpSkoVRD?cEKHr5Y(F!eJbNvH30neF1@&tBOgfE2M^OAYsUB#9
z@Tg`Z6mdBCNa3>^!wc`ha{<|fcrY2^t~ZkLxz+^9NZ#HlBHK}L+b;D$FJq0mQ8B1W
zbYYK7bzN@zOkSFK+QwnG2dSrbZ6;DCSMDP=SZwhChT8htDUi+W9Z@#D14oTsM~bA7
zOW8TJ9);XJOxDO!?8sH>(KSQnu~u{h06V|s%9!lrNXT!rNi7*YtSBBoY)*bzgZ{Pc
z)XF$l0RJP_`_9T=DG4JN0Q?>q5qVbv)oN;(X;ue=GOD_PvV`^NoApX()Q_fOLE6>y
z0y}|ap7rna7msc?pNrGI3V6@d4gxP=^b+bT<ya}-7Wa6&*@WbAv}RB!E=Od+1{je3
zxZ&P=rGsxWO9G@HfV4w{5rW!WjPpT_I<9O;?+L|TXi<uP6Qjb>Xb?)z8VE5)XvAMx
zfW<^`jB2rac9aXv-WGdBk#C7jZw;T*D6*PJIy!@f_ztYOn0evydGkC3VWGj?ce0`$
zP9})N_n$&M!+!-9x5b`XJ%0Xb#^5`{dEmvJz*$jV;0OYcGTM-6r%^M$d(Bg6oXb)Z
zs9WxUZMFl!kdn=;sh@BnHzb(OL#EBXnkW$*#9XH(P8C)#aSp{y6Ob`63k3%T4;DDE
z{@C70%$I`%47|KbEpnX2@zXYWZ!@V?#uuCLGT`WN8W*Nj1*k^I+=U#dwI*RNCiA$7
z)|Z=dvd>}m789BfU@yK?3Pqu~QFY&seh+wnM*9xuJ-rcW<J#504FcO>XOAy^V%&iN
z&2h%#v2$^<7G4jnvCy)b3OWeZBO%7Omt`aVn_!(q*~o}OvFL4Rz}yLup3UybsJ*Dk
zgA!qvu~`BGi&D_3y{>-8(uL6&8W6>XM{PtJ_e}2l;^!*vIoIdCYSh1IM90pJA@ig=
z$9xgO^lwjPpMapLVI@biakt<+%kBHc^+;`UTO=V3yIdJ^Ujzvg{YlMi1OhMv22F7N
ziNb6iL!07fC1g+0y7;C)b&CId?=!Ia=yx{6;b6Fi11C3)8RiX)Aoxi?hw@wGmjs0@
zE_mBISOj;c4W8T21)G)?8q8HNCM7^Vds^y!oT<?%h0yBn%Es4x)ld*S!XDkkBfe_Z
zqXl3OAh+P!rd{^hd*V9<$0|ylle>=55(E&Wj@PVOYg;u9IYIEmJ(zz=d9xzo_yGWJ
zkiZ1iCv{KBYpH6?(e6z8r31ZTPtgVAmmMW}I6`!+n8oo1Ic)qMVdmb7PNk8bl|9uT
z(F3@`f+cH<%qWe&<ex+D&xC$0)lv#VQ6v1VzY-<Z5`IyV{Z-D|r?}HO_4y(N??Zu%
z?F}-psS9#YxUYlkcNx$!1MgoDwckkgAsk^M1K*~qcG@DqYK(QxagCb_7SJizf!MMp
zpJl_;g+2W&xchhD(-efHBCW(1PqViIJ7&I={0W|in(=ZQA0<kiv)Gr74)rX&C@^f4
zt`@s7xJ<CCzTWRcO-wRLi8?pcJ|G>_k&%MZ$>}a^*EvdjjU#A!NQrGej%ns7@cL64
zbWRM4J;@%;nYKkfeV9BEy>x#t=QTj_dgjWgEn^<{!HRAQNx|zH@!;}Tgcs0ykAR45
zy{6SMh&D}Xo>`0CVN!X0BeJ6rkI0mkvO17w<OL7J;rL$Aa+WdVy58cF8jv%26O6lx
zXVxH@EH@y@GakEnS#{;Wx}EdE+|n=F{#Y9cIEVh3MEVMRVi|ra(m|NWW`&f_vCN>k
z67joIa32Zy3%P5IkILO8$7XMbEWi20pA{6h`}TTHW(#NEzxv%nCx(HBO!4xy^Ms+7
z%+hr86<sECk7rftxZv%PP(lj-P)uAm#t)O0sF!OBX1XHAr}qkCXE9(Q@byj_6S%HF
z=8gKBf(zY9se6XNQ=E3N#x;Hq=H!BOhZYXuB3J}<h7i;lMH_{&)xjIw&_eAi+tOir
zQ94EFai~ogYx=}~U%3+%F<R(*TyZ#vt2nB!FByM)Vk_`nliBtSybafKoBGKWzVw^1
z>Mnt=x&4PPO-XXp%vyAcE`f%vdZ|qdUZ*!;Zv>h_A)||q5Tvp1_F%yW%^ACZtbJq1
zFKev}&DfoCgbi>%JSo7b;Pk**^)zsMO}yBzH>c0krV3d%LyEK%l|<J^m&xWK_MFQ&
zDB8}}x3hjwQR8PqOKliGHJDb(i1df8TJUyiW=(WJ*5kcPvH+va&!$I^e!smF&z~Ck
z4iI3b#VD2NFGh@dY!7Swlb*bLt>)6<_(n9aNcE3jeTIL;8F-&$MC@lW6GEtr{wOg3
z3GtjiZt42-;r;gqXAAMnpB_K!MxqOistjcdY6N%d#8bArD4E15KYQ{ih5CMM#RG5=
zLCHT}#7HrPUHT^RBAMb}jgaSGFWY-04vw+pt+Npcpk_<`skrlExi+vR>vor7v$Oib
z?||=Z!|D1)Jcb$8^e&CwMR?_`?<VrC!bfbr{pc%9cpxHZhjR@Wbn2hR0hUAZFQ!RJ
zkk*QND<E}^B0YZug(B!&f#%Bvn6*y%Jv7iC6mQQ!OtFL8?mYEPBq?f$pGv4j!-8r%
zuPbq%`u-+Aj!;(HD)j-iACxRI`bh8l6Sk`m-03n`kP5nHI}|>;soYbL%9y@XILEy3
zLrq^=+10%E^`Y|)sMMddx?q_RK*d1qZs?<^F{hV!`Gw!*<zx~pD4}X{@uNeU+5{4~
zMox8vYufbjgMuATpYY@l5Z!A<Z?jL9gljTEBvc_v1#@8(f~YWDWS&aF9NRPH=KO=Z
z@ab6ed{x>av__xRp#KVEzwxKw0j>UE75rPX@jK2g+Wq#bdG}>IHO0<}L6B~<gj$>E
z_JL_;*b_`VX!^rwIn|`#6~8TEaY=CGPaKTfP14_#Cq|CR38p{3i1yUyhs|0!Na+dC
z6rZX2PMMgzOlkb!c}2YREePNPa53?yVTO-yzp78WN+QIxkxl1c$|`UOK)={fc)Wg#
zul58F@KL@ihR7}IXU~53mGZA&>RunCTGPHJQVMm^#SPW{vR=XQRBUH8Tsg|kI_87l
z)w&8CBoMH{2WJgN!UR&{hOLqpS+q+*ws2J6Qj-n00;?St9bx3Vy(>#yO@3Dh3DqaE
zC>y=Op`?kqz{?n${LgaI1$_8p*Un7Gw?zm58ZMZ~?uJhUIl?k!ntMItN_@Ydr3f&*
z-Hf7!=!OS~=H4%Yy$gf8uA(tfiG*X<?M1VWuw?WfjcgsBBKG<Sz%MYKKwlH{u21v*
zj>I<|I^#01kh<8Bqg>-t<{+Akm4r)-nPU&Z(;E;MTQy9fscV$DUoRw@|N2Tkdr@d#
z>Rn3mtd*_7Tkss{QAj7Fb9or<aq!y~I*Ay#zQ_txcfTaUF!0pqRzZdbqTX-&cb6(0
zH*~^(ap&&29j%1YuLeHF305crgD5dIzZ*#((<7S4NjQRZJZ_ni*0OJGas)IXf;5XA
z<2v3%q9?3i3Qiy$<oy)!MusRy=H%$`#sr~^lWJg!g$7=<rp$UhVxVyVFR?{bD1(WW
zY13%ut8jY7-wI*<GHf8nj65f5aHu@VJyUy*T4iC=B}kLovpvJ4z|>ilx^;sFP`m|s
zgl5@GRdV)=k(Az?FavkLr#C88*p)_Y0Ky%ttR~k~*lOvTX0rJ%>rhVQ_28D-s<RQr
z3M0KE6G2R`&sLdXs6w*e6ac^bbCjk-(%FRssV(2E>qi$r97>S1&$}+=Le^0y!@)H1
z1P(6xruoeJq21s%%;{lUaou}ho?{|#l0EQB*i_Vdc9vcFsf5XFj7>>ksGw%*!BPd=
zQ7hLmZlvnUPnq`1$Ug<*6feQOOF-#g+Jk4O#JP!coYIWvQiMWsdDAJPXk1~-{_w?M
zexg&c+Lvb#_UMO|^#$TgsnDnFs4HP$DZXCJ<(*roWx1%db5r|M2AIJ4THSwI$_{EN
z;U??r3+SH|7OIbhe`Tvv;18X(X8Qb*Cu{m69vpa^kdbhZ#$*-!<t_C=U(12a<4ubd
z?W+%X9W@`%v3{J~CDnjEtvKMGN~4<S!Udo(rVLVcO)xNOXH-*1e~R;-zny?}BoH~-
zMC&;fh4td-)y#{4pO>!Z?HE5@?RC<6zp-UCs~~c_g1{F}4FiIw6<Qb%INJK2<hISY
zP~Z61+3bdd6&i)R+=jh?u0~Mq2eEsaRLTNhk=^#y_GIi{rGtGmV?9!jliqNl&Q<Pn
z1sXqgYtK%VD|#PiI>Mk~r<aC3S24@cSzlb!oOfB&G~RPse7R-u!)udB?Enz~_ZGbY
zDh^9me)>f!;e9-?NhvUCdzS(^BPd#)V)YLUOCTFBhYH?<eK5eEJ^>u>6cmNrq&PRY
z%}8DK4(vC*foEKeu=wQOC-ib;tX&3Q*!}JT=kK-SXH5;yu^`&TZhndr1J~zJxOE}j
zK&0_2I~ie_c2nmf(v@H6^1qmO-35m*wr29!%^PN|3R$Ik^0H=+yJfzmQ{b63Ap*Rh
zOT9*<S>vNc7E#O~vXw?hL*9Mr5B(5v)irW8GvnM+`Fv9h6)hHvyX1{L=Pv*VTDAsw
zQ<pChDpv}Nd5nl|qK|ri=0Xx1TP{1uh+H`=5Y$-dP?2}>cfIEub5MGw#IE#Xsy&|w
zyX)t%xS7bLutx-Bm@d=*rc>4Bx>LF$*WM!~a7f-191IMPQ2xFu0@nPZDXXpoGo2l?
zWCVeoZJb~xM#|{CG3}`F5VbY6bqKA2GI_7S#Yw!~W1X5#`n@bg+5{K%Ef4VDay&0~
z!A!$@rh1J~A=tn;8pDo;o==E&>1@6&fyNXA&PR;(H~2uE>diSs&Q{~wbuo1F#<gbS
z)`kMfl!-coRl9_OVst8MS7o!7jp-&>N|1)Og36+i`EH^gB<<k(p!r~fxzh<n&;WOG
zDiAy17fz2?+0E#_H_+PNA@n{LY=2k)Z#u>VOPpho9ACeskw8#wl0K?Dj_4gDpD6{W
zvZFN>R-+mA9v7e6*kAA4{TBb3&-!q4U1fUz$1C<yS}T$z!DiYV!ZDp;%X5}-PyO@x
z^aNA#8sbC-77g8$i|6=_8$C>E>FtCN0swGE1YO$epil|2PS#4bN7O=5UD&^{8>m1h
z`x$uqBKE^d1<M~{J0n=2l_^?AfxLyxq@rB^_XA}o^Do5Ak{PDsAp?tPa?fOrp6Twy
z^c5yM%eDdtiE4M}kh6&24at$N^#et^&uMa9r1GSL?i`9{xr;2;>*)t#%x9Rg?j#7R
zt1Bzqt_9-!j<BTuWVc{Ue}KL4*{3H&SyMFdRAgNAUaTS4$y$VwFdMK+4ygYF=AYu=
zDW>9ssV~c;H&R_9_NdmTViop=$#0shgwoG$-kHAB;9GU0XYh0aI?Ilh?jJgM8yY)2
zid|c%RR=ouw}tKvk~@ui-I?eMnt~1r-_>d8Vx-`8coz3t2X`cpF79Mq_Eier5Cede
z|K$HB_G(thPoJ*Ci5_pcSALZ_h%3_4-?<Q<Dm*83L}gXVqA(&Su@Caw%QbkTaW@2!
z8{&xZTesTumPJd<6`Yvk+OXSZ&UTS_84>?tAjCnp!U;udADg^0_;r<r%1}>h={+yF
z0~CMo+fYy!7j+|tY-X!<2$GX-J<dRb#>>-tT0=4~^b<^s>sFRhGk7iKGLU1EbMUIH
zQNqzD{F|m$dzSfU@+1ZshqfUPR=Zi-ej|^}9ntlx`&TvQtWttiO!5qwj$Dlh031O;
z7jDxSlP<}GGFhn4+==t?h7xVnwDej<|8rH2uOGS;nkB~mz|*60c@Tdh9hbVz>Xj*Y
zOE9gXHdFTTgoFG7O|i1s%YX~0VyDGHBO4ZD_qWsaczm!ci4*LRv0n4Ipp4Sr_2)6F
zzGbGn==yEmUIEcO;6M#yeND}Vvmt3~ZHbWC5~&)fnPd)Qv^3O>s8$Fv8`C7sHy!_S
zB08G-onda40M#bUG2u*}dXF7w#rqGT`@VdMRZvu;pLlGK%1x2X#4yeHicj-rl#Z}O
z5Ih^X5@%yAvRVGjO^nw)SduAQvZ>P`SVK4Qij8uXHYD3#!g`0CLXv_i@0i2eA8fe+
zyCUqN>)u3Y9Y$~B&GxWg*AY#jpss&eBb{4`2pd^5cdBN?KtL>te*Ad$?9gsljk6{D
z)!u$9FD}jAZljz?VFHBPnxz}VVfKr7N^MPMpUv!tLMB(f$)|i^nKuO$oiYaae+bI{
z1bMGmF&>Hi9tH2<sUU)eK`PMp`%7|7uge6P^6MF0nFbN4Q6FZpuQoVT_XgeSJH0zc
z8D70&A{eUIb!R)bsk4_jcBXX2TOoXISBOM$bCeVRLx@if0F018X#PIsgFde8*L5v&
z$h^(9(T;Zl?(L@ndLjY7gHi(wKR<tEfNa!RG$0sd-J7-hz}i+&2&$2%J#6kBTq)WW
z8fD=5KsmQP;un(UlE<p;+2h0@+KCKyP{F}UV57$*RBJuRILf@K_M8I4A6Rdqc5a+b
z$(NF%)masGkq|!`f6ppOhBSjy|14zo?fp1zyAaglW__x&daE6y+$wy&Z0$A13)!*U
zxWgw8IUvUf{HfKMI~1i}r^y_H)soXd-ppA{fYg_DyrKC@^Xndq*QagQciJlY@qH-x
z^gc33Ps;3!Hv&yXZg==Q7_^d)3c`M{HjjR+%lMlD?)UVGjb_%?)t8ce=kVQBDvm$=
z1=-C?is1VkEdbsLvL`mL;n)!vJ}fvG$M0ea*b2EF9VqKXjkWhPAr5C=Z<AXDIW+w6
zf^b(?)-@>(Ru-?`_P~O8%ae14CosyxEaeuyNUloB_IQH5z4@Sh0Nb$sd#t`a1v*F9
ztQuj~q~Z<BWb?46xms$`_Ruv+IolY-f>E1OlP?tYJqDaoN)Vb-7ACo=GaQZ1WfeYv
zHqE}qIVMQM*du>>Pxt}vbIlBb$0m^no#ezxuf2JSA(sB#7~rzFg9${jfKnjM;T|oQ
zL1sH5Gv_PaYCy&9k})}N_4(=Ywx927!?Yc1{?byqI%0OnF8K%{ZwiEj`V88la24GI
zB5=n%S~D3Idx;vu0*a^@N7yxjf%E`kc0G+t1+E>po(VQE^M_Mmm+wPJbcd<9?A$x?
zRUT_ccam@VMj`h^h<y~#2ow$7*Wla#VK4b(9Q3V0DG-ovKNRo_`z5bC5Ym77q2%ll
zZBXWR_Q5#FHPQd3G1qPA%i6BBVo^;~`w%r%6MLfSep|HYUS^1ga+c;OdTT|-((|~6
zM81!K3jf*yXkCPhiIs(~!qep#U{fUREU(72J>q{6_^nEKn=OPS94j)wFfR*H+3XAS
zDG%BdO4*%5uX=4<o1gg=_7)B5ckI<(Y>Z(zOns4bPQt$lo}u_)DwgUSBqhe$WYuc;
zjD$uoSGA71@%BS78j~)lfJO(cY0#M$4)D#xr{X7C!IR5+uQkg)-SxKnlt&DgZ3dp*
zH~~V`UQda{4@SVlbA?i1s4^s@J`e)xUVlPJt-1Eo6u1;vGt)(4<5D!u^5@(6R#h5D
z=C!&eZZA*dTbeOD><FwqZ`P$R>T|JzB-s?`>z7!aK|mSWA4abXiQ8c|EBZ%w+>=NI
zo4&*c;mw>Pt%aG_5{#^PnJEwkVS)3ktdmh3(#lhIDOvfV7on5UNoczgw8l|*q_mw?
zPhDYPW1q}x_@T|Ycgobh{39XeY92KTqJ&DYLk*;_8<v}3xkH}3p7fjbP+1o3H%%3;
zPl~X5>n_XlZNWUz{+)y1;x&nX;LBp^YB#ZTBjac*F<%*xk|ho&u1Z@TQ!0%`R%=w7
z-cZvj_3xJN;?+5~K0A27$O>rQIgJ15l;+~q#4b_LU;cqZOu@wD<8_Lhi_PQIDCeC!
zc)-_ershnk$*6P_Rr*mg01oz*!u1b*sp&oon-Z0zn7awe!{Lh=6-tSNPXg~@%xx=6
z38O<Yty`{OA=FC7D<9&U(NAq1QaQ%r(}wY$&!_5!3omjRXU*`h!8$llf(@z`#>)8-
zE0O^bV%wv$nb}@E6{bzKYuTF;bC6MDn-s{}T9Ufl<@3<ciu38OzOgLZ0Xh^xh|i;V
zlOc6=b#ve;H+v&8TI4C^p{@&Nd;4l!b9P)-g20BdwUnqDQ3ktk3L(ZfO6Wq=6i8zS
z*w|h~*iYggj3ZII>Gdze>r<V*&PnAxkqiKsD7d(xP)^E&8CCJ-M?KC&)ae^DI;xc(
zNyxa=ihn$wnmy7Gnmv{f^$minkO0rYp&6OC@0G4UuDsuB>9=<V508a`Qye_%bQX#-
zx78nq^gcHo3+wr7JdPjxWsfV0+??GcbRYMB??692>r&y2cxfViK&^7OSg7MZ(flCE
zAZ1lWX|`+fW?^!-QB~yU7Y2EH0cr21z2|-#`%*!KCXS7IV(+YOip+c3$Knlg?|P#{
zu!trY2KCY3zKh!|UMkt@NxOrff=NP=|0V$!6=3|v30!0uezOg_7)%*$YlE+eRGo0O
zz<}CL(D~d%`>Wm?ZdRSVaA{6{=*>%)2H^jXrE8AMvwiz<)w0{dU9FbgmR+lsmhIZT
zY}>}NZQHhO+pXog(|hgtz5TyF*UnBH$5%(%KhER>(LMo&Q0X-f_!`0wRMH^UG1vOs
z4j+Iecv*L+vKK~6r^Z<7RO);y>FB(0ubZi9*v!Wk#&p<;)Yp___v183i}v?8NunGD
z=+947eMJnmH|mXxNM1{oXcDn@nM^0?GhTmBl~*(pZpz<O#RT30-Wbv_plSfr(LP{t
z1dSTAM;7sj$ywaK>P`PpWRt1u5o)<gmpJ-!kgT)otvXPFJ24r4qJ+AF0{(^Z5%cuT
ztnWdEGdCR`x=H>piUv%$mJdMO)#~)eOljNfxQlk0r<PU*;7{Xw1$6NPm!FQU@=KW2
zANnKCS%%dSvB4be!R6R+>GhL7PBNOOx?UukkGf>5D<?6OrG@;be=BYR67R)U);Zs!
z=I%D@(#p5f9-qEBG<n`ey-nanNFzaZyilm%yy^cfAI)FJYUAvS2*#K!aFNA!VtuCm
zCFaIgL~nY%OB)P03A^F*ErP)bq;@W{+753!YAE&p=&&C<hDD<P#BvDy+s87g0AG?B
zXJH@^)AD}7Be*10-#m;B<>~Jwx(I@QZ40vwOVor<nF%)t*OC7#`4z*Uaf+$RNj|4P
z@rUG=d7C|d`{G3n?Iu@Uo)d=<)Z^~SmYaRmA)c1bMBRGZJ$boJf&0atcG-=mUheV<
zNl5LRyHw%X%i3ljHjK9v(88)dz>GEea)ay&Fwi#>I<k+whrWGPsS^0b4ax%!pAV|8
zpkwg1-?lfqeq-hb0;CI<tH4SxLMg1nA5g4NBHuBPEB*McC8J;jvhq_ed2&{{VO1Cd
z9E98-|Dn{f#tdzpS1*EA)Q?kiLxJ;?RoC6nhx3&lyB&X-ZBcSYP}4DT4dp-EAXdu`
zIckzG+mN5ApX1K#6Re*~hK|M51C`F>VFyNN)U;nS-6nG9M{ZIl{<2g1gT_td8#p2!
zGHfuP>*J=d3Lcc!7Uut;Tc|ytCc+ra1~j5?3O*`_l-eHU`K0*_>G^DW_U4KZ(o^L|
zSgpqnIgubZqG<m9P8&O(13#}wz;ecJxIreR^3kbBG7}}LT6=QalkBrwKp%1LEHw9m
z)UkilnPRT&Be=8aLh&M{OBMm3cOQy!5^Hok<_jO`pd~YZJ^PExy|M?&LFGYv36bJO
zLerqWr$t4iEy?r~Tz_7_l}6~F7$I)UyMUcET&Yx^yQmP(+C+5D+WoIVxPFZ(ZQ$E!
z@dC-bE7Lw5O7|jQ1UT~fkkUqc3wi0(3P_wQO}dyS<}I2+TM&|t<dK=Ki)ggcVjHA@
zqVt>5dNgFf#xWpA(3}DasXG%b*!U)Rs*wjDPW!DoYrZu`<&awQiePDQ`gEt)o6zbV
zyJHzXzqFnRGmm)p2(87-A5gT{^=D(@G(?+6cqcNH@L$1&C@V3<I+oJrn#GZp%PNPi
z{fQa#>Xm4wXW~4H=&nk9w8z7haYZ$Z$LpJ=$EZ324~q`m?klqsR}YdxR2jalD1ul3
z%EQ9Ds`$}#?4b_cXJu@pP{$HSQ}EJbi23iD^P_6dzHzlN!`ys0@#M-p-S$N(Clt%f
z?0=VkKHSjvd)Elv+;BT#d&*FLS&~sV>DACu@LYl+ror{=(?cyk^^@n4!bdb>jBaj`
zKrcI@w2pdmlfOTlE%(_EN$e4!RDh1%uwdf7>+>~y%T-C`J)B9*e)QMrtu>aH=hp+q
zcXiDaM+d9=pA%;@yOe7Zkq$mi3}!xa&2PR%r|H$e6|D+-w)%cj;2|m)N@zg&DT;Md
zyjvLiQ2Uj;g*f4&^kK$M$sK|TbH9z$DBh`-6(J^x-|_jz2(6XW%n26KU8#C}a}A&G
z0FNq%be;qSH3qzi199`i&PKvNT$q=|Sfs#R4@(6EcP#RNZA|xObnK)-Sw|KBM-!-B
zuBxphQA?6i*!<^MD>nc~*JPwOnuHld8qcbn*?b%wzB_%ATTb9$sra@chDC{jz}-FZ
zZqu2qXXPigV9H!WFkZ#Gla1VJ!<hFGO_)%EyqvpGhQNip?PC9AAej7H(A{W0sBiy6
zbiwdRL0*&>&eY#+=F|^I+9d+c%IlsI$<c+_2HZ~>Qyx`=dg>~G*3yR0Z<60wGvF{2
z(NI7w*qa#9*c2uF{mVSpldX>mn36%e)9WRb@8Wbd86TOQRXqnYGLiQ^6&w_xTlHqi
zCsUwO&#C@&;u!wREP1iBK}mwgT8?u>b{D0OSh_C`EF8*j@9HI#egvooy;ME<%KUCe
z+QqEz8R6}moCZCX${wl2WrOuKi@}F7C+Zz<OWke%FEF)zdSx>oP34!}C^lSyhfd!P
zSL1~qtR{6Q-lLE9y!CuMh{+?gg4jQYz{Ceig2U+bI)$OgyFVZh^s*hf^IS!X4^Ft)
z-@ga<+Om5j`#n_gsda`#S^jO=*Svsd_tg5~Ar&a~KJdz!tt$?)Pq*@j_-9PvfWna&
z%gFRZoQmS{_m_|~jNRNulBQloB>vMLLv)2pnm$ia`5&M16(@0vd9nVK_U-s3*0Vm_
zFuD^><=B49cs7yzBpy(~rZm8X=S{tZ3u6{I;KHTS#4+@hg$}Mv>;0+k!#C$K{n*Em
zhE9V)pj>KI866_qQMb_vtgrtJ7~M*cseDL;Xa|9SnWdMl%wm-=PoHfk>8}GztR*@`
zffgU;*}5GvTgdIq!(8ub560iMBgm26^+KJ0q*FE2>zkED9Q0dbGJj4le&ghOIlp(;
z&>wW6TVq3pn)(1}wZ@NFgbFigiejyIk%F-KLi&OXZOup8rg74DjA2vn=fc-vc*tT7
zD)w0CQBeL~tYxogS}DGa;ND)-!;i}#XisS+;3{UMNrc3fF(i7q1NE_QgVAbv`x+`{
zX1l&7J^yzmKB*lbGRXjQ+kbV+Kg(p_pU4An%eQ?C9Ig4Ib9MZjIe$D6)Um-sQM99l
zEXcup3thG?ttLK+bnY)o=J<l4J+0zS0NV8ThY=KwtM6QBNL68!HV5T(2u^6D73pk`
zB)Kg~)m>-ziNEYndLZLznA61OvopYEF1gzBpex<)iV}|&FvRZm5!3nIZ+__ASEE^7
zc|CJGt36nX0@6S9F3^!j{6xh;mqpANQ(Ml4QTgRh`)!(#z?M$8<wcW8IeyJ7k=c{l
zJAbra2vBQ)_*IQ9=j_nX6wvjYSZY*S^#fXe8}kme4P}_VYir6iGeF2p!~fuziC2GO
zdc`j5UpHSnNU9iH+kk~ZCp=`{8I5Cc4mnG2e_Pr#U2J$iJ6QB^lM6StErS%rUow%I
zp#&rP!$IT^25DMCV*}-_OGC|PlbfQfYD0J|%3@DcbsZL{BLU#NebN6)FNBR=6npte
zxYBK1<J>wxinA)Jx10^*G1|8IXWo&YjF!}oM0T$)Q=Of_3lVBO(;Y)9kTQWV=y7Jt
z-}7;9n$<3jk|rL}#4B6cI-Gtbl$OK9?1_`*DWJ(#*x<w>vA(m(Ow*AZsJ3jJQKTxd
zvJ8dhicGBknXCbA_jvxg4~a4*zsgT85{e(>0Y?wmBoUci&3N7_|Guw{OkW+<fGQqp
z+gz=FTo4IkBtT0$hy)+hVEtZaVu6+Ud)-&2VqO&J7!II1M~;?CY?)&v`0<%cj;b#J
z(I!G(Gm`&_HyuL?vxmkB87yal2}3WRNoMs`P33Tk8I2;V`MMQe!Z_U?W$64KRzoqC
z?~|f15(W%Z-9I&}-9wysVFv1b(v~JRRW%yTdTbf1&UsPh*E^BE&aYo~JWGZ?$K&QC
zJHc}2{oE^<z#96kr|)BG)hS|YxBD7P3|?4G_|iY#p~?NO_#QdCh`*jh>&$iKnG?@}
ze$&fdGeX#;>29W`ykXOPWU{4RtKkbd_k!W2y!$a%Iy^FkU!tyzfF|iNUE;1scB1}e
zOzA>YnwM5<>F2|_$Gd~5fRsFBd)iItAgM&?BQQ{{8QidzDGD<kL%;axs`}b|(+FYT
zde>ZCybeej-s6z~;5w!R+68hCwso4NlOBaqbbS5#Z9{^He3Ns8HkOb$Ndka~umi<w
z2p4a8*^$1SmGvNbU!^~dE}+J>iH)2{gMMNKPl(?qCM`yti`>&FXZ%ienJgsjay)YB
z=1EuGbPw}Q<!Dho;jFbh_OAocO(12b-};ZUgrC$ijGkyjjjh-{zcwB#F-;&zE~}V1
zY@c|X7g0HJlDQ6E9~)9c_B=`;gX?zGFgEP1A<Zxh+UUPLHZC;0B$1+TTSbEkkn5_c
zO-{C9;@=^8!vcCHiLoI#8fbERX`$@Ack1<lfJX~;&==^)Sp#%cbcVhMoPgW~ZH<$m
zmXJ9Q{L;}`P)1ka>4hI4gR+PS>Ehisq+?eonRfGe>ypiOpLC(nNC8xY#r?|oaCa}3
z#G=GzejZe;p3xtSZR2}g?gl2^(brjooxU7ol&izlysK8cb&WT4e7MJ->bHV$EAQRR
zN{pu>)qhmpbQa+uZM*rtVCGo6bdJ~WsJ?3THKKG%LGBwDS@FvEEq@BjQJZ#04i?6*
z6hfM-nUS%RE$uphmHdI)V;k3~(-zTc!0muCsx2mu2j3r$JYE(_{bOfkl-E|N+2>6-
znKV3h{`2Eu9;kn*UZgBKyQ^&O5>rQUT_JUHdWj_s#f)*M!iBpGQ5Zvr>XoNm*Lo%<
z;4LmFuHK2DpnX-@ZKNgur0HhKPE@M@gESLzE4C37g6$(wV)-w~3<y+;PT7regTT4p
z-U=RyMfHBZ>#bW0V8hUCl&jToFaz630{Y8^<L<=wkat8g*8F=1pOtlA$Rux_@rw;e
z=nX}%jD|yB&?rqudyoL!t=~rDw0~ikQ7XY)z~ArwBArhxptf)rFjd+4BRX;Zb8VfC
zinUy4blT}gH*CuNIP7ab^Ph9LE0cR6r2NQDfAcPA34}+)4s3JNz<H!Nxw5uu;v<mr
zm#LGqxyKV9P2f&K#^^xMs|0$J;FlppAy-LUaKPRZbqY(2*|m>jRy}DkHWUKZzyDb=
zn?Co(puolS1Ivu;&`+~63m9<Mk^^)8V2#f5f?VuviKbzj&f^3^Td9>G=($-gNER#I
z6;D<|<GEDzHqbBL%9G8#NAg>Nr7@sxI=lRnNO+Em<q$sPT2roNPan$v&DXV<|Mbq4
zMK)L_Q~K_G2=_P9_MTjH1)w`9`ZlcgDHS(&1V9Se#8xzZ*ICCq2_!0oso6BUipTsY
zTygU1{MIj&gYQ{ic8>?WMG7S1D)a)C3Sd<xN}Ay5AJC-{V(NlEt`D8Vm+lNi&O!zZ
z)g2b32e4;+!bs>%pk)Q>61zjaCaD%xL^dO_ZrNZIPCdRZ7knZ$;8tZLc`^5i6kt5Q
z3sO)YWBu*++iamUQqD*)bW9V-rC{HVYrA*}kwb6anNF7LhL#*1sC;uhgrWM7yh3s@
z60<*rgqfa-@pX{7ViAup20SL-=sI9ATxI*tqF%LrnaThDg_{4K!es;+db!6J(0ZqE
zemWWpgHLJ2nOc`>olV<xs_W6+D!=dL4ti}W;9z$@z~55v@dF@m>u=fgbCT@5n0~L_
zmA@llFRSsEk=mZ$d-D}4%}oSRZ?k4{p{}3NP~~w_H?WBg$Zx(6kifNZ-h%RF!?qqs
zmJ8^(-oJQsb-Ibtxh-xmt_oah2rqi8)k%&^)%$&Gx~LcdS}NB_y}t!>KYEJ~l3?1O
z$>$kW8=rSR*8fGow?Q0tU8p^=1!qeQW>Jc$oJu(oe^S;Q_iJ2b(zbq77mpuY_}yAQ
zj<Qm-%#`PR$Bz0Okx3Q}!Up$*N7GaP6j(d<7ES-N`L@d}wec`qwUDdpj}3DtKw$j;
zy{9cK8lv?Cpf#N;22+l=5t?g`@h~UF%gMmpcPC$CzOVw;DWmpKXAL&5FIgO(t+N08
zFGLu~w3`n=(x7Iiy$RT--}R;9Hq9sn>tGO@SZ5<&_Jm&jckM%DIz}o#mx<KI<e&Ym
z+J@nn{wL2k9e3GHKlcwA7!b#+WEjAfk@7~Jb^EvwZZ)l2q55W{ya~!+Jc{t30`<5M
zoaq9a!;TYM#U;rp7}e_byS>M$Ph0;?`I0fYdUOpskkx|KnU2P^=YErh8Rm3dPP09f
ztj(K7)-toUoYpVC$9VYJZ>7n#H~~xY<RRa_u5FlFf^YJ^voQ0^xo0)T1pnzmuVX3I
z2qo4h5eUf{s{6(s@l5LB<d3icFBW5wdDVryi6Y8Xdrqh)BoroZyNRy`LOM?jQ7T|U
zS&4wOs!h2*19Uir*jxBwOZ{iG!+L?M=oY^44^5Uzx*p15Q6n6k*%zN|SeRwSgfVwQ
z`z~}Hi}c*W<8BoUsfA;R)n_5aI0aFPkA<BQdKaAbGr4R`SQhkfb&ld6L(mJForW|!
z73X>KKyC(tZ?+C8x=?o?fV-8Avs}+1Z$AWXG{+IA(1u?+Rkrq-cK?a(8v(uB@m-w|
z=tZN6hx=;QDe(Sb>i&EpC~RM5+?5eL=A?%6ffLY|=dnLLEydn<CR%XW3g?_$dV==J
zzY?86r=SII;O7l_VGCb8O~*F?$pr`;wb-_pVVF>NK)5w2TgHQdP^KVwLmz=ceXPgP
zCh$;~YUdLZ3V0@ln5QkZogiAbymYcnVh{uf<7An6N(01i-8N-1Y&k5q%vhOa=TtNk
z%Cz?_k4PQJAx3U{2hb?XwUaHx%L;$)=zD^006oZ~T_x#VE&?a}`SgQiPIDN4`8YzI
z4U;SQ=FY13dEfhcukXf7J2z_hd)HU1Enbv=vP4q0z|U~;@vuNLTs6Gte1kIW0IU~2
zr}aDO5(noH7mr^4r8r6(U(R>_Oi8C<F#w`{%Y$O*Di|!_mAKbV857TV`rhJa_3pa?
z9J?|li~1pdl$Gz4!ZV!E4HoqzLDadEEiarAa)UyeJ&Jy_-9lJ~pC}?_W1N1ab&<?I
zyx5@Z0gC5(866$tgy{VETNl|q@?U1^rc?ow{ZME4zzgw{7oskwE7<vzl{JGgSqA2G
z%M2v3=2rl>+fD}!yo?YRTjeTtWO(pA`HCx|8@#KKyj|TiA9ar3N&W5fq_?Q@t(pvI
zMJRy39wUqCsJwYF76^Xd^>6v}qblMV-?Q2L6K(pf1cdidr7$8N?Up>U+joB#G-qP1
zHFZ-9|B)Bx=OT<=Cs0GOim<;DW*{zheTA1unD7p!9z_Ave2-T(qE^E@Z00rf-|F0X
zvc_@cz(YIQer>R2A~&|Bt}+oQrjHXzsYInO_uzQS5e0v>hW$Py78AOXwLFGd!+c1Y
z$$eOW#G;CjKTifa8paU=Cz_E25@Ls3?}@pf=X$PdJTVj6cPc0diF<_5Pj*ENe#kPp
zMY%-F^LiJApg4i@7P>h&${!)Um0)TmEiYu2M1~)OvG?j7<QY~@=%p8(T1_iU&!N7Y
zLa5;L$=?_qJ7J=Rcz}j-C9UP(_bo=0yj-sR3n(&qSS7O>lbpM9e@hCnT|nY+Hk4m_
zbmvSInx_Uc|K7Zk_QqQ03H_5Ivx0ZxvQBU;1~JVMcJm$H<)op{CKhOW#*;GkrWElT
z5v=wU$2KldFsw(jJ4wy}zPU;h{$Eu`Tj8p&V>0jY@PzU6a8o~Am$9)XlZ=Y`FV*2H
z%hlnGrZa_HU{jjX)kKyyNJs6rlz-!sr~S?0A&rcF5spRrNd(R{Y(~#{lHBMf2>@R{
zv<%}de*N&bLt2zj3YtL+>6Zkt8G#1&87#1I4tW0#`|^jNf!-rnjuKwZM)7UC1BT5B
zNVNHK%5t^S+Idcir<^h+9(V)U9?R#-+UHxrmGf+SfY)^a9;n`FW5IwTBlrOh-%nXl
zD;Q(3kBhu`(@Rs&L(XfVT+|)hB2TMu9B^0%wVggGbRkcI3vpEAB1;{GZ)x8@orn7B
zrJuIoYL={ujtF^}=O*SKmqj!X`cKHsTBwF5morF$B-c9XWUKZToy~@Ejdcw%5mi}x
zVmtQ}3UF)e$+sJwbhH1ZaZ<*(0p0RieWV8s5Kyts@c3*8z)_Q49T`G0IaSO<T%M;&
zZ2;_zr=w-7rlpBs&3&x2D1IN<p=D?qEgYQW9~3+OTAML-0ImH4C612xhl|A25aucy
zKU6#LOb?|U%#@1r6Gmz<Q3k^g$HVRPa&i4^4=rVOqM2&MmMcNdsCkXzJDA=&>TdpP
za=*LYIGh1!8>OUISLEdya5+EA`sWP@``6vl;1;j4Mwwuqd3QY?ArC|bun=;1oy`v9
zJf3QJJ+)CdzQ)MSD@Xm@xH8#cN5(+e7JT!yVj4n~Jsb%yPR{D0($GC}&u?&eCn~D`
zGdNPf(IuyRsU|99BW52=pgMt$Un$x0S^$)9p|E6vzRNt^*Yg$Q?<{sOc21hZT_jdj
zwA#a0LUG87H#!Y5XFg?q8T7=NNT>99;fDlnOGaj1*SWP-Qx_Gip(Il<YWuh5Zn`e2
z`igSRpi2;gS*^W4`~N5Y-WyoP7e+ABLc-`=Q1{Nm0=fJ)%vB%oOiJJYF#?HTu2CL3
ze4LU`BwmgxtsvVk+m?r`qA1mF8%Si|aWc0Q{PV~GgNzR{X3p{7WxB)MhrL|xTgD@F
zPRB?qdcl$26Nqa3o;YH3QGN)$pd}r256!e3|EJrl3LklxtP7xFYn~1#<h-VdqBRqB
zVm#9djn99u#(Vnx?_@P~*<oHPt=!{xukeBmk!~)=D`Pj(tOfo=S16vBXue!gQVLqy
z<C6mF#`0UE@0bR&DoT8@oN%fhHRg2j*fWl0>1v_TI&Wpm&@NY)=l<TW21iv{)49GS
zguF<Lre^(ZRpq>h6y;3s>`&8~`7V~KVUXn1T2G<QD1n!PBjw3AO&P|>Lw+W;)2*|N
zDlLkAr&k?x)*s}69T#5Ts<1A+Jw#{_L5vE%z9W>ml<#zBLE(4m{ui#5)1G$K-!cv-
z8*HM<m>yk0o~O<T*{ZcXd<8Ro1Qd`0!q7X@SLHZ1(a%mjQc3u|`D$$}+sATv_CC~t
zKHmRi`Fj(7Ot`-+gl29OJ0?Or4|Dvo>@lzFeC+hyP8upes|ZhL)Ur=RR97G6)|}bK
zdRSkKj=bXn?-$2X1vsSUoi+>fOni(^oqEL%a?V|e-(7!Zc^9(sU9Hm0YX$OWiRIUe
zv;R5J-VT3jUb>81XCYc-dt9gc&cqwG6vh8@FIe2CQfm+WV<>xnpN56@%*eT*9E`R{
zy(~5uncW2t+1#K5i{w;5g0fgfl7Z*ym(PQbbb~qS1hdja_3I@g?=sdmRHUvF4(L{S
z>;StY;)QeMQq*fcFIEdFJVNOqXQZuEy5HgfjK;%IZZ`|%uL(e0{7+1R?Q9h--uWY|
z+*mTP6HF38WiY}<W(+cacNnr6xsTOQRMRA4L2tw8niJIlI?9Kee4QtE5qYT;1QPqk
zRNXftbv_z#`gQJ#$=|kc0l;!v1Fnx`ySBp$ZKK!p>`n1;VBfz+*7Wg=xCwjNbAhZM
zpFi7gO32IBIfYHXcym|KV(_{_%2E@RSN4GQS|vTyPt@=G8cHc)7!i~!aoTWK#t&*_
z?$;!Mcz}bTIF_558r4~^s|vW??8z0gO!Ftj#nX-IFVH!l3o<Ei+Jy?fu`++YX|7Ro
zq{1*wjrp@4$38q!hxLiK837sw0tCTVQT;`M9E16Yk+0@-8K+uOlsoTSrRXI>>#KQA
z2svRJYUK*{zj;`~;ZvkQhf!NqCY{*M?e{OvUzGkpJG!|hiI9Hpy5&;I<TsNlbXY%%
zU+lS$H9Q^>dE0O}N<7GGt_mcR*pui^uW04$iY`=#quOlyCy~%yy)<%>@ja*~3Ov2{
zc@5g#f4Jv!UCS+3mMp!7hObA%Wgt5J9JUA(;~6p0rZ7hcvbApci|CyJsL{`mfI`|X
zM=GCdAj+_z0Hi-XR23Y@*U{{OV{2!}bIZ61b@Zj}Q!E@L^CvB22dZ_zveIqdp_mQ#
z%A-sVAqxRG`fl$vrzc;ue*a0kKOVXFXcysTx(qQ!PlG;<bukA)KAkErCUT4jJlpB&
zszvdiO&?)O?ovU_Rw9Ve-&9sTL725m6Y0^!w}Afxa}|`6XYqTPzQaM&V%T1r6UgI;
z9&$ZLz$ZTGBmN}R9t`k@#fSD&F2{D6pR2r*cz2J@Hi3d;USkdah+n>NFVt4aG*!qj
zR#<2-<L}X56E)H-CxFgy`z|yNN6bz0%R<Nh$zN}z;6km1rXH%;q8NEg1@sDY`sLhr
z$Tc16h!SwvK*JyydC7moWM}>+yIrQ%*1Ubd%0mV%!-oQ7-2lKuRiQQg(44xEaD<-j
zBS&am(YxA<=hG+kYIE>@`qt0~iK)0&e2p@)22U@GC>Ig>gL*IdMd*aihfLHT0vKcK
z*(;Ce#h$w#=PR|9RWFjv%`LFwYwE)__!G<Sky*G@-q|2JT8N$58C<=}+`Tt>%szGE
zjOUr=ZOX=;xnpf@pg-~M0N*4{b?@jrA<~aLe&s!!)I~blxD^Bm6yckk`Db<8jGg&x
z0G5a|E_bYhwxcvM|9b@Jmw=nz>6wM^SwFWRhr(25x6E5@3i1MrfX?;G+po7h{G0BX
zueBZz&&;zOK_;-NU4h9DQ`mo5%Mt-C>1n%nMaLgSW0{$gw8zu*KF>TG+@T$u&ES`+
zW53w@!%#+q&O7c-CdtY-m8Q7ke%<5oS>&)tJ;wx7{K?sc#?I_Sz2!bK`_%NWk*L+l
zEQzg=9j&r+EYhdVY3Ub7TXi_x4`r*T`!%`}Y1jt#^sKH{`IGHAi8}F}SCr<wP3Rv4
z{`~e*i%~Uw{$1F1XJ7L5I#O=5k&^G04S;8)p>6=tV-}cT4o#v-ihc&+4wM#^p?|q_
zDiQfqq|#Eb2Cv7YKRxXV`-Ci~y=lavsoGN+m|Xp=agiC--3>7Q0A1UsMMB1XRh2dQ
z(n{HKr;%m&zCmb#j!OHej$-qszo42y_syzjU%D`p?#N-NFyoZr%P|wf%t7WG%W!=B
zpZJ`g#fCrUBhsTqE%x{PljbC2O-9(i_RVx#x9Ca?p6TpuARYezZ_(*eTnyp(ybNz;
z|4kI&ESM0-6`GHjC?1QDZPun`n1nm|=Bi+|z3WZ#s`dL@l-^JD8?=Y7F#l`d|Ml-j
zG=NXq_1FsU_^)JU!bQ=CG<C^ZeAb3&u{{y+g!lG#O#4s{uLWRM4;A8j_V74gEHNFB
znbLjpN^tma*uE*{9NT?j2B>sQd`?su8{$N7=)}zXD(KZsfR1tQcRDgXv7ZtPsCcEu
zUYP!=&f`XU!|UBs0Q*Ut=T-c-lXZQVr_2z9_apGzW@&Psx3Bo3XOL`I`NNw8Dyug2
z!sJGOogLoCtUQrI-osQ<m}t0vp=UH`F|x#JACEk9*!d;H&Rc{1MBDgXY)be!UA!tJ
zrh-~pvaT%5l-E`69EocpAGgd^H3@mxbdF_t-or#ZQ6jQFi8cQ!t<M;?k%m~zB-}nF
zrl00pl)BhsHjw5<?iQ7^XVz24_YfNJqv5pOG#P%S55sPe{8IdxzJxF@(G_}%4uBQ;
zhNpvgA!D7&Imnw=#|J#E_=6u&CuRX@*e``XPou|IY00Ia)e8jIm&^M%FNcrZ6i0S%
z8zkdSKz_aMQ_=|=M{k)bu}Lt-OX<jg$R^7n6Se2JHPf=GZuu&{1cy|brRfB7x}gFF
zNla|0;ti+K-`JiD8^gQ$3#F4C!U1fBNJJ<Da)E?c6FP`~0~RcgW=J2i;}&&GXV+YW
zdV~15-3#y;>9?|A5#VkpgvU_;+uFOpR9OkB<q8ltu{RbM>_VZjAOBLvLPKXyj_;8Y
z$*nEIV)01JTynMZ-S}qi`PaHr%!f<d+o$;TQ4K5k!6PxJ9ympWPu=$`Wf^r=qdPC7
zO}iA9Qh&<8!l#@27->;{5=Y(?oWm(;JS%?*BiS#|XU5bJyL4dweZr?$;Z4pkh4#<K
zxuAD9m!(1l6W)rK+0_EW;J*!690#KA)Z3Q_on@u!973^RzK?vgN7a+~)N*R>4~qf|
zETrF#mX9m0_m7`uXTseUv&GY*l`M>JmIr}{Tk?-Tj=*i2Ex#<;^*prUUQr<th9vwK
ztlj1hxfjMTm)-O_sH#}fygv#rOEV@y1Jp385&>;gMxde4XIVFPM=@F|e?>f6A%*Ol
z;GrN{gu@WIOOv0;vmFAgf_-xqbZTuAJo#?iB>szhi9xlU57kG}si_ctHSKC5f&moe
zR525U70X7k%NG_4x{3@sAhENMcQq%8o54ct^{1l8f+M+TE6JH>IBJl9n_8T8G<5<s
zijt{kA^pTDc|jW?QLHMF9*2Jsb5@6K0V6q7B=E@ajZfQUL?Ub>Kc1B3xi!vN43vAM
z<v&PBCs;ILfD+6Zr6AJ+rL)z8a~IcINx%2|fU+gJZCLO27W4~R!R2NC37%0bn1*;|
zY!GSLlB|X)`L~+CC@0s#;E7g7`=HqNFnJDBz#wRv1=KGNnqIE#>Nb#Qer}_WcL1L`
zu>?H)Vn}Jlpym*rX`aNteGkh*devM2J@u|YQ&4V}C=C?|EBI@Yy(=WWVB(ZM7m|8z
zvyVb?<ac?+iuv|cA+{SlMrd)lvfF<eqw>d}&Dt}awLdjYp-~b`#5B-FFWT{TdaE`v
zbKrBWZlI**=S{iO2BSNRWTL^NT3LWf;q%)S{W)wqlW_L%>d-(($>}^C&ynvhj%4S9
z@LO1&nnM@~N{%3N%3|_~r-JTez_(E6cU^JX5uir!iE_rM;@B!fa3#c)CC_kASsZ!q
zdhK*-f5>a>hkThWiTSi`5S@^KvZD9N$B)znys_w+D?1pEZF@x%_5C|G&Ooh;B0cvA
zljk?4_L7eKNJr6UO%5I<3cnB6XX()kwACIpmUKC9j4Q1DQ|V1|fwB^s?Mj{7Lk*5h
zqo&-Td^vu=p^1?6yH_|<!+%@J+Te1!=b)V#Z|{K6fi?=!^_h;6HV)By%?B0+;@d4R
zv^xkdla8AB$C`HtZRh8Hs~+E$1e+nzp*!bMPet;=9mIy5&d{f~){6JK0YFy(e(pTb
zD1sXO$MI!hk@FoHmSIpG^RhP0AJVvo`_<|Uq=E`zt#*sESs~5F{i*b&ut|-9+rfug
zOE5Tr-jV-3dYbS-DQslnyI0<a!z4~7Q|_fbtAQGKhp$;~Y(5jRYMbOxwt%DV2>Dxq
ze@FQS!;f~_qvu7BmRhVI6@6$w6*xg3<{P;kzad!hmFZg1ca|W!-bbC>Y)O&G2*jTA
z1``lubu+!;meAezW~Kt7Cx_Fe=^<Tx@k(<2!Xh(GnuBN5l@n|#BYYZuHz_I#+(21G
zz?R~1f(?iEb(~&_C`_Inxnu#`9U40{mgf+b*zKxMcbhgQ)K6VU8)^xa2xCgs`s3np
zAslhWPkF3AxCWw;6W+&r=7ad30XMf7`EvFuog-gdKZ=dac$1B=Pk>pBWjknUX8f!_
zkK=2`29Pjq^GCLih|Yo!A7C^zKfnB`pKOt@G?PZGum7ODQGPT<;Bw|}5BcK*D`hj3
zJ7bOSrnUA~ZRmo(({Uvvb7EX=<^-dWG3>&M4h1-3A>oGT$lSIKf_90pD{AC?-uvDy
zN}M<#^;;j-Ge01W86F{b^R<hr6YofAbXLxVd)+pn3Tk=VzpR47UO0vGFReiuMq%e&
zyS4C6WPDROTxR>|eS8(eMV}2V|0^}Cjt?<PL~&&p(=U$>yJoY??zykMPs`{Y44wr7
zzb{^;p@6Hb!?!lT`DDcN3{!40N0kFcLpN?+u89WW?prS*S?+0QQX%S7QZd^1*&rNZ
zO-+iHOT$cG?&-c{?}13fb6aC1<lk3enG4XzZIan#DiRfy=Vfy%0ZlKY(sxzCg#Jz@
zW^PFqKr7u&Y>|&c#?|)LZZ0*l&P0~VU*9&BnnKR<rMMqqvoG5QCQDLZ@Ti?ji?oIR
zNXuqBMJ^yFYuru$NhGM2?+s?^Uu}e&#^5hDBoE|re}f-yD=gbZjzelr3w+Xc@9U(G
ziGVM*;$7P1(>;=8!iQ%4I|D9q+<y7tUxhuxD5w!a98Nekkcx6-andZ<h+C~2eBD5_
zuv*Xy-t^0KD1D&jzodqqdXJD}ha$mKJkcB&YbMnu(&AcGj{*&I1vDu)l)Q=)G+J_e
z*xY1x_<0-KU(ik!&Ca&gJo(yZecFPHSE)uv0gSihS$p6AVLOAVZ4=o_C5bZN)1V_a
znPw`tV7n$LLJK}e+VNdtx$)dg-fY7P5oYQ=uJT58x&N*0Lzz>Y{OX{H_UnMb$_}Nf
znk#%uH63z7Hy4|-3vvzS4ZGL!VieT`IXl4~(ApkJeRC;}Y^jFYUhhk=nvaJfYXl%V
zE(fytYBuhsd3`Z8FJqM%h;QXAog5$aQI#vB8T5_}aXz$z88=z?j<6H7foK#LfbRyI
zvDy*eEg^+)y(?S~xQ1@o#x<Dv+pyydv=V)pKuz!Ee8U9z#Q+!=@)1pkL_o;?Nkq30
zM=N=Ej3jR4Llton6-mX1vdJ$Djjmos!NJ9u2f|A_c>Xl5;K@>uQ96n1@@z4U%k{9r
zz%d!}Ec(<|VyEythnMI)2Of39WHsAMO?ifdGf}AAycuSYo|4=jPFK<NU_-U~_#D|m
zR|}~TEKs2v@rQl?x*ozY8JXjVcgR$c*ih?R1Sstv^yau0`?>gXuFf7?I&_<rXQa{x
zWxJ4|mjN9qB&D~|*-zNPOJQmt2B=N|IfF+4$=svpXq&jimn@N!dzT%WnGZG2#o7th
z(!uu+vr3(JEYMq68^3;7aRz7EVfZ)--f<$$kDUKdx7t89R2Tpv<&*#Y=8u)sHz7~i
zlQ>6HCQ&+1<J`($&vsc2LRV;kXeQbT2Wy{fyWbBl`A6Rzv`@m-RCTX!BafrJs*C)C
z#q)iv6PTq&<OJk4`fwgqD}TJ~(&3a<*z3d?^ObHGtH*{IV$3E~V;eZP5qrcSrAJd(
zIFu%qlqbBdHn9a-h#3#hObo&n0u$&wuCVtaqA-D_*jME}l5;?d&~!C+-spVM(JtKQ
zyP_t-_?O+0RcSKuQ6}T^m$mfC*8cwL^59Qz^CDG=i_;qG_@WxTso_EjVi&daq6cZ^
z9BG;~?s6+!2eLc-KPAOo?3bilq@MzX4|8dRp@N0MPG`R@IGq)RJ0fACzXWua0+adD
zvRQOODSsJqtuc7)H+&pY>6Ij&RVOE-9!QUFHPCWOP+SbLOyw8NfUKe|TrSyMu%JmH
z^p&8KAMkzEIbBrqj-D`gJcOmBM|ZP#=FSYu;cyK8j?UM^<5$e|yT;JhyGZ4HWwB|}
z9o)=ZmRMH4bWtU#S?%DjoH-<xnN~+Rc0y@hT5r0&jbQQ@3>3JrgOYKqDic{3&2Gbs
zg{jTR+Kv=aw&|#^9Sx`bDhh|%4in`Sk}0l5S&G+aGJAF#@CfLfq<Xr{%p-4fn6}bC
zEU>A@JcHnabz|%7KmEWL{qnjwG&#SMgt60vK`Gid#XLRo9$L?jgV#T<wTpm{X7&?Z
zhMKpvF70=JoDn#|pTM>1(Z8#ERn)RQeb(>2Z~)KXPIP3<SFGjJVBZ?DFSBM|(r+dv
z-E4dr`MqEw==ut6e({jfg8tpj?q>1ypuk_Cab-B!cXavTd8er4`fO6t_Ucih+br7s
z%sV`SC+3(G67KRVL~4jsN-5_>I0u+EBMBHI#qOAP4ePxYTa53&U#d4Io=Rk1+R*a7
z;NX&ojA-i#acFXmJTmh)cyJAFue(x(!*g9KMU|n#J^?QHs!9R@+MCc18zR1jNP=F9
zt9x(xcrltI!9@oI3ERY6A2)#T0Cb^b(vLcp`!X^oXR1&)bVAxQ2Yk)#KpdbtFd~N_
z-^WAbdi8m_lz+Z7=JnxCM^QF7ue^TZc6?#rPH>^L4zP9cCr$Kl9f>p+-pF>t3l(PG
z-@J>{eORST;Y^PrjgEgmY$|qvjkTtkccN((t@TAM!PxcZ{H?!G`rdxEZ1DJ{(PC~Y
z=#!16`W*H<hn&&&Vj5b$w=^9`<=3&okT?uX%CQS*whc&miO$u{cIJX3o@n|#Y8oo0
zBFSZRD6DTF8Zu<fi?b*V<4dy>7B+Ji1E+|uhnQV7k9(5elvt8B+~If!)!)a@w_+(?
zgD%`;2707z5#iBA=D(>E6s*!v??7j+)^~JGpf0=&&eQs0JsqL0(L^r1${)2_0?6Lb
zfV)Z8_C~9SoByekUcos1NcT;^G@b6-#7L?v9U)(OnPtL+IADL{U)f$FaTM;jGgJP^
zb26RymWvf3y>DkjFYrb#UtW-C2p@K><gpvH%}>T?K+>cThaJP(K|bwCHgJji;Sc@{
zOG57zLG|ql#0sL`*p`;4jryDq+|rU;;St;_)gk*_Y(nu5EhSk65c^y4jtzplzaPf-
zpd0^fP04Da43!`>7bR-_4%g1E?%!?v$!vx$Ivt&=j{FkqH}o8X{!mjeY3R{$I`)7(
zwlNQj+*;7v4RBR;93*F=vN@0Gyr2BD`FJ5rC91{=N(QKUI=NH#XlU{R%aqsDxKyPJ
z+i=Yb2;_D0O=@2n1sM`!D;$V^X!o1Q`H~HiAqH4b;x9aV$B<005-CRmuxIiafJm$r
z<!Isziblq7BljQ=A}K1-?TDz^zi%gA*<v(Ml9!Ehf|=Nseb38;+ke}}cXme_P1$dw
z#b+m-m56ZIJz-=KtzkO=6XCy~|HA9g-~_I{(40PHwBr1b8UO0>EFy6C1^8G~no0W(
zAgV-_Ma<EGcznj=FWA9DfVs@l(3k(dbHZv7FbH^Xoqg{`T85l}g5XUINE${716Z@M
zu@r%H6aZLp2muf>ol_~+kjJAR9l7n!RQ`#YsP2#dW{Yy(JwqKADZ`Xl{&|f#NWP^F
z9XH<hcW;$79C-oqL9_rB3mQ(EN>!1$apmPW`iTWD7D~89fM^0RMFuYY!gL`U&``y2
zHmE#$6*I++O=U~(ucM`mTJ`A>;FC9gq-mLfb8TD*;Z_-hw(n`!IiWO_QphuKx_#Al
zU$D7NaWy*fY@nlWhhf}mFU=dC>9tqJ0(Ge_4yUwOLgtx2=`)?Lp+SIrLF=rColV9Q
z522qMs5c?@4h;?(v2S5b@Ap^=WXSkyD5MU1`>Qu!{g@l5emz4vt}^3jm{M5CxY0zb
z&;64yc$gE2b)ut#8yv?TQ_&V8R=?8-Yw54698P*}jq0Kh|1;M|7L{Wvqcq^11lC^L
zB|_@ING^IE&F+baf^}#WOnz{6amikJzW`}!oIh~pI68j32LM<YiU2-~XYx^7x<fw{
zOqSc(wjelA3bW@-mbAP&x>qLQea}-H%xv<{pY&D+-#0lp=;CfQ-1QtZsJiMhhOa)W
zdaO*V$jVA6rN>!(snrVE*5?Fu(eT=k`{8`&9ldTqe8X{`G~EabNyn&!yM&(cOP}iY
z$!(Fezf5(|(2c%9cr&p`yw>O8Y}8IMVm)eb;5Uavi@FCVT`er8a+5af^cJ8`zM9L2
zMP))Z4-ft8u&6JqUngIy@j$QjQN2{V6Lp`#J9k?V?1Q02ROBPlQJC6teQ~KopTyeX
zEKpz9Ccpl9qlMmz5qJ<VpK7r^NX4dBIrP$ZoN%zwCIs|f-I>%gdwwM89$cNyJXHcf
zvxf#CAlr2leJ0r)J{s+-`+b`fClu!fdKn3I9>IhUtzA`7A)}gVx^b>5Hh*lhKp;1Y
zH~7(j<5|(~-b3-Y45D?P3|b~ygA7P&jujEaDJJhTvw)<WP$=sV6kzZ`I#Dz`_V>?~
zjsAv0U*lV~)Xn<jN_O4gVMsB7(K%i6B;>9~@h|{s?z$bCm;XW1>Uos(5k<apSHe9H
z!b=nq^|L+-7%(&eInkj03Y^xIO6;A8h}pkP&oz<An7O++^o&1lBUINBVaFqjBK!xB
zbSg(9NsdFX$&%~iJ9Hk|GCxCU_4qmK`1Mq{Tg+Znl7TO~l@=-1Jy#2b<$s)0I%LoR
zp-Fqd)df4gtteb+Mx2bbx5$Z6<+1H_xX4AE6qW>$HyjHev`PnwDznpZ@HdzMNx+<R
zc)Fw)Y3N0~((Epuy{P(9QNR;iU;6;=cV{GBVtJuKI@k`R#Ra%lM=oF7#G_&e2uOhp
z>qr=Q;tp5Fi4YU?sbVdMx%*yowRq*?&o05={XXdUcXa@K6_MJzL3od=O8i^a^j45$
zyLk|JqH(zQQ35zQecwoz9H*Nfj$kf!c&;N;%IU6S%Df{5wx{LZGOvgwZ1uRo!v+Zc
zmsaa}D^FIb=;4MeLoxdb{!KYiY0=%f&o1q|Tu>Fo6#>fxG*xceC3cpfEo-Wd@v30h
zl-8};2PC21*eXFmR^UGN0ZSnLR-AOLvsYXm1r;F+YXdn<fBVtWq5mBp0I|T-%+MQ4
zI_gQjVmr8<CzOi8<3~H%;cy2)vR{y7_jHVj6eUgvjW;5ChUm4mx%<^7Yh39_Y$y8I
zZr9Hil=i}`XR%H@oFN7wZM{K~;gMKS75Y>{t)vaHn0SX4r;-eQvFw8~wo~pd6ulv^
zRsVAO5J+T)pVk2Ehc&%a;L5~IZ2ZccKns&7!>ybNE1BI?{ZCVK<Sm1t0C#t-BR%8d
zK}g3VdMW!$T_K#rtE|N1@!@bLXAY<qyGF#D(@9jJFVLQbgR|zcnmZAE<`1N9^5j!x
z&J;T#&r&&@Xaj$+#Sc1=Q6J!wzJJ~a(#0KV>3mYAE&;K$xb^?9Qjj6scM(xT)1S|C
zb|Xm5m{TKKhnS&Dw<7o7lhN`~n1m(M(7`Ww9}+cp=)h)X5x=%I0KKESHWP^RCB&U6
zLx##99_fu}Y%;dfS~WtHmV!>x(tspqMDw&C$&b%u)CW7}>p||eI}Qma@t=G7jZ~!x
zy*EX5suJbA4J2(tYgrtY{p8#gz3IAQt^M(R!AwP|6S+KZ0c^d0m*3wis1FC)+VkFf
z&6#vF`xn5AUB4b?8__2&u=VeZ5fFu-r}A45e9-w?RZo9h`lcM80r}421l{9vv>)<x
ztVLJ(=3P?RSz;(!2W4~OPXKp8@rw??NyOXZxS9aJ@{S1lNnC5b952EltkFmcH)uW%
zncTx~gpZuRDfy&Di6isdT;4ilU9ggzfbsq*0CP`f`+^n`O3>Z(Hgx+T2<*(+il9r{
z6GJ?o6J=nVEm8e#k@{LW4wjr;7fT)7D&5yahZ(!{8GmHyJ}C?R{JfS_y|ol>xFjzG
zfwvaW4nAxyIUfb%gi)M2oCgxLF~2(1T?Zw((yU@dOimQZy*qbs`~9o+195K|2@fbB
zc2>#t6<dkZBc<(=xP<c~SOyj$DAj7=)_-WHo^(z16HKn1!F^D*7h8c>XSwpb`M&j`
zIPrWkA(0?keiz>k0D^)-{q85K2#}z4TG~ugw;%-j!(Q(i-ty2Ts+CZ-lkoVioU=#^
zwf+?1Qg>hG!`Jis-pz|y0mZx^ubiUvsy3$VHg&R1Q&q+o0-~AjWB$Tpz({M}inPp0
z7E)zN_j}V|Q6*Prd&#E#tmNS+cgY@O560?b@JW6+iCzsr4;W7zi<J}sX@9wQAQdUU
zH()}As$+5I9E+jI&vI*|Sa1DMq-jVI`*UI5C~)L?|98Bow9DElc4DEZ`5V46nHbRU
zUv2`uL&tqosr^Z^?`!+DIZ4<wFU{YYGP@`f=pGw{*LC&U7vYo(4p!M*GGW#UV5%4H
zP9!CF+fWU|E)XLh@*5Ki76_2~OOFfkBp9s_w5)m>al;)ZJqWY3*{HQ?Hykx6QQ@Fu
zpo6x32roS3zH*Ef@&Xh}f`q)OY`pTMl>}1|VIO{#$@SJ_FwcYI88<~w^@9Yv8Tf&m
zDl-TL)t8G>@Zqjl_cLH+%hgwozCY0Tw`?~P$07`}!E8ohL&0i(ha*VU_MjcfZU4Ty
zei}!Y%&47YHYN8n78p_{+S0s4J@|wt?=#Zu2MK#mOTTm@N5K@r2Vs9GLLjc_sHd+k
zi^{V^u>YH5lp9-2f~8}VLSHi)4YT%0=G>#T4JIoc1y!xJI^$c*ovZfXbDBqYE?Xw;
z)Y9ei;uriv&=1QYOb-w>i$bTZWLOMD9ADDXg4-6w$cb#HD}8C(Uv!R*TgMIG7@sUA
zr+h|H?l{#6!VVpcK-5S4(DqC_)mZT-&Cl+qMh;|4RRkf3<FnQ?r&j0}|9!Q886HqA
zT|D`jN{#WW-<_#1<(K~gAWW5<dK0b5rKxr^NbKAnh2-<c9CxOIGhgO$8RTY{iRk^!
zjt}oKwusO;nP`Ef)|s2{O6MRevkGJs?(A`+hfmE8tF0z2<`e{yt^bV~H{iU?D+D`!
zo0Ea?Gi;*WOsvQl$TgDhWbYcw6)fF#4?GUBHc-a$*R^*76*u4uINr{ro~jn7l;NZP
zb+)u&BtpygOfVdqy2?|y$mA|qqr(>FqPJrpvXfxcPm=4VDVrXx?O=2W6V||aw8m3m
z85;{HnDJK+_$U<Lq?ZB|$-0}7Y{T*H8&|`gZ5U#1F(?9{gcU>yl121r<-ndg6B8o5
zpNo}(yQ68F`?C=)2c)~c7Zb`AkB7=`*I&hKCbkS!eLIRt`C5)C5k&x6{Da^-A+({i
z{6#I>z8b^_hmgx5h6_~y*bnuFdNUcPF2(s=&ohx*u7D(QWb+EqG;77N5yDCDkR%9R
zt=hu=aY-tCIu%NlefKQ~Ov|l~;I%3rH{^#8$xDp5q4@m}lFKtqjz;?T!yeM(6E~~V
zC5(+eW~@*6xs(;cpE<t)!>E8PX}fmf!D>q>N0YpEou<7KYX}Z5$VIwv4L1l46UwC~
z=X0&fA2=MPRwZU$EAWo{lJczi-@wpQ{)qIOqhQ+zDG^g^uITOI4`;~0Hf8n)_W}Gj
z%`ae99R%yA+Z`J)I__p~ThTW(-}@xKeKl@^5E)@uMm*2+v;d1#1nx|qN)k+*n$qV9
z(v#C0%&KYdssItq_7Ku6%ssv5xTJ(Kz*P7io&#0oZP%!}7y``yDY3s??a+BUanitz
z?SdZuayqXw;EBkZs=qqQ&;CA)QYe6R8a9)Dst7AH^C3THBD!BVj)Q3pdFe&(>)UhG
zjmJLECbN6Cdi~H6Mya7a)dEsA=fcF*#LIzm)J&{g;KZO|REfNT`UO8auqhjQWDq)^
z-YTYc#Gx(LkR8p1f2H2iA|p`9gFYV4C}z^m6wBcqFJW(F^pE?zLW6ao==JXn_0-m*
zuAr3a7!0AGB3xxos*pFTlb9yjJX32s&nOLkIS=v$i{nT^uCaPW9@c;$tJ!vX&LsxD
zMZL|bAu#5g3#)8M88<WrGLY61ll2OO{U1cU5Al4vFrQxQANZ*dAycSX9s0to{nhN)
z`zHKa=pH<_Lw63DmI2YPB`$#4$r19gWq7H7{gR#<eft^m4jqa6u4t+Db^>_76>n?{
z%0|pK7w@a9$y6%*{y|A=#n?5uI?j6q*Ycb%Q1KaVXcGuaY=7Lh6dfXf5P{;UcEbC0
z^xW%?eJ&eS`H!l6f0j@=Js!6a-ro!*#v@-nkSzCRS#_xscvA#VAzj9eYLTeCA#GAS
z;@vRuwhpo-;!fz~68%m|>84j{GP3FmvTW{0`t3&;G3d6KsbjSINC1pHyTfyoO})gj
z%{1yV4WQbtgRMc|B(z2c)<A*vO<oWf$}I5Eirez?`od<FQ>JS2vzYvScqE9U_aFy7
zGCit7jwkr#*OSd+Y=6xZTVmz&s#v?pK60l*j*c$}b6MLftViYurp>;8-<;jKf7Q|!
zDzQs!cXuY_J#^J#+jV709WVCOxsyk`pELIFgSO0tN9JP@t!K$ypF&xmw3g?9&M$&s
zT^ia;IoB98Z`rL<=>N6YL#V)H3C<lzYTCT$%*9*!`*NnCjZs%z`AYSlCn)yL1fVDV
zrU|ZEel?QTQAKU^t3A~o-15w~`%!M5g$JbG;=yneZnD?%R3*2I^P`b(e?%H&5e?4*
z?WqqV;Flk_RspF1AzYVEZ3o5=AT;^@O+ElG&agsUduHwN=!$t~I3cI*a3Ai{Kk7v(
zBvL><YZ!s=BbTp><<y5hZ>TNA4fo<Y%3lx6l^3{m%WG79S8$m~k+72yqvjIVT=^be
zpM%1=pbD3a)ZAT9#<=H}HY?cc28HvKPz*^228L6U$cJg#O&xT~WNH=?1Xd`biys}0
zq+OKKlNuWAxR&a~BAlbT$*$(yaA9HL1ev<3LF3oGUm0bQP5Q(G7b%HP6=I)$7CJ?v
zS_?v1kF|O!6^5pqLsly1fBA_?%!&2fP*?$bIPO|sVUqRk6EN+kH;P5>kEa(v>Y@Ip
ztB9|o_Bab`|1#MtD{<T2Sme2%)1rN%RV-d(^nr&=x>&To7QOosif>SSvvj&A-n-(Z
zilk+iapH+49evg8`(R<~99O|w)Fn0v-}5p#u?yUCl^}QvSec|6*}6}_<UX*k|3*_4
z_5ijt5|gkjfDcXwiC#>@n?R;i3fQ`f2$=@yRQ=JuGNAbj{U1$V9Ta8%y}fjIODrkf
zARW?;k|HGy0@A%KDIn5~5+dCV4~=w}ba%tju;lOd`_9WB%rLXV-f=$XT&J$<=;QIu
zIrKNnD`9%BjJR1}7Y%d|$&_Lu`lLJB9Voy@(S~djvdL<aDd;q8ULn-$v!n!J?2&`;
zOySj<om@kK*=Qh-l~#CM4=6eqJ!*%q&5EY%@u?>FdGM5{T7lK9H|P%?$!r$FBP>z@
zC4=j~@}dHq`v4q1q@IB{nb5n-aIxucJD$WiSWpkgPhos3B0C5LR^Oc<8DTFrFSx*9
z#TmS#UH)p;N&lda(gOhAG#R&$*;3qoUJ@`iELP_g@zrBx(m@az?>LTVh!ZM^g4dVP
zD6%_5(;BWP?92BmtublIB-8=;xF<JbspM;@nEyHm*aA2-hfOZ$OT(z$06D^_flgZ|
zbOKM=)<IS_HMK)^>~3`&A<uv}E-vJDnfQsmj*bWZ5=o#9D}7*p66A|P3O)|i6Y_S#
zUU11>q`pdHSUG16kyOd{k|W#Jhx5mM-jt7%<`A_7rlbO+yl-r}r7FLcnlY|=J-;7)
z>~$Gayc%Af9=C%y{8j5{lcMo9g2!m)=LA}>o*&Fx$n=LgXAD&L`&u}<o88y7yPSEx
z?GI;R&a3WjrdFT9=a>K4>6S{3+fA&IJv)YGE-B4j<2+L+@A15+9I>dkFM51kCtKoH
z^yhn%iih!2;}~qIu)ZZ&A=X!i>NpyMY(fkO%5l6zh26#3=ZPB_-q#u<=p(&8qDa7+
zSQ9UkI+)-2%~T&$F8T7<#v>6}KA3n+mc}589m++(iSZp`V(`kg!~CJ46yG~Afsazk
zF8W5UqCB*k&P78}j-$WAmvu|^hu9DMJqzqHw0%zRLbsTJq6$_tOJ1M{nmuO$5NJ7%
zMI9lD(PltDXb^x}q_NRKBIh;%l_~%kxzqEo@2v{`MUGEf8df~O#K@18qI(>!3s&PQ
zz__p4Iv`fvkIA^)Uj`!82v$}zNmu$Ljp6Tij>o@Ed`@&RQ%FzAKJ>aqq`4+q92Z;-
zSlmOeq#k220!>0Jr0#A`VY+s^oY{xE#YmGD--K;|ab}nP7#3i(*YmJDJ@vZ){8P{v
zv{|}4@k0oTCorfIn2`pZ)h1&SynC#!v<XZ3A{X1~z)f6j^?HZP#C#HM39Foc>gvzi
z7jM0@3Rf9nmP{K}MW9iR6NAN;Qq0UkC03Cc=lUeS6ULD!B{n?x#G3XKhu?+*5b5IZ
z6>?Eqw2)L#(Cx?6fo!rO{?gwV@lq8+EGfEdGu;#Tz{{=VvigP)%%wEZKY<|ghU4E_
zI>>c%A)ECO>+j>v?^R~Y0|hT;#`wDd(B3(PKZm4NdFB@jE4>oK%v~u$_br@Q4PX=S
zAB32e4OZ~8+*|KT5C4{{qoOHQ3D$4F^q|gYzs_{Br8AeyKwo$tyZPXLMgGk^9yQ>Q
z9a*}v#ub<_5IQ<^5(n7jt1H|><o3wB$?`=5n241qRf9~RAJ=-`5gnthVS-%q0Og2(
ziMRn%^LayM$hF>{=V@Ad;U+_Li1}nk4_UXs6F;k0kUaSF&kgL2m0_(Wb4gu2K+lQb
zi}ip(4QBqT!H*qki5xPIRH<`B<qn1CQ64K;%ye6(Gp)>zF~x&;#c!zL=z%7AzsL3j
z1)J&$#ttARu>itTXR$i|;Z>?tsZ`<+`j7*f0K~dMB3iE)eNXqkiwbTn=hOd_3Ew#K
z5IS<h0r<13*t{mgldiSfq!j0`f~c!GJphI936kr(<KZC`pNZV?fEsDVaUWl0y&Oek
zrQ+)2;)4i>Cph@vVN@8QE036N5zazV{0QehC<7WM^psc#oW&hGz<$w6+dicjIGvs<
zOP2HKQtPkpS9fe|&<Swhs?g3}tH=81NZ(pBa<1)`DdnN%d>ibfsH$042S}<b%nfTY
zSOxaC+R~*9Pj@`Fyd;U;FN%X#RjoG>Vu!k?wTZ{S2Cav*zA7=vy|0gL9QKxqHy%^M
zB+wG}OA%jfYYSB+-IwqV%YL!ew@JPehOyJpuSCSwRfA*3(;NKH{GSt37cq(*RoJl6
zFYv$PzFK#{?WF0{7_a~|79RT%R-UV*OE0HX0xeG<fK)IuKSbpAj~>KQB+mA$P${-k
z&?;|cB1shZ*4AC+Y8m%X9ii)%ic$-HGuJpQY_Hd!Z%{y=*H+%Av8QtwkM0lph>lqN
zq6b_`tv}tb@?=`Sxj%nG6<?{{ceb?GqHhTFv@R#5CJ`c|+X5ZvP8vROy8Ux*cIIts
zVZXj&O8v!Kj6qku)m~MO`%fpL|Iy2XqquG`4qn&X1ykQ8e0dc974&{pQTh-z^o<!)
z)Q!|;<ighiEE~;|+<=0Si!!kJG~MRv@%6E3LBZ-t?M_zhrgUmc^T1ycvp@EzO$xa{
z^DSOYlU&_deC5~{f+O4chQ7H;+-9qzpQH-p&~^XLwTyvVa7URKFIQs%WQqd1iQDK7
z&U%Mv`J5a0r42SYrA|uADT3vebu|W=9oru~9?<$%I9)Uusr?FRf$QBF7nKp>P!kI9
z6ntTj`A%?1hae-*Bu!&SK&t{l=0pr#iE~7#iFPpfG6Vn1E&)$k&0i6NYT0xk{vuld
zRF^VulR-@zM?nn-7o}<$u{ZY=)Ac0GR9Ib!_PL5DE1FdSe<-K}Y1Su{V@)1+=z(?1
zckfaG%F%<>^0B|&O(<4{oQ<WkP0-~jc13@+5JzxAYe%m=0two;)w?|{ZVPW<Be^XK
zmNOp8mM1V_NakU}rR%#3Kk4sPq$%9w&wlsYGCtT)D2cqOc8S1_fPN=cQXMGc*wFm$
z+9uAxn@Duv`bwnWMuyb<1T}^|&`k-?a^ZXK`4eb=F<7pa{9Y4H=i%swAJD%p2~<h0
zK>2jP`MF71(5K68#Wq;me*zUnys;GdVL5v*Sw+i7KxaI|zt}7U`<+t`ReV?RU$Jb_
zTNXIKP;o($LRHh-x?uvl;!>00F-5ln5>OjDcl0(>ZD}LY$1E3O?{>Umec`6CLgu+M
zjG)*5z78KCtk};t<(wQ5``o}%^mnmUCuHnhgs>ax1?JsY4PMGzm0F64)p5JKZ<B-V
zlzk;!lvM7G-cjZQB&hZGFQVAn8|+*mk>bqwu7z+G?*o2f7Uobdy-IeoE*FOU{%oDD
zt)Pc7#bP<o)hjCJv7lWqZWMR;Pr=2HwPMT|!@*f1V5X?sC9SczJ(PI@>SKVzE0D(A
zQfp#h7)4^D^}Jz%-@)DsBQP!uWYD}yoWDWx6hT;#Pj_t;i-qsWk6Ab!E#s93h11U@
z8~#R`O;u{4X<b7H9{aHKqA@@paPs9>t;rvFMA+XmJt2VR96iu6G>Py?9JJ3nOU6{r
zt<-#mW4dzat8c7vZYzpp)r=~vYC}6&jdOUg$?@>%)l!1larlAg>BYOd?s*r2zR*^!
z30`_U{{kf@L1?z{%2IIX-`k5Ux<5aD(3j5){1jC&|0KPFib07QX!2FNpyOPV{bOL+
za!Ox7<+qI30kYFF2j5lL8P9BWgNWyYwgt0<Kh*bNjV2GHCnyjV0L%Mt>xh1T@tQt<
zWg|7k3j)>s`RT{|T|XaJ3oHgiY`(Txuzk75XJe1c*M;bT(42}4yVdjMe%v*MyL#s?
z)W0jgNPx)ao933y3P`n!H=hY+j|H@8u-BS}m3`^K6_M}aOLT4doOvYXDp@wnF|WPH
zfM~MXSuG357_<>iR1sF$n6lU+_#KLv{ARj(rUM09ma#J5Z_bt0@p2dcmXpLQioa%S
zm+T+*F8QSw5A?%+{uCym_WoC{GV|^nXcPmZ3J7z#zY_&zenP16;eLdk-G+7uStz1J
zmn=;xN?hQN<4Ci8#@P$$4Ue(GCs+K5&w|GIi^ns}83PFiAHd1rcPs0(lD`|FjS50D
z9Jklt@RkChL3h0@Dq!CIB;??$<nTP+@a^3BLB_+oSETpFf2w(Cwy&76+-*hO<%Sr{
z*i1pv0+8pnNA)ok@`?P)ukq6g!<B@6OBB7(Jc$N7iEw{!cz~o*BA%ZBw5vO29|V+4
zpFovI+<2uy+jKdpQ|sF}YDJEr2sIgR+&}4jo7IgCYgFP6nW&8iff%kBCz_`zGS?xD
zA-7Ecz2gKxZVd8A=0+%TWMke2xN(}OCmu}8j~omNs=-hD*aE}+!+aL6mdx)&+#zWI
zt#RqxbKHXo3r3rllJ)?N-}K+c+*sl(!t&p2Ayp{rlF7Qtuoo9~X1}}r7GSqpKH^}_
zehEmAUNz%#dZ;~2Y?w8Bi<w<bS1@P8d-<sKG>*U7U;_G!p<?^%u1M10_ph-3mg7=u
zk@XAaXlx=&bBK@XR2DBR{b<(TV#(f;kNFx@C!Ra}1>O2LFc$b7KkII?3n@>Q@C^g<
z5-^|rE4JyvID0k;>4NOVqu$N4jKgN<=ZtSB#s0+vNB6MF&7U*=v8fF+Axm4MiF7hd
z0iOn+Sfi))PnYGUOIn*5`@6A*v)Be%Sk2<eV+hJiXo+s-Ms1<E<4xnKwpT@pv5?}T
zt-qarQ1vPPUn(vA<r4-e$bY11ob>9c2T5jp_G$H{UgqeMFZ1Y=Uhja2(lVO1P3M_2
z^^xJtTJv)6Fu%Lx59c@ak<CCMSt0+j;xNP_c&z=ygx|M}jR$~igwR0)_eSg(!lQ4x
zLcz(&@v>U$EO|^Rz^uBL3$I&!r;jX3P7MOi$qmA2pycXpkPzgfIv`_qye_zfl|#UK
z|A}9?vF!707Y&?h>1Vv8O-X!R{p;&hIu2VA1vzi673AdbE`8ch)W^&}>_$CON*Nic
zGWt{Udz3zWNMSyym|nQ8Is}K(q9O<S_he#8!Ucbuz8G$BWSxAS=%Y{)U<Yd#)-PLT
zx&h0;cHm}l_nH{7hEjuOwP=}D<+gX&D{nLqo(A^19*V<1g$F)$mv)#;YJ6(Boa_-`
zPjcJK&5s#{+s<jbRD<nSv(3P-X|Qj>_g5Wa(ji;I0aWfh;V7Mwzdzs#Wo=mrjP%Le
ziI3abWB1Az<WZ3C&Af0t&{DbcA4%2`tC(v2%D@{#>``U*PW8~(5fi6?7OKdaNj#q-
zOmGX_=B__UkatY=KJ6#ES#N9mt@JCTuc86wZVLhczyCm2y>tAD)8hk4=8dyQL-qcf
z(K1NwkDXqg?*?!#ye%6z_Th2y<)mj+4_6}6ZT_$<q1ilv1U$7HUdr6AHSHIX0cn;7
zU17DxYN=E7f}fSuzsD`X?x|6fNFU4qj+}V0$#d5KZ$6&y|EfzIzt54o%759S#nJ3U
z5ZMRJ^%Gd@^D&<5g|a3V{;JaVg^wqaYoGXeG~5HlGlKTJ3y*R`#!lci$A}aSwW~@G
zkXw~sC`!cylwCnbRA4^gABr}hDhqp&iD4bs3VKmZh0$iOG%R*&ZfPtUqrJT$#Eook
zWcov0yaI$LhlsoC0Dt0tUs~JFa{Wvj@O)b$syZ9pC<;Tg;ujXmOooZr(B@*5A3TiY
z<0+7)_=SfH^+49)!6q>tr*87EBsuo4m^UH<SGTrcb9Vl3rxe#JCKp%w6f!dPOemf?
zJ??mSmA`lBn*|&j%lj>a-?%AnosT~OIROM2ci`>6-p&I#E{*l@!#qP4DeZTmu7wWk
zKC(W~PY$}f2^kqdJx=`_4%XJ-;`}=vuG|UZOZ#BbN)(kVOlpTAje*Xhdt5e7Et+$s
zB3{KDHPAL{Htg}o^l(`T<;E0yR?@~LT|Wb3R**W2r1`LxIX&=uamcY{sT|t<Lc4|+
zz?~(^jlLg;1Za7I9@74tJ8On>zrcn7kzp+><InorTi`H_h(rUpc!ZneZ_^ZI|56K^
zD~hEXy5vg&1ial-2zejH%3ZbQj@jFA%{e<I5xT-G_WeBcej28y$ARfA|L>g$mj03x
zL|k*Gw|2_kw!X4U>dsm=C#JJEPPbA#+{ZkRWZW2p(OcRJP9ze^647ipI6DJf*H0Y`
zT?Vgd9y8%dio;$JDK<zFl>b5JAkW-DI4?Euv8Xu?`7SBytWVg{feercUhYOVR}zjI
zW7;~tQ48gP4CKoG9ipZnxqleA4Pl5jco!uXDEkI>-sgodYI;+TZjgnMn(L~zob@8n
zAO8z)ECWwl3!>33$~`TjEk14l*kTosyEubvxxT`uxl^wQ)60_s^^))YyW(8izi#7W
z&&Hq-MN&Q#WBy~4adS&OowWX6_om*|rtWI*yF=ea44*P3A{%62N^!bPk;!8VRu~>f
zupYN#C^^cee|itB#zFRl@o3KaAWgTA5INQtlfSJ}+FFryhp%4RR9{KFW^KeH`L}%q
zqT=zkXwn|{sq4}h;4D6N){;WYWVaE~W^w)RN_gL5SB`UU|L?;5%Fbi<bZu5?trqN0
zRsHc%c4`Hh03T~n9u>j!mz1GkRNS|5AhEla{M7V{u*R>kz|^!)Rp~d%qt|zaH3m=;
z!`<J@D6IFi=0ga!z<6*V5^Cje!0Dc<JFFmY@%7}&@LL-Mb00dFLx?-{+fs81*`nG#
zJI4`o(NB__?CM{mRwkrwYaZ;eEH+={n&@MJWs;HH+*n@$=yEOtQWh6dCow75ZO{Kl
zN0$(;0G_Q|wbi#`3!6?oNZ#{Lb)P_^Rd5el)jza|go#!pbr-3atB8!E1+2wQA6)~(
z{K;PO8e`Ha)tm)8>Hc|kDW*GT6mG9GhQjjXcY>WiPtP<-V<i-a%OG@_hiX5@WS=Ps
z+>pMVrRsFvM2cbG97RsGfh>ObG{-RF+`AdVte-)6d)urS;x>8xR;L#ksAymYA$|n*
zF4qJ3gf*@v*ii&8srIkA8dtIQD+p=J%?|uL+Xe<H=I3DCz#zNQG;cR?vE0lxk4c_E
zz+n!LlK*{Eut|z8`FfiIG7h$EU|`&RN?tx?&ia4nAB{!qwF>7QnHpiW4CnEOk30LL
zr%n7g|Ac=)RL<Ui|2wb|m$)1`llMRO{qJq+n2x#XBA}97>s?^EfEMnrrZda3=5`*x
z>9+WvqNvrCBU9w}5V_kgsEY{cqDKe*j`IX&eH!T0?0k{xqc;*6ZIP*wu8;U=$!;0_
z$3CWFwDy;OD%^Crg^~Jz#gybb=X08=I)ohr(SP|Tf{nLKkzr6@ta6)TcQOnvDxEfW
zbje{sWiUP@#=P~XDX}qb#}_8PYC424x!ZQ;*XZeGlhUvi8;-brb!irQ&-#9Xk}`&d
zu4>$DqgtGZtNean%;V&TORM7IA0O_hXnqe#fEg<70!DF<^iS6U^lKxZtJT7{nqz;z
zn`Y&i_Lm_nJU!CeNA^Lt=M9(a2a~UaIFZ!j?$#najRr%2c$rar1726z$JA8$8i--S
z6qU|O$x(kT84_Dj>3k1TJM<Z#2nV|57K+^u@wqv&U2Ap|QfC!At@N6O&Uze>?IQ`R
zFQu0R+zW;+lyS#Z+2P4&&MuNF0K>Lp)A!W=r#6L5yfYgVE#Ps0CY-Ba^Spx2l#GC8
zmUIxeoudh``tw-`Ij}9vt8Vu8wyVFjR1gb$b*Z2zBaTRu#P*X;(Vt?)o%hNi;W}pN
z(%A$o)da|9Kb&`GTiRD!Psh+WW5`iEKv#mm%MPSI8{u`rzFi-d|KU6FSG7i17Y0-y
z%13^{GQRw??-Jhcxq=g8#>unyDY$VbTDZBAA4m%_>vwT>y$4CfX!&8G!In3z_qeIZ
z(Hd|8o{H0CB0e5ME}`uY%&S3};M9JZ@SaPaA1heF&3V#K&w5|J^4bgw<!9Zk3p7JB
zg=&F8w$(v#TjS`|T%xpZmjnT6clR;N$s8F&vqN{oh%3Me038R;&0jhmzRmf9l4%u&
z=ELqa^XXFxXrI-xQ`NI8o84YLhzc_y!Br}Z8AV-u-R|7wuC6(o@m0DdZ}(4dNlAq1
z8vp#x`r4Yc1=61;tmu3_7(AMhoI>3j1iAsl+(6QdHh^r~V4;NQ@8vjnDIgBrTzQKf
z9~8A_oOW!pQ$B<<I7NZab)WWTfPeHsb*|2!HzVYO#}n-`d+=>|Q5mY}ho{}8cEUNC
z{%B&#Vt<3Uvl`BrpjYHOI0W5$c<NZ3J!p77^T`rSP~F3MoEWCkio{jeSLQb<6=*t!
zsder6`)efM@vT)c27A^f>jEl8)E1U*{>;LiFG#hXCA#4t7r^&}=qV&#X%JK{W7u%0
zh6FVSh9RMhXa;A=hYi;XtJ^!43VxHhE1yxV#57}D0~pt~vb~+CD%aope3%*_$T7v6
z>gGbTsV;F0<sxyS9iY_Yf)E${wF(+Q%*Vi+IG;qS>HxwVi2M&lO~Kd&s>^3RpZ5Uw
z+MVxfH$bKK-I8&@IKXh7ymyRk*_0wqB9U=D^q*_+!bb@fz1(Gwon1!`mY{Ne2b9I9
zv!vbK%y|lv=3W`A3pLviDoLU*VZsHWB5Sc!?|<i$N2Z0KVhLoYs_1sRA!13GX!QEI
z?px{FBRH}KRoDLx?T_?re|^1YvJ@v;5Ry8kZ2H;d_W_*^$e>bG9LEqz7+6&Sed?Jo
z1L(x_KNxB!RCFAa2BDb7^XOyT<1NfX+6siK5|>;UPuBM02X{}8Hug(}w+|R`^4}K2
z(E8=KKA}0gzy5g|HeKwtu=0tn>-Olgxw!?a(dS-s%cHCw5Gkk%;QRx1Mq0%Bh-#0{
zpV!dXw%$+*wlzHLF1@wjvFU`4`_5=(<?}n^*P6&vOHIPWcI7FDb7-5M71F}K@jLR4
z{)b<g_a(TtpMYdY#%b4phICP6Bg$3uayD&3#_Chj^tei06eb()&8%$(QoDhg!!p(i
z1}RSKT;HPk-Lnu`7qGhs3}Rll44oH+oe~fsoB1!`Gu_)zTXJn(8*~}?6mG9$=ZTVC
z{eiR0T74!C8wP*XKt{mj0uD8Z-FsHZ$d&%um8AWAl@4^QXqkrLDaN4PW^8%+YnY)8
z4+E{?GA;79Z2c`M-Zor>l+|oP3~$6ws6mdqxFmctsP^5t1b3>rcTEuZvR(SIeHNg+
z(9fW6WKJOrZV!4<_$2`Xo)0ZSH&GT%734q))64f^Qc722pd_#HMYrwYJJo1&zY9<o
z@$mM4U7MAXU9H}f_$&%gkWP^;4E)y{;%PFp<HJW~Nh0G?<8Locb}B(A`udR{%bU|2
z^=pOp=Vh|+uDPnjuB0%^9i-8)ZWN1;0t3#h#5ONe2aZi;8jIx<7)Ot!HZozskw_hl
zdR-DiA3ju9t*2qXMnCOnob8{Uk7cl(R=t*}`d4`af>*{e$xlh36;W#OGvNZZu=|v(
zAj6E$T7~!KpBt3hM<Dt7wpq<bZW){ha_+Qke)U#;>L&Ybqm>vAuNGC@qPitQmrR~8
zIVRC+wj5Tp%7?~ux=0j6sSfHoPD3LsiUv-@9)$^F+PB(9%Vf0mqY9sw{UDD#fkyY4
zqLk-6v46zoy#!E1gy;Vq<cu?-7Y*e#%}kt1j(}%Vy(`dQrAI}!?JWQ3h;PPa{N_X^
zn>8OhESM@Y2$$PUTV%tb{O_7qWDwjF^jHA2gJw<=?kp&8>8vyL0rh&zJA*=(i>a|8
zSj89JxO<4D*6HSsOF$+4?g93gvTV5cD=pmS)P6Yb701Lc(Eon^H@^*X9%6>FikGX2
z)7sk0PkFs`0Q?9XMmS^zmIv5*Ki;_s{7ld1kGsYlQt+=T1ETJeT!&MC?W`UwmLp(o
z7R7QKG9Q<(oxcCnL?6xnX%gV*hlOgG`ZXoOj8cd6s==~mBZ(T8TC~oc|2&H&qD?;=
zcASRl5Td0ASL>d~p_nu8yb@2*h`DptA$bfcrE)b1>Dch#pC#wIRt|Rahq&JXW1&vC
z9edMx1T{Wa+>YQW0F&07$`-m+#5@fgkpT=J|J?U$HL0xw>G^{~@ixi}ycgx@*<B`M
z_!rf@(p>xZC4cn6M#q^rT+}m-ueoh#OQ4Ge@Feo<ocfBBYW}k9-k@6tJxRPCgn$5J
zBbiljICBAhAnw1>AY4!ljY9Bu^KTag>F1XF@Ad<+sY#buRXx_|^`b0!Esh~oDKbX$
z*ecrb1$eCYHeyvOI;7gC$6M-?;q$Kg*)BgiFY92j*lP+39edV3DTyd=WM-R3nBEUp
zQaxrRPK`XHf`D<bPRa0I#ituZKtNDaQq8A}jA|9(hm+~uT2ejIMJKa=GcW|S)M3ip
z!B$+ex>LXOB65tkP`y`s!R5>tSo3&Xb5G^^;18SpHI{z2ZzSPks->&)!T73k&Au;S
zydm2qC0mhDMK+@%b3G9B&#$3@3>MzW<jK`jk|`yj4T!m`86%5dyR{ppk*NT^Npws|
zY>dS2!Adf0CCRrVc<BBKL5CZA%wqU5EqUB^b|r5)R5+D&w5+}q)=Be}#aWUk7P!@U
zDfa=QF<X04F>rUA@vCG>l93p!*Mrd(AyY|v)h<-^_J6ku?bpX0SjOR$X;{5`MG_++
z&0u=}lMywn$om+3XK;e<H!vS4J)`~`NG}BZ=zd`VJM)Jbc?;h+XxWAhd`ih~1Nq|)
z=iAPnry^p<W=>p$tJK~}gx}6d8C0s3SclEK*gVvz9g2BiTslh<M#Rn*3S-T776m?o
zUj3`Z@ctPRM=|^fh?zQuY3i|Wv%jSeQy8SAvNjBEjh9VgM&Yj8LhJt)lXaQI37(k!
zF+o|Qt7r9yK#0v&OGVC1KiWC}c28~*6BqSnK@4ek$KbKu5p7S;DuvNA9eATe0L7{7
zKOGt_Lu>zmf#Nz&O$N!UEd@*ybo@?q*8D#+G@yxChsW5HmBB<R8{f5aW`6slLmX+L
zH|BIoamd!_F#`mgUk9t}qZDfLs&>%99O1GyB3(kJkBWDj-<@zxA@$Ulri}L0#0dj&
z3RBM3!hh&Qkd=aw-xAsj1)6*woPKCEm?<(T;j87lEH62QG5?O+A1W&=JIuehm4vs$
zmR`IHFRD{J>gccSozmC_P>S2meyB~&Nn!2eb{r0j2J8=wcHj|7VGn2FG0f~)Xnijc
z+B^-GI&LFgigLqd5E{O`FW4ZT3X9FVZ4R^`Ke|4F@cy9~=YOnFJ0`z4_H({wjHMg>
znQ?WQaaF6cenjb-YV_C}DWDH&x}ek*qbH4d!$#XlK;r1Ms1k9Js&~R<@aLjLEjVRI
zvu$0qk;Ihw@y>)%^HgDK5s+#?)u`gj(1|G>8IoR1yj;e_1m|D$8KqkLqZ(*FXO?T`
zuAjyG8k1T<v)eXt-KO1=n%mJW!eV<o<PRhi>koc*T<<!GXHL)|<F7~TXBUY-)lCj5
z9p!>ms-Sbw^5i&Wl(?PRtR`K{9r(Mkj#q<)0bqTZ*)QQWO|LD6?-zbpJyhRFY$$Zd
zeB4o#DhrJsUUIZsgC%IF)Y8aS=)t#F$m%B3e)?&jJ#mRZEG5FK*}%AhDH7?Xf(IcW
zM7lswxK^S5!u~4U!3H<Z%#)2yR9W#|@+YW4E1c=P)PtZu*K$VP0J@PkaF7bYP-EJ@
zdB`dYSV-=F%}}33>|Ae~MTDx)nYoH=xbv5#WyW`^|3jx=+5;0KO?OnRZ}RTBJh=xp
z%5z23lNRQqKiA30h#!2Wlk6Jq{lPPP;eYr2?z8{ZSjEm*c;LxJ{X=ZPf1a8jcX$1(
zIg+_0Q-;2#<^f9wa;ofv!nm}fL|JXr9qYrFP7#o^7qN>Ov@FOy9(B_T<6gMKTV^UN
zebm>pY*MQQ+B0m~hi|PKhQ#LE_o-8i;b}u>@uy~AolN(Kv%=<5=^!kbp`IH8{5cE?
zAcUBVC%T?P2h}-bf_OYfmdOp@6DtGeF<~&%Fy4ripf*O+=I~>rjpB&!nS*Wk$%-$r
zvdwU7r1Lt0RpSsD-8e6?v$KIF6+i3~8ynKWT<ZDLjpwuSqU~Bh_%d&M8FS154NH1V
zHDDulmE4?8qN`igx7zD4ttr{hms!+LC47773Y;!y{iL;lAVexAv4A^u^lzR4m~!Qu
zepnCwGP9;NPjaokZuAIpPOqdjOuva0OFfDwjiOOwh~<QyC^c`K^`{zmkgNVdC_2_V
z;#2#3^z(oB3vcvtzwjzDHC-J}Nbt%w9m^z!-wOol9MaD;8_X(!BM77eW%Jw~!cHzH
zpC3cEquZY7EaUXvE?xxT(7uv;{T}1Mkk9I2F@u}?*_D<O57Km9;i(b3y-_|f_B*Rz
zKnK<Jl29+%-8@4%>-K<02Q|_gr&jV%ch!P?i=1D|@>Tn|ast$p`iU!A;7t6w_dkNc
zO3W(2@TlltNy`wCirOJY)X-V(eK#)cy7s%7#iLCu3ccdYJU>>K8dx;kofoAxXrUg=
zt)#zvDP*>?YRE{ffhDM`mqL`89f7~bmydpi>A|C5vI`yBiBNw760N~YLa<)Dgy!sr
zE34E03N11~V(7G;X0tD}5mOGe0YxVH43-N#7na6s!=+k&a2z7J6idP%{r(3(BKwCs
z466O#zADfyucsKiZWLSmYSj_^(NOi_*SPo8K2w=y8`p4wNNmnX;$~fc+)j16Nw%fb
zDu<M+=A0shmvA31qvOvD37Vs85;_>ne#m0v`e}^lQ-aA_TFXkKW8Yf7K;<|tm+MsH
zlXRdYlnNA;w;_9FU4pyu#GUVYu=5Qi?sNC@NX@yHc+}97Cc^@hYLGbqmJZVD<Y@y1
z%Gm8)_Wkyh^NN?_We4twZsLBM`892w)kCh=@s8F5_VWtZGb`&I21a(7@_+SolmYOL
zUegaLhj4b!o2x^&xt>zBf`;egqagQ>@gmy;Ix=z<9W(W}Yc0IZ;mwJ$e-PP`tV}-Y
z&EE5|BYmwK+S9SN`Zc0DjA@$fSR2x)d2cW@&z5cHuVSTxs&yh(-r7;=V`7R}9>|e#
z(S(fEf)Ch|G&|AaZMTNF_d5_pUN)5lG1mQqW5CqUAIlk652~eFP0m*42-|n^{=5bE
zlT)<xo@pi?urLrX`m4)Mt13EPo>%}LK}SS)Mt7%KZtxdibST6k&uv(MN9{ZESmu@8
zFvHR{lOnyN3r>ES_-*K(kG?Q(^z_~(l$k!q3bs5>!eh##9*@UcZmb-fdMcUht$v;9
z!GEj$;)|<yQl&V;YBQ0tDfL*q75@J8q^a;ipSsA#%^G!1lWai2((0r-b)V63R17|i
zo?lh<o=1M60eqzxveV^%sQ9L>(t*@~o5r6wR5ahQsN`j*6B~y!R)o6GooaA+Umf7b
z9&TfuU(AbWHh}xxzRBf)KvaK~(dyT-^d$EKW7_WGLDnCRwz8G}tLDmF0{0{2{f*H-
zc8q_ed!kg9a>Y&;g6CWEG>Nx0Q-GM8YAzVCvo@4x44;XtYor%}{%Op$XnQH7rc`Qw
zt`d8E>%$JV+dg@SZ>)*4uFdFww_CqmoV3Emy<q8BA2Y{1ejEQlp3tBUmyEf-k6bU(
zN<Vrqe#cZ5m)&HxWG&vuQR#GE=k$z<rj_aNKwLnKfShLbO7eJQrwudW`}mW8|1Snh
zip)x>pC3%)+DJ8h2tCEclyhPd@>4<EY_}i~U$LOa>FaU5-$8AyERlCQsMgbq;Kw=d
zo6GVL?v6z=rK!gTMY}oJs-G0Zd(^w8tFN)c{vlkTqKz(bDx1lFI5aNWFCOyPXl(Gp
zmq6*{;qcYc&GFL~pP>%Fo@Cf^@&mcb(rprzYzgTWFFUP<$=Vn-(eXok10T<z#lR$6
z_Whm@=IR#T19aOrR&kY;aJt^_7c=+Ea7l(wK7BNkCfju#_bWvd68ujo9*YYA0W{;G
z@iU2?yn-cLOliQ1%Viq#dwt6<3$=e_VYd(j5V2Z!^s4(R9u5su`mPg-jQ<JMG2ra!
z?xNI0@hmf?NCKzu!TL&RW^ZfV|HLqxM`d;v?|Ed5$Q;bKp%U@K^n81$5_#$b*%dw-
zx%?uKq}jffT+dD4bp2&nC@)D5PM@^Emi<`uh&M35*bF<Rub3NO06Ngu`TqNLK%lJ4
zDIJn-UZHM%3T+=xr!g_#$;HR$cYZXd#bTB)5IyQg3jI&`^#~OrI#0;&Fa3Jt?jq&)
zByo1d6Gfl@*7em3MnyP?UmXh$y^St?hwzJKaj#oXvPi%J`92WDZ+jB(L}nN8{PrTu
z58Bd3a5Qylap!5W34PolDeia-GpNvLrJuUpKLTuvCuuxjI@xgd=I#q)EKF{&{6ZEs
zU+X)?*{6&XXGJ4B)vmZ|@NO;kHye^qiQ2dF*#m_(l?VR}fbqFg*C!y0S_ict$7#3+
zYI=go(5`m=VP*T!Nq6U%og`gIH?EH!0~u|L841Yn0{L~M11YAOpJUulA+HT#ZJZmq
zEf)H`t;9=~CLK;K8>3dwT1FVf8O^lyWTuIB_@49_EmNhTp{`FSG0Sc3FY<+syjk8>
zv|gpHQ98T4#8Yfi=3u*z9xHmN<kz3CP`6-yHi@i7*o<t|UD~@v`cJx8HW0Df{_kt$
zL&c?0ZQm#Nkq}4XihjMXMAh{2%(RUb#Z&p1EnkVJ`XBs*j98GFeBm&OCO6Fuvrd{|
z_C%Hl!L)s<$MOAcjm<-(`^R19(em0AzjId2`3vxP+R;=Q_hydCx1;aAv6mU-Er3~m
zjpdD@I~w%)%?B{M!L&99eOBjEZ^$nk;`kxN^u4^tS5LHlv?u`F#P)gYf01o~-rt<z
z^p%-=8{EXho5BKI4>IVIllciV#GTmbK2l($=v%#7?w}az3sC}Gw_tMIyY_q9q?2ow
z>u`Sx&OVc_`!>py_Z0I6#$88C27-TXg=gCFydjIEThkj1FkWIEih)Ho(|tejUEbr5
z$+uosBQUr0{y4j}Cr8zLgID5VZfg|-K5k;?8QhnpMjNq{MS%5mj#W<z&cuE(9QsxG
zwwRsntE7ui#bMnWCcfmjqK~OXs{#S(B}Cb}5)-qnCB<PJ5hLKau_WJ0=5prqpG03}
zY0}O^plikynkD^OyS$7iE$CM5{qoA<eN@S@h%{OXR==G2O(>L%Om#>&$NPLb#;x08
z+;>nu_S|?XbRZ_YLqM>Wez?5N_di0iGk(cAK}xXw!A2*x<n!&J$jDo{gsh^c@w>j|
z56B%Ewa=pmBkaXMk;Xs2wnz~z48LLy4Bv?mL(@~Q(UsisG>A2uE1_3t8}i5Idu~@O
zW7)DPP67ZMIqoCh=%7#6eUSX7rxi_i95+$-_lhij909j}h|j4QiQ7+yZUSpG5t1~7
zk96pBFy;T+Sgn41V9L8Siudu~STsK1s#N)uVqQWA&1L;7ZKJJUIu7S+q#6Y3KsR^`
z5OO;gHh8&;C@+^xUt;md5kK2wPxYCQL)6~D5}Ro#5OL+}hf;wcv>DCO61hv_oF=cs
ztWY*w+VTCfEK-C`s`STyF3J6LIjKCmog~0ab5{G!*6T5AF0`?%EY?s)iMcV)$$co@
zZbe1$>HN<yP9kYQ9fuWzcF#20PcQ2w#VMrUFJQVp)uQe8rMZCcH-oKJOWx~YMGrdg
zk%*>s4UO!^2E)o1*TE+VRu;&%uRoD*Y-V=Gp1@^g<+{bdC6puU+Hnof7|{*@jjW`H
z5&EJZb<&e#k!1QR-q%cp#*K{Ev+o|)YGc6f`M)*1>EWmAsT}_%)FFP}P6zGf4-jTB
z5I<0!tGh2Zd6OK+sX5ybS=qn%U-qN5m4weIB@L#J?Lu-;!pHjBEa8}XuHi3*{?nz5
z0EQ={_4-=a34rH*o(-7Wi$p`N$!Kpb>S4JS8Lc4T-?YotZe?vG=60ZMn&0Sm&y@U3
zboSkRiW^GX&q^}+<!m%>3oVn%+WmxNsC7R;?*cXv5LX{^$DX5GkzRf}?`5e>vzSDV
zx@NUDlgl#?3K&&NZ4c*o72tyD|4RJ>vr3ebd^Ft^EyECh1{ep7obSZ%)0UCcIBQIf
zPgkM^8B)`kuRPZy(PC{pjudYvF_zF=zS}K-3&%=?O8>sEeAb@IyqlBQ$;*`Bi=#h4
z5b(X9QP`zVPv)_ycW{()O53z9#AmQce{;(l!W}nZ86R#M>S^BH6RDInyI6M-8XLvj
zYe$o?Kn@($He%<IZ~gC%i}VfDWvyr@1#~78wtcdVs7zr#HDNzpX6>_M^J}x7DA6tD
zK?nsOZl4|PR=jQGbj`-My(f#TOIPqHnvJ#4&rwgw8)Dn=sNhC}baLFiHI0h5J(WTy
z9EZ;rk=`~xH`Rp!U>BABva=SG`rQU%=O}lv3iiya<#fG-dfkO!iP8;DO*O$P@`l1K
zwfjW$DlAxpA(oB=v2$d8Ie$LD6c^bN)mEq5+k<637I$cyAOO!xXm+wvBOsT4G}-R1
z_)i&-ermHUf2iG%)%>rV;16wWo_{_Do%|ra%rT2W>|B0JJ!^UFD#18%d{I58Nj+Dx
zO;)z|T3ox;U;24RKi8cW5<90}l7!i?fhSGbOAp2+dD2=cASY41YtU%!{SU~u%)(8r
zt!Vq-T4VYGhE%9Kb=%a@{>b+6;=X!qi}=2!h|`<TNv3~_za0l${~7^#;h$|2OB&Sr
zOLkqiXu{NpWbQf3?Gy%8FzNi2lVe(QEq#zQiK8VGr*7rzXMJI+{$qM_!P;a6%z@%W
z`Z5~gM@t|r`$+M2B4uD%F&d3<ydbM9PtzLHE`QX`{8&1jb1%PM#>Ts<c5ZKTTf8!X
z9Vx$~J$P+Pd=0i3JaDcW#F1^&=N1(ULR6zWxy(V33(s9)Qb$*2KWV>lAj4TU7UovL
z)kG3Cg*XH&$^+y`TpozBf}DxusJq<9LrkWmn%^?t8#D7f9(KQLn575Or-jG}lY;ux
zTD>E2$`k;E=#32zFYBPH93yRSmO)>!A?12WRm0p0j7QFHefA`7dd6HO8;^&-?V*DQ
z!hy@kfud|n7ncyB=*mnv=<6_9+O!Z{em#cQ_&tOfdrCg=v~NXw^Kv4{NtNkmeV57`
zZ$BOm`I#f2>Av$ks`Qc4Wi4*(=2WQ1pwuJXI#0J9+n_tVRz>;kP#GlZ`iblA!DtAU
zlxnv0g!i-P&+YtW{rg^igdYBYImF^I^=<-ly?^gy%HLhT+3SnS?WzqcchezfVz5#`
zw{4U^_QDs~?n4;9`Y|MHw~6qO(=Vp@dJ^xX2EN;oU?<RVI0&CZ6oCh00K}<|9dQrY
z3N=u%I+Az|_Jpd$RxtM#>enm*Uxp5j(8LW}?F+t2{6XF8fexPnqRVO%)AxPiVLFxE
zK0<JC#>QS<9c{|(tzfr(5%+$pVi1+zY-(6>uy+v1f&nmO#>LgD?o&^+2Pl<-oy-8a
zdtRAcL8hRl(UsajgyoLbCqqK^dLV6n-Zr6<ltBnQ$W5%`lH39c3r;0q#BCYiuqxf{
zXHk0!Ib2&yj%(?X*F&BETNRMKjyX4a_)PL`CLM`J{HjSoUnY2~@E8K1B9ml;b(Hh6
zKfXSCm;W`e-CrCZ<t(?+4CRW&mq*P@|F*N7tGiMbqj597S5Igh{rlP2wbU;?IjFjX
z*rGhCYjw+L4HlYR&u3!#+DZ?#)f|@>26L;JOWi28zs#Z%vkC5J)y(p#rx)x(!<7Da
zee<P(NhNuQ-HbkNSf0dXKu&2%y3eGL1wSg4H8^xUcV}E0_u15!D1IaB0~tq-dq4yc
z75XobYCPO{J|a?&zsKf|{4MLrhB%UCYoXrY^xWeqzV>`z$VwCE*0%E6>gR{W4t&3c
zOTyX0h5oaX%{KeqZu|Y7P^U=!?8Hw#OwdzbZR>AiX%&s-_3xS;<NdPbYR-8QxYv$`
zW}636rDm)cCs!ktOn>OXYk<taE}0unj=A9V<K)6qbo}x@?-!8V3`wuDX?O8PEjx_B
zE-PBFps}|rnY-#7m$&t|U@3f&=Q0c$8wk9wqe%JP<?(iiGywpDv}q6PPBOrYXg%q&
z?=a3&^m>F47=?XSxonc56G!DzrOkVaHP-{7aS!o2CBENFU)PZ$ILnKrAMNP7sb-Z!
zLGtv#KasYPT=RW#{N_G%Ebxa`^(8WANj~=>G|#_+ZVBgvi0`v0Yj)}V^6=(3x9Zm7
z&)41~qwaUY5jR+N9DWzkOw}au-=)9<H;GDLo{AVw)Z-1MFiv{HGRs52u)2%)$YjYu
zSK$dRUk5m%eFfdM$#9mq*pLz~^YZ|%z;ZB<YQ&#=S*;n{nS?ayes;O@KKIit=3VAe
z`S)t2Eywt}6~iIhrHJrL)lK{BTzv!)hq{pFlF{Ds80aU&{uE|=)4Mm```(W<I0+z?
zPy7+!7wV4!$#~iP@=1Y`w4KNYR{Bfe4vW;EQqO-kC`l*x7+l1-d*38HGy#|%dueX@
z=gUX7ehpRUU4!%bEqK)RHF@gI(}bZtKa_z6O1dD)0DkmYf<8@}x}W%#Clu*$ZYGrP
zK<MX&@C?!qd_Bp^XNe9UcyVAZ0mkX&7j{5WcXKb7uvf598{5S$?z6TP?{zAPTHfCN
z?1v+Ed^*VfG%SO{Lfk}}#&dE@a2>wVs{}2l%PNL?N9Phr>U3;Qeu&Z}`oaEDkf?av
zPwWWDo35G0^(60>f@=WS`MT(-_o3!F{AIM^tne_<L@u9UOG&#a%f9DcAQ?XZUgzz0
zIpuw#pYM5!5?_KrJJWnpeF9g!O?+PpE@>m^aKZcXwZUnCIpA+}YQS84RX+8C;<05y
z`2`cWTlj&-RX@@q0#ht(r2K-lVeo&roMVf+yS;@dhHkjMzhpZa@+;}#9<_lrcc|``
z9Qu5F=vh~KxfZij*z|~LwV{ZW##6{(I!exccJfH$+mzf<#};uAVNkMK8nJIg|4RH2
zt6+vN!d*k*8=|q|17YA*-37rkIt>dLNQw~!+#R0pG>=F23}JX95AT&sW1^Or^mw%D
zT}b&&m7I)s9!ZK#{-duSu2kNcMtSSING2D4Hj_Rr^#s+=TAs5~{8;d@2LA4%afrE0
zX^BnBJ~_zWi&;k!_YO93QUHK$vzF)Py0w10Vr<K2gCAFZ85`udt?WDPruWw+#?Yg+
zX%y$6(uaSoK4DoD<TMbq>+SQ)<M<$@D8&=o$s3dA6IUG>hn?I2eiHocLX!BghJkLw
z9LV{XISz)FWeEAgl4PPoGPs7%@q2aRQ2!Y9=>t@)+po2@c|;E`uw(^^<p;}7jZbFz
zC%xYoyrzk5y*yZ9=6WF1k$~sF2rvh#8ttG7TPwCLYKR;r4-elFWwR%2*S3_6AABc6
z-nAZjNshdWmKSSIzIl*#xsp@R<SS9noDr_^SJ1KTQl9_DEEd%qALS4;<ZPbpExuJT
z;)gB&ct<L}{pHX@p6ybCYm<%%l8qH$-;WLcb`&YF=@Z;+qc4;KIfuJ_$Dj=a;a&rJ
zyfJ%r-MvL8H_*=Q3^Q?E^z+uPtYLdDm7N|63=NaG`<=Pto?wK`Q7kKFV*9j|F&hcD
z(g@Nv58`rwCc|}Z`P4{A)y?ol%hF!LZxPU1N-ouvjOF!8IWY4v+%X=2)1)@})4}A&
zfK;#ns2OoFQgsieLW<MyFBUsWVc6CP|H*+6sFnm_iUf3dwp>}mLdkeiJS4FqL^3Ho
zFkOk5i;3pbv=q(1k!zI^`^)r~%;E=i^yf&L&6)qPO)W%jg1`QTk`|Ql#kI|dHeI|Q
z;NA9-&iYq#^RuGu_Did4ue%j?`fAg&#UAaQuP5nV$NotRoi<SR{jiSo>QiEn-%Um`
zlQCjqK``%UDFe3Gvf1p5ZIvumZfl*-t)FGt3J%}5xU27)yr>6>eeAqYn-KOfaMDCt
zTcOOM7CAb?YrfboRcgO;VfkqvkMnvlmT@gH#SqS7`5?|5!xSFNXSjp4Q^dEvna?QU
zGJE@970fH<EJwox$MQ6dc@A3mx*q<mEfIu(47lqouTjTFi6^!M8HFGCT1kXi!+Ih=
zh6^sLKXJEQq-tUm$mY|Gv)6d@Wug75-x$8hReD3CFy*MYdsLVl@5~do?f9NcCcS)C
zo;o0FZ{Xp6isg*F?_;*Yx8xwZ2NCra$rH1sNJU-CKabFH>f7U&AY~vb#5nk$&BiIC
zp^aa(nae0?F)^OBPwCj)rxq}dlKtj(gc`hB^X}GQo4?zo3_fr<HZsByaT7H8B95NF
zMA$S_8Vb4hGNzO~^>=7rGrQ4)1J(cQdjrFAH`rxhG^WHZ$C4;!4iMgrwqqHaOiY9S
zazF5OY!2rOvGH{b<yWMMBciWI=t+0!O_lAnMGgq)iMrAxnz-(agBv%xowzUxewv;;
zpvCRF@RAT{>T&rLAQEO@>=zjr+MnZfTRVvy@|9O>niS^D7`xs)4fvQFMZE8S*X5T6
zrz0xE&U<4Ww_+fe19S8LQ=9scGi_qh=)4&6=`+4GNU;vloJo&aD>RPai5+*D>vmtC
zwPEQX@94x&qoaxA^5xO1w{$$6*7ZHl(<@I)Dow7e2t!rV$Namk&F^5ZlUDk<mlBJP
z_UAe3rT4}#F{SkK<^-2DEAcYG0my7aVzK<CTpfN|F^`$9C57R6?0%#zO-1tA-9xma
z%Do6Wcnd;MnQijYuKBsCqO`b@K9S=u$<{yK<K~xEf_CXQ(!N{moma-s-OtQ##$07u
z89=fiNw0bOGyjHSVZ_DNqzl7vmPiq(ezA`8{wLg?x5s)eq49LVN7WtVg+pWU=@n$e
z@jQuJQ2R+3${Y9fuZhtlasfI5GK;~{mKUx`{D<BFe3=9`YP);atH$A#4n~ulOF!GQ
zG{5V)O<LE@vQIbEE#zpi$$MKnaD61hr~O-kNvoyM*qgmv8%#|X{giN<wbgU@d=A8)
zOJj^_PnC7wKdoSH59dkdm`QJ@*@hj=UU1=Z9LzmRq$A1Wk4`H+*jE~OmHZAvc~O70
zou-rwXv`lAUdjtrKzKU(b95bT1LD1ZcUMuqlr?0@@pZgX_7bb{ms>P!Yp0Lj22be&
zRq`{;dwI%SUB8TeQlASpyE&3-$!MT)U`JyZ$+xzJAi7XAc3xct^vjerQ?f8NCo9s9
zkhCK~h)HGFgJK77sC+l2@MTb8uNq?C9}#8Sk3;*sD;5DKS2OrXqbOq`f}1MZ71Mn6
zo!&6kT{8aN8?eGfZ;P?dRXqkBrLh>2pO7L**+cRTPWgpJ)(288xKc|SkUNwOm5D}u
zI5@t}*FEEnT5NCMUy9i+Y$q77bF;7TZa@;{DF<#}S=7`xRVf^6h@Lf+n3Xlua=F5F
z*XKOAL}%QXfXXqsnR$4;hreuYcZ;A_qiuii#v;D|mVV)5tb>xBM9i`O*vFmk-nIbr
zkasbnAU~38UWd7DIN8T!*Z0ingHDbJbR@(+gx;KgkR(epG<ve-@YBk<2x3(P3~dU2
z9~K&N3Ei&Hz}8?oJ!!FDlME9VSaGwOueddOCH?5DU9wk+8S_R0ZUEi2GmXEorV^!D
z`*s|4geN7`XqY3;cO)2koc1on5vaMOV}kjt(*~OTF&tT~9K_roFv9V6(6VsuUkQ)3
z8pRXs-a72Sr2`HAnfMeRNu0BVdSAjDa$-6gbfBGnxv5L3)pDj@TZUDY>KRq<Mv#_G
z*Ks&JWuU2He7-ZdGO+rFCbCXw(_pUtB_b@jthc+ZCQG>QsMaF`d_;aBP7AxI-)`%1
zl%eC8Y0fw@)|srU?C)Y-yMaxVDE{T}_!P10a=&2K0Ca>B>go|M_D`S369k^v@0XaO
z1d=~sP%O_Cow!nIKV2OCgStCcm)N!6kq+>GU4Qk~s_WI(gspkwFF@&UC2<|`oBZ0?
zV_}ihU5!i7`w%_PQjga5`5r0t@(U{#eu$dmOCo(&`1S+TenvKYc4oMI3|eZAq--&G
zE(B%)+&asLquT7W+QJm2`G#z&_fqUsbuKBNoYy@qkmEDVdQCr8zs+^IRu`L7+__>h
zgr7l%UIq%jl+Z(GYP-WcKgMS?AF6s8Za<H{`$$P|-?*ua2ZoZWG_u#&Ar1%S+WNGR
zqE%2fvss-D_#WssSj6eXj+5vlaJ3Dczrqwd8283DFv+gqeiCR~zx~j0rDm4fbhvZ|
z-E8Vah5yw@Djw^7k`jMDHdg?W+~BWxPO7$JS-&;L?3IQRUWZ2gdJ=bs;!#^S)(n;H
zl4s)*lr)A>m+u`c%yqGjqsDRX0X@RJRLfrea;I_<fhcYs27y4^JKL-?HEF!#7k!g;
zSN0J}MD%x6C*y0$&KAa=T;^-q8NyY61d9n1{#(gXiNId^@=1y!TXpaDELAW6=Fxsr
zD0RfqAdrtE<I0hsy8RvRABKclZZuEyH8CkHN3-*>XyLERFT!<F*@m&TOowWz!$4!h
z6q$RYqP6^DP&2jFKkIM^;mA!lnDEU7-znqF(Xf(Y<5ZJ}yRb3!#K!WG?o#aMgD7s|
zA1)sN`v~@+fdW8Ytt~1U9|!zO=JU8w^$XgNmi>=_5y$^)=gJ?U{J#GfBO*%^C2LtC
zgtrpeOIb@PnNOlgNY<ewjIn2=Wr~o7B)iFwWh9L?OO~-EW6;>9hB4XKv3#!Y_iy<8
z_MTs!AI_a~U+3I&?>YCJ^E`j4f0KVTpF`*!d&6WnO!xSFs)y-bET8U~Wo4!~q>~R)
zYX>_1#5cx3prYqFVKI0wxS$Xj9g$D(JzWSdE&C*;RsbI9DZa+?q#g<*X~O*yW8ymc
zdy+YW`@sFp!L@zf^8p6bdd93>&d~n#>AZOb>?5OTs&%DI=)vYh8DAXr##K7*c;P-*
zpCk=C;<h%(8uzK>f6=mM$QQ+ID^=vWo3?yMePcW?(?BR_a=9)C)9ugk$>=@<$IaP>
zPAoWpFAGUZk6G!z^XiXo<2_bjT=pQm%1`UB<OkE$=^O{#tY_Zvy+2xF@$zqYB(%KU
z7s0zO6+#5U>Xi5II%kl><NqM`l9Db7uh6;b^h=aIcx*uV#)W+YLqq2|fol(}NwocM
zZZamz;A&EN3y}p`N*)_W16SUnRArQqOFjdH&WbBEyk5M$!P{h;2r8}niAZD6!Ig?h
zFJZEQWJg^-!m?<gN>$*vD+V}B%mwj9@p4?EPzE8(!9y0;pVCn(Lob%Ft`StIa|W#v
z{2x+~56gg~rwTVddGBdnwQvaGoq-#r9@k5AQSl)Dq*T#75FP^e#pBg{gGnmzH<xwO
z-Or~&bE>+^%1XHtLCKpF$h?Q54NQDfq^mzHwHeI^vG5rm@N$@>QA6kQh-&H5pSV3{
z-z7+wdX3DC222!PTaCUpN5R_jvbP&FEVi1EA{s_z2V$M-)p}Tr<w*)9v1D@48C{{*
zi}cj&kz-7t=O?i&<b_f)vAtVe=Px$P-MOnz1u2S6iqztevYj#r@Yf}_sadyp9Gd1X
z)dih?SG76$G|~?Lj@_Q^Y)X&`td^+fHW%iCwI0`US0zmZ>$H58?Ypw~pR<iDgmHe*
z?bzjF-5>T0a$)Xr)yNOc@xjcnN>KHUwePUJT@Cl25{RE@GJYD?G*saC<Ip9|Te&nY
zVC`unR49^$q>EU-4bpTIlb_x?X|Jf@Uv04>w23?uXa;H-mL>OXJiE5j()*=!KdL<c
z%&4kiwzoaFqcK=2mhj7GBFq)LK}M5Z!gFTne4Sbe)y6ERwhEc9v-cIu$r7LByck*K
zuZoZ9PG9$_hwVF{cCafbkn+MM1?nNQnB%V|GiCMkJE<|)WDNo=)P@B8VCdw&+`ACg
zpH*YR;Ksh|q|+XK)7h~TrkQxVP4-A5Rf<|;#kvNZuTJSl^^}L)<u%jne^IzsPVgGL
z+qzidss9|$-0MS!p-0Mo<{7`FknyF<V?(_dgYI<$Hp=d`FWye}rtQkIUR(brzShgU
zvQSE26_Q}McIvM@=-+Xw{<>qS%r3I{iNccZFHG8e_TrpVkwugA=h~yp$qVqGvY$f|
zT1FL(w?>T??1DTO<&-WyH06k&csHeCsPFxElYd~2%VQ1?1sBP>!=RM*=I72rZTkuq
zB{+OCd&>I<iYnd@*I)#qJ$M(}rrP0ZM60bf$?^WTB!1|7alPJT`wJu)a}xYc3evoE
z{zz#f7s`Oot#5PZJGu2mbA`MMAv7~d`0%WF;7EFD!^wcY>^AermY0F_#oV`<S=Z9P
zNKoq`2xRyEkLO<?`Mi-r(w+QuYoN5~nvF4<xnZW%l8?Z6nAciWkCb3nV}Dd6?)~|D
zycom(-sF|{Dx8h{kp7(T&dy6up~QmR&tl4~-^H)NPoSZaX8!l4E)5d1{HKh*jI5;?
z&RDE##bCJ+3_(!(*KKxYakA2v%?}HQ$mjBq!_MmEJ=-I~wj-{3)C#2|npxqAT)SCC
zEzj_6BAe6WR83YApSi?`emmUTG4|@zwO~}u%Ww8y>zn&;F--N}Xl&DF1tJpG^zx!x
z)6V#(>!03KxNeHQg?@js3}qbe$_`YJ9>OWyGFmezcb=^O<5V4Rly$tKZ0?n7hfBA!
zUq$2EW+BS7bqWPL?Uz<{P$|LXkH>qiHgLVv>gBhf(lp!QZ<P{3ynC*7^!M^_8%Co`
zjOy)>AhK=Z#hmQZhf>WRN}o8MKwAIe(eTvXm^M@+c*I$J$2Kk`DIuTUMfQ5Cd|63L
z8#xmyZM8A>rHt6F%u>95<0pCxFB{Hg#kp<hhL48PDZJw0Ba+Zck_e)(oV$vg?Ws--
zT1yk-vY8|7aV_^zIWlA{$>e-mHNB6#px~zH>KAT#G`sRi%9ZP>Xa<^8+%jBQqXfx~
zekaN*W{wwpPQoyHnLk%r$ge8OXDz~sa6xfeW%f|K>w`|UA#zT^d*)<3`<K>?nSQ<7
z>e<$y7WF~bL8HDJH0I_h1LkDdR_ZL<HC*6w%5%l3qNmsfADPuhQF{_?poyjsuB*L;
z*d>&k_23^DP)sa^KT=;kgaFA!Sl!I@s&#jmyu>kyk<Sfdsfn|`Eoq@Trq}-IIZTq*
z$6BKM)GD=O?inYv#)`0`KP0MKp$aPV!;zvocbRWaZ8>`cCXfAf3vZ%q+0GKT>4aZ9
zjrSgb4~gxxQFgXIJNy%rG0WbjvC=bV*u|M+^{W$yn=6p9H5y4~27?)~g6ZrUtv^FK
zu_9zO;%Rr1PQ5ApWE8JvFihk~&R;UVcY9%L<UyE}D>BE$LeX`0I(jAXgF|n|8S;*Z
z_D=IG&22fA0KHA>lwL$zfB#q}jZOGz<~n3k-euWhZ{E=xBP#vey77tq6>sdCy_9l5
zn&4LT2OMs-kMl*0t;Wd;o6?#wyrm&sHkN&WS3yuv0;19%v?oMRa9)dO!Q9xs${n*u
zSa81tJfKD}r+*(h{A3N1a5(Z-VPFG2byK*`du-sKg{8axtkD;*Wc>KWHmkV2OPE|s
z%{Mnm=@i348OZtZ8^h79?@ADo67Y+V{3a!~gv`cuG5rO5QLI?MY$RbQ9<RL-nLYG`
zJ5Er|7FEskn2QV9x_auz8hQIrph%6CW8@rKdU=A;SFHWfPSU@-*Vuid@-jifiaUPK
zbE~4tk1d?61g;KsDBsJgob5}48~FpP@@@|sRPMa~81ttvP1ilTWw-)aCf$`v=Z=Us
zA6-%KO;q+#Q-QRCh(>`=XJF=pR;ANbdnNvVCaYB1-PwoDmWOqkPIf-t9-r<LXa=8Q
z%N5~*S#$>n4S_@|!^q?JoVddnwePV3gNcad6sCp!qIm!Myka=dat!1Fqh-P9_enU)
z7Cx81RK2NP{9K4i3uPJx1)0U${!Mnq`5~Y)V-RG7;Q>E=ty#bIU6$6xOK+|CrWnT9
z*)%2LHgX2W+lk<?y$j=liJ#+Om~P-jUFyhHPUthCqqI+se-`<YyW;4M$+UH^@vak3
z{RYJ#kT?H;>UEr-i_gwiS(oQw+IOm*hNMK!g?O+1$MF^IvDdsC;&(GF8Und}5Cp<U
z%b{@Im*{GQ=#9OWd`I;%u)hl3i>dgPt+|i~7%+E-9O#2NFJ)q+8{|gpDHfvoBKf9@
z3Ofheu(7yiYx;g5#0_xmTIseX2K*pUqVY|R_qEFx*XJ<znBCSzDym(Uj7i14$u<JO
zYd8aOSK=Hb=S!-iF~(AT?L-wNb*tr$r@0&c*THN;0#LpZti4G9b)*}%U-c4?i(j>r
zSYq2WPeD5RLe%5y1!*->xB#GEzc}FJ4MAUyWp|ye{!Tpmp0AQ53}iT6fwRR0>Y+wL
z5Dv&1<B&$VjnF01qKY7ZkUIKF6k&9c*<B;spp1Y(Mqh%gkfCj*bk}-M?NS>L9fzf*
z>)(yWO5aIYHw1W(f;cRs99oLmD*8NY-g7|c?+8UW`~GGctld3ly@*~ahBM)yudD;;
zSSc<TbFd&ncQn|aVX8sy?b%(r7yMAZ7Y2gTLm6AQcS@=ipM~XQX3aO?1J68KMgX{@
z`Jgfj$AyaNO>>+%q10(!(-|lvA`1(NlE18a1aG`Gaf`>L<k_b@Si*xPLJo3Un1nN5
z-giJ?chpk|)WVVLINsjC;DAY}Srzci=Hgs1^W@kUg4fom{qvy%OhivzdMO%T-*0MX
z<3x&d+3u%*LSEZNtogij*OfzFT;ZyJm6J}<30vajAZA$YuZrMWtDUn=S&o>+fvRDC
zU@O8v3<W{i0uE-zF<I>Uc7RsN4Gr%l^W7<TTezgoNEm?SS7tx^<7vk?g}G6^_7)~U
zurmVu&5lHjTEx>N;Uu)!%e154P~R4V*Zz+J-*4hK8AIl8Rgb>#+}7C@!S_TPxqhM`
z?Xz-ZmF+G@@#jDh7qG)4+B1L8!d2w-%8540`HLa<mK#wrI1XlQp9(Jo@-haDa$_|k
z^H`xqaxrTHgwvs?inmBVC}cUn!~(a=gdjU;UOSEPD3l6$a@}qH<JzJ>AK2C=rvduo
z4qUdk+6Fs#O)VH}7xeMLTMd;y8N%SduI7;XSKU?N4+aKIa}OH4bXSp%tny%Xi!@jO
z1A<J*1KFtCo4mZmR2|m(_NAo+!t@kj#U7;60T~pRqvV}YJlQBs)KXcTG^5S#Oq6F|
zpl<O;$I>L>jZiRVa(9Lj42$>45Os=2`|*<ATQ6q!whV2|(Onk^XR@naPj>Zy;xy@+
z4<Si7eZWO2h-o0JM-#*B>X+XXQlEL3yLGgrTHdU-7xE$L#q{^~-p`xTyW-9B9nt-9
zDgxZt<STn@i^``n<(>K|HFEXi$9KtRcFB6jA_6joM{4ifUQ!i2QqhO-S<ad_NCNUl
zXn`eCuJEl@wXIaNrT)^%3eG@yv<_R-#cFGmD^AiydwPlihAS#iKA{{6?GU&-oi__G
zSuK26V7wTA<8<nv6tIx}yVMnShQ%YV6U2!8Uy6vNrZQ4*CdM!zfN=YJD4TcJ5_JHU
z*Lti<;TcY*Da$iM_~v@I^P^Nppvk)9fI|{{$T<04pPiAz<!qgb_vG67O?=?s*9A!u
z6n^UB+#Ci76mb)1OGUy+s4x_^IDhfC-mp60d<^U-Q#Z~^b#K)prv^&wiLf7C<hvSG
zo(4NzU>8CX!5U=3eb~eZ=CF)j+x1!~z!ULZE@Jcw^~<_WVDkg%b*mGh9QsgdCn1lh
zRSVXoPV{KRs2`K`sFRZk(zj|u(^zm0f&*9rAmC50xY?^+nG?Rj8P~Z?oy+8t1=I9B
zhD0?U-KVJ4FI-G}|Nl%D<xBwuQpa`o&Tg;X=svDpKoKWlyY>U2fb1>!>4NwLJpLvA
jtAu|o;eWmzI*g&gu62;ASE;z&&1ZSR<~;G-?Wq3(I(xn5

literal 0
HcmV?d00001

diff --git a/api/core/tools/provider/builtin/xinference/tools/stable_diffusion.py b/api/core/tools/provider/builtin/xinference/tools/stable_diffusion.py
new file mode 100644
index 0000000000000..847f2730f240c
--- /dev/null
+++ b/api/core/tools/provider/builtin/xinference/tools/stable_diffusion.py
@@ -0,0 +1,412 @@
+import io
+import json
+from base64 import b64decode, b64encode
+from copy import deepcopy
+from typing import Any, Union
+
+from httpx import get, post
+from PIL import Image
+from yarl import URL
+
+from core.tools.entities.common_entities import I18nObject
+from core.tools.entities.tool_entities import (
+    ToolInvokeMessage,
+    ToolParameter,
+    ToolParameterOption,
+)
+from core.tools.errors import ToolProviderCredentialValidationError
+from core.tools.tool.builtin_tool import BuiltinTool
+
+# All commented out parameters default to null
+DRAW_TEXT_OPTIONS = {
+    # Prompts
+    "prompt": "",
+    "negative_prompt": "",
+    # "styles": [],
+    # Seeds
+    "seed": -1,
+    "subseed": -1,
+    "subseed_strength": 0,
+    "seed_resize_from_h": -1,
+    "seed_resize_from_w": -1,
+    # Samplers
+    "sampler_name": "DPM++ 2M",
+    # "scheduler": "",
+    # "sampler_index": "Automatic",
+    # Latent Space Options
+    "batch_size": 1,
+    "n_iter": 1,
+    "steps": 10,
+    "cfg_scale": 7,
+    "width": 512,
+    "height": 512,
+    # "restore_faces": True,
+    # "tiling": True,
+    "do_not_save_samples": False,
+    "do_not_save_grid": False,
+    # "eta": 0,
+    # "denoising_strength": 0.75,
+    # "s_min_uncond": 0,
+    # "s_churn": 0,
+    # "s_tmax": 0,
+    # "s_tmin": 0,
+    # "s_noise": 0,
+    "override_settings": {},
+    "override_settings_restore_afterwards": True,
+    # Refinement Options
+    "refiner_checkpoint": "",
+    "refiner_switch_at": 0,
+    "disable_extra_networks": False,
+    # "firstpass_image": "",
+    # "comments": "",
+    # High-Resolution Options
+    "enable_hr": False,
+    "firstphase_width": 0,
+    "firstphase_height": 0,
+    "hr_scale": 2,
+    # "hr_upscaler": "",
+    "hr_second_pass_steps": 0,
+    "hr_resize_x": 0,
+    "hr_resize_y": 0,
+    # "hr_checkpoint_name": "",
+    # "hr_sampler_name": "",
+    # "hr_scheduler": "",
+    "hr_prompt": "",
+    "hr_negative_prompt": "",
+    # Task Options
+    # "force_task_id": "",
+    # Script Options
+    # "script_name": "",
+    "script_args": [],
+    # Output Options
+    "send_images": True,
+    "save_images": False,
+    "alwayson_scripts": {},
+    # "infotext": "",
+}
+
+
+class StableDiffusionTool(BuiltinTool):
+    def _invoke(
+        self, user_id: str, tool_parameters: dict[str, Any]
+    ) -> Union[ToolInvokeMessage, list[ToolInvokeMessage]]:
+        """
+        invoke tools
+        """
+        # base url
+        base_url = self.runtime.credentials.get("base_url", None)
+        if not base_url:
+            return self.create_text_message("Please input base_url")
+
+        if tool_parameters.get("model"):
+            self.runtime.credentials["model"] = tool_parameters["model"]
+
+        model = self.runtime.credentials.get("model", None)
+        if not model:
+            return self.create_text_message("Please input model")
+
+        # set model
+        try:
+            url = str(URL(base_url) / "sdapi" / "v1" / "options")
+            response = post(
+                url,
+                json={"sd_model_checkpoint": model},
+                headers={"Authorization": f"Bearer {self.runtime.credentials['api_key']}"},
+            )
+            if response.status_code != 200:
+                raise ToolProviderCredentialValidationError("Failed to set model, please tell user to set model")
+        except Exception as e:
+            raise ToolProviderCredentialValidationError("Failed to set model, please tell user to set model")
+
+        # get image id and image variable
+        image_id = tool_parameters.get("image_id", "")
+        image_variable = self.get_default_image_variable()
+        # Return text2img if there's no image ID or no image variable
+        if not image_id or not image_variable:
+            return self.text2img(base_url=base_url, tool_parameters=tool_parameters)
+
+        # Proceed with image-to-image generation
+        return self.img2img(base_url=base_url, tool_parameters=tool_parameters)
+
+    def validate_models(self):
+        """
+        validate models
+        """
+        try:
+            base_url = self.runtime.credentials.get("base_url", None)
+            if not base_url:
+                raise ToolProviderCredentialValidationError("Please input base_url")
+            model = self.runtime.credentials.get("model", None)
+            if not model:
+                raise ToolProviderCredentialValidationError("Please input model")
+
+            api_url = str(URL(base_url) / "sdapi" / "v1" / "sd-models")
+            response = get(url=api_url, timeout=10)
+            if response.status_code == 404:
+                # try draw a picture
+                self._invoke(
+                    user_id="test",
+                    tool_parameters={
+                        "prompt": "a cat",
+                        "width": 1024,
+                        "height": 1024,
+                        "steps": 1,
+                        "lora": "",
+                    },
+                )
+            elif response.status_code != 200:
+                raise ToolProviderCredentialValidationError("Failed to get models")
+            else:
+                models = [d["model_name"] for d in response.json()]
+                if len([d for d in models if d == model]) > 0:
+                    return self.create_text_message(json.dumps(models))
+                else:
+                    raise ToolProviderCredentialValidationError(f"model {model} does not exist")
+        except Exception as e:
+            raise ToolProviderCredentialValidationError(f"Failed to get models, {e}")
+
+    def get_sd_models(self) -> list[str]:
+        """
+        get sd models
+        """
+        try:
+            base_url = self.runtime.credentials.get("base_url", None)
+            if not base_url:
+                return []
+            api_url = str(URL(base_url) / "sdapi" / "v1" / "sd-models")
+            response = get(url=api_url, timeout=120)
+            if response.status_code != 200:
+                return []
+            else:
+                return [d["model_name"] for d in response.json()]
+        except Exception as e:
+            return []
+
+    def get_sample_methods(self) -> list[str]:
+        """
+        get sample method
+        """
+        try:
+            base_url = self.runtime.credentials.get("base_url", None)
+            if not base_url:
+                return []
+            api_url = str(URL(base_url) / "sdapi" / "v1" / "samplers")
+            response = get(url=api_url, timeout=120)
+            if response.status_code != 200:
+                return []
+            else:
+                return [d["name"] for d in response.json()]
+        except Exception as e:
+            return []
+
+    def img2img(
+        self, base_url: str, tool_parameters: dict[str, Any]
+    ) -> Union[ToolInvokeMessage, list[ToolInvokeMessage]]:
+        """
+        generate image
+        """
+
+        # Fetch the binary data of the image
+        image_variable = self.get_default_image_variable()
+        image_binary = self.get_variable_file(image_variable.name)
+        if not image_binary:
+            return self.create_text_message("Image not found, please request user to generate image firstly.")
+
+        # Convert image to RGB and save as PNG
+        try:
+            with Image.open(io.BytesIO(image_binary)) as image, io.BytesIO() as buffer:
+                image.convert("RGB").save(buffer, format="PNG")
+                image_binary = buffer.getvalue()
+        except Exception as e:
+            return self.create_text_message(f"Failed to process the image: {str(e)}")
+
+        # copy draw options
+        draw_options = deepcopy(DRAW_TEXT_OPTIONS)
+        # set image options
+        model = tool_parameters.get("model", "")
+        draw_options_image = {
+            "init_images": [b64encode(image_binary).decode("utf-8")],
+            "denoising_strength": 0.9,
+            "restore_faces": False,
+            "script_args": [],
+            "override_settings": {"sd_model_checkpoint": model},
+            "resize_mode": 0,
+            "image_cfg_scale": 0,
+            # "mask": None,
+            "mask_blur_x": 4,
+            "mask_blur_y": 4,
+            "mask_blur": 0,
+            "mask_round": True,
+            "inpainting_fill": 0,
+            "inpaint_full_res": True,
+            "inpaint_full_res_padding": 0,
+            "inpainting_mask_invert": 0,
+            "initial_noise_multiplier": 0,
+            # "latent_mask": None,
+            "include_init_images": True,
+        }
+        # update key and values
+        draw_options.update(draw_options_image)
+        draw_options.update(tool_parameters)
+
+        # get prompt lora model
+        prompt = tool_parameters.get("prompt", "")
+        lora = tool_parameters.get("lora", "")
+        model = tool_parameters.get("model", "")
+        if lora:
+            draw_options["prompt"] = f"{lora},{prompt}"
+        else:
+            draw_options["prompt"] = prompt
+
+        try:
+            url = str(URL(base_url) / "sdapi" / "v1" / "img2img")
+            response = post(
+                url,
+                json=draw_options,
+                timeout=120,
+                headers={"Authorization": f"Bearer {self.runtime.credentials['api_key']}"},
+            )
+            if response.status_code != 200:
+                return self.create_text_message("Failed to generate image")
+
+            image = response.json()["images"][0]
+
+            return self.create_blob_message(
+                blob=b64decode(image),
+                meta={"mime_type": "image/png"},
+                save_as=self.VariableKey.IMAGE.value,
+            )
+
+        except Exception as e:
+            return self.create_text_message("Failed to generate image")
+
+    def text2img(
+        self, base_url: str, tool_parameters: dict[str, Any]
+    ) -> Union[ToolInvokeMessage, list[ToolInvokeMessage]]:
+        """
+        generate image
+        """
+        # copy draw options
+        draw_options = deepcopy(DRAW_TEXT_OPTIONS)
+        draw_options.update(tool_parameters)
+        # get prompt lora model
+        prompt = tool_parameters.get("prompt", "")
+        lora = tool_parameters.get("lora", "")
+        model = tool_parameters.get("model", "")
+        if lora:
+            draw_options["prompt"] = f"{lora},{prompt}"
+        else:
+            draw_options["prompt"] = prompt
+        draw_options["override_settings"]["sd_model_checkpoint"] = model
+
+        try:
+            url = str(URL(base_url) / "sdapi" / "v1" / "txt2img")
+            response = post(
+                url,
+                json=draw_options,
+                timeout=120,
+                headers={"Authorization": f"Bearer {self.runtime.credentials['api_key']}"},
+            )
+            if response.status_code != 200:
+                return self.create_text_message("Failed to generate image")
+
+            image = response.json()["images"][0]
+
+            return self.create_blob_message(
+                blob=b64decode(image),
+                meta={"mime_type": "image/png"},
+                save_as=self.VariableKey.IMAGE.value,
+            )
+
+        except Exception as e:
+            return self.create_text_message("Failed to generate image")
+
+    def get_runtime_parameters(self) -> list[ToolParameter]:
+        parameters = [
+            ToolParameter(
+                name="prompt",
+                label=I18nObject(en_US="Prompt", zh_Hans="Prompt"),
+                human_description=I18nObject(
+                    en_US="Image prompt, you can check the official documentation of Stable Diffusion",
+                    zh_Hans="图像提示词，您可以查看 Stable Diffusion 的官方文档",
+                ),
+                type=ToolParameter.ToolParameterType.STRING,
+                form=ToolParameter.ToolParameterForm.LLM,
+                llm_description="Image prompt of Stable Diffusion, you should describe the image you want to generate"
+                " as a list of words as possible as detailed, the prompt must be written in English.",
+                required=True,
+            ),
+        ]
+        if len(self.list_default_image_variables()) != 0:
+            parameters.append(
+                ToolParameter(
+                    name="image_id",
+                    label=I18nObject(en_US="image_id", zh_Hans="image_id"),
+                    human_description=I18nObject(
+                        en_US="Image id of the image you want to generate based on, if you want to generate image based"
+                        " on the default image, you can leave this field empty.",
+                        zh_Hans="您想要生成的图像的图像 ID，如果您想要基于默认图像生成图像，则可以将此字段留空。",
+                    ),
+                    type=ToolParameter.ToolParameterType.STRING,
+                    form=ToolParameter.ToolParameterForm.LLM,
+                    llm_description="Image id of the original image, you can leave this field empty if you want to"
+                    " generate a new image.",
+                    required=True,
+                    options=[
+                        ToolParameterOption(value=i.name, label=I18nObject(en_US=i.name, zh_Hans=i.name))
+                        for i in self.list_default_image_variables()
+                    ],
+                )
+            )
+
+        if self.runtime.credentials:
+            try:
+                models = self.get_sd_models()
+                if len(models) != 0:
+                    parameters.append(
+                        ToolParameter(
+                            name="model",
+                            label=I18nObject(en_US="Model", zh_Hans="Model"),
+                            human_description=I18nObject(
+                                en_US="Model of Stable Diffusion, you can check the official documentation"
+                                " of Stable Diffusion",
+                                zh_Hans="Stable Diffusion 的模型，您可以查看 Stable Diffusion 的官方文档",
+                            ),
+                            type=ToolParameter.ToolParameterType.SELECT,
+                            form=ToolParameter.ToolParameterForm.FORM,
+                            llm_description="Model of Stable Diffusion, you can check the official documentation"
+                            " of Stable Diffusion",
+                            required=True,
+                            default=models[0],
+                            options=[
+                                ToolParameterOption(value=i, label=I18nObject(en_US=i, zh_Hans=i)) for i in models
+                            ],
+                        )
+                    )
+
+            except:
+                pass
+
+            sample_methods = self.get_sample_methods()
+            if len(sample_methods) != 0:
+                parameters.append(
+                    ToolParameter(
+                        name="sampler_name",
+                        label=I18nObject(en_US="Sampling method", zh_Hans="Sampling method"),
+                        human_description=I18nObject(
+                            en_US="Sampling method of Stable Diffusion, you can check the official documentation"
+                            " of Stable Diffusion",
+                            zh_Hans="Stable Diffusion 的Sampling method，您可以查看 Stable Diffusion 的官方文档",
+                        ),
+                        type=ToolParameter.ToolParameterType.SELECT,
+                        form=ToolParameter.ToolParameterForm.FORM,
+                        llm_description="Sampling method of Stable Diffusion, you can check the official documentation"
+                        " of Stable Diffusion",
+                        required=True,
+                        default=sample_methods[0],
+                        options=[
+                            ToolParameterOption(value=i, label=I18nObject(en_US=i, zh_Hans=i)) for i in sample_methods
+                        ],
+                    )
+                )
+        return parameters
diff --git a/api/core/tools/provider/builtin/xinference/tools/stable_diffusion.yaml b/api/core/tools/provider/builtin/xinference/tools/stable_diffusion.yaml
new file mode 100644
index 0000000000000..4f1d17f175c56
--- /dev/null
+++ b/api/core/tools/provider/builtin/xinference/tools/stable_diffusion.yaml
@@ -0,0 +1,87 @@
+identity:
+  name: stable_diffusion
+  author: xinference
+  label:
+    en_US: Stable Diffusion
+    zh_Hans: Stable Diffusion
+description:
+  human:
+    en_US: Generate images using Stable Diffusion models.
+    zh_Hans: 使用 Stable Diffusion 模型生成图片。
+  llm: draw the image you want based on your prompt.
+parameters:
+  - name: prompt
+    type: string
+    required: true
+    label:
+      en_US: Prompt
+      zh_Hans: 提示词
+    human_description:
+      en_US: Image prompt
+      zh_Hans: 图像提示词
+    llm_description: Image prompt of Stable Diffusion, you should describe the image you want to generate as a list of words as possible as detailed, the prompt must be written in English.
+    form: llm
+  - name: model
+    type: string
+    required: false
+    label:
+      en_US: Model Name
+      zh_Hans: 模型名称
+    human_description:
+      en_US: Model Name
+      zh_Hans: 模型名称
+    form: form
+  - name: lora
+    type: string
+    required: false
+    label:
+      en_US: Lora
+      zh_Hans: Lora
+    human_description:
+      en_US: Lora
+      zh_Hans: Lora
+    form: form
+  - name: steps
+    type: number
+    required: false
+    label:
+      en_US: Steps
+      zh_Hans: Steps
+    human_description:
+      en_US: Steps
+      zh_Hans: Steps
+    form: form
+    default: 10
+  - name: width
+    type: number
+    required: false
+    label:
+      en_US: Width
+      zh_Hans: Width
+    human_description:
+      en_US: Width
+      zh_Hans: Width
+    form: form
+    default: 1024
+  - name: height
+    type: number
+    required: false
+    label:
+      en_US: Height
+      zh_Hans: Height
+    human_description:
+      en_US: Height
+      zh_Hans: Height
+    form: form
+    default: 1024
+  - name: negative_prompt
+    type: string
+    required: false
+    label:
+      en_US: Negative prompt
+      zh_Hans: Negative prompt
+    human_description:
+      en_US: Negative prompt
+      zh_Hans: Negative prompt
+    form: form
+    default: bad art, ugly, deformed, watermark, duplicated, discontinuous lines
diff --git a/api/core/tools/provider/builtin/xinference/xinference.py b/api/core/tools/provider/builtin/xinference/xinference.py
new file mode 100644
index 0000000000000..7c2428cc00534
--- /dev/null
+++ b/api/core/tools/provider/builtin/xinference/xinference.py
@@ -0,0 +1,18 @@
+import requests
+
+from core.tools.errors import ToolProviderCredentialValidationError
+from core.tools.provider.builtin_tool_provider import BuiltinToolProviderController
+
+
+class XinferenceProvider(BuiltinToolProviderController):
+    def _validate_credentials(self, credentials: dict) -> None:
+        base_url = credentials.get("base_url")
+        api_key = credentials.get("api_key")
+        model = credentials.get("model")
+        res = requests.post(
+            f"{base_url}/sdapi/v1/options",
+            headers={"Authorization": f"Bearer {api_key}"},
+            json={"sd_model_checkpoint": model},
+        )
+        if res.status_code != 200:
+            raise ToolProviderCredentialValidationError("Xinference API key is invalid")
diff --git a/api/core/tools/provider/builtin/xinference/xinference.yaml b/api/core/tools/provider/builtin/xinference/xinference.yaml
new file mode 100644
index 0000000000000..19aaf5cbd1398
--- /dev/null
+++ b/api/core/tools/provider/builtin/xinference/xinference.yaml
@@ -0,0 +1,40 @@
+identity:
+  author: xinference
+  name: xinference
+  label:
+    en_US: Xinference
+    zh_Hans: Xinference
+  description:
+    zh_Hans: Xinference 提供的兼容 Stable Diffusion web ui 的图片生成 API。
+    en_US: Stable Diffusion web ui compatible API provided by Xinference.
+  icon: icon.png
+  tags:
+    - image
+credentials_for_provider:
+  base_url:
+    type: secret-input
+    required: true
+    label:
+      en_US: Base URL
+      zh_Hans: Xinference 服务器的 Base URL
+    placeholder:
+      en_US: Please input Xinference server's Base URL
+      zh_Hans: 请输入 Xinference 服务器的 Base URL
+  model:
+    type: text-input
+    required: true
+    label:
+      en_US: Model
+      zh_Hans: 模型
+    placeholder:
+      en_US: Please input your model name
+      zh_Hans: 请输入你的模型名称
+  api_key:
+    type: secret-input
+    required: true
+    label:
+      en_US: API Key
+      zh_Hans: Xinference 服务器的 API Key
+    placeholder:
+      en_US: Please input Xinference server's API Key
+      zh_Hans: 请输入 Xinference 服务器的 API Key

From 7c485f8bb80d170b5756575c150e7cdb41613145 Mon Sep 17 00:00:00 2001
From: ybalbert001 <120714773+ybalbert001@users.noreply.github.com>
Date: Tue, 24 Sep 2024 10:33:30 +0800
Subject: [PATCH 047/235] fix llm integration problem: It doesn't work on
 docker env (#8701)

Co-authored-by: Yuanbo Li <ybalbert@amazon.com>
---
 .../model_providers/sagemaker/llm/llm.py      | 20 +++++++++----------
 1 file changed, 9 insertions(+), 11 deletions(-)

diff --git a/api/core/model_runtime/model_providers/sagemaker/llm/llm.py b/api/core/model_runtime/model_providers/sagemaker/llm/llm.py
index 2edd13d56d4d8..04789197eec2a 100644
--- a/api/core/model_runtime/model_providers/sagemaker/llm/llm.py
+++ b/api/core/model_runtime/model_providers/sagemaker/llm/llm.py
@@ -85,7 +85,6 @@ class SageMakerLargeLanguageModel(LargeLanguageModel):
     """
 
     sagemaker_client: Any = None
-    sagemaker_sess: Any = None
     predictor: Any = None
 
     def _handle_chat_generate_response(
@@ -213,23 +212,22 @@ def _invoke(
         :return: full response or stream response chunk generator result
         """
         if not self.sagemaker_client:
-            access_key = credentials.get("access_key")
-            secret_key = credentials.get("secret_key")
+            access_key = credentials.get("aws_access_key_id")
+            secret_key = credentials.get("aws_secret_access_key")
             aws_region = credentials.get("aws_region")
+            boto_session = None
             if aws_region:
                 if access_key and secret_key:
-                    self.sagemaker_client = boto3.client(
-                        "sagemaker-runtime",
-                        aws_access_key_id=access_key,
-                        aws_secret_access_key=secret_key,
-                        region_name=aws_region,
+                    boto_session = boto3.Session(
+                        aws_access_key_id=access_key, aws_secret_access_key=secret_key, region_name=aws_region
                     )
                 else:
-                    self.sagemaker_client = boto3.client("sagemaker-runtime", region_name=aws_region)
+                    boto_session = boto3.Session(region_name=aws_region)
             else:
-                self.sagemaker_client = boto3.client("sagemaker-runtime")
+                boto_session = boto3.Session()
 
-            sagemaker_session = Session(sagemaker_runtime_client=self.sagemaker_client)
+            self.sagemaker_client = boto_session.client("sagemaker")
+            sagemaker_session = Session(boto_session=boto_session, sagemaker_client=self.sagemaker_client)
             self.predictor = Predictor(
                 endpoint_name=credentials.get("sagemaker_endpoint"),
                 sagemaker_session=sagemaker_session,

From 1ecf70dca0b313966d8955bdaf43fdac5515cf99 Mon Sep 17 00:00:00 2001
From: zhuhao <37029601+hwzhuhao@users.noreply.github.com>
Date: Tue, 24 Sep 2024 11:20:15 +0800
Subject: [PATCH 048/235] feat: add mixedbread as a new model provider (#8523)

---
 .../model_providers/_position.yaml            |   1 +
 .../model_providers/mixedbread/__init__.py    |   0
 .../mixedbread/_assets/icon_l_en.png          | Bin 0 -> 123637 bytes
 .../mixedbread/_assets/icon_s_en.png          | Bin 0 -> 37303 bytes
 .../model_providers/mixedbread/mixedbread.py  |  27 +++
 .../mixedbread/mixedbread.yaml                |  31 ++++
 .../mixedbread/rerank/__init__.py             |   0
 .../rerank/mxbai-rerank-large-v1-en.yaml      |   4 +
 .../mixedbread/rerank/rerank.py               | 125 ++++++++++++++
 .../mixedbread/text_embedding/__init__.py     |   0
 .../mxbai-embed-2d-large-v1-en.yaml           |   8 +
 .../mxbai-embed-large-v1-en.yaml              |   8 +
 .../text_embedding/text_embedding.py          | 163 ++++++++++++++++++
 api/pyproject.toml                            |   1 +
 .../model_runtime/mixedbread/__init__.py      |   0
 .../model_runtime/mixedbread/test_provider.py |  28 +++
 .../model_runtime/mixedbread/test_rerank.py   | 100 +++++++++++
 .../mixedbread/test_text_embedding.py         |  78 +++++++++
 dev/pytest/pytest_model_runtime.sh            |   3 +-
 19 files changed, 576 insertions(+), 1 deletion(-)
 create mode 100644 api/core/model_runtime/model_providers/mixedbread/__init__.py
 create mode 100644 api/core/model_runtime/model_providers/mixedbread/_assets/icon_l_en.png
 create mode 100644 api/core/model_runtime/model_providers/mixedbread/_assets/icon_s_en.png
 create mode 100644 api/core/model_runtime/model_providers/mixedbread/mixedbread.py
 create mode 100644 api/core/model_runtime/model_providers/mixedbread/mixedbread.yaml
 create mode 100644 api/core/model_runtime/model_providers/mixedbread/rerank/__init__.py
 create mode 100644 api/core/model_runtime/model_providers/mixedbread/rerank/mxbai-rerank-large-v1-en.yaml
 create mode 100644 api/core/model_runtime/model_providers/mixedbread/rerank/rerank.py
 create mode 100644 api/core/model_runtime/model_providers/mixedbread/text_embedding/__init__.py
 create mode 100644 api/core/model_runtime/model_providers/mixedbread/text_embedding/mxbai-embed-2d-large-v1-en.yaml
 create mode 100644 api/core/model_runtime/model_providers/mixedbread/text_embedding/mxbai-embed-large-v1-en.yaml
 create mode 100644 api/core/model_runtime/model_providers/mixedbread/text_embedding/text_embedding.py
 create mode 100644 api/tests/integration_tests/model_runtime/mixedbread/__init__.py
 create mode 100644 api/tests/integration_tests/model_runtime/mixedbread/test_provider.py
 create mode 100644 api/tests/integration_tests/model_runtime/mixedbread/test_rerank.py
 create mode 100644 api/tests/integration_tests/model_runtime/mixedbread/test_text_embedding.py

diff --git a/api/core/model_runtime/model_providers/_position.yaml b/api/core/model_runtime/model_providers/_position.yaml
index 1f5f64019a166..79ebd007647c8 100644
--- a/api/core/model_runtime/model_providers/_position.yaml
+++ b/api/core/model_runtime/model_providers/_position.yaml
@@ -38,3 +38,4 @@
 - perfxcloud
 - zhinao
 - fireworks
+- mixedbread
diff --git a/api/core/model_runtime/model_providers/mixedbread/__init__.py b/api/core/model_runtime/model_providers/mixedbread/__init__.py
new file mode 100644
index 0000000000000..e69de29bb2d1d
diff --git a/api/core/model_runtime/model_providers/mixedbread/_assets/icon_l_en.png b/api/core/model_runtime/model_providers/mixedbread/_assets/icon_l_en.png
new file mode 100644
index 0000000000000000000000000000000000000000..2027611bd5e8b4c7d06a5d00e515a0db70e67a17
GIT binary patch
literal 123637
zcmb5W1yo#Hwlxd{hv4o`;T8z)?pj!I3YXv#+}+*XJxFl(Ai-S{2o{{+@)x(eU%%Jy
z8@cZv;}lgKYMisrUNYyJYj47p6{X%H5+FiAK)jWa7FUIUKrn=WfW(7`1)o`(ohSys
zK{}~Qi9*y&5*>hl5eI3>m?<bg(1Sn2LqLXFLO{QM1^glazaSvq<UvBb0l!1Op38&!
z*Iy9~^WOaHGbG;Y7u6<(;2|J{A!Niw)ZHO}Wg~d0&o94BwQ^`{_jS=l`T@w`$syyJ
ztSX!5nSpC7I@Xm<F4axPy6xMmi@vEXz-s$rYiovu>WJgjm4%hTF$%H=ngTLNTxEE&
zpdhNmuL3Wd9BYKMqhC!EmQTJuDhkrxtT|0xWSt5)r8?z25LNF>LqLChQ<2x&&AZNJ
z)iz_5#cl;%Fo?<~n`}`WfzBL&+Ortip#m2TTt)Ypm^tJGRVqlNmTA^0NkQ?*7;a*N
z3Y%20C6@~utEH@pL-B2IR3vYDPfL)yu}OUxxCrdeRxPHRJ;yCBX|7vzSE|^UsE=To
zD$2)YxNWcy;voL7#||CjKH~H)ga1(J{HQWbE4(ZvuAc}@2SA(p&{YI|7vH4PJOZww
z_?~e-!NUKm5>z_H*@-#qO`Q&b5R^+rBw7E5+kAvEC}OE5AmLNgvAu`ch{<nV493`H
z89!+0|MS5v?}w;`F9>;1)AfwN`SnzDQ5az-vFwv-#E~x?@vjZh!e>o`f|#N&%s5I=
zGL&a0#hp{rDQ%F55xK(o#WD<E#6-+Kf4hh5-t}#KF$uNeBQk^jzs|wyZ29dWC_|24
zmI8R7U~_=jB87In2;I=sI8>!`U*yVzHLBn;Y=3WdGOKVp>=}z<NOSmSjC=Y*kA)*C
zS8D~ymG(CT{p>2pdzM))SP?s@N^^<-Kbru2dn!rr(}$m5dY$yBSw&ROucI~6&-RJd
z@5Zqo9t~d3+c!#v#X$3mZuMYymaaBGvnA#I1ZfrN6wV8@74LV$F;)1h;t>D;dfvs?
z=RGnn%e<I>7WvJBT=61$ZM|J>hm%-gGw!VHUBA!X#H4{;_W>g;RCoeZ+#hwl`NpN^
z?z~7zlfZ(uE1Vjw``!NeiutSt?GZU72H8Ije{lr)OfK7~1y7<ee2XY_p^=6v|7P2p
zp=+{cPHtwOZkzgUm-RN6Qz{*n%fx@p(GIPDL_`GP^o3LBh9^I!K&*MrCk`gsocU+e
zDZ?)LadfGtmRh~`>s{^{HiZBAn0~_uTLL?u-BIs~h6l0Ef!hK*lj6Hqv^o%%pJIh)
zXAd3u%AVm8eqZJ|lZXGr%hlODYi#S<G5$>$J0@N4nAOuI_wd0q=}8V?@cTwbEgzpx
zbosFMf%)jk-9dSM1ccD(8UjH5*FF8vdd8jbX|Se-kQ(K^a^QS1u6LD3$n{5kZh<J$
zS(zvx8;M&|<>`CcW;s?Kk6-R~&$R!Zn<eP8Umtl@Kj85t)5efo1YybQL73&_qV_m#
z>yd0ru>D84U{LwF#UTLtE>CUQL$Z<g?r>{N-?1Bvv7(j*lsq4nrcHez=UJyL-ic^Z
z>9+e&GzmD&x0@pcD$rF&2PgW8ZLLospBQnz|6VEfaH$^}uz8NHAjs*ox$7+?cpHAh
z^+`{EObUuh_^%_56uLw)n-`O%Xj8;mD9pCtGzn8>NW3G=7iJeGFYfnyt)ctcw}f0(
z6Q~(B)c=}~e;M$9BphL8w6m2N=hmKgl(yX0pInUdYIP8;1&uxVGS@i$d7GB_ceSnc
zmIo3VTt&CC$M2&WG6-4I&G~D+uosdN8xd-jd_9(_jU12@{_FC>D+o!cskx9a<$H$7
zhA>Gbg3yqC$+akkzC7PkjUO;N0I{0=-Ef=825(+?A0+|0)#7c`@U;Q8!ESowP^4ma
z*2;^YRdYZ2>Hf=6Au;xY{P-i?M~SZ!ShpLtw2@6rHXP$*a^4>NhZ!6*{i1O98j(m@
zrw&4XI{qNY1=5<dHuw@&D}Ehv6+PiFd_D|r8=h|!s8xPwsDGQZ`~Y3Ivvr`e>>F4l
z*2!WEhw#ZTJBQYq`BcDW3$y9QnR^T#gUmmRN?BW*r=z&ISQF=c4GE6T#)$D9ohKrK
zaJV^g6%lrET|^Sj+vQlBl}1Y?S%!1vZO5)W*55*DczoU2TDs`I5a(iOLMT2sea8un
zCn)05UPlQ*lksHDEdAp}<@}}DP%M_>y+>kYx+ydr9#8kbSE+Jh{w5`NNP`oRK0kwA
zT={{e3;j>-=V->|GIflNnbBv_Cajc>NxG*ZPrgTpeyTVCI0fFT<AYuhcCQp0tK@$U
z+drK=rT%V)d#brMPfrST0mpYyC<GRY4+;ZojsY$oY;+0YmnQ4Q2xAyxghmWM$L(}w
zL!FT4?4|{GZ^6w%i)nKm!sLEPt^9<eX?q8SL(B#FJG~!-h@4CzeHvO9X(FUCIX_E;
zJ*@M*G$NkD6|)$fz7V7)T7ML{%g&7XZ?^qsHu|s7`|HCQhJRJ|)fvWs{OHV8QU*kD
z)L@y=YI1E1<vXh^P^I}qt~!_Q!-%?@VW$fGulAx}Z<M<{Q9?iK+|gHg+?;eb-n=z>
z_{%B%G!d{w;=k92W73W#7&}v;QLsAw{E7iE5C|gf3U%*R(o*(p!yHp87&xIfu<h?+
z)TTsKn^pY=on#R1C~;>++)>%HjJcRdn}_U5T1!`+H_85Iw8WqmM$!t$_^SN8GCU7q
zQr=Bp2x~4Kq^`kNPM4R%u2hb6MJ<~`nNR6&hY8QbS9ZKv=vSrp9bo*Usm*P(7_<r}
z<gr7F9?OvcElq}tz&P0;Im&KQAk5ZUR<<9yw2=Y2P+P-7x3IFp2qw|4P+Y^g4a`@@
zc?^U?6JNY+5S|N?$P)DS^kwwF=WD!sV`l`<BJuzV?{)4JiwH=4+l}YL+!1_l4`61l
z$zn4Z8mt6XdlEW0|1-DzBg6dfpIZn}2L${#;wDZK@RtD>?}?Qo^Sn6eHCN_pi1*%G
zy1f5xXN5bZhY^<9lYi$e)wsX)c_fY@rM7Q120{Arc7x$wp78p1`+XlOt2`7{+-K;D
zhAPmh@J9cccOI+5c%$X)bBEn}8*W3@Tden^|HyAo+h%k)vizsyjHDRyvW}c|3@8^Q
z3fkKw^zJi7GTwo`TbCB62NXDy8pM4p{pO-~@{GkAy^)+%iCYi7Q+|$1LU}<hut{}R
z^<bv<=jX$x?I{*w_d#r9hDfSCk*KMt+51{S^qxzwSTI2YLrrT6&F?}(JCX%3j)HO-
zr96sT)V<m4+kHf@k@Z-hP4SvzspJg-NnFi5#TGWce4sHbH-yhYNam2sQh5~>6-y0O
zXT@;EzQ|);DLc&rpUox?VJcmWjW~+Df^9jmYOU%7DEpakt+k@R)O36wVIz^%QX>0$
zvy7aJ2UPmw*gs}Lz*T*&qUDyAF_>hpZ>OyRcK4|UT_p1nBRGeHpGI89!7G&=Wd7Y!
zAj8v_k58g>4m=mFiV(7d9N&g^75S8|9-dw@k8I5z3NX1nul9Yn_vsWP=J!d>C6y*9
z(<%I@DR4zc%lpc^33ThJ=u=8NwnFkZk%P%<eovwHw{66Bkl}&;pHckZPSP8KR7m~U
zRh7JcJVi2M@TjPS!ZNyO76h~U`k<G!RxL_>`#$wkp)$J7_656BX)^P=As;-<Wl7<b
z)KyARvGN}?VA?Tds{00D{~da4e0<PzMl&nH)N(97Jar5sK>=b+zGd6ZGhVIc(e6_Z
zs<f*BLdqiUUEtZtOkj)1$)m@K92&{z|DrMf0|ozuT}3ev*mbJITnE%(k1l1ma-TuE
zw(@V*F(*~yI3xv<UQ~o7jq5^R>u*^5sobY-B+&29VA1J5G43|iLA|2YJ_%%fjuXU2
z$}i=UFQ{V}lvMJmq|l}ZB3ev{2A&@obOgDDbaDl%2Tm{4r9YKYI$$;zZ0*cnz9aCq
z;@iQLV07s2!6+>hC4Z*;pZV~=&E#LG=#Pw^2L%l2acf#BDV;anY0a;^Yk23Kh)~wO
ztYDSU%kHRC`7rUO;+@>8X#K2okG&|)_YPj|G+B(je#C`PHEqao8t@(bq2aS)I^?n+
zLCSsiOnqLk3J?}tpuw5I&Crg5wZgo2p9wOxF-k6p{2$0w@}+|}_vMz{J?YFx0OM55
zm>%1*jN=KrrT>Y-{{hO<WG2wucgbv5dX;!)ly*`vYeEK+wj9w`?Fb=N`oT2*(<dr#
zsSiKtQ4Y=BNg8fagGhVkP6qE5P8O~So}w36-@L}|mH<?N^JPUC40V=AUGt^OX}EMa
z*y?*8uwyoQo@lZhl!<z9<8oOLDPbLG#aX*aSFk>X<}~_NI<j%aOgFL$ad(9l!^@WC
z$;tmu+yDEdFf>$a?KEcow`H99TRu)jLQbL8os)u&W@6>3)o80EHY?SjjuxU%TG?pi
zIVxa4$z%F`+#zZ7b?r^+b<J06SQ~zDk&=plCCsf-W%n_tqp+dDF(MVF#3mdakI&Tx
z9Xud7Z$+68o~&#|bmMQ2R5qQFLKCt8??z8K2|us}Rk^}Q@a+U*^ZuIv{ipr^@{SRB
zNN`v|e_EuU7yL~+65qZzR`xzH+Uhf6h12w80>IgH=Gs~%hK+X$0SuA2jFo8~9x)=z
zI_oK?$V)!piz{K`y=CCECPF~6UG=G|sF1}1>^sH@=L&k89`sJLyDA^XUrrXo3ApFv
z!+X$7PRUC(+TKP-Xq4MH8do2=pyM5L!vX6`K#6Bb$$w-kWNz>~HL6N6S(KMahvrH5
z>M`$1#W;T9G<~&GelpswbYj5M(Q~?bZ)F%NiWh$a-B^6NdO!aluvO4f`)mH?NiXEp
zr+Rdu^T^n65lO6x%W3B3=br|2F;p{5(aOro(qf#vE7cazvo6s`0jVIWTnH?3q2rW@
zOZCt?*PG9#45y#r`b|6O8NZVKkq?7eP&7CP%!IUk!sgUI#Mr?)QsUXI`?}SxW%n`L
z3h2Z-F-ti)lzfjg$LUZ4YTzeVcCysw$?){adDrsU>>laO;g4+U?+&k)8=-;({6v;|
z3KfS9>PC*_+ykkSEM$LxYPcesw7#7QWGM{k9jy*7rVfN)oW4_;98|@b)F$@YBNeAA
z{JxV)?zkvduT0Ty`uuoz(vUh@8&3UilKroY^>42QpAp7_Qtyd?8e;fXhwM_UX)2RJ
z#AMZK)u$RL>Ea!ZR>&-uLBt54g<o1})Ug{0hoxSDFWjTmf&Qzw6OKR#W1mDRMcJ3z
z4=qv3RXp;h3(2YZK0Y>fBCLm5BnY_rc+l(Ye<~>;6-$zAejewd{fN>+@nJgJxFfC@
zUNo#A_lwB<{QOBm#IVV2?jQZZ4|+xz%w8@1*4F6{b5zCZ9t|hu$$AdwA+<VnK<_KH
z;Zt2gk=*62?yCtlvpc>fW<q~NZwI{BU6=q`x=hC3bw0Roy#M@KxC^(ze-)M(&l(Fc
z#&0wzsWe_l<b_fzRjd|1XyM_U6D^mj%Kqk&+VhL`8<f$u4Ki1|XyN9BQ~%oEO=cUs
zG9-guQM*d4Ql(RS{6wO*ymB>Lw3f{@ApN7l$nwuyRXPOhzOOtY!YQOdsv9F&7V;Ma
z+xAzPt)mPv)PId_41NTo!Cv;DF3DL%LSk@F4Ty=P$m0*@qBuG!O=#Gy<K4EB?$Ton
zW-#kC5E8{4{p_&_NaiMI>hRISebc{96}h<l*#8V#rtiU7z~k=G(j_N#Fj9@nBxBwf
zQ`s1uoM#-#l9OSwEuC=~0dIWzUL(lKW*IoCH5ob0F9{yr4vIgCg)Eboat1VY)hBue
z1}V8T<_CBzT-@=QKuih8tnsF$AYIJ_3*K1t9uERq3Zr(?Od}G^+RE?9!XbK!CCWm-
ztRUt8OizTN!qLD;MRBE?rOAN(tl=49E$4F{_malQW1*vtEjw)INF->B3u5NQP+lpE
z#T1i*k|d-zbQ$Qw1ILg(2KQUzRyp)*WZ1yiuem~-O?f*C;dB~sP`kUkvuOBTS6c|g
z7;Jr!(4vU3bst=pBKus#O5P~uV1hYUM0<R$_J3m0f2dq}tI%bB^R~|A6BaqDAT4nd
zbt`Kby@ToI3RJF&V~c)#y0T*5e*1YDIzXy~8M^<|CERro`$iC$!DJVVHR%6Eo8y?E
z2UKT(=)e^5)mEpymd_F&#}E=VwLH=GY{|85iAfr+wv=hVMoS^c=L+gE&Zb1T*ZFDe
zsj#(E;jDbyGM#pDaiMHaN)P{YVIfT>0*-H4G{cI9_xE!L<@9{JyBTp8^GC{E83`3_
zHpN4w6*eiE#L!W-4(wD48K9<L7A|Y5*Gr!~ShOQT|Eeb&{QUqYBaT78DHxMHx}uBW
z8K~v=yrHP6sW$_Yj7Q}siBn0gp6QTeme$tRX0fp&<f84kGh6n3rzeC?&`yp+TB-kr
zm^i<hn-}0#ic&b_N~=3`i8pWho=&Apwj3%QG}jZ=k_=^<;<rGjb~bL)pP=^1-!@uW
zQHC-2HDAn#8e&;#pU&;H*gkx-oGl5ix-2So1<o)*aZBwgZ``#d-U0ePc-HAs&E_H?
z%2s!?Fp4Z?`5+`1P_E2Xq{VgtWV8RSkcs1ZokYUJ8VzWeLUO(F?eMuFXjYF5<6Q>J
z^7=Br-EfW^8S<!nC?N84Ge`}R0hW}60utkl%AaRCSL;}s3uW~aKOJ>2B03nw(Y4@D
znFYQe7%>aL!)t#>{J@i6HD|>$7Ck;=TM31;V{dDa(j093Hvx?Q8o*aUMmSVQSao?`
zO>;|DW*KoPvT6cVLxs?+{jLpC5|=?*TWUWML;!zDrOy5kSHeR*i`?2ZbSy00bx6zU
z@1mi>1i35nkcLO2rr*_-pwe6;9FrRd7{tPu)g+=dX{|vO$Zl>SGh1bg4D)ZL>;GlW
z5nk{dUF_2ZUWhBjM&PJ0&Q_(Q<z?WW8zNGiXA7p1$wMyM(9hgPJOB)Og_0T;L{AnF
zT5PPu=t1*C^s=Bk#>`i9s*2|cjjbQ2yrqtSqgBe0FSf=FlWNAs##VE%>)H-;!qV00
zx)^B<*F*;JhA<e0bAHZp>@}=oE+rTk`X9jdztGM-2Mp5uo)LKp7*}N$!~+eP^X#()
zKV7-w`rw^E&>I%V*>hpNr5?o0_i|xQs-uspJyWRmj;wdy%eRRWAZ8~3Ma-(89=)<E
ze^X39V<qZG)zUQ)7#NstMOR|R+6U+og-jsCZgw!BSI_e!WnX`r0fm!UfUvf9IzLA%
z;;-WLIPzr|os@a2Q6%px;f8T@k5`iVhXXAVJY?f<qrd7?z5+=NjtjhsbwuSRg-L0=
zCoSd~yqoH2Vle;?yEC$}?^3$Qu5pv;4Yt{foi~))EUVqU-T2^Jz?avSf*8ZZM)ORi
zyoFJ|zf)b{q;P|IM#<n>w%qWv+1s&Gb7}!a;<DI@A8k13Eb-y$L>z^z#@vI|z+V%=
zT=j4NG(>Q@CFX~DwIQIH)8ZElf=e|My`y&?iz;v-J-uFcmcc1UB~zU8$4F7qe#V`p
z4x>F{YuiY4Q~|uQov>IxtsQijcr)5NB`|#uWbUexPo{@TPEJ-+zV?;AQ+AD~-UyLS
zW41<6=SzqLY2UP@R7ay*>(|8)IPzsPz5nybsQwkV%$ih*F*z;;Uf=-^eizYUG6;sG
zLqy_8Gl%}#Q~{x2r)NeZdig+qoLODf*CctTAM=s_rqJ2KnEz&V6bB6YJM8}F7Hf{B
z{>4m8OtC->jn?svUB`BTWiD;c>E`2LG{uu%#Bp)Eo)9@y#He2`jwplSeLohI0iRwv
zp}swY-zuYg$n2cR+xhc=mu0~OheW_>+G^!g<y#UXUz*V_g9zs6eY|G8N(~hz_cS|M
z*JZ6s@b#S%2k@Js)_@Ko+r<by7It02+OZZ9nAw=Yhc(>bloKEdC4&*d%~TS30TF65
zL};D7w;0YTPfl26RO2I7aWSR#MiGIg&v--17bCmB!w@|%;3YH9=v56bSQ+4cE~0K-
zFGvB|EcZ4u6~R<B^Iuri4&Fwxp}+8({kR#J{a8JZr!|mK=_S#;0267{a{Q71u7M}|
zTUSeiw$GU2tIMN6*1IkZYo914iFf&f&L*j{uQ1YhT`B<j-fC7ZlY?7VvOO!RA`p9y
z=~=3DgtBsCYARA~V*^e`m7H8b2GW|HRiS^w7C(UVk)w>eO{MKAond;OPGQEpP2SjC
zLtwS}DBk>ZTfguI`Wc*e!@w4g&O?ahxaQk#ar{mJ6tvDLkOp1lNZ&<}9GgfBQw+a$
z`oJ~z@rZ0M-MiO9z?q7bXz@rvu@QBAOMu;ap-P`Pk&OoXwPXkEF^Vs5%c17PMYL&|
zc_P>>)fut)=(5OED-RKbM$-yGga?l4OZn6%wKz^>aH@;Oz9C5bbK0Ux1?PZ-aSqGk
z?SYQW#=ZoVIF)t{;p~6{_EIm<oc?^7e%;YLjg$Ph3sLq->*f@r%$%70{aDX;kNn4%
z(X%Tr1W64kc*Ur$GmRT*`I**LedlAEU*N{H#*?r65ra0?Wx5!7s&0@yObtkwTGFUX
zgN?x+PbF%I!ybwyQAfPKwB&9;5zJ=)2%Wd&$fv60;AqnWOgcYs41$<EO*B#%l^t|$
z%-1*|+Lr2H=_mU?)=Jjcy#qbgBSvmanh~mm8ul6LNDH(F*#!i~I7rr-O|FGH#-wUb
zzw&sI!a?|1KB1?Y-X>-{y;;<TUe$<NwyS0NRAs;U$1iR|G^#XK8i14{7M9t~2>mx#
z2qPSF+Fd;084m^_-t^ZBXFbAxo?EWQjMrX)h){TSzj*j8s3WeL;G8L-y1Ey;D#Z|W
ztgwg2ohj=}idq5S8YO5di{0lUV<=&R_6#pz*c5k64tt3x;Ui9~_kLk=Oo!AP{~k<j
zDn5xnQ=1X;zd&(xE;T4+=`si-m{^%GADWYEy-eA>(@Hhc=$JmN+oP<T`vWYv>kxFx
zktGGa(c*_){0){j$}>cD!5Lt3JiVe59Fz?(Zo=V&&*w10wkE==Pq7Bs+6p94L%w^L
zvy1n+h!ZREH5_yhu7U$bNPY#^?7<89?Vg7;lYq#|p~1xVQQ1a2Ow>VXP*3B?Who0_
z)V4-m2NqA?aJQh!MAjXLZ_cMMT4(cLCPjRWd&1Riw((#Y-zoj_=Y+*03QkxXg7KLg
zOgWY+<3^#@K;GTDXkG%<4l88^*T@Rdb#x}Z-bW~KdF)}#ubz6Hd>h65SP8tJm>lS1
zeBv&y##bKjoTv|**T%W|qIH`F@`AxM!argHm@;v$v|u=VT8G|PKg8{1Af$ekxBP?P
zlf-09%hg1EZE*W$DH7jMrDqmO#g5Gcs%5k1N7?_R`s|gmOAH0h(?b656e!Sr43rIT
zmQ?pCv`lY!3)XLv+5!RsawQdqv|bVD4~Mh>m|_*@I8bnoTv(<%Np@Ohh>brLOTd23
z31QUlz1CzcatEn*Yhha}Fd*08?p(0hKG0ZboS_%Wg#z>t88l#C29k^vs<_0^kJRhK
zikV)x!1)@J!FJfhliQ4HkK$BbYmfl88y8V~JQ-9i(rtT^CRoVks+`#kQilTml_u$^
zASB3C(9j2w9QOJU7ZipAA9ER@40#bi-s5yiDI{2+_w;Q%!j86|9exb0Tg>|M9n*(k
zEYvVs6!QDZY&vh6cehy0)4PSwFrwvB{JDDFA%Jx2;ayxwCOl?4jK#N+{%oo4Qr2y*
zV_G-w+OB`>gjIV#puLCSP-4<RL}}RF4XsN4Gtj)3U;!>aY@QHNn<1Im4}|6YPBL-0
zp8{Ci;KkQ;A+tisHY3UR4jEDVv$AIKIre<N=mLAvC9rY8GRA=OS9l@Z%m>=azf9|l
z5l0>Fz@Q9SIztL>Rl11s!i?;;fwO9(JvskL{g-xPupIg&=)f_f2X(q_eMS_(GdTb|
z#g*gQ&~bGT6GRBs7`$ssi;F-0jngW<qO9QHU}{hmpQ$OnC|@41$5>g1hC^jhwOu4y
z+`F*dk+)8roc2vS6l3Ykq4b!WgjNC?W=OD8e$?*!Hd7v~YW?U1n<C;u#(Jw=h5(7G
z-8AD3VkTSKb7j0ar@SyutrR-()=*xV+?9UG7tSW^I_FncJA?LjP{xP{84Y6Vru<$S
z2*(sd6dNwQ?QVyzyV`mm-?BD3RSegT#6Eset=n>f=OdLHu*4LzpS)zYBS3}X92sxc
zoTw;>*Xs4HKtjYkm+(*i3%=jK#J<O5<LXqQfm~i0x<IQ=#EMoKdL|$n7glYGG35L5
zN&v7=dbr2zLy`N?t{|i>ubxkXID>VHDs?ISx=Ep`kM<-cuMfH&8&;E<2s#q5`cW=U
zl;)!CS0j6HgM9uu_PvoFQiN8zcE+^ULT)=FKgTPg3FBb+fl7GL2p4bTGF3l(@G8ap
zS&|7&PfxEXwqVb<+<ImBp2y6(nG+14J*S7dF^|TUc2phhYo(eQFvT7kyNGW*gy44T
z%Mfquc>k86%)c@et3HGjQX7?S5%iPPeT0++4_AvaqK3?5E!==-X<5A<BDKmoKV^yt
z{n8S3QJfPQsQp4#EseV20sDHZh#865rjt9^X1Q-=Lx8sob2<~VDKBX=UAXZy8GXQ{
zU)A3_^G9GLstQ`j$}91vJy{3+L4mpMQ<V$!s%y^RRV!3+q0l=#u4o<Y;4gzI?BzwA
zm}&;29BN<1k|-8V169ZwZ;++!J38mjrE%ac=`-G*{)W&%a_>!`0SjNcYn&X`uK5>7
z;k!e^xEo8i+`nsgf~7*RlAGW|<dUjivk^cuclupABZ;a+<=Ag!bgKvpr!$V`A#KGb
zs?X5@p;hD+Kr!k4(?7@b_*a&qaE{$LR7b*DYdpF!Qr_zDN3)Ja1yD0G@@VP2fuXAp
z!=8BC7qnpmFzyV1<}09%p!9i{LX`z((BpX=3s_S31BqIPq2aiF7_VXpStUGjpz07%
z%6Po_cvCTxOCd>ezfC%?rsjljgk7P*|A1x<X|NAZj|2mc!|ZF*{+lUo1zeaay0ggb
z_R6f7Duh<gtKsCc0>gvU1hP#i6&$%Tb+gxI!ygWRDh#k{fMHZp0bp!#)vOq^Q?rZI
z39(FGs2y7+lG{-1c~>@3R!V^{abLZYl!&aYCFu#j)Y~JYs6Z`wbeTj)@@5>!(nGN@
zelBeo6~Y-?aG;g*@$JR0$ar%KM%(BH#!cg*a5GK@N!OdWoInslOJR+eoxD}E*=63a
zb*_TKGUHzYc845<EWWuYuXc+38vN~p82NK#CIn-sLXhaVTQ>4=k9v}ws_k1&ySp^j
zPG}q^9V{CwtBhEk_05D14q(1PRqsL*cPev!g~~U<E2W$MXqpj+_50sPr_!kqf6H;d
zlMBOCxdTAl3ftiRFx!J1{h%Lbh1;<oy%LBR(7s%wnHg10pQwSFG;2DO%p41U8|8{!
zw+8IT2ynq+G!`+caED1jjj~TCvemHC(<!Qga+}#;VzIGC%;VMo-FclFleDdlBN)Se
zO1Q?J6Em`L)7qAKYs{XoP@5o|6bu#wP4WF@;@oN}FR|+aTqIIPbJ-mBrgBdM9mhrC
z(R4h~ZW5c|oep}VveE6;J6w(ydfHr$YNR%Ak*RoTRLS{?1tGej+uvfOifaCt<G^x>
zay|5zZK&d@^ZG4}cm(?P=S)sD%>?FGhd9m1rtF7Yb1PX58E>p`e-T|tYu^Eu?35Ym
zWil?3N37mq*9D0Q@~COOo1Ez~Pv)v^v~CG;VF2u4rx0p1h*%p`fne9Gyp!m%K+XrU
z#V+cRC;+#lW-m46pOMM~yuZv&R)DCBKqI!K!E0wG1OD=f5=f>drNP7m%fH-T3gJq!
z{knfu@xceBQc;~E;)m(!TI(gDva<tIqMfL_BGGahl@ff<%v_<xA^urDC_O*#J(>^6
z{iuq&%7AQ-tc7?OT$bY}SpjL4??v~aQ+o^sZ*bXgDgUl8>wisq+n8wfzXwyx7UQn>
z&Pz^Zd5P7J43o$7777`(e<-b1=?^6BuibJu=ga_HrtQ>!|17P><j6KRj`&4uI9D1y
zbz^kQKv3%ique;deuvxeb&E^UnsdIa@1QQCNde!W6w^iLAYC$ffpB;AR|>jA<!_tg
zOHOOsaF@7~FXStG_Wl@}3U2IDV}Y9@Jr*y?p7Ds1yp!Cou{%NurIz=l{;Q=u>1s2n
zWZG048*k5Cxf<=3;7Z#SS;BSjB3r|mG6{wsbrq9^E47b-n+aiqSFjJqh)P_A*jgla
zGt}w{bIT*mm`=oF<(6EwquwBc7P-cIu@Ka1{my%`zQ2?1wq&5Pk}h^E!x`1%4c7fB
zPsIgSj+aslg?}Hpo^Z2U>2^l!zSSE?blBPF54Ti#0sc;dK&()jTRHqb>BK6*jK+Ny
zd~u$JEjVh4^mFe@9bh!^84-~%AFD|V|F^z7U%l6sI|#e)Q0H61*{!ru5+5?6ZYlY(
z@mB#c8TtqgtGt<ocUt|TZ9q8&6%4ni5P8a`ZsnaQD~C#xntisF{nNKT-#eEgOul`-
zgnzA^cCaAVaK-aKBrObPYpAF=te%s-w9t~tp$0694&aWm0j9(HF5nSn<Apy#Df{t4
zS!=ZzBlLXRe}A*}#rn<)qVF?0?RY>W)F#&zwMvN&G7DmSfjTwfht=)(6zgNI!<(Xy
z@B^cv1RPs(+k?k}FIGlI+ZJ<A96`3>MXxOEPe&~{;ihO#$wRkZ(asrZwYOr07BGa0
zkKi(l*#R$_+A60p7%W?2M9PcN<xq{6o8nG^_}l4?Dtsy%ax{69Ll=RxV{EqNmHo4<
zS?$ny9ar1Z)f}+0L4v;6SebdD*bQOx3j|5cKk@bUbtJ;og9y#(j%!uhpc8KvGVM%*
zn4-)M=s|byx0n3khfO3!HXMdDuV{Qo{nHTmguXcDdGB=dDGg!0ri{4v(1_nuNxh_b
zpk;|SLz7EqFH3&wSy;HZ{9epM{=xjV6W>V154@Zv4Z03$tKE9B<_QradgxS+(Ks9`
z`RL4|aEQ;l9MMR7QZ=c9rZ+Yam%)W&m6$Z<_9N|>P60Pvgr<;Qt0QRSJJ1~}<Zq(1
zClVZ`yMv!G#7YwG@>xb-C?ylVpo*kdF1%yVdX&jV;;PUnGgoJ~@|&4Q>V<WIsv^q%
zR>EKSllZxwV3sC^k4L7bG?G?q_k`5qgWdqF)x4B8%w=P(AtBAp2E0;XvOP*tT}tS%
z1nN<^Z1NRH8=0BsS9V1S?@$#*@2dt1QjrZAWr9z17vH#zqj`~$K7yObDC+W};AwJ$
z9{NsFw_@o+&UnoambIwIwwzJ?FD@?5;h8_Wg77!l-PbbqrO5=Llw&-l7>$h>fU`wH
z>rgL@UYfb#rrJM;&oiGGoBNeFOmEw_RvB5MP5w3@Eunmr<fOJKFaSBVCsrSd%(k~^
zDQLntX5Ll2)v5GwpgDD9H>tE$MoVAcjUwh)Dv@F06mbC#i{;~0!Ilb!PPY^inLDGP
zmEFx$sr|}3mEG+hdp9%(m~E8BCyY=vlMel9=}nMA!(S3GAr}v<VhN&2&JW_0t1Q5l
z<Q!$^sm}%qG|^{)8hy*TqNVLA^jVtNO`@+mi{zCV`JF8<RsfqHOv-mx?BX|_G}qV)
zC!}Cmc<((m+1Z%>3Q#0St>t*@7v5XTGt5{LfqP0D`;e(mv|yfhKQ9LH4V>4K3{<cA
z({6e$As~{9^(8n4H~DN=8UqYj-2*mHNrN`Gh0m1(l0LFreHLrBJsobA^m1~#d{)Th
zI+hSjgiBiz|6>^X6Valn<dY4rW?@b+SgC4qVh5*O;;mN0;xppTY$T|+AVy@+84@I2
zv1r4Vt)i?jzOD0YLTCUE(nCj>=5iv<x&4L$4a*LjEX(yE-V52U-SJez(d&7k{5&Fv
zCf2IZ`1lU1T2xLoq^S-orS7MwjCChMGqOa7JE+m6ggv;j?mibF?{e7K@Y718VW!N)
z*n_Dc0GkHg;cLrMnxRH1OtT!saj#=>;3|1H<HT%vKQc&I`>>!lmQ?SXr7_KGhVg%-
zCR?;_Wg!YUh4VE5H#rUJoL7X<C#~Uj8MHShI^pbk1>}?F!H!1<ofkD972wJzNC#>_
z`{TTT;RY7!LjWt`q$)K2?@|q>D!J@fv3Yb9Hv$mc+^RHIgV(g>_Y*1$)dC_%N4JJH
zk0=$mV-qX$4-v+a*;=b=*?T{Cxmw`UU^ILH271f&8f{v-5>Ck)ywKxF%)y@OVxjL=
z=oh>CFaXT976K*S9F5enTIW24zvYV~QU~Oj7Q`6ks;ZsmlMV6vF-&<eq$G6-pALFs
z7oP=U8ehXr80t>ZkQKtljm$!8OPky2r@YG0m@eFrcxk~o5)p*s)GwH+8%Q|#zmv7W
z32s^sD(m9`GY{$xLzn!qjU})lY2_h;`g{aiUDt>f5s|vwVFjprjFVi(864;u$ctE<
zKKngcRrm1FeU@F%6`lLbdv{2xl<+Hy$yz3@_6FG-GJi(U?1JU|@C9y8Nu&Nx64s&j
zZ)9~@S|76Qn(RqZkk^PPiIMBIX{~HKaVZs&#5})TU`HefGa}hdzu_<-y4s1CP(d{M
zOCgODmVz?wJd>dH)FMN3&ofEr2#?=;4&90@9#odrjh+C2@#}x}Xa2Vg`p;$a<L)Op
znq}S+!B!;D9)kKtzs%<G>{m{Tta<c3DdbSOMDDCm7f=Z1i|?cFx3)>ky7b2S2bM*I
zb#r5tlm&oyvia~}xjTfRCn3nGu~?T$+j5srtTMno^+IgnyI@^|3u6dlV13eTC0ooe
zcM(Ircr&*n)LRDX`wM?1B5&3UhoOH6Cj(@xFvA;;F$N#>^?fv=8hJj!A3xI*&>~uX
z|8lkEfb#fvU2U)f$h;EvBEzpX<o70r7V#>iTsi{F<y0bC<Lp*Cr3Y>4J?|=$SIWZI
zzmklZ5zOn1aJMpmmYO|z+_A>#xynghs7Yr%b9;I#S<);61jZWcH^P*31m<pK$MS-)
zwb91{Fj8h1)*GU(bIYCMt%(PtW4;7SG^3AU)rPxIhrdG{$F?%~u-n8kY;j~`a`!UN
z{7gK~t`X?tR&=@P)*iL?4>=PaeLwsqbUUyCPGUJ%Ay@E;3Slp@vaF07=J;8uv=fXU
z28#TvGJot{y`u0x+op5CA{G_@gvfc&P#H@eO}c5+egI_hjFTPQoL?CBFurBRoUuuz
zwgXM7smOeuzVdIY4j}npt@TE`SoVDIqx_)*R5O=QNZ%aM>earv{5cZZ`vrDyMQdkY
zlVlXNF)NKmMC+2`N0?n@t+gwkEKXf*iSdoiSoE1Lu(IDWBu$?9pvE@X%?HdV8*cQi
zQSI6tj?ENmwBYdQ`O(xMkM)n86~Z?N@}j^S-c1HPWI=Y{XD@@B$nlD-i(kXVaGwv8
zRQ^scB=fhSj8g?=*#yQTU8H;{@F8xTgVu_>HwnFlp63P7Wp9uRYNJcqO5M@AwQy@q
zd5%JBXsq2u^O`P}SG&qdZ&$crVB>rtS!)+g9@0a>^h^<+;3T~RFltZWKz?iSv{d_;
zQvXTVL3YuZx$&ghi<%=@;|pR+B}$W=f$1Hh6#2>T1$UDPe~g8Mg{OVUaRB4vftG4`
zLRBc}?t5IliCwjP{>9xxVRY)-`}J>1D4)LX`^S{yiT!28>9<ghLd!pQGBxz{o+j*)
z;$5`C-Ava?HnS4S+i_)#oJ5!Cyq>?Q#=s`ua(ETIZxIe-+yw;7H|V;=y`cygNF|<$
z?eLY=5x;CF*<{Y;oTMO)>)EFZ<EAh-YP?OVMUl|1_~?vU+_kBvoi;@XZpzIf=E*M_
z$;eb!m5-f(2-WJ?m}OjgkpafUk_P4Wsa^69Na8B?zSbS}Kj}2kVlfs*C2>3EGFU_*
zWDOeT_8ve|(wO8G%NE`(Leq}0sYFr2p(=k7Lp(^KhSVNXr--`~a3R8g!7#JhT0i%l
zTY6s*lanG!s%TC}>^B#>?B6Z^1z)WEHJ0Z-{uVDE#H|#<#djeHYJEF?&otO)YhI+x
zY63UEWS2t9`nCcE(WLYHR|D&q@|!p=D{p0=*4+^(PJzjbw6B5CDA1v6$n}cRj2ZY<
z3g=Xy?+sePHK6F`ssK*n@tJ-`1Al-_XKhy5lX>6D!Yc7>bM&_gL+&c|rp9R7&*u3X
zX>!=jGSzdNjH@~Px8I&xu63_A;$4To$o*y4I3lP*<*&S@Wl-Y{B*DP|m7xg<^QOi|
z{-6+SxgUlw*4SLMGHHLeZQ!pWw-tH-0|M-q)2TsNawiquT1;2gF4R>J-Uuszb$uFE
z0(wT`Tf`WJC|7=9how4$15rvSubXm|B{GRdd%U9~ZF#+X)_s$9UdaKB(HLj8>31W0
zH13b<FtlYjmz|15=SEFk9%HvS43$RAB`GHKNh`o@8%&!Gf5Ewum;Uc_ma@8$$@P_4
z_lPzcjcS!8m?n$0Qh3LoWHM9sBcf9{#g|=a2g`8dx2{EE7Z@Y6<D~w<%M-*6QIg^p
z4C%ULliUy=_09}rRhM(<U?V^V{;cM_P!0zoQ(5?kf8jIL7B<x7-qcO!`4A!H*$394
z=Hq&IF1)wPV|AODt0YUtmL77=`QG79{Q3;%=;?^s+BkCKzejXq*+p0D-0$j22!{1T
zH(SjnGTdI*kJC|;J$c@otW$!QopWEct4;e#8L!ISvo+78is_abgkGa-!W$s5cmp4|
z&sb;dhV6~iL_~U1k{L9)w(y7%c8LAQ#>U8R^%Qzf_nqJQ;CZG0cxXfd=h4>pBi<A{
z=Y$nvU2$%Gp&zG~wlD9RDW99>D>X^Qb$DNuX6?^rzY(FKcJ~X}7dfCvNXbK+-_#|e
zE)~T_rq<t1JH>r=$&KC-Y{2x^1DbgdKY79B`0tpQ%P&qBM?ehug#_GbB?-am=U6|O
zbi*|t=bH52JV`SEzl-gmAFT3^Q`A0qeZm@c?H~78Zv}TB-p7|F%AgkIQTp$o)qO-2
zguK~O0oak<gM`laBcJL>SH=YU91T*;SD?%ZNm95;JE~B~^mEL`jFM*k9_>0Fadxt8
zQNQ1i`JmPqeL6fkx&b$*G-|+m4C3UWaDaPwF*j9e(~Cp-Q_-T^o!h#bi)f4YD75Ei
z5@%A|v0F>%JJy#QA-CCIeC-ELK^$3E=55I6afFO|Bxw6j+SDwLZs1)NpAJb?_P!t7
z68y?V+GL=cut(5FK5T%(sWPkXE!6oirWrRo)DYP4YYpFF%!{CgzlHKRfP0oB9S;6i
z+hK)qd)I;f_+|%6gvRGQ)}>LGG0oy%Gs!6~Rcj$KRd@<#x?SH*XqEshpV$pI4jZS0
zmK&i55L5V*=++t)dl`(-L_7eIhlVgnmd5zKsy~{5nP=^Zz3Crm*%|$@gaW?LJ8i?^
z{X_Hw_ntoNDTy+{ZGj-pd6CthK;S)fz6#zG2jxa%NpuA^kKUzR_~LOQs`y03|18`6
zJ1AK_E_O&HBqTW4(0CxDNhAGb(~}6)^z@jf-mv45I)=Fg<}=NN!HjY_81WTE%^y-_
z7sc-Eq`C9kk3j4zE>e0+vsIRt%fL$ANR5qiV}wZxV^zBzuW-H{rp>pHY?$W`QWEJ-
z311X#5o+F)^-#HQ;%7mL|8bn1MDD>#PAV?65pzZ}U_0PyA|o>zu{)kRP6YvWV`NB=
z+`b@d>igS+{Uf#MI=b8Jw~oF_4)4^H)OPeZcGkQu_%LX7!^P6`y&r2($m{g@V>&~^
zzZ{pn6l+uhJBej`e5v9@q4s^>OrK6lb@QRFjtvb-w&sfalpO#<*Z5#vqfnx2Lko&?
z3u7dcI$@rkJ?1XGz{Q#VhYt7pD2-_se_;)c<nHF?W+-G(jS+HM%pkC1?Lltc`fVTi
zqR(cueM_<vUYmZCQtX9puKjM>C!#&IpDM2X-6*Gms}T+B49bA@U~XQFeGJIGHvwwj
z_GtR$d`W_zxe&Ki;A4ASih7QHRqL)47K4Y|9oU-lHK6)prN&Q(#Ns8e!D}O3<OP9C
zZ&<v2C}To%g4uFDL_giY^r3)vNl_Z+zsScwNBjKnp`yIeY4vGm&}<z2e(lgg+RWqb
zmF4>}$prBR{+#O&`L{+A^^*(c#xBrixnZ}{73A<<0{W4TK6TJD$uBBhxUoC(y;6DK
zkR&G|I_c1J`kApt#RM6}^$*87zj@l4AX`k%Od1QumgvYfbJD<KvcjK|3TWF9V_i?v
zP`{nKVGeiUL-BXnpkP_h;eP$3HHtqhq=DC~Fzn^pt!0@3ess1`VL~ps(aNv7$9G>i
zmyokgI(6=#?+DvIT|S<<>OUHgedISIiCq+nYF@B388<X@Hot-<Lhh5QR-=32Q=Umk
zEg(D83gvy?ziHxMIBhu{O{Ms!T#!yq7ARM_y)JTY@FOx&hLFpqY&42nrh2|FZ&3`O
z0{a9~HpbR(_(|OaizE3@-bq43mO_{}_m1#Jxb-ExwkOZ~=vK4Qf6P^ZeJPcta5r)v
zHi<ehNa1Ya%u!x&BOcXnpgo057CnY@fcys&`w4~MI|n~U>?*`jzh^FSrlDtyxfZN9
zckZ!8l>W`oXC1tdfd_kexn2+vKn?B{UDbi*QtI;Pj^Q}{B*=A^7Z3(#nESyvy9a{-
zaC|tUVRK`22~P0c?2r0R`NKhRZ_N8sq4<{@WoIu~oCcJpsR8fT0=2G&To#xcRE66+
z?ry-}&E)(PH;AVq>g=oOnEUY*oxPmq2KGGvhQo%Er+&>L#Qjb$Up{zuXLsX_;Q9ro
zH7C>;5@>!_%|?tsr3i78=>H}6L-_M=nX4B3C7qem0v;kvBs;jRH}K=$VqMZH$OR+J
zH<1(rq4#_<Bi==w43b{d6b#=|Uw-U=D~#UXWQg12;fH&|84Gz>G2aqQ$xP?2M~fhx
z?X@G(nt0Q~fsS)jr8hVm#P9gxcKXyn+SB#bI&8+3{5o=wl5vr~)od{B6khL#xQDP;
zC8h*+)?y2KJv4~v{NYJ1FRrPR6Y>oe1Qdt)Bx7%T27p^BIQ7%1I{%_#Uq08E%D$N1
zHuaM(V^ixQZL6frOG>o`pPUxr6=gK3<?hP7=aVS6GjZaNvu|1aZjq^bTSy=7ruQV+
ztT`TGliJabwOf6jJUv<p$E^(rX#VxT08F4_*k9nLl_o{Hxsd<^j=EuCt8To#luy7R
z@&R_Eqb5R!{3EXBVs08BO0iP&-LauTmRFc}gL_dr3wm&H@Wg&C3)$xBdDfkj4>XOs
z>Ynji<l-7&$}brxt(qysDD@u!f}98~1uAHc&><+HCaa`fXh!^D<rKCb6K5*$v!(O_
zZ<ZFIkHkt&cY5W0<}OS~;aToXg$h1r8bOxvC?07lSk)6NYoH??Qn4=?KCNh4`&Lh}
zU~%&C86}VQoAp8+3IC3fBJ0OK-m#eL>Y?A|<Eqv33e_tNWU=u6RfPiAN72aEe%8eX
z?cFia6N=Ex9}0q>a!dG5g7w%5^p7>{R%oj$5EUsTXtBUdf)igP9(sGsgms2<=#GSF
zzyhZ+z8d`8^SuiR)hTbFcmS+u375Y1WOI6(cKQ@C*+p6q{7%_FeBP6Wgcl0L-HCN+
zXN@%#@4iD^xlohJs$R-LB(>i9N||6DjK^M2ftkRLf4RTXWG$44t*R#dm*(#PWilMY
z1YR2c0FmK=Q!{n>ku)pfW7Vfz<bm!t-J}N~OYAE7S`d-g5bN01`lutaoAdJMmc>Zn
z*M5L*3QSx$1pe34=rhBS_wd`3V=1E*(>vHSG^x{wh!#8f{SbG`?;bu}WlnDi^fi!9
zj20za9eGRps^4?cEP@R6hxSa_*NzwAhjzucm2<pJJ&^9IM@(&%l6Z(#dsSpDdg}+J
zA3ha5nDC~oLl+u|q_A*be8I6R3`@PXz<=W?q~(nlk*ry)7vPE=kQ%i+w(~<Lt2Dwj
zl%owPJw>==dWrO23HyF30Sl=pKgbdG5F?o3ys&VN_9klm?19V#$u#K(vNdKL=rvbw
zPlV*g9uGui8TalCbwtVMZxzBaT*ZB$rC?07_?9LAr8>};<ltq#-R-ovj*P0|TfFcZ
z^z`18NNzU{z90%qM1U>P{kEWZhrb}rlllqa)7Gva#gx=J?`S)m{{0s~-MGDYj=y3r
zFsEgOPv!9b4aMnj+Si6_xJw%Y6Zt*)Gj6!!l%s9=TP-Kd3uJ~x8XkMcsg#G0=-+oX
zDEB5iZ-zaYB^$yNbKdNH7;~~Ee*6SzM}LBwWys%cEjAc&0<W*eCIX(4Jz<PsdDavg
zN05!O&@NQFo~&tycTtj}OK%Edb^%ZjFMi#j+)t7hB|ns|!4-FpHp$gcUpPea|B&^K
zE#;>=nv?NOjdj`;WI7GMSfg495|>d)(lZ_H$3&Z7ZeJsmzI<QgTa%x^?ZIpSz;ap*
zH|ChUX*GKC60gFcJZ|5Y`@XwkYhqCI`y&4fKT_MG9rO>!3x0f0`+ip;FUqmR8)1)T
zB3RN<nQRt{iPL<>hfwFJ_u$RsFYqot>*yr}s}@rm@jD8-M5d1%Lytlk7QxwiR2WRT
zcb92?+7O?_Y0LcGle7GWf5`|tWVk!?E?=|1wGzN;8{-oTdSvNSUHCOQBf*Oz@LcSg
zeJT&=>j>GKx_sEg)uCroRUHDQ2~Q;F-W!CS4O$p$0Y9dnFHZDed4#6+b9p#l3MJb;
z8fJcR<>OttdhaE7O)lR37PV5{3cRWqs8w<hH`C<?r=EJeI7t?A^+Swr=(;h|OF%=;
zNeQGY5Uq40o})jsB*28-Gr_oIXmgpezI3sWC&76kD*9J-ihv33fA_)vpcB^L+68j}
zsv)^yTsFKbIRx{<{>UiC@<#lIak)(7l+nCWS_3mOO)g%K^kkWoEMQMxlT=p#$-!^T
zApg4LI-M4FVn^c!RT?6}+?Y01ZMZaBsLy+9+EQFe(U){M`dwi@F;e%xTrB#zkP)Al
zKIz7N+4*uBUL41uG6OpEE=&T*ZG^#~<BOk$pOF<Mii}LsL>6GBCRyMwZ6f)iUA!}l
zYldERSaQSWq|j)9iASY8HaSqX3a5hjF(cNat-mjNwaAh;bh%wf=hhM9^LaQO?J(B4
ze|s07mH32=-s4Rdjt~4}`Pp$cAoCHD{TXd)vEW?-h7?RyjZraFit24Nc7Mdzc``S`
zM;UbZF}{KiP7<v$G<{SG4K%JQ>hiRwCG5*yc-_S5zXV{mEpA%e;$6uf1C>u!!<qE5
zpwGHEux+LyI6Q_Dw%5AD%sb~t(IC2?oKz->&&(SclxS=XB=wAWBxxckpJ-M<2iG-G
zDmH;a$SmDVt!|CJB%kmdACWH@#k*1H1j!U8`G!@;%%_#l3k%+HhusoyQAR8{Dqw^Z
zz$esv5~?`d+To>}YUGTQt7;4@NR1K+HwM0=y=05+i!{Ed{D#3M*pBcujm?CW@z==G
zjK@B%hD9_WS=?Eg66d;Ndv~KWUi#!y!4~bQZNn)i?Gyq<Er2@0tKb*lNpQtprRFnB
z-+23j!P2P+xm#ASdBo*6)Gj|7J^0huxpzdH(QeJ}ce(p6HfSPL3WU>8Zb<0S2d(Z{
z<xdDPw^m<jcEobb9UkG_Pqbn_$`XOMwFV>yC-gi(=Q$S>K5M>3BEjT{KKy)JKIQdd
z(XOm~JTMb@(L+O!tLJ!2V_B(pl!zNTEuPL^C%ZGd_x1}d8S1?=3Ate&q;Fm0jpjMz
zU@R~0=!Plw0?P2ZFo{(`t65>JZLsQG5<CLIx%3sXAoNr8x@wEhz8FvaoO+NMd&KK*
z_SY>{w&v&9b5w~r7YXX9i3e)+jr}o=^En+6rUG}=0Oq2`m{gj(Hs!9gA~_QOLIC~M
z2ZSyjYe*!Pffw-y!8m@KaykU|4S#=sm~y&KlkzV#LVlZbO~kRL8uG-j-io0Yk}g1*
zY*V9e!>3{|t>mMFCGFhlTWKMiJ=O;`+_S%p0<5r7R-WGNe8G6$fxj5>`|=Te5Y5A$
zA4X6QbFZvoRS{OeZL4EUvo07uHaHb>=30?v^Eu`n>f*w}^TF>$1AD7nzlUWRh(#Ke
zq8F&5ZbYAtLBKHMC+4hANC!o}BSlG?Sz3^(;ed4IH26&)<(J&B2S}PMed*>N@cIUV
zZ(hJ2(N&53=kc?ao|cy<B9XQ6GoRbhvu`vi4uI-ON!!}(8*1SGj({n@*Qe#3=X$(h
zUzgjnZK1O+2!d?xtKiJIoejkV&?D>QLrY%I{R;*<w4622IO+$<RV>G9@@Gs3`@$HR
zuvpH?4Q5?cH;H9q%$>Nq%)~x9$w!RFPI`Xk#lA!ug~Q0*JZ6c9TJcZjk`}puywTGa
zvW@M5gm6F=aqeD3+>(iky-3b`qI5}PK7O!U8s4&UKgP)6Ns%W>d~?i|5bA{YzIP1s
z=IQ|KG`U?cQ%1NIeEfjSLOLY<V)61TtgOn>v3uj~OU(aA(^m$?)kVwV&fpHgEw}_2
z+}%lV_u%gC?(R;4ySqzZ2oN9y8Qk6d&3E6uwX5dm)TuLj_U>M*yBBrqtKnW}M<{Gz
z`Xk?Yrgqnh-0e#Q0GV*h8>YeI3-<C}HRa&}SU`W%DPfdz>x7REWNikH9P%BFF8Ki5
z4szP<5e?YTXVu_eOT>K<`o}7~weF`ekdleQk1fjt9bbN`R*Xg}MDae6$YFh~Q8-9;
z&$U7hf6-;BJ&An=u46KS|0AH<my#LzKrY`(RA=9O3XX>-xb3fhqHV`^;F1CNRy!QI
zs`3iMBMbkCyVJ5Df599&V|z4#(XIR!EL?vq{-jEGJZ%zOF3&L#(DI2)sU-?sF^73j
zJt*<Vc10!PTDv^L_W!w{2WZ)$gGVpFT?(=Pd|~K|DM+`U27+m&5t1dQv{oEOV+%;e
zSRrL<(Gs`@xD}RQQBgY9ksQait^z`H9i**lutgC21!-Nnl^OfdGpY6C-4i4u03y51
zYi4qCd?B6P)DQcIF*gl$P|Zgdy2iO;04TIqUgZXy5fadOeHr5PphD4^r`w{yAtIO7
z6N$4RzzChWNIEc4RCkGkHEx<w%HwNn=;(Mfo$yD06@29JCppj*#^npT5yIBHqNj+9
zKUi~`*i#msxKOK^oeaxih>3nF^19W^w#JjH5iyCSB!RqkA?U+Gw85K11_OU&;Wd~@
z242}h<_LoFYMo0%S59gEPyO_2J=n`|{aF2tx+D1s4Gig%Qd`U;`Qa}^l&`a&#2z8i
zh(~D7f!*rdFSeTXV`F0izaQ#ljF8D~jWFo0nVtEcqZLuC=I?^-TrP0hp$(rsX?rlP
zIO|hW0T<K<#U{%|@dU<tVK<l?%1=q6+EFFtf(r!^Zl<+lgEcPKieO=aDDZV~FoVss
zzz_ZIHDLD<lFGR6{d$LoP$)NsnfITACVN4E{e%@geh}+OXEzF_aF}bHcPEcB|HVnv
zW-Rx7{aygw)N|+lF2BI))jZG9$u(&|x#5PI#i>qR@3n6M@+YqH%^jahhN1qE<|}-l
z^nn3PQ`&x=LAP+i4dhV2cEAa6{Kl8~&l!ZbN2gO~@GM@2Vbe1=P@Lq*u#2k$7?{2(
z@q648ww)ck0(Q{FhTLMP@E?^)If;>0o=`y)fe?pvDt)2o-{n7mv}XfBnSlCphcC-3
zl-1e<|IlrK5oWU&Ht`IVvvqgVnF`lQs~dM(tAc@tbepdMBB*o*q0RmlYl+H5`%|`K
z@U;~pJEV~aStiBz)Kj0|DjO5OQEG|Z<{f{<Vfa@SYqNJ&>R9G>2-~5P_KMK9<YC?K
zQ}gjQy%Cd^;QYV60x)J)XszEBa}3@>Bh9F1BVHjMfbG<jV@p{^o2VsCE^IE=Uz0mX
z+$TNgmbRV+*$J1^Lb6k}PjN)cA#<Cly){V;0WiuEk@zV18L@YlJxD$1Bn<6&Qgo`8
z5d0FIxvLP#JS(~AsH$aMhAfy7604~1vp!kWsz^9ZQGj+D4c`|@n|A_0`TgG2#ovf>
zY2R{3(<0O}IgZ(n<;*!D0{_o@3(=K^R5&US$v(}(8==Jh9BfS<Rxk1D{%GJheE)Jt
zqC%yN@OspvB)ao-h61$0yeac2iIGa0;p2fbto!$OA3?$)+bPL*+2&E*%X<u;9;fyr
z^1X}5KCQgp2xat|9Jwqgi1|yD?}R5ikq1*u<^%O|G~Q(EVJPRPBdKyGtyx*UhXVR>
z;;qc#hcJS01m}_#7p_%L>Hw=ms#h|>n_fo5Ln<6m5p5Vl7m>TLP-Yp5g*C~r!A=o^
zDXr^YzT(<I60A0`RPn!9hbrpDPj<t5_a!y!DRQ%@obU9naUSp!BW-!)PfDnfaXl9$
z8(;aoD~Ptg&IMhfgjr;N`(7VsfH!0m2XXLxYR3S{dQ!f+>h~v57R-ji%@yuFoqvAR
zbNZx`Kv4ooFc|Sp1Foi)mPIGASe5MhFoCYNM<;LIn81`~avBgAs+p$hIB<_}>4&`K
zo9_wqj;Of7gw{_LR)h4FV58jRJGA*KoBS#mX6|LYV1};kvXzu^>&=xV^BAlZ`qSST
zm2q1OMY}IVU;4uy+5*WF_dmQKNoGuj>WY~zFeSed<ujr-UxK=dW%9cau&&cP(RjvB
zVD7O#su_M#Nk^#HggpKV&VAcF3v`^uaH`xvXF;g>O3Hs3eln{mcH_&SdiPzU@7`Zk
zM<~OlyAPXwG=Ie=b@wr$z&gjraSmQa{^|!+D8`lpeT{*C^=AG1J1)qc5C4A^KBkyS
z(B#=+nf2(SKHF!iEsy^q&`C6@PTeCP1Xx0=IUgyN&FfMi6;B%30qFowPWr);lnkL#
z6q63Ro{-?i+_@YhAWO<1-qTsDe}y~1=X>ah<d@xKG5_qKSK&Hmki2gXjaik&pycBE
z53@22kP8~!jFNl)Bm?<@az>gv;-CjNH@6=n(VkBy+QG@Z3J~SD*^qg<=R^FZOFw=&
z6A{}4!7^J7NR#>%A6Hyua4G*6apL7phx9stf^+MA+z&d{{%i!uPr&(swDD%3fK%q^
z7?Rm1gzhiERHu^kB_lPOR;>e@-rhF2kWRoU6_WJQkada5+<!d-spZB(8dAtK@lbRz
zvtvTL(71=XCaFyrgN_k}aVL9%(GxE8Ah$Z0zYebz8=~DKRZ)r+lW?o)mzmo+HzL{_
zCm&m8FyBQC1Yp2)H-G}~gF)chs`5zPsV?6pYSNb8<vijDnP~RE-?KGdmLYkVS`N{j
zJC_^ofP%m0qOoh6wm+P$t8H1NaKEm-n4NPA`Y6adC9J<6>Nb)^?ABo?GPryYzO-)x
zx0HKcUR_R>YJ(3)uVnd=!Nk-Vt_}mh9>879ykkcyE!XC#vv(k>jw7l_)`5vyo6E~v
zt}*3>Ye@sw#Fh)fJqT*E-t0w*q5D+`<n3MZ<W-Otdh(Fyq-@!SBP+Zu9Ct%6UseaB
z(M_Y%PNF_1oooIro?n*j!ueX57vKykbz~j6+0}XyB>M1nCm-1c4W`i~A;1jx{YXgn
zS6M${LhEJNO#Ez7RKaDHtNWqwxWtulovs(5LgIc<ocjvu^UZg0bUEj6(2z9#$m|!~
z1%Kfc`uS~(O5w#E`$uT#x82UM<IYj*^+_mF2ZzA|g$*`Q2=CuRf2;Kze+ax}`w<%B
ze$i|a^*`=73INP-vSn!MY$QRAOapVApVH~~NP;2~O5;Nd3^v*N4ignrr7c;CSY*+k
zTViO!f=Aq92r)*}Df@8|Yt^JWvCi=?wb}$z*ne^GaY7%+#VHFtQC1_*!opCMRG|M2
zA-tGxZ*eM$P8e&~@uP<36%J`-)EBK|lqT0*f5Y4&*s^HQMF5U+1|3}dPU85c900l*
zkR9V(8t#X?nAI$Zxcn#v)kAU_F)MD6pz{*H^Raz|@NpJze>VDy)a|dgZSJC(_St17
z+9gzxwLd1Hd+d8&_~o5y!g|EP-^2sm(7y%0FtdQJmLj*6k>({x*d0?^G7W4ykx~#S
zbw(eDm44@d8=zA1RTF_tvvI>km&X46a|W;qA@c*O5SzXrmy{ZcwlJ7)+$He|&?9g_
z|3Dz}+%Ir+@uXQwZYC?jOPV2)ANlf*-QzjwjjYv|;#|pMU9D&{eW6(+y<d%E!DK=K
zIf3IEaQ`77j$xi|MwaJz7C8n|7H$~i-}~B7@01>bJjEH*NV~}N9t>AQ-tbm75dvde
zg;gZ8Sowpo_`bqx4yiI5)s8OtXXeJd9n2jRj=KvLc%L7(Udpy3gP(=44!|Hj0@E++
z-}+KqX)f05ho=P|?7W_$1S&~anWkR3z!ra3(oN#KQ9!BnQ(j8iq@f&09T;XDDVLbY
z<_V(p|Gp^|!XNv%Ey{Itz0qdS4C-#)c*N$AFe!zZ9T4C?^OE=c1qDwcIi=u*yH;Wt
zxGuQ&?ZY(cIw9X$z7S1|=;~13#F14EhWmH5`G8>@$xfZvnTJ;1U6mRhRoDyK23VPi
zxWI!}feCo2xx*>!vj1D2H=v|oQ>hCXBJ@iRbkyRk6kst#urPf_RPKdd6G6qgFV_Ps
z$-q2G4;<VW#PW2L<NQhz%eFHP3dU7i!eye_t{CJDQ=HrF$>p?=CvSPva*)6Zz}f0M
z?~80Nv8Kge+FfNmw^fD0waCqLBb*Q*d1$SD6cUO-NN^=48d=Jh&G5aIh5W=jaOR3C
zeP4xO4G}`ttq2KodGL%GLm#HiVKB!JIVGVEjN#~N0wQl35f1i9wa~((u;k8ZpBCFy
z;3MT9)2&5>OHey;E{!s1Ut#HTBpb66;JTzl>B7uSk+SPuP?P#1oN%6V^$}eiQhwQ6
zt+YXH>0CCgWeOH{?NpUy-N9b|g;v(0RLqosb2`g|q9yl6;G6!gDfIk`61Mj~KOQ|D
zgZ|7?fz|X6BD3+&W6(EyS{UWS*mC_w^1Er~Qe%tW!jnmZs#lO2sb2TsnU=gTiUj(D
zsi)h)mF|7-DU;P7j`{Wg6iY?=+{|FuNscm+;#0DwA9~|X*S51juRgoDE2A)Soz9?(
zY@{~gvi`TDu72$1O5^aMPQ24M?jJ7JfbR|F?ZYPcnPvGHb*Kt!RL6TP?cv-HNNzSq
zZ#6V6aAN4?4147pi2HU8Tsu+b|1|pX8|l&#FCJ}2NUG=o=FVml{#cw@!;2Q~#F&Ut
z`{1S9<~R9i#a!sO0Z-SNKSBkyon4hLCG#^SS+zpJZ_Dd^cvD(z+h<1vs=FBk?>ow7
z@n+S=aleH<u_uJES0T;Oh4lK{qWR~19>gKf7F|pD7{|j8aI+^`S75TD>fHZI#~@rt
zSsMDK472B}emudt3XQID%yq2&yHTlKb{+veY|Ya(HlYpsN3v)IQjKnF6@I8mX7PK{
zwn#wiwfoAw;^acO&)%C+UsD=w-N<@^t^O>V=hnl^g@m3(h|yYXN1N6s44)T~=XA8^
zcw9!xo5EthJ1$3sHKC(jTN^a+8;t!cfjr+I#?diN-1?2M%;;UUOSeYkiJMsc*tTGc
zNAR0nHWb`WziKBf&%jsGQK<Jl22ZX4P^M{hKEGiG-+^vgCddKIfpf+N6KQdIZvH~=
z<meCoOZeDnI|6Q-irs81Zo{kK3FmpSbb@_uvW4JHG40c+jhJwqpK4HBELG_1EdFpe
zD1bg_LE<rADAYG(8YS|mDIMzk;$VA~(G2=r@&|+Y#jrPicvOw+Vf%S6=ZSgbKoi_<
z?Vzy}^(t1K6N@YP_UQbcxmVW;9D=}dJ*+Gp2$@Csy3r*caVP%i5<o(t=992(s0P1r
z;w=10GF~E2H_<o{c?h15{HB5Ic*Hf85V|pnK1SQ6Hpvs-*+nQX@o|O^0|I+5ji!*m
zT`t88hQ)z_fb@<ou6oKIK)`1|!_Pz|ACr$dL~YmIna;${2U~^@oLT<IkE|auRA7%1
z*>OS{m;&k`3uN+O1H{gi=p~DZY=o?IfScZvJds|%1MicC>VpK~!xJxV<I>rOALxk?
zI)#MPfEQ`eFwp-aba}&NdXR0S=Z_V$U%7pPBx6Op#gnvIr&wpSyPI@kvs<F&+Qo*o
z=UfEmMm}r;{Os|^ikV}(z?i}R>PuzAf*9y0dTF_d!U*DwwxyO`VNsHf5GLqC$+p`D
zL4febL`Y$Z1#W4exWth^eR5Y5x<{aMPtw<gi&VW+GtFEsWz!Uh_`sy8f|3`d+MyXh
z3Ts{HF_lz*udGtm!@Z9u#o@VqYOh!`Nntz-L|e&^um9OzWU@TieX+?#*+XKu@mnRN
z@`O0}>hOlsi%oxYKIB9=B;b~lPF>PFjdc`Gk`k_le6Z7t^jjx+$v5gAly`0~<$+HJ
zoQM$rxf?-G2PgXrb=~@=*h(*57~ql|S(1t(!%|mQS97c2N$tQ3bBh*fS~6LYaIgCW
z8S2~R_<$K>8>JM&>n!VtkO^v?1|e$Yv?dJX$%&doP|3SM;GP(mHU_!42#|ba05Ngo
zB+u5KZrFjY8h#gQ-$-&GQya-|I`5B(3{OlMHx;`GF%)m*Bu~C4_ThX;tWyQko{kA5
zl1g}Ffi@dzC1daddpV6jwdJ3u#ZR%$#0h`RF{+!vD)W@dv2S^o8J_UoR!N?S&Mr!w
z<OjQ;RC<+JneuDQu&-~h%j#ZF6sJUs9CJSpSZ+Pl96liy-jK8K+b6UhL4UgL4#dk_
zOH@(7!Y^MbywoKn%EtrdG}3?#VEp?^Ju=YI(Z}9b;BGm)2kBv(*N>;Y$}qAbhUj)L
z$d<NUveLYc!GDB!$U3@qalDQXic|>mti`szx}GUni5(qZiN*#d&hI;v<zvb_>X45J
zwh1zLy6N6E`N3<Xf!YZX)fDv*_iC_KmrEs?Hm^{-$Fu~e@_4~p<9)RREGTVCGeZBj
zzfgj|do?Bvayi3-$x*@uuQBM7Eky4S6ZoL*P{D7r!v6_?zf(cQN{WoOZaXTwmoQ-0
zohH#Sc72;6onXP?-lj7KMhYY%x}Jz2gzP@Q1ucB`26++mWj0^z#iO-M$;fPnJfhra
zRBP1>d-?J5z6)s9pN|Kv_s^B6&40d~B(^gzdP4K+1pTI!HpXs>Ee}V(JO}$5IR(Jr
zgsCXzaQ}2P9p&=mhEcW7f)7tpGCunpxQY}jKH=oms4l&n5v0hC?n*d`MHKp?;^e1W
zhledD1M$>HU1Rq38O57b@{ey0c?dku3)%rWPDP!0U<WN&hXlZ$Kxb2neVsL&^JxnU
zzsg!*dor>L_wm6cB|L84bh%eu%7Kg_?}BfeUaev9N4D~xcs_EM%}m==!Px&E6s~Fd
zb<^R|{c_tJ?Np??x>_shT8|;(BSjqQAq-VDi2=iK2S4hanY^$!TA1|#>&ZHu^$u<~
z--PDX^wHAGnE)3T*KX_v6q$H&9asW4pnjg~V$eCug8JsiI0Tz#+trgf*Q;U-Q-{c@
z*-W_MQPLk{d(}*8<YXOY&!>iiKy4KIB@>;n!fkp|Y#OaGjwr4X;PnM5hHaD(;ZBTJ
z`Z*eddP#=4`~i%3-`034w5nZ?7Vs+TPS<h77U96Z$>&w-&VjZdyt-a(+)!+cw@ROL
zOGdgH3$1X1@kQ(2$H}PK5fVVskKw<aB<ME>g!>f&zf|lA0I^T`offGR=*h<?OJL1L
z8tE*#Swzw`Q(xVGKdV*`*!;YJO_)eEgtuKbA*P*xxqvZV&;&cf3vm`j`(tM#Yq?IJ
zNx0ADAno-=Kl&>UL;@e1WFKHcci1|ld7(sM&3zI5tr6$X8eT1BT#GL6#2(mwOo#od
z`JY4q*`gD;%aSi-ogPEey=Q_wY3mITMbO8QkDMH+n1krUHtksXbB$=n=cCpJQHZkN
z3km#Z!ViBrR!NpQ<J#QVtFw7@tQ)|I5$P7W^)j#-61*f*XVf2%?MZu{Vg5j^b%?6g
z;z=}{jz;La{zY8#scWToY_rj17=Z}t7g1!N2hS?QL!+8Q>D&fhRI1!(BI^nTR%oKH
z042DP##e-f-<f#$__L#(9mM@F{q*f!uPo>&XaNi@0R(SP9;!%6QD)r)0U}wTX(LYO
z)*e|x;4;inGcA%49IE%u<<{W1u^T6~S)#MttRxX-kY{s!Y2pMTBVi00P>qKSnGMA2
z_8WO3>x830FJg0zR4(ESy~86vo))=&fGbjPVct6r>iGjEZ@8Y{H0BF2<ot^`&`(S~
z%kfHp)SZ&1zxAZR7)B6OQ}kS*#_x9hR(8Ig)djF!C@d65bwl?@I~w;W<6@%=LaweN
zkq5+S-{d<D2aj8`>`-5vzWMkao?5tF{}Pk{lDa2aKJrpb{Jo?863|Bvq+7ZKF^Ese
zbh;VP8wYRM8$bD*v$#R@8SkuN-Xifc8n;qa12*W-WpP`#AtD4jsJ3|zJjbehuMi_r
z&_w3UhHYf}^YMw51n(6oL-g0;bsW1tr0S%>B}fITLtm{F{~h1~VQ5p6JRss+@rio_
zH4miVFS<ux#1w=?cF%^{jz?=xI3wrFu8lrkLR84R{(i7ZzTq4nrQEkVT=y5*oz!Gq
z6TNkPZxn{NRi@2A*=W7EZW8ip4J73%)QfM+u#kslsrbbx+;LJ>9Wxy>EleS=6}jPk
zA^w$@uV#6qs#jE}IP>LA_V!cPys*W;oy!Wgiev-c;cJ2(6GXlHEu6#Dby=&Ik)YqD
zj(<?`SI)LI+H4(q8V`#;=fcM7k&BOyn+IhVg;Y5n>TG<!ed#Bnr$Uovhzj(;yzlm#
zIR?kKUwj+WwPC&?eHB62xcS(S%f4jJ`j0cKtAOnVtz+d3ttZ-Y%9Q<UkE)?#4vVdb
zy|Y;)1p?!urw%yOI$8Xw+-i_H{fHXB&0-~6wBgLA&-mO{(_Lm9Ks6dd0_h6qf}*R4
zUs*-s171{+KC{VPr{6d{s$Cj-gi_w1ye^p!4TyVDGRNm7K=<T58bgJC&n&jT$;Y91
zsH$NnNxW$wZt8*!h&8b|Q)crlshdKuMQ^7BC2mR7%Uf47RGfxj6m}fjG%9WagH72B
zBF#A)Ss#{gz>$>az`Rr=3O|zPRSvtOE;-Q**kG8@IE85V_9WQN^uryT28j~_K<X^d
zY%BO%S!&{S#9(0r1_mVgM4iDTdQ`AWkp920>l9InOWW5rR!_B}P3FNqhC$bY;c#qU
zrYXu2zgFS^G{3+{WbG8EFXzb1I{}vz_eX;AP#0v&rTu{*ta{`F7q0ttE15)jvVLuA
zg+n5achYRRXixOUfd`t$>5*nWHIK|a)xCCizinpRnLtJginz_7`iMd}Bb57xt>-(n
z{>{L<s&CaB(xe6ikpG63T%dHu+_-|W=2MVdO`7^s6)Lkvnwwx9i+x(9vqK#lBO?&c
zjE@|Y&XI@Pc%sT&hu5_Gv~Wr8gh9nTB}qTn+~2Lr|Chg}YVe$5s5HqBZR4Nekn2=>
zmcUcLa?`=6Nnh6P;CG$|CO2c*iyJ!yqvGRZkqEj6q~&IL%wPtgd_;@98)iL!-;ai-
zhkJwK!tkM~!ei9L*Qc`ZDvrTxftihqjSK>e4Q)P|?=R%C%&^h<-9e5z4S%)%$cf|F
zU6YMze5@?F5$};$uaLZbVb=I@Z?xn~knr}m<u&kr7IwC-YW3q`O@J18vMgC#Mf?XP
zx;PB79D>wWzl=(RBPA&~aScr9LNRzesd9!Gj3m0g@rC}9fo35MB*rvuaV2pIQJ)Jg
zSI|>npw)f$T6U-BG0VYP_wqqAh}9~yb?GnUV|k-{I<QmuwHCHOzqyanA(>CpNU+sN
zr^msL`6%AZVn0Y*Y805XF*ET?K)f8m<?Uyv%qaF1%kB%x`A$fjsx|*FxwH%MN&AeG
zrSGC=*XVs455-3B+)=3t|GIDIuhH?|L_Q_@64+gvl9#gPt{*Uvdk~W<=ibbU|Gi>H
zEU$^za5NlgxvrmksO<CHtiVxZ)&~5yB%@&kh|CM)es3S~W2)r;kcm#na1ugUrI}kU
zDfb`|nw7LC6+c?dz}`D16PQP!#(gTD>_^Zudom>k*^;_5ye@5DevX5m4wiL-+8X-!
z%_FO}uSXb9_AT%(5kcU{DFWkhd2FFhP7G|`KoSCVp?q2&hOuyOG9vya3McZA^L4eS
zSV<zXg)K8=q+q%o34e}9bhy))@|7B6-@M33bRttDDqHC-)dhvbfO$6>R#)<Gur~mQ
z!6^{$;b$DjEVxQxl9#z971q*^&O41wU|VEEY<&qYOdY9lE&yOne~0kP^cRvM`Ree8
zh4ZDKa=RBye>nEuKo~NP&*^%%R^|QoHL8_Z@8Ccs*Dt7N2LswE)~p{?f~28(J4;4f
z6+%f*=bo#D`AW&QfGl(b&s-O7Y5jqATh-SY_ot_c!e=D8ze$4!0KHN;PXbyRMMi~&
z?$B=gSeNnp3Lv?3(0lMv))ir^5&0I1wE^f}XABjkeGUOIP+xxuT9;UKk*fUEA8EvC
zSUatQgn_7@UiszVaM<&iy>j$@;nxmN1md4d(;>nGsD>A#<Ia7;ko69u0l+SMK#_k-
zHmn4AljnFlc7<4>itSa%FGIYa$0wZ|cOgVcE^xwaq#oGs`=QMlz}QO*^95Jr8kV2z
zP(Nbwz@z%8(|aAbK`l4l`w=wm*^@K9{j#BPbG&ccl@~HJ%RB|_C~sO@w^%Tn)J(58
zHO%M28O_><wF04f@chUcyu8Ow>t~H0Ut_hcye%!gF^m<!`-tj{CUbY+sE)pFf`2R2
z@2sHB8@hIQt8_R_UJ0C*1-XxPC2#?rv*VTl*OAR$5C#&S{)?)wx5uj|ubRz>?qWyd
z`b_#!lFjhmO;G1SbrUi0=_L(<TOA6YX99f9qU-JByfa743`NN9GY|=*IWxk*yhU-X
zVX3?43LdrIHJfCfb|8trZAjD5kQH3x6cp?ph_D*l(CW&ADQZD`!EO#?kIjELg4DWP
z%CR}R7B$Js2bn7%S}e`McQ1H;>_?qEeBJT3>Wvs+J7hWNLg7M6$Z0-N%cNPuhc{Bb
z=)~>rjo{w(q)%m@7JQ&a$kO+dr9TZm{3BBQ??(jMj7PDA-V@WITH#1>*E^m-@du#j
zSy);g5oSh`b=$2K!_R1h9BI?Xg>x=7=?Y6CCJj;qhk>WSv8xqy7NswWec-pZXaA5F
zS(CZh6a>NiB4XBNnZWrZrG{p|kK|R!qcP;iqT3t&2hS|Mek3Vj_?_kqS;R_RrIjLO
zARvvOxLIDC%*rNB!(~Xqvu^cwhz=y165prnht*5UtiN{rhq>i|-i)ui3#eAN$b9P*
zWli0w0TTu=x9OL5af|fhkObM%>~@BEHxbtPJ-$$N?T=XUEAMIp!K;8+L+i~j^c(7*
zEZ!$T&I4I=H}Cz&b^jq;#>bc=B`ji#nB~7Vdy#M>tqMtRK>uz5Vr7_)RR<r%-cfqi
zdR8cVSA^@*L^8EVll5<^NnB=g@W2nX(gg4yBaTp|wDc8L_l?g+L=kRtRVC=cpM+-z
zGEUL)LRLUWY(irx{intSQpvwQ8{>^y*(r`hpgw;}Q(RgP%lSp%t?~EK-q61nICFU1
zk=fxh0~E5Nkba*Xpl^FOs$)NVYZhYSHPCU6An+_KjjQ6`TkP!qTxfNyoxg8TO2PKC
z8WHEnV|!`NhcuTsJJmUJXSswKd{<r!9J|{#Bf9@aGnJ=fn9u8c)=%CR?O9K=7Gu@Z
zG$p_6$~#ABAJsh@!Oz=Su<lol@p_YsO}dNQ!>I9j4(=rTP}c*uQV0E^=g|p;G$HL$
zh+gL&NZ)u=Sf_Q*@g>=Yc0rW44qcx#b{rOci(74#U}OwGf5L<GK%c|eu1A9VlR}Z7
zyQ6Y?{gOUUZJ@W#+1(e0&*uVXRsP)OjQ37ya{*YJ1_9#{)eZs7=UxwYohKe=k1fW)
z{}!eXV6HM=k;ppo)20U%LvaE=eY**FUq`_0;crj_yD*ODMo=^=%3Y&|QgFQ)!5>pt
zcUe?BHPw`ff4uGRQn9cq!cXQYshP7T`u&){UI+(={noIP!imDFFHhNMoJH9E9xrz?
zn01<fzhte3AY(hE<}Mj;47L7+=YNQ|0*3t>x3~`qsebtAvXuakfQVwD*2w#FdQSBt
zxZlVTgw(<sU&*UHb6gV!&=$25JWiYMtJ>TAmB&I;!CM+5=ik*27pn(5s#h~Y<Nu%_
zPHi2~{84zU;V>-<+ba&c>(-oKU6vYwW;Ajh8g5SF-blJ*^#EFCUFlV)lCpoS=r{cy
z>R!TTmDp+wgt&h|`NICs1<IC=sh%mY6EJ0EY)>7>y~g|%#AcK#r0i#NIJggv6%#aV
zb*}B4af1G#_%Xngi-CD%MMN%HqPi5GGtXCEb(-Gla5vhQn5O{8wEA}v;e5-iq_9X(
zc3+YC6E@`#VldNm4@a#6c>*xvgxAUyd+8|8x@Lv~KgS6Q_#1iAF&+}vOVAUMtd}dy
z>{bB#gXHn%OY3mToop|dpk=sGU2~`jMLWM<yLi&%?C%S@lW8<bs=Iq;@Yl>a3IwWG
z`&92&*L9%K#~$#!9YL^<%p&(xqR`hi!!9d^DjCN&Vl7k-S=lPLyZRNv8FZH|_bkQQ
zdDM_+XS`%rsBkBh?ZP@wti^cX^DIVg(%S`>MF4si&q2d9yjYvT{O?40Zeev;$E~D+
zxGE$5&F9VIF6kPlWc6(>4ubo)H`@lhrAZnxzh?9`+OvP;LZl9Kbqy!-x`?z4^qNZ+
z^xhBD)pgDvx}xVj3Bc2~#ZLA0?v#(gv_Ax+C4D1*$<gc2Eu5|Rb1kY)RBmX$f=3sh
z(n|IJw40E}f0R58yY?z2WRx@`8LNqP>YMGoriMmTPx1$wE}i+G_X}7}Y1!8IeHYC>
zSsOq$dN%5^NY0zJ{FCqKPt?<4x3if0PJcAwcpUifuyJfx!PC13;aGd3b-)2pJye=5
zu3l%6?30k2)Q@1?(#BY2t|-{VI1*lmksj}xoH%Hg&^Voe_u|Q3O+Mq8@1!4e?gwRS
zPN01oV3G)+sHGuhv`XOzD3H4xFlrcor6N5^PItyCCoc~&p<w_vl$2~AtQ;6q!M98!
zx}J@I(B^e<|Aw15;QC-*REN|0AI9cgUMY2!0+9Z3(`UHI2TDwLuRODs#%=0`A3(XF
z1hNVe=)E~%q{$YFiTZO?A_y9K-44Y<-@R}dSB9^MI&fpO^X$V1<l#3BLL=WCX<1Fs
zi7=W7St<`b8}MAWdxz!1QcK6JBsxzn_&ta^d*Tt=e!{(P{MJE~`$^yq7L`MS{oCbh
zvw+9xBa`cI6*m12CxvGk9`0pQLVh=ET8kgOH!4H>-M`9&YG&O#c|QC}!RFbast%8+
zl!JsG0-Ip;zR?l?#+vE@(jP^B{)tCfHfRbqjfEVBd`RVIj$ocx;|_V!vMdDD3bO?6
znw6N|UvbTfNv%`^!NK4kO0EhV>arq=^d?vS<E58Pr2X*2&^(0sW26rlUcOmH@GCOL
zT_pXjP2<q5!|`bi){(8=`&xgQsQ7`SrJ+@$+&G9m-y0vhT&wG%fEJEUp3`4OP-_Pw
zlV^V(A_%qL=M{izL@E!Ev#pO6AKxDq-(W|}*$};pf?9^wCkIs=NbtjYS9m;COi`&@
zBtmW(Jy@nR24}}2i~~x{LlGl)Xw>n|cEm~z$ww6)Q_B}08PDHTNd+=ql`Bcl`1r}!
z*eJ*Afdw~0OS*vrb5|U%7-|!tBi>N_jwX*@7$^ds1OL7g;>Ps+EbAqeYF{YcaeJl5
z6pm_m2G9~|9IPR!j$Lm$*WGUQ4A?;Tm``8m$*xW@wV0ccCye$!&An(}z`g`jr$3VU
zoEi9h2QI(8Ax>#IrVhGiu`?QdDV}^1oFCN!Vx9E4XR6rVujfd%8UBA%BkmgX)K!Z7
z>m^*8MrKb%bl|>cQV&Oukzp4$Pm<~i3LYM3_-_ZSn4Itw$el<%X^tueZW^B%kqQJ2
zhZ-0X<%>qC?w}p{o!f~0V#rd?nZmjsBrSXW)IhmHEX-i^vleaN7W-Lp%RDY~kZXme
zB|l)lrd%P@2GO$&8onS=e85c+f+`5xpBo1WQMa~(c4hd{Z7am8Ngqo6hEImWExj6o
zd5sWqKL+8Ncw2eR)L;2n%u`_(CntA#*LR5+O!CubIPDC#cA#RT)*)kIKEyW<A5s@X
z9|o`+zbodNG4Dpes<Z2a>QtYO`tFx-0k1&U`e=+$U%!2O%QF3<Ou!5lK}y3qk%&2!
z%Qpj33yKOrW`1d@Hv!V>g$H`^1^-;=A(10iiOKCieB?8pT?LMh3>AWYKlaI~SFZ!o
zo>denH1wl=ap>qzT6S$I+er_V?nBvA7?HRi0`_-PRWeAYU`UNOJm{vhRCaEckB%e`
zz!`ZANsT)#NfrRbFpy28ngW=Q2-J=7w5`>oFHO_5P&O(a25dO5F)Ur~|Le+Xw%|)i
zjmaYuT#{Hidb7Ri>MPV`BmZqlAnffzSr`qc!Z~1ZZ8st2{PlRSsDyUoRPLNbFWN<;
ztK(Dq;H*+62E8(^I+y!e({S9T#Ta|bZ-H@SxVfK$j1Kg20sg9G3j>u`VZ+g9G>WKc
z=s_V%=Ljlhk@F6=q~?AJyhzIli=KHr<%5~J!@@Lvymtl~ytVu9O2bVjiR55+_j;!a
zrP3I40v`RhS{&YQ_rB5kj-!$$ygh`f6*!I9(W@2=3I3jcT9>8a9k(0?T?Sl=-1v&f
zG(rr~YW5Odw;NTIQW0^Gl;$;=2>C%~R1e{FYE9=-s%}Kt0Tbr02p)~d`hVCkIBvhY
z{0E`0Nmn4LV|SR1bC@OYILPEf0BqUX@yCziFq%AuFBGS;h>t|Uvqxk7Xuf`ztRwdb
zkACcrPcB5yS@pQX4&OYo5+XB1M74_Ys7AL+*>wsZQY*txohXzI(n5yaQppPEIzu@o
zK$pB33Ry_yHouqh&>dSsd6Qt3#)UXW6Boo2_*9kp7L+#nur-Cpwkrf}_p)JI<W`xV
ztQ1=rpJ9h~1t%VB(**#stcIxlS~*k^t}wToK;&wz^OzQ(6?lOD(6x^HqD7NJiZ5_2
zi!xw0w3wHh6S~yv(KG9~UEmd`etT(EvGlHBf~)rtDQluVMSO}?Vzu+JFq_>Zq&`T^
zZ?`R;%0NI?@0}OI%yBD(s}0DR`@<((j78W*05}tc2{pB~Z-M%`x>!0fO<{aDZCr;M
z0SX?%Y(H+eICfwHGh!P5mTf`#6~JB4B7Q48XENb%?B-TlECGBzhSuW<?0nX!_X$|o
zX`!&Wj7~03ADxc5eq^$pJ%5|DTbO$|@6zjC$0N3GExFyz%-h|q9+kzHF&{9)zo7YQ
zi%>w|nDu(YVyL7B_CftCx|s&&&Y*=Ze^u%1^AGt{dM8cVd+Ogm=HL@cL-0K@lTLZr
znRxSqLyXa^1iQP-9A8~}KC(4SBCRGhe4V1R0A@kh{mt`+s|xY6KvJ$8Re5>5*2XLK
zxO@4QV9yQ^kGmW}qus6g<o^1(y3fLs7WY<TWrzil8FAhTEIhQ&l;`oP{cDZeIfhbp
z>U<UJ6cxT$0|iA_5QI!|DG(QT!JfTrYKzs9%-QBm@Xhv96yi>`dz{!iQ&pID4-29?
zumR%E?9IccfS)n`#XK4zN|8bQ#qM0K#g+?Q{e`-tTf`2jVpwLmWj^504Jvq5e#Ul)
z)#<CqHQ*JnUxnf1vx*ZYMT}V$&n}?Y>Qu)Oe7pdsSpJGH4hptmPaDd6ryl&nVDi?p
z@x3r^+HL(_!byCG<9_Dpu6xymuAUULzcLpa$1haISw@S_2B}fg6>wR<E_E3BD;J|7
zH7jqXTM@XlIvyPQ)L(VkXBm>Phg$@}Czto5O?7!LL5}Rg3_*dWz;o!}&DGAmAmt>E
zsn4NGa{T^^=f<Z6dT@5<`P)#N@@Ap!9P0XudAlt_zW0K91_oqym}*XwL200l*#{k@
zPqZal<GER|PV&&x;<rGGzvC3YP?E(1IetQ-wlF3Ps)DZnSU*w_xeiwl9rx_IT&eHL
z^&M#ng&m;FYj<1>=VdqwqUaRV+J)~1AWj<wF1;hx`8FSHs-1V{GshTX%?b=SuT?G;
zXxv@y$NE^bE*v0Loq&_+j#z>ZmS|LeIgeGu;y#!zkhQ;eoa|rZ^J4X)BM<Euni_yT
z3<86BE=|WbD_>4>u;<o}fa(J!sc!e&{zoI*eIZ{<ohl8#SE*%PQT*P_XYCRGb%R*Q
zc%gQFMKd?xq9=PW&JatxPTR-y4`aNu7sM7w7wouXypX<0oG?F6)wi%b$$jY5oDm4k
z$<Ke@>B^(;zA&?q#yBdf0<CQw+d|qvIsuJXhA9ZY^T1}A^yUYcLYBzL8JXdZy4|0u
zj&v626s?)4|D=yjp2+RDK(#NG)!GRzomp@x!X(WS3G55Y(aLJ-wE0Ki#6uQJ0Mm*;
zz1zQ-Cp%<=Y$*~hJ~PTf$MC9Pu<$uN3XJ4ELz}=VsLNtz{G4p8F={)~SG9+`6&PDG
zyS13zAM%~OG7!K|A3^RKYk~uiHF4mutVI6!40B47Gr|aDyL++5H~m1`3g+BGJ8lnq
z#J`5@z?eLT7?yEB-9D$-(r0V{DJB1*7Pm-};4S78%WgWXKZ8Bg7&$@J%*o&81L{Xt
z&W9N$3oItI4G`F<VXDIW+2KF5SYJ`GPZUE^LACW!t({`f(o<JDfj=|??JrW>TFWV`
zw9@DAJK1W&fa9?aF~=E!2l<90+kY<DY`XjBhi`8M?SqL49C79Rc5brzUnAGb5$60E
zZbi`7`{Ixu(ANHxk43W2p<c8G%gJkp@a=nText&`vZE9y)eiKipLFQ)z+wnw5Vvl}
z`gRbK)4RcB3p(@mA;D=OwjT%;AmCtn^+v@yS<87m&B}d$cLVgMtJd<HJ)3(sR2wvd
zW^SF)Qz_{)tU05i9VfCteFS&0oX(Zm^I6@F1U3?O<E}`vaEAK}jixhrJv~_mJBVb*
zKlI>0wbj-f;^T(1CtMB7p8Ofy{#EE2j!(?YJ_YT4U^@RA*@%prv;Ow);AF0LeczoV
zS#{&&Hpp&a;0HL_&vzrigF@Neg5D+z@Fg@S4D6ut=}k7c-Tb2_*WLeCB#+M4vm-n;
zTz|3nF4m$gx|7*Dx0jgSZ~++!WfdFFrx3X(IskoqN8MpT14R;~!&>=DiNL~qI`nvh
z^$seO3&PX%8&rx>_=C{D+^E8F&aS7hWjj}4yE8*@C^337AwGr20*j7y)1WfO-rIsH
z29O1+?vQ@aFt65;r(J(@WFEPXBmf!15G969{2@dQ-x2*HC8E`_Z9nO~J@9FjoP{RB
zgEWO+7`KU#Wj^zYY>9+swoIg>ISDD<_^%_ogWylNBUYrzuC~~PjWQ=PhU%K258};p
zEzB^x;{3e<6Z4|#wonP23iiNPH7l-ob)Piki_mvD1S`hU?|uD&Y1R!LZP~Dqn66)>
zIS~fiRRpg_uI^^%y{-dK_EBd3xeetSv)X&9ew;|0azT#BkME(DyRCCcxcC^zpYY>(
zoQxqgoBl)ojm>D?9<s9G{8htKqO=B9TUdovTv|iTD{@KwT)JzAIx;`X{jmj<qjJAd
z`N>`GqgM*v?5x4#N#S@U`o<)JLE~5dLu<D``3EB;X|0@l(SEwRPojcqju*cC95MoZ
z#A4VpjndFRKY?uVMZC_u9dc3^j_dK<gm^J?5P#?XGqiJK-akgEup3MWJ196&%+0_$
zxW^swbmlR1pXR3Df5SNPQ#!aWOH^IEn(P{(s?&Pa4|8h!p$jRY<8mOzVb9n{NZv)W
z(#^+C5eaM2rXt3dKIA;Kb4=PUURcR4R~TQ<JLHQpe2vYmt96Gz;<5OYAj>6YmL_KL
zJA*~f0S!?XA9XtT)HSFYP)DI<WP&p|9Q`_5^<ffeLcstJD~7jEskrYQX^IEc;iH<%
z+~*jh>01L-&hbu&#sb)q+=+@0LV;}6`gE}Yi~JE%IZV3TY?>sScFO-6vF)m0oP&UY
zZ+G)6U-b@5@Ym__kupQ@rn74FAi?{EV#W<(+q8C!1iB1@h+`q*W&c1!&Fr^?hk*h-
zx7zq3A(3;4c45Ersqh|J?eM$W9w!0t;VG%Mf_Qm-geYY$(QvNiAv3@Vrv*t1i#^Am
zJC#Jlo8wEZtLl<1m<9mSwABc)v1RMwx;R##{WvK<8|8gCom7MH>hr46zt{$j;ql-d
zQMny@u{HPg%|qN4A4dpnZ%@Ce|M>Ko7KGgslB{21{1%hc08S*MpYQM2hL|#jlEaf+
zMW<-=L7w!rBY`0{-V?uP5PTq`jL-F49Y7bc_p_EO2k$d_le-aPPkeg4y%yX9%KU@b
z89=?Seb;LvcFG;{fAVVMJEP5rr!Gw)ST{`IO7FvimEaS~ns)Qc4>Hh=XW{e34E4s3
zuti?n^J4J~Xs?CRBc}1I7THg(`kw46f$Hnr$}u8<glYBHe3zk%T;KJ;L}j=`?wY6P
z<%fT6do%qq-sk_!mtZ3K4Zdn~9s<%oTDPilK^Umw_!$2cspt0xk|-rA27?2MSkXoe
zjn_sk(JcnpmoaIB4|gOY0W%rKU(Nc8>I>pu7&Mf3Ss>9Dl-sxx6AoTp_e~blFj8$3
z3s%Stq}Efx5SDgcv{5M4VFYrveZSIO@S?m`fm&WneXoc>8|Uj@AV;7#%CcTHy8{*@
zv<8EH8j?odg{HLn(Y_4TQfL_%tcM%?QNdxsbg6PU5iqY7v4e+w;U6c>5`AA#B1&J}
z6lqG*;5JL)xm4y{fW$K&;^fJGexBZKS9Jn>4H|?#SD;S*zbt?(43vauDT)))qTXz=
z3Y$>ouuxc;8>PkZWn{83cF&SDUrRn!CPQX$vpIAz`KIL!sw#-H&dR(w&&|8{?fzeX
z(W`Ol4Z^lH%h>1YGBeHFp@gkKAH0@Sak2-4xbGr(LY%7!9QL<Jhds+>nCtwCTS;vP
z#%9u4u*4$p*T>XZ^?@QoEiH(5Mj8hqcGx|k{uC8|`F(vkTb!H6g8AM~l6TvvY=P1g
z)YU<F*NMAIv7^iR8g~>W1vN(YI4<*Q6fx@^=W7LtZ28!!s^uDU4}<xMoGtTZ@Jwy9
z|8l)Z$_sH8%)12keQto6ZZ0{V0ZBWfMp53#WD>b0YdD38;?BcCYpy$yJPH@1h}@{Q
z*rNPKgyae%-Mrr6?N6;?;CaCu<g2_cd_$jOKVs--T`@nq_CFDG&L^*Oo~4TX`#STU
zH;5t^+Vf*Sg*)7I+nag~3ihX8eRp`JE)*{NUoCeQ_4@R0({y3}gU=9vN?+*5Ia%`f
zm!fNxqB?rBW{!FT8ZKK<k?QX>-O(jv3Wc(Y4OWW2d;qyx!7ke#;#Ix2F=DC-W5yF|
zW=FmE(_SX+M7uFib&(Y8Cdyc1=oq!QW+|dbqCEbE5ydS1Cefx(!z||GeWlTimK_<M
zf*q2=DVMEuiE_5bcsF|~Jm_qRGJDZLatFz_ZL`m^|JQaFU|L{qY76)Ybj<0ji8iTa
za48}#E98oLC<chw*6qepv78NM8!WabgBRpSKDrTB-KmOdH%(=|j008wCj7<wl*UY_
z10H~RZ=ZWO(JA6m*SS<03vy~GQfNLD6c^~cu#A#b!wE3odhqeGd&;d;8C~0!_hQwF
zGbXX5w!|D0Pj(9WIFeL+d!|73_Pd~eJnu{KPfl799s95OCwh_!2S-mXM<>uk8RDJ>
zgILznJ&Eaqr-fyyf5D1idC9g3MMglZE7_jr{X*E^Iy4A9Q(hk0KudbK`zcm*i|T>)
zFZSvTbUnXj+j(t4BIR-CkM-zS+F$~KZX+E~e_f7?6T9{yrbP{@Q%|ZZ{8G$r!H9HP
zz1yv^#tH<3qioA}eirhYovzcXSnsu9#Ofew(iboExfI05&Zi7sZqHx&gdh}39RB6X
z$@EWE76EyUv2yqEzg>4Uct&0&!ovzg(4D%o9rok`>);rAe!EIB+Pj@YN>TRmUC)$d
zFc}h<^w&>7>>ZixD#K)zZT*?9j?Rl!9U1fSpd@t(iLW)N*T}x(IHMt7d5%!wr1?L@
zgADkU%j?L3I&CFpH`k|d^mIo>WB7I%jO6edklOd2^Se|AvD<!Ci52ElVsb}@&V=l#
z>VnPDQ+i9O0cjEN<?sz_kBMbX?|;6^Y5up+>w*fAsr$^~HWFq4oYyRemxlsfD#nRT
zg7{ASr4yl-|2Vjj#8*Ud@X98fZF&H>0`*mf-1~kg9F+t#t~*I#SFc9c7a@G&w*_y1
z(%90yhyol<+J<HP+iN~S99QyYMA&>NUr++Sn!P}APC}LR%mROH&xbhWNa;q(hZ46~
zX;VB3VL=Tj*a#*=8}%}U3uL9DG0qE0>XYH*Z1?X2b2*-|97`m)RWl0sF2A77`BD!L
z0&QY`N4VO5R^2o$p5PnsTxu^cnK+Jk0oid9>zkX8-JkxwDV*DNEC?ARaLzpT(P{Kw
zy`O9DKaxp{P)J;hjJf+P&Zw>q>bJ^jMtt%AMX}Sa*}+$IQ6kKv&4xnYp)(c%?~7`(
z8x(l`Cs-&j_ED5xHjTUCqN}UK@1;~$J#UuX^gldI>zOGeHQsztC7MdvC8MC~yQu`N
zUMG*rN^j)t(7C~j#mx^2)*UWdiyfu)bAT(*y60&s!VgCTi;|L406B`(5WEDJe^c3+
z$*Sepyo$@~*BZVwG^r7DXTpfctJSun8`>dV2`sDnG;ZF#-wAPrxCq$zYE)`^Kp48r
zeBh5#Z(KQkfX)5+2KQYBI4qosayrH2zZR1efLXgil-}QO0bD2b_l;e*=Q9r>Vd6OK
z9w*|%@Ngu?#kv?(pc4NX5*B-36xmsK3|P#->l0_lLeED1j~|bwSV<OgkXwrY5t3EH
zMmOjyKayuIxWwE!j0XN0-lm0?K+j^-x27e_fIF-2trc;3X)?83Qiw~<7tP0`eU2Hb
zZ0XECSW%*PNPQ4SxT&dT)_C{zcq~zIM0c&)nS8}YoP@H9lmYgt^l{cP2Ffm<0+acx
zbb3XcS2O0nM2~%}WE(1$!Fz&6PP!&Glo@D7<XHx8?6%ig=O7QMCRvs|cQ@8QKsdYO
zHkYEDt7TNWru}NqwlBr;%a#D^Q?g(A|MDiJ?L-h&JA~ngW>d=+wPnmmCt2q0*OqNj
zr3Rq;8q*qX^Woo4j*s6JTz~mE-WD>H9um)Rq0^F!)-PR#K<+N&7F<XNfAJ>u={<PO
zgk1~E@b~@*R`idBjiQ_Oq7z5M2*b(92pPCHb*US#J`E>g6SW&r91>K@-GMIN>Hsud
z4e==5V%6uhu4do&qJ{8Gpl52N2Pa%ZZ0?teQl}Bt<6oW>zJ{iCI~m_E&rqL3N8$`J
z1RAGs?w^21Q?ZG%S1+?(#N{;=&Vszfqu&RmxobZk=f4zOdDUdYWTcjGx3XFGPRG*b
zf>-$vV)*qmBuIbS5mDH4!8d>nVuS`<*E2O<&)mBlsCa67qjSa{SbJOiQnl_7hg2M&
z>%P7n?yWZ-P36uqetTwF&w$N<-P(epCw<EjyDLPOjx_Ga|8&|E8BN!^i?8?zAl`YH
zc{Y<jddikecR0N<#IhG>8vJcB)EK7b>emxQ6ZQ8i^N0>VGvewL^dH|<)cD^aN@zJU
zHP*JzFry1}*an@nv{Wqeez~cMIWwNOtmF;;FZa0&^jU!{FokkY4DG2FJ(ADEN}mD`
zxLXAHvyH@36N$grU?P>VewF*7*TAo)YMG|Fwf?=oiv2{2=f0fz(yxk8XF&W^V$MCN
z_zbv86o11URtTFa_WSdB%X|J%4^nmDS=5qT0|#XPtd0(X8wSD7L}tS(_{5b|pN&3#
zq=^l^*5ku>t92cP4BdT%X1zxg9T09xnmDN8zOqqZahoLUrz(WOp5t_C`sb{vA?7<t
zlD$V4OW698aQK%fXO8yokQNEc0=A}I8XVJx)W=;>ji!2iAKqbVV05Mh|3ySzkdzNk
z_*F_Q^fFbjLy*i+edqbwXKuB8TZt%L6Ezo8-Wp*0L8d_H$bd|HU;K|tj_WxnbGpwm
zBs&Fq@d7*jkZTCSwCCKh_OcL+@mpf55xY|t0>mif_1VmP;NJ)C^CY>#tHAk<zrABH
z?$!<M3{pLVm(fYld04)Mc1Od)S4%(gc+c3!Dk<;E&rIF{9!xh@E57GEleO24a^{RV
zZhvU>znuG&%4v7^<}$s!g)trxohfVYA`6Z`-BnaO@YhXDw(ROaNEGDVeo8Gy6}zzy
zJ?87GBifSbmLN<Jb-F2#^QMbUIcWbvaMne0o=?|@J6fLidv#?{5~G*ar|kAj<$CKk
zhN|$R)?s5Ck}hC33YMW_mg?j?-HXG@^QQZ%M3xDr{X;CmAj<oWOe8u<#sg1Gn>vZ9
z1Bch?NCrD_F;?U9*xM=3+R1SGkNv@pa=v%A*6#+l8C9y}Iq`2sp;ye*g0TqD_|vm4
zqLl%CIOap`A0SBVVynfCW)jUUkNfNAq}MRa@V@l_N7Fm7$JMs|+p(Q=;>K!h+g2Mp
zX&T#_*tTukww;EJZ5!`g|ND7Az^rZ7+7`~^hyBzjIA}(LM=2GQ5`$RiC&fJ!D6Oom
ztSf2>fP7n!fN}JqVd7J=`Vi)Ii@PZ0`6X9O?&p!zQpYdD`7~exG=@Fs4%a>t1jqzH
z^0W%ctgnkRSg%BlUXub(38)Cn8&q-^ned<N=Kkarf;KmOynr6?>jtSb-PIbcByGHn
zSyUd+A<G7YnskjxaD?K5my8||1!~I6+5hgyN*6c2v3K4DF(Rn;_)5~k^3^?D5z;3>
zR@vUgZ!+&MokqhjrWyE==k=+RLDQQ;R2b<|Briy<9Jk)sn3LmHV@9-a*c;8Cf~=KI
zBW<m<OgBg9N`T3eoccI7u_Yodhbo-vXD4Fdp^ts`UQ%BI*zSh=v&W>l&o?79;pQWc
z0wbGScB*VMmnvh6&c@5u^^>fvjN(l%4r={04j}$0|KT(dhe8%d@fH8V{4&i1+5hY=
zK0AZ4c#4cwbl#w`Sjw-Yl~Z+W<M4|{BtlO);H7@w;MDMDS0)CF<xYPPExaZAit~#l
z>CZH&8AEk^S7wc}`^l@#uD6y5KWf>41ED`$z8Jd~bhp4zK7}Cy;f_*TZSAZKr{c-!
zVrWSk^sY$m@2vT<DGUp>6x&mXQackm{KvLn;5jM%)2uiR2n3LMEv8!7pcPWhl01XE
zq1g!Hi(<SCx<V}$>$K~)FJn&faVgDuKm#^eENOluz`or8kK@k{gG{bMjI?>ftG)F^
z0NN=nop_c(^>+ezWWUUO7@mi&h|i2-FHV}tv-0#TuRk=i%R9v<&Z}F$80q^>YjUth
zehjE(38Jxr8TUElu5jhLeCke`Mpm!M7k?ry0zXjF?<i`Eo=dug9MnEADe5*M{<TMQ
z;0!!gR8@JytzE!qrp`*6HNwn|b&{rJ@(kG)89Lg4<1Vo$7Db_}wCcrlpiTp9F~{g7
z+$Xw%;~V3|turHso1`NTHS+bX(_Wu0Q#)w}l+{SK4G@wtRanNM9=4SgVDY#;AFcFh
zD$wAZi{<taLlQnc5<-TZF7#{8kRXC&JHqnWM>Z!oe|rCYW)CxG3;yO2Sf)@Br`%ID
zWw%CM4%8oa6W87%(Y|$EI{V&QgAa14yFm2Utx&=G+}CN>_cjvW28mAgR!DAflM@Tr
zUEa_)fh|YgM$64&o*eT^tzq7#D@bO%60K28SQ#UcSSmf2_S7ZS_9xM)BXlQ=MQ{zN
z89$cx{;g_$&z?OPzPbj{{<F^*YwhOC%JO9QbMC7JR7S}w8ARkL9{p-p?Gv|eY#P(x
zWz;O}21b`s2Fp+w7;>=wPn1`Y%d(v@mZ4sgTTiB`rfi7C_EDU})B7b}eDH&rPdI9Y
zzj}A!D*F~Wk!9Y+kq$ZjslYp*c0%IiyfiUhNmQHsOV;T8?ZOd$<=Ls}ebw>D=I**V
z#Hw>wd!RG5$Kc<iSP}MJs7B=>_-M9@N1=B3fAbtjkX7(<56hxaU$Fha=w!Pl^3_i0
zk{Itb>~9NbHB;dwRE;XzO^lU{<c-2$w*`B?6?eOt^U$}#eRSJnfU!gugOZ_0KhI(H
zesz*ZqSru4<bfqe;)_WzxsN%90{@`G!_UDo8a68@_M766`*sKeV93t*zFJJp_F99I
z*paoa^i>TRsH%mDZrX0m`n*IV+S51>q2BzMY9%wj>0*SwX+XYBkGP7NBbBf7tv1!@
zb6p>FO)%&N)vo$qjZIvK8eztb3C)|1>k{s@_D}uea#fv4oX4X$)4{x{{9h0vO>q7Y
zgY{&-engn5sKzu^lkA2sZ0#i2X+_jIKKmq`Z2|r{!ZvoHMDa55EGa!`-_p|yJ&;%W
z8ML`?jWQ})5OkJbU18X=?(A;ar>dyv1n7X)U4Fk8_32HV`@Z{Dap0$E@t01Qd?Nbs
z)&{t(q@Q(0#EFb31SLEg&^4uq!_NMlB?7g#acOn>R<4i4B}*wljCJ4sro=IzNoc(P
zfyG82UjdHUFWy4?^x;CS#AS`2cP&0RW7{6aT98$^RRd-^odInXSPM<bmG(Epkv-Xg
zgwCm8x{limV`waHGuix4M*29)`93h4eOxXiPT$tK9wCoazuI97+Dq9Lr|Q^zlFSzs
z@8t1c9`p=xg+7{Y)x2xSvTp_O_U<rwPji7jpp=s85KKCLZI53Vxm+3xMV;5nE!DG{
zN-pN);Li!=K;oDFGn}!V#Y>nwj-~_2qQ57_&fh=LidKPn?%TXj71fraQ+2ILB|!vs
z#MuFS*2#qXzN*TPXyfV_7TRWMkZ>NDPb~yR2Q6&;J@VL)-o-_F^<L3Opk1ll?dby-
zTU03PF&&eKx)7kGqmpXgm#Z7<&AoyaqOFtvVK8ASykO&hTa889m^b5o>8QlpE#%<~
zB`%RKGnC)J^I}}bK4dpHOOTBJ{O5he+R6tA*STs!vR#1Xz1b=R&Na_IaBj?!0bGHK
zR<w(jNx!F$IQ5(t(yAmH(CI*r03EEuq#w4v*Nr;X&QlN4b(88RVPwXh5n^1I%MA3#
zt{8xVRl%aIc*nrazC>6>gay9sN2K8CUjSUJ?+no%Q(cGu(CUFQ)%dP0cP#z?n^#IA
zrCRZ&z+?J8e41}(lm#&C7&zR|217#3_Zl79N__KAH4tKBm5BRPkB=rBFJ=aa;2WSE
zJ@1cisyzO%<+T66_P_&_=Nj=#c)KT(z~4sXS;9PO<G`=P3@~!{ik&mBXnEPPC{mKM
z9s_a2*5XocS}T~T3{i>{q-uN6SP98UEUvHiA_=5$t6&Q3Beq8xp|uJ_=|E9G{!v)A
zPLoq%lM{3QXT4+hIG)n>ZY(lr%%v*}RmG5<N9^iaUWix8;DnOTEPEW#=`bBgb5Hmm
z2x)NF272%y?hEuH9RO1!`B6O4T<~z|@~Q#=L8wESxxoH+Y|+}HSc;+Pw9W;-Rt=#K
zzgLI{c&+p9e|qB73L0><usAh|4e*D;*LSGJv%s(poWvTo<g92-|2V4?sY#mIqZZ3n
z_GJ_Ksv9lrPr2OgcjCy`rv!Nc%xg=lb0hM*^ys-2HESUg@YlhNrC?vgQZm(i2_&u@
zl<6<iYoCZKXPHGaseqc(Roc{(**nTdKb)zGa^`$Rgs&{zp7>7i#^F*ACcoEQpeeb8
zW0S#UGlJ{l4iwz=a`?`Xx@05{Nje!kXj2dvWGcM<Uy|W@-LP9C7f%zZCpb+<77;Nc
zKaCe*DmG<b%0<Dp+7px<d%`T%>wcjE_WttKQRqVGX`3;^zWYgdVy-@~9`&v@R4_<F
zKVh!Uv`INSWhz$jwk!vI5N+Msg;3K~yy;d}pN|wq8S9M5cE&nk>Sh68Had+%41NR8
zO$m9&7aWyWWR_>i7C{xLr^(Xxn*sk6o5>Rq+$rMA!5@ofoWUmI#R*mBaur5n`@<TV
zh%AuH+C&k&csfkdrf36kDqKUDke3~9sK{QpRwJ^17rE`nD%PQzem6}Mtz)r@77kv9
zdv>L|{tRB`XoZL6g1^|b*3WnZ{&H1VpWRP-_|I<8KmoaZI<j`=Ano1^;n!9H{GgK@
zdvU5z0-1Ao2eU473n|hI?YLyR5Ie-Pz2Jv)ZS5NYV+Fo=XtQL*1V%Jmwh;K$oJWcG
ze+Ppi*zR!)e%zsY?g>`bChp5l)avI`T+aOST_WOKlC<=N<ngDRLT{$b9a9=?Li$ka
z!c>QYijO2qkuEUKDM4<He$3?dyfo)?#I>8qqZ~=Tp{^o|H#`{-$sw}9)V&0EF2^SI
z+?LLDg&u-g-(^j=axtv-)jdKUPb!PO>ol~9>Pft#RXGnKOqR!SMSe~YbmRA|6$yk?
zWSq2r$cyv%j*QHjYsjKu*?GtHb=FG#*LlrwjIW#KQZ&H@nkbSrOG83AAByL|f)EaM
z<hU_LuGMyb$YHlSH~I{VOE;%Ipf)R2xA4xIVpw;G28Z9SoSzHri)r_Jgg27Ul0CRX
z`HETp{@>~gO`Hd7Nh3O}Y)D*Mksaj;L>P+!{eCHQ!`8BrGUy6bw-ApHyeqvSpM)<X
z%m*Ld5`5dE#n7ykq`-=(D^shcuP$a$xP>IH5M$JSGreQTt5@_*#%p6sWotwU$Gz>X
zK3r)ogjZ9coY#OKhpjEfK<4;Fjlzg*$i=0MvWiperK6tEZ;`77QXRvk6M5#;qeO;s
zJUoll=6u)L%o%k{fJ@4Y@3*sIf}d{5wFWxtM6OK5ODEiA#O^7t-nf3l@D}QOIAbsU
zd@0afrnz;`>o`$+OOl`As`I9n16Y@(zJk&0qo<xwjz^{2IpIf^ZX8BGd)~P*9#|i`
zHW(y~|0XT&xZ3qZc#_#<4Y$MhxaBTfSVZbd2<nx6X0*9uzdmdZF0A~d(nA@n`jR{-
z`q6xR9Uk0<f2p3xd863-&n!Dc5MdUQyu`SinIA;8o$mJm!=y3TEWTNHMV^1{&Zo7o
zju!5T`<6M#6RFfxkS28TgfQt~e1>I|rBcdM<DR0hlRT%p`fJfum$%G)9eh!LYPGO}
z7<1HRume5N_+@!s=5GOqzCE2&JmioPGw+yxKh+UW7q}EN8+Y9-3d*d(yV*kGCh+`j
z|FG&p>QM6HX{CBeNA#JlM8-L+?RUU&|HzBH++k$$aOZeH7$#l(WyHx4VDR7I&HXE$
z35TBpOwRNQrC(HMqI>}EGI`hcOgB&ieD5^8mr{!5F?L>qawSI$53OS7BAEf>H<ik~
z>&|BXFXoz^^hg}!mZQOi*d;brRKM)A>8PPSDt7Os$*aN7xjVlc3*tqR6~2dzgny~7
z!XX$bW~ANrJ!|P}9VB0w7{C1})<GdFVsZkFPPZEcDBPSs^}IxhY=&8Zc{P#d5ZZD~
zfnyVL!$0yB{hvla+3fwjqscTfk`z>(89Qcs0h`W=+L8Zeod#l~d=iu2nwqP>hlCP~
zK-nuvf@ezWK9jWw=6xW+BhriGPqpTH{>(_W>?_D>_1Wq#bnpkkh9q|jflL`ke4nqM
zcy4zP2SHjd2s+FDM+R2Fbm(q<tya4onfEy8z)WnBg%Kq1)0)OR77HcV{yQcugE%GC
zaPoSHS?4$mj_yf<XNt)QuNK#v4&lJZwd*RXy)46hc8t#xF*^EsPCA*Ls}OyOZQ8C>
zE<dKmPx;WC?TT(6bW`#}x6ftSI-p+|-aAfU2VNmN1F|Rn1_toyT9<A5uhINm>&GK2
zg&$VA)6z83>5<%|zx&;xwXFM=a*1e`I-y{|`wa1yeJh1rJl!}ECdMP_)$8^a)mJ}g
zU?Nr|m}S>#C=1b=wB||HkX_Mt|DwwX`W()z8ep051?a{!7!8FB?{jCfEESI;A#WO^
za@vepEqxttaSaEGPy0)Zbr)3<4^2j-oqhUOz%6BK)e^uV!<MF~boNNY8OrXw_RbP7
z3y3)pcC0mGGQo06YencA>t!9jU_@LoE-fz7X}3F>%O|*5Kax(IhTonj(-><-y*0vM
zRdPx<AJxG7E%R)s82s|2g03#@HM7eB-?XUkWEFT7cnne&%J!glw`j5p3KpM@EoUp-
zF6})dE@v)#O(-Tli0?68RIm8#>8DhG$kBw}$NPAGEJsX`(?w(WR*aUFNcY`#*vGFH
z9p~YnvOk%y+TA*rlNbD(A2kW1&d!JR_U8Kb2cJumz2uyVCqK5wt4K;>fyzi&z9d5Z
z9bAs|8*hNi^8tlyveh(9-sCzP3i6?~Rt)ukREhzFL@n9^^nw#ZnJZvz6r6cUB#(QW
z+y@N7Jjn(jN^5lDN?m(r6DgY032vtSwrAdG>z{?zjOOv+kELg4Ma(DXnJA*2i?dLC
zaZo4SLQFfb6(5+QS0fB|;C2JKaM85?DheNg4{qoeTdUl9BQ@n5vR9X|dE;8j8L`)T
z=~uUyfy{4<{BMOsSgd54ztAug9m=$}tG(S%K|Ba*oxX)O`^hAjP7C?*KSffN?lVG-
zYz^*&%kD9_tc9=ji+}4)vOR@k67<x8o;!U4gSgLMP*D1Y5H~o#hnclXAA_1+`bmfm
zn55^EFjW9taktT?RM#=JH9uF$B%}F4uXDzc$(ivOp%OfT#Bn!aUWD?hqosGbHMLeE
zSQP0llOb`reYa}?J0qH6Zn&!A>%>FyZ80N(-x|;C7$p2%9xVrn_S@5TRzH__A^=h~
zeNg_>q6a9<QmJitm~dTJdx;HOA^(Zmd#8PVjeGKMv@`%o_`=A19(*`V$c?SO8>-W*
z99PiZ5Dlz|s=~o#9AA%W?-&RZ&q7(3V=LPOZVE^GUr;W!W|TwLV^E2vJQ4_LC;O;S
zAqZ^I3;h9r2e#-m3GHi*=5>hMDcoy$1&TYGAHKE?@mIMOIQejRiqn0^_k<5*WuTjE
z<T(YX=v|`eQTL1cm&GneJHO*v#cR!#eU&s2>$(R@ka6H@BGs-!Z`sf?!9UR;zu63W
zTpT=<NfzauSuFuLF!Y7}8Rq#yMNT#GLqFT=R{A4hBFg<I!7>3FbPehqWA+U2n#6xG
zSjLa!IESryK8zrpND{eSWI@F3M-Tv(%-vH+ebKD$89)NdRn&kaegjJ~MKq$KkG-9Z
z!>m%2n=kJ@5&-;hnsM`vdx!nH*<<QSV9J4E!K^nNk|!FX<}lP=&9#{Nhy<}z)3A7W
zKN4S7#Ge|ZkMFw35cs|Km*e^`oG9(-Kap-Jxa)y<uCH}xJI=lZ!~4pROO{i9g`GJF
zJgY}D_i4b@X6yAHP*$Uv^Witc#%RTI!qrw}7&1Obr6(x%P|~+Y*DnxGGeUomV-d-5
zZ6-&X7W8^Y?fWTCBPna@IS&F0M;jyE%V42lN56x%X}@9|L%wRdk-vSUhA8!HPu8mU
z4O=N=J1)GWVp5#{Gd1z6tHfGcB#%XqK)}n*Rza)-=ji~aX%%9uk$lSo#SuzkPZjZE
zUuYs6<6?4B!d4LWZ>F|01{I3#%4Z@+&<r@E%R~B|=O*X%T$OE03R$%($G@9pKGtF#
zoV)SId@=T9oG2|}!Jxg$`%?m_w$EP>Q>-_c&Y5r&;HnX^_v5gICf$DVqpZk<h(~ZW
zY|N_u8}f@aTFK%n{VG|T#9kdt>DQP`>p}g01?=2r?APzjBrg#s_itgiZ5)EJ73HR}
z22d3frEBL%R&?gBv@pJ}=M8e8lL=picx^H-wk+0IaU1`?!cwd&N-h4QJ*5K1sL!{8
zA#C~QSHt|~QpE8zKj}r{O`E~{-?vWC3F{?k8v+qZ=RW-&Mo<k*OAc7zQY&P6p@P-D
z-QU220gxb>Lj$F;7d((?!knm^R2XrN;9S&9ko;Y>14k{^*MmdC`X%ptxC8OcNKVan
z=<Q?lm^0OY^bYUxCY5PGF1CULj=@lQx1EV>9k7Aswk632pH`H4OYl&C(59!&GNqsS
z4XX*c;Kd6pcISi9={@u7QWJvW=On(;p&G(l0JHtDNQvJB#r8dJ{|{z7{GY}g=o|sA
zS~y6b;F)=WqDa-{_#bA7HvQeqiK!gI<8;F^(!H>NS`MKJ;=QQ&q6CITeKT-sUp%up
zqjOu3VgJdwr7r4ldciZ^ou%sQ)9Ur&K(AA#XI58&)JbV*+Xofi$pt#+{Xj_GQrn0!
zVA&q*2fknlu!DDt>L-I%<FR+CmlzE6pQbEq1BO9hV4j|F1-ZoCKwc^&QyjQi9K;!E
zq=GkrB9SY~$EGcrsIPcqDR-~P;{#f$j})<Z{-VGFD$<Q8f?AEyVsa53*^r)!XR!b1
zm5QSF3@o7p&2~;A#U1iteSx_;<Ia{KcF!S-%Fj727slpl>Lt9iLs!FK8M~#X=n&t+
zXD_P5MJ6&ni^Wpm?Kk&;J%nrQoC1jUR|gI*OP!TQy{+&KjlC6r2z)0O-`+izH67ph
z@uh7q?oYkc@2x={`&jc0aIX%TdBJO7yw_hCmzEH~QR}g>%e?MqN8^o&UiJFYQa`M1
zCuir=mBwhMF>0R%&!<blKc2=KmPjlm79W9bf(7(nIGszK@}1vmmeel?`TlHZ0M^F+
zY7rgQWF$@Y?8WAugan!b!szIsMLPSXUCyB@{S>QpA*PAg`gp?t7?}UUgQAyzV~wlt
zxoy3_ac!<?u*7XIeag>Y-SPa!o0JVrN1yfAJR#3A_J^acs6p9?vYNpraJ9w?h7Br9
z{xVG2b4sYjN{*Lbz!N+&9sYandBiahgK4e0c@>d1eeicFTFlQ<wcVA{N`6cK-Q*YI
z6>GPa>r3NhHs+heFx;bf1wn2%VGqbuo3UcgkNRKTFBTTX>R<bVR`}eYXs%^Tc8x1w
zNj1VS9qSb=3Y9eUlP8;1Zq~sHzB~dV*A^v^z&bRI-DEWO3LDxm8E(Rl&kpQha!Yv<
zSWXSaXu->7YWfVYm}P$N)7c>nWX~A~G-smO!@3jN+-9tc2_;n-)u+uEu`G}V7+_!Q
zU@N(;s9r@KCM^8u0Xxh=2YKM}wz+NgKMeq(D1(tQ!q3vxhTI5-pSW~zg7U2yx7BD0
z*nFSEJwP?mI4-#FTQC^w$fWI?jx09x%~ec>c}=~)yioSRmhHSp4D-Y}pqT;2O}@K1
zD_~#xy<+BbGKp4wlS9W_1peq!zF(GDJHblJf-x_$-AOfGbd#bbV15HcrDs(ATM_Um
zCVE|VIbx}wOaCa_o9JNc$NZMZtc(+GK%x$BGLO41=r`#YkNt>@mH8h%QNTT^&w!a*
zy^Ahlzd1LN65{D$G30h%yV;)4*f!giQgnkGdI1zA)R^FMTl{es!>1uO{VQg45mWkH
z!UalB`e0bDv+;82Mj~UgneXE*x4(EkXS3<SQp!eO3c1Q}mOLr-C|WJ#Dwt)7zgTPf
z|M*qxJJ#S#mN#(sOhGqIl2nuFKj8a6F$)vkK#II;K-<`+z9{(KO*{))pE~z}{_3ql
z24d+Er|Zf5fzown($gy=wAT6(J*;w^c@9`Neh`;kw?adOYw%l`)vk^k0X026_<AXg
zKD3P7ZqF;ZAEJEB4H1!a*BAN2+0%<!ce!N2&WZm3>$U@eb)JdvRVt@S_OH~_XB14+
zDh|8cw0d#;`4wu--QIRbk0{a#)IT&GD$|GkPYlO%EyhUKpQiH{_k2fJ(dupfXE0Rx
z%Z0WH@XbS~yV~Y<C}UWDb~azMDZMS?p&emKW$0#hO=Qj7<h%Iku=)Pugn=jhPeBxx
zAXtld=Ss#~KVHXdF9K8%@jp8}Nl|jFR*7x%y<NYBiHfQ@BL{0y7#ft>rRhl&;rz-+
z6^zFdC8G>8BG}9o*Un?4;inyI-T%C0@e%NgRO5JN=zj9hxPHGt7A>(K9JBc=bfSsF
zFoG<X<js`lajEBcncDSwyZ9DQDF%ooQkj<3o296AivI=in(ekNNTCt!M)tDjlEt7S
z3(gkFj~~FFKIrVuiJo~a3Oe(+Z`I{JrUIgm&Gg)5N_0>1dGl_~5KHR8Zb0rWbe0ed
zp)2lt+B(m=ouxASdG5=-S0tb&Kzyt1e3U4L_(l$}v4ube{-4PqmDLtxKWA|LMnahi
zl;xHh*Hkt`B!|O3zV{_s<Q;h1AN18TaINU8g~@yi!Cr4nrKkKEF-n!zB6nx3s1g#a
zyv)c>b&;x`SIc+$Yzs=GIFheygrRp^Fh10j6S-^qASOeuZi78oP`*gf+%P3<{do0T
zc;@^hBuo2U>I>Zas<ZzMW&JTk{1`1<4f_by-?-BBFHpRX2Z8(#z8LDE_68bsA$@cd
zSMH;xI<Tu0I2I{<FHcbs2QekmHIV)1v_wH_Ns#m3O8kc<K_>u{dT+-HKel9{IrSp^
z!O`f=0s9sR{zP&iqnkP^{mdPZFX`8?FXf-q+9Sm4!g3)&Y+Me+j4#>u+IA%h*AVRk
z{xGXzEhP-3AA;N=jC7o;xYoLUL&-|TlR#>F>_F&bK)2bZ5r|3RJqN3H>{TxLL}6{U
z$*Vpd9K=-mbo+7s3{u-YdEFEOP^`&6L;70qFK=)@8Vo}mu8|Uv-tIxy$6*Eikmy{5
zaE?GT*QaR1Rs1MVz@Gmw5i9Y_-xJIE#0!)|%z&L6dFP~RCqmw`g`O$A9;S!pI>G29
z=?VJFjb_Uq9u-H6!j%5uQGLr$=hczC#I1jUGZj{nsj+L0Rnt6ZoJ5p|%blO(_2@<|
z?_Sp$D3R9Z_<1mt8Ne!ez13=?yE?nbm6==mFMO(RifL6sve|t<5p9GOlhD>!#*B;#
zz0ms%@`$TA&DX69ZW*dVetW1BvujOD-t;|~n;rePImTa`>tgu|1M-(97pg}1dHd~A
z5Y%HXKN<ReUa7yMPy!OnHMT$804Wh0xW48UU3m!>&8E3X9F*b+2Yf`iy6d7P(oT%*
zbFu#4>dR6L<C|szYAjeZ9t-9VIyvIRWM4|O=x+#Bf5-X#fO-)O{@|%^Uh1$!7~}=L
zFT`U3?$Zt%-n^;0-SOrP6Mn_DRA~9|Us90j2Qs)yTdMvXAe?z3ri%uoKZX@d1nL@=
z6Xw@aV5v?BtvKCv7kaj(XexjgTE#3qk{&qALCyCQEE;398cU_M$Cu@ht8D;b@rJfn
zUPc3ghKt~^qyK9*)tyJe_%QdMCj1PQitBVCN*O-_{qYa%>vx{L;g|rET58(1RKTtt
zGW{-~j@sUU=fM&KP94sCL(%^CEcL4&FgOj)Q7f~i5N^7~kCyaj4?p3|7yZF*(XAvD
z+ep%G^E2?_zbK#>)Rl_F9pP!(@SRHi{II|1i;FQ+0^Bl=3B)0JB$inr8XJr!!PVl^
z*-xYo$0d+}2}F|ElX}pMf05b|`8q^gsbF@8tbXH*>K8CU&_Q6pyi;=<L7aC}Tt*Ck
z>dQPfOiT^2RPNkcJ!CAxs!PTEC;R4!#0R7+GP%*e>u91eG;};o{IjIlrwsBzTPrtl
ztEs5oeDES4G1ay!n12>+kOY&rPC=+=u}~f`@9@Jyo73d+kpL5Yf%dqEPG+s&I+y;2
zWskkLu=Xp!x;>)UL2boAU)HW=YI!lYLf?I#3X4IZt(L!GdDyY*hfQ-mg%lk3I!;KX
zprlJoR&YAn%7k%PN);Nex$(oYz&4_5?Dun{y}CFfu-^_q^dOSycn>1@t?#hvP#=%C
z;}Wo!Tovhlkdy2AvtSO7oS(Bh^|JBzPQwUTRrN$!omNZ4*(>2?G!kyX;-ZE$;!rii
z$=<MQ392o{Vnq)ob|!nTG-C8w`Q0rnJkp8h2mzy+9R6;{So@=VGQVr5dV?$YYPqO1
zxT{Zi$<%ZdzU8!^x-ixHp%aIZ2MSlTtL7^w$KwBn_b&nirW!&8*VK4VCiKd_e|0{L
zU&k#?V({W@-{G*T)VGso^Qe8tmJzzOfufVZTludNYPbe5P-Nd15~KfGP}ew|VSpYQ
zoW3&%U<TQUgE>heP_wtg8&0=u`VrVTndBX7XPs#uGS33h1U7O&?JVY#$q(WmY%X@3
zArElVtl<M>4ml-c5V;vBSGF5BGCdx=F*mtfM!oL~Y-a&-5K|S*JoKu$LyB{43fX8P
z++5HR<|voEqM-#b^J7#w8%ZACNqvg=^AbVdmMPt&s|#G}@Dh6A8V!nqXoDnSG-_XI
z$D7H7pTCgiy=zeiI<4OS))RpI44uz2P4aRM__J$$S=~3bWk+iZThpNMnu<TYFM!ut
z)xJI=ZQ&4(XOuo-$!Ms1=sza?0{0^3`?PVr?f0=8lWw+@CbDf;Ox<b2gevtao|*7$
zl&jHYQ&@S@t*_;M`}C`l><3w)-e5jS{6wBGK2g>VFz>Q0MJ5dHa(qKX5=Veo4@68W
z>REa^o`+LX8s-r6on$+ew+WmBuWrsWTjae#!{;WzgO;p0;tNW4#lMG<?be>DVgKx)
zK`PzVa%`HNkdoGc<Hax=E$mFQ*oDrEH&EdyC(!?%*HElAipzcR%6#?_clQndx1HiJ
ze^=BA;<pP8gvW11ed-ycB8aNRa>|0Y&H=fv*#_N2TA9d3JmOx#WvHLUVEteIxWb;F
zEwyAh;Y7~@B~wVVw||0+T}R^cANXGAO%8=tU<uNgW1cg_bXEKt>q8UJDeprn1#6CO
zF2t$linO(1UoySE<;)Oz_ib}}E2Hq*oc)N|rQm!UYUg&|G#PntqO*0BQe*r2%5>{F
zM2ELG(u%Z!dC*dw@_&LJJPUHjK9a#n|IT*>C$=;L%lI$#O>Y~;3T1wm`rJh^bhx#N
z{ear-Bf^WV)S+vQ0aVSz-(^1XccH3>XL0O}@b<V^;zS-L=<BWnC2~QXHAYyjzmEWz
zYm86xl=C<ZWpsv?5aoef`3U10;WL}_@keqLOC7UxaDI$Jgw4{bs;**CKk>gs5#a%t
zNT+_>Uv-mxqJ(SC#(s0rf<!q^W46N-p~LUh=vGgMVa&$EnZmb>>3oHw=q^J!oVmXc
zfB9<x+@{kxZL=9iD^bG_F?ZIG4K^nXz93w-VpZ!Q-Wwx0OC2$wvICg}Klu3e2McqL
zNY6Xh>H5!rXvaA#VVaDJaE1h2^ek@jOkYmNvz29KZ8(KqL-uPLO9{E%At=jOffnYr
z<ci!wffxPVp*;Dt+9Fkj86u)&R3$~#B-_TRTj|(uhD0HC{ayoTmQ87p(w_%y?__=b
zm)|#XMB`5WSWlCfpq~twI<$|$&EAd#tmrZiFgr5m!9Le~j_XGbcbl%%8uslPAKNp6
zL#f~ONIjDN-j`l^RV%xu%+|lV)au@&Dh-h|@*r!VDd{=!FYEtlXKji*?#z!<QYy$F
z`AqhWNE^X*tIr4R%{JAQISTYqb{7O_ha;#E%yCK9W29A9rR1^Q;ojm$pi2$2&ASFY
z)gfnBLh^r{`J-+xkgwH$RX<5QjlDnOV%1P^Y7zg)wIohHY^xb365)a(2zYb>UUx9a
zaajx?J!jh3#aC{}GVLrwVbkPG?WlyO`d2mLmn=kY48rp|KM2qHsPyFc56=jTN97gb
z@s<iwQvXa|TixE`#ao>+%<E@>PHWlnd6VlDm5V)u0|nVpDpVcnj!N%N_n_J@eSSE9
zCZr;R4&#CMTkE0h33sxom>K3S^G2H%8MdAyS302udAt`KHSM0~vWfi<%J0zM0qYZf
zCM=Kc^Np4*I$I+t*4j@m0Ti8bHP7Vad@cm;VT3Dv#KL(edDR3#)i5NTL?PM`p4?N$
zjMwoun0U~h-+AkIh5yVSf1~m$n^S;XECztl*CGf8pIt213bN>WadMLb6r}=MGuirt
z^6cd}U-anW%u+b`J#imh&ny%V9v`6<^52tuP&ezk?2D`3ICMPkg!Odkjw`hk4U8Lu
zAb%fZ-<gwcZY7U$MF0QLk6dX^%<cyCQ7N1SqJkj`<lO*q%FH=fN*y2IFge}p;G*g^
zfDL*a7Eds<!R^L~9tjjvuO&Dpht)WKb|8(56*5LUB>!wmYb4Ynho_AM8^M(rUBurY
z0-K9z`KAdUC;fJo_^eeK7b{Sin~>+ko>?Oo@exI;^c-d{KOqFQs~&bTTe1MZ)-^^m
zLuAE%;M{l5d>|0O58Zdjuou%Q-jzP!${lxgi!~V8k2Fj-V8&@D#5p}H6p$Im<Akhn
zk`V~g7~wG#{!O0~PP$Aw9Mt{lx^`fef?+AE+WvqCcs42#ns4hT5uEY?#yE`>hoq!M
zA9TXK4E3pD#Ssm+!M*}MgA5`Y3BO971&c$#Tc)>b?RNU<3v>F%@&)~hE$z>~W>=){
z6?y}4A7PIS&OMYs!etxrn^Gz|1Na8VwY*-D;9MuEGq^U@!aLz>jsfeEAU3_$y^pqA
zo+WEo%K#3=+={OEFV^k-F#&2pSz*~_?D8nV?_@(Y24f~*HfdHOnUmcq@rEr}t~gnz
zXIn6JxmTW9bz37iu-%vvCcOTb;mZ}rSI)l=$QGe#D(-s;Lfs)9u_%A$z9NWdO<-iF
zAQST3yXaQoz>u4(eh~^JBVt=6r)L@W`>Mkbkdu{iTs``q{V82uU>%IIs!c>%^>RRI
z6o5y!SdSj6>?iPtCo@C3gsjhq*?CU72_E;)*<yl|j?G?k4MLc~GI4@9^E!PI!)(U?
z>R~;4v6RUy<!KT=cUZHhiStGza~~X=eS)|T)^t&4^06c~BAFr3@U&UtN^~zguqKHh
z*xsL4bf)H|x*JotPh`yVS_*KUSA%8NtGAMvq(=mhoGFKmsPQNa^l}T$L@gz|BJ;hS
z4{EmtgpG6?2+rzItWYlkNemq&u61S-7wqVjf?GGiGZ!D?^*o}NId#cLD)1HDJPs<I
z;8psS2Jr|hs!nFb!c_;SNom*BimqAl71&+{(xSYqzW?`f2TgdpV3>9nm%JJ85Jh{K
zc0FlT@74mEN4+)<i>*WI0=?Z~VHzD+I}OosQ?m}ap<KD*oC}n%9U}HIL3~r#B8c0i
zZLwIQ<Zz!j5TiAX-!v6;+e2D0dJ#?I1kNEcgW%F&;5D~#p{fNdTqWM%w`~0&fPM%?
zX7u^Lf>KGdUK=p?mhtU?$qTpw^7=p7#?`^Enhb(e4$Gd4N*Qp%HAB!p3?b=D8JMNK
zq5d(yh(c29dxM+6;0GQWhUmuD3<X2eh+IT1KA>U;+$O!i)bo9=SLB{?U>4UM5KlUQ
z8WS$u+a{bH{fEOnc|re#USvuvyh=gY6qUB|6V7Gfqbf?d2;DtqiyjeVFYKl!X%hI)
z<%#%pkxlMg9}__F<U6g(K4#_c#;Bkp_1EMGb4=r}mvO~^(IE=cxmjP>zu&RmL?-RV
z35B|1>-z&tC=RMBA?{)w$Cz*ej>*%F>$QYjSQC)syuCrZ6Qtj*V?%{r-)bXldm=b4
z<lcMO4ZCXJ4At;}sg>D_GeW~lJ>})6nYF~QdU!LhLS1v6#Fe?gnl<8*3V%lGW&xh0
zo7qRtch_{LsRGWDZq&55YcQhj2&`HWSn6EFnp|Bka97CbJW&@S5yxjLb_GnU6jFL?
zJ!WeCCDp+Mmv_P1Zc_-1*$T_wDI8+8(J&|)MZqv?`wKfcd<ioUvW0X<E2z_#bwJNB
zaqdxla<h08WP?BL8FIC(1xtI!U8xzVC#?qw8T;st7}LCg?AZn$(|6e^T<zvMjNi8w
z*yyfV5mLa)xw-TvIyf^ucDQMYA*CyW<^jD+21F)Z1-0pj4qX~Vf41f-==T$Lk}$V2
zcI0XMrjNA44D?-nAs63Jfp#a2$!ChNw@uRji*NA^fKyU@jpXV=y89)DQV=YHiYp|0
zU|Z4&TSBvUoRkKF@VMas!U?}oM_&%*<&zC7xD)>>@fIf96tij4Am&4rZj8S%%H24m
zuCl%c7hCQ%;uw9fI2Rl!tcBTpBwmbzX7;r6O#eyBaQZz~p&AiygRz+RFwf8*vz_hj
z?KDFUQQ|z^83{@XN4mrteCf7@s{SIdjoQe@oPpEoAxM3;I9~TN!MZOprPIAqZNF+^
zhDOKkkr|gfBhKJ;Rlq))^2({24<Rh0wzO9Fg!bP+3HwIBQFVM!hwIz*k&=cRvzxvg
zM@0KVE|tc4eC7+69FIFkU!I@nG3p_{RkSe4p=w@Vr4w45(E=?_)*y*EcHdiSg_0Ig
zKf!0n8r)IH;frwFS=qAU>?07J4`mMuP$ema`$V+zsgqwoZO<dg{jy;(Z`oC#FL^hv
z+Mbq(dH-N}awa%FwyV;}sAg9N#GV&;??YRG)pi`6opS6`c3gdBXB~kbN=<(466ErR
zBb)8qCf@qiJ%Xo%kVqW#&-wid^ipMOh(LFK%h>0d+2`?ZHdAns9o4c~8^p45NzH81
zX~>w1qFFW9L$k>3Qqr#eRLfl$R4k}b)tK}93s$=Biqqe|ll<A&Auee~1t0Fp(j~tM
zj?#Ot=t`U>KhFc3DaMD+W>+=B!HgpeO&CSA>pnff%-J1D3sC_-ODZ|BO=&t5|1Du&
z(gCmJ$!NW64MCndJplV^i?3E7kQQ#imC43R(C9g!x6wQ+m~i6wmwQcdlrcnCAyp9*
zywP|a`J@eup6&%21Us>fAlGL0Bjx{-$O~~7Hk~+pGPq`&>hp{LeaU7xG{^nljo1es
zXxGAO2mEWo45tkt3c?okdUY!^J!;1IUW?4qrVnM>)vXz+o)YBBzoYX#XY2$F?=psD
zQl+a(2}{pr4b5&D{7K}M@{Ke&O$kRTC7d(zM0FQIpzWhoLfdP0Khg<Q<niafRZ>%X
zBtlar5i#3iCpxEe#Q}dWOl0}HUHtOilhpsMwcPT)twi|BGe7H<bMsnca~6#2WE~2z
z{V0l%x7ijambIf)5=W;{>ieng;<YlD0>`sV_AR`EC?eLioSnU<XfV>t4=jG;()s>b
z54jN&B}z=oZ}hNwCCGf>r-hM)kJk}&4j&C>WkF2zE`}eyv#mtUujll`L^2UBm97Xz
zeh^kNvCfCFkawfnGU@~JxE8X&-m(TsrIbry^R9mm9tnyHl`nTkJv5z4v(IqQ_r@sp
z)>H_uq~@?0_bNVcj=&nsKael)3~Ti<3(GZ3t=LsX@|d9dfdD?l*YD89;&5mZ@@F)=
zc{e#G4Dor3tzW&c&SGCuX-c_K$QOKAd|S}sK)sl^Kkr+v@JB|l4J5}7$M?%tc^^z0
zc^0;+-F+5&m;4_0W;?ie+xqy3SN`s?Yh$^Lh{)0MXquB>ZuRgLkWVwD6B%3N3d-&$
zHMW7dR7lF{SIY)k+)%@V4Cu54#EoLdGrVbM(-nAY$!F)c(He1B$E`NCI4h>%TuUdv
zOwSNotFNx~SnALYD)yB7d5Yz6-KH}q$JLSiEp*|Jp6CfY;Q?})rX37snVj7Az!J$d
zqb~38pC_AFGT6r<?$IvDe*-cOe^zW~SyBC^I>Z^GrD~&pSQ2P)Wm5k({qM*&WWw_6
zGD)n{6oTRza@-}+q@XzqlB$WnJj(okZYL*-8y!TBqIr5HqC$SoqDV!Mot@A)I!5$M
zZ%Z2#_)dDSt@VG3bfu$^^)B>&IZ)9K;p^Ys!}P3xB6!0xk_)DCy@fgWfxzyAx=!HF
zhFXaULX{OnJ&bINIn`_aK#K)$^+6()*gU-~GL~nQD??D8&kgw<Vy8IJ>ksKH>cy^T
zb4Bp0PWKYl?So>7FZP~&Px^%ePg+hjsoTW<T++f=_x#v<eLO*CBQd_vaxc2doAot6
z!iLuYM)ylf*^!}o?*$uej@RuN`7%S6v;=t7$c|6-0`59ecogJ1iIhx#QgE&99Fd@w
zl~{2>h|9L=TEL&F4$-<4<9G2*ryEf;$|0{WAk0;QrqYt2H3-2a)o(ebIBZxhr+il%
zQ}XGRa0cd@Q9G-(-PV+Vkfe^c%ceFD#86ui)30$?5`o*zmeB@CP!wif+n&r9vERS^
ziF40wtZ^ouf*pX-g$R6ZsfOPOs_!FFT2am7u2phtRJqi-Ca~PC{K3fBQ<n5B{-9;-
z(mgd0OzmAD6vp_|$=`V3*-sU&DVmVdEVV^Bg**hFwGxKg`wt4o%}%>f944*@i{>Us
zWM<fXgR<l}?sj7|4c7_ZQ3I#l(*GA>No+_WNpMY2VhA>WNE-ipz0JbTX_7YK(T$e}
z?~in4wX9`&%fHo;u3xQo;Y*&(EnZ$col4yvQP+`QWageFvL#G2=PC+co3tg{Sw4K6
z?k3*%_sri;$-7>me~wJhMc!MWQd+P$GT?4<ArmSE!%YQbY4FWD*nG|hgLsl}00y3%
zl&Sw%U7$cq1cAZRo<#k^ZUmGm#*fmkjgr|)A;h3<umKc(u41xX=(W7pl>bYA?uSF(
z9ZqS>^12DUoDu!ZDCL&QEiBKi_2W=7Q~o(F<G;S5>~7WNxRoH@+Hm8&8Q2+7D?z0h
z2<TOeP}h>en%pf+zFOoh_>!Ycz!nbbiJc0AoUlLpIr_uB?f+*1L^1g**eZP>TzBX}
zTr!cS%Vr*X9>sqx%i$h0!taK;Vd8BAap74y0@0pztIvE9!Uzs59oKqr36c$4C?{yh
z+v8hGKkl)ocMcR^4r9Jy<as4mEu3?m$jb!x4fjg7%9;xE<a9>cY0CEfQ!n5Catrs_
z_Czl(>MC83?>LG*rcbzrazm2{3i{b@0fE|T$!AI8N}#lLgYjWg&iA2EJ;)REWg>kU
zj;uJH>B5a}6%0@IoEx+)x{pHBIsU~vyZxvQI444)$>tW7Rj0qrB7Z|1uG7lU&S-$(
zO|Bbmv`sC{0YOhNL!9U`DvMMIv90=$9&~p+T5}f!p^NYfa-Y-uZPhXt&ukNv=S=|b
zN19{2+G(?&_XxN7K%0?vx7Z(?kF$BAw0-i9;C&WuxE9>#To&@)KoUegIiROvSHp9T
zN};&>6&<xMmBoviMGkpIEf_4z4=;^N5G*hID0)<WnnQiSZ-;H_v)ZrOA#kCHO~V}3
zEi;~-RZZKSE3CaG5iO|p7qw8-FM^pU#-|M=87aNAo(g7Ny_?j6u=2p569Yzg$Q{S_
zC#>@&#H!8}A)aagc8xA$7~1^mnTKA_eOefbZ7inak%>nR^M)n#3>&;|i;bYd#w&J7
zV%2t2&vqhZI`SzOY$N`-|Nq8uYVO`Zam0Vf5fQCL_YE|Oqsee8kz}SK7Ir*S7*K7T
ziEt;$aua_3j$nT8Tf)1rFmc~kc$vH7%i$3-*aJC4FIscshwdfID=Lb}cOm^Pr{pi$
z_2A9fsSp*CS^DzoK{%9j6Q;GnGLxg@!ek}G6?+pl?2mBiZfqXWid2!TG~{eaIQ5Mg
z7PZ#(ZC!_C^m74gH~YdJ`PApxh<2=VUw$~(Tq{j90{ItRxt*(TP+=T^@fNS>(h+JN
zw;8_o>wy|u<N5rs;<J?84;Il_lVw;V4gj)Wax99Y&=*vgA`5QkzzGeuGgm_>B?zs#
z0|3YfLcV$;!E2@JU>Z5F0V}sM&$(2-H=HTAy=cUL$x09EhxBer0B^a65%oFMjSh+z
zV-M2jUCk<2*iDX$@*8kwpfIl$QW$jo1sK^Cq+8RDEWEySMXgY~^Zxqv!<-?D*spC>
zAOvlIlw-QK(i#J)k7QDtVfG=B9XLHl8qZJ~;CJg;%%J|F*aMbiU2)ydgg)WBWpR+?
zD!Li~kIL~9BmrIlno}^hm$bPC1~}{b`D|u8tsCv20iX;4du+vGNMf%lmD61oga4wz
zhu+=FldSK(nlzH*koC70GYPv&2sIl)ouSv;^4tJhRcUHzXk^i|8-4}vQd)f+Sv68G
zSEmi<h;4GCicqEKvb)pqk3QveC>y#sE*|+-qJ5q^1K(vjgwQK9z={Z1VsoqAg`2`w
z>RDauHew6Q;r`zxW;46S;{{XSnI!9ID^iD|ee6z?e`R{Q|6(=dr~YbKU+iks+Z69n
z{_!uN{BiLs+maDs=zGudD{YB%toi@F022wAA|e#=_EFgUQ=6aG^9p~|OP!3D+`zo?
zQ>$i15p$xwN+wfBv>pg0a{12h0k4UsnYoAp>{#tW;*so@E?b0V-TI>md2Qg)#O*#T
zH!z4T?M(n`sYhc(z6A=HkKpYLhd=9;Xzid{n};NDF9!w_lr)HgaKAd9yY2YNSMZIM
z)&ej#zU_!Gey!L!tR^M4%TnNPdc^|iSW-rTj!d>6_V||*tRs^j_-mPW@-+P=JHTo0
zu?q&_2zILUpB+2v6yk;Fs|=>K#BWfYZlmOQzS@XP0GSNPt7}Cle0a+;%s&j8^(Kxq
z^?_%rdO?@#;=Eq7hucXGAz0+7$w=yUOjKCE-Wc9SDPhU4^rPma`604!So{_+zm%@%
zqo_)BFHu1NlcvURv~?l{O2nBeLM8m>x?}3uJKDM+rrh}j&6Vu*7`vLFuUSvd(Rvn=
zA_T=}?X`$;12Rln#Hz+(5nAX4h@CIv>C+`lc{Br#;3p~j`}X1$`U)4QAJJj*yLIuv
z6H>1?rOJ<2teN8x+97JTG3ENS(^vBP+P9ZbMEmS<1h@0_nTk=f*=ltYZ8yy^F|~zX
zmPC*p`h+R1iJOrP0$KyQ@*P;ngqVCc6Gf1C?2(p~HvetJTp`S9M}D9c<qDc=6|{rz
zHyFGs`2x1bk*p>KzDq<H0>&FYMs|2Jn9SY>gn?K<;D)_?{yanhJ@Se+lmu7$@#{84
zg5-{w`|REl$>O7i$y8<KP}omMqv|*Rv1WOI`HwZMIWaBZ7&0Chl9|*VSC?Ik(2YfW
ziW5~{C_6+@rCwb+4~=QnpdE$DzfKKZ2Y}P`wT`Q%sMeX8AEdJ(K%m99#rrpCv8VKd
zz<_ESJC7Z0a8`22G!fx~Bwtl#83RqGOvjFpS10)|@5RD@pKN`C*-K|EhUwx?`X88m
zi(+ye!}~A&ONnIe0f*ccqK#+{tPwU;1`Pdgwg`|;<I=UMT0|JtB?@h8V9{O3TCOBI
z9zS%bT~P2(7k|@mo1p;IX5VcD12!&n)Zbk+(5JBVGX3l0>$@$-F>RCaT{Y;eqqWUq
z)0iCgM*|T2a1iT5Zc$;WR;1y#dquYAhFO-;6^$91ZWdM|(VA`7Bv|S(tBQ+Fe{jFl
z6Vi|F>Lq4nS~Do`gE<@UyTP(d+s+fs#E=rM_NS==E{SV|m~Y587=P1V5$vme+aH5C
zaQ9nZ)7<Gr`HW)UCr|4!z+DEH2YsfCHnn*apY50zVXOyvNs7HtT9MhG`KMZ5RYUEk
z5G~q=!>^u}68jfT8#{xOFt||_mMXS$%8iWYB=3|=8%WIrsbKuBD{Qw>;^+?yfX1lL
z{6Y1#rygSvd(S6mHIJIP?5#{o=0r5#iEu`8l?x=yKbmqu>~gy?T>-<`SsL3s-42uL
zVs4q<2Hv;?*gL%dA9GFWc^}MhY@G6#k!*D>0yBlGU%eL737J8a{GNg|3HLLa;2g#|
zkkmg3|9w~EA`o1X@v=FLn7Ou~9dXkO*zQUq?{w(`himxq^I1k}Y1Tm;#wD)f9zeuA
zKOcSHjEij&w=+^$4u;ba4W0jsj|>9bp&IJz-U?NdAT_=&4EcAjsp7D7#(w_3Q^J~J
zQ?h1F{52R5CN*JBporHLz%-z_ggZhQhRMxsS)U+>0@VX>TXtwJy5&IMYRTyNY1ZuY
zU!zAX0@9Rf`>4;OrB4UZtj||KYfubZ`{s=)kC6mHB=9X?HhgHqfq0ETD70*Ws4(@1
zK}>)vmrJT^3L%6KBcf(563LP+vtY&3aD3}hvn%r_lU9R}MWC>p<T#KlQpyF33QHAb
z8{ccKFa?Y`x2Zt$@JAM4uF#7_!w`^l=Kbhi;%2mx3;k?>?Wn<pIXOnCK!2xLL=9$i
zq(qsIM%0YU3u<6!Q{fB(T&^}hs0_tea=7UAI^C$O7k|<Gmf@yP+ijVi-|6)_?Z*+l
z%qI~IhB6MD-x0!;aUO?=xKS^JdK=H6`DEf|?{?FNV|Mu<ST?&z_!R;_alDg{S(b8R
zOJJ#_FQe}%6ZWHQ0Op7`7}c;_r2~c_fIS8;@u1Mm3LM4GX#HzW;9YOBH+g=@t4$ON
z5>lu4YO*&=F~b28nr<pjv?R4(aHTd|>g!%iz#VhGzR>QW(Qa?O?6Z@>byUcC40eso
zI-7?2r<&U7Ryi8R^?86l>4L*fO@Kaej)t1Dvc2WRT*OxumO1x>!rr{loKi+Mn*xny
zErN#o`@J3?Ggr(|l9h7i#$t7!Wc>kVt%Z@z3*SgpCXy(mpYFQ5cFHvf^JdPf;J>FZ
zL5L5DjXcTfyPSnhPpGUt@+Og>HDf*7J{?;^#MgVXk$n3LRRbV<N$A(VfTYq&|GU^#
z2e`rjO}FgWl)8v}{M|8uqRM%L@~~|>mt&~m|6~z_QE)7uWTv{djojRu3c&q9rJa(H
z&F>boXMqp`Mdmi2-d1v@_fZ>R4$l9^Q4Ysm9jaTzgk01|P>yZyh66kHBNWqIR*!Af
zqWbSb>mPIhy0@X<^paB)lxZ6-wQYxp;%<eVYxgX3Qg7c)!H<3@$-ya}PCkWS^M*E}
zfEyQTVfbv3|CLU%b)_!YpqS6M%VlDv05ykG?_Lfs^v-L4;35pc;L?BX{eLulV_YQs
z`}H(Uw%ul%o3+`t?b^+Dn~lxJrm4+#ZPsRPuFdu{_wWCC-p{<5xp>aGzNd-*Zf4tX
z4wp7m5DEHqi18n#4YXATmxNr37xN~uiOj>J2t;vO#g`orprro@XBx=9Gt!%F()owW
zuJA3n@{|fm3}?y4zX3vv=57Z)!{?rS9^1R(LCP=xrsyF%COaHU@Qu*#^ZKRuNAQ76
zA@uuf8`*w~^{q8pT1A7w@C8u0dwksFs#|;bd$@O@#HUSm$-u8iGlge*rzcEf4mq=*
zKfLy#^F35DsH%KXDJ7e<*%;aN*`YpM4sdKj)_2MVyXfXX<}IBR82$Yr!g^bdhVWN6
zhs*s)uZUmuTp5&mcLaBrye_^kc#?lRHNJt$an9tlRtlxiM>mRYR_Y4<lt8fB<WVeK
zXHjVS)wqFplyZknf8<bilpPM=^xsw<#ou+bOyTYp%Jga-Liu!<tm@-pOH{+t&p5lB
zJ@;GHmu3juUkLgfB`;!N9nH%IFN;>hk+pA{DoK6VJ+;gHi26%4G_?g*s;d5wIcrLI
z5yH`$XqF&2kyqT12siSm3ELNfll}MkjCzu(H!+<ctqq+XwZP?)2tA;etft#2M6!d;
z5<i*p)Ssp^_$z4Rr!N#z7tvRZ^vOj?dn+IzKAJ5aS^8cpi5HA(Wc6H|C~PB-YP#9L
z2P4GX-^8OW7nSE(exRdAd}T?q9mu!T1%njnElm>%V`O%{IrM;ld-+wzAhIDKL!&Rp
zPf}C!Pcg|@X)BfxviIK<6JiJtQ0L!AYg0uNODd!%V0r^fo99(D%&rE+Vj2V?76gcX
zC+&AEx!f_%37(sEYhzyc27p<%cYVsnvP~%cvIIUd_13^7p7|TcDzN|I75g#3uv6;i
z;q_I2dni{sJg8uuXEm~HP=I&HF_&Z;;8^$`T2FIqr<BbX7e39#IPV!W)=XP?9!O8^
zk$~-)nbU5?szO(>Pqax&v)3Pl<nDzb`24F6Liyi{<6p}4?Ktk9&qCZG6H3Dl*P{fL
zc69^&QMzwKl~b<rLsSJ^h^^42cGsyc7R^})Z2b5#9y#mFv4pX11kW?rolUgN8@`I8
zSawDrERuXUIdPJ`%)xf+pJbK_I;<B6yCJyzeZkRy8aL1u>D&GVFHm<g9E9cwzPkXu
zU6?I0zAr~`QDcSs+@gHnFonm>@2AQhJ@x#4*wl#0I^j-q4$bVUw3qJJywIpzkgibj
zf@b7;(%w0;qf;yp?YPdU-+vdV$z1mE_oC@BFXJk67p<COs5n;f$JSG5LvGw=2MbWf
zS-u#|MLPX$cFT^)MQ;(l&dE9dV|p{YcU;xrq2EyG+u>#v$r+P4BN_TCZ&;XHR}$;=
z@Ds8Aa^|Eyc=v87KKH?lR?AGlwR2y-vlDLf`&9@fn7uhPj^rB*ye<(CAsA4I6IN2?
z-}Nx1u=F6}-=XI({mbp)_IBYG@|=L(&$5n_g~~w9C7%=Qb5ZYVjo1Q0unTO*3F;?z
z)7qqQAt#+h*lIH;Xf%QCZe+Yr6<NaWhSEpJ^uyloFyO3GFkgPGC=fkJZYzO|hzo&|
z6B<kM;!E=d{fAVzC~25YtZ9N}&|OA7{{5CJhBbdIfKr77I1@V;cX3Tp@O8M*jl_U@
zPle=ab&wRgVVbYR7Fo$rVZ}QhtC<wFAg!@gNigLJ&EcL#>BxF2hmSL{CCUT!rS`sQ
z&G)c}Cte+w-V8p*`(scqc#4&9E(9j0d#|(bLeAg*5UzFrE0t)n>o>dudQa0d(KM?|
zn^7s~!y9g4p~I(R8%6y_7`pRa>}lmpA_!y>Jo|CD&5nFdGhAki{yDCpB^)bhH>H*I
zP+c+QTRy?i@F&OGD(_XdFY|XVnV?Oh5{KZeCH_@|Bi8I7ctFqHBTH)f;_hRhP-Ed4
zybKL{U%6tol=&xWWC?RG!d*LS6{V6NF{oZMyqEs5TlY32C9ueaxgVWo(02wc^=mq4
z&f{q!sCcc30Nlw!#vfcKcxMj^!MQNLaq~uQ+fQmb{2<A!#@e9LXBpsUw|&ktb>T>`
zhO+j6AxsNa`u)pd|C}DQzp=E*G$URZZ6WJg?+53;vj^U3MOaK19CB>c=0z-_^5q_m
zP7PyVtN6p+Xuc3F(3rMp%0BGY#@X=dDdF;EQ#v6L&QZ3|nX+$cS^ybU)EBI%Z)HEB
z!`_!dNW0RxcO}VyuzdhS9+6YrghUlYy#VnI>jYd3Nb&sI7;QFYF6HCq>Jaa=i<<Z=
zcU5Yu<&aa#1)Ek(`6avxh~nKz7C^`)?=)4A*DboEXgWLi*x9E6#r(h;mIXUzaN!=2
z_Cnd7V@6_8psq7Bfvp(Hx{pbjpARQiT{*MBjc`uwB2BQeZbF&u7<@ML>uywLsKJFZ
zn+%OZh(c^-r<x5IE834ime!5m+Jmyv_(GsCn%<|31m#d99fYfE1C@^p$J-k~^GuaV
zLD>th5end0w+X%-Kt;h+DGXmM{xef;Y0yZy)L$z`NF_)VP^&3VBSw@dj3|7__ggT3
z(n>^Qm+DlUG=z=U3$qy}NASb_%A=TRVc2(R<(KuvY&iSIqm8rk2b7l@@CR4Ml2t&A
z^H_%|_Nf)lTtKe6mRlT*BiFG_v+v21xlUui7ipT)Nvzt2C%j9StVhu{rBJM>Vc4P(
zDPxezA{UzK>G4)b;_m?ewwh1Yy~jv-3%-H@HJ#hv@&{(f9_wbK+lEaKG);I3{{5WX
zde|xG<%}B$xXnSuXEUL%ScO9fn#?;={~kfXwS#=C@|-m?ZG;B2Ike5UD!#E{Sr5%y
zy1Snv*7{SgQ%64N`HmHf!HGUdh(KEV#n2e><q&)k`Ns9Hr>N(Z{GYRl>~}lfY9Xf(
z{olx6Uggoh*EW=TUz)WxZmvSf!*${IPX*<|3lOJIm@bom8>b(om&Z+h@3C3ENakk*
zl7KY)yRkps(q#>{Rs9=K#sst8bR^jdVs2#t<2)4!yJ^Op8DTL80)mcxp1=DeBoX1^
zjy^~d(D>>qNLn6Tqd|(d#coC4%5@LGtDF<01op$)z?H&zm+m;3mYg)9)<FF-8IsNY
zZpw5dI(?&ldo?J%MNwaz+!;FSBYY~906D@O=+A4O>{69vR0W#gyu9fI?zth6Kt@(n
zc!~22`ZksCuPNW<Ls>p(@_UBaZhTQVZzwz*DqS_LzF$AOhGR>-lqCm<Yva&_{^@uE
z*M%6rV9RMP5HfL`ECs=u|5&jE;Rn-51AuUAhm+Y4qzc35-wFy~61%{=S;Sd%zm$>a
z8b`$sG*~ixX}QoWUf!=+A<=gf23Q@ukO&1ZYjk0y4>z2e*cug=immxm;}exofAd!m
ziU}9t$5AMQ6PLSK?)3iU6jS2AuK5(xpaRtCB6D|7+q1**hlXFfvAcYEgZQgr);UCz
zi7xpSFzZBVw+zhveOCv&ofyJNAiYHA4{hIwA6hsldQ0##%LMTxjCKO5N&&bhJcdt#
z;hACbF@GRm4hqGWRw>)+t4e?>S9SRms=51zg8fkyh(20a{U?G20q0F0c&(Abl(<dP
zolq{869(x3q`@SpRcT`0?dGD7TY!Tj0ju3lP*DYjU`{$jwpSXSzzfPiP9k`oN_J+`
zA`p*zs0EytdL$yhs?+r8aDmmUAKL3oHzqcMO#7-+Bg<<rWZsM%UX^e`*~#zqXOr5j
zc0XNA+Q!l~*0bd2=U-LI7gp59jSnlm!ipRJy&7<b?5#FkaY5JZ`W-KxeJ9?f+NW%t
z)^?)o9c=%J*o*WL6Z=*oWH047fhJ-sY0+0-1@D9*DE!v=&_uj{wH?t!bxrZq$&GC9
zbT2VqF%WbF5aOy0guV-aVTya>Hx)3+9@DUs$wrLLvev>6efn0-H<k)HU#{GnE{iC@
z_<p@GsTHe=`?A|dY#^M|hQ%2vyWbD5ey?S-Z!5aErPEwM;EsEUOUA6=Q_4pE7*@qI
zxZlsgBbO9PxU2Ee8t|^W3~KQi*$D}bL%+3y?y~)kTVcRP9M$IEQ1`V_Kv_grK@NX`
zBo&QFDhR^_M2G~%q_vR{ESP)#%;=)@hLkIM9PAiwsZIhM!N{fAz6SuX#u8?lSpOvL
zE&tbN<w2{ftHLuI8;auEjln&DEoNI6J-u<RssfTDuma9;Sd<TDl1+e=I8kHp&C=%}
zMRCd~PEM<d0>#)TrQy6_`RpcsLQP{C-&sR`jtVSgbo^Ig9&%?-FaA<2=k4%rRt}o6
z>l&9H(~qh%wA{rO-j`XHbLsZOMG?DfC-f8DOUYAe3si}5{hVws|4N=2){}>xqP#$O
zrg<L>#{fNWBf==n5BS6~^QP|S(9a-VA_A$#Z=S_MPuhFr&>y&u8Mc|9Fb8Q(Pmmvd
zh1PqP5~Xu>^-vBLf-h(jNCfT?a(~1nwGh!UFT51z33CtcG19_x--n3!CcwQ5Fpk)q
zqd)n*kIdZOY8K4Eb~f@mAOv*wn~e7rFPifjY$*{+Jn-WxNm7<6=L;jRy1yVm22}<q
z=C54=n0rNhjU4u=ksHDw6>Fn=6W58|r9OFaQ>tZO42KO6_+UC;1ZCH+iML~52QJ=2
z>g-Rxqpj^zrN_GMAbM@;E9~mEaxis!#3h3toMMZ!1eO*LJGqH?mSh7+i`WJZ^Hx^O
zFa^i3tpLNec+yzEqz!TxWtxeXft82ZPZeeJrvi^Ua+pz@h1)<iGp*q}dYWaIU1nJa
zAgXb0rmRVZ8u(!_6M!(T_d-^nNaEr`S<@!fjMu_frj=cpx*m9113j|&ZQ3?J?5|^f
ziC~-|Zhkd2E&j=;KiWKyu;_*7A_n5OexUs0KB}s?$U=O<XUY$$YoH1~{)Jjb<#(y>
z;I?i^EgL@t&j#d{3gX%lrMn|}w8|G%Yb=zElt22y)D-97M-kMVz{6PQ9wGuT-TDmr
z;`bHLk&P*$3j^D+5N7WQ`<m1dIIH3dV^QlN^G5nMD?gj4$btipER2Xk*O54Y0Cqde
zSL&x<Af*qec>nYIBEAvO^>Yn<wXjMvvD)K6*5ylxnMJ^$-sW=<-Cc+6eA!0D?nl(a
zMHmTMRny{#+SB%7HwOMrSR*+2#&WAZi;HZv^;6--3uePs2+xJM;#w!vc0$;ib#TuU
z5*`b21H=g&OHt24?3X`greQEgnHOK`mE&5AbNS>%8P8PN+ZQ0zvLKQ)g5yaTQnb=f
z?-g=*{SO7&1GhbxNI*hi&j+;^B^2!{(HgWbeKl$XNLIGMmU!$=d1p?#a-R&#FP#1B
zcsjE*M$$i|0C<IL9u(eq%TyIygd6Y`S&+|l!@S4*JSPMt^Xl7Tzm$IIv0KLxFh=k8
z$(6%?oe3tn)27w7B`keE2q5iAg-o)Gk?vy`P*rFH$=fHpV7c#|etTh^O?c~=$#`eK
z$V=1y<o&i_%S<PEZ=#h=?ui~1wZ)?9oTTp_MX1MlKuXt)K??4Q4&_aX1f-iLQu(@?
z>vG8mP0PstoFBXwaC10RTn*!*8TXrxzoCHh1rUH=&3>F2CLcSYk!gf4>~hWVvwpbJ
z7}qrhebw)pJ8&a_f|Q^NSSW`<D_4qhshQ7vGvqsr{1{_2g1`iGK~bQig$LyP9_`lq
zDFaff+@t~87*2yl!yjKCOYEm?V-Fv;L(f-YLVHGn@-e$t*Vfiv@;+nMXcLZB)zAmc
zNq0Cq>-~wzP&*E(J3AqH?9zT(VU7Wqr1X3hM~;*yN@~aFC;6aYhbFBv?Dq3cQa77z
zh;}o&P#G-bdU>9=!TKJN)o~s?d3zx(!EvP_LmA;X!?Tx`zcENIYM-Q@Yw(6=zgJ_>
zjjr-fT>k+*Jj|DmxaCs~vA`m_+>10}h<jZW_dcu=vgI??s^E$IZ$B*YCH<Gq$+s9q
zUrz-jzsB`?X#^qY7$Rv0^7{OGFgLmCO2Je|R_XSZE}v92xd&fmToC%UW_?+;shLLk
zEZkYBf0hA?ZZ)rboe%zu)4%94_20~mzK^kgkVqFg<HE9E&+10T3Uf0V+2Xq+4;_Us
zN4imLtu${of`yBy7~qjihIDdT*A=-f*vnRPWJ&OOhg-BhA)x4}7KCY8{Nmz}f149;
z$GH$6-oE}TcUwXpu|s_PQ*DQAk6Q`<>x1_N^UEF#f)fcxNgHCrvm`Bjf4@|9s@rp`
zT?)r_H=b_ax9Yj=%pyrOw4bo_PQVaulOYto@)1bJ&yokt)3AoFPJ-n!eQ;k{B{Rg;
zRae8;r#!yqjD$T{@w+c;z`Ixx1j9-L2(xWmcbAlSXnOs}u70VLLY?p)@QYQvb^PYL
z(-+nX?at+}M4U|9+6F6oe=0ZEQbDB>{SwhA_*yhQ-Xn~z;s}7z6h9Fz`Xda#$cy^9
z650E>G{a^>i}OFZaXX@<ZMqL?012L-ZJbLMbpJQmuIq*0#ulikriqJe`j-Ir=;2yg
zO3JA)H*u2$GuZuCiGAx9LEsVVg(`WFK9zl2v{4v~tr!@EhiwBz`!05)2>#!2s|*SD
zQONdcbWvf}%}#cX!=cKdX8uncOge!J(iwj}_HuU4-ZNg4X`|E7{9$)~0#9R-v&@S(
z(Oiz+&cR5)$;I)b8d4f2{Uez~<f9oc2?>d9;}(fhKb7ElJeHYtnL-HN$av0D-Cw{S
zrDe!9BOKI(vuAI|>XQIgtsa^SDfODGK(vLmddm7Nj<tiC7}=WtVh;dU!$h(dJ1l%P
zA~=4_c?`2?e@yOJ+~z`Kh%G`!Ec=fIe9&DwA|p!Ga!ChDh^9TpA@Dn?q5a-a1LhSu
zZ(7XWLaI#p>t6}8`g%kY*>MU*s7V_?q#<s~&pFnR)SdYR<<Kt}D?4*YtYrHw4ixP@
zfm3~C1AWO<-NhFbb#X{GT1X?)N9StuU!L;<FM!Z?|9bG>M@}ab502H>A}b@swExc7
zj!)Bcy%F)d(M`U<ss_w%rjY%Lw-U*0&%&4ZQB84o$InqosSHtG9fOST4G5QKH8EH5
zFL$Tf$f;E3Cm9SbB+iUie|LKBsuKzJj{XAMNnG`~nJIW0&3L>m?<Da==Pl+MAFyHr
zQ*w36_4M@(ndu7YxK*ia2)*!Ent0V_4~z<I7t2FfUkcE?{Bp|#?S^n&ibDS|4`Q$O
zDBp-xcK_9Fw820)WklI5v9z<<BI*iIwaVZlCL!RmnW6M?EobG`ITZk%FY*!3b&8g|
zcSj-er8+B)G%}EQ5aB_=Pqw$<6?S85hNvE{(?c<>swd?3sFy2EinZV8z`y0{jsN7N
zI;8%f7;|D0f=Qqvma~^}cqPB2q|HP?Zh4AHj#cRnCr?h=cs>9c@6A4coe`{TA@ue%
zty8Nljon5RUuh7-KPXL`<A2CIr)s$n5_+ikj<z&F!g}4hog`DNont1qTOWdVZfujn
zz>q`zYY_+jOXy0d(>4XEtTw#dlUb5>FlnxAKFZv|pl`J5eNQsG%J644NK^lB`!&Sq
zwR#dM%75+}7^Hgirlb*#GP!;Pg)sJA{z;)#e)kJHB2I;rXD;y|+@IvjE}S|I#LZR%
z8<iL_di5diX3+Rqf7$NfZfQTNaYDIGV7;}7G#kAt_N8zd6ybvY4H;kh2BMhAF#VpI
zQqtSLC_x4a{9L=5s-oYSlS0rycV0?E)p{dZaO5|wicP;G#$N`)4^pi<>Wy0OVg<YL
zzhNn>ya0j^MSO`qL1C7Fn>hapJUwM`+so)=#i4-;$e%%M`e&e0`+C1>#uqYTNA%So
zb^8hJiA?s;o}KGQQW&F~9I}_0o()4`uba-Z<73^R^;KV#UYU`0rEoEGJ!<q1vSC}f
znc&U(%Gh2BCG^!(-@>n#%i=0ORzoNrX><XMJj+C!<nYNIyZEtB=6EhY!w65ndIB6|
z%W{!KX52Frhf`}a=_%-$9<cal>I%Wqe^@G||D}roBB8bZx~GQ#M%XPwXwEg!O}`IX
z3ZEm(k|zx3TKy`^u(!S1P$~bp5CqfLJ{e(y(C7tk%8cpr(*}LYDRT^F5GRZaz|ADD
z^aV?ctCLGkR%^T-y?@mL-n+<v&?^JCjTG}QgItfe!9$ZmxxYN@uGAn4uNf1mYiZK7
zNPLNhgh`FO+{*&kIl$aB3Ck3O|6f0*3l9oMm=U;AV-7Ni5qm2x1>W2Qz&ijv?D~#A
z({%IJ8MJXAo)x@d?ubC<KJ-sOm76X^$L+%$Y+Lo1;VYpl*BLhYH;m7f-o%qaPQmVY
zpNMGipa>-M{?L%_-!%ly?joKWW(It&{gRT5I`)V_%i4JGP$cz42xF8gO6Tvp`87Uk
zl7KWM!{>F?s~+lG;lRXkv2;SZph`K)03kcD#!G-bK=Uq#$I-8*-ZmA+suUSWr(d@?
z@F@dftPZkq3X3ARmF>_f`mPdOmU69RqF3b9l7#xOe~BD-P=qWN@dLP;HIt`lO((5l
zLRocmy>(UKrY<Q-zYuBC`Jcd>W3%d#XP@Z8?8FTgS|Ys3T!AGai)}&kNQ_I0?dCQ7
zT^rgIsJwM<6Pw~_LjaPf_o1VCSMTv(IH{irzaQ@J-;MCP34_t3v)fKSWROm3rM3kS
zNt<4li$_yyxxfvBKpl1*C3)cseSmx!b4w^fpHU0>S|piKLAV6{T;rwAtWJMtff<?U
zvUhlMwQNM%1FGCc1$ysTYapV~+;bp!a@%m74%zf0Qgn`MWlcgGH-h`CgM5r{|1V4o
z`kDs4HijXzD<jcp$Ec*q_>7fxx}<T`pnO<qrZ60!LVW-Ka(W%!VYRUm2}#dcy8S*%
zTp=E}|HEM2k4wkLje1%VeE%6Xv+L96b0@GMTgQJ(A1F8qi#7n&u?X`^7jVmyd-suo
zp6j(ZBMEvASq%1abPZg1XScOA_`v{GZ8%DgaCB@V^9}2V5M4&bbLZ<R9+cn9=0(?>
zDPxaCF=9Cnh4fUOkiP9!Y;<!*d{<(EbM_*Rl$Wv1UlHU@7Edp4i<kQ|CMqJZiOTfH
ze3baAn-M&IZm4}MU@ZNB)<S7pPjhqcWT5|n6?H+d1h~5fJ{&&jVPP3t?r^%;@cI)j
zY4})ht?s{X>+rl;iDb@y8T?!`+}P+*R|-HrX=j`Od{FXlFUklO2<f8*oOKaw^NMnA
zZiEU{#JGWjP$*Tfp$y{AH;iweV<Hk>$h(<aSS0A$h0bruZ7PSMtdpAK;jCo_Zn^ql
zww_lZPZ5E0;DRxF3*z52VBjKrQ$>zh{+qP=Rai!88_fCz@?|q{KQrw0@h|mZcmKRK
zCS<ms^-`$7uHEu&*HX!(I@X;Pg-++k(B-I3oFf@E|IAMcHJxo>uHE$7-%eM{cl!Ee
zAM?j}E&oSbKJDMk)>9Vk)ebde(=5ZZ{)K0u+lX0Sq{n4Qss7&@1_ZY9+i^d>5=Aa)
z-Oh&YJTSkm!F$JMj+_<n4i|bqq$mp?D}AR>iu}zEDee>Si8KNg&)o#SSK*jvNLj)I
z9VZ-)vWO=$jzD+lk$rVVZ|bWQZu@+g-BwdM-t6#|>ZR?7IcI=`pU%$`R0Pf(6q}O_
zOfKohNZ&sZc?QG{dDvx8@XkZ*P45aAzq73hV>T~L1=Zz7(?9CIqyZ^|Vtn2OaES4>
zzJb+}OtsNKHtR)UE}!;K+e7dWQ5O=2skq<O-)vpPxnqO7W#c3(a`SWtN(ggraV<pC
zcF?@$TLN8+&X<8Y70Mk~5sLwCoVfUEzM|9O@>c;H_zSt8hzAS!)@}xA$8>qgJ6sWa
z^x|xB{5^<>g75D=Jl~6Ll?)2m;7xMZ?Bk<!{nq3F_ISw)s#f8sVNK}7N1hAh($C|=
zv_QJut*V0F8OE?WQrfvc^&GUAHe(twmUF)ac-YBItV%b?o@K{ER_R1`5C;mD3|M1t
zllu>GKC%xu;ZuI%Po<w{;3(S_c?#0Lu?~UOdA#0-qf;%pZFfX9R^dea#j0Q3TYoy|
zUR)jmVKFO`(~+^?4LD|>3N-)8xGmfkh55`=>QzfmKy_8!{*7(>Uu7=f2Z2RzxK+G@
zDola`5n%7#C^mBy=DzXp@Nn50Y@1w}d3rub<Ksa}xC|~3;xg$^aJJnc)J72;=AZYW
zcnm{BRV^_2TX53;y;(QN#@1Ff3s{jrCB}7z{OqW00vmwai>N9NvGwMmJ$Xc@vLmnz
ziS$wgfjNFmq{$2zVTlWg9Lp*bu|D#bW(JeZ!tPryi2%_#3akG7n{;Tz!Fn6nOXR2A
zp4b5I*=0++9cX^r{O1AVh=RbBGn=S=l0C)mR~e^`vQo_||46znjO;(ub+<27x8S-@
zM53p?dhM3G)d^2}W?!X}zMdCeCvI04bN}0o;S|Bf{Opafm78EbA~M#T1Ne?1`tcWb
z>hPw1EOsJaBp*Z&$%nk$zstgz<#p-b;aAAKRe_+7G6cIwdvv+=4RyYe#k+QnTSF;?
z{nYq~0^bnT5ENV}A{_8<TOe|df_B#klu!3btRh-T@R`Hb_JEnY=>caT-Ob-<?7j+p
zJ%LO!@MnW%fpV@KaBb)XeKD=<LV5Iwc2M-9xdCh;h5MB{Hc4`q0W;#(ow^8XJ@x4y
zmua+P0JpcQph2kDOw`3VJBsVfwbg`n^>=859hPV*g|>jbuqMUEeb208VrID>Sk0b}
z0LT&4|MGVMIoLCpLou4hyy%O#R4$kV?C^}f94}jp**#{~@HPcOWohcBcjv4i-5NS8
zHy{cnIuIKBQ)p>;O&8cLx!YwpNLta92a>i(d$cZkPc%7F(>#(>NwSR(H%Wm49WIfl
zna;Os=l_TWlflH<F7OUwjf}xNHm!}p0-uS0jG8Cm@~jFLE6&?fmZOHuyz@fejnc%u
zo~|yUxLmXJQL^qu*1<Ox`eTi~V7B_2u%akOo7;s1t<e@Iyob^Va#IkU_TA?9su8u4
zXz`NMw)4Buu8QdC({5H=T-dknn3_`tnPzk2;Lqkmqx1Xcxbi`kpJ_ojBA=sXqC;_;
zo#HO^RU4Xt{iR5WOk=`%28oa*xl1LXs_YnknJiZuW5?@Q0!}IC%}&0VckK;)=a8^+
zg&%YLkOSvF8)>+|=xU7X$PSDL2z$<wM9F=hLk-sY^Hl><?HS1$J_G^;I<HgYuP`E<
zjKI6<Ue6cx=N|?v($g#0VgSpXc;z5V(pFpq%`NHeaU3}b-Rr&|i9=AtA3@ipLbiGt
zRHjVVw5wn;mDr(}od!Ws?JQaC{lM5tUTw89nWx2|G|ovU^=G`+)QDyCQ7t<vSXfCM
z`QN@XMlV`5Op4wbGO9*^@nq<Nml;c#5T)(2^}s_(fuCFKg0$V!oHmbTL7-eg@z2Iq
z1rztEoE$Y3b0It~41{~;)<dFs0Pk5}nOhRQJHfaW#<<N)^=ieh{hv)6_jodYtdmKM
zddpb+l!KSZ;%PGrr`M{!{J^Lf{m1cNODLcRmOPnmc6yJ`_6R}Yu9?(@mR8~WJ@mn6
zYe<B&a1~P<$oOAdgEPmmWPqV4l!+>!f}L}9M@Pp6)T!WH)HQ@Ps5OTO+`*B7)fDPm
z3{TNVJ890~AfEgKpv5&f#UVdHgo^D8O#;ly{?T;hv>6sU)PnOIa^b}I<xyj=Y)C<)
zw+DSK5;1lY`seF&mOaWivdBqvLDEoXSHZ`U!Da!%@xQX4y(<%grWBQ$JqEhCymcLF
ziFZ0w><gr4q~~Ab6yZTL!wxR1_*oYun(yc(cDOKpBp%<<bipQlVGuC?tXkJBPS%T&
zt;kitmf0VGTDRsihOu11&iv=9=!GP2+x=?J6K?Zg<Bo9{JM)7oxg2o0ZI}O1nO;F6
z@uwvAgNZ<xpUeOUlF1){B{3aQl=sdMdR{xe`!>aLj4Ilw2=Tc|!zG&3oRR$nngkdz
z<7sAL$>^g5TDJC+YxMJ3q|&2V08YZD_ZtIKM=<(jz?KAXD3<k7hqiJ+S?F$|_Ayye
zOYZUhJ;?b^B9+mP7j9?C*1w3x<QG6o(S%t-gV4~cN_d$A>PeFuUdD7KJ37FoL^2@3
z{XEfeg$|$NVjzH83GE4spMeTCxJZ=t+6BhQGAe-lc^^m6O0HVv6Hug-hRU|&j@SfD
z>k;mk@hGBod+Wrhrw_N|FKNnkiEXn(m11`~#UEOXdA#(>&JR@uO6OJ<hz5OO_-M45
z{KT%-)>iSWj#={8_~*}n3!NKT`k+U?|4>~jAQ6wlG8cp<PbgyUK_Vzf>EU`Wa?MO9
za@}Uc4V9hZH)%-441L+^JsyXgg+=|hF>as-xky}8>;Bpdh8d4D3BjeVL|Fz25H;%}
z*EA-hBO(+gJ@GV#zBNv-Y4l{`EnCqBlz$G$%^_tg1`SkPyi>5+NvjD;Qxsy7VDP_6
zN^Q6IR=O19g0zvI8rL}TFy|)R3XRs_R<_kTCTOUDCFHV>CE-+kv$x_wjdzM}IzGrw
zl&ru1;OIK=)Rik*b397vv0YR#iVR`GDcb;FqVWaQIuy;Zcn;A&S9kWFSdUHD#sp#Z
z7O}ruJXA`Nrmo4L34GGL+qjAC>cg0+F?LBKDDbNj_aH^cqmG9Hfx@bXF(MwF%J|Or
zy_5)*j#XqC?`%(q#PJ0imbe_Xg<3uaA0!;n=$*+>{Pq|`Hi9IlN2%rYa<8|47$DUK
zkZ@#HXF$<#3MSH#zxvdxGZ|*J{H(WK__k<vOnt9wZ*Ony?9AG<b4zfWVk=`;23auQ
zlP&3w@YyoT$bvDAtrkM3^93)<DjUgk?-5~be0;<RYb0icmys;Hzxx8o-#zH(VgwV$
zd|_lK)|OyWks}v>;t6>S69I%iw*Pju3ZnXK7QDUAf!XR$vvVGNRSy!dzy3@*Gh)<R
zJt8cg8$7mT$OLGy*zW@y)Fb9y>c|ooT^vj3T^_t>dNeH>G717dXBmy*#<3UREtxd{
zv|8o*>mj?Ns67bW_U;I8Y8V;(R%4R-d{J&v*;B5#QO23MTnmRQpBwvWxc}Aha(*uo
z1rl3S#E`>lUyQRI<06zt@&>uk_ZSmfb8C<TXQp)0Z&cc1ee&C@daKb!-h^M)(`6jm
zL}GB;OR4~SxJS8HSsCZH!M&w3J5p_PK9Gd%FNTw$e^IElzB@`Tul#&<Aq4JZCNWB}
z+=Q$pL+C^vI;R{FW;p^te}iJn;HR^MUeedx1Hskx51bFA5C9kP#XyUgFM9Enp#eIT
zlY!~gB&w>d7$Amcb>&vt1bo8_q3WOH4&Ik%L@-w=e<68bdUJ7d-OnYlPMRPNqpotQ
zfeKPw9}vjN2EXBdqW)zwTjXJ|+YC5OLQ=`oE@aJw%>WaDmdXkYKJC%LlWg-T;$BUK
z7@-;Xq=#?e`h?`yzBVrjUnzVBj#0tHj@-NN@A3Xd%eaTOoy=V4dQU;;e6oEPe*wGY
zl0E%SMwgWo!yJ@m`<HP2`Ok3Q3$%iUfb7eST}?kx$>zIuZ;&3ZciH9y*VkwJ#cIBg
z`eT+oQwk;oEt*Z0fj$}~P>CM_vttfDkF{R<zCK_a%7%CSD^O>Rx!4gDRw3@Jq9gST
zFK5KPm^Ln1_8?v&YuKa0CHP2=upR)^!Ix7JZXUobp{93`rDV3xk@-QrmElJo7tWD!
zhJBnKpKN+bk)5oZ;cNWJlHoL6Ud<Vv@Dzvu`ZDG-W;340tzb_Y;s#n#oYtpNJh8=!
z_%9OJfKd{Flr}Un^=~G&?18urv<`3%zX;SWs%V<It^(1RCwPhAbBv`5eqaQtoFss^
zuY61{S(e#t;EsF<vjnL8L*#&{@+QA?hC)u)>L1lToEGr7bbIJ7<c0X=6v?FPrLysv
z5{qWOPr1K*9LZ=&{x;MWBhrYOWWeO1`c^$Xd;Spi7yb+bfux}l6{;gnnur(5f@v8Y
zN&owh6dcIrW|GH-$SXxY7-J45V(rRsru}+|Ty}v!%U2%yaUpF)&s-&1nECXl`<pe7
z^QJ;Si~}(ULM_kHD(5x6_$=f~RGhZQefzG?Fn~v#z7mBB1)=+7%+U~D8jDEh3V}Q@
z@Fs7jRNeDKy^#mI@&!=VEC$+~l$$t%E+1+gEuYdsXSOM9aD-G;e-H)W1G1D8se|bn
z2Ck4%g@s3`1IhiLFZ_+ilwE&g4`29>^<YqJJC55r%JJ^TEB4_}ul-~q@l&4O&d!GC
zSn!zRRf-nFw4aNb{82{0IS(D&wvKR*Dgk&%WvliW7@KZJUQzxorG8Ho!K|Jc3<A$g
ziX@zSdt4z$Poi6Ng9&#SXEvB7ZbSaw#)u5L0Pl*63Vq88)2e!Ph8cCfs$p@zl(Jp(
z=_>6+f@#ce>v4;t!(kQ}KANsc`Z=`aIh$H-OicfEUA}BxFg(`|4$ND}t|vxc?^07p
z-z_)Ln6%&GpF~KmsU4BQ;+M$+m4u^d7-+`klgj+)xeE>($*B-<!Z=2A_%(4Dgr&{?
zL##xtK|D_DURIC^$9V{l-^e&)Hp$U=czPAsJ>@4oLDb$5pzZ9Bq>Uzw?@Fi;{~zTI
zfXXTOaFLfErvUrFtoA{C&1Yy>>;a1d4Khg0jfvKE{$xELScK}$5!8#@SDio|H#2Q%
z2}{Xz#RiOmb{TUlen-+{D;yqg8~;J0QJ6DMn=Q95$+eWK5qQ@)m^+600Hj%45*NIG
zARQi*?|^P?q^+7szQ$c{dt(yI-(4WHrGC`lQj{+n&)S%%>cg8snt{trb&v;DqAwL+
zldg^udargzSR(+xFq=!eFi23&?i81j*bnH{Kr-r>ar-QxHA96VGC$%%ms9E0^&n<y
z_xv_Ie!qTAv1#6&QO}74ANrakP!Jo=N}o2DkVHO*{|48p?mi>lu3!F}C55sZ8Ntk`
zV5QY~IvM0XS=AlFdlhu9>UOc&^nhkkob&9PGkqE)<W`Y&fV~^f6g?!T>7NQc=tq^X
zz#Y?AQzh5aIvufx{zg?JI(IpTj)wD_*=bo`$PE8X2OYaOy7S><)B&aQ1TMHPI_59D
zx?eJRjjRdFR)QEgL%>D)gsN~?Oz@jb>Wle)!%dB!lVUyF_6_e)UWRXa_qLbG#Mms-
z(>bPu-V%Z>Ggf-j#OUGSZjW&J<=hi9dN*N|<tYLA%@Q8-5Ku>RXCw(iA&UBf$Q*73
zX%qr?*rI7RD9HEAX5YhP{}zv45Yv9R-7Ojm1b9FKrQkPu0oJVz^j5sXysRO=i3_ts
zGB01g1W%e)c-jom18={B0A=h)m{F?it*k_vtVffi3E*UxagbzPhPR6C;L7Q*HMeB1
z;ptBJtBl-ehm=GX<}w!@@R3wT<+}o|c^V!B>e7b+QU#U9#(y)$%%T$Wo|7sXwu|1f
z@-)5!fERbIS<x@!ZD+M^WgR&<VnuV&Mg0Z$sDwP8Z;$a7qT<?V<l)F%F7IesMM(JF
zZ17fZjiEChLMtm_VsnQ+%1s5X0-)ES!3z(@D83|l%n)zVMS;8#V;<QrCJ^VfMm)z)
zsS+!ATZfBuyT@)>FyNo8Xqdff5vXufs~Xs&GH&Om$$Z{Mw8%TD?EMb94q0bMZXkDl
zho4ft$W&};j!8XI$+f-7iCh4iH3rF3`^)unP=5o6zQ}PVi1cv~+Uqh8#E>Oug?~T%
z+nco%KXS#s2a5xptVfOT`qJA}2(2LK8>Wr$jyLvTHpUZiE#<$nFI0pIZLqc<880K{
z);7*Wii1s@bjm`wuJemlvzwd*{o052qDbGf8rLOjgV^~#|F0Lsg#yCB=R*{|yx+|g
z{@Dn6U^i#H)2=&61mN*`mmr)ePI}#G^^a%%8i5Ci1C|nm<L)o9luB?PZD(&R$S6?I
z@f|-Q<QwgR+p%gf2OR<D_Wgf9qiJq5nt17$J9Cl+p{P^SaYiYPNvByiTAxQu^6lfR
zswTw7V{-5zDAq($6t9+qsT^#E`<?MRXEkQ}SrS=RhPmX8v=ol4X0uK;q3q%G^4HR5
z+&d!R|Eg?QN}XI@`Fbh>CO3uQdBeCQWpD}4dV$*sj1?7xFSz*na0pS%PzdRuE2H+8
zHh5kA?Soipwn2+)_dvm5wm-A46Pd#K(j^BC>2>PEWPlEA^WAA2;PNIz1z^(^Qz#Or
z4>!Dt$lyz{sbG&5NSoXt(PGG6l5s@ER_1!>XaimS#TxgAIaKwdeCw@vIr@#luNsto
zovK}HMqPUGnTn#zWZsdCSHCq=x-NYb>PJ?@;wJUz`?g0xnd^2RN`)pcewY@TM<lIA
zG+=W5Jgl0n8vGAyYICSuwf|^C6TkMd$JR2*nZU!$N!-{Xa>C;iwpJdGAlbC#t)38c
zfN`c$=|@|2>3*eyMa4Omz-;oCh4>?4f3_V!1!1b;g4_#F#-yH!J>d6$$aByz{O#?n
z9}xQW4mD<e`+~N*n*G#NS0GNXnf&W8`6a!BRclHv_zNE?Bow%Wfis{x#l63WN=n0~
z<bzdWC0Cp__^btQ`r8&^2MWA}Z%6_ovOv;QxF%i6Z(K(Il;kRToP?DGs>@x}-mfbf
zg6+rG!~NjBE3cF__9zQO@3kg-WZ6;`)8HW<*9YXPf1u>hNbZLdMC%UMmPrGzf6HTR
zMQ8)6_MTmHzts*~;f0{(cw6q0ZPyAF8?Q}P+)4u+!wKi-cG|_>wys5_$c#WLa2uf@
zd|x9R=aZ{)*x)L@84q`B2oxv0_on1>Negz$hWSvs6_MLM0lpf4cNyQFGpgbxes8d~
z0;wpkdlQ{}Pj6t@w$D_Ts)L+=Hnsd1XN>MFj`M?Zvk{7Zc3!QMX`F3Pqa}15pku|S
z8dNcmmRSWaO0=3e_t#fg-jD4XFzx%DdLqvT)N`qkc#*7>OpB|lF|~c24ofNuZ;eKh
zE-+}Am4@R{k|}mECkU}zjM<-SM&={#FIP@OQoqXO8s3$I#AS<hDEZA6LD#i@Im0kL
zyh&+kI#A~be-%k5?yk;eQkY@YN48kc;!%ctcsL6|DHhU5@q_uvMIo3#!gx|=%>l0C
zv)Ze%I<3cF84s-M&wa8-OX$LnKEzR92_Z1V3_kqr{e3rH&N;c$EN*gH=g4S2Kq7>M
zE29C=EwM$p_)R}M$WJpH#`hPZ7cr3Bv$#O7h&UeuJ`^xad+tMZVquZ3S+H+7n&2=+
zOp{RhH%NLlmSTYL^2Md%1W%8oQc>OB+}*IQl7W)=$CzqOe)kwe?9(eO33C>|cJR|W
ztMvpnd;)?UrMmp2ss59OQYv<^(C&UZwrQoR&2pO2s$F&Z{kgu6o)dU4=u-k0$td(b
z=tb_v4G`D92p+C-A6<~TrT6S5*|SItqi^o_=syq2{qEY0*{b%&Px{Lg^h_D#<5uqj
z?5RD)lKXZgHpJMmgVEbgOagyIfk^j_dl5ld<tDQowVOK<^1el+GpW+2s-7VL`9X}v
z3!Q0mO$ZH59d5Bp6t8i0FRjnKns5jNCREGlXM@gfPwl@HXk)0#!q6T!3@!4cUH9Nd
z*BS~(CclYc`+D+;Rf8n?#<H0=l`Z4P_*|Ln%4@h;<%vvDBW}k}+YIVNEa-Noa!;lI
z{}#X}ydUdoVL$0|wA0o4<l#^10a{Qrum34*l4AfowjPw8Eq!woP(O~r0*=NZ0pRU=
z*cIKs7PcZvKRN}(R(AiVi3Y(ttTgP?)gHAYlq_Us*Fr6mLLK|0EQ4Ta_!M4oSG&8_
zoFlkow{UyKKSZ<F9q@yFC<~OZ=8u=gDJ+H!N8uEC<KU0A<slHn844^D|BK;6CC_BZ
zNICmk0BwpyO5RGz0oqxws`4g7t-eoxLkkGlx;nD9CEXscL-Cs};Sc_TjiaMGjLiVL
zoDMPdjF>&1!<2L{>HOb%7$PsUBfqb&VHg#NP(Kqi@O~R?9#ZXk&K5}yko(<Dk4_55
z>i5KS`*hP<FKB!A&*3`|I>1$Dg$|5IV9+H17K=LHx@-GCpKK#fuZ`hZC$==NLyh$q
zz5EmNs$pP4GSP%F29J94ksSqrQ6hmErE`vk;w4LP@--_G<5yxB_*XLr85rf9Hl2t&
z_2{XHf|ux4W4S@;<*Z|oDN;mXV2``v;90ev56O2j(cwhCx}PT$g?>U~hZgTp#^gxQ
zvi6-5o)>sec}fyiAAlNzlC~;}i*O5y_k0-WyCZz1AEE8}xb$sQe_wBtunv_FmcoRw
zAbMfQB*Y1F`TKXt<Tz#!7Z?3#L5uVr`AzwI;SfftV#xE<8o8HMKAWt}`)49ZLZV{;
zPnYTffoDV3BwtU!iE~ousti|2Vn%tQc6Jk;BH5Fh5g&JX5`%D}QK?`rCCe(Fn+10f
zYVjL$DKcD8Cn4j}No7TBM*7bi)!F!v_?I`+*&le<#iES~5fDmGxL3cfsCX*XE*>y;
zo+<|lpjIg-ttSEoPiFI`^p^Wq5PjQhhR_V8Fn(>m>Yv&6a0+U?R|#pi5pj0(my6l+
z?=0d#MmEfmT**fa20F@qpU=E7Y=&Cw5{^}9v0O(<#I+>~uk2vBtS&&QA4pmVt@@*k
zXX+_5OzpuH(2D~1!^S+ZLC&B~_i|F-+BE!9SB+tW!JB3|n<u{no<U(7C{OGnx*g5@
zdLqV5fl)`MW};Mm&QCv~BU&dun_3+p{d!Z~>paFwcuVVUy&Muj38)E4{0sMf^ny%1
z>7I!K!s!fs7E(vCW1s?UNaDnTV<q7m&J5Vbw+zSU^uicXtAuQSo%E+M4mdp;F(B_^
z+K!?QjP+>gbY(>OJ&;pwbr8Pnam_HY`gyPns6H#0j-IJvR?EnbLRa<))!~az)!_3u
z0LW<~;>NBZe{B$kjRuHDG3(ZIyK8<Y5OG~GzGs#Mlcson%8u!Z8$L&9{pg3;J7gR#
z@PHSUSiq~-dEHV%Tp)F-Ftp<?Rrr=cpoF+Q^2@33fU+tYj%+4mr&iLWP~Py5=6t`&
z-&j!?kbxM%!tY*t8cQ2AkIlPY5)~SQE~8u(1j?;@h~uwFhbFlR>s~QyvHFR|y#paK
zRBCR^`oC)jFw@ak?OX`AiZm>{)=jNy)l*e`_3&>j5@dPRkVlr*e7g8fp;}3iF`~ai
zS3BkGlz_tL;#8`W5`n|$AK@~|**uP`WCom%Eo9AD-cL*h?XywWl24~${y1lIeyvC*
zXC8<wL{enfKz!)syYZ;%UHt9LG5l_*b7yRP&2}4!f1#XY5{};q%e7=Rr;nI|aID)h
z`>WJgUB)Uh<tH_)G`GJ`5{Mxaad`R7RL5wnO{3HDqu^V~XCgR}nrU`xTuMBTO^Zs4
zA}zK_3J*MN{A^8QSJvz6Co+AtULHm7Z0}@axkjqgv9C{>p5ndU5I{dD?PueD&re=i
z$2O>H)jqgs`oIl$ognsVmL^=p;hxOkqFG86#<KojN=ksVY7KsjQ2bUjQOpG!A%B<?
zHV~##j}0Kx&Xt{?JLw6U4*ZXj=L0cKa65|usH@#b-s|Y`#0EU6G~8hQ&?q*y12D2a
zQdNFL{2Ly}#6>m~!!}7dQjfTz7_Za}7G?aqx>}$@OSx7JWG?+Z%{ZY7lbs}eSIc`P
z9`NO|X#J14u&Rc??w5rh3dHaQ`CngX9{;>D#h0YyMQluM6V1AsrwrSlw>Ou+Df#@+
z7|Ykjs%e?7z^?nTS>?(c_p!nTbe9IaEy_kl&AMCD5JLkA>o&I*7n`9Sukb`eK&!uJ
z^2})SlFHgpbrv<?(U-r30#4Gj1!yNx+&ND=cyzwGmPqdAwva@nHdqbn02qgX!73=S
zaN4$WOFKU8hd-(2uxBmxK)4J9ciSf^Gwk>IssPZL>M4q;5lK#Z+Xi|BYajDGcs<Il
z%?&&#tc>r!S{-KktTn(|^U6K**w{jR#&Ho(WW5$>aNxh9E#qmm0%%nyTEY_NE^Vq2
zZyhy8l87C6P=KD+BShF?T!1mH(yX-<FA)0o7~KO*clf{TVgb3_Ee0qQZ3Y{pEYfjs
zY58v6b017|n!NBw-pyxDY&N=|Y8j<EOf*>35(MjD6x^dDjkJ)f=6HjZ|2wnsAQh!D
z<F>hwG5IGahyCPbHd7#$AD0F-i8a3eRgO;bo6?Ui8VOP6R;e;5q^+K=M6?!tb*2qh
zf>q|UKt)|gy5`REa$S?c!syxrT|YFP=6{@!9j*rjMk<ZPq^Sp`F-i~7gxq(tNt3kd
zHmyZFXqhoq)6!8!!>yyXAix?cjd4nWlCoepm%!j{Q^l7F61YHZ)LL<ot3XwNN`OiB
z{Nq^|_0AFMavvDdjAbxHiE2S$n}Cwy>S%%T-iJQ(`6Ib4UwHc2U1AbOcekMm66!W4
zV7i)kWAcomP_1!H&>Oz2PEj&IS=9<+;#JaFHMEpmq$nGWr&zuxVM<E(RJ$F{6{fBE
zt#2(LP|>Dub>ji_Q{}BnL80*5%1T1=hlppq&!ul|D)_@@nd@m(S}%!L1Q>(&-Oq!n
z_IMpDEP%>%nq(>_%zzx@SB*6e=;{DRxbDw~ftErIj$_&Q|JH{oc+sOEF2aD}og|vT
z#!Tt*PQuBY!992U%;xZ4J0HstC`5$1nbotG)&T^;5&sE>-*FvBv~yySgXJ9y?}>;q
z>#8F)znvA24k+g4=Kh&FB<vPM;wa#PNes&B_-%+D@xcLfh5U~C=9CJ{Mx&8IU5rQA
z<Y#m;^o4;%X-6=JbiLQ|&}Uhzk0o@g5BdQ&e=K&^yAi|(Hm@Bx@F}KiR|Pe)bCPIP
zsKvAaNL%%_f?7X@dS|<UIzL~5Sh99u^y+1a>DW!>$B<@~WjUn{O&0W2F7WX0_&DmH
zi5T-Q)WRxUcw>J;$BTIxX_CE(zalkBY9i}U^7gTi|1FDb`}m>j%}8+$Q3Ct5%^&$y
zkOu|YEa`LaTrRZl+7yKt(R&hMHJ6{1+jQ?vcH}(p{@?e<&UH3U<u6|t>n*_5g0u<;
zQX-Fg9yZ490tyDa@ocKXX3;MqL2!gW|HY|`sFHc@DQPf9WO%$4PbtWlI@aNDB9K`+
zL3O^oMR@jGhEKft_I`G{EZ(?odfb&q!mlMKwjKSYQ<22l0s4ChKd+fU>izY29kzH#
zkC1PJ12ju8Nb^rK0wrMRe|63feYw?jw}))Um@mp0@u97qO--^Wv)3WGYmnUaQ&?TM
zJ`7*{804dh(Eu*#H=%}riQmGueR|~r*gB1_cx^i$46C<7?Ki}{fxSbE*rLbcY(pD+
z@!4of2@ru+kWVE~I-DuiC=TXr1j0_sDgW>n$Vra(5{BD8JLunCT*&pkQ*>)~!QFne
zI$8)|Y-XbvqJ^0KO35M<;D%tYn{H=aaj~TOdDa)aOj30eVx0S^Vkz~zDxg8B`Y6M}
z*SAB!NVH?eDgJbX{@a2AWRG<J6?1i_^N>)M_rzx6NZDODJ)o)G;}{;u`&b^TNNEBo
zpiNSucQ1PJFPF?YFJP2`HQpt{j4$&eKB)rE@;)NxN=}%DMT8bC&Q|F`C|e=iX*xWU
zT(!+uBaGqp*AImtF#N9^HW4^lhhm>q^dIB>W~||SQ_cqDC7wlk!dYCf65xOIh*Yti
zYoloi7Cnxk=!>YIcGk%<|9rUYjiuGKq5*xF2K>Gy(k0RnS?hGoXp`nM$e1qLBwOpf
zOhC2Cjs|v~{jy9fAo;(Ope!7ym9;frB95MMC=+rTs47DE*gu*0MyHn@c;0Qr!Cux{
z|G{pQAf=qoSG*=8!pG4b`!J79cyn<1=bg}w57Kxn21ke%wjjqY-76=pA%W=%&nZ5C
zM~IL#jS)F0e7S<BDiszdqge@1(3KqhDZwPnXV6Io6jaqQ3?qVM5<Xa;H#-y9;jhlC
zAmlBpw)#lgnVn_2)Mk$m)lrLLr1@eveoGkK+_95=8~2+nJeX1LPefNR|E?KM1n!6_
zs7M^(KC77w=vn-xO6>D!A1fx?HsVJ6QJ#A@&3F|kr;*7<BGn5A{JFK{>L;Exuptb)
z)s!yy+^f?;W)s-C`}ze*ts{l#Ope?XSxj`#J;_Rza3Fle4r%_29Ix&m&6%OFg;yq%
z;jR43!fngIAPz+nDTPuYtX>&;ytvn~+Saig<gta1zEUs|m(3L6g~0#g>K&si(V8vL
z*tXfRZM$RJwr!gob!@X^?<5^N-LY-kZ=dts@4k26@0~IB9$BkuRn3}n78+#gV#~4f
zyMuUsO4+*{$tz@Va6cksBJEe&tqF{=fG&>_J84PsxVYGu*ih=o5Fl?4z^m5tpage`
z8Iey%9Qx+~=U;XRI}Dg^s~u0wR3CnSoj~Qlz75N}E?&@wKqf|&Z3atXqeEdgJIL1L
zKU!3)2oNmOjDm|dzDNe0sM2sVcQOq&hnYrjLF|6PLw?-&>e$O*VpTX!IMCGUE^Wvf
zl~bOsIAH-);T|{amYa<T9Wk1LKAq7e6%hD6_J5LDQ93_Xe={mkuDTr9k~q_TwIAh_
zgzxsFyr!OWxTh(lwRSw8&?cf-y^w?8_tK-f+9H3q5n3RD)KwA)oA_1$*2@2dJyD(3
znEp;Qx*wKW-8NZF;YGZBf_uG~Yv~#xZ2f_HPVA?s)ug8bL0SGC6O`5F9X@1gUZ7|4
z%(&h``_7wCINn%)!B-DO&kSU0G7PFS569bJZokCSY8Kp~LD?Tm*n}}-%aA^be`nYJ
zEiN1D?7vK}?frrrwT4OEi&FnRSsRBsfr@$DHS);OydJk<XJ!JIBx|s60ye1%E){DL
zD#L`dx(Ef!gZI<=Z_m?Z^`zeoivHExv0isa`}5kZV*eIe@{C71GUlAFH9}033NbFJ
zp*-U830Q#JY5~#V?O^`rbGy&CN$dacFW|w>i~$pdJntne{dkaH#`@261JJqGLa>7B
z`S*&z^4F(Xnmge4zsXS>0w^2{00n!;(XbD42nkeKk;dy0=abIrG;q=K0msVtk|ck0
zGaA|edVp2Mxo^jMTRA|icSiZ(E?A@ikIDcZ?RFCW1$7~?2C@T;4_FPOR<QgK>8N}p
zX>&RFj#U&3^A+MDDOIQlJkRc`N4}zUwx8dCS#uYz6L1B`LB&#^Qyw8|6t#=>z*;2+
z8Dq9^nZx6HSf)}*C2~ANVDjzy7U$I<ODu9V()GAu5GZ&K+Q{3>o&(0w%Pk!AO?Pcp
z%HwZ_%r=TnVoOkZ#r~ynU#}5~ddLe|8(|^SnAX`4MviSMz7!M{;8wUYMn5Z6Z1xO)
zYOQy6SdI?#U?XQ5w^OqM?s=G19>kEKnY&b|I?{D?YK9oHrU`_fRHQX%h{fyP_xp|Y
zkR@wO?KHo$-vnTurSM*m22WQ^v^3;7e#2T52A5S~vn;?<mT1xo(Y%P-7;#l)*HmX|
zVl?1|Veh5n1F4TdND2eSZoU0RHxpr9_Gh?kc5s(*&kshQ(pYbTI5~0)W@CLeuj}o=
zS)dDY=s#m4_+UB8JAEfR_`Tu_U$9S#_%v`|BuV?0)v&J-;4a)7%(Wx9IX<ZtMgDx>
zAwkIkLEz)y%(X_oUi0gW*UFYADBQ2pmgf3SG{8-LW(ckQl-D7Vu?@B)%taMU%R!rx
zIlIvkHVe3b_;i`khojbKgtpOi2$mB8?6NQ<VkZ;V9tz@^`3I{qA+6GVm5Vd;3HY(8
zX6UGR<g7tEpK-nGXVW0eUwBYo#wc&yS%FKf=ONF%>9wSSbpBy5@!!!vj;ZF}Pwc%Y
z)hZz#EI>>l!fepGMzz^2jGigX-HFbeqgSWOXXQ6UGt#MJLq|Q>#^SWzBKXDyloMP3
za0Xmzj^_PpGA4*-ZA@>yM^uM80dY_#{GHKXpt9XzoKNNr%`~c+ococ#-xz(5Ngcs;
zLcX7jkf6Q4_KUixC~%ow(GQ+(T8ysm0mizvm2i0f2I_&X8}GYhjOlr#Hr(bGMk$rJ
z1;nmw`WIqIZa^PtmHEjY`#FgNdtW06v@l^s#Ti2qzO@|sso=MYh!t{?y;3AYD(2vh
zj;5xj2FJxDdy8n`U%H8bP??;}M&m(W4iUI^I`F^lAF+n%UWM8B^S^I~!((@rC;QJ!
zn=u>hz;%*@GDJ^)v#i=d5!T(S%?Mr;#ba!ez+u$-%Y+s}{2~$5cps}K7AAWyDYs6c
zbZvhe#gJ~NzxU9EyfO&m1%&q^`8)v6>1T1+1dmJTpq>P=9-7%NRKO|*PU#c52F*hs
zUg(AkGabk&c7#yFuiJYrA?EZuuovR4di^nR(9CpI#d;h~pu;>Q7TRNu!rktNp$~g^
zWFDgS0k@~1vb4sY#z|sC-q~u{z-^Qh&|jNVf*4&FfE@cdri5FeP;3<ihpb5YsE#>s
z@5_LQs>jfl>c4~`V>O9#D=ZWe@HP4ld|vEDJ!W!Z$$EKR4irC~E(4vmc02?92=-$v
zp%b;qB7Pgqk25shZb%`L8?uQ={A}YW0dA0;G$lpOz7Qx@z6_lRi@8cekwvb1F6bPl
z?RDd!`1eTI#f;7tYs(uNo;VvyUwP2nRx41mr5Y>5Lb`0|zcPTItyQ81PQv{I<imhz
zrRF9)?$aT4X+J9h>PME@2_oE5GO*Ie<R&eeJRiNY-?06&G=;gI34V3#?5!^$f!1?!
zwy>z#+iZw~*#FG%XkjOhMegmkm83_&#r-801w88SB-{3uZg$u`kzO<Dqu@do5Zy2h
zSZGXm=Sz^FHiir)P$9v}@z&L2*#Oe&`cS|KAsNqg3*2XTLtR7WT>_zGfdS0&hhUl1
zx2=L)8kF0Ig!1OSG~9V~&=%`rDo@z$@6D~<;43&<do_C2nLige6pD?<WMzB%@D4Ak
zVGv*XZ`psF&Wd}uzc3$cnfN=gi9+BN02xqP5zZ9fhNQ_c9CDfrsAtrFwy_Es9yalz
zfDOnGtCE{P{xKJQ^P!l3!S{niz=5I5F$J8d#RYjn#+`C`RRmY#%)-Avv||%Y0I1Al
zFLxH*s;Yd>v6+$nQPRZ)UBc8D_65*U1`|6ym$2N4FKcq_yg|xA$biIRd0zInL<=p3
zF<JbZv0y-94CgS3o(5K_#^&<)LHTNfe<{T*&|RAl17DR-r2RYpQZUXrR|+M$%<Exw
zbj<EUZFlTn%1q$3;i+R83_;;fyl{9p+yWQB5*{@srl2zVp>7L4y(ZS^_7{aatb8;U
zMs^q`%AiYiK;Tw!+oo#M9l~xl@Zqlss~`D9Z{=vMVG0zwLVPONYLZB-A=j5*8rPWr
z%KAF!kJLl~bLNNP^^I{P^zNDUyP<|u9=Ek*0S?$BSEz^n$PMhx*))w)LHv>Z(0AtF
zY|Rq#Y>SRraS;{JbK<DCOhu$>c`-Eeu1=-voK_TBI)Y|wy=fCqTfE16rdK@f%yySj
zUqBaTXot)X_4T!<XfpOze2cXlvYAi;G@3S%Cw$bUhcQ6sOo<3S!?$w|aGk<gsFHU|
z&j4;?eS58*nO5n&+TG8$4Vk+aP&%PxI*pRVbDmk!$#g;`68=h;I3rRJhw9Xxw}v`C
zQY|^5!q9v>l(H7|M?9gwwgi;6&1S@G^Br^~`qx4YL~=itxE8qODQEtD<i$uDj0Yop
z8=M*;e)RJJuAP=^8`jRo_`MqW;i;Hf+S-1UDf~xM9VW89g?d6uwJX`oAUlXM$2`q^
z-3;x4Krip{i72X-D5m9(|2}22K1F*^UISXKZ5tf$;(@)U%v8KuqvEc=^q_FnnV22;
z5mVO49b%+Q_sX1Y-A;L`YHA+TQ7T5kK0^A<E0k{FRTw`a?@DBZb3|VjD*sW^E^Nh8
z1y&s7LHyzFuS3DgwMRAVB(r5@GD^Xt9muJ<S|r@S{>E>-Lr=h9SMI1A%#rtUhjElM
zfn>ojOhtg<L?dx8`JX={$J`NB&NqJ}_=n4u%dfJl`;lQI_ovs~^ph&&$A;w|01VhO
zVTWUoEEIeIl%J&Y$AWQqd}L+p_!K|ml4U<2Ge1N*5EApcPE?^er+>yYVLGb06$IPV
zv+JJ3Bvdu${ERr3U_;lsOJZ)QE6If2&YI_*_mu7Y7A(MPE}11DiYtqkPvr6S!6z#5
zD9+n`1bAa{)nUBrDUVb0!Cah7_coB)9cp7?K@)`(SAtn7_#-fvRV-hab*&08IOzjI
z5=9igq+s0|kOSQh@lt)){o2@mueyqrmj7Ya&Q8J*05`=Ii$x;kA|o8ftasW*+{k^p
zT=%&?oGwshcldE%{R6Zkfx>)akeYl_v7`5LD-`I!*enp!iK(mZH%2c;Zv|gmyyq|<
zPYnmk#-9w_Lp7(2m4+`ZFkL3Ww5F8L<CA1PN%8HK)UwrenD)u@9r$9eR3@e(BrCYX
z`F*f&gKdu=YDGKWg#;n4tt)zEjN<0O3au9xfxh>U7jbKclES>SXqzsN3F<{fGj3=R
zAP^2SMN^s`E^lfNLHd#f2{6lda`%?Nq5>udQ;*M!YxP4nH>!MR9#lnV!fX)cI4Wr4
z>W|hCo5Xws(CNp6)9CvMODd-pq%p+7#q`_GlL<kXoLHE=+u^A`n6>IA0Tr?p3L~q;
znnQ^U7>6A{Oy+=BpPoT{71g<9O;D?s*gDLwuMH`k^R`I@ESRc3Q3brirUJ1LJ<zF+
zp)L~CQuCx)#GmkC#sW)6h@bGF%}ntfU!Tw+Bj?0>*Hgr}A-VZ8yo%7cFYBy>AsUo}
zqYud14}I{3TPF)ul^AF!^8$_P&sF<hY5{s<o-F~SW*Kv<J)i@cOJtSN;bX9}pZYs&
zpZ|%&CNkmd2d!;i{M$e|HiS<iyv{}IVRT>-#SDfR1OU=kDTAgQ`D*6pZUn9qT|TMC
zv|}K1%X1N?!k90XQc^oeyg`vR<>d{SV8DnXYbn4shn9eitsN+Rvu!6Fmp50eH>pt3
zDsW^ewW3*gC_omGDQ5(s6ydP-cVpwC6)_mrIZPCy`ywvkqKGai+))#D7V9whiw(>G
zLb^E7UdhtSQ8l>VVFOjzzj=-AD)XB{8A<nhU;kjOLYAA7D>OYuSc5Udy*S8LOjer&
z2XY*5@uqinf@wv?9X`B(BR*_k(VO#Pka{t#`?Y??)c8^D6E4|TLy|zf6l<f}H@9D|
z2y9K?b$XI*k*iSE8k54)E1aR(1~=n+{^V=^*p}!6lB%w5CAgaNPC9gf>`PJbg?pRh
z4rhteuruV{<Rn;la!twu%z2dTxzKI1_65v3pBTf$?$^^V`*lx4l!Vxbw4dkYH$_`z
ziOu`&6(fpSCzBZ6_$ursnUNX}eC&2Kpv}w})3M+Lq1MExx6JqOa=eL31ju16vds^(
zA^Y8|V?77gLIGB*QGH*Bi>$`anE%p4<U;|k4`-gj@lWofPX4QxU?t9{e|&DosJ^PV
zS-$=qgcXp10Tt9I|EN<TKqFO*=)L%3c+6Ce5=vMD5Po?>q}L^XkgVQDl6JqsKWC2@
z<jjD<^UYjW+va7#flt?@!?eS%ROBba=oJKbs_Ks@$xQtCyFKHZ7M>l*AX^@s308NB
z8FqcA;|ZRYYl68|OBV#(=8W>miNUPi`AoE$^O=v-e)??R4(-@a2OJk_Z%(1~w6#6i
z>FRD72%N^XDcVPzyAp9of;=e0rb@DwA*QIq80sfp=X>rd>2~6CX8FrTQuD<L|AjH~
zD|-D{?!(RZb{>&`?)A1iMQ{Gz-<=h!3~iE}UHMzJcC!n!LIakaTp#8wYB2qsG6Bsn
zsAK9ur6dpK7C{DI(_KE-kY1imfX4A*{x`>5+7c6y1Gj7dmiZk>dhSjb*PcclPuEpM
z$CUzZWqz<4J2)<CdO1HQhA7eHgjgp#1xbGzF=fb1KXL2o+7_cmjT5%?0GZEy4>4YT
z+buC27O^?U$oRu0O69$C^H8!?9Gc`MnsrIBz88uG2`D|A{e@_N6vl+Beg2qO!nHaD
zJ!I9eC166n<sh<>?cXkp8^&nx6W2)=SS0_$1VxMm39Gj<e#9FG!Ai~>3W2s3m;Z$u
zJDJ=HfJ~LyfAIwxZ{&0UjW;q*0?~0bb%oInel}GJfnADn_D)G-Z8jG8Vl^W)+b7Rf
zgND^{;$8FBK;37jd>dc$brIYr(^eHNS1sFU(SgxAe*HKN%V09Nwod}##pQjjet!ff
z445@I0u)cdz1xs$Krd!O%cHiRR#i`5E;kMyBDqJd`N~gJ;;xALPtN0D4D(S|@?a!*
zoaK{{wqSRIeo}BKW}+KGR7nBhK6a{-#BxVrC4r3!>ULLklzneq@rvDoR9*t%Gglf8
zq`0VV4Kuak8cfxApoO0O523il0sYZo4*R9=EJjp1$h_XVuAu<|1Yzo&OZ<hBvcfoQ
z$eZl@ht$iI@>-EPiM_bZBw^}3uyjM;5YN8f%NKHnhO^fVo2Mlf8YoC0`Me*AZt~eY
zM{&jX91(R&b=sJYy<>}%?C-)oXMelx0F7uSn;Q;ubS-LdsjjwImlcNSM@F4`cXqtj
zdLQ7nZT2J<)=3Mi$Kd12dtNs*DEw^r7lHprWB9MsaCRAHGEd;meYBq{S1;$cD71d?
zB^)@r$K|+7ABea9K^RWn%u!&o-^*>;YJ*JqA7I1)1~zf2V{!V-Aa+Fn`q08Xab`{9
zO~J&lJr%~+hv?nAvy|D?C?ItP{VAZzvu1cVWI?QM8Im{l*OZ!W*K1o)%mS!{W5?8m
z>vu7Du(Py6K)8zLpZS`M`|u)r*y|s)y%&|tYgYjjVLtpWs2Yh_td??>l5a(d0IaRw
z#n(;OX??UJt|qJK6hm;4Mo;~h5gSNv1m{@G@_<x`)x`mSu*<!M#KIcQ<nX>9HFx<9
zfe0QP19ic&&eWHx4dS-=^mCVC=HL)P*=<{z16sVsd$fN=v~Rl^ic&z+FS`)~7>>?S
z%{kgvvn$XPXmbbK6n|RC?8%0|03q3hbesm<tOhI6g2%M)W(oLyP9+UEBw}=eZSEa6
z`F}L<dXD-K9?e$AYkOQ?Zt;=tyIruyp9K0%o{;V5^~;mk;VkJ9P(-J>=sUzSO~@Sz
ztvJGDbXIA9Iwo}b3`-Ty`boeN`aBWBlQKU)evdX`XCD|nS}laNC46pHtk;ZDy%ZQ8
zbS2gE1wnkzblLIi`+-(U^(0bIyTm*c7IUMbaXeoq-Q@)`VRsgQ$8Dhc?mkyeY0;cN
zoV^y)dl30A{P^#|{~M%0ie9~>vFRBRvmM$Wt@G88#s*)j3S(BZ6y^Pl;r^LH1JJMe
zTt3fF!go5O>;W%qF6B$GuWYidoUsVuAWlIYWh#B+8rlwx)y*gaBu;&bo{Hiw_?J;O
zrhTa}#s-25(xG-AfE5Ol!r+fX-faEUBWf?JyC3ooFw5C`fJe255*yLrRcrHZ{KivU
z;5A@@sHkf$*sxN*kZcrwk3)^Z1*K5JTgYrBdt0%4%!XlKgMCJLogwiR!aU&;4l>=o
zQP)B%D&mm<8Vc2VD!mE2f^??vgJ+C~J?@XGf2GOJ92@B<gUekUjh70AUnuMZVrq0i
zhCN}wJFfP9cWnd*nc7+903WQyrk2e!6j$R0xHKPSJF_JG6V>NKnzh_EMivI|qp2FQ
zF~sEX5|A|{C2V3*px<sz^G++(J;IF>YB4XM%7J@UCmQ<VXpzCn8HWX*fBP1aBrBvu
zGHOnY&h3gylUHji$&KZ76V&vn&^WKBw42=fKLv8r?o@|PWFY@?5g-=t48R738ufFi
zGd>`${cU^jWRle2BeTno7HYg~MxNaTg3WvRATgVI<3ay3Z=@8}4<w<b9zXH(u8Z<{
z93A>!4)s3>%fDmLSxnFx$EYt5S8QGc!Gte)a=(LpZHCsovM;Usmq{QJSm(M)u_+R;
zAumraIpUA=?dI79s#+8vND#^UeamVA%6?5m0wRESsxBM11P3he^28A6km3Y0E82_x
zgHy;m^cz8(dwlp+zkoKJJ&}4%@*)R*CxyaC&xe552q-TLY;P(cB=`o_`oeM$=2-+l
zv{bZFVWIe2k(VF>6(%GIjhdm8iy$m1zrpor+8=Rs)>BohUEitV>RVyaH@vJtp?JT#
zkT-66>(rvPGHKJpv<7%1Ylp31W?5tGLjO!;SvacWk8jJdMpA(JUOhc-<=ldvnc{}=
zYZ+>2*wi0wTjfCeH3W5pLdKhJ`C#Qzc>0CzcxnnfO`;MtZSC&4$Xw>$s7r^4C+$K*
zJTR3}`(x_P49wox`#xsI$nT}(>7?lT!GlInB1JRjeeRQ<!<()E13zT3V&>^g!@AG!
zOWF1QNJlNwAdb>IZ{NS1VZ}YQ5VbPpEvwF3B1IZbRFWB22y((nroRYUZY@34$0N6%
zT!!H8W)rXjqrfWzEe;gB&9XutMVb~c1$WC!(hP0zG|Jg^kMTtx2k|;!?%uH*hQUFb
zPZ7owp`4Nc#Ty8oQxzD>usz_|Yi)QLaHt1&avo{Y+QzH+)@Gec!tn1*$X{<$u~L5b
zQdC7|vDW`0qyLlQ0WiNqL9i@^f>S^L%Ggusxqs&*g1VP6C(|l~T%%~9p?`nJS>l}3
zYqE$>i1kk{THM(1FR86{WCEd**KJtFrTM6cDwIPC;?fkRpzV~CvV}Mwl%al0fkzW*
zAA$N5oc&c?z&O`dyS5|lNlN_QSfG?Nz<`cr?!dGU{%3I9kdnJ^9tAX2zxrs+9JB^z
z+zQBiZDz42?uBk`ZLozfvy}wnzpTb}rp#f7Kq412HL6TFvsnyowC-db@mqiZMSCPg
zAIoe_RJ++qNJzLJ`$5%Lw^{VD4*sS0AOw0Tvu{jCuYV?O`ZCtQw!jS<AQTanKMj~C
zYWQWzvUv=eOVqxx@AGkwWIX2a<1FlgF)sAQ4f&8@WB>5(CC`m_1JT}W|Iui3gUxET
zz@df(g;3XcWFQ=VWZ$LZ<9Vh`2y8RJaHMAhVY!r-9mGB8$o^xZE-UamNNjI1;1|Zi
zHv-S~EKewB%L;=N!glxZeBpv!)82XPN^>+(Em039iDy~}>G3j3Gun@u<eB+hR(xcz
zc{IgXyU#D9u@{dZ;+65c3h4ajhhF@2IYrnP!f#JdiYb^qTDJT;EK1QJD*J13RjEcR
zY1cMlI*Ot&HVIs~FbJRh(g&9~AAg_?ZUyTXw_5hX14AX-g~)lGs3zV|ipe&53HScc
z1X$^&pQEsLtxC7hi+cI8^ec9UFqe_uQN_7Np2HbHf6<i`1nYa7j%O&J4W}Vv%&oaO
z<Vqh%zX~Pw`?x12^Mt;V@KKv|Qnjem9eWjP&1v%l2CM7>{_iu=A`YUHXGQu5uYlJK
z5W#o4#MGL;vv5f_x;clXZ#IaPGn@x{dg-@2%J72dwmAquigcTf9x?9=8Q6D&0w5jT
zL^C=60-X<xQLh>I-xgX4Icu43r2k|9`DZ|Rs9c0dgBO7>DKWnB$ZSG|#K$h_OJzDt
z!MT;I?9DKk7F;x~o)sE?WfAxq!79OnCO$TQ$0{=0&$zaa#Rw@2aao9CsF-S_${s4v
z$Od2VC>^#x=dN{Gpc7$*Dkpy{6G5OUw%f2i!(jmjB(GIRvrGu?w@?z46o?n4y)a#b
zJBH%j8OF#)^v)n)%Fs^N$0QI1K(|Z^n(Nq3@}OU9kuIu!$2tZh{o5c~d2N*`Rh#Or
z9K7`I#t-}Vr2sH3)YPfqHOgrK^;J2^fiERE1J6W>RUnwC7CZ^K+7B0$Uyw20($bmn
zqQW|d#1eunIA3d>>50pxHT>_m4~%e6MTUY3^M^~sUUw;cN_B?8TU)!FKH0M;vTp`^
zvvV6!F9y~n`iPZMq~GqBOfUPcd@tXk`1RX1Ze@(_c7O_X=itfw3)Dk^CJBUxbDt;0
zXMwAM-TrMZS?(z{YRRrCDFaN7jGpqW;)U^$7X>&>Q&CU~O3JD96l8;(Dp4v!k0zNT
zC{^KY)MSo7q+g0tHtSM8f!9gfHS|N)e@gxTUIt-ul7@e1@!Y}QOkPtceVHybLpWJT
zpd2lrNPj><?AK>A-@kIX95*Am?3S<NDM7*yNe7d@lJmk9Hh#&AnzVu@Rs-nlff`CA
zh`eaB>}4-+i3P{M#Jr6}kSi6|vm7jY#<snTC>(pd^SVH%H=B_J!qzgjR4!{dqd%E2
zvk}#-9^O?#=MiEcS=^LXt57~{B}kA2ds8k<^yl?KAn|)+%>bgH1!b?TE~y4s{!UZ9
z=ap~xppGk2d8$&`^Kg;O);6~-8CX2zb0{TKXYt3?(tHmD%jOcJmm_cPUTs8-D^y(e
ztKz2CoQMe^VF!=Kmt`)>f374c&<4~+Mfn}!K@Fm*tEP$qk|7(u2mMYA1_VFp364!_
z_k}O<Dc^_3Gf&!vTYVL$S{{;%j%${X1c$MpSD#Zk!8#Q!w7*3BX3b}DskBD<(Nr2&
zuyOJSdW-Y4bneOBe~Sw-A9RHO`B@)&v-=u5srC)J<VjGRPPaFy5D(J2Ln)#tJY(}T
zV_(lIY!xD2Y0X^rk=!)kmPFo&j04)S>*WH)>oo&Sd_-rlQh$KJO&sP=xwvxxY{3Sk
z5Bm4nntL3b2B70hQ%YwoM(?{X>D>YNwmq+u%AXc4h90IO3wop!lt(TD4xO2KKrUzL
zEAHQ>KJJ@R-8STxzXFD*Pa$9bTT}jf8?2&%a~@>aSw+&r2BQw7*fe=2zXJ>m!j!^z
ze4Xpkv*%s}dE-REm;E2vxlc0g)H7^JhV%5I@HrejFd7tImt&e%mtG@4eBh-jf)>=i
z_1Nlc{m0*SJn`MW1e96PPcJQ<Q7R}g;I?c^eDDEHbP#a+0aZg!JbMD~6?PmjYA&=+
z75^Us9@Y)?RwIn70N5(Turr&i`Do>9_u#v!-Xe6v$=FAj*-VX}V%NviK?j$Rg?aS=
z<3wo8K_Du)KoV4aZsjO{@$`xMnP79gK*Om2Ar(1T(boqRz=ZGiux0A?sJpziy%xQn
z6u|}hXd4<-{eot0&$|a|pGhtWSb~kL-vPR(;EQJswU{*YXZAlVyQS?(o8Y$-LG~i1
zhZ>!C?uPdlIa`Tt1LV<1gr851y~Oe-gHdKVBIaJb;!y}8<aHGnqddfthgw4UO4IOS
zuR8K2X=(Oz-!c-t>={_u4*Of6hu>Wupl>VR-SiF~LFTF5eV*8-9~9~!kwL8;q-D>@
zi?6!AhO=M{H$JCJ+a0cj0IFjem{(NNX9(`y-yrr!N^12S+VX=a2Df{UQm(*y0lbEl
ztb$TB7vz&2vxR<S<`+q25=&Ljob#!^5@*_IB-vWDXd3l>UgvaeJ+*&*1OD*={crpO
zO$wmF+8r)E=&ja8*Nv1z72<C#EWV05;4=Al_-()?pTO8u81l`EIjgP2V0407*qdJ*
z@a-7&ir#_$^8nPK?aHpCMD&Hr!}6_XmT<m{l0PvXRQEn6a@{uh2&-2Um_NR`mA{{J
z5`bAdz5g0~Fh?Bei9jeu%-J@!%^azZ$2Mg_gQMDO-S7_tI!#i?gPcwg&FC_45Qj+V
zf`Z(@3ME<#Kn`zS|J(7>{_dSB6b`9|A5^WDm4n}Fl`>A;?J5{F#-?Gs{(@b85~G{f
zB8c-D9FE8D8yL`p%Qfrl>FKE_vwU18G_Nl$;6B81%dN48Lcn1*cv(S=6b330F=;7R
zL9ulo=&c#$9hSp{x38yEBeZDtxT1J@#P2_q#eMRaeG&_jZz><k1Mh!*O}Z_PcG+fP
zG(*ed<sxuHuw;rQg^_;IOrVsLeshx)CdAg^AEns2w38m(hO>VivNyXog!?UlLXM~?
znMD?QRO^-6_$52&O}Qp)N=^1>KY=|o**0G?R2P~YDeFZIkAEd44HBUl*e~&2FV4x1
ztOqtwQsptdz7Xe&WjXRZA4;gw8wX%y_TYa((+C#8UO*c0>wX}gf?1l?(%ykV{1N24
zJ?i_khG>pqbva(ttxuf~LD4x#z;HjYArJJ4PsZ~?j0!HjmBd+WsYvA`C17v#8%1+u
z7$x{=o;|P!@{Gb8CYH%U2qned$Op+n+rEp@Q2a%8aYPn4n^Z|mhTJ@Bt5YToi^1zh
z^OwzPZ6MG(N3tw$*w^oB7fg0^$4hccQ06N|c1uouppk5$3z=ULj4)7At7|g7>3%~)
z=Urkkj+o%<w$?W?;5M)2Gs>0v-=XHeUC=+u7H-r`4qGgHy?{ZPKLh+f*w?1xend)2
z!|(a(+;70}VPoB4zPPtvh0B+Sd_I!R&^ku$IAeD&h#~4_Nv%NPH4G0n^}kFYk#EEQ
z!+7@;sgSX5vwgy(`4qk+Bebp%xov&{0KE$U*KiB1BHLyptDQ-y+&Vp|!Yl5D^i)BK
zvW4a(;0)l=TbY#<hkzdqYLX*#2GaN>wV^4-pf!R7&r8r?n=B0PJypops2r^%+haWp
zBi2hMvf1e08NqOl{BSy&sN5<vyOh6}917NZK@6I<e;#g+CsEdX3HSDPI&ARz<V%M(
zR`J0qnn8(K#0-09wk7+4!J2lx9gvPKe8W4XY|VQ+#B#FsH?fsRvq38Y+_@m<;re|=
zPkU=uhKe6`$4ut7+n+`ev3=AJoaUj;H50)@)SeaImlwmeN~?b{E^C%x)C<em5SIuT
zB=g)+sO`?cvjvjmDiF4BUz(0oNt3(%_4m|cCviW$*HVpfPB)H6FT;uSi2Ylu7?ca;
zOJm0%H$7VRRGedxN-Br}#)^EQ8_-`At3mXTS3r$U@mxKJT(^>-MJ({k?{e#)Q+F$*
za9)?R`w?pNU^YT}oiefPLNnbs!1f-z@15`+RsG<u$$8tKD?YhrQOrvWOtKdNLxutE
z*Bf$#m2Yh#3`2UFuS|&!<tt2e2thda4&!mldi<5Cs=?^i&vb4G6eJOYQ>F1y!ttNG
zvlZ(%gS&NIRyUlS<eD|iiJqRw9P=+b-am=KdnxyP^C0+e!SsI2Qppdusj4W!&Hs2a
z|IP>lxP~PUJrqkVJ1GtC43V(A*=}Uo@|kK-+-&;umhkcaCd7X_um_1CPI}UI7p8V4
zc&)^NEoH;HW_dy}-8~h0khzE}>gA_x+JQ*t-N*7(7d=ALOFe9Ffl932Dso(JJI^M#
z_c`AFkK}Ry#E%4!TKn0BwC2r{I?FkRnM<7-a;tVpE-3tH;%3V_DEel6Q@m);!xka1
z<y4-%d-=-N)>av>uG&DDPizQKKYVmT{zUS;wLYjq<-LY)2-9F%98f&b`-LK=OwIet
z+G1V7HU>@h@iyDV&QXxE2q}wXoN>bHFv3dVt2b)|K=-i6fhw3`?4N?O>J73+3pnlA
z?}6x#bU^CU_ihkIhz2Oy_GHU>u(HozYg9q<U|Kf(1(=l7O!b_DusSSDL6h6M?RpD=
z?Y#$(c+AzD6IK1TYM-V@U1tJKOs8W<0npBA)g^Ng3B|ssnWJh`Q9rXlvn-7!^nHax
zTd_&$?l2%cTMI%i!*r^0y_-Cbj=Q?7f@O4F=k(vEgE|Nk1QKTkxsZudI9xPw)7$Z0
zTP;MVrd1%akaX1@zuldTh!pH~L(Bv^DUVAFtUf#@FD1U76lQ(bMK<h6rtJ0nap7~f
z+piSg{EaF;BgxUqFXbCGmb7lBc<Bx7JS@b!<LsA2ee%l!h0qs|yqm2J(pL97?OC&X
zej4Xv_5cC#Hx_GiL(|O41<_{uIvmAuidnK2a8dPU;Xjrojl@-Jts-4(oKZAF$o*A>
zG}h-S7mIbMhF`C+m~vd2HCiHDFUljm5KqyfL{GKtrhuSLP*p_h>G1b=y||UbLcH0h
z6c)CXqa8-NIMrRD`^DUqh-VRq)~LT$R`;(DH}yT9W3!2Y845Lx|6~97Umyqs)J`h^
z)^5^SYX#wn7jb$1CnQ?;Dh!m2E^GJhG$cIMHppdXE-xEl1Q(3Y`A|p(A*KVCH=<bC
zqO(=gIOldES2YICIh4dUrg%XoJFnI6%mL)b64f3Jf8<8W2ZNLSE@%gG>kzB?UD|B>
zHd9ez;ShKFAa$&|Tc$>9?gdy8oyTT$09fG2Uo8lSU9Rnk46HuH5RiE5G~kK>W!t27
zmWUxx3sWxoi^y4^gTg=etsvJA`Lqjdzwl!swu%f#aUXsA#rt}5N9GdsStv?OQU1?}
zn37vVRwOw5Nq^BvPX+N?(3l*Eh^STxa7X*57XpY+c$n9w2F3O>mh@eB(Qe&G1+<JS
z(0)ZYnyA}+6+`dnNmu*?{T9XKC-Z6mllhbF%_M^G>3e}2<WD>OaD1?g%8ZJLNYbc#
zK6-FZIBS80wY=3`7}iRXXlJSSV;I=q?{keFy#~GZ*HP(yHZ;8>NHCQ8iQ4U1+^bh~
zA24(qu)&PKAMA%s1zvA7fz{fda607rJHqOipkg9D`K6^~59(jT)1L0lh}<?NgW{7R
z4L}C+@>E#R!h%;RlqN<^zDUM~A{hL`AxxO5$DU2^hF2r}vOsQ%%f_`K|2+Y-n{Hcx
zxXty^HlL%Ojm!=FzG>#qHY6OHJD%r-fy{i2<nT+}sHYm^5?Y2{%aY_k6t1t_u!e<K
zJ==hfV@H2k+d^?6kW@I=xtK2Lf*YBUfb0PgSB*ojtH$b5>mFdn#x5=?pS}B4R@$YS
zdCOrwm>S?+I%-MG^KziO#r*u^iGZFkZtMSB`2jJeU=W3&Uepu6Ntsx7;vY)N<1aG`
z))0p#s8*O*pH_^582G$GC&$NH^D8SS2Iv~RHex+1XFFJ`VhmG|)Ulz|v2#w|P2VTT
zY+%ZcmC%sdSHPgZc{vqY=yk<TRU9?%0DyMl%~*Uh(F&1*p*1|=ifN{cz_f;3mxS`}
zCU70fAyWYx1vI4e<^NU5iD(-D!i7+Be5P3Kb<H&;**(q%)NyJ7X{y>stF&2XUg24T
z=(X~4tJwvP&hJ1yDb#M3S=;myB~MwF5rp)Q)ySOpWWdWk)F_?xNfDjs5np#+DZlXb
z=^G-zw}f8w`MWh0R}d)4oI5R%b?XOO@#mMKv5ZDI1iTE0A(fbYI|ELH%bvGx8b#=9
zJ^Rz#YAPR9T}n;soHodW<s3op$LU$2%#i_v<_~N>N(emzrufCk@Hy8WMHPxrnPehL
zvzfhin{(r2t^l^%Umy5GaJuT&W85+U&SS_5^AI;j2^^gcxV%z_2erpV4g*ar2e(N5
z`SKP9^t9Mzzq^!Ce0<IUYm%QA$8Sjf3<|*%yOz5fxf@P-PV@p~(KOD-Ghb$Qf?K|o
zG%9<ArD%0lk<94(cfSLEvgD=lIL}XJ>ywZ4&Qn~jN)w9ePxWUzbfcbsfXG#-Qkx>s
z`OY)5xd|K&fo9^<yIMlW+y{XtIn!$#`W8N%`j)Zx@$K?$-8)c2NR#1;6!KJK^C*jI
z8w%VcUKyJG+{OBv+_v`;HK4bvGY=$ie0B~TiF_72Ob0>LG7)|~arxUGO9f)a&YGZn
za?}5JNcGP%Iu!!Kxq1w7zLEDXh}@^Me$8*;$O6%j>A{fa7*0``A5Nu^^9;7$7c3tK
zWV%jH%qB&U@*gHYw(R;?Ae!I6ag$F;WQd_K<_e2&3j`}-X_kD<aH{1u%=#9-)<^RZ
z_)+&W8;sT$DLL8G6KXYcwJ0s>Q`=$zs=yQxi{?cJ_~~I2?{Lauz=9(Rw4aIW;GFKB
z1*-})vMaXfil0wUPMp^+R!fG}(4#;uz>`20wc_tQ7I);gvr9g$Lx;OrZSlxKg2m93
zex5f-KUg&mpQG6>Pq4XI?NI?Jn!rV+0~5}m@-N@k^!fnC4W~bL$_K&irb)LizPhfp
zxIpRgs=I@3y9m;c?2pD78XSQ}5gGd~Ji449T9=S@K!PyN5QOA$=f|lcu$44S;>ZHY
zvz)aM<x`y)RIwO@Gky@BJaqV};BS5ZWKfE(3tZ{Kd>T1ozk}AX{(M$9iGmj8i`^{L
z11X~0-B0@5v@}_2eO((C*j@ke%%+v{Io^LzTo<X6za~{*bM=tWin2TtARFsyw)8qC
zEHLP{asQZz9<W{bb#4$tLzcf5<)aX5D&~Jjz3=stugRbUS~-VPODFCnjeV$1Dgpm4
z$<;jH(PawFX>@C=1Ik;U{Hl8|km$_jg;?~^Q6zai$nsu4hnaQ5O|*iRRBz7ZkTm(U
zhj44Xma5NrFt@|*)RF1A8d`b{>_19}|E>0)QGzHP?;@oWI7yF*dZZc)8rggI(UZ}p
zXd(eA5xf0>>mR@s?LYwxB7Q(d)H*3T5RJST+Wm;VHb{7Np#f_6wZRRkf`o3;dE8I`
zs<qEpjZw3eGJsWVWiN0LJ1O!#$j}8S&lo}(e#7AmZ!u9*XJ4vV1~eaM)%wghh><X(
zY=!rXK{{XqR4fZDElYWFtllgtTahoB30EKuzuCzjg0O=(T6_r7KQB>%RS4<MUCu~4
z@w&<RQg*`X{wTxO97O@sVgr!F8-YS_7NWGv4;66|y32kn%u$r$x%p0rn;^YA1SMTO
z{39Ru{*^sfpx6t+T~YJ2Hlcx>y}VI$o8Tia>U&zx$@r`iVoVP7d-J~eqbMW~J-iGd
z4w7{Uw4gXV<C$7(4C<gG$|+@t-7^uo@R>P}?R3chMCpV@qom<W{`y+sLuTfQVf*lW
zvQt5#MukX0QKp7Q5uG=|#|J+oNV1>i>|Zw88$8L0w?BV#ZAIR58(!HbLE+wYqpSfT
zO$CQRO-&L<funZJ`uX{_xp`^cyvoeR*QCm(yLI<hP2SUeLDx~!Q`7%JJ?$#BCl|AZ
z$eyZWQDNb(@*TMN7N6t!@gX0_=MH}Hxr!rQhrsoSCz{mr;^VEM>(LMBsx!EAeJ1WL
zGv_5he2qWhfifvlyaqJIe5>{O_z0a*UDa5%G<t((;<s6QV``7rA|7gaEe)F)Qy^;$
zxznko@pt~H4A|Yymwv?Ow;Vz0Jjl2rdharPU%)eTmn-yLsK<xa^Zk?m8n~C?`Yuip
z5q+8FAi<!=Ig@qFovdNvb<l#`W&Uslqj;iwJNE1L_ubpqlPE$@jTRI=csl4rkU4<W
z2)6;j4HFz^!7sjh8tKj#RU!h4K|bvRL>3RufqPwISn%BLax1}DoSoZ>dOyw>5?m<N
zo?6q!;+J#Ce2aGq%(zXhoW&4p5-K>gTQ=s4p4C1H^aKBD#3($pIU8+p{Q7S<lLF8G
zdI5kub!sQiB&Xy{Y!F`ZZIvIdv4e<N{m~Gi&e8?&S%|`D6yK90LNNARI}Z#ADEeFC
z!z_rhCWD^JSF!xDKxern;7n8uUTU?L@0O7WvIZ{G!^;q)3hNB+EWyl{$MkpQ+$p-6
zX4UBCDd)pCJdwO%oGmGm4v}%4%fGyQx>(b!PO|Msl&PnjsWHs!@CvKw?)LM;XbrGH
z0GPV=#b0kv)-g^mFM?FZzqNkMJ7e=jx2M)0wW1z04Az8jOgNjuBk|3xf^Kz8#J*4=
zbog&@ggOR%rY#mNn-a1y-{*m^MQi{E>ybI_mZ=%{OxNfvQ}2uCk$Ph&t?zFyXV`=v
z3%SI2wYYN{I?Te=;~N2UU3Is6pG^+)9lx#ly)cwlIVtZyY=npcqquF7O39(4Uur+{
zJXZ>I18%;cw%*y^x=RLi=$xVSp7u6A`fl#F?k^0TA+cNx$zq@YtXK&3Sls6BO&tvf
zTxL;MubvHfi^aZQPx_UJ6LEJ9*#G~7p9K5`8Z1wKY2c`?=~eh^V19DV>4t;dI->`}
zqA!>2NjyiWtEsJ*4f1NjiN2^nR~xW4%Z-$6^U_h+g{B8L<7hQoB_nW$ET~o7Nz!`u
zm!?&M_eW7@@YYbTIP4U6=c8h&RWp?I@^wnQ$o3>vf^(D_Eb^y1+GHML`PBwFX4!!$
z8Ier9F}uH!{&CXbk6@iGlE=4yu*6$U2AO}7F<EPt=U?7Jz+G;({&S+oDx9Hw^`O9|
zeF*O00^{T_CN{`Jr&-tBRm!cvu0Sb9&Sy+g3bpk2@828EYB#?7=mZ%abGf9Wx(-5#
zHMbNNKn$|<U&lWJ#QJ{MY*Onk*Zg-F-_K)uGIIQWC&-R$M%s^VU@$y4m@ECpsoq!Y
zhq2Y>Tk78T{p5q*YP=kjTK~0qk8Hs=#C(?OoBM-+-}A~aa<l6Uz3$8qs#A}_8u#ga
ze93C_-Lt6I;m>1iM4;(Xj`GJOt^dxfUB2s%%o<atkszJ!M~WrocK9_2Re(~i6~okW
z6?2Gl9=f97^68iu^5)qP8WnQyUfoq(_k%`HX6=3uXl=~GQ*7IoQaU|f^!|OFtf5If
z-Lj7o;>qF9lUhB!DVxG-=HH#0UpqNOwK;8(|0lS$pm@gsVX)}0O`=JwF3MeU!yF@?
ztSf#7y-~j{!@#S~?tUs${n60n%2Wbdbfp4H{=*f@EGG}>-}0)eXBSZ@V^Zh);i#$k
z$SHE8xfv)!akp3Ep@7CTO>^uO+kA|Ws_x&Hq5hZ!Ef7ES@Ug@ZA*`2pSH@d#f>(w*
z&Q+aRkaHps(nZQFdAh+&{B4X>%XPl*<&2YoA2EYsKG%X3#AO?$E^X`AS_)rQ*zMvN
zHOj%t0%}FJ0xcopAB)G_KOmi^5gHe7^rlZP=?o>F5QSI)tKZwb(Im%TE{PCULH}ME
zqa;^5f9Sc;nid}Ws^(6Fb;;`NJTd811~MEV#a+wxyf1>#bq~@H?Ok@-oqInxys(8G
zx$O?$fz@t@p9(CI#h08t3&}hSj39Kle+cR1GzGej0+QZ5HV-KX7=~WTr>XF(ZwM7N
z4pTsJu>Gknxs0ORC)+xkNYeuk_HQq+bH5NZpmiaclm}@^WP}3di7&YpuSmQKUbc;c
z_g`x478aYX(W2Fmdq3)siJT|&RMQ|M(<+vc?q6@G>hIrnFNj?}_UYqxdDXny9o~8B
zP5ST%)U8unLbwh2l$4*&^YRV70`P9^G5$Aa{~O7{2slP-Ra+<mIKQR(JCd%AFK!W1
zCOp-qJ`>^wOWf=LLwC8>=_gMJjSFW!?L7gzzoS)u$xT#D;MC|K6_f}UZl9ZDv|}(k
z5#a!wtuV$r4=;38ik1t^3-qBX#FW1@#y8|IBpD}VY6(77G9Ol+-NtIoMgfpCH}W5O
zCBn`9?6^2G-BZTvu7kB1^WA?X(ZT(;wz38w1_=&fakBGQQlDD9g&D|9akYAUT80VO
zqpEr9pCXdMT03EI-2EjnT6@h&fM3OR6ydC%p&<X%3HJLB8!r6a$4segMjbT>)ZVWg
zLH}1_;dm70OLOKiX=A!b=A=$c=f6<=-$aAItSoDe!h$_t<K~K`Wb@4_z@aMMSid2j
zC&OG#ZO<1atX?2C7<5~>9Zb_v1|qako}~LHc2#1VSe9Z}M?dYxzkbi0bsZ}aa-F?6
z6XH8hC~#h0xxaWIWM+PT;xi!RbsXvP&a5Y~Dd4`^P8n2qOXwPZn3xef58JyhfAYyi
zXl&JYn{=<A#^r2J<EHSon#u7*SpN)e^YeJS?C%TCzfu})72*td#}G33qIU*6+tz@&
zSw3{-G_0jknd!f{n&`fT4ZUi&3^;YhpC$(P?{hoa^MBQlclLYAus__T<m^;`dkC;I
zaA}%qr=e=kCpQTgTzCugJskh&<GZ_ibNtCmf6$Dm^MBuSvwjdsn*8lwv2HZ}d9~Cb
z?4UztbEn@bn|c*@zDPen1_Ajn<*}l22lSCbp1?iWxGkZGbc!IBQ&}8uu>8a(_^7YL
z-jny~!3HGjk!|9}XSz`H=0QF`sz#Ap$>p9_8ROb`CT&sH#uBau{A9`KE}hMSD2i45
zRclgVz}MI0jg_Hlblax!=*_$gb^W-=_nr_`#~j^O)ss=Ziaf72gHh!JT$?L1sKMvP
z>!g%=tq7(oislt<FTa|-IO7Mb`!mM?K+Sf9l4CP(lNn;26bkE=B})=zM1uysZfng2
z3*}wI5hf~nff)wjit*n=jwUksoTvR$7~UNL<z?nT->C8`$F1I0`>dU*9RaL|HM8Fy
zdJ!4e_V>sAOsimK@O~p{KNL5ehf1!Hxn~{OpuqY_b5)nk+-mmHCB84({XnWCpW2~E
zyZd$a_T}g)WH~U)w<NSu32`x`&BDCjd%;B|V93_rW|0Xb3{@r>tDCiKtR8&mkG!NG
zW!U?F6ct~sd~O!c9QL|a+?&1SG+8>n0aiR0d=cv#;Fc2!+CIe+d@FX;s!J7y%Xy=R
zA#3w&d6x&}jVcb#$j=tVHp?T{1D4*P)_tls@H+|TZ=Judp*_CM+vo2a=*hoMvCV~a
zukTkkR{5_J1xgQlcTN9MIM}qQ!=H(<<l>FKt$hpR+q-%TWcx7r>OCp*z++`B>-kTq
z_df-IuqF#&k+nQ|33>2+mzw#n$6^h?s>DYyo2XOvH3y%D3*}`viP<*tc%L>Pg;m?d
zPhVeOhB~E^NmiX?y!`Xh7A~98NGyN;Ti6!^obDED5+isGN=0onSRqmi2Mv9U3JS5F
zx$-a$p^ek<_bh}_60FbrgmkOje5kG3fA7eqvNB9Qad>S$7C2E(F5&Qcl;bXKip(k*
zRTxfwb!BCxY_%C@ig^Jf_}_;Dd3>UTxGde@w9i{VDz4_zbz0?huh*3IK*q<%?*%R8
zD37}&{H@U8=h(o)gD;L?Cb&1hqk|ubi(fM`-((Z?(`6zvO?_|*r(z>Wzu}gju5Hc-
zsQGfH1)rYUJa}&vny#P8dyzYBh-UG<Z4@@w8?sotO?on3>tB5+4CE8}b%o3<WtV!I
zZcdsF_#+HfR*-)kj!lZn%s-zL@GWiux2JM?)Zud9Zsy9i>hqh$paU%_$%$w_*CwF&
zs(u~mZN;||*)i4W2-~$B2FyO?AYZm0XJ#09zVp3nCF%kz2nYWzY&<`e2s1uGEdibr
zFKnemro*pj2BI|9rD}ox#HZCT(#5F9m|ZG73z2!hCA$0#9e#~@-T@i*8hIbga^#*n
zMtf_k(jbKUuX+1Euf2$gMIVLR03(9_p0xvwixmmNQ9pIat(7aa^E$HJLrHGqIJ^&6
z+#8-JfPp1(waEstYj<{acdhKk1ebQHEa+9w)at*j!GG?{9j+1V8_{G@H*`rd=^J`j
zHKJrMqMb4y>zMFZq;jWKX-2lweKGT#XO3h}Weg`GN^i`MU-w)86chCa6B%3I+pQ=F
zYF?$af-<fnakVm)jaH-Zqxn36JCQJJHW#{S(}TU@Lw~paY{CIi&Y!j_6s$9kP0?_N
z%mwW=8%tTJ!Dx?Vhca1rqknOu?C92;=(-k1TjW#PE=lc0wWL6Vvaz*g+Rr=1Co`ST
zb^1B1&iNgc#R*S~V|Ee={j%$(PfvUq9xiR+etJ6>`}`8rQ`d|0o{&R64}7)~46;{B
z9^CQ<0~vVLO8?%$G*t2)uriM;Y0;GtNWIy;$6(T(KTEkS&0ky3SYtL~wvV$HVY7ZO
zYHW0NYZtIka8mNT7eEMmDy5vsq-Fx69ncoT_v?)&fsZ=57|(ud%YCES2uZ_2XfRo|
z$h(UtAc9q_{3ym^fq#9%aH$_rcOjk`>T4~qilM&yczenRUk@PN&$yNf+evx)z6dhW
zH+|#L!~cLG^z(14j;i;NHRsA$z$~S0pXsIdfu#w1QR3@wdfTA5dPYLGH@3xRT3Ikx
z|CL!^gRRHcoc<%Hqp$C)7kEn@0p~0T4HlN6!S{(`AFj1)$2lLbzKXoCdnM)1Ul^FX
zm0F=)Uf22gTN(15`p#3f#NNJ4ZG|_4_qx+tmwZQsPy~lioq$DGONyF81?>wsDrJiB
zo;bVJbgx1BHvY$N!--i30Ts{F6TP|~C$oH4>rTh8Z}mzD2+zB<99@pjqiRdeSI%^v
zviLdaEGDwu4%!#2BeKm;^bdtG`gu<H5d>~;-Ftvq)BIx4rVy-?7zoD8)LH_km+=jT
zKP_Jj|EFQqggQMt3rsDFQqOPM=#?+@77cEN>{bl}|B^&tz(Vm$QSaWI;CYzipz{+v
zTD4_6y%ik;VO=DN)(D9YzOw`p9(-MHc87-R))!EfQ;l(Rq~_u_pg`lY`+3m{{Auav
zI416iQLWBACqdTzL$%6~{<32Sz2z{Gv(z@+QuB8sbA=!n_7jG)$u!&iVdF;B>5*8Q
zWVU+hlmU?+8W8xf3X6Y>Bbxpm&7#!rpJp8H7DdHM?THoY%F5j~4+@S`;>R2V7THtO
zVBEXu8Vri{^lUuXpv^7u8{}e8T~2O}5eaYa+>zOhlAH^6_~uJ;6wgQ^tYqho@s0i}
z#&kYP=a_oJ#oDiQK<QDH+E=a4v^Ypc@Yw*;(nu@pOZG<tCLX-x9QrdO^o5#N$|vBt
zR~wYaH>z7GLG?=-*o#hlI}!QNSX1k)9Yd#-<2F7RPh%Bq2L7<?#dXI}A>6+n_%yKK
z^nuek%lQ6Wc`0to$V;A-#%3JeM_86%bcCGl?CTZ>CO%w??&ICJx2R*o{|{Af*%eom
zt&Ktmg%uLC@Zb=f;7)M2;O_1a+%34fySux)yE_C64#BBgJ;r^<>HY=#%i3$s_2@9Y
zHWP1GM7oj2L51nJNkm{;i#M3|iu)|9t~{j2PX=)_`NtAXMc$OpuL+E_x02?rJ?=V5
z?FE>F9;9o=Yl$^Rr+AP_mC?zbsHXhWM@08=l@t1ni*U&O#@`PQ4I|2DV?5WH%u^Wr
z_-*PP*=XaY>I3z?h>!|_OsvxAJY0^=GQOuy4B8sokw4{0bQl;qj32Q%be>wYRLZDr
zK1LUqJ)I*zRvy&7t2LUr>JA~du6qqzw`K8Qq#&4ayUyWpX2!Q9<vx27F%HMKpGJU_
zC(7!vy^fnr`<{<G3bj6QhS*@rIax%46J%mz+{nN*p5I+<Zy0ve#b2*lfu>#u{I9q5
zY!#^2{M5TKsTvWf3qx=Cb9Ns&Z*+i(>F=wxtq#|TuOj<Y=wY-VKezPS=D2kK9_$j0
zV9Ief!KK}Dhpt}!6s)vq>Se4nkH^a{>z!_II2vjB%{fl8U=NZbD{@WfA;@vCcHX_K
zPVMU;Jt;V_YHm`^|1Qb;#PrjzSB*@7_(xmOsIOjXvonf^86HKVFBK=Gr9)Bi#^)^u
z7j{;V3FsCHrCQjcCF_>y&qQA+$I|OGM@A7w3oj6Ny93ZN6SV2)+M{X0Sn^?(Yi)UV
zdqhzC^jPWuMW35*@Sp78CXhJ;>@_ZDRci<83Y=MbB&-6V9LDNS;48|s6dthl<xVj2
zFF&CepI;Ai)I>XrlpPEH7gKeN7p^-uJ+XJAs9<wZ(}SSLXGxOM3mNR!KnEx!c=&?~
zTIKqc)s!T5P%aUlh<6Hf@=It9I{OrqGzMbRC)KakuD;CBuMDWq<`YJ@`(I9G7=SZ1
zBP%WK8xnm)FZ4MX+XeXyzY14QfcE3#FPve<dPG&(rG({7C}p=^UL!+XDy#A;d8*a@
zK}9839qJ9QFieJVDjl1CS<@BxYfAPk2H4hf^2fnmfY!R(ET2vX0faN*^yt!q+@zXI
zDXRLYg$E$PLy8)TZJN_0ESB8+z+%BE4SXu0hH4=6vX)Y`3D#imi3NkT4P`rbzZa*v
z7;|aiPc;|LuG9wa5g+1%N}a`<70z448{J|`JC`gXtF@AXoDmSLOZ-dl)#x$HOwE0Q
zE$Yrg1uxu=<iA~7g(4${k>B$JQ7YE?+ICdD;Wff~p>VWrGKa9}yetMxG!!Rdo1?`r
zr|&Pe?;8?D!>)!)7dk@8SaM_`>_<s#(fAd`1UwUl9~*#e=Bwv_)eXmpBZA-x@Qc`m
z{Psh%+oU@>rUmM>LQRgtw|_FSNNxCfKf0FB;B3YuG<C`ou8_Ac!&c)1T?shiYPpwQ
zXH~8anAc~VYEcXNy%)bnJ1~s-M!WkAIQg{c$$M{{w~cXp>pY}<eBXF%NlN3hg@B9y
z_gjbnK-!$xqKkGvtobjpq96&N9cA@cUKo8eUJm^!kQYNVgJeSO!?1MX>D`-hT44(H
z$f*~_v`894sd=uip^C~9@*6lZOpvGlr?Rp#+#{HDyJt_X*Jg;BYlL&|48%f!C&l;j
z6^1EiK`_ICdg}B21w3|0G$Za)wR-Vu_}#&cn`{u<3jg?asA9DE7=W?hB!-6HMGMAi
z{s#V}=b#cAo4k?A#Ka^;v83R&t);8=pSI$6Cg^)D`vC_h)5+AEQwR?&fofcY3|L+N
zYb}SoV6jba<tC@OAlu^Ll0sMwqh_qo+iyDY&sk%dy&&G$@8urntOHm(ev@+R+wKp$
z^s79gj~&?>qwyQcMq?E#T8Tp}Q}@fKT+#yWVYE!k0l{TIGm+n?bF(vHYLJzCxG-(b
zz_1ukziu=L!{E)cPEKKb4B09U{@q6Q*GnwQ-J9LPbIN-F<cIMq5a%44)n#&IxRk+c
zk}Zlk@$UecA=iG*<I5BHFql@OMw#hVy}Y}*7-DR4QiAklc{P99e7?2r(k|w<bqqgu
zOIU6HtNzPz<TqlRq!i2Ay*islk$&R*ec`F|Y{9a;J>3%y;nJx>CmEB6`}kvmrS*}V
zwriu7YV)f6V9_e}{I6g~Z=X>c2okOn59H+X+Fo9_!9jIL9e|4PccP1i_OBo5Bmn7^
zx%sdxgO+~2%PTvNeF23L`0rT5rr*DbyAf^YGs<7<k0Nyodl;}0T6ypW2=qAx%!L_p
zAS4fbzBiB%ABXChcIb*b6_JZN7&$B99?PDAp|5w7=gUGL#w@P0E!NU~79U{*Pxp3Q
zc~c%0BIAt;#>~8#+<jzBcl16QA^ER-s3yoh={c1DvWiG{2F&2;)bLvtmZ-g`@4IAw
z^6t1t#+_xgkc$ytP{w&X`wE?1axqqa2=!@jL?mv&MRKOB%k$({wRfDkw0kzTkzi}>
zneYaCAzbfp{ePWwmf0rYq`}ZJGCupJY_$+E-q4UykeMrec%-9ps4?7lQg107*dwa8
zNycBe2U6<Go#IqLv*NX%uE~Ha4*Yb0ufOR1bYCc$@X-YEv%zU4SAag0LwXdjarGrh
zI{oeKYk&g`gp-<?K$jP0sDdSb+W`%WQ0TKGnag9X0qPmFWc~^AxOwh*ce<J5WY2Z%
zfWDZiL#i%$)A#9WBN^}GqgKhyiqlDSqA~0e{0@fK{c)N%B3+{VWhoFwM8#)ph*(gf
ztAU6Lrc$@EoCIb_Z?Cek4{FEB0Tccz;REu<*4ug5ZG)|GURI@p$ufizz7tU1sZk?+
z0<xsj=qrP%W9?NkD+<8Nc3J-RM~PPb4IMXBO+*d$;%!yJZYc<Ul<0iv(&TWcL<FK9
zaeMtAOgY-7bEl2e;E7$t_0=G@z7eTWC@(K4TXknjH;&T{wW`{rWn4jy#z3EqZ_{#q
zEy_3e4(Wwv@uR7~NFG%~x~T$P(OZ{d8s^zy)PhRz_88`&8wooGrzP@|oZ{w|qAN_h
zT}DRC-VKjw8*~kfy_hd4e|n}E6|!ani)tsU+>*K+GxSr_&t;8*^G6mWUil+yeA2~u
z@rR%QA=`H^?8&k1Xw-JK9QxCdH9uK_ZDN1Qf7`O8BBZ%E-tz!V9P6$|<jH-ULchLS
zz3WD+RR9dQ8&Hkpgx{xx|4i$|`Mj{3lHi`am-SQ?Dv1zg|H~zCru}z0w5lp~7%jSi
z*)IF4?@GWGxwm3f)XX}RQZF~GNiS-VDG5GVh~(S~yIA&!tk*5=@LY|8yCP!I*{My&
zSq1+WhocTa9Quz>xkq;Ud;i-?ers(1h`JET186s2^6SJiF)+LlGSupBj0jSG=!6KG
z?WOSw9Z)lMf2~CCZ>e1$RFnmWpQot!8*^gczqSgTRzK<d{y~%za5w$~in^b-CnZ25
zP)H16CMx^J+;C>+7CHKj2<0x8B&RIF7DbG=mQ<bl{gT?F<?fltC-%ogu3SEJ%-cyo
z;Ne9KG*Ci34GpDR1xy3O3&Vtq;4joG)Op2#($^Iruq=@q-9XhJ{6*^CCmKZq0wplR
zNT)EUbQmTu9P;G}z3yh^fkVcOvlXlUetSlnv7RZ;eYc5#-u|-U$Z%FtNlpN2XvFG)
z8fl7hZp`2D0y3v>IJNcucrIwf9cFbZ{iKbD3M4O@A|yO(?p~+8z6p+n!VR%pvY;jW
zj@E><*zH?=aH2yTcqMZ-@N^16qj<Sk)^4+Mqt**1;8DL3-M0y)Yie+3?GI0sS2XLf
z-C=-q>d;XdK-$L`<liX7w})JLZH;qB$J}+RaWXDEbY*!EdUgl#3OSpVM1a7q#|M>a
zj8|iZX2Qtusc;^Acet4F57%{99xkT=`DlRfIzXio;7l<0z%uzGo{X^NLxaIuKwziG
z<r)U{gCYX>mjc!}_saL}sAY4_bKSp}TlrOuNB3atV$gzrNXq7%w}xS%Z{VUjBNA|z
zzF<ka>k%fy%}-^HB|H!I*c$WJ<>zd<l5Ki?nERTG^*`t_tP#}F+Zy#%H{PdNOJb#E
z4q9#`K;=awM}VuQTYL6(5p@BZXPt^hlVy_Wlvp?|a;4Oo>V~E9ca-et`HMQV@vn)~
zuISTcx<$0%ZRT^N$WZhscihOnzG{aVqC^92l&P)4d|Vh5*|nM$fzVadRN2ny9_6@$
z!!|B``bu|qvQJkwm=tEAdwK4`SgmX64M9#BvzYN4W#oNNHwYM%_cJ!P6bX3utiGsh
zsV>fAiIdoc^Z?_-x1VmepMlq;Fa^!~uEz&>fu|&&+~jl<OKSbmWy>oe`!%HZ`$j`z
zXskB}Hlw{plz17W8N{+qZY6B(-DkODcCktQ(mOJ+vSg4*_+&|&M#+K}w%z1A!N$2h
zHWk6!D{A~Z`P`a2iKsO{SoPQJv2T-dSkmkeK8I~*;(jI`wxqZ1hN<kq$E|K8uojc&
z1i#2ZLrr|H!(2M`v?qpGV0(Lm{SxXbw5BAJ<(4oxfRV8;lN!;Rj@wn6>s;CUTYSbE
z;0)($(wz88gj$?HQJ$LW#gO3{Na8f>8yrTp)7)+b<Q48Q8xj%dPjc71B~457Ftkmx
zYdgQOS@u~|dlHfF0{?r{Z{Me#T!-eH>w^nINn?S6$%vdiw;OZ8{c9O7GwC!n@Y;a<
zD%nu4pI%?MqBYgRx8?efXIl;L_Kn_4$Iiec5U)y8x+*>#A63~Jg9R$5p$uDTRN*@B
zU+$<8dq+^r_LAJZd*?6PoFTZYi+BE=>hL8>vwuPVKWF@(weHzKumU&MvD`mq{l~}=
zRZC`J`d2L6UU3+|9iUlJO9rRIr63muFcZyzJiKL;sdVcc(FNUW!G2Ck-OMRR2}?RS
zAjzRSkt7|n;NQb{T&h_Y999txMfq;*;3{8gn8g9dr)OQa#p93SEjw=3qPGtiHP+Nd
ztD$hLh}#VO36hi~$sLRnmEDoXB#iw13cWYyrxFg@#j)6aDI~zE?%fsv9tP2}tKrWG
z>VI|?EcWC>@qJm5w2gRDa>Fc%x@DuO%trU!;GgQ*FrDX=AkF5LUKgp~7h#Z!za{+e
zh5m0WfK;?EO%^78A0d5^ASbzNewgEwzvVs`+cZrUfz5PRA7%*n9Bwmwk&!h7>$Y4u
zKLy)f3z;zQYm`4S&<Th>xo=|%1`jJjRyonFr(eIxq6HdfQ^P61At8V$T5)gRCa?@<
zl`VEuOA1*I^B3bhX4x!m7j}?o5GQNiW#bQ{Zz3OAmyLfrE&ttT!8FrDzs=as(S`g-
z49-?xV{QC)ew);GHEhaRSFfhARq-4c(6_cUVC&7Tlz2#Mm0?SS`#UAcJMvHqZ`mZN
z;5^yP`g%@ROue0h<8VD<<)h8mboqE}P=sa5CrrG>r}<!**aj9OBkPrzkn<|<&rz{f
zm=P-Oq~K1)M}45{_AyG}2|NE@@<j#zU$=8Oye;($(`KSbypyUC?l<%KN0ef_&2yet
zFDl$AR!-ig<>=NA56Da1%E9uRNq5i3;kL9o7Of6U1GeO9&;Jardc$4pYVL0kD&$SV
zKTgF^$A0eM@Cz`yz}X9Zpq#icgG%bW;y3=J<KcQ*kXnhS4fMWbnWk7b3wwfBL{g|s
z4~MLTvJZ{0@qKYYSfX+<Y{{pSBD6&~rvBdmJJ|sZbKV>{!CAfiZ3c2{QhEGdtOppd
zBqu0na3&+DqD!=*vb2T@1`0!THQ#l-gw4%}bo7UzVJ+FxDx0Q5UTm!Ow$4B=^|pn4
zutwJ4<3AXk1e=>fB@K;dl35fC!&c3Z6?ub&&(?VI2*deq*~Co9IYY{wKqem+;(Je(
zPiM<@fRw|N#K*O;%J{o0fDt{2l@cpc_76@l6npPl8Z6Ahui>c7@b&d`97f%@9~+7d
zJi<EcTTtgNQp>b-TM=tX_4n25Ep|CTN7g>Q9pi^Fv2WHzMgY|sU3#)x-ymlB{+Wy_
z@e%X&Xy{>@%&j6&P|bdpiE&581^OXahBU5~odSo0VzVkh17H8S643uEaHlvxGLCr;
zZJC!*RRb=$lg2@N1^*%`g;?C`l6VnPC9~E2!Y({+YhXOq8|TDflkd(@@gj$?fy#gJ
zTBqNer;JbpxrEfSe7T;(+@#_yU=zl~bd+;c50^>ZAIv&z-@eEd@kV?whVWmnglb0E
z#$chsbk#iHW$`j9C3A$WT$gIKJyxFBg?6ifF##SLlf!p7Jk<jA1NQ~+**zhYv4G)u
z9FDrU(JFo?-kk(>taC}#!Ivs<iBK6~?)^dw^Z$T>g=0`+lsOb<um<<CPsR6b!V;nG
zO5Eb(H-W$)?O||Ycdx+>DpxH8M_!qw_xi?5%_r)sEk`*<5U+Uo5lKpe>?mcZ-{bjr
zwNBIot(W~vx<s=itRSnOe)*SVRrBp{vA@rTIRazfEli<`uIgwy35<J9MeZaUK4p+`
z>Ranpk#Z+9=5(G60QPSx$Dw*faN(zmNNUM&JD&p}AaF23K}JmMqw&!UpLRfr5H9Uj
z5Oj7F31$-4O5Zzp8$xsCL+{x33vLWSh)xgGr{`ZFGD%MFE{%<`KS#g{OhR1B&ED`Y
zcen9jvW;3SXtu6I$m{30tH$5H@D{Kn179S(X1Wd3`nd1qf@xpgdASkwco58rcVZZ(
zYkEFe3+<o^FxqIIiC-m>-b)~4pj%&74D|g{7Q{k_^R7=A18b209>(P8$g>S6@u%~B
zGIqtep%%jDF$Cv=`nq{5=J+^}&4jM62{LEAy|hxFzdvb2!s2}PnEsx0`^{+WeXIfS
z*Ue-(uo97UC?&(;lY3{c>$sdzQ90FZYOsfwif!c^ee{(ijcpMo$o||%hM9Xs;d-KA
zEu70JD3rFs>>druhvmQ@YDF^p5BY&KiX_#50OV9)k5nw=4oay;Y32N#dny7Z)XpKB
zd?mW$_g{AEQ2tdrT(#=@tDT9DgKOsj&$Jj{fGzucNu~JXiu!MRR2O&n!ai@-PHS?2
zt^9pgM}Bp)&rsj%i<}i>KA+itLlR%L03h@1-~Icqdk@G@>Bfl`8<ZP=8``SBbhb0B
zQhtJ6O!-fm{teyge^?A@yvI<eXo{)PUK(EGOb(1lB(hE*<;O5*5Ft!B00=;5WBE9F
zDzA=V&Z6V^rp4(C9t^`6apn8$<aC(!sy8e7g-#=q=No+5LxIf~@zB_R{?GC{u#5yJ
zD;<lH%upxge;20V<mCD<P#K@TX>kDMU-!|YIgN%R`LAyU48CA27Iaf+Yil(w<IZ-F
zX^vg0XP=&H#w@=_Ln^r~9|lfzNTTy=EsQG<xS|31;AZ>1e*Gi(1wl>Tp$Vvm^BI~G
zj-nu0r07oyi@u!xDzyyXk@ZP#?oZ|pA`ThJu*XO*&lihxHF_FF(Z7&V5^doIJDnzo
z=Oo_CEW%2Ne}R9bvFs5GZO@uHYcknv;GuTYrxRB<s`&NV2ekT;!|z>~vx=`<T*p1h
z9GMuF8}DCwkFk=Q!?`4pr~|Zwd<JhIsY;Z5>T3K%(Dn8^`{n9GUD^R2#x~1h#T9EZ
zb&Dg5XI59dyjLQ)S*WGPgDb~<wXq}mKO4}Mt2ekq<M{Vn&Us2s_QQjX_6b$iExTNi
z@2V`3{Gq<7=nD>w=|0_v{N&8vbnLpV!JkyW^u65w_KJsY-S|$a-q0Ja^@tAl`pJJd
zBq%=*&urGOpEc%rPO}L7`ud(!k0P0hTWzZym3}{HVj{-D%-%IBTesTe%@s4|V4(}4
z#_n@Uvo;>%Sr0bM|7Uc~8wyr%=k{hwAC9I<BECo<WU1o8{%(YtoXcG&11Yi1Xik5c
zplbW$q9dOE{YI7Ay4=qJxY`U)9NzNCydnP-A=c|9jksE$SZg6vrG0M+kBE<Vr55yk
z*(^=9RD0K;sT0+Km=U<386!Z3bNYt_SrzyggRtQ3xAe`C-@l4}a!(Mv<3a@VP^+?{
zwe^p&ilt=Xy%_V`j)EOF68Y!i;<EGUw<>VXQx!l-MPpsG!JO+~HEMp`;AQuH{yC9U
z3uJ+Eko7OA)_T4z{mt4s>Job9Mn`e=CD~t_S+MH`#^M5Dfd^TDsdHN8g-h<(d>#o<
z<Q~#9^J=XSy}&qrm>T4a=jBEAxyE(fUKpt!Cv%Q-Nb{vzaf80Zeyl(~H5aJh@rU(h
zlRWt)(!~>|62frlP4H!)D|Mt_Pj1d`b}2!)-m(Ml4?%UyF4`RSL}(kHC+pv`!>$6%
zm}V*cXFEM$oKXb~s>LfG^cpSs%n1pDri=21VJf~y#mE@3a6#qZ@$HEy6-lm=7-fF)
zFOOYUM`Q*Wrh6cX`$%y5W<iL@x}evH)+uGVqpcSORQIv#Hq^wUXPwJ>NF!1r-dxPg
zzU;HeS)R;5Cu8=7?+#gvnMYQX9WRWcaH(+aVoy|b%hhMnVeUvTx6eBdDF680>@&pK
zJ8a)u*=L#^`kl@Kg{Il%T7IRPE%ZF>@_8?o)*AlL9YY1IEf0)L`D8?ii7kDyF+DEX
z|12#cRU*+=niigNP=6U}lvo$&=t~+Tt=5kl@LeOX;#c`gtV8CZ7%~wn>=an1-MOrJ
zMjb?&71T5Kb8&a^n|j3$uabegrDDU2Z`ubVF#>qHjW0<e3Dqe3+%>`qmTf7fI!H55
zgl+QO@0fe1(}|#|5Gb0FNcey|nWr_(fHYw~4%b^N9472;c>zMxCL~*y%X3#~nGDet
zCZ<^H#Xqr$dp=uREtgW{=mG(u+oYf8r0^}jr%_wt0Bp40__{nFeP%^sv6UUZqv8G9
z!Amwo+d)cD`$Y)dZQgkV-!DLC`;RAvh^*1ktu-t0Tau<5IQq<M++9-WJ+7R|#9%r_
z1fULiiLplKFbZB2T5)&(%60v~1kBkLO#A52>`A*aL#<SCeZg6iicmR9e~n!$P8m68
zy|Jj8FF}1|myH}Z(uJG=gZ|In8{OES_@Jl?BH^;k{49Dt|I%f<TTH!rN8&xZPGz*U
zW-0g=a|BoF`#77CVr(`WPb|p3M8L%KH`)dc^sSYxpwR1Blw-|^oNug?noFn0;*Y0T
zs%F?<o)4=k0=Z+f=I_W0?hgD(a)-<NPtl%aWK<fc{f)`Re5!7M7wtBEWa{~o$rsNz
zkd`h_G@yh9dqtN2v~?_DzO%M3YU#ph^A9emh0dSXF^cYtQL#8_p8mGH%WZy++OvEB
zeAWM9;}EiaG~Hjpz)3Q-w^_l!z!6_~14ZJQ0JOGX_Cyy*oY6q}wKG;@Ful)5?dj3{
zEy-Z@r+h;bPSCCgS~idO%RNzXfK3<@?D5Qi5gp=1&*E*T5cq-4{3~gy4f682iJviA
zcXi0VL9IJ}np7WSQ4J+ul*+2$TnB0TdwZ~1XK@8{2+T*^C%q7;;xp>;wiF&RpVH$N
zW@>li`yQTW$PhY}eLhzFBe6qCC<4A5^rt^J08Z|Xo?V~dLTa7BR|AI52{0fk-Jy>C
zKm!N8;hBlSU4pqXPi$PJ&6*{8$HZ#cVHSIM7+#E?UBPQ_DgxXZ0`!qYf!&M%Bw+uI
z04WTLaxE78h;Q80Ak$Tnz2Pl+p?TzhnaXzQOuEzh?$<N<<zrsRMFP4PO2oWqIw8!-
zNQ|Jy%HCqWFS&xNj~rN{O1SJ_T-x|N=Z5eaZn+_`*n|86?h|P6G1e=PRB&v66e!vG
z)!Dx#+xVcbp^^4t(`V^zpXYF3uoV@Wd?Y&oW%W8T!#(%7724)zKk-@S<I00?x^TH7
zFk4!Zo4IA<*R`W$VVFgmt`hP?W6<)O>B!Yu%&4RLmmNt<!GgPZna{-jbQt08J_qEr
zB2v-!2{&14a8xm)TEeW}U)fFhLAegzR&b}w253!0WaVE72d&;4uL$fr*YEM^bu-t5
z``DSPN><htceqYGkh56mhM&%mnjZgsNLr0;82#_qF%S&LOT<*v6R~+zvllOPvdP<@
z#nu+tF6e(F7E$~qU0ut{%Bn4_Src-ah4szeG0AIs0eXiHU5AL0w-|x~j&RM8%OhNF
zb*P2cUr1#;j!G1(8VHL%T#TV<=_PtSS`K&Ssox<+IO>gje_Ccra0!|jwF=3S)<#@4
zua$c7pFW|g|B(;v%jxm8Q$0|mqF$=hiXmNwJtkBG`A=JCpuOhHMz#M6TiZ_+mG|gO
zITR<Y5MR}mLu)7IRA}7&Go5;QkUM+#Z<6_7L|3Z3NeuTKQEuw0-Zms0<i!k$p>24~
z$G_9m*4`6>)wud5E648atRXaWx$j6qqmwzhy&m)dA<!cijTh}jeQXJ_+pws5h;6g4
zI!3b|I~mmL!`$gV6&fHLKfU|svZNaFHe6zBDcY2tspe0h{4$%Zi^KZ`dE)VgS2aYU
zNYuM1&&XPvnqS?Nr8REysgaa=bXQ*S?$QQ8n)7byms{la5}q?E2)6oYG|pcX1~5b3
z`#xab@{jv(uO0X#lE3==9X2_9PP1y$!SjCLva2mGC$QtaH&>6*lL78@l^kfSf8D8;
z4Y%AMom-c`$i8wlc6WesN^NU4-4UIU@Seq-eU|ATcA(Ze*2)b%MHtvPe$+1Day(@v
zL0cX5(-}B8@$G8=eSH`zswbfeW`)bvOQ{+5u1gH_VXU#6ulk(Pervel`_#lEo@#CX
z@=coV3Jd?t8C#RntM>M7>$}~5kE2{NKw++s21kmUcYmwWuFT()_D?x_uz%7>LUgEg
z1hS=h<ptjF@e5PZ|1iJpBuwx+W+np=f*eBUMfaR3U<z?)tlJ?}om@k?$;sEa0Ob%A
z%89haE^|TIio8v8mQ^o(qujGk^IhyRuTb4*dL~ahB6jBRQsehK*ogXc*d2W18D@>l
zi1=snuqzS$faMm)nI1#x>CWE$jwcIrAK%ngV7&PL&zUOcY=6UJhtHUm2o|&BQ7v^Z
zgQe38N(405*7pUIESV_V&p|k7L{>~5O%eC}3D(1crasEEADV#M0Vtcn3?-%QWOR0h
z(YuO1OG*s5e20`*2=2vqkukUVy=f-E4r9*@o|ury@Q?i6D(LY_?Kw!|<#phmjeqqh
z0cV~%1L*5YOQgPpda2s~%j$g%@kt2LI};HFPTKw~Lvsuo#=Z=JI&M^d$(RetULNV|
z6;qRY`_d==lP>-rX##n-WIPV_qSPxnmkhe%pWT?{iQ_SN!p;7=ukZ~=sqkmedSxGo
zeV1`eDwG{z1$c=E5?_qzHWO|hJlW!~(#3mkz{Fp2%}8d7Y?jU?Uz*VGKL>i32r(+q
zoGiq(-@Y6&E`*sMU{b-rFx0HcKAvO`%)Lr8cbB_%4qAlfNMFHgz1rM&wmZ`oaq#GN
zow=EYi*P)*o)0^%F?~Gyd%nm@)&#L9{}+&#>~9L_jtHQ!6y_aJq_LMthVe^KqL`)f
zd$%+Nv(pwh!De0n!Qr21)m0*RjT0GdWs*sDw>}H%eeHY82*aQPAZomGXr#7P_-GZb
z5*xFq)~fW{H1y|ySD4)H3<j8wfkZG)38iLdfw11ht?Ht!3m8(4fa~j^WQUW_gn&n4
z_v#g5q9Uv4>+JbnFyavayUPhFL=?mw`)&H15Wms~YTM4~YG<pDjluzs;)4tG5YE5=
zkGccs$ir4vOa6OIha|x6=Vd&tgp}F!r6=8Fo4`)3fce+PuM+J{$~JbBch^uFVpnfN
zZ#i4RX!T=KHXL0u<ao+7!7n(N!ygd2X{lmCW9(ouS{5bj6&r@d_wq5`k-5Qu2Dg^?
z+Gm+CIP(=?{$Jmc)V|k@Yp-#EtzAPqI4xs^m-(9!VqahWy8cjAm<V!nR54$fn>Ia`
z_!JpJNK~_5vHmG|euQ{(Fj}0Z=LIAFxJ|EQe<Z;7U&#Wnw*AEt-JM*<oX$sockwjn
zF2S!k=ck^DeCnI;txpH5fJR!<HxE$L7#Tj)X@YJ)p#}DMf1bq(WK5|v)ZFVl?)Udi
z$WBA#Rw3tu`Oz9?=sFzNuF`A`gHj0JcVC#N$iow7=%qD}pH%Rll7Gc1=I3C*e(KVq
zlY<$up13u>-;u5F%E+0GkFmK~$Rt|d^+kCi18}QKr9-bg`eZBhHdRtL&C+iO4Bf<w
z7ZJ>`y3M*+owxq>*vMmb)8!-lDp250s#>RY+WJu9n(ATtrJR@Jrnmdm19kc@3;xFc
z$p`%}$Dt4-oM>K;2X3Ve$cBnAIVmGvU<QQ!?i8Z8G8k*Ao%qGq=e~vmk8zG^YqSe#
zDd-sH`R>KPc)Gf{G@U*!Xw(@FN$bB`jrNT_;W4pkgao=0LIwKJ?Csppbxj!ew)H|v
z{Tky63tktsP8C=FDF1V&syc}eKvcJ*qlRLE8vqA^6f{`voHWhVZH^epB24xqyTmVt
zI8U=|L>Yn)fU23YT^|@G2+q)(H6>31#hBx!6R}!=epfwK5J$BCtZzM*f*fVtA8ZR-
zr2=}zp?{OUhwg4xsrl<->A%hE37|Fy(ROW{vEHM`^y!=D_4Og`@Ce(eJ&4i^5oVN3
zo$Rg7e7&o@bx?o7eJ$7cfP*Q7Mu&sp2T~H;i6|5dZ3NWf7K;_XCMCBvc>zToq`Kpw
zz?gT%{-7w<-hj{*;Yh5bqa!EM&Z56YUM`pwzB46}7(eAMrzm-&_R8h7#^_;oV&V&`
zo9d9{F$;D-=Dw=q=PL*(zl5P;5~vVC*M7SAc{w34ns1gGb+4arn`$ygu~iJc933;o
z9qi)etK>{!2~$g4!JCDpl5KV1tQp$&@vo6|T|#uKcXy}f{+pAemh#^fZvKk0=epg4
zb<tqD1KMoMi)@!>pZX7TU>#n)+}HNm7<t$DC_+5ad<3w!bJ&A)ioY(guVET~1bCWW
zpmm&n!xeE@AuF?N-_EIXJ*pdDfP!%O_hao9B~=Lq2gNicqbf%LrK9M0UFK&Cor`F$
zD7_g;L-sHK#acnm<~`Pcg8HTyNl|W`1mrM9jgcDprIGa703o`99DNOXGt^UGx^WoT
zWBC4#d@$w9rZO_o=W&kU$EzK%4TeMW;2<I?M-jofR}V8qo!93wq;v-Y+zyrAhj*ZQ
zR!&T(j6^~`;S4*@T^|eG;2ow(bhaqQ*X9r*F-Jwc3mVm2kNPGF9e3-)(YU)&JK@yh
z)2EtWCLbDyMV{Iix+P0HJ~1p{dW}M(HwvKn)Tcs#5gQsI2&AM-*kFij9tDj{apv91
z%*#bj6a%(_=wJQ)Ev;}c&+8xt2DSO*Czz+5YL@FuUu4Fx^Ad0|;>?v&Ei9<-?GB<m
zITY!!v2>?ra02D}-fGQ$w$V7tro<iv%^HVMVh;+(?(AuS*%F{#0U6s1p!2sZ)@k?!
zMNbEEhODsFJqn%6rTBwZrWL-Vq@;4A=VVhDx-$nS&Yd*V)Q?$eevkk`u=G-iz%Q0z
zwHW1Y^UX8Sar%J@QzEIRzroZv?YUsGouu&$_MxAh4~Y15W(vfMdMShOJCa%Mt_wT!
zFGY~>5#rMi5t}(G_!)*OrC;5wHhB76*<j`Pnlmr#;=)~3Zrdd88>L-Sim^gZJyo3b
zi129W9oLVwPtH{dmD=Y|COWo#r#w^1hxA`f{;jyglW9|RGc{DKhi7G;QXNEWHNy$)
zC)J73OjkO?0X8)wF)$8zbbjI-$@~?>-S`wONkxCQbCA6PqRyQUBOY|Xt#z`hW1OBk
za}51+zS64uE-@qNQ+(N7?}Wl<2G;m*%LHc-G>%a?4=0O=xJ8CU91cyM%c#g&AbQ!F
zVB~dD1P<eNnq96^XP-P%LKbh+)Xv@WbVI_a)pagaDvT)Chgx`_en8UXB3v{jY1s~b
zRp){v_pOIJTiHqw*Uw!6sFf>Za7yGzxp;{ph{h?jLSeB?#VsAQ+$<;m1&MCgi!Y$p
z$<qjvp^&&*70rMK?f@xxFu`<tX*V+QS;hX@>}=58LRSlOPGLW1%5PZ@x<ZI<>1Kqw
zL}edmXp5G5wC`{mqlHU^IZ)^Zu1tPTUqPtvh}6lZYPCkMN8m;#6Ti46%ARoVz%kB;
zXu<ptm!DatCwi$v$~4}Wh^s;J__seKx!9)Q<DF7QF64`}W1{nR{9ar3C(}J(WG%ej
zOv?xNN5KaJp#j?pnaBMJgp@Z&@1qTRAmVuGntIm&4*PTB_I<QR0Hq4-GS)@iube8Z
zHGTu*mV;KZf|=j78YBv+8TQ;eLAf4c;s{KShZ>zioK2#Yhhfm!EmxtE=Q`#ugSbmb
z4|<kACe!Kc86o!;M8?{Gpphyj^bvecj4ct-lo>`_cBP`;EazVJ&FYsn7e)o|qfM67
zT5p6m$FEF`iba{0VVS+>$x>XeJEaAzt~Bh4VY>+Em6Y2y4X++UHXeTogsr3f_Aw5}
zWQ%%(<}30g{lAT^a)#+S)}=<YEux_me_mt$^}L2{?(x&Xc=l2QxO+04=3ViRJGMW@
zst<_cRM)i4BCeLQL^jnmZ!$;?X4HH79Nsr+%ijijn7a3$9bZ^Yjz0Q(iV`DJzw57w
zHT|C?C|4K&B6wpNFD#wuioQ-nVf^M5WTrrE(S$Ho1(7?vJ1H}PdmbJh^uDIZjw1jU
zZnBkD^|_)wzmegFi9exSwqZ7kAvJ({K+vBX)|)g92As<PR>ql1lIpYu?>M<yORj0U
zsmV{&6VHp!R9r;30@a0)RM$KA2njR($}{E0TcdGg;8{$%H#<?$A^O}^26Fwy4vzD%
zbg2T`&zTJlF~RN@(nW9s{%iR)$B73{iPC{}bIa*5%FE{Z&9o;??wA}fEY-76D-+*B
zvavcTfS+>8!9#Ss_!t^vX1FG9q2ql{VD4?`P58`HQk|PPl2I)+2toB1A`j9ThQiS4
zmNlI#w^By9I;o#*n3V7)$OQ7DO9foFxS?(eK+pw+l}wZV5bfeemPeDiaT<xvnwI^t
z8i&LWe7e`Z@-eh{kL?Ko8^k}4IN7?t9eX9VM(N>&y<Qk3we7c+EH|L7u%up$I2;5n
ziz9`!uza~(_SN5@IceRE^^F(KhVTdI<VJkyW7dwoom!%oBVa5@Z8?<*bJS-wu2+cC
zdnHuwYAED5=OX@4_dt{-r+?uu8&9y}icE}3QBh<5@jjq!W-RAnf<aj0h}-Pel&0VB
zd_|l#*y0e8;ZHI>!J@QInArbF<n->?wTVR5ov;8%PN~F~^`sVG?*CA<t>1HQCKn;O
z{*!&yeXw=`hor297*8waHo+5jV+%}4_SDTvH14!qh!-2F@ZEBGC$Si^jZQ3dh#h^{
zEuIP2iD|7Hb{eI`0^cBajdw3*HX;8Ha0HJl_?+Jz3=|>P91@%_82kQD4aVa84`;1w
zfsG`Il}Zh=p~s`c)6nPa-*zv8;ZGi<gBAY4S_8Fp+YoO<<C#otJo7lK16^?X%_r3d
z3d<k4mI-<jG7E4e@;OA*fPn+iBW4M<RM+5Uj=6Vuz&3RW;D{BWh-%t7j)?4p*#Gr>
zZ(LbSU$@mh@7{lISA(&iHWF$X8s05)?2bzyvREanM`iLM^JNRemE$NDg&P!`>*WEL
zH3ZJRbQTQbv%&ohn&nUpe99Prp<;sXEKOAt0m8N3BNJgITYR4x{zwGx8IbX9MT|a4
zg^&v{p%=lb%mZhOlky(GLBW5A?(VFaF+rIN4v@Ivm)~AwJ6dXLC68+~Lw-GMk$KP)
zO`R!A(NWtL_|zFAU?a_r;aVf<J`7n<PZ-v6^DkI_%3<D|_?WSae2d-0x>{q-4QOy7
zjrv<pqcT+;ViBc}4q>ub`)|C>&)UY3qO(%nAo$l<MV=oOI3@fz&*SVkr>Y4O$PDv(
z-aZhec8m;!zz2<l%dK)Z2VBiE?&-aRjn_gUJ?Wl4&lvkh=F@tbL7kJi{=To4&>q<E
zIwXOUzKDUW@COg8qwG#Kg<7mDuEcU2)lq71z}CBfG(e1g>vQJ$BWFP_#clkYG&dEK
zgmrF0r7{(Hs%%_>NY<e-)f&^>!-2HdihJSa<41Cav(2}mUcs7>a<*OTQOt^S?wJk~
zxPM%>!eU~I)hVS0(o<u&e_EDr&ZZ}?)1&jL2sl1CFX^j92UDMsUN0LaW}Anj_)PyF
zXz-uyU8>V3@tsJUX*H5j^xiyktRBODBoxX!@tD{|tCvBKIRvZ4viG`&1O<1w5_N#m
z6GKtS1`#@Y#;;t}gIO~*?E?!+%v%&Wei64m2?YE9vH*~uWbK!zp>S?FQiElmvaH?j
z31&8n?8KM_SRru`pz$gBo6Ih#bz1vAKcok9eiJ3)QD-CBxJm@>1Wc$$^yZe#l3K|h
zTdfGZET*x))lMa3aBy{|hR3K6a(hD_ykM`ABV^!m!ejd29;QrIJssS%UxG$xHt4((
z;BbXom%qpQ@9CdKL;(Cf@4U=K2Z?bO&X}#aI0Ni@-+7VncEeS<6Pd+)GrPxzz9K>4
zHsKTJPN@SgnhbZB3@0<8ai0JOzP!pWm)7IFVbwjKDi?$+=T?tsI%ia``Cr*-OGU1N
ztKpJGqxJcTVX&j{-*Bo+;&bU(%<+PFq`=52HoFP_{>B^?rMf$qli;iIBwCIH=`P27
zxszAZdIE%29g^M&^ORq>^3PvkjS{vK_rwdUu4*Ey$U90qERB91@Mb1480HS04dGSn
z%!_KcvS(PL$9CJ<rOwI#Y1Nck81{VU?p=F%;;M{HtX)*pqd#92J_oVwG0j_rlvJjS
z_mVIoR{0!pUmh!yCEBj?EYY8{^*T&g)(QbJoiCEuE?Y&J$*Q|pR-MwAcioSC?c-Ho
znAOWi|Be%k$Wgu=a*D=FM>@R^9<<4=JYpzsiM;>zzk_s<oHyEd{9)PEZL0j@JYyUX
zOHYSWd$d~@os$?(2g$Q2VOsUjdcpb3d`?iK3WIfotGlA>99e^xWB5Qz^Lrzn|48n6
zWrN!yM(h6;wApu%pv4WX6dwy+o2l|N-6w3_&jpEx@p^O`^@n-I`a(WEQq|}EnN9Jj
zfM*%(sn!(egh-@9UN8~N7k`8BvRn=3U5(l;f_)yuH<oy6L<G1>U0CZyVnUfG)x?lH
z+*u+s50X-RGn>^~h!qG7ti-cUkyz_N<H)!s;qKh6bQdU9*w)TN60T>w(Hn`3FUj$F
zL58x>S0&AwxDzUc_bElf6IssvTa0#DxA^W}eykK5Iudyg$>Zb*^6;SzM3slSX7w-R
zt=T)}Q3mb^3pE=uz<Ai5;g%whSlXX|uOzzTeYo5iFcR~?`wM&E+t8HMZ0D^9&w{s`
zKJE(%*Co?s8+DU<qMGhFZ9fTmpoh_!f*ST^QT<R&tpS8RbkSgQ;pVEen#<J3Uvp>(
z9a(fg9H;5RZ<~!Ae0-XQbBGfvei3Ht#JSIANOoOE(vJd;bWeOM?|9l#i6<{8-SL!B
zGf6k$Kl^UGyWU!TOynE&3bvA^8)NDEq4OueQw}1%Z%Ev0@cVtuNToeeM-%*MIqMEt
zjWIg%<(?rjQaQpO8`u8w&+}}D2uq@eGk$xR=w47_aRJ~X8Ulkr%>*{@Xs0VZKA>pJ
zx))~DLX1YC)OH?flni7awJ6mzMGJhZ(Y5Pg54^jr%o=~o6qnQg;yT{NM^$Cjj!30I
z3+RUYm?l&@L2r*1ASDKQ_4r_0RNTZG(WIMvb<)1P%+_8PD@V3oc}&IN;S8)oGQmu0
zzRvSqi(1MVg;mNgqVG=7pQSFDQH)oq0ezO7lz<teLgk(9y!XJH`yWfaR1$O|4sikB
zoqw&D5}st8ZTo1iWCM`m93LOQXAzTbvId}}@K0Opaty5x^kxI$V6KcWU;ELtVL!By
z8`BZl2aEd3_4Dyo+0NC5zYUQVKzn>{Ejo~dXWi3S%t+!5M9u%pIzB2)i|cm^0||an
zV3V}$WRmW!^niKOrwH0r|MJF*FnqH0uf_g|TiL*uSU;{V`{mb6I$|T1wh-TOD=^I~
zOO0=YzKCHf0KdYhc=WU+XB$UGkD8HjWp|*cHCG;Wn}QSB_Eg6f63K!DU|m*>@<eQb
zp_&^fa8-mOX0xe9l6kjZVOpgn6TwZh(pRj9ew5+tO^QRIYi$=4*iEXF&~KD~GOP4!
zZ5C9pZcg_1^H(eo+xnp|D-rs_Iuo>Kcn$dE$h$K(a{U<WCw>R6>EuW*dpA$@j~KyT
zw;K}ti&7hYhxB9e_Fm7X+EUwi@yD#WVup#wZ+ok&?JnQ-CG)nGNQ=+;k7qAT)`f7(
zG@Ml+tcL}G>B0Ed2xDotl;cxAUNzc2Ut(J~iCa0ZQ(;#Bv(oF-2gF`Mh|0j0bfdJU
zILyr*v@)&L$ExuMl(~K6?-Rg2|C6$iZr7VD<*`^X#G4n6&z*>WXxu10*O_e^DQ)gf
zG4#7fzx?kdct%pWE|-;uEJl!jd4*6bXw6NffqAeTNp;XMBSjh*+dH*z?|fb9Oa<7t
zeWt`1kBA%Zm1&-sglaf`b+ME=bSrx|wb};eu9k|<_-*(UgaU>cMj)(TdlPlsnW>55
zwpdN&9&+ROnO8A=XGe}x*NZc-XkIqKPAHRt<G`%QT}YHp@!uZn!TNt|2@<JcsF~#Q
z%mol4sEQ&I-QNem#U3oy;eeq8>M#bKE{_NQAAv);!cw!uA)GG@)0^7nGK5E{t_^){
z<dLpFOiYeuc-o6&kC~l4litXpU%xEpN-~G35@jIoKD5Gj0BUEBpNy!naPt3lBAOiU
z{=f?FED79Gt2DmGOZ^jYKuDt=<WAut7RWDNNWj6@N;p{(@)aZTe1Z}tS%}K52W0s3
zeb!Hr`ShdFG3BLbW-ohKvhcVI*pmM8Kz@l1==Qko`P$!N$2zkizkl<j3bpCC@G3GK
z{L%yR{pQn`Eo@MyR1X}T&Clfi2l8PBX%nLSqR+GS`Q5IeTRLx58};TZ{lR#^S}H&{
zSfBAY@6|uDpN{M};8g61-%E75bI;d0?c9s4clw>lZkxA?{4uGtuoHt7H=pE1?{hbt
z>FFFBf@sKdoXkF08>p?j>-~?-+pYjZ=K@lAd?B)qzj7zHF*sg=+ShWw8^>I>?Qqsn
z#;DB{d6-pJ+QXoRt?OnZD=Vv09Olf<|M;{=joiOKZw`Hs)+*{(iR|C4rPZ%25JiBm
z$OC14HKK=dxfAHF{afqZpUU1+O?;>6I`ILq4{Am)^KrsV=~gv|H2ZevUSIw!b^8e4
zVS=o-lTkphEm;B1<fo?z3DueZo;s5kv9(s;%lEh37%?||PPQ>AjJrw_Qu0FDIk>^K
z=Q<5Dr@5O`UG?;vP5-`dbvaSKjRwBB`*TnIM0RL;nZi-(DZ>q=rg6bZCq2u0ahnI{
z6ry`=--xT$Hs7{qe#W^FtY|Hgf@Fdz{_BS?b@NjoeYPg7D)VyGRyjv5fC7vcM|EGv
z_l82)BX2So)iIfx7#`hO4$F^alQeKfO54o{&+4!1sn#iyN2LHIcCG*NH|FBA<YL(}
z)yMja&eT4|9H@mF81W_2X(=>+Zn(L&10v@{_D%pt?Rrn2{kUMQ@p@AN?jZ?zP0*h*
zM-B=*@xsAh61Bi1<7lqUU%GZfHKGmx%{DL8g3-faT2+*efa=skkzUJPoz<ekrQMiA
z;N<`=HILmsKkxYa6W^u7S*z*NH=pU?zOp#rn>?i+pYl~-mPD%wzb1__YW1oc$~biB
zIlKrh>>E6P4uTx$ZQ#2b@%SJ4Pjz@D^f#;$>eevyLR}$Xsxsvff4!MX)xr6$)al-p
z7@C8eCotMZ2Hzxmea)#?_veB51GnaTpKi3rUzw2`q>6CwG_~e`UYO^EAc+{agMtEQ
z>41)V+ocZg$lJYD%xoNQU%Op=rj+A-JBYKz%NmlWfy|wQx`#`5jL(#1c3@Ljqhg?0
zO*9M7V8g+K&A~$qYX3U5b;vCBtPoOji#+={8E%rh3%suKjWxU0gW?K3mbs{T7JbUJ
z*eK6S6rw#C^V5DO_SO@>0eg7i3Rd;zjI@cLpP#23pJxr1CktbQ4U*rc$!s5giC*_H
z-l1c3@*zB4Dn?$Xm@*6h3iXewm#8xWVWtFS3sOYu^_^&M<BKtJ^k``nn!@}vMrJwc
z%Vqgh(amI^y*u2_hBSgNI>VA4b*Vm+8#<=|;J0)iHYhE~&+Bcd6`J$Gw+b`}<(&%Y
z!hMI9xauRUG5hWzha9l>C2`ImXlR%C)5n)RU)-e+M8Zi_>Vq6EoQI9(pyRv!C}M#R
z^Yy;tn`(C)_iVV3pZL&CvL<dtuHMB5_3nLI)?vwPF#k3uBCPV&zTfhG-jv#*=b3Ps
zNPYQZcN32sb4Y@bQMd<$IE+of&5zlQ5y{qKw3mk^>SODfYvN6xPexAfJ?}K$>8S|4
zc3gBV(}LgiUV!(d73wav^+J4~W*pwTtY<<gT({i%zb`pn1dTpf#vKebJdILeKgqFB
zJ!n4ZaEv*j&ro(TBR?j|SG)Nh)XQD#GyXT6BZ1qia3Yo~-ayl2{a8s#jVN|}If-`M
z05KnYgh-~c!Qc*^n&kbEkRrW#MU<1$RSr&y^4{{Lac9eTKc>#lVG%;u!ZSd^rdu6R
zl}*XJ2MbIu(El=?Y(y^U%P~Ypt&I~(02D8p<)IQxnnybjvw}X@FQWqMoly5g^R(pG
zIPxPm2mi4#TEW6bx6n+gnk0RM`xKV+0oo26654dd)Dl3Mc(cPp$eDWMF5X;>Fu8_F
zqKRa9ep;wFJ9Nnbey`uJ<-u&JiIhc8KH67?eZ+f(1ZUqXvddAVcX$v=%+y@2SjJ0_
zH6FZ>!Nimwh?tj*5Cdogt!m}!bX}wgM9-J>{_?P(|F5i&@u2_Ce(sz3)TzUV%k)IA
zqDyD{a}Y9S0Q6GWC8d8HX|S%G&f=L@dBy&oM1Rgf+Y<qdF5DBPz@#r9K8w9Nel;MC
zS?z@=Han8^-a0cNfye@TCs4PFyckPXcQA^?hq)0|i{^*D#8uY9K*9u5cY9IQ?|-c(
zxoE)aH=VwQ?)1*ryUMbd21Jf-G<EK@V~bAZ#Gpv^<~_=Ta{Kr&$HbXE!k;rdHn@Ai
zgTDT^c88ulVHbybXQNEg8+f)`?M=isYPfad0wK<%ffZwg3)Tu)h6pUj;=PM27QJJ$
zO<Y9a2uW51Nq}v<=GoN94a|HXna&eg!E@~vC5)l>#NxCxTkOdXR$LQ`O1NDIeixAt
z8O5A)jqyA24{(6p-vZpC)v?hNltlnv%<HqAzEZ|B8gvEFaFIBo+axrz#dlKQuY1Dx
zR?dUUyu8)S<JiLF*-(?2o755SscHO8=`8CW!z1N<Z5eTdc>gK1wBoF#tJk+lKIIlH
zIu15$POy-WPJovC8l^m@{2Vs2%QNa5FjXHB+nqv{aP{|30XxJ~<qy65FJ^kT_~mNi
z>CD#N|Gag;yQN%5REr}wZv#8--<fH;8Z1Ze4}sYguE8hPk(NQb)QeW4?|fajS+Z%%
ziHUHk?YJ<PDibI~tq+Z%+Ej{r9WDt>o#mIjupd6an(U#I(^@eHJd2(~kEjI{3+_S(
zMSMkdKZfVjgWF7F8p)NmAn6nn_qMLd|6&G7O`s@7I{bZH;Pt{I$sAbVnKhd8{SXo#
z{UR~Cjknre9*Ut&?d_Pu7q0@Ztn?G=f5LnbPYY>)%|>{u=~Fp)$(>D~5WGkuL=Hl0
z`HDWx@=lZH3*b-)7kFY|hpGH|Sy;GqMpbiARm6LCE{#i!OFn9Uc9G`<sw6$(w7x<=
z*4UQ5);~ds70WzVxpJ2E)n<JXy_O}}&hXs^T3~%eej9Soydcl`n9eA};vf7>;3}f7
ziuH5=>4dDL=dHX))Bjv{@#2H<A@2~Znwl!_UVr&olw!4QDCt0qOkk5ohfh6$Kp^eS
zVTmR`OcagzT7JBE#rNhL_cVX}m#UJRLAVhnucG&Ag1M`Ol_{5-(TEKf&r$*Tib+u?
zc5cab+Ujcqo>|c~u<=Fd8*e3pQ0@|J;vD7M)Z(#2kV*>0di%d0Z%6~29t=NSS8g*q
z1A1bdUJuXLdLOS%6qYjVub2q<a7G+>hnac#nL-7KT?n*f+uf=Gz2Qp|B-F$L9`Yhr
zpd3e?0bgWF%{m|mqMcZ$_d2hEXn61yEO(LnPurF`sz9PU`dr=>I11H>+f;ZGE03B=
zW4=v{O%$2y-6(;Vj$_z1noTzUA5-5LU02t39jmcz+g4-SPGj44W1Ed_G*;urNn<CC
zof9YLI}hIHyT@2#|33SywJ)qWuQ`{Q)Au^pTR*Iy=r<X=jL=`EX>uTZ{}hsa@bCtq
z?-kW;kt;QQY=Q6YK`3U|>vy^y#X>U=3?DYM=e`n6xuiOekoH#Mw19VESID`u*Awzc
zC-^=KzJDS-(G<)g5l)r;6uf2+{llXLV{(l4r=!zCXnW+gVvv<EC}&2WEjMH&7RRad
z>K!a?W&c4zFl14PSxQx3Efs>YVZUIk$_c9{=!S}B5{;#@`>b0vFrXVG1gI5an|cN6
zs)(0~7kR$wJHvi(5510_Zq>Uh7wu@5x2`80^Y#&0y*W6brk(Dw8}4R|%?ys0N|)n!
zWj?o#FTph}5aHeAGhzCP?Zp<TKyrt~9wEs$Um>nsHjR62$>?TkYE6^^rTz_e6@c@!
zH_#ZTwdiT~TI%s@&l;HtG1yovJ_FVsZ;-V55%B+HTW*mHw@@5>X)$;?n=bX&&X1k#
z@fnExhoB#=2h3{09<C)EkeQw)CILVzR{EP&w={kYe8vn49q4OW!OVf>?BH<Ok-hi9
zd=oz+i}%6nNIheY)>q*&B#8IEGXhl3OCIt}G&aTR=m2*_^zcTwVmljQq!=uRM|c<i
zPEV8EtS&_lKgpfe2-BkX*j*-&+LU2H`YSQ?^tt<cD|`nI4_L)#K*j5pfm7VT21jso
z+<4%|GRL~xR|y}V7oN51>#$M}37vjzROZL8zxL8ByvJvyu2Qw!D-Zj%sOEwttF6<&
zAb~-FOdZu{xuOKBpA0I#f-Oh!DnY{Fv$lT2V+G-)aS31=5FPjxUafIO=CxV$AJIHQ
zZQIhbQilw#w!DD-v_|5}<LHj~)>qU&laP8Zq*QXMuz)^97!kUa{MyQn#!7j!!AwR-
zh3%hFA$9-)=F!t^<KWy72Fv88rMR1b=BOLFf?>VSn%=5w=P7sycHz_}Yjb}&R*HEr
zpwz<7L~xL@{}CJWZ;_I4g*Zg7d%CF$)Q!7j3h8Z`UOHM0?}FI*Ulq_0_5c$`+5$#)
z?QZp*jj_RoXyj@RbO*?1alV2nhxeeAMlE}N4K1zqw)*<3c+ky}o`;@PU|sZ>?X>v!
zAV!+yZJ~_DzUzYtX@>Swzm1&A_^zCL_Q*efZ0A25nd*fssUdfu|CI4=d9{fY4CRA<
z{=9@DAN)b_47C9t;DPe2a{IWdxT{VUI@LXimkH_1qD}7ELRcvo<ht+B*nv4EofVZd
zA{2~H-*q2`zzHFhuI~h;fnBhsuNPM@`aIUQ?Re^ZtknabQQ3<>IJ-cNXUu_?;HWJ?
zmthG*DH~r(tBVZMzcHI?<0Z5Oa8A#nC6bpBim!_IB~lx)hpM@Lm=2Gd*YUu0Oc5%k
zH0?ka>Vx9^x&c?)M>Q<X4+Q`GkGM$d0@LaN*%EbKo&~9?u`YXmYsp~WT;!kZq8RSR
z2W;TMYtTyg=6PPWYXGG*U73z9hfq_yo@}s!g-(%5;e5se7!F8a7Yug=UWdnHI6QrK
zeH*3)jx<@WOnrBza4dB(X@SI~6{r%QUw(tPFVFL%b%QiT(g5rRLHFe$>KtPJ@!_Mp
z<02)eDyP0-i>A;al9XZ|+lAQ+^#Tv9iiT0QjEwNb1ICW%h-zJ|k7x-M7BjjHvebMp
zab};SO@HG~X_lO|4|VMw0Ol|0(rp5+(!l@B6D|hVPVOXIvtDr3`#>9W)VHlCYE1})
z07rzu7F5?RUlvS7cMe?WMpW~+heV&gzE>*M2+jW?b(}2w+L*kf{qxbI?CUC3tzF`<
z4MC>=nfpZXc?7#nqu@ezuiUCm*j>HV*k5aMNP9yMt{Ty*t>VY*xQ2qE^hq!Ij~I;u
zr8ZULf!@KEs>iC=QlXn-Py40vWm{c6fX*#|IJLn`hiPoda?UneZ(7FTZ;A6D_jiL5
z^)kv?m2!?@2(AT`XKQKwBrnN5i=~*51--k4jP|p0Y_m6wosH~W91Hfql)=8Z&*j1t
zYxLiKywo_gbCaHb^-iTP6#?0rmWIC%fhJ^lFX@{UJ>rrp9pqg`H6(T;WQxe&wSL)C
zRSTK(Ln=J-$@E_yY$~5ByALn-sTE3GSE!-32^~GmXi*tQEl&FTCQg-0U;CzOek^G*
z2#%>s&d%8J)}3?}T9LB<5(>F#UHZ<Qrw;=G$%rU0Gwsk^mYxIBwU9fHv{b=IEKu2o
zMG8Z#pH$X}Q(L7byUqt>rP^m5c?(PF9G!(%x}+r|5;!}1N#U%HQ-QLZwZ{xUReh%U
zT!Sh9sJ0q}6{8@{g`II>Y#l5{!OG7CFZ9my@Hg22J8dfiTECEgaO5P%z(%!qpjy-3
zQjS&Z59<}@(=3Gpd442CW3tvtoEgQaCZ*j-eZkc(3hpNESpjdwxsuF-(OgiKLaQSs
zk62UZUH*bJw;dVkBRKg>`%7qfn8F(NH>ziiivFm@f%vqbY69K1wayvyc{vEZd<lfv
zIOp=w6~LA#CoX0$$9}ko3=#7>yso{ZH_K?!_3pNzk}jus5d2fC24bUHP054cpcZx&
z?C@(%OwBUW;X~fkgpB;Fnb4c7uhc`mP>fv;EWg^h$Z@R>OiXf!>C%De{F^|T()kDV
zfx&6dVXJLq-xj;k_ku4%7_UKu6$!BmyFqkQQ)%juYv?49@$%0w_D-*MK}?XA>;s;z
z-KRVTlWwo5Z~#I(k7S2%3otcX@`DTvP1?y-$F+4|E6J5MkG?I5*mHC9B}{-rU+~@{
zeOSzmK;t`@_C&9D%p3a#seQIsw3fo+)4osMdV%zT)V`jU23`yz9FdXUf+%C1U>{L^
zT0z0({7<~CB4YL5S7Z+xpV5K)(rJ(Pyo#i2af`h(-h#Py3fMxB)%~taRA(Jk4tlZb
z{fYqNmC68m5S@ABa4Psw<_&Q`3`cH_ps<oSS%$hSuXO<iGP<;8(yokvefMo5vn0}T
zFSby0uguE|N2b{Y`*Bx(RruXJN9c6s0liW0X=N&or1TU5rrQRanZz&PsRd2s$~&4y
z*Tq(gTl*XzAGP6UQ&%6ZsisA)lsged5VnbbI#&j5zD6N8P~aITQZ;73`j47-!OR4t
z!)hz;FT|J))3nSzUz9hI0+!b(r%FNf!NX?nE4BA6sW0HoXqMXtC)5Mo5#n;_C2i1s
zgj0W@J26BPV8Oa4aI=#jzaNJIqv<3%;Uhd?uH%j12<`|oBll!38mXmrB6Z!~pe9{F
z>q);Fd4u0hq0uXxY~^Y@$)R^4DUZ_+Q|%kUyk%!EI3-%e5iI;T$JBiRrq!6AhKQZ;
zXF4j8fb~zt*I1BVz%3%HGZ0QnN8~R7kS{MzpkP>XbY%3P1I%p(tbx^dBiT-pdO~o$
z6@<;_+-|I{dc$5WUWb+w=HX9i2Q|gQAl#3FhW#^QC3ez0c<aPu*$nX(#z~c4;oB6K
zrIt&BN9RV={R1knq-R_D<n@>oJ~SDqeuHgj1h!RISLcw+!Ez#Nht-UF;52=34_~)t
z-e4Rcjp;ahdqUTS?u3Sdo6OXt#Yn-8n%Ek6WIs<lYm7B)B(P1DBeJ*E7+Z*HRY_WK
zNg}OVF!u!wJmG71=Ck&ZFA~*jD^qNeWbg%xH<(z=z9bN=i0afIF#;cpIFioY9en?n
zvl<a`hXx)%{)WvE1BL_9mfLltf<vg@mxJDUsb-kq7DuVL#x{`-^Kcu7BdQBV2qx3d
zso-lGJ&FjqjLuAXYMil89p82zEW&Q7*V>QJZ!Y23CS}T~5%nAV*owZAOzV&kW4|dQ
zYG5r3l+<ms`2uVKHY}si*XRXPHsQwh41NEVQ0k(CwB<?AT(faw8Zq2Z^l2PnR)cxG
z+IrJ0VnNv}{QVTjWU46BwD5tOH7mf#-scM*mcXD;68w02Zt`mO{oMOKqTKX1ctJ?%
z4^J<yCvl9bnW6lIRMYrLK?Ir-`j5i6MZS?L<d5iQ3Q#9+300TI4r%wN*2Y9$7qC&x
zCaZ|)*OfM9Kc!`W=<T@B0;5B;$>F)S#6JA5>cH`L-tz3~htA}5uc047O39}A7TAhc
zMtODv*MMAVrCMCZGIn_VeHTH^&VDC6S6&OcLhp|ACclBJxfVR3?>zaKgNhjwiQ2r5
zYM`w!db`a(<!oUfYYSXCWD?NuF<+GTW$CQDdcX!a#FJoCCk`#bElgB3(Cb$Lb*S#j
z?*Y7vQ$O%}Q`V_xUywQOX5ofMee5@yqhDQ25w>lI@Wa~{;&FGk%KhrW(I;RK@0q`d
zQQd{HWy6^FV`ED*;MzujYx%YR-|!Y|BLdFp^y4D)&#b3%Ey|{Ec(nL1f9mwLQC7b=
zv%tGs2bc8areBQ&YdyqkDQX2>6oV*wl_)lOtP|%O!7;Dx_%v|@unQ2A;otx!PYc(O
zct2i95S1)2r0<k#FB`WH@g+`I<H-5fqm0+L0eRnPKM$`8S1YYQ#BGN54=H$p1Oh>b
zCJ&f&L-(1ouXJWs`}Q&-zV+W;Km38alw*mr$`q1sF7jceYVkw+joI{NJY9~4VF#}?
zR@2ef7x%qgE3AF|WapeiUvvlU>TqgS5QXVh@5%d)ZQCRNDFh;y6hfz7{ti3x+R<p8
zt9oeXS$?_zr({|?hqA|FX0F~^(;J7+bp{jDBxCV)G%}K05%~Lq2&vV2NFMKLmso4)
z%!+;!i$RMqd<7+p1s^O6`7d05RU=b9A=^N2;p|5OO!XrN?t}0)9u*JT7qd_ju{B^s
zE0J<`8<icoH0Qecze_~XCU(-^R+vw&@hZ=Qiq#p1{2@^7I5G;h^Qz58Du}YFDj*dZ
z!8M|V9Lr+|WA5Al5B=3Di~DM<3S+dsV3@nXZlE%~ooZ}f&IPNs$a!m5`ons;?Q*NA
z)`x}sGD+V$ag)H(opW}AdWgNb4YZfEkQepGi1f4&_Z&~K#7*P%!4{tZlV8Mmt}JX-
zQg8^JsB>u3*ouyyky--!Ln00RHhy=iMja`Rb1ymI-sG9LZdyfISlUe;8P(Q{jB1a$
z<961;;Tkvj{-GWXY<;)4I>H)rAe4brtf4wob#0c@Cca%O{m0z;cw`8OIG<2KbQ^Y~
zR|iq#sd{a{aJFxU9sKr-byiwO@~oVBMx<g5Wag&~gPyr}mc|;x>!4UPD7GjGDXbhp
zD!mJ&MkF}>ariDYAWRa<P=Z_~^qH4z!c=L%-WY6UtCeMv_ZEm2_LadcJLId)3vQ+%
zLP6b-gf5q;mG&qq-oHw75VvE%4%46p!qx-CLu(~)aza=)o)UFoANUn&^|IYh@PK3r
zUwzv~pOU^{x!c)@VnB?2U-RHZen4#|KjUe)0?w@ClEnQ2AHGN&%hs;4g^KHO4IjfK
z6p>EI23Db+tihU`WMw9(1d|b%kkT=s1JnEstOUVIR7=~lFc$ge3fJt#g>YFr6%y{v
z<N<YEa74Z=d*e9E(xuDCv-^tE>O`jelp+apy%SO+vQ^mzp{y4)<BH;vi9Z)OMQv@r
zh#;I3Z&O!YU+P?pW`4TfO5M&Zy`H#0u8Q=^aIztI4Moa@&;RJdn0S!2T1yxs9|k7c
z2>g>INDIe9q6QO3>IGCTc%L$y^9Qa5^9)s4os8l|4F{C?v>Js<k-FuvXTPV&#`)|e
zpIuVHhwqgu&qWZ}9*49ygom8i-D8;qV32jGz{(!65H&tHx=UjY=;Nl0O7gCuE(=hb
zwlld*SO1ewiUin50~)9+NwBc>u7M20jQ~bS$!V~iw-^6wmJ(D|5o`KAEJ;J!FMUEV
z?F&(O<3cFZ@_RcH{eS%~4NWHb&Z?=gKeW7Qhgln0IG}S!l8GTVJ6<+Bti|Mgan^K*
z*atXzVDmRSB{s{)Z)!EWNX~D>2nN}cq+ta0us-gy0_i9n4h#CrMD5wq$hNKjs72T(
zdrJK9x&Qdp<aoEMGp4*0e6*|~=(vDyQg%1LHPpBBi5-67*a9V9(U!2Hg2ZF==lx%r
z9yOBi1*z!`I{0LLdFelue&Z`s<GfzoC=_q}{72%<ngyH82SOiBh{dE>*(B30wTh+1
zr4}=p3r8*3woIxZi0K9@@Nu?iu){ogqq4tAZy;X{jX<aRKA34~322(CLqq1Vr~pae
zZf)4Z5Vu0H!W2gAT`WdGO1BRQAe*vJJuh8AxU2AR%d_Jok%O74BCl^pxVz4*D9l9+
zD$I5oHz^fgJ$E#g8(tFG8}{Z2>i+)d=;+h@9|SE^Hk@`zOFWrE@V=)f)VG>FEUZ>1
zWzrt@4@~>t<ffw6>^YuSsCy$op{LJ9$xxwST~$>VrjV6~C|`PU;<m*)Ojt-|)#KV*
z&4R0RL`)y*8JZUpX^x>5m;3v)d_`_A#5DVUMSWDuUI^K%XD)s!yI_ll%Lx|VN44)6
zh0{(_r{!1FOV8S49dkoxK7i!}<BgSD0-&VZH6EC0(7`1$Cu?rd9<H^=7qOkStFDQ$
z9K;6O3VtJ+`0`E4$*Xl!jSe(Ex~lUqCSeL1o3cz9w%WcVBUyXK|DAOmm^@Ha_eZ5#
zmzIQ{%wKV#fLoa6)*tAE3oj8%Ln-#|TeJU#S0b+_Rd<5rKo>;4hr^*%cFr<)R-f!5
z-Po)L8oWR}Qb;!kaoR)>SM&giu8gzdc%PBGg&%ajFh4uwX~Pl0!|K*wfE+Me{rV5k
zQ$=7sBzCk<R0imEaV{_be;XLt)k?#u0qm<pL$qQ$zH0@?h3?65vbAi9i#8uK(!u)&
z&Z?RKavx%^&$HzkdCx{h+Om(+48*y<7hz~%1cGw!%9#vJ)FmlVv)VhN{cfL@N?gfs
zvEVicWI;h=uD+ffiumHC%A&4KYMw`_GtJMK0YJ16-C2O=J@Al>yvxL9r2fuy@bdHf
z_<U)*?t7v;mQ9%{g$Es^UQ?~PcHR;Kr5^pZ90B4pE=sG*B=HptA|-t_z9PsPSJ*P|
zl?69!;>pi6*drmeR%jYvx8keR+s@Xspon*~WfIIV@h8--ZMx`yyViBe7M9=WqgIK)
zh;K%|Ub}zD<$R_%J16xTn_;qUh}GVAWxL*-)2+*{rLFTHyz?YCm~zEjZgh*8hN18^
zo7F8jZ<QI2$$-$-5CexYs`(iCDum5l6T0^ze|>4be!>jNNl&~Xg5`d@uwz}qM9a;@
ztq!Lhl#TCJzxHx+a=PyB?z|wVbu8kn#KOQnTxRjW=R!{qA#=*=LP7`Nl5P|QwI$o+
zM6D1CeVLB>61zTLZgAZI@9J>+AJX))mf>8q*#oEfPG#P>ykUM?hzy$8C?%cZT}gf%
zObS;==Ik%XaT=b7%s#m(76Xhhl{>(%`H8PynymrUcRR@2xgNRAUrTOV`!=V3?2T}N
zqz8_xNKzHQMf62_o*EB0>wiC>^~ZbCVao&u#3qhr3pPy}JhDrS-rp)Szvy`5HMn>o
z?4@wsO(YSg7EfgXkqjD%YGnspBP3yT=!-xKmnPYT(++vfQQhE>=?Nv-I#SW+G@b;|
zv{2%Zk4qBlXw$4bXvHIGRxmoUwjJlomv`ShDy(zAuE@vHOzF_HIHa*<o-ZQ<@_+qO
zDhK>Mss}b-a}f@QSQEta%pS2Fntja}J9Jq1M+9?6gG{ASW>%&|cblkp$~y9Ymo-ca
zuxd_DF(Xy#0;^m1;DSVovS+A2C?l^mxCd?sV6X~QvtMg(T`5_?Ha9}eqoQMrW21`&
zgV`9sIK#XTkdgC$M-H2c`mz(pe&4zG^EV>Da7hAWC}|ji%qkS73sg=x7<{BXCncpR
zjVShbfsxbnb&xuifNaI$R%eIPmLKr?zUN6}xsD%DU<@Lfp5=ASyujP|@G2$tePfA0
z#x%?oSK%9z73Q#6D0S;#DZlr++Uz9AMm|7%-X_tn6fBqeJ965(s^t+fK=_y)`gc3i
z`sXL{)1pN!O^K)nS<Kyf#OOPaR1!#BAx8pDm&g*cLWEfuyxnF@(U8E{N7r7|4fWG;
z(I>K!*+Lc;oUitr&solX#<zYx8-!(g(-Hk$L%^WL4Pl&pQ&uN7#47xe(-TBK<%!OO
zdlxGG&bZ>nYc&x-I|6tYx@ID=wzgh1xZ%Kggfb@e4FIFg{%dvOiAwh#5Mx&&OxsH$
zzeH?C#|pmw^E21)qvqD;!E{vBAd2RTzn=7t2LwnDaO_t}cc9GaWmRPFa+#IqwXT9y
zCsCs>ir3l9QkocSEv@x(W=2)sC?iH&Z&FvRb(n9wC~m73v}xL`Dhd9Lwl5n(?isEo
zL!vR9L__#jw=WbXNS9o^x6$mXqD|k}z%f|=y=XlkIJf<CIVmY1I2-faRm%8nRzEOd
zh_MOyQjsH#E7n0Czsy6bSE&*$va0iNJn^DPLN=LUKeZSzD4O10L4lZZg=sCXI_Z?P
z-Sili&DPR>m2`;B{!P6av5n1-qNqL6_raX=i4AGxCYRSZAR7>agMZ%-6|Vmc^$m>}
zsc?*eSm74j`n4#+nq5)vmZ=$fi+Z5fJ?FZ@(Tw32cr}MmeDw_FT4cero9G53wF(-P
zw{%x%*R%HiuWA&4#-B?{Jzs6(gwpH%HCIvl*R73qhqt0_eK{670eyvhcSwEV(7!NA
za~x?$G@jd2pq9SqohQLAm0&MQ#eqeHxmgj*&G(FHhpb6^4E_oQD9gB!%}vB7YO#Fi
zo~6diz^-s)4qH3(9QmXF;v`Nv8_45t877#r43<qEjcjgv22ICq9Yx0#Kp|jqS)}Uo
zWsP=nYU)dZYTVr)ZiaDtA!@i5vpZRI>G~1V@-mi$Zxr`<PTB|iZnx%?VfDoO9|{9d
z|5<8MyUz!SK_c-yXW^Ol5?lycZCYx~_f2n0tD`(M)z!kOu73C5wn1sZxODiCWPjW}
zzqE?r)O+QZYh8Jftn@6n!kgRu%W+|UL&?;YZ%$ABBHCQ2k!b147}`_V2UG^h^m#O)
z?hm+Hs&DanT2YaedDsm_x36+c`mRvLARgQYlKjeupzXv#jxbZFc6=2a)NaEJX1vNb
zItK`{6}5ab`Y6T`vLaR09nr*C<^GJ<F+ZR7H$J1h^Su`AWH2S2oudaJ&UE~zY`(Sc
zd(X-^QHDE)^<UrVAql>e$IL2&ZhcrJe0O0A<FZd7l}ArW4erz#`TYP*FOo#zjdZhK
zWmEWH77@bMRnfB#;@ef8yuYy?FA{{v284>)-`{S#*437XwMZHSXpzBhS)x&Np0Jdu
z<HQ6^JjB>?rvq;1Cx}05^59kMRH5H4Tg%wu^Z*{l>xY*gPDUyGS~g)CGmqh>%QbIV
zeT>N3EiKuj_DMDi+61pdv`*-j&46r$MIA!isAHxDM=8vRK=?Ci_G?E7YLkDhDstAq
z_h(2Og*RxaqdW{9N0XOE4ZO7`HUjHBA<mcBBl+X$+zw`4mZv1-G&|V?NEd!-3}Gz<
z8|t9^@cfDIh82Rq_xl}M@a`5)_fI0vPttbH_!(Oh_2asV+Qc8FtU6+yMaUzii-r_M
zzl%zCt1;L0G*T^R^46q?`!Oj}-q73y_$}P>TP3*~o0SS8J{WHsYRxk}Z0%&X&`#c)
zl`Qx%p7Ibl6y$}&4JK%dX*e#6q+jHZ0nM7dqQS=>L8>JE-_?)u>i3W12@MWwtz~IK
zG_%_B&dd9tqhIIy2H$ogKm$6R=UE$i)FMmEL0|RyF=#48ro?pv`xH{NV<t@ELTRgV
z&$97V9Sy$f-bpfLehZb`{0sN^>V5a6auPz8s5+5nvR;r>(I(Ruu~xo2`K6g9HNM%G
zRF|I`HDXN*$<(3B9K!TQZ9Rnq+F?vuHL2atOLm=itzesK&g<p8!FiN8KvK-XT{g+a
zKY<Qb*#fGT%%|O79<!CT*20bGR_loG%29F*&S2L49PHZuzdJ*nL=9o8s5q^N>#kzI
zWRJgiNUDx-$B7gII~KVuBYr7VX5udFP+MRBXJs&_`fD)@$&p~D=V@3-z&uma4`WmM
zAvI@C74hR8vwM+Zho;L4)an*ni=Hb|Dkxe4;=#0%jdV1mGzayckrVDIbWNOwmIKFy
zdk-BtZ6)g1k**pi^8zkiYTD?3IKLeUX3-GUUP;hJPyMM%p5P)JFd`OeQrGYS-qU>(
zEZuT5VWu$7|3zCobstDAZ4E&s1vZv%?CO*kwMJmJ$hGCb<N>DKlHX@XcWl@za0~9K
z^e{NDTit?|QFs5v=fSR&McawSnB7mQk>r)o_S6u<*3LV{uCc^%@^0j_sLYE+Oxmgh
z@fmGpfB1a`6WCy5fzz1R+T!k>)p3*<eD?fG+Q6KkHsfrePd#RX$o`f3AVy5g?8*7%
zhT!i>jfgxn23$)OVH0&Ky#AtQG(w3aNXakH9WGpZUy>Ncw)M0td@%QGxsOuXnlGc$
ziL}?pK>Zb7FQH|Q(X9Ep(#08C#AbuBVoek{X#Q%<;lKL>P04DrK&fcvCFLJ;3{Qh|
z<Ngxo%baZU3LQ=O7XYvrnb_=fGp3q<y;CKxcn_@Pzgc9aU}Jq@WfYG(X;9+b3;$um
zqxU#J_u>R~Exj|x?+WG1$}i23QH-Gz2CY1Y%T&>qvznu?r>9rGpxRh_@+UnQ+_?)p
zkmT2r!GX{eO$(zZy%S>u#SORW6dR?_><-+U8-f@A*rq(x)nEQkY5#iLA@#Xe@2#(<
zaC5*cx1n_LiK1uBelmX)3gc5eQ&6G|2190kLRS!5Hf6yZ>Mija^yPe64VIKohC9ix
zCl*2`3YGi&Bh{PaMclCtR1uVG_k5@PWLEmCa(C>Uv4t}4w_<1!whdc8TCq9A#hdE|
z9WY=e3l3gCa%q6krg$PmUNplOA$n<MyU~J~5PbFJ1-a6QWi2ugZuzk}LxFp-d{jqX
z8fs{FB7xga-vaGB`&asW%#unAM(tuMCT$ADp%I&=5l_tY0jF;N$4Uq6mL$a8FjEgW
zTbVyS2oyF#Se7}k2MR*`4QNli%{TEkn<fE?C{H2(32mUKB*j3bxVm-Gi;{gXzBd3`
z8(56D1i(12G3M))DORAqm&O0VV{cs7W4!RWV<7EOBqKEsrcHlIR?R}ABp<aO4!Ac!
zklTg*i^B1`D)CbUH&^~c<QVmt_6PnBC}WU*x^~GC>g04v6(0B+{`9`xZ+1eQrg6o;
z4;mBFmCma<@E%ibznE>t)xT3V!BVY0kU*wLS=kH1{|)>Zu-s3jKsxo|&i6RAC-g$!
zm(U#DP9eaHx!ct@8PI~;Cy?i?`w2fAa<8I{X8^XF(A)~|Y+Sq9Xc{IZ(#Nk9z21C2
z2lJo(V!@b;qTe@;vQ(K^j*pM8aGf^V?$i71S7?w)A(*8q^P+1@JBCH=f%GWM55`o=
z@y@X2+>1W>{>(Ys;=rGf+t0FTL*%9m=d5UXzwL}LHhC*o3{Hj|eV!w`#qkssbSGwl
z)Kp2BG2DE}v`)Drz2^V)xP13HP3R6Nxx4af=}BE1Ua}a5_4O`AVW5Cm&OS3Jql%El
zS9%*xwP|8NSUj!Fz!pybCe2+d*VlR?IAN+Ept9_a@2Cwau+wA(v>e*KC0&#Vv#;Ol
zwn{z(RzkldF0kbU!)N__Vde~gj3i;^W!%0TC%1@+n&tGB3#m;}kxn(a=-3}GD+|Av
zk9esp{`901+(yU9P^O|ympX-CeNDNbMkx^TxjUS->BJ15AKNZf|6pl+le?H6<*wOt
z<JwTWS0dQZyc&$9R3Xyb?-!xho$$z-;d-6DtogiT2A7JNkMhI+u<c7Iv!C~?$DNvR
zBeI=90q2X=KwGZl?J5&2R1O6eze`-}EQkgxWd|Me2Vv^xSfCuG5?*kw`(CZpIb;M-
z-_R;NMYWt^qX!t=_!K~9qEXM<ICOyCt(Btj+ei%cWaX?Y&_8RM;GlmL<I!S+Q7kbM
zOZ+s#MCdrF<xRq!l74N#<2kEdRa{*B3a>f9&ElPYIxd_mY^nNEj2%uWr`gD7F(n9M
z7VU`PCY_>%ke!O>@p8k9JVChAEA%q=J(n7yhevR6eVyKFeXYsxgO*3hOgF&GvpoLC
zA(@UGVYM+a##mW<txc<}Fm^*`K+;;lP^BHwUyDSu3Mi>MO(xHQb$D5edc^Ft8}`Im
zcVX6&+CE~XupZCd-x0RI+BA4WK1qzqVq*Gawo-fWBA60kP1UywJXcpPO>w8AheoS@
zrJhr6tvug(`s033JGt<Y|F=kI4Kbn^$ShrPZZ^lP>f~&!aw*4@prFpjCFn}A8R;`X
zkiuD)!uQ|a#jg%iE#^k!`(eK<Hz|os)XZ_J@JtNZQZ@WTgHm4nfyZXkUj&AR1iUZS
zjxEB$b;!3<16{4yv)#l?$d@(eFjs5Uco{Sr>0*#e0|Gl?A>1G9N*CoVt~%yQAwB8i
zj((Gl8OqPXa}M94A&X?(kV=X_<NA2v=hNE$?nuJ}Ck%1i!!ML<c11?MZ7EKlQ|f;6
ztXpyYIa#7^Qso<AE}RgTN;e$X1Ny8%fya+RwD**Ik~aP%`5|ij$1SHj)rQ>BK|(JT
zM;mZhm*}~!GH(l`Ge5%YCwIRbx5l?|8J#-NU>-%#Hflvb%9L%t%EmyuoVi*sXUCuh
zU?KVBg!i*!JTnOCP5E~=hn0HI^jA4}M!nV(uKIK--ChWc)W>vL#;NS~Z-n6p?mj7R
z;trKjBgb<sca&KyOmlqdcn&k*v4|d4^mPp7s;19NgDZj=h$Kb|^8QzxsNaVK7EDI3
zxLcLr!c0c4HvGK4m~nUptaom6O<SE?K#<*kN~#v&6y#GK%8xW&CX;URvm)vrni`?b
zw}$b0<%=FRg}HoHT1=H`sbgaDdO8#(G5fzVMnw=Jm0GbYtHl^s|Av1r`POwElJ93+
zeEjc)#s?Q=hR0)hY_ALIYwzn)h_?z6VD4F&StH<fbx4n2!HGyD&|gn@@)VtyUy}Hn
z+z|ZXgfR_lqhN#upXO3Y*)C#KHWfvV+}9aRSLYh>w_;x#zm#`3!Igs;kig4kK3ixC
zY_vdcpOVJn+)+%*`}E0C$hImV-Gf~ZCN_f^eeMK;Wl*)`3+44o$I_oaeThBiXvtl0
zO_*Mb_o2;a8umd(kEyW8=7zIhw&e9&rx%zYe6i=^@)#1`i}G*|S*ukK3+7>B;j=_n
z;8mpROytD)jDQR|=K~dS3*E(1zfxM{?*_qRGyG9^`Q}6gl?PwUhs@Kg6f1m@iQbM&
ze2qI;Q^;QKoxpD;H7t51Jk)zpWSE2~MU7G^p1-*Cwki`&18<DD61XEv1Aas`y)_z9
z7gaSh?B`TFc2Wv=<k=`&ZkW}+36;Fz97-bV;_r}f-R?cje{%GqPaSW@p8w4_6M^{V
zhBzDoCozA83`MQp1T|Fsz)vUj^}PC<ah3w+rn)_W|0gCIT#5dM?jnA;ZXew0I_y~G
zDJKB4HVp2##wt{y#mGH1%1?kJrjOTZHzM$s`Bk$K{J%MN4haJEPrs{j5>l$W+Hj_F
zoq%tBFShPpFnK_r*%9I1h2><d!WdXc0E-hIuCA}*_gEgRX}D8kS*@<hLXT=H!38fr
zl&bw+LD@r^I<W*UXme3(p?iKVYqoDbT3uEFHBNIzsuNl5cJrOLVA{SvPbnKi$U$GB
z6bu~cw_CI)xvygPiRO&?-t1cw5z>as+G24*H{e9Z9g$u)F%Yv4XWfN$tmGnywg#I6
zy(-VM!68#;Oa2V=o)*+dxHnhQ@dIj?TTHBbcG$C&4vb#;tI^PzK<_3*imEL8tnA?6
zfdAw6(^GYP&BEmU_mj{cWCawQ%)vWZcg8WhG<;d?54OeL-dwAZ-cF{^wP>)THx$!H
zV(vaPpbO?+Bd;-ag02A3LAnrnQpufILB8uBS*c#z(4!+~;jpLb;-~qmc~{^(9r!Yr
zOL_iK+Xt~gwSnqeQeNY0qfVlu7*jsRcN6Yd!w1#;Nt+35+`yQ(1N-$L`a6o!I9kF$
zvjB=hk$1VN48oB1%C<}D7S@pPft&29jJ-`Vl(z@jC0{}*2f6~&d4l2X^_@iz>}!b9
ztKT2O&9Bj<2pMX|*+k*-3VUm`W1zuu&NQo4wY&zYT_OFs`v+bmQ)jEhM8N+qrFxZ2
zK<x20PVDh2e^LLtweZ)RPE2#|c#^TPNMUia1#S7klkN6osRb_Q6pW1U5fx+A%%Lco
zW`OgZmE$kRA?01h=VSwVxr?Y%QW88a+q7+~{@}?bG4@bSn_rt{;fZKnZlW4pH1})0
zA^ye4p(Ux}x+?+`I{4YZFxHiydGE)pKqzbb?U+EAD-zO2a|xIL&AQ|~ZnWy;_55a{
zlFVT^2ZXOg)iWp)z}EJ5;xCK29SyG?YH%P)J^Rt*clAw<AlaN$rNdzzVeY;}!AXU&
z>^)uXb01c4BAn}DRkWbE`$renWsPub4Au}qRGpvBd#-2NBt=)g$n!^1mlTAkBKIqe
zhBGTyzpjPPj|j;t9w|(#d>xc_UnsxOsM9?6RQSdIxF91nt|k&!vYUX_{YYn-Kc07g
z^tvKrl3ah6dsp&BJYE^YivDd*=F_myvaB`)b*LcUdomSg?#Vylttz|ssUl)Vv<?Ce
z;~DimY39HRM^0?smsvjbkEWp}3m-{TE&|Q7ohB$xzi?z9PHmg3k3^bc3bMXO^3~8|
z2V-oLh0u=s_YaH2^s(VfY@Dh~avqE(b(Y0?2Nv>MK)EiWTDEWFn&T^5Jk+R0a3|tl
zxtW%qism&uvOIM9;s(icS8J7Re*0hOCWm_5aO$sN<+3qRmu{5Or-h!!M0B-IZf*8Y
z#!qc;c=ghT;B4L+Y}aok(z+(qDLO(=`oRg;7n#pTgKyt&IhjTW-Qa~2It9vE8z+J<
zXz}^d=s<w=LT~`5&C&yTjclPn-6OK`_2F5OUZ|4PSaI}wM#RM76UyN0k$K7RHW}E{
z8y?=l*5UEB>nPK$E(*DEO|sq$rESs7*3@lKI$<S+%B8+5uI$~7mX4v+wvgF;u_XNy
zBbHpS->H|Trsi16duad_#*TE@!SOGBtQgA`^9J)_j7t7kfS}AF8Mac&aHw>Wd=f8b
z$UAyb%Zrh4pbt+|Akr58m~bfnfIHKeh30-55%97I5<+?S!lm7=^hXu#j?u9>I3lPt
zhflMpe$wj_1=DIk!RM?*F4t~+<0gU_qnI~a<&IrmUbkF}9+1EN@I{S#P|7*{%Ga6?
zw}@9+(2>DTJ|OMe0&4CwRy6SX$duMm?}GENS6b@NulsH;O$x%I^Ipc#&$bj7;yx*!
zMs4~;LP=A_hh~%P&I+l$d375OSrJOk>G?jC{KlXnwV6Jk6$hvIK{bMY$=^N6X407T
z>L9IBx3^Y0>bZuxzqY>_ybrW7CBCp{my5$imEw-gzRj-wjoimHCe5|{?}lH(E97vV
z+W!%0aNz9fRSYiiQYizZ#S&`&S$O`z0-)R}bMgJ$zZlK~10(+;EAdt1Q4)9+4OSv*
zFd8)R9lxU4fd2QPbp?#ugbAUbW4@q|>-6LD!rSt9bIt|Bq1Ny-^G*~+!lq4|<(DJ>
z^#1eC>lL(^t)HIVA_I?A&L0dKN=zB#XeQ=y0NzCZvaYUIIE81n>QS2s(-Y1Ip9fV=
zm$`3E#pcqG!c|J;*yzDg&OiAQtk#rulqR7ZIy7iQeD43&l?dj3y%UL{zGVEMJug?c
zIv;NLZ6&YZBilOL$POb;u<aU@_`7%bF8s3>r)^kFY~%+#=66v+roE16Bc-G>D+MV<
zywc_vfNAB`vil>4sZO$az{O2-=)Iksbo}>R-d;viLlSg_awTV{<O_z=#XlUu5FYb2
zmD&@7A0fPK-TdtRx(m^pUI#~<Rj)sV2mG}C`>P$v=9hYbQG&E%=pynR9CTAV|Kh#x
z;Roe<7OSQSys~n*=m}<7wBb~J(n;6K)j9PI^pe7^vEA^6u+prLzGAL^5D>h(prLy5
znnbH$hVQ89tf6xw_If@#(dUM!@1TV^Th3vJ&-3uVHau?>msG`{$q#O)v;I6N>b^YP
z>YkH7u-_ZoULR*zc=}@NfIB_-f{^GQ50R;7lMs4#UHXTg;QtN*bt6Uy$j60HZg_Au
zHJ(vq?2c^0H%A;_;4m&VZoMoyZ|TNq?F+b6i1!!I%=QfjsHw7rEKh{?>S0asL*=jX
zecKbQ;mQlvUfi+^C*Xc(t-hhH!lS_Sv)bYYN3O03e~p*61VMjPQ*xBSV@%!xS?Sn5
zxwg{q0nl~H&S957anwW!>uLXd`)ccaco)ba;8{0MpZ=PJ>-V{CSx=8TWM6x7#S!D3
z5uxJ)VSCnq64W>W4@XYaRc^oZ%8Bx%5MR;Ck4b&N34?X#RBq`T8af?~()ySya=vU(
z;(GLK#Ii<S9*OxHD9X*nLj8O)l47yoO!snG50Y9Lt!N-iyKg&#i2WG_K<>b=(@@Q~
zLSDzgc2!YdD;RHCf@KaaDR1sdjaR*k8ZJO9G6Fyd&V~abekJ*br!}r<tW731JTg9}
z2+3V9woKSni&P31#QiOSnxMOES>j7L>YWUPBv&IV_9UOoXOO8EF6zX0h)%6%H%!o>
zpVIU^4Jb#SEHAAMq<NC_gI$ZLEu8*vVeg^eBg&ZC;;M4&-?#xeyn0_&DES4~x=OuW
zrneGC&UZMQSB2-6r7+uVFYv$596Rc}ZL0z$!Y@gNd}4!)$zz5xI>KZu0vlx0kTxFf
zOxQw@;U65%fF+~C`YgwwDXXhS76Oj7S61u#jZ9kKqgB?QKXzU=jIPq%UMP>0ii9KJ
zHH6fVq_8o@1;FtGjCP?NEw(APqu`ogTiTV08d2uc7JgLg{Hgf8o4;-R#Y8`HUN%s+
zM;wc1<qfI6yU(*`#W#m)rf_1#N^vz14T)41QZ+|WIl_Sxb8G5E)iv^&mp)2>tJpzQ
zoyFiy*x))ZpAiUo`b-gj79`WV=f{@B$!L)&KbA-WTm~VgQ#zlB)o73!-1#zk@U9dH
zzb!hEU=DP(kNrAh`o^`6MKMl=avlnP%8T(d2XbsQ9BpfVwv}dqQGDzD?t{azzu(Vx
zw$ag9e%xG7-~*P!`>l7p`N3Ml*dQ~uM@Fnvx_Iozr3D&epIo-T@msu*j_`N0Ge2ho
zBb2>BMA_Mg9;vSD<RfC-tPBLH8?<s_wr`w?QqJ!tASa3yP}BVs%Q%H+2h=%-0~G0m
zJBl3@88OeoEngRe8p3#{@{gM_ta(yB%}-xc%$Up5K$)~5B>?g-7B+$awg-VuH1+7S
zW7hPp^{=C{H2Dt=)6>&9E&hD`FzE`UDD+Fc^j5V<wKe&Q7Ve!fd9wsE#~b1J7BU{s
zj|)RVPmO#0lWe_<hhS|xgHvxx!A}wT-cIjGYytQm=+#}8Oc%-&m~OqwCm&|$>nKb-
zKJl&fq(CC6p#;r+9i@fJ2a(@E9#9vwtvg`jcO?4h>$>nQqErAF;58zr06rt%9bP|g
zRiTi57a)bH8T1TojaB_7Hq#37=QujCeju{$9Iv4?hvrAvYO6JxY;PDo_8bbJt3Ww?
z_T@xU%T%WDv>TX-92BuA-u=j<Or6C@mob<{<}?5E8@*}S5<3Jeao1WAH!wb`fj;Ck
z#eU<Ew0136^$n0<I-d_aYtUTkHFMOGMjmog%+tZS&Ht>gfbv`-xIQbdcW!1nr<Xa#
z`vpr=y}j+VJIFJEWyyxnDFU^!Cdr*Yn>vX?)KETNxGKdno~^36CV2G3u&F-ywu0Ff
zpR|8Lc=}MBOH59k<?^0p%DCdIW&(c371}dIcR!A1*qY0hK&!W3NB^VF53PgTa;1F!
zuN0S<v(@kUd{?cn{P|iybT7Stj8(Vz!|;IlDz-4{46gN5{&GONt?Xv^>RNV>9N&YW
zzynRZoZHNVYo^^N)cH!&P90}pb?PrQF{l$Ae{44q_f40vD#*KUbmtVR6XQC|s(giZ
zJ=wrzGCxW^TZi^D^JO>L8MYj1H_COR0<wUe@Thh3r!sJn=zC}EQc0`@jivkRy~AkL
zGv_pH&GE%I|If(C$Z3hYOORaUNBHTtrD%V=>f{Bi7buP5I~U~Y8TO+={MIycq!(GD
zkviNDM^2wdDjVxVV7JRjPYC~RnEh|+Al{2SFa_l+e-xk{#0wd!fQx~my~t7TJ)u~y
zU*(qgth>#AT$!sc+tY*ZUB%>S@<|^~hJ>1gPpq3ACqoKmH6nrX&l<k7Lm(cB#Fh^8
zHCs0|tHb$i_b`h0Y*)@xt$fq<)Crsh4vh~<2FAU;J1IZ*WmraWkL;(uY@glmbMlEP
zBT1LMn33g&RJl!$Dn-QQoiUk~GMz+eLJKPtZl(V95y?UC<<wG#@XXyv<@OZiq`X@C
z4nCSK<QpQzQWU4AHQj_%oMGHqfwqv|rP-wg6~hibm+cY@NU5mm&Bj>LBBAxdoCmfR
zvLU<8<g%>d1BY&TN=_193Bnkm;o*y%q~QZGrROt|{B^+nCWX@;_(w7^q0xYNap2cg
zZkF%W<v2H#vP&Dqq%{cHBT)QQD&6h}%J_-+oME<p!t`m1s&mb(sWt*_i>c*@%G-4P
zD?o}3BiU#^?&1GpqY6bB5S*bk^>^_~o^;Xu*1JA&kxSQ^;Z1ZL=#B)q@qf{`?*j|j
z7rX3czSh;%@wm)oSK@<BiLHC+8BTvxNB+Jy)0FNi#TFv#X#KFlTDha@Q*)pu94%+F
z&JjS=Yf=UcCMm_}^QloBT4#7U+?cyL(#LJUHBjulme8WnxtRJkX37ll%m<Yd^5>s_
z%4W=pA2Ak7y-nCtj8FImW+qGHuVqZp)LuWky)jDyHKos8*JO@|$-=vUo&ky42cHh5
zds2i_e=)bd5vbfjWiaZJOZNk3IrcXWYfzG}nJquuHJ+q{-0|txLaQ${hzVoPvem-M
zWz#Z_Wx}LDUhc3&@hWVEAC5%_?-{%wHgaJOAY9@kOElwkW9!|6=P11(*>}5ewzvMc
zl(0#|^3qvlKPVm6bwFdSjilPDntybW-WTW%@29Pb;zUL`)Rk9X-qcewoh!Q&V83Lo
zje?Oe-91$guXykKUH};G>gp<V-28b95~*AkUlt}$7I#yN+L#3B{d(1U^r6|?v(}UJ
zm_Kf8!OH~*bHyAbNM$U#X+0ePJWZ6&%M>yVdphcdii47?iIU#|@ON{Fe>9nEl5)p=
z2RyOE6vO4=J!VS*;`m5s#a{C9aK5BduJrO(LD+7^$FoEy@q+zl@DySY@>Woue)~4Z
z%Y}6WSvC#_Q>8bC`9uavKuJ^Qjhn)mS7*Mm2^%`X5DyX}p`Q-=%VIN0)pAmp8bTzI
z0a94ZM=rljZ_**1jGwnc3&&_^zd*!yAQ2N;v}nwTb9n<axK)jf@WR)kh2o2N5=(pm
z<~w^#G*K0>b-g}Dg*82JtgV-(?nP&k9hl}wETndFgDM+YMi7TDVH5YZdGft;Nk2_N
zT~tHa_8c*8r7AD4y(kLu<)!$Y-jzZY>z)dK^m38mf>_A}eR;c{@j^6h^DPFA5~298
z`a1u&ZpDb1RjfFKh0RMy`HNr-5Y8ntd3ZnZ78HM)-^;*x*H0+6z5hD-LlUXf_4k5|
zZg)W5c)M5(Ka^I9T@zKTqAM6L#AU1}rEPBj!UxmQQh<NJ8%;*rwAv_?_1sS4y9@Zs
z0S2Y|OaxK;UbE23phTr$eY5;5JD>f7R=(Wv$`avuljdTGE&j^9we+Lq+9#<m1qIet
z`Zb+l<#1&fPhGR{RU=(>fQ_X@!RmM=O~YS|q_eiPTpY9tjhAMMwaHP7O8%lH$1--^
z65NX_g})6wvR~M2T8G<K=>!f#a@_+b&`K!g;W9g#caV4)8LzBiozC;XN_r7(?CqEC
zMV7@1bQEEa*EI-i4(_Ca(d+?eBR&$9{Fm+;AjTl&REsAxV25Xd0uwrUdkgS;<-UVw
zvhI=F!SV>{Le;N`&=3ZrTB*iaxj987aEk58VWa&rS<c&d4dXkf({UJ>5C<23O+jIl
zhdBeVcknT@^U61n?wSsz596{DdHipWd<O0^a|b>_`$^z#-`x11tBzLppMr}ySQfy=
zn~A*q)MfAKfI_cCyR150`a)!q(=*73gk<0(fMYxiCl%+Yr?rAVZ~d%5++$Xw#MenZ
zptR#Ie5N0P0BiDVJ4Xd8k}hxV+u>+=d3pBj_ZGqol0CyG>-}wviub(Is3vD7p83Cb
zawsS}E~^CyMJ8JnXESld7aszTjMotz8_ir}gbyvSQ{TRS-t}pm%5D*m;B#<cSP+kp
zW2K!YH(oC;F(ojieZ|B_8jjQ=!Z8vsj)%g=I6)5?)rScZMkpciv^n1L(jzd>`LoXx
z_;3|e<~o%@n#xyyZOp@@T;{<&`F~;c-`tbQq+osoYf8b`(tOPGac0VvbhGUuG&Q2U
zGl>OQOy98vOFyoSqq*YbmeMmF84U`Ii9tmrmW6U}!;Ph^wV|f<^5><H@|h+`6)soL
zqv5ppizm|C&8W`=m;oLgP`?RIh!nLzCy+1o@s!k{1S_sh7=Nfx4D;AAAwHR@$*A2_
z`)(+{8s!mBSkc3%*{2b>6IrcZ6tJ!Jrd-@*Wy|Lo)J_!R`DbEsr&Bf35<^(R?9t)H
zvvD4tk$Mgb0>3QyLs;gsjsi$L);#%FC_2voV6QX%NuRi(^T$o}Nl8g}<)g$Z^onRp
z{MGnr8e8Kjxd42k<?B=8{v1aLt(XwTp*0u8<!|Iq<R#fhxL4FyR%E``w{>_N7;Fq$
zeF)cuqP%DM2Xk~E^y=f!ZT@Oa$2uu4i&LqDX`we=?d&ps(YIy>y8aY?QG&OsceS;$
z1J5)=!rUbr&Hm{*)9u99@xWgd{0`c}CTjj}D$6!OQ?%k!I_+u5A;146u#Ug-rl24U
znIC*L-w^!4Y4KQ7S&*MYtgWP0?ML5my4G>toqkuWF)MwPUfNfs@}fGcaikb;yD9Zq
zvO%?&=pEH`NmN<T>xU5#tUjxRf+76dPjh%1`36Wmrw(x|&FNY!l<)saJmhjhe2tf1
zu@!L7Jw+en<?Cc-_+s!hp#)I6Yq@GVB%au!xo9~M-WN%~>^gqf_Jn(@5h~_T^&2-k
z5#O}p=+rK_wzmSV@P#O`@T;R4&F&Rm2*$8K^B}O`du~zm(}tz-!uKke06L`&zs|oj
z^!l&QDC1}?Hfp-r9t!}Tcj`1;%ZdG0zV%+{D-=yZ;kt0xzXJx|s_gRSkHT8baC`A#
zrt88fx;+fgy#6{or37;5J-oO62ipAK6EiwkO3U8PP9S=8IMv!;<9PYD5}#?2+V;di
zH0;EGy8~E?>XYDb*4UUe;0VPj2jlbR{wah1D1F_oe#<?68|zuMi+1PsaWY?m^E;1E
zJsyaSA%g^d^<$8it3b|2ARk1(<ApzX{0q0N|L8PG9M5}$I$Ixw$9GAmbaDLEY5vJa
zfiwb2!}4~su=<tpDD13_=36=c(jv8M&OBzreUmCh`;66L8Wr&oy+dZz@na?OWTomg
z4ancxh*A*Za%r~c*y{W?xxqBIAU}F|nLOjD$eJz0h$dM459{55`=kV)+3Ck+mJ!dl
zSsD$2a6<F*riQ?q-NB&hhK5XzQYsWEekNSw&iB&5kM<?FzCHvx?Uz~8l^Kv<5B1C`
zJtKI*`d-d`LSN?blQH(A!dA4<CTe(__i&_-laA%@!9)`Gx6s@!v-l^jkZ{q5DY7tv
z8hv65L!%u(tSgqk9ZrP5(k5Pu%PFC9gW~U(p{Y!3<2nD2D)=wLU<d-a=s`u_=Ti4L
z8YgQ)DcQpfrTIp;N<g+D?R;isj^!EFP^Jrl`cDmHZ3VKjWU$RnWGMMrxD$E1yM-*!
z+#B0r6WigXfsDP`9GOt8M}u^oX<~Yw{32<ci4XM4E1p!l3Qgwyz0s{t6en7e8|31Z
z0vXDAL9%#>M*if2m>|JYCxD#ZN?$RtJI~LC@9E|abe+}vi+e|0v=MYnE|NvZTY$8a
z@f4|DwPKILgqo}A$yKX$(#3bD9y4`S$F2aqWoT#ATxGtl)fp|JZ+gwJ@gN;P6X5@A
z?_9&#%(ggQn!1##5|e7kn6`>^TsnfHl*TCI(rHPFju4ktOQK3cqG()Gt?1~u-#VC7
zDKS)>YD?USpe|8F8;rXWZ4gD`<eW3@bNbBbIrHg!n)lOs*RwzTpS9n$*M9c;J^K|3
z+MGZ+7Umf?dHc_J{j}GB;q6L~y&Te>?j1o4_IECQ5LAv8op=8%tje%}O50}}V&usW
z3s;vY1Pwlu%r}ySpwtRE2VakFW}QkMo)ujLDDibg6M{%mTtvyxYficwr34A4s{(&U
z4aua}K5w?4xU@(4M|@z|FJ>ES(Tu3jQ)8(-PA`>)8RG^unv~spMR-pgX^YW2w0C98
z7P$3>ZVD2*3lE0ar`mWfXba?I^jX30r^oJg_k9`Po1zwAerFS!<iUH>v6A2~Ny|}+
zR5P;Y<*mKNedr?pYa)q21ycR2c~1Nb__4qpNk3<u|IW!*(;kL&NqCl10yBuG__JGu
zPkhN|Xl34;oiJ=2b6F5r$>bjp6&ZK&-&&JQ@6%{F*`%#`DX@i5#RiosrxqwTU|81H
zdF|XNqn47~8}fxQi}37`Ruakw+Mr~wUi{Ekgl;YY8bxPjcV_z^^!P2s>?_U_dlIgJ
z#MYG0p=tB4Z*eobc2|?J(~wc7pypQhIEc@Q+6Ik|BbJbDbNa-IeHq7tCWI$SrvN-h
zMxJG?Kr1T6B%s_8oQ}2?oOdc(DsDT{sglmNR;J})sPZq5ZZ6hv{B`ORLos$!%(n@;
z33o5awLEd{>4Ep|Y_~fAJLM#`=8NPyxxE}1%$jiHo<EmVM7gTSetnQ!(@}5xx2$y(
z&J1XhR(a^|4QF`dNm97ieFi4(FzjJC&Exf1?j+=_yAv1lgz#X{_&mSIP6U-4747Zc
z;f5QaM8G0T%<i%)KlDLfL%UJE_c15&%ku@ddKrWE><&o}-q5@bWbBo_XEI<gB4lNZ
zhC%GDjG@3Z>tb25Z{6;lxge}SFH8rwj@B1ET)$Pd>la7Q&azDh24EOj1lNsxY76w)
zc1Ez?VWA*`^w&z69v9fx2<r`=P7n77m^yr*N<ZSIpEKUU@7u{C;M8ov4sYLxXs5K!
z<(keI`>qgEk0}jB<21GX8>LF%?80TH9_hfjhC^3O$>BImUO*;m<cI6!xzu-7{k?#l
zIz!7qPGo+Z<TAFY!5Y0{b)^49d2<f+i)z4~8@@9~XRx1sga5vyd}<flh`V2-D`cWi
z=3ZCqR?qkPbs)h$O6Llu6pZVFQxTI&p#?J5W2&Lrz64?kKR9}3H!elzv<hO^upcr#
zl+#2$3uMY%pb1!{@EQN_cHNeqOtPp?4#-lySeZo&W|3_QXM633U_Y*osIvR&m)l0B
z)A}X4qy~#P&KLYX^mP77eg59O{g3wdQsVA7dS6EZ%5?LRB@lI93Qo}KesWfxwWl84
zoo~7oKG$sS;48mBa;A#<(kGje;^r`n^tt@%^renjvY*A8SrZpIt;I^bq29T=Qms$>
zT6(0!@kqpRZR=Gu;vS`I;2B5#D>QJ_c@x(u*_gZwY}e=szWnF)$%`|0up*G5b1S?c
zpAF;Fa<jtT7USqxl<DN+^?aF2hA>c@cz$(?l$DD_ChK-`ZORQkqVHAVW&j5tR~lWx
zB|(}M<Iw9#=o;O1!+36?qRZayB{-B!k7rBleYgi4rEJxUR8LvQjIy?0@VC;oq5Hxy
zl;2sTj|Ka)hJ~2)%FwaNGCN0APMS3LNOY5opNDzDYnAv0P=YXIaP>Wj8I)lWD)q>i
zTO6Fx$^{+wmtE=(!n2Luv^^#@yudc|p@kba2<y2rQS2n|XQk3_^u?C$Rb0NfI<<Z@
zGQGX;#IErIJt^HsxxHIVbB&RGpVjh#p!zuAr(Hy>gXNHU3p~;=JP6~^1+;G5h}c}5
zb8EGHWD-`BOYOUkfOce!hw41lX!jwvi7g1dicDn-(rd$Bf1#p8Wvul;Ff*k(3Op=c
z-9mVKquWnlnX$)U&v?Nz>%9DSn%9KU6-k<XH4hPa(?rxu{WJ+a_o@jJH-m_kn%y)=
z&%=GSPhiTH!`43!>m0AtOsqSxRg2^UdnEFQ0k|V{Ki3EP`5Z(*YK$${hSGD$(%Ino
zy?fQ;Dma%TB^x6ZrqjBMJ}0+*#h+WN)oRc7t2l3;lxW%bM#Jp>OUKc-x59D&MyFY<
z?V)Wr8?_h)3N-k05=GLBeZ=uWbOm~iP6Q`%xZ?d&OW+TTE7y<RPsxbOs=LxoI=v)C
zm91>4XqbT!yQRi%DwUh@+xMPEudA+sS;KYs_qxohX9gTcwaYbK!9LLoC2nq1$VT=9
zkKx6rKs_Q*p0u=kEihn__lsU)-A&`2gyF+^U-iCH&eMvx_?XF^3f>I)Z@K$vLaSSe
zikGWx)=T$ue6l7J)mm(Ft$9A_M?1rQ+enxF7b8Xg;7i^*<COqcw9zNd_H`Ke4`R$g
zMh*2-DEybSwJ?(x&x`cIaRSt&b$XQxSm+VeoH<2VWgBviO$(JA!i@)`vMr!e0d>a5
ziz&4m3f9dVU|ExryVgO}NBtxh%Z&H79A$^j-FTgk)a?IO|2+ioZs}&KC;qoUp_c_>
zo05F7vRATDuh=_$d$RsSZ0Xg5_`}&9qih9?u#4nB6W9gMAvXdNPkFNNZO=-fiQPT1
z4a#Tba})Pf<6%YWyZy_1tPGAWzpcfu=kh#yyhq6l03p|chcVF5gQcqc+I=dyJy~pg
z0li-!CfO3cVb#Tt(~q9sYF=>tTxt2EA4;@nP^lPc3Ll@St&8y=q39Dj=I-!<5BCGf
z(^1!RxLsOyxnsKVXDh@t%9Ubok5pzNB~c#9`nOsDjAVLAl?I<WUp40w-*liy^Tcmg
zrHs>@A&h}I_!H#j8?BZt4s&~<NzJV(Vhz@&eerND9@7;A0S8f4f&At!wp})zVj8n^
z^YN;&7DGZqKV{U_8m?lQSR`5S)SN^6RdlsWX;|K)39fIZrk^LUwxt>bJK2d7flGf_
zAU@|=A<@fK2h0+SG$hCM6tK0Av3~w9M(WF;^PuX%6`8`p5<REq>P(4jX(Ls=iRA3Y
zqjQjdAdY|qa}&bVfgI%l$hQ}q@(0V{hJnYoT0vlTAiq-D`5WTw_<aH|YwRXb+tUAg
z#mDr5;X*gWHS*PwKVRwpI30g^{))g~5%@nufDuRE7LTHn&;d^+(|3p-+ixAM%1?Pl
F{~fjj{-yu`

literal 0
HcmV?d00001

diff --git a/api/core/model_runtime/model_providers/mixedbread/_assets/icon_s_en.png b/api/core/model_runtime/model_providers/mixedbread/_assets/icon_s_en.png
new file mode 100644
index 0000000000000000000000000000000000000000..5c357bddbddb15ac2076742ffd3eae597e47316e
GIT binary patch
literal 37303
zcmX_IV{m3ounykXwr$(o*tYGBHnz=;ZN0H=Y?6)bY;61HtGf68m^!SfIX&Ie&(qy=
zB9#<=Ai(0nf`EV^NK1*UfPjGRd^c=p$nQHL7!<qT2hg7?KSV)lr}0m|FT~BYq%Gv-
zL1@3vp+P{ytw8|)z4F~~zZ(b$cs?iy`1cX?zkB&$|MwJhCm;O(&$0h|@e@tw90Wud
zL|R-#%@g#z8zx&%qFKnm?dsw(W1Ll*k}Vk^DheG4L~}<?0!5ebsRd{U<F%pqye{M+
z-8(Ln&4kZDcp_n7V7Y?{LZCpT(w`g464Of)2PT-e&?iqV`8wu&wR)9$d$rqMZK|fu
z6kpZw9xYV-?7XV@dAdL|t)`3`064~?N4GT@g*BmKoBqda0#%^%DNizT%46!*7PrF`
z)w}gfRPV572VQHq%<Y{6)57Tj>J9N!Ne#153|M1w$R3W&k&>G_2%6c6Q6yueWZpx&
z?zA3qlo)lCFuD%emK4IWqjUDcd%|N^upO)RuQmU5^qGkshpT1;v(s5PVU^m~*L|+k
zB_|k^KJrEqjCMnnYS&}VW+48QZ1l9u(Ww3|XYo%~X_pQrMHY;Y$(Eof1N=xN#}B|y
zKeWBT|EnKAU0^bDn+g8zlcx2lCxuEN)L<A@3AQbY$bhl>X=}U`2eK24DKH-F>sZeU
z(6<`MwDMqVB6NfW>;}C?FHaKIrU2fV?kgVRcEX}9AO$Ce?XifRMcf%;XFBW7Uvzp!
zEyow?MgG?%YMu;=XVliDLQ}*C9pN(sF%hjOmIJxB5ylrO<qFcMejo29h<a9%Fb~D-
zuifW6W*&V=!r~w0DG(0U9V{j{Uk-X(ja4ZGRXr0_pd|s5RBuW5_}$?IRNQYcytZ`G
z;fy&u!!2;OlZI8gQx3cMn%`npivV;m68bkFzw=V}OPE_!K-rE89a<cO+tjp8DHZ}Y
z=^*sLNYaa<bA^DVz~N7DU&BbD8hrk?`}R!TbGAgd)hb5jN@?l*W7Oj=AgGAih$0Jt
zp5Fa5m}|i5$+}aZf9+vhZZAMsHD8_bdr(#9fiK1$*9Pt=qynE%NFGT<&#e9MMx|e&
zuXn2q=BI{kd=nJ2pY<?AyLub&EbQu))cFxt<&7&UhiSD7R4GJ;GN4CPy-@>cx{;7s
zEAcE@y_JR7KP~^IC$iYl_!Hr|LaElsfQ2Rs^EdEtItgt2$)klIE>vBW)B937%jsOk
z=%!i}k*3H;cak2~@0xzSG=)0F)J3eiUD$h2!Jmj=_oba$`4y3((N8suQeVhRA4q_u
z@VkQT2H63rj<w*Pby{fE<!H;hGRASZHX4uu=8mG!%s&3n1@!uY@k6S{%(Z{kvm9CB
zF=ev2;IyxlktTM8={xkH*PB*(`PZ4w<!InS5wJ7N#Ie14H?MbtVPGyTOtUlLMHtxL
z?{&4yUhGr#SdxU#-e5yw@^&5;-{6ttb}A~J5ArQwCsDix%@3I35v%~cc4?pb!`fWD
zi9aEyLjEw+5ZS%(laj{Qgzf;n>A72b20XOTtLGA_-+zvX%-&!Xb(HaAQLTFOZ}jzw
zgiOy_Q*G?gO=NX1FJJQ3u&0g*LRI?IU(dZSjgkkATOezXGSqniTrFR?B#(a8FuD%H
zmGaVkvS5P_N9qAp5R+VPZD8Z2GxpuySA!L)ryU*SO4e^;RkB{@dufE`lH5aS%jLJJ
zD!x}cO1BBD(pql1aHuQIo_iu%cd2hyw(>6%qeadN9(KK+12H<D8bxmKGN78f$wcQS
zE28~tl?e~T^h^Y8;W=0Uk$zeja&nNvb*s}QLuVldWzFC``7q=s`%h01MCA&R8@11b
z6tLQN%aoB_pvtR{j(N?N)1z!hSz&VILghh~pPo{WOA>GEDb$`o0NMzA+*cUa@1Sz=
zZ;9zeDh>xyQ5V%q(Y~;S|I?|?#a&#9w{bSIksD4HJ__9z1UXH|stboLk~^m(z={`Y
z8fa5Fe@+Nc=@Sf5qaoJ9`XF5LEz^V*k~k6r9*Gn2l{|TzZxlY2@Rq~C*!5a%n3?F6
zrsK;~3;<NulPa$xzpaS^3E|B*9<*vP#e|h;98o1jfE;r5?^q}w);N@dXyzZ%-0@7L
zD%jQ}awDY;F*Vus+xMuaZeY;%5tdfJdf2zd)y5p{jE5SX<AN-H-jP1iA^;K0hHDJW
z%45<qJ|50>2bWL9r290$TzZ*I-+k6c{whhof5nCwFU?^u3hvN)W`cXv^T@>$ZE8-!
zYDG|cqx1PIj*?Zh{-RfI+qUG;n{HWcs-UF+U2-<wkAmf%X{}ZiC3C+$==5GnNrhQ#
zcZ6BEh1+aJrX@%?J3rg0chAG9Z><4sL7aH9i<wt=nfO6ovs*U-r=9{}rI_{WZ+j`y
zf04-`PCe6PH7~lP(*&CUtKG0fJCbu46uDkR+#uhm$C!3zeheAcXr#^><;L3wZ=7LW
zopR1I%Mmof7^6-S9u3_m8gSa?iwZkuS7XC1rGfhNGi*f*sI0?>S*TT4=V6_WP8NyP
z2La<K^KlLsQef%JHDz%RQ#_<QTrmn+!T1HEMzWj@;1ycpMhaG$@HigF-KBn0`Xg|F
zOOT+_hw*9TtZ|tHYstdxP3fJDFt8|h9?~~9#0Vc#glwY&7BpYNA<&l}OxD9q9TmhB
z4@iOQ;CFdiJp1%sK{p-yIEU+*3p9Fg|4Zx*9`%&s<GF~#N7_J6n39Hs6RJ!0pY6V5
z+G%DflMT`gF;7Jd>)@q7ozntXB#Jyyl`x%d%(A%5a~gl%QmT*ufHhVKEm$``sde~q
z#GGEOnQCKq)##kXZ`w55)n&%;*iFn;8tQ7<|LH>8kE8))%PZ;jswLS%Y~8>iTq+X~
zc?cYUWrU}?gfxir46d*EIEgM;=B&bGtvW|4C1}wcw#s{q6v!F4i=y2k=Jht#wt;wc
zO+s+MFj;Cltqd4E3o6mTlSnE&3^G+o0GpXgWb5~g4V8iA0zb-8F4r<JCUU8Cm3q3B
zdhnRTR0RTq!Py_aatZM9<&77T0KAx7@u9Z@g=r&def48wPU^IJVBgk#430ayT8}Z2
z$)$^~*^`y;6MqLejR01`bM!B)0%#)vR*W?Ee*F8a!h|f4<<<x`%S&Bd<UhX{L@0?%
zH)K)vt*;36BKU>GsgBB2$WragdJLq>KA6&A*;w{Ln&fe^GMwa1!7^(Fr}V&Y(IC<{
z%n=Ls%hAXyLR^F*+dcTyr`8?(W-keVdp)R3#7<P{W1Nw<5~VT1dQlTiR_RN<Mz{0q
zThmyiF{0j9C)cV764J37Dy@`E1nhWggwJ;JJC2q8Ez#ys5BN{=0pUiYV1E3j7pHX1
zh`O;FCUsYZDmUT9gcA_@w8olNLdDDSMqbCp*{ZknT5A2XI5UySnB(Ottr(P`%?K3(
zgi=8SUGmjKZYBOv*obP(+R|K~rvJ*(ol5*!qNa!_E1*PQ2uEEJ*3!4;LYgz+&aktg
z*(=8MH6ebD71z>Y6Aa7I!nL^&^=P+O$2xIV`(WimUJcG+?Ahs^c&Q4gmjjZ1Xu6D8
zCRjAVkj?eoU)%T)bke3ACDjlALLf5MVWq)rB`XPtrrY)@*X)9TZ@r_{>Y>o=aV|hr
zZG7|T(Qgj4eIgJh5CWIcvdp~EvWI6IRYt!0gQ8lHD)l{~=~A;iu51b53gb$}w-jhU
zVA57@<#F<Lwq;O8tU8P7w4qH$+VSJeN%vpDN<;G~N_)?t)h7|MAD3^hZH(<~5ZXlZ
zNZ;EDQgPK;&)pfzZc`QiOutslmQ}3KHUT=O#y@j~j;LHVjl@UrvbEQLGasV-HiiA8
zXg;S3+s2wV39UH+Mwloep-+-f++B!`Ya#jjCZOuJEoUC0kbcMYS4<^eat!bEixBqF
z`a3-SSwVOoslnKN#14NG7z&*C`;IyrytA(@$DI(arzWOTsGvDxet<0Q(g6eXuVUip
zRO>%n!2)(-r!(JERwM-$Rx&s|H8Y|Z;zT-_h5f40Fr=n8bAvp{J=erJLMp%<t?3-a
zfu<~9%7Ni!RuoRJ_A}!@l#xLcp!(z_e~zvYdst4@5-PE3eB9wP4`#ByO6;w#bz@IC
zKS=_)Xc;OcJiD^%NZJ$cgV~lj>_!iR=1uT*fleMbusiQCY8V#|W+=rEueJu+#Sjx>
zWVY)q!A576@(bACD&O9$4`F;r(yx+F3!w)~b2JM@DrZ16w0bAZ@Wl#>sF^QfVIuG4
zgD5tR2eIUU^y&}FSELA9&xQtg+8Cl#c;;WvY17Or`T<rvujLup0GFOD?)Q5n7TZry
zQ7psdcrh8|nAQ0JDj&n2wN&=A9`W7Ygmeu+r>($S1%x*L(CAmC`<m>A4N*hHe1UMr
zvXyv`5Krlbrr_>ea1qYy26}2L5vg4%rPVa`;pz9VtK;OLCSU?>Nv4c>DDBq9GcD{6
z^nzK#a?a|;*vrbug1iuyVu(iF*B(A+X`x`jH<frcs1>rw^XZMK#?>T0c!A1GINU9`
z0*HJ*el)`*5f}4?=RDOsMBo$~{}#C9a4Q0S9ATnS6~rWY)h?d_?T8kr-Pu_M2@W{z
zsEc@FY~}XXY=^{=t`*#a6|Ivs?tJDB<EYkPAG$NfV4WK*@=bJL^&kvxM`;sihE@La
zxYNi|p__u>ewgX8R3?t5y^hN97;*%7Tkdm<-H=O{&PoIJ7547S5q1KzwFZleJ%F~^
z_<+u5{=EX~l>L3^(o-^vJy&uYcU%GVn_U-)aND&it*cE%bPsvHK+A=@6DL*0BfTEh
z{klIFPKu5eQ}91Y;=t`IO6W?Uqvyp^2V(XHwZAQnZ?g2AVdyi0=mTkTNGb;M_r}<j
zjXrt|S3hTjpT&;E5@tZDiwG3521B69%TwR3gub})&5KyS+OCM!prQ+AGz)8dIIFvi
zCa<Q>5cGf-_}SRH5r%1!w1^^NCZvy6KZ#_=#+*143mqp~UMe#F^L-QFQ|zl+_Tshg
zXsc4HOseoE1T?@{uA@NK0C@QZNFk|?*~7YThMn+$e9gir=F%EZy>5UTXb58X4M65j
z)h8W;iQQ)&gu`)Uj#w5#lz7wOTv0+WLJBnWB3ZtNUWy<Jm)zO|oGA-|wpk_+S`T;J
z(9Uiif*NjE>>**cf=hHO{pSIclIT$hrf$EK$mgcb7#{gI`rvXvVV6krJvEIT>(A*>
zQ)tchn~&J2r6CL{P<Ia6-<9HbXoeUUh^C)^W9@RLP4fKSUrpnsK7COd0@}o;28_{!
zlC=ng?(1gV1N+$tqOiBmS<%mb`sKrKRx<wl58?4Z*Q8LVjXQ|=TU5Lb;n`rR$jh98
zUBwrsQ5R<ANd{;NMiJV2RmO0_?C`+`a{!j>9itz(N64st3O+LU)CP5B-&TjCXaXZ_
zC5c+d1t7VpLN6#>&RIAy2S~SI6$;*Ebu?PH+B@kD8k6E0BzeCVwE-ftK?yrL_FerF
z6ST7)yqh&^1{eR}!uT4o5VnobU#XbwSlmpou<fVhvk&9@WqSdkNX`b!6WW{GJCc&O
z2v<Q5yBWZcT^EPrkf0RIdR1$Y%t;8Pp*j$wBD93~S8XTh{jR-hn<*S%O)o&gRES!H
zsWKYuKmGv*)9~AD$Kg$x<%G(5l79wW(LJ%6c0eNKwZ2yVd`NVbvuR+K2EY_0X-!UC
zecV4Nu!wea)!%+Mc8C899kafU!CE}R$^yksqS6o?zcu^p$G}`}h4m}!{gD~R8+MZ(
z5u&58p@acmsT+gG{rfPK7;Ld9`S<!5$`8%OIe^7{GV<DpC)Z;Sl=L9QwlU~&<-ay9
zOs}EnSGWzh+l(0>Kj{g?Fp`%Nrpi_7i@Y)73U{ZZH=&;DpxP1i6a2gLI^LYj?HIw#
z5CTG9CQ3p*%_+a^F%b1<H3yz)>8~ErgUJAJxT4JG%($8EjLVQD4cm}r2NjP0=rY0=
zqHP%NG@{{;)&s*Si#>Fk7pf`s@7vS{@>+CY%Agf3dfe3CwofelE@J(miqOH1;ShTM
zc&!h(>@x?4e}FB(kb*Iu>R_@~-DXz#)H`4^1A2}Zk-eUC>dCC-t@CKU9kX`O%s5Nv
znOLdGy<zpIA~HOLuHSzI<kNB|8r)7g=}hnI433)^xi!H}=AfJKY|A!AAkDH*9n)8Q
z%rzR6{OXydYQ{zlqnQrWBt5CNH~uWKT@Fd)?`<l!o#AN!ikusNJu_I{X@aYmVe*=V
z2SY0pitX|6T7x%8vL4>j19o9wZgbIokpt{Bm825I)jjAL9H2Lt+SwGwFs>V}QZ4mA
zL{m75!LriYe;EhXXs7e{`9u_gl1~(I9MK@yyOY?q2c}Xg@kP-r>T;CsXiy+1C$C&H
zubCmRR&gc^&f~Iwuqfp6a8wM2X-;$(lzJE=f<cNe8~k#=9XcdvnSwO6$rg0eT5076
zs5=CD@!d4I$)A-LSFmd1K4TpmU_cqp2DGdo0mCIRInRw_#n=%sY8Ad#NdVoEwx`R0
z(A1+*%Nq*_Z!bQ7MKQS1<X^OM>8{y1JUFND^feH@8K=(SuiMfHj%_aCE1IZ=#vMOF
zB~)>hBE^65S}5hPFv7pOXX2EHnB`15{z48z3gEX!%-4w2GKjX&i<5a{9#pcgVb_I^
zN_($E;7YPNnr7BxG6yy)1a9-Jl-_gt5h<q&zOxIg(m$rr>E;rBqR0F%^{3!mWRfgq
zH8o@cI4b=U7<-7p@xO`U=*yK9VuSk&o^n~FA@oR%#>Wv$STEsSYZ_j>Sdo>P5-@!h
z*Pn7`4DZl;0m73NkY6bQ?dgTFbfa6roPA;`4ATkFD-j`zq&|@z<G6rDg7D&hO|~8A
zu2}J_4fJNGL{NMQYz`$U?rZw(BCrly#!H(DRvUMeGGe~D_uNwaC0jMM3+e4w@2A24
z#I}@bI1FbZf>>={2#r4n1amlI`%7GNG6m*H{3dQu(ox3%<Lc+|X&iJW`+Lg?aUS;-
z<>DQQdOxS7=-{2Yt#p_6U)5|J&yG?cYVM4-xR-AioTVvGGN{f&{_c5_BxLpK&LIV|
z+~dBy4>XwVG@LY>UTPP3jbOH`kHO?^t~}yJWe4nkwvNG1&WA}Be7;{k^D+8Ej<7Dx
zL3+(C#XD!?D*A#%ZYBp{;2rZYc=!$<M``!7C^&NHunGRhX1Piu>MZ@f>Uh8N)>KnR
zV}^kcD5z9bzSXvLOROGyAF4q_ei@chYD^As?YNZnFF{~kP)Zsc&m%neigN$Jnjl7F
zUuqAu1qi4k;q!Ka@`WcHWnpoBeA8C*%(nPIVv-GYD_{YxYD1BD^z)^p8fxj#JMhzj
zksKMF!cZ&lj_K{vmbs_Jr`*vKPq;eSjm&fjuZI$L3}J&Ut$Q>qrm_+gX10MdJ4Hga
zA$;Jh4O1IB+7otTQfUZ!YLZ?rYxKcM06kr8NG$ZM%t$uDks{`!v#kGRuhxjiaESe)
zT^tKk@N9P^E%Dtir~wdE%V^(SbmRznOg9Ug`t67?UX9*?>K-KCgE7NT{c&Ad@5DNG
z0w!EHqBII33wK_|4Q#i?TL_aLL<X5X%jN3K8o8)-M)#K&OK<(Tx>0K2+EvA*TaaxR
z+G71{K#P$0pptgpzmmmyU?ms+2x({0{4F&7C`AMj)WFS1!FPXBKCA|F&@F~a+&hq0
zxyeoM^tZAqV!HnTi01A5P5=5zw6-_-6ecM8neitqY2KYw)zM}+kGKHQ@*Q!?6=z$X
zrPos$n8^92_owP%m_nk|g`SK3(8oD1tstK5kW$FTG#h%5>}$*}^$9mdHB8=w_E;Sq
zR=FFRk$Kq#&$Lk)#^v!c>#5JyZ8CGDmNC+l@#<Z4=GDd%-J83*ZtQ=605?wua^^kQ
z&M;)<4ajhG8B=<$3+KRXLS@E<UJ9wasAy7!Oe*!02^!HGlb>{L(H|wK+wLM**EiuK
z-Jc#p8v*tB^WaU9%vO$!>Vm=l{kEcbnM)LIolX=%=JyR1S`#qLhu1|~D6P%oE#${P
z;)C3@v;H^;-4X>nLHTi%kqXeWc}o)Nvn!knKK9r(oy)bekGE-IwboRI?;t1L+spI6
zIK#fehoeTgdWTnQVxH0ovANhn!JpL(ijRMq5RCZOw1F?2@(Svh8Xttil*k$RBD9}$
z?*4p)3>Sd9(<(VIt@|!S*gJR;Q%aOY0ocdQji$k(efeSzxj}=i;zD_xmXVlQemSi}
z!mZsx1|uEvlU#*`H9waR`03b(7W=`}8vw~?ysR8mXc0V0CpSKhygxG!0ozL)p^!Dl
z{7HFdR659(cI84Q!0XCX)h&7xpt%}8HxibTtdt<p=x5yQAUwN}-6ROU(dXp{>tFUN
zNs_I<0awMfuH%O$I8ao^@Y4<B^->ZvO%W_dGy4hDQ|&Wl0+i(jZ21*YT*>99P4_;>
za4sc^06J5rG!C=(c#M0YgL^}6bgCTu1O-7|kSU+HCFq!Q1;j@U%78d#ZT5uiR1~AC
zhND756>f2%j+i=GD6XmN)Q?rfgSby-^lK>Qrh{Ll&@oY%ZsoypkbMVgM(vIgg$Q4@
zw*9=rel3%dFB6}ZvBu|}JS>F?w31-1hcG16+tm~h0A7NbA#M?LufJH0{I0MqbQk>|
z6l~g(WL%;M`F|6O#c)#`t;}z?UiOs8Pn-~Tf;x2++<#c{&CA$FlFXe_NiOiK(`>8N
zFZ^p(>1%$`+b^o*%M%D_HpY>7#~>fpf*8&5KVxmvlhUF|6KD2mxAYXEgx~imd{8pv
zKe~!IObU!h-|oHCA5oYt*JxEI4`j5R!0Nx;eWX{VE9)rw?uA;`r@2Wx6H_NyK3Z}9
zK0z&sELpW0X3)!*3XiiJvR`+86V(+&N38d3zFzN&a4v+q^r0n=Nl~X~eI|`u)1<Qg
zmLlLPgSJQnOrY0P9@0|Q2*L-^v8e0uXl)!Wl68s9bD|d_OBCQq$0BmQ{|J6&2hy+M
z(ixUVT$7Gx9C~d|!?c{@w`hJSlYiEDB@~cB9$Y@+_0zh>NJnT%J~4No`?uMH!-xm<
z*_&codr+d4Y(83KYWA<r!bk+c;c?fjl^&Ua?5up-w{d3`Y}4hYjinPugtfGY%4$g%
zPhaCIXSpa#JeZi?Y&R}?W1$Me0YouCwA_v}(dXl8z&{OhFI{-|5Mz$In<}3`rXBY*
zhaW-{n(iLxg@0=ku`1Q#0i6Et`)-9E8xK>J{)lLj^I1Q;V<W@LoTCQvp{ff`++5&L
z@X1dgRvQjHXssd0B2Jk%bWfC4J%B;oM*m20`6yQCq{VK5U~CgV3p{6(Cb($QmE8*H
zca$I6=AmStfY*H>x&d7MZ?i7Ep9iwglV$yIL3Oz!wihV@zrWGfm!IK%J#3x07H^;o
zJX+kZ<h1&9aATUC2C0CxwV2vq#DwHQBzP8L19M5o5Zfhs$FO6V(uX}hsklJpB*@lR
zLNSdRh>ZIX9B<TLIms|KMw^18(rr)-s3a$n+n%#7m&q%Jk#amI%|MA0p8K00VqA`S
z_ivkymbdwrKhnS98KM5nx?2-#x{E7hmgxz1N$83k2vG<Q?^3%K+UGzUBAFI#P#-q9
zF1N|^?T`}Duw~gIo(Bf!j)9l@>Gg7qP~_3Exrk^&YW~jTL(E4v9diFBq*q|{V%>+%
z6^dcbb>n@b2{@Vxv_vWhgKnu?cl(WotBbSb4{*JB!9VwcGgq+AXUfxW(^GwlormIC
zk*}pYd<J=IAYAu`yJ2ZjDww1Vm;3ZygthZKdo?s)D0rTc$!@MPi-cOUx+22=bho1z
z)R=Uqe5mZiHRawf2)WI_GlbYlAX_=#c_{QBU@8SqgngMPn?8UAppVFS0I=w~Bsp&%
z+4F%P{guCNEAR{Z5z11V-oUdVr_d*Lr7kZHYxj&>@}*mz>)$`xr}^}t@$a=rmhpv$
z+Giv(s#Lpbe1(Z&1BfQql{-pk?6b4!z`TL$1HE<e6O=5t#wfMnNPaGN*2)Ph$?81&
z-NF>gu%7htGi`fcAI~x#1I3q~&D1nwb|h)qSp~IckbK03p0}@kQ8B0`3iPM#<gLP6
z5$YzkihYsFw`+`_7&X`u5j(m4%xkoE66z&dP9JPg==f;hZd}205>S-rZmI>xV14oW
zTSNpXc6JsIQZKxFlKZs2Ula~OcT%Phj>bCMMA-j|wc6?J;zPA<cA9ae&oa9_k(U_O
z$nAMhiW)7^Kr}?^^Zwb3>>ADdBqa2}#pJryS58Y2m?ZVh;gTY>-S#L0yO&95)h*L^
zCp~d?EK2)Etc2D96ari&)5{cv4IUBQVIip#e9f+GdsmJfERSk7B^`|M0C)d)Ln<ww
z9m4xhhc+J~vqD9klDJ9CLA{=lL2n(Wi}DbZ=X=`yH^iXAG!_J{TMBWGJ6-L~op0F-
zZnDY!@TG=HQU5O}sQzAgrv@W6>G`F&A}TH+TlMUNKX%P+VuWVvQjuS~oZo18&SDl;
zJxBoZh78^4Ift-hh`*-p217pEeh|g8!Auz7f<K-sxkN_(x;E=OmII+PLnbgE!1os=
zE~~X~xginn{IN7k!`>}K6+s5fJn0)DnWJ`{F6o1<`}gCHoSSJ1XZS%Suud#EM(TLS
zxJmK46hZ|-XdZkmyLP|iJ)`s~sg$Ik9>3mO&WOq}7v{W#O!?K<c0@&QC0{F@vaQcu
zoj?$-c>^eI6YsqqA*(lSsXk6;p2{t=NO0VAmH&eug@oZIl&WBFJu{i#xdzq+!^qw?
zMIJQDLj$Gw8*CK^$kodGWvzuSpNSdpGI+TY4gJ?t;n5=@A&GdLWz&f}M#d?Ipb&dx
zLehvllGB{%h{J*h?{LI63&E+Fk0W@FJ48E-FNNeKKepZ%9-M-@s3c(t+3y*S^5c(c
z4*__k)ek1#?1VoL^av@ZzgRyU$1^8)O7>dp!Ed(?Ms?sXU2p#<(_GX5j!6$g=LbvR
z#5FiKenLKc=n7%siUsAdC*Q4BoBpK4b|5g~(4F(B5b0fd{>>0LVCuKvGVg*Bcur5V
zsABxz#|{uin2hR!dH4Qd4A%XQ9j02~Xum!58vnTU-l*5tF?|&?QNmq$5C7A8?!N$`
z!U2&YJZhOvSf~5$A3&MhZI(u8tesUK|F|+0_AF3_AVtK3Xxp99?<%&FU9%+T?mU0d
z+G`qJN1F<u)%p)(UJoi)ugV(Dc&w6)MYvJvC?+8<%zTl&KCn5TU<lClH^Rd7M7|EN
zR&v%az4I#+zlhw`J#9x5Md%PsC;j%2n8E6qMUM_;G|jFBxkr&AeMP3^VjbY=L|&&7
zw*%k!)1qCb^womw>xpP03ELoCeCN;kAkjOV&h#V$l@V0R_&tsv|BYk1Q7Idc0P+zO
z`gm#d6)28ZM)EgIZO9FDHaUOn6r=f-lzW}yjTq*g+h9QUOIN=2{!5u=uXqffDWBxm
za;j#lb+%#PS-h}CMZYp-%Jn=*#+7j20B7;yR0C$g=4JOne|3+*P&JR0eYi@G>+RLm
zb0;{1dJVY}1}gBaP!nQ5YJ5OC|88J>9-9lPnLV=*rFrioTxbc54p|1_ms_G2>JzU)
zrFY`BA0C+57KyywesTAl*XR}(+^DpZ%%<$aUH_3i4wLtGh`6B@2P<CAZkTZCBVE)<
z9c<}`{b^3;QfGf$3ulI)%W|ARnNjy+F!tMvCt`Tp7;)IcR@v+&<yF~u%ru1D0A8X@
zZE$rD1}SD|$HO$=x+<Q`ug#abWx2yNcXCR{AA3*EfFpK@-VEo2{@HQ)p*I%kV<<(w
z>#SAiw%AM&Jk!6|rD<7d6&$B3g4M8BvNSue{=%Uaj&G^+xc6II5G$oWISo+l^Zfyz
zVFvi^ZY^+Dw!jP1*phmGxnGturaqy0Xm&1VQ;4?9Sg{aVSf6NX_-Je^yEh{H%5<0>
z`s*A>izwUVkPqNaszi0fEeLU!d~L{Qk+xNn0-(+oU7ifLO@$*JBOb%4j5q!wiY^0~
z#l$)|zuF_42Wi73CJXu`VIQq}Dd^dv+_uGj*z9XAbN%^!Yu;mrGrdeLUr-`Twpo1x
zf9=Yca!#YNTWERIJ(_Fx*F8J4EsFh>?vv2r^W@3HvyeI<^FPAI@n-b{Qqq3vIFt2L
zKy;sVyt_%w%st;Eb<<P%1A?hq=ppP$(CON7Rk)s4rewXFJ?0~S&v-%F{6xR-)dq#b
z%+I=Y8$=sCYzi&Az3iDtZ+3Ms$8O{Z+-zx+p{>RF^frehea&f<Dvc8_^B3#mMK0LW
ztr{lc<dyQ^_4D1w3V6rW4$av6PDKe)a`sUOb4fG(HGI$9eOBHVIYz$Lvg=`LL9Nqd
zd#2>=`}tF0p2?@dl8MmkHgk^<$L4i6RTKxML@O)B&f{%`TpWgG&vyk&Lp{UPT<wt9
zC;t5u`*uVq_y&T+yLWVSR@qk1SL#4wb6L}y;hDu=Xg6*R)S4LlVw9gLMB_at7xrqH
z-W=Wd!eOSFS3jY88-^lb#lhP+tTdN^-eVWu?>Za)cMr<@k5cKnW0L*Qv_oC@xIWEG
z^&_YrPKCU#BP~E_P7HOK>h)&ges;!yGZE4c`i(M)g0d#49`e0ztyfP1Ot+<6qyC}b
zO!LpuleD~Ttlv(;A-=L--M<n}zQT^4q+>dHh`mVq41=gnEH!-ZTmGHZu4_UEMM;{=
zf^4&?+sK3I%Vr`1>F;o-wuD0KOCSeQ?8~*_bYc8@*bb+=zvwB^I%{ExVjoUMXTvvf
z3u#_S75hFeGs=VM#U)2wE5-`c=4H@9g`MK#+AboSn&q%ZDf5AnTYt1axZm3bl*X#Y
zqv}`&)fJh}NZ+woNm0o2j|URn^fPz#OM$$h$b6*&zydmvBCR~`XX2O);UWV^d`R0z
zt7|{@S6*C)9gjA1pOmv#At|=4vn3%e(X-#B1pAO^O$ZH9q!1^#W7K3)(RthVlV{PD
zJ|2GV+OlI2z8D(Zl$8?A?jKR+Le+cg><U6zt{+d!FA^$Fi~m&zRb6H9xNP|S{jD*5
zUq?oL(g-r4h9BoyV%LXtsl2zBRnvOq-KF*6&cLx1%0TN6`Y|g-)CC2qSvQ!0v)#e2
z;`mdddJvZ#vj6OUN@Jl$gjg9Dp}|#av%@uY1a7G)Z%K<_39di^@Wr?EzQwTs<~BvP
zGwEr#C%E5;)#j*7e~TWpnJHbhjZif!p4WLJx`j~Xjdk?E%@}{m0=lCVl#S!oo&Prm
zMxk5Mxz!0g6KEixiHWPr#7zKQRSA<DLD?q<d9GDzoN6$?KKJ$y$r*(gD>;b6EM$Hp
z9Wk8mk5+%I+&g`_9xY(xi<*+#UU&&*<NiaMYunSFG30%x$^Rx(L+rW^RO8bC_E6<%
z-VFr1V|j0lgMsUPbOb~Swn>JZc+;X%tsP*6GtbKv_l2+bYkN?(r9LSf{Bt_Q!L#~f
zZzLRO{gm~`<J`A)_FbXp@4NT%;wy#{e{X8j%}ePb9%n(*21Yc1a)}Jm6~BD?-`IK#
zs+f^dnhy&i9S3d1&u%E!uxL(3XN)g$UwCX6TGQDp6f44bKk_Ni9RWmy!{`C;062-j
zeF!rKqw6bk1-;yQ{<*_hj2^@YHA{Srt`J!o%&eq#Q{3W%mlh{Ppx8A{c(5@KdeYz`
z0q$L8m@;E9_=&VQPA|(%fb{lMbS8L%-u@<r^`}S7?HNeFSxFL{kRjEp((>Z+r_ApH
zN3tj{ONyUW-{CRPBO+~PUf2#f)Z;d%ftwsW+`NHx4GY`}6-0BBNyXNqJ%&%aHwrX|
z%??kW{9|*WA*yYo4@9R)BOyWa3#9l^$OhH&jABTm7l**<#^91?Ok6x(AoqL|3f4%B
z@c^r*mAefUkXcvds<Eye!Yg@aIN|T13RN8Xba8VQr*&9R>bm}U)X+t}Z9vdB(`dS2
z26ARm&{xk$3;XCA@L^E@w3f!Iw^FSz@F~FlJFRc{{!jY5%xCV&{$0K8Rz7N_4{+Yt
zLAbuqeZO|)@G+NO8G3a9vJl$+&7)e<Wrvq#s#^DxM*K%L!Yr>v?_17k-NdEJkrAz#
zuDY!DdeJk7OYx>0RsGwLTm@1Lx-8++u6U`5%9{TyTFG`~sQwz}Jq3;Q=8x7?L4TIi
zN|TS|^vP7JWxF|1k%HW27gk=UR>YkWMJLTvPcQGQkBOX&nc<a_Vog1F?fZ?2XuEQO
zb8X$KH1`8!BGMh#4v`CU#Ls#@mTN6W2+C%VTE*JU(Cq>VMf-!iGHyCYMVrN({gDfI
zD(r}_?(-jv`JuR*0rG$817w!8v<SKEzx4$-C|w)9@wxN#g>x<{<y%&?M|+=-(W74u
zPoq?31d8d3Ow1y^@II5lkdftaSZqS8h={gTbifO%0)0Fu&OU{lRI7QZD=9s=e%6i9
zHBM?qkF!uu{#Ojb*y;;it@~KIF~G+&mB-f4*ztg5H+Z!qOn009Ndj**bCT?#a|3wp
zO+y!`@%q%eE5+9rML1j{;8~?3Awo$r=FAD*4(%V^M6&clNj^QY8?$e1(~HjU(7{%0
zKh$pdOa)`iI{taqc2@`!9fGL~ci!kyq%^MOBkh`C;LuxqlP&7J>Lq)nqgDHLCOU(z
zeBlhRWB>*So;gy`4{&z2c;;Zp=mZ^Ga_#jbZ9xNu%*pL$-^F>5Xe47T<#Bf!;>~XZ
z9}Ly_FvM;7((EAOQ}?+2v46;1X9|F#TCDU?)UB0K$!A>WbWxiTdoX;A{2(r*Qu4Ku
zRBhf4yFULiT~6Vtdh|It8;3D!MzU$Tsr!y&3*6*jarwb2*Te@Lj;jy3Gi&gmS|h)~
zr2u2(7^MEjTz^4{x;rfTwdJ5JR`bvrKAU3ku#nLEQdCd<8h5YljW`pZ3$w@mky(Y7
z(U>&B`%RQt%s|N?X(t={y&TgZ?1q1;nW5lR5>Esv3hzY-;mwRE=;G)W&#|gfIL!%0
z62xoOz{3e4kRJc>%Ne53dU8A9ffnJ%+VXwiN+WrFGuVkm2+}nNmg$!EU|v4NYv620
z>HkL&X`gY&(qT3lg@Vj)>40a}Icoh0U0NwGW&SO9Nh4n=rcYpxHhHdhm)*p-&92-%
z>5#GMBA-q7ug?TA2JMck+Am47MB$-Xr)iS8roIkIn73bhys;jA-a>q!;jQ$f-gr1m
z295MacH%Ddz=SZXRpESdJqr%7`n@IgC21tcv+j%woF`n*jTGhdJB9<_i;yfwA2LfY
znh3t$+f<g$+Kxv-{$9s|H3?A~sYPEN%%U#5#JP0!CuT{ar*BOD%tnP{=XUd~drg>V
zn^9CLEbT)OXwh0{NROs-pqc4#(v1<F6+c>Gl*fc{nsqa$*b%K-{zSI+f@5Z(lcJLD
z9w|n>v=al{KEri*SSt|xVmIePUO<o%M(!wO`=@fU9Aqo-hE~6^eyw41ZigaxV2r_e
zz9g}MwMfK}YUhS#3%|h3e>hYd*%o8L&}OZB(>|jw7c^__LX**!8S;#9DES2_|C2%U
z<hlF3#ruS-w){|JX>A2ns7(h?Q$7ZYE?#nUWMut1_sdfro;qCwxwCxL>V17_7^(w`
z>DO#nX~?h9-hg<n8qjXMi6?Hn6h`W8E>ODEDqoV%F65C!D>=7a6LVtr272Lp+T7J@
zcCc&&;tmfB9sLmm`lTABIjBh+txkEIKk%@u4h5x4MTqBXSg0uzrjSbS0dE4<snCBF
zHpB$AX55*^GL;kPZ8MQ&;hanBVqD|IO_2I7&bfcN5GgOV?)fV89(TOr$yWK17ny;4
zOa3@GPPb<s_bPzyu@Kh0-po#9xG7ZZ%^_5Xqk`DwL=VzT`(xxoDTJ(c&scO;fT{oV
z;s=yU`x++bWV*Oi3mJ~2&2D^P^B^imNEdbF*$`D?Jr2`owcDn6=Q{u?AQB5+d!`{R
zQJqc}RJPbSyBbML+yK@^78jLWbB|Ew@7t#E&GA*WQ9S^`FmLEG^LY|a^EJo~G^q@e
zWbP#%gl7`Mj-B0@BG6%_`98BQF05bU4ZGXlY*BE4_ORD<@c9e!EV?S?!vy&8H!d32
zKY`0QcpC(_qIO=;?nN6k^e1_B&xyi)<b7F!k(kaO9GTeOe(i%0^e#|Om&Qj_TTU4s
z2r9owx5JLz4hO3>`qoZLO?@_4)*lq97LsVV)$j)MB-e#oSvtYQb>s{?7V;n}#^H6k
z(h*1Yrym+H1&!u@fjK*S88ue#M6-TzXj4{Iqh046`kMMbZ>;b<nqxJ;Ns*5UL3+#=
z7=0y^I<f7cd1PZ92{T<CM=ee3WoW8&t1+?4YIV@W$w=-5ryk!YX1-6oymb26u(gO9
zvq0;>1AU`itcY%xaHLRtFHGdy5S6olYyst3m{7r8)9Q_FwX$Usg+n7ca4Iv=YU({C
zNPeHIk$~~QHJU#)DPIi8FwW@09}B+}sQ)GfKHc-8dN)(+FVM-W_2*LRE`Arv(~!A(
zEnJeb+3>RCHM7XZzj%>PbTnPz0kNWk$FaBkW)nD8TX(699`8*(l~o*bdlG1S?yaOZ
zGn8zL%%juY5dSO!Cz_r(XFVZ9Ybc50Qw*iyTwvZ-wx3N(6yfgqKJg+X2mf5`*=UbR
z-{i!g3(AMrr^S|r1}S5BB!5CQ@^g)sZ7Df6#R8$~UtP(@x`9c~3pXm}QrA!F%r-`-
zYW5U;*}87%yRfP`+C_O$Fa*!^_X6*ehMP#(2W{>DLK#81qjAHgS;pZXg!Cod=X4Rn
zGDZ@F;!ffGFEpAKIu2*60EtUR)4g?|x;qBpFxY+aBVVAmBU}noN-F=cQL!*eS8R*v
z4;xg?S1&zt?cvC!XR4ER)TeCQ;K;L>88tjNXoO$PY(`0~5i_jUV188EiN8pWjnpVw
zIs4NTviP0Y2+LC=Q+*vX;$#HzxRu4Kvi#ilRU<p~1Uq}L|B#U+82$_5i*iC7mX7Zw
zl-RGSEa3h0tg=Dp7@$YO>kSl1-k4A!8>Ag!Ms}{=&PFo+UF)&3E0$M7MR!jgR8vY!
z`isN}gjWTq&>>yKtT*j*Ds^$48elRVkj{{emy0sY?azJK`g^H_9EEA;kHuVqAF?=@
zh9BvwBy@^<$&$I=B*&3FnOg>xfcDRTuScoQL5zA;6am61*new|1m!Q4m+x!KBP{Y8
zP&>tM+gN+$fW0FRL*G9TMJ~q}g2yRDzM(&L_RpIQ*+1{J8u{u*!nNCWoj(KPWGVG>
z*+d4r;M!o)8Cy_J6yfN);)YwXhm(Jr(eP0hEd%c$9Za^eSDC6u%kqIG&Dq+>6xBqD
z+8J(4O*oOfcIin`ELFrOVSjiV%vHHUqO7iNrW)+5DI(!c<@Ct%WiJsF#?0DuJn4*}
zL#oR&^7r@IEQX$meEWEUjz4z4MD~s2Kq^V(!LCEz6=!-G@K$Nlmnt15DE`|gLGhB}
zw|}WdUZ7Z^k}gOd@LWW6sl#9Y7s?7#>dxMmPw54@dcr>eV=KmKLp9$hejE(HLJO(&
zPOT5VW<G9VaKVhpN;EH=Hfvk;nCF;k8H;G6+!4%eDCZByI!h!xR=pW$)V96<y_r^N
zIC!<<z#7JV4)5RcU|RW(0g5i<<3+_pYpzhw2_$MCSbxX&2gMEgpPH<l>7Q?;fCvU>
z2$fiC)V`+{9?aOy1dR%Z2}aN?EXmbhxW4dCA5hp$7qTArFl4h4S*FI$_QWx2Mr`+7
zT3wb!rwI$E)wX*~E|tNd6BoIEZO07UsHY@#WmDM5+D9@aJlaIjbTU&T?d7z!E6J%h
zwzbd3%73c%X+i#bV-fKjuW&5Rk-LhvvZSTw3bjqpp{36#)f2pF35iTA+k<FN(~Sen
zJHQ4N86~5(L5Hh$(ful8{2tr4K;NO=4Ol<bx^$SB*zgwD1?rhDyQ*(?^dO1b>IcBD
zBqT@XSB|{tcN$d1TzW$35tbNc9FjvKJ&0Xi#N!2jOB5^wG%#A}yp^jv{DcmdMP5?s
zHNh&((5D1tM{|;f!%3VSKjRks!S^?eM91K3UXHPPv^)Q|!}})M2&j=qo3_Kuo%;#j
zPW+=yNXr+N9xii)dIjJ=y!@{W1np3$FX&a|;1hpfEaAObZ?=ryF+FYf5nc~2>L&-h
zj`af}6x<_vlcf+*oIpbw;!{nucLp94*fIwWq(PW&8%I(G9HaRO?hNCQmr{oLg{X1I
zR~W>_QCb2lobBHhiPBrCdap({BNT0#ix&2z@7g_rXy~6@Pq)*hS$<s{k;prR`*)Kg
zCbo-TZ~=wFY~%q~>-_j?&LWb{9Sq5U4gW546PqU2b?2LqedeXcACsQZ+|!fW+N_NB
zidA1d$GgTW27pTb0h%Da*+jPd<%33NT7jvBM-zm#KLT)9Or#Y8uuE~R#X-bQv<!Ue
z4f3r*ve<#ztI5ISNnOA~t>GL1#hKNxC^!`cCAwZ(76I%#HZrah*_r1cl@k3cWTt(M
z1Msbhtnx>&!%!_i>JhgKD%WN@0mO1!S;HwzZo;+dyQmew)Hxi#LuY>8lh#qoCHL9B
z0a?5q-!+LOdUN?#>tGXTRqr_03{Px+fO+7ZyALbzu6zzd2vgkvg*m`(QO0SE1pL6Q
zG<s9nO<D%4pLC%36#(wW7ij#^!Sy#x$|p-2LS9%PX#5=mAeAFtzk(}lsl9njM+G4_
znHIVW`^DvtPfd*kGu2-qZL@bpQmn%My|Dt8X8pe2cYG)~KS;=8zPQpQ0ffFv@gQI%
zVfz<c5ps+F4Y<#aE_9ntct&y)bPn7NRfe_)WppGj0HqGF;|FDgLy}WkT(EL>Zrtqm
z<`u7lyJZ+7p01EE7^w9dsnCfNC2>}p(Zo1oKl|GLOgdZPZ$H*G!HiQyw58py0BvVy
z5ms=!HK9@EY8k-EWYE$#h!#jH^9{O5VV7f@9CQrDJ9i<A*1)8VG>%GQUFekZ3zs>h
zJSy_QCb29(QH549em5#uNFsvA=;_)Eb5=-5{d%cph{Krv{BUNljd#XRZ!9N>C8@Dq
zX4NZ0uxCq>=D{zBKpei@rb4|4iu6ETj#R{+8#VG>tE)z}qKeXD2S9vL(;B^-sH#;n
zQbAO+^pK0F`C<x#4GFG%4%_O$2}>uAiGi&$1@K4EsQIJz$)Z8xjm8+Hipumgth|Vv
zaBTKSLvm3hE3fRK-3r3_vZd5b!CV+jA*U=jDI6I~c}6z2AQhZIh%4nKNX8uKRCChR
zI6OldVpsIpMRmfC&=ceTKxH3+{8cQY)v*2C!Qf<<DuAYuW<p3}<R2PPJq|s^t2zX^
zymwH%p5@p+y_M<^Pgr;HI@b(kloG8ctfuM&gH5tW1Rqz*3`_I9vcS>E=QXV{;rPQ=
zlg)q-<Q<@|00BE1Z^8-#JJV?dm|A&*EIs>1AAcSS<<sg>-=gYk=u503T>hP}q~QLd
z(V|z&Qx4blCHrjDpAQ9V`KodGx6?pa7&nL-_O{Eg`cbkptEP2&<7ObP!Bx^iUdxm5
zV}H$nhbjPHuplDdvffGo6RQgvE2#S8nc!7*aSOFC=F+gM09~AgBEM&0IHi`%p5x9H
zUEToUJ>mi@HVZ!mkTOV#hp!dv6AHZF&-b~YUr|F33hj!DR!U19ar?vsnNOv?p>jgq
z*{uK^a9C%W7PA<nJ+5bPbC}IZcODHw(Ibz;<c;z+MCh&p14`foaDJy2hg;bIOa)TA
zki2$%#zLSd2v-_+MU*(iz@E4Ax<4k&FSm&(DhmHoy~3#FT<(xsJ)y_K@z%7nOg~8_
z(pg?>+pLwidw#Sv%C?AgVcDl*s{6$$Sf-}%NpAYU6~U;`lAuY(Vji6eV1PZnMFR58
zAoN}oTNbg6WRjO~-5gyD^I)pk@Vtgh9_Lc@i5s65t7)T>BT<EYc!JL8xN}e<Y?>Ft
z)92pj&;qZUJ~Pb1kAc5^z3D*@wa_%J3KhNJ1nF^k7o$1aHa#{aLV$N}YSAN<fl6-(
z%2PO}Wx=33f5VX594LzO6PGyE6qK2`2`n3W$^*Z>^yKd3PyQwojeLwp^sBv0hq~W^
zF(2*w$?OeGuMJgUw$v29rNDX{RTQ4bm0RQCNrrY#2i0TYw#uK4LF0aoKzXWhTK(wh
z(O6gwIho=etTN2(LegX1?+r<h43%J@NtA}vmFUu7V}2+;l8!iUJ2rxZ)>QRKzuPHx
z#&$A$LoOs;^cQMkhQ0jx+aar0JbpvxI6%Q4zR{i%_{aC<ScOxa88sk6=&XV%L2Ec|
z<j4&qd8e0zo@uplWW!}rEydXj(@?f5Xmc`Ls_W@2^}|}~r0EJ_LqJ^6f?kpVXHx@g
zrhb6M2|fEn(CS&j81ap<OxIMFfS__`V!k=ymAL}WTofm2@?@)DY^t>AU7!Ol?u3z4
z`$LTA(09gfG1q@@5%1=Qj7b?%5ssu2tpt&Q=i#bWsxUzX9v23H%CMXLA-zUQdyL@7
zl`((TNeG33AVL?{*6;$u&kf4Rj-*0+>Jb_bR>Tt>Q)7k1d@8zD>58;i9uG$7sJu<<
zquC;cE!6rCNoys!6;Gyuxu?jB;11*z+%yI$1tr~onNS_Xaf&gq(t&AmJ;59G!vf`O
zNEn2Yr)CooePBv0q0A|aLN|2JmHM9j!YS8nVqDYh2c0NflDUjqFXsm>!TEw<=lRk1
zXAl=$h5{ZB`1j`Q)<Y;l;}gwuBa_oC>aARvqKB7QCdK=*nqhrv!Bu9*lkJ6y+FV4B
z)F(@0_iobNj&1Bq?)Nld+s0V2#85_DWaI0=Y!L%B;vlV(AU6tZfm_iWK7Kw8a~Gof
z7kcwX#>6xfCyK(_Yqg4lVE#B!e-FJ2q)UDU=xgq*u@S?5NTL>zP4dRrhGjdES2m0%
zMdnj|s8jFaL}>m_K>#r;)X4to46^R}Q@dG<$bFgS1*!7AL6s9xy=MHh;9~(T-;+d{
z2f%UwO?5EyuL%$14cJR~w#yHj0joy@@-m~aR|N+J^V`MRq*JV@lCkhRK*?C>#3tQp
z$S&C<r^9--=xa<7;8C}+cNbt-IR;TPSauGHI8)t{31j`xL0u~c8>GW9`I<E?<~9JI
zld{V1QYU-3tKw8zXS#a{!x@l{DT2Vo&VFU9WT?HKpQKM5jKx-CEn395vqrnaT^+$-
z#NP!;&=$iWtqz$4(GU;|<iyi%mxSilcTT!|A;A)`fEPo{<-Yc_%NxmiV+}-G{i{-E
zOOA0J^_O)^2Di*bB>mbtp_E8(;8N0F6S#2zH|$ZrhY2L}L=RL|{E4CsfVQmq3$wWZ
zQ)?M{c3ivEc6rJ=3M0)p&LmYU(rKQagz2(|Nm(s+<!ZuJ(t~kAwRtQ$lh|L~ZwhSd
zM)l0iMAh&khM<=k<9PI$&@#L7Vj%)60@J+CC^E^$;Fc;0kbPDnwi}gNLTymK)9;7O
z#5YU8eII<hZX#kldYEGEQdL?*vpV9p!+XdE`=n&?^_uWqqVTEEcB{aEu0B=hR_qR=
zZ$Kwg5q*6n9H86j>ZbV{sa9!Whw=#38&g>mY165w)o0x4b;R4?*2y<$*=kg6og8lm
ztsXpO83*bi!#n(6(r-A`5!dYiCo@5#z#)nhw7iPMquB5?&Ebj9Gb77wg)`(r`WwK<
zr0)z6Ttz9Y@zq{fiK91F?THe(#ICdmW1?~iXvshWf)g#TQAr;QdOuR1jS>FkPo*@k
z!H|_hz=w-%!D8<V1!nv+)vEKu#C)^e*_HR{304HOBsN`}D!j*~d5uZY?63;hHACu&
z_U}huPTo?J$a1V=G)WS#5=*Q7-@gDL=!1cKnPJb62f-Ty^e3oG@wYX;=fz_$i2Vb1
z7nv$&34W+y1>XG)|Fqf`3E+-Qh~ijW?ZOLJ;mp%IH0%QY@y_a5g@@E~8lC@SXCw}9
zgSe37mj;uuhtC+RE<RIaali*16!>|I@;?BeKwrPe1r5vW08B2D^mD!-TEK8phb)O_
zn)n>18AZU>T4opsJ}X3r!&Ix}1F~qU5>{)DO<zB(nj3YQ283-G$^tHp>;uV}8=-z?
zhMz)GjoENHp9AHaf8;F5Q<2iwUhRes1><Vn0!J!nnfSu7U`@B*u`%9YvzBXFGB>ns
z5f*PpnvGR(ZNoCMp0@>ZtF)|BnHby)OY^x5uuO%V(VTvZEm46cQt8-%?!w^pB>eeV
z8xKUd8i4Wra*R)?2L7v=M`Zz*^67ClE;$7+Wxu1ZuRt$ut6v|FK&L{c<XeyIiJsQ)
z&|}f$RvVY1ZH9FwEm6n9HHv0fbeXW(Ys4ZRs3!A0AC?3bY&Fz9mc%)t_4ynefcPA7
zb1%<GXPtz1yAW(T6Vd62LVgaGWgY0j?H$ZB0b8`h@29vcf2!NCYOxL99XuaZ${_(|
z0hdtfa5j4G*(@PDwcJ%81*|hIcv?kg=?&t2CS9$5e5c5<p}DSea;|w=Ww2CTzgKWk
z+plv9u_JXD=*xftmJ^3LvL2T831retxU09P!Ueh}aRQDZJOary9>-}APwu?zvQ$^N
z7pc-JV50Q3axd;`q#X)4(9%GS47(zRp%icl9|M9~>|cZ><YAO_Wl)=c>xncgD^fUD
zIl*Xq5@A#RT-cnihkw~ME&vH9Jd9gtdMmDg2ON18@wMEnX2e*fx$2?FF?mQYSl2y-
ztid<aYXje54q(lvSad;npgw?7Men67B(la|%I?e67o@B#;4%qcDc}+c9?m_bfJ-Fr
z_UWy@XY~kRDu@LPZLtpNj%@>*?1qhg9ix{M@KA~eB(9x4k!K6pq1Z~tcFv~5`g|*^
zUBr(cl#06x!DlOBv)fwj$i;jb<^kKX@N=lp5#i<+4Lo~qgYrwv2ONUZ3i3Vh(N$%E
z1Wb@pS<F{AH%bAQ^7)NY7gs4ooaiey(C=RX!vs#?<?#t+Q70`*zrQKLdC2p~LrmIS
zvS1By3%#ksnmq&)0LO$K*wbsLK<l7bp9YnQP(7zJi!X8*KlG-#$)PP!H8*UxeA&{#
zcuJ~(v1iyTAkLQ08mRO=DXM|7R@4yuzAP$=>1t&LFO=QY%6ql(9YHv=U!I8y;DV;m
zvslkP$Zwx;9KV9U7B6_X2eF07t8jUg(rk?oMCDLmP1-7FoL77pP`dEf;eZ;i#!-gz
zVU2HegYnbWXY>bo=eTF$^3!1u7)p1^2SHV^N&>k^Ajpp)7y1e(MFQB@s1$IDJc@Z~
zhC;qAU_DO@7!#{?jEBw$G9D;vsq{Px;5?Tt;DU}%HiV#qq10{Rv6gJ>UPh?Z6e!^r
zaG1f+k%i728QdI{O5B#dGg^8J8S#T4(kmVAVA&&CMkx!Qi~Yinp~tr{Vd=!`yR^XZ
zcvVpnD2@b_0xphRZK>i7s1&fRWC~vyvMk>Um<*AR3wnT2z1-jGpy2GsM%?g~eMoT4
zHApK?eRvHI*77XmK7PWkM<`%tj{u4HVhuTat5OCbD5N7PIw)P%xfYrd3hnWWjG#qV
ztK2W00C333A{OAt!x<Dao4FZlz_pxkC?y;~hziud1e5~yFRQ#EH6C~M<xNKs@RooT
zHk=hsCc40m*BlEm9Mr)1%s}YUt<1H<{oud?_S~0>^p7J^ti)!3Zq~LW%nsbQ|8RTD
zIh#jXfV|Huv+MoT@ICzIn|)i*I7+a@7I5|$;5U$-ia?XAY$~coNx)tLN&(x;r6{!_
zp_O_@UzBlF15qtTA-{a$$h({@V7!&rVB66Gzx^cPG)*2{;3x0}mKS-Z2?WqcCL`iA
z;cd>gQi)tGX>29-zdcA;XTVkaOmaN_F9Io41eOJY_s@oZNO%=IBh~#=t1L)-Zw7ft
zJgPNBbDU6B)_AIs8>N6t_|zKj*el&{S=LDjSr|24L8<z-fQ=m1fMNB%j$?C=fN|tU
zk*3R?Lhjb+2sD-=eI&u)X?dO*whd*Eox7h9g^ql=#D`ha$ci0_OdYR{u<CsUH+Y~4
zC=Q_r1n<p34iMZs3Oxs^<wfD*HC`WkNFUd~gVzn7$#v_!<z;B}_O29g37<VICdA7q
zH*_gPjx%<BTfnBKIVZ8OVHOYZZK8<Wa^zQ$CaJ)s`8LzEW0Msg7o_<J+!NH0N(PEf
z@|IcZ@%3|De?^3|OdjHqQ#y_07$nx1fWF$(YF%!S;&QV~b5@hZqrL*wvJeYdLC#yd
z3fWE3X>r!W!sb7<u>;R&V+ZGO!#o##&XoPa60W0V%aksV!YFZ3LNBqv@og`=TdmYX
zE=PaF#mKs-2LlGT-u3`VHe(XCowGVvy&Cc*2QQq3WM3fTufPR5%@dOIw6TJCwgo&y
z#7&_Y=qu=6-zj`BDgz6BrEtWHey!(NQUJnwU7(3k3TnYF;pxaZKA>Ps<Ut75?OM2)
zAJxY8F)b213(@W;gIv~ozms*OP2%$CQlSQrK-Pl@fU7`rlR)|1YPs=URam5P-!GKr
z`(Ky)O+DB8uq_71^hxEL+c~ws--KiOPV+5DF8Aj0$!_paYI18(nj0J%!{LAjc`TuG
zL9!0dNa###Y^^(-Vgqtb8?3k&IO_u`U^6DlzFY&3h5xm2_u-eU6{l%*Y7uh<T+Ewz
zL-bgtTW8{XBqQ}5OGDK6vWO!M)eu|)N&yF#S$-He3_L#q3fq?igz&9^hv~iLL3fj2
zbqI*c&GMDB1v8|QPg9nzeUh=&wH=L6>RE1MShADQWMj=z7HlegWP+NWdhh$1*eQM@
zqhmS+Q@}=jPQh`$`K~cQu8SA>n87F6*ntTw9BadjNG>b*Dwgm%mCF|y1r$&MRX4i}
zIKGMl*NiccIc)}G-N?YS9~%$r5rliZaNDO@aigewhBU4PQowNABB+WTZ$C1=4$EXy
zq3k>(lI`*1wd-)$kaYj~F0kV9{qQW(P1~3SCD=AN8q;3DNOf$IoBrqGka%iJ&QHmr
z=T#xBHZHWsNz>=sf^xQX>TiA%X!)hLkD24soa?rfi(4tgOxX6T@!b$|;Q%h_2CU(q
zh)#Vxn&PsSppH0!H}r<VUbvT!OVkDZ#NZsfCP67F_#hHc3OI<g3IfaJzRrW?ALm2j
zC!N~4FgFp!>|b65>35V7Au%?{h!I%z5GF<!X5hfXs;?j(IRo|i{Z_O{;1@O&zGomA
zU<RNZvJ1%*km5P6M=WlfH|XUk7GySSJS1JefyXP%*h%;<KdI^2Zg%kb*K8);YmA9|
zIqUZnfjg1>TMtCE_kr<;mk=~(FJT||(fN}u00qN(Nw13?1m$~3Uet7@fb)`Hj>yUd
zoulA#N6X%iiI-C=*Fy6(t^Qd|tbIq7=FY=tz|?|lt6{MaJqW2JU7m>8#j~ixae<LD
zQ6E?trAint@7JIV1HZ7rgJ-U2=mh&C^oUP~HTzU5AIK~<7g5Wu>%7>GGz&DJNW}|q
z9p;t^cXV_GaMIk}==(1L=)^%AqjiD27(wah3XkXMJDEtXefYvN#5!I-cnFklwg<!)
zfe|3=xprM69)VO>g_M9&z=f2iKU#8q&c1MdqNdf=K^Z@851?xON&t$+Q+cJ*I`IEq
z>xNVDaaf2Ok=YSeZ~X~Snn$I=<b2JwojKqR!rwX53lHfb9hCtWTpkV~UDS!v%vp3x
z;>PJy6ETe2Jz&cT7xfy_Ikq7%s{>~Mufn~w8>3>d;qNXuwoM4a3cStBmQaB!F`%-W
z<ve7)ph5X!@?MZhO2B>K1(xXOC8rc{>7Tv3V&J?H$J9w{7+IFMXAn;5*{_-&-Q{dI
z3@h?0#_JEjI%yl4Vdxa5V0|qxEzb1K=-t!u24&k*jwI*U+dCZrS!q@<sxh$;8)dsw
zK|I1FSgbEPwfm|^N=&$<7b17`vy7}?wJg7Bzc0KLAYo8NDqySwV^4MY3XjBA@w%w=
zs1!^-#QBu+jcD}ttQ2qwpMCcwlcbx+-;bB`lek{r%(c3eFe*~5m=bba)m^m)<=!BA
z6neMdVL!;{?6fb}kwt8Oadj1^Aj<Vx#{$>lsc77u2EfIRw0oNyPS$C#^wRNc;JmBo
zBqZ#}NttVAOin;t$LHODF(A0J7poZEjnFiC!@y%aY%WBV=j3bz&HD-yHKl;P#Z+(z
zrGN`AQfGn|H3U@(b%>i;q#Jde2-sTVOj)&LOMqr%*SXg;J+$r}K&)_w3fu}krq2qB
z#Wmfh;^FRa*tiTOyd|xFn8s=MHa7qWGcw<>X$o3}&}-|DL6PF{BxaB`u#;PZD;UN8
z39DRM?w-8o)R%vx7#Jvr0^W}-R%^LV3;eSh0Fsk;!j}N#6sSK5C<W|KPKANetej?<
zcwJ8CTKFWc)vp30<$eJbMixb<!l!OLm-#-DFu@K2pMIaXF32Zo&%gs6bm9f~vF0<d
z2DXbwK`tkm{NuF|FyU_vlf-FoalbW!m($hCsmqsl=7h=ELGaxli1qw7Y}Gr9i|C=;
z8t7tH6fqK(oRGRtK632T0}Y3o(IyX=!Zvei*T)TRL$DQ=4~kScKe$Mg6<HSvC<W{y
zoPsxLJg-?M)}9t>dKfNgReVT*PZ&RLh7Y(j5U$t#0Tq2|mHFIgIb|5KpjHp{(=D{`
z$TlNq1P&{>C(#UrHZF1P@d>ZICA|W<7tZd<6&ZZa$;8EY5#284Gb?ryvj*M-Od+iF
z%x>K7b(|GE8*cTjWLd@kEpoO;8`D3CEM<qeF_m{4vHZ3%<;}q8B%l;<37<P&swd^&
zopA=-ldtAar}in3!2`ckn(u#nAAAfsD~(jC!g^eT{h~QUQfo+A=`X+rA;*3cFiuEf
zC&HS1OY(;`Y1j0Ysbe5voeGO@YieD9>%FYM{*D0fX`tx$*tCl5n>lb#!yO##RHke)
z!D|p8DFfUJx(;i@1wr}ZyAY3Nw`neD)r7svzElp^dO-(KZN!uUF6k4aMFr?i=R6``
zy5(@c6@dK`k`UbM1#Zg>au$ED0P+>Y>n7TE;8n>r-+RH4K*XOr8IgRjg5wGps~A(@
z<MtcmrDMr>vlq(_?Z0Tg|CW6YZLZa{bZG=*^D|+c$gS99EYrO0O0XD>RNQkKw7S8S
zm@UbyTO}!Gj2~W-j#MW}2`B}e6h(0_h9jUIQXK~Fdkz9?p9qXK$`BAqJTI3ax7`2y
zCRks;%>_4!!!<k~K0A3I-~tw^FRNCPgyQjUg;6}!rcB&{rEcMS9h1mB88|`rGRHWX
zH%K0HDZ2;gEa3UtW&=0$WJdik@JpO$oomI@z=ruO3kezRT(SaXP%>p-vO!gNM@j*g
z=mA+;cmcZd{bKcEOVbRvHu+vJxKZAolLD49Gr9I%d^qm0^^hy_@`in_;o8w5u1&YA
zx2xlv{93>=d~X~k7V#}X$O}DSN6(?I4odh<hc?@4Xv5nkFJHdRw$VqWVA=<KfF(gN
zqnOS_6kdPYm8>lC1=q+bN*9{gt~HpmK)bZ>XTjt|OO+BzptKLTgafOzuJD35yP!Pb
z#2c<fa@t>6P(CV~*sQIK$LF-JHSK}t-p0^=nal7v*Abx`Jn~ugr7^x6%IhHC{USQY
zH!J{&cwA|~a-I<c5BZQ!2l7gvIQwxxz&Cut+JZp_|6<$VR!MXKl<+HVAcbs0!j^C$
z>BVVjv==8D?S-s!`rhWn4=+TFN?sg{I%K7QOYq2PnS8IZ3kuy8_6VPb3^r0xsA4Yx
z9tyE@?3=u6g+<>6RIn{HobMnWKE<25Jp`U-AWM^JJeZYk(J|=RgylRF>NrqeI4oo#
z)`LBjFqG%_oA0`7yZs2gYdb%EwcKuVfH!0|1o&HkHQ^Q6Ek&pyHnyNtIqkryMxBt~
z+=9~ePi#*2wO{t_eeq9LK{`o5DPSkb6tKSg&1q+W+<^ronIIL`cwNpfx&)D0{=_8Z
zyj;gO^`RF!;ko!iFnmKQHGkR8YzgbZf|guzLEDmNY%E^zugf;Lp@UA|;NgueLTX8=
zgs~&YuF&S)`}X|UuElSi#(5z3b0%G47WNGfzJ#3AMfDJ2Z`UqZsv8jv9c+@8$*{cF
zX@fpy0ADxCDQ>`yWG;1hz#q&7PvPVy0i}R*lUpDx%*vMpLI?niy}_qqtIOH|D5DFM
z=-jvmBF9m;JowylXsYGD&^cevxN*8KSVth;X9A8P0#K%=BPoXl1n+jqCU~QLZ%0Zc
zg}uRx@b?7jGW&-9;GaJ9p@-~GXi6pGat$KE-Oriyh>m6yJ&uWi*O1ZcF~}*Hry!G_
z%{?-6wK3B?l+0K&4xU(H1LCg=M+$2715*(+m{X!2f1UV)!gjR}`-4<ag^++!z=e>a
z3yqG&&C*Tv5f_lWn{a|s7(^T_lDx@RJAtOoO%vzmGYDalw1RzDIbrz>@UG9f><9J{
zL9nq^CwL$b(U9wu*pY~b_PymPjW4q1LkYit;r5Nx^B+xKe&;_&?t@=-f8xt>e{G?5
z3&>x^;Y^qdgf*}m*|Ul^Q6&poyK*U<x^b#e3!b~*<S4oH;vWaeema(YzOb%!|Gkeo
zNTq;_{K$DEbRoJJQ{-;uVq4PVMed_&$z%WPXYye_?uusgd3PFob#6&FMyMj+_d47K
zN}q{@5Wy8N))+BUkQO04j$*?I2LZ3hI*F{9bh$&hu@@&(?Am}7vcS1v<!}9FQH{@v
zTEXk#a2ap0X)Z)nAkRgjL*5XD{BsM~OVB9PCql@3C}chyG4!JfMN`pnmw-~h?()gM
zmD3gZ6Y-{?abq5$Z(s9<9E}jPc;{tfTk;us%b#Dk7uMpB9B6D}THQQ{HviRJ4{SZd
zT^ro<L978x4S!cG|NDsTM#9BWh&<jAvZ5d5%a`Zn=A~uJ46>qauz;mfFL;4)BSg_6
z#Y|A3iIW^0wh?c}->C&3XmElt7K;6o)%EYwjM)2xFuxI;gu=*00!jhrA~l~l?Ux>K
zEI?2837#)l1NNSdM-**$zJzKJKg>)$?}<>$2`_vN>vRvM$_EzcRw!TTKFrI@fiQWg
zyaHPuNZJux0iz>e4eysFnf*xMFXtkNT|42v`|?(r-EX&l2JsS$g%xeV`s&J@8FE3y
zeC1%5=A4@_4T=Rt<`MI4-~^s@1?|Tb-Y}ow_5JT~!#oRc2S`|0!sqS&-sg{V#Z!$1
zk$_Ub1reh?5ksR(>|2Uci-4?Yh>Cs-)bo0w7Kgj?QpFKApP|M1B|_oL5Le5%+)nn7
zq0#W2LJPwMoWBc^nDz?jNJSp7;B}cpK_J%DVYG>yiJWtbV*+wr>wWkAF$WgXP`WQ!
zIv(!fZw*(DWtO>!iP&n~X9jwu;pK2M73;z~a=>#(;S}8Z*RAR9M2pzUsgp$pp5>v~
z>F6bEQo;p&vxvPb(!&x^3b;sLg^rIOdYnb|pC%r87+sW1Awmhy!_kHE#idG$1Yjmk
z?+8!GYkP|%hXfsc(fY{0z)gA&*4?pZQ=6I6P41Um@5u?7Fek{R5FG}Y$Tepa;UmCo
z2r~VD{@~=Jj{5iXy4>z49A7`iNk=0EYxideGb(w^O*SK>(yxTMDZL{9JIsQDZ68DM
z6>Q&G`S+Xy1unMmx`6>?@^VW(`Qh&GeBsC(s4J8J5>N^_fUx|6;>Cug9r|A=Ls%Ea
zIm8>D0&cHg{Cwe%rc(Z}b9*u8+j%A93;LFPHu<tRefg99rp7;k&6|Z=PeeFWZhTpR
z&l*nlW0Cy*LxWorl6C$s9v1cWALGlG<^4U|{i!d^K(}AC>lzPh9+YqwT+&<Z+H<Mn
z%sdcyg|)1?7DxeWhS-t|@A>0U7q1)CLC4GbMnr=6M*jUN3T_fm3fN6LIkt5!Ipiw5
zO{HrmUf`xX9ZGmJUM9oqZi$xSd}2LnIucL}TQ|k&qbjL#c5KH*;|sDZAT8mp$Zly-
z|Dh1F$W@CW6D6`H(aX8;04p8=RBScAmvS9Az}{WDt9eq>`={J`Ye6QJf~+5)f5#vu
z@HdNFW&<L&*XJ_MOdR|+59Y!w?5+VGM1J5%v-W24RrGt#`dR&e+;--rd=so((DU8T
z9hSIJpM{rzQose4Y4_!`kCf|TnPgFSl;92Tpnf2;ttj_AT>2k)Q3g_AqMUzs{bDBf
zP3K*XT)sZq5}9@ZPUE>HUt$`?t<fCGh^`zFVgDj^qjW`+?=yA(xA~}JF5x%d960F6
zj>oP~TzLv6j32XWORa;_Y$D4%7<@jHTA%AR(rTW7fUn&4K-+bSB-@w>ZHGL!*fnR>
z=>qU2EkdgsSc^`BxRo`Fv1?1p<tXb+Dc}M=@Qq7nM{o%29Uxg}M;1!>7|t7Hdc0f>
z9y_wOby%C6V%wl9XEuc<e6CPpz}6dU%KcA#3J&D2f@aRoRWP$Tcb+JptwcJRGzYA2
z^Xea7@?|ae-gi|Y+|IJEj>U_|3Ug5Iuio#~_i%dyl<)~mSgU-(yMXbrTto}D7K{nP
zAuMK2<ie+`T%#NU?p%iB)f66l3=_q`Tb>Z&cfs!e{P;{S@~S;I2`B}eo7`-%CiF14
z(&EYPZQErfG8f@TP*?mQDbD6#YfDL7-j0K<&;}@yZ|uCRHLq<$nL}%R;PL;~7)h6O
zP60!32M=V@^m`AhCSQww0u<XpF8%f6Mf{oOd;jol7nndz*z>FL`tWU>n@fb%(*eu&
zb)N)u64pX4Vk-a-y_Zj+(~!+;Lp~G|xOjfka9uCtlDc<6(0gu&O?Z>Gfe6P|ay_TB
zg^7g+p5%sjp!<eP#@kR<Re4H4Dd0Q>m_Tp8Z1#KbI<O5SNx&pOxxR>a-Adf&`2tB&
zw7cM>8`#>SN~Q!z@U?uv`Wb`|-Bq~KeE$;<TEveA#YcitMHZ~xL2(lB$mwkBaY?lM
z`Gg>jf)ac_6BZN3f0Aw;-Msw1n={+;x$k^)!}Y-bprirM&>>{?wjeIbw?!%hwJ|I9
z1kn8`JlTz1!JNs;X~JYjOx7V6ws&xsM(^O4Kh=pwLF*0k8LpdW1f_`y*C3DdJ;k}z
zCchiTpuU`zwdLb!3ChB#fL98*0FSwS$-&29pXuEo<@<%snN-5{PXfjJ2b&|0IIG<^
zKRI?$b!a5Cr?_Rz2jiD@E^hj6p$9PSfydr%z5nrZ@LWFw+WsSB5i*}MFT@M;JGP>!
z6T;n45}#_l_ujKx@BZT(IrC<N*tz7=uR`fwmDa@BM6&wFViMr#3Qyo+4R=17`MINU
zqR~9egF6y5^Zw+T=kFOuLB5@LKRyU(8g=kcbS^*xem`gzh5I&OHgQ@?H|OjdbI?sG
z_v*3$5(xNU3eaPXk!b(qoLL&v|At{T1|1sKX;0_8{#oEzL2-S-BkC7vZ3;TW_tkHu
z0txM9WW<s|C~=Ds*4?!o=Qn?%80@y*{}?Pr^H3<UUx2>n^Mwx0c?~#KYvwtY6JQa(
z02<3F6r0;}-+i~_NiPe%W6325xT62cY8a}RBE<d}wj2dmI<Xc2b&CT%Kh>~;MLziH
zeQ@%8*32+V6CfFiZp1_|&{{L+Vh6A5+lxtp^yH)!e9x}vx$d&BrB$k%yd<C$a311o
zzv$pr%@hxVM9u+ELu<m3$MRsq4Haj<>RE0@H-uVkgWGvC34IFe%(NF0aw-2}=i;Wb
ziZSli`=4&d+VE#BxNu=j$ZSzg?$TZy19wO<cnC=IS%bt<1X^9tdguM^0dYOSJ1)O?
zAuQkD0npLfNBc|?ecJ)!cTL<7rumj<=WYN*e*1Ex5^qC9x6?Bo!8Lrqj6mmfg*D$d
zNfW0BK_-p1@<_j&KpgCZhg7#e*z?~@&Ub28^}!_I_(KRrX7oXKUov~V(Ai3i{I~*)
zzL&GULAcZ=6@XS63A_&;A#32>DV2k;%H+0{F1!Z2@CA_9`;lU?7fPZJxpMn3(}iAw
z+X`<(c(3IDm=Mc*Y_-6Q0WTQ?7h%$P81)fUh2!@K*%+#$I?!kYvP|$GJp=MHgugNs
z-oaz~Ao!f_z`$N6w1utrb*yt}FGzLA2Tq$UxOO#U_6ZOin;;5L$P37-b$sGJ_^0*~
z;;^zTmH>1%U}EyE$^ZA@-FRjNG^j^7=(_B(xx~fesF^rY>wE>7JjwJirvd^jd6t-_
zhu(p5IM@@?4uJS7W<(a+H4WG8g;2V%O3;OLnKFZwTw2%<HHuHZAF`A8`(!=F*nww>
z1Uw{@U1&m+`7E$Rpb4C`fltDH!(cDhtvP7N@10K+oF?HrA%FW)Hu6*cX7VzQwD7=V
zq?7ax_$x9yR=B?oWAHR4Ad7yC@65)p-ukpZ;tI-10)D>SPPn7CzGundO@o^86D|fm
zj4xyIMVgmD$XbC(e_I|j`hmLzf7`GnaU0a+cKq7636sCUcz++)taiiD+b8~Z&z?N`
z$QgC}mD3vZ{k>B(;?of+I-P5Lx=a`Xf-~_a-NFaquNcK-v5*W#JH~_fc8gi3O<TVE
zZ4bD-n|Ccfq0t<Od>Ajgk782rZUkUG%DC~~+zIETfGzy^|16kxQ_CM8cm&r)QM__Y
z?VrA=Rf}r!Wj!Ov7`YbuC*yVgyc3-No$H}@1)#qh)5$H^fBlx609i&@Xo7b^2wl*Z
zT{<(nknOCi+1%S*s6Eeqd6N<1dj|#@nPE+`$Z7>&hv4@-3x9_|fzN`%KLo!WM@5O*
z^G@vfm5My<v(FwD>D>joGkwBKunxJcX~I`;-5dZ%fo4bmzdKceNTBnR$4^6|cp2q-
z1ZKl4p)_Tng5NK?b5|T$vZ%KXufYaqpt$P_>pJuzUfeHXM!p$l>1IodZf^N)=S~;A
zif`(^WYYK;Gml_`9|e2+Sj@1cMIx=byuPDk87p9cwD}akTbjnsX}-H_gHQOp!8$)M
zf1Iuv3rsT3!?<Db#ka8q@)pFu3oPQt0`7q{{{g1|6lLK*ANYU&*PZtOJP34OviJxg
zw0ni2^=Mo>#RELE8Ge_#CpZty&up-{AJFvhb|}?Z$u?p;;ktPZ-ksOrws{?j^-Z1F
z7OEFpnr^(YCka=D7u%Oz+0YPe!1>172}tR20+i>8cz4L}K$u6OE0Jkyk<VelZ`HR3
zd7)4C37-n~D$_#?8o&JO0BiQC=;}JJm)mF)U&p0K&gQx%1Fn)kawv+wV+l#l(zU$h
zK_(RXVDg89*?2eTKY|zY3PfwZ7_HY|ns)Q{zP$UX5(4im8P^oD^o5x1w?Q$TgD>mh
zL0IR)SV5L=<(nYgDHy$f>hkV_WL!!R_4~Q`<85^fyR~_kcN_<0c`PPa$HLM$3i!_Z
zmWi)pJd(8V@x9pvcjsF;<>)Oa=Pi)kTAk>bNq>6iRmc#{r70a3Um(|f|AgM;5=1-G
zja?Yi+<?9B<9fqZ<lR2P15a>WzX5RjplsKe5o=A$O*d8bq|{@)S<SGV!+WE9j-vYL
znV5Kd7EgM17EoqIXy8$}!_UmD_Ir=-(>lDP-aYQCw-zpv+;8VK00;T0;zt7AmmhTw
zX3}>7Hfs$FQJW8A+uS*R=(xe)&A66RFV>RiVQ#U%HXXY2Ilip;0UK^SRZ?5$rH%8*
zqRX(>cM<q%b>j^g$zIS*L%y^lZg385QM|maX}P!S$E8$Q4!lT>$-6%Co+*|UJqSyR
zGw^-GC$EPx!@G~~q!4?B%({J=rFBl8vaOTf{9Me5R;lVcF1p|XcxZl){zl}_X%p5*
z!8aWch!4VK!f2?&9M|rdZ-v!+i$>b7n|^Z>6tyY|3G{sDGpAt<`ciyHK8CucM1A5D
z%Tg~Q(T1Gd<+Z<;hkJiVeDJ2mtN&b<No!&})%d9tu#czKdHGTQhIin90aSMa#|bMB
zBfo5}N_Y^4$!~@6|F`!oa8?vo-c{9oA2Y*iW|#p{RD8c*_(BB+H^yjU@|gsDM5CyS
znlG-2Yy9ng`F7*(W|MC<yGe}i2NF<JM$l+9F+Lau5EOZ%JQYL%VSoYVabMkC-@gWC
z=03Xb?Y_6~y?yVk`u*m2yQ@x}I`yCKK2@hqol+hcHg&^iR&B`qg9wOoecSwl#DM5#
zIImod+ys*?^wJkRrizcsVYy*OF|Z;xH8z*Fc|SJ{oY!u<?wWfrH24i_rhH%Uzgfyv
zMa;~cZQJ8?2~`IO6Dt&P&$g~V!{P6#p(BU=#jVho;9Hi7KQYC(ug!NQc1YcTf5}zo
zq!<((puw&-{2rT2r>U>|FI0_O0Uieap3q{-|Ec}yt3~lM$)z8~s+;nH6iW~<wtHy3
z_Zi6c4LkC25}Ql8Ho*?<AsTq*lBcMu{JLT4`t1(tNzy|I?7XgK7^mv3c#x+c>nQwo
z@t9An;hyahNV_nTo7_07?M;{Xl`WgLYrj3hFpOU_rDil#uW+n0PgqhRiXyhi_{|Bd
za+Q~;QnLn*#8VrlJ}loCE11&cx8MJhAykP@r2g))@XWq34Qa+A<_o<`)Uf#8#V^0@
z!e)to9B|ib&t+sF2~tO>5|e`nwB2+D%$)oY+@+YD(_o|%5?zd72AQ2kIC-9_{GaQe
z{ZyLG(g13=h~H1x?sbE%Ft~aZwiMlNx1K#Zc*^qN6H1`!k^1L1m9+DYz5Y6*>M@aj
z|EekYg_Ve7`x$K~{_1vcImU_;Xy2r&DrQa?Gi6NOu*3gXr{a5H)U;{gjMS1KBLZSp
z+;JfM6Z{|k6~vBra$wfG5VK$5RA*(rZh`N{orCUreO*RjDwx2|!(UPGh%H7@>JxE-
zCuZ4kn3JPxpoB+T)c+hC-tTCf_UT_O%4K$r2<)0ve-pN1{T6JT44$l7)fo3Lg*&eJ
z*tA2r{8pj9hw7R?+iz8&lJeUoe)~kMNk4@?kB#p)wwZlFde&`8nh0O)b)qt15>1z0
zVv7E=`e$F11J!gWWm(qsU$>0VnD!v5UvHa*5?jvvohO4{EC-}ZQ561WKs5MEdQWpG
z9bu)GNd<i9KM#dV+iyPo3Y`;Savq-J1JOlz1Ro6Jq&Pm<Ex`Yv@tU6xpRuYdE?-{e
z0fDyf4m}KkQXT{A@8rf_c?U7Izm~*fA6}f_Db)YFN>#*s%4py2y~F<rUe-ScHvNuP
z@jy{^E%gPCM3$8Yw*k-SzhMq+w#4%}LPXxwFyn=J@hs(aeh@hDzzt&&C*)y#2m8F~
z&050gRqJT5<8YJ>G%GjD!@IIYXY|K|fArcLX62QfSMds1oZ(y7J@`gEc=zBWF^wDA
ziFhC!u1%#rLICy)oG_<gukZ+zuEXgqw+?&m)Av&;Wlla3*fpu)7DO0(6m6I*i8PHv
zxvXS7AiP8s3$lH9Zhmi2?UBQpzY-2DWuNv>yu2>TVn5{Ee-<&+n=*Eosc@bOU|W?X
zh9Rs6FU3~7zNk&WbdHbtZmpd*O*->r1lT<SVU|9yKk^W~)bB9mO7+bw*nz<mZ!|4W
zlSygpO%gb8I0jqO35(4fFnP&WX~-+4sA3iH?kS^BB08T87yX;@Q1m@eTsP(u?)>;J
zizj9YHM-wwA3}e5{2w;KG%Zsx5ZG}|{dtHcGzV=T;h=q(m;+sKC=mmwv<V6$3hTxk
z#B+Xf=$vgWc&U(8Ua4*3gz*L=Q_#2Xp=V++qY<MT{to?$)2hc*@wXyi!#K-w!uJ@8
ze;a~HoCqB689M!WF<_FZNC@nE;HEpVj~IKZ<JNa^NP`kQ0H;xYv}3U-Sxp0bjzzud
zLv`URL*IslV|vYC@+zIQsAA326{Uc8-FnV(h=lVkoIEoL-^ejm50atBTCthm@`GVj
zFDV}(UinQ!r>~LTfJwj<<96qDjl+m04n*J=+GrmNRQ-^0Rjm6uN_e3V{DRNmO4YPs
zFKznNQDqMi0|2`F+NRM6RrxJE$8bm_CnSr9_3)iss|jJobaZlA%Dw8JhtQJaQ}PdG
z0J8<0`F}j@x#w0|(a8M5A<+Kd%}tCLFQdhJOi=dHVEYa{+ow8~fFovhVBZexIX>Q^
zu85)ivO?I@Ap7zoI(bXjg03vH%T@skuh_lY&O4bK212ya^YG#QJN%>L*JQ2#p&XxT
zJ-QkMXIm4cEs@GMEj-03+%>Ub06z3$9V+eMigINhwia6|3tQ3~!G*B8o(Sc%7`kjR
z9*D(Clz!Oo+WHP#^@~9C)s0`Tq0!jon2CN9bNDMEf(>5?WjWD<b{ujz^HnXv;P;}G
zu5q#rVXVdtn>q8dlrou7Gz9iNc;hGxvR<IUeOMvjjt8ScaH@pn5np%acZq6mfcA1!
zS%+<Z%dvE~T*seA)D^XpSCs8iL|G_cK5?SbK4|mF2qB2`OvyQjj58V?I}86zE9VdK
zY@|(i;M$E?tP?~d(-Ni@O68l1iDz=QUxh;<g3;F@(4{6q;n7iB2v;YVl-bV0?6I91
zH9=IUN~&bweiJ@R8}Wg`FW<-++l1KSn;IJjZ_RfEtNjzl)I|+;3{LDB#ToWYa|)9@
zABnZ@Bk;bAL|ozZFcmi?$(-?)^1wUXP?H_iQtF|NDECz@rIf{6pLSwD>iCA|n%AdP
zkQqfsVBe$PtW^vqd{^vEUg9NY+5+m|e@T3n#q|NIcESwJhgp<w#t5KCDT8J?QS@as
zH?G}bL0eMUcqfXJzWC{wDt}lxjK#23L{$e8-53OLt07PjABQ00_{C8u_?QI|?I?(?
z-a)JnRbzX}5YW})BZl9e7_+!GHWNKG8u#w@2OmkU#fF2c-F<Rc+aNbJ{Bia3#X(iJ
zY6pQYCLU4cCy{sY9K|aFgLAco5!FmF+hN;JLs=`)tJ2Q0`-%Im@TiLz&ipG}Y{&IB
zJkDQIQnl8gbTCe~!h`?;O2sb`zleSS4>6o`T!qtbf`}Pig`+1C<`M4<@-Tqx8317Y
zbqo6Sb&wx}KfGgUlSWuTZ1)+Jmdd1%>cJLHlqQp+q%0bPXY@i$jhfJpJ*vW+>SnyO
zCW%PidI*8g@2@&iQ;1N)!u#1)!m`C0Qngo@orS+Z`b4#NG5h_y^hEKcd<j)LV4bAG
zn$!V{O$YWIcj7%j9B|r!_Xl|>GjNYQ#-r=FkBj3-aUh7%?q@i9aZ;O@TopaJtFTPE
zBTT`2wv89c@jOC}b8+<7IWPkriF+tc_YyTii?JWk&ibCSM!E4J*R^v|iDjyYe%;;p
z$2G!vK_-tQaBxl75V32;v#3EywYs8w!Hjd?yOjGKd{C^x;w<<cJ}Oxc3b5|_4kszw
zsH_{LJ{<nRF0&@ED#3ZuM)jAlK%Hns5tpw+3BL}U{=M<V7njH7%S(?U(D}%fXK}79
zf_P$Jmnm(hfp!(&`#$H^G!<f0>Dmt3qtehM7AY`Slh8L4GcE7hpxlM7Ne&p&hP;hX
z+#g}k`yov2i-JGc`I#k`JQd_c0rx1~9b+fpXu0oW_INd9w=ylb)-9a;vYP855i8d9
z#0`BOF)|+SH48J}Yg}(!GRzwSJFjUx6I;*b!sNUhrsu4cT(W?aSAE7g6EjX1Qc#?R
zm4!k))HGFA;G`t69AhhCmhd@celaM?;@iMriK?nc-AprWERB<IH_ZO?>$ZfFM4mw4
zz{6MFhKJ&*48n9%|ADR8YGlu3&9u3cg(hU+nOMxaKBUw*KQYtljxVN4*zUw^BNXxL
zIIsNWstvn7panPxCySTNs-uL6qtpfj^qVna|3^?AXHA)>61^Lyte&IW+;70pTHw&t
zWWjg?QKIJ}o^yvqxy<$`0=p&-yM}Pzi*AVr1-6+Q)uY_v2j%$|j4zjG&yRKzRwaBP
z?7!>G{}ITA)yH`<RBTNOT!4z!?QuS>arPUNoimp;ij6?WAFi5-|Lk=Za&V${et<BQ
zwiF)B!<#VsJpz_PJG1KECNZP!7>pZ#Z$`Rb%Cz?gjX4`!Q~eD>Bct&y&%rs#FH}wH
zSdgxbT*_E3m3!+7C#c<1#{KVseq$???_<a>yz&*oAU;~JOs@YjwTUUpLhwzFV%4ov
z@_ng%T}Qki3>%wwz64Ey^<Z}rHP}{nI347I^)0bBoI~Qme_^vT2D;*mK8~7c7ZRqm
z#^oela`J`BB0ovYly{|q0QZxBfULVs<yE8e>?&W>-%D6`z@hTRI-s!vb!{nMAtPqW
zL&dL?L`e+JiDWoly&vI>r}?252Y%T7(tz&?%dQhXc4%{ExaaoqWA@bUT!O|P#$SV-
zM)O8P)${SS+LSN;IECe6gwET6ZB2b1n;3BXX!x|%tMk>S<ePs4G*!I~tt1buvTB%8
zpW_!kT|+os$a*H3u3|!&u_NI&(~Y5@jYijHBC{-~k@19Tb4|*z<i>BQYkp_HNtw(l
z0R#r!{`ziw^M0BlDvd3mYDZH_zIg`LeM4KZ?mN=F*fQ7PtlR$Nj>9>N7^hfrGb^C{
zQ|uBrBdO>-^!eq;NduO2zc|R&>^`Z!<!#mqSd0eRZXf$^2=gaqg=br~P}l;-lt36w
z8u@A#TPTiWKHpIoTKyFtsh5XNU-id)wI%r$27$v~+_D8@<^Pv!YcN@@hNH@<x$=px
zj~Q_83K(95@jUHZ8%9jjiP)hvrD2f?h#UMK%HrNd{@AeDZ@gnlFEdL9frCd4QY?)e
zN-R2;1`b@(dz<l>MHXX#X4@~JpCA7a&I6<R?=xI6{xOy$)eKaL#R#{F4AXZG@Ao;F
z6~B)e@zeZILv6>xPHZ6YE-MAR@6JnVcW)o}XN(H|)k@gJW(s$QMTpZdQ(chH7BR~X
zQpI;E_DxDtT0Vg;BySr29hwuKk=ZBi>(0Kic};u?vU*1oxN)O4;oim#oElFfh9)dq
zStrgSml76U3I;1+HTu-3irtqrGg;A72xtns1MdE9MD1>&{*DQDg~K)=L_~{@wM!pQ
zM_D)!f1l|Z@uzDt(Ik~)u&5oMl$U%fl=H3OP-J=ctU)#%!ID|@-#dP}_ue-4jF`@r
zLJDz~r#BM~jE1didp?Cj|Nl62`pRwjY(w%c90HBaZJpS=`0K1&#GntSCJE;wR}(jg
zFK33-I2JtOW}<dE>X%^K@jScKrgT(4k0@Pd`joC)hR^t$D`tMvc4THj5vaZM)h#5@
z{xDTTO|~0mWGPp=?9Y@F81p5VHQTqeQdnlG?r2vGqVO!kYjRXVv(`7qQiCrDy*TKH
z)+Cur94X+Pw_kFm!RTBlpHZ1v7mos)uxqFi6zg7?tmY-@8fQ71Eb}6nWSTVsdbRc>
zD(`p70A~fSKp+a&m9LNkXGiC*fH|*$>FhJ7{pz9T+BnQ~a<VAld-2JTS?|rys(o3>
z%LvT2D;%D`&4hJ`*=05keu!Y&opEbhoGe=@jU=z*#ub}iG7_=v+;xZp-c3HO`-Oq`
znsdyYYg+;Dxqa*@3TJPFWnE^r9-yGgX#Ce&^CJnLIc(~Z>^^Au5lC`%0fFH&cH$go
zGR+YyPBv1bY7|C#xhqPOhOuAP=|{nYix`RWuS^RvXX4u#9RpvF#hFRXDX@0--%!H&
z8Vs8^Ds%IWz`!XBvBpZUWi3<U`?=B$cZSUS!_7Fsllgr(Y@KLUW|9}hzF!>hz#Xm8
zB8bxx1e$ZKw9!81d*s3JO9LN-)fHWImTW2Dw!gpR1cbp8TQwS-u`ez)WmOje9`TU4
znQ(|W{gs;xB`@zFK$#)7(b-DSC|eO7RqRO=e|bsp;taaN*wC{uLHD(+Fdjn$J7ZN}
zj@=7Rb~G6E&)X%Jbftv=24>STNF0Vj{1i&`=nQn`#T>g6SV%0P*h`2%jj^6HM@UY(
zu`Zp{#1hK)-q1?}es536eyWTLxb2pU2g37LM64O!U#(=Wd<c;{7-R1PE=oLm=#xvc
z-=g87Ad;2)2sAWr7hB90W#Wt(<xa}0@&3@pOtiU_!wI_g4AorN;1w`(`@j(Hr-pVI
zgm5;Krt~4&+#JnGcv;!A2xw~Hc|2Spa{>7tn2tAL?e-k=^31uM#W2G*Z0Q~Z_E-qV
zs|lI8CQ@(~U5**HeI=h*z3&e1>Rs@Vp4^vV<07L1W(s=(d`_^z%bO$=C%q_vPuvm?
zLGK-pa@=ZL@+lAkoV=K!r2{ZK?}uma5gF)QC>Dp5<enJOg$b`SH%`2Pljt7sh}h&4
z_C<`xZPH4z5=5YOatqEXBrl|j&Od_kjS!}vmRe>n2d3T4jD=H=uQacY@Rt$sj5{*}
zQ)521%Ux{FGOOH2yIvT4uHBkx6>!__<EL;+ud~a{t8|>I7gD0Mx**@Cx@SIG=mH<f
z!ovvQJm6-CXJxVtBXS<r|CIs_O8F2spgOol%~$5i3!a~+!ZUhV8hrYiz?bl?+E?*`
zmZ_`}pqxD#*TJ=zQ7*#qg6HCx$q{k+PA{=&915e79p8#5W05wPo*f0jffWmJ(j#s=
zPX-y)XLrmRU~9S7QUUM2ZQRM&mQgmdJ+1^Uz=$d*(=qB#cls1b`qD$7Vb13}Fpd@{
zB_*d2;v9cMxN=r<K_2f_cp<wIW1Jh2!=Z?CHwSZV3^8IE2cj6k%*i7zlC?(=sQLQ^
zOR1tgLw(&V5e@5PD8n(iiN<k8bZ9fD+R1kKfbLyHuNrNKHjZWsQ^H3>X}jX&7{^yk
z!S^X7_Vf5Wt7}UIOc;L%mE3Uvv1+}b`80szg~b8s`?GBpQ2}RVTUXXAEd=0H@m7+=
zAz}p9BCt-@>sd+sY~RAUa+hu5G=vDQaWo}b#`WzqEaHlcvE0ybj2mM5nkOxw#DwpN
zOqGBD4PkJqpP6%4DgOxj##NYwZ8=im`J2yJ?CXx&rFdp(2=hyKtV@Y)^A^e?qY$(D
z?(i)2vc$5~J5vR`d&=1V1ylOhQprm@2NAN4B1!<}LlSe^6U-boH%$F#z0-P<zC;nA
zJV60-?q7u?MGLNc1$yuxZFdbtBRU=qS!T0)$4yY8pU;hG7^b`rQYv#kcgV~I?#Z3f
zfXwg|0-P_;Q67WS7@}l#j>f*XdRx3G`GT{sSod|Zw;1Eu$Y<E5bd;?&oDgySZ75&7
zU*ymhDEb(GHvJ?vMFEQ=1tH=8IBB=EXw|a`3oM15z`}FSyf+wW{nflg=9VJ@jbxip
zQU}*BIcv>Flq=h~VVi_)S{sDS2p<QvzQYEo4H7<r8qo`2l3ogPUf)L;!AYd~*#>&$
zLV`iqvK;!d4~lCU9InQL2R^m7QVdR4yWsika2JBnN{DXTNy?Tac<@EsbE?F|j5l8X
zW1}_ded&0lDBuBp{Vs4Am9A3Bm%%9uXQXb1nQ^<_z98<kYe&}EZBx<}4*}XtbhQ2+
zLNTwSyz0W@NrIM{fkCx<3C!w<5r)%BPaj=kj5RnTdN$=60veF_e4M@M5@X4x1Q4i+
z21UFD$NmNIt65LgE(FeqI#P<1=`C1Nak}oi7O~0gdR5y+UB@=7`S`oktO6#p_jb&(
zj-QdBfW_<=dztPrV^n^5>=dfj)Do(;U>{porv7P!_Kof{(aQ4jK!7pdD~N+p2ZiOj
zqlP1d4Bb&wAA7v@Gbq>R+oQEl#kqbS)gwpa$ic1lYe~vdM}XeGR4nVQancVgVJw8n
z`AEVY9!p3Om5oN7E|G@e*d=^`?08k%?FN)@Y%ylP<I+?Z=p#N$hpYqzJRqRo3`WD#
zFe|sBDww02h*I4}s4@?ovoiJVZ=AYhbGdb1_9s08TvM@zncqMg=_NQ=iZ!vL?a(=;
z;XE=9ymK)fS~SU}uY?6M<(+MNl7euNu~wpVUkuZ=P>Qw_9H+KXC2|5P#G86E*5^cr
zpMtCJk%3w7B1-j(Gd51hon5oZVP<rF3K*k&nA6F<W<+J5NAZzm!SPVU;hq_>@REy5
zz-Bf6+xU^Gk`U0SZ;KcDi;q;DljRZzGBeJ@BSA0B(Jm&s!9`}0RjDA*bDEH~uuz`x
zJ~rOR{@%|B36Ha;v@D4+HQJ7Cd3BZ)CI$F*ufrO%9YtwU+cXQhKtt3m&F;xgennSr
zFyqvxfcM-sZUS_0&f&Ps8Y!JTTGulpX4N&61r}f}3yvj3$3Jj&>GVNL>J|bDqo2FQ
zD_1^@@-K4bVOpQ_1{}8VCCbE_w@V2p-^f&P2*8?RwoKQBwOVWy3aw#kXDwx27ZIkN
zlm$3C!b~nfd>p5dwKS~zissZT+BC0%HDAiKvF3$Vx%4Kpl0F4Y7@cfJP?>o&faqD5
zP*yXVE7jX6Ro|;j94foljm;l-VyK&74JjTXh9j?iM{;tUv&uyfdv_8BuAUjVxYHxa
z#SQ{hVZVs9X7tcfegH}edty7I*w;H2!7_y6#9St;B$V$5P~2HKQ0WV??JgUIZz?FZ
z1kTH_=4)CHnWsi)`BI$5dK9oYZ~+$9?^)HX<a{`#jpQ`w;_RT3(JZ~b!%K6k7o>)B
z)1bV{!K35x?Of<a8w#b{+~6a5lokST#DWhEe+R){S_p+-D-AVazin-9z2KdMVT)T<
z%9ngjS?v7W+V86khh}3QoWJm%WS0y|ewbXdM*&lxav2I7ErusED(8EE)MLXa=RV55
zSs=hkeoVg>%~@yn9Hx-pfuIq-5aEH~tdd(Y3=ZctnJN(hs&rHnCDa6mqTC0J2($q&
z<F7E|%tndgIQU3AX1{LeHO#=pns4^Xx93^lbYk)|(W8KI*6$U`1@gWh0W1Pb1>6ru
zMP0#WCe}y2kZQ2TwlU9wD4d6|A5&}1r5&B9S%-VgrCnLK_z2*1`ixsT?MMm7!<9Hk
z7`{0AadzmFX`p!L!}HmVnC(#v0+ztz*{;J?vM6!){IJN&{xjk-$iW#f;opeQl-G#}
zFsXnO0jZ!n*quBw0~T)h+?ANn8UB3=0&2iR728Czbj%uF+kG&CScZmVarq$tZ+)w+
zoVGhlnbFFW$SF>}oqHR=d$IusLS!>*?q$*Yt0T_z3T!pHus0*;Uv4DmTGoC?)6rmB
zfOAcA7QJW;bc5#40j6lZk`3EIfOEq`>zTTHo}8IFZ^W=iIJyj4zatZAQhA6z*%Oo@
zDO=uF1_Cq(1<c{G4C@PxU;xN{+2Dn(PTMI94}ua7W>Y`Djxal87VU;2?_&#V^W}Xv
zxL~>)Cs@RDEW>L#11`gVs?9E82J9JgDh2}u9Vp|pqw@$4x8Y;9#U@;WOwN8CfK_g!
zSSHEN1kHIDXR(#=c82oJI_<`~Zv@JFc-9;aRTSRNFgxcWi<JnM@ad$cw~$eTo!NUM
zf6Wn~2I({}k-6oL09CwW{lI?8k-+;vVKFz3p+>kCL3%#2*{)Q;Hi0TfvaC$dl<Y!u
zt1obneabc+yBNV&qn*$&{)|y}>!hLk0N&0H_xYybmRRJE`F*GKW-){DQ+sdZuLS~C
zp#p#OEaZ^c1xEl%&r8b>_|A~e3JwC{gE{X$2YZF**=XA{76TiyN)ne5z_CD1P0*Vu
z4a3oiXTMh$%Op6K@GB_a<8$o}X7FpV{(L(4^!P{R%FnDmXY`ofd{|e>(B2#QYl;A`
ztN%dGfK8z(|4gberyrC&gQ6E}Fvt)wYdq)CMKJNDF*8+G@Q45a5eZ2|K~z#X1J3gk
z;AQqWh(67;ktjKqHkL6ZJOs)&`=u2qU(BAjU=RDriQKcg+YXOsgni}JS=F{(sehJ^
zmx|=LQQCH^rApXE?>`jXrBr($<<1%b=F_}11IBj-+xd%_aNOV+%A<P_A#A0o7?eu|
zYzj<eWZInp!?U;r%J(cN;jHx!UW9g{gujHNl$f#RnnILDIx&kLmMf!7^?_0U3>Yuc
zkW3XSyxgf5XobU7UftJX1Wp~^NidPs7h4)c<vNlY-KP_-w?LL1@$yhZD&TlLD(xCx
zzG*E+h%C1RiVo-UMw@_5_GVtnKjkch<qlV7zKGhTEyGM)oU)YNUDaC?4i|<oU^D}V
zpmcXiHxiB#7~N9RqaEEyPst(OAkrbDy9K01hk}Sdq*LnS@A%H$xjQ%KdCv2mHy#Em
zk9K}TxmCd?NQHo`wCh^^9{YPTZHz2L409Wkua0Sh(##J^WunsC!CS*6*efP=c`Bcd
zb|OY~gP)hjZncUAaQBgo($u+u#*sRxf_iVpdfS|)H}FH#eLgtjgoHr7V1l$Wqr3l=
z>)ie+`(%VhC}hm6G4XJ9s(;B~gF|JX$ETcf639v8j5u#jPps3^!GNdlda1vQVy+#=
z$*Fkuk=ywSOa70(th-ks1N<y9F{he)9BJGiw93>~u<$*7Y)xjvocv{57Wbq4TQj)Z
zBw~A$_+(@M(k>C@Wwu5O@T&9FA0iUXZugIc{(W&)!cJ(dPP*VWq=&>@oAK0Q8iqST
zKv3L8Ctqil%6+N%#{1c}cwQCfQx8kY!OZ|uN;_x1ESNUPnjMYQ2^O$XBr2~+P5srk
z<R=;<`ZId*nGe&G8{H*gXr<b<n0B$6f-12yQKmH$=(2@<-3Gj_&An-<oDzbaq2B{V
z>Cf<`!RK7=c9i>vwIB|>gDwq1HSmj1ntFMCOkt@Xo=ilnt(u$)ou%i)ibvNhj|9l~
zNDj&6S1r1-&_e5nMt#;b0?I*46MFvX5dph8ezi*alU|}PLU~&Z&Z?IX#kI6J4z_4<
zJ7`+p(``L$4kbO^cX-;T9>x8FB>kBRaJDZucH$S8L*r2<XWX<f>ZsZI@5k5CZ9~F(
z8#F%?y|^JB&$dY-zd&KU31hU_sPdz(Gu~Om=ki5TJA`^cWM30m7TT28X;i-u)Puh{
zPTTITur(=eyvV3I5H;BgTUI8Qh-}_0h=B=(xe0RdzoOy=dc2ZGp?Z`4c8Z^f<6;}j
zH9@%0!o>~b1>4@WzSPN_yM2n&E{6xswOgz^T)cqZB<X>@Hzs^5)^+Wl>W=VGIm0%1
z=DQ8sN0Z5XF^*cBW^<*_Cwk=f{yt|(TuM4#HY-mF5f{_6;QY#DFcZ33QX7lAE^Z^r
z#cxPutG}Z2Wk_qK^oTRPvIzW}z*KouSdCGqf$}Eesn-Gt+==DpLs~}CN%c|QdN1mc
z!D`x)jxHE0KM#IQPGtJP=$%ej+EaE+_!qtJM+?YVl8a@mF}xR$ME0hs+FZj&Es4^m
z4K6s@46;~`V*cKHemPBXceS78SEbX8Skpp51&Vp4Ks?Z4^thy%`x#Gw??c=h*F{?*
zIh?B8r3Rk=0`P*XzT9v~za_#>udH2U=9K|2)cZ^kQ`EPxRXat8;>lo&6w}4s2%<gG
zlzXKF*l=9GOSASV;wRK4Ckm#oYJ8`s%;6x52K>Ch@&+lyrFv_{L2=E0JSd9~)Vkaw
zc;qNF);4#f$lR?jHshMA>;0q8GLHtji&~ZRtr{&jvwZFeq$MBKE?9|WLD2Z%5h6(v
zE3A9mwV&Db@@WrTd1Y$Ywm5e=SDOALKu!y>+vCg9_pL$Hq|7~gu5hKoT!QwsGMjNc
zBhMR$VH+Zzl_?%T>?Lu!lBKp&l+6vvyytZAC6>Cat=^c$7`ek+&dbx@gK6L_+Yq`@
zsL6PL9w8<R@Ee+Y_kjz!nxGbQI)~5MWo~Ko1fYL>zz7MKocSh3coM4Q5so38`QTde
zg1q2%P(o4EmG9I0W=r`eKM}^MCl1ZwCwD+n3yuOe__rXkh<cBn^SJR&Swh5IQ-+}G
z_o-oQ!@W!?%-LQ#X~cf0Jb>ief0&;;_pS27FOAG8!}r?bIlvFjzt@?(&}E1Rtf^*m
z-?f^-Ma6DX%WtRk)gTRtH-(7PhJ}?C|KD}KI$bPvzp^oY0W03O8qoMsG(29LCU8-x
zU$|O9(w}Gr8827G2}5YKnB~-MAYceNat5ilX7a0lM_w$?6rsoUym;(^0Ta_-0-s%+
zClBslqJl1QH5i}Q%~yQ4B{e!;OQtb*OP(Uf1QWI%98-aho!o^L+19<s@;?uRjLVg)
z6)2jBKO(2+dzJC}W5q|Y{OWzjJ%46ZoJ_?2H~~NVY}k@}z`LTAtrK=)-F>M;)=1a4
z%6?B7z9w|ja9Au4Q^6*sze3tYjJ20DVB{s`dtuclf*)kXUWSRWu)o;p7Ubaxp_eCM
zZ|%z-;mVTTpD)d)>VD7z)G{f+7Xz^zc5h9SHvbau1U_qe`b8m6OILgD;YBLHBDT1U
zB%B|s>UT43YwZKRoL!yc2rKxN$|DvLV={7XH^AZgm(z#V!CmKW1E!lINf90i0<z4L
zZGGkwKrq&bl!C!M{RyN-cNS%K&hsVH3-^ilRX|+c1`_pqliRaY{ki$&wtiFC{uv}X
zY~R%sIIu~KKlPwtT4V!x_RsM}XSc_<Q_22pqw%Y$Xq=91(<1rRrD%|oyP2V<RF)XS
zFo9S)79;p2EYzOuZ9Kp|sKdauC}i%llH{%}$=2Jm`*^ybwA1c|`7Y53%S$OSsjf|K
zm=Wrjd%uxaUoerJV<@;LmM<$!6cFx<Pam%38Lp9=xTzD3<eVL98xW{<l_)1Ow0cT`
zua7Ewg3DV$oc@MtLSDRDw=4@(EMB!WdJRu=y0rUhuBuQic7Qehjlc0*7M=LnM@f<T
z{;(7)((F#tsQPp$^9L$yp~^zST;klB+zdhQH~6LBn<exWm0M4O`lNoPSlga$%n&$#
z4tvhvwl{!I{(%GIs$826e$e5{O~0F<3xA~1&<R#}(n<I`-;l8V6S|t-IVW~OZOOY=
z!00FFR7MN+$U)jtb#Y!o_2W{1Rn2$Z7H5167ibC1>WRz(Llrnm9^_Em)=Y2jDEUGM
zr{?;`%u5y|Is+GH565u0;(?)+V$wkuLuA&bH{I*0R^#O&hF{&OF*6*QLme16*T+mS
z$UJb9jGDG<$5?f&JF3xu1F*HdRiQLIv-mk|>2R(0P)*E%xkbOH$fkn*!uB>+bjwSd
zj4arBJ<;o+VBhSX+!0}ahXbNLs4gX{$qy&9kYMPH!C%hpw=zk`(D=!1XUHTV!$6bZ
zx!Y96IJaS0&IYY0<)Fp;cRt5~)~^90w|es>iz0p%v)r2>yDgsRQQz=W?*xl9)4f?M
z3+JkC4gua&t^~1lk11kJks~>CDGQ^_I%i$vsJ9-)#}eKrXLXsGuP?RbOIb@}!WN%T
zpf9vs->rRw`neKVB*}KYENp#m|Cn!hy}B8;sO7f<tqQ|UT^@e?txVPq+h@6H{m36e
ziG5rbc=>xN*qAj}62Ow(P0VGmY`}Rpl2Zgi>%V(Ej5jXcOGT?_vd1ZCgFv;`k&W<h
zXeNXOvDiN=8NfUA_2m~w8C3TR^4Q?_mQZFt`|*k}+~R-bzq+Hr<96m52^8zEQ1A{J
z_vr&$<~?C1#+-(toREGId@+wZ*N(<^kx{qC4ub@4BZh}etrmG+v1|1PLY0DF%Xj0O
z@}7^LVmB#F3Y?pRi7Ky$4ZHtr3zKcsCVpImoJp7T$0fErsk4?ii(A;cYF$m%K(%T}
zx%EKRv|ciV!p)$UN`!}!z+{B|8og55IOAf>529g+&Cxk}EZo~dzNCKot7=Er3k)AC
zI-dEzOd5az#&@E8g8IGtIYUtYH9(c6Z&p%AX)Tz8JSpk9#Hjy(zzU&pSUC0QYZg)+
zuIxfNg}Nkm{~28Z*=06%=l+~#=^$0*eOOTARgsG%X~wmyrLFn$(Q<%V@xLFJX6icJ
zARxPMHG7*EczduQ{i_a}7k^D;DYLxxWH!Ac(gF%+#BcaWLxr{LG?jKtSvZoX^6lAf
za{>&9F1Vp4c*>&b`lq8O>xXJX3czhhpI^1qafr-2n`}M9Lz4r<_rFniYIeP9Q7gBD
z)0vqM6cM`>1C5RrhLsY)i>YV&PYbqoH)RkiaXxZUU;nLd4lEX<N3ze~zd3ZXotmZz
zh7hanFe_mHVXgpOL+>c7TUPss7?G|;5#)|H2XPxQH1F^%7Zg?Y#Hu4qa^|ERZ!{AB
z96F*J`^?O$Xnhn?GS~r#*N!D7gEddMZ8!p#fSnY_)wN`r@gXTZGs&9M$4T?DvY`e&
zyV7JLZ@_i^#RNFO^c?2&TD1pR2%}+t&Ur0(XfhgbuOZY%)SWct*7Cex;T*eoSpv#!
zl^qKeHUjsY4o9XoJrv?W92^DmBo;`(p<rtA_8yWQo}TdlJX&PXN}4Ao@_Q%?91#?9
zT(cK|m|nvi>Hp_4n5x*q3x1`H)wo4ax6AB-4`x91vHr?eYZaE$vX%IW>ga1d5qSAl
zz#=n7n+ggfjKRu(w8Jf0kAxlvJ2tgglVpnneU%)%4DQPR97cD&;27n{KyZ0KtxNB?
zM3OwagBT{3J?b8`b5@-TMR=mFi`)k~mYA^+?yGigabY(g1?<^1qkIn%Dw7AwRitNk
zG;4C&R_^0I%?u3GtHgaQPeWZ8-#rHM<SnEaL)RZCNWcn>CMIyX&u5|<TysQE%c)H>
z-5TKK{>Xn~_6eESUKSP@z2a>)3(k2aKU87L`9mquTnc-JC#`|z{Pjd&yAkO%5BNwJ
z@JOPwTFINN64LB|9FAcaT^7a_Y5qBRm^R}IwCP+(ojpijp|m!h{>kCHJa<sgrV7Lh
zpUkpDy*7aNCtj(7d~M`#A31Ys69@u(B#=YJYke$J+YkfsfZJNFU7bHGl{&$3I6B;{
z=w{z;M^^?&%Q2r44_P9QC*_k6dI!4ugBJH0^njJ_z!P1<T}m3Z%>Hs4-hj<A9Q8!l
zhDhQKIV|&=E2_~)ZI6kSEdzAMLY})L%>aWH#zqgnD6&U?{7o9wb^9EzqW0SVY|4j%
zYs5XPgKHD<N^E2G3GMdmu<e;J_Rm{N>oPgg9YCw5x=C1=4$6))O(MCGna4M>Sxjy=
z0vV@C5q|R0`o^HR3KBZ3f+Eci*S*@iV$6xLq$i#Am9;eNHbCmqt6a-T#5HZnnneaw
zS;zg;L6Qbp;eD@D()yJt1)%eORcv*M+`>bHGD=`?NFXZ*16C~eDd%P&e;ceex_@@k
z@fZ<pxfqDE!RGRB)9t4+Y(&O_*aN1<!)av!NsjYveO0Z$s*22EoUrx^MXBQzUFjdW
zq>nb;b)j*tBR@g=*S(&Rh%}PKr!eGI^FiC>U2&P)==h%6vD>IioI)jI&n`uCW6~UH
z?3Ror16b{?Dv@XvA0)9PhXOk;Oj&`M5&Qx43djohe`+})z<YVVgo=QQw<R4|j|ElH
KR<2R9j`$xdv7Jc(

literal 0
HcmV?d00001

diff --git a/api/core/model_runtime/model_providers/mixedbread/mixedbread.py b/api/core/model_runtime/model_providers/mixedbread/mixedbread.py
new file mode 100644
index 0000000000000..3c78150e6f806
--- /dev/null
+++ b/api/core/model_runtime/model_providers/mixedbread/mixedbread.py
@@ -0,0 +1,27 @@
+import logging
+
+from core.model_runtime.entities.model_entities import ModelType
+from core.model_runtime.errors.validate import CredentialsValidateFailedError
+from core.model_runtime.model_providers.__base.model_provider import ModelProvider
+
+logger = logging.getLogger(__name__)
+
+
+class MixedBreadProvider(ModelProvider):
+    def validate_provider_credentials(self, credentials: dict) -> None:
+        """
+        Validate provider credentials
+        if validate failed, raise exception
+
+        :param credentials: provider credentials, credentials form defined in `provider_credential_schema`.
+        """
+        try:
+            model_instance = self.get_model_instance(ModelType.TEXT_EMBEDDING)
+
+            # Use `mxbai-embed-large-v1` model for validate,
+            model_instance.validate_credentials(model="mxbai-embed-large-v1", credentials=credentials)
+        except CredentialsValidateFailedError as ex:
+            raise ex
+        except Exception as ex:
+            logger.exception(f"{self.get_provider_schema().provider} credentials validate failed")
+            raise ex
diff --git a/api/core/model_runtime/model_providers/mixedbread/mixedbread.yaml b/api/core/model_runtime/model_providers/mixedbread/mixedbread.yaml
new file mode 100644
index 0000000000000..2f43aea6ade2c
--- /dev/null
+++ b/api/core/model_runtime/model_providers/mixedbread/mixedbread.yaml
@@ -0,0 +1,31 @@
+provider: mixedbread
+label:
+  en_US: MixedBread
+description:
+  en_US: Embedding and Rerank Model Supported
+icon_small:
+  en_US: icon_s_en.png
+icon_large:
+  en_US: icon_l_en.png
+background: "#EFFDFD"
+help:
+  title:
+    en_US: Get your API key from MixedBread AI
+    zh_Hans: 从 MixedBread 获取 API Key
+  url:
+    en_US: https://www.mixedbread.ai/
+supported_model_types:
+  - text-embedding
+  - rerank
+configurate_methods:
+  - predefined-model
+provider_credential_schema:
+  credential_form_schemas:
+    - variable: api_key
+      label:
+        en_US: API Key
+      type: secret-input
+      required: true
+      placeholder:
+        zh_Hans: 在此输入您的 API Key
+        en_US: Enter your API Key
diff --git a/api/core/model_runtime/model_providers/mixedbread/rerank/__init__.py b/api/core/model_runtime/model_providers/mixedbread/rerank/__init__.py
new file mode 100644
index 0000000000000..e69de29bb2d1d
diff --git a/api/core/model_runtime/model_providers/mixedbread/rerank/mxbai-rerank-large-v1-en.yaml b/api/core/model_runtime/model_providers/mixedbread/rerank/mxbai-rerank-large-v1-en.yaml
new file mode 100644
index 0000000000000..beda219953745
--- /dev/null
+++ b/api/core/model_runtime/model_providers/mixedbread/rerank/mxbai-rerank-large-v1-en.yaml
@@ -0,0 +1,4 @@
+model: mxbai-rerank-large-v1
+model_type: rerank
+model_properties:
+  context_size: 512
diff --git a/api/core/model_runtime/model_providers/mixedbread/rerank/rerank.py b/api/core/model_runtime/model_providers/mixedbread/rerank/rerank.py
new file mode 100644
index 0000000000000..bf3c12fd86dc3
--- /dev/null
+++ b/api/core/model_runtime/model_providers/mixedbread/rerank/rerank.py
@@ -0,0 +1,125 @@
+from typing import Optional
+
+import httpx
+
+from core.model_runtime.entities.common_entities import I18nObject
+from core.model_runtime.entities.model_entities import AIModelEntity, FetchFrom, ModelPropertyKey, ModelType
+from core.model_runtime.entities.rerank_entities import RerankDocument, RerankResult
+from core.model_runtime.errors.invoke import (
+    InvokeAuthorizationError,
+    InvokeBadRequestError,
+    InvokeConnectionError,
+    InvokeError,
+    InvokeRateLimitError,
+    InvokeServerUnavailableError,
+)
+from core.model_runtime.errors.validate import CredentialsValidateFailedError
+from core.model_runtime.model_providers.__base.rerank_model import RerankModel
+
+
+class MixedBreadRerankModel(RerankModel):
+    """
+    Model class for MixedBread rerank model.
+    """
+
+    def _invoke(
+        self,
+        model: str,
+        credentials: dict,
+        query: str,
+        docs: list[str],
+        score_threshold: Optional[float] = None,
+        top_n: Optional[int] = None,
+        user: Optional[str] = None,
+    ) -> RerankResult:
+        """
+        Invoke rerank model
+
+        :param model: model name
+        :param credentials: model credentials
+        :param query: search query
+        :param docs: docs for reranking
+        :param score_threshold: score threshold
+        :param top_n: top n documents to return
+        :param user: unique user id
+        :return: rerank result
+        """
+        if len(docs) == 0:
+            return RerankResult(model=model, docs=[])
+
+        base_url = credentials.get("base_url", "https://api.mixedbread.ai/v1")
+        base_url = base_url.removesuffix("/")
+
+        try:
+            response = httpx.post(
+                base_url + "/reranking",
+                json={"model": model, "query": query, "input": docs, "top_k": top_n, "return_input": True},
+                headers={"Authorization": f"Bearer {credentials.get('api_key')}", "Content-Type": "application/json"},
+            )
+            response.raise_for_status()
+            results = response.json()
+
+            rerank_documents = []
+            for result in results["data"]:
+                rerank_document = RerankDocument(
+                    index=result["index"],
+                    text=result["input"],
+                    score=result["score"],
+                )
+                if score_threshold is None or result["score"] >= score_threshold:
+                    rerank_documents.append(rerank_document)
+
+            return RerankResult(model=model, docs=rerank_documents)
+        except httpx.HTTPStatusError as e:
+            raise InvokeServerUnavailableError(str(e))
+
+    def validate_credentials(self, model: str, credentials: dict) -> None:
+        """
+        Validate model credentials
+
+        :param model: model name
+        :param credentials: model credentials
+        :return:
+        """
+        try:
+            self._invoke(
+                model=model,
+                credentials=credentials,
+                query="What is the capital of the United States?",
+                docs=[
+                    "Carson City is the capital city of the American state of Nevada. At the 2010 United States "
+                    "Census, Carson City had a population of 55,274.",
+                    "The Commonwealth of the Northern Mariana Islands is a group of islands in the Pacific Ocean that "
+                    "are a political division controlled by the United States. Its capital is Saipan.",
+                ],
+                score_threshold=0.8,
+            )
+        except Exception as ex:
+            raise CredentialsValidateFailedError(str(ex))
+
+    @property
+    def _invoke_error_mapping(self) -> dict[type[InvokeError], list[type[Exception]]]:
+        """
+        Map model invoke error to unified error
+        """
+        return {
+            InvokeConnectionError: [httpx.ConnectError],
+            InvokeServerUnavailableError: [httpx.RemoteProtocolError],
+            InvokeRateLimitError: [],
+            InvokeAuthorizationError: [httpx.HTTPStatusError],
+            InvokeBadRequestError: [httpx.RequestError],
+        }
+
+    def get_customizable_model_schema(self, model: str, credentials: dict) -> AIModelEntity:
+        """
+        generate custom model entities from credentials
+        """
+        entity = AIModelEntity(
+            model=model,
+            label=I18nObject(en_US=model),
+            model_type=ModelType.RERANK,
+            fetch_from=FetchFrom.CUSTOMIZABLE_MODEL,
+            model_properties={ModelPropertyKey.CONTEXT_SIZE: int(credentials.get("context_size", "512"))},
+        )
+
+        return entity
diff --git a/api/core/model_runtime/model_providers/mixedbread/text_embedding/__init__.py b/api/core/model_runtime/model_providers/mixedbread/text_embedding/__init__.py
new file mode 100644
index 0000000000000..e69de29bb2d1d
diff --git a/api/core/model_runtime/model_providers/mixedbread/text_embedding/mxbai-embed-2d-large-v1-en.yaml b/api/core/model_runtime/model_providers/mixedbread/text_embedding/mxbai-embed-2d-large-v1-en.yaml
new file mode 100644
index 0000000000000..0c3c863d06b89
--- /dev/null
+++ b/api/core/model_runtime/model_providers/mixedbread/text_embedding/mxbai-embed-2d-large-v1-en.yaml
@@ -0,0 +1,8 @@
+model: mxbai-embed-2d-large-v1
+model_type: text-embedding
+model_properties:
+  context_size: 512
+pricing:
+  input: '0.0001'
+  unit: '0.001'
+  currency: USD
diff --git a/api/core/model_runtime/model_providers/mixedbread/text_embedding/mxbai-embed-large-v1-en.yaml b/api/core/model_runtime/model_providers/mixedbread/text_embedding/mxbai-embed-large-v1-en.yaml
new file mode 100644
index 0000000000000..0c5cda2a72a99
--- /dev/null
+++ b/api/core/model_runtime/model_providers/mixedbread/text_embedding/mxbai-embed-large-v1-en.yaml
@@ -0,0 +1,8 @@
+model: mxbai-embed-large-v1
+model_type: text-embedding
+model_properties:
+  context_size: 512
+pricing:
+  input: '0.0001'
+  unit: '0.001'
+  currency: USD
diff --git a/api/core/model_runtime/model_providers/mixedbread/text_embedding/text_embedding.py b/api/core/model_runtime/model_providers/mixedbread/text_embedding/text_embedding.py
new file mode 100644
index 0000000000000..05d9a9a0c69f4
--- /dev/null
+++ b/api/core/model_runtime/model_providers/mixedbread/text_embedding/text_embedding.py
@@ -0,0 +1,163 @@
+import time
+from json import JSONDecodeError, dumps
+from typing import Optional
+
+import requests
+
+from core.model_runtime.entities.common_entities import I18nObject
+from core.model_runtime.entities.model_entities import AIModelEntity, FetchFrom, ModelPropertyKey, ModelType, PriceType
+from core.model_runtime.entities.text_embedding_entities import EmbeddingUsage, TextEmbeddingResult
+from core.model_runtime.errors.invoke import (
+    InvokeAuthorizationError,
+    InvokeBadRequestError,
+    InvokeConnectionError,
+    InvokeError,
+    InvokeRateLimitError,
+    InvokeServerUnavailableError,
+)
+from core.model_runtime.errors.validate import CredentialsValidateFailedError
+from core.model_runtime.model_providers.__base.text_embedding_model import TextEmbeddingModel
+
+
+class MixedBreadTextEmbeddingModel(TextEmbeddingModel):
+    """
+    Model class for MixedBread text embedding model.
+    """
+
+    api_base: str = "https://api.mixedbread.ai/v1"
+
+    def _invoke(
+        self, model: str, credentials: dict, texts: list[str], user: Optional[str] = None
+    ) -> TextEmbeddingResult:
+        """
+        Invoke text embedding model
+
+        :param model: model name
+        :param credentials: model credentials
+        :param texts: texts to embed
+        :param user: unique user id
+        :return: embeddings result
+        """
+        api_key = credentials["api_key"]
+        if not api_key:
+            raise CredentialsValidateFailedError("api_key is required")
+
+        base_url = credentials.get("base_url", self.api_base)
+        base_url = base_url.removesuffix("/")
+
+        url = base_url + "/embeddings"
+        headers = {"Authorization": "Bearer " + api_key, "Content-Type": "application/json"}
+
+        data = {"model": model, "input": texts}
+
+        try:
+            response = requests.post(url, headers=headers, data=dumps(data))
+        except Exception as e:
+            raise InvokeConnectionError(str(e))
+
+        if response.status_code != 200:
+            try:
+                resp = response.json()
+                msg = resp["detail"]
+                if response.status_code == 401:
+                    raise InvokeAuthorizationError(msg)
+                elif response.status_code == 429:
+                    raise InvokeRateLimitError(msg)
+                elif response.status_code == 500:
+                    raise InvokeServerUnavailableError(msg)
+                else:
+                    raise InvokeBadRequestError(msg)
+            except JSONDecodeError as e:
+                raise InvokeServerUnavailableError(
+                    f"Failed to convert response to json: {e} with text: {response.text}"
+                )
+
+        try:
+            resp = response.json()
+            embeddings = resp["data"]
+            usage = resp["usage"]
+        except Exception as e:
+            raise InvokeServerUnavailableError(f"Failed to convert response to json: {e} with text: {response.text}")
+
+        usage = self._calc_response_usage(model=model, credentials=credentials, tokens=usage["total_tokens"])
+
+        result = TextEmbeddingResult(
+            model=model, embeddings=[[float(data) for data in x["embedding"]] for x in embeddings], usage=usage
+        )
+
+        return result
+
+    def get_num_tokens(self, model: str, credentials: dict, texts: list[str]) -> int:
+        """
+        Get number of tokens for given prompt messages
+
+        :param model: model name
+        :param credentials: model credentials
+        :param texts: texts to embed
+        :return:
+        """
+        return sum(self._get_num_tokens_by_gpt2(text) for text in texts)
+
+    def validate_credentials(self, model: str, credentials: dict) -> None:
+        """
+        Validate model credentials
+
+        :param model: model name
+        :param credentials: model credentials
+        :return:
+        """
+        try:
+            self._invoke(model=model, credentials=credentials, texts=["ping"])
+        except Exception as e:
+            raise CredentialsValidateFailedError(f"Credentials validation failed: {e}")
+
+    @property
+    def _invoke_error_mapping(self) -> dict[type[InvokeError], list[type[Exception]]]:
+        return {
+            InvokeConnectionError: [InvokeConnectionError],
+            InvokeServerUnavailableError: [InvokeServerUnavailableError],
+            InvokeRateLimitError: [InvokeRateLimitError],
+            InvokeAuthorizationError: [InvokeAuthorizationError],
+            InvokeBadRequestError: [KeyError, InvokeBadRequestError],
+        }
+
+    def _calc_response_usage(self, model: str, credentials: dict, tokens: int) -> EmbeddingUsage:
+        """
+        Calculate response usage
+
+        :param model: model name
+        :param credentials: model credentials
+        :param tokens: input tokens
+        :return: usage
+        """
+        # get input price info
+        input_price_info = self.get_price(
+            model=model, credentials=credentials, price_type=PriceType.INPUT, tokens=tokens
+        )
+
+        # transform usage
+        usage = EmbeddingUsage(
+            tokens=tokens,
+            total_tokens=tokens,
+            unit_price=input_price_info.unit_price,
+            price_unit=input_price_info.unit,
+            total_price=input_price_info.total_amount,
+            currency=input_price_info.currency,
+            latency=time.perf_counter() - self.started_at,
+        )
+
+        return usage
+
+    def get_customizable_model_schema(self, model: str, credentials: dict) -> AIModelEntity:
+        """
+        generate custom model entities from credentials
+        """
+        entity = AIModelEntity(
+            model=model,
+            label=I18nObject(en_US=model),
+            model_type=ModelType.TEXT_EMBEDDING,
+            fetch_from=FetchFrom.CUSTOMIZABLE_MODEL,
+            model_properties={ModelPropertyKey.CONTEXT_SIZE: int(credentials.get("context_size", "512"))},
+        )
+
+        return entity
diff --git a/api/pyproject.toml b/api/pyproject.toml
index 41244f516c95c..9e38c0945601c 100644
--- a/api/pyproject.toml
+++ b/api/pyproject.toml
@@ -122,6 +122,7 @@ CODE_EXECUTION_API_KEY = "dify-sandbox"
 FIRECRAWL_API_KEY = "fc-"
 TEI_EMBEDDING_SERVER_URL = "http://a.abc.com:11451"
 TEI_RERANK_SERVER_URL = "http://a.abc.com:11451"
+MIXEDBREAD_API_KEY = "mk-aaaaaaaaaaaaaaaaaaaa"
 
 [tool.poetry]
 name = "dify-api"
diff --git a/api/tests/integration_tests/model_runtime/mixedbread/__init__.py b/api/tests/integration_tests/model_runtime/mixedbread/__init__.py
new file mode 100644
index 0000000000000..e69de29bb2d1d
diff --git a/api/tests/integration_tests/model_runtime/mixedbread/test_provider.py b/api/tests/integration_tests/model_runtime/mixedbread/test_provider.py
new file mode 100644
index 0000000000000..25c9f3ce8dffa
--- /dev/null
+++ b/api/tests/integration_tests/model_runtime/mixedbread/test_provider.py
@@ -0,0 +1,28 @@
+import os
+from unittest.mock import Mock, patch
+
+import pytest
+
+from core.model_runtime.errors.validate import CredentialsValidateFailedError
+from core.model_runtime.model_providers.mixedbread.mixedbread import MixedBreadProvider
+
+
+def test_validate_provider_credentials():
+    provider = MixedBreadProvider()
+
+    with pytest.raises(CredentialsValidateFailedError):
+        provider.validate_provider_credentials(credentials={"api_key": "hahahaha"})
+    with patch("requests.post") as mock_post:
+        mock_response = Mock()
+        mock_response.json.return_value = {
+            "usage": {"prompt_tokens": 3, "total_tokens": 3},
+            "model": "mixedbread-ai/mxbai-embed-large-v1",
+            "data": [{"embedding": [0.23333 for _ in range(1024)], "index": 0, "object": "embedding"}],
+            "object": "list",
+            "normalized": "true",
+            "encoding_format": "float",
+            "dimensions": 1024,
+        }
+        mock_response.status_code = 200
+        mock_post.return_value = mock_response
+        provider.validate_provider_credentials(credentials={"api_key": os.environ.get("MIXEDBREAD_API_KEY")})
diff --git a/api/tests/integration_tests/model_runtime/mixedbread/test_rerank.py b/api/tests/integration_tests/model_runtime/mixedbread/test_rerank.py
new file mode 100644
index 0000000000000..b65aab74aa96d
--- /dev/null
+++ b/api/tests/integration_tests/model_runtime/mixedbread/test_rerank.py
@@ -0,0 +1,100 @@
+import os
+from unittest.mock import Mock, patch
+
+import pytest
+
+from core.model_runtime.entities.rerank_entities import RerankResult
+from core.model_runtime.errors.validate import CredentialsValidateFailedError
+from core.model_runtime.model_providers.mixedbread.rerank.rerank import MixedBreadRerankModel
+
+
+def test_validate_credentials():
+    model = MixedBreadRerankModel()
+
+    with pytest.raises(CredentialsValidateFailedError):
+        model.validate_credentials(
+            model="mxbai-rerank-large-v1",
+            credentials={"api_key": "invalid_key"},
+        )
+    with patch("httpx.post") as mock_post:
+        mock_response = Mock()
+        mock_response.json.return_value = {
+            "usage": {"prompt_tokens": 86, "total_tokens": 86},
+            "model": "mixedbread-ai/mxbai-rerank-large-v1",
+            "data": [
+                {
+                    "index": 0,
+                    "score": 0.06762695,
+                    "input": "Carson City is the capital city of the American state of Nevada. At the 2010 United "
+                    "States Census, Carson City had a population of 55,274.",
+                    "object": "text_document",
+                },
+                {
+                    "index": 1,
+                    "score": 0.057403564,
+                    "input": "The Commonwealth of the Northern Mariana Islands is a group of islands in the Pacific "
+                    "Ocean that are a political division controlled by the United States. Its capital is "
+                    "Saipan.",
+                    "object": "text_document",
+                },
+            ],
+            "object": "list",
+            "top_k": 2,
+            "return_input": True,
+        }
+        mock_response.status_code = 200
+        mock_post.return_value = mock_response
+        model.validate_credentials(
+            model="mxbai-rerank-large-v1",
+            credentials={
+                "api_key": os.environ.get("MIXEDBREAD_API_KEY"),
+            },
+        )
+
+
+def test_invoke_model():
+    model = MixedBreadRerankModel()
+    with patch("httpx.post") as mock_post:
+        mock_response = Mock()
+        mock_response.json.return_value = {
+            "usage": {"prompt_tokens": 56, "total_tokens": 56},
+            "model": "mixedbread-ai/mxbai-rerank-large-v1",
+            "data": [
+                {
+                    "index": 0,
+                    "score": 0.6044922,
+                    "input": "Kasumi is a girl name of Japanese origin meaning mist.",
+                    "object": "text_document",
+                },
+                {
+                    "index": 1,
+                    "score": 0.0703125,
+                    "input": "Her music is a kawaii bass, a mix of future bass, pop, and kawaii music and she leads a "
+                    "team named PopiParty.",
+                    "object": "text_document",
+                },
+            ],
+            "object": "list",
+            "top_k": 2,
+            "return_input": "true",
+        }
+        mock_response.status_code = 200
+        mock_post.return_value = mock_response
+        result = model.invoke(
+            model="mxbai-rerank-large-v1",
+            credentials={
+                "api_key": os.environ.get("MIXEDBREAD_API_KEY"),
+            },
+            query="Who is Kasumi?",
+            docs=[
+                "Kasumi is a girl name of Japanese origin meaning mist.",
+                "Her music is a kawaii bass, a mix of future bass, pop, and kawaii music and she leads a team named "
+                "PopiParty.",
+            ],
+            score_threshold=0.5,
+        )
+
+        assert isinstance(result, RerankResult)
+        assert len(result.docs) == 1
+        assert result.docs[0].index == 0
+        assert result.docs[0].score >= 0.5
diff --git a/api/tests/integration_tests/model_runtime/mixedbread/test_text_embedding.py b/api/tests/integration_tests/model_runtime/mixedbread/test_text_embedding.py
new file mode 100644
index 0000000000000..ca97a1895113f
--- /dev/null
+++ b/api/tests/integration_tests/model_runtime/mixedbread/test_text_embedding.py
@@ -0,0 +1,78 @@
+import os
+from unittest.mock import Mock, patch
+
+import pytest
+
+from core.model_runtime.entities.text_embedding_entities import TextEmbeddingResult
+from core.model_runtime.errors.validate import CredentialsValidateFailedError
+from core.model_runtime.model_providers.mixedbread.text_embedding.text_embedding import MixedBreadTextEmbeddingModel
+
+
+def test_validate_credentials():
+    model = MixedBreadTextEmbeddingModel()
+
+    with pytest.raises(CredentialsValidateFailedError):
+        model.validate_credentials(model="mxbai-embed-large-v1", credentials={"api_key": "invalid_key"})
+    with patch("requests.post") as mock_post:
+        mock_response = Mock()
+        mock_response.json.return_value = {
+            "usage": {"prompt_tokens": 3, "total_tokens": 3},
+            "model": "mixedbread-ai/mxbai-embed-large-v1",
+            "data": [{"embedding": [0.23333 for _ in range(1024)], "index": 0, "object": "embedding"}],
+            "object": "list",
+            "normalized": "true",
+            "encoding_format": "float",
+            "dimensions": 1024,
+        }
+        mock_response.status_code = 200
+        mock_post.return_value = mock_response
+        model.validate_credentials(
+            model="mxbai-embed-large-v1", credentials={"api_key": os.environ.get("MIXEDBREAD_API_KEY")}
+        )
+
+
+def test_invoke_model():
+    model = MixedBreadTextEmbeddingModel()
+
+    with patch("requests.post") as mock_post:
+        mock_response = Mock()
+        mock_response.json.return_value = {
+            "usage": {"prompt_tokens": 6, "total_tokens": 6},
+            "model": "mixedbread-ai/mxbai-embed-large-v1",
+            "data": [
+                {"embedding": [0.23333 for _ in range(1024)], "index": 0, "object": "embedding"},
+                {"embedding": [0.23333 for _ in range(1024)], "index": 1, "object": "embedding"},
+            ],
+            "object": "list",
+            "normalized": "true",
+            "encoding_format": "float",
+            "dimensions": 1024,
+        }
+        mock_response.status_code = 200
+        mock_post.return_value = mock_response
+        result = model.invoke(
+            model="mxbai-embed-large-v1",
+            credentials={
+                "api_key": os.environ.get("MIXEDBREAD_API_KEY"),
+            },
+            texts=["hello", "world"],
+            user="abc-123",
+        )
+
+        assert isinstance(result, TextEmbeddingResult)
+        assert len(result.embeddings) == 2
+        assert result.usage.total_tokens == 6
+
+
+def test_get_num_tokens():
+    model = MixedBreadTextEmbeddingModel()
+
+    num_tokens = model.get_num_tokens(
+        model="mxbai-embed-large-v1",
+        credentials={
+            "api_key": os.environ.get("MIXEDBREAD_API_KEY"),
+        },
+        texts=["ping"],
+    )
+
+    assert num_tokens == 1
diff --git a/dev/pytest/pytest_model_runtime.sh b/dev/pytest/pytest_model_runtime.sh
index 4c0083a2de343..b60ff64fdcd90 100755
--- a/dev/pytest/pytest_model_runtime.sh
+++ b/dev/pytest/pytest_model_runtime.sh
@@ -8,4 +8,5 @@ pytest api/tests/integration_tests/model_runtime/anthropic \
   api/tests/integration_tests/model_runtime/huggingface_hub/test_llm.py \
   api/tests/integration_tests/model_runtime/upstage \
   api/tests/integration_tests/model_runtime/fireworks \
-  api/tests/integration_tests/model_runtime/nomic
+  api/tests/integration_tests/model_runtime/nomic \
+  api/tests/integration_tests/model_runtime/mixedbread

From aebe5fc68ceeb30c8a8958c644197cc8690c0429 Mon Sep 17 00:00:00 2001
From: AAEE86 <33052466+AAEE86@users.noreply.github.com>
Date: Tue, 24 Sep 2024 13:06:21 +0800
Subject: [PATCH 049/235] fix: Remove unsupported parameters in qwen model
 (#8699)

---
 .../model_providers/tongyi/llm/farui-plus.yaml           | 9 ---------
 .../tongyi/llm/qwen-coder-turbo-0919.yaml                | 9 ---------
 .../tongyi/llm/qwen-coder-turbo-latest.yaml              | 9 ---------
 .../model_providers/tongyi/llm/qwen-coder-turbo.yaml     | 9 ---------
 .../model_providers/tongyi/llm/qwen-long.yaml            | 9 ---------
 .../model_providers/tongyi/llm/qwen-math-plus-0816.yaml  | 9 ---------
 .../model_providers/tongyi/llm/qwen-math-plus-0919.yaml  | 9 ---------
 .../tongyi/llm/qwen-math-plus-latest.yaml                | 9 ---------
 .../model_providers/tongyi/llm/qwen-math-plus.yaml       | 9 ---------
 .../model_providers/tongyi/llm/qwen-math-turbo-0919.yaml | 9 ---------
 .../tongyi/llm/qwen-math-turbo-latest.yaml               | 9 ---------
 .../model_providers/tongyi/llm/qwen-math-turbo.yaml      | 9 ---------
 .../model_providers/tongyi/llm/qwen-max-0107.yaml        | 9 ---------
 .../model_providers/tongyi/llm/qwen-max-0403.yaml        | 9 ---------
 .../model_providers/tongyi/llm/qwen-max-0428.yaml        | 9 ---------
 .../model_providers/tongyi/llm/qwen-max-0919.yaml        | 9 ---------
 .../model_providers/tongyi/llm/qwen-max-1201.yaml        | 6 ------
 .../model_providers/tongyi/llm/qwen-max-latest.yaml      | 9 ---------
 .../model_providers/tongyi/llm/qwen-max-longcontext.yaml | 9 ---------
 .../model_providers/tongyi/llm/qwen-plus-0206.yaml       | 9 ---------
 .../model_providers/tongyi/llm/qwen-plus-0624.yaml       | 9 ---------
 .../model_providers/tongyi/llm/qwen-plus-0723.yaml       | 9 ---------
 .../model_providers/tongyi/llm/qwen-plus-0806.yaml       | 9 ---------
 .../model_providers/tongyi/llm/qwen-plus-0919.yaml       | 9 ---------
 .../model_providers/tongyi/llm/qwen-plus-chat.yaml       | 9 ---------
 .../model_providers/tongyi/llm/qwen-plus-latest.yaml     | 9 ---------
 .../model_providers/tongyi/llm/qwen-turbo-0206.yaml      | 9 ---------
 .../model_providers/tongyi/llm/qwen-turbo-0624.yaml      | 9 ---------
 .../model_providers/tongyi/llm/qwen-turbo-0919.yaml      | 9 ---------
 .../model_providers/tongyi/llm/qwen-turbo-chat.yaml      | 9 ---------
 .../model_providers/tongyi/llm/qwen-turbo-latest.yaml    | 9 ---------
 .../model_providers/tongyi/llm/qwen-vl-max-0809.yaml     | 9 ---------
 .../model_providers/tongyi/llm/qwen-vl-max.yaml          | 9 ---------
 .../model_providers/tongyi/llm/qwen-vl-plus-0201.yaml    | 9 ---------
 .../model_providers/tongyi/llm/qwen-vl-plus-0809.yaml    | 9 ---------
 .../model_providers/tongyi/llm/qwen-vl-plus.yaml         | 9 ---------
 .../tongyi/llm/qwen2-math-1.5b-instruct.yaml             | 9 ---------
 .../tongyi/llm/qwen2-math-72b-instruct.yaml              | 9 ---------
 .../tongyi/llm/qwen2-math-7b-instruct.yaml               | 9 ---------
 .../tongyi/llm/qwen2.5-0.5b-instruct.yaml                | 9 ---------
 .../tongyi/llm/qwen2.5-1.5b-instruct.yaml                | 9 ---------
 .../model_providers/tongyi/llm/qwen2.5-14b-instruct.yaml | 9 ---------
 .../model_providers/tongyi/llm/qwen2.5-32b-instruct.yaml | 9 ---------
 .../model_providers/tongyi/llm/qwen2.5-3b-instruct.yaml  | 9 ---------
 .../model_providers/tongyi/llm/qwen2.5-72b-instruct.yaml | 9 ---------
 .../model_providers/tongyi/llm/qwen2.5-7b-instruct.yaml  | 9 ---------
 .../tongyi/llm/qwen2.5-coder-7b-instruct.yaml            | 9 ---------
 47 files changed, 420 deletions(-)

diff --git a/api/core/model_runtime/model_providers/tongyi/llm/farui-plus.yaml b/api/core/model_runtime/model_providers/tongyi/llm/farui-plus.yaml
index e5de586c1caf1..d0ff44382789b 100644
--- a/api/core/model_runtime/model_providers/tongyi/llm/farui-plus.yaml
+++ b/api/core/model_runtime/model_providers/tongyi/llm/farui-plus.yaml
@@ -67,15 +67,6 @@ parameter_rules:
     help:
       zh_Hans: 用于控制模型生成时的重复度。提高repetition_penalty时可以降低模型生成的重复度。1.0表示不做惩罚。
       en_US: Used to control the repeatability when generating models. Increasing repetition_penalty can reduce the duplication of model generation. 1.0 means no punishment.
-  - name: enable_search
-    type: boolean
-    default: false
-    label:
-      zh_Hans: 联网搜索
-      en_US: Web Search
-    help:
-      zh_Hans: 模型内置了互联网搜索服务，该参数控制模型在生成文本时是否参考使用互联网搜索结果。启用互联网搜索，模型会将搜索结果作为文本生成过程中的参考信息，但模型会基于其内部逻辑“自行判断”是否使用互联网搜索结果。
-      en_US: The model has a built-in Internet search service. This parameter controls whether the model refers to Internet search results when generating text. When Internet search is enabled, the model will use the search results as reference information in the text generation process, but the model will "judge" whether to use Internet search results based on its internal logic.
   - name: response_format
     use_template: response_format
 pricing:
diff --git a/api/core/model_runtime/model_providers/tongyi/llm/qwen-coder-turbo-0919.yaml b/api/core/model_runtime/model_providers/tongyi/llm/qwen-coder-turbo-0919.yaml
index 6ab39cde2d052..d9792e71ee454 100644
--- a/api/core/model_runtime/model_providers/tongyi/llm/qwen-coder-turbo-0919.yaml
+++ b/api/core/model_runtime/model_providers/tongyi/llm/qwen-coder-turbo-0919.yaml
@@ -65,15 +65,6 @@ parameter_rules:
     help:
       zh_Hans: 用于控制模型生成时的重复度。提高repetition_penalty时可以降低模型生成的重复度。1.0表示不做惩罚。
       en_US: Used to control the repeatability when generating models. Increasing repetition_penalty can reduce the duplication of model generation. 1.0 means no punishment.
-  - name: enable_search
-    type: boolean
-    default: false
-    label:
-      zh_Hans: 联网搜索
-      en_US: Web Search
-    help:
-      zh_Hans: 模型内置了互联网搜索服务，该参数控制模型在生成文本时是否参考使用互联网搜索结果。启用互联网搜索，模型会将搜索结果作为文本生成过程中的参考信息，但模型会基于其内部逻辑“自行判断”是否使用互联网搜索结果。
-      en_US: The model has a built-in Internet search service. This parameter controls whether the model refers to Internet search results when generating text. When Internet search is enabled, the model will use the search results as reference information in the text generation process, but the model will "judge" whether to use Internet search results based on its internal logic.
   - name: response_format
     use_template: response_format
 pricing:
diff --git a/api/core/model_runtime/model_providers/tongyi/llm/qwen-coder-turbo-latest.yaml b/api/core/model_runtime/model_providers/tongyi/llm/qwen-coder-turbo-latest.yaml
index be6d9a0e07272..0b03505c45084 100644
--- a/api/core/model_runtime/model_providers/tongyi/llm/qwen-coder-turbo-latest.yaml
+++ b/api/core/model_runtime/model_providers/tongyi/llm/qwen-coder-turbo-latest.yaml
@@ -65,15 +65,6 @@ parameter_rules:
     help:
       zh_Hans: 用于控制模型生成时的重复度。提高repetition_penalty时可以降低模型生成的重复度。1.0表示不做惩罚。
       en_US: Used to control the repeatability when generating models. Increasing repetition_penalty can reduce the duplication of model generation. 1.0 means no punishment.
-  - name: enable_search
-    type: boolean
-    default: false
-    label:
-      zh_Hans: 联网搜索
-      en_US: Web Search
-    help:
-      zh_Hans: 模型内置了互联网搜索服务，该参数控制模型在生成文本时是否参考使用互联网搜索结果。启用互联网搜索，模型会将搜索结果作为文本生成过程中的参考信息，但模型会基于其内部逻辑“自行判断”是否使用互联网搜索结果。
-      en_US: The model has a built-in Internet search service. This parameter controls whether the model refers to Internet search results when generating text. When Internet search is enabled, the model will use the search results as reference information in the text generation process, but the model will "judge" whether to use Internet search results based on its internal logic.
   - name: response_format
     use_template: response_format
 pricing:
diff --git a/api/core/model_runtime/model_providers/tongyi/llm/qwen-coder-turbo.yaml b/api/core/model_runtime/model_providers/tongyi/llm/qwen-coder-turbo.yaml
index d2aca4f514ae0..2a6c040853005 100644
--- a/api/core/model_runtime/model_providers/tongyi/llm/qwen-coder-turbo.yaml
+++ b/api/core/model_runtime/model_providers/tongyi/llm/qwen-coder-turbo.yaml
@@ -65,15 +65,6 @@ parameter_rules:
     help:
       zh_Hans: 用于控制模型生成时的重复度。提高repetition_penalty时可以降低模型生成的重复度。1.0表示不做惩罚。
       en_US: Used to control the repeatability when generating models. Increasing repetition_penalty can reduce the duplication of model generation. 1.0 means no punishment.
-  - name: enable_search
-    type: boolean
-    default: false
-    label:
-      zh_Hans: 联网搜索
-      en_US: Web Search
-    help:
-      zh_Hans: 模型内置了互联网搜索服务，该参数控制模型在生成文本时是否参考使用互联网搜索结果。启用互联网搜索，模型会将搜索结果作为文本生成过程中的参考信息，但模型会基于其内部逻辑“自行判断”是否使用互联网搜索结果。
-      en_US: The model has a built-in Internet search service. This parameter controls whether the model refers to Internet search results when generating text. When Internet search is enabled, the model will use the search results as reference information in the text generation process, but the model will "judge" whether to use Internet search results based on its internal logic.
   - name: response_format
     use_template: response_format
 pricing:
diff --git a/api/core/model_runtime/model_providers/tongyi/llm/qwen-long.yaml b/api/core/model_runtime/model_providers/tongyi/llm/qwen-long.yaml
index a59a3350f62dd..bad7f4f4725c4 100644
--- a/api/core/model_runtime/model_providers/tongyi/llm/qwen-long.yaml
+++ b/api/core/model_runtime/model_providers/tongyi/llm/qwen-long.yaml
@@ -68,15 +68,6 @@ parameter_rules:
     help:
       zh_Hans: 用于控制模型生成时的重复度。提高repetition_penalty时可以降低模型生成的重复度。1.0表示不做惩罚。
       en_US: Used to control the repeatability when generating models. Increasing repetition_penalty can reduce the duplication of model generation. 1.0 means no punishment.
-  - name: enable_search
-    type: boolean
-    default: false
-    label:
-      zh_Hans: 联网搜索
-      en_US: Web Search
-    help:
-      zh_Hans: 模型内置了互联网搜索服务，该参数控制模型在生成文本时是否参考使用互联网搜索结果。启用互联网搜索，模型会将搜索结果作为文本生成过程中的参考信息，但模型会基于其内部逻辑“自行判断”是否使用互联网搜索结果。
-      en_US: The model has a built-in Internet search service. This parameter controls whether the model refers to Internet search results when generating text. When Internet search is enabled, the model will use the search results as reference information in the text generation process, but the model will "judge" whether to use Internet search results based on its internal logic.
   - name: response_format
     use_template: response_format
 pricing:
diff --git a/api/core/model_runtime/model_providers/tongyi/llm/qwen-math-plus-0816.yaml b/api/core/model_runtime/model_providers/tongyi/llm/qwen-math-plus-0816.yaml
index cab7233c98d69..c14aee1e1eb87 100644
--- a/api/core/model_runtime/model_providers/tongyi/llm/qwen-math-plus-0816.yaml
+++ b/api/core/model_runtime/model_providers/tongyi/llm/qwen-math-plus-0816.yaml
@@ -65,15 +65,6 @@ parameter_rules:
     help:
       zh_Hans: 用于控制模型生成时的重复度。提高repetition_penalty时可以降低模型生成的重复度。1.0表示不做惩罚。
       en_US: Used to control the repeatability when generating models. Increasing repetition_penalty can reduce the duplication of model generation. 1.0 means no punishment.
-  - name: enable_search
-    type: boolean
-    default: false
-    label:
-      zh_Hans: 联网搜索
-      en_US: Web Search
-    help:
-      zh_Hans: 模型内置了互联网搜索服务，该参数控制模型在生成文本时是否参考使用互联网搜索结果。启用互联网搜索，模型会将搜索结果作为文本生成过程中的参考信息，但模型会基于其内部逻辑“自行判断”是否使用互联网搜索结果。
-      en_US: The model has a built-in Internet search service. This parameter controls whether the model refers to Internet search results when generating text. When Internet search is enabled, the model will use the search results as reference information in the text generation process, but the model will "judge" whether to use Internet search results based on its internal logic.
   - name: response_format
     use_template: response_format
 pricing:
diff --git a/api/core/model_runtime/model_providers/tongyi/llm/qwen-math-plus-0919.yaml b/api/core/model_runtime/model_providers/tongyi/llm/qwen-math-plus-0919.yaml
index f82fba0c012aa..9d74eeca3e878 100644
--- a/api/core/model_runtime/model_providers/tongyi/llm/qwen-math-plus-0919.yaml
+++ b/api/core/model_runtime/model_providers/tongyi/llm/qwen-math-plus-0919.yaml
@@ -65,15 +65,6 @@ parameter_rules:
     help:
       zh_Hans: 用于控制模型生成时的重复度。提高repetition_penalty时可以降低模型生成的重复度。1.0表示不做惩罚。
       en_US: Used to control the repeatability when generating models. Increasing repetition_penalty can reduce the duplication of model generation. 1.0 means no punishment.
-  - name: enable_search
-    type: boolean
-    default: false
-    label:
-      zh_Hans: 联网搜索
-      en_US: Web Search
-    help:
-      zh_Hans: 模型内置了互联网搜索服务，该参数控制模型在生成文本时是否参考使用互联网搜索结果。启用互联网搜索，模型会将搜索结果作为文本生成过程中的参考信息，但模型会基于其内部逻辑“自行判断”是否使用互联网搜索结果。
-      en_US: The model has a built-in Internet search service. This parameter controls whether the model refers to Internet search results when generating text. When Internet search is enabled, the model will use the search results as reference information in the text generation process, but the model will "judge" whether to use Internet search results based on its internal logic.
   - name: response_format
     use_template: response_format
 pricing:
diff --git a/api/core/model_runtime/model_providers/tongyi/llm/qwen-math-plus-latest.yaml b/api/core/model_runtime/model_providers/tongyi/llm/qwen-math-plus-latest.yaml
index e2fb6e0e554d4..b8601a969a7fa 100644
--- a/api/core/model_runtime/model_providers/tongyi/llm/qwen-math-plus-latest.yaml
+++ b/api/core/model_runtime/model_providers/tongyi/llm/qwen-math-plus-latest.yaml
@@ -65,15 +65,6 @@ parameter_rules:
     help:
       zh_Hans: 用于控制模型生成时的重复度。提高repetition_penalty时可以降低模型生成的重复度。1.0表示不做惩罚。
       en_US: Used to control the repeatability when generating models. Increasing repetition_penalty can reduce the duplication of model generation. 1.0 means no punishment.
-  - name: enable_search
-    type: boolean
-    default: false
-    label:
-      zh_Hans: 联网搜索
-      en_US: Web Search
-    help:
-      zh_Hans: 模型内置了互联网搜索服务，该参数控制模型在生成文本时是否参考使用互联网搜索结果。启用互联网搜索，模型会将搜索结果作为文本生成过程中的参考信息，但模型会基于其内部逻辑“自行判断”是否使用互联网搜索结果。
-      en_US: The model has a built-in Internet search service. This parameter controls whether the model refers to Internet search results when generating text. When Internet search is enabled, the model will use the search results as reference information in the text generation process, but the model will "judge" whether to use Internet search results based on its internal logic.
   - name: response_format
     use_template: response_format
 pricing:
diff --git a/api/core/model_runtime/model_providers/tongyi/llm/qwen-math-plus.yaml b/api/core/model_runtime/model_providers/tongyi/llm/qwen-math-plus.yaml
index 8803e747e5eb7..4a948be597649 100644
--- a/api/core/model_runtime/model_providers/tongyi/llm/qwen-math-plus.yaml
+++ b/api/core/model_runtime/model_providers/tongyi/llm/qwen-math-plus.yaml
@@ -65,15 +65,6 @@ parameter_rules:
     help:
       zh_Hans: 用于控制模型生成时的重复度。提高repetition_penalty时可以降低模型生成的重复度。1.0表示不做惩罚。
       en_US: Used to control the repeatability when generating models. Increasing repetition_penalty can reduce the duplication of model generation. 1.0 means no punishment.
-  - name: enable_search
-    type: boolean
-    default: false
-    label:
-      zh_Hans: 联网搜索
-      en_US: Web Search
-    help:
-      zh_Hans: 模型内置了互联网搜索服务，该参数控制模型在生成文本时是否参考使用互联网搜索结果。启用互联网搜索，模型会将搜索结果作为文本生成过程中的参考信息，但模型会基于其内部逻辑“自行判断”是否使用互联网搜索结果。
-      en_US: The model has a built-in Internet search service. This parameter controls whether the model refers to Internet search results when generating text. When Internet search is enabled, the model will use the search results as reference information in the text generation process, but the model will "judge" whether to use Internet search results based on its internal logic.
   - name: response_format
     use_template: response_format
 pricing:
diff --git a/api/core/model_runtime/model_providers/tongyi/llm/qwen-math-turbo-0919.yaml b/api/core/model_runtime/model_providers/tongyi/llm/qwen-math-turbo-0919.yaml
index 0dc5a066f0300..bffe324a9649c 100644
--- a/api/core/model_runtime/model_providers/tongyi/llm/qwen-math-turbo-0919.yaml
+++ b/api/core/model_runtime/model_providers/tongyi/llm/qwen-math-turbo-0919.yaml
@@ -65,15 +65,6 @@ parameter_rules:
     help:
       zh_Hans: 用于控制模型生成时的重复度。提高repetition_penalty时可以降低模型生成的重复度。1.0表示不做惩罚。
       en_US: Used to control the repeatability when generating models. Increasing repetition_penalty can reduce the duplication of model generation. 1.0 means no punishment.
-  - name: enable_search
-    type: boolean
-    default: false
-    label:
-      zh_Hans: 联网搜索
-      en_US: Web Search
-    help:
-      zh_Hans: 模型内置了互联网搜索服务，该参数控制模型在生成文本时是否参考使用互联网搜索结果。启用互联网搜索，模型会将搜索结果作为文本生成过程中的参考信息，但模型会基于其内部逻辑“自行判断”是否使用互联网搜索结果。
-      en_US: The model has a built-in Internet search service. This parameter controls whether the model refers to Internet search results when generating text. When Internet search is enabled, the model will use the search results as reference information in the text generation process, but the model will "judge" whether to use Internet search results based on its internal logic.
   - name: response_format
     use_template: response_format
 pricing:
diff --git a/api/core/model_runtime/model_providers/tongyi/llm/qwen-math-turbo-latest.yaml b/api/core/model_runtime/model_providers/tongyi/llm/qwen-math-turbo-latest.yaml
index 2ac0e4692a701..0747e96614219 100644
--- a/api/core/model_runtime/model_providers/tongyi/llm/qwen-math-turbo-latest.yaml
+++ b/api/core/model_runtime/model_providers/tongyi/llm/qwen-math-turbo-latest.yaml
@@ -65,15 +65,6 @@ parameter_rules:
     help:
       zh_Hans: 用于控制模型生成时的重复度。提高repetition_penalty时可以降低模型生成的重复度。1.0表示不做惩罚。
       en_US: Used to control the repeatability when generating models. Increasing repetition_penalty can reduce the duplication of model generation. 1.0 means no punishment.
-  - name: enable_search
-    type: boolean
-    default: false
-    label:
-      zh_Hans: 联网搜索
-      en_US: Web Search
-    help:
-      zh_Hans: 模型内置了互联网搜索服务，该参数控制模型在生成文本时是否参考使用互联网搜索结果。启用互联网搜索，模型会将搜索结果作为文本生成过程中的参考信息，但模型会基于其内部逻辑“自行判断”是否使用互联网搜索结果。
-      en_US: The model has a built-in Internet search service. This parameter controls whether the model refers to Internet search results when generating text. When Internet search is enabled, the model will use the search results as reference information in the text generation process, but the model will "judge" whether to use Internet search results based on its internal logic.
   - name: response_format
     use_template: response_format
 pricing:
diff --git a/api/core/model_runtime/model_providers/tongyi/llm/qwen-math-turbo.yaml b/api/core/model_runtime/model_providers/tongyi/llm/qwen-math-turbo.yaml
index 9a7f1312e9724..dffb5557ffb7c 100644
--- a/api/core/model_runtime/model_providers/tongyi/llm/qwen-math-turbo.yaml
+++ b/api/core/model_runtime/model_providers/tongyi/llm/qwen-math-turbo.yaml
@@ -65,15 +65,6 @@ parameter_rules:
     help:
       zh_Hans: 用于控制模型生成时的重复度。提高repetition_penalty时可以降低模型生成的重复度。1.0表示不做惩罚。
       en_US: Used to control the repeatability when generating models. Increasing repetition_penalty can reduce the duplication of model generation. 1.0 means no punishment.
-  - name: enable_search
-    type: boolean
-    default: false
-    label:
-      zh_Hans: 联网搜索
-      en_US: Web Search
-    help:
-      zh_Hans: 模型内置了互联网搜索服务，该参数控制模型在生成文本时是否参考使用互联网搜索结果。启用互联网搜索，模型会将搜索结果作为文本生成过程中的参考信息，但模型会基于其内部逻辑“自行判断”是否使用互联网搜索结果。
-      en_US: The model has a built-in Internet search service. This parameter controls whether the model refers to Internet search results when generating text. When Internet search is enabled, the model will use the search results as reference information in the text generation process, but the model will "judge" whether to use Internet search results based on its internal logic.
   - name: response_format
     use_template: response_format
 pricing:
diff --git a/api/core/model_runtime/model_providers/tongyi/llm/qwen-max-0107.yaml b/api/core/model_runtime/model_providers/tongyi/llm/qwen-max-0107.yaml
index c0eef375570d4..8ae159f1bfb13 100644
--- a/api/core/model_runtime/model_providers/tongyi/llm/qwen-max-0107.yaml
+++ b/api/core/model_runtime/model_providers/tongyi/llm/qwen-max-0107.yaml
@@ -67,15 +67,6 @@ parameter_rules:
     help:
       zh_Hans: 用于控制模型生成时的重复度。提高repetition_penalty时可以降低模型生成的重复度。1.0表示不做惩罚。
       en_US: Used to control the repeatability when generating models. Increasing repetition_penalty can reduce the duplication of model generation. 1.0 means no punishment.
-  - name: enable_search
-    type: boolean
-    default: false
-    label:
-      zh_Hans: 联网搜索
-      en_US: Web Search
-    help:
-      zh_Hans: 模型内置了互联网搜索服务，该参数控制模型在生成文本时是否参考使用互联网搜索结果。启用互联网搜索，模型会将搜索结果作为文本生成过程中的参考信息，但模型会基于其内部逻辑“自行判断”是否使用互联网搜索结果。
-      en_US: The model has a built-in Internet search service. This parameter controls whether the model refers to Internet search results when generating text. When Internet search is enabled, the model will use the search results as reference information in the text generation process, but the model will "judge" whether to use Internet search results based on its internal logic.
   - name: response_format
     use_template: response_format
 pricing:
diff --git a/api/core/model_runtime/model_providers/tongyi/llm/qwen-max-0403.yaml b/api/core/model_runtime/model_providers/tongyi/llm/qwen-max-0403.yaml
index c12444bd7b57b..93fb37254e4c8 100644
--- a/api/core/model_runtime/model_providers/tongyi/llm/qwen-max-0403.yaml
+++ b/api/core/model_runtime/model_providers/tongyi/llm/qwen-max-0403.yaml
@@ -67,15 +67,6 @@ parameter_rules:
     help:
       zh_Hans: 用于控制模型生成时的重复度。提高repetition_penalty时可以降低模型生成的重复度。1.0表示不做惩罚。
       en_US: Used to control the repeatability when generating models. Increasing repetition_penalty can reduce the duplication of model generation. 1.0 means no punishment.
-  - name: enable_search
-    type: boolean
-    default: false
-    label:
-      zh_Hans: 联网搜索
-      en_US: Web Search
-    help:
-      zh_Hans: 模型内置了互联网搜索服务，该参数控制模型在生成文本时是否参考使用互联网搜索结果。启用互联网搜索，模型会将搜索结果作为文本生成过程中的参考信息，但模型会基于其内部逻辑“自行判断”是否使用互联网搜索结果。
-      en_US: The model has a built-in Internet search service. This parameter controls whether the model refers to Internet search results when generating text. When Internet search is enabled, the model will use the search results as reference information in the text generation process, but the model will "judge" whether to use Internet search results based on its internal logic.
   - name: response_format
     use_template: response_format
 pricing:
diff --git a/api/core/model_runtime/model_providers/tongyi/llm/qwen-max-0428.yaml b/api/core/model_runtime/model_providers/tongyi/llm/qwen-max-0428.yaml
index 173c55b6b9796..a5c9d49609041 100644
--- a/api/core/model_runtime/model_providers/tongyi/llm/qwen-max-0428.yaml
+++ b/api/core/model_runtime/model_providers/tongyi/llm/qwen-max-0428.yaml
@@ -67,15 +67,6 @@ parameter_rules:
     help:
       zh_Hans: 用于控制模型生成时的重复度。提高repetition_penalty时可以降低模型生成的重复度。1.0表示不做惩罚。
       en_US: Used to control the repeatability when generating models. Increasing repetition_penalty can reduce the duplication of model generation. 1.0 means no punishment.
-  - name: enable_search
-    type: boolean
-    default: false
-    label:
-      zh_Hans: 联网搜索
-      en_US: Web Search
-    help:
-      zh_Hans: 模型内置了互联网搜索服务，该参数控制模型在生成文本时是否参考使用互联网搜索结果。启用互联网搜索，模型会将搜索结果作为文本生成过程中的参考信息，但模型会基于其内部逻辑“自行判断”是否使用互联网搜索结果。
-      en_US: The model has a built-in Internet search service. This parameter controls whether the model refers to Internet search results when generating text. When Internet search is enabled, the model will use the search results as reference information in the text generation process, but the model will "judge" whether to use Internet search results based on its internal logic.
   - name: response_format
     use_template: response_format
 pricing:
diff --git a/api/core/model_runtime/model_providers/tongyi/llm/qwen-max-0919.yaml b/api/core/model_runtime/model_providers/tongyi/llm/qwen-max-0919.yaml
index 692a38140dbed..e4a6dae63736a 100644
--- a/api/core/model_runtime/model_providers/tongyi/llm/qwen-max-0919.yaml
+++ b/api/core/model_runtime/model_providers/tongyi/llm/qwen-max-0919.yaml
@@ -67,15 +67,6 @@ parameter_rules:
     help:
       zh_Hans: 用于控制模型生成时的重复度。提高repetition_penalty时可以降低模型生成的重复度。1.0表示不做惩罚。
       en_US: Used to control the repeatability when generating models. Increasing repetition_penalty can reduce the duplication of model generation. 1.0 means no punishment.
-  - name: enable_search
-    type: boolean
-    default: false
-    label:
-      zh_Hans: 联网搜索
-      en_US: Web Search
-    help:
-      zh_Hans: 模型内置了互联网搜索服务，该参数控制模型在生成文本时是否参考使用互联网搜索结果。启用互联网搜索，模型会将搜索结果作为文本生成过程中的参考信息，但模型会基于其内部逻辑“自行判断”是否使用互联网搜索结果。
-      en_US: The model has a built-in Internet search service. This parameter controls whether the model refers to Internet search results when generating text. When Internet search is enabled, the model will use the search results as reference information in the text generation process, but the model will "judge" whether to use Internet search results based on its internal logic.
   - name: response_format
     use_template: response_format
 pricing:
diff --git a/api/core/model_runtime/model_providers/tongyi/llm/qwen-max-1201.yaml b/api/core/model_runtime/model_providers/tongyi/llm/qwen-max-1201.yaml
index dc234783cde63..6fae8a7d38f36 100644
--- a/api/core/model_runtime/model_providers/tongyi/llm/qwen-max-1201.yaml
+++ b/api/core/model_runtime/model_providers/tongyi/llm/qwen-max-1201.yaml
@@ -66,12 +66,6 @@ parameter_rules:
     help:
       zh_Hans: 用于控制模型生成时的重复度。提高repetition_penalty时可以降低模型生成的重复度。1.0表示不做惩罚。
       en_US: Used to control the repeatability when generating models. Increasing repetition_penalty can reduce the duplication of model generation. 1.0 means no punishment.
-  - name: enable_search
-    type: boolean
-    default: false
-    help:
-      zh_Hans: 模型内置了互联网搜索服务，该参数控制模型在生成文本时是否参考使用互联网搜索结果。启用互联网搜索，模型会将搜索结果作为文本生成过程中的参考信息，但模型会基于其内部逻辑“自行判断”是否使用互联网搜索结果。
-      en_US: The model has a built-in Internet search service. This parameter controls whether the model refers to Internet search results when generating text. When Internet search is enabled, the model will use the search results as reference information in the text generation process, but the model will "judge" whether to use Internet search results based on its internal logic.
   - name: response_format
     use_template: response_format
 pricing:
diff --git a/api/core/model_runtime/model_providers/tongyi/llm/qwen-max-latest.yaml b/api/core/model_runtime/model_providers/tongyi/llm/qwen-max-latest.yaml
index afd7fb4b77696..8e2096885957e 100644
--- a/api/core/model_runtime/model_providers/tongyi/llm/qwen-max-latest.yaml
+++ b/api/core/model_runtime/model_providers/tongyi/llm/qwen-max-latest.yaml
@@ -67,15 +67,6 @@ parameter_rules:
     help:
       zh_Hans: 用于控制模型生成时的重复度。提高repetition_penalty时可以降低模型生成的重复度。1.0表示不做惩罚。
       en_US: Used to control the repeatability when generating models. Increasing repetition_penalty can reduce the duplication of model generation. 1.0 means no punishment.
-  - name: enable_search
-    type: boolean
-    default: false
-    label:
-      zh_Hans: 联网搜索
-      en_US: Web Search
-    help:
-      zh_Hans: 模型内置了互联网搜索服务，该参数控制模型在生成文本时是否参考使用互联网搜索结果。启用互联网搜索，模型会将搜索结果作为文本生成过程中的参考信息，但模型会基于其内部逻辑“自行判断”是否使用互联网搜索结果。
-      en_US: The model has a built-in Internet search service. This parameter controls whether the model refers to Internet search results when generating text. When Internet search is enabled, the model will use the search results as reference information in the text generation process, but the model will "judge" whether to use Internet search results based on its internal logic.
   - name: response_format
     use_template: response_format
 pricing:
diff --git a/api/core/model_runtime/model_providers/tongyi/llm/qwen-max-longcontext.yaml b/api/core/model_runtime/model_providers/tongyi/llm/qwen-max-longcontext.yaml
index d02ba7af18318..9bc50c73fcc3e 100644
--- a/api/core/model_runtime/model_providers/tongyi/llm/qwen-max-longcontext.yaml
+++ b/api/core/model_runtime/model_providers/tongyi/llm/qwen-max-longcontext.yaml
@@ -67,15 +67,6 @@ parameter_rules:
     help:
       zh_Hans: 用于控制模型生成时的重复度。提高repetition_penalty时可以降低模型生成的重复度。1.0表示不做惩罚。
       en_US: Used to control the repeatability when generating models. Increasing repetition_penalty can reduce the duplication of model generation. 1.0 means no punishment.
-  - name: enable_search
-    type: boolean
-    default: false
-    label:
-      zh_Hans: 联网搜索
-      en_US: Web Search
-    help:
-      zh_Hans: 模型内置了互联网搜索服务，该参数控制模型在生成文本时是否参考使用互联网搜索结果。启用互联网搜索，模型会将搜索结果作为文本生成过程中的参考信息，但模型会基于其内部逻辑“自行判断”是否使用互联网搜索结果。
-      en_US: The model has a built-in Internet search service. This parameter controls whether the model refers to Internet search results when generating text. When Internet search is enabled, the model will use the search results as reference information in the text generation process, but the model will "judge" whether to use Internet search results based on its internal logic.
   - name: response_format
     use_template: response_format
 pricing:
diff --git a/api/core/model_runtime/model_providers/tongyi/llm/qwen-plus-0206.yaml b/api/core/model_runtime/model_providers/tongyi/llm/qwen-plus-0206.yaml
index 1111298c3712a..430599300bccb 100644
--- a/api/core/model_runtime/model_providers/tongyi/llm/qwen-plus-0206.yaml
+++ b/api/core/model_runtime/model_providers/tongyi/llm/qwen-plus-0206.yaml
@@ -65,15 +65,6 @@ parameter_rules:
     help:
       zh_Hans: 用于控制模型生成时的重复度。提高repetition_penalty时可以降低模型生成的重复度。1.0表示不做惩罚。
       en_US: Used to control the repeatability when generating models. Increasing repetition_penalty can reduce the duplication of model generation. 1.0 means no punishment.
-  - name: enable_search
-    type: boolean
-    default: false
-    label:
-      zh_Hans: 联网搜索
-      en_US: Web Search
-    help:
-      zh_Hans: 模型内置了互联网搜索服务，该参数控制模型在生成文本时是否参考使用互联网搜索结果。启用互联网搜索，模型会将搜索结果作为文本生成过程中的参考信息，但模型会基于其内部逻辑“自行判断”是否使用互联网搜索结果。
-      en_US: The model has a built-in Internet search service. This parameter controls whether the model refers to Internet search results when generating text. When Internet search is enabled, the model will use the search results as reference information in the text generation process, but the model will "judge" whether to use Internet search results based on its internal logic.
   - name: response_format
     use_template: response_format
 pricing:
diff --git a/api/core/model_runtime/model_providers/tongyi/llm/qwen-plus-0624.yaml b/api/core/model_runtime/model_providers/tongyi/llm/qwen-plus-0624.yaml
index ef8dd083ade5d..906995d2b95c8 100644
--- a/api/core/model_runtime/model_providers/tongyi/llm/qwen-plus-0624.yaml
+++ b/api/core/model_runtime/model_providers/tongyi/llm/qwen-plus-0624.yaml
@@ -65,15 +65,6 @@ parameter_rules:
     help:
       zh_Hans: 用于控制模型生成时的重复度。提高repetition_penalty时可以降低模型生成的重复度。1.0表示不做惩罚。
       en_US: Used to control the repeatability when generating models. Increasing repetition_penalty can reduce the duplication of model generation. 1.0 means no punishment.
-  - name: enable_search
-    type: boolean
-    default: false
-    label:
-      zh_Hans: 联网搜索
-      en_US: Web Search
-    help:
-      zh_Hans: 模型内置了互联网搜索服务，该参数控制模型在生成文本时是否参考使用互联网搜索结果。启用互联网搜索，模型会将搜索结果作为文本生成过程中的参考信息，但模型会基于其内部逻辑“自行判断”是否使用互联网搜索结果。
-      en_US: The model has a built-in Internet search service. This parameter controls whether the model refers to Internet search results when generating text. When Internet search is enabled, the model will use the search results as reference information in the text generation process, but the model will "judge" whether to use Internet search results based on its internal logic.
   - name: response_format
     use_template: response_format
 pricing:
diff --git a/api/core/model_runtime/model_providers/tongyi/llm/qwen-plus-0723.yaml b/api/core/model_runtime/model_providers/tongyi/llm/qwen-plus-0723.yaml
index 87a4417df59b8..b33e725dd08c3 100644
--- a/api/core/model_runtime/model_providers/tongyi/llm/qwen-plus-0723.yaml
+++ b/api/core/model_runtime/model_providers/tongyi/llm/qwen-plus-0723.yaml
@@ -65,15 +65,6 @@ parameter_rules:
     help:
       zh_Hans: 用于控制模型生成时的重复度。提高repetition_penalty时可以降低模型生成的重复度。1.0表示不做惩罚。
       en_US: Used to control the repeatability when generating models. Increasing repetition_penalty can reduce the duplication of model generation. 1.0 means no punishment.
-  - name: enable_search
-    type: boolean
-    default: false
-    label:
-      zh_Hans: 联网搜索
-      en_US: Web Search
-    help:
-      zh_Hans: 模型内置了互联网搜索服务，该参数控制模型在生成文本时是否参考使用互联网搜索结果。启用互联网搜索，模型会将搜索结果作为文本生成过程中的参考信息，但模型会基于其内部逻辑“自行判断”是否使用互联网搜索结果。
-      en_US: The model has a built-in Internet search service. This parameter controls whether the model refers to Internet search results when generating text. When Internet search is enabled, the model will use the search results as reference information in the text generation process, but the model will "judge" whether to use Internet search results based on its internal logic.
   - name: response_format
     use_template: response_format
 pricing:
diff --git a/api/core/model_runtime/model_providers/tongyi/llm/qwen-plus-0806.yaml b/api/core/model_runtime/model_providers/tongyi/llm/qwen-plus-0806.yaml
index 967f258fa9162..bb394fad81455 100644
--- a/api/core/model_runtime/model_providers/tongyi/llm/qwen-plus-0806.yaml
+++ b/api/core/model_runtime/model_providers/tongyi/llm/qwen-plus-0806.yaml
@@ -65,15 +65,6 @@ parameter_rules:
     help:
       zh_Hans: 用于控制模型生成时的重复度。提高repetition_penalty时可以降低模型生成的重复度。1.0表示不做惩罚。
       en_US: Used to control the repeatability when generating models. Increasing repetition_penalty can reduce the duplication of model generation. 1.0 means no punishment.
-  - name: enable_search
-    type: boolean
-    default: false
-    label:
-      zh_Hans: 联网搜索
-      en_US: Web Search
-    help:
-      zh_Hans: 模型内置了互联网搜索服务，该参数控制模型在生成文本时是否参考使用互联网搜索结果。启用互联网搜索，模型会将搜索结果作为文本生成过程中的参考信息，但模型会基于其内部逻辑“自行判断”是否使用互联网搜索结果。
-      en_US: The model has a built-in Internet search service. This parameter controls whether the model refers to Internet search results when generating text. When Internet search is enabled, the model will use the search results as reference information in the text generation process, but the model will "judge" whether to use Internet search results based on its internal logic.
   - name: response_format
     use_template: response_format
 pricing:
diff --git a/api/core/model_runtime/model_providers/tongyi/llm/qwen-plus-0919.yaml b/api/core/model_runtime/model_providers/tongyi/llm/qwen-plus-0919.yaml
index 9d44852ac9cd5..118e304a9723a 100644
--- a/api/core/model_runtime/model_providers/tongyi/llm/qwen-plus-0919.yaml
+++ b/api/core/model_runtime/model_providers/tongyi/llm/qwen-plus-0919.yaml
@@ -65,15 +65,6 @@ parameter_rules:
     help:
       zh_Hans: 用于控制模型生成时的重复度。提高repetition_penalty时可以降低模型生成的重复度。1.0表示不做惩罚。
       en_US: Used to control the repeatability when generating models. Increasing repetition_penalty can reduce the duplication of model generation. 1.0 means no punishment.
-  - name: enable_search
-    type: boolean
-    default: false
-    label:
-      zh_Hans: 联网搜索
-      en_US: Web Search
-    help:
-      zh_Hans: 模型内置了互联网搜索服务，该参数控制模型在生成文本时是否参考使用互联网搜索结果。启用互联网搜索，模型会将搜索结果作为文本生成过程中的参考信息，但模型会基于其内部逻辑“自行判断”是否使用互联网搜索结果。
-      en_US: The model has a built-in Internet search service. This parameter controls whether the model refers to Internet search results when generating text. When Internet search is enabled, the model will use the search results as reference information in the text generation process, but the model will "judge" whether to use Internet search results based on its internal logic.
   - name: response_format
     use_template: response_format
 pricing:
diff --git a/api/core/model_runtime/model_providers/tongyi/llm/qwen-plus-chat.yaml b/api/core/model_runtime/model_providers/tongyi/llm/qwen-plus-chat.yaml
index df9448ae04ddf..761312bc38123 100644
--- a/api/core/model_runtime/model_providers/tongyi/llm/qwen-plus-chat.yaml
+++ b/api/core/model_runtime/model_providers/tongyi/llm/qwen-plus-chat.yaml
@@ -67,15 +67,6 @@ parameter_rules:
     help:
       zh_Hans: 用于控制模型生成时的重复度。提高repetition_penalty时可以降低模型生成的重复度。1.0表示不做惩罚。
       en_US: Used to control the repeatability when generating models. Increasing repetition_penalty can reduce the duplication of model generation. 1.0 means no punishment.
-  - name: enable_search
-    type: boolean
-    default: false
-    label:
-      zh_Hans: 联网搜索
-      en_US: Web Search
-    help:
-      zh_Hans: 模型内置了互联网搜索服务，该参数控制模型在生成文本时是否参考使用互联网搜索结果。启用互联网搜索，模型会将搜索结果作为文本生成过程中的参考信息，但模型会基于其内部逻辑“自行判断”是否使用互联网搜索结果。
-      en_US: The model has a built-in Internet search service. This parameter controls whether the model refers to Internet search results when generating text. When Internet search is enabled, the model will use the search results as reference information in the text generation process, but the model will "judge" whether to use Internet search results based on its internal logic.
   - name: response_format
     use_template: response_format
 pricing:
diff --git a/api/core/model_runtime/model_providers/tongyi/llm/qwen-plus-latest.yaml b/api/core/model_runtime/model_providers/tongyi/llm/qwen-plus-latest.yaml
index 32ccb8d615903..430872fb31a44 100644
--- a/api/core/model_runtime/model_providers/tongyi/llm/qwen-plus-latest.yaml
+++ b/api/core/model_runtime/model_providers/tongyi/llm/qwen-plus-latest.yaml
@@ -65,15 +65,6 @@ parameter_rules:
     help:
       zh_Hans: 用于控制模型生成时的重复度。提高repetition_penalty时可以降低模型生成的重复度。1.0表示不做惩罚。
       en_US: Used to control the repeatability when generating models. Increasing repetition_penalty can reduce the duplication of model generation. 1.0 means no punishment.
-  - name: enable_search
-    type: boolean
-    default: false
-    label:
-      zh_Hans: 联网搜索
-      en_US: Web Search
-    help:
-      zh_Hans: 模型内置了互联网搜索服务，该参数控制模型在生成文本时是否参考使用互联网搜索结果。启用互联网搜索，模型会将搜索结果作为文本生成过程中的参考信息，但模型会基于其内部逻辑“自行判断”是否使用互联网搜索结果。
-      en_US: The model has a built-in Internet search service. This parameter controls whether the model refers to Internet search results when generating text. When Internet search is enabled, the model will use the search results as reference information in the text generation process, but the model will "judge" whether to use Internet search results based on its internal logic.
   - name: response_format
     use_template: response_format
 pricing:
diff --git a/api/core/model_runtime/model_providers/tongyi/llm/qwen-turbo-0206.yaml b/api/core/model_runtime/model_providers/tongyi/llm/qwen-turbo-0206.yaml
index bf976b518a911..2628d824fe824 100644
--- a/api/core/model_runtime/model_providers/tongyi/llm/qwen-turbo-0206.yaml
+++ b/api/core/model_runtime/model_providers/tongyi/llm/qwen-turbo-0206.yaml
@@ -65,15 +65,6 @@ parameter_rules:
     help:
       zh_Hans: 用于控制模型生成时的重复度。提高repetition_penalty时可以降低模型生成的重复度。1.0表示不做惩罚。
       en_US: Used to control the repeatability when generating models. Increasing repetition_penalty can reduce the duplication of model generation. 1.0 means no punishment.
-  - name: enable_search
-    type: boolean
-    default: false
-    label:
-      zh_Hans: 联网搜索
-      en_US: Web Search
-    help:
-      zh_Hans: 模型内置了互联网搜索服务，该参数控制模型在生成文本时是否参考使用互联网搜索结果。启用互联网搜索，模型会将搜索结果作为文本生成过程中的参考信息，但模型会基于其内部逻辑“自行判断”是否使用互联网搜索结果。
-      en_US: The model has a built-in Internet search service. This parameter controls whether the model refers to Internet search results when generating text. When Internet search is enabled, the model will use the search results as reference information in the text generation process, but the model will "judge" whether to use Internet search results based on its internal logic.
   - name: response_format
     use_template: response_format
 pricing:
diff --git a/api/core/model_runtime/model_providers/tongyi/llm/qwen-turbo-0624.yaml b/api/core/model_runtime/model_providers/tongyi/llm/qwen-turbo-0624.yaml
index 060e7fb4c9d09..8097459bf0906 100644
--- a/api/core/model_runtime/model_providers/tongyi/llm/qwen-turbo-0624.yaml
+++ b/api/core/model_runtime/model_providers/tongyi/llm/qwen-turbo-0624.yaml
@@ -65,15 +65,6 @@ parameter_rules:
     help:
       zh_Hans: 用于控制模型生成时的重复度。提高repetition_penalty时可以降低模型生成的重复度。1.0表示不做惩罚。
       en_US: Used to control the repeatability when generating models. Increasing repetition_penalty can reduce the duplication of model generation. 1.0 means no punishment.
-  - name: enable_search
-    type: boolean
-    default: false
-    label:
-      zh_Hans: 联网搜索
-      en_US: Web Search
-    help:
-      zh_Hans: 模型内置了互联网搜索服务，该参数控制模型在生成文本时是否参考使用互联网搜索结果。启用互联网搜索，模型会将搜索结果作为文本生成过程中的参考信息，但模型会基于其内部逻辑“自行判断”是否使用互联网搜索结果。
-      en_US: The model has a built-in Internet search service. This parameter controls whether the model refers to Internet search results when generating text. When Internet search is enabled, the model will use the search results as reference information in the text generation process, but the model will "judge" whether to use Internet search results based on its internal logic.
   - name: response_format
     use_template: response_format
 pricing:
diff --git a/api/core/model_runtime/model_providers/tongyi/llm/qwen-turbo-0919.yaml b/api/core/model_runtime/model_providers/tongyi/llm/qwen-turbo-0919.yaml
index 97cd34929b4f4..e43beeb195aec 100644
--- a/api/core/model_runtime/model_providers/tongyi/llm/qwen-turbo-0919.yaml
+++ b/api/core/model_runtime/model_providers/tongyi/llm/qwen-turbo-0919.yaml
@@ -65,15 +65,6 @@ parameter_rules:
     help:
       zh_Hans: 用于控制模型生成时的重复度。提高repetition_penalty时可以降低模型生成的重复度。1.0表示不做惩罚。
       en_US: Used to control the repeatability when generating models. Increasing repetition_penalty can reduce the duplication of model generation. 1.0 means no punishment.
-  - name: enable_search
-    type: boolean
-    default: false
-    label:
-      zh_Hans: 联网搜索
-      en_US: Web Search
-    help:
-      zh_Hans: 模型内置了互联网搜索服务，该参数控制模型在生成文本时是否参考使用互联网搜索结果。启用互联网搜索，模型会将搜索结果作为文本生成过程中的参考信息，但模型会基于其内部逻辑“自行判断”是否使用互联网搜索结果。
-      en_US: The model has a built-in Internet search service. This parameter controls whether the model refers to Internet search results when generating text. When Internet search is enabled, the model will use the search results as reference information in the text generation process, but the model will "judge" whether to use Internet search results based on its internal logic.
   - name: response_format
     use_template: response_format
 pricing:
diff --git a/api/core/model_runtime/model_providers/tongyi/llm/qwen-turbo-chat.yaml b/api/core/model_runtime/model_providers/tongyi/llm/qwen-turbo-chat.yaml
index 8d77ba7a2a698..c30cb7ca10096 100644
--- a/api/core/model_runtime/model_providers/tongyi/llm/qwen-turbo-chat.yaml
+++ b/api/core/model_runtime/model_providers/tongyi/llm/qwen-turbo-chat.yaml
@@ -67,15 +67,6 @@ parameter_rules:
     help:
       zh_Hans: 用于控制模型生成时的重复度。提高repetition_penalty时可以降低模型生成的重复度。1.0表示不做惩罚。
       en_US: Used to control the repeatability when generating models. Increasing repetition_penalty can reduce the duplication of model generation. 1.0 means no punishment.
-  - name: enable_search
-    type: boolean
-    default: false
-    label:
-      zh_Hans: 联网搜索
-      en_US: Web Search
-    help:
-      zh_Hans: 模型内置了互联网搜索服务，该参数控制模型在生成文本时是否参考使用互联网搜索结果。启用互联网搜索，模型会将搜索结果作为文本生成过程中的参考信息，但模型会基于其内部逻辑“自行判断”是否使用互联网搜索结果。
-      en_US: The model has a built-in Internet search service. This parameter controls whether the model refers to Internet search results when generating text. When Internet search is enabled, the model will use the search results as reference information in the text generation process, but the model will "judge" whether to use Internet search results based on its internal logic.
   - name: response_format
     use_template: response_format
 pricing:
diff --git a/api/core/model_runtime/model_providers/tongyi/llm/qwen-turbo-latest.yaml b/api/core/model_runtime/model_providers/tongyi/llm/qwen-turbo-latest.yaml
index 4458c706aad40..e443d6888b7ef 100644
--- a/api/core/model_runtime/model_providers/tongyi/llm/qwen-turbo-latest.yaml
+++ b/api/core/model_runtime/model_providers/tongyi/llm/qwen-turbo-latest.yaml
@@ -65,15 +65,6 @@ parameter_rules:
     help:
       zh_Hans: 用于控制模型生成时的重复度。提高repetition_penalty时可以降低模型生成的重复度。1.0表示不做惩罚。
       en_US: Used to control the repeatability when generating models. Increasing repetition_penalty can reduce the duplication of model generation. 1.0 means no punishment.
-  - name: enable_search
-    type: boolean
-    default: false
-    label:
-      zh_Hans: 联网搜索
-      en_US: Web Search
-    help:
-      zh_Hans: 模型内置了互联网搜索服务，该参数控制模型在生成文本时是否参考使用互联网搜索结果。启用互联网搜索，模型会将搜索结果作为文本生成过程中的参考信息，但模型会基于其内部逻辑“自行判断”是否使用互联网搜索结果。
-      en_US: The model has a built-in Internet search service. This parameter controls whether the model refers to Internet search results when generating text. When Internet search is enabled, the model will use the search results as reference information in the text generation process, but the model will "judge" whether to use Internet search results based on its internal logic.
   - name: response_format
     use_template: response_format
 pricing:
diff --git a/api/core/model_runtime/model_providers/tongyi/llm/qwen-vl-max-0809.yaml b/api/core/model_runtime/model_providers/tongyi/llm/qwen-vl-max-0809.yaml
index 12e9e0dd569f9..fd20377002042 100644
--- a/api/core/model_runtime/model_providers/tongyi/llm/qwen-vl-max-0809.yaml
+++ b/api/core/model_runtime/model_providers/tongyi/llm/qwen-vl-max-0809.yaml
@@ -69,15 +69,6 @@ parameter_rules:
     help:
       zh_Hans: 用于控制模型生成时的重复度。提高repetition_penalty时可以降低模型生成的重复度。1.0表示不做惩罚。
       en_US: Used to control the repeatability when generating models. Increasing repetition_penalty can reduce the duplication of model generation. 1.0 means no punishment.
-  - name: enable_search
-    type: boolean
-    default: false
-    label:
-      zh_Hans: 联网搜索
-      en_US: Web Search
-    help:
-      zh_Hans: 模型内置了互联网搜索服务，该参数控制模型在生成文本时是否参考使用互联网搜索结果。启用互联网搜索，模型会将搜索结果作为文本生成过程中的参考信息，但模型会基于其内部逻辑“自行判断”是否使用互联网搜索结果。
-      en_US: The model has a built-in Internet search service. This parameter controls whether the model refers to Internet search results when generating text. When Internet search is enabled, the model will use the search results as reference information in the text generation process, but the model will "judge" whether to use Internet search results based on its internal logic.
   - name: response_format
     use_template: response_format
 pricing:
diff --git a/api/core/model_runtime/model_providers/tongyi/llm/qwen-vl-max.yaml b/api/core/model_runtime/model_providers/tongyi/llm/qwen-vl-max.yaml
index b811fdece48f5..31a9fb51bbbb5 100644
--- a/api/core/model_runtime/model_providers/tongyi/llm/qwen-vl-max.yaml
+++ b/api/core/model_runtime/model_providers/tongyi/llm/qwen-vl-max.yaml
@@ -69,15 +69,6 @@ parameter_rules:
     help:
       zh_Hans: 用于控制模型生成时的重复度。提高repetition_penalty时可以降低模型生成的重复度。1.0表示不做惩罚。
       en_US: Used to control the repeatability when generating models. Increasing repetition_penalty can reduce the duplication of model generation. 1.0 means no punishment.
-  - name: enable_search
-    type: boolean
-    default: false
-    label:
-      zh_Hans: 联网搜索
-      en_US: Web Search
-    help:
-      zh_Hans: 模型内置了互联网搜索服务，该参数控制模型在生成文本时是否参考使用互联网搜索结果。启用互联网搜索，模型会将搜索结果作为文本生成过程中的参考信息，但模型会基于其内部逻辑“自行判断”是否使用互联网搜索结果。
-      en_US: The model has a built-in Internet search service. This parameter controls whether the model refers to Internet search results when generating text. When Internet search is enabled, the model will use the search results as reference information in the text generation process, but the model will "judge" whether to use Internet search results based on its internal logic.
   - name: response_format
     use_template: response_format
 pricing:
diff --git a/api/core/model_runtime/model_providers/tongyi/llm/qwen-vl-plus-0201.yaml b/api/core/model_runtime/model_providers/tongyi/llm/qwen-vl-plus-0201.yaml
index 188dea389a7e2..5f90cf48bc999 100644
--- a/api/core/model_runtime/model_providers/tongyi/llm/qwen-vl-plus-0201.yaml
+++ b/api/core/model_runtime/model_providers/tongyi/llm/qwen-vl-plus-0201.yaml
@@ -69,15 +69,6 @@ parameter_rules:
     help:
       zh_Hans: 用于控制模型生成时的重复度。提高repetition_penalty时可以降低模型生成的重复度。1.0表示不做惩罚。
       en_US: Used to control the repeatability when generating models. Increasing repetition_penalty can reduce the duplication of model generation. 1.0 means no punishment.
-  - name: enable_search
-    type: boolean
-    default: false
-    label:
-      zh_Hans: 联网搜索
-      en_US: Web Search
-    help:
-      zh_Hans: 模型内置了互联网搜索服务，该参数控制模型在生成文本时是否参考使用互联网搜索结果。启用互联网搜索，模型会将搜索结果作为文本生成过程中的参考信息，但模型会基于其内部逻辑“自行判断”是否使用互联网搜索结果。
-      en_US: The model has a built-in Internet search service. This parameter controls whether the model refers to Internet search results when generating text. When Internet search is enabled, the model will use the search results as reference information in the text generation process, but the model will "judge" whether to use Internet search results based on its internal logic.
   - name: response_format
     use_template: response_format
 pricing:
diff --git a/api/core/model_runtime/model_providers/tongyi/llm/qwen-vl-plus-0809.yaml b/api/core/model_runtime/model_providers/tongyi/llm/qwen-vl-plus-0809.yaml
index bc623e2f03211..97820c0f3a1a2 100644
--- a/api/core/model_runtime/model_providers/tongyi/llm/qwen-vl-plus-0809.yaml
+++ b/api/core/model_runtime/model_providers/tongyi/llm/qwen-vl-plus-0809.yaml
@@ -69,15 +69,6 @@ parameter_rules:
     help:
       zh_Hans: 用于控制模型生成时的重复度。提高repetition_penalty时可以降低模型生成的重复度。1.0表示不做惩罚。
       en_US: Used to control the repeatability when generating models. Increasing repetition_penalty can reduce the duplication of model generation. 1.0 means no punishment.
-  - name: enable_search
-    type: boolean
-    default: false
-    label:
-      zh_Hans: 联网搜索
-      en_US: Web Search
-    help:
-      zh_Hans: 模型内置了互联网搜索服务，该参数控制模型在生成文本时是否参考使用互联网搜索结果。启用互联网搜索，模型会将搜索结果作为文本生成过程中的参考信息，但模型会基于其内部逻辑“自行判断”是否使用互联网搜索结果。
-      en_US: The model has a built-in Internet search service. This parameter controls whether the model refers to Internet search results when generating text. When Internet search is enabled, the model will use the search results as reference information in the text generation process, but the model will "judge" whether to use Internet search results based on its internal logic.
   - name: response_format
     use_template: response_format
 pricing:
diff --git a/api/core/model_runtime/model_providers/tongyi/llm/qwen-vl-plus.yaml b/api/core/model_runtime/model_providers/tongyi/llm/qwen-vl-plus.yaml
index 8977e12e4f2d9..6af36cd6f3d63 100644
--- a/api/core/model_runtime/model_providers/tongyi/llm/qwen-vl-plus.yaml
+++ b/api/core/model_runtime/model_providers/tongyi/llm/qwen-vl-plus.yaml
@@ -69,15 +69,6 @@ parameter_rules:
     help:
       zh_Hans: 用于控制模型生成时的重复度。提高repetition_penalty时可以降低模型生成的重复度。1.0表示不做惩罚。
       en_US: Used to control the repeatability when generating models. Increasing repetition_penalty can reduce the duplication of model generation. 1.0 means no punishment.
-  - name: enable_search
-    type: boolean
-    default: false
-    label:
-      zh_Hans: 联网搜索
-      en_US: Web Search
-    help:
-      zh_Hans: 模型内置了互联网搜索服务，该参数控制模型在生成文本时是否参考使用互联网搜索结果。启用互联网搜索，模型会将搜索结果作为文本生成过程中的参考信息，但模型会基于其内部逻辑“自行判断”是否使用互联网搜索结果。
-      en_US: The model has a built-in Internet search service. This parameter controls whether the model refers to Internet search results when generating text. When Internet search is enabled, the model will use the search results as reference information in the text generation process, but the model will "judge" whether to use Internet search results based on its internal logic.
   - name: response_format
     use_template: response_format
 pricing:
diff --git a/api/core/model_runtime/model_providers/tongyi/llm/qwen2-math-1.5b-instruct.yaml b/api/core/model_runtime/model_providers/tongyi/llm/qwen2-math-1.5b-instruct.yaml
index de237842af7d8..158e2c7ee16e5 100644
--- a/api/core/model_runtime/model_providers/tongyi/llm/qwen2-math-1.5b-instruct.yaml
+++ b/api/core/model_runtime/model_providers/tongyi/llm/qwen2-math-1.5b-instruct.yaml
@@ -65,15 +65,6 @@ parameter_rules:
     help:
       zh_Hans: 用于控制模型生成时的重复度。提高repetition_penalty时可以降低模型生成的重复度。1.0表示不做惩罚。
       en_US: Used to control the repeatability when generating models. Increasing repetition_penalty can reduce the duplication of model generation. 1.0 means no punishment.
-  - name: enable_search
-    type: boolean
-    default: false
-    label:
-      zh_Hans: 联网搜索
-      en_US: Web Search
-    help:
-      zh_Hans: 模型内置了互联网搜索服务，该参数控制模型在生成文本时是否参考使用互联网搜索结果。启用互联网搜索，模型会将搜索结果作为文本生成过程中的参考信息，但模型会基于其内部逻辑“自行判断”是否使用互联网搜索结果。
-      en_US: The model has a built-in Internet search service. This parameter controls whether the model refers to Internet search results when generating text. When Internet search is enabled, the model will use the search results as reference information in the text generation process, but the model will "judge" whether to use Internet search results based on its internal logic.
   - name: response_format
     use_template: response_format
 pricing:
diff --git a/api/core/model_runtime/model_providers/tongyi/llm/qwen2-math-72b-instruct.yaml b/api/core/model_runtime/model_providers/tongyi/llm/qwen2-math-72b-instruct.yaml
index 1fda35abaf1fe..e26a6923d12f2 100644
--- a/api/core/model_runtime/model_providers/tongyi/llm/qwen2-math-72b-instruct.yaml
+++ b/api/core/model_runtime/model_providers/tongyi/llm/qwen2-math-72b-instruct.yaml
@@ -65,15 +65,6 @@ parameter_rules:
     help:
       zh_Hans: 用于控制模型生成时的重复度。提高repetition_penalty时可以降低模型生成的重复度。1.0表示不做惩罚。
       en_US: Used to control the repeatability when generating models. Increasing repetition_penalty can reduce the duplication of model generation. 1.0 means no punishment.
-  - name: enable_search
-    type: boolean
-    default: false
-    label:
-      zh_Hans: 联网搜索
-      en_US: Web Search
-    help:
-      zh_Hans: 模型内置了互联网搜索服务，该参数控制模型在生成文本时是否参考使用互联网搜索结果。启用互联网搜索，模型会将搜索结果作为文本生成过程中的参考信息，但模型会基于其内部逻辑“自行判断”是否使用互联网搜索结果。
-      en_US: The model has a built-in Internet search service. This parameter controls whether the model refers to Internet search results when generating text. When Internet search is enabled, the model will use the search results as reference information in the text generation process, but the model will "judge" whether to use Internet search results based on its internal logic.
   - name: response_format
     use_template: response_format
 pricing:
diff --git a/api/core/model_runtime/model_providers/tongyi/llm/qwen2-math-7b-instruct.yaml b/api/core/model_runtime/model_providers/tongyi/llm/qwen2-math-7b-instruct.yaml
index 06fd33c5f4740..589119b26ea3c 100644
--- a/api/core/model_runtime/model_providers/tongyi/llm/qwen2-math-7b-instruct.yaml
+++ b/api/core/model_runtime/model_providers/tongyi/llm/qwen2-math-7b-instruct.yaml
@@ -65,15 +65,6 @@ parameter_rules:
     help:
       zh_Hans: 用于控制模型生成时的重复度。提高repetition_penalty时可以降低模型生成的重复度。1.0表示不做惩罚。
       en_US: Used to control the repeatability when generating models. Increasing repetition_penalty can reduce the duplication of model generation. 1.0 means no punishment.
-  - name: enable_search
-    type: boolean
-    default: false
-    label:
-      zh_Hans: 联网搜索
-      en_US: Web Search
-    help:
-      zh_Hans: 模型内置了互联网搜索服务，该参数控制模型在生成文本时是否参考使用互联网搜索结果。启用互联网搜索，模型会将搜索结果作为文本生成过程中的参考信息，但模型会基于其内部逻辑“自行判断”是否使用互联网搜索结果。
-      en_US: The model has a built-in Internet search service. This parameter controls whether the model refers to Internet search results when generating text. When Internet search is enabled, the model will use the search results as reference information in the text generation process, but the model will "judge" whether to use Internet search results based on its internal logic.
   - name: response_format
     use_template: response_format
 pricing:
diff --git a/api/core/model_runtime/model_providers/tongyi/llm/qwen2.5-0.5b-instruct.yaml b/api/core/model_runtime/model_providers/tongyi/llm/qwen2.5-0.5b-instruct.yaml
index ebf809955337b..dd608fbf7675a 100644
--- a/api/core/model_runtime/model_providers/tongyi/llm/qwen2.5-0.5b-instruct.yaml
+++ b/api/core/model_runtime/model_providers/tongyi/llm/qwen2.5-0.5b-instruct.yaml
@@ -65,15 +65,6 @@ parameter_rules:
     help:
       zh_Hans: 用于控制模型生成时的重复度。提高repetition_penalty时可以降低模型生成的重复度。1.0表示不做惩罚。
       en_US: Used to control the repeatability when generating models. Increasing repetition_penalty can reduce the duplication of model generation. 1.0 means no punishment.
-  - name: enable_search
-    type: boolean
-    default: false
-    label:
-      zh_Hans: 联网搜索
-      en_US: Web Search
-    help:
-      zh_Hans: 模型内置了互联网搜索服务，该参数控制模型在生成文本时是否参考使用互联网搜索结果。启用互联网搜索，模型会将搜索结果作为文本生成过程中的参考信息，但模型会基于其内部逻辑“自行判断”是否使用互联网搜索结果。
-      en_US: The model has a built-in Internet search service. This parameter controls whether the model refers to Internet search results when generating text. When Internet search is enabled, the model will use the search results as reference information in the text generation process, but the model will "judge" whether to use Internet search results based on its internal logic.
   - name: response_format
     use_template: response_format
 pricing:
diff --git a/api/core/model_runtime/model_providers/tongyi/llm/qwen2.5-1.5b-instruct.yaml b/api/core/model_runtime/model_providers/tongyi/llm/qwen2.5-1.5b-instruct.yaml
index e9bc99339de0b..08237b3958b32 100644
--- a/api/core/model_runtime/model_providers/tongyi/llm/qwen2.5-1.5b-instruct.yaml
+++ b/api/core/model_runtime/model_providers/tongyi/llm/qwen2.5-1.5b-instruct.yaml
@@ -65,15 +65,6 @@ parameter_rules:
     help:
       zh_Hans: 用于控制模型生成时的重复度。提高repetition_penalty时可以降低模型生成的重复度。1.0表示不做惩罚。
       en_US: Used to control the repeatability when generating models. Increasing repetition_penalty can reduce the duplication of model generation. 1.0 means no punishment.
-  - name: enable_search
-    type: boolean
-    default: false
-    label:
-      zh_Hans: 联网搜索
-      en_US: Web Search
-    help:
-      zh_Hans: 模型内置了互联网搜索服务，该参数控制模型在生成文本时是否参考使用互联网搜索结果。启用互联网搜索，模型会将搜索结果作为文本生成过程中的参考信息，但模型会基于其内部逻辑“自行判断”是否使用互联网搜索结果。
-      en_US: The model has a built-in Internet search service. This parameter controls whether the model refers to Internet search results when generating text. When Internet search is enabled, the model will use the search results as reference information in the text generation process, but the model will "judge" whether to use Internet search results based on its internal logic.
   - name: response_format
     use_template: response_format
 pricing:
diff --git a/api/core/model_runtime/model_providers/tongyi/llm/qwen2.5-14b-instruct.yaml b/api/core/model_runtime/model_providers/tongyi/llm/qwen2.5-14b-instruct.yaml
index 3ed85dade8281..640b01970351b 100644
--- a/api/core/model_runtime/model_providers/tongyi/llm/qwen2.5-14b-instruct.yaml
+++ b/api/core/model_runtime/model_providers/tongyi/llm/qwen2.5-14b-instruct.yaml
@@ -65,15 +65,6 @@ parameter_rules:
     help:
       zh_Hans: 用于控制模型生成时的重复度。提高repetition_penalty时可以降低模型生成的重复度。1.0表示不做惩罚。
       en_US: Used to control the repeatability when generating models. Increasing repetition_penalty can reduce the duplication of model generation. 1.0 means no punishment.
-  - name: enable_search
-    type: boolean
-    default: false
-    label:
-      zh_Hans: 联网搜索
-      en_US: Web Search
-    help:
-      zh_Hans: 模型内置了互联网搜索服务，该参数控制模型在生成文本时是否参考使用互联网搜索结果。启用互联网搜索，模型会将搜索结果作为文本生成过程中的参考信息，但模型会基于其内部逻辑“自行判断”是否使用互联网搜索结果。
-      en_US: The model has a built-in Internet search service. This parameter controls whether the model refers to Internet search results when generating text. When Internet search is enabled, the model will use the search results as reference information in the text generation process, but the model will "judge" whether to use Internet search results based on its internal logic.
   - name: response_format
     use_template: response_format
 pricing:
diff --git a/api/core/model_runtime/model_providers/tongyi/llm/qwen2.5-32b-instruct.yaml b/api/core/model_runtime/model_providers/tongyi/llm/qwen2.5-32b-instruct.yaml
index 328519c16822f..3a90ca753273d 100644
--- a/api/core/model_runtime/model_providers/tongyi/llm/qwen2.5-32b-instruct.yaml
+++ b/api/core/model_runtime/model_providers/tongyi/llm/qwen2.5-32b-instruct.yaml
@@ -65,15 +65,6 @@ parameter_rules:
     help:
       zh_Hans: 用于控制模型生成时的重复度。提高repetition_penalty时可以降低模型生成的重复度。1.0表示不做惩罚。
       en_US: Used to control the repeatability when generating models. Increasing repetition_penalty can reduce the duplication of model generation. 1.0 means no punishment.
-  - name: enable_search
-    type: boolean
-    default: false
-    label:
-      zh_Hans: 联网搜索
-      en_US: Web Search
-    help:
-      zh_Hans: 模型内置了互联网搜索服务，该参数控制模型在生成文本时是否参考使用互联网搜索结果。启用互联网搜索，模型会将搜索结果作为文本生成过程中的参考信息，但模型会基于其内部逻辑“自行判断”是否使用互联网搜索结果。
-      en_US: The model has a built-in Internet search service. This parameter controls whether the model refers to Internet search results when generating text. When Internet search is enabled, the model will use the search results as reference information in the text generation process, but the model will "judge" whether to use Internet search results based on its internal logic.
   - name: response_format
     use_template: response_format
 pricing:
diff --git a/api/core/model_runtime/model_providers/tongyi/llm/qwen2.5-3b-instruct.yaml b/api/core/model_runtime/model_providers/tongyi/llm/qwen2.5-3b-instruct.yaml
index d1ed3c2a73edf..b79755eb9bd84 100644
--- a/api/core/model_runtime/model_providers/tongyi/llm/qwen2.5-3b-instruct.yaml
+++ b/api/core/model_runtime/model_providers/tongyi/llm/qwen2.5-3b-instruct.yaml
@@ -65,15 +65,6 @@ parameter_rules:
     help:
       zh_Hans: 用于控制模型生成时的重复度。提高repetition_penalty时可以降低模型生成的重复度。1.0表示不做惩罚。
       en_US: Used to control the repeatability when generating models. Increasing repetition_penalty can reduce the duplication of model generation. 1.0 means no punishment.
-  - name: enable_search
-    type: boolean
-    default: false
-    label:
-      zh_Hans: 联网搜索
-      en_US: Web Search
-    help:
-      zh_Hans: 模型内置了互联网搜索服务，该参数控制模型在生成文本时是否参考使用互联网搜索结果。启用互联网搜索，模型会将搜索结果作为文本生成过程中的参考信息，但模型会基于其内部逻辑“自行判断”是否使用互联网搜索结果。
-      en_US: The model has a built-in Internet search service. This parameter controls whether the model refers to Internet search results when generating text. When Internet search is enabled, the model will use the search results as reference information in the text generation process, but the model will "judge" whether to use Internet search results based on its internal logic.
   - name: response_format
     use_template: response_format
 pricing:
diff --git a/api/core/model_runtime/model_providers/tongyi/llm/qwen2.5-72b-instruct.yaml b/api/core/model_runtime/model_providers/tongyi/llm/qwen2.5-72b-instruct.yaml
index 0e88c24aa8347..e9dd51a341624 100644
--- a/api/core/model_runtime/model_providers/tongyi/llm/qwen2.5-72b-instruct.yaml
+++ b/api/core/model_runtime/model_providers/tongyi/llm/qwen2.5-72b-instruct.yaml
@@ -65,15 +65,6 @@ parameter_rules:
     help:
       zh_Hans: 用于控制模型生成时的重复度。提高repetition_penalty时可以降低模型生成的重复度。1.0表示不做惩罚。
       en_US: Used to control the repeatability when generating models. Increasing repetition_penalty can reduce the duplication of model generation. 1.0 means no punishment.
-  - name: enable_search
-    type: boolean
-    default: false
-    label:
-      zh_Hans: 联网搜索
-      en_US: Web Search
-    help:
-      zh_Hans: 模型内置了互联网搜索服务，该参数控制模型在生成文本时是否参考使用互联网搜索结果。启用互联网搜索，模型会将搜索结果作为文本生成过程中的参考信息，但模型会基于其内部逻辑“自行判断”是否使用互联网搜索结果。
-      en_US: The model has a built-in Internet search service. This parameter controls whether the model refers to Internet search results when generating text. When Internet search is enabled, the model will use the search results as reference information in the text generation process, but the model will "judge" whether to use Internet search results based on its internal logic.
   - name: response_format
     use_template: response_format
 pricing:
diff --git a/api/core/model_runtime/model_providers/tongyi/llm/qwen2.5-7b-instruct.yaml b/api/core/model_runtime/model_providers/tongyi/llm/qwen2.5-7b-instruct.yaml
index 35313cd1f7094..04f26cf5fe12b 100644
--- a/api/core/model_runtime/model_providers/tongyi/llm/qwen2.5-7b-instruct.yaml
+++ b/api/core/model_runtime/model_providers/tongyi/llm/qwen2.5-7b-instruct.yaml
@@ -65,15 +65,6 @@ parameter_rules:
     help:
       zh_Hans: 用于控制模型生成时的重复度。提高repetition_penalty时可以降低模型生成的重复度。1.0表示不做惩罚。
       en_US: Used to control the repeatability when generating models. Increasing repetition_penalty can reduce the duplication of model generation. 1.0 means no punishment.
-  - name: enable_search
-    type: boolean
-    default: false
-    label:
-      zh_Hans: 联网搜索
-      en_US: Web Search
-    help:
-      zh_Hans: 模型内置了互联网搜索服务，该参数控制模型在生成文本时是否参考使用互联网搜索结果。启用互联网搜索，模型会将搜索结果作为文本生成过程中的参考信息，但模型会基于其内部逻辑“自行判断”是否使用互联网搜索结果。
-      en_US: The model has a built-in Internet search service. This parameter controls whether the model refers to Internet search results when generating text. When Internet search is enabled, the model will use the search results as reference information in the text generation process, but the model will "judge" whether to use Internet search results based on its internal logic.
   - name: response_format
     use_template: response_format
 pricing:
diff --git a/api/core/model_runtime/model_providers/tongyi/llm/qwen2.5-coder-7b-instruct.yaml b/api/core/model_runtime/model_providers/tongyi/llm/qwen2.5-coder-7b-instruct.yaml
index 35313cd1f7094..04f26cf5fe12b 100644
--- a/api/core/model_runtime/model_providers/tongyi/llm/qwen2.5-coder-7b-instruct.yaml
+++ b/api/core/model_runtime/model_providers/tongyi/llm/qwen2.5-coder-7b-instruct.yaml
@@ -65,15 +65,6 @@ parameter_rules:
     help:
       zh_Hans: 用于控制模型生成时的重复度。提高repetition_penalty时可以降低模型生成的重复度。1.0表示不做惩罚。
       en_US: Used to control the repeatability when generating models. Increasing repetition_penalty can reduce the duplication of model generation. 1.0 means no punishment.
-  - name: enable_search
-    type: boolean
-    default: false
-    label:
-      zh_Hans: 联网搜索
-      en_US: Web Search
-    help:
-      zh_Hans: 模型内置了互联网搜索服务，该参数控制模型在生成文本时是否参考使用互联网搜索结果。启用互联网搜索，模型会将搜索结果作为文本生成过程中的参考信息，但模型会基于其内部逻辑“自行判断”是否使用互联网搜索结果。
-      en_US: The model has a built-in Internet search service. This parameter controls whether the model refers to Internet search results when generating text. When Internet search is enabled, the model will use the search results as reference information in the text generation process, but the model will "judge" whether to use Internet search results based on its internal logic.
   - name: response_format
     use_template: response_format
 pricing:

From 4638f99aaabdfef5ad751d725ef5ca26027bd7fe Mon Sep 17 00:00:00 2001
From: Benjamin <benjaminx@gmail.com>
Date: Tue, 24 Sep 2024 13:26:58 +0800
Subject: [PATCH 050/235] fix: change model provider name issue Ref #8691
 (#8710)

---
 api/core/model_runtime/model_providers/jina/jina.yaml | 2 +-
 api/core/tools/provider/builtin/jina/jina.yaml        | 2 +-
 2 files changed, 2 insertions(+), 2 deletions(-)

diff --git a/api/core/model_runtime/model_providers/jina/jina.yaml b/api/core/model_runtime/model_providers/jina/jina.yaml
index 2f2d6e6daad73..4ff6ba0f22c84 100644
--- a/api/core/model_runtime/model_providers/jina/jina.yaml
+++ b/api/core/model_runtime/model_providers/jina/jina.yaml
@@ -1,4 +1,4 @@
-provider: Jina AI
+provider: jina
 label:
   en_US: Jina AI
 description:
diff --git a/api/core/tools/provider/builtin/jina/jina.yaml b/api/core/tools/provider/builtin/jina/jina.yaml
index 9ce5cbd6d1afa..346175c41fbe1 100644
--- a/api/core/tools/provider/builtin/jina/jina.yaml
+++ b/api/core/tools/provider/builtin/jina/jina.yaml
@@ -1,6 +1,6 @@
 identity:
   author: Dify
-  name: Jina AI
+  name: jina
   label:
     en_US: Jina AI
     zh_Hans: Jina AI

From 64baedb48429bf7332feab61d7a83cf1b7c91cda Mon Sep 17 00:00:00 2001
From: ice yao <yao3690093@gmail.com>
Date: Tue, 24 Sep 2024 14:04:07 +0800
Subject: [PATCH 051/235] fix: update nomic model provider token calculation
 (#8705)

---
 api/core/model_runtime/model_providers/_position.yaml  |  1 +
 .../nomic/text_embedding/text_embedding.py             | 10 +---------
 2 files changed, 2 insertions(+), 9 deletions(-)

diff --git a/api/core/model_runtime/model_providers/_position.yaml b/api/core/model_runtime/model_providers/_position.yaml
index 79ebd007647c8..80db22ea84fe6 100644
--- a/api/core/model_runtime/model_providers/_position.yaml
+++ b/api/core/model_runtime/model_providers/_position.yaml
@@ -39,3 +39,4 @@
 - zhinao
 - fireworks
 - mixedbread
+- nomic
diff --git a/api/core/model_runtime/model_providers/nomic/text_embedding/text_embedding.py b/api/core/model_runtime/model_providers/nomic/text_embedding/text_embedding.py
index 6cccff6d4601c..ccbfd196a9522 100644
--- a/api/core/model_runtime/model_providers/nomic/text_embedding/text_embedding.py
+++ b/api/core/model_runtime/model_providers/nomic/text_embedding/text_embedding.py
@@ -77,15 +77,7 @@ def get_num_tokens(self, model: str, credentials: dict, texts: list[str]) -> int
         :param texts: texts to embed
         :return:
         """
-        if len(texts) == 0:
-            return 0
-
-        _, prompt_tokens, _ = self.embed_text(
-            model=model,
-            credentials=credentials,
-            texts=texts,
-        )
-        return prompt_tokens
+        return sum(self._get_num_tokens_by_gpt2(text) for text in texts)
 
     def validate_credentials(self, model: str, credentials: dict) -> None:
         """

From f42ef0624dbb8a1286a4fd00b15eaf90ceb70921 Mon Sep 17 00:00:00 2001
From: zxhlyh <jasonapring2015@outlook.com>
Date: Tue, 24 Sep 2024 17:23:11 +0800
Subject: [PATCH 052/235] fix: embedded chat on ios (#8718)

---
 web/public/embed.js     | 60 ++++++++++++++++++++---------------------
 web/public/embed.min.js | 57 ++++++++++++++++++---------------------
 2 files changed, 56 insertions(+), 61 deletions(-)

diff --git a/web/public/embed.js b/web/public/embed.js
index 8ed7a67dc8a31..3c2735b6fc8d1 100644
--- a/web/public/embed.js
+++ b/web/public/embed.js
@@ -69,38 +69,47 @@
       iframe.id = iframeId;
       iframe.src = iframeUrl;
       iframe.style.cssText = `
-        border: none; position: fixed; flex-direction: column; justify-content: space-between;
+        border: none; position: absolute; flex-direction: column; justify-content: space-between;
         box-shadow: rgba(150, 150, 150, 0.2) 0px 10px 30px 0px, rgba(150, 150, 150, 0.2) 0px 0px 0px 1px;
-        bottom: 5rem; right: 1rem; width: 24rem; max-width: calc(100vw - 2rem); height: 40rem;
+        bottom: 55px; right: 0; width: 24rem; max-width: calc(100vw - 2rem); height: 40rem;
         max-height: calc(100vh - 6rem); border-radius: 0.75rem; display: flex; z-index: 2147483647;
         overflow: hidden; left: unset; background-color: #F3F4F6;user-select: none;
       `;
 
-      document.body.appendChild(iframe);
+      return iframe;
     }
 
     // Function to reset the iframe position
     function resetIframePosition() {
+      if (window.innerWidth <= 640)
+        return
+
       const targetIframe = document.getElementById(iframeId);
       const targetButton = document.getElementById(buttonId);
       if (targetIframe && targetButton) {
         const buttonRect = targetButton.getBoundingClientRect();
-        const buttonBottom = window.innerHeight - buttonRect.bottom;
-        const buttonRight = window.innerWidth - buttonRect.right;
-        const buttonLeft = buttonRect.left;
-
-        // Adjust iframe position to stay within viewport
-        targetIframe.style.bottom = `${
-          buttonBottom + buttonRect.height + 5 + targetIframe.clientHeight > window.innerHeight
-            ? buttonBottom - targetIframe.clientHeight - 5
-            : buttonBottom + buttonRect.height + 5
-        }px`;
-
-        targetIframe.style.right = `${
-          buttonRight + targetIframe.clientWidth > window.innerWidth
-            ? window.innerWidth - buttonLeft - targetIframe.clientWidth
-            : buttonRight
-        }px`;
+
+        const buttonInBottom = buttonRect.top - 5 > targetIframe.clientHeight
+
+        if (buttonInBottom) {
+          targetIframe.style.bottom = `${buttonRect.height + 5}px`;
+          targetIframe.style.top = 'unset';
+        }
+        else {
+          targetIframe.style.bottom = 'unset';
+          targetIframe.style.top = `${buttonRect.height + 5}px`;
+        }
+
+        const buttonInRight = buttonRect.right > targetIframe.clientWidth;
+
+        if (buttonInRight) {
+          targetIframe.style.right = '0';
+          targetIframe.style.left = 'unset';
+        }
+        else {
+          targetIframe.style.right = 'unset';
+          targetIframe.style.left = 0;
+        }
       }
     }
 
@@ -146,12 +155,6 @@
           box-shadow: var(--${containerDiv.id}-box-shadow, rgba(0, 0, 0, 0.2) 0px 4px 8px 0px);
           cursor: pointer;
           z-index: 2147483647;
-          transition: all 0.2s ease-in-out 0s;
-        }
-      `);
-      styleSheet.sheet.insertRule(`
-        #${containerDiv.id}:hover {
-          transform: var(--${containerDiv.id}-hover-transform, scale(1.1));
         }
       `);
 
@@ -167,7 +170,7 @@
       containerDiv.addEventListener("click", function () {
         const targetIframe = document.getElementById(iframeId);
         if (!targetIframe) {
-          createIframe();
+          containerDiv.appendChild(createIframe());
           resetIframePosition();
           this.title = "Exit (ESC)";
           displayDiv.innerHTML = svgIcons.close;
@@ -255,9 +258,6 @@
     if (!document.getElementById(buttonId)) {
       createButton();
     }
-
-    createIframe();
-    document.getElementById(iframeId).style.display = 'none';
   }
 
   // Add esc Exit keyboard event triggered
@@ -279,4 +279,4 @@
   } else {
     document.body.onload = embedChatbot;
   }
-})();
+})();
\ No newline at end of file
diff --git a/web/public/embed.min.js b/web/public/embed.min.js
index 0e023cb5d1970..eb2085814883b 100644
--- a/web/public/embed.min.js
+++ b/web/public/embed.min.js
@@ -1,31 +1,26 @@
-(()=>{let t="difyChatbotConfig",a="dify-chatbot-bubble-button",c="dify-chatbot-bubble-window",h=window[t],p={open:`<svg id="openIcon" width="24" height="24" viewBox="0 0 24 24" fill="none" xmlns="http://www.w3.org/2000/svg">
-    <path fill-rule="evenodd" clip-rule="evenodd" d="M7.7586 2L16.2412 2C17.0462 1.99999 17.7105 1.99998 18.2517 2.04419C18.8138 2.09012 19.3305 2.18868 19.8159 2.43598C20.5685 2.81947 21.1804 3.43139 21.5639 4.18404C21.8112 4.66937 21.9098 5.18608 21.9557 5.74818C21.9999 6.28937 21.9999 6.95373 21.9999 7.7587L22 14.1376C22.0004 14.933 22.0007 15.5236 21.8636 16.0353C21.4937 17.4156 20.4155 18.4938 19.0352 18.8637C18.7277 18.9461 18.3917 18.9789 17.9999 18.9918L17.9999 20.371C18 20.6062 18 20.846 17.9822 21.0425C17.9651 21.2305 17.9199 21.5852 17.6722 21.8955C17.3872 22.2525 16.9551 22.4602 16.4983 22.4597C16.1013 22.4593 15.7961 22.273 15.6386 22.1689C15.474 22.06 15.2868 21.9102 15.1031 21.7632L12.69 19.8327C12.1714 19.4178 12.0174 19.3007 11.8575 19.219C11.697 19.137 11.5262 19.0771 11.3496 19.0408C11.1737 19.0047 10.9803 19 10.3162 19H7.75858C6.95362 19 6.28927 19 5.74808 18.9558C5.18598 18.9099 4.66928 18.8113 4.18394 18.564C3.43129 18.1805 2.81937 17.5686 2.43588 16.816C2.18859 16.3306 2.09002 15.8139 2.0441 15.2518C1.99988 14.7106 1.99989 14.0463 1.9999 13.2413V7.75868C1.99989 6.95372 1.99988 6.28936 2.0441 5.74818C2.09002 5.18608 2.18859 4.66937 2.43588 4.18404C2.81937 3.43139 3.43129 2.81947 4.18394 2.43598C4.66928 2.18868 5.18598 2.09012 5.74808 2.04419C6.28927 1.99998 6.95364 1.99999 7.7586 2ZM10.5073 7.5C10.5073 6.67157 9.83575 6 9.00732 6C8.1789 6 7.50732 6.67157 7.50732 7.5C7.50732 8.32843 8.1789 9 9.00732 9C9.83575 9 10.5073 8.32843 10.5073 7.5ZM16.6073 11.7001C16.1669 11.3697 15.5426 11.4577 15.2105 11.8959C15.1488 11.9746 15.081 12.0486 15.0119 12.1207C14.8646 12.2744 14.6432 12.4829 14.3566 12.6913C13.7796 13.111 12.9818 13.5001 12.0073 13.5001C11.0328 13.5001 10.235 13.111 9.65799 12.6913C9.37138 12.4829 9.15004 12.2744 9.00274 12.1207C8.93366 12.0486 8.86581 11.9745 8.80418 11.8959C8.472 11.4577 7.84775 11.3697 7.40732 11.7001C6.96549 12.0314 6.87595 12.6582 7.20732 13.1001C7.20479 13.0968 7.21072 13.1043 7.22094 13.1171C7.24532 13.1478 7.29407 13.2091 7.31068 13.2289C7.36932 13.2987 7.45232 13.3934 7.55877 13.5045C7.77084 13.7258 8.08075 14.0172 8.48165 14.3088C9.27958 14.8891 10.4818 15.5001 12.0073 15.5001C13.5328 15.5001 14.735 14.8891 15.533 14.3088C15.9339 14.0172 16.2438 13.7258 16.4559 13.5045C16.5623 13.3934 16.6453 13.2987 16.704 13.2289C16.7333 13.1939 16.7567 13.165 16.7739 13.1432C17.1193 12.6969 17.0729 12.0493 16.6073 11.7001ZM15.0073 6C15.8358 6 16.5073 6.67157 16.5073 7.5C16.5073 8.32843 15.8358 9 15.0073 9C14.1789 9 13.5073 8.32843 13.5073 7.5C13.5073 6.67157 14.1789 6 15.0073 6Z" fill="white"/>
-  </svg>`,close:`<svg id="closeIcon" width="24" height="24" viewBox="0 0 24 24" fill="none" xmlns="http://www.w3.org/2000/svg">
-    <path d="M18 18L6 6M6 18L18 6" stroke="white" stroke-width="2" stroke-linecap="round" stroke-linejoin="round"/>
-  </svg>`};async function e(){if(h&&h.token){var e=new URLSearchParams(await(async()=>{var e=h?.inputs||{};let n={};return await Promise.all(Object.entries(e).map(async([e,t])=>{n[e]=(e=t,e=(new TextEncoder).encode(e),e=new Response(new Blob([e]).stream().pipeThrough(new CompressionStream("gzip"))).arrayBuffer(),e=new Uint8Array(await e),await btoa(String.fromCharCode(...e)))})),n})());let t=`${h.baseUrl||`https://${h.isDev?"dev.":""}udify.app`}/chatbot/${h.token}?`+e;function o(){var e=document.createElement("iframe");e.allow="fullscreen;microphone",e.title="dify chatbot bubble window",e.id=c,e.src=t,e.style.cssText=`
-      border: none; position: fixed; flex-direction: column; justify-content: space-between;
-      box-shadow: rgba(150, 150, 150, 0.2) 0px 10px 30px 0px, rgba(150, 150, 150, 0.2) 0px 0px 0px 1px;
-      bottom: 5rem; right: 1rem; width: 24rem; max-width: calc(100vw - 2rem); height: 40rem;
-      max-height: calc(100vh - 6rem); border-radius: 0.75rem; display: flex; z-index: 2147483647;
-      overflow: hidden; left: unset; background-color: #F3F4F6;user-select: none;
-    `,document.body.appendChild(e)}function i(){var e,t,n,o=document.getElementById(c),i=document.getElementById(a);o&&i&&(i=i.getBoundingClientRect(),e=window.innerHeight-i.bottom,t=window.innerWidth-i.right,n=i.left,o.style.bottom=`${e+i.height+5+o.clientHeight>window.innerHeight?e-o.clientHeight-5:e+i.height+5}px`,o.style.right=`${t+o.clientWidth>window.innerWidth?window.innerWidth-n-o.clientWidth:t}px`)}function n(){let n=document.createElement("div");Object.entries(h.containerProps||{}).forEach(([e,t])=>{"className"===e?n.classList.add(...t.split(" ")):"style"===e?"object"==typeof t?Object.assign(n.style,t):n.style.cssText=t:"function"==typeof t?n.addEventListener(e.replace(/^on/,"").toLowerCase(),t):n[e]=t}),n.id=a;var e=document.createElement("style");document.head.appendChild(e),e.sheet.insertRule(`
-      #${n.id} {
-        position: fixed;
-        bottom: var(--${n.id}-bottom, 1rem);
-        right: var(--${n.id}-right, 1rem);
-        left: var(--${n.id}-left, unset);
-        top: var(--${n.id}-top, unset);
-        width: var(--${n.id}-width, 50px);
-        height: var(--${n.id}-height, 50px);
-        border-radius: var(--${n.id}-border-radius, 25px);
-        background-color: var(--${n.id}-bg-color, #155EEF);
-        box-shadow: var(--${n.id}-box-shadow, rgba(0, 0, 0, 0.2) 0px 4px 8px 0px);
-        cursor: pointer;
-        z-index: 2147483647;
-        transition: all 0.2s ease-in-out 0s;
-      }
-    `),e.sheet.insertRule(`
-      #${n.id}:hover {
-        transform: var(--${n.id}-hover-transform, scale(1.1));
-      }
-    `);let t=document.createElement("div");if(t.style.cssText="display: flex; align-items: center; justify-content: center; width: 100%; height: 100%; z-index: 2147483647;",t.innerHTML=p.open,n.appendChild(t),document.body.appendChild(n),n.addEventListener("click",function(){var e=document.getElementById(c);e?(e.style.display="none"===e.style.display?"block":"none",t.innerHTML="none"===e.style.display?p.open:p.close,"none"===e.style.display?document.removeEventListener("keydown",d):document.addEventListener("keydown",d),i()):(o(),i(),this.title="Exit (ESC)",t.innerHTML=p.close,document.addEventListener("keydown",d))}),h.draggable){var s=n;var l=h.dragAxis||"both";let i=!1,d,r;s.addEventListener("mousedown",function(e){i=!0,d=e.clientX-s.offsetLeft,r=e.clientY-s.offsetTop}),document.addEventListener("mousemove",function(e){var t,n,o;i&&(s.style.transition="none",s.style.cursor="grabbing",(t=document.getElementById(c))&&(t.style.display="none",s.querySelector("div").innerHTML=p.open),t=e.clientX-d,e=window.innerHeight-e.clientY-r,o=s.getBoundingClientRect(),n=window.innerWidth-o.width,o=window.innerHeight-o.height,"x"!==l&&"both"!==l||s.style.setProperty(`--${a}-left`,Math.max(0,Math.min(t,n))+"px"),"y"!==l&&"both"!==l||s.style.setProperty(`--${a}-bottom`,Math.max(0,Math.min(e,o))+"px"))}),document.addEventListener("mouseup",function(){i=!1,s.style.transition="",s.style.cursor="pointer"})}}2048<t.length&&console.error("The URL is too long, please reduce the number of inputs to prevent the bot from failing to load"),document.getElementById(a)||n(),o(),document.getElementById(c).style.display="none"}else console.error(t+" is empty or token is not provided")}function d(e){var t;"Escape"===e.key&&(e=document.getElementById(c),t=document.getElementById(a),e)&&"none"!==e.style.display&&(e.style.display="none",t.querySelector("div").innerHTML=p.open)}document.addEventListener("keydown",d),h?.dynamicScript?e():document.body.onload=e})();
\ No newline at end of file
+!function(){const n="difyChatbotConfig",a="dify-chatbot-bubble-button",c="dify-chatbot-bubble-window",p=window[n],h={open:`<svg id="openIcon" width="24" height="24" viewBox="0 0 24 24" fill="none" xmlns="http://www.w3.org/2000/svg">
+      <path fill-rule="evenodd" clip-rule="evenodd" d="M7.7586 2L16.2412 2C17.0462 1.99999 17.7105 1.99998 18.2517 2.04419C18.8138 2.09012 19.3305 2.18868 19.8159 2.43598C20.5685 2.81947 21.1804 3.43139 21.5639 4.18404C21.8112 4.66937 21.9098 5.18608 21.9557 5.74818C21.9999 6.28937 21.9999 6.95373 21.9999 7.7587L22 14.1376C22.0004 14.933 22.0007 15.5236 21.8636 16.0353C21.4937 17.4156 20.4155 18.4938 19.0352 18.8637C18.7277 18.9461 18.3917 18.9789 17.9999 18.9918L17.9999 20.371C18 20.6062 18 20.846 17.9822 21.0425C17.9651 21.2305 17.9199 21.5852 17.6722 21.8955C17.3872 22.2525 16.9551 22.4602 16.4983 22.4597C16.1013 22.4593 15.7961 22.273 15.6386 22.1689C15.474 22.06 15.2868 21.9102 15.1031 21.7632L12.69 19.8327C12.1714 19.4178 12.0174 19.3007 11.8575 19.219C11.697 19.137 11.5262 19.0771 11.3496 19.0408C11.1737 19.0047 10.9803 19 10.3162 19H7.75858C6.95362 19 6.28927 19 5.74808 18.9558C5.18598 18.9099 4.66928 18.8113 4.18394 18.564C3.43129 18.1805 2.81937 17.5686 2.43588 16.816C2.18859 16.3306 2.09002 15.8139 2.0441 15.2518C1.99988 14.7106 1.99989 14.0463 1.9999 13.2413V7.75868C1.99989 6.95372 1.99988 6.28936 2.0441 5.74818C2.09002 5.18608 2.18859 4.66937 2.43588 4.18404C2.81937 3.43139 3.43129 2.81947 4.18394 2.43598C4.66928 2.18868 5.18598 2.09012 5.74808 2.04419C6.28927 1.99998 6.95364 1.99999 7.7586 2ZM10.5073 7.5C10.5073 6.67157 9.83575 6 9.00732 6C8.1789 6 7.50732 6.67157 7.50732 7.5C7.50732 8.32843 8.1789 9 9.00732 9C9.83575 9 10.5073 8.32843 10.5073 7.5ZM16.6073 11.7001C16.1669 11.3697 15.5426 11.4577 15.2105 11.8959C15.1488 11.9746 15.081 12.0486 15.0119 12.1207C14.8646 12.2744 14.6432 12.4829 14.3566 12.6913C13.7796 13.111 12.9818 13.5001 12.0073 13.5001C11.0328 13.5001 10.235 13.111 9.65799 12.6913C9.37138 12.4829 9.15004 12.2744 9.00274 12.1207C8.93366 12.0486 8.86581 11.9745 8.80418 11.8959C8.472 11.4577 7.84775 11.3697 7.40732 11.7001C6.96549 12.0314 6.87595 12.6582 7.20732 13.1001C7.20479 13.0968 7.21072 13.1043 7.22094 13.1171C7.24532 13.1478 7.29407 13.2091 7.31068 13.2289C7.36932 13.2987 7.45232 13.3934 7.55877 13.5045C7.77084 13.7258 8.08075 14.0172 8.48165 14.3088C9.27958 14.8891 10.4818 15.5001 12.0073 15.5001C13.5328 15.5001 14.735 14.8891 15.533 14.3088C15.9339 14.0172 16.2438 13.7258 16.4559 13.5045C16.5623 13.3934 16.6453 13.2987 16.704 13.2289C16.7333 13.1939 16.7567 13.165 16.7739 13.1432C17.1193 12.6969 17.0729 12.0493 16.6073 11.7001ZM15.0073 6C15.8358 6 16.5073 6.67157 16.5073 7.5C16.5073 8.32843 15.8358 9 15.0073 9C14.1789 9 13.5073 8.32843 13.5073 7.5C13.5073 6.67157 14.1789 6 15.0073 6Z" fill="white"/>
+    </svg>`,close:`<svg id="closeIcon" width="24" height="24" viewBox="0 0 24 24" fill="none" xmlns="http://www.w3.org/2000/svg">
+      <path d="M18 18L6 6M6 18L18 6" stroke="white" stroke-width="2" stroke-linecap="round" stroke-linejoin="round"/>
+    </svg>`};async function e(){if(p&&p.token){var e=new URLSearchParams(await async function(){var e=p?.inputs||{};const n={};return await Promise.all(Object.entries(e).map(async([e,t])=>{n[e]=(e=t,e=(new TextEncoder).encode(e),e=new Response(new Blob([e]).stream().pipeThrough(new CompressionStream("gzip"))).arrayBuffer(),e=new Uint8Array(await e),await btoa(String.fromCharCode(...e)))})),n}());const i=`${p.baseUrl||`https://${p.isDev?"dev.":""}udify.app`}/chatbot/${p.token}?`+e;function o(){var e,t;window.innerWidth<=640||(e=document.getElementById(c),t=document.getElementById(a),e&&t&&((t=t.getBoundingClientRect()).top-5>e.clientHeight?(e.style.bottom=t.height+5+"px",e.style.top="unset"):(e.style.bottom="unset",e.style.top=t.height+5+"px"),t.right>e.clientWidth?(e.style.right="0",e.style.left="unset"):(e.style.right="unset",e.style.left=0)))}function t(){const n=document.createElement("div");Object.entries(p.containerProps||{}).forEach(([e,t])=>{"className"===e?n.classList.add(...t.split(" ")):"style"===e?"object"==typeof t?Object.assign(n.style,t):n.style.cssText=t:"function"==typeof t?n.addEventListener(e.replace(/^on/,"").toLowerCase(),t):n[e]=t}),n.id=a;var e=document.createElement("style");document.head.appendChild(e),e.sheet.insertRule(`
+        #${n.id} {
+          position: fixed;
+          bottom: var(--${n.id}-bottom, 1rem);
+          right: var(--${n.id}-right, 1rem);
+          left: var(--${n.id}-left, unset);
+          top: var(--${n.id}-top, unset);
+          width: var(--${n.id}-width, 50px);
+          height: var(--${n.id}-height, 50px);
+          border-radius: var(--${n.id}-border-radius, 25px);
+          background-color: var(--${n.id}-bg-color, #155EEF);
+          box-shadow: var(--${n.id}-box-shadow, rgba(0, 0, 0, 0.2) 0px 4px 8px 0px);
+          cursor: pointer;
+          z-index: 2147483647;
+        }
+      `);const t=document.createElement("div");if(t.style.cssText="display: flex; align-items: center; justify-content: center; width: 100%; height: 100%; z-index: 2147483647;",t.innerHTML=h.open,n.appendChild(t),document.body.appendChild(n),n.addEventListener("click",function(){var e=document.getElementById(c);e?(e.style.display="none"===e.style.display?"block":"none",t.innerHTML="none"===e.style.display?h.open:h.close,"none"===e.style.display?document.removeEventListener("keydown",d):document.addEventListener("keydown",d),o()):(n.appendChild(((e=document.createElement("iframe")).allow="fullscreen;microphone",e.title="dify chatbot bubble window",e.id=c,e.src=i,e.style.cssText=`
+        border: none; position: absolute; flex-direction: column; justify-content: space-between;
+        box-shadow: rgba(150, 150, 150, 0.2) 0px 10px 30px 0px, rgba(150, 150, 150, 0.2) 0px 0px 0px 1px;
+        bottom: 55px; right: 0; width: 24rem; max-width: calc(100vw - 2rem); height: 40rem;
+        max-height: calc(100vh - 6rem); border-radius: 0.75rem; display: flex; z-index: 2147483647;
+        overflow: hidden; left: unset; background-color: #F3F4F6;user-select: none;
+      `,e)),o(),this.title="Exit (ESC)",t.innerHTML=h.close,document.addEventListener("keydown",d))}),p.draggable){var s=n;var l=p.dragAxis||"both";let i=!1,d,r;s.addEventListener("mousedown",function(e){i=!0,d=e.clientX-s.offsetLeft,r=e.clientY-s.offsetTop}),document.addEventListener("mousemove",function(e){var t,n,o;i&&(s.style.transition="none",s.style.cursor="grabbing",(t=document.getElementById(c))&&(t.style.display="none",s.querySelector("div").innerHTML=h.open),t=e.clientX-d,e=window.innerHeight-e.clientY-r,o=s.getBoundingClientRect(),n=window.innerWidth-o.width,o=window.innerHeight-o.height,"x"!==l&&"both"!==l||s.style.setProperty(`--${a}-left`,Math.max(0,Math.min(t,n))+"px"),"y"!==l&&"both"!==l||s.style.setProperty(`--${a}-bottom`,Math.max(0,Math.min(e,o))+"px"))}),document.addEventListener("mouseup",function(){i=!1,s.style.transition="",s.style.cursor="pointer"})}}2048<i.length&&console.error("The URL is too long, please reduce the number of inputs to prevent the bot from failing to load"),document.getElementById(a)||t()}else console.error(n+" is empty or token is not provided")}function d(e){var t;"Escape"===e.key&&(e=document.getElementById(c),t=document.getElementById(a),e)&&"none"!==e.style.display&&(e.style.display="none",t.querySelector("div").innerHTML=h.open)}document.addEventListener("keydown",d),p?.dynamicScript?e():document.body.onload=e}();
\ No newline at end of file

From 9ca2e2c96896e038a9e676aed5f355fd54188591 Mon Sep 17 00:00:00 2001
From: =?UTF-8?q?=E9=9D=9E=E6=B3=95=E6=93=8D=E4=BD=9C?= <hjlarry@163.com>
Date: Tue, 24 Sep 2024 17:33:29 +0800
Subject: [PATCH 053/235] chore: remove windows platform timezone set (#8712)

---
 api/app.py | 8 +++-----
 1 file changed, 3 insertions(+), 5 deletions(-)

diff --git a/api/app.py b/api/app.py
index 91a49337fccbd..1b58beee15819 100644
--- a/api/app.py
+++ b/api/app.py
@@ -53,11 +53,9 @@
 
 warnings.simplefilter("ignore", ResourceWarning)
 
-# fix windows platform
-if os.name == "nt":
-    os.system('tzutil /s "UTC"')
-else:
-    os.environ["TZ"] = "UTC"
+os.environ["TZ"] = "UTC"
+# windows platform not support tzset
+if hasattr(time, "tzset"):
     time.tzset()
 
 

From 1c7877b048d2131c8c133c26c60c8ccd342f0b0c Mon Sep 17 00:00:00 2001
From: Shota Totsuka <153569547+totsukash@users.noreply.github.com>
Date: Tue, 24 Sep 2024 21:53:26 +0900
Subject: [PATCH 054/235] fix: remove harm category setting from vertex ai
 (#8721)

---
 .../model_providers/vertex_ai/llm/llm.py              | 11 ++---------
 1 file changed, 2 insertions(+), 9 deletions(-)

diff --git a/api/core/model_runtime/model_providers/vertex_ai/llm/llm.py b/api/core/model_runtime/model_providers/vertex_ai/llm/llm.py
index da69b7cdf382d..1dd785d545408 100644
--- a/api/core/model_runtime/model_providers/vertex_ai/llm/llm.py
+++ b/api/core/model_runtime/model_providers/vertex_ai/llm/llm.py
@@ -2,6 +2,7 @@
 import io
 import json
 import logging
+import time
 from collections.abc import Generator
 from typing import Optional, Union, cast
 
@@ -20,7 +21,6 @@
 from google.cloud import aiplatform
 from google.oauth2 import service_account
 from PIL import Image
-from vertexai.generative_models import HarmBlockThreshold, HarmCategory
 
 from core.model_runtime.entities.llm_entities import LLMResult, LLMResultChunk, LLMResultChunkDelta, LLMUsage
 from core.model_runtime.entities.message_entities import (
@@ -34,6 +34,7 @@
     ToolPromptMessage,
     UserPromptMessage,
 )
+from core.model_runtime.entities.model_entities import PriceType
 from core.model_runtime.errors.invoke import (
     InvokeAuthorizationError,
     InvokeBadRequestError,
@@ -503,20 +504,12 @@ def _generate(
                     else:
                         history.append(content)
 
-        safety_settings = {
-            HarmCategory.HARM_CATEGORY_HARASSMENT: HarmBlockThreshold.BLOCK_NONE,
-            HarmCategory.HARM_CATEGORY_HATE_SPEECH: HarmBlockThreshold.BLOCK_NONE,
-            HarmCategory.HARM_CATEGORY_SEXUALLY_EXPLICIT: HarmBlockThreshold.BLOCK_NONE,
-            HarmCategory.HARM_CATEGORY_DANGEROUS_CONTENT: HarmBlockThreshold.BLOCK_NONE,
-        }
-
         google_model = glm.GenerativeModel(model_name=model, system_instruction=system_instruction)
 
         response = google_model.generate_content(
             contents=history,
             generation_config=glm.GenerationConfig(**config_kwargs),
             stream=stream,
-            safety_settings=safety_settings,
             tools=self._convert_tools_to_glm_tool(tools) if tools else None,
         )
 

From debe5953a814f075aec5f878bd338829cd63ca9e Mon Sep 17 00:00:00 2001
From: Sa Zhang <55871322+Nick17t@users.noreply.github.com>
Date: Tue, 24 Sep 2024 21:19:49 +0800
Subject: [PATCH 055/235] Fix/update jina ai products labels and descriptions
 (#8730)

Co-authored-by: sa zhang <sa.zhang@jina.ai>
---
 .../builtin/jina/tools/jina_reader.yaml       | 42 +++++++++----------
 .../builtin/jina/tools/jina_search.yaml       | 31 +++++++-------
 .../builtin/jina/tools/jina_tokenizer.yaml    | 16 +++++--
 3 files changed, 50 insertions(+), 39 deletions(-)

diff --git a/api/core/tools/provider/builtin/jina/tools/jina_reader.yaml b/api/core/tools/provider/builtin/jina/tools/jina_reader.yaml
index 58ad6d8694222..589bc3433d947 100644
--- a/api/core/tools/provider/builtin/jina/tools/jina_reader.yaml
+++ b/api/core/tools/provider/builtin/jina/tools/jina_reader.yaml
@@ -2,14 +2,14 @@ identity:
   name: jina_reader
   author: Dify
   label:
-    en_US: JinaReader
-    zh_Hans: JinaReader
-    pt_BR: JinaReader
+    en_US: Fetch Single Page
+    zh_Hans: 获取单页面
+    pt_BR: Fetch Single Page
 description:
   human:
-    en_US: Convert any URL to an LLM-friendly input. Experience improved output for your agent and RAG systems at no cost.
-    zh_Hans: 将任何 URL 转换为 LLM 友好的输入。无需付费即可体验为您的 Agent 和 RAG 系统提供的改进输出。
-    pt_BR: Converta qualquer URL em uma entrada amigável ao LLM. Experimente uma saída aprimorada para seus sistemas de agente e RAG sem custo.
+    en_US: Fetch the target URL (can be a PDF) and convert it into a LLM-friendly markdown.
+    zh_Hans: 获取目标网址（可以是 PDF），并将其转换为适合大模型处理的 Markdown 格式。
+    pt_BR: Busque a URL de destino (que pode ser um PDF) e converta em um Markdown LLM-friendly.
   llm: A tool for scraping webpages. Input should be a URL.
 parameters:
   - name: url
@@ -17,13 +17,13 @@ parameters:
     required: true
     label:
       en_US: URL
-      zh_Hans: 网页链接
+      zh_Hans: 网址
       pt_BR: URL
     human_description:
-      en_US: used for linking to webpages
-      zh_Hans: 用于链接到网页
-      pt_BR: used for linking to webpages
-    llm_description: url for scraping
+      en_US: Web link
+      zh_Hans: 网页链接
+      pt_BR: URL da web
+    llm_description: url para scraping
     form: llm
   - name: request_params
     type: string
@@ -31,14 +31,14 @@ parameters:
     label:
       en_US: Request params
       zh_Hans: 请求参数
-      pt_BR: Request params
+      pt_BR: Parâmetros de solicitação
     human_description:
       en_US: |
         request parameters, format: {"key1": "value1", "key2": "value2"}
       zh_Hans: |
         请求参数，格式：{"key1": "value1", "key2": "value2"}
       pt_BR: |
-        request parameters, format: {"key1": "value1", "key2": "value2"}
+        parâmetros de solicitação, formato: {"key1": "value1", "key2": "value2"}
     llm_description: request parameters
     form: llm
   - name: target_selector
@@ -51,7 +51,7 @@ parameters:
     human_description:
       en_US: css selector for scraping specific elements
       zh_Hans: css 选择器用于抓取特定元素
-      pt_BR: css selector for scraping specific elements
+      pt_BR: css selector para scraping de elementos específicos
     llm_description: css selector of the target element to scrape
     form: form
   - name: wait_for_selector
@@ -64,7 +64,7 @@ parameters:
     human_description:
       en_US: css selector for waiting for specific elements
       zh_Hans: css 选择器用于等待特定元素
-      pt_BR: css selector for waiting for specific elements
+      pt_BR: css selector para aguardar elementos específicos
     llm_description: css selector of the target element to wait for
     form: form
   - name: image_caption
@@ -77,8 +77,8 @@ parameters:
       pt_BR: Legenda da imagem
     human_description:
       en_US: "Captions all images at the specified URL, adding 'Image [idx]: [caption]' as an alt tag for those without one. This allows downstream LLMs to interact with the images in activities such as reasoning and summarizing."
-      zh_Hans: "为指定 URL 上的所有图像添加标题，为没有标题的图像添加“Image [idx]: [caption]”作为 alt 标签。这允许下游 LLM 在推理和总结等活动中与图像进行交互。"
-      pt_BR: "Captions all images at the specified URL, adding 'Image [idx]: [caption]' as an alt tag for those without one. This allows downstream LLMs to interact with the images in activities such as reasoning and summarizing."
+      zh_Hans: "为指定 URL 上的所有图像添加标题，为没有标题的图像添加“Image [idx]: [caption]”作为 alt 标签，以支持下游模型的图像交互。"
+      pt_BR: "Adiciona legendas a todas as imagens na URL especificada, adicionando 'Imagem [idx]: [legenda]' como uma tag alt para aquelas que não têm uma. Isso permite que os modelos LLM inferiores interajam com as imagens em atividades como raciocínio e resumo."
     llm_description: Captions all images at the specified URL
     form: form
   - name: gather_all_links_at_the_end
@@ -91,8 +91,8 @@ parameters:
       pt_BR: Coletar todos os links ao final
     human_description:
       en_US: A "Buttons & Links" section will be created at the end. This helps the downstream LLMs or web agents navigating the page or take further actions.
-      zh_Hans: 最后会创建一个“按钮和链接”部分。这可以帮助下游 LLM 或 Web 代理浏览页面或采取进一步的行动。
-      pt_BR: A "Buttons & Links" section will be created at the end. This helps the downstream LLMs or web agents navigating the page or take further actions.
+      zh_Hans: 末尾将添加“按钮和链接”部分，方便下游模型或网络代理做页面导航或执行进一步操作。
+      pt_BR: Um "Botões & Links" section will be created at the end. This helps the downstream LLMs or web agents navigating the page or take further actions.
     llm_description: Gather all links at the end
     form: form
   - name: gather_all_images_at_the_end
@@ -105,8 +105,8 @@ parameters:
       pt_BR: Coletar todas as imagens ao final
     human_description:
       en_US: An "Images" section will be created at the end. This gives the downstream LLMs an overview of all visuals on the page, which may improve reasoning.
-      zh_Hans: 最后会创建一个“图像”部分。这可以让下游的 LLM 概览页面上的所有视觉效果，从而提高推理能力。
-      pt_BR: An "Images" section will be created at the end. This gives the downstream LLMs an overview of all visuals on the page, which may improve reasoning.
+      zh_Hans: 末尾会新增“图片”部分，方便下游模型全面了解页面的视觉内容，提升推理效果。
+      pt_BR: Um "Imagens" section will be created at the end. This gives the downstream LLMs an overview of all visuals on the page, which may improve reasoning.
     llm_description: Gather all images at the end
     form: form
   - name: proxy_server
diff --git a/api/core/tools/provider/builtin/jina/tools/jina_search.yaml b/api/core/tools/provider/builtin/jina/tools/jina_search.yaml
index 2bc70e1be1934..e58c639e5690d 100644
--- a/api/core/tools/provider/builtin/jina/tools/jina_search.yaml
+++ b/api/core/tools/provider/builtin/jina/tools/jina_search.yaml
@@ -2,13 +2,14 @@ identity:
   name: jina_search
   author: Dify
   label:
-    en_US: JinaSearch
-    zh_Hans: JinaSearch
-    pt_BR: JinaSearch
+    en_US: Search the web
+    zh_Hans: 联网搜索
+    pt_BR: Search the web
 description:
   human:
-    en_US: Search on the web and get the top 5 results. Useful for grounding using information from the web.
-    zh_Hans: 在网络上搜索返回前 5 个结果。
+    en_US: Search on the public web of a given query and return the top results as LLM-friendly markdown.
+    zh_Hans: 针对给定的查询在互联网上进行搜索，并以适合大模型处理的 Markdown 格式返回最相关的结果。
+    pt_BR: Procurar na web pública de uma consulta fornecida e retornar os melhores resultados como markdown para LLMs.
   llm: A tool for searching results on the web for grounding. Input should be a simple question.
 parameters:
   - name: query
@@ -16,11 +17,13 @@ parameters:
     required: true
     label:
       en_US: Question (Query)
-      zh_Hans: 信息查询
+      zh_Hans: 查询
+      pt_BR: Pergunta (Consulta)
     human_description:
       en_US: used to find information on the web
       zh_Hans: 在网络上搜索信息
-    llm_description: simple question to ask on the web
+      pt_BR: Usado para encontrar informações na web
+    llm_description: Pergunta simples para fazer na web
     form: llm
   - name: image_caption
     type: boolean
@@ -32,7 +35,7 @@ parameters:
       pt_BR: Legenda da imagem
     human_description:
       en_US: "Captions all images at the specified URL, adding 'Image [idx]: [caption]' as an alt tag for those without one. This allows downstream LLMs to interact with the images in activities such as reasoning and summarizing."
-      zh_Hans: "为指定 URL 上的所有图像添加标题，为没有标题的图像添加“Image [idx]: [caption]”作为 alt 标签。这允许下游 LLM 在推理和总结等活动中与图像进行交互。"
+      zh_Hans: "为指定 URL 上的所有图像添加标题，为没有标题的图像添加“Image [idx]: [caption]”作为 alt 标签，以支持下游模型的图像交互。"
       pt_BR: "Captions all images at the specified URL, adding 'Image [idx]: [caption]' as an alt tag for those without one. This allows downstream LLMs to interact with the images in activities such as reasoning and summarizing."
     llm_description: Captions all images at the specified URL
     form: form
@@ -46,8 +49,8 @@ parameters:
       pt_BR: Coletar todos os links ao final
     human_description:
       en_US: A "Buttons & Links" section will be created at the end. This helps the downstream LLMs or web agents navigating the page or take further actions.
-      zh_Hans: 最后会创建一个“按钮和链接”部分。这可以帮助下游 LLM 或 Web 代理浏览页面或采取进一步的行动。
-      pt_BR: A "Buttons & Links" section will be created at the end. This helps the downstream LLMs or web agents navigating the page or take further actions.
+      zh_Hans: 末尾将添加“按钮和链接”部分，汇总页面上的所有链接。方便下游模型或网络代理做页面导航或执行进一步操作。
+      pt_BR: Um "Botão & Links" seção será criada no final. Isso ajuda os LLMs ou agentes da web navegando pela página ou executar ações adicionais.
     llm_description: Gather all links at the end
     form: form
   - name: gather_all_images_at_the_end
@@ -60,8 +63,8 @@ parameters:
       pt_BR: Coletar todas as imagens ao final
     human_description:
       en_US: An "Images" section will be created at the end. This gives the downstream LLMs an overview of all visuals on the page, which may improve reasoning.
-      zh_Hans: 最后会创建一个“图像”部分。这可以让下游的 LLM 概览页面上的所有视觉效果，从而提高推理能力。
-      pt_BR: An "Images" section will be created at the end. This gives the downstream LLMs an overview of all visuals on the page, which may improve reasoning.
+      zh_Hans: 末尾会新增“图片”部分，汇总页面上的所有图片。方便下游模型概览页面的视觉内容，提升推理效果。
+      pt_BR: Um "Imagens" seção será criada no final. Isso fornece uma visão geral de todas as imagens na página para os LLMs, que pode melhorar a razão.
     llm_description: Gather all images at the end
     form: form
   - name: proxy_server
@@ -74,7 +77,7 @@ parameters:
     human_description:
       en_US: Use proxy to access URLs
       zh_Hans: 利用代理访问 URL
-      pt_BR: Use proxy to access URLs
+      pt_BR: Usar proxy para acessar URLs
     llm_description: Use proxy to access URLs
     form: form
   - name: no_cache
@@ -83,7 +86,7 @@ parameters:
     default: false
     label:
       en_US: Bypass the Cache
-      zh_Hans: 绕过缓存
+      zh_Hans: 是否绕过缓存
       pt_BR: Ignorar o cache
     human_description:
       en_US: Bypass the Cache
diff --git a/api/core/tools/provider/builtin/jina/tools/jina_tokenizer.yaml b/api/core/tools/provider/builtin/jina/tools/jina_tokenizer.yaml
index 62a5c7e7bacd7..74885cdf9a704 100644
--- a/api/core/tools/provider/builtin/jina/tools/jina_tokenizer.yaml
+++ b/api/core/tools/provider/builtin/jina/tools/jina_tokenizer.yaml
@@ -2,11 +2,14 @@ identity:
   name: jina_tokenizer
   author: hjlarry
   label:
-    en_US: JinaTokenizer
+    en_US: Segment
+    zh_Hans: 切分器
+    pt_BR: Segment
 description:
   human:
-    en_US: Free API to tokenize text and segment long text into chunks.
-    zh_Hans: 免费的API可以将文本tokenize，也可以将长文本分割成多个部分。
+    en_US: Split long text into chunks and do tokenization.
+    zh_Hans: 将长文本拆分成小段落，并做分词处理。
+    pt_BR: Dividir o texto longo em pedaços e fazer tokenização.
   llm: Free API to tokenize text and segment long text into chunks.
 parameters:
   - name: content
@@ -15,6 +18,7 @@ parameters:
     label:
       en_US: Content
       zh_Hans: 内容
+      pt_BR: Conteúdo
     llm_description: the content which need to tokenize or segment
     form: llm
   - name: return_tokens
@@ -23,18 +27,22 @@ parameters:
     label:
       en_US: Return the tokens
       zh_Hans: 是否返回tokens
+      pt_BR: Retornar os tokens
     human_description:
       en_US: Return the tokens and their corresponding ids in the response.
       zh_Hans: 返回tokens及其对应的ids。
+      pt_BR: Retornar os tokens e seus respectivos ids na resposta.
     form: form
   - name: return_chunks
     type: boolean
     label:
       en_US: Return the chunks
       zh_Hans: 是否分块
+      pt_BR: Retornar os chunks
     human_description:
       en_US: Chunking the input into semantically meaningful segments while handling a wide variety of text types and edge cases based on common structural cues.
-      zh_Hans: 将输入分块为具有语义意义的片段，同时根据常见的结构线索处理各种文本类型和边缘情况。
+      zh_Hans: 将输入文本分块为语义有意义的片段，同时基于常见的结构线索处理各种文本类型和特殊情况。
+      pt_BR: Dividir o texto de entrada em segmentos semanticamente significativos, enquanto lida com uma ampla variedade de tipos de texto e casos de borda com base em pistas estruturais comuns.
     form: form
   - name: tokenizer
     type: select

From 4669eb24befe840789de1dc9ae32f631a31a765d Mon Sep 17 00:00:00 2001
From: Jyong <76649700+JohnJyong@users.noreply.github.com>
Date: Tue, 24 Sep 2024 21:53:50 +0800
Subject: [PATCH 056/235] add embedding input type parameter (#8724)

---
 api/core/embedding/cached_embedding.py        |  9 ++++++--
 api/core/embedding/embedding_constant.py      | 10 ++++++++
 api/core/model_manager.py                     |  7 +++++-
 .../__base/text_embedding_model.py            | 23 +++++++++++++++----
 .../text_embedding/text_embedding.py          |  8 ++++++-
 .../baichuan/text_embedding/text_embedding.py |  8 ++++++-
 .../bedrock/text_embedding/text_embedding.py  |  8 ++++++-
 .../cohere/text_embedding/text_embedding.py   |  8 ++++++-
 .../text_embedding/text_embedding.py          |  8 ++++++-
 .../text_embedding/text_embedding.py          |  8 ++++++-
 .../hunyuan/text_embedding/text_embedding.py  |  8 ++++++-
 .../jina/text_embedding/text_embedding.py     |  8 ++++++-
 .../localai/text_embedding/text_embedding.py  |  8 ++++++-
 .../minimax/text_embedding/text_embedding.py  |  8 ++++++-
 .../text_embedding/text_embedding.py          |  8 ++++++-
 .../nomic/text_embedding/text_embedding.py    |  2 ++
 .../nvidia/text_embedding/text_embedding.py   |  8 ++++++-
 .../oci/text_embedding/text_embedding.py      |  9 +++++++-
 .../ollama/text_embedding/text_embedding.py   |  8 ++++++-
 .../openai/text_embedding/text_embedding.py   |  8 ++++++-
 .../text_embedding/text_embedding.py          |  8 ++++++-
 .../openllm/text_embedding/text_embedding.py  |  8 ++++++-
 .../text_embedding/text_embedding.py          |  9 +++++++-
 .../text_embedding/text_embedding.py          |  8 ++++++-
 .../text_embedding/text_embedding.py          |  8 ++++++-
 .../text_embedding/text_embedding.py          |  8 ++++++-
 .../tongyi/text_embedding/text_embedding.py   |  2 ++
 .../upstage/text_embedding/text_embedding.py  | 10 +++++++-
 .../text_embedding/text_embedding.py          |  8 ++++++-
 .../text_embedding/text_embedding.py          |  8 ++++++-
 .../wenxin/text_embedding/text_embedding.py   |  8 ++++++-
 .../text_embedding/text_embedding.py          |  9 +++++++-
 .../zhipuai/text_embedding/text_embedding.py  |  8 ++++++-
 33 files changed, 239 insertions(+), 35 deletions(-)
 create mode 100644 api/core/embedding/embedding_constant.py

diff --git a/api/core/embedding/cached_embedding.py b/api/core/embedding/cached_embedding.py
index 8ce12fd59f576..75219051cd30c 100644
--- a/api/core/embedding/cached_embedding.py
+++ b/api/core/embedding/cached_embedding.py
@@ -5,6 +5,7 @@
 import numpy as np
 from sqlalchemy.exc import IntegrityError
 
+from core.embedding.embedding_constant import EmbeddingInputType
 from core.model_manager import ModelInstance
 from core.model_runtime.entities.model_entities import ModelPropertyKey
 from core.model_runtime.model_providers.__base.text_embedding_model import TextEmbeddingModel
@@ -56,7 +57,9 @@ def embed_documents(self, texts: list[str]) -> list[list[float]]:
                 for i in range(0, len(embedding_queue_texts), max_chunks):
                     batch_texts = embedding_queue_texts[i : i + max_chunks]
 
-                    embedding_result = self._model_instance.invoke_text_embedding(texts=batch_texts, user=self._user)
+                    embedding_result = self._model_instance.invoke_text_embedding(
+                        texts=batch_texts, user=self._user, input_type=EmbeddingInputType.DOCUMENT
+                    )
 
                     for vector in embedding_result.embeddings:
                         try:
@@ -100,7 +103,9 @@ def embed_query(self, text: str) -> list[float]:
             redis_client.expire(embedding_cache_key, 600)
             return list(np.frombuffer(base64.b64decode(embedding), dtype="float"))
         try:
-            embedding_result = self._model_instance.invoke_text_embedding(texts=[text], user=self._user)
+            embedding_result = self._model_instance.invoke_text_embedding(
+                texts=[text], user=self._user, input_type=EmbeddingInputType.QUERY
+            )
 
             embedding_results = embedding_result.embeddings[0]
             embedding_results = (embedding_results / np.linalg.norm(embedding_results)).tolist()
diff --git a/api/core/embedding/embedding_constant.py b/api/core/embedding/embedding_constant.py
new file mode 100644
index 0000000000000..9b4934646bc0e
--- /dev/null
+++ b/api/core/embedding/embedding_constant.py
@@ -0,0 +1,10 @@
+from enum import Enum
+
+
+class EmbeddingInputType(Enum):
+    """
+    Enum for embedding input type.
+    """
+
+    DOCUMENT = "document"
+    QUERY = "query"
diff --git a/api/core/model_manager.py b/api/core/model_manager.py
index 990efd36c609c..74b445236268d 100644
--- a/api/core/model_manager.py
+++ b/api/core/model_manager.py
@@ -3,6 +3,7 @@
 from collections.abc import Callable, Generator, Sequence
 from typing import IO, Optional, Union, cast
 
+from core.embedding.embedding_constant import EmbeddingInputType
 from core.entities.provider_configuration import ProviderConfiguration, ProviderModelBundle
 from core.entities.provider_entities import ModelLoadBalancingConfiguration
 from core.errors.error import ProviderTokenNotInitError
@@ -158,12 +159,15 @@ def get_llm_num_tokens(
             tools=tools,
         )
 
-    def invoke_text_embedding(self, texts: list[str], user: Optional[str] = None) -> TextEmbeddingResult:
+    def invoke_text_embedding(
+        self, texts: list[str], user: Optional[str] = None, input_type: EmbeddingInputType = EmbeddingInputType.DOCUMENT
+    ) -> TextEmbeddingResult:
         """
         Invoke large language model
 
         :param texts: texts to embed
         :param user: unique user id
+        :param input_type: input type
         :return: embeddings result
         """
         if not isinstance(self.model_type_instance, TextEmbeddingModel):
@@ -176,6 +180,7 @@ def invoke_text_embedding(self, texts: list[str], user: Optional[str] = None) ->
             credentials=self.credentials,
             texts=texts,
             user=user,
+            input_type=input_type,
         )
 
     def get_text_embedding_num_tokens(self, texts: list[str]) -> int:
diff --git a/api/core/model_runtime/model_providers/__base/text_embedding_model.py b/api/core/model_runtime/model_providers/__base/text_embedding_model.py
index 54a4486023691..a948dca20d69a 100644
--- a/api/core/model_runtime/model_providers/__base/text_embedding_model.py
+++ b/api/core/model_runtime/model_providers/__base/text_embedding_model.py
@@ -4,6 +4,7 @@
 
 from pydantic import ConfigDict
 
+from core.embedding.embedding_constant import EmbeddingInputType
 from core.model_runtime.entities.model_entities import ModelPropertyKey, ModelType
 from core.model_runtime.entities.text_embedding_entities import TextEmbeddingResult
 from core.model_runtime.model_providers.__base.ai_model import AIModel
@@ -20,35 +21,47 @@ class TextEmbeddingModel(AIModel):
     model_config = ConfigDict(protected_namespaces=())
 
     def invoke(
-        self, model: str, credentials: dict, texts: list[str], user: Optional[str] = None
+        self,
+        model: str,
+        credentials: dict,
+        texts: list[str],
+        user: Optional[str] = None,
+        input_type: EmbeddingInputType = EmbeddingInputType.DOCUMENT,
     ) -> TextEmbeddingResult:
         """
-        Invoke large language model
+        Invoke text embedding model
 
         :param model: model name
         :param credentials: model credentials
         :param texts: texts to embed
         :param user: unique user id
+        :param input_type: input type
         :return: embeddings result
         """
         self.started_at = time.perf_counter()
 
         try:
-            return self._invoke(model, credentials, texts, user)
+            return self._invoke(model, credentials, texts, user, input_type)
         except Exception as e:
             raise self._transform_invoke_error(e)
 
     @abstractmethod
     def _invoke(
-        self, model: str, credentials: dict, texts: list[str], user: Optional[str] = None
+        self,
+        model: str,
+        credentials: dict,
+        texts: list[str],
+        user: Optional[str] = None,
+        input_type: EmbeddingInputType = EmbeddingInputType.DOCUMENT,
     ) -> TextEmbeddingResult:
         """
-        Invoke large language model
+        Invoke text embedding model
 
         :param model: model name
         :param credentials: model credentials
         :param texts: texts to embed
         :param user: unique user id
+        :param input_type: input type
         :return: embeddings result
         """
         raise NotImplementedError
diff --git a/api/core/model_runtime/model_providers/azure_openai/text_embedding/text_embedding.py b/api/core/model_runtime/model_providers/azure_openai/text_embedding/text_embedding.py
index d9cff8ecbbadb..6b270b65ff00a 100644
--- a/api/core/model_runtime/model_providers/azure_openai/text_embedding/text_embedding.py
+++ b/api/core/model_runtime/model_providers/azure_openai/text_embedding/text_embedding.py
@@ -7,6 +7,7 @@
 import tiktoken
 from openai import AzureOpenAI
 
+from core.embedding.embedding_constant import EmbeddingInputType
 from core.model_runtime.entities.model_entities import AIModelEntity, PriceType
 from core.model_runtime.entities.text_embedding_entities import EmbeddingUsage, TextEmbeddingResult
 from core.model_runtime.errors.validate import CredentialsValidateFailedError
@@ -17,7 +18,12 @@
 
 class AzureOpenAITextEmbeddingModel(_CommonAzureOpenAI, TextEmbeddingModel):
     def _invoke(
-        self, model: str, credentials: dict, texts: list[str], user: Optional[str] = None
+        self,
+        model: str,
+        credentials: dict,
+        texts: list[str],
+        user: Optional[str] = None,
+        input_type: EmbeddingInputType = EmbeddingInputType.DOCUMENT,
     ) -> TextEmbeddingResult:
         base_model_name = credentials["base_model_name"]
         credentials_kwargs = self._to_credential_kwargs(credentials)
diff --git a/api/core/model_runtime/model_providers/baichuan/text_embedding/text_embedding.py b/api/core/model_runtime/model_providers/baichuan/text_embedding/text_embedding.py
index 779dfbb608163..210c274bdf5e3 100644
--- a/api/core/model_runtime/model_providers/baichuan/text_embedding/text_embedding.py
+++ b/api/core/model_runtime/model_providers/baichuan/text_embedding/text_embedding.py
@@ -4,6 +4,7 @@
 
 from requests import post
 
+from core.embedding.embedding_constant import EmbeddingInputType
 from core.model_runtime.entities.model_entities import PriceType
 from core.model_runtime.entities.text_embedding_entities import EmbeddingUsage, TextEmbeddingResult
 from core.model_runtime.errors.invoke import (
@@ -35,7 +36,12 @@ class BaichuanTextEmbeddingModel(TextEmbeddingModel):
     api_base: str = "http://api.baichuan-ai.com/v1/embeddings"
 
     def _invoke(
-        self, model: str, credentials: dict, texts: list[str], user: Optional[str] = None
+        self,
+        model: str,
+        credentials: dict,
+        texts: list[str],
+        user: Optional[str] = None,
+        input_type: EmbeddingInputType = EmbeddingInputType.DOCUMENT,
     ) -> TextEmbeddingResult:
         """
         Invoke text embedding model
diff --git a/api/core/model_runtime/model_providers/bedrock/text_embedding/text_embedding.py b/api/core/model_runtime/model_providers/bedrock/text_embedding/text_embedding.py
index 251170d1aec49..8c4c50b26958c 100644
--- a/api/core/model_runtime/model_providers/bedrock/text_embedding/text_embedding.py
+++ b/api/core/model_runtime/model_providers/bedrock/text_embedding/text_embedding.py
@@ -13,6 +13,7 @@
     UnknownServiceError,
 )
 
+from core.embedding.embedding_constant import EmbeddingInputType
 from core.model_runtime.entities.model_entities import PriceType
 from core.model_runtime.entities.text_embedding_entities import EmbeddingUsage, TextEmbeddingResult
 from core.model_runtime.errors.invoke import (
@@ -30,7 +31,12 @@
 
 class BedrockTextEmbeddingModel(TextEmbeddingModel):
     def _invoke(
-        self, model: str, credentials: dict, texts: list[str], user: Optional[str] = None
+        self,
+        model: str,
+        credentials: dict,
+        texts: list[str],
+        user: Optional[str] = None,
+        input_type: EmbeddingInputType = EmbeddingInputType.DOCUMENT,
     ) -> TextEmbeddingResult:
         """
         Invoke text embedding model
diff --git a/api/core/model_runtime/model_providers/cohere/text_embedding/text_embedding.py b/api/core/model_runtime/model_providers/cohere/text_embedding/text_embedding.py
index a1c5e98118e4c..1f93068a8c9e5 100644
--- a/api/core/model_runtime/model_providers/cohere/text_embedding/text_embedding.py
+++ b/api/core/model_runtime/model_providers/cohere/text_embedding/text_embedding.py
@@ -5,6 +5,7 @@
 import numpy as np
 from cohere.core import RequestOptions
 
+from core.embedding.embedding_constant import EmbeddingInputType
 from core.model_runtime.entities.model_entities import PriceType
 from core.model_runtime.entities.text_embedding_entities import EmbeddingUsage, TextEmbeddingResult
 from core.model_runtime.errors.invoke import (
@@ -25,7 +26,12 @@ class CohereTextEmbeddingModel(TextEmbeddingModel):
     """
 
     def _invoke(
-        self, model: str, credentials: dict, texts: list[str], user: Optional[str] = None
+        self,
+        model: str,
+        credentials: dict,
+        texts: list[str],
+        user: Optional[str] = None,
+        input_type: EmbeddingInputType = EmbeddingInputType.DOCUMENT,
     ) -> TextEmbeddingResult:
         """
         Invoke text embedding model
diff --git a/api/core/model_runtime/model_providers/huggingface_hub/text_embedding/text_embedding.py b/api/core/model_runtime/model_providers/huggingface_hub/text_embedding/text_embedding.py
index 4ad96c4233fb1..cf18f84ac8cb2 100644
--- a/api/core/model_runtime/model_providers/huggingface_hub/text_embedding/text_embedding.py
+++ b/api/core/model_runtime/model_providers/huggingface_hub/text_embedding/text_embedding.py
@@ -6,6 +6,7 @@
 import requests
 from huggingface_hub import HfApi, InferenceClient
 
+from core.embedding.embedding_constant import EmbeddingInputType
 from core.model_runtime.entities.common_entities import I18nObject
 from core.model_runtime.entities.model_entities import AIModelEntity, FetchFrom, ModelType, PriceType
 from core.model_runtime.entities.text_embedding_entities import EmbeddingUsage, TextEmbeddingResult
@@ -18,7 +19,12 @@
 
 class HuggingfaceHubTextEmbeddingModel(_CommonHuggingfaceHub, TextEmbeddingModel):
     def _invoke(
-        self, model: str, credentials: dict, texts: list[str], user: Optional[str] = None
+        self,
+        model: str,
+        credentials: dict,
+        texts: list[str],
+        user: Optional[str] = None,
+        input_type: EmbeddingInputType = EmbeddingInputType.DOCUMENT,
     ) -> TextEmbeddingResult:
         client = InferenceClient(token=credentials["huggingfacehub_api_token"])
 
diff --git a/api/core/model_runtime/model_providers/huggingface_tei/text_embedding/text_embedding.py b/api/core/model_runtime/model_providers/huggingface_tei/text_embedding/text_embedding.py
index 55f3c25804ea0..58baf4933c241 100644
--- a/api/core/model_runtime/model_providers/huggingface_tei/text_embedding/text_embedding.py
+++ b/api/core/model_runtime/model_providers/huggingface_tei/text_embedding/text_embedding.py
@@ -1,6 +1,7 @@
 import time
 from typing import Optional
 
+from core.embedding.embedding_constant import EmbeddingInputType
 from core.model_runtime.entities.common_entities import I18nObject
 from core.model_runtime.entities.model_entities import AIModelEntity, FetchFrom, ModelPropertyKey, ModelType, PriceType
 from core.model_runtime.entities.text_embedding_entities import EmbeddingUsage, TextEmbeddingResult
@@ -23,7 +24,12 @@ class HuggingfaceTeiTextEmbeddingModel(TextEmbeddingModel):
     """
 
     def _invoke(
-        self, model: str, credentials: dict, texts: list[str], user: Optional[str] = None
+        self,
+        model: str,
+        credentials: dict,
+        texts: list[str],
+        user: Optional[str] = None,
+        input_type: EmbeddingInputType = EmbeddingInputType.DOCUMENT,
     ) -> TextEmbeddingResult:
         """
         Invoke text embedding model
diff --git a/api/core/model_runtime/model_providers/hunyuan/text_embedding/text_embedding.py b/api/core/model_runtime/model_providers/hunyuan/text_embedding/text_embedding.py
index 1396e59e188bc..3e14371f897e0 100644
--- a/api/core/model_runtime/model_providers/hunyuan/text_embedding/text_embedding.py
+++ b/api/core/model_runtime/model_providers/hunyuan/text_embedding/text_embedding.py
@@ -9,6 +9,7 @@
 from tencentcloud.common.profile.http_profile import HttpProfile
 from tencentcloud.hunyuan.v20230901 import hunyuan_client, models
 
+from core.embedding.embedding_constant import EmbeddingInputType
 from core.model_runtime.entities.model_entities import PriceType
 from core.model_runtime.entities.text_embedding_entities import EmbeddingUsage, TextEmbeddingResult
 from core.model_runtime.errors.invoke import (
@@ -26,7 +27,12 @@ class HunyuanTextEmbeddingModel(TextEmbeddingModel):
     """
 
     def _invoke(
-        self, model: str, credentials: dict, texts: list[str], user: Optional[str] = None
+        self,
+        model: str,
+        credentials: dict,
+        texts: list[str],
+        user: Optional[str] = None,
+        input_type: EmbeddingInputType = EmbeddingInputType.DOCUMENT,
     ) -> TextEmbeddingResult:
         """
         Invoke text embedding model
diff --git a/api/core/model_runtime/model_providers/jina/text_embedding/text_embedding.py b/api/core/model_runtime/model_providers/jina/text_embedding/text_embedding.py
index 6c96699ea2e3f..9120f26b8d92f 100644
--- a/api/core/model_runtime/model_providers/jina/text_embedding/text_embedding.py
+++ b/api/core/model_runtime/model_providers/jina/text_embedding/text_embedding.py
@@ -4,6 +4,7 @@
 
 from requests import post
 
+from core.embedding.embedding_constant import EmbeddingInputType
 from core.model_runtime.entities.common_entities import I18nObject
 from core.model_runtime.entities.model_entities import AIModelEntity, FetchFrom, ModelPropertyKey, ModelType, PriceType
 from core.model_runtime.entities.text_embedding_entities import EmbeddingUsage, TextEmbeddingResult
@@ -60,7 +61,12 @@ def transform_jina_input_text(model, text):
         return data
 
     def _invoke(
-        self, model: str, credentials: dict, texts: list[str], user: Optional[str] = None
+        self,
+        model: str,
+        credentials: dict,
+        texts: list[str],
+        user: Optional[str] = None,
+        input_type: EmbeddingInputType = EmbeddingInputType.DOCUMENT,
     ) -> TextEmbeddingResult:
         """
         Invoke text embedding model
diff --git a/api/core/model_runtime/model_providers/localai/text_embedding/text_embedding.py b/api/core/model_runtime/model_providers/localai/text_embedding/text_embedding.py
index 7d258be81e058..d8878c7be85bc 100644
--- a/api/core/model_runtime/model_providers/localai/text_embedding/text_embedding.py
+++ b/api/core/model_runtime/model_providers/localai/text_embedding/text_embedding.py
@@ -5,6 +5,7 @@
 from requests import post
 from yarl import URL
 
+from core.embedding.embedding_constant import EmbeddingInputType
 from core.model_runtime.entities.common_entities import I18nObject
 from core.model_runtime.entities.model_entities import AIModelEntity, FetchFrom, ModelPropertyKey, ModelType, PriceType
 from core.model_runtime.entities.text_embedding_entities import EmbeddingUsage, TextEmbeddingResult
@@ -26,7 +27,12 @@ class LocalAITextEmbeddingModel(TextEmbeddingModel):
     """
 
     def _invoke(
-        self, model: str, credentials: dict, texts: list[str], user: Optional[str] = None
+        self,
+        model: str,
+        credentials: dict,
+        texts: list[str],
+        user: Optional[str] = None,
+        input_type: EmbeddingInputType = EmbeddingInputType.DOCUMENT,
     ) -> TextEmbeddingResult:
         """
         Invoke text embedding model
diff --git a/api/core/model_runtime/model_providers/minimax/text_embedding/text_embedding.py b/api/core/model_runtime/model_providers/minimax/text_embedding/text_embedding.py
index 76fd1342bdb92..d0d1d2aea1ccf 100644
--- a/api/core/model_runtime/model_providers/minimax/text_embedding/text_embedding.py
+++ b/api/core/model_runtime/model_providers/minimax/text_embedding/text_embedding.py
@@ -4,6 +4,7 @@
 
 from requests import post
 
+from core.embedding.embedding_constant import EmbeddingInputType
 from core.model_runtime.entities.model_entities import PriceType
 from core.model_runtime.entities.text_embedding_entities import EmbeddingUsage, TextEmbeddingResult
 from core.model_runtime.errors.invoke import (
@@ -34,7 +35,12 @@ class MinimaxTextEmbeddingModel(TextEmbeddingModel):
     api_base: str = "https://api.minimax.chat/v1/embeddings"
 
     def _invoke(
-        self, model: str, credentials: dict, texts: list[str], user: Optional[str] = None
+        self,
+        model: str,
+        credentials: dict,
+        texts: list[str],
+        user: Optional[str] = None,
+        input_type: EmbeddingInputType = EmbeddingInputType.DOCUMENT,
     ) -> TextEmbeddingResult:
         """
         Invoke text embedding model
diff --git a/api/core/model_runtime/model_providers/mixedbread/text_embedding/text_embedding.py b/api/core/model_runtime/model_providers/mixedbread/text_embedding/text_embedding.py
index 05d9a9a0c69f4..cdc2d58d0c7e3 100644
--- a/api/core/model_runtime/model_providers/mixedbread/text_embedding/text_embedding.py
+++ b/api/core/model_runtime/model_providers/mixedbread/text_embedding/text_embedding.py
@@ -4,6 +4,7 @@
 
 import requests
 
+from core.embedding.embedding_constant import EmbeddingInputType
 from core.model_runtime.entities.common_entities import I18nObject
 from core.model_runtime.entities.model_entities import AIModelEntity, FetchFrom, ModelPropertyKey, ModelType, PriceType
 from core.model_runtime.entities.text_embedding_entities import EmbeddingUsage, TextEmbeddingResult
@@ -27,7 +28,12 @@ class MixedBreadTextEmbeddingModel(TextEmbeddingModel):
     api_base: str = "https://api.mixedbread.ai/v1"
 
     def _invoke(
-        self, model: str, credentials: dict, texts: list[str], user: Optional[str] = None
+        self,
+        model: str,
+        credentials: dict,
+        texts: list[str],
+        user: Optional[str] = None,
+        input_type: EmbeddingInputType = EmbeddingInputType.DOCUMENT,
     ) -> TextEmbeddingResult:
         """
         Invoke text embedding model
diff --git a/api/core/model_runtime/model_providers/nomic/text_embedding/text_embedding.py b/api/core/model_runtime/model_providers/nomic/text_embedding/text_embedding.py
index ccbfd196a9522..a797521576b23 100644
--- a/api/core/model_runtime/model_providers/nomic/text_embedding/text_embedding.py
+++ b/api/core/model_runtime/model_providers/nomic/text_embedding/text_embedding.py
@@ -5,6 +5,7 @@
 from nomic import embed
 from nomic import login as nomic_login
 
+from core.embedding.embedding_constant import EmbeddingInputType
 from core.model_runtime.entities.model_entities import PriceType
 from core.model_runtime.entities.text_embedding_entities import (
     EmbeddingUsage,
@@ -46,6 +47,7 @@ def _invoke(
         credentials: dict,
         texts: list[str],
         user: Optional[str] = None,
+        input_type: EmbeddingInputType = EmbeddingInputType.DOCUMENT,
     ) -> TextEmbeddingResult:
         """
         Invoke text embedding model
diff --git a/api/core/model_runtime/model_providers/nvidia/text_embedding/text_embedding.py b/api/core/model_runtime/model_providers/nvidia/text_embedding/text_embedding.py
index 00cec265d5ed9..a4ea28bd1073b 100644
--- a/api/core/model_runtime/model_providers/nvidia/text_embedding/text_embedding.py
+++ b/api/core/model_runtime/model_providers/nvidia/text_embedding/text_embedding.py
@@ -4,6 +4,7 @@
 
 from requests import post
 
+from core.embedding.embedding_constant import EmbeddingInputType
 from core.model_runtime.entities.model_entities import PriceType
 from core.model_runtime.entities.text_embedding_entities import EmbeddingUsage, TextEmbeddingResult
 from core.model_runtime.errors.invoke import (
@@ -27,7 +28,12 @@ class NvidiaTextEmbeddingModel(TextEmbeddingModel):
     models: list[str] = ["NV-Embed-QA"]
 
     def _invoke(
-        self, model: str, credentials: dict, texts: list[str], user: Optional[str] = None
+        self,
+        model: str,
+        credentials: dict,
+        texts: list[str],
+        user: Optional[str] = None,
+        input_type: EmbeddingInputType = EmbeddingInputType.DOCUMENT,
     ) -> TextEmbeddingResult:
         """
         Invoke text embedding model
diff --git a/api/core/model_runtime/model_providers/oci/text_embedding/text_embedding.py b/api/core/model_runtime/model_providers/oci/text_embedding/text_embedding.py
index 80ad2be9f5687..4de9296ccaa92 100644
--- a/api/core/model_runtime/model_providers/oci/text_embedding/text_embedding.py
+++ b/api/core/model_runtime/model_providers/oci/text_embedding/text_embedding.py
@@ -6,6 +6,7 @@
 import numpy as np
 import oci
 
+from core.embedding.embedding_constant import EmbeddingInputType
 from core.model_runtime.entities.model_entities import PriceType
 from core.model_runtime.entities.text_embedding_entities import EmbeddingUsage, TextEmbeddingResult
 from core.model_runtime.errors.invoke import (
@@ -41,7 +42,12 @@ class OCITextEmbeddingModel(TextEmbeddingModel):
     """
 
     def _invoke(
-        self, model: str, credentials: dict, texts: list[str], user: Optional[str] = None
+        self,
+        model: str,
+        credentials: dict,
+        texts: list[str],
+        user: Optional[str] = None,
+        input_type: EmbeddingInputType = EmbeddingInputType.DOCUMENT,
     ) -> TextEmbeddingResult:
         """
         Invoke text embedding model
@@ -50,6 +56,7 @@ def _invoke(
         :param credentials: model credentials
         :param texts: texts to embed
         :param user: unique user id
+        :param input_type: input type
         :return: embeddings result
         """
         # get model properties
diff --git a/api/core/model_runtime/model_providers/ollama/text_embedding/text_embedding.py b/api/core/model_runtime/model_providers/ollama/text_embedding/text_embedding.py
index b4c61d8a6dc79..0501c8b841666 100644
--- a/api/core/model_runtime/model_providers/ollama/text_embedding/text_embedding.py
+++ b/api/core/model_runtime/model_providers/ollama/text_embedding/text_embedding.py
@@ -8,6 +8,7 @@
 import numpy as np
 import requests
 
+from core.embedding.embedding_constant import EmbeddingInputType
 from core.model_runtime.entities.common_entities import I18nObject
 from core.model_runtime.entities.model_entities import (
     AIModelEntity,
@@ -38,7 +39,12 @@ class OllamaEmbeddingModel(TextEmbeddingModel):
     """
 
     def _invoke(
-        self, model: str, credentials: dict, texts: list[str], user: Optional[str] = None
+        self,
+        model: str,
+        credentials: dict,
+        texts: list[str],
+        user: Optional[str] = None,
+        input_type: EmbeddingInputType = EmbeddingInputType.DOCUMENT,
     ) -> TextEmbeddingResult:
         """
         Invoke text embedding model
diff --git a/api/core/model_runtime/model_providers/openai/text_embedding/text_embedding.py b/api/core/model_runtime/model_providers/openai/text_embedding/text_embedding.py
index 535d8388bc7d7..7945723636ced 100644
--- a/api/core/model_runtime/model_providers/openai/text_embedding/text_embedding.py
+++ b/api/core/model_runtime/model_providers/openai/text_embedding/text_embedding.py
@@ -6,6 +6,7 @@
 import tiktoken
 from openai import OpenAI
 
+from core.embedding.embedding_constant import EmbeddingInputType
 from core.model_runtime.entities.model_entities import PriceType
 from core.model_runtime.entities.text_embedding_entities import EmbeddingUsage, TextEmbeddingResult
 from core.model_runtime.errors.validate import CredentialsValidateFailedError
@@ -19,7 +20,12 @@ class OpenAITextEmbeddingModel(_CommonOpenAI, TextEmbeddingModel):
     """
 
     def _invoke(
-        self, model: str, credentials: dict, texts: list[str], user: Optional[str] = None
+        self,
+        model: str,
+        credentials: dict,
+        texts: list[str],
+        user: Optional[str] = None,
+        input_type: EmbeddingInputType = EmbeddingInputType.DOCUMENT,
     ) -> TextEmbeddingResult:
         """
         Invoke text embedding model
diff --git a/api/core/model_runtime/model_providers/openai_api_compatible/text_embedding/text_embedding.py b/api/core/model_runtime/model_providers/openai_api_compatible/text_embedding/text_embedding.py
index e83cfdf873cb4..68b5773e16afe 100644
--- a/api/core/model_runtime/model_providers/openai_api_compatible/text_embedding/text_embedding.py
+++ b/api/core/model_runtime/model_providers/openai_api_compatible/text_embedding/text_embedding.py
@@ -7,6 +7,7 @@
 import numpy as np
 import requests
 
+from core.embedding.embedding_constant import EmbeddingInputType
 from core.model_runtime.entities.common_entities import I18nObject
 from core.model_runtime.entities.model_entities import (
     AIModelEntity,
@@ -28,7 +29,12 @@ class OAICompatEmbeddingModel(_CommonOaiApiCompat, TextEmbeddingModel):
     """
 
     def _invoke(
-        self, model: str, credentials: dict, texts: list[str], user: Optional[str] = None
+        self,
+        model: str,
+        credentials: dict,
+        texts: list[str],
+        user: Optional[str] = None,
+        input_type: EmbeddingInputType = EmbeddingInputType.DOCUMENT,
     ) -> TextEmbeddingResult:
         """
         Invoke text embedding model
diff --git a/api/core/model_runtime/model_providers/openllm/text_embedding/text_embedding.py b/api/core/model_runtime/model_providers/openllm/text_embedding/text_embedding.py
index 00e583cc797dc..c0a3efbb003dc 100644
--- a/api/core/model_runtime/model_providers/openllm/text_embedding/text_embedding.py
+++ b/api/core/model_runtime/model_providers/openllm/text_embedding/text_embedding.py
@@ -5,6 +5,7 @@
 from requests import post
 from requests.exceptions import ConnectionError, InvalidSchema, MissingSchema
 
+from core.embedding.embedding_constant import EmbeddingInputType
 from core.model_runtime.entities.model_entities import PriceType
 from core.model_runtime.entities.text_embedding_entities import EmbeddingUsage, TextEmbeddingResult
 from core.model_runtime.errors.invoke import (
@@ -25,7 +26,12 @@ class OpenLLMTextEmbeddingModel(TextEmbeddingModel):
     """
 
     def _invoke(
-        self, model: str, credentials: dict, texts: list[str], user: Optional[str] = None
+        self,
+        model: str,
+        credentials: dict,
+        texts: list[str],
+        user: Optional[str] = None,
+        input_type: EmbeddingInputType = EmbeddingInputType.DOCUMENT,
     ) -> TextEmbeddingResult:
         """
         Invoke text embedding model
diff --git a/api/core/model_runtime/model_providers/perfxcloud/text_embedding/text_embedding.py b/api/core/model_runtime/model_providers/perfxcloud/text_embedding/text_embedding.py
index b62a2d2aaf06f..1e86f351c8ae5 100644
--- a/api/core/model_runtime/model_providers/perfxcloud/text_embedding/text_embedding.py
+++ b/api/core/model_runtime/model_providers/perfxcloud/text_embedding/text_embedding.py
@@ -7,6 +7,7 @@
 import numpy as np
 import requests
 
+from core.embedding.embedding_constant import EmbeddingInputType
 from core.model_runtime.entities.common_entities import I18nObject
 from core.model_runtime.entities.model_entities import (
     AIModelEntity,
@@ -28,7 +29,12 @@ class OAICompatEmbeddingModel(_CommonOaiApiCompat, TextEmbeddingModel):
     """
 
     def _invoke(
-        self, model: str, credentials: dict, texts: list[str], user: Optional[str] = None
+        self,
+        model: str,
+        credentials: dict,
+        texts: list[str],
+        user: Optional[str] = None,
+        input_type: EmbeddingInputType = EmbeddingInputType.DOCUMENT,
     ) -> TextEmbeddingResult:
         """
         Invoke text embedding model
@@ -37,6 +43,7 @@ def _invoke(
         :param credentials: model credentials
         :param texts: texts to embed
         :param user: unique user id
+        :param input_type: input type
         :return: embeddings result
         """
 
diff --git a/api/core/model_runtime/model_providers/replicate/text_embedding/text_embedding.py b/api/core/model_runtime/model_providers/replicate/text_embedding/text_embedding.py
index 71b6fb99c4a1b..b6cf89bcd9868 100644
--- a/api/core/model_runtime/model_providers/replicate/text_embedding/text_embedding.py
+++ b/api/core/model_runtime/model_providers/replicate/text_embedding/text_embedding.py
@@ -4,6 +4,7 @@
 
 from replicate import Client as ReplicateClient
 
+from core.embedding.embedding_constant import EmbeddingInputType
 from core.model_runtime.entities.common_entities import I18nObject
 from core.model_runtime.entities.model_entities import AIModelEntity, FetchFrom, ModelType, PriceType
 from core.model_runtime.entities.text_embedding_entities import EmbeddingUsage, TextEmbeddingResult
@@ -14,7 +15,12 @@
 
 class ReplicateEmbeddingModel(_CommonReplicate, TextEmbeddingModel):
     def _invoke(
-        self, model: str, credentials: dict, texts: list[str], user: Optional[str] = None
+        self,
+        model: str,
+        credentials: dict,
+        texts: list[str],
+        user: Optional[str] = None,
+        input_type: EmbeddingInputType = EmbeddingInputType.DOCUMENT,
     ) -> TextEmbeddingResult:
         client = ReplicateClient(api_token=credentials["replicate_api_token"], timeout=30)
 
diff --git a/api/core/model_runtime/model_providers/sagemaker/text_embedding/text_embedding.py b/api/core/model_runtime/model_providers/sagemaker/text_embedding/text_embedding.py
index d55144f8a79be..957f2e5d0f78f 100644
--- a/api/core/model_runtime/model_providers/sagemaker/text_embedding/text_embedding.py
+++ b/api/core/model_runtime/model_providers/sagemaker/text_embedding/text_embedding.py
@@ -6,6 +6,7 @@
 
 import boto3
 
+from core.embedding.embedding_constant import EmbeddingInputType
 from core.model_runtime.entities.common_entities import I18nObject
 from core.model_runtime.entities.model_entities import AIModelEntity, FetchFrom, ModelPropertyKey, ModelType, PriceType
 from core.model_runtime.entities.text_embedding_entities import EmbeddingUsage, TextEmbeddingResult
@@ -53,7 +54,12 @@ def _sagemaker_embedding(self, sm_client, endpoint_name, content_list: list[str]
         return embeddings
 
     def _invoke(
-        self, model: str, credentials: dict, texts: list[str], user: Optional[str] = None
+        self,
+        model: str,
+        credentials: dict,
+        texts: list[str],
+        user: Optional[str] = None,
+        input_type: EmbeddingInputType = EmbeddingInputType.DOCUMENT,
     ) -> TextEmbeddingResult:
         """
         Invoke text embedding model
diff --git a/api/core/model_runtime/model_providers/siliconflow/text_embedding/text_embedding.py b/api/core/model_runtime/model_providers/siliconflow/text_embedding/text_embedding.py
index 6cdf4933b47c7..c6c681c15d9d4 100644
--- a/api/core/model_runtime/model_providers/siliconflow/text_embedding/text_embedding.py
+++ b/api/core/model_runtime/model_providers/siliconflow/text_embedding/text_embedding.py
@@ -1,5 +1,6 @@
 from typing import Optional
 
+from core.embedding.embedding_constant import EmbeddingInputType
 from core.model_runtime.entities.text_embedding_entities import TextEmbeddingResult
 from core.model_runtime.model_providers.openai_api_compatible.text_embedding.text_embedding import (
     OAICompatEmbeddingModel,
@@ -16,7 +17,12 @@ def validate_credentials(self, model: str, credentials: dict) -> None:
         super().validate_credentials(model, credentials)
 
     def _invoke(
-        self, model: str, credentials: dict, texts: list[str], user: Optional[str] = None
+        self,
+        model: str,
+        credentials: dict,
+        texts: list[str],
+        user: Optional[str] = None,
+        input_type: EmbeddingInputType = EmbeddingInputType.DOCUMENT,
     ) -> TextEmbeddingResult:
         self._add_custom_parameters(credentials)
         return super()._invoke(model, credentials, texts, user)
diff --git a/api/core/model_runtime/model_providers/tongyi/text_embedding/text_embedding.py b/api/core/model_runtime/model_providers/tongyi/text_embedding/text_embedding.py
index 5783d2e383e2d..0eef0db3e7df0 100644
--- a/api/core/model_runtime/model_providers/tongyi/text_embedding/text_embedding.py
+++ b/api/core/model_runtime/model_providers/tongyi/text_embedding/text_embedding.py
@@ -4,6 +4,7 @@
 import dashscope
 import numpy as np
 
+from core.embedding.embedding_constant import EmbeddingInputType
 from core.model_runtime.entities.model_entities import PriceType
 from core.model_runtime.entities.text_embedding_entities import (
     EmbeddingUsage,
@@ -27,6 +28,7 @@ def _invoke(
         credentials: dict,
         texts: list[str],
         user: Optional[str] = None,
+        input_type: EmbeddingInputType = EmbeddingInputType.DOCUMENT,
     ) -> TextEmbeddingResult:
         """
         Invoke text embedding model
diff --git a/api/core/model_runtime/model_providers/upstage/text_embedding/text_embedding.py b/api/core/model_runtime/model_providers/upstage/text_embedding/text_embedding.py
index edd4a36d98f58..812bf92eeace7 100644
--- a/api/core/model_runtime/model_providers/upstage/text_embedding/text_embedding.py
+++ b/api/core/model_runtime/model_providers/upstage/text_embedding/text_embedding.py
@@ -7,6 +7,7 @@
 from openai import OpenAI
 from tokenizers import Tokenizer
 
+from core.embedding.embedding_constant import EmbeddingInputType
 from core.model_runtime.entities.model_entities import PriceType
 from core.model_runtime.entities.text_embedding_entities import EmbeddingUsage, TextEmbeddingResult
 from core.model_runtime.errors.validate import CredentialsValidateFailedError
@@ -22,7 +23,14 @@ class UpstageTextEmbeddingModel(_CommonUpstage, TextEmbeddingModel):
     def _get_tokenizer(self) -> Tokenizer:
         return Tokenizer.from_pretrained("upstage/solar-1-mini-tokenizer")
 
-    def _invoke(self, model: str, credentials: dict, texts: list[str], user: str | None = None) -> TextEmbeddingResult:
+    def _invoke(
+        self,
+        model: str,
+        credentials: dict,
+        texts: list[str],
+        user: str | None = None,
+        input_type: EmbeddingInputType = EmbeddingInputType.DOCUMENT,
+    ) -> TextEmbeddingResult:
         """
         Invoke text embedding model
 
diff --git a/api/core/model_runtime/model_providers/vertex_ai/text_embedding/text_embedding.py b/api/core/model_runtime/model_providers/vertex_ai/text_embedding/text_embedding.py
index 519373a7f31a3..509b41d9517bf 100644
--- a/api/core/model_runtime/model_providers/vertex_ai/text_embedding/text_embedding.py
+++ b/api/core/model_runtime/model_providers/vertex_ai/text_embedding/text_embedding.py
@@ -9,6 +9,7 @@
 from google.oauth2 import service_account
 from vertexai.language_models import TextEmbeddingModel as VertexTextEmbeddingModel
 
+from core.embedding.embedding_constant import EmbeddingInputType
 from core.model_runtime.entities.common_entities import I18nObject
 from core.model_runtime.entities.model_entities import (
     AIModelEntity,
@@ -30,7 +31,12 @@ class VertexAiTextEmbeddingModel(_CommonVertexAi, TextEmbeddingModel):
     """
 
     def _invoke(
-        self, model: str, credentials: dict, texts: list[str], user: Optional[str] = None
+        self,
+        model: str,
+        credentials: dict,
+        texts: list[str],
+        user: Optional[str] = None,
+        input_type: EmbeddingInputType = EmbeddingInputType.DOCUMENT,
     ) -> TextEmbeddingResult:
         """
         Invoke text embedding model
diff --git a/api/core/model_runtime/model_providers/volcengine_maas/text_embedding/text_embedding.py b/api/core/model_runtime/model_providers/volcengine_maas/text_embedding/text_embedding.py
index 9cba2cb879418..9d800af5f4d38 100644
--- a/api/core/model_runtime/model_providers/volcengine_maas/text_embedding/text_embedding.py
+++ b/api/core/model_runtime/model_providers/volcengine_maas/text_embedding/text_embedding.py
@@ -2,6 +2,7 @@
 from decimal import Decimal
 from typing import Optional
 
+from core.embedding.embedding_constant import EmbeddingInputType
 from core.model_runtime.entities.common_entities import I18nObject
 from core.model_runtime.entities.model_entities import (
     AIModelEntity,
@@ -41,7 +42,12 @@ class VolcengineMaaSTextEmbeddingModel(TextEmbeddingModel):
     """
 
     def _invoke(
-        self, model: str, credentials: dict, texts: list[str], user: Optional[str] = None
+        self,
+        model: str,
+        credentials: dict,
+        texts: list[str],
+        user: Optional[str] = None,
+        input_type: EmbeddingInputType = EmbeddingInputType.DOCUMENT,
     ) -> TextEmbeddingResult:
         """
         Invoke text embedding model
diff --git a/api/core/model_runtime/model_providers/wenxin/text_embedding/text_embedding.py b/api/core/model_runtime/model_providers/wenxin/text_embedding/text_embedding.py
index 4d6f6dccd0cf7..1b5a0904db963 100644
--- a/api/core/model_runtime/model_providers/wenxin/text_embedding/text_embedding.py
+++ b/api/core/model_runtime/model_providers/wenxin/text_embedding/text_embedding.py
@@ -7,6 +7,7 @@
 import numpy as np
 from requests import Response, post
 
+from core.embedding.embedding_constant import EmbeddingInputType
 from core.model_runtime.entities.model_entities import PriceType
 from core.model_runtime.entities.text_embedding_entities import EmbeddingUsage, TextEmbeddingResult
 from core.model_runtime.errors.invoke import InvokeError
@@ -70,7 +71,12 @@ def _create_text_embedding(self, api_key: str, secret_key: str) -> TextEmbedding
         return WenxinTextEmbedding(api_key, secret_key)
 
     def _invoke(
-        self, model: str, credentials: dict, texts: list[str], user: Optional[str] = None
+        self,
+        model: str,
+        credentials: dict,
+        texts: list[str],
+        user: Optional[str] = None,
+        input_type: EmbeddingInputType = EmbeddingInputType.DOCUMENT,
     ) -> TextEmbeddingResult:
         """
         Invoke text embedding model
diff --git a/api/core/model_runtime/model_providers/xinference/text_embedding/text_embedding.py b/api/core/model_runtime/model_providers/xinference/text_embedding/text_embedding.py
index 8043af1d6cf11..16272391320d5 100644
--- a/api/core/model_runtime/model_providers/xinference/text_embedding/text_embedding.py
+++ b/api/core/model_runtime/model_providers/xinference/text_embedding/text_embedding.py
@@ -3,6 +3,7 @@
 
 from xinference_client.client.restful.restful_client import Client, RESTfulEmbeddingModelHandle
 
+from core.embedding.embedding_constant import EmbeddingInputType
 from core.model_runtime.entities.common_entities import I18nObject
 from core.model_runtime.entities.model_entities import AIModelEntity, FetchFrom, ModelPropertyKey, ModelType, PriceType
 from core.model_runtime.entities.text_embedding_entities import EmbeddingUsage, TextEmbeddingResult
@@ -25,7 +26,12 @@ class XinferenceTextEmbeddingModel(TextEmbeddingModel):
     """
 
     def _invoke(
-        self, model: str, credentials: dict, texts: list[str], user: Optional[str] = None
+        self,
+        model: str,
+        credentials: dict,
+        texts: list[str],
+        user: Optional[str] = None,
+        input_type: EmbeddingInputType = EmbeddingInputType.DOCUMENT,
     ) -> TextEmbeddingResult:
         """
         Invoke text embedding model
@@ -40,6 +46,7 @@ def _invoke(
         :param credentials: model credentials
         :param texts: texts to embed
         :param user: unique user id
+        :param input_type: input type
         :return: embeddings result
         """
         server_url = credentials["server_url"]
diff --git a/api/core/model_runtime/model_providers/zhipuai/text_embedding/text_embedding.py b/api/core/model_runtime/model_providers/zhipuai/text_embedding/text_embedding.py
index ee20954381053..707c08ef1b5f6 100644
--- a/api/core/model_runtime/model_providers/zhipuai/text_embedding/text_embedding.py
+++ b/api/core/model_runtime/model_providers/zhipuai/text_embedding/text_embedding.py
@@ -1,6 +1,7 @@
 import time
 from typing import Optional
 
+from core.embedding.embedding_constant import EmbeddingInputType
 from core.model_runtime.entities.model_entities import PriceType
 from core.model_runtime.entities.text_embedding_entities import EmbeddingUsage, TextEmbeddingResult
 from core.model_runtime.errors.validate import CredentialsValidateFailedError
@@ -15,7 +16,12 @@ class ZhipuAITextEmbeddingModel(_CommonZhipuaiAI, TextEmbeddingModel):
     """
 
     def _invoke(
-        self, model: str, credentials: dict, texts: list[str], user: Optional[str] = None
+        self,
+        model: str,
+        credentials: dict,
+        texts: list[str],
+        user: Optional[str] = None,
+        input_type: EmbeddingInputType = EmbeddingInputType.DOCUMENT,
     ) -> TextEmbeddingResult:
         """
         Invoke text embedding model

From 91f70d0bd9d96e07c61c5a37cc36d456526758e2 Mon Sep 17 00:00:00 2001
From: ice yao <yao3690093@gmail.com>
Date: Wed, 25 Sep 2024 08:47:11 +0800
Subject: [PATCH 057/235] Add embedding models in fireworks provider (#8728)

---
 .../model_providers/fireworks/fireworks.yaml  |   1 +
 .../text_embedding/UAE-Large-V1.yaml          |  12 ++
 .../fireworks/text_embedding/__init__.py      |   0
 .../fireworks/text_embedding/gte-base.yaml    |  12 ++
 .../fireworks/text_embedding/gte-large.yaml   |  12 ++
 .../text_embedding/nomic-embed-text-v1.5.yaml |  12 ++
 .../text_embedding/nomic-embed-text-v1.yaml   |  12 ++
 .../text_embedding/text_embedding.py          | 151 ++++++++++++++++++
 .../fireworks/test_text_embedding.py          |  54 +++++++
 9 files changed, 266 insertions(+)
 create mode 100644 api/core/model_runtime/model_providers/fireworks/text_embedding/UAE-Large-V1.yaml
 create mode 100644 api/core/model_runtime/model_providers/fireworks/text_embedding/__init__.py
 create mode 100644 api/core/model_runtime/model_providers/fireworks/text_embedding/gte-base.yaml
 create mode 100644 api/core/model_runtime/model_providers/fireworks/text_embedding/gte-large.yaml
 create mode 100644 api/core/model_runtime/model_providers/fireworks/text_embedding/nomic-embed-text-v1.5.yaml
 create mode 100644 api/core/model_runtime/model_providers/fireworks/text_embedding/nomic-embed-text-v1.yaml
 create mode 100644 api/core/model_runtime/model_providers/fireworks/text_embedding/text_embedding.py
 create mode 100644 api/tests/integration_tests/model_runtime/fireworks/test_text_embedding.py

diff --git a/api/core/model_runtime/model_providers/fireworks/fireworks.yaml b/api/core/model_runtime/model_providers/fireworks/fireworks.yaml
index f886fa23b5bd8..cdb87a55e9466 100644
--- a/api/core/model_runtime/model_providers/fireworks/fireworks.yaml
+++ b/api/core/model_runtime/model_providers/fireworks/fireworks.yaml
@@ -15,6 +15,7 @@ help:
     en_US: https://fireworks.ai/account/api-keys
 supported_model_types:
   - llm
+  - text-embedding
 configurate_methods:
   - predefined-model
 provider_credential_schema:
diff --git a/api/core/model_runtime/model_providers/fireworks/text_embedding/UAE-Large-V1.yaml b/api/core/model_runtime/model_providers/fireworks/text_embedding/UAE-Large-V1.yaml
new file mode 100644
index 0000000000000..d7c11691cf9bb
--- /dev/null
+++ b/api/core/model_runtime/model_providers/fireworks/text_embedding/UAE-Large-V1.yaml
@@ -0,0 +1,12 @@
+model: WhereIsAI/UAE-Large-V1
+label:
+  zh_Hans: UAE-Large-V1
+  en_US: UAE-Large-V1
+model_type: text-embedding
+model_properties:
+  context_size: 512
+  max_chunks: 1
+pricing:
+  input: '0.008'
+  unit: '0.000001'
+  currency: 'USD'
diff --git a/api/core/model_runtime/model_providers/fireworks/text_embedding/__init__.py b/api/core/model_runtime/model_providers/fireworks/text_embedding/__init__.py
new file mode 100644
index 0000000000000..e69de29bb2d1d
diff --git a/api/core/model_runtime/model_providers/fireworks/text_embedding/gte-base.yaml b/api/core/model_runtime/model_providers/fireworks/text_embedding/gte-base.yaml
new file mode 100644
index 0000000000000..d09bafb4d312f
--- /dev/null
+++ b/api/core/model_runtime/model_providers/fireworks/text_embedding/gte-base.yaml
@@ -0,0 +1,12 @@
+model: thenlper/gte-base
+label:
+  zh_Hans: GTE-base
+  en_US: GTE-base
+model_type: text-embedding
+model_properties:
+  context_size: 512
+  max_chunks: 1
+pricing:
+  input: '0.008'
+  unit: '0.000001'
+  currency: 'USD'
diff --git a/api/core/model_runtime/model_providers/fireworks/text_embedding/gte-large.yaml b/api/core/model_runtime/model_providers/fireworks/text_embedding/gte-large.yaml
new file mode 100644
index 0000000000000..c41fa2f9d3236
--- /dev/null
+++ b/api/core/model_runtime/model_providers/fireworks/text_embedding/gte-large.yaml
@@ -0,0 +1,12 @@
+model: thenlper/gte-large
+label:
+  zh_Hans: GTE-large
+  en_US: GTE-large
+model_type: text-embedding
+model_properties:
+  context_size: 512
+  max_chunks: 1
+pricing:
+  input: '0.008'
+  unit: '0.000001'
+  currency: 'USD'
diff --git a/api/core/model_runtime/model_providers/fireworks/text_embedding/nomic-embed-text-v1.5.yaml b/api/core/model_runtime/model_providers/fireworks/text_embedding/nomic-embed-text-v1.5.yaml
new file mode 100644
index 0000000000000..c9098503d9652
--- /dev/null
+++ b/api/core/model_runtime/model_providers/fireworks/text_embedding/nomic-embed-text-v1.5.yaml
@@ -0,0 +1,12 @@
+model: nomic-ai/nomic-embed-text-v1.5
+label:
+  zh_Hans: nomic-embed-text-v1.5
+  en_US: nomic-embed-text-v1.5
+model_type: text-embedding
+model_properties:
+  context_size: 8192
+  max_chunks: 16
+pricing:
+  input: '0.008'
+  unit: '0.000001'
+  currency: 'USD'
diff --git a/api/core/model_runtime/model_providers/fireworks/text_embedding/nomic-embed-text-v1.yaml b/api/core/model_runtime/model_providers/fireworks/text_embedding/nomic-embed-text-v1.yaml
new file mode 100644
index 0000000000000..89078d3ff69f9
--- /dev/null
+++ b/api/core/model_runtime/model_providers/fireworks/text_embedding/nomic-embed-text-v1.yaml
@@ -0,0 +1,12 @@
+model: nomic-ai/nomic-embed-text-v1
+label:
+  zh_Hans: nomic-embed-text-v1
+  en_US: nomic-embed-text-v1
+model_type: text-embedding
+model_properties:
+  context_size: 8192
+  max_chunks: 16
+pricing:
+  input: '0.008'
+  unit: '0.000001'
+  currency: 'USD'
diff --git a/api/core/model_runtime/model_providers/fireworks/text_embedding/text_embedding.py b/api/core/model_runtime/model_providers/fireworks/text_embedding/text_embedding.py
new file mode 100644
index 0000000000000..cdce69ff38033
--- /dev/null
+++ b/api/core/model_runtime/model_providers/fireworks/text_embedding/text_embedding.py
@@ -0,0 +1,151 @@
+import time
+from collections.abc import Mapping
+from typing import Optional, Union
+
+import numpy as np
+from openai import OpenAI
+
+from core.embedding.embedding_constant import EmbeddingInputType
+from core.model_runtime.entities.model_entities import PriceType
+from core.model_runtime.entities.text_embedding_entities import EmbeddingUsage, TextEmbeddingResult
+from core.model_runtime.errors.validate import CredentialsValidateFailedError
+from core.model_runtime.model_providers.__base.text_embedding_model import TextEmbeddingModel
+from core.model_runtime.model_providers.fireworks._common import _CommonFireworks
+
+
+class FireworksTextEmbeddingModel(_CommonFireworks, TextEmbeddingModel):
+    """
+    Model class for Fireworks text embedding model.
+    """
+
+    def _invoke(
+        self,
+        model: str,
+        credentials: dict,
+        texts: list[str],
+        user: Optional[str] = None,
+        input_type: EmbeddingInputType = EmbeddingInputType.DOCUMENT,
+    ) -> TextEmbeddingResult:
+        """
+        Invoke text embedding model
+
+        :param model: model name
+        :param credentials: model credentials
+        :param texts: texts to embed
+        :param user: unique user id
+        :param input_type: input type
+        :return: embeddings result
+        """
+
+        credentials_kwargs = self._to_credential_kwargs(credentials)
+        client = OpenAI(**credentials_kwargs)
+
+        extra_model_kwargs = {}
+        if user:
+            extra_model_kwargs["user"] = user
+
+        extra_model_kwargs["encoding_format"] = "float"
+
+        context_size = self._get_context_size(model, credentials)
+        max_chunks = self._get_max_chunks(model, credentials)
+
+        inputs = []
+        indices = []
+        used_tokens = 0
+
+        for i, text in enumerate(texts):
+            # Here token count is only an approximation based on the GPT2 tokenizer
+            # TODO: Optimize for better token estimation and chunking
+            num_tokens = self._get_num_tokens_by_gpt2(text)
+
+            if num_tokens >= context_size:
+                cutoff = int(np.floor(len(text) * (context_size / num_tokens)))
+                # if num tokens is larger than context length, only use the start
+                inputs.append(text[0:cutoff])
+            else:
+                inputs.append(text)
+            indices += [i]
+
+        batched_embeddings = []
+        _iter = range(0, len(inputs), max_chunks)
+
+        for i in _iter:
+            embeddings_batch, embedding_used_tokens = self._embedding_invoke(
+                model=model,
+                client=client,
+                texts=inputs[i : i + max_chunks],
+                extra_model_kwargs=extra_model_kwargs,
+            )
+            used_tokens += embedding_used_tokens
+            batched_embeddings += embeddings_batch
+
+        usage = self._calc_response_usage(model=model, credentials=credentials, tokens=used_tokens)
+        return TextEmbeddingResult(embeddings=batched_embeddings, usage=usage, model=model)
+
+    def get_num_tokens(self, model: str, credentials: dict, texts: list[str]) -> int:
+        """
+        Get number of tokens for given prompt messages
+
+        :param model: model name
+        :param credentials: model credentials
+        :param texts: texts to embed
+        :return:
+        """
+        return sum(self._get_num_tokens_by_gpt2(text) for text in texts)
+
+    def validate_credentials(self, model: str, credentials: Mapping) -> None:
+        """
+        Validate model credentials
+
+        :param model: model name
+        :param credentials: model credentials
+        :return:
+        """
+        try:
+            # transform credentials to kwargs for model instance
+            credentials_kwargs = self._to_credential_kwargs(credentials)
+            client = OpenAI(**credentials_kwargs)
+
+            # call embedding model
+            self._embedding_invoke(model=model, client=client, texts=["ping"], extra_model_kwargs={})
+        except Exception as ex:
+            raise CredentialsValidateFailedError(str(ex))
+
+    def _embedding_invoke(
+        self, model: str, client: OpenAI, texts: Union[list[str], str], extra_model_kwargs: dict
+    ) -> tuple[list[list[float]], int]:
+        """
+        Invoke embedding model
+        :param model: model name
+        :param client: model client
+        :param texts: texts to embed
+        :param extra_model_kwargs: extra model kwargs
+        :return: embeddings and used tokens
+        """
+        response = client.embeddings.create(model=model, input=texts, **extra_model_kwargs)
+        return [data.embedding for data in response.data], response.usage.total_tokens
+
+    def _calc_response_usage(self, model: str, credentials: dict, tokens: int) -> EmbeddingUsage:
+        """
+        Calculate response usage
+
+        :param model: model name
+        :param credentials: model credentials
+        :param tokens: input tokens
+        :return: usage
+        """
+        input_price_info = self.get_price(
+            model=model, credentials=credentials, tokens=tokens, price_type=PriceType.INPUT
+        )
+
+        usage = EmbeddingUsage(
+            tokens=tokens,
+            total_tokens=tokens,
+            unit_price=input_price_info.unit_price,
+            price_unit=input_price_info.unit,
+            total_price=input_price_info.total_amount,
+            currency=input_price_info.currency,
+            latency=time.perf_counter() - self.started_at,
+        )
+
+        return usage
diff --git a/api/tests/integration_tests/model_runtime/fireworks/test_text_embedding.py b/api/tests/integration_tests/model_runtime/fireworks/test_text_embedding.py
new file mode 100644
index 0000000000000..7bf723b3a9374
--- /dev/null
+++ b/api/tests/integration_tests/model_runtime/fireworks/test_text_embedding.py
@@ -0,0 +1,54 @@
+import os
+
+import pytest
+
+from core.model_runtime.entities.text_embedding_entities import TextEmbeddingResult
+from core.model_runtime.errors.validate import CredentialsValidateFailedError
+from core.model_runtime.model_providers.fireworks.text_embedding.text_embedding import FireworksTextEmbeddingModel
+from tests.integration_tests.model_runtime.__mock.openai import setup_openai_mock
+
+
+@pytest.mark.parametrize("setup_openai_mock", [["text_embedding"]], indirect=True)
+def test_validate_credentials(setup_openai_mock):
+    model = FireworksTextEmbeddingModel()
+
+    with pytest.raises(CredentialsValidateFailedError):
+        model.validate_credentials(
+            model="nomic-ai/nomic-embed-text-v1.5", credentials={"fireworks_api_key": "invalid_key"}
+        )
+
+    model.validate_credentials(
+        model="nomic-ai/nomic-embed-text-v1.5", credentials={"fireworks_api_key": os.environ.get("FIREWORKS_API_KEY")}
+    )
+
+
+@pytest.mark.parametrize("setup_openai_mock", [["text_embedding"]], indirect=True)
+def test_invoke_model(setup_openai_mock):
+    model = FireworksTextEmbeddingModel()
+
+    result = model.invoke(
+        model="nomic-ai/nomic-embed-text-v1.5",
+        credentials={
+            "fireworks_api_key": os.environ.get("FIREWORKS_API_KEY"),
+        },
+        texts=["hello", "world", " ".join(["long_text"] * 100), " ".join(["another_long_text"] * 100)],
+        user="foo",
+    )
+
+    assert isinstance(result, TextEmbeddingResult)
+    assert len(result.embeddings) == 4
+    assert result.usage.total_tokens == 2
+
+
+def test_get_num_tokens():
+    model = FireworksTextEmbeddingModel()
+
+    num_tokens = model.get_num_tokens(
+        model="nomic-ai/nomic-embed-text-v1.5",
+        credentials={
+            "fireworks_api_key": os.environ.get("FIREWORKS_API_KEY"),
+        },
+        texts=["hello", "world"],
+    )
+
+    assert num_tokens == 2

From 68c7e68a8a0db9efe5c18c85c64be8346d3b2574 Mon Sep 17 00:00:00 2001
From: ybalbert001 <120714773+ybalbert001@users.noreply.github.com>
Date: Wed, 25 Sep 2024 09:12:35 +0800
Subject: [PATCH 058/235] Fix Issue: switch LLM of SageMaker endpoint doesn't
 take effect (#8737)

Co-authored-by: Yuanbo Li <ybalbert@amazon.com>
---
 .../model_providers/sagemaker/llm/llm.py         | 16 ++++++++++------
 1 file changed, 10 insertions(+), 6 deletions(-)

diff --git a/api/core/model_runtime/model_providers/sagemaker/llm/llm.py b/api/core/model_runtime/model_providers/sagemaker/llm/llm.py
index 04789197eec2a..97b7692044384 100644
--- a/api/core/model_runtime/model_providers/sagemaker/llm/llm.py
+++ b/api/core/model_runtime/model_providers/sagemaker/llm/llm.py
@@ -84,8 +84,9 @@ class SageMakerLargeLanguageModel(LargeLanguageModel):
     Model class for Cohere large language model.
     """
 
-    sagemaker_client: Any = None
+    sagemaker_session: Any = None
     predictor: Any = None
+    sagemaker_endpoint: str = None
 
     def _handle_chat_generate_response(
         self,
@@ -211,7 +212,7 @@ def _invoke(
         :param user: unique user id
         :return: full response or stream response chunk generator result
         """
-        if not self.sagemaker_client:
+        if not self.sagemaker_session:
             access_key = credentials.get("aws_access_key_id")
             secret_key = credentials.get("aws_secret_access_key")
             aws_region = credentials.get("aws_region")
@@ -226,11 +227,14 @@ def _invoke(
             else:
                 boto_session = boto3.Session()
 
-            self.sagemaker_client = boto_session.client("sagemaker")
-            sagemaker_session = Session(boto_session=boto_session, sagemaker_client=self.sagemaker_client)
+            sagemaker_client = boto_session.client("sagemaker")
+            self.sagemaker_session = Session(boto_session=boto_session, sagemaker_client=sagemaker_client)
+
+        if self.sagemaker_endpoint != credentials.get("sagemaker_endpoint"):
+            self.sagemaker_endpoint = credentials.get("sagemaker_endpoint")
             self.predictor = Predictor(
-                endpoint_name=credentials.get("sagemaker_endpoint"),
-                sagemaker_session=sagemaker_session,
+                endpoint_name=self.sagemaker_endpoint,
+                sagemaker_session=self.sagemaker_session,
                 serializer=serializers.JSONSerializer(),
             )
 

From bf64ff215be9e9094afddbd9523188c78942b039 Mon Sep 17 00:00:00 2001
From: crazywoola <100913391+crazywoola@users.noreply.github.com>
Date: Wed, 25 Sep 2024 10:09:20 +0800
Subject: [PATCH 059/235] fix: . is missing in file_extension (#8736)

---
 api/core/rag/extractor/extract_processor.py | 4 ++--
 1 file changed, 2 insertions(+), 2 deletions(-)

diff --git a/api/core/rag/extractor/extract_processor.py b/api/core/rag/extractor/extract_processor.py
index fe7eaa32e62cb..0ffc89b214c2d 100644
--- a/api/core/rag/extractor/extract_processor.py
+++ b/api/core/rag/extractor/extract_processor.py
@@ -124,7 +124,7 @@ def extract(
                         extractor = UnstructuredPPTXExtractor(file_path, unstructured_api_url)
                     elif file_extension == ".xml":
                         extractor = UnstructuredXmlExtractor(file_path, unstructured_api_url)
-                    elif file_extension == "epub":
+                    elif file_extension == ".epub":
                         extractor = UnstructuredEpubExtractor(file_path, unstructured_api_url)
                     else:
                         # txt
@@ -146,7 +146,7 @@ def extract(
                         extractor = WordExtractor(file_path, upload_file.tenant_id, upload_file.created_by)
                     elif file_extension == ".csv":
                         extractor = CSVExtractor(file_path, autodetect_encoding=True)
-                    elif file_extension == "epub":
+                    elif file_extension == ".epub":
                         extractor = UnstructuredEpubExtractor(file_path)
                     else:
                         # txt

From cb1942c242b17d930111cc97947e78f2efb9d1cd Mon Sep 17 00:00:00 2001
From: =?UTF-8?q?=E9=9D=9E=E6=B3=95=E6=93=8D=E4=BD=9C?= <hjlarry@163.com>
Date: Wed, 25 Sep 2024 11:27:17 +0800
Subject: [PATCH 060/235] chore: make url display in the middle of http node
 (#8741)

---
 web/app/components/workflow/nodes/http/node.tsx | 2 +-
 1 file changed, 1 insertion(+), 1 deletion(-)

diff --git a/web/app/components/workflow/nodes/http/node.tsx b/web/app/components/workflow/nodes/http/node.tsx
index 5bbb10fc3a596..4b7dbea2571ae 100644
--- a/web/app/components/workflow/nodes/http/node.tsx
+++ b/web/app/components/workflow/nodes/http/node.tsx
@@ -15,7 +15,7 @@ const Node: FC<NodeProps<HttpNodeType>> = ({
     <div className='mb-1 px-3 py-1'>
       <div className='flex items-start p-1 rounded-md bg-gray-100'>
         <div className='flex items-center h-4 shrink-0 px-1 rounded bg-gray-25 text-xs font-semibold text-gray-700 uppercase'>{method}</div>
-        <div className='pl-1'>
+        <div className='pl-1 pt-1'>
           <ReadonlyInputWithSelectVar
             value={url}
             nodeId={id}

From 2328944987881c95e7847ea670de6e8bf0be8232 Mon Sep 17 00:00:00 2001
From: zhuhao <37029601+hwzhuhao@users.noreply.github.com>
Date: Wed, 25 Sep 2024 14:48:06 +0800
Subject: [PATCH 061/235] chore: apply ruff reformat for python-client sdk
 (#8752)

---
 sdks/python-client/dify_client/client.py | 137 ++++++++---------------
 1 file changed, 47 insertions(+), 90 deletions(-)

diff --git a/sdks/python-client/dify_client/client.py b/sdks/python-client/dify_client/client.py
index 2be079bdf381c..5e42507a42abc 100644
--- a/sdks/python-client/dify_client/client.py
+++ b/sdks/python-client/dify_client/client.py
@@ -1,103 +1,80 @@
 import json
+
 import requests
 
 
 class DifyClient:
-    def __init__(self, api_key, base_url: str = 'https://api.dify.ai/v1'):
+    def __init__(self, api_key, base_url: str = "https://api.dify.ai/v1"):
         self.api_key = api_key
         self.base_url = base_url
 
     def _send_request(self, method, endpoint, json=None, params=None, stream=False):
-        headers = {
-            "Authorization": f"Bearer {self.api_key}",
-            "Content-Type": "application/json"
-        }
+        headers = {"Authorization": f"Bearer {self.api_key}", "Content-Type": "application/json"}
 
         url = f"{self.base_url}{endpoint}"
         response = requests.request(method, url, json=json, params=params, headers=headers, stream=stream)
 
         return response
-    
 
     def _send_request_with_files(self, method, endpoint, data, files):
-        headers = {
-            "Authorization": f"Bearer {self.api_key}"
-        }
+        headers = {"Authorization": f"Bearer {self.api_key}"}
 
         url = f"{self.base_url}{endpoint}"
         response = requests.request(method, url, data=data, headers=headers, files=files)
 
         return response
-    
+
     def message_feedback(self, message_id, rating, user):
-        data = {
-            "rating": rating,
-            "user": user
-        }
+        data = {"rating": rating, "user": user}
         return self._send_request("POST", f"/messages/{message_id}/feedbacks", data)
-    
+
     def get_application_parameters(self, user):
         params = {"user": user}
         return self._send_request("GET", "/parameters", params=params)
-    
+
     def file_upload(self, user, files):
-        data = {
-            "user": user
-        }
+        data = {"user": user}
         return self._send_request_with_files("POST", "/files/upload", data=data, files=files)
 
-    def text_to_audio(self, text:str, user:str, streaming:bool=False):
-        data = {
-            "text": text,
-            "user": user,
-            "streaming": streaming
-        }
+    def text_to_audio(self, text: str, user: str, streaming: bool = False):
+        data = {"text": text, "user": user, "streaming": streaming}
         return self._send_request("POST", "/text-to-audio", data=data)
-    
-    def get_meta(self,user):
-        params = { "user": user}
-        return self._send_request("GET", f"/meta", params=params)
+
+    def get_meta(self, user):
+        params = {"user": user}
+        return self._send_request("GET", "/meta", params=params)
 
 
 class CompletionClient(DifyClient):
     def create_completion_message(self, inputs, response_mode, user, files=None):
-        data = {
-            "inputs": inputs,
-            "response_mode": response_mode,
-            "user": user,
-            "files": files
-        }
-        return self._send_request("POST", "/completion-messages", data,
-                                  stream=True if response_mode == "streaming" else False)
+        data = {"inputs": inputs, "response_mode": response_mode, "user": user, "files": files}
+        return self._send_request(
+            "POST", "/completion-messages", data, stream=True if response_mode == "streaming" else False
+        )
 
 
 class ChatClient(DifyClient):
     def create_chat_message(self, inputs, query, user, response_mode="blocking", conversation_id=None, files=None):
-        data = {
-            "inputs": inputs,
-            "query": query,
-            "user": user,
-            "response_mode": response_mode,
-            "files": files
-        }
+        data = {"inputs": inputs, "query": query, "user": user, "response_mode": response_mode, "files": files}
         if conversation_id:
             data["conversation_id"] = conversation_id
 
-        return self._send_request("POST", "/chat-messages", data,
-                                  stream=True if response_mode == "streaming" else False)
-    
-    def get_suggested(self, message_id, user:str):
+        return self._send_request(
+            "POST", "/chat-messages", data, stream=True if response_mode == "streaming" else False
+        )
+
+    def get_suggested(self, message_id, user: str):
         params = {"user": user}
         return self._send_request("GET", f"/messages/{message_id}/suggested", params=params)
-    
+
     def stop_message(self, task_id, user):
         data = {"user": user}
-        return self._send_request("POST", f"/chat-messages/{task_id}/stop", data)   
+        return self._send_request("POST", f"/chat-messages/{task_id}/stop", data)
 
     def get_conversations(self, user, last_id=None, limit=None, pinned=None):
         params = {"user": user, "last_id": last_id, "limit": limit, "pinned": pinned}
         return self._send_request("GET", "/conversations", params=params)
-    
+
     def get_conversation_messages(self, user, conversation_id=None, first_id=None, limit=None):
         params = {"user": user}
 
@@ -109,15 +86,15 @@ def get_conversation_messages(self, user, conversation_id=None, first_id=None, l
             params["limit"] = limit
 
         return self._send_request("GET", "/messages", params=params)
-    
-    def rename_conversation(self, conversation_id, name,auto_generate:bool, user:str):
-        data = {"name": name, "auto_generate": auto_generate,"user": user}
+
+    def rename_conversation(self, conversation_id, name, auto_generate: bool, user: str):
+        data = {"name": name, "auto_generate": auto_generate, "user": user}
         return self._send_request("POST", f"/conversations/{conversation_id}/name", data)
 
     def delete_conversation(self, conversation_id, user):
         data = {"user": user}
         return self._send_request("DELETE", f"/conversations/{conversation_id}", data)
-    
+
     def audio_to_text(self, audio_file, user):
         data = {"user": user}
         files = {"audio_file": audio_file}
@@ -125,10 +102,10 @@ def audio_to_text(self, audio_file, user):
 
 
 class WorkflowClient(DifyClient):
-    def run(self, inputs:dict, response_mode:str="streaming", user:str="abc-123"):
+    def run(self, inputs: dict, response_mode: str = "streaming", user: str = "abc-123"):
         data = {"inputs": inputs, "response_mode": response_mode, "user": user}
         return self._send_request("POST", "/workflows/run", data)
-    
+
     def stop(self, task_id, user):
         data = {"user": user}
         return self._send_request("POST", f"/workflows/tasks/{task_id}/stop", data)
@@ -137,10 +114,8 @@ def get_result(self, workflow_run_id):
         return self._send_request("GET", f"/workflows/run/{workflow_run_id}")
 
 
-
 class KnowledgeBaseClient(DifyClient):
-
-    def __init__(self, api_key, base_url: str = 'https://api.dify.ai/v1', dataset_id: str = None):
+    def __init__(self, api_key, base_url: str = "https://api.dify.ai/v1", dataset_id: str = None):
         """
         Construct a KnowledgeBaseClient object.
 
@@ -150,10 +125,7 @@ def __init__(self, api_key, base_url: str = 'https://api.dify.ai/v1', dataset_id
             dataset_id (str, optional): ID of the dataset. Defaults to None. You don't need this if you just want to
                 create a new dataset. or list datasets. otherwise you need to set this.
         """
-        super().__init__(
-            api_key=api_key,
-            base_url=base_url
-        )
+        super().__init__(api_key=api_key, base_url=base_url)
         self.dataset_id = dataset_id
 
     def _get_dataset_id(self):
@@ -162,10 +134,10 @@ def _get_dataset_id(self):
         return self.dataset_id
 
     def create_dataset(self, name: str, **kwargs):
-        return self._send_request('POST', '/datasets', {'name': name}, **kwargs)
+        return self._send_request("POST", "/datasets", {"name": name}, **kwargs)
 
     def list_datasets(self, page: int = 1, page_size: int = 20, **kwargs):
-        return self._send_request('GET', f'/datasets?page={page}&limit={page_size}', **kwargs)
+        return self._send_request("GET", f"/datasets?page={page}&limit={page_size}", **kwargs)
 
     def create_document_by_text(self, name, text, extra_params: dict = None, **kwargs):
         """
@@ -193,14 +165,7 @@ def create_document_by_text(self, name, text, extra_params: dict = None, **kwarg
         }
         :return: Response from the API
         """
-        data = {
-            'indexing_technique': 'high_quality',
-            'process_rule': {
-                'mode': 'automatic'
-            },
-            'name': name,
-            'text': text
-        }
+        data = {"indexing_technique": "high_quality", "process_rule": {"mode": "automatic"}, "name": name, "text": text}
         if extra_params is not None and isinstance(extra_params, dict):
             data.update(extra_params)
         url = f"/datasets/{self._get_dataset_id()}/document/create_by_text"
@@ -233,10 +198,7 @@ def update_document_by_text(self, document_id, name, text, extra_params: dict =
         }
         :return: Response from the API
         """
-        data = {
-            'name': name,
-            'text': text
-        }
+        data = {"name": name, "text": text}
         if extra_params is not None and isinstance(extra_params, dict):
             data.update(extra_params)
         url = f"/datasets/{self._get_dataset_id()}/documents/{document_id}/update_by_text"
@@ -269,16 +231,11 @@ def create_document_by_file(self, file_path, original_document_id=None, extra_pa
         :return: Response from the API
         """
         files = {"file": open(file_path, "rb")}
-        data = {
-            'process_rule': {
-                'mode': 'automatic'
-            },
-            'indexing_technique': 'high_quality'
-        }
+        data = {"process_rule": {"mode": "automatic"}, "indexing_technique": "high_quality"}
         if extra_params is not None and isinstance(extra_params, dict):
             data.update(extra_params)
         if original_document_id is not None:
-            data['original_document_id'] = original_document_id
+            data["original_document_id"] = original_document_id
         url = f"/datasets/{self._get_dataset_id()}/document/create_by_file"
         return self._send_request_with_files("POST", url, {"data": json.dumps(data)}, files)
 
@@ -352,11 +309,11 @@ def list_documents(self, page: int = None, page_size: int = None, keyword: str =
         """
         params = {}
         if page is not None:
-            params['page'] = page
+            params["page"] = page
         if page_size is not None:
-            params['limit'] = page_size
+            params["limit"] = page_size
         if keyword is not None:
-            params['keyword'] = keyword
+            params["keyword"] = keyword
         url = f"/datasets/{self._get_dataset_id()}/documents"
         return self._send_request("GET", url, params=params, **kwargs)
 
@@ -383,9 +340,9 @@ def query_segments(self, document_id, keyword: str = None, status: str = None, *
         url = f"/datasets/{self._get_dataset_id()}/documents/{document_id}/segments"
         params = {}
         if keyword is not None:
-            params['keyword'] = keyword
+            params["keyword"] = keyword
         if status is not None:
-            params['status'] = status
+            params["status"] = status
         if "params" in kwargs:
             params.update(kwargs["params"])
         return self._send_request("GET", url, params=params, **kwargs)

From d0e0111f88da0fc972b07b1f97893df18cbc0522 Mon Sep 17 00:00:00 2001
From: cherryhuahua <68722306+cherryhuahua@users.noreply.github.com>
Date: Wed, 25 Sep 2024 14:51:42 +0800
Subject: [PATCH 062/235] fix:Spark's large language model token calculation
 error #7911 (#8755)

---
 api/core/app/apps/base_app_runner.py                    | 2 +-
 api/core/model_runtime/model_providers/spark/llm/llm.py | 9 ++++++---
 2 files changed, 7 insertions(+), 4 deletions(-)

diff --git a/api/core/app/apps/base_app_runner.py b/api/core/app/apps/base_app_runner.py
index 1b412b86396c2..203aca3384376 100644
--- a/api/core/app/apps/base_app_runner.py
+++ b/api/core/app/apps/base_app_runner.py
@@ -309,7 +309,7 @@ def _handle_invoke_result_stream(
             if not prompt_messages:
                 prompt_messages = result.prompt_messages
 
-            if not usage and result.delta.usage:
+            if result.delta.usage:
                 usage = result.delta.usage
 
         if not usage:
diff --git a/api/core/model_runtime/model_providers/spark/llm/llm.py b/api/core/model_runtime/model_providers/spark/llm/llm.py
index 57193dc0316b7..1181ba699af88 100644
--- a/api/core/model_runtime/model_providers/spark/llm/llm.py
+++ b/api/core/model_runtime/model_providers/spark/llm/llm.py
@@ -213,18 +213,21 @@ def _handle_generate_stream_response(
         :param prompt_messages: prompt messages
         :return: llm response chunk generator result
         """
+        completion = ""
         for index, content in enumerate(client.subscribe()):
             if isinstance(content, dict):
                 delta = content["data"]
             else:
                 delta = content
-
+            completion += delta
             assistant_prompt_message = AssistantPromptMessage(
                 content=delta or "",
             )
-
+            temp_assistant_prompt_message = AssistantPromptMessage(
+                content=completion,
+            )
             prompt_tokens = self.get_num_tokens(model, credentials, prompt_messages)
-            completion_tokens = self.get_num_tokens(model, credentials, [assistant_prompt_message])
+            completion_tokens = self.get_num_tokens(model, credentials, [temp_assistant_prompt_message])
 
             # transform usage
             usage = self._calc_response_usage(model, credentials, prompt_tokens, completion_tokens)

From b0927c39fb589b9df5c8278dc8bdac115497fe2e Mon Sep 17 00:00:00 2001
From: zhuiyue132 <fenda.app@qq.com>
Date: Wed, 25 Sep 2024 15:06:54 +0800
Subject: [PATCH 063/235] fix: expose the configuration of HTTP request node to
 Docker (#8716)

Co-authored-by: crazywoola <100913391+crazywoola@users.noreply.github.com>
---
 docker/.env.example        | 4 ++++
 docker/docker-compose.yaml | 2 ++
 2 files changed, 6 insertions(+)

diff --git a/docker/.env.example b/docker/.env.example
index 7eaaceb92834d..f7479791cea32 100644
--- a/docker/.env.example
+++ b/docker/.env.example
@@ -568,6 +568,10 @@ WORKFLOW_MAX_EXECUTION_STEPS=500
 WORKFLOW_MAX_EXECUTION_TIME=1200
 WORKFLOW_CALL_MAX_DEPTH=5
 
+# HTTP request node in workflow configuration
+HTTP_REQUEST_NODE_MAX_BINARY_SIZE=10485760
+HTTP_REQUEST_NODE_MAX_TEXT_SIZE=1048576
+
 # SSRF Proxy server HTTP URL
 SSRF_PROXY_HTTP_URL=http://ssrf_proxy:3128
 # SSRF Proxy server HTTPS URL
diff --git a/docker/docker-compose.yaml b/docker/docker-compose.yaml
index 16bef279bcc68..414919063a739 100644
--- a/docker/docker-compose.yaml
+++ b/docker/docker-compose.yaml
@@ -207,6 +207,8 @@ x-shared-env: &shared-api-worker-env
   WORKFLOW_CALL_MAX_DEPTH: ${WORKFLOW_MAX_EXECUTION_TIME:-5}
   SSRF_PROXY_HTTP_URL: ${SSRF_PROXY_HTTP_URL:-http://ssrf_proxy:3128}
   SSRF_PROXY_HTTPS_URL: ${SSRF_PROXY_HTTPS_URL:-http://ssrf_proxy:3128}
+  HTTP_REQUEST_NODE_MAX_BINARY_SIZE: ${HTTP_REQUEST_NODE_MAX_BINARY_SIZE:-10485760}
+  HTTP_REQUEST_NODE_MAX_TEXT_SIZE: ${HTTP_REQUEST_NODE_MAX_TEXT_SIZE:-1048576}
 
 services:
   # API service

From 2ef8b187fad67c248fc9ce20e440503ba24c0a42 Mon Sep 17 00:00:00 2001
From: Hash Brown <hi@xzd.me>
Date: Wed, 25 Sep 2024 15:50:51 +0800
Subject: [PATCH 064/235] Add GitHub Actions Workflow for Web Tests (#8753)

---
 .github/workflows/web-tests.yml | 46 +++++++++++++++++++++++++++++++++
 1 file changed, 46 insertions(+)
 create mode 100644 .github/workflows/web-tests.yml

diff --git a/.github/workflows/web-tests.yml b/.github/workflows/web-tests.yml
new file mode 100644
index 0000000000000..5aee64b8e6da0
--- /dev/null
+++ b/.github/workflows/web-tests.yml
@@ -0,0 +1,46 @@
+name: Web Tests
+
+on:
+  pull_request:
+    branches:
+      - main
+    paths:
+      - web/**
+
+concurrency:
+  group: web-tests-${{ github.head_ref || github.run_id }}
+  cancel-in-progress: true
+
+jobs:
+  test:
+    name: Web Tests
+    runs-on: ubuntu-latest
+    defaults:
+      run:
+        working-directory: ./web
+
+    steps:
+      - name: Checkout code
+        uses: actions/checkout@v4
+
+      - name: Check changed files
+        id: changed-files
+        uses: tj-actions/changed-files@v45
+        with:
+          files: web/**
+
+      - name: Setup Node.js
+        uses: actions/setup-node@v4
+        if: steps.changed-files.outputs.any_changed == 'true'
+        with:
+          node-version: 20
+          cache: yarn
+          cache-dependency-path: ./web/package.json
+
+      - name: Install dependencies
+        if: steps.changed-files.outputs.any_changed == 'true'
+        run: yarn install --frozen-lockfile
+
+      - name: Run tests
+        if: steps.changed-files.outputs.any_changed == 'true'
+        run: yarn test

From ef47f68e4ad5713b4694d9e0d0c646fd47efc55d Mon Sep 17 00:00:00 2001
From: NFish <douxc512@gmail.com>
Date: Wed, 25 Sep 2024 18:25:06 +0800
Subject: [PATCH 065/235] fix: the translation result may cause a different
 meaning (#8763)

---
 web/i18n/es-ES/common.ts   | 2 +-
 web/i18n/zh-Hans/common.ts | 2 +-
 web/i18n/zh-Hant/common.ts | 2 +-
 3 files changed, 3 insertions(+), 3 deletions(-)

diff --git a/web/i18n/es-ES/common.ts b/web/i18n/es-ES/common.ts
index 2ba907361fe8c..59a05f63d8a3e 100644
--- a/web/i18n/es-ES/common.ts
+++ b/web/i18n/es-ES/common.ts
@@ -202,7 +202,7 @@ const translation = {
     invitationLink: 'Enlace de invitación',
     failedInvitationEmails: 'Los siguientes usuarios no fueron invitados exitosamente',
     ok: 'OK',
-    removeFromTeam: 'Eliminar del equipo',
+    removeFromTeam: 'Eliminar del espacio de trabajo',
     removeFromTeamTip: 'Se eliminará el acceso al equipo',
     setAdmin: 'Establecer como administrador',
     setMember: 'Establecer como miembro ordinario',
diff --git a/web/i18n/zh-Hans/common.ts b/web/i18n/zh-Hans/common.ts
index 52ab7d6f02416..7947d32f251f1 100644
--- a/web/i18n/zh-Hans/common.ts
+++ b/web/i18n/zh-Hans/common.ts
@@ -200,7 +200,7 @@ const translation = {
     invitationLink: '邀请链接',
     failedInvitationEmails: '邀请以下邮箱失败',
     ok: '好的',
-    removeFromTeam: '移除团队',
+    removeFromTeam: '移出团队',
     removeFromTeamTip: '将取消团队访问',
     setAdmin: '设为管理员',
     setMember: '设为普通成员',
diff --git a/web/i18n/zh-Hant/common.ts b/web/i18n/zh-Hant/common.ts
index c1f3ed2b2b53e..8cd51b1991f1a 100644
--- a/web/i18n/zh-Hant/common.ts
+++ b/web/i18n/zh-Hant/common.ts
@@ -194,7 +194,7 @@ const translation = {
     invitationLink: '邀請連結',
     failedInvitationEmails: '邀請以下郵箱失敗',
     ok: '好的',
-    removeFromTeam: '移除團隊',
+    removeFromTeam: '移出團隊',
     removeFromTeamTip: '將取消團隊訪問',
     setAdmin: '設為管理員',
     setMember: '設為普通成員',

From 02ff6cca70421b6f2d482989a581a98117b64e7d Mon Sep 17 00:00:00 2001
From: "Pan, Wen-Ming" <pwm@google.com>
Date: Wed, 25 Sep 2024 21:27:26 +0800
Subject: [PATCH 066/235] feat: add support for Vertex AI Gemini 1.5 002 and
 experimental models (#8767)

---
 ...5-flash.yaml => gemini-1.5-flash-001.yaml} |  2 +-
 .../vertex_ai/llm/gemini-1.5-flash-002.yaml   | 37 +++++++++++++++++++
 ...i-1.5-pro.yaml => gemini-1.5-pro-001.yaml} |  2 +-
 .../vertex_ai/llm/gemini-1.5-pro-002.yaml     | 37 +++++++++++++++++++
 .../llm/gemini-flash-experimental.yaml        | 37 +++++++++++++++++++
 .../llm/gemini-pro-experimental.yaml          | 37 +++++++++++++++++++
 6 files changed, 150 insertions(+), 2 deletions(-)
 rename api/core/model_runtime/model_providers/vertex_ai/llm/{gemini-1.5-flash.yaml => gemini-1.5-flash-001.yaml} (96%)
 create mode 100644 api/core/model_runtime/model_providers/vertex_ai/llm/gemini-1.5-flash-002.yaml
 rename api/core/model_runtime/model_providers/vertex_ai/llm/{gemini-1.5-pro.yaml => gemini-1.5-pro-001.yaml} (96%)
 create mode 100644 api/core/model_runtime/model_providers/vertex_ai/llm/gemini-1.5-pro-002.yaml
 create mode 100644 api/core/model_runtime/model_providers/vertex_ai/llm/gemini-flash-experimental.yaml
 create mode 100644 api/core/model_runtime/model_providers/vertex_ai/llm/gemini-pro-experimental.yaml

diff --git a/api/core/model_runtime/model_providers/vertex_ai/llm/gemini-1.5-flash.yaml b/api/core/model_runtime/model_providers/vertex_ai/llm/gemini-1.5-flash-001.yaml
similarity index 96%
rename from api/core/model_runtime/model_providers/vertex_ai/llm/gemini-1.5-flash.yaml
rename to api/core/model_runtime/model_providers/vertex_ai/llm/gemini-1.5-flash-001.yaml
index c308f0a322fdd..f5386be06da6b 100644
--- a/api/core/model_runtime/model_providers/vertex_ai/llm/gemini-1.5-flash.yaml
+++ b/api/core/model_runtime/model_providers/vertex_ai/llm/gemini-1.5-flash-001.yaml
@@ -1,6 +1,6 @@
 model: gemini-1.5-flash-001
 label:
-  en_US: Gemini 1.5 Flash
+  en_US: Gemini 1.5 Flash 001
 model_type: llm
 features:
   - agent-thought
diff --git a/api/core/model_runtime/model_providers/vertex_ai/llm/gemini-1.5-flash-002.yaml b/api/core/model_runtime/model_providers/vertex_ai/llm/gemini-1.5-flash-002.yaml
new file mode 100644
index 0000000000000..97bd44f06b514
--- /dev/null
+++ b/api/core/model_runtime/model_providers/vertex_ai/llm/gemini-1.5-flash-002.yaml
@@ -0,0 +1,37 @@
+model: gemini-1.5-flash-002
+label:
+  en_US: Gemini 1.5 Flash 002
+model_type: llm
+features:
+  - agent-thought
+  - vision
+model_properties:
+  mode: chat
+  context_size: 1048576
+parameter_rules:
+  - name: temperature
+    use_template: temperature
+  - name: top_p
+    use_template: top_p
+  - name: top_k
+    label:
+      en_US: Top k
+    type: int
+    help:
+      en_US: Only sample from the top K options for each subsequent token.
+    required: false
+  - name: presence_penalty
+    use_template: presence_penalty
+  - name: frequency_penalty
+    use_template: frequency_penalty
+  - name: max_output_tokens
+    use_template: max_tokens
+    required: true
+    default: 8192
+    min: 1
+    max: 8192
+pricing:
+  input: '0.00'
+  output: '0.00'
+  unit: '0.000001'
+  currency: USD
diff --git a/api/core/model_runtime/model_providers/vertex_ai/llm/gemini-1.5-pro.yaml b/api/core/model_runtime/model_providers/vertex_ai/llm/gemini-1.5-pro-001.yaml
similarity index 96%
rename from api/core/model_runtime/model_providers/vertex_ai/llm/gemini-1.5-pro.yaml
rename to api/core/model_runtime/model_providers/vertex_ai/llm/gemini-1.5-pro-001.yaml
index 744863e7731e1..5e08f2294e2eb 100644
--- a/api/core/model_runtime/model_providers/vertex_ai/llm/gemini-1.5-pro.yaml
+++ b/api/core/model_runtime/model_providers/vertex_ai/llm/gemini-1.5-pro-001.yaml
@@ -1,6 +1,6 @@
 model: gemini-1.5-pro-001
 label:
-  en_US: Gemini 1.5 Pro
+  en_US: Gemini 1.5 Pro 001
 model_type: llm
 features:
   - agent-thought
diff --git a/api/core/model_runtime/model_providers/vertex_ai/llm/gemini-1.5-pro-002.yaml b/api/core/model_runtime/model_providers/vertex_ai/llm/gemini-1.5-pro-002.yaml
new file mode 100644
index 0000000000000..8f327ea2f3d37
--- /dev/null
+++ b/api/core/model_runtime/model_providers/vertex_ai/llm/gemini-1.5-pro-002.yaml
@@ -0,0 +1,37 @@
+model: gemini-1.5-pro-002
+label:
+  en_US: Gemini 1.5 Pro 002
+model_type: llm
+features:
+  - agent-thought
+  - vision
+model_properties:
+  mode: chat
+  context_size: 1048576
+parameter_rules:
+  - name: temperature
+    use_template: temperature
+  - name: top_p
+    use_template: top_p
+  - name: top_k
+    label:
+      en_US: Top k
+    type: int
+    help:
+      en_US: Only sample from the top K options for each subsequent token.
+    required: false
+  - name: presence_penalty
+    use_template: presence_penalty
+  - name: frequency_penalty
+    use_template: frequency_penalty
+  - name: max_output_tokens
+    use_template: max_tokens
+    required: true
+    default: 8192
+    min: 1
+    max: 8192
+pricing:
+  input: '0.00'
+  output: '0.00'
+  unit: '0.000001'
+  currency: USD
diff --git a/api/core/model_runtime/model_providers/vertex_ai/llm/gemini-flash-experimental.yaml b/api/core/model_runtime/model_providers/vertex_ai/llm/gemini-flash-experimental.yaml
new file mode 100644
index 0000000000000..0f5eb34c0cdf0
--- /dev/null
+++ b/api/core/model_runtime/model_providers/vertex_ai/llm/gemini-flash-experimental.yaml
@@ -0,0 +1,37 @@
+model: gemini-flash-experimental
+label:
+  en_US: Gemini Flash Experimental
+model_type: llm
+features:
+  - agent-thought
+  - vision
+model_properties:
+  mode: chat
+  context_size: 1048576
+parameter_rules:
+  - name: temperature
+    use_template: temperature
+  - name: top_p
+    use_template: top_p
+  - name: top_k
+    label:
+      en_US: Top k
+    type: int
+    help:
+      en_US: Only sample from the top K options for each subsequent token.
+    required: false
+  - name: presence_penalty
+    use_template: presence_penalty
+  - name: frequency_penalty
+    use_template: frequency_penalty
+  - name: max_output_tokens
+    use_template: max_tokens
+    required: true
+    default: 8192
+    min: 1
+    max: 8192
+pricing:
+  input: '0.00'
+  output: '0.00'
+  unit: '0.000001'
+  currency: USD
diff --git a/api/core/model_runtime/model_providers/vertex_ai/llm/gemini-pro-experimental.yaml b/api/core/model_runtime/model_providers/vertex_ai/llm/gemini-pro-experimental.yaml
new file mode 100644
index 0000000000000..fa31cabb85abb
--- /dev/null
+++ b/api/core/model_runtime/model_providers/vertex_ai/llm/gemini-pro-experimental.yaml
@@ -0,0 +1,37 @@
+model: gemini-pro-experimental
+label:
+  en_US: Gemini Pro Experimental
+model_type: llm
+features:
+  - agent-thought
+  - vision
+model_properties:
+  mode: chat
+  context_size: 1048576
+parameter_rules:
+  - name: temperature
+    use_template: temperature
+  - name: top_p
+    use_template: top_p
+  - name: top_k
+    label:
+      en_US: Top k
+    type: int
+    help:
+      en_US: Only sample from the top K options for each subsequent token.
+    required: false
+  - name: presence_penalty
+    use_template: presence_penalty
+  - name: frequency_penalty
+    use_template: frequency_penalty
+  - name: max_output_tokens
+    use_template: max_tokens
+    required: true
+    default: 8192
+    min: 1
+    max: 8192
+pricing:
+  input: '0.00'
+  output: '0.00'
+  unit: '0.000001'
+  currency: USD

From a8b837c4a9143385718f87fdd4863844607daae3 Mon Sep 17 00:00:00 2001
From: Bowen Liang <liangbowen@gf.com.cn>
Date: Wed, 25 Sep 2024 22:55:24 +0800
Subject: [PATCH 067/235] dep: bump ElasticSearch from 8.14.x to 8.15.x (#8197)

---
 api/poetry.lock            | 13 ++++++++-----
 api/pyproject.toml         |  2 +-
 docker/docker-compose.yaml |  4 ++--
 3 files changed, 11 insertions(+), 8 deletions(-)

diff --git a/api/poetry.lock b/api/poetry.lock
index 184cdb9e81e57..bce21fb547c8c 100644
--- a/api/poetry.lock
+++ b/api/poetry.lock
@@ -2333,13 +2333,13 @@ develop = ["aiohttp", "furo", "httpx", "opentelemetry-api", "opentelemetry-sdk",
 
 [[package]]
 name = "elasticsearch"
-version = "8.14.0"
+version = "8.15.1"
 description = "Python client for Elasticsearch"
 optional = false
-python-versions = ">=3.7"
+python-versions = ">=3.8"
 files = [
-    {file = "elasticsearch-8.14.0-py3-none-any.whl", hash = "sha256:cef8ef70a81af027f3da74a4f7d9296b390c636903088439087b8262a468c130"},
-    {file = "elasticsearch-8.14.0.tar.gz", hash = "sha256:aa2490029dd96f4015b333c1827aa21fd6c0a4d223b00dfb0fe933b8d09a511b"},
+    {file = "elasticsearch-8.15.1-py3-none-any.whl", hash = "sha256:02a0476e98768a30d7926335fc0d305c04fdb928eea1354c6e6040d8c2814569"},
+    {file = "elasticsearch-8.15.1.tar.gz", hash = "sha256:40c0d312f8adf8bdc81795bc16a0b546ddf544cb1f90e829a244e4780c4dbfd8"},
 ]
 
 [package.dependencies]
@@ -2347,7 +2347,10 @@ elastic-transport = ">=8.13,<9"
 
 [package.extras]
 async = ["aiohttp (>=3,<4)"]
+dev = ["aiohttp", "black", "build", "coverage", "isort", "jinja2", "mapbox-vector-tile", "nox", "numpy", "orjson", "pandas", "pyarrow", "pytest", "pytest-asyncio", "pytest-cov", "python-dateutil", "pyyaml (>=5.4)", "requests (>=2,<3)", "simsimd", "twine", "unasync"]
+docs = ["sphinx", "sphinx-autodoc-typehints", "sphinx-rtd-theme (>=2.0)"]
 orjson = ["orjson (>=3)"]
+pyarrow = ["pyarrow (>=1)"]
 requests = ["requests (>=2.4.0,!=2.32.2,<3.0.0)"]
 vectorstore-mmr = ["numpy (>=1)", "simsimd (>=3)"]
 
@@ -10498,4 +10501,4 @@ cffi = ["cffi (>=1.11)"]
 [metadata]
 lock-version = "2.0"
 python-versions = ">=3.10,<3.13"
-content-hash = "17c4108d92c415d987f8b437ea3e0484c5601a05bfe175339a8546c93c159bc5"
+content-hash = "69b42bb1ff033f14e199fee8335356275099421d72bbd7037b7a991ea65cae08"
diff --git a/api/pyproject.toml b/api/pyproject.toml
index 9e38c0945601c..f004865d5f343 100644
--- a/api/pyproject.toml
+++ b/api/pyproject.toml
@@ -253,7 +253,7 @@ alibabacloud_gpdb20160503 = "~3.8.0"
 alibabacloud_tea_openapi = "~0.3.9"
 chromadb = "0.5.1"
 clickhouse-connect = "~0.7.16"
-elasticsearch = "8.14.0"
+elasticsearch = "~8.15.1"
 oracledb = "~2.2.1"
 pgvecto-rs = { version = "~0.2.1", extras = ['sqlalchemy'] }
 pgvector = "0.2.5"
diff --git a/docker/docker-compose.yaml b/docker/docker-compose.yaml
index 414919063a739..95e271a0e9e4f 100644
--- a/docker/docker-compose.yaml
+++ b/docker/docker-compose.yaml
@@ -630,7 +630,7 @@ services:
   # https://www.elastic.co/guide/en/elasticsearch/reference/current/settings.html
   # https://www.elastic.co/guide/en/elasticsearch/reference/current/docker.html#docker-prod-prerequisites
   elasticsearch:
-    image: docker.elastic.co/elasticsearch/elasticsearch:8.14.3
+    image: docker.elastic.co/elasticsearch/elasticsearch:8.15.1
     container_name: elasticsearch
     profiles:
       - elasticsearch
@@ -657,7 +657,7 @@ services:
   # https://www.elastic.co/guide/en/kibana/current/docker.html
   # https://www.elastic.co/guide/en/kibana/current/settings.html
   kibana:
-    image: docker.elastic.co/kibana/kibana:8.14.3
+    image: docker.elastic.co/kibana/kibana:8.15.1
     container_name: kibana
     profiles:
       - elasticsearch

From fefbc43fb035fca2171d949ac5b4d261c343d8b1 Mon Sep 17 00:00:00 2001
From: Qun <51054082+QunBB@users.noreply.github.com>
Date: Thu, 26 Sep 2024 08:18:13 +0800
Subject: [PATCH 068/235] chore: fix comfyui tool doc url (#8775)

---
 api/core/tools/provider/builtin/comfyui/comfyui.yaml | 2 +-
 1 file changed, 1 insertion(+), 1 deletion(-)

diff --git a/api/core/tools/provider/builtin/comfyui/comfyui.yaml b/api/core/tools/provider/builtin/comfyui/comfyui.yaml
index 066fd85308281..3891eebf3ac7e 100644
--- a/api/core/tools/provider/builtin/comfyui/comfyui.yaml
+++ b/api/core/tools/provider/builtin/comfyui/comfyui.yaml
@@ -39,4 +39,4 @@ credentials_for_provider:
       en_US: The checkpoint name of the ComfyUI server, e.g. xxx.safetensors
       zh_Hans: ComfyUI服务器的模型名称, 比如 xxx.safetensors
       pt_BR: The checkpoint name of the ComfyUI server, e.g. xxx.safetensors
-    url: https://docs.dify.ai/tutorials/tool-configuration/comfyui
+    url: https://github.com/comfyanonymous/ComfyUI#installing

From 5ba19d64e90a9d5c94b8d4e6cd93cd4043fc8943 Mon Sep 17 00:00:00 2001
From: =?UTF-8?q?=E9=9D=9E=E6=B3=95=E6=93=8D=E4=BD=9C?= <hjlarry@163.com>
Date: Thu, 26 Sep 2024 11:22:18 +0800
Subject: [PATCH 069/235] fix: TavilySearch tool get api link (#8780)

---
 api/core/tools/provider/builtin/tavily/tavily.yaml | 2 +-
 1 file changed, 1 insertion(+), 1 deletion(-)

diff --git a/api/core/tools/provider/builtin/tavily/tavily.yaml b/api/core/tools/provider/builtin/tavily/tavily.yaml
index 7b25a8184857c..95820f4d18b05 100644
--- a/api/core/tools/provider/builtin/tavily/tavily.yaml
+++ b/api/core/tools/provider/builtin/tavily/tavily.yaml
@@ -28,4 +28,4 @@ credentials_for_provider:
       en_US: Get your Tavily API key from Tavily
       zh_Hans: 从 TavilyApi 获取您的 Tavily API key
       pt_BR: Get your Tavily API key from Tavily
-    url: https://docs.tavily.com/docs/tavily-api/introduction
+    url: https://docs.tavily.com/docs/welcome

From ac737637264f4caf10866ab3953cd5c203961939 Mon Sep 17 00:00:00 2001
From: zhuhao <37029601+hwzhuhao@users.noreply.github.com>
Date: Thu, 26 Sep 2024 11:23:09 +0800
Subject: [PATCH 070/235] chore: add input_type param desc for the _invoke
 method of text_embedding (#8778)

---
 .../azure_openai/text_embedding/text_embedding.py      | 10 ++++++++++
 .../baichuan/text_embedding/text_embedding.py          |  1 +
 .../bedrock/text_embedding/text_embedding.py           |  1 +
 .../cohere/text_embedding/text_embedding.py            |  1 +
 .../huggingface_hub/text_embedding/text_embedding.py   | 10 ++++++++++
 .../huggingface_tei/text_embedding/text_embedding.py   |  1 +
 .../hunyuan/text_embedding/text_embedding.py           |  1 +
 .../jina/text_embedding/text_embedding.py              |  1 +
 .../localai/text_embedding/text_embedding.py           |  3 ++-
 .../minimax/text_embedding/text_embedding.py           |  1 +
 .../mixedbread/text_embedding/text_embedding.py        |  1 +
 .../nomic/text_embedding/text_embedding.py             |  1 +
 .../nvidia/text_embedding/text_embedding.py            |  1 +
 .../ollama/text_embedding/text_embedding.py            |  1 +
 .../openai/text_embedding/text_embedding.py            |  1 +
 .../text_embedding/text_embedding.py                   |  1 +
 .../openllm/text_embedding/text_embedding.py           |  1 +
 .../replicate/text_embedding/text_embedding.py         | 10 ++++++++++
 .../sagemaker/text_embedding/text_embedding.py         |  1 +
 .../siliconflow/text_embedding/text_embedding.py       | 10 ++++++++++
 .../tongyi/text_embedding/text_embedding.py            |  1 +
 .../upstage/text_embedding/text_embedding.py           |  1 +
 .../vertex_ai/text_embedding/text_embedding.py         |  2 ++
 .../volcengine_maas/text_embedding/text_embedding.py   |  1 +
 .../wenxin/text_embedding/text_embedding.py            |  1 +
 .../zhipuai/text_embedding/text_embedding.py           |  1 +
 26 files changed, 64 insertions(+), 1 deletion(-)

diff --git a/api/core/model_runtime/model_providers/azure_openai/text_embedding/text_embedding.py b/api/core/model_runtime/model_providers/azure_openai/text_embedding/text_embedding.py
index 6b270b65ff00a..8701a3805002e 100644
--- a/api/core/model_runtime/model_providers/azure_openai/text_embedding/text_embedding.py
+++ b/api/core/model_runtime/model_providers/azure_openai/text_embedding/text_embedding.py
@@ -25,6 +25,16 @@ def _invoke(
         user: Optional[str] = None,
         input_type: EmbeddingInputType = EmbeddingInputType.DOCUMENT,
     ) -> TextEmbeddingResult:
+        """
+        Invoke text embedding model
+
+        :param model: model name
+        :param credentials: model credentials
+        :param texts: texts to embed
+        :param user: unique user id
+        :param input_type: input type
+        :return: embeddings result
+        """
         base_model_name = credentials["base_model_name"]
         credentials_kwargs = self._to_credential_kwargs(credentials)
         client = AzureOpenAI(**credentials_kwargs)
diff --git a/api/core/model_runtime/model_providers/baichuan/text_embedding/text_embedding.py b/api/core/model_runtime/model_providers/baichuan/text_embedding/text_embedding.py
index 210c274bdf5e3..56b9be1c36534 100644
--- a/api/core/model_runtime/model_providers/baichuan/text_embedding/text_embedding.py
+++ b/api/core/model_runtime/model_providers/baichuan/text_embedding/text_embedding.py
@@ -50,6 +50,7 @@ def _invoke(
         :param credentials: model credentials
         :param texts: texts to embed
         :param user: unique user id
+        :param input_type: input type
         :return: embeddings result
         """
         api_key = credentials["api_key"]
diff --git a/api/core/model_runtime/model_providers/bedrock/text_embedding/text_embedding.py b/api/core/model_runtime/model_providers/bedrock/text_embedding/text_embedding.py
index 8c4c50b26958c..d9c5726592113 100644
--- a/api/core/model_runtime/model_providers/bedrock/text_embedding/text_embedding.py
+++ b/api/core/model_runtime/model_providers/bedrock/text_embedding/text_embedding.py
@@ -45,6 +45,7 @@ def _invoke(
         :param credentials: model credentials
         :param texts: texts to embed
         :param user: unique user id
+        :param input_type: input type
         :return: embeddings result
         """
         client_config = Config(region_name=credentials["aws_region"])
diff --git a/api/core/model_runtime/model_providers/cohere/text_embedding/text_embedding.py b/api/core/model_runtime/model_providers/cohere/text_embedding/text_embedding.py
index 1f93068a8c9e5..4da20806904ba 100644
--- a/api/core/model_runtime/model_providers/cohere/text_embedding/text_embedding.py
+++ b/api/core/model_runtime/model_providers/cohere/text_embedding/text_embedding.py
@@ -40,6 +40,7 @@ def _invoke(
         :param credentials: model credentials
         :param texts: texts to embed
         :param user: unique user id
+        :param input_type: input type
         :return: embeddings result
         """
         # get model properties
diff --git a/api/core/model_runtime/model_providers/huggingface_hub/text_embedding/text_embedding.py b/api/core/model_runtime/model_providers/huggingface_hub/text_embedding/text_embedding.py
index cf18f84ac8cb2..b2e6d1b6520c7 100644
--- a/api/core/model_runtime/model_providers/huggingface_hub/text_embedding/text_embedding.py
+++ b/api/core/model_runtime/model_providers/huggingface_hub/text_embedding/text_embedding.py
@@ -26,6 +26,16 @@ def _invoke(
         user: Optional[str] = None,
         input_type: EmbeddingInputType = EmbeddingInputType.DOCUMENT,
     ) -> TextEmbeddingResult:
+        """
+        Invoke text embedding model
+
+        :param model: model name
+        :param credentials: model credentials
+        :param texts: texts to embed
+        :param user: unique user id
+        :param input_type: input type
+        :return: embeddings result
+        """
         client = InferenceClient(token=credentials["huggingfacehub_api_token"])
 
         execute_model = model
diff --git a/api/core/model_runtime/model_providers/huggingface_tei/text_embedding/text_embedding.py b/api/core/model_runtime/model_providers/huggingface_tei/text_embedding/text_embedding.py
index 58baf4933c241..b8ff3ca549a63 100644
--- a/api/core/model_runtime/model_providers/huggingface_tei/text_embedding/text_embedding.py
+++ b/api/core/model_runtime/model_providers/huggingface_tei/text_embedding/text_embedding.py
@@ -44,6 +44,7 @@ def _invoke(
         :param credentials: model credentials
         :param texts: texts to embed
         :param user: unique user id
+        :param input_type: input type
         :return: embeddings result
         """
         server_url = credentials["server_url"]
diff --git a/api/core/model_runtime/model_providers/hunyuan/text_embedding/text_embedding.py b/api/core/model_runtime/model_providers/hunyuan/text_embedding/text_embedding.py
index 3e14371f897e0..75701ebc54a74 100644
--- a/api/core/model_runtime/model_providers/hunyuan/text_embedding/text_embedding.py
+++ b/api/core/model_runtime/model_providers/hunyuan/text_embedding/text_embedding.py
@@ -41,6 +41,7 @@ def _invoke(
         :param credentials: model credentials
         :param texts: texts to embed
         :param user: unique user id
+        :param input_type: input type
         :return: embeddings result
         """
 
diff --git a/api/core/model_runtime/model_providers/jina/text_embedding/text_embedding.py b/api/core/model_runtime/model_providers/jina/text_embedding/text_embedding.py
index 9120f26b8d92f..c7b729c14eaac 100644
--- a/api/core/model_runtime/model_providers/jina/text_embedding/text_embedding.py
+++ b/api/core/model_runtime/model_providers/jina/text_embedding/text_embedding.py
@@ -75,6 +75,7 @@ def _invoke(
         :param credentials: model credentials
         :param texts: texts to embed
         :param user: unique user id
+        :param input_type: input type
         :return: embeddings result
         """
         api_key = credentials["api_key"]
diff --git a/api/core/model_runtime/model_providers/localai/text_embedding/text_embedding.py b/api/core/model_runtime/model_providers/localai/text_embedding/text_embedding.py
index d8878c7be85bc..ab8ca76c2f097 100644
--- a/api/core/model_runtime/model_providers/localai/text_embedding/text_embedding.py
+++ b/api/core/model_runtime/model_providers/localai/text_embedding/text_embedding.py
@@ -23,7 +23,7 @@
 
 class LocalAITextEmbeddingModel(TextEmbeddingModel):
     """
-    Model class for Jina text embedding model.
+    Model class for LocalAI text embedding model.
     """
 
     def _invoke(
@@ -41,6 +41,7 @@ def _invoke(
         :param credentials: model credentials
         :param texts: texts to embed
         :param user: unique user id
+        :param input_type: input type
         :return: embeddings result
         """
         if len(texts) != 1:
diff --git a/api/core/model_runtime/model_providers/minimax/text_embedding/text_embedding.py b/api/core/model_runtime/model_providers/minimax/text_embedding/text_embedding.py
index d0d1d2aea1ccf..74d2a221d1b57 100644
--- a/api/core/model_runtime/model_providers/minimax/text_embedding/text_embedding.py
+++ b/api/core/model_runtime/model_providers/minimax/text_embedding/text_embedding.py
@@ -49,6 +49,7 @@ def _invoke(
         :param credentials: model credentials
         :param texts: texts to embed
         :param user: unique user id
+        :param input_type: input type
         :return: embeddings result
         """
         api_key = credentials["minimax_api_key"]
diff --git a/api/core/model_runtime/model_providers/mixedbread/text_embedding/text_embedding.py b/api/core/model_runtime/model_providers/mixedbread/text_embedding/text_embedding.py
index cdc2d58d0c7e3..68b7b448bfec7 100644
--- a/api/core/model_runtime/model_providers/mixedbread/text_embedding/text_embedding.py
+++ b/api/core/model_runtime/model_providers/mixedbread/text_embedding/text_embedding.py
@@ -42,6 +42,7 @@ def _invoke(
         :param credentials: model credentials
         :param texts: texts to embed
         :param user: unique user id
+        :param input_type: input type
         :return: embeddings result
         """
         api_key = credentials["api_key"]
diff --git a/api/core/model_runtime/model_providers/nomic/text_embedding/text_embedding.py b/api/core/model_runtime/model_providers/nomic/text_embedding/text_embedding.py
index a797521576b23..857dfb5f41e2f 100644
--- a/api/core/model_runtime/model_providers/nomic/text_embedding/text_embedding.py
+++ b/api/core/model_runtime/model_providers/nomic/text_embedding/text_embedding.py
@@ -56,6 +56,7 @@ def _invoke(
         :param credentials: model credentials
         :param texts: texts to embed
         :param user: unique user id
+        :param input_type: input type
         :return: embeddings result
         """
         embeddings, prompt_tokens, total_tokens = self.embed_text(
diff --git a/api/core/model_runtime/model_providers/nvidia/text_embedding/text_embedding.py b/api/core/model_runtime/model_providers/nvidia/text_embedding/text_embedding.py
index a4ea28bd1073b..936ceb8dd2c60 100644
--- a/api/core/model_runtime/model_providers/nvidia/text_embedding/text_embedding.py
+++ b/api/core/model_runtime/model_providers/nvidia/text_embedding/text_embedding.py
@@ -42,6 +42,7 @@ def _invoke(
         :param credentials: model credentials
         :param texts: texts to embed
         :param user: unique user id
+        :param input_type: input type
         :return: embeddings result
         """
         api_key = credentials["api_key"]
diff --git a/api/core/model_runtime/model_providers/ollama/text_embedding/text_embedding.py b/api/core/model_runtime/model_providers/ollama/text_embedding/text_embedding.py
index 0501c8b841666..5cf3f1c6fa87f 100644
--- a/api/core/model_runtime/model_providers/ollama/text_embedding/text_embedding.py
+++ b/api/core/model_runtime/model_providers/ollama/text_embedding/text_embedding.py
@@ -53,6 +53,7 @@ def _invoke(
         :param credentials: model credentials
         :param texts: texts to embed
         :param user: unique user id
+        :param input_type: input type
         :return: embeddings result
         """
 
diff --git a/api/core/model_runtime/model_providers/openai/text_embedding/text_embedding.py b/api/core/model_runtime/model_providers/openai/text_embedding/text_embedding.py
index 7945723636ced..16f1a0cfa1117 100644
--- a/api/core/model_runtime/model_providers/openai/text_embedding/text_embedding.py
+++ b/api/core/model_runtime/model_providers/openai/text_embedding/text_embedding.py
@@ -34,6 +34,7 @@ def _invoke(
         :param credentials: model credentials
         :param texts: texts to embed
         :param user: unique user id
+        :param input_type: input type
         :return: embeddings result
         """
         # transform credentials to kwargs for model instance
diff --git a/api/core/model_runtime/model_providers/openai_api_compatible/text_embedding/text_embedding.py b/api/core/model_runtime/model_providers/openai_api_compatible/text_embedding/text_embedding.py
index 68b5773e16afe..64fa6aaa3c5a7 100644
--- a/api/core/model_runtime/model_providers/openai_api_compatible/text_embedding/text_embedding.py
+++ b/api/core/model_runtime/model_providers/openai_api_compatible/text_embedding/text_embedding.py
@@ -43,6 +43,7 @@ def _invoke(
         :param credentials: model credentials
         :param texts: texts to embed
         :param user: unique user id
+        :param input_type: input type
         :return: embeddings result
         """
 
diff --git a/api/core/model_runtime/model_providers/openllm/text_embedding/text_embedding.py b/api/core/model_runtime/model_providers/openllm/text_embedding/text_embedding.py
index c0a3efbb003dc..c5d4330912782 100644
--- a/api/core/model_runtime/model_providers/openllm/text_embedding/text_embedding.py
+++ b/api/core/model_runtime/model_providers/openllm/text_embedding/text_embedding.py
@@ -40,6 +40,7 @@ def _invoke(
         :param credentials: model credentials
         :param texts: texts to embed
         :param user: unique user id
+        :param input_type: input type
         :return: embeddings result
         """
         server_url = credentials["server_url"]
diff --git a/api/core/model_runtime/model_providers/replicate/text_embedding/text_embedding.py b/api/core/model_runtime/model_providers/replicate/text_embedding/text_embedding.py
index b6cf89bcd9868..9f724a77ac040 100644
--- a/api/core/model_runtime/model_providers/replicate/text_embedding/text_embedding.py
+++ b/api/core/model_runtime/model_providers/replicate/text_embedding/text_embedding.py
@@ -22,6 +22,16 @@ def _invoke(
         user: Optional[str] = None,
         input_type: EmbeddingInputType = EmbeddingInputType.DOCUMENT,
     ) -> TextEmbeddingResult:
+        """
+        Invoke text embedding model
+
+        :param model: model name
+        :param credentials: model credentials
+        :param texts: texts to embed
+        :param user: unique user id
+        :param input_type: input type
+        :return: embeddings result
+        """
         client = ReplicateClient(api_token=credentials["replicate_api_token"], timeout=30)
 
         if "model_version" in credentials:
diff --git a/api/core/model_runtime/model_providers/sagemaker/text_embedding/text_embedding.py b/api/core/model_runtime/model_providers/sagemaker/text_embedding/text_embedding.py
index 957f2e5d0f78f..8f993ce672252 100644
--- a/api/core/model_runtime/model_providers/sagemaker/text_embedding/text_embedding.py
+++ b/api/core/model_runtime/model_providers/sagemaker/text_embedding/text_embedding.py
@@ -68,6 +68,7 @@ def _invoke(
         :param credentials: model credentials
         :param texts: texts to embed
         :param user: unique user id
+        :param input_type: input type
         :return: embeddings result
         """
         # get model properties
diff --git a/api/core/model_runtime/model_providers/siliconflow/text_embedding/text_embedding.py b/api/core/model_runtime/model_providers/siliconflow/text_embedding/text_embedding.py
index c6c681c15d9d4..c5dcc126107aa 100644
--- a/api/core/model_runtime/model_providers/siliconflow/text_embedding/text_embedding.py
+++ b/api/core/model_runtime/model_providers/siliconflow/text_embedding/text_embedding.py
@@ -24,6 +24,16 @@ def _invoke(
         user: Optional[str] = None,
         input_type: EmbeddingInputType = EmbeddingInputType.DOCUMENT,
     ) -> TextEmbeddingResult:
+        """
+        Invoke text embedding model
+
+        :param model: model name
+        :param credentials: model credentials
+        :param texts: texts to embed
+        :param user: unique user id
+        :param input_type: input type
+        :return: embeddings result
+        """
         self._add_custom_parameters(credentials)
         return super()._invoke(model, credentials, texts, user)
 
diff --git a/api/core/model_runtime/model_providers/tongyi/text_embedding/text_embedding.py b/api/core/model_runtime/model_providers/tongyi/text_embedding/text_embedding.py
index 0eef0db3e7df0..736cd44df8888 100644
--- a/api/core/model_runtime/model_providers/tongyi/text_embedding/text_embedding.py
+++ b/api/core/model_runtime/model_providers/tongyi/text_embedding/text_embedding.py
@@ -37,6 +37,7 @@ def _invoke(
         :param credentials: model credentials
         :param texts: texts to embed
         :param user: unique user id
+        :param input_type: input type
         :return: embeddings result
         """
         credentials_kwargs = self._to_credential_kwargs(credentials)
diff --git a/api/core/model_runtime/model_providers/upstage/text_embedding/text_embedding.py b/api/core/model_runtime/model_providers/upstage/text_embedding/text_embedding.py
index 812bf92eeace7..b6509cd26cfa2 100644
--- a/api/core/model_runtime/model_providers/upstage/text_embedding/text_embedding.py
+++ b/api/core/model_runtime/model_providers/upstage/text_embedding/text_embedding.py
@@ -38,6 +38,7 @@ def _invoke(
         :param credentials: model credentials
         :param texts: texts to embed
         :param user: unique user id
+        :param input_type: input type
         :return: embeddings result
         """
 
diff --git a/api/core/model_runtime/model_providers/vertex_ai/text_embedding/text_embedding.py b/api/core/model_runtime/model_providers/vertex_ai/text_embedding/text_embedding.py
index 509b41d9517bf..fce9544df0a41 100644
--- a/api/core/model_runtime/model_providers/vertex_ai/text_embedding/text_embedding.py
+++ b/api/core/model_runtime/model_providers/vertex_ai/text_embedding/text_embedding.py
@@ -44,6 +44,8 @@ def _invoke(
         :param model: model name
         :param credentials: model credentials
         :param texts: texts to embed
+        :param user: unique user id
+        :param input_type: input type
         :return: embeddings result
         """
         service_account_info = json.loads(base64.b64decode(credentials["vertex_service_account_key"]))
diff --git a/api/core/model_runtime/model_providers/volcengine_maas/text_embedding/text_embedding.py b/api/core/model_runtime/model_providers/volcengine_maas/text_embedding/text_embedding.py
index 9d800af5f4d38..0dd4037c95856 100644
--- a/api/core/model_runtime/model_providers/volcengine_maas/text_embedding/text_embedding.py
+++ b/api/core/model_runtime/model_providers/volcengine_maas/text_embedding/text_embedding.py
@@ -56,6 +56,7 @@ def _invoke(
         :param credentials: model credentials
         :param texts: texts to embed
         :param user: unique user id
+        :param input_type: input type
         :return: embeddings result
         """
         if ArkClientV3.is_legacy(credentials):
diff --git a/api/core/model_runtime/model_providers/wenxin/text_embedding/text_embedding.py b/api/core/model_runtime/model_providers/wenxin/text_embedding/text_embedding.py
index 1b5a0904db963..c21d0c055277f 100644
--- a/api/core/model_runtime/model_providers/wenxin/text_embedding/text_embedding.py
+++ b/api/core/model_runtime/model_providers/wenxin/text_embedding/text_embedding.py
@@ -85,6 +85,7 @@ def _invoke(
         :param credentials: model credentials
         :param texts: texts to embed
         :param user: unique user id
+        :param input_type: input type
         :return: embeddings result
         """
 
diff --git a/api/core/model_runtime/model_providers/zhipuai/text_embedding/text_embedding.py b/api/core/model_runtime/model_providers/zhipuai/text_embedding/text_embedding.py
index 707c08ef1b5f6..14a529dddf82d 100644
--- a/api/core/model_runtime/model_providers/zhipuai/text_embedding/text_embedding.py
+++ b/api/core/model_runtime/model_providers/zhipuai/text_embedding/text_embedding.py
@@ -30,6 +30,7 @@ def _invoke(
         :param credentials: model credentials
         :param texts: texts to embed
         :param user: unique user id
+        :param input_type: input type
         :return: embeddings result
         """
         credentials_kwargs = self._to_credential_kwargs(credentials)

From 0c96f0aa51e65d970d26f043be1f592f89dba507 Mon Sep 17 00:00:00 2001
From: =?UTF-8?q?=E9=9D=9E=E6=B3=95=E6=93=8D=E4=BD=9C?= <hjlarry@163.com>
Date: Thu, 26 Sep 2024 11:24:03 +0800
Subject: [PATCH 071/235] fix: credential *** should be string (#8785)

---
 api/core/tools/provider/tool_provider.py | 3 +++
 1 file changed, 3 insertions(+)

diff --git a/api/core/tools/provider/tool_provider.py b/api/core/tools/provider/tool_provider.py
index 05c88b904e4a6..321b21201414e 100644
--- a/api/core/tools/provider/tool_provider.py
+++ b/api/core/tools/provider/tool_provider.py
@@ -153,6 +153,9 @@ def validate_credentials_format(self, credentials: dict[str, Any]) -> None:
 
             # check type
             credential_schema = credentials_need_to_validate[credential_name]
+            if not credential_schema.required and credentials[credential_name] is None:
+                continue
+
             if credential_schema.type in {
                 ToolProviderCredentials.CredentialsType.SECRET_INPUT,
                 ToolProviderCredentials.CredentialsType.TEXT_INPUT,

From 4c9ef6e8301b2986653f11ffad837cf9e45f3284 Mon Sep 17 00:00:00 2001
From: Aaron Ji <127167174+DresAaron@users.noreply.github.com>
Date: Thu, 26 Sep 2024 11:29:35 +0800
Subject: [PATCH 072/235] fix: update usage for Jina Embeddings v3 (#8771)

---
 .../model_providers/jina/jina.yaml            | 43 -------------------
 .../jina/text_embedding/text_embedding.py     | 20 +++------
 2 files changed, 6 insertions(+), 57 deletions(-)

diff --git a/api/core/model_runtime/model_providers/jina/jina.yaml b/api/core/model_runtime/model_providers/jina/jina.yaml
index 4ff6ba0f22c84..970b22965b5d2 100644
--- a/api/core/model_runtime/model_providers/jina/jina.yaml
+++ b/api/core/model_runtime/model_providers/jina/jina.yaml
@@ -67,46 +67,3 @@ model_credential_schema:
       required: false
       type: text-input
       default: '8192'
-    - variable: task
-      label:
-        zh_Hans: 下游任务
-        en_US: Downstream task
-      placeholder:
-        zh_Hans: 选择将使用向量模型的下游任务。模型将返回针对该任务优化的向量。
-        en_US: Select the downstream task for which the embeddings will be used. The model will return the optimized embeddings for that task.
-      required: false
-      type: select
-      options:
-        - value: retrieval.query
-          label:
-            en_US: retrieval.query
-        - value: retrieval.passage
-          label:
-            en_US: retrieval.passage
-        - value: separation
-          label:
-            en_US: separation
-        - value: classification
-          label:
-            en_US: classification
-        - value: text-matching
-          label:
-            en_US: text-matching
-    - variable: dimensions
-      label:
-        zh_Hans: 输出维度
-        en_US: Output dimensions
-      placeholder:
-        zh_Hans: 输入您的输出维度
-        en_US: Enter output dimensions
-      required: false
-      type: text-input
-    - variable: late_chunking
-      label:
-        zh_Hans: 后期分块
-        en_US: Late chunking
-      placeholder:
-        zh_Hans: 应用后期分块技术来利用模型的长上下文功能来生成上下文块向量化。
-        en_US: Apply the late chunking technique to leverage the model's long-context capabilities for generating contextual chunk embeddings.
-      required: false
-      type: switch
diff --git a/api/core/model_runtime/model_providers/jina/text_embedding/text_embedding.py b/api/core/model_runtime/model_providers/jina/text_embedding/text_embedding.py
index c7b729c14eaac..b39712951256c 100644
--- a/api/core/model_runtime/model_providers/jina/text_embedding/text_embedding.py
+++ b/api/core/model_runtime/model_providers/jina/text_embedding/text_embedding.py
@@ -28,7 +28,7 @@ class JinaTextEmbeddingModel(TextEmbeddingModel):
 
     api_base: str = "https://api.jina.ai/v1"
 
-    def _to_payload(self, model: str, texts: list[str], credentials: dict) -> dict:
+    def _to_payload(self, model: str, texts: list[str], credentials: dict, input_type: EmbeddingInputType) -> dict:
         """
         Parse model credentials
 
@@ -45,18 +45,10 @@ def transform_jina_input_text(model, text):
 
         data = {"model": model, "input": [transform_jina_input_text(model, text) for text in texts]}
 
-        task = credentials.get("task")
-        dimensions = credentials.get("dimensions")
-        late_chunking = credentials.get("late_chunking")
-
-        if task is not None:
-            data["task"] = task
-
-        if dimensions is not None:
-            data["dimensions"] = int(dimensions)
-
-        if late_chunking is not None:
-            data["late_chunking"] = late_chunking
+        # model specific parameters
+        if model == "jina-embeddings-v3":
+            # set `task` type according to input type for the best performance
+            data["task"] = "retrieval.query" if input_type == EmbeddingInputType.QUERY else "retrieval.passage"
 
         return data
 
@@ -88,7 +80,7 @@ def _invoke(
         url = base_url + "/embeddings"
         headers = {"Authorization": "Bearer " + api_key, "Content-Type": "application/json"}
 
-        data = self._to_payload(model=model, texts=texts, credentials=credentials)
+        data = self._to_payload(model=model, texts=texts, credentials=credentials, input_type=input_type)
 
         try:
             response = post(url, headers=headers, data=dumps(data))

From a0b0809b1c1598db7e8762d78c6d9f6d34070168 Mon Sep 17 00:00:00 2001
From: Shenghang Tsai <jackalcooper@gmail.com>
Date: Thu, 26 Sep 2024 11:29:53 +0800
Subject: [PATCH 073/235] Add more models for SiliconFlow (#8779)

---
 .../siliconflow/llm/_position.yaml            | 39 ++++++++++++-------
 1 file changed, 26 insertions(+), 13 deletions(-)

diff --git a/api/core/model_runtime/model_providers/siliconflow/llm/_position.yaml b/api/core/model_runtime/model_providers/siliconflow/llm/_position.yaml
index 43db4aed11916..a3e5d0981f743 100644
--- a/api/core/model_runtime/model_providers/siliconflow/llm/_position.yaml
+++ b/api/core/model_runtime/model_providers/siliconflow/llm/_position.yaml
@@ -1,25 +1,38 @@
-- Qwen/Qwen2.5-7B-Instruct
-- Qwen/Qwen2.5-14B-Instruct
-- Qwen/Qwen2.5-32B-Instruct
 - Qwen/Qwen2.5-72B-Instruct
+- Qwen/Qwen2.5-Math-72B-Instruct
+- Qwen/Qwen2.5-32B-Instruct
+- Qwen/Qwen2.5-14B-Instruct
+- Qwen/Qwen2.5-7B-Instruct
+- Qwen/Qwen2.5-Coder-7B-Instruct
+- deepseek-ai/DeepSeek-V2.5
 - Qwen/Qwen2-72B-Instruct
 - Qwen/Qwen2-57B-A14B-Instruct
 - Qwen/Qwen2-7B-Instruct
 - Qwen/Qwen2-1.5B-Instruct
-- 01-ai/Yi-1.5-34B-Chat
-- 01-ai/Yi-1.5-9B-Chat-16K
-- 01-ai/Yi-1.5-6B-Chat
-- THUDM/glm-4-9b-chat
-- deepseek-ai/DeepSeek-V2.5
 - deepseek-ai/DeepSeek-V2-Chat
 - deepseek-ai/DeepSeek-Coder-V2-Instruct
+- THUDM/glm-4-9b-chat
+- THUDM/chatglm3-6b
+- 01-ai/Yi-1.5-34B-Chat-16K
+- 01-ai/Yi-1.5-9B-Chat-16K
+- 01-ai/Yi-1.5-6B-Chat
+- internlm/internlm2_5-20b-chat
 - internlm/internlm2_5-7b-chat
-- google/gemma-2-27b-it
-- google/gemma-2-9b-it
-- meta-llama/Meta-Llama-3-70B-Instruct
-- meta-llama/Meta-Llama-3-8B-Instruct
 - meta-llama/Meta-Llama-3.1-405B-Instruct
 - meta-llama/Meta-Llama-3.1-70B-Instruct
 - meta-llama/Meta-Llama-3.1-8B-Instruct
-- mistralai/Mixtral-8x7B-Instruct-v0.1
+- meta-llama/Meta-Llama-3-70B-Instruct
+- meta-llama/Meta-Llama-3-8B-Instruct
+- google/gemma-2-27b-it
+- google/gemma-2-9b-it
 - mistralai/Mistral-7B-Instruct-v0.2
+- Pro/Qwen/Qwen2-7B-Instruct
+- Pro/Qwen/Qwen2-1.5B-Instruct
+- Pro/THUDM/glm-4-9b-chat
+- Pro/THUDM/chatglm3-6b
+- Pro/01-ai/Yi-1.5-9B-Chat-16K
+- Pro/01-ai/Yi-1.5-6B-Chat
+- Pro/internlm/internlm2_5-7b-chat
+- Pro/meta-llama/Meta-Llama-3.1-8B-Instruct
+- Pro/meta-llama/Meta-Llama-3-8B-Instruct
+- Pro/google/gemma-2-9b-it

From d1173a69f8ecf6680b78d194f8c266cada2042ba Mon Sep 17 00:00:00 2001
From: =?UTF-8?q?=E9=9D=9E=E6=B3=95=E6=93=8D=E4=BD=9C?= <hjlarry@163.com>
Date: Thu, 26 Sep 2024 13:48:06 +0800
Subject: [PATCH 074/235] fix: the Image-1X tool (#8787)

---
 .../tools/provider/builtin/stepfun/stepfun.py |  2 +-
 .../provider/builtin/stepfun/stepfun.yaml     | 17 ++---------
 .../provider/builtin/stepfun/tools/image.py   | 13 ++-------
 .../provider/builtin/stepfun/tools/image.yaml | 29 -------------------
 4 files changed, 5 insertions(+), 56 deletions(-)

diff --git a/api/core/tools/provider/builtin/stepfun/stepfun.py b/api/core/tools/provider/builtin/stepfun/stepfun.py
index b24f730c95a7c..239db85b1118b 100644
--- a/api/core/tools/provider/builtin/stepfun/stepfun.py
+++ b/api/core/tools/provider/builtin/stepfun/stepfun.py
@@ -16,7 +16,7 @@ def _validate_credentials(self, credentials: dict[str, Any]) -> None:
                 user_id="",
                 tool_parameters={
                     "prompt": "cute girl, blue eyes, white hair, anime style",
-                    "size": "1024x1024",
+                    "size": "256x256",
                     "n": 1,
                 },
             )
diff --git a/api/core/tools/provider/builtin/stepfun/stepfun.yaml b/api/core/tools/provider/builtin/stepfun/stepfun.yaml
index 1f841ec369b5c..e8139a4d7d6cf 100644
--- a/api/core/tools/provider/builtin/stepfun/stepfun.yaml
+++ b/api/core/tools/provider/builtin/stepfun/stepfun.yaml
@@ -4,11 +4,9 @@ identity:
   label:
     en_US: Image-1X
     zh_Hans: 阶跃星辰绘画
-    pt_BR: Image-1X
   description:
     en_US: Image-1X
     zh_Hans: 阶跃星辰绘画
-    pt_BR: Image-1X
   icon: icon.png
   tags:
     - image
@@ -20,27 +18,16 @@ credentials_for_provider:
     label:
       en_US: Stepfun API key
       zh_Hans: 阶跃星辰API key
-      pt_BR: Stepfun API key
-    help:
-      en_US: Please input your stepfun API key
-      zh_Hans: 请输入你的阶跃星辰 API key
-      pt_BR: Please input your stepfun API key
     placeholder:
-      en_US: Please input your stepfun API key
+      en_US: Please input your Stepfun API key
       zh_Hans: 请输入你的阶跃星辰 API key
-      pt_BR: Please input your stepfun API key
+    url: https://platform.stepfun.com/interface-key
   stepfun_base_url:
     type: text-input
     required: false
     label:
       en_US: Stepfun base URL
       zh_Hans: 阶跃星辰 base URL
-      pt_BR: Stepfun base URL
-    help:
-      en_US: Please input your Stepfun base URL
-      zh_Hans: 请输入你的阶跃星辰 base URL
-      pt_BR: Please input your Stepfun base URL
     placeholder:
       en_US: Please input your Stepfun base URL
       zh_Hans: 请输入你的阶跃星辰 base URL
-      pt_BR: Please input your Stepfun base URL
diff --git a/api/core/tools/provider/builtin/stepfun/tools/image.py b/api/core/tools/provider/builtin/stepfun/tools/image.py
index 0b92b122bf59b..eb55dae5183c4 100644
--- a/api/core/tools/provider/builtin/stepfun/tools/image.py
+++ b/api/core/tools/provider/builtin/stepfun/tools/image.py
@@ -1,4 +1,3 @@
-import random
 from typing import Any, Union
 
 from openai import OpenAI
@@ -19,7 +18,7 @@ def _invoke(
         """
         invoke tools
         """
-        base_url = self.runtime.credentials.get("stepfun_base_url", "https://api.stepfun.com")
+        base_url = self.runtime.credentials.get("stepfun_base_url") or "https://api.stepfun.com"
         base_url = str(URL(base_url) / "v1")
 
         client = OpenAI(
@@ -28,9 +27,7 @@ def _invoke(
         )
 
         extra_body = {}
-        model = tool_parameters.get("model", "step-1x-medium")
-        if not model:
-            return self.create_text_message("Please input model name")
+        model = "step-1x-medium"
         # prompt
         prompt = tool_parameters.get("prompt", "")
         if not prompt:
@@ -67,9 +64,3 @@ def _invoke(
                 )
             )
         return result
-
-    @staticmethod
-    def _generate_random_id(length=8):
-        characters = "ABCDEFGHIJKLMNOPQRSTUVWXYZabcdefghijklmnopqrstuvwxyz0123456789"
-        random_id = "".join(random.choices(characters, k=length))
-        return random_id
diff --git a/api/core/tools/provider/builtin/stepfun/tools/image.yaml b/api/core/tools/provider/builtin/stepfun/tools/image.yaml
index dcc5bd2db2f5b..8d7c9b6586b6c 100644
--- a/api/core/tools/provider/builtin/stepfun/tools/image.yaml
+++ b/api/core/tools/provider/builtin/stepfun/tools/image.yaml
@@ -29,35 +29,6 @@ parameters:
       pt_BR: Image prompt, you can check the official documentation of step-1x
     llm_description: Image prompt of step-1x you should describe the image you want to generate as a list of words as possible as detailed
     form: llm
-  - name: model
-    type: select
-    required: false
-    human_description:
-      en_US: used for selecting the model name
-      zh_Hans: 用于选择模型的名字
-      pt_BR: used for selecting the model name
-    label:
-      en_US: Model Name
-      zh_Hans: 模型名字
-      pt_BR: Model Name
-    form: form
-    options:
-      - value: step-1x-turbo
-        label:
-          en_US: turbo
-          zh_Hans: turbo
-          pt_BR: turbo
-      - value: step-1x-medium
-        label:
-          en_US: medium
-          zh_Hans: medium
-          pt_BR: medium
-      - value: step-1x-large
-        label:
-          en_US: large
-          zh_Hans: large
-          pt_BR: large
-    default: step-1x-medium
   - name: size
     type: select
     required: false

From 62406991df10d6eaa721df1b56e8df957802759b Mon Sep 17 00:00:00 2001
From: =?UTF-8?q?=E9=9D=9E=E6=B3=95=E6=93=8D=E4=BD=9C?= <hjlarry@163.com>
Date: Thu, 26 Sep 2024 16:28:20 +0800
Subject: [PATCH 075/235] fix: start node input config modal raise  'variable
 name is required' (#8793)

---
 .../configuration/config-var/config-modal/index.tsx    | 10 ++++++++--
 1 file changed, 8 insertions(+), 2 deletions(-)

diff --git a/web/app/components/app/configuration/config-var/config-modal/index.tsx b/web/app/components/app/configuration/config-var/config-modal/index.tsx
index 606280653eeec..ac1d86c1a2c2d 100644
--- a/web/app/components/app/configuration/config-var/config-modal/index.tsx
+++ b/web/app/components/app/configuration/config-var/config-modal/index.tsx
@@ -1,6 +1,6 @@
 'use client'
 import type { FC } from 'react'
-import React, { useCallback, useState } from 'react'
+import React, { useCallback, useEffect, useRef, useState } from 'react'
 import { useTranslation } from 'react-i18next'
 import { useContext } from 'use-context-selector'
 import ModalFoot from '../modal-foot'
@@ -40,6 +40,12 @@ const ConfigModal: FC<IConfigModalProps> = ({
   const { t } = useTranslation()
   const [tempPayload, setTempPayload] = useState<InputVar>(payload || getNewVarInWorkflow('') as any)
   const { type, label, variable, options, max_length } = tempPayload
+  const modalRef = useRef<HTMLDivElement>(null)
+  useEffect(() => {
+    // To fix the first input element auto focus, then directly close modal will raise error
+    if (isShow)
+      modalRef.current?.focus()
+  }, [isShow])
 
   const isStringInput = type === InputVarType.textInput || type === InputVarType.paragraph
   const checkVariableName = useCallback((value: string) => {
@@ -135,7 +141,7 @@ const ConfigModal: FC<IConfigModalProps> = ({
       isShow={isShow}
       onClose={onClose}
     >
-      <div className='mb-8'>
+      <div className='mb-8' ref={modalRef} tabIndex={-1}>
         <div className='space-y-2'>
 
           <Field title={t('appDebug.variableConfig.fieldType')}>

From 128a66f7fe8383cb30dec5e5e3896d75a225c77e Mon Sep 17 00:00:00 2001
From: cx <88480957+free-cx@users.noreply.github.com>
Date: Thu, 26 Sep 2024 16:34:40 +0800
Subject: [PATCH 076/235] =?UTF-8?q?fix:=20Ollama=20modelfeature=20set=20vi?=
 =?UTF-8?q?sion,=20and=20an=20exception=20occurred=20at=20the=E2=80=A6=20(?=
 =?UTF-8?q?#8783)?=
MIME-Version: 1.0
Content-Type: text/plain; charset=UTF-8
Content-Transfer-Encoding: 8bit

---
 .../model_providers/ollama/llm/llm.py         | 19 +++++++++++++------
 1 file changed, 13 insertions(+), 6 deletions(-)

diff --git a/api/core/model_runtime/model_providers/ollama/llm/llm.py b/api/core/model_runtime/model_providers/ollama/llm/llm.py
index ff732e69259e2..a7ea53e0e99c5 100644
--- a/api/core/model_runtime/model_providers/ollama/llm/llm.py
+++ b/api/core/model_runtime/model_providers/ollama/llm/llm.py
@@ -364,14 +364,21 @@ def create_final_llm_result_chunk(
 
             if chunk_json["done"]:
                 # calculate num tokens
-                if "prompt_eval_count" in chunk_json and "eval_count" in chunk_json:
-                    # transform usage
+                if "prompt_eval_count" in chunk_json:
                     prompt_tokens = chunk_json["prompt_eval_count"]
-                    completion_tokens = chunk_json["eval_count"]
                 else:
-                    # calculate num tokens
-                    prompt_tokens = self._get_num_tokens_by_gpt2(prompt_messages[0].content)
-                    completion_tokens = self._get_num_tokens_by_gpt2(full_text)
+                    prompt_message_content = prompt_messages[0].content
+                    if isinstance(prompt_message_content, str):
+                        prompt_tokens = self._get_num_tokens_by_gpt2(prompt_message_content)
+                    else:
+                        content_text = ""
+                        for message_content in prompt_message_content:
+                            if message_content.type == PromptMessageContentType.TEXT:
+                                message_content = cast(TextPromptMessageContent, message_content)
+                                content_text += message_content.data
+                        prompt_tokens = self._get_num_tokens_by_gpt2(content_text)
+
+                completion_tokens = chunk_json.get("eval_count", self._get_num_tokens_by_gpt2(full_text))
 
                 # transform usage
                 usage = self._calc_response_usage(model, credentials, prompt_tokens, completion_tokens)

From 008e0efeb00faecd44877245ec48f08fda6a048f Mon Sep 17 00:00:00 2001
From: zhuhao <37029601+hwzhuhao@users.noreply.github.com>
Date: Thu, 26 Sep 2024 16:36:21 +0800
Subject: [PATCH 077/235] refactor: update delete method as an abstract method
 (#8794)

---
 api/core/rag/datasource/vdb/vector_base.py | 1 +
 1 file changed, 1 insertion(+)

diff --git a/api/core/rag/datasource/vdb/vector_base.py b/api/core/rag/datasource/vdb/vector_base.py
index 1a0dc7f48b803..22e191340d3a4 100644
--- a/api/core/rag/datasource/vdb/vector_base.py
+++ b/api/core/rag/datasource/vdb/vector_base.py
@@ -45,6 +45,7 @@ def search_by_vector(self, query_vector: list[float], **kwargs: Any) -> list[Doc
     def search_by_full_text(self, query: str, **kwargs: Any) -> list[Document]:
         raise NotImplementedError
 
+    @abstractmethod
     def delete(self) -> None:
         raise NotImplementedError
 

From 6df14e50b2c71e2fc9b0601943d37a78f47e69c2 Mon Sep 17 00:00:00 2001
From: =?UTF-8?q?=E9=9D=9E=E6=B3=95=E6=93=8D=E4=BD=9C?= <hjlarry@163.com>
Date: Thu, 26 Sep 2024 17:50:36 +0800
Subject: [PATCH 078/235] fix: workflow as tool always outdated (#8798)

---
 web/app/components/tools/workflow-tool/configure-button.tsx | 2 +-
 1 file changed, 1 insertion(+), 1 deletion(-)

diff --git a/web/app/components/tools/workflow-tool/configure-button.tsx b/web/app/components/tools/workflow-tool/configure-button.tsx
index d2c5142f53c10..6521410daea01 100644
--- a/web/app/components/tools/workflow-tool/configure-button.tsx
+++ b/web/app/components/tools/workflow-tool/configure-button.tsx
@@ -65,7 +65,7 @@ const WorkflowToolConfigureButton = ({
         else {
           if (item.type === 'paragraph' && param.type !== 'string')
             return true
-          if (param.type !== item.type && !(param.type === 'string' && item.type === 'paragraph'))
+          if (item.type === 'text-input' && param.type !== 'string')
             return true
         }
       }

From 3d2cb25a6704ecd53b132deb4d7ea77b34a43d0d Mon Sep 17 00:00:00 2001
From: Joel <iamjoel007@gmail.com>
Date: Thu, 26 Sep 2024 17:53:11 +0800
Subject: [PATCH 079/235] fix: change wrong company name (#8801)

---
 web/app/activate/page.tsx        | 2 +-
 web/app/forgot-password/page.tsx | 2 +-
 web/app/install/page.tsx         | 2 +-
 3 files changed, 3 insertions(+), 3 deletions(-)

diff --git a/web/app/activate/page.tsx b/web/app/activate/page.tsx
index 90874f50cefe0..0f1854433552d 100644
--- a/web/app/activate/page.tsx
+++ b/web/app/activate/page.tsx
@@ -22,7 +22,7 @@ const Activate = () => {
         <Header />
         <ActivateForm />
         <div className='px-8 py-6 text-sm font-normal text-gray-500'>
-          © {new Date().getFullYear()} Dify, Inc. All rights reserved.
+          © {new Date().getFullYear()} LangGenius, Inc. All rights reserved.
         </div>
       </div>
     </div>
diff --git a/web/app/forgot-password/page.tsx b/web/app/forgot-password/page.tsx
index fa44d1a20c689..bb46011c06302 100644
--- a/web/app/forgot-password/page.tsx
+++ b/web/app/forgot-password/page.tsx
@@ -28,7 +28,7 @@ const ForgotPassword = () => {
         <Header />
         {token ? <ChangePasswordForm /> : <ForgotPasswordForm />}
         <div className='px-8 py-6 text-sm font-normal text-gray-500'>
-          © {new Date().getFullYear()} Dify, Inc. All rights reserved.
+          © {new Date().getFullYear()} LangGenius, Inc. All rights reserved.
         </div>
       </div>
     </div>
diff --git a/web/app/install/page.tsx b/web/app/install/page.tsx
index 9fa38dd15e2a8..395fae34ec169 100644
--- a/web/app/install/page.tsx
+++ b/web/app/install/page.tsx
@@ -22,7 +22,7 @@ const Install = () => {
         <Header />
         <InstallForm />
         <div className='px-8 py-6 text-sm font-normal text-gray-500'>
-          © {new Date().getFullYear()} Dify, Inc. All rights reserved.
+          © {new Date().getFullYear()} LangGenius, Inc. All rights reserved.
         </div>
       </div>
     </div>

From 03edfbe6f530edfc4fb7781adda3852d6ab643a2 Mon Sep 17 00:00:00 2001
From: AAEE86 <33052466+AAEE86@users.noreply.github.com>
Date: Thu, 26 Sep 2024 19:04:25 +0800
Subject: [PATCH 080/235] feat: add qwen to add custom model parameters (#8759)

---
 .../model_providers/tongyi/llm/llm.py         | 102 +++++++++---------
 .../model_providers/tongyi/tongyi.yaml        |  43 +++++++-
 2 files changed, 92 insertions(+), 53 deletions(-)

diff --git a/api/core/model_runtime/model_providers/tongyi/llm/llm.py b/api/core/model_runtime/model_providers/tongyi/llm/llm.py
index f90c7f075fe86..3e3585b30ae33 100644
--- a/api/core/model_runtime/model_providers/tongyi/llm/llm.py
+++ b/api/core/model_runtime/model_providers/tongyi/llm/llm.py
@@ -18,7 +18,7 @@
     UnsupportedModel,
 )
 
-from core.model_runtime.entities.llm_entities import LLMResult, LLMResultChunk, LLMResultChunkDelta
+from core.model_runtime.entities.llm_entities import LLMMode, LLMResult, LLMResultChunk, LLMResultChunkDelta
 from core.model_runtime.entities.message_entities import (
     AssistantPromptMessage,
     ImagePromptMessageContent,
@@ -35,6 +35,7 @@
     FetchFrom,
     I18nObject,
     ModelFeature,
+    ModelPropertyKey,
     ModelType,
     ParameterRule,
     ParameterType,
@@ -97,6 +98,11 @@ def get_num_tokens(
         :param tools: tools for tool calling
         :return:
         """
+        # Check if the model was added via get_customizable_model_schema
+        if self.get_customizable_model_schema(model, credentials) is not None:
+            # For custom models, tokens are not calculated.
+            return 0
+
         if model in {"qwen-turbo-chat", "qwen-plus-chat"}:
             model = model.replace("-chat", "")
         if model == "farui-plus":
@@ -537,55 +543,51 @@ def get_customizable_model_schema(self, model: str, credentials: dict) -> AIMode
         :param credentials: model credentials
         :return: AIModelEntity or None
         """
-        rules = [
-            ParameterRule(
-                name="temperature",
-                type=ParameterType.FLOAT,
-                use_template="temperature",
-                label=I18nObject(zh_Hans="温度", en_US="Temperature"),
-            ),
-            ParameterRule(
-                name="top_p",
-                type=ParameterType.FLOAT,
-                use_template="top_p",
-                label=I18nObject(zh_Hans="Top P", en_US="Top P"),
-            ),
-            ParameterRule(
-                name="top_k",
-                type=ParameterType.INT,
-                min=0,
-                max=99,
-                label=I18nObject(zh_Hans="top_k", en_US="top_k"),
-            ),
-            ParameterRule(
-                name="max_tokens",
-                type=ParameterType.INT,
-                min=1,
-                max=128000,
-                default=1024,
-                label=I18nObject(zh_Hans="最大生成长度", en_US="Max Tokens"),
-            ),
-            ParameterRule(
-                name="seed",
-                type=ParameterType.INT,
-                default=1234,
-                label=I18nObject(zh_Hans="随机种子", en_US="Random Seed"),
-            ),
-            ParameterRule(
-                name="repetition_penalty",
-                type=ParameterType.FLOAT,
-                default=1.1,
-                label=I18nObject(zh_Hans="重复惩罚", en_US="Repetition Penalty"),
-            ),
-        ]
-
-        entity = AIModelEntity(
+        return AIModelEntity(
             model=model,
-            label=I18nObject(en_US=model),
-            fetch_from=FetchFrom.CUSTOMIZABLE_MODEL,
+            label=I18nObject(en_US=model, zh_Hans=model),
             model_type=ModelType.LLM,
-            model_properties={},
-            parameter_rules=rules,
+            features=[ModelFeature.TOOL_CALL, ModelFeature.MULTI_TOOL_CALL, ModelFeature.STREAM_TOOL_CALL]
+            if credentials.get("function_calling_type") == "tool_call"
+            else [],
+            fetch_from=FetchFrom.CUSTOMIZABLE_MODEL,
+            model_properties={
+                ModelPropertyKey.CONTEXT_SIZE: int(credentials.get("context_size", 8000)),
+                ModelPropertyKey.MODE: LLMMode.CHAT.value,
+            },
+            parameter_rules=[
+                ParameterRule(
+                    name="temperature",
+                    use_template="temperature",
+                    label=I18nObject(en_US="Temperature", zh_Hans="温度"),
+                    type=ParameterType.FLOAT,
+                ),
+                ParameterRule(
+                    name="max_tokens",
+                    use_template="max_tokens",
+                    default=512,
+                    min=1,
+                    max=int(credentials.get("max_tokens", 1024)),
+                    label=I18nObject(en_US="Max Tokens", zh_Hans="最大标记"),
+                    type=ParameterType.INT,
+                ),
+                ParameterRule(
+                    name="top_p",
+                    use_template="top_p",
+                    label=I18nObject(en_US="Top P", zh_Hans="Top P"),
+                    type=ParameterType.FLOAT,
+                ),
+                ParameterRule(
+                    name="top_k",
+                    use_template="top_k",
+                    label=I18nObject(en_US="Top K", zh_Hans="Top K"),
+                    type=ParameterType.FLOAT,
+                ),
+                ParameterRule(
+                    name="frequency_penalty",
+                    use_template="frequency_penalty",
+                    label=I18nObject(en_US="Frequency Penalty", zh_Hans="重复惩罚"),
+                    type=ParameterType.FLOAT,
+                ),
+            ],
         )
-
-        return entity
diff --git a/api/core/model_runtime/model_providers/tongyi/tongyi.yaml b/api/core/model_runtime/model_providers/tongyi/tongyi.yaml
index fabe6d90e6b20..1a09c20fd93f6 100644
--- a/api/core/model_runtime/model_providers/tongyi/tongyi.yaml
+++ b/api/core/model_runtime/model_providers/tongyi/tongyi.yaml
@@ -37,14 +37,51 @@ model_credential_schema:
       en_US: Model Name
       zh_Hans: 模型名称
     placeholder:
-      en_US: Enter full model name
-      zh_Hans: 输入模型全称
+      en_US: Enter your model name
+      zh_Hans: 输入模型名称
   credential_form_schemas:
     - variable: dashscope_api_key
-      required: true
       label:
         en_US: API Key
       type: secret-input
+      required: true
       placeholder:
         zh_Hans: 在此输入您的 API Key
         en_US: Enter your API Key
+    - variable: context_size
+      label:
+        zh_Hans: 模型上下文长度
+        en_US: Model context size
+      required: true
+      type: text-input
+      default: '4096'
+      placeholder:
+        zh_Hans: 在此输入您的模型上下文长度
+        en_US: Enter your Model context size
+    - variable: max_tokens
+      label:
+        zh_Hans: 最大 token 上限
+        en_US: Upper bound for max tokens
+      default: '4096'
+      type: text-input
+      show_on:
+        - variable: __model_type
+          value: llm
+    - variable: function_calling_type
+      label:
+        en_US: Function calling
+      type: select
+      required: false
+      default: no_call
+      options:
+        - value: no_call
+          label:
+            en_US: Not Support
+            zh_Hans: 不支持
+        - value: function_call
+          label:
+            en_US: Support
+            zh_Hans: 支持
+      show_on:
+        - variable: __model_type
+          value: llm

From 9a4b53a212054901f9c597ac64c0eb97e461e641 Mon Sep 17 00:00:00 2001
From: AAEE86 <33052466+AAEE86@users.noreply.github.com>
Date: Thu, 26 Sep 2024 19:08:59 +0800
Subject: [PATCH 081/235] feat: add stream for Gemini (#8678)

---
 .../google/llm/gemini-1.5-flash-8b-exp-0827.yaml         | 9 +++++++++
 .../google/llm/gemini-1.5-flash-exp-0827.yaml            | 9 +++++++++
 .../google/llm/gemini-1.5-flash-latest.yaml              | 9 +++++++++
 .../google/llm/gemini-1.5-pro-exp-0801.yaml              | 9 +++++++++
 .../google/llm/gemini-1.5-pro-exp-0827.yaml              | 9 +++++++++
 .../google/llm/gemini-1.5-pro-latest.yaml                | 9 +++++++++
 .../model_providers/google/llm/gemini-pro-vision.yaml    | 9 +++++++++
 .../model_providers/google/llm/gemini-pro.yaml           | 9 +++++++++
 8 files changed, 72 insertions(+)

diff --git a/api/core/model_runtime/model_providers/google/llm/gemini-1.5-flash-8b-exp-0827.yaml b/api/core/model_runtime/model_providers/google/llm/gemini-1.5-flash-8b-exp-0827.yaml
index bbc697e934e05..4e0209890a336 100644
--- a/api/core/model_runtime/model_providers/google/llm/gemini-1.5-flash-8b-exp-0827.yaml
+++ b/api/core/model_runtime/model_providers/google/llm/gemini-1.5-flash-8b-exp-0827.yaml
@@ -32,6 +32,15 @@ parameter_rules:
     max: 8192
   - name: response_format
     use_template: response_format
+  - name: stream
+    label:
+      zh_Hans: 流式输出
+      en_US: Stream
+    type: boolean
+    help:
+      zh_Hans: 流式输出允许模型在生成文本的过程中逐步返回结果，而不是一次性生成全部结果后再返回。
+      en_US: Streaming output allows the model to return results incrementally as it generates text, rather than generating all the results at once.
+    default: false
 pricing:
   input: '0.00'
   output: '0.00'
diff --git a/api/core/model_runtime/model_providers/google/llm/gemini-1.5-flash-exp-0827.yaml b/api/core/model_runtime/model_providers/google/llm/gemini-1.5-flash-exp-0827.yaml
index c5695e5dda8eb..faabc5e4d13a7 100644
--- a/api/core/model_runtime/model_providers/google/llm/gemini-1.5-flash-exp-0827.yaml
+++ b/api/core/model_runtime/model_providers/google/llm/gemini-1.5-flash-exp-0827.yaml
@@ -32,6 +32,15 @@ parameter_rules:
     max: 8192
   - name: response_format
     use_template: response_format
+  - name: stream
+    label:
+      zh_Hans: 流式输出
+      en_US: Stream
+    type: boolean
+    help:
+      zh_Hans: 流式输出允许模型在生成文本的过程中逐步返回结果，而不是一次性生成全部结果后再返回。
+      en_US: Streaming output allows the model to return results incrementally as it generates text, rather than generating all the results at once.
+    default: false
 pricing:
   input: '0.00'
   output: '0.00'
diff --git a/api/core/model_runtime/model_providers/google/llm/gemini-1.5-flash-latest.yaml b/api/core/model_runtime/model_providers/google/llm/gemini-1.5-flash-latest.yaml
index 24b1c5af8a3fd..6a0344699a78b 100644
--- a/api/core/model_runtime/model_providers/google/llm/gemini-1.5-flash-latest.yaml
+++ b/api/core/model_runtime/model_providers/google/llm/gemini-1.5-flash-latest.yaml
@@ -32,6 +32,15 @@ parameter_rules:
     max: 8192
   - name: response_format
     use_template: response_format
+  - name: stream
+    label:
+      zh_Hans: 流式输出
+      en_US: Stream
+    type: boolean
+    help:
+      zh_Hans: 流式输出允许模型在生成文本的过程中逐步返回结果，而不是一次性生成全部结果后再返回。
+      en_US: Streaming output allows the model to return results incrementally as it generates text, rather than generating all the results at once.
+    default: false
 pricing:
   input: '0.00'
   output: '0.00'
diff --git a/api/core/model_runtime/model_providers/google/llm/gemini-1.5-pro-exp-0801.yaml b/api/core/model_runtime/model_providers/google/llm/gemini-1.5-pro-exp-0801.yaml
index 0a918e0d7b1ac..97c68f7a18d91 100644
--- a/api/core/model_runtime/model_providers/google/llm/gemini-1.5-pro-exp-0801.yaml
+++ b/api/core/model_runtime/model_providers/google/llm/gemini-1.5-pro-exp-0801.yaml
@@ -32,6 +32,15 @@ parameter_rules:
     max: 8192
   - name: response_format
     use_template: response_format
+  - name: stream
+    label:
+      zh_Hans: 流式输出
+      en_US: Stream
+    type: boolean
+    help:
+      zh_Hans: 流式输出允许模型在生成文本的过程中逐步返回结果，而不是一次性生成全部结果后再返回。
+      en_US: Streaming output allows the model to return results incrementally as it generates text, rather than generating all the results at once.
+    default: false
 pricing:
   input: '0.00'
   output: '0.00'
diff --git a/api/core/model_runtime/model_providers/google/llm/gemini-1.5-pro-exp-0827.yaml b/api/core/model_runtime/model_providers/google/llm/gemini-1.5-pro-exp-0827.yaml
index 7452ce46e7dcb..860e4816a163c 100644
--- a/api/core/model_runtime/model_providers/google/llm/gemini-1.5-pro-exp-0827.yaml
+++ b/api/core/model_runtime/model_providers/google/llm/gemini-1.5-pro-exp-0827.yaml
@@ -32,6 +32,15 @@ parameter_rules:
     max: 8192
   - name: response_format
     use_template: response_format
+  - name: stream
+    label:
+      zh_Hans: 流式输出
+      en_US: Stream
+    type: boolean
+    help:
+      zh_Hans: 流式输出允许模型在生成文本的过程中逐步返回结果，而不是一次性生成全部结果后再返回。
+      en_US: Streaming output allows the model to return results incrementally as it generates text, rather than generating all the results at once.
+    default: false
 pricing:
   input: '0.00'
   output: '0.00'
diff --git a/api/core/model_runtime/model_providers/google/llm/gemini-1.5-pro-latest.yaml b/api/core/model_runtime/model_providers/google/llm/gemini-1.5-pro-latest.yaml
index b3e1ecf3aff37..92cd6b310d07e 100644
--- a/api/core/model_runtime/model_providers/google/llm/gemini-1.5-pro-latest.yaml
+++ b/api/core/model_runtime/model_providers/google/llm/gemini-1.5-pro-latest.yaml
@@ -32,6 +32,15 @@ parameter_rules:
     max: 8192
   - name: response_format
     use_template: response_format
+  - name: stream
+    label:
+      zh_Hans: 流式输出
+      en_US: Stream
+    type: boolean
+    help:
+      zh_Hans: 流式输出允许模型在生成文本的过程中逐步返回结果，而不是一次性生成全部结果后再返回。
+      en_US: Streaming output allows the model to return results incrementally as it generates text, rather than generating all the results at once.
+    default: false
 pricing:
   input: '0.00'
   output: '0.00'
diff --git a/api/core/model_runtime/model_providers/google/llm/gemini-pro-vision.yaml b/api/core/model_runtime/model_providers/google/llm/gemini-pro-vision.yaml
index 075e484e46930..2d213d56adb9c 100644
--- a/api/core/model_runtime/model_providers/google/llm/gemini-pro-vision.yaml
+++ b/api/core/model_runtime/model_providers/google/llm/gemini-pro-vision.yaml
@@ -27,6 +27,15 @@ parameter_rules:
     default: 4096
     min: 1
     max: 4096
+  - name: stream
+    label:
+      zh_Hans: 流式输出
+      en_US: Stream
+    type: boolean
+    help:
+      zh_Hans: 流式输出允许模型在生成文本的过程中逐步返回结果，而不是一次性生成全部结果后再返回。
+      en_US: Streaming output allows the model to return results incrementally as it generates text, rather than generating all the results at once.
+    default: false
 pricing:
   input: '0.00'
   output: '0.00'
diff --git a/api/core/model_runtime/model_providers/google/llm/gemini-pro.yaml b/api/core/model_runtime/model_providers/google/llm/gemini-pro.yaml
index 4e9f59e7da94f..e2f487c1ee921 100644
--- a/api/core/model_runtime/model_providers/google/llm/gemini-pro.yaml
+++ b/api/core/model_runtime/model_providers/google/llm/gemini-pro.yaml
@@ -31,6 +31,15 @@ parameter_rules:
     max: 2048
   - name: response_format
     use_template: response_format
+  - name: stream
+    label:
+      zh_Hans: 流式输出
+      en_US: Stream
+    type: boolean
+    help:
+      zh_Hans: 流式输出允许模型在生成文本的过程中逐步返回结果，而不是一次性生成全部结果后再返回。
+      en_US: Streaming output allows the model to return results incrementally as it generates text, rather than generating all the results at once.
+    default: false
 pricing:
   input: '0.00'
   output: '0.00'

From 3dfbc348e3f53b26a3d385e83dea10bf0ef73736 Mon Sep 17 00:00:00 2001
From: Hash Brown <hi@xzd.me>
Date: Thu, 26 Sep 2024 19:41:59 +0800
Subject: [PATCH 082/235] feat: improved SVG output UX (#8765)

---
 .../base/chat/chat/answer/index.tsx           |  13 ++
 web/app/components/base/markdown.tsx          | 119 ++++++++++--------
 web/app/components/base/svg-gallery/index.tsx |  18 ++-
 .../workflow/panel/chat-record/index.tsx      |   6 +-
 4 files changed, 95 insertions(+), 61 deletions(-)

diff --git a/web/app/components/base/chat/chat/answer/index.tsx b/web/app/components/base/chat/chat/answer/index.tsx
index 705cd73ddf18d..8184967edc982 100644
--- a/web/app/components/base/chat/chat/answer/index.tsx
+++ b/web/app/components/base/chat/chat/answer/index.tsx
@@ -85,6 +85,19 @@ const Answer: FC<AnswerProps> = ({
       getContentWidth()
   }, [responding])
 
+  // Recalculate contentWidth when content changes (e.g., SVG preview/source toggle)
+  useEffect(() => {
+    if (!containerRef.current)
+      return
+    const resizeObserver = new ResizeObserver(() => {
+      getContentWidth()
+    })
+    resizeObserver.observe(containerRef.current)
+    return () => {
+      resizeObserver.disconnect()
+    }
+  }, [])
+
   return (
     <div className='flex mb-2 last:mb-0'>
       <div className='shrink-0 relative w-10 h-10'>
diff --git a/web/app/components/base/markdown.tsx b/web/app/components/base/markdown.tsx
index 443ee3410c4de..39a399cc9f73c 100644
--- a/web/app/components/base/markdown.tsx
+++ b/web/app/components/base/markdown.tsx
@@ -116,59 +116,80 @@ const CodeBlock: CodeComponent = memo(({ inline, className, children, ...props }
   const match = /language-(\w+)/.exec(className || '')
   const language = match?.[1]
   const languageShowName = getCorrectCapitalizationLanguageName(language || '')
-  let chartData = JSON.parse(String('{"title":{"text":"ECharts error - Wrong JSON format."}}').replace(/\n$/, ''))
-  if (language === 'echarts') {
-    try {
-      chartData = JSON.parse(String(children).replace(/\n$/, ''))
-    }
-    catch (error) {
+  const chartData = useMemo(() => {
+    if (language === 'echarts') {
+      try {
+        return JSON.parse(String(children).replace(/\n$/, ''))
+      }
+      catch (error) {}
     }
-  }
+    return JSON.parse('{"title":{"text":"ECharts error - Wrong JSON format."}}')
+  }, [language, children])
 
-  // Use `useMemo` to ensure that `SyntaxHighlighter` only re-renders when necessary
-  return useMemo(() => {
-    return (!inline && match)
-      ? (
-        <div>
-          <div
-            className='flex justify-between h-8 items-center p-1 pl-3 border-b'
-            style={{
-              borderColor: 'rgba(0, 0, 0, 0.05)',
-            }}
-          >
-            <div className='text-[13px] text-gray-500 font-normal'>{languageShowName}</div>
-            <div style={{ display: 'flex' }}>
-              {language === 'mermaid' && <SVGBtn isSVG={isSVG} setIsSVG={setIsSVG}/>}
-              <CopyBtn
-                className='mr-1'
-                value={String(children).replace(/\n$/, '')}
-                isPlain
-              />
-            </div>
-          </div>
-          {(language === 'mermaid' && isSVG)
-            ? (<Flowchart PrimitiveCode={String(children).replace(/\n$/, '')} />)
-            : (language === 'echarts'
-              ? (<div style={{ minHeight: '350px', minWidth: '700px' }}><ErrorBoundary><ReactEcharts option={chartData} /></ErrorBoundary></div>)
-              : (language === 'svg'
-                ? (<ErrorBoundary><SVGRenderer content={String(children).replace(/\n$/, '')} /></ErrorBoundary>)
-                : (<SyntaxHighlighter
-                  {...props}
-                  style={atelierHeathLight}
-                  customStyle={{
-                    paddingLeft: 12,
-                    backgroundColor: '#fff',
-                  }}
-                  language={match[1]}
-                  showLineNumbers
-                  PreTag="div"
-                >
-                  {String(children).replace(/\n$/, '')}
-                </SyntaxHighlighter>)))}
+  const renderCodeContent = useMemo(() => {
+    const content = String(children).replace(/\n$/, '')
+    if (language === 'mermaid' && isSVG) {
+      return <Flowchart PrimitiveCode={content} />
+    }
+    else if (language === 'echarts') {
+      return (
+        <div style={{ minHeight: '350px', minWidth: '700px' }}>
+          <ErrorBoundary>
+            <ReactEcharts option={chartData} />
+          </ErrorBoundary>
         </div>
       )
-      : (<code {...props} className={className}>{children}</code>)
-  }, [chartData, children, className, inline, isSVG, language, languageShowName, match, props])
+    }
+    else if (language === 'svg' && isSVG) {
+      return (
+        <ErrorBoundary>
+          <SVGRenderer content={content} />
+        </ErrorBoundary>
+      )
+    }
+    else {
+      return (
+        <SyntaxHighlighter
+          {...props}
+          style={atelierHeathLight}
+          customStyle={{
+            paddingLeft: 12,
+            backgroundColor: '#fff',
+          }}
+          language={match?.[1]}
+          showLineNumbers
+          PreTag="div"
+        >
+          {content}
+        </SyntaxHighlighter>
+      )
+    }
+  }, [language, match, props, children, chartData, isSVG])
+
+  if (inline || !match)
+    return <code {...props} className={className}>{children}</code>
+
+  return (
+    <div>
+      <div
+        className='flex justify-between h-8 items-center p-1 pl-3 border-b'
+        style={{
+          borderColor: 'rgba(0, 0, 0, 0.05)',
+        }}
+      >
+        <div className='text-[13px] text-gray-500 font-normal'>{languageShowName}</div>
+        <div style={{ display: 'flex' }}>
+          {(['mermaid', 'svg']).includes(language!) && <SVGBtn isSVG={isSVG} setIsSVG={setIsSVG}/>}
+          <CopyBtn
+            className='mr-1'
+            value={String(children).replace(/\n$/, '')}
+            isPlain
+          />
+        </div>
+      </div>
+      {renderCodeContent}
+    </div>
+  )
 })
 CodeBlock.displayName = 'CodeBlock'
 
diff --git a/web/app/components/base/svg-gallery/index.tsx b/web/app/components/base/svg-gallery/index.tsx
index 81e8e87655009..4368df00e9d38 100644
--- a/web/app/components/base/svg-gallery/index.tsx
+++ b/web/app/components/base/svg-gallery/index.tsx
@@ -29,7 +29,7 @@ export const SVGRenderer = ({ content }: { content: string }) => {
     if (svgRef.current) {
       try {
         svgRef.current.innerHTML = ''
-        const draw = SVG().addTo(svgRef.current).size('100%', '100%')
+        const draw = SVG().addTo(svgRef.current)
 
         const parser = new DOMParser()
         const svgDoc = parser.parseFromString(content, 'image/svg+xml')
@@ -40,13 +40,11 @@ export const SVGRenderer = ({ content }: { content: string }) => {
 
         const originalWidth = parseInt(svgElement.getAttribute('width') || '400', 10)
         const originalHeight = parseInt(svgElement.getAttribute('height') || '600', 10)
-        const scale = Math.min(windowSize.width / originalWidth, windowSize.height / originalHeight, 1)
-        const scaledWidth = originalWidth * scale
-        const scaledHeight = originalHeight * scale
-        draw.size(scaledWidth, scaledHeight)
+        draw.viewbox(0, 0, originalWidth, originalHeight)
+
+        svgRef.current.style.width = `${Math.min(originalWidth, 298)}px`
 
         const rootElement = draw.svg(content)
-        rootElement.scale(scale)
 
         rootElement.click(() => {
           setImagePreview(svgToDataURL(svgElement as Element))
@@ -54,7 +52,7 @@ export const SVGRenderer = ({ content }: { content: string }) => {
       }
       catch (error) {
         if (svgRef.current)
-          svgRef.current.innerHTML = 'Error rendering SVG. Wait for the image content to complete.'
+          svgRef.current.innerHTML = '<span style="padding: 1rem;">Error rendering SVG. Wait for the image content to complete.</span>'
       }
     }
   }, [content, windowSize])
@@ -62,14 +60,14 @@ export const SVGRenderer = ({ content }: { content: string }) => {
   return (
     <>
       <div ref={svgRef} style={{
-        width: '100%',
-        height: '100%',
-        minHeight: '300px',
         maxHeight: '80vh',
         display: 'flex',
         justifyContent: 'center',
         alignItems: 'center',
         cursor: 'pointer',
+        wordBreak: 'break-word',
+        whiteSpace: 'normal',
+        margin: '0 auto',
       }} />
       {imagePreview && (<ImagePreview url={imagePreview} title='Preview' onCancel={() => setImagePreview('')} />)}
     </>
diff --git a/web/app/components/workflow/panel/chat-record/index.tsx b/web/app/components/workflow/panel/chat-record/index.tsx
index 1bcfd6474d7c8..16d2c304a73c5 100644
--- a/web/app/components/workflow/panel/chat-record/index.tsx
+++ b/web/app/components/workflow/panel/chat-record/index.tsx
@@ -90,7 +90,7 @@ const ChatRecord = () => {
   return (
     <div
       className={`
-        flex flex-col w-[400px] rounded-l-2xl h-full border border-black/2 shadow-xl
+        flex flex-col w-[420px] rounded-l-2xl h-full border border-black/2 shadow-xl
       `}
       style={{
         background: 'linear-gradient(156deg, rgba(242, 244, 247, 0.80) 0%, rgba(242, 244, 247, 0.00) 99.43%), var(--white, #FFF)',
@@ -121,7 +121,7 @@ const ChatRecord = () => {
                 supportCitationHitInfo: true,
               } as any}
               chatList={chatList}
-              chatContainerClassName='px-4'
+              chatContainerClassName='px-3'
               chatContainerInnerClassName='pt-6 w-full max-w-full mx-auto'
               chatFooterClassName='px-4 rounded-b-2xl'
               chatFooterInnerClassName='pb-4 w-full max-w-full mx-auto'
@@ -129,6 +129,8 @@ const ChatRecord = () => {
               noChatInput
               allToolIcons={{}}
               showPromptLog
+              noSpacing
+              chatAnswerContainerInner='!pr-2'
             />
           </div>
         </>

From 063474f4083c6e73402c8dd8b9862ec8dabd14b8 Mon Sep 17 00:00:00 2001
From: ice yao <yao3690093@gmail.com>
Date: Thu, 26 Sep 2024 22:21:01 +0800
Subject: [PATCH 083/235] Add llama3.2 model in fireworks provider (#8809)

---
 .../llm/llama-v3p2-11b-vision-instruct.yaml   | 46 +++++++++++++++++++
 .../fireworks/llm/llama-v3p2-1b-instruct.yaml | 46 +++++++++++++++++++
 .../fireworks/llm/llama-v3p2-3b-instruct.yaml | 46 +++++++++++++++++++
 .../llm/llama-v3p2-90b-vision-instruct.yaml   | 46 +++++++++++++++++++
 4 files changed, 184 insertions(+)
 create mode 100644 api/core/model_runtime/model_providers/fireworks/llm/llama-v3p2-11b-vision-instruct.yaml
 create mode 100644 api/core/model_runtime/model_providers/fireworks/llm/llama-v3p2-1b-instruct.yaml
 create mode 100644 api/core/model_runtime/model_providers/fireworks/llm/llama-v3p2-3b-instruct.yaml
 create mode 100644 api/core/model_runtime/model_providers/fireworks/llm/llama-v3p2-90b-vision-instruct.yaml

diff --git a/api/core/model_runtime/model_providers/fireworks/llm/llama-v3p2-11b-vision-instruct.yaml b/api/core/model_runtime/model_providers/fireworks/llm/llama-v3p2-11b-vision-instruct.yaml
new file mode 100644
index 0000000000000..31415a24fa8b7
--- /dev/null
+++ b/api/core/model_runtime/model_providers/fireworks/llm/llama-v3p2-11b-vision-instruct.yaml
@@ -0,0 +1,46 @@
+model: accounts/fireworks/models/llama-v3p2-11b-vision-instruct
+label:
+  zh_Hans: Llama 3.2 11B Vision Instruct
+  en_US: Llama 3.2 11B Vision Instruct
+model_type: llm
+features:
+  - agent-thought
+  - tool-call
+model_properties:
+  mode: chat
+  context_size: 131072
+parameter_rules:
+  - name: temperature
+    use_template: temperature
+  - name: top_p
+    use_template: top_p
+  - name: top_k
+    label:
+      zh_Hans: 取样数量
+      en_US: Top k
+    type: int
+    help:
+      zh_Hans: 仅从每个后续标记的前 K 个选项中采样。
+      en_US: Only sample from the top K options for each subsequent token.
+  - name: max_tokens
+    use_template: max_tokens
+  - name: context_length_exceeded_behavior
+    default: None
+    label:
+      zh_Hans: 上下文长度超出行为
+      en_US: Context Length Exceeded Behavior
+    help:
+      zh_Hans: 上下文长度超出行为
+      en_US: Context Length Exceeded Behavior
+    type: string
+    options:
+      - None
+      - truncate
+      - error
+  - name: response_format
+    use_template: response_format
+pricing:
+  input: '0.2'
+  output: '0.2'
+  unit: '0.000001'
+  currency: USD
diff --git a/api/core/model_runtime/model_providers/fireworks/llm/llama-v3p2-1b-instruct.yaml b/api/core/model_runtime/model_providers/fireworks/llm/llama-v3p2-1b-instruct.yaml
new file mode 100644
index 0000000000000..c2fd77d2568d2
--- /dev/null
+++ b/api/core/model_runtime/model_providers/fireworks/llm/llama-v3p2-1b-instruct.yaml
@@ -0,0 +1,46 @@
+model: accounts/fireworks/models/llama-v3p2-1b-instruct
+label:
+  zh_Hans: Llama 3.2 1B Instruct
+  en_US: Llama 3.2 1B Instruct
+model_type: llm
+features:
+  - agent-thought
+  - tool-call
+model_properties:
+  mode: chat
+  context_size: 131072
+parameter_rules:
+  - name: temperature
+    use_template: temperature
+  - name: top_p
+    use_template: top_p
+  - name: top_k
+    label:
+      zh_Hans: 取样数量
+      en_US: Top k
+    type: int
+    help:
+      zh_Hans: 仅从每个后续标记的前 K 个选项中采样。
+      en_US: Only sample from the top K options for each subsequent token.
+  - name: max_tokens
+    use_template: max_tokens
+  - name: context_length_exceeded_behavior
+    default: None
+    label:
+      zh_Hans: 上下文长度超出行为
+      en_US: Context Length Exceeded Behavior
+    help:
+      zh_Hans: 上下文长度超出行为
+      en_US: Context Length Exceeded Behavior
+    type: string
+    options:
+      - None
+      - truncate
+      - error
+  - name: response_format
+    use_template: response_format
+pricing:
+  input: '0.1'
+  output: '0.1'
+  unit: '0.000001'
+  currency: USD
diff --git a/api/core/model_runtime/model_providers/fireworks/llm/llama-v3p2-3b-instruct.yaml b/api/core/model_runtime/model_providers/fireworks/llm/llama-v3p2-3b-instruct.yaml
new file mode 100644
index 0000000000000..4b3c459c7bf2f
--- /dev/null
+++ b/api/core/model_runtime/model_providers/fireworks/llm/llama-v3p2-3b-instruct.yaml
@@ -0,0 +1,46 @@
+model: accounts/fireworks/models/llama-v3p2-3b-instruct
+label:
+  zh_Hans: Llama 3.2 3B Instruct
+  en_US: Llama 3.2 3B Instruct
+model_type: llm
+features:
+  - agent-thought
+  - tool-call
+model_properties:
+  mode: chat
+  context_size: 131072
+parameter_rules:
+  - name: temperature
+    use_template: temperature
+  - name: top_p
+    use_template: top_p
+  - name: top_k
+    label:
+      zh_Hans: 取样数量
+      en_US: Top k
+    type: int
+    help:
+      zh_Hans: 仅从每个后续标记的前 K 个选项中采样。
+      en_US: Only sample from the top K options for each subsequent token.
+  - name: max_tokens
+    use_template: max_tokens
+  - name: context_length_exceeded_behavior
+    default: None
+    label:
+      zh_Hans: 上下文长度超出行为
+      en_US: Context Length Exceeded Behavior
+    help:
+      zh_Hans: 上下文长度超出行为
+      en_US: Context Length Exceeded Behavior
+    type: string
+    options:
+      - None
+      - truncate
+      - error
+  - name: response_format
+    use_template: response_format
+pricing:
+  input: '0.1'
+  output: '0.1'
+  unit: '0.000001'
+  currency: USD
diff --git a/api/core/model_runtime/model_providers/fireworks/llm/llama-v3p2-90b-vision-instruct.yaml b/api/core/model_runtime/model_providers/fireworks/llm/llama-v3p2-90b-vision-instruct.yaml
new file mode 100644
index 0000000000000..0aece7455d625
--- /dev/null
+++ b/api/core/model_runtime/model_providers/fireworks/llm/llama-v3p2-90b-vision-instruct.yaml
@@ -0,0 +1,46 @@
+model: accounts/fireworks/models/llama-v3p2-90b-vision-instruct
+label:
+  zh_Hans: Llama 3.2 90B Vision Instruct
+  en_US: Llama 3.2 90B Vision Instruct
+model_type: llm
+features:
+  - agent-thought
+  - tool-call
+model_properties:
+  mode: chat
+  context_size: 131072
+parameter_rules:
+  - name: temperature
+    use_template: temperature
+  - name: top_p
+    use_template: top_p
+  - name: top_k
+    label:
+      zh_Hans: 取样数量
+      en_US: Top k
+    type: int
+    help:
+      zh_Hans: 仅从每个后续标记的前 K 个选项中采样。
+      en_US: Only sample from the top K options for each subsequent token.
+  - name: max_tokens
+    use_template: max_tokens
+  - name: context_length_exceeded_behavior
+    default: None
+    label:
+      zh_Hans: 上下文长度超出行为
+      en_US: Context Length Exceeded Behavior
+    help:
+      zh_Hans: 上下文长度超出行为
+      en_US: Context Length Exceeded Behavior
+    type: string
+    options:
+      - None
+      - truncate
+      - error
+  - name: response_format
+    use_template: response_format
+pricing:
+  input: '0.9'
+  output: '0.9'
+  unit: '0.000001'
+  currency: USD

From ecc951609d4988424bfba58094d6e256ae86144e Mon Sep 17 00:00:00 2001
From: wenmeng zhou <wenmengzhou@gmail.com>
Date: Thu, 26 Sep 2024 22:32:33 +0800
Subject: [PATCH 084/235] add more detailed doc for models of  qwen series 
 (#8799)

Co-authored-by: crazywoola <427733928@qq.com>
---
 .../model_runtime/model_providers/tongyi/llm/farui-plus.yaml   | 1 +
 .../model_providers/tongyi/llm/qwen-coder-turbo-0919.yaml      | 1 +
 .../model_providers/tongyi/llm/qwen-coder-turbo-latest.yaml    | 1 +
 .../model_providers/tongyi/llm/qwen-coder-turbo.yaml           | 1 +
 .../model_runtime/model_providers/tongyi/llm/qwen-long.yaml    | 2 +-
 .../model_providers/tongyi/llm/qwen-math-plus-0816.yaml        | 1 +
 .../model_providers/tongyi/llm/qwen-math-plus-0919.yaml        | 1 +
 .../model_providers/tongyi/llm/qwen-math-plus-latest.yaml      | 1 +
 .../model_providers/tongyi/llm/qwen-math-plus.yaml             | 1 +
 .../model_providers/tongyi/llm/qwen-math-turbo-0919.yaml       | 1 +
 .../model_providers/tongyi/llm/qwen-math-turbo-latest.yaml     | 1 +
 .../model_providers/tongyi/llm/qwen-math-turbo.yaml            | 1 +
 .../model_providers/tongyi/llm/qwen-max-0107.yaml              | 2 ++
 .../model_providers/tongyi/llm/qwen-max-0403.yaml              | 2 ++
 .../model_providers/tongyi/llm/qwen-max-0428.yaml              | 2 ++
 .../model_providers/tongyi/llm/qwen-max-0919.yaml              | 2 ++
 .../model_providers/tongyi/llm/qwen-max-1201.yaml              | 2 ++
 .../model_providers/tongyi/llm/qwen-max-latest.yaml            | 2 ++
 .../model_providers/tongyi/llm/qwen-max-longcontext.yaml       | 2 ++
 .../model_runtime/model_providers/tongyi/llm/qwen-max.yaml     | 2 ++
 .../model_providers/tongyi/llm/qwen-plus-0206.yaml             | 2 ++
 .../model_providers/tongyi/llm/qwen-plus-0624.yaml             | 2 ++
 .../model_providers/tongyi/llm/qwen-plus-0723.yaml             | 2 ++
 .../model_providers/tongyi/llm/qwen-plus-0806.yaml             | 2 ++
 .../model_providers/tongyi/llm/qwen-plus-0919.yaml             | 2 ++
 .../model_providers/tongyi/llm/qwen-plus-chat.yaml             | 2 ++
 .../model_providers/tongyi/llm/qwen-plus-latest.yaml           | 2 ++
 .../model_runtime/model_providers/tongyi/llm/qwen-plus.yaml    | 2 ++
 .../model_providers/tongyi/llm/qwen-turbo-0206.yaml            | 3 +++
 .../model_providers/tongyi/llm/qwen-turbo-0624.yaml            | 2 ++
 .../model_providers/tongyi/llm/qwen-turbo-0919.yaml            | 2 ++
 .../model_providers/tongyi/llm/qwen-turbo-chat.yaml            | 2 ++
 .../model_providers/tongyi/llm/qwen-turbo-latest.yaml          | 2 ++
 .../model_runtime/model_providers/tongyi/llm/qwen-turbo.yaml   | 2 ++
 .../model_providers/tongyi/llm/qwen-vl-max-0201.yaml           | 1 +
 .../model_providers/tongyi/llm/qwen-vl-max-0809.yaml           | 1 +
 .../model_runtime/model_providers/tongyi/llm/qwen-vl-max.yaml  | 1 +
 .../model_providers/tongyi/llm/qwen-vl-plus-0201.yaml          | 1 +
 .../model_providers/tongyi/llm/qwen-vl-plus-0809.yaml          | 1 +
 .../model_runtime/model_providers/tongyi/llm/qwen-vl-plus.yaml | 1 +
 .../model_providers/tongyi/llm/qwen2-math-1.5b-instruct.yaml   | 1 +
 .../model_providers/tongyi/llm/qwen2-math-72b-instruct.yaml    | 1 +
 .../model_providers/tongyi/llm/qwen2-math-7b-instruct.yaml     | 1 +
 .../model_providers/tongyi/llm/qwen2.5-0.5b-instruct.yaml      | 1 +
 .../model_providers/tongyi/llm/qwen2.5-1.5b-instruct.yaml      | 1 +
 .../model_providers/tongyi/llm/qwen2.5-14b-instruct.yaml       | 1 +
 .../model_providers/tongyi/llm/qwen2.5-32b-instruct.yaml       | 1 +
 .../model_providers/tongyi/llm/qwen2.5-3b-instruct.yaml        | 1 +
 .../model_providers/tongyi/llm/qwen2.5-72b-instruct.yaml       | 1 +
 .../model_providers/tongyi/llm/qwen2.5-7b-instruct.yaml        | 1 +
 .../model_providers/tongyi/llm/qwen2.5-coder-7b-instruct.yaml  | 1 +
 .../tongyi/text_embedding/text-embedding-v1.yaml               | 1 +
 .../tongyi/text_embedding/text-embedding-v2.yaml               | 1 +
 .../tongyi/text_embedding/text-embedding-v3.yaml               | 1 +
 54 files changed, 77 insertions(+), 1 deletion(-)

diff --git a/api/core/model_runtime/model_providers/tongyi/llm/farui-plus.yaml b/api/core/model_runtime/model_providers/tongyi/llm/farui-plus.yaml
index d0ff44382789b..34a57d1fc0c9a 100644
--- a/api/core/model_runtime/model_providers/tongyi/llm/farui-plus.yaml
+++ b/api/core/model_runtime/model_providers/tongyi/llm/farui-plus.yaml
@@ -1,3 +1,4 @@
+# for more details, please refer to https://help.aliyun.com/zh/model-studio/getting-started/models
 model: farui-plus
 label:
   en_US: farui-plus
diff --git a/api/core/model_runtime/model_providers/tongyi/llm/qwen-coder-turbo-0919.yaml b/api/core/model_runtime/model_providers/tongyi/llm/qwen-coder-turbo-0919.yaml
index d9792e71ee454..64a3f331336bc 100644
--- a/api/core/model_runtime/model_providers/tongyi/llm/qwen-coder-turbo-0919.yaml
+++ b/api/core/model_runtime/model_providers/tongyi/llm/qwen-coder-turbo-0919.yaml
@@ -1,3 +1,4 @@
+# for more details, please refer to https://help.aliyun.com/zh/model-studio/getting-started/models
 model: qwen-coder-turbo-0919
 label:
   en_US: qwen-coder-turbo-0919
diff --git a/api/core/model_runtime/model_providers/tongyi/llm/qwen-coder-turbo-latest.yaml b/api/core/model_runtime/model_providers/tongyi/llm/qwen-coder-turbo-latest.yaml
index 0b03505c45084..a4c93f7047ff5 100644
--- a/api/core/model_runtime/model_providers/tongyi/llm/qwen-coder-turbo-latest.yaml
+++ b/api/core/model_runtime/model_providers/tongyi/llm/qwen-coder-turbo-latest.yaml
@@ -1,3 +1,4 @@
+# for more details, please refer to https://help.aliyun.com/zh/model-studio/getting-started/models
 model: qwen-coder-turbo-latest
 label:
   en_US: qwen-coder-turbo-latest
diff --git a/api/core/model_runtime/model_providers/tongyi/llm/qwen-coder-turbo.yaml b/api/core/model_runtime/model_providers/tongyi/llm/qwen-coder-turbo.yaml
index 2a6c040853005..ff68faed80810 100644
--- a/api/core/model_runtime/model_providers/tongyi/llm/qwen-coder-turbo.yaml
+++ b/api/core/model_runtime/model_providers/tongyi/llm/qwen-coder-turbo.yaml
@@ -1,3 +1,4 @@
+# for more details, please refer to https://help.aliyun.com/zh/model-studio/getting-started/models
 model: qwen-coder-turbo
 label:
   en_US: qwen-coder-turbo
diff --git a/api/core/model_runtime/model_providers/tongyi/llm/qwen-long.yaml b/api/core/model_runtime/model_providers/tongyi/llm/qwen-long.yaml
index bad7f4f4725c4..c3dbb3616fb96 100644
--- a/api/core/model_runtime/model_providers/tongyi/llm/qwen-long.yaml
+++ b/api/core/model_runtime/model_providers/tongyi/llm/qwen-long.yaml
@@ -1,4 +1,4 @@
-# model docs: https://help.aliyun.com/zh/model-studio/getting-started/models#27b2b3a15d5c6
+# for more details, please refer to https://help.aliyun.com/zh/model-studio/getting-started/models
 model: qwen-long
 label:
   en_US: qwen-long
diff --git a/api/core/model_runtime/model_providers/tongyi/llm/qwen-math-plus-0816.yaml b/api/core/model_runtime/model_providers/tongyi/llm/qwen-math-plus-0816.yaml
index c14aee1e1eb87..42fe1f68623bc 100644
--- a/api/core/model_runtime/model_providers/tongyi/llm/qwen-math-plus-0816.yaml
+++ b/api/core/model_runtime/model_providers/tongyi/llm/qwen-math-plus-0816.yaml
@@ -1,3 +1,4 @@
+# for more details, please refer to https://help.aliyun.com/zh/model-studio/getting-started/models
 model: qwen-math-plus-0816
 label:
   en_US: qwen-math-plus-0816
diff --git a/api/core/model_runtime/model_providers/tongyi/llm/qwen-math-plus-0919.yaml b/api/core/model_runtime/model_providers/tongyi/llm/qwen-math-plus-0919.yaml
index 9d74eeca3e878..9b6567b8cda4d 100644
--- a/api/core/model_runtime/model_providers/tongyi/llm/qwen-math-plus-0919.yaml
+++ b/api/core/model_runtime/model_providers/tongyi/llm/qwen-math-plus-0919.yaml
@@ -1,3 +1,4 @@
+# for more details, please refer to https://help.aliyun.com/zh/model-studio/getting-started/models
 model: qwen-math-plus-0919
 label:
   en_US: qwen-math-plus-0919
diff --git a/api/core/model_runtime/model_providers/tongyi/llm/qwen-math-plus-latest.yaml b/api/core/model_runtime/model_providers/tongyi/llm/qwen-math-plus-latest.yaml
index b8601a969a7fa..b2a2393b365fc 100644
--- a/api/core/model_runtime/model_providers/tongyi/llm/qwen-math-plus-latest.yaml
+++ b/api/core/model_runtime/model_providers/tongyi/llm/qwen-math-plus-latest.yaml
@@ -1,3 +1,4 @@
+# for more details, please refer to https://help.aliyun.com/zh/model-studio/getting-started/models
 model: qwen-math-plus-latest
 label:
   en_US: qwen-math-plus-latest
diff --git a/api/core/model_runtime/model_providers/tongyi/llm/qwen-math-plus.yaml b/api/core/model_runtime/model_providers/tongyi/llm/qwen-math-plus.yaml
index 4a948be597649..63f4b7ff0a087 100644
--- a/api/core/model_runtime/model_providers/tongyi/llm/qwen-math-plus.yaml
+++ b/api/core/model_runtime/model_providers/tongyi/llm/qwen-math-plus.yaml
@@ -1,3 +1,4 @@
+# for more details, please refer to https://help.aliyun.com/zh/model-studio/getting-started/models
 model: qwen-math-plus
 label:
   en_US: qwen-math-plus
diff --git a/api/core/model_runtime/model_providers/tongyi/llm/qwen-math-turbo-0919.yaml b/api/core/model_runtime/model_providers/tongyi/llm/qwen-math-turbo-0919.yaml
index bffe324a9649c..4da90eec3eddf 100644
--- a/api/core/model_runtime/model_providers/tongyi/llm/qwen-math-turbo-0919.yaml
+++ b/api/core/model_runtime/model_providers/tongyi/llm/qwen-math-turbo-0919.yaml
@@ -1,3 +1,4 @@
+# for more details, please refer to https://help.aliyun.com/zh/model-studio/getting-started/models
 model: qwen-math-turbo-0919
 label:
   en_US: qwen-math-turbo-0919
diff --git a/api/core/model_runtime/model_providers/tongyi/llm/qwen-math-turbo-latest.yaml b/api/core/model_runtime/model_providers/tongyi/llm/qwen-math-turbo-latest.yaml
index 0747e96614219..d29f8851dd390 100644
--- a/api/core/model_runtime/model_providers/tongyi/llm/qwen-math-turbo-latest.yaml
+++ b/api/core/model_runtime/model_providers/tongyi/llm/qwen-math-turbo-latest.yaml
@@ -1,3 +1,4 @@
+# for more details, please refer to https://help.aliyun.com/zh/model-studio/getting-started/models
 model: qwen-math-turbo-latest
 label:
   en_US: qwen-math-turbo-latest
diff --git a/api/core/model_runtime/model_providers/tongyi/llm/qwen-math-turbo.yaml b/api/core/model_runtime/model_providers/tongyi/llm/qwen-math-turbo.yaml
index dffb5557ffb7c..2a8f7f725e936 100644
--- a/api/core/model_runtime/model_providers/tongyi/llm/qwen-math-turbo.yaml
+++ b/api/core/model_runtime/model_providers/tongyi/llm/qwen-math-turbo.yaml
@@ -1,3 +1,4 @@
+# for more details, please refer to https://help.aliyun.com/zh/model-studio/getting-started/models
 model: qwen-math-turbo
 label:
   en_US: qwen-math-turbo
diff --git a/api/core/model_runtime/model_providers/tongyi/llm/qwen-max-0107.yaml b/api/core/model_runtime/model_providers/tongyi/llm/qwen-max-0107.yaml
index 8ae159f1bfb13..ef1841b5173bc 100644
--- a/api/core/model_runtime/model_providers/tongyi/llm/qwen-max-0107.yaml
+++ b/api/core/model_runtime/model_providers/tongyi/llm/qwen-max-0107.yaml
@@ -1,3 +1,5 @@
+# this model corresponds to qwen-max, for more details
+# please refer to (https://help.aliyun.com/zh/model-studio/getting-started/models#cf6cc4aa2aokf)
 model: qwen-max-0107
 label:
   en_US: qwen-max-0107
diff --git a/api/core/model_runtime/model_providers/tongyi/llm/qwen-max-0403.yaml b/api/core/model_runtime/model_providers/tongyi/llm/qwen-max-0403.yaml
index 93fb37254e4c8..a2ea5df130f37 100644
--- a/api/core/model_runtime/model_providers/tongyi/llm/qwen-max-0403.yaml
+++ b/api/core/model_runtime/model_providers/tongyi/llm/qwen-max-0403.yaml
@@ -1,3 +1,5 @@
+# this model corresponds to qwen-max-0403, for more details
+# please refer to (https://help.aliyun.com/zh/model-studio/getting-started/models#cf6cc4aa2aokf)
 model: qwen-max-0403
 label:
   en_US: qwen-max-0403
diff --git a/api/core/model_runtime/model_providers/tongyi/llm/qwen-max-0428.yaml b/api/core/model_runtime/model_providers/tongyi/llm/qwen-max-0428.yaml
index a5c9d49609041..a467665f118a6 100644
--- a/api/core/model_runtime/model_providers/tongyi/llm/qwen-max-0428.yaml
+++ b/api/core/model_runtime/model_providers/tongyi/llm/qwen-max-0428.yaml
@@ -1,3 +1,5 @@
+# this model corresponds to qwen-max-0428, for more details
+# please refer to (https://help.aliyun.com/zh/model-studio/getting-started/models#cf6cc4aa2aokf)
 model: qwen-max-0428
 label:
   en_US: qwen-max-0428
diff --git a/api/core/model_runtime/model_providers/tongyi/llm/qwen-max-0919.yaml b/api/core/model_runtime/model_providers/tongyi/llm/qwen-max-0919.yaml
index e4a6dae63736a..78661eaea065f 100644
--- a/api/core/model_runtime/model_providers/tongyi/llm/qwen-max-0919.yaml
+++ b/api/core/model_runtime/model_providers/tongyi/llm/qwen-max-0919.yaml
@@ -1,3 +1,5 @@
+# this model corresponds to qwen-max-0919, for more details
+# please refer to (https://help.aliyun.com/zh/model-studio/getting-started/models#cf6cc4aa2aokf)
 model: qwen-max-0919
 label:
   en_US: qwen-max-0919
diff --git a/api/core/model_runtime/model_providers/tongyi/llm/qwen-max-1201.yaml b/api/core/model_runtime/model_providers/tongyi/llm/qwen-max-1201.yaml
index 6fae8a7d38f36..6f4674576b442 100644
--- a/api/core/model_runtime/model_providers/tongyi/llm/qwen-max-1201.yaml
+++ b/api/core/model_runtime/model_providers/tongyi/llm/qwen-max-1201.yaml
@@ -1,3 +1,5 @@
+# this model corresponds to qwen-max, for more details
+# please refer to (https://help.aliyun.com/zh/model-studio/getting-started/models#cf6cc4aa2aokf)
 model: qwen-max-1201
 label:
   en_US: qwen-max-1201
diff --git a/api/core/model_runtime/model_providers/tongyi/llm/qwen-max-latest.yaml b/api/core/model_runtime/model_providers/tongyi/llm/qwen-max-latest.yaml
index 8e2096885957e..8b5f005473345 100644
--- a/api/core/model_runtime/model_providers/tongyi/llm/qwen-max-latest.yaml
+++ b/api/core/model_runtime/model_providers/tongyi/llm/qwen-max-latest.yaml
@@ -1,3 +1,5 @@
+# this model corresponds to qwen-max, for more details
+# please refer to (https://help.aliyun.com/zh/model-studio/getting-started/models#cf6cc4aa2aokf)
 model: qwen-max-latest
 label:
   en_US: qwen-max-latest
diff --git a/api/core/model_runtime/model_providers/tongyi/llm/qwen-max-longcontext.yaml b/api/core/model_runtime/model_providers/tongyi/llm/qwen-max-longcontext.yaml
index 9bc50c73fcc3e..098494ff95012 100644
--- a/api/core/model_runtime/model_providers/tongyi/llm/qwen-max-longcontext.yaml
+++ b/api/core/model_runtime/model_providers/tongyi/llm/qwen-max-longcontext.yaml
@@ -1,3 +1,5 @@
+# this model corresponds to qwen-max, for more details
+# please refer to (https://help.aliyun.com/zh/model-studio/getting-started/models#cf6cc4aa2aokf)
 model: qwen-max-longcontext
 label:
   en_US: qwen-max-longcontext
diff --git a/api/core/model_runtime/model_providers/tongyi/llm/qwen-max.yaml b/api/core/model_runtime/model_providers/tongyi/llm/qwen-max.yaml
index c6a64dc507497..9d0d3f8db39c2 100644
--- a/api/core/model_runtime/model_providers/tongyi/llm/qwen-max.yaml
+++ b/api/core/model_runtime/model_providers/tongyi/llm/qwen-max.yaml
@@ -1,3 +1,5 @@
+# this model corresponds to qwen-max, for more details
+# please refer to (https://help.aliyun.com/zh/model-studio/getting-started/models#cf6cc4aa2aokf)
 model: qwen-max
 label:
   en_US: qwen-max
diff --git a/api/core/model_runtime/model_providers/tongyi/llm/qwen-plus-0206.yaml b/api/core/model_runtime/model_providers/tongyi/llm/qwen-plus-0206.yaml
index 430599300bccb..0b1a6f81df80c 100644
--- a/api/core/model_runtime/model_providers/tongyi/llm/qwen-plus-0206.yaml
+++ b/api/core/model_runtime/model_providers/tongyi/llm/qwen-plus-0206.yaml
@@ -1,3 +1,5 @@
+# this model corresponds to qwen-plus-0206, for more details
+# please refer to (https://help.aliyun.com/zh/model-studio/getting-started/models#bb0ffee88bwnk)
 model: qwen-plus-0206
 label:
   en_US: qwen-plus-0206
diff --git a/api/core/model_runtime/model_providers/tongyi/llm/qwen-plus-0624.yaml b/api/core/model_runtime/model_providers/tongyi/llm/qwen-plus-0624.yaml
index 906995d2b95c8..7706005bb535c 100644
--- a/api/core/model_runtime/model_providers/tongyi/llm/qwen-plus-0624.yaml
+++ b/api/core/model_runtime/model_providers/tongyi/llm/qwen-plus-0624.yaml
@@ -1,3 +1,5 @@
+# this model corresponds to qwen-plus-0624, for more details
+# please refer to (https://help.aliyun.com/zh/model-studio/getting-started/models#bb0ffee88bwnk)
 model: qwen-plus-0624
 label:
   en_US: qwen-plus-0624
diff --git a/api/core/model_runtime/model_providers/tongyi/llm/qwen-plus-0723.yaml b/api/core/model_runtime/model_providers/tongyi/llm/qwen-plus-0723.yaml
index b33e725dd08c3..348276fc08f98 100644
--- a/api/core/model_runtime/model_providers/tongyi/llm/qwen-plus-0723.yaml
+++ b/api/core/model_runtime/model_providers/tongyi/llm/qwen-plus-0723.yaml
@@ -1,3 +1,5 @@
+# this model corresponds to qwen-plus-0723, for more details
+# please refer to (https://help.aliyun.com/zh/model-studio/getting-started/models#bb0ffee88bwnk)
 model: qwen-plus-0723
 label:
   en_US: qwen-plus-0723
diff --git a/api/core/model_runtime/model_providers/tongyi/llm/qwen-plus-0806.yaml b/api/core/model_runtime/model_providers/tongyi/llm/qwen-plus-0806.yaml
index bb394fad81455..29f125135eaa3 100644
--- a/api/core/model_runtime/model_providers/tongyi/llm/qwen-plus-0806.yaml
+++ b/api/core/model_runtime/model_providers/tongyi/llm/qwen-plus-0806.yaml
@@ -1,3 +1,5 @@
+# this model corresponds to qwen-plus-0806, for more details
+# please refer to (https://help.aliyun.com/zh/model-studio/getting-started/models#bb0ffee88bwnk)
 model: qwen-plus-0806
 label:
   en_US: qwen-plus-0806
diff --git a/api/core/model_runtime/model_providers/tongyi/llm/qwen-plus-0919.yaml b/api/core/model_runtime/model_providers/tongyi/llm/qwen-plus-0919.yaml
index 118e304a9723a..905fa1e1028bb 100644
--- a/api/core/model_runtime/model_providers/tongyi/llm/qwen-plus-0919.yaml
+++ b/api/core/model_runtime/model_providers/tongyi/llm/qwen-plus-0919.yaml
@@ -1,3 +1,5 @@
+# this model corresponds to qwen-plus-0919, for more details
+# please refer to (https://help.aliyun.com/zh/model-studio/getting-started/models#bb0ffee88bwnk)
 model: qwen-plus-0919
 label:
   en_US: qwen-plus-0919
diff --git a/api/core/model_runtime/model_providers/tongyi/llm/qwen-plus-chat.yaml b/api/core/model_runtime/model_providers/tongyi/llm/qwen-plus-chat.yaml
index 761312bc38123..c7a3549727ce8 100644
--- a/api/core/model_runtime/model_providers/tongyi/llm/qwen-plus-chat.yaml
+++ b/api/core/model_runtime/model_providers/tongyi/llm/qwen-plus-chat.yaml
@@ -1,3 +1,5 @@
+# this model corresponds to qwen-plus, for more details
+# please refer to (https://help.aliyun.com/zh/model-studio/getting-started/models#bb0ffee88bwnk)
 model: qwen-plus-chat
 label:
   en_US: qwen-plus-chat
diff --git a/api/core/model_runtime/model_providers/tongyi/llm/qwen-plus-latest.yaml b/api/core/model_runtime/model_providers/tongyi/llm/qwen-plus-latest.yaml
index 430872fb31a44..608f52c2964ea 100644
--- a/api/core/model_runtime/model_providers/tongyi/llm/qwen-plus-latest.yaml
+++ b/api/core/model_runtime/model_providers/tongyi/llm/qwen-plus-latest.yaml
@@ -1,3 +1,5 @@
+# this model corresponds to qwen-plus-latest, for more details
+# please refer to (https://help.aliyun.com/zh/model-studio/getting-started/models#bb0ffee88bwnk)
 model: qwen-plus-latest
 label:
   en_US: qwen-plus-latest
diff --git a/api/core/model_runtime/model_providers/tongyi/llm/qwen-plus.yaml b/api/core/model_runtime/model_providers/tongyi/llm/qwen-plus.yaml
index f3fce30209e48..9089e57255bb7 100644
--- a/api/core/model_runtime/model_providers/tongyi/llm/qwen-plus.yaml
+++ b/api/core/model_runtime/model_providers/tongyi/llm/qwen-plus.yaml
@@ -1,3 +1,5 @@
+# this model corresponds to qwen-plus, for more details
+# please refer to (https://help.aliyun.com/zh/model-studio/getting-started/models#bb0ffee88bwnk)
 model: qwen-plus
 label:
   en_US: qwen-plus
diff --git a/api/core/model_runtime/model_providers/tongyi/llm/qwen-turbo-0206.yaml b/api/core/model_runtime/model_providers/tongyi/llm/qwen-turbo-0206.yaml
index 2628d824fe824..7ee0d44f2f283 100644
--- a/api/core/model_runtime/model_providers/tongyi/llm/qwen-turbo-0206.yaml
+++ b/api/core/model_runtime/model_providers/tongyi/llm/qwen-turbo-0206.yaml
@@ -1,3 +1,6 @@
+# this model corresponds to qwen-turbo-0206, for more details
+# please refer to (https://help.aliyun.com/zh/model-studio/getting-started/models#ff492e2c10lub)
+
 model: qwen-turbo-0206
 label:
   en_US: qwen-turbo-0206
diff --git a/api/core/model_runtime/model_providers/tongyi/llm/qwen-turbo-0624.yaml b/api/core/model_runtime/model_providers/tongyi/llm/qwen-turbo-0624.yaml
index 8097459bf0906..20a3f7eb6460f 100644
--- a/api/core/model_runtime/model_providers/tongyi/llm/qwen-turbo-0624.yaml
+++ b/api/core/model_runtime/model_providers/tongyi/llm/qwen-turbo-0624.yaml
@@ -1,3 +1,5 @@
+# this model corresponds to qwen-turbo-0624, for more details
+# please refer to (https://help.aliyun.com/zh/model-studio/getting-started/models#ff492e2c10lub)
 model: qwen-turbo-0624
 label:
   en_US: qwen-turbo-0624
diff --git a/api/core/model_runtime/model_providers/tongyi/llm/qwen-turbo-0919.yaml b/api/core/model_runtime/model_providers/tongyi/llm/qwen-turbo-0919.yaml
index e43beeb195aec..ba73dec3631fb 100644
--- a/api/core/model_runtime/model_providers/tongyi/llm/qwen-turbo-0919.yaml
+++ b/api/core/model_runtime/model_providers/tongyi/llm/qwen-turbo-0919.yaml
@@ -1,3 +1,5 @@
+# this model corresponds to qwen-turbo-0919, for more details
+# please refer to (https://help.aliyun.com/zh/model-studio/getting-started/models#ff492e2c10lub)
 model: qwen-turbo-0919
 label:
   en_US: qwen-turbo-0919
diff --git a/api/core/model_runtime/model_providers/tongyi/llm/qwen-turbo-chat.yaml b/api/core/model_runtime/model_providers/tongyi/llm/qwen-turbo-chat.yaml
index c30cb7ca10096..d785b7fe85787 100644
--- a/api/core/model_runtime/model_providers/tongyi/llm/qwen-turbo-chat.yaml
+++ b/api/core/model_runtime/model_providers/tongyi/llm/qwen-turbo-chat.yaml
@@ -1,3 +1,5 @@
+# this model corresponds to qwen-turbo, for more details
+# please refer to (https://help.aliyun.com/zh/model-studio/getting-started/models#ff492e2c10lub)
 model: qwen-turbo-chat
 label:
   en_US: qwen-turbo-chat
diff --git a/api/core/model_runtime/model_providers/tongyi/llm/qwen-turbo-latest.yaml b/api/core/model_runtime/model_providers/tongyi/llm/qwen-turbo-latest.yaml
index e443d6888b7ef..fe38a4283c2d1 100644
--- a/api/core/model_runtime/model_providers/tongyi/llm/qwen-turbo-latest.yaml
+++ b/api/core/model_runtime/model_providers/tongyi/llm/qwen-turbo-latest.yaml
@@ -1,3 +1,5 @@
+# this model corresponds to qwen-turbo-latest, for more details
+# please refer to (https://help.aliyun.com/zh/model-studio/getting-started/models#ff492e2c10lub)
 model: qwen-turbo-latest
 label:
   en_US: qwen-turbo-latest
diff --git a/api/core/model_runtime/model_providers/tongyi/llm/qwen-turbo.yaml b/api/core/model_runtime/model_providers/tongyi/llm/qwen-turbo.yaml
index 33f05967c2a46..215c9ec5fc96a 100644
--- a/api/core/model_runtime/model_providers/tongyi/llm/qwen-turbo.yaml
+++ b/api/core/model_runtime/model_providers/tongyi/llm/qwen-turbo.yaml
@@ -1,3 +1,5 @@
+# this model corresponds to qwen-turbo, for more details
+# please refer to (https://help.aliyun.com/zh/model-studio/getting-started/models#ff492e2c10lub)
 model: qwen-turbo
 label:
   en_US: qwen-turbo
diff --git a/api/core/model_runtime/model_providers/tongyi/llm/qwen-vl-max-0201.yaml b/api/core/model_runtime/model_providers/tongyi/llm/qwen-vl-max-0201.yaml
index 63b6074d0d7c1..d80168ffc3fb5 100644
--- a/api/core/model_runtime/model_providers/tongyi/llm/qwen-vl-max-0201.yaml
+++ b/api/core/model_runtime/model_providers/tongyi/llm/qwen-vl-max-0201.yaml
@@ -1,3 +1,4 @@
+# for more details, please refer to https://help.aliyun.com/zh/model-studio/getting-started/models
 model: qwen-vl-max-0201
 label:
   en_US: qwen-vl-max-0201
diff --git a/api/core/model_runtime/model_providers/tongyi/llm/qwen-vl-max-0809.yaml b/api/core/model_runtime/model_providers/tongyi/llm/qwen-vl-max-0809.yaml
index fd20377002042..50e10226a5f5c 100644
--- a/api/core/model_runtime/model_providers/tongyi/llm/qwen-vl-max-0809.yaml
+++ b/api/core/model_runtime/model_providers/tongyi/llm/qwen-vl-max-0809.yaml
@@ -1,3 +1,4 @@
+# for more details, please refer to https://help.aliyun.com/zh/model-studio/getting-started/models
 model: qwen-vl-max-0809
 label:
   en_US: qwen-vl-max-0809
diff --git a/api/core/model_runtime/model_providers/tongyi/llm/qwen-vl-max.yaml b/api/core/model_runtime/model_providers/tongyi/llm/qwen-vl-max.yaml
index 31a9fb51bbbb5..21b127f56c47d 100644
--- a/api/core/model_runtime/model_providers/tongyi/llm/qwen-vl-max.yaml
+++ b/api/core/model_runtime/model_providers/tongyi/llm/qwen-vl-max.yaml
@@ -1,3 +1,4 @@
+# for more details, please refer to https://help.aliyun.com/zh/model-studio/getting-started/models
 model: qwen-vl-max
 label:
   en_US: qwen-vl-max
diff --git a/api/core/model_runtime/model_providers/tongyi/llm/qwen-vl-plus-0201.yaml b/api/core/model_runtime/model_providers/tongyi/llm/qwen-vl-plus-0201.yaml
index 5f90cf48bc999..03cb039d15a7d 100644
--- a/api/core/model_runtime/model_providers/tongyi/llm/qwen-vl-plus-0201.yaml
+++ b/api/core/model_runtime/model_providers/tongyi/llm/qwen-vl-plus-0201.yaml
@@ -1,3 +1,4 @@
+# for more details, please refer to https://help.aliyun.com/zh/model-studio/getting-started/models
 model: qwen-vl-plus-0201
 label:
   en_US: qwen-vl-plus-0201
diff --git a/api/core/model_runtime/model_providers/tongyi/llm/qwen-vl-plus-0809.yaml b/api/core/model_runtime/model_providers/tongyi/llm/qwen-vl-plus-0809.yaml
index 97820c0f3a1a2..67b2b2ebddc61 100644
--- a/api/core/model_runtime/model_providers/tongyi/llm/qwen-vl-plus-0809.yaml
+++ b/api/core/model_runtime/model_providers/tongyi/llm/qwen-vl-plus-0809.yaml
@@ -1,3 +1,4 @@
+# for more details, please refer to https://help.aliyun.com/zh/model-studio/getting-started/models
 model: qwen-vl-plus-0809
 label:
   en_US: qwen-vl-plus-0809
diff --git a/api/core/model_runtime/model_providers/tongyi/llm/qwen-vl-plus.yaml b/api/core/model_runtime/model_providers/tongyi/llm/qwen-vl-plus.yaml
index 6af36cd6f3d63..f55764c6c0550 100644
--- a/api/core/model_runtime/model_providers/tongyi/llm/qwen-vl-plus.yaml
+++ b/api/core/model_runtime/model_providers/tongyi/llm/qwen-vl-plus.yaml
@@ -1,3 +1,4 @@
+# for more details, please refer to https://help.aliyun.com/zh/model-studio/getting-started/models
 model: qwen-vl-plus
 label:
   en_US: qwen-vl-plus
diff --git a/api/core/model_runtime/model_providers/tongyi/llm/qwen2-math-1.5b-instruct.yaml b/api/core/model_runtime/model_providers/tongyi/llm/qwen2-math-1.5b-instruct.yaml
index 158e2c7ee16e5..ea157f42ded91 100644
--- a/api/core/model_runtime/model_providers/tongyi/llm/qwen2-math-1.5b-instruct.yaml
+++ b/api/core/model_runtime/model_providers/tongyi/llm/qwen2-math-1.5b-instruct.yaml
@@ -1,3 +1,4 @@
+# for more details, please refer to https://help.aliyun.com/zh/model-studio/getting-started/models
 model: qwen2-math-1.5b-instruct
 label:
   en_US: qwen2-math-1.5b-instruct
diff --git a/api/core/model_runtime/model_providers/tongyi/llm/qwen2-math-72b-instruct.yaml b/api/core/model_runtime/model_providers/tongyi/llm/qwen2-math-72b-instruct.yaml
index e26a6923d12f2..37052a923317d 100644
--- a/api/core/model_runtime/model_providers/tongyi/llm/qwen2-math-72b-instruct.yaml
+++ b/api/core/model_runtime/model_providers/tongyi/llm/qwen2-math-72b-instruct.yaml
@@ -1,3 +1,4 @@
+# for more details, please refer to https://help.aliyun.com/zh/model-studio/getting-started/models
 model: qwen2-math-72b-instruct
 label:
   en_US: qwen2-math-72b-instruct
diff --git a/api/core/model_runtime/model_providers/tongyi/llm/qwen2-math-7b-instruct.yaml b/api/core/model_runtime/model_providers/tongyi/llm/qwen2-math-7b-instruct.yaml
index 589119b26ea3c..e182f1c27f7ea 100644
--- a/api/core/model_runtime/model_providers/tongyi/llm/qwen2-math-7b-instruct.yaml
+++ b/api/core/model_runtime/model_providers/tongyi/llm/qwen2-math-7b-instruct.yaml
@@ -1,3 +1,4 @@
+# for more details, please refer to https://help.aliyun.com/zh/model-studio/getting-started/models
 model: qwen2-math-7b-instruct
 label:
   en_US: qwen2-math-7b-instruct
diff --git a/api/core/model_runtime/model_providers/tongyi/llm/qwen2.5-0.5b-instruct.yaml b/api/core/model_runtime/model_providers/tongyi/llm/qwen2.5-0.5b-instruct.yaml
index dd608fbf7675a..9e75ccc1f210d 100644
--- a/api/core/model_runtime/model_providers/tongyi/llm/qwen2.5-0.5b-instruct.yaml
+++ b/api/core/model_runtime/model_providers/tongyi/llm/qwen2.5-0.5b-instruct.yaml
@@ -1,3 +1,4 @@
+# for more details, please refer to https://help.aliyun.com/zh/model-studio/getting-started/models
 model: qwen2.5-0.5b-instruct
 label:
   en_US: qwen2.5-0.5b-instruct
diff --git a/api/core/model_runtime/model_providers/tongyi/llm/qwen2.5-1.5b-instruct.yaml b/api/core/model_runtime/model_providers/tongyi/llm/qwen2.5-1.5b-instruct.yaml
index 08237b3958b32..67c9d312432af 100644
--- a/api/core/model_runtime/model_providers/tongyi/llm/qwen2.5-1.5b-instruct.yaml
+++ b/api/core/model_runtime/model_providers/tongyi/llm/qwen2.5-1.5b-instruct.yaml
@@ -1,3 +1,4 @@
+# for more details, please refer to https://help.aliyun.com/zh/model-studio/getting-started/models
 model: qwen2.5-1.5b-instruct
 label:
   en_US: qwen2.5-1.5b-instruct
diff --git a/api/core/model_runtime/model_providers/tongyi/llm/qwen2.5-14b-instruct.yaml b/api/core/model_runtime/model_providers/tongyi/llm/qwen2.5-14b-instruct.yaml
index 640b01970351b..2a38be921cf3f 100644
--- a/api/core/model_runtime/model_providers/tongyi/llm/qwen2.5-14b-instruct.yaml
+++ b/api/core/model_runtime/model_providers/tongyi/llm/qwen2.5-14b-instruct.yaml
@@ -1,3 +1,4 @@
+# for more details, please refer to https://help.aliyun.com/zh/model-studio/getting-started/models
 model: qwen2.5-14b-instruct
 label:
   en_US: qwen2.5-14b-instruct
diff --git a/api/core/model_runtime/model_providers/tongyi/llm/qwen2.5-32b-instruct.yaml b/api/core/model_runtime/model_providers/tongyi/llm/qwen2.5-32b-instruct.yaml
index 3a90ca753273d..e6e4fbf97808b 100644
--- a/api/core/model_runtime/model_providers/tongyi/llm/qwen2.5-32b-instruct.yaml
+++ b/api/core/model_runtime/model_providers/tongyi/llm/qwen2.5-32b-instruct.yaml
@@ -1,3 +1,4 @@
+# for more details, please refer to https://help.aliyun.com/zh/model-studio/getting-started/models
 model: qwen2.5-32b-instruct
 label:
   en_US: qwen2.5-32b-instruct
diff --git a/api/core/model_runtime/model_providers/tongyi/llm/qwen2.5-3b-instruct.yaml b/api/core/model_runtime/model_providers/tongyi/llm/qwen2.5-3b-instruct.yaml
index b79755eb9bd84..8f250379a788a 100644
--- a/api/core/model_runtime/model_providers/tongyi/llm/qwen2.5-3b-instruct.yaml
+++ b/api/core/model_runtime/model_providers/tongyi/llm/qwen2.5-3b-instruct.yaml
@@ -1,3 +1,4 @@
+# for more details, please refer to https://help.aliyun.com/zh/model-studio/getting-started/models
 model: qwen2.5-3b-instruct
 label:
   en_US: qwen2.5-3b-instruct
diff --git a/api/core/model_runtime/model_providers/tongyi/llm/qwen2.5-72b-instruct.yaml b/api/core/model_runtime/model_providers/tongyi/llm/qwen2.5-72b-instruct.yaml
index e9dd51a341624..bb3cdd6141f1e 100644
--- a/api/core/model_runtime/model_providers/tongyi/llm/qwen2.5-72b-instruct.yaml
+++ b/api/core/model_runtime/model_providers/tongyi/llm/qwen2.5-72b-instruct.yaml
@@ -1,3 +1,4 @@
+# for more details, please refer to https://help.aliyun.com/zh/model-studio/getting-started/models
 model: qwen2.5-72b-instruct
 label:
   en_US: qwen2.5-72b-instruct
diff --git a/api/core/model_runtime/model_providers/tongyi/llm/qwen2.5-7b-instruct.yaml b/api/core/model_runtime/model_providers/tongyi/llm/qwen2.5-7b-instruct.yaml
index 04f26cf5fe12b..fdcd3d42757ed 100644
--- a/api/core/model_runtime/model_providers/tongyi/llm/qwen2.5-7b-instruct.yaml
+++ b/api/core/model_runtime/model_providers/tongyi/llm/qwen2.5-7b-instruct.yaml
@@ -1,3 +1,4 @@
+# for more details, please refer to https://help.aliyun.com/zh/model-studio/getting-started/models
 model: qwen2.5-7b-instruct
 label:
   en_US: qwen2.5-7b-instruct
diff --git a/api/core/model_runtime/model_providers/tongyi/llm/qwen2.5-coder-7b-instruct.yaml b/api/core/model_runtime/model_providers/tongyi/llm/qwen2.5-coder-7b-instruct.yaml
index 04f26cf5fe12b..fdcd3d42757ed 100644
--- a/api/core/model_runtime/model_providers/tongyi/llm/qwen2.5-coder-7b-instruct.yaml
+++ b/api/core/model_runtime/model_providers/tongyi/llm/qwen2.5-coder-7b-instruct.yaml
@@ -1,3 +1,4 @@
+# for more details, please refer to https://help.aliyun.com/zh/model-studio/getting-started/models
 model: qwen2.5-7b-instruct
 label:
   en_US: qwen2.5-7b-instruct
diff --git a/api/core/model_runtime/model_providers/tongyi/text_embedding/text-embedding-v1.yaml b/api/core/model_runtime/model_providers/tongyi/text_embedding/text-embedding-v1.yaml
index f4303c53d38b8..52e35d8b50afd 100644
--- a/api/core/model_runtime/model_providers/tongyi/text_embedding/text-embedding-v1.yaml
+++ b/api/core/model_runtime/model_providers/tongyi/text_embedding/text-embedding-v1.yaml
@@ -1,3 +1,4 @@
+# for more details, please refer to https://help.aliyun.com/zh/model-studio/getting-started/models#3383780daf8hw
 model: text-embedding-v1
 model_type: text-embedding
 model_properties:
diff --git a/api/core/model_runtime/model_providers/tongyi/text_embedding/text-embedding-v2.yaml b/api/core/model_runtime/model_providers/tongyi/text_embedding/text-embedding-v2.yaml
index f6be3544ed8f6..5bb6a8f4243d5 100644
--- a/api/core/model_runtime/model_providers/tongyi/text_embedding/text-embedding-v2.yaml
+++ b/api/core/model_runtime/model_providers/tongyi/text_embedding/text-embedding-v2.yaml
@@ -1,3 +1,4 @@
+# for more details, please refer to https://help.aliyun.com/zh/model-studio/getting-started/models#3383780daf8hw
 model: text-embedding-v2
 model_type: text-embedding
 model_properties:
diff --git a/api/core/model_runtime/model_providers/tongyi/text_embedding/text-embedding-v3.yaml b/api/core/model_runtime/model_providers/tongyi/text_embedding/text-embedding-v3.yaml
index 171a379ee23f7..d8af0e2b63565 100644
--- a/api/core/model_runtime/model_providers/tongyi/text_embedding/text-embedding-v3.yaml
+++ b/api/core/model_runtime/model_providers/tongyi/text_embedding/text-embedding-v3.yaml
@@ -1,3 +1,4 @@
+# for more details, please refer to https://help.aliyun.com/zh/model-studio/getting-started/models#3383780daf8hw
 model: text-embedding-v3
 model_type: text-embedding
 model_properties:

From e5efd09ebbd3e327efdc4c96075a4723c17927a0 Mon Sep 17 00:00:00 2001
From: CXwudi <cx2010017@gmail.com>
Date: Thu, 26 Sep 2024 21:14:33 -0400
Subject: [PATCH 085/235] chore: massive update of the Gemini models based on
 latest documentation (#8822)

---
 .../google/llm/gemini-1.5-flash-001.yaml      | 48 +++++++++++++++++++
 .../google/llm/gemini-1.5-flash-002.yaml      | 48 +++++++++++++++++++
 .../llm/gemini-1.5-flash-8b-exp-0924.yaml     | 48 +++++++++++++++++++
 .../google/llm/gemini-1.5-flash-latest.yaml   |  2 +-
 .../google/llm/gemini-1.5-flash.yaml          | 48 +++++++++++++++++++
 .../google/llm/gemini-1.5-pro-001.yaml        | 48 +++++++++++++++++++
 .../google/llm/gemini-1.5-pro-002.yaml        | 48 +++++++++++++++++++
 .../google/llm/gemini-1.5-pro-latest.yaml     |  2 +-
 .../google/llm/gemini-1.5-pro.yaml            | 48 +++++++++++++++++++
 9 files changed, 338 insertions(+), 2 deletions(-)
 create mode 100644 api/core/model_runtime/model_providers/google/llm/gemini-1.5-flash-001.yaml
 create mode 100644 api/core/model_runtime/model_providers/google/llm/gemini-1.5-flash-002.yaml
 create mode 100644 api/core/model_runtime/model_providers/google/llm/gemini-1.5-flash-8b-exp-0924.yaml
 create mode 100644 api/core/model_runtime/model_providers/google/llm/gemini-1.5-flash.yaml
 create mode 100644 api/core/model_runtime/model_providers/google/llm/gemini-1.5-pro-001.yaml
 create mode 100644 api/core/model_runtime/model_providers/google/llm/gemini-1.5-pro-002.yaml
 create mode 100644 api/core/model_runtime/model_providers/google/llm/gemini-1.5-pro.yaml

diff --git a/api/core/model_runtime/model_providers/google/llm/gemini-1.5-flash-001.yaml b/api/core/model_runtime/model_providers/google/llm/gemini-1.5-flash-001.yaml
new file mode 100644
index 0000000000000..d84e9937e0b66
--- /dev/null
+++ b/api/core/model_runtime/model_providers/google/llm/gemini-1.5-flash-001.yaml
@@ -0,0 +1,48 @@
+model: gemini-1.5-flash-001
+label:
+  en_US: Gemini 1.5 Flash 001
+model_type: llm
+features:
+  - agent-thought
+  - vision
+  - tool-call
+  - stream-tool-call
+model_properties:
+  mode: chat
+  context_size: 1048576
+parameter_rules:
+  - name: temperature
+    use_template: temperature
+  - name: top_p
+    use_template: top_p
+  - name: top_k
+    label:
+      zh_Hans: 取样数量
+      en_US: Top k
+    type: int
+    help:
+      zh_Hans: 仅从每个后续标记的前 K 个选项中采样。
+      en_US: Only sample from the top K options for each subsequent token.
+    required: false
+  - name: max_tokens_to_sample
+    use_template: max_tokens
+    required: true
+    default: 8192
+    min: 1
+    max: 8192
+  - name: response_format
+    use_template: response_format
+  - name: stream
+    label:
+      zh_Hans: 流式输出
+      en_US: Stream
+    type: boolean
+    help:
+      zh_Hans: 流式输出允许模型在生成文本的过程中逐步返回结果，而不是一次性生成全部结果后再返回。
+      en_US: Streaming output allows the model to return results incrementally as it generates text, rather than generating all the results at once.
+    default: false
+pricing:
+  input: '0.00'
+  output: '0.00'
+  unit: '0.000001'
+  currency: USD
diff --git a/api/core/model_runtime/model_providers/google/llm/gemini-1.5-flash-002.yaml b/api/core/model_runtime/model_providers/google/llm/gemini-1.5-flash-002.yaml
new file mode 100644
index 0000000000000..2ff70564b2a95
--- /dev/null
+++ b/api/core/model_runtime/model_providers/google/llm/gemini-1.5-flash-002.yaml
@@ -0,0 +1,48 @@
+model: gemini-1.5-flash-002
+label:
+  en_US: Gemini 1.5 Flash 002
+model_type: llm
+features:
+  - agent-thought
+  - vision
+  - tool-call
+  - stream-tool-call
+model_properties:
+  mode: chat
+  context_size: 1048576
+parameter_rules:
+  - name: temperature
+    use_template: temperature
+  - name: top_p
+    use_template: top_p
+  - name: top_k
+    label:
+      zh_Hans: 取样数量
+      en_US: Top k
+    type: int
+    help:
+      zh_Hans: 仅从每个后续标记的前 K 个选项中采样。
+      en_US: Only sample from the top K options for each subsequent token.
+    required: false
+  - name: max_tokens_to_sample
+    use_template: max_tokens
+    required: true
+    default: 8192
+    min: 1
+    max: 8192
+  - name: response_format
+    use_template: response_format
+  - name: stream
+    label:
+      zh_Hans: 流式输出
+      en_US: Stream
+    type: boolean
+    help:
+      zh_Hans: 流式输出允许模型在生成文本的过程中逐步返回结果，而不是一次性生成全部结果后再返回。
+      en_US: Streaming output allows the model to return results incrementally as it generates text, rather than generating all the results at once.
+    default: false
+pricing:
+  input: '0.00'
+  output: '0.00'
+  unit: '0.000001'
+  currency: USD
diff --git a/api/core/model_runtime/model_providers/google/llm/gemini-1.5-flash-8b-exp-0924.yaml b/api/core/model_runtime/model_providers/google/llm/gemini-1.5-flash-8b-exp-0924.yaml
new file mode 100644
index 0000000000000..2aea8149f4c79
--- /dev/null
+++ b/api/core/model_runtime/model_providers/google/llm/gemini-1.5-flash-8b-exp-0924.yaml
@@ -0,0 +1,48 @@
+model: gemini-1.5-flash-8b-exp-0924
+label:
+  en_US: Gemini 1.5 Flash 8B 0924
+model_type: llm
+features:
+  - agent-thought
+  - vision
+  - tool-call
+  - stream-tool-call
+model_properties:
+  mode: chat
+  context_size: 1048576
+parameter_rules:
+  - name: temperature
+    use_template: temperature
+  - name: top_p
+    use_template: top_p
+  - name: top_k
+    label:
+      zh_Hans: 取样数量
+      en_US: Top k
+    type: int
+    help:
+      zh_Hans: 仅从每个后续标记的前 K 个选项中采样。
+      en_US: Only sample from the top K options for each subsequent token.
+    required: false
+  - name: max_tokens_to_sample
+    use_template: max_tokens
+    required: true
+    default: 8192
+    min: 1
+    max: 8192
+  - name: response_format
+    use_template: response_format
+  - name: stream
+    label:
+      zh_Hans: 流式输出
+      en_US: Stream
+    type: boolean
+    help:
+      zh_Hans: 流式输出允许模型在生成文本的过程中逐步返回结果，而不是一次性生成全部结果后再返回。
+      en_US: Streaming output allows the model to return results incrementally as it generates text, rather than generating all the results at once.
+    default: false
+pricing:
+  input: '0.00'
+  output: '0.00'
+  unit: '0.000001'
+  currency: USD
diff --git a/api/core/model_runtime/model_providers/google/llm/gemini-1.5-flash-latest.yaml b/api/core/model_runtime/model_providers/google/llm/gemini-1.5-flash-latest.yaml
index 6a0344699a78b..a22fcca9419b9 100644
--- a/api/core/model_runtime/model_providers/google/llm/gemini-1.5-flash-latest.yaml
+++ b/api/core/model_runtime/model_providers/google/llm/gemini-1.5-flash-latest.yaml
@@ -1,6 +1,6 @@
 model: gemini-1.5-flash-latest
 label:
-  en_US: Gemini 1.5 Flash
+  en_US: Gemini 1.5 Flash Latest
 model_type: llm
 features:
   - agent-thought
diff --git a/api/core/model_runtime/model_providers/google/llm/gemini-1.5-flash.yaml b/api/core/model_runtime/model_providers/google/llm/gemini-1.5-flash.yaml
new file mode 100644
index 0000000000000..dfd55c3a949c9
--- /dev/null
+++ b/api/core/model_runtime/model_providers/google/llm/gemini-1.5-flash.yaml
@@ -0,0 +1,48 @@
+model: gemini-1.5-flash
+label:
+  en_US: Gemini 1.5 Flash
+model_type: llm
+features:
+  - agent-thought
+  - vision
+  - tool-call
+  - stream-tool-call
+model_properties:
+  mode: chat
+  context_size: 1048576
+parameter_rules:
+  - name: temperature
+    use_template: temperature
+  - name: top_p
+    use_template: top_p
+  - name: top_k
+    label:
+      zh_Hans: 取样数量
+      en_US: Top k
+    type: int
+    help:
+      zh_Hans: 仅从每个后续标记的前 K 个选项中采样。
+      en_US: Only sample from the top K options for each subsequent token.
+    required: false
+  - name: max_tokens_to_sample
+    use_template: max_tokens
+    required: true
+    default: 8192
+    min: 1
+    max: 8192
+  - name: response_format
+    use_template: response_format
+  - name: stream
+    label:
+      zh_Hans: 流式输出
+      en_US: Stream
+    type: boolean
+    help:
+      zh_Hans: 流式输出允许模型在生成文本的过程中逐步返回结果，而不是一次性生成全部结果后再返回。
+      en_US: Streaming output allows the model to return results incrementally as it generates text, rather than generating all the results at once.
+    default: false
+pricing:
+  input: '0.00'
+  output: '0.00'
+  unit: '0.000001'
+  currency: USD
diff --git a/api/core/model_runtime/model_providers/google/llm/gemini-1.5-pro-001.yaml b/api/core/model_runtime/model_providers/google/llm/gemini-1.5-pro-001.yaml
new file mode 100644
index 0000000000000..a1feff171d48c
--- /dev/null
+++ b/api/core/model_runtime/model_providers/google/llm/gemini-1.5-pro-001.yaml
@@ -0,0 +1,48 @@
+model: gemini-1.5-pro-001
+label:
+  en_US: Gemini 1.5 Pro 001
+model_type: llm
+features:
+  - agent-thought
+  - vision
+  - tool-call
+  - stream-tool-call
+model_properties:
+  mode: chat
+  context_size: 2097152
+parameter_rules:
+  - name: temperature
+    use_template: temperature
+  - name: top_p
+    use_template: top_p
+  - name: top_k
+    label:
+      zh_Hans: 取样数量
+      en_US: Top k
+    type: int
+    help:
+      zh_Hans: 仅从每个后续标记的前 K 个选项中采样。
+      en_US: Only sample from the top K options for each subsequent token.
+    required: false
+  - name: max_tokens_to_sample
+    use_template: max_tokens
+    required: true
+    default: 8192
+    min: 1
+    max: 8192
+  - name: response_format
+    use_template: response_format
+  - name: stream
+    label:
+      zh_Hans: 流式输出
+      en_US: Stream
+    type: boolean
+    help:
+      zh_Hans: 流式输出允许模型在生成文本的过程中逐步返回结果，而不是一次性生成全部结果后再返回。
+      en_US: Streaming output allows the model to return results incrementally as it generates text, rather than generating all the results at once.
+    default: false
+pricing:
+  input: '0.00'
+  output: '0.00'
+  unit: '0.000001'
+  currency: USD
diff --git a/api/core/model_runtime/model_providers/google/llm/gemini-1.5-pro-002.yaml b/api/core/model_runtime/model_providers/google/llm/gemini-1.5-pro-002.yaml
new file mode 100644
index 0000000000000..9ae07a06c5118
--- /dev/null
+++ b/api/core/model_runtime/model_providers/google/llm/gemini-1.5-pro-002.yaml
@@ -0,0 +1,48 @@
+model: gemini-1.5-pro-002
+label:
+  en_US: Gemini 1.5 Pro 002
+model_type: llm
+features:
+  - agent-thought
+  - vision
+  - tool-call
+  - stream-tool-call
+model_properties:
+  mode: chat
+  context_size: 2097152
+parameter_rules:
+  - name: temperature
+    use_template: temperature
+  - name: top_p
+    use_template: top_p
+  - name: top_k
+    label:
+      zh_Hans: 取样数量
+      en_US: Top k
+    type: int
+    help:
+      zh_Hans: 仅从每个后续标记的前 K 个选项中采样。
+      en_US: Only sample from the top K options for each subsequent token.
+    required: false
+  - name: max_tokens_to_sample
+    use_template: max_tokens
+    required: true
+    default: 8192
+    min: 1
+    max: 8192
+  - name: response_format
+    use_template: response_format
+  - name: stream
+    label:
+      zh_Hans: 流式输出
+      en_US: Stream
+    type: boolean
+    help:
+      zh_Hans: 流式输出允许模型在生成文本的过程中逐步返回结果，而不是一次性生成全部结果后再返回。
+      en_US: Streaming output allows the model to return results incrementally as it generates text, rather than generating all the results at once.
+    default: false
+pricing:
+  input: '0.00'
+  output: '0.00'
+  unit: '0.000001'
+  currency: USD
diff --git a/api/core/model_runtime/model_providers/google/llm/gemini-1.5-pro-latest.yaml b/api/core/model_runtime/model_providers/google/llm/gemini-1.5-pro-latest.yaml
index 92cd6b310d07e..d1bf7d269de76 100644
--- a/api/core/model_runtime/model_providers/google/llm/gemini-1.5-pro-latest.yaml
+++ b/api/core/model_runtime/model_providers/google/llm/gemini-1.5-pro-latest.yaml
@@ -1,6 +1,6 @@
 model: gemini-1.5-pro-latest
 label:
-  en_US: Gemini 1.5 Pro
+  en_US: Gemini 1.5 Pro Latest
 model_type: llm
 features:
   - agent-thought
diff --git a/api/core/model_runtime/model_providers/google/llm/gemini-1.5-pro.yaml b/api/core/model_runtime/model_providers/google/llm/gemini-1.5-pro.yaml
new file mode 100644
index 0000000000000..bdd70b34a21dc
--- /dev/null
+++ b/api/core/model_runtime/model_providers/google/llm/gemini-1.5-pro.yaml
@@ -0,0 +1,48 @@
+model: gemini-1.5-pro
+label:
+  en_US: Gemini 1.5 Pro
+model_type: llm
+features:
+  - agent-thought
+  - vision
+  - tool-call
+  - stream-tool-call
+model_properties:
+  mode: chat
+  context_size: 2097152
+parameter_rules:
+  - name: temperature
+    use_template: temperature
+  - name: top_p
+    use_template: top_p
+  - name: top_k
+    label:
+      zh_Hans: 取样数量
+      en_US: Top k
+    type: int
+    help:
+      zh_Hans: 仅从每个后续标记的前 K 个选项中采样。
+      en_US: Only sample from the top K options for each subsequent token.
+    required: false
+  - name: max_tokens_to_sample
+    use_template: max_tokens
+    required: true
+    default: 8192
+    min: 1
+    max: 8192
+  - name: response_format
+    use_template: response_format
+  - name: stream
+    label:
+      zh_Hans: 流式输出
+      en_US: Stream
+    type: boolean
+    help:
+      zh_Hans: 流式输出允许模型在生成文本的过程中逐步返回结果，而不是一次性生成全部结果后再返回。
+      en_US: Streaming output allows the model to return results incrementally as it generates text, rather than generating all the results at once.
+    default: false
+pricing:
+  input: '0.00'
+  output: '0.00'
+  unit: '0.000001'
+  currency: USD

From a36117e12d06707e63de5b8dc51d97bf060b4e7c Mon Sep 17 00:00:00 2001
From: Shai Perednik <shaiss@gmail.com>
Date: Thu, 26 Sep 2024 21:15:33 -0400
Subject: [PATCH 086/235] Updated the YouTube channel to Dify's (#8817)

---
 api/core/tools/provider/builtin/youtube/youtube.py | 2 +-
 1 file changed, 1 insertion(+), 1 deletion(-)

diff --git a/api/core/tools/provider/builtin/youtube/youtube.py b/api/core/tools/provider/builtin/youtube/youtube.py
index aad876491c85d..07e430bcbf27e 100644
--- a/api/core/tools/provider/builtin/youtube/youtube.py
+++ b/api/core/tools/provider/builtin/youtube/youtube.py
@@ -13,7 +13,7 @@ def _validate_credentials(self, credentials: dict) -> None:
             ).invoke(
                 user_id="",
                 tool_parameters={
-                    "channel": "TOKYO GIRLS COLLECTION",
+                    "channel": "UC2JZCsZSOudXA08cMMRCL9g",
                     "start_date": "2020-01-01",
                     "end_date": "2024-12-31",
                 },

From 6fbaabc1bc4b8b7830d55bac7609afdd703242f9 Mon Sep 17 00:00:00 2001
From: zhuhao <37029601+hwzhuhao@users.noreply.github.com>
Date: Fri, 27 Sep 2024 11:13:29 +0800
Subject: [PATCH 087/235] feat: add pgvecto-rs and analyticdb in
 docker/.env.example (#8823)

---
 docker/.env.example | 21 +++++++++++++++++++--
 1 file changed, 19 insertions(+), 2 deletions(-)

diff --git a/docker/.env.example b/docker/.env.example
index f7479791cea32..d43c3edc7e722 100644
--- a/docker/.env.example
+++ b/docker/.env.example
@@ -346,7 +346,7 @@ VOLCENGINE_TOS_REGION=your-region
 # ------------------------------
 
 # The type of vector store to use. 
-# Supported values are `weaviate`, `qdrant`, `milvus`, `myscale`, `relyt`, `pgvector`, `chroma`, `opensearch`, `tidb_vector`, `oracle`, `tencent`, `elasticsearch`.
+# Supported values are `weaviate`, `qdrant`, `milvus`, `myscale`, `relyt`, `pgvector`, `pgvecto-rs`, ``chroma`, `opensearch`, `tidb_vector`, `oracle`, `tencent`, `elasticsearch`, `analyticdb`.
 VECTOR_STORE=weaviate
 
 # The Weaviate endpoint URL. Only available when VECTOR_STORE is `weaviate`.
@@ -385,13 +385,30 @@ MYSCALE_PASSWORD=
 MYSCALE_DATABASE=dify
 MYSCALE_FTS_PARAMS=
 
-# pgvector configurations, only available when VECTOR_STORE is `pgvecto-rs or pgvector`
+# pgvector configurations, only available when VECTOR_STORE is `pgvector`
 PGVECTOR_HOST=pgvector
 PGVECTOR_PORT=5432
 PGVECTOR_USER=postgres
 PGVECTOR_PASSWORD=difyai123456
 PGVECTOR_DATABASE=dify
 
+# pgvecto-rs configurations, only available when VECTOR_STORE is `pgvecto-rs`
+PGVECTO_RS_HOST=pgvecto-rs
+PGVECTO_RS_PORT=5432
+PGVECTO_RS_USER=postgres
+PGVECTO_RS_PASSWORD=difyai123456
+PGVECTO_RS_DATABASE=dify
+
+# analyticdb configurations, only available when VECTOR_STORE is `analyticdb`
+ANALYTICDB_KEY_ID=your-ak
+ANALYTICDB_KEY_SECRET=your-sk
+ANALYTICDB_REGION_ID=cn-hangzhou
+ANALYTICDB_INSTANCE_ID=gp-ab123456
+ANALYTICDB_ACCOUNT=testaccount
+ANALYTICDB_PASSWORD=testpassword
+ANALYTICDB_NAMESPACE=dify
+ANALYTICDB_NAMESPACE_PASSWORD=difypassword
+
 # TiDB vector configurations, only available when VECTOR_STORE is `tidb`
 TIDB_VECTOR_HOST=tidb
 TIDB_VECTOR_PORT=4000

From d6b9587a979573fc22b380a13b80770883fc08e0 Mon Sep 17 00:00:00 2001
From: =?UTF-8?q?=E9=9D=9E=E6=B3=95=E6=93=8D=E4=BD=9C?= <hjlarry@163.com>
Date: Fri, 27 Sep 2024 11:13:40 +0800
Subject: [PATCH 088/235] fix: close log status option raise error (#8826)

---
 web/app/components/app/log/filter.tsx          | 2 ++
 web/app/components/app/workflow-log/filter.tsx | 2 ++
 2 files changed, 4 insertions(+)

diff --git a/web/app/components/app/log/filter.tsx b/web/app/components/app/log/filter.tsx
index 0552b44d16b3c..b8d7ca5a36b78 100644
--- a/web/app/components/app/log/filter.tsx
+++ b/web/app/components/app/log/filter.tsx
@@ -55,6 +55,8 @@ const Filter: FC<IFilterProps> = ({ isChatMode, appId, queryParams, setQueryPara
           className='!w-[300px]'
           onSelect={
             (item) => {
+              if (!item.value)
+                return
               setQueryParams({ ...queryParams, annotation_status: item.value as string })
             }
           }
diff --git a/web/app/components/app/workflow-log/filter.tsx b/web/app/components/app/workflow-log/filter.tsx
index d239c39d2cf61..58b5252c07446 100644
--- a/web/app/components/app/workflow-log/filter.tsx
+++ b/web/app/components/app/workflow-log/filter.tsx
@@ -23,6 +23,8 @@ const Filter: FC<IFilterProps> = ({ queryParams, setQueryParams }: IFilterProps)
           className='!min-w-[100px]'
           onSelect={
             (item) => {
+              if (!item.value)
+                return
               setQueryParams({ ...queryParams, status: item.value as string })
             }
           }

From 4c1063e1c589d239df7287350464c4a7d729cab2 Mon Sep 17 00:00:00 2001
From: 8bitpd <51897400+lpdink@users.noreply.github.com>
Date: Fri, 27 Sep 2024 12:05:21 +0800
Subject: [PATCH 089/235] fix: AnalyticdbVector retrieval scores (#8803)

---
 .../vdb/analyticdb/analyticdb_vector.py       | 19 ++++++-------------
 1 file changed, 6 insertions(+), 13 deletions(-)

diff --git a/api/core/rag/datasource/vdb/analyticdb/analyticdb_vector.py b/api/core/rag/datasource/vdb/analyticdb/analyticdb_vector.py
index 612542dab1df1..6dcd98dcfd14f 100644
--- a/api/core/rag/datasource/vdb/analyticdb/analyticdb_vector.py
+++ b/api/core/rag/datasource/vdb/analyticdb/analyticdb_vector.py
@@ -40,19 +40,8 @@ def to_analyticdb_client_params(self):
 
 
 class AnalyticdbVector(BaseVector):
-    _instance = None
-    _init = False
-
-    def __new__(cls, *args, **kwargs):
-        if cls._instance is None:
-            cls._instance = super().__new__(cls)
-        return cls._instance
-
     def __init__(self, collection_name: str, config: AnalyticdbConfig):
-        # collection_name must be updated every time
         self._collection_name = collection_name.lower()
-        if AnalyticdbVector._init:
-            return
         try:
             from alibabacloud_gpdb20160503.client import Client
             from alibabacloud_tea_openapi import models as open_api_models
@@ -62,7 +51,6 @@ def __init__(self, collection_name: str, config: AnalyticdbConfig):
         self._client_config = open_api_models.Config(user_agent="dify", **config.to_analyticdb_client_params())
         self._client = Client(self._client_config)
         self._initialize()
-        AnalyticdbVector._init = True
 
     def _initialize(self) -> None:
         cache_key = f"vector_indexing_{self.config.instance_id}"
@@ -257,11 +245,14 @@ def search_by_vector(self, query_vector: list[float], **kwargs: Any) -> list[Doc
         documents = []
         for match in response.body.matches.match:
             if match.score > score_threshold:
+                metadata = json.loads(match.metadata.get("metadata_"))
+                metadata["score"] = match.score
                 doc = Document(
                     page_content=match.metadata.get("page_content"),
-                    metadata=json.loads(match.metadata.get("metadata_")),
+                    metadata=metadata,
                 )
                 documents.append(doc)
+        documents = sorted(documents, key=lambda x: x.metadata["score"], reverse=True)
         return documents
 
     def search_by_full_text(self, query: str, **kwargs: Any) -> list[Document]:
@@ -286,12 +277,14 @@ def search_by_full_text(self, query: str, **kwargs: Any) -> list[Document]:
         for match in response.body.matches.match:
             if match.score > score_threshold:
                 metadata = json.loads(match.metadata.get("metadata_"))
+                metadata["score"] = match.score
                 doc = Document(
                     page_content=match.metadata.get("page_content"),
                     vector=match.metadata.get("vector"),
                     metadata=metadata,
                 )
                 documents.append(doc)
+        documents = sorted(documents, key=lambda x: x.metadata["score"], reverse=True)
         return documents
 
     def delete(self) -> None:

From 29275c7447c4afbe2a92fe23e31fbebf79b1541e Mon Sep 17 00:00:00 2001
From: zhuhao <37029601+hwzhuhao@users.noreply.github.com>
Date: Fri, 27 Sep 2024 12:11:56 +0800
Subject: [PATCH 090/235] feat: deprecate mistral model for siliconflow (#8828)

---
 .../siliconflow/llm/mistral-7b-instruct-v0.2.yaml                | 1 +
 .../siliconflow/llm/mistral-8x7b-instruct-v0.1.yaml              | 1 +
 2 files changed, 2 insertions(+)

diff --git a/api/core/model_runtime/model_providers/siliconflow/llm/mistral-7b-instruct-v0.2.yaml b/api/core/model_runtime/model_providers/siliconflow/llm/mistral-7b-instruct-v0.2.yaml
index 27664eab6c817..89fb153ba09ac 100644
--- a/api/core/model_runtime/model_providers/siliconflow/llm/mistral-7b-instruct-v0.2.yaml
+++ b/api/core/model_runtime/model_providers/siliconflow/llm/mistral-7b-instruct-v0.2.yaml
@@ -28,3 +28,4 @@ pricing:
   output: '0'
   unit: '0.000001'
   currency: RMB
+deprecated: true
diff --git a/api/core/model_runtime/model_providers/siliconflow/llm/mistral-8x7b-instruct-v0.1.yaml b/api/core/model_runtime/model_providers/siliconflow/llm/mistral-8x7b-instruct-v0.1.yaml
index fd7aada42848a..2785e7496fb06 100644
--- a/api/core/model_runtime/model_providers/siliconflow/llm/mistral-8x7b-instruct-v0.1.yaml
+++ b/api/core/model_runtime/model_providers/siliconflow/llm/mistral-8x7b-instruct-v0.1.yaml
@@ -28,3 +28,4 @@ pricing:
   output: '1.26'
   unit: '0.000001'
   currency: RMB
+deprecated: true

From bb781764b829b83471a0546cbced7c842e826231 Mon Sep 17 00:00:00 2001
From: HowardChan <lonely829@gmail.com>
Date: Fri, 27 Sep 2024 12:13:00 +0800
Subject: [PATCH 091/235] Add Llama3.2 models in Groq provider (#8831)

---
 .../groq/llm/llama-3.2-11b-text-preview.yaml  | 25 +++++++++++++++++++
 .../groq/llm/llama-3.2-1b-preview.yaml        | 25 +++++++++++++++++++
 .../groq/llm/llama-3.2-3b-preview.yaml        | 25 +++++++++++++++++++
 .../groq/llm/llama-3.2-90b-text-preview.yaml  | 25 +++++++++++++++++++
 4 files changed, 100 insertions(+)
 create mode 100644 api/core/model_runtime/model_providers/groq/llm/llama-3.2-11b-text-preview.yaml
 create mode 100644 api/core/model_runtime/model_providers/groq/llm/llama-3.2-1b-preview.yaml
 create mode 100644 api/core/model_runtime/model_providers/groq/llm/llama-3.2-3b-preview.yaml
 create mode 100644 api/core/model_runtime/model_providers/groq/llm/llama-3.2-90b-text-preview.yaml

diff --git a/api/core/model_runtime/model_providers/groq/llm/llama-3.2-11b-text-preview.yaml b/api/core/model_runtime/model_providers/groq/llm/llama-3.2-11b-text-preview.yaml
new file mode 100644
index 0000000000000..019d45372361d
--- /dev/null
+++ b/api/core/model_runtime/model_providers/groq/llm/llama-3.2-11b-text-preview.yaml
@@ -0,0 +1,25 @@
+model: llama-3.2-11b-text-preview
+label:
+  zh_Hans: Llama 3.2 11B Text (Preview)
+  en_US: Llama 3.2 11B Text (Preview)
+model_type: llm
+features:
+  - agent-thought
+model_properties:
+  mode: chat
+  context_size: 131072
+parameter_rules:
+  - name: temperature
+    use_template: temperature
+  - name: top_p
+    use_template: top_p
+  - name: max_tokens
+    use_template: max_tokens
+    default: 512
+    min: 1
+    max: 8192
+pricing:
+  input: '0.05'
+  output: '0.1'
+  unit: '0.000001'
+  currency: USD
diff --git a/api/core/model_runtime/model_providers/groq/llm/llama-3.2-1b-preview.yaml b/api/core/model_runtime/model_providers/groq/llm/llama-3.2-1b-preview.yaml
new file mode 100644
index 0000000000000..a44e4ff508eb8
--- /dev/null
+++ b/api/core/model_runtime/model_providers/groq/llm/llama-3.2-1b-preview.yaml
@@ -0,0 +1,25 @@
+model: llama-3.2-1b-preview
+label:
+  zh_Hans: Llama 3.2 1B Text (Preview)
+  en_US: Llama 3.2 1B Text (Preview)
+model_type: llm
+features:
+  - agent-thought
+model_properties:
+  mode: chat
+  context_size: 131072
+parameter_rules:
+  - name: temperature
+    use_template: temperature
+  - name: top_p
+    use_template: top_p
+  - name: max_tokens
+    use_template: max_tokens
+    default: 512
+    min: 1
+    max: 8192
+pricing:
+  input: '0.05'
+  output: '0.1'
+  unit: '0.000001'
+  currency: USD
diff --git a/api/core/model_runtime/model_providers/groq/llm/llama-3.2-3b-preview.yaml b/api/core/model_runtime/model_providers/groq/llm/llama-3.2-3b-preview.yaml
new file mode 100644
index 0000000000000..f2fdd0a05e027
--- /dev/null
+++ b/api/core/model_runtime/model_providers/groq/llm/llama-3.2-3b-preview.yaml
@@ -0,0 +1,25 @@
+model: llama-3.2-3b-preview
+label:
+  zh_Hans: Llama 3.2 3B Text (Preview)
+  en_US: Llama 3.2 3B Text (Preview)
+model_type: llm
+features:
+  - agent-thought
+model_properties:
+  mode: chat
+  context_size: 131072
+parameter_rules:
+  - name: temperature
+    use_template: temperature
+  - name: top_p
+    use_template: top_p
+  - name: max_tokens
+    use_template: max_tokens
+    default: 512
+    min: 1
+    max: 8192
+pricing:
+  input: '0.05'
+  output: '0.1'
+  unit: '0.000001'
+  currency: USD
diff --git a/api/core/model_runtime/model_providers/groq/llm/llama-3.2-90b-text-preview.yaml b/api/core/model_runtime/model_providers/groq/llm/llama-3.2-90b-text-preview.yaml
new file mode 100644
index 0000000000000..3b34e7c07996b
--- /dev/null
+++ b/api/core/model_runtime/model_providers/groq/llm/llama-3.2-90b-text-preview.yaml
@@ -0,0 +1,25 @@
+model: llama-3.2-90b-text-preview
+label:
+  zh_Hans: Llama 3.2 90B Text (Preview)
+  en_US: Llama 3.2 90B Text (Preview)
+model_type: llm
+features:
+  - agent-thought
+model_properties:
+  mode: chat
+  context_size: 131072
+parameter_rules:
+  - name: temperature
+    use_template: temperature
+  - name: top_p
+    use_template: top_p
+  - name: max_tokens
+    use_template: max_tokens
+    default: 512
+    min: 1
+    max: 8192
+pricing:
+  input: '0.05'
+  output: '0.1'
+  unit: '0.000001'
+  currency: USD

From 0603359e2dea807f8cac2cfe5ac68aeb081627e4 Mon Sep 17 00:00:00 2001
From: CXwudi <cx2010017@gmail.com>
Date: Fri, 27 Sep 2024 01:49:03 -0400
Subject: [PATCH 092/235] fix: delete harm catalog settings for gemini (#8829)

---
 .../model_providers/google/llm/llm.py              | 14 +++-----------
 1 file changed, 3 insertions(+), 11 deletions(-)

diff --git a/api/core/model_runtime/model_providers/google/llm/llm.py b/api/core/model_runtime/model_providers/google/llm/llm.py
index 3fc6787a444e4..e686ad08d9d35 100644
--- a/api/core/model_runtime/model_providers/google/llm/llm.py
+++ b/api/core/model_runtime/model_providers/google/llm/llm.py
@@ -9,8 +9,8 @@
 import google.generativeai as genai
 import requests
 from google.api_core import exceptions
-from google.generativeai import client
-from google.generativeai.types import ContentType, GenerateContentResponse, HarmBlockThreshold, HarmCategory
+from google.generativeai.client import _ClientManager
+from google.generativeai.types import ContentType, GenerateContentResponse
 from google.generativeai.types.content_types import to_part
 from PIL import Image
 
@@ -200,24 +200,16 @@ def _generate(
                     history.append(content)
 
         # Create a new ClientManager with tenant's API key
-        new_client_manager = client._ClientManager()
+        new_client_manager = _ClientManager()
         new_client_manager.configure(api_key=credentials["google_api_key"])
         new_custom_client = new_client_manager.make_client("generative")
 
         google_model._client = new_custom_client
 
-        safety_settings = {
-            HarmCategory.HARM_CATEGORY_HARASSMENT: HarmBlockThreshold.BLOCK_NONE,
-            HarmCategory.HARM_CATEGORY_HATE_SPEECH: HarmBlockThreshold.BLOCK_NONE,
-            HarmCategory.HARM_CATEGORY_SEXUALLY_EXPLICIT: HarmBlockThreshold.BLOCK_NONE,
-            HarmCategory.HARM_CATEGORY_DANGEROUS_CONTENT: HarmBlockThreshold.BLOCK_NONE,
-        }
-
         response = google_model.generate_content(
             contents=history,
             generation_config=genai.types.GenerationConfig(**config_kwargs),
             stream=stream,
-            safety_settings=safety_settings,
             tools=self._convert_tools_to_glm_tool(tools) if tools else None,
             request_options={"timeout": 600},
         )

From c828a5dfdf6c4a6caf7230ed879ae688608eb373 Mon Sep 17 00:00:00 2001
From: =?UTF-8?q?=E8=B5=B0=E5=9C=A8=E4=BF=AE=E8=A1=8C=E7=9A=84=E5=A4=A7?=
 =?UTF-8?q?=E8=A1=97=E4=B8=8A?= <hgnulb@163.com>
Date: Fri, 27 Sep 2024 17:31:45 +0800
Subject: [PATCH 093/235] feat(Tools): add feishu tools (#8800)
MIME-Version: 1.0
Content-Type: text/plain; charset=UTF-8
Content-Transfer-Encoding: 8bit

Co-authored-by: 黎斌 <libin.23@bytedance.com>
---
 api/core/tools/provider/_position.yaml        |   4 +
 .../builtin/feishu_calendar/_assets/icon.png  | Bin 0 -> 5488 bytes
 .../feishu_calendar/feishu_calendar.py        |   7 +
 .../feishu_calendar/feishu_calendar.yaml      |  36 ++
 .../tools/add_event_attendees.py              |  20 +
 .../tools/add_event_attendees.yaml            |  54 +++
 .../feishu_calendar/tools/create_event.py     |  26 ++
 .../feishu_calendar/tools/create_event.yaml   | 119 ++++++
 .../feishu_calendar/tools/delete_event.py     |  19 +
 .../feishu_calendar/tools/delete_event.yaml   |  38 ++
 .../tools/get_primary_calendar.py             |  18 +
 .../tools/get_primary_calendar.yaml           |  37 ++
 .../feishu_calendar/tools/list_events.py      |  21 +
 .../feishu_calendar/tools/list_events.yaml    |  62 +++
 .../feishu_calendar/tools/search_events.py    |  23 +
 .../feishu_calendar/tools/search_events.yaml  | 100 +++++
 .../feishu_calendar/tools/update_event.py     |  24 ++
 .../feishu_calendar/tools/update_event.yaml   | 100 +++++
 .../feishu_document/feishu_document.py        |  12 +-
 .../feishu_document/feishu_document.yaml      |   8 +-
 .../tools/create_document.yaml                |   9 +-
 .../tools/get_document_content.py             |   4 +-
 .../tools/get_document_content.yaml           |  35 +-
 .../tools/list_document_blocks.py             |   5 +-
 .../tools/list_document_blocks.yaml           |   4 +-
 .../feishu_document/tools/write_document.py   |   2 +-
 .../feishu_document/tools/write_document.yaml |  30 +-
 .../builtin/feishu_message/feishu_message.py  |  12 +-
 .../feishu_message/feishu_message.yaml        |   8 +-
 .../feishu_message/tools/get_chat_messages.py |  23 +
 .../tools/get_chat_messages.yaml              |  96 +++++
 .../tools/get_thread_messages.py              |  21 +
 .../tools/get_thread_messages.yaml            |  72 ++++
 .../tools/send_bot_message.yaml               | 114 +++--
 .../tools/send_webhook_message.yaml           |  46 +-
 .../feishu_spreadsheet/_assets/icon.png       | Bin 0 -> 5417 bytes
 .../feishu_spreadsheet/feishu_spreadsheet.py  |   7 +
 .../feishu_spreadsheet.yaml                   |  36 ++
 .../feishu_spreadsheet/tools/add_cols.py      |  22 +
 .../feishu_spreadsheet/tools/add_cols.yaml    |  72 ++++
 .../feishu_spreadsheet/tools/add_rows.py      |  22 +
 .../feishu_spreadsheet/tools/add_rows.yaml    |  72 ++++
 .../tools/create_spreadsheet.py               |  19 +
 .../tools/create_spreadsheet.yaml             |  35 ++
 .../tools/get_spreadsheet.py                  |  19 +
 .../tools/get_spreadsheet.yaml                |  49 +++
 .../tools/list_spreadsheet_sheets.py          |  18 +
 .../tools/list_spreadsheet_sheets.yaml        |  23 +
 .../feishu_spreadsheet/tools/read_cols.py     |  23 +
 .../feishu_spreadsheet/tools/read_cols.yaml   |  97 +++++
 .../feishu_spreadsheet/tools/read_rows.py     |  23 +
 .../feishu_spreadsheet/tools/read_rows.yaml   |  97 +++++
 .../feishu_spreadsheet/tools/read_table.py    |  23 +
 .../feishu_spreadsheet/tools/read_table.yaml  | 122 ++++++
 .../builtin/feishu_task/_assets/icon.png      | Bin 0 -> 3917 bytes
 .../builtin/feishu_task/feishu_task.py        |   7 +
 .../builtin/feishu_task/feishu_task.yaml      |  36 ++
 .../builtin/feishu_task/tools/add_members.py  |  20 +
 .../feishu_task/tools/add_members.yaml        |  58 +++
 .../builtin/feishu_task/tools/create_task.py  |  22 +
 .../feishu_task/tools/create_task.yaml        |  74 ++++
 .../builtin/feishu_task/tools/delete_task.py  |  18 +
 .../feishu_task/tools/delete_task.yaml        |  24 ++
 .../builtin/feishu_task/tools/update_task.py  |  23 +
 .../feishu_task/tools/update_task.yaml        |  89 ++++
 .../builtin/feishu_wiki/_assets/icon.png      | Bin 0 -> 4252 bytes
 .../builtin/feishu_wiki/feishu_wiki.py        |   7 +
 .../builtin/feishu_wiki/feishu_wiki.yaml      |  36 ++
 .../feishu_wiki/tools/get_wiki_nodes.py       |  21 +
 .../feishu_wiki/tools/get_wiki_nodes.yaml     |  63 +++
 api/core/tools/utils/feishu_api_utils.py      | 401 +++++++++++++++++-
 71 files changed, 2755 insertions(+), 132 deletions(-)
 create mode 100644 api/core/tools/provider/builtin/feishu_calendar/_assets/icon.png
 create mode 100644 api/core/tools/provider/builtin/feishu_calendar/feishu_calendar.py
 create mode 100644 api/core/tools/provider/builtin/feishu_calendar/feishu_calendar.yaml
 create mode 100644 api/core/tools/provider/builtin/feishu_calendar/tools/add_event_attendees.py
 create mode 100644 api/core/tools/provider/builtin/feishu_calendar/tools/add_event_attendees.yaml
 create mode 100644 api/core/tools/provider/builtin/feishu_calendar/tools/create_event.py
 create mode 100644 api/core/tools/provider/builtin/feishu_calendar/tools/create_event.yaml
 create mode 100644 api/core/tools/provider/builtin/feishu_calendar/tools/delete_event.py
 create mode 100644 api/core/tools/provider/builtin/feishu_calendar/tools/delete_event.yaml
 create mode 100644 api/core/tools/provider/builtin/feishu_calendar/tools/get_primary_calendar.py
 create mode 100644 api/core/tools/provider/builtin/feishu_calendar/tools/get_primary_calendar.yaml
 create mode 100644 api/core/tools/provider/builtin/feishu_calendar/tools/list_events.py
 create mode 100644 api/core/tools/provider/builtin/feishu_calendar/tools/list_events.yaml
 create mode 100644 api/core/tools/provider/builtin/feishu_calendar/tools/search_events.py
 create mode 100644 api/core/tools/provider/builtin/feishu_calendar/tools/search_events.yaml
 create mode 100644 api/core/tools/provider/builtin/feishu_calendar/tools/update_event.py
 create mode 100644 api/core/tools/provider/builtin/feishu_calendar/tools/update_event.yaml
 create mode 100644 api/core/tools/provider/builtin/feishu_message/tools/get_chat_messages.py
 create mode 100644 api/core/tools/provider/builtin/feishu_message/tools/get_chat_messages.yaml
 create mode 100644 api/core/tools/provider/builtin/feishu_message/tools/get_thread_messages.py
 create mode 100644 api/core/tools/provider/builtin/feishu_message/tools/get_thread_messages.yaml
 create mode 100644 api/core/tools/provider/builtin/feishu_spreadsheet/_assets/icon.png
 create mode 100644 api/core/tools/provider/builtin/feishu_spreadsheet/feishu_spreadsheet.py
 create mode 100644 api/core/tools/provider/builtin/feishu_spreadsheet/feishu_spreadsheet.yaml
 create mode 100644 api/core/tools/provider/builtin/feishu_spreadsheet/tools/add_cols.py
 create mode 100644 api/core/tools/provider/builtin/feishu_spreadsheet/tools/add_cols.yaml
 create mode 100644 api/core/tools/provider/builtin/feishu_spreadsheet/tools/add_rows.py
 create mode 100644 api/core/tools/provider/builtin/feishu_spreadsheet/tools/add_rows.yaml
 create mode 100644 api/core/tools/provider/builtin/feishu_spreadsheet/tools/create_spreadsheet.py
 create mode 100644 api/core/tools/provider/builtin/feishu_spreadsheet/tools/create_spreadsheet.yaml
 create mode 100644 api/core/tools/provider/builtin/feishu_spreadsheet/tools/get_spreadsheet.py
 create mode 100644 api/core/tools/provider/builtin/feishu_spreadsheet/tools/get_spreadsheet.yaml
 create mode 100644 api/core/tools/provider/builtin/feishu_spreadsheet/tools/list_spreadsheet_sheets.py
 create mode 100644 api/core/tools/provider/builtin/feishu_spreadsheet/tools/list_spreadsheet_sheets.yaml
 create mode 100644 api/core/tools/provider/builtin/feishu_spreadsheet/tools/read_cols.py
 create mode 100644 api/core/tools/provider/builtin/feishu_spreadsheet/tools/read_cols.yaml
 create mode 100644 api/core/tools/provider/builtin/feishu_spreadsheet/tools/read_rows.py
 create mode 100644 api/core/tools/provider/builtin/feishu_spreadsheet/tools/read_rows.yaml
 create mode 100644 api/core/tools/provider/builtin/feishu_spreadsheet/tools/read_table.py
 create mode 100644 api/core/tools/provider/builtin/feishu_spreadsheet/tools/read_table.yaml
 create mode 100644 api/core/tools/provider/builtin/feishu_task/_assets/icon.png
 create mode 100644 api/core/tools/provider/builtin/feishu_task/feishu_task.py
 create mode 100644 api/core/tools/provider/builtin/feishu_task/feishu_task.yaml
 create mode 100644 api/core/tools/provider/builtin/feishu_task/tools/add_members.py
 create mode 100644 api/core/tools/provider/builtin/feishu_task/tools/add_members.yaml
 create mode 100644 api/core/tools/provider/builtin/feishu_task/tools/create_task.py
 create mode 100644 api/core/tools/provider/builtin/feishu_task/tools/create_task.yaml
 create mode 100644 api/core/tools/provider/builtin/feishu_task/tools/delete_task.py
 create mode 100644 api/core/tools/provider/builtin/feishu_task/tools/delete_task.yaml
 create mode 100644 api/core/tools/provider/builtin/feishu_task/tools/update_task.py
 create mode 100644 api/core/tools/provider/builtin/feishu_task/tools/update_task.yaml
 create mode 100644 api/core/tools/provider/builtin/feishu_wiki/_assets/icon.png
 create mode 100644 api/core/tools/provider/builtin/feishu_wiki/feishu_wiki.py
 create mode 100644 api/core/tools/provider/builtin/feishu_wiki/feishu_wiki.yaml
 create mode 100644 api/core/tools/provider/builtin/feishu_wiki/tools/get_wiki_nodes.py
 create mode 100644 api/core/tools/provider/builtin/feishu_wiki/tools/get_wiki_nodes.yaml

diff --git a/api/core/tools/provider/_position.yaml b/api/core/tools/provider/_position.yaml
index 40c3356116770..e0a8e7511e59b 100644
--- a/api/core/tools/provider/_position.yaml
+++ b/api/core/tools/provider/_position.yaml
@@ -34,5 +34,9 @@
 - feishu_base
 - feishu_document
 - feishu_message
+- feishu_wiki
+- feishu_task
+- feishu_calendar
+- feishu_spreadsheet
 - slack
 - tianditu
diff --git a/api/core/tools/provider/builtin/feishu_calendar/_assets/icon.png b/api/core/tools/provider/builtin/feishu_calendar/_assets/icon.png
new file mode 100644
index 0000000000000000000000000000000000000000..2a934747a98c6680065941bcd31d2400da1eaf23
GIT binary patch
literal 5488
zcmd^@S6CClyT+3cB}hx?J%AJ?fE0mHMQMV8&_R?YO?nelY7hlNk>0Cxkg60F3{fc}
zH9!DCT0jJ80tr>&<bSTu?KwB+Vt1Z>_S<*on|*iYn;+KLNRROn&m{l=z^H#)+mzb=
z`Og8lNIh#<tfx^MV4$hqEkMIC-#P%mE}^e|)BK6uR-s`ETVH0Mz2)PP*iM*g8O5=u
z=gVzzzUD!>4qnG3;S|irW=|=<j2DR<b4O5KP?)3Ik%Q1@$SCYC=t~jV@og5Go&l+;
zQFk;0Sx(5&rKN{k<|!i`a<&`VhktVq$&Aa}S{%RWkl(k5UE;6YBZhWmN7`8iyHpC_
z7X*NYS&WB6nEGx4zz~qGH~<(+r-^40XKsiFfD<(UASPA}hTllwJ}nT!6%GLNi2;DB
z1#0h@Kv0<z00fBwfP#@v{rJIx4rCzIF9!hL^xIohfoS5_XapB!fRO2>l}|xX42D3f
ziFX7+o4)i%ZwLZ_Wy~1N7x0X<1J061*y29gfDyJ2fUTeVMQPHq7@KLU>F~{ov1emn
zTTR^lX{pNZoq%B_m`)+II}6z;t*n3|Mo<iR8#eBehwYo7A3tR9cJY}v1@1o?0R+81
z#7^wzs_W1|*B*gTRrjP;vLHEqrzh1NawpkKee*^mz`ry=Ef6Zccj8bWk$Xm;_~Qd+
z!$Rpq6vkMw2T_W`z`@_hSLPi;t-USqYIZ!O+oWq)HnGaTYZ-CEyis!gE(t?eX-|+*
z_pAj^pNK_U;<C>U1jS;dIL*?`4jeJEOe)(8nc{{SYk($W#)H9s-F>-9L)q>vds%qH
zkcrltA7vCoxnI4*96v@UvWzE-7Cx7P{&@x%&*6D0Sj%Vv92rJtf5apSYHAh%#><`<
z2e{f;;;o92hA{8!WL}P&cvjHhvd7BO#hQoR;KbjEOqS{00syPkr2de>038=`YCj_O
zPd5aDqo-cUk32YjA+$q>s&#{_U4>P|VUB0oV{Y#n(#1N^fI)(I^SU~-b7vfzi7wtM
z^w6|?o(Z6K3k;+Kp}J2e-ZCfW^Q6Q6wEFUk_k^MfZ-zo4F__Xr;-SE%Op=u@zWCtx
zh*)#))OFs6a3<iYeNAfYxhqdoBBoK@Kb<X$K2MBhD_YptazumcdXlv+J|{|Xh-M~M
zk7}-Kzv8AZQ(d9guYF|+MzzF%vhYD1)kG$8FoP#x!NorTFXPmO0l?4<)N>XAAOzW2
zaQQ(p)m{RktgoK`-$<zF;pO`bTxxeUpD1f~3#O!N4*qPQX&s)<+#`ItT};OS{52?=
zci}ds=0~_l42A}!e(NBLnu8VxOsk2SgM(QzKAh_L(-fU<s3bLq>50pJaB2=3&M;Z<
z9qKSBQXoHQsl&j0SDWDz3kDNys17>(|0S{QY(fc-ee#?BF<;wCO^An-3EEzIBkWv}
zKsIk^<Npo;Z8x=$$tEqUr{$_OAps?4=Vwu(Mtdvjr2cDZ!(8b5VrP4b{>KKf7o0v7
zo)OJwp1u-Fzm8t+m$NYP78%<NdndTa^tF^@>(2mN{nOOaPn7rk52RKXip9_VJ&iy2
z(`YX=^0A&q8oV%!9lT#qgwn1%Cj_27!_4M<F4B-l+<dk6ey%)#i7tEJ&T4Ju#TT{(
z<ajUO1>D;@oHK;g{)Qou*`7VPiE&kMQBhO8^r17_RKNU12a>1wxuj`0t0s!bY)4s!
zD?#4%fVSZ!q_)Gvm~T(j8f<T9wI(=i<4)-;?aw}-GRsC*K21OM7F^_5btc+Kd?fpC
z({Y)hgA^KyW=Sg@@Y)@@@39y|m(+&v79Umiv!6K>p+Ga=%Bt#<jqsO@aQ4iDNa;Q1
zVk=_%nWFKw1`D&|P-r@DMh$T8jPR)%rF|z-K2>I3m-oc&6^U8Hr;PRv-1eVCti15c
z82eYGwqp7}`)hNLUtR=Dj$X5rBizuen46llf69Agt=t;*q;xz|RhMGh9D57TaiOlK
zW54w4=tz9_xY_l<09!q@*Q*o7el@U{4iA>_x$Ur8rgLtd1}wROt^d})#PqKdN^EzJ
z7h0^03z&m4`l2JG1D0xSRrOoZb~{KblW?<gFvtjabwf%$oN_pSaY4pSfmnNr*s0rS
zD$X_zH~T37GJ0`!+%TZ)u4_mgVQ9CTy|!q^!CC8#Y^|?B9@NjknA-=j)wN+xDWtCt
zbgiIDK30Mt6>80uK9ILNkR>xjR`b&#oVou@Cmm6V#ukF4sbg!;3bqq=X%cB7Tqeui
z9NUQemDNQ#3o>0InrU5C`{tK$c6a^jxOY*iE@dc1HVm6dXZbe&l(yRQe9rZ8#aZZ^
zw6z&J(neQ_c6Wlb6zOxXR{3k(gvGzbShj_u7a8vdgN?)Gg|#jWL%d$hipGw#^159)
zqu^pbdyA};j?!M<rO2EA^NZ!nKxI(_m&TN|@VTi{vn5AkdTil`I!@6xovqAou(z$2
zOY!3Q{jvPM>fU@>0yNb)<?5Sw<r1lJ?FR?SclRC#AT55{Zt6NZE3qqfdGn19PA{}9
zsdEvPtVn{3_ouTIf3KG{5E>JcavNAAk|JC>{Yv@aK_w@vn<`L0BfkXk49mu<G0*J8
zM12N(`ut?kQ?K0n^~GOrQHBbyL6Axx`gm#;7OK*g^6as6B6K%AID4yucbgb;*et1E
zN7H7T12PSme^yEO>LbB}i?kf}5yCmO@z)g%j_r`dtUNCy{`O2m4Bp};<onh|CQEto
zD7+Ed-Cobbn1m;NV2IF_>pnCLqyyd=Jo}x_XVBrx^@8}}67`{sD|1|051pQ4fgjz{
z37jS;^1aQzysFD%sxq~W{vU)KfX}hm6@2}C7_E=!#mXw}$z_AVTN^$9LQ=ct=!sU(
zuSqmPdWU6%TT(R<S?r}Jmx8z<)(~gfoYa}ynk3sW9X<n-b;iwo$uSo}2Uy~k(i@Ms
z55cxd2U;YjG*=_-^FNo^+YPq%>kiv}7Jk>bHrt%5g1z{g|ItE<ctv()^<UTes>vD=
z&CY);`g;{Wtx^Mmwlh#Ghh;4r7i%m$_j#>HO{r3ue5tQ-+PnW|wV65V(kEG?>=#a5
zZ^@jEZ+%y9ROh%3mK@ueb=Cg;(Tdg~{7CX>JD=m-e2)k&u!ziJnD<xscxmD_Jt$bI
zknq(`!sjg48qmi4(K32OT->5%qZZyAg8W|eHJ=)BJ@en|8qf1zrNdV9JR(N58HEGA
z5spvQ>TQu~9oAo>J*ex}I>_7SGSXdv6HEDxX!>EArh?9@3GhfiC31nXHaY5>MU~w%
zJ5P<!laGOPB`4B@DIETkABv=FzO7>ZibWvcY>CH6<w`lRAbavgA=`YtS;jV5ggW7}
z9Vc2$G`92#F1D1<L>80FGhbf8?IUW7<~$VM8QrKLtS?;$d$sO-<8GI6&twn^qBJ8P
zRT?$m=--sZ5<akO=!hEGLYrvT-^jQV;}7kG6+(6Q2ADL(Jmyp$-sFQe38L;)mW$W7
z<WDYyQnW%&Yg*0&mu=;w9K9W~+;w-b-htY}{0v}8L-Wa^O#RmhPUVsGt>OJr+$~${
zo&TPY0`_c{1}^{~`_y`&Y(#fLb{N_PA7+LHiH48nhxfhLK&boooRx_b0f2XmTO7U+
z^<OKqp&Ke2l$^WR-fF&orT=g15oI&WQhj(+8QR2s15q!9vRO|)qU7pXx)pHZ<8%*<
z{r6=NPwIFJi^_8^kGzw;%m}=<!;}kO+$<+5rAkPwM^BP993OCs(__i;SM&006H7*J
zZ^K7b)3E9lK5;b%{Wy+Ida*@Cuq41_^!Y%#<$?oLKx6-le*6vfo6Q>HPYDdWf4Tcl
zsw9+!RONFo0O@N-Ml)v9E%!g&DM}xV?9?Hi{}_9;)_7uN=Km`){jBTj&u=n3AP)WZ
z_pmAPdJbrcinuatr-OGK-a0dmu9%We3Gw)<1`iotEL<8s*QJJjZLOhW7OUqFPC=3~
zx}>Li@^{PPLqu^hr#@b3T$Lv#jr=%A0kBt^b8g!<+GZ#)1v#UG7S4^+J3pvXVwU;n
zVU=vQrF8D-(d!cmnU-E~ovcUM*s2<}qG!f;MP3FIi0uL7T-P<rqmDs6kv1h|8HgY5
zDp>NFY&OiE`r?be5iJRKDTVtwq<2%6lC6V8e-6cGm?YS${34;8=(N+NdW}0m<Ldq=
zYy{^d{Z&-M32r<eg_`!I2hO^6uF$oK)vGv6n!~+4$iTiY`gzJ;5e8*vhRC~gnPF?G
zJI^rk0ilm%Bv?%%NZiF!4YT2SEjEKq8&wLI>jFw;&v{V8oR>uqO1c-?Anj4^3)C*f
zT#==b66e{+N-En+9>_wbg@5Q&h8MJipzxr3eZ7aiqN>h$_8<^0BCi&X^Vp6zq=z;c
zexUaXRdJ`}V5T~6SHQ!Z!Pmb#Y)hpVr?|o!ll_&+k-8mGi*DtKHRJ>S>1BO{J8&Zz
z>Dr6)#$})#Ze9R00L8&!MJ*v=kbi2KcJk_{w_UFm9|vowb@b1n<qk+33onIQY!@}q
zHd10$k#DYUb$K#OJh@sdpmE-@9;|`Ne0vYOQ=s&+ua>$i6R<<1Tbk~~DSqDm?*#!Y
znZI?X0(UB$X(t!Ti4K-o$1D)@KR?Ih3x9j81xU9{!mnNO+S~0e7;=`|Sh>Msh<06W
zY=1LH$!%s{U2t*>R6U%7`uRK>P->BL*D-(UX4}_@mhuTrCO?b7zShW?ES9TOyJ&t@
zqLVDkKKb|c?(*`8kr}0o>6}hqj}FNDD+CLqc1dbx;Cn{QiVvUu`NZsE(d^6$m$43c
z1;E~ZRzNM;DuqO9s|)6Ch5hORd)-w80tY~CvhF?h^FHerJ$H+JSi+yG!nv9c7mxck
zk~Yb@WdGxDV`i)&r`3Np1>%clUC8-6PM_k#lM9(0D=ndZV)qgAUQS~M!yV)-)CgoO
zVcpw%O<;p@iUM?Iu1c#9oRH`*yta{8z^&l~Da4s6eH>3UJId>QHDdTdMz=NLIN2`M
zIw8LojX0Ds`}uI=U_|jHkbRoV@j*$`E|gs{=^K~lTXr*`85*9+^dpmZtSu$hjL6Au
z-QcE)Pk#F-E!LCB35vn+!ekooGu87X3g(_(LT_=Q|8iB?3;i!p8{KXr%L12MoUo>d
zdamz=FNS;v0Bv5PJBn>ta~?hdAfMqYL;O&`%Lc2S0pm?dxY*oUxJnc4qcTk)hh?%8
zi|2*s{kj_+=dtGKE{2Ex2vqB9ah1sPUq4AP7?%*R#GgGr(S!J`{xMTT*y93;o#3P4
zbi@T_;J_nB$P<^B)MEH{WW}UD=lQOf#ll?1*=B`NK*TU7#99mi^$UC%1cS4vUu-!x
z<LyH3%?+5Vw-0&m{vH)v46}9cM?jGLeJ6wD<^ya_uE!K##4p=KZjjN`Zr<Dlpwi{*
z=Z|;ahMB47>BHIV)Y~(kklJf(kp8S-$#gLY@^T-w>Q-y-P|*7Qy_VPhFZNW+)qF9L
zU+U-9f2zC$pYe{z4}%GR2Hr-R95YC{S0J|8cHP{n4?5CwN4PlxKI6E}9k_dgfE#o?
zhu1zHOgbx=3V>|BhR{85bhdhfYKm}y?Qb^lqq<qMYYt-fyQfA@to%))_8&E_dZ&e#
zoOF%<)#cpI`TLA4d=hLM_~7MPuvCpjq+WT?+~HB`U9}D|>qf@a({}yhHT>bZ()T#Q
z<9k;)CH$`B>K52hH*p_Ph-a5xP?3p2C9Eo&3P`Sv$Cum^q7s%g_wIF3>Q6Y(B3x}u
z-3PvM=J5~F2Y?Rb>LC(rAn5<11lz*q+0{n*B&mUIrvVMm)e;^z(~*3Y=mSCMC7`Ro
zj`C+-)G|Bl&zESzdfdZ~8u^FJpjS|b%4Kyu(v;zU?nJ|NLg`UH1rToohOL|U0zhDr
z-HtL(F%@VCFXk#ejH4$Qnz@Y%fI&j|T8_pI6@&KwC^iL?2vHBy*eJOw&kqC_)q$%4
zhcJPRI5QM}oVIk?fFM9uMTHSzVkJ5AJ1Q(YDRCw!JeTgBavDV{ppfl}{^*<b6~^-b
zG`Z-;RDGvzf*vRW6nds94N4<gsR~Lg#?rbAmR1)a<gSB&=X7zF_WnT6XsUF3X6eVX
zUm(~sO)(H3YZ3*=o$TU4TyZ5_sw?gZt?{kAXDeKBW;bRs4Fa~){n0+m|J(A2fdral
z64P=w((bgGah#?B{({J4JN59Ei#2fc`s(0MfRPc=F$N}&uxu1nT0W*|3>$E@5h3gQ
z7ZH`L$n)OW5Wg>M2`^*0L%BHZ9}ft8&$7>PkJKuMZRL`Q88-@2Mk2H@nWTFPt5S+F
zLsHzguCWJ@m*dt_$Y?9MPR9%2ZivX+A@%WHx}bu}?mB6wufL%mOccc8b=|TB+Q{GA
zZhnp!1zYQl3})ZmWtz2Umtudiq(x~`fywQMn0h!I_(C6z^7u&~X}5H%u_J;?jT1)h
z)JcCcc(qkZcuVYIsjsrR4uv(klxqC5pVaib0cm-rseQuh8gxy%x4l-Ld@rBX+@0j!
zff+|)Lp1?l9x4YU$o~%vPtYMYop$Ob0Hh25UxJ56k@&OI*n~FT87e{r=<671H{5b~
F_Fv-s=P3XH

literal 0
HcmV?d00001

diff --git a/api/core/tools/provider/builtin/feishu_calendar/feishu_calendar.py b/api/core/tools/provider/builtin/feishu_calendar/feishu_calendar.py
new file mode 100644
index 0000000000000..a46a9fa9e80ca
--- /dev/null
+++ b/api/core/tools/provider/builtin/feishu_calendar/feishu_calendar.py
@@ -0,0 +1,7 @@
+from core.tools.provider.builtin_tool_provider import BuiltinToolProviderController
+from core.tools.utils.feishu_api_utils import auth
+
+
+class FeishuCalendarProvider(BuiltinToolProviderController):
+    def _validate_credentials(self, credentials: dict) -> None:
+        auth(credentials)
diff --git a/api/core/tools/provider/builtin/feishu_calendar/feishu_calendar.yaml b/api/core/tools/provider/builtin/feishu_calendar/feishu_calendar.yaml
new file mode 100644
index 0000000000000..db5bab5c1081d
--- /dev/null
+++ b/api/core/tools/provider/builtin/feishu_calendar/feishu_calendar.yaml
@@ -0,0 +1,36 @@
+identity:
+  author: Doug Lea
+  name: feishu_calendar
+  label:
+    en_US: Feishu Calendar
+    zh_Hans: 飞书日历
+  description:
+    en_US: |
+      Feishu calendar, requires the following permissions: calendar:calendar:read、calendar:calendar、contact:user.id:readonly.
+    zh_Hans: |
+      飞书日历，需要开通以下权限: calendar:calendar:read、calendar:calendar、contact:user.id:readonly。
+  icon: icon.png
+  tags:
+    - social
+    - productivity
+credentials_for_provider:
+  app_id:
+    type: text-input
+    required: true
+    label:
+      en_US: APP ID
+    placeholder:
+      en_US: Please input your feishu app id
+      zh_Hans: 请输入你的飞书 app id
+    help:
+      en_US: Get your app_id and app_secret from Feishu
+      zh_Hans: 从飞书获取您的 app_id 和 app_secret
+    url: https://open.larkoffice.com/app
+  app_secret:
+    type: secret-input
+    required: true
+    label:
+      en_US: APP Secret
+    placeholder:
+      en_US: Please input your app secret
+      zh_Hans: 请输入你的飞书 app secret
diff --git a/api/core/tools/provider/builtin/feishu_calendar/tools/add_event_attendees.py b/api/core/tools/provider/builtin/feishu_calendar/tools/add_event_attendees.py
new file mode 100644
index 0000000000000..8f83aea5abbe3
--- /dev/null
+++ b/api/core/tools/provider/builtin/feishu_calendar/tools/add_event_attendees.py
@@ -0,0 +1,20 @@
+from typing import Any
+
+from core.tools.entities.tool_entities import ToolInvokeMessage
+from core.tools.tool.builtin_tool import BuiltinTool
+from core.tools.utils.feishu_api_utils import FeishuRequest
+
+
+class AddEventAttendeesTool(BuiltinTool):
+    def _invoke(self, user_id: str, tool_parameters: dict[str, Any]) -> ToolInvokeMessage:
+        app_id = self.runtime.credentials.get("app_id")
+        app_secret = self.runtime.credentials.get("app_secret")
+        client = FeishuRequest(app_id, app_secret)
+
+        event_id = tool_parameters.get("event_id")
+        attendee_phone_or_email = tool_parameters.get("attendee_phone_or_email")
+        need_notification = tool_parameters.get("need_notification", True)
+
+        res = client.add_event_attendees(event_id, attendee_phone_or_email, need_notification)
+
+        return self.create_json_message(res)
diff --git a/api/core/tools/provider/builtin/feishu_calendar/tools/add_event_attendees.yaml b/api/core/tools/provider/builtin/feishu_calendar/tools/add_event_attendees.yaml
new file mode 100644
index 0000000000000..b7744499b0734
--- /dev/null
+++ b/api/core/tools/provider/builtin/feishu_calendar/tools/add_event_attendees.yaml
@@ -0,0 +1,54 @@
+identity:
+  name: add_event_attendees
+  author: Doug Lea
+  label:
+    en_US: Add Event Attendees
+    zh_Hans: 添加日程参会人
+description:
+  human:
+    en_US: Add Event Attendees
+    zh_Hans: 添加日程参会人
+  llm: A tool for adding attendees to events in Feishu. (在飞书中添加日程参会人)
+parameters:
+  - name: event_id
+    type: string
+    required: true
+    label:
+      en_US: Event ID
+      zh_Hans: 日程 ID
+    human_description:
+      en_US: |
+        The ID of the event, which will be returned when the event is created. For example: fb2a6406-26d6-4c8d-a487-6f0246c94d2f_0.
+      zh_Hans: |
+        创建日程时会返回日程 ID。例如: fb2a6406-26d6-4c8d-a487-6f0246c94d2f_0。
+    llm_description: |
+      日程 ID，创建日程时会返回日程 ID。例如: fb2a6406-26d6-4c8d-a487-6f0246c94d2f_0。
+    form: llm
+
+  - name: need_notification
+    type: boolean
+    required: false
+    default: true
+    label:
+      en_US: Need Notification
+      zh_Hans: 是否需要通知
+    human_description:
+      en_US: |
+        Whether to send a Bot notification to attendees. true: send, false: do not send.
+      zh_Hans: |
+        是否给参与人发送 Bot 通知，true: 发送，false: 不发送。
+    llm_description: |
+      是否给参与人发送 Bot 通知，true: 发送，false: 不发送。
+    form: form
+
+  - name: attendee_phone_or_email
+    type: string
+    required: true
+    label:
+      en_US: Attendee Phone or Email
+      zh_Hans: 参会人电话或邮箱
+    human_description:
+      en_US: The list of attendee emails or phone numbers, separated by commas.
+      zh_Hans: 日程参会人邮箱或者手机号列表，使用逗号分隔。
+    llm_description: 日程参会人邮箱或者手机号列表，使用逗号分隔。
+    form: llm
diff --git a/api/core/tools/provider/builtin/feishu_calendar/tools/create_event.py b/api/core/tools/provider/builtin/feishu_calendar/tools/create_event.py
new file mode 100644
index 0000000000000..8820bebdbed92
--- /dev/null
+++ b/api/core/tools/provider/builtin/feishu_calendar/tools/create_event.py
@@ -0,0 +1,26 @@
+from typing import Any
+
+from core.tools.entities.tool_entities import ToolInvokeMessage
+from core.tools.tool.builtin_tool import BuiltinTool
+from core.tools.utils.feishu_api_utils import FeishuRequest
+
+
+class CreateEventTool(BuiltinTool):
+    def _invoke(self, user_id: str, tool_parameters: dict[str, Any]) -> ToolInvokeMessage:
+        app_id = self.runtime.credentials.get("app_id")
+        app_secret = self.runtime.credentials.get("app_secret")
+        client = FeishuRequest(app_id, app_secret)
+
+        summary = tool_parameters.get("summary")
+        description = tool_parameters.get("description")
+        start_time = tool_parameters.get("start_time")
+        end_time = tool_parameters.get("end_time")
+        attendee_ability = tool_parameters.get("attendee_ability")
+        need_notification = tool_parameters.get("need_notification", True)
+        auto_record = tool_parameters.get("auto_record", False)
+
+        res = client.create_event(
+            summary, description, start_time, end_time, attendee_ability, need_notification, auto_record
+        )
+
+        return self.create_json_message(res)
diff --git a/api/core/tools/provider/builtin/feishu_calendar/tools/create_event.yaml b/api/core/tools/provider/builtin/feishu_calendar/tools/create_event.yaml
new file mode 100644
index 0000000000000..f0784221ce796
--- /dev/null
+++ b/api/core/tools/provider/builtin/feishu_calendar/tools/create_event.yaml
@@ -0,0 +1,119 @@
+identity:
+  name: create_event
+  author: Doug Lea
+  label:
+    en_US: Create Event
+    zh_Hans: 创建日程
+description:
+  human:
+    en_US: Create Event
+    zh_Hans: 创建日程
+  llm: A tool for creating events in Feishu.(创建飞书日程)
+parameters:
+  - name: summary
+    type: string
+    required: false
+    label:
+      en_US: Summary
+      zh_Hans: 日程标题
+    human_description:
+      en_US: The title of the event. If not filled, the event title will display (No Subject).
+      zh_Hans: 日程标题，若不填则日程标题显示 (无主题)。
+    llm_description: 日程标题，若不填则日程标题显示 (无主题)。
+    form: llm
+
+  - name: description
+    type: string
+    required: false
+    label:
+      en_US: Description
+      zh_Hans: 日程描述
+    human_description:
+      en_US: The description of the event.
+      zh_Hans: 日程描述。
+    llm_description: 日程描述。
+    form: llm
+
+  - name: need_notification
+    type: boolean
+    required: false
+    default: true
+    label:
+      en_US: Need Notification
+      zh_Hans: 是否发送通知
+    human_description:
+      en_US: |
+        Whether to send a bot message when the event is created, true: send, false: do not send.
+      zh_Hans: 创建日程时是否发送 bot 消息，true：发送，false：不发送。
+    llm_description: 创建日程时是否发送 bot 消息，true：发送，false：不发送。
+    form: form
+
+  - name: start_time
+    type: string
+    required: true
+    label:
+      en_US: Start Time
+      zh_Hans: 开始时间
+    human_description:
+      en_US: |
+        The start time of the event, format: 2006-01-02 15:04:05.
+      zh_Hans: 日程开始时间，格式：2006-01-02 15:04:05。
+    llm_description: 日程开始时间，格式：2006-01-02 15:04:05。
+    form: llm
+
+  - name: end_time
+    type: string
+    required: true
+    label:
+      en_US: End Time
+      zh_Hans: 结束时间
+    human_description:
+      en_US: |
+        The end time of the event, format: 2006-01-02 15:04:05.
+      zh_Hans: 日程结束时间，格式：2006-01-02 15:04:05。
+    llm_description: 日程结束时间，格式：2006-01-02 15:04:05。
+    form: llm
+
+  - name: attendee_ability
+    type: select
+    required: false
+    options:
+      - value: none
+        label:
+          en_US: none
+          zh_Hans: 无
+      - value: can_see_others
+        label:
+          en_US: can_see_others
+          zh_Hans: 可以查看参与人列表
+      - value: can_invite_others
+        label:
+          en_US: can_invite_others
+          zh_Hans: 可以邀请其它参与人
+      - value: can_modify_event
+        label:
+          en_US: can_modify_event
+          zh_Hans: 可以编辑日程
+    default: "none"
+    label:
+      en_US: attendee_ability
+      zh_Hans: 参会人权限
+    human_description:
+      en_US: Attendee ability, optional values are none, can_see_others, can_invite_others, can_modify_event, with a default value of none.
+      zh_Hans: 参会人权限，可选值有无、可以查看参与人列表、可以邀请其它参与人、可以编辑日程，默认值为无。
+    llm_description: 参会人权限，可选值有无、可以查看参与人列表、可以邀请其它参与人、可以编辑日程，默认值为无。
+    form: form
+
+  - name: auto_record
+    type: boolean
+    required: false
+    default: false
+    label:
+      en_US: Auto Record
+      zh_Hans: 自动录制
+    human_description:
+      en_US: |
+        Whether to enable automatic recording, true: enabled, automatically record when the meeting starts; false: not enabled.
+      zh_Hans: 是否开启自动录制，true：开启，会议开始后自动录制；false：不开启。
+    llm_description: 是否开启自动录制，true：开启，会议开始后自动录制；false：不开启。
+    form: form
diff --git a/api/core/tools/provider/builtin/feishu_calendar/tools/delete_event.py b/api/core/tools/provider/builtin/feishu_calendar/tools/delete_event.py
new file mode 100644
index 0000000000000..144889692f905
--- /dev/null
+++ b/api/core/tools/provider/builtin/feishu_calendar/tools/delete_event.py
@@ -0,0 +1,19 @@
+from typing import Any
+
+from core.tools.entities.tool_entities import ToolInvokeMessage
+from core.tools.tool.builtin_tool import BuiltinTool
+from core.tools.utils.feishu_api_utils import FeishuRequest
+
+
+class DeleteEventTool(BuiltinTool):
+    def _invoke(self, user_id: str, tool_parameters: dict[str, Any]) -> ToolInvokeMessage:
+        app_id = self.runtime.credentials.get("app_id")
+        app_secret = self.runtime.credentials.get("app_secret")
+        client = FeishuRequest(app_id, app_secret)
+
+        event_id = tool_parameters.get("event_id")
+        need_notification = tool_parameters.get("need_notification", True)
+
+        res = client.delete_event(event_id, need_notification)
+
+        return self.create_json_message(res)
diff --git a/api/core/tools/provider/builtin/feishu_calendar/tools/delete_event.yaml b/api/core/tools/provider/builtin/feishu_calendar/tools/delete_event.yaml
new file mode 100644
index 0000000000000..54fdb04acc337
--- /dev/null
+++ b/api/core/tools/provider/builtin/feishu_calendar/tools/delete_event.yaml
@@ -0,0 +1,38 @@
+identity:
+  name: delete_event
+  author: Doug Lea
+  label:
+    en_US: Delete Event
+    zh_Hans: 删除日程
+description:
+  human:
+    en_US: Delete Event
+    zh_Hans: 删除日程
+  llm: A tool for deleting events in Feishu.(在飞书中删除日程)
+parameters:
+  - name: event_id
+    type: string
+    required: true
+    label:
+      en_US: Event ID
+      zh_Hans: 日程 ID
+    human_description:
+      en_US: |
+        The ID of the event, for example: e8b9791c-39ae-4908-8ad8-66b13159b9fb_0.
+      zh_Hans: 日程 ID，例如：e8b9791c-39ae-4908-8ad8-66b13159b9fb_0。
+    llm_description: 日程 ID，例如：e8b9791c-39ae-4908-8ad8-66b13159b9fb_0。
+    form: llm
+
+  - name: need_notification
+    type: boolean
+    required: false
+    default: true
+    label:
+      en_US: Need Notification
+      zh_Hans: 是否需要通知
+    human_description:
+      en_US: |
+        Indicates whether to send bot notifications to event participants upon deletion. true: send, false: do not send.
+      zh_Hans: 删除日程是否给日程参与人发送 bot 通知，true：发送，false：不发送。
+    llm_description: 删除日程是否给日程参与人发送 bot 通知，true：发送，false：不发送。
+    form: form
diff --git a/api/core/tools/provider/builtin/feishu_calendar/tools/get_primary_calendar.py b/api/core/tools/provider/builtin/feishu_calendar/tools/get_primary_calendar.py
new file mode 100644
index 0000000000000..a2cd5a8b17d0a
--- /dev/null
+++ b/api/core/tools/provider/builtin/feishu_calendar/tools/get_primary_calendar.py
@@ -0,0 +1,18 @@
+from typing import Any
+
+from core.tools.entities.tool_entities import ToolInvokeMessage
+from core.tools.tool.builtin_tool import BuiltinTool
+from core.tools.utils.feishu_api_utils import FeishuRequest
+
+
+class GetPrimaryCalendarTool(BuiltinTool):
+    def _invoke(self, user_id: str, tool_parameters: dict[str, Any]) -> ToolInvokeMessage:
+        app_id = self.runtime.credentials.get("app_id")
+        app_secret = self.runtime.credentials.get("app_secret")
+        client = FeishuRequest(app_id, app_secret)
+
+        user_id_type = tool_parameters.get("user_id_type", "open_id")
+
+        res = client.get_primary_calendar(user_id_type)
+
+        return self.create_json_message(res)
diff --git a/api/core/tools/provider/builtin/feishu_calendar/tools/get_primary_calendar.yaml b/api/core/tools/provider/builtin/feishu_calendar/tools/get_primary_calendar.yaml
new file mode 100644
index 0000000000000..3440c85d4a973
--- /dev/null
+++ b/api/core/tools/provider/builtin/feishu_calendar/tools/get_primary_calendar.yaml
@@ -0,0 +1,37 @@
+identity:
+  name: get_primary_calendar
+  author: Doug Lea
+  label:
+    en_US: Get Primary Calendar
+    zh_Hans: 查询主日历信息
+description:
+  human:
+    en_US: Get Primary Calendar
+    zh_Hans: 查询主日历信息
+  llm: A tool for querying primary calendar information in Feishu.(在飞书中查询主日历信息)
+parameters:
+  - name: user_id_type
+    type: select
+    required: false
+    options:
+      - value: open_id
+        label:
+          en_US: open_id
+          zh_Hans: open_id
+      - value: union_id
+        label:
+          en_US: union_id
+          zh_Hans: union_id
+      - value: user_id
+        label:
+          en_US: user_id
+          zh_Hans: user_id
+    default: "open_id"
+    label:
+      en_US: user_id_type
+      zh_Hans: 用户 ID 类型
+    human_description:
+      en_US: User ID type, optional values are open_id, union_id, user_id, with a default value of open_id.
+      zh_Hans: 用户 ID 类型，可选值有 open_id、union_id、user_id，默认值为 open_id。
+    llm_description: 用户 ID 类型，可选值有 open_id、union_id、user_id，默认值为 open_id。
+    form: form
diff --git a/api/core/tools/provider/builtin/feishu_calendar/tools/list_events.py b/api/core/tools/provider/builtin/feishu_calendar/tools/list_events.py
new file mode 100644
index 0000000000000..8815b4c9c871c
--- /dev/null
+++ b/api/core/tools/provider/builtin/feishu_calendar/tools/list_events.py
@@ -0,0 +1,21 @@
+from typing import Any
+
+from core.tools.entities.tool_entities import ToolInvokeMessage
+from core.tools.tool.builtin_tool import BuiltinTool
+from core.tools.utils.feishu_api_utils import FeishuRequest
+
+
+class ListEventsTool(BuiltinTool):
+    def _invoke(self, user_id: str, tool_parameters: dict[str, Any]) -> ToolInvokeMessage:
+        app_id = self.runtime.credentials.get("app_id")
+        app_secret = self.runtime.credentials.get("app_secret")
+        client = FeishuRequest(app_id, app_secret)
+
+        start_time = tool_parameters.get("start_time")
+        end_time = tool_parameters.get("end_time")
+        page_token = tool_parameters.get("page_token")
+        page_size = tool_parameters.get("page_size")
+
+        res = client.list_events(start_time, end_time, page_token, page_size)
+
+        return self.create_json_message(res)
diff --git a/api/core/tools/provider/builtin/feishu_calendar/tools/list_events.yaml b/api/core/tools/provider/builtin/feishu_calendar/tools/list_events.yaml
new file mode 100644
index 0000000000000..f4a5bfe6bab94
--- /dev/null
+++ b/api/core/tools/provider/builtin/feishu_calendar/tools/list_events.yaml
@@ -0,0 +1,62 @@
+identity:
+  name: list_events
+  author: Doug Lea
+  label:
+    en_US: List Events
+    zh_Hans: 获取日程列表
+description:
+  human:
+    en_US: List Events
+    zh_Hans: 获取日程列表
+  llm: A tool for listing events in Feishu.(在飞书中获取日程列表)
+parameters:
+  - name: start_time
+    type: string
+    required: false
+    label:
+      en_US: Start Time
+      zh_Hans: 开始时间
+    human_description:
+      en_US: |
+        The start time, defaults to 0:00 of the current day if not provided, format: 2006-01-02 15:04:05.
+      zh_Hans: 开始时间，不传值时默认当天 0 点时间，格式为:2006-01-02 15:04:05。
+    llm_description: 开始时间，不传值时默认当天 0 点时间，格式为:2006-01-02 15:04:05。
+    form: llm
+
+  - name: end_time
+    type: string
+    required: false
+    label:
+      en_US: End Time
+      zh_Hans: 结束时间
+    human_description:
+      en_US: |
+        The end time, defaults to 23:59 of the current day if not provided, format: 2006-01-02 15:04:05.
+      zh_Hans: 结束时间，不传值时默认当天 23:59 分时间，格式为:2006-01-02 15:04:05。
+    llm_description: 结束时间，不传值时默认当天 23:59 分时间，格式为:2006-01-02 15:04:05。
+    form: llm
+
+  - name: page_size
+    type: number
+    required: false
+    default: 50
+    label:
+      en_US: Page Size
+      zh_Hans: 分页大小
+    human_description:
+      en_US: The page size, i.e., the number of data entries returned in a single request. The default value is 50, and the value range is [50,1000].
+      zh_Hans: 分页大小，即单次请求所返回的数据条目数。默认值为 50，取值范围为 [50,1000]。
+    llm_description: 分页大小，即单次请求所返回的数据条目数。默认值为 50，取值范围为 [50,1000]。
+    form: llm
+
+  - name: page_token
+    type: string
+    required: false
+    label:
+      en_US: Page Token
+      zh_Hans: 分页标记
+    human_description:
+      en_US: The pagination token. Leave it blank for the first request, indicating to start traversing from the beginning; when the pagination query result has more items, a new page_token will be returned simultaneously, which can be used to obtain the query result in the next traversal.
+      zh_Hans: 分页标记，第一次请求不填，表示从头开始遍历；分页查询结果还有更多项时会同时返回新的 page_token，下次遍历可采用该 page_token 获取查询结果。
+    llm_description: 分页标记，第一次请求不填，表示从头开始遍历；分页查询结果还有更多项时会同时返回新的 page_token，下次遍历可采用该 page_token 获取查询结果。
+    form: llm
diff --git a/api/core/tools/provider/builtin/feishu_calendar/tools/search_events.py b/api/core/tools/provider/builtin/feishu_calendar/tools/search_events.py
new file mode 100644
index 0000000000000..dc365205a4cff
--- /dev/null
+++ b/api/core/tools/provider/builtin/feishu_calendar/tools/search_events.py
@@ -0,0 +1,23 @@
+from typing import Any
+
+from core.tools.entities.tool_entities import ToolInvokeMessage
+from core.tools.tool.builtin_tool import BuiltinTool
+from core.tools.utils.feishu_api_utils import FeishuRequest
+
+
+class SearchEventsTool(BuiltinTool):
+    def _invoke(self, user_id: str, tool_parameters: dict[str, Any]) -> ToolInvokeMessage:
+        app_id = self.runtime.credentials.get("app_id")
+        app_secret = self.runtime.credentials.get("app_secret")
+        client = FeishuRequest(app_id, app_secret)
+
+        query = tool_parameters.get("query")
+        start_time = tool_parameters.get("start_time")
+        end_time = tool_parameters.get("end_time")
+        page_token = tool_parameters.get("page_token")
+        user_id_type = tool_parameters.get("user_id_type", "open_id")
+        page_size = tool_parameters.get("page_size", 20)
+
+        res = client.search_events(query, start_time, end_time, page_token, user_id_type, page_size)
+
+        return self.create_json_message(res)
diff --git a/api/core/tools/provider/builtin/feishu_calendar/tools/search_events.yaml b/api/core/tools/provider/builtin/feishu_calendar/tools/search_events.yaml
new file mode 100644
index 0000000000000..e92a282091efc
--- /dev/null
+++ b/api/core/tools/provider/builtin/feishu_calendar/tools/search_events.yaml
@@ -0,0 +1,100 @@
+identity:
+  name: search_events
+  author: Doug Lea
+  label:
+    en_US: Search Events
+    zh_Hans: 搜索日程
+description:
+  human:
+    en_US: Search Events
+    zh_Hans: 搜索日程
+  llm: A tool for searching events in Feishu.(在飞书中搜索日程)
+parameters:
+  - name: user_id_type
+    type: select
+    required: false
+    options:
+      - value: open_id
+        label:
+          en_US: open_id
+          zh_Hans: open_id
+      - value: union_id
+        label:
+          en_US: union_id
+          zh_Hans: union_id
+      - value: user_id
+        label:
+          en_US: user_id
+          zh_Hans: user_id
+    default: "open_id"
+    label:
+      en_US: user_id_type
+      zh_Hans: 用户 ID 类型
+    human_description:
+      en_US: User ID type, optional values are open_id, union_id, user_id, with a default value of open_id.
+      zh_Hans: 用户 ID 类型，可选值有 open_id、union_id、user_id，默认值为 open_id。
+    llm_description: 用户 ID 类型，可选值有 open_id、union_id、user_id，默认值为 open_id。
+    form: form
+
+  - name: query
+    type: string
+    required: true
+    label:
+      en_US: Query
+      zh_Hans: 搜索关键字
+    human_description:
+      en_US: The search keyword used for fuzzy searching event names, with a maximum input of 200 characters.
+      zh_Hans: 用于模糊查询日程名称的搜索关键字，最大输入 200 字符。
+    llm_description: 用于模糊查询日程名称的搜索关键字，最大输入 200 字符。
+    form: llm
+
+  - name: start_time
+    type: string
+    required: false
+    label:
+      en_US: Start Time
+      zh_Hans: 开始时间
+    human_description:
+      en_US: |
+        The start time, defaults to 0:00 of the current day if not provided, format: 2006-01-02 15:04:05.
+      zh_Hans: 开始时间，不传值时默认当天 0 点时间，格式为:2006-01-02 15:04:05。
+    llm_description: 开始时间，不传值时默认当天 0 点时间，格式为:2006-01-02 15:04:05。
+    form: llm
+
+  - name: end_time
+    type: string
+    required: false
+    label:
+      en_US: End Time
+      zh_Hans: 结束时间
+    human_description:
+      en_US: |
+        The end time, defaults to 23:59 of the current day if not provided, format: 2006-01-02 15:04:05.
+      zh_Hans: 结束时间，不传值时默认当天 23:59 分时间，格式为:2006-01-02 15:04:05。
+    llm_description: 结束时间，不传值时默认当天 23:59 分时间，格式为:2006-01-02 15:04:05。
+    form: llm
+
+  - name: page_size
+    type: number
+    required: false
+    default: 20
+    label:
+      en_US: Page Size
+      zh_Hans: 分页大小
+    human_description:
+      en_US: The page size, i.e., the number of data entries returned in a single request. The default value is 20, and the value range is [10,100].
+      zh_Hans: 分页大小，即单次请求所返回的数据条目数。默认值为 20，取值范围为 [10,100]。
+    llm_description: 分页大小，即单次请求所返回的数据条目数。默认值为 20，取值范围为 [10,100]。
+    form: llm
+
+  - name: page_token
+    type: string
+    required: false
+    label:
+      en_US: Page Token
+      zh_Hans: 分页标记
+    human_description:
+      en_US: The pagination token. Leave it blank for the first request, indicating to start traversing from the beginning; when the pagination query result has more items, a new page_token will be returned simultaneously, which can be used to obtain the query result in the next traversal.
+      zh_Hans: 分页标记，第一次请求不填，表示从头开始遍历；分页查询结果还有更多项时会同时返回新的 page_token，下次遍历可采用该 page_token 获取查询结果。
+    llm_description: 分页标记，第一次请求不填，表示从头开始遍历；分页查询结果还有更多项时会同时返回新的 page_token，下次遍历可采用该 page_token 获取查询结果。
+    form: llm
diff --git a/api/core/tools/provider/builtin/feishu_calendar/tools/update_event.py b/api/core/tools/provider/builtin/feishu_calendar/tools/update_event.py
new file mode 100644
index 0000000000000..85bcb1d3f6384
--- /dev/null
+++ b/api/core/tools/provider/builtin/feishu_calendar/tools/update_event.py
@@ -0,0 +1,24 @@
+from typing import Any
+
+from core.tools.entities.tool_entities import ToolInvokeMessage
+from core.tools.tool.builtin_tool import BuiltinTool
+from core.tools.utils.feishu_api_utils import FeishuRequest
+
+
+class UpdateEventTool(BuiltinTool):
+    def _invoke(self, user_id: str, tool_parameters: dict[str, Any]) -> ToolInvokeMessage:
+        app_id = self.runtime.credentials.get("app_id")
+        app_secret = self.runtime.credentials.get("app_secret")
+        client = FeishuRequest(app_id, app_secret)
+
+        event_id = tool_parameters.get("event_id")
+        summary = tool_parameters.get("summary")
+        description = tool_parameters.get("description")
+        need_notification = tool_parameters.get("need_notification", True)
+        start_time = tool_parameters.get("start_time")
+        end_time = tool_parameters.get("end_time")
+        auto_record = tool_parameters.get("auto_record", False)
+
+        res = client.update_event(event_id, summary, description, need_notification, start_time, end_time, auto_record)
+
+        return self.create_json_message(res)
diff --git a/api/core/tools/provider/builtin/feishu_calendar/tools/update_event.yaml b/api/core/tools/provider/builtin/feishu_calendar/tools/update_event.yaml
new file mode 100644
index 0000000000000..4d60dbf8c8e1b
--- /dev/null
+++ b/api/core/tools/provider/builtin/feishu_calendar/tools/update_event.yaml
@@ -0,0 +1,100 @@
+identity:
+  name: update_event
+  author: Doug Lea
+  label:
+    en_US: Update Event
+    zh_Hans: 更新日程
+description:
+  human:
+    en_US: Update Event
+    zh_Hans: 更新日程
+  llm: A tool for updating events in Feishu.(更新飞书中的日程)
+parameters:
+  - name: event_id
+    type: string
+    required: true
+    label:
+      en_US: Event ID
+      zh_Hans: 日程 ID
+    human_description:
+      en_US: |
+        The ID of the event, for example: e8b9791c-39ae-4908-8ad8-66b13159b9fb_0.
+      zh_Hans: 日程 ID，例如：e8b9791c-39ae-4908-8ad8-66b13159b9fb_0。
+    llm_description: 日程 ID，例如：e8b9791c-39ae-4908-8ad8-66b13159b9fb_0。
+    form: llm
+
+  - name: summary
+    type: string
+    required: false
+    label:
+      en_US: Summary
+      zh_Hans: 日程标题
+    human_description:
+      en_US: The title of the event.
+      zh_Hans: 日程标题。
+    llm_description: 日程标题。
+    form: llm
+
+  - name: description
+    type: string
+    required: false
+    label:
+      en_US: Description
+      zh_Hans: 日程描述
+    human_description:
+      en_US: The description of the event.
+      zh_Hans: 日程描述。
+    llm_description: 日程描述。
+    form: llm
+
+  - name: need_notification
+    type: boolean
+    required: false
+    label:
+      en_US: Need Notification
+      zh_Hans: 是否发送通知
+    human_description:
+      en_US: |
+        Whether to send a bot message when the event is updated, true: send, false: do not send.
+      zh_Hans: 更新日程时是否发送 bot 消息，true：发送，false：不发送。
+    llm_description: 更新日程时是否发送 bot 消息，true：发送，false：不发送。
+    form: form
+
+  - name: start_time
+    type: string
+    required: false
+    label:
+      en_US: Start Time
+      zh_Hans: 开始时间
+    human_description:
+      en_US: |
+        The start time of the event, format: 2006-01-02 15:04:05.
+      zh_Hans: 日程开始时间，格式：2006-01-02 15:04:05。
+    llm_description: 日程开始时间，格式：2006-01-02 15:04:05。
+    form: llm
+
+  - name: end_time
+    type: string
+    required: false
+    label:
+      en_US: End Time
+      zh_Hans: 结束时间
+    human_description:
+      en_US: |
+        The end time of the event, format: 2006-01-02 15:04:05.
+      zh_Hans: 日程结束时间，格式：2006-01-02 15:04:05。
+    llm_description: 日程结束时间，格式：2006-01-02 15:04:05。
+    form: llm
+
+  - name: auto_record
+    type: boolean
+    required: false
+    label:
+      en_US: Auto Record
+      zh_Hans: 自动录制
+    human_description:
+      en_US: |
+        Whether to enable automatic recording, true: enabled, automatically record when the meeting starts; false: not enabled.
+      zh_Hans: 是否开启自动录制，true：开启，会议开始后自动录制；false：不开启。
+    llm_description: 是否开启自动录制，true：开启，会议开始后自动录制；false：不开启。
+    form: form
diff --git a/api/core/tools/provider/builtin/feishu_document/feishu_document.py b/api/core/tools/provider/builtin/feishu_document/feishu_document.py
index b0a1e393eb811..217ae52082b82 100644
--- a/api/core/tools/provider/builtin/feishu_document/feishu_document.py
+++ b/api/core/tools/provider/builtin/feishu_document/feishu_document.py
@@ -1,15 +1,7 @@
-from core.tools.errors import ToolProviderCredentialValidationError
 from core.tools.provider.builtin_tool_provider import BuiltinToolProviderController
-from core.tools.utils.feishu_api_utils import FeishuRequest
+from core.tools.utils.feishu_api_utils import auth
 
 
 class FeishuDocumentProvider(BuiltinToolProviderController):
     def _validate_credentials(self, credentials: dict) -> None:
-        app_id = credentials.get("app_id")
-        app_secret = credentials.get("app_secret")
-        if not app_id or not app_secret:
-            raise ToolProviderCredentialValidationError("app_id and app_secret is required")
-        try:
-            assert FeishuRequest(app_id, app_secret).tenant_access_token is not None
-        except Exception as e:
-            raise ToolProviderCredentialValidationError(str(e))
+        auth(credentials)
diff --git a/api/core/tools/provider/builtin/feishu_document/feishu_document.yaml b/api/core/tools/provider/builtin/feishu_document/feishu_document.yaml
index 8eaa6b27049c6..8f9afa6149445 100644
--- a/api/core/tools/provider/builtin/feishu_document/feishu_document.yaml
+++ b/api/core/tools/provider/builtin/feishu_document/feishu_document.yaml
@@ -5,8 +5,10 @@ identity:
     en_US: Lark Cloud Document
     zh_Hans: 飞书云文档
   description:
-    en_US: Lark Cloud Document
-    zh_Hans: 飞书云文档
+    en_US: |
+      Lark cloud document, requires the following permissions: docx:document、drive:drive、docs:document.content:read.
+    zh_Hans: |
+      飞书云文档，需要开通以下权限: docx:document、drive:drive、docs:document.content:read。
   icon: icon.svg
   tags:
     - social
@@ -23,7 +25,7 @@ credentials_for_provider:
     help:
       en_US: Get your app_id and app_secret from Feishu
       zh_Hans: 从飞书获取您的 app_id 和 app_secret
-    url: https://open.feishu.cn
+    url: https://open.larkoffice.com/app
   app_secret:
     type: secret-input
     required: true
diff --git a/api/core/tools/provider/builtin/feishu_document/tools/create_document.yaml b/api/core/tools/provider/builtin/feishu_document/tools/create_document.yaml
index ddf2729f0e4b5..85382e9d8e8d1 100644
--- a/api/core/tools/provider/builtin/feishu_document/tools/create_document.yaml
+++ b/api/core/tools/provider/builtin/feishu_document/tools/create_document.yaml
@@ -7,7 +7,7 @@ identity:
 description:
   human:
     en_US: Create Lark document
-    zh_Hans: 创建飞书文档，支持创建空文档和带内容的文档，支持 markdown 语法创建。
+    zh_Hans: 创建飞书文档，支持创建空文档和带内容的文档，支持 markdown 语法创建。应用需要开启机器人能力(https://open.feishu.cn/document/faq/trouble-shooting/how-to-enable-bot-ability)。
   llm: A tool for creating Feishu documents.
 parameters:
   - name: title
@@ -41,7 +41,8 @@ parameters:
       en_US: folder_token
       zh_Hans: 文档所在文件夹的 Token
     human_description:
-      en_US: The token of the folder where the document is located. If it is not passed or is empty, it means the root directory.
-      zh_Hans: 文档所在文件夹的 Token，不传或传空表示根目录。
-    llm_description: 文档所在文件夹的 Token，不传或传空表示根目录。
+      en_US: |
+        The token of the folder where the document is located. If it is not passed or is empty, it means the root directory. For Example: https://svi136aogf123.feishu.cn/drive/folder/JgR9fiG9AlPt8EdsSNpcGjIInbf
+      zh_Hans: 文档所在文件夹的 Token，不传或传空表示根目录。例如：https://svi136aogf123.feishu.cn/drive/folder/JgR9fiG9AlPt8EdsSNpcGjIInbf。
+    llm_description: 文档所在文件夹的 Token，不传或传空表示根目录。例如：https://svi136aogf123.feishu.cn/drive/folder/JgR9fiG9AlPt8EdsSNpcGjIInbf。
     form: llm
diff --git a/api/core/tools/provider/builtin/feishu_document/tools/get_document_content.py b/api/core/tools/provider/builtin/feishu_document/tools/get_document_content.py
index c94a5f70ed7e3..e67a017facc8d 100644
--- a/api/core/tools/provider/builtin/feishu_document/tools/get_document_content.py
+++ b/api/core/tools/provider/builtin/feishu_document/tools/get_document_content.py
@@ -12,8 +12,8 @@ def _invoke(self, user_id: str, tool_parameters: dict[str, Any]) -> ToolInvokeMe
         client = FeishuRequest(app_id, app_secret)
 
         document_id = tool_parameters.get("document_id")
-        mode = tool_parameters.get("mode")
-        lang = tool_parameters.get("lang", 0)
+        mode = tool_parameters.get("mode", "markdown")
+        lang = tool_parameters.get("lang", "0")
 
         res = client.get_document_content(document_id, mode, lang)
         return self.create_json_message(res)
diff --git a/api/core/tools/provider/builtin/feishu_document/tools/get_document_content.yaml b/api/core/tools/provider/builtin/feishu_document/tools/get_document_content.yaml
index 51eda73a60095..15e827cde91ee 100644
--- a/api/core/tools/provider/builtin/feishu_document/tools/get_document_content.yaml
+++ b/api/core/tools/provider/builtin/feishu_document/tools/get_document_content.yaml
@@ -23,8 +23,18 @@ parameters:
     form: llm
 
   - name: mode
-    type: string
+    type: select
     required: false
+    options:
+      - value: text
+        label:
+          en_US: text
+          zh_Hans: text
+      - value: markdown
+        label:
+          en_US: markdown
+          zh_Hans: markdown
+    default: "markdown"
     label:
       en_US: mode
       zh_Hans: 文档返回格式
@@ -32,18 +42,29 @@ parameters:
       en_US: Format of the document return, optional values are text, markdown, can be empty, default is markdown.
       zh_Hans: 文档返回格式，可选值有 text、markdown，可以为空，默认值为 markdown。
     llm_description: 文档返回格式，可选值有 text、markdown，可以为空，默认值为 markdown。
-    form: llm
+    form: form
 
   - name: lang
-    type: number
+    type: select
     required: false
-    default: 0
+    options:
+      - value: "0"
+        label:
+          en_US: User's default name
+          zh_Hans: 用户的默认名称
+      - value: "1"
+        label:
+          en_US: User's English name
+          zh_Hans: 用户的英文名称
+    default: "0"
     label:
       en_US: lang
       zh_Hans: 指定@用户的语言
     human_description:
       en_US: |
         Specifies the language for MentionUser, optional values are [0, 1]. 0: User's default name, 1: User's English name, default is 0.
-      zh_Hans: 指定返回的 MentionUser，即 @用户 的语言，可选值有 [0,1]。0：该用户的默认名称，1：该用户的英文名称，默认值为 0。
-    llm_description: 指定返回的 MentionUser，即 @用户 的语言，可选值有 [0,1]。0：该用户的默认名称，1：该用户的英文名称，默认值为 0。
-    form: llm
+      zh_Hans: |
+        指定返回的 MentionUser，即@用户的语言，可选值有 [0,1]。0: 该用户的默认名称，1: 该用户的英文名称，默认值为 0。
+    llm_description: |
+      指定返回的 MentionUser，即@用户的语言，可选值有 [0,1]。0: 该用户的默认名称，1: 该用户的英文名称，默认值为 0。
+    form: form
diff --git a/api/core/tools/provider/builtin/feishu_document/tools/list_document_blocks.py b/api/core/tools/provider/builtin/feishu_document/tools/list_document_blocks.py
index 572a7abf28419..dd57c6870d0ba 100644
--- a/api/core/tools/provider/builtin/feishu_document/tools/list_document_blocks.py
+++ b/api/core/tools/provider/builtin/feishu_document/tools/list_document_blocks.py
@@ -12,8 +12,9 @@ def _invoke(self, user_id: str, tool_parameters: dict[str, Any]) -> ToolInvokeMe
         client = FeishuRequest(app_id, app_secret)
 
         document_id = tool_parameters.get("document_id")
-        page_size = tool_parameters.get("page_size", 500)
         page_token = tool_parameters.get("page_token", "")
+        user_id_type = tool_parameters.get("user_id_type", "open_id")
+        page_size = tool_parameters.get("page_size", 500)
 
-        res = client.list_document_blocks(document_id, page_token, page_size)
+        res = client.list_document_blocks(document_id, page_token, user_id_type, page_size)
         return self.create_json_message(res)
diff --git a/api/core/tools/provider/builtin/feishu_document/tools/list_document_blocks.yaml b/api/core/tools/provider/builtin/feishu_document/tools/list_document_blocks.yaml
index 019ac983906ff..5b8ef7d53c23f 100644
--- a/api/core/tools/provider/builtin/feishu_document/tools/list_document_blocks.yaml
+++ b/api/core/tools/provider/builtin/feishu_document/tools/list_document_blocks.yaml
@@ -46,12 +46,12 @@ parameters:
       en_US: User ID type, optional values are open_id, union_id, user_id, with a default value of open_id.
       zh_Hans: 用户 ID 类型，可选值有 open_id、union_id、user_id，默认值为 open_id。
     llm_description: 用户 ID 类型，可选值有 open_id、union_id、user_id，默认值为 open_id。
-    form: llm
+    form: form
 
   - name: page_size
     type: number
     required: false
-    default: "500"
+    default: 500
     label:
       en_US: page_size
       zh_Hans: 分页大小
diff --git a/api/core/tools/provider/builtin/feishu_document/tools/write_document.py b/api/core/tools/provider/builtin/feishu_document/tools/write_document.py
index 6061250e48e13..59f08f53dc68d 100644
--- a/api/core/tools/provider/builtin/feishu_document/tools/write_document.py
+++ b/api/core/tools/provider/builtin/feishu_document/tools/write_document.py
@@ -13,7 +13,7 @@ def _invoke(self, user_id: str, tool_parameters: dict[str, Any]) -> ToolInvokeMe
 
         document_id = tool_parameters.get("document_id")
         content = tool_parameters.get("content")
-        position = tool_parameters.get("position")
+        position = tool_parameters.get("position", "end")
 
         res = client.write_document(document_id, content, position)
         return self.create_json_message(res)
diff --git a/api/core/tools/provider/builtin/feishu_document/tools/write_document.yaml b/api/core/tools/provider/builtin/feishu_document/tools/write_document.yaml
index 4282e3dcf3977..de70f4e7726a2 100644
--- a/api/core/tools/provider/builtin/feishu_document/tools/write_document.yaml
+++ b/api/core/tools/provider/builtin/feishu_document/tools/write_document.yaml
@@ -35,25 +35,23 @@ parameters:
     form: llm
 
   - name: position
-    type: string
+    type: select
     required: false
-    label:
-      en_US: position
-      zh_Hans: 添加位置
-    human_description:
-      en_US: |
-        Enumeration values: start or end. Use 'start' to add content at the beginning of the document, and 'end' to add content at the end. The default value is 'end'.
-      zh_Hans: 枚举值：start 或 end。使用 'start' 在文档开头添加内容，使用 'end' 在文档结尾添加内容，默认值为 'end'。
-    llm_description: |
-      枚举值 start、end，start: 在文档开头添加内容；end: 在文档结尾添加内容，默认值为 end。
-    form: llm
     options:
       - value: start
         label:
-          en_US: start
-          zh_Hans: 在文档开头添加内容
+          en_US: document start
+          zh_Hans: 文档开始
       - value: end
         label:
-          en_US: end
-          zh_Hans: 在文档结尾添加内容
-    default: start
+          en_US: document end
+          zh_Hans: 文档结束
+    default: "end"
+    label:
+      en_US: position
+      zh_Hans: 内容添加位置
+    human_description:
+      en_US: Content insertion position, optional values are start, end. 'start' means adding content at the beginning of the document; 'end' means adding content at the end of the document. The default value is end.
+      zh_Hans: 内容添加位置，可选值有 start、end。start 表示在文档开头添加内容；end 表示在文档结尾添加内容，默认值为 end。
+    llm_description: 内容添加位置，可选值有 start、end。start 表示在文档开头添加内容；end 表示在文档结尾添加内容，默认值为 end。
+    form: form
diff --git a/api/core/tools/provider/builtin/feishu_message/feishu_message.py b/api/core/tools/provider/builtin/feishu_message/feishu_message.py
index 7b3adb9293750..a3b54737691c9 100644
--- a/api/core/tools/provider/builtin/feishu_message/feishu_message.py
+++ b/api/core/tools/provider/builtin/feishu_message/feishu_message.py
@@ -1,15 +1,7 @@
-from core.tools.errors import ToolProviderCredentialValidationError
 from core.tools.provider.builtin_tool_provider import BuiltinToolProviderController
-from core.tools.utils.feishu_api_utils import FeishuRequest
+from core.tools.utils.feishu_api_utils import auth
 
 
 class FeishuMessageProvider(BuiltinToolProviderController):
     def _validate_credentials(self, credentials: dict) -> None:
-        app_id = credentials.get("app_id")
-        app_secret = credentials.get("app_secret")
-        if not app_id or not app_secret:
-            raise ToolProviderCredentialValidationError("app_id and app_secret is required")
-        try:
-            assert FeishuRequest(app_id, app_secret).tenant_access_token is not None
-        except Exception as e:
-            raise ToolProviderCredentialValidationError(str(e))
+        auth(credentials)
diff --git a/api/core/tools/provider/builtin/feishu_message/feishu_message.yaml b/api/core/tools/provider/builtin/feishu_message/feishu_message.yaml
index 1bd8953dddcb2..56683ec1680f4 100644
--- a/api/core/tools/provider/builtin/feishu_message/feishu_message.yaml
+++ b/api/core/tools/provider/builtin/feishu_message/feishu_message.yaml
@@ -5,8 +5,10 @@ identity:
     en_US: Lark Message
     zh_Hans: 飞书消息
   description:
-    en_US: Lark Message
-    zh_Hans: 飞书消息
+    en_US: |
+      Lark message, requires the following permissions: im:message、im:message.group_msg.
+    zh_Hans: |
+      飞书消息，需要开通以下权限: im:message、im:message.group_msg。
   icon: icon.svg
   tags:
     - social
@@ -23,7 +25,7 @@ credentials_for_provider:
     help:
       en_US: Get your app_id and app_secret from Feishu
       zh_Hans: 从飞书获取您的 app_id 和 app_secret
-    url: https://open.feishu.cn
+    url: https://open.larkoffice.com/app
   app_secret:
     type: secret-input
     required: true
diff --git a/api/core/tools/provider/builtin/feishu_message/tools/get_chat_messages.py b/api/core/tools/provider/builtin/feishu_message/tools/get_chat_messages.py
new file mode 100644
index 0000000000000..7eb29230b2ceb
--- /dev/null
+++ b/api/core/tools/provider/builtin/feishu_message/tools/get_chat_messages.py
@@ -0,0 +1,23 @@
+from typing import Any
+
+from core.tools.entities.tool_entities import ToolInvokeMessage
+from core.tools.tool.builtin_tool import BuiltinTool
+from core.tools.utils.feishu_api_utils import FeishuRequest
+
+
+class GetChatMessagesTool(BuiltinTool):
+    def _invoke(self, user_id: str, tool_parameters: dict[str, Any]) -> ToolInvokeMessage:
+        app_id = self.runtime.credentials.get("app_id")
+        app_secret = self.runtime.credentials.get("app_secret")
+        client = FeishuRequest(app_id, app_secret)
+
+        container_id = tool_parameters.get("container_id")
+        start_time = tool_parameters.get("start_time")
+        end_time = tool_parameters.get("end_time")
+        page_token = tool_parameters.get("page_token")
+        sort_type = tool_parameters.get("sort_type", "ByCreateTimeAsc")
+        page_size = tool_parameters.get("page_size", 20)
+
+        res = client.get_chat_messages(container_id, start_time, end_time, page_token, sort_type, page_size)
+
+        return self.create_json_message(res)
diff --git a/api/core/tools/provider/builtin/feishu_message/tools/get_chat_messages.yaml b/api/core/tools/provider/builtin/feishu_message/tools/get_chat_messages.yaml
new file mode 100644
index 0000000000000..153c8c80e58db
--- /dev/null
+++ b/api/core/tools/provider/builtin/feishu_message/tools/get_chat_messages.yaml
@@ -0,0 +1,96 @@
+identity:
+  name: get_chat_messages
+  author: Doug Lea
+  label:
+    en_US: Get Chat Messages
+    zh_Hans: 获取指定单聊、群聊的消息历史
+description:
+  human:
+    en_US: Get Chat Messages
+    zh_Hans: 获取指定单聊、群聊的消息历史
+  llm: A tool for getting chat messages from specific one-on-one chats or group chats.(获取指定单聊、群聊的消息历史)
+parameters:
+  - name: container_id
+    type: string
+    required: true
+    label:
+      en_US: Container Id
+      zh_Hans: 群聊或单聊的 ID
+    human_description:
+      en_US: The ID of the group chat or single chat. Refer to the group ID description for how to obtain it. https://open.feishu.cn/document/server-docs/group/chat/chat-id-description
+      zh_Hans: 群聊或单聊的 ID，获取方式参见群 ID 说明。https://open.feishu.cn/document/server-docs/group/chat/chat-id-description
+    llm_description: 群聊或单聊的 ID，获取方式参见群 ID 说明。https://open.feishu.cn/document/server-docs/group/chat/chat-id-description
+    form: llm
+
+  - name: start_time
+    type: string
+    required: false
+    label:
+      en_US: Start Time
+      zh_Hans: 起始时间
+    human_description:
+      en_US: The start time for querying historical messages, formatted as "2006-01-02 15:04:05".
+      zh_Hans: 待查询历史信息的起始时间，格式为 "2006-01-02 15:04:05"。
+    llm_description: 待查询历史信息的起始时间，格式为 "2006-01-02 15:04:05"。
+    form: llm
+
+  - name: end_time
+    type: string
+    required: false
+    label:
+      en_US: End Time
+      zh_Hans: 结束时间
+    human_description:
+      en_US: The end time for querying historical messages, formatted as "2006-01-02 15:04:05".
+      zh_Hans: 待查询历史信息的结束时间，格式为 "2006-01-02 15:04:05"。
+    llm_description: 待查询历史信息的结束时间，格式为 "2006-01-02 15:04:05"。
+    form: llm
+
+  - name: sort_type
+    type: select
+    required: false
+    options:
+      - value: ByCreateTimeAsc
+        label:
+          en_US: ByCreateTimeAsc
+          zh_Hans: ByCreateTimeAsc
+      - value: ByCreateTimeDesc
+        label:
+          en_US: ByCreateTimeDesc
+          zh_Hans: ByCreateTimeDesc
+    default: "ByCreateTimeAsc"
+    label:
+      en_US: Sort Type
+      zh_Hans: 排序方式
+    human_description:
+      en_US: |
+        The message sorting method. Optional values are ByCreateTimeAsc: sorted in ascending order by message creation time; ByCreateTimeDesc: sorted in descending order by message creation time. The default value is ByCreateTimeAsc. Note: When using page_token for pagination requests, the sorting method (sort_type) is consistent with the first request and cannot be changed midway.
+      zh_Hans: |
+        消息排序方式，可选值有 ByCreateTimeAsc：按消息创建时间升序排列；ByCreateTimeDesc：按消息创建时间降序排列。默认值为：ByCreateTimeAsc。注意：使用 page_token 分页请求时，排序方式（sort_type）均与第一次请求一致，不支持中途改换排序方式。
+    llm_description: 消息排序方式，可选值有 ByCreateTimeAsc：按消息创建时间升序排列；ByCreateTimeDesc：按消息创建时间降序排列。默认值为：ByCreateTimeAsc。注意：使用 page_token 分页请求时，排序方式（sort_type）均与第一次请求一致，不支持中途改换排序方式。
+    form: form
+
+  - name: page_size
+    type: number
+    required: false
+    default: 20
+    label:
+      en_US: Page Size
+      zh_Hans: 分页大小
+    human_description:
+      en_US: The page size, i.e., the number of data entries returned in a single request. The default value is 20, and the value range is [1,50].
+      zh_Hans: 分页大小，即单次请求所返回的数据条目数。默认值为 20，取值范围为 [1,50]。
+    llm_description: 分页大小，即单次请求所返回的数据条目数。默认值为 20，取值范围为 [1,50]。
+    form: llm
+
+  - name: page_token
+    type: string
+    required: false
+    label:
+      en_US: Page Token
+      zh_Hans: 分页标记
+    human_description:
+      en_US: The pagination token. Leave it blank for the first request, indicating to start traversing from the beginning; when the pagination query result has more items, a new page_token will be returned simultaneously, which can be used to obtain the query result in the next traversal.
+      zh_Hans: 分页标记，第一次请求不填，表示从头开始遍历；分页查询结果还有更多项时会同时返回新的 page_token，下次遍历可采用该 page_token 获取查询结果。
+    llm_description: 分页标记，第一次请求不填，表示从头开始遍历；分页查询结果还有更多项时会同时返回新的 page_token，下次遍历可采用该 page_token 获取查询结果。
+    form: llm
diff --git a/api/core/tools/provider/builtin/feishu_message/tools/get_thread_messages.py b/api/core/tools/provider/builtin/feishu_message/tools/get_thread_messages.py
new file mode 100644
index 0000000000000..3b14f46e0048a
--- /dev/null
+++ b/api/core/tools/provider/builtin/feishu_message/tools/get_thread_messages.py
@@ -0,0 +1,21 @@
+from typing import Any
+
+from core.tools.entities.tool_entities import ToolInvokeMessage
+from core.tools.tool.builtin_tool import BuiltinTool
+from core.tools.utils.feishu_api_utils import FeishuRequest
+
+
+class GetChatMessagesTool(BuiltinTool):
+    def _invoke(self, user_id: str, tool_parameters: dict[str, Any]) -> ToolInvokeMessage:
+        app_id = self.runtime.credentials.get("app_id")
+        app_secret = self.runtime.credentials.get("app_secret")
+        client = FeishuRequest(app_id, app_secret)
+
+        container_id = tool_parameters.get("container_id")
+        page_token = tool_parameters.get("page_token")
+        sort_type = tool_parameters.get("sort_type", "ByCreateTimeAsc")
+        page_size = tool_parameters.get("page_size", 20)
+
+        res = client.get_thread_messages(container_id, page_token, sort_type, page_size)
+
+        return self.create_json_message(res)
diff --git a/api/core/tools/provider/builtin/feishu_message/tools/get_thread_messages.yaml b/api/core/tools/provider/builtin/feishu_message/tools/get_thread_messages.yaml
new file mode 100644
index 0000000000000..8d5fed9d0bba2
--- /dev/null
+++ b/api/core/tools/provider/builtin/feishu_message/tools/get_thread_messages.yaml
@@ -0,0 +1,72 @@
+identity:
+  name: get_thread_messages
+  author: Doug Lea
+  label:
+    en_US: Get Thread Messages
+    zh_Hans: 获取指定话题的消息历史
+description:
+  human:
+    en_US: Get Thread Messages
+    zh_Hans: 获取指定话题的消息历史
+  llm: A tool for getting chat messages from specific threads.(获取指定话题的消息历史)
+parameters:
+  - name: container_id
+    type: string
+    required: true
+    label:
+      en_US: Thread Id
+      zh_Hans: 话题 ID
+    human_description:
+      en_US: The ID of the thread. Refer to the thread overview on how to obtain the thread_id. https://open.feishu.cn/document/uAjLw4CM/ukTMukTMukTM/reference/im-v1/message/thread-introduction
+      zh_Hans: 话题 ID，获取方式参见话题概述的如何获取 thread_id 章节。https://open.feishu.cn/document/uAjLw4CM/ukTMukTMukTM/reference/im-v1/message/thread-introduction
+    llm_description: 话题 ID，获取方式参见话题概述的如何获取 thread_id 章节。https://open.feishu.cn/document/uAjLw4CM/ukTMukTMukTM/reference/im-v1/message/thread-introduction
+    form: llm
+
+  - name: sort_type
+    type: select
+    required: false
+    options:
+      - value: ByCreateTimeAsc
+        label:
+          en_US: ByCreateTimeAsc
+          zh_Hans: ByCreateTimeAsc
+      - value: ByCreateTimeDesc
+        label:
+          en_US: ByCreateTimeDesc
+          zh_Hans: ByCreateTimeDesc
+    default: "ByCreateTimeAsc"
+    label:
+      en_US: Sort Type
+      zh_Hans: 排序方式
+    human_description:
+      en_US: |
+        The message sorting method. Optional values are ByCreateTimeAsc: sorted in ascending order by message creation time; ByCreateTimeDesc: sorted in descending order by message creation time. The default value is ByCreateTimeAsc. Note: When using page_token for pagination requests, the sorting method (sort_type) is consistent with the first request and cannot be changed midway.
+      zh_Hans: |
+        消息排序方式，可选值有 ByCreateTimeAsc：按消息创建时间升序排列；ByCreateTimeDesc：按消息创建时间降序排列。默认值为：ByCreateTimeAsc。注意：使用 page_token 分页请求时，排序方式（sort_type）均与第一次请求一致，不支持中途改换排序方式。
+    llm_description: 消息排序方式，可选值有 ByCreateTimeAsc：按消息创建时间升序排列；ByCreateTimeDesc：按消息创建时间降序排列。默认值为：ByCreateTimeAsc。注意：使用 page_token 分页请求时，排序方式（sort_type）均与第一次请求一致，不支持中途改换排序方式。
+    form: form
+
+  - name: page_size
+    type: number
+    required: false
+    default: 20
+    label:
+      en_US: Page Size
+      zh_Hans: 分页大小
+    human_description:
+      en_US: The page size, i.e., the number of data entries returned in a single request. The default value is 20, and the value range is [1,50].
+      zh_Hans: 分页大小，即单次请求所返回的数据条目数。默认值为 20，取值范围为 [1,50]。
+    llm_description: 分页大小，即单次请求所返回的数据条目数。默认值为 20，取值范围为 [1,50]。
+    form: llm
+
+  - name: page_token
+    type: string
+    required: false
+    label:
+      en_US: Page Token
+      zh_Hans: 分页标记
+    human_description:
+      en_US: The pagination token. Leave it blank for the first request, indicating to start traversing from the beginning; when the pagination query result has more items, a new page_token will be returned simultaneously, which can be used to obtain the query result in the next traversal.
+      zh_Hans: 分页标记，第一次请求不填，表示从头开始遍历；分页查询结果还有更多项时会同时返回新的 page_token，下次遍历可采用该 page_token 获取查询结果。
+    llm_description: 分页标记，第一次请求不填，表示从头开始遍历；分页查询结果还有更多项时会同时返回新的 page_token，下次遍历可采用该 page_token 获取查询结果。
+    form: llm
diff --git a/api/core/tools/provider/builtin/feishu_message/tools/send_bot_message.yaml b/api/core/tools/provider/builtin/feishu_message/tools/send_bot_message.yaml
index 6e398b18ab3ae..4f7f65a8a74fc 100644
--- a/api/core/tools/provider/builtin/feishu_message/tools/send_bot_message.yaml
+++ b/api/core/tools/provider/builtin/feishu_message/tools/send_bot_message.yaml
@@ -10,53 +10,53 @@ description:
     zh_Hans: 发送飞书应用消息
   llm: A tool for sending Feishu application messages.
 parameters:
+  - name: receive_id
+    type: string
+    required: true
+    label:
+      en_US: receive_id
+      zh_Hans: 消息接收者的 ID
+    human_description:
+      en_US: The ID of the message receiver, the ID type is consistent with the value of the query parameter receive_id_type.
+      zh_Hans: 消息接收者的 ID，ID 类型与查询参数 receive_id_type 的取值一致。
+    llm_description: 消息接收者的 ID，ID 类型与查询参数 receive_id_type 的取值一致。
+    form: llm
+
   - name: receive_id_type
     type: select
     required: true
     options:
       - value: open_id
         label:
-          en_US: open id
-          zh_Hans: open id
+          en_US: open_id
+          zh_Hans: open_id
       - value: union_id
         label:
-          en_US: union id
-          zh_Hans: union id
+          en_US: union_id
+          zh_Hans: union_id
       - value: user_id
         label:
-          en_US: user id
-          zh_Hans: user id
+          en_US: user_id
+          zh_Hans: user_id
       - value: email
         label:
           en_US: email
           zh_Hans: email
       - value: chat_id
         label:
-          en_US: chat id
-          zh_Hans: chat id
+          en_US: chat_id
+          zh_Hans: chat_id
     label:
-      en_US: User ID Type
-      zh_Hans: 用户 ID 类型
+      en_US: receive_id_type
+      zh_Hans: 消息接收者的 ID 类型
     human_description:
-      en_US: User ID Type
-      zh_Hans: 用户 ID 类型，可选值有 open_id、union_id、user_id、email、chat_id。
-    llm_description: 用户 ID 类型，可选值有 open_id、union_id、user_id、email、chat_id。
-    form: llm
-
-  - name: receive_id
-    type: string
-    required: true
-    label:
-      en_US: Receive Id
-      zh_Hans: 消息接收者的 ID
-    human_description:
-      en_US: The ID of the message receiver. The ID type should correspond to the query parameter receive_id_type.
-      zh_Hans: 消息接收者的 ID，ID 类型应与查询参数 receive_id_type 对应。
-    llm_description: 消息接收者的 ID，ID 类型应与查询参数 receive_id_type 对应。
-    form: llm
+      en_US: The ID type of the message receiver, optional values are open_id, union_id, user_id, email, chat_id, with a default value of open_id.
+      zh_Hans: 消息接收者的 ID 类型，可选值有 open_id、union_id、user_id、email、chat_id，默认值为 open_id。
+    llm_description: 消息接收者的 ID 类型，可选值有 open_id、union_id、user_id、email、chat_id，默认值为 open_id。
+    form: form
 
   - name: msg_type
-    type: string
+    type: select
     required: true
     options:
       - value: text
@@ -65,27 +65,61 @@ parameters:
           zh_Hans: 文本
       - value: interactive
         label:
-          en_US: message card
-          zh_Hans: 消息卡片
+          en_US: interactive
+          zh_Hans: 卡片
+      - value: post
+        label:
+          en_US: post
+          zh_Hans: 富文本
+      - value: image
+        label:
+          en_US: image
+          zh_Hans: 图片
+      - value: file
+        label:
+          en_US: file
+          zh_Hans: 文件
+      - value: audio
+        label:
+          en_US: audio
+          zh_Hans: 语音
+      - value: media
+        label:
+          en_US: media
+          zh_Hans: 视频
+      - value: sticker
+        label:
+          en_US: sticker
+          zh_Hans: 表情包
+      - value: share_chat
+        label:
+          en_US: share_chat
+          zh_Hans: 分享群名片
+      - value: share_user
+        label:
+          en_US: share_user
+          zh_Hans: 分享个人名片
+      - value: system
+        label:
+          en_US: system
+          zh_Hans: 系统消息
     label:
-      en_US: Message type
+      en_US: msg_type
       zh_Hans: 消息类型
     human_description:
-      en_US: Message type, optional values are, text (text), interactive (message card).
-      zh_Hans: 消息类型，可选值有：text(文本)、interactive(消息卡片)。
-    llm_description: 消息类型，可选值有：text(文本)、interactive(消息卡片)。
-    form: llm
+      en_US: Message type. Optional values are text, post, image, file, audio, media, sticker, interactive, share_chat, share_user, system. For detailed introduction of different message types, refer to the message content(https://open.larkoffice.com/document/server-docs/im-v1/message-content-description/create_json).
+      zh_Hans: 消息类型。可选值有：text、post、image、file、audio、media、sticker、interactive、share_chat、share_user、system。不同消息类型的详细介绍，参见发送消息内容(https://open.larkoffice.com/document/server-docs/im-v1/message-content-description/create_json)。
+    llm_description: 消息类型。可选值有：text、post、image、file、audio、media、sticker、interactive、share_chat、share_user、system。不同消息类型的详细介绍，参见发送消息内容(https://open.larkoffice.com/document/server-docs/im-v1/message-content-description/create_json)。
+    form: form
 
   - name: content
     type: string
     required: true
     label:
-      en_US: Message content
+      en_US: content
       zh_Hans: 消息内容
     human_description:
-      en_US: Message content
-      zh_Hans: |
-        消息内容，JSON 结构序列化后的字符串。不同 msg_type 对应不同内容，
-        具体格式说明参考：https://open.larkoffice.com/document/server-docs/im-v1/message-content-description/create_json
-    llm_description: 消息内容，JSON 结构序列化后的字符串。不同 msg_type 对应不同内容。
+      en_US: Message content, a JSON structure serialized string. The value of this parameter corresponds to msg_type. For example, if msg_type is text, this parameter needs to pass in text type content. To understand the format and usage limitations of different message types, refer to the message content(https://open.larkoffice.com/document/server-docs/im-v1/message-content-description/create_json).
+      zh_Hans: 消息内容，JSON 结构序列化后的字符串。该参数的取值与 msg_type 对应，例如 msg_type 取值为 text，则该参数需要传入文本类型的内容。了解不同类型的消息内容格式、使用限制，可参见发送消息内容(https://open.larkoffice.com/document/server-docs/im-v1/message-content-description/create_json)。
+    llm_description: 消息内容，JSON 结构序列化后的字符串。该参数的取值与 msg_type 对应，例如 msg_type 取值为 text，则该参数需要传入文本类型的内容。了解不同类型的消息内容格式、使用限制，可参见发送消息内容(https://open.larkoffice.com/document/server-docs/im-v1/message-content-description/create_json)。
     form: llm
diff --git a/api/core/tools/provider/builtin/feishu_message/tools/send_webhook_message.yaml b/api/core/tools/provider/builtin/feishu_message/tools/send_webhook_message.yaml
index 8b39ce4874d50..eeeae8b29cd93 100644
--- a/api/core/tools/provider/builtin/feishu_message/tools/send_webhook_message.yaml
+++ b/api/core/tools/provider/builtin/feishu_message/tools/send_webhook_message.yaml
@@ -15,15 +15,18 @@ parameters:
     required: true
     label:
       en_US: webhook
-      zh_Hans: webhook 的地址
+      zh_Hans: webhook
     human_description:
-      en_US: The address of the webhook
-      zh_Hans: webhook 的地址
-    llm_description: webhook 的地址
+      en_US: |
+        The address of the webhook, the format of the webhook address corresponding to the bot is as follows: https://open.feishu.cn/open-apis/bot/v2/hook/xxxxxxxxxxxxxxxxx. For details, please refer to: Feishu Custom Bot Usage Guide(https://open.larkoffice.com/document/client-docs/bot-v3/add-custom-bot)
+      zh_Hans: |
+        webhook 的地址，机器人对应的 webhook 地址格式如下: https://open.feishu.cn/open-apis/bot/v2/hook/xxxxxxxxxxxxxxxxx，详情可参考: 飞书自定义机器人使用指南(https://open.larkoffice.com/document/client-docs/bot-v3/add-custom-bot)
+    llm_description: |
+      webhook 的地址，机器人对应的 webhook 地址格式如下: https://open.feishu.cn/open-apis/bot/v2/hook/xxxxxxxxxxxxxxxxx，详情可参考: 飞书自定义机器人使用指南(https://open.larkoffice.com/document/client-docs/bot-v3/add-custom-bot)
     form: llm
 
   - name: msg_type
-    type: string
+    type: select
     required: true
     options:
       - value: text
@@ -32,27 +35,34 @@ parameters:
           zh_Hans: 文本
       - value: interactive
         label:
-          en_US: message card
-          zh_Hans: 消息卡片
+          en_US: interactive
+          zh_Hans: 卡片
+      - value: image
+        label:
+          en_US: image
+          zh_Hans: 图片
+      - value: share_chat
+        label:
+          en_US: share_chat
+          zh_Hans: 分享群名片
     label:
-      en_US: Message type
+      en_US: msg_type
       zh_Hans: 消息类型
     human_description:
-      en_US: Message type, optional values are, text (text), interactive (message card).
-      zh_Hans: 消息类型，可选值有：text(文本)、interactive(消息卡片)。
-    llm_description: 消息类型，可选值有：text(文本)、interactive(消息卡片)。
-    form: llm
+      en_US: Message type. Optional values are text, image, interactive, share_chat. For detailed introduction of different message types, refer to the message content(https://open.larkoffice.com/document/server-docs/im-v1/message-content-description/create_json).
+      zh_Hans: 消息类型。可选值有：text、image、interactive、share_chat。不同消息类型的详细介绍，参见发送消息内容(https://open.larkoffice.com/document/server-docs/im-v1/message-content-description/create_json)。
+    llm_description: 消息类型。可选值有：text、image、interactive、share_chat。不同消息类型的详细介绍，参见发送消息内容(https://open.larkoffice.com/document/server-docs/im-v1/message-content-description/create_json)。
+    form: form
+
 
   - name: content
     type: string
     required: true
     label:
-      en_US: Message content
+      en_US: content
       zh_Hans: 消息内容
     human_description:
-      en_US: Message content
-      zh_Hans: |
-        消息内容，JSON 结构序列化后的字符串。不同 msg_type 对应不同内容，
-        具体格式说明参考：https://open.larkoffice.com/document/server-docs/im-v1/message-content-description/create_json
-    llm_description: 消息内容，JSON 结构序列化后的字符串。不同 msg_type 对应不同内容。
+      en_US: Message content, a JSON structure serialized string. The value of this parameter corresponds to msg_type. For example, if msg_type is text, this parameter needs to pass in text type content. To understand the format and usage limitations of different message types, refer to the message content(https://open.larkoffice.com/document/server-docs/im-v1/message-content-description/create_json).
+      zh_Hans: 消息内容，JSON 结构序列化后的字符串。该参数的取值与 msg_type 对应，例如 msg_type 取值为 text，则该参数需要传入文本类型的内容。了解不同类型的消息内容格式、使用限制，可参见发送消息内容(https://open.larkoffice.com/document/server-docs/im-v1/message-content-description/create_json)。
+    llm_description: 消息内容，JSON 结构序列化后的字符串。该参数的取值与 msg_type 对应，例如 msg_type 取值为 text，则该参数需要传入文本类型的内容。了解不同类型的消息内容格式、使用限制，可参见发送消息内容(https://open.larkoffice.com/document/server-docs/im-v1/message-content-description/create_json)。
     form: llm
diff --git a/api/core/tools/provider/builtin/feishu_spreadsheet/_assets/icon.png b/api/core/tools/provider/builtin/feishu_spreadsheet/_assets/icon.png
new file mode 100644
index 0000000000000000000000000000000000000000..258b361261d4e3366251613141efaf200cd492db
GIT binary patch
literal 5417
zcmd^DS5%YRwoZN$O6VAxB7`D>B2{_|NCfFcC5m9^B1Ni1YN8;ag8~B5K}5ljNR=At
zDowYWE@T6ufDnpw#PfU4%XzwQcZ_?-eMtT>zS-xRYp%IgytSn<C*n8)0)cQ|GBL6R
zziWSftjyrQ-j&5i;0GFHYitN14~i^8Ajr8(Mh2JhPAmE08OXo$d)phDo_jX22sv$J
zzcoQ8X({$I4M1mYUXN33N<B-3jydw#j5@xBBVluXSp)LMJp7KQ;+1ZXbQwJu*A>sL
zdRI`?%X-mSlvNKO|Ie?xLgnG*o=vL-$74~@Z_^H692>&gwz#AF8KDeDRMbw;%S7@^
z0lEPo#d95ysLibL=R-lQaee?Fs}d9;GP^(x<}0F2F-h^b07#9l0oO$-H1Q3K391Vw
z<4=0p`{@z^2;xpdlh~~3JnE^>9z);@q9`KIu_BtdpQzbsD575%4Twvli5I4HvMd2)
zb~piMXifm)8t=*F!(<e9AZ#(&5I9%BCRU5h8n+7lA2M9bi$6TmHEA~ID(;`3wwo8D
z!HrgO5b^+&;dDiUJ{z6D?pevSmy#Cf>(F(ty-{~Wz)Z}Ndy@Eb!#j;_(=_NxxD1SZ
z`^;W2^5auNzYn@A2>xJh;sP2HSC+;woK59g(<;NCo^!c#Ehuot?D4Rd)w8!(U0P`&
z%VC9Q-96X{A0A-3$C;#tp*U^SO+lV&Hxa7rN-a66lFyi_>RrN%R|1Vq#cqxM66~=V
zqiED^xRL<xd16`2lz17QKLy`C;C5AQq}G?csiJ71;X1vkeAk?f3PH~lK?0n-*-Rh@
zP#+}EP<KhXST$KEuy5ja(B<qpz+U~Q@K@4cocJ1spl?)AdL-CEDd_Vhj4=~WZC!a~
zCD2iFc@$SoTjqPB#rFZiVeiN>X95ANScBy+VLmy`yfkVWj>n$$_jO&|;@RMZxW`Vz
z$O_A_ZC28OHeIiVe5Q^a@Eh@@xS_-0ekSH|OjbBisUizBYwacbqfha(|5(|VJ2}WB
zHpATzRjsPoJQBe4PIt1hgrI7Y<u3`Nkh4!riJ3M^ie0e!&bM~m=Y9#23n-n<2bDN3
z#0|w)^M+00X~2BYNeQ*hftMbP?ScCmGX4g&>50}-!}5gFUy?Rg9_E&^py-is{-p;%
zTvMdnGy%=Wqn{@HrmR@{jrI>dJKTelw2a|nx8U=k7~^*xb#q+marab_8vS2fSE5+Y
zn~5{LDooF9tC$SGL27*G@;{=nn$Xky=U;8P+72;ajTwrUrC1Tnxpj$-`QP(e{CvV9
zh3pN9onxvac@}^z>)z{0S+aM9#p?S7+MZA%@a=)}H%cBs2SJLPD0zo*4!;C}*GG6$
zTMl`6(oQ&vd#E0UZeZkChdQ~D8u|whYR0Tkk_54k46T{#q1zuLx1St1$09Z4@?G-x
zP;12H+5MiS`8ypOtP=3S-g>%7u{bLB#@8}4dCm`<qDYPN4vDb~$%Ks25&O8gW@kiR
zgX7!yjtl6cwxQzB^NKCexj>w!J@u0gpTDCfl`36#Ce`k^1%{YRQF_6`gSF9DyZ3Bm
zxr|mHO4@>BaQa#*$Cj2S)LE_&-nSsJv3#y!5=1E}zv^mBz1y3!vXp@@&b|hGYttj^
z?UFMVQR53YFP6~^{CEd_wTWBGlH@xTzP_*PtPwumo>==enAF|ol)S&NnfqRA-FZ#G
zf5A^UVYdF1#whcR86TO2%c|Bm2kp)#9~26{2Ge@<yY{ZgBQ<1jASZIHhfI<7W@U|^
z36W4PebMAzs*1_CW`);N&sM#Y2E2{p@~C%jETyXPhg1n5d|d6Zk1?=$Y4Lnv?@EVg
zZyTFlvtpXs-B{IeBV0RA;QqwR&w)=9W`)Ok{3z1wY$O3$rPku6tzH3vUiHASG31CF
zFEpM*#Jsxt^=6i7R80M?Y6{Yftufo?=HnY4t|9++sF~Fo_bckNnpAn`Tgm;LeiTs-
zY<c@=-(wo987{Xz=8%j{w*>9z!sOtO3(_ASdG)a5Jouh*GPceIQ8i;V2@|!6kT2)T
zP3G<Xwy01}Q<1t+HYJtfksF{91YtDzb}Vdj;%l}zYnzNa+QUGR3zIOi`DbURJzM<m
zja@guSjBiaPm`3txU(Mi!Srz9fhYC3YdD^2gZnjYgNyX2cQ-K!?mNF7@}o<EOE1Nf
zD}vi=+|MEB%jgASvuAK13_jxN5#F<WPT`*;$50oCFxN)%V_C}4NC|Dfzax#R(#mPI
z>TrKXy&4^~U@c_IEJKMZl%eR=P~7mddg1f~Y}6Wy#^*OPN3UfmQK^3@itZpKbd5@0
zc3B?kLSn(>gN`_VSnG<!@3hnH;|l8!a<hTuJLeEpLvl&N(>c2%ub#NtO@@Vkcuw1I
zE&gLUmn!~7K;3W8aMa%NswfiMeh!H(yq2vZu#y-1l|`&3hz-OPVp0Cr>%V@TkU!%>
zig$7Eo6aFiCaPc7AQg38k{ODANjtgts?q(!v%Aw&3~?*`#9gB9y+Zv%yYux6M}lWg
zZx<xJi+nC~VS409zF@4uNVBb(Mk2Z3f?3ZToAWoQPPAG2xhnrv;b?lb-g&$Lj^C$s
zPQ1~y0#}&SDev!H4o?gn=wXkGE^o&VB)#zL@8~aP&#cLPRP8*m&~BSz&nCyG#6jZf
zso1^MFWe;5nPB{~>}2Y?UvriIKFi_e_K9Y{*4iad-AWN{so#fZ7hA_A!jW=?o>IKy
zVmr>iln9}Jc@{D>cn$kyL;F=r9*Z3=5=?1+js+lA8D{cPQAguYGbtg}#)-6p{L3|*
z@z$zI?_#(<*R6L3#H|ws#J_XKWhkN21!JDd0#QqszIwaZd#vflMa8h(9b$wT5L8*<
zBX{8NtDYXEG^ec#34>iz2?zd{3O{5DZf>LtQt;n=e$*{$Wdmn-7<kklzDioRrO1qi
ztD2v1M8k&5NPXM=ZtXk0S@jJWL4l$s#4TPsT!=?KLJsBc{`*C?Lsa3$PS^6|JGhU7
z<9O}Ai?~pu>EK!_{#(~HLVkHiwO`?8f|1cmvooXKl(DSNxarLW);a=FHJAic)+i^G
z(e}q=C|W8Y@LuK-u<Tk2rV9klmYN`{Qj>v76%eROl8~k7sr_FPLKx5)D|=j5MXweA
zsi4aJnHaAJ-_CgY`b`pY7?BvwX)Ah}o;`g;d}iT<<{|np=0<vHKqNO<$dOFni<#Ck
zwZf90(-K<F91F>|Z$nPE>jWATfMx145R;)u$P_vwvE1OY&0~oC274T(v@nP)MM42l
z#rob;2tLAOLgc?^gTq+sb;?jg(TJ+PMBwcG4dIK#Er=}+<AlWWW(+AJs#wy1vl}I}
zgrbOtHaJ}Z5}PgvzY6=Rk_0S6%l>;)jzkBQbA6MY987*4wFXPvD(6q#>};6+{c!gE
zuh?~G=IFzw59PEky!+o7(m;57&_TyPg|dS<(VkwmfpJ;)jR3<_=3?xKgTG9sg|mT8
zmbR_XHYesG&?uhyDlEiy%>7r}!@aL&dpZ>1HoqDf{|1sCM-k|%3`JO`Qx1$Y1q`W#
z=265^17<`61QLL!MCE=k9QvOWA$*wMEB*<)^nabuVB#n)s@pcn@<vexczgsf$eP%I
z@3JN)&qV%0wWW7v<hh8>^cAx%#VAfr_bI)ekLa(&LqeVj$|@{;Y-L}+dGMs<i-qjB
zY?v+X*zMz*NG#KcadADHPGbeIc_wRiR__VO`E21E)ooPd$Kn_{&y~-FXi?!cZ;j)`
zW+86OO8FlQV|FKxW@02IOxZW46GL_i{b-fv6gpxnGhF$M{)3r=^@jeM+LX?&{Lh4+
zHcrFkpPy#et=tCH7Uz|oFd>G-d-d4iEN~E?1J;#SxCn*<otzM7&3l4eSClz$*8lga
zHdVb<<+LYwGXBl=3kMz!{TGQW{8#AugApQit$Hlb*59<5={EQdBR9^%cR7F(q@d+?
zUF^M=vW)+TbwL;kQN=xWk=T0WhWeVCx!7w+w`KfxNsITadP-Xo=L=PhVJeGo@a@!q
zmof3Jr(FYQWZ%24n76YqqWd-jE}xdBZyen;Hm_KpwW|Ccf?=F5QLnGQ<p3A84VD%?
zx#!K1n7me}#|Q^IXhd>=oaRy&&#~_7uBTMaU0*MMYxCUXn$sPyZ5ihR%j-)C1<G`f
z{%ZU%>apw>-l9JRzB?mJ5v?YFm*DN3ll)!1?9R}6Z3b@E^)gymONw#o8R29`3SRVA
zsi#m1t=0|{3Ol$DlC?xn@?&-b+n4<(hog<Jm9_sPW3HV8fsafcLt;}$drNc!+O^ii
zn#l17#w>e6@h^J8p`^7HP29r&!qo2Y?k_>r=O&21`&6h_Sx#E;kq2Ot#^)<fXdjDi
z`++b&abECiZm5^L^ZxzgpuM`se{A~JhFa*f?21g;L@CJs2)UGXaphhosJ6fkr|b0P
zVz~Ncyx8;V<NlJ&{F$XDNbi}u<112d5zuI^8K_;>I6TxR^2Lm5BW3iP4X-x4dph*G
zm6EL0WgHQStu6)~e{wh2tBkuxN}e`;DpR8pkKD&Fm)Mta1e}IKjl)|Et;Qyf)CwjV
zkPeeGRL>U?*#_-`_0#OH+8^nA;z!>_N-_;=+cGoY(3B$D=}mL@>(pjfZCk%evCrcJ
z&uIHP?BLXa=%?HV6Lp6PoN{`SM8K5p<G?dQ#bcfX!YmHW@&W8O<Q@<vPR##&j$>x4
z06A{wSth;olH1c3?@~tp92xRpW^+pix#O-i$m?O6I<%QK@99QH^!G_a5mi<-Wa`gE
zzEx$`fchvk6%$k<3><;_n^f0_-b<{r5(4(e|ICEo-FaWJS52Vs7-pVVc5U=O(B%;n
zv7{^76nDI(5CdMo4L}z>=DnZt{YJ#!;~_mwrpfX30Z_0@QJch%LX1fVD8<h{FpqS!
z0JzL5B-O2Ft{9^u80}UKZ#D+FhC&kuP?md|5)OKgclinJ|AbJ4FL<~UA#k-g>|}N$
z9N>nSWT$qR^Q{>WfTx|6jt&uOCFxfnj6UV(edaeo{}m!Cd`%DeX#7VPoz>WcK<6LQ
z^-L@MNLKzL>o41FvMqpzAe(q&>L>YY@!!*<)@KY!myRyFfXqBaH0EJT(-*!;(fo!w
zbJ0>MXdoA4mZsRnjJ{g5U${6KbuK2G9(^7p90v)^Re!~rYW$rXub|1*azcCG!`X<-
z0JsfOjz3>5x!8VHt5HK<(3JiZa<lHQlL|+=X5foM7FwQQx0%`HX0lgA&^T>Y8YsjE
z4#P-|aP9d+&hJ+~<z#dB<q*DSwx(<HvE@C4zz0`HDqdJ+Xxr@7X3V-wOXfS)i>x3)
zRoe}iD|SBMZC((X;TavAVgj57V|X*prE0P64W*hxeg&y88J~hekQ$+rMdeC*R;&Sn
zgx6C89qHlcp@_WZ?QnH`UrBv4bP7TGG&Rtb9xeyMog-$X+=pG(yy-#ELtWJFls{L%
zPG7OCf6C<aseo!CDL9J7-;W0!km5PNrW+4BXw}eLb#_@Wb1@Fq^*3b3Ek|rg7bk$o
zYnM*FsVX<^6xP=)U4WqX9}e{AAhJM#Xm9VOIr~IXHcQcQ^!eZX{yq1_(ZtP}_p*HH
z$rh*PBzaWmq_g5SkA9!ga5Vsfg4Wq|ytpo?pa76Fc|svVeAcJ()LLM<;@wI6OII06
zSC>;n*PmYab_Xq25TBgp_tTPf3g;+@O>2dfO)Z!zl~Sem?k8imS&3FK{d%ffu4JAj
zAjB0MS+yF#;WI_b8j*swkZ%2~-+wRp)VGRi>^J?MGjd<>nRdAr7r7&Dh`-Pv-FK>h
zIQo;wPAv-XS-wED56lI^%4ogV%jy{%p)hIaAjY;Y9uVr$e&h4y<vp@li5+4+XwhTU
zugYsO>)Uwd^W~3#Sm}JL6e>&pB)4%2Uz#8r-4KnR5f_ww?1ls1cbrIlQ*~F)F~(XH
z5fO%=Ejw1~Pfjtv;bPV{EI26c+pUV(<q*48d+&s2jam-S^~thP|Au<~N~=#wULsXy
z&8s4;J*m-E$Yl}RjtviJl6n*{6c|&%)b+-&NfN@50Tl9~_0elAeL+<1&GY533Fw|0
z><A@QAn#|P`l^6%xuASUl@q=loBB?Q2P_2<x#&SlpbgxtH+Qqmfw$mxdb{!MhZH>#
e^8bzLQ2*ztuUWBv%zNP5C*%^w(uiz`yZ<i?3E!pw

literal 0
HcmV?d00001

diff --git a/api/core/tools/provider/builtin/feishu_spreadsheet/feishu_spreadsheet.py b/api/core/tools/provider/builtin/feishu_spreadsheet/feishu_spreadsheet.py
new file mode 100644
index 0000000000000..a3b54737691c9
--- /dev/null
+++ b/api/core/tools/provider/builtin/feishu_spreadsheet/feishu_spreadsheet.py
@@ -0,0 +1,7 @@
+from core.tools.provider.builtin_tool_provider import BuiltinToolProviderController
+from core.tools.utils.feishu_api_utils import auth
+
+
+class FeishuMessageProvider(BuiltinToolProviderController):
+    def _validate_credentials(self, credentials: dict) -> None:
+        auth(credentials)
diff --git a/api/core/tools/provider/builtin/feishu_spreadsheet/feishu_spreadsheet.yaml b/api/core/tools/provider/builtin/feishu_spreadsheet/feishu_spreadsheet.yaml
new file mode 100644
index 0000000000000..29e448d730f74
--- /dev/null
+++ b/api/core/tools/provider/builtin/feishu_spreadsheet/feishu_spreadsheet.yaml
@@ -0,0 +1,36 @@
+identity:
+  author: Doug Lea
+  name: feishu_spreadsheet
+  label:
+    en_US: Feishu Spreadsheet
+    zh_Hans: 飞书电子表格
+  description:
+    en_US: |
+      Feishu Spreadsheet, requires the following permissions: sheets:spreadsheet.
+    zh_Hans: |
+      飞书电子表格，需要开通以下权限: sheets:spreadsheet。
+  icon: icon.png
+  tags:
+    - social
+    - productivity
+credentials_for_provider:
+  app_id:
+    type: text-input
+    required: true
+    label:
+      en_US: APP ID
+    placeholder:
+      en_US: Please input your feishu app id
+      zh_Hans: 请输入你的飞书 app id
+    help:
+      en_US: Get your app_id and app_secret from Feishu
+      zh_Hans: 从飞书获取您的 app_id 和 app_secret
+    url: https://open.larkoffice.com/app
+  app_secret:
+    type: secret-input
+    required: true
+    label:
+      en_US: APP Secret
+    placeholder:
+      en_US: Please input your app secret
+      zh_Hans: 请输入你的飞书 app secret
diff --git a/api/core/tools/provider/builtin/feishu_spreadsheet/tools/add_cols.py b/api/core/tools/provider/builtin/feishu_spreadsheet/tools/add_cols.py
new file mode 100644
index 0000000000000..44d062f9bdded
--- /dev/null
+++ b/api/core/tools/provider/builtin/feishu_spreadsheet/tools/add_cols.py
@@ -0,0 +1,22 @@
+from typing import Any
+
+from core.tools.entities.tool_entities import ToolInvokeMessage
+from core.tools.tool.builtin_tool import BuiltinTool
+from core.tools.utils.feishu_api_utils import FeishuRequest
+
+
+class AddColsTool(BuiltinTool):
+    def _invoke(self, user_id: str, tool_parameters: dict[str, Any]) -> ToolInvokeMessage:
+        app_id = self.runtime.credentials.get("app_id")
+        app_secret = self.runtime.credentials.get("app_secret")
+        client = FeishuRequest(app_id, app_secret)
+
+        spreadsheet_token = tool_parameters.get("spreadsheet_token")
+        sheet_id = tool_parameters.get("sheet_id")
+        sheet_name = tool_parameters.get("sheet_name")
+        length = tool_parameters.get("length")
+        values = tool_parameters.get("values")
+
+        res = client.add_cols(spreadsheet_token, sheet_id, sheet_name, length, values)
+
+        return self.create_json_message(res)
diff --git a/api/core/tools/provider/builtin/feishu_spreadsheet/tools/add_cols.yaml b/api/core/tools/provider/builtin/feishu_spreadsheet/tools/add_cols.yaml
new file mode 100644
index 0000000000000..ef457f8e009b2
--- /dev/null
+++ b/api/core/tools/provider/builtin/feishu_spreadsheet/tools/add_cols.yaml
@@ -0,0 +1,72 @@
+identity:
+  name: add_cols
+  author: Doug Lea
+  label:
+    en_US: Add Cols
+    zh_Hans: 新增多列至工作表最后
+description:
+  human:
+    en_US: Add Cols
+    zh_Hans: 新增多列至工作表最后
+  llm: A tool for adding multiple columns to the end of a spreadsheet. (新增多列至工作表最后)
+parameters:
+  - name: spreadsheet_token
+    type: string
+    required: true
+    label:
+      en_US: spreadsheet_token
+      zh_Hans: 电子表格 token
+    human_description:
+      en_US: Spreadsheet token, supports input of spreadsheet URL.
+      zh_Hans: 电子表格 token，支持输入电子表格 url。
+    llm_description: 电子表格 token，支持输入电子表格 url。
+    form: llm
+
+  - name: sheet_id
+    type: string
+    required: false
+    label:
+      en_US: sheet_id
+      zh_Hans: 工作表 ID
+    human_description:
+      en_US: Sheet ID, either sheet_id or sheet_name must be filled.
+      zh_Hans: 工作表 ID，与 sheet_name 二者其一必填。
+    llm_description: 工作表 ID，与 sheet_name 二者其一必填。
+    form: llm
+
+  - name: sheet_name
+    type: string
+    required: false
+    label:
+      en_US: sheet_name
+      zh_Hans: 工作表名称
+    human_description:
+      en_US: Sheet name, either sheet_id or sheet_name must be filled.
+      zh_Hans: 工作表名称，与 sheet_id 二者其一必填。
+    llm_description: 工作表名称，与 sheet_id 二者其一必填。
+    form: llm
+
+  - name: length
+    type: number
+    required: true
+    label:
+      en_US: length
+      zh_Hans: 要增加的列数
+    human_description:
+      en_US: Number of columns to add, range (0-5000].
+      zh_Hans: 要增加的列数，范围(0-5000]。
+    llm_description: 要增加的列数，范围(0-5000]。
+    form: llm
+
+  - name: values
+    type: string
+    required: false
+    label:
+      en_US: values
+      zh_Hans: 新增列的单元格内容
+    human_description:
+      en_US: |
+        Content of the new columns, array of objects in string format, each array represents a row of table data, format like: [ [ "ID","Name","Age" ],[ 1,"Zhang San",10 ],[ 2,"Li Si",11 ] ].
+      zh_Hans: 新增列的单元格内容，数组对象字符串，每个数组一行表格数据，格式：[["编号","姓名","年龄"],[1,"张三",10],[2,"李四",11]]。
+    llm_description: 新增列的单元格内容，数组对象字符串，每个数组一行表格数据，格式：[["编号","姓名","年龄"],[1,"张三",10],[2,"李四",11]]。
+    form: llm
diff --git a/api/core/tools/provider/builtin/feishu_spreadsheet/tools/add_rows.py b/api/core/tools/provider/builtin/feishu_spreadsheet/tools/add_rows.py
new file mode 100644
index 0000000000000..3a85b7b46ccb9
--- /dev/null
+++ b/api/core/tools/provider/builtin/feishu_spreadsheet/tools/add_rows.py
@@ -0,0 +1,22 @@
+from typing import Any
+
+from core.tools.entities.tool_entities import ToolInvokeMessage
+from core.tools.tool.builtin_tool import BuiltinTool
+from core.tools.utils.feishu_api_utils import FeishuRequest
+
+
+class AddRowsTool(BuiltinTool):
+    def _invoke(self, user_id: str, tool_parameters: dict[str, Any]) -> ToolInvokeMessage:
+        app_id = self.runtime.credentials.get("app_id")
+        app_secret = self.runtime.credentials.get("app_secret")
+        client = FeishuRequest(app_id, app_secret)
+
+        spreadsheet_token = tool_parameters.get("spreadsheet_token")
+        sheet_id = tool_parameters.get("sheet_id")
+        sheet_name = tool_parameters.get("sheet_name")
+        length = tool_parameters.get("length")
+        values = tool_parameters.get("values")
+
+        res = client.add_rows(spreadsheet_token, sheet_id, sheet_name, length, values)
+
+        return self.create_json_message(res)
diff --git a/api/core/tools/provider/builtin/feishu_spreadsheet/tools/add_rows.yaml b/api/core/tools/provider/builtin/feishu_spreadsheet/tools/add_rows.yaml
new file mode 100644
index 0000000000000..37653325aeb16
--- /dev/null
+++ b/api/core/tools/provider/builtin/feishu_spreadsheet/tools/add_rows.yaml
@@ -0,0 +1,72 @@
+identity:
+  name: add_rows
+  author: Doug Lea
+  label:
+    en_US: Add Rows
+    zh_Hans: 新增多行至工作表最后
+description:
+  human:
+    en_US: Add Rows
+    zh_Hans: 新增多行至工作表最后
+  llm: A tool for adding multiple rows to the end of a spreadsheet. (新增多行至工作表最后)
+parameters:
+  - name: spreadsheet_token
+    type: string
+    required: true
+    label:
+      en_US: spreadsheet_token
+      zh_Hans: 电子表格 token
+    human_description:
+      en_US: Spreadsheet token, supports input of spreadsheet URL.
+      zh_Hans: 电子表格 token，支持输入电子表格 url。
+    llm_description: 电子表格 token，支持输入电子表格 url。
+    form: llm
+
+  - name: sheet_id
+    type: string
+    required: false
+    label:
+      en_US: sheet_id
+      zh_Hans: 工作表 ID
+    human_description:
+      en_US: Sheet ID, either sheet_id or sheet_name must be filled.
+      zh_Hans: 工作表 ID，与 sheet_name 二者其一必填。
+    llm_description: 工作表 ID，与 sheet_name 二者其一必填。
+    form: llm
+
+  - name: sheet_name
+    type: string
+    required: false
+    label:
+      en_US: sheet_name
+      zh_Hans: 工作表名称
+    human_description:
+      en_US: Sheet name, either sheet_id or sheet_name must be filled.
+      zh_Hans: 工作表名称，与 sheet_id 二者其一必填。
+    llm_description: 工作表名称，与 sheet_id 二者其一必填。
+    form: llm
+
+  - name: length
+    type: number
+    required: true
+    label:
+      en_US: length
+      zh_Hans: 要增加行数
+    human_description:
+      en_US: Number of rows to add, range (0-5000].
+      zh_Hans: 要增加行数，范围(0-5000]。
+    llm_description: 要增加行数，范围(0-5000]。
+    form: llm
+
+  - name: values
+    type: string
+    required: false
+    label:
+      en_US: values
+      zh_Hans: 新增行的表格内容
+    human_description:
+      en_US: |
+        Content of the new rows, array of objects in string format, each array represents a row of table data, format like: [ [ "ID","Name","Age" ],[ 1,"Zhang San",10 ],[ 2,"Li Si",11 ] ].
+      zh_Hans: 新增行的表格内容，数组对象字符串，每个数组一行表格数据，格式,如：[["编号","姓名","年龄"],[1,"张三",10],[2,"李四",11]]。
+    llm_description: 新增行的表格内容，数组对象字符串，每个数组一行表格数据，格式,如：[["编号","姓名","年龄"],[1,"张三",10],[2,"李四",11]]。
+    form: llm
diff --git a/api/core/tools/provider/builtin/feishu_spreadsheet/tools/create_spreadsheet.py b/api/core/tools/provider/builtin/feishu_spreadsheet/tools/create_spreadsheet.py
new file mode 100644
index 0000000000000..647364fab0a96
--- /dev/null
+++ b/api/core/tools/provider/builtin/feishu_spreadsheet/tools/create_spreadsheet.py
@@ -0,0 +1,19 @@
+from typing import Any
+
+from core.tools.entities.tool_entities import ToolInvokeMessage
+from core.tools.tool.builtin_tool import BuiltinTool
+from core.tools.utils.feishu_api_utils import FeishuRequest
+
+
+class CreateSpreadsheetTool(BuiltinTool):
+    def _invoke(self, user_id: str, tool_parameters: dict[str, Any]) -> ToolInvokeMessage:
+        app_id = self.runtime.credentials.get("app_id")
+        app_secret = self.runtime.credentials.get("app_secret")
+        client = FeishuRequest(app_id, app_secret)
+
+        title = tool_parameters.get("title")
+        folder_token = tool_parameters.get("folder_token")
+
+        res = client.create_spreadsheet(title, folder_token)
+
+        return self.create_json_message(res)
diff --git a/api/core/tools/provider/builtin/feishu_spreadsheet/tools/create_spreadsheet.yaml b/api/core/tools/provider/builtin/feishu_spreadsheet/tools/create_spreadsheet.yaml
new file mode 100644
index 0000000000000..931310e63172d
--- /dev/null
+++ b/api/core/tools/provider/builtin/feishu_spreadsheet/tools/create_spreadsheet.yaml
@@ -0,0 +1,35 @@
+identity:
+  name: create_spreadsheet
+  author: Doug Lea
+  label:
+    en_US: Create Spreadsheet
+    zh_Hans: 创建电子表格
+description:
+  human:
+    en_US: Create Spreadsheet
+    zh_Hans: 创建电子表格
+  llm: A tool for creating spreadsheets. (创建电子表格)
+parameters:
+  - name: title
+    type: string
+    required: false
+    label:
+      en_US: Spreadsheet Title
+      zh_Hans: 电子表格标题
+    human_description:
+      en_US: The title of the spreadsheet
+      zh_Hans: 电子表格的标题
+    llm_description: 电子表格的标题
+    form: llm
+
+  - name: folder_token
+    type: string
+    required: false
+    label:
+      en_US: Folder Token
+      zh_Hans: 文件夹 token
+    human_description:
+      en_US: The token of the folder, supports folder URL input, e.g., https://bytedance.larkoffice.com/drive/folder/CxHEf4DCSlNkL2dUTCJcPRgentg
+      zh_Hans: 文件夹 token，支持文件夹 URL 输入，如：https://bytedance.larkoffice.com/drive/folder/CxHEf4DCSlNkL2dUTCJcPRgentg
+    llm_description: 文件夹 token，支持文件夹 URL 输入，如：https://bytedance.larkoffice.com/drive/folder/CxHEf4DCSlNkL2dUTCJcPRgentg
+    form: llm
diff --git a/api/core/tools/provider/builtin/feishu_spreadsheet/tools/get_spreadsheet.py b/api/core/tools/provider/builtin/feishu_spreadsheet/tools/get_spreadsheet.py
new file mode 100644
index 0000000000000..dda8c59daffab
--- /dev/null
+++ b/api/core/tools/provider/builtin/feishu_spreadsheet/tools/get_spreadsheet.py
@@ -0,0 +1,19 @@
+from typing import Any
+
+from core.tools.entities.tool_entities import ToolInvokeMessage
+from core.tools.tool.builtin_tool import BuiltinTool
+from core.tools.utils.feishu_api_utils import FeishuRequest
+
+
+class GetSpreadsheetTool(BuiltinTool):
+    def _invoke(self, user_id: str, tool_parameters: dict[str, Any]) -> ToolInvokeMessage:
+        app_id = self.runtime.credentials.get("app_id")
+        app_secret = self.runtime.credentials.get("app_secret")
+        client = FeishuRequest(app_id, app_secret)
+
+        spreadsheet_token = tool_parameters.get("spreadsheet_token")
+        user_id_type = tool_parameters.get("user_id_type", "open_id")
+
+        res = client.get_spreadsheet(spreadsheet_token, user_id_type)
+
+        return self.create_json_message(res)
diff --git a/api/core/tools/provider/builtin/feishu_spreadsheet/tools/get_spreadsheet.yaml b/api/core/tools/provider/builtin/feishu_spreadsheet/tools/get_spreadsheet.yaml
new file mode 100644
index 0000000000000..c519938617ba8
--- /dev/null
+++ b/api/core/tools/provider/builtin/feishu_spreadsheet/tools/get_spreadsheet.yaml
@@ -0,0 +1,49 @@
+identity:
+  name: get_spreadsheet
+  author: Doug Lea
+  label:
+    en_US: Get Spreadsheet
+    zh_Hans: 获取电子表格信息
+description:
+  human:
+    en_US: Get Spreadsheet
+    zh_Hans: 获取电子表格信息
+  llm: A tool for getting information from spreadsheets. (获取电子表格信息)
+parameters:
+  - name: spreadsheet_token
+    type: string
+    required: true
+    label:
+      en_US: Spreadsheet Token
+      zh_Hans: 电子表格 token
+    human_description:
+      en_US: Spreadsheet token, supports input of spreadsheet URL.
+      zh_Hans: 电子表格 token，支持输入电子表格 URL。
+    llm_description: 电子表格 token，支持输入电子表格 URL。
+    form: llm
+
+  - name: user_id_type
+    type: select
+    required: false
+    options:
+      - value: open_id
+        label:
+          en_US: open_id
+          zh_Hans: open_id
+      - value: union_id
+        label:
+          en_US: union_id
+          zh_Hans: union_id
+      - value: user_id
+        label:
+          en_US: user_id
+          zh_Hans: user_id
+    default: "open_id"
+    label:
+      en_US: user_id_type
+      zh_Hans: 用户 ID 类型
+    human_description:
+      en_US: User ID type, optional values are open_id, union_id, user_id, with a default value of open_id.
+      zh_Hans: 用户 ID 类型，可选值有 open_id、union_id、user_id，默认值为 open_id。
+    llm_description: 用户 ID 类型，可选值有 open_id、union_id、user_id，默认值为 open_id。
+    form: form
diff --git a/api/core/tools/provider/builtin/feishu_spreadsheet/tools/list_spreadsheet_sheets.py b/api/core/tools/provider/builtin/feishu_spreadsheet/tools/list_spreadsheet_sheets.py
new file mode 100644
index 0000000000000..98497791c0fa1
--- /dev/null
+++ b/api/core/tools/provider/builtin/feishu_spreadsheet/tools/list_spreadsheet_sheets.py
@@ -0,0 +1,18 @@
+from typing import Any
+
+from core.tools.entities.tool_entities import ToolInvokeMessage
+from core.tools.tool.builtin_tool import BuiltinTool
+from core.tools.utils.feishu_api_utils import FeishuRequest
+
+
+class ListSpreadsheetSheetsTool(BuiltinTool):
+    def _invoke(self, user_id: str, tool_parameters: dict[str, Any]) -> ToolInvokeMessage:
+        app_id = self.runtime.credentials.get("app_id")
+        app_secret = self.runtime.credentials.get("app_secret")
+        client = FeishuRequest(app_id, app_secret)
+
+        spreadsheet_token = tool_parameters.get("spreadsheet_token")
+
+        res = client.list_spreadsheet_sheets(spreadsheet_token)
+
+        return self.create_json_message(res)
diff --git a/api/core/tools/provider/builtin/feishu_spreadsheet/tools/list_spreadsheet_sheets.yaml b/api/core/tools/provider/builtin/feishu_spreadsheet/tools/list_spreadsheet_sheets.yaml
new file mode 100644
index 0000000000000..c6a7ef45d4658
--- /dev/null
+++ b/api/core/tools/provider/builtin/feishu_spreadsheet/tools/list_spreadsheet_sheets.yaml
@@ -0,0 +1,23 @@
+identity:
+  name: list_spreadsheet_sheets
+  author: Doug Lea
+  label:
+    en_US: List Spreadsheet Sheets
+    zh_Hans: 列出电子表格所有工作表
+description:
+  human:
+    en_US: List Spreadsheet Sheets
+    zh_Hans: 列出电子表格所有工作表
+  llm: A tool for listing all sheets in a spreadsheet. (列出电子表格所有工作表)
+parameters:
+  - name: spreadsheet_token
+    type: string
+    required: true
+    label:
+      en_US: Spreadsheet Token
+      zh_Hans: 电子表格 token
+    human_description:
+      en_US: Spreadsheet token, supports input of spreadsheet URL.
+      zh_Hans: 电子表格 token，支持输入电子表格 URL。
+    llm_description: 电子表格 token，支持输入电子表格 URL。
+    form: llm
diff --git a/api/core/tools/provider/builtin/feishu_spreadsheet/tools/read_cols.py b/api/core/tools/provider/builtin/feishu_spreadsheet/tools/read_cols.py
new file mode 100644
index 0000000000000..ebe3f619d091d
--- /dev/null
+++ b/api/core/tools/provider/builtin/feishu_spreadsheet/tools/read_cols.py
@@ -0,0 +1,23 @@
+from typing import Any
+
+from core.tools.entities.tool_entities import ToolInvokeMessage
+from core.tools.tool.builtin_tool import BuiltinTool
+from core.tools.utils.feishu_api_utils import FeishuRequest
+
+
+class ReadColsTool(BuiltinTool):
+    def _invoke(self, user_id: str, tool_parameters: dict[str, Any]) -> ToolInvokeMessage:
+        app_id = self.runtime.credentials.get("app_id")
+        app_secret = self.runtime.credentials.get("app_secret")
+        client = FeishuRequest(app_id, app_secret)
+
+        spreadsheet_token = tool_parameters.get("spreadsheet_token")
+        sheet_id = tool_parameters.get("sheet_id")
+        sheet_name = tool_parameters.get("sheet_name")
+        start_col = tool_parameters.get("start_col")
+        num_cols = tool_parameters.get("num_cols")
+        user_id_type = tool_parameters.get("user_id_type", "open_id")
+
+        res = client.read_cols(spreadsheet_token, sheet_id, sheet_name, start_col, num_cols, user_id_type)
+
+        return self.create_json_message(res)
diff --git a/api/core/tools/provider/builtin/feishu_spreadsheet/tools/read_cols.yaml b/api/core/tools/provider/builtin/feishu_spreadsheet/tools/read_cols.yaml
new file mode 100644
index 0000000000000..3273857b709bf
--- /dev/null
+++ b/api/core/tools/provider/builtin/feishu_spreadsheet/tools/read_cols.yaml
@@ -0,0 +1,97 @@
+identity:
+  name: read_cols
+  author: Doug Lea
+  label:
+    en_US: Read Cols
+    zh_Hans: 读取工作表列数据
+description:
+  human:
+    en_US: Read Cols
+    zh_Hans: 读取工作表列数据
+  llm: A tool for reading column data from a spreadsheet. (读取工作表列数据)
+parameters:
+  - name: spreadsheet_token
+    type: string
+    required: true
+    label:
+      en_US: spreadsheet_token
+      zh_Hans: 电子表格 token
+    human_description:
+      en_US: Spreadsheet token, supports input of spreadsheet URL.
+      zh_Hans: 电子表格 token，支持输入电子表格 url。
+    llm_description: 电子表格 token，支持输入电子表格 url。
+    form: llm
+
+  - name: sheet_id
+    type: string
+    required: false
+    label:
+      en_US: sheet_id
+      zh_Hans: 工作表 ID
+    human_description:
+      en_US: Sheet ID, either sheet_id or sheet_name must be filled.
+      zh_Hans: 工作表 ID，与 sheet_name 二者其一必填。
+    llm_description: 工作表 ID，与 sheet_name 二者其一必填。
+    form: llm
+
+  - name: sheet_name
+    type: string
+    required: false
+    label:
+      en_US: sheet_name
+      zh_Hans: 工作表名称
+    human_description:
+      en_US: Sheet name, either sheet_id or sheet_name must be filled.
+      zh_Hans: 工作表名称，与 sheet_id 二者其一必填。
+    llm_description: 工作表名称，与 sheet_id 二者其一必填。
+    form: llm
+
+  - name: user_id_type
+    type: select
+    required: false
+    options:
+      - value: open_id
+        label:
+          en_US: open_id
+          zh_Hans: open_id
+      - value: union_id
+        label:
+          en_US: union_id
+          zh_Hans: union_id
+      - value: user_id
+        label:
+          en_US: user_id
+          zh_Hans: user_id
+    default: "open_id"
+    label:
+      en_US: user_id_type
+      zh_Hans: 用户 ID 类型
+    human_description:
+      en_US: User ID type, optional values are open_id, union_id, user_id, with a default value of open_id.
+      zh_Hans: 用户 ID 类型，可选值有 open_id、union_id、user_id，默认值为 open_id。
+    llm_description: 用户 ID 类型，可选值有 open_id、union_id、user_id，默认值为 open_id。
+    form: form
+
+  - name: start_col
+    type: number
+    required: false
+    label:
+      en_US: start_col
+      zh_Hans: 起始列号
+    human_description:
+      en_US: Starting column number, starting from 1.
+      zh_Hans: 起始列号，从 1 开始。
+    llm_description: 起始列号，从 1 开始。
+    form: llm
+
+  - name: num_cols
+    type: number
+    required: true
+    label:
+      en_US: num_cols
+      zh_Hans: 读取列数
+    human_description:
+      en_US: Number of columns to read.
+      zh_Hans: 读取列数
+    llm_description: 读取列数
+    form: llm
diff --git a/api/core/tools/provider/builtin/feishu_spreadsheet/tools/read_rows.py b/api/core/tools/provider/builtin/feishu_spreadsheet/tools/read_rows.py
new file mode 100644
index 0000000000000..86b91b104b702
--- /dev/null
+++ b/api/core/tools/provider/builtin/feishu_spreadsheet/tools/read_rows.py
@@ -0,0 +1,23 @@
+from typing import Any
+
+from core.tools.entities.tool_entities import ToolInvokeMessage
+from core.tools.tool.builtin_tool import BuiltinTool
+from core.tools.utils.feishu_api_utils import FeishuRequest
+
+
+class ReadRowsTool(BuiltinTool):
+    def _invoke(self, user_id: str, tool_parameters: dict[str, Any]) -> ToolInvokeMessage:
+        app_id = self.runtime.credentials.get("app_id")
+        app_secret = self.runtime.credentials.get("app_secret")
+        client = FeishuRequest(app_id, app_secret)
+
+        spreadsheet_token = tool_parameters.get("spreadsheet_token")
+        sheet_id = tool_parameters.get("sheet_id")
+        sheet_name = tool_parameters.get("sheet_name")
+        start_row = tool_parameters.get("start_row")
+        num_rows = tool_parameters.get("num_rows")
+        user_id_type = tool_parameters.get("user_id_type", "open_id")
+
+        res = client.read_rows(spreadsheet_token, sheet_id, sheet_name, start_row, num_rows, user_id_type)
+
+        return self.create_json_message(res)
diff --git a/api/core/tools/provider/builtin/feishu_spreadsheet/tools/read_rows.yaml b/api/core/tools/provider/builtin/feishu_spreadsheet/tools/read_rows.yaml
new file mode 100644
index 0000000000000..3e9206e8ef70f
--- /dev/null
+++ b/api/core/tools/provider/builtin/feishu_spreadsheet/tools/read_rows.yaml
@@ -0,0 +1,97 @@
+identity:
+  name: read_rows
+  author: Doug Lea
+  label:
+    en_US: Read Rows
+    zh_Hans: 读取工作表行数据
+description:
+  human:
+    en_US: Read Rows
+    zh_Hans: 读取工作表行数据
+  llm: A tool for reading row data from a spreadsheet. (读取工作表行数据)
+parameters:
+  - name: spreadsheet_token
+    type: string
+    required: true
+    label:
+      en_US: spreadsheet_token
+      zh_Hans: 电子表格 token
+    human_description:
+      en_US: Spreadsheet token, supports input of spreadsheet URL.
+      zh_Hans: 电子表格 token，支持输入电子表格 url。
+    llm_description: 电子表格 token，支持输入电子表格 url。
+    form: llm
+
+  - name: sheet_id
+    type: string
+    required: false
+    label:
+      en_US: sheet_id
+      zh_Hans: 工作表 ID
+    human_description:
+      en_US: Sheet ID, either sheet_id or sheet_name must be filled.
+      zh_Hans: 工作表 ID，与 sheet_name 二者其一必填。
+    llm_description: 工作表 ID，与 sheet_name 二者其一必填。
+    form: llm
+
+  - name: sheet_name
+    type: string
+    required: false
+    label:
+      en_US: sheet_name
+      zh_Hans: 工作表名称
+    human_description:
+      en_US: Sheet name, either sheet_id or sheet_name must be filled.
+      zh_Hans: 工作表名称，与 sheet_id 二者其一必填。
+    llm_description: 工作表名称，与 sheet_id 二者其一必填。
+    form: llm
+
+  - name: user_id_type
+    type: select
+    required: false
+    options:
+      - value: open_id
+        label:
+          en_US: open_id
+          zh_Hans: open_id
+      - value: union_id
+        label:
+          en_US: union_id
+          zh_Hans: union_id
+      - value: user_id
+        label:
+          en_US: user_id
+          zh_Hans: user_id
+    default: "open_id"
+    label:
+      en_US: user_id_type
+      zh_Hans: 用户 ID 类型
+    human_description:
+      en_US: User ID type, optional values are open_id, union_id, user_id, with a default value of open_id.
+      zh_Hans: 用户 ID 类型，可选值有 open_id、union_id、user_id，默认值为 open_id。
+    llm_description: 用户 ID 类型，可选值有 open_id、union_id、user_id，默认值为 open_id。
+    form: form
+
+  - name: start_row
+    type: number
+    required: false
+    label:
+      en_US: start_row
+      zh_Hans: 起始行号
+    human_description:
+      en_US: Starting row number, starting from 1.
+      zh_Hans: 起始行号，从 1 开始。
+    llm_description: 起始行号，从 1 开始。
+    form: llm
+
+  - name: num_rows
+    type: number
+    required: true
+    label:
+      en_US: num_rows
+      zh_Hans: 读取行数
+    human_description:
+      en_US: Number of rows to read.
+      zh_Hans: 读取行数
+    llm_description: 读取行数
+    form: llm
diff --git a/api/core/tools/provider/builtin/feishu_spreadsheet/tools/read_table.py b/api/core/tools/provider/builtin/feishu_spreadsheet/tools/read_table.py
new file mode 100644
index 0000000000000..ddd607d87838f
--- /dev/null
+++ b/api/core/tools/provider/builtin/feishu_spreadsheet/tools/read_table.py
@@ -0,0 +1,23 @@
+from typing import Any
+
+from core.tools.entities.tool_entities import ToolInvokeMessage
+from core.tools.tool.builtin_tool import BuiltinTool
+from core.tools.utils.feishu_api_utils import FeishuRequest
+
+
+class ReadTableTool(BuiltinTool):
+    def _invoke(self, user_id: str, tool_parameters: dict[str, Any]) -> ToolInvokeMessage:
+        app_id = self.runtime.credentials.get("app_id")
+        app_secret = self.runtime.credentials.get("app_secret")
+        client = FeishuRequest(app_id, app_secret)
+
+        spreadsheet_token = tool_parameters.get("spreadsheet_token")
+        sheet_id = tool_parameters.get("sheet_id")
+        sheet_name = tool_parameters.get("sheet_name")
+        num_range = tool_parameters.get("num_range")
+        query = tool_parameters.get("query")
+        user_id_type = tool_parameters.get("user_id_type", "open_id")
+
+        res = client.read_table(spreadsheet_token, sheet_id, sheet_name, num_range, query, user_id_type)
+
+        return self.create_json_message(res)
diff --git a/api/core/tools/provider/builtin/feishu_spreadsheet/tools/read_table.yaml b/api/core/tools/provider/builtin/feishu_spreadsheet/tools/read_table.yaml
new file mode 100644
index 0000000000000..e3dc80e1eb0b4
--- /dev/null
+++ b/api/core/tools/provider/builtin/feishu_spreadsheet/tools/read_table.yaml
@@ -0,0 +1,122 @@
+identity:
+  name: read_table
+  author: Doug Lea
+  label:
+    en_US: Read Table
+    zh_Hans: 自定义读取电子表格行列数据
+description:
+  human:
+    en_US: Read Table
+    zh_Hans: 自定义读取电子表格行列数据
+  llm: A tool for custom reading of row and column data from a spreadsheet. (自定义读取电子表格行列数据)
+parameters:
+  - name: spreadsheet_token
+    type: string
+    required: true
+    label:
+      en_US: spreadsheet_token
+      zh_Hans: 电子表格 token
+    human_description:
+      en_US: Spreadsheet token, supports input of spreadsheet URL.
+      zh_Hans: 电子表格 token，支持输入电子表格 url。
+    llm_description: 电子表格 token，支持输入电子表格 url。
+    form: llm
+
+  - name: sheet_id
+    type: string
+    required: false
+    label:
+      en_US: sheet_id
+      zh_Hans: 工作表 ID
+    human_description:
+      en_US: Sheet ID, either sheet_id or sheet_name must be filled.
+      zh_Hans: 工作表 ID，与 sheet_name 二者其一必填。
+    llm_description: 工作表 ID，与 sheet_name 二者其一必填。
+    form: llm
+
+  - name: sheet_name
+    type: string
+    required: false
+    label:
+      en_US: sheet_name
+      zh_Hans: 工作表名称
+    human_description:
+      en_US: Sheet name, either sheet_id or sheet_name must be filled.
+      zh_Hans: 工作表名称，与 sheet_id 二者其一必填。
+    llm_description: 工作表名称，与 sheet_id 二者其一必填。
+    form: llm
+
+  - name: user_id_type
+    type: select
+    required: false
+    options:
+      - value: open_id
+        label:
+          en_US: open_id
+          zh_Hans: open_id
+      - value: union_id
+        label:
+          en_US: union_id
+          zh_Hans: union_id
+      - value: user_id
+        label:
+          en_US: user_id
+          zh_Hans: user_id
+    default: "open_id"
+    label:
+      en_US: user_id_type
+      zh_Hans: 用户 ID 类型
+    human_description:
+      en_US: User ID type, optional values are open_id, union_id, user_id, with a default value of open_id.
+      zh_Hans: 用户 ID 类型，可选值有 open_id、union_id、user_id，默认值为 open_id。
+    llm_description: 用户 ID 类型，可选值有 open_id、union_id、user_id，默认值为 open_id。
+    form: form
+
+  - name: start_row
+    type: number
+    required: false
+    label:
+      en_US: start_row
+      zh_Hans: 起始行号
+    human_description:
+      en_US: Starting row number, starting from 1.
+      zh_Hans: 起始行号，从 1 开始。
+    llm_description: 起始行号，从 1 开始。
+    form: llm
+
+  - name: num_rows
+    type: number
+    required: false
+    label:
+      en_US: num_rows
+      zh_Hans: 读取行数
+    human_description:
+      en_US: Number of rows to read.
+      zh_Hans: 读取行数
+    llm_description: 读取行数
+    form: llm
+
+  - name: range
+    type: string
+    required: false
+    label:
+      en_US: range
+      zh_Hans: 取数范围
+    human_description:
+      en_US: |
+        Data range, format like: A1:B2, can be empty when query=all.
+      zh_Hans: 取数范围,格式如:A1:B2，query=all 时可为空。
+    llm_description: 取数范围,格式如:A1:B2，query=all 时可为空。
+    form: llm
+
+  - name: query
+    type: string
+    required: false
+    label:
+      en_US: query
+      zh_Hans: 查询
+    human_description:
+      en_US: Pass "all" to query all data in the table, but no more than 100 columns.
+      zh_Hans: 传 all，表示查询表格所有数据，但最多查询 100 列数据。
+    llm_description: 传 all，表示查询表格所有数据，但最多查询 100 列数据。
+    form: llm
diff --git a/api/core/tools/provider/builtin/feishu_task/_assets/icon.png b/api/core/tools/provider/builtin/feishu_task/_assets/icon.png
new file mode 100644
index 0000000000000000000000000000000000000000..3485be0d0fbd85444995dca78b51634899807537
GIT binary patch
literal 3917
zcmeHK`8!nY|35Rv$S@chMwYTYii(M{N3v9;%}$eNRLVAE$sB}1Y7mMPO;kkLL)K0@
zG`2^jO=OvlD0z^vt1O@MT+jC(`2P60uFntW+}CyA@7L@7y3Tno_dD^ZgDsJ;i2wiq
z@!$cg<LD~*=h=Wq-+P_L)6oSJa@^JuD1WUu4giuO2d(xwg0!h2^Q(@%<o1~mAGXqf
ztGlYTYhr-~`PPll+!l$gM6pB6J8gxkL+|z}ZmZ6>+S5TCNFY4gu9+HkUNu91z`Dra
zn3cS--JiTCJttvR&menYC;#MG7C&+c`aK=BZR+>*>g%6^Z=3aOmiJ@TulZyD_>>XC
zl-`2D=pMGfifRJLrR4Uv2LX5<5LLeeuw`~C+iGHf_d@E|m{`D}9#qW10rryDELubX
z!{IMMq0)e^*?%Q)LHc``oXM3Y*0N()fBE&rG+tIJ`KiB1yhwnX$8zY}=>BIz^Um1E
zw40?jys$$>?HMvvv*na*Sv5@?T3lv3{Uy}*adNd`M3UZU541ZHUw_#hYw)ni^o;vd
ziE7wioKS-QS~fY-d*O4*lB}%5-fg-~iz-058GR3j8f#Ffd||*Qg6?1U;b*DoroI#f
zzpjZox`jA$yvLg<A?n!nG-`&Vf@ex7b?)Bi0LwE@b1<Iu^WyeqnBCAn2Y;XJ{3)h!
zFV#Sr2kdl=#uwbN2msYkf*U3o)oF+Cg+Z|-A<$n;cS{cr8YXfma8a1%*j`=CG$^Ek
z*OjWW+d|<39L$}QzxVu(iyK+EN`YVfSSl^wy9rVz2^p5i2L5QZy3#8{2Mp@QsvBm{
z%|xK={{8jYa?)n}+a#g-?5Mk~PMYy`Hay^UUM81C>KrZ;rLwubzdKI_Re^q}`RG7{
zo7^>1+8G`Y=)A5ORr3jPvJyugcr+Sa>6WOFM#<YPs$6dGpb7H;huSi3kYIAgSQXFI
zQ&bPnOlh8AQ{aQOVQW_c75Pc1*yjnO<&P^t?Jam_@KnxHtU_9yIvp6BQYk7CnhyEu
z{v_<p`ov%fYe`$3o#8ukQBsT_mAE!U8Cak1Hm^7L#j)phUw@eyy96J1BSrR#JO9p)
z&ZoegPr`FWgc<i6fBZhW2~L>xKB^?ju45K|bVvW+jBm{Jk=xjM=4J5@_5f7yZgA;O
zeYV}#4zCm(>|t!}*UGOJ8dx7#DY-j+hlsp$+tsXr5jY=`gfr>h>AqN`Oi@$vzP4?<
zRVEIe*qBeg{^?=-!_|bJ8C|BA!v0F{JR<`zKJjuDx(5!tCRJG!psd?!b23Uv*>$x;
z&&^pY&HNsa)Ma!^woeBmD4uWgp$Dw54N6X@xD#P5`RWcn4la2k<E-rA=i_qbLfCZ3
z=cyh{=`}sONe2KTC>;Z)nR0;k=~vR9H2_*d;8u<tOmM5lfliV4aQsnB_sPl(g@Eug
zpvn}{%{b=$WJH?%ZdX|oUCa-NKs+wZNN1Zjo;{)%*x3+U`CRJy37ZGiN;x&NZzv|8
z18aVGF6V^*&|oES+~`(V#OYiz0nET&eClJOQ_<z&Jh__`9$<gE%Vu&jzC^oP&fK)|
zfs~HjWCj3|-SbuXr-AN;$<beJNRq4G4%G~6wQA>Yk~OcyStEIgnF=EO;P4d1DYk(q
z>@@QObS2Ug_6|J7fn>`IJOwi^7if#~xm~$)A7%2e{6`qSVd2Wa^`J4QjuWDcC+wP;
zIM@5*Ryc@Yr1d35nt2l3H~H}*a-oY6QoD=)9q3Jw<MrH@f_j1nc9-CxW0=JRIo{XX
zB<SnxCZonbptryxYSDUI8fpzTW;&*T`G6upha(y2AZSZ7Fht&f`%D`THgoa(du;D!
zu`fff2>eh1UPw`aCKba4!1pZ*IA$rxO{GBN!wV9IVbEb~alZacgzy~#zfdK5ZY96h
zgYLsS;Br5O+hFyIK~~yWd`qXsj>tG674I@U&1Hkn24hdOa!u%h!nC?DxM<`RnF`Lq
z9@>bf;M-*i^b;u~G6TVoed2sb@DXe10u`y@8PQ-5W-QgfiHE$Bsn9|Sk;6n3)mnnD
zKM$Vx$8@F`2-M>Dk>U+vpx10}s(~{P3Bc1dVPt4Xy>jKM7|4yU+vO!l4U|?}eBIyX
zyFN9cW9oP=|D7GXF7$Nwkqs-KzE!y#9p9{Xab;qm$ho9PVGZ*6MAX~-eEUbw^<UcA
z&qCWxXTr-6rvoNbDDwx*WqrOdp4g{Nxc;@C*Yc~)2Bo|rck!emNKSW<VralIP;HoV
z|5Pr?s7%Tu)o#0j_u1&hNWF=XFL24V;5<+pvo>n5m=jy0*VZAL@O{_Gmcf-@3gf35
z&)=MYECgB)xVrp9?DhHFXTB*pR8^&f<topX_GC9SB6D~%&e)ZiNCmwe-j{W}W_<4A
zE;F_zmM-Zu#f;7wTVg9Li{+6rT$YJo`CHMPggAfZXPVM=QIs!S!nD@R>Ot-pjO<8X
z&7#T=m*JN4{rraZPb~^m^gq)L^IS?h(&e+V@5XoqZhJCNt%1rtdlr@3G*q2)tGB?X
zm3D%o|5W4e<pUZZxnF%hlnF>`M$LLeo_$>T^N{|u@bvc`eqSp&Q>QLIFI!aBJ=G||
zkC%Qlf9$W}lgq6&%}+=nJrtutsUQhES5jnGdHK9fG#GktBo1{%FJ$qjPHSP59h#zA
z9h&;&?YTFz@s+NAqq*Yqd}FK*n5=$OUYB#XRgABi^DWZ;8;dq(B3Q%{pUcGe8oX5f
zy&K2)Dp$F9w(lqp*%8;dl!YI55*mcByhj#@)g95q7bS%HeVk?wT*fq&$SaYk$8ee_
z6wY2>-IP=(3(KR?MzU{e*QYQ7*m_ptw9mT9lf6u+2K#yg6-@S;HN!*PM3bm%`3GJ}
z$0UnF)=86*;^18knhA1gn)j~{WGFLT4i;7Ymx{gAz<Y*qK;cm@wpVQiZ3{=qOzi(D
zK!yy0rrByX|2h)k3!NZf4K4pwP7$OYHl-jOsd$z@7e@j3miI-36)|*ytglRWJnUg8
zK-fa4li@uxhEh;LODdKLpvay-a`EwXp>NP^=TM-60p1Ioz!|L=u(DKq-Er;*Q&A{W
zC{==)f=XyO?0aVe?IANC=PC4s>^Idu5w_Z<W1h9#vaPbL^CF|#G6}gLL3M|dI4-r(
z)M5PdEM!n8LX7RA(e(&}cCCs-+^srTT8%}%$RwKr`3B_GOM&DU^MVvb_lHVr_xg`%
z`<(WwNE%vk;;pXDRyvQ~cQyI_D#nAQGo;_O$p<m2KwUU(2wIXpI2g-P4xiJGprwbj
z^|P!-Z%}tNEUe!ZAHF@D>=fVglyNF%_`b6{nt{>~L1piAt6(<k&*<JT+OH;i<>+I^
zs#O=8Dnjap8bm<yFA3Bnm8N<gt|@5ou#R{e=c3a^(#C*y!G<}lc_p;qnu4N0T89Wk
zh!UYCNVV%=5HtB_L4S^idiWpV<tlC7tY-5EKx3p}5T#?RhssABvNGkVyQa#p4FNb>
z1zA{-S|1a<_SVE^Qc56N;nZ<PPY_X|&aP9PUJIScOiWD7rTolX(i(lN0>(R%V2{eX
zM(f?xAD=|nycC?Et!zUZM%RM^V_DcEXjafbKNVX?XBxi;pv)1jTvOwqbT;X_OK4sI
ztc4J1rvagQcbT|4_rSlZ0ccg~SIavO)p&We+1tTwKX-`m&-oINvg1(DaM$~Q^9xp&
zB$<jIa*-7Lv|tOsk5(vp+GiB(pXYPcGr8E3s)UEAPTq6d0sgUcy|MKRuajO<ydA7v
zir$D|8Q%$42RFke^l|huDbt>6pPW@fA9cM!77N-cxP0K*n_Pe|^Ijn1W@{|tVm(?#
z3cr6e$w(5*_|PLP;LNf??TrGPlv`u4$c=|HoMr_x8j!Z_m^WIOMIm74(ZV(rJR`Q8
z<FlbVR_sF_6%E0A3STy1k+_8pWhy8ZX6g9?@XS;GgobtEPLAJJw%tcH6EvtlVS<Pv
z%!G*3@f3($yt#KH;8JQki^3~sgANNl-fxy}El#^q+`gL(r!E|jpn^);%Dth0_6Ie2
z9^&*U%+n5VEmD4~hOanYFyfAflNLC#RFGVv*n1psF1HQkG?#b|*WVOk5*IjlD(Lqp
z(9;fEUl@V273bg9Qs}J$$}5$d(FAHMZmZE}`;}i|@sP5Hyr3PDUXlv}v~wvJ%lFCx
zi;ijir1;(w8YBI9Sb5@@0$mVMO84}@^o5vxKo3HMNBSjT%-{D>1g2Q7_a-3WDCxsP
zh>i{QwZcp|!qG;Y&vr`o^8-Z6ZCSoSvmC|jJ#F@K%lm59*TFnfMSt<Y(p8_SA1y7N
znz3o!i?K2Q6TynttuMNeBrmi+*_oW#+nUv>WWH)_rA<F`@M?}+<oA#N-0M{h_t<i_
zU;x{rmvU?kO=>{;ehF3z)m(XvrB-%^G}rq6q4v7#a#v?TM`~9P7M-17VXsJOCOSRs
l9H-vG{cq?0|1|{WeK*e(NZd%(N9SYUptXZlxusX!e*o|W00aO4

literal 0
HcmV?d00001

diff --git a/api/core/tools/provider/builtin/feishu_task/feishu_task.py b/api/core/tools/provider/builtin/feishu_task/feishu_task.py
new file mode 100644
index 0000000000000..6df05968d8f17
--- /dev/null
+++ b/api/core/tools/provider/builtin/feishu_task/feishu_task.py
@@ -0,0 +1,7 @@
+from core.tools.provider.builtin_tool_provider import BuiltinToolProviderController
+from core.tools.utils.feishu_api_utils import auth
+
+
+class FeishuTaskProvider(BuiltinToolProviderController):
+    def _validate_credentials(self, credentials: dict) -> None:
+        auth(credentials)
diff --git a/api/core/tools/provider/builtin/feishu_task/feishu_task.yaml b/api/core/tools/provider/builtin/feishu_task/feishu_task.yaml
new file mode 100644
index 0000000000000..88736f79a02e8
--- /dev/null
+++ b/api/core/tools/provider/builtin/feishu_task/feishu_task.yaml
@@ -0,0 +1,36 @@
+identity:
+  author: Doug Lea
+  name: feishu_task
+  label:
+    en_US: Feishu Task
+    zh_Hans: 飞书任务
+  description:
+    en_US: |
+      Feishu Task, requires the following permissions: task:task:write、contact:user.id:readonly.
+    zh_Hans: |
+      飞书任务，需要开通以下权限: task:task:write、contact:user.id:readonly。
+  icon: icon.png
+  tags:
+    - social
+    - productivity
+credentials_for_provider:
+  app_id:
+    type: text-input
+    required: true
+    label:
+      en_US: APP ID
+    placeholder:
+      en_US: Please input your feishu app id
+      zh_Hans: 请输入你的飞书 app id
+    help:
+      en_US: Get your app_id and app_secret from Feishu
+      zh_Hans: 从飞书获取您的 app_id 和 app_secret
+    url: https://open.larkoffice.com/app
+  app_secret:
+    type: secret-input
+    required: true
+    label:
+      en_US: APP Secret
+    placeholder:
+      en_US: Please input your app secret
+      zh_Hans: 请输入你的飞书 app secret
diff --git a/api/core/tools/provider/builtin/feishu_task/tools/add_members.py b/api/core/tools/provider/builtin/feishu_task/tools/add_members.py
new file mode 100644
index 0000000000000..e58ed22e0f479
--- /dev/null
+++ b/api/core/tools/provider/builtin/feishu_task/tools/add_members.py
@@ -0,0 +1,20 @@
+from typing import Any
+
+from core.tools.entities.tool_entities import ToolInvokeMessage
+from core.tools.tool.builtin_tool import BuiltinTool
+from core.tools.utils.feishu_api_utils import FeishuRequest
+
+
+class AddMembersTool(BuiltinTool):
+    def _invoke(self, user_id: str, tool_parameters: dict[str, Any]) -> ToolInvokeMessage:
+        app_id = self.runtime.credentials.get("app_id")
+        app_secret = self.runtime.credentials.get("app_secret")
+        client = FeishuRequest(app_id, app_secret)
+
+        task_guid = tool_parameters.get("task_guid")
+        member_phone_or_email = tool_parameters.get("member_phone_or_email")
+        member_role = tool_parameters.get("member_role", "follower")
+
+        res = client.add_members(task_guid, member_phone_or_email, member_role)
+
+        return self.create_json_message(res)
diff --git a/api/core/tools/provider/builtin/feishu_task/tools/add_members.yaml b/api/core/tools/provider/builtin/feishu_task/tools/add_members.yaml
new file mode 100644
index 0000000000000..063c0f7f04956
--- /dev/null
+++ b/api/core/tools/provider/builtin/feishu_task/tools/add_members.yaml
@@ -0,0 +1,58 @@
+identity:
+  name: add_members
+  author: Doug Lea
+  label:
+    en_US: Add Members
+    zh_Hans: 添加任务成员
+description:
+  human:
+    en_US: Add Members
+    zh_Hans: 添加任务成员
+  llm: A tool for adding members to a Feishu task.(添加任务成员)
+parameters:
+  - name: task_guid
+    type: string
+    required: true
+    label:
+      en_US: Task GUID
+      zh_Hans: 任务 GUID
+    human_description:
+      en_US: |
+        The GUID of the task to be added, supports passing either the Task ID or the Task link URL. Example of Task ID: 8b5425ec-9f2a-43bd-a3ab-01912f50282b; Example of Task link URL: https://applink.feishu-pre.net/client/todo/detail?guid=8c6bf822-e4da-449a-b82a-dc44020f9be9&suite_entity_num=t21587362
+      zh_Hans: 要添加的任务的 GUID，支持传任务 ID 和任务链接 URL。任务 ID 示例：8b5425ec-9f2a-43bd-a3ab-01912f50282b；任务链接 URL 示例：https://applink.feishu-pre.net/client/todo/detail?guid=8c6bf822-e4da-449a-b82a-dc44020f9be9&suite_entity_num=t21587362
+    llm_description: 要添加的任务的 GUID，支持传任务 ID 和任务链接 URL。任务 ID 示例：8b5425ec-9f2a-43bd-a3ab-01912f50282b；任务链接 URL 示例：https://applink.feishu-pre.net/client/todo/detail?guid=8c6bf822-e4da-449a-b82a-dc44020f9be9&suite_entity_num=t21587362
+    form: llm
+
+  - name: member_phone_or_email
+    type: string
+    required: true
+    label:
+      en_US: Task Member Phone Or Email
+      zh_Hans: 任务成员的电话或邮箱
+    human_description:
+      en_US: A list of member emails or phone numbers, separated by commas.
+      zh_Hans: 任务成员邮箱或者手机号列表，使用逗号分隔。
+    llm_description: 任务成员邮箱或者手机号列表，使用逗号分隔。
+    form: llm
+
+  - name: member_role
+    type: select
+    required: true
+    options:
+      - value: assignee
+        label:
+          en_US: assignee
+          zh_Hans: 负责人
+      - value: follower
+        label:
+          en_US: follower
+          zh_Hans: 关注人
+    default: "follower"
+    label:
+      en_US: member_role
+      zh_Hans: 成员的角色
+    human_description:
+      en_US: Member role, optional values are "assignee" (responsible person) and "follower" (observer), with a default value of "assignee".
+      zh_Hans: 成员的角色，可选值有 "assignee"（负责人）和 "follower"（关注人），默认值为 "assignee"。
+    llm_description: 成员的角色，可选值有 "assignee"（负责人）和 "follower"（关注人），默认值为 "assignee"。
+    form: form
diff --git a/api/core/tools/provider/builtin/feishu_task/tools/create_task.py b/api/core/tools/provider/builtin/feishu_task/tools/create_task.py
new file mode 100644
index 0000000000000..96cdcd71f6d2e
--- /dev/null
+++ b/api/core/tools/provider/builtin/feishu_task/tools/create_task.py
@@ -0,0 +1,22 @@
+from typing import Any
+
+from core.tools.entities.tool_entities import ToolInvokeMessage
+from core.tools.tool.builtin_tool import BuiltinTool
+from core.tools.utils.feishu_api_utils import FeishuRequest
+
+
+class CreateTaskTool(BuiltinTool):
+    def _invoke(self, user_id: str, tool_parameters: dict[str, Any]) -> ToolInvokeMessage:
+        app_id = self.runtime.credentials.get("app_id")
+        app_secret = self.runtime.credentials.get("app_secret")
+        client = FeishuRequest(app_id, app_secret)
+
+        summary = tool_parameters.get("summary")
+        start_time = tool_parameters.get("start_time")
+        end_time = tool_parameters.get("end_time")
+        completed_time = tool_parameters.get("completed_time")
+        description = tool_parameters.get("description")
+
+        res = client.create_task(summary, start_time, end_time, completed_time, description)
+
+        return self.create_json_message(res)
diff --git a/api/core/tools/provider/builtin/feishu_task/tools/create_task.yaml b/api/core/tools/provider/builtin/feishu_task/tools/create_task.yaml
new file mode 100644
index 0000000000000..7eb4af168bf74
--- /dev/null
+++ b/api/core/tools/provider/builtin/feishu_task/tools/create_task.yaml
@@ -0,0 +1,74 @@
+identity:
+  name: create_task
+  author: Doug Lea
+  label:
+    en_US: Create Task
+    zh_Hans: 创建飞书任务
+description:
+  human:
+    en_US: Create Feishu Task
+    zh_Hans: 创建飞书任务
+  llm: A tool for creating tasks in Feishu.(创建飞书任务)
+parameters:
+  - name: summary
+    type: string
+    required: true
+    label:
+      en_US: Task Title
+      zh_Hans: 任务标题
+    human_description:
+      en_US: The title of the task.
+      zh_Hans: 任务标题
+    llm_description: 任务标题
+    form: llm
+
+  - name: description
+    type: string
+    required: false
+    label:
+      en_US: Task Description
+      zh_Hans: 任务备注
+    human_description:
+      en_US: The description or notes for the task.
+      zh_Hans: 任务备注
+    llm_description: 任务备注
+    form: llm
+
+  - name: start_time
+    type: string
+    required: false
+    label:
+      en_US: Start Time
+      zh_Hans: 任务开始时间
+    human_description:
+      en_US: |
+        The start time of the task, in the format: 2006-01-02 15:04:05
+      zh_Hans: 任务开始时间，格式为：2006-01-02 15:04:05
+    llm_description: 任务开始时间，格式为：2006-01-02 15:04:05
+    form: llm
+
+  - name: end_time
+    type: string
+    required: false
+    label:
+      en_US: End Time
+      zh_Hans: 任务结束时间
+    human_description:
+      en_US: |
+        The end time of the task, in the format: 2006-01-02 15:04:05
+      zh_Hans: 任务结束时间，格式为：2006-01-02 15:04:05
+    llm_description: 任务结束时间，格式为：2006-01-02 15:04:05
+    form: llm
+
+  - name: completed_time
+    type: string
+    required: false
+    label:
+      en_US: Completed Time
+      zh_Hans: 任务完成时间
+    human_description:
+      en_US: |
+        The completion time of the task, in the format: 2006-01-02 15:04:05. Leave empty to create an incomplete task; fill in a specific time to create a completed task.
+      zh_Hans: 任务完成时间，格式为：2006-01-02 15:04:05，不填写表示创建一个未完成任务；填写一个具体的时间表示创建一个已完成任务。
+    llm_description: 任务完成时间，格式为：2006-01-02 15:04:05，不填写表示创建一个未完成任务；填写一个具体的时间表示创建一个已完成任务。
+    form: llm
diff --git a/api/core/tools/provider/builtin/feishu_task/tools/delete_task.py b/api/core/tools/provider/builtin/feishu_task/tools/delete_task.py
new file mode 100644
index 0000000000000..dee036fee5203
--- /dev/null
+++ b/api/core/tools/provider/builtin/feishu_task/tools/delete_task.py
@@ -0,0 +1,18 @@
+from typing import Any
+
+from core.tools.entities.tool_entities import ToolInvokeMessage
+from core.tools.tool.builtin_tool import BuiltinTool
+from core.tools.utils.feishu_api_utils import FeishuRequest
+
+
+class UpdateTaskTool(BuiltinTool):
+    def _invoke(self, user_id: str, tool_parameters: dict[str, Any]) -> ToolInvokeMessage:
+        app_id = self.runtime.credentials.get("app_id")
+        app_secret = self.runtime.credentials.get("app_secret")
+        client = FeishuRequest(app_id, app_secret)
+
+        task_guid = tool_parameters.get("task_guid")
+
+        res = client.delete_task(task_guid)
+
+        return self.create_json_message(res)
diff --git a/api/core/tools/provider/builtin/feishu_task/tools/delete_task.yaml b/api/core/tools/provider/builtin/feishu_task/tools/delete_task.yaml
new file mode 100644
index 0000000000000..d3f9741367662
--- /dev/null
+++ b/api/core/tools/provider/builtin/feishu_task/tools/delete_task.yaml
@@ -0,0 +1,24 @@
+identity:
+  name: delete_task
+  author: Doug Lea
+  label:
+    en_US: Delete Task
+    zh_Hans: 删除飞书任务
+description:
+  human:
+    en_US: Delete Task
+    zh_Hans: 删除飞书任务
+  llm: A tool for deleting tasks in Feishu.(删除飞书任务)
+parameters:
+  - name: task_guid
+    type: string
+    required: true
+    label:
+      en_US: Task GUID
+      zh_Hans: 任务 GUID
+    human_description:
+      en_US: |
+        The GUID of the task to be deleted, supports passing either the Task ID or the Task link URL. Example of Task ID: 8b5425ec-9f2a-43bd-a3ab-01912f50282b; Example of Task link URL: https://applink.feishu-pre.net/client/todo/detail?guid=8c6bf822-e4da-449a-b82a-dc44020f9be9&suite_entity_num=t21587362
+      zh_Hans: 要删除的任务的 GUID，支持传任务 ID 和任务链接 URL。任务 ID 示例：8b5425ec-9f2a-43bd-a3ab-01912f50282b；任务链接 URL 示例：https://applink.feishu-pre.net/client/todo/detail?guid=8c6bf822-e4da-449a-b82a-dc44020f9be9&suite_entity_num=t21587362
+    llm_description: 要删除的任务的 GUID，支持传任务 ID 和任务链接 URL。任务 ID 示例：8b5425ec-9f2a-43bd-a3ab-01912f50282b；任务链接 URL 示例：https://applink.feishu-pre.net/client/todo/detail?guid=8c6bf822-e4da-449a-b82a-dc44020f9be9&suite_entity_num=t21587362
+    form: llm
diff --git a/api/core/tools/provider/builtin/feishu_task/tools/update_task.py b/api/core/tools/provider/builtin/feishu_task/tools/update_task.py
new file mode 100644
index 0000000000000..4a48cd283abf1
--- /dev/null
+++ b/api/core/tools/provider/builtin/feishu_task/tools/update_task.py
@@ -0,0 +1,23 @@
+from typing import Any
+
+from core.tools.entities.tool_entities import ToolInvokeMessage
+from core.tools.tool.builtin_tool import BuiltinTool
+from core.tools.utils.feishu_api_utils import FeishuRequest
+
+
+class UpdateTaskTool(BuiltinTool):
+    def _invoke(self, user_id: str, tool_parameters: dict[str, Any]) -> ToolInvokeMessage:
+        app_id = self.runtime.credentials.get("app_id")
+        app_secret = self.runtime.credentials.get("app_secret")
+        client = FeishuRequest(app_id, app_secret)
+
+        task_guid = tool_parameters.get("task_guid")
+        summary = tool_parameters.get("summary")
+        start_time = tool_parameters.get("start_time")
+        end_time = tool_parameters.get("end_time")
+        completed_time = tool_parameters.get("completed_time")
+        description = tool_parameters.get("description")
+
+        res = client.update_task(task_guid, summary, start_time, end_time, completed_time, description)
+
+        return self.create_json_message(res)
diff --git a/api/core/tools/provider/builtin/feishu_task/tools/update_task.yaml b/api/core/tools/provider/builtin/feishu_task/tools/update_task.yaml
new file mode 100644
index 0000000000000..83c9bcb1c443a
--- /dev/null
+++ b/api/core/tools/provider/builtin/feishu_task/tools/update_task.yaml
@@ -0,0 +1,89 @@
+identity:
+  name: update_task
+  author: Doug Lea
+  label:
+    en_US: Update Task
+    zh_Hans: 更新飞书任务
+description:
+  human:
+    en_US: Update Feishu Task
+    zh_Hans: 更新飞书任务
+  llm: A tool for updating tasks in Feishu.(更新飞书任务)
+parameters:
+  - name: task_guid
+    type: string
+    required: true
+    label:
+      en_US: Task GUID
+      zh_Hans: 任务 GUID
+    human_description:
+      en_US: |
+        The task ID, supports inputting either the Task ID or the Task link URL. Example of Task ID: 42cad8a0-f8c8-4344-9be2-d1d7e8e91b64; Example of Task link URL: https://applink.feishu-pre.net/client/todo/detail?guid=42cad8a0-f8c8-4344-9be2-d1d7e8e91b64&suite_entity_num=t21700217
+      zh_Hans: |
+        任务ID，支持传入任务 ID 和任务链接 URL。任务 ID 示例: 42cad8a0-f8c8-4344-9be2-d1d7e8e91b64；任务链接 URL 示例: https://applink.feishu-pre.net/client/todo/detail?guid=42cad8a0-f8c8-4344-9be2-d1d7e8e91b64&suite_entity_num=t21700217
+    llm_description: |
+      任务ID，支持传入任务 ID 和任务链接 URL。任务 ID 示例: 42cad8a0-f8c8-4344-9be2-d1d7e8e91b64；任务链接 URL 示例: https://applink.feishu-pre.net/client/todo/detail?guid=42cad8a0-f8c8-4344-9be2-d1d7e8e91b64&suite_entity_num=t21700217
+    form: llm
+
+  - name: summary
+    type: string
+    required: true
+    label:
+      en_US: Task Title
+      zh_Hans: 任务标题
+    human_description:
+      en_US: The title of the task.
+      zh_Hans: 任务标题
+    llm_description: 任务标题
+    form: llm
+
+  - name: description
+    type: string
+    required: false
+    label:
+      en_US: Task Description
+      zh_Hans: 任务备注
+    human_description:
+      en_US: The description or notes for the task.
+      zh_Hans: 任务备注
+    llm_description: 任务备注
+    form: llm
+
+  - name: start_time
+    type: string
+    required: false
+    label:
+      en_US: Start Time
+      zh_Hans: 任务开始时间
+    human_description:
+      en_US: |
+        The start time of the task, in the format: 2006-01-02 15:04:05
+      zh_Hans: 任务开始时间，格式为：2006-01-02 15:04:05
+    llm_description: 任务开始时间，格式为：2006-01-02 15:04:05
+    form: llm
+
+  - name: end_time
+    type: string
+    required: false
+    label:
+      en_US: End Time
+      zh_Hans: 任务结束时间
+    human_description:
+      en_US: |
+        The end time of the task, in the format: 2006-01-02 15:04:05
+      zh_Hans: 任务结束时间，格式为：2006-01-02 15:04:05
+    llm_description: 任务结束时间，格式为：2006-01-02 15:04:05
+    form: llm
+
+  - name: completed_time
+    type: string
+    required: false
+    label:
+      en_US: Completed Time
+      zh_Hans: 任务完成时间
+    human_description:
+      en_US: |
+        The completion time of the task, in the format: 2006-01-02 15:04:05
+      zh_Hans: 任务完成时间，格式为：2006-01-02 15:04:05
+    llm_description: 任务完成时间，格式为：2006-01-02 15:04:05
+    form: llm
diff --git a/api/core/tools/provider/builtin/feishu_wiki/_assets/icon.png b/api/core/tools/provider/builtin/feishu_wiki/_assets/icon.png
new file mode 100644
index 0000000000000000000000000000000000000000..878672c9ae5a511f246d3211863ec786596481eb
GIT binary patch
literal 4252
zcmeHLdo+|?+t=Mqw_}5n^I?!j4%5*oatPzljD$2|kVK9XVVn&i#W-|O8RUs3$w(3z
z!-NtsJ<f+m3>D3Cnw*9V-s%0`KfZsy_5JmIYrSi|f9!kjeeLV_yRLOz*WPRIJL4=G
zDJmo@#K*@cYHMS8j@Pq)H%Nf@J>|m6<8`~j&mk@NC=7WPA0O<zt);m$9y?!%%Dguq
z)4nsq5J`D6cr_S)HONTsUY;Imq+ISqLsHwpmo+CxpkdKcl+f*iQ6UZf7izDo2f(#G
z%1?S^6i87WGnMwKoO$PV)B05D(bldpr_Y7L=S*kDqD_m!8KDxAl8ji!GPh_(e}=Z$
zuAq^i%tyE^Sbten0|%LzL1!e*l)V9ng!>680MMDe|5vk0p+;|VF?MNmt_+19WtP?o
z)~Kxr3F51S!y1^ShKNEJ&l$Dum@1#!w~h2-|0=A+y}e8huwqgbm*ook6m2hl&(RxM
zxT?>tiDyMbE94GMzUNmcH2I<XTJGF(h5jM-ie7VMcj=ptfmI|uy<%OhD`%9U_%>Ta
zmLUy4C#YX=jWXmE=*(O_$yHr?l-rjk!OZhH3AV86?uc;;#dI07_efFGAHeO7uiFVj
zLtSJqC4cgZ7jE!fi>nPJF`^+Y=PPCGCcW=(Zhu>81A5fsl4Nd;AW*Yp)X)0ig@)_H
z*}vu!DoS;WV^PP$Smvaq-tE+Ah0e+T264jTC@ZR-J>u%-8pFwH76o`h#Fj7XIo1aA
zZIoZpJsNv#IG!{5hk~OS(}=K)jTyJy%vQo7oxb(npvP^as*efX8nLfhEsM8bPWgh@
zE$bIIC<_n|ECd-G+8P2ouzeR@E?&>L@3l}0;aRtqW6CFnnj{^ls+wRnUDG*3x<HM&
zUlCwcL``XO?CuC>zmfv7EwX&VG;%WAAX2PW-o!_+<-HDK6)lfLWE$Ic?}6jGlOCQ5
z1_srwnXfl*|CrVA`gL8FFdNEu-ljMSlOb-fH4RU>0^t`|=aTyi;(HYWuXKcB`3cPr
z#Z{FLE5jBfH@~NwtYM;ucN4Cnp*HdL?EE#IuGys2c3p7;3O+N<Y&c_QAzx_)Umh$E
zjYaZbFtq~4?bD@5F*Q1g94UZ2e5x&oK;K{s^i6S@-_r-jy)3+hE6qe~CD^^*&86cR
z<S&zNB+Qs^KGuXRfoZN~9MLMfx>v%gxTRoAAXvhGGg%6#!^96V2xj$<_cy6k4Sk?_
zgN%~Yk0mYu)@Z8m3fj1b8>|k7)UIe3?I!4%@~~nGmSUBX{<F7^19jLK;NuOm^;@Ik
z)IDR{2bay62d~zgXMk?eM#6;f5Y4c_yJm6+YGCy}5;YeXpumBe`^~2VJ)NTAIHyt!
z83eM5tyOV?!B#|+IOIuNK9{($@x!`FT9Ht2NN4u01@mrWOndS5^hY+*go4LivhQL3
zzrqg#2m0L9Mqvc&;w)=s+Vjq%?_unvn0K+O<MEv|Y}6p+uJ1jKAe!CDc7fxr&0DoZ
zNct~*u2#l<8MwSHA$UAeU6x=Ugb1S)nE4>|c;d3(Vr5*0i^fb4bVI9W(wm6=B2&G4
z2=!pYUM-2Pe!h@Yqmo1)xtnKc_dm{aa_Na{PG6&*RS+ZnoQg=2n7r#&y}lhDc`H5a
zolQdxkLVB?LGG(!+2XUV7R=7SOUnhT#H<oL!xrru{xH0;ZS)Hoa}IsQfIwY4Bh|N<
zjzXo$b?#nsCO42|p62Z*1}}CWFZp62VCZrO09;#hF|sN+)VHYN-1J8A`rHpbxdDP;
zu7jQKP3}Yh86BhtqF;F3h*c%*RpSh@lU?hddyMY|@?Dp^j*3yQ4QI3_evj9)W_FFf
z?YiEwAV+a-Hf$uji}8cH<U#(+G3YR)dK4weJ>SWadCAVh?oVIaOnTW$gkyA?#@iM&
z*Lw}4id$3HJ@F1#plL(m1Rn%-Yny96pL8D<a(&vGS{YY`<|$;khb6nFaWb;)JC{!#
z0rJNttiCBKuHk*wUm_C(RdER?)Sn<=A34j1fxZQP800Ani=1TZIbls9lB-F_w3gtF
zA@NohR^zyqwU=8H3~)^cls+Bu1R2jsYRjwO(x!GMeI?HPzXtrZh;jj-J)PVej$xWH
zwL5Jjv@~#$cq^l9{rfQd_j`VqrNE_z2dQ+-adlfuap}0&c1K&5scIJ8@tz+}5k&8G
zc+%Ibkx7rfu7<G`|2VVZ4-yX<rz|VO#?J{k_|gjEj382+psjKdz4t7$Zffs|NDz?I
zf+mk!da?i9k?Z&k7F`woB(0j$anB#r8+5(V2iv6|2e-d!nw$F1r3<P#?YBjtpvtP7
zWEB9C=O4AoQuJo8d`bw2v+%H`A`M8V?NpQKj{A|!Kiowqg?s=~#rofDNbz7+R+;EW
zY}V}zD1_U`skZ0eoetPJ=z(oS<x35+o>WBrm})W7SqUW`x^zB8+~7H>R(Q;$HaH;Y
zQeIU*#v7CTmr~b6xn%z0z@}>gwx|ngY(Zlb)98-%j}+u1zttg3o*gR=(nekei2+AJ
zV#@%7Sra}RvC`6Nq+i#RG4Xa@Ejri*Bu-`XAfs~f=*rd*E~6}cX*`Q(tpjEt<VsCR
zY1`qhpQ44c%)52IjD{SJs3g)IFN^cmHdFGs>t<8>h>%XFOaC(A>x0W#pSRyDs>WEj
zF9{$|$N<R6SI5>nd!qyZrJeOtIvau~36**hAQ66(XEjI^+hnv=F*$JcVs)0hD1a<E
z%}e8&K)P0+`BV_P>|cK5)80*bXDj3I!>YHHc=e|Go7K=DPD)I|?<Yo@@Z^yQ;R$X;
zp1Y`JDhF{Vx!J)2(vtgu_P?euKmKsD-jiPUFqy7_m<T-Ha`Pq30FXi)vwZi;$YAV%
z7`Z1DJ4V{RdK^I3%8>tZte+DT<D|6e$}jBI*h-;0e*f)J<euW~Z~40kpFOuqz9_p&
z$N*J0dRLD?@~NK|yg)q{IYia?PYsW%g=et(kQUMI;%+C*9NkfYhv)HGdpXs5@6mH1
zg30DyPF5CPPTy^mrFk$lUi**me+x)y960`~E8MBMei0XSDJ-N;v+W^&3jL|jqip)F
zz{RdLnHl$?eN$b(gUjD-&om`*=ZmVVHy&31d|&(Yu-`rrpy1FIEU@#FR(P_Jb1clw
zPbl2_o$|}_OVjG?F4p`-52S}iUC+Mh%-<C6yX=`0+<Aura?_4?)O>gT;*nItCtk2&
z=OfD!c#dg<T<F{RCkt)g1}gs>dAL^j_Ou1_mkjbnPCi}b8pRTMH}OUZWDLCiCD!4<
zoM_&nRtAPXS|mDX<5VirKA1CQ9}`dcCIHKZ_?3_g9l21Wgk4q^Pj#U@IB;|8BDJu(
zjOO!II4f7%SA|z7EG{U9k6kp`e%4=nt@pL&m<-4Z!mS5ndUi8+D7J}{()Xn!G`?|U
z--5_L>#WO(iIe>z<5@FgIG-#3*~)b|9CBCb8gRHrrW5QrUBqsM1Bw;<9DFOA94ViZ
z==)pTx12YG8do3AJk=iyFO6N+Iq;^kNb;CF@2q_qqvA<;`@qHFi3-ld9Z5P7N}!S>
z(b=eVcCsmzeN$m+qZL*U&lxy#r;h%K%3ee6ndqI2@H^4fS;S934VWhK^xvSeP-oIS
z9xQe~f6Rf^12+dyxy|$xkM$J>DlFlHA;qTZO%x1Y3u>7(i4gJ^g&h{$nh-A*&)<6a
zvV)g^cb{Me%EgAY@u&SPJUX^UodqR?1qi79FXRb^_b5+)!s=y-`@7HGr<3fenu1~Y
zRS_*xi8f((JujrQV{MO26y7T2g*=B{5qDB~S$;HMgKaj&QEc|>T5Y9ykDtFahXje>
z`9=0oenp$OD>J)&LfCf)cds601V-`_@Hvbh+w`o~=hM=bcN3IQ7MwxXQXlwO|BByw
zYW?78CjQC^MFQPMeh3T6w^vN}KyjjnvoBI!48ZWdBw5*1x_+{v`*7Z3Cu`fm6i!XM
zmD7I|3VQAiaW}$5LXK=C<lrSG600%d(gc+F`YW#%t(%2EPf$tWRQ$%|jv&wDHQznk
z&%;C%sqzJ5e-hUWG`SC@(B<kXhk*h(8};t`L~PL42S(l+<1HbX^of3i=+0&|Vs+ev
zxi&-#>7@=a@5Wr}mg<;46t90}59Q*$XB*k^MlsXtyEsApj;bPPH)h}l@5<F?OI52Y
zY3mCB6-zc}4ck7kqN7)dyeGvhW~v5L16x0q8-<Sgu$gVQ;5Z|Yn5bF7ZKBBVPCl3O
z#)&@U&nkav%NaL5GBB<kKVn?V&>IR2PNi!}y<iZs7Me!%BMfQ>*GID$8%WZS)XeLU
z3$(d}Ybv-xEX6-$hb4DCeKUR)ZTuaJ=@H*A(n-l}o?h!~U_9!es@K`ZH%H@d6?P+z
zfE0RK4;kXbd&;{<C-l;cejWdblu~79`M1z6U%hnNuTfI5yYV6;^m^FjvE{79BS$p#
zAo&Syg)w}te$wv^Lg4~wsAME%VeZgO<|*UygINsSsib$TwYIj`vs4zEY^oEH@Qp;J
zEp9mrB?_&Mjm0R?*e@B~TgTynDW@Im3iZ`I^rZhjtcLKw^;Rc%R%z@07~t_)T<n44
zE`Gxt0QSSom_o`&`N2_m5<NKA&jWHc<0EJ#ecT80zs(2KSrtX7oZYnxr0~I=%pU6i
w(3n*8V5%8E_<LuZ*+qB*DgS>G(RcWsy-Z0jI4eEJ``y52YlXI?Sa>D>19*6ea{vGU

literal 0
HcmV?d00001

diff --git a/api/core/tools/provider/builtin/feishu_wiki/feishu_wiki.py b/api/core/tools/provider/builtin/feishu_wiki/feishu_wiki.py
new file mode 100644
index 0000000000000..6c5fccb1a31d0
--- /dev/null
+++ b/api/core/tools/provider/builtin/feishu_wiki/feishu_wiki.py
@@ -0,0 +1,7 @@
+from core.tools.provider.builtin_tool_provider import BuiltinToolProviderController
+from core.tools.utils.feishu_api_utils import auth
+
+
+class FeishuWikiProvider(BuiltinToolProviderController):
+    def _validate_credentials(self, credentials: dict) -> None:
+        auth(credentials)
diff --git a/api/core/tools/provider/builtin/feishu_wiki/feishu_wiki.yaml b/api/core/tools/provider/builtin/feishu_wiki/feishu_wiki.yaml
new file mode 100644
index 0000000000000..1fb5f71cbc516
--- /dev/null
+++ b/api/core/tools/provider/builtin/feishu_wiki/feishu_wiki.yaml
@@ -0,0 +1,36 @@
+identity:
+  author: Doug Lea
+  name: feishu_wiki
+  label:
+    en_US: Feishu Wiki
+    zh_Hans: 飞书知识库
+  description:
+    en_US: |
+      Feishu Wiki, requires the following permissions: wiki:wiki:readonly.
+    zh_Hans: |
+      飞书知识库，需要开通以下权限: wiki:wiki:readonly。
+  icon: icon.png
+  tags:
+    - social
+    - productivity
+credentials_for_provider:
+  app_id:
+    type: text-input
+    required: true
+    label:
+      en_US: APP ID
+    placeholder:
+      en_US: Please input your feishu app id
+      zh_Hans: 请输入你的飞书 app id
+    help:
+      en_US: Get your app_id and app_secret from Feishu
+      zh_Hans: 从飞书获取您的 app_id 和 app_secret
+    url: https://open.larkoffice.com/app
+  app_secret:
+    type: secret-input
+    required: true
+    label:
+      en_US: APP Secret
+    placeholder:
+      en_US: Please input your app secret
+      zh_Hans: 请输入你的飞书 app secret
diff --git a/api/core/tools/provider/builtin/feishu_wiki/tools/get_wiki_nodes.py b/api/core/tools/provider/builtin/feishu_wiki/tools/get_wiki_nodes.py
new file mode 100644
index 0000000000000..374b4c9a7d149
--- /dev/null
+++ b/api/core/tools/provider/builtin/feishu_wiki/tools/get_wiki_nodes.py
@@ -0,0 +1,21 @@
+from typing import Any
+
+from core.tools.entities.tool_entities import ToolInvokeMessage
+from core.tools.tool.builtin_tool import BuiltinTool
+from core.tools.utils.feishu_api_utils import FeishuRequest
+
+
+class GetWikiNodesTool(BuiltinTool):
+    def _invoke(self, user_id: str, tool_parameters: dict[str, Any]) -> ToolInvokeMessage:
+        app_id = self.runtime.credentials.get("app_id")
+        app_secret = self.runtime.credentials.get("app_secret")
+        client = FeishuRequest(app_id, app_secret)
+
+        space_id = tool_parameters.get("space_id")
+        parent_node_token = tool_parameters.get("parent_node_token")
+        page_token = tool_parameters.get("page_token")
+        page_size = tool_parameters.get("page_size")
+
+        res = client.get_wiki_nodes(space_id, parent_node_token, page_token, page_size)
+
+        return self.create_json_message(res)
diff --git a/api/core/tools/provider/builtin/feishu_wiki/tools/get_wiki_nodes.yaml b/api/core/tools/provider/builtin/feishu_wiki/tools/get_wiki_nodes.yaml
new file mode 100644
index 0000000000000..7d6ac3c824877
--- /dev/null
+++ b/api/core/tools/provider/builtin/feishu_wiki/tools/get_wiki_nodes.yaml
@@ -0,0 +1,63 @@
+identity:
+  name: get_wiki_nodes
+  author: Doug Lea
+  label:
+    en_US: Get Wiki Nodes
+    zh_Hans: 获取知识空间子节点列表
+description:
+  human:
+    en_US: |
+      Get the list of child nodes in Wiki, make sure the app/bot is a member of the wiki space. See How to add an app as a wiki base administrator (member). https://open.feishu.cn/document/server-docs/docs/wiki-v2/wiki-qa
+    zh_Hans: |
+      获取知识库全部子节点列表，请确保应用/机器人为知识空间成员。参阅如何将应用添加为知识库管理员（成员）。https://open.feishu.cn/document/server-docs/docs/wiki-v2/wiki-qa
+  llm: A tool for getting all sub-nodes of a knowledge base.(获取知识空间子节点列表)
+parameters:
+  - name: space_id
+    type: string
+    required: true
+    label:
+      en_US: Space Id
+      zh_Hans: 知识空间 ID
+    human_description:
+      en_US: |
+        The ID of the knowledge space. Supports space link URL, for example: https://svi136aogf123.feishu.cn/wiki/settings/7166950623940706332
+      zh_Hans: 知识空间 ID，支持空间链接 URL，例如：https://svi136aogf123.feishu.cn/wiki/settings/7166950623940706332
+    llm_description: 知识空间 ID，支持空间链接 URL，例如：https://svi136aogf123.feishu.cn/wiki/settings/7166950623940706332
+    form: llm
+
+  - name: page_size
+    type: number
+    required: false
+    default: 10
+    label:
+      en_US: Page Size
+      zh_Hans: 分页大小
+    human_description:
+      en_US: The size of each page, with a maximum value of 50.
+      zh_Hans: 分页大小，最大值 50。
+    llm_description: 分页大小，最大值 50。
+    form: llm
+
+  - name: page_token
+    type: string
+    required: false
+    label:
+      en_US: Page Token
+      zh_Hans: 分页标记
+    human_description:
+      en_US: The pagination token. Leave empty for the first request to start from the beginning; if the paginated query result has more items, a new page_token will be returned, which can be used to get the next set of results.
+      zh_Hans: 分页标记，第一次请求不填，表示从头开始遍历；分页查询结果还有更多项时会同时返回新的 page_token，下次遍历可采用该 page_token 获取查询结果。
+    llm_description: 分页标记，第一次请求不填，表示从头开始遍历；分页查询结果还有更多项时会同时返回新的 page_token，下次遍历可采用该 page_token 获取查询结果。
+    form: llm
+
+  - name: parent_node_token
+    type: string
+    required: false
+    label:
+      en_US: Parent Node Token
+      zh_Hans: 父节点 token
+    human_description:
+      en_US: The token of the parent node.
+      zh_Hans: 父节点 token
+    llm_description: 父节点 token
+    form: llm
diff --git a/api/core/tools/utils/feishu_api_utils.py b/api/core/tools/utils/feishu_api_utils.py
index ffdb06498fd51..ce1fd7dc19e80 100644
--- a/api/core/tools/utils/feishu_api_utils.py
+++ b/api/core/tools/utils/feishu_api_utils.py
@@ -1,9 +1,23 @@
 import httpx
 
+from core.tools.errors import ToolProviderCredentialValidationError
 from extensions.ext_redis import redis_client
 
 
+def auth(credentials):
+    app_id = credentials.get("app_id")
+    app_secret = credentials.get("app_secret")
+    if not app_id or not app_secret:
+        raise ToolProviderCredentialValidationError("app_id and app_secret is required")
+    try:
+        assert FeishuRequest(app_id, app_secret).tenant_access_token is not None
+    except Exception as e:
+        raise ToolProviderCredentialValidationError(str(e))
+
+
 class FeishuRequest:
+    API_BASE_URL = "https://lark-plugin-api.solutionsuite.cn/lark-plugin"
+
     def __init__(self, app_id: str, app_secret: str):
         self.app_id = app_id
         self.app_secret = app_secret
@@ -42,7 +56,7 @@ def get_tenant_access_token(self, app_id: str, app_secret: str) -> dict:
             "expire": 7200
         }
         """
-        url = "https://lark-plugin-api.solutionsuite.cn/lark-plugin/access_token/get_tenant_access_token"
+        url = f"{self.API_BASE_URL}/access_token/get_tenant_access_token"
         payload = {"app_id": app_id, "app_secret": app_secret}
         res = self._send_request(url, require_token=False, payload=payload)
         return res
@@ -63,7 +77,7 @@ def create_document(self, title: str, content: str, folder_token: str) -> dict:
             "msg": "创建飞书文档成功，请查看"
         }
         """
-        url = "https://lark-plugin-api.solutionsuite.cn/lark-plugin/document/create_document"
+        url = f"{self.API_BASE_URL}/document/create_document"
         payload = {
             "title": title,
             "content": content,
@@ -72,13 +86,13 @@ def create_document(self, title: str, content: str, folder_token: str) -> dict:
         res = self._send_request(url, payload=payload)
         return res.get("data")
 
-    def write_document(self, document_id: str, content: str, position: str = "start") -> dict:
-        url = "https://lark-plugin-api.solutionsuite.cn/lark-plugin/document/write_document"
+    def write_document(self, document_id: str, content: str, position: str = "end") -> dict:
+        url = f"{self.API_BASE_URL}/document/write_document"
         payload = {"document_id": document_id, "content": content, "position": position}
         res = self._send_request(url, payload=payload)
         return res
 
-    def get_document_content(self, document_id: str, mode: str, lang: int = 0) -> dict:
+    def get_document_content(self, document_id: str, mode: str = "markdown", lang: str = "0") -> dict:
         """
         API url: https://open.larkoffice.com/document/server-docs/docs/docs/docx-v1/document/raw_content
         Example Response:
@@ -95,45 +109,404 @@ def get_document_content(self, document_id: str, mode: str, lang: int = 0) -> di
             "mode": mode,
             "lang": lang,
         }
-        url = "https://lark-plugin-api.solutionsuite.cn/lark-plugin/document/get_document_content"
-        res = self._send_request(url, method="get", params=params)
+        url = f"{self.API_BASE_URL}/document/get_document_content"
+        res = self._send_request(url, method="GET", params=params)
         return res.get("data").get("content")
 
-    def list_document_blocks(self, document_id: str, page_token: str, page_size: int = 500) -> dict:
+    def list_document_blocks(
+        self, document_id: str, page_token: str, user_id_type: str = "open_id", page_size: int = 500
+    ) -> dict:
         """
         API url: https://open.larkoffice.com/document/server-docs/docs/docs/docx-v1/document/list
         """
-        url = "https://lark-plugin-api.solutionsuite.cn/lark-plugin/document/list_document_blocks"
         params = {
+            "user_id_type": user_id_type,
             "document_id": document_id,
             "page_size": page_size,
             "page_token": page_token,
         }
-        res = self._send_request(url, method="get", params=params)
+        url = f"{self.API_BASE_URL}/document/list_document_blocks"
+        res = self._send_request(url, method="GET", params=params)
         return res.get("data")
 
     def send_bot_message(self, receive_id_type: str, receive_id: str, msg_type: str, content: str) -> dict:
         """
         API url: https://open.larkoffice.com/document/server-docs/im-v1/message/create
         """
-        url = "https://lark-plugin-api.solutionsuite.cn/lark-plugin/message/send_bot_message"
+        url = f"{self.API_BASE_URL}/message/send_bot_message"
         params = {
             "receive_id_type": receive_id_type,
         }
         payload = {
             "receive_id": receive_id,
             "msg_type": msg_type,
-            "content": content,
+            "content": content.strip('"').replace(r"\"", '"').replace(r"\\", "\\"),
         }
         res = self._send_request(url, params=params, payload=payload)
         return res.get("data")
 
     def send_webhook_message(self, webhook: str, msg_type: str, content: str) -> dict:
-        url = "https://lark-plugin-api.solutionsuite.cn/lark-plugin/message/send_webhook_message"
+        url = f"{self.API_BASE_URL}/message/send_webhook_message"
         payload = {
             "webhook": webhook,
             "msg_type": msg_type,
-            "content": content,
+            "content": content.strip('"').replace(r"\"", '"').replace(r"\\", "\\"),
         }
         res = self._send_request(url, require_token=False, payload=payload)
         return res
+
+    def get_chat_messages(
+        self,
+        container_id: str,
+        start_time: str,
+        end_time: str,
+        page_token: str,
+        sort_type: str = "ByCreateTimeAsc",
+        page_size: int = 20,
+    ) -> dict:
+        """
+        API url: https://open.larkoffice.com/document/server-docs/im-v1/message/list
+        """
+        url = f"{self.API_BASE_URL}/message/get_chat_messages"
+        params = {
+            "container_id": container_id,
+            "start_time": start_time,
+            "end_time": end_time,
+            "sort_type": sort_type,
+            "page_token": page_token,
+            "page_size": page_size,
+        }
+        res = self._send_request(url, method="GET", params=params)
+        return res.get("data")
+
+    def get_thread_messages(
+        self, container_id: str, page_token: str, sort_type: str = "ByCreateTimeAsc", page_size: int = 20
+    ) -> dict:
+        """
+        API url: https://open.larkoffice.com/document/server-docs/im-v1/message/list
+        """
+        url = f"{self.API_BASE_URL}/message/get_thread_messages"
+        params = {
+            "container_id": container_id,
+            "sort_type": sort_type,
+            "page_token": page_token,
+            "page_size": page_size,
+        }
+        res = self._send_request(url, method="GET", params=params)
+        return res.get("data")
+
+    def create_task(self, summary: str, start_time: str, end_time: str, completed_time: str, description: str) -> dict:
+        # 创建任务
+        url = f"{self.API_BASE_URL}/task/create_task"
+        payload = {
+            "summary": summary,
+            "start_time": start_time,
+            "end_time": end_time,
+            "completed_at": completed_time,
+            "description": description,
+        }
+        res = self._send_request(url, payload=payload)
+        return res.get("data")
+
+    def update_task(
+        self, task_guid: str, summary: str, start_time: str, end_time: str, completed_time: str, description: str
+    ) -> dict:
+        # 更新任务
+        url = f"{self.API_BASE_URL}/task/update_task"
+        payload = {
+            "task_guid": task_guid,
+            "summary": summary,
+            "start_time": start_time,
+            "end_time": end_time,
+            "completed_time": completed_time,
+            "description": description,
+        }
+        res = self._send_request(url, method="PATCH", payload=payload)
+        return res.get("data")
+
+    def delete_task(self, task_guid: str) -> dict:
+        # 删除任务
+        url = f"{self.API_BASE_URL}/task/delete_task"
+        payload = {
+            "task_guid": task_guid,
+        }
+        res = self._send_request(url, method="DELETE", payload=payload)
+        return res
+
+    def add_members(self, task_guid: str, member_phone_or_email: str, member_role: str) -> dict:
+        # 删除任务
+        url = f"{self.API_BASE_URL}/task/add_members"
+        payload = {
+            "task_guid": task_guid,
+            "member_phone_or_email": member_phone_or_email,
+            "member_role": member_role,
+        }
+        res = self._send_request(url, payload=payload)
+        return res
+
+    def get_wiki_nodes(self, space_id: str, parent_node_token: str, page_token: str, page_size: int = 20) -> dict:
+        # 获取知识库全部子节点列表
+        url = f"{self.API_BASE_URL}/wiki/get_wiki_nodes"
+        payload = {
+            "space_id": space_id,
+            "parent_node_token": parent_node_token,
+            "page_token": page_token,
+            "page_size": page_size,
+        }
+        res = self._send_request(url, payload=payload)
+        return res.get("data")
+
+    def get_primary_calendar(self, user_id_type: str = "open_id") -> dict:
+        url = f"{self.API_BASE_URL}/calendar/get_primary_calendar"
+        params = {
+            "user_id_type": user_id_type,
+        }
+        res = self._send_request(url, method="GET", params=params)
+        return res.get("data")
+
+    def create_event(
+        self,
+        summary: str,
+        description: str,
+        start_time: str,
+        end_time: str,
+        attendee_ability: str,
+        need_notification: bool = True,
+        auto_record: bool = False,
+    ) -> dict:
+        url = f"{self.API_BASE_URL}/calendar/create_event"
+        payload = {
+            "summary": summary,
+            "description": description,
+            "need_notification": need_notification,
+            "start_time": start_time,
+            "end_time": end_time,
+            "auto_record": auto_record,
+            "attendee_ability": attendee_ability,
+        }
+        res = self._send_request(url, payload=payload)
+        return res.get("data")
+
+    def update_event(
+        self,
+        event_id: str,
+        summary: str,
+        description: str,
+        need_notification: bool,
+        start_time: str,
+        end_time: str,
+        auto_record: bool,
+    ) -> dict:
+        url = f"{self.API_BASE_URL}/calendar/update_event/{event_id}"
+        payload = {}
+        if summary:
+            payload["summary"] = summary
+        if description:
+            payload["description"] = description
+        if start_time:
+            payload["start_time"] = start_time
+        if end_time:
+            payload["end_time"] = end_time
+        if need_notification:
+            payload["need_notification"] = need_notification
+        if auto_record:
+            payload["auto_record"] = auto_record
+        res = self._send_request(url, method="PATCH", payload=payload)
+        return res
+
+    def delete_event(self, event_id: str, need_notification: bool = True) -> dict:
+        url = f"{self.API_BASE_URL}/calendar/delete_event/{event_id}"
+        params = {
+            "need_notification": need_notification,
+        }
+        res = self._send_request(url, method="DELETE", params=params)
+        return res
+
+    def list_events(self, start_time: str, end_time: str, page_token: str, page_size: int = 50) -> dict:
+        url = f"{self.API_BASE_URL}/calendar/list_events"
+        params = {
+            "start_time": start_time,
+            "end_time": end_time,
+            "page_token": page_token,
+            "page_size": page_size,
+        }
+        res = self._send_request(url, method="GET", params=params)
+        return res.get("data")
+
+    def search_events(
+        self,
+        query: str,
+        start_time: str,
+        end_time: str,
+        page_token: str,
+        user_id_type: str = "open_id",
+        page_size: int = 20,
+    ) -> dict:
+        url = f"{self.API_BASE_URL}/calendar/search_events"
+        payload = {
+            "query": query,
+            "start_time": start_time,
+            "end_time": end_time,
+            "page_token": page_token,
+            "user_id_type": user_id_type,
+            "page_size": page_size,
+        }
+        res = self._send_request(url, payload=payload)
+        return res.get("data")
+
+    def add_event_attendees(self, event_id: str, attendee_phone_or_email: str, need_notification: bool = True) -> dict:
+        # 参加日程参会人
+        url = f"{self.API_BASE_URL}/calendar/add_event_attendees"
+        payload = {
+            "event_id": event_id,
+            "attendee_phone_or_email": attendee_phone_or_email,
+            "need_notification": need_notification,
+        }
+        res = self._send_request(url, payload=payload)
+        return res.get("data")
+
+    def create_spreadsheet(
+        self,
+        title: str,
+        folder_token: str,
+    ) -> dict:
+        # 创建电子表格
+        url = f"{self.API_BASE_URL}/spreadsheet/create_spreadsheet"
+        payload = {
+            "title": title,
+            "folder_token": folder_token,
+        }
+        res = self._send_request(url, payload=payload)
+        return res.get("data")
+
+    def get_spreadsheet(
+        self,
+        spreadsheet_token: str,
+        user_id_type: str = "open_id",
+    ) -> dict:
+        # 获取电子表格信息
+        url = f"{self.API_BASE_URL}/spreadsheet/get_spreadsheet"
+        params = {
+            "spreadsheet_token": spreadsheet_token,
+            "user_id_type": user_id_type,
+        }
+        res = self._send_request(url, method="GET", params=params)
+        return res.get("data")
+
+    def list_spreadsheet_sheets(
+        self,
+        spreadsheet_token: str,
+    ) -> dict:
+        # 列出电子表格的所有工作表
+        url = f"{self.API_BASE_URL}/spreadsheet/list_spreadsheet_sheets"
+        params = {
+            "spreadsheet_token": spreadsheet_token,
+        }
+        res = self._send_request(url, method="GET", params=params)
+        return res.get("data")
+
+    def add_rows(
+        self,
+        spreadsheet_token: str,
+        sheet_id: str,
+        sheet_name: str,
+        length: int,
+        values: str,
+    ) -> dict:
+        # 增加行,在工作表最后添加
+        url = f"{self.API_BASE_URL}/spreadsheet/add_rows"
+        payload = {
+            "spreadsheet_token": spreadsheet_token,
+            "sheet_id": sheet_id,
+            "sheet_name": sheet_name,
+            "length": length,
+            "values": values,
+        }
+        res = self._send_request(url, payload=payload)
+        return res.get("data")
+
+    def add_cols(
+        self,
+        spreadsheet_token: str,
+        sheet_id: str,
+        sheet_name: str,
+        length: int,
+        values: str,
+    ) -> dict:
+        #  增加列,在工作表最后添加
+        url = f"{self.API_BASE_URL}/spreadsheet/add_cols"
+        payload = {
+            "spreadsheet_token": spreadsheet_token,
+            "sheet_id": sheet_id,
+            "sheet_name": sheet_name,
+            "length": length,
+            "values": values,
+        }
+        res = self._send_request(url, payload=payload)
+        return res.get("data")
+
+    def read_rows(
+        self,
+        spreadsheet_token: str,
+        sheet_id: str,
+        sheet_name: str,
+        start_row: int,
+        num_rows: int,
+        user_id_type: str = "open_id",
+    ) -> dict:
+        # 读取工作表行数据
+        url = f"{self.API_BASE_URL}/spreadsheet/read_rows"
+        params = {
+            "spreadsheet_token": spreadsheet_token,
+            "sheet_id": sheet_id,
+            "sheet_name": sheet_name,
+            "start_row": start_row,
+            "num_rows": num_rows,
+            "user_id_type": user_id_type,
+        }
+        res = self._send_request(url, method="GET", params=params)
+        return res.get("data")
+
+    def read_cols(
+        self,
+        spreadsheet_token: str,
+        sheet_id: str,
+        sheet_name: str,
+        start_col: int,
+        num_cols: int,
+        user_id_type: str = "open_id",
+    ) -> dict:
+        # 读取工作表列数据
+        url = f"{self.API_BASE_URL}/spreadsheet/read_cols"
+        params = {
+            "spreadsheet_token": spreadsheet_token,
+            "sheet_id": sheet_id,
+            "sheet_name": sheet_name,
+            "start_col": start_col,
+            "num_cols": num_cols,
+            "user_id_type": user_id_type,
+        }
+        res = self._send_request(url, method="GET", params=params)
+        return res.get("data")
+
+    def read_table(
+        self,
+        spreadsheet_token: str,
+        sheet_id: str,
+        sheet_name: str,
+        num_range: str,
+        query: str,
+        user_id_type: str = "open_id",
+    ) -> dict:
+        # 自定义读取行列数据
+        url = f"{self.API_BASE_URL}/spreadsheet/read_table"
+        params = {
+            "spreadsheet_token": spreadsheet_token,
+            "sheet_id": sheet_id,
+            "sheet_name": sheet_name,
+            "range": num_range,
+            "query": query,
+            "user_id_type": user_id_type,
+        }
+        res = self._send_request(url, method="GET", params=params)
+        return res.get("data")

From 55e6123db93ae7fbda06596d6cd0a163747e6559 Mon Sep 17 00:00:00 2001
From: zhuhao <37029601+hwzhuhao@users.noreply.github.com>
Date: Fri, 27 Sep 2024 18:16:20 +0800
Subject: [PATCH 094/235] feat: add min-connection and max-connection for
 pgvector (#8841)

---
 api/.env.example                                   |  2 ++
 api/configs/middleware/vdb/pgvector_config.py      | 10 ++++++++++
 api/core/rag/datasource/vdb/pgvector/pgvector.py   | 14 ++++++++++++--
 .../vdb/pgvector/test_pgvector.py                  |  2 ++
 docker/.env.example                                |  2 ++
 5 files changed, 28 insertions(+), 2 deletions(-)

diff --git a/api/.env.example b/api/.env.example
index a3db406aeabdc..7e5c636ff8975 100644
--- a/api/.env.example
+++ b/api/.env.example
@@ -162,6 +162,8 @@ PGVECTOR_PORT=5433
 PGVECTOR_USER=postgres
 PGVECTOR_PASSWORD=postgres
 PGVECTOR_DATABASE=postgres
+PGVECTOR_MIN_CONNECTION=1
+PGVECTOR_MAX_CONNECTION=5
 
 # Tidb Vector configuration
 TIDB_VECTOR_HOST=xxx.eu-central-1.xxx.aws.tidbcloud.com
diff --git a/api/configs/middleware/vdb/pgvector_config.py b/api/configs/middleware/vdb/pgvector_config.py
index 395dcaa4208a9..85f5dca7e2322 100644
--- a/api/configs/middleware/vdb/pgvector_config.py
+++ b/api/configs/middleware/vdb/pgvector_config.py
@@ -33,3 +33,13 @@ class PGVectorConfig(BaseSettings):
         description="Name of the PostgreSQL database to connect to",
         default=None,
     )
+
+    PGVECTOR_MIN_CONNECTION: PositiveInt = Field(
+        description="Min connection of the PostgreSQL database",
+        default=1,
+    )
+
+    PGVECTOR_MAX_CONNECTION: PositiveInt = Field(
+        description="Max connection of the PostgreSQL database",
+        default=5,
+    )
diff --git a/api/core/rag/datasource/vdb/pgvector/pgvector.py b/api/core/rag/datasource/vdb/pgvector/pgvector.py
index 79879d4f6394b..d90707ebcd5cb 100644
--- a/api/core/rag/datasource/vdb/pgvector/pgvector.py
+++ b/api/core/rag/datasource/vdb/pgvector/pgvector.py
@@ -23,6 +23,8 @@ class PGVectorConfig(BaseModel):
     user: str
     password: str
     database: str
+    min_connection: int
+    max_connection: int
 
     @model_validator(mode="before")
     @classmethod
@@ -37,6 +39,12 @@ def validate_config(cls, values: dict) -> dict:
             raise ValueError("config PGVECTOR_PASSWORD is required")
         if not values["database"]:
             raise ValueError("config PGVECTOR_DATABASE is required")
+        if not values["min_connection"]:
+            raise ValueError("config PGVECTOR_MIN_CONNECTION is required")
+        if not values["max_connection"]:
+            raise ValueError("config PGVECTOR_MAX_CONNECTION is required")
+        if values["min_connection"] > values["max_connection"]:
+            raise ValueError("config PGVECTOR_MIN_CONNECTION should less than PGVECTOR_MAX_CONNECTION")
         return values
 
 
@@ -61,8 +69,8 @@ def get_type(self) -> str:
 
     def _create_connection_pool(self, config: PGVectorConfig):
         return psycopg2.pool.SimpleConnectionPool(
-            1,
-            5,
+            config.min_connection,
+            config.max_connection,
             host=config.host,
             port=config.port,
             user=config.user,
@@ -213,5 +221,7 @@ def init_vector(self, dataset: Dataset, attributes: list, embeddings: Embeddings
                 user=dify_config.PGVECTOR_USER,
                 password=dify_config.PGVECTOR_PASSWORD,
                 database=dify_config.PGVECTOR_DATABASE,
+                min_connection=dify_config.PGVECTOR_MIN_CONNECTION,
+                max_connection=dify_config.PGVECTOR_MAX_CONNECTION,
             ),
         )
diff --git a/api/tests/integration_tests/vdb/pgvector/test_pgvector.py b/api/tests/integration_tests/vdb/pgvector/test_pgvector.py
index c5a986b747925..72efdc2780eca 100644
--- a/api/tests/integration_tests/vdb/pgvector/test_pgvector.py
+++ b/api/tests/integration_tests/vdb/pgvector/test_pgvector.py
@@ -18,6 +18,8 @@ def __init__(self):
                 user="postgres",
                 password="difyai123456",
                 database="dify",
+                min_connection=1,
+                max_connection=5,
             ),
         )
 
diff --git a/docker/.env.example b/docker/.env.example
index d43c3edc7e722..eb05f7aa4f0b2 100644
--- a/docker/.env.example
+++ b/docker/.env.example
@@ -391,6 +391,8 @@ PGVECTOR_PORT=5432
 PGVECTOR_USER=postgres
 PGVECTOR_PASSWORD=difyai123456
 PGVECTOR_DATABASE=dify
+PGVECTOR_MIN_CONNECTION=1
+PGVECTOR_MAX_CONNECTION=5
 
 # pgvecto-rs configurations, only available when VECTOR_STORE is `pgvecto-rs`
 PGVECTO_RS_HOST=pgvecto-rs

From 27e33fb15c01befb47d02afe257746778b1164ed Mon Sep 17 00:00:00 2001
From: ice yao <yao3690093@gmail.com>
Date: Sat, 28 Sep 2024 10:54:04 +0800
Subject: [PATCH 095/235] chore: fix wrong VectorType match case (#8857)

---
 api/controllers/console/datasets/datasets.py          | 3 ++-
 api/core/rag/datasource/vdb/tencent/tencent_vector.py | 2 +-
 2 files changed, 3 insertions(+), 2 deletions(-)

diff --git a/api/controllers/console/datasets/datasets.py b/api/controllers/console/datasets/datasets.py
index 2c4e5ac60789e..5a763b3457c69 100644
--- a/api/controllers/console/datasets/datasets.py
+++ b/api/controllers/console/datasets/datasets.py
@@ -563,10 +563,10 @@ def get(self):
             case (
                 VectorType.MILVUS
                 | VectorType.RELYT
-                | VectorType.PGVECTOR
                 | VectorType.TIDB_VECTOR
                 | VectorType.CHROMA
                 | VectorType.TENCENT
+                | VectorType.PGVECTO_RS
             ):
                 return {"retrieval_method": [RetrievalMethod.SEMANTIC_SEARCH.value]}
             case (
@@ -577,6 +577,7 @@ def get(self):
                 | VectorType.MYSCALE
                 | VectorType.ORACLE
                 | VectorType.ELASTICSEARCH
+                | VectorType.PGVECTOR
             ):
                 return {
                     "retrieval_method": [
diff --git a/api/core/rag/datasource/vdb/tencent/tencent_vector.py b/api/core/rag/datasource/vdb/tencent/tencent_vector.py
index faa373017becf..39e3a7f6cfea4 100644
--- a/api/core/rag/datasource/vdb/tencent/tencent_vector.py
+++ b/api/core/rag/datasource/vdb/tencent/tencent_vector.py
@@ -56,7 +56,7 @@ def _init_database(self):
             return self._client.create_database(database_name=self._client_config.database)
 
     def get_type(self) -> str:
-        return "tencent"
+        return VectorType.TENCENT
 
     def to_index_struct(self) -> dict:
         return {"type": self.get_type(), "vector_store": {"class_prefix": self._collection_name}}

From a2e2f8a8c98e188dd3d0a4669b4775260756a106 Mon Sep 17 00:00:00 2001
From: Kevin9703 <51311316+Kevin9703@users.noreply.github.com>
Date: Sat, 28 Sep 2024 10:54:50 +0800
Subject: [PATCH 096/235] =?UTF-8?q?fix(workflow/nodes/knowledge-retrieval/?=
 =?UTF-8?q?use-config):=20Preserve=20rerankin=E2=80=A6=20(#8842)?=
MIME-Version: 1.0
Content-Type: text/plain; charset=UTF-8
Content-Transfer-Encoding: 8bit

---
 .../components/workflow/nodes/knowledge-retrieval/use-config.ts | 2 ++
 1 file changed, 2 insertions(+)

diff --git a/web/app/components/workflow/nodes/knowledge-retrieval/use-config.ts b/web/app/components/workflow/nodes/knowledge-retrieval/use-config.ts
index 994bf4f2052df..6d856003d4110 100644
--- a/web/app/components/workflow/nodes/knowledge-retrieval/use-config.ts
+++ b/web/app/components/workflow/nodes/knowledge-retrieval/use-config.ts
@@ -136,6 +136,8 @@ const useConfig = (id: string, payload: KnowledgeRetrievalNodeType) => {
         top_k: multipleRetrievalConfig?.top_k || DATASET_DEFAULT.top_k,
         score_threshold: multipleRetrievalConfig?.score_threshold,
         reranking_model: multipleRetrievalConfig?.reranking_model,
+        reranking_mode: multipleRetrievalConfig?.reranking_mode,
+        weights: multipleRetrievalConfig?.weights,
       }
     })
     setInputs(newInput)

From 6cd22f3bca2509c19f7b4dcfca4b4785b139d677 Mon Sep 17 00:00:00 2001
From: zhuhao <37029601+hwzhuhao@users.noreply.github.com>
Date: Sat, 28 Sep 2024 15:01:27 +0800
Subject: [PATCH 097/235] fix: update qwen2.5-coder-7b model name (#8861)

---
 .../model_providers/tongyi/llm/qwen2.5-coder-7b-instruct.yaml | 4 ++--
 1 file changed, 2 insertions(+), 2 deletions(-)

diff --git a/api/core/model_runtime/model_providers/tongyi/llm/qwen2.5-coder-7b-instruct.yaml b/api/core/model_runtime/model_providers/tongyi/llm/qwen2.5-coder-7b-instruct.yaml
index fdcd3d42757ed..7ebeec395393c 100644
--- a/api/core/model_runtime/model_providers/tongyi/llm/qwen2.5-coder-7b-instruct.yaml
+++ b/api/core/model_runtime/model_providers/tongyi/llm/qwen2.5-coder-7b-instruct.yaml
@@ -1,7 +1,7 @@
 # for more details, please refer to https://help.aliyun.com/zh/model-studio/getting-started/models
-model: qwen2.5-7b-instruct
+model: qwen2.5-coder-7b-instruct
 label:
-  en_US: qwen2.5-7b-instruct
+  en_US: qwen2.5-coder-7b-instruct
 model_type: llm
 features:
   - agent-thought

From 49af18fbd6549e1ce756544529158a5ca8a7c338 Mon Sep 17 00:00:00 2001
From: takatost <takatost@users.noreply.github.com>
Date: Sat, 28 Sep 2024 15:54:26 +0800
Subject: [PATCH 098/235] fix: customize model credentials were invalid despite
 the provider credentials being active (#8864)

---
 api/core/entities/provider_configuration.py | 2 +-
 1 file changed, 1 insertion(+), 1 deletion(-)

diff --git a/api/core/entities/provider_configuration.py b/api/core/entities/provider_configuration.py
index 4797b69b8596b..807f09598c760 100644
--- a/api/core/entities/provider_configuration.py
+++ b/api/core/entities/provider_configuration.py
@@ -119,7 +119,7 @@ def get_current_credentials(self, model_type: ModelType, model: str) -> Optional
                         credentials = model_configuration.credentials
                         break
 
-            if self.custom_configuration.provider:
+            if not credentials and self.custom_configuration.provider:
                 credentials = self.custom_configuration.provider.credentials
 
             return credentials

From 61c89a9168e54f79be54866a1e88171da533f91a Mon Sep 17 00:00:00 2001
From: zhuhao <37029601+hwzhuhao@users.noreply.github.com>
Date: Sat, 28 Sep 2024 16:31:02 +0800
Subject: [PATCH 099/235] feat: add internlm2.5-20b and qwen2.5-coder-7b model
 (#8862)

---
 .../siliconflow/llm/_position.yaml            | 16 +---
 .../siliconflow/llm/internlm2_5-20b-chat.yaml | 30 ++++++++
 .../llm/qwen2.5-coder-7b-instruct.yaml        | 74 +++++++++++++++++++
 .../llm/qwen2.5-math-72b-instruct.yaml        | 74 +++++++++++++++++++
 4 files changed, 181 insertions(+), 13 deletions(-)
 create mode 100644 api/core/model_runtime/model_providers/siliconflow/llm/internlm2_5-20b-chat.yaml
 create mode 100644 api/core/model_runtime/model_providers/siliconflow/llm/qwen2.5-coder-7b-instruct.yaml
 create mode 100644 api/core/model_runtime/model_providers/siliconflow/llm/qwen2.5-math-72b-instruct.yaml

diff --git a/api/core/model_runtime/model_providers/siliconflow/llm/_position.yaml b/api/core/model_runtime/model_providers/siliconflow/llm/_position.yaml
index a3e5d0981f743..8d1df82140b79 100644
--- a/api/core/model_runtime/model_providers/siliconflow/llm/_position.yaml
+++ b/api/core/model_runtime/model_providers/siliconflow/llm/_position.yaml
@@ -1,18 +1,17 @@
 - Qwen/Qwen2.5-72B-Instruct
-- Qwen/Qwen2.5-Math-72B-Instruct
 - Qwen/Qwen2.5-32B-Instruct
 - Qwen/Qwen2.5-14B-Instruct
 - Qwen/Qwen2.5-7B-Instruct
 - Qwen/Qwen2.5-Coder-7B-Instruct
-- deepseek-ai/DeepSeek-V2.5
+- Qwen/Qwen2.5-Math-72B-Instruct
 - Qwen/Qwen2-72B-Instruct
 - Qwen/Qwen2-57B-A14B-Instruct
 - Qwen/Qwen2-7B-Instruct
 - Qwen/Qwen2-1.5B-Instruct
+- deepseek-ai/DeepSeek-V2.5
 - deepseek-ai/DeepSeek-V2-Chat
 - deepseek-ai/DeepSeek-Coder-V2-Instruct
 - THUDM/glm-4-9b-chat
-- THUDM/chatglm3-6b
 - 01-ai/Yi-1.5-34B-Chat-16K
 - 01-ai/Yi-1.5-9B-Chat-16K
 - 01-ai/Yi-1.5-6B-Chat
@@ -26,13 +25,4 @@
 - google/gemma-2-27b-it
 - google/gemma-2-9b-it
 - mistralai/Mistral-7B-Instruct-v0.2
-- Pro/Qwen/Qwen2-7B-Instruct
-- Pro/Qwen/Qwen2-1.5B-Instruct
-- Pro/THUDM/glm-4-9b-chat
-- Pro/THUDM/chatglm3-6b
-- Pro/01-ai/Yi-1.5-9B-Chat-16K
-- Pro/01-ai/Yi-1.5-6B-Chat
-- Pro/internlm/internlm2_5-7b-chat
-- Pro/meta-llama/Meta-Llama-3.1-8B-Instruct
-- Pro/meta-llama/Meta-Llama-3-8B-Instruct
-- Pro/google/gemma-2-9b-it
+- mistralai/Mixtral-8x7B-Instruct-v0.1
diff --git a/api/core/model_runtime/model_providers/siliconflow/llm/internlm2_5-20b-chat.yaml b/api/core/model_runtime/model_providers/siliconflow/llm/internlm2_5-20b-chat.yaml
new file mode 100644
index 0000000000000..d9663582e5ca2
--- /dev/null
+++ b/api/core/model_runtime/model_providers/siliconflow/llm/internlm2_5-20b-chat.yaml
@@ -0,0 +1,30 @@
+model: internlm/internlm2_5-20b-chat
+label:
+  en_US: internlm/internlm2_5-20b-chat
+model_type: llm
+features:
+  - agent-thought
+model_properties:
+  mode: chat
+  context_size: 32768
+parameter_rules:
+  - name: temperature
+    use_template: temperature
+  - name: max_tokens
+    use_template: max_tokens
+    type: int
+    default: 512
+    min: 1
+    max: 4096
+    help:
+      zh_Hans: 指定生成结果长度的上限。如果生成结果截断，可以调大该参数。
+      en_US: Specifies the upper limit on the length of generated results. If the generated results are truncated, you can increase this parameter.
+  - name: top_p
+    use_template: top_p
+  - name: frequency_penalty
+    use_template: frequency_penalty
+pricing:
+  input: '1'
+  output: '1'
+  unit: '0.000001'
+  currency: RMB
diff --git a/api/core/model_runtime/model_providers/siliconflow/llm/qwen2.5-coder-7b-instruct.yaml b/api/core/model_runtime/model_providers/siliconflow/llm/qwen2.5-coder-7b-instruct.yaml
new file mode 100644
index 0000000000000..76526200ccdcc
--- /dev/null
+++ b/api/core/model_runtime/model_providers/siliconflow/llm/qwen2.5-coder-7b-instruct.yaml
@@ -0,0 +1,74 @@
+model: Qwen/Qwen2.5-Coder-7B-Instruct
+label:
+  en_US: Qwen/Qwen2.5-Coder-7B-Instruct
+model_type: llm
+features:
+  - agent-thought
+model_properties:
+  mode: chat
+  context_size: 131072
+parameter_rules:
+  - name: temperature
+    use_template: temperature
+    type: float
+    default: 0.3
+    min: 0.0
+    max: 2.0
+    help:
+      zh_Hans: 用于控制随机性和多样性的程度。具体来说，temperature值控制了生成文本时对每个候选词的概率分布进行平滑的程度。较高的temperature值会降低概率分布的峰值，使得更多的低概率词被选择，生成结果更加多样化；而较低的temperature值则会增强概率分布的峰值，使得高概率词更容易被选择，生成结果更加确定。
+      en_US: Used to control the degree of randomness and diversity. Specifically, the temperature value controls the degree to which the probability distribution of each candidate word is smoothed when generating text. A higher temperature value will reduce the peak value of the probability distribution, allowing more low-probability words to be selected, and the generated results will be more diverse; while a lower temperature value will enhance the peak value of the probability distribution, making it easier for high-probability words to be selected. , the generated results are more certain.
+  - name: max_tokens
+    use_template: max_tokens
+    type: int
+    default: 8192
+    min: 1
+    max: 8192
+    help:
+      zh_Hans: 用于指定模型在生成内容时token的最大数量，它定义了生成的上限，但不保证每次都会生成到这个数量。
+      en_US: It is used to specify the maximum number of tokens when the model generates content. It defines the upper limit of generation, but does not guarantee that this number will be generated every time.
+  - name: top_p
+    use_template: top_p
+    type: float
+    default: 0.8
+    min: 0.1
+    max: 0.9
+    help:
+      zh_Hans: 生成过程中核采样方法概率阈值，例如，取值为0.8时，仅保留概率加起来大于等于0.8的最可能token的最小集合作为候选集。取值范围为（0,1.0)，取值越大，生成的随机性越高；取值越低，生成的确定性越高。
+      en_US: The probability threshold of the kernel sampling method during the generation process. For example, when the value is 0.8, only the smallest set of the most likely tokens with a sum of probabilities greater than or equal to 0.8 is retained as the candidate set. The value range is (0,1.0). The larger the value, the higher the randomness generated; the lower the value, the higher the certainty generated.
+  - name: top_k
+    type: int
+    min: 0
+    max: 99
+    label:
+      zh_Hans: 取样数量
+      en_US: Top k
+    help:
+      zh_Hans: 生成时，采样候选集的大小。例如，取值为50时，仅将单次生成中得分最高的50个token组成随机采样的候选集。取值越大，生成的随机性越高；取值越小，生成的确定性越高。
+      en_US: The size of the sample candidate set when generated. For example, when the value is 50, only the 50 highest-scoring tokens in a single generation form a randomly sampled candidate set. The larger the value, the higher the randomness generated; the smaller the value, the higher the certainty generated.
+  - name: seed
+    required: false
+    type: int
+    default: 1234
+    label:
+      zh_Hans: 随机种子
+      en_US: Random seed
+    help:
+      zh_Hans: 生成时使用的随机数种子，用户控制模型生成内容的随机性。支持无符号64位整数，默认值为 1234。在使用seed时，模型将尽可能生成相同或相似的结果，但目前不保证每次生成的结果完全相同。
+      en_US: The random number seed used when generating, the user controls the randomness of the content generated by the model. Supports unsigned 64-bit integers, default value is 1234. When using seed, the model will try its best to generate the same or similar results, but there is currently no guarantee that the results will be exactly the same every time.
+  - name: repetition_penalty
+    required: false
+    type: float
+    default: 1.1
+    label:
+      zh_Hans: 重复惩罚
+      en_US: Repetition penalty
+    help:
+      zh_Hans: 用于控制模型生成时的重复度。提高repetition_penalty时可以降低模型生成的重复度。1.0表示不做惩罚。
+      en_US: Used to control the repeatability when generating models. Increasing repetition_penalty can reduce the duplication of model generation. 1.0 means no punishment.
+  - name: response_format
+    use_template: response_format
+pricing:
+  input: '0'
+  output: '0'
+  unit: '0.000001'
+  currency: RMB
diff --git a/api/core/model_runtime/model_providers/siliconflow/llm/qwen2.5-math-72b-instruct.yaml b/api/core/model_runtime/model_providers/siliconflow/llm/qwen2.5-math-72b-instruct.yaml
new file mode 100644
index 0000000000000..90afa0cfd5b96
--- /dev/null
+++ b/api/core/model_runtime/model_providers/siliconflow/llm/qwen2.5-math-72b-instruct.yaml
@@ -0,0 +1,74 @@
+model: Qwen/Qwen2.5-Math-72B-Instruct
+label:
+  en_US: Qwen/Qwen2.5-Math-72B-Instruct
+model_type: llm
+features:
+  - agent-thought
+model_properties:
+  mode: chat
+  context_size: 4096
+parameter_rules:
+  - name: temperature
+    use_template: temperature
+    type: float
+    default: 0.3
+    min: 0.0
+    max: 2.0
+    help:
+      zh_Hans: 用于控制随机性和多样性的程度。具体来说，temperature值控制了生成文本时对每个候选词的概率分布进行平滑的程度。较高的temperature值会降低概率分布的峰值，使得更多的低概率词被选择，生成结果更加多样化；而较低的temperature值则会增强概率分布的峰值，使得高概率词更容易被选择，生成结果更加确定。
+      en_US: Used to control the degree of randomness and diversity. Specifically, the temperature value controls the degree to which the probability distribution of each candidate word is smoothed when generating text. A higher temperature value will reduce the peak value of the probability distribution, allowing more low-probability words to be selected, and the generated results will be more diverse; while a lower temperature value will enhance the peak value of the probability distribution, making it easier for high-probability words to be selected. , the generated results are more certain.
+  - name: max_tokens
+    use_template: max_tokens
+    type: int
+    default: 2000
+    min: 1
+    max: 2000
+    help:
+      zh_Hans: 用于指定模型在生成内容时token的最大数量，它定义了生成的上限，但不保证每次都会生成到这个数量。
+      en_US: It is used to specify the maximum number of tokens when the model generates content. It defines the upper limit of generation, but does not guarantee that this number will be generated every time.
+  - name: top_p
+    use_template: top_p
+    type: float
+    default: 0.8
+    min: 0.1
+    max: 0.9
+    help:
+      zh_Hans: 生成过程中核采样方法概率阈值，例如，取值为0.8时，仅保留概率加起来大于等于0.8的最可能token的最小集合作为候选集。取值范围为（0,1.0)，取值越大，生成的随机性越高；取值越低，生成的确定性越高。
+      en_US: The probability threshold of the kernel sampling method during the generation process. For example, when the value is 0.8, only the smallest set of the most likely tokens with a sum of probabilities greater than or equal to 0.8 is retained as the candidate set. The value range is (0,1.0). The larger the value, the higher the randomness generated; the lower the value, the higher the certainty generated.
+  - name: top_k
+    type: int
+    min: 0
+    max: 99
+    label:
+      zh_Hans: 取样数量
+      en_US: Top k
+    help:
+      zh_Hans: 生成时，采样候选集的大小。例如，取值为50时，仅将单次生成中得分最高的50个token组成随机采样的候选集。取值越大，生成的随机性越高；取值越小，生成的确定性越高。
+      en_US: The size of the sample candidate set when generated. For example, when the value is 50, only the 50 highest-scoring tokens in a single generation form a randomly sampled candidate set. The larger the value, the higher the randomness generated; the smaller the value, the higher the certainty generated.
+  - name: seed
+    required: false
+    type: int
+    default: 1234
+    label:
+      zh_Hans: 随机种子
+      en_US: Random seed
+    help:
+      zh_Hans: 生成时使用的随机数种子，用户控制模型生成内容的随机性。支持无符号64位整数，默认值为 1234。在使用seed时，模型将尽可能生成相同或相似的结果，但目前不保证每次生成的结果完全相同。
+      en_US: The random number seed used when generating, the user controls the randomness of the content generated by the model. Supports unsigned 64-bit integers, default value is 1234. When using seed, the model will try its best to generate the same or similar results, but there is currently no guarantee that the results will be exactly the same every time.
+  - name: repetition_penalty
+    required: false
+    type: float
+    default: 1.1
+    label:
+      zh_Hans: 重复惩罚
+      en_US: Repetition penalty
+    help:
+      zh_Hans: 用于控制模型生成时的重复度。提高repetition_penalty时可以降低模型生成的重复度。1.0表示不做惩罚。
+      en_US: Used to control the repeatability when generating models. Increasing repetition_penalty can reduce the duplication of model generation. 1.0 means no punishment.
+  - name: response_format
+    use_template: response_format
+pricing:
+  input: '4.13'
+  output: '4.13'
+  unit: '0.000001'
+  currency: RMB

From 850492dafaa8ef801a332d5a787cc9818c7b6072 Mon Sep 17 00:00:00 2001
From: zhuhao <37029601+hwzhuhao@users.noreply.github.com>
Date: Sat, 28 Sep 2024 21:40:27 +0800
Subject: [PATCH 100/235] feat: deprecate gte-Qwen2-7B-instruct embedding model
 (#8866)

---
 .../perfxcloud/llm/_position.yaml             | 21 +++++++++----------
 .../perfxcloud/text_embedding/_position.yaml  |  4 ++++
 .../text_embedding/gte-Qwen2-7B-instruct.yaml |  1 +
 3 files changed, 15 insertions(+), 11 deletions(-)
 create mode 100644 api/core/model_runtime/model_providers/perfxcloud/text_embedding/_position.yaml

diff --git a/api/core/model_runtime/model_providers/perfxcloud/llm/_position.yaml b/api/core/model_runtime/model_providers/perfxcloud/llm/_position.yaml
index 37bf400f1e347..c6930e54f50aa 100644
--- a/api/core/model_runtime/model_providers/perfxcloud/llm/_position.yaml
+++ b/api/core/model_runtime/model_providers/perfxcloud/llm/_position.yaml
@@ -1,24 +1,23 @@
 - Qwen2.5-72B-Instruct
 - Qwen2.5-7B-Instruct
+- Qwen2-72B-Instruct
+- Qwen2-72B-Instruct-AWQ-int4
+- Qwen2-72B-Instruct-GPTQ-Int4
+- Qwen2-7B-Instruct
+- Qwen2-7B
+- Qwen1.5-110B-Chat-GPTQ-Int4
+- Qwen1.5-72B-Chat-GPTQ-Int4
+- Qwen1.5-7B
+- Qwen-14B-Chat-Int4
 - Yi-Coder-1.5B-Chat
 - Yi-Coder-9B-Chat
-- Qwen2-72B-Instruct-AWQ-int4
 - Yi-1_5-9B-Chat-16K
-- Qwen2-7B-Instruct
 - Reflection-Llama-3.1-70B
-- Qwen2-72B-Instruct
 - Meta-Llama-3.1-8B-Instruct
-
 - Meta-Llama-3.1-405B-Instruct-AWQ-INT4
 - Meta-Llama-3-70B-Instruct-GPTQ-Int4
-- chatglm3-6b
 - Meta-Llama-3-8B-Instruct
 - Llama3-Chinese_v2
 - deepseek-v2-lite-chat
-- Qwen2-72B-Instruct-GPTQ-Int4
-- Qwen2-7B
-- Qwen-14B-Chat-Int4
-- Qwen1.5-72B-Chat-GPTQ-Int4
-- Qwen1.5-7B
-- Qwen1.5-110B-Chat-GPTQ-Int4
 - deepseek-v2-chat
+- chatglm3-6b
diff --git a/api/core/model_runtime/model_providers/perfxcloud/text_embedding/_position.yaml b/api/core/model_runtime/model_providers/perfxcloud/text_embedding/_position.yaml
new file mode 100644
index 0000000000000..99163d42931b1
--- /dev/null
+++ b/api/core/model_runtime/model_providers/perfxcloud/text_embedding/_position.yaml
@@ -0,0 +1,4 @@
+- gte-Qwen2-7B-instruct
+- BAAI/bge-large-en-v1.5
+- BAAI/bge-large-zh-v1.5
+- BAAI/bge-m3
diff --git a/api/core/model_runtime/model_providers/perfxcloud/text_embedding/gte-Qwen2-7B-instruct.yaml b/api/core/model_runtime/model_providers/perfxcloud/text_embedding/gte-Qwen2-7B-instruct.yaml
index 03db0d8bce850..161d5ea9a2657 100644
--- a/api/core/model_runtime/model_providers/perfxcloud/text_embedding/gte-Qwen2-7B-instruct.yaml
+++ b/api/core/model_runtime/model_providers/perfxcloud/text_embedding/gte-Qwen2-7B-instruct.yaml
@@ -2,3 +2,4 @@ model: gte-Qwen2-7B-instruct
 model_type: text-embedding
 model_properties:
   context_size: 2048
+deprecated: true

From f97607370a2dae9eff8bfeece505b417d232df76 Mon Sep 17 00:00:00 2001
From: zhuhao <37029601+hwzhuhao@users.noreply.github.com>
Date: Sat, 28 Sep 2024 21:41:02 +0800
Subject: [PATCH 101/235] refactor: update Callback to an abstract class
 (#8868)

---
 api/core/model_runtime/callbacks/base_callback.py | 7 ++++++-
 1 file changed, 6 insertions(+), 1 deletion(-)

diff --git a/api/core/model_runtime/callbacks/base_callback.py b/api/core/model_runtime/callbacks/base_callback.py
index 92da53c9a464d..6bd9325785a2d 100644
--- a/api/core/model_runtime/callbacks/base_callback.py
+++ b/api/core/model_runtime/callbacks/base_callback.py
@@ -1,3 +1,4 @@
+from abc import ABC, abstractmethod
 from typing import Optional
 
 from core.model_runtime.entities.llm_entities import LLMResult, LLMResultChunk
@@ -13,7 +14,7 @@
 }
 
 
-class Callback:
+class Callback(ABC):
     """
     Base class for callbacks.
     Only for LLM.
@@ -21,6 +22,7 @@ class Callback:
 
     raise_error: bool = False
 
+    @abstractmethod
     def on_before_invoke(
         self,
         llm_instance: AIModel,
@@ -48,6 +50,7 @@ def on_before_invoke(
         """
         raise NotImplementedError()
 
+    @abstractmethod
     def on_new_chunk(
         self,
         llm_instance: AIModel,
@@ -77,6 +80,7 @@ def on_new_chunk(
         """
         raise NotImplementedError()
 
+    @abstractmethod
     def on_after_invoke(
         self,
         llm_instance: AIModel,
@@ -106,6 +110,7 @@ def on_after_invoke(
         """
         raise NotImplementedError()
 
+    @abstractmethod
     def on_invoke_error(
         self,
         llm_instance: AIModel,

From 74f58f29f9a3070152b0bb097eb2bebb9f9cd19e Mon Sep 17 00:00:00 2001
From: Bowen Liang <liangbowen@gf.com.cn>
Date: Sun, 29 Sep 2024 00:29:59 +0800
Subject: [PATCH 102/235] chore: bump ruff to 0.6.8 for fixing violation in
 SIM910 (#8869)

---
 .../zhipuai/zhipuai_sdk/core/_utils/_utils.py |  2 +-
 .../rag/datasource/keyword/jieba/jieba.py     |  2 +-
 .../event_handlers/create_document_index.py   |  2 +-
 api/poetry.lock                               | 40 +++++++++----------
 api/pyproject.toml                            |  2 +-
 .../integration_tests/tools/__mock/http.py    |  2 +-
 .../workflow/nodes/__mock/http.py             |  4 +-
 7 files changed, 27 insertions(+), 27 deletions(-)

diff --git a/api/core/model_runtime/model_providers/zhipuai/zhipuai_sdk/core/_utils/_utils.py b/api/core/model_runtime/model_providers/zhipuai/zhipuai_sdk/core/_utils/_utils.py
index ce5e7786aa293..3a7b234ab0c06 100644
--- a/api/core/model_runtime/model_providers/zhipuai/zhipuai_sdk/core/_utils/_utils.py
+++ b/api/core/model_runtime/model_providers/zhipuai/zhipuai_sdk/core/_utils/_utils.py
@@ -248,7 +248,7 @@ def inner(func: CallableT) -> CallableT:
         @functools.wraps(func)
         def wrapper(*args: object, **kwargs: object) -> object:
             given_params: set[str] = set()
-            for i, _ in enumerate(args):
+            for i in range(len(args)):
                 try:
                     given_params.add(positional[i])
                 except IndexError:
diff --git a/api/core/rag/datasource/keyword/jieba/jieba.py b/api/core/rag/datasource/keyword/jieba/jieba.py
index 3073100746d36..a0153c1e58a1a 100644
--- a/api/core/rag/datasource/keyword/jieba/jieba.py
+++ b/api/core/rag/datasource/keyword/jieba/jieba.py
@@ -45,7 +45,7 @@ def add_texts(self, texts: list[Document], **kwargs):
             keyword_table_handler = JiebaKeywordTableHandler()
 
             keyword_table = self._get_dataset_keyword_table()
-            keywords_list = kwargs.get("keywords_list", None)
+            keywords_list = kwargs.get("keywords_list")
             for i in range(len(texts)):
                 text = texts[i]
                 if keywords_list:
diff --git a/api/events/event_handlers/create_document_index.py b/api/events/event_handlers/create_document_index.py
index 54f6a76e16723..5af45e1e5026d 100644
--- a/api/events/event_handlers/create_document_index.py
+++ b/api/events/event_handlers/create_document_index.py
@@ -14,7 +14,7 @@
 @document_index_created.connect
 def handle(sender, **kwargs):
     dataset_id = sender
-    document_ids = kwargs.get("document_ids", None)
+    document_ids = kwargs.get("document_ids")
     documents = []
     start_at = time.perf_counter()
     for document_id in document_ids:
diff --git a/api/poetry.lock b/api/poetry.lock
index bce21fb547c8c..85c68cd75f129 100644
--- a/api/poetry.lock
+++ b/api/poetry.lock
@@ -8074,29 +8074,29 @@ pyasn1 = ">=0.1.3"
 
 [[package]]
 name = "ruff"
-version = "0.6.5"
+version = "0.6.8"
 description = "An extremely fast Python linter and code formatter, written in Rust."
 optional = false
 python-versions = ">=3.7"
 files = [
-    {file = "ruff-0.6.5-py3-none-linux_armv6l.whl", hash = "sha256:7e4e308f16e07c95fc7753fc1aaac690a323b2bb9f4ec5e844a97bb7fbebd748"},
-    {file = "ruff-0.6.5-py3-none-macosx_10_12_x86_64.whl", hash = "sha256:932cd69eefe4daf8c7d92bd6689f7e8182571cb934ea720af218929da7bd7d69"},
-    {file = "ruff-0.6.5-py3-none-macosx_11_0_arm64.whl", hash = "sha256:3a8d42d11fff8d3143ff4da41742a98f8f233bf8890e9fe23077826818f8d680"},
-    {file = "ruff-0.6.5-py3-none-manylinux_2_17_aarch64.manylinux2014_aarch64.whl", hash = "sha256:a50af6e828ee692fb10ff2dfe53f05caecf077f4210fae9677e06a808275754f"},
-    {file = "ruff-0.6.5-py3-none-manylinux_2_17_armv7l.manylinux2014_armv7l.whl", hash = "sha256:794ada3400a0d0b89e3015f1a7e01f4c97320ac665b7bc3ade24b50b54cb2972"},
-    {file = "ruff-0.6.5-py3-none-manylinux_2_17_i686.manylinux2014_i686.whl", hash = "sha256:381413ec47f71ce1d1c614f7779d88886f406f1fd53d289c77e4e533dc6ea200"},
-    {file = "ruff-0.6.5-py3-none-manylinux_2_17_ppc64.manylinux2014_ppc64.whl", hash = "sha256:52e75a82bbc9b42e63c08d22ad0ac525117e72aee9729a069d7c4f235fc4d276"},
-    {file = "ruff-0.6.5-py3-none-manylinux_2_17_ppc64le.manylinux2014_ppc64le.whl", hash = "sha256:09c72a833fd3551135ceddcba5ebdb68ff89225d30758027280968c9acdc7810"},
-    {file = "ruff-0.6.5-py3-none-manylinux_2_17_s390x.manylinux2014_s390x.whl", hash = "sha256:800c50371bdcb99b3c1551d5691e14d16d6f07063a518770254227f7f6e8c178"},
-    {file = "ruff-0.6.5-py3-none-manylinux_2_17_x86_64.manylinux2014_x86_64.whl", hash = "sha256:8e25ddd9cd63ba1f3bd51c1f09903904a6adf8429df34f17d728a8fa11174253"},
-    {file = "ruff-0.6.5-py3-none-musllinux_1_2_aarch64.whl", hash = "sha256:7291e64d7129f24d1b0c947ec3ec4c0076e958d1475c61202497c6aced35dd19"},
-    {file = "ruff-0.6.5-py3-none-musllinux_1_2_armv7l.whl", hash = "sha256:9ad7dfbd138d09d9a7e6931e6a7e797651ce29becd688be8a0d4d5f8177b4b0c"},
-    {file = "ruff-0.6.5-py3-none-musllinux_1_2_i686.whl", hash = "sha256:005256d977021790cc52aa23d78f06bb5090dc0bfbd42de46d49c201533982ae"},
-    {file = "ruff-0.6.5-py3-none-musllinux_1_2_x86_64.whl", hash = "sha256:482c1e6bfeb615eafc5899127b805d28e387bd87db38b2c0c41d271f5e58d8cc"},
-    {file = "ruff-0.6.5-py3-none-win32.whl", hash = "sha256:cf4d3fa53644137f6a4a27a2b397381d16454a1566ae5335855c187fbf67e4f5"},
-    {file = "ruff-0.6.5-py3-none-win_amd64.whl", hash = "sha256:3e42a57b58e3612051a636bc1ac4e6b838679530235520e8f095f7c44f706ff9"},
-    {file = "ruff-0.6.5-py3-none-win_arm64.whl", hash = "sha256:51935067740773afdf97493ba9b8231279e9beef0f2a8079188c4776c25688e0"},
-    {file = "ruff-0.6.5.tar.gz", hash = "sha256:4d32d87fab433c0cf285c3683dd4dae63be05fd7a1d65b3f5bf7cdd05a6b96fb"},
+    {file = "ruff-0.6.8-py3-none-linux_armv6l.whl", hash = "sha256:77944bca110ff0a43b768f05a529fecd0706aac7bcce36d7f1eeb4cbfca5f0f2"},
+    {file = "ruff-0.6.8-py3-none-macosx_10_12_x86_64.whl", hash = "sha256:27b87e1801e786cd6ede4ada3faa5e254ce774de835e6723fd94551464c56b8c"},
+    {file = "ruff-0.6.8-py3-none-macosx_11_0_arm64.whl", hash = "sha256:cd48f945da2a6334f1793d7f701725a76ba93bf3d73c36f6b21fb04d5338dcf5"},
+    {file = "ruff-0.6.8-py3-none-manylinux_2_17_aarch64.manylinux2014_aarch64.whl", hash = "sha256:677e03c00f37c66cea033274295a983c7c546edea5043d0c798833adf4cf4c6f"},
+    {file = "ruff-0.6.8-py3-none-manylinux_2_17_armv7l.manylinux2014_armv7l.whl", hash = "sha256:9f1476236b3eacfacfc0f66aa9e6cd39f2a624cb73ea99189556015f27c0bdeb"},
+    {file = "ruff-0.6.8-py3-none-manylinux_2_17_i686.manylinux2014_i686.whl", hash = "sha256:6f5a2f17c7d32991169195d52a04c95b256378bbf0de8cb98478351eb70d526f"},
+    {file = "ruff-0.6.8-py3-none-manylinux_2_17_ppc64.manylinux2014_ppc64.whl", hash = "sha256:5fd0d4b7b1457c49e435ee1e437900ced9b35cb8dc5178921dfb7d98d65a08d0"},
+    {file = "ruff-0.6.8-py3-none-manylinux_2_17_ppc64le.manylinux2014_ppc64le.whl", hash = "sha256:f8034b19b993e9601f2ddf2c517451e17a6ab5cdb1c13fdff50c1442a7171d87"},
+    {file = "ruff-0.6.8-py3-none-manylinux_2_17_s390x.manylinux2014_s390x.whl", hash = "sha256:6cfb227b932ba8ef6e56c9f875d987973cd5e35bc5d05f5abf045af78ad8e098"},
+    {file = "ruff-0.6.8-py3-none-manylinux_2_17_x86_64.manylinux2014_x86_64.whl", hash = "sha256:6ef0411eccfc3909269fed47c61ffebdcb84a04504bafa6b6df9b85c27e813b0"},
+    {file = "ruff-0.6.8-py3-none-musllinux_1_2_aarch64.whl", hash = "sha256:007dee844738c3d2e6c24ab5bc7d43c99ba3e1943bd2d95d598582e9c1b27750"},
+    {file = "ruff-0.6.8-py3-none-musllinux_1_2_armv7l.whl", hash = "sha256:ce60058d3cdd8490e5e5471ef086b3f1e90ab872b548814e35930e21d848c9ce"},
+    {file = "ruff-0.6.8-py3-none-musllinux_1_2_i686.whl", hash = "sha256:1085c455d1b3fdb8021ad534379c60353b81ba079712bce7a900e834859182fa"},
+    {file = "ruff-0.6.8-py3-none-musllinux_1_2_x86_64.whl", hash = "sha256:70edf6a93b19481affd287d696d9e311388d808671bc209fb8907b46a8c3af44"},
+    {file = "ruff-0.6.8-py3-none-win32.whl", hash = "sha256:792213f7be25316f9b46b854df80a77e0da87ec66691e8f012f887b4a671ab5a"},
+    {file = "ruff-0.6.8-py3-none-win_amd64.whl", hash = "sha256:ec0517dc0f37cad14a5319ba7bba6e7e339d03fbf967a6d69b0907d61be7a263"},
+    {file = "ruff-0.6.8-py3-none-win_arm64.whl", hash = "sha256:8d3bb2e3fbb9875172119021a13eed38849e762499e3cfde9588e4b4d70968dc"},
+    {file = "ruff-0.6.8.tar.gz", hash = "sha256:a5bf44b1aa0adaf6d9d20f86162b34f7c593bfedabc51239953e446aefc8ce18"},
 ]
 
 [[package]]
@@ -10501,4 +10501,4 @@ cffi = ["cffi (>=1.11)"]
 [metadata]
 lock-version = "2.0"
 python-versions = ">=3.10,<3.13"
-content-hash = "69b42bb1ff033f14e199fee8335356275099421d72bbd7037b7a991ea65cae08"
+content-hash = "c4580c22e2b220c8c80dbc3f765060a09e14874ed29b690c13a533bf0365e789"
diff --git a/api/pyproject.toml b/api/pyproject.toml
index f004865d5f343..64b35621b2ec6 100644
--- a/api/pyproject.toml
+++ b/api/pyproject.toml
@@ -286,4 +286,4 @@ optional = true
 
 [tool.poetry.group.lint.dependencies]
 dotenv-linter = "~0.5.0"
-ruff = "~0.6.5"
+ruff = "~0.6.8"
diff --git a/api/tests/integration_tests/tools/__mock/http.py b/api/tests/integration_tests/tools/__mock/http.py
index d3c1f3101c54b..42cf87e317ab6 100644
--- a/api/tests/integration_tests/tools/__mock/http.py
+++ b/api/tests/integration_tests/tools/__mock/http.py
@@ -17,7 +17,7 @@ def httpx_request(
         request = httpx.Request(
             method, url, params=kwargs.get("params"), headers=kwargs.get("headers"), cookies=kwargs.get("cookies")
         )
-        data = kwargs.get("data", None)
+        data = kwargs.get("data")
         resp = json.dumps(data).encode("utf-8") if data else b"OK"
         response = httpx.Response(
             status_code=200,
diff --git a/api/tests/integration_tests/workflow/nodes/__mock/http.py b/api/tests/integration_tests/workflow/nodes/__mock/http.py
index f1ab23b0026aa..ec013183b7601 100644
--- a/api/tests/integration_tests/workflow/nodes/__mock/http.py
+++ b/api/tests/integration_tests/workflow/nodes/__mock/http.py
@@ -22,8 +22,8 @@ def httpx_request(
             return response
 
         # get data, files
-        data = kwargs.get("data", None)
-        files = kwargs.get("files", None)
+        data = kwargs.get("data")
+        files = kwargs.get("files")
         if data is not None:
             resp = dumps(data).encode("utf-8")
         elif files is not None:

From 4ec977eabafefc3859ef04425ec669ce87b5483e Mon Sep 17 00:00:00 2001
From: -LAN- <laipz8200@outlook.com>
Date: Sun, 29 Sep 2024 16:12:42 +0800
Subject: [PATCH 103/235] fix(workflow): update tagging logic in GitHub Actions
 (#8882)

---
 .github/workflows/build-push.yml | 2 +-
 1 file changed, 1 insertion(+), 1 deletion(-)

diff --git a/.github/workflows/build-push.yml b/.github/workflows/build-push.yml
index 407bd47d9b0f8..6daaaf5791dd2 100644
--- a/.github/workflows/build-push.yml
+++ b/.github/workflows/build-push.yml
@@ -125,7 +125,7 @@ jobs:
         with:
           images: ${{ env[matrix.image_name_env] }}
           tags: |
-            type=raw,value=latest,enable=${{ startsWith(github.ref, 'refs/tags/') }}
+            type=raw,value=latest,enable=${{ startsWith(github.ref, 'refs/tags/') && !contains(github.ref, '-') }}
             type=ref,event=branch
             type=sha,enable=true,priority=100,prefix=,suffix=,format=long
             type=raw,value=${{ github.ref_name }},enable=${{ startsWith(github.ref, 'refs/tags/') }}

From e4ed916baadb10ffd22ec1772e56b4fcefa1b57f Mon Sep 17 00:00:00 2001
From: longzhihun <38651850@qq.com>
Date: Sun, 29 Sep 2024 16:12:56 +0800
Subject: [PATCH 104/235] Add Jamba and Llama3.2 model support (#8878)

---
 .../bedrock/llm/_position.yaml                |  6 ++++
 .../llm/ai21.jamba-1-5-large-v1.0.yaml        | 26 ++++++++++++++++
 .../bedrock/llm/ai21.jamba-1-5-mini-v1.0.yaml | 26 ++++++++++++++++
 .../model_providers/bedrock/llm/llm.py        |  2 ++
 .../us.meta.llama3-2-11b-instruct-v1.0.yaml   | 29 +++++++++++++++++
 .../us.meta.llama3-2-1b-instruct-v1.0.yaml    | 26 ++++++++++++++++
 .../us.meta.llama3-2-3b-instruct-v1.0.yaml    | 26 ++++++++++++++++
 .../us.meta.llama3-2-90b-instruct-v1.0.yaml   | 31 +++++++++++++++++++
 8 files changed, 172 insertions(+)
 create mode 100644 api/core/model_runtime/model_providers/bedrock/llm/ai21.jamba-1-5-large-v1.0.yaml
 create mode 100644 api/core/model_runtime/model_providers/bedrock/llm/ai21.jamba-1-5-mini-v1.0.yaml
 create mode 100644 api/core/model_runtime/model_providers/bedrock/llm/us.meta.llama3-2-11b-instruct-v1.0.yaml
 create mode 100644 api/core/model_runtime/model_providers/bedrock/llm/us.meta.llama3-2-1b-instruct-v1.0.yaml
 create mode 100644 api/core/model_runtime/model_providers/bedrock/llm/us.meta.llama3-2-3b-instruct-v1.0.yaml
 create mode 100644 api/core/model_runtime/model_providers/bedrock/llm/us.meta.llama3-2-90b-instruct-v1.0.yaml

diff --git a/api/core/model_runtime/model_providers/bedrock/llm/_position.yaml b/api/core/model_runtime/model_providers/bedrock/llm/_position.yaml
index 86c8061deefac..47e2b020fd09a 100644
--- a/api/core/model_runtime/model_providers/bedrock/llm/_position.yaml
+++ b/api/core/model_runtime/model_providers/bedrock/llm/_position.yaml
@@ -6,6 +6,8 @@
 - anthropic.claude-v2:1
 - anthropic.claude-3-sonnet-v1:0
 - anthropic.claude-3-haiku-v1:0
+- ai21.jamba-1-5-large-v1:0
+- ai21.jamba-1-5-mini-v1:0
 - cohere.command-light-text-v14
 - cohere.command-text-v14
 - cohere.command-r-plus-v1.0
@@ -15,6 +17,10 @@
 - meta.llama3-1-405b-instruct-v1:0
 - meta.llama3-8b-instruct-v1:0
 - meta.llama3-70b-instruct-v1:0
+- us.meta.llama3-2-1b-instruct-v1:0
+- us.meta.llama3-2-3b-instruct-v1:0
+- us.meta.llama3-2-11b-instruct-v1:0
+- us.meta.llama3-2-90b-instruct-v1:0
 - meta.llama2-13b-chat-v1
 - meta.llama2-70b-chat-v1
 - mistral.mistral-large-2407-v1:0
diff --git a/api/core/model_runtime/model_providers/bedrock/llm/ai21.jamba-1-5-large-v1.0.yaml b/api/core/model_runtime/model_providers/bedrock/llm/ai21.jamba-1-5-large-v1.0.yaml
new file mode 100644
index 0000000000000..276c7312cee00
--- /dev/null
+++ b/api/core/model_runtime/model_providers/bedrock/llm/ai21.jamba-1-5-large-v1.0.yaml
@@ -0,0 +1,26 @@
+model: ai21.jamba-1-5-large-v1:0
+label:
+  en_US: Jamba 1.5 Large
+model_type: llm
+model_properties:
+  mode: completion
+  context_size: 256000
+parameter_rules:
+  - name: temperature
+    use_template: temperature
+    default: 1
+    min: 0.0
+    max: 2.0
+  - name: top_p
+    use_template: top_p
+  - name: max_gen_len
+    use_template: max_tokens
+    required: true
+    default: 4096
+    min: 1
+    max: 4096
+pricing:
+  input: '0.002'
+  output: '0.008'
+  unit: '0.001'
+  currency: USD
diff --git a/api/core/model_runtime/model_providers/bedrock/llm/ai21.jamba-1-5-mini-v1.0.yaml b/api/core/model_runtime/model_providers/bedrock/llm/ai21.jamba-1-5-mini-v1.0.yaml
new file mode 100644
index 0000000000000..3461d8ab71329
--- /dev/null
+++ b/api/core/model_runtime/model_providers/bedrock/llm/ai21.jamba-1-5-mini-v1.0.yaml
@@ -0,0 +1,26 @@
+model: ai21.jamba-1-5-mini-v1:0
+label:
+  en_US: Jamba 1.5 Mini
+model_type: llm
+model_properties:
+  mode: completion
+  context_size: 256000
+parameter_rules:
+  - name: temperature
+    use_template: temperature
+    default: 1
+    min: 0.0
+    max: 2.0
+  - name: top_p
+    use_template: top_p
+  - name: max_gen_len
+    use_template: max_tokens
+    required: true
+    default: 4096
+    min: 1
+    max: 4096
+pricing:
+  input: '0.0002'
+  output: '0.0004'
+  unit: '0.001'
+  currency: USD
diff --git a/api/core/model_runtime/model_providers/bedrock/llm/llm.py b/api/core/model_runtime/model_providers/bedrock/llm/llm.py
index 77bab0c294588..d1961784f2e84 100644
--- a/api/core/model_runtime/model_providers/bedrock/llm/llm.py
+++ b/api/core/model_runtime/model_providers/bedrock/llm/llm.py
@@ -63,6 +63,7 @@ class BedrockLargeLanguageModel(LargeLanguageModel):
         {"prefix": "us.anthropic.claude-3", "support_system_prompts": True, "support_tool_use": True},
         {"prefix": "eu.anthropic.claude-3", "support_system_prompts": True, "support_tool_use": True},
         {"prefix": "anthropic.claude-3", "support_system_prompts": True, "support_tool_use": True},
+        {"prefix": "us.meta.llama3-2", "support_system_prompts": True, "support_tool_use": True},
         {"prefix": "meta.llama", "support_system_prompts": True, "support_tool_use": False},
         {"prefix": "mistral.mistral-7b-instruct", "support_system_prompts": False, "support_tool_use": False},
         {"prefix": "mistral.mixtral-8x7b-instruct", "support_system_prompts": False, "support_tool_use": False},
@@ -70,6 +71,7 @@ class BedrockLargeLanguageModel(LargeLanguageModel):
         {"prefix": "mistral.mistral-small", "support_system_prompts": True, "support_tool_use": True},
         {"prefix": "cohere.command-r", "support_system_prompts": True, "support_tool_use": True},
         {"prefix": "amazon.titan", "support_system_prompts": False, "support_tool_use": False},
+        {"prefix": "ai21.jamba-1-5", "support_system_prompts": True, "support_tool_use": False},
     ]
 
     @staticmethod
diff --git a/api/core/model_runtime/model_providers/bedrock/llm/us.meta.llama3-2-11b-instruct-v1.0.yaml b/api/core/model_runtime/model_providers/bedrock/llm/us.meta.llama3-2-11b-instruct-v1.0.yaml
new file mode 100644
index 0000000000000..029f428776e0b
--- /dev/null
+++ b/api/core/model_runtime/model_providers/bedrock/llm/us.meta.llama3-2-11b-instruct-v1.0.yaml
@@ -0,0 +1,29 @@
+model: us.meta.llama3-2-11b-instruct-v1:0
+label:
+  en_US: US Meta Llama 3.2 11B Instruct
+model_type: llm
+features:
+  - vision
+  - tool-call
+model_properties:
+  mode: completion
+  context_size: 128000
+parameter_rules:
+  - name: temperature
+    use_template: temperature
+    default: 0.5
+    min: 0.0
+    max: 1
+  - name: top_p
+    use_template: top_p
+  - name: max_gen_len
+    use_template: max_tokens
+    required: true
+    default: 512
+    min: 1
+    max: 2048
+pricing:
+  input: '0.00035'
+  output: '0.00035'
+  unit: '0.001'
+  currency: USD
diff --git a/api/core/model_runtime/model_providers/bedrock/llm/us.meta.llama3-2-1b-instruct-v1.0.yaml b/api/core/model_runtime/model_providers/bedrock/llm/us.meta.llama3-2-1b-instruct-v1.0.yaml
new file mode 100644
index 0000000000000..51c8474e54846
--- /dev/null
+++ b/api/core/model_runtime/model_providers/bedrock/llm/us.meta.llama3-2-1b-instruct-v1.0.yaml
@@ -0,0 +1,26 @@
+model: us.meta.llama3-2-1b-instruct-v1:0
+label:
+  en_US: US Meta Llama 3.2 1B Instruct
+model_type: llm
+model_properties:
+  mode: completion
+  context_size: 128000
+parameter_rules:
+  - name: temperature
+    use_template: temperature
+    default: 0.5
+    min: 0.0
+    max: 1
+  - name: top_p
+    use_template: top_p
+  - name: max_gen_len
+    use_template: max_tokens
+    required: true
+    default: 512
+    min: 1
+    max: 2048
+pricing:
+  input: '0.0001'
+  output: '0.0001'
+  unit: '0.001'
+  currency: USD
diff --git a/api/core/model_runtime/model_providers/bedrock/llm/us.meta.llama3-2-3b-instruct-v1.0.yaml b/api/core/model_runtime/model_providers/bedrock/llm/us.meta.llama3-2-3b-instruct-v1.0.yaml
new file mode 100644
index 0000000000000..472cc7403e2d3
--- /dev/null
+++ b/api/core/model_runtime/model_providers/bedrock/llm/us.meta.llama3-2-3b-instruct-v1.0.yaml
@@ -0,0 +1,26 @@
+model: us.meta.llama3-2-3b-instruct-v1:0
+label:
+  en_US: US Meta Llama 3.2 3B Instruct
+model_type: llm
+model_properties:
+  mode: completion
+  context_size: 128000
+parameter_rules:
+  - name: temperature
+    use_template: temperature
+    default: 0.5
+    min: 0.0
+    max: 1
+  - name: top_p
+    use_template: top_p
+  - name: max_gen_len
+    use_template: max_tokens
+    required: true
+    default: 512
+    min: 1
+    max: 2048
+pricing:
+  input: '0.00015'
+  output: '0.00015'
+  unit: '0.001'
+  currency: USD
diff --git a/api/core/model_runtime/model_providers/bedrock/llm/us.meta.llama3-2-90b-instruct-v1.0.yaml b/api/core/model_runtime/model_providers/bedrock/llm/us.meta.llama3-2-90b-instruct-v1.0.yaml
new file mode 100644
index 0000000000000..cecd0236ca9e3
--- /dev/null
+++ b/api/core/model_runtime/model_providers/bedrock/llm/us.meta.llama3-2-90b-instruct-v1.0.yaml
@@ -0,0 +1,31 @@
+model: us.meta.llama3-2-90b-instruct-v1:0
+label:
+  en_US: US Meta Llama 3.2 90B Instruct
+model_type: llm
+features:
+  - tool-call
+model_properties:
+  mode: completion
+  context_size: 128000
+parameter_rules:
+  - name: temperature
+    use_template: temperature
+    default: 0.5
+    min: 0.0
+    max: 1
+  - name: top_p
+    use_template: top_p
+    default: 0.9
+    min: 0
+    max: 1
+  - name: max_gen_len
+    use_template: max_tokens
+    required: true
+    default: 512
+    min: 1
+    max: 2048
+pricing:
+  input: '0.002'
+  output: '0.002'
+  unit: '0.001'
+  currency: USD

From c531b4a911f34357e2beeff77bdc5e1ecb4333de Mon Sep 17 00:00:00 2001
From: chenxu9741 <chenxu9741@cvte.com>
Date: Sun, 29 Sep 2024 16:13:20 +0800
Subject: [PATCH 105/235] =?UTF-8?q?fix:=20#8843=20event:=20tts=5Fmessage?=
 =?UTF-8?q?=5Fend=20always=20return=20in=20api=20streaming=20resp=E2=80=A6?=
 =?UTF-8?q?=20(#8846)?=
MIME-Version: 1.0
Content-Type: text/plain; charset=UTF-8
Content-Transfer-Encoding: 8bit

---
 api/core/app/apps/advanced_chat/generate_task_pipeline.py      | 3 ++-
 api/core/app/apps/workflow/generate_task_pipeline.py           | 3 ++-
 .../app/task_pipeline/easy_ui_based_generate_task_pipeline.py  | 3 ++-
 .../model_providers/zhipuai/zhipuai_sdk/core/_base_models.py   | 2 +-
 4 files changed, 7 insertions(+), 4 deletions(-)

diff --git a/api/core/app/apps/advanced_chat/generate_task_pipeline.py b/api/core/app/apps/advanced_chat/generate_task_pipeline.py
index 94206a1b1cd67..897b6fd063c76 100644
--- a/api/core/app/apps/advanced_chat/generate_task_pipeline.py
+++ b/api/core/app/apps/advanced_chat/generate_task_pipeline.py
@@ -231,7 +231,8 @@ def _wrapper_process_stream_response(
             except Exception as e:
                 logger.error(e)
                 break
-        yield MessageAudioEndStreamResponse(audio="", task_id=task_id)
+        if tts_publisher:
+            yield MessageAudioEndStreamResponse(audio="", task_id=task_id)
 
     def _process_stream_response(
         self,
diff --git a/api/core/app/apps/workflow/generate_task_pipeline.py b/api/core/app/apps/workflow/generate_task_pipeline.py
index 93edf8e0e882b..798847a5070b8 100644
--- a/api/core/app/apps/workflow/generate_task_pipeline.py
+++ b/api/core/app/apps/workflow/generate_task_pipeline.py
@@ -212,7 +212,8 @@ def _wrapper_process_stream_response(
             except Exception as e:
                 logger.error(e)
                 break
-        yield MessageAudioEndStreamResponse(audio="", task_id=task_id)
+        if tts_publisher:
+            yield MessageAudioEndStreamResponse(audio="", task_id=task_id)
 
     def _process_stream_response(
         self,
diff --git a/api/core/app/task_pipeline/easy_ui_based_generate_task_pipeline.py b/api/core/app/task_pipeline/easy_ui_based_generate_task_pipeline.py
index 8f834b6458ea4..917649f34e769 100644
--- a/api/core/app/task_pipeline/easy_ui_based_generate_task_pipeline.py
+++ b/api/core/app/task_pipeline/easy_ui_based_generate_task_pipeline.py
@@ -248,7 +248,8 @@ def _wrapper_process_stream_response(
             else:
                 start_listener_time = time.time()
                 yield MessageAudioStreamResponse(audio=audio.audio, task_id=task_id)
-        yield MessageAudioEndStreamResponse(audio="", task_id=task_id)
+        if publisher:
+            yield MessageAudioEndStreamResponse(audio="", task_id=task_id)
 
     def _process_stream_response(
         self, publisher: AppGeneratorTTSPublisher, trace_manager: Optional[TraceQueueManager] = None
diff --git a/api/core/model_runtime/model_providers/zhipuai/zhipuai_sdk/core/_base_models.py b/api/core/model_runtime/model_providers/zhipuai/zhipuai_sdk/core/_base_models.py
index 6d8ba700b7b1d..69b1d3a83dfef 100644
--- a/api/core/model_runtime/model_providers/zhipuai/zhipuai_sdk/core/_base_models.py
+++ b/api/core/model_runtime/model_providers/zhipuai/zhipuai_sdk/core/_base_models.py
@@ -48,7 +48,7 @@
 )
 
 if TYPE_CHECKING:
-    from pydantic_core.core_schema import LiteralSchema, ModelField, ModelFieldsSchema
+    from pydantic_core.core_schema import ModelField
 
 __all__ = ["BaseModel", "GenericModel"]
 _BaseModelT = TypeVar("_BaseModelT", bound="BaseModel")

From 42dfde65468ec8b945eb086d88441baf1fc22e2d Mon Sep 17 00:00:00 2001
From: zhuhao <37029601+hwzhuhao@users.noreply.github.com>
Date: Sun, 29 Sep 2024 16:16:56 +0800
Subject: [PATCH 106/235] docs: add english versions for the files
 customizable_model_scale_out and predefined_model_scale_out (#8871)

---
 .../en_US/customizable_model_scale_out.md     | 310 ++++++++++++++++++
 .../docs/en_US/images/index/image-1.png       | Bin 0 -> 235102 bytes
 .../docs/en_US/images/index/image-2.png       | Bin 0 -> 210087 bytes
 .../docs/en_US/images/index/image-3.png       | Bin 0 -> 44778 bytes
 .../docs/en_US/images/index/image.png         | Bin 0 -> 267979 bytes
 .../docs/en_US/predefined_model_scale_out.md  | 173 ++++++++++
 .../docs/en_US/provider_scale_out.md          |   2 +-
 .../docs/zh_Hans/provider_scale_out.md        |   2 +-
 8 files changed, 485 insertions(+), 2 deletions(-)
 create mode 100644 api/core/model_runtime/docs/en_US/customizable_model_scale_out.md
 create mode 100644 api/core/model_runtime/docs/en_US/images/index/image-1.png
 create mode 100644 api/core/model_runtime/docs/en_US/images/index/image-2.png
 create mode 100644 api/core/model_runtime/docs/en_US/images/index/image-3.png
 create mode 100644 api/core/model_runtime/docs/en_US/images/index/image.png
 create mode 100644 api/core/model_runtime/docs/en_US/predefined_model_scale_out.md

diff --git a/api/core/model_runtime/docs/en_US/customizable_model_scale_out.md b/api/core/model_runtime/docs/en_US/customizable_model_scale_out.md
new file mode 100644
index 0000000000000..f5b806ade6f49
--- /dev/null
+++ b/api/core/model_runtime/docs/en_US/customizable_model_scale_out.md
@@ -0,0 +1,310 @@
+## Custom Integration of Pre-defined Models
+
+### Introduction
+
+After completing the vendors integration, the next step is to connect the vendor's models. To illustrate the entire connection process, we will use Xinference as an example to demonstrate a complete vendor integration.
+
+It is important to note that for custom models, each model connection requires a complete vendor credential.
+
+Unlike pre-defined models, a custom vendor integration always includes the following two parameters, which do not need to be defined in the vendor YAML file.
+
+![](images/index/image-3.png)
+
+As mentioned earlier, vendors do not need to implement validate_provider_credential. The runtime will automatically call the corresponding model layer's validate_credentials to validate the credentials based on the model type and name selected by the user.
+
+### Writing the Vendor YAML
+
+First, we need to identify the types of models supported by the vendor we are integrating.
+
+Currently supported model types are as follows:
+
+- `llm` Text Generation Models
+
+- `text_embedding` Text Embedding Models
+
+- `rerank` Rerank Models
+
+- `speech2text` Speech-to-Text
+
+- `tts` Text-to-Speech
+
+- `moderation` Moderation
+
+Xinference supports LLM, Text Embedding, and Rerank. So we will start by writing xinference.yaml.
+
+```yaml
+provider: xinference #Define the vendor identifier
+label: # Vendor display name, supports both en_US (English) and zh_Hans (Simplified Chinese). If zh_Hans is not set, it will use en_US by default.
+  en_US: Xorbits Inference
+icon_small: # Small icon, refer to other vendors' icons stored in the _assets directory within the vendor implementation directory; follows the same language policy as the label
+  en_US: icon_s_en.svg
+icon_large: # Large icon
+  en_US: icon_l_en.svg
+help: # Help information
+  title:
+    en_US: How to deploy Xinference
+    zh_Hans: 如何部署 Xinference
+  url:
+    en_US: https://github.com/xorbitsai/inference
+supported_model_types: # Supported model types. Xinference supports LLM, Text Embedding, and Rerank
+- llm
+- text-embedding
+- rerank
+configurate_methods: # Since Xinference is a locally deployed vendor with no predefined models, users need to deploy whatever models they need according to Xinference documentation. Thus, it only supports custom models.
+- customizable-model
+provider_credential_schema:
+  credential_form_schemas:
+```
+
+
+Then, we need to determine what credentials are required to define a model in Xinference.
+
+- Since it supports three different types of models, we need to specify the model_type to denote the model type. Here is how we can define it:
+
+```yaml
+provider_credential_schema:
+  credential_form_schemas:
+  - variable: model_type
+    type: select
+    label:
+      en_US: Model type
+      zh_Hans: 模型类型
+    required: true
+    options:
+    - value: text-generation
+      label:
+        en_US: Language Model
+        zh_Hans: 语言模型
+    - value: embeddings
+      label:
+        en_US: Text Embedding
+    - value: reranking
+      label:
+        en_US: Rerank
+```
+
+- Next, each model has its own model_name, so we need to define that here:
+
+```yaml
+  - variable: model_name
+    type: text-input
+    label:
+      en_US: Model name
+      zh_Hans: 模型名称
+    required: true
+    placeholder:
+      zh_Hans: 填写模型名称
+      en_US: Input model name
+```
+
+- Specify the Xinference local deployment address:
+
+```yaml
+  - variable: server_url
+    label:
+      zh_Hans: 服务器URL
+      en_US: Server url
+    type: text-input
+    required: true
+    placeholder:
+      zh_Hans: 在此输入Xinference的服务器地址，如 https://example.com/xxx
+      en_US: Enter the url of your Xinference, for example https://example.com/xxx
+```
+
+- Each model has a unique model_uid, so we also need to define that here:
+
+```yaml
+  - variable: model_uid
+    label:
+      zh_Hans: 模型UID
+      en_US: Model uid
+    type: text-input
+    required: true
+    placeholder:
+      zh_Hans: 在此输入您的Model UID
+      en_US: Enter the model uid
+```
+
+Now, we have completed the basic definition of the vendor.
+
+### Writing the Model Code
+
+Next, let's take the `llm` type as an example and write `xinference.llm.llm.py`.
+
+In `llm.py`, create a Xinference LLM class, we name it `XinferenceAILargeLanguageModel` (this can be arbitrary), inheriting from the `__base.large_language_model.LargeLanguageModel` base class, and implement the following methods:
+
+- LLM Invocation
+
+Implement the core method for LLM invocation, supporting both stream and synchronous responses.
+
+```python
+def _invoke(self, model: str, credentials: dict,
+            prompt_messages: list[PromptMessage], model_parameters: dict,
+            tools: Optional[list[PromptMessageTool]] = None, stop: Optional[list[str]] = None,
+            stream: bool = True, user: Optional[str] = None) \
+        -> Union[LLMResult, Generator]:
+    """
+    Invoke large language model
+    
+    :param model: model name
+	:param credentials: model credentials
+	:param prompt_messages: prompt messages
+	:param model_parameters: model parameters
+	:param tools: tools for tool usage
+	:param stop: stop words
+	:param stream: is the response a stream
+	:param user: unique user id
+	:return: full response or stream response chunk generator result
+	"""
+```
+
+When implementing, ensure to use two functions to return data separately for synchronous and stream responses. This is important because Python treats functions containing the `yield` keyword as generator functions, mandating them to return `Generator` types. Here’s an example (note that the example uses simplified parameters; in real implementation, use the parameter list as defined above):
+
+```python
+def _invoke(self, stream: bool, **kwargs) \
+        -> Union[LLMResult, Generator]:
+    if stream:
+          return self._handle_stream_response(**kwargs)
+    return self._handle_sync_response(**kwargs)
+
+def _handle_stream_response(self, **kwargs) -> Generator:
+    for chunk in response:
+          yield chunk
+def _handle_sync_response(self, **kwargs) -> LLMResult:
+    return LLMResult(**response)
+```
+
+- Pre-compute Input Tokens
+
+If the model does not provide an interface for pre-computing tokens, you can return 0 directly.
+
+```python
+def get_num_tokens(self, model: str, credentials: dict, prompt_messages: list[PromptMessage],tools: Optional[list[PromptMessageTool]] = None) -> int:
+  """
+  Get number of tokens for given prompt messages
+
+  :param model: model name
+  :param credentials: model credentials
+  :param prompt_messages: prompt messages
+  :param tools: tools for tool usage
+  :return: token count
+  """
+```
+
+
+Sometimes, you might not want to return 0 directly. In such cases, you can use `self._get_num_tokens_by_gpt2(text: str)` to get pre-computed tokens. This method is provided by the `AIModel` base class, and it uses GPT2's Tokenizer for calculation. However, it should be noted that this is only a substitute and may not be fully accurate.
+
+- Model Credentials Validation
+
+Similar to vendor credentials validation, this method validates individual model credentials.
+
+```python
+def validate_credentials(self, model: str, credentials: dict) -> None:
+    """
+    Validate model credentials
+    
+    :param model: model name
+	:param credentials: model credentials
+	:return: None
+	"""
+```
+
+- Model Parameter Schema
+
+Unlike custom types, since the YAML file does not define which parameters a model supports, we need to dynamically generate the model parameter schema.
+
+For instance, Xinference supports `max_tokens`, `temperature`, and `top_p` parameters.
+
+However, some vendors may support different parameters for different models. For example, the `OpenLLM` vendor supports `top_k`, but not all models provided by this vendor support `top_k`. Let's say model A supports `top_k` but model B does not. In such cases, we need to dynamically generate the model parameter schema, as illustrated below:
+
+```python
+    def get_customizable_model_schema(self, model: str, credentials: dict) -> AIModelEntity | None:
+        """
+            used to define customizable model schema
+        """
+        rules = [
+            ParameterRule(
+                name='temperature', type=ParameterType.FLOAT,
+                use_template='temperature',
+                label=I18nObject(
+                    zh_Hans='温度', en_US='Temperature'
+                )
+            ),
+            ParameterRule(
+                name='top_p', type=ParameterType.FLOAT,
+                use_template='top_p',
+                label=I18nObject(
+                    zh_Hans='Top P', en_US='Top P'
+                )
+            ),
+            ParameterRule(
+                name='max_tokens', type=ParameterType.INT,
+                use_template='max_tokens',
+                min=1,
+                default=512,
+                label=I18nObject(
+                    zh_Hans='最大生成长度', en_US='Max Tokens'
+                )
+            )
+        ]
+
+        # if model is A, add top_k to rules
+        if model == 'A':
+            rules.append(
+                ParameterRule(
+                    name='top_k', type=ParameterType.INT,
+                    use_template='top_k',
+                    min=1,
+                    default=50,
+                    label=I18nObject(
+                        zh_Hans='Top K', en_US='Top K'
+                    )
+                )
+            )
+
+        """
+            some NOT IMPORTANT code here
+        """
+
+        entity = AIModelEntity(
+            model=model,
+            label=I18nObject(
+                en_US=model
+            ),
+            fetch_from=FetchFrom.CUSTOMIZABLE_MODEL,
+            model_type=model_type,
+            model_properties={ 
+                ModelPropertyKey.MODE:  ModelType.LLM,
+            },
+            parameter_rules=rules
+        )
+
+        return entity
+```
+
+- Exception Error Mapping
+
+When a model invocation error occurs, it should be mapped to the runtime's specified `InvokeError` type, enabling Dify to handle different errors appropriately.
+
+Runtime Errors:
+
+- `InvokeConnectionError` Connection error during invocation
+- `InvokeServerUnavailableError` Service provider unavailable
+- `InvokeRateLimitError` Rate limit reached
+- `InvokeAuthorizationError` Authorization failure
+- `InvokeBadRequestError` Invalid request parameters
+
+```python
+  @property
+  def _invoke_error_mapping(self) -> dict[type[InvokeError], list[type[Exception]]]:
+      """
+      Map model invoke error to unified error
+      The key is the error type thrown to the caller
+      The value is the error type thrown by the model,
+      which needs to be converted into a unified error type for the caller.
+  
+      :return: Invoke error mapping
+      """
+```
+
+For interface method details, see: [Interfaces](./interfaces.md). For specific implementations, refer to: [llm.py](https://github.com/langgenius/dify-runtime/blob/main/lib/model_providers/anthropic/llm/llm.py).
\ No newline at end of file
diff --git a/api/core/model_runtime/docs/en_US/images/index/image-1.png b/api/core/model_runtime/docs/en_US/images/index/image-1.png
new file mode 100644
index 0000000000000000000000000000000000000000..b158d44b29dcc2a8fa6d6d349ef8d7fb9f7d4cdd
GIT binary patch
literal 235102
zcmeFZXH-+&);3I0P!Z6ZD&2~7REl&I=>pPwmkyzKLPS6iq&KAn1SwKN?+}WBfD~x~
zlF&OO^p+6vM$dV^an28)*Z<$k7(3Y`ti8utbI*CLd0lfR?_a4aQeI=aMnptJsjT!u
zi-_n7;kf+$3K`*(ByK+wBBJZk4svp@l;z|&UU|6MIyl=95h=Y-(I+?1?xW8(*1CP^
z3J1si!v~S$L|V`7iSrpc6qHE9{;}dHn))e!8OBdvky~;niuHzsm7S*Z6axi!f0IE4
zkE@m}47d`whFzV-UwE#{*bQYFlM!LAX1_>`b|i`C<WKGiCwr_|nEzN=gZLK58!ES)
z;Uxy8sYj0zh-TW)c8<|xhv7aGn&n5s7kEb_5=&xsqI+bEUeAbSs253z=5>SK*W4ib
zoLT=-=V5~N)SFn^2Uqrj#D;Vhxy97=7p1O&!=6%o<0Fcn@RxLWK*W@Gdzg38?m1<Q
zS!7l>NdhC0IE@vwr6q{?o5KzGbM~v>Jb-!_YVl8?v6t-Q%X^Q>D?RwH-6em2?OtOG
z7v4_nqJCwipUldjt04P6WpvW;tx1%vdT<y^i5%jtdR){y1Lj5QJ0YFyR>|r2DO7bZ
zc?#zwRHe5@LZ=MdDWcv?U0J;04$#Su(ipxG?soef&!sG0?&SxvrPO;tme1T;`t9<B
zWYp;>%7)`TrnoOyG`>di#Ce;iCVki2fDZlRp^N#b;I8-l@i(1FzUUXqnm3$ew{_xa
z*mL)98D@6BXWC8ZUtB&je$K*48$rf>RQdV3gHmQclow)Q(Kgk?ds9>;{DjW-ei+wQ
zXzi|l)cJx*#lZ7N67I5!27=-q7W^cZuPQvyQ%`cAJ~I1q@Jx%5v5+NN`+nmiZV<xc
zBQy8f;B6nChf(@5-AUIL-s)^F-Ae5Ic*!Q)&s{`7QsEPI+ocyIR98q|a)k1+E{8gE
zIBq$thv@fb-)JY*ioY7xuGxM4gOc81<iaC1+EAI>|Ga(M;T{rYcKhMmbJ#OIf3Hio
zqGs=ON3!f*T9mv&?6FY$@{x)pBi~iYcOTZ-fBIMKS++G+^kMN*GVQ^3Q}OS09#w@z
zzQnxYeDsm=#r3=IZlK)c5Y#6fcNZQJi7;e_2U7l6YL;2Jnwzy7@0K;9+#vi+=MIi*
zKr@GCGor{b;li5F&8cW3mVB}86)X2rK*MdfZ%7Lk5j6GPX1Ub7eq%GbRgD=?MhCU8
z+=`-mK~D1RWTGh4@Ll0;VFbDEa_vQ>bic_M`PSgA?K3gywTMlMd9g)14QJ^(c8Tb$
zZ$s9K6=dz-Vy$BgczkqX$@O8@-J492^o8F!$u}b}Zo1A7_463cn`|La=kn9~4>Nf|
zilpjoucLI5Kk0sd-6ilgzK`x8>)j*MOj8r$c#wHA1R`1)wL|t9hRMn}ASY_;#bUA4
zpta68(h3-j^hHFQXC3e9#w*f!#O=11M)}KbXCXv$|9G9|-y)*1Aa-5|GI>M3OH9nc
z7#3<DPo&lE_Yp{MvZf_?^(ATbGp>i^s_iNj^v*;X9X=Hl+oY89SDs!SdJ{eWxPc1Y
zJ~4LL?2XwjRgl7su&|LEWL9rJ-qDLBU3^<{L*^di`xlGxwD&0cUVMwZ{^{M1XZ;B_
z-;G|=XoYUeQ{CoFI8c)Cpzwd=br+JL@$lW(Jn^DOjxkh<h7XxzWNAj+?H`EW<WpGt
z96d5&FWf{^rl|3m;pdh;Cx!&oMO5%^!_xNS-JZ^;212P(HWpgWWP0y(EhY?DCok<B
z=$6uKbx1G1$P9VnuGsLlYxhNFoW?i1d5XQ-Q=b+#*AB=c^{&}DbBKcAou@Ziu9|e(
zKYw8PPj2$7XHzOqm7Y?9zwmcT+({@>thpV~Y4<4nqgtkliIRyTLQV3SGpF>O;{>I9
zq=}?`X|!sK3D=ar<O}C#<uB#)<nxb`S20Z|Z$GEHU6|Ot{Ex?FfATZNvukHuXHVpl
ztmO9!UT9xR=1T5P=1YFie_w@R<%w>zqukM#2d}D&tBXsDy~oGKo5w{R9UWI4%^WKo
z@6;MPmP{NMWoZZI9BY~yBh9k%r1eY-+LVA9n5##|o38dk>B4+(QdM~J`94pKsQ*Nd
zN!tqwWePPgk;HsrN?|%;(ibut<?`YY&ydbGdkM+eDK!6lVy`7sCM=dNIe05QB|Sn&
zLI_*KXmq#cbM1Z&y(4Nw9CGwBBj2TrwUlopyQI|+n|yqitf#`ljWMW)rH8)9K_QJw
zvsWRaB&kfhyjrzTtx#Xz7^Wtw>{|u+g4RT;OT8R6iYrMgsW)&1&r-H($bE>Yh`18L
z%cv!~+#i2}kvk$i49D1DnvWn`pvM+Wmzb9sd4=r-8|)&PXPA4K)E}F;`ps0>6%D-t
z=bq=<$emEOx`Z)JGk4!o6P|G%cO18KwqkMQePm`G)ZftiV)$PBJ^1VJzDG#_?RUlC
z;_qJJ8}4IkFHIk0d>SwrbhbQBzx65nQ{NAfPqm+bBK40ngajVVxdactb?m76USVMQ
z)A^^9gQHl#N4ibfiVIH^-2nH1W6FeK$;qV1WZ&dLDRU`!dL8cO66NCPf}R<5#<_6L
z=)ujZ@6J?idu&fmo7O3!kf`cy9+ZGs4iy{Kod|C#RM(ZRmUr(Xbg8%8bbMb+(2Grq
zOS_GFMtO0&JGupkAw2vxcfWKEQ{h*)(V1H<-Dg8rP&XIuswfD>_@S(JI(>`18dp_3
zDc!R+X4WzNgB@z?b+f1fi7sF=R`SI51j3fhrWGe12hdO*78>Rrrb?yac~=|nJrq4p
zw=O)k6$w*xrQUhI1Dpk&J;f}aZtwbwjN0bAEvF{L1q}lI+cJVAa36ymf|+sRxWRM(
zi-`+rQeKiv#KxrARj8@y?X2yzkfD&#_Rx?WvDwUxOqyo~uk2qLzA~=xakJW;naAE<
zz5VHS?d{fFbgs6XM%c~HJDsWTyxn3Ng8-EEqKGF!0e-4s0WX5S%|u2d94Cx^FVI{s
z+SMFTM8;G_{i7)SBGevmxW69wNaPWXHC69F*49=$*5Ge3ks0*9BH1@(ZU-x#H;1){
zbzk-U$A!k`{h;z;ex<Uo@=j7{V%mM*ST^YZ%z8A}JFd^kJvv&BTAP#OZI3#v^<p=e
zwxsL4wqx8ymL9}O$jrJ0p-=N@)@l4`jtg&&CXV^!0~h2EzWM_<5Qkq6&7>osQTXRK
zI1FyD(_^lsT?N~o+VT$grEX+2N>)n@$dueEQT%qZjIOnxYG5`VC(S8tXRHnac%ade
z9%e?fw=3(ODk2hH6J^Apri&b0DPHP_*$p065TWNywM1Nu`XdHKq@ZPEo4P9V*VIwx
zQQXneFICcOKvk2k_m(HQj<xAu`Wvxmox^p4xIvBwLe&n{g;F1fvfCWU9^VBj9K4MF
z<Z0(KcrB|IDB|riv^}SqshXP7l9eP`=up0gF!W@{{ew|;DOM=1sj<!;ZlOHbJNta*
zLSe#CYPehW@|906qOs-k`M~pwP6}x3srL~C7yX#Nmhaf0CHSiQd%wPzFBhuy0%Jyc
zE!#i$H3zoz1@)85{F*&F`w!><bj*61`h9xd<*_Enh8~Z>L}Bwu+p<=(TLz8I{Y8$?
zCa~Ubz4a%_OCfNJmfWDO!kEu7qN3q|uaLU(@~M{ylbW->)O<cWkTp)Pdcw`=Y!U+h
z1{bgIb#X1;D=nV3tg&|IGvX;h&5HS~t(ZP*X4+!e5<yMEP20$~6HxwbqAoizz?<0g
zlj>s_YlZkL?|Ao$Qnq5S)*L*3Vr2T`=v45-^H-ip8e<(@>Jjy<k0iAGbxzh7t;elB
z<Br$`q-TAfdwgDdz1pWZNRz&k$pqXqQgv2^Z$yr*4TWWjd8Y!8PsQdu6^8=BGF~Cx
zxeK3ryD@qvlcCUNd~n5JUfD(B+LxXIv$TvE#X9!pMbOIfL{5oaxjV$Ph0VJar?424
z!D=lb?s^R%-JBA*HG%pJ2k-o7Qf)By(Ks;c2IgTWKvTHN=CQnk_0SyhGU_GvS^saK
z!$tRwB>_`U@JP0fq%hFv;K#~z0y-W&Fx)G-6L8yqud(n@qfDm6Y^F_eSHIo7tTivN
z*-Y8|2x@lPGe1~Y4{*)ET4B)8fxwe9;52zCHC(1jIvbaKzOk{UY_4Rk2dc%+9P%x~
zkryo&!B;i<wa>3vROH^eMbt7&M#OiShztSN?0t9wx+a$#sCStJf0gGX^4bAW4%G#;
z_qZLJK}B@z90C_8(qN~My-xY&lE9m$CaSuS)^sa#F3~#(a_5)l>Dg&i7cpEXnKH|`
zgInKqk?q54AmZHFAUXa-UMm8jaJDf}wpCLj;wGG5A-Y7&M0A;OMoc)QiJAYua|Pmu
zM5MnzCm|w=a3H$$?=k9x_g{bUgyYvW|9U6MA^D#XSG00S|L6Sj_g_P~SZwDBZ&%%v
z3_Xd6=<okJh?TYO?Gh2m5-GoUrsGGvjl9|-;)%oW?TD?lH8#2%Cj#eApnPEui<ddW
zO5eZIRC*yFsii!fSay#)F)^`bs#@Tt%$F4N=_=ZVHrF8_7~cY&+j?@4)o5G3lDQ?m
zX0jGMxFmh@BWSHz+OB--hwo~jEv6jT?JXC3mH5&PS)%{@OZ$=a8zPb`9Dnh0;}WA%
zN5WqW{553!*Wld$H~3$c{eRZ@PZ#_Dfi)f+z5Q0>FoyUiS^&LjZb2*xGx2N;33-Le
z&CN9{lJ<V_D>ND3wFhah-BRFA{NKaxN4C_m?I4raSfPfmK-+0z;OZwkMD?%~`O9dM
zB_EFNmV==!Ju`><qAoLUhmMX`(z}SFSjc()CaCjCa5!K&6jYTV>0f2c;cwCLcz_p0
zMox~T5IBe9aFtN3Khvjh@ep^>c&ltKZNxR6()m<lz;sjpKfi-paPP8Nunp#$>pi~T
zYx)!BC>UCd<}0IPTusLDEr+A3=wZ-BG!|Ap9P7askk{ZY9RFM+$G|2fB~UKb=+0l@
z#z-hhfJ*l}X25(l=)(U58^PxD#P-K}hn^RtABP-hI?61(FX2$FCTep-e(UwO|8;Wz
zMoNnpB+<N8k#MVnL8QmPeorK$$LPqb7IDmgDqn5Bu`AMB0cxT0S9hXf_$cY0XzV#C
zoSCh)H_}Btyug%umSmPC?Gv~DYnh0Et<u5RKUwR)pgZIdtwF%X7leuD;0n4KfAOfa
z`}ASJ4>ME;N7SA>(JP9-LHc{UJ)-=ZA_$96F*fT_lO)bK@wW<9yZ53r?Fwc8<_1jH
z>;=2Uogh+*&{gYkuu|oGjRPR`2<FsaJr(&^JBan&oQS);_1jg}+NVz+dXJN9l~xZK
zE#*w>5dYuzDMGw_y`*PoLnms}pZs-@Mm$J|aDCHeKOi@pK64~Ti0Uuy5n}eQ$Ksn+
z8`yfQ>npy9GXJApwm)LCq)fb2CGpB%+=KX%g1$w9JiWWCiPz7cd|oWcBHW~0@7}#b
zsPL2vQl+cX{nx|$(;ahkLsIn3rZ=+jnL^&)9N~gTPv(NOkx@MSGMtU0dVljo0?aIC
zYAz%0MDuR%;D)(Kxtm1d?~s|;r051*%;ymQP5A$KPck4-A2hS=$1BV9w1fVEcaxNj
zJq2HbMPc-9nZKFD4aUpFQVDmmolRuT&COXPEmI{hFRpb(dDFin(>y%#t@-#gF!ko&
z_<2q=Nzwx~k7)iTd(*2n)|WQ#99hUJDm=YXc+I}}e%kso`mZ1U%FiBDoLafobR$}R
z`h}Q$V>7z-vqb7EnZ`F1Sj5GsrT_7NZbg3t-ouZw)q!}CV2e8~XzMABe8DhVtPh&Y
zQ#88M=kM&&OLFkziH2&elQ-`()lV^+`IeOV_Js?iq;~_>yBRBA@&463KFUg*T)=N>
zTb#ue`ol@N0$Mwku7aEXIPd@X#|XtQaCgzIPV1$w5stTdO3A98UU&HZ@2t3_o}=GO
zQ6m}uS)chSI+m6@>@SSp&T?lY+M3dcP3cj@wp{xAzqo{m#Ir(Hm9Q}Pu<tA_8V=%r
zW0oa_99R#F5UbKpCrTz!MH0Gaf061E5etl{T2nJo<*$Ev=eB+OFE*Js?Wc6>DUYtC
z5a#!Vc<JOuj_u!s6~_&V!%l;+(Gc3!S9^@<Hh<+X>y#mT(hqV09P*qzh`yEpYhP{E
zzq#;1fh=b)k=J90y3-n&mRHs1tRE#~F%SMOdhN|~lB|ivKj~xniq#XS@);p?rRCzr
ze+OY{mwT2B!o3I+w(PZ*lbDv?fBqMb!9tzx<GMA;xop(?+3#XU^IwpEW6}@4A$uaW
zxu6zZ`<5Nq&4|D9tSmZ44Qqy`eaE;Vj|AqwajYvGg5o6C^BGy~RQ}4PLdZy#R*Gp^
z*_9rdF-^8-tNqPdI0Q*T05O#7dTl5nZ!L+0_tF1eO&3GM(oi6Bjx7*jSJB_k(<JIr
z6YOAHe}1g)JBOpjxMa9gCBVIsD98M$W%d<Mw&oDqFa|UGXD1;iJJ|qcR5*-~xq<QL
z^)*wvdS)GJ*^*+W5F@|M4Ligv`_n1lX*&=(B+&*R6o#2N=+qjcU_yd|^HkXNkPQ(&
z$)oWn=zaE!dz2D~@ubLnWzr3d61`c_-u;uUEmt2<(BW|3usOoWseDx$c(`!4ud1v@
z|KijRjSd8@v8nY3A8LTyeGwjI#t)}^U;ZzK5NskbWmS5_rV#akpYzMV;_wxp+ZmmH
z<BsXYUG@bz@)oEEN4GD3E5a04y%p3l%;M<>;OM>^C%M_S`(+GQH=$?RxFH(_lHSW7
zBZ^&6hqpR0i_VP~f9V6Olf7uM$vDY}qni&z0gX0k>aY@e{xdWk`h;3)Q99tX5OI|=
zYqK1);EP}?fh^|D;oIdLu(YpOlyXDJCWBpMPUW!P&K%v_)741L-pqBg;1xO>R5^Bn
z5)gNS+SW4&HiXWi@N8#YdFap4P%ynGN&w;;xKpXxG(_@eC$r|uc|-&R0!<-3_9@yE
z@#U2H7B}SS#1{T#spZRuVeepOfl`z~n;W1n0|Mt*`IA#T?ujYrxN&W4<<k?p)jEf+
zw`25fV~Ok;Yyqd#=y|!gx{12AIq0(FNo>6<iri=YJ!N_w!fMQtn*XCT@SViTc{ETO
z<J$A#6JX;@ni8zn<CzBN6tSn+d^b+)c3F?un(4AScJ%#5fCPnx^jS^d_8zDv^K~~}
z+HiNP&@H#BP7f62qCs$@4I)vMKYJj8?$3}U&Lub*$|63U`n9F$alDoKtrxk>2A;!G
zzA1eFUZW5lBTMGW<g`Uezu%gvTjiL>bbiJmGXQgpD9~UDt{Bti6_;YYcyB$%DQt2+
zSp7~aevwlKC!Zd->DPV{*m$;<0U4VZT=B)So-|ZL_XbT2Ci>?~4I9-M%sm<d7vC3p
zSjOo2z*nTutr9LfdFPAzux4v%PmjEH%6eo*%j&XJa|_fmbFoJNyl?2+9vsLJYN+@}
z%*x3l$$RwCN~pi+`NpYil$gzpa6l=`Q85kj^4zIoKv(|tvpKx~PvG4%r%AoSdY9I4
zxbs*{h5~-E*iKyPZ40wD2!t<cqiJJy;m{!1&0>h2Q6&Sq{7>{5O#9)yD+q)@0*7%M
zBCU0u4_bY`6%2|7CK=m)`f|Rx7|oadqx$?gOBr0_N?$~;<#g51&9-24ebt-S8*gyO
zDi?7FacOk5DMwZ;Uj_;a!Tvos3?`%cjY!&za`T?T*8N6fp!2M4$la*&$Gx8sH+9p$
z{UIXT8LxTah3TxP8p`RVQb}2xW-9I-#8Tyd)F#u+4D9GeS91pU?PO51dgPBu49XB9
zA7EjT0{}^#Eo-9>tA`*IkZT5RGc~DXT0uu2JWHz9l^SNcerF3Kz>tdz>H3+kJk4uh
z&~3i$Mtj2$SA*9JsdZeV<HZO0?zI)OO-_vvuzM6Wc0(ApLZ)TVAkst1CGGQ_99ai<
zUK#V+hr)MSdZjRh(0oHj4suAYl9<6C^GQaUkkhy^MxCE%{*OhWT;d}C@<UQPH#b4w
z^Gx^S%NMNcYuPwDQGbtKB~P1*x4)owB#<P)rPrUqY}@w7#|<9*fce}<k;QB*n_FSl
z1tm0ixtDrK3c3D<8uS~F&P7@a<RDuW8&NzEp9mJ859Sj@AK(DYCl{5~LG|4zc=T3k
zN@qzOSFDRbPo>)cr;H4~Q`Li*U3wSSw=+JCu7#PYcG(U6NToa7EO|2OVpKFIE#?1Q
z+b2<Cr{yf<-fj1`Y%Kaf`WK=#21(Rb9U0dtRb5+Qt(cPr;DJZqVlWOn)GV2SDn0D%
zO-R0$k+(-TlWw$9$&-q@V7BdIy&~8=-juDb0)a6n&-fqzNeXCZ5r{W6r=m3)HrGbq
z`<bhly<QDRvl=)0LJ!JZY=9^MS1;)4#Gv%xv^nKjP0BNdyS|=+@}dWT7oDbf1#vn!
zmDI9qedqo_BaTW^C?+?9W#;`F5Lfdql&daqO$0Dgb2jE>)NmdJUM48(XB%<+-psBw
zZ5qemW!*bYrS<;l-AQq{*`uQ)>``v+G@x$<rdE$z=zt@h*Jg+Wj0ZOFSZX}6`ppk>
zbmqfBczBlIzP?I54jmGa87zdVz)5(La;;_uqJ;)OyBgdxoqT?Nm?vnT6Ugb+_*SvO
z1J3H(>;!3+9<&)46!qRJZ7pctScN0X15>H5y;`=>r-IInhJ}ag!e)6H_J^TT>jf|S
z&OMl&c)$`D_M-TpgA|udzmH~Vx*|Z{TUTyEAHP_;Nsfm4n(DOnl$_Fj#hoC#bXSbi
zi2uyDC@(EKZHp9juI|;58D6(eqKvfn7L=Eav|{W|c+E({dP_+YClrNdHR-Z<QcOA@
zZm#zhz1Zvp)VET33`?D~3>$-_&U@tN=G#usmPV@%Vc++>Mz8iZ(50G-P<hlN0ty57
ziwiw1QKjb#aUc~y(WR|4vjDNQ!&xlb%ld9mWr-5C<Y{~c=)&#pM)_-w%QU|>srIR>
z2=5ahx{BRog+~y!+y?+l`5q0NAn}2PV81GtYuM9-1>Yh14A(^;;0~3$P@#TU5RQ3H
zrE3x9_^`q;zW0+kv>oY#MM@({D1NNPKHa|Sz}96r!T}|fKHM|WqP#aE1XQ?qPt&Mm
zU9*qbf;1E27{&)hJpQ#QvuW}P-B?`FH!B;k&Z?c<J(D>;l{v3-1+6_kX*2aLjrJWV
zt6wvV@E9RKQ6`f+X-vDwzyl#+6yw2#Fq`LpO#fslNPJx3ggp3SgEruwaLUgW{?#3{
z-Op(It(`f=K56PD0i?w~XZYLkeHV!F<LJ(O8rgZ}D6K}{{EYQv2tSQ%J>Ta6w4D_+
z0gq<~MX-%Xj>`5$b9&=brPW@RKo6UJ1&vQro^uIE#uDCS%HNo3W|8e*(14Svu=~1K
z?G(I0F8WXwvf_7O%Zzn#f;2Wk-SNnQ@|R0>%{hhIwB%a$L|w4qK{74cgVGtFHPaI@
z+|*%ZXbG6n(w%8API5J@21)PQO_y5T9Xo`};8TcS@@FvFZaYk=PU@hBG*Gi25}$+*
z1Dk|IU0|}wgCf)4cY5#Bo0kgBKEVe}mQ_DHNHU`IFm$htQ8|bt6=+#+R+q6ElexgW
z8GF?GM;a&iPl#X68?x(DM9~S$w?=1jt}?%{^rR3-eI*k4nH%uPrGB#d8D&+U#&lhf
zpFlamc=7@wO-bI9F@mi|)?gf!dzxuLwy{&pgvYeGCkis=O<6NS{V!k3yqyD63|fii
zJnc~jOLuJ)lZv?1uGpfJlj3T1eKCe>g#$DW&orh-XfnPQsLj>nV$U?%(g2e@{EHzZ
zdm{2pEA_rj0$?8>CVc-uTM8B1h(+RQ?L&urGKv69m|2<?q~C94Vk7uQc;flKzNijW
ze#6NuIifqw1P!7+A)jj=aXFaetNs4R#g812Sj&90lBD{~IDizN2^CThBJ3Q&O!7hT
zrOiX>=#vv9E&18X9KtQSd2<t7xDdEqr`=>yyBi0R)~w}X2*wV|_)c6Hj)zi9&EPI_
z6@%IzFuTsT9j7E!#n7n&@8UdXqk^s~5OR&x*CFP()dG`)+Tb6W5ls$b^xwmsY<8r4
z+<Vyrb@(LNetu#<3K;UA^RUZ07;HME&uPJxa5xg{n*v=huOsiLnIK&7<@Ltni1~1r
z`sseCPp8Q73H6kf@c|TvbN5}pxuW4!@qymR{#m{9<By#Mb28`T#)k)f86?@ktJ8>M
zwIt5xFF!r@<u6U4JHyXM$n1qm4oP8k>f!u5dwn7^yZA4CM><@%bvpL$jr#hz6bd8H
z7+ue;(*MvF{+l}=HpxnU{9Nr)S*S8a!{rQESOVUV7b&{k7IE`Kl}nbeHURmf*#Lf%
zGXJ41{p5kDfB({<ZIRE1;Qj}c(vjpGCTC0F#<-yEq{bq4KRjEbx3Qt^#(iXxh7?L4
zdWPC*Y~I0T1$Nz`z#g@pASO@^UCrlJ{rsyA{yKx+Gy?_A=*Br9>c@d|?S_fyyt>rN
zX~ACP?<tO$%wyB=4;hq2*M+~&FS=%DNmF**LHG{?(G4_E`;wFcW><{Miipd|R{k$|
z4Pue<f2I*p;Q&F-ZAl;XbvX-*#xI6qQ7b|pt?rF&g@&TL*lRmD?)DW+I6xZqHxHF>
z1z5eK%r^kF)_PucY=1E=8<pYOdN-ziRAr#~tA1d9f+L0REQX?fpU^PV;UT_QG~6@*
zY3R>7oz`eN73K~Ntw!|5i4VELmw*z@0-%Lzz~;FBSPe9A{&qd8`2a$zDq8j5=Khc8
zFhWes{4zK2TUIMIAoPLrjqpL;4L|>8U*{EG06EimC9i_#E4f~7q=G^ho06t*PEJf!
zApa)4E3#vQb=77J>x29WTq+gd1aV;nKuha&@SP!fY|wt}k+7cmeFj?|3DhZ|@EMiJ
zG|j!5wuw~NIt!V|B}IMHq;hJzT;!p-3ZQ*@X(&RQ&EdCXkQw-kY95Ks8Y#M?ykX6r
zL(Sh=1HOkex`1$7*6K1@nE>d)>|qLBzwcJP^Q_)6L0At9yf|YBO4l<3=gy@y{gh#Q
zKzZkVVaW1TW><KCY9TDrzKP#Iyf<2)Z5d2e)lPCC(ONeUJm*x=FWr~Z<EMpaaB+n@
zM*DdV3<Mwto{@t((;zQ6v(M>7Jy)I3-{aV~51#I?FX1?@mD%TXOP%0`Oua5X0m_8+
z58!9if6`|ygf1)i3Wghrn2~D=X@jqX@?tWCt6EW6tl?v3)5tg}>(w6x;k<(Qb2JnW
z6k~9O0#Dr7#nwq1EBtICP;&{0k$$q5-{6ZjQhQc?_vzz84SogxjdfyTsV_y27{r=}
zXFW23J*cqZKrL(Lu#kJ$sv$eTt;cX}?X4)n0*MHyPJsiOdT@KWa9o^J+2WV9edx}j
z<fMU%t~n@tgG;p#?BmhmuajmD^bRPOY`yk6dQh}4yc=RR{dmR)M_J!AU2V6vxY?(<
zr}KknR<E{{QvuZiCm(%A8^u|Onr&1|e+`kE&0ykw73x$MfQ?1>%Ji8rrz+U~QI5}1
zB|)~5mlvwpx4&@T41Y$uut0%4;%fRZNy`52=tb0i|B-In@#m?A_bt*pEe8z){9<!5
zj>eF!5pC|nnka>_#pu?brv7L0=*r%UVd$jwWa;+1LE}1vklNP}S}1g<z^iq-(-qLr
zjj9}u7`3NHf;<P}*feM=e+Pizoe(>4;4JXy&~s~lMD~4}xqo-WPU}twf$h11dw$Yz
zt3O2-y*AA=7ES?&1)+X;BDu_^pS*C`#s36L_aOt|Sq|4*A3N?Z_iY65P+*X*qPs`U
z07os5)cxDptLZ!go)$TC^&5#hW((!#9rEaKh6@@+)`=DU!ZmWOb!F87{_13qO@vwH
zXRhGZ-X7+%ox`tuYBP#Vhc}?Hd8gyWgsm$f4&D!{wjK0X10DS1yS-Pq{8^xJc@e$I
zBjC=%3w35{>&p=F`X!UVPu|3>WWU+Vm5V)uTnaoWU@mkw{3ElGwcx$vg_k|4tPW(#
zr8hh^7<XK@p>pEr{vM4QWC`9yQV>dl(rJ1ZKOpAoLI$}{`Rb-VqXXy8-eK2hGzb}&
z=?rd%osCbS*u>naYyu}jSxz8g;oM6OUC{!q%X&sQHOi%L%&_U4;CtFFrUM|{5)v>&
zLP7g)hRj&d&d}x;AZ~GI=(z^->DC<E0pxsSFo5!D#YPqT`4rN9O{&gF48N4O0?CM?
zL3kawphtIX{GEt;*J5SFV@LP*6BNO(X6Oejeu}nq(Bp#QkUky~oDhgn;NS$34?0NZ
zVUXX^VTfKeb{&@5TS$;O?-RkEK_F1E`f`1jZ({_%nC^x!eUfxYga?$IJE<;mYONPj
zi?s6iW4eB2Ncu+sar$Qd!nZgc6S{(M=VRh*e!B|@*(he$q0JnaGjEmB(i}`Ldvym9
zd?GH_s<PZa0M!+ra({oH-V<hWWa|6FPTgm&b-Q1Pr<E*BKPaDt_dl@qzd$(eUXIKg
zUi?z;g5La$`Rsb|Udef_+lq~4fYs%(b&49P-|0j61(Wba^!NbeRC$H1d5w^g`~-^?
zdjwWf76Lb36zabf{H=Z+mbU{>WIoQGPtuUr`5in!6Nsfcb3QR8Y|gI)1neqr-k#@g
zTlRpW8rRI-FdN@$boV!l?#cvLU^j0q6f>ro1;rW%6yJvr8RaR@wGA30G=eeFqcsGg
zQlBB@z)_D<V|I-<7ATiI+`S7pX{F{STbn!y!bf;+b*B)R8!tnC$elRcS)0jJv;a=j
z6kKN^V&GQ7C#p)W^{`W_k4Hl2iuKt_>K^u6Hsfbp!&ad0u7AsJClra7PNa+v+R`)5
z_ph(A@b7Wg5MN+C8P^^@1wWNE3&zhjH_maIK|$*YLQl&TnZEx~ACqOcB;?nbeFxdN
z9Ub~`n~q&1vU7C_C@6n?_`XGxepc+s%6M7Ol2M&h{2g_Kb7Rnme6@}<d-)rH)ds`}
z{)r7YW<J`RR-Y8x@H;1uxqO&jpuu`_CNyfvF7;iC7I_KwZbIL<n>rmX<R9Jz!egtT
zA>ao^d@aYAEp#6zSUpsj!@4{X=22QRy6=$b_l%U|m&|!_Rr!hC4=-atcI#vASl{U{
zYt8ucHF;mT{cjNzZQjM$6^N^sDqmxm#A1arN2uA=&jCGSK?iEyOoDqIxZ|kM-idvg
zi@g^{V0rqXY^BPL2h0TvMyDv0?^y`#WB96zF$M;FcwQcehW*eGoePNI!<?<U0&sn=
zRCZAm!(Lc+FF*b`+fZeX1Sg&h*W(Ai<D69P5<7L~)3%*|)aiQ~Wv4l!{My5Q5ug^A
zOi`ETI*_suT3yn+UpqO&Mg*B3*<^Edy%altU7d6?%-dvtQMozj5s3K&68?9EE5!0j
zOJYm;k}tL7vi2s|)#jZ#YozQo7s;dKB~uUx8dr}eKR!IzHD|0nbK5@*Et_l8X1Zu@
z2Dp)u-VSD(>g=2i+#VI2{H5a5JH0_tgATQ!EogD4RWhw@0&mX>78f>bmPi5be|l_#
zZrZ-shd~Q117^b14u(a+V{gLX=mt2s!;L<7@I{#U<a~Sy$S2EuQk$bd1H7=R99)lI
z|LVls2;Snli+br)+V-`1F_59D&r{6@ggG$B@kq=lgD?)9&5dj613M8i(wH@$#D5t6
zxZh`95hC!Nf8|hoL775#^<#FEBTeGXuP$T(Syg~sS=0_r9P<8qD{+rLjgjkbw{67&
zGRo=?MNML5HU_wmto0^a^dCVNUP$^O{OWylsV9GpZav{~e;Fv_H}Tp#$qN4X`+4=4
zS%h3(^_6I~lfZL-AU&QN*eC!JgwP{jWEO}{o<)G|QV#L0UN(HVpC_`4jHt!85aqc5
z!eW=uU=7I>iV`D#f=Zj8adBMVOTf+6C#J476*~L*A5!#8+BB1!T(-)ZQv=a`GLzw3
z-kb-+GT^{5g)NwicDpyUsRMVIZr%h5+Pfs`z1i}`eNmQ@s?s5RC9PIoW^UERRhd9f
z+rzT^E9c*iA>J{UrbVB8XJ_+PDXDUq@jS`G&0jNi^}Zl90dC|ua>z^SO6e{`KtSG@
zf<l`X0kK|-KURR&)k17TAC~Hh*msg?P9Zw^W>3@8;#e8Yy`9TMSQ(Xm8a?wgDh&xC
zuAj1s7RPz-6`Nz50?$_TB|~QU@O>IVLcM!Go(!*&X*!<_xg_?sLNcSr`uh`AdC{(t
zwg12t*KeXUq*J6G&vt&g`wOZ+m{<@L4N6k3V|er9>XPz{#^&RuEuQ*bW;69{kG(-x
zkO0+J<H@o2QKVdlwBf(lr76Oen46GLz#)c0P+UbE+j^|baZ&1`NL2na_T#S<QhbY3
z8te&FMSgqTX$S(z6F&-ANzq5jGh|NhUS-vg!o4hSut_uIuD?j-+lPoZ;t188=DFG!
zW4p46tmC;JX2_0+GFhR0{C8g{zRP6B)CU7bV3hQPA{WbzyJ!TX&PtnqPBy}tY3r+F
z|D<t@Fq4!~eXf|C`Zq(b_xm9sFtRTjwa@wTu(4{ls$xjLRn%kLx}BP!&Yg5?4<FJ^
z#_;0qZ+HywqhL1P+gA&hnLO=1o;W@o`(%!N8@Q#HQJ-?yli}*wjPjQ?c8L64A|6R^
zPc?R4RX^Y6L?-r)F?~#@Sod8dY~t|><2<5=0knaSyW8`LPwWqABf$7YSl<rp$tcV{
z|7u?NoYbyc%_^d5AZ!jw-!L@eBrdff7pfqH44m5`6#g35vf){K6T`+rGWd=Z{lYDk
z>a$f;>*QUvAAj5++F7_PNN<myqGl^ooNQ#nx+`naji>zN#=V2345u4+F&-u6^nlw-
zpjm<?EQRG2Jm>QjL$_M@MNsuSY#XxI!drb$a7ak`S%~rLgm7eEVbCDHwyJx=ymf~1
zE`^hKIq}VdLDAZYb?Iy=uDr4rFMj9i_R0jYdk(lg-xE6FGwn}ae}Oly=ZE$t={pCV
zRPk(e^3CtYu_gzt!Jz#2E^yZFY36`i`bRTYqOA|?vV%ofb#f~7#`s!xpQm<Ltm)q)
zr319`1*UxeT}5k{Gjj-LY9ZkKkcaE^s+7ajEj#mhJ!5Yh0v9(lXnJHg(jMvei+wNY
z;)S9$fcvl5k5`wkGCGx48UdQi&gW#Z-F#<LTHo3bDpSi3G&T_q$!2rz7vXc2iyjA@
z;N?C|u^W1S*d;>P80@`ogbR<6sZUv4qhL?+$31=js92$%g7qgWQ49;0x{YP`QyPDq
zao=57oY?W}ll3m&26;!(6d~v#e68VRhuv%=H^Vxx1>Xh8SZ3X@%ks4B&wX~o4!qax
z1$;_z{R3kx(=TEeQb{xA9^pGd7Ysw5OdxgUHl*MyWV5CxCCo`Mqs#^Jo8vaGu8eSm
zehncDFG3%fpQn$_Us7XAcJC4ZnS2;cIV#K!=wUthl;r6TT2bYkq|_Ucjr1HkPsFY-
zfK`^*g4TZmQd0E2@UE@3V}iyFM|I84KOHQ3Y8)u?x9WBis8M#&;f+I{okb&Hvm_9z
zhfoA<zTvxE(wi+<<@qK7_J-vTMk2n{#N~FZufgJT%=FTdl>1#)-@|*?HNfk^&yd0W
z@`T2PesV>jTaN#rCA`e-Gm|;;PFKgWe34sqa{zda54_cr*>}4z*3JBMbVK@O95+lu
z5{T`rMnv)OGLYBSo;MD&p+pdFt#zXa-?>n9SdxYd4`ssdhMA_jf=-jpCxg-aYG!B2
z(A?j0kw)-h9-!45<^jf@JGAX8;tv_>&nEFpZ9bdI>Kck4oyr@Vw2PnWGs|wU574<X
zPVcJ{=;<ws(_BQ7;4yLyn4)1n1FX+NWckkR@Z)Cb96HE@mm^<~^y&gG&I%{lm{$=?
z_>*D8j!~<ljhkC)GnP%t@bmD|$=2&H7h&?yF}vYx)B##`@$lrRS(SidHN1S+htg95
zpNDs@f<e;t7)KtbM}5sQWf9axxI&;LkFtcI;0<zWCx1c}(&dYG^LD2Ck&~TPlc8`@
zNMfBspr47cM5C>H%IuGl?W!phAD1gulNIg~$n&KbU2PXZ+#_B-!;=?m;lX1g6003-
zBC0+%GnufxHmPFIGGNS!)fvn0AJ*h4I8x|NEQxiJ$ew`@cOyz*N36NUxFzAU%m;Ii
zlslILOCX0&xMU{GwK>}&f)rS>5Yer<p}m}jW7EPyJwh7n_{{+e`B|B}q}bfhy7O62
z`bmtVEj%4m&ilagKdYXM#>86#26)R-ELNVf1iU#>3fLqU-sI$R-QrPGF=Isxy~b5k
zN!J|hMcIg|jHyqTebY76RrqTDq*<?@gNh`Dr0M#|ZsK!bn3)8>F_V(!)o<@~2r?YT
zp{X<+4uDFC#y_`YmpPPE2icrwFav%%0x;)m(x};{iPgpmAN}BydAYurtd6=wBv9ib
zVa?_KgWo+(vWl0o{T8_jVMamN(E-yvVX8oMTQ&CE0y);>SDgk_eY~vQmY#}zpBLv_
z7<6{s_q(F>`YaSy{;P-XWj$_ddJS=ZZjg*QpP7mGe0#?Hi?MFnm^xm;Kp(;_#iOGB
zCQ<$6bUvqDY0Ky}n+%mMGZ4=Eq$8Bd6Oa9#w{a|b57H)n{4uUjG{ZSmhV$Ua4V}Lf
zw+Ge6AFe?$gK_sY0>$}nz40JqWnV<eL0%KXyv1PGsyz%LuaVxEtsZD=++8T$Hc04A
zij!J){cOh>*8QY!xu8wJ?R!~Di^6g|_0EnhZ06q2uR}52Ka)g-7++-FlS;Xlty56v
z04wq=^j5RfKRWY9ZbPQLgPTX!3lGW6E4gYr>0}TsYW;k-UTa2Adg+w_#{5tUuLA7`
zzZP?b*NDB#cW|F!5Q%)A(6zsvO0IpK>fJL5H42qG>pk`+dWKAJe78$5I~{36<t+Mg
zgeC1yNoAD5c%}aFCkI<5fZ!I5zSOik->{!sTzEus+0NQ#VwCaWt?QP$;0dd1#<0qT
zu}vtk6ZmoU%8jw!K5|3)iv#wI%~N!bj>wfS%t!T2@OfHe_8}Ag?f~r4Gli=MBO}{0
zzINVd45bAYhJRc%MXf7VB>4r_k~C*CDE&nDi0fUws9)a4bnpN2OFrTwuEi}1BR%%J
zXlkE4{>|anFb27+U!$qZTuRE`#q{BUz$SHFsnZUOa(B?!bw>(EV|(8~p#KEtd9r!v
zNSzg4szkeze!2-C`o+q+da&_JgUr_!7n$(m<Q)l@v09Favi@Q7F~?WrxCDhtb{72~
zWQE@>QT(D`&71vivWEb!c-e8tXE1~{{jB);{h)SvRb%2Xl}~iQ1>?F8t6MX3bKl^&
zCO_UDT@OX^xgdl3>EcVzmkXiarH)%0vVN~7WRG(<A&zaOyk#?eGzz%-u?w-lj#WbU
z{8(U@_&z8Fd+m#OZ>};Xd2&vH&KJmB1xb6gW072(>|6CxyYpRWRdX?(7#RBPzKdw9
z8Tsg@?6s^B_PxF(q*^^9yCmH#LfB=d#w3ED;kMx?yKFA>4@mH=%SpD>nA#k^BMw0*
zK}EwYvA*{#foN>nnz?G#gz@Y)p}-g}bYai?Eo-!<B|r2_h90v^LmOVXD=*xN^Ev6g
z^2Xyd0HG_vgCOQ@TDP6|^Eh0il;)C7x##tL8<J=&Ay$ofN99%G;3#UODjw@_xw#={
zNl7N=fxl6*9^iV3oMYCg!dVZiLQ1ak%6C_LB@H$QyugqgKN35)>EL~I|GszJNu6Oz
z56Wyp#dC7dL4rT}X8*B`pg@naZ>z}Dhsw6}si89bVJX{dw!K-S$rC^KtNin#|Di9O
zHt~MMi4vXAw#2Y0#jYZDZVn7=LK<g2o$tW!%Sl|gWPI89E?EqgzO|R{(WD9DkwY<W
z(1O?&TO1?gc8@h!Hiy=vXBjMC>F#?aG~B7RaVlAaCxu1*Buji|c%}h-M7H&xodtpk
zm&Ro(`Y?e;`g6~<8)9z)Mk3l24PuutkuK+jcOry~q>2+Um6P=@daozbegveu2S#s{
z0lRng;9rB=R$RY!2uPZ0oiX{g;lkVe|LI$#+tM<rzbNY0GjxZxGQuV&*uGTbM&@MZ
zVJ^+9Z5NY~exZC3U0svhYd$Po&vyr7`Bd`*7*2{&646~PikH1I@AW1e153bpW{w+z
z<XT}Ao`Xz|<_`uJj~0(N#_UczsJojWHH-V`#OSu{w(8)X54SE9hLrovsiEf{K3#bq
z(G1X**^RaNWhqpENB5mR$^@UnvLxmgs=Sp3#(jtCm_^Fcy2l*9F}OYQ68*<IS<?Wi
ziG|26>Fr4)&)V0<xLaF_iF0&tXrS>)xIng<=oRb!<i@RiW(YB%pI>V~;6tG2gv80V
zHj-=$l0gwILFZ#tPAD9SoP*H6Sjdq0T86{<8W`_e#qxtng)D_7@SH0+BYQiS_(~o@
z5Gk9w!Z7~K=!yT^@B%-gaGPda^W1_2^sbm8B`Ys;7KCPQl@mHCrX|(cRxD#C$6C!i
zrnmOhznXctX@jL++uOcyX+j0D3JSkDW8GUI{?>=oJP$`#&d#^R#oAVA@Y6!IaPWBi
zieY+LYFX<jN);>=KP>3Ad|+wg)IR_~z-8l41+H(um9^JCji#pH8GyhoZ4LEe&(`YX
z8HlW!PoF*-Hfgf&J`R1mD9Ibua+u7RBEWo5gW7AUT~bRa$83LV)aBcH5-%mzxtS*t
zQF-%g$;Gw%%|-dU4O#9(Dh$jU28`(Z)Nr0BG16qnGU(SFoPo&n+DrKOs)9?58m~b3
z#uHxEVN*}tkCR&k4|%6v5Z8XUS+Um+;^m5^OfBKbqn{rTD0}fwZl~NCec9r$`t;hf
zZe3Q@+OL9g|Iz+K=(K;B4cQ8a^QNo1P@L?ob!pT1<h8!IMUN7a!<AU!J4Xv~8TVqL
z04>6o0DG-Oz!3qf1Kf3DL-Hsdc)3rVX)rL2pTFNXqI%X|x=QN)b9x__jusu6S|<bV
zD94<~$@Tpog(c2H!JWdw(MG(AHfgsE&leHpat}>bxLSD!pKRydNR7)NAjfJS=9HlV
zhbSsA%sk(ldpPJ;XR~sh(|busjo`ziE$$CI7bl$(6~QuR<715lJA{%L;#MJWc2HtR
z5`8RzRt5lYKX9j<G6Jl!xxM~71P=78S!Z}9uTXDYQ8egy&UY9wc#n{ZK0ksL?;S3X
zo_M<=0|uTJZR`_(GL=~=S8^#R+B`??$-Up4OxBnLQx08U+-d$*fn+`G03y)MsO8y)
zb>>7#Ca?)_^`REJ#Z%pr{6XmDSXD9#L?AvCjog{(VZ|0}<gW<SF8>3-Q@_@5RuZkL
zjUShj`cS1T^|^CUCS`iF=9k(K18wUE?ti4gy}3bo58;w_@}lp66V*MZjA`}zW=Y;$
z0%;slOBd=hJ&l2gN!{#vfHN}*suXP^l|FZ`U&HzwXOT5Z9T#E~W0e}7l^(FHSZgD6
z=2~XRuD)hv_a3Sm(hVx7^(CW~wn1Y;_;T-lLXOnLDo;>y>po+R)73ZZqy9L~@!c<=
zx6akOm*wmc`)k%@zd(tBuc57Ea{Q}0DG}3jaGehVu&KHRDKl4Lv&^2<C7`Wk*$__C
z%)O}B-yYf{DyUl!t0wWhhDE$3cQDlBF!Xj6kZn!BZkX1#H)TI=>07VmBX(7q2!=Eu
zxs<=rJFp{)k76vC@#RI6bS#6O!U?F>)LQfF0|~R`<wCCxy{>mt1vkE=3jXkV+3Vi}
zc8D$#lyZNu_Zrrpb63?^gnA`~J;=S4&!q?|{M3Y~O1yV22_uPxk=!@j@|1GOX}JzI
z-v@gn`+k>6p4^sb-_#LKE8NVHEZMIbj^1DQ!j+lXU^U^(@P6Nl6>VvlMO4T=u-i+6
z%+tWm&c((Q<_HJ>(fn=fJ!I<mSZsJjT;j0&+w`ykP3Ml-a6g%uI$O^#@1d?$r@1UV
z5)=@=jrn4Tim4FEf3hhHugg`zFP_5m*(WjAea6g$_i?@U8AqJ7`u@o$H0kwUN(F3T
ziQ`xKW~cNGg9Eif{NBjxDyQlr=$+!eREp4cxda~gw^h^BD<A33sju`dOv+@A&!^lQ
z96-JN@CqXM@pFSfZMj1O7UsB)t>0vD{JP?ReZ-LJ%&byQ6ILO(gs&v@9kLw4o>UEM
zL7_dlPAFt%RIioj0F%QkP^J-+eqJwoTM^5#5?USIeCB2<(^sWE<J)F!%x=?Gd_bFb
ze25{mG4AM@`ySL#a%Q=G0HuE;=r?i%EC+u&@7_AtsPnTV)c7<29zX=D>VT?GzKZx{
zvt$gZ-{uo5fn}aHAM4rkmRpyY!!#NHs%-CZ0;hpXEv~&8&28t!8f>ev;*0U+X>z}s
zh~8XnIq&y^KK&W&TBNBVxtEO9*rz2^qY!AxF}B*<rOhP=;h_K&p=U-k{#*eM#!1&!
zk2tHkD|3qfDi{2!BQ+OmukiucrZL+cJsVhKiDJw!$jpmc`cNqymLXy^2hHvx0kj=a
z)puuzDpPt<<@%g??j**h)D<5TZV;q!kLU-X8i!4BRU6cx^J-(86}m*Vj#0q)#)C=#
z>(^=;@#_Qvvb6kKx=)W)p?Sm3#<0oiR$VJ&x%bKw>z*Goly$lJu?Lup*Jnez>$0zp
z9u9DYNgY7!lFn^UG2C(T^k!3p*5qkdUZ0tBwcUI^dce^KxSn3(;&fl>g!Y!tn1D}(
z<WfbIWVj=#(APRFFw0p7=WL$R%_nZiN?U(O_S*1)jivJC^9zl-PlhafzAq<5lDn&g
z+*@f2eEO=V#E#A;nPhY%3g|@FZoXnNdr`4ZI}vMI61{rXtK4JUS(Q>4r@RmG;2vmx
z?9nT-j+JEyJe(bFv7A9KdpbjSDxP)hT~t;`vqOSjD+FbhY{tZHnV5WH*wmf$LK?uN
z`O$YN1XQ)@>FWphhYTdK{9^%48cxi$YBq3RJ0SKjYsP*D_Sn%n70rGt#YC2A3$vxn
zdW_WB`Kci1q&Z=RVi%k7b!%=+2F88L+jrE{coGwrWb0iqtZ0~iY5?Je&lsh)8BRO@
zXKQK*`f8CtZd)$q?Vd1Oih&QX)A6&*CPfjx=-}MSBpQa6M8S*V4AvEUfz&C57mC$=
zn8!`90<IcS5K0zSo$QM+uk^FQ=?wH@hNPF)p^M&^S0I9vKJ|SFV~5aw%^(!}gl5u2
z;}p5JWOu#WlW4fD_roC<bEAeYy0y_%b!CQatM>l%Yd)rTK65GT^`^3Bcr*99yeS9U
z_aFNk>}26<<Lrj6!%7Dn%dD-?%hZ>cfybbL!QlhA6K}VWCGtnrXw+U6UzhQwSFm36
zBs$n&d&Z=Fut0B7Xuw+171YH{?;hCV%6USdyxt7veg%X^_n08>qcn(3=tEk3Qs0{}
z;PK1^d(fnaRO@2cQmn@gas!D@)*kkNUJR<$^$pwI*NA+TZ7tk-IERnb{h9o$t^VcH
zkT0o5og0Nu%j(fa-8prW-c^k#5#!pFf-m3u?#`vF_o)@-*N+ILXG>R3q-N;$RJ(|S
zMl=wHl{eRY0J$!8r8IjssfOI3QN4#YF%$ZGq6yxZ&otb)^%v%(4?2lP+(x$sO0yPU
zbM6`%s?dnbqO$Hey-`H4@eT7=)@Jy3t4HnJn;2>pa=VjN%2X?k4RX?8608;iBwaqx
z5qdIO6%LoA8mn4-5CH}+KD{=qu}Q?xRmG^{L@NB+94s9TpV6ZHQu|%GbhByW;Z-be
z|C9G-+q(?W*HB6`HoZoP2yqzce(|khl*pUT{UQ_qx|zn8bEY_06}c+@uA1V(&v#xw
zVYzzm#4ln9EfG4c35Mz3z4Y5pX#0ozC!4_MEi4aXl3sieaQ>E}%_2cY2Vl6r_;k<X
z>G6?iVlo6+vJ(-@((KNK66keN(~ef15aMy!jj&iP%GqU2G-*-xaXV%$;!E#zr%h$X
z&#j+2WLCXWj7B72jBU^SWA6<o#2HjR&;}|-MB6s4r?7_F%yr!kernsblT`KLfbRX+
z!@4tW6PfP5x6hjdUkSw0@c&pDs6eS0|9D~+m&O+ZhUwp}|MXu?=Zwt6N9<l~u4@>h
zdTIGeg>JX42!<s}J5%phbIL~}!r_ZCfphE=M@g@_%@|6!HO>r!qLU^K!|!R^2NLMW
z3GA&Rqu<R4W9eD0w%L`ftmRu5XF+8_=IS0UIosS{ahlQnk-#!Dy<R~W$NN{VM;*e&
zQUTWr3aSDylzbNqn74+ntE0A3TqspN9EF5Sk|@2n&^rqLRW4qam?%HLqV(G49!lm~
z0X5jyj#nbq8MRt&^lBZ>2H|N&U;oc<0WcCu4$lDl8yiv)mJ_ER^Qb&~ff$;`5Gr^Z
z&<Sr8I}q%aGOQi<aPFv8;+|;S>^4|dQ+I^$?aK}RW<sz3|Do(HpyFDZwc&&Sf#5a-
zclV&d-QC?KI3&2cTX1*RV1r9=ch>}Wcl{?PIp-wzyYIX1znI0^uxIz~u6nw<y1MJ>
zHczLiv`<GuKh}w_<l0>JK(m$PfaQz)a26X+&+TU=qw%<tA1M-4x6u@TZT4hCKQS5b
zU%h70<@Y7Cs4`kK{@$6YQ9vn;ySfzCX&<fZPDB}sv0da5JM*}H6hp4A!p)vl4e!Se
zn1|M_&njdO67<Chc?C9QI@ZFtsb?R~;cnYYS)!C8;ji^wp$>C~U(WVEMF>H<yQA7Z
zDZ=P=KfCGh0+RK$eT0?Q0?;W#J#njXywvi}l!c-k$2z2h55>2?!R}PJW$#-ZncAI!
z{Pv=UTCegPOX_oN+rgX_iTmx-Z~<(uV_C<ol$QGBHa)C*&~PU=l#p@8{2gWPVTJoj
zD8qvz|C*t~6GuD${`PktzQbKMv5)Scyy9AJCHEk_B_ueekT#Ju2Q{Zu-#x^XuG3D2
zI41V@UKcfJwGuIr8O*FBBY2<sKc^v_oATf9v7mw}8gpbkj~9DpC8L#mEzmbG{qpF6
zA@0Fi|8#;b2w+%22LKlDCJub9k@mJ~zp4VX3-w(r7Oy^(ABW$}eUU0wQK}5F_PwL_
z;I!VnCx{sRBiS&ljF;^h;T6SN^3D>^gP-gr693ZJ?w3|S|6bEYE!wMwxRrD>m7$Vp
z?%KVAr{L+)nYfJO$=P>6^N(52>2diV?Q+;Q<UEN~0Z);-(ZmjE6s^j-?tECPpEpx=
z&RX{GXn}QfSK9-3Hr?*Aqb)9FrN_pQUnw=W6(4ChN2CsCbL&~&3hDt^{X}L1*Skmt
zja92C;N({vmI>huv#7wX*dvXT%vY(bOtjCl42m~8$4ZecEoN;^d#$kglZ7pB%n;Dj
zdHi8|FTWFX(ZZ`O1mBrO7GF_mMq0L==4&i3t^(sgyNqu!h02z$K&dN}KjJY~y*MjM
z7=b@_%C-xlCiYcaeX#bNKmKO3wxGQL-*qJ;`GtMzta{c7Rp&a34;2Sn6v>g;Z2Pc3
z*tJ!Py>bQ`IKp~iKz4SksxD!L53igtSF)Z<(aL8^cIZH>K;2ZVfbIyZWZjhX;d?6z
zV43<7Lgcv!o^o)(IHfdPofT`-h{Z^6pi^5x|JLZTZ_n6hjzt5npVP{<<hC#zh*0bX
za)lpr4?mjbeNrA8+|wMAI0b>+Lk=6qq>2(3rTZadzq>{`eJ!b_L(%h42IJ;q^x<8J
zLp5|{F}2%UGyh@niMSRPe#<I@TtKar^2|}AsGL^>#k4&WX^F;C1brd}&x-->!BYmr
z2zMNmd7Yw-HmVL=^kbRL<=L(4q(gyoa_+6kjzcuu)2BL178MRl&qb42;C!VPtjg^$
zDGoDx{90qr{rLTx4!r$K->EehX3CN|7OIN3xZC5ycO@&~q;ZyV2XD<#SAp}VIv`_&
z6>5Dx-~X+19@Fl#WX#2VLH~AN1cpRKsOD9~a=pcLZ&lT3(@eEX0F3TH!2!oiL|IL$
z9xapkGkrA)#_#z{umZg90VgQP8hEmLGuGtB8UPW!<Wf&L$w`alYpv*36$dne3!Syb
zqUY?V#sMhznjTicLv<;&`Kx4vPy^@tbk(v~ZxhNmsihcrbgnmC((>!u4BVHio|IBW
za-0S<VlE{F?gh+XLhp;JD7H*4L-+Hs2kxAdRR#q(a>rws1#s+O{n`7+%k3fC76QIB
z9C=#S?kkpIbA(KaH#hq95VtR+<4haKhwA}~XNC9A#|LuVwJ#Olv@8fHG_Fn<<gJi&
zG1QfAFj)eavh$?33kG1<gHhoY@Kd_Jw!XmZN*zl9qyoJboqG>Org}ExwH9^*fAXey
z%Yp1=-z6DK+~v+U1&S%tonw`GhmN=<#VMVT0+Yjeu~vHkf6JLCbn5gpyNOy5joP*>
zj;~Yko}@bV^4u9~`viLc&FO7qaf#b{3UGm&Z7#sPj5qDY9@z%ogvK&jw>C;Pyp*Ok
zjk5;hCX(aCf|$iCz7W&WbtTveb-JU-DyOqAK!8kCpoV9h!jObs7PkGj&P*(SI1~Am
zJvZ3f(NLV#!>uVNU0F4&tR>m>h^!m7@b&$~>8DvxPL<n$+rXN`yCF!lF6V%@#8J<9
zYHi!+-ovsA&#5^>22N8!9oE}pzxsn>BX`ot;|RCu;%FwgiVACI<in==G^0w3wg656
z2m$wfDs$VxGr2@9x8S<{Pbom2m9dVGtow8c%KJvWvelHmt8HhofgW^)c%@YrbJOCT
zK^md`tgd3V-|H2ZLn|IBf)9Kl?ewaZcVy2($dZb!uDFVtF}X}T&Xg9Z!~OkhzB9@J
zkI^djD*CArw*CksHiC7x$}UA#CKtl!JxZ@Wi{2q6Q&C@#@#FR0DfW{zF|}@v4`@2i
zd)b6ma4N-<TGq8T^FP3?<XUwzaeqAI$%D3s+pT5~FlQxrMpWxoyu6OKaC;V8EQ7aw
zA79Z?gMH^Jmc1}Fz)T&gky-+mYz8Z)Lp{i98V9^N1dLYH+~{~7BJ`rVlY_PwVW=ax
zw>~gS_@fu-uhOrSvv*%sxnuMOcdpgjrWT*D!4IaPf7q_KB(b`oChfI2{312Wp>xU5
zUwP<_dis@zlsRAaSD=2Uyq(^Pkkn+ezkj07ipeDW4bJPdPjkZ*mN2YNH8d6oWQSRu
zax+Pl4+mX3r_&5opD^|_{bC&UQ~eYj2=VTivN$bIaVm(5+3{<iB4jc-*QJBO%A~n{
z9utg=2<ABz2{h=ayUEIx+xTObiXmN5_me_p8zu^)<AO1*$QPJGmc*ClS`$^MSh|_Q
zX9PTwg2P)QCvOaDuh8Epv0~DA<h;aifMPPs*knv?*u&5PC*LA)y>~;cgS++-W5^iS
zPH@!Gsjn;@m93|bK9q6iQLVSYyk#nGI-8!PzQ9C(6zWx#WVZ_G^*n77TfAJ}?u3bp
z6`zxSgkgeHF*25k7@ria!8@QJ4@YKcj7qpqANP=YXSUf?Dd%kQM<njONc$b|)?6p+
z5h&%3@i?bA<EhA_@?AW~8GCi2mJz{u+UkHz2t6YxbD?mAQsx!qNB5jU{L9U<{W8uN
z+?eA~qr=*Lhpb3IwDF)V2NS)fGkHo-CmFP}2}B@IIcOu=gS@foTvX{oT>2f&o63jK
z&G=q;_p1})VPT^H9@I)s<N0Mv7||^=Pq#FN1;xe>v&w@k#c)cG8r=!z(^tl`V*>+3
z78_s=OSRDo9JW}S1#aqGBUqb^fW-w+_B$XhSMrz2-;%uK3Ccbd^DvTPXG-KJQ)W+^
z4F-~?8SeUV*?p`++Sk+A88dCz6GH+;Xj>i5(Akx<CbJG*{B%5;!gXc#vPM*+J$G{b
zrSmt7J~E?m;1yTzRYdrt5mSb<b122MgydMbju5f(w@^C(01LU7T~oZ(!*B`wN#T(Y
zwMB9!;t+mpn?gpHC9AY`+11@4o$HUriz4hQPqr9D!xftsyWGV%WYC_cb5+fDoysz@
zZ;wzRQJG}4rsplws~8#7%L?L@g|iJ54-qS9d*=Mn^2R+zXosd8_tv!Jr>-(0c8J!_
zDNDSR%KaN6Oi}`3lupl8AFwr^LRd+A>zT{-S9b5H+D>9~?!bgBd+Q>go#$NY6BRN@
z{ZJ#pIU<Ib76X;u*C3bQ?Yx!6N_HQ;J-%M3Gzg5EP;~dKk>|GkJ(85^2r-rLrIoMX
z&Le)+Mrn$tk`H^&7wz6N9%Q@Da(Utkn3PVbTc3$a4M`#jI7=(6A9X){X^oz7Z_XXf
z6Ei+Fe59yW<dJh1epJ<rGrjFGwN!Ify#*M5^4Gv)DsQr+ODUnwQ>ah?t#dU|TVm(i
z0qNFGiZ?Okrj7AdK4bt;dV2Y7vG$?D1bKTMaz;07&a6J+`#*nxFTkGowNtd+30+j0
z=^=)scyInd42=j|RPUIU8c;F9_h1;Zc4fyETnHazA3I^wChOWLR8=|>!A{<~F2~VO
z6%lkS_`#LNHYlca3X(3?-h#h?e1nQ#>boC}-5tN^&JyhQG_H7s%Ln=8N1~-i_4DTe
zcA>*mNp;s^iDKq@cAq+OI!0%yZfbKC7W|$o|0?QpS^!B(l>=bCwz?c%-#KluKd+_U
z*fP!b)0Y!{D|RY8F`PpmUWV0EpcQ=C6xEy<DGvpL{f>R=PQeYibf^~Rik-#0Q23fo
zKe<%%@>R+v^OExuiqdMl@GT|C3?<w?@TWa)lQfV|?=TdF5A4nWkEYv}7(DzXZcU0f
zwUq7;Ri~RRUPR$Yuq%i3sw<^G?UIUt-Q&ojRUB~0uHp<&pq(o`2(Q9j-BRSAtDNqE
zLI1x0*}D??%ahm+l^+lJ7mvt0;_fG(OPreF{4t(Bm$*HZI*wb;=#I5nfI3E2B<k7q
zvx!h%w<=!OUb<Bw&D<sb^zfRYCA2$5oWfGM+hJrHwN`3ft=6t#T+r>XIzm1HaKJhj
zc_N^``?>5GZMN)ffyMG&{3)=CrE4*Ca4%+WA!t)+CQu2ptwrutPMDFguX^^kvu<Jf
zO$w+qL@@S5aqgY<W=l)>+z(^P03J!WkGg*)JN?*tmd8u=iX?g3TF{q*QaYb<q@r{^
zZ^UP$&Ay|&ft~O+MX@%5FIMJ(+i{;wZE}GlPt1yaTKY*T-<~%%VW#X{nNzHk<WxwV
zW8y6IWj`bwyj&B1FS(9@eLa)fB3kd7zw?YunTA_Nq5rdDT9gc6zO*{(Dxq0Po_Vu^
zeYhB|c&v&&Zzlbrm6}e(IaaEudI3@B%So}F+|-hOv*#zQY-!#dsNOvP#H=(HL9z^X
z?dmJDa_O?tb!iuDM`<<qUI7PemD;9p1@MD<dG#bf<Zcyb^{DluNhY-iSx4!RL!^|B
z1OJpJg^mQ3%=t;qaY~P;9LmSO1jCFXw<F8)aI?plRWxN&cOK$z#^c6ilRI|@`>NBe
zFz8uT$KGt#adA9P9oJgirSX<Aud*yYEtHnt)I_KkB8*(+P4w6glSnEbyrl^Ea|_(M
z1n*<$Misvc7oR&?nTBWA$YJRA^9-F_y;i7fTjxaWqPUK<yaQ+4C5PK8>FwyKyWyoK
zKaSm})&qo>54B*MWrbq=TGf?Tig2(&`{N(pYc4|1Uagt)J*j6%lyi`7CPi+>TpYL+
zaMlo9G*?}zmReZN-W{;Z0!cK<urS{|D%Mg&<m_XSVkT=J81%eXq6;)2Zh|auTknSy
zsySzxa)Y<Ex-}~C<c_U{Cdl;F^*Q`D;Ej6aKY0{^W!2~~NWlHexwez#ZAp|)-JH`6
zYu#cE3$4Az@!k;}WVZ&4veSm3JK|Rzx)kV5yjcanz~vOIXF9z<8teM5w$~%UEhV3S
zMln+-M>YxhQL#3!bgGJlD!pUc*#1m0!%4SKlp{hpQh#1gN04KX5Xe^7Na@004WodO
zrJ#=eHij;jx<%jb9X}2k32T7Bs5+|*O2I;A_}cJ@+sLJPw?PSaKm?Ef%p$oaIt^$A
zBEYSH(Gc0n>CU%?yRFzw9j@XdRes&57DMy(OeX}!IIs$@7U3{k^nnYhjw3^+BmssW
zUfi5#0sXG^<aXM@{`5ZgATeY=+i1GFUgyz2Z{ajHG$WRgjg1Q3kg8b3{T2xFIw<DN
zdgFylo{DWOmPdrLgu^eL^O4SaeYN~11KC$Zu82u%TOLfsO#|#<t~i7@7aWt~4;f1i
zZPn~Croh%Z1un@`CfNF~s>pY*s=EAs=X7aLgesKDG#5N&eyc=$fyUp~H?ld6{)Jmu
zQ_a6_q9SOrYf!v|2RWE5&LHNFZm-wRNzG}l{L;o!)G}y4tJP{@RJ<x9Z6l{>4&FRt
zb#BeGIC28*H9+;_^G3TNN$XR-C8m??GC(3Nqt`N|&d>d`)?`=(nNnWxWL;xvv+-#9
zW<}g(lQ&ZmYayJMX>GHOCUgBs+*W#<F*-!`TqhX2@!F_^ZA-CfSx<>L59W6e$Bb}}
z&A=}M=xs}=`wAaZsTWd&DYRroxbk?Ce02dhCI$s9Kt{*2)|luoT{9&%E=8X<6&Uk;
zFz#mRcA2eMuwa-*>3J?mRofqlEfc>U!@a1?j9^bIv%asg-=~y^fYTP8k|b8UHhu+K
zx!sH3j;p1!Bd&HpTQ4D+lg0|&6=^KLQ`0D=HPhiN@>l@4)zGZ!MXUpXnup`+cZgK%
zntRb|D45+v5~vX5Evus>zVz82;Ltm5k<&FNN^{IvHL#4?Mpy=BXjN6G?d)on07kkT
z1MdnG7a59&hp9DgThDu)$}ENl_OLzSG2Ml+tOr>O!%1hgYzDTO*v-2~W$N}av~47k
zr~9g>mCaY@vy_JDR+O=8vg(0a=lTo6o4u=f?CBss%6V=|T0362*M%YNtY*FT0F71L
z-1ij?-k-AwZt&8!tc5O6tT1A>#H@r?xXC?EUt3waP-5L9a;>n!z&&ORkFJRn_1c|(
zabOX3O${f+o=W4|HfHd(l4+@@qGU;<`I}jksyj8#OY6{C^fhw?0A0EUZ5Bd`p)7+E
z++AT%It*~%8gr-O<&#-pZKdj}W=Ds3eSpfx@M^?JC2$uhFAxn7u(y!zXYWq~oV+4B
zm}*8zB6FsY!Ezcmjz1mF5@sE9CnKZ~k2Jx<oapy>=di%zjIWbvr3pv!DqA2>e6e&L
ziQUAsw@?Voo;sG)PQC*l-adexJ#MO--EepS`Vkln`w?2B;>#Hf4v@{)__o*9Qtg#a
zw7U}4xb}>9O=5CMb^asffQqvb&NiD+bLzq3i_ZroY|W=tVV5Woad&=7o($RthWWjU
z>uc<-%u_s?8#~7dYVKI2W6B)!!|a#wI*c5ltZUACfmhR4pOjklJZE}jn1aon$-`?k
zx|I%P?okyf$!`7HL@QJGx7~Mh7^~maE*1}zjuODgokPCv)cJ0hH7#-FzPqnuQzo6)
z)@XW2CyBW!2eho&tAOFBl!$rIsGYgiWlQh4_UI#c!1ELV@MwesZ()>+CC_9|i(3uW
zsx0Pd$@f&101jD$z`90>Y0?yRDTC9+q?C+_zUDNTkB6}Wn!oRfDn$R{A^!Bhvzk(C
z$*zOIlU1Sjn0cqEfgp$bq=eLZFX!$qa$4#+nQ_5&7+DZPQpA%*rp2^BI$xWWi!MJz
zSZDRfvR|MGM~HAacqiK+X+{+?Igsn5Y*l^p9aG4@)xDp2&KN)`U}VPAb*PXMlYFyp
z|5OxQJFt^jxt_1~SfjOBhq<aVSf+(%L1MjzPahAq3}UTXL+?=qlV$mdh)>PD)nt-y
zWlK&`JriG1OnHCUgv<{a8|G{x$X9@;35-x|{e;Tysg`K(Qi{zLLU**mwC_|%-HmV?
z5aF|@kZLDT?9|DyBud!>8wq2fY?U(-G%fbNWU0s52yn?vfxu95RyHiKiB{J#luw=D
zgjWKOsVWDI3^jFkH-8-L0d&VUCKdS!SWOTwQ(LVTbN9~h8|vd+>4&VrO{`s=U5=MD
zU0ExOaETc|ED;aWP8o`aM?kN}Z@Bj57rU?3?fY&QyxBJ_$d{uwMpoKfl%ZjBJk7nC
z0_&#OGT05^s+cB{w$1>#)Ot-Zg5j|1oJ~N^p}W9UjH}H|vQ))`rjU@pN8Nia!RVC3
zjBnAK7@o)e$0rsVqG_sRD|)=TS50}@H;v!EWON)$mG}~psO@<^&f3X8GS|lBo96pN
zUA;kpfj9nMkwzW%{`csWoeX}euZcT)T<j8$%)P@xy4F4Pqf5!K=)LE_VmZbKde@vM
z<E7ryiv16WZK`zUzFWJO^|?(st<pQks-XkhW^Iy8Xy@=*1a6=OaJ08?A{V!t^e7pN
z98ZIj;{%74gw<tWsD}=d1MXs!?B%0{7pW!AD=(eO8{uc362-osnbl;h)y!8#RLtCE
znUb+f49O$B*-13poZJ7JzgfIA!}})dXZ9B;Fba;mvvusSxhyV2@*S4!d|U^5R(cLA
zBdn;C;hQyYadpmGNKSEQ`kO^Jj1Wb(G-#thUZpzQq{9PO`;S#Or*O>sQK};Hte`Df
z?Zs7S+O<+7g>coz{w#$#ljK&)Ia?C8kAwPk@(~MZ_C1{hNbDg<(rR1W)OM7e!!os#
zkVA_0%>uov1B}}P2SNZPfrxoj%RVqy{i1mf1)`xsEVuy<2cVzBIWT2PIaC<k@?l%l
zm)tCnF9!xK#5_=f$od&W44!0~sH=phJXWQV2X(f5Nj2}0lbhue)&$z7j$1b0>3M=M
zty<rLqE7sR9=enI?NWl_DpO&~B|^{q@mu$!PNA~9M>!?A&8(t1513gXy(+&%f9_*)
zKJ@+cTF<#UYKGu_Nj<`20Oa$olbE=8#+89woXYJtxKJbyN7dVI2ct@L)nXpQ2|=@x
zfw8Ar0}XqHL|}N~QVlBLk13=ieyX*}qe}O9L&He#G=kE#f^De`rgim$@W&C{YcUNB
z&CGY5_GR=p{Fj*GPpX4=Im%=yflmRTxr8DXQ;~&K*^~}$XOd!hT+UO%-qGu5vdi?i
zi%E41$?-x>W+A2OR>hq9Vc<MBt@gg?A|i(0dd^)xfsqkJ=K+Gkz?!P6Vk`th#w&2U
z()sP`gyh8@-Z%GsSMukDRKLK>gnD8^Cst9ERv^VRG{!=@4M&rX(Va$_r(88=TJ<UN
z0fnd(JSA{M@X*-(BWZcc(VniFh!_sUKD8?5vS4-FT|}LdgJd*H%NT|EwIWX}&P2e(
z`c+Z72fa|)$|*}p$plNvj}^WgyDbH@f{AQezV*BAOSyagZIyfAGzCv0f6*GJrIy}P
zzL8>D6$^1)b)0YX?TxJzwazqSN);canyM`Y6_Z`&Ck#{~GRTGW$ed-1SLS=jxi8f)
zr>FO@SVVfATVI;)Ic43Tqn-ktoNd0`-qYAq7xe+)r`h&_2^x$8CMAs!*7xEAu+@oQ
zX`i)peeJ748PpyQe8M^~-Q$aCuT-@c0-xw4&0b!$#IDw16362h7}0!r2ss~ck2X}H
zCeomz%aY?L)7(VD^H56%<h0!O?d!BzTOtZoa@n?AzK>|*nTj}TG|b@0tK&pl^+ca?
zNSOehsmV9%s<tA8xMXRIEoZL_*iz%+OnRKimf7mXZ{oC8@RuuJ>th)0E0+UJ8^&X&
zMy!>elmoU1(tt*nrm+Z%ze7pa32ZN6;eB&jLJ6I2!P)8imOH-F#<b2ljjFSBE{^7M
zB~!~NK6mcTN9efyilkeY06sL^XZ6d-`;!5cN1SOL(?f)h)TOOUrN?=gs%?4H+DfXL
zf%Wc~*ctHoA5Qy=m5l5NIae~?)wrAW3ht4&Etn^lSyn4OgEJiAsZ3L_6owdl=l0}y
zK@)Y0O_*u^Vl(6Z^vzfgZwRDRq1Js6PZXSf_1`?;J#oBeq0KUpdHdhQBoDiqmW{V$
z8Uv%TWG;!44*hsWI8w(<(#X?dK-(y=#250}PQ%UPv?L8q?X2*6r`<F3m+HCIX}2oq
z^Mb9j11~*1JEe?N-|y}C!)HHH;_oj}Mj9*ZykK5QAb5VrflArb04}pDp8~T(ufh+`
z?pdVA2db8;3L?93>eq<Y<5Ct1*t#1*ix`yyl+x%|5(YAZGM~Dwuhb@T;FWs6ONyi8
zQ5iYU4Lez4rN+U6_|ZM6Snb!Fjq*bAlLsO)oSump)^ZIRv5yCfL*^g=i@10tS4lON
z-3`?06gi2srQ6m(Dmo6};b~g89caA%s!-?Y&R=5AhW~NZRF1E<P)^i5)M(W?zDg@c
zdb+LNdbqxsd)_Nk_?}BTMcYOk6Em=6`Fm44_}(^<u9FrH{SH-PwIn=49As+ZHY*HN
zVy^rxXYoTf&hilIo)5rsyXBTcFYD@!K5gRFEw9CHe%?s6kX|D*y)%PYh+#)&mKebP
zJ@2rQT$Xct*5$N$<qpMY`|?m(O6K!P!E--FTlxUvb`f`^R4uI!h`mWXr)Cc?uyN~k
zGgZ}{63EpM?Vhg6zhVD;S=kwX;#nxMGr5CSrTkg~ua|~4C25PL2)4TV%ME7mU7oT`
zsEPyE@I?IZ{#_r*=U7KIS2U$f^OS}1Dy8Q_=+-COg(nX<$f5=^BzrFMk<~NV71At^
zm8f#QTKYWGYy-YWf&CI-6ogz)wHJ42zx0ePfZKY#p=x-4CdnIhc#r_{m$eF0@vbS5
z-M0<<$kiF#$d1;H^*##)HDhi+T|#}CpNj?$qL=J>K*bk5gv?4^g^ZWgcWs0>wXB1=
zn>9?)Bh?d|M>}HQ)Cg@q^X82NhpL&y8C_TPV^)Jb!y2-1m4$O8F+z!nVjc0KK9<1!
zS=!e3c*^T1ZiHCltcS8!RG-RNS9tW=&^f~6vo`hilc`D<V+P1e>h>`3JVO-hW3=q2
z7&Mk%r<J?b^yZXKLnCM$d_u82=*tcZiR@MkR(b`dPBWD<Lu>Kj{grvmamkfh$tI~t
zJe`D9t7X4Vz+9|_826>~u0D3C28+tl!qG{JrA>Ip`luv><;?@}=KzLoE8CA$jPT~b
zrJ_65g*qd9SvU8U1JG_~RgN_Wvj)0b>1cbmrNEP@g*B5;{<BzA?pJEFtrd|9gK``U
zb-wF-J_Hdgfexq{Q#psq=vuiG+hpZUd36mTlvW6bp^>8}EoQgs>cav{cV9yRxV=Ug
zzvD<)gm-&g?7H`w&u(OO_x#gkS@crOyDE?JTIvfW3mQ9XMUVMh8?D!}$!v5q35IFL
zSqG>E)H?n}ktnmBYXulgs<P}LFLJ7GG&<nSUaEklkoZ)z+-%M`L!@z{8{+a_|FyYe
zH4Ay#+TBa!yLHS%Uv;jSZo4zra<}U|+Yp#eF@+&tdvIY1&<rPsMC$#o^&43VM8|+N
z(e9I*E}jUJ!S0(k*0DY8%+t2~+r@I*LjyCMJRjM^a+Q0Di~OI=*@|ez(7PS=*Bv&_
zEfTxGuogd>8Pp0m&xGL3R#3Vt)o)S~4U##{$45I8@PDDi3Q~v$0%C84x!RaqJaqh7
zjB%Ek)B^$hPgz6n$e3B-XU}pscCj~o&Cl#r&;*v?m4Jj;tCi(iZ^>us2*otMekZjm
zv-E$gkmYcop8T@1BV5xBd;SH%PLFS2sPlmVw;=3V0()ho3a71sg%sn%NkY1&zJl{|
z$BfNOa?e!5UMFtNLl~KEQ;!3U`NN^mdw$cD_x9CK^R2_wShZwbb@i2>{Wlc?Z(cpE
zMO&;Fohp#5CF;!&%rvu6P(N|ydNT;r9v>7N2$;IDXQ;hB$T|*Rf~X*lxKdL~Qp4#u
z{%XN>FzK*R@UAK}P#lletjD?P55C?Us!R{w9XyAErFtFA1X)<?B|PZINMqZz2aVkl
z)P2MPI$90S<-M_Hgt$$_FoBHm33GFfO_K&VYYJLc+NQVK@l6$;A5;AXQPl7nN{@Bo
z)9JxUeO|nxAnnl3{Yhs9B`NEzEWIbPa3LKpH&jx;(p-MYTAEyd;S4)n{n!}rF))Qz
z$2=|r!PLa^24@TjbwL^etIP68#fb6BJ~AaQrMj`6{=pq-J6^}iSaAt=X7QY23u!iU
z9LTm}lQ1}3eC6s+QLIxL(R~@Mc*%zYWvXrEPOx15Y_o@IHh;&!^ziOoq_8t;c*wpQ
zt&)TT=R9ZVY~d31Okcu4<oV}q3HaX^3dc0Sa;9JJ#)?<Ac3&zFU`Qt`pW#TSexD-I
ztTu0n<V<z>4C#GJIo1MNipyvWo1Iah`+b7=_OloJR1E3v%SgLo5pXg6-|zhsi*^yj
zZ{9Q3*FmUJS=ew0#vJ*RM~VquHqcbxF*z9v@zwAB^IvCS0qO0r0kZgO`I<@Iwf_sd
zJt%?MfMjgf8w12eTf!yQ@Oz8@;Gy{a2K^B@h<SN@7YTb^+UL#hv9SM?O6<ie0^wp?
z$~X9mf}K+6|Hf_X)y{^hi{H(*M4|X=v0PG@T*L5hN8{f{|9jQL>tzl)XM{Un2*e92
zc}wwgH1wwQZY_rM`zu-df2Y|1h45?+JtXIPGsiw7<QC#kwcKdOFr4Iz{;yj4_kDj|
zk*2qUW138?@rFQ8tpz)>O%VQg${U)lov4%afi~a!chUVJ({)8J^NYiy#Ir#9i<v}U
z07z{T0&c(F`L==AI!!UGxn<*zXZ|Rx-3jem@}tu-<8mSTD~iS}@${d!?Ihc!hu)qp
zFv!3Ak9qREnItcpSgdFFG{gmSq4CGkyio-lc%@ztB>#M0dp**(>;3w9G)3_L9cv(-
zQ6Cr^sn&6LT2FV89MqQDG7b=Z^`O_nPvFsyK&*W(PpuC>PZPTQA2R#*PZw?cRsfz(
z&hc%)xEBBdl%G$CI{x$aTMh3UK^ksvtVMOX|H9UdB7xruiKmm2p<0&99R8z(xxliY
zatI^xzVVYH`*_;F#8ynI^q-lQGq=G*W|!B^&$wpaQgZTpixEl6eEeGxN-sHVtNBn_
zR7flq3zdWZGZ_Zrj-2-mW)riw`5`F-*(io61KF|r!e9CIW1u+fXR7pXXtS>M&Wbqy
zNun=68$xsuy9;9?C&56EOFP0&|5RXR1Fsu@FMi0DcGo}F;$I>Dy2>O+)G?>CGH{nh
z19hC##%)v9AKp_tILbFcYo9~eWZ(bTKS5@Ic#KDQQ~v+P7LN}y0ZxS7^@}8mf1PE(
z3ns(;h>iFMIRBq_;UFeH?Uh(=z8h_xTnynECp;Gn2?mOoBK#gr`QJMzs4Q3wFEV-9
zlnB4SUb2kp)Y&n)1WI7sFzI_{o1GnA>%;nwQU%>e`~vn##_AXY=6r<1dRV^?L<KtS
zu`rmEz<?n-KIj_R_`^J)sPeYOrtQmkFd*D2oALf6u>618@pmhE??JrxhmT*2yN~y)
zzoO{T3*D1k8^3{+zu5A~nfW1jKlCxWoh*V!zn7>Un%#cZ%y9qpP!D!GP`cTea(?&C
z_aanFqw&elpt?;qMGpK`H<fk|f^+<k;HF5^KXy<34~d0EcyZ}vPlNVAcgA#SNG;!=
zf_lXLeH7rw_fX_Pg)F-9vqTI3SF@diKX&vO|K#DC*GsNwg$?-18)93pP>B3PQ(m7=
z>FiuvUeo?9hyPtbIaG+N)fmgsz2!`h@?mLQB!Ai88ghEFCm*ZN0U!R`)4U4a4#%}M
zSZ$^dF4P^VkDL*>i)xDhkyGWj%qmbCJyknz$kAOi$Nvdc7f~RW#N~sDoYMcr)FXnn
z)#M%-%N+fxJiRsq=s=Dh^XWqL-`++8GqihrXo*oQCV2N+%4_alziJEx>;l0>pY>9Z
z1f*G%reN`(rgE(H1v>vCD-py$rSKj@GjSYm@|(>S45zwSk181aR=+)n<h1Jvo<dfl
zTuqQmyh-c#tC$$DQ?a#Dwd!T_ik2g7SNE~uWM-<{q(kH3^uskau6a?)@PC!X8|+0J
z1b=be<o)syw!VQ><|F77nfVHhi}#1wrfx?v&+QRlzb$fn@_F&-`(BUEu`V1occ+}L
z;pccdX74>4<BeQ_K+4mv=}QXv+Jr$hdw&o0-$9Fh1)S-FE<FZsDc-iWV}<mnSz%~l
z45|NDw3wKefEY3oo_;q%>1tzBeESEY$b(uB_kMN$@1CUk_ojcZbN_%YAMbWMJ39QM
z@2rlD3v|CsVl>}e|F?Z_Z?OivMbNw#T5<@N=pOKw4e)-?tGhKcyBwCpxO{<8(J<d1
zL;8~w=Wk1?$?Hq~@M6R8TP3L$Q`)=hjSUkkR>yZ4P?O)j?TcZVkn&WJyeiqnlk{^J
z$NJaMHj>cC!gDJ_KR&+dZ{P3l*WGV;Y}Uw&wzh_*kIJsQLa!D23ykeNwrg>_j7wp=
zjE^Pk_`eT4-dErkdrd0ZX`9BxzR?AeNsCfxjv=1<Laq&Ak#p51qaPVfMnvzR88LLZ
zxE-Kt|26kJ3@_6Z+OE{k?3%vLMtgWeYRH`WQn{33DAsj?&Jx(6C!hb>smG+elYRie
zV{>|g-}DZ)$_gN|0)>C_JkNo^MiFxH*w(a>`Y8%lTwWJCI=nE*2V=y~<VuokFYWqy
z+1b&>W$U0QaGvoq10i9Ep{$RnH0Aqkqsg}P<g%6Il_TgTM$pgb{rvn<`O@mq?ryou
z9St)6%U>Vzaqy~Gle$hb_$WzcG~sE=xIF&bjWg+&bf~+#B8q`@83^JR!2@YTe?I_v
zk{lWA<iYLXMvg>wL{UFv;6hrCcnYO1daBD3Q$rYxnD?)bE()6qwH7=!rzDa;od5o>
z|Ngv(_zDb>a$OtidZ}MOv1fg*vJ+K~c<`kGhnN?*=-;2{fERuP^|L)(tYsI-nNisn
zxUNU<p`%_vNAV&fK{;i!6>r;1XB$@+{ZVuj6ZpG#@8YiOa(rBm1?#-XLULTRw*{@K
z7DfK9*<W?&7sZ!wE|ZW^S=rxtq`^SYkz9}@5R#;?Li{C}AUyA(A!@F)*X$0af4qSg
z?34%*Q9-vjtn@}nzAu`=lQI9Cx0g2j->Uy;1HK11n0^S6S#e~SJ9p6PO10q|3Mcp<
z)&J1~9~3sXjkg8oq9+DjTc3aa^RM5r*kFlU>nGy|&`konyS+EcnjN?}EMdQuotX=2
zy|2X5wr_L~$LFg+{t!WSObeL5%6$P{50!~>t%^K6d0@uP`nxzq@hPCOIg-ndIwQ$6
z7srEBlKP#KQI6Wb@crI8lH?ahUm4#aAKzvQ3dne+<wOy_ga=FcsWx6PZSapDKW?bH
zG#VP2xC$};UacQZmIk*$NcVWC|CY`v2W#2iC#quQf9n=S-s^Yx58{3R2<ON5l9E(M
zU+4L{ogR_+qjnEw)W-Z2gBRV)p+PsMm#aSKmh?pbz~RS5COA@i3)@jYN;f?n<EoED
zY|dAQUlRQ(JQlEe^j@y_WfzD)0E2CQMDaIZ+82#`8Q<Li)sS%*S>k>dry)|ihDkk{
z40r|VhgbQRef?RiA<_2nw${JDI@<rmQTMCsjo|fKTD3V6a}J7J(HWv%ho5zhRRzn>
zVId-_?m8lJ{8K2=-d+u4tk*l;<Hib(%76@n5MQ2s;@^4`{!4-OrI-{~JCqTleqvq%
zp?=QtC91zb=uJ%Kg;f~_{*fqkgKggJ-(LGehKMhhK#aX)6brnp9T4n(LxO+6n}Nv7
z^-R!-J7^xhGBXOCJQpfB@bfQ4%esLhn{sch<+LaN0g%iaDB=k&XJx>)0c|UzH}fxj
z{L%{4algc@nk>-nkoG$Mm;8`ZUQCr<kyUC8&iszyDBgZj-J{`Q?fl%_#A!mc<hx4L
z@0vqGU}<r`Yb(l&&YSnSC75<^`;`J+f{=h0E=5QHqdCD(>3A1H85&i_Fj4ACO8(I{
zTf!))>#g9i0JiT&mA`|RI7$)R$vRyzutYXIFDhI1CNb$(MfGJ~Sf>~n8g{kn#*6+@
z7%?LA%cYdpocw}m%%m7MZipyh)PBFg!wX!WKeIjqLB9%xnwSs_RbJ39IQSPb@SniP
z_{`WpFur^LZ2J&g4f%(Lpk&&YQWmfGzkfpG_SecV75FY;1pP;KdBOY%Kyo^~zwC(Z
zZ4Q?3;t%QKVzI#ZiCGE~uieP%D)L<Nyugemso!cHr3-Fj9pQSh_Kq7572SD8;<tOe
zV7Jhy*XMjs00v@hklusP)#CrY?~fl97pE1pG$j7V1H_0_V8J%l=}?wIXfp8vf86Vr
z^t#^ace}_o=kP}zKh-z;)kM%3pf-;xj|%-kng5Rm^q`14TAT@M2w+Kt{{Rqg5q`Vt
zxx8)SrL?@LoJ$1rb?jea#)pPVbe{f_`v-ttK%??_uQiL%jWmX4aBBQ*ett?XSj6jk
zC)iPPZJO{6#hOt3X7o>p{@;ER<Gg%B5q%U@`SpM3;=liSN(kzj&3!$EOG86LZX`JD
zcB)m?)#v`bl=xha_1GNo+B<8q@Z3y4JMIZMTbziKuq59fvI*wRBtKdX&5jV%?PvL;
zC-7FGd~0#7!sC0sO1O;Fm$XHhtGQ)$K5KXue^!{>C}q5pxHMC;kHK|6<EgTKerY;U
zpS9#V3m2V?iN!SpJmlCsHPN143q(m~{_J|x+{z18RaIrUlsme4|31_XtGDo@yuADy
zMF8Xfkn$gu_#g7)l>?2sgyIXY<5+`++a3T$w8tC6DJ=134Qo&ps2MUTMO-pNsmwCT
z%$CG06Vmb1=Byu(ASM#1G+2xY-^9*-;o~Y&&kF&_Tgcx#eQW0{(r&@b&VC2Q!D8{C
zRQH&mQ`Z-k^l77puVC*TsdO6Yss}4w_L=-(?!Y?2^ZV=?$v-XHe^*Z>K1e;WsI%5{
z)D<nKe6bD9F*+Wnvb&1~fhHR(?)OHwA=B*T^4%fPGcBRo8CRFoWbSPWE$8genpD->
zw9yrDg#@vDoO%bJ$+Y7K;;53OMmN1YxQuvL)A?*qxNVoG6Z(7+cuG3SByZ)})2E~{
zrz)s=BXDD2BX`Ap-q2&+nlLjkX#W2|BCjTttSAzCsXv|FUD&)dfl2y`jBC1Az5Zsg
zz4VQV&PeR;>PFrt!qPZQ=HmV`R?Q-TtAk~_MaP3GeSwOOdr0mqi_vuV^l|(Bsd$)C
z9tl!PB#48!%IBEZM=eUyy9;t=Q-w!Kz2LO>rg7~t5dXtH`P<9-eUSU-IwbjpKdKz^
zaV{s-lH+xFYlCvdGC2yVXV&^C&$ZqAA_1(`np*Sei!ccp9@^Dm?aXz0sescPdV{T{
z%!{V$1Gt5&g}J(8FO@G=D%~Myoa-GBxZG}v%4KR9bZW}V@k`FfJa%kXk3*Ap{ml?O
z12dK!x#DpH9~GT*KCG6<J#7qHK;#w_D2VEvexv;5*!(xB{zNYCIMNTm&VY`B{G2a^
zdAr4KLqF;((9&2dkbg<Fu^yZu=HLK~I9g?uqEsZO3xvn}oK(H_UJViQD+KJNV&ONM
zkisHPd8c_Jbhx;M+H*hb#0esaX|)n#I5KBx=?DP7-SLdtK-@5g(p5=~r}GIkS=#KI
zg-8l+tLCFO^Cex_mg;HW;==U)J5>L;8zl>VCftistd>{DAQAsa{b~F8+lw7fvt(`0
z`xfYk`fQbj{)u|ISH0)mm$TZm;g~ayXM#;W?5l1jsj*7>!EH;PDT)=!O6cCE0;E!D
zq$txS8+o0%5(V(%0GgA=({7d3{_9VGfT2*k6Meqlz4(6|GIDmX{h6{DnSAN^_tkAG
zlfmV&T1^gOz%w=s83`r2qNy?s7518Ae0S+Xs%<zN(lU*@2$MMKls^W^pU}zvVI&pn
z*HWBTpI*JYDvR`?q|hl(rGBz)C{=9`4m2o^gc;!w08ncKVriqp9V|5lj=1ja<eh5R
zK3){MF+DM=)jJiz5B3ekCan&C0*5<Cryc(u7{?l>qT{KfPE~>2*7N{IQd3CE6`SzJ
zDUB`5Ip&MfXi^=UMRGE8F~?<q)DZX|^8I_{@F#rnvRgrhn-rxdl1lMCL_cg+PsHKk
z<@Tbr$v{pdCv~TYCb8o(SAb85&kG~mKL+w32Jdwd_`a`}DI0yQhtakhnaCO@J@o<K
z)&}A$$kB?Yqbcn%B`$@_e25x^z`6xN!?-@!aXDb}L-WIeYs4%xis#9gp7XXR%5Z)v
zLw$wf;N@FAu+RC?$a?i~gkd0S^Zx7MLJg`LsDG0waHr9Jirt1`^oK&$;(Wlyn5#&L
zd6KzaDlNEr%d-pf-sK>%RHID{*X>5_y;iN>h7mfr)b)46H<e(Of@$ovKJpgxRayx+
zsC{sMG~$=Kz3Y*juG8u^pYD6Rmd^#5I9~Ae=8B<geip%vk!x#ums~3Ox3}<1j)+8H
zBG-qpY36|bCz^eIMg5I#?4?RIhcXR9qd}Eu_OeFN$15(v(V<vk#MdDIsOzN}Ux?<H
z4@D0^BaPP543E?Bb_H-)e!ov6n2B`5pXgeW=wM%-u%F&^dcudKoE~ao@a&#v(#N`<
zU*=rc5jY5ni-%ksIfvxh9~0_GKTD4R<}NT+2M^wdZj|$cMVN_yW|ZE?BA((^JNBTb
z_&|x`H~x>|`u$=};M=@O`UEcinu5!BasPr4*WM+XCdcFG#pC)C8-E7YKMf17BDh1T
zHu0ioI+jwp6Y&VAha#{dwV(^zJWp(vPrkvS(@C-Ooftl~_tY1P#7t2rCi4Z+S>Ga#
zMhEWkqH{?SZgLftOveVXV4Q)iat>(zQbQA??NH0<>?9QB78hn_rC$HDZ#qXqWZ5>z
zMRD)UjwV17AvQQ_<||e#IuM*LSDIAYe)DIgei=4>6fd$jk~yDu0$8%ke!3X{%pkn8
z3BDNZ)<F|wf4RPz^Iop9nsItx@1BD*WRFb#4(viy&ahBGWvWomy-2AfK$GjEU#WHz
zUp)1_nu(X{cLY|QvSuZV1k!L!C%`)=Uc4fm3hCDRkDF(BD=5jVpA}pk6M$tJqSp%S
z(9P0<X2U7W@vrgf6;|$Tswi_(uAGoewR2OLjFR21AAK^ZU9S#c7PcV&*G3#bmul2W
z%WNavN3Sao=Hu=zZ!4NEH`=>V>K6TvKAYTrm9gD_1UHBrBP<<Zk^J0HgKP6yu~?3d
zVX;p+P8IA{WI^Jfob)Mzi+)R)M%pW?0~kLL3eN2@Q!q;*CsjFq=9hn#o-F?=!$-X&
zemK{CQCZHY7nKAwdWKLUp!{@9iE+IlV7B&!TQ1hlT^e+s+{E_h6<jVvCl~9M?gG2a
z?HKAlT-2&S)V?Ss+`NcvduAp-K0!j4D3DI`?Wq6FwfhdmdAVWWQ^XkEJ*Hf9dyB3u
z9Zy=F_D7?l?7-m<AA~><OB_2K=^sIM8v3yOob6{thNW@_@&rEf_SW>>4k-N11;For
zCNz;uBoYZN2<KmUaj`CEV!fE3S7SQONxr&<=b2qk7zD-o1Mm2i(7cE*zWXgA8h7Zp
zoI*6*#%mA(-qa6Lnl93{o}_m69AK#mQRdDyJ68{-F)kXSK~a48w{D1VFF~Av^d6c_
zYW(>)Odp~mSO=|JZ-q~SYqN%bSVV0KeYp6|KO9G1T5p^ZAFh5f%0wbpV1IfzkJupq
z2c<UHLIpI#ArOLQE99ZR6=SiwV_r?q&CeIFsEIlEualj0g*}+1(qRJwTCP%*%#_aB
zT25Wyc}xb|eL!z9A1+Vh*eG#4ZheI+?}Q5t`NNsw)kpF|f8&%%uODnv=9>rxC{o4O
zbneG)v&p*dgi6fMyOiA0GFbd)9vkh~PNqQCo15o(zqktGpoY7O_xF*S^F`n$7e$c}
zShQ|iFINThSN(1%K-Z|=olPDwolSM5a4?|R!Io@XVgP8C0tUYca>On`r=AlCfnWrI
zN9Rl42}Lo&S(cGHl@+e?F5stFVhO=c3U&REuB;5$5#Ht(d8xM3RPo~aMY=!IaA*{%
z2dZ`qcfatWCn-a%X27ohXZjz+;|&^e=@dy$I2jCj!ZP1kAfWaz`XecJ$xC_z6-1Hz
z(y{*46lg3%L;<_6gVZjK5YrQq?HhmO*aY=>`AU~G!oqN+C@&~~R8E+Xu=ZXQZWOm)
zI_?*0mk?yQ3a{gAp{N%^M?}Iet#Qd$rT0uf{(6O?dSk2${%?0hjcOlIqJl|O=%UGf
z%jxuGn+dl^1ELG{)oXazV(YD;B=$&}Rw!{q{TzBr1Uc#)eaUYiw!|3cb04nCWsiHy
zi6%!U1dC7GBMBxM?%ah!EiQTeIF1hl<U=jr%O}f6akfXW`hiVwL?YT~vBTwb<Q)q(
zb)8g^9N}?XAJ7gN01@G0CUlk~=UWiz3S+pc71}L??8;V<4!2j0rX@NJtjl&KX_^%v
z3cmIdz1ASamvdx^?>&b+vb3*HVdP+QeyKHc3Yo2n%(+=E(btcHn>?4EsXmMqH0@iH
zs)rhAy9}Vpv(34?H5V7_$Lskg-skf|5E1wlMu;2%MVMYMl<rIVOm#xRNe68carq+l
z7te1dFyw44B4V6-OJu)>R#8r{CWk{xo1CyZ=pcOj?4T}5+!(+jqAIH4USvnj^3<r&
zI3m68Tt3hw1C5ytFUgVJQQP7kt|R4)U<qmj9T8%q`#v;TD)~eFS9<~z8+UD^hDA1O
z?c@So0?+wdiRwvSwiif(#6E9=;Do~)-+6jdc0nNu5>ky?`I>k)csuWcZ8k=3f({NO
zA`*F`EetZ~oygPge#57i#e&b<Z|-R$J{2j$l<Fh#4e(3JvG8EKfmdY95x6|5ZkE#Q
zeHLE?y!hX|7P^VEHW^Ofy65-g^L(U{X<>GJ0x>-$&=A0sYZhgv+PgeQnRW=x6v~0<
zEGhuBYL#6D<P5zarxX`>BtGLSZ)~hQcOm)H-;!klU*dluxsEh~pKl{``8~Y{3=A%p
z0n%1j7c<ORTAR_nZ@6v=Pzluu#UuInIo8`{oZBFudr}w*kC38wgH>V@Q0;zBQ)kB4
z?{n*HUDeQ#So7Q5ikBMN&;;$^{i0J6uLi!)$vdKG`Zefv`;J`J?}+k=2zPG?Ykh!R
z6K6@ZEa1(MHRq6%Fv_nXpJ&T7l)en*;V0<fic2DiTYMAtJ3Dc{UasTn2GQVswtBsQ
zp-I~;lxjC2Rt1t`K_WB?X&;)3-?4xoq*G7O#Y2c+%p_F#ubBuh1yNL*Gw1FV8x<qb
z?!pix865~>*=!C9JooAmxEui62sEl$DDJFMkS+HI$oo;QMrbs;<++_XpQPqHaLyp$
zaIb+s9(VT3H+&>9Sv23Q@pjJz`TN>((y9s4IH~q+aUi>2<#_DzKK(Pnpi1dXzr2FN
zV>DyxU#@p16qFP6%dp;Hn`wbCT^L9UI@msQKlps?sZD@oA}9H4{s7r@dRb{=qqm>^
ziqIt_+w}H293pYg4KJ9T^#gg6y#`flAbf&E-{V&>#Js-O-|LLlrhp}&{l}Oh`+Fhr
zBvP>BhtPvTPYUOH(|6<w)hF~qbKp&Mg3iQ5s&s@&^L|E7egV1v2)lwvn6%PK$1?y=
zf_gHjqTWG(Y~$)me~&9J=*{WD@8<-qI;OlR&>w_^4hc=s6PfY$a}?P9=Bz{8V|HRk
zD%#%)=Cy!!yO{3C1d}nFR%I`atjw6P{*1O~kirp;gFf#fZP=d}0<O;s^0LYC^W%db
z+da(04+bw5^A0&2`s2$MAG6Q>j~6|8PoRxlJDql*;$|aK-kYaO-r6<d5R(C|QEG;N
zy2joc1X1xbjH^t5EcIi6T>T`!4gK9hiCZ^n8b<*7q3O5}GC%b_#Rc~F?+f~C;q)EL
z%GH(=dRS{P23DV;Tjbxq`@707pozg=F2eYmM$!cEzV$==4vtD%p_Z)~YjCEIg{OXo
zLhn1J?Yf4`c`c4Y348nK2bOil=k}ef{ljvjyX=grt1c8}^9!uG)$0{kDjn${d3<~u
z8T7U8x27!-p{_boNk=s&H{Yfq9dnj8lbndkD+h%ZE%o61P!YH;==%GiVFReOAAG6R
zsuUX?4)sI&>n@N3z0mXDS3+=T?~cX$q)Oaketq%Jd_d|2A#`077?Nj;ULsMxSK-=M
z71Gy8O&ejj(r&WgtfKfYp<|`mSx78maBo4g&_$pqrgED$j_GujGqD(vwC_6uJw&NF
z{t!wf%E4XVyRPgH@jtyG&|JZk%N4CSSCo@3ahFaO2^_NL(<@*ld<BZtOW%ZT*jy2I
zMy((kvlGf!yUWI12%sf?Q4<-7wUIze$CLvm#H2an;~bn3`+us6Ol}4yyy4|S;6kqt
zU5dr%n?5I8pHcT_!ybl2(uJ`*kYi=)tI-)E{j06~?ze@&7Kr-Sqf(6@;@`fg<H8V(
zsUyzwmwka3gd`lEynT>7ya?{%?;MOPnbe0aexW-MZf7b;?T~cT%xpFlg+vrESZ}+Y
zR%p8pN<){gA3R4%%k=2_+7w{<O&^B#fUW$WfTp0xH-iM$H!)ba<Q5gEh|+UDFr`Fc
z!eISBU<uAC4hjU|`@X)Ip(#<V^f7_aNFP%y_}ZKeikX~9uEQTJ+zNgr(+&WpI;-<&
zxh>AnWEba~#sqwvJQmsGWG+^T0fjGw;xdAN_C$S|b{c>U%!`icxf~JoC`3ZdyGrY3
z5(ga9incs|#49E&M8P@!H&7%NpZ#1H;0`f}RO)pn8U|0y5}L554C;tdZ@x|$4FDIz
z=v`e9EJO(3@v&q6JLY$Vz7T{HX4A{x`67hk&?1cJooCr+EpzQ{uMjU5=rV+gCwffV
znY;e(J@mz1Tt1c+!iETyU;7KTp;&Jca4cgRZISpj&wx-oFgj~gt4O$`W?ih{0vN|u
zn?>yoOpU%56{x@{uu0d4O*+r$v(3S6ss)+_KCpv?j4a$PgSN2webO+AzZLGi1Ko}V
z!L7Xo5U=+~?OT^#SAy>p#Yck5CFLiY3(y6#T&#&jLWdpC5RapT6xp(>wHTM|MZj|a
z{)pcIPln$1wogqdVmPc|+)b(n^JOyx(X*$|NqdrK2L)hH{^e3WK>4A-2x^@F7+Rzn
z&dp>{X=UI<UlI77!s?yPt>AvE?O>_X{o2*gdH7Bev)ZrkYuZW)4rz-kw49&i(G?N=
zr0d+)3Ia#OZ;P6U)?mQ;uQYBa#jYSPi4P14ityVo7~b0fddP9`fx6)!vZ%=TCGaI&
zFnauB+$DI{lVhQFEyz7@UN1np7y7ZNAi34nck?$H9VIjl<tr_&{lR2gEf@9|J4oJ%
zgvbJ}mU00qxCgBH<dr<yyKnzlWktk!$zzS0kGl8@qGJh{eCd=FFM&Af1NhEo!tn^G
z#lM63(ERO(%mnGfZ<sBi)|PN_9~O|s@xLeZ;v50q*RKx)Gth%&ef~ev-ZCt%ZRr|?
z#tFfb;I6@gy9Wsxf;$8V+PGWe5+t|;_mJSOjfLO@3GVKVJKUB1zB_xLbMJk=AK$;#
zz-rc-vu4ezF-FzIiP!Y%R|$AZt<f=7E)*a7XHwvC!+0UGQuv2FSK;i~cB>SHuoI;f
z*2p;^wmhJOVsx9@&mDxrQH9wDE?z7W14E%gtuyl6)#F&pO_>-6pwvYO_)lJrthNN9
zg*00rKl>o$ebsXt>@~TO{0oye;Kl89s(X1Il&MD&>)s~rM-Up`#>oJiOWhE9s~iFD
z-(FVi7|-4-5BiSGm7wqv)#6MZCmMSV4+21y;r95`f0Hk=AW$&rHWp(N4EEYMR?K<X
z;&qtOD)lqHh9VY@Ytu^S`i_Hz&Iz1@3Hsk-^*2C7gKZ5m{}T0~pPMRN#UmDZk$%~u
zGmH`at!A+Rfu=VQM)WIMvp0xZy~~to3J9ZER)m%zAlw$1MZ|oN!mtD5*>ylt?RrSk
zKP0Unvi}T|qsWEfjw1%)KRTw>+OJS)H#w&|?~Ki$b;)xn1iW$#{2M-_C`7dB2r}$A
z<T4ody4o<~L!-fkS33A4D-&q@Hxq_P<qvu2Ows*{Esu?(D17Lc*GGI*V?f3s;I>a;
zKwb~@$@S{K-LCkDB4h(yoDpG*NSVCGhyT>Q+!1(=vdD~`86QvW$}bye>(zv%hO+vh
zPY!wz1@bb++HOV}<Uq9(dL~ufd(?L8s^W;sg#K1Zg_8C?LMI|`R_A`&ZAuPvc}2i4
z$PRbQSM#B?S0>%CY=-h3;rT`8`8jbU`+wfZ4sm|m6H44;2qs7>7r8tm{k%v`PK8VZ
z5AT^T8xYt{_rEZ?m;<)GatQ?TP&3sk*gk&{m<>H(>CB+}HJ1`WGiJ$eqVL8GoC!)+
znf>F{pM(v-=Mb^6Ur$S*R@+HE{}9Sj3~jn50GzNxinsJnyI(k>AcOHYM%i;0R4Ulz
z_yj*<r1_7hPi2u$4(?;jh*j#d*x|_-_B|KKUf#{(utCsgcU^p^5BxPFj@Rh<#I^jE
zA-4s3jgCqwUP!kDZ*oGLz3u+^{dIacWjtWAUu6B1q<8kpuvcDKiI$;rOS|KK6@(a(
z4A|nom~635Tz?d_Fuacxx$Fq{yi1POKkhMpuRxn>VYBH2$uupG){4DI^MwO5s<aka
z;wUZJ*{amczbylx9;E?cOll?wmHIQg2|6=n1N%zx(${n@?6)ckIHlBT&KuKecG(&>
z&0X`~Dp4*M9CGR?34oZQH6C%p-J{zc7yl=EH~b3niA;3MOW6r*x$zb$+-bvfFhdeN
zr1$jGyJrdC_06+CXVYBs+>MD18n1S8NaXn3oF!qB3j6?oNf**kEOL#`h;YXlvt&M(
z=kF+1@uAQDIBSXD3sMXMlZ6*#bSvR?d~6KHwd3ePryb1xBu0g|JV6%j?L5u%zkknQ
z4zk0-$}4Y_pYKF=!&`~!WMOF7_$As9k7Uc1lEv%is1h`5uj~P@O!(bK6)A|!K+QU(
zCa=6z9)>pM8!s|j(0u40`cXGA`Qs>gGa0o`WmkjJKLpz`jnNareM^yjfvBqv%qR7k
zadElQ_}B^!MYm!euZ1A)I*vFDAH#C84Y20Pp4zC2JN{(SvgWXjz@baQr4q)(mX+#U
z7{c!iwhvhY<wV{FUPy@HgokxbzylkYVha1u#liZ=Px~(j#0O3M0?;@9pj-fQIKK#d
zGbq?-VKF-2hz;w-r*%^zT!xnb9E$(U^4H&9SWXt|UQm&a$q}r}<?oJ)Pw~z^lA+Th
zPt0Kb_$8ikT?5=I;@n=4ZZ^cRB?z~WBw=I9w<eE$bMdIsiU_}kZ64+JeY)oDf%`v!
zNUT^3YJ>Ds7#+N%yLdPL57T7h`njMvV?AX|&>?ULlaBsqi-82F9wqq0?5B^Gt=%*R
zAhfU-^SP1s^VHAGy5rMzWh;$3B(&>n8r2Gt8K@_GjkB<0PP@ISa6<W>IR^i2j1*!1
zx>68X&TpY|+ZXciG5B>I)WLJxiJgi;bK$=s?|+L>Xei+8HPN|nav{T{g4<Z#cv@47
zFB4Qcc)}RwO;7|E4P;Uop5&r<w@4G#&uE`D=@vs=DcBIvd^e<K3%wJs5wpqW=54Fi
z=J)K{?661<&*tg=!8^6^*mIHH^DB_>wA{Rs!s_grg#^owWg!ChLdMfOjhO_1ub$_C
zlJA<<3*oYXTs^%OSZFHN<hwk%Kxn*7_Pu8ehifR5$SvFrNN7K>j^$N#zUkDyzk(T+
z<EJck)wwPC%q>oU&@O?IfH@rDy?TBwg2bKlHxLuM{@c5`dt}3K<%96l8rI@T@X`%C
zQ4?5<Z@U)$*JA@aCI`H`zt+XLJd3=&m%f;KT`y|$?guG+I`V#W^~{5)&UFZb$x0`Z
z_%&!l&nym+B@c%m(d^@R$nlNwg>JomxLC^tx00gKKLUq>&p@1ptp+^}I+iznLQ!ES
z;%n)T9S{aVKuK~;E3m$#CbZAu1w+VSJnz||3(c3h=A5gvL5Rg8*8W1P5^+M57k(4P
zEtWV@>^%qHI%1wpOLvwxAfmwJjg@yeeD4rFetSNf8L=02IPpg9wvq4Kq0xT=4Hb5f
zubREkRsVVgk_3bC6XtmXCKZ1{APq?MM~ny)v%;C|vijqs<37V7KKT@e&92@v%?ldW
zk!l){J<43|941yufLL--S?bmtRLx;qEwHeM|5^*4W9kXG?jQ}!*)=EmS6~|8e~b;O
zqDcDD*Y|T3t=T4wk)lCNC%=QI8}Di`g+tP>LW!vnaXt|C38SGIrD7p)`RAyF&!Ht#
zZeP~EWmGJ55$lqJ!@+379K3p$WVCm2vz*LBwo;TiD>GYby@l_yJxb}|Rp5TqTA#bf
zW*@e$=kjv%4+1ZS4PVcTE=A-J>_wQs;nsB_u(HtyY_#sGrdm|ks6RaLa%gQ3{(fAs
zPCS1xIC6z_W8~YFry%ogIHcqJ?ABKd9EkngYPx6Jp9%*Q_hy)=jxavF80#G68D~|C
z&T8mb|4Ar$clsujXO0T9-I+4tm#VLs^WnTSEaxA@RZIxY9gX6n8#*dI3<8kUNp)`M
zL8*)q3L+V!TUq*9jC1sifp|gwUOHj2C_6*T?JU{Qi6^LRVL1kKOp@aLvYSGO_m`*g
z5Ff5pc+tw>AxhP5WSIQ%59&wPw2LB9VTxn}&&ukk#19uwI%5Met;G}n#3%9`@L{wj
zmVl8YTXqv*YmdrOeEJPn|2sUPq0A1(vg(h~O2>JJU!zC8%03|5AMR{;H+CH?^R`|N
z@5@Jka~Ysni~HM~_#>_=!R3Z@1fp1X1mVaVp?wO%QEebL?!7K>*y4~tr*5<FbrL*A
zy}Me&C1lTTA7|%W2*ATODF82tyv>pDpzq0*j{MwlFfoyk?dWy3eVBi$(~*6(;I>o}
zG#(j{tExm}euOaA%bDfX-kNg0Gp-B$r6qo`v}xJT@g@ut=CKl-?2O&OqE#UqK8f&I
z)lf$VYiEX&;yv$mOA?aVmU@!mbQn&T*}-hB_)jK{J7)R}XA>2zhWn=LO=9ZVqHKxA
zKlp!aj6aL}j~9f-{$9J73g^aNC1M;fd2veVAhFc1KTz`(K67#R>J!iNBI`!D?hue9
zH7DMFG&$KWNPOI{9vMR`)hJm{&4B!GBnrqTov(@)nQhyNH{G}d1$C>co!)O9`tT&-
z){6k>M8Z_q^{_9?tz|*=@SlwsaE_2^+z(^L2Gbx)e}AJ+R7M44Ji9tdZvsC1Y%x?w
z&Gh%N6uGRpINoC0Ll?yqxj>;h)UJdTgNHhoBEjMx^?Q>gB`EW1YHBVE=7y`2*j<CZ
zB1Tg>Eg!1&{@(2Y%~=LwlA<T&ezYjy1}Khjum$zL0$>NmwN0Qqh};zWZ9@+}xrk$Y
z_=Wy2yPv9Q&ZO(ahWI!HoeaBMfm#PsD_x;^FTVhZTjmbEV(M<t+6=?escP8-l$2~5
z^A}~qI3Lmrd)lN8q|68OX>9ZVPqrH1F=c&0uQ7~;q*H_gwmU9}j0+F^r_j9>MDk?}
z!k8trYp5br%s8U|F}2@b2~cAWCJ?(m#a0zlz&SLHq>Ei*NrRq|**O(C^g5*@{-t=3
zR{VVa&#L>k97ZevrveFh-{}zBmPrKh;4FU=;6Rvw7~L>qF!drThe}p6S+ml=w&lNW
z#%F`XBu=5$KQq61s>@&;L~wZ0s0a#|%C@26ro)*H(3X2uH6hceKGOfZ@9o?#1={dY
zsXEB-2L}Sj7Ynbh7#j8Xxs`1iU*!sjbeWri{+Wv3bS~tr0-OXIDk+em*EN8&T7s}p
zIag8h#mT~0KF&y>SCjQQlhKL(E5-RM*XfajBV)9Qd@c2R54Z1R?fu%t6s(Uj)-0_3
z8)OAyazI3pEFgu~`^t$%xS9_W;V-!&U|;@l3i@L-YWj0y(%%?Lv&*iY4HSNGkhZZl
zwWProhE$ffKl7p6{G%%XF2b1L!>HJ~Q4q0UWH|D$W!h{p&-$xQE`pV?(CY#Y%f>%U
z$p75M;0v--4V-c&$<eu<A`a<8PbTP6I$|Yvz}AqWo~{^J6F}?!Elr8o{`8TAKMBXW
zhB@Z<aN$P61)1!~>#J6Nmfdi4R{o?V6Z-AXR{YmkbyL8Barqm|_-&Q(K9C|}qjF}_
zDx_rjcRtr0{LCK@is<I+wn6&aoBuU^Qc}=2+t{k#=cPU=EaCW&lFWr|60qB@pe9&R
zBDJWHh-Bt1O|z1`-uxY$0AM2AidJD-eX+cUXOV1X-uI;c5Ce&IpyekqQqO2xk&UCW
zNDn>v0I&@yodz<hIYJs@HpXM2@-NEdJ_!DoOo!qE@$EoT^8VmjB^zHfZ4zVYB7VUq
zF&rF*S_$-qX3><)yu~U8s<E#je~m*JTp0vqg1bT|7@sjcP3)<u+Mg}?e=bavvh5a{
zs!VS$iu;3OL1YXRVNhrCx*0OYeS09EA}}Z3z@Hd({TDp+8|@a`M|)SSBN=@-M|c~s
zre$*dl6zH4Ad(JZ1S)td3$KiiuZoz3I!^_-2p>cWny%W4CS0|IoRr+N47hpIe_1W$
ztz@?3EvD)dMmQ7@DjN>q`=M)yBCbUBKR{~1)5kw0WacdY$;tmR*h+epM%TQ2ytnVl
z7T7N+w5ZUouo)61EmDR9Ga0;xQIUb{Ea4@n;rVKs?0>lohDOAj&$GORf+&<og?sGC
z$3fRA?A@jjv7KQIDjHws>}Gpu|6|G%TtVDKwel|!6RjrVBPRD~)YE6C*XdMtgYacu
zIU=$i{O`lTgBs{oXa8USE){W@Sb%v`5Sarh2=_{i3xd_6yd22h-P;Lt*3znpH2V(`
z6CnC4ep?^@u4J6S(lyUDGFon2gQw;_HiNpFmyvvoDIx6|)Pe2yyip_}+S`6rdCk`&
z!<<va{I3B8ydgFSP~;@7P_xFNt@ZX4NLS)2HV}vM<Q@FCP~Zx?cu7Q&XE2i_tJ75L
zW#JjH&_l$u-$^0LC(1;arh2al<W>j#7PrdN!W&S(*)ZNx`jhzrPo97a0{Gz9ZwDC8
zo#O~*BY4LD%eYo$vaMRp-?r<=T&ZIR#;Ls>kX>U%Fj2vcq=0OrCJ<3dM~|uGTf9aI
zE%}XXW^>^{xpAlD|J&I5sNqI1V}wq$#`Xzj<-z!JP7Tv4|9WQ#4c+K+x53*Yv#suV
z1a8}9G~rO$0-|%=Y9AZ*a5@#MAbdh?+8aK{t-r>O4$=MFN3_?^|5K-mCTBEpRF-gi
z_~;~2H)MxD8&)^=U#}Ap0w*NG#aHR}^WAe}mS?3?pMKrHmaRo*p&IOXPlK%sGBefH
zTTM#(G&J=U?{d4s3^3eq3Am~LlejpBA!EV+`Z|Cs1x{d!6L8t>zrT$6wEwl+a}ROc
z1B!<vjn5&n&HSG`78Qj`yS`hAMtkGFUbXB6So?)|^c4-GZd=@h<)rj(z4AVz%xd)|
zGxw*z23EtB%Gg5pulMCZgXyHE^ag(emg7Ip7@~#iuae`IrM6K*Pagl<QmVE=$=u%E
z(vDLLrhFMP5VTj+|FIVEGsOzh&htHHvH5-1{YmEQ<f8ZISP5k?U*goI5er3Y)Okyn
zsC-KT>{sgxuhp~OG%YSPDW&W)RL1RCs*~yfMqrZVSs=^?Om7Y<x1#^Ulvw{V+ATwO
z?Z_7*sRG)YgAWE8Wjc%iof^?xgzlbMtXp5^{9#EPSihRUqWTDm7<0LC0RQ2pfLF!+
zEOy;LnWz~y{8mPQ-u=yt5H3)(@s>LO$I7zN-7r-WARAXPL51mPn_$wq{imp#iV98~
zsqhvW`X|Bp0|WvW*c7qy*#G5*{*NEwfovb9@Xv_<d-VVRt|5BAZ+iECc@yNVECmx&
zDgGh1+uwrs|FXS5A3O0FoUWfyKRy@Hj_`}aR{b$isA!#0?|tp~V1_R~`12!Gagh37
zE+36Tr9>~~iPz`LWB1pEWjPIgysItgZ9QA->tQ$XfH38{&{(REj@{>m>-zmJeu>7n
zw24y9rJ0%ieVe|)!Px!%uN7GU6fY--cCYXI=y%lj^{73bE-GENc6ZW6_{ru-yc{0A
zd~n3GeDS~#rG+=ite+gn{g@d`7{{~<uq=++PU{A54h1de^DEztu*?ltyX<Lr9L^NI
zA4potS1THJeK&ocrFlm~?sJv%uI#P$?$pRy;K!>6p~rd7Dyt3o9nX_KCBVE`;$lP}
zRIDby(7NnSxOAGff*ksj4SiX4eQy<&KK!Km)gz;~*bpNcNB*4cS8tu;MnBG$WGA>S
zl~RpPIgV)3T6V$nT<FyI?q}#LhgCiw-<yKt!5?D4+ES#TUTgSj*sioWl3v;_^NdM@
zh7gDZXed+)9Ho8y%nmCDUo9%u12rFw)+~$7{U><-(C<;{>%lJ!l7zin--Ke43hAC5
zHXrqEW%*G?5phpkPGyGUGTV<A4q3Gu#2Na|FzYu-UEUcV^)r>X&^>v<8b>DFG}qzj
zv)FQdvJy&eN$0YMoXHQvxZJO1`w{VsX+1&hBvbjsELt48J6URWvCqGIU7F^b^ilki
z>WfOROS0?BcJlZtvjO?j&7r)0zblxr$HVmtvze-mu~skMk(RkS+hiFs4+V{HIxLo!
z!b0Qhv4}RHE6j(Rz49smD4NWQE4e|~o&Kpu4;VEBga^fb?I28!-|(}Mkn>e8F*}=K
zTZ=_v=|pQNhY<0#FnMFw!W=^v(_n&rtQqqFlWr_BJJ!GajF>m-{zA2;@BPLLe#doL
z6{PX@G&)1Y+gA*O;I&MzqbGn!a{OSfUNh<LDiCvcxX(4$ZGW0FoyRVgRlixlSZk)%
zIvMkFPoWks&&gybj@(5BD5E@Ju<$wGRj=|}@?}(`yV3o*(CoHXM<TGk;IP?)N!o!b
z>UmZ?j(pkwvAk#i;H2g|GUq10s3%*Ok+N*>&(%5=O}qmE6JjOndBc&);nOq$9u-5z
zAX!i4U4#&1_bfAiH7!!2yex4{eE9{T%0ean;%M##-(xg#mY-{`%igcuHO*Qbb-|S1
z6cHJs{lZMMMa=$8vyA<{PI;Rsli$OPoCj=Fyzy%F%|`N2dIo!YVn`H`Trm2xyIjKm
z0X0=XXe0v8vJ@YKRhk%V4rX1d)E1T^K5uK226a7o&OGM8ZRNhS*UagA+{1N!vqyMn
z)Ors^xU^n5-=A?Z8_Vy@w48?k+o;M!#sL94)K~ng0fBDis>RaMY9I<$|K?D~h=aEF
zOqFq0NNiyJ?In5al%bzw_{M4GfJCjskJ##s{CFk&u<;nuH_1sXy4&ka8eiY#Zl{h(
zpfwSfc!jK1ko(=KUX*MJg;7%A3)?UIgE@iCjezE3ZzGZ~xAiaBP_@~>^A5LMo0(M~
z<MDdPM+Oa0Zwz^ItJftXrPgwr*La~~mGh3{`RB~CxnQTg$NP2^!~l}~r)IU*D4EBS
z1;+n_28r!ZVEY`o>@0V-{z({Ml-Vd%yteO1X>QM;90H@e6J^x4vvukSYi8b#v~KJI
zF0Kc&jJFe#9s@}%1>PwTW>j3q1TlCUlk-9jMD&X1-*n_!n-()c%c@kk>4Fx~N+DHg
zJhso}lETLC3KuW@xov+be0g_F_QCM_6g{Kxn}7wn_37p?j{9+iZ4A*@@2i}4L2+zb
z=yyhC^ely;z3kT0GB4*|u2t7BYfmbrY0O*L@A89Qz#i;}8@KwhJBp;^xDPEgUV=#~
zW_K0S(d9WFcIef7_JWr49P++b+5ZIX%$G#gSbY=f3wC{<D_|XicGbB%1zy&^SnY{S
zNHGt$AOQ^|F)_8@E~vL1uTZ&4PFH)x@6yas^Uv@HbZoq~61AEtBdt;+T^b;rtKY*s
zi%2{Xf-^G8t!oOo)^EQ!TqqTo1!aI&X3bYA>%xB#hxJ9$@_`qcr0ZWyeR>^5Y+V_h
zK^J#^e|uS4b%!Lyfbi0%-IVZ4G^t>nH5+@*Si{;f4l<bG`+&&p8D1*qtr!1wg4+9T
z>E!Oxw6K1pTY56DiJb-)3#+-Z3?jrGhJ!h`SX(QK^_~}*PV{~C_o0G^%rNd|iY4N^
z!)3v0r!7pM==`KU;%+9*vKHU9Uq>u4t0JNm?Wx+3rJ*nCr8f^`8-as>ZUF)U>i%9K
zmvHz4i}cMBWE1p^Oqvek{o`rY;~4+LRj*(mvSl6vopIc7+Ij2A7ajww&LzLc*0)Lv
zl09?3%JfT1VR$NCrbrlIm#Gpp-SW`e&@C#?*d3)OF-+{Jq_M~wo*p%lM5wsvR-4*i
zB#@4HUp{%u8iWPtE5hDyJS>t7oxi;9Kq2p2?^ARBl$r|J!9T{huW#QF&N>6YrD2g<
z<_u(s4_(}Gt|{=#MOnT|oL)CO^i+KsG?tsg8RdIBlch!8=abGG5c^Va>v_UuGr8}T
zYNR-8!?JfG>tJA*ym7zv?8Mv~s_B-~VXxhktH`%EXFCNgRIX>Xn#HK|d4Lq|b)+ff
z7w~jN4wu?U>-pZ)sSl=h+a0~f?Sb35$I}&++}XS9C-3JQ(E1U~OdHpt=OVtQdQC*q
zq$g%y@or4wY;~j2f5wh0GVCJryJI}?=wxcM951{{aoZj>h#VvHV!m_R?0mu&D;|jK
z{_F)0#%<8vGepuWAE|r5r9QnDx;z^n`Lfk+D{tiU$#W*hYg!<JspIy@ueiM)u$lNA
z$q|+^^9qCCQ76M&&~ZBg+t`9!TYu8mK5TCz%?#p<y$$z6mc!HfL(KVo&5{(V+R`nT
zv{cXfqrNy&F9hB~dM;+_$x_6z_wP4<^eaM?O05#r@}={-g@-Dz#y5)wf)QTIsZVN{
z%0wHJgL(^E(l)czN`(rXZjMQ=EhQ<+Y*O)e#^~6F*JE76Z+?OM5?tEN<@1`%^0(wF
zy(*mQ&45Nq+6B=g-@m0+S?VUu`&z1(yrTp!E$jfe4B)9l;E-tx6;hc~bUFsU6hk69
ztk@e3AK}|FPz|o*DEBS6VUHK-;O)Xp?)PDzeQ#1J)1x{#rwft|bLQhby_8V3#ln6p
zw%#(-S1bsr5yaqTerkmDs-tY|A6e{t6GuP^me<K{cuM?h&zs)m?C`V1*7%Lfo^x8#
zH;@|iSus_-+*43GX8+(v=n-eGLW-PuQpXA;TDM!@vCr(LTD^7SY;at;Kitaybl19U
zat7dO3p4G^wy(R+u5J6}yJ-v0{LkNR$pwZ)O&Fc@yfQ?yp}+;AAjUBwlW@IKt&%t!
z&A|L>i=nYq5|(`SPa<zHqB|+*j-3uJ0bT5k0sUtazpby*EdkLC3W8OuMH`tO_D=)E
zy1=4?mt&mO=r&VrtwjDuNk@XW?U##_1}b+a9>G~oiz<`HJkv}s-xIWvll#Kj#PZj8
z!c5EB%xexSTzao<l?g;y?<!SN$$i(C+9Ytxt2hXbyG<W2BG3t4_sH!u?J|{7Qw!NY
zEr*H|aK@^pRH3K8Ft8$e2^j(K!kQl_CK-)R=6fdOcfFG0RDud9h=KLtYE2tQeEM}Z
zB_tYP!{^=Tr>!07Wh0`uHQ?G0;7~l}xHg|#Bt?sN^IIqaYliNJghdO1%7XeKpSz+B
zP*62xDKHUzZ!Cj9lHP`}Ue7y){?N7vZqsD_fCpOL)E5veRI-nx;Ha=g-j{ru|GefQ
zcQ~~E#+V?eS2%rX+2QLfQXAm0#(KoD*UVBxK(4Fdi{rlT8xRyEZyT@8)Ggs)9E%K>
z5F_lx0SX;|qaq6PmiuR~w|C24m`H>%44ww_U&amxiBp&i1{n+LhC`xO^n7<qde1la
z%G*m8vkmD03aW_AU3ytx{_ZKK<(<?=0acEN+w-ZVv*V6?_#kxm9HZXo3@dr4T$4Hg
z-8-o45%Jp5U}r$Pq{L)#@DCax9j*i?o`mj!h78tAZTiI@M)<sW+wL6Gyn$e0#C`v?
zSXI+qi8zpFWgc}SqpNz93Nr{5>7e_V0lmxZWKy%&(*Pxn7fi=H5A%W$@VpI_T^-E{
z3^>9=ylBwq5bS5EpQiz^@pI-v@RHhT@MCUf`U+tIg)3}XJeA!zj@f_&_OQc>Ao1yE
zUir^)s>U@;GNJry8Kz~oWV|EW(-t$sqg8$Sfxkk?Rl3*)>iHhULe$;%t(p6(_qLhD
zCH11B%$k@bao7P7a!X{1$@GYJpKqL5+5Fl?+I4IF+I&ZGi0?LLcVEF0(9>q!M9Gr<
zTA$f3{HoK|_Ha;OF-*gzu~|AO$GsSsF1Pl8Ks#(ri^!6&+R+<w$;O&<=RG{Q-Quzj
zF53lOs#%aGy5ZlR-e|<lv6g*7ev}f)a8h`PZ+=vNS@5G<W>4VSoJE*5X8n-dp#SF1
z@~A!_1H)J&I)Cd&*sWVQ&KuKqI-$3{PLG}V&s5JoO7EPEUA3gIs?8?HS9@-@CK9Sr
z{w`%y*+sRPZBTu<on=z$r4c|+E9FIE$8J_@-H58U<oLm<j>ZC8_Si8??mhpumMa1m
z`JC1HOAt!XxRM}sp$Sm6!w$YC=c_QB3Wm2v#T<p7Y>>k%dZ6vse_0q0d*wcaY9V@8
zp+zo((O-A8d{D)!M`&<NVDmlwi0`2y_{sAh<{R>ONohK-S=&wW$!PT2MJRH`uDa3I
zDq%GStzLV)wK(CtlNU!zMdLY6pb%bvc4@T5&BMvkdQ8ct+oM()FVlvj2hlzDyRHO=
z@wPh&)~5BhJG1BS>E{hNNaM`I`Ks-b8szMM9SovX`L6#kwtyouA;wsjP&sxw<N_^B
zz}bf41;$*jN=J6qk*QS@fJt=|k`W1zU$$jFsH-djc@biwUma0{%^Xid-m8gh8x`ZS
zr=K;;AGQjztw+&VbUqxhUx}nxkNO$-s)j%%!*DsIyuNCb>(E}2ykiZs#iBMFNOGv9
zyE7-;*r3vp+<G|+Zd<yF=6PpApr9Cd-^IOJLQfW-4sKM`VHn~2l+u_P^oE$F)PCc-
z=Y_hk@A(UtE0+4j=kE$|-Yz#CnthwANd!Q({uf9`&wZ)BdxwAXclQI2LLw}4VT~(=
z>4_?(QR-f2rImc5>{`dbHYjb#?*wGd_z`fMnY=t+7bN_rgVT#l=~WyIx7&F!J43gJ
z#B2NnW@AY~8#_f<&*yJ>z>9imT$UvD`Zw2Gi6;wg)d6}7?Z@ABm3-*4=G8qf6vpRX
z&sPG;`4Ys~-G_kNDnDKnM+2QYAIhaqAXM{|Y{1CxX51$l<+;02iom&A<9^kfeBF9A
zXYY_SF`vPd+nS=N$IC|`BXrI`AGs$Evn^+5deix>vl)?@<y+`;J<}T;&P4RZAEM_I
zBC7R-fw!C%zI*Gs?IO#G?-QFWdfy<{tvvNGP3bf~+qNDmG*}e~J{8iDB&M+Mxa`(?
z<HBa3sOWSx2iC0_MZ1*3ju)5WdH+LOn&(-JrE(R^@%-s(B-<V%&!{Ubh|e6zYG+Fk
zgD{kZ&%e$C#R9vjb)HhqC`YJup+TAk9*R#GUzUh4i8imt-CZs$d7s)sg78x?&kK?b
zer_S`N@i(CKBxcCvDD<UckaX-WOVhESonrW`Dc`&;NKZj3sCURK=)vW6KsjxjsIrK
z{Ks1T)(h4YF6qj_!G5*Z={FaQ%E-E|NG<Ou_D6pT!FRLGRlk=mOAZ;DLs(>p-K;pN
zG3u<;*Y>Ed5PsLfbqJgU^*Nc$lByExL0jhHU@(rh6rumHZfL-&90wL+jEL5ZwO%_?
zI+!O9h0W~ymzu7pwy!pL%JwOy`dzk0gvn`{Z9HAwbt*-bqNrFS<Ui@V?0tVl@klIO
z9<hS-{@ck1oMGsXx|l+8b(dMnhrOuXeECbvje9(H0ME~s!i6q>Wyysue|UVjPqdzC
z*@Iuy@d2Qe-3}3q;t8;d;V|FYSKJN+@cU^n)HE$jF`eg2=;{yTSJV)=82Ir?fW_yu
z7{vBwx9!k(TUwJcX8%TkV8sAO>Tgbjfq`!-A6VXY^Y+(96RaoQjvXU*4%~gT*!m`P
zBtua5*H`UtQ**V}Z0-Y71f2)Gg&f<~-ITI{_3jkTSFaD^Ihrx?%##RcLS;c018nZi
zSK4?$Amn{fvW(z0;W&gA13DD%7ufxl5HH|*;)3_}2_yXDbB1Gd3=_04hL5%y<3`Zk
ziEpL>xjJP!wQtRR(x)KdgM4*{-VfR;{6$~t$82+u7m))>=Q@c5$_>eH)2MU#G(Eh;
z^f&qwogx<NY)dcJ{Y%V5G{*(G-i;?gOT&^Ig?qBsoB(*}jEe#L(Z~}ujG)PR=S%gH
z{o2rVgmz9_D7Z~TBlPif#Scjz>y$Q{9WHPmTQz(9;1#XVZt<|T>4pE*Wq+PRz65$$
zjZttaZf8x0fr%&Pt(OHLdJIH)fqG!qdf<vTXvTpZc7C~OK&uqk>|zS5sdOD8;?BC4
z@;UI^*hpS)6BW*0f@2;&UE3=!5?rEISTQYheRZ%2@ZSKTV_$BAJf$kI2)9@)HBCT}
z_bggWko)OGt2_+}xGQ@z;+ZIlx+$1f<x-u!WyQ8o1M09?57t&8^UBMXh)^%pxJ5sC
za{AiOI0zr8)2dr5LAVX`>O~>HD_UHLWFw=zzd4`WOTo=?+ENm~Tq!AG6c_|I;bv4K
z-gP<A73M+!c%W;BdQCjnxC_d^E+y3ff~?bLA&NRt>;b^RY2s+Lg|f>bBI4+FslO(|
z++Yp2s}{{3Ou|?S-*&bpiK(nX=z|V1ssM#hryGI+69B9+rZfd_D#HaN+0KD{u<92*
zt8tgM!X35-Q%i*tp?Fx`{74xFUF{dsz00(Lz^{98R51uf4Tpu@WL4S`hbcePzji(?
z5?x^0oF%$_UMck#Yz+X`06&q$mB%pPp1VC;*rMm5WVSO~_qpYK2L}>MrFt+&KEBZy
zb0S1Iec*jM-gRPb#N36M#3H|M->U|})RPjEI^NB?6MlzihFQ&F_KV{{5!4rQUXIUo
z|2pNXzzyeyOJ^7>w*MM77V))TGcC2U->P=7?mR3jJaoO)5`Kgzlh6@Cu%4Cvkyjvl
zkW_G^;;!Pxy?8y-A!T(f6${hzdn1CfrDmoCIxoImyci<|tY<4r5FzCr8AD}}-~@SO
z=#ckq;Qf@ISB9@H5mv}}+m(cpNMpKd>4!0QmSbpm$EAEC)2(ZY$A+ZLQ{)ZFR{iBf
zEMo#a36w|{#N#)Su!1+q+uKCD-;#jc(FzBlP&uq17IfECgtT~~Q__OskA_UOH(Acc
zZ9RcuwTRRubER~^anW8bU_cS@YQqkS5JdP+K()~{^xz%%BDI9aF@EP}s+6bUsfXf&
zd}50q?@`049i#PA`}T|70j11DtB3?eV%}|t$!9jRfsOWO82~K8tkWR&_G>@o$L)!N
zNKD8RZ;Vw*qktXS{zR4{0pq%U8FF8v4Q!9$E>dA+4`ZL(b4_r#i`y7@sZ9}0sO{Gr
z%o$#{-3~as>hwAH0?K}9PvrZ^v^vnQtqHNvl0{j6ziq7t=Uw~-TrW5>a1}9RfN1i%
z)%f1CsMdDYvzgBYcfIVl9A<Z_w7mX%%&iuivgn#ZQM`MeKq1UVbpaRLFu*X#(-NNI
zLr^TFxV*;WZb!hsEAO%Y({HDkP`+Y~KHq*T*gnT>K*RC=oc9ZFR_5r|R794qV-Wt~
z+41!EbDNdU>!xk-T_S68Nkoa*S4N{0KDPQz>oNx`JqN?{3#?8)((biZpIZAiSgC*3
zQ=SE?m+AG?ml#Egw*iQ&l=p&+@v<8r)Gm2j@C7`RL&e^TgpAudc=&@YPjCKOF@7zP
zcdP=49iT|#NR!}El}z6ghlQ^$)>E<vnXOc|CWGU~R)yv5X@vu%t}FH~`a0Wp58W-V
zo&$|)ma(sndjaTU5e__-Uvii4EBS&|kEAvh1okoA4g6g>Zq{psG=hnZE#uX?(EjNE
za1U(_NG9cXe3z)r`elni4RaPT$070ML%T7ZZ3AsTzw>s5uO!*6wn`uc)VGw}V2Dl$
zV_tOa()nav+4(0>=ch74P4A^~dN$s%YC#jA83ODOFM%_C@Ja2%kG7tR7Q<NsPR-&2
z*hNF7ZBo7T*LKBZ=~VG4GqquhU^Y<sx9f3)NIG9-_T>I9;1q9*_p9|3v>L4tn?Tgp
zWEA|6F9&ZfV0<V#A1`{nKEA8RNMFC;>J-}x!k8c)3D}zgkMgq%i)9FO+!p~t<jm+&
zom!o4HREa9u&tXKOT!%z1aHk{@t$e6cErpK*P}&h0t()cEq8YTG>fZY1J9iv0m5j~
zGG3UTQ8!aq3YTR>88I0Z>?`>j;0*0hPaI0v=P*mUvc}CX-s6fBnj`ye3ofBXO=*US
z#l&P>o|lJ-BYCn7t}9*`9WlqNIIKR#s}5O_B{WdxS!?8z)kPwq+#&1Pk2~2$utnoY
z25e-Uen5!?=!9Hmt4;DHBBF?SORN<a-H2_2zNs0AIJ9zaKA%p3buzzT*4lQj<U8)h
zbX*DUI;j-r3vzf=QM>D)KqeBo3n(s?_9HGVi|7RxG;>_(i?}lDXVEh<x?)8MnmyR+
zU9MEtFbr?q)pGttW0}06@H=v`8L+SvK`a#3RR$22Q4o6k`WGhzpj7aseB{#uO*x&i
zkyzn=FGL{;g0318_8>~kUkwI?JGns!6%qZBlEw<ck9{bOYW~@u<-fFydm-*JD>NE-
zxE(&JSz;Py!vJBq!=YKsxrVBK2x9U-{!M>jhlsuY?qH#G`O9|phsRdNQF<Og6F44h
zFAWT^wPue)IgyZcqdF#e^z1Lx0;h|d_X-zp9^N%0@E)`1)W$g_)ISl0fCQemse=X9
zY%Ms8XWu<#HSvrf6ZR~wH`)W_0=Nh-0ed5F71><8b!h_v`u=FW@+a<VcZ1c#m_#aA
zFDcwiR|25R<~llTmZKHMWw9ooFoYlPijg;3`gx2Aioc^=9LtdRgEn$NYOes%op~JJ
z8)v=zqAeFhJ{&6~tR~+&t#CLBME0b#TMD85fF~O@<0GRTob9PO(ikv}pWnQ&;@`1n
z^j%iB#jD+(t7S+O?b^3vZrtofakoU-d{sz=de#Z}syU(DaGXn&{`#gEmrW}mV001a
zd4KbC9iVA`tpG*WEo9pw>S%;e)H;u9NBEloHc!%3bWzD(qG3{nM58(5$>aUZRS+ru
zA-lUE6y;Y{qTB6(W2EDKwQ`ZbWMbvjnwOEyST~xvIvX1~|H}`$;9@VD`W3<Ok>L!%
z&kWx6Lt+mdlJMemo8D2i#_U8}WEfM~l)FSL7BGLvj_&u8BWy%L*eZ6E%*T^2YxPA$
zh!1)m0v*nD(w5t2oLkjXmt?jj#^GP=N!jo2XD!`mWH{R^3_U7QLt!h&y$|0an|3yJ
z5tnT1#j!(_wyP)38{M*?=g5|1zS~Eh{!ar^zx4;A3SBRI8c4gWJ}eoMJHMa~O<+2+
z>>rOwRYE|cfOhh~4|;#7aFrq6Xq`x56m`?u=>pgO48RPMqca3{3It=8couWJoVg4(
zF(2=r@_N!T*=wo#f+-MfN*7Z?RGz5u7CeiO1uccgh@hkHj%FV2@?uD!15yF8JG8+J
z;s(v{P-97#auV#f%wq_xzMaAC_bm2`?Kitj=M&GCd@fsUrm+dcz6;YK?~VwaI=hsU
zZX*N?oB_}*>xWnjgp06V|9u3k`J=`>f>FH+qfZa|;vkIdf`W5Y+$R(yhu4OljIeq8
z$9vQD>XEis+BZqN9TEzb9FFT7vV{i#QKANxd(2)^;awyYaQ;w^j4<~F9%^7^Us@&i
zCE0c6!xliXAt7vCW%zp>lt-&PK4;a-55{QdZ30kPg^Fq^VvZkAc(1;*GlW~x0ywX2
z0F}Y2Zw};y`NLpXcx4p4&5`%><-dCY+=);R0n2TTQ64hgIlw+iaKWD-p5o7Y9n}#J
ze*%r~Xxz#)4LBe3RT6_6Tj3SN?eZ`mu2tFTm<MXre?!ywWc>ENYbp9(w#ze!IPT44
zy>KSEDkeVEWF6onVYsJkBzr8Coc-N(OtM)6VUqVR^p+p>qPe6kVOD%THyHW_*TZQy
z9O?*zL!@6%o-m$di872RpqWJLULG!FRDAr1aSiK=IrlK(<UW7ldYzZ1F+S}wrJt@W
zWxP5%>@>`;doFsX>7^s4t6TCAWgp_Sk`6ZCJ!)#qT)wKanUQFPzLH(=s>7RGo?KpV
zd))F`=(=1ce{|QfXxu3z&bJ(=b9J5hA{4HadLkE#LWBX!@_QJ~b~YfsdholLNsdlI
zJ=mWp$umMvK3r(_FgwPO1cnz=*`mSDldzi)`|a`#VaW>7p$jRy3A@_xQ{KD9uitI>
z!3I!Hnz9&xPWh32RIe`wG*6jyAZ5n(Te;i;Ai?dU>WuH!q@Z-=1dF6ZCQbf!(ksl4
z)A7-qEhG<Fao@WTpT~QfgQr_n1FU_T>uKE9NSUs%fS*eb7k<~P$!w2uBKK#Uy$Cqb
zE-_>awJ=&k6HPeZ<U+;t!@I*(zBLlaRhfEeQrX1(Z%zxcP7ymbkU?^w*rY+kh6g}j
zRM_&<o3tq~bWSq>Mc8*Vsl_Zum6>pUQDNG9YG;5&u6`OvKK#ohkkuB1RVf=-Lc{(P
z>8p9|JIlrFj|NM%hKlgHLaGGSr_+03p++F`1hXLg-2&$8wI5ajw~}CzSnzWcvXC3+
zmu$RG%dH|<CLB9PJU~zJoo5IWk<RI=Yb|v&Fv-xK8@?~?s(z?n-Vuo%9@w`Ohvu5C
z-R&sZCA;0V_8SoZ3$l;8LP@xVXm7lz{0JdOZv3swb0Z4UKwa3n(Gx~kp(t|YF-&A{
z#)5<+sIOV|9KSUmx$o#^P529S{wFloq<Fe$O@tc{a3uvOaO;X$k6n2v%Om;Sx93;)
zVtOG&{@ekP0EgR`J8y}FjS{FCi1l!pNun0GmqG^}sI#5h#Q5x{VfSFoREdg#we2oB
z(C~2CPh4KX*dHYz@^jj5^7^CC`KuNri^U+$0jtSUl*Uvpz0#8UEc9L96B*jC9W_XN
zqMlpm6fLuB&d|d37m9!p$*_xI>fUvk2pd_rrF59~UvJ&EPeMA5AKPx^+pKhfvwE8!
z_+E$lU1se5NE0<|x|EHFM?l$O$M`P({*^t<Xp#eHpA8j0a-<!UL*yi*Y?(coOEPq^
z?^VhukPOB2yl*~gn_N<LKSiZ&b4ad#-FO^YtGO{9c%|zo*f6?6Vy#>6`c3UU%~W<x
zP3Eu>uRv8U_vc%7Jweze(8KH)XSQiB7*aT{EILlQZ?onF&ql}|7CWCYp(J75(-IsU
z%|^mTVy!n0Ggkp9R!L8Xiiq%?my4ONrZ$+*HHwmR7G&2BQMniG%}Y*s=g(fJxUSOD
z?sjqa*4ge}Zbel-Chmb-<YsHo4jYix9M(im?!NgGAY?I>x4m{DaI1?v(=Y#ud~tJ5
zzT7Q24FU_Y3zE!Tv3?(mxbZS=`Ps1bGG;?0nCVGxFY3Z;J!gkl0yaH^?|}XX|1<R0
zx4F+9ab^0Gt&RsRUK`Jtd@l$Xo=yZ?QW~933cm=#tXsLky1cp?A_Bl2Qly;pYkf}3
zPnhIx4dGcFYsBil%T3mu?9)+L$c7v#W*8tYwP6%;Z+tHGaJfKQWC4JZmJ^gMFV}07
z$Fetu?QEulQE>^U%}nk0=D-LbRkg4gLxo<mCp_5T3#|4HyTBi-8Ez*_OQ!x9<SiLd
zyw?X_JZ^-a?6%h5Gl`!S30zCRc#1VE+KjO!vWWR^4+RrZsvgLKpHs`T%YD>j#6bV`
zHT9A>(#sZ=hFxDjnbl*3E~}M<dl#Ly4&(5)YRv)b?BSMNK%Af@2#z<wqqVlp-N8tO
zv=^zD1Vl7<*tli41aZv|ltKUOXBU`_ickO?__2HEDKE3+&}+LZ!nK^S@l%F;-1u_g
zg}XszmfyVDvtAw{M|VQx0gfw704j@gHUza<5MX1{@*Tf?5)cTajC;kvVZn&#f(dTY
z1VI_c>=@`>6x)_FRlz!^yco}y+?Is<o#+rpAr|P56Oa<NEU!bPd;Qw+)YB(h(}|kV
z=(vo_ddQMatyQPSq@NW&-wMK)GqAP5s_?)k5}zShQl`EkmVB9BhdtT1PBnO$)G)|p
z=C~5^da^JB$rNW;zXV~o6MH>2&AZHp`&*>&1B!S4#&G0Qc&2qm13}Ns!+lw)gLZ%B
zrH|LIYHrS0WR*yxg()jO{aA2SAbmBQ(VS2pzjZ>&dlzl8i>G%4Zrq?dDCtQ+DEH*U
z8s%Z@0%!9p&mUudYz2DEWgC%LqjGAL8{{FGihufrJRDrV<U56q!*b_8jX|zAMe&gR
zThJ|5h4Vx7ZJLc&Z1w^7D$00Efi>r#6(l(Y^L?LFZ_k?q)n$oq9}QV#lj$E{xHnV=
zAlbHGDL1=}uK5-3xc|KB!n;j?qpC2h*Vj`=&ct=ENN;1;C-d0zS6cc_IILczxjOW;
zMHA0X6>ToaigB@f8_{xe_Ze%qu~vw-YfE6G%6%8AAF;f|9Oyz6l9_hU%v6(}H^HIL
z(`mF9qYcd>H5he4)%v(#Z~Uw_Lb<}&D)AXH=vKYpZA1%SVbZP^hElbsQ{nv0l>xo|
z#EkkX&t$k7nkJy?iVJ`a>qDZ^xay4aLU3vuQ$4-lB+xOmayW<Xzhlpjb%kV#Dlm=<
zqA!9BtIcZ_NJN!HQ`T(P1&?k_M=8jWE2F|pU^_+DY`6qxhV6E(X=5us2Tiv_yoSld
z;3nE1!~vG<Quwece_pc{{<Rky9m|0KE$YE{zikO6@;>)8J(S_r!s=S~>k5I~9~Cb8
zLyABcVb6f>tBEK6%jT9BBds+{L6W_vUPrBY@zB7~<N4U5UZ}q8$NF?`>#{d3%bhqk
zEn>aV2aTQYlJf-ojLNcJlGZs}X4=Ddrk5u?acilyv*~E@sN%S^eT`?%_EUVvZ2Kcu
z@09cG@^o5-RU?R0A;ze)rJpgH-8o4SI6j;W!PFfOl93}M%VjqvRX#HGv{bjeEbR-;
zNxe5a_mi)+)^ImdhP-<)0T&53**z)Mhw4@%DaWUsZu#az0ZU$KH_J2m#Zve=vCBm0
zKc~*h+uuq=h;1G7_vgK#LFRzRxD62BR2EGonz){k3TI@`1}i(^$u;(5AXB014Iq~?
zssLo%s)s(8M8NojM+~HOchZ`nzRlXKe?9!>a`hXpL9*b)5yoQdtBZ|fLq00>u&O3E
zwan;ucbh}$6S_%;Xl4B1_Un*&$361(pyw<KJWI4v{99N9c&<Y!$F=dU6*t1KK-`EX
z{kd9c{4uKLs@~wb(hj9*H2Rv+f!v4C4w*Fvi(-?*3l;T=17jZ=l{8eNQN$4O@2?w2
z&@C+&YWYA^A$Q=<DC3Z@RjG3xF6265UOQ$lAo-0EB`ZfMNoLhATr^oKMqyhKcm&*!
zM!>#pKW9&MQ>@@oi~iJ(o8QM94t73w7{vyBN#<~5hcm{$Ql^l4`_IF3U91=!_tQFy
z5we$tJ{H;5Bbluyze#|m<691!X%*Uj>u)GuG`?z;2+1gnffr1%(H7BLkeexUp}M=j
zNmJ+?JTkq-(2YTRRP+Up>ZKI|$^}-)dSx(#l{(iS1>OA!N<0rsf;hrbdBJV2sWCo`
zUkmc{t$xB?t)-Z#4856Oq0fiHv0~Utc`byj-2oc_5|Sy=Q-T(5XkfT4pXPpje-|1H
zyR40Fo#!5^fgsQ!0L1>mavxw<6GR`cAF|<{K9d`a@Y2O3hk+ajoO@9aa@kQ0T9Ju|
z+!W~LWBsn$mI*v&Ot?wuz5)>ISUngzK!B)GL8nNxXO{m6#oQtVdiobC@t~F&H6ZuQ
z+KUjg;kah_o-ga*M`a^)lAvY+KkqC>$XuH$5y4!03a8d0wRX@^ZcD)Tc;sj3hCIU7
z<{or`>kqjD!cMCkoHAkqR}~Kruj>u}dKA<xnU4)BhVAV7Ps%KR04-wiONugC9KJrW
zhDkv7ME3#BRzyPsAjNPQ>!c3DdFxLuWYIkU?a-p>C7)7KP{u3q0c0p!4J-9X5>#i$
z%U-;8^N6IOmMk(+M1+=~+IB`?oGd-I<*ZTK5sTbO_zZpLCQ)PU81Ej=@bC$86$cO+
zdntTAEG5x<uKU(Iz=F@u1g{Q!t~WHz22(^|QtB>prsyM2{TAv{sG<QaPD82Jo~uIR
zyVvTc80-~c@8LNRg77QqMX|!*=E+#m^~CWBFG>58Mm9;t`H3c!)UiS!^+h~3eF@aP
z2@>Ljh3R}HW)v-UQSa#Q2`#M*R;xU^sEF&$`$vcoA{=xZ;2Z)<OiwRbcqe_eGFF17
z$`k1+MiXBjA9d}>bYHZ=#SkPi75)HmNBww7%K)n)pF#{mU0Dq|d+mjAdze^fTIx7f
zZY^fN`V1?EVk3cIa&tYAHLap_^4e+X2_#e_Ou|>oN3W@tjr<*=6wUX3hME^oqwhcG
z7kdBhlM@#Sr<X?@=^TrXpc_d8!;a>)KM?m(`{f}ylKg;xoTexot66KMe!v0;CS}vE
zl4{I*>Rh*+Lyq9j?v6_EislHu(J7g=|C`sd>ybBX{hNyE@3uBP&$Fk78K=IU?N93N
zcl17&NAg>N;k>%2)pQBPSk&PO9HhfSE?!wBMa~JXj8{G^48~^y1pI+0m|-4)3*Eg|
zjc>xW*oRP%4_6k+#)+2kN>)~gz}XP;ZajNGzaj3tjY>b!Rc8rl9SyD>079cR8(3d3
zSsi+*B|M39-^+3Nr54}griA`%PN%*7;k!4uZ$oz3w{zk8V>C18Pz0;WeR|paMIL99
z_cau3t-bN~*<r^0;Jrq)AgPd>Q3d`l9hZ-xoC-_`$4cL3YhB*953kv8;Pdz%WeR)t
z2W$>0$6lXq3iPnN&E$glbc!|cw;yM4fo*~AOl<A3R7V;^j3`X*K-1dY_kp$GC2_W6
ztX2F82>cchHYtTLXpTBqFppWs%8?)7k}0=c3$RZwl=<9#sP5&u4<&CY7vT0wQaHco
z&iMWTuaLUaA-a1|_ch-l-ULd>+WQt^3O%ncs)qci-Elqa!o5#<7x(%7O^+m(eD5c9
z%OyWkC}h1WawYZH+@a?lzdqf1`C@Qu$6dknycXBI1KN3EGhU|eH952NZo^9=Yj26c
zcxiG<uXOzN4c1aKq5MbVeNmtel6X>N+5{c%MWh*kXe6;2ZlazoCw}3xaV$%w3k$rt
zaZh;}?OLgX%di7*K|*eJ%La`WBS0{0XN4{b#F0jxVhc=jU{G_a5eyGO^7SYN8@xNh
zEd=F3nZ5`52GBFse83D;z5qH#&*}3Z5&>f766y~hW^2NUJ~T<7y&@Zz<FG|Kq+SSN
zlwU?)U>SuP<y_mg9yVELK-j~YDI5H!LgeBLKY*VTcuV`(bwMY9WPPor5`zN?`2*z`
zlz50hh?G6su&rrf!~4NwnfFORG;l=8Yhf_bp&#YG|K1tUA9nAA{Rd%GEU1%<`sM2k
zKHu-Zm29<-p|$nc?!TZl@OCj6zL%Zux}@epNhz;}aMk=)4y;Jjgz!3}$)27Go3sK*
zS~XH8Q*7CnleiaA9cri;t7sU=5-q@?C|rxcq}m+zsXYI+<a<kcPsdkh45Lsj(_3^>
zq2M}s1k|q+;mywBHtaWr>1CP-U%uc?RzjT;s<jcB_$-X5Ut=Wy?&ZKO;U2(9V^oI?
zM+Kj9>0dnpsO(}GC5&OLM&!IePJD&=eRerxSS#=KPc%`5HWfe!j4%!nLtHG?T;#?N
zd}wHk2J}H~o6CZP!-4X}p4cd#qH#`^Z+`Wr-Jnbf2}w$lteXKxfVi|}@h-Zza-v3C
zlay6990h6Z074$dnpn`_ach8h&Chx;c{yG1)7$m<k<VPPGPsqD)Gxi^VLv0T^~D^H
z-FHCaEXK10cLp;OT&PG64{HLCrz4Of8Dlrg;m;n!E{8XJen+U8qeX6u9#IJvUCsyE
zouy=Vz8w^x=G`w#{ABn)H;L^UlpgMn=+CY{?vY&lPgO&oRQ9r)TVG9b&%6dI1w|P{
zF$~@#DvT<#`E8qC#4^G_e?QKx!F&+;TZCif_N=KK+gVc~vJ_jd>w}mbPdqiJiLXg+
zYIpZzYbW37JrJ&3$=F5+QgS<kB#VE3kvl6o@}#|Tw{^77@DaFuLcGt@<K)wqhT$QD
z_MT$8LH_??>Mg^f?4tH>C6!h{N(4m#0qKzLknZkAI)-MD5D*kpTBJKi>71dvYk;AV
z9D0BOhWKCib3D)SzF+1m%r*Pkd#`hy=WhknHJ<(+yef-;KdOZp^Z{V+Q#mUZ90q@!
zzr=l$f-!8vC;V&nvCmvK8ZAzx+OfZU>GLv##d_I%Ez<uxv?&7I+O5gCGx4c|y-8g#
zb<YnCz7-ZT3;~DxC+NF8ZYM-(2OAMZm44=HlLj0SIkJY3pw)Ne6{Ola?2W`(wA}vT
z($cLn$TtN$dCN+^CPOI!^u<Bw0}!w&i?p`6n2!5tg%jHVmf@Mq;2klOydtjMVmzd^
z-3~Voy$@fJ2D~TBUt96~*?B=pdHeGdGPYH<meG&*JgyJq7>aI62)}Z7B_`mD6u13^
zLsq2P>hM!<`)c7oOLdrsaH&X0m49UhVgJl)m74<;F6~Tw>!+=*62Kgn8QGhLa3Mf0
zzahvxqnPn@C;hZ=_4iPy4u75T5w+-rK*iwOrDnx{ZqK$w@S0!h&liv8)-~x^q;)(n
zGyHwO7WBnQN&Ns}5I~={F&!1Q#w9`OV{b|S>W9cj53Fz|L&BLS<lm01BK8)erIW1u
z$Y@CU(%fFXu&<IXfg@>g5_nZo{<YCNHmm`<w<aoQ8>IZJ@!j5axX|s_Toj>o{%#ZY
z?xTdysdzEzH+U49_#$CT5&Vy$>Baopo9=LcDG4^1>1jbaJw!%8HTUlD<l!{3pbD^%
zo0Tm4R^%cgn<`eoNMiBx9{eL%aiseBw3NDcqCi2A%j?*!MNfin6N3ZAK7_PNDE5or
z-Y7XgW?~$9vZDDZDMk|aIa<-}#a#OtBkS)(@bQxki4?%lAhPzO%2MPhW`Pi*c0!@k
zy%AL7iqHp{+wPe4X0+l@XE>GXS0)`fgcNsqWUpp%zR{NOZFYnSkqO)tbU*(CSa{-h
zTCCuG08=b1oZPVduy5FXWPoLRiM_!&)emyhMXDkbX!r@vhz=pIYN9^w%vfkiHYf_G
z#WP((llbQ|M4UG=%kHnf0Tx~n{0LRF7*qnjy%u>z(Bc|_N1QY!d9R&%n-R!-@>Kh`
z_b8d7oYm<%*b;E*E==Jqd|nCiA4l<taY^|+1TN3s&!2q$kPrd|_G11PrN!QiU_%*d
zn{FQ8_p_eAKf*bWh~llrDRqoi-}%@$(h+@i74={|_97kp(*strDT1_L1GLTD){@L9
zy^x!=kb)45@5kmOIuJG9cHh?~4T$13scR=c>mQ&RuQ;+|j16ndbB4E~1a$*gHxe}@
zX0Rjur|2=!Z<UXq-PJk;$P57XJl{x0^UW)@gRokUj0E3~eFqKtRTOEK@CB=M-4j@a
z8QpuXSLs-z#r-7_Jejy+<vvJXukYWjvrZ`-t4JG<y2W{*2KHrAkHC3<v%$u>iHJTT
zG?7B^x&WmL|7K93ng(g>c+C5l9BbD(s-V#Q)0a4WcviYrpDB({A4FUZKGRAQ87Cq7
zv~lB`@x)TfL#x3pGsn@?{Th8#3;@95j$$#*bV)C}@#44X+*=FBSYAgQk31WR?5Cxu
z3p}SaYVkGGq^^2%@>C}c<EHC~{){V-Lq^x=c10s61Sr2PQB(|zuwwAlO{Wn1SS7dr
z_mfjv@4Kt*0D`w3i^M2^Kp+AmFUs00V*s$0{F|HZQBTis_P9O&2RbFiOBMruEkbj_
z_7Gv1Rw_=AC`3z9j4olkY(ES!AE_x6qiX+OSE5H(qn$wtiArxM8R5<5W_wr05*Wk)
zadm_#d!5F&-@P+QT~m$LQuwxE8f={P=ixHPv&!L?bm;!q)<(x}dc|b)7iCJUWZ4+&
zR`X%=l*6|LX=2EdU>>2ID7Tf?3b9a)JA7w<+`sZqF2D=OwMqY;{5Bxq88)eyQ@z(8
z*-<C9yp>>kJe-vX)h=vxF35$eHrb;0e#?oq8c8XIHH>|o=LD82$I%EDc|u`8*IlYb
z=--IU^C)6EVWfQzxSu})%}g@Xv*!w)1I%`1qr;#29=dE;K~2yvcpWa#)lvZ!F`vI%
zcSW67`wTo4XLRYN+sUCTiDdG}Fmz%!tVz>Ez?AS`enISqyb#z^*4Wx9`7G5z-Oco8
z=A*~(iwC_@ltc%S{vC!vZ-T?#cM8S`4jOug{Myx0{l%tNHaKIDkm8UXplly1sUn^u
z^!W5LjD*6FMn&o2%XYw}cn+2^#CnN4I+80J{TMGyEyF$pQ^}#T{!B7r8%MHBie=P{
zXs)G)9vnb+mqPosmtPRbG}G-L`&S0lQDvrXnRbHjF%zK!{aP>`?B1~YEC7TeXG6{X
zS-^38eD_h0u=N~T6-lY)`J`B2Sz;d<n@N+Y&?pfwuk|MWg!ebg<>frF(9-kQY{x28
zoHx0lcT2)}=1%S{x9WNtd=OTF;;YSOJ`SA_hW{Lt5YonFso=<tL?y9TtF+N?xoK}g
zFAlNQV_zL?_#gaCX$}AM@bObsIb2=2ymU*M;ZJ!iTl)dUP`|*e;$nx{79AdIDKXrR
zNad#;N}6O^KP+EuKVP|qyNp~6pAC;HJVP5~j~$-*7#N^~S$yGB8uJx`4!=F^&-kRW
zVFXcHg2@;!-sv50IKDRPjshfAcQY&KKKY~PUrs{2)7Oaa%Y6gA%NI!YlK$-J`}5R(
zzjz4^P##ysP8}o<oji#s3d+;%_{1=`FHB(mD|_k$)SK>*z9xu_AuW1@Ft^&G?87KX
zI}gN$v*K9<1`K|c8kqf-ot+KLiJ&tN4QX`F1ENtMRy0l!s}fj`D879<YS-O1sr<cf
z`q2oMjA(G*uDXUF{m_-wd#&PrC>du1sUKSbM_Zh@?H^scz_qrc<73?GvgjwdpgWlL
zJ_aWGuLDsA<LGh`@7v^+4))=Ypn+Lp5|x})sBRd=Ued)IqaXr>2@7ys)dNyrCHC(<
zq~fcHz1hn0tAV1AaBeCMv6Cd?qDDd5_qn6Bf2zIgUORWFaJBvXV@UI1FqJEn8&+ap
zf^GhgEzpI)O^fj>)-#0!FoKxwyE@*Nw>lD@kaWOSWt$c36@w6HtD?Ng*?)egjK&<c
zNN&2ZXDAqINuiI|?!Du#&qcfL)K!9sqbiHzRbY}Go8ZuJRIIZxJqq}F0{I<f5EGtd
z;jzI#Cc<$Em$)B7b)Z`L3SiHy2Itv|Z}vKm?6WkvzsBPx(a2-8rVG6`D;yIg4Bv?4
z$1tDG=jFcixs&ZML&6~q^M-=Wh*t}d&b`?dkD;&)1RTS}@!fj1LP-G_U6H#j2<0$u
zeJ!GrtuY`Gl%Fia{EypZ&O@jo@fj(tkMvmxrMCDkM;knuMjp0%o(12K!0hvl3~`Zq
zcYN%5qWWaGy=-iF{ZaFT@Kd|&ac-qSiUN4dU6F<c_>k$X_D|eEk{b-)ykAW`*Jkr{
zmeAvr)ms3W+1UPT)k_WOtt-D$xS(I{9bWbH9+<@!rp+wHD9<DL5^a|5n&g_3GtXKZ
zMEk7Uq4xbn*&J8d`rPdd(oZgeb^0!K#Z6#H|A?Emh{^EdOh11pMs_b|vHZ}2CUO2%
zBX}cZFrCRIg&?`qyIukNLDIXmaqL%JHo<)E8Adi}MR|m?8}=hz5=YGr4hwSdr!BU~
zn;_duks2SpQxJc5G#BLC5;bu_e{}~5)%k$<1HE2+2BloaxK&;lxD4@xT<BK=i#snZ
z$!5BVbNV%CC_P2ZI~rRnM}fP)vq3<Bto2E@RwsPzznO)sAH#rjrDDc(hn5I_4-BXJ
z11p_O@Pr4}K&*(3Zm?x!5f#^fxDoa9da0rHp~*K*_$M89yEvb4kjS#|QVZ~|{kt^D
zu|-deMR;6(X8xYk{YVk+GQyl#H;os(Xt)tU87vy2F7bxeGKXaB2zEc}@`pvA|97Nm
z>RjJ7B<P5Tm{!myo#iIbs$qacn=*#ydgRCBa>qybl3!&#az#wew^fJ{JPZs<p0h^V
z-8hRJrj3lb`6giugRf5Z*ts&+AS5*gu>GUzHzj7FSc3j<h9VztWHv?FOz4e6S+RQz
z1U>RpQmgE5F8_k-TV@70K~Q+D0qK=)#a(@h+7W|A;B)T^QJ#YGnT=?Ct>)|Hiz`D{
zY>j0GRF+m0`ATWoOGhB}g#4B0EgPi$hqaof`^W-(6w2pZWFl`*D{5%43OidLcjOy1
z{F%KjO6jwS<`eW*v_SB1S=`Rl(USx&_Gn#HWCkVYt8?`F5%?gNyL{e@3B}kHw`t#<
z^u6$G`FV&jXIVItSb%-d@skm!j?{gg@U~&(pLAKr%Lv<*)~knZ4EGMvm<jLlu{dl}
zWqP9j3ipiBmXO1(^`hwyh3&;$5}09VR48{KY02xA@<u%0i6tU(Ai!RCbvagOX?A}H
z|Dq2v-4WQ-xN~}xXZuFf1N0|cZSrx!xL$*7R$!+<X*&dPzm2OQ118YF!v`o-J@-nB
zSr9H^TdC8SqN_0ch4QA1IsVZ~)nl|wn6Jn|vX@0JFn`3at8zy?c+r!kU<r%ziB!%p
z)^%mYnyBvdi7&G9FD&^sd1sG$I(gj6p#I~Mt-&Y@=AaBT`ECG=j}vNBgh&BL>;%ca
zq2_-IgNg)DDm@wYp{Pi7F}tmYJEYe?xl&iYqlfwIRy!sREX3*eZ+BZo(|?TC660)6
z34gOS7Wy*Kf;`YIY1E>lX;AKXD<j`bu{q^x<>X|noHwj=gTwG&0sMoO7`($bVrtNm
z$GSViEd@qkK|e4km)AZ`Dsjm@Mu7@EJg1%sDA^~3vRdw(O=;9+uD}-deGpHpxxuIr
z7lZQ$8CtR+y(DRwH+MYcOc#6QklJ|^l#gEY#o8Xuy_0BlAbJUH+DnKVQNr&s*G>#t
zLzZRw9*}_gV!_K*je60L3xf_~hpd1oVljrjsOuBvcP90Jj&9FdUi0$tnF<PnVuAD@
zIvp4g+(&H0XVs~AxCYwyxBV-o+Qt$x;b5di6%x<ZxKj>+a_$w8FQ6V8mSU-|*Qk{|
zxi>nbJIh?V1K(5v3d#0wt(!n5MJUZke_s3ZHic7@vn$jMimQ37P)A*OzAbw?YT1uq
zPWu_XT~XsnFA|0D8N3E!h;f{7PrpQFjN%bhPvpXL-hcBu07|ZwW>x_V-M;yo=<YD^
zbMcPtmiZB3{|V?^yf`=G=DRLfq5+sjE9LY|>E1xa2~1Jwvob1~TjsP>5srylk;heF
zor;X|(4?mvLA{YU?p!g*;F^2Mcq0M@Yb`_Xo&K$kH<a$<2#Gfjig|h;&NLd=cjED~
zYQEAN7#O8DF0>g^RtROIT6D^H_pX*^W$y(!D2g23-R;xzfj`=YHq+}55RnkQI~gxW
z${>g{N-b0TysL#H-*F8DFu5{*4y|AS{~aKDUs^^?^d*FyjV`8P0$OR@!o>Wj;I`w(
zgCt!{h}ZVuRxGqhax07a?=lfXOUHD{FQ*qxoQQoL7B9x)m$Jd_aSiA&T0cn`vsKA=
zU4$(#|M=z(Cmg>+8A;!oy+6Te%DwV#a;CvbepfM((oRWJtFM9!I;DcfxA1MsrU%UO
zA(=U4x;$uGMb*#2%jtHaY!N*c^9;YYi;LB=jd!GCpS?3)Nc0A&Aq8EBi?f5*dClhd
zT)^aq7nuz!Ui9vvH&gHAeZ5r2wg>9-+j0uIGq<DXD_#)HrtyUQYP<eN86<$R<0Rgx
zKR_%V+HU>!n11{Ny%IErE7$*lJ^{M(`d>RCr;i`yJ{d=ukXJWdkx9dgt^W8QiQhUl
zPo@HI&-RvBfPE2?OL2w<7IxYf^4^wh1aR9rmI*Da1njCaZ7j9!UXzda;Vvw#u~BfX
zi`1K>u7vlK;)l6hV%u3lcD%=#w~F*LZza_>%d4W8PBs<k{!H?6565pIZ7q)@SAR;a
z>Q$?I;-2iB8!T@Aa)vZ?ZMteS(}dQ!P7-($iuk-=#Fg0nH-$x)lVrdszqps{haMf6
z&Cg-8n11Vx%632ZBX`@-xMR^(@>U3rj<tA~oDNBZAH-4)CF&d5@daP5Ll{wz8ru&4
zvZJY=&I=B~<381SE|13~__{%7&W$;j`%L0N=e}K${WZa#7jyP(4>ld3p<>3xCCfrW
zOE=qTXk+jNizH|&WTo*4k8Z9gZ>&};|7ir)Ez&qWHECX+KBzbotN`|nZhpl0U2YW#
zu?c1G^Q*@-$PfX2oRo^SY?~CT%n@bS$*GVVh-HYCvss9yq(|b2E(OzvG$1wJKvrW`
zHEDmS4=MG&UeW)sIG7iDk1{K+ujXZKYmIt!Y$UPHr|NUGEMV1;6ef+t!#3ep?JfB5
zW91rJDeMj93%y0syM;({<^Fxmmvp*p@v_S5aququ$j2MldVZmo={BPDm3+_Z_SjLe
zSs6A72S3Oj<hI9JdVF@bSkVf-*7?^kz1k6~WH<&VTo!a&?8i$L;<eh`$ro)}6FqE2
zF>soECM_6uc545G>e8PtL;-!jf4mvL1NYt@@+|pGo!66Nt{k_^%(<@K2L2}CrxEy=
zMBi)qlSbt1P3nolymsyWcUh8A4ezTmWP}^I+Q=}rSq!~MIKhri){{+;mGx&WfoT`U
zZUMdaAvE*(0wFsAj`r%*VDiA&M)W%rEP19V{KrCrBu7O6dKr#^GCth3>1}oAicQ;D
zH~`>9LKSj53a4qMIdAR8G!-5ZtL3)8wlHpRvlH<<v<2Kr-gI(iJ!70d5vvYfTKJ2s
z*fA32x@S!Yz`QH8s0Czx7yl^HDtw&#vvvi3ZUsqP#U(6-ZKu3(;YyR=y+MtE(zDmu
z-*zR%2T@QbC+qQVAFMZYgaD(ja~g51gWThs+3hOKaK$g%qispeI>^}gUd&q?3spKA
z&ws@MXKk^VNM_d8fvpBh5`>s5?xxlcwxCq-x!W?@8V<c2In^EtKMLJdY(m|G68+tf
z^cPMwM!JyayS5@?EH(1g-ZNK^eANFIUs+I4dAQdy40h&zrDC&@_=cmc%jz!r-o>8a
zh<g(`Cl%|oBHzn&5#%xVFaTfsvKg^|6(sIASdg1X#2~cLo<`*kIjH_@iWx6+oF$u?
z_`{4g%{sW~Yv_1~^(e_K3@DK0HE7ucuB7qle^--R&@kN`KL&_Pk3XAyCwF{OuQ^Yz
zd*M9J*zCU|6%8R7-l)NqV}G{RO_d$GWvsJ340On77EW0+tV1t^YKB=7SrF_E>aoI)
zsheeD(+^znOSuRNVz^B92Sl?AnM6fJ?f*_rN#4!n^Pu=m73Mrhd6M_(LG>UaA6?|@
zOF%-h<5!I!{i#{JrAW>)S6;vM3xy$7`o$>;J@7fq_S!4;<rq0~bwCOFg=>!;_=-k;
zjfMSGV=*9=DEQp=3KmduGM+nz06i5g550Ob$s^`8smQ~k6^;R2?}n(={jOlAZQkM&
z^^qy7o--%T%X4`ElqMgH_2`ShX^{oVu>l<qFO*Ld1ZV1x&fMV8h@jxWHlGHg=QlTz
zOEGPR-1eS;CN?4fUtY!HO-i?}8Bxg1dyu-$CbC;x@#@-?hHuNvTPf~YE(9BP8z3yp
zvI8acJsDEI7L*Ko8mOlBi+{p)9xYX!>tB0sB-ryuean6R>6KaP9#5%_QP{(GF7*#^
zts*JS)(Z0Av0Q;;xEfJcSMr`FRi^f}>BMy25&ijm2XZ%>!oYP~z5JVPm)6w<+I{l%
zwZ!oXU-is(j{0C9Rp^^+b?c%HKCkXaLN`CnVT0Th7Cwb%0o$b9q6d%NR|1|{6z8Ca
z1~I)DNB?YF*_)F1v8;S2thWVH(!gac8%%puSur-`T=n?A@}`&WhkE6QMCObvaYD37
z({Ye#<Fm7t4_$9cbR>=HAG=B$te1D=k2R%VthmG8`F2#^L1B7D!WO3lrf8#1T??a^
zol)rVBR{i)s-v@Y5nt_N>7!+F*X;`v`^u5_g1nV8wt*<P7a^cowSU0nLvm45m$Nln
z&!>8AzAA=*=&oMGi@!KM4{z9`cp>nyHe&{MAHZWE>72&Z#=H`UbVa<q&ucmz{MV%;
z;JVO#%jpOJUj76KiF<ExAb98kxNxWd9-P4S(jAcP$D@)jPA=`lmvbzdTw-sLuU?Nx
zo~<-F3V-x-X3D$D%FedAJa(BY)lBex;Ib6#3s6}5qb@h@J;y8%^=sRYwKwH;@Zx~L
z>I}c@!gTB9`cN;aaDII(>Temg`Me6ncRtV_e?|cJ31~GFM0qDmDjz?@)%4Gr%Jg4q
z&&uAdaL`4ATfIxX<XvML+fd49d|Z5na&w2fw0X#HmrH75MDBz4l0Fd-hSu|Ji;jRm
zaR4$hZMUGM`58+kwwg6<@1S7?q{c=!$^$TdHBlD9Cxyh-SN)Yi9k|$)D%Pn6K)B6y
zRo9u$@dtWa5$xNE|GwWKL9y!&CpzbgyGptTf9>}{Gr!w+FEiBsKqaVxn|AUDm%P`G
zkh2WgINMGle0-;O7J%Y-^U80<qMgFD1YJq`T=^MU-o7Y7{}--~wV2oq&{LTBA=9S^
zSa05azpT`r4AN9+*ydB&SsS~wio>8NjO)t>9ada;V4Q0EU&rrC62)5r4+k{t-o4-L
zuSjYU`8e6@$P!{GPi%G4*QejZ@ev``#$GKlS3!h1GFh%zX`Irnc}Ph>mS$BGe5pOE
z@c!CMGoL^KQ@&3tlx48AD9Z2keLIy3eo*x8Z`%{SYTn@|(YmpC+8jnrT><Q(a&#H#
z-T2z74K)1Wo2Yq{JU`G~(>!>$4V3J<e(|#WTm6QLov7}>QmQ<dzpOIUEct*pey2%O
zYbox~_(|mkd;=(e2a7(D{1){)*yF`2VX{u6kIlXR6gDwhoHH1Hg&MgxkCW>{T2=$$
zkdB@SCp*8(MLf3J2KttcymQL59PQAR>k0Xw`HII-)pXu@5Ece=r;ma!#JdvHc*Y#~
zaH(qb5}9?BPqvH=OuS@SM~EFZZY<rEVebb8)kTO@#g*UO42?|`C42o1RNsc#f5gy&
zUmJ#JCU;x~sKErT)r1PZU>R36UmF{k8Az*Ui4A~kVS>Sp!pMDZne6-41rihULS;Rw
zlDwp=yMpiPk4kk}voS%d)S~~Ss`s%h#?qe5`uLdt!G0&g=a4rxz@q5j)WWplFCHj4
z)4d&lWBPn%cc&nXf_A93QQZ+AZNwrIE{hBnuxZvJ`@Edd(Vn>~;wFOL*6DmTGXCWW
z?k?;1&QvNFZk13vTb;U(*4s4@opBS8VD@xPL%=+VN_BU96J`znD8nXRobycJa!P}6
z#3Wy_No}`=Q8J7OcL|nf47tdvTiT9Uz%1j5^SgtkWTs#X?D4@l#y4KL7ds$Iv2^27
z?UbPxdHns+3tt~q6Fmb31}CD@5w%V=t9jNLvq9z%=@7TKCQ>)P0+Cu;l*H);aOFRa
z{S4VN%+&i7H(4~mzwkJWRAj#LLU+Ap1qVT@ZZOca;0g@x{h;yvx;*}T@tNIQ!Xvtn
z-b~w?z&X>C54+5&F+_wjf3$2ch}G`}>5~5>Lwfl|gJ<#3qB1az^%brULY-JRgE{S|
zo9r%(F8>-%0k6t_RSt=b=5|gM$*cnui!AAH&YvFWdA@x1u9Y;-1cAKR#5f6dQQye`
zOZKn)1PQh^PiN@SzJ9Tt;Y2bV0Cjt;d<Hl1^0#3QF?UtOtT7{abY|cAayM#v1s`Z<
z)*rQ^vRY!zRw|(-;%2*dGX@q2BHN&h47cVEF1Awyu-Ps2O%2-5BBtsNLi_O1Eb|K9
z5u9vuE7)qot_`=sMP(v`$x#*$@d2*F7rJ<47inKp?j`@Ud|BSgdjKS6VX)b_5l?(@
zbFkc2o-XM6i`e{fgrw%Y0HH7kW-mvyhuzAG+RxCGj!x;mnLk~fSuwVdrm^XZSD8sN
zDT2>ALkcDAC-jgV8}ch9RWk6uX2;YWT<m*@RY3}V)HTw*QSEuqK2MQ#PwWRlD*p(U
z(g^^0gSkRQ45{X_F@*3B5PMEmEm?h-WnJ7CbT!TpTBcs3ogjNd%lvB2Y2U=J?<W&e
z-%A35Nu;3OWBbr?d9o9`-mj2#hb9cbrTQ{zuqwOJ`z;SGkc$p6JVA_qCz>iA;|?_K
zI6Y|a&dWx2T>Y9Q<E4#Nl$OmDWpIsdD1V|OA7*I*7M>ZIa*#Q0N+_lFQ<TMj%lAzg
zFt#7`yzT3oQM#{+*V6k>06lP@`$bUYIB9$TNB(nGu?ziOTh(eQvZ{so3{=G-BM~EG
z<aGmPK|IXfR2tgg+uo@$?+~}@fhkb_E<^_(ix;0mZ+L^ni9-t7bNjuQ-mV3yOEX%@
z!!qisus{^!d2krCJX}6Z7PURr&{v_S3a`qHh)Yv=@DNA2O=^F>RvXp=%ceg|f8*xw
z58{uJc}Ow9tCiHv1n>EchBgXX>42*F^1GC8+)Z8TV@CwJU+N98+w_QYRean1C%G2G
zL3RASY-F`M+QQ5^sqJM&ZM~61@I;4iUkat#)b*xXg0InI;MEtg`FxAATU%G6Xgg$Q
zUr2y$EoW1u&7Q9EF1c$$THi7}zTua6FOAY;GZ^&2<UGK0&woHkd@?y`&rEac`Fb5Z
zae821L)=-S%i~3Mvf0p-aPs|@eQq>DDX8==p<bGUR^hcFe?{MhkbOpAotN)8r`R8>
z0MuD!-S9anBo=SNaV4p)ll0_?ON0|3M+RX5jtrqG7iDvh(tj!&VbPA~FpnE6fzwXe
z3IzJgN}=_I@T1vQiTZqj{aEvNIspeWR#h9OG@MCT+ALTW0atGtR9XIzJT_gc!uBNA
zIQ~MM;(wIcuoMV*ok1H@5DVDoS31#IJJs>QTYU4?1=e9nVu7`|3ofJizO3W9Df4$+
zsiitFj#;T#VeeN4<&G8#2vur_Y;QN8hF^2#UYv~{nrBdEUS4ZbW1DBpO{gV5Z&R|*
z&$8|L1TOPNjX*dOXREjN*R`dGd$BJB|GYE;Gy&3B%}iZ`$Ib!VxNV2-3lvS068?W6
z>L<3FvxqzE8xO{^eE0^8A#|A4QW5edvlULc7D$v_t{d-EWIwW4t~LEI1n6A!a3sJ`
z0BOsVyVL3bfgrC|Tc=VyVrjwJJZMnNc)Zb@j&~t<b9IZKVyFlU7YhEalj^o)t-cHg
z9^9^5_E-pZhsp&0Q{I<j#nEEBDbv*~k2l3uvRo#R8=cZreFffnX09!N<mX-a3u3FQ
zS^46tpklLlyaProj%TzO&*g+e?;iJZix*>s5Cda3f!YLfDgs}c(|_cLb#s&UKgXR7
zA-hc%ktcD2mOZJf$T3oz{ARd$Xv9bg+3u$wi0E0Bv57>DJ2=o=4mkAIaovgB2Bv)@
zQ+B~9Y?5)cFBG`1Sh1tAo&m5rmn33YmW%Dt+m6<o$vF9X(j$ZwgO68rA@SZ+rHe)A
zZJi?~8ZXN4hgn~~NE2gRg6lInkc8uV65NfO3j}2Amd9q%p-o>uM^~iKtmQ%>o%77>
zhZ`^&q7)<X4FdDpjJ-4pSiWzAxl<@j6so|JNLGgLdfR$IAhXn}B^e~!CloMoH=a``
z=QkL2xmJ`%m<H(oC7(pHLxlf*{|8H^RLvn0-fNti>BOg$W|Zev@)}IiaJeXLcB8vi
zWjzG;wOsqo$S;Qtn>rfeuA}esA;>>dNAw`Fw$s6KH!2E>KR_ll@-UtwEA&p@4+PyS
zN!$m{2;YY+#?l6b#VEsg;i8ONx9n|~h=bC^!<sRP1m7ax-AQ2|$)6*-zx-EdN<<_q
znT<kjNy-z+<#I*#{^&fui}&M`xqDj^t}Fu5(=g&>R=b|NDxcEZ#dO7nk1_=9uZ$|t
z2lWALqs6=pp!5J@KcqCcu3}gJ$WYOerWDPRf$Cj@0E{t<yEgh}+26)i^&A-BlwR*@
z(l<Qio?r<uw)7nF+aUR?WOn!CfdRNbyBjJ6Dg4h?BN7@dv!NjxaRX6n4O*rR8@*~C
z27g;tOz0E7&VQ{Y+Ywf2kmDs)^_rplm-lHcSJt8$27fls!Nj7FG2uQ$OlO@Sd*f+e
zt4&KzK>IzF)O-RdcXur|6{Po&0x(JWbg}Q;R!-(CHbkobyHtBy^HuEZDtFWgS7~QX
z0doUb@8xSsyCKE7lEGS^`f#a;JklqV-bx{z8h`zn9JWtB&h8tNz_glZWG)2~5vWdV
zD^*V}zrgHyD@%*r-$KOI|7M2-x;rg3d58LBhg@^$D26t7kgb44`*4GuGDO^`b7e&{
z%&b1_aPBub9HgNLt&-0SDO%xwH=bC<C64N7eh9vL<MBl0rrH)Q(4A17%KAWohK@pY
zqFCd|@i-z2p3dj+>;CrWn<M!y4mqpiC>8GL?(HisyB9R#*OiX?<Q9D^Y^#d$5D46Y
z!K$X|_qqf^&}Gh&@1&KS-o6ixZlD*b724?2h`Gwr5oB&tELO|-aduvK4p3&TEfi&!
zB-(CH**V`UK)mUP*j6_-Q=^+k3Q&!=cOG#Mfa3qzL#(&HCw#diKEy}4lg&n-&6<U{
zpbg8cIZ|)tCRV2HCEu55MWw$FsTqnqzSa1VgJtLE@1R_;0m~5uk9z8Tt$tThdApa)
zXOL|qZlPA^q|f8T_?h9bOUS=Vc*e{!oYe9d1;X@3e%mVKS=;?gN#r@O#2zt|KlPCv
zJWm)}YIjja|5(=$1`t%_Xx>uTdU|XSxIH2k;J-pDG0Mj={~f!3o}nfq3#~A;i4fb*
z<{Iq&q5e4NcLPzyVYukleF`1*I^tUAo_32Qqs@A^Y|(W$AeRxSO=3A=JL{5jqf^C;
zlxtHi<Dc$)Rw15CGaslpST6SMro)mdSiQS~>E82juz0+JF6YGEksa)#hJ0{u%JkUH
zwh@SnSnt~_a#Cn~Rkh^hs_e}Y))Je@3hB}Kps&mpNTA26p6Rg7Z^iQ=Uh(fh<G;4F
z;~&192pY=ChtUz(k3TFW6`XiH>@&Gtgym@<NSyF|&R3+lMoqZYLKOtU$#UY;+7YHh
zEw>yh|Bm=+2)mSVb8uH>a$5Xx|MN|p*~p-<LGr&kV-R9uH?C?h^r|p*(OJ8LD%IS#
zsRgqCm<xItj@(JmpxcMfAjM{o`se(uUqo!qMcDg2p3&kd$h!{z5}1nTLrg4jjfRTk
z-kz$+E)9IY_U-?{X=W`-j-0J@(BRJgKFhfyH#`o|3uU<+=Q32dS~~HdF8mtv>VbC3
z{;L+~^dU&A$U+KM(V4+qUZM{+7qx9frMb^T+(3yCg|C-m4_(XMVSF}NA_OhBKaDLq
z81}~XJZ6urAdQm3_>wl~{#@5xt7*gJQ~M6#u~`h6kIv`8XTrJc7v}+`mZ>d=tzE2_
z^5@~M<{oSEZ}w&?O-Rv3RkxSujfCFd6y!;-<W~44J9Y=~lMMGiac_iO3XUQr+a<Qz
ztnJycfAnI{IwQ^KYKvi)UQr>g0`6Jn<6o9du7@<VW|KREqwpH^^p}M?`b@v?P6!*h
zR5foIBRVwUUjw}hCb;ctT*(xy_)IBnLk;o6COLCfKI#+&@A>MyJx2lUs!<-$!p_G{
zAb~c5SFgu8Evsw!8@6{2<T??<$~8;qesT1oUWxCNkx}OF2z-luZFG75J}_in#Zb-Q
z;eSo&Np;EeT{aiHy%?^+--#7dk`+})i+Uzj2lW1{4+WWKa!<;OS|N;|EE(ko2VW8<
zSIQMCd}|#$xMDw@3Q^l%Z0Z8(vT6Ps;)3~P`kb`n2u2eH2>uGZ1L%n#u1y`y?4p|Y
z%7#W@cjq~Tv{z%HS+z#9qiUk(uqqGqeBpFfcu+7|&u1!<+WCdDX84yV>WQFY#RRtx
z*mksYx$n+XD_F^7!YG3Ff4GWfCNem)m{`o~`V1H6`HC!j!lPGw<~I?fOAloWlvvL`
zT0LeCIEY)&9*>L_8A*1P>#7Mp?l45HMcbHJPSQC1K1@MUUyp@CXVj=JR<smB`~W>J
z)@|bpOO}@!TN(?P^&7c{SO@Erl)3!Ho+kLH4gC!(;IJ8)mA2ANKs4_C%%V0+WM?JV
zJE`UR{sM9I)kq_A89bM`+AI>)2t>e!+tABL@wiz5TTQFhLI;xB>r||kYSIeV!#a4W
z(JumDLt_!qKnz!O7byz<e1Eh3Wrw4U^Zq0uRN<&kW6`6lE?4#yF7hom8`wW-_0)EX
zfbDIW*R0X);&<i1W$3MMJSH&+H)wPihOT`#G8Odl&hWL1WD;n=fl-?={jL@opRY^t
zkMh~EKI{&vknVIAQ04OBny@N$ACzZFjk~>b{^UZu7~UFw0eYz$Jn={9nrL==2w)!X
z-mN29FL!~0DlU8Ee}v;{LNX;SB8`c!M-9K|UKrWMGIe#M+}*jkT3ev~2>tD=aX)YD
z$ANxjE($o+G^T-p0;m~NOrr~TryZIZkFwiuft`%Wa4>o3&yNI}4kNeb0PGCnwG{DT
zv$0Bn1~WsZUjrQm`;+xQhq7{BDoSM5^}O(};KN_A5^1KV{jxaA%{W+1#xeM3ydlW0
zW-RN6B_3?p!S5KYvhKS(c)*h}LH3%B-b5eJQ*Mq?f>cU*!wj^(T@HUQ7XsP8%Or>z
zF3{`p_zUfxX;reN13m>ggiep7M7@a9?Io^jhCyvTjeTcre*&Kj+xq9m1gZI_;~kTB
zsTNKUxim*+wXbp|$kh>s_7C|+sI4OBhQr6D3mb(=GPm92K?@Q0PbOaS>nhR*hqx%}
zcPV^ZH}GT$7!^kcXE1w&UyYLF=jL+U)Yq9Oyy&AgGpmSaviPZn*=#<?PlQ%Q)gIGO
zEEIXJYjr#$DA2sRK+In#=g%c3L(L3cj)Z0i_TjxP(@~UvNTTR>=yC`<GH)2skVxML
z^w;Vc^QP)$iIj>o^3Tl#iik01hUtkrfJE68?_uXQo_uE|FN@9kz;f_k!jj&G(4vpG
zqXKRl-r#%=j>QeKXGH@fI^}CwMs8$BD!QoHED!k%f!z(ZI*F<YR5HfiVe<=ql=nUi
z*hY-rZ<`4wN58mC71+0q-}l#5?WM?^JA)MKA{ML80NUeH^GxRJ4|+mV3kJ0;!r}4=
zeFfZPdBJiS&s2cBv9RqM69jWK*`7mx$5POyEUYJGAGADPtofl1eh2Hv6WcA*yfq7$
zsS|u)8N)Y`tPb6|EmS|TJUN1_w)zy+WyWzryH%+rM?*-iq&}^pd{?FM=U>vOt5MsC
zLT_7_$xqI^4yoknP7cb*(SzkxJV&jS<#F~+=z`GSUM>KhWp}<jRlJD-vpkM8I#LUH
z_QGGvWx>bnuSJW#D<%LF<;=AMY5%pM@t)_P{nO<wam$fLfAL^!J`n}MGBqHKr`mN(
zMiYSFx4B!O%7|Z0Y^dA+c7a~>xu5n4iT@u9KoSujQehQyc2%mX4DCO8O}Ag{!f+yw
z%nBJ5zWBItvLzaXHBAbrJDyYaBTH$6m9ByX>l_P`c9yYR1m;8%ekVhTOI-UHg=`yW
zffpKd+I&8yQ3h8ucm3zkv~o{`S&X6#xj>34(ijEnXI;>XG)fVG47JFl7`0Y`paCc|
z+1P8&>Mtya+?g&z0{V@xV7z+1#(||GalfW#4BgOCk5BY|g(Qr7`I|iC>MrQDHp}f;
zr6r<ccetJVZ^yw$V|rr)Q_u9F>l}5n72%@MVr<tx{;8ab#Y7bnBj%J9Bvm{%5;61c
zpw=ue<?J1F+Dg@Z>sfX-80t3juyRz^EbPVpu->=_=HkG`VTR}N0WO?0V|K%Nxz)($
zNA;&SbZm^qZyjX1EP*AlyfjoEaooqSEgwODW>yGF&Xvi!I|3LGzOJP@r?V#aPwQrf
z2NEU!c8UAO-RX~ejsbk%JR@w1abAzr2Y#Pn&aZnUar;d}&XcRW4c{Kj<OYojJ2d5d
zqbTp;m?R?%+z0z(69?qKWYy(_-Ds!2l<mL?9N7stNq-Q`BvL<qOjQbSwtsj>APycI
z25jn02b9C3xtgub4XTF>j?w}7nt5bz4gPk&vRV=7k^yj!ruZdYSGPrK@bkPQ=QfAs
zfNiseW{DsQxm>g>#9Z8!Tg4T%SKf6Hbo?o8YXf`d)+V7esh8fh*zYKdGX{B!mh>^=
zt~EXmDJEG5qCu_zmf+X8^)c#Aqoy<!0f|mVM~LY4j`~ILz~c$2@0Vi*K-hi3#SLLr
z+<egyiuVHSkH3A|0J~jsWd@7e)liFFl5(<#^9EezjX{7y#n0%TVuw=y7ytZ<sqefc
zLin!fmLr=S1=h7pFo$=z&{(-r1e+gj2X-JHS@3aO%4B2H4!(xTr;AZ~<PFPT$DfXs
zhJvTFc+*+uSg1d3Hf%_4-)=*<kaga`fXT*f)v<32liBeaffr%}BrMGk!b6nnrAZwK
zFzhbm>X_GLukqz0otn&tf)L(b5{&(j4R@0Uw@nFb;BfRAMpcaQhiJUjlzHQkwrT)+
z=tjVZxkk=CkUVg2a5LNYv)xWw`7sd%6s3S#lrGg9^CO@9TJaqFO}%Yn|NLBlFxm?`
zbR%*o_y>9~Q>I_%p|zch80{OTq+4guk<cG?&|}!onOP2fkGA!ywvTUPo3!#+NuCd|
zFFXdti70Kr?GPiML|^^b$VBM!5_tIUjhwrsCe+GWFM4_cPIkm=9Ffbuv+YobZ~7ab
z`(9L8#)gRPmAt$craR``{&KFLbkAN;ws%9QH%n@COo8Ievs2)ScO@F#Au!U}gBS^2
ziD}Cdk#xmFlH591?u^sWmTlx(HFXt!4Je7n<enrhEq@M8@ONHn;VO3`5&e_bv?*Es
zEnGpDL_8<2Y5Vlo)*iloy8D(Z2t~nv3Jaa~owB+=47OL}KVJJcnyiNIVoUE~_ic2c
zvod1sTo^xQ5f`;jCl}SPm(8i*p$(LiM^y>4AC(hyF4!&~zR@gKKA|rtVEXG8^7?kY
z>TX%d!6L3Y{{F?xha7rQ-(Q7829$Ov`Gpj1)LYlhkD|=`v#us;=O1TAUR-?40#;_x
zVtf>K8tR}yL6-2hk2aTq0ZSM~T~)-pLTwG-pD&?SmBstm#)|kAEkRn!K_Ig7ex=QI
z6;1(@1{Yt05@b6{qhWFDMY+181Tga9gXoMdoS}jgdghhSf8BA~M+foqbH@xF9CnYO
z#>AbI7ECvr>D+4PIWN3`uT9}|(te9>a%J~Wn3uir_U$_e^~}r;z3&s}$vjVxyeLt7
z(=xs2h|RU0%}zXX4A?vx`4dGZRz(RYzk05J5byHsP_6C|+j}lvc|=pir<*z295|OM
zD414wMB~tHyF2de+ZpL;Kiu*o(Y>s^J*i4?riIry_-d1CYsSF*LXBdeG7I3Lx3cHH
z6Nd$Iv3keVTO`SIp6>e1&e!MXWd>a4;+rbg)${b*Cp#^RUHr@2Lh?B@hL8-VLM*4;
z+`f?<d0izN%qM;(*6T4psBIU!=<jkfPEZ*O3T{s{r^`6NXlLas%-^QHpZFaF4h?$R
zRA(}FkAlF04sTPl^>R$w%=!O?tBWD!bPpDSdTpIpHS-Hy-V~b8%=ldY*0Gc-%Y!$n
zr(n=FhD&u9d?*9wqQF6sM7Ulz@Q$Ci_B&jC#85lG4wj&psXtqorrx4tUtqMtT%WPL
zMSTp76aP@kzi<s!@yl#oP`$VYe8v`w$M5kOCOe59(c~$oE1wC6Y|g-6+Y${rXLM}Z
zt%Z;Wx9+MUMb*MwsVsQs?Cb-)|2@io38mQ2U*#ack`5E8t2x+~RygB(Ipg5w(H5>A
zP0z5uuH%dw!rD`vC6aY6nq%5@dLblrWj(J@rf`0<^Iie=&4(kgg7x4Yn^f%IvlS3L
z^Ds8}_HMHI`cD7&4>X^^xR${gUR4t|L$`lo*a6j$1eh(UzGhBW$s)^CW+P&NZPxp2
zLh-ZtLFJk%g<MW9ZG=I`O2GC3kMz(rCwRBeU=da8G9SA$GYs$u?M3%T1%+0ERn9JZ
z+(qy7JDLds4Qm~LU_yLm^bF>5FK7`PUpECaZwL?ff-CdY9YGXZaNV)!D4}Bm<MJ+z
zhfhubuTPcVSW|Tl>o*vdkxw!-7LzHq6P{*CF=C(@Gl|61-;zC&n0qCmDTkz}-`Tso
z9DWqIB!0QtHM)0)Ir;`pq4YH|wt(@Ry{D+i-L{Zsg68Yb>*JB^Gh9pP<%|xe*DRt6
zNnDJ|YOa2&aK;~|M144ZC?2Rf*sBc<-VZW=_<gS*PKML4cpB<P9cSX!B)kyC+#65t
z5D}+{up&_xAlPShoTvvP+*Z-ti~r(fDH`+JlRJ|D3f$aS&ASWn-}HM)6|R}&6Brek
zCkpiojw)O&+?#O~Y^2DxutLl@#78~RR&0*firw?`nyV+^(||~}kpOo>V>%B_$)dMy
z33@Mrxlhu3ZZ5yrcgOq?bX>I!YF!q#Z}U0Y!F-TS-l3K|Y5^#)jc^C6iq@73>2#+_
z*P&2@UHuiI-Kyf-S)_`z>4G`f++snIX2(#Fy}UXolM?OD?3x%Z^?|U2Fux#yG<J_?
z%faxSE8t`9QOW=fQMHYcFC&GQPESxwXdBG_2J+IEFwi9FpyzInb?3!j^IRMXsoZMz
z=nKsYJWvk1XlE14+?t@52<jFa(d=3*e5)~ZyA)}XEiX-YNZ&jwCL$)rnhRd?dMr)W
zApTb*x9yU~`v82Emn%PMv<;z}@sUlsJl9^jIMAQp;WE5CpEORLSx@3jHY<7rQ+=k9
zVD7$r<Cqw*ALTOTc=XRbf5qifEZ6)D-%^6=5}cb^&($~foQ{d%89|mN<7ZyYOQ~3|
zi_X#+q%UBlu3i31&$i2akoQuK#LdMz>RqWG)6{eoO^{v+@6SssvoqPo)Da~aY2Y0E
z`FN+`^Q*aYK){G^4X@2VSC(b5v|@l+Sl|~*e3C)Y&adKs+r4WWf#yz@`(-<_<hNGA
z)|iF5J2WJ@_*GIl1j_XxyudgX9~Ld(mh#d-cdsv#+k+4LUe|!dOb>59{z6xpqAdSb
z3^iP1CUQ5JT;3RA6Z*6~g5KD!hWI8%k{h;v%3k}?aiAXmv}2fPFlM6K&UvayC3XDz
zgs)Dr#$LWk@W|uxi9%OPWMTJ=#Kn$hMVrWxhY&w7HYeocS+OMnWm^g@zYFzBaPDCr
z{<jhy5UmE#_N_xy%LAiKvPTDj!`GX|JxN@qt0t0C+ydwwCE)!nQBN#VI~XWhobDV<
z%5lLDRK3MFC0i(kY|DhQ0YJY$kQL@!dg!ey%U=kP13u1HpA@Vy!UB)U8<gyFWv;J8
zm+rCkZ$7{n{8Ih^$T8<d@u@e@?2<}d@0Ifa7fVWbdr~D+#dFW@AN4wGjF@G1f9|G*
zcZ5e7+%R1ff5A}V{cQpUb<U5IUahMMn5KLPS-CYsTgrC+xy<SH?CvLkLwhXVWse40
zfod6Om6<Zv_&aCeVj&u3Hbz}VIR6EKwe6pbqy3KueeH1GpY}6ZhqF#+LF3MY`22B0
z0i%aYPUqXh$e?olcP^jyt&vU9n9R~qqr&eAH%=lAQ{(s{s@W1<xZ12S3iK%i)|$=1
z2j0@LBGi!9Q(h|SljWczgw6E7j!kF?u4&nZMc9);vb;VDU!oj!Q(^$vKnNQDNRK;r
zDMj#Z#W7(swmK5z&)U-0`B}2*LENQHecodI!h+EDoGgSLq{M>A3je}y8pjp*LKGA`
zGFr&-mY_h1^1)io%S+1?<>aRkiN>|>yKxP>VWYS8Q*AT#Hy4v;ERaBga5g2a2lCRf
zVDdc(qkOh-oim)RzpT)n-flw4S9PJnNC)Uot&F1xb|oH!2<>`Hhsx^_?3s@rv9&pz
z&zp?DiU@xmLFp4z5e;a%6vYo{-o{D`rTuu7$w9t*7|y^ecOECg^m*YqYMz;I5D#C>
zE2N(0q9ZEpD@6lIg;k(AOm&7pk3YCa#jo35_t;2~V8)|=@JLLPg+D^UP$8BCzwfRc
zykdMag%??(Tk+lMj<+OAi{5VYJ?1_|F@hxN<@&A(np<Eait41rdn&(tr1t0)4Ku~`
z@pxU6q}2a<b6+DOuvD{}>Qc0%x>_w7fb0QV<LWn|UU&WkSQcH<=}HEg8*`K>=D{Sv
z)I%A<BpJ(f*QdO}r)?UbATYJPGbi;0-s}3+{@)|}n?_6kulD73HOJD=AJ%X&L@bGc
z*0i6~n39%e?w_v|xUNWaeO8vA-IeCVK*Q?eA3bqYO`eFv%-})D=c-yg<++qf<0f{8
zTGz}6#Dr2xv-Q>16t5pJSBPhdP0d6$TffXq6mBlFDy6s21K=KtvZFi7ZlKd`mOl!z
zjXUD09F2@Fl7ylfqki`r>O!qDW9DtHf0W>zg6V(mS_34T%g+GjxwVGzw09WY9YJB_
zU>FTB`|B@=3HITqoupw~U|mbNc30tlZ-gb5<o!+JU$|rAS~S5dmNRAaCkbqa(ceyj
z3+*zXo7&oRpV&RohW8h#f_|A&A1pkd`36fV#G0XRsC<U3Xx<Vm=sTHcz}(M8%|rK+
zRA+fgZ)Yjd&DUB;R;An=X>J0nY*DmgYF#`2PfpXuae7J0qA4!N7XK4#I!kk|_v|9#
zfFp_XQdXLYK-t1NVqpW~o1jmUKGzJ&z>GRAHS2u=!!I9D;2r}NA|+2>#><xUBO{Tm
zOTa$9=$r$jGu>`=<H6X*?h|=EhgHy^#5O?+4jqX39k$J_Z{xclEZSe%eT{?EKlf@%
z_Xmz+fG|J_PQQ{B5wJ%0W^{WIO=~~&ia=AbThMatV$rd<x75FLn|pq=<vAmiw?5&0
zR*_GEZ<Odx6p6BK+KmKe?711b9d}+Fh>qN06vC-o>SdUUpO7>CST8HL{L<$($Hebw
zqd)rnd0;;=I<`%Fz5iLVDR6oi`+TI|P14Nykh#A|66+^1Pcfs-`~}#zCt?Fs9ZM3L
zVE^ndHP3NjJ<DVHkDvYLZhA!V_mev98&RWFgOr$?LuJ5<ZKFR?j{>@C$#HBkGOoKg
zC6bTJFaeanZO6RB;Jal^8+x7vGD30X1&Z2lLrZK`)Qk#X5K__Wtl#L4+|RaEnhh}Q
z8*=7vAh|)O#@U;X7%gjX=3nZu7HUXkhb=XOxI#pGU%UkVc%i9O-;==yIk48)sz$CI
zJ#<HXa%H)l>Fk%A7ehD*!VGm^)1qd4dLK(4O7X)R94$wAyK&8}*q@vP=123W*WS}w
z^kyDDTJ9F~Gap3_wS}((jkz+3p5SQt^SDh>GSbO3YfsAprz2Uv3gcDbK|tleX#v))
zMr^@q$lB6*)#z7-dTLym%c~Cts@1De(FSyNYjM?D6pFyJ<m&djz_@;#r$g`W9yKGf
zQ(c6FB%y2~V)DZ>L5eigZhzA`%4|3BG<1mqdX{{piEF?*|CIMH1#fPJBo}s~=qEyO
zDTwydgKs{-)SKoe*l+g5+>b&7GUuxW{6g`5>(1=cP!rf3JI(B){?6FNYuZRI<V0f3
zO>Nu`88Xxl;DrdA1*atrAt@OosQw|1V*MhY{T$f{+HQAH$(JDX{*tNs#x~c@;?^T#
zf&urxG?P=9j=J~=iR9k_-E#A`0-=!$vbjyH=DovkLav2POb5w-Eh8Yd0DL6<{#5k`
zwibn4#A7$${Dl&E(|~r_MZY3r_;(x>U@ArexPSk9Zh!?CSEjixE-2A1KLl5eb#Cpm
zpI|qRbfkD7u5#^1<Cji3pHtp`&wC$*dW6WfMV0A|630+GL>DynkzjJHc!Qe`_xCtU
z+8vxPd+lu+&$8J^T&ZTcT`T0P^FD>Q4r;_TC}O|;h{%Z|PvFRk7=!n^@iT-{esHY}
z(T<7%_jSjsRX+IE!ddBrxA^xJE$hh|RmRAZcY+yJ(OVKJkEgs5S|))nB-5LENJxLu
z735wG4E}?7M>y!R4B4a7_?ha}aLZZ$6f_PN&6%-Qjf`^wU;Rm#ah-LvFB=nSM)a5e
zdm#Tk6+2xigGD$5=1voGCg$>gPXKrfzV1K=G4|w2hfyABKM;MEc5<uV0K)ySGv<(f
z+yI7CBx#MiJcEAAYcy+fiw+1HWahK(RH(65V+`*UT;ARpoLI&C{P-&oQHW<g{jzPg
z$mVcUp8?Mj=};_5SZ>ZIQi%j&BBlE5m3Lnaovr{ubi^XtII-g6@C}oRF8{8*SDGU>
z!=30WKP6Iu0uFMk^mj?Q>s%2Z(@gGaM4H=LTrEGuG1ukKvny!C65Zz#=(&<9`_>V8
zBgmW1^Vy4pPlk7tq7p4Zb|@tlvO@Vp0!}ifMz^`z$#{5mg6n@CmZ_9X`)hR54~os9
zZM@0I$|14;-`D@&rSL!N;s5(sMKek<_HYJ1yY-;2%-AA9ll?#TzA`S#bbVO5M35Aa
zFhGzFNhuMK5|EOX8oFU<5Rg<#nxR2JX&AahI)?7<4(Vq854)?2d(NKSPw%JqgZpEr
zp8LMKuTHI<%LaRJZ=#PS_+*WIDQ7&&kEy!ALCN;8utJqAZ$%xZSR~?!>=)9*LVR>f
z)0Bd|w(IGUM@c8eAUiCxxU^B<1Ks?oeunQPIw%PJGhTvHjii{GC&4t+2<Y<A&dkc!
zcVCVuK>t%iJ)sO2#PiuGk77i^JeX`|r}bczsYgG6Jgy<S%iM-t>DY<d$;H0T#Ms<o
zCvbDAx-O4Jp;jwQ&SQIQ5pRhV>58`hyvu+2>$nC3_eJ6RBzc3>?@h93-t^~1nr_*%
z9Hg6?Q=M>g$T*>av7FV{6V8!0b(I0+8^lQ(gLfSqqnu_UOF%;*6{8QG$(2S?<Tut@
z@arxlIkoPOqdMfMkzu%V7a54!PGMBk-`hF3xXA9xu@ERkxJSk%?z_%iD)R8nvj+1C
zS9g2cBP>%*CH^g7ENgV^w|sSiGS$dHPO0ZLYKziAUt&S9(+Mivx}?eR&9OJQb(9nL
ze=_%<Z2s$CO>{s9r}bKf(OGS_N!;@E)Eg~FE^s+2ioDTi9@FR`n@oZk8KaI=4gY;l
zJVSDY(Wj<IC;JwWnxW4lpYraIZ7Fp<b(ASR*?=%EX*1_Oky(%C!YugyDHEH?DE9um
zcfLx@IO@SS?Kwj8=9}7&M;-?q(OD*z#j-(@rs37F3El!<?fBi&IP#)-U)yXl1N>e&
z$^#@}YW^|93LUn~i&N+RYGHjrv58{C_Cw4nIT8+{e1BwNj-pExI=A31HvA0HBD08&
zPL{JV9n~(uLdnlJ-_o_t>j7)P|2Hm$s3L!h5<ZCTLo?Z&AQ7Z*c-YmUdM{IOjM<^g
z6NnRCa;p@o4YqqSMw~AMV()}}UO2ipN<=8PGl-+EX=s1N^&el9UT(2W*F7|Q<@-t-
znqSr69_9Q3Y*j{>W#HF18fh?j8f39ob?TCqj(vW#0&Lss1pdML7PBRtiAN~@5WK%(
z<l*fs1-^0pvG>?)><@?&XxH_vqYCD|HC;_|xha{<f78pqqpn~a{SQAE)9~VS1nYdy
zq|9V+0&BtBUObqTye3;O?cID@yv51!N!VGT+u@FVDmJlID(yH{okrbTzGf2?{1uN^
zQz9=2au)lFm06RtY5UMB`8G|%1guDZ(~SQcax7si&Q$*d-5=b8o<8&BABpgH{o{z>
zd-xsiZ)jFX(zEG(bW8%DLkzHlHas#ivWN-m_Vt5$PAa_>3Y<KM)vNpE^&ebKtK;c$
zK2K5BWZTU@4z--ldsjv=Ispvu>&%cLGp7<aP$&ikX%^ws2St@$U~G}~|9-09VJ^o4
zNPEnsX0_`t{&}wgEM&0{cZ`7nz!qbr=6a|%xjl$Cgg0oU#xEPU$d9fAE13#07DY9{
z{`zY_fc2lRc}}O@@Fb{fF^p_^cU>+0IL22dqLD&9Kh<JT?r4Rr;h0O2`-L;GnllQz
zo~d9B?12R}_y(;i%4DTo49_QDXMG?)Ny{$cqOX6)rAN6mQG)_#iJ=PBMsQPp_19yx
z=6`+9T&oZ2PSmTk;~Bf_IODlA>Ch4bJXI9Z%W^j}-lpiYYO$DIV$zj~@3aT&1$D)5
zcLH&oi49<FXn6`>FyGc4g<97If+hInkuO<a!*K6+3Sf<+clLCqEOUQ~h|6?3cKpbj
zDd%-^AZTpc2`p}%Jm1)e%dnnr&QQru%m8ny^;X8kJar1Z+KC8rKO5v)PH+e&;jxK0
zS%QrAUKkH#rSNP*6{LYKmDKe`LyuH?R?q7&tMU<w#Ym?$<U=dF{i0C4ns!tI8F$1U
z@iZs>I$*B85~}S96x3EFdpqqdL2v}_m`0%7f9=}S08|l=Dt=L<t6Tvx+?>DyR!3-U
zl<_1mm^F@7Z;FfEu#00@%cbPAT#6S~@ONQ?S_BWlm?quTK7*wcKZ*cU;p0kRq7B&d
z4rXonMo~Q|$?fp)JwSgD%?n@*d0b|qb8{zzi@8!7e{Y$Y<!BlnS~2TuDg&&2nPfvT
zJ>EQPzjU|qIA^oCPT_BkZqxF45`b~T&Gl!;-Lzsi=n&C2@H}m99z&%jMrp508LICt
zx8;b?4BYQ<7$qA6gKS1C?XyX^EWCgz--S#oZ~D?hWPr-wG#r%<0&e*#uz6J%3RAfb
zFliWcx=*qcsiO6y?(!5Uxt^{~-~lEJ$ke&J!Fp1dWAp10jnzucW{bqLSW(oDcY%eP
zVn}zH%HO;-?$h@u2K=SyooA{K2{u?PCb(^Z;kd_2sb%e2DRn^YLdRXZ9?z?)XbbhR
zjP{jYOGv#pd?urRl2Rqs;|HZ*p`}0_BO+EP5HpqpRD6%R$>Ly5-w^Pt8d3IDC9VQq
z*9kY71@S5uISO#QCVo3hZ<T{AA9BkPX!cuDCorNRO=cT9|K0s$QPH6e81>kD^R}w0
z>XQG1TMi=5v{}MOO2AYB_}-ECD5D1TS-w_H<x(Sl2-n$mU7hhz&gY~}-^=5ZW$n(!
z0LluT7pYitG+lyl4cieNKe6RfQ->H;s8Jydx|2foxRLq$S*kkWr0-@(e)koR{gfIT
z0*D4nWAm7c;G{=B{G!15!I3wTBC8}5H-FXJbvT~QL1g0Iv|b@AUy*R^9HJRi@e&fk
z!e-gzjeHRwE9YsDXl7ubu-x@Hw`SUV-s$yp3+SDM<r3da{hkdqvi654DhKK}Skm^3
zx*Cr`CG3Vy57yKArT;t&)!VPz3dx=N=>;tY86VxCkg^Mbq!$!~z<K3fBe;pJ6@T=)
z&P%{GgoIHJ$1SIOl(JRwmzFH2DfkYz<i*L2r(A8J9m>`Klait^+_e(2br~PvFd68f
z$RyejQW9~ZJ!`B6L<Eo{>I7~zndEy@LTPQ_^Stk#f4)Mlnm+uAwz<M}OwJrJ%7GB+
z_BlY2WeqN7dX6E}M^bU(K*$G$u*dM>m7DX69{7f?m?d}5?m!>mb~}Z9p{^*GSu?P4
z7KS6(Y42f@4(twIdWRp(4lJI=0z$-{qVI`f?};Qn0Rj~=%G9gp`XaXaps3i!S_$*l
zKXC<IR~3&|!INNdW9MZQf7Cd%gn_^5sWR|rDoX@6&NcS`gpUN4-I`ql&stV6=<*x|
zv^7Bq%Tb*kk;dD*l5Q|Sq?&=ZCX9g{14%;`#^u4l-0<N+CAoLLt8%{Y0r14+<g_mV
zd3n|bUu@ZI#>ZFQ@elG58YR3sjdNVKVk^sd={Co>ZZUe)q{@wuyFd$1wU85;P)fHL
z=FiXn@y#kpArktlR@O_Pnzrf$n2Xv8lqVTSd;${60#baTz(|&5QOa+k#F|FH0Z41|
z1AqWh%f#}nUw#OeMs5j_?wwqTuCHpbQ)Und=Yh_ZEh7g;<`~Ekz^D6h${**AfPMvb
zCcr+BuuF@HW$>JTKg<ADOZ94w7$(&K@!MS3@eV!=n7Wa+=MC7!hw9(+Aj(8o+5=(*
zw8c#WpZmRv1MPUtJiN$W19~o&Z4p1(SfnA!Zo~22XRR9N0?c1kwIp)8K9x`3uu6V}
z&Jdooyrh=z1}Y#CF%eMYSBgXL@q>K?#-INxE?V2a=!*q8cL;aJvbV&$Y>j4!1(9xD
ze)GYC5d&q5{g_$kQGngy-~dE>ev^d2tYkLA$BG3AB<&`>+-e624pZ!p;;rI8E9b^_
zn6WzVtAh?VEYmvPo7^9~JT}jHUD!8a+g$?!RI+;%2_HEAq(AA>sHoUfZ24C$^+_aI
z8ugXG*5?h=kz0)i>UWmX?5YVyu>fCQ0>KUSF+Peg0rlsu(qCvE0oe-?(`uzRt?EjS
z|E7H57t@<!v81Jz(@ThyiYmgWz~*;Ht0(={PVDR!6OLTlK#8zJncG<+h}WLC9SGWJ
zF~#UkRXc$I6)>s&Nis4O;U<6^R>5-?u8f**QxYXW8!9k*vD}S&J;yx*5a>&L-E}NL
zJoyW(oOLSRroF9}^lAXY52RDcNg|u6sEsnN3FU-E>J&YW&_tr(0eo$K1h=Kt;t;C9
zqN9mDATD+d6P~Gfa)*9z+G;5Iria@!x0at<0HceIfee@Bm52Zt83hAN(yAEnu8M!T
zChgL$#Jip{?@BUK5aOe6Y&=GM(VE}@o;m-9AVgZl2$X02nF-)8tD}33F{S4Tot@t=
z4m-SYxGoFR$0+MA`#wocrc)?X*eq(W_<(@9I*ls7&4BrNB+0Pfd<2K)?T}`y4~!%<
z)><!v;`MIUO-!N=30@x0Yc!cC(YJ7_wr`nF>vU$A^TC$PKbf_RT)7{~{>YC<?><*u
zl=8UIRCz!)`G9G2noM*b`W&(|oyG^=U;+LCX_ovzCOyRKLJXWMlrT0bM!p#v8)o`(
zj<4^%^Y*u39LbZ<CHL<E0m^LE3Y$-D?O)9Nj<XV#AwZQxaf{rzOj97?WDiAs<U-cH
zCyIw7BHcPW+6qx-9%8|F7mJL5UDo3$POiIcnM*J(@zD==C#8XwNv|px$4}lCq9c+c
z7L%Z)Bp_@76vW=3A}|X>WIg2W6JB2J7BP$#ah?>^-*Ra4xO3~z$(UxI%a^Kv{0{xg
zh9gHMZMB=UKG)b+T@)w0#>^YBNa_&McyVqrAr2r|^3&VA2)azcvr*;x#EopZt`X*<
zR~6YC1V11C2id$#e&&Oj0km?R^kvBjXL2>jTCY7c0d~vm4r=MpKN?R0-YTDcO7JvA
zkkDXYWvoO>hpX-M7qYvFI4e%lRFE?HNy45(8Ty#h*Ue$n;1K^hi2vjs$wHwZ(kGos
zaM$8kUi&XVM?@@HUb<#ggV|wF&a3L-4bp2zG*AItjH0uNz1CmI_UA7rzeE0__}L8L
zs4wqIn?Zj*1)v5N#3n(OH#&#G_W%96*Vy2<-}A&HpbAYzE*$+27+t^ZPj-v206Y}Q
zqrHZy>w4lRGyeSC9}FslfeBP&dX1Q2{yh9&4{-er$uhtV`E+)ovwi(efBTJOWEAvt
zT46IzpT9l8^*27L17PjDI4^McuUES!Q+~UKoEIYUs|YGi$g{saz_m9#`xt=`$Ev&g
z2;Z-U!u9)IyA`eQ6(DgTGXD;at3&{n>;wfj$gk1K-(Mpl82}`;KP=<g6Mu@E|9B0^
z8v<CWOrkuz^*5*e^}CXC03h)ifpKGhh1p;6e+7QI$}22QcH_8QrO*EP8`QJ_ATe}{
z>HZxYC$F%ywd_~(FJcNnFVX)e_rJlx{}XcOEC2kt-w<_s>#!e4`m+KF=IKpJG@NUE
z^mhc9Y<5MXUfu9eLv2<S4HwAj$c3V9lYN-FjH!D8ZR@jxxn;EVh>TSk2p1&$CHWio
z7{2b!q2c1%C&b!2;(P0s@1^=JikSZe0irL^$q96mD$?R1pordda(`{49w%tt%wRkF
zmK~O2t`=j=8LGSXp*1Or-x)qVao!Ut1ymh`&mQWomzr;tTBo6Ah?STu&2_+yC@PQg
zGmOK<jJFrXDLp{Vu0ZAPX?@^il^w_A#&4;UUpebPSo4Ob*Odg7t8mQ$+^jnTaOpCT
zIvGpK?!Ik9sxbZ%&paXxD)!S36lL0&XQHF@G$C#cGSST*iyhInRf~RyS;{4WC(F}t
zGYa=}l79G{sX0|E`Jr`*wCOPM`TqOZ-BCsM?-RAggGZ5KA1L$=EhoNCFJ29(`^R%3
z0zqnds?vRxJ{EETd>=1bf`_*m+J$C)cMIzAFyrNVeAb?xex06xY7ckg<f*G=u(LmS
z&}i1t_9Ertu}WTi#ZsW~)?wiKWq0*Ovn@>(C<qF!Pz`ThsxFq5%FfH*|74%>&%)w2
z#}SB`(Z{r$wsWW07_Adw2oty~p(p?(A&h^zS5H*mNl<^{UNcv%X~Zt#%5VCzmXq})
zoEi4<Q-|J=e3XCj=^-nn?8!HnrS*4&3>|M^FQygRzkxvCeLzqdJJEdH5Fa1%E0-oz
z%jqP<tfURty&MVY>}Z(gINMNS-h>66Sf!rWj18w8)Ww`eLbOXf3=rv_`NaH^0s8NB
zDMyM7$j&80;RgFGTB`d>#@RP8AVJTl@37pCPcpffdp(q*NoG5&Z7G?(TDyFU(=eg8
zWuJdB9lDnJ1qy<xnc`A}hMCm>*6btoM%RDFDgtsJplI%!fALb<7XeE{Sm_Or`prr)
zw7paxo%c|w5yfw>M@5bEn{Mj_-j7KEMhk&<fT&Z3Iy`LKuT&fibIotS{O?!!RW)2`
zZgdlZd1$wEm>Ir6&s@(CHM^a7(FN_wxmpRqi$WTm1M~4xNTzv4P``Ob;8;l!epE|0
z_xwPjm<PeXxd_+?P=QT30!A~cpch@C<X1AI!s=@*Ml3bQG@6=GMWFAI94Drrb}XZ$
z?Pr5x;T&`lt@*vr?489Teh>v`O6@p@B)2Q2BKhg7fXk-p<D>Tp*Gz$A28{DcBh}B2
z<xfm)T;(1oqJ#f5ual**a;9JN2II8vhs!;~Nd(m8+rtJIT3X#Tdv+O?^>;WCJr+i_
zzT0xRO@zafM3d|=6(KQWrmJot_hW1KZ%n}e(eM;Cq-0nmjA3+mvAdt1aP3Xs6Y(y_
ze9C{Ex+i_%cEt4kG{^s+b*dcrN&$N=YYF3;o4<XB6O4{3dkmu=`GzJXUF=Cvi@bd1
zL~YOKK1X0Hop1Y$%i(36LZ{EgnP~K8l`Q|>?spW>a#sS)g9i;T?K&NAr|k=jrM$W+
zg4}2`-OWmS(8jwqP$;Ywn66);KlrL^x==6EX>BwbMv{4OKlJ`5f?osZ|DCfw0@5T*
zI>UdNzQ1vy&@$#;cdc#)EQZj$#(Z650*aH`Lr&?K@-}ehd_|I)qc{0dV5wb-`Nhas
zKRD|nFiWYq({doWcC<hzO|{ZYl;f;1Jw%X8>5C#Kd596Q0@q-|UKcM6h^ewFMW%Ov
zH%4X=={_pILHHY8_zSH%p#w)llaptD$;I-_llXL8Ni>%FaJnvqXRWo7%AZTD#uS(*
zs0YqX!A_Tpt($p7{37K*+szHS>Na;Zg#nY*9?N;TOjN4V%cm~p0WXx%2>-WAqJNTZ
zKr&)v>U)He>gy<EAoynfMX?Lht#!QALh6otT2XFQ_tZh>1SN5~ee<Y}FsS>aet*^8
z%!mvn_T<n-59S>rpcJa4`mZDz6et-9BT|^9mnmtmOIH9J5CzU$UhJ;l@Bgw`U$5G`
zYCc{D2_(eR=JhKvH>UA8{lv4mfDda35d>PSOEQfotm;;reG-m!wh;RZbRO=B>Wu4N
zHDmpwUjK%S=hDz%!kRCHD3a*bbfs)-hxwuQSXa7M4@lIQ`!IGKq4vF=H6(jseO{>f
z%+z9{yiKwFOGH7=p1dD~%<`nlrE`7Al4bRl-p3xfk0iy*u9nVr>k;~N#s9_wqOAtR
zgb$arQ9T;}LxvZ&%>yLtuz__DDT>T*>PxL3)M-=ZKAL=KNaU34HC?mGIp3IH$H?h2
zn7wY;G?AxOJfxVdMh!~bx(hav;<8-n$cK>Slbe|+=10k8CbI+W#70pJnijWslZ}NT
zZ}|TWSat!Q?7nR3xl#`<MLh!nrlXwkc!sDNwhUHIeliA%0d282;^VyXL(Vj($L?gJ
z2?xUF%$lXoxH7aR`6tcK1yNK7vb6Z!OGhj8GC!+lmSQ=s*!HnvbYQVJ2yQ7e-g+#}
zo^#co{Ld5!I4Q4o7d&SByBBXQ?yOvRyeCO~kT7(s1c4Yc_`52j7Xm(JhFo$pAjFp9
z`ikbz(_dHXe>SbNNN5xa+^&W9GG?nVgGV!9kjVmycG8sH)#ZMfmIt?N`!A=trrj;4
zTs*ug4%^?K=RhqSHYSV{@9ci;5K9rl{*R#Ie>-dwb*QQac61Y%aAAi15iMYa+{aY=
zi1MOb=zTl`v57hYv8blNnXa#~oChV47vf2~5l}WozK%#G7EGaMQMWGKZwD94keK)F
zg;`8hH%+*%3wD-G!=7{tp1kF$c2nOQ@kn9^!YBSiy^T4e)3G6fC-*^GR}r2M$Eu)j
zjdV+Xl*<tJ%mbSafjeAnAIkQimaNnd8lWcD$w1_UfE~j3-;Vv)*V5^Mk~6zcyQ<8S
zDNkQLaRH1NCKkQ152~z#8snT~qSg`EWSnt2a54j?B2`vbC&NzeVqjcyv2j1BKgU?l
zRtN4*(}2pDMH@34Rd?Rs2^GeJJFPJFLnR*Ao={)Nz~ARR+5SLc>73Xz$)#6&>Rf8v
zzc#aSG1KUY#_<Hd`6d@(drR#@kIO}_y`Fj5sv7}%l@2!9u9u59+1EC4ChU?DTjJ)O
z?ZBHL5GW4*^gQHEvGlc6@h{@?`sG5IJx8?GU2`QD9X*XXANVl)w294*1G_Zh&Fjp)
zX@EZc^>-8&d#aI-NwDSdlm!xZeJYOX;W75>T+h8}6=g~2PZMoEo}39%vYw5061C%{
zynT<;;$db{vD#jcA)SHU*?JRMp0=auOiz#vM}V_&pjhO?g!MRfhDL>r#iZb*RmGK=
z32M(&fVoggTZtrWglYEg^VLN2lyqZch#i=ujK87zXFj-r=yau@qI;M$rb-X)^`<<N
z)a3#;hg$g7o>%g0=P1YkE1JSBc88~lAZV22x->E;6L#Buy&6udC|5kGx2A1B>H}(O
zS6ZrsY_q70F76SZZ$vo=d=ZhZnIRucI)FKdk#JABp1BeCq`0kmSZFKcnnb=)QsANQ
zFjpVft^r25xwCdvfA3pgsuR(&JM(BZ=9fv}H*Lvj_~#k~Kn(BV0TP8d<yC>}+J^~g
z;-(5d%nz@N>?(LJBJjg4K)rLHGPWzcVFE70>^PTu%o!q8oGJDl(aeUI&K>qkVeqGV
zTbIN0d0hs?NtRz)4Bj8FPgAmB5nY<O7Xd>i@qHx$o@Q4W7y9s$e4NZ6t~GlgM}9JX
zuq=e9<OqW)d2TeN0oJ23#bS8GH>$zo_(m(c7)T-fqlWOz0;IEiU@C+gYX3gfw6m08
zXbO<Tare4$*v-cYCQHYlydxSOgOq0A0A7hBk9=qQ&Ki%pCR8~nClEJ(p(cw33{~DF
zB#x~oPB4rbyAgzXwh57|>kr!uVzV45Az>tal4UvI?n}OuM)m1xQf9<d3q<P?g`U8j
zBY8X8P%mf`?{$j)8&Wq=pweq>?d6-NJnc%@DQ1(NtBx<RIjz31aUgpQKb{Y<CCpaJ
zy;~PMy?6mQz;Pn)S<3GzxqObgV6t|S1DqHd<e}UU{Y*OgX|hX+gJ0sdY-byUd}bf>
zJAO&)MZ|y2e;*YKw6w^8K)x3pc4SN|qExF}Pg>AcCFODcgstaC!V$t2G|>a};e73+
z{ERQkxuj|y3t7FrC%l}J+yd<wPyCZ8D=5kQaJ3(eSs^?MEUIbk)kbi5H;9K4$mahn
zH`9c9u&9w=@7MIq-#`L=fg13Nw4O0s#bf^&zibbHPsciaaZ3IdPzEyQ(EvR-VHExw
z+5s|DAAnCAEPY<`{58}0H;)q1LR>5}n~ak)1k7FQz2A`dm?TWSZeqIm$yd#PzRlHz
zlLxRu!%_zXMK`Gaja)&Gh!8s2UBuI)!z-3tV1Cf_5<jUoHJ#$t_+K;4e*;cTCOHB2
z2M<H<-^2Ze`s!h4$7mMi*aJhSbP7)b6c5SQe-n@X3ueHPLMI<O<6rpCP&{lD9^u8<
zvr+z@eYSWA&P=*h+V*Gg{(l+`a_uPXh9F=4^cVi`(8<OgJW~5+m_9-h+fzUlAmSZi
zg95%rm;Z}83J@3B*<2+96e;f6&lSIXrzkOazvC}Q{+s{x?4x;uM2A@_)yo|ie}FY2
zUC-I|H{~>crC=de^oPOg@;Av+B>o+)-#Fk>h3W#8cO&Be>I<~DuVTLMVuSw4?)nYN
zU$?HR9JjtR6o&i_!TyR2x%*d<P;_>(e?>W-lfqYZOtY%Glt0X5PfaRdGkCCO!QiBA
z`0LHYZZemkpN|Nwy;0{>D5?E6?#E%iZZXTbA9<iZMf`DjDwUb=?u=NNdT9((!mW_5
z3Ek~{wHi)m!O4KfN{=BXO4;v};fuqfOP=CM;0`A4x_q96<fW7YO!mD5)_IuU{38fW
z5LryWEKl!1uEKcE>x%XXr=zqC<y^D=G*fZgg;_r$Bm8OCGw+k5a1j|WjC5(ky#`i~
z#wlgAJnJE7xCDc@Z7AgjpY&r#(=ZtNvWD>Qc&Qf55Ff==9c_EC5z;HSa=m}~0)CL^
zC!Nb*1Pdb{i)$D!lQ7bQ#g%~iml!ej<_#6|l?N5`;t2&IXFu*L6iYaBOjVPaIPbk5
zzx$<WNBFvg15*Nc1t%>B{#IUrEA*!9w&U@Rp-ewbP^UzT17ok-rDHqsQsl5$l{wS-
zs*3Ga>?25Lr}W;YgJ9gFBq!?Wv%b%Ie6~xgHP42}&G<?VGJ!<)P!F$9Y=y}I>E6~E
zeh}~Ia+kW%;YeDBsl|w?Wo+(vX>5~JRO}ZCucv*ZTk+c-c7xD_v690VuML34b|(L4
z<IE~~8sBz{*z_hEWv_}L!UZOh+%I8yubE<g%mOf*fVMGu%r0_uPPq(9sHtGxGVlL}
zl7UPQOJpC;(Wn@^cw*kiQclZzQEl!L>_)z<q*nH{RfFSuXZ4pqg)pr=LNE?yH*80E
z7W;iFN-$2~voOvd{OLI<az$wKPP)v$6WWM5a%~dd^l=?XhAbAunVgy~SySl}hq>5M
zCFI>*8>8c{4js_2ZabkaIk`XC#X^QVr?8v|-o@CmzW6?mxCbkn7B6F8s8C7kZv$eV
zD~4(%+$=hU+Yu@(z7uory|~T~aRjoA&*$z})akF<6uV!b7LS3{jfy-((#$YFw-+ta
zcNQ7Y!)zdtX_o>n6;veJbs*T<l{I*2@h!LLBZF4a1T0@UH*oX;A70EyIaZ5izU`n8
zRLMq0$#TAnY|GC5d^}J~ryMG-!7+MNcU}DLnyN9-OA;<?$^BYHPp?s>RIHEE$IA?D
zx}cwCGBwjo2)13vj&S`6kH8D){4rJUL-$)PXgl}e-BWH~CFBjT?h1tCi*%_9$stGt
z24d9vBUmf3q$cMf*w;f*2^!V}@GFZzxbj3S5AMi6N01Loza!`MJS^U~jeR?jPvOZR
zKYv(aPe1(3MMMQ?b;w@JR@GKHinV?HQxi)v1Il(a_s);N<DiC6@tY47%apqt?*U_C
z>UEt?_y?ESCnWasEGI4gIs%4*Zc7XhYhk|ps#Cmx_wV;c^XLycqo>gN_PuLl1t97h
z1W4UaRZUJ%aC0~)*_h!cTvbV+)E*n0)z#JAwROw0PNg`UCL3E@x5HjS7-x#WSmj-!
z@`r7^Ny;Oer`;(DUX(Aw42n$7*L&ea$fhxU_nj#NO}p#P1t9@xS*)Z4sh)gD-Mp%#
zoLSrK(%y23F2+Th>thw+8tfbQlQyYyjf6zxJxNJ#zS01P-nGSna)>8q9DVnr^swhL
zUNeR-VFS7-Bzg}mVOH=tjVxqKjcDuz+7HEyn}PV0lAw0yDdj@!cUQS%2XBfWqU;8O
zHo()a{ICDWDtY1(_b^vk)|@;h6DqWr6stI&6UH+yQiKX-m2^jLjuv<4nfktM=!`!8
zWb9!vb)W8Rul*^vCwk>{hG5x?B5lrhoVn^FmqOYfTte*oFvQQ0U#$>tF}twI+~hq|
zJVQC@@2-6u*QEkMMYLNR>Ea9O-eucsI^U)r9ue5K?@p+bn{r%6N-CUY(s+D24}C;Y
zp{}c^r@tCf$E~?LP>!)x1ow&(=*6QoFn11~62gT+!=j>?c6NM_)Fk=JP&tg0{k4^I
zBF4opHA%Tl<i<3m?m9dV$+L+)(cCS)$XIAu_}stfw^-Ls7<~x?4O*41qZK<dmqfHo
z)toKlv4-$LT(yW9RV~=Fj_rtw@se;K)r71l8@-!S$sCYQ39|dC(R3KT?s=kjqdld3
zx8s3WZubMA$rD%F2@*{Hhqg(s#59^|3>wyAayk0cTI+UZum)jweW_&2VK!aZe`pLw
zg>O#PlaG~!J#eP5s7iDPr!tDpPUdw`cAxfLI-YIU?9pt0VT9c#Av8W?V=s2~LXY=6
zZ93PJty!?WiM{XWG{*ta6lpQW_>i4<$E-t*LF+WUnwN)Js>vK)uBKLM1j_7ti(#It
z#(9L%SCBw`9u9%%2^d|70@YHQ-&(6-#xx3IPGi+5)5K%uy&owTNAx6#BdibKURp?e
z92nrhhwetZ@@(_1e?dSA33d~vps<BtW763rv)MJqB!3)$a-6ZxH*|jz-J|JSodI(q
ziM|8}=N!A0By3hhKYBl9dy%YF98k@2=~3!2xMptytjn!5c#)c#I+12d)i&NQ2p*=a
z+$Q?0z&EOCi15=|qT&W<_PguXJz=ZMvPOoV3W^(WP;b9<fS6uY_yK09FILX6d5Lt~
zZk~q2U=wGSkT)*W?Ay5|5uwu}shRu)7@NQfsCIk5*H`L~J{?vTC*-#TiyP^p?r{Hz
z5k9p~5*ij}%qCbp2TJ&EA<gcIFb3^xu~*TYbEiuR58p)9Usza({61n>#I`X~*bywa
z*7_|LYXfyp79JiPncj$8bqZ<I+F1`&0;?`Ntxtx;>)3p=uAv^=TdcDfu|y<}mnOS?
zNxZhDy6*)F?%Aat+mDM+4syaGisTJV6))@R%Sy343062Ie{{sZ9|8NGTu+caoD(j>
zTT<3Sw=IY>$=ANcYq(Xm`_jFqL6kV&96L2F%`70)FwtcY<n3sEPQbI;m}ZcgiTr*D
zK~~y7I+L=htcR}i!a0SE!*(ESQi||-q05n0EF~4XsJE@r4?x~PtOWq<e+6OBZTRCK
z{13pCHW%<FK1J%@{X={xntB2YWnyh8!Z&zKl>KhgJZP%l93T6D%t|BT)6Z7yT4yz6
z?s*!&6L3<GTIrsP(PLFxaH+S68hf&1#<y`Z+&cL}*nGh+$vrY(TKU~BbXq0%HX}J+
zmL}i%eW$HO_~`C!%bKIGhcUavnfrpZJw4cDaE@5k+K=I9<n)u36pyjV2Ix7OSGh-#
zUkyk~LbFeaonKNlIlPamsJ}}pWK*Av(8aWs9reWuiEd)5kC{Q^+?%i5V_G0}k#Xp}
zP;39(VQY->U~R~{EZRW0BRd}4j={E<R(-ac-9}`c>`Cx4xpjj_dt5<+4P<Ry$VM&L
z(UyjDp@13>{^%-TSZj2MX$hn+q7#Aj2Oj-X(N){!^^fG9AsJ?L$1G}10o#)FccH6W
zmBl^v`c%DCSeeE|?>}3@?3A5;r_n1yrG55x8L4nP-$`KRB-tMx(DSFoDd#0rUxrD2
zQ=k2Zyp~Hq0jdgVNqfHcW@?WyV8dArxS`0+y)`Z)l>82`SKmEk1(?;#HVdL{8lFUt
zGLH@CPU}mWR2!X<4&*;73#k^XMj!xtpJYx7=O&1rMQe7o-y<I%H}9$*qa+DaX>K4<
z=rdgw;cE|QA*Y^ptCQ|}I{z>!2iMx#dN+hDFJfEfh8Nk_H*CUClUSAZhFF&|b;?%C
zb<UB#yM7Kl1Fo}_%Vv)c)}v|myh>YJU+};u`jNJxA&}^cZx#&VVu4QD#1RK8qc!i1
zsQgf08J#UveL55vi?liKZ#<{qmf1FC87sqzuB@kSIiS;UPVdkmUo;IrXlyG=0dt*S
zu=URPe-b=5lbsEKS2H+FI@dHpsP35G4!hl#ZK&cRh(9+!iFXg2ww$y{Q{~SU`mH!X
z^oR7l;i-%4D1bYAA~a$1_B8w3o-ixiTL=FKVBZSe^DJ`vrfuZF`-j;^br;2*%YY>C
zT=m3HYPU+2$wtH77|p>7mxzP?>y?g3SX~V0UvN+;eWo7O_Izi+L)+S$b|1_2k-G4@
zGHmf<J<?|RCLg#xO;39tBcg!p{Bkt<<J{rEmq`8SC+cs<&m~>TFPa$6VaQAq4#P4u
z;qNTlFeeSVfqvkzh@*|EIKtV^{xib3)sOd_hLg!?q@5cW&*Y8Xu#yueI%G^bPkrs*
zh`x+-`&PH;h)CB?Dqd)QcI%`=AE$Tdi-D~KCz$`1U?=}O)Q5Lp?4cGz53xAtx~tVr
z4b<FG(Xm6hXGf+4`m1-k{FaUuQMh`Z<r^;e>l{|z;!pR!>te^ur_~Yt!NvU}w=iuS
zhN#D~1?7oDh2wXgzEmd#Yqrc4YdcL3wGyK>+v>5Ba9|nwr0Pn`=&&IR{SPrBM2$gk
zl&GDA5#PM$`9Qu>B@w3)l_B~i{dd|#9K@=gKcnw*ML<8fB&RO&AINY-7C6@rZaZaZ
z=~viK4L)1xPc1WD4T9v=rLG_>j`U@Vs|2u&QfWDD%B;Rw8fL&dUA&OWmkyWlxHK>{
z>;K}T;W%|eUF=!i+aV!zn8r>^We}T3{ZVQ?fkTLs25O>x;6DE)EhM2HuK4QJVU^~~
zfDamUtpm9C3_{X-*zib!4j~!GO-3#j-H}D&k&o1)qlsU!ixmK0CVe`frFDauc;s!E
zYEpOcCniJ~m&2t49x<6ISMD)C^T>0RTPEYL@~&*fz`&2aI8Tic<%$eC`MIQYpLy}g
zSiG#aUyHspFvhCYoVx{R3=6PovB%wk(ny$Oq3IDtxR+xwbRUb#%Hc6c2Y#t|vQV5q
zaXXG~K?njuu$HVdY7QUEaP@x_+z7(34z>oYs19IQuo=?*E35GQe0{rb5;*|MwaBBK
zt3_`*Y#Bj9>CUtq!}_uxSR9fz&(WjnLnr<!^JUTbUX*FVJ%N0A`7jG*tZm;H{mivW
zrqvMDt0CaOF~DQ#x69s7)e>=b?>Y+5((1MD3Hw|a`UU)Sl_hTj^;oZBc+h{x+7t1w
zblSKd==+5H0*&~q9HPcbOeMyL81;kcly;^*co_vdAFQ?dn=(hkYrn@adZ(z~hDP{t
zV(0mDP!lh8X&ZSHO}WAGPh7i(ZPG~cg)yf}O^>u3p4_`P#pJ^NYJ|CT@-|v&!A6Vr
z2JsGW$8CO`XuyB$wLLp18{BFvEMx_%vMafowtKIjRj>;tnzxJFVZ_Wn6a3Efg&f>k
zznjO}Z+3Hx1AP%B`&H_th42k=RjC+;V{oC%YZ&Gn2c!0E#L@BTxagABf<d!Y=~><7
z?z-e?n0E66M*c}uZP&{tSFROdB>n17ug{N|H%}!V-F<MdK1g`*MWrnG{HSSoVVE%6
z6+GN`mvkbvo^sWAwa8+f#?}(ES%clo|AW(tv4o6NxN7mT1iaSG?5i{wT$-MdQDMu2
zKt@Vc%sj2dwYDAY(et4SAL}Ov5n_FJ4Rt_r6lk@bIr&=cv3!yo2+wYu{fUDaATfLU
zt~luTcn6gkGKcrOj4j45#+6!+^qNRA4<o(#po}J2Mq+`1Mrs1W>}Y^#+40CfRee}w
zvFJUjgoZ|rzDtMILejg-g9vv|k9Tb*DF(tEp+a>%LpIo#n;|7+X3?O~9%;O>YHe`z
z0dwI7vgmpLVnHxx|2vTF9h=Ko8mw<2oa^t{H3YM45&1ci=jO1wS?h5WJ~`xD;87AU
z-`sN=6heA5@_?Pa0|uVGNorI?WAybcXr~9KMexEkFMbT}ST*XB5O*&HDuYBFKXWrP
z$FJBj#KNy_fXi@Xa9f(H6~!C1Jg@7{_Bv$SeMR#?b;z{JZD87b=vET8fky9FU&`qG
zP)lN_l3V$*F#6m%!64KZMsG!vhf>_gAh7wdY^Eq<OAjg)EpRQwUVrgR3EP&Bgb%o%
zjmcr^C}v#Hb;YM>Tz{=6LzQa^!qYLi#IVAj{<A7k=^kKeCh~TpIo$%xBJAX)ljO&j
zrj4wBVgaa6x~(A#`rT2YVnCL}d~4P2WP~d!p+}D>E%o>*d>5V$6h)=3D%KsZk*U8-
z(ev{*!xc@|GXdYgc8@tbc&0F=7iPz$y|J#XASy0iq3z0dX1o+2$YoiWjaFj3Y6lSt
zG;^q6proqgTVx){R#%VXGSwZPo1Zs}4H90C2jvS(SzNhI7X2HfFbHQ@xa`_IyN}jU
zgPujCKECuTMVFZ=`MHRP@<|RE9vke6jL^mwBpi4P$f@TeftvfX<-@jWdEY(?ZI>eX
zoaq7yi)w=p-K+_b`};RfnvkvUf?uFzW-I2CQuU{3s*~7teeQcGs4faDO4997-*zV`
zpxL{ZF!nx*tC#I=(-VF_0r8_E5e~lQgfcCaiSN2)o1<0T7Aj=0YT=(XieJKoB(_hS
z1aS8fYxfaaFOQ;@krHvvJsvp<d?(f+|3eJnqSpd8XFZpJ*dK~KOV=}hi4kg8V6YCZ
zSX|0p7zmdw(DoO++?UEYf5uK?-*1(~9)EnQUZg+T%{DVoThy!{*TZlV0v~%U`_v{H
zXiPf?OO)~Dx>a#@rOLh)R@9Fe*B(N_?z`hSMAug;gC!v)6K&nS3@noYGDGP)4X126
zed^$LPs7CfHO>SPEj`SmnYB;)^>dDeuk65h>$Ly_erWe`rM*c9wFqi^<M4FpMSk^R
z)PxQjzax?uM^8G3j5jk;tTgE-b8U~}tPvqhSY_`BK7)MO%OZ{}?R;o|w24L?euy;^
z9WVLBOXzv)tmZ9iZ|i-5O-^d9p}cWgl8gz-gt0bYgf+Vx)ND!-4<A2}|K<B7ndika
z+%n!u`wq&*@i%$k44LX2H(qS#_9Ghb?Vy%%&pZBUXtv_doB%TY&AFu~@LyWCKZ5|g
zh(2{B=n1yV(m!eX86q9snJD`!qyA3p01PK>P&cb8G1U0;Eyjx(b6{&tp)l?qkt?%G
zOdpou#V$m17>`u^Xls_-Bsv;?4jCVw(kHc21A0)3X<1ooTdcSu3PPizwosW?PH+Hk
zpKDKem?^=Gs;OB9vdKVtqc6w!0hf!9JFv>zFKM$BNAToXr9FS+r1CeN_Y)Hx5j?SQ
znC^+&A}Rj6K1HaRq_VQ`Qcz4x%>APS3psYi$XO0-f*nh>Qr)2qr_!^QB;^-kV!m|U
zw}E7Ko`cb}GHGn`$3U|<`MKtGU&)BJO4hx3=I87tP?OP4#x&sJdh^q-6J4B&F5G%w
zJ$`g|czih`3bKAY6V6k*9j7Cbt~JTKb*Pr1{{eJjHCCNpSLR8eme@32XFF_R=#L%T
zOGN?)uWRC*A*&x-8vaSTDhP=<jsoaI+XCHa^&j<rhE}LJklT|Km^ku6ivMVj@npe2
zK0GtnBBXTaybX2&mK_-6V<h)1!9~js6C5YxPw!WOw+n@dx`_9S>one+)xeFY>f<Y>
z$CzL@dn%P|9c)1$;bfuY9E7c}x<njrzZ&yb5$i*&PC29+4T!OAloy*Y6gH!eO)p<F
zkJ5zB_1zess9)5$<J{(S<tVgKd!tD{fBsHZ=(#bH7*r#$pXm<A*N6u^r9Is|hKUSn
zh$0YFAET(yFBWmw&-$10G!FbOdf$W;K{aXgntX#i#wab5WmK1rPok^8Qca2@y}0I0
zej)}OoA1RUs|Sc73s)5QwT@)}XFBSM6c8JYvlpx&{hh(TY0YFiO{*ed>#qPoE=_lb
z=FB{)<eT_$74>beSdZ3yfR&IfBRgu9_3F(cd>*QWulDoDpyalSvI^_v$xu-fwYW`Y
zOI)9=r_&KoD>W_!vHNm-sfH?Cdi9C?E(YVJ<hx5%lx+z|0*^`ktzTpwg3GRq^ys}^
zFM;#rSfh5kr$>GIXY3B{iLvlIhMH2!+_om#Y9>?mYc}#E59}cQ938&l)X&eH!P^^V
zkN3YUvd1r#Se<@bRqAlcQLmGzJ+b)gGHR|~W)-daXZa7UD&XZM;z(XHVg%g=B16q?
za}0klTk;Zm1oN^TtinAH`46dQ68iRXu?`i-sNd`s>B5!K(@5vj3uxVsX6<yxT;#*W
z;>0K5*;lIi?N0IRzzCO}E;R56amGV`kZw=LC@sj9dz@)qQ0>9+_>kpse;&=lMwhD1
z-FexXhlSQo29ICyd!kQyR>HpwnwqOXlo=n=(e;AbteTCYP%|}o^^-Al{ckgd$@5i2
zVQ2K1$Fa`#-CJ(wKN8gOtL4&*x@-U3K|;Ed-g<7MZtiHXOE*@VEA|F{CiMwyecwf!
zm{D^pZ_VCV;(e;xQ3FmSon^HpmuXZ`4Gr1zqKaB8qd58JfX8IE+nZ^tKeW7;UW%b>
z?Vd?CkguXYBl!|szb(f7+B)c`6=V$fZ2$#v)IA!szuAp8c+h2~^~amIOh8EOT}Q$a
z@9(fgM*@@svt9n~h@SK4VwrRXYG#7V;peVYY;G%+&SxPO3FcIyAmykWPVJd8%i(6!
zC$5H)K1Oty*lNcc?@vmfdv6Vw+DqZ%hSJ8B+~?&Ti%2<cV_GjgJy=_NFWc{QkXg<N
z#Dg97-u9o3Eta3pU}^U5-XTTDyB!v-xKSW-D{5Nh@?M9>rA)lp9ia1)2u79iiGAPo
z$?>ydQb%^l6%82ViFL8D*=VKAem*47#5w3p+BhO6=FW*}2j|KDUd2`qul>i##Z;n!
z9^;yjjU@(DR~hLSR-5*3qmHcF3mw_rVlW>Vn1r2xD#INT*<^H4W#i9FA2yz$wsRve
zo$?jv^n;S<5~$FVx^B>S$k2ohnhuRJs`)N8$S#c-Y+s7+udQvxCM1||yJ(o@q#No-
z99J!FwIcrTgK@=vkNw~t6DI<k%Je6rq7wkdru>_FygxpQEIc!FXukMC;jFz~$-|#X
zMY*3js!n^~n?D|8v+@S<9@LmKfcTMgSe_<t8%6l!Okt*0#A|%qWNl?r(}}EFH~w@z
zzB806ce=|1uJ?;Z3cAnWUYil&!R!cVVVp1KE^q{j^$wDxn0bfY+XN3_Bo;}<Rv>rW
zcg1=6q^)ut@zJHZ>JF`A2V|Q?M@bF0R;l9XQm>3GVZ1RQSZ>L#V8nA9Ti4Nkmv2fF
zlaVq`duV=N;UyH|9`Q|H5g_`hY%~!xjBVd-SYP&j(c?sw>N%!3EfA4CqEYbkST`tg
zG@22SFbAAsYV4iSOCb8Ouj-(@Rk`VGSlF4r2JghzYp+g%-{N!Fib^ONuN%)ZP5$6P
z7xNYg$!?3aF65<<uf@<?5=QcQ<Xofg;&A-4au-mnUd>bMxT-A%K=n+hE)f2(a<Yhi
z2m8!fcI%#JzIgHzz(@Yv^Rrz<F%PhUBG+NPzXvDKu{;qn4|4a{rDdd|fdy7$H-gio
zao_dtW9^@bhd{H{;zZb)9gCTbXfVx7tNrNEJ{q;##P`9u)tS<cZck2Zps10i0cEwN
z1I;Hi+$li$*Bp!}#kMh~lR)i$`d$jT{a_|G+b9bM);ms^28&J~L}1+hllHVgA1tEu
zg@nYxD0qq|v&{~_vCyzvE6II35G2u~=fGQ@usLMA2Y*{6-*5`9(#k%XIKXnbp!>3V
z;D|)GFLhDcJO-If%W^Y)R)<<_^f_i13iB{$RI>6x-Ew`pE@kW?XUbFC9oE-XZPLME
zAW(a3;(6C&BW#Y51fQ7r5n~*eeaFzs>x+x0U*3J7<PG-yk(*U4y>S(9S3KO-5TyuN
z_6{5{u=vAX6iPltfA0EyR=naeGfkWAxD;I-e`0(&c;_rlbkm_Mmi+W+-O?dDQgK(C
z-g()O<O(BHPf-pRJv{GPz&#pH;0dS1t2{gvMfPV*##2xo$4k7sGfI}Dm6H)#{^1J?
zY6ymX%-QU0Y%)B-wCY`Hs{PFfEhuZ!?DeF!OYRTiHtUVaWa#|dXw=`KvRzpm5tFoz
zw&^68(1*U9jAgSC5k2#n8`IiZ(_n(VZVUVuvoN)pn?Y=z*d?%Y!omF*z}h0Jalg1Q
zvuO7-VnNN-_bgfGY)_>YlzdkqjwGEyWO+lX&1E^LXSza*CbR+W;5@v=zqPGWeUdPY
zV`uzBRJwY~K*}W2Q0-wZ73fD7@kjjH-$Q7(3{9U>-YsY9Mnkh$5k%@nyS)EN8}Kvt
zl|TG$ngCe}#il*s0)|(wb_KGTHeDqoBoO>wCb9N!4jZOD>lQyrLA|_iTCcT6zto!u
z=%~Q$JOOI0mu22?U17j85C{`^2?Yk8z;v&gxPOEf+8@agQbynzY1X<l{gCEY)d-4E
zA*H93$m?;yZ$R|*M}65zfAGl9@BaBK*Z#_?093)igiA-{Psi<t2KhlHKQlDI8$u*k
z72yx=H2ko5|L0SZm#$h+Y_}Ozev;hvAN!977+iG}et2{9XWsGJLI2|#o^kX*QW92E
zrU5MZ{+~bl`@0eHRFxm*N!;HezK>>dFnm3y3g+7*cDCET&kw`F{-5{w0sY6afYF6n
z3PyZHQ_TybFy1KwP^62)52)&368Doge0Il#B7u{iQaP6v7GAu3nfF{=oY2KTC&z5G
z#HEbGOs;Ok+Uzf#-T!f~tCIlK0{~9gK@5qO!!dQh6Td0G<@h`q(`_(0Fz{y9NHmZa
zG#_Ii$FW#W?EQcv7=9!eLp`75(z$uIlc_C=^a!Y9-Z;$FC=v73bbwT|FNzwErH74c
zkNJ4;pMDs4nMci~<zQg!@VUoYFUd*u<43&3I342XGj5<&KlI%@M+r%3hNHV1|E5L2
z37{Xvw+F`o3QZX9RvFu)jQP?n1u0IW)X=)DKJu4DSeg5N4osRuN!GFguNN-6SUZ4K
zgMpDY{kcbNkbJ=_@r2DSs^(nz!F|S>#Bu1efF}VNr$y;J8J7vJRozG4;-X?vkMr?I
z#kCQi1TQzHHRqhue6jU`AY&AFUKYYC2}!CTTAxL&{Re%Bsq;d;N4v1O+KF_D&U_M^
z)0m5Zn}H^e%hxuWk$Y&-_FT<B^Xh+3Gz9l>Ri6t^qEXFYyY}3`#CE>J#kno+{U|;u
z_13J}a3|0mbg<#DdG}=i&ZK2yR<G!$yU>OB{!<O-gO$h3HEf?+oib{khyck&{Z^Fa
z;}4?+tFk;$1+;(<H@ET!aXLK0e1HnP-TP{~>2`voxNOVWC<4nWg!Hn40%B`Jxs1^o
zPT`%GyQWDBBNe7)(n%+u8=D(nq_(NuUJ=jN8``KqC|f;VP%Cj=(*NspiT^^!Ghcxs
zed{;wK8N5Ui51^BQL|QF^id3FZH&aJEq#}7%E<C6CZbl#kH2{ePz&sHt9_aC*$H2~
zbP<zy=33pI0^~v-ZjP7itYQrXYAq*}P06jT3LHC@8h!QikOG2pDAP=b+qcT|NFKbB
z{)`X5;s4fK#t=w`NQO6KyK!=v$xS2IKYiF;%Pp~00cWi^@zr4i+`Y38y{~9`d*ub(
zy4rzi76wf|%2(#?3Eay~^#91tateSH%KFNeEsUeKBvLGioMt@JIA_q|&8k#rdsnwe
zGTvQM+a)I$t!$6vY<$}^9f+y*!PDnmPwhq7#i6F-Y8?J4Ss5hE=ib<Xr0TaX&JI7d
zdL$;uefbQO@maJ)g3mV}<YPISlRmWWKTwgJZ^m9&g})31lD8FeeOQuD(HK+MEbE^+
zozJ>R-tD+?O+f#R#Pa+-<k<4I*r><*(&#fzCWE-D*W>sNB7Cg`&~p9?73PJP$EDZy
zX5s1G!H!yYPM5^1z;uyY%gbH^bHfDdqRva*TFou+aY|p74$^3$ZM50x9`ktFNq+93
z5`47SA>Rm?x)=qc=Hyfcth6OmG9x%(V3zA~Sx!BsJ3H(co@OX=%T<3;Hb03V{JO{1
z&2p^cQ{3vb>{HX0i@m6Iq~LBCEYjcK|Lr)U{Vg8&-GGqy;h!yQnWJwCl=XxvWe>M&
ztCv<>m=5KYb(|e#R?>7egJ7f`0JHIASKogtgiX)`uU&|NcRi1qtz&-@>M3=-=}8i0
zpWjI}@q&I$-$@{gR(0vm2#ZkiKH}gapgavbTaf_0Hz%~;UXYKfU}UT10;Z>J>rGQP
z?A;)|%uPjlbRH62<WVQ>WLqhF#7wVTxqdlr=Z(r|Wc7`?ZYc<|-}hQVTH5MJb+3iD
zLVpCd+m0ps;+dBvUFf1<eyATM*>dpUTLkki`y?kdK@z8VB_}r;vX*psjd`@xUWlzg
zQ_uyFXLZ_3h+Vp)LaeHmmUZC)^iNfs9I~?YYDA;@MU(=bk+DseY9{KruH9*;>|%p#
zT_{ZuX1ljhK_h*B%`E$cwtCs94W%A@8!+*d0ka5-vhOXA-D9JssGq*D6myW(hi1le
zwi|d||J4J-low8?CdMB>DDYZF%IEI5CY}^=Jij;#WKpP4x*Gz!3sI)_09i^aC4J*9
zsswjELf35d`jmlT6+9bN>n0iR2sq@<lQ9olxgj#nEaoe%kRXMG%~ZQ9#<Lakz&s$p
zLK7$4)K}ZA_;I=SE-?}Fg*-3RbRC$U3IT6Sn4g_{)7q)m;h&9AY7T+n4`d}a7u48n
zo@;%74$K9wi(^d7n(=2|c6$J1E}sL7!PX~5UvlS_byvazqb3|3njiJmx~cPQ$M;`w
z5m#NduUo!oI6X*r+RFeoZnw4qBK-UzPyw>Vo>5-ML}|4q!M2u>?bdvjUoEgmuBz{z
z4}?JEicwNi?^DzBsqaVrVmT;1kkgzB-*c6Fa+H+zGZunQHA9#Mm||15x)pOk+hZ2D
zI$v>FgkRKJ&sR+Ishb2ZseHJke1feJA31}WTeW5*&U!WpM~jMz7VA9(8JDr2irnJ#
zM0E}Z>I5R`^C+m(wyRZnDaU&Gt1}8u9EYNdfPj^RY<FmR`D+JA)Va|VG@wdr%11AR
z=h;@J`SY)nXZvIi8~kpCJ#ewdDzC|U343V08Y&DFM};p8xVK2<s!51qAFOKcoSCSb
z%y+3<qQ|IyBRRjAqg)uOqG0<ZciNH51)q%I46x?hxgv`QT3kXx>eyISycQ2WNvQ61
zN+G09-flanaZ=(+#QAc1%j$Y_W}`bun>-PxoH+8TI(m_NJk3bx9T#mX9*$;)$fr-4
z9Y839s_JMXlI+sG!u51ir7uGX%W1dwOuwD1e{qd$E0mtON-nQ_-!&j0AWYl+WD-?=
zMHs3fo=>4UId48*s}RR&ypX3-=F`E%D?8-`*K>mDJC$#(bx7Jf=f8CC0F82gg6}j1
zZ`;~2JIzg}8SEp4bO7O!#a@p|M(njS5bS%>8|`-Y(NZ@*4}fMsC&~BT3c(0_2x^z$
z56ARVdvMHunBKot*~$U)&k~<D>q9$TlC(|?jvNKDDqXt6u9=_TiL=(0k>GneHM&<?
z;q`PV#0@rfVACDb^3VY8?zB67Bnjv-7i_hUb06!M+fx#FTQ`rMGE_;U29Aqq(F2u@
zk&N&#^H)908qm&1p}H->34P_(A`Zd34Cx!PKuOBG-3_%lfwqi@R{uJjNbwc{W}vc6
zDUOR?z{MB!dFP3V=Y(rR!CUH==Qxhi+ljmFd$O&qd|T0Em)2z;$j*|DFd|dOT8fuq
zHYjWZ$2JZr?e%dSVo{KBvtki@dwYwz&yPZWCG$t2iD5bdWD<Fm_^pC$-DpbgSlrh&
zdwS(lN-O%u;($5YZhdP^KE97i7}CSu>$OeP!3cMi+m||JbqI;zEkW{-OICf7A?+$w
zRBS8F(nR63H;JP^5sm`{03ZvKtCSrjDw=Em$KF@|Rkf{cOG=2eY`R-ex=UKR1qr3Q
zyO)57DBTUx-5t{1AS}8|deQmLwcTgy<Fn6s&mZu7;kSSRbIvixxW^sWb=^btz$s#H
z8|XK8TK?3u7Xxi!+NN1byTRJ8@py(%eE$weIazW(x`zR3D%4&I++8iRCbzpVw6jYA
zLf@?ESi)YgX;fN+=VJJe;xpr-)tM4|A@^>tBt8)_O)WUaHs0)hQLd5S^Ys8AYwUGL
zSB!Ah4!j(UJ%A_V1jTfK2pBN%@!zT-<(L>MFMN5wV8D#}_v5Cyqi%WF$;}oV3>MTM
zc`8o2_^$r+;rttfo!kYH&OMYgrxeJ<!?K#P_yFe|Xe;v;@WtQRM|Bz(A4LKML$}a%
zXOgck{M?<4E=I_!4FhYM6}tUqd&Th{oy6BEwIe#~_z?njkWrxCiG`Y<px=O|!Y?L!
zq!2YB{e>T!Joi3bs+Ww`rBUb0c=k&xv-$iwrkXM#j>(wEm+Mo};`{laCx9D)GoxE#
zO$;Q-_65q9X2tjGT|_FBb{8rn1+Pv}YdrSqKRbiBW#r^4z+6}70_uW^0#n4BI^4|)
zHhF4{y|YHKlcE($jb|+`RDyy-XIWWUvr|%$&5LF`B&4x5^V<-g&n!M!@}dtw2~25O
z0_*`co<KS(I~!X{sltX!<L|@=nW7xRNmFeztxKPU>ZIARr=r?15#HZ(`TztA+5vuO
zEAdpV^1)({y6@p(|HE0|z|Q`d;5(rI>6sa$jA8(yyy>{_BP-t6@ubV7(>IOgU;Xb~
zXKxS|`@Qbnm5m#DT4(o|X{l*u>`AZE-^Fx}SfV8muapN5MWzC#!#I*VV{>``BIC)}
z*tnH*u!6A{y-hO&l;k=$YX9<nb?-1L>el`=8m~7#=N!G(Bh}l_^2w(^P3ZjgT$w>E
za3biKRP|_})lQrB#c*4JLxp5A=$Xbcl|)A$;Q_|FLTJeOz@nx&5aChg2n&HM%W+r1
z+_F}=*ydy*Av@re2_u}eWX;&iNsw5Tm5r8rZIa|ACz;-%6q9ZBS+9})>bK=S2PGvD
zfo<;6GDMo*eTr~{!HqtDGz^sQ=XGiRF7E@OnJnz&@@1MV?Z8qRw2X`+u_&GN2HEBe
z$pq}SXU59ST?By8fGV28Q2zA}T9Pf_;5$Sm?(V}gjgI36Ho+JJ_wy)!{|n!7y^ER(
z&#Q+?3kuoIm(ALq%fj(BZ_-aVThSH@F&0`AL{<_HVtSJ5Pz}#w^=9Rg(ZAEV%yYIO
zKIJx;vbh;I%rRXqjJd7?AW25YjQ8|*_q6wrh(gVV1tv%t(1DE*K7C#OapiXGm!XH`
z1Y|2rN9ka?>?Q%x1^0t?Ec7qy8qky00pVCMZ4k9k>ba)nSypDkn~!Z^YxCj<zeeoB
zlPFsrwqDQn;T5=tkJaqdfs|1IZ}%^OEKow>6(Ge)!j-{=irF9_Tcr8jeM;q?M3laG
zDDFDd)GwYK@~3n1>YdI|yGNs*&ui-T;-<;jb!Hy`5Z18!bj}^ZVZZQ|ElKU#{meIV
z{W4ZBwnPs3GuSwlk-6!sNrTJ!a`!!>L$lVK3I%@KVW#RriT;>D&v$g}d>!7Ab?HXI
z#tS-(mVoM^PYvi1fi25k5&jwXedD@&-C_Cntb-#-nMLdn`t!1?iQc7tn^G{FydZ#p
zYp_Y3$V=WG;+mUVbQt0X>r4fl8+)7gwck*kFRl`HgolMKv4!9Q#@Y>5{%XPh;uW~e
zqJ%2`5vNj$vM|TN$2y8%mQjA7u(NXrnsKonS6yGfJ2wSCDK{429RGcQE7rkI9$lu{
zBeb2H0)g{k(>VZQR3?@=?ssx!Q{)EWEIh?^*COCfFQV|H-Gw>6Hyf`HkaY!4lJEB^
zuHUjFTCCQptW%jS7aGO)2_U(Ea^cq;Jq`A2e?*nJ`$HAZ2P?G(Ax{Bt-ycWk>Go4a
zT(w!xH+-&}Dtf)<Ywaok%)VjOtHXJHmqeGh%PXW*{qPK~ue%@bWs2O^FBI#?RwZ@A
zZ8Sc^_bgU;9*AgXDjgC?;sq3Azlb(ecMhpcbOEj(W8&S7Og*W09C7*E_Ecyg(8U$-
zaf$=R!SzO{nr~sX0XWah_6m@vHLvwOe>Q*8uFSNZE`sB6mK}HB^#opT(iV~V7fKI^
z0ej?p;fh3iF&F@?xGolGeCu@!?S+UK<v9@jGQBA9`Qs3gR?RY1d}!L?z5N{vJ6W3$
z-3vd)C!t!_lB$kfK@rRx<bt}7x2G2#kMH6zdh0Pg<A;2h-P|o%r+vIPA@UI=qIRi~
z&8RCdH7F)E_jOHLOm0OBWTsIQ+oY^#%88J(uXPXIzGm@w_~BkT_N#Qp`kPkC)LBxK
zG(>U!tCEI~{E{X~kb08?!Me+DEMovyHc`4`+eP=;ACB3U&#5Jfo3@$Vwp;Euu}zmh
z(Qi8)Pq-L;HKPTylJh#R?+9g{V>VOs3SL||uuKZA6|w;qK)%Auk-s+igs)8D>z*SJ
zN)2M!3RZp&4UQ4GR;KPW+#3Cvko@5lu>fAt>wCp7&MyiH#gtei8LkWYNu%aJ3H%a&
za$|sYZai(traGImu%RMQ@^87!=OEa7v8O<=z2dO9m^gr*I&h_&;%)iqR=Cx~;O?cP
z>O#pRQ1BD)*Ax>MM<7toZzheoXV^nO9_&hMCBEBS7Xrs`>MSp+t<}e%j{>Ynw4Gi`
z^f~e)J3E-}1<%{(5WMAULx9nEJb;ZZ#n%Essa5gbEGPAe`GF$W*?j!E&b<hzxf8hv
z2ndLFg%CcQ(s7EP-E2t%(JS!ZW}NWUYlLrZx1fJVBEv|4b@OqHNxWY!>U|lGLECVa
zDJAPz+0oVIzaeh-?bi^=0QBjTEZAznVX7$Ea3PQnOqba+EW`j&20TVttB_xWzI>&e
zyaku;R6~Rf*NNVL@sIf0Cf+Z`W<*6z-4|0dJFjxTy}7uS{rUiVaLCrM#M-*WMFftP
z<mP;QY&z<8<SVFmd;wz3JZ@hsJ$M*RAE*@L8^B8Z_0c!s&oaf0(tu7R<-PY(Y9}JC
z>C52;Yr?l|R@%Hb#+R07jmo6^J~py7U@6{sORSl1MT@2~sg+jaeaWi@my0vaDp#Zj
zXP3`z_U2$B!Z4oMw48_E(|iLIWD$|qyrdFbMC6&)wzlY`<zI#+H*iQX<sj-X#r~pu
zDm)3eN|s(FMF(}KO)k1;f$t#4&63Zyijt@g%7Cbw$RXtCXp9Be$;(T!#sk@PT>)s7
zWY3}U2s@|mgVL9IL7i{S`PMqTE)fkjElTReSE572p3!sO=k2VW@UYqfE04Dq&tE}P
zIpQf9fS2%~(}$#GF=SjBrzIYXeBG=-whNJI6p&qL>aXx9y9N;7Jr*hpELpt!I86(Z
z`z@j?EBe9nl3Y7_??Uf3<?P!R1N)lL1<0YvNI()JcwWJeW@G4H{=K$hn3sO5OR$x6
z1e6Bh<sduKGr_eO%cgKZTNP=6%aclSACp)TuA%E8dX}*J2XN>E8ErSxIf*qNU&6P{
z>}`e_A7bDmIZUpH=HXQ)=3Wo~lIT_~_w|;_KPWqGs&3x=cD@j>7^-nBeMww=czNi#
zB-z36OD0ML6DeTRFHM^+G72g)79@ttCJaZ@rI4U*YQHybiJ=C@T3Ox0y8_~E*%On6
zcLB!(z*eJo5^Ahqn!?<=VK)HWsX4)8Ex>aO7}OpG(3k*YotBD&gK31h^K9_-VH4Ba
z(cHIw`rXk^U#=dv>~p#~9lz#m@1Uf?7MGRX9$)(&hzI0iG*T|kO)r-s!S>)opB|u|
z@x+b5rV#>SxA`Up%^eZ5Z5G(9P2ADeZi@(AY&j76vR>R&%(I9y5sf$GRI)UyEG8&T
zU|jvVthRg{C@4=QRZt}p(n?H&LO(}a@i~Jh^f|Va>wXo>VbBpCKi%=L@Cq$Ds5%v>
zX+HmUzI3JASPvB3@@I*cJ%8FA^9bQw0n!Po>m64n0%HYKV2+hqGX38m0g&%nTGGnZ
z<Ywd&c}6=Qf%RpLN65$V*YfA{=Q8WP_l<^pB`q8FiEhhvv3ilnG}EG?UK?h&b59>@
zJzm?G?V8@j1Ol(g{V$~aa}M2d_PwQvmK$hj{JDTo#eoRFMIBGwe!^Yg&Y+8N-`<yE
zr|YBuxx~8&t*VHmN#Bx$-nJGq;}7@E+`YzuBLnCPcZ*G&jQ;nvW5pK2?GqZ^z1RvZ
zR9p-8C1{`iu-j8Kn@(%y>jbdqYsuTBy+t&^{_dvZkXL7`!>pcLZyK5b)UewX7}*{y
zz9}8(Y-Ob<eXs4^ySO&waj!5$==vT&DAi;xTS`6vVc?FSOIHnqbKXAExaZTy+FA4Y
z%D4fyIIrA2VI%)i7HO1mt@4+MHkTI^b(>f2q41GuGd-Rzi44W&M$8E%8>r^*cx!O8
z{=F)N2l0@qHvJNOB=hf3kTa2v1xA3ebs-Tz8ons_WH;r!0`No2UBUL}IhkG_YX=%*
zwdIym_L$HL+kp?a410UWcFsw(odKz%AKZ|Ti*MuULf~;BveZZ+L8zc(>`j$Z;+(_1
z!PvRKf#HA-rPrPQ7oh5SygMCexJHcLQ^mDFuOPrF0b~VKx1uU1x{gydrr*HNb6cEg
zOAq%P_O2}LIsz#qo?s()ueBo+qs<|mV?75!AjV=IZx%=UmKG#QQrn&&z<K88wjE7x
zzX|v&?Td+*c~l9{xF<)=N))^_v}!D^4;s$oQhE9j<Fwr_1BNipoz7F5_RC$CF9e&c
zVuAre86}>cYq1M9kTp|mhxZh!9dz=eKk-|rSCWp1NU&yAF^n%4md3&r&&VTizOMV}
zQ-Dw~MpT;szeGyixq5;F)O#g3LnxMh(3+fz!hHiH*9&QdIV$^`NC2F^0q`#_FVjXk
zM=(_ZqSw1V`m_F{<yiE`ud2Ufike}Fn<ZlVD$N{a;h*oyl=OaASt!%HN%iXLweQaT
zj$$Mw1r%QZL}j|Ea3?_Eczb7WP=b~*@)Re*iJXBUe<YokWKLd}=-ei;?1hYZqrf^s
z5SNoAhv_PC9#v>E0k9Mm_TUZ{$<xYf1vVo_$01>AE_Of3*mhoA-jEuDk+6cM0_RIT
z1M5h`$X<$(qyWU&vM+$|`zP{uYwO+jsX_u^QYG(<%Oy;o2UPFPy7bAabgcr}{yKaH
zc3=Gwo*YeVV#vZL1;f?s!HDp9dQwmRSzgnYnr*$TmYYS%Ecs+Z8!qTXiz0lyDpkFY
zOiaSDfgLWFQ=EOA7ajG+C+QJ4N0eXxui0O1*egk*Yh~rVP-7>VkkDTAxE{to*Y*RH
z#q*_gUhROrH4kfq)^9*)@dpp~&-i{q7?c16s&F=PRdtyk&;@UD(^6jCnemu)pbnLj
zRn?9IC22;yyxV<FwpyQ-J7>IG`#^QFG%!(;G6_PHn+(ARrBNfko6x-PS}6wS`%TsS
zlG@sqY2*P{?5~zqH67NVX$*8i6tqJ}OD+z{P0uRfJLi4;YOd~^V$YLaJZGMv35va$
ztVnjVq}U>Tz^-#zh<;aR;%EEtiTJ!6P4@8M9N$pgaqfYNduAl^3Z2ckgOr?HB)~t?
zE2b|W)T_>wJD{wW0V-CzL{}}J0c1IimKK%bNYLY}rPbl}nrMsNi*HP?d+2XNxqQrY
z`->syu*+m=q?YEZ_cl1qKxgvMK@VJ&jrNJlu*UnAM1wxctSpn4=34bsG#xT8UL<vN
zcHyM>{|%A@YL#uWWb>ZR;pVJ$dtAd8$VYY?==k~rq+Dx8$<NcV&(MiqP=9Fi_znO@
z%}YQ19hvwFZ^N7KKt#-n1*xgC*xTn=fRZ(^$KGCjir>%2{<QDB^}_oEWP{<VyHwx6
z;`8vHMGz8YKph9clu`;I#3<B(MNm;y!w@em4srvI0}vc8;B>l*Y_=;CXSHGvxm#O@
z=zJrql8sEKpk$O`Pm_V0)Ys40Q>eVeu0n!<>@sV%9F773=<iRrj+`>>Ai!bJ!;lFa
z9v^SyQ)!udM=okv>83zLXtapB={1tG&VT%lo;~+PVg(5+o~8VR8_9^n#wA(QKGEjS
z1iRuW?DB=DVgx;)ZHjzpol_ZDBAHce_AQ8HzM&@~%Z*yVu`cH_{YMF0G0K-KVNCUs
zg*0g=)Hq};l_=;rakw33wxx3AE!pHH2G5{j{s_8yh1?U=+tObX)WUK=7}KY;Ay<wl
zVR|U0s_H*c9Bn#vs+BAp^Z^OLkEacTNz7>g9NP@>s^?FO&mcoal#0&K&#4_EN+yoG
zXoi+V=9=-*1~w2{34b9gy!gQUxj!U5k0y=|Dckm?<@T92WZBNpdr3_8u0T(&7^A^i
z{@3iz3PR&SBr2<1To|$fSpf4Uz;H_8bzGzcGUqoo@UMSfxlNPoCI!9i()!``3~+!{
z(?R~417z5ths_9qeIBB(!qQ(%L6mPdyri_l0^R{l{LG*J7z7PPCWv4b&wCcgP-r1y
z3kAAmY~S3_oC4aWS|K92iz-8;f2W*)Dx~KkW#AN~71Uc+1Ei+|C|mz})#P+(Hi%7N
z`V*{USh{0o2{twELFBX^)IwYLUkG2qX~xK|LeHHRu6ezNfGfGwWg~5I00}{reVVDC
zjWjSG!5m!Q*?I^fz1QO&2C608i(bWlVoTE`kX#p$c=1#ZVe-enfay_UG=DWw*(et+
zLTAMuvF+<ur@Vl-oad7G{pSEW>96<rUmNyyn9WG415*K9(K660S05Ujf(V`}&Jnh(
zR8PLaV*ZIFAR9nvyY#ZPo6?=DetsXleEf6T%}XB~_*ondq8(>pN?{r*X8(Y9r`01x
zoxC5RKULBhiiP}w3`paLrEWT48cyP{ON~eZBAOtVWQ)J=txX{<^So)9whcEHs{ef4
zS4R3{50Nu}CG5j!0VN6kvZq9FmM+OY#L4~=5j}t<e};j7tnn4$zrayIhlpS;>Kf+$
zOA7!H2$Bku_4{c_{==tqD3g3^QMCA9n2Mi{<3D|RM+FQIBa27;dma6!zk8tvJpM>a
zp!R<y&xZsr4Gbp}VMY68c7EEP-#__75%uXl=m22mpYHVcpCLs6!$sa1{)Nx|{nx*|
zT<8-ac>%t}i+d7#gx@dwk8cEn3hH{m_p<+YqpzTF&pUxNl>g6N{J1GHDPTA}a)jA`
zH~Ivqx**+p^x*#Eo&K4w_~nam7*L=?!lD}g??!I{+P#=p=V6S$&DTGF3nGJZnW?0*
z|2rX*5j<e!o*a;C{BnQ)Vd;uMm9YMAq5r-r{%@iG<MR2xh5pYb`~NKZKUzKvD$o|$
zU^}}J#;ZohI^rg(dncfN*;mRX`on9WOBS{+mMXrK^s#sBWY@|5>4E+J?-a)oC@3g)
zsU;FLbcU|;wlxM%5ujv|H_|coDbNJph4i?m#P=(p;3wmRn%grp#Hm*s1RxbJlWfFG
zw(L$(TjI7>d;}cq?WtyOqS%WwhGQ>*n*i-lEy2qhZmuCHs^&PXYX5ErewyUmr!br7
zs1*V}Z*p64l)~i|#dXufjgry3k(pp#dZ@14qTq-v6Y2WVC&}W<>#4Ogb1(a_UV*y(
z><QxI<0BGc)br~A+O}wM-Es(4xA)05QJv-VAa#fV)Hq{2hP@Je&c`RoT2W5j-^go&
z(Y`!W_b-~;&zt!B(@V;<)a36bTUVO(-r4&>Wpso4pO7g*EB9+cpU4<im6K(01R9*e
z--<qz8Q$0FAk=1Ao;0g2VI5i*4oL@7lvk7w_!tiOyaB!fbxb3GUMxMD8g*N4F+~Ed
z_X@OR6N1XN08OT%fzpBWss!3va^AC`>|p@Dtv$YwLLU2=XV|1C7SbGEJI&2g?}Gj(
zC<r+80cV$8^gkj`U(%+f7JOF#U_)Z!;XY}O=F4m!g+IQmpIWfQ22v=_a5R-k-9Y2&
zUI6>XxGnp5_j!qen`DaDc9lLY0D}tKkV}*Tng(WBeUGu<9hH>oe=?Y91j+6MKpmN>
zgDEC{tV28y!&HM$UK>ULIdL0a`69Oz$Ph0Bol}M)yeQ^;uxq{lowoxzMgbBCcjRuB
z$k`)fr0npjz#>Dr6R?G)i+Rd~6t(r1Rmn9GQQed*;P&X#_$S=SqCcTiYnb!YoHSbc
z`uYP3cNUJ7o>!OJ@q!SQ_><#l`{ZhGOz!p1+x9>SFw48CexHbt^N#*?wRLQJ1JTY|
z8PwDQC4c#lY|{S6E{MfH<BQ+@LJ-`OkpudwIY~#=C@~2vS-Y=k9B*J@=^2)2%n$uz
zw&@gNf-t{GTqlhJ8zI=ZD3`fNy@+bt(YC=i2b+`>ue7ByIXLEq4uHGG0_1>Wu+cWU
z(CjH7BXYiN0!3XSX@H2Zi3>R9e&KBWFLw-sALbUW%QT3+6({NnDaG7$NDT024-yfT
zE*^?rkhTep`wuFzm?hjlZxCEKvDgj;QgM%J1>%w!HEzirHkfU5ZtVBD&PtN?Qn)+M
zoW8rq5rGR58w93NZ5jnspm0;-nX1$Hs|InIEXZ48YI4DaoK&D#DGbOkAY~wRHw<z7
z%%%KhDO6*8@5sM+2gyGdHww}iUIkpIwd)KZ-}tTKH6A{vV`+uC-mM>9rOI`ez&0UW
zdX_M|ZyWt3i}#9Z-qrw)&s~+c`hD4gfZpfYmvV9r*E7}D)BvBfz;fm!manVfu>Dxi
z@H!;<xVP|G)4X7R)px3XmL`v@lkJZ~e1Cx_J*&6z-UHlDsj~xxXrOq~Ycg9r?_s!E
zJRfkCEYeLn`?|PkN7*D=1Nua32*-VHVL=pYHdJ{3c;rIOgJKZ^kPo1sgD4dpD0ik1
z0Y3Q2q$Q3#PiT3}+97(W?KjC!L{^&<130+EQg)uH3mHQ49{X8ShVxf)$A@vV$;|lo
zJF4-gm)b7#4W-uYG!0jM7B;R8UNPtbCrd9tI6%LrNI;;#x|`Tt>S5}U-r)5&M5Q^`
z`RItKxTOlKJFdmx7&v^-JFwvG(916pXi4I_P~#bfH*#f3uq5#}kN1C@PtZrak>NE1
ziVP|Wii{QXF3kL7Xb90rogm(3){5dj^65#i<oaL%h(O*qz#*y6n^@5C!<ItYS!;KN
z(adW$!?At?I4kDzgVD0#exQ(rYBgnv1E7(`<%P0Jv>jV{0;(m&#d~L}2-lL>6zD5I
zLmOtNWh0xOF#Zf$f8Rzh@C*vVLW!Y4A^DAP_L=6C7@Af?neUT7APT@Yn59XG4-oD`
zp7whZcphCQUE239IiT)wn(T9x8}GT}J6{F_RF6VqbN@);YPT=h`1Jp2vnaa3UV$6E
zB@|GFfbfQ4D+RzYYFz=)x@vaGuy+iVje%?j0M49oYd}WGs#@BlVyU6?xI{8Fyl581
zB#O=~8&Ln9@fs$vEq}zj>6p6Admg+>+wYd$0zd-<XJ0{sf;MU4RoZg7_awh|P5OP;
zsy6$2OAIgasyBP6k%o_vh%~|p)VpA^v_4*_c8%|)c>B`flgDwZee^nAepDc$nklL}
zGUfpfUR6?fI6WY#a6A=b1+et=dJ+#j$p9KkjIKa^-%xUcm-96sOtCf_Ame!j#S?9N
z+>BN~V`Ed<_t=t#;=}9Kvn>o~&xi#LI3Qj1Xo+j_w4{fBSsN}GCI_+qw0iQfM}~L4
zWb`O0I=Q4Bhsu(PV+?Wt(hD?6*;(94zJbu9iw0T1&?x|2pN}9};Bm4>TcBf@ZsQ4s
z^O`R`QgGY24B}+#s542X@q5;y3qU(5H&eV4lI+a&xC}ZD3hr#tqqP~O6WFb{ihYjO
zr<{|prOI6NJ{EF8<WmkS(S~dv0wltNZ<-{{_m9mcN<n|tE+2qOF-z&&fuJw(dNAM@
zB*g#R!m0`|gD)}h@yRhUF()DLJOydc$j?yOM$wP?ndVvKbRiB6TPKL!MAvI_37P3g
z34gEOZSlh$P%Vyq#1Fyg?Hei&OupLTmX<G8Tch7{y|qg~G;n==>+ZCEdZdu}CRk#@
z+J%RPcV<v`*}44Bb0ppuQF-Kl@58_S*0ddj6E+W6`usaV)o=I!A)|FnK-L%mEpsG$
zogvBh0~}AG%i}@CWgVd~H>X7+2&r!t&4s~+WNJSE_L~NXZu=zkl4m={Il2|jO<tcL
zz1T^}mV<q$Gz}N$Q(jRKT*Q@0?^r$n^ii}liG5K*fb#?wS1mUcx-H3Y;L9C=ISaV%
z;Q6<6HsJ|#J0KXbL&9+{v8hK6)Nox3{BRGSd#_BO@O-a>v_$5lCX)eJjeg_E5r|x|
zT0dGwZkXQ_mhI|2zO^LO_jqNKHP`+s!{>X4g~Toc;m2Do`^P7xP}+z8dtaI_WR&8j
zbBse+f6vHo)*)>Jb<5=J3M?=wV1C#1jS&!&?K#)1kT;D%0^lBly%H=p$F1vv5o|f~
zOHn9xG=$!r7$!%)dTbTtYip4qkq)xRp@4Q+kHfPj8bFC3h5O*{Hc#+22pL)MJJime
ze0o&#OrYrvc<-rky8NDD{Nc@N=O>K#mN)6LZV)g0^EE&55R2(F-SF9I=o?ivfq6%l
z51C({`?qhkP09C|WZ7W~NOrC?eUrsprIll3HyF0o$!vr{ClF)NpV)fgVBQUaH-S$m
zT@3OP^LaUrsOuU%T&iFqd!|XsXWhtzwz9{csq*l^ysV#vg~eW~`>ZoWv&y}Qy-=^D
z!>&`l4eE`iB|s@WayE~0{zsziS6hQo4eORVFX_K`8aN&RAoxHoui@wb=8pD0*+=1=
zYzK2f;1Gr{P?btsjhf)v&ME^y0SAv|WKRaASOC!Ude`@2+nW39Ad`8cCx1Th%O^df
zH*cW8x8GjW-{0nP0w&f=wTg#OK9d3pmnEAoF==_dGLT@E@w62~t1aW{sti8}?vIf|
zO#>(k$o{>Sbn~pf+<WBzWHJBtEGBgTCy=35^p0AXy<betBrX>g@$sLkk&2%lWC?t1
z{Y5$odR+exBcMk1)#O~wS64jnQ(?yH`@Oe-_kp8-RKCs%yp>Xp<+dN8(7%1SKYx+S
z3~QK~+rgKf#SawcjYGE#(_gJ@6SCM>LZP_B_}k7oYO&=c?t(&3@71udSCUmKN}!TI
zw|RNxyWOc?S58KN`(dw`uKO=VtG_4jpDvuM3r&_Cd7o@8h*JqEzVM^Dfg^1c!Ib~%
zAEH^lOPd%?@gWB*D(`NlTrML6mX@h*J+&EzbOBD+1v0;&=2B;JUcAqW`rAzW<F~4k
z(4?QMMk5GYE8pN^-OmV!5MT2Dpl1ifTpV)R;A12gu+D5fB&h*YDyxjI#o{uq$0Qab
zyH0De(aI?6>wmQZ@jB`L_pg#{ksQs{w$Ej&phA15{bZYqZb>n5k76li$vI{2fE!$8
z5t{7SQKcaTfH1F!(ttf$cU+DN`1f5oEt%2G$$ZiPO+J`N9iYSm=%&yPvDiC}@`8AW
zk+0*yNs(y(_?-W9Q6NRBk*)Zl3-z5Y(sNC<Fuu6Co;yWaY^qwEH|9J&m|A?C8y&U8
zSBhBT#wLy96t6ezJvE<cB+Tvy2-*B_WNp(~$w6IXP;l%7Jjjba?~Ju=``!k=-UdHj
z6I8$bpH5W<oDZMrZ%N;4ESo*yb$t2qWqR`pZ$v}{taXZ@LG~;uCMNmS>iDrJ+RZiO
zo7XDFP4~rsN=WpZA-n{aIwI5Py5Jbd2q{`F(!@=LjHz^(a2B6qQjNA$THV%vxa?1Z
z3tEDvH2M2qu$es<H+!ww72_=5_3wlKU-pwQ7@ne^Gimg{nPY(1Ifx5b`~QkfshS4G
zI2sxnN*No+7#f+#Sj!i$sV^OwuS{TWWKd90Qt>jp1S)$_CQ?R5&qGD*11CBiY#@2=
zSi=qV<p1-AvWG!Sn@&w&*xrs;^JY%>h9z*#fT6?Z(TsQt3+RsJ0OCuZsA#A#6ft+W
zxcrAxg+F36yCw1dZcKiEY}I06Ev@>@%*^_DT438#2{L~(RQa$Rd~d&if+pKLFc4LF
z@(4RE!WvGMRW|IUvhur*?(S#@b!XAE8(@n^ErtmE|I82BCwO4fS$4rU8KR|06-SF5
ziec=x#2kP4$O=PGWBNkwd|##v2KSo%4K<er7S@WNElxI+`vm;=g^%IC-|%nSfDF)_
z<-`u4=Swe=-=hLdGm=Ws$0kd({M^^HV8e?KG4meEmP>#d34A;~LrgOG7M|kvo8~Vh
ztA97^he)Q|$N<pfGZ_lHECRE6+D9!56_2Ke?C=cGv{7Hg2KWE>)*@i9Rpf$@9X7Ov
z{A6*WT~H^r1}Ak7V9-E)SZS4#M$-R6aki<Fne;1rtRs~4Q<d9sk$&D{*0?CF_=!sU
zWdTDEbUc_r|FD!4GX>f@b=~*=JNDMgwj}1wLLBo8ejWZqqtLtqgBEa`AD<Zim8ob`
z6!xPjk7+S4nL|!HWW6}wp6=<IxAH|0!PTXY`R8|-mXE;7%DPK$Y7F1pQF7CezF}U{
zisSwacu9=+RNVU0&`s!ElZ5&M2%feApY#CbUO1@ds))K=ULZV)Jt#NwX<dU$kXXkD
zF&e3%V&aI3RrFD>q}wkIk8oib81@BOeg%>IGNVztWPrri<B-r`rd#<>kMezIEXV>D
zg8lu|<OdCr;GqqD$WW92`UAx$FtiWa2!$~<Yl6f$KF9Qk0WGYa-AkIxGmHuAL9|qL
zAT!m?hBhop9d<?37L#0@i2=rWWNB?mP%!R4ZN2=LWQ=^XB)l|TH+Fv%y@V(7*wj_+
zm9`xzg_b)Ct-0`iC=LoABCxws7@SGDtZY>WN~_)u8tYgfZcL&%ay_ZQnB*R`yp_uc
zF=Ynxy-x7`cwi|kJ``hwHEUzZ;!%$L&zHu?<p#|qY4h<I2`~)RX2uf(myTjj*+U}A
z6<!0$f4%$bp)s0aHT!AnMR3ATXQgCw5HCe=9=_4<^AVfWq#rHPnwDl)q_o<RuHNm2
z8R#qiWom!jLc(?M$8B_dRg8QHn!X;;nk%C+ux=>K{`4H7BRi2zF#=*1N1n_V2X<7>
zGlomIdJqy$(y6VubB6U?Jp>z=enYj{#+dCD`Zw43)1wg%gBHYC>BfIpX&_%oFtUWG
zL^Pq6u~S@>)<47nL(o8dwS>KUGsMEU^+xumqT~Nu{D0m(5(wu8>#ye!jz?hU1C%Ai
zFx6}f;?JLbf2`EoPhBp;ml)p01E}S`;r~xYP>}$%HTFpV(Z9wlA(={}Ta7+EkIr2q
z-L*Hm!Jf2DU9-d^lugHHp&)cA`P6O?phJfFxmwT81TzAOoD4%BP;UKQb^c$AiOdoD
z;PW2c@jp^dircURV#`reAMzWts2Yyt$yw=T^dAc;v--TEONgfM?!k+8u?<;6(dN9N
zxjeh3zWiR&1IRklbLnyo%n!X7#Owk1r~?C!Ad&Rp;o(?>gp5i`{DKoneA)sfYg3Er
zy_r*sF=u<TJ%{Oj{Z0{p*2M7!AKT!8MO2cIsc@MCR8@#Z&igg5lO-))l8bSf0r+=9
zyRZ>5l)jY-FyjCmoD+I{N(woj^S=F*OoJjLy;4M^1p-mCGa%Z1S$(}*?Qw7uc(S`d
zy~NFni6jXlh>Z3#zh>(TJNnRxLj*@W_T<&o>ktqBePS66=>c#~?NRt>0af{(NJNFf
zXXLFox>22C+fWruRQU}All>Le=bjC=L7tn4f2{2(VZ%K_c}I6#d=$Sa_gowaLHS})
zWpm-lsn(ppGxOAhk<{d5rq?xh)QOzMk(HI~K?+h~01Z6r?P@l_uhI$n@F5pDwwL1q
zd3Q0D0g4{hvjJ;xY#AS3nnZg^R!EwgCjinlJqu0r>oa;TV`OOu&3eY?&*;PS0U>!P
z$q`&{C0Q5R9rFA7hF4jt07KRm0#iy=?wWVd>+W{?cS%uiF2+j$9>1Fix)IMpbgT}Y
zhMt?!R~`0O2L0-hh|}y|Bsl=P4+Lbf5T5)CGa}3at0mt)D4|rcMfOp{#IUb`jcq>O
zxFu|h8L#U3PLQMK)oWe%;B=GDo{<WJj|2=#gIKa;0n-k7ThP8>dON+%WcCx7GQm%c
z073zQ>XKDD&YLBHgOwN#xPWusP}X&U7kO^OP3)Uqt5qA#fp0G(PEJ$-KCq24ZC!Oo
zm)J}5mme)>s^mCr7RZu$ouVaZ(lRpi=>ayVAw+L>chBrwW!Ys~nq%v$@r4BX#b<XS
zkNHmCK;7Dx4jvW*abDNThNlTm<$X6))HG7p&Q_Un-(#?diPf?#CUz-hd@kx_ui*&_
z@n!o6r7%sK-<)0QrQWT72AV#(JgIlSy|oytefu1JSo`-mj-nfBbi>XAWHHfhQjfbF
zH39nFr#cVSplg-~pnHIAu#}l!{4gXn8QeFP=!+wg$)!~G;0Z>V<gNs!fy`4MIe-d6
z74u5pGKiW07q`r!H2sy*m$VFDGP<Vo@+c$C>Ve(+6u#^CTor5;6%A~!eYjHR-42>P
z4jK-wzm@VG4Adb}sRI_V!=3f7_~2As7>>R&)0wZwQ_s4F(#zqY(v}@s;RoD5(z}z%
zP-Yk4-#*&ijy8>_=f@}P7{7XOWTW|%msVj`Bs^P0Cj#E`JtDSZMm+tiZ<NwKRd1%}
z-5QL}YQ|;NeYLlzvm$Z!>BExw9M;3yLjZPp3^+?rtHvxbDt<>9HS~THf$!||)GhP=
zLvR95`@#9_Cq(J^S4>2zv8ArtlQn(t54Kb};+8-5GlyC0@O2A7o?`pkEr4^n<^>y<
z_vZo7ZZBz|#=~<c0$Q`w*34FDtcw{^kz=%1k9IslIV@(5RX9FC7kjs%9mzz^_ex`-
zt0R_H+_K!DTOL`6(^Qxg|N8qlWp&Q)$q$erGSDs2xAA0-(!N{Z9%t|!-|cL?{q;P0
zZ<EWNbc)w8C~REb^+@jW_`8kOe9OICtRc{6iD)Vrphd>!6-WW%eXnH^`_c8C@jVu{
zm15C@z~O#K6Z$}KXMQyJ*8Io1n4p27B@}<TQ5|M>`E4`UN24s$*_?_#-bm<{zq@Kc
zQmT857%c%q0}<6FF~Tzyrv%KDLX)>*6^qNiW~N`R9EV=`J!wo5a9jNI-lGi;jbN~`
ztHN0o@rWzE1q~z|Wfkhy3qN6cO|1!zA1hGs*-nTd-#gnT6Q?og$mndDPD9W+7G;yd
zNnj5X?-XBwT<^kpmUJ_2Z{tUVhm&h+ex_n#a*v7Ea#(h<q^Q5?8y{|B2#$q^oh3+o
z^Nu<BJUS_9uiU$x#>0o5OkBm~Y^TmgTO-zbf4tT2VaFt=G9_Ad5Uy02^By808D^f1
zYcgV|>2q;6>S1_YfCOc(q`k`~6zCBfdu7y$#U~bcW|gH`k<Dt}Q5vS2mB?uojSw93
z@d}M(f@{q&Vbbq+|4~>DiJi-*{cd&t^;$>+X$;t5qyM0!UTjQRB}*f8q~)k)EpDPI
z`-DF&@eu*Wm5WuSTZv|auDzVI3{%2h8$HlsnnR|Z)5&?y!X@(59@y^38JA>|h4#nD
zl*pMM5IV|u=>cF;a-)^(51k#*4z6vv(w3$`{H!1C`2k&&KDJbPma9=p=BXt^J$s;W
zmh0A4`RH(KNHN+Vf4bGFTI|Iirn<?NIO4_w5%ZlZ!%PuTg{nvFsQjPgdS})NKysgD
zr<B1D0`t#Gwuna{YR1LUdw3S!_3XWq#&r><S&eQq`trEMqO%tCHrAuXd5u@3K5;Y_
zv3FF&yp_?`0v<9gt$_)Ji1S%?Uo1znNUQgzY4VKgLOX*ugSj|#Ufy(a#=Ri*#ETD{
zR#v9C!)kF==3hB-m%|W$?>Jra#c|%ohtN=I;gW|W_3LoU)5m}Z1GLexKMtbwBX)`p
zrA_f3MO5*;=#8~|sJ!)n;E#%##`Lis_+Bb^&ERo)VkUwMjDn}^B50&OXVGZxwa-H=
zF_?5x^WBc)WBVT64nRs~>Q^O+slP}(j$IAq<INmwOvM8WqcnZfOZL?6VR8Q5y~ve@
zK})J#9iHkvBd(*d*kZmMz8GookQwlJ4jEjm(n3KE?jv#;)`@*V?A?L!`m{UoAVQ6X
zo1*dN`;kd|f#DdVLcqj!1f`i7sJlNw&HjA<t4BNe=V_6&6sSO(f;Cm_4-la+D?*a+
z#|gL)mYjHOezlkHduf!kl*{2-)VEkk%3?NbXBza9zXk;s58G-4cYoSu14J$7z|2W+
z9-HWrCz~J1hl1Kgjq%LPJ{@vmr@B$Cz(FKaUDb_mFkb0X8jc|Gg%7EK?+l*RbuoWm
zPLhDbqzUoq0OJy$kJ6^1?(;#W0`EiP-F2`!=)*VGi5zs>yB&5Yo0t<IYMDE6PNI)V
za7MYba$Z><X@`as7))!#M(*>85QoE3Wt;|`C!P7*dH2tU8^F?@Me03at^{GoC9q=;
zK2HbJ+)xW9>SP!}p3vyNfTMXyC4O%b^NNn8729|UUOn#91De<GjEhGCj{#-2Foc$Y
z=yx~sV-3R(W;uuQsItH$kpiL>PuPd7Rnw0+6b)l9D<#IW<tWa^x!74**~BV%J5(6>
z&`owPTi_Sy0FH~16h;8a)H?rr67Miijy;d-JWMY(Y5{>}w{^=|@=yU5k@7)SQkT0^
z_f1D+H>>gbtSybga)4CfQEL|2RjwEN(a3*mHYe~F%*Y~>@>IKVAY|2W)pW7gs&L%?
z(Qg3-NQ?4t{071PzWqL&@6JBikwd0<XN*W_Ac+!N*PU&LD7HkK=?b?f$>~bE6_9L#
zG|M%LFGB2MY44nB9QP|rimr^ip4Qzd_&AUn6N22Ko#>kClePbHmZhH@?mglGA#!D&
zThn-B?=D$YaUr!Ke?l_T4kB*QmZB^{%a!Uq^#8PEHsAjApiGKAu5G;fEz^Ng_|iL!
z%CY49vru;9!urD{)SMzjZPVu;_%PaevO#zT&zicxMMqPv1Nu6l4ltunPihh4Jzcgf
z?cOP@Yy%2-H;cG<e1ioyGd3gX<+ckhV|fMA56xmW58e4!PGj!QP0D4r>`Atbdqeb7
z5qYwvpJJ1olTOmawb{I*M;TNm72kmI&NgeXTSbiLvdONrp4Xb$$G$LV=)*4h`heiU
z(FV|mU?q#ua5x2amW@(Chu7KinMnwb(oD4(F6+$1v+RD!)VpD0!;8THlivQ4CQ3&2
z;xc>^{;t)n<-lI#FQL_~6e@yO+Sy)g<0RuE?4s|oD~P6BTzcZ_RYu*S@V}~+8TE5k
z?3=xRM5K7eKO-BPH6o0{!~-<wHqx8{I{Dn}0lL@Kv73;~^{&?kWwZ5_K`M85Ut5d{
zG_N{z*o7tf$BY=)-@k_Sn(IE*O?7NTa>1Sb-F2dpdozFa`FQre0UY83R^#ez*>BOg
zI@^C6qacSjh`5Zi>4!AbOcJcRN#qjKIdKb++C~ja=_&)CFzGU*3u5)~ruPS)>`%@&
z>TyYSDIgiTI$l9O3X4MJ$jBF*ivLu%djRS~x#-U}ROmhaIOR!QPz%TYVrcmp3fPK{
zuZ*Ol+W<pBSeSpcV>Y|-4H6;$n|CyGntx;<D4BL4x<~}OZYn14IeMED(U+2^c;w*I
z%sn|ek#>A8p6E=RWjDT|Y#MX)s@XC;(;W_<mY*i>#k~wyx*R7qAdHposxpv*{PcMl
zwzD)kw#Hjs4!!^d)?1qCHp9%G=kXXQD9-&eRg6ZHJ4TZvQe|&fjomIrkDQr#<#KT1
zxE`G>U;5p<YZYJZ1#FDt5GRRZ^*m8}IxFYw-<|47zV;c(>rn8tfBT!sV8Us6TUAW%
z<1hn~iI&4LpY^fep(-0eZa;Nn68S}`5)C~TW@V>R>;9Op1bJGc*-oZuQl(00R$DX0
zqsl(T?BC1G!9WXdWW~$$fk!I*^!A5KFm@(D*--J7Va!JcfD1pxHya*ewt@S<dh||G
zkEGuUX277`dwMqB&T9J6-(7~zu9;R8=6IWpo-=v-wC}OHEWkG0ceL!6k#3D-2|lY`
z(|VnwNl4m!Q2ha)$Ae^KuzJ*{uFkD>4+MAhfn@yrdD04tPoN5hX)cgrW1rv~cKRzY
zg@L)x9$V*sxq_~HVHd8)5D}k|u_wNwAff@1`3zPijR<_!gjpeq_4xfmtlC&-B=m?6
zm8$-bk@<Kn@oT<YhO2ysQ7l29qH$`WxmGwf?S7vvyS?v*MM&u7w#7=bYSmkD;U7KZ
z3I14~&Wl%J|MLFHxszGLq3}m37))Ll9WBETZfbB$``$o*yVaT1jky`!V2`{gsm;pI
z3`cS*oSc2u6nQiw!ldMQ@xCZJ`J!<2%F3|l(JFF>{@oVu=w#ocFkO{`@19+ZfP3}B
z^#xg$xiks1;(Sp{Eayr`t4oG#B6B>SJj&qS9Ei-)7~XgGS-25DUU--~vxhEp_ky`)
zZOSM++T`2i02lt#A)#~j(JC<x)D8&tVkrn$PXXP~W8^9_gp_<Rkt@1Rh$(Dp!E-nS
zvOcvK0FbVYNQ`LI<pY}2k1t~vr_~dz+2WKklf3ua-S-qe`+#-S_N$r6u@u}~j%mxs
zp9WwRY#m7-c>Yl)jprbD?QDv7^s4gZ?dZ+-!Lo`u02@?%kl&wXO7p*$1P8h889wc7
zi-%3^skO(|;-YIH|LwF_Mk{80)J{>&E#)6#cHiP6-)XWq<qdntRtLSdz7HI@cMHE+
z&gNT7_qeB!8#s<1?Hih#TO!vnRta7-ba#AtF-9UUP=dq`54hU|Z8{qSKN5mcVZtxj
zZdD-*3OwBeX94FQ*XqRu@m=f_fB@$+fhtRD<v8nG`*40oFXIX4<;~;DIfWaZ0)mM#
zl5moX2GYhr!fjD^m*>K2*MXs=2hDu1M6bnA7;Hn=;c08XCs{R-pb6W;K00w)ArIva
zWV|6j3PFNr@ppNwr!RWW(L*`Mul&_di|?#{Z(xcikdIsJds6j~otG0VS5p1q<(Br`
zszks4@=3{7fuQH`Ku?s_Yc45u^=fwOF~FrAmlgKEq<R;WI<n4pqQ<5imxRk1*CwI6
zg>7-mI(Wi{s%w+a-e+!jXKJJW4(;S#G{2^DF{emVrZV8<d1~&K#XFKD`?XSX8qqGT
z(Oo2&L0ZlXT(pco3<bl}vsRn;*2;s2BYpT!GUsi8Q~Jhu7Yu`~!nU>qD);O(Z!jFp
zR|5^8RHt96Pr7lhMxE!MVC$t&CD@4uBZjJ00|$3uxsJgZp}>w5xZbJ3^Jub$Q&cQ;
zU+1&cb)1(Yv&EEEsTBd&Pzd;&mu*!q!SP+7ict$Q@%n|5WF2}mIOT3n7TN~N@DOyn
zcH<z==#Yf+dm=Mm=FQgf!+t-X`(W++oR;36FM%(INv`_=ne_FAYtmSJHaJpuLgx$e
z3>hu!`T9X`I)J*%ZqW&Jf1)04xp<G3{fEQCI%j()ayPNn!Wa5PHP_9dCQ?URSmqmQ
z4><b#@9sS4uvJPxbUfp#5%*X3n=09_<IZVlujs&L%1u>;(`)R)F9xR4NObY1+YS5}
zHl98*j4bVD;M6Xy@|h)Q*fZm!do?w}1sji&g38)~B89|hZ>3t*Vv!p*c=#0#?a=n&
zgW8%ZgA2R)P_;P`)Up}1(>S!1=OCheEPC!vFYTIkHcQj_$>PE%^`yFO-IHH~58t3V
zp`lkJk>)U56FeBrG-HeBQDGy@?76#ksp+RTeew~@C+K3}n3tLa66lnF8QI3)Skq4t
z)f^KiR12g_s!tQ|aq+GJsE7)_=SWz-Y@V{00`FU~32#5dvnmiyJ-oDtNV<*<d!yZb
zbJL6bGqc$l#L%tNow7rmSb!Z~3UqoR^Bb?rhuC(}ADr%;3q~NI#?Baj2^CPD*&EMp
zWX23N$;|1mEi;keSJyRCH#Q!i2}HxU49&@QUtb+=j+oPN!|Ioeg|Umd{DZ^3mfd28
zYU84_d1dLyN04*D8qm>^h<U1mSr}Sb=As!s_vW>ZLA6@JlkfzUpHeqJvZlhy2#>jZ
zghG8*ytkY0y^mWZRn`yC@AzDs7id;Dh<%#nty?zVUYgH5;#n#6TLXFAS%!YM|8nKs
zS@EpVI`(1R=3V#N>Aif6$3bJB?ECH_p4oQ>#$0{nv-?@kC=vAF`AR{;0svzmBW3Z;
zU)mKQe-u~3PJtvuxf@efWp44By=FvpF?w806fxzoj874B=3Z&UMg9-r#Vj<ySBDR`
zc|-U<eAOgr!xy@F>!o(hx2k?_;VS%LmPgrFq<j;v+5crG#O%<2l~~Zn>VPPCLgRaz
z-?Nj1{@3a>V{RFrNo~9;vYBK#gJ_KDgNd-8g)`7QyM^K$;o{Cx3n?pq>@;|X!9Ms!
zXQ3TlIgSn$+a+Az5AB5is@o++lbO5|UJXzR>h{?2uO82mRxi}1432#o3srI7zY&PA
zX?ENjc_pr*C8701w8D1FiNiwu4CF2GiJF{&FZ7jzBkA}f^q~mKhv+0^-SI$O=n0y=
zZArWq*iQeH?<%+5C|~oP;q805<J*Bc`yLTE3Qiv&Ph<7|a|}~rza32&h4Sdv==chs
z2p$I?WJD2DzS}wh^NaMe&lx}{$)Y^P>Le3e6h%7_`{WjWWV?m8S(cFD_ZU#B+IKNC
z*>a&U50F3Mp!%RcK>ua^zWC;|PK@%R1=6#t=FTj<WZL#X5(bwAh=OLbeN`b_f04x1
zo3>%pC$sy>e9ol@2U{V-Rh16`{#N}i7nDasFgCxT_2U@G%@ij4-VT3cv&gW4oQYF1
zjHj@wdQ?4(r4ZQ!gLXHUv2<6fGPY|^ysvU5Clh`tewA5eIr<5Ae}zQmyVpjNR*v}a
z_G0<-Qk~EE^lh;QrPy@0Pt4pNo*#)>eP_kre+m!v_?ewBQ#wV4@wVT3iC@v058)Ei
z@{hkm#N4tcbN{?=QL<GVKJ#6G>Dg+)__LG;C^=SzQ?FXN%dRFn%CZL&`av|o9c>5>
zd!*yKkignI7M)kG$O1YMXjc7^;|>$=od*k;H`7}Y2o&E=4YQXJsD#&e!~(7M=DlHG
zB&xY+Ryh0dbKFt0h2Ef~@^)(5WV}?{z#JxG$5?JU7pjs&e=cK`<nfLE$gyCQGvV|_
zh4I!Hp-mGcU0P*EF6s1Da*)^ACA(&+Y2N(gMCO&TVrE2QgUsFaw<y-YJLlO`&d5F>
z!L_SZZbVSaz-c{4@w$5B!%%_8`R#7_{(OC{R_VW*i2loq7d}P4F^!>Yx~gyK`FiKI
zb{y=(q#$rzKIb@(dv-88Ku5X;==gaFtwt)`)2ayGX@{_o>i@BgPh)15ht;f_-5mOb
z^yLU^siUUK{I`K*`JfXR?9D$)kS1J!m%DRKym9qo9Z9A6`g~W8%oI|*Y53UL!mXh!
z_dEbm)iliy@s+9FPFHdxSKLa-MqUO8r^f0TX9)=p-z;Ux=H>e7JScY?i1$mq=2uNi
zCA31b$qg(h1!`j6*8(kfWZ(0p3B#eyX%b9Gz|l}de>@4r`G9FQU^pS?xq&%KhM6N)
z-}MT_B;RtZ%kiO&B?!oVVFVZy%4F1T5(vt>hRIwsq1cdGwKP8X<O=r!*yTmdIt^Y%
z*Jn&L@6Bu6KI~P+(;Yh>01#pi%LJ(suE84+NspA0qT(Ak8h(0Rvx(d?onuNammJHv
zT9J=l1Y&AU;W!CdOUsg}WwF`lOQ-3si-H1~b(+QLO6u8#+EwZ_3YSl3vS<qh!H~Ff
z2!-3}XO}%szIcT&QxpZlfrwJZ+zVlXFhxa0LspSp?1~EG;0vQ=tVH6q=7*_a76Y{A
zj+Z#tfs~O8DHP)^iDuWSKv>PvZ9bo$1&yVfy6X-5MML?muCsCw-rYJ51V7`Sg+JYm
z%Ia6|U4pxgj-R@3P9Z;#y?b?awnx61>Yx?Bs`_iK4q8W3Ook;uN;eu(5bB3JUtln~
z*@3@22{yA|&ho*~Ia@vOxmM|MB=O$GzkVCQ$V|+LiuWuUpKY_JQSV7_RE=I3yG{Z}
zA-9aSxzpl;z&lAVny+p2hMBfhu2vsljqVG~KA>p-t03EfGvbS&0NDrnFk;OzH`zE~
zGlf$GI8=j|kZFC}@H@byE-&BcG~MKR1Ihs19BFOzfYew^&$C!<Y^02lPuTK$0E6E*
z=$$5yyn{3!S(rpi54^3ZbZ^M5FY|O6jMY}Vv(DI;auJLfp@pS)35Jto?_FWZus9@L
zmOh4P&TmdR$P?XvXjb)rYD^XmNR->PyVIL0nUaje#(n)zRo)G+zEzH#U_54=|J=uL
zbA4S}e#p6kNFu+|d|pCXAkP^2vnPc9o@uO1rN%t<Izznw-R8~UbVa7Z0o=Rm`ZDX=
zuCQCfJt_y1dj4Jq<3SZB;+X?VjXUSUx@YP5-VshZWnIa9HlRi!qx?sFD;W3=J6pMO
zr)cquI_}N<w{r!~+tCm2E~B*VM|qB0N}AbhuHg3<5jH=E!^zo-xvkF>z`_{r<7Dm&
z)+hDroC`eZNur>(9|7IY2ZDc5PWuypkoFB}cP84zkHYcA7a;uy{tPR_+`_Z%#n+u=
zH7ger)D-`&${rF!bPW^Rt0EW;G_M=;b~UW&9l1QM#0L<dnANG*vxQ*DyM`wBUD7#b
zI=V;Je58s<>cHs?hc$go^#w+UJ@#MB?;}*4VGrGpn{r6oippEq3K9vK=$3B*Wr(yI
zx%ZjQebxOo+@Qx=B{H5*O*bN=Y2q^a7Xa3=)RTS2<1HP~8`OCKLaoLA=0)s*El~WZ
zQ*PW)IXF|<tB${j@iA;szWy9J;f;sDU~%Mvrfh3M8F$nkN?tKp_;4N9#TcGgfwb7?
z{MK@oDP<`r|LmEKcM$LX&|w;wJ_&1>gVzfCbmyYJyFdyf(6|u8ZD?NSM6iQEG&JC$
zP1F?a8a(pc=$oJoCawNqq2X{H8pSp}<);qbNtBwO%d?Ci8-SV}Cp~*S9Xqk%$6m2?
zy$0vK7&&A50Ey4#vc@nimK-3Qhv+<G@UO&lW5_|{X!z!lnxYf*K*`INbFY9|gmpEL
zhuo=mM*15+v{e#RdMVj?ma}MWhp`K6^fbbW7#H7Vb4d8>Yhz<FMP_w9wCviO704Ps
zwrqMUwt~?0%hEdIsjc|}58D<7LQNXfpltTZ^%uI{d&GpUW6e6V`H!Uae)rTPC=NR{
z77@FN1&-+&Zh@3tx#zlh!)-xiC@P5yr##QmtIHLi#p#wJ4!BNV?<O&!t>@eG=>0&?
z<7)~V*#ygmoolZ7dS|=$ESgeq{frjkmDHSmpTWXuBmi7kDb$8J_J@$Buo}WmqW2?2
zCD&Oe%*FT7&QkZ)`h;6)AT)xLrN>Dl50ID=G$}d_qepIaB@+}RzsJ&}-fg1pChoOc
z$t@)se?ejfEAhTZG`Zl!n|9c$tHdc;AjI4=EV6|a*m?0>e90Vckt6p7X^6#W8j4Gp
zc$zr)V1EhKgg)>iJf3Qm4;Y`inTFr??R&eAxQ<~L_)$U`*b`|b*5gtFl=Gx$5hLCS
z0%F0sU%u6`>1<W=#kI{#b~}1IS9*^ol~5$}zIC#it^R-Py=6dDYuh#~Eh?daN+U=~
zDc#*2(y4SKT>}mvNSAaY-QA6Zz|h^@-97JO>;2q&ySMv!etthbf6d6uthKH<uQQJG
zI2iaIL-Z!gjKg$2i0!vKBepEB{maa!M&YHFW4AFc$aGQ#OK@4<B!zS`C%JWoKO^TP
zdAQaW1^DjWe(BwMQ^0EA21A3E62`p)<Alae1C-RtO<rJcYz&}w!_3tBxnhcXvTY=w
z>Gx1}hyl4=v(W&S9&%P-F`$=Q5~ID;C8)7&2dJ!i5sTI}`tWR0kyR$1ewYOGS(wiB
zE&J8J2_qLwmr<gB8!Y+sT?bQS;{9r3B*CQ^4yaU^=iOHxIA3v6=avs5-E<El$IV_Y
zCAmvkG%VzppzZs5dXoL;^#21haSs+nV?N=Grvesd*$T)S*<Vcz27IYuSX@u3=)Exq
z+{yG$s=c-VH2!eF5Xyi}ECYld42IB`K%u5w;ywG0U3G{epx81~D#OF}vQ~d~ccwzC
zPl_-31+XjzLiV`Ff_`8#*kGkU@=rFkep}LfIs4RHibx>nykgluQ;0~EAC44A+cttM
zKYC(l$i0tT4H-<nzxI|nt7YOteNCxcRgQVaLvq$uMD^um-%3zqS}Yxnn>HF{Y$*%>
z)J$cC`If-Z9xaMqu+{3So9ymz+f@$eM1n!V>lsQ(zvTT;f<+GV+{rm?@_|C2YE4_b
zsv4m$IlcC|9RGWNj6$B<u5PSiYl6N&u<^WPS`reu(Mx$HKT(h^HF=RhWw@Mec91lF
zZKraQJ23}@a0wDu9s)YsG4gm4E0(Kd-W6Y>pF6tBBi&eLG$0<tbt<S@`rh~4&)=*_
zNL>878I!K=c#dATma09$-TTq{+>|C($xOB`5|F4*S5<o{vU+>$e>3QF6w4*xZ}7p&
z(Et8Md{cs6&r-VbT<GFry|LTv&G5G@25yUGt2jhP6Sy%L%?u%lM6nxHS(VOL2de+!
z0(j0K+crX@rvQqIB^4=Kw<am>iY>?0^Fy`=Tt?%f2|mxO?nA6x^Co}9ru8Q;I)d=X
zT2Xlf2U|U8JQ5bC`;!_aAcHSzzWFp1ml=Jh-<V-|gH5*>4-5G<7Whn=tV+HLt?R5{
zJNYKZ+{woivgCR$${@pNK3LWB{PKkth6o`EUHxgrh599f)5V2|r#YEBt71W&S>DVT
zTC1e1ll*>k&>F43c+{lXOv$I(Nek@{JNbiYA1Y>M!!R!3@b%f{z`Xc(EG9v-tQkM_
zg?a7nJ5HivLc~+R53mJ%;`+k+qRI%Q%qx%~{0r30p@S5Pt_E{MoHGW=R<MMORG*%r
zxPDe-{7lS=`w3CXi1~r$pn(7G7TXLTW(BulE!VCwM`y$6s%0oaXYE=p_WiPTflDJH
zxR)&q%3gb>bG({H-f;PaM(&0e=sW6~r1JN`i5!b|k%$>a$^1?YiZSkv_Mi8QTcXS<
zZKm}?C;+&5@U7gohQHRB?&{i$!E%nB6GeA}_<N5WK*W_NFl7eJ<AoB+sx`L7%PYHO
zV9d~7xwmhP2t49eI8nLy+3h#YJZRKJ{Uh^xN(e*Md`1;t_P5N8iZN{z3>|5@@95JX
zM2MB>EXd>Xrl4Jv?_@o0O_PWUIvhTag2S+aK)8h0xB1VA%l$Q!tDU?OX>93~GJ^Ue
z$<RcNtppuGX`p)L8oZ7~E`-Bj2D<JL0mRt(PwT2YEghZRQ|9z}J;i(kn|PD}pKzd&
zPWLw3kpQf!Hy)=fqpp_$%A387&S0Ch?%Hsved%3V0H*|l6+5Hb|8aJ((2~wL9$$s9
zq6bf$hYdq(Z+olnfE380&#_t~$!*hSCJiP14S5<yxxdh%4evv!&yAO|;+uKq_VNAn
zg&|__9)+gbP&O-&CNL-Z<mRZ8a?#)N`D;)}wzhPX*q&slL`Q-tPxAimRZ=GlgTVwX
z!2Av5nOPzM%_s>qwH!&PMTQaLLmGUJips)WOZEacIdr4@cPVJE+8(AK-5eqJz#hih
zV(sYR7s-z45X6*jlPaJemKsyq>Qm{S)*{IuBO`AMa+Q?_nYb%ybFrORuqZ&E$so5p
z{*Gz>D=oZ&V3%Yv+5%G(AE6f#arajwc_Un0FhOMbi?IArHa_fLfm6TJF{JbFz4qwF
zy9bE)tRSW-0s5#)t7|-&wK{egu8cDp&RVabcBc{hSW-bak}@wOk#!X{fNjiB{j3A1
zg~yy$6kn|NG*8HkZV8u`0nU>$yZwaBYEKL$E=xg_JwAKv{C#D3{*CX?XkVjsvJ`|T
z)?daZ`+o{+bcxbq9`u>jpYraj+;U4;Zf=CK)zwVj6gP5f9#hpBA$m%*BM`(|&OSMZ
z`j3*SG$Gf0wxtC{7itf5fgElqCtn<eGzQ;1u#}B)@(Ny<ZQPBOn5jjc8U2!?Ed$k3
z%2U&Ae8q8IM1qP+8&^D~S9L!Bwa@(CCxcomzgZtXxT{V;h+CIFn$af5ydMIFyXv|N
zEPYgM^e-zJAaU$ldIw=Er(1zygm(ZQ{k!Dhf)8rNjh2f|35cQNJALYg<ktu-uG6T9
z+|jPDP&e-AmG|FEA}XAwM)Oncz7qUee@uIT`Rq!s<U3hPdg=GF4cB|~XUf{IDyQv}
z0vIKnjfh$!8XHHbmnK9RT|PyEILb;3Zy-o$wqmF^9M|F5+7qWoZdw$&eH^U~&+a<s
zs`paJ&D3w~cOss3v&{8|HO`*s6!lJ=pG;)3#Wjxsg>xE64-!$?dRVr$`w?jWIqv;o
zmb++o3V+4<XS5sg5s|Da>U19<`9xqa@UK*$J7Dg86GJXI=u#fpf}rjZLpgZG{7ejm
zlfo|Yx?ktvP1p13lAfOUd&peL;Ac3D3z+0Yt@v*j^lGd;DC8mJ;Svw%zg0Se6HHAS
z{o{9$of24YL#sLle3pA#PgN3*Q>u=d=EORl-Of+55g6m4m{<`salky6=0GXxU&KQz
z9J>L0;J%jxK8KU*cr!2$r9bkWpLl6!a)f&hF<&!9nmbPIgA8XQsx5m}oC{~pI0jXu
z)z5O5cFV?ecRx&<hZiT5+U52@jT6hZ#)X(I=$r3TW{1s;U$fx8`27Crbcg<=v@gvP
z>Z#XwQ}bOxhKrJI>ZphCVB+x<xo$&f`*8K9JOPuYNjK+vwg5J}GCW%y_bcQt_#F1_
zCO8KwN2^j&R;_#`GU}z@B#Z~BpT%Mcdfv(J7zSIUnpU1}%#i5eD`nOnZ62gW-%A6Z
znvNB4#<lS?XaPV&xlzk;OpOeoslBf5Qog4(`D@QV^gOP?E?<vlRQaY^CLQ%O<!=!#
z2JFSf1-+vz&&4(n*e)KBcqlwKiev`qx9HkymzpDT3l=dsx@G~+Jt@-&RSlk~y?tZ=
z+lgZ-r>O#)^?1svC#seU=ainqwi7Rls!`<L`4(hsQZfu18~Zf+NQx&qS;A|bv!xqI
zb5ka8?3m^<<stxSFaGT@a_|>GG3b3*hWAsc4>)S_W|ImInnjS=HT=_}QKaCQ*iaJT
z$;<Xt9+jV!x)tQ@u7|GT<tflwIh8VpnXAK|Fgp!6$_JUZ&8UQrJVbZimx>t8^P8+O
zFPiUAogGUT3FfHIlDPT?rwVcG0PTj%*kO~DMHO0kFdz;{39*Ed=pSwvN*+em&O-wP
zZv8~=#Rc}z2;xzucJ>Uk0j47lg;W~D#{{3}sr%$cZhM`=Z#H`y=6kpX*<2nyN@KPS
z5<0R9I#P3jxw&b`$`a)yd(H=y-@kGVOANJuCE?UZL@A>w<HN|#HkN;@VsvFB>@B*}
z{fJkn9;?3zNptk8WfDPHVx&^~*6<3qEkp^DxPY18&7>WBp_f#V-xDWN%)^rGeoCud
z+_gcAy=qt^aAm)`#;DSU_#XuAf8e@GI^Cl?r{w~*^@KR0WP!IR`7xyN+h_}GmwQtR
zdNO?Aw3tRvFzYtaMwR7+K5NSr>(#}Di`i%=gEU8Tc8Myk!|JN|*F5Tt$?{SVS|?l9
z9`ep~w!5YOx(|7LazIN<>R@B5z*d6WcSp_2cmDVlc2q~;1{~@Q1MDv~Du+EjvajSz
zN=X_Gp}j8Qo~`RX-d(QaIgG<Zb=*aSI6mmJWUN-CEKk_9X6m{lyuRZ8y#H@>^IBwI
z9c|QW<9&eP2=q0xa8Jf8?0Un86$g9r#@nt$bo~&2m-him(f-~8_<(O%g2ip7y{gQg
z`AZc(yPCpLOxpwjoUSfL!sFHiHH4H@;P9B1YFSNYL5aKEHtCU~CszxDlpuQ&qQ6&^
z=+g!B@xu6zw=iLei4B!D0Y3BK3KyR5Aq1z{-u0eEBmMBUl&Cl2f`C-j2LL=0lx=s3
zWjXNC*J2g&;%^1xa;fCH#-tfObgTpyC8Z3nv%NvK5Pjl772{P<vXVi}@O~;`tnXXG
z*&q=7W)6TrwL9s3WBDT-%~9ordm*R+gN1Jkd!QdbsLyWegr=?X`cqKy1jQodtUbO2
zgAO%)lK0p=a%d{sTx_~9?7)01-`5l@&D$2_*nS=48B{d>{s7d0NIc}dLbqK~CcHkQ
zn}j7_R}?hw{;h#WSZb$;USH~a_g8qHu#va_7Bp?7a_1N1T|q{E_yw|*;9=%C*c<Vv
z(;^rwX-zzm%gSeyN|b?Tx0`bEv4Jf^(<hW?D!N#OjM=Z_(}fz!yItTT@g`n2j1;qv
zCh6kzqZ3dsz)Q0h?&gln!#bb~DqhugNUv7(PT}`2Xv7`ixb5z5J>u#8Gb7k9p%~U<
zirTCXLe4`^f1Yxi<sVDmdeiQt@I3;jb9HI=lHj#fRuk2U$0t94ynXEOrWrFa5+Vdg
z!8t2b5!bkW|7x>`X}5UOsMU9gF!Zp5Zk!tP9Uo@zic)zagEYN1)4c8?L-1!Uk+MaN
zrBpWUPw@VNrZQZ=Zma*^*1?0cI%xBku>E#Y(gj7zCtA;@S+=@r{O?Q^9EnIknGddE
zN7UsM*V7-+5EL05<wRP-7lZxK84BuG1xzgNeYqX*Yl}O>67o}V?n6u*g7HF8@B^e=
z@`EKW)5SP2?iUxH+ekj}#Vq`?ycZHSquyrPUbfWEmVsz@{*ZoqkYG~ClVVVW6VGzb
zd1SYC&k(7Q+*LJT9{e0vafeWM-3veGp^fBt8$J3zJO*z*#yC*>Tpt=j-BBB>sb)$;
z&Yvn0Iv%jiRf3ovmkbS^8fX;^_`cm=c5-sU+6+xdz%Rbx*posX%}h)C?Dpx)pseTq
zWs{o`9=~H;$cI<O;hhJYtC|JyIGR7>qGW4WoMc%}OOJDw8>7g66`iPo{*Mo6RuB}=
z?U?~sI$MRgm|Vk`Woi~a%<i3rt2mUh{WO0E!opg1TaK5G#IG}$MR!<wT{ST$D`0jM
z$sAiC3M03>#bM=|&sL-JObs=DP2~U=2FE@VJ%~n2xRUs`cH}=M(E^RFA~7ix+CxA9
zZ=wTn6Z;popK{rDDnubW??O=0tdQQWlU_)^N2W<^&kh?*b<yc?rJJY_ysl-;6(*W*
zPN3n3qi=~?poqUjqSUHwb&bH6@MMA2x7_#IWjVg*Cy%N1Ybf}mp^WG|Y_?V*=B0mn
zR#qpxyxz+TvqywtygTLV^~prC_`p7U>th{Y({uSjYkz&7@5W=wY&gDYl}0_?sWc-j
zGE&JVe7*-*bG$T|R<N&ZO>iEi@L}~<pu~=^z;lk9T+zrE2{@vEVzw>OTCWJAdE7qe
zmRJE-F6P<G`wr&(FZwP2UAZG<AqJ7w{VUODn3&=QWMIq@#69-VNz(xrKE7mj_;ayb
z)Rp^d{rCI~$?M2USGR%Qa)T%h#~>ipTK(8rR#6p9A{-EUhf;gAJ(oHC!j(HI-eCax
z9bqjf)Ss3$clwm^?&)!IK5}I7qnPb`J|vMZ^4hYBfd~dGufqQ=tNx!OD*{tgMkcBb
z;M#C-0|V5ikVAq9+dMNC%~65B4%j$bI}bD`z<r^-XtJqbbZbl(?MA9o_Hk&Id;}>e
z>4%e}SX<sk?kh$o-N7re=tWyGsX!jC{xVCq5C@*~mjy@PhKz>aw62UlZe27M#|0Kq
zay!B+99FqmFAOd%r?n;4f7mZ~xf*n-rmYh|k31F=6H{!rfHQoePOL0rn4FInQohdB
z|6}}s24>WiBAY5t3PiMRawODkT(bqdTA+|LoTKA+&E=9do<`lpEN7qi<;%S5SM&25
z9PA1a##RAm=Lu5!5_*QOb||A;&k;U-T0C(i33_Z6Js+qb!>*L*bc=Q2eEEp&va@xG
z0MyzJIUFg&w%!n0YnvfgYwhEPs<~Y{!GJLbK4$%^R%MP{vYb*RW-OpwACTNFp~upd
z(_YFlt+2AwTe*5^E58Q=Y!sl+Dzzp}qVMYjMf?+UY&7y{{;1n56je$$)ha>S>P1A3
z9T$tnHwE(^cL8YQ?GE3Xe0N=i(QJ>_R1*wnIqRK|c-&}ySJ{X5JHX@}NJc-1JU7)Y
zH`02#lVk-_$@6(jZRThlV6g>MMkaQ1#LWw}Y769V*1dja-~PK`|NJV%@vi$WA%V2^
z@KC>_m)9HKgYwDVHi4Wcg~w=nq8X+Ient45-bcx&)^-Pvm+?e?q!P-VD3!A$`Re5%
zaRs&fwAJexQvy@f42oVT3i-8!Zr4KluHih_)D82nW8yf0EyRcRanPYDKn}|Fw2krV
zWcp!YcX;qLRRVhjT4IMqRr`Iekm&kP{!!n*%5XmWP;`g;?mhA|wG2FW?+WI0G}MS;
zBSw`b{mm-+i491==<Ml1*OcbAJR`XJ*x0FOWBI<M`sgE1apmCJ(G=-h#w(x5ZQ+>I
zQ%v|A1pww@GMsp_R7Y=k<^O%M2oO`bIvJ}a_!5=mGK{Ez$C{;`-zhAoqVjUjoso%k
zZ)2|a3yr#s<h)3?j%Hq<=)}wGRl;Z*b!)xV?drqva11lqw!MOCRRZ^O+KEDm4wcpj
z7ity<K;Oo=<VK`t=6vRiCxXxbQmz<#v#3l?NDdqVsb|X$QCdU<GV9Lb0AgZ$m!Kpe
zc9Zlov7i_JKUm`biAR&)-PhCC2l|x#oO&@;toaE~S-QGHdID@qmql>+up+MC^w0YO
z;%XBLvPMyd{qJnq&>*n<=t87&8%m@=1ex<q`JxwxB1OM1k@sbUh2?CWqPAH!iiGpJ
zV1_%(kb%Mabg`!`BQ^<R$l_36W28+VC)4Wzao4kGcMSkEN=lCB5b%iW8lBB@dW1Q8
zu^7rphc<n*CLT~#*&UQnU1EEYQ;(IdzQtlj5ADD$8eQjgV1vdF_3)TzIVh%A-r;l`
zFLq8McK2*G>h0^}ZLjYeM*S^y^Oq9<tVW!ScFJV*-odQ_FcuR?Q0Y@IHTpJop1^<r
zMvUYZYXZAl;n9VMz&^vVOypo)oIoy|2*5a-1Nm#Xog{4wxM{8#ULO@AMxAA~XA5zS
zAQ2*#$i-U^QW(z_<fEiH_USofi4%Rg(vyl~jKsa&=crx*`oYFy<x+jKbIky+-lcl&
z6jz{z@Xh;x>(emp?qnE{z-D;yP^uFHP)s|}_ny@+(RMwWp=NVFUsxp^ErCuL8$5#e
zy)oiV(wV!dv3+1s`d<p${}|~%??qV`cwnjkH(f4L`@zG9Qp}&;Q=VR=ebXiu)2eSc
zC3<N@7VAXZ7EYYK>e+XRN5jRH&_jS+q95J$@Czta{i1js+ox$Be4{rS?{wOY92?EP
zqB!7lqt@QIA3GeW&sAYDLk6T0>J!F_J03+CKw>A&AvHxh3r{viN^55dH!E*JE(kv3
zRZbES00{xyNWZX+u5Au7J2D``SWw&@J?*lVl=i{?>war$N9FRe;I?|k8>~M+z~3LL
zoSzr&m2EXzI42nB-#8I%QV1?cm)7WNTk1vbRq!NZ!~jQ1&`zrtPP{S^0BXO5JB+C8
zMUC4-6@jW#lmd6fB6XIPaJnbVZ5dJ}p1N4ul#l+mVf<+i?Cz>Xkj0uhpfSU$%Zxl^
zG1bQNgjrX}xPGXBOlvOT+LvqR;A;?SqT6w?$i&VWB#iLWC!yxc9J4&FQtt73{qEJ~
zotlaT>ESX{<wPT4v^CSKApBEnYwOcdwvJ=PvHmUy%Ffpl?r>PYUz=0@!RE?>V5xWH
zK%>nJ#>rj^w%BNyMiqXlV+cXUN)J$Kva(Bb2IL1wIm_^vw)^K95mDAj0M*=Zprh<A
z0U4|7gAlA}zf=xwFs$9IGM_5T*?#@uq<gAfeYNLCC<q8VDaH0-6<tm-qnY9cw>b&i
zHxZl+4?FZ%ZFj43yN9c(q}jIKpA}5oAMNQMXYuZfcf$Ow|KcCI_QQw)+b^&36^;G&
zLfI3^jt$MB$jibsLG!Q`UfoFDA^B6SxmCzZbAU%Z%6@rCn1C`6cTdx~b?Klpn>;WK
znXaC4PbNk!o!2jQ@x9Xy<ZQVfjk|oUX8)Su+tvx$il6;<PsEzGtAduRT^L@GyE3k5
ziYk!wVj@l5o=u>GNWK0Xnd`<OO<TWSH(vx0wS`j^)GKN^-4>(EWHvMOH4+0j^X@?R
zs+R1M!ifp$EG*a-)PFrF?xrS04={sakG2WiB2^0nJYU9J`>OVm)EUhH>1d%+rSXb_
z(@7;b8sd|Ocd$DlApltWh1bdxQQj*`j>LP4*RQbgLaSocEK2E3<We*gfLEN|G(CCI
z%Zk%f)K0y2uya<b51`d6oVT4ayV(rm7a7GmO?}1KimeXlIYuGw^)l>bb0Ti;Bc$|~
z<4sIE<DsUdrT}lj%Vk$5r`mMYAu{{C_K85j#&`gD(``{<GD4g-%a#}sLy3*O@rEX$
z;?HtZf8TXK_T<YuCF_^tCQX8A+mB17>J=D?h}AyO9Xy^D_M|$dih;qJb7hfWtmXeI
ztV+K;&$pm(*JdYiW^wgBOrL7G6tMiIeXP}zxDIV|gx8fWEPCpFKs)AAFAqanLkWi)
z2IoOHo&QR=@!a>0Bsa=9X3KH`O{WPJw!LlbPkMFZ4Z4Q>QBMZ+1pws;jfw1`KWRbz
z&r$XZVL|d!pid`sJ}#LLQ(61$8OJ8ziPAimP3g8}ouKG?=wU^7@WxSxhs{c7Zk{^f
z&9P$C&1cM_$7UMe1J8J3O<YV~w}p1Jp;|eD2A8ejD2Wlo$}Go}E>AWnuk1FWi_)2F
z2m6eZzk7|P*>s><Pf~VtfBuRZWz(Tab8~t62syRC0uaYWhoFaM-DPOS#11+$6&h(7
zmfIofW;1Kr($65=kF_F!^J`1WK1dYggTf_9&p_W&TNL~c8~&Gh3AuwKKMC0EWnVzG
zwt^)(wmqPRq?o6c+q656!au6&3;irHkj~x^%S^)h#USAGTgFa)e_BXNw;|BNrk-Xn
zSSf>cu|v8h!WkLR1#C1@5fq*tgB}XYYd~mPJa{*BO;c-bf2k&*bK2>lnYaHMSpxxQ
zo+=%qjsP4-js3>U!iIwat_*-!rmUAzt5A6wlyKRbz%Ck0(IAa`$6xomN%_wop4{n;
zzQ^<Y%F2h@JBTk4Sr7ufS~725lCC!>_?(%iS;?0OTCg)VrzgLIbum}HKaIbjH|^!T
z!s>gDWz2BWZCv5xb9&W@OUM0=o;po%=hi1~_sd>G5)C2fP=A=$FDoJ3Hd&$L^h2PA
z#x_Ph5&#JDs+x=%GXlUOQvjS3iudH(UGD?iO<nhjHrh1pRhc;k-YVc)g^DCPN)tHE
z2Ypt3li|*x;OFOm=QAT*75)2>`^OxGxZcf-5gsM$qPO~9C_Wh#SRtMoqxX<d8st=n
z%VHi0<mQv>uBX4+XQKhpTk4vmo|0eX+)ZbY%cj`Z_lR;vUx7lTE*$0T%{)f+V#!0G
zaT0Qx|8OpnR@*ppBGVz*OH#RGXd54ZHTas=h08pryhBD2fDWCiZ$^2%jt5qFEb5os
z*k<6JfNoCyR#aFNabCCy#wnD4)BZN`;mt!!84JRu_ZLKVe8Pmj^o6cJ!;Wgo2yk>F
z=UDmuw}k_Vhsg>|LRC{+Rjn7tb-g1v_KRB4ip_0ff-MbuXe&YHCi@VF(rtP?U@Q|t
zj6tUD9NQsqZ5Sk;K39SS!6F5Xr`8werO@wG>D{1nS8k`H9W+=z%uE76(#s~M_Tk?e
zb$(5O14BGQ#4GFvPh86&uzrZEw%d61bRZ8?RTbNMEvb;srlN!3Qay&<;Ztdu&*@k?
zSBw8u{6fd88fc1YszCQ0LL};xq7d+SC-X1~U5~3JZ}%Cnxot(AG)>w*pr|<N4(hSl
z^!Osj1kbs?zrJ%=CERVaKDBQRZQZLrKx+hW?bm7B(=nzVat7b|F?y^|_S(NY8Q$bf
z2M}SgV+R5vNxjHMI+4Kvn0C7jZ{GdR^#)&3q3`+RY!&jhR?`Cu#<dU$2uN$Tz1@sg
zku|%P@|BYwBnoZad0R7M6)On*>jw(boXqc`611G;D%vA~zzH}&%3cQeyPZ`&xS(x0
z0EMV6754%-!6X3MR-lnp0VNYyycOnij~CS<s;O0~z6Ck#gW+Cr_84adJWcAO0~(pm
zv7J6#H})J7B~Wpq;sdevEpiA!-tCM+!TsOu;AD%3`?H{#26wC&Km`nI!0+o0;DCjy
zjzLOk5<Vhl#u;Sy-+P#8H09WpM+gT>OxRZwg~CaN2&Hz#SM&JO4y3FAtzd0jd5F$s
zKuC)C#OPzocXD`-*POFWId8N|y@d27h}$WUuNW&40{}1?75(9Q%3zIRcb7Cx+8lr^
zrSn0<roIdJvV$v10n6&N=P4uAB=`&)v>j^QZcHxf>>N9N)wE<ileEjI`l=v=1eJqA
z&0!Om8u0!7ZWXxd0S&U>o$9;^kLT;J*lluiIOTxgM)EoPN5rQ_8Gsf?ixY}DX@!&%
z5Oj0johOD2=Q$D{-==uRDcjqDCgslKY|A#I#}^@ETM=jET-vMiKnXZ&*l`#9<W0MW
zki<?%Ag<xmm@l177-_dsCzr$YIyVr&C>N+v5yp_p-P%ES9(RnFQ}vA&s?|$3#NO7s
zUW`=I+nmY_9(}Z+FLF=PA%w64XDj%yuOl&nXVy9j&$$z9KY@wTZ?#Hb)&rCo81KWB
zmiTBcl5-N}@o8{A`xQC+--Csh1_O|O0_`OZz6SX9MO#3lE1bvO^#MLMIGViz407La
z;TQD2W|_HXY700g><`8rLL3-&Nc$2hdGDgOtGI16Ku<NwFwtG8p!$$y@Wb4BiO83=
zQ=o@kVvIIH5KfL53!;>=GKp&uHOp1oy`SRKh0+Mk2k@SjF{^-iAUT_L3hy~z5OH!j
z?698FM4Mo~6U1(uwnc%#(*KqPEX9Y8T$J+LRN2sHU0A>6G$fMk287~W#9JSWIzc@t
z^%MKQ+#H+wd5s~QwYq7QaXa>~2H`zc*-ZIhuhc4sQY%mXoOpKLw`kX1xg$#)Dylm=
zR+F?!7oy*-Jkf1BTbB*l4T%YTQIpqUYS}=(mSC4-L>4Icg>bExMIv?Ew+7RPV6K=!
z8E~noAhFEZ{gg~`ZFkA5bSkCq5mABpvf4N#vQY^3D+rO;CY?|V)-xF>7tO0vw%UJl
zvv^UE)6$*I$u@A<K@l=>QO`Ql2={m=CcpvEg`GJkOY5RnA?YNXIz(2iuGn9Iysi<D
z*PU;Qm=;}j1oD)JWc`0IvJ$ZYSxc{)OINppi<^{)?SJ9@;w?vZ<RUoYGnXk(pu;97
z4oQ8VV|Qhtysc*Z6S$Iqu^+A7OYtu;nXC<LowBygDU+Q+V(;)5&b1%xQADU<%<#t&
zB$|}}_T3xUOM?0aWf@}?Sk4$@m9x0qKfayi2^&jG9lGT_!Uvo2C$XXbF5N#~@s<P1
zg-3QEv0xLCTypZ4K$pd|C@tKb(MooEm^OM)KDK!_v#|M>!2i!CZS(8BJ;+!`EzTSh
zps13Bk^ke9S#mIe&Qd4Nrnac@cz^!<Uw#PeO$=zNhtt%gOz)J8i4h5@Vwn8=egrH*
zFrvb~j6;4Mo8*Jv{X+8W{dJdv%S6ufg$MhL84I&61%F&E8TEl9CxHj$<~&HZ>$jt4
zCqRTQ*sOb8^Pw_~MZyg>%&@!jr<*QBgf7`kPbk)`FZ<i02azhd+iBfp9~1tZa^CQV
zM+Ro6NTzHXlvhp4N1bK={lLot&`NpOz3S3<5t`aJOsx0-2?ctR!p}ZQESR-t9CQQ4
zitp0sersU?p6#>oi~09(w=kcrH?rk6vwvcK!R*BIC(dR$ZyM+5xQ~SXMUw+O4vWK^
z((M?nhlvT_1-SWg-d;bOyfza*-ljggpCuX^7WM=~R_?v_)@;OU-QNuc415?#a!g{q
z!G;-Vgzf?C0ALM2lL`E+E!oL7U989`U%fuS1jrBIsIF^t{VWvzvK0O@!&!T<(#c+#
zj7G%!w@#o&lL9(9U_-tX5Bu<Qb?C$5NaCIg4G4~pmQ=s__rzSqgXq3o`QKmcuTC~V
zrvAVU_5b#1usq<Sp3D992Y&liae$?1UIgYZ5Bl4$QUK^T4?;ky{<oj_(;t%mxM|G)
z(`WDf|2O@g*y+c3{eNeq<wJkYDfp67R-Rb?eZxkSwu9t=`QM80|Fvi*nna|eZCa@v
zb({9{xeeM(V^I`$ffMQXpX7nVurTCSEo*4-vcn#fZ~kRL<!Clt=%n<WzzCpZ>^sLN
z`|&0VmCFQRj3NWSxB1Qdeb|qyetX(K2Y2Jvc;55u-{PRI2S!#x#dIOn6I%zLJ6t2k
z3Mi(cV-fGF7M=|^8h<dR@q*CzFYYI2p`C@A{%JY?b$+tAVQzF%-HdrZjcsK&VAy1^
zzW{zCp-M{f2SX6J1Qy(>+Q_=`y{jVm-{h*^BLlJtb9y<VOwE`z$4Ce;SVU#sNpL^5
z*k<y_Ic;L0Hf|4!&H4I&IWx*7czKGl_NupB_Au|17yidsWa+}XVc@H8AB_~NGHCvO
z<C;`{Y&a{}YgPd2v>r)6t-)ppFVSZY?ZwT08r_`#&h9U->_n5t4I?j!qEk_4ikBrL
zc>R{E=wJOfAdsK+e|dlT5bxjSB#a#Weuhp=@G|n)Z76bD*3H>x>>myxQ4Ke4x6{7o
zFX@p#z2!XtPTIrES<du#0Kf9U>pc$TKl~fn4){(%xeklxL%*fG%sPUtW6@r-e3woK
zB)`Om1|I+M)Zop>`!Oq=pPlmwOnZNytNjQ_5kj_Hj=y-51M}WmyQcBx?umKvpY-Rw
z<K}5h8%a@bfBV~1$!6%s29dr8geU9X?M{^N^52k!7=BES34fCex{y?njr8v)%L8!C
z2-A#S;{srjX!V~o>cdPhar6Pael_+}(Z4^QxiIeQrN5?3qeKw~o*L06xB3rH0$f9u
z3<gfy-y7>JCz#pN!V4GK1ctjxemUE-A2W;vY!+bB?md(G?Zcvr#@^eTAP=liLIrw@
z^Xq7ScuvVF4~^&_q^jdy%D{~MHY+Bp4i5wXPsGL|YeTPrlt-wmZtsr;!21c%w8(mm
z4FtA--JAb?E1Gj}Z-zXgQVB&GxNr@=<d2;Q{&=r3;2``@>*!bibLR;VCr;Y&gt6fQ
z+28E$@gHmct$dS=TakFz!0#0jz{~f60NPr^bSVp7ez?81CI0uf_um`kO%?`RldQ+~
z(m#5Yc+Fr2J=;zfD1PT>0htn-CYQ4j8ZDbOl91HZZMOfkul++(<xd-Sm>#g3CKD*{
z*#GpOzyAsl0{~AE&t>>)<JLcpI9QDg*bFqaFC>xvPyU37G_VF^OV{pj7Qf8EKfDE;
zQ(%bnS-dv>FP{b|i~*KbIh)h?A9OAM+YOn-0I0C?ZSQXk2_7E&-@B*qWfUg=Zx=#<
zm#6BlXl8-*&5=~s9=>M>5EL}3sxgO+#th4imhKc<!6LggH_paiBCtxeIgwZ_zlUBC
zILkAJq$&Wp<t0Me;7nZe|I;!dk_M)4(QD9P&JFri(s`{ln#-HOKIE=LS02cNHqKc#
z+D>T#u#oGc;x)bVegVDCX1E!j_ZF4LGxBPcCX@ibLI?m%_XUJ_@c+#y|8X}t+D&Um
z67SxnU}9lO6gwfREYgLBZ6*mPaO?WTfc(gX<m<J>BqSn^fdsX)s!EFV+_3X&Ct!We
zlQmqNVyYIK1l|lL`?aFs<I(#AC2_s8U9J~)y}k2%AOK3g>%Kqm%(^Gkhs~#}Zg;36
z0^pGTlAQ_rzs=Mux?~v4+M^>WvWwI%G=3SXWH33-NBMQSNPdivFvG-f1HH6~04V(1
zeF1reJDd;ytZw91)oPbNzjMR&Bu9zy(P&7GD2L4&O-NxTR0mC8FHHj4z}3y@qC1Fr
z;nciHf3kqZ(-TVtP(*3iU8rn~*g%bZvi}v@0gM;;!GrAy1tKS>i{>7Tz!LhGL!4(4
z5)vfg{0KA&t2oi}Z{NOjdGJ8$qc|8vmXeY0)x?BqSj=~%4t#R0+`9sdxSGRMVR#+a
z1qLOvoE*>(X-tyZg=;dhX?J-2jl~mEP_+$fJk82c)%Q2d>8~4TSPOi8&Bd(i(JKo_
z5gTm?(v!-%ypb0u9D`4bPDW;e%tRQ%y|OoM5*PCY(xpl_F9SaEBHZn-dy%!-b9+(Q
z+LD!u0%$y)Ad1BbgP`dB!L<3}J21@Y-XvQUp)m;!lOioA^jOp<b(BV>Ofmz6(dqcO
z{4vu1wv3bc?|7?K?fE52GbX}jD02C4jPa5_A}Ruc_22Z)MI-PzjsW6naq$N+l#5GW
z6+M8P5=Ga`dpCPE&v0SizO56!Ab=JM)YcHrzM7}S^7`??T?%sG6>Gt90>PcAz`tWZ
z?!iVp=}#vIisv-&a&P^LS`y-Uv2Z2dZL6}Uw|LoRc>q>GvB|$OcWeon8Q*30IkxqF
z(>yosuLJz6Wst2!uwRQKCF{Wi>Yi&b?mLMeZlV*^o&j78`rO9^>m`H$ltciSWD{BL
zcgF)^jhVbbxNMTAJN*SXy`7Z*+o%yehK&tOO-}AaZS%Yq{H$d5h-ofCTW62&D|O@g
zAn@B^cP{J`ue*FB&|wcRqE+vlfX8($OG!-~wnC+Zv}Grt&~H1<Vr?aKo5*$8p}vc3
ztUW8u$M?=jiuGqVTnqeDT6N-Zp1820gK4;vb^DJqI|Jpl$J?VbpwBz-_j_*i6D3`*
zPUc17oKGT(b$gL_e+N^50k^a?9kD=!n+2{nq1m!F<zqSz>Yo8dafAoCcYOYCQd?-;
z=e1>g0xpvbwUH+r_Ke_1fVAm+0YFm1QzfG#7~+9)=Z}EZUcNraA6ss#JmvP1X&>gk
z-SnR?f^&}Zxz0?Wfpb24I9e>n1ZI1<+C8nkii9+<115`xT_QsM=Q&e>(NyD?Fc7W*
zt*D<o%&JF}U$5`%%DFwVY5&8Oqp4ftuwR~3b>bwz?k`kMHyp2!$`wCLC;4#_hr%=Z
zH=~@*J_S06w~k*xQ5vVoo&~VQ<sl=71!3O*+KdHc1nx(_`=<%r4CF2DGb|KhfGC&<
zHx)bV3PE~2N}2CSCtf*K`uoF$8cuQnC_23_xwM2-4DRvdbG^~#TOA4RHFxC2+%cEs
zGf>Dl11SSTs|5>S3DOF8m(8XuwrXr=m1(gZ?Ej~4UKRs(B=l2i-PWKFR8NXR#&_wT
z*Kdd@7#MB*AbqYU*=ohbe~^9WvE)C_s`{9@sJWg3J0(#nyl$^s%IQ;kyjk4{lt`IX
zMwUKWSYKz62r)<+O~lknTpqJGx!G-wbB{~2F!Hvl_I?$(0`UhG|4;ztl|ml&CVH?o
zN#2k!O)3<BwEv%O>|bvwi<c+>)_;<ysXhzEA>~~U7_f-5YXBh$q6>4c1?F^6MQCCo
z?akF`nGesEt8<w_2@`opAV03H6@sqdPSek0BZrS@kRE-+YVvbtzdpCE@8f?7Q-SLK
zqr1V+t^tTg0m3_y`kq080dLO5qUO^O#CO!H74+{*0fuFSEsnCexDjHN?>`Oi2sHo=
zH&K5<C1hSp((`*~2Yg8qqUKIZdZ9enXo*I7a7D_*q-X$%v6KN;mFYx8rFsw!w;<~D
zgb2ICCb}JQy^H-bl)C<O4(^L>(~a4DiI|G+$ifGs<s>9R{Xe%uEJl_TaB*oc!;iGw
zKYC8Q0*?Xx283N(K$~9w8=&>q$jzl^Y%J_6jX%|$aK=eIck@~#h;7sAjvPl$qz5uX
z3NX3o1Vig|ihC~Kpa(42Qqa;y_BCF&@0v6TaY}fzf_{o`6Jf)aGkxZdI!pf52L9zv
zzN6pWfFg)HiSu^&n@-`*l!G{6dG4A8FZQMgyJB>u?-STC+wV@3hG$AfST@9wj%mE<
zZMc*&@JFV*QLeGwCCXQ;D3SuoToQhguRle~bv!C9bp4Y|Ml$n*<+Z)g+Fh4n>7Csv
z%dBHeU}40XO~jJi@rXv!#ERgHxT8Tk1fXvw7V4~93EgnZ@sUiF=JOqRO5AZZ0Rn?T
zV0VXn&r>NE)^fd|!5(j{1b|+}&AN0Ahmm<I)xtpegV=boOst4|02e6lhSJ@h1){j|
ztT)~I5+ok{y30kgn&=)zbrxtk-ErQK1l3yH2p?QrKMpodAKMB=-8u7hc0GwbkqcH_
z{^+(I@4*YY;}nE-up3&<KO8PC{M=vyI2-QO+bnCHJOpmyKpW9ZZMT~QVBZM?%$Y?1
zTL&}iAqb5NdwTYr$E#khBR!{!69*xHAMa+^-?|$`*{@?M6piO%uwrQEYcMDExX&Lp
z6b9ZzTxYjsGUnnP?_Pr#Y0OnOXUzHOR<jK&_Urf~)O9RY==hs8_^FrH0)0G&&8{(~
z4P|f7E?NuQb`}!-mtOZfK3bolt5c6SJ&9LKok3cgj;Kcp`rs5Ez|=4$)bYWL_PI^P
zx6Nl8@dNGF&>qrOwn2@>{GmDhvu(=Hl}^L<fIDUC-pFm+34Q}d=_f_>2~lHxuj|Vs
zRkAzM2Iv*=k3mdx(F$nkE@3iM?|cJV65cu?5TOdby>gl;W?-3I2)T4K5NjWj#-_nx
zO#!j*Olg&r(ChC^qR_YMJsCc_D~7lK0_OqBV9`fIeb#N|m6NJ7*B!abUog(xh?-+I
z^_6Gzpo#XUt=deV7a>un++QE@xb=Qg_EuIjalag{d^Hj*eSg^g&e^UL0B|-dyY8G8
zo`hxoZt|Oh9}MGUzNL3TW2iA4rjO_bSP->AHYLY<V-jxhS2mrUm)?6oK}@udU?xp$
z$?U>yRHnr7yY*GZ0nU;4>?uA6>6l=IF<U#l-1;EF&yd?ht;y-N#?aAyp2j0|uDXVt
zD~vg3srNL7i;k>?lNI!6kgt#-r|8`P!r9(z2Q3WH?fy~QVLu!E+B<f#t`ttV1l`VR
zi9PUW2W}lYa30_}$SnnHofp=2#q*#WEor!S@|7;ZH+hZu>_xEzRXFsf6w*bahEQzn
zUkq1(G{yTaZdBHk-ePu6XTabsG@5Shja@lxJsBfiw|2&5w(uQSD%C7AADV<WO%vqQ
zN_w?-+o78}apGooNP%tt?S)uvS*8@xCD6&eb1r>a4^-T~sNY@&&$B&z@Nnl8gvL!d
zJIEVzNBu<T?(S3siM7KO+|%SFWntNOh%o5rZ>)B6*toFfJu5h{P5+zzI)Dsh0O-uW
zkk&e^b=L!8i#%Vf1`rGtjV!XYEH!qR=2Jnn6%{)@6Fd{jTn>AQGdri7Oj{m=^bXw7
z$%L=R0-j^SZ<Yd%-G|wk?QP+^7P|qNo(*Wq_37e~WRZaPPQK-|*7?@Z%Ep>jxr7VQ
zGCqzd1=MF9Qg4D7jQxD@-9pkb_fUe)Aqa+@<_<Rs($7I<s<XP<#x?oHKszIge=vS?
z3BiZir?G$3<QQeaj)c)2-CxnlEUV5c9OHb-hLCnH?Vu2cx9n5`oM^)W(wmbXdXqYR
zwjyF=h*v+f*1Mx1qE^usPfKxCDlA@1SXDY)wmB=JkuM2z@;#T{7sz|FgiWsI-cj?4
z$7Z12e$zd+lK0LdD8)TcjGDCS0vL9d%!EN)2j{wJ%E32Z3D9`<PhAK_ZaiHVyTkco
z4dZxr_pY~2Uf2!TE9GmE1DxugjC{=NZyR`quW#MPLz>@AjjVWH3I`yj8@+Y}NdDqQ
z{1f{qr4vdE3tx0-@@PA611vk?7h3P#X?f^lu8%D<fX35wpXH=rE~-0OR4O_`Dgd<-
z`Q^(c;eB1YcuU}b7i{OA`wthupF?_pOFl`A`|8_`^-}&>=PQiP{ozu&xJMyV%QdF7
z7&7B?^>^jshkYC*Mh87!$}2#JO-Il5b2BsMULJ>lfHyJt<r>y)D5(;D0US`tJLdEX
zi$$vNW#3Hl$tFJ$!n&c+`jc4^fB|@*RC8R|RUkR&4&w>Nbv)f57aya=HxCg!kF$Z3
zeCo%e<r0J9*Z)F<&?OD82<n!4vS5DVh=E~tuohSF^d>DL@+nR1cN!Kg(xaZtQk=N?
zhd(`vaz5TDtmq>f9p6uruL%=8xjCivvb*5S&o~*;ztT59@?~$tz(bcI0=vvEOerG2
z8Ys3Uz|!<cr3d3K9NG8gPY|^y+S|Y0e<2T$p9PVRH*e&2=NS<RrS2-Gwr6km#oK5W
zYgNWbBA7}cesq-JG-olN#Cep~uY(9V&uCL6(7QMs0eY*08BIsOA<`1EZOh&XFrr?H
z1sKm=>s;ZnnSBP?H8Az|+B0Ju<|JhS=oh50Qgfep*2`$G0M#U<yVziQTJ!YD(b3ts
z@(p$uBY15s-p<b`xjOO=Vz02I9SPukyM%Pl-27H&ub}2paMcskPU9s)^Lr#y%_lVb
z1af|?FP>boD4ox#hHh#I{GH7!;mP{6v9{A<sLQgSsYP)!<`kqPT6#dA$u35c2JY%C
zvH|4G>XRmPWYE0h$jnHzcmS0wk~;Bxrp9RXzD2^)d)WA6TQND5J*{_l9|i3E3W#$M
zEb<C+ac?v7YikW6t^5E;3UtX&5olG<3`CU0V_xR{FGl=6hUhW*{t_rTA8oe$%h`T?
zYV)ImM4o^bn+1v5aLBujZ0Oho<?p`0O^#K2i%Kn`@J63keD8k;i?&F!q7V1sI)PO4
zJw$GAK+qTDhEi6w01#0V4d?``tl4SjA`S~ut;`(ln~u0V*;<ocdkb*d8QIv35wd0#
z+U^cFYbqLWK%o=0X!{g7<=`btcCqz(e@qz4$LQND9)4LAM1ES2T|{oE*V|@s$AWG=
z9x$6_jW_wBP>IqF^2s?%z-G_YmG;ora+HD;pleQH32ZjKF$|5q(c5)biJG;`?!v;%
zG|F1`1s6Em>Yj)vbv|<-sKP*5$m!t%xps7|xmx8HDmorz=d(?V?X9TH^+Plq(G-c*
z1na>MN_LWkEE+D^jK<Ub9NvinaO|_LXCxB_)b(coB@ZrmkS1MX4Rw~;4E}-S)gtbN
z$>pOYI6CQlJ-_1JOn_+L`dBI%@USk*MyQ{}(lY`mw#>x8RXlTQKi<;ZER#5V-?DVm
z(R+`#01-YospN4sc3$>^a>rMO7Y|g)!Ce(pn}f#Hqg2XanT{>(b7wbob_@*<7Oc9x
z#&YiR>m#+98jpVjU>`7^yLB035tDr4)(B!|Z9*Z5N1C*VLGd_2Icv5rdSlz`3~_er
zJDuh6J_a`!h+4U1St34>Onf<@m$C1zL)eXHn>cFyd{wB*b!dfi+E3jZZEJG+(Fwb)
zMr9Iq!q_nK9(??$9p^JkXfhK1YZY0~sq@Lc5_^V+qsnk{i}-qJe9f0wc`DeXWm32#
zrs^-#T`Hy9pE$-CL<#vv#T4eNF|%wjTBX#9dlzGBX9n#oLkn1CcHQR1Y_fe39zv~9
z(kN)KCaP^RtCzK%o%Q(c#lF*>I^8j-@E#s+;^`I-Y_7DNduLEGG#@JJhv>DAOE4OT
z6M7t&VGf9i2ljySdkS(#uZ?RqCUas#Q>Ee}4`<LCJICg>L7r<JNrt{|7f~I7ci1ro
z*Q_#=YL@ZEHEqN0z0$R~j$5t90oR-6`#GxhmohRlJh(l|nyBqHs|-E~??(@Q&1r+X
zX#2C?!BuY_9@Rx192GWpmqK0IrpJhUZYKA(xH52!jT6tX8J7TsQ#VSK&9QJmoM#Dc
z{wi*VF|IAi{f5M0cCkTn(?iI;GrX_PsDjX~9epEo=^{Bsb>YnXbo$d^nV7&*>#M-)
zpdx?Xm8bVzI|{6|(Mo^j*}$(}qHiu(Q%7w+{fyqrcEZ=-w$bVgorg<MV8zCs?k&9W
z<iZ>Sj#aGf>jgxBlnIzibkB;))nq<6i;0&lKxfN6avm;`*Kk|~Z4a`nx+asCym5OL
zuptbIL!g5sT~;2F(dbdrBHqtdg-qjIXdP3r5<-qUF2gCQ+G+%@=h&)^@4_FkU$72r
zBd9Uzwgn^^#7>P?0>A?s1#OWT+a$#ukojz$uZGPE!t9L-v&mQiX)Nixjgp8GM_=;(
zR_<9P67{w#LN_v-0G0=NscU0BSR#aWpRsv$Mz*%1w4p=_Jka(Yw-2M(p38xi8J#91
z^Ss^3;0|mFq+QJ%z16)Vy-mZpXbH_|lmN@v?*$HhyVmi=uF3Az8x=lj<FBzljAk3?
zv>#i#9iEQd3-EL%BqC38%=5xsPicN(*}x!VaJ!4F!Fn|(kd{lsl?A=9YwQEFtwv=)
z>rM)MAY_K|_9dZCsoB*x2g8LuGR5nz^rm{zcG7c)4m-@Auk-Zz^O$RF185(<2h<6w
zueY`(dd`F2?R7K%T9m+2)Byq*>po8z(T&72zC5Qy#`))L8G8$dWyQ<wRE3SEirxCp
z5@{$YDFJNZy0KnX^x?sp!Aw0GmvO`9)l<vH%e?BF?`4IMbdH{c*9}}ew<96>w6*JI
zY*#%mUQ7)DWa}>hGoW9i-t$AZ3{3xYpY!$J;$;zs+>q}opWREF0iQ`}p}wuiwx;a*
z3cCjx_fODy64xADcnP~bPjvV&?U8OS`ibsk>iEebvIi;jdyM*Es3h-{8Cii5C0S`k
z9tVPNADtsfJN7PL5EKN~2V0mH`6vt=48NI9<*~UDiO1>Z#M}EQ{q8NbcCp&%P5o?f
zx&c>_A8qx*TlSqX>1_#T*76TIQ{_`fyfYioqpEyZr%SQTFA_=Z`r^<(2?=qxZca1W
zv;YABX!AsY;IiE4yY-R+@mj1zhzd%XOz4~542b&j#u_<}Dz$aK0C%Bf-3j>ma0)Mp
zeP+U}&NmhO+?kG1-LqchY{7+O%hO#4%x=uvRb#gWS&`qqnXVwOuXjN8l-VVu|J=ge
zo5!U}Qnm|Ngs%Q(@^Q58q7?G)U}4_R>UZ?X5VzZoD_oeDN{!Fj8_9442cY9$0igS-
z!I1J?G&SJyM-F+%6Vb%F(5tehlb;YqfqpFwzKU(4)ukLsazpM}bkr%tv&q?TDUn+5
zwIZ=v80Le6rCr!>b?y#PKzvU4_Trn1WT}-}y3kPuH{oW{=1w6{z9@AVyM;lwOrP@(
z65~tGxhvL;>{5*V%o@cbOcPAEUofs(`JyHlt!|&*n^O~SHF?&Xd+K}qqG9Ty78jd<
zn8ttzd=N4VU2vyH8_kzQTj(L3)WNOP2@{2TXdA?iw6nH{;}0Gx7Xg?W$2Q&FZLf_w
zY#wU)l39J*09e9!aNMh)%r*t}T_4mLL|G#+kd99b60O*G-)0`5L%DA-kGeOi6f0Zr
z3yX@2AJVSqjC3q1dSU=vxw@$O?nH%P<i3Xe92S58LCkh%a6m(N>6z|@FI^^-$aM6U
zv=A5Ou^|dB=IaPe;dX@3nf5rW`gHXuX+WFHd;l|E?e<M%UXcd{BW`;WMo=i3u$p9o
zk3pGAg=<7lOr_{8b7Zm6<jdEJ8HjT0aZlMZNAuO9RVlSVOI!?9x+|OIhaH1hgUxg!
znNIJgD#a|_*ArSdG82vTO4r|U;5Z*;YP+2=58gvXNs}YYHA?=LIh3JZ3sdMUli-K5
z5rvp`CvGQzwn@o!M$Hrcj`~HaP}K8cI^_3ChkrW<#9zwRDoGvHe+lCve135mhs|Bu
zI3LgOlE*#jc-QVD-gL*s-syH8Aegb~ADAvd_j<OJg7GRdV;7?Kt`C#lWfxYp$}zi-
z%~oX2ryng~I9I*HP8IFuqi4B>Xw$v7M6!Pr9$;BQC4Lk{i3VUT&@7fd%$7VzELz3s
zLpo<SDDpWUW!oui?j;nv=LzaD@?hJ(*_pZobetCLZzqI5`lweR00;I~yd3WH7|Qte
z07EOjGMmx@2$X5^VA?}D-h=5HQ2TA5C6~)oqPdtk|Eg9>y#53OX<Joh$UG09!@-%b
zeT#5vA<}%xzEgeAP0!#dz;9UUX+M<CYb%Bh3UOPb8G|>+-Dsv!ukb}AQi&lz<yhw3
zL&wQW20u|_lm&=k>L&yljnb;pR5E}-sVAOE#v7HLkDmV2ZL$?du#^+e&B)Sp>?$@c
z&T!O}>=C3who&v1`Mwkm%NG}d{j^1m!;S|XybsE&76%`{2ek-#($I24C99WaJ9=z@
z3>7m#k3YGEldE=J2+4bNU>;XI?F%wSF~2>`-Rp#~1RbB|y`$br(Y6qZLOg&#7p4x!
zxF*;ctJ)r<?wc>vIxHT5AlrnHGuKxcX~U(Q0hUy!zvi#9SYgWn2!V?Q;Yvb(3JA%i
z(oTTqt$fG1PWucR=dJ>w*;-Is;Tu>wPHdX%shjg7>ZQH2%A4H!EA<MC@H-KpTFW`s
zwY(H~`EkDS0_(f#TeW%@!cXl(cbG~YuuAXSQ5ejWRh%6GbNsPQQ|*Rknp&^=K#=t4
z^V1ru%B2e(u2`ok1c@rmT-vsUx2edBMZrsD)FO2r58ubcDSY(lWr|GY(Uu;|sIWLA
zi9Iq#gu9t^hoaJ-#zg6FjqX?<hayP;+&EiSGF&XpoTEAVHe)i>DhVg8Sfr$hq-<o*
z(Pqla%myTF#U5v_aOAngM)3L5Mitvmh})=_J`s!!<ej@ODm8OO9%>NK|0X|`5A#$;
zDUaUcF~G*=>Vp1hq+xErCkBjJQwZ~HXR5py)k%AXn<iD2;<!88h*!K7@dQGR0kpK*
zn!N8REoJ%KVv=mxV6^X#*ladl^*smG5oCBs-X4aNSe62WTRFBY?S5sh`vYqR&$e92
zQP#-`-3~w1y}q^E7%on5GMh)fsgtl}4cx0J^Qjm~^8HRI$yAHKeCUe{eWPAsWHRkz
zSJJBDtsiw@oZxiyEKX{02o_G``l{qtNCmGDRzDhp(MN%*^yH@S2;#R_KD+glZ!wO}
zQCQ}lDRJPukexs{Z7@}tDD=AcnOo-NM>vLY2|CzU;#t@`zQ+<DzSXJ=x|g@a%C)U|
zmb*uIuRK;?6!Mg*1)F`FSsJapbr&z~T@}oIY5QXCw`0Z;Bc3Nj@b3r*C9$N1+-cua
zVhuFW7zwW7NXX?X<00OUMEI8CG_H8HJ0yN=+Lpt#+Pb}9c<F}Ig0mjC=f+IvNKQ`f
z&mH1GWwT2AWukqo4ne{d_8IDWLt{)JU#XOr0^|Nd%rnj>I7Y9XhgTMCk1C2IC-;@M
zC;Bh69cQrziv`q^1pXg;Zy8qAx_%D}QX*2KQUVepE!`y{h*HuW($d`^-QA5M-Q9xH
zEIOn?x}}@-K1=sGXCL==pWpR<{(oD_y4IZYnNQsJxW^bbjUv=&XO|^59hZOMRhsTK
zcYbiywI@zu>M&8w!rb!{SiH?q923(XXJ%3-sotlv_z@xrX|%=n_$FsOUSNbI8;!$l
zpnK}Dp*U-H6~@VwM-n({aveFhO8b;=W@tvq>CJn`rYo*3niR!~IGDuB(z4VMU7c%b
z<|H}Ja}Y`4KIX3XTY@^Z!25T$I5)m~nLeLpAKl)|ZbxQWYoIJ4o4@fFajYCXoO&R0
zzeDr<6-IVK`6vY&dgq>NvC(-of1-r7p_fG&mvOzgxzykT07q<L{h61EqAK7;+>WbR
z$0I@xcMJbXdJjLs?Np_cAahI~N#mK+1kgMBK%-I@J<q~5=b^5hhXveBAF&x!Yx~=t
z7H_Nv`t~LcIriJW>kbyN=ZE?5u@*P=F0*&;vt&+iiO#quxgVf5x_5tCTq-SHd^5In
zwt}C0EeKy{_wfk%vWt?4>}aEwVRNrt^))<EqNde?a#`A5xEjz6Jj}A`S4o$Ag0Gg2
zhgxAX;m&_hDN_D;#e{qg3IE`&ba7F2&(yA~4n1Oo{+rx^Pgu~ho!z6^yfaZuv664m
zaHL%*0~Rz+jX>FxVEFU}3k4nIOp5!}8J=0f^^BBNeCd{$oq~RidhiAL{&~`F(stOk
zv-e9E*Rtx}TrZ?fn&MZne2vyie#8b&pd31`JX%1O7+Mh<vG7?VnJHDxvJ1trzf;U|
zrQ~%m%g5s}(&vO@s%)l{q|9!KMo2bmkwkR`FJ6&Z2O_O*J%!h-SSzFAoZ*&Om3-)e
zto5s6(enlQ!0M-)!p$I9ptD|~V}dp{eT%dvckCzbt#x$VUkAZ1&Cm&Zo&7e>*yzOE
z2}=IW5X_u!@O`^g@_Rum`+scBBy4c5Id6HD#?n#gu6sBTdU!InSuea_fA=OjVBm&M
zQ%D3CUrCw~T0ixHeyt-mvAdDYNR}(Hbau5XQQ}BpMRa^n2ca4hK)c|iKDWy(8ilYC
z5!JnIFTDy4?{7LC>xE}*?c~Vl5jRZWG)ZmXTIaP^B~~MLGbFa2+Me<hU3vHJ9lUjd
zUp1xsTUzc?<)Pu>21q~m1q;;^+GtIKz6dY$tw&T;O>ii66&>-<)0p{XEU*2C8?Q)b
z2BhA?z=f&8uMQ>rdzJPOk}xAp7qU7YpT-gPcl0a-BBD-p!MHogmhW?NzIJj&kWR8i
z?(b!5Is%`yn)D>BmqYq#v5|TRtZ&A09@V`Gbe)J0Yeyedh<keG)?S>6$CXNalBnj|
zb{3hu^`LN`$-l{{9P}7muh8T+8c0u-qEU}r>h9@h-8>v-I^JIzX6Wi3HCdis5X!R6
zUpW>EowB_XWgU`sJKIXDe@Fi)q?$(y|KUOL=UxHSLLQJcCD%#+rL9EP16;Qi%t1Z2
z`_NLgkGRBr9LpaTdrh1s9WYu{&KDcXrP%qxF=yvaF=lzWXrsy*E}Kgx2T?#S4a;Ae
zy?Tj%)G0l!?w7z*x2A}a!{J=((w}KHvTUk9Pkgv`Ep89xyJs{<$1TXSvKGBfMzpBl
z!$_*U^UO>9(u>d+BlKF(Rr!K~3wPyIH!Ko|EWb57?N!{Z;wG}h!HfYVgV!gOgjUj0
zcwDF6;c>i6(Tk0y)7SCD7rj0EtD5df65wPweBH2pw>CqN<aLP2;3tpbkNx3f7@K63
zCgc05R&#80dPj-;S)%%Pm@Pgzlb{nc_p57#4&`XOp7kHaT?FHjbNWfu8`=Snj9Cnb
z;ocEDfA=K+c$f3*kBPPm!Q(?HC4*Vg=LGx&*UM=~1R~{O9QSNTGCf(SL_eZHW%CuP
zzIwemOkJ~j+7hT=Gr)8;mDcOLG*+b;R)mFIR}&)g7w0EEige;xhwHZSrN{%Q={23}
znqXrvM(vjQWn+T?k;@XE(MX9WmR#w}#{}-<CPs5rA%w0=kF=%2ZUmI%8Y;bl(OMp}
zC-w;cAgz(m!*f@QFA!UQSn6MnjFx{bt701-jh-q6?}yJHdjs|2Yhqhw8Io)bU311!
z^cR%)o^D(`xmG#_t?-zM7E|#=iw>F|yHZ*Rb{7vTjYwdG<-&WZCKkT!@3qw?bMMiA
zWDz?Zb&TJ=-S)Sj8ZmSFeKA$0DtJ8shrza;CFdq2ox}Z7EK^O;1p_|3JI6*dLpe}!
z&=8g$nAYEe7-}=rhtLuvDrbJE5q2VuzpL;OF&{csI0b=>6{;%9&yE8z(_>l!k<HlD
zx0sZ7INDLL&wA9DM9N=RFl{&!BnUl=Ijmq|o^PSb_tvi|fqy*RKB$G4ovF;fB{HRK
zB))wlIi*Rk`3Z?8aVFEPxRc2jfgnux*J?bu3JC*)=C$?IRr^pvv$!T1M`qMss<&Io
zLKC-n_yK)w-=%ct5&O(yim!f+%lFhl{jrzkdj27s(A`K(Ivzjh#aC<X%J*3CG3tQz
z((Q(v!2c6A(0f#3oxha|4&bZn#{LOZgO|)F!(jxEjT&c&)7xqMTe)gOAM|UXZf4Z7
zx+yn{Ml!aTnpxlPv*w9jLqiOu3@NkP@V3v~!kaTwcq-!Z{OF-+?;3YjnLbt*>$OFH
z59FkgM7J0O;dU!4;*hNZ58XjI8^h*eLfbEb`gWf7vlnv5CUXiK<5HKOckN71BC0N`
z*lwb_rIibu6S>9%)#8^11oft8aWew~fqM+iRG)oQ$uI0TND>kfz<K1p1SWiV^VpSc
z)%HcP5Lw-pt4d|~z5KCwPqN04ENUAfk+wEa<tE<WAlMaQQD6yS?a9{naF|~(=v|hW
zb=bTK>nIGuPnIQ%6z(&mXr+h8jPIUZmYcj~iM2eNlQ55eb$EvuLYlEznQFHYutFf9
zpb2r&VZi^+XqZkZlG`}jbmfWl_5I=zbL?@BJoA)Wr-AC~U-UHTOt%^iyH#h``_xOm
zf~o<ky29(5BsKvU&b6Ddo4w7O+qM~1adfJ9dAB;Saui^Zdj1ro?I-aXvsPfQZnW&8
z@_CkxVyemNJEN!%4sE2lmUZS_PTQEfB&gQCFOp+gZ!r>i5(cQTeKG0_$QjlO@5Wb~
zUs;It`BLaAQZ(13o9%MFY|pfqov`xRI3*v6+we?x9hG@rFiviu7pvoldm4@7DMtJE
zSLNii8f9wN*9KlL7oYoQEBSPBnF@-s(eX)iiH=<AzJYpuB$U9hX@Y@_OueY?Qq6H~
z!9b*3%5dKM^&qSGQAn^@H<sTCC@oin>3=JEhi1bMh!)_i5@zNhFMES85vyi5XyJIe
z_n_DLviJ7g<&QFoVs>aPjCFDW@`tNSS&Fl4_Ml~)yyO~`;cR!YKWY4GjGMSiWHpV-
zcuNkSC5Q4ku0$aM4KcnlplyEpE`)xl97&->9QvyUPg({sxs1bDRQk$Ur3ZbnF;Q_g
zp7yS$JNrELj_aB@p5r>_5H$Y13ttW9z<`1_-7TGbox7gKF!omEzXS+3;4=_Wn_VbD
ziq86MkGMSRd-Zzc*yfY`ZkN$44s4k!+MXDv?P-3?Et|f5H!;oGReL6?FfMg^-I3O*
zp2d+z{hXHj<b$7LId88zPdz4e{_^EyHkvV|-l<<khMmPhXQx3VGXXEQSZO=uqdq1_
zTPONJo*r|#WSn>WWs96Oo*%J$D%E+`1R5;Rw%Hw1#%DL4-`;GFh->a??ygTuP9}el
z>g?N_(%k9b?-pLIF{~`A8p(3hnyNEbwQ*Q!6eYSceN;DIa3v9opVkSwbcRd_$NAK3
z1RnIJM^d=cK7b5p-@AhkG^$XvJ7%k;=j$e+3pOpBpM{wP*W&Rp(}4-wV=EXg8Z%Bw
zpv*8<-*mQM7fSwFItXe>v*=|z{q~K^u*vwrE0=9`W(%R4#6>l4ylpx1#vw`7-G{qw
zk-KiEZCi67{Kw;3LKs};E~-B#1z#+;oVJB7W=eDrI$1CbQV12uzJ0Uu<jIo?JE5i(
z0ys*es%y!Iva+(h^ElbMBY%0hSO_0o_d@zviW&~p$hcPo5}HA;_T#qd{k@9fFV{rk
zaaSi%T$LsXJ??KO>RtiR|CZy#Y5jHZup}FFwy$sh4OZo>^*{??g*L99?i_qCYa?>&
z&DAH~-h9~O&`dFLoEl^=&4!ut1zw!k^|EUg`s`TE|B4@9N<OmRT})-wwEV<XtUPzj
zISh{#*ibNUn&=xO{)-bs691`>cl}%b1(Y2XHRv5<%tYz8tb;e>A1KMVG>;+9<DFAy
ztMtnIE$9B{ep3!}i}=EXVi~$&bx^i6zooG=9ap6t{+;hF*+IkDSl8PH2OV_<B2cVU
zeyxJN0>x^}l*xL?1U}8z{9#+$-D#2b;3k^%B>hxeTX!uIAJq#FxMy--<2VCY97JwX
zX(U{J?8nndiHou=&{eq`ysT2ln5@Ci%p`TmW*z$WX{>h1gFERQmI9g+)vsurEA5>p
z_T~3MWvDx|t>^gqktt!Jdj3Z|=?n?xwVPMF2A9T223xc0BxBh$J9S4Y<Gsk|m{(_(
z#p1>)dB6*$r(50x;&=I@-W7s|wX9r)-{WHV(f|BtqkX%j0dV~GAJqkmRh51;(xVOw
zd1#K5rsh2A0~ooqyG_vCf-CEk6YS7+=NdGQ9mCxwvz!e54@wnp5|oSaI70dMwK&HN
z^QRaZPV3H?mxrm^A3{D_t@{|XaiGrpB6`RR`UKh?K_pCPE3^~T%iKzqB59sW#O}?0
za^J02#%If6AS-T^2eh6gX5=f(_T>j*Pp@P7zM$z3oandnoEHEsP-C#`68$=`@a0_j
z;C_Li^>s?PzA%B4-{~aFu-<QGx44--(UgEq>~g=j<#CKz!*h8&2{SV@;bBndy(U?_
zc+Gbh`hNH5%-!Spu`u!Jxbv~A(A~=^2joyIveZTSzIs9m!kK?%7fha+KFuOyEr#l4
zIvM1<QxjV!<K|OjJ85g^W)qk?U6@+>j3s(c2TTV1_9Q`jx7G0*zDj|qbGd7u&jXdF
z-w|vnlVm1TzxELQ0C#&&Nx#i~hYJ~MGVGdCoa6(=mAsH_p)vDhmb~&BHJxo}BM=@e
z6JcJr)(TC?cr=2Iu|2s%b}kQi(FgZwBQ)JoO!)}Md+#Szzdbs8-<`SuAhn_*K^!hA
z^)a20y{8%-m4WoGmLpis<#I|#dRKXP)Z}IIJkKO$yeO$vZ}<$0tP>aE3um-oBA={3
z@&m)dp+_krq&}wG*3HVazD>2#Kf7E00M`@edHiR#98IZ>@vvP2a>)aWI;6iyiJvfd
znyGVj8F@)XmqU5Kpk)qVlvPjJ6PuHWNLF8k^u~8|GeWsqPp7!sxFPSuIF!y0yCbO5
zT<Gs=xOSlB3}^+>r*Jz>eQ_3w6UZWw9)$&=#`e*hKXtDpq3~@q_l0>>_7byZ1p@uK
z;{=hbk;e!5zS?zbrN=GYCtPs9Rw@%vQoF{(DLCIrq1GqySzcr@zk_bkt)s2}x<oOX
z_Cd>J+$dxAJRYxogO~j`M_$mT53H&IJx=3Mn@}DMJ3d#7q8_KUxa&uaARsAAD^2#2
zN8cJ%pWGTAnjZB7{7#mNh0%5^{KPYWfIT0~6R}QXztwxwg~r~~lk)KHo#a8ZhY1O+
zOW$r^(``>xF){~uCnoE2)$Y~xguJCVfOrsYzDj?TO-P(a)A7qnz|4t+A6IkCIH^0l
zdHdY{z(CWD#x;y<L``ATdB9w#TnUZaHXV)g3=~djZvq1d>#j~Qc2(EQogTjY+o!f6
z#_XxeH!>4%q+p790P4;y0X?ywC5XI00VA^|Jc<|97xakXPl9@#QZ+B%<IPbCd4%3w
z_cDb%na4rHG?uIj_LtxCyf^sfiA9fS#v-@gmy$l!*<vns%6cVumE@vaQiP~AUc2*z
zVUy8|2kWkZEDjxA6mluo+JzH0zsWES4QD+MHJ1(>t2CTt$y_*!x*HwHkd7z5mGz#^
zqg`XS%CvzJ=lj98u~?^HYa@9{9{Nf=CSTJ_7`W-^UP<ya6-zloF2x0Jcn-jKWh4#!
zOXHB2FJPs=lqMWkkf>3l0%2~L+MCUh;Xl_rWfk&xJ%~I7%1*|7({@~D#nvjKtz1OZ
zJ>IJ=b$n{O#&(0qi%uGl9L$L`b89rw0l^t29VP1mlwvb$IX^8p(&Jj^9j|hsoC_`D
zj>)k!IOsRI)w#zbO;i2JNKxL-g-+<zB&fXijRbs;|6)b{Zz3lGYIp?cfH(dh5BQZV
zd?e4SGcy6pk-YkAQWJk>C>Bf6M5ZW0;msuD8(#W87OJSNsZ@_Tl!!i~It)Y{wgA5$
zlw%9tNaRULjP|3;^DNIB#|N9WmEK(LEN}=bpY$$YrL52<?i%Kf7k99{_m0X|>17L@
z+Kqu8rkpFA(n^?UYL1Q;Dz%_H@m~gAt$v2aR~DNdHgP}{HxGNDZmO<FtRwv?$63E6
zqkee~@{dyvraxFa^0FkAx}iAFam6AgOZ~<(Cp>N6%rHzfT<lF6&gR?o3L!d|6nd#*
zG-w8Ve;L6yFs21pt;45f&%GvFyozyOV1B4@I<F(RsRzZd7Qz_+%uD{Tr|v-=k8iA&
zAnlmX*Ep%tjRHl;IhIlB%KYI0W>KxJW<lxYs#f#dSfz22^f<xJ*;RkXZw;3{%5O+G
zbJR-*{Xv)V6)+z6qM&-W%sUjyL-napd6x^$wp4D-p;Z4vNnbHW71<q^l5?)cjnB@T
z+O@-$*<95W#L1JsdHDqxBdqbH$roCP3JBH%HxL>z%#=wBda+yD{STE&v~Z!Ntyc-<
zMXYj|A}svY*VZ~ep^5*N-;mS@$bSD!NGAlhBZ8-+c<>qvIOVzDsB!r@C7#`SgT})S
zFl2`p!wQjZUp{$MCB~3ywBT@sf+P-m$7@?_747IdpTk@clWi-vHX?RSh;9mhBni56
z`^{=g-NJD_&Q`j;dJOwoQ07ugj$J{q4>j9<hRA(-5;Rv?ha7yG7WJ3fBRIW0*AOEk
zdBb*tf+RADy;B))DIJeB`WrpvBE$3*=vFqk(@bVlPg2S&h`wH9dU8|{m71D*|D2`b
zM};Zp7#`6_<fEW16e?;d-gf50Q!98^+-OGdMYUP@(6%a}yXdgfLyC1T`tT1ptMZ?X
zS&JW4CK`0^q_oMFe6|jIh-{TRf`(&yEw71hlea!ZTD6hfr+%Uz`P8lY%wW;C1ZhOe
zdy@g*=-6~9$5P348h@RzVl93_pJM<x80|6WqAR}muxdV865(<@Grq!EXs&Bk*E1P>
zI%>p|l4h&EfM(c?v4Fo;b9{lh!2^A(Mr<*Yaqi7kG{HhzEBTOtXdHi?7sK`;;onJ;
zEbc#HnXMi>dCNTX8Z-(x<CR`b9ch<HPfr)vx4$QrxR4eRK~${$9-opT%W8%C2`z2#
zn$z%bO3DKU<!=uusi?$38RK;PV%LN?n=8AtwJ>Pa0Tzvl=RyM{B(Cx!zclyN@2+MZ
z&FZ#)f(4;|#R)0v?)b+)4*(R*uL()!WAZ0Zb0s(mIDus~nF_nw=pq$Tip}vCk$Vzy
zvO#u{V||@m()ngznR`?<NoomD$<42J(p}5&4dOe>39=*<bXPUMzC*`c6g0Vo0!2Pg
z)NAQBp09Ed4$fw)uTRm$Q0ejfvUmwl0p|V(VWl?d4x!n)5Dk{O<g_VQf4kW+9!QJs
z<^65KOQ}t}{+<nkK6k#YzoW}LAnYI5i<`~TcXjuM*V!7k$fComR&R$lT+T$M$O|zs
zn?!#`J5FWdEwn-!URqh%Y4m?uRaRd9P_>bNS;)K*i5>xgVe1y__u`YSi}pvS)qCiK
z%c49TWZ|UUNdoM!MdJAdI9M!L7Up>31k`Q@|4#iNrv|64paVJwtRN(@<T1psk_oom
zL+4{Yf*O3*FPCYotL&Lc6A0V@>yPsw6ai`;@cviG<}X{IfEoge1t}RBgwUj;uaj?-
zIPR9*fgr7QBQip%+ySvKQ-cdURX1kR>t7R{e;)vU{G`D0-OldbXmZ`jx&~phRD*zx
zZu@&N=IzFXE5^kq9`6K{VdZ;;q2%97A^zA_sIxrupLO;f^K{Nt@y2NdGSviuh;2+|
z1o#K*dGt{=z~|1=vHcOA{;kOLj~^mLG#=>;9~zP+^Vf+x^Jp$95i=g`U4d(UuW%UC
z)<%)p<7oI8m2N_rRd|~UB{~bl5PG?g+W+;-=|iY*^vA+MPSFK9qUFrMxF~00{vaw^
z_aQ3_OAc!}WpqBEwe{IVW&J16^XSJl-aW|_&Qg5|5R@-po*Voc3;cS)97*_dElsb-
z`}+raCp*s$TAE{cGq^a(_f?pghJtV0i#-tgz^IE1It!E5U_AddsQQy)_>aLM_Kc2x
zW33zpHg>8fPo9=h;fTvvmPW@$zsQx6uzPmPa3$cw61n&*aYlGKS&AP%uo(j!sTP3R
zjxE`H>(|@+^#bYCsD=VQ^e5WnQb6w^a1UFo=-GJd?@!zRo`Z~FT43EJPWj)D0fsG|
z8a9byb{>!aUUvUy4)OnwABWpn>782*aFbt=RZ+o@TYqr>zAvlfb&<L(cDA)Imo})|
z0jaB>@%`aFAOfDKF|vNCJMe$d50lt=j(qjg@-_tPP3N*F>dngAxf_A0oa$N6B_Gtz
zPcvuYu`Rf=Xz82<jxsH-CJYryP<E}4P4KKHtOiFf7@$W)AssT`gl{@McgC~Y-vj>h
z_Xm?jXKacqX6%V!47_*8^>3C8*PRcwt#-a=2u`&v&RZ{PM(On^bbrQ}#t&N{?1(=-
zuirmc-R<$)ZFs`U#6&nH{d<S7e-9*Z^#(J+oY2D8lA-a@;^UX7Ipe5z05OGz`o&Wb
zw`=-@Y9I+ZHTvkndF2da!~wZzw|rd6>YWcT&v?LwJ4n2~+%$K*%-!2RQ=IGKtqzHN
z974dQk?+2|GlRH)U|znw6cu{Fzf|6=z4%zU&LvCD{-}-bXkqJ8Cz$`}U|lfDZsV&4
z;7U9)B!GIsI=8xynMZ@1>MK-Q=>jA*ah43tkJ2PBRBV#|4^u;$4=ly#(t3AHAN_O=
zXtFkwN&gU$f``wM%2*2yjY1(~8ACy2O#XN$vFLr@Hi_5-&%50t6DUt{S-v}P)||kV
zY5G+~CO6pY#RucB%f6xY%%I4gy54W^(OvdU+KnpjYkomgxB3|5=!yqQCFZy9UuhGK
z1rxS(_c!;kVeHP;3YF-Gc8p!t_MxHRIkYl{c02&ZqNdZ03x9Yj8}ZQ1=W}H0^)|s=
z;hBP0bykz}cb8+>ANPRL>89=`hYa=qVYOhxymp@{r65-#%P7I&D0p}t<$2yEE3O^w
zeUZ_E!y9}9VKe?9XRpw8Dd53UwX${Ad#opeMcaH#BMvpq#z!jFoMD#z!Ir!U^v|{R
zJ`9M$9NPE#uN7gK6`tdE;r>7)d)I>Bq1|>dvOXl$JhfV5hD&hj%8-`E>Wj83IUQ>q
z@to^aMWh4+;f_P-$;X$hObRZG4!yZ$B{^bdwbsF~ZB*5)W+*p6SrOnY%<gF>xG{BA
z6WO-69ZmM1w__6%i*f5ZMrnz7u^H<dAeyu?{;$0g9HvhYGM%v>wgs$xyM<Fi*v{uv
zU5WP{00H_M-2DZ_(ESfgz;$yA3y+nFoRg}HySRGp0-)dD6*%6$0V;28!Hj+^_3Opd
zsp2?fBDcw2Nz1v1?45(piM|cDHys=lmoJ5TIZfolSFw_4`(lY%o{1PB69#<*1TP%i
z1GCuzah93ou58U{&V&11`3!uIDnX|GP`7JndEcFVdfjQ-Z9aEZqiUDStod;Iv%Ugm
zhnF0whd>Ob7}NDbdOBN-rabHiT_XRzN&fj5mkvNUY;=Bc1nZW8wtLV!kvi{xTNLfK
zlLAB_1COSnag@@~kArBW7j-ZisP4OZ=I#x!Qc#G9hT&5kZH~sg^+Q9&$JzT15bUdK
zm6I@9^`;jeT^EwDbcVdMg*yhB3}%71K~J2es7F_H7(e^wbW;k~8Um-00Dtvo{l?Hz
zmp0?ti>ghpsssKU40cV{*o6U-DCC{<u&sJ28tCKD3r$xhah5gm$uiq1J5KHIoP&1R
z#J}=E|NKNhLkR&Cu**EV(YE=$caB#OXCc&}|3!vtkCkJX#j2rswuJN2^72G}!&KtT
zH>Nl=>?Af*tUKv7u7^;^DOG~}@oNeONtfYM;VOADrdg$(4?WixT^_n&v#RSQN3+pJ
zD>dkTIz^MEy4ln0D^!3C5N0(a|DY$rj~M459u0N}r?=OdQ`(N*Z(I@lqXDlTwO0Rz
z=10nvH+rywZ9-Jg;Zr6_Z|ZVVv<4Jb4UP_pU}QUrHVU{`w{Hgl7iM7_<vV~897)S<
zeekYouqko7y_;S;WVBAtdfk0XW~}JMcIQ$AU^nuh6>xI^Qq0TKx%Wd&rUdzudq?BS
z)*;qy{FQx~Z4%4;)x0Fe`_oY;64mr6g#A)589Qda0O|FeAX~}9zYFw#9@XhwFvHHH
z+6%Df4K`L8MB|#%N@&kvu=<Ei`U13eucIS0s`|0}(&SR-qHi5hIl8xsneP0pj?T5s
z;*vz*`L3o$E5*aB0a|7N;^;GEIyx#=FSb@bLEJmHg96_y*;aSK&78!knl-%G7|VSq
z%8&?~zJs@b*QSlx;cr_R36RLZH`u2=nFwU32cegPl+%h$OhWmpW79L*amGHS)@MP-
zRQXJA*97nepqg;m7h65%%fm*RJ+R}<qBk%8e8MK-xQ$(K%5VxCCo1ZsPFLTf7=+`_
z9X3jxKW}7vKG?hVqXjku8csXfM!?!Jyo$15N`MRj(_c^^YGbWgH;Z-`%ZpU~_uHv|
zm`9UAnuqu4UIO9!FR(ggO;jC*!!_sUH!GC~y+~%tR5t6Vy!tUjoazM-?`AghG2z3W
z=3dnjC16xdo^oum>fTx3+>T-uzu&bgIORa_yp(b8$bPvjs4o`jXw!GJ{)1fHlK|)t
z5uGZ_?afOT*l>M&6dsvOF4C5@ha1TF5AhcX5}a#)s;nzO4xfB>E|joQd#V#}2s(?t
zv^1P+S!t5Y@Uwh*^GU@l_2lt8>;4iunnRQKzDE?GY-_hYO9vu=2J6wQo}xPOqJ}Ls
zUCt=8*>f>=vzirIpe!yg&^X9$INNz{)z(Hv&Rp)h`7||)SLdSR?Pk?BD1eYoRquws
z1A%UXUjF7;!A+pt=NmU4RV_Y&o=}NPN><sCxzVw#2m(lJgA8-GmuvKSD9vxZmIO$e
z;2R8xHTwhk@WEto@hkX*VzQ3DDSu46to<sR2~lP(If=#WjpS8v^EIp>TSYJQxRF-`
zMzR2|B@2ZIb6<fiCoIl@8c&p1mRA?`dm_hA85tSvj~)^)?YfWkc-e;m=^2hW!3F3|
z6%mQ;`srP$?M~Zx_xYWqhR>-s8;-N4)ge$d``ToeZ>4a4_Pg{?H#f%fX~Y>t^HW&g
zhO?MVa?BHg9ytP|cLy}PWrOGQb;xpp{wd;E1};#ecg1A*j+bo}Kq@6Ti~2F1)wxLu
z2{q+Ke{xZ#goojAV4w<%#OF+ko;U4M%*!4-7ZSxl85&S}=ohO;qu_Xxip^}2wm-yK
z9<V)WjlQNgZ7##AN0J0?37eJ|aCb-?4qK*{rX`A^{D@o?SS(uBi3>@Y!&^B!TlqAM
z!h_GGc%8xw1={nYW}_%XpLV8HtC`r1E-M~R--VvUf5<w!ru^WEEvc7{MVq;t*CT|i
z%W;&A4u_Ce?4e)!C_M9g*8QhF<=>@S>6eH)T^yR%<pxNDD5rdgq525t!W3GFuMG`c
zpB874B1?g+xa)-3Ee14mk#D*_em*$wH%6BPcCm+$T1tYn#J=^aW;_dvw{v8@b=hfY
zPmVUmBA6u!40q~XO{Q}^8<?&GJU-vPPagrTU+D8Vzv8Nh>FAh#q!?Bwl!y;UxSHbX
zB#;DJl#IW<ciav-Bt}L=m=*7vRU6KZOc%Y*t}^uSj*%Dgdd3p|Ebe@5hNfeR?`u0C
z*^3lM!Csv(wbtdq3(z`w(C}A6<iBHJ4}oL^9?<KJ1^2TYJ~*hTKNaQ~!r}OcVz%au
zn&<&Vhp5&URAvXAqsitsBaoAB!A9rKGcsSiVq`Jk#_V$Fv}6;pE*jo;FD{vk*!4Wi
z<$Pk?s(<@ENF05oxS<buHJG`3npFUtfH0zb5C4<Ja888PTzu(J)`6Q2-Iex31|Y>7
z85zK%sxq50LlRfIYITHkcSnE5Sf|r;XC}K82;AoWfCaiWejfrBR)h#PxtZ8xK9WWy
zEL`YUIlte8VK`YYMNP=b8r3MwVuq=1RbebZ*||(g*h<CuI8f|dWb(MMi?}O$xCy5D
zOH^moBFQ!}$W^Vbs^#CAV2n(;hx=VCWp$p9cNtX4BX{*LNl*9=KP0oA3a<mH%-BLr
zb1F-ES^{M8BR5gxxTcdVKI?SSUdvrEtn&_4W|NAEZ22$N+#CL0%Y&uRD4LKgu4jmW
zYxC~7f>d;gj!bI8E=*jo6UA2`Z!btiTZZxeU&DugzbR=^c$nFn6RNk{kQ%T0H>c4>
z4JV=YSMfMU@slU;WVeMgZ&Cy3e)%ccaH}JkWE}eTB(8)E3r<Lx(?w(0GH9-fcqelp
zXmhC6%kX;}Aif&VA6PEyt%|rp7VH`TfnBe!!DoM|oS{p-T5WYZ)_`^*A~kkp#(qkv
z*)3{>HFn3pg3$I<P`|_<DHRR>EB_n&l*NSX6zsBxcVK%tJKWU?2P=h0VBoEf7R;Hy
zc82{Axa#R}>ksETC*WKcF@NC(Hg!u*FQnIN<6oce#0t|PTI~mPw6lzZWC-U@dnbt6
zoOf~Wl~lBAZe9cX(?|~rK^@LqBr~4lfo<9MzPtO6&d%z|$;f!M*7RQyNCd5<_F{6|
zZAEAGC<cxxXGw0Wz{)Hp<42YHkhh$=YhC-@1W|e*5)cdA!^l;gfVHHnG)cb&vxV{(
zA>o`r`569&EIyMwR(tJVX!PH`8bJh>yFN{nQn^ha?;`&`Yw{R1|FCzRdNqxiHjf0w
zSu|*mK`PJ~&4ehx>B@9Wt#)th0s?9e9UH9fFRq+yT_sOO*_oF%%?&zl>k=R4hS_H8
z1?j^HtwlaeU^PY<A=aVg2Z}7WwE9mjrsH<R3Fl|*fI~sHtl2J0etu69&#HTGzkQ0h
zGj7;DT=#-#VRAVxp`3S>vCKspyb@0(#K@?q5G~Y4@uUi3=YNL+{*1@xN#<{5YKdbY
zCmPGQNeR3&tL8lwj9bRNFg#2kw-Lv9DAdzCaR8HLVY6(m`MQE6#%kHNIshr?rK@{p
zZ9xYI+5G2b+H2i}tvSZ0G@nC3e<RD`6_4V=sj3SMtT8ujnTUxZg~CL^Dfg4l3`656
ziAwc$DbtQ~@0r|MlLV(Oh(~UOUUv{tfC@DwcZi7v&8|DgD@o3E4si(zvo44~C^Y%M
z{Dk%qCGr+!*`VQ1C+B~!jh~VBAOEEOmkF>#SyFnMr*iqK^YbwQf2E+%3|(Uq5s|JJ
zjP5PijthWPw>qca<FlGdwTez`eEr#KLd2jaOEHA)CSQS`n2mTC*w<ebHJ^XGfeInA
z+&6lFF!Efr6l9R)`SimJ3i{W4iXUArYo{H1Du?u-S4%4dMbXa0PO?A^M<s-uGCu}U
zEfyqNH!^|+8bHrBygXr9%%zc}zcTPCj)tDIV=4}2Q9(ikpD0U9N7dH*QU*XXVUO^B
zxIO>fk#xh;m?}K_4Q-(*?P^Bx54+1i1Sp|ltZ%nQU{H`Nz4mo0h5ply;5365tD)O^
zTV(clJt2S^zCK$SMQEu+Tzgam?LIZ=5Q}OL5xxMKM~^3Oz4MeZ0qkfBSR^X9Yb?I~
zfG$RceQt9G!+tU4gpV<)>h0i2X%C>`h1BxWx;ib#<55)%26UXn069oEnN$N1OR*o(
zLl^iz0Tq|t5cg|9U73pkN(#N>=AFEt<ps*rIIoMtaImFs59?(P7P(HbeT?e=(OJin
z17(fD#wZtfxT%r0iGITWCucU&yJCz?(jRaIfnDPk{H5?$)^9b131k(Ao$J`uGK-uF
zHBpm|^#c013FI52`MP$>y0l)FgZJ3GeM6F#Gn~yMp}be&Qh-{tQwXh90n8o}MNN0V
zm(ZQ(jvKXckWW6ocQ#aCH56AG>q~&ud9!I_tPtjAH*1%{24iS7T(4YYN*Ea`1mO5p
zoLH|iMBXm05d=W&GCCQ7Pb1_zvtQ?42azd~$7Q*j*NF|jSmyY}mtltP`41!o|MCL(
zl(E{y(0omAKVy}GN1GXfbcaV0owp`#+Ty^q%&?2eoF5{*H<+#2-6_!}x*(zrCIdEU
zdh`7HzTkNe(LW8(GvFM)vEsamAmiPh=S$ZqCUeWY?%s!poZLBw&8z*Xf?8Q-ZVAA^
zc$2!)#BKIeQ?>H#V+OvH5*SC299i7-y(_Wnce~ov49Yi$rWlE?l$X1M9#keJKKjKL
z)aqQjlxgm#iNlY~Q~HjkcHNJ6Hrp+)LUGhJzBD&nk0I|bF*77eI8HkrRKj##(8nwL
zgn<4X4bX=6c&KT6c0&;46B^{>V1+mt*+u(|=)AvqEY_X&flq-F#dYF4&V~hRw#LnV
zM2_vfN`rtw4&Nt)2ulUc;jSo^L4&L=4E2roH6KP?LQWiUe`THj>3Hd;2FBN~gHHyJ
zV)S^ZH`QsA94TE7_bB0R=OY(Y?l?ZtO^d7!nPXy9+2z%@SsSlTn<&(`J54~`TFot(
z0!WfP%lyJStj|jzweauJGxYU}7@x%-XH0oqWpnmIHzCM$<&o=?YE3^oeQBT9Qx+R6
zxMaM%5Jrnf$=5BDmuKU1nG&{!FD~TB8ePjjhKR(WT2ceOt0<U4jcEM3B~p)3p;bMK
z1+mi&JnLgwQI2a-Mw)mM);8g67Zr!*8$ZCpiCSjGW7A=k7r{_;K=#f<<Ew41tq713
z0{y5+az6D<yBs*!UO+{qqn}{lLva$IQe#kt&1MOeEofM%l~UdB>l)Gvhy?<-t}JTq
zabys`C@2C}B<z_I{eVOB7J40x(RdV_U;T_75NMRH#~AyY%njeU6Xcs#6z>s+2?QGM
zKEu6Y`QF0sEQmYvj!pbayr5L4APm9|25#C<bS)%*<&FP&la|*2fmo**V^Hq%m3uOq
z4#+ZVU3VGvg&&CggOB-t^2EFnP|A=$U9mFyyTkY||Az_lwTy5a#Q)FS_=ivb@5lcZ
zgZzKP@|$51FuvEsM*mYW3RtYtD%7&ZF2lvxGmQW2q`v_#E-t=8@tB~vZu=~6LlJV+
zCIME3gE-S47EbpBKA|Q*q2EK0mh<>WgYa&(Pv=_NrQ0`s@n1$nzD1OGfT!zr(Mc|m
zg*6VKGP@h~2kpPc<C;hb02^)X|L717kH}J!ne#QOmXm^0IWnY}H-HA7;PvZ5u&{H+
zs?&mgAaOaU@NI42)_g8HS$_-WMyGhVJ(X!@NhhVCsNIt6O`!GxOEKmf8>9Ho(S%3j
zi72Pg$)dddR`4&TM}Jz#2zLjx0<p<ni{xRw%#d)G#*~J8W^&5;2mXhR8XX<|0_07C
zI5FR68TJ15O~K;eKp>+~8abIhYP3g%2R15*{K!8qBFE&3O~zG~Q&STONnHPjUM!`E
zhmH2G7^yIm%uBSN4-OD(9zmQ-Y{dXGyw|@4fO3T4McUeBi4k7QWpc;H=VFUR;3|s#
z{5r`F2>o8Vd%pkeNCHUa8v;)Y4(&k%$3h@K6KG5XC1Mo%`J}?*2D+RWZkYesA%Fe!
zgaFP}N@Ff5TTTij_j#+Xt9d^#LF$AX2~Vge?&#M2X|VtMCSo)^PF85H=z`DTfw^d6
zXz0&xMgu3u(qv?rVm)1m@~@ZsPYJaIfccKDVX3lLF!rC@Nixk38)YyUiL3=R-kL+y
zga2f#E-xl9_ujxKjM_%%a{~@MRjYR<-cP7~`H?he((3q+CgBahBz!k3hYgs7%O(BG
zBwUUSj#&>7AY-<x7&U)8Q>Y)$cfHX55dKk)05)Kd!j)$~2Q0}0Nfwq<{q6(-XpuvI
z)8($9avxaDWg{zUKgUa$L|W|`i3oxk%Wr3h0Esq$2-BWUqH;b3?{~lIDHZJgJU8gm
zyPo))74GmOs{NrL`Rm&f5WyNa*%BvLm)Q~hwkk*>@-2k(=Fg$Rx{WD}#rTgV;W;pq
z@aXDiFq3eff18BEh5av+@HUu9c%UsmHq0a(``;$v!c@}F+)AD-ul=S8IJkWQ9R!D6
zKE|SZmT~}X`lhZe`PXOqbI9<=U#00`mCf7Slz%Lonj7ma?=CEGe)l?Vc$VD<w^(+c
zq4@ytM%@0dUf9&}V1eJ;@_qUjL!ZBW_fPo2Ea~n~{U<}p|2&NV;$2YOdN+gl@u$_p
zfBy3KpO#?;KxY%y`oEj5|Mm1aT40FeN(fqQ{`OXXox16lFm#>Yx$pPAi2i(6_g;c&
zjcO+p|BKP!pTF~`HyND<#&Ox8KjHT!h(Fy>`a2l^p~Vbk_$NmgRCanblvGj}(oB9|
z{sFGjBw&0HYi_;#zx8mT286=4c+sQeHo%1FqFwBv=0#sKHang^FdjYl!+hQrt8+bB
zPMtnx^XJ+5x9i;#6=ubSH-~#C6&Jcsp1|pBe#6He_3mRwb_QI=-svVlduXs-t}J!7
z_XM?9<Yc$uk94y2cCUbA_-&5ZwLk!iNI<sUiobd#ewd5<cjUuguA7bz8z513)tgU5
zHJnYp8W2=d<~-A6&B<(lr)JU8UQ>?ie)V4T1rXsK_3^up)8jczp6#b?(#J(f@r<cz
z-Y2NH+l{@8lt#9ghDQc^%*E9nJm|XH-|)nK)};{mPk*5O_=)p>%$x9U-(W<W@YYsR
zDr)-Jp-5S(HVRTW$m`_jWb#}ps&?d>*MMfDE-M>JhVRvm@<kOE*936|v+rcFe?bi%
z_{?j~BjJ$j-uJYWVKJlSz6$7h?Cfk(&-0Gzal?5ifFmT*G&onL3(Nuf)oCOo84a|z
zYAc>N;@`wCG?e7tNE*zs6=vB%60Kla{i5zp`0@6n((uwxXPp094S$+lH#}=@8{QDn
zaBBiO3fz;@{3#E7hfKI9h!i5SS_mJOR<@bSmb1kb^dH0B8{PKHfLCV#tm_6m$sPwF
zNH9yAzjd)*n2zc?ntsdqhA-0V-b7waSoytC;#u{0fjXsm!<8rJ9IJcLZ4SlktH~~4
zp+TPJev|_AbS(2VjhSwZbweL4g`DL_rpBy@L`2$Fa9PXCmtcSd@QsiWRkI0!2_JO(
zW-^J=ejc->x4`_Tw!R+|@0HBUu_y5~=IBgkID`G?#{Q>u_T$W!v-4E1w*yj$CiVki
z1%2NUz$Y)G`mi_8-<_22N#*0?Q)XiCiIrY9g(gk)^qpK;Idhh1Nbr;ASp@Kdx5<Cx
z#uIuk{>^X{N1spegeh$*VI;?}Q5Lt(d{xehU9DXTlly^&UL;#^_ez5oQ(ndsyf=52
z{^)+98(-+fJ{){M3~z$;o6O<rW{I0nFKA9`n!r?X-DRM`<i*dIdRaD7_?nw2TNe!h
z^NFr!N;^ZEO5P%reQN~YR#myjlI&PiD-17%nlt6fP1&NhBp@nC@7oe~`PFwpD(#b?
zV;B12^xx%j7WX>)=k)Yn4j|zR$mHv@wcIUI8GJa99{7;k8sdTa@DX>uFQ8@kn}NXO
z1$!)UZt#z^*T}%YfV;}wz1wg0`~$fz9o%A1_J462@hAgR96}X2cdc*?^=3r8hzqP*
z{I&IcZ<+89DVm65KwdCfAM@ZjLJDP&^}L;T0AvV3%d9aP5#u!Spu}y@-4!7-kA~|Y
zq7#zDo44WEssr{3(O~iU&1ev85RL;m%y9O0Q{i%O24xSZC9!VS@65>nVQg#m9qu3R
zIpE&Fr&|#042mHV)b01K1`M>3Mt6s7*xv-KxI9A6g8Ta^K{8LNrJwmUHf2CI+pcQQ
zqxsx5d^(K$Q9Tt^S3B}Y6FfXVS-q)aHD(JD;_#>h(nr&18-Q5jbNeOj=rJ00DUyuR
z!FP<01A0OQSpo%Ydrs;lYZ$fHADw#_Y|cW<o}{^+6~aKDp$4O<5-IlWAafYi=HV2S
z%Q{p!jh8e^0QLDJ7@f!S%Pm8;fSr^V!Np4@Qke-)E`yNdyzlyBLC!Aw+kVFd$9ySJ
zPQGSEq=J%(^7vk}u0nbs-b?MT<u)%a-vaazoGlq4q-GfjWCc2i0!kOhig$bm0N2SR
z>}H*l7ujL_*wn>l73=Y79HrybWJj>A^XLZua^9wS+_C9Y=g>0Ea>9ehFA7yy)&0~<
ztwLU&omwX_n@5-=wTIw$0%ei~ui3*j>v=dUK#@_Z7u57iwarbooa|z<8Mq}lWr+#f
zY0|e4Sxz5iTfaz1sVV4J)PK+(@b~*?dMnD(@@mk^!ss$j0rKTNeJgnfQe$;PTA!4-
z=Bd+e_Bh$x*0`Q5c@hvyd~;#Gj61X-m`$y0ozj=O;j3Q?WXxMF_a@B%nA_87V$K?O
z>Tj=rs{VL!L#A3H|2|}?`DBLlrWE&F=gI)6Ko#t!WJ23_tyf@1Fg9BQW6*H>W+3(z
zKr?@j!VYv;l&aNcK_w3$KjYfGPw|7fOdHpvyN7*5$%G1QsOHGOBcXaCfOi=<Q-2#W
zXfoO>IiQR?_DTT48vbloP(eT7hZ#RMOwI}Kvpqq+m2HWZS;7izo+@6H%YKdfKClf~
zfe>&a7-TsB@I4r%j@I#Q@54yt_r9F(@18>?CaX<!&A44J_LGP|?8^9c0|zaRhmuO_
zSLgU*rMy~~nVC9x<_I5>jo;)MN;jK*vZVeo;M5;I^kYDp$Qi^V@y@&a!}GQVY<-`<
zk4xlV0W$cf2J#>7{AJDc#zn{YFEdU^nTHPu%eM?#@$#d<+_G$%Qo!`qkFlUWCPBU9
z4<pj?d14<gN?uhK0al}wXy8O@mf-ZHh>H7hg3yPLScCG}c65uxkEG;g%9is#lo;+(
zdS7)+m+2)HTsSw<sx^#&m%z_U_iD0orlX!XEacpFvs2$M7^k^6nU0}xKNeO_Bg)nW
z8zWdTOgLHrp-f*^<bK`fe<o;{RUY(o8CjyDPG4OOaW&y8#|1!tfrNy;he?B&7T6@=
zt>P5jTzB4~$QKCQzLSZ2u{Hw-VGU|4m^gq41bBEkQ-fo7q8v?(C+@^>CQsxORR_<t
znm|e;DWFeiq#UNcOl2C?*e;)$MagtH4!CgL(3#0qEG`0sDkiOXn2&|u;W}tw&x*n!
z3d3hT1kO;wX=SXRT%vDzJdb0-yktV_DugciM<=hptO^o;0l?FBWPzN1f12fjK<m7z
z5#3MAVPXsJom9|%%19aGj#Eh>7=N0?ES;4~=c&3%nxJz7;${Y58~q@pK?aG?*+wn)
z_!Jm&S@nuT?HUT))A+>th%M_*l!8`XEge-Wh#MK@l2cG590AJ_rlaz8*iv?wD*e&u
zTKXCuA7QWCxh=E!nj%>%Q0C+T;$aF$tuR;tZ(V$dgBRR)9CoLQi%IEUWggsSE0*q9
z+-`ID+t(p?-*X7N-#LkPQ^w7{x2$?c<P74?toQ(y7kroOFGG~&KvYqBy`GCl`3~p8
z^5Jk3@ds5O{t-Zvs@*R;fM4O9?QD`!T2nUu!u9o|t;FD4T;M@>te<webD>F2L-QJ)
z_f#MsulUEsEinfKd>6cA-|=~4*d=lXy?}>*RMChe)T=B<tt(m&<zlidv%cXw+U%UY
zl(T9(?f)!+?U5VS(#D6ja^Zjr)`XC6r*v{Ryost;R_?oG_Brv}?96ae$miUM;gDs-
z`P-vJ<`#mRFm0ap8V)iL!VqHOk<xmWImhuAtjq0)9#gPWpTumXT+x`nj6H8>!Z*_c
zg2VmgL*h5vY3}u`dt(@UXAVZAWeN6kO&JLkllT%Igmz+Dm|vEt(q33ul{MCG=_1G)
zD%To^;E82i&AFTi;c?94ic5vf2{EQcmBA7vx!0arp6D531;~eH+lP<bO9XDX)({sM
zhKowBu9eAtu~NUb9@?x3L?VrrFLhzP(Yo!lrTl~^fybqBWX@{LwA!+JBas8+Z=deR
zur9+kt#>;f^ot8=RINmY3pUe6)aCm9_IZ*H>c+-_$&Qm<WhUch6OGQ*n`h@I+(SI~
z3qnR~D@LxgzsAx{k<*m>j*hHW=9GxZ%DNDrzri<6VmBcL{R<-bkcmh%@p|5YJwVRv
ztY4nbb>c^gBbMyJXEQE<1fh=br$yp&%gJ1|qn=cZyIcAWKRNbBDKaic2$p!tJ^yr9
zo2jN<{?HgiA!Z%#*9FItriGj1o1^Kd*_$J7i2<#NQSRKfWiOPGWoq!&c#?O*?s4cq
z*p+SsjQT46K5rxi?JP-UOF1fG#8D4{+a_y(1Fdb^4a&R?Q)9kYHS!O36(KhkDtdjd
zc!rxVWZ2B6Nr5YYR&l8DMSt^k?9R-w{g$en8tuA~Cx2d}lT8${+xaWsWF~=B2d4%Y
zM`FT0O(9hGClMRd9nPc{<>3UXlZA6QF|^!Sv;v-DYjKLWO7$R6?-u68SIH@-rhV$V
zrS2`CR0<Im>DD?)n(!}18k2pfp9U6VBg{;s+h}&zW3$Og1={ha+%)q0oI)su{aKNI
zE{(rkoWeAir#4)Z5o@bt^Egj3@Tlo=8D9d*#2X|rLxC@dtEmgy*~-)vY3|>>K{3R%
zGUT{Au(4TQQK2{}jy4WZl-el5bqcY>U>=c#*{NyG5=_Q#5D2;VOT1uYm_#DLeRQF{
ziU0b?wt}e$mpu=6@!4_n4bVA!;U$ZxvhL;qj;Ezs7}J&N93nRnthhR$<3!E_zI~&p
zThg~pW{;h+Yq=g)pV)8cjv8q?SKYc}Hi&i~-(M?b&e+7=mU~6Pe3ziuH08aj<+-wz
ziNs$VV*c~od-wgJcYPfCORz(-J+XPM>k>UNuUjz3^prd?@X@qgl{>bI@hvidDwacG
z-uuY7a+qz87VM5z(D>|Dq9GN#rC9C`?-_%`92#<Nnx1Ue(g@3M#kNqf;m0Aa8-*#R
zQ}$wzw{v_>1&aM<`caA+?uw%7*101^^1~agG(Qf4cZkXG<jF*>7yN5sHY&%#x367!
zbuZ|zKA3|4RE0yo*8O6wxteD^lnHk~)~|7P-9=KFc<U<V{OcymZU1`?8LQ(58&{mp
zC5<s1ZHZAxnufq1p-Mh`>NoBAo@m0U54o-OsjUO@lO^f-+rpZ!SD#;i7<Z^a`EEy(
zO0E7j*SKy3q2XwmGC9O3NFWQ?g^~7EZ(LgYcJ-R@_E;JoqueR_&Ty=S3e^R!cG$CQ
zqGPDg8>mn=NnPRBfaFhlQDKHxVFDNohV$>}GX@(R?<#|s@$Gbl@=C8NdG;O7K@^8h
z^hwSB1x>cM({B}6f%6kIytWuhmZP4p@TnB!$wq+fRWMf~F3GnL@39DYc3@W{m;8i^
zTawnYE!a>w?MF|wpqboWT0OW3Ms19-keGS$Y#cXx{<hS--zc|Ru>z(Sl2DC$5ReJ$
zKgt}RZhRe}U3mg6=S(><+pJ~rZ&g7O=C#o{=5n5M=!2NjW_vuyFTD3%d;1Sy<Ckzq
zz!p26*WOdj?g+9%5!Q5#insQia4VC>EE@EDsVy_aY?~_P+jMgI>ffUtCSo=3$3Xdf
zD7gv+aKu(AF=O=4@0n*F7vM<tI4d!afnws2R*-UgTr83EJNm7b2a}?kUX`Wo!AF&V
z1g2&)n^5!_7<%Xz49xp<K%M;>F0Qj>n>EyHO%F${Cgj<BzESl22x3`HkT5=e=WpTP
zWRI5*AwWcur9IQFbu19SJEsfq3%0l;V4Fe5f3;r^^RTd<I<L|iB>r)@Lt-AN$p7p^
z{zXP0E_|T(*0}(W+!&&(Sh41$M$l_%pQ(|8rm@dQ(w?`Rcp__SQ`o1Ssy#w2E;`<)
zXq3O6_oh$7dgbA5o0hpwuydqzGF;e5gkf7eb1wV0+&KZg`vWiU0&YNLTpR<87&=km
z7?jL?S>aDlHM;N_B6-vBlSui;)PlWcNTLbbptb$Z*V$=}@?(YrAxPbDz8|IPWnJ@l
z(!>e`30U+8!iZ2F1XE<$LH3Se5#!U;y2OoBN3{C$K0;%t&VCk?U#BQ_oY(hbtj1l#
zBm@OLdG_YGv$Va=XNT8hAF9~*sO!EtMO5s8ne`(dZ|(3B87~6!Yl+QR2xz<`!Xof3
zsJQ9DRfl@@oXe^Z%qXe?DEfDw*4sbmbT_NjypQKea96`YF4f8H9naBpkCjCl5%3lh
zmoExgWl$rXgPziO;%IQfG6T=k+rsDRGqW0AREOHNd==H>^$?~Zzm0|9b-N;^O)ouF
zf7>mK^Zd=6i=c*AWTgWa+K6>Z2{a8iDGY_mL%+u1i4Pdz-v2wa#`DFa^nRocer8d#
znlnCSQv@a+{%ck;q09TV$9X~IoL%J&CAN=%Io)&7WXHTp8mRNxoyD7-<sWjs9%QiX
z!o`J@_u<ldAs@4cX;@4=8Ke;D<21=hc3r_NH_MF3y#`rdqUSv+K9@LUcE@fxyn`I)
zs@rj|WkLk9V6i7$bx%EQhqt)CKpTgyBYHkPT(-noc-`BMO><Y0;^wm^!tq_P+J!_3
z2nq)@3RqI95;?D-|6mfqa!Z(y>DrT5{#jwrhX|Xf*#y!D_zCO<`@r29W*ECQS*glZ
zF{cO%Q^~i_(%vB{PP<$PG4-QFIvnp^d~mI2vudLtd-KI-yps}e67PoXn%p6Sxj){l
z+ufZp_}K&g8_JrO>t;*s(SU}5q5{UwB1MKEvZUbM<F~huJ&^iYH>@bA_7dMMa-^c&
zOU>dQVV}-=#p6krN}*K1NGOG=n>F*?Z|KZ+_d`$E)}8BtxA@n}Y&tx2zq)z+4BB&8
z;9=?*gVdV-PE_|OyqBe;Rl$CI&E>I2o$F)Dbz9A8Cv_X6g^M%j20}MU4{!FU9M-1X
zzmx26Hd+la&6m7M8_Kyp$7j2OK8Kl;2_wm^wHwcW0jUVDCyuCqPpqLARlzcU<~A{e
zf*yEF&oXCr&ARRCu7Axa<l$RE2m3nZnslPfHyEuPxGgXMi;3qprde*!i609gr#+de
zvpMX#uYKpR6pJMTdLzxog}?ko+ylIe&Y3aov~l2`Laj^WtN{G5fU{nOr+xu0kbKmh
zWKf2D3RxLyG@+Ez4=QesOERw+GC%{&3IMxOhkDQ>>+o{S5xiV*^e-+Sp+yWp=kS;R
zjEZ!NE!2HLprmS#NB?9r-P;1(JEtL3_+Cm_qv3#WbJS~(z($Oz|D4Ew`XpxycPG$c
zKcYSugU6jDPLroKDdgLxx%;7pQ&@7*?6w#ipd(=pH0ayKDk~}}E!`ObExZbx^1UjH
zl{cwS<m5p<f$d;mD~_Q)sk^e}$kd1OaiAFJW1=i>lCpT{ELB!J8GQXLw)j5R1zX;3
zr#B`y&7BO`sn!BkUM-%0<#^#Za5BSf<$onz!U@a?#YFKEX!>}raPWXLXzkWUk}x!b
z|GMV=^5G*&@u>DYkR1S2?gvJo8`eZH<pA2`_#-G#)$p*{CA+XMK7bKmtpIt{v7>}b
z-S+IP%&|wI_L$$kOTK+CCVwo^FO+2nPv*yfc1wP8!X@fsBD}^JzB0&d`Tw|k%dn`n
z{sHuepa?3Ugo5-@Vn``zB$bd3X+h}$k#0vpN*GW|kd&?g=@>vl5TvC`=^+IMh8nng
zz~c#rbKd)Z?x)MAc{Y2mz1FXLlkIYOeV+gBZ4fv;dN3E{3^*&lGfI3Ah)nQ#?03d@
zP$Biyg{_(`@S^*&le5cMX-D2EmV3*~UQQ$9&0<kL!ausL{~F*uipSXCQliB@TyW^_
zt4<JA&yL26-kO%h<CB(Y#3SdXk(Q}eRyp08mXaH^hV%MLh^*>b&<!in@3P;wgDDRK
zes}e~lk)L#0JDFP>Hz^O!@zhE&xTJejbU$Z@0_BGGHs&#3vaX`Jlj1{_tDaUZC{6Q
zns&{;UsH$oV+2v}Ra)VU?_!b{QJq>Yob0BiMTr?7QPvdui)Fp{P=nk`ioi67twe(+
zQLUow_K;cyJGUv8w6b==n%&{VG$QZm*M1JoG+qyd9=Cft<VV+3+|;A@MvXal)Krb6
zA&X6ML^-<#1ulyo#wehEBS{a)6J?MlN#j)^SygBtA$c3{L}n`%W~O>896go9rqtg-
zEQ17QSh}P_J(qY8EXuAn?&vJ0j6HLZDqH&KVuDAi0Q^AH4=h|0pLfx|yNu(moY>Ue
zrW=e`*7oA;?sj_jzYd;jCGy^@itBX<b+a6HfR4S-lNx_}QE>a|I;$V5kn-b1?Y9!d
zb92v|@C{lq{adJY8>g3jIog`JtH4M7aPf<^=xS0qC+btkK=RYe{RHZbTa+#^3+_Gr
zy!k4UWO<O8$hYlJf@S@L&m+!q&W`F&DOI>=T*&D8O*(C@&*J^#tRWAR&~-9S`Mvx;
zWf2F%b&vIdeNV;@(CU0`V0~gbiKo9|Ui9M!n(lEMt(}6M$H3X1&&W|SgT$@rhTd=f
zqO4xGMxHPpyn142xHUP|)6y~;`DjYGX`}X_T{2RqT~yw)(|NqYoHUfPdVaFLpW090
z!v0jQ@>{(!ty~?S(T&htJ!L!``0sRgx*<cY+DR=9q}`9!w)3CLlsstk8G_IIz8j`V
z2{>Tf{vNW?sU#)kaA*cDDI8lB^GGvbW!qCX6hCO&8C<FrnaU`>_C4%un!2HzTD$M?
zO4ahK@r^Bkho^%0ENyIZky~!fB<Zg0Aj7wYyOXB9?e`@Zw?ctZ4myb>G*0BR^WD~P
zvg<(0Yo9EoQKlSiWX)GtpG&beBEbq>Jn3m!PXFu__ttjcW0{i2%{KAiy!o0g*<x+G
z5~#z$Bi0-q)CB6m{j4uM)q8I;=I{hrODN_{UWP<uYSR>0Op7`F0mz>a#l=Ua)XlcX
z_4*ukl5M>3{yxm-=P>1x$4_E6=}{){bALw_>S-ag{zmzp(ZPaWlXFxElmfb1F}H%~
z5&~5kts99m_rtUWSY_h~Al{dWQ|G&0%EmBwn>Yon_uXK<ko@{oP|pF?1=U8;a8EtO
zp$B`*tOpdHZgEi0?GG~<RyBz_HTnT5>0!aZmMj#LkZmivMrH#9($Z9sHRoYlX<jAa
zk@?191A)%go74j)R3TQT&^^Y-DxAz=DD6eQ;h@Ko)v|$NrcMDpNpfQST1ahcwlzgJ
zqrQcvTN{P5PT6DXom{ozhVNGS_Bxd6q77k=`jve0<&n^w9?=exe-Ie3iFrB5^83hs
z)3`mvo1xCDoCh7KS&`X5rfc!QsNA<Sm0M4d=Kj}i&FLPP%><DlzG%ls7<#P-T0VK=
zyg9nv*5VepHO15#scjISU$HW`lQy}yS-Y(nly}>C^ub>Ci2KssQ?H;AKT<nKi6kg4
zE{l-Jo1F#HZm`V^$%yOp_0pYyo)vbH^@Hk5n7u)Caa>@$VLdYD+J<DA;hb<r;{vmZ
z{=-&PyFNeNPV#~Ex!~V!m(1l1$cp$_K4pOUEESm@`roCAZIL3_v3?6BV<lH#=Wce0
z=#(n?-eR(AJw3kfX5Gh5Yn@j~c+TvLRVUw{{PmM8#@LAU`fHclVW5j^B>vQ@BB9T>
z3A;zkeN=afT@HHRZj_Agqp1vE<8hj47c;|9Xj=VnF|QT3nT+67JNWVz5<=wje!B+w
z)z>tE@s29NWcezt?qQzn1WCy%gn^5mw_D7J{W5Ss2^ai`{99qr%pQWvJ0<e4cF!F{
zcb}@-0O*!Na}-QkRp<piMPv3wlJbC8{ukB<dMfj7*=kosekAXtf`qC)m*14{>?5da
zOpNR)jYT`C4MR<_90$$(cn=Vi__ug<OKsx~B2x^}F12X8MNJm{!yJL?cggOoB=b(E
zZU-4A-kxGQk~!c@0c4jagO@y={N%?Rt!AFfqor@rry=AUSQFy%)IEjglEUZW9HjgN
zNOW7Y-vn^W5+@CpUbkl#$uZ!xt9cCxSg(omGP3lh1iy0(rF_#XJE95wt1LT)wS+i<
z$Wrpz^pK$~8YA?uj9+swJO4$UqKUV1XsAMs(O@j&y?ou0CmkhLb~TU3cltO4m5u4}
za9Peg85ZR|5{;RhUGo$>BzIaGa3gi|zTETW$*(rkOIsl8)YR=`sHuh~_K=;Iw+prJ
z+kIBsQ!c-1hujv0C8C{JwRAIQ!pgM4)jON}C1g_`^=lz%?HghWMjLrj@|`79<^9+A
ztolOjEnj2VxB$VmC*M!8cpspm1yX6z+rzxT?1g)VE|QcTrRae752ijMVH5`Xw`x|$
zQTK@4zMm#+-aN4HwevB81uf#9k(UTFNlG7SIHvM`!fB)JXn*`k!gZ^h`;8^rmdN1s
zYUp}*XQ?<JEZ(AOSab%=w&jH+n~@vfC{6oNKH*ILoVzpO)h|0sg5xxct+tSk*yy8*
zj8Zo^w`JXDdlOXVzc-jC7>w>t`8A_uC@Vq6s&>4qLd%<**`)cjs7Y-B4WK0z&H-qE
zMIH;FY62sX332762zQ+6-0F-A8-gRRXGhMPw%5FCTE3|}=2GIl7!TECC12kUzJEJq
zJjP0J78T|HV0|OclbxuuQHyD3i>5GJ&Mk14u#nBJ*#iq*!r2PIBN{@BuM?FP4Uzl2
zH*db8St}^9Y(F59*EV!=av#_ed*wXjmcgcB@3q^ybExI|*=xlm5g8-y<JtgaqwD61
z3_8s~cZZwq@-i?k0>e;Mc=8RTAxKR-59s!~$O7;Bb@w7Kftxkn*_U60TqzO0qe_#4
zKQ$yacb0kg(?rn^&H9t`fgd_A>pq6W^-NSFkjSxBqo|8*^&~`|bz+vTuD9S*>^_l(
z<tq_Lj17@-usjlEkgLNTKE7ExkkQ>8K3U;@HlxPf0M6>SxQcG(p~}v<1Tmxl^up#<
zUhg9ibiKow$suZtMiaRi?|;sh=)UtbMUU2NYbYC*+MPGG^3gtigHzp4Q?An9=A-oA
z^#3tdF5@H$%>_5v=2z}kN`^7qTPknZV!<IUY?8BdTclFH>#<a0jmWBO9xys=t8~(!
z<GO?MjPyD0NJ8Whu{b76;NCC+z4T^PsA!e6i6qf`8xi|ao_KkbKef98MAXUap6LEp
zc>K^2XOzAQ2pl+%GtzF&%cW+BtQGogxQthb#c%AuQae3Y+BO#8=rLMFm@DEOBG6yF
zvfxQmH38YZX}xC}<<q1P2<be`j{VFNf**XVBLvxgCRTIJ?%_y)XN0|S?#(c}4V7I#
zAHxhX7~xet&~6And1b7u?fWkK>3qUvpJX%#vSupuN&?E9=^0&`-b=RrhsmZ6w0|*>
zery|(Mp{0vH}R$yZB&EALuAg0h@z)s6VvTH7hfSagAYWTqG;WQR&S$}N9@Wf3((DZ
zWn&5VJm${!WONF0Qlm%BQ=7$Cx{9AP1xfZH5LIPE@o(OIvQs|XY3C$^T2qTGd^e}y
zXofeHzIf5|c<UYUJQE;1y<?R+C6DkW#0;Fn#dD1s5@2QYwK@9RPV|17)Jx8u#mIG!
zchEtcR9)A1)!y*JzC|DU0jO919nE0fWp?Mj01?Nq+F38Z`!BN$-Nq0j3!3-8Z0fMQ
zE7pi@7|JtPIiPh&+oZ-Qj(_m|6;jA^BKJGH^GxE=TDdWIR=t|3dOC#17^!!vU-zuq
z)jYcRA;DuQ%Fw%K4Z-bm@Le+I+oVEG;`qh^kD4!w?e=B}Vr<BM-*eu}PU$RlfscDV
zPiXw*x)!Ro)0cMk++pTETM8`RDUIZe6rMZkex`?XU7P?}p>MuUcz#S)Ds(lOoCbV!
zm7z(Ev0ihBR;rZc%(Nyu6pJ$h?-JKJQB(Rfq?t$=&)l$Zqt1~W{?dK-TDTTLrUqe!
z<N8qBT$<WM<j1QhO(7$M_Z&7$?Rnk7o+y_nDj*|$D#7O=y)+3(V*isDisKAR|I%R+
z@^gl$81=!RRPX%!m`~sWxBtO?p@++{jt71disp%XX_83{_2hfYdrwOXfd<?XFxWb#
zPVL<1!&1{M*%YH=`IMA_m+f`ekfxm3tED`nKYE(vX{J`nm<~LNs@53F)eOzg<6rMj
z;B<i%8^|KBK%#6}OJ>>gBgM4^LdJBk?AlTCE1%EKq=|skQ2$jQuhzC+ff}#7LQCDy
zI~Yb?)7`n9YhwSUC@~?_^W*CtnIV2Uk1>zVVw)k?<l%wZeS}73n%oYY(2dqq9UmF@
zluo9$J#&u!hyoOTSFt4Xl?8NMK_fIjzDiThGmvI((d&(&Wm=%To>Ze2T|8S!J!9TA
zxeDbHVwo=>*hs5AZ;zJSVf*L1xa$2&9xmB6;Rj!im-`#jciG`ot-Mzkn$0xl+t)EQ
zHAQ483)$R5udNjduEOahlHLPbc|>7;Bz1p-P~sGtNO)S(>Nzx0Oc$+EY{|GBT=yw<
zS}Kx#&z~^|rcyIy-r$IY=INFjkB6{}keKS+VIiXWV$JB&k8jJ?<j_xLgxI{YC}S?L
zCEhc>vD;dk@==Q;vebcPxtaS{7C*Y8lPfW!oJBE`z4v1mH53-pgv;_dQX0<!IXgDK
z-_8{w1~*%gZ)!PbmS*2(G|k}+)rt?6xN{)VUcBQw#T1uim`T;CYZv_bibBa_NlSl1
zL@m<0`EQW;li8AJ0fTk-0lDbuqt{!!t6RJ`p{2uOLjU2Z><Y^T)d(?JTwIJl*IBmU
z%H64Hd7gW`>C7!kgTmzML#vJuUncnxt5R<SJoI?Oznn{=79U>?*{0}z@mN>ffHZkm
z(CFRM(7T_a8^GWbQBDK#m)A37a|Q+b<x_@3^C%2+@Yu3L;6uB8!}baP5DOTED1NR{
zwuT3dln~u#BOKAF@TVQ@idF>7k5BVXibrzfj8u|i<O>lM5~D!@%7+15G6;U!_E9|r
z9gC!>P-|?kn8uN6kg_y$zoDe0lo1~OZNGTnB7v`0L=2&y1QW)nPtGMeIsn&K9Jd$-
zA8-`oirzfDeJf4umm_1gERQEEC)?AH00VBsO+qy`5PZ5^3g-ds^;t>!u5+itv+gl3
z9*Y@bi#tg@<TMw?GB06W3HYKGKmVRfuYldqa5@dIo-$oo=Uj@R*T6Y8t-2u6U@}Lb
zm=$xR0t-k_o^VPbpioM2no>0R{zUVi!c_u<`?0yX7wD|(1;+mNENFtjtK9@&ZVU&2
zMeD`nKl@20UB=N$o%ek&cFg*aSe(S2)};eyw=`Efry%_X5Vl<^uyEbH`8JB33UvNk
z$0k+O6Xw%HF19od#16^C^S)2TE#p=C&mKK^g$rUPN!<P=@8e()e|E{T2)5;LwG<Uo
zlV8$n$qp9SUk<x?&eti%-|I6(GD#f}(ebbPiH_GIh76`pIc3;^_IcM^thYjpAal`H
zb8XiABSt&uaV%N|FaZ-Ja(xS)%q&stluZ=gj!B`>A<UNxV90&v3-4preVQHk0x-rY
z)78R{DbGThZX%sgE6~C*Enq4%W)F0CXHOT+ky@*}Tp{}>T>m6fh&)bPG^YYcn~hUH
zh;!Sn3f;lW$LBIKDUT8ycq<X}0?aKP{>C3Jne-axUHp=w(J?tn91=?7Gt^&L8=lX;
z)J)C&s*jchmy;6C6s>>9X9sSGmA$%N)-^V+^YZdW@Y}?UJz*$~*WoT?V>jYY`1tW-
zZfw_iv%6#X%8K{y0Ue&D<fZEY3WZVjxzhB~syH(RrT{3Z-}RFmgYrB?gtRa$1I#Z?
z;>DI;>-Vx-6dRwMFO~2tq)iSCiQ`XW*SkSzv@e$=X&GPl{*mc@i6&>7yvu!RMZm)C
zL0QV!pz{Z|>m&$YzC54Nm!oB7w56Ppr&nf{!(yW_lAaF4ffQq}TD+O171K(%LP;b4
zj*;=c#3SAiC{%Cgotvynv5rJoN>frSBR2Z}0}sYrxC|;JBVG@KnV5l)NO;|2TG3!R
zEL!28i?Ct*7*$W(8^{09WqCZqM-I16%P7U73Xg1U$SgoGu>uaq^i9`w143&cJeD5C
zr*I+wEC6Yh!p&6jgT*f5Uvp`^bJo2Z?+@_&ag3iUBvk^+$%#t)q)q&M5ONBL*<2|&
zhbqU0zP~vK5*P=0fj|Fm7k_y(W(%ip03M0)<rRCZRQOjALKXnz_ks2Q8`5Vm61(ok
z=+u$J_;df*e)ItV&K@a0Hbwsp0_S6%(ltF*DsR62*F9ld5;vO=UrnRT-Vf_}{|dS*
zTmY0##XW`|p%YHe#moDJ>NKDHv32zmL4W+<qQeoZq74h|pivJqc|1G(`oE9o1n8HA
z-ztt(SCpjjU=XRohkWG|UZA2`XO7ae++BG_Co`dxb(R4*D#V+(KtxCUbBg(|hBW31
z#QW0sAusWEvt@IV)Xe26)+3PUTsl?zlKouOy%Vr1p?NBCOgjxxMfi-=OCXIV00XPt
z+jP!b-wE%Fe@;E|WyArUOUtceL_9--JTLjRl=8F)^$!|oDPpT9^9Opils!Lmr`a7D
z{dG?XUY;jP1FQ3#-bK>Hoq!BSsD!5(?0hJ|0{(NmQ;EC42R2$0Q`=x~FQ{oO988(}
zBtPD5oPx?rk`+k;v)J#Jjk|jco1zRr;ev9&>1#*}7-Wp@<@jS@^(!?%Xh^@m@gll6
z%<#g-8Y=UZ&%qj(nZ|yQghldFi6m(nSaz&lbO;{LMkQb?zY}ydeyL2$DDl?bVd|9O
z{z9>3@2qa0J|K-GjLtQ3K4NS*&>Yi#z3>1#p(f4YP8PY}4oZkCmAL?5I(X80X;!;>
zc6HX5>7wq&*ZPn?9J0boGKg$4lFR@Ut^1*6h*NosZuI^DQIFoe!)C{&Zn;z^CYsl>
zva&(!hKIpMX`NncIZdecYiv}s$}q1Ph1uePd32}u?CT!E#JwlGyR~Q!_pPFY(RuI!
z5FbsN!+TL=)RCT^;c*}+f(0H&5SrJ7Z08zYWjn<){&6Q%liEZku%}Mi_jd+j`>(O+
zAW@3a?=2m3*NK!GcdDhS{7T>Rj~FCuZ_8zrRj-vG-gf$2k6#zaNZX|1C$GBJSx+py
z>Iiic_t|csC8YP4A6R$D(T2&Zs}t_sd|YPWa<|iS_;QcA7%D9{Bx}<38X%IACh*?N
zu;j{jufAT|IRTc|cxKw~=7+ys)rrG>!LhyT$1?-&{2Y-;DL*q8v|7aJDzEUC)3A+0
zWTl{)j02`CsHzL53uW96i`Ma*lwrCSc~QvNnJ;KxmtK19**n5MdF}k|uDyGp{t)nO
z6V0G%Pt=IfC!&#EAf+YCeP-m3Spee7uv;@qh#FLdi~C{K_*<<!ebtg$A|E6VV!_%w
z)?{><706R4AruauJP<-O&SeD7w2kzIIG-cdF@XROA(Q0_I2_NE5w8GSb<`SJ_c?`}
zga(~z37a`?e&BoP=wLg0gzjL$11a@*n10236ErO0uw-)Q`<t?vPX?;b*Efn*=Vpmp
z7fGoMzupFo6qPjNFeQ(JNrGAwEOi(<WVjeOna0Ud0#GEy<@+T?pnIN^y++JuSk{FA
zgLZhOk)IY56GE6C?yoqmTaU*WZVcGnHTV9C)Bq2~=6HL8b&C#bc1-z@yE8Dun2v*`
zXv%CoUBpD~##m1oA2f083sWiu^&th{VmW32z9I1y>Z|ui(p9kI4!ENOztMaHt;X<B
zrIJoEZ_erwU9y(=n5R-$%7voKw*x(eq3gslj?rfs<jYn9HVSoRa`j12<lgVWBI{&q
z42mRr-qQmqP!VGiRB6tr`Km{nkN4pS0a}Yt)uydxuxdR!r?<P)%U!*}^$vNpYhRnu
zmoK92S~YVCRRSKtyMT<JmxpYpJ#u{bJGcn0s@(y-jU{`4aX5U=HJY!ZNtpvy-7_fn
zLz9P5^=Q6^l0@0Cl+w;;`LnkJ%PT~dR4OT(ATHny)#1B{Y|^y3sdA()Qgz(f4Xp)-
za>?vZ&+Nl|cnmgQTW^$EPJJADv^}-&<F(o|xzqD}06v>G^j^T>uzcB}b}w~X6$7*+
zZpgQ9Y>MG8iny8_O$9P#un8{=v+}YNgH$nF)Bwyr%8P2!W4AfcMdWv*KR>ioN%3q7
zT&JJSQHn_D)J_X<D1^&d23n_qqqa14VPnn0Hnj%26(%xkb8`cSH}inV--=hKXz<+n
zBJe#>MWd?kdKV7;-;bFNVEi0+13I-Nba8|ZtkZuDt9gw9*5?<@yQ|j7a2vmn?JRBC
zqS|W{ci`^J$lc8f1zhBAuj7a9jXq|R;81yx?DT-bX_CGd@peRN?xMq=%eDgsw4z3A
zYZ>#5Y9HKty+#`W3^IDuQ6WX^b|)yY8e_`F+3lFR_kG$ZLu_|LVM-J}a~^Jeu`o<<
zvX|3t+o@e>lVjz5+dV`(%*P3Cl@II_h=CA12;zMh2Te$jd48M^_^XaT(vh~81C~X0
z!PkS-rq<ThirqG5d3ku$-78(M>AjJrzWxa03j&ke0#dyj!|;87@dTIrp1>Y<k;w#k
zkanXNNpYqVTy3PJ<4EjT=kM*jW%XDo)Dv1A<(Yj8^11vVuvv5~d?q8#+ow)yO953i
zF|IVB?cTs4Jn;^hhe>iE^Cb2e1o(N?24~+5bBcKFeW!VOf?Oz8p~Jxiv_$omLtt%Y
z1;dD&@klAyj~XvM!@TSgbfb_!X8*d!Bdw>>gonTL7WI9E4DnMFvZvh+_A6Bd1Xser
zM?t;E`s-tdcjvqhK#|$xo#65J*U|eFwThhLi0#P=KbmXagYU@QbM^pI<O;)CBODu?
zzq2(l)@!+tZ{&RDQoR&*0{9JrNAQwV5*J+-DaRg)uZXYo<<ffw8F0LeA-fS|GUz=S
zODjx!VdZk-L49>^+H#|P2^na1OlPh!RM<v5<KxyyUcyFAPu!x<T24J;?-1byGH?#X
zz(I(P>s8uAMs~_is*s?IPZytI2adZT9S-T*kb5GhWcS8Ku?CT`Oc4JkpZz7a-$=wY
zTS10<E}HqDMGXdPrb<G<=We1p`fIB;IyMo3og(=r4ay!{grI~@Wty$&cZB|!JO@xF
za7$ENbZ(PsqUQIpeSf{#x7J(j3z|7Sy@^j;i^?#b4kUphzzAt}noJ59D3hLUp1oq<
zzd^Kk>IC6ehl6X9HWzNQxa7Kw0*jE>O+{S-%*;#Y{Y&s}+_=%RH@=fT9J!!xP~%x`
z6_N<Vy&chq9r=${OVr{C0aZ$LJ4i^D<1$^>1kMhp&;0;?Lbb+Dn<NvV@rT#PLH{J}
zBZYpXQP-0izZ)b64s~|!p|eDdasbOy_xk<*j=*Nx$crLhKR9%J<fksImj>pEtQHFn
zv`$Qev^1=kN;jBuT~)o+<0hb?p{e4+({-hr>+hpA43|^D)K)P79=te+b2L65q{Lje
z+1tM_O7r8Dze=nmO~66Od-Zvk1(#`MxPvj2h?P)HT01)0rdJi=o_voyqD3Linc+_I
ze|sf)Ko-IRZkgPq<CUVTl42O$j(hzcU|Vj+TDNUN%{x&IwCL1trKmx#0c07>L#A)H
z@XP-6bZPcLTm=wQu~L3#R`=yUJOxM!V04C?tequ!d&{EzdGLr!M41j{t_81!s=M;|
zIGHv-jd9DW<>uny!TO70OzvmD5Sl%Lj1;dCLVxX{?ay^-IIQ9pUB&BrPR?h?QMFy6
z?8Uqc9}sQpa6j2W{9(o~08Yvm0junf5bG@vE(=HPdNTrJ{<0czr)YICtj<rlfQg<=
zuF1`01i`w^imB6jHcguNscnYd^Resk8(N?beB39HOXi{#w#ur9a@q#!WriBt{w@aw
z!?v5*FBYP-E2jX;kEj*Q+b>3)KsP^;nDhXk(HB{iIafXeir|yL&VFPi8&WnjbgWlJ
ztPdQx^4qL6&~{dkJ8^V&^s%~)sAh<ysxH=O>F*&c=F;?REBN4Bk(`z?4rC*)U(tuZ
zF9UfVGFYFm+akDZAZJqOM56hzgG?4K{f+Ek!b{8sEDWXJGqqR^s#1Oj@?Q}Fea!1>
z!*bsF2C!Zp9>i(dgNH&JQ*-HFwB^cGmE9cwwX)6=;AH%Yw{huA2%*e7<g{)Ag+elj
zl65lXbuy;b=a)!QT3}7+8T#OAv{tR{aIqkK3djs4imx7$myWrvU#1yZ;3x)sjembp
zrEFb44LCN42!2?M3}^^tE>1X0xBgwK1}>-nQj=vpPkc7_R=6K`igXvlf;1f?XSXv`
zE4xoMUz79IiM`df-XU5KpW!)g)SKSrcAst2-Ay~&Scm+cIPr0Jy|iAkPW1M|Vvnfv
zX!2x<JJV{pgBP3?qyPSOUqFQK87ZF|xyzc=sxUizvcyN*;`e&ewaXwSfcvFyr8qVP
znS5o)xt!)m7E~X&MB8na(dts`bKku#ahwxAM{RfK$haRf+>c*P+qO(ys%*O#qM_h}
z{@9Dj7XCX(fFcopfW?G0R;_rnVz*byYF`<y#nyuInMoc5g7Ww89u-%0L|$afj~>4~
zU?L2?mnR(#otbuO&H(hQM{JBvLSpR3ZZVTMM?5lRu2WDO(JoNqba%Htx&!8DIbH$t
zJa}i>OkJAjQWJi;(ETHcTy1K2_QHS37f28Q7P9-g9yf7_({dE1L8=;yX^vFT%-1jz
z)jaR2j%d3X)nR2{X<0crP`-w$E@voe=k;Eyzqk!}2O+rXGWekCD6I=vR}4V#s=cCr
zKCT?qB}b#ix7}?ajB&;O4KH<H3C<gQ-y}3|rS&mO-~Xb*Dz5D*&Wl>PWmFwy;6C~Q
z3ewOid17MQbjnMcGBZ<#7^KpeO;&gz0Em-7ZeP-=eW3o$9_?WoW;RxIVh{pO_q}hU
zR_oeqSC>jqqToz^=Ix!GA(Vpy{*2RXanVzOJe$rSQo@+0O{14S1c+QCNe)_nVqDrI
z6``|s`@e%cLLeahet%UEJ_897WCD^b|8`;w^F%2cyMA)-KPf-*7my{j;pX1yestO7
z){)o(^a{_AmMPDW_@p^T3x<-N5I^Z!#gUSd@{_yPVL0Wwl`)Yxyio$g{8BVu4IgjM
zfG+GT6QIIRlNxaSF7=l;{%!Bn6fK7)-!wiMalPbfYI#hP9dkj~(bHIH=^P+Q5@K#V
z|NZE!C3uslIs^j28;J(ukl1Oi&J%*KZ4(!3P~rB__0|P*aTj)d?%MY}r`dt1TUYPu
zH%xp6$ke2#<1+6)p!t!j`DJ5{_vj8xO!lRdOddWgcc5O(!sIR7O=8a<msEVFh*9de
zm*MI5euwYviHNW|CmmxVq?cvFOQ_!1PfGwHCeB-%J3cuwsSXLmW=Hw2OLsqaFmJ(r
zo-PBNH45@69@O@d0X;FmLkAdd9w14Y%P}22g+PSY-R8@SlT=@&9-Wv6wDT(&=E3tn
z9=sHDL3R6jzJK~>73`s^B!AperAK;R`0B{6mnG(z&q$Z;`;LQL78Md~EabIyEZBDo
zcz1iaW(}mJI~IwLQDtm*y3q~~eUI*?YdWwg-S~bMa)MBIl%*4U4egwb{6GpsTtB;^
zzF;2fGR4tKnJ+xYqmP~b=G{&uwxjk*Ikr#XvXCpz$sXO-brq6jBTSSv(4n(<Qb=+d
zXKl#SQx<kU0SIz_1x6KPLevnNOF)ftdOG@5zm&RJCiX*i$>FSBg;R^|T1ZlPlHLwF
zldN$ZJXfISVwWx?UD`b+)?c5NIhEK)!E$C62-O;=4?o9<5ee8i(x~&7Q|C)8c??fD
zBM)RqljZ&O)JO;k0GgPrt)wH+SmThSo*WKHqxk*?-P3uSNe+-1cfnpq@)2?aAbUKP
z*sp5GwI086P_a^-`4;%ya_b8`mdYo(=YuflI%7ab{w~ax8YzyKkcb_Bl`?c<MP*@*
zWuImDn(6O!dz*?LlUq8TFLUe-hvSUsFI+}!k@K{3Wu$dPGlWphvCr|!`6K-nL-xHb
zM*Q5b5Ink)RE6v5JYlfW-O-J&R;;c)a&*-gg45f7Rxj!8$O(c$!UOXd&48_qW>^Z9
zc~i^@_x*W5$aMhVI+vW|j|4ts7$@p%&L834Uym|h3>-z6y~YuJPK_-do`9XX{v!4u
z9GHU8_vg5;VMqL5USW&?q)$|bVdn`T`Eyl?=yUjLjzhjzF8-ILKv&=hhy?|6nE8Lc
z9?%ZBER{Br=l%Z#&I!xw2mum1t!RMfg#Z6}sIEuA>katnUr7DA`xBODiwBsrXK3-w
z3IG3-j^}Sex;dzz-~PPv=L`Q-mq|<jAEfsJ%`3ft2U85TVr^IKOnsAZ#SL}-2{Tn|
z*LM|4>Z^B6P2X91dWwX+)FDdpFtgy2W4OZ|v$2x(ub9<IlE7sKM!H9o+Pu*>CbVtC
zV{pB6Dk86Uq^F-P3@H4Y1!^~Rbh)|dRl*tFmdgQit@)D^5aT-;U=lC_oAK)F;K72d
z^7Y1`z(<xl@7HYj{J4$wIUlPr#@rMXe8X=y6pN=(N{Yj(R}|LT8hP1rXZ;bOO&=JS
z94Oi=d@Ss8qSb<-DJ}^d;G-KHxiQ4hYo~&ML?VH5vGIW^2pS&F7WT^TMm-WpL=!-Q
zHc52eyt#Y%a+_XP@C|%kOg&(EWbD6E<@rpB2iJpaMgp0EYO0%g`rO<}mzKSj@HLkT
z&<i$x&~C>)YAS*ck;%#QAob#CgT+TU4Ubmgf4GvxnqSk4Iu%_E-7NL?x^7{Gk}eCe
zo^qozjjAuRE{h&K4?o-2DWU)>sDkf^F-FvTc7LeoARKXFr+6+I)rM2p78>Iyuvaf_
zlICw2>}ZfDjgQMKxVV9BomqT*_0QkFzrcbt2nSg}Br$+|wwHiYI7D`-(lOIZFn3gB
zwl160OR&jcM+G2zLH0Ap0mw5_boHc#eg0U}aS&Uw!Ad9V2Ns1l>sta~+QOpB4`{Yj
z*DL0cSC9g(I0vuATpd;dkLXdN8P~Sll(k7Y?@6zD$EEa4ylG&T{jZo&VC535$m=>>
ze>9~a6nc9_1So(o31b)WFb8;CXeyWD=ozlZI3!88P_asuuw~f=u_`VZ0|qi|0#k<p
z0#+~bxo-pf1<EFkjEW@^GT;sa##XoOgXJS{337_GMT)dtq>@X9M|S}m8FsNi-XPvc
zYL{-#M{MtVKiTu-m>qchLc$tIRq8Hg9hYEpea9g@r#fv>Pg>!Nt@{DDQpT8HwwwLk
z)g}9w;13Ch_r&0IGE60#$AI3|Pf`nD=!lHcwC<0piFaXB<c2n$!vdfqkR?3_gcEgR
zKiLro=gonpydXa@7*~_z%;<@ZDVr9c*$bbo{0NDv<q<bsuN@MSv)GJ%K35Ifl8s4;
zeYx=~N{+8c2;<3)43(E2OBcuiMJ8EF<`%LL-n<Lq%yF)86{{yz-MZy%u%^#*mO}sW
zZTIGp3GUS>0r-v*B2d}W<lBB)v|;Q$_fK~>7I>eSj8!^Wjw#GR*@uIjsaX1kTUPV<
zYl_6&wS>06?j%P%zbL|7TapIr)<)hpY~8bV)bCq}9O>I&p@p6lx-c#Tfbz3M=87>y
zR+3be$$M+qHkG-g*`6sa(LhlAsY<TOlY&96rgWt^d0p$}Z*R;K?sMPiR^*H}&=OU7
zGJDlLeg0jw1|faKfc4BCzGI`Zn=9kE041g6Ly;M9o~SFBu`COVPDp~`Vp;|)(YqHA
z$<_lJp7~UEy9TkbvH7m-_d<d42T{~u-ooY~YB4U`(0aBWzFH#3)7N*siwD@XD39hD
z`6C%ysgM*-EF(5KI6f7#d}F<gr4(i1s2NmlS@<Bxq}jl=z!^M~B#|RYI`hHb48K+C
zLgb35<=AMSK!;G#y8PA16?Sp$tlFlRwbfp$SQG*BSskXi!Eh^Efb_-guDbF)WuMk>
z8!ynWULOjLxsNPwbj&d^NZSMLHq&-<a2lumwz+t{mcUi-JrH3<yC`m@At139Gg^K-
z=4uy$w;oDUAW;q7$jO6MQ4kvW<u&E-BqJ`6viTN_|A1F_Qem6$aI-$P?_I}?w!UGe
zdQG16PAgGg{e>CoMz2VDT)fsYlC0uM$rA6@X7w!Sx&w)$j>VN^&ympeQV<+Q`XW;U
z)>F~O5nm-FXS0YcRCt9KqGM^&6cku&Kb7vd>v1SJw%)AQOBm{9tA73+qYI2!fAJ%}
z{!-yyz8`C}fxYH8zTI)usXfxr!wHz3Vr}hCRQ=lUX%uJmIC5knLx)6)zN-_b0loh?
z;YY5{0$O*yc6`{ep^#O^Ef5eCGv*oj3K*f^P_|S$Vw*Lr+z=9>p^xh6_bD`@d1CP>
zAm!maSeT90$FaDJOC}F|Z+PoB${%$me-%`KnQ@l-`&VQjw3=zz2_{O&D7<(%QE3&O
zQPhL{{JA=uX17JEM@}4>KiSzp<8GYXQJR(I<^+nrH9ODPbN29Co?oJfh<!bGCrg3{
zkoQ$9XV0J6oTTxSs3p`1Pnj?u+&H$a$qA(cCdEO%3=2-`mH<J`CE4jm*$VA_p4_6|
z91e_RiM<bWR(Ql$8`?Gbt&?%D7qUOR9yAFb(xNIHljTVp(gf0Qg<apJ5#_BTnZ>$t
zmR{Hj4R<w-VIC7aUel*n?{Ip|Or%r2*QB=v4Sk#<D%&qC&uflFqm=eU=MQ9FP`eD)
zxQ@1b@EG@z-JRnRH^`&}8B6c9$cewkbfY+Q7&we|_I^|LIC$&eCvb2(Ng0s0yN)KJ
z<VRA^BDzjel5|JML^#ZJEB=lLSit3eg=lD^0fJug(q=0ll|Vp3Iru@4E{LO;QJN%b
z=9{m3jXiw0|M$L?kgqb9;v}P)#cDDodsChWv>V)Pq-AyEm<j)iAdEB4K)Ut;1?Tfx
zIozK>uD=ddZUBM`&6V4)Rw-y#>-dN`XapPz7CD65b84(+g6b<ieF8Tn$=Ym$CwAJk
zOu+QH^~A+jBY|2*pgOYG#-02~taq6~{A>`aVyT0Z7*=mp&cIi1RJIudLSeeeL54t|
ziwGMrh@-IrUsZ!O_<kUR9cMkh^+106M86lfiIsS!^L|-xt|Yuqh|*Gu8jGet-zaGN
zqXn{g_d}YvqPgeN+xk6x!BL;YJ#tvnhrz+6fm(raWR>HGT5sWHM<{h~W8?c`v%|Aw
z3r$w!wH0eF2zEP>M;%^icKJ11=7{ICI}g8oDU_pqb5Xy@2r(nr3^Y0@4SSWv_u5vi
z+?=Zx0MdqlfEY0E+CCN#XYt63C}qSx4^A@TT=gHf77^#0<I{=_NbFIO(O|xy#aqNG
zlg7CjZIbq!r}LIWhr`FXD{?Pk!vw^6SImn>96wV%Jk7)!!jq8Gvt!lXr<?a~ShNF6
zlyja5qIck`=RQ&;7G{>7F(=aBo8ajxv|k{XMr5vM`*x#(lL7``YHCgk53k5Mq>0E=
zu|F87y`Q<U<>FKDy>gDGDs7`f^xb36Z!#1Vr7O(*YI9#PJ%$;w@|+%@>=EU2)c4C0
z2yor+0_Tf1YtuP8U)?1F-uIME?^Xm8wmM3e1K35ZB1OVf*%iYl;`~`By>LMo7Bfjo
z3D;zqY}=R55!Lr@nsK90*s8Q0TXMe9?t6j!=HPlH4SA0&BdXCsgCLLhiQN~L+A=bb
zDerI@wpND9+r}HS?TFgLxga+wLD%Kj`iiwGt+)VzX0s1|=HiQ=HWtJd!ikjR#5O)5
z+Kk`@uHd2VF`umE1do-3i(b?8P1cbYId)*Qq{;(1RAldz09k>d%#oMwG)H7q(1_#l
z=ynjk#`J?gtA<Z7Z9;X=rC@=y=QPEv)Ws7ECYMRlbxO>@yF*z;^uCjRH;7^l*pola
zg0zgX$tXHRc=XPh9++GqYc7v@UAQh|d1W$}r_;(+Ca7m`ox51r_!Vjji-RvwhHq2m
zu6a~j4#f5S=na4nbNkid|D}2F{bP4W<ccOJ1-%{kWJXgv1IWp`eJe-O4jfKuvy`40
zviD!kv0W@&J-x+ZOQGftGsQszjW^w#JP5%fW|xS=TmLFmV#^>B?$0o}`9e5D&1On9
zM><Y*HO7mHRZDn>U7Y=`_*#WWI)%6kwB9~f&Av;3Lm^t+8{Fj4s@$)~LSH)CB6V@b
zXg)gMtZ-11t#!l@)l=0#XP?*8N>a!wRRnqF-l@0Jn~$35)XPa+H@Dk*V_mY0#R`TP
zlIX<av5arc4v_ipVNwS)%^!f6?XdUWa*yTeeum(jHGF4Jp!x=(;pcn{Rm0Ui@p1;Q
z-O-I!Wr~RjK#LzGMb^9!AI>$4$c3elHrh{MFgzCF4y2NS19ws5kc8es*A5I&EBdmV
zjD(5!hEEWG2!#5qeXJjG4N2^gXiBU>eRiCSj84bb(g3abdG1TorSwzmmE>z$SH0OF
z>1Sr2xealp1>O2gCFeYHpBC)-$bX)3VGd2E^^O^FYvSJT&luCZWDczd`iasbpjD}e
zXAa}uh#lEbMy~f9d$rYr{fyDyu?Sq_{JQ0`iExImT^7Jsf1EXg+&3rCTeW%GX^+vc
z@$aE87P6$11^`{)?sX>QIN;+LST{ZtMKg1~o^;NzC~F^CjVin;;}*ui+EhvteK_4=
zf)>#sp%2Dakq%3tP)O^N#HC}!6N?>UU@Qyh<5=gCSsSfn%YHn-dL<16Zrr@0?9C3V
z4I8imM?kL+hjLHncuGsW)4F3yZ)$&Bj{G;j9pX*uIw(b?NL92fTC-acOaP_Ov$<Aa
zRx1eIrH%HiE#Q-Tt6yR@^VYg=-1A%YMvlRgggOZVe6{n_#U2*N7<nrJXBSE-fwSVW
zOzQixa0If8EZA(241h{h#d?4Z?rfS(G)Eu0{gx!s(zRqo+_%eNbTvg6{<^>2NRUx0
zG{C+|hmxH{pLvx>E4k$~X<CjdK)nJJ<EdJ-+%;yWi6TF~LUk`W%%I%<@Lm5Bh2YA>
z@xSIN5tMj{p1FXK6R+V{QdZ7*JrE!Eq6jRWVl!CIja~!<W9VuX3f?P4vtL(l5BC`b
z%q2`N4+6Oag#I!STLs77i6~A94zV9slXDSB{R@Hme)b1C8Ijw*EI^r+A@fYK-xUX%
zQD#~nU()qR7I_%o<@8K){Y>7_busw~ZJ|&)8J@IZoppW_Mb(U0Im1q@Md%KAp<d(R
zETu!`E9SL#=8V3FU>azWRlfbBPmmk6va>qwG`u*MnEQCPoT9d}{(c5NG7*St9JXgM
z9Ty2w1I|oX%h}g<8{K!t_7ikrw0o0Eq9p0%OFeC~BC>EYYF$9$MS}r}(713p?L*Lf
zQI9j@4EPI`hZ}lble-nxyjLB{C3Bq`Tj$Sb0-F*3$d27<^WI7dQSp$MFH_}2Y%mp8
z&wDV&`1ocGG}XX)oP|(RKe6V?K`oi>AdvRE3*Fi97x>JCtxZq>|C)Q}n&S#}9P7oI
zE{~8-VV@o4mA5V&;`U8__IUzYX)VpyU?FG7v-1>H?eP3%an1pG&Yf}-#gNhH0lVpv
zYG^qV4o$bRKIp;|APKPi!8&f7KQ-Jw6Psr<fCo$#i#V87%E(ytK78+}Lmuk-_4L-=
zhK2?NwcSn*vfe2pbz;I(L1@!%Ld)|r8R$k&)%eyfuaJ<Mi1zB&FWLhKZTEN3n5vx(
zDH!gqE}%kT83PT+mT5o!m*GmrReG)1Ol{86)l3bVNIbSyg*>3+C(i|54Nr4Vgq4S(
zXdw_;`P>8I56{>4Hc)-fh|*5^$?gUU&44sCs$kt@EDTGkVxc0^B;aOp`k4VneUfkC
zvU{2QvUAw{pt!n>jvbS=@c&~QVzKJiR}(f$%{Bt6CI0xj`{r61)zaH3JyxVMo9=5~
zZaQ=c$#0S*e6MHY-N<&+&^?!;JGSrnvZzOAV{ylt@wPv{>huHXevKsa?;cUrjeGJS
z+qX{XbTYQnziLL~0z7YrDZ9B-*Z!}qivV-BSfaX)nbQq|FCu*<MK{Z7`IV7f-j$Sq
zNak!mW-;$^rM^nmx8fGHeS9#NhNd<SxJ?I_iRKG$jz~$;wQ7=BW=_AyGDAUBB(p8r
z>CCWI(O&~NKrZsAcX6ZKXQziE^<BaS68Rb^ldM_GQ|KAs+(nTBg?x`{&yI<MYsVJq
zw`j}8ojrt><7w|Cip^C!A{`RLX%5S!dRxgr1ezHc85?>%pFb<EZHw0ez4-Qd&*i8O
zmg;B9%Le<qR46;VNpaAasSnCUi2KKCoC@z^yCl)vxIKz<HnK9PPCCt4`fYF`zFy-N
zjfZ*iOWl#D7*l5Rb3b+DwRMNWN3t1h7`L>n4IdPdFQHnwEISrwt-)**$dIi?pgOHI
zodb>-r<e+okOeft;;+bgT1R*8nr9crMcb(;HffS}Ulkq?a~?;~AX?_)BfNb!*d4BH
zjhdDn`3)GBrl>zpVYH4$CZH0-85Hds01tGvBXnv`YSqJ9X#Od>_i}>$(%au;$8~7J
zS|$cOME8n}<kv1=FTGOnwJnkL+TK}+HxT%rkA8!^c8t+>oRVimZUco*a(Q9W#j3FQ
zHlHWvyW13YpW||O-6tKkGTrmW{RSl&w1O>~qBq^>l6#`4cS?HJG=#L>;oVd6=q5O!
zr8o|4$f$N+=X_aD6az_Drjzn0g@ekt4^6H`)4nt<0TyH%ULN$RJ)1s9v-ZVtbOXxf
z%WQ%dAYQRL-bB|Fi>8f8*ePz#XrT45proc|5gRwXh%Rz$8RkRm6V|@z5JM^WNyzxJ
zZ;hZ%H3{X34Ze3ei0a5&k@CP5!o=q+?a{~bfrdPQSUL|VzU%IQo2h4)Jd(?6=JaJ=
zutAWZnoDVdhlUv?Pp|a`aV{`Q0JF1+b66vyB_)gx5bIV>6WR-$wOMLO^tD>ce5g!k
zQE}QF1q}k(#-gaR+_Cp(CR3dpv|9oW>a73~2-=E<He=B?5N~qVcB7i|ZP)*67o?z`
zd&ERl7>b)8V{4yBjhuEnHnk;^yl|SAgu7Ly0DHw|qgyd>S>&E~n=;jK&%2>M=Kg#G
zVLRqXoKqXDe6A^Kbc~X}Ewe`MT5B6^4u=rS)I5_m6tIcAYWf9MjCW^;{CBrKY`sJ0
zQ|j`omLft07w+>(k*0C{XoqBLS|E43N=pm!Wo6Lm&Q7#Yrw7_5fqt;EtD7Q!P*&lv
zH173<$V`XccD-J)DIPv;XuxkxWZDx$Do+`9^;@h+*V4Rksen?AY7GJ*>8~lgj;bs+
zBp~*PDEZ~aP05&~Tw8<}p=f@;(YWExHqc-1%8qK$PG(*$dIQ9{@}1pgmOZc-BuCkh
zkUy#-K+V7y$F{EO>xMy2ty)%r8%@^t8KJ^U_I~MFc{WRS1s%%+!WcUW2nAtku4&x5
z4&jAPPxQ2vy=*2_?Si2_XUnJNT+#ve;OlcO3Si(%6BQw>{a`9_!t*fp!%MRfb?m9b
z65NBwZD7APkChsSJgfYY`(Hm!93*!&bO<kZ%_dk0h>P(LT9bKh?!(9+CVJ6sdR3!j
ze9YqFd)~GeldJC)e#4_Feh>eUFoSpR>2d0>E{X73Y@WRR3UyI(v@Bcch`CcxBEC8s
zuLKGD?e?gsA2a1C?GrOuIt4(k=^0}5I3sgL+ZYId_i*psOVxRk^5)WmAoI8DPmxSh
zSq1UUO-6aOtW6y#!|;Ul`S>1n+9LHJZ|s#4j%sw|^ND+?dEud|8hN}{eg$Uubc}=o
zCcxdA#eHO<a^1oqv2xWYz1ie*U}oJ*<mKh8K$9I1x2k%bcC9&UD4q80CZlvWZxPks
z!Y4m6M;4GeAZ-=<CgHBSdaor=w1CHMm2w|a0KIhFH3)>lBQ|dz-Mp*u6)<Ts;6)NL
zfc+TZFbxS@-&hl?r|gDePeY_(s?YE0Wt63T`t*tIPN5CU3$E{8FaBuglurNI(Am|D
zQ>fOsNL!js>}oLsjN>iz3LusNX_jCsx0Wcv0j5Ox2zYHo&1%nqOF1_|gwtC)`WpE&
zY>@tQKy=M1O83}+Jk72`>8jJ>Oe<)J%e<v4${-Q`+!Y&D8{ftIwI%c0&|e!efu6^o
z{gcFy9xfS?Cp?fKe3fgl;3*Oq7BLJ6W+1ol@Zl|>3NQ-D>?peAPlP!SdOMY!2h)8d
znN>BpXu?5?6Hg2%jXzpF>z<yQh6K67zlCNYbC#Gm1x5uuuI1cY^3t}DbY&P=!{U}b
zl@K7vjfuIVS41b1`;vG>Ay=QfJEUQ$iW@vwo@Fc+A$tGUrcD({y%x3Bn=4&IsKFzV
z^7lmIt`wb?q)x>XhP;UamND<<C4zJHq|5Om_^PCNJRUi!#RGba4ge9(y1Sm3);Lb&
zyB#%R^6>CjHhgUN<J@LqDoX?MguBmt6245{Q&du-C{+pU6Hm$2FBI`R>c<4&{G}h5
z0gYu_e-N-Jh>N>W<idpuX%~iCN!Mh2_H$!&qM|sdI!pJjT+Y_{`dQZ11nBt%%Bf3?
zDcgY*Um-yNaTAW}SSd4*XO>y3_}7#$U@Qf2ZZYx}+V`dxKWcl(t-^ih%(8d?qqdLl
zp9U?93@_)R|ECtaIuQua##~=Y1G;gPHZJNTEd~0l)G<BI?Vod&rP(1mZYCD4%psvK
zAD)Wr<`omOeFW7t&UcaqEOExqLMN68VIYw-@&8SG<sVSR0`Of-P%Y7{7u!ISzj@lH
zaDo9>Q-?G$4vSI$7k?I0E#nQ`IiE3vWW(~pzwf0k1xPNEH_e&-8!;TMl9UMCxogQU
zO8BpYgSofhe<6XRRsJ!5<L`U>dk**S9jnVES12j1lpA!hiJBu$Bmezt(H>sm=Cl%!
zWD*3QQNmnBbI{3lukxJOX*GbUM6u>6{k}$^#x~j!e|&N<j#?i0T_4v82iFM$>PsjZ
z`(F#3Q$tP{Up*RcgsggM^l{hgg(#LWCzy4RvA{zf;E&{DhPQ+KMk<a+6#-Q~oTmTp
zUsb14+O4~Pa^+nQ;r6IEVlL@5BS06ar8P%zmbi=qH1Y9`;YMelKqg6G9I+xhDzG;H
z5>;&eTk_|}BEOKq?>fuA;6JGI=Ne;lUwQUyWBfWoZil5f^t&HFq=ac2yp*pLo^{*+
z`YWn+IfxVc2-S+;mgG>8rp_-QxNl*I@`@wm!HCECUD{KfC$w5CuzfAsy`J#&`n+M7
z$+<44yGQt^Bnuc51=1jyQP?8MnAazj;1mIZwp1z4sTK%|pgO}zFgftLr$Enk<Esa!
zL%*(_z-kiycmzH}+<Pw#uZo;1`kCraI*S>H;Qsi_i<8diu3cxo0T*@&rmmoCGQk@=
zJ`vFI5z%CbuJ1o(VeofO{z-8VMXXhL&v6<WNYcN<R%qS|Nny#c{V1L5oSfoud-NF3
zlZvO@Vf7Q{RKQ+e`1|{@xIZ30e}brXmony;{$`CFkICB#=}COLHDq;sXEBh^SLE)A
zD&@nbEHP|*Mw6~|BLGw1ctE<%b6nr!u0j_@N<|fd9&&&D{X_^=-k}q-hIWpE=elKH
zzi*TxN!6J=*LCb~>txocY{dZ$Iu?}Sp$5#!n4X>Ewg2|RxDIcUXKOoGRLnsee;ViA
z4&+BwLXrR;F!GsDAp{fQ%UU|QPu_}9`YA){T3BM{GH<3f9?Kc0Sy6!_u8ow~PZ6NX
zydn?iTfm3gO2N7V3<0{FBNbEN^ksUHEAdTPMhQ@gu_7GD^=jjn0@9i%nbe9&QUg?E
z)|fmZZfX7P1V;pXbOp#|l8SM5(%d{8_1r4KQpJ0N_|x!q6W5-3bjat&CoBT&W(>qN
z3YX%!`JZC`sf0OTbqFoy3H;}+Nx{IKm99_DpV-#<6Fc*{0E3ysH?fYA_(|lLAC85<
zofOKU$wwaW-~RPaP{)%2(k!=Z{Nz*rkVy~@a^TL%Pu#r!eJ0FQhxpBY;QNX{1se7r
zT}{B91|otl{)aOG<tHgc<L>`D%Has-m6)Lq!A#9v|9z$oAc*eRuV;sScYk3E2@0US
z#&K=>3YL7!KTg?oA0X2X_r?DKp(enc90}K?{`*WqAg<w|732>vIm#QG-AH@0MB9{@
zJN0j5e;%+slb7fTVfOml2vo+FXC7eGSVX2eAM*tT{&aF+76QS>+l_bnChiH?D!|3@
z>uh^B@rtOeh)ss_Dh<aAP?WJw<q`?7eVRAS4L{1)aWUoVY$7D52u><XCXV4GPE50l
zi$z}(#UX3(Q)$|F?*ROY@8-wMiY%|w*iuORO$Mr*@(oLt7GMPlIlM`S&Fw&mrk0k+
za(n&vQ5eJ=Ub{+vgoaFJ;<i`dxL8($$XpXqrV~`pZ;)0TlJuB_H}`;nA%5)QtOHM+
zyKtqg&wxLQa$?p&`W=_A-;nO6c#WXPPDp?=X3_@T{wylyp^fkRl+I85>=7L$HRGli
zIojk(V=Fc}=H|T10iEMhJg$D2r%oR~<&SG7Md}^6n9RbXfgKa`^guP*kQQw{sVJ}M
z^3`lQ84vxEW9<RxeTASV&d1pOJ7E7^Vvr#&Kd{&^(<qu9s7C$?u{%6NXTblb+I2}%
zQT_)fRp*Oxel6NbNUX;b|7*ZOh><H2JV%clzG@BsrzHBpgAjc>u{dbweB=kraFS|4
zR3qjeT~auHMm_&ox~@KCjy!m8N~>h?9LjK~n=ZoYXjNkt+;l(YsAmR=KVI?=mX%0i
z2CN^3rn+lDPVKXXhYAM%^L&suY?G9CwNIpGV2B^4cHK)X;^f+OZmr>+KV|{^f>p>#
zb?mva-MCRFT9hFei;~fnf}j0Px8G}%J4c5Cir5Q&lHi}Z4`WDR$TvwC*VA@_dZQ~#
z72p?nUGd|q=T0g3@<Gy$`yYO4nx8-BpJJ-MpN?KP03P}zedWy2YRo_ZyH3^%Y{OPD
zs}>-Eu_i{lg7aVS0VQ}nkM?K>UA}-VA6s`~ZN8%fFsp=GV&xSFpbsB1oIjp;6+!Sh
zqQ?FtOAw!}0@z=)CxH<EZ?}U&RTq!Oj$jIR`-V?0+y$mqiB+xoK5;<;rTo+z#-vA}
z_!VS-UXh>x29r5{`27mbKMCj0dwu{KNR2-u9pfgFJq|`;K>y1<Kdv}JFj`vi8U4T9
z^W$G*^gu-W|A$+B>HOi!`IDMvew<yDiHnafzSMRs_687?`6DcW993sqe(L{z#s7tW
z&wR@rN-8R)si`T;hmG#PDF}nyDSj%I6QQewFa6YHDqYF4E$Y*a6}6IA61d6?Pv1Ma
z>kY@^sR18)!3-%=0NuEO%gPE79KdM1P_9MaKR5~-T7ls@gVk;d!vzo2?PmhXF1Pgn
zg~&rz1g16HMqhE|xE>uxd-w=5ZCG$MO3mmaBSB0(>oaZSgV`haZ2CZj?KkI@lR9jU
z=+yWrE-;H)Ux%BXMi7TIM>~%+!Q#Vyvlb07`iky4p^`5meo=rFhis9t;3m)O8#kiL
zk|>mjEr6v{>_L1dDa0eXz5;~X<VlB0FRPYL*ggV!5ltkn5AO|<WDh>r>l-4VeTUc*
z<|lmy0(#=K(k(cv`rB|{w2gb5)SvYS?O(?!+?2MNBasP6!A%eGC($^!$ozoTA&GcC
z0$_lQ)J-f&<3E8bp-f;i8Uf}G4-qNdD7Yjum(G>jAhKk75eP|MDkzcr`TQ9T<An^P
zR_{+CMh>g2L*CA7&ePt?ccrF^EGNs~FIN((4Zr6Wj)f=Gb|`k%Ku6TbZ_y0x)C|Fi
zg7pX<EtpRz03X4c7+O%ZbY~V$fM*PXx4bP(G$}o%i`GYu2Xg<D`;+J)I_Dk$<RdxO
zm$P!9`2SG$)^SyB&Ev45NK1o&bO{nlDcyMp1t}>3De3NRkVcRWK|s0=9n#&>c>w9|
zJnu&I`Q7JUc)s`j=YB57wfCMiJ+o%k6sN?=zb)bWY&uzz{tOPtrpXyyq@d}S0o$zd
zF1?OZA1J*$9f5{{FV3s6&PKY0K9io+#Q;rzSnhr`L=m#$p<1hYxM?v5u5BJ@%IjLR
z*>OMoN9;7<g(P)C&ClIhqhM(+Ii=JOHpUEJ^SVHwdyez9ImjF3aU{)j-mW*+pL^90
zLwl#4>)YjjTo417U-lSDk#m<|k%Wc{z_?R*@xdp<kV#c$!gIY-9BMtp62>|}hrYXq
z&5xk;WB?t5gVZb5i{idGOqyxOSEHna80ePm*Yz$V(_-f0q%T#Lz<uyQW!PCB;|t~)
z?u^>$Qg96VF<RX{v#Ft>uI}jutV6nhz`)8Zy-NOMox2bY`U$i9zbC^~KItns<U!Bx
zK}+FBJPWX6nR|yR!m>x;N90dbjO7J3I4>gP7VWh*o>bp`vDO#d6g+-FSE8A(u-=>}
zb3I4;Z&Q?kKRHOzklD(M0}$^eyo{5p6eKYrcEy?Mr*<aki@n1vTk^H>`R}1=%O!mJ
zgL{yynJhV(Q7){vRz3()^7@(l_t4pjh29p<dj1$+$F>m=k}+deZfd|Cu5RY?QX{DV
z87?<qZ|0;KPh^l>jd{A^KO4CE@WL522WJ->y#7&Bv}k1^yn!Yg@jYnzBFQmzaK>M;
z<^Md#2L~&#wq?mIW5*(fUYP88F|3(inEz7{y<eGn-I426n1eEBg@HdfNWC=*&0JVx
z(pnA!N2)jd!A;|JfZz0`NG^Rf>DKyVWqXMNCt8z)d)0%c(a|(T5M2<jhraM7Prx>)
zSg{4nN~6d&31FG{z@zYc(M`RJHsnD=Cl+IQgpMSnpYBbrs%Dw-L9u_^3vAwcP&MJV
z(aO1a+a%^?FTQ!sPYEw$<SOOe^!jt-t%d`x@a@q^`Ul_fe9ISS)(;TX1{$Ld3K&)T
z8(r4S@^_L2Kg>jxiY0oJJumiRD|U!WC*#l%UbXLk=kudtUuZx10b^?2tJc-nq)W0~
zavMu25?PfPLqXl@KPdbM5!gxUKVq?Ohdv^^CQk*$yW-bEA8eaEFoXCXyjYBdB_o?t
zFcsCt#rJ?;v*j#%t`YAi&Z59NBePzvLp}M`85BQxPE(;oko5$&H(67Ss1k{M=#6b=
z{_L8J_*Bgcf72nQo!}TcZ#~r?i{@4(o^5%#vIoVz0j{U*7%xv`38CA%Rq)1dpB{uv
zQ6;7_?L<b3&+b2+vMCPAx_#_8A1u7=w)P*M0@&tle(`H1Gg2b;zA;h^O0bZqQ>~wn
zvy)(4rFNm*fOL=z-qs5jX$b29hshnB@q(CV4dSgB8=STM1s@89K78YF(W35o^7&=)
zBI-Hz*TZROL|-L#p5Dx8yR6fq42L;(*enPd3Eb8!+ABXXU^?0FkSUO_R%hmJNIYIG
zU5)>>&93k{T|hAV@hKTXZJ}5|{;~gYi<AimRzfM+=6hZ~cl&5R{Y<hNAzE<hqZTpK
z&fBK_y79Rnp{2ELEp<+j{@P#YYxlOi9~fEFL@11-k!U-21cl1zu{EJT6sHVjwS3j$
z7CcS-QEjEI>_gJTX;q8sTWjeKKl#fZZ5v~ZsU`Y@{B38qe|p~CsHu#Vzu0H?-?5=I
z3Ug|oB|yS-e(P*E!j_N#!EKkqFAF?fF~U}?ED&k@7EQ3GGPx@@x1v~z*8|Vs@~RON
z3A5%{FJUpKA#NlfLrE`DU(+U_2|=77rL;5sas?J+x>8-D*4(jk?o*oO=&h7%GrjA9
zb#mcU!FkQ<OpLPvsOsWs-vu$TaK80HTN<%e?inIWHU;ZXH0=_BDf}*5wU?@-2YPjR
zF=4>*B2H`p#|K7W-%ugF7Qc~Z4da!thCyCx&4dcF=TC*s-&y8Im1SiaZHlG#1Zlpu
z5<$)`F}(j%1LA>w5Hk6#SSnJY{nHCD-d8`kC*iF+#g?N?q*ynOhICMQcMyA&R#YSr
z9K(}7mSQwrBqfj%75)0~q#BrfxYs%N))$ygY5<-IRv@ygZwpQoaJ+9J09Y_-E~*Fq
zza)+CA%c>6gNu6^9t}iJM(K$_U4w#Yv7CNpgKn}PF}SL@PpGM<-@&%8MrcCWF(eVU
z(8`BUhBmBbkwdTn9J!b3EJQG74M`tMCx*b;8>{h=8g5n5=&%J0%Pe1h!AZ@$yl|z8
zQUqkCqR;E1TFI}0zA6=zKg4jHg5^4=C<T|yi*yi39O_f1VRFdhk<UmW*8e65|Azg#
zBO}qN6iVcs?AlJr*EvruEp~-;n=*Mk<%QB`jn-s{J6AUdU>z|Wt2vdmufZl&g2mPJ
zXmNNi`1GXbAKx!h2lhS8OZLF;&%T%8#-pqYxV$K-x%^`&=<aKR9E3Cm8_{fhZ6>=V
z(Bmo^x_MHFom3PVkkC12#;Kxb18qTOCuAt5`p~H2f|1deLS@2hMtq6xjcNNTwiHR@
za~r|&8hdA^>0J%4PG?+uS9Hk9uqP<0b<~C=i6!RjHL#&2aug*tF+=Bb6M56RzE3WG
z5@zL{(uD__u2Ud%RZ(O~+`~Tt6ZZKHtq;029R%HlCls7b3&doW!BX3;X?iITH!MCy
zUba$1rbwwqU5V$k^PS3?%>=rC0!g~?r)tgQ`==2_<0Bb_VH*?@7cYFHIMNp4dTBUT
zkuAXEymCT_FXG-il6Qzx{)k=cv*FokTz`z-#<=K5-liYGQLb`F*_PGBU`XtOdYKVD
zZXCL&2OrOs>@Q_>)?CYe9}*0Xn^{uA8r&3qkY5l9et3a0<)k^%i?mO6W(~sBIrm0F
zycy{tG-tH=1MQ*<_vVuZt8gl{^YhAW`u8T?FyJ644u`Dh9DSX~B0=3NXx`%XhXkaQ
z8s*L1@N>Lx1>x{h3ox>booOD9JXwiTJQI(>qqw{iF`j1X!<PGLyvk+tQEK0V+UKVs
zV2Ett;!7jsumFpl<=VJU0wHj?P&p^<5G)x@0|#6G3%g8m9WMVbMmHFFc#=L&n8B|^
zG-c33z1{AuX%gZOp>yDgttdjyrwfJ}m9o{F`J?!)ta^vYCqu$ZRJ507OB#!x6lWrR
zUOjm|GTuUon@F#hWFKp2(abu7rXstr3+jpJWZ-E1aj}?qwpdmN4Mb9y+;!zOvqt4Q
zXFM5`qVKioA3u2M!<PpSp!5@U^UGf|zR}uX?;`Q)za9UwwngcI#`aYBWo*1zv8+-(
zDLwL>Nq0{v>W@+b79&k&&@wtFtHM_B3lYjm0|sLBvu8Z`Eq(REMhZx{9cH;qXY?gD
z_K<tdRYfB_`@K7|b^7i;*kJMR<?Re5uWYv%Ckefcl|R{K<%ZiiS9iRaD?irF(5S%)
z9aT%TaM7O%h{WTSeK;O`xjr;?1VV6H?}!w{0dM#6fDnT}Q>vqsH-Wa(;nMk9SMi2+
zE5mlc*ApH8j@`aNpslIA?)>m4!9pyfbE`Lv4K@9Iq4rW<QxTCI6(K1m_2Z~VJ<0C(
zFeWy}3iz7xO=|{OOYxrND49WeV2{2IDc5UV;^AftRS?#ahGxLrr`g^AZlQARFzg5L
z(H*0iuQ@7x<QK$>^6ER)pXIki_YW0Zg{MEC)lKYsCBy=+nP5<edtpI}SVov_3zu74
zD78<kkfK<8uU#slOnb>@u{ml}Kg)~NEL62&O;OBnW8V0p)m4CZHVJR0J5jL|Iag)c
z!Hi$ju$ce8hQ?<m6BV{E+06|oU>1<W=hR#m@*kf7>ylR(bEy>Rv@?PqB}Nc2t#OfB
zWyKz<!YcV(E`Z_8WuD#WT4w$@1;1UMJ`fB}K=~prZ>ZztWvHs-j|`0B*IZ`};Z1(Q
zu@>*Os>lf%eIA+&ROgg5{ebmE@lW=l-$Tz2NkW@pr1uAhjO+#Q@gM*!l#$V1bB+B_
z6c`*Ovy}%(uk~LF@K)Q*At1@?JI`YfWqAU(=f6y&ZuP*)L&D7~xXkHP>LbH%7vmnw
z!%M1Y2yV(Fc9t?8S94?|7jOjWTMqc1<i@{w=IruR^z1Oj!=D+m(^d(sLGBF|a0P4B
z-Zoa0KRq5q!rWhF^NUOWl&szKNE7F~%=;Qn1@c^R2IRFAV((=>z<>RL$$qVP|K{KE
zp&}s@GAY$sZRlYrDxZ$4>{jtkE*{b^n2nNZW3(DIZiTp<SJ`*6+3)*j=xn=qrj?h!
z`YIVaJ3eNF6USy60lk<Wk?R3XD<AaKEk?~;w#%zI%&}VxMf$h<%7LSEmFUTRP{ZaL
zeZCGAd-CO7X|Dt7w&UNT#^f&MNrg|xJvY46C+TS-;)(Wl3kDr*+U{Etjuq?I^S1v{
zr5P!r@uqHifpj?Ntp*#OPh40744#P`T<9>BgWJ{yPT$gsu0dv&374?c8c~fDzCulW
zW*}qST$0U)ROx_#^vkfB7n`=apnMIxmLD5DgS=;$9BWPRJ2G*MfM+7xW2jW2gh<sR
zF(mwfmh$+goWHY-Yt*6GyMkY1ivPwefUlKMWz%`^Vme-lmfdBCnTJE;XD_F^%NLNz
zq;>Db&WDPhmeKybOQ7|M7M6q*$Xsh`am(d$Mp?P`7m<h2j=1MN#=Msc#U?8c%Lao@
z%cg3RyA?)M?PFAJ`q_=aU+%(GQ1FG4MJM8D3;(huBonavjR(uSkE3Gxj4$p?(vQtW
zjr*ZYd(+TOhB`3ggcpGO4;@Bf&^3+q+~KoPg^Tf4mXsOM6WYl*28DA5h5Q-_OHH}o
zI%!k;D@G?RONwhCeqnZg4~bt!M&m#d${Ce?)J}?+4OS*x#hu=WI)d-26Y@mP(AQb@
z1d|9hI%duJ9MX-UXN^&p+Nu%PIdQ6UGl+%ls#-n4xh%-n4ZJKLD^B~n*7=3DJ_4B1
z($d|IY*V(1{r<s>@=87<Fv3K|8&9Mq7swZ>aU$NcN9;>EAsrW}Jc%zA9>G(9kBC_X
zes(2>jhdRwdq+o6eVh1Th>|OV2y;ROpk*fU$<fu2a}qZ`Q>1b<&sO^SxdtpYA|&x9
zo>#WS*{z-r+qFyYDVW{`R1GJif?kBGQX3h!7<z&`nN4s4F4Tg0IXH1-usLeMEX7?~
zHM5vMU^{PG*@LXeFneg)HX@DD7=rgw65{SBCR{OeF;9h%ufpIS!6^0G9Hg&5o7yPc
zD_*CY?c9`DC3=scUEA>BF}FU*@73>M@V~A9GK%}<MaTQ;N##?vteSS3-Sxeo;_u^^
z*wGqKSXoOgy7sdyCO>}3`6f<7PZPFQ>-5QKaZgd~*(0k~Bt(3@+=9!E&W<AN?Dv41
zIcc+7R=UQ6rh@;zul_DQ!<u>1u^>jLH$_q>8_WqAfE$fOilb|m<zs_i-Z}!)`RY8H
zl)PW5S+_sYg~ibS4tl3xzTXutEm&!7(g@2&vJuNf2MtNODFQQ3`pn!qD5Emb_ySGp
zO}VRv5?W)o8+w)4IwhuiK+EgJ83p|f3j;gtl=}xyE9@jXD9ldfO%8p55N_oo77~Vg
z!0cL}medB#foy+LO)994@!-htSbMKh7furtZYLUy4P7s`5)bX^rW-J7eM9y)sk-Kt
zyY4RdY2}x7bzv{HhpUS*C%~Fp#Sb6#X}BCmK%-@6#(SVnN7?w*em(kO(<bc$FBY%N
zlOpxuQ=6@@IM=<5CYRlo#prm^P%=EPjcQC1u`t_^NCJUhL`k-V<pO3>=x2VPKM)bb
zG<n&NN0ZuVI=W+5NotU@<K|5iz_6xCaH`Yk*aY*)1RRhir=?~{P?ed0s{s)VQL=07
zu}qqqusQW#FH<C`KH<(fX>?^X*u-Oz3}8*0C1mX(+>_xIMr@I`(L}^8e`OOFMkcF5
z13UQh1$$iu;?|{hT()%Xpi8HT9bKvPk$_?J%FDW`wCOVFOPWhae(ccoRP?_ihA$6c
zeXLlKZnke)R)aX<Q)=!Z??u1u_y<U^>t5B1l5D~tv^HxXVyd{jkH>o*p(|6&?xK@|
zSoF;V8c`S;ljkk*9%L5>*(H-+;uF{DknTUl2lshaZZx3dbig-zpsg&8R^-V!`uAFC
zDwg^bwrr+XU~VitX0Q(G)V4zeIqr*8lehYtuBSz&vl{U}+rlVJ=P&{XEI<S>Pe~dp
zn?xO>Np^zb;wOd6j9VMA+19)=<=W+We<)M?u&&f;{?MZNjFvD^(9TbLqm;B^xNC>7
zOMJ7o4}l}L9|&rT8vV?nqt&sT=&0$h65B}I*)e`JJvX!I{4yQY25or_@IX#aW$qkZ
zkDBJQMO0G%-hg&W97ot-GgU-{W8v3KtLjX59%`{NYJ`D~q@|SE0D}d7?+Z&?GHTy%
z=6w+R#%QEg>wDnP9q-{P7T>cdE7>Cxw(9+4mg>aH1>qz06a!P7cYe4s2Fgywdg|6T
zA@$XP>0k)ej!e>4115jXQmnO<UX~a!pX0bvLCSudZCcX!ZY|GNhpB7M;(YgGbM?|7
z-5n~Nl1$fA6NFq1(q_eTAqUz%^g5#a8#66d!tYx8<(g}U>!2xgu4jY^IWOpv>P{_o
z#*u5@e*2ca=ue$wJ)!-O=M@#PeffmR+C)_DqA}~-OBiPfAh1eun4F}$ram`a@?8X6
zFR@BV7U>jrxHuC5-aEE+JufiR%<cw0Fn;H(KWO);F$&SVSXxO*fK!b3$6;;_l3Y1k
zB_XOzxhI&IH+h2ll|bCc0bf47E#7ny+hVfbY+vL>;wV7maCpuk#>#AJhOi2rgC(*O
z6MGiELK9&~W$DocnR$r6#MR8}e`eF#1l&Q4z@-5(L99@9YE<6S$aw3PER)E(ii2MN
zbY33IG3qa_%frDtGslfzK|k4S$90m<BrJank%rn{Ih>ne7J|PW&OAPfG-mOpf!=m<
z@h$}PIQvusG69t}xJd>f%?h{+ixarGF)}b?ioSD%%anPQhjCK?NU<p|_6;F#d}n?;
zkOb0ik{fLVvT5$}&osZ5r!$Z1M>~~k{^&H(4P2-UHk#;qv_h#`NUw7ewm*M3kM1-t
zK0+0Vm29@LE}CBZwgEq2Cm4TX;myt5ye#GJ;}bf0m{GO3+N(~-lO&@{$f*W+OeCEO
zIpGAZJ1Te_?+birhdb_}LOW6RLJPkvHeu?59oz0xj7teO@&KwFxLsS{1oIab-9N&Y
z^LZ|k8qtKbD0#-IHLMzCJ<K*`7=345I7SW8L@bN<U6ZTxKA5y47y+%vahgn2YTkXg
zHn}=!lOu&%rxaSx`9U%xPquD*^I;OML!CJeEaw<!5iaks;%)<T9*#=lmnq?-*KsMn
z!yLtiiASYW^hPnkvDUfNOL2CF8i^35on8*md2ug>%<C;5+1N=e?3)<aSv|S3Qsi>4
z&V?>8x|(5+<+1tkv0&5MK?e%-OzNcTheq`72IpesmTHn2Ee%Z+K|U0Wg>{T!ZEUQ1
z^eKA+&}Y!|=rZt4`fQ8}&4T6u;io{E@_u0!JoKsJP(ixoP#zOb87P5Thal>zBo+f&
z3ZvHDj|&$Imbtu~U!_!MHh+M%gN$YfKW-e=2bx06AxSn884f19-(5Hz#FM1HSGh7i
zpc$&TOo!v(A{VwB0ozO(AW5yqcGEFpjE%Vi+}-d(w+L&w3+K{6O0P_RuHy~)-?j6V
zClYGz&bfIi3W<HHkZaD5nEi;w%uVS|L)5Vt1Ml(lh|YrF!}Ga_#A=(3vv#$DAigAA
z;QVpP_5Ph}IZ2L#HBvy3dOqrkA^ZaKtZ=^FSB(=%u$<AhQUyYeU)d*9txXyrMrRbZ
z{KCGGVMze<@EZY6K><X4U*%k>giWtbbRa5!dxZ(BhC5II2nd-s$71dT5Tso`R&paV
z+ogmMaMC>A^pOv2ejz{)hZ1smP7!>gYwO1M6PC?*dVC>u?`KbVNP<mx_Dos6?B?m(
zz1?WXP1Au*QQjKi+HK<KPJ_Lr%9y9IEc#!`fB=KpH&Y;41`#M}iU>)F6S+3GzV})3
z-)Z{l>yn;M!C?7cM4AbXKcah=vQ$ZR*bNLBOy{8!Ls#1*S4gt>P;wNxEff*yllbmp
zkv~hLoMe%q`U^qh9U8b@XX?YzxJg~ZUgchzlZJdNyr>d9uKlI%29nY>g>p0DFa`n$
z47L<U2FyQ|4&NR3E_kkr<&#Jx=N(=idfa&1uQ{q0d0eH-u=eZidQDO1e$oDz4<qhg
zoK-CH(gI!t@#TVU#a^D{V#6LULq<H8s02ttb<k||B0X*bOc^`B*zA4O{6fW<+vT!i
zvpgYg83ZlpHgug7;U@{X4==mW-?*S~tpI^BsAn}30;s1D^SYO<0$2)j;b=t8{%}ua
zMGX&zo5HaU&#GDvr?N{2z4_OZ4+dq)8t|!cj~uI9K>0Ez5!w@M1j5@v-a%i6Cy=@>
zh2~vMjaD8Z_FjCmgIkq+xWRy)<+^@^F<zZFrI*GsWEkZ#9~?<LVW3ndxw=`CCGKqD
zol#C_UuIqf?lATp&JPun`t*PXAGOf?Zvo)7GyN4dIFU`+1)AZQmX@aE;?iiV(>8mt
z5FeNq4v4Ls57UPQvx>uSGVT7<T~?YtdKyY(yj3f<TABf<FHEA6R>95}t3Uv4AV=Mv
zMO?>1>4v^VM-=L}Rd#T?RJDb9R-^^`$mIYfG>J2XUjPQ}1~}P=Ps06nSIB0~-E}Y0
zCcZ}aEoU2lXd%V7^z(kA8mIdpXAZ$|xv=H%!$uCnwCK9sYRvG<tTehcbk5;MgXL3E
zlZev{S-k);{MTj1UwcL7l6ZKZ>PkO(#3|g`@6L)#KPmfWoHF`n-V))c`*RAlDf;Me
ztW|v`ZDCmzw(AJ=Z^o+4t3J)&xAR0Vi6L6eDH!SXC^Fi7%j0->S;-g{{vM8btQgmU
z$23p2DSfOU@$>In_Gxyv=aCkBkv1*e5s_3RxwM>8rEE=yevH`KM(_Lb<pTll@_}gm
zoolXWBX-|TG!_RrE;}oeH54JAE*7B`j*(7+NH{*JLN*2aZA{oh?A|rtkGsyF<po)P
zR?X3)FN|(N#-iz#wMpHm+NtV(M{@YXlk)gfPD<~~XIpVp^}+<Tr^Y~5G9aT|`@7$E
zp7BOCA7|`JuXq|*wn1g&g~2*uB#WGgQBG}P<zoriQ1X)9#U}gOeH~s5(b2yl;QCLJ
zVi=7J$Pxx_37%Ws!GYmoR)$8Xy`^4z8#+Doby&BS`S%LF4t8RwA6HhV+heQQCp2`(
z<+VgrP-$nRQpXv_pdRV?N3;a+3*G>&tL;8c-oK{x7%7_5&gQZZ76W$}8`V)&h0AQN
zKO8hIqFF8y+~na+iJ^%&HSGGR{Ejyzg@8>_?RcOwW%M!Q#ma&IJ8w2=-Y6%Rhei`0
z5u|j#e706l=n~O%LdnMWgs=Io4Bdx00D_O75^w7pTgD-F2YM}>Zv}B~&fOQidN<nl
z-J%_KLt}Mh>!@>uI40VH+TD#;Cy5X^_q$3~?=cMhj#GeRMf@nn@NJVg9g^60ei^Y-
z-|!jQ(2xZ5zkPQ-Z<cZ_`@|<Ow>&mJ?fnD-axC9Fh_fc#sNHd@a{E!i7W5;JiKs{{
zZ_|Cl1EKPAHuHJHYDM!MK3P+gwJ6YgTl>_Os8aYYlX+svRavX)5&T{nX{0=g;Qe}$
zQ|1qw7*dIfmO<l+O9b4?1YaM}IDKiUh@dafW`5;0oUi{=k0Q&Ju((^j|Eu8ujnR#P
z)s@v+$N&g&?vyu)eN6~F>;fI^)g+Bs-AKP4_OK{hKR8mWH?R@bYyo$>B{?%v;G%sC
zw!L6U$SU<F@|56g-qgi9-7z)R_KgT>7R1Vv8xaq2-jm{zJ|C1`c>~=*kUh>wB7@bY
zu2QD)Fr|x3wX7S?8++n3Pxr3RVF&r~rpJSRg#h_@@PfozMwO_nQ?pD{)r=qbEgR8@
zo?O+eFUELLUSS@$$|0SaxKTMbl7!uZhY@Xw7HV2)lGWrbAN&@w3xWJ_?XE~+`fGvg
zsAUY2EKq@+l|QPCe<HI*4wj^$?$=0tMkmo5+|OTJy)1YYIP^9CRJa!XWg7R{a9@g|
zrySZI3~lH&AL+Z%kwnW8bHS(^_beF=Tpr=a`i)``?cDK(C54c}$5}U2292=J2$l?&
zmbMgzm*C~E29FGwPWO*Wi8jIo=}C-*5QD}I?FYXUQ(`4TSrNMG($q);q8uf-wMbFU
zR>cbONMQEn_3WCeprc<wrA?@1+xW+$pEx|l*Y!j^k&UrK9<X+p=zo_;vTu)eU98kE
zdM}wX-!rMFq78X@O}BsgUVOkwbXr^QQis~_)4=EMehf?f+dt(v9KEP;RQpKr(+jq;
z_8u%7VadJy9Mm9USS;fe?n8(w1Y9@n(B(}NK0d0;^(4$L$hfH&WH_x>1+Eq`#X@?s
zZ&Mh<FwQ(7Y&gFk6h~G&jBmBA1iBmW&_>Ymh6AFz_haGrw^f3q@<fF@VOF--%O=Cg
z;jN_VCrSs|34WXB(@MUJK_m*<LVBFES12fN?C4&hqbf9kjoDG5Lrp9Xq()PVU+b5a
zmeG0Dy!EdsS42`wgMM9K)%pY`WZc;c{|523yUvUKlg#5!XLm$q-XnWs7^0UQy&1QT
z;G5*Rq^rzQ0m>zWff$*7!0gXc;}`{Zgtrg8O~ix!=(|(q5RS&t9Dm?Z(UE9cUsm8&
zrE|aC7`)^Tf9xdoP-N79$4pp)I`*^Q$b_K-F^%!qtpyWQuVw>#a`BNLaN3aNR!LnK
zObkCKB<=G61Q6$4=Cylt#~lp2iaV9wX|8Yg-DC7n2V^;#-Tgq5S+;4F>jDr$10etQ
zv{?YtM!lK^H&YB%p&Z<fQrd9pcN8(LS8b|(qxAwD9vj5t3)nmzV>M^a-ehH4q^6De
z-qfnI4Bq<#u&K~&nS34MKaLV`0leO<*%gx>M8#Te2gOV}5OCvvwvG~7{g%os(d3EN
z9luD(%9cL-V&~-!GJcfol1=y?_IOf8W{Cm8ha2NO00OrAFz#+|<VPf|r4sg^KE+N7
zsgpPL%uzv|C?e7|ZBB%Rb(g$4aPeXq`E6`uU}5$cH^7{Gz3f$T&e)QlDYZ0i*4k)5
zo5+2-m#>wUqbn<u*RylN-S3?q)XOgyQQl=P)9Q6Y1A!!lY|HMLXKmu7^WMm$e}TL0
zse3p?t5Dp`Sl8>@YHW37nNclXF`H#^nI0cMi5#FVm?*og|MI}i#*!M=h{|hVU>WYj
z@lXN%Yg@fs37WUE%5z8EFKu@?qQ&3~N<=+8Q2sAy_>%42Q&R=}g~;$zGp&iY;XF|W
z@^d0lkK>kBt&E>w*?=-fC9xDFKmkqbH1O|b0TV@4T7h$$UijnZE#&Z-n}^wP5oLV~
zU|`oIK8(h~m3XbmckaTQ5+Kc^j3jg4lRqmr+ZJT6^a333+bzWJml}~7KTHWoj2dMA
z{r?5`G6?4Id#yLsH^UEUc?`=u$05CmebJ8{cGZ!PF7RJzvWFzHeBBxb>M=?-8(;rl
zZ{s5ou3y@#)Zvy0Xq45^$S<a05fgku5oRVtMngYl0IJtb+_vMEo!v9wOXq`ezr*ZU
z;@z@CbQM(1H>v&+?^|iM`??+m8$9VNoyf<3$DIpf`)7ZR+0%b&)j!z#YhIf#8gTKg
z{6ivc5xzfy^_#7M0%_TW!NCRfTY>rOKO}<ycB^b^ZU0xWgaTs<X?9l(EExaQ^1gpd
z%U__kQ3I4e*=MQ1{XdCyh41BXK(jIIm^1xfocIfu3lYF-Cx_Tcq<^*CN*rMG=;N1o
zzmV&mh`w#d#Xb1E=2k}sI2p*eZ?{z|XT)xg!5<LsWQ2@=^(@H@tl#T}Gt8ftf5rNL
z@r#9&+tH@whQGB-HBfAA+8T3>+x^dXvWAJiq-)=f4GzkB$D;(T&h`j^<O|x`eZQVO
zVE~#ZPcFSGZdF9?s+hAiXR3}`3B-&*qsxRj)iqMMsemWBML?~$Q`brY0G6{kDuNn4
zt*yPaCHWUq?yA*)LCPJ&h5c_{{I6{~Xt*71I70E>TL2BTOj+Hf*9f_)Pw#|E*}4UO
zz5S~IF1E&w9vLgxbWyV?pC~=Tx(1ky99B+t+w!LT$n|f)k;HbQGaHw8SCBi2@XwqI
zjOzypF$y#~#L8{_KDONcRjVqu8bz7V)CuL{N<j}S<qNw<kFOr*)4<&>sutCNqR(F5
zDw7V%Zbz$*CwKQ77J+tH7>?-c%A@2O7;q(Av22NApw0jIay}nW+PAq8F<qnqFybLE
zxyY|$)V@|&C!RQ_oDg!ev)iA3uT+0g_@!1<X2lVP>j1dKt>PCCxj)8wvVCG<HiUf5
z)i=?~eG+@N=48^@Mvai2KOhg$;&(L|9@@74`X`N1raRq~w=y!?tGkZCl%%+feJMba
zo%bp%M}7gx{$%#x?Pm|@TF-aU*3erqW+7HT9awg>K;x|FCZm12S-jNqYpunr`n3a$
zuWjF#TK>nwf+YB<v)cQUc5k^~NVvc_^h70Ie`y)rUH@~#2)Xkh%{A?BLY%XFebipT
z(JT%8wp@MhZ!OIJ<egl=36haB`0-;d!QUR#4Ok-yyuV&!<Jg<9$W^qQvcLNZxM*+p
z-t6x6hpZIpztoMLAR#u1a)67y;(c$@@mIeA#9imMam=5Zw#*<w!+qjsx*hO)Z3WHj
zp8eDt?Kv|gfm7kzyg-03?!&Nje43qhp-K8vS@}eiV*QFqTt`7yuL)IzISf>1c{LX%
zMgiomP6htgYd*U<GER9@b9M23ml4hel>XbrIg)GiRZzZ@3mJ<}*QXcLObKX;o(?dU
zyPYN4Z0{bqR7}4=civkztes&!8d1A4IG}Ks>u1`S?t+HL8GTN<J4IT|ZhsH3wRP-m
z-L1Yyc(+sWuapf{26!;$;OB?G{&=gR2yF(SDmtEf#&Wy1{R?nyIe<0ka<F9mcdNAk
zP?XQj9^wBh(BCTnBClIh@h?{w0-(^jyhKL1fiwS00sn%Z=_znzTRg4$?^YKn0ifXU
zKKAkX&4~XSnK9VF*{8B;-oIO|Ob37xNc>Px@SkA*1=qiwU-z!7cr)jP;=fx>hXa7p
zjOv?mW9`k~`G8Lsa1-AT$7TP!)i_t>BoN=mzJG#w0}+6%7Hmw-<3E7AOCku7k)dBi
z9S<N)u-$Gk-JNg&rvuU-XO$1$sNKEDtEk(=0NhQr@crMV2srRff|x00N`jcm&q#d!
zzNi03F+N84@}tg;N^#RD!zckKbyG6cexyc7IQ!M{y(ke@<8RZ%diRX`8Ib!#dUbPM
z!8Ik4v)pQ;r_lmxLrmNe^O59Pj73oZ;W@_p?SHny3Y>z7?*=?}^+Wzo(D$2&|G^H)
z7nsA^N@wNdhNV5-{c%;gvnv8XnYFyGIgI7$yTNc)FM;;raMp=niT@#H`s(!!SaV`Z
zh)z6Z3!;0unHz>MLWZZ~m`g6d^#85^$>U(L0(UfFbuLMh8`JeyP$etE8nvT3_D)SZ
zW{|wQ+7J7m{oHSe4K9s&^5lQ;exEbU;dE6EoD9Dk2`ulu>y-<kd@MrW8O_=E+aA&0
zCiy@iE{m5{7HR_)`+k!HFw%#{q2<%>FnihNZk%kFi~E<qMg#wktiS$`uNOh65sPlG
z-?CwBhM7NL-Bsix+BKT}mtl<QmcFCH-1hA|Ffr%_M++M@_7XE#0B5~O;<UXs71}Uf
zPb}8mpR3-6c=vc*z)y&G^-8D(bX&sLd7<l-Oyp<-i(K$Ex-GHUw?mDFTX6JqgWEx3
z1}K*3KT2e`{dl#RP(hf@=#po3X^Va2+L^ba+nxe#;Ar_oyOql!d7~N|nEvFI`r9bf
zkfH(oz69&4*Qc2OVYk017<aEbTR*6uoOV6n58&C$toRif0uHgCe$$WeyxZdcs~0SD
zyD#TpRrbK8_X3H8W#*7rf?p}2HrBgHvj0<Jj$rM2YmbunsYC&Rix^i$`URN#L-=O|
z!f*xpf2{mJcg%;12)M7P5M9j?z&vfIJfpZqvk?;_Rod**U$*tPt-7zH%{rg0+dYiI
zhvg5D6t@Jj%UAr&K~j&UNm!ryKU^U^G$1xWCedrB1SGQE^WCFsuqcp#j?tnA6aNJA
z4@d6T-r1Tux@2w}0$MHKOT?~SGeF<nyDoTZcW|U`b^OrU+<+(b?y!Rx_(RD+ME4ti
zi=Hw`cy%M&?d%J`UwQS(Q}VXaJHRJD#v)zo?>?8ivbh^&sg$=r8gsQv$tQZPtiXe<
zmhZ4`-`8LGI288EO_(KpcK<fP_%AxNJp_bfZ0GSK*xLd5%Xi<qUzu_{c;vhP^_IWu
z&OcNy5(JF14HulqO$CGhv<Au~0P`Uc!3f^=zW*|200RLAl(5(R^<TFC&zJszo)70L
zx8~tnkM>`0xosaLoIsM|{}<X}m7hKMA13Y(4=%*>J;91PJ#|2MK#)d?6)yIYEG)h%
zaGJ0{lyPgvA+3H1v#t)v-?$w+b;0#j$#`dDrp@a8!RYO<15qmMlj}^=sC%*gc)S~|
z44s=Su}P={O4?P{Q8PtDA|Mdq4GWxzn#k4WykY|XFU;}rzz+=%i_6PPY5nvMhX~Nn
zA}<>uy6_xtE!+dYlX&(ha=g+3C3C34O`^u<jRYt#Ju|Z)qQ(LrCgCdYWY5`iYd6bp
z8oTpQypoe!)0Hex0pAg>E0S%%oZ3D&ra?`TxkrN^`)ttplRU$#*9PvN#DuOo{N`w_
z-1)9damh?T)Nk7MCBUGNBFL|YjsIAqC^5*k+s9q>xF;$$ThI?7V0qOvagJ{{H-$Y=
zU=WyC{JP<XJcOPW1L}nUAz+u~f^|`o>{S@L)q+huoz+e4)1l5mUe=&Zr)-{byz5{I
zgY)gt%jBe<?bjqgX0u{8Ky>CvM;FnBkBR25`t$Dnc)rX^k<}UXr=LQ&8LhXRNiaV2
z%?b%Qxx*lQBqkSbmHIR=ujhw#EZgKk>_oyULmzqiBU|{xH>;`Yf=;PpoC{EZ47B`y
zFpo%ex{eo(?9k}VF_C*7;^+V-kfg+ZZencm=Pm#NEVUm-GbcOA0Z4M0+@Oq--XMi3
z^1kyV2OF80eUCpZ$%5TyFe;z|2K}(i7vJb8#pr(}{JoD~_L;t$kkv)i=?`dBC8g==
znq#UZn6d|TYl8YmFBfB}DSdQb9-_aUGlZZz$?t18fsIjgC2mqgU)_v(@f!+ht$Vi!
z*YX4RReUuqfwRIzE(r0qtzel|zmJ=8T*cS~b23RN&0XEJwn&zdKRcyH8(uKGIHi|`
zZlO}9(riLjL?wV*{V8(l?vK=#zgZg*H*2Iy$kM7gMkRhqRL3xqlCC**hViya-Bl$I
z2g(m8zGKeZ#Hr;a)KMnB=<NDw)^`gu#hIOEcPxe;_sQcwc=($%Sb4g2i@a9Ewc<W_
z`gFsGXeE^#Mps=;ZYhyp++4|M1r-s7n;xi1%J5$K<3-BeCu1$e_%@oY@d-lD5Dxhg
zGW4wZK8oy+xR72fyvaE2+&Dgh=?qQSJ3;MGEtLwdBl!SoG_;w_FpSS^Qci-t5_Y!U
zqXR@K0UHj|6+-)}h_|R^V5sH6!9m{BI6QaMohnf2#l`1}F_q5^iNb}o`I7v=e0jgM
z{gP)FzYBBt0{GHM*Q2)kX6%HaKS<pLQ89KLnpNJ?O>25Kt0<6#ohEzHW;bC<tB-5v
zdwIs5G7Q+98B|?ZDWh`sf<-@Kh2#TM@v?#s6D)*nZqY&RJxKQtP-}Crupe;DAn4hM
z|2pDPTm2Y;uI~$&4I7gF|Fq=3Cc!`v^<t&aR7p%T80CWHP>vcQCo-NFuS%o8he1!&
zEpyy7Z(uY>vn_fyuoYBlKv)p?hg<(nC4pp8EDOC8?M^h?R3zz@)a#?IQNj%JzOTS%
z+5GiyHwwSs=`2rz5yu{_!|kv|{R`@v7Z`byihz>b|3uyG{6A8vzg_@bGKP;!vbmiV
z-`5Sp6=?Dxy$a__Ly0xsA^?B0Z_9>&&HjJ!z%BUw&%2U^-92zZ{&Qf5w(RGYhdEkM
z_BM_52~m<vHKi~F;aHf9a3u|H!lw}<D-R0+_JAywlAabvMl;1kuk()M_VmiH-0v-B
zW&Db*r{1vqR^IMW_wj8uB4*Df(oF^*`oH<%SJuY}+X&|ebxVN|vnp}H)<#Kpi#eBD
z<a?DlJ(=63Co7dF!{6UV#;2Xu{L+-60hw1!{m2E+aGR(B8Ow+8+gbb%xhQ;ox52R?
zc^R(i%phf_JIuw&mMME;m0CmScLQCOcxe*36i$<#%bmkL+$dFnANe6+)@(RXu#bme
zJVBQDAc6HuG9TBy2jbsTFq@grzCZ|F-sJ~{Al+D`2JprTO!i1M^`}qrK-oFXitps*
zPx=W8EW{U%U9`?T`5*DwtDENu*j9-XtOeP(-NL5&kbsH2&R1^MR?IW9Yd6b6gaiXw
z{K9eK{fIF%+)OIr_!PGi&6~w5&16H@;K>aAG{g<L@|?V;*h3_fUqI!BcsDOJm04#!
z!lQ`kYM3m>Ly0t}FD&a5G(@g^h`rUS-74x3ax6q~;bxN~RW70sEL_}`Xa<zjiIwz}
z6#j4DzEM0fGV?{WVheV^2Uocf8;hTnpKsfGehJfUi6v%Wki`ks>rHH8zpI9ko|g6|
zB%@~m-CrmT3|9D-ZP*0PMaE}&1%15`@ql%vhLb}ie<XVe7nH)Frlno7rXX+NCKNcb
zW<G*^AErBarbSUrZR51~P*~VoR@a0+a>0C})M#B-`zT7)sQT=1W6*Wddy;p1ecEX9
z;`9Jy@)Nwax;ps?0!1rvq@kfP7zjqZ(Hq@;{cbEk-l99YZp~z4V#Ha!=X~^>s?2_I
zsg4r6>WC#PL{Z5PG7Vc7xW#5)Gb=a3mbE6?49Or~?lOq?<Qqmxt8Gj;Xw-50?FfWQ
z!OX3`9ot*Z#`u<l?FuEsvpbZe;d-AoWI9RuNv4j}dhJDLBAiC}-KC@_)8VPZMTp!M
zDw-&0_i-7++iGhd{iyRuy4}$po~U7MiG%~&!+{tsP(p;-<Ine{PtxoH0<4jxJki+-
z)HrQBpcjy6{5PZX0$6fiCBp`J4`1VP<g7>%WGm!GMdQ}dnbwwec_a8dBp`!cXw29i
z9Sy5=F0N!GMWGO|bpTCe?}yiwEq=!6m22D1T)a>&)(ukfd)GDEZ?Q^~(KeKoWnjs9
zSkFrwKBDd{--uqC7sBX#63C?01V8q**m%->Nv&wE5yj+V1hq_fAY*NtH;uZ3c_G)~
z@jQdubFC&b_QO%_g@p*2;Xc87HO@==#}7Lp(KwyqicuYBTdv8Z0c0`H=O1k>HP}T&
zDk4QsI;<Jz<q{U$X^qGH+1525{Hgy7s<eTjGcVBhTx7M9U1X@}y}zYFHJCw6zm#Ab
zjP$07KxGzEb2Qx17{W!il@G0#PjXmwI%yM?CYMAPh1}G_Pm&M2j{0#uByb(|o6Gs0
zMyIqh@5W$s;!KJoLm#i!2bse9<~H<3)xnV&)j9R+eotf5f7kCh2*ZISu)#?%B`>ig
z^T{PcI@%-5lcK;f-wx@bXb&|+tXx7*)6VqE{ot2M)f_7PRDrF!X~*w!tKQGOE>3^t
zi8W11P9dvEK7DyF&1TTSe@GEzcc!}otk%oWcX4t&+>FRbd`qEHW@7DJ1GMfG$*l1}
zy$x0ENL*yt<{LDihqm^|DZXk-^|}d<m~|z|#CGO<P<AIRdpkQmWxqczm}4^QOBM=+
z$fJl>?+<`XDr+w<&N_XGou0gyKAvy-h;4%sw*djh%0y8oIdyc0YTga#Ro_vzJyFAU
zT<~7f2*rLnZBxvpv<!(z;+g?w@ASL1U_u>a5B1TSbUo}G7bQlv5UAtPh$|gSNm`Sj
zPSIQ!UBw)nB7NOLuXC7X#h8KnFGMG4H<t3Ma42-w3Z!ke#^@=Vy@7jI+0L4l`@bws
zJ8nfRCAq$~l0dQFX^z%`G$B|lgkfUNE7e^(+n@VWFC3L!^mo^47ON_SHul_nHANrp
z{jXBovL?`t-i5fC8h)N3l1AgQ8G23|g|?Aet8}}Ai=qXlKmDrT{`9L3iF*qAR}Ql=
zRBRiupLGqYMm)_s@fi&#i`KA6yy^kV+hjbNWURZm&`>RaL_yEovmIyWT-$q1KG=f;
zhx1yW-?tD3;>|oRnwH-!<T!{z4d@6Xuv2a4o)WB2=By`(P$ca-tdtQh`gmj+wU`zC
zz}-&%VnX-sDsZLiK>G18Q#nspN8H?kWvrYqat3u6QhmqwoI9iFce|K-+jcB<J70@R
zS!axfV?&S6oH<mQ3{8Fp9PSUw#&vW!@BWN4Z5Kr4Cc{<FEY`7lZ$@v8pIyds_920F
zo)~J!M<tUOuVm8{wR{ZxNny&MvI#?AY(?&P_@?~%TRV!19E~z{y-1~{sAu9yJCN<h
z)=mUV={`IjoA#o`r)kyIsxBv;af>5$wPW4En%a(Y-ngu2yQy(8jZxg}J53cmFKhA_
z42yo{$Nui}@YNy!#fX=?6LF@YhPC^IxL*s#Ear85YwK_{hP6A=$lpV}8X>-ubYTAk
z8zUtW$Vl1r%G+^b8Rf#+rUax-sjOnR=5uBS=`2f5CqWV!z`a>m@mvI7`zmdp`M5|@
zNWBXfg3tqL^)2`#5O(gk_3-qq3AbYO@38))3~w3v{(Z~{Wh+;W+IIQ`7civv-naFU
z7O8sPm0?iLB=tCIiYi=|K*-MWM&5Ltb%}A=XJKFO6Ra9g>!<)*wbVL)+6C@VY~v-i
z6Jw}ejxa$$Rk3JlAm9W?uf0dl<z?hlS5~I)jLJnr;#Dft;a;z`-594_vlf3CS2LgN
z`B0()>`kism{SnEHMx0iH2jeg6xFttUqdP?6-R6M!>W)Wu*l(nGuC{k{h1oa;;xL^
zPu<?-o@e46$v1~wBp_HL`&kv2oqH8aTd3`&iq5pp37U%;>BK)CXVQpLfjNtw8!gXL
zb$3dDRMno1hQfasWg^}=Jx18M7zo!tv*XT(|HqO4d!c%d`&G7EQR$%QN=oJ)UZ`rt
z*{0*d6zkTiByvCAvr3N8*x0xBM^mny&PP*>(34zKo4p(?(y<bQVy@j~ZKj<z1SX^7
zXD4U8o0xG1jRKF2KLNRhF|c(YQqX5_`l{zP>q=t<nw~0U=8Zn6I^D6Zz+K2nqgs>Y
zg-m;++N>polIqI40RYkx9WSZi9^1JQ^cLxKmV|sL)B#CzZoG+MvbN~hr7;c)n*Wnp
z`Rzqt|G5xR-$&}X2pVyrG6#V)96>HezEzX8-j5%0zL$i2HXgNTQ~)L|Q-$#Clps%=
zhGMPBR?AUInH{jjHGl1%Sa7X7vM!sMER)|lS1~!+NSaImF>;9p#}1UUm^z&4JXPrQ
zn-jF6nNL~q9Tg?XxMIZpdF}1%kG~fok)28Iv>Xa7D-nxcTRxY37=vfrtKKYdk2d+~
zCn2k0Z{yu#OqsUK*W^=@h@7??R?IY`@E#~AYWRMzScl72&b?e0IjR;r&71gYtIupK
z`1$uoctM}?)ob43Fci5;Zcsm+QBN<Hd?S8tC-`(eKPM*=B@WznNXV(0FX)3DX29{K
z$wZCleN|6dp&ax%ShpLWeI>2J{}Q6THIgscLdYZ?oWu&6?q5@UvmO;~g%hv>j1H<f
z{bn>%D+uFK*Rx+(*A>ey*U`}d$=JOy@5-l+v_4s?l2^B*>z~$b=hz~;)Ol{-+G%Rl
z9#pki!K-Q3y~JSYx)7P!7i7R8*xVRGZza!rx<bEcJ=8X5vE?ySvC*33G)zA=XEVOD
z0iM7u8C#8g%6N4tKt{I>3m?f<InTntpjOT`R`<jC%(T7V<c>dS_Wc-He|fKYB0n=)
zWZ*2Co|ZP8ETJfmPfu~Q*13w#{scQKH!C8D;6;V&EHWM&&`1eeWIos;jXK<bOvw>6
zU0ZXukYGF9253n-T=i{n8amo_tE3&wvx7BiZ*O6aHNo^Z^75;@M5rG|zc1;IKA}Sv
zgI?ZarND^VHJ*TEu-NBjX7<fCPZWuIVk&%<PN1buwDZVO+5t156iv}@?HmStl1=G!
zLY5ENXR=rjW9Of!J8X&DodmQu*jQ!pkawCv#VQI(WmEfj8=dLm^GfAp>+6+_qU)Bq
z|KemID49kC(qA1<^J~WzE`t+!nt#A*6-$lwq~oi;Ji;V49ymN6Yi(oKbAGewy1mm>
zD4=HXLjCZ%PZxd0`0(JWPj~gUK3!D!b;nJ@F?(n>)4R<%rsa5TiauXYtaP1pCWC=6
zruW<L?6<%BHeZ?-zZbWXYM5*E?W9XENC8}b5P!fu>zKfPZIQJBT+kE~u50y$43l;X
zGW&j?VCA_LD%b2C_TyR4bSK9OI%l1%2YjjSsBt>l=f^uH{=wZUVDO_f_y?YV0~u6g
zu}G`-fMUIc*>)4b;&=}8eL;^TdY^0L#zdLSw5w}EHT1i0_XC=oZl#(!z{y2_f#hAP
zQDu^_mi^`)MFcf5O-5G|<Jn4y=gkKbrvZGf3EkE}r>_Z{YSofnk}Lb-aDMr@Kp~}j
zNDPmBa9sa6G3lpOVt2Kp#q5QrNp%A2HC#^*ec$Y8T}(uO0tCK&m36Bbl~R2q@2s94
zg-L(b4JZ2GrPevA|Dpf}b&)$C(bXLTUwqqhuTukkaxf^4h7_`xk+Z?h7~itCvN{(e
z>u{`I8f>a6gN?j<9_<7Zlqu?Ee&%)rKXzPy?qAt6-M2m0`2H{_hU1H%clE(A6AEi&
z=gj4y?a96#!<JFB|DMxn$X9>NkkAOx?ufD(^%VhPn}W2GH$5C{^OZiS8i(9R0;$fY
zEhcIK^PZ@@k=AQ7GLMVKp7<c+<?x4?Z%&s*n~v!49Cl0q68XIKae{tyb$^|!YYej<
zSd!PFlIyIi*k90>$RU%XXW&_Lnh+tuwebBd-LVgQR|V(j6nZHFQPAt=<SaBMAPNzE
zfV|Aed)^KOWtkc;xz?V^qo@n%m57;^y<ds5Sn_Zj%*H&!uH4`5t~-&_;<cX#t^gqy
z3qx+OnZ80@m1qab?_&u|m@q#-=Ys<+km+NGQO(_SMrZZBs!bJ>VZWVuR6s*iDIr=p
z)Ua;VPEeAC*uCFh?%(xhR3+Cq>9CenR;CHzKK%UHenDvL3)F7q!wT;^x5wGcs`0MV
z?7K^KbY{$;c&h<_Cevy0T@wQBnR0`ObCn68l054Dj4SlQ+9%L+G8B+AHNDSvf^=e<
zo#E9dKbb1T^wtX&UkwDp%D~`{7q|j5x?z%fyN$xsdvCr8%31AQ$mQJ#*vI24t7~6?
zN~Rx|8N8>`DpHd>rY=W5Hl<(CVJaAaMD0it$^3OyJP9Vu2E5FEshcqL@?yaY2!W33
zpxnweAJXF;R<oFFCY8pj4J^yJ!Y(gPS7Zr%+l6s8Nb2e~LKq~?N7VZJ1f9f#@>G}I
z?j1(=Ivp|($Y8EJvhJJp3*!cKd-cB!B}+xMzc{ttovu9C91sR#`u4EnmtQ4ga@I(L
zBY$Ko8YVgFoCIJO2}dK~G3#1Yi$hACCfMxHcc@wHcHXA;uo?M;EG4?~e<iYxluY1Y
zUwp)QGkaM7SIhA0Q49jP3`t2%`%Q036N87@bc5*u6LOa}Ta6nAA_VtOKLQ!D|C}-2
zy<bW4YSx+=PUuhplvX6taDXzdEab81NNEIeUQUCV4i~uixzQkOu{Bgw<YIBULL7!E
z6HhMa`|wjkgTPpkJ*aF-EXIe1;dlXd_bPOqs<LUn3O1QNgwj&JpI0duQNaUqT33re
z>A=ayYjV*589fjE+@aNVjIEk7>ETu&5Xo-Jybx^%BBn_LN1H92q@Ll|3JO$J)>Q&^
z>dL1dB1pV*yN~C5f_6tSp%QfTOuAi<Xc%X{7CR2LeUr~#nDZqL10vTb_sWOlw^S~q
z!Y~IX(jJyxUZConOl*~p>GHZBEJA|1&)>?c(aUh1<QQ`OoV};B8|GUUXwo^n`-@?m
zuVM=g`+-U7>y)e-8vMt{Pau0@Q*fN`w~C*p*E)sRIjd*cJLdu=(@iRGDctY*0*SI2
z@9c%?EElyF?d{p`o7vPNrr$rB&$qPMFU8X%;Y`w>9m?BG=m5+PGWm$l`fbYDWMGVK
z7;uw0Z8xo!hJO5LPxTIbKUTD8G6O@jHd3<{#Zg!<YAh8*r(7+>P(6Fz#G2y*%-StF
zPLI}EoOn9LyYGWa%#c)j&SbyP-}?SjkM3)Abp}urUH=$EaFm>MO#VrHEq0&yyTTh!
z<@k_bqSdW}N0|r=FKeWUT=r?$*FMUrb<RwlE;S7hCsnLRbTi~ms7AlI+@9!`f00$S
zZmD4)D+&v9@&phm;w0K5uU}@A?*$T8Fad)Liri4{1eH8!1Y|ZZy@Pd{wpo))mQsKv
z^-|2~Y-?l*xXHF_GK~|7^FGYogagQz-CZpdqMWx=^u@1PDqeBY(%mVvnR+|+4P-Ie
zg{43zwgdJdeM@U?UCN>2{Qx+XXx=)iaxIZpP^fs;@)V?juF~7zVctsk)KVsXP4~-a
z9%%q7&m;+fRSR%+bsb%6*GrK@0+!5l(5stGeiE9#K76~eYo$uw<$AM;?wE3`bTmx_
zRU2^K#6pY4xq{C_yquJP(%t$-s0ESlM?!k)#An4*Y&SP4=fPtUalCmDZ&$v)=a}vS
z@*c_=OU<?yYFaxIn;Qj;{<=2hGs)<8eaf^?)NuQ{JaE()DHQ<4Kt`QtmAc*+L&CN;
zK#KFB&8qNO)9#{a=hvXPsdQuT#}q-|w&apf!IzJtIyxkb%ftoni<HY;YK+E;<VAuX
z19yKAyfBH3X2IG$T{Ep5d-4DUKeIDBek*F{v+b@pK8`YF6>#fBhw~Y5fg+^^WU_(@
zu>TBRn%^;QK5}rDs*h^C=!KpZ(y8xOb`6>}&CR2l7ne_O*Ud!ss$N*r%+$)|o%I|b
zM1_ZIRqULag82E@KqpkveL-ur7S_8affKyY#aw!#`8zNtR~NY?B~l$-=Y#oB5ss}^
zg57b6sND-2(?0CrXU%|Bc)L(Rg*^hVEzaj;Ip$yW3m-cyuR7;X6pXWZ;bXkm+&ftS
zaw}2TrgijHv@C6@P_quJv6?CYg<Mtos-yMcSDtN`l@x8*NoP5LfpzG3Wn=_WS0^ij
zt~?wAeAT!bNIH-YqvqK$M|j}5+Yzp~a4<4$rWpu<kj1cDetrC6Pw7Hb&Hb^<u^s#7
zD4v!kk_y9uCn`^9-Ngp6MTz>CoE*AEj`hFU=ie@f;{DC?HDZ4iwZacUNU}=JK)&z)
zWA81)s@l3gP(ctBl@4h^x|ME}ZcsV}kp^kllpBx4rX{7Nr8lw34XA{4cPSwa(hYY3
zdOUjIecyY(-v0-mXUntJTyu_|W6a+i+P^`cpEjvjTIK5;c*I(o0LOJtk4;4-aPFK3
z{KEK1Rj)nhVjt)>2l0tM+!6^bcuoVkQ8zU^;Ky75b2bw-c9di`YzPSLkrz<mv>s+$
zttz74Qxu1BSq$bBJ+*iPq=Rn>>>BNU{%R$48%-z`A8m-yWd_$}rF?m%^J4~H)BB9p
zD7jd!mZA;;mX=7Cfzdz-z--(-v1#iwDAH1gvevf_Juu2<UVXZwt^-&_fs)SG^g%q|
z=0y1hqOC9VJ=4(I^tOr|Tg|szPLXa)#bMgZz^>hPK#we>X4KA8lOdGsXcKSmqo+1=
zMqA?I^E{L8q_>dWH{@-^$v41)bhR{O?(0yyp7X)&MQT^E6Fs&>{pM`%0#O^2Hl|K>
zvQK0|C;A$2-|3{OyVM)N88_u<6xnn7>J%n28`%uL^N}LQIQlNsC1GJ$`x4>4$DuQi
z>9~|t?#CksDSu}kVPhL^&8rBHd~Wr4Y==EU?|I8_V6;9hwL%8mI#UuuD<tcElY~%c
zS{JUHALqWUVdd=;rRO2Gu&~H(BPeNXM(fbX<Zk7TPI@0GMniz~m>+YrK1w6WebsNo
zo>fm&RhsVl1cY5hADoEq87+Ov3zG_?mUb@~-+FNDOQ~Jtez4sfDj{z_wV>9okggyk
zeOquh>03nzlyJY`4to`En74ogk?SMc)UUoyX`Uwt_wD*+?_}l|v9q%|G*5>MHjp_+
z$GJ4z!7?)m>4VmO7T2Mk&P++iN|g*XK>52q^sA^to`{^Fz3JAxd?Ctv6FrpBI4_?2
zE9;%knnbPR6oVboy{`&Zk$kHaYRvp6Q?|`bcB#dwbz8<V8x1n+SRYxKt5|f$78chw
zIvg68M@9AaJ+hC!;vj?Fu)g!}4^GJmzLPvAliPgmfV){-QigqCdJ<c(U&kh}H|;`G
z$=odMztI%vdREs3+6**j&9Qj6ADwV5yb5gL-NbMRb|e~BqwJGWYiZXzdi+tP@BXAG
zb5yeFl*2Xum+Thf%n=GK5dxmB%2luSn|#T18IE5`l&}58uq`?Bm_^#;;P!Q58{hEO
zx9=Y4?v6p-J!=%q5?s65UMk%;hJ`0Mc*{Ng60{88qkgYfB8!+JA=9KWEAs&pTgUxb
z2lJ5A#unD_Iv6L>5r*_Fj(YrvJxRYNe~CV7AD=9&Gl)sqZ7r(qu?6_AVm%9O7K(R@
z>W<69^V|G{6|_uDHe?H0!sKvv;GkT;x7qvbY`YrlJvZdS36EFWMb~bjJW*AxpkmHR
zE-T}PKC53Ds~mIq_%+0v9Z~toliqvW`Zb0_0T`T*Us<|tlF?k{kbP=S;?16^-?-G$
zU}8wbZjxkf=-{Zs09@y!H^>pgz!ccEp9xfZ^mOw(wuBTv5pjBN8W=+J;jTO4jqmeb
z`P_7zRh{p6L<_U<8%tpt3LW<u>N88K%QH?BJqcCQB)k=ns^{;1VNC|4so&T>wt|%+
zesg|ahH76Ax=vEQ??2JM(*T*7<#JQ=zB#T?{k>#WW^a4$Abia`#s1r%c|LG&iStKU
z{O8Wwuc_9HVm)Z^(j}Q|_UqA~4fjgjKMC-Ej7ZRA$y;gZIlVP^<Yd3S>6lpRvUq#I
z@$(n(vS0=c;kEr;#x``pLrJf`@ew&s&w-<~RG4Ib@Jfa_cK66V-j!)_r|rf7oOn0X
zxSz*&$5_5Bt5K)XkY<aZELUbK@<WqEwbTl~t>E92dpuHzMJM*^x@r#R+S?k-O|pC7
zDTOG>NAb=O#=tRl4E<S+HQ+i@c1v&HxzO|prru5k<AQ?xO#>OVo5X{_MTks&G&@gs
zOx__LNV5pWAANXueLF&zvEZrRG96POl+t}reaYOnJt$P$;(GAzebT0_`H^MIC=b$$
zaSzk}e*);qR0E0ZUQkG~8FeF_##qmerR+w@H0xIn>UbKP0_Go2e{IBXPQ)LbMa0E}
zMLUc3P=Tgq?YBypw^5XImVGubKU1Hc-r8g*_yHMm*Aae%TK)7rS#_yT9amBfyUFLI
zr%tc`a$LpIQ%OrFtHK&C7NLl6)xod6e~Z<b$a8CL*FoI-!zWEYh2q)f;-4oj{uB(5
z_?K+GL+xb41Ilb`_ajZ0g?aqb%e(H|))mx2H*dw(PkPyPgxd)oW4b6uA@I!AD($1l
zcik}~190zL{ucW@Izy}N8doar-)<L6;zK4BlACMSoj*Th2-q#)R_C&G)sG`iQe}C%
zZU<XE`A2WkLruy*dA(|s9Mj78d!i!sVyoGN`(tFK`bpkj+1dA!-(6o-Jn|7sDwvVy
zzO1bL3|d?8?$yj2G{LSF%zcNU=^ml`ENSlK7&_8CX>U&9?5B6iT)+PF;+9S(#F&u&
z6HYCP3M7f-dmJ}bGeffQBc;pSe2{_&XsBp9FctXvW6@;!mCf`#bRQB4hIr59%gn+1
zu+|ppcb39^C5+1n?4oyX6Hq36Os+E;`lHhS`+sAwzjLZdk^DU3(TR%?>`lPGujQRP
zVodeqG5)gr&L@Wtz^&@>S0o)HULy*^+hjXrIX_}OMExry<RwNw`3Mp!rdWu3gcxg5
z61kOv`+aiimy?L%iDh@uF=h}bBnNM5T?-MzLcJRzJXjd&PF<vszj|Cc+do>ZqoOkg
z{OjA&$$hW2ld#2CDyKEOYS-P_YrpNnDHBEo8Teqh;2qmjSDGj3%2L5D=Qi6MUKl~A
z$zxFBx7H{i7&tnWVCrbk*w#Rg<La3H?$(s)W_x2=paMvjkax+~#i2wwkNvW)7qwr2
z-lPWK8?=MZngs|qB764-rhk=-Rc?+9k}=gA_Uzi}Za_(VW0}hqNu9qh6|>u=P}oQB
z-Tl2`%k<e{ry~(mx!z)1=+h8`|3`IYz5$`>r>2{`*xgSoW^JYdHMZZg+aB<xcZJDI
zQ9AKEN*PIA+TY`5-b-P~2NG8S8sV2f-vttE-(~KG6}L)uj}z>kgbsIs<^A4)?&;wC
zHbe&Rp59xfsc+*eyFV+%_U!k_%RiE^u+riaU$K{kM4h)2`Lh?qGL|n%K-9Le-g1@(
zj|%u&rIcFOo~(X7A|jotZ0408ty2<e2cZ%fl!?_;a+WnrB)i|x3nOR?RVcb;VOaID
zc%nAW!}j~`-A@&qpAiz?c^@^#asJ}zl>b5H&A)^#jQIH;#HMEyt#V4-wp1C#4kQW@
z|I3}24evJM7zT!9*DWK5vpIrAa6Wq^lE7P*dzN(%H%vQV`LkM}n{x7LT_a?n4f6cV
zmSrbWLs_^^tsWapCzGzUd3?uYVJ?jphskj|{?jvz8(4d1H2cQ{&PU$95%ApeCMVm2
z2x=OdB4NlQE!@&lyfhAM)B|n<Mq;FjyW}CsM9mB5jZ0q9h9FA@SPX}7#<^(|3b+}@
zj6)fW?v9uOH*y`j%zgeeRXXl+{}WL|Pv@L^%-d+`p3)V`umRn~{&?q7_6Izn)cNIt
zX;-mLcfYrpI)CyG7bWc|+5rDfne&OYqXrmh-BEbvc@GBkKeqjJ4U7}7WTk61*obNP
zmAvy7f`TSjc%R4_Re|1kXz2vaUQ`W1m5+`!1D;OLQFfcamK%Hbq3jM*ZKe|^+HudF
zR^=bKI%d4j&XX<z+L=yoJ4t?WDs#t^WD`O)gcgpSOVu9~s<lPx&((_BUK&f>Uf~P5
zJyO-QP1+)&Es`+UPq{SZ3w?nx=M9LZ6>atpp)!%upWc>8bJ(eJjTx_|lyf_gFg-D@
ziKb>ba*P=-VQ3qN2?Rb9!84?Zj}~764MaAFoj0oaQydv^3GUpHUn~+j=6#`me|Z>A
zL*i2Rlz~b4m<shT7h<++|B@2~M5gOEbM+EA`PW70EQh0?Z?DKb7%9zYBC?Wx^yq=<
zTqb4rKFXL-F5hys>E^aX>5~M}Kh}HB3KO5&hP5^JP-Nx0W>hG=E<JE+nO60VKkCv@
zB?}i>b9X%fH>$miFu7Q1YTu)C$u7RfYFjyywk9FSohvl|m;DzAL(ftW37yf%tDMRK
zKhw>!%G9qG^HuDO3uRQYen0zx{PX)J{=zwb6`1xo55tZ1DlBJ5Z(R4H8-6%a5Q3VV
zsbRdWnqne3BaVWEq-=jhCG`Bw+GZs0q6j+gU?I-H=PW3{>DW(-f<*K`d2g9II&v-#
zRSG43NsNn=_A&C@HS(%~r#ui42(x~NmF-yvHJ$Rt*#>kE2hcq^ONam1FR!x{acYXF
zFxS_oJHf&T5<M3@(6DCFGP$)_$RyN&q{3+TjPk6X@3IhUXskFJG+rVHk;2#DZDwJv
zq%e6sWcS&{-6emlKtBY@z(B~0dV6Ig>Nat74)T+mZ$vT<Dm;$V@sfZJ>GFIGHBdnC
z8aP*eF7qOik_+pDg@JY}BOnXnL@8`+Y@q+lV^cE_Z(Dn7`VjJa9TRE{28Elq(9^m?
zLm7pQ*mJc?+nY(-YDX6#!036b*?(Q2oqtmqTtbb3fD0XyfH`};x&r37$6DpRpUm)2
zre56g2#k9jHB}umvGu<`TC0hLAqkBXf2?=$`u<^8|8LNL&@|}*BG8&Ao=*M!WhAaG
zL~?VgVT4t;-U~@)@sUe8b@BSSXs9Xbi<CK@7r)kNjaXf|T8gRV;fk!xI3X{8=F~db
z5R`IBQdBN&=o|TRbtS64Kk6-#^5#r|=~+YH90!D&fZQ$2exZLl#$VDr$AvI4wlm_(
zR0Bt(i?pVpITO3NXrOf%uXr|qth!Jj3K)U6Sp0KY9$ZhIYkAsZa{8S5BI2VVaq~6X
zU!(sam)R^|To3xlURl8atHxE*Cw3-Rk>srRmxOv`zPt&)L=Eg^Yn`@9QkI*%+B!F<
z3ruj%Ch3by$0+sn(uTfp@-KV`x&*IRN<@GzxjB8J=O4R=Q4{X>LI>;ZS(m#D9Cp$&
z;+hUz7y>KWwg)59fk4jeL8tS&rlUW&P5SB>bx!DS)%@*8PT|Yd1vQr|s_#Hxoa~o*
z(&t>u5?)$)ssd3hNxiJ;<T7rwR$wyRh(31*8DQNpf$3Rm#hYcdpCuXZaDG9pxTJ+J
zQUznk3Dj_#1a9~pDPaFgU^(e&4@Y6L8v#K9n@_VjE@_dAx|#oS_1l5dlIpN67>5Bj
zJn@Y1_9FVUXs%mmTKGvS=T86M%)bxbl^3jzb-I~;8YdDv!A}h=SP+MRG)Vu($^6Qv
zcRkN87MeLq@Gi4pb(j+nVh%<B3^C7MCB>8N@|j9+x>(pc$q?;CaMHUD42yXOq!ol`
zEh5gtS6&KMe4wFkXd&)@Zs@!Eg#?KoIpLgHoj!=r)WJcMrH+1|z=2EVdfNSPO~=)3
z)`~BflrUVy3L*?qA-&Rf)*%x-bK|B_%}`)qP&#H8w1p}>_+qiWrcGXL7`!?>#pi?q
zeyCr}y(WKdM>jKSfmWK;y@Rc-vN7$tw=bp_&_S?V7)h~?lXQ|MhVRyb4jAVZ@B02(
zLvV9xa*wW9G@zkxQBjc#(9rkWADZ@ylDBV6Po&L$z<#8f^!*7tF5q>;8|BU=Qx6fk
zsa{0^;Y6gP$V}dDuE3hTY%T@;$m^>uMe=wPu^sUIE!{=EN+3cgmQm(8`x4NAlcJE*
zgRS*=85(JN`!}`tr!@esSN3)-HPsaVheNsK;kna|$O<U?e`xCZ6akDQz|S4?-$%7q
z0}PK7yzjq1xmf^wo|KeW?lPtQMd1ntAbhVJH~a5Tg4BV}qoV_Ib1A!iYTVc<r*y1K
z{`->xjKJr8P$YVOzWQrsfBXf!asxGWF0XX>kH7z$17K1-;PanerI=ni*w2@QodBC<
zKcHp#-=9QB13v%tLl^sH0{^#EBvk_Y!HUIP&j0?T&#B4${>9{fXcMdh?pjgJ_%UXZ
z7GhBmUEA}MUR{lPo0ZELVy>faVnTb%D86-Kir+K$-;$4l1WEFnfBTlvBpHP;_4*}L
z(8omr^e1J!J8DMbtfOxpvRcl=q9;Q%z(V(*C9#K{*${SY--)Zng<t$olV1E&Z~i9C
z=?^sI$;!?5^X02ult~8M=vnH~?YO`w{6D(db|9x0>FpyaZtjtr3I62L?-t5lg}hoV
zSPkh(K)_9>603-<YCzb>$tao5`ye_X8)y^)O^VmEo;mcCOaAupf2^+;-&Y^<cIYiT
z@BKH{drCqP*{`U{i=TIjYkQsrIdZ72t*z-zbRszB2CNQVzfa-k>QsT2>xaIpqf?cD
zkp%YRKmEBoHEWnyV}mr#Wfn;|$sKWR%+(ELm7F}}<Bw+T`;z+1z8_g^+}WSST;dbl
z@3C^kTW;NrP7W5&c_4=?*>y)P?YzM`46W{s!5UiJ#hD9W<#Tbh+NUjjd1*!usL)d9
zj@Y^%o=IrP^P*eB7PDVPE)%0MOyAY%sY;hdeLw{XT5SdZlnXH40|XHqcTWA7OBy=5
zvwW%Ct0wzs<OFPH3g@4lvu6x5x*fKGu{lg)Uh<QSe`qxLA$fId@ai1ND;n~UzcWio
z!AR?R-Omr-^h-QSza(NU@CI5bug)Dorw9o<@hT}olErA7&PJ~_Lv2M(R=?%wa8b*{
z(<mSYh=0s+uG&$A80dRzN}LKo(gf1;B8&lF`|~8rMdFxE!&*u&2J7%K)CXM3%~r)l
z+SA82V}+5Fx&K(#-|vI_@W00&Ed+~jX#v(Rc6fo}ERvJKC?KrID!4~=_P>9~>JK`R
z;CzpFN3n4&FabOUH*L&$OlZJ%15Mfa_9fpEwnp6@ujXoar<DuXL?+X%v$4QbsOl6J
z*VW%#Mhs@u-IZ#z&jQM?0Y7pFwe4IojDTo7f+OzYPTN;bS2cFJs!~m0Ra)H(cYmzv
zd&}T|Sd|XO_c3?;5ZBZjz^d+J#GJ=&&4@q~y7se-;svS)RyqAt7P`m)A@C{lUixza
z=A!wUS&m7chpzu7B9IcA7;}15&b0uLf@mrU_unM`<1fPKsDRN2i*#Pr>-^V8lDGi=
zZ#Hn7Apgzg%YJlT1q>cYAA|j(wASx0&D{c`1324l&C4R%Uu@redFt~CkmN4Y;$Le6
zJpw$rb-*jtOBC=Qa}WURQ!>gm5$s<dy5uGB_-R7&e<f?cwlL9uTfaLEE0B4irl#(m
zo_-P&lOPXt5H*FZf}j<*NSm|B2uY=ZcI>7W7N;2r=D+}TbbSX2vE{#g$z_%wL){CA
zOqa$=SxC59%m31T(~s?28-^<UHu|kJwuW$X0{oz7f6LHxYPakq1f-wmAO=uNEn4N_
zr2_&KuhqmDtT_=LE;81$obd~sTI}%#@;rSw4YG$Ub}uQ<I{BXFd5|-SKVN-9T}U1w
z7u(R|eQPmFBg+CQk_9kydycl3sN*?5ScMS*fL4c^tS_^~mHMZ#T-Os!KJA`&@)efG
zW5-OdbX;w^O7AF9+q`5!NH)>wt%)O&<N!!eu9lS4f1Lfx8`BI*&qn0yZ0f$ob_2=0
ztMpw%RuYQ!p%TQAx!>7htVYUd(6#Rn5*FSUGadpi4I#*L+YY@zHUA+yNYXz#`TCGU
zuw+pbC>5!4rce(EuA+dgfSSQC6M=N?q5jBg{B*QZooJZlNaPNeU?X%U;6ZmJlUDZ~
zt-_t9(Tn6?_yl!nEDC7#%fSUZZS}j;z?4Hnld*0oXmHbD1ReOwsrGso@}9}Aa3Y%a
z#-vY42}&b;O2)-y)_q^el4(c9*rPgkhAFK8HJW>N)J_?bTd9p`V9LRYmj?cPvA@`1
z(L7tq?V?0n5(=^b)n8Xps5c_teqKGk@_KLF9c7(ukXta!U=jGbJeqSeFH?@&c@(JM
zh8AbnPL@&B4$}Ta4xz6tpN$La&nJ=z?YMAl)@ZV_veUN$`e<b*ojh|s@-I!%HY6Y9
zDo!Px)?0qYDS+dG-4JCJMG_AWR<}v6crB454rE@!d+=sMIahFf`si<5U;U-|e`~R-
z9yu)1*YMN6L-s^8f~Ck{V-^k@-qh4pNYaSTY4omQ74ai_H<wBx6g^`HGQL}u75<cN
z6@_GS6-W%}ah)UCObTj20kWh<J<mTmfBqeCA7*q+LOTN~5s_B>wM-7&?Rs_cyPc*8
zTCVn*>U-zYs*S;=MI=prg}K>D>zjNN;{!gFSBya-+!`%JH8u6zhy1pHP9E^k+o7r#
zBOv>BXXqQW0|Q1pnN>k>#0`KhX<q9;+|yM$3<S*juI6S;OnyuTr6<AaX%s<BTCmni
zmD}C4?Us9N!JkpidGniU=wL>prOZjI&})d+sxu-^JC2<06$6z;eoQwBRa8+Fa}>kf
z3hxVsX$Q|v9CEhe^oU#4bMxvki3g?036pRQowjD1OE%W1K*wFcpqR@m|J)rZS2H!j
zEH}I;3^*X5M(L}x%?Q|H7-Ch40MW%>u`HA{av<K^L}~hY7r<}S_=gALIMs9>sq?%P
z>dgY-6iXh$P|;d3%tOUgQwL-Sh5H&Q0`7?P2O?-iJu*t^Rl08H3u_vudOb|TI$E6^
zeja21`7u{SOdOkjjbQUZsQTSxw_dblUlj)CvM-Mv)ZX}#2Rd``HM?j@P2@1{0L<@{
z{lajvvGu%k?1FtID+Z!!DuM6QeyJa-!GqYB<ZGbfr;eMf^!<?w6)@4N0lusmiCQri
znPA%lfNQ|P(r;NfThuCK`ynXdvESU@A_7gu^O@p`J!V?;3R|%TbYU}so@9b2VH>PL
z<vk?bkI2?*T?Qi8-|?zu2<;+hxR@`_lJD<&73oD3@v|>ZR8Dw|EBmJY8Eg&HelNGw
zlRiKvmK1XBPccs;<z@}F8mms(F7FF#i{(uJbbCabYcDvzg7c_hwVJtRXIRMD0rcg_
zGs}t>|N6~jzzd5IJGlR_7<jn8!hW`jbyeg(j3EZllq!fQhb70$<TA3YdUIVJ6zhg&
z{Mq#SMW$x60arGtCiCjpiT!v?tSIwNe$Oo1(<3g=ouJJKUJ)Fn9X;b6-MwrUc|>|;
zOM@DI>IwMt#!Rq7htnwR0sL#54<f?&k%FfR*O&pm-uIS#GgkT4&qaA@HW1@N?#ZEm
zS^HdMf1IO@A9<XgZXBfX<#FQPL(p=$XDHX2>%@r?s9r`dL|k0FJ1t5lL3fMKP6S=X
z^u%B9d(LJzzEG)EBx--zTDsP{D+)FBLb^?8ZPJawVM9d;-J0`w4V;dMZ^c(UCSF}w
z*FV51rw7-t#)Y7^`&vGPTnUuK7QtSlC$kaqVrrHsuM*FG8s1Ynb(lY~-LXt7euoOo
z8u%q#Tm(m~8#SGKuUH~Yg!GlTd+_yRc`_x5-O~H=E(K!v!@cu@iN{|KI<g;P)brma
z``fOa{SY2N3KH0spU<QW{=9fly5r7$G~UmZHs-qRnCpRyNzO{6FLpx#@rS`m3KGU#
z!YDOx@QWbE+*KUEeGk3Wxvl@n?I7Rv7Ka9xs)by6=o?rurS;)Vg;al<jJUoqlH`-%
zLGimmps?*w4<;*0C63IU@&uLC#fA+v9ltb!6jaXLg@I}|>zwsIqV{8unNeB*=x8az
z9EvE-Zw?&`=u-q4Jt_WuuXau8%Xmgy`1nT9M)%^xZogao=ab`gvH9Tf?@{w|o`(nX
zi6<)^v!PWRdV(R$haaOh%smhK?SLqP=-XH@y2QlFXXJrM+ppe4*77IE?}9nn*1cv0
zbH8rim3vSjmGnohp@0{0rjI-;Ta2+ptk064!r1`8MxuGq!k#ne)@8?36DvY{E7W4D
z`9n#to9ab91#K*bPN>%U0ll}pn<au3n#Zd&5m*bRa!5z>i_PU#;`fO+6L+8r5Fc@R
z(rC|Q@fs6B*(;TI>lt0t?{Pvg^NVRrpA4>=ubvqLHD8(L7o)Tzb&t$z;u^#ydgy2p
zx${F}t>GLeZuwy=p(~y>2d$L@bDuZ{j_;k;IBSlkvh&_kRW2-;Kkn5MTyY7mOv}zM
zXFh6iJxOtM`LKuWxgrY&?snIna{?^lk)YS`j|_CNj{XSDl7YEsupxz(7Lj&Ra}BZx
zIDCN^<R?#h9M|m}?(-aD)$oJInx{RAn!bZ9SropgZbYs|Y;ej9Ea`UDh<mz_xlFNa
zOl4l<3tzO}sv-`@av_v!$6szL_2*XiENh%~syV81>0|1znSoW#xlRYCgt~V{KVPPu
z(8`HRu5xBs@;6^S&e&LIQi60SttdA$te31;>*o|k5H66gsm>JV9?ObOL>g+XSg#%8
z8EQKQ!6JK9N5&Z~_jHTyC$~y!1w8HwGSg`jRNIgku3xZFO>q6)A?W8`7yVMVsJEeI
z+GttXou^aTGf{ISuu;Ly#}~_du<f*Aj*T6Lwi1OsanKvqokHpX@|brz{xWK>f=8?*
zaJ+EBIQnhp)yX#c#+&*`N<%}to)+lX=W+PKx3L}YpT#yX46PT<(IoB7X<(C(QxPZu
zofLIR^8-1A=Zg)mL%UV(Hyd&A2CHB;OUM$csY%mL<R_?|Xx4yg_IVJ|n!_^L=yZ^*
z>dXjcTxY6gm<TlKQik&=Uc93%0(q|Nk2(@}1?lA%7kwUBf=6#MceCF}rd?@t?cIp5
zsPug<0k7_`KItA75VG!dGa-8O+)}rS#0S)s6ZhzSch!~Gps5Mz&WGhyGYJx7Z9%Bt
z)t6`yuQiL{8;#JWjBk+aHN(VoX3X;s`^vw(LN%d{n#gQ5G3W6dg7F>)!sy?p$1Zv<
zc4$uP?Rv^+cGN_d9{f(~qzPnZ_|p2OvGv};=6-HXgnZeE&PFazwPP36eEjO3@Ta_^
z?~Q61&K$f6d&P<Q$jY*E3S82#FZ-RHomJ-62iaVyQ>&n;!{MlEkj?(`MB)j2rm9X5
zW6lozRnWTT+t01`Ff&HBc+RHp6$l#h?vrE4#-#5RzpEO5Ib+_$6v0{?2%OOZ;g~{p
zB#H7VEN$Q<49{_lI6y331d|@J9*vk`2lbdchRLOgLHpRXKA?iZBR3;oz!6f6Ne3Bl
zcs)0&s{^@U)!?qGmn8GQ)eWDJt|XdaYWB6Ilw**1&`4_xolRdL%6?Y|eG;rXtjQu*
zl3?cA?6PjdTSg*Yr0)xwQVz!k-)dW+?XGv_K}~jp@5|ych=upmx%5>ZKG2cVTD2>J
zwzUNJJ&xJT_)OOE44x8UT%s#D@Vnuz)x<1Yq2Wr^bD1GaW$^r-ve~V70Jk4<`lcGE
zUgMzy=_7}zmDJe4fyGW86|}{gl}>z@IAH*Q<TZ4jN88!Ds+FW8K6P7q9$j{3192Dx
z&kz`tZ?4ZR@|_#|PBt)VvWcVUOHm|I91)`WENo*79?FQTEXe_qL=vIY>vhd|^;?I}
ziB8_jFR0R(`e)&3;U@9l)tROn&&GZq2rMvvQ(%E%K*Lgt?Socc1}gdc<1IH?h?-g<
z?QC{}n%Yh#*H_&(?8NMB<+^F-l&Tp{@yeD&|H729ui8wMi1)HMF{?04TNqx6#=J1=
z-?4=U4r5|32deVX4r_X_EM$Yidx|%zDgxyIGMdO_Z>d}*gBa7VN996zV<PmJv-b~G
z8;iRh5Kd_bX6-faWX*1Qh$oYLRaudl&fdup!;t9eNM8C#x!wra<E?dk7obJtuMiQ~
zEtD|6Sc+c_r%3vmdosJc2fKqQyVlQD!z=)5`LwRK>3#yfI^S1wATb|1QNf?#Jfa6@
zi+7xO+=K4ntO7b_IT80*7Tgi!fzK%DMFGdvqdl2V>;N;ZbWmnKkTASwx>M-{aUGa=
z@pyUo`}ad!Ix{YS3s(R;e3kWc#gA+(qZ~Qn9mo|Q=lpA&jJd@P?@KL)-V8kM#Rp@}
zrrWLK(gN;fjmL`-!t{v35}xGWP*UQN!8@cO3u|zdjl&CzuXlyNiIqcb{L9Q`1QIRF
zw2-N$%9qbEyq+{VJ2zE1A?VrSZDQv_;R=1}WgFWf-*^4HZAC<!1j*L0PW)lsHN$Dc
z&>5y60{`{F`m*pA+PG#@NY&bSCJ9aAcBXA%`m7j9Xy%fNyEq=&w5g_4SZr4*2#gR)
zHH>Eo`2rTdCf3!5%bx{vanaD1?wWAc%V`A};cV5zlXJ@O47E74nu^q!`I_^O86y(I
z%Jkg(vqXsd4ns4Ddmp!v?U4L#L-a9_61FyJ(kqEwW%tK~+#ZdOsh#8xWOyFN3dvy-
zv+;059DjGP&O=t>-*_*Sh@vF=(KYUc8^{ngTz7Ju!j;ideKfWiTeGvj8D%{jbDPh?
z?hy>=-`lrB;=wxBx+%M?>0ll_e2WI@ns9AD=`S{@(!Utxs}3LDqaWVLdLWbt{Akvh
zKj<H@$i=m3FGt(i^S|OdS@uGU2!_zqY;IH;j9NyE;?QfZ)s;6>-8gN&=i4;2t~7Qd
zDW*COHtv$jf^x4eb<-fMajRLQy#$0^sw5Be5hsU8Vq<q3`7fl}rkd_3#UrsCo7E@@
z^g;dXRYuGzTf#;H8z`D-QP|nDYRYsw@-6V^4Z7juN^nvHw3(FAAk-U~K{`Ck((Nhe
zeVq6+V(_m7AMW;T<C#vv35yj;{3Rd5`*W(qr?6-0l$aP00XNf^HNK%NuP2!Q?N;o<
z#+|LGwr0SFv$(SFYcQ6+kE<EQoBbCHNY@HbGHQ0i%2#PS+%_q@;qroPV=t!U=ABNy
zjBcjo>&9X8uRleXVsT%y>tQOjP^ffVZqKi8#;w^rH2yXVC{gchHDmnAnnK3>{CtH4
zpHuscf2>$-ee2RYVMxVW?R%``K-TZtvm}S}VhM(Nf-v;H7>`rV?X{R(yOUtqoST^0
zqIIf73<AlcsF}20RFGW$=9$P>{LKR+4!ZGfIu<Q~>!NN9mYL}+3URPEo>Zn;*fnAr
zGn^4L3;qZ72L^az74mSkS2vosL@SB5^7+e2T$;lsXPTC6UHUd`$J!8i3HO)=@(76)
z-F2G{)hsupg(SwH$Wndz+EytqA%!-n`LIfDampgw7}bszBbCPYW_@!m@72^etW(w)
zTR-Mjxyx*w<gsT}g5V44`IdP<cd+MH99%;Gy`YCIdi)qBB$IzMkuOhnFm2rZIB}zJ
zv(KNosnT@;zVcT0BxXW-_Hf>1UaG@Q3Um}Zu4%}95S0+7yZ+I5CUIwFENXn$*fUJ`
zkfA0l55BUm88bN@sa1-fVcOHMcIfu}n<9Dr=g*SM)!2xysP)Z*MTl&U&UbMw7|zYE
z)*9k@4H|-*aVZ<I2STo@Q=%<pMPJ$Q87T)oGS)Yv13POZsooJYN?Y-QN2v_m%#4;s
zYRRJ%1r*ZsrMa1<Zk|RZZI(KU%qhIkJBfycu;`a8<1H?P@ljFB!R&#@^l+tPEA8Op
z;)+{hvg@ku!$S>L3KXToJ81zfU%Rp%yWo1+Dhij!t#7DluTHyGcR$2?f++zt5G{}2
z+p>5KL1i}S?O3O0bP;LU%^+^=rgcrFYshfUkAfvw!tDd|I2MdkHv1NHj{}CrQ|M<O
z{FYX^`KGpY$%1#?_fwA?*VvomX_jMH|5k?J{Mvj(zO&76frtB81hu!1PtQ=z=ICao
z1V8)H8qq|I$L>^ikHf%3ND%j7f##k1rc#y@T@TQqO=G$NQSf|sKPe*J4Z77=wEAMd
z?_O#14>3MqZK=l5s)o?!%R@~*2*f2t-(%)tqvp@fUqG^Cz&=)T<k5rYA~bI~Zd8rt
z$@gj{6zfXs^wmsS_rsotEd{Egn;4D8^bM`J_RX^7WOeAeF%vboa2+Tisq|{SPp&fi
zq}-6Vq^={D?3N#~VWPK3EBh&Pso&ytfbB78c4oGG{0KClI-CzOs`#Vy-Ct45#azxP
zaO6_+rhPALqq{|Sy{f9RW=D%fLBV~!<C_T1);^h*v*?uu8kRCv_%hje1gTxar}ZQV
z{?X)v?r&|e-;DesXqfl+)AmYwR-9Ic;qTwSAKv@y`+l0xI}D6nKv7?NtNiv4Cz^{|
zM~{w4u4Iv~RHdet?prviSw%6JELrL?rk+rvyRR-a8D}@ET4<8S1W%x*4v9T_4Uz>e
zxbYIAHps@V&K_zsQDAE#p3RlN&$Vb*bo^JN?K<ZJb55l(6Xlv=@3c+0#z}>{HFP%r
z@v`3<>dxd4-FQlM`IiW%-mIY7cj5?4C8CzReih9^)=CyUpD7#4dADWWSB!S3ZbYax
z`|?8x<ro^Jhliy1x_@r(f2gJ|-tMlaJE!$HRg?$c(O&E^tH4o{P?X?SIZHZz!ci<8
z>}WU4&cjo8V`81_BwTRcuOl@o@yN@*ey2Lo`s#Y_svDR4;iFY)!xXfPzdYTxN=goQ
zc($g8_lk1`%<$x|Y3YAeXw!F{6ZpBSrhd^7`JlUeK5lpx^^LC@l6o&%2yLF_2#jfB
zAq6$d*A}wO)%*dk#4E}fxRoR_$bw2a-p#T^qm5dX(j=)WS44KQ`d~R%E*Wo&i(aVN
zs!=InI<R8=C?u`K8p<iE-=Nkm*xUcz7xz{S3G|Jy3;)O5p{~!g>+EQSv{83%XY<(T
ztXP-p_{VcC9vHY}?_{qmk9s`ku=rH0aYer5Bv3HQQ=CGsNIz-4!>CA*t8yTdB=3({
zNd$8c?y5kFgL&kBs9~HcN5u6_=cE=zD5hqj!d-%ph^VE*-_xpMufN!7wc7r;Q7FP=
zwNx(@?#X*{_?pFoL+kD2SFyt3gL=Ia)#bzRFom3WGHzBSK0USfDUP46gAp(aNL#Ml
zEM)$X_iw>2dol9b;eq%|(H!nBlgi-JV4u_i^DX7FhOq#%0+9>k3j{TX){9|5wR_q?
zNj&VHOtp#yg5tA)8nfW0WAXe2z3?a+1^@e#@WZI@Yvp}?qz02Pw<ja`J6T3iJ^9mB
zN_vx^y*@J%wx`;+{&N1nQj_&`d*PYKy&O<)i*^UCwKX2GvS2gzAX|HV7h}lXW?lIS
z|3ZA%e3&~RlnZw(%^Q_;I1Hf1IZ7B3L(Wjgk`WjoH)e-sd3^5iO-^&C6czbo04WPm
z8_HW4ruFRA*f#0EcA63VM-W=8iMKmCTcf==KGnfbDO9sr*wTaU>E_^ZJWX>fyX;}J
zo=U^dECio)75c1}R5qay|Hkw9i>F7jq_S&b8|=jS1f`=x$a<7JVa4X-nAr|cAu^x+
z^eO#~If48&`Pv?#7mxBjoqKrTaCY+tOtE8BlT3VBU<>2`xjJ;xql6@b^8tO0rx~J_
z!wBuAgdx+BiS5mpFLUS1yD_Uicj+#c)QYbSY+6f`WyPPVbFF^;(6o?0(IOWKHN1yq
z#_dC$9p`mzYXiZ3b?Y}}5fxcs18h(d!TMC2a7>491BIYwi87TD^=iF133Abru|Avk
z*T`{P{C6^IK%#Gzwrvk|nexYH7qag7Z<Q`+EE~glHOp0Rf&6UfgO!xT*;MOiomo)Q
zo(TEqg+CUU)Et6=Jj)YXEa^~`>ULRXjSF~5xAw#MB0Q*D3~?MHN1m{FZk8-4xhF&|
z-X;6ZUtr8}fzdrTYu2eHLO#7BbFMi&Oq|%N*5o2Rn3L#j_YdnKv)ANZ_)NJToOrki
z+w9fbEk8U=&ew~{qqm&G#5b%A$<8Zht~trc5MT=o3Ywl>>qvmRZfs<Zi0FqPTP(AE
zPRajKUl4u_9!ySg!_jiW5y7-!X`fOVTcJcGZ3i;s!+6A)-wU=lw9LV{GqzE-y1G=9
z3l;;I*s(K)(58NakLA&+fjiNgjO!{iDk)2D>&40nO2)F^BTEl`{YGqK*Vko63`MIy
zKkn5FFB;v6o@N@r7rCnB0mTyC)9qDCxM!8tZa`74EkB;&HtACBUS4JL$N{U;tp=o9
z%$GM%^d448TWO*;%>(2U`}Px13c{%yQ%2Xh>t`K_XJF!q_<Rcjp8hkYTOJNX{yeee
zY-Zc7xRGK4ea76)bco`~L0OjJI){fiM#Tqk-eVcMn*9e@M}BH-x-)D!YAf0=`rjGD
z7cdiBBzgx-gKgdas9Ou?qt<0%N4XzVy4ZPqy>2`sa(izF73JhJt!e_`j@!^;`0n3-
zlkk1FF?Hv#20bnK`3UH6@#JVxGiF0*?d>Kcs+#<SI9sXeKu<`$QZ471lJ>3@ELn`W
zv}R1HK5h7*#d8Y12Pg>6BwANHxGq;<BbqDKLU7D&o^1J@r{3nGg`wS|9KLtG=pf62
z1`&|K_7rsiitcX=iqAgaFz_Ug+G0px4np7Dr?f^A2oGM@3yR8D>xh+W#=~yt<CobR
z%^_iP8EL<M6up_(kB|MS{+UhrSWEnuK;A5^<nWf<1eUbmuJq)it`$(&aNsb>GClPt
ze@-0t{K&_>N~z|Ju3LKyUTTq=j+Lq`nLCU)mhVA014M!5+AJgD#~@f}`cTJXC57zO
zE+0p-Ky8R%#A3@qeC|UfL5P0xdI^O!oE##*`A)mk?)QpKXEtKuZhl8!jAQ@X;O35r
z?cmis9oH@v!4yZ2k#?Pu4Q`iR1tIQ}D)&aqdC01Vkid3;&upU5?$_W9gBGA2_Z;%P
zj2@i3<~UO#F5$M2n_b0zOk(Pybs1efCppwhjJT1+Y4#9)s1Q!XpiyVOI_^HB)qXvP
z2iVY>TS+?nz-5?33FnY25ko6P_e}5<m5}&wMx2W&zufh@53x-l=@SmJ9Z>~U_gV@-
z_J>R050!LfQ&nk_r*vh%mD+W?K=&aBF8Yr+*{a`MACHzjUZDaqMFKcRQ4{{4NJ83f
zlVrZCk5G-ulsr(sSY>iR)fh>1k{+wyxB*F@aZ-GTo;vQ-liPpM@=prn;WG!S%KI)p
z9mx!!r6Eg{e0-hX*Y%XY>zJ16rX*jxo9((Y6YAIT)7ujIY!m-D%K;qLXXW6CZ|_hI
z^PSqA3X;nhUrgBB%l154avKRU=AXGv=XEDxeW|F2rFvJTB!XeZc8SfU;$VC>{E0=A
z%?D$_#P(L$mGUW)AbA#{17uTYXKo$4Wrgm<<C&55;<XOzs!1w<;MLp$2e0!b6<k+Z
zx)W=*R?T-3c5*UK?NOq@V%PkTf~WE^X!;YWvi7#TzTAzQ?wf{WiMiX1k@jyrZ>BAj
z49~vXqniRP>pCvBHYU8K8LVm&<lRVfJ3R&h1|e4Nw2DD*?|X7lQ)eu!<qVf(RP98<
z`j6ht630iCo)j6El#JdAE>Fsx5&QlyhY?c(6GtiEOA=qp3I}}K-*QSxM}xwP0tilK
z`WOwU{MYgB`j@2X9%yR;TegoPDN<i&A}UzbloUoaq;;Uk5;k?IE-vfkw03kexJLB{
z(i>4NnhdkArMePl1|wV4<&TqBbX{2Xi$<$)_ELmem{Qaj|H%CJ!)n_~mp=13FR_)b
znn(cm-Nla|M%KtA9*^Sp<R=^*Zho8<fU8clFX<9Lc<^BUx`MmXGAOQUt1Um<Qpt6E
zqMFjO`I}MVKAKQ$5dY+Nw{cF`kuRketH<KbF}F^r<;2>s&~6|O7_)S=f>L1qj?Xdq
z)M<636m_pe?``&!Ilkuu_=1ye)U4+!-w8E|=RoE3EHta$`s8?&EHNT~mWyw*1K2;r
zMY$cM1TQJx%Ehg;vM4V3%bdN>N66MA*BC$&(CKajy?S@8Z_zFheVjk~UCeV`o|fIt
z_H`s>M#C%5PD{Dqb^z27U4$zMtuoFT1LJf-%oHCXNwM<+dtJgbzXPBSNV`~DH(f<x
z=pQ03QS>rM+(lb27O@SHuK#)P7UcIr0tPuy#2CAgN(<TvTeNOfi{u8bT7<}KjGr7m
zB2zdm;Tfi+3^|gbOE-=gedQ$HxF0UKG*T+$z7&=yr{LM#(U*6;n`Pe9u2b0!pqcDH
zUT@6!+FiFhJ^gIB+>Y6I>N6#}f|#PX1uhnt%qBd4;Kp*TpD^Th5_8IDk?W}hOE2CU
z@e>sO2yR|cVT`!OMIZ~;VT#~FABv_3`-u+_AuEdRuW#S=Z@Q7JF8dj{YxwtA2w2b&
z%ZpjVmBC`{8FJtB>x-krWD~3`z7%gpmv_rms2A=j7#bSR&!q%wzA)F>Zq(n(W09Og
z1PkI4tn@gVbN&psfy_sh;%#7$rr0HY71-06B~Xwj>ne&vv&71mZ1>k4{>XX!s(Ax<
z@B;<zi6?$PLMB3N-<Lk?m|#py@~$P{m%f9^Q=@j@wy!gGR|5wk(z7+S%Pg(7{3Xdu
z4M&1O2>4<yz|~X#jPHIi3A_zd22<b((m$%e{h7i@4@9qs*#HSxVB_@8%>vjCaA>I{
z$H73PAMVW5`sZrtX~|7K0LRomB0GFJyf>e=F)T*&Z=#+41;BBDaas(*e)u>7cJQDs
zrK-<qY33@~!dubm-25S4Yu6CaJjHIZwkn5kiZ1EEHLWjW)<V_=2%8DN=6O!UPAc%b
zoP;6fzO<Nch!;TT%*B(`sgk>zvHOCoh6;_k$RTDg#7W(razN<QIxd)^s+hisGwmsi
zYJUDL`~CY>^p*sh(w_u_2+;mGn*!jXs1%VNi)K)uCUdtB+KN}a7&(lfYR;Ao-@qO5
zjI&$e*?=MXkwXT>>n=Vmf(l?JQvKe(XXoL0i>#cp8Yq5Co#ErpPn{W1y#lPZC5tU(
zJR_BI)dTJ8`+kcht{Uj=ZcA@nggC7USH(Sh&jxpAE^zIru0nZrJQ3*E^ge;G>CAq@
zW*>F%@UmE+7<ZVNmlKiNiD<QGNj)`=-zsxKz{(A$@=eC2GX0l$Bca->=}%N2nRbjH
z0cWcA?Sa~#L#b`a3?iS^NW?Y(ndGz_4QXiP%E@Z3fPT=;{_=^8r{<dwJA3#f$0|30
znIc?p$HjyAAxB%4ZRFa!_!$k?=Xb7>qy3P4eKxSV$|MJH1T2!VMh{6*-7lGn`6BS3
zy^8)=g@cPbTd1A|TUXz;o6bg0Nw;lq4B5WitC&Cr3I+BE8fRw$J1bt|(tXPTS&BjW
zB@~-Mi1piKK;bYuR~hpIwvB~5P#lgMXj#1Q&~`RU0pQK-ztpCx(5XJ9h4j((=ab3!
zSdJIpV2ajD!=<}sC<ghU>-6mS!7BDo<)uW_Dy@c5&d^_kAp)<US`d>dU~7<z<pi)s
zS}qruPB}Z#lYZ9(FsD(mjo&cYt^trltb!7WUs%hF6;fF525@LPLoW6k^}021qN^UM
zCZ!1RDRIBeU+w|mSHh)8Ilm5x{+Qe^L-9jj09+#xXO04Z-O#|OKX&u-w6&$5F#Cw)
z>FQ$-0#Vni(nI$ph+{deB%iL5cgg`eI@YU-f7xhf`$BgHR>A7Kw{G1^6L9?oi}P(d
z{&|89KuYGs6tsp0e(`H+YAQad6=ix!0o(A6phZ$*#8!d*PawgYZ-ix+Bs%NH-4D9K
zSHt$|JfWbE<UIC;`HKU=MhNt<jk#24VJvXK-(+Qq`9IDS>D%=>M^b>D;d0v@seXQb
z796kNvvUZ<KR15fvs4u>$ao=I(*IrggbhFm5*k&6|2V$65;-EhL1nPT8WMn1HJ8+P
zzO^N$c7am;$zX<43?zl%+sjPx-(Lln+yf39G6kkr|2f3JP1yep`irK*#z@}Y-n@?s
zlrJMO*y#H1iid`U`&%jCe!1JR;-~4PT1|WmacIAU`pLyi1kRYdIZaqo!Vb4@q^eiP
z$DMH@$q+GFlkfjMgwf-U*8c}!W=()Yfa-^Tv;XWb`WgV7Qx8u|?>QE5vDK&auG7B)
zh$`>G!2ZdZi~T+zY^r*-JKNbp*TyWs0QTK1+yBgZFpB6u8}v+%s->d)TOfb`0f0AG
zm-xg-!jNO%%{M*mqCJb4Fc_58-+Z+DM-u5b=M548AeL~}8)N3L0N9mA_IX6WC`Il;
zhyAtU-hIDISSCCIpiuz4qowikJR-94=H_=3;?wEetf(zm7Pxi<+@5+tzy=UN-3Vc#
zFMs_?&ciL>>IVd!h%-_pselK^K4szLnn_9g%WR&{eVBo@lu*pI<%#C%DW2#j27}-Q
zI5brKzC|sj{W+TP-@0k9L4IP`xWor){BoUO>jROUz)$cZ$({L6s^<59r{mv@nbSj<
zXxmvG%q^Y)rb<tHj{ltVs#V0>UciSj3uj*fZb5y9xEAl`#<fZy3s|mKJS(6x^qxB{
zD?jlEoa3G`pGO%@XY2mgT012G0HT{4%_@O5LJ`|5zHQLip|jIh`5B{=(HrH$F6Ky*
z5Yjg<TG^DHMiYm;h=iBzL-gk;P<v5*=4gXdN+X7gT;v(gvQAO#6Ce0cop1a^6QLs;
zNI{`pgNeE5MERG@RXE89X|i&Aw|&f!RDhi}dlU!2zw%$nIzJfMf=kfsQA9*UB1f5Q
z7$rmdM|$lim2Z{_yET$dlr%c6+(vv|Yhp7Qe=eYN3IJojy1j?KefyR@9!`8Qy#Um!
z9T+vq*OI?Zf!&F)w-TrnI!tfHo{rHUE_|5Gg|EmK{@mDj<0WLN?V<?q>}$k0pKD|~
zUeS2x2!DNVLJyqWrwfifm&_U@gL$a47>*^9vZ$iia)bj_`yX}aznKW&yQe3jjN=)r
zxe{IV7UBIhj=-kPBr4VR%z_CAUl)C5d1w%QG-AlqLwvD*0CW*vy>WBLs;4;MH%-0J
z2zKh61_@iLM3;#0PbvfFN<W=KC!Alf{|~^*|AS!u{~CgsGL|d<1J((|wa<+o$Fv>4
zhZjPoF2oX?X<9{XXQv`zGw<7)MwesFe@i|(Dvk(t&w~dVoucUMl9xeZYUxn{{gHsD
zv9uJaL)h>0Yoq0<zkdF{P~j9I)Ff<5eJ19>wHh@j?aw}7V+w!$CE$N}IJ_HmPV<j@
z>?TqqgzI0BK7q2vGXSGFUprUDL^N6Bn$7D;a~mSPKX3+rtMQK?b6LV}`K0-%nZQ*s
zD`sQ#i|&A~!G**%&(Gp4+^8#j_PD2KMx|*X-R@FAM<xr#2b>2*4@~W!-J4iX@t{{j
zkrOs)OA0JIKCNBV#$ZIheJdCC*o(HmPJaREVm;0QEc71cM0umr>TN{e)}WvMd|y_d
z+CL~Tg7$LjcmSB_o!#5xBEsaajn7xM8Ct82@mg<k-{doU#KkrS092asjx0Q46(b8&
zH;$nXNq&$geSTEH65p4qeiEM>2(m9W$pFfrT>r+1o4tJ5jtrh8fZtDjefA`I+jFAg
zq4rO&{$SPLmaT#kW1^|iX(Q+Za7zqZ^i<x!>3f~Vh+ee&7Pahu9ESbr==XJNoi^Ns
z{%TnfZoEndM4FLf+;`5MBse_1Xqg`g4#N8T8^8YmP|TB+d%Hp7Y&5L#x!r8E62K6Q
zPWNep-*}v|N0{a(bn?2E(T`f=pQ{Gf|AbD)0ed7AyeZ8GJA7D@O1iHNo<Dz)tlZ!F
z_#NfhLLMMK@S+7hXJo(R-j{2SYn`_2rmCG*$A_;f23^n8(SAfjay^yy;HL5u>GFy5
z>PW3LvVls*{{pRSd`Sj$cOWGI2camB!Zu*w54}ofOzV_K7%?n){8;19wfx|z{vKj=
zkpgfCmM;>0(3nT-W~w@YPS1ie$$+k_PGYFZ`A!dj*kvpIC*U3cx|A**wUv=Ue|V(s
z^_mTZ2^S{qr+G8gq`N|FztT_wFX9c8+)oU)Ccf{9{eFk>qOS5~l$O77bAM-hq|{o$
zb!Wv~zRLR}qAICmR~&tf>tfI{Q|g-^gr!Ad#8j2M6rlmX%ldg@OwVpie_6CELKwop
zTvNrBV3B)nOx!70Ki33oHDUZPH~%s*H=}FGJ)JzaRMgOWO`NDg3_8Z)^C&`xUBVw0
z7|$rE?pI5Gavz4p<#KVb_EI<l>DpH1LG)r`bleh-xkBWKEqOg~vK7Mn9nO%it>kgO
z8?7b3p7YMK?waY~<|T^SUr?Lm;Z)OIejL$2nA1Sn2aF?4h|db(-DD=9^^MX5GpA|`
zM!+KI&Fn`YG|>4IwsEcyU_oqxy44lKAfa#j2W$ieRKaR$h;)!^x@%*rX+RqUd@}g&
zCH{-FUw{_F?trav0$rl=s=&fE<>JaW_m`-mbTyCX;x4u`|Io%cHn4pAveSL)cO(H0
zSLue8UZe!A4%%q#57*<Drra?4sqX)H<qytS;`#C%MUUh=u|tqEqhNaZ4vZnmcRs7A
z>BKW1^&7RlDZ6ZM@oP+AukdZ%P~I~n7>1%;jOC7HTJ1Cw;YD|Cuc^oB5`*{CoXybz
zM`V4=4^?N6T(j0JEo4Wo(rKD%fl06G#|Ep`U4+Q``Re&f;<0_p528N>r$qOxYFol}
z*5{;S@STBl5uMRVtanV*6B#WjJQ=U0(eejlCmtDTBV#L>%{pY4$k;E6gXOOSt<APK
z1Fd!6I4+gtTH!JwtBAG=I~IpKGMg%#RQ`oacYC|)>X;wG-V%T&Ty>jWza5?@`Crxx
zz`W@(>^9ldCII+ZjE74iAHI1()iz^-xfFvkyQYf4ni*btrdt1uns#U&P#v`Iz+MfN
zZ@v8DlfON15*%^j<QTm>VhE@k3ZUJvn$5zpGs;~Ph9G<0V#X1{s&@^|*)K06UA5a$
z=2Ud;p_-^-*KwKI>7mO=h_Q0&RWdbkzK&)bj^^+OX6F~D&*TYw`?{>cSvCqt|3;*z
zXY?$!S}fGiP2K)ltiGfeX$7RA1A+W7uoWt{8mnNxZ{SfTR!v3a&%~cszCCQ;r=m97
zVaX;zwi+0;0H3s58>iK)(bMaO;4Cj$An@_I&Ygr$pBT@4X%x^9t{21C30HHO4wXoq
z3gTlX0e+btdEYX9<~*(crT$&jvPEp+AM^V|%Q_{=5ov$iT4oVmw}ppn(7=VLNc}v$
zTNE#((0KK@7`Vf1m|KuJDOv23Ss`#DtJw1d?1uRYIM$lJ3|tt<HDN0=93@aRL-}oC
z%G0~>gf1w$k>%UEzxk`(YBMm-R>1~BlZP_Jmoyx`P?+^ZXbprPsGHTBj_ugK5I1ct
zx=)0p+nYxQ5?;X5t{6;8RlbpNi4*zBO;%sT+zx)9_76j|*BBnPqFK<&J?W^_MR6%B
z+<3&}&PJnGt(xHrhWEaZ3s?O>m}b5O;?QwR@Hgk)MCz4!3|A%n;IeZY_p$u_6aIuZ
zhwm(?*0lSb0hrbv^8QbZ-5-|V?-{+KP)y(p+UO6q2&&-JgG=Uh+izF#tVzSl)ZjuV
zcCg!Gy32==?F!gH52%{V=kEvn-RfOa4(pF}ovjHZq0<c!*K+%+kQLWR_tw)M6(YR9
z_BgzSvAfZAjef$~UU%BuUdU5;!z3}y>G_@z!7Z(+YkzR;cLfKoH%O{&p7k-5=M8>t
zGF&M(TZaCP`CFU+X8pm(kLbusb)#-JIP?BHlOUN@ABZWPi_kqn_+h15I`)dA8?2+X
z0&lpI9V{HQ%${Bdx3B`vEWlQqx9t-jE405NBeW{ADZqucejddm9ZFzCxk^2$n&Kv=
zBSzDF)$LOtSK6Xg8CzgrKzd?D2b?f6bhd2)GNmZ6m140jDtBxeW6ms{DrHsrpG8nu
zUj_MgJ4tEjEjzoFN00pTT@({G+6_qwWgm9PsF_;q>9V2dj;*J>p*m4-SF4_w1M##S
zZlu){w02a262(vU;!jwPzA@-UuDG?5@ODF9R(<!^lS({po!wa1+x72|%@|#*L;iph
znnzHUe7C-&OmHXQc3Kl7?NmzQu6@O-yS&s6_t#wt)1IAtJzI_abv5SZok{M!RM%98
z#Rj}y6{aNTE4NGDe|W6XHrPBHbXZ%W488OU)*m*q(;ZClGP%TM)}Wf#dKSwNlVY{%
zPy<JNt@L;!zk}{czKRgOYG5Azj@lv|ds&meHLZIjjgz8VX!0Gpyy4<E)vCe&r?4xJ
zhkEVXj?+0RPoxxO>&Tii*2WT}ETO?cD%+4X>sSU^h8GowkbO6@O^z63X_%q1C2C}1
ztV3C*u}uudU}m0Q?|VKy=bX3ad47N0pWlDqd->kiec#vhy|(XzNTv@X7dDdUE{D=L
zSi)lmz3&ntIXaH}Dc6YFGjulMBjs6R<Fa9!WK3;byet~>nH}tDJ2CK;PfC{aQ-@oI
z!YsuXpLL^~>`K4)D)#4rQR3oOZoS0NQlsJSwo=}@gH&`Jv+R+m8fk91MIK9l0zsKi
zK6%1j5LH7nBrk0lp>hj4EH!HakxgRRKkIb_>Hou}n5$qtMPztRoc?-q-6C3s^%)S{
zd?hW3#Y{DX)YN6$wAue0J$Q=W0zCBv<;4I{j;c`fFVg|U6e`ss+GV0O#;I+40yU3K
zx>H~!K>gCie+@9~Fn3}uflvCnT19WKMejQjAFOXJJY|Msj|6@kGCi?gZH}I6{HuKL
z&w>5lg8A8Cc4&}3wZCzglFoku43CZDex{Wm1`w8OFXiHu%+e$MU&iU93Q+3jDb6Ay
zWH`*1Ko)Bw{3xzD>FblGvf%*Ydl|#55km7`Q-IDhl06?EdmAwO!T=~U056fcfgZz!
z;~rR<6gZ-Y??!8k(;{~CiTdIQJNZ@LiQ;TJI|c$_&0dAP!=UcInx?DKST7Rjgz;yV
z966mkyvGNpqnHD*yfs8$B|NxsNZ$L-he!9D+y7&o17NDL&eR8d!e+)ENEs-BNH~HV
zx9p<ofJ@CM{QO|@P}xU(yx>Sg84l}1uU)E4Zc@<Hvh?VX>`^Dda<^9}Z?^g^Z~@LX
zSR!r&up%aB`r||T_M_;1zawZzhvvHta^PhMjD}8@Or61*p43Ty1^-1_Y<)Pfck`7~
zRK)I!1l@v;kns~<2!xd1;0?USnqtG7o|h_(tv+>y_Kzi1#KTsK>HE!lJdkWxv9gu*
z^&^IdfOkByh!NaPUGbryS1q*q{}0t){*=ld;4>Qo6$zpsM10)hjl)S=%VuC9KY8-<
z$SpiLD!{zm%252vuV(g&m~nhi+kMJN!O5yy)a-@U+VHo`B*C{CS@B%atL`Avu=dG7
z+YPLZd|fB|0fKb|Od0o~ni8N63a4T1G4+j~=pP25r{A8nUp&gMb08Q+yPKe0p)C>I
z#|&{pNphe$j517&s?KcF-r`F^p5%8kEwr(8_69CykU+RQ-!+_{7H+T_IzQd)W3fvE
zaV#R-T0Zqu4KE2BmX#rUJijOX|7)mX+&!0eagFB!c|wS8r%pS9a&645i~6p!p$*#I
zFH4gq@IUkNoDc21Rat6$v+4F)aHal*7Q9?}Hp3fVJMSfJ)*)%AaA>(wQwHdKSdS&I
z?rWpjBpMPj89jHF$|NA|+V-YX37LY7WCk3zTfCW`7Nw#?8@DiLXi>W--W&<+G**oY
z=ubSucW*pjb44R<L0^2MbWdo6L_RA#2hrXCZJ+)z0d8(D9taSAil9JEV>){qMeLq8
z*zMl}d(_xpa`n&8usUjd3U_A!cO~lI?V{hd(+h{QD5Vtl>TVG=Xs)|=C`L|o_&PDW
z>Q%B^v)V|LVce&Pb(jMNFX0>3wby))6_&wcwZDi#6$85Rso(iW6j#XY;IFiLEBw?Z
zoj(Q<64BRA&P6j@88I3xHoQ|s<et7Qw@M3R=XIZ=@$vCJ4wA7)%tY;|QXlh&>y-ck
zMui(iiHUg!s}ffLEhg1zh9N<H5!=5x<q46V`84-yb+D(v=9g>^R^zLFkRk-o2GcB-
zM=>=&d35(=*{J7oaUHBOF}QraA;gtRc^84${nE!)uwdj9kjgr14QU0P%U3J$!Icub
zHv@Wp;r09$`}Z3G#?J3&<4nI8%=_k`-{T`lHKe3!8c>elR=smWxm~Xd)gC+Zn~HsP
zn5Nevpxu9S<-`Nt4rY(wUd3RiR;&9}ymQ?DN^n_|9BGRV0Ap@mBgAau&Lou;HLQ>a
zoOWMrZilh#+qOn0hZ6`K41#`Pl6X9NQMtLLn`<#dG%8VXx3`k|RecWYo@B2VQ|}WZ
zJ=%z3a3{h1PN(fvT2)%~CnA&CWgg?)M(St0CJe-@<kuPp4)@DexPo+L9K%>!R62(>
zf@p0Q&o$T03x&n5Kjt&T;(~S~WES@R#>xEGvEmP?<Lq}(tpHp%*&g)8z@0C`WI^;i
z4ptIDN^0rGFR_*=#`ZKjOCtqZFms^TPntALeA-m}*4yFzsQ5@62EV<#nMT`b#b%2&
z#>VvXqKW}r@Hkts?CgSGe0iGgIsE>LP-Pf<V2`LyVo4#_uW>Q~SXxrTdWZ&4dAi;D
zGh5pRnH~!iyR->D!``G4kiV=>VYOLZlQWs`qSu3cLZVj7;2uAmLJbl<3gasP=jn<x
z<R?`xMv0CWsZNKv2auTPHcH<2B`z_CuIi@4m`S-6DI$H8X5~&t_(KK$P)nFBPtB9A
zKa`MvhjiZ#rknM8c+ExA;F~I~4j}lJoo;3E+xL9}bV{u#?;DN0n9Jn&9vdRan1mKG
zTDwb`<9xo&kNj23mtNK8m{(K;MeXh9ZbW{?1yq1O1<70cRqAp}p%DOaK&G@^5`P7t
zkJ@J^mv9CV!#cERCBM1Wt??ES24s~?-4*eFbED$9`ZR)GOYIebLm60xyqAW(booCd
zD6&(m9YjKTnt3jglBA}iTCB)i3>&USPpSI7o8UYK(#k3sFN-<P4RVQ0^QmK*w5?Ya
z5>qEbtMGf<BNIZurD`oD`Whx{S4=iW9=C-s+<p*3mjqeCJU`gfwFmMqO=%ifQ>w0g
ze~EnSuOS9^ZI8_0vWGH1eg*saX_Gg3M^xLaTP=i{Ej_rb7$>*1pa8CmfUJ$Qw@h71
z&bCo(&8COu-mH5Dus_OvSjd)IJe(JKZhKYxk?syk!fXb&xnYMTgfC_iueTtL4rmsI
zc3S(8<|^z)nGzht#V7JQA3X`sx$(otgq${NH&+KRP#R~JNwu|zaXASMyc3fnJS)uJ
zn`V3fNJbg#EhH!XCp~C|n09ETYvcbU@<6_H*cO+2S-1S&{b2owm+SUv2?>HO99no0
zJ{WhQk+Htg?B$xom8wzZgu=}RQ+da|g`GO*8xf<f1mgs;P`M`+nLFV>?*CE7``b+U
zdxIsbB!nIs4FwmX{1F&`)pM5x2fW<McTYM^x!k^I($$zltXG`(l7CcYsqK$jU))TM
z33mcgf>?OQ*;I}5tYJDTyMvtquk>@;Hx*wCBF(K8`G>WiT|OE3RfXb5Ss&RafBt+c
zV#2!QQSIplh$DMife7y0Dpu3d$|zxHp`4CJ37lFBrezX?OW3?jytR49RambDWImP}
zi=GBgKWbLE;+NV!P)uJ2Al{oBg=2+&lMTvZ(L0O#c1__fE|R<jV@2Zlg^$#n6}qjb
zt~czcVaX8Wzf1|eQhn~)1E<tXiRTl`rl!2vJ-WH{i&~f*0qSDi1>mCadT$VhR1=AH
z9~lU{gH4DC6T)6pRkh|Vdk&;BD2Cn4nIW_A1}6-D5QVUm!g!Cks<xB@=5~&+F51@D
zCmf8ckT}M2mOmJAcsH2)2rsluHbpAk_|cW8uzO09zx%N+zqj{|O_Ej*M)+>8ML3ov
z)YD#*wci!xPqU_wg<XR-<ELjNt9u_X(|gf-(I}k&Y0MZ%U3<?lLVVLF(%t>O8W=sO
zc)Ug$$<A7t>M&~+J*9l;)Of?8l)$6@lLC`@O4`;RNF#ST+i^&Uj9+V8MQ!<jx=Cv|
zN(8L!erHT7?0i-Uj5tb`2VHf#CS|2a<@KzH2!={LjRXk|hylJJdjGOA`fe+>Q`(tS
zF|FYHu0C%eLvCuKt&=Tw5yjEmSB{PhScPEVKe5WSt)@DH38L`=BbXDHUb(LI6}1S(
zC-v1VIQS$-^&_*(uk!Vui!@4*Zy1x&4F8n3xvi7)(BxiypNFWw_Xpjua5W{lQ1n52
z^H$N^*KI*M9<F}2`3~l&?CsR{pckSe4Ur^1umNqGjy<c}D=V`whbCk^nUOuQ_?*wa
z%F|;LX47m*+uk5bMLoA#iF~fh5r4?rw$t!I;J-x&biggsGtb$$>qS2N$)Ezg+dPV*
zl-o7Y#|+0D=v^D04vHOPc{6^ojwhj4<>}+=9IN*RNe*j;V<c2od_SvzOF(oo%k<6X
z8gPafDl38Y<FOPYsqsfc^=N&N+YU2vblKadB9taxYp+*We^lK>jD&Gi*1rC<zi@QB
zU3gM?XFeY{;3iGKeMQmHcVw)(+6xa)d`fb>a9AQUl;AW-d%1TZT<rcxeZsX2!^c7w
zzUvBh=1*4SFE~<1TdEwA^yTO1*5Wu@)_r(8pFd~+BIpAq=h!o0iME{%;bzHmom{U6
VyIkFKQF&azXJTk>P<O>S?w>N@qbvXb

literal 0
HcmV?d00001

diff --git a/api/core/model_runtime/docs/en_US/images/index/image-2.png b/api/core/model_runtime/docs/en_US/images/index/image-2.png
new file mode 100644
index 0000000000000000000000000000000000000000..c70cd3da5eea19e6e3613126ba7b42ea33e699ee
GIT binary patch
literal 210087
zcmdqIg;$(S(msr9@F0N@B)Ej&0fGbx5S-xd?gWP+1cwkDg1fuBySr;}cNk!p;TzuF
zecxx7=eK{r$2oK6K11Kt-PP4qS5;pfEH5jLfkuo50|SF0DIux|1A}-A1B37d1qu2K
z-VhEI49p7wa}g1FNf8lpc?Vk)b1P#Q7>VFmRb;iV0|cp>iWCSa<mA*>G(poaio#~F
z*@PXRCE)yU4aoE7C&Uov-Vu~%6j6qdV)Gf8Dp<`EqB9IOs+BX^DhkbcZF}s3c2*xD
zjyr;;qbZt5F!!jbqEVq1a1pfcVtW0N*u?X)*(Bv)Uy*;uv`Zh`B$SwEWsQVcY`;6b
zIYGMecb-uwy&i)+T4=!O!;-;JA#Hs71S^QO0S~jT>=|5*4U?H%_gjfB(s2I!ua`6^
z=bk*HN*fG3vZ@>WXlZ^QFuR#yB4*q^n$y4#$5V_kZI}vUgy{sO^uR?D!SLc4v^6(-
z!giZuFA0;Ob~|{f++*=3cuz@_O>dpEAy+uOLwk!Hj7HVaO#f)g1F75I9z-%w>&i`~
z#t4lbdo_D(A`9?)T_m#nRyI5&K#gPri_)i)%pfL_8eK~1KI`)X9A@#Wao>6McJz?%
z^C%nG_BKk{A#!8b{&o}rj0h=A3|llp#aQQ_`k(BY2Tij$1ZDBjOU5D`<{2Qb<qCrs
z!<}^FqWe|$+eUF6l<$9kwpS5m>sAV44i%MDz_t`RQi{MM%eZ)@p4<~md=@#lv2~{@
z{F?S<AQHoMMdk~0iR8gHrZPRf*7;s09PTgvKztKwKl(%8nlrbMhd)~7L&B_l_Cn%n
z?|B{c-ofe1mph!yvuAu@)ycX1q)0@R_d4_|bptDd_p;V+5{BIo3TH;T5Y@1rXtX~+
zlnypuMRop0Fiv%~=VJNzIRUE`K@<)X1x}jWmzi|S*NWWY(0tEFbubmX9bPd4)vsNl
z=fzJ6m8+mXtZ!cW3R2+y_|akS6QV;w_v2yilZx9ngjXS}ls!SO&k!~~V#7N8sgY*=
z@{x!c^<%)#J+cY6@^k&xhVlW>Bfnrfz;r$$Sc$dLCrJAKJ1y&PBGDIb1F(<nM3%9D
z9dG}z!f+8L`+HyvZ#D`3LCr`xi?B->m#pXfq(lj(A5uuiI|$6Th=dq2+gaxC|3cnq
zl_zD`bg!qd>t4}&&2@tHwn-!|Yfw`Me@AZ7JE;T9MPystJ~usvA_gYpc<e3b5$7_p
z@>UI`LSRs93i)v4)zKY~z;56H`Z~{sshpKSEg4^EO82OtcsWvg_b<aRHAZKpU&yL+
zhCK(wK?HgIw8#fR5FFd}(LqM_b*;nY;|H+?Rk~y*Z*h3pR-+K5m;~j1qb`;o5d-*_
zDFLk7$=X`55#G8nWo6tIA*V>0bN4Apm&h=!eIO7B%X_yo9KL)GPXH3w>R8KkyDtyF
zzD&_7t&zQDcjp7MhWqU{`xOkH9<0?LPp$9BXRxs3M1H<z5ipAFuD`tqw00HWqe{b9
zeWIsBmTLb}PGAL-)Ztu?egu!Pj`9I@^n2(!TRrAU`^*%g&Uc+NOwZ5QetzTFNCw}3
zQ>p~PZ~Q377NjBy7Tt(=Nrf>W+8y*FAz=8^V5D)sh7q2k?~xcL1#RS|1fK)C+xKs8
z%Od6I0t&Nu^I0vzFvZpBNWz5h#_i2$cyX9N?`DRM&zNyG;+2TYWfD#tn$h0F9e2Ux
z2JGvb{C?Zp`9X~%F2q<*(F#c=Kv{1_jdT{_^isJP@32E)Lp0fk-Cn%@N7tEXa=2W#
z={ov(&8>6u8{3Yu4XiVr2NIvp?<qfEH=}BGnhDeB<7ULjf13aDLE-~OTF$#pKFY{^
z@oEb9PE%I@-!jQxv?R2|mt{VFvZ58Byor>cf{%h9h<_=w5s4<5lg*i(lD(PDnEh@N
zxsrGx=13TmA}^|E3)cbB4f&4f4(*Qqj$JI;K<qqM^eaLPeN0aba}3QO^%ufzcI8kD
zk?R~9`Kp4df}#Sa>8a_aX>JP(iyaFciwX<M8g+}JnVbBSuO8_)3fh`0Iw@HKD%!cN
z5?)F7sMneYwq_iOoXp?jzA$DpXU>evPMk~$n7!vn=BOuz3rir5CB7zB<<Oa=|HjCh
zB#^2jU6y{Dr<(~hQ{*V&<VpNE@+vVlF_43g15{0<@wPg%=AxRw;&`05>{>c0+q#6b
zn0Y+4s6`zVbMqFdw_MMT$g}r#FF~*Q=XgwozR!V0(IsC?tEBQ|@>Eqd=VZ7gT`FyI
zP83#T`K8A+!i(aI>eOu0Rxw)SM1BUA2ciTr5h-$S4MqU(MXuqm{O^tLo36_az1j4L
z5lFU(m^jVe*P8~BERysR%d%<Nx-OQR=8wv!WjtgUivTfNto?`=NP1q$a4uR+TTGi;
z8N9Y&V%0J99IWpX9iwWeS~Bt<V2!r<8c>i{(ErVU-+pRWTAL;*VMuGlO8+MDRf2!Q
zz%W-rO@bF!9jhD%3+tLSVC;uQM`eGxn*M~<gr&I!&!9u1amltdV+j5b!;nSnjCv7p
zmTPuk_Oh6yIBj8X$<8{&+QRx|ancHGO}nVFq*L{FvEs<#Xm&xnR{VJ7xax@Un1v@D
z^9?3tpcCeC7fM%iKyaWk)}ft}ixD3I&n&Ni-K1m4HwJqPJ9nOC2iJqMoUSp<$DN~-
z<iqBkyHS*5oIh{BeC7yqJvKP)bSe1Ou=B+c!#-tyaqoU`q(f$}cJ(-yugfb2^bu$R
zEPIpsMlqZ>+(u4njAM*p3^NXsF`y>GX*6^le~)wOFlbKP7VA{_)N9rI?!*1o?a`SV
z*Q80d-Bw&=xaWwMTWgXhANV)G96$o*1&=(qL1rLW@Jw(Bu$u6xmB;f7M=3|~KBGR9
z?Y=&zJgdq3$#|dC<jv&O<u%Kl?F`Nq*Fh9J6bTeH6fGGi8DB-@{BSxcJL3YJ?7|v6
zZ7}M%m)SktU8VfoMLoM0g90OOA}9NE754Ja6o$lC!YV^>#W_WN&1|kN_B>d*Sn&)o
z`)~~n4Hym6y2FB!2wb>QaRey<;tx%J?S4I|F1Xfs#=#?!SJ@SkoRX)}zESbiF2CLg
zxZm%E(g)CI#`G#FS|}dOPPejl7^?i*Cq5LY{dN>)&$USt&L_BP=Xr9Qg|~<2ig%NT
zGZ{7IobB~T?6T0!Yk&Ew@JdG@ur1_K7@J&;Vvpb+Egm(^<ko~~$Te<1so`T4-;iJt
zWs!I{P6>X?Am-3&1XzHU*Hly1+s5JK<fDU*#wtZc?FaGYDBCDOo;K|ba{AbBvRA3~
z4hCf$!i_aB^hB!TYWeWqTbjnTmAXb)ArB!8p#suh605x{XN}IcX6bLf5=gsgyjFZz
zQu1W*wz%Y|GOx<x|2>-8YL3MA*6Z`7bZCO3sq+Y0N{ttnlk@1&npCn>Tx@en^v68&
z((`3?M=~((y_9vq=Yr~L!_={6jLY-8%sXpzEy7~;vl3}}=Zxis=FEp7VMr%>+pk-v
z>oRaC8$k{8jaoAR)jrsDAS?$WRmU==q0)RbaiK7DsQO+tro^?$p>yyO-v*yVML~5y
z#i{g{)=GV^!$=gT?yN~ki_R;xhNi)Mi%&BkryowLv&h9|OM1;2o?UrinPJ@A{x*eW
zwWX!=(#u-acLQ<R%%<LkV3n#FJIlM-vZd}N-nu?(+k*4rf(89*LwjZo#-ig@9_QU{
z?N3d_hp!L0j%Sy&Tak|<kKJ0itxv<ea6pN`s+&2|a$b3-2>bHlHw9^mYfBL`;|sqh
z=K*vN@{ZARQypEhfpw&;e2Q*Lz`YH_X+y{GYcdvrRTp80%w40M0r3&M#M5MAuLBJ!
zE2*XZpsC$azhoY#IIo*qo;64DQI9miZ$3^Le=_@e?p1)ZzHLp9fbx;75=hi;PVbOT
zeA1$LEm_ls_x9FIdXZ_VeVKOi8>bfV=Z&x=QbRsoTQnPirdW@|nd8i*wA0~6sd`Ok
zxl5fMuPo4v_dK|wX)5b-&o>>p1Z$IQ)veq6YQw%`lSSJxZ9G-!Bd3?f<?!}G<VnQI
z&{*HcQ+Eot^M<@Dxe~!5oyFFVXR7VGB`sMVO*)dg*KIntz3U^jbvCw1AcOmpwjmGT
zo!0`gFV>P^r9dh;=3#$-S5j9(SH-&qw0On5v9tndh5%6I2ERU_>6K@^dIi(GiUh-q
z2!phorqD+R^hOhj@lZj8dqiaf2BBTTq+>$b`fl3Wk}zRztbCSO^5w|Tg<fEMM_~Eh
z*oax{(~@YQOV3TYjBF+SkeC{e2??VICJSzXFJJX5ue6Wtdc$U{dWyV@Vlsdx6jsJ+
zk|r`TFbvRV6c_|pVi-i|6D;&2088?}&!1uGVBr7fcQ_cBKyw&`f3}f@en0(1KtE6K
z{No!g9qwN(5O34r|MiUU<*BLXXd5B)3)N0S-4O<cfcohJE2&6z1_L7mBPsex$rbh}
z1zGQ<@?3zNI|{j&ypR%8I$aPxB7T#a{3nH%FG*C6S|}-RI$4}6C`0HN&~APJ7{p%o
z{K)(fBoU42^9|y0vNJkr;~XthDake9_>G;tZj3#e0<_A$OJmJOLxeyM1B)UA1Bd+`
z=5HQrLWCSjh?#x;|L&jv`${{m(E9G@zZLMWHq*jzC=WGfnXE?t@6MoY;RJ#IwWo9f
zXjf(agkcmS|F0oG^@x4jfc)R*6#)b1&D?DE;?@6S;{P-5@9ks!|4lpC`To$ZEdPiZ
z_Wx_1f2&CZz{!8nG4^*MW~%S)y67y#A_4zx_T*=dSbr=3Un;erfJK04xK0qU{moMT
z-w1uuJpP;V$mP*d$<O9GLwh4A{@d(P9+u(%Pv~<!`e;bN__Pt|a0I;gZ?lJ*&YS=L
znN9>6+`m@kFNS#V#s|YK$>`wp?qc(YXF^l9d|WYo%r#Gv(Nu1uOKCC=0(k@xLR6F}
zu`!Kg5e<E%NYTp$>m-i-SqhoTNtw^2)iOIiiP6!t4Q_W)y7J`y{{D(e-6FMIItN=@
zRFaaCk=4^m0}^u4!^7VsCrKI!)S?=28V)Tw{0)MhEtQWN%J5^E{Ayz4Te8tXvEgt;
zB4ctG9|r1lV)L<guO>~?q|N8pa=w5HOM8`v$3{mZF-RN>i=^%gVByp;*mCr{q#5t$
zgn+Cr!R=)13~0j7j0!iYJztKy|91aVN5N37dF8HiLFLy&CG!;yF=B2B+|BHuPtbSr
zap4Y{TWvOJ)SDaScbMKqBzWWZ7q?xuwBDa2(eIMRZIjORrqmo(8hhP*w%t28PKk_s
zw{Kn$m6%95Fd!-lT%XQN$;_mrr=yE$et;Vo!Dx>C6R?)A55G`sFXHmsP7?v**F^OK
z<(Y}f3&0)JJhpdR-5E;yl(@#6#~s(022@+?a5ozdiTh4McP1r{T79KF&B$;p*083S
z#j$42AZH(EvD50qK~0<ci2DFHT)9RU)8<ObOnb;Zmk*g!$oCNN0tr{5oZjTe5$Saf
z>q$C_mOGr=!|9T~f&bjTLSAprE(H<UzCG|BayE6v(L1bWs8>3Y#4Q)u0%3R2eMD)`
zmNMa9T`wRdG*yqc%5YTM8tzv+9lNS*ClG-cRm*D5ua)<#4VoLK!gd8zfT(*lx`wNZ
z<H4k1Q)|upCa5mQ(5@j~{jdIWd4x!*(^ElnKh-UJ!0^sr;){M+OQ59kx|i@Uqf8^h
z5qIupxKwQy@wlmMd9PoM;I2mN@&i?iwG4Z}NC~edi_JG1?dU%QI^e3+xTjbR57baT
z?u;Z+#n1|V${2KGrN?}>=_Ovm5yy#dY8Tae%dm8VMH}Azgoncpk`|hSnJ|qNsJdFn
z87mj%#u)l)WI}@t!~=g(KEC~CAbtwlLU@%bK_#*-mtEAM_Lu$i&qr3=aYQH1oz{SD
zg)y%e&}9~~rbr)S@^ap1mrgSdwrET*4KSG~|LA%`d!Ss{YQN8)4*ao-=Wa_CLUBer
z7_0aC`Q{)(H8V>+pMpn!G#~c54(WO7YMKD<et6O}K2c!kDk07+H^Xh#)V8{7@;8VG
zA?AA%;nnK62Nli>A3>Z_IK$^YEylN9lO1jYqM*&DCSUh@nGOvWg$ff^`-N7TzQsvG
z+=nc@fl<tlrX(iDG|QQV!QBO|VPb@;vmU}DD{nbRpw2Y63<mr>@C|;_6IrFXvr{Q8
znXnxD{z2u}!ON$B9q}Gv8L9yDdn>7zOV#GlR`YuZ5Tsx>rrt<$1Y#4!6Vw**E1d#!
zk@d2DMxn>&I&i@LeecnV^HQ|YDv8EADK93AM(imhP!Yg-1?*S-Q}oclK$XbA`_;vQ
zf{&(n+l^F%AFfizJw!%U7&&=rrxTBoTmL4G(2@C&5h93|^Vbz6mR)^?fD}wh?nQP?
zS)yj#4s*=DD%>Hdecs>7HZo6Km00AodaIpS6Mw?&!kC4$09{~hn)I<sb?@C!fhEkt
zm{jb%#3Vx3Q4=GoURPW=G+z{obYQ?L?Y;SUy2)B@QDPgCuxvd1-+1H=F};(hc7qtc
znqQ}??)W!rI=BAC!`dWe8K14hX<t-hfrqv3l8!I^YLO|f#IwaH5e@`kflQ8LwW1#k
z>s&3|H9XU*g6(X6QV1X{xBNMYy{*=$P^p(<><!a+u25o%aw8d`b=wsVfd1Rw(MNd!
z%?s<W<Zv|b*jNgkw<$~9)Z%q<B<nam0=c{+!gAFx|B{6bRYxj1I{cV==OscyIpLS~
z6_OJj)RB_aw+3gbEVk~x&oF~hNc(+q`wu`E36*joscHjFj~6%6$^nA!FGOhL1Hykz
z0F;^8?#RXSYM%qARGjix$CKcK0al0jSt_sl=TIaj`un@j1#z#Dy>z=Srek&i>rBGl
z2WsD>p&(M;8>QYf!J~F2jeS>|0-Fqq_RjxEFg{m-hPaU732h?v#&6s(oK2s9jQs;#
z_qq$!{C~PA=vZ-JuVc!Fe>9RN*)Y=DEOp_6oJQ81K|2qR%8BW}B>QGtwryf<LXn_>
zYYzbcq3#<xs<zCn(@k<AU{<#YT+3AsUGyaWNeZ`3358hiv|c`+)-(oOl?qrP1COUd
z8%u891tWjHEceG_lJ}z=1LM`<JoYWAb}e=g3oQPPu9tPqEh6AJV~t*Bo8z(r+Uuei
zbYD7avwTK>=I?^geeM=Q2UmXSE<ZY5uc-{rL2K%5mj^&@9i5IgO2bQFp8x_mc8U;{
zy{=R^d#1PP=76qh^IaAmc^^*y&`|u*5k{qrS1q`hW%b}Zj+s0D1a!O|QYoVc)h?>1
zJrVIUn3rCNQaq}uTbc{r0My5~hJZ(v*0Bihh7K2WS1TG?7oZw8;J;Mv4mTNf)b)SR
z4i&k~uP?iaK0g7XKh!PPQW>yoVA(zfnOS_azyC?LmF6Ex&CMd`;NU<_%dP6*c-btM
zDmgYZR@aw<!NtR)ezRig0)-vHYE}<^z-^!)B;>d1YPWUW!WX0~zKrG>I$dcPpEOO_
z=b;x1^`$jK;kbHWC}=Ur{cLS{`$buc-6??f{_tj0oh8^?xV>m=sbFv3Q>Gl<2Mx#7
zbi<?4?A$=J4tU;@tz5*+I4vuqtavVvNJl0>(<X6>0MYfdzw0BOSd&5Wo16L84DKqR
z>OD`pgv6PzdFmE)aP$wy6c;1yxacQXt<*Yo*Y1pJl844Kg+2}0EE;9oI<GZOe7&qp
zHTCkjm2?ZWGv^+5p51mG$MlC{d#{CnP0qD}NZ;}ARNG#PjJ*u@Z=cK=+VaH`j8qo{
zGp%_6TZ*-tWE;QjQ>E}7GpsZnGp^hm5=*gD7`LXf$*h9bvm&a^kORrw%U?U#=~Zl-
z8C!h4N6*rYd$l%NEwbtuzG9iuUbYIn52hIgmHFdcD)+nWe7wI_LwvA*_txt09#v}R
zJuJ%6Vnxd^$kHb}FE2?h(a0oQ%ETmqfAl>F9imR<J@VN1>+1W7$IwaBF(pP-Www47
zlF3qg9OIaIe%@m{#LFrcpHX$F8z@i%(gq$-Hho*3I2nUn%iZlXsXMogo%81mn1mfc
z_EcA+GBZPDxR#J^^`Ys3<J<p)GSHnY|3j$WwG46EV=ck`h;L%e?I?aHp$re6G|9^H
zFn|6ujn2+`cIFr!D=en7Q`p1ZmqkFAnA3_V{wo*~NVc8)iE!uwpHX-bbXCP7D)`8K
zVc&2R6}nZ(h$N0tg`-$;D};D5!4ei0qfBWX8pq7YeCEx*$(Qipw!XS(2wd6fe~ZaD
zfrVOuM+ERFQbY%y@U86}Qhi?vSNbIOfV#OPHb}_~i^`-~t7S5FH}&hN9SjWS6~<WG
zwCV~7V|DxEvY6vrJ>T3=L{O>hqeF{5P0TrbaWOfm*m9M<Ma8lr^T8GMNGTDTI_BU^
zKJ&Ul=_fdL6fR2BE}hMMQusI5c#`u#xfgzcmlJ0n4KW*gYIhe4v6ceK+&cM2d`y!G
z2z|U4uU<J^6u`GQJLNg%X=i`(MPl=~v8@!~TiOZpyh2?lr_i)%y6%mr%pomStEj9#
zs^9hEzC?DrI?uCiGJWv@-$(GoPlV*i<;&NmbQ`n6pj<pv!Q?N;JgKJ57i+)IJ~beH
zybRP86*N}g{EU%>_MibAFx8=-=$@Nu#RK~-&Ck&W(AsNBLLEvhf`rqv04b#XHvP?w
zg>3qq;cm&Ppep`R%ALR0?!c+RX_NDV@ruAw2#zbHhgC-mqw8|MA34W(h=8jkJzdDA
z)#-Qh#lhjt?To!`{|L7>ZGi&8&=oPms`|~Ex6b68t|t{nSUuqgM>Voa6H0zi*&mm0
zpR`T8r(CEkPHR8;4l|H=Dk_e{uQ^qAIWGmd?~7L{7tQt`a9Ud5m?Uw=jU;i0sVLQ&
zFVR@Md1Z_j1i&9Gl|p?s0cZ#zhsEESF~G>$?7Yq|Kh+hTBXe2PpJa=qLDvE(+YmyU
zBnaL;%sE$d-7LBNaJ>n@h>FI;=1a1wO~}qg+KQ4%;u;uKF0#6tj!`n7G-7XR>kU%5
zge2OwY%NvWP%qUisc>M{tUWk^_Zd}+D_KjG-&?L0CkQg`cpX2s0DW06#QE=nxVfxF
z`@DBaMLzoIw%mC?oSQB<hQDLIK40@5YH?Vys;a3YJ0-=ys$@0ID%K@)T@qSa(<|)>
ziS^+|sU$W^;Kqf^+`hj)gnHvK%nb8r8XOT!F~-d{0?1+h+|6#9X;P19mC<R-O^~}g
zB2Dw;nm0>?H$>e`ba1C~<>uG`fG%>#tkK+^yy<Q}k?C*Mrri`_iC;6tFiIW(z=tpO
z*=hQALKpt=gVGgXYy{(>RFt^aR989@D0ts#a?Mu|f7D{SE74<>*n5WMU0b46qX7VU
zrB<6)<$y(tK7!^1jjrWjeY`==no@l!E<aVKbLA2QzB(@hG#4_rcXHz1R#8D~6H(F<
z|BU#5@gCEfaDY>ACHV(fsKY3-TeFuT{}w(wg9-<(cQCF|Rlf6v@Tl6by*d7x=Y==f
zqri_9uP<U7m(g`{gK(LbMno@8=H$oBy%to(kI6H5QfD*sMF8ZQx%{{;;dbJ`o`jyb
z&<e(Mm?Nzl0DzD-b&80ZA3upNHbTh6fxHuZ4|jq4__!XbB|6Ne7_aq~7W4y;DQhz*
z+d>9-aWZyhVakz>AJiI|);n}#H(D|E`;T1`<yme0e(r?Tt>S?5wOt_%4wqqb^I#uS
z-F0RFXy?$-D6Iupxep<ON5YIsSVJbb@U(njOf@3|kkU5us34^XssY$GS~WsPxZ_XT
zz=LZsb%OgXSX;9cT~fu4c6T|AwH?&d-Eyk-gVXzL52c`H0j)+GkF;^AfEDK|4oiZS
ztOecH#v4wW$0Cys?K+_{S-8E-Ccoo?Qbz1FK<k}z+)p<rVd=0i55uTE+Qby6d&$bV
zHyL;^je0ibPLws*UEx(?g2oB5AIQ(*#R`lB;JhVivj4Is|9m{)hoxPNM-t`-ga5cB
zN06nStRmRiB6UwXn}g$M<p|S{y?e>4BDSJ%X9RA~@b17s!JL}&&Tykf*X*i?2hWD1
zki)G_Ioc*FTTmaFqxHv!F+nf6s}UH^>&y}c%erGOhrZ*wn8k-5h>xxttpv*@m!&yz
zKvN6t8Wl|&{W|xP%Uza;$2gYw6lV+DmU-*X)ZVvuDwPF_MIg|GW5ZTu(Ob0BKVj0Y
z9#PF#7e#*tZq(+K`z7S!rC1uuMO$Urt`5Cbg1^P5A49I^*4S*0(4*)me)SUTdne!B
z%J>7o*sXnX)aO7>QB~zds9On<^wYgrL$f8S2)SX_<N_X)<MBc5c3xcp_ZftNfhF^q
zY9l(~Lvuem6mFCqcG+mPQJ#1C<N1xPecPV{x6k(@-sJWJItw}uaK`en9o#e&l_Ryi
z0u-pGtiN<%DX~nUleceE=n9xDnp3Ded`D`0e`PDi-<*=dzGskdbz?bMZ*RC@5Xx+{
z6$G-HHCENx4e5{2v^gUvVq&gsQO!u^b01xMJj>E@zpK$1IV^1uW@BS(-kfyivKc<F
zgywwAwz96`Cqw|<AY>2dzGOrH^M*lH%ejJZw7*E{SIQI!$Uc{u<asthyGrs>8@M3F
zp<UtyA|0C0uoAoGJrFP3%m{o##pg7y5E4Qu*ZaZ#u^ql~sDmG~95|fqB6xL@2YTbF
z=FAH0(9b0`<<;j{|LM4r;T4)Gce;6`LipL)di4IBw!!Qy_Tp$NJn_Shxz3&I(Fd%D
z$wkFe&o=NzXkVEdgloF!Q3<bZ&GyeV*zJfG(`U7hl>XSojF2sTe+@$Z2x2(nn!*Wu
zL*8^+D<uIbP6Nb-3xL_EM!d^4^HRhFEl9X*IHHXm9~M?2jAbo`fI7J7Z1FeF>G5w`
zry8bYql^iV`xs=QKz-vVAN3hd&v!&9b^+HKAl~z-wi%6$<v3Xxuo-Y%AE|z=SaZ9x
z<Ri|~6$2b@+loFGR@<5kUv)N`31ThtY2BwdM9pm*_ktR2+k7$&ESYjM&p@ZG^ar2H
za`hc#@0!ESZizTk=v+NEJXF$W=g%3A3hxjhI(;`UF@Sh-PcX-5{>;NF`cE`I!-+2M
zhxe^&_unir%@*VY5||FG-`KQ4xWup6ZC5>C-JOA1Syo9_f0d%9r%_AAyTLumk}i5I
zcMkO()<dzwRt}+f&lSy&0F+7jPot&Hp5?D+4yo*Ic|LwHFbZmaoX5*LT703QBl21%
z`7Ch{F3X~i%;o26`&)`tFx%bb<G|e4+<^{cF+xG>9tTMMb}cpQ3=~E&+6%Ynj?aDL
zv=*Pb#w(EV2LI;GmaoCP-H+!nkqK?*r7XQ{+hO_E#0D%wjEYV=Bt0DM^fqkru09%!
zEiyzyR?_HCKP+n$s;m`WS53v|M1X`uWxaWxxM1Rz8?^GUkjSldPtO^G^CikMo84FY
z3`$dF(aiW=j%pz{hUR6I#()IUpLd>HID!5_phX=MliZ)~7TMp9oh774)_6RFXiu8Z
z0$(qSAf>R&r@Glx%OvuHxZ{<rmuf8GLm>e!t`=?4!;LT=fwb}N&6Bi92OG^6zAU_B
zez*HnjkhOx(QVlHxid7apjHm+MbQ&tKoWl`v*1aSsjKt`1hN8*Mjz#)-KxGi{z*$l
zO40XvZ7$dHwKqm^rsbP_FTP8UrWvxgZ|axufZCa>Eby{?Vmfa?i8A;5dAb`6XCQy~
zIn(MAJRtZEQeC-t5r19R0ocI)&39@$A;*zqiU0t2cYUL@>^p{!PBX3n9@;vNz%B6@
z<G&UB3`J<HKQf;m*QB>%pLKh_9p+;JVB}?y5t<6B5VV!v@flSL9g2Mkt!cT4N#Z;-
zEvxs2)nb`A%I9uz*We4c^47Tp0KqpHa+Q71%;XUDL6Eqkw%K{6hI;^X>aAo^W3KB>
zV9%;uQeEPfwVo}|w{$<GS{$PbxSI*caC2HF$t3fdN-zEmF1=JP`mx&Dczfi>ty6K_
z^uUgM6W?^|CI80z;#E>Mg=xUDLrV(N1cG+>`mKNuHeadi?Tr>Yg~Nz)yg#k#P(`c%
zg#9ii>-EfLJD8fDzCd(Ab_WBdZ6pO4k9s|&b9IdT==IWi#kW)&c!2<&<4d`(6JDLS
z&J+~R2IjhlJhtFfuU(H+u!ibor^r?JB#p5}YWuA9lEkHnj4%>3WmK$6?fsH8n|mEM
zSt(|=>=BmI<~mRr!CUd#C_66!b-cuoo?!^0X}<bW?e$l6v{)IapRoEI^SpjH%?QV^
zRkYjUz5o(AqC+6$)@`}yx-`IloC(!62*3HNf71qrG^y5UH|?e7@SRRmi2lB2xN}w_
zvYD}v(*0uR(nsoYhRLxk_eo#Bp>wuavu+gqaD($@?N>D=qBz`--!g2pLT(;=N$|E3
z58=_zTi@RsH}#(kc!BCE73vst+Rl!mxx7KgOHk7<oeZ1PwVr)*J8qb(a|AkS$r~AE
z>1Vmr;<B1wPb|DCabM5N&hDPJMaD}Lj4IeG+vVF_O5oDmJSl1y^l*Mp&>%~^PeK`K
z1v*}I8-OCK&Er1Dyyn$whILTa*=-BZg%i%(Z7FoOVO*sL^|iAd_k!{64F!fLWc!=j
zY1aVP@pl`Kp;ZrO%z;N}rpN{<0g6l-@Qh3Pazj0KTxLu#J2Y;7fY0H12FZoA$<sGl
z+3_!DlIzPf!5==Y`L9p#j)K-^!m6+9Na+Q2%=xU^Qm*OdZlJ)N+oB4J|F|tuhsNzV
z8}*DF94e(QYW-<ML)X6Cdx9LLPU;pNq(B6p29ou@SbRCcl`X_6cPfDm>pT5;tjqnw
zn+eYQ(qu%~U=UdxyU``$+{OjH7XVrTrnZ-b-WAqsCvw`pIbhY&SV$aZWW%GLC?1dB
zrFo(F)fW%>&h2iZFo5ZR#2S~}Vw+)59=0tF=gz4@oA!in%&QyB@Iwb<r1kwyn4*W<
z@ra9<;CZ=q-RhS{mnFvB++1m}ECfP(&(4$knQjXw2&{`-hQ{}-AO@`}Bg1j&w4zpI
z<x;i+mVY!Pnp`F!2%qhp@9Vyp3u88Q$$UUbdmp4&JUP(hvLN#Ww3bF+RIQ&Zn5ODv
zj%+iks8KYFe)jVd3nh?Rs@06`zuO#h(@Aun;_H}`Vlou(+Fe%o{K~sFphLZq30v4O
z?}l%XK(qSF)B~FGFD&8Ra?W}%)t;@1ReIhzXnt#QKF$n4yr_#!OWP$q$R3>2jd^3!
z&#2Xr5=Qd2?_~B&^t_tK5rVDVM6s;2oOTChVsXn$fqOFvtsN{#X*L@LTg{g%R91d@
z_lDQS<_s`l#~*Bv_sWONS4a$%iTluH7;+u{yRb~7iYwRQWk?t_owiD*Lwp{V;}MeQ
zT2q_x?wpD?THQDPr!KT$&GdblvV~Vcer9%okWrh^kALU)+nhO8akrJ7-Q6MTc<oKn
z_?pXwt;d0Oc3DP+qfI&ZKkqn12(y?+l&(sOWn6c)U)}BD)h4hbB@;VEE(t;&rVhV|
zOm!QS{qE5vb33Bfyz0Chi%qg+8Ze#8mAUAN*Yy~&u2CM4#A2yR=37{Z>;iIF<NkbG
z8jby@*8Q58#Yo;&M=V_M?#QvBpXq(Et?2ED|Bk(RBhrt=cWjMd=i_G7to^lNY;7g(
zWe9NQC5t3&M&Z;n-{Z`jN@&1}g2;54?ETEt+*6HaiOBZejLWMR>!L?>%KBzP{0&u#
z3x=$;|J;J-<vgv|m5a-)Tw4Hlhdn1Y^e7yu{~PRc%~>NPe?RRdeG?}>)#dMn|B&L+
zTC`hBJoUKq&U9W9R_0Rfbv@hp>PZz)CzBGI&#E>(A+$SsT}f3uK`;MvF0Z-n$hqHl
z6^kPx0RyQXCpH_AE?0%k*4;OAeQn#Mr;SqY$E2=~r5|@R6<C0S^XGk}$EXj)G`sn!
zfuZkE4(oW@9x3Gb!Oij2KphWPP-B)M;P4olWAwJN^)tC23irI8J2aD9_rBOS-Ts{w
z9>ed9g`aWSaGHV53Wfq@|7z`3-{IjGWrN~m-o&gg6?aFQLXsq28IIm<MY~IBXe5^8
zSz_*Www-fbPqsb!JmP6PwkNcnX8<GF!+K8BVCIMHX2&HgRJ9$WJ+B}L<_nen{Ig{4
zZPM?TQ=Cq!i<U|*N_wGK-DTN183z=&wi-eNQkHFV=e_;Yw{S6$ryIMtQ27XFyJwbB
zFx3{HW*UFu<h`Yj93)E|W30{V60`Y>{;pRl+Z}HW6f^K$LXY8hK)dcIJEiB<_-SLf
zw#4y|D&@AOofp4dL8Hmzr@x#I{k&ER7smM;BTB~}+-Cl?lxcIDMbM84Ui!UuIttG0
zEMBaiTP-K^&;wLMGum3N)zY=$+fnM+M*haR;8{jZLS?Yc4scbtNJEkqPxb?39`~B>
zq&fQGUQ-Wt_R#h<rLJ584XLzQO6a=}s*_VwQZX!ypFlTqS}%w`^tUFA=Sb0-|Bj_@
z8-~c}aHZY?rdpE=ZoQ!ZV4;g8jN2`Vt7NNtwQtFg)Dm>V=W0p#dMz(4+2tf5Gj*p(
zr6f*{)|(<)5Nj1{(WV;vQ)ydet)}hDEMul6>1&7E3TJ~14<NW(=FL%~)cBAZt(EMD
zbEU7jc?mD*V!{PdwMv7D%c_W+zhS_qsEu<Xb0Cb<rb%O}i{9pv2(Q`IAB0F5Rk}@y
zoN|L-yM43eS(J{lT5)Df&_7}2mdwxw3S`75=<O4C(J|;672{l_<ddx?x?dIbhtB}u
z7*$Imy$RwxyQmivO1Lr{uD&LkCsq7*KOF;Eeqs`%=EJmQ>zP*Ir}t~um3zmL_dFk<
z!-B=Xm^oZ^+X9IJy!mW0v}L~P>-Ko8@J?|}Q*Jk~DL1S*3^8xTvW^Xo)M29IY}Sl*
zVe(4^U^Ymec%S`Y3rxp@MLudv+cT(Pumg$Nlq~1#i@ds?&T>|yXs&?dr7+d4axF^%
zpe1eR$6_CN2C6g5?TIl#W<p)}Sm~6TFl)<653(RYr%d1VWN4Ns|9T4Z3p^M?CbhlF
zBj+nexq)`acW0syg|Swms8dW1zwouDm7I8&Cf&J~W6{vw64kF*x6qBFfrf_0@AEh=
zf>nEdHq)gmBL)DP1iU%y_Df8Xs6dW=V4e5dbvp`Hi1RJ=z_tOd@=JHF{WNP*CULut
z-OT|8=Pd-o;NR&IvvV~UzMkHMaodA5QNCN$%CzIbdq};HEaVM;LSN#8ag_&_)Y`7`
zl0HGV7993@m`g(x{9nAYK8v~5wOfNo>!NE-ew1-6*(|LiXqj(wu+Y-h6;pEQ5;FB|
zh{N~@0+`HEpC#Dxc_{g6-i}l3&GR6@xsJd0ZfqGX>6=pVDW!)wHVu|V1fRGfi?8Rv
zkvZKsXt9fqxQ!orCb5`Mzs~?Ex)Y-x2{i#OJ2jo#tSNjJxDX8@bg2V34wi81vS`+r
zlU5Y=4ycaZvzGltS5M|N_(O!JSv{|%<F>EvHvbcvr3+qlyQ;SI>?DQ$I^0PrWo9k)
zbw6Su*7$5VU|TgZI|=N@)8cHJ@%LK;7#1%Z(TF|#G18();ix~F3_T)2c~`==^f`=p
zU_L%AnozvWKm@sfuua1OnlTm!bhuspK^_(zIlh1(lLem5y%!b<T&DA7!3dZ<*A;)v
z>@(l@DHqV`(&v+u>=3{bJi;cqH`^<35~0D2IaXGE5eLajeXYW0mv5`qEO0{YX4d_d
z0-9$O9o+*3d#2mE(OF+sM556Y#sfIyysQ=d-N5zWvx`4{wR(PJrOB3z!Z=Zd3GF|;
z2?iLd2zoG=kZ1jZW?SOy;k=8OHDlBP$BoyS*0xnnBum+)9T{Xb&~ri+An>p;FxmOS
zzELwb%*!iN8nok~DZ6xr4uPz#u#sebr;adsCvo-#2JzWK!ua^nj9-?<I(y=k+Nc!R
z>%0fs&@aABBFtHn5^SC_+h{=p5j*NKYb2c2F55dOR(`BG4Ld3CT-qK|?LX4>uDWJt
zV>8m-q<>?Bh=^D@20-?p@Prl^nXEOCKWglbf}}77j)*kH_;1@1co$BJh-%Cx4AkoY
zE%Zwoi5v?|%?tT`Zf<T-w)eyDGu#xvZtl7igaR%0QkIvsNH*8Vfc%SkaqqNEHoM|!
z^36R#ZjO(Y){I45Cb)*ccPV0SD*gr1Ie72fZ4a^~sHWJwS#JAchH$Qb&!PS(<kIeT
z*{?Lt48^x?5R*EWBaWE`x;ut5=`<@j3_F?6RtHyG)EH}(d1SY8z8+)wya!Cuy*hRI
zqW4EBtgVdTB#f+}wT*d2aHbkZ&b|%kTVzE@wmSW>xyXT(MFTg81$u&Mx8m!-DX;ZG
z=U6HGG-TH8?2oAtV1DTXzBOaBPLuWG;7mML<$y>)`1?PN2Ixz8tFr2SVI-U}O*U<{
z#RjO+X>>rz2MDz!QS}Rizge3~6b>0%Dn>`C@Rp-RJA0rqbcpWffq`cV`fH~vx;pl{
zOLG-HT@GQtOV9UHR_sB{Y-d+f&E<3q>xaUX?*!U<Q=yfwB0Te=ANB4Hs@)tr0I`je
z_!(6d6<^%YKXJ-^EgkY2JlDym<ND$t`Vy_?i)mk-Ia#`q>YuGX3MsSGL6`URW&2wq
zpqulCG{a+1wE^s!u8x=o?Os#K&|1xW)*7jJ;pxS^a5{M#3<~OJsKe-Zhtl=Dl&pm8
z<BYM?&2-juE{ayQDGH#yYN~83392u<{$_Rl9wWi=LbAnr+0$BXbA8QY4Q1cp<M{4K
z63e@tro+R|%TfC#D+;LB`7j}%wXDkG8LyfrTmL3?7}57W$%(P>vnR04n&b8dJ5^hV
zD5RPr`ygbwQN9R;^Wb^q)Pl$7-N@q5aIDW6Cq93y-58)Qy%Y@u@pm3tYW>BmlN-91
zo0BuZSk`KjG<|yljnxE`&AnpY4-<h~5xyd8@+kD(um~1B&`5N>i8Y5e_MWfiwOk1q
zAS)nJM571PUCUsou<8mF_b|p2T4zbZp?*$_arz|$jB6TLLT{t1BlM=ij{FFE8yiYp
z;?`@DX&4%BU8`=`06fv31ZLwIEmps}$P1<|%W&gpfmV}@;(zf;&6Rot_s(P0?TjQ2
zxb3EzM<;P&bTebIbg?2>koYt*9jJ2M0IZwheiFicfu8b$p7FoP<>gS5$8X5p04Lpk
zydI=qmFERKPO0zd!#OQym7s+I5w}8lf`6E#d<jBSQ4KfmbNnQfu81&vpE(=CrCj(*
zQ=NYs%euy8n<V5bI&Yk-`)QxK8V5WqiW0>?SPcwEA-NRtv`-AifuTXkCxY2uGkGCx
z>Tx>u^s}-hB4qzHA>I=0F+c1(<>k2wJt(ULOu3$Fe6kRSVw#rx+I5rwAD_C@jz|O!
z*A5uaq-x<1<hNHKd4GlR3~|%Z`l!e#8m;ugJ(1WFje2%1ThI=dxU5L)em`vUP=_^?
z>ARgRvtV%!3a6i7kDDeW3M}FFC@Lyu=jRJS!!$KN>Ej-h5f#jV*}^^j#L5XB(}m;o
zBDlhO)>%F+g0opHr8g6O1Wzcs(Jt3HNTUp{36T2$%o^6X4jdlSp(?tZm72)CbuXu(
zrIwny`aD^xPqIH*%-;#%Z?NT&K@okMKE}rBRZAU`lau=e0PxvLkaKs}$b1oMPlXNp
z&mk1u;Y@SD-UmKA>YS#8YqE}}A$y2mWPfCcz#FVHv6433_k;XK`}#z)6QV}ltCE&U
zvko?R^?7=&oK+l39mrPb`Um+)2;pi~e##>!H_S!a;vQ{W@Od@xHA1741fKR!huO8e
zN;_k6Q6ZkfuP>h@cS4o-`-y**3#H&xKzWXQG0V-KB)PN-y*(dreZL3nPW+R4#D)M#
z%$8gGx8;ADiH`$r{};KaRBbk^5=nN)A@!t<r#85O-#r5M+5czIk|w<rMW1p>Wp4t)
z{>hhG=Fb2Ituk_2qdahWA?}g=yIDOQe#^pn-Jqh#kqr;Vt`Q)1YQ|KdhCGqA@)h9j
zk0wojCwk=0?0xU#wVfYFec*{wJt`qp@#}Jz5IXfg!r-|XUg1}qB7)M1|6U`>2#Yn+
zk*y%4?mUGw>-Xq?;;LZ(A^y|9lhQ)-8gD_Cv&OWGXGC%sICfB`8%c0^d#M35Tg*uE
zv3btLilwky)mNWvRW2OX)}X^`3B6T%6*moKkQK&%{|77Yy;mG{q4jKiMs`llld%+Q
zw7gRY9H=i0{=N$D&awy9&F4%aIPXj?)ZXXmD0FjJ{<5TlvXKT<9WS<gBo|yS;-krW
zC}4aJo;Zm2#gnFS0!>%1;`U~%CErTQ%0`9pGz2{a11bvbJQ+?7|3TLaHj+&GNj>?U
z2>qG8z`#}D-QK0az2XI=6ff0FI?uC-jq-6wIom`Nd%(x|JuMFg0>?iYg80C|8*8SN
z+EUUnhOnkqhieUbHujeXEi2uwwt0DZ%J7_x`Q~u$$DB;z?k7d{|4XB^ynmPBjUtr3
zimL~xS=Diuef_^i(4Er$@QAkkX%ipR7uEiwN!)25!VuUWQ9FXiQtGrUmx=_PtOadE
zOw3EiG;~#h!R&;F)gO&eIwU_3kM#s#P8$oQ^A^n&*9n)!S^kbS7r>z6jNP>|U$(H+
z`SdA6UD_oJ8bD-^@^=B5-77#xv31~--jfju;bRN+m%-(8{3VaUj%;U7vxzug+j4c&
zu9b}Dua$~T(a8I9m`guvIs1X!;K|+#{n}|HD{b+Zb+qXm2?+a#+t7@lyG=*8G2Mo;
zZyAWBF*vHRInpBFk)=x6<$7Uilr7bW^IiE!PL_tU5jl_G_$5AmaLBz|?^i)?ZaIgr
z-BxyOdG1JqlP3EHX1KPu9Z#NJC>fRfsP7fs$X^_Bx2RCd-P#lFcw`X&$3?H-Phn_7
zDmeVq7cWZkM<M(pnqpoz1I11{pvZEwME6Tm)7}1nunyE2Gh@lJ7KBRT-*p_sCN$Y|
zt!WH!ZD~lLr};2QlCNe8==L+Xeaino?&$%mPuShRGp<YjhjEP!K0N2|C=ek(K@hI-
zRA6mWFo~+Rh@#}H{t~$|tSP#}n=Pf{uGUeHm2E<-gh5VAyMA1f>8-Dy5jLgEo|Q2-
zh&_ILPft$|rFGS@v9V8$fVKEepj6{QEw_UwBHMce1Dxv`X?^|3yTDb?;nsJ<S!=~=
z+sT^Gc*qubu#9fRf4(A47&Xaob(%Ry8<<(+NLx-_>lH(5JpZ7_%cH{O1cMUu&X>?a
z{J~$oi6}oWtP10e&tDdN-kbYym4kYhvrT=4%Ju<;iVj+dMQC~d0v!%p;v-z+I2zKk
zkn&x~6Lk-+=84G13deM)#cDJC<C)k03(A>(l1bbnxBJYy{zbG;2oBWe5OG^T*Z%sa
zka@+;;VU#*La0^zyJ&1xIM3Z*jIzW<rqJsYhYN_i-=MqY0&g8F?1DWD>Dg{i|A=sV
zqWwlA#$V{^8b<`8$CJb_*y~t7aU~@FLpYSKs)P#L>Z^*XLR_->B>xt1=L#AHLZI?P
zL#nc9p3A31xP7OCV=-MY_SF7zW_KAno_vH`UrGjIw&yrY`OEjyE#;8--a{g&R(PF1
z@sL8j#03H+n4-e`>-JI(;<KfK4i`7*Dur*-<uZy>5(;Ghbz|igMf4FK&VMa1So_&a
zLa$LpA)};VKKIv6x%4R04$w29ng=*GPiSuksAtGxGwiv@|M^z_+LN-{$^=by?V;`Q
zO~qxP&41wpUG<4(&mkPnn>+IU=Yl}<HFE5IKG*Ei15{QuBv>B=itp`d<b*%;p(YS4
zPJR|Bw$9nn+1X!)PP3Z-3VLe+s^z~}hfumR4hmp*l`2vcvgpM1lK^f;3VNvG?lwgN
zWhJL0bn?SJYycD{Xy_Rjq{k!Knf~qNC&VUnOq0IWaNXq6{sns1e{I1MxWML)(j5<7
zZcn#KDDM$^M<oUv*hxdr#8l{S`gb=~g5}81QkGC3BKtINT!8)868=Kkv``M{T-p0w
zWLT7#B%ypDAkvM0-1Y48mtF&p;&YNBHaaYL)Vj@skY~3sO6Q@XTI3bVv^Z3Y*XM<f
zqtH>lLV(kOmz~0DrBVCLS@}G&V&`MS0#}y22np#|vf6}Gpqtp~$z(~J^OJF*)K5O2
zGu4M5-V$acK7ur7tKliM)X#<BQ(!-wkiNVTY0FBo-T#A%zaPYlop1PLn)8U;-=L;E
zh>PNWdPs(OUkC3kq!dSg{ai>6-kCQt*1_$kingo8E#Y*?(><!Txm5N*p%z-~cgx6+
z-{m<@!cpE{5H^6_!(YHWw+RLc?Aw#$R_~?~-W7jmLWs5iNBmoqX~8FboF+Q<LiO<z
zVVMj^-&?9%Ey>`S09cgnFYR7H=d~Wg5^)}sX_u!L!I<k0(9x3lw5Q=g1Zf^Xtv`;Q
z(}2mbP2nJ?>m6RS;Jya-L|-9NiU{}f@Ax>f)C0MQI&KrXneFxiKN3q>3%Ml~qfTfM
zh1G9E168^Eas$Eukuk|ugM_h@fb_ogB5of@W;kuNXVkC0hg}P?B0{UR>AFo$&XBQz
zl%&$FlDvGZ__GozJ1~A)>Rj33+V;|MZvU=zlS>!fn>Ny~dU`&1KhN0hFgsG4H!?65
zFQ>AjXbmPqiUQBdz03pH=hAt&1~M|FJi*A7)b&Ckvo3dyWhSNJX;qT;CCIfW_tXo{
zTE}LZB}b*<)~(%G$1FcR%mO<=bGb>WnOlcMY{XPUCNfb*p|kXp&Py*B^z%6O=c#2h
ztQ2mV)z^VJ-7u0tk>6Lj{dX1=j0D`V4!m;BsZUq-Fb-O9>)SinN?Ui_juU*RtE=@W
zZvJyyK7mnOI1(-^4u?pE(Pp{yq$u8yxPmB(aXJJGQXeIdlwkmAuHpSJ<zF}8-uw$=
ze4DS5*JCm#gbd}Jfj(B6#cJtJ+xX#J%F7AXoLhAh@dd_V4fcq?uJTIoKUO13Cvv}A
zs!(Me!0ew!qc<(|=<CQmO0{ft)w0>3q#ysz!>u`yniU5OI2}wnNBn}Y=p?RFwR+1b
zCRqB-zNsg$XXUqj62Vow0~cuLqmy)pwcd-HD{IRh{)8|c?h=xD0$}cb?itNg!10QW
zwSnHpQbWgV0$yJJS$B=9`Od+^555bz*-h_CJ>O5|&et9Z4QKqysc*KN)owHTI`%6^
z6p*gJ<j1dCmSrH@vrB5Wx>{E1cVoS9@O!yDdgRykSFN#}q>@~!1W#JHU7Ka9q2b*&
zYO#w;;2jP3+^-6+KBa{+@y4qUY$C+m!Av(zhr7L<wN`KpdxQipiS^sYQ9$n8-E&v7
zXo+tS1SkeN;Ew(Rdj91ep_Y74`Y|1*A)tgjOHoGN*HPUHx&frf_cI|wS@pR^&Bg$a
zFUoaS@zO0OD=g~r4?l;w)pQQ2X{j}_v_K>p%K`Wca6MPti#N7`>%8Z+J_6Qv8>3pB
zjN^$MjO&R2;ku5<SdCTgH(7jvZq7|yaJwYf%U+wTzLTs?H%=twdLj71D;Y!YDllH2
zr5OfOa((K<*-1D-YxKGO#VqIYx|K6w)v1ow2KQ+8<+sgi>k?$|bNdgbGAd&SdiGwT
z$jC3Hu~>GI@af8q3wiWcPLjwVkfVGle%2D@N>UjN4LwoV)ed{4OL&hC``;I(YPYrZ
zieq?e+w}nT9QW(^WPi-l3dkGAL%6pGgf9tVu&<*>NQB%i7x^g44S&JUeIPLCI3%JY
zZ9ehwV16I0w~ASAdy23qBi9!#NKhN|yyg=VD5Rs)%3QiO6FqY`Y4)KJ128rR9w+d1
z%h-SYDZLd(qS34!?YaSTir$x=+Ru>Ih*xn(O-z%qGjWXrBSGK0Ly386?W5{y+YCgy
zH4Z(<ssr)^UklmCI-f}O%iYGgYmIgr7lklCGnxDi+<jRb3)kZ<6S9jysZflynWN1Q
ziuvYWl;ob`*v(#>HCZ*A_LZ*R;~!yn^i{%5zIMDxBK437xkK_h+Za+<*6DYe2+<K<
z&;Ce*Go~<vTzuqULRBlL+SPUq)YOt4BH*@l^y>3*_C#Hb;hQdblPG-Bh*;ZkSsOnV
zhkmskTh;tyyBRhKRxQ=$w~by8cpfqwzVV0Pv;o$<lq9@=z|>?Q*whFSH|ya_caBc%
zK&Dcs@8JjBz!=NfsQTR=E%_00fd}GX10LYC(a<GL-DmCf{F-u$+fRzKcsw>}M)vP<
z90raR8wKa~ny(%VDO>CD?F7`@-~wKYaKH2sN}-CVw(!F5Zp@F)|4?u*+YYJ1z2*L_
zILyW47M=ZDuJTSi?-6InmmT3AD{|^CwC+Av1f)H=l{`c|a6_t?mAOSmd+z5&ThI4N
zx9*%{uGMmi6d4hKx3;Dwm{R=>dNGbq<7sF72R1l=yN&<=l4|9|w!7FPKyDRsa~h%5
zr?yfZ_0gh5lx(6Eo*5LdIX!HcF5eqFBfWEryh~&xyj$`8%tPV<*g(@gqi>=`Wt44@
zx2f@IEdKt)HVkK24S2w}KMIX9@vgMlPv)(iJ~o!6KA=4O!qxnb?x8{M0o4!fp@aLj
zhBhy{zT}YqVLqqIA9|4wiY?YkKkRDeAWVk!r^OZOfPHN4$@KjIZ4tV~w`unJB0M|U
zNI?%hs#)!Rn*+?h>hd0iiqcO+uM6w-WV>Cg`O3WbXFu}KmTSTK=<-9)kU0bXKf=B{
zs;RDdSCA?qC?X2dR6waBO?sq=bQA%RCLq##CnNz85s{)ONJpyFNUtG)6p>y-P3Rp$
z4-k^v<FDMi-cR0j|H)cpaZdJ`J$vSv^30(|wl*^Sy#!NQ3QD3<0rhvun3p4)PnUkB
z$8hc~`le9tnw~TH<C3={{U4JK-P;&xxMPfNe@aQZ<<IL?zj_Q0tgo3`Z6GWwYQ2JW
z#g6R|+1QF=t`C5=A5@=V&(ktfAB44}RKqi%z+{&%@A2NDY)gL!^#2X9<^GVY^;9Xu
zy>uTsd<-9ie<Jc;>+{UE6NDCQndlO4T?<6ep)zeWo|~nG_jGs0z9ttFuoC3q$sT`0
z*rNUXj^l(iqy2>C9Vv_&)=?LG`S0fpr1`uxEj6qD$(_6<L{TBoKHG(<F$C#iVbN1e
zMQeBah~B^Blk5K4E&n6sxo+1M;!8+tmrLw?I3Cq$0lPe|cbDv=v{&Pdl9!#!hQjY|
z`uBJVYz16Qm^H3l)WDwev}7YM&1INol^GscfNFWR%tRgz^bIQZ!x`ra_9;Ae&T@nI
zRG!XP4U`<kYgF-e)ya__`D{xk&e7z1m|Z?9T7m)rVrK3vG@~JpaV21Agw10SOS#jP
z6<OhZ_<`*|W1m&p;MzU2`c%_<I%RMF<Se%+)A7d-xT?6G&ZB(0K5!ygsyh<@cNPFG
zD)yEqOdzJpyyp(p#RoG|(%$OxDAeYOrT-;^NnhVr&ZM78sstC7qR&k?M}C)KsScRb
zmo0!oNa6E}@jAZizE#6u*vf#-4#uPGh4<GY<Fb*(g~i3U3vSDNF==Tc7P^FoJ6ewl
z^Y0-PJeRsz5UJT?7BGS{+}Pgc*FS}|DiS<SWUnWhOZU|C{#iLd87nrTcskGUW=%BL
z*%Y_0IOTFLZCj)`enGRl<kj#DtQ*L4-ogY9T^jUM+@XkIcN01h^=pw4SPf$^3x#$U
z*bna0`;2=lj(OwG&Z97kD%X2H+l#%7E)M%`sdjMZBh0Q>LDKd2&^4Z&{Ck!FWy53i
zG2eLa$q`+CX@N!+!reM58BAON2o+0@a;F$RV_}DPvbvEhf`V1k3QIV}g>PYLF`Tf(
zjSBBY>5zp+<$#X)4vR5yNEAIAF{obJ^2x9H?tLlxGL+Nd6obF_*nc=1J(=LzeKW+K
zCl3qx0$<E?BL5I@68JShKlg9(Ie&^*bg;ZKCBf0r>bSP5Xn<|_c+=KA*(S*_(t_pF
zcG9_5;hP(qoUVrr?`vw>z!uspLq*+7m!K};eudoruZSGeMRT&aQ7WIW{F?>#sc~av
z%2p?0j2qC~zZkS=#vX(beu*@sHGk5)u0Wjk+=1Op<Ns*>ym>K6DmKpo$`$%bH1dvq
z?nDOA(xVS_0cZXAt_LOT`zFOPEo=lQg-!DGA`dm}%|0jKn#H1jeN?;t<W-FwWgPsO
zeJkM_2I?AN3SBuWZQo2m%v!f2?ig7}sXTW=o0<o1mc+=x_1>gZp`k_(6=XDWw2mIO
zMT60CdOFagJ+#`+{C5^0xBK?sI%~FM`53bM=%-87czy-){(t3V{sV+QjR9HRmSw&!
zH_WBSM*E!)W>|aiNL!9brbe{=9JqTU@*-9-0GzpZErvTuLR~aBd;75a{kG)|n6sRP
z)S?^{uZobg^cd7sW+I<Ey&3WT80_<i?&3D}zQYUa?XX;<?%TgHt6`iFUS(``Z^73T
zMAK%$v#E{7%zCe>e({P<c&$3C<aw>A6?eEb_;<IIX<pq(dF1|>OA%c>>xALYv-@mr
zsX`}H<+|D4YeLrfU(K3`Rt|D+l`s%R%^n5ZljwekFNQZ4rzoAZxLN02I@{nOckpI#
zzCR56kqL}DcZX?Z`oxo(iUPXp(^|fQ`3=QyvOm!>#ssdVNNz+;xR&2Qax70!Y$k40
zLu8YB|2FS{P60J>q{qp3`d|+xJvEfv+~;IVDE`x@NWMCI%h<;|23@4p*g)z0=yTJO
z6>ol7m>>@SBr6Z|<0}KOs0#)1@G}ltgh#It?~gtTJ`X7Oxl!@hz<w?AD%{fghH<$U
zb7YEJ;l5$9Df1J;EM4zM1w?d({c1@u$3|}IfSS13h6vLuHRZ1Fh~34j_T%N@S%y0b
z^7t;I$I!>HamlIZHJAFSsNK=b=$P1Jl`8|t6lu-9$g>p}HhkR-tkJ8r6gM9#+|gAO
z8y|U$J>la3gATAG6)J2S+W6IAOce$Qf#OY}Azeq1VpCXmcD7ElgezLBxW?g!+*Y6U
z4lG)Z2mY4!BwBGh+1VASz|POdM&XVL;3h>hv2wy?Zleq+`1~)R{(}(GZ;=sn%){zA
zG*h|mFur30(l9`-nwzFY2ZT7(*O%=0ic9KPQ7d{1gBTpyjI&5ToYs}1S0$+xR)(L*
zpigG+3alccTcd-{?Fi3Dxp3Bblrg#zcJ{Jn0?O%8sR1OlB%0g0bqhjZw`hXRz?&C#
zNeLLzZDki^yX2fM>WzM+Eg~So0=&=T(HA_0C?2nLp4(W93cz8jfBHxk4ab@InZ)PJ
zWF}WQRz>bR__i}o`HVRm^+#hiYhYQ-4F=VU`Ae2WL06I#6UJOS)Sq3`Uu$l*V0lw&
zV=ZB+?bkaJxLGhLNyrgZ%nZqFQskZ$A^oz>eA%5S;j#3?r0b>W9tP#v%gQBYOSD6u
zTV8{^;G*xJyf`|Fx9Gmr$ivM<iR`*iT;}H2%furlmZ@T<UA-HXv~Yj-3e{=Zz%d}Z
z(D_P*7Ra3i2j1MV%>sHCw$UP2_Ld0dw*=ytI~E^?oPRe{q4+d909)`&*Dm!C{jl~?
zv0@wa76Wl-?>jdH5M0hV0*3VX1us2od~U+M&Yl!03l7N1H4_~yoh?p&>z$K}igQJh
zmS0MokYp8pZKSBmgt;C!X)RTUg?<MWk|W<H()N1VT(UROd-iP6tsy~lFm46yoOZAC
zEz7fdKYPDV%QsqW652Xk>y2-kXiZ#|UzF(R?vBM{Yo-L>vLCla3KxPm=#>$?YF{hp
z_jU(+%bdrBs-|q3uU3@eizl-~sMX=tQI}H+-E&adiXXM`ZU14d{wUeI|Lsr=u4eir
z_S~YDNX7bb5aE|Ifq@Fg5E!&k^Rrh^grzo>zwb~vX5$)Wb=>GIKt^F}SNe;(`fzrf
zhkQ-pCk7>q@VC9!aK~>h{XCg3Bl+9MTX-~N<z@r*jwo?g^iY78ZDOml1V;nLrFzq&
z%*`{k+^&CeF*_$GLH+eB;tkpgq|*w_N3I|J^3SLXz?_f&kp{*i+UzIsPkU4uqg$=n
z`Rd?=GaX%BpQ-5Y72$~4Mw2flhPGq~_>^15i`{SE8XKqPLK}?vtouTZez&Iuy1*an
zmNmK_CQ&(}igsI09-TeU5T9ieCrE2r1@LAa3zqL2Z29_Nl%vO0BLEhQ3X?{jer}X-
zgwQ{-etH<%A#?St1`FvPBr5<1jBYyBz6Ra@qdbnL0cKB8E|K{p6gKqXH0o0mto5}O
zES-|M=j=Xtd-$<FrKSY4xZ_VhO^-rK2iSHxb@Vd<{WKcC#(25tQhn2Rsim5R#VQj;
z;{f@kxFix_LTYe;XN~J%&yh^o5c~#DYoX>hx-lQc<*VWIfK$Y>%lgFVD*B0Ky%Q#v
z-C73)Lb~0|=U8RUl~IR-Ua%^8uwO*8xyLrw3V7DNQp*Om%zC*PJBFlE9zL8PIQL-D
zGsG#)wl3Vg%=Klfsu?<TZAR4v#KdNP6_GNe=e+Uf(J$&^vHZI72kDc3o$MZ_@KlfG
zChA&2*9m`A)IG}+J8+@vtz_>W9<;xGMuWZkZlqC%Ps5hdgwvE^oG`xk{o(V6p(i)H
z=Yz}Q&7o6^8yJYi%C$<DV1b|iQxN~3$mhE*#p{4a5E;iFW^!g8AjWv?F%$y2PILJ(
zPY2Mb(Rob`PP2TJ(+y*}Y>b8QZJ7`-d<X5{CLA0y()$W|oeF*4<A>koJz+okxeAaf
zzWMMw5+wyD4E-R?SP?m_V3OQsMOFQJ8C5DjKvSM&Us_t2fG~eBap`Z?j4m$=*iZZ(
zOUADxYbjsvV!X22&ml56it9m>w)Dq-SZ2Po2^D^Qf)vTEy|FWqtl0IsD%kHt*d6E|
zP_J{Y@(J_752FU>(RfZ;^$4%6DO{OTX=+)qnO<DvA(BsflE~*=0>*OO7_WLB%bTQ*
zf*h3Nag(GUHsGL(i$;adZBzsHr;$5}ChO#F@rksTgy51srf#zfx=cBDmXRW*K+kgJ
z2Q=H?tcWw@3pe%rdA<h{9<!1&Ze6^qgLv|iH^<<Nulr%V$S|eilvBk-;$;)zXVU$M
zL%QzfY}Lj2y~&KwTePY4U}m25Z~lAw#+DXqYc~)*re8UHO_?X8(24UB_>ng_B?@T#
zc5_*D+%*eQSugVQU~l~|1KP1wr7BLXfbJXsQ2(aoOR>bk8)mKgTVD(l&9|aW-L{8)
zV3VDhA8t>zY`XTBhSr3I&~1&koFhJ~aqxcBErLvZ-fTjc(?xFmPPSEAcr$_UrN6_y
zC33iA;XVuTJF;tE5Q0bS9BX5`j*uU9fB#Zb8+*x)3XL~)#C=&A$`hO4=1YqO=+Uuk
z#dTUe3T4Q8Ji6VSiG=rfZyMM%hThIl)m(?J8NS-S_-97SKWi{RRU9bRh_DEsQ|2Q6
zLGgl;tnXTpue`r)+%oNphV>duSXsLJan$4HddGy`6$r)i0;IQk2O@2dEidJj-r>D1
zN-u&l;3g6kyLnqy!rBH&4!^zN6?|$4t|{$ji9Z=OX0lhJ-fr0s=X>>TrooORvZ6@o
z?wJouO_myy$oTVR2k&ofy#M)Y*$UAk!>a2IacoRrN}F<5ttw~+h##7B3~F=eE2Fkp
zd8OKCWv_}(i0!6HIp18Gi||nVuh?h>##E=YJH3d*_g#kv;3U*Hfb_Xw!}7LXJJWUa
zJiAZYcNAir9%XKBMgD+=4EwTq=RV$c2I<kUAmz7#hV|-=xzstERVEMB#ZM~C2gq7x
z&|!4~aW|#c`$AOsER$D~&Phwxyn+Osuk&_&^4UAZ3{v@*7|CcT0P9C>{DV4}m0Ry+
z^WHNC;;li%_CRkh-Tnx}Il=>1BI`9fX;~q`b6QWQ5f(abqE{GLmVOo-vCJB{1~RXE
zU!1bVOf54D31#UTZ-678x-H2w<?v1lOPWuo!FDRk?FWB%4!B8fm+JCSt1NC*xh{p{
zZ7PcBh7M$Fs<Lu&GzN3XdWog0D!)-z_n`VsCh{w|@yrPAkwcs-f|!APHwClYEASXB
z`ngeihBZ@R((6;R_S5Qm7>iJxqwCex@GTkpyT>){T9bsg9uq&ns;D#7*30yU)u=w%
zcAWIo+i`$?ZmU9xXNpoBmoTuBoj(90e9|DtmeRFLmOQ`hu6_%5jTAK}>UP|L=*Gmi
zeJHb;Yzf!7-E}gf*55&!a@VcWf49YqYb^5PKb^=QlgHIh!?ndX&TJ{S9|ii>zai!9
z;f_<)`n6>x#zEliRjC$z1v`v7t9nU7D>_OOAf11t*5SJQ0&2)cwqI}53*lQX_=RPt
zo5QWbD<3fXeg?|wy$YmK=Asnb!S<`iC|_R4Z5FeDqI%IyO4mmWieB;S^@CAp_bm4P
zC&-1$x|=rs_82HGM2?{K!hO`d{rXIHe8pg)vHrE~#;q7DM-&zM1C6@i9Eb&Pp#Rb2
z=ct|5F}0eUF#H3Q-$9!`boaA=*#qb4mE5oDZ!gzzZiMOFsNY^xxdz24n|~a{rwJ;%
zB6QYYS1*~*`VS!Gx6cY}Ve9q5{_XF!L$n^H`}0aR?4;e@hDa}U4Q7F>AUr%gV~(1>
z#dWKfj(UFVh3O1botN3DE%Uk?ajUK)VQ<f?W;0RKd8*%<ocmMpJ<E@Af@TiN_`t({
zTf^xz8KFj1J~O`wJ_-ryzjQVDE6p3IXLR1VP@sn7cZufRVaKOW$yOA?Q@j&ftlR6E
zwG#t}cn3eMk<?KikqU^g=2q+6WD^UTi9>ve5sSm0BmIvg1b8j{nvBtmJ)B-S7vK5*
zK+iA`FmvX$Xyv>+dD+m(m!FL8IN1_0q(>w%;U9UCA6W1~gU4^b+$QMpF>co4NH5=n
z%fi3CDMq?~HTvd#Z6?IDN%#HS8Wu@v@F=?2F|0tEb@Xqi+2{wT9EwtFX9dg05W4qy
ztwaG<*(kv_%{O7aq?bFj8vgny369{xO+k_b#hOXm5Gub}BIQ!b9Vs(dXo8ktX~|B-
z434kC0rRvE!i>Qug!<t|mtOcU&+yfY+L^aUvA=8^idG&`?enJhYF5JB8U@3rDxSSt
zx%#=Tfwq75ow74})CluAlFjqw4vWRuTnKob^GVogQ-!Q<#;}@*Y5?hf162?uG_sq?
z%@5_=XoAVP0^XvQKPqszyzPt<(t5uO4XiQDssNoQP^ID;EY{(Vz+Z~WtXlbN@aQmS
zF<`=jDu>^)XkA)rDz7g+QuX*i)}-OYAraLL0K4f8Bd?_}PG~8&T{&@ii8Hon!o{FC
z=Zc6qKWhW;q<D)~EN@e2_|M&`lDA#%N(6<6XIyQr#18u%%2>M1@SuLVyyAXA?>Hy5
zkA}^cJeE}FQ#f(oA}m~x&ZtTebhqcvvXJqMNmjH=?~>p+iZ}Dm{2*mc=ou;?zSk6C
zbS5RPh5d4z!&JtSV|B0FVxC_03sAh~vvl?zAqnf`1Vt!_)ZaIJ1;K^x|3Jma=;s8^
zB?6X!l}tcuImI20SXfxm0r;EC^DP|UbzauZM=bIqRaeEedSzo3e#Fu~-yF@fh7~Av
zTK}cEfRP1^geM;;{VnbH2CUUqK{eLAfcD+7Buc7z%i6+_e8YwRLhqas9Zz51YM0(5
zd94An05X3Z_M#=z&`A-uPIigI(~NSrfscca#lkPD{%$~Fp*ezgKA^~~3LSuO%xHt^
zYSIz-3Y{;{m;?O+AH3E?8!!#vu<|_WqRL4n4o?5H`T51)KW1>6hK6TrPCn6P0qFwg
zv10uC9oE%0SMWO8I7)u;5$wqjgyro+I<J+1j<~w%ld_jXWg^B`SdxC+cWpSxF!LIb
z$SB!?ecdmUukZUD`HG<R0bhp&Ka=qtZ+(B<e%-q|5a6nfWLjVt-74!=a82MaV}?Mp
z>4RFfuz+S#=u(Oe%MC;H>SGwzOj+gU<_F1=U5uaOC)2=wY4gM6v9hIKUr^}PMEbGS
z0J51JtA@5cj89S{dDHR3wi??sVVuI<C&l%PS4>R#c$!&QF1UtS|0A;gQ;a`Daf9j6
z^9r>7?2ULXW}t@XK^~u6KDqZT6;DP+N}!E&_G8-*^0jm^4RomT4mqdjvm#M5Wk;d1
zKZ^=gXL1j$eHet9ESX9aEISTnN1AAI@$fL>O%Yv8O<wMkK&Ep1o1D+Wt5;-Gdf#LN
zKNy%P)FbWMHRFY2<!h;w%C!hPXXn&0564}ql7xu;V7S`;6W<MO;)%dDx5dzrvnTBd
zB+m}_iC4C+F$Z;C6&ncRs%I&zsu5}<TkW<r>;oZ7x%-tktY$6<iR^^U-mx89WKgh!
z`ERa#1ZLZzwOl6uq-M-eynZ_+)R`9^w4Y31VSW+&4x>rJVD@=LNjjRuV&+zva9<kT
zHV6>jqCLh8r2NfE@n#;WlWrReBbl&iX%;Zk?ZJj)y~?owfi-)~U4hO3%o^7#!eNBm
z?xd!V-0l*|n&sR`2mpjG?ns6d`=`4K*?muk^Xir^OhuF8;6JRr+HKRuEH<zE#LB+>
z=Qyb$WpaQ2qk}>KslMsxP&d@nlW$o!u4%ESb6H~y<*f6nWCQ{Q4WrOiCR!j+kk9~;
zwg*~$&Q3Jy-2ktl-nz-_+=V~R^4h%EM|&oPrcc$UGx7Wuzd@5u#1)}JqxyiB0<|i-
zg?X&Q<QmgFlH}D~kfog<`I9#%Aw2F8x2I|3M&(4M$}bjrqdDRWuZ89+fO5a}X|d}_
zIZ?`SkzH-PU)6O2TMfw)6am5~*mHLaCaMd|05S<EbY)sjI?H0=lfpdGD^0rB|3g)0
z(#ZXU2zEtfJT9{QDX2S%QevgDU5;6sk05D(EbCUP!TU^g5-kg31&-Sq=XB+<75jYz
z-(PvTdnN6wkFjz@IFKu1#z`p4ZxM31eK?MgV5d)Sb&ogk#+4+NG;mi^*PwbBM;N<p
z=W-J3%t&eLVkO2X;^)!c`)%!aTR!*yp6>(#Q3R*eogt@v(0KS^ViaFHOh(yCD5Zcf
z4>CEojnkflOBTPW9+D8JVLottGKm`0QCE+9u$25b-}a=aYHGtjO-(slblbas!Y*ff
z@~3re+5pM4KcHk>=+$+bg!dR5gt{B%Y_wl@vQ_iSo20i41Jwttuhb}g_(~mX9_xOH
zX^UhHHM+A~P0Hg7hkB7O)FqV7`!gw%imEXCtM>Pk!#Ha@+XyJt05yf3a=C++GD(YF
zVYa{qBgyfofE~v}j*4$pW=rxRd%mz`cAW7DYrxV_>{%w7nGHcH^OGL0i!V+?{r|T`
zi>sL8)7cWgmfIezvF(l%BAa*|f_!}lOwO9WG-=KDE8U!aws|%|oo!*vSE?<$=uxEU
zQw=$PAz10fCRCA$QN}7YC)7~s5cT6OFfr%O<4W+GJ{cg2EHx#Eg+IJ`YdMJ`dkpu=
z9fMOm8M@Lhan@9ev?ar8P5NSl*}5`2e6rgkuPn1UsZej1(7sC|s`0{z*;+}&lrboC
zv)0_6g~xWyE5hra8%Y$vOpqV04Asv_wNwC6Q{mE1+=KE9*t>MwGT_A;(b9t~Z;wVI
zp4FWM_4i-*OWL0%{H*U&C@ysIwwjLZea#(LjIjK^Y{K4uc(_*jcYAdZ=r)<vR@lxo
z4Y`EU^gsjar&}FuP}HGECfecYbudP-*5o=^XV;s9tD63TM8DiZnG^-ZY<c+rW#9C*
zr`S~R&&K`LsEv_&xsS`Dhu)JjN++r816o&8+crP*mk)<}OMHkf^V7LrElDuhzZ=a%
zg?yVNz40N)s3xEZCV8aE#%4a(84V@6izI(QZoFS_7o$R^PUM~KiG}BARVg1G@A|4L
zQ->zC%^+tMW#@5<i*Icc7-b0yZ`Xtc;#rVyMOP=V>oJv4vX!~X?#FjLKsbnVK(Vow
z-{hKW5;V+z!HaxWlu7uunAlHE0p}%v=@iG(ol(1a(%fn8+nw#vn$7Ztud^A(((z4g
zuY`#~X6A?A_9d{8x07}Ihf-Jn&)-$Z|C#(*qOFd}Z<&Vg#wRirl9N9$n&G!g<^`JO
zp3#2ygAYKb6^pVt@E?U79?s8Zji{5amcYsD#h4epMW^fxpQn!WpR8WSWkwFB72i0z
z86COP0%6|6B5l9d1Nmn_9ID$t!Kt@v(;YL$HhyApE!d=Dq*8M$fk!j0!^Q}0$1>(U
z+t4<N+^+3ZRi5}RPns1dg+*HflV=pF+DlIKL;`KRcIl0NXVJocy{;*n&uix_pCvwF
zI=A$vBGd!k+iHPqIJTVqxsZ@c6nW)t(f*A)#+(gb7<kZcg7zcK;c=K&Zr59O=zC?A
z%{5CFmaUKt-@V=F1ocp`Gce$z8jrrXSku1S-TM*pm-_@6Q*2F(L@*vPey1WQV<5hm
zC*)$Boa7G;_80=T*f4}3I%Ji%vggWXbxxJvmJya&&lS2qEh5wgt%@)lf3RXz*Z;e9
zKu4}uFp(MtGy?}ZQ($sa{re--;UzO_^3qC`c4N&K9<X$84Em<;6Gz_6f%iW*v7`vB
zGH+HjH{a;!={ew2oibt|A`WD`7LkYRnLfee-$NnSR`*j=Q#Z3eTs_JH#wBGQJmce0
zfWqpRCj|M;btO3PnHB-VCVnV`+T-^T&WFh#KAbl$w+lCQ-|t2P+`BP^CSQN6D}(fM
z&?O|PIP%d%eZ3r07)j0F`GjJwLXkV`5i%B>Bt;?2LT~ZZ+@!%OTPqN!TsK+*b3RL7
zHm9gigbDXpX${Hh(;7KMJ-CY7xh=gF<Wgy3rrRWxSm<iEOre_5K+reW`TS1~g4N9+
z>OZ7$0~s;`x(N3KXi2|(`HIHJehTrT;<``6GQp(yG3k2~fj(nA`aTkz=gGZ;8*L!n
zuVarDB59l86S-~FN_#+-vU`{wcgd&S$1(M|f?Gvb@MW@5!+R2IfrU0DRAPT5+Gr9w
z@q!ZihVEoz-1#O@(RgltH=;Q|Xx%BDRzz&Ql1)j)s~Zn)Qo81Wm*?i#q~<I~Ui?iR
z9#C2k$&Sl0g-kKlSr{m8*3Z3?22a~p9<gYRCMLPCyWqSpnnNaDtQAfCUXE|?0MbZ;
zyIn1BFb6=SK50v_vE_&Vy%^eW%@bH+!F=q>>48{&qjqZ$d>E~`1dM7ymc-utSWoKZ
z6z>!C39K!j%Gb`~)zD@G65t6hf$DmnTtK8SdthpU;AnHpC_n&{KYYixw>IBt8ym7l
zHtCCV_6Q)wBA`3F-3Y&#FKF)}$HR4f2DJ|utbrBvKe9vL_5XAl4lNWO#P;{;ut_ay
zxVV_7u<YJ!D)Nq4kK#m*pF0V-H)rn5vXY#UgK$|LnQsZ|H|t`}Pty8e$@WNo^*$4K
z=8wAv?!`c-`L%264?l3v>Z`)1a<a6oi1FHsBYi?tKntUuJ0|>?sOo8!SZNW5g4wj|
zeLU7x*neGhxA5H0vpQ>Q{sCcCz{DtLRrw7Arv~Ry#LgNVY4X@aew*(Dr?dQ&r&j&Z
zDv%wq9o-bC=w<^#K^gmR8F*PHMVgB^e6m|%vc%TVcO^)vu@gSIjOtw+cy@(Fj=|TA
z(N}q!mE~j4+IMrpd@exs{w1?FeKs3M+y3if82c^lL<(Mijmz)3X9hb+yDbp;c4+S2
zm6W@qwHt=!&JeNm=#{mKZP(5{*M?ZJ-uuH~>zPnnj*mB^@3#mP$gUQ_UlaGGrd;@5
zE}jtiz6B`Jxc}AE(obV|nU)!sVw07t{_0`vQE>u)Uf!KnN7LHL+L|H*j+0e{qSNRl
ztj6BP#8Vx~v}anakd*6s!K6y=BCEVhxO-Rx($GB6JzLx*Sr8E*A!?-=TaCMjo<@=u
zJ1bB^r5*AC;-Be6Xx>!Oh%71gblmym*+Wp1@#%_TigLYCw{hje?uLkz`vB`TS)YpK
zm!LGX_WY23(otFFX7AUM2OJznY2C=f*((DG6E1Xl!Herg3wvV3$_)qfLVQ&@N(cT;
zC+dTroxqq!=W{8k1s##ugPqt&YUEo?zcH7!Zo`b+VWcwh;;|`B>Q<3sS9aZ8BKlHB
zZVnujPTErSBxJsC&!{IJ@R!1yanef`E(K_xT?xbUu+7kTo?bt%w9&eFmMFfsN{P$8
zhf<i2M&^*8ZLzLnR-Pu#b8Xi<&_B52c==6pEDht9oQr2dAAY@$Lqtgr$SKSEIXfgn
z%x*uEN8yuxNSL_PgvPUU_WYUD+0U2y<yCj$Z-aP?%N6efHq!!Er8Blq7&c3gUK6ga
zE0tLEK)t_H@6vaA7F#>JtjE1kahYHa;i)>`FXcm{sjaM)-`j(yR4~upy$VpYtAW&Q
z^5A{fSc|T%{3BuovEKr$70~i)5U2qR4R;|>^<!ro=*#YP=LR$H7gVpkaa5>uU8KF|
z#Jo(%X)if<LJW|oP_(z=QfaI8RxSN`(B#)$62&K5_sE=;>AcCsywo;F#ksv#sZxm5
zR{q9<g?Wm~kE=&O@F6oXa&)wGU0ql6J^vOKogMt)TG?+y!cM{ZS4v`qBb~NjVdJx%
zU5zMpRSv1elh=JEjsY&kZ6$doIkul)_Lz_cZG1#&T0$DFjU=x`7GQDHC?9WkbmE2H
z%(=esyd|~BA+z2gN5qY|PoD%=rHeU{5_T~w;aK8c{<sk~$PwO<rLxN!L8`M)i+dC;
z@qmwF^J;)~LaV{mwB1f|`^oQ*6Kn*Yyr33yck4}AVDR?>^0RScDQJdlZS&7#XDI)(
z`j&3a*mo5u4BBXZ>>n7?N2<@g<t=o5j<}C|#Ahr(0o|V*sm34ER!u~zAp>^QfMH@s
z&_PLYZ6L$KZ`wPMO_|}!cIdqDo7~t^p@)cvH@4g|$6nBM@?DMLHgRSSur4;%za6WY
zH^(Y{@U>fO>U~(}(czqT@g3=rdi!aunw4A1w@e!98yoJV>nuapfGL|}j&MQPCHzI2
ztlRh5LXduJ-##d+^SLw|<9Elxf^McOH9+&hWM(MEiB&#5gIVZ)8m;`_{me?V&w2?~
z?#KtP-ikYp-!BKr%W2pJV}!o5ZPs9sI{c%Kqa4g@KSQBL01p;7_40w=(Uu3ZJaMOU
zKO^>1_L#nIQjB2s*M@}jxdIn_FU|h$($bhql7#k0ojNL(d#Pf4sNKS7@nT9k^8FKk
zI_-Dqh=ZM66*OtL*A$L<Ju7wLM*QW3lcRvKeC-!IH|6Ek1NOeN>JHlxj8tol2W0`e
zm?Ve)g>3%X?Ct;+u8VR9YZqN~fYAT1ssTS&x~2b+C#Sn2jnp?vDD%Nh{|#WK?WCs?
z5;kMDzv_FVhH1azt!9Dt48&`)=y`U>pssUKU3TJow0_RZiF^E`vL?0S7)@V$q32l|
z)Vq~IWgTgh82<a?^|Ucbj<4IkS|oOLO|y})Qp3X7Gs3R^Hn+G%68`P09bjS>@?wO6
zZ6OWz+TT_CnRWZiX4_6(Md!WWah1DhaIl_QaaS<4Wd1&S{dhJcVBVxR(AUam%C#F9
z+x*Y=f&Wf&aOojwmlLA_BKpsr=^D;@mP^m4jo>HGD53H^e3R6=y7gBH!d=mo#wpFj
znJ|gCo;pgzG{Kq<gKD)!-YZ$pOfYsdNEBBOkK4$0tiE3Lt|;ElXR*6e;3K-yIZ^Jg
z(a^j^SAq7;lDb$44$Lt&Dwt8Hp>U5Id@v^4xr@el_ELFt5BIaYGkw`Y5P%<}90xbL
zl50xdyhHlAH!SA@W2Km=NWEytnYPEK(VQ^f(pxz6-PX3RR{H1Lg1Gm-g18v(@_Y-u
zp#CVR&t#f9_+;f4nGzk@CGxy-&@#Kew&*a``LDBR=dpZWl7Cs^FnT55%eQui{r(0z
zZfvagx6I1MI$IS?T2}%GT_e%z#(tKgHGcN)3^@%~&_6%)$jB2DZMzexz&VMYX#>1^
zLu63;mNm_g-MoQ4yBCUPedcHJ&nL?xxsaaYvd{sa1jwMo*9UI;l!v<$Pd;7~TeYdV
zcE+fan;IwO+tNaX^L{bmn_IPCjC$}INvws)aG(^}L%1e9=0BdbX*`2nRE@ay3|CmT
zbgM|oRrmwV2VK1jHOt;F%-CN%soR>cA1Je>{CkzaqQCG=zs4g1eBjGu-b`lu85g$d
zsOmtZXy{3{$-J&et)1ThnGYed>w_jrfB6haj%l5^m{j>$yh*aPSLAuJ%-sej;|NTI
z0u(cif2Z!d8xcz#p;O^opVQQzftbe;p%${F_9T`u`}^zJLH&~EK1YFdTFAg-`w+S9
zc@2`E8iy!|7`Su^Z|+Kw>~FYprrtt*va4q4yae9y0^y5CZk}dmN&AIQX1*&z6&22T
zc6)D?^6xo4lTxIH<!@*>nEg7Kk@63Cp35`r#~Sg^P5AFU0YNFmA%4mssIh9JUXCN_
zN3^Ktm&l+?1@ikL@<;m!s!cJWm4GAL=em*;Q!pU?Vq~kSum4o%cm;LM$P<fs_<r$s
znxu0DvbSx~{<WFbU9*S_J8b8(P$)H(=$YSayk;>1AyhuG*u<yGtkYPnl@Xy^Q(eBM
z6$!NiQxpdwO^~YkfgEnys>Nww#;x3!w#K6y|Di4a^(W)azR}T_YWamld6Ffqdn=vR
zKE**&Jj~OmGmyKD7-lkj@o{P&`jGujVl?JU17w0oRz2OI=~~uARq-n5eGL6hUETzC
z)S6nu+Ik4d{iB0-YN3x4=Ubq!6GmU4Q)6zl8cRDKwJY)!XtBwWMn#ZZQ}@<a=#Rl&
z{by}5u5qRhX+G^zS5saL`=Z`&opYIj5p;&^uMb8h_}*;Nl%oLneD6yP-JkWv_I;LP
zo$nPN?g`@@?#zzwDLcF6(@GyKRJqJO*DVlx&+`?60^CE5>rcf_xMCM}ES#Pk1@5S?
zrC;K;Q*;?T&=swg+l*iVOYc_{qapR_em~4vrco=yMv;sN+H@((9))(zCb0moNcwcg
zd`_c=<NU{|4(uOLLYk7pP5yk^8VXcR1O`DCZI-M%p|7QGej3j0G=fKs#`lOk>+?CQ
zjEQ%k4O?!q$VWx_+&o40^SbHAp3tNxW@aAW)=$1sI0#EKYj>38(eBtjSGY$8723=J
zN#23nlYf}M;d7!Rq9~-3zu{%Z$rPk}Wg6uro0}rjVz@n|FF-g$`h17*5{@_?NAE{m
zsd%@T%MkaXPi`~><IHrS<plM-;CZ{>7dF?Oo!vD_;Ibi?T0ixPNOJlgwlY`UuxJBO
zGU7s~$$VLFeG=32?RtN7<XLvQ7hPYUg9V*`78*NGjD`k_tQG1DP1S+vIc3BT0?#IW
z8FCs@sC*Hywbl`vnVVJ29;tfD#{EhIWwmPJ9t5Pev@Kj@*A8^q5g8m*{5qhvR8}J$
zlQk}EalM}QuBDQ$|6;-yk6uh%Vlw_l)l_XNMaV%@FqoRb0ai1`RkxW%+>Xx9kl(A&
zl1@(xlvSf%G2?~c%d>Ir69-}lLOgtnoRoM^?JVgz)gJs)Ht&Yc4?$5W8YrZH@(p57
z;u-Q_eAwg2#&OI{9MWeQ)_Y--LkzZhNBJdU|M%L*>-d&X|MxetnjpYyU;b9VKPstZ
z=s3Y`sMRIbr+@9LMLMJfWnlmn8&9$bA%<!VXRg?~kOpTz+!gC6!%SZP!l1a9s?o<m
z8V?~57*+|>$C@HV@BOW~zCO1l>?6|s){wLlyzGaD5yJT0l!%26KBIBmvFhN`yJgLn
zokA12*JOs}9;f32ceP`ePlOgtC4Rt&M?dF(7hGwf*v0SDwbxgDNfZIE_p6CVHz<75
z>u5cgkh1vIWYK}4OqM;x=z{8l+PZIdP-?7JhAj>Pr@J;|!pv%ClNyYzgLoP9dRK3F
ziaI;IA8&FT)Shar%8NQegB?$rT&mFHY^De8z%<vb15VPjS0Ve)+jG+-fBpVq4iC_F
z%nVs_pyf~l6Y6#*JqsqCx}Jt$i&MWc8`t;COoAGIhE>tm%eI~0ERd2k{K`jmV&s?<
zYk+Jn#8pY)Z9J@R8Jo41PMTj6V|<MYL6*AuT<(<k#{QRQ`*VCQ*MZdcr%@b-tm!B+
zAEt!?RAWQ2uRkt!W=--~W;vbckL`x9IE+X(?Q6YX?DLggiFGD9anZn!2o-jXfl1B*
zzmr^x<61QnH@qgC!M|g9jgSHlL=gSc8qSd{S|bm9_!7z7Eg%&=Xqv0Cm32-H#=c`+
z^$#QSzj*fSQ%wzn5OojSR$Lcr9On54tuaE1Yfi2vFq^fm1l#jph;4H<ONPIltH8Pv
z%kKMa&FrwA^*hu>t5U&ptN4QA>)bch;(UH{AZ}O=!Yo|dmp{Kg&2IqTWij2ZIfG1%
zq1p-|bM*ZbpJ<LByX-4AJu67H(VYXGI_OiTZprFWuOQKh6kgI^i7m7f380^K>_;Ca
zhM9-V!4c+sSR^)kqu(2cu6kNIo*$+K(9B((X>&&Hot)^nP($VNE$)XiXI0XuXbyf@
z{75%`)$6jiyW?8#?%d3FTeun<6>lfw_F~#SScUEBR_PD)%3ab#f#tthXAoBkNS)%)
zOoJUIW|8Y+d^fG(VBL2jOV+*09Cs%)gKY1`mIZhw&2Iojn^f$NbFK0Ih{-gk{yF6!
zAd(WjIZgF*_$YC_(rzcF30<YEto#%W1y{)JET209I-ru4EUJb}Td&9m&UM7%e<isV
zR-WD11m>Cz;$am-D7E$J`e?<HdyDtc=y2q!b_gf?K-{ZG`kU2jNp=j~6N+x2iSF0L
z;3JiZ_~ueOajU`I<#cXIZQVQKiIS)KM7?k1RSHk}st8p&I+IoL$$QEogYPr_9ACvw
z9{fNZ1&nt+tQytfb1}WiB$*!<o;37pgv~bYh$rWDwUjF7GztViC}+4rKU)Wa-ZhIp
zXwfZ)xLHWS@%HaiRdmU#ojpV$8x=Cf_0`gu`3eNV11oCZb79K&AjgT*H(=(2iyzoc
z9ABfS=2gXh701O}xU3<NBr<}E?7O;qj)oKN$9pboO{D3N&IN-(v<TA41dg{8ff`j_
z$0!qtxa@<C3lq3c0G4wOf(>Mu?`+CCY6V%m1ecw(#;@MHu;mz#^5Q)^BH<%{BT1fV
z8kNiPJMT@wp3K+2hA5xw_0JZ(J|)wgVnEK2+i<-&d&pZ2U)a!1JQ4E<CLE?WAcGV7
zy^jZy&a$p6x+Bd~&h4>4f-sb`b0NnUcoaZz`Q!aBf_>Q42u$yA{MyO{?{X2m=St*}
zOSNSY{dY%@#ZuLHX_(zAJAKqzfYh_1oFU`d!d8wa`bX<70VpJi$eLF~-+6MR?SZq1
zJM6wTIqqn|a)N~kIQ>8u4MFhx1)o$S>d%mPlf2@aLM8%w>NYNEn<P#QoLLEg@K3k0
z7^%a}E&ZiqWN!C2e5n?+5;nj6vJ~s6OSGHNj@KxkPQG{=-ncD6hRRe-@|k!+8jhq*
zBmY&Nn!&K;$c3hy6A_g@km!RyIDX-fs=~K6(vBmdAZC#81A#>wHNwjv`q^j>Gl3Zj
z(HCZwb5S{bkn~(qha#tw?n-IrQk87Q=G4zp{3e31@bj&285xn57lzDf0unT-9zW^n
z09PKYe|$8NlUTD}dPIzTy*Y(YqZZ%qO53zkjM=1O8_LPgw;Bo6cC+x@a=ZeS*)r;~
zSyDOe8GeJf9=j<TV*{rQBMzp0Tf5tHkKNbwwIdW7)AF}hM(kr0uTG;r%L#9#xnN9-
z(yRvNoDkk7MEuiz9Y}W1L56N+B-Ozn^Trq^SNEA02#q*cbsSt+R!AIJEPw-Z!p|>T
z@U(NtJ`(?h{q4R;_+dGfr50Ujqj)c5vw_sanwNKB;skgVT7WK9HALJyCT=s>l-Oxh
z?kr|6D?<9rp_X>VP4_HtyIxp|w@jL?#v1>Ah7`z2G<MXMQ-jcD2QZlbGr$4dz=FW^
zlg(<ep+;hnZAGNiYFu3&-R#<4Rzdujmwat;<^$rVb6bu<=HM+`rYscX^#dq%LhnjW
zl23fwLwHr)Vh+P%ysq~onwJymEhX}z*xM`XJg2!3%A%@rlGNxSz4I;@%50k^@ziU}
zj6K=N_44^gckga|f4s3f?g&AdmsJMtM#gLA=kN-Gu~R4?N!0WSfEc%;-8+TW#*gd=
z4LeI0gd{`z(jPcutZY-HpT-F~m*At&=?}fOFe<99tS4@Yy{>z9aP7qRzvL_WdMA@_
zS7optR?xwQPFOPNU^QmuJ=CW_kaTj2OZ3%AtL6I-K|gTF?#{G@>AG*(1vhLkmZS8e
ziQSRGPqzw{xlrV`*RLcsEVm(Xl2YPb>2<uLoRz+kUgqtXeU~#U-DhAU;yKR##U>*q
z$E$|mIi06QEK5wRL?))9+$Hox9_0NEhXG5~uCFvQks1bHeD(8ka_;x$S{sEU{u7$t
zz7Pxu)!JTDQ&UT>MMY7ZVzym=c8+r!8!_i(r&>&TMacYB!FAC|P1V%3W7#KZ7jnR}
z!<yDU!>M>C$@h`$HRxkeHeNrb^^UYn+MvU!iCUv}_?=7~#X;w?Roq(MS33ms_SgfT
zkdJr|dCVngcs5X4>cLFuD)`|RF4>EAtDFKkm(K+sK9wJT8-DGe?T<~m*L}XSKy|I3
zjmE_FI=HN{SZga=Xe{9M7eS>n!Q`}GQ)i)A_xNZeC~Wa=2c;G#)Yt$eIjz1UWE^se
zht|78o<N1k?Y3iMlK2uik}Hoh>+n*XYPb=wC3`i8=5hh(JAa{WvON`4R%pb>@Od4;
z*n8R6NcP2NJXz!vjRvs>R}saB7JGFojdKvVDPoiPi7RY2yU1hhi8k$4?tmpbdFZ9S
zKx`2=)b;uIv!2OM$sqx_Pg3Uo{KLG3%BKk6yI!!C=elh(Qy@QRaPQ^wNByyRmnXPm
z)X3OZWLWRn197Xr++~B)ViK+((8Y7f)vh8F{8HUn;S`5rkFD4!b_CN`{Ivox2Q#QU
z|ALO+8%>Xdt-f5FW69DQjCEey4?m+#i0zIgqc|)xcm<}|;eJ8ndE2}gL5vvs*LxUF
zSCU4?Dr@YGBtBn1G2`iVzesa#)EsY)BE)D9F`0JAog%<L$S}6CCDO5F;WrHM(G;G`
zxsJImip~#r<nUvA(|nC-I!XI9mqA>Y2ZT?|6FF$@n>m2jfSzJofZc0+6PXVfVP^(Z
zN0IG*@AtgKaXvFR%nY6vaPQ7Md5224Gvlu^kT~nb@UIViGILmv2~Ym>Sp2s`N%y1V
zb$K!;r32n^gr8_>xFMfT*gof!Kh(cddI8Y8H_hC#jm185xg=<CpY-`t7sP%U2y&am
z<r5BOQ3L>s7<E0y5L>d=MWtwEp3rRhszgE<f0cm5FNr4LB+6X_wwY#oOI$Yd9i6(?
zb(}V%9%-NUn63P7vrxp#@nQ29=4ba3p?t9?M@`ONX;D<T;n2(St;bUj>i9LdUhZs$
z?OS9fILJW%O<Yx=K-a+T6xhIU=RKElmn30MSncoSrZk;*fWTFNJ767ObL{eHd`f*C
z<2%iT0Pin1ny>#2eDf#_hl|uJ61{0IR|Ks^V#)W%*<gS!9oqX%2KZh|G_e>wYQc@8
z%H`SBhn0gc9rskz=_s|@A%p%%iC=Kn+owGy4VT^}a^ZK_*C*q4QrMh49hDILb2Ui>
z$qYhqm_ngXQTvQ5-|SG^eNfotLYUeq2MKg0_&lXJNYOnu@eN`(*5F&no&yI=dw;NB
zDF~VAu^t8zrqqHHeymh;ii$#0{!J0eDa7?)YsLiC!7Z(%{;zizqNw&OZk=1#9sNF@
zv)Ru<NT4FnZpDw_Xo`gNgYB|ZpNmUGue4&U?rrP;i*AWiir-JO4^r@ZE2$mre~Bwe
zDM*n`s8W?nmwd*pT7nujq_gXFHUG4T9C)aEm$(`&9<s#h-!UWWuES<e`wkAsvpkPF
z#oPfNH}oR#-&bv;UV}7(V{(>6Rl~-39z<tPpAHgfvM$o(8uvO-Fy8(g6uurtgJsc3
znmg^V0i*m=EcGpJk)Py2o<)EjP?S?6N#}O03%HiAQ=H}imnlsbBOcv)vgf?<p6b#x
z_MCTAI4Bu(i}mC8BkqM9_X7XZQ*o;2-W1d``gnQ*ZU|Nz%L^EUKl%ycnj8<oXw$hp
zx)6NYlY{!{ybAJ5ot+#UgpVtIRL|0G`G(50t8nRXWn``;UJ8G>0jfjGHk=+qjFAry
zPYHq`%GNf@x1_=3C+aVreE@M&=4+t|G5!XjKQj=AcTQJZt|$8XLaaiLc21WD_R2|7
zT#Xp?8y=d=ri}-aQd5WC`6B;v9^Wl<)m2nT#AUebvp$-opxJ5<-ntT8%y>5ix4|Hg
zG7y3kC_dAEn$Ns#7|bE>;5@i$h|<hl>Cm-j0)F&W({*w)-suH&bot_Rw3o5${kx}H
zsa9S+8>(^d2gC(|wf9B-X5$>>CuYBUn=Woql%t@J>QC@8*{3N0;B1mW@KpIMlPZ3D
z2Zy-qPQ%Kg?^`u41h8njZ77Vv?fBX`JBPsTC{9uKb(3`g{%)Z~$b~11YTDhbv(bqm
zcIKl0B8irBdn;4=aUvExj;m}capC8eF?>ED=K@hh<h=>{57U*&S$$D`;A)MQNkQ>e
zx)#SD$P#G>^F=BUbqL{`-e^I$w(O9SkJ4#IAw7^%+(0)?0r4_heGtH#i>*Qfz4h?C
zJYh_%j_0d+a`m$<4ip2lPZMCO@(xM^aNA!x-szKt5jA4zq!o5@+3s3$pVNX4Q2GgF
zenFi&5{e*r$w$Nq#se=AP%q0F|K6y?Ks;B!oa{lQaOm!C9$$?al6sh}5J84eUMvpM
z?q}M0vOf<G9r|#p9^2E|=@N;A<404g5~Z)SLlQ<NjEYt|Fo?>Bd%Qu5YYbtXNSG`c
zae!!^v#-kEH?_AcrKtAOTX|@4O&8Vmu46azPLOyj!=1xH`x1^*PUac1i|l?zXOCIu
zUG3~Xzis2i=ig!@Kl$QSA&q){&TchTL0(Dn7yRf*pLwGR9y$%>=NaKXB_qnkD3Fdg
z0w*>3g7;UsBY7J!LqmeyLP8nr4w;hdKI;by2T^ALfc&^8cbOt!BynDrcV9>O{bzZ}
zdU!wK-L;Y`at?O>+6a}9)6AyEbFP<Gjukml$Oitr&#+4w8nubZj{WDBx#|<s6z$#3
zaI~9zJXWVyENEKJF-}MCPqUNzGxc%7v`cMYnQlT-7K<pKemJ87efz(&08ZNgAZWP-
z3PL6q^DaHvBTp<=^W3cx;0=ww@IydGg7o;bwF2J-dy$7LMXnrJ%9GQ0DS<*xwpFQ5
z?$RD^G;#vQJ~a$?y77ns1)YJBU%tW}qf55Ygpmha^5{I8AuPY^u+~zH%hf=EL3ir(
zOZ9lkDGod5l{{FpdHSa2=NPDEcGPZO3|G-+fQm&X{u^v#e8dL|QcPi7yZ9>><m_$o
zM)4(hGwjyO4B|Jvi5)riM2FMDQ$|qwdB8%&H>u8F*3eHM$yx~yUzn(FI8q8kii_MX
zdU)#5YXnwEcfMzak>=cxmnSyk(tx{9_nA8MkdW>B(?t)w6`89sZ8S6+03_)-CXT)k
zrpf3;Qa#Bz3(3{K#Pu(ts3nJzW(MRbpGSdK&OJ>qaq{_LwyTP*RQ=VbU&|63cPe<+
zdqGBiLP@pcM$-bgj(eIJkC`T=`U!R|Ub9P0iGDZHe_aJSqEpchyu~`~L@gt6?Lz(7
zj-q=HW3Ubl<206nqwYowT`>O!g7X46&G;LC+h{i>zxSz@6sY`B912X|@O<AYlx`nz
z<KsKjBW)zyH}_v$Oc02b%p*99%u0g%=nCK-ob?p5?gZ??$WM-VbD@8gU;bFD-%3Fo
zCpPZaq{E+F>GNw~8H~O(DY8JPKJnM?rt6by#ueYS|A2TDuH$U`h)Kddg&To$4toc~
zRKAr4X!L-JE|@y@ls}?$YKP#}vpHYPFW(vHR4F_4EZmj>#H#A6Zl}j{P9@`_r=pBg
zGag-jiT*w?nD9Rg>365#insKo@`n|+&}xZXca!!<74~c0u9&e4v0t3q#%_Xw(y(L{
z??GHnuYs|Re3yqsRQtFRPcufqcd8=5xm^~yNMA@1{BEU=^WK=h)9cjl(0_4fXUOv?
zX|{3?EQQD^FMfAm2fe+1LOIj3%JtZ#`sG4v+iAKM3V2$J`;Jj>Ku?0-(cTEtB0=C-
z1^L5RC1yX$5(0UM=l#DP5-<*hz}c~F1#UU7#^uDc<#++AuPA-E8<K2Ov^Uv%=|lHl
zy8>+CIpG%%53&HaYeZ*v0lEPcn9Qb9H53(^2pS|{;M?Dq1P?H#Q^V{{`JNjuk~_;E
zYU3aQTfQs9KH;Bg%P(HPewr@`{&TEtgIaiL=1oB0O~(L^q;G0Sy;_x^RJ>Vp&x5~z
zg^WVxHX!E^xAnIGpl<#a^QKGlY?n;-WpgfLH^s4?=Bk*Rr-Cd-0l@H#g!s~UeGG~Z
zvA}OR3PUyAX&5I(qSinH|EU^(H02WA-Cr(NEO9r0qY9@+@=oVc!-7}3XD(6h)bu|&
zTx+~|+VV<)JxWVqF}WGTLdHI75Ypw67$2f%-$rikh5>)h$maR94Rt!LARhz(KAJ6-
zCrQ@aF)EEVQ@(?@l1lC~kcS6g!9Ur-@_05jwq3WU`KLiLQ2H+U{4}<nTY{_4KDe@6
z?#-cWWP&PX_-lOAkX&s;O$|>3>)7Q}x1gDpaatr<LQRP)G>CTH(J770NB=6b!x!I+
zH$Hw}E=ag6ANqRf{%PuH47j=fQlDR8AaRF)a5ZH~YUH(7Qk-GwJ?AN`z<Z8Z!s7V@
zE2dMFl8gdil$?Q~sy&}lSWzf(yWT(l8ZRp74`E{KikAOu(tf6{M7aL6jKytXGO975
z;<X2k{DzI1bWA*Iy73U|>huQ5#_lSW8)hRM7fy#ZGGqZ;xlC5_<dlqk949h;#q@$n
zcX*KG!a-@vw_Re)QXbN7@!hbxb_#9<eFTiG$S1l7V*Dmtmm(iMI{SP4W7;9K33<&_
z4<q{nCmveE3e_dPmU?&gRPBoiK0`)4A2Ui397tC^iT6VHR(X{ko9SUpu;j!N&Df1j
zkN=;LfanpjQLPJ-j^p9P4rDKFj{xEgyq+TS|FlrU7z9&uvbQgRKf1+#d`lQ96}Q8j
zy+Lz1KFH^oF6nZ^jN7C9{QUKZsGwHwNIQD+)<0+ID*T*t{(Um~O~5b^)*5vLbkf&!
z?jJCef}Sego$xN2`}O}ezm<zo)LKkJLPAsDQ&UttOcF2<3Fk(W-YvFSYO0*1Nexfs
zXVumw(H@6&?iET4Q7|qS0#Dv^u2hj+nz7G(PBL{pNBVm#zYOm*IkCtn%lgX?mj5eo
z&=`%lq@;Q#?dG3Y76?ZA*dmVeMIR6o6IYAv$bo1i_y!AL0_R!YEgHNHIv7XA*QkqF
zJ{%L;QpMG-)-^&WnpT?sD;D{mCF}W<k2@m2ee2lE8paI#ad05ix#h^VD}(=suJ4Yg
z`v2RHtdb~|Yz<UGWt{Azl1iu~*(!S-nQ?H8N+qLYk0U}-$X>^xg=`1MKK43}&DqEO
z()fNp_wV<;|LM`ggLB^R*LcqBx}K%zva+NRnmPRBd1YoT$+|VQKVHKgw(>0EO!CP@
zQZ>o6t{II}AH{ud9bf)u5m-S~H;_bO3!$>|&OkZDqJf0ATZu-pd>Hq3#moMnSY3NL
zGqaXhdg$!!L`~135{{(X!MYrqecWwEVyrp;Ii9x*gEof>VD?taL%V@^Q1`wQwoHVV
z*&3Pab_cgV33zInzo^&^7UrZ-OZalN!#?@k)5+<z_4|o&8sj17TmBOx96Xio;5&}c
z|9E4!9*|CXaT|)bJlB3|JL3VhU=mp!ntp=CY_$LSu=CB5@XP;-3vMVeE2qmS5v6AQ
zJhD076a_Ipb|H<op%p<pG;ljP4gitWK7mW{e_6;A2qs;hx1o2pYZ$?|_H`hQ-_4(l
z&C^oZ*N6n0P3ylU$IQ?9`1)SPjT6=uBk~i%{e>~iBNO{wqO9OM7rcvZdH+*EIiVRW
z5keJz)MI~+nPa&)95`Gauzp0~P{q<K$3hE^+TXH)1j}EI4WKK*|6G#y=a~r0o!vvX
z9?rDxIu-<-<D6~JFVl0%La3)OF>_c0SF8b-+k}eEFlVpT&Y^IdKgmq*&qy(iQ;i>f
zKKu8;-B@LomS!R9FpdHp-rI#cG`^l9jeE><9Nu1c!R>V)fLp>Uf35qMqz?6W`rOZU
zQxA=L{<(7L8LS+UhU-%^llxBiG6g=3rHN016aNNUu19EX0|$dQ*sX8K2_r-kFtm*x
zPD!}6>+e^CfFW`BrK1EDf9opQ!R-XcT)M8kFu#^rNHjSAAn>GEiiao5N9d)U)aj@f
zSLBaQBmP(Zdf)~W!JV_cky)p}AKpT_mk0(Aadd=gKq9+CpgAmG2Aps;UvBDt!{vVH
ztHdy`kIc&-IryKQ7ngwg{n%e^CC|qM3^0lR(ARoFmQ#AbCVw|~*Q+?pTGOJM6pAZ}
z)StdzbE#z4fAt+1+X0DZ{9`u53~1vd4x2_TTLbZk)QV}4$A1mV#hk#>8pnLz$y~|z
z%&|>hxBXZ)CpvYSakx9-a!`w^L(p9y)$-3(`+JR@!66POyy!@KaeyPN^$}L0R#c%r
zfY73^`QJi+@Ut%9)$}F5`LXO4<9VI{Wa|1MWMbyaBGVlY|6Rd>j<Kj)Drj4n(q6Tg
zkz-Fg-j$K2u)rj*8J;*D>O^h&5zrfRHS|Z*?8q_Yn}XGUY&-NMCvXQ-dD?@`m&MF%
z{FamJYWYTTI1&A3clG`$N2QOmE~e5;YAW3P^n};h>S>g$%`C4aRqGldMR#|-C{3Hc
zuP4APV7MJ?@o`={CQ(v=A`O>>4}Kln_x@6Ftd!Bu2T^Jg9@8yN{c{C(KVJak;M1+G
zt-J6qlXAS*AFa)_!^#8ay0W~@0|El7y-xp$azQ*s`U={KzHS<wiB8d$QCff2D){I2
zuK8k9<YX>0>9+GK%0i@nD0WG(xSuF=>^Y@0o$z3C3bc}q^4PxO=jWAaMJEdedXoI=
z<1GVE22eL+QN$6`{OZN3Zb7?>JA<7CmKyx3j|yvuBk)~BKL&ZmhER%)lP>Ga81^&I
z#(O4TI8-XC5m*xtu-R|-L{+}wccmo#BTLB)L%SJ$x=(5D^LoNxG}rsn;E2R}p_B&i
zA3TG4q_9;RkZ3_W!1B|4H2UJ+SKxbfyaya4E9skq5A=v&u-&T>6x@fjQO_>=tvkLc
z`DW(j?QwSw5ES4$UV=d<iyp5}n}>9DcfYa9Er0*{^Vzwo(HD97RX;~kIs^|S)++W{
z?KIWW(Fs?juVXf+(hhXk_+7kU>1w?E=&yl=`Z2XuBJm}vb5W-@6a1Y2-o)FHJCsOL
zGbSF{-%ne)mOhbZM_dW((-q=4Jw1(4WGOmWF`E%S`Kf&5H7n<##^;i0CyH%bUrXgz
ze7S?^=!kalQtD@X;go$NnfB;{ykUC)aad9^#qEOLJlCC7F!HNuE(f_N>kir9UC%l^
ze9b(2>V0Np5kzHZHI!Suue{5~B&@9Rm=<Oy;d6tCH?{!HFzGe6OO3q!RsGM&;*m_>
zId5z5V^j2`!R5u%%02E|3w)@#_>u0~rsZ&vWKAEruZ|AS8~H5T;@WKkWWGM#8yf);
zIi?0*Z*-nvF;$S4zozp}ygq3&-0?$RmTCFw?j|6lAgi)i7Nyo7nzJA|lNR7e<+b}|
z*r08!Kk=C3Gpe_rL4#T3iM48Pk`K*R2(bmr>`#z)cwGbb%)O3@9pEx*Pa0YHX0;k1
zHhPSj_h}0qp)OX#t~XD&rAqNi>XkIhlWN!6R2n{V*YcmUezvtKiy(*BknV3H4&z&&
zVl5HF1(H22w#hCMIjP<zA9QJ60?)Q+YwdJmeWG=<LdQ}Fy^5Zp6&_OE)NhBB4iolH
zF!!ddG&FbT8cg*)h>9(4Q*M^Q$g8v&V7Y|O7cMXoUNVhw{kA+>VrM+2jQGpPLsqm0
z>KF~pw&4NNNS9E0KC^lDcirJtYm9Sb>c>cTO9q+G^5^$>vy-yA0iT9*UC-N=q|k~d
ztheZ!uH`brAu)RD=4ps~j#aG-luX3x12WnY=)06<72W(cGz8?Yp1Uk|*)(6v>M!91
z_7(ULW|}x+mr}4h566?Q67<@+dRW)`ah<7~duN)I#CQ&L&{o^kh$*yDc8x;c=>pWI
zN#Z%17Yl`7_jctO6^v#BT!`lueX=RW#R87de31y01@1D|kPGZ77%O;D5S%s!t|0Q|
zPFfP%Y$<kt5#8<U;~`dI7WGyQzwxv#E8gN1jygi5eT@{F+gR}6luzxHNOr!Di#&Yx
z3$n_l{_L(FFZqvM)LoMQdjlqotZsf?eYy9H%^R$fC42!V3P2lc#0B@ae6!sC>kY8@
zS+PVBmCoAlMGG(@-P9FDcLgd#BW|(IWpjNg3WTh9%UC31{Itr5LB=%=`ho}GQF2;O
zx3o{TU=T+P4aY7ei<>FJYKbhiUt%9}DqM7uxu*Wfy;B`bYyxZ6h~ADk1U*<-x^`bG
zWk+9a&m1NZ<Ow3qK~P@~d$Fxo+9I#Ov1{L)#*Ws;_2d~f&^kB3o$H$tNa+>4$H^yC
zvJ`45-!Xh_@s%UWH=Vq(W;_(b>i05H&W<wYQIeWZqhQg8h}OaC!e^ZO&Qv>+-It;e
zY6ZZ$w7IzE_{j*2;=n;1_!QnB4f$)(J(;CDuM8+UxLIOWzp)dB-9JI$glbPUnTV)G
z5Ar%%g3yU<`ixP`4~N{X@=WyK*npHB1hYe$jGb%)7+FHTE$wv5-jt$(g*CdG^*SZf
zI+#pf?ZIX@ZJ^Qp`Dk*Pip$S5XTMbhy%fTz-#ZACit*8Be9INF8tRK6JXEHNTi!`^
z;iLYF5D#p-^|6mi`<bt@ela{!+_o{QU&SxQb3%uSYgJ>dMegj3cZVi>6Yw*Kim&7B
zKO|YJb-*g84k^vOi@Cq4M48QeFfv1p$Zqef51nYvj8dPMtr`vC1WGlRdlNM7uyc!?
zhKxMQ^jsP-mlq8zGz%n8WlZdwCC^tm#^m%2ALnpw_}h+s6BoqUO(k!d!WY@&ovbWF
zN%a}c>scoQoTCxO3zX5a7Rf`TrDIW4DH#;WjNG(OV-*)AP?_(n@-XCOzg2(|`Asol
zbtVm=G;v!E)O!Qj4{|Mv!q>+Fhg*IJH=vjaN>klgt}N#cJOi#@qGoAo42EA%9vg9f
z3MKEFbV?tdK_V;!1O-*NvLEcQzoA}hPf2mgulym?o3qE$nMoQ<c}{sLtkSDGZf?U~
zH0<`3YbqI5W8vT(x~JCp^Y=q%b6<G!30p#ux7@ySmBIUM;+vKbTkz^WX$*ui2}P<0
zEUbF8s~JVAu3k}T=lc`2o_HCAPhASxSbSeURX;)&SjWce;3k;2>TBrvk!JcvAMZNA
zJDs>!O0gC?y{0hpIorT^fKi`6W*cN8e7p=!G}-E)uskpycBERwS8H0w%Ua%JFaDHg
z`T%d@lUmA~UE8T?VtPH@UpD|{IrG8LD??U#816|N9EC-Xz08)THK0XGOt}T*Z|qfB
zPk#(wB%_AuQ~Daz3<v$=_e|*#CU$W$4$Vp$o{giV9wh`aB6jc&f{$=-Ps~A_k{iL(
zR9LklH}~p=vFP>re#~M%L|&cd_7rHjT~Hi!P62%rS>w`8>QuO_ZS!k!9h*<3Pn#Pz
zdOxQcy<X(G`_=`Uqjc!>-4kl9pToEX9noRT$&>!x6sl;G8j-G@jo$h!<U4ynX|?bh
zPO53I(9qUA&f>%bL^R&n;8zJ%3G^0`yq8aDrh70a`QT_UgY`a#nGX+OEHCarLw^FX
z2^VbbgGi;zo}=bQB@NZh!dso%qW$s;Rg1u|z23zuQPH~KN#DRm@jq6ywHb|I8}b=+
zDX)7S+3x6O(ZKTzpq7gxQy%`{_g+c$8JR0!klPU=7+0SdP52DAsQM62rn3dS*Xxlz
z_O7mF4*vTIe0vtyvDH)gT=ic62^9iaLr(h2UlBcBe_Jgdg3k9@b49U{ODFUA&}<V0
z6%?X7PGhmW;i2VvZgaS-s=?l*BBXOMU8Jm>qK8X7jzBf*IAjh6tj?Y8_dtdcWM)#l
zUR>|rVt3EV#WUCk(`>tA@t>+A&D<6Xjvu~`onjYS<xM<0VUW7f{8=((yd8;XE13K?
zffz@(+e+MbRi~{^^0vs_<X3P<y@Tvtmv}m|qoe-*hJ;ohOR3z+@YW^g{5`_T-&&_#
zlE8j@(>Jt#dg%K>l*R=&rxprquuatd?`Y)>JNJ26y>*Dv#DZn>u`PI4?bhU1?9>pa
zOhjdN`H*MC!Uk%IGFZ(|x1fC0EpHa~8JRfLpG!&GY>t_1{L1~@Sj0b*YGKzUK1|1=
z+4%;2g*nJ227!(ksZ!I#P*%HCM51N0<<c5M*jiw63y<15R33FF1G1ZAnVJGFHZ%=X
z3TJl3Iqqw`T-V$~$eQiEtp<WZ!@3RoC=Il^&#;0-o#=wF#)BVw5FV{QqGmG%jaHZ8
z(Gi02$s=<45R(sIM?CZgMbCjS6?Q!cN5~1?(J(XOgCDvLI!z#?U36)=t(L8uY#ZN`
zoaU;sSdp0u=#t?!xBz15n)2P{BVmsas%jhE2Is8mc2t|ePFEzK@Mg07lNXB>Fc0Wh
zhT|iU3s(6)frXN2{MeE!PlXtwNLjhc4DPVC5O23VE3h;FMOBru617Ihda;I;&|Aq`
zfA-n7-Vgg0>8?h6?cn8{fn(HnG40Xt#ab=zp?fL_LmiKB4WxbcM&it+ye=6`9sQn$
z`Wb?^Xgg_}cXQDt+55-OceCWGbB~sxDtT$RwnWK|Y(7fi*hPa)cOR;)GU1uwhk&E6
z6se<L-J2{p%fRm&XY1+Kgii&qXrvI;Batx@Mo{cbmP9kqTXx9h^Xu9M1D~(~`lPvn
z$p{BnxoHlkr<ir}<vDJnDe*>)z_rh1#9EaK&b}PZ)dqH=7b0MH&^A<K(iC`_+9$c9
zptAdgqLWF-CEFHqCX3y8&5R~}mtA;(&0i5<@lpFZTtu1!QJ@3goK>+fC_S+X^f8vd
zZUS{5QCYqv1v^G@yE@*;3uF=lIqV>45k(}n&2mKnOe_n;Zzk0z1omK>x=D~Sg`rB#
zH`g)xRD9(5aSnbt+m`{NLXYlZXe<q9NECblJs{z4-b35{k^nh3S6AU@%{4-0qD@)k
z2Y{djwtKdjmEr}5kn-DRKA3qn9Sn4{gd2;xGLI`;eayD^CC6eXN6@$>?hDC_2?$Ct
z<Z_*Da%Akzz>4zgLw-7f2?4a{c2je9mW9>G^)<Ok1ozf(wsR*d1otQcND^IUc;nDC
zPh~V2y=F~2l-y6iioBvUX;<HrtZhjk=L$VsZ>I@FGMe5p{HyqNH72eZ8bT`9SDhx?
zedw*yMFs@lz}5E1=JN%JmbQnms-M^3-FS>p82!-;XQe<7m9p+byKZ%;uHzb<!#5lv
zF7Bngb=GhYOoSkURjD+cRU~!G37hIO2D_xY4*z;k_xU^hJ%E>!k94o&w(ct9wW;Ng
z!h<kO4kx-*mh2Rb=uj(fen*g~8umD+Hr}yQ*NRb;P}Rjs=4B<8GV88f++NRBX;5Op
zed&&ZvN2OC-tQtLr|si+Zjs=4Q&`6Z8REfEKu&X1qOQQfGryKHpWw2H#=%c}FAi0@
zZ}yJ_+0}CaD=oc6o6)f8{b=mXc&#IImaGNuBaiawI>dy(&S&4+$*tP4cqUB5?>(x%
zaIp$c!S8pODQIz@Mr=`$!w)he6jwv3b00cXv$aJ7N?L=J-O{m`)5Gc?{}h*kX&m+i
z{#*mqz}0qOM`+tYsrh0E&Ce6HFCDVS-FU>8NKYtHMaZG>kp4#(w|J@-YC^A%1n`kN
z#~P^3U9I+Rn@HmD5j9hUEC!0oCXss*`F`%?R&&JA?DZGXp#NnWyVosm>U;dats|~A
zD~GQhfR+h*kLwK2_qgL-eEJ3Al2_P9{NPpo%lAxS`0AyKt7QZ3YcHb;-8zpVOu1Cq
z@$OP3Gh_!lGb<#|`Nf1|L*sjOIUNo<r?V|ARZ?w}ZBJIHcGip>r(khgr@u0v4BsR(
zM2Z#%j!pVNh8%s!Qw}pf^JLW)ml#oqVc974%{Kw)-sO9%QShbc;GGmIUg2Oz_Jm^G
z8X}<5+<WM1PSaq4h7<sPFcec$JKrb^B=~3s7$M=JxEKvXH;q73BR(J&02PY|1s9Lp
ztL7Y@H_|_`qc4*by7e{F&#}fyoA%1oEghF%3h-aXeZu5!CitMH+cmam-b{g?2D5T@
zEn;8!qV}DA<ThT9S--iUK^x=tK4VJs49RZn6yy<@#0CeNSAuM>HQm}GO1}Jbpt9He
zZKbI<2IG7UHHfKLqe*aU%bWKqpIFnZW8)GyUjLc7gxM?OPvr_a#;UG}+BA2=37f||
zbOzNPg7em~FiQKx(ELG=K+)==ot7h8qsI^hPWd&*B`TKgIF}Q8-N$@$nr*h0l*>;{
zt#6tJ&ae|h1uYoKoX${lgN+(emz1K5)p?=4uCIw?P;u#}Ew8Y5?I=szDjU51%rbVB
z{zWUCz9BG#^(ShJk{_&jeyLIv@6<7{jLAt09C2Ms##uNaudAsxLD7L)TIVw#NWH%)
zF_n$8<mVmop>Dh2)|jKP0SgL{<d4_-;krDk>Aexk?W?x6(JBow7u^0W(anj_1%|Rr
zP<CVzWgZXIja4dploG4zYB=YsuMNWy0cLFh<>uOKGhWnl+i0yKP-sSeIy`tRnmyQx
zn{nfrE6rCd`{(|B9dG26#a!%OWUtHtrhD}%`Q@~{E>LeM`2ftNp*nezw=;)Bp073z
zqpxAEP8!|WX@!#U(b|gECCWpqRLPX?FBWiuP2K_~Co07)WUolPZ>OT#YU8mC6nSH#
zMXsHslA|L!32Xd|M0|I!V}vY&u4S82XsFE|>J*_+Mp)M-q`Ic6J`H2n2hjfgmMBlP
zPbUM`17v<yV{pX0{(D_z41Zcnn+iefwPui|!P-lK{>{nzg(8GbG$t~Jc3a94A=jXf
z;T@9cr<LQ{V#{&9vi*+LX2t3AVzoUzJ-f>Da&tdxJ~Gd(oZGMsJ(Mcq-uJL*iF@;n
zj*K$oUQoa5VWQ30m$DsOpS4A0W|E*4B41;XLMz8qtQNk<TQ~qcu+FJfR8w_R^brdy
zctjIsYh6yZqBTaZyx&%eU)|ZhYlCd@neeRz=0D5z_EOe|Q>SCl)#1pec0VMMO_2a{
zGSoFLi;;&;l=sPZE_(O-_~80QNj?LnOVg@#1?(%Kq)$5uRX-}k(6!uc)f+6tZ6qPU
zx`CAS!c`9;mRlURpR~e|6v6k?vuAmCGsm9V&M+(IB$oCO%mnRs3k2*I<I`e)uY0@1
z4y~@1KJ_j}Z*?b)(vX^lb*&~y)SHpDMSQI@9E(gge+a@z4w3RVG#tC%f7~>R0Y#>a
zVRhx@hxuub?2VK-f5$2Bp-h6`Ai@4z-RvHCsVDfiqSpLQelvqL^iL>KDBe@u2w;>^
zn2{t`JB32!qgwgY&P?q}^ILii5yR)FC=n2Wi2|g9J1)vhw%B5<3gZN;u~{pj9z34t
zI}w{>JsbT^-`sH^ekRSo17{H0jLAWW8Cm*d)@)Mb$xFxd6K*T(Or=%3#$8X7Nhlfa
z&5pUMKj)-umg4<WpLfyY&0J@JgWqB$(Q>mQN^PaYe$=u#<VpBsJX#ssmP1)>5X_H5
z5tC9JV`}m}c5>vXJe%&(SAecu7IYmoI{S5x^T(Y6><wqH92~+5E6`)4I1iO%X}dSO
zE3_VO6jeP`PfurC14z#kAU}Nadip%aJ05#{{nFI9{^yf<Jns(&!_f3Qz#D-fM2>DM
z#!)2DD=XPKVCz-LBIpzqpxc~z((CCS{{6*I*jg0Q3nn=13lUAP#n0kFA||6O5gfO3
z@A-6dSh005T$gHKYU;%H&Qw}uX+f=4^y8y%&-S@5p@=5Ua_y>^y)V4$S@{~3@sALf
zmMSJ9-N&CxviZDnDIdO5bajKbSPu3#tjrDC{DG}Qan4rPP+WJI$HOe8+y;;1`;qHn
z!e%#nKK;bWeR%Nr$?J91pKhatUcLNd*m!$@k{`pvYIRGlYQ}rXp`~2a?$y{Y(l72+
zxs#11z7=*OCC*s9R=W;c^VDiM)L~RrM0HgmGck;jh(y=5*XA1MnlOTkkDqeG;JY=E
zqa7IzF=ob)R-*p$$)^BjRx!MKPWRAcl*ZwD`lEdSr~-Wpf^)uWNfTnx7oZ5=-mRJu
z>eawkdY?7&cuqi=bb}|DJNi=wtCwn}CK{sjE5ftWd_f+@o8Rm`rl&$UPT3u&<T2gY
z3c)$^rTWe1)*CWtmW)+Cm8G&ZqlCl;QN$1%DQvIsDOanpQOfN#gY4^9tEv!N3`OGB
z2gAHcu!6;da|t`zj4;dgCP)5Lo3gCiY=r4HWHR{&u)^1z3j7IU1(;(N6)oAzS9?#P
z0B3UggEOk<5qUhkUonRa!L_~e!a4}fDn5GIu1-BlQk<hL6x8Zf=^6R2wVELUs;mmp
z_wX%5A|;>eM}uq>LgSr60i3pad&#S?H%9NgNg$)Sc)?aQ;q$I@0FM7buEKReLXZqo
zAl69X6N2rK2+BR8;+*_>E0?Ya<-((1sfzqqnr~7vAFj;-L8opICa@aRJQ$69Ne%?@
zGqQX4e!<JOtX!`W)2OElg&GKh-j+BByrg6lF9Z1Qbb0Rk=Xki9&8p{~4_fJuS=MnE
zw6BHxmM9zh&gbS#{**$c<TRBvd6Jc8QZO=g1#?g2?eo>fwWCB_*u%pMmOvH%weQ*h
z<M6J0pU@>oW#pxk9z$h<t`%>E9*qVseoAdiQ8EB^k_U0$(Va|<W}PcLUUThVDC?9w
z@*`3hKk~~BfukBMw7VjW3fe`i6!V|Lt>sJ8F~YZC%=Vpr?uKQ3{wXwSplluOOL)Py
z8c#~$d=4_;$k;0BiD;nwvA~-x)I=cMg67V~ukd9jbP6$=R;iMME)=+Eqg=b61$$E3
zX6`L!*3j8{sC!YB0;*jXg1cuF1&TraGgSET!P|h5SiLgF7S|J@ARACN^JZP_>YdXT
zMe<DoXVIB5SE}-lNo~h^Z$c#zAgx9ZJp2Svqa>GH&9B+2Z*Wx{TgS8yY>EmHC!N0t
zyBD4@jz*3K?sdy$24EAg+Gqp+g_qDs7H9521EZn_D-Qsc)=ckXm8qIN$`5f(Spdwl
zKm!Fkx=i(P#A&~k@$cjA<96;3pzuauP&NbUfEY@d&nvRh9M=exXN}$*!m3i6UH~-r
z0rlJ$)890fYSv!6X4KP3JGMaSwphRtI`}I8J%&bJ!?^tL`%^q>ydV~sZqC)=bjA`s
zf$wJBG#5Z#0%YZgs5rCD5iCW%Zt3R}zO6fogPy$o?zCywSE=eC`gaG@aK4|`*|h!f
zQs3rj(<gM(55clQYx3&B)k#eI;8fC@CA>gVsyTj2f_^P@ZqG_svTJ#mu?t%XYPkjj
zc5%cXFN_#|R{E5@)?_P>Y7gb!YFB4^J5vn8ldDl8!@REf6YLeQ2{D|mXlso19l?(p
z*Vg^)!=esK-MNN?>iZfQxkcNQe=Nq#Sf)5ikK-L&UE>;%kA7Iz+{8cHGhc!6k!S*e
zRIy9XXC4U`NGGnXb#`fcaL7lv)L>F@7^h)~CSxTzD)!i3wHrU!8H58LgoU<APaFV#
z!tHn_3Tm{Aq(*wa;?pBOT464guC}DFLHDd=Q1DZadG(&{T6WT3wf5a8?Gd=^BkquY
zw>&s8a7G2!q|6#QXl^j3x?Hz&(WN?fx<cPDQhEMFrx|r+3N(|P`A9fBB$$@R7Ac$5
ztWo>1wAaL#l{7)rU$;OWPyZW^y0V(_Si%XGwF?Dh?&U4QSK%|&L5M;KS_YFb=;VeQ
z-s48~^8Qdydl02=#_-4DoDIW7H{|*e=%r>w-tu#5pEFliyN#w8?4qsnXCxU5p1zay
z@CVF4UtPJ>rb@go7xW1t(zz)SB)Fww)^4FjpQv#RsexY~=|SK+6QA>zy(xXl9aFhB
z-5ju0VJWdx>+xlDrNOoiJGiu*A0gsIy*fw=GqVkhs>aJA;wwF7nmRR<Zimcy?^V=i
zHS;2HDV_-+EHO^Zqy{7rVwaYjZ7uxBhU8&U*A>30`-MZ-KPn2BKUbyn8EjT;(TJh$
zw9ncUmkxg2ru7EXrI*HW;SjVcLDhG>6H!(jytb83K?zSlqdqez$HWuv+ii(1KG;dC
zTAVa!f3>AX^Vy|-^KCPXju{?LJI*g}Z}kQLesC(!J{nk#HLpGMzrxz>I$y-0!K2Z#
z!M$nJx6#?+N7xay3AyEuSUs2fubqGlxm~Qr_^&oeP8A?F7|*%|cSNaeCInm?4VSIu
z83rShu&{D{S_mjf1M5;vObSKoi^>+vfs#V<>eWzhmE!Hr!FKz=U1HR4rJBjhJl!Ic
zA@i(YkOi}3e@aw(rP-tReea7l4WiL%Y*Rpfy#awhFQ+nGEbAvy0)EI&?e*ZTW1S7f
zkI$ef6hZQsruI*6L}5Ir4QH=rUDaPBO|&-(Vn)i69brTq4Oh8R{rz2gr{G>?b<B3H
zUf}(ry+E?t`P16;$p-f)yJYZ{bkZ~4?Lyh-`g@-|7$Hwq(|}xx52PeQSKgOsexI2J
zW;c}&4GC}{HcA~N>=+j_HXQqj?C^md=(er7cErI}O+>=nFlsD;>pCTwoS{Z6*g`=p
zVq85a4jS>J(9cyL!G<`VNQ@n*cJJS+)&nBrw(y_$bGlE~sDiL!3D;yZ-5zonyU0m)
z;lhIL!iKT6>%4u^alRDn#1N@k>i%Ac%~rr<1O)dQLBnfO)ymNQ2+-+@+hx0B(xy4C
zX-)R}2JFskq|ZoT=xrmThHii;0MOko5F8i{9GvkFzn!j72$Q5d2?0>n+007NdwHqU
zxm_ohsa`k-Q2j5AEh1qh#H7ewZlH}kh{1GUDj)VvCTCKlsBh{)mT%3zr?=k6r|({v
z`|;6~-!ak&)8G#`?wVx0dOVZE6aWdzXWOu`Ua+l_xhp?gFlYmEwg)~X_AlPxEPX16
zNDV5J(}RmIb8S&x*jFBH%33AZN;VQE7MT?6h<rp*N-mQ)EYH2@LllPmG;!kMmML0O
zxjxXl<Wj&n6!CPeO-WZznepCx(5?zf{G_NK#+WY04wy42<%k!x`wwz4Ed9rf&*rU9
zhe`RZWxRdcssBAQbpIB0$>-<I2KpDaW`)#Owfke`0yjL!Q|<a!N>1Tz67Legh#~i(
zv)YZIYWe`bH6e{S6u_$E7U1TqJ;M!}K&KPX9Uu%8AyQt*Ld+0PP!7Ci`T8DaAb)Me
z?F#>dRDZZ{3pHDY#_L0Vyo+;hPTU46#7Y3))wMm&@D1}3qRx3G5*R(v@g7t~J|tVI
z`w_kee%q(r7s|Sm#%JcH`IG4$hD$~eM**W@e!Sc9!V)=8zsRlj#?k9Ce*MSO&2%&<
z!Ayb{b|KY%ARUxoa2XB{x4VI|wWN%NnzgnvNp!x-Wc0;L?3<}Tqod`leCf@OGqyHE
zf`s~|7i~`5@2_9umV0LI;^HEaT=b{a%2Bd=Pd-6%4Ije|o~km~+q@uvuMvAdO;i@?
z;v_H#dxy-I?8`)NoY^Jmx6pQgU|0}Llhe~Mo=y%nQjl(aw$G_4h+%;OAnn@PBZMQA
z4ZPGAxxsujOb}M;;1Z$&x{P;$4!`Aq#c=(UtHl}?=hIGbZ%1tv&3TqcK9YIC7Nroa
z=DP4hag;V);xo8mhLvQMSLi&#UK+7$D)LjVONnHyI?8RaKZgQzaxCDv8kCLz_eIwD
z7t>Kuvc0`2MEyw8P1=XO==Sb;<~7gD#KqE8bOW*sVm`#A@V%9FCQXL7Y~Tx;DSan{
zP3Qb#fSb9SdN)NrR0i(~ggOb+_$lu-`LA9QHa81rTQfi<GcjQPNX1YqJ|bIi%Pni@
zep+0`L_U1b{8(!<VpTsXHr0D*gz#k>&H3IwIGXI1UG1iY;B;I4F6K7<P=`ROq7>YG
zK#L#g_^LY3UsQPlrLFE|flZVxd$eReI2czgZm#ml6!k+fy-y_8Q)6*FXCTJpWx%)f
zqAuUovjamzNBu;UH9Viiv!i?pM9eu&ZM6||S@F(1rkEe!h_?y~3LuCxHcNESqAKMK
zQBe#~hQeEBGVBZKFQo2^N=6nKq2q^0J3#DH#x;aLyc06)Y0v^iY0&5Nr+m;<^J*~)
zXurvT#(ffvh~C=B8v!OjtNj7CSC!^7^e{{aL0e%^{9A?|O+>f@BR`YrIR6{;nr;uc
zf4*q+^ao91TcC-B*?BemyxA#{!D8IP@GF0+3OB!OSVT(7#%H4f;|GD-#@0lP=G{==
z&d^wg_Kj2@99D+Xf3+b2N<YbP6u#|KAS!ephj%wG&{8?yqKLxb=kGKR9D5qh7Aq&+
zntk!BV=<Dkw}9|-1E-#(zm->jR(K%FbtZS8_6xRngM%*m9tOo+aXABea3VtWLAJjl
zo>L{Qo5N$oH1XlyRWjsO_MM~*?RGdQ_DsW4>y;o8KAqlR){$AKo`he^k?E;hg5P0+
zTNXs~3b$8tDMdqznG+|D3miWmh`#1;Igco)LVh)rICSghO7u0_$V5uD1whrCBe?9D
zm@kVt2Hxu#MJ0!9+tP&+j9x~)3)kH!A*FM3yF~q$>x+8sn2G~3>IvFrIa;?$J@9+S
z+lwxvs)4=|OG{OF8S-R|{<(hnF-5GIM+BJl0H4AKeTk8jWbE``pY2>-{>nW#!a!4c
z3a;z*5uHK5+Y+Ueff$;P;Q5<f9YEH-)0|K~FzSO^A1*m&m))><B@xz@yHM7^O~@Ar
zLQDpuOLdT$^}Sow>!;9^RJjGs;nsKG`>obbAkfaX@}p_(LHXGn*WO5+FcvITn{#EJ
ztp$b2jGcV=yc^;MQYhb;1vWkPfe(b;kq4t&&w5&N&Evy=5kjNyf;Pi1X!czuLMkkB
z)^C|_b5w{cIG?wIJe|xdVit-ObSZgd?$ab5-`My00k3NgXS|HLBW7^c|HJKb7oYC=
z>?HzH9QP65nYn4>c#wYf1;~@q@@JK3%V8s;O_VnwwK8a$d*_=A07Ox-M|4CT=bWMJ
zgA4HKl-wczq!Y<lmjwMm#sslQ86c)E!Gb1>*B6IXkUa*F-!KxJlV_@k9Gz+F!hoO2
z&!d(t0w$BtZ{nNMaD(P)k5-M7Lu`K@(QY2W;V%A2W-_5B;U0JfnF)V0gfBaFChy=a
z6TrQ?_?i0;9B%SP&;d22Z|_dAD4%lcU6Ig=>hGUwKB}d!EpD*HhfHcfNowgkLVEvT
z)Bur4X7tv@Ns8*$@*H9^bINkMYT)rk1RiotuYbAjG5(zSdrg{UieIGv@>s)3pU!mj
zw2`Qy=am-krd#XZoWB9z-59>(q~Ge7Ygrs`9zz!zPl|V13p@uMbo=yv>)>)<ar0<2
zt?jkkznE|7MJ)28`Kb9d^Yz22psgdQ9bsl9LL*u(h*b-G<31!T=AE;2{WV69oJ*sp
zX_Rs<;>#2iddlq3n~O_5d3J6-(|@p)AiZ55ZRb7Rb2K_42pg*Lj-Nzu1wXkgyXImZ
zMJJt9BgsLmb^c^W&~zcz)t*qCaX_tOWK9|ASYqX>lyP^VnMuX~DrX|=)PcDAQo8ko
zI{x|e(2~7Lnx}LJE@r0HRBY@+A=m7TU(g^>?g#8CYZ)jvqvccIelWICjQux%>^Q4@
zA?zac_?T=HUOV$2xHZoe@pLAVKkJx*aCtHQ%UI#d!>{@)G*i_M{f1qDu4My=eSBIX
zP&|#f&_?UgawSr0ddmOBaNNN>2Y&r8ME%X#&fc5egpwT%lRST5CBUiw`+qkEK#^XD
zO6jDW+jmBo=pxFnKQS;6kG`(=67vfj^Z)%5T(j%(+8w!rpk#B1S@8?398ek#IXnMX
zIuw+Zvm+E8BCDQbp~`$D9*)xu&v7XE2i>?MxQMSq(owhf`nE*?4L*w8OK>@3r9jl-
zUZ7@YafunEFdpj(=u<D=-<1EC7C`XseH>w5y7nueV}9Yh_aEp$-UINz_h#G3KR7{Y
z?dX1vFoi0Zg#<gAZ%#JjN9SK~;PxB+G=m~UVMl5$r;QRZ7XibvCgB1(P1Nq`Hl?Nl
z^kTwFw+ZErNcPC2h7g51eF!5Y-Gk@vNdcc=!U`&Th1;KX?M@K#o;ZBY{yg*lf)|8Z
zh9`UHD{lT9#2W0(!Vzhc`MnIhtur%ik5rk|1vt-%ZG8I|KM#49{k!|psM!;>kSun@
zh2w|*a}NOySdxhVGZ>Drp!tqxnSgth8N^t)A3dzfY@om_E(R8iMP|S5X#~@t2hRT}
z+<;*mw{>u4&p#*ag>HIP(vkTWgrp7;7M?4zb?y1<mz{&5m7F|RPZOnD$gz)u@Rdv1
z;+^&a&;L0jQ2t$yzuwC|(r9_v=#7CUpp35D#W5B@D+Ox9xonVLRG?u8IMo*I^j;yF
zg)-x)1Aptk&P=*Hi=*R*tQS_sQ)X^$2ZHw>>OdM&<bIi8p3xaNK;!dtq*3dQ_Q^k&
z_>&KBfBB79%%!D0lg0UsQteld-B4yu@~}kQ`Ndt_hQ<?_wxQio-K>^c9moI8rh;ni
zo?V(6c5>RaYBbq1^KRdXK)_1Cx1YGRc3`VC;|a%sYe6?3?-p-;$aCcg&+|(yCatZ!
zE~(jl35Nf^7STq*V0YJNEu^@(bi3{&jN0*I@3-kkjaQ(URr~gU5Gwz}&-<7U0)7|z
zzKnvTNei<KJNnYqI?ER9^Rj=)rne)5*gAl!YV4065)RIe`6;|J*Y^P+J*y|w+<lR6
zD(me{4$n%k{W544`5nP`m<Toh3y&O}vA-k8q}<i8%Q}Rn+bXTqqbLAmLZ)PIxCAn8
zCy%P5Hk-FhJf3oR-UjYWd#aBc)X@^odb>pG3)^Eir`zuh-+uDXZth@C<?X05adC;5
zQ<LiI>bh_F>dNh-ozG^MR=2N*7$0+zrG`+~S;~c5hK4qlRol0b;~k*yBd>)4v4ksi
z0am<2vOMR*{)L%N-^omYR|9Mu-+%7s2{v#Gzq=i@zxbVVR&+PWpF{xq_IIxJufqW@
z5BN@A@Vkj8K65eq9_j!X=s_T$VdwbaA9Vk}{vaJ(pXK=0tjSB)jO%n-Gu23_G+LIw
zOo{NHWc>g6irw4WbEZv*HTVs4{Rs8l+|{e!ul@Ju_=j;}#opf|Yr)4jYQInSU*G-f
zVSoP<e1DrkvS9RV)9^9$8_VX!=`4Pd7nt+(-=E^&IPx-cl0pS+udVXx{~@)o&BJ9O
zVRn5M+1Bj1Zmn_cF&a7`#P;W-ubD#|q(2nEwpq^(j;?WWafhB?dbGc&yFiMty=141
z#f0`=H8hO9?Sq#nRp{wL!6P1<sIc5A;+z3`3;*y1d7nA92gqvc*GD!5|6R4<Tf7}s
zf#&9@j}`{90o3)^7AU_2=oP~pz054}qZ?ciw&#z#@_6okoeIoK@NT{iD}UmE0Pj0)
zpKBSpm0AKVGhIXE@GQHi#p&YwYA%Y;hg;s4fsF?@*0L-<j#o;YiQ4dv8u}V7RXyER
zDR|o4!!z=)<KXA`c-PpDnaXvp5Ft8;EAU~1zr!8hf20Cw@m%`X6Q0{ws2`tL=g*B<
zQx{r%6m1DJGxEV7LAB~GTwZ%OFfJ~2JssUU(%1G**csfBSn-Y(q+;yUxBjEiMojck
z?%GwqGhn3h-<7-$;FPjvHP@>W->`lS+xtc!rcBHztM={lG-~CScfdx1&8H@2Af+o$
zQ4&G5nZo}b`Gk^qWqX@Lz#fGPv>fZuVjl1~y<b5#HR+$z53S=}_LGr;sBKdFM#3vS
zJ85*<Vyw>1`CmcK4JNvCXM2L#AwW$>J_(OMU5=u77(={{1VJ6PDFFn$Hoo~1w4tR_
z)0?ohCmr+M$k1csQI<=a^u51p_2fW@#S5LdPtiw0**ulo1O2NNNL}5RWE&sx{pS2m
zUWRHsa1Y=$HXzR0WdtgA#b`jV&rNtvu!GAs<#tdh>>N+WjiRTdT_^c4-zWVJV0zM}
ze;L5o6NS#?Q~dIUh#YpQ>(<!r?$`Ebn}45>#zd&Px|8P)OS*G+`;~2%{g)WT!EzxR
zTT7^bjZ>BNPwr>R*Lr2fk@QRE;>L484sr#u3+^KxmeQFw-_AS`4>WiPs4YRzVAhg&
zRN(YvmA!r-{bhIiwe8RR4`bF*ePu(Of>2kdhmQ({SMMUgKG-{J2)UwbD%wN;5c+S3
zFbnV)qq2|8C=pZjbU0DORMu_q!H!_s?n%6yw#O>oPCVCd6;6LI9DIuh9K_v5&lTg)
zLC+-y)2OZ{3_G)KyV)-kKn<as>-<H14L*0UBQny<YXclaWKHTl9i8JaFM~gXAI)I@
z!0B1LtU%l8u_M8wC(lFyl93FsGL5o4t$93v0>9*iBW53TAJBp)Nqi~L3~Y7zbz-(3
z>B}zrprej#XmeDy1@MY%-lK;+_Wa3(p71oz*hL23nYYUh^xi)y`>kj4Fvo$M9i4gO
zH8Le1x>%f{{o$wN3Z0I~);?YF8rtd8()*cb;r>e|<G)$}7htL^ek&@1*}py6q<$US
zW~m=$?(N&RS1FrwvFOnLuRLr`QfFd+A$m#7WWBF>$V5tiF7hp7CKusp3o32<frT~t
zVmK`nZVA<RmYFSS{LagSCM`2~=*M1nq1UHYm*WYg)8Bp_h!+>@xJ#39VI}hNidS7+
zbS(tqm#lOw1ofwWJ%PA*^isPJy`{Td&(uUihu+@%`gO)<4zZ4#;1CMOKG0xsKWXK$
zl#Ga!ND$h+(ta+!M7W?C-z&P;$g5`e<{y$OFx`PWzOWu|xy{b`aq-6t?ArM|zvH5g
zQ|a4JKZm>GhRo#v_vbqY&()6!I<3zSl?IXVz0LJFW$F1Jajp~h;WO+8Yma8S<38jN
zgsygCwxvdXJ!9p8XWWa`Ri-8;L#Kb88(<w8q3OeRlA4S?-GFPkx~)c|j(sh6@*We{
zX{W^ii1e7j$+2web5g5P#RxSxg(mKAEC1U81>fhra$v|ujr*8{&5^&}@Eo5TGIg=I
zd+}<th6l9lQu;T*(Tyjuz9UZ^;W;Z>LqVpN1#-kUB*u;#Kfa#*&dN&TFiH0N+S}IO
z`v{nqqO~(VY2#IAN`AS4bQ%lCYCKM1>Lt$=u@33_bBDmY2eZnje-u}Lay>Zt=G!m>
z>iYn{FHv%&vjM3qxZCf!bAD5zpo=W8S&0w)gr*nWy5GF58}j9}>3p4a&fW^af7ajr
z3re1T@w`x6cD60qe){xWe`d9_yzT8TE-FQZoqKqc2&vr1ts_4Z9=E>x@L^)&%xU8)
zvs#L6{yG7pb2-YS+}xx0V=iQ7xe}<M$W04*qGfFnPQ*dC&{u@b=7tcPW+NcmoT+3B
zn;StX0*bF52-wds(XU55m!1`^l9*0M*3-O9@8R~OxPrYLE%>#U#X<DUJE62sqBY-#
zdfj2k)q6$?)NcwC6Mm)fAfHj<xpHWzU9(5a{CkL&jx$xALXuPIAzozXesg3x%LeoL
z!9s}?+FzY2NcL{?H+CXL1Dp{T6tTRLlL2A6oUoA`F?dCtyr3hf+9gK9;@ZL;{D8e$
zhzLn|!e0ZTw%4%hf+f!Ct{3Nq^X%B*;v<(LN+SYZvyi;#A7yJ_i%6u4)Li{+zjCYE
zEGqR>T>-(Y@`KDuJxDwSxtt0VPS*>(GQ#;A-O2wLqYL3TIDq5aCD?QUpg01ULTFhC
zE~T9S#sRoy!Cz0Y>=qv#Hix{j7Q8NzVRphtP0tjHtCh2?Sq`;`xp4)qJnR0)U~qU|
zPQMjO6-`QKgvr&^1$OA~Y@u{z;KR>W40aEl)>k9G_-Il(0Hf5fhU<=2*CT*lvnAZ4
zYRdvdS9vyDqW%!u#M;he7N_14f*?AWjbxt4sK0I+1sB!$l$#s3lI$XcK0il|i+koz
zAv-_6bksYUm6LDK;Q=1Q;i_F;Q@{Qmn{&Wyt|v3>r5WroTnd9j9`VY<l9n*7-Mzhi
zJGSC21EiJ$NYD51%R|d>-p!uXzvNsRKkd+D4Efy>WY*f&DS0NtrkiWV)QVbSO=>N1
z=ilu<El{B!$CfVLr@xmGVzy`RuZ_O3k2&`65Ox%`+2^C_kLswmpi%<wnMGUTqU2Cy
zuj>0JS|vz?5bA&wB4zGgEu*p2hWx@XaP7j|!aYJA?X}n)sKF7O*crJsjO?YIgv0{@
zNn(b%A9mXv9EF{8fwvFIKa=$f9S@CCkFa#c@UX@^DQ8#(Aqe}vzWjR4!NK)xZbV!^
zJ!~((H3Er+thGk)NGEycXh)f>$zPDA24+<n-vZMaAJ`i$Z(n=>(9!%2(Aj+oykq*1
z(4OFLfM2h_aH}HA&-6RD1_@%k9@QIdy%}adHt_S+Zx4Gym}v`RxO4Z_Yej-fR~2O`
z-vzqC05k~AI~&0M{mA%0&X&OH#S)iJ7V^gN5|3}{@WpYYzT4%UgczUU5t*)a#r9f~
zpPN?;jdBe8>gm%T2{!c*%D1{2a`-~V*jF|kukuO62ixS@oDZh4_Nn>yOK{!r%o#=J
zS+VlIcf-cgUS4^+jIBcCg{ww;6(8`jG{+^w?&BN_8DkA;#cS36b{U`aakRDI211cb
z`N$BO=|11@7mkKjtJ3Nx_D}xmry6q{-};n$S}!UF3NQs`FB4^GtY~`|HR49cG_$><
zh+NMBwIi>W`DlJ_&|mKQ9PgEVLs1%-$#SOg^3x+k8EkFYnHVcnbN$jS@+{+`ZnOr@
zW+X)G9Dc1};ohO7(VDm?;c-8^u_I-{^PkENJH=1UH$PsPxOJ!WCrj@ZaWdRna9D9r
zjL~HyW!*N!>b<43R+Gy%y21A1ke7&j9cOnt!!NQ%odAl4w`LY<zFw-fK--Tnd~_+P
zBE!`Ur-=}&ZEJhtKGRY`-yXp}p#ARVq<Hb)^zrB*YIv`%)ND6rz=y<la}SB8&UNSx
zuE|Hg4fmViAbPCC;Ar5XLB3<J*ka60L<p5ghbmz%b#DH19h)>Py;Ovc)Q<IagppyU
zJI5Cfw`TL9OsAP?1Dxx@+G%CU>vAXU%0?XJ$}3aVS4nFbsHJ1CCeh#Y;4?cN6zAM-
z#Y)8NZ;UM;q2G{)1$^lG^x>KW+rT*=#ux7iVf)m@A6Gqeih)9Kw-P>It;T%GPZ#p8
znM%aR%y;}!a(rP>N2`fU8gb$$^0#<X$5ow7K(Yu#6*O>c90$Uork5)3n%urgu;*JZ
z#~w=vM`xTrZ~QAT2B?M`&+7vO6JWwOAm%1%mn!_^phQeiNv-@Sa*iwH(ngl4Vs|&X
z$}9aY2z2K>KvmLwL|R5IlCxG);I4PXoH$aZwTrJ9ty`^Ih>{5la5JXWFcuH(S>~R_
zV)e^2QDi~E<b@axKXntfq!DU+?Q#-Q&N4oLPEhS+3YZ9&L7){NTMVkdk>?FdBXv1_
zwF(+d;?4A-W^jcw+Yyu{uW52TF1qk4`MpiEbWMr+Cf7F7KG=VkwO~oRH`Tv1*$<@R
z1|XDI`QFgRF?pqqCY_GpxuhU|eMpo|NtwCl+pZGMS@vGMvh=5zIsx0iWyMseX`|e5
z+z7=P#%6uTX{_9&!tFVtBXYAAOE-)1Ko_aDzC5gcdp*l$9>4T#ud|B%4MhLF>Q4(r
z3QqHc+Xz|ZQT7<q87&|}_<3!y`AvXi@+c~yj(p4a;k<nL`LWyZ2BQ4V%^&fxdGoY&
z<)o#+j&!D#WYW?`_G7luiDr|pVb5wyR4Ly$6#+Op(3{+*$jBD;U*LM5H}lSS(;(JX
zF32?_10wGU9V3mqhOorLy&i466F`Vna0BpK5HyEXnWqHXKHg{gCe@YoAovN>wr&Vd
z^2H<3*Toz~93tm)a_M4vWe1zrq<I9_%lW<uy|(mNCFs$A9c`Nv>HBbdUhoaITYcR>
z@U9rO2+;ciP}pJ5vV618UUH$`5>kc|hLMMK6Jzq7E9$TC3-Q$;SDI#4n`)kT8>zKr
zKY*;m$MF|*w|IrzjM&$7K?|V40##)(6vLK;MF7LyRRDvP59pOA7AC$lfe{jHR$z>c
zQQloh<-^5{Dux%?x;OqD5NH(I+pISTBN(lG>4p$$vX#0#p(Z6<$g)UZtKWvHF(-%|
z)N}aaWkl<|o%g;C*JKsMXL98SY3}}9EK>nt&TlC|(1J!SQ)oE?gbaYJ(KZ)X>om_f
z#DUocqdVl)7bO2{)ef82`f#up3&fNz+pLXWt8nm0wDDaH=*`Rf7*8Zl6xcl5)iIM%
zL2<unWH;Zjt`em|FS69qagw2vrnn*^7>bE4_EDRBmWdKcdfMTei#Tt5t8TmB(&b>b
zIV6Itn|%2S`|xlbORwBXWWX4g(8obXK?U8mS~$TmblDXle1(|FUpzGU7-Q`a=C?j8
zL+M%_3A6A+YD)NQ6}}A}=#gtwjY#*U|5B(JV=a2V4F80QntbWXae!fc{wS#9Y=a`$
zp7nsrW*u#1j3MTY2@xv)!o7d+0q$!@B&R?h_JaP7_TYlw=F3`qRQ(EOVKQJSt{tZC
z>NYG7+HIvHVvUC84q)EHPQnVKZ9U5R>)B)UQmH%nt23c;O>ykyo#eBb*kUCHJgKf?
zL@Z(BJ67U6<2io<<o+6csc<;%+_&`w*g1!XcapoI15y?q6Hd|Q2vJjmr7X^_D*1}q
z&1d&+*(f(2KV=!X;nH#+^=f%uaOtOygzj>@Q($0#SN3cSYja#di6IC&E?qX{f<TxA
zfHP)WX?+$R0TLJtvw&UWnO%i}-h5YO8TRk`%S>Sqdn(F%uH@kH+SHXL6~bj!jyGwb
z=GIkhTT-!ybzR?-{_zx2<Dt&XQSsfY<@D+FnX$m|6R92<S-+P8v@R0q;}bc(4umFL
zzf*LcBvX+?7+dc~mQ75=<z_vVlC0E+8{LUQG+46W4vI>5<X|~#>oCPiS-AYVK~Kdw
zYMxSt+N`;QV(_YUC6&59C$6G9VB)vo*gS}9o?PD2!=Go$-wUDtczNsw%ze@+kPgDG
z{(>CT=VihWYsOA`Rfhg<PkF7V-nS5uv_OU-h;Q-v#c_IFhfvuoSi`}*C#2gZ>WaX;
zMRHVFUDVzrWCVD+trpVX+TU7F$i=Qi9#Av)R66hcd|QR)6{nV5vuPdY5q#>#e%iY9
zSDguZ@YJyTSk$L;HFYWbw{s(?1&ecydh-Z%#o@-!)6=*kGZ(U_jf*@h{TWk`b1+6m
zLfPI~oDn72h&YgxFh#ixCsh~qo`KykC922d?tPpW3?~E^RN@uHqc-QUWfd9qDbf(!
zC3(9oJZ2=Opf=G7rnpB`xzDA`^#e`^R!grnSNH3#TAlmYujo_Ewq|)YltUl*xm*`D
zWFj(fj(dJEKckm*1(#e;un6#|`e<fs$LFHx`ZeJ!bI72064p~YDe%gg4xV?9-?0nL
z9Y+WTwBrNFZG8mjxnPog2tx>iwkw|dZn?)H^y{>TrUv_ZmWCA2>!}_;t(~q5%BRV!
z9vq&jLFA5&FHP%wOE1FH(PzGG?XyGdJNmQZV0`F<UwP-7N@$TF)5vO<8u{UahQ^Ti
zky%ea;rKMW#^|BRRecFc<3#(Q=lS{#fgL#Cu;&>}fvC>qt$r2mx)U$TBha~zM|J_K
znC1(yuDlR2E%MIYr#3fth4dAMbFs4QyM5`V=FJ1IM&JPFyW_1y^k%C{eE-5Qp>&d>
zmKmRF5tv-qB)3#sY_*~o@cCu&iPm6*PZG6#KR*^RRyGv&iAPviD|Mjk&G!X$yTSYb
z&+$p)#A#g9&o`&^yd3Xzej^D&T$?`M$?w1Z$yC0mD9C7HVE`We`t|E9S^ueH$mVOi
zebjx|CbPc{mOoJui$QGVVe`WycE`Yldq!fVkbG_?HPUs-2Yd2v7drkF<ZB7T>0RNG
z2`jQ`me4wt-}th>^d{vdTak2o`|9$&JiS*;%wjv3BRQV2=Z@7^jpL3Xcov0qEn|}}
z1w7O{!V>s|#fZo4$_{4q#Y^vZFsDh1rIjB&p+P%0M%FwVa*<g=hbgS*M*UT0$TunA
zd3J1nSiMf2d+|{3HC^<kHLW&tAZtt6I?ot7wKM{M=EzZx>!~B$wu;wC+w<yGwg&5i
z`$boeC>xW|Gdu4Yr_a>gSK?gW47piZ<Xqkp`DvqHpYisXq{1sT<x?*{+?LUfeO$ok
z?v|euMSdDuIxc?m`|hl{)#<!jXJ#(?EV=Y_&ySj<HPr6Zo{9#mndQOPg1cfBuikX>
zOHMIKqF8m$qpkW(PsSJVpW;L0?gG<c$AOzSdlBb~>yR2S5jm_T<tg{Ny=)Ac#we&6
z!4!QiJNwWmQ`uHfs-G1>fBLvq^qb}m7HHi;W(SAdFK#zd^uP3m#s!NOT_)ez@JI8D
z4X?g4UHlq?TV*@K#r%WiBf|7-{{PVR7f?}s-}^WY;~)b_Bd91TT?zt{gQQ4zx0JMi
zG(&@gASoc-4T3bxkRpg6(lvy1cMb5r7%zQ){@>qPvs`z{%)R&Qv-h*>Ip+lH5}={u
z6B$x#30Vyd`L|_QT?2XI$0E`>g1XA`d>=!6(5^nUzCNnExTkMVs<Clb&$LK!cDW1H
zv!~ARKI7}mG|&uG*7n)(*+WzDaMy;Ys&o5d1jtHA*{w%u^e9L9&o>uwEf4D11|n{Z
z*bLW}HiweJZ@EY$3sCx4k&H&4bL<TJD9v26atRf4IWghb-g+U^p+lr6P&L4csuK=7
z)a?u&+5MzEvm_Vpq;d3|X|$;Am79sbgm({u9Cuy5OF5t~uBX1#R440EUm*17OwZf~
zRpSkRjI$HOF+aK%u4B%|%Q_WEe#w?fy&ZOr^>#76^aV*zRrjZxVA0oT;0wk23C$uT
zkz(u1nG5N)8VkLoTbIV^n}IU8N@Gl*Iw#B@d_!n)B7k_&z#u;#=>(U}np+3MncBCY
zE^-@uQtc9_>V7NzN@@&IrPd$&N^M4azj9KGS1N9W(?Lf@!p^={eP$+;VmCTIqB(R!
zs%pM3)2~s}9!h)c{KnNpgmRSOGW9&zl$6q^r+F^|g!5z7g{^rvIFMXe_Mxoq7}+Ho
zb*^Jq3QHcq6plY1`ylZ;BE(|uxH`on2Yw;ZGoqAm=wj8jO^*?&r&D|Q5POZ+hC$e3
zME|kd>D`@AjN6NyUV58Dp|(@cvN2@TEiI)8z4t`wFgXi*a)s|rRQqh(fic|x`zR$Q
zz*jfX5XxjRXT?4-V!MbqR8#DOoX0J-TAr)XJDo5`us{N6&>^S{*O3Y{IIST+^>m$i
z2~tkDi@qIZm?MMxq?(fz^GZ&@$5RGTg<*L5f`>TKG|En$$)2d24BDqjys=TcHXDJq
z;9~I6VdxYY<MfOqEN6f&?6bdZ<Xr(KiQ`4NHR&f8{cnk@oImN;J#1^0VVR{x>Mwg4
zU{=mmE^$O(q)!`MYEqpUXEcjf(7a^b)X?f%B30jGjVEuq-C{znHM8f_=$0NpeGU%4
zArQW58u|JgZ<gxpb$`8EeB`cor4H<n(te07R>i?<=mhp6B7!P5j3{}^+BHM40_h$f
zJ<3Q3>BLQT$aa}hE$i{pc9|TEibDf?!<x9>X*?yIffB`*f6)LVi4Yu+=7q>_SEh#s
z_KWrL4YEr#O~ex?Vii8BgVQo42y~m`FOkU~74@lK+}&>#PAGa5c2N?Dd__w=0UYHU
zng)aNd+&xj!S<F=IjzRRm>mX9^m%Qhn>IRArYeRt@Au-4t*RtzF^rFRI|8Es+M8Do
z^8j1FA6-$-1N}w|!a|L_J2#nhquiZrO%_w7xU1+k9r^mwpr;7$qlB_)#>(Rv9p_XL
zuZQx1FLMsmO6K_5Cm`z@NF_rT(Ywbb*PW|<OLz_!?TT=BuSOxF;FFf&&fUVvwsyyC
z)nmu&eHw3W<mVR#=U%pMYFujne>y;Qn0xZM!!EZ6GZ88G0L;)1zkf^1IgkpQ_4aQ)
z5R~!>ydD!n(HAkBl^U0P&$Od&u8C-yEiYU5#GWtg>@G^~huuJV_2YaViPKecD<`_m
z%E!|=UMvml7K83~lZ|xe>uY+m<sSF4qFm3aNp{_ext7?w@EdLzXQX}l@@&*&CS-9d
zP4CQ!Q22brc9dRk-+H<_xU}7%=A}e7WhC3PA+rc!r26{~FOAiSRm_tsZ!jZ>zT?AP
zHRsZdX)yHMvVFJBaaQnrD7L#x!&3JN6$zHe41Pz$8}74v=--jmyDZ|S(0YMYG6*z@
zC>M0)On?u%G}U>{-O9S;UCs}(aD3Bjf9g&eC~P#z0K>GY&yS;FZZ@6u;G{eL?AZ?_
z?f6ZHR%P9Tl(8ayi{X4P)X<vK{xOqflbvcL8*=G&n`2Kr)02hn*z88PFBoUBv8rPi
zhzyf<5FDGY!?&7ktB?c#M9uA#1!pT|Vv4Q(&`LRc$=~sW3pV@gb6DU-qua~1<j3KT
zXL=rZ0}Ct~rg{|f)59*GkM-uzW*_aSw=g*X+lSEbdmRc6ksOtJjwzd6fBg9~)9zig
zVE+~7``D7JxafM-YO(lei@n#tfMX!&B0=Aw`Z&0MGL}>tmgllEZq$&NBt=SIQIWFJ
zHrK@TRqv%PEDlSr`Pn=+C<E}tFve-;;;}7AK4#n_LoV3q>6EURjp%q2*2h4Tre89n
zngOH7Fmk}h%WNt`dqHRC7@7qi7G2E7-QOZ()R6+ODARGsI6gBmwdmBno~^-?PpcEr
zDATLq6gDzyei_IKbkvWU+XfUaQnK$Gqo+X$31?#g2l-;Gn7YES$^4Rfyx}sW+Nrv^
zOg{&tM&xBbbIn~v<vZiUhf3sOXNtZh$SvFHC?X;v%k#y1#$R)JsYzM&*?1s#6e}_V
z)zp$T^zr-TwlLH5u5sHbD-eg*I&i(6{&<xBo&#ssx)<&>;$5g-C#ZIl@}kLps?%o9
z#xIQL6DRXb&9(PfbTTA4X2x0Q=vRtADdv+_aUJ9zVO{yKV~pC8wKUn(ouTD&I7yh`
zxJm487Wf(fq%HPj6+u1>>AEdlORO-UKX@hgtB)@gwc+Vd=M33hz4&l5Pps%TbMM=%
zyOoqMg`oEKi6lCK$Tnkaog&9tCq8}0mNrY$1sT%F9Ok(0#FH`eu=83OjS#)*?PM(a
zwL{fVGV*Jzn~wR~1%ev__OaNDeKOod8qwbSrcX}ENT(0%u5Sw(2bOvDATT$oZNF*0
zskHr$S9e?SQFW9lciX;11yg#v*Yu+T^&G*92FfhON^QCdU!`~+n+dr*`Lq}-e*5EE
zU4cHf$xh!UNiS;rtMNC=hzLFn#DXW@ghF*nRNXJD=ay_RzQY8aZ*p}cGAu3on`#>b
zJ%$8e`#zv%?Ka2}Gi}jifE?5M5=QF1g{kgD(q7I!Sp0r#$Gy~5c88L1ms{^m80Kx*
z+m={$tIu5;e9cX}zLs}0Gc!E}2+6Z_E-Lc|cDv!kuP=d3=@%&9SSyiIUqsDEv6#Oi
z7os8m4Mw6dJ}9x)mNepA<2@*!>_$5mFPoa*_q=&_gRD7;hz<=GrDlik`wgKDr#S82
zrF!rTtxPGLhs{o(ja!+K_q!{P;SOu;4wRu!jgVStM<Zp(?D@Q}%5T=ks(L|YS>B(x
z=c+Dels2qE-u^zkw^^r9w?4|q%MUMFX?;z`4USCB_G~?c7mmgn@<P{N1_ycw6O4`_
z!mBqXrEph=34*cbk3`JtGAJNEKHTUEmuzl}x_56fM@d;+pTlJw#v8~pO#J#Tu`W02
zZCM!B`ICdJo&!x)q2#T}I$CFfc&H}6l6&xinw<DT(W6o-f>1EkcJQ3ur=;^YXgH``
z1Z+C!+gh4wFk-8vAsg()9q>$SMnl{l<G0!^LGsP>o5rg#yeo22oB7gL?eBo#LP?B(
zlLT}UA`Y4+KGurvBiPD0wqYDJAFTNFJ$bH<=r|@inreHZfeHBHHPG#^xVYGFF#w3A
z1ECNL8<x%o1Le1_|CT?X8JseRNqS^0naVx?q<U7p!CK-Uem~>Pt58!Mu|D=9UU6#M
zW9^MUm-F2tzdHbMq|&2NqEhL@FP<Q7gIcwF_h(c!%<p92{SHupAA%{+J&b*uJ|J|M
z%x?U~=V@9*pP~-C^Lm_{UC`Hv_Rh#${0Wg<a<Sd<QATB#I4u7^F(B))*ov)5$%UMV
zaIvg9>Ox5A3*O%set+LkrJ{zlTvlKVTIl=dRW%99J#)2lzZXq(RWIL=222)>3e6Xa
zW=oWNiv~R4)sJmnMg75$mH5w%s1&ifkm{Ug0%4mP_f(+#*!7%xr4w1**!>&#vc!HH
zemT|5NAouqx0&Zov)x*a{SKp(YS2C66t4sFv;q+vdwF3#5QaQJx2)|m;M5nmcIyvd
z0Q2JsGTZe-vVO8M4nr2-B1Zo?Yuh)`FTECzm$=3?j<Dwa0Md*P{)I@h$2*UHf~PVy
zKK_^&>2*1pkl&&V5JaaPL&TbE8NXkX_92{|jRkf;{Y>4@`}^-tL8vb~>N5<5<>VyP
zwDd7FDfG`|O>^&NY2T)64%-QWcs$_IbNl+f>p@V}<bw*?Guy>mx51uL2#?F|W7`Pe
z8kOpE(GnOOgG92+68w=3LJ!RP@j~{Dz}`lbknotRi`zIK_vbW5X@RCa7BvM`*`gCu
zpyPn}kD=%&v$e}32-gjCkTs%D9fd-v4Wp>btyu~G+>dGyvfO@MZ&192xy{j@hK5FP
zgwOr@Z#BP%UMB3wnwn!KT^V5O7Xm0BSP;NsbQn&Bk0BMo*Ru4$;J1yt4Hm!i#w*W7
zunb0NKAkbB$)uSrdDIW@cmTVS{I&}SRlKtX-8nzj+SocG-MfYkqe1gUrfptdB>DhW
z{;tf##vSp4dx{d))rx5S6(<5VAo|9KV?QvTQFAAFlJ^ejPmR^5p9wRM;7Bdd?+l#<
zzQYO;nW;hZyAL_`^ZASk=NIoJ<3haq?MAOc7+cUHk-yus2-EjP&xG7?$$}{)C0as}
zG2!2S&Q<_$GHd5{Wv5Ha>FqVyH(QHgZ-M6n)|YTO@ch+R3rtXpl+XEA7(>Z^o85Uk
zZu5b$#D8~xB8NCE<EobNM8B`&6dTz=W>9>`<d>PcCZXY+p~82{3{f|Dz8;h%n(zQW
zI|L%8nm)`Ns+pT0dK~ZLnImlL8-F4tswg^8DltfKq_PR))AKN*i?<HfqlF}Y1B-ns
z8dT+YY%X~w(Y(Oi{Q#Fb92?Yf6;C{vRx4(jfk2MV2Hc@)+Pm}LZrk$`5dWS?Uq#xP
zH(dVv@ZIHkeY4N%@eju`sdqgR1%j>s%h&L}A<<Em*V(Ig7m0<D=A^X(sV9l|$TtAd
zzQ39yCfR&2D3uN&Y0@1^N%}n1G-CR{`4K@xi-?OqHjUh=pfcfO%Kr5SP$-(+t-Z>C
zG@wv=H34#<V~tPARS!uaqcl{nh62gFFKF0KY}CYr{{g0J=q|TN>XWbGaPfUGZ)<7T
zydHJ|yn+Z7K$q~XN8Ok;e+|cgc`lxk(|*AOgHlAEr*oj~Vfv<1AxKek@T&4hZn&Y`
zMuh)O4OKdWp_ozm+Y-B#p~sY=w{G5kA}!4n8~c8JsN&(RpHCJsBn$}+Cn0Ym^k!J!
zs~p;Px}A*S+i|ZFK>7y^z}VkoDpSpngVsILq!x<^PJs5rAIkyqHYp#r2bHEcc^1@%
zdD+<^3d}j#x!t^M-)F)`U3jn4`U42)gO9~nQ%^W|yT#TOu&#h2JxY0q*wtS|_yJQo
zR!2jJQBlEIqRddnJjb{#l(txlKV#-!qd>7}XO}N70pxN3y8rte)1qHG;w?tTK42i&
zeuN%#(!Af2cG2;jHuI;g{fdmX@1i$GE%Q|`|4A<<rP2DVT?WJrsAh<-0MhTCzn>2V
zx)DX^W#UCJ1{~ww{lC-t&oF-t`33|b5s6On+}EcDi_xHTCH1&TNU-yl!~6BX|MM?I
z1bE1`?OdboE9+@UT0hLW2is;NU;+I`m%o7g-!JV2&})|+1eBd1K#pyQ<NeAbue9{Y
zhxG515?Fxu<zWXqB089hIKSQ3g}BJ2?9iuk%Qx?E0Er7=A^eajkG_#Xo=1y3PTp?A
zgwd~L-2N-R|K#}VB7puXQFlXDynxb1F|!_P&1*CtpIf`X`^RfOe~q?|Ss8#I@moR&
z8}M<W^-IF_YW8o0z%wHOqdwZfMQl?cz^wuG98Ftcd901w^B$nT>hiDZTwMbmdb?%o
z2HhY!x&kqQ&hmPo!L8k%qP$?S%fI8D(E^o^9kCkh<NY4&CXJ7Ib;Hw4N&>)PX>Ol|
z3U*?k-$P&VOS3s@J8lV0Sa7l+_!Fo6KSW&q1SkdKKIpoRZbzipd^8qHx!cV9*@X4~
z@zr0`0@1yi$D?VZ#TIc9VBa4rKl}^B`U*`z)W1~HrxH9vh0s#Y$Nc?@o!}W_#5<gq
zPpC$*5SXxXq){=)14u~T1D&m&%P~UW5Xk>#AOIU3wMxwFa%T<vQ!BPf(FJvyB<{@n
z_!toQV89I~uglW>z<P?~ow0>;DQFKIi-W{P`^P~3O6uo-$`kq^wn*qzLFo9z1nOrm
z{PpyN1ZnCq|G!lZVt+L})OF+2C!!)r8+&B&M@DUw_g!xG{HK-M4}7z)HLQH?4veib
zC(lWR%2~d~nj2wyC*yic;P<_^xj=6V3Y>%{<KxqN%(u|T92Lc_u|kWL5;AqF^SXA`
zP}ba63guG-M4AoHZX7&85QBdpGgc#&oDbSXmj4TWfXZ1K;5K;S7a`Hr#aO5gLyX^J
zjFJF#>-HzwYC1d>at@E?Da#Gmf@kK_moP4sCop}`r87^;x_Sa75rVqG+g0U34#9n@
z)J`K~hLnh-1#U2*;pe~7nz`YtQ`wwx{K-Rf(&ako>Yus129jP&P(k%m#BMJs;JwJa
z1?Z{4m2!^=_`3y?qE`f_c6uLwhh}_qE#`^E(uI5bvm3n5`|*QB>;{U<VH1T>@46GM
z=7t)w`mlbuF`;hS^;?bPLA;%jImt1?0pzJOjZK<2e`<kG86-QmzqfBXJsIh}SajPE
zsI>aB68MCrvLeJ(2<*wx5B)!-<h(8luPrfm0)xC|sML;MvufmIyhlc3IOe}9Sygw`
zYcjhwHF5@1%~yOfUSsl6>txEN{(EBOlErYO%JQ^_c2b-hr^A{$$a!^G1Mi{Nsw=Zb
zb=8o4qr*doT_<;^E-_Uqm~hd{I0w25X1Xb9W$8An>US31$qbQ3iQTfYve{t{32K}@
z8s0Ja;yanHS&TjRzoGwxn)_c;q_G^EUCa}fasHUPh!mAkr+6r%+8g8XgZ>2?Y`_C<
z0n>hVcCQsB>W`+g85XzuX`k*<e1w{x4~?Of2@4aSmRd#qwDbp#@S^MU44z{1!pZs3
z@_IO=tRl%H?AfugB-$U%nykN9Ib7?GXG{Jv(c}XEA|qSn?csxzTt1mG9Iv-w=smSz
zKj+^gXErDepRL6|-@?*{JA?s4cr-q#Xu8_%weM~p@~nP=|EOlGz^$^=b_FNpYT`xI
zv9fg^%kVb4^A=5|3;i)qpKBmqs7FjD`N!;SPmH^R;tzo#!wCQj<8T4(K`cS;@UsUq
z_$zjSQ3fa!MwsP3`IgmP+v!HSc?XAt{1&~9D$h6e6u}H>_nrVNx1t;_2C}TA?r^P2
zu=RRP);g%jG!Pjbi9CCZtZ=)1lS6&bXs@daa;Lv0f#f)n5-M%Ozn{%lnX7c}{;_17
zzf!to_@p#ox~bl3NFkM9Y;D*s+R{td9L0a8kch1Aor*O$iudhSkgsT}qamOB2~VzX
z5xsU$hMn%1*$^YQU6sRWf5!TqDUwt;6{W><P5i{)20>5^{%TMwd$P9>-6!&0rT@o0
zkyC7fK8cObjNv#cWxCnf&OM9{hexrm_Udm&R|O<Htp#JnuC09grQX{L0EV3JH5;l^
zOH!!`K0ongVLKMH-q2)+EMt`u^kvp6mB;mo!imo6rWSfMz9d0~P)e)LFpb2c$*`=P
z4ndr41zec;;yodEzthVthufD@cUj}02Ku4lU&iMh0Ls2<t}N5))D~e~9fI^6_VuEr
zAD0jxwXV`;u}q8qnJ=FMc3ByvfBT^N8NXM`WBub?d3=wfd)e_e^-86>e9~!s9gT^t
zjRsx|e3fnmo<o?WR|5y~q=J#TmBTssjXXl7X6%KtWgTn|#nJM+|H*|298KF!?_z7^
zo5tI80D>!0w1$!7<THO{)cK_Ev~d}vxLRPM9zwL#m-}AKAiMnC`frIqL0T#gP<BG^
z2q9p4bn0yxD=uf;FKULvR8giN8||8Oj6E~UbL;UBDUuNn>z?;7IR@%`1wZw5Z|~{r
z7b;1zvUwKAn$f}6I(#0-9G~DO5k~s0lsU2wq?pmTo!7_iFiGet?^!K!q>ta@w(EQ}
z9A{=nzKz~?H}qy$PG&`JT|vd(VsBO^yWXWvt!xqOc=_RNb83UMzQom%pdl;X{yImp
z=R#i1@xDJ3y5o0<6uG$@@z1Z-SS!tiM^_M2ZJr%p7`py+i@u2a1cW;8U&8rPOYxc_
zpWgjoMM`9+G;oMZGL;>t#X|&0^FkW%tFURZFU;)_Mlzw*JH|bqUW$bvGTNk8_Ba@N
zZszR-sk^%#wQ=MtwxBq<7l#t^?Tuo(yyfny_xMU2T|Zdssy!tlCZ?5Gb$7PXiYwr6
zZr*Kaf%d(V)!QN$v1>krnay2%v>N$Xb2Bu@PH=gNoPwfMf&ZR{9jzu$dE8~vSGs6Z
z@}Chnf)a9{a@^!`gtbk*vUr7ZkZIR1{|aESy)4xSYr0U|;h@c8Z3{KNlYV^&N?kPy
z{eOlaGKyb8PSM)abboRQ^=Qq71%yM!_I+31^$;sG^bO6SkU_bK_vuW^2&0ZHk1_1{
z^uddEN2jI3d(5ouF!nGaTjyhYT^Nq%<E`n9qK|So+;BVinY2WKL))pgAQz!G`D~?w
z$p(**$DyFt_KvI8(xtv<Xee0?EY-)ot;IHv@#p(;5Il|Gx;!$#y{%T68af96=`YtG
zB@KGZy?PrPRh<e?yn7v34FwjwYH?!DW?B*0$NNl6gjCun;)6kNMGB%jR!Sb0brpQl
ztCuzK<d*29Ts=ZlD^>MvrETqaQe82auZ79)c=%pMrdX*2o6SIQq_}t7nY;ig5(D#M
zSG2Y4^x}Eje6i@ErLo|U`C?;nz(DBJ7>9QnCMRC;LgSa~2a<j-;GPS2?wn0CaJs=F
z+12NuWa8rDT`yCSZ&EzIpnmj#b{4*-2vnI)$nQwnFvoG|NSc!C+c8WJ@3N?|N>xaF
z(F0iquDmt*Bu@le9V|6ET=X>+`ZbgD;LYc#vm8nm*1T<Q;7F1`5}*TBr{-Dm;MMdE
z7+L^n@eCn=K|+2vQ~Z3az^$#VvwleK=lS>2!J%X<&)O&{S<Ru2!*^|Nid{e8B`)8(
zpXkX}X_Y$K8d&8n`Icoz!^ZadeTF=X%w|)art&fll@sX|$NmJqP=%9M*HidRnjSvf
z&Jb3WKR!};_AHkPc_CsljS_j)vF34^@-1kJ$MU_=5HeRxq2w!Z+x2e~cHvPg#0&~D
zM^NpIn70`UjmyvFqLb+x-2k5H?-BgEAP@yRv8#z$3~4e$!1u3;3vLq+Rt#yQ;^P5E
z=$w<sGvX0LoqNOAfL5$`%6Xpp_QVz}5ero~^J#_n?&*j@)pLvkFN=3Cj9-=L<k7QT
zxKw|UY4l32-(TjKU(d`HYw&oE8r_N6?=tE8e7vrORBmo`Gi`kOR`kc}1UJe|M0b*I
zO#d=xs;ML!0e2c>=5n~q+hUOB6+f+K{u;a~eEG%1Sj;e}R}Q;$Z+}~giLPT#k#*?x
z=4seWuj9O2pm5WB7?<zVePBJ>AHVs}#W4dgqzgBL2So(51oU{{ZxCN%xfys$p!v=)
zAcXfcO11OYoLaLj2R~>LzZj|h82{-lNAqb=I9~dY#(14onXn~kUy4A2)5EsJbyBIo
zY0(+8Fn7F%t#CNDb-`CqEp}etdo`uJ(j(`!rzlBZ>`GW#Ugqw2vz`^xxE-=0x%m`z
zO>9K)^m>-0paMr&ZwRn!|Lb1p_sz6%pBeYMmkZe=Qt)i<I#0UkMKi=<N<$6;nXYeh
z6%uKa_dEsmm9n<w*Bq9(^C!0XH!nogTm?1pP)SUn!NaDexh-RvMhBcU2k{00?x_F8
zTd`pH)1rjlETWDuOH){1O=;fE`deC6u7TD{8J=2llP-f2hwEbnNdF_1=g&zN`V#t-
zoM&y9xJ&SKE_JX`x?P#)@g)KlBdK26dge3|*ksvvhKj~ZPA{Yk`fI+MkC#*$&MMDE
zmLQg>AM=-wsO$XrnOK(-KX$pG*?_tRu<M29>0n{Sau`rBv*NjysT3uR1v!*E)Ux4y
zgxl{B>D7Bm2;%Q4Cl1i9v%l>N;e)m8OX>g~^Er6~t|H_gRgJO)%!?}m5BCWo0^B;V
zaX2QY;6519l*K}@%>r5v_TC+&%s!~wb%7Zi$?tzk+p~PPjSF=-#2@#BpQ5}9)jxiS
z=Ls!F)&@$xp|DcTUa%RhildiLW&b|a@T!0Jo&@%fKB;-AS6}h?kkv@<K%#Wj*1)F)
zvoFiZ(rNrd^7XS0J3ccQmuxw0Qg8T%pv+h0Rgrwc$Bu`By~l3sDT2<O_A6Z87yd~+
zrBY2>Te9E2DFr!$_#8j)a_{%~nT``lo+D#aA61vfSBKx_fBBwX{zS=aq);g^2zSL?
z3&Zl?1d#VCjiQfQz7GB%ipXB6ti?Y&FSZw-Ku~Q@g=;B4^<K!yTk((l)FZvuQUL@S
z_es~Fr$1zDsApTf)Wn|$07HGe8v|P(#ajJ-sWcX=H!diM0FDe?+DEpU#tM1dSRf!h
zXHhR1%2g`XEGBiKYu!^!CjVMJ-&UJ~dlenjbT-aeJ8N;UZ*qdY6GAdLuLesn`f&+O
zF7dFk%~5OEW|aBjo{)T>5O=^Xgv_CZWVm?3xcaXA(fTYZs7}k|{=ez=<TJoI=s09@
z5p>-}k3W#3I5tA0@%dTYWCXJVZove3CNA&N-{!^bn^tlU>gjUhz|}B4u?)qKdI4wU
zu<5BzNqYA+cs!on6j|c4;A8$ZHLDN)-a1jPH<_Chb9b=<AkWQ@`t~|)_Y>ByZGQc5
z3WxP(2H~!1a+}J0bB~zlLg8&_>%>iTA1~C@wNw7=8UiQ&5!Sw3C`_!Sym?#tU~MRS
zoV%!-%lL2uDm&q_8E5O^njL0bxu3jJxg3~B`FR}SJ$U+~pppc3kTe$V$~Awnxxpmc
z6!=0t6#NhL`0!q(YxFkZ7@%rMNE=t&MsxjmVW=ko<T+>yC(Nr_m-4WyO|{(6yE3QT
zaC*s?VjsXY%hi2KCVdx8HjPGl-Fa56>Sxs}0|oaaBod*m&*C1DZhB0QNYu)N<br2e
zPD*nlzI6T&|CT9Xu^ZqJ8};s~$@FQ|%O73x9D-+$fk<9;*k-&I9oMsKauweEIKSE>
zp;(Juy-+hX)+)0GNiO9GJDYxCYuCgjzqq5jxn?_lqrC8WREr&Cf6Wp8<n>1JB`HVV
z{3nI+QXir#y@WggK_Mdor@S4hWUXEW(YMVzNZ3_<w0Z2rVHZsyr0frcvffY#NGDgX
zrw1KaS+x&7S$Ur!?(o$l?Xbp{F0`$8kC5`^CgZj)yWX3;B2oJ#*c#4jbIa^erw7=Q
zCNDim*DFoK=!GW&k+F_|CLxQ}W+QA!&rT9Q+qyEVF&4#9|5u>}{1#9U?1bZ<nz7#k
zh%pI1<=%6bnu|*;APAV&*KvuXziP`nTE=7jn{SZJ`|S=)&dd2DWI3}zqf-va@faNC
zq9Rx7Rm?orwLD&z5<!&N=(bB>2byej3rqDfiY<0AsF6=X7!+HudhO2Y@JyKJ%#7s)
zbys4lmyL;Bq8dW@U95|oYk-L3)N2#TdA=K_fQ|mCxrnADBg7uM6>+bm@8zwfN(`!P
zf%X7wz}r%b&(x1b3A--mSI-5XPMgyM`*_m~l<YNK_0{>Ej^BOJZ+_{Yh=5uhBCFgt
z=(P@iln1X&28Qh~HA(>GQr5j=ncF)-V+eGiob6K6Rhqp^9Z4LrGES>DkXMgcM0Uen
znTm^E7kCsoWU@nQD?HYYbQ`_pw7)`+Wp%JQPA(&j7mLSsnmTzq7cmn|nXUjwsppOq
zPu%b|z$bS23q0F|fVUI|MH)Rpg!??0P%ybX+lB1_Yd!p;M2x7vRMNQ<8xa}+sSsD#
z(~pyBX6N!`?E0$Gl`O#G#mubC8+wcR-3vW&IktM^McIuM{GPF!lhT|$m9mrdap|TT
zQ?Mv(k_WIh5f<6V@)`X)WsDx{V9V-MRT|lnq@jY5b3Mz$tv!pGs6307My<hg?|b7F
zj`#;a&b%I`b+9rtWduLPUQZ3?y`hU$nu;~^44MD41nK+KaP4qvylU)alXO<S_Bdy&
z`iYnOt6Udo5>?onl*e?_cV%LiYV3oile}Bz;>IO?n$zYP6uz5_3Q}_I-VS~JIhP}|
zZjW>zGpDF`_i;;K>_X9SPX)!M-NmB%3++O6*c>l(;B6K|RaGde!M^U%gO%}$h;Gh?
z%_&%jEwY(*Z*{2P2{7>7^@k4!{9rxRQ;%j^D0qOCIBO8Bzv4whkG#V5hgHkB!0#Ch
zUm6tRjRj)B_kJjHML~dol|W~+#$vjCu}Z3$M}05Bu88QY9d@P{LkY{`XWaxGve4NL
zdfO%ITYT=tZG|ZMmgalI#$ioCjlLDO=k%wj7?-LSX<mZrA?%+liypJ2YaprjI~cYG
zxHuY_d&bUNi*LsSvS2u7NLg9awr;TA9?dT_<uIrirjyhb6TgCMs;5}*1=1>Jb#tpK
z$RD2%n1qoGFyufl2694^7aaUbP;OZ@4sHCZ6`r^GX>?D^6+B4SPD*^s;ODPX?+HJ(
z`V_zWY;x(99{;r1&47OI)h~Rf9v8|?)pnzeDcEG}BGzy1NHizQv~xh*`8q0Jtv)Wt
zAa&;vb{JK5-<4}C!#=zygzDD3e#a3Ug!|%K_YnDlr_1`d)PY3WKMgK4P%|yi7;ssC
z$O*9kD|f8*v92ZS?put7_=xlqy6$4dsMVJQ9&C6~>NR+`6}~I^#>Vz#;-ifmP+j;~
z*wc(+6SKB9stFrG(fd~VvhjWb(u?V{ZO6F?ba@uaFJqOT`Xj$HC{M`ack$ufJ}e$)
zYoICA%y=LFil>#`y%TntS~+CeG~3tc=kKOR6VweDc#FIb{-|2Z(|iq}$TR2hWap6V
zY-kB#{%I7uEm&AwjX5Eq0|<4mIoNL<zeaw3WQVprda{?C>ae%S@vYu%b;mbJ=RLNu
zWBSXEm-`pB(#qBbAFNB{eBtU5%#W;}LQ5!L^>7(~2_E22<0WP9)XOC5kTdVgf9A3|
znUT_P)N{XKX?ExFBhuj3N<^P8FB;Xh49W37fGK<>F!SYy4<OG8AiRm&VK-c7`qKS0
z8HrpkQ$~v8o19Kw$hmG#ED38@J*VryqMi$|hix}+a~b41Nf+_N-^RH-8RG0=?u`#O
zWgbd6fuT-Qrg(9S?+A3c{@7VscweV&JuBL(*ME<z2sK&f`1+a;`7;fn!lEEi5zEQ`
zN0f0(>v4Mf`PW{j*t7wAL;MV4IU)D&PY)A$LQ1(ZeDd?a7dovHkxKbG4fXV!MK4KL
z`Ct=oggDpQ3%HbL%z2^KLwy;=uG3l+FLY$@k{En^i7{Tem0tB@sn^~+tcP=QqCQcJ
zl=6#CmNuO+ufW~o8GRXF*xI4c%Xy!kF`>xuckDX3*^9ZsUZ<5|0>bIQnd^T86a?>8
zARQziiGj`q8sK0FL$wx;Jnykn4+ld6?xR;TqUMV1EY>~fT?EybHL?|i-IL|hgk@6d
zkFw4&@sx_Hg_aW%kr3*}ROG`@xVv0(3{l^iV84H-*LE)b=H8v93+^<@=EKrz^z=J6
zM4O~ZRm)FJkawA;HW-a0;wBmjEAfDq7^_aLYvPa&QM_~1Ro~E8ktRxqAvSHOZmr_e
z3@Y$($>>vyrPi0GGM3K!Z(oW&9sys(nCHuvyl_IbTb5v-9l+L_H~@&!4R0vUbdnt`
z(`xz<ul_cgF0>&3)q>cf+EZz1C8dpSTTj(;L0Gzy|9F-0#}(`M5jw9Fi%6@dwhDy*
zBUoI4*7gm+8>G&L>OVoyf#_%DA*i{_eGc}IQl<nDdJw?Knj5-wezdb@nRL@VM)JUN
zJeK1Jdc#6@e6Ab3FTL~#dS7DatLQqs%!9#w@pFc1gSs;p39II0t>p*LM=TGXG8(k&
zI_(X!)w{EarSs;On(i3+fBW`LTU)~Wc!%s}+z3AMo*0lWCpf&C*gir@>~FYm<!pJW
zRRB4q#g}}t4)nUPm6RjsknZ93DJ3OO2!FMmn3B>P$|=vUEdd<Dic|c~fngMR4>E{=
z=oLcPmCR66Jsu-!??MtC9n(9x`=dAcn?ABy@yWc{Bo1&is9J;lw|(`YL`SkJ!8gkt
zo|S(40VOZ9%%wDIQ61m&$yeC7@*k6n1opv`ax>LOfGU>QOiiZWjlT%LO@aQ2+4pGC
z@2v>c^Q*2&-Dy>nsZXFkTEoqEkKOp?QiQ*MN%8^;iu;>7Wl67>*Au*344~hOOr~=%
za1sV6VP6x{cuI8oD;02Q7lJSI47~G}^zUwPPNH9Ww65?_LTW@KZ%`PjrFm^Q&a53F
zy_t}lnjhVTdV!OP9r?2F>`Y*7HJ>wod|~-A1F=$?GV!^t49Ix~gU{|<j8fg%g<lL{
zv;|0iZ<lNl9m2gw8hr3_h6|^D|A7s?MIasxun8!N^^$s;@v-Z2h<dUfW(38FU%Vvv
z%2q@J(Hb(fnqM^OZaPqJuk|TBG`1Htx9Ph$4kO_a!em|RTw&cjFm-p(|2RZBnap8F
z;XZRAe@f5ky(Ay~O2l?0x}&6<E2hGGo+mFk<z_O0Ljt=>hTm;+5_w)piN7)g+7HA&
zh&0oHWi;>yc`8;`-B+!<!j0SJDGctAg7UHEKYIb_f(+zN)g6i3)KBp7mrGtx#}_fH
z<dU2pPl;>Sy|@?QeJ*6|n{@uQlJ24r>K)tcXSar8(5XIhsyB~hud+V5<1SS@%A}kf
zb@!|Cq}^C16E2C>`fv`dgrwv<xdP>m-Teet6{DT#oBLzOpOt@@Ye3UHg<oIa-=M>|
z*bs2(@vID@X-TGQWQs~O9IMu>82#epe?>5$BG?8W8*1M?9@||)hUH6*ykhQI?=~-z
zO?#-*Q>30lBY_1ylAH}G54ST>Os{#T5quiYrYBnTIe7EK1chC|$2G20>Y*a_>}PY~
zXgv>H^gFNbjUVU~=q!1#6;b^noBS%?%wPa<u(P|7A;2%8ih6?h;ct)~*qTOqo#o;N
zX9xxt`YXAi0)OcnU2!Z4B*jt@uBp)hbaU5)8Z0wwX|J<JbDTt$%w1;R=7Qy{Aj3!S
zK<gX<+Z0p${HruF3w(~+lxAJ`%=#$AROsJut(|R*7`z?cSr5%~Dci)uhnV5?zp~Nk
z=?jw6v21)Cldh9+sQO^)+3v=ORDV}O+C%cTsn;lZ3f7fPhi{Zu`O$HbNVWC1LCw3a
z%}?IHZZ-NgFw}mrXEjp%g8O3u^8DFprN;QQi92miVP@B8vt!v~k|Tl&`_k5CfO6ja
zj70|Zp3jj0=IRePxMRf$fIwLEVwZmlN$2z<|8vS$G5kX+;bmTy+4qm_M4IFfHz;6E
zK8LzPM?IJq-l!9j#13+xf^gtO@J}yt>na=*8oq@EwkA>xHqXV<$e)7&^%E=*%ImVN
zs_e@!KTwD<vot3rBqSW%YB^hwLIz2^x;U0GH1eK>!p&ardq$PhaU7V5<noKt{?g%K
zu=~+EkU(zaOw??3)vM&5%O0!^N72A!WqInIm{Q|Qbn~$0bb(4Nzp7FQ^00M8#hi#z
zRj1lX8X#CdvKf6oE<VF^vpa#hdRQaFqybo#M|s)(Bj7Vv{!|lNej&jLuxjHqmle-I
zsSxJ#+i6%a90r{&b%jkuBiw78#Yg0WMW&3Kj%lXt6BvSKqQ*3OWOi@v33!K&mJt5a
z`StN}xyn=7?ZU5vg20@f+cLD?ZTx@B)K?CG9i4Q?z;jXQ)U(yeQlc-*4p?aB9c~8n
z1kT-ExR?iaM?q<9J7axKIb~<Si8TZBk^3X7FM(c7uG2!dn)U}zww&DD&`NXluE=I0
z_|8FlKYIH}G58|IX)35hn=GP(KA@>T-^fP0;)Q?l2lpv62<p%<<HldlEGSN81xI(c
zP+leli$;jtIa?%rq9<l^uF3H?;XY>ZHI@kLU)D&?Ty){CVoKjbuk{KnhdS9CIOK+U
zkr(omBrBz*>J@wLCork`rFy)4u#@X&S_}8UT%F9|uIRbjgDCZuY~BxWqRi&NQGm`%
zr{zTkhw-kcDw1DgXg6xLe%t5!3Fzrn6&kyml1VaKPc`dENJFbvby&oILgVj90#O29
zzdT$S&gcio4nAz4T|>QBeb*?3stKqm|BlI1k}6@_A9>_kI;aBK0P2TXJsNX46RCoA
zj!GG#82J{~MtMt}t?oUG5kG?2_Us^+rx-YkS$D@OCpNNF&V~mwHl|1@VRH5DMr2Nb
z*o!rFE8Cs<<{O8fFt6)S-1%bb(i!Q|W-H*8QF5wQq-tM4>6I38RhRqosg(qZF;<O=
z<Y0jCjz4;{X+B)9k6N<tZ(TU*CJ3lr4JfDLPdG~%8l1Vq&fc6pw{jI0=Pklq;lG2q
z@b|lDAYU<n05vno^5szo)>BFpD*jOz>b9K>=1yqEqkmxi|4;yYi-`b~SM#&L2iZ~)
z#BW}^N#Q<;My&VWYyRI0kr1HsNlC+dr)S^pinwJ|pGN=YDu03@S^p(L>fZ04osJ(Y
z_huU_1jw-L)soKt^A_cD!0>H&7}3$o2n}I{q7!6z$8Jv<QTo>+eg6pwalkXYNSvsL
z|CDwD7=eI@+I_yl`jRFpDT$fBy;Zmf^*^b}zjC>{o_Pl>#`qMjcnvHl50E#451KzI
znV|j8`v8Xa-v@FBH2b8Ih!qS5z4O^`;aKD-doYJTBk;Q5-wFI5$@m;W0J@Dw@k1iW
z=>2ro<Dnrl36lT$#-Dlr{h$Bwo}g5QA7(U!j~%*7DJO|!{|$)$tumtlL`Yx|aJa-f
zLq>uJTAM6J26|#f3!1clRqFpfPs9wsmOfU#s%z-HL{DS9kK*>coSo;<ma+fZ!CzVb
zdK%R-WF~4Vy&*@F;Q#x^S4CIGD?Pw=xo!1dOYJ}T{d)o72LW^yry3F*Y;llftfbK#
zRM>h1IF5rr!<X-`IDIY&pUTN)sqP4nUZ!^Zt_%X$cN6`>BMmx>!uoXL(Oi#ziwx`+
z0W85XY{=475Fs*jNW0d4?$hY??$Mu@^+_W%q=p6to)4-*lW*r<`&qmu=mrm(ReJL0
zL=l{{F^NLX`$3{74InE_-{Pwv$2n*j9X$Eer=+_)Y)dlzuXw21A=%m4*bbDf-XT-3
zSjn!$28u?r6W^m{@e90wxPHUyN)D0s_o2L^17Px^tDvFpjdbj+K5nM{D_DdY2z93V
zf#9eYR`&8E+29}7w?#Y?Vfl)Ij7i*Ea}OA$B-mGk!DpjI1n&P1rbsY@S$jcD64pxd
z&z)qznu7|@<b7+m=%#;jq!H)fZB|F9%02{?Aqqf1{-kB`_cE-fGF(QrcJ-}($TB-p
z`ahf`su>M%EY+MwOzZRCccqE}mm8%>+gn@*QOg29haf#!r(lpSjqQLnukSa7kI$FB
z8q)t~P#5w(Q@d1uboc=wqLS_1CjBr7SWyWiw|n8h=2edY{XaTkp2yQk|JTu>QRass
zQ?|W+#Y_a&;N!8*K42RxKw0dA(D25bgNRmre1fr__IMF_A7gF035MU9`cFZ?IOwF%
z_Kbi(W=;?eoB~fqVas2}acd-)8)yU6MxwDEPkka1@)gO&3V|y~7_ND_eRhIc{L{b4
z5T&~LiZJ74QHA`~b0JzFCfCux^`=j)Pk@z>1jl+Af%Q}~)F7kvR|cL+8u+;Wwtu;R
zR2Be7r^Ov=VM|Z?E2}qrN_Eq&@b=3m^njz`12}sRmwZVe#{oW{Um7H46cj1s`aSvk
z`+uw%6(CYrK4-2W^m90dVSr)lJdeT$>EEYfI|{_~e)XMJ_N3`CXVCwXB)F{sDrlv7
zaF^c~K27lj2aNK&0)|@LcjEcPGxyxEi75-V_JCttAvPcsS4%0`-~`)o0c2wfvzP*)
zF~K2Su$TpCfAvzFnWAks=XqG^0X1h%R?7bA^<Hayzn_)nE8zh9Hunt@X!67Z0-ON>
zn9Gh3RU_CY@x6gOo9oky;e+P|iNdL7YQV>2BP6pX%C@d28>|l>{>9N>H~?g05Y)6u
z1dxm=0(gW!C=l$1c@y2TBv*ELkUL*Jq{GL@3&g}!pgdV)fRpo|s{Nk}_p7+^5fC?0
z5<>iknj^bRG!2Litzw^3Z?-XajJSa2e<{l-t^M(-rM8y?*+!H5`#GX{CW+d&J<mcQ
zIC-KEPx7^=IR_i<<qD%3#MSG?bpCl+Ujll7j~`$s9KQ(#Yfe6oNQ*eiI5_L`YLOL`
z5eYp5);=v`^}PxB!}_J#12;oI_{677B%R>h(B8e?auU$9+vZint{r(je#=miSnGd=
zV3-C>)0w9tHb|cupz4Yu$yS?DNEQ=g0)lNi^pLaW?8u5=+VY-m`*W&R70X3_<q(%%
zq=dIO*G%9lzF9r~qkq8`A_tnysIxA<14f0Rg{ADm_lsb?c$MhrN5oOKZj(qh_dnmR
zAvEmqvI{wT(JDw932eFg?|YSAK^HdLAO<=<;e(SIe6mfq(KiO;yB-$zop3<S3<%8C
z-CWW?mVC^Dept6<TO;_jIPTMrivJ-3iy*WBte<CP0#->X5E+(m$$GxkP{0dlegd>e
zHAfYsd&gS>n)T3uM(I!M!vJ}vm$GDNH**eFewk|5z;k?4_CPN(v6OT62N?x<@FqW2
zPv#wR!n=I#;dm7Mq7@Z^1=?}lv9waJ!nD6K*dd$cCiA$cnfXIAY5dpv)&c^=Uko%h
z>3XM4AAMVX%kkJB<oq9YHp)`fdSBb&6uf~6+N42jx-mwu9SsfX31tBkQVtNo(fOLU
z;m^Dg4d}5S>I>QG8b%2eer;18MUunb|B$LOtuVm&I-g$3j2s?{tFSpBd~p&Sc{fYR
zWIE#}%ZN5x)5{#}{VytF1xsFfCEa&T1v%IGv_DwqWWD&{*_Dp|=&<*#{19&JGo}J2
z5|@<^?>FOEc6>Q)z7T^2pUot@VX4Aj#vY5v?=f090W8-UB6L*P4<dHe>*xxE1i|ar
zUQoFZ)PPK6UeFW3=uAKb1*jxogRWmte9Vh3ygi)QUs?S5MH<`T(sJ<upK#WYMNGdP
z=W4=~)-czdwy}zc@l&hmBQEcZ*9I__JYKgs2Gt@n=NdWZlThwQ`#Vc3oxRCHUKEdA
z-=(_00&h&YL`gQHSS7?CZ{-}WpX>#B-7&52IZtQMy4_3e2I$aNV%Od8YgG#(L~`c>
zzQgjYGoE6DKXnctaI0HJh-C`!ds-^%<L_+8avDq88Ke*}5uoY5PBjlSeyV^Ch{^KJ
z)Q9rgzWCGl%%*O0KkJWWlLw5v!jp?B8_?0JWf!w`kp=<lb1zgo6W@kT4bddexma+d
zGCVlg^8COp9n*PZz?V6mLuQFXH~Wgw^eN4gt_Pm<IId8w<VQ5!<!9pDJp%1VR-zl@
z)x(_+LkIJfh8QOn1QjY?y3tqwwqPhJ^7_my@#&Q$ebTP1fq@tF9pV6LwKm{i_QimO
zz-YloyDBLJGJeTu>c3nm!3}V%V%f=rLe1pJQe}A`O|!5)1#C1YlH_E*x97l_(B%zc
z61rq{6Y1tbf=e-~Ad|qRla%zxRs`U2<$Vf*GwzXbXsuKs(%!$%Y`Te<yXcpq7(TtK
z@laJkG(*oKoj#++C^97QiU1Zfzy}5(lqSMK1VMrBux*hfJ*QCfqekPlDHXvxaC;5k
zA3Xp+LN`q*Yj^*6%d2^sKTQI5!FVL}?0DpCL&MAU)O};z*+>Ew&2I1}FPbh)C2=5!
zk&{ytuSfl0-A;&^Gt}7lgQbemTLJ2C&mGdZr0}rudFyKUU>AOmzZ?cQxv12uN;7L$
zg$zql&WdTrs6>Ahu3bNkNS<>PQ}3aO(M$GkPT|@a98^u>v#KuE&NoDNuZ2;znWgYo
z4egS#Yt!TRI(N$T0jo``UR6k7TRA95=o;m$N=k|=AFe;-$;*EHslEN(J+v>um`Leo
zncL;|-HBYm)_q6rBb!Z52dmtLdPV#TeQAAWaMXEg@NI*7r@&A$j%TJlv6=b#jL8QF
z7=kVr!h*1i%)7~OmuK;erDcAPA;&k-X&o-Vl>+3%ReB>%Kyksolt?hu3??{x`*tVh
zdgx~$&IJ3tMI*T|Y%J0wzA~x+67(h+3euH>CtHP5T6>Olp+GhKgwM4nG!0{-V#`Yu
zda<8;SHtUb92Y~=K_|qxhR2z_I(~T7rKT~=JN2Q2T1FzYFPAWDI(h}V^o79K^M2Fe
zsYjLD+N%;gsnNjRWE1_27ema+3ocD?=*E->g4J$jW>)+3qY}qAfIMVrsX?=Tx^-es
zLDu=uQngg~b}~6e<Lef3G*3-dJN<y*J&(#a*TP1aNtl$MN(jAj4vZ_*(XXY^{^*oC
z`yQw?Jn7YTjyL6K+P>eky-uv%WNzRZck#YPBscQjq$o$zF8?{t5=RLtxeS{1^&S_G
zq$)a$!|Sx;eiEwph*?HIaSNvpH}A={v!sPJ;>p7{2ZMpCEwj~1)A#tI-a99=gL1VR
zqQI&R8Z?KWtCxU?$S@Lca2`&;+B1C`f?WAw=glo#j$Z6GA0H1a!?o@v!EfY0;R+~u
zl*b8Unnu`JIRV;Q^FxF5;)TD+YgF~Ay@q#m`M}~e3yl&jyCFKtXJ@TFW#0e_b8)w7
z(8bJXL~_lyi7n6K&@ZKvm;To%asu~YS)-#jtS(55ZrXAzpVese<%b8J1)m=AFmv`y
zRJy=Pj`)D4cEi)NritQs=edC#8Ev!KR(jtww^U)T_)+W>w!RImI;)73@aI8l&t87$
zomOcl$79xfIXt1;$aiSM<zVo!qLH3{I)3S%^+a>eho`$<2Ole1L!r^v_dO7W?<erw
z_bs*IMwM!}&9|R)4MYrp{krU)LK2&)0NIcYJ_Mq}|H;H2@GWD-kUo93=3;X6H`f5d
zL#i-v;R2h<o}un$Qiq2e0{&w6V_ET`n~D$g+49oq>bBk+59Y?xae<^-j;YZb%VNFB
zycrXs_If~_wfEgg(JM$v79XPnT<nZzpuWMcoB~;E$<li=ZwGLwB1zmZ@x6k)FAoH|
z<9mNBns;X_yq>D_gqo0^KeBprcW$aiN~Nj(<MT%G*T!#xdX-P6y)X;5d@{5t#jYQX
z%)Oy;$GHdzfw-TS;9GR7?l}^V?xGU&g7`N^kkjv)jXOvWP&}A+@q<MgVIL%<R8q#d
zi*rxUCgG;~0IjINE-iJiQOktiV^fe#|AU|RiA^c1hg&;&yX+nA;=GRv($aL^CIAlW
zPB&<{ycn;_aQa?pyl`jdd?&29j|SGFg^{oIT~gaaTm(S|21u!X{GEWK>B0tZQ7a?3
znrem~<k?>OXj%bkJt8-iN&Zh<ht5H0Sma5%5$7z7b86iS1>9m{nYec<o*j5Q$d}<-
z!?$k@3yIv0%xXp2(&W3T!)zh=ksAjyRBU}p*+_pO%gJE+1R>oT@22rVxB<uNU<t=P
zzL(*n^&YELl$OWyV$U@^5p0-;-^)OKEl)+%7ZRJM0TKhf?5R3WB-y4R&ud;h*ZC)b
zl<;J6+|?l^>EIC@viT~0R>8G1WejV#i@{1-Mu6di>*18%{dI6Cxp<%WdCSETC7jTl
z^Z6vLSDo<JWN{ekuxU)^&boS@>Sv-yn_NcT9If*TP2$NTY{N;ts1x4PHM<yrbmI<V
z%dZ^Y3q*WT+I$V5LZv3=rrpAovlsn9JAY~T<~aRZbwWadF0d~fW^{qK>Bw`Akh4Uz
zh^q`wm|<od?j)SXP7m}_rA3-~SV+8MI4)*%Y1M({xqzpf@di00%$eo|Vvh^o_;#L^
zDFuj`7UU5))J$}w#<%2P_$1AL5=_tv#bnwIyXML^(Iv|3Iaa|XHhHl>Utu=ayn4ZI
zByoPaoUqbcL0Q0>tWn+Qy;P#4zEJ15kXg`bekQcB8DpAU^X_b^=EY1#_}Ylw)64Dn
zsY?6aHM{Zb4AZw$`yI76$b?uoE8%BS*AMbI=S0Tl8RWl(##vdmz+|3N+BKeMY;RGD
zU87tzsZoHjdOEl7$-h1mJd!5vOniOzLa6U`CBdj1vCc}GkVon}y;)HV(}(ccjMl+2
zC@=Am+C|%3Z}d>6P#O-RG)uiyU%_VLowu3lJ>4fj9`7RLpv6;Qz!%!=E?lG1OT8%&
zh!qFKn5ZZK`k;+W@06Db4Gq$Px|l<!?J^rB%bzu*fS}+Eh~blNxNz9+x=EVeW_h4+
zUfesoI)blD^&d*g$6*W?m@y5hP6@*A55-68or8yMaAxB_sas4H)~SGC6COIB#a=|5
z7AV(clDyVBHmeaT4N}SL>tF0W<h_1;v><<P>!$Z~){Tb9H*cb81j#<wxu@b_HXp8Y
zg}6IyiElNY6lNe;S+qUt@+)U3PrKS4J%h}CYxRo2_QyPo!R&l<XRq|jTw931D+@7+
zf%604G@(mUcC{fJ{p3fsROcrX#WS-Y3gno5LWEp&NQnIl<R(QgZeOz(W`6nOEh??|
zE$O;@<>cx0&M(IA7V1rSRQbbBJRl&roGPRJTSWXd>Gswp+P9Q*&0&U0_kcvS3|Oyv
zR5aoHIe-=2;nd;|%UNpx6j%OO>~I1N=TD7<VA}ws+)gGcjC0_QJQXxC%_vLr;x<a*
zI6utO*;(s(yvfw##r=k8TuYNpX|lTP411xcfc{mPTG8{`qlbl>m6F^{xmTR$b$GL3
z%MD&u$%6wos}Y-M@9_gS$(<7N1ZC^Dt&$STwFZFWRPUVIt=<q=Imd_Qaj)?V01@1G
zy4hz-OO$zvuCBG}ed+;h7mnDkuJ)}WOQW<MHQT;C?YU+QKQoG7U)(yuc^wGDy9z`Z
zQoXf0H@p%BcZgTV>zbOFDcDrv58qq|s56InjHSwTe#0H9O_)D?I}_NGS@Y`<skMCz
zaV&|Gd=4l-!(cR!7TxyKg!M-UTi~$>+P6^6cCSMS8Xt8Lk`JiV`br~q?09_Z*uUHX
zgF<~=@o*drF48XO(t({<eyzE^ie*Kf5-6&lvMyI12xA_WT=wr)4EUv$oGd)4&y>|V
zJHwV0e;&k^;Dl*uLetvVF8TkX>ph_1?Aq^PglGvON<@u_UJ`<6GeYzrh+Y#x^xn%L
zL=q95=tPa)JEM0ddhbT>ooU~lJkRr%@Av-Ky4NhrEMv~O&wb9m_O-8lefF$wQpPFn
z^_$W(3H31!0cudJrqTgRK@^L!fw?e$C>5W(l@;lEb9h&?NR*M^5r?#+u`A$`ti`Da
zbh{m2X{F)-%8BM5QF7_M<^nq&SAMOSWua7z?Yl(3^1QYl{!ykE1$!k4=>5n&Dy`k#
zMjGQiIxeWJzZqIvGGTg*D99!QO3(@J;Lt$Nf_nnXUXmRu;fy#L54`+>9709H<9AxJ
z{uRziG65`iDAL7UTz!pN3c8S(R&=a8wZRpNwz}IwK&XG|tgjj@D>b+N!G1dTs#n9J
z<9SUN&Q8V$x#YyQ&(XP`e9vcz<igW_eMuH6)hys<HWaOTm|BvVKDYQRKbH>%6c>q~
zx=-EZ>#+eP`Oo*(a*gF_$?R~dY}BL)i%)Q4<hUv*KZo(WEnru=Iu~-~A3a)b@7@&}
zwcpMg1k8j^-rIg-1pn*|W6o7g8!snb>#pO!f9pm03x63ul4|a&4=Qsoa-OHa_5kP+
zfy%q9{Wtrq(*E-&r5C8jR2OE<9ups9|BCJY+{*~y4D&_HTccczulwL6h*@n%{r+`j
zERwm9Tp+R$d_PeF>DiCx)tA9|Up{ZoK<ifyJYM8?UPOP0(>33s2NI<&WpB#>39wNG
zzuad!k#Y@Pe5#c=Uz_=Ajf$=Vlq4x97YVeA^2_#%(@_ilpm+Lb(@ijHW97Kc-w}(Q
zq3Be;;&oV<)E0ifli9h_!FHkz$=ghS!w1tu8$&5InlV3-gZecZfZP0F4Gjt1v#c}y
zrh3cKhmhu58Ag!J3N1D2=@M)E{@c;TqCc3Ho?`1WgWX%SP@JVz(ENhqKbnUC$p-)h
z^J<@Bon{U<tC(-_VpImbWW@VB=vso{PK_MAO;%v<NfM?u8>MI29I#pC?HnD{X9edz
z3Z=f&CL1I$SSea#znB3S{)~tmDX1u9#|>^<n|%W2@H;^=)Is6$2F%R$*A~zX!4lTo
z{<_(KgHu;ri@@PA)#2F1SNQV=?q3z(VH1}=nr)!m>9)ll#kBcLhr_Hh(1mfX*6rq=
z$4;lSK-4UlZ(LH{7d0uPr1cA*5%MO^g6rrYuycf30&!Y%Yu0@?r@}%x&<Rz)ZmKg8
z77n#ai2Zq*BIdq1%E#JUT|-X3c{W^)nzx^-m7&e_D7&T2ua_eeG>;en(Aw2a+>)9i
zhxTX*@2@gE8cs!r2PJ{!2HIPNMz&@7>_f^^T*$eTS(?dW!(8!|WUT&DvmKTGLhju7
zPhrfN?e@_f8C^bTe<ID!jDEWBtMfbuKLCQFukKivU@%cgeY}zjLvjp*hE`T`;bW}A
zfTvCjH^u?L;gi}SwSW1zNiqRxA;wScx1C8qI4dl4pj-Ioh*o#&GG;P|SfH4PhL?m>
ztMuStV!U~89Ak}k(rdIyDO8uI`+C`TBF!!;+GljPqFl8Ll1qpz3HJ#kT5})taz3e+
zcxrtP)k~|gdi}%4vj~s42#5KBoIgZpozFmKd*g>j0~P3v!;ThnN(0$OWZB8ukW4U<
zxlZFYew{PhXVA6H{cHl_8NlyL4Yq%ws5ArA113H_qpk-$ZM5uPm5AQnoitpea1<tg
z#S99)IOxDId24xE-Gh6lot#Y|xzC-Ke5fw2wtHLN{tmZuI5d<`cEwEC&l;h`f_ATs
zb>8L@4!hp|9T?^2Isk-&*A}MhR|g+hNX#ZyX7fRN>hSe)yAF>uP!LebP30+8X`h)b
zK)dXCX^Xv}bLqHlKUQpOEq&wxTR=UvJa)1DZa#S5+A$R|wgnzZ5ZKs^Y`u|JKQuq|
zVWr6bq652$`v>Hioj9QG_)*z6c|Wx0n9}<FT<+aUZg;x4om&Nog83uH5Wa|2^iyjR
z387QS4Ba!Q<1x!?r#4}8&wcq@kZm(P<RV?g303W)`vT$x-TOu*618R@UC*Vm`z4d8
z`K%V$RyR<7K)ivGjkEjA>L(FKl>=2!2SRfzxA-$w|7}ujvL0@1rAXtm8;-`2EtEhM
z67id*U26!b!@rHa^-zH4B>|HA2x#O$5X2d96m33|@sI`w?Cfz}hE7c1;^0M)(2aNa
zGCWjiJjR^WKu0vf&jju|@{9RNNiqA1T(LC_PSXd1zTTz1>FK1e|9N}I0r~R{%0^F9
zZ$d`Yfp8NdGzKFeMyrAs7I#9g-G@2#nysScx@=!VStZ753f?S+cHE@X?{zY<Mj1)W
z^j;=|sjXK)76uJ&(dOu=<%<S)*g3@`5`7e^q%$VwofRK#4unIa{+^)a`^6PKV@Ahl
zxJ2ls&F2#8?buF`z=}^3f*u~a$iFx$ZtQ*bNR-tlt-_;E>I@*pA{kM&4>1S2b%bO9
zLkF6DjNWL(qo8(A5Oudw%2tfc+LIbfK_QG2&p`SXqzvy+&bqQ2pb%j?Kz{mL1fQWN
zm~S={rtIPIOoIPCRv6_9f5~0|KK^KWfN!S=cGxGLEm#VaeM5M^;udzQFm~c%?1Mv9
zvbJ*aw#mYI$%0^*&wjjg_<l-u3+$*Q*GDddlrcL5|1?7}mb2#!a2JXMq1U*0;{r%m
zc>o?p(foS*RsjIGtWZ#q^N005kBs#CoLL*%u1+gHM}5}nRlc)6m+cMSekcU?_z(=W
znCgrs-|+mdVC?r!Xa*3V4A-9}eG+D__1j=T<pH!MJb_rdqw%T;t4g=^Hsm`f^2d*4
z;m40@ZxI5ncP;8`q*{(?XTguFnShQNWCS$8!6m(}WtxG03G|J~>E>f-L;IvX7A8An
z!EW;Z$r!>DmsV`T6SukZ?vdZ#44@RhwkhD%d4bZ^8N2XKkJg-?TtrJ*3>n2-!N6Cg
zhgHel`QJbv4EK{o_Y(O`QmG+yc>A;uY6%Zv{^Vz8TySyTlj=v%xtK}S4q`_)NR>mc
zo^2MoNvvNOQ++U_S}<WZd@SO0c<5O1RR6l71!`mVk@k*n@MXay)8S4)R`<iUHjYPw
zmy6WZjoSE|<$0>C?LI}+yrcN=;%;5Xv~D+T6KW1)-y>{akRQLVpkuynjSty=!NXLx
zYEf-1v^ga=xqTBTu&eP3RE7tFaNp1$g-9&ly+WDmPDp*a>YGXJACqn0)Hcpq4!>Pp
zxbJdJI4q>@1A!i$hm;SV9za;p_wcANixk$s@KLQL$d5P;p&G>=m#<dpDj@0DHyGpJ
ziZlqUgsWXb<5P1sQvC+<df?&~I#o7~)IfB|qzshbO=B)ga>yX3&y9;=Fq>o<6ri-1
zxAs}M_W-RE^m!HhBk}F1EaT-tGU<YE3a`=EXyj!WasPpgCZx;#?$2+}#DG5W;tBu?
zd35USM0ff}$F#r#<nI3EOENXu!3UkPtidC=Vj`QvwfT%_w4!=4t*%@5$dQc@ExfT~
z3ANH{^$fB<XeFJO6K4d;+s}X8dtLj=&uX^*=X5EcM$Z7|+$>c9$2+L>W2H-t^WL?x
z$j9xeaszmQ<Fi4nSg!&<D%T&GM+INsQf_K{gYl~;&P&KHuS_Y#&(4(qYqicdsgJ&X
zeA+L@CDdw1%(zT<iI^Rhy*J6F%N)n2wVZ3bVGZVs<%)~NFcdCUwhySM$@ow?+Y@Vp
zkSF~C5cA884nKd#fIDeG&S%pn&^Oer`qxIdQ&VP;Pxbzy_ju*7I*&cgAQ-yvWAXjV
z*lyDXfl|^dHK`s5Jli5j(mQpENBQ$gCp*M%^cj*{{$=BygM9UU<KxAHr|0vkLE~qN
z71sb{%)GOr>>t?HnWSbi>~jtq-E5$#$P%Nk#&wr7Ir<*Fd%w+WutQ<=lkcAX`Cy<Y
z4AG#z2)fX}En>4N`Ihn8Mi=C@Mx0n%!dV`B4Z$E40bZ_tIp<~ty~v8leeu#Xt0;Iu
zMPb)hFNM_ZCkJW4$Awsej~i-f9u7TxuLJ#7jcww|t5W+6Ybt@Pc-97x(94eIe9cwR
z$QhMeZGLJ!lq7?^3G62mve*WZ7~90t@n;;nC+1Ob2|!Ap>gU8>Ro{@H3;jcV|IsHr
zlWQ5aM-H6@>H+b-E;EyHS~SF%O9{>G8qx*)23tz<ibA+tWrA-xo$N~D!6EF*InrDg
zCdEYj1E@<&fAfJc0@UP50zp;0yv8ljFUa>(<n44``;;LpCt*2ZzC2R(R29f5x0rtO
z^8BtV!Zx*WSzzx|Y%6)oD-VDCr*3CH13zH=fZ=|2E#lxBn|fVmIg!PC6mkRc#6p;<
zGgP+${{6h2vC?(}q+7Bl;ZDOq=eyfMQ)g**Fz^Mu&I9PMNqhV*`#&^^MxZxBPCVL|
zX_kvW$utsb_~oGR!ySdreA6>2CY}x1&)y5h!JFe^r476D$&-6PjQn!EK8DZN74Bm>
z%~zb9d(mDpGu3cBwNFsfk;F*NYY=02^q>%rXpK5az7tTOVSp$?u(hgs-^9q5o)S>g
zlb6K`UihwUKKViro*tF`MPuKf?loo7WO<t{31g6$RR{486p!W9TbE~rgCx5YrL**K
zuMpH$g&@dYD27j43qkYPY9ZZ-ey02l$0OnkZEh1<ywnT(iX4R+sfLT)+z%n&2&nlZ
zouY$aB8Nd@G!{BX&fY+scpnTXrte5{-RQ{idCsX08093c4md^|`}Dh(WlKQlf&s7}
zZN>-#pFES0@X5PGE#I22b(H?5TkG&(uqv_$RL8e~>w%tT%M!%IUV2P_B$aTRya*^N
zizNL!QP`>i2m}83moO2?D~Xx4rtKH9F|mGA0D{eTfV)%TDl~QLH9`@tB&G?ARkS&4
ze0yeIL>RikUFn<$m;jz_O-K4AoRqw*cy)!;xbf3wYlVNWj~XO?0fWZt2`b3<lu|W0
z9Pu{^ohuPhV@S4dAb<zCTLe=BrW#4E3q93**XSMZo%eixTkEQafDWfIV0^eMqwZS=
zB-0Hku=-i~iA+o|N^qxU1Wc^e*PgxT<65ghb)7F^To<meIPe4ajfi1`x!Domi~K1|
z<V<r7wmCvcTiHlrn-qGx2I;bt-YaTA`Hs46?UZq2WH5=)<#JOYR=MGvE(*{AC%zJ4
zTM+$qpIK^<crwduqq_d3l<rnRK|lSco8r}CZJx_2*`7PR<y_U<X!ixa*Nq-(6~$M`
z=t(1pcE9sPU}C49k~X&<8ws<`;GV+s_{sNmD&86^SBKqdJ#jp=W$7?Rdprx>qxIX6
zMoJGOGuI8yT7KRdGk0a=za1VONCB#IfBBP<B$E<u79N-wk%OriqrML$y#a$W?G9fm
zzImS4yt>Us*Sy+U`;;qb*=9Z`Ij&nBy#keXt0cO6O3T$2Yf9nj&#8-ibsfeUrqxXK
zj8cwj><dW31IR9MZLUVZuNThE?_qqVtTq&wfjs3?eH$ST?PnNXBWYi6R~O43p4kg|
z$28i#DOX7%<TVMeaUExuPM+D3{*9E`nwTv)%w%+z-&0+&+_wklhk6peK?ryYco9h4
zzoYBsTH}3Q6gtr#({oVkJ}c<)l3|uU!=R3T>}Pw^Q%LLEt-u_881`6^5`MmZ11tQF
zT_%vlc;I_bQ6+}IfNN62`;YeHPyRYU23V6O^LztB@`MDFxl(h~#+MN18xu^1PvY=!
zGL%{<9icx;rQB*fdhwlx#&yTE*|opayEb<pek`Q~H;{?AgH-fJAVP0$hWQ++wbxhk
zPctX*e3PIVOTtg3yW#5{f`#!ZwWe>vYcAifIjyqaGEJL3&S)@LoCgcX6<9zniu9hc
zm>ELdP(s@iKPll9T7yF$@j|t%HIsykEX<nH0O|EsbD;1Ypel*g;KU`S*d1+D8aP83
z4|WY2$Mgy2+;A(<=25kPbq)ibz2&OtnuiY_M5?4UrH2K4+Q9`R1d@)QkAABFcESXP
zf0#Erb(q!*G_sPENw{8f@6A<(TsmmDvKPSb?HWS>)&HlWvHwcP;{dV}bW6SQciRZ^
zp2Jo?DuMv8`~g2OJ1m`YN|7T})Kr_PRG_MP8l6JS*#%~xzL2Hk7iEEP0(|T4ZgB{n
zA4$#KgI7_9p$Ce2+PV2;Y|3cFG$pL;ZmyxFU4>mQc^lq`XS);+Y6;kDG(#>CP^qp9
zF!4F}b%B$TV;V;u|2e#5k?D{kqx$0MIIbUqX>a-2+ls;}jdt7l=J3mn=Xei=rI(lA
zyw22<QB+XK!JHlNqGB<pJ7Dz=%zDGLIl7pUb|l2@K;lJoE{e6mC(mBn^Zhm=?Pcs1
zVEQ`z!)#~FL2D9-@HMJ8-xv72BWrqvdVY=W!3oztu7Ec;bow?!cKdx7<`SNI_PL*4
zAcHGyy+5kDftafOP%|C)y(AEuPvh7Xb0CW$x{>m<)!|EqH@E&QGcq#Vk)Sgsl}qAJ
z`oRi;{`vq}=}s)`f`g;`FoIS2Qijo{7ih@GL>7ZPRXzoPN9<btxebsKp;~g$)+(>z
za@H-(cHt+@RCtwESYO-5#S^dPt$-NqTGt+%sZvii9)ohBiowB|Dr;wI%7dvsZqm;!
ztS#V&oTc(GOADT(Gut{+l^tP%*U<gOb@Dcg=#N#FDAs7&f=6xIAVd$q@s^9dR8-8L
zdas{8{$>XyUN`7_P|%h-e~|67V)5-Z_9YLtSq99WY^=yIR><zj=PFv^gCBQuhw{xk
zEKwyCY^rIQ5{Rz=w~;A(W-YZexv49s<6!~4Ea2P!(WD)&RDh1De7*C9ckM+`utvsr
zpAP(Or^S7ze$%ITXg|;6&vI)aiv%Lx`;Fps!a_o;kF_%0k}o7-cE_d#bXUKV1^-;l
zzlC8J7gz#V*l()%&L@K5Aj}hz8-700{*n&lCKnHeZdR+DzW`E8KwA7Xr-T*5!u^5k
z*S70y^8FmWSf{5tbZxYEmM_Vqaz$S}A}eKT32dCfTCLI6S=y)+h`W@vt&_kE)FF^W
zp_atYp$V*q6%O@IgXZ-$t{VMT=38c5!aW~6HFF}*nmM^5z$IT7fYkCLM(;=Y&jJU0
z3U>}8CD?aC)llc4`AeqScr4<>REp(!G5y)5;E)gv00Y`+OA`LcTmC^v{*o<#T%FOO
z=GI*jMVz)kW4)Qre21-!G4A(5aj-2v7aDw;YJaol)nVkT-9KfFO3F;l0=|NkvQpfq
z>|&~kLxbjn7B$_FMSiE6+Qqu|`L^&{-v_UgOuXI=kGPWY@?_vD=zT_P|C&2md1YbN
zdC)^{4E5}?Z_cYmo}6EWJQP>_M4DnQEV|1Of;yN5kxGH!a05<FZXiCT%8t$*ij>#^
zLKpN^(s}!V3u1^qAy!PAq`txPM&ilFTTPj7DS+T$yqCfBI5RcELw|xpJ16UO74{h~
zbmG1J(V_W^;i*7CVe@r@n)zS1@UI{Kz2oWtH8tqR4~~h!o;H1do&pM79gnpe($j+k
z`s5WJ_Nr?am9_T7R-fONbu^!!1>h+?I`m#y%!UL7myx@kt<Hp21V5EJyz{GJQjOR1
zvJcFqxTbv-SFllS5b~isNZVE<ywgv$wqd1Nb8{HZu*{oFBv`1eaoJgOb_>?S$J2p|
zPo;Sg5}ur;^PpIa2#gV09B5KO)sGJ%51bM1iyjjL_!ure@we+A(iU+6r2LW23KL)o
zloMBY$#5E1XJvAv)BU4?Z3)2@KWeBYjsK7V$gjjD0bMeF`VS?5=#oq%D;miPXLxWV
zz2r{Gxc$N3nEY46?PbF9&G)Kn7LFsM`!A5UDR8|Z4G$O6!r`#q3d}|GI7>_cLjg`?
z#v;o~Eoks;elOo5;8}%~t&i@I#oqOvxo$4Z;MYiKTyZwf?D>03@5ubveg$K+L5SgA
z;q3e4qmr3>s7T2AF5cc_o8oSKQ#O|Gg%_G)2LF;0;zauoSAZXD2kHW+zsDCK^`sgp
z1Bil%um=Ov=8r6NuIfhcwgGT<NsSFC`&X)8_`%~kpju1Q)2~bPQWFJ#Zj3PLS0CJA
z-ZF}xNq2G}TM~XV#Rvs$@?g%c;v9=NdIrPIX!W;N*<imSO>=E1X9vQeUMZKWFMCqw
z*Nj4B5E!Zw#~ec6!>t<gGgf6(CwJJt6xxzYKtFU@vRYwM0em3$3g~Gb{Q|VAbazY`
z8|cQy)qG=qV-`kic$D11U4y|9`8A8>$f9%GIz}~HU&p4{h|S?QK(}0G61MrRk$Y^a
zDXu?3oCq*>W*Q|Xt;*f}o_<!kOJ}=%{b6gKuIFm!yv2Ku5G>ih>}vsD!UU(<5G#=M
z_aplq*tNUeaR@H410IuWA#7@4_rbR=QA^)Uh?tf{v0Y1Zo8q*XA;IYLeD5l55c<9{
zg%(t#yj%W>+Lhpj!V3b&;rR#qBb0-rwG9YaNPx8D0pH@dZiNg6s`At+cc%a(Ax1PQ
zJY?&4^q>}>gg#net=)z;8`0wa6u!G+5pg5*Sd%M0`*%&jnHA+G;MpQ<J(Df11jNao
z<AgU#IxBrQ{1O<1YK3FjwRE`plZ0!iNOR;pz9TTUM695{%>^Zw?k*_<h9bb&99{sc
z^Pv>bT$hjRSK+y`EYj%S9cf^swb5AmD?50O-o6z*u(m8Q89DD&5&bwZcNGD~g=@Ds
z{KM^jgU>+aMRH#p$D0}jMMa_%ao2BOsCixvPyeu;2;<qkbWl$UQ(F?oO}-MzWA`)i
zc-S&BlQ!#=keMdo-uxh1l>m-(WIZ%bAOcEwGv2Ov&x{{BK=++SZ8!FBm@BKMiKuc-
z+6nfHIh|(388z8)!F!{Xl$^8XC6s@f1o)VH2BemH<~XMF-8%U@y9YF&zT+{q><wt%
zTL=L%DaAjClWEBkXXK!{axR1rnyZ@$A#)JMrlOBEYp(FE06SU)DO(_%<s_=9cn$OK
zr94YmzRAYP9|e>gKHo1z&sRG$&ezAvlb&zOcfKVauCf+>zu21Tv@K&wtU>ccs3IA?
z%E-?r&YYa;R|x)HV8#<ZAH%Iz|1FN&SoJq4{rz4v)Tu&I9+v{U83sf(md-<}U^|2|
zHE+em`K=53p65`7bmQ*d>4Uv%+iUKM%l)R1fsy1q^|F`02$3sG%vT@PzFfNh<>?M&
zs{+GXFct2sSVEc*It-2Xx#2_^Rr!JSYbkjabq=L3>hB6{)Lv(;l*DHJP$tNZdKW6C
z{_b=6i@0}u4M*U`ll8f}gF)Zf0q@np#)FcDhK*E|&pz$Ry1U=TMLUOH*QS!SYV6K~
zfJ3tMK(M&MG#4SR%u@OBVD!F|=w1d}-p9h1I4lmGgp#_9c`Y;dzMx*|QQ(WZF`Zq5
z$20*iCO&FDqp>5&atm0_Tu?{PKsY$QP%+X7h?O|h3QH-t)a4O4osn<ul8tF&O8X!e
zD7-k*qXh}EpHFyu@hh5t4FVb~<{Q@wT!7_7ga#&c!V`fOWf`>?iZ+_JbKkk{>^&gr
zW71emc<3986YtnPN0wbri+uU=J|<{0`mt3sKGjt#vQV=3V48}|$M#m1sC;s%HqNCm
zr%LCOausQ>qFx+wChXsryaaCgRQJI%BEkqPLR=|-BD5rX83|!fl0W|(G?~QSiphY3
zMH0vIo1_x_EsHuO;-(P#Q|)#>ZVfXlWY%7+YQ?uR|Lpkmuz93u$2NyOedlA?EH@ma
zf@h#n9P>=~-r~gi&7wwOUTr^3B21B$4KA%OW;IwHz@|Sno1(Ip<eGpkkV;8$llV_b
zwZ7QkTyJE0j;w!Au2!Hu>Z{IwT~_$@{&hquuFx;mmtJy1$r2XIL|`@^J1ce?0al-J
zAfs5InL0sEGX^!J9?N+Uj4SW|`x*hPKkYJsF-1tG<6iQ~rf2Dd<F?6TQ6aA(wgFUK
z0aJp&TRns=X?|Am0P6KHY_p_%HKsG3Nsi+uN@%e0jPMwYc$=X4x<rN27G-aihf5Jd
zro$=mD>JjjSY++oZ4eKQSr#JbL+Uz1qtZ0VhGzteN8uj&$i0j5?scsmPLn{~k;efq
zX#}>@?jj3)8hnb9oEZDjj_u|Nm)p0RXg-R1WGL%VB4;Cq+f%gl47ny{aDb_1i8v=*
za=&kQl7-qBpp24aCuI447F4c%4!lc75#aKJ{_rH;1Nf>5xXd3UAM<)Kn|2lJPGR0p
zo$iqAje@+-3+8e>&UZa^8~mbU@gF^yTShvseMjaZw=9NY)c6Y|YeUeWl^l~G=&OgJ
zu}~x+^GMLQoJ-G9ndJ#xcCwi6bDdFV4CaZ2A6*esXG@~Ry^wrVMU^w%obDi7eZg3I
z$ZpxS@6Nmk|8{SP$Jgl}F`{=5vw<h!Q8@Rxkyw=B8_zUXhewzcP^N12lxq9y{8w6T
z2UkicpVVPKq)iNmrosO9nBMV5e-Zw(vja#gpx(8jeT3;$a&(!DR!pVIRroUzODb5<
zev>+TOzOVT2lwAov-HxR$j^dz;86g@qjZUrygOUTY%*oekIChq;6!ew;$ypp8N32r
z08#-fyC$Oy_8_0j-uRgx3qRjh7nIlI!3|OEHw*NdW;D4U`E99p7W%M&P&S`un`&q)
zrI+Rr`*v*6MoB*l3^jNs^j+6;fi`NTHU<KC{ALhWwQr+>kyHbs5mk8V7PIGu$8let
z2c#K31WH*W?6x;>9G{XEdI45G^_$^_-s8RTw#Q+jdpKnc$473=a@sG0?PYr*3LIB7
z9>SC$NtJ}(YxKu=tXuCkoxxwgSO%hmEYVn$$X?u@brv!roEbmky@^~6Uav||8+K7M
zMquz9QdnOUi~~(FA~(Ik)i2ccs#gK$vF4luEjQ`W`fCOvS9_v~`{SIoRWzct@dFs7
zzAj?5WTg1)wieA?@SKOQp>DZPSnLL$T9yO$PxY<)hD9zmPX?=MiLQwY%oMj-v1G)B
z+K?!5yo$cc{5kv^Hcr#Ah_fB7=plN&_u|hqPfWs9%AL+s`2n*Y^-@VARK<BpeGR5@
z^~JZcZPJg{R(ZsD3NK%ZX0JhQIx+IMPuA%a{yd96!X<A}AhzA9iq10+rky)k$P{s2
zABIG+BA5eM$bZElUhPW{jJtNblQRVnuq`QlCg^M~Ip(T+T07&ZSXVXuv(VcebrnDP
z^PPN}krRBNctTCxsiWcWr<0-QG-C33jnH(tVYbhzcGO52FBe4PyNCws)>uZiOKnx1
zmSUohfxYeg%S1zCIU!`Za^2by13p%<Kj4VP1@G^w-wTj>@KQkxcSJ9JE?K04P^D7x
zc-^rxR(^X_F%9(L_{qWb3<(}ch(v6`t-`0`)Y~@^kJc!4r1tr3&5~N8$|N-|pkqWn
zH-X1(ytXR1lERHDj)*wCoq-4=lyq(k%JJ3d&NDgA?K@D5Y<TxjRpjq2spBr7hHiX&
zsQ$(G&&C@R=+A#|y=Lx}|2syO2s)hbd3Ww+kr?FG+O%-_w(<EjHT2vpj1RjQE2j8M
zv9|#0woFvZ1s|?rusj6}az4=1oG<5}-c4}pi#I#o2p1k3i?>kx4_5NYnEUIvY1<UQ
z#Qq|`+~{$ff1_8`LBhjxPQ{61_mR*@R1m7a*I5h|HMn_Oi}BJhRPhKBwG`O0AVaiK
zlbA|(mY1xLj&-#s7m-zo;U2VmQqvd(4B+<|#>l52hm5^RVj&Z>?+K2Q2=uhmM+ooV
zV5d68P!&`VgCj9U_utGXy;#?xf7HEHR;*tGdKvc12!n!U%hkGUMJ0h=z!;prO?rPX
zpqW2eSUXX~8GhkUg&aPuteg1s2rIZ8Q8M)@g_BM8&u#!b{{QzPdw~0v`5i;@O~Mz#
z*zq(;Tzz(pc=|3kmbd>}dVd`T|G!t{opd}XSc;o1fPyKV=aO&r7ah-)29|KhUq|Bq
zcnx^^95@-D9^I?^G^WDT`j8<83C!uA1N48s9;*8DdLaeg-X<py!`{n~3wvAqhB4yC
z?|J)wTv=W?l}oB;;St30#JGidho@s~Y7PH*0sM-lVqly6zoY->HyppuZHm@`=khJW
z0sWk%#pf+06iAnm)g(%#zvh9QU=0>8w)~Of|9tC!yoZw15^5JcbxZP((p`{x1A^)s
zy`I09>HoQ2@CF1ugG7qS4SmIxXL7-P-^wT&xBD44>Um4{e25d6UimV*HB$NiIrU6l
z1g}~dC18nMVefGi3re0Mw#dxm@1sBYsV5tZ5WW|FtzbVC_djm)pMf&qZvN=LL13@2
zq)6sZ>Ziu%+(hoY^qqyw4RE(25_?PX&vg9P2ap-!<}HO6Zw!vhvIn@l@@WeMqwi;o
zu2sr2>Eh-ss#1Z3SpVmZ40r+jp=wYN4NJr^ZY92-&Y|ghof8dtrrYG76uON?EDy_^
z6#na9{qz5|Be1G?w8=mrtk|Lc!&^xOnk~45S@bwho_u9WaZ6>@minJdnGpBhJ7Mxv
zoDH%w+{y!AnSd^Y8b(e21S@?lJakn>D-rU`fa$;QN(<$KbVH0$WTXf?VMI$Tvm_oS
zmKo-#&?1Wk^$&wMIag5%DBxIrl)V0aFMA0ZV?NOy{MT)LfgeE6cG2q|BGk8T+5kVC
zUUk}XGAO4?HL5G(ig;A>y6-Mm7?EbAlSt^J0`PxL(I+Ed_m(@T*ywxs#s3%zq1jg&
z5d6W~-3#C;0uupZMYNJ*ug2FP8>!t-f1@!Z+4tsEV%>5K8@MWvtLtWKtAN}YvG1Zz
z)<+jUcO{3EQlbAE7(32sD=ZwF@C6RGZ7V4rnEPrgSDc)IaDxnlvo+d%=*$3W-2HQg
z$}<JwcG+Jut@ZilZawb}FJu3&TSCT-&3XCZQP(vK83|6`BhNraziT#9{I};cy14w3
zy^OjVD>i@c4lQ9=K+AB3`K>tWA=<COPkzJRe_v#`et%F#XFUs<EKVF8#B<p$pI|6Q
zgzd8+@J?R!5hP2TZ0N(E;Q}AzvmY|U6F!S&WcQf0E9dVANivGyRayj#jm-<3pD9?r
zhq>o$Jy*GeIHe21R7lQ9)`<?82!AjRfD?|?4cayqjUUEE|Mdt;5|K$t&YNUO?MIDd
zyt~^<$?py$nslw%$+ySxX`i}PPs_v|ZWYgK7zK9QkJ}S-Qv}qv*>8i7oyIkFC9LrL
zy!Yos>av=1ahYiZ%wxXNj83FYS_~z}ZoAr#^&gD<Y!VmrdBw!eTt3tp8o!3TI`Ohg
z5U_|k-5N!t<f5+dk40u6OX<jR$BNyVv)p&EROP;yZ?SrU1&Zqf38-&oC5H&chP9eK
zXn(2%g2g@6K)r6H%4tv#&Jun7RZYHy8hiWPclZqrVU{h9<qoa<jGvZ)p&o0x-`xhO
z-8N!yFz$~KtQaIs>?1JR!+dFe({tZ|<M)bqAsR5)?S?Pj`9`zBsuMl2>iq($vOC=8
zH;R4Cv~G3WWcYz*R-bV=jx>1lZhLB^8J8BD-nAgSzt1!r#7-|(sr-7KZ?!o!yhnKP
zX0^g-LAS*5-1lX4{Yy>w)Z?p-w1-|p-QVa{HJxiz-S)z1*U<QkciER7l&CWBxzG6(
zG%5Z#9U!$(!UPb)oKd8U^1bAp<m1v^Pz;)F6>wTL@^<$6am=hu9apAG8e3gtukw05
zp=|Xc<@j`p8s0CXBp!N-ik7}sz{oKWymxLZyJ7y<`I)5)JVd$Ea{a;CoH(UR;o3(H
z#Xqgiler05lCW?Nx;nB(716cNQr{{Fy>k0&Ax3cHgurg*4D6Eal+T28q79O&KqCSr
zjQi@R4PUR%e8hJ|l=**^rOe1$Er=cXvP2uF<>U(c_OSrk=C?gL^+4mut%V)6qIa&_
z6mh?r5F~Lt`ZfKfY}feNbP^6r5gPH|JSQv-x<^_B+Iu+<b+f_}OJ~xaT`7B_gE4hH
zJ(CrpZpffpUPz7v%(S+hZDUF~2(lb_>XDx2)37Ucaed@I^}1!_WLyEJv>|5aT=-#N
zP#+RKvA2JGEh3w;(WM$CXg%2<Gie+J*l93_{<8}v0=O5+6d}n%_LvG84=e%Y%ZJOK
z9Z0nO{oSTNtd49$*soT3+g-yGKP_N2;M`q_`0KG$Q(B6;re9M{p9|4vP|a@YmGjow
zqji_)3L~BQiZc7C4K#S^Iv1{y=K|*!LN9iw3vykrPq!kJNvUA&E?bso?iRVTS%yh*
z>eV)uA}*Wux^-4zUy~l|#$yh~f6kbxK8{0rpYM;_(hBFZ@gxdYbU7-%E`Ht0A-37*
zP)6j%{Ijuj!55Kz(f;AxV0{{nQSSw~FnXW#v_^#ocjU0nF`66hhelqlbvkBdDP&$d
zZ7)L1wks3(uc~TXgh?;FrHVC*-8eNHh16=gRg+yb_9qMUl|D)=sP@HkJ!Df!jnXWb
z?D;gjT6xS!OIVTz3IU5SseuRnLLC3G3x8@c%N}?t)bG;D_?+~kM%`OnlR?HC%D_|g
zduwA_A=o_N_G>Qn(Cew6G2tk-7tbLbb9SY~81|?_BeB}qw|O9lqd3I55BqsWM-B1u
zVW1LY(HeN)$gZT{#RM0Wf(h1ioSV8cQh}JC@clVOjw8bnlO@5pTk^Zuj-j-GK?l-)
zMugjkT4>rIwQQCFVb^NbLoZAxPCg{%B-vY8am=P@9gbuPuT4ANEn==6Yqe-eYM^^9
z?y^~`pLSva<+?{41(cuCS70?G23bchsf}aZbL)hXU9vB0IP9A&2CMA$?w6Ykd2he7
z{>nQT&sP$|tEp~(f2?rPU_&`sT;>|&0xr7TBTM*wmItu2EnOGgfs@txkk^jjR`P53
zeioSIl3@I%`X4)b#CNR1lZ^C`d%+STzlR>?WXy%qq<tuR6z)^;+@!C~SHkf)%Vw@p
zW@+RmXeXobsnhi-jfm^Ev55D1Veg3c*mIDPqa$JNeG<-{%kjmMXvUa^h#UA#4Q0<G
zv??W>*T0UJY>{ds=n9FjSs$k*Fb+&qy%F&>o)n|kueO7$fBE4sypqkiui92mPupR^
zAmKH@$@@CbKbAzB`JNV`>wH@L#@wwPVrt%c$_F0YeP{&V@>9Ij&W~xrTQGc;JD21>
zMnb6VYi`@cqmUM5<@CTh`}u0=@EePlrIn;rOCm{w35f9=n6U9sT;Jps=*<yY&biRC
z@hRFe0dxjYPnQI7!8qR7pIoNPal@ON_D#djsV<j8xRFj}u2G$1*h-fN!u;r~T;R>#
z_NsTiEBxr`NcEp>g4cN3693@e#{denmp@d*(T?!%->B{XF{o2**t+HL{y&i>K8{<w
zxa_cf{<PuQJ!`+1gi4N(D8?vtEu;hHiPr#b!V#KVqSzBSJ(5V!)7cu{KXr;CwLS^i
zj!2qAPOE?Zpx8|7NoSMYLhw=f0L0I`2sXK|k&Kg=x8u)ocx}4A`l)j<?BVgux=b{8
z?s*#S@7d0OV|em{VSlu}o|HA<0QP~wTRsJ=u8=2L;GF~LD#XVIi%G{Oj$4;5Hdkd8
z?Yx#3YglV?Nr`MN<#+7|4mZI?U$eo{)Tt_j>d*NW$|=gejn*3s)nAr*%@+ds&QR!i
z+H6ev9e(R`p8{k`4XaX;KDm(WOKgx5Uo$eX5d^1?&MsLJdvUgyIcsQlbxvF9YjhO>
zvA#x}+oXIPM??hpW*g4NeeR1&41fs-ZdF@3nzk64FpMrFvrihWmbYC@{@QnkDs7N_
z8$Y_*nMUz?u209iTx~P@jrDnHc+OwnX5}C~-A&xI=;3@#czD)1WfX7uXVL>U0U!u*
zk+3fok|J!msHNJx-1-2PN+Zwxi(mqd6S6^ElWfv5{885x>`rjZwHp}G#Rr{cdaUS|
zd!8}y!Auaxrx>{(@?yTNt!~l(O*sV9`ofI0ec#G*zW(;{n82!X<SZp>aiK1RTOauo
zu*?0$*|6EIx3B;mXK%r!$h}w??X}+<)UVbBjhEOP7<V=h?e?F8<-Y0oQnN=G=eC}d
zKuLNmU744++k#o_Q|Gi&)>xz3`m|{g3V5nR=ZMI@ng+wV16z8Fu`QAgG5=U-uF1)}
zo?p3-OuKH4ob7kOaM`5yx6UJwrR#FAoNI=F2t&IV){ifGA=0C=$5WZ4dT}S;K6t(R
z;n#g`d&(Li7rDwa@owV7V9;Bc61T#T0V~atiasq&saCy<U7ACQO4~il>6M$&6v^e6
zz8iMFeK>4bUknV{mj*cs?y~*`5`TOI&MLK&yt-fVtp0q{oQ=iDiUm)RoWFSg;T`|%
zxa9<|do)L^O$-Jo`n&<7#x}V1qRyVF1{PoDrkw9xAL=cK-c{+evKT~Z`Tj&z#Zx;S
z)gey!@^NedyeXlA8?%I7Hv!)K^wHIQnm^kGYjoNfJX#D%U)!5Oy}3e;CPaxio6pam
z{*1=um0$2Yc$jdN(Vs#x_jN4KNG2Yk{^dGJl?e{)GV!wqyOx@J<Lz*x-oC@gXKyQ(
zV1SeiP&kWtnr=7P@t7PHQMS|1)Wg}}>xyN8G7=PHQ`|pwoASNf66PpI%z2zaC<C@-
z&h>pU`*_*gJADB+Lf#!knH>)ode-ljwLZ*le2W7<mMtCjzk5TLhQvJO;5{2qj(4Zl
zORxHEEZQd>6a(PNCbrwOV^0R7U2b<-E#7iajsM0<^XqJ>6<Xwl=0=B7DZD}hy)(P5
zLU2KuvWEgyo;&vZE*(0QI_^ci6WR`pQ_%1^I6TW=-)!OU_=O(8<C*)B&aIL1EoKV1
zlz*Rge+VaeCP8w4O;F5vmnb8Vug(dq(XVz$SkbTWHZAtlV}jh4uy6A`u(|MDrA{Jv
z(F1#%2^~El=ySU!35RUqnDo2Gbf#x&6|$OJUX%1r8U^2uL&}9dx?aRs3as6s#G(kq
zs)-<f^ks`{;W2)LOoe8|04}mu)o+hyzW->I4HrFs$GN4a`#|+KKu*+$PD(>CrsG=~
z6tMH=m)P7ez|ovLWPDvV4Zl8Nm&H&cD%Lmf6&*~|F2Ui9ZHh(j*@173#(jqZjUOIQ
z`on!n7H+&-y*AU*{uKQfBM?vfK<5YFcQNT>_@HU5X+<93j=+C-l<TrEKY!f=R1N7T
zz0m3;P)dI0*_Wu+OWr&^O^I}yRNZW0*m(n>6kKF70N<OAH%+C_EMw#t_u;RDJ^#j2
z0J(~0_q#OkS|fXP6RCLkk$N1N{`|qeqT_(E7bXSeGs1no0`_aq(|?6*Dr03|hP;0S
zb%L)>%Tj+_+KUf^>NUOeyAlzGH*>sy1+5lGjjiGc^Pq9#CSl?9sd;+e=%(TGmL>9;
z=C&>gxe)Y=+ifw9$737`>Bx3(B8th9Mr7ewiS}$o&jxC_osicHfOk-YQJ<Vw;WlaC
z1OV{+<3r;%L9$=HjLJDN==uGoQH!)loo7y)E&efx?}$oaBR7rxq_dWCpJ1m^TrLz~
zVoEfXfK`>2cb7qSAw_6fxz7IhK!Z(5tN8kPH=h4a9wiqveqC5lu=9?mgx^Y#V9847
z7ahyo7ZZiLryPWZd%qS#=JTg0(Xw^nX;tTYc6V&j_vR7tAX`%*g;WN_52xUmC)2cw
zcF{|>|AeuU`E0-joDT^ykW6I4x(z+XzYkHktF*%Wcc}RrmGRv1cWpi@A<Y)|*=>iD
zaGyHb$6Tda%YwGItNW}VyGJy<?h;45z7C8ey3jY7<2->&UeUCPLL~C7kP7u$X8EG%
z78yc{tyeznL?SxV4far?xg;{1v9hM?;<<WF=iaQCB$X%*4c4L{IU1i)Q(iKIS+~u6
zikzaeo5Ay52HJ<m=NHSkGByWB7ri}iZs3{T+nPq%(LX-_{06qkjr*`U8?LNgn%-}W
z0YX7|L5U$PR!-PxmC?N3Od3=}S8;7zdljE^ip(kMv@ck#Aj!UT?$BoXJjuW~J;0#R
zqqpw^ZTj{&r<mZR=Wp<ggIN)HM}f?^wDF*SN`rT<Lqc$w{JJa+SCWyyMZ)yjI#3z=
zp+nm1hKYt5Ws2j{jInB0;xs>uvcA#ZEb<3n?;So$uInRUONWxz4_wAZfwzpw9{MCC
zGyI!9$kM~n`&>eC^=fR^cgOU3m(*hOGu)YRKg24456$F5`kfZPno{hcas1m3+?mvu
zXPiI<nsvp2S~>01VC3XRfBd9{ulMry!K4*<?_50hbp1A0_NxpW6`2`+pFwwC?Z<go
z@MBf5eM}h)T8K)3h(a=$9QNtp-&k2$yUIwtS$#SzU~$S$@IDUdo9C`J&QSH^c4gS9
zJ9tbPtY0giei=fij)r@m(d3ObFD#_;E7;kNwrR(f#?dHwLq4>WG#mch)5YAc>NS0L
zLY2K_KPL$xd6~)w2W_`l&3~Hjlb}euJ!;Qh0@U`F*CU({KwNt<{s?xR18E{k7Yy=k
zJ0Vzr3|Tnjvmw|W_SA;ocDB0xi`f%MYy7c88bdw(;@_*V1tUz$&y4+>qV!norZ#uJ
zjqs=YYybRv7k!V@rTBXTJ&<0~$E3ndEAhaESFTKkn&yS54-70XZj;@c_{4XE+}JEf
zqPX+RwVSj$Akr)lEi5jneAg3Uv^|#3TM={Mn`vP5bFRk+HCXPSGm-7=?P)jQ;Q972
z^IM<qXz!q~k$qG-5Hf3t7CizAGM1ec4omFUU|>Tc`&wmwb@i*Qi$~hjyzeL(!G0r(
za#qaRLv#~k=N%=JeTVWbwvmxY23q5~@taY1F1Ih-6mPmFE>h5^3ATUG{tFGY5WXX5
z<g^Cj?x4^Xv&#i2V6?P;zc_M=@87E?LQ<XZeJIkyU0koXgaYMe?WY5mE(_<1L7U%g
zu@u+2%(Sh|QrX)UJ>=mI!MyfsRGPt=pXr`MBN)&1ODx%+G}XO-o@v;CuAI3^UOJv%
zNbVr4E~onmad#EnWh~+C3GgJYzZMxtPe-SrHgnuOv0@3Y6tWd3^$Z_O72mP(DAY$)
z)t!0q804i-makFZ);r$aya&aU^;?YlRx78dXDe>kyK!9@dXs|~Rp|E+TyHdb_GLMe
z({AT_;wEa>I<z^T)Ra{ml-$s&0ZfXkN91F84M&`;HLDaK)$qfTh$iD|%6HQQxuOaG
znw7jqz!o(I^5PYbB4}|k?w$yaV)XL+la2U?uj(N89|sxD=z$mFn;q+JKoG2Zw%Aj+
z{^{Wk3RRqwy&;T{{05BMmg1J5F?Dwam2cD7lSKjNo$9?*UAv1{$<ijhDQxb62=W?5
zBfH{y&VraPXSNUar%%}^O<QzU7d^Xe$#*iF3~8*jS-j6J_m>wo;FzWi`+gw&1zcsG
z%6u+94-3P&I?;t6DlzUR7ItB?E)n%6(LM9%(N~K)z9TNe%R97V`O$XPT?b{PJzTXF
z$);+@1@i$F-U{!iUi7PR+O_*0bv#`vN@R*dEBDDZK;yyGi`gcGYDrq_85s3;Fzd|+
zQLX!gHZ}=qqhD!!3M`gL`sAOJYJ*wR{$4(tEM#|6`q%Rnn92#*8+_xB{E&j>=H1hO
z&(a@;&L4|=zsla%%(vtqHjshH*S52y@2eA|YkG(K@gS)%Xt{c!doEwINd<Pt{zDoa
zkK3b64xPfead9!bA9ZJ*hL_{Qag2_zFJR1Z!98IXLF?B?+-?nEfNq)fRkt5X1M}TA
z@>7N`e7guZC@KM16nz<o4=bum3l`^duy=Doy(9KZkDs>~*q1AIl0JURJvr#>RI;EG
z%cHO8e!KJdbM(-@%><_MNnJae6?9F1<MvM=s_e5MzN^AzK+9NyO?T%hRbXAQUP?>f
zf`|&O4_Zdb8<#=lwi5uQC!SjPJ)2dWy$fVjf3<#BHUb*vDXSaX*SEsAiG@SQV5@b`
z8*_bo!b)SZDUs#brpi%&%^Ws48RzBI1EzpaOzA17c?sXGU%dPh{h#@g(mVcUAS1U<
ziTyX9?F-K{ri7`k^{yE0v+ad$zJjIN?|`y;fnq%(9OwIWs38Xa&ChCjjuJq`<+(ZB
zo!#Hmq+GVRlPcj21Y*l8Ti1Aso8I21RWo9BGre(poCR}9QIq+AZfW%%^x&Wg++T3)
zaS~^HyiGe=WJD{NXPy1z;)t)A8NdhBV--sAL-eOS6r-@PG!}qKEvMc&9f23Pcx#Y-
zF{ukE&G#{s5y)!9EYoq>k~{Q1-HPQ{7kW2oM0$H{KUG}J$7>gN`tXD`#ZZ<bb&uC=
zzZRuc;00R<?|n|`Go@kV>6^%{*ZA6ELpzsAyl75hN3IWL=)Pd5Y__t-QN9J3O*AQd
z^?p-M`QqaI-1ap$@x^V{4hBgw3xdGCg+*_FllBe`bDzV{jVB9oT5gD0yRwm&e=k7)
z0|ApYX+So>j2kojg&bwxkkj*vF6eLO{ZBkXfKxgVg2NbhDNox8ib20xk*<`14x+vt
zbsi)zc9bgQi&}M%O#_i`k$CWr`$SR=QDiA35GVh}CufFAgj?3>p2W=*muNfZ{=vZX
zpR|6{EEVhArG*z6-<zNaiO==yn*6OOk}lDL!WdfNLN>rf{$)9B()aK8w#H8wXe7)7
zt4Zu9kBqj4_joOV@|qYe>^GzgNd3VBZDBm0(DlJAwRT)Q3|EPA1;sL#c1fgb0hKQA
z`<a>snJ)2*D5HLBdzE<lK78Mr;uQFOc)0~8KpLmtSLc%U`C}vRbe(9%T|qLf@q$Om
zLqPM0WI742m%4qzW74gZL&SX0?Xyam(vz~uZi~(JiIlY=l^O-Zs#cufQ|-RaXtklt
zeSQ1d?+I=xqSes{kV`d_<!{%*iz5o6Ztn{WMqkHgc7G#Q!5hHra%mWX5|47;^?e_%
z4`0I-;!jWRX1$YT8&Z7S=B%m!c*MiUpqRejLBgIps5I_zdfLgy8EH~C>SQVGG(Kvc
zNEP=*KKGxuS#d*TTda~RpBu}{lF+`lwM!8H2jV3ChBzo>-8I332$B=|+=Yy-?%02L
z(%&fEp_3p>mHJ^kZ@0)d(%@#o^a7qMLs8!X-o9O&Z16T-{#^G|26?T#=gBzX@o&Nn
zQljFzLlxj4M?%kDDjG5}pZ3+)EAJNVRbh3EdwE-kAZpfL1;_~8#e60-hO-OX`3%Um
z=PiH8K9H6`-QKHe1RQ3fo?r1WKH2s(#P56EAT_=~)`uy}44J8TcaaJ%@=kvB%zeM{
z-C&~3#sl`Fwkt;YjRtQ^Gm)4=50PGg`tEH(`zC)093lp)yTWMYfvBf<M`K|a)$T>d
z*D%$;ccq{#{WDxS8bJ2MLb67YmY?erudhSjf(`wNw#F8P?skR6NrzyDk%u&gNbbaM
zphS_*!K#?v@U<O(Ki?QQTC3=L_f7dUxsT6O+^@$ly#V_!8fOONC@{OH9;HyN<MtA;
zkxJBD8ci-JLGBL&7;#M)v_EUQZP7{@(L(t!eeC+Q71FusZ1Qi?%7N__laiFftBa^0
zv08WlKjXezWk0>yAn9=Pg=x|zR%4eTsSqB{?L69^DZ+}ww7_FBDkwf?j+=JEk4p)f
z$~7yDVcLNj+u(&ucuzEhdkWwbe%AIwBux2!OezO51#xv1IZDkwp2~5vl!_-;&I@;S
zI88wr<m}-YJ*gMH4PHhnzS3_%TiIp5X}8ZpjMw774424n9*-S9k8@faAl2Gdf}+O?
z`tN-ew%h#_!Ey1#F1=7ToQ^qM;yS(M(Ko}=)Je3kN}BoOD}qbw`0w{FyuA}7!ZgTb
z%ir}f3*<mqpZ#3onBw+Yjit`=ZJ0=vxn`hM9$oOzy(w`Xv3k<o8?FjUUca1n+I>uz
zu=SSwYu@h8@px>C{#ShIQ4;I+u8)otrpJSnFnsHF*e5<qAlxg|JFhzv*wUqjox!P7
zRVF@uw@7Pb`x8OR0~!ks#xHs`-`(OmVk0ECA09fyXWB)l9-eD%dpL;N9<5yCe5vfm
z>Czl_wZFd{o+KZIQAryf?Y)=LZ7qSB7Pz?1N6Wlm397Aqwr48DyI=mD?_#>bER1C!
znfPxgXK)9Hw;gv8BK3W-$ULxEsMfhkz&AbBQS<%rH#)ILfT}@Q8NQY$2cJ8}rDn)s
z>hAd*m96W6e5Ja06wE!1iKgLPDsb#H^*OcfY#t0#Wul+|lcS7Bx~6C?c9Wodfwry!
z79_&Z&2!HdrzUgMx@dv%BK|wT{9EhD1ule(?e)Nz`|=e4eTzpo5#cEfxQa0H&M%5U
z-0SMcWDf`ffl*;{)4X{FOI|_3QPvpj=?aUq8JnKWn8oavV5q&B(7N-{3SI!tY&_HD
zR!LG#?>W8LnJl)?&&gMgByC$Ntgp1>VnS0<R`WpFT)Y$>aZvJ4HtGTE#b#(F?ftL1
zubOZ)6EpV~LrD9dz*F&QZ_9k4h%yjy_9BVoJUF7Owz*PHd)*tuks0%ahhtJRf9XPC
zvu+BIf%R3}&c=EdHC<O|0XQCDfvkR^JkI+OmC7cVRtSr6D&5ROe<d-K(Fg3l&jY(8
zvkfr%l0&@67ie$GAV4INpzj((_o2JqZvOhQpp-Z20vykP{X}*5noPobK*Z?aO|iKP
z%sU#!Mfl=wK*t7KrsbId@G}86d*l%@^%pZl$pS$FL+<l`v-D~*>1r9A3hIZnT)xiA
zL{Goc6Fq-CDtcRgv)BaHuvRdwUirz}UzJ@y?cSY|Mm51C!-M11eS2g<abIE&WhhiA
zhmPpNfzzc-7Ur5*mQwsZ9IA{}*$)VPm>ixMaXbnr59IUkHkoyRk)C`BQ~p#yHsJA|
z*j<@b(&@47!zSKTY?ZGp+D;pN9B11TBPDC+o#%g%CBwf3A)6jU24+Eb9c*9Fj)PvK
z{-MNwL!1CEnRJ7pw>V1ISvp5-7DM;$nubUz#&dEM=?lgO8}m;1HR)IF{g{WejFWI*
z-|?YtzIfA0+R7@qr#CM3<SF7mVUqmw|Btcvj%#Au-iIkcniMGlQWcO6A_9s+0w}!+
z(yR2|i<A%oDvHvZNL9LYkluS2rG;Lk_fCKW@*9rl-0MB}-0%DTGoJ}0Gkf-4d(U2L
zJ<qc)bA>GSq)%{W2<EJP#2gxTYJBoK{r<gLvA|<)Vu?No9$<<OmG#Q%3een-BYkVf
zgff>6rRDXF<%tbVtgx=Ud7Cq@o^(*{Y~aJimkM<x?}GU6K9D)3+?$V-;|GhIaxBV}
z0R#6B)ZgB#*WwTR_hv$!7=AN$XD|*upCZ7E6x7PejB!mbT|Lkh>*9|&@stc0%Tm6v
zI+sSg+!xUT2wJ`K5Yw;7`)UEsM+sT=#N;H={p!D8N9`0WzLO%f(A-ry2)>)J2)g4b
zS;_AkYXfhb;>BQN^bV;*43G%6%ezpFUx*SeIb1W%fW%FybE)QUKYZrsopd`^(Cmu$
z;k?lqH4RNR?<tx5*`c7`ryYSi&$Gj}Kmz~a*ep^2uwK_x607SLAaWBL)550}e*E$O
ze#l#KSX8_7&swsFTg*&NGjyIJA+=n>BAx#Zg#lhO84!SMZH=vuf;fW=YR=~-{)S=r
z{V^gS&nSUgeqjdFFbjfWCZSc!@FNMs!^eFZ^N;@>6#Cuy2F_VN69RCVAK}5#?{47q
z1kP413nFrKAJQoNdw%@hnV07gY=Eh75Yp2}Lf#N&xQ%bMYJ}fi&i;?K{0-I<U<~{O
z^`wo6jocq6e&Oe0srr7eg+*gH&+jqqzcnoYgeHHpr}51cu=JQq5Y_w}<~TH9r6AX^
z-=^>1ng_rh_)auXl#A%0E7)gqs`LYqdA}R_f1Jaq*Z>cesAXA=Z7vq;J#qwAh*kS_
z4%zb-?{Am={o{3DRQRZr05A-z^8?ZbWI0{q;34bqf5hki>skDO!I>3hj*EhXkjv39
z>mW-j7F8o}J|OsQa{R5=Alw{HEVa34Y}YfJpr)l>{*_0jg*Ilo26+F+Z)gd>2ZG-_
zj|&rm6q+oWi~M#wQ9sGPGH(1M+UnP!Jr=;$`x9;R2Eb%{;f)C+2tLwh|Gq2yg%9}G
zkv!oN^3=D|qWgpnz|T0DZhW+h5u9`VpSY|5P~as9ta;DWfr5Y-UR709eAuh<e|&wA
zAL!<x?XsmF^=sJiWEaowJy(hS_x$<0MbCv$roBtyxgfZYO`&#yjHUE{u`H;4_xI`a
zA47k+Uw{F5#hq~SUU?Qgf(%8PMnoKG&{`!GZLIYFVayqpmn-FRc4u`q9nQAI37v&#
z|05m#-<_fV9Q+nJU>U#DVu4^I=L9U0EQad;8Vgzgmh@+?6U%Q$y6=cc<(2fG7@fb_
z#~&XvC@w*L>|`NBN|PXi>*zH$#_65xE&TqG|2g0gg3CEC^<gsv6s3R_hx)!{tc4Nm
za#ZIkW&EF(^_T{`X7%kK7`HpvuSY}Yf?IR{_Z>i&s~-+CQ8v{p&{t36@tsa%04N_1
zFdF`4L4P|TX9KQJNI{zy4bdTJdpMg-c2WG@BKv=SKjQYKp({&h@pG471E6|us<SFi
z-5UPq_W^LTDliT)R^P74H$?_K-A>FiWB(H<^UtFVc?1~X=~DnN%wmGG*+IWA+w}fE
z-f!>2e<dw`oxtTdAk65<i<kEsb66r&hGA<4io%E-o%=Mupyq#j;9nmj%r6&HI!5{^
z$SOhA)PBZ0WHU36(l<l`n8#l>ek-mCE{$2OW_;v6_R$PLT`jIlofi21nJ)Z&9wY(d
zy3_Znnl0ipun#VS7K~<}`iHo#nEdJ$?blc_XJDr(Zkg)T7uP4<wT`$E9mYItg)Iye
z!2AE9%69ah@4(*qvd*Q!r7H8#6bZxkR9#W}+)Hm`WnO)et4?Z<8eyjY-5`P014xr0
z$K95$JqEZB9b_<&@;i;2@Hec*)_X+NFV>={xIY-%j^r?WNAnxhaWebWK@fb#B3kQ+
zf@^BECOpu}Gud(2ZUg|mmUOh6G%z*w3$E5hZ+W?{55uB>cdm_ZZ)h`V=NVfBH{pr)
zX03o-t4#do*+s1)sV9xko9%k8{R#ccjv#$Q5*TC^woY@_P$EaNX0#m_5!U?e_fGo!
zuNeOuj=BBjaF*J8Kf_Og+X-(6HNk^mz3<6ZytkgKHKLZ#r^G;VLq-<w^s|!|0o5^p
z0H-bv#0<78?k}j)AAR@(P_v7<=yXh|owF-Ckd$nwtIk}3Ci{&_&Wj0md^3ct`|u@-
zMglv<^|JoKrP<zG7-ncABXE|3`+C+j9Qv5{hU}IZ-u|iuuKcNehG(PaDRJk=p3nB_
zZR@f?@V!Kd2w<!A8p_Fe>T=gaZmG6K;yV8C%b1-0C4f|aAUaRJ92A3xu~^Cojzi}j
z6T1F{U%L-Id%rndIcvAX-mBHSV(g#CJLeQ<sl48u>Wl0xS>SL#FS8o3S(0!Z%vKd>
zw4YcXC8Jtj>S?Q4I<0-+y>6U(Ac+dH)wuP5f`*1eT2?s%z*H#&`3Psow~BSIg*nKx
zXi_+<Z?~R@2^gc)_FDxS4Xd37pY3P&giC%&yvK{wkg9fSGU0tMujnuR@~?>9W&7cf
zOO!4F`{+%0G3A@eIqj4an`UOuF$NYXg2lcBX;<@s+r{g_6bv}T&<wEtN$9<2J-bEm
z`AQjGvh~p%s@?iKe(y8bo!<(ry^0)#s#DMjv7z<a<l$;vgu)2S#lr$)6d9XS84K0H
za{%o=2znHYDjqG$GT5TcgE}2ZR_B^gp>phiw)(PmqSeR)2F}%Lmk7;`Xrm=gac{PZ
zvCSnmogS<#t(Gj<vR0?<gz&QeIm_=sSR|t(uUv0skAT>+r$Ki+AI=9SG7%9z#)TjE
zs6q>FXqqp6h+i5jp;XUb9Z@e*5l%Q?i=R-3?N#QfQ%U=J*v`GqBFPuTe<aZ%CTL=D
zCGx)8a{gs-BaBJhvAic%s5X|=!nC(ywjea3G*eMw>!YUBxHwATHavXAuTZ4W08=$P
znpJ&az*g4#Y-mcOnk?_!g!&*SJicCl^AxI!$-V(evTmdJ|AXgzmXz`*p3}D75y(`Z
z*Yq8%wrm{(H7H%z)J?Esg)IAW7S^jgB6^&54YPa<wUg5LlP7vLTJbYOsJBj_+4mZk
zNKdt^r5J$x`J)FJ#$P^&a{RU6$zLV_)VoU<317wp>@~x5b$59#VLfjsa86pU^+3A-
zGJ|6JS-JUqyy6m2ICR3N5z{(dp2q|lg)W66-NjF4cSZ`zR3>LPB^~CM@fDojJYWns
z1OTAs=w6KoD$W)&#p@$$84c&3{I;Dm7a%{Z<Fiv!^ZAa0N&V$&X$38edG&7hHtOd&
zJs%-!MyJp|Fv{L?GTgT2(d5CuuIApkQ)_-iw&CG+sEF5?ycny>id7bw=wwf&N#tMV
z=ysGjg>l_eI4X;o-ME4^F>!u?$@e=mo~|od`>0Jp!1V`kbZWNvi($DbY-3OSvoXi9
zwMF9^mN&VZ@aarp^ZTuWaO+;odAQ+zTd|e-c*w}xIDh5S(o5`!LMN#MW$bBhRnZFT
zet#w0VeoL>CH#@7WmO};hUNnirHj**Hz@~#79^>9<^Ue3+?DwN>hg1AAHY&Em_5vl
zJ919>^K~~D4g~cs2uhMb>MybUc5AN$3SZ(~Up0k+J@pj}oE>zx^f9BOqJ+FDic~ze
zfSOl`IIoRS4juK;JPpGh&ZRnmQwNla?ChCyt(hs7dEW;>8uEg9)RDYS;kp3#nove%
zmU#>SJw)k}GJBpsw<l6gsO}u$8farGMwopBsZ(Otwu>*5Y(Sj?Yj&%{;{|&Cs*WF|
ziQuqgHMlG+6gVwML(~aa<{VfCZ;o1&lf65y7O{h7SQ&w*GFL0ld-n5J?#ZN?JuD9$
zt^?HY<Zd@gD62Z1#mK8?y0%9sd#vqi4%Q>m<D?K}3$qm%{@;+L8#^~rJnE*I?>hqt
zo6l%9XfPQ^-s5PMbUKm^j<$!kBR@OdJl)PNw@Mrxjh2BiRPkP-_ih4)8S}`RUuck!
zlh|MQNzSoNBlFro$N@1se0KPn4{N5`yA3v+AC}!wy45?m_938VuTWmy8ihE1%*>P?
zBEFNP@*<!&i9I^DLVq{CwB&Fg>Ou5~NbcKK48w@>!BQ_D@Omh%;nKgY<*%JcV;2*M
zL02DW`8U*g=O0F<cjkW;pEbWF0SCi30^Gh}gt)OvBHemcculDwcdfqV8-3e&youX~
zX?gW7O2EmHOYdv(m>Kzp3uoa4mQa%px@Y+~oOj7Vr*B*(pM90P$)&@RI!VaZw^4MC
z?-5~MW!z#J8h!sL)atl$g7(2766{l#H$wP!6MGke={BV4m?=>4VZm~<?=}^?U%FBk
zo_arb)KiZC3rV_ZNaD}|ZmPx?GSvs(`^Gd@9(p`hr18DC-xG@BL|nC9So@N%E1tL(
z&6Gtt1!+2uzUFlGynvNf%Q8R1=)8K_B{^_Bs}`AAEK7HG*Z;((J$0C)#q)x|pk{NH
zk2nV9+`EwbRrHh=R|~}%Oe-p3Ep!}DXT*1gOQ-MkA!KHb7z#i~d^)mKYgLM0nVvbf
zyUNC=XlZ+S)+kSu+i+R*#sQJqi@4al(XJ}OXpd8jQ8KslIOThGOiDe<eX=8TT^@+`
z=es0P#PZ;Iu7Ard5En4W-!#iOat}Hyw-_lMu$VQ*2MzG6vlWO>eyvE6J56$Cx|~N5
zRF_+^S2rdj0Ke8i3-*|M<<S>yi*8kF*qgP|TDjZBTb1C$!;tG-`P4grH0%!~b*wOW
zQ0=z#+pAAX3?IcFAK%SDNxUeE751HpF9oWbL&+3TlIOm<7T|MqUaX2j3+KtfqKo#f
zc9Gswo|CP((^wuuR<grYQxc)WT$;rg^>x$M=flMrr>~PG%);!ouD|uf9(tZ`Jgn<s
zryUIlcJv$Z(!-{+ROEQSieM|<T>fED9d0wUe<j+&aMs5w?t|4xuC!e_m$T$@9;EAd
z1&jRxcX}rVSj&5UIyFYWGUoH(>0EAxBBxTiwj!)%t6i5O$SnEf+`oL1DOYe+jXyRL
zy(v{v$f|(-s-Ly2@VK%fg1)e*=mF$;qY^Z18G2l<>4*h4gnE=_xOIw*-o_W<?0E0n
z{;Z@>K-p8b_|`)8kD#ND_9Fd?-r+UoH8CKi&gs0h$dxF9vBZnxclEgG?W)F{7s4TJ
zya7qiC@#H1eCv74uJzBgj-E1l&TZeRp&6mcy_p6&EFK_FoN*63J*N`A)v3QddGKj0
zm|~W{tCg4^`ZWqu<y&3PKLo#a;(kG+)$U>@D>tz1rH{-sRg?pi@H70X?+zX`Ux>l>
zqL#3EP6UkV%pDq}s%L8vP_0>|#{hWPvgKdk@E;JsF|dBoV_?=s`TZERTTe#m&wavr
z;q4Om=_VGj6Q|@&fW>F~Y8RoXRpUXL=a+u+7@I2B6NBp2E-0rOa3oo9O3Qgg%*g6{
ze8H}$?N@|8tBt9+4bIgz<4RTKI`P_?vGZ^<Y~SDSo&sg!5`Apv2`NyxDF@mh#o~|L
ze6*#3mG|VGb`A_aD8CUmQ$Phf^~h_&i^^_N*dE~$%CJ1@*>C@xnbIz1vIlH@@<ojc
zpYB~xGViTm#IdN1MjswnGOq6W2cHms(m_lx!y>0)HN3*f9jW=I5SCGRbigz>HST<x
zn9Hu*j&C_C*d#Snqfx&5pfXIYn8@5HCB(~E>lA)yRkScp)3M)IhOqhwl9F%IS)sS6
zUz<#Es{9r_$P7>%Iyo=4{87rx^9?EYz12HEHxrir>>)9-@1|IPtmPxlD>5?g#JIgt
zYA%2OCWsxGAXM1z)z;L!^Sn#ag0azUh3Tz4Gb_6yRdgOP&ERQ@)0^_3A(bq$m<9Lf
z2lFwS$vSu4cI`KIm@voR&?XgyabTE4b<sw0n6h4_t>cmaklZX1g8Cq83UGqko=(cw
zN->pqGJN59$V01@#ji3vuIlcs(&F49)LteW7;U^VVElc6a)3I}svbywu0Q@MpR1P{
zwWvAkP)8^s0Nb2m$1ps|QrV-9(<x7gl27I*etZWxy?zg0LYkI7mC=t$ny)#Z(QIet
zJbSK9X3z9V=Hoq6Y#9H%y31t6)+OwzJ|Twn!aW5lzU+;+?>2WR6ud)4z(nF`q0mVr
zQnY(ud)3T@d|^unYLvX$=H-aq{AlHnnkasbxWJvrxst8QW1z3Hztn3dP%+q?-)Y+h
z`osH%=n%4@<H$DdY$ky&X)Of^ZR05e)4~S5;`bnRFx+jOOpz6~{V)x>o#^Y9EWX2Y
z9-LL($$ap@ZTU`R*tSN#RBXVA&_zy3@6_qLuf$Y3l^n8t%wemnMXoMp@w`34q*OoD
ztflX)xLTNch{ZUG3fr-aDV?~q-X71zxsaOU&bBc_xVTq^3%c%pf6k*OxIcF{+kg5-
zl;ODxN#@xAedX#GV?Luo!{bZif>e{)q0f3^)wc~EEOn-ebYHw+Zg<W#i$MXZ3{MqO
zZ<TuOVC&Xi(B$Q1^7yM=_xMKSpFSl3$oV{L*w#DWyd&U7@iSgozq~e!K2fCjjCiC5
zfHiD;0%8}r)HmA~P9?b<wYL}UH3K=f)zjxHFiv(F5Mp*@<D+XPyXv+<;1Fqqt{crd
z&y3zZr(4yoRT8ZYF?Y3sSSTAtuW@U^l?ct~2p^x4ve(-z@Q@ePs}3lGG&*<)UjoyD
zoh;_iv!r%ia1_lcYNJ8Q4ra^DKlxP52;`IdUclm*5OvK*<73&u3h^pinWp736MFSF
z(O2%G(kqpzGa;DASHuMsW??0B(&syvXx5Id@Cq2>DS;@bnGBdKmfIkG<6~u<gIHd)
zWn+Rqbj#JrX~X~U&CN@U)F|}M+p4PGyQ?L^Uwf+tGZlO^yG(Jh>9Gi|6F;dO&)5|m
z1<M*Zo~RW$H{)D<;i!FdY2-2Wb`7W=mrX;3=Z3Y-{WVu5E%C`-jD0~mH*$dB);CdJ
zPB9ljGF|!O2L=0HGaq>!9OcJo1@TUJ>reLW_RnIxl;Ro!cgROz;f`L_OOsCDk5Bia
z7Yo*QCd)i-#2o`@vRI%d@ojEJg)2Y#G>hDV97C{rTT@>2t#>pWHzFN=@F&eAKj<a8
zZO{Glv0SC^&f%B)q8BTn8(c$$0D3>LLUj6joB3OBBdym8FHjG5(FyvMjzM9@&jT63
z-^=nIq5AIV?r$knZ%7~p2EU4O_*jUVbjh7sk3JH=mm(~3>ygEtXs<b1v8V^2AXHOR
zSJxDOb@YgS&mhv}z{>UE2oT|-|04tgFTMtZnCz{`1qneaO-pd_?pZ!SLxe-VF^q#8
z?b2Qbn|=k{)WJ`N#HarKSGHss?F^32!{5f8wX|`)-LJD~$_$C_LtCd@NaOWjD>{eV
zCV-E8jI8r^O~~-z(-zT6%Xv+ox<NluvM-&$sa+8>hsT^F(@aJq5K2F)a<4Dx)L6IB
zV(zC!ml>H3?5k;+*~QAcC3@;vth)eDi0kANE#2<w`2_M2fIaVbqclQX?7NVyWf|->
zMa>e3K#E<k1c|IQmc3u?*ZCqI=5IcnsEjIEFdkuutcFxyDdjnlF6Td-ZrJK~4RScZ
z@{Rx)ck(GGRtAa%#PGDxdv>2=^m?etL_}HiE6)q5(Qlw8DM5zKWZDuc5-4TZrz#(P
zSx#L%p@Ehjz&qc&+OfgYRj2_Tqa^^ieYt!GaxAns#e>1Lc79F{p&kv>QL{QHi6l7b
zi^x&goc#fwcv}KU?S3sYg~!_}rt>RTqtBTfcQ>iGvLn;1{niuqj!k(&FPwI!q~rK5
zo>eb$TQ3<u)9Pqwn`tcaQL#{)n)OxX$i7uscZtsLncSd9R2t+~Yr0keF3m>F%@86g
z=kteGQee51cG)AjS9DTPrLJD9gH~GVI;Xa{#7?X!Nn1!(*>E=iWh;_`5V9Ire}SK(
zU@{!E?+xJdkCtB`YAM0dWnZ?$r<tOdN7*inf{zJaZn~x~8a@s6frmh$$M+^CcZ`!x
zYUY;M!pVGP^vl8-M7|f$mb^P7VOWl(qXGmU4+3V?nmdTv5N=Y{P?h7uDOIr76!{Ma
z1KGyNL$2eUn&SrLbf*Tx`HFFqeV1u~pQr@&%?jZ<itYoFTVe$+<@}_U;>Q;se$jgg
zn1M)GYiFEzG^7b+kb)e_Ewb5dVor(@oRidv1l@{&<RvEB;(@bX2c$Za9AuZC$lM4t
zo0W?f(C*3RVKN+FAi0hpaJQr3Z2a`5T^_6IO9_t~kkYa)S{HR7v){;=lT91$#6OJl
zviuhJ`pEWdOJIr`DDT#L<1-L-x}?z9BeJUV4eC|&Gh0n(+8}q9mP3^3P~BJzgt8Fx
zlc`{H&Cy8DC5e6V=qApSAEFmcN-M>m0@D|E%baKqF>*Vo%g}{~NDAYWA}zaj&M0-B
zfzlY5?D)Z)+gh)~HLBAxu!pYH_ESOD!i*C=*&H3b+J+#1^rCMZ{^#P(ENs4$ASgwC
zssRqQYwY0-S{kUdw-eW{YL(v>eIC@FXwHJ_JSV&q@Bw~fCQSI{b6CZ`ivwzl_4@X>
zj)OiOiPEG`IgR|3ahNu&P~s{$Nfbv4jaS*c)7tmkx+wO$SW{wLWIR_+6_dwEz{?N%
z$PBmLJHQXsZJ!l=4wgp8Irm(r&R#x#=&L7vxIXW{-oouH$55iXYo*rs?lt%*=@N8Q
z`LXy>@T(Vh;J(vh#nT2{TE#h?Hl+4(!}&&<;<ZZZ1?wIa3Dr?<F{N}k#7DI|b%sE>
zBOd%=!iILmINj?%!%3L9mZ+m^%I&Id<a*iPYqFi;L^kIiFm7c?){9Lj50p~pfl|o`
z{-P$l?7SmIX6oB&#hlFoz$}z)zeERT$D-26B$8NnVE1GZQYOp3+v}Y7Z<+!j2|K<e
z3EnAVpLv^+iZ4X#emV`9=gC+7UB4Ed1(v?S*RTh%AB}$;B1O8daeAx|t7YFi*r5Di
z$=_Xnf$lBx5o$Rc$~Ak)s8myY)==lyZrdyrY{!E310Sv~+=U(c*4U~e%sS?O<x^M=
zfgs|wFE$B(X#tRlYaGkrG*ha{w8eiyAheHUZUP>&THV;lL_cwuL+k#y#@n;+xX7++
z7F>-_VsCcw<ac(8A5=ZD`XS20o9D)(!OqVKYa9Y(?X7MW^8;y&G)RN8uFZhv=%yO^
zN@+E!nfYro0S<QK$&}hw6nmL#_X68$*)w|=Z|FMeZOAnf8k|>0FW+iV#)8Mn-rYZS
zQGO;EaP>W2??cuO#_(!U?b?T|IuqK@F0{onzv_i&k4@jE=2Gt})2T(v$E-zQi^@1-
z7PyL2I%#h&W8(-coW623q~?h^Vi?_H@_*-rIgR(U*h}4)ybAv6{Ls9uTkKm*X4}Z5
zm{d1+gvS`}{8`@Y7jLsu*S(%Js^gSIoQn>V2ZR|QYyhxp;{-7Jz6T3q`Zj_${L&XY
z4FH58Nsu$q)auyue&6|vnrHcHGdK1-EJHIb4(Jz0CvmlXq<`GsYZugFR<rl4-yp5d
z)s!tDc=?O$5gELV@F1VNt}f;9`9?EC5>;2$qg-u!mOr>2=&Jp94ac1%dV)4*sFU+t
z^*l>Ww+*DNisdI$vdkB+qlPq0yk}~6Ogwrv-)K<)v(POyDu44Ju*JX7wfAKnxcT;v
z7Vn+f<iP-a#*=(z%x9kB|0a8Iw$bz@KX({Rr!c_QEPN(;MdBjH(aoz;Vt+DgWb}6A
z$WvMhorAy!LTyZD*ol-b@e7Ns;ouS0pY9=s&6rb~cLfQ301*7Lyir52U4vZ_S6-vt
z&wIEkUA$*Lxt<5nNmlO$`24k``F4+BRtWq8$5dK$j;FW9-6m{$W|G8;A93hBts>3Y
z5ijdmbhmE!+F!WX?-X)qZFg&wl@s8gW0Qa9hMvqJ-eTP-27JStm+!B@*%Oya0lsXa
zXRm3f(Bz!0Rs(Lj3Jd<o6lfJ&)M+YT1-hcL@#s<w$vsbHsKaCFp+)Y@!py^b|EH^q
z(ZjZj>~Ikk1!v!*t#vl|xJmKK<dDkJeH<q4@rCBKGJ&|0ts4sN+2t0+cx!I!ZQXal
z-{<C!qp?$<G}Sd}nRBywrXqkYpFIA%uL#9&HE6t)R+|Vi@M#~>RR?C#sKJ@hYdr&y
z;ti6vK?$9Ae!i)N{Vz?d+5xEm9~}MIkp>_QH4}*smty`biXHQl`Ex9q8>`5Q-jFu1
z(pcoe_UP{E(#li}Dl1L$d}9AlBb<2Z>j|Zf5g6?G;<N48dw;J9p1~Yldc)<lRPfWF
zI~GZgJQQwoVs=w-*tdBEz<H*fY@Ws|x(I*RiF4pPCQjt@wQ0puaJc4qGfD4Vi+S|Y
zEyT{ywe%DO$Jkt%czJi%CtO_hZ8JjvqIJn(85JvrG3$P!ng{w?pQe~6%@jvc;wi_Y
z_jL~)N)ugY7iY6o;_?lSg2ITg+D}4aXB7_;KYsjKY#1y)GYkCw%DT&}VO$^#kpfn(
zXkdB|wg^dlPYQqa94}m(6C_B|@@cAbYBb-_sk%^VGZg#W-~zpM&@phl5#7myK^OKX
z7ED&y;WgHi4*iOBMqWZ<x?YebJ&B<V*lW-V$P9Wt>yce{B7^f3lBU+|$BmWRyFLTb
zI4a*1-ta%x%s1TS#e7@x#Z1`sSc`&>uc$%DV0A^F*KlDN834f3&L-{Sq4fgA+>&~0
z8b(GaJlI}^sWq*vQq5u4gNJG0-;czik>=tXYudXV@32Rq6wnr<hw#dS#e}a^8BmU`
z;A=U{za;r07awD4asS0xTd>rO(7-Usv90!Yx<k$xi6>J&GZbW!?1bKs<|1Q8IalKn
zL#OC18aU_{(+eH*@{c>2Qs-R}lskj?$oYsPdJ(<pTLguo)t_t@0;tfUJg;~Hhs6ka
z<3Gs-|D?aD38Dr)Id_4!?3i~RuOq!SwEEejgVd>yO4dl(AAuxVly@~QNjThFljWoI
zR2}V!g{uo&JFSGxExt(g!k;k$p0$DyAXZ|BSykmU-Xi~#grkO65<K@3@HA1gYDL<W
zah?mG=?&qPXSQev9=((bj+ak?V?(oRYnjEQlPeT{1&7CF;Um6dz!&Yj&0&L;a_gBs
zBJ{g~h^M>3Kg;GF)Xbnex<*pT7x(s&vzQMV4Wqe+(-UJhv$=`ynFMw8Dyk%tiTc;H
z1gJ!vv#I$@FiAgGdi*N)8OXMymPMVbm-qgH4M7$!_3>O&@T@4>K|klh^$M6JPPo{}
z`9S^{v+y&sTlMPOJH>HH!dUFSK()exGt{%Rhqd~2RAYP}=EFMW-w&RDpMMFSAnMTn
z_+UhqDR9jHk=3VtRo~=&dVP&j9xpJwtCh^%BVPmv!F_#4hn5ruQc7VSQn7@hPh{@Q
ze2BzlfX*d~X|IQ9;Lez*@)4N7{rSYb9IyhKcX4_{Eh|M!%F5527j$^Sn#t@$yHLyn
zmFsLAE*EIL&SeZYg0edjFpX4JsvQ|R<~PRBA<xzh2Ql4x1-$4b&nLz*U$m=7*}U0-
zJlJ;yzs#=S&okTl%((Q#P!t<C=Y}5;bDRm8HG1t{#f$8}F9XMK{~}<F1f<a3JacxQ
zv~<#5sv*+L4GYNIHBJR8MERBs9RA`JN$}*h^wVpJ=^3OmQXwzh#&w*YrBweJPP0^)
z?wrx$&?zw<+--%^>)uREf|glqK={e#c&U7$!2^q~riL(wZEXers<niR?cka<D`FZ#
zDon&><kbi|h5L0KC%m#UTPaF^^6PGwa<#sh&5qOKDtp1%)TDKzbP@0O_A{TU(4q9J
z1Mu#H%qdyMByvI5)t#5H73sb4JLf(|itTZ`eS=6(nzt#1NU`Y|l{rs>hP6FiSwC!o
zQAEa5=s9nmR&K^Y6w1<zF=nda(Im#fY;(tIvW#!%7ii>GI7Det`K{0`_$q4}?glpr
zxRBI^Bb|&un}HhSKUFNBHE{A+f6A;`tr&$0qQ4kNfv2W@gzeP9*LXN*it==fs^htM
zo)oEUlFiJ_%}FAdBWMwcnzgxP@eiC?L85n3nthd%<*IHuf6j(f<0`rl<S#0dn+^DA
zpZN#72I0PwRc-d$%6>>*iycL=A?CALHS<Q$ZGEOZeZNG%Lw%bSQ_1-7k~}c(wn8~l
z;-tW+;h!!ZOK3gU&#3wNTb+3Zh-cta-Z<5v+628tqm}p+{<@`AYe#{p2v*YBbEZpu
z#FD{aDdtfW`N?Rb)hR6?Q)B1PD5o7uXNPef=+?&&T(xUkn~jC^C@~tmqxNtpoH*EA
z`&rR%u48Cy(;acMemGat(2|RJS*99b;~K-tc++w{)~1Y7=!FdC5~mQz=tQTwnRTbm
z9XLj!_(o>b<6QM`J<dYV6NKn~cE|!Qe3FfoJvP-Jkc9gx6UMBX5be^Cl9%vqg<8Su
zWJ?yyu$JYbEkir2E0oU(;lEK7lgEup^2^+Vw!TTO?_T(Gd+7maVF8(Nqa=qL>{zwX
z7b}8@do@b@??6)o!#SSGevKU-@0sIXA*ZLC3#Jexe*3K<UTYycjCtku{`Iax(cL9|
z=29+p26=$KRHwOb{m-h#$PBck^~CFik3fDtZ<4WbXOzfPm8Vv{q&@!wF2UJD`>xiW
z5h=^OX`s@>V!{fCKF6&qM;q+yiflR;`qZr-nrz;44S}O%QBzjBgtvBLMJcDW1H^I0
z=?IZEk5YxvM3X^63+9~w1y5vX&-ck4n}N`5Q*P2%*X*0}&-Z9@^=m(NQzQX89q>hk
zDqZhyWY^K2JQu04t{1k{Y6y8XveA><FA7Cp0SWQ&vzdH!a)${ppn(AuUEyAx>9J^q
zA5N9a!I~0Os4A&7ClTzVj@Pfi6R#EL6hK-XyEYfuL?Xt6*H=d+mXBvK#j`%jJ5Lq#
zo%r>>TgRUvzss>;>nj@?OjbDF^oV&?1r`L*s3T?Zty>I%7sr59%C`%P1xuUUsqU8w
z+?opa<zJfA$`sF+I36LEM>ooyKLr}~Mrq91X>f3zUtDzZr}iI@i+BLExZI89Z?U_m
ziJ|y<D=XB6!WdO)?Ng!^t5!Es_>6QnX7D82qA!QTVr_pXhW_)B&jJD1i9BiDdc2~{
z?_g=O7!)%a1Um8TK0Y6r$p|3&U~i60^~DWqHJJdk8!lhq<bWL3TzLLU^Y~*f>@7z9
zgzaqUG{cX(vTJ}(6lYkdlz-1oU3GtDfTy@GI7waM4cK$5#H7jat)aIAk%?sE$>dI7
z!D4rGZ+{IxdHcPX92xGhu-L>$0~@c12@5sqPBBOJ(dtM~q?8&tICl#3tiV8$Ta%MR
zv(Sx;$+58@cJsq-jX~9mr%v|oeLSO{X?Tz?o<!^m)T-k#iAd}=3z5H`*I`cCntMJ#
zm_n$cY=6GxWoJZN<R+_4E-U0&g^1O!wrpApi?+oD?RR`7%Wy2Po%rduvR!|P2<d-T
zXSB7o0&CE5`pRPMTr1LlWW}8L^uaI}vERi_A@IlS^-Z;Gz5R{Q{8UlLy$P>7c=MJ^
z_h@wXn04yZX_^C>ub<4s?9c2gH(J}#I3;2oiz!7zstB5FnFSmQn0FLstES+M(CA|k
z2@!0TRaOJ_f*ud8wC_&wg)!Sm4;Kd89|%Rp1972iCPVQM7g9KTC6skkn_j~ckZ5X8
ztfkWEJUELE5j4r)+o@X+d3*HJ5pRo#Mn&!OBz1QR^rLST()$$Vv^u%J)FW40!KeK6
za~{;Jwea36<1-UPOpY6v&iU1%GAYBO#R5+G(_*<!BxoSiX=og%yQ!p}oZ${dc13ek
z##%`*YrU2=O+14-wQYaA%>__NRjZ0V(4!FxM}YYZTcUlu9noxC6hvkNAH~7;ME+2m
zNB(dQzc!pwr@ArkF*>L58pGOWrSU?_f0#`~4MAH*;hJj&Tf}QZ&gq6T@C6@9QM4i}
z`~}WfZ;jr0vqKCD#?!lMm6FKY8OF{)$0T$?y(EClIWtq&)S`%<IkF>Gu>0DY_oC$1
z;L*3qpS~oUfnYUlzsk`sR~yeRK-afrcJU3_<nflmy8KPCx9tJV1|yM+n*NWsMX=Af
zkTT0;Zo|5n>wO<<)yQMUIFUcdcvp*F4(IA>k`d=YIm<=!!1w#>)YKXsNo>-LJmued
zlsk9?39~I%n2zZ6_j4i@Q7-!5a~gcHT7lInYKaHYRSlHJ_TQ=RJqH2gAHVV!)yn}2
zh!`G+5oOz3>^k1x`gN8Q{dDFU#Sb(JRz<xp3dd+q)Ey{Z=<TFC{gE5fnX;Qb>p5e>
z!4s!Cn+P$Er33i8kheRFiOi0EEe`eJ)tY%9cWWJXqth+URSD@QIfHpGEE6sXfg2qI
zsFtFJAKBPRU(SJmK~Lzkq~WtN8R~3qt(0luH?avX+W`2B1yI6!4KcYVv0Lv^Cc|YE
zO2y||F7H#{Qa`X+o!Si~U^8c%o`-EOC@pAm`v+Vzt9IOu_C4zOVn424W0&;pKA>*1
zaAyN|EWp!#mDR1gl7~UD<R)gkrKq6PZec+<z-;I?iMzydhz@~D;Lw#+SmbvmcrdHx
zuF|fU@3H-7E&!gOae@4)!vmW>OD*_n)iPOjl)tjoU|m-$6|<uS&qxOKi8`J4=gQp}
zEizjgNUl`-HhEPyA1Ag+$PS=`*$iDU9lgIk`EF{k+b~?;s_qval;9%(5)P8-xN_--
zsR8wQe2h7Efgs1~Ev&{3N@H*RZ*7<o5c@r<TD4>F0h(vxBRb2Hq<rMiXQ`Gs*l=5u
zANup>GwbzUm-X;Tjh-l<XrRPsx4H@HR+Rx4GPL|2)kFz9CF)gwnh2T6Rv3*bp8a&(
zTnmeKAwxqA!on(c-*eDz=S6{4N2ge4{b80_z3kZ(kG928V?ux!F4G#<y9V7{>J}NU
z?X@g?ty8s`ffQt{C(4wWs*ip+?`Gv!>#TzA>7K!A5v_=LqmhFc+R1L#Y}bLanQg^5
zUhdEv!-Lp%E?=4mmPe~39oD{odqD8KYSf~mNISd6L<3(kE|eYILynjWBduH|;jCJR
zzql)F^e*v+;%I(?#dwIYD^Lo_yt=J}#Q(f@dg$q8;&(}mD7M%$pMJEs0|DOOq*z!v
z9ox7a8(SZ4GSqQ*wlGc9MrYnvV_3R`Oxqx<kS@N^`zoPP5fC$3UchTVA6wu@KEFNC
z*LgA{+wk-x{i*_#V87&MfZhE}%g`@_KwfsgW7zBHPUY;lG(d~vQIq<KYy%=_<*!`Q
z+{V&l-kjr;#OD?@E*?8EQu!i}SM59&7HvD3>_9}VT6~UQ*ExjYF3x&b^uz!QQxV7n
z-6o-D<K$CrOsuqz)}H;~=s2>IP@{FjEmh3n^`<s_%J-;Fmb+a)C0z<2Wt#W739KoT
z=vaV-{UY?y)Wf<pezEUid(>>@FK_8MvgC6cYd>9v9mg^qoRh6|0ff8oRQLu?ywWZ{
zv+df^cZhC@IXtGgUB8J|H9VPRSCWiNLre?*(WvP9R@kJ|NUHD@a`XJdrHsv)M7;B7
zoYSH@3FYg2Mj;tu;|T+M57t|F(64u}mbhyqz#mrj+Yc!4pcR;uBHd|d8Bc^2{VGLo
zET4lW6g5p|Zi108-C<6=ePS=wy~FlK5pIQ7DfPlrHIymaOl54tl<$Ql!R@OF_vh(+
zOD11@I;-}i4-oz3K4V`p1Wcag%i+wkVp*_1HvTyTSY$NBo&3aP6i0EMRB;L4$34<x
zo%2Mt42)2CW7tDYWOM=AhCGT*O>syYKCzz{)jAGUrmy1SR+40=&Ozu{PeSv}q?@*b
zriYPrEmmN^Z2Vg@>xMN{G;(f+*>8wA-(wY@1d$eoI+vryL>Xhb4Aq&Kchiz2pfX&E
z;k{4F;n?wybfTkOZ)}|%)=b{>s*}PFix<T%>lO&x*V>FT=B}u^3b^bQaWA7V0QQ6}
zH6#*C8^3{#VWTuWScCP&^M&PIuq?Z5m^}_sT*g+b$Uf?A;fa!ngMn&e@pE_iYX7be
zg3_Ob=h+C2#&!cu>0vpn5-(4ce9cyOPG6{Sa^*il5ggCdM$LNkfD|rT)-C*YeTlmi
zm;8$tb-@gvw4`7PfY<8zL_Fp~=4}z*nZvf%djbCev~%CU%u3ER>EWFLaQQw}NW&#^
zsd3ZeHTVI<35S+FrM^!{+^IQ!eYS~pa5*j!Pf|k^T`F(cuWO>aBi-x|L;5~Sl+TSK
z#MUdB4+z)=5=+tCu{;Xu(_?e9GMr1Y;R$K6AByj81DsZeg!j>!_?iL7B<tpIL+o(n
z02vZWDs)kuNWc2aB7=IF(E>g5^Tf%9X+FK?@b<dnW%uz<e&4+|CUlc?xYI{{<xHx{
zh+(V*9rPh4nE=<g@Fy)DT`pCbZ0xNHo9NeK#D&gAFHSX!LJWuaxwQRWJ33CwZ*RIM
zhEQur^Y}Na^QXLbWr8*wbtpO-daaMCG+^ds6He#t4Kd65Vr!wd+ulGPu_Y1RNu;(u
z`SD(qebC_MN4(}GrcqqP!xQf{*PWnCX*cy+OJPF`KqJPyEoVj9qq;n_)-D^8QZO>6
ziu2w!H#*!o**v8e(5;HyXwN$uSiooQk94X_s)8U+z9B!ELZ#o4)|?MlTCKx)ichtq
z+BceV<5gE+&$*EEuXF4Z)E&!+mVBnfmjBd9`3`J1$!^fmM?h8yGo<6!6|X~ggH!l?
zl^8)EMGEuy2Ihn7w2=?XXU|{v6cAoOQupQ`ohs^%(&f23@dN^zbJHkFw4{bK*HZ)l
zeXBiz2hPgtRoZ`vo!foon4#8;9_&~yY+rn6DD=f_O-69z8IxSfMdi(ikP0y!W~y~y
zU$GXsY}+NVlCMy*<6hcg)XBxT?eTmlqiRd1ZivBei4LLOux_$7wn|X>GBI`q{b@0{
zy&)R#3(jU@&Q|g1vBsJXEVp(ZT;GvrsRj+^MGM9c6AE&`m+AppuHkNsd(o@AwUD)N
zA2>Hs&@%_buRi1F5m<L`jpTGTPh+~~3CmQZG(I=GIdN~#Y(ZdWKak7rDUKYljb~IT
z)G|@V94As9UCENwu$jt9v2X~#TlV9Q?TN`6N=<cO9eWnJHKc*r7F8IjoIQo?x>9gr
zI>VUOwZ8g>ZyZwJb&<k$9NTre1Co48^9=(Lo1IqVfy{p+5a953l><9SZauqhZ^V*c
z2&ju(COnK%_rtdR0WNxAyy523F*+mb({N`Np^CqW*&=m9TuvUaymQ_+G@<bmNVwzp
zl|c+0z8p+Rlal&G1CyShF=e2bfR?KvT}9|agVFLZBN4}mTR>*gEs2SDePh(ci-)$W
z%ww}paoG+-@5HrwZLD%7=zQAtrxki)-(p(CWRqGU&!npi?5B?pl93PrC5hk_CkO$&
z*W)ZP#`Pn%8v%dv#;_@OiwFs;c(mAn+6CO$Iuo|3tgB5Q4miq%Dp=tJw*vwSD=Ql0
z_2Z>)I)SdYCp$T%Q=PR>#|-WM056u{yO&32U&OM?h7s~EJjdSRDwgu10IfNUW;D2#
zyGe#n*X!x?PN(}Fw=T&8x#|(iz_PoKU=4K>uIEuV$h5Y{9_?B@Aw@!VT}UMj-#=V?
z)dsuzpg+?4GUSz(R>`v_KE3n8HO+K3nqWsG1On%+UYbitmk1Vudxio5<;B(-*v4D&
z8i!W~?B$w`p1o#etH=?-4$@3>?Zg$<V3L?m;Yb>gk+?s)ii%I~@qPzBq#dco0sEQa
z7N8g&JL{jtyKF3%0kWSwv!pw2Gh1)dG#aoxWBP+X9KF_~EitwB!j1mt*qH3?m`*iw
zZ6IS%>n$0fG0^bo<x9KIK&DLmsVbCdwUBoYVZ&olH7D2u)wmYYpJ@#HVo3SG!8Txv
zzyk+IYwA-JL^qV_!Q<3n{oT`BTbAwb$q<67fcjJU7kzrHlv!hd2#^{g1RW=#F6FFr
z2iU|}Q;hX|1C6_2jVH#R<owl>O{`}N%k>+on`MRoO{UQhXwuxSpUnN!ADBq|%<&@<
z<!E!mq0UZwDUOs5X6yn(XyJR4ZBIBgQ1CE|z%kKzzo{+djt4o|WH%s~A!zi<zWvrf
z&vxLzp;M!jWTVNqtI)y4NsHO>cZX%P!ln_EWeO|vKgmd(iRRwnG507^8mk_<k<dsK
zPr*!gS;J=IadpEdGhP5n375{U2u_d5hl#f09~PZ#<ScMseM~yQ#;RR?iV(H3(vEBF
z`D_fWc88Y8Ol(h%dpOm|)&p$hm<=*H84ve+6zvD8;%-GnV|Gt5&3X*T-95(!Ypa3f
zSsEX<!g#k^aJk#N>oIhtTcX;2xF=;!gLg_Wi*hO8l{i0~4n)Hmq1FXCN+y77LHI5~
zj&ZC%2}R$ImXMh?{j{1@f`r<@5rV)U#r;%NHY_>n)+@?D&eT7I*thpZ3+X!>muMe6
zExlLUo<d!Ec*tnn6?G2An@)9UPv=8vZe*NHHvU1b;6L36<f)K`F^}2>*hTtzAKWj*
zkJ}L55Jgy=pGmN=z9ubRm%JH{RSyuKHaIf{K~PMITYMWQ(uO^RTGO$&bOQ~U;s2t1
z_6J+JRsiV?KU{jY0$^BUJdPx`W(t(n)D@aH3e%44gH8>e$zZ9OprYJ*3cmkESmwf`
zB^mw~VR<*U!MLCn0!Lrg$UYZ+SyawedG%nTaz|jvXwO*OQc}1OD%7Nnu|^^1Te-gt
zx{Tber{YqVt0q)(_Fl<zV2f)Oz|1rrD~eE)5f(ZG&Xf0~;ySm#r2;())1>E5Nz{ZF
zKg@h@)0eAzc&$%<00Pc~_XT-G$^2q2Gf{IWtz9vgz)mD>?vb8V<6XrR!2iFP%U85K
zp>K??Ei)7$?`R9vtkid}#DYG*1>BSLH5<2rNJT8z)R)=PLN7$oLRT`c)gL<bpIFt7
zYKWr5^+l$i3wpE!Q9wbD3eM8Qv9uZ9f^JZMiDCmpl+JE0e~c-AzNyyLEDcjkeqA`X
zSiEkuKVrX?Iyoa6<Er}pjrL4VAcQaH)5(E*yXg)>`^FzU=J_CoA_(5WUL|2b0{}(l
zB`|oetstM4U+lf$fXl())?R07M){K)By?q>C7II5MzYvji)%$#Qh@!t6}@%m>cEh!
zf0~1`xc#u9ud=Qi{jl)x2BEF9)-R02pL#8FENNb?v8Kieg)Ybk6ifswmjGOK&<)`n
z_7_ht={hV?1Tnq3N+$BDR6G}p`M74Ln`#78UQP8KEdn%lH1`T7JR<?55|P0cgzN2G
zt#dpMu|I(bwk^UHkC>$WWb>C!_Ae$Liv_?oNxhr&fQayf%~`h^{ucb9aJuCrzfbBn
zIu{Uu4@gXEwF>X;NPpNWas5h8RHeI|!NC{@Pyp6sJkKDPg_hIrRbpEHv6!6n8rbYC
zZ$e?Z>^<_=-ul-uQkms;X?Ev$B(SFr7-=}CX3SgW{x{2!rIhe$u3vn`A{mnRd3AV+
z6klcxee9Fefk8>nOI=)?wo5(I*H<_5a6PUAf|o}5Lzi_Y_)yQMQ^yt6&K`$;_?`L+
z@Vu;>Q$JTdpe4Ld>5TU|Xb1W-KAp_pN}=yhy2>TjCWZq`(f_0!v;AFW+|RzUV!hG*
z0lzM#d@8>Tk$E>??0(yNgY;5dr;R$G+#Lbv^<^5nxVZF0VIneAnOW6VQEnvT0GOcp
zUx3*^R=4WQP~MbPYr>Oj*aw}|I9uxV%umQt6#sQ&IciW^09iJLL?%mHTbo?TXE=VE
za;~=v!|#2bap{q)ZH`5Oq-0oBPUl3UBJMu;tNsHPAv(1lh4uck4(|O6_D>4YGkMP<
zzu>5E-t+A&%z5PdMC}??i3#-BNUem2+jB8@OwFD^;aUuTJu+^wU(~B?ymrcvzk5mU
zbintymzAZJ7mQ}M)oG;xD9s8Yhy44%Qj3Lsv3Yx#CBhwBGzh##vm8jk5`cyCUk5dF
zJc3=C@$fKu0TrCZIRTZ_+XI3B@Vx+tD@Z-sjjj;*=L2H%33g9h;;rSWM5h0`;vaWW
zE5-ZJY+MS4r8Gnkn4eMXpgf7Yr(Nj($G2%O8_}Wq%9r~o&W$M!eAwHwE^nRxc%r|a
zjp7z00DSZP)j-G$o_mKs3qI^nP+k3V<&S@CH5LxLMgRkv{lIM{mpC#Yxmkb<u8Y2T
z>iuS&=5zR4{lE9yA9qRnPIxD%@jH}lN==^fI~MHqDQmmEaKnIpWz;c!)4{EO>c<FC
zY#g+=+X63$;&$4aw*jeta+z_3-!~zo{CM;o|0_~|wg2}~O6wxJqg&PG?>_ay<(iz}
z!LbAIvu3Bj^H`o0hB*QLfp2dAIA{UDvlHyDOos_w>#BtKy#6qbf<IUKFuVP%Xp3wY
zg*<>hxbmAZ{B~=&DH|l9krw*lk-`tkuCA^ax3dC5wpqlasNls}=wKfy{QNW0#<lYw
z$B9iF=*sS+Wm|%m1lB%GPDG6rAFBMLuiu!<{^w?5;gkUF>^%NN4acDzOD^2=Q?4t*
z2orBjq$}B!IE7@$pdCrd&WJZ#O-%lAutU-yZFfws`H?!(!9R0E(8)FJxbAFtrfS)!
z*ITkg?6@9S!S3Lkk1OZ=^FS^VlFtcv1GV4`2tb}&B0|T?Nf(<9&1jV6^~N|{-x$(O
z5JsuwsXxxyO-_b?eue;QmM3J51O#llEYnI;s>*~ihss#Stw^63ThAJ1rkbWg*!4yj
zxVy4%>Gz@icb}(`69J8Q$xKxFtPbz)-MjX8FuV_aXW3V;JwTths(*@ljha?%x|!YN
zZxRRF6IodM=ga(SMyr|NeEcYDwQlEZ+<KDGOo#UQj=G;*@6>{tFnL$Gz>QNLnJ0bb
z{7*&-{H@&>7{hjq&@v(MTL?=Pvr@mAh^Ii6_<Ucfr;8!Uw$=Fw>*>Y0)IMbD%73pG
zmlxf9ify^O^VFZ;NF9W3J}*NR3QM9PO%g-)A1hU&(^Zv;M=k2zh}{XvHiMbtsU%Y<
z|8r^jt7lpO(+Ga&{2mgBhtmu#D=qA9b<;`}hEt(Dm;JMphuk{SN-LBF5~+vDK>I;E
z*nvkALH}cp6M#Sjy9bE@5NyJX#19SnY6v`}FrTWz9X{0%!X4um4i9YVR6{a8zD0@U
zfqBvfl+JO^bZD>tQ@h}%tiS{r^NV>T_ng9+%iCRYTF7J6KkMzf-Jwws>PD^TBsNWK
ztYh#hJZtI4T*Ed#&bepjx1CC1xb)=m)&!%MiZ!3?KTaPOH431a@h?>qtdu@BZM(bW
z_Feta)Noz7+sw;hcV=_#5UL?G_xY}Cn_#ZUa#0ylqn{9d;?MG(`TfzPVCivC(1hN_
z|CrWM{6MQSFjscbr`HKGe*W+~Tb9K=xqS;}xsz{z5@~T-ID;plXpHprY7U7&hZ^cQ
zWk})Gq9?fLh30t2h9vs`IQ__h^4<YhusA$lf;Ot}ybo_=SL#oTK7^+=bd5uCg^nB;
zW)BHLD`l3ORv(3sHnF#@u5z=7e#FB$&9z)_I6fGdK|mV43hNVBv0;M;ZRjfN$A69f
z??#?BMjTVK$(w9rANa;7%$f4@*Ushfaxv-f{stNT))2xcA8ikqeHB>9b0r5>+eys7
zmDwn%Z{5n+_PORH9V|qWKxf%L7(v<~J=E?=JQF^nHtI6EWo0;hL?)pn2P2tScf%VH
zdI;~e4dp6qW?FSy&3Blt@t@!rLl}G?oorYp^TLQV*b#Vo7)n*{eORXxOW6~gv}a%R
z$1T`CE<G9vF~KygY%@VBvy_984o5C`{Ti9EgYSSJW?>F?%WLO@TDjxsjZ}$YDVvS@
z%hmmW_WIX0KmJM9U91O3#9xDF-ilA7T2~5^w^nQmnQ0}8%c1?#bj-;auom6SOicHU
zIs~2vF4*TI><TPn95unAR>!E_{Ey~|4N|CraFM7?eE;;0ZK6PiN!j@>ZS7b1DCb@~
zRjBR92KW#zd?O;EonK|$AqcvuiHkkZbOI;6S@~RoLEKZ)9i@A^tvXP~_0*jn&Ehd&
zsE}8>jw@R@k|&xR6i!}8GV~cEd<0+F4P8z)?-;bzm@2DTL7-f2T1|@IN%B%n$Vwl@
zml5_#ry-ijED#MoP6#UB7TP_-pHy~Bs~Xyp&e5DpNlI||R%5c;k~R8JHTVZ<v#IBF
zb;a#nlV<*`AJa~|REW*j1TRh`@SuC-ewK^_P<8lFe<Z!YNm}!B+;|b#x`QRPgNG<n
zUFMvdUAc$Q!m9WSN27W>G6SJrt~+6jBUhPoL}fh<if#cFJu(noK66c5<Mg7iv4YwK
zr3d>}cU>1txYWyE8-a!I>l;5@gRWlnpRO5x1cxFYE-@>iP87A6VSK3nF-N073O0{5
zf)XP`uH40q?3XLh%YG^gy*5WLiJNwD4XRuLRLQQt(t5sfHZU|F#Ga=*=Tj)LF@vTN
zEN>PRCdfll_-l=N!y^q~Ua;fpKr`#aE81J5?g-n5DUJ^HxmUMAg=RN!Fy%pshO;8h
zU`Id1^0sWaL-Hx>_v@R?`B#S%L@Fs69y*I_<VxDOnM@wt`JM(_mrNV;5g!ei!v(37
z1s})Q(|jk@J7f@z7%$I8g|%jEO;1|Fm>D_hK>lPcj;-#lsi}_2g2>Ftw2)Cyn6gFM
zv?t+c4HPq)w1&zGI-fe4ntN3*=zh<4?lqCPXW_ORnZ2?xRQLHst?I_TaZK8?HP>Rf
zOm6I2h!_*e9KOGS57YSwVc$<g0bT2j^-;!2K$(j9pF1!hz_=-~xoEG?BuU6V@A*C?
zVlXAA*_i%2X;OsE=XCx%HhB(BcgdfPm}Fg%-m*8V54b-2WHSdIjq3nlQbTKt=AJtT
z<g?PhDW~y7E>*w_;ZC$4dMDt0)LZ9?B0FQEaoZD#gAOJfcQo`;f27P!Uio(5IC~_D
z3-Y-#86-PXtV=^+)OSQ0#{B52(YL8ex`8WUM@;u1jJtgkR~&^mN&>a-(w|QqPBn?_
zTHokX5d>bH6dweSz>;bC4s*x#R^2tdCK!`iuat)sG|u>0MfD|f;?=gZA`vYeUnhyp
zkC5QTnUt|-kjM$#|3}w*M>Vx|U!a131re+$ND~zi0TmFbK@pLzpdd9W(o2-yBXR{5
zX`<4LRFM+tC4^9<_ZoWWEhK?J5=ebVz1Mqx-+S+$j4&8U&e>=0Rpy#=tq_N}`chji
z!Ht8em3bpkccEa*ezDEEn@_%#pKqGHn{E}Al9pOvmoWbro--wGE~RnFSnE?_xnY)T
z=7XOeNfwggV2i;tRZUWCkIPk*61(@?qS?vV6R(I3oTL6j?&aY0mFfn|>;Epb|Lw=P
zmh*(EcdOT|r&vwP0})vhTB5$lPqJIy^@)q5&lL}o?)Hvjvo#hue1+c29Qorgw9C};
zVwl5|ScOk&_1E%Ec8V*0d<!2sxq1fG>rz^D%D<QBJz<Y%nrMo0SN!#!-h{~{dJ|_S
zHnyf-_<+3E$D1mPrd*mPriz`s%}s<{FnC@GN&)?8IcBFwC4fH&n><dpew~H^dCgbV
z_(Yacp0Q8wy}45`;~#q9$G(%@-P(xE<=0CS+-y|h_-rtv-X|E|QDZf8W7~kK@>j+4
zzy2g1(e}Yt-9a(ZpBbO~o)Fm^2rf?8sK~(Eh@NZTo9`b$aD<5b32;6h3N}<siU(qY
zVm4M5Z{99%bFKj@dFpI+@SXtS@32hAdeBBx0wztR3TDmz4rATB9U<i0UGw-mi2sk<
z3x0L9iX4S{Mj7{Im#dX8;@Bl7v8Yd^tPU54{JE-^YRar^JM;H54BnpD|6`DT(;K8=
zA2@O-h<vNQdj;iQ2IHi_w;C%c*Xkqi<C9na^{T<o4z6_3pRtjrWLC?!6T23?&pQA6
zXy%Sw{Pely5W^>Z#poz|tvx*O=bF_wrd?`9?C(<x+Ht0<nUhRuxHIyuq|IC1bnF+&
zGyg8YIYTvfoS34dfigK4bxsVbY`FbzIQ*P7dq1Dz!HV;y_3UG%hi-*XO^LyXdXT(n
z>AK~2EAy$K$NYk$l`4&t6$Mk&7Jgq*OWMENt^If(yQjO+`1;f%eaC?BE5@C0PUHb+
zUbYI7a1UsE{BjMynDL{^?QM@;yJ^-08Sm7gJWCyt%b+1?JU`$OY=_%>eN<$1b+f=!
z%&^A8Vaw<eqaj?a7wE37lUt-Ys-!Ysc5HEE7R4J3j|>Qhd*^lngMeME$4YF@NlHq(
zX@aq6W>*|?i6#m3-i<-z8~=nzbn=ZQv)=C(WPU2`fj<aJQM62ybrntv+_5cl9TNSu
zF{1AQQL5O-*oJpGYR1o1rm5aH15jva0aeP<Abkq4|M|ot7J3``Rv%icRO6qZ<e=5#
zmKydK;Dx2b%cJ#-0wQNK!}*bY2;Fs0;ow|Fk?Qz$c~s=}J6icDwe<*jv{Y<eLNd1B
zyxds|W#w6%>^4w5HtzQA2EI{Z5ejZ0f3WRix9eO{3SE|c7UPKGy^wDLD<;<hjsDJh
z;u%Rm4~;yx{$^CifK8_hF*a&iA~!W48L898m_#nOe^nBSHV_LqVu(NN%I8;Z;l2nx
zTX*hJwPU$lIPqwGMAPp8A^7%z_|vqk6|LhJ=7Ez?&`y#%mG<egCOZu~8w?%f`!ZF%
zJ7HnD()kkTGWl=Rbx>Q4w>Ba-dFb2~C$QZ*5|dO2qr%QyQ|4LW)?KS?dQA@Zi12|4
ztWGwXk>@%gZJ7qD5D2781`vRIQ7^0;BPkUB(>GQwjn(?<{J84LK{{`8C?=tFsSW5%
ze(dT+{>|G))y`w@?DfXhWMpJYhZnOmgbN97AdTHKDdo(*OoPrTvdU4ysbbaq)1{0L
znNc8_HGJ%0S;};nO@{%pL&kt%LMk#;c;<Zg_f?M`W3`R3<jtBq7Pp+Z|2RkRBwc2P
zdj<~YhII;#a&r%bacpIF_eB}wi#q>}orE`8R}&O+ZGcccXYl~{INT2*ky4HyAXdq$
z`t_s?w%keS7wRoP#{9vqG-oD5^`C8QCY=QoetM(L9y!E3JByVFZ;FFh+h*%_>{549
z`@0hoixc?VHWf0+aYD$KW|~wYCNHzgD`oYx7du><w2cjn_fEO11hnFQ`yd5V^QxqS
z7;${_*hy*AOK}!K2;JS?%Br+y^w0>A2D+;ycXSe@%FL`{RP(y?<T<oK>RC}&cq<1J
znY^8%O#0z|jh2I%6HJTzeI$>6+#|EYsr~syc($6cFj~cGW>L*P(8sdiagcoy|HDc)
zGM#Vz_W=M-!E5z;_BBKRwIPAZ@Z(x{;z5JU^r5oZQCtKVw+oqU15ynpw?OASYnOYD
zxXsp=*3JWM2;Xwp%iZvwZAOh!-hu-JyuS&i(6MzV(i`YH94X5Y-TJd9`^vO=E!_&q
zyAFvts+xkAq7q*Q6c7N$f&a3H%J<TMD>xf1J+5cUU#$>{vQFcV4&!{I*P-_yN?qzv
zs-h+qUPnrItiKF|{xq!7$61ce7<716?Pt}vD@+0%A7dsQ;9kLG-YMc#iJJ*JLdl*n
z{$na%lx|smpN`qV&`Y;~`~@Tx7{y7WZjs<u5BkBM%-?}OT1hO&0PV^BY>)KaG5q!*
z5wSi}IMMaAWYsHX6qqjHTbXea2Uwr6`GY4&gjXq@-RnEPTPL+0&A!RH4ymn;<Y80h
z(ft}`-uP*3#21&O>n;_b8Zgjt7B9GcL;k#|sJfLeR={67f^T@S)?Gi!htf~q+wjOG
z`O2mQHF@*$_nMI#)rN2{y-M@P6O)gs>$0^!T?;i$Gui3RD?drsrE~7V%rz)^<4V$U
zgNn?yOBwUM2^poMN@Wm<;cqw>j41sJQkvP7B%O~ch@}wM?jGB^c78f}%MZ`lRjIJ;
zeHT`URH9Jh+cMHh{T19}d{^Bu8zT3&R^ga6nJH?$HSQh`K{dvcr}AZ$BKYM$`r~Uf
ztoZmJshJoN7q+p7nY%Sen@;0?G+44`zlSn1szL_(8~fnVsAt=;-Y1iptSQHO{;n_q
zBs#|Ie|z#fH!SKnt#@uAt^G3RA0rancy7j^J@cj?bTfDDMd@z&6c+BqMR3v@xg5kb
zI_=d;zeXCaT=$$3Iw4^kt)$5nE2FP;pM)J9<haePQU$NwYT+yjKah@;M4@jC2to-B
z7{hV4j}X2WVKf3p78ZrGKbWNC3fCI1a5!Oz#>cZ{D0I^o-@8k19)|E=eISM0X2su-
zS2}$<!Hr%X_jbt$l`xF6(bzd<29w_7QT&L@1AcmJmfjd;0yD)t9L{^V!NU0YM5wY(
zXN~HpU2p?t4~;#1p&l3LJj{gxi>nvGC!OkOYy69JZq@Qu9bVlyry?N{2tsDs3A$OW
zxfq{WV)ed?gKp_J(;&9cZfE83&?~=Ay?ci@8EShM!Bp!-Oe_>ZUAyubj&PXFdf4H*
zTalm&a79I}@e00DDrW+TKk|ZWa|>pK2x3-mBs(KJtun{@m){c$7<IE$NFBZIEAk|w
z6;ijO28i>7rX#F>hbV660oc>T!sAZ;fggtlFoziH(tEZi?TNuJ4&ao)75{i_wb`S6
zrWCVDXPUkDvh6a}jsQQ+z_AB&5;XZm?-P~0u_c^+LDZ?JvCyH@Jtko9(#w`L%H~P=
z8`5_uTfIh$BNKQ}JrQ#O<^l_ETh?2=R?2vJZN=1*IZN?7DD4sbY`Mb>W3tpBeCW8L
zy!DbZ|0JqI-A~}SH(%GV(}>UfP=RR^E`J%xl!`4}GUrm<QXaYNjbFIoO?Y|&bB2B=
zRARP=F>kJ`?+bm7PB~ULRS^ic#6MocL;sRDMSp3!GX^=3>~2izeFj$`uWw1m3gyy1
zU@Z#!UPteB_$a7-H=4Pj?%z>vVtE<pE&J97araX!bZ0Cdhax|RfRWe@axJ^p2qMx1
z=8CNXAN37k^PUzY2lGy%;?4>o$Ojwf&(B5vGAD#{_d2qVwhxXa|D|^I#G5n8y{k{G
z+40up>@DZ3(qg5t|Ey_!>M?tx>=#3?Ul5CAeemoX@@#HcxG9xls0Ou22UFqA-k!+x
z4z4CwUf^e5&h0KAindazY4<u-Mx#u$69^Iz-|)UA4{oU{m`sMJm-kj+;yAl5+Lx@J
zz-ZOwUr@1|S-fs<TH$P}G#PIvbW$&n@YCZ;!1fKS+WW=!q>N9H0IoN{e+ek>O_tNt
zSEb*AyuZo&R}iMr>CN}j7Wok51GB_KNY655;<}EAisw=_#|QV)=a9BNbs;J?*)nUM
z7pm(nk0FiriJMov_gMd)hVNm8&reg^aQ5#KOZ7pfex#L4ou3P%--2~N$WQyBD=d{3
zPC7W~%0dwml#H8p1@M`=M6Me|pVjHYaJ@m9)g(L?A4Wn6LnGFtb)s;VldHbtElN&k
zP!*%7Dg!w|>5hFOij(EUFc!XHjnkkTa&a9Ad-V_I^{&oLkwKqzU}qfRtKqEOcvxP<
z#oudv?iK(}zP<PC{n2wg@zpTy<Sm_-$+o-uM)$ABrGte8PyCJ8y(@g#(}7AyB&EPu
z0#mw~9U@IG2u^~;%ewu2yd3PNvJsIW$zC`lGA3}MY+K6r#}h;r7~iy56+6%EmloI}
zPUzd~?PI-;t`13dU&?cP#x&B^C$VRg!qa?%3M?E>3Jzy*PC7a~BW%Mn$UNdN*T=VN
zhHIBxj|>wr^1%E@-EwKcNgqfZrQpnW=sxfGd9UV<VXHJw-5FEx)4zCAe1qCgs6YFp
z#dKeV_HaXbhF@GOq{-*7V7&FrE=(;w*goIiibwJhB)RnR2fseqD-ETm>lDSVBJ~NO
z1D;1)pr`JNKHEEikNmVY{LK$bDfwzSasAp(sJY9O=`#2dlzw%r)_#BOIfCNJjYw$s
zYjF_s0_99h$F0(!`iH2)rP5_ZH39dsq{h~!`I*JE<&uU9tCUvcR>|@0&O&GTo9sWK
zUA5wq!Wss}3?JXBg@X$hc3{%`ODJMw)oQ*!z^_^cyKRCgRjLZ#GNLHS*N1uvW{s~$
zp5$MU=iCTQv*tgx(OR|&^~+SI%&bJu)H$sL^#xyB8{xk_Shn!MV~IHBeYkg|&@lTQ
zVPGpttG6k#*kyKiw>0IR+6jJnmk(6Epq-5jmf=>xTAjmdUXOi@y5=9|ml$r_KL>A3
zabK$*ON!r=ZBtrJYM*W_6umPoxu_S!-0D?PS&DyMy!rV_$oxSZ1_5FYg@^!6fRq(8
z>HM)dT6U|_mK7F5MYi2*6}v85T-+vBevE$L`tm`&0i_EQcxV}h)qd?OU~X^v3Hg3?
z((f2SF1_N!VequKtwQH1XPq|Zpx8>Q$<BB8y;UC@9{tp_*&Im{@^BEYf+5xCySoCR
zeEC_*XLx@lS{F&I7c{upS=ijfc=+Obb&^wOf^a`pBmPMo*Bw|=Qt$+f(>*Kqb*B!2
zlrOMWsd%BFG&5Krw%T@tk%DRE7$CA4QC42u{0Pp~Cx;GeJkR{p#Em^+@)8+!pla=G
zd5kuojB|;8v=^;NRj9Pt{E%%sJ6)imFPaKr3<P>Grwv=!9*TatX45e$TwCGb>WAQ!
ztWhClx$CBr?5VFBLH?6oZwRd&5A$n31Cg9q<)reXZN4Il8ylD5kAG$}uK6BU=Q1e*
z;^mI)1R`Xj!cvt=<3y>fRhXmpI;%3j3LLDO6?UilgxB%BdoZATFq08@m^`r2l}wL2
zV^|7bpH9IIJMg9J3ruJzPK^+OK)7-P#yDKP2YNP0KErAV_5I2G1$4|CsT(R@UB?XT
z0!Hw)Iad`oUdB1sT)-Acdqt|$K!(P4go0?k&s|uG!%30;TkP-|_qNwJe#02TEI^IY
zV9#0g;|9UCY&dHzo}s90b^d?$>%A&=XC5>gLGB*Dw-DxywOyg6rAH*BK(<Kx>^7+r
z9>{cC92`@n2Inxg&feP}Fu(cndZGD`^^qXt_3QC<<5~IrG~#f+N%$BVePcH)t+rW{
zPMoF6<OE5SJE0E2Dq8uf(g>EqC*B*|Ki{oH=XR!egY37yIwc;ryy2a!r5T%;$oJw9
zG<)d{?9rE35AqC6Ll4)Tom1dV5Ju*#p57Y3kcPNHslt3asFJ76+K{D+Lg1Nj_9|a)
z6&RG)c4{zzCy*2@wYJt;DdYs>rSnk|GPa3(pfDVXTtD1=5a%S6B8l&K_sLY+<me`R
zwk4u`7%fDf8qQ#L)e`289Rb^~cBYP%F8euYzh}-ERnrDIOSlTr&rSiU8jR8M#pK8&
zS$6c-uDx5P_KoA}Zo^A-@H%qu+nq2y=F#<_qgMs3|E+|rE3nAxfcH`q{lVc&BOoe9
zY5h=8N;iaeK+eS#p!3$4Tb>&E)YYY1ndHKs-;J)mN%vzkIa9g#N62d)Bg9mxZ0&5V
zxV?()<`>9yoKl!qO=(Psp+%)v%gbXn+X79CEg~iwhj8$llrt~hOnj<__HL2FC-et%
zih2JUrcsJ6o0a{*IgXz*;Y(7c7i&)`mLrw`8|+7IA0L~jZF`X&gs3?ZLM=j&W2%{5
zo-A`VohDTxqrYpk00<xxk_G@nZ%>^u>iee`z{xjq@bmqsr^^3=KMVUuPwzH(C|_@b
zu+kZ0G(`#UUTV#7I@13D&nnI%6q#y%FS|21WxcU7aWEjJU1)l=vagVt=njaDpr<^c
zUk;4l(do|+>rO+qDaog@n<IK`H3ySDuFCvyza&N1)2Gw0HIMS8DE<x5qKe_coZjf5
z3n{u8+7lIph%X++mL}V;TYF<c-eLDE**`_8O!La$hg>e0#Y30EhW`?j+C&UFjF%O=
zOv1HKzy@irO_4%Y*$}&Sq^b04n#0F~1n(wqUx$%`;y9~s>eLEGjx*K8ta5vemR4<Y
zda{38CtD}(Xkc*``tS|%D~W|nvXvHvwQ-i*eD#+j*gPg&52mRarCxl?%0`HN)1Cbf
zP!N3UY<ek)*xEC~y)@u0?k$V1n~3s)Q;w>IN8Z1eZB3;w5arROP0-!p%4-|7GH0zE
z8<)Wv1>CS?Ie&ehhi%0iJ>@I9xNrV;>|W}M8VaLMFv38Y*81tW*<6P0$f@V0n+6~v
zo5Rt0nxdzpZ!UFjyOqsoWDePPBPQ3`rSl{e$h=9Mm>AC$XfZHd(i3)bYham_P#Zfx
z+5g~fzjlUQIA*T>^y$<0k=<APdo1wAqzU<rflS9?1VWKaoqhNhwu^y*z*^Em#ifz_
z(`wbD{k*!kw77xnr(s&MlCcuyjvD%>v#}5)DLE^NC>*)sT@pBU9l&`PLZ4K2s&h}y
z>f43?GUaNw*tCzN>%I{U`S*kXe)KBuez3jC-fg}v)YwjqXK<j($Z1B5?Wz->kR-c;
zKe!b?`UEsJdTk+R0+<*QXv6Fgu=q5kFT3i9t6BaMU1Psb2ucDag<EaKhQ*<C>Eem*
z7M+Zgi7@%FGca?c$ifZSJ|JQINEHfbOs(_8TtDrJ;kqA|NEn1!CFOwI^X&c7yr=1X
z95zH>&)_G_r}>O6q&#L{)tqe+e4gf-X(|Vqeqh^0(b=9j&41nX<E2lf1D#b)42t>4
z+0G&<<ODE$x5b<`0vai@glq7A^vw2rj^OYdYIYP0lo^}GkYk}A1gGFtuG7XymyLcg
z(`8H-`B|JOH+R<L07m}`(_($JSaeO7D&55?W0ENEH0PyWLCNT(_tdJA929EtS|c-`
zQ0OD{P_;5%SCNpvi_>b!ta&d;nSadpFW~N82D=3h8u_(oq*@I*x3V4}uMW1`f}2wK
z*1e@!X+(DRZw&#KJVT#i4UlR@$QM~$7D!@`PxbmY{ska4e9^7o$}44Y8D#;oXf#6F
z^OzY$RUH){)_;A_zuFc)Q|KKXC4h+Rq%sKk5DfwEmaZ|vj*)e(1aW<Ocmm~*BnP)H
zBsMf@P<J~|@6|5}<GU>sh;Q<`P=T~0|HQof>+%hLJgwe*R6YR#du#a;i7%$>Ci*|>
za)U_km2^kF`fNg6YH~MNG^c!#E9)MT)cBW`pyr-DF`;Eo<i-aln`+a58@0-$g{^B(
zp3Wqllk2J!K8s-64-)Hdt>Jhnl~bTGxN70i?wKJi)z~?Tn*PiS4r<C*-TW)FmTs@q
zt=o}dm*!LNG{Wl&jMRjE5cqAz2qHXGI3THwH~IsX{QO7o+=Ut2O^j~9ly@m<>V06K
zRiQVqLA|PXnlR;s!kukL)NL5AG_L!63wGqvrG&)#VQ1wn#SB!aAGjQsJf!i^jXe9F
zP3y=t(Hi@;AqQxi`Mm0eE#B!dDI?g<jz7m?zF=F)uMlhN<pC3%TCY4vNNl+vC-;!U
z(x+k)zyQ2#w>NW>C4dAuXZyz2a9J77Mk#RJz$7|Ngk`4=!5y2bQo>4$QrWEK-OuzQ
z9N*og9<Vx?cq|H}jXOmLQ!g7RtK3;iJhQ!=CZlGS@KHr4u8=1M@DubXzus@*n-t!6
zL_W9bV^I%tBTdmgRf`q=&#>d3wmAXZ{^#{xyQ4#*%wsLA6kBl=rJ?1O!G-NX{pg}4
z_Hd3ykc@b$#dE8DKDfM>3H<A2Mv-wq=%L}UR71kM4>`DXXtgHF@hY~Ddtx+@C$2D>
zdM&(to4UNDjcUrgle<znr35cZ*KXG!dI6kPBsaw~CfUJid*y6*&GWLL`Nf3gR=-M*
z&h8LIoUWD4$CT+Mb76Td#CYB10ja$`eCJ}3j2pPE%tJbIzvCC<mwP|}(d56T<)wkJ
z<oSmKGBN>jrf|MKZ_btR1X@JtxQD?6tvMU>R=x^`(tpr@T-*Q@grt#Ya~EP-)3{C>
z^EM)EMj!W^F7tWx=1nR%#!m)xT6*Qw=9nFjR!m~N&J`*Ne-~OwTpR1=xMs;GM9z4!
zup#2sY{MJs9Yx)Frz8E*@B*0fCdPZh-O6WEu0QRrMhMe;`ugQ8hla+2*I*+5K_##C
z?SXGA89#)gpxqo#<-Qj-Skm7<kbSPSlc*yqw^$=3m=AQKh|^)Z*D3FIg?rJ$1OxAE
zw}NR{2Bj9~bPo(pw675eyh?~yPILKcmlmK#Lre7?PCUB9@OO<pk#B4cq6_S2T6*k7
zY<9?q<Yij~mQ*B0YC*%Kt9JPK7f{VFh+>BAop{_&G+>Y!FD?XC?5}&N0Ko629`Z+}
zJb$fa^~t-FXhHWt2mL>#uwUNoV47ODn`wB+9k>z@k2eJ9RazV2;!UY$vMN!Q0hR3y
zRyBfgEn?%37!;o(EIw_pjziKjXOo~I^K8uk++mW2lyJADT|_dj^qVVtQAt<@E^E81
zQ0ijXT<=4iz9MNwm4Gw1&hjHV7F~LlU)|Ok<!|Z{>XCJZA(G<$BE7+y>lu<s(jnEI
zjVNT<3S;``PWq!H$E;K>C~3hN;!nyZ6={CbRe0B%c6m_ZuK#_bmK7qd+WmIpE2`g=
zPoXQ^=Bs`EBcr*UaMj*pQIpol?wK{u1~3%BUR6kyyJVvI(V~wMCi*RrzPGlM4$f+T
zPpS|{1Ks8}HN)h`tONVEX-3=~M3mLZ%`FrpsFqIIIs&acdpJI<<KxhA{AH%Q@27?o
z+&+PeXah<c)lRh5GEk@%@?eP9$iac7bik$r%6St;T?z5_O-sm;;|oNK&c@B0`jw;y
zTX1H@pLv&1tr~vprx`lMYTk9r&P*Hs1%_&=lAAcr0QuuaUAviQZ4;T#o;@2qQwrdC
zbBye-PhS%hxb&2MpNVFxdPb-EVdlYE26FfbTTg#xJ<_l0Y8m+PMLHDh((_Rcb6pFe
zD7W4AIYJb$g<!kvjpxKquD*9W1U|bnoKPV7iGK=-Nk-1#xOc6KE0w{4V7@dT4&@9h
zyVIj*nZ4PZy+x0zJFH6F7PUou`|0E_GAJ_WYh_^#<sGx_%@dSwx3zA~(Qe+N#aM2u
zIH`ReB~(Q~M!BT(?Z6p%$9v#Q55L)C=69{hIFGRs&v5!=emr$RIPh#IQYWsx@wHA@
zXXv*J(c?p18&LvXBh)d3f1LduCF6b#6K3*N^QEp7JMeMYuV53!B+tVKucheL8awH}
zOU+)Yb{Z)VDR&yyt#YUepGJK+&+N$QPQuDLC16?(CQ^G*?j}WM=P|l|f`R?_U{=^S
z;cV32F(WEQ-uWYt1jNglgk$Zdzds2Cpr5ca&|K0G-EHcF<adu<$2_a@4?!+>Sef~b
z6}GJmNP$W29Ww_eCfJ3!<1$7-ADJ$m5T9Tk&y!u~sejUTsz~!`Wl|;E-2T<o)g`FV
zHp6J0SurZ+*85cIG=s@M+xiwEaYsGfet&dEu6=<1<oSP+vY%f4oL0IWmgIYI171sY
zO^TZsb`r~2d75MCw_UEjS}-rI>aV$m%ICjU#$gUQbL7x;{O6{_2Nr<tg2!hc!w?;I
zI?YGFnE$xMRb`TAq43oF>?PvqrZkz6DnF0{f~{|>^RuKx&GTZ$x%?ka+2P|9-U`VT
z>LAr|;GC1?>q!fP-eC`Y*WiW8<fQdFf0Mk+Q#FE*p&tTWio$+}b*=y}Yxz!i-G6QG
zy>)i82d=`3v8vLgJ6j={s3G8*e-<+$ap;L9>PWL5RmeY7Pbq^lbnShRWq^r09&;#2
zl5%vWey`$%xq<h=zrOqDOZ07k8>;nV31}DoE5!L7=g6|j>h*lXVr+ntvsU#iX*lux
zIk_mNt!Ct@v*CJ|PFRi_@ydIKmr$rHbsY58NZ-jGLvG0Y@8k9JjHN&4d{&rfRE3X)
ztN&BM{AJlbT)I*=mEd&&dK3J>4u>f}YMGcQd+C^}U#C<_{>=brKl{_4`1>J&fY$+6
zTnm7TZ7159Eb(Bpvu7POMz`tw*gm5F3nKb;3-zhIk2#pB%tKXs|71+S?_bPfRg491
z%|R=l)^t6ptnXPf=<RAlP32@~X?MH%>r8ypTOWO^-Fa*^M>|gHyosf18v{7#@d-2u
zptL*8vF&aazXAAMUiz1Rm*uz>uvD+frL#T1?&R;6be{e8d+LNkHroAu3=KVSxCiq1
zGWE<d3G<&1zE{gjfqU_GwY`6*c7UhP)IfYm0c%Xg8XWL>E_mc`;PAg+3M?6^3Mi+M
zcQos{j@twSNo@PjzANS?vTbRJiDw;bLyrT9&i{Pk_n#Lp0$0g?zO(AQv04aYb(IqP
zB!otJXn$<$|Cb1R;Kbv1FK#KEJ`{W}*rPOhtY`|ZQF;9T`|H(&V*r}LvAp(2`)?oV
z8>7L!a+7pbP4faWJhJ)ze=h0GPokcMuE#B5CcvNuVA(qyHyJdiYpnl0hW{>CbVwSS
zh3&)T?@B6`1Lr<z_{6UF{N%3&zv6jR@w8mPYxmv5=?zau)JLEjpik@izn}d5?%qA%
z3V#+NkGv5{7xypAN?`X9|Fyj}bkss9NQ1QVQdMbmoZZAXeXRMuN|zGvFC71K>A`IK
zd47R8!qL1eT!h?sW>!wl!lFmd&Ga@-AZHUwAau%7&uEgB9spesPZARVb@3&4rm3(_
zTv2orH%5wx8HpU1Kf;x0=meBgOB{)2rcYi0CbKBk)l%_oPgkVz5={jh6n|~_IvErA
zeW>V!Vu(MLBtRf~8)d1(uaC9#V-U1)_`uV&3j=xT=HlQwi?Ysp$bIZ?hu6cVCx=#p
z)sXqvp;R?w>EquS;J;`36tD%}zIb+=jc_h~)lWM&OH2FUO@=ayn_<k!M`h3QXoLw5
z>LC@l5RXhf$a>_=?z6^o6Y_GB4}18Y2SoPPIqzzZeSZ?0W%ulE&90qZ$V$yrwR_z9
zXthxVWI|-ZYdu<E$!=?@umr%&#;RVFI!~804`I8mGm??KZGBG6*uv?Dmv_CZ$;$gY
zvB!_jptdD5@z8;Sh8U4SMtq7(vMH4A(nD`z1TdPtuMw8=r@lLLzO8>CV+~S%jCbwM
zp3k!$AY7K0{;ScBMC}!A!w>z9n*6a*02~iMcgIif4G!49eqCl)Cr-Pwb${|ZwN2g|
zv}hb*><7g8B~45Q1Fdh@)X6g}W=gNOra|MR#q{(~^#HD~bMAa6P>Zdp7BsTk!bH`1
zJaV>jrYRB)#*l+S(>IPGY10#U|7X&Ww3J2>9`P#Js`t~yag;E!Zgg`Wc}n#rqj9s^
zjrYH1^QGDgo;Lyl;)e~|WQ6avv9A|AmC26_puMJr%#l7^*jOlqb##rnXIdH9HGut_
z1jfemP4fA!ZPDv{u<imguzOFW^DOc*9!$YYcJs~xP~RtDRI*lPC^5@y7*T2|GkuW$
zxu@DL^&FW-7WeN+!UvoKD)}F-*=To;mtz(ou|apkt&o^t0jw(Y_~GD<nYx}4yuBf|
z%9PrVGn}v#;vGfRxuyw2Wfj)`Pa+0@!vM1wByYB@k^YiRRtV~-{Y`7wNR9XLBer+Z
z^`9lpw>&BngId9SLom_hw7`-=?W@dyPuI{NwPaQU2Mh<-gR0$ZY!<(j6I$OR7wYK~
z?RACbI|gJDzH0gEm@Q#hXoC5iKcLq|Tas(W=~DXf(&;-uFA5;LYqqV$Ui0q(k=Pqe
zz|0xTe0or!o?qNUQk8M{+R}u|CLO)qH3m%@0+p`&b|=YQ?;V`xk4>rN{A(W>;6A(V
z-5Li&Y1R(WppkWEE&8h6|0p0n$_DS8-3KIF%Nmcl_pcN@<5D%`x!A)BL(;$Kxo7rP
zxeYtw?}G#91I10GzJGU4GN#_uJ2vU1VONQHRcQvWJWLalUfYX)a(tNoEWgO^<os|l
z3BbrcigM#cG>Ix*m`nf)K25Dqv8u9vJLzUb+QU2^yJXkgQ6IR5z_3;UBNR4U`sCQ^
zXcSHVb!aMTnx+Quk80N|A)Nh_bY_?NO3J)fs=!w01hE37yW4$vh?=CLy7+_PcDJ|4
z%Df<@+BO*74g!j{E1XAJX~kuP>5~#w%+((dy}r6TL%e(Zia;PH{z5KvYC6lhgPPlr
z@P9Vt9@a~0FVoO_*$%#aDaEG!5*ZzEYVuC-fv#jZ6z2$4`loHeZw%bD(*j1!tZ#bB
zxCiZg-B(-DDXrSt6F|D;HroHV(!9;Pky-aC#NQ+4T|=>XJTE(?Q4K7(LS!Z?8JLyZ
zFGYOi#MaTL#ilp4TFG1oa86BYau{d*K$y{UyY6w%3=l9v`VizD_0gdBeNysEXbWQc
zb<cE^aT#oB5|#EyW=pQT_uAH3wQvl6VCOb}lHdQjpQnIa;uoH-b@Ka7#sk|;v4oV)
z^Y_*4khvVtR-(B?QI(<!DYxF&eNr<c*h?}@XnYDNnw%}?3ji(t6=YjH?j^E3+S35?
zs9>sIu!-nNM_~dt+IiiU99n%XDMW?a(^q7+$^&R)CL$~hUaO!9xYxri?4Cdc9;-{n
zUwO8CWx4KEN~L7EB=te|txqG&RTHkXAI^2~p^#N%1MD593$!X~Hr@$bAVyK_4~m$0
ztxt;c%e$6@0Xv|ZWY|a?WPeON{{tAvn*GkV|9+*H9)2+lGhaa|c`}&OO}Zi{sdp){
zoVCpPy7WDd*~bFvD2cBoxyD1ijj7RLAkqRrVOaW@s2M^zTk1l7#NzKVSZpB}WV@;K
z<WGhDZx7&mLbyK7M4?ALeGeNT>d*_?U_BKta^&JU)+R_*BRuwtJle&*->LawT#7VR
z)@7l5wFhRzyi$s1e)m)yv~r(u>VHYXK%kE+r4f{(A()BgT4KFhncKLQW;;-|DC<sS
z<n$g<$-YB$BvmDu_IXYEr*%jjdybmiVM>{c5M9jDAnGS60zM@1BYRPRn-QjD-dqup
zk0M#^Uy<iByAI*Qe*Pge19N%Rgy6=XE{7C^UbSv*e{br3tLCS3t=_GuuN3rXm@YK{
ziDc1x6Wnj}?s_^z(IO~Zp<6wQT9*=Dhtf5`9Fv~tV-?!qf%kkb3Tx$X)?26Pt9V<a
z?~BjXRy&*#(fZZOqedvNn?8sswLKt8gwqV;RjI-I=?2ss^mKaZ&hN0`KgVT3^X0v#
zLEATYgV{LkqoWJX9JpgsbxZvF<JqIqRWSLTsf9FkocmhYX~LYuymLwq4oUdn-hGIs
z%l|QGnQnkk^IMK|vF0Fq?&)qq7t`yecfjoKOHLYz2@yf-{hD|4KX)(uu%?dgA@-!e
z7fO4PB#M(%faMu^;{<EB890Q`9%<0MOX+pL88>OAj^A2c28W-TMEMWPYLHcE1@?~T
zpqpT^rppHzKL%mXIB`x)y`;)Eqh5p~_~TA*Z!aUY%FAa)Wu$;MscOFSXcpCh_1V_^
zKM^?3#WB{d8R@Y!kO6o<yKMZT!F|L1kQnn@!e~~bpk0+`hf?y;U!Mdjtk1ZX40UCo
zwr>>KJElzFeY*tw%(-Z3Zgclb0H7D1UWnp(vYa(f>8SDKU)WIRF5v`#6=YJgH=6l8
zLq++!yMNx9<Z9Kq7Kz1GUaPqAa>pO#eTC*TBN6$Wdy%A(eQqN!=Us;3^PpJCB~{DK
zTrsn9zlisNbQRRI%0%O@{-_=i&zQrJ$^~%Kw#sbU(>m-G2;=NT_Gi=doj$9n^`H=n
zXYC-sUb_$lQW;SydOWax=J5dkaG{YAP#R_ev4+jW(CVz1)yD=C%Zu5%l=O`ZwAZ7m
z9dJ^jJ$j@YyPPA<LE}6CqAcjRg<l<d9eT|4gQp99<fX(ci?0bD?HE0sa?DA@$~!Vr
z>oMBl`yV#u?^wGnN{s|{@9niu1|l#(+|p&N*G|51=%w0|6Mg8-G-ygHmL$7>I^D{}
z;pFFUV&<)nU*}BstL;9Ijm&vMuW|dP;5zv=2<Y0S&ZIbT*f<Tw?Ufo$O-;p>15Fi*
z;G;^B<(V=Jxi~+c-Ob!Q!_qr)u}8ng>0ZwZrpeIgf#r-`aL#VNGunrhCaKFOOgDwQ
zB6w#|9vTS0GpDYl3chFhWRMFMJ3qx9xnTK$W4C<!#>$>uEcx^`Tj!bPjt|hdw>gAO
z=vIT*A;uSJ6W22VZ^_uV3|q;)^pV&8S2M%;q0D)hufECx(s(DR1ZLUFHgElDquIw=
zvV`B^OyE_XUY)=0E*t(%RsY1FFZZwn1?}Ux_>7g|as2iANR<HF(X;8n3-%2-v9!R4
zvIfl6+%wj8wQ(}xKh|!QbMn$9&;iU?Z#n?ER@`(*aVTQ(OSxZk(*ydF8LkzJFkPK~
z`$S2?q;S(wwy-A(6c$uk?{o-P=1?BBjd!=dOSC(NB(W}wa4pTKmwtsoPX_w<m(34k
z>u5T4xYIK<bP(>npylHLTy;rqDyGQ)3ca`fY4O&QbR^#lXza4<nL6Cf+`}%5FMCQ6
zp2xOjB;I^jcarWEF$nVUS7z}Kbf`bXc(sFltowg_Qu@CFzBcxQs!rY_`({Y*%3t<;
zfKm|#>oT9OuYzS4-+c~3a3A6Nh$-%2CHGdOZDXSVZe@tSSBy19z|{`nbm@$w;>dy4
z?FuT6+fvp9Iwx7P;+*x`jt<l@yrHHD+d^F$LCO3f!t9}GsaMJrs+NwAZ}#`BsLt7~
zUMBJ1awFi!%gsX%BPU41a;%|$1;W)(9Cx1q4cy(^68Ci{y_T|HFUk&Iw`JeiD-N|t
zg2KT`pBd^9FfYoJQ&igQ1MyY^Ci-Q{Gm8l%kE?#=B+^@Y*JG}-`rv37@_5@SU)dkh
z=HG^rNAIPES7!UYsCvPpK=v*8YPm$~NS`S&u})(z8ljs${b|J3wsy%KH9$|NW4$mH
zE`)*n`Dc^Vv@vUXW+y{n;X$l7biPH8&@v(9N8H=>o=I3#55dp2S}g~xXXS+E^!hDE
zY<iWI*mTB2j<ey+B|N;SSLxhwv5|NQpgjE&u~g~}%GR+%fK50}@D*>Z!qj=?fcCD2
zAH|JY1VWuL!-b0i($^jQ&YnP^9fwJgL+6MXn?)MRf+0n}|D(tE>*O(1q+L#+zC%Ex
z&@jk$#^#m}_iygi@1y+-k9}hxEYZjVhWh>Hvqy^scVGC<+_VFaxx+nI&z)#I`B;Vi
z(R#7a8}!-Bw*OHIoq+`$1nZBQ`1LWnWZ@)vNDNibZzPEh;~el(`EFg|Kfvxfb+(dm
z$MO+*H*T6>G8V`+03W)EvQWL}G5Kf_reMn1={TOcksg2+<&O09Gn}BI_qVlmz45~y
zDph?Y;cy{ZD9^sn3Vb^h1S9DgM5%gZ6}y?o6*z@0|3JMA@Ue8oPZ7!c$mLS^=W8~{
z{x=-H45WtVF<VOxKmrh~a^brd)=&B|n%rh(b&1zoz2<vi&mN)cgdh`}N<JUZ?oLpK
zG`q+BNq-@~7M*OPyr=Bd3ym`Blvc%fQtM^FBhEWIrf&-AmF1rCA38$V<wcJ{_9+t<
zbgu1u3(vaW<aa+TAB>FD<H{27^90?iqt3=T2Qrg4<7(~eUi0LyysAKP!zIk)hCV*1
z_8fgU=C){uIi;y_3_aIGIcnV+DPTBK0O7G>Sk^S=JD!>!GbywJN}|SON=*thPRbij
zO5#+USM5J<iW;ehbGEbsy~k9Rfz(*;kYYJx{@bkz#m^Ec<CR|U{|FlXHr=smbIfif
zgUSZ2U%R^5QY~chC;p~*3~4spCT{f&XE|owH}jmS8rHUQUwtvXD8=RQ?mj_ZeOY4o
zNf+dJR~bsfXI?zfoKek--mW>Uy!>FSGj2q4YX=<AaN5fvMLa0YHzvK5p84tB3DO6d
z?n*X~oS=kjF}%%F)lPAAV>RKgMZrX%O?(x@uM+7(_3GbpH?K&V3G~T(jC%;RNKl|c
zD)~!NuDVBhPZnsYLqGv`K;$v|ko1W6o>%sZ%oS-(i?mwjSWj$Z-5vx7*77@OK&3Uw
z)^2Uvbk1teZN4zY|6{R$Fu+X$uw|3)`h|Yt9m2h=e1{+dRY5~r{%1U1(ykE=0FpUG
zVfu1nzRY$xP0l-2nK@@J!rAP|69f!ly5H)?Rr4JG7c)@w*oC0^Z?}0HhS{HpF?aG!
za+N4YpZ3)cE&>5dDKW7lSF@6?Cq)lB60)<;!<E-9Xy<0c+VCyH0Ne{f251oD>Vfya
zC{RUDbhir`Y$6}WQ^~L1h<otK7f1usN6MQb0<kI%j;uSB#^z7>3(Th}Q^ggU5<-EL
z>@w%+p~zXv%C2ib;UMP3ZLN`VPYxvXB#;5!0HZE{<<GMRr9|LQ_C}oC_<0PVWu}an
zqMHK$i#5ft&4r)2tr+|F61j=@l~P$0QeO~lmi+_0aFgfpwah%)l|^}24{~e9TKb1>
z_7*_SN^JH49rV4LE`9p&>ew(yx2>+52&Eh#bnxRRDEMxjmuNl5H1}TRcJuu=7XCW5
zkJ32P(h&CyQ#?`YQAG+>at;IrrY(kIq<Tig<WDd>6ext59CJWifE^!=JB1~4W(D`h
z@nzzZASkK6IKfPu-|g{4h|$njtB2d!^^MI;zc}aNjD#|$i4LX6vMsb`S{y*XOyoUP
zZZl2O$jEcI=;z%>DalFKE-5N3-HFqqji`q9%%D#1^uHqXucZX1Vx;@6Sa+!3qyFQs
ze=d%iJ^-7!AI%W>wKY~*0WAHxCUU72{p{qhr$)$ZqWe#0yokZ)j3zS@E9~0jLY%*k
z5eF?cV6|7RWRVD@eYBGCb2M>S!(RfZIoJ`L>I1?}g?HNRz%ij*YwZqV0ycH8`|3R<
zk561LOiD_on1-`q=xde7hWgoeoJPK0Kv}#9{u-wm%dsn(Ej-}KBUQc9^(K5}t!d4Z
z<S^dWHNu>rg!1n*6ib0n&-00mp}b_arn>y6)(l*2(U@?S6gu8uGjka;`u_qi2@-0D
z8MK3Ls=w~Ekqzi(by+BY20DGsxh&--o*PG2$cRhff`xB7rdx$<n8`k(qcu}`nXE7a
zWvb&{fboX`d#?JvU?)#CtvkJ1QY@WLg9iaTH^N6PbZkPxYR!pvSgXa(PEQl7iz!Mt
zI~31z`A;7{%z0srk@ufRm{tlKQqFL7c`Kxcxb%w4^=Vv)=#L2_?yvrAmF`u@mw~SD
z8@kQ@5CHyD5@0O8&;x7~^$-%nYsy=O0g^-S+0TtzmYo0vT>gJMsSCi7w!ijUTXgQB
znq3N7nZR&H4Q<QVB{VxHC78}LGCVOTNSW`v=XJ_nZjc@C!m#P|UVT-wpRfbAD9@22
z+3kH?FB)`5pGt`czTX6kshp<Y&Y{uZ*r3Jp%Xd<yQU`t5$AI498_7oNo%{4$vZO`M
zN~#r96JhdbpV1nZqykI0RT)I1bR@XQKl=0MlfVo(&SRFMvsIZ<WqH0Hl58sA)=geI
z$IUx};Lw%)w|Axa<`6(CPJrWC39c8Ak-Vd6E@2~MJDy!poeg;I6)G$xDS^q@8vUR2
zMNK#n=pXXm8}1Mq$j=X;V~_h=a;*pr;;%S7Iq4qpz{888vCJKPKU1jKApl75f9P;#
zIylIlFo4s|-Y&)V8i*U^yiay<BY5LRjlJ;tQC0S?K`jQ`)oU5j3T~GaF;Ey~(S7Y?
zsU>W&mnk&88+wG?7Ac_Als~T;v>hCCpSt9M{VKLpd^3cWC=Uft>4ur05jDQPsFOn|
zD+;UH)^6-)vCj7@Z0{25zwMsB2GOp$oE9sRWMY>HiCV-T=!(n9dKz&&Z}8f~L{Gig
zM7`aoo{ZZH{QqFl7ps92se`lYjr|#c`9Dw_tR9_#Cwwa`MV7A=JWuhR&YWOm$PBn&
zQdFnA$z(F{(38(}keLnvZkG}v(~`7&YH<eu#Qkur2p?Y4Il!fCBIEGa(-V4OgIV|Y
zjsZ@{emr^3?ApNAX%>lRSNPJ&`&Q{P6w#?m!SuVAhD7ddv*pwl>nsJtrS|coc-i<(
zY!lNwl)O4q9R`To1OKaZ{25uyyDy3Vj4W=6rRTW=^>kT!N1mAABCO*Dvd;HNI>=W%
zf`6Eou64T~!d1oY;?8HVmlgu6yO`^$QHS{A>s#DjkRKhFap!e;)kl?@;RK*50{TSC
zBCMHtw!XYrV-L`2W21Rq$!4|S71T6On&&#K{iCQ<?=Jr%4TBh^JAV@0i+6ukLGn6B
ze}}Bc_XcZUsARCDOiYON(|tP2O)PjIsIPQ_ANK3s3soWk#?V(`b~o5~MnhN?fyT=K
zU-yllB;tb1Zp!}u2qk@hT3l=9ZK2;@Rr|YIw)Tm|+EV>-QC5G#xc@?M^+neG#}Be}
zSEsF%DQ_yrTg4{RHt+WD`{#+O)dMJ&_tF1}Mnuy&*OqdITw^W?L|!^yGklF6JLRP$
zSyjI81>+0KwtwnI@l%fpgp#EfuFGr%p__0G5D%HXm$CoTxaD0Aa3nHztIuPLkh?g|
zu{!JAZH2kV_(Y|y%^D(q?Yxior3a%r8QETs%67n~1Lm<_Ujc2Gp<)GuAOC&(KP5Wh
zFRY!*KhnIBy~8%nkD<+zhhBiNdf(BX7IUn^E1Hem+_63_5_y3=@Qs5BfUF3=2|45T
z$pVVCnZBy@cYXli;78ghNp>%M&!&n#VFl|`tnNevwMCM9L7WR9-qiCB4i49>m%wgQ
z5t(j*Dek?B8d5jB)qZSM;73J((KN*_6xQJXw9}|hJvv&eC{v^Z2pbfw(I9g$x!1QK
z**+cg{gu|^s*Ru+siXLE(G>GX7^{F?F5m6;Wh^$rcj<`(+7-=;7_Mv$<T*Nt3b(Os
zAt#(%iMV>>pJ3qLo}Z<nJ#wb%3{{s<#Dmtu>@pJw*Qs$sPy8jX3P#er>c`i%pXqP-
zVfrKUFaD-8{fZz-Z`AIbiN(rQ!B}V;>4<z3KMEkDZ-85~@Wm`p_nu{L{CmYrP^uID
z$j1N+4{Qyw<oWDVrN)s$^R26fJ-J1XinT*0J<NLqR{vCEj~@eet;SABt?kNPe{$@F
zT1#Pr{+x&IzMBdA79;(IV$z_ymO->t`D!{YY>{Lh)+Q4*b<C~17#0-^K{DG%R05bT
zhIJOz(OxQFjXm+sp4EAPC^Mp#+>m?KK*7EK?ft<$&B`%%{u36?@$B!d#Ix@FtHK*3
z+G`619^jeA;X8w-*+Zw?a-FAZn-5>{*f`;a-y5|Uc%|o0uZ(v<vd=iXA%bl;FEGsh
zFzGqL<+iL6)mFADv)G}tPkYrcJIf~AR;04r7VGCfS>h|R%Cl_Z%CxMpxb8CT+-qLu
zlP5ntcV0&WhWQCbUe?j2V;~V^6T1n}H~cKyPHEn~KkFkLy`M#12^>PC%)diUgMfDo
zMRzoU?RhX1Ivpf`=O#qndD7mtUphw438mG<LCn&czcZat3P>hn0U5BkWzENrTY5wN
zuN^C0fMFhbs6lN#uFGeEiCx+W@{liCIoaCtcet`#Q^t=#pg_P|;!k#k9Kx144BApN
zsf(R)WZccDEpye~b?7yKakYL$><kJVsgxtc!k}gRjxB(o>bM{96kjt$$G7CavIOdN
zmjYF?t+-7|!elrsZ1Uq$X|jS-WZ^`$yIqABboqLr+1;p+!`~muM^@<oP}UNpf*Pel
zd|FW>->Q1VFgGtmJApa?WWvT{d^FyuX#M~80*Y2+focl^it=#K!>FWg7`6rv>P4Ko
zJ93!;jJD};@t%~n1_yRRfl&=W>gGOlMFEgB+n^&=umxmwZk!apcJmbq{|VsnVIXH5
zqCmSdo=W}8IY+$Pq>=htlofqg>eNEygzJY27qizxWF^~|UGYSdyHlr^t>y&K*Pnj1
z2wG*Iei7ug&EDdv4k=e(DUGNLtO`tsNxQ1R`;XPTcn~;c;f=XvVYp3$(s*1*l(K|q
zsHLc0SJxT6;w=nha1mf&M)+xRo|HU7e`~C-I(t#NJ}os>4kK;`HA(w$G+uHcQpor5
z#}wbJR->TQOY(=1opOLX_~#eYglAd#qEU0U^P2FNNh$cWcD(HlTn+iJwCqE0NYMuf
zl6rL_I<aMG4dePbd&#`K{~h$ncp5+rhal+FZ!A4&vHL89V=vT+0o7#!W&Jw(;G2iE
z`PMaFgNYS?!Oz3_sL>*FMBPKjkplgfAi*1yziZEGF}wgvc>&b;%J(IZ-E^El9j*)k
zdktUloxvR$$noU-DQ&X!L0NiEQMg=Du2UJc4DCYq?2KJUn1pB;SnQ|fGR&QsD$NaI
z)J`GiJ_HZc4672teG@k3M*9-(;4J<N@}DzOvzu!@*ac_z@VtN|4XPoNhGEj|+pX*r
zVqJM*dMBXwBEO=_IX;Nn>OC!t@bicT23oeE(E6X_z`i=TYG!|xYZE(}N>;fQsqn2*
z7M+`yV84}AyM+t8MvqFkUKP|+-?<1-v9B_F?u|q$ERBU}3^HqX2>iAm`jjeg({R_A
zpVBWnkbAUvVv`kD<qxLvZ+*k&pl{t+%(8Y@eC_p*jY%H`O6k(qg>0S*0Vm5WBbd~h
z%4a=@w(uifsm2%0x_#fJ6f*4lp;xOIIZ6w?r%SP$ncS07Sdx?E7VqT`!MxlRp418l
zaq^UJseMn+M#e71@&)IU)fZH+LX8RIf=7#q=jF~fPuTtIX7>a)UP!+g$TbI#+fxI(
z$#55x;61v6qK?F|P#V*P!#NOi2jA&H_xCI4ZcBx^?^Xja@~qor`*PK{GnIB}FWxC{
z*9%x7=}O+w?mfy6BiEttXflvsT{@lHcWJlSZPA8zOjXi+-MTWOt1{TK-+u*QQG3VZ
zu>eQeW~0x%_5}{<pOv+NKb*SnAFp_i#(t0l`P7VtJp#>&iwU03)SOyTnU8+r?Nd$i
zPG~N1MbKHKtDpm$beHE~C3%;DyLTe?v+c;qGozF6vf-<)n=Nn4O8YNYENU-m2eW(0
zTni8n*wLw(fAVkr{~mzof5*R~;z$-L9NV6xIu4@T2GFj_6Db@*8_Z#@x~dBXKKHo$
z94+SoT2cE+*2Y~~s(9GB<VH>jhU`$3jK4j&4=9qZ!Y(O;kf?iVO1%-j&7E6N%mya=
z?XLHcgMWPI2R5LM-Ww1Wo!cGl6dKvJd6*mP?>MB&J*70z745)Zu96!XQiGEwPhRa>
zsseQdOg*%lEqGLIhV$RhO1h=l3#gd)RL@MZTyM2WxMq{}!E{L5Za2++sr?$M`~@ky
z{$;83CkZJPFh$V!O;&cc4Sc;N!jvUQ1*{WlS{wJD072~u&vO}q+(lThKi~^*`lo!s
zv+l;=28qPwB~-CRAGlTg4FNN>I{m$7ziXM>&KJbO=Ly`~k@aD+0^y#xNl~)*I@)3C
zTeYm~<h_$p7S4mDor$?IaP6}^(<*Pr^&;EG-hR5jPQpMoAdQu>ya1@ZT%w5VKbk{&
zw#I)eR(=A8+V22o_<^9cJg)#39Mwwo^C(*%EO@IMe?trBzakDP8(ESb5Ou9J!&NOG
z*=2;KWt+$KxRJ5)W2E5aWLqOl?e?0|_3mB-GtjUFL4G?KgyG3EsjQF>#T8gq_9#sQ
z0tGKLC<G)KA)Vx-_8;cUIm>$X%_esYCc^%^YHihNil0^1(Lz;2bnTEsP=Mr#?uCR$
z)i}AhAY~2I{JV$acvK6=j^~#NIj0eA;@a?2Pi}`KT9bj>?yxEP%wdOZfv$K;;ZTv~
zVE!gD-e)E?Mx+4>JND03{wXVHW61YEWz=h|Zh9=yvkpKs5jOJfAmg;JV#|Oea{zCd
z{jR{Q{KB~^=wyH3LP>b<CpX(8h1=WP&Z0RvIUQ;;rASTk-o+w+!B6brpWjGLn~w!V
z$k+bVLzoYLziBCLmjms(Tgqr^{hPJ<%hm_KI^LIF%Yn001KZ8Ph2)dalBpQVaSvh8
z_||7apwsn1TA-?#m7hy^F2Growyg~1!uVqqMM3f`LHHjQ>j6X}`IqZHotx(M$^+x4
zCe@Q8@n=g100BO4T5m~x>NARv%wnR5sD|C5h0pXVfu;j#o5`2H?yGD*Iun$|>GWK+
znQaHTLoiR+K*AQIt)>>PY)PNa*7;OY`i8c=qDw~y{XguzbySqy_dYBk(y4@W2ny0l
zr+^5OQX(ZFAT1pOGa%hvBA|eDcXy|PbaylK05i<IH&1<@&-c52zxDq8u6M0jEbf7s
zbLO6X&fe$R*S<E<I`)KhT?NQ(@-+rJLM4Fn?{R#9i1yQt$Fq;6XkItJd|L2MxCPHe
zI*sOf4a*mDjzU~ZMG`?&#Ko%M-Nt15mEDfo)Zuq;!^lw_5#D$U*S%?{FiB)*9vpA@
z1MTn(een=b+G9QDEMEVlmu^9)fo58`SHo(eVCCFn`&VxBjdE&D-;tg?sMpToDm1}P
zN8<X@2|kQoEQFfpmmSAd{~Pi4U1b22+JwA9?;`Lsl)ru`1rjB?e6!h0!Yg&)On!5&
zrghI>XX0+j@uJ9+?Kso*aU7%mZ+Ii|0OP&3{@P;X3&+}$M?2Q#jkgOVCS_$U6jj)o
zz3ssMuk#LWoDbgiUXeG&wj2NX^W58tozXukb~OIqRO}oBVdcI<bAwCv?xNO<Sx1qd
z_tTV}>V*yg9fc-iNu+aTExoNpPf=@Aw>yz&6&rvPsQo3n;v*G9#!&7N1MiCkb&9NQ
zRgSXFgk4CizR!P;JSh%aZh<3nYwQ*p2;R)r{!Gv=dDlD%FcTP1IK3CuH#K&1tUxJu
z_hAr<7Z9K@+S5ys^D>wzXSQzVMVY)SN&C-Ndj}oRrSWMREH0gzFeTV-)I|WL)!xs7
zp<`#4P<4t8&g~hvXLBU&#VMC4P;|3CwUpxld@-vi))dVzMD?A?b5WA;j2i)|Y?W43
zxJDwEo?Yq%bY9{OMJb@oLa+!1yIM^YF#^6l>MCoG=rU{%5P(?k{(R#-^AAuG8VYa+
z!v@%@gAPwqp}Folhu-eoji{}mRvQU#aiPsT=4RJgC+F7AjZ`Q&UKlbO_lEV+DA-nS
zFoWNG@G^>ZT2s6v!|wUk7Lc1Q*8(LFrPWtDO>vbLZvGdGu#D}lvv~`0d&rw$)7ru1
zu;p19V9~{HXb{_R5l%qq0Lxh~`nIHKJyv?FACN5n!0mLhL(%tyjh17fVW0b2c}8JL
zDY15uiXO)KRKM<RNj<nu489JLIk~cKL(b!s9QB06Tjjn8Xl}ZIvY}XR;B+grTP|A@
zN$Eq!nrYRLK9b%!u%JHV2Y&tG3C-wtjK7zd+V9(?T&?=F9|r%GYc$Pab=2(C>*|X7
zp|f;+xcbwN=;UHy1kHRh63nTCsz6D``JpaNucz;{Zq7{It$iqvx784ubEDhzAw^u`
zsw-;<&-YoW?OtNBt5K=)e-w<QzVTpx{AB2NDMDssacI5h&H3;QQ1cPYT3RTrvBUxD
zJnvT13Y+8&8I`?aRYDAmv_OtG1Z5!zvrFZvnWQ^_=+3c@{5%BZr#f8gGL@_#;hpW!
ziN)tKOMfz1N7OU}7jJNj3ByWHJDAUjp!RUN?l6W&(vB6%3*H`X^>a6mu}uCG_p||W
zo?8@*1#Re+(UU{EPB!pgH&JEnFd!ApDv3G)eH$72U8<jNQgiAeJr87CV+0qGQWDRF
z#f>~Kq)Vf64z5N6EM`p8no+Z<P>`)W3@@#pAC_@dQ|x)(+$>?YRnT6W-0g|M?>kwf
z3oZC}b}8B(?dBveWu2~kdk9qPQ!?C2KlyqNA?JrgcDvu$#$1qHb!(-Got~}11|jrd
zMLG|}O!VN+i@I8T>e~X-sniYnF~<Gn5%tdXO}#_XG~ZIm6*(tp>(QAEtGC4kq<bGw
zu=&{ddTq_{SvZY^Y)RdI$I*A8|6o3S0a6#O-W#hQ8V+@RgAQqRzX={)p;|6_Bv;Gm
zTv0O=hNrtDA~!7dkhPrQ#g;>w4=vY)XyCMN)YQC3CjSka_o-hKY6Vjm^>sX3DP=5k
zi7@%(R1;(f9M9qvH_h6CeEAII$t94A&*Vo|1H70iyBo-M-L=fW?1AAI`UC6izMg~>
z-k^yT#Pyo_!4wYGiHmD10L<B5uZ^8_(eO;?O{skNC3bdO<{Y^2DdMSK%ZtfUBkuFP
zsW>YwjJS`^W5X8*-?<+#8MMY-2|#O&z>jMe95SvF4+R+wfJ)@3d9StVH_e{a(VlMK
zgrT?98$Cd=PZMHSPu=ON5g9U@^TXU*^er3Xzc%;Bz~+t(DWDat;w)=i0^lU)n&w0R
zm{;6O96mrMBoW5z*|jcV?F-sWKh`mV6dJheSir6GjY=aIY8Gk}K3dTl=)=MkmimqF
zxg}BYZ}kC^3B<#>DCpS6Nn5AtT+4Ou21N#&2KZX}7Var@63xx<{0gwzyoWxwCUEFf
z^aABk(EyWewfetw%fOhsRZ7p%kY{y)pW3}@C*tN;8{Tb=cG8dfcpk{c7wgsbhb9#6
zt6jr30mXq?%@m|_%4}U9LL7X$$ys9*bOE2St&*IrvSYMwItkyNzSXD6XZ^UCC+#h&
z&NeGt(qb9+^0x3FsC3vf8=T}c=j4iEZ*LuKln7a@v76A>=1xo7Hl`x_`!xAj^A;7F
z=v8;OGfz$F9xwOSyBc`fGEoXMUW}-oE3Ixi(`UfWNQtp@`h<n7X8d^@#<|M35*Ar{
zv+rvyt}jD@1c+vA`U25?K__!=jU@o~Xe5S~eb@O?x<8o+Nd0(D{h}RSE6oR=NpuVf
zTcq!AoJ`rH5ib)_tp0~Y&rpg!+>q1%6}pOd1TX#0e_dTy+{(G=-R#OVaPolT$LpvT
z*Sv?%3kE4Nn&S>q%SH0HdPv+nlSKEU+iG2#Lm!J?yIr}$52$}UcdbT(UL*`8n7U>D
zEmQ&I|Lg#u!a*aUC^BJ4$HTt)>RtV1X63LF!jX{(I@$<1-=G)(@NQ)=@{Oa~)W{+}
zSX|m-L!RSd6XX&(w_X_%Cu^~0{i_tG)y9zSV*`EK%~H04jDm9ib!bJ4J|*m$zXrOf
zRe}8RO(W=rX>u+S@gF3|M~dsNm4I?WqWCHK`sl`w7Y%#Q&u2p(4}TF5AhYuv?9EWQ
zZy;5RDz~_;u_aRmh%nbK#|8#RuD2psj{lcZ2CmSg0*yxVp*e};Ixz2?>2oWtk)u;;
zK(OXK9^iz`m68X!{q#a!Cz$?olmXE75YlZ=Ps^DmR*l4w(27~ROs@MhxrV8lzpPm+
z!fAMA$+@p|a!|bET%X|U{M;{T4E>O%y<%KugMr{I0TT=M|L|#X6QoRXg6<c{ubTfI
zSp6BgBJgM&Jg7wDBvH)WbkctoKRkop_(v^PgzZ}a0$9$+3o^qZx-$2S80o{QIc_ls
z!MPjU2N$GiGo_MdgDgm7g+Vm#-v{eEKphj0pXAhsum7;3Z{dhXa#9u()`kcVjWM&U
zulO@fA{u-r!ugUDN|gm~ndO|N4QNHxEj^gLS%Jdq*g`J)4-meg2tbsgqbMxF3gh}o
zi*s_!i1<PLnjGXW?UUQL4bTAy*DE(Rb^QN~4}7f0nCK6ExceBqD(HA8nsh$`9r-G_
zdR_kW!DAVuLep6ooN)gAsz1mT|NJ231h~h)GQ<BDi|B7(?jv<8PRxdu^7=2h<$q)@
zZ!e=i0=67NZuXzT`M+25KjQ#X@sY&o(?|bbfBAd5zOr}#>%Z#p+JE#5|L2I&n7}v+
zQk$fIW8MFIE-deWJC!N-pZ(;Y8T>O2%WZW5I$zTN`U_1w0qfsd0RK!A@F_sa1?PbK
z|7A>RU`*2Jddz?B+5ehX2sywC{ixpaFR=eV_}L%7G2j5>#Q6+8`g@-L%;2AYd{qUU
zaAm{z=>E^+C=dhVV1D5G*ING{a{-iFfUomi{PF)~oX@~G|G(M%_oDIrKbp-yU)28x
zO-1pMr_}#2+R(WFkxx~8`4XAV_16vuL=y?-2IJZ08k+_is_~V7Z_U2396$uL4!#f*
z{<rn6Gw^Ig81QX1p3!)2l;|1}+vulQg2xV~IYBuJQ{%6<?1(-8#ejQ{nati%iuUh!
zU?7%W4(oMe6|19P;k$U~$@y=`Qh!Uv*5keY*INMUMif6&;IAz<_$gGm;?LfBD*x9j
zNPt&Rb6)?Ox^?@Cuka9+K{BE*h8njg5Z_G3GB6-XR_CDct4{Zj^wtEvP2)!UwN8Vp
z#3=tgzjZk&ir>S^`3p>1A~afx&V^>zY>fx5?><=*ts-Pfhcg=;OuDygKyD5D^C<}z
zdH*CBPo}DD=Un@ThT?z!z7N1;`jLKS$pMr#_Dja_1a<?(f%p**g1}p0O(UnxTRQ2j
zB3riHsrguWCs^2Ua7h|^TAM#^)}Gm20baneoG7RVh$}Cxrb>38JzyGiXawplmVwCh
zh1dBHHq_{%$DLQ|PmSwXvV2~R1b<-J`sy^IAdNPTp?RMqp5a??US4qM=g`n#ADn1>
z_HUSI3RqIn48jUnkD8J2?fS*~na1X3@w%fa5NLdSVVwVTcz@-_z~B+kpA%>KvFo3`
zNEv*8l6`BV_spS*!MtsJ^Us5Uj!dXZ{1f+eU{vnU4&;htY3NnFrb%nLOaWbBjnq?-
zEy52?U9(#)hTdi?tI)JL`e)0$9=jonBj7w-3U0&rn5>?GE~1=mMyC&>tf^Bd;+>;=
zM6`*EQ3UZ1Bugf54iCl*7b)SNmrPxuiyVF#)CW$IIZi)t%{DfEb*>+p$SEakT?`~}
zocv<T8008zWd`VPQ9rj1GXpkmV1`q1SR@a1R}_~0qH=`KGrUCn&|78v?*7r-#g|%{
z0nsXEf0p7Q@vm+#Kr<RN(qKQ#5xPe5pITI<-tPNzi`S3cKpY8D+uW8B>xUN4g}bFj
z4`_&qaETLpRWGHCcGg4IvOk2FlzUF_w|=FS#t(Ay&a>UCp&i>>0S^>{gZR%NNlc?z
zJofFRoQE0Wpb=r#3hOz@^0#MpGu;~Ty?3l9kxawymYAI^yjh8Nn(+(oVu|s;V0Zzq
zTtZH>YFP;+5KWu{fmTebT~RGI&@x?FyHq#V+G;Y-q`K_YAFaHk&;=9k+(l9krdlxn
z@gacE@#iCW4KtqZ;ScHgARj3CfwDN3&r}NuvNHcj{N0gG@6(5{45KKd)ClcHcXv4<
z8&dcQ=mJq(x{J6vE%$T@BBGx#akWXu{4B=G$?)Z#6?O!_bJy$ZI;p63kB0N@wP3iJ
zVp243R-oc9rhpGgC%g2*yIQ&@3y$p{q#ISl%{ssQh!t_^I+6nNNZ+ICBps)vfJS~L
z^I(w3%+-_~l^_6;F^-*n?=_n%Eo3nb6h&_ylAVM0_tM8$X>k;kDJf)|T=9h@V%UHE
z!|;|j$==5<q@wWM@BX`~5%Bx5|DDx+=BH6Oou-&xjdUHz_!bIU#wF|<;tS(s%x~R<
zz7|0dW1j_ILK+}T(HWSG9u2VlJ?29quU+;2_$OIQPK;eD2Ux|l4wA}W=02=DVcc8@
z5`-lp7*cByC2prnF1bTU$Iy6AjW=PJ-J#^9UEZbY%@ua?(`lm;s4<#d8{T@p)c~Ee
z8}3}l<7WCDMz@W@LOCz&@eJ@;q}cSSVeaU?xLgr4as*l3MxtJ}W4Ggd*bTD@9!~Tf
zoai<JCMjPOmhYzJf7aI@3vE>p4f~WBW=wo=>^pN#d>j#N9}C@8*}Zq2MR0yO|4P*5
zx{5rk)#xG)C^mc@?LRqvbxN^5lv2sQvnQwVqjuZR(~Eesd-{8){0E@gtm1m>{ImNB
z65nu7=%OP4PlbY?uIA#o5Uq)t)Mcu!6Ztj?&_6Qs#qhJp_K%A}NT-wdA)`7?L8e8C
z7i;+L6{^_^j)FEwEzrQ)w}FI#K~+gF>YvrkqC2KrOg)iQB<-S-mSfo`vxCfO5^IZJ
z*5j4YI0(Xh6^b8f4VP2~sND>WN4>ss{NbFO;j$+vV%tt`a(%emghIkg!H|(OJR?8?
z6Uaz2xkvGOteEviW&8~B|8*}U<e1Jgg^C_(^1(Rv(^G8)|K`CJ)2l{#SUu0h5}a`R
zlSE(R(Nu{zc*z^2w61)_ciLQ0Uk6WtrlftGT?m5zX)HKFcMRl7M=ICTZg2U&9$idm
zqAXD)*TXyeYT`9zQ5D`_#h<)qMIWLpFBkXNt-4rF=bI2jiZu1fM|f#BINLCHE9y+4
zn;(49+6Wt!bmm$cA2lD)@C)3#@u<vIIQh`Op7Nc$ui-rPtHk5BFhl^!sBAFWQk@5z
zw2j+bgEORShL-ota_6))eFUt8o80Dpva@wYTvz*RlqhD64H?oO9YZ%)AZJQd*mZu!
zKBW8&i{$3Gg-D^|gM!%-(l3S%i)4yrWjIx^mhkSHi&Tr~DR%lC5s*gp^$T*|ru$8$
zSMcJJlA3qlQX{-fIf<+IX)?d~3-7WAyxwk~7A2yHV(raGr;<LshnerdeygP%cR$^V
zmGFc#J+q%}n~_)XSLbd%RB96<DEJ9qq4Ol|Bs4yo?@QXa*(-cdR?%*Kf?@iev`}{D
zou?4X^PX(N{nVM$^_Uh6XOGP)B23<^g_^g;9qz_wdn7IQl4)K9XTTfXE`+=2#*HNW
zua##-J?LHSjkw*HY-KKUaJ8s+3wp1`9(Q8FYEyOp?7{}D0=uHQSLm31dQ5+BWq)pH
zuP!l_-qMoA_Ab%NT=w3(VOVC?B<g|$BVILyIfVB??Zee3m{+W$?$MLg)<qq!#2_x?
z{LJ>L$q?xy7gmOiu71xx&N6<ki}GWXWyTT{gWERl_hQVPmG=s|wA;x?QEhb<K+MjZ
zVudSF(ZcQN7`;D~<6bU$IE8y1EmEU}!S8Ci(<()uL1B`dhxZ(j5veso3j;}<kB%x}
z77vOpp@7h@TSI#owSv1bagNhw-(l5?t}(Es&mZ`=ssvGz?^{q0HTsRunc`FP_Q=n>
z8PUa()E<7y$gPWNfi?;1USCkJSCV!cuJBqMGQpxpkchj+90F>;N|c>)7h5Di0bx5n
z4=|H|I;uK;r;X!H7IO-c>A)p21!zFe<t)PBRdFJq^YSI%0xfo--U^^?-EHT(PdGpQ
zBv&N#7ODQr@S~-IvD?Tb`Hbn%TaJIrO9mr)3h7NGC5U0a?giFN*Z7V~K;3<rUm|W6
zL=+;d5;ih%y(?_G71P{Zh6>j(8p5~IA-XW@vApz)FI>@*8RSu|msXL4H-3~yudv|<
zQyD<O^TcJTKduL-1I3^;lsn$?(}lZLNU1lbSEc0Z%I8Wu{F8gy_bF9v+$(GC=a~9p
zG9~aA$7ttjLmRRtvSg1<CJo<rH$$LqB8&NAm&}#=^^QWr5GbA`VfuVeXhHH#@{xzz
zs{F}pof~uTbzN?|-}RLeda|i$1HmSM|B;PS>^CmWsShMP`LS4j|4oCY!vX;rl|iHU
zs?yOj>oKW!&#WK2Z4NW<foC4aNv7w>urPX&?FLG~vn5I#e>kG6dIlf3N0sj~dz=K5
z_Ulw`3Mcs`Q2SD{s)#u6Y&d(bx5Izq`m)EeYq9HAo2R9JYU>(kYae}`J}>-3`ewF)
zyM?T;)@QwU>E<ZQsp-j6qw?AN=j7f)<vGS1yBW&r1#ii)R372|wLwKnbwQd*=Q&RT
z|A!lR#o&q~^*booo)t|ku#c$JTW`BpOKLspfD%F6Y&&{Wnxjh2V*XsaRE5BD{#!|v
z&D5}_TzzxxYmkTOKAuXe<3MGT-W1{IXhvcd5^C?&dm8J%7Kyec0&q8{=EOW4g?pDD
zhV$%c_I&%aGiJ7fI^(hM*D-oXmtWgB&KpTVytnjLlZ??7j>GZvmn2)K>2DcXyUBt=
z@Jn-jP}onwql;cl<qI3#gPiF2EcjuaYABDKQT~xSKy%$bN}}Pu;!2zCOD%2QZo=r}
zxTD^p(plq){(fGM?1$}Mj-y*SQ!I-D>5i-Fy>=Fgt_v+7#X7vJ<rO%^K^xCN)Y&E7
z6Hetd+w4j#+Nw7yp)!6sC68A!NcrNOR>52CQcvjxzh?O0h$NVNUDM~et-euD>uC#8
z#dvi{Wx44LeiZ$BTht9nN_Epge4kdY=~)pGYolG`fp%{CU(uldYad2W(!<J)w3vT+
zh3GbAwogY#ybl^Sl5OTsg05{62vTczZ&Ja=j4gMt;V<U}Wl}}3ua6dk@`zUGKT8gj
zKRK~~6Jih{{l2Q(TO;a-mN|p^qbu_Q&?!;M@Zkmr%c>8+jOy#9Z`z>QiyVbM>w2?t
z)1M)-dgvsR9xGaWd0Or~K+U)RBtf+OEg?r$WPW9>>ozUOWO=>+h<CX<3Wsu9JZ)}=
zM`m2YL+4ujI{Lyy{w7@<^7>Q&Hn7quUrx=3*WJj!KZ8tPl^~T#onRzJ#Za`<j$o6K
z+c)#h)@MYdLYxL%`)Esf#9iI#^s?Qsjvf~+mFZT>YHURodv|${(;g#9=;;Id?}Do3
z4&JSVL{GiLiHh-Y>p%aB3&b_wqq=u7SPq`oSU@E#9?3K}JXd~R`<MBCy(4}}c8Rfh
z@ZYyFg7wo;pTkVm6_IZpQDY8dJHykcB#44rE3-_$(NkHm&{kh?Erv*)u0wovPo^p;
zsD<RD`LO*gKu|KW+ktX07+Ie1k%B3U*Bu{T2`E@#ucpeI5olqP_){ELFa_X!Jbr*)
z9&&X!F>15`L<4!#NNU9Mpov1tv!ib+F=F-OM8|zA%>;hIh%#J1QUe*%-<4Y<a-NhO
zBzMWo*~X-^K32(DiB2|785<ri{UTbe<t4MA&upO`ySs4@<T}nFT!xD=dIS)z$cy}W
z0z-TkKpmYo$)2E8E(4MF0%C69malRD9@;cQeNxe_cJ#=Aby}(g@J)4IeY1-rz2xmj
zYveF8Ov|_fJs9G)*y9m@_1R{B(0+55sRLu8*kD0ZzM*<;Rj2Sxzr9fr7th5UbPf|r
zRb+Bik@HTw*pLl7?CxCMjt*PZn5k95P&l`_HSaBV`dZ@W*4Cadh1x~ZGfLZMe;#c7
z<$TZ~9iHh;)PJ011Tn$zHPt8D*;1I?h1>IO<!fy5z2g%7G0cVQLv+_0>%Zy~Ung-7
zZUM;&Bu3E2eElh(*U{R@n?0%C_Si#OQCJ>jWo-dv-3${=pybh#JE&q^`&gGl=V_J+
zh5WuL;cPv?u-KmEAlSV|@-|q*mMR>Fxvwu94~J;)L+1B7x(Bs}W;-2MGc-=tvzl4t
zLO->4>toy+Oi0un%$tbBGWrL;5qdR@aXGu>tjHvGI=28Y^J=5%-Wr{<zBe2k5hM|L
z$_F_&^!Cf=e<A9xAnG3mddke+u(tr03%l>EbG4uJ)KD|56Z?sy{tcYw)nhXLwgD~4
z*BQO{QVd$>mIR8Ti*)%#4=%d8rg=MWLK{cW=mO-mGL7|}K3qsefqw4>Z1?_(vKh*o
zwQChm8#hBC{ZwWO%TPcUKaoz_oix7mX4_X1Hev7_0woKti<bC8eIp%4i;l2SB3F0t
z<*@~y?e$bn;}*HY+7zO$t&;OTG_p0j|H2AGKMO~7&4Y1@V@lWXIB$2XzBoO3%-B=Y
zM30}*JZMWg_&d<vS5}~C+;#gWNiJfWY_u;}y;KD5+75>D?7Uxv=c=SBO0p>d#Hp4s
z;q{;m$sImYR%RR3@t$*Mx`=K-j_PJXbgw4C>cySFi@dtMNW`DrSECuQ)s|I>qhbXA
zxIPZLBfC3lK9o2iX&=4XjoXg_a|_kP4odb~>5#h-_F7LSFLpy#>s+wgORkkC@^;dm
z%vVNT1y2B8ktr;(B$#xFxw&BgM6aT89lN3XuwQ+mFfr-N1pCYF7%^QwVeo1;>18Lf
z&#-kxXT$fwVakTYaz#Vwc<$(cLNz8mtU@86*oPqsq*#b#fUMLbMFfhx=j~hVr@Vhc
zzbd|SUlO-HFwYgc8MNQ%O)!QmFQl;L-hWEGNfJnimsF(o9lypJ33Ih8ZkkjS?Y;Z1
ztr=NR(mlWT?!i|5yMz;8x)Cs7PdM>ySdaFZ5s^t|qp|8b7K2|Dg0^_mZ*rilkIt-9
zjBfgDAK^bga4^FIHkxLVvy~8giJsIYmeJ?lCVG9aUe=j&_r;IBtqwKvb>2$fP{@q|
zs`S1IiN$!QLP}`QatuAK?-C6~+5v13Ov<7hz=*H7ftjdu7I{`HDzx8Kuku4{x)$YA
z=@#_pDED>hWWN0O5_%#~)rOV_X&J7`+y(7)f}a&Bz03EqX)iaGblg!bAM0v4C&-jt
zP?dDXrZTl+D^kW8CVWaRjE?#oup}%a%jR$z0_kv~fA1qzf>Wq&Q&^h*026r&3t%_;
z!;Aq`#f5@jLBl9u`D%TfssU+~AVl3?oCKyIhtM0MkS(Vd9&i5&s3|cYAF+jR^)=px
z(Z237cUGTuc~i=+mfM3FqF^nmDD{ei(bf;M;lhsUH!&wmttgzX8>#&|#mBYaPVy&4
zHQD2|tEbX~egl@D9;jpp7mEKfm;k#)(W81T^?r!0px)3O_iz=;LV8+KDqkVz>*w}&
zIa!~5Qa+{NbYf-m64750&N3+nl-abcv|t94%~Br2)SdJoqv3ZST_?s)O=49&);rmC
z9<N~|GQ2LT!zss9Rmt1<305WWnY`=yx?N_FzBqvtJxxDKxKD71M0=p&rJMb$(2k~o
z{qlXs)I@?Gnjx944+a5SAV>3pZE{^Ml9;3l=2Mq;QE@x8D)lqPK7Y;HUn1*$L*0#d
zOtM)@6oZNmJ#YT?<^J^LUX2jvQR<Nx@d4xVnkm~)@$B7|m8gp+-rT#y5|gGEPfVWj
zQ!S4ZXQqYjE}{GDw?mOYSPJ8qv|B3Y|HjW$BWQ8~uQb?Qax*myQBzKxs*$UF`9|I2
zk-VAiZ}^A(h+}9<=2u}Sl|V7U*%`sR?(eDLu7^y`6?fn;Hz*S}2c{k$3a{RICzcBi
zHwKNv_sDiVQ&-ZP;mow)j4x`uiK7G3%jr+U6wA1^PVD5DQSu?bb5KW&80-DLC<~R9
zdgYX{RyUAMi|CF$Q3iItc6Dal^2{C+3J#S#2*{XyV8!12juRlq^Iv&R>OZ(JPizu$
zEMrg_ovPg0F)kBf6bbo$pml##?557AP?CQtexaJQI#Ds&+UAzxqy4B#K={5=9>aWT
zdG(%vBG5wRH)K3e_)D@xPMY3dzDN1i_kdfPME~$TPM9Z42|9H9zjP%o-2=(%S2hH%
zC)_srdp&n}M2Yt>c-`wMo_@-`&+eXde+ZKMp`T`xcGtxQZ6aM*11s0^v{cW9l2((U
z3Lo?FT-lqp<VJ$RyWz$(N^jh$WIg#~jAvcum-V_z>oT|u-aW}Y%>5XHc$2LLc1m76
zyn*}dtD{Ghr$G+XJBH~hId;Z#oiLG21J?9~_rQj=3Z=xCU_dobr?CK0<=H&mD&4%h
z8#NV-qPi1vyj@PF(kg{3#9FXru8NJx#|n>@GGX@762B<OGLP8nqV`?lJe2v#tyPgn
zS?UcLB|Vtov-rt-SswaCBITgk-XFY{AC-Kavky(@!MW2ZRBra;HMv_yffwuscgMkq
zVpM}u|NiEK{h5zi$npvJlIZwd#+UDfD=fwrdvYYA4~lDZv{8W@IN62cTTSo!&20`5
z-8tKCc&Z=0-n(c%bzwe|CvWM4z8C!Vz3LHe>6J$_L%7i!Ypp<UoI<TeUQi|`P7p~>
zZ`YE%mp5p&jmETQlgebt(AT%3+5Y-oL&UIKFX2L*d@+-+4v1$mUxtJ8$ZrDBp2_xk
zcXK&Dka4cI0NtT1W~}^xL&O{a<4CrCvHjF5iDA0*U8Uhj;Y1OS*yySo#jlMytwoGr
zM4d{}M`6#Wh*qOQol_UmU-ypg{c*;W`uvFam>Z4>Ds1BY(?056eEV?Bbh7WWcyg^r
zZ}pva^~ic4j^1I9qrl&kcv|DJYFWIZSCBkqB28%%m&%gsTruC}j)X|K@;BIcj+{(?
zoM_PaO+0ScDSVRFlO94zNHslWDYJ4AqVZv`ChdCuP#`1n*Stc>VebWLk#2M5y88FW
zeCDr8O>s@fItIB65ju<Bkb-ecES|1wLD$02Dcnw6v|zfU#Gp2!P8Q4;=Th5(fkMb+
zhQ4f82|FD!Jmk>Q^Q%2*DY8bdvt-6dbPwp1&cx=k6c?6=%oAM!;Sizf-rG6Q8^X=v
z>*zmDQ_Rckkjgp7E<`A<FZm=R9HtriWWEZfBc_9<zycc)QN<(|zapc>4Gphf4d8_-
zL%qZ8K{Q0}=Kk?d)s+UGaKE1;MqlsdwV&1k4?0mijmd?>CeX7Q$zmZ7M^_vttwh`;
z_k*lD1Yd6|iHy`u2`kaY4n1_Bb_sq>C=-5KXLo5&?`{l%bw+(TAYvzP)FIXL|IPd2
zSr=%{#s07qI05n*ugZH;FqMT&*v$@;Gny#5)BQFZ{L=lb+nu_=t#|K*VgfcS66kfE
zK)ACFy*kvW$3JSsLD8ClW=!X2pBL`HFjkyfw2VDN$)Bu$EwN2CR}qeWk)XKo)$VnR
z9%GtFmFvX;dJ`-{1Bd1n&o47-ll@jnj{=4ZWXK*NPqE&!W1N-gJZah+1jW;96~K>V
z5y_tzigv|`v)*ppw|+bhK(hU>^gVQLuZg#C+W3(>Y;>_n@i1iTF40#|qyNqEc(FD0
zsI?%v<cmRB9cS=HL`b^n7e)6Ph&br`Oaes9oq@6jP(9Anh>~o2N$Alr)2oQ-f2ms}
zE9k$s^Sh-Umn_>vRlinM67>?PMt1jA@LaaEa`RI}nEhs|dA@q}=W7g+`N2WJ)h(hv
zHoTX+Q)Jxya4geO{0iFon(sr`+xl-^c!=-szU#kr&&<ylJvlz><rQ^3ob2~d7YJ;^
zD69AZO{7~8Nw&&3NMpaSL-o>s+KraIgxQL+B#_ebNUzmEG&>e_mb{E~ri!)XN*{iL
ztw`WROMSwOUh72A$-0NsF&#BRu~u6o^}ad!FkneydUJF9*d@v@6L*+lSYE?8rc1x(
z(Xc3)OTSare8qT?icF^Di9_40Q?qB>lssAXJ~_;eI{$ef0=wTuQmSA6)ldPeopR+B
z`g3;(3}h9o^yTQw=q@St#R4c$HGYn6-Q0$sH^MvkS1=KsM(Xvk|KlK^UU0Ezv2&}h
zsyI7F`aanPuS0$(ith0HU9kT6_-JH2J-Iih<#>(**KsT^!R5M-@d&uJg-FCqV;eTz
z$+uCN;+eUYFWss2^#<|ghvVJxqS|1prRq6m6@yWDqvFZ>_c&aWlS3hjjodrD^um-9
z8OIUn*dBRqkQA{UWm=P7iS^1IqkdCWvBJWRSXQFyR2Ni9+U!88y=v^J1(?~5g0pCj
zw?9@~PdlSU{UDxOr{}muemaBiN7!ro^QX3*yS+$^3@7awFCmAoKpA}BVW!R(spX}n
z4Yve8%P70Tt#{wma<yHwBwwnIBnu%4_NI=hf<*ta!wjiy<b=v)k$cdePcX9Hf#yr!
z(8p-xjs}O&)30h^KAZ24%cWIc^7bKtt3%EQv@$kX6k;DoAqFcDxn+;$t>b%}6yD(x
zm2<Y*j<0bUmA`z)t}c=M!Q!Yz-VU8I(Susu{_r!jsxIpKqL><PP!o0W{vh;vZJf}z
z%pS$)8%!ixs7=IDzd@eyJ&H!Wg!EpJM^yXVv;_J0=%yFwSNLtTx51p;okch`XB5aC
z3a^YP;5Ip2^TRR3zA?m#kt|6)UJYWCXgXS{C*r4rKTfSfJnsC7<utW7@D{=AwLM;8
z4KqyhCrjU__dcmdPUq5`48?hp0xGB9mWI+xJA7!S6ckQ3M;=iN*v3#9jkq?!NVwL0
z>0q;+dyXwl9$!V2gczn&0BrEQ;PZCirURBDW`Ia_K|ZSFF~@CYi?~0vwng613tR@A
zb=06yl;;I@7`X0GkVj}IKMe<WU<74WNcG1#Wu6&dN~on>qcT|ARQ38uqZdTKCib26
z?hD)Y7ExdQ-n^L6g3GhkT~KoCr~E3bDdjH$C8av$%dPU8c?dk>*$DBU{Hvm&HLkX)
z*1q&h4|sxHcGeHvGpTP7QV67_#$v*|^Ukk2ZO9~o)aR$@?CFw@CQ@~Ryyxxn)?<Z=
zKTbxdh7{u;PH>32JZAWPUpr?RZhif`K)pMc$n8%wj#L5-7lDJKIwb_$z$*q7vc7K}
z<I5N_?o^`t;Zp(3$6R7#%}@1ep6*6X1b=^jxbs`l!zh6Jc?NPrUwR+HU8q3@5xT;i
z0!~%??mru&q-{`40jGP>5|G2`A9A+t*UVQXDKvTCUmU;ZW1y6|q%*xoNSDaSY&72Z
z@>TE1loR=$YOd>Ww(-qFb~>bjQ}81AxwUWQQjL~rZukqrurO7>S+p!)nNCbJw>kb!
zLs#HzTzY9)ylh&O_QgNXpc5ALX58GKBxqg<Klw=MV~`u;B9T(86it;A<8-g7$wv4)
zEwtB}CJ$0!XdIlb`4#}GsraPmgr<&wjN&aT<YHb}noq8Yd-JWE5r}?EG>4QM)eQYM
zpOX$6q0trZ)S2#0c61)uiNag-qlFx36$-uCv#<oYemSkH-Qw1d_PZCA$3vAn(VR13
zoXPY$@*_QcnsrT7nBB~hQ&$)1q8}bRxkFF5+ct>s!zw=XH=4`5!iHry)I#6^{S0E!
z!6>!WX${^a6Hxtv#zxFSN$sVJo%=^R4Pnkthk$WCo}9q|OAfiDSVXa8_LLYl?keRN
zuj0e}_ux7qIlxwVJ*U@5UxKp~qLz#^yk=iiHle8JQW@twmE|PxAPIFTX-pPCH!N6$
z+gdTPcKXbHF%`2x1ldwZF>B+u)shxIWp~Pc58(Swsm*d+HP;#?O!YJu2E1-0lkGMh
zy_hoW>=|6Lb4y~(B&55@yl4|-G~Sm8ra-QM?A<-VGV#xjyBxw(K1!WTwlsrgI?MK%
zj()+{emD${y!+W3>aTs!*y!osZI*%Dk-Y)2zQ`)r{vmckat|_lR_bz^tka7pVb>{o
z41nq{=S&yGBSDg^&`8kJQtz(J4B-Rs{JFd|@&?8cYTv~LRAtoCgRWW_pZmd@spIo+
zs%wGm#RBz>Z%M8kuY<-)G*rFWl|C)62?U=SDZ$%LL4NZ}tUVHug+-VU7R(v~6k)a9
zueYj{-{^R+U*{vJ|H78KZn33R&&8%c;{7^HOqGH<7<YuX;AaXku8;3gLDlc!*Ox@J
zrOptFP#hlzsMq>TxEepU@LG-O0BDeWHQZ<aFz7~rle&Mu{`H6PM9HdGfojk05z=ZK
z8bjo^k;lKetJxYuGC~RKYMgh*gm|^oVW}q_K|)5F9=bCEYlR+;L3>2c8=F1G8zdOL
z7|3TkT6U~=KN(@)Tdxt`1@rKJCdM%)qdcT@uz++ARG&-~m3>lpbjVfIN#cDnGxfQy
zzQ*{uOetLC6>GxfXpKxz)IeBiQ&u6vJJW!^x?0Jx{y_B*V%5?Js#TiL7J_mS_G~-?
z5q+jVi(gZxg|+`q;Y`2V!Ff|<<?Wx{kDW1glV0<t>8(4$(21s|{p<Mf4+p<@BHVsB
z3m<y84JwkIccx$1DIZYxQXCMwk_cF>Z7wOhS>Ta;e;q{5*lccb#DIIaCFHnGHt^cL
zq-g1=aP%Ygl#A-tq+;}wZ4G-@;w5)LnC4`#XGtlcvRl+S!&BvuN7rzRPI4s<8_GZm
zVpd9myMMh?WC0de#ZjV8J551f9K0w_bqDJx(}Ti@O^mz_IB<;-?d-A?_@j2eoo+T7
z%{a^!JLsKh7u%EM3C7G|Q83B$5;hifqY@>Ddeq{@Ar~5SZV_%sv+uiHzO+W?-eFi-
zxMm})oTImX7SmVyARTd;{8KczlZcU?(NpP?wra)C`;4IL)1lf!%vNJR_+TXlJRhf`
z(<*Z>8^}a=hDb*$aCQ|$4GhF{F><t|GBX>x%O!f%&gX){%n8hnzOf-#P2?yObdA7F
zKnjL6Ln&>={UIEm6i-wkj>YBWdD6GZ9pX;(ll`_gEJSlJpX-lhTaoWH7{8d!Fw<_e
zW@w0_d9H0<FhOc(t=3v2hMJdZ&3t~%oQ?wW1RQKWA}0W_=6=#ERNBh-=0*k2E%;Lg
z$Mntnr8Jhw)3Dw7WT`KLd;CD_4>vIxRTrBj#ZslmZ<kA61?{@p%TqnO^jOGERV>-#
zPL~vE(^yJ25OArs(o85dS4u3mSNe3qk}TlGje`BWIeA}dH{$lIYGw}0#ag&mK>PTj
z8NZ&YjOk8bGQx0b)rDs7FF!(f>qoY9PgHKN0YBo?TPz7ha3qy>?N)Q^?{mn?x>RHh
ztdVEr*~H-rqpi#fU|(HMV@Xrc`=9!5mGZhH<jt!mtk78$0DV+ge=3>8qygiyiI;z;
zZ=LR&ke`yp?<i(?q-&*vi{9lIIUyR1YJ@nbnpkmF(j*G~fAgS`(;dm-2B1jN1l?3D
zzV_iNS~%{n=kxPvcb^3%?+9QcOIx4o1`>GwmiGx#Gs9<0SWLOhXFg2ztPIdWO*=s0
zX__4@MAv6~<DC|Fu}<ilMC@WOSOiUH^w#Cvp?;%h>zIzL)8zZY#dBjE&lc5i3KpPr
zBS6TSDOoZj;GPzx1C5@N#kw|*#h#k4(Dk-dt8MVkGbhGp%aG2jCtDG2f#q*iuNlVp
zc>dHt#eJRWY}FH@Wi?(*d8Wk{cOfFa$jAbt6Dx~S&9sEoH+Ol52SJwyOPkrU<RuPH
z2~85(@k|#_;BZ1HKPzq7SdOl#eR<EGXI6m>9U=448tptBhL*9uDnwU^R)eBKf^r=L
zg2_Isr}b9Ri+Rlh2|B|ooc@XjikX<mHl!_|Nz2S<41X}??!FMBfiIL^$n7zHTuV?C
z&4i3GfZSv_7Iqz8PS?0WCFGAH22N}9a0yQ9Pww|li=t+K=qPUVCycXkHHp_4ynLJ<
zqZmvlaQQ^o4f$2km0e;<ywN^y0NcoMUut?FMlE8;{_A-mIo$fBNM0W%RZ8H^77o$H
zB5+r#Da>w=*AyZzV*1SHsrk9u+fk{D!GWVg;N}7@a6@x!c%MW2Z4g=&Jg|XgsFB~Y
z<88tzqYWQ_^j<xXeDK&B=DlTQoVEV47)?(&+ZivMVAjPTGz-2&Tw<EozI#sQ#Z>ua
zq7YjmS0y76;~pWN_ItDbcm*B@vS8zY*P4a;Ji=t#jraS6XM+Bq;xHgd09F_(ovJkX
z_ImXZ8jH1z5VdJA2ul8##E6az&G4c)RG7%`o@>_i^X7(ET(8B51Xv=s+Q97N+0DZ^
zY^Dv<CRL*Q3Ji=!LqgLJiJrm@3$(a$JN=*PSLjz1U^c(sB)=@LXlo(tq_SZ%R4oZR
zW+be#AIFsI?bvXL*&hfSCETf0y)pLOyw1kd7MU(&;<<yV!F)u)`9^BC>f?4ewNksk
z`e#AMCkb_AWDipz3%f;;K99wBvPOT5$}BMH<BbHtdt?bNH(EJL!ICA01dTFBzOM)w
z?*}?*mcBDoZu_7SvVSoUWi^J0&f9VAC3en{5G*u5NaaC7%!(CQu2%@}`hYcZR~247
z=ek={kJpkU!al>ZsQXrFB>!F=vQvqH@i(!Mkv5q;P{VwHJs_$@N#$S89*t7*#>CuH
zh|7gzuVZrrJir2&F7fA9Z?K*kzioa~;7-z?*}>5M!e!XAi~+JNY+{056qXMk&SDLV
zPC|!S;>S@W?@Q-BG<_vx3(L=BEySf5%8jdA74_>!6N4x*5zzhYmCqJ5U7mL@T`4wt
zSxoqXGACoo*0t&CSIfs8T=5js$Sd#+`<CwV7i-NzhF5PR63~)aWW+vBf9@W1e#LnT
zPF`D>%VcLX@2wgB(E4SXf5|00OzTeBGs}*SSmWn-vQJOJ%k`W62OR6j7MGyPRm|25
zT)ntYho>(A0L=6ZPj+7ZM4!j!>UBjS^;%z4Bm1L-14%5_qq!uTrnBKmJG`oJr=Si(
zS_!5Hg#nKaWGSP6;I*i=u2sPCy>*q#h#raz<jV}fr0|!jNxQ@Djy-=tkC+RLX0O^N
z>nj8&Et8RTDq<<&U*c(y#g$_S3$Q{}7}l~>+0X4g^#>1p+|r@m-&CV-{@idgUju=Q
zb4a=5pTLwFCW8b4Xp?vAZC3yafB`M&1HQ0pH%s!POLh!Bvs&gnn(M7mQ?NY_QKlAN
z55evvdQ-kr{swJzrjT2L`a7+$Xt?;V+FWu%ZVu?~_bJmo8EmgQ9f}03<wPHow$$={
z&W;$VuddX)(U(({Vj>IZs7D)Vc0>NuJn)Ojb@Z}5uE$X$YSbpfgtWh>LOAe=PA!4-
z+rY%$c|NfSPCvw{aC*})aP8h1NG*uBd`_oC7g{vt(`MBn`<<vGalW_<iW4U>5%hj6
z(<Kbc^Z8x&Ar`CwmjoYsZT|NY=(0}k9n4&k`}P}h-a0K}4K#61yJ$u9Z>iq5E<G>M
ziM9xAUKN9-v2(SOT()o3O4f+Vo4)BwXX$K?pb=jrNlz-A$JuPg-gqubK`e{*S)hY6
zMgO7l!^t=O;*dBZ$&`HgPB+5uHbc7@N^OWHaGVpX|Jaj!R<3ZZ!Z#`aN`#K#ahE4i
z$DQM>*R8^6i7pbx=BV&q&QF)Xw+>PB9T_QVH>E|qa&S6XY#@h4NVVpSdM%!tC+n(I
zvs`JQTtek5`6DzfDg>j}82u+A@=7j%-%+(<2FfLdT!?c%JVsxDCGkwrM15$Yjt+-i
zzlraQ-Tz{0cOl@^!ty51^Ai(9sf2c`t6s+V$lPu++jl6f^=<DX`?1WhsyF2gBwTk}
zQ7hdM)9_b|xScFg0Vh9X9IPV$Oq=~lmidH%qqf}YTjs(3Cu~VA^M0~6%&Z?KdDoc_
z``E=w`*<gen3^XzG={3ljoSd1qU#rhv2i$KO1^ErCcggMa(5tQ5q5^j)0B5RJ|d8d
zi>@asWKC6BimrRk7mu@>=(wltJ}+DDYd_$h$)Jj|ha7w_fSkn~M%LN)>`&L^w}mb0
zcN!+(PiM3(Rf2-BS0#FU@7NW&3<i?KUUX>Cn23*H%vFOj=G8@%s-E{6H=1tIN%y7S
zpVD1TyIS&t%^G}l%o_=DnqHMq@ks_$SQ6D1H;gkc7$1-hOqcAf(q1q|&#g-A^3Vwl
z8=YaeKFWWfA*4A~bT6`<h{aN1wS_82DZg<-+<pScTFzCvwOWoTlnfK5XK_|!(UD1V
zrcSzb9i>@K1WNk0YQW7S>!M5<Lz%tEwFlZF3jhM$xp!S-^PWbJ6MN2!uZWg&RfDL0
zpFB)(23qFy6QdG;C)GZX<NhTMo9bO{N1H`bx`J%0ACBtb<nH`Ur_7ZyLEco~A)0aM
zF%fQAv+OP^TozGAaTpMPlws7c)7IlJvY$rOW{<vOXp$b??9QM{wB2Z8amXZ3Xy`Ig
z@%{FN{gz9RTrttg*ng9iRc_r?F5Nxuz(1b7gNa;pQ$uR?^ETsWHq4ol8}r*7j}C$s
zte^g#`P{Sv1WhoP9{e5Uge}+hOfwO&ART6Z$2koOZw3S-&F6A8kI^s3L$322ol5HV
z88nxl33dD?h_dAIKg~9cK#rFnX|8T2^`Qf6g7Oz_u~0@YB_#r5C#5_}&{Kbk$C*8X
zbhE^+EA{&V*i#;->&`lRHwK^!;qzbx@Uh9IcPf@>`H9`5*i_6BInBQNU0Xp1+U14!
z@aZ^zXkmZvC}>gh6{VG)2sf?a70#u?n44!8E=rnVEQ{8kr0l?_;>dpS+nwq~z1V!A
zwSSJ9fs?;)macl!N5)a+XnhJ}`}fJ)yyi>%tWStv_S!-YaJNigZZMf?>RPh__aKe)
z4C~lYOSQYf`*0$fH-(;D*sKU**1=%-wnZ<zHVUsM90ya0GTDuy&_jt1+dsVns!G`e
zh*yT*^*k)n<tr>%A;XHHoywMWNojQtpb0l&VyNJ?)7ER2c=BO*E9o=t6j_>e;1)Qq
z+xsiMu_8HGR0angXErh~S7B^A4ZCZbti)wh6R%mMrcHNn<72_6j%Y+11ygEA{(Qxt
zhVJEt{;2yl;rj<X?1X;*(sU%6`A<5hfAXEx0%n`QB*y!_b+v5pXB&<DwPIHvxpN8W
z8Y%QYAC$5`BQUY+Bo3QPo-3s55R(;tySbMcxI-OUoFdlt_J}F6#ARpt4TzmtmVoq;
zs~Yc_5j9G>t<yc^`l5ZSh)48jS$@c7jVi+QCS>q}@^OZGzvWcXLY`?(u5xP60?kh_
zAxG|KxQ~8)g)q7#tA7)|2R*|{Bf;pjJ1UF7q?u%Y!Fw2KHz-Q8Vt|~kk=eqK5@n5N
z$X>;e>3DG_eMzVEvdzR(MZE8v-csxE9yh=vn>(c_BnsbnBaDS!4Rjp-wsp1}WRcuH
z`hnV!m9i6xFNO!i?6&w07h6MmR<7c;uRb4`y<J&5CtO(w9ZI~Zds!U(cC%|Ws>nWX
zu0ZDr6dRF@;6ZQokaTFVO6Y{#3tUC`V#AK6iL#9zm$3adm2#KYN3>4r$H>f;4$PQN
z{ca?$tX$U7_@Y#cO}cBcuNJUz+`!&c8g=fW205aYP7lmhh|ANuubYWK{=3|pqdn*)
z`maZAZfp%*(;0uQz|U;NS4M2@QG?94$Qkfh4kY;$HhC9sUBlSEWB^FyD!$yX48u$y
z<_GK7tFVb36v#TxMdMFQe{rE!R)54i#4fLS8S$DuBtMC<Jzo#ums_SQ{j*M*Go#aS
zbBI+wG9aKpYThk4=vOuVj(rjto62tICol5^1b@7%(xcGliH2!-3bt9K^2(4>TcZai
zxu2rZQ4g%E{XjNiMy7{b_PZDh9fLoyk1u%5N$)I<P?hv*%8QP`+HTUEB``$uON7Z4
zG-Ha$SG+%=3>7r9Kq~yMJbGjj*uAqHuQEj^Bbv<Nvq#>aa~r3eYL+nIlx{|1GTfc~
zFR8(j{sXu9Pn&r}<LIYLaWv?bX1>OCTGN>@1D5D5HUZ)GrFJBFZ$gGs5aT=rpqsKe
z;?cGjMJ=r$hQ)iaHiFURYbNyAN99jCmhwP7jmd+f?BA$e2*il~#pRUx2skYwPSwbl
zf36Flte+#c;{p0I5;W%#iT%qP020z|^JH!IvJ{JYwocB=?&rw|huUj~gH0zAa$Rmj
zCbB&zP8@fCQ+aH8nrKyZYFu9>^;oOI+f`vC{^%9OZ3Z4%=p$eG%x0&o_!v?nX9(M;
zSZ3rXqpDkNv^miDUIEbL`FEjvPvZ*kdWh4pVVHi=HXwGC(@B+>#QP8a#>4&?C4+_D
zL#G&zv<k}V25Z=yE{amTJJMw-g?vIkSTQlb6?MN1UXp<w5W!|gQ`v%+z=j0Ll<m7X
zm2a|YLsO!Npa0^;?hD?FB)UA*>UWN5qfDPW@(Nm+vC%HX04Sc{*_=!~N4E7d7GNVy
z^;JK^8ob%{mA;YVs`#dJ6oB2gD&3XL84(e@MPGTy>|WV~_X|^=JfP%u9T%*rN#-f*
zz5zHbia&6FxCqhk@9=mGC|Oj?+vwk8E!_9Qrxv^^A?4a~(I#V9R|^znB9bKEc4_BW
z>RRfx%8<H?j1Zw{d;Y|LN$l9++tvrkKVpafP?wF++ffq-qJ*$NBhpt4fTG-p*|D4C
zK(w!Snghe6G|f35i$H|ZvOG4rY3&lw2fgvPVw>J854goZva|}q=}j1%xZM~V!Vg*D
zGYBL`T!1vz^T9;B-tb<io}18$Zg`pXEaMp?xqw7*s^IT9yq7jGQjTGV7+VT1-NR?h
zEdAO28KUs15(5`)BB6wm&%&(+w!LUUuPmNgCn~*B>y9r4@?L}r{2l`oISQqrKEWUN
zLrLqmxDnEcsI$ua*e??tir`08Sk}uSM^cAE*U8wk1<4NwQ|ClSS}b9+FU|yBJY`kR
z<T#o$T@~+kzkFt`FCW+{G;%z)n6nB`brXu;sv4!&y5f4!VKwo^$hmgucL916#!GC$
zUM8#?v4js4Q1m8{o}m4=RJ+ONse62L1gYS#Xc|ge*J&zH<~`#^ZB&yqJ#uSgK~QNK
zbskl=UN6Mkv?-eB|I-fc)NALaRRz1*c+88Tr6v4Csl&|uFhhh)pgh>M)a$`!-L@U$
zPi`SZmV`~`W-<HvYx7nn$yxM}nO28Efg8AB!&C?IFQ@XSYl(gftSHqY)pVHh?UHf`
zZZmmQN6kCJF=NS-<awl4rxrmh)&rUppM3IYQ8oG51l#eWE>-_yY5ePQ_7?FbN5+$=
zho?1S=a(1~Ax46ap>Ivi6c;!p^QSL+*yA~VbXL8mNB@zQOt$LrZIipE0q%ZwO4?Zd
zr9HV(+#__|YC_g={OPM#xEOxGY2DH;UxQ5x$2@FM?opj;Dz9EGGvb$#;{C(1?*ZvF
zDVHnHEX>eEq-6JhvS!xh6GiqADHVyq_Dw$@@$^WfUni_Zgf{@4qg#4;kB$V;QBSKg
zG5Q@2${qj}YB2y#%ffnPb;{~e-4;H<ph^T*;Q5ZWH`DAF&#tbSbc`{7sP;WVi4X^#
zau1{efR=c~I}t15$ko&<&(A%-jkWMtLzN<NMMpb^8_b<-&EW?K=x_dkT++Zz0WFdB
z#I&+WaP6rstjT3g<;rqGB!S|4lhadNY;m(#=%!Ag!}a;j$~fJS%A0dU1Y<^bE4!u3
zK)i9GsT%6$WP;FQocm`n?JiE-Gx!Dl8>o#tAv%agsc1#i4`(pV3VqP<*hF<`rWaEq
zuoqnOlqzldnpzOWI79vWhz55NX{Jc-AcpM3L<JiNA{BilnI263yYM602Li+oH<iZ!
zRN=hu^8U&t!YGR{pku`$2F*^N&_i6s5*cHg1<Zpj$IzQNNjvt)GmA+^up}4s3_+(D
z(%TPN_P+8RQ16t=f+@Qem8VQ?b5#-sgnRHKKeYB}vlIP4?7ekRT-(+?3IqZKhhV`y
zK!OHu9D+-5f_reMan}TBBm_c`5FiA%;O-JE!QIn%;|(<Oc23T{=bU@L_x=8S^{ReV
zR9DmNrgyKs_F8j}ImVbf^QD3--w7^&E&muRY)C!JU_D~_qthA2z8bE-WI1*43bf`y
z2W7A}cD<Y)bxml8D62+|hQ306=e&ezRC~Q@1Wl-1y-;uLI8f;+)U2+Wjr&ewAue*Q
z&8o$fo#hfD*Q{?95~7s)gjq<tKXL03GwsS`JRB7ccx)$o)Ki7?gp9SgP5RQ%k(lY2
zVhCyFxdc+hqa$bnc{J?`Qy$=F(_I1Fjt&+Fjfg6y?<xir&}V(<O%VyN)w!)bdtPTx
z4SMwutVXC7TlV)8oAQ=AiZ*@l95#l(w)~M6F6o3FF3yuvG)B4#aX=RLt9{~2(<!!h
z!r7{w?9isPDh1xYQBjPpWod0^d8nhjxycm?w;QsDZMz7`|BRayo4>_2@iD(bYn1)M
zIS&kI0PBn9igPcyH10a>LcQnj3{6Tcbx4Uhyi*r;9`OAX<7h+xe43Q7H@aJ0_)$S&
z<_6Cvee`XWBU9=>e>hdEG4HNb5%z!>b@BXddjx3R5?TNQPEIy@|My*9DYY`1=|>-n
zWk%+@A16s$E!~^u>p<|SFI!R>OwoJH6Mfk8`k@w`iPz2{+nJs2nL@p<{m#VkhDD3z
zDLc{WD&E=78g{iIX8vJdWXduA*A;TS4!q8+BUCvPZyyE%YmX;2FI9s5g3<Y#yVSnm
zb#LFG<FdnM85(gbYqEWIg1(ba6>ysOQ?LhuFE+vpyGTNXQ4cv=v&xGX*KQf`JW`t8
z_|n)zh~&wQC%td+vsWYb`7yHKi{n#9RAXXUIG|d&mg9Ki9OT-=Edk@3b6ULftr9d#
z&8V~kkqN{bzdqN)!QMW{XhdLCN`?S4)B{4bcysQUz6C~CNH$BWkdl8=XSVrTl)#g>
zeAbY{4Y{3xYMH_N)BMubRv2u2J>US6@U;=}g<vc1wY4F}>GtFdp-O&XrL^p*N}B8V
zGDW?yIensIgA_X%|AaIm;;K`ujF^bS3xj(C-Q0F!gF%NYZJFs}F$`Z;WTJ?IGla&>
zH@4lmxBO*832USS7H6RMPX-x0Yy*FF(pRxJ(mMe%hJi$*;+CX+A%AX?zhgyO0{oBd
z*%;J3&VL9{*B&8bOw`JwBjd%W65w__<^qkUjEFaHf6(Y0SV=ql%L_m_AS-aq%*Su-
zVV^YlCo-v$4RzyojC?o0#%D>|5orLm;nC)5tJjTn-@Ld&lEKDcO0h794_`_wsd2^O
zUYS8b(ZZ(@F6UMFhz}PxQZ><)IgRVqIO}KtMgT?%a`7k7D%}jaIsWOgR*t!95o)6~
zPsg$=20IavL9Kb5dp*H6ovAeT^C$k^z7~W?p0|68aB(6=k*AtPX%DrNDH^nnS4~AD
zHH*93!mh;W$5Y^ZBaRfzR*hbEg`zwAz8GJ6lNJb#`904`ruif`wbL#qLEN2ni@45!
z{pllFJvt=VA%XN2Yx$v2m23o&t2pU~?iRpnIp3-4lNOqBUo(0Z$yj7;4X+gX$f;vt
zYW1L#;HKU!bCi$bdqZ`jwFqjwkX*&ILp)y3IWpe>k6OgsmD4O+HXM)LjEgk{^vJ18
zRNN(q<<fVA{-(tf;2`4zB|-Ia%L+)%?4@56{ce~-K`i<BH$1x^ecmG${>=h#4?8CL
zZb#HWKgzSWd*2S9i028?=8@|#r&T6@L7eg|(1=;*-y4b0rl>UQwOEe67K~zDC%;r`
z^#jBKqA_&@$@e&%Z)fUN;}!l0VSn$!03Rsv7a!<C%SY{xgs%^?y_lSikJcEfh2jE>
zI!N&tJmGs-8_>tGpBPt3YrQ#=V)Ul_7lZV=aL)9wncuvrEcz!N#&uK5iX$q2_)S{d
z7W(Fkpov0W*E+hl>%6akMydLbSVs>eaB_lTC7>{}@TCxMGNL2gtmB`c-rZ{;4E~mw
z+jo<x^Cl{-z25gMofiAY5z*aN>rAKnj5nSI3VM$A6=*?aPjD>|z4p;z7La!rX9<RY
zIln4LH;|2gTcHvS+;b^s9u<zo!wX<M5CsU+8Eq~vSmb2){7sG~;V#N!r(08!x>?%n
zkK(EB<kaSXWG0ILSd`~UhO)<~e$#WS4i=tb&Fa?=;<S34NtLlpQ?O+m!!3l}e`d5=
zV^ricTPr^^!qF&+(2C=dck1#RS7f?%oGob$RO9>teakKF3y}RRBmJC~8)LpY&h|Ih
zlqd!tX41<Z!th(26I*PCDW-}24@Gcca*-lGM2L*!9|br=rwBNJmuJ%ik>K;sMaBI)
z1Mdupf`o**2%_M#WXG?(EBW@h7oRg9TR3#@$WaEa=R;h?xRz-}?NSmRh{*hh6d5B!
zqPy)QiugOxKa)V(S155i(-TMMBDIc>3KJeA4t?V=YIuT0$_W`m{XWW=Dii&XOlG0)
z7MzVcT6vKAv<eQp_cE3rB_LdqzEC1IFggx{>2zWpZ*YUvME(u)nPtbCm6+=t`#UU1
zOZ+PBRtvntM)(<oV2nX1ZXA`40vfLV{zc44FFPs9vv4O2>)UWen*^i>_@7l?5riZ9
zvz}IG2nR5CCgO*^c{s?KqV4aTT!56hGgU+%1QYqP-rV(VIer^=1piwu`@uX*6qjxG
zs1uefIX2zgeG7b}T*CC0-napwmR_u{xT7er_|$1XEelk);d@c>uFQ>}9*q`@@+49a
z5qRt7+H*HRx3az+K$b->x+5<?Z16x|HlNC6!_2CCYg9B#9O;fNl5mJ3d&8D9u$Qi$
zPY8IP^A}$k=_slUT7OHDjK3o~0z(#4t^S#2gt!3VOqYIz%pZ3U0AvDXj3wA?LfS*>
zUe=VYxi69`qk)HnVr9*r%^Fg`_1xi6<e-mrL%N|NAS*}deq#SvI_~|36VkV2m*s|9
z^rt~P_sVs>X)OFyd_Fe~;|hcwMd1Y+f?0Jx*Pna!ZJ-b;=#**oLe;ykVBbqyNZR}#
zP2kEX5oK+05L?VUa>kRbl)at`3``L`4?k8N{m3kSO^&J_P69AFo3wN=t+;cLIL#R2
zdLE!LHVmf<)TJxSA2?&QyhNdsH!YalKstVM-QZsLTDQ#rvI&>qXMBvJtF6aAH%u2y
zd=Fq@{YZ8^XQXh$M&->R8Zx4#gis~*Jw0ve8!7`Kl-L0I{>AQcy=2*~3*=kCKL$$v
z$uBniJC<Pt;~#8z86*YeTpo-uV0Rs<rXV!(a+n{g9xe;7h_6RnbWg0@#H(3G;*%TS
zcfE;e6>UkE(wK(kI3YtUqPz=r@|#zPV(DN=%%X=mOdKc;?ojpBpppBY;};xbp)4yY
z36EAQfvp|Eby4k0V@tSajx`z}VkWGE(((06IqD6UiuhW15xl>I>91_@*U44@=|`Py
zh4Q}^Ll;fVjTb^hLox~-DBz_D`_+Lk2K?6pY~lGlFyz6n?HIJw;>^<;MJfCkDaj@g
zgurUymS;AGLLp?`vhp3`Vm@tjdS-e$1tHg@{v)IL9R~rAkw|EP(xJae+`oq7gX%|#
zTM;yxmut+y+JYG5vTvx7H(1Ed4DYt6x|ja_w%<>V(vhYm`*!>K-mv`h{C~@J|NSUm
z@@vSaSVj@Q1quKD!+$)9o(7KlzyI&Q#`6D;-#_l)f6to#59dn~(-Z|XAo=f-(O-E6
zEo%HhBU?~u1ASDjNlJBJ*}!CR2<2v*8sVmO3iJ5Wq*8UlhGGqkfec8X?yYTQ#NW#l
zJir3ZklV%o&o3Jyqhbm`=avYFuGAnB5o!U%P&1p|cTZ|ESuuZPK-w1<_c!kMCv+i*
zF{aJv3bI*Z`i~sw8RnA!L|eiegmG#xCThl;zrRZYfNdp0#Q%V8&EK~_ty>`cX0vkm
z_xt??K``_-MBJS04NFn7-o7B|h{8WM`H(FjRHvbOxu=985R9~WLxJ?Z;NMq@HbfqS
z_GjRiu45TD;Q~ycL}!bc+8%Uw!xDu{gSm~<+nYw$B7sX1f|RX~%2@LIebLAOSB5*d
zPg;SN9|O>^^M7$u{%dS7GLQ&&p7&Le05xwKHM+$m#eDJnbY>#|(8{eWku-P4NDB^=
zR(j~DiHq1w+w3_M5ay{Lv1v3btB~Iu*C!aNI>eGEVXe~!VjjM~l<m6z$G;At{e`6x
zQsOa$%mB6=;<f#5hJX9a*!G<L5WnxzGkxe@7o){hK1)`th1~)Ew8zrdeJtv4MjGx>
z@p7LxmFt=eXnu@6@v*q}6QJr6sD^%<t%ja@#s*k~8p_sglDz~zyh~|q^#%S4?;8~S
z55L2rPb##6bxE(XBhdk*0_Z-B67(@|2J+V&F^_5oP<fO`GZ+{6X99%~?zgMM<q?6J
zg|!EPI?0;Yrwb#<M;{M=dE#7P(rmG3f?P8Ib-@!rDE%|F`j-<=z=-SqOViN_=tq2*
zR<Xu+@=AqVUqFGqLL@#CQYPgPyDf|cmR1fPebOx<MES=I3RYi%D*Xb%kh%kY&csRM
zg{MmWYhy}yjETCYhvQ0v(rWNk;0y^J63B3EqiKBITOIVh{?6kA>4L<_zNevKnaI(o
z0x$&50Wr|gY{;aNsQ<+mY;S9sk}QUqj$)lcTJf91(!Kd0b)Z8$<sO$<Ur%qE7|V)k
zKX7#-C|S|}*sNvLfFb!5goZ)OK_H(L&g!-P<7B1gFH|S^15)Qc*=WF5a^&T*s0uuS
zW&Q~&4S+e-adhC-#QCj*$lGh?F$?y{_zhz4o5k{jzzRLVDwiCTP_@TT`FpW9&s<BQ
zQ7?w*dCYMB`0f&iz>b7I8ZVb5(T~Z9)4{WtVDj*fbNSaIJ%$<!7VGx)kDWcM1&hf9
zc^`aGKw`L--jMqs8cEJxTMKY%bc|0C8y7J_jNmd#H=89;(i44+0Aa@57PrmUjA~iR
z8u^NwhA4#N#`P|euwT*_Rr&V-Ke68=Xg`rrAz6M2)WGOhTQb58*f~bLal1UWvcdkE
zDdesbq^_DNT;q;211TIfsF1N9*8dn~yI2(i@ZAOT07k=g8rhYT4$$`&_JipXzSh<_
z*Q&N4j-A0BDUN_{u+BDmHs#4DN<APR;smJtN7Iw5<dbRM)J>C$4?yA|HIP6vAcJW&
z@pFi+w*BUu^u(5DAp1_Wf#GKVHZt0)rOiDH<oWVf`#0#sgM@_!H$}kufXVhp1$;Vy
zupCaZesFDH*LrsiUSz#I7+M!o9?2BpY=YlDm5Zkwhrg9;nI}Jxr^(3{Z5?KklG8S8
zz4xj*+w3E(<7l15a&#%cZReDHAf2hY^<bi~t1b6u27le#LJh^&8v`oI)iuis2*jR;
zI+}aSKoXPY@@O=)5>fR#WWonybc41`78*#6F)_RCQ69qA&(_y_{=Hl?qM$G{LQ;W-
z4rgSpG;1wK6Mfv#eGmntG=j6?KnII3uB+}~v!ph8S%6ww>aq1hiEa87J=$ySw}sh3
zg+whR>_$e%#2lRPrR903o7j!R=Q6qZMic&l$T)*M*Eg!a78%>wvJcOk10Zi78z{|r
zG@Aq>EIy1@&e$$ACyU*l33LH%)3<M<d4KvPi^aPrR}lZ{1hZ=UV}Ut0n}%>tJrE+|
zzwyO=e_#0QbRG{!I*Vfe)Ch16UxdQ~&@&oS`{Lms>dOa%+P1W4Z?uz09Get!cd=j8
zZ8U8X#A8}2H{T1pFV5L;lKSqEr@<EZqwO~ijnC#rg*r_QysT@~{(w={$eC)Ip#JU5
zRlzg+No`1&4FrrlJzEq2o~`jKM;v}UUxapB%E0>WD~5a`=6h31<tHCegp+T6*2@FF
zH6L}aU!JdH@qa9<q&*D_fcnuSDbOe%3E}!@kspU{Jk`^&o3ziroN)eO{9w^Y-&b^D
z6qC~mx6|h+vmILFN~Bx^t!-R$yfQ~YCh6zB7v&^+CBI6joo`khf0b@GWnVapN5TCZ
z>*ySl;Cvv%X67(;kGTMJ3^*-`u?N1~pO}x}N!i&J8-59{DgM(u2JpNBLdYZZ?r%yU
zRH;Nu_M@@2%%p#B)9rCgrl>yaaWL2({`rxlIL58!mGOk@m17P7O5FBhXp4fp{6H6J
z%I}34Iui~7>*;|Go^zZ#6sG3TxKg7g$w$xjX?Lfbd9+8odLYGUp#t4{^Bl$6$cgVo
z11)b*Y$-&oGbly848Z|^_{g~TO5ncH$#&4Mmai!?i*cJSxsg4m5!~_tftqdnxs~e7
zs&as$Q=`LjxzY1=z+$u};b!9FDgmn1SB5C$?rrY!D^={7^%*P5drjU)pR($Gf{rID
z9!!rq&VyeO%$N9AeAquazily5+gNE`JO>nfHCqRPw+DeCd-G6`*IxKDO(M+5`~Z;m
z!u0M7{aW*Uqf(%DZ4z)(3W8sM8);fSaA;*n7Q3EU%!7C2J3_NUL4>_KlZ9rld8C;<
z&TZ08c^ZI5)2~W&RX&yn1bEHHhQW(GqY#$2o%x>!Z4e`V$@$K0lD=*u&rF$~$M4Oa
z9sn9?9st_CeADWF9;8Q-NWpA71peYRY^Q2M@*CV+`z5Auy(2$*K!hBUXXnG4yvrPl
zK5X!7jsN2(x(@ZzgF?W~D;j?IQ-J!q-Yoj2c8q3HFDLZb;VcwwP2>sCjqZ4YsKsiE
zG`s(_FgXv-Ad0%ii)umS#awkF#Tdb7tEIL?4NrP<?bR6T@AUDP$ZlGiEWGci?JhDR
zUvhIpVcl}znsZ}`W8ahJ+WWG{tP%GtQ>4~w$3mk$FN~nVe!7Hhf4(+)zy&a5s?jtk
z2<Yf^>6j?gikP&Y5>kf2mV|z&gMJEnUsBLfNbN(jRo{vSnJ|b4-j$kmcFwSuw>$+b
z?f9D#od~=NN7utn&ptA&wc=4Mn2-#;=b&j3H+FICT`9!PS;{uELe4r06NODq*ZOZe
zyyv>Lx-oul!8Jgr7?yw&LJ6BW=|Ug))WB<by_=i=B20M1mi8DHZ+RB%!+D6?dU^Hc
zYOcw{E*BD2FJ`*p0u>28i6!ULYMU3tTVf(B%kdj+Jzfqn^Wi&e;kNj=U$<Ph6L>%a
zv~+eY60X79Gm_iNDqEy>4L`S5dDP$J-0~@V7K<SZu-FJSTQi$ba^9-C{N8MP&q(I+
zr(C%%<|bu|Za?1HVBzp;xB1bj4VBE74=)z|>(zcm1+MRd?>+{(UKaT{GA0CLEPQ$!
z5D8U~z<XQw_n<*yCMv<X`Q$}pKLn6nUspo#u#=JTFD=Gtedn_fCMwxa0G~BCRZPx}
zUYwmBbrDsohiK7x&anX<+uCIgtlLvHVYC^TGP2=^@4s|rgXR#(i|Cc>X3<to*wgBO
z(N8tS`cu#{LR@r+9I(HSWbb=#ka43<3dn|F*q4(6?)kb9zl;FQ3$G=5&9}J7s{H;U
zoS=oF0-Ba}<qX=^fW%I;uW&dZbDJPS$07}YF^q_?xD$)r>|wE7<1*t|2l2`BSl4%L
zqe<3xFUscMwO3$<`yJe6JIq&UXna=QKCQ%Jh{Bm5lKzZ4N(htav5WJVEO937ub+PY
z{%3*%H!Gm!(A=Cd(A+&5e8*bYb9N_s^>s*V=_D-!C6zq81qSy|8>?zLQ0e1rP0#Jp
zGNXZBTL0`z0_4t%v4~EY7pYEo1Ayt*jzSWvrj76@Md;1fZIj(r_mqSnz*48t;jaCH
zOBrLLj4I7kOe<oVd$o)!=J2OIr!dSNuwX-FN8WknP|SkV+rsx^uGVQMgW?mYHK?Ly
zS0MVP4A816PJf`eXr(VvE6bQwNhvX`nPSy_Y%<4M{U`MXj@f+>q}_XifD$JGRyhgE
zKCtn>1NXnI>wcN!fX#Pkw-~zdWCEhE@vmRRt0V@E|1=r4dR1;?upoe=U92@`d|@qK
z^%Qk1^*K(=7+Q?9`0dKmf+hBF$T%5R1e_&k2jkNyj*sHOseIO>!SS^~ky_{cD4F{7
zd+2zV*ud<`9=(1(y=e5cf}!=w<9J5M<1wiDg&PF5P!M{szq>gTQi>dd5bPP+Md2N!
z3F2CkTacsLYjw&*#f&k-|F$Sq;iN8p{Bw98ptxT(9dD-GKeOL8qJ(&#=R8yPB6D4)
z2JfX2T&1hS7xBce!iy^rXU%QJfMJ(d8ZB4k32?v!9p*6t7U4HgzFaZ;Od(}(GjkGa
zygbUn?UD8>T)cKyfpfyMtsvLthe4D0AV$II_E%01K5s~XPag1@t@5$gQxv(wWdKR!
zgJzkA(36J`*#PGk=tQpRv4Ck3E)j0UrufzCTjsBLI`XtWTycM-1GKmJAJK!ZeRP_C
z05SsoA9)0~xlfUon4M$($`;l{k^jqD$LYu4taWA}IMqG<fb0z+D!VwxhpHfwTNSC;
zea3Rt(M1``C>8A8^9o|;z({3O0E%be+f0GDFtI*4qNp5PJ;Lrn?e4-`BHhL}j(d}O
zO=~lnd1B|b^I&6?h3R!v1ih410d8rAi!jiX%Cougd+kr?nW}eInbBMh8*Q;(PArjo
zi%K8Ds(TFiNBs+<22rEN;CL{&Q|6u&S_kdcJ65C4#h%%ykBH&DOJADNQH*m{bnP!%
zIVAwOkmF-#Z`v6;#(MJYd4)ahc|&>!0S`-dHYP>%?V+CB7VXSZ3j)Lfw$nNZgm<75
z-Qx-`KL%4TU@bk}V5?Z(&)b33Q?qpwNHK47iLxz;F?+<$Vg}#CA+Ncd79U#G&{m<H
zfi;eQd|*{V%=O`IdypEm0**aOzU&#kS;b<9|FXaP1ZV`EbzOYU-Eg!BXYUTrkcvm+
z)z@xwQbBk;`5k6<-viS5^$zJZCT3-n=%#E&g}tXVs%aD6e`uS3-$yR)7-R%AQZ2gu
zRB&x_yZ}n5)$LBp)qPEyZu#gyFEU)X9)9_~%1^Qnp3jw99({Vb72fW5eStZU#x0$9
zv)`Li95_|Ewl7?M>`b8}KUMNRliAPQd@pS^5bXsN3DTL%_MNLth{sPNV%5i1tG?Z&
zRFUcFgLOf0T9%MIsjzg7?d$}roc7jCSqg0)u2g}kVfRRsWLmj?%WEKokXKA;7AZ2z
z@B-blNQ^;j-9w(AS!Iw@`EFxS(QYJ$IFFw{S^SEIFdRO@pn3S7_$lSBD!>SDDrk3Z
z<2=`6QN3Q@3k4@8-AZqz!S6}k7LFHWDx`WjrUM$yd&Amv?;Jvw!=cCq$pxK4IoX}7
zU8{Qpw>C#VHUoVlb;}Lfxc%i)9!-?yqXmXvN}8`=gY)1Du|}nevO?g4W|E85Nk~k*
z<v`MFsvz+yOv&tI*FV(>BlxBc34$k&z3F0NfSk2=QB7Gb1Frb*oE0B~?8TDz@_N=v
z90axTRhWQy^VJbIH9}ywwNhh(EGg!yx_@rZ!fkWN4EFpnbR=`UFbwMDK+zdP`~5NY
zfG@i)8_j%yM{g9Z?{W*(nHaN%@(%+M5MJ|IHQ@e8Nu-S|b-Kq{AP?VNPrvQA57yf~
zdsLTfSMvc8UGo_-eSZ3$SJiWAVO0n}L$Nrxg0ojO3*`7(Z2rS)nq`}&7^!;zi*DH=
z*bQ>L0*ROWQins|+6Gx1C0XSbimbn=;B32{+CJ}_63;vY!Mz+|6LQ*uK09yKNQ0&=
zm?_6D%Z>c};|sNmi>lk4jM6UdZmKF$9>_l`X}K;hx08_1H{o}hy3s8&F0SjJ#NW!x
zcBlU{EOQb@_B~uam21b6cR)Stn;O2r+vx6n4_Erv$Vs5&Fq?LsNDM{wLm>OZLGqK*
zGjbv_K^Ql;*cPiZGd$t`Ap(Ljpz!5>d#1*k(T0FNRljLxaul{d{85_Dg;g-7iKy(q
zmn3S%GNQ(ke<K-MCcDcje8+1uGLQi6Tu=HCDIL|dJ{KecPBhjJt_8d!zsK0WF#c_i
zLijpx|9qFuW)IiVKp+`VV2x>GjECvL4nyxd>p|lkPx*z#X^-1Jua&iNU6Y4AZt@-W
z)z6c{SQYBTXZ{T2WmoCBidq@B=ksTPWrw}mMYStiX=m%rDiMztp>eU2dXH&Valog=
z7lr~$_7?qU_*CvuVprGWF@bn;(H{Vr>DKu|Lnlj>`u7@55z*~X8`N?i@1yxupSiCO
z{E4G(hKjUHw-}4h8@b$73XLj{D~pinqPQ?C6eC{Q&3vqV^71;2x-;B(FheNUm^^eK
z*NQE~rRUhna`ud|c{qInE=={ON^DIK`R=ZVypJ;@RSxASuKg*S(_?^tsq}ZiExHXA
zaD~G|^=CMxfCY9>2v`MyqZN?IdZVAR8&%cQ4xwXL*7_|>_nc_`ieU6%W!KUX5-~&g
zSd{niRz!Y8z4?|%Wif}ZwrGOwG2YiA8gv_~X@ym!nqQ1`L`<WXIX6iaB3^YUTSn*a
zMc|8;<4l_bXpXZ;E>yg!y%hw$kor@)A4;?$`)ARJ$?Fu?-{zW7vxt{o>zj08r%dp%
zh1}s1r}w1!@LMmpEH*kuX>M}n#f{TFiU0;#^!nY?RqpM5U(M(A1{Sg<E~1|~svIM+
z!Ggp1-C=Pe%Tf`lzQ@~Uz9l{<sjoto8DTwXvsKovBf!no*;EO#lU3%LehU}}+Lq`*
zX!T)L#O!Fvh9iFEZX{&(3l(L7VD1nP$RKgZp>zeHNdLgG@H+?WW<bIIP6VMQ;pDp|
zl#tKbuw&)yxw$}K`)vM&bs|YFbx)z=>`9La7+*hMkfVvdz`t9J<PxjgZ-Kbap&SO=
zJnz3U;cwuSO)xzN96Mybz>CtFOk>tF51f{NxwT2!G*mGy4vQh~GEQ-%*T%`>GM~KX
zwN<l<uD%t+pL68B(I`;0(nUAoA5cjgA~B2kJdE){eS4-gWjVb2R`ou|g<-Qdw^ZsN
z@wb5w7IH#o_C~JLnfHr>xuh@#P`Zjxc!f+ozsrWdx&i{&f~)6|!fvWN1xRLdBuD#B
zENYovgnes5()f`05J(Iko-?>4{?JAaLOne?9d2HQFykrq{Knk)z!koK2r?3`scCwy
z&!N5MR4`h#S-jC15e*C71}q3915~1`Vf@Ob9sJIhi<lM>F?CXCCn83>#;`EVq6Rzj
zPCeq3h15Sur+0ws7$j&kR+kcjK|2x%_~4Q`i&Km88ngWcLtii;jr#lVjfu=ejX5`;
z)9q2<1z_CvDlIExf6O+5!3?=eIdyB_f&c@H5iY>I;%&iw9Mb57LnPSEjKubj8-fDS
zA{3B6LX=Ntfg3@m2!_p03oWouj5yXU?tDcMrDx(IlY9;YN!B{uvi=xd&<89=Z~&9%
z0?wqV$j+4($GzE{+zd>7#-X_QOtFCH4@iph#Pivwh(t!3$f5Nq4{HtI_p`hh;-;Fb
zc_zDk)=xC|VZN~5*xSAny~@RPh7DIanD&{nVU?38VyheQPBv=s)u!Hu$0k#f$5@29
z=2c~DP9ttv@Q)WG6(T?77b(1d?V0b1ID{oHmM5o=h0rNOM9rou0v(NLw(yieblIqT
z?H$s`>abg3%reBq=;t7&eD;a&)2GM9>x;u0Eiai0aM{I855`2~ukD%JpSPSg9)w{+
z03*pQo}k~bXE!Gbfsx|c^x27>0__MvdypZ0uUPU{Z-ZPH)?fCyA!J0srg+FSSr>;3
zatna>S5Wi?*ZTKz@B|(&-Ab9jeq{}n*s1L=)Q7}nwdV#vkar;gt#qRxR+#G541v-<
zKll<b`rZ6;?B%rdHHX%9H&VsGvNyGw3Kjs!fluqF1JX9WedsIfYTy72ohQ^|#OQdQ
z3p|7Hs#89uH4gZt_KklH6AxO|2$1a;$Dd@o56>@9O4V68gT#<}Cxf?dD^7P@AzF`h
zUxH=nmvg#I+;Ht@n=3WrOM<y|ON|S<YD^a*RYVjKwNJHL!+20eVfo8X!JuaD5vb-;
z^232-IsOeC=dNzTvL>SsKQ%sUxn3%@jIP9K6LCdPa6)jg5{VYygr&8*os9~=B_0{C
zx3aM<Hvefs+%?IrtMw|+#3R&#*Xb_sj#oC@(UExS{&WZ|%?IxQwxl1b&&p%BQO)Kt
zQtm5@+-o-pAbWt-XnB7JBi_NyuD*Nbi8x?D0mp&mW71f;Nlj?3hB_z}9e1uS%93%8
z7X(r*i+KgT`;GCh@&5up5SNELz$zbo8U3Za?(wb~W!o!b>c8%k1c^M<33}xhs-0Q)
z@%)n|bo-XUXHp*=@o)Amo4r2hl0$J*=&%R;oZHyS<NS1&{QZ(fEAT%>iO!q6WaIRR
zK6pVM-q+D(JdtDE+<ZuYve5-!O4t}+xQ)RaHT~MfPHxToC=!pHHoy=5dS}A5k!rw$
zuJO_CRE@Oy(uSJYZL~yx)L85o8$?`a279|Yq9163rk^bwByusoG0DxaaroUrn3CwC
zY5AV}lcb?JvrOfi;RP`u4S!!CCEdY$yIJY%V>}4pGK@Cyds&Vzv_nVrQYOgQnOhO7
za>D@pxvUwrx^90%9U!mPc`+5aznTdPbe8jhE)J+xxA=AzVt)+YXyYwt;q;`a8BZ4V
zW;r~n8WjuF1@a5%Bx9TwWnlO+w@!^)H|p%!?v&fH$48U>mu7UCL)`yb>1j{~G`z3T
z^{yNIjamOTvVUNK;qN2>O+)`S!!$kBLqOC{R(|6!9pUc{|KK9(rAJLv1XzP=GmKCv
zhS?ldblhythTNcXcdKQAScTkwCSXq|SI}?7>n?!deC-V0Jz#gsZmjZt!p%!Z3Q!0-
zp05;-^^mEqh?O_b=+!aN`|G1;mzF!&eHJs@_(~CZ3Q58YT^2(8*heFAaQ<HB`NsFl
z<FuTqm6mL=1&;W3x=%7oleb6GtJ~jWa|f(wi>(YSX%t76V^9Ee9Fk-ZO*5jA<eWRv
zqPb?cUeEJ8?0b~9+vFi?$>(9}d28Fb;VEnTR29F9d4N1T&^$Z7F$uJ+R-g&o;l<Vo
z!qz6zN@0`9*sCa~8+f`p6mY-3HH;XG=1WYWtFy{mIEz|bdmjxLu17U7h{14ZaSV8o
z@*HrCv<^?+P=6$1?)q7Hs8QZ}8*e<hO6lDaoI<V!I<I*6#3i+9V7A<-I=%UH@PjYS
zfN$dAj3b}DmTM7yR^h8esS&Y1jgF7rv_IVu-=e?coWAfBZ+;1o4g@7u7A~=_*pm7x
znOg71A*Qotrwz_1j?%Aa%>B&ob|&r3Z{_9P4Z82R_$}Q26^_P41mCXHy%wh%YAcOY
znBktmLX3CCh6I{PubP-VGf7^*%Sfc0<uOy6-(H0bw0$E<GqU;MHA9h9q^fBn{x*Uy
z+azgU+&Jbuf%zdsp$sdBrn8mc`W|_O`dOYvq5Du*UWV?SsMkgRivY#&V96O=&agi3
z1~wcIfFDVt@o76*RRH5yehe#Y%uYrrw&kSUC`l&fr;poatj@y0x770fh7;(idz+CF
zY4aCb0!&(tyJZAkzhj}MQWmH|3uCN5VkeqGrUV(mAO{%~X9KAb@?9n}$mxXic^mRD
zPL}F(J^>>w$u=ifD5Y|II}r8z<<OeMZsV1yCp0mQBL-;D6N%U!s@nC40(GS%9;-q1
zZ048iX2<(0ISh7dt0j4A*~T$<`k8GipZM@aSiDgHw~gdjaGtC;GL@ACu{t(_YIg$p
z9^2U+6*DPx<gI%-j_0`womKLAl2`i~>av(c8g<TMmMH0G2E!M0Rru#*cVa8x>{bSM
z&YiahK>`bioEA1fbaE6HTs|*xU2kptA0=lb%<e}afczInaYexR%9>&bZEYc^0;p`s
z_t=>!n673vd32BxNWGUBqs&X{dUSHU#_xDeveeG>nzXtYg`+DMR2DOcY;ujW7p?HV
zX`C2$o4kX1Y4xI8K$<M0A2cQHD!vltBHj^sjoHIj-zT5KmeLIc5g#Ca*=hYM$M5|r
zNsAZ?{VfLh<8=^If{+Pe2ulE$@iljYRR$3hH)1h`v;3DU_|FlO5(^&z+wZc!VDDsS
zDbVv{qi}~1baCutB;vs+CQqE6OQHLJTT$BM=zs<ey)HlX{|<ox&({c$!v}=?uG1+m
zy3}eH@t2k2KPgBNruVg$tV*NzFC)%(^Gf9_gaKbRuWHTb{x_`2%dOk1@)=m>A%DR*
z^0cM^l5p>NPyAmY>L2hyOpFAet7Dw;zW%=Yu&5q!;7<RwuhZG)u~|^n(EBENeusrM
z4@vWxV=E;E*_6Cf!1en}+InFIzh?b@twE+9lvOj2YjPn;qJ%&`k-p(%B)?(5R!;;E
z{ZE!oNVp78Zk6&)Zzlb#1pW)44@n>lW)WXoyQ}&6nsP2s8|azBtZT6~0r4-i<yoYk
zeZchHB8u$^WIB$)>(U~CdLp}crxTA7RFdSuoCwrHWaPEB#9A#7I-z)Rs7?<<P%cjf
zriR5Azl<M&{``*r;bQyOH^(<XrETx~j6q{Cy-zI=ldj*N!rRo8{%4zq-^U%LNxb3_
z?6E*<_W0+C|Es?6yUGAOMnis;i~0(B?f-XXlKd%f9Lp8Z)$Knn@qhff#sF;YM@NN6
z>3{MZf76)%>yc0tI4<Uvm=@`OJ;9I*dEg58``a~J|L@2ByD<J=pY)Ftq!l9sZU~y}
zN9FJTHoyM+J+$A60NVAI2{OvRN9uok^8d5p3%RQXlOX938&Qfs!J_}MxVVTz_#@Oq
zz&!YEG_CyOIKt3RDB01^2sM6CDT0;2;Aw$7{htQ`&dc)w1{Hg^qwtm%cblIhEr4mO
zLF!$G&~Loye;B?_VEAZO|Bu7xO$3llf{6alm!p&DFz|ZAHatoKi4k85C6Qh>Nq)z)
zMOHs3RQM6(@2J3kU8Ye?Z@;CKU$ol{%x3Q+DdUeCj&YJ3@Z;Bvi7oXKy<DsB{@<qp
z?Zz)QjL&CE?oguOt-C#PvC*1-Sblh5{^dX5{(np|U_5ALFvofXWnA632wfNt_b%Oi
zj!z3s{{dh8=ZS_SFamTNMq9Thgf7xZOV?^yVw#Em{?eEA+k<xh+vGot1}KaWu1X?I
z)NBCli@qMs*4)4oQ;Gi#vHp*b?PS;f=<+MXMBZ0>?}2WW%&P%C3qHdC`k0UkrIpT2
zd-_52wY7ZqBp%tS3cs9Yo`3!Mk5vT20$35ED$zT?#gL)mSD};WGp*$R<$W(9!-(`E
z2%Z(@w)#Vx>7XsRwOkxvcdgC0`~7{T!C(W{0ZFF5b$Qw}RN7-Xjk}ZM*AIshy#+Cw
zXHkE0=rO}#;tAoHYYGu=OeV9#dm=6}{I}IABS!U%#LMhVJ#RMf65U}-e(L<ri%fvp
zKjEvbMr41QnIk?Uo=KyBW`^!w*>XKocAfLn7rD<d?NbMKf!1Caxu~>TusJ_@pIwv5
z&_J>lWn+@RZD>fC%>=-=Ya;+sGjd~0T0^zB@`+3ujIXa(&&W?&QJ5>N7u(nbdfXM>
za<zuFi-T`9gnak3+V3^*5!~#*I-X@21|86_TVsdx;~9MQPI<_l!8x-m8;w{G%FgLq
zXu7NT>hMy|X{nK6-fM5tGwU{*RG`*>=2NRQIK1@2V#hG5apmBvosJw~H-vP{?31^j
zy%%)EWTMY$vDM%;=~MhISMOijdp84;Sr9Z!n!_2S@ackMSO1^t0mhXqU~0e}W~5Nc
z!`n4C2WWakj@gAs)ud3&LfW$eqBeeP79V?keWmd!&F{s<#QPJspL>_IM#rnUeVfZq
zn?PdcBa41K)(e&~M<XK0)S9#=ssmYkxzgf{-;L{B-wdTU%@mC$K26>j&NTWwl0mbD
zsP{gq;)`cpc@#F-6}#Hp;(Ag`6GD^x^d{<>%v9=-Lcm$SYdGVi?ukjw_c^fk7yXvd
zjgj}0+R5{c-*!ZWe6QbT=)gL=qVSkhcHROxEJ5APtON={2$Bx9QB$DivK=O0B?+f-
zhQQBSaG`>_jjOoY$WW4=%AwSPTr_p~PJ`F_2!*>^v-&^gs`uirt9iiEej?|Mf`31o
zRWk1Ag5kOt-wbKUpx%(fq=DJFg?LhOMQ(A7z=99md!4~@^ztN?vL2*PU(T(JKQqi;
z<*fRKh-LI~`K9RjPo!@EZ4avgPnc1M{;knU)CI_O&>=w0IvCVYrBH0`dG<P(3??(3
zO#XS;VIPhH@peArCK0`w|Gee)W)inhgL&F<x@;MLf%A!Nl1vgvpU|k)_pC7k6rhv(
zB8q=zn-@x1rmZ^{D6!J^DIO<wmCP_I$T}PUusqwg>*Rx=Rs8JaJQiuW`U;q~Gc#J=
zbTeIL3j9p^>X~zFdz(_vT-_Ul0wIffa-cq=LZ|oGY05SY!BOII);~67T0K<%$#7Gg
zt}x)`P-&#2ixzqxR&FlA+(SJ=?-R@!(Fr-cUG})$Ak)Q%%w59%M{!G0v|=6-;Nszi
z6$TQa>B#Hb5NWwhsH-OBeGZMs<yveeit*WkPq)YO-{CEBmW$(NPK=yq`fb7osy#h(
zuO{;)S`9~x77j=Uvklma+}rW3x2Eq@CO3;x2K9`O&jI$AE(g2b3nb~i>j)*C?J5>=
z^Ii;Z?J$=0W+Kr>mPL<j0L8l`fG@uwVj4(hI`_u*Vho9vM<R0krsFhL7Z2lTMVNJX
zNi8h<%{0CRDb9>0v%gU@U(`Lrnim%NGXeWEmcnlZ5Ia1R(PZ$+eWGA=2g?uW#}xkT
zK*)r_gS?*N+<fG`)L04<31}9EEi~A~A!?6Zz~?~Y0ITK|x{6eGBW3F$<HzrL^0e1{
zRCo0>FPt_6xw$tCo)wmMe|D&nig+JGCGr`!bp3|MW2(guZnY3ju0b25Dk+G~$+?_@
z0$iYHU6`m_Bysk;hA3e`&*Q_fsA+$9zH^ZsUto5WAV~VNm6BKjc*A#|6>7+*#?!on
z3*L;Pxvd-q5du!xd2-XWIxVaV>X_g*8a`v#yMVecf%iW@<uVAjsWO0biUnAt&6a<#
zBJ`d_#c#A3Z<{}}JwiMy!MGlwJZ5uU&TPB=kwS5gZ*?2Q5C7-P`1H##<a?GNOkjv`
z%*+>9=x3o_SXcol7$W1vetuCzXIEbJU9GbX0f_k11A$J!?w)YZdL(V6hypSHO!CR?
z(|NCNXvE{xS*Of7XUEzUPh4rT_AgI<z8y9I>u2o;6>>@4Lj=;*!4`;>QCIviz+nr>
zXLi~(BLtjIl*CmmA(pxp67o&CS9rI-3f2XFa2ojxJ?_o4<DWnmA(rPYxVh_$hbB#4
znj=PsIC2jKyq+9!D&v6;?~H4E8*w!+_95abFIYJEMb?>>1!SXqi+`vfx1}Bw`*C@l
zFUSJsvQ5RzV$B)vsOW|W^*UA*9K`wUCLT|r#vZJULaJ>z^-68W^YZGQzng}jE`d-E
zmzJ$gmZIumBVNU*&7>SA?$bknQ|v$DX;NU3X$tOSBlI>xC3|z{nF#eUAr1j7$1tTd
z$o;NP3t8>dfTkWB{T)hTnG!vRfbq8U&`vd!Q?dB7r`18I6&o^4b+;M+Lbn_t_Y*L*
zn5@w`I$6a;WDa7Qnr|}LNDP>xm5O~HJNf42yM~nGcB{_4zJJr?@v2l5p<)6yz&Jmz
z5c0IwNz`C^(6C#!Te;IWYdDz$Y4U}xXTk}13JuEXMkamlikGB~ralawQPe#U>$RKy
z!kDjgl}!r=dgvTq99cBvGb_fLSw+5o*3Rp#R7xF87dx6O2k|6NrqgQ)L4bUAO0v=v
zl7%C>ftzq7a)5e+%=>twuv>RPA(=((6JaL$H57j&!lJT>F|fJb&hxocfNBX>M>p<}
zQcHDzgX!LF82YhRCOV1qVpPDHoaxm^PG+T%^=q~U?gIgZPvK4$<xNe@3td5Mw%n~z
z7~pVBHr{QkvOb?TtxzqF*j)jq*L96;6NO1iLuuR@P21d+bFv@OFJd-7;mvVnZKG4K
z5K?x@jJxwOKY5p~9JWd&)q20=IEI3yHc$r$0VW-9Z`xrr3+&7VhhjYt8r`HqZNUSw
z2H`p%^rI}(2s~ya(MzMqTS0*`x(HvEqj<Dl7s@39gkk}dGs|F^Y5{MFqWRv%LA!Eh
z@yQG##OUiQv1Jk6fP0$gt~9GY3;O0~#&0enk6ou>o)acQM~3mvfy|^3P->H|$(Ln;
zj;04s?!QT-O=29_VA_5l_b%KvKsO4_+43K{6#VZPsMnlx(a*zy;$wMY`w5g&t7>uD
z8yUoh8BUKJow?Y0PW*<B*7IO$S(rga7ZIq$)JqZOqB|H-A1IG`aYrR$+o*4V;fAx*
z%t@|=?a$aAK_%v9*n1<^rFOSnP4coYXDm+=XW&ccl_^><!7f2yfqel#hpYgGyQa+u
zsb7*YMrx+GeLk92R7Y54vR<jn7eISWjhuC?TfUUwpbrquWKy{F;>h1Q4Al?lbeg&Y
z7Cuyj{-#3JHt!%W`6f1xh-_XCRTizK6Mob!zv)XK0`~e(w8{zGvD_Bt^h$@%&b)Ot
zzojI=*;Dwm18afZjIiXYLhY@o#>eM14|a?c81A)E<U%@7yW#iz^!kFzb%N5ud-lcb
z*z|b{>gJTqwhk+GLf1ycbc)q^@_AK!u_}EBV|}VDrURpV_wwYkw9*C2@*^z$)_at6
zDa*F!jWexIM}zuj-CNli#%PMI=e|0uK)rdVJ~0(r<-hbf?&c?m;$L8qGSx0qpYzZj
z_%QnI@cUi$IpTDS7IUqy`dr0qvG?he<)c>Ju(q2Y&(Lh=4N8x>r>BeOZK~Yi$SsXM
zw<U+MiP@i|8j_qwD=jZy0(`?3zIWTvn4hn9zzPQ$`YI_ko`Q8R1HO9n2&XLgm3R4}
z=nOxaOfnR@em$2Kt)D9ezR72%+Mj+yw37a&$e&uUt_k!0u9TKp?t^c<_6O#EDI_O;
zfxODotKh9*F(b>#IhAAh557iRI{$@D9&&Nl$g`WlY=I~srt34?Yu}q`f$y6Q8^$Fv
zy87FGuV}myLzJMpO+q~<9J6j^;vEj(%ReYQu?Sr`-z42RQHEp|%2={4&??H~jJp0k
zsiDUQ;&g&7vlE>zG309tn$_OSPsZUlSOEPg7#(#l3qkDx-bzyUS!XDG0y-);ZYr~A
zXV}0sKWieq819H}p^@vEpt<=*lf)*=b9vg;`@_g5k3w?nl-D35*QEpGTc^<Vh{tq`
zj4BdveLuqG52%!-Hwf=>C4v6PSIGzECC@Zn8T1N`7hA;A36)o`Uc=r!eqy6_k?w=_
zZr4v@373&RQhlc|K;+fiW0bbi7BF{K(0$h`y;Xl5(&$w1!O}R;t@ZM(x6Kn#M)(Q7
z?G;WxXiikDS)l9`NAv?N+)N6C3GAh%f->9na2ByTOLyeLcqSg*J`xGW?JTWcrMhxI
zSk8IGPC3p0oXJ(}bMt`_IVxf`_m)MpsU~@5=Y}it?j4z5na)r>A^dxM%%_~LpcF|(
z%f9N`c@XjE!SIx->0H2aCAu5`%s`;LV40~TE!fn7xrd8xRZ)GpwF`RFE*OEe(aYTJ
zaJ$TY2L9nA;bQo4?ogD~n6b^hT4JHkGLpRuA!e+CEgXg9ejtJ<X9~@mAGxpN4R7?g
zP;c*dx@hq@Z6i8oD~`2~d=>N}0?od>6XH|#pj~(-TS_wJUN^?Iw~lut5PiA}q7);?
zQ=YYy(W9LRe?_;3cIFoYeKEce<FHkiNgL@c>tBvr2nBJa9b8c_Ff!vG82iy}&ItvQ
ziVD6p`p3Kw{53DG(N^}OL{adqK}S<qe!3&<^vHLZw9Rsnr?%HHOM8_JGO2`IIv++E
z1#dO!6?nJH$eQd4K!R~yY4W<2FMWHI^hI%ulGm#Ww<**$LH)e<%-~oS67r)~g(yVc
zM5Sjk3cZ%ym8Oq&0VBMA<x%({%B7pnGi(_dUdc-Eotb1-C^L?|Vk>BtDn4pjNa`)E
zOtjG}AmnNlqo(UHd~f8dlrES!arH52a=+vOof|Zt78B10x~z?#5!as0r*yxt5VUa5
zB4x0ReY$KhVG)75Lrf}2GVRPvk5AH<Lhedqroxk__VmRkZzV<IpQ!hp&uLN;?6B8~
zwUbFO^1~_=o6af6RPcQWstTfy)A=Lu)eg(}BjLE0Bra6wxVJ0?BN>|1^M1<i(ar_R
z*-+Yp^+sMB<>G_wB8O|64_OW&@*>Ji{0Fl&nKX&iq~Tz+QM9EipGlF>@KJ->{mNg?
zl@6P}fZ8&{^TvMaZkP7JqUyf!g7*^sfr^{m;ruBKmrC`4qwJfFWNkYs)Y{rsIjVno
z0R%wBqTslmc+{YYA@#`BK&c;GtM|LW<i!T~$3NR8mtj~*G7KFxMz59|10Z4b*-w?8
z4`i|z<DOLe>^i+63e@#LTw5sod@J5%DBhZywh|1`)UcBke=bU8zjwcV$2ke3ws-~P
z8yC0MBX5n>vR1@~@^EW0iY-Su&3h}HtBWiulh7JKhfeJk4?e2mWNV9an_noSm}{dQ
ztX_Z1pqU5npD-NK2%cu2`QZQ}oMOd2D9)GIW~SaG<_F$AV?9dKr(pj4^aLEN`=d9y
zbO#^&dKz2+as&napOvoPbzpC)V3anNw=+?w#&pOuqu=Dq4`V^Zn9Yw+oZF~WglipF
zU8(@5QF+*{*nk@yE&aK_{G(PiUg(n~oDcq}oZU9g>UHVy$ha{5ZmlTES2BXB$kIVS
z5LhV!&f7^5p06Hi(N6J8-3nsO%~o6gxB;w1^UZDRjMo}9F>bzG7_$xlJR3S09P@GM
zkC|NCYBg8T22~OV)v&w{44QC!2CtCiuTAq}FGjod9`ezxjkJ0tB3*_&r7Le*zb+pA
z%#3D=UrL~X8L4-~^oDf8>MqcfC#U2A#kxZ^^l4?cBE|=N{WyVVI;u0GlZl429EYO@
z^_`D!*8<DS=XTQ$zLaQG#+J-i$od@_WJ*;8gWbQKFHS{c<XhmKgenx@N(WAaauL$N
z7Fw29=SuHsBZq<uMqj>!#~EUD1(G?(?BgfIt%~r1Z+WVVSvH0J=N5e7M;vsFlj0x1
zTVnPC4z<U!rZrIn7j(2;2r5`MH<2ACA}tp4vEdsb4zBs<XGJ2)l%_puZ5`q-VeQSb
z?396cDw=R}Y$63_Y?2*CVot_ocGZ5_kQh+OMnx1ZZ=V^KYg=EcV`(#9e@=+@y);nB
zdj*1)+`x8eF{gzY-@0JMW22TGzt0L4!uLm~qSw(=R;Ux&v*lITR@`M&+Q~!io)U{J
zPlC~l$ttc;Yo%4X1=YIWD+qtJ=<K|NQ}2YCD;O%z&GpUZ6YkMx;4zbTg|<V|5Oh@U
z`U2Yy1_fCxj+=73{t?3K{)*o}bqW)D$D#V8-u@YF@q~OE%yrKtL|1zF^ke7i?g~Nx
zx7j3eCwI+Dr2G*Lm29E(o`&2wHcHQ6hi?U=9|JJTSFYEUtC@I^d2P19CT{O3`lTlB
z5=sg4F&(x`j%M??jwWK`mgopv?#0;^QQX8)WyX(GW@;-re%d}o6b7w!x$#s^=asTj
zCc517-{0}8C~nVJ=bm#!^<y7#nq*PFdQeN37hSn!r_r1FzPs*V&csR-)j@YntZa3n
z(0wYOHY1h*uEL}fTC)6AyA98~S6=J2Wgb`4<9X$tM!8O%Cduwazm*E&;)NxWh2x`S
zn-aPNj5&^jTx87fVqKq6LOV#N;J(gExk*7$SAcZI(^47B<m|QAFNrO%*>#4B9yaef
zzMO(H98TSS{PZ&%Muu(R(gVj>{i?hK^oPV%j4?+W8mn@I8W+c6zE0qGaWf`Hf_O&X
zCKuf0tAP&|WOu*Hf40}@T4{=v8T0D8sZ$?q$tIjP1A5Af5&4O~P3-FsD&!@j2%%AS
zF18$EzU1nZY<Pfa)f>U(Y`&ANVdf$A1E_=Bwg_CbL@3}Vb0P<Y?Y)p{sx*&~J`gh7
zjM;d`fLJtc-a}AdIR>2Z$$z0BQ_A?R<feN>;$Y@?d#+H?g4pEevRGYEZloYHFGGk0
zTq&3La+k`J?e9~~?nPdZEEfdWq0O}bEr^2Jc#9^v2X3x!Phoi%M!pyQ+-KlH>JCf4
z*5-x@ySyP^?ewPgM*??RrD=1yy39}Yxw?!5rv9z93jbx(*mVhX25s>{-CN~eYA%Kn
zCrh-W@)vt=A0GEW*QeqralJ1N6WbZcBGsQt^2MHXr7f2pJ|aV3YB<EFxg1B0GNaR)
z@2W`aU03AV&m-K7*I1Xx2zxBP1C9A{u>M(CCSe;LNaz<|gk~+;li$#u&!YuRbj;6&
z54DS74;bvDh_4M%tWFk09lL2mpMRl@4x!t$LL7c);B7XYj97fY>bzvBKr~yt@b2ER
zpqHO1{-XsZib(W^7)*%SKAq-XL8Y1<?E+rsSBWo`CnM>zHl86jf^XfezfrQG++lub
z58Au1ReD^Tz04svkC;b%++!<I4!v+Kb%JDHl+9H@?p2Ci&5I>d>DXv{@0QkWbdj9x
zYwM5`f!$3a`9eQBqXt6j59p|ZXh==7-&S+T6AumIPpk$=bUH7P>YS2wi3H0G-jXk2
zmIzCI;j~dIU|!4^ZuLcWluz3mq%MRl42(lJ%6v^=g_z9YJTF@gOp5L0)vni<2)z1A
zk}9D}?J!Z^?T~@dx2y|yfz=#i*(l5bpc=)^*WZ$eDK@0Y2(CD(U8#D9CU3O1$h5H#
zsqtc@Cj<H2ffl6NqS%gEdZr`$1KtERY9~|Kg_P{YrSq_*nlS(G4W%_SglN}b^9i+e
zs~;2Mli^kCjHcK%p@oL;lXyJTX<F<C7B3G)>gCFjdF)3}Zv5ICOg`wS<e4m&8>*@r
zwz_GT8MiRM$i28}*#P1Gg~D`lA}_arx?rNmBl0V8^!PvS5`uD_vPdcDFlgUKS56dC
z5|z_`ZsrbGlYH{3N%A7pr`Uz{mRx2>Y0z(p3{*<sdyhU>Bn<v^?F_aUB=6@gja{Z&
z{My5AS|bv2n=OXD294Vl@O+F7UZ2^e3%@_P7b=Pxt}y<yk@7+1<p0CoTmMDX?|q|y
zARU5~bSohx(hbsrDBUGUN_Tfimr{ZV(%s!9CEeXM)C@4p%vtU$_P+LgpXZ!E;Q58u
za#UXMU2A>v9ghs3=Hs9Mp0gk`;_hcmV%;BA(%*r4xtapGBnh<4`eQbR=p;|Ko)3YZ
zWduZbRehnSmHzbx<9_~$^bPSIV{HLPBgzl*0Z;7Q88VrJfbx|PXYPS?j`F4U!`^8I
z)j&4hTbv1!beYNbrBbYp<&SP2iBoqAmUmIlQL#N=-l^l-LbW(L^*V>s^(H<l-{{ME
z6WUZ=k=S{QVj3fs9#Jy@lB4>$B8&$HN7Co4iis!C`g|G24rryXSmb(ShU&*S{?eqC
z7EHlb?!+3B?@7PR5EI7&le%bL{^SZNwtP<7Z$m>3b#w4d&VZAt&|WR^mI^?eUHw$Y
zPY&lRbs<oiDUEK297O7CGHCGj@Mn&9oY&(EmsEg08hbvXzdx!1oYkQod6w4LN~*k=
zYj;#{z~g(Nz4PgeXm)&++^qKQ$G1}Lx-r8OHEPI``AH?0+fOcWBA`lZlKu-zOkjf9
z2%j?u>Fc8SLp%aQ1%)6=<pwX}z(LTn`?+ELs~qY)ZNOZH@;Mqkxk$Ci`8%F>2pm)2
zJ#d*t2;pu5<E+OcGgGAO0qr~*J<r1K7QgfAI<Fs)VUjsRH~|Z!C^8I~t75gD(Jlix
z6s7Q+{a+i>4Vz-ErWwWN4V508W;TV2pOs$$GhXdC0p{8ARytCl%#JB)q*D7U-jMt2
zi>cu7i?|Y2ndmA3Rz_ShPkNs(W|<8+zQ2@t!mB&ad|ubk42^K-L<+wlFA<=XGL!1}
zC|#1ir8Q3$c1|muH}F?IIw2Af=<l$R)7u_2%*H!$=y<x?Zuk|h6469~y7u|%KwVG4
zG$DfYmhaecB|5b7qiuO#R(z}*)Hy2OucMQ`2(iFH+rMw)&v`uXus)`hmiIm9l{49@
zwDMDvIti;Mn!@Qr=%r72ZV5rc!|9?lp+(roFN@wT!=lqF6bbPYPac9rawm;oEG!bi
zjrhoGIsi$V49O7s>P9uM8)DUEL{)PfJ^=78MFrghNJ;yJ^gA4<E<SctNX>^7m;`kk
zX|09qKD(MXayhe!gklo*re&`P6?Yg^A8<K@;~`iW<PgPDhE!@q!KB|Mlj6+B!F+z@
zMSbBXGg1Ax&0;NB15+iIq*Pzf>9$_NYGfB(kIR=xI1Qc!hrwyrzz4gQIGTTyumAdr
z=GXq<kLh$p3{tDcC%Wb66#|q2vexzuM;rc#2mOTTv;<vBoU#{;JmwPa=bBYrCx+f&
z?cE4Hd;Tk^1XD|C?q)TGv_YVcAliA`H(76`xE=QPQ*jTO$n_FNyDyJ}P*caVMbmN_
z`VZ;ixro?^<Bv(*bXA<p-)tC~{_sKS_j`xwk5407j@E$e+q@r%HWkJ(1{n&E-H3@d
z>J*op^7Ansc2foAM-mNxNHKmPKr&$+y&gj~QB|aIZ)Qfo3x&X!x71#_jr%CKr&m6I
zFH$4w3=VpB*W$a)EP~z2Z=lm$Tf{zoYOOJ>Lp@u57iLbix-oO)^GXC*qQ!@WxeK&H
zOqMj#vwb|Pg?nA`i+mo9k-G0Lqg_*7QLqjOE~_hEw;+eLGc%GgQ)VRv2*=SRLj5_f
zaVdf@6BJoJUv@iv#tiXqIId0f3E%b{)fT|_tDw6gwVgYCV2$?HnvDgkTc|Jk2q9BU
z&i0i<296!kQGvr2wE8%3tpn;5+u(re<1HHE8*$?%P1Su(n~%?ooVj%3tAi&LNZp36
zoonBIeDN8FYBx!aVV~hnl#r;0*K%|U_cO0Wd=sy~`$wd~WbQV+@weaXofiOqXp&pN
zyNvk<f4vPvOlf(}?B^FN-!l-b#Hb-GUnN7g_?^QKe4SczG&H{laeu4OAfuzi9!5$N
z{pr0WPrHrLDLvhrEhkSLa>1n_lr?TDa(Z&z@~-=}StQ{d&0JH5Vifs&C8@mCheTb#
za%!==#qGxzfy$+*nG6>xO-75+T+?bcQNyv;$Nj2Vf-Gn=<_|ZwE7lq{!ZAN{+vL)2
zlvLeBg!{d#H5Lb^bKx(q0z1z#T?DGdbH#r)N-jtR@P^$i5Ktw6AYU(DWpaH;{h+=P
zU2GjV@vc+a>--mAwfz>!TPBQ8$%dFS6(j2^v3sEe{OLGd&^=b_wvr=ho_F5bq>M4H
zuh*l7a;SuTMYFR8xJ!`L95l=4p#pmtl?{?UVT0Jh#Zpb_1gv1O+&A^%Ig*5n6#VuI
z%V})tKX17f&Igj4F0K{;TMA;GonKb_1P`COmzQqxjk3HLg$K)BUuS>>(w_#Wd0CNl
zk0i4i4vtTNhotXMxpQl{KAYfdWGYu1za`bxQ`Y=4BQGE~Qs8<OMGw8@+91_;<}Whu
z=O|!i_|%C%4%2f#O=lK;^Wg-wCSKT*gb4)oczU+%EYD>2`l$lZfV&{;qKuDpUDNT^
z7q(A-iXR-Etlkp9Hkurmkz+s(aZ3*U7$+xXOARTcFr6-UVT&kp*5!AsY~{kt6z5bD
zH|17`kOFSTV&AXrms~#nTsl}S9tYKwT11|P*4Jv*#BOXK5__9hZ_%ZWO!(r-ADm!p
zgyA3C)!lJ060O0@WIht{6u!D1R)@h9bLl~xI`xW1;(oQT{O`ey4o^BuUYA+x0?W*1
zT!awDy<Wo^kHOBjSrrb)wc>Z7QfL>7f%i=oOW+kRwBE=hKx@i;9o`Z(gVaxlq_90I
zlLr&|lfnfAC14Bh1DEoyISbVK7rm7c^pHTJdkAnMv~sLSRm6$*H-7c$iN;~E0K{!1
zi7kWay>vhX4QDxF2%!LDmQ0X5#t;GHM6Buyy;g6=ZXXRdEm#kERjmm@NX>ij_1IJj
zr@yg9fKE!l4O|W`oGZI_#9AP}7-JRmk~bUXp*q}<ESKwB7sN6k0Fji^A}eQkovD2%
zNVo93wVYzM#NHJ!4+F*!vnK*)pf_7Uf-W`xp*MBDpo`wo9c0i8^IU#R^?b<BOCL%q
z`cM)2-1mB5n?fSzaXsp{shRv3--`z){w0{W-aoyj7dTw4*N9PV<{aDxk&&w`7JuG?
zPw{qYi+ua`0Lb1@zkj`8tr>c5(oa`mmN@X#LQhm9D|?css^g%l8O!bGhzudKcH<c=
z)5hQ&PxsMhAAP8qE_y!ZY_a_0Qt@L}C=DZcYCO|fopMC=syjc!cN34zJElu|7T844
z9usL7;Ga%%oz7qDDKYY%NG=FnHg7%-rnx0Ot^z%9#&Z@Icz8Ts?DOChC&__;*2K7>
zKXsmT6-H+PT35~U3#X`2ZA=%cQOOHIhzRmztZ+5)(Ru27t}(oedjtV^2CBZ4!5;VN
zr~#iNk!xrc`J*)0vs%EZS2|b<3sH6X_C2%e`Q{cw@n!MKpT+b#*cL95<7+=-%O{c8
zQ9ZY;R-}&h@~J<WFY7y0oH<mAZT+6_by9HQ_o+XTcdbRe#t{43yX$OQ`8(w3$V@Kg
zZstjYNsVeLw?7sCWHG)yG7(vh|AHKTFy^WxtR3$(-ac+Toh}Q9;^Gq=5Mk-E6ZiJR
zD@CbYBD!wC5`mmQ)6EZP8_JET6z_jt6q(XFA0v)PoF>2guqt@jQIhot#aRUUx28cW
zIqH5F;=KsVALz>vY?Np3f@U2$PHmYYgX6E~?k#59=pMRFh4GcbIEDT5M+i<TDXAYi
zn1&=tSmAo0_)k`SVtkWpew@nh{&FXGpW@fR>U&P<O1^A(Bw;^<H@2<fnOR=9{fJ$V
zDe_Ly8T;zzcF>boc}5}6f-um>_hGDJx5agceN3!gCOLik`QB&F+Zh$YO;jT>AL>D|
zGS87e9)fRKghU>-?me<cZ+a-g=Rwl5M3t(jQZ5iq&#Ebq^F=<gxB`E31CW=v5=}Sz
z=@}0}%8l#}nkj#y6YGblF%(2jC^_q-VN@L}6$)g(0s8tGLc4SO1tM{?Ak%#enm1Sw
zVdH&Op;D#poAWDF6$!^*3CeC+wv%KNkc_}@?`(G{FQh)z0odzq;q|dZW(CGRjUN;Y
zfJU-Oq1PFA6IxFcN+}-q?yxh!XLMU?Rj?yRw8TBde<j*MUVsX<CFE0lyx_iG*<&37
z8Gh_8C{n%>F}(}x^v1`BPrOj}vQ<l>BcCcv08qbz%AM;iI(lsxZZ>%r@0HW93d@5)
zBA^!Tpqr~I(>^sKyFyeV`~J$q!4=5T)wy{zCEbmIrC%ZJqN7N?QGyx<MUV79Z^J5P
zQYIMRP*ZN*c^4$etYjcsk{5b095q96&5a?8Hj>Gm6-!s{F10V0tv?*?9~wt0-hJJf
zMB`e{4j@9DPQuQtQM7ywwXz;<f8uX%IpSKqeRV7+zO~`Nj1e<<W*-w@?0>%v2*pfE
z&a$03T1pen$L2tiFRkf+W-YLMYOr5<=6i?hx<I@Itl|}C*>g%(cVskk3avE){NK~8
zb@*l1;y$?u)^9sZs=aHG^kvMwI}_4m3klm9O?nH$987(Fi94XtXe<_f3IddoXoVd&
zlV_Nu?`ky#zoHQ0W{|-LA60%lHCM2}HXxlixBXi{&chjiCHOmM6IUev#-(+mA+zHT
zIGxWNAyU4N0BoNrOB((eK=O9@Hd=G=LPM;Z;E+z89I)k>p4jLZCbhN*5WT`eeXoj>
zgxcRi8g$dbgyrp&K7($eMLR*k_PrZ)Q$^a&hjES)rxTH>-38jJ@ZW!hLGars1=9}S
zW)k0&>$E8^YJ@7@Bxv)<AzZJj2A@i>0L?FEZ}?s*vssIY^kphTiwzx#6lZrdnf<F)
zz5zO|0A~TZDX>s|dgH--`F9*G%rv@RfQ!=2nVWc8&`=&;&P(?na+GjpT<d+*k%Em7
zLgRDpR6P}v#!LBE$G5H&_1g5j*+<2E89#G1H}x5=ba6ScuWhta^bW}wi$m#bqe~_@
z(v`cKq=rALYU^AU>gFO({nRg|l_1n~-wW-u?`HUXG+fczEiFaMza{7}8iK+6T6v$`
zJ7@QkO$ir$dpE=4M;NdMs#~=TapcMEJsvQ1s?PSJrn8t~O{+6K-h!ly))J2eQsSf|
z;g)vUkZ)xD6pl*4Ot-pKE&J)n1Y%H*dvr6}B*lQq(|U8sx<TD7W;?ekefNM{+?@<E
z&#zw@s@3k_-Ky{vTh;F>c*21YWU7ut!*FFLQX5+DdNess2%)>Ex6(BVjbWn)D(pl+
zYhZ{b$#OXUOwQYd{MGVro8nfZ@2$g2@tOdbjjm>wePYO*uAJX);WhRU^W)hNNAdj`
zw+;li0ki5Q=H*KlV}~qx5|%!9Hbxve+kVybyl_^7;&i-EM1)%pU2#H$5BaSc*>zj>
z@!!-{<5zUXHn^`L*ERR=bx;(&Eqn|Z{@IQhMQ=}mjSkC)5}6ZjYY_1-0&z~ZhCG2O
z`kz}KTHD_c@|W2UDuRB0M30J$VF}_{t2e9Yfpyn3xJKWl-aL#GW8BNEXmva>93wy>
zxce^OCBk!2xaOMajiJ33N6C`oZbfA(71+~vTf=J+rWz#b5@_WR0>n%HdX*J)k)9Lo
zuniNOGt&@6faH}-V|<uv$xy{`25G%+=*UMFP>z_hp?CgK>>)a;wpl;++%W^38z&c;
zc0EITKVsv|7K6d;XT%jjA7E`X!AJ0g`{Cj80_RrYAu*=5Y=Q=p((43vg|BUaCv51)
z7Er-9`dh<}67MQVsFqkSRL7s=4V@BY(k3wf7-c0MiMTw(iFu>lW)or2uEzAHUiJNY
z7VN98=TI_TnI3>-6ZqZz0#Zf=l<MP)W4I{xHLv(~3XNEF(hhfshgiP_cq7z&Ie!Be
zGvEvMqQX;{tBZWa8U)^(R<#SVE`W+gCW|j%i-?&>%AsTD`6XlPTuh5yOTB`N%XLgo
z-@6N?=?XRcLl0K`L~ZpM3315FFJ;*^FddV;+*j9F_ACrsRpH*UgG_rvB+B>eD;`=Z
zNOz<Si?<7(p|yL%mEzdSvBuRzg8avOy>+HeXj|lgYq!G20D0=vf4Vx5UCZ!R8)o81
z<59G?F$}`nFmsYwUvc!y(WfF!79y%2inu2;n_KM(5}fGOmL!(AxcZuTsX_DL4Gn$!
zXH*Gtdhzbd`A|Q7QsxI-CInv|PI~`J{yh6^_h&loLiOG2)l{~(>Z~c8CDJF_Stw6m
ze9)GAg?Hp|J)FbY@WSq$)@q5qT&R~&k@QUrY>n~58d!Q|fH6hoyXFh>9=`3Nte)eC
z-q7u#w+9y%4mO!*=#c9glu>J(@7>iqHq+*d)KtHYBfiA+ZDVT6cdoMODfY+SicQR2
z$*Xjy$4ir0PBe&CVNUFh=`qsOo#a7H)Qwk5#Y;O>QAE<I{$5(H$^+xrwZVb@=-)Wh
zK>nmoK%==&uoP3|Ox$ito1ng%Y}1e8d}%t@sL+k0cJN_tnKM+AM=|m~A`ox#PMGNe
za$lX*X$I9GPu%N%OVZ7eIQFDwz&HE`o<?W|;{|*uO%joa$Uv0gmB(YpCeq3@9w)PJ
zHmuaYUy+YY$H8~-*)}Qw$#WCo<FJpiueO5&u#fmXYKglCSLc%Ha7=f40JKMKyp0r{
zF?NOV7CBDB<0P7sX)IifEo5>{nnt}q<Jdl7A~L0%=XF>{S5;s}*&F6k^G;v-kOTuR
zS3c%md&Vd2j+dzyKRUj}@t&2iIXFPc8X>AnFD9EFb&^TF3qp15V_ZLhX`p+%i(9lF
zd6W<8<tEjvHWg|%$xP<4J83uagp@uJEv+bFF2Tz_M!pWDjl<tOHp|1M_aNMf{8Yx`
z2CY{mRJoFsf71G`ACx6&y8HD@3j|}0!UyF2a4trK&_feh)Xe7$)aA(UusWN>c)rzF
zWjfu`JhmZ?5_@)HM1yf~BjR%-?+-EU_ZirL7{go3d;G5sS663tv|T})6-CT`pz}vJ
zMn8Is-jup|Bad&uM-(Kc8Ro(Y$LGXi3oSwhf&`=Q53%WQ01&3QQ#oz^M1~as_*0`p
z`5Hi74ma=-#0;o@bgM!$9h!f#S~LE{{whKeZ?pZwQ_b41K1+oeSMc`{MNTcUr8iv<
z?%N>|yyEBn;);<xXGdn-RUlTo`;Z~cdxV&g|GBQnOj?C58xiyCOXsC5VITIKIy*Zi
z2zaVX1+6`U>fKF#R*MfMJ!>i_1MzU0+tgCS6U?wBpF@VzLe;cayMS`K_aNh1Hv8ZB
z=nxi^Y#26&$O`w_9}<}Bb&srO8&hMRT(*db%QPtNC>`WyxW$>hkXo+8zSm$))%0C+
zB{-l3NphX?Z_Aw^V<#+QOx>YYtfQlL>9ZR0Z@ym|$K4<{?7~2P9KW%;foWeFc(I8;
zxbA4*O(H2EXY0@S)5QX0CFTrn%_-;x_Wsi*ruhIKQwYv4<;kd$sD|AFz%}uD8`wYp
z!WM}?!BpB1_1!&2qq0UdgpZYAeatzP(b;nj*%fOoIL9oNr0r+L_6af-SQQjeHtt96
zKxGAsiva^zOu}@1yhbr)p9PjYg2aA%0D#vZg}-nl%E54~{*dp()))xdr`+9mv{fwj
zW1#8D$dpiNc~`!n6&Z8-ea#LVzkrWci6-lb5rp8wND^-}Tr>!i(*v}0lBc`iBI(;*
z-02{4q4I;ecV2Fg%45HU5Hi}9i|cz*b<0hH>PHz@KT&xi4AJgMW#16vR9yJQBuaS0
z)~cmj_AvVWrw^RwLz$B69Dab;4^{oZ>aGGgB5SeUw-8}4bM_{rSyGI^3_x1P8~UOj
z87NEmEEhC=xbSH4>3h=TJwC9sx*~}oXRhjYw9E<g=O~g?ZU+~{Bi=Map#!j&I)0Rb
z+Tnb^s6=kKjY0CCMcs2xG0VTb!odOkK#NXQIEoWK@B@~|sSCV*mjd~!0l_s7{sNs#
zq%#BPe`RrM!+&W`hH#=pjC_ixClIPB0&$$;C{V9)=xDFLkzx;Vat~q;sd}xrpV9EK
ziI&^S7|lE)wSP;z5~ZNY&Eq-!`=%O6mkIzXK>2p)JX_rPr0BlBxntYf0_Bu}>tO<E
z60LkAmL^=sFJ<}~K;mhVhLi-x(@5{2R8pXB?^>fUYYgWLp%RJ=Y#`r{?`GYJeC8?8
z*icosK68XRf_Y?y`>(|T*$H8%cTCSj0iczlrk8r0#o36{4I>D;t>cw9z*y05K!5d!
zaC#9JZ>g&)Uil|^=8*?cTpuf12x(79rY}9-Er#I?Dxv9x86O5tMR$2x)vc>68Qet1
zXK((l(R?e{p~FWRHGK%lyX{$c>RvlJ&H1^B_Kb<t;OCdn*{R+#of`96n%%-oHaSfW
z5CaLK&Ss9ukCvH{A6C<+h;d|Ttk(z6I%2BQ{cirV{la)Yz8!C~<A7+QzU?BJf5>y?
z8dhmxR_8^R(enPbb&KWBOvt!zX1mCLxZy$AhpG->eyc%wuMk8`gbWlC%)|KNu`?^6
z<M+zcCDLoR?f_Sl)Ro6#st}q+_4Jg2>@OUG1S&0X6H!FWSLuJ>*AOO@9Yxa1eJ6s{
zULxsqvfgL-QFJEkV!0Ne>`{^cTEb~(t%|#Bxp4YfVX<u8&+4*e<?^~ws%kiNI$TWw
z_#B%A^G(nI>2l=WWVr}>W!eSY&C?w=trLwvRyqHrI#S244<o;3(H+tV`TX2CrGwAb
zg_1j4;q*~?T$)ugH|S!zO77e>=C|Gj3*Qn&rImz-J;k3~#kd3Y^3!aMIwbCW5zFP+
zZfUV};Co|(lerRFeBsi~E_-4AULb8L`B{?QHE0kYu&tmXi|9VAqd=qkV2pb>6Hgx+
z3|=dd$o#ccu6;6-Yd~=_v({Rts$Evme#&X1b+>!!$T$B}=W_c<EZBvp7l!7{d|=*R
zes`xe*JR_6wTHxpZS^MxTLNVx2XHt5A!0nk|Nbaf28oTXk{*SGH*BjoLQTjbotoqv
zk+A;qP0&rEh$ms&uU^kO40|LtPY`A~@mrJX(-nwLZV<6}Z2;zP9q{#4^(IPCYkW?g
zauK~Bvk40IGjnf7jdO@=o&l4LQHvvsIXZsVWQ?4cj5o&3%{<XQpco)>KPh%l6E1`r
z*D__s&P%>tuKMv1yGYGe-Y43;qHsJjX}Vsf0qhuezbCL^=1`NST9nBT_xze8;JA1g
zTgekxWv#)cm>Y!5viUl{yO*3o@J{P(vM{T7<fHyFMU6&ax>wht7FyXgjG_%_0ic<L
z8ncm%tKYuK-mh+<sj~A7TE2GFh>MrPSP9;%M68=gZbBPYl+~UawCgUq?Z3z6bw2d)
z??KU_G9uyXdc+jf0C=WX7$!`V&RAA|Jl&QUK)6$<{hI}7^7~vX5iuU2f1j&KGLqB_
zy4pvh;hkDWK6i?5Xdjbsd_*NkwUwaBVayB^PHa0tJqC6o`~2B=5k(HeV6$TjWt2{?
z*H${&`hw<*F-=}K8YOc|BfGc;+13ZuGtZB**zl=^HuZNZJG{sJ+H%-G5dH?*O6y4o
zWAvrH{Ipjx%C!nhOY(losoP3P;g5?VELO?$T}d0I!+#@s8CSsRN;`!Xn)_u?nN@{i
z#&u3*j-e10)eo5M>#yvIBkWW9M}7VM=`&SxQDU2%u8`pOH*ZxiM7+Vu!y3h9<Kd`n
zD_&Pg2|agE=f@d;*Ld9z094w>#<L6Rzd!opLMjUh)ia5mlwJ(HQI{j*xn*V(;_I=;
zU>?WD>C3ni8d?Am>mN^A;+N5#t7uQq7pzf$`^4UI#T&b=$0zJlf924vmioN_h&?!4
z_MAKQ4H5uV-g9d42|~D`{PwFnea3b~z~I6-(=!63-DS)ksB3`gyeI2$hgJ*lSCYWV
zavh30BZ^Xdj=RIuqF#(Pvv$>A5*c5cr>K~mwLefV(|lV~lC*W#px%9G#b&JDw>(v=
z(4h#-=&pQKvtKGX?G2y0ISsr-jsE+#I&dMUp&)Lt(<nC!h1O^Y&*&ucgoV9NgIWqC
zX3+aG=+#(^_mD74-ztgFh<ULS52sY$iK#;+Z+q^ki2IQu!TRK>xb6q5s{Gf-vxa@W
zW2eOwf6DOud3r!?5)crF-|Ovh*rS})^c0F3u#!Pz+HnvUiOfrG9E*Lx(R;7D_=1$t
zE_B9zLh*3Am{}HUXAUtN42nrV%U&#cYneYqYBbWA6E56)InaSkfY+a1q9%Of9#Hiu
zqh5X|YUihYe8XF2IOTd3Xx_hJ8NJ!4fpbx5ZhQ&NBZh_Q$!y67%zGmU6(d9Y5sP-H
zaYLbMjsilGp_Dabf}6gzNW^b5Ik-{$B+h7iC9zN0IllhjS3#<yS(rl4`<(qP?xrLT
zz4y<KB{A}8v*OYJF2h>gP@bc~-Lkyy*w&eqzGxS~;op{`9ex!DFV%OxDAl>MG;XeO
zz#!7Dd;iTxcyQp`-h5Tk#hau4prI_l(r9FFr33eB;R|0mL)n)Zch6<%ofW5@L%x+T
zOw=cTrS?Dc7GzQSqkpT9bcgSW{q#8gX1jUu4M&>dXOp2;*k)XNTYEAQiz@bJOHUFE
zB3^J52TL~czi!G#ZM?g1aP-b_zOA^8zFs>%(<-7ZdRq{m)dCs*xp2vG2HWhz8Vcvg
z5TIfH@D>`j))n#q4(8NpjWE<$XEpzngWRo+#{4uMl~mbs+~6rN?|E}nz<R9uieuBv
z)C6e%OBImx{VJ$)gqQn5qto(VW$ogz;kXq$f*dU|THfXQd~hGt^u2{Bj%XRVB}+>C
zPq+fKfdmyBQcMxphrge}jko|veo>9aa_3@TbIo9aZ>uVb5C##l{3_Z~)yN)x%DzKK
z862cgaWt0MW*pgkdvU`~*XD5$b@$*r(S8pffqA@mLbS)DdOC6Lrdpg$vuWAQX4!v?
z=A`Jk06AQpL`0ji{o_DeH&9!r#oe=^`(VFQ)NBZIV0moE+4KWDx8S{YQSyqmQ4K&N
z<k=-Smc0Xh)rN!EASJ(^5(qa*z-Vxfdk!p4qEgd3q|pnrkk`QR*=^`pBA|1}mk|RU
zFd=moFIRguoxBK=G!-%$-pL8bQTc_=((Kqu&%M}1MX8>k(U*FwlC3+2e!qrnv66%=
zPHL0Df|5!>aJ<5Pjnz4pjcG3>uS>c&i_q1F+5Bud6!AN^Q&83Mg15vTOQ(#?<?P94
z7gFQD(jWmIrU$}d)1}D`7Up6#r`t*|>(tn{DT}SRb;noQjG2T;229qAwngpXi*;Ui
zj{#A*=QAam6T&&Q2t_=qTh%WwN|`<x#tORj8}Wb}o&j>u56Q*y<4*bgCwa)lhR@?v
z3r(7!i4ujJ&_i{rGMB+Gg2DG<ZXs7RLW<%LU|SmR7wjMJ=i59yPZKU660R`ye&p_x
znEyTk1mnMOdyzAr2Y!9N*<W?`u-WH!nD28*YuJ6wt81HP+y~ohj<)|~;7ZfJ3X*q%
zUSQW8IRVo!A7#KgV<S91jzE<89&ZuUC3fd!D&Q{vL}b((a!fDU><S?6)>=K^GKH%!
z?>bzPavJW3R3T^aI%H>7`^})-79iZKO=pC{PiLk(ZoS;j*w42HoPY2OW2o=~NL(J1
z%@?<3!^-Ftu2U>8>J!@N^?FR1oaA+jlZPyZy2*LW=dE<wyn<6Y=Q1A<J!Og_Nwhau
zYPrgcqYU{>>EJMccGG|UTygs8Ztd)>(kF2*K=|F|=0}Q)`-J;9)Fjpyf6;sbL9!Bv
z(2tk1c1PvqhAk$cdqXJ<(dHvxhODD0gi|%Z`AE6&G{+>@IVe8$J_jmnXlhqrTJoks
z{o7~htksRUx!1IA$LTzv40|5q!KVvtdjYkG+V3)N{I%Z?r;lwcmK$^wL3l0XS(-FZ
zI8~3AZ=;>|P7PMBuRdSS(pipWlV;A@A84=-)fD?JDkHW+G>f!TQ97t7NB({ffr~)q
zc*dkO!kZjVm&`|(qa^5(`{B7`)Zs>24?z$1@Gx8a5rO*6KC<}4Y?{iEz@{!=IYHU~
zcq>QDcb0J3xrBfVt~1L{ZgIC=6;u^RB~E9k_(C;>zNzz@D6RC<YzL^z+az40K>!&i
zGs-4!>uQkuK_|KA*UV+O)XC%Y#sK5k@nCfPoyt4+lPwn2>5i0NN|{LTljUm99%B{&
zIAI*t0SVh{_v|6`hgSZ})a-YbSs!ItH)xoJUN{~xr$XR&pEAl?hO<U^b7zItE$DJ}
z!}e{NtA0%ts+M7*h8%HU7j!K64?I&y_^=0_Aqw~!HzCqszm_2EemZosRw$^C+Ukc#
zDUx?YE$(+*xUwo(5~ZZmYK3et(>5m|na*qpJx^Q9DlzZR=f-ozpfVTHK3!?ye2+4e
z`}fNI4^lfV2I?>3Sa!LW9?YPn@ZTC9HV&7yG0x#32CXhSSi}`S!llyFQ<UB-T$n3R
zg}tJmKW%%c?P{^gVV?qHqs_G)qm+a;(04!VizOBq;6{`)mDWlgyW{l`X_kMQ``dV(
zpdHhr0ovH)8V>AgQyn{0W}pOXyalQ@b-4sgQE@C-rrW@Z_tj=!G{a<G8&i{$`3!6w
z<}X=y4KGgWaRg@tha-}>%~%eL!&u*`Qv0{IK~9AkBMF(}dk>|(uTjB8=FJS#tTBe7
zLJD8rJ-gyrJrM?LvvKJgOe*Xkhr|h&7<`Lj0(n0<#59wQ+TI^}VLjiF3Ypw36aKz5
z_h2O}@%ld>F-C%?D40`T3I&9cM7IxI=?ljMSjMObQ(u<_l<79L#&*S`Hm8*ECF2)s
z5QZ!UTJk5J(QljpMZnbo7p7XJM0UIY-@t!z=R87DTeZp-o?0l8rd1&_=-*nxEwY+b
zN?vVJ7ILYUA6Huqv94HxDs!$#B3%mMr%5<gKt&`}(9S216p9e^3~D)qzbWb0{OPB4
zQi@H5Li*e=bzI254BcTVHynpzS*el{Qr3`n*xRyOD&J|tpOie;G<H1T!P5M!_<L#2
z$ku@B6?_u$dJtb6rkW-W&Hw(ow^c=j`CjZp!AuU?Ujg7>Y(EbTR4o1odiRR9_)B!r
z5pj~`BgNDirH^Crwe4-q#l$sxF4Y>0jWEySv*+RA`{EzERSz9t4f^%}v}8Xf1hu=|
z7Pkv+WyNK^y0(@dJxsaRcjZoTX?Rr^bukz8T~lxRX}f=j)UlkYA;NY`T(4|5;4Y7$
zKflcjMInii5aUD~&LbJ#aa(JKe`~)ExF5+pf^`v#`drDk-xPNoFP#z1IbBuySr^%p
zoZwDhY(2lrOX%5XK^BFDm`X8hUQ@JqL$aSKmS^g?EPYoJ<CdOW&qP{0{E709L>odF
z=yZ}dynCP;^thabD-CrV7Ec0k^y_UiYW`S22xb`!k{vPMD{q^%pu4pX^lAF8*5&i7
z{UkhyfOjkmR0KK>ZzJX_cIN#zxDj~ayOEG{(uEJDbYGOQ;8TlT-7<`i{!;GkJG)v4
z*S+!0OGNJ?ke<{&P?trcna}n`*GSAYD;QcdefCu9zH>@AE`tp5nfH4`H(VRN8Z?YK
zU8E!1cc9E}bWL44zEUUWPtNNU@VfP0W=DO0$|0TUbnfnJH3?+$4a>ygjI39QTkr1K
zqH_Sw!CKGSc0<G24dYVZd6nrM<B}JAd6$Fu8R3_|a`PYh2_Qwxe|dv%W#kO?F^un=
z&XAfSc=?`^Zhg-K^~w_{#H*2-g}>9L)|x|+E5|!o+{cHv_jmNs%YhRaBZ-Q*(&30t
z1{GuIzZGBpEUSO(L>v=VqWaxChMUU;w_jp5Yz;B_ue1ekk7tZn%mEzQTZ$5@Tm!#5
z*Srz4j2ES{fkp{{L|uo=U3G=hEAa3y;-f`(s$qt3u*Oh!weU3B`$DUYd)-XA7%0#8
zD20NU1GEycc8c>F+7P@_?F5ux#X7#fYYZ8UY*FqMegY9##>--kq=K<odVd&m&ah=p
zeSFke+Y!qZQGcWFlGggbdF39fC+C0u$N{K^fVMR#_gQyGG4xWM7yDVFF!Q3X#@?>y
zzkja)RH^U)CXm^UUV}rlv$8AVwasQBbPnC_LB|1!8tKaFw{e2FyQ--YCbdRm!&{3h
zT7{hFpV$rBU$s-*M!O>G@L0qft_Uk4=)R$Dq(wa*W(kJ+eqYin)}X^E<8>Tnr`omL
z2jy6Rmpi^2<&^tLaT;_u_B`Qj)_sOj;|=N=iTJ0v&p)CUa7WjbIwjXR)5YUNJV0tB
z_ia*;0FMk|OU<>_;#$8clD1N!6z<jY^IEt4byLgPW!dM}I>ZFX9;+2PcZEy#{myG0
zV)poWG7m%!wt8bZ03%^%0!WEg8LgeTJ@e!{o^pWndrSU^T8bmZ(*N8P&0;-%(&^$#
zNAhA-o7?Q>z<VF$-~8MnDt}hp1e!jEnFGV=#QH~+`OoMGTt^tAW}?EE)w%rb1sxAd
zW6SXH595|z)DdQ`wjd-@%k&E$eOE}4axx!3epDV^Z(c-A)%n^asuQSd^WV<33)!k4
z?c#3x69IyQu{0&x_vCerXNRVAV{+;!h+&N*i-V}T3;Q8@b!I2?$juWD`F#uiR!d>G
zlPQfRr<>PLMSTy2KS%_wy^i@W&%J-}p9$3joeU(c-km)-nB0q}ygmE9d{i&i@Kkk&
zy55_!g{ao_yRxg=Xqn9)-9c`DwI9_lA0|DwUz!KqVX(h3@idmt6p;^ID|*wGSZdhv
zM%Vu$l-5VppV(=@{+oX`ZFE|BDqsLQ1O)1RZJ_(2PVubqn?mVm;~Xv=&u%ELem`WD
zLY_Au)%DK^_AhAVzhl}B-B&a}4@E>lYqeg3)mNgg^sz!BMRfak&<!nxblPl6K`;GX
zPTc}@`J9(^uk#KqU+@z@K3xg$d{BJ0+Qy=kS~jVr+pho2bI-!(UeKRo)|KCWIq~V6
zJb~q*l#)5upMV65k;|Pn+^|{k%;C;51HA3NU+U)8ezhv+`#z<2ZXRG$tGvk*%IdKO
z$mRd}()`yo{Zr&Kpp+b(Tf<2KYV?Usn_cAtw5q9&gBe*d>p7Kj{Wj(mZ`^9YrUPn6
z{xxye%UiDg5QWvnevSF-^Jd@PeMjnRvqCL7<CXr=*`eVyj-nAL9_8%$Rn;YYWQIzU
zMdOEVQEBNCFi1k)ER(v}9?X^v_xaNBv4|!R5elePTCT3$eQebOsPNTcUI1J|qhI)+
zrdR)b0DvbI#I!E@;n3E`0nx>P5=cYB|JwFo87@s+Y#A{VD|JSl#i+0ug6E2SLMiOW
z`t`Xj{x?*TVgePeh_l;^?dhG!x9<-){F0w*OtWT40m<YF>b$kfp_6M0@(1~$0khsQ
zI5}}F3jn7pC^2LL<Vw)mj9T6khev*cLQapoGBf@^-h27Zs4o26dehrq&DXJ5BRqDy
zMm#{%6x%kl8gF8@O_bt6y1Jb0rI9fq=J(J{Hl6t+@Y)J~W%f1JU`w0YgwCF{SO92m
zG1uX@;+j9n0H1-tkVfK>4z?*oCTDb7S!(3gEA3H?rAuH&;6<|v1-qA8e|ORpe;i9K
zoD10+h}%+&%fd78Dlu=HbcNN~@WX0r{<T>DuUr53AG?K-IY9n*Z+MIJYKkcZT;6Gx
zuhJ@|Rm^ZtjgDH5d{Mw|!q|FHaV)t8dzw@@n%Z1ftkA^6RN8n9&i_H+@Tp8!q~|eg
z1qcyF4O`karSb{AO1@Tpm@$-#a}E|5v+R0Rvro$OKW{dRoHi6`8f-nRE4?@ob}!?>
zlEvPcLlIw1>tXeOz#KnY3YCNFR=)e4u>l(Wqf)&2R@XwYa*xR%!nZ0UnWFBJeOC}o
zx^He+B5)HxoMT20iXRjGk97A(QT-oZ=>f?Dns&r1K<9?#I-`E(9DG8X$fUwfQ)kWo
zU6EvxUU;)Aj*Ra$b_qrakI%F=jGVhR4p5tsUagqvHZkptu?X?5p_y<SzG@1>$H@)a
z9V4TepnFs;s(N~y>$WD}isRs|J^WaKNvQy*c^N*H`m1FPqFdSFdQ|UnpBI<C9@rqx
z?pe*~l_9G4o@y3dkXo~Rf!0EiI0=RxEC20|?`k;Cd5`lDb*#DOZt`?%w^~Pq79bht
z7gg8e9%?oW7pO@x-G3rxD_9lv-$=B*bI<s&`G3%p`WJxWY+0}dQ)GAQyNU(o{qI_S
z=<P_m-ZCAtSJ7rD+W8E^uv!rt1p3;~C{fS;SqgU+h!@g!ws^<coxy-G&hL8IYlXS!
zU3<(7w`D9lk7H{b24eFT8#bn2r7H_0%R)$e2)T*xY9zcqs>|q&=|Jo=rYPdaA!mWI
zTGxPJ$yB6=x6guMhaAvMd|&#BuY5{7jbGlg@a1K(UEjq3D-zcISjZo4CjXr{|EPut
zRxxF~u5Kb@zdHldF5+y<Z{VL`#>i7Z|G@?Ec(_G4{9-mt)bCzKl_Z3*R-RP^zq@I|
z5I+!z;7?WWtx_9WFExH<b;xjGZk?QP$YSr+h}>jBn-={+>$&VY?XRpgE6n)-)W>Ur
zQ}jsJseFY4f^5vQOiZ-LEN<*J)ktS6SuJ98LC4p#Umc9;im~2aF)y`2@vnB|A_-r1
zw15dr8`Wwm8V>n4v9HBLFni$6d<hoJ1*WY%A>$4g-a|HjGA_tb?r8AqKW3J{;zQW+
z*@+DptETz*6?0FFTRy#?BOT@3Ui>7S$gY5L{Gnx~Ay`DWrT18ODZNcjv6e@p)b~%>
zczN19LIMBt;@-HDJj$oIqb!4Ui^V7}_|}9by4v9#p@$oJ;ybG#!`b#{tr|_Sy~jeG
zmUB<$YQyfqr=m1e8w)j-tQQZ;jY5yX7wT!yi!Y9x!&jmUug_W?Y)lhS>2Z*iWR=rt
z+Z(Y$H)BylS`>*Dg<rglaPNHd+zsW+aaHtXVB1E+OmE=jIl%397sgp_x;fvx`!L|W
zSG=k!_n!oVe`mozlcIzG{W@{%Y3RSrY5rHeFhGI{FiM4ET>kL<_P;+8fBm<O$lv>%
zs~4V}|L1@Af4|>fM4-P;?vL~OzvuVApUeOM=Ow`1qnUga`@bLckQag98pemil)TLm
zpOTz*HdH*po1T@g9|<O=F~*R{RS&2zFcI!0##E`J+K5p9Z9NR!9s)-|1u(90Jlp;C
zKYus^q(DqLL~6>OJL&}6$I*1!#+&tKxbd;m(1pTCj0OrnhWFnz(BCgy0DUKcs1NSn
z$^FlZPPD*W@4~Rg#<R^Aca#F8|7^EVlmH1#<H#!UfBr@ijVwsxFwwcVj%sKxK#2e6
zz5(_3>)JVCYD8H3q9WrjEr5S?6@edBk+>n!LhNxY6CYn^W{_}Wium1YKR0Zq+_e`g
zHPlLf(D#>udf=Q)0QZd%kw^A_B{y1ILJ6srPVHHNqVJlA>OmZ~V@0lFOG#pSZCe}o
znHdW1q9MdF_WonpoNevod2;F8o~7GkDWYWG&=3>f5L9Uw8i&IxEs%VrLe-rTPT`0C
z69i;*wG~I;6X9y#J!(O}`+N~!u=1}$B?iyS?PMUZdHGzc=7_yDJ5EK!X2ny?IRN{n
zXH0Tn_QaPetZ@Sp1D1#P1Jka*%Xnb1#t673B<@eH;Z2hr|8-9*f`P-Clg!%t+i&fU
zgTbgl1+NJloY!6dTzQ8LMR&+1gf+P;M#J7CO?UmAoo-orxVwDXZ|vKN`g>leRbe<q
z1+3H_gOGEVw}AajU1Qu74#jPa3pB6I!?O*J2FZ0b;52NFk4$<mzXo5^f!N6{*LYk0
z2ELtxV{I$IZ&xcFYg~>NRPktT$5r)u?)bi|j&m|f2Hp5A9uudN?U(m!*Vreu>89~o
z4%7Yy9=Td;Tu$}>)huGL0i5;l6qc@|f81I?;E)KlT%`42(sf#KYpV`F4kq{#U^}vO
zK<nZr&bFp22Aog#4jHqvmu9NQT)n~6@u@Vo9zhL=+PYe++0LdBo2Bpumn!!X>xz@S
z(94lH2K0`48%%7L=Z3ZMuH-3z%RJGh*z>2g<wwv*X6A#CZ)M(Y&vD2^qyRro28Q-C
z7C@rY$$l0JuqhH3ul1U2rFUw{oh`<*A5<HknnwZ!(}-WmoPBbtols+@c_(tZz|D3|
zxzqakql1OEB$?P_L`d*RQf1<w?UB@H;Et(1{JHTDnX&-XRd$qdLlJyX5Zx&Qdt8l`
z!yz|PD_EtGny4Rr0B2rws7$VdWe*RuY3SDIk7y@RsPcw|StgNW$5Ji(4F~FlOLqwX
zITQc%qr?3<;JB9j$%bTj!I?DQJd)mI*H3DzIkiH2i}uui>PV?bYwtlMAz##QM6DN?
z`Ue_hXzKmJgaStf|5eX}CahzCS$Qpfc2mUoI#~_iN-!C;_<vigfT80n0n@Q<|Bl!z
z-iO`zGj{Wf#SUr}VLl)~rlEMcEPeH@3ayd{KN)*}<n$lJi(YLve60E-DG-GHmZ+1)
z5B|5|?4N&8k4cuh8r6XK<-BHyZvvn*SJcAsrS7Z}5Cvu}ZJZ;Il90N))`@1>87r`J
zJs<k=4>$Ma_>r*#TJe=&3jQA)Me2-cU#@lco9TM|k{@1M4=*{$)!cpM+MUCk73+mf
zGCb6Qeph^)0)RJ^iXdVWCh5Bz+VokAH1U~&)jGav>B46YHvWSFhQFFE%xq}Rq0x_@
z(*jVTv{@SWU)LjA<rY7vqY3_0+%t*nJi<RDO|=5){*Ty)i|nx{{WTmAwzl}qyJI-D
znKYs@`-y@0!B2!*z`@uMmm~6fuzwgw4(S5$N-GThreh2*MO(8UHJbw)czaPq!u)jR
zR#$NN?hC(qu#G1mM)uk~YU^~h#q!$se!q%u76!#Hy;~>A7WaS7qMDt0O#x2<hHJy4
zdJVxZ7)`GcEuI;+OYKW`Pf3OGmka$L8ir-H<rF!KQIBKWdAxM%X`@pfJtrze-GfY5
zelde&!|uo#aPVZ@5kFU4Z{1^B>PPkJv8Cj?XumG?`8wlha!J&N2f~76l~j^egVxfG
ze9+*?YgZqw243gMV7lD<PI18-o)!FP@Ze`w&H6qPF}c=SbOh1>TFi!6y=I>}KmRhU
zzl+P5-$rL-o&>Ep|C|s7R;qMTCk|l1cB4*my<+&q7nK)krMJUF3rK+#PY;zyreph~
zZ+<g0$!h>qwm_9SI?sZjBmC+3A+TVmKUCfbwob_hR-uVPu8MD&pkSJ02<zM=l-snG
zA!Cj+p*NC9_vERBOqot$lkIA=^qeZ(H)i$ahe=I`QT*D%QT@*2VPNS~>ZFFvrdgIG
zwuX0|<Wr|RgN4^!(Q;0g9(Q}qPY&<l^pk;I9sH7Ciqm6#)7b7RAT>g!dRCs(Fs0yB
z_Nq`SvU;X;LZLABG~RtR@;R{NN^1=rKFpRsjAE?m2k5fT<=wDbe82&C6ugcX%4`ai
z2k4wG+d3afS>S_{W;_7Sj_Tn6WiUs!(a@%>xp7A}pjOM{ozyY6Gcnmi71G8%J1o0T
zH8or7C`u(w13h|q|4sTPy`D-KlAia%Mt$7Q2L|}RDM81K4%(G8v}s!a8z8Cn{bfy6
z`+TL0MLQlN#@jwrB<3f1ve6{(ofziNLjRh9eMJ{v`ycZUi~yw8D-*DJb2(lh2yiFz
z^|;{ISu#qa+dt}Uv8H6WXb}=)JVdz|bQ42wJx;KiPHBv5$8c0Hkz_-6fQvLa%ZIz!
zU8%XBttJ6<Qvv)E(fm)1_*Hf4q1DJ<(NJ<m3}Vo%!^Ad&_@-ExZl%@d<Z`vQuX6<V
z+{~9%cMfs4mymnzBcKBN^XEywzU$E(h+8$fith|3*Y;-!dbex1JvVkmZQV6V7YJ;`
zuMy|<UTCtLAIw3c!5ABxEwdEoSQnl%qanhWvEjdDIQ@nT0eFdu&x)9-;kWK=#q2dk
z(QUS_80IED%tnFYma(PwaburX8tnRtukm1TjS{37fd4Y}7-suuw(w(@PcwDv>mS7U
z{j|$wPDn{F;5#Do{_F95cPojqE6n$NTYq~Q{J*7Sk8c>}pRyK>Oja>&&>e0WXF~@a
zavoP=!$5~}9TU~{Ikm~e(UwCKfo)}Pf5Xk^a{bg06=hx#^#(21GbXlbpV=&52D04z
z3gvatzx?$*>Y$zAdq&-zGO!u6rCK_P9xqKlPe+x&ATpB=T7)jqd6802NTXj*?hy82
z9UW-{>&gRG0JB+NZg*WRsOhVQ1?-l7&3%aF&}$BfxoOW~*L(4HjTF@gV^RLg;aL2#
zD-i3O$xBGuX_mE85wOBi13}fA-~nOV?foh*rU%16cug!ivAQeSpx)<LCCz{{%KW9K
z1ri#gZyy$lK9M%Z7k&+bhE=jtY&!2_H(>Tv^o`l-XTHR_X{h;pyu*v6u0tJ)Knnfh
z8SCO6<b^o=p^FWMfzpjc7A5PHENp_k^VQCeEBG{bVSz%P!&*O76P6wO0xt!?8Z4G$
z%{kMKVt7zsbegb;tE;DCr^{D2%>Gv^sQtS-H97e3KY}y>nFkGr&$XYAg%^s;_<5XU
zfJFMCG)c(mTo@cmUl^;_%}8b6epjReHM`gAIFk$}AJn+=B^JT2_hi_%0R&Z(tNLV^
zig+U5(aQsYtE7Va4@?ZcfDVZH@p#YniWANLmayBcZoRE`UfkJ&_tza5d}~mW5KsbL
zVR`?P>`v&I8!)eqJGxlRte8oH0oYtUDYxI4l*hy*gCipY6&yyZ`oC%&4R(fB-wjN;
z^RLz(h$Z0=r0*#sM(JKWVrKX~Qv#y>T)4wR^OWK$R+k?DW^49KAiSut9NFA^ypG!w
z0vIgy!D`XpA$NJfj}`DX;<-MY-}gBmEe{F?RPy#WvyJ<e3I&*SzrWjd)Rce1b*sjd
z)`x{xY9BC~-J%wLHz`ba-=oa44*v##GuI}k@?V>Xj_5U<?h^<KtlwQ{%sK)?v8v&6
zETG_MD0WzN$q7cj#eEe=UIATwM?|lU2GBLWI`t<z0wS^YHtc`LfA&W}{MYt+(n|7Q
z@t;2tN#p{6yW4k8OHPFIi;DmRBWaKvdCr`j|6=zL?JK`hZEmXP&lDd!kRy)dOYB=X
z;DW4l9uQb_iG8?-@W(-Uop#etct^pCy&-Zx5n2`$FYP7^g9H$l`?;x&4r4TY*6%)Z
z*(Wc!HiZ_7T_8~|)=Ww+^Jk5uf^i2f&q#9aQLNYis^*TXnAxnz)XHKv!!pLn=)0KP
zA8+!s?XpR{+>W+sPcyc3`5ernoVs^6@G5*Tz+4fB9kJx^BcZd%#au4NG#v1r7R2;H
z4Y+`SZ7GS{@@M?>PDj9v9p<;(e8M&+9l*c4#)@>MU^4RT<5B9F<E}?do9|w%kGN?J
zJSK;$j}XG0TWvhXSQT!}Gw|^*cNCLQB+;9m;iPrf|C;}Y=%5hnY$R0l96fdL?VSs^
z!Us&apr=x&U#WestA=g-J4Y3Cx99RY2;o(TW*oz-l3Ck{!f@JbpuI>-BLpT;R8f1N
z1Vm{_6F4@<XIg6Dqg<<R#J4HLy%N7PC9GgB4O||`YL>Tt+U5<XPNMVG9SH_-J4dMl
z?wCyPwwevdHTh}TaYa^wO%v#Esa+9pUmNrxJH2AePiJ9>{H<dKz>-rUhVD^C6=2==
z+4O5YM(+0TarR*vr%_!qEqOI}#56#s;A_EK3c=hOVXxDr<06^gGus0nFVrBZhBA)!
z(=Xt^OtH2(Gb^c&;<PpUfpn+Y=8Q}vj^T^yU0go8t<Xtg5^e3LrtuK}W|2eOm-bIH
zt@utG`gPdNdD5JV_3u`%lPd7X@Y|^W9I6wzVOTvf!n`FoGzeXGY*#<YjcbJ2o1O@J
zLxpJ&tl$cx4wI~hs}LQhCVX-;sS*iL2*2yGcKfCvVQMTnpUed*N5w$!oB)n1b7W~b
z#&#Sy2U3gb?H*3w`g0vwEDzEpEmvm0VB6iS@qvvtvcw-nymfT0bm=L;r(zYJE)FfR
zzF40<5k@vtJ^6A5hzJ(yUO}ADe>-VPK4o;5YC9n`AX!oAroBqw_U2hudReumF-YsI
ztX3cBD>j)YTl0D=lcm$TlK9y7_)Je^iLtNgl9Wkl7pMLkWi;4)*wJ*r&9?T$WYeq$
zi_-!p$p~13y-|pk8+=vAYJS=(dq~yDTr<NMfq83w_wV$Tfh2v^`deW9AN~HnbmU{S
z9i&dz^By^&W4cT}bnx@I-ZE>2^;~>f=*eCeqT6hN&Yx}l`pyP6FbHc9>pmw7jxhKk
zVORrItS_`vc=vWus-{s$(21~-7h@!4wSrScuRbYf0A8-eL0cR9dHbU~<zZD`XRkGX
zuA^Gq5IRwIp~5$zA6SWKn)>a|kaE}T$3rVDb#+3aPF;*+eT?y+F`V3of!4d>i|GIu
zm&)Qo@as8%XxF6CLX!(zb5|eMF!H8iE(5{G-c>;^cZZ5Hxaew?%UE;&L`_O)Gptkb
zS+8%wL+diuum9`qikGk=k}%aLiX8}<ukyb9hTAPxPXMF4YT@;`*_xmG;8tO!KX5f<
zwqiI2Y(vmYl#QRgWrR$fb(pbi1HebUYm7Z!yIuVc?>bU7Nawqw`f9d^h!9$t1NEUB
z3MI_sF<I}}aPMn?Npp*>-_?FXa+&rBVGj43H)Ymn3R$e*fwEPpLO+81lIq#pE|@Q0
zv5_9jdFWS%?yLEp%jx^CQ6tXbuPl}Y+6|9bmY1E?#wq6af-koRQ*~HuUy*N$+%#81
zAvD_KfTb{FP{>&SuSG(P9O1o=PXEK&Gw%R}ADvIemH>3aufuP#cM~h#V5RR^77v7R
zpnu+Yi64Z<l<{hPBVGU1s&-Q#XFb*OJhgG~Wt`BUA#y@L2dgZs_?i(Sg*h4)E=Xr}
z)(HI=`uJ!JU}Uyefw@UqSA<<o=>R#vH$*5yK`t4A*8<~-Oi8ZU(&4^6*V-xd-G+L$
zUV3%5EA=CUslWTLOH0HC3**SBUOT4I$-oz<ItbhFo)LJq#h=qKx92L|yx%yJ(UE?N
zZ@M9p==T31>#L)pYP+{d0TC$?0qIg{kdPWu1WA#Sk`n1|7(lu~N*V>Jp`~kRq>=8S
zyPFvpzQd!>`@a6ZwdRju&El+c&b{xwuWRpX-#gu6a2}o3y{TE0I<>*ItTmw-3LH{W
zjRqXQ5YMLD`rx$c)#BASaJGWbcdc0W@uo%}6117B{c*J~45jN&#U>e-BVd`<c(Th(
zHyHO7Kes0?PC2aa=Dz=Tv`Q-mY6+(uNdxxZaSKRJEMqL#l^AWa?0Y=IaL0yEhv`=W
zhirP$EJ<Ft@cAFQQO*Nm>RHF82jY*27*Ybnqk|?Ih3f5BlT%-K=&|cGtmO7?a4nnM
zLM<fFdu@DjfU2HxKjR#mldwSKX`$15g1ZOmLD9I{weF2I6r}TrV8~t%a;l-f0Ga2c
z3Ls~4e#==)KyTmdG@7TU#ZSA}BqFTT_P+3=ty8hj7BKks89;-w>aDI!!zLxiC!;VS
z?h&GK_*(${f<=_`2Ye(W7VV6Vo=Q_l6>c+kuYRin371IMnJT(A<5}x*BHiZ`q_J))
zySw^2<=ecWbk8|`=f<E7aqf$X7yFj;<#Ppc={u)EFUr{i{}mUx%KD&z^#A?qKiQlE
z9~9x^!!q3S|Ct!chJ>CB^An?Qn3kL`REX~piMTs%PpVn)X_|O^)!W4{oO{_1BEh~K
z2;}6^av;4WRB{o5{glaoC@hmW<{J?(zFI*S**Kig98t4Mtb?5wLythJJ|M<TZ?AR{
zC`(Br>cHK8=xS4KhNyk-<U1EFT<x}fcJ?gD-dx)ftZ_I;8+}e8(6{(VL+Et-!dCiA
zy07HRce>A>@RbY`sOQi3TA;>0q^?T5sdjnjWB=fsIwRLjh3*v*w*SKC1#8Sui&u6Z
zzdm4_e#p{+>pdG<uPXdN!ZtK+X-Ldg7MnqQ34XM5gBaN98`#^OulH~VZriq+2m0%J
zXLWbZ<%2gmzZCeRrRCc%!7Zj3#MDz~%m>?*43XD3J5-oo9@K58EB~zCZ9Eu;^z7;T
z<cd2_bz9C`9+zqDXDA6zGmSmX&DdVz<96$L>$Fx&I+HDUb&gdlKu2L+TAXsP4$&0!
zhS{^z`zWbyuTw~8u74E=<IW!ge)1g6@t5jH4gY0egvfuE2>9|u6HT)Sm@dh3NLbw}
zdOKSZ+g%#E4(@_4Ma|407`(ndq!Z+e2moCg^x`|wZG#az`SY&FRsr!t#2q%*k@fqz
zXGGM8wn`(DLz(lQ2rsA8%C9zHea!%6&(X5nnvt5VBPb*^^llmwQJc$xT}cqHB(m&T
zSy`h+wsxt=!NM|9Y{BW~Q#$Ay1VKNe_?|#=gFGA?uM>;3S<mn)&@;gIB{e)%v={Z*
ztBQT+_T9jIwkxBc>t$3oJo}T^H`KW_w*l$o?ka7YuXQf_#q&3|sddO4A&2Dfi!t?-
z?f7GXo3g>CJV3V)T<=oSeJDR&JXvd+{;kvNxn4cQHzj@7WnO0Y>4fW?Qy}Cp!?QTu
z3u(e^Yg7p;_Vx&brn`Tew#bllBJ~zMwLD!`0ztew6cp9ezPrPdNXN{M+{zYrc#unH
z;8MQFhmIWz)$>Il{j3{abzWZHkPYwfclIHuTOxK)ha<CfL5<fJ;ay(WMn9I?@P%Hb
zA<lapUkKM8ASCZAr&yJL^;Aq$Q_V$YixTjQC+dKUww;4^tJIy*y2;4?teP(x(53}N
zVMh!4yY#}KEP(N>r$<dv)o96wW@84=@l+0+wOL>O^B7&qJFZu4Rt9E80`Iv~L4Q8k
zQTxnyD4g`Fc?Qpkwn87=?Y7lL?LE#?Lwls})~~>g(eQ1teL2z=(h}ovXfh#K_Sns$
zuqqUS0y<v5+BH5+92xjkwX7(tD{{CYGR+pv<~aEkOHlumAiXH+LZ8X|L)ePfW6n&d
z`$9U!m^yS`b)w^$-L~jYzB~M)SkyJu;WyOdE0KP8*uN_ZoMH3lqw;gxX>c=BdK)+Y
zRO<~EE9@M(H{k(+o>NtNysVAneE(*#NytWdX1}f^y|g(yG)?mm(Pe(=v0tNS5a8(%
z$i@$oN^=>{-zq%ryXz6W(G+cs2n>eaG|^je+qXM`51O<`**zX-?kWjihPSd%9aJ<k
zlBQGgZ=aOOP%~w8$%(|Ew*gjAqAwdji`2K>(~m9x_(vA*tH?a$45Jddek<_*jr;m2
z;>&0b=bOQvJp#sR*HiA=%cvjIe}o|<O_p`tv|L2KJoxD(Ii~}+cDB{??PV9$4$JRl
z<8S}*GHXmtZL2D!_zl5k5!4xqp?(a4FYIpF<lb~Afi6#Qo%)@sH=biS6sAXF1#gT~
z_G<4pa46vjlRvcwINE5$gkC|$!(ihI?y}A;NI~vPSS}wCKkK3G5E81w5$+|$SEt)*
zq868ud%tk84~mpsH1j73#H(cT@<#!$eqYZYQF_rbRYoMtw$nIa&5nD;uj5Ub7%AhK
zC{2)dtxQFck+HtStzv|JrXGxiy_7}W^qPr}n?Jk5ErPrj;=X$MYtMZOKGkW`Te0HW
zrxH{^fwT*efF1e_E=ZH6<`00@D_qqXBDN7bs`%y;$}!<IcB6~X*pNc)w^3AcvE?L8
z)sdo01pSz?da@Z_!oTb)cb?jH2bgX$-y@><_OGy7FiA{n_gh_#A{BZ)Q1lKa%u65v
zbv$0{*Ut>nX9PXO<ys)r15H+l%Bja*Vd-bT;^*V5mFf35l}FkwRCfscsW~}1p~2b7
z>k~T%r{=Yb=0?SPX8o)vo|q=L&XG<jm`{hlFh6MSh#59Gd(j1%*5KAyI}_c0ruM@L
zQA1R_+?zWV77MGE{CW9N_fV7i`rYLD?8VCPP?g_=T2S%m**fO<j}L={DysSEBJMBM
zhY`!42I17+a9M>6QOWx6WaWMD$BMN1p|9%O7khH_bkN2eA=V=#D<dny2aag|TdP6r
zDI0r?GdTUJyxI!MCyiSZB^ei%`p@h}<}6Qfx$ukQnUpS7bVmBdz#Ftk7j?}V+plof
ztM#oJ9O6@g1GhE|1wh(mPemj;lkrm(N8d|Cygvn`zxXtKOkG!&$u6O<7EhZ1!7iVA
z(MZmjnIV6`ngQQ$Pgsw^xEs*Ck`K*8y>@50is2vevX8@NDwm*m8CHRW;dw272^wo)
z8E`bC*sL0{dAlFW!6`dY+;4?go^^XE^D){(-0elDe$LI>7Ek#)-Q1SLQC(DSBkN1q
zOk?Z^)uEc)i6P8#f7cv`R51AT&85OH!1Z!4Ds67PlQ}9>G;E!-=cJ`@kIfdp6G&~j
zRZ>QUdmY;}^$;|p74jXm56)iY{u)AREFwSozC4QnZ5H*CV!z5*{=(Ae4NA#);K0RB
zwR2zL?<2m%&^8B&KjZyfZ$!a~qw$T0vu2fkZpuQ3OTq5_Rp5_i0b1^AKO4UID=^ad
zex^m?e?bq&-!46t*56%B&TUilq5Mz(6+j*w?*jjr_9-pxKL<MfW)u$ueN_ViF%QdS
znLV`^*IzV4tO;Yo52SC%`tSYzzLhkw_dQY24ZYl+z^{Dwbbwh=AM<S`-&w4e?P9-I
z6S-P~B9{et!JQbV$E`^PU-Qe<DBs4#LjeC%zzfG4vl?Mjg4mXG|M$7$o&)#B=ymCK
z^uNy<Adm758MG{_ran9Pc-wTkVgUO`?TR41dM~TQYY2{*PcAmhJlyhsWKOR#DS<+c
z4OX7z?}8Q$2o0u?T$t{0$XkgRm?>aPl-CP^cWA&0y3$?Mt;&B1n)^A)<WY&#txw2G
zY?ci+eq-_iY2i^d#7*1(z*EbZZzPf<$=OU0Lgs^$_$1kcj7z7Ta;-)Uz1&IH#Eu`0
zv6Mi+gRi;^mFG=MbnS~TyG`};qd=EB@!j4+menRIx?QPCpD|xYiXFFtA9Q0ahY#5l
zJ%F>}Q2oVPGA)2Pez3|Cj={*=LfqZ-<{MG?+d21jKlfB(C_+5VZMv+Wu1?TsHcD7j
zT_Fh`+iEwE>h^AQ;K>TvyqitxGY$^b?INC8z+0Exoa^#>oIO7Mbul>BNl{4I-^X>@
zTd`SCTrAo(TuainBF)!=oWY45t%-{Gf~7&}5jFd?IxyuaX4APar=UfCQT425<LN+<
z`B`eMR_|p}*JHcrt$IyrSoQ~;bCw&s`gd|nj(kLC+r^E6?#H|MW9K0GU#n+SwLhM*
zu!N1uzV<vl#W<bzp8qvPA6O#i3&MM_D=oQzu2n`&y0g1bB$V#Elwev=xEEEjbH1&z
zQa$SWSAE#G__h!~(O4zSb8F*I?xN9;>!`8u8=LXk4da`y6;5pxBdAn^sO$8Ncc=mF
z>cew`9X-{oDwuQFtEP;P=H_oSbR1HW59ZFNbyQToj#HJ@=}UF352TcIcro<!_Kq!~
z!%gG5J8b*$T=L0yKCHfZzRryfKUZB(wUaQv3^MuL7TDSO(rFx4Qr_JVIb&HV>%V$+
ziF-nfH$~p!gf=rhEj<+~eYCaB!z?BmuXa9?GdNVTnn20AIuQ&!zkCLls?xz*HMN2k
zzXUm!mma=^&FhYRJRB@ni1v28bir;XwA^OJ?{h)140~<YihTMAf=jNc^)LjVVhnVS
zgG}guF)xj;A|n8!R1beR{ND1xo-)Wq6sNra4U1ooSVlmXJFhF@1uQQFqC^hE!}hDs
zt;<`XJ>QoJEBGxZ%M}-(yTw7HI9-pwsp}Yg%-%>`I_e4(M^CKy^+|y5!&JlJF-KE3
z!Qy(ivGi++Dz%F*CmRoiWy{_#N=$evDDacYU=Xv*es%suhpEEP7JhT`I`MQa`{}!@
z>t8bhj(tBkEE0vPh8DzM7dVZ-lN7o5sd{vIA(v88w;~^a^Ej}T*+-VDX0hBLk+M3c
z*~Y-NtEMJu3~mvo<W+ADV<8WrgC&%;ir$?RkVay&GyPSy%zO<@$AWzaN{NBDyCT#N
z1>*`p$Dai?SEjC{ot(DT^>gaWIiN@S0f~4sR`s>5swm(`wKHetYkgB&*Z83EHSby1
zYUk_alHnj_>o)w@+yd2snx9NRsLshMf4tK_LkSKl=#g_8Leej<H4LAN9@G@P*4(wG
z@D-ww*PP@uPHLIPf-0<g=(u&PszNLg5QVY5`m1;AFvQ@mX*PT!ik{jV#29}irM1fd
zXj@5;%DY}>V0G8!h7`bStpSzh4Z`sTZ5K*VW!KBasdfF+Soc;eH1DiJ6w{9$XkVWD
z*0cJP1hM7vX>uI%%%3%uOJ|7b%yt!Adj`10b@kfd(s@^n-5?)5`p>Clzb)g5C~&q;
z*nYBI`>nrJKEVLP-IBEpG90<b8x`5zY^Ody0J|*cO%Ea_VDh6Z-HUSTeIt>o1WSaa
zW+l{DYFkYrRQffXQ@jdp(zR#w)EpT~aAZ)r^0SeY2y3(@q;w+ekzK?OWL=w@r?eOQ
zHQod0d0T=$U*fmqIRy2W5Tcs)Lw@nV$@_^cA+#Og<x$~ODNc0sWg_G=@hIsZ-In{$
z9jRUE{L#GYmj?0=Etim!>-xmMz=xwaD5Iy>$KFpwud%EW_oi8MQOgMa8wYNKkC-?x
zI#xeAF~E-wAb-BYv_n5k@yDS0u%9?a=<1f>FcebGFb=BID3Hs1!-wnnwhjQ)QDzPh
zPJjql7n!N|bnj2wC}Eb(dOv#~qKSxUdDY4xWV=9gxZ|vpbrV(+9E#XDIrJ8p`PRh6
z6%A>Qy2js8#niScQ^^MDWG&y=Tepwvpiw+Eiq)ft1i@Y0ow$ZwhtVjm>2z!d^xLJC
z9u*V}&v6HQe?;%Q#IR6O(?$0F6#Pz1>teQxWv0io<}8Vh7!|v!!D%^moNxDk0~+wX
zLk93|hq^qGUv5ph?t!o0*_>oiZRfKXhgP{IEGqDunX_3dp~}t9ZRsNr)lF^P<{1`U
zRW^otHMPfzO{Ktmk#^-f@3h6bNx!H_yN)u{eFNxB*Z1X^z`#4Fxg+dDEK|Qa@4hJX
z?D~*6E~4Df=<42fI$`eY2SZT9HH=q9nK%l)fphrXyHe!k`9nZgFC5kx*Gzbt9@i|3
zXKa@MCAe=dRt>mK#?;i6GrW+M)1D)zV{+*!G_VZtzMja;n2eBTtC6Y!hR%fbT;Dv}
z7x_y?J|aECLsaw4me<mrSAH*;Xl%X82#=yo?nobvp!DgNiYcTY#U?YRyuy#Wrn<vc
z=|4aDLJ!>XDWE`6=MAk7CKGfCHHBhgeB9jOU6N~>rCCJ)A8GIm14XI#r?mIzqzAHh
z7knZ<bM1As>k<z@5_PH*?%Z7Uu3mXCrh=HZz*n230kWcVxd-ykg1EAEjZ$YcV4lpL
z^^le~P?vqZ{`+L9KwOtjzncuX)6GEb?{t;&F={B)g6gDb)U<Bf@{`EXP{L{3pxV{0
zUF9UxUk2$YdW)O9@dVKRiAH8<vC^Eo^I0DLEiSQ4_0QWIN}$&R-wDOjblSLI<Bz87
z+<#bBC3C(<x1%t+R9*y~<4f7z@DG7OSV|goyIiKhES1a7+M5DbpS8ItMHk9KY#bJY
zx<Lg1_!!pVO2dyChEdSu4<n&dYPz<Yrv(xi9mGi2yt@4s=8B<vzK2&lN^%9C4cY?h
zOw(PY14U8{;3r92kP_w%Iv^tFSOR1H+<)CXOtFrhY_B=YU=%g^C0TsxK0IEi<skKC
zH($^75>$)~Px_(r_)K7X%-ArqsMW%zA&bhRRTz~^hv`dIj0l!>(=IAuT^nA&;mOZM
zC`_rSqE$J7nXHsl+*5Mds%ly)V%96D-M|aE%D+|YL9LLgsRiC3a9Y_vj|#rX%5a~r
z?H2c2Pq+KF1p|53VeUmE_sc0N@*=O))X8%qMYgJgs|0542D8Wz)bEd0!yPy#tgI_q
zrs4UdQ1sH~5l$YHRyrz9rkbuE>F4L$VEUWRk|w0$xeZ=MwIn_=g5?tw{e1SOCizJ$
z$6@LC!Psa^`Hi$E9UVC0&<5VAz293y^<`zoJ7ewy7UVi}q3yV>i_Zj3C_C@6!54#m
zuz#Y*pHV@wsM3Xfu~|twdxOl3*~Dyk?9Gh=xgh&12mzav3NQfO{p6$Y+aEIXu!S#A
z!nC>&pDpvkG%Q*}>4xaUXd3m~?I;q6^iO*4lNH7iBo;;MlU;1ebOLYIBCL0<ZVd;;
z^e30$HF9PNubgQ1%*1Lc^1#D8yN2SOrx>Vmg$P|orNXYYd%f0CTgGz_8d|$9uJ2{M
zdTjG@FK&{}|ENO2xl8!C{9zD)pR(^DdDjB&{eZTD=y{a<pia$>S|s3Wg>wN7sj8=`
zZIw%W*q=@*&xH;9W9T2LscGcvy38^edhc~iysa8Mt0rd32ZFwOn*A{0fb^Iix5`0F
z(DEEZn+W$;+g7fi8fX3M!5PKbu9cn^I`(S5?7gqi!<=WuHJeokRr~nX?9T5Ruy06s
zqLKYPHCl=eVo8)q=l+Y2aogyzs)_nukmV)rPuNV0%4<?dS4SFShzZwZd39YzV%PDx
zI9Mlu=zo0!Z6OvL@kQUb^k68wdvtziy0wE_tnMwe8dc<OPs`q5svV!fm+;)9f$fB(
z`uo`Tv|D%i-%q9hfb%&bx7$^IvzZE1RP2E!(6!gg-KBS@a}$pHdRe=rQzSbCn`NJc
zU<Y0F_s7}IuG}?gz2^GQFLFLCLrUrlf0#_y`OG<z?6$JW3X8q<T0N~^`jFPCdar|s
zhLgUq{*2sw!@f6<1Dc#EO2n^>{Y9#533<0IL6u(0vy+PO2E5D0FMB-PYa@jLu5oB#
z2*m4U6?tXzZbnj6V;OY)M*PvsdROv=v73ghvsou0-3Kf~i|kv|DSP8)o@dHHR(4%Z
z3(=0seIA=D_^|hvNUSyM>h6TLO~JU1m&s{y@RX)>l#gt-zExRS<TB19iiK}$f=SqA
zsO<g=XLmCqlva80zEYxEJJL+vTwb)rfwVqsU&>h4;2q8@x)zm9_iW(@?QEyd`ygnr
zkI%aA3ag<`D89z0%IiYFy<uZcTn6?NUQyEO*NAKRbVJc7*VgyEm(_3e0lg)7SK(ah
z4&U+}8zHjbx~vS(d0sBZDU(sECHBIX#lJ}P!I1Q5oF^W5TsXnG;CJK;TR#JJ+FgQA
zsuZgtT(9|-W3l<7Wd09DU1A1s>_rLf;6E!Fn%sPgr3^ZXsTZL0bv0A!I=Pk&i-O=P
zJ;tk_%SQB7Iz+Y_s47IQTOXDtyiEpg5ITb(FNFnrBOMyeZ@ZQ45kXKvfkdR>x4ui$
z>!-QaYkf(%?6pTZsXfyY1QN8I^uotikh>E-_yhvl_LK|io$%!InfmL!pgD}br_7X3
z=}IynZ(&Vg`AVV)Zgb15QLx8;4v6@p^)?&73cd*6%SNW9dOSM~Uix;nd;X>vg?GD%
zX|9buv2CALP8xo&!4PKr)StdPf_1?PI&0_Y*|R#>G}V9MOYNkg<@Tku>E78U)(+IC
zJ52_d>PaUcwV}ZGj<iFYvJ?wVZ^X3E5$%TN@HMpKLsD^K`lI6p3UvQ79wm|S+DFrt
zv8N`mkV9^=qf)AoTY6<3v4CV^^O-Wym|IdGTNaKYNPLBa{@|Z?AM|44>~`MDh#Eyu
zXZD{OD+Ti!+b;!b?~Um$?B6-tDV?C5-#?qN{^~sArfoWJc#Jo?X{x(5uxWAFYl8{4
zY~QV?5ky7>JwY(sXR_qPIl71)+^ot$eyQ)}oN$vf&a8oTKU_*VT(Wt`9*b6`Ab)Q~
zb$)#L-@MnKEtzQxD7QIgE!tvkQx0OX1kIhPb{QjH9UP5#1j9?0(4Q~!E-`9I#8Z0N
z>MPDHkbTD`d1)@Qg2V0dhd4IqJ(9xyA4Ak;H2Wjsp#hI*Lf475j63X`3~s!fH3%|k
z--B+3x6N7XMn)b&I%(aRF=2@>)FZpR4w%Qxa4Fi&<u5J_+D^wqv9AtErY9}QK9x9K
z9*roCq?R@S)F}TL>HHa{!w{*ERn%;wVTiyyo3bE1{neGRt+xFoyftvxbqgJJ>HKRt
zf*h*n(FH}Y&r&QC97a*YFdiw1H5WysJ%F#x$Uak8=o+C;)N}ixofH+3xER<&dTu95
zFlROCaK4*XV`KH@K{GM-Bq_?`3=RwewnzzisC`GKVmk~`YM^w#X0AN;scOfd_qz-D
zc1R4fT`8BVr^7)--Q{F1-!d~OJtORWlq`Dj!r=O|Qr5pP$OBR!1a&wy3cS1BQXBRd
zRGyCE?mZ5spNQK5nNm7A;<K}n;fTOMI%Vi#lgQJi_bn2;>KKszGC@+&{&Q@dZx^dY
zEL4*~$pf9&Jc{xpr~HdfuoCCRsi4sP35Hz~_wJGVh)|YCmh53DiHrv_Hs~077i>Y)
z=V9znl@k9wdTXP|V?J;ipI4yeeH{Zx%`(YNje^1zd!0dn)5s?Y=5+fe77Ka}9ml|k
zs!h8o6D-q;iCiTPQ$GRWOm9$P6=5pN06uMhqL-wkU+c>vO*~P3(J<irO=Z>GKz1!l
zZ6w!TL02f^$q%TonGaY<P%OII+J5Lfv@Jp5F&a~=gfHm9`f|WxLRVc!E5rORZvjrp
zS(!8&lVb~t1#;8H)^{n6U4zhz&JJez)NEb07m@^IbdbH+R-&+Tmo-39SS!HS)am3C
zMuSl^+S>kr@1!FZ)V6Z98w%FQ4<I}~?+|`EqC+fXRUJT%xUz7=IXjf~U4y?D{c9Pb
z#q0n^s;@H<mc@9>P1t2nT66dB(t|${pcWB48ti3^P*5(i-3fC8>&l~DG+nA7`&~re
zYFk%*)h~R%bRaHjHGK~gbUrycdZ)(O_X$kr5ffz40nb+9c!6GNGk)tPw*OT;^MT|-
zMJzR>xZ#s?hF>@7MRE`}UNwiy9OBD`K?%XcGZ|_1o8{!0?a$8>EZt;RwT&!TzK&C)
z&koLd-7H(<V}Upl1$0gep?lb;dx&Kny9?i_dWe0}?&G8doL$$YuRtQHtP4m@IG)Ka
z<uJ1uBlvB$#dOY@j~A9<@XKnGr7oN@S^5-My^9h>#A8T70i7AhZYKt+rV&cPgwVQD
zQNJ)cHg4eD^lK!nn)%>Rtx~OD{ryl!eR{o~4rVig^sw5z(2V+@aD9uL4sIiMgGDRh
zKOQ2Hc?hbn&ug62r#5odZW{8+GC%~oV`{cFv#6r<qe`0l;p$Dz4yAYF_b@~Z2Q+<w
z+36}jSN{z~SWOy}@rSpsxMpp)y~;l|3V*hktCI?MW8geQF`%YsoJNUpzg!KRG#)on
zQv+Aukf>`a)(5xL<jO0Y@rxQ>6byhwnPk+pO{5#Qri$U6IX|$1do!|868W(8LoFD3
zJEsr)KUvY|ThmUL8d4)KcWWoPmdM9O9aT?H@m{2<8X5T&6-14g45KJvXnG@K`kWOn
zy!iNdbI!(}C^<|L+{|iwTm?rsQ-iO}D);krm$hP^r$0LEGT^y}l3XK9>ekCvk3<aH
zRvm7fzqP(>y7AgZKj+8U6+5|6W@<PPluH-YG~zub8>q&Qi$b&q(NI@cX6ud~c6epa
z6zEo{VH+%d!Yr^>+*5KSY5nrz-;(J^9N+myAtcT~LU^nb;FJY}#K6!4A=B`k_<z<R
zGmNmgN26i5mBOYem&)}*Hm$MuKJ?9j*8JxKIK{p|49O?E8)SDZ>J1_4;GLSqW;xy9
z9;I+M@m&#n)0DAq^!{S?Xf@J!?a#)-e%`;?z4=^riE4P{7bUuy@@84H#Rl_aR6_5L
zdZac#SCf)sNiaD(c<6@g5zQJ<A_`;oxcn+PD(b=L;7<{kg%46rY`h3fdBA5XEwF`;
z^^y9jdhRs(p-N@ht9e_m!kQ?Yf|z=%5_Nqm>D^<>T%WOP(=>0B@gHP2n~Clt@1O+!
zJIv|s;=7D{S%QN1fCFGGn|t8X)9>sIM*5Ry36<i3)Xgh6!@PQmO}2Rl<Na9!+6H82
zz4>>Ngs7T{7E^Q&)5$Sp*JH49N^wH<a-Xx*P5oX?1lac`aVd=Y6LfzVmx5N`_AKKE
z?}#OpzY9y}`jQ5W@1jLudD=9M5Td2Jvax?yIuLvUt&NeB|0`vVg7bg`aMsdJwE~*|
z5Jqb>fVxy21H;(mn=Z^h^@7*0qbe5m?mZ>Ecm=aR9{PS?>eMvddn|rli|OuXH1&ON
zI*M+(=H1SWNRYg)*Ib(UQ5-c%XMK0z_A|?iWTi)Rt1fe9DX9gl91wp89L-uh1H#tY
zQ{24?zemXr$tlRw7Q(r0zmDkm^wM3t$zWmsT!<Ye$jjS(E%q}ojK)%POX*{C2mVzj
zb5Vn}^}UW{km%$A$ui6+QGYu^=i0%S<I)lpytZG%JS$9=LIEWbE;>Ad+>GtArAP}8
zjytov;Yv3|VHKu#;ie$z7GQmX7Da>{IadzUOvFNFsId<ws~1vipsJcJQMLfQ-14?K
z=(a&u|78>}+^PYS-?8<DMWV}ic9vc8>Fs+Rl+_*2%Q*7+d+!ttWy!ceUG)wtr86_=
znobF6rrY8br_B*MUT5B%d#ZEldgxEfTY1Y>?C~_7Kb%rqw*B8E`gAhTx>Ci|bM_mO
zyuoRJXH+{7ZC~FheKW4lw!MC7SJ-fl39bn#E7nb*5>{qgFA}RdE#<yQnAxp!_+B;R
z?9kW)kM?_+kIT`hM_PZDQ$MD!B>M7nlhV4*d^?viF|KSUdg=a?B1MO@+dO!(z-G{>
zp<lwgH`gCl6v@^@atW<#Rz>;CK7~a4ol_@GNA`>9t7}&BJx+`ryD~#H<ih|{IL}(M
zRQ5({&lq<<?q!iklwa7#vQ!U?$odiBEqaul(%#ZJ1f6nV=Cv4q>>^^_xGnB<w%Jv^
znJw$&6i+dn$lY@1u(7bPZ;C+6`seTf-iSy!6eYc{`K}8^>0KC%80~Tjn7M~^U1`~x
z#Vz?2EAiz~E3_J#W?xv-$-Ue%!gS_v(XKk!d#73w(AUp$k`n7{=zHgurzjX17O^3B
zBfR!Vl4yWqtGo{~zmMuGUP$^^Lo=|L?ku|J^P(CyK|^mO3D>c@1M6f_S4+uZ*6As0
z8nEI$8Hy(vRXw15{OSL)wBI;C2w0b;TJ^_IZ)52L5#Pvn9e8#`)`pg^-Ze%=cyK`B
z>ijq;HyblnjkU$x9~VBGr|4T1=)un^+&hj)CT4RKZ+cZi*kvuLcfDqvw-v^AT~gHM
z2)xd3`itF<4QfhTYUMBXJe;P+s-LVxd>;EH;&Gy^1UldUx#dw9jvTQITCEs%Z6}8m
zXz1a4T}igj8cpoRaah9Fui94fr<E?pwXF+vv1##BVp>T?qD>^k$4g>g&nga2mfMaP
zO6vGjoehW<KivzR{|rn$BsInl*3VIx*fU*V$0VjXbU5n5oe<Wfv~Cke?0(HcgilL@
zR8R>$JHxfyKch@!6@7x5Q+cI=&%T=%Ekev-N8@P^nSgZ2n;%MOIs|FVtykps#bNjx
z%m`-rV0p2k@?lP1&j__-N~*IVfvchGQHAQclHR4+J+}orr^)@ZUz`2yA`{*)U<l*2
zkDadj7#p0Y^4kus@8C7XjTc8b4uAS1v#`?nX^Xtv3|?=^1->o&ON!E8cPuvJPhIQQ
zyScvN7F5a<7H1&Mtf0te;F|50yI0RuMEk9~eCoijQZNpBIsZ@lrE)6J{YJX(w48Ul
z;WD4#q}uk;YIP`hyxDI$k8$53`_V&&ESb=^sp*#!Q4bZfX~Vifv|9z^RFm6%C4_jD
z)L&9n%aJdghIAqIFo1{Zvo0sP(6z(IIHv$=C(0Un9zlY8c`$Q+#wzL?O%08Gbwb3M
z=aTLB*oO=EO?vy8pJDGq3WeoK+hWny)7(Cjv$3QU;gSmPqUfI;mR?tBZ;YJ?Oz5Dg
zD=*95vBbtjC8hK7Y%{<f`-0&Hw2IX*V?+wUUCf8^RZmq(35-jwx~#TJCC~ajh=*ll
zKv_))aJf;a-Eus`mXSnmG#b;3dD0jvI)O(GH;Wg+plgyqw%n$NE<A@Q6T$!t%i}R(
z6ilxX+n9solGh^6?&=V5+Lg`=wjDcDe`jHCD5}~Dg&*6@fm%rL&sv-By0IJ>*<;-(
zvP0L@z4Qb?^>y<$L`r>=A`d)0-aRI!bC$p}8CtVAN}4wOfp;Q@B)$RCePZc(El}wL
zDexQUmnR1qPhM?vkP0EKteQ6A?lvfK84MZH4NIur`qNuaN$R;_#SLAAC7$XqU``?A
zrU>L64$vECaV%lh&$}6R)372|3Q^LNOMBZt>FA+(3|pIi-t0N0-eWU{i2+TZ?wBcp
z?Qj0dqhf&Xk6Ue~N1-08y~qc8FmoXRH6$<6%94Jn8num83WcORV!12X;Q8jL@R4=O
z3q2;PCDDE^1KXpJAjoSe)-E}NLS@j+lEqk=43bafZNPzRK=9Nq*%h95S@&jKJ_xeU
zq@a%bkacv4zcc!w=tJm%Eh%~#7)PtSBnll4Y4Sp37mEh^JjS22oCs|e5Jj#y#L#JW
zdyN`k1=Dz>>6zUpya3XCswtKg#nak+(;cxywHWeLn$;{%qQ?5q1P!Hm+mYl`E0wZH
z#M@3|Nw>bD@XD^J)p`7cZ06VYN!W)o2XjYa3!=LmGAzQnzGL%znf+F)-2@k-H#4<~
z+CPLuxz^nHQSXR*=k8Y9y9K{DdI#*WYeIs(rW>wZQ90&!m*bM2Nz0p&ruwc5Ve4l<
zjFV>84)-xW2{Dl`v1i$Ww@ZX{Y~**9YLe=vylM8HD$y_eFp@qAqN2!B*LK3ZOg)?M
z<~guAT<l68SK^v1o1{;F_~@xp5i47<<JE$GMz%$GIXLNp7v?frY+k9R{uU^*KtR3V
znK=vd(RW8l?NHfm#h-&VX0ye8)pVfgvO`PoCGEyC4x#q=S;pXsPtizH*Yh-Jp+{7^
z0Wb0~5+TPbM*oM;#69xefz^CAc#ah2<;p7bxJY3wLx}*415+}y&{;*d>XaLkNBO=#
zhFpl4lbSV}=2z|{<`g`-w`u{RX}D&Ehht5AMNvnjgr8{J*|I4pQc_{7dLs3t!)A_~
zW$#xH(D6a}gM<_n*3Ya3?ePd2I2yM;=sq?)VkAw?q2ZI5ZMrz3pik*H>_8GiVft6@
z92wJBj}Mc;-SPY^7u(f~E5uQ@#Av41*FE<R2NNtd%k-fgqi)_p-D+KrtZPoeye{*?
zjY+vAle)f|80&Z~wYBznRi9H4hvi#TsF0qLy1o09Fzb~5vArc#5LLSOrqgK&7e{;l
zGzgJHQ3Ao#D?G_ygcOw!!Zw`I)!qu6D#_G#6ZYawER^Hk6wI{6ViD{UmXcGT@%g_M
zZajdn@xQTM|Ky#*oa&j*9(&?E6_D#v!1Dd#ODP&vMx)#zKeVRMCo1gc;n8JIOBQW!
zyQ0G`qe9Js4eXn=6Wr!KAMEIulvQ~ocW~6iquamT2kA8j>s^CenIVCGw-vLPB=aVD
za|0p_Vl}J&=Vq1*;oI2kuUTnh?{m`*r_;{dsnv`CDeH}6L80$p)T$n#2zZwykJ7-T
z2ZS;<-!CpQ$;AwP^$Iv(mCq=WT$s~y>-~Hqf8J%au33tGSHa>oq-o;Ox#GZ2?0*e_
z#<yG5TLm3{>Wpj1@QvndyMtVO_B|ow5~h;uc0pCJm@=q4{nDMvC1*>N+ct{KtS#Z=
z%s*<H=5tfl*5=7W5v2^APF6DvkBY{wk?{d~oH!!iC3V8rn{tObzdeXOoexQ;W{B|1
zE7KC7fWWqwMdumLL2vzGr{Dl@Li=^elR>8U-}C@U(5RWWs2l<r2%)k(s6AQOkTZz$
zQ@*S5ya=ay?()4cT5DAR*J!Vt>jksY-HWBLu|`{If6_+_VS9itd3xd4wdIw{FV*9!
zY4Uw_Ij7L9M%A}DO0>nL{XlK3j;>j9;k3<gsSYEB_@Bb<*so5C;^lU)I3dX+8u`S;
zc*K~_<8^}oZ__wzcn5DOF)EnKv(qJ3?K<E=sr}=WiYkqo)kyEH?U~^-rRcE>me5`r
z(87=9ddSi7m4MV9)$8h#q22Q!0j8)@nU^wHo_kGVP0c7M%viEg;%^1Z^$S<Jpqtrh
zmjovxQeL(#7%1tkPhG2p-E`Uucw?E}wng1F*gP-I)P9@}xK(t1`%lR*p}Ca{pE(rq
zv*ewFu68((PxVE+&7n4P5CqBGs+GRKLNP3xJK<q}yxwXAP^=0TN+_#TZ`O@%X5H>y
z&}igR6*Idz++_JMUBu(()-gTFY+!SL`2OB<l+5G8t*xz!qDr)dZf44_Ug^vt*Prmz
zx<nB+NwypLH0f<J7w-uLvF_{8jqM9``c0Hn&&!v5PGeI-d*5+-LdJ9Y>|=YGcpjd>
zpg}JOB=$D1$14;05g$AmB-b=Z`{v0jnj06aNTM4AOK*EyPYj)`gz(wJji=VBH0-9@
zA?PmbO&*WpI{d|+S~E=d(QRQLXG?j;JG~*B77RkcaM#a{s)vO{6&(0`6_)crfL@M%
zeqCNq5w&p}yRqlXiQxqf?1F0f(v<}tf?~+rp8to0F%UPqG*2_%<ge*^h#UGIwe%MB
z#J9q5LIOSlGLt8dZl%SXiJ#DEeWOuyPyAV@=^v6-SiYl{4$BNnKg;|l`c3;af1E-$
z0JOW6HJfeV>L}qD?$VA^tuiAVO?^a_hEV>h62@f1nXvg(Q_>V;Ealy+ock|LEUWNl
zJ*;%r<plS>VVzw@&3<3gO=Ojzff{be6}Y#ty>)}4x;j%c{Y%=PkfHrp?;|Gt#5>Iv
zCAztwm$x5`^itXT5i#Pe?qt(gKn}g(i&rX%<#@|06mj?Xd&F5&{jjuBa0(aE&qM%L
zdxn$=<2a%Kd~+9f>Mp#BGzHLDgdDv3qh#B}@Dvv<daJV_^8)r^=BL_%O@VgyhsSkc
z<`)Kr?-DEi(?e5%>KjU;uI6Rqt`i>;CPDh-K<Ghuv=}YX0kZPp(t!Ze|H<UPVOCaG
zD}twbr#;)In%mDBtv$H<0~#{faN<t&<WFC$Q>)lP*@WI}19rMA0aU%~-<$m3Gi|1D
zF7OeE!<BdOy(O%4H<bm2Skg#iiHbiGHL+gcTAL8Kj;&vT=JZbY*inLJEC5upfMoOj
z$EBVBn_`m2D43X$KW#)BsL>jN{yXzyh9FPJO`#Sl`VFY8lvt4)vjF!*DVJF;>0x9F
zeEsVaEk_d+f7M%*2cLs;s3CTi&5`K*kU-hZ6pS3_w*U5qH_CdQqes4-tC~|g%d<gJ
z$r}Tg2>`kaybbh7BSZhne#u4oe8gQ{?Z>@0c~SBqR|T#HI%HA1>Arc}76r=7!Nm1X
zns2Q>nLG^t{TAgCnj743-zq3L*&^3O3t)cQ1@%7?;(Nh-B5yC7zpIUHSwKQ*dtaw<
z6P~|*NU4;HQHKE-f$kFb5yM~fE7J#C?8#4Y(=PLtoScs>p`oD-irH_}PnJ9WeQLmk
zRR6jh`H+_^Dk5DV6YX*d)q$KbS?@4$NcQ3LI9W9T`ML+xsPF}jCW}tsxP$-6h>9gL
zDnCW@*7cH16al3_`twT#qi-EtP>oip>Q+n`$Bv4`d@^S-$n%$m#rG4vvX<6Ya|=?E
z*%|?wKlb%R3JupndLhw+8CX2A&08Tz-9)<S-*kjhhNc5kY*t50tN6oU(uCczq0*lP
zRL;g&Sh3!ly)5`f=guFI(-8Z*|DcBx5Oyj%-ZGdu#$}&BX8I2`eU({*V?07?3;~Tb
z=A^u<$SbZ<`keLWK7E~;e3<9!fZ}Or`3cF@)L!kburMDWgd7qFuH{1T(fL0k#(%%d
zXbNBJv3PTM--}w34^RJnB5^d$Mv%j}P5?8kr}yb{;)0A%t~^?#*!M6liN6g`46T;n
z2{G+goM8zTv<mDQDS~PyY!cqNg3&9?$^w~l>!OWpBS_QnCf#Xq13~0GM5D`jVE*@V
z%Y`-r0yRK#7ErU!1v$6FA8V0jSncxLX~3^p=egzGw+c&4FE;^!zqaxdR<w&G8>h=E
zO0A)C(YSnTx`V+q+{15g)JR&^e1miabQ_PgiX`pZHePIPh<K=OxJW(vt?vKGf)=*5
zH42!innFa)`cu;N$BKR~F;!b~)gz)sRhtA;k8l84XWhor&ECP-zt_HsIhq|Kdwo(R
z4wD5_rCcV;+I;#15Q={N^?`JSi+yiKTw+Cr>filaLE;OC->KyvP#>Fn1|0o@$Z2%>
zuup$Um1@~PPvihB=-n{D7q-?>J^out>qv;E(oK_Gp8G{N&IghX;8RFOZoje<-Ny&!
zr7Y+Cnw2SayC<mTvvYnYTh{|r`>*%e-V*1IyZq#QnRfVNdNDg=^wad2TT_;#BHeiR
zRe-|QL0B%tZ!jgNDRtC7l5~zod)NJF(7ZLSj}Li1M%TZ4g!Mtj9kdEhnnoUnpIy&-
zUrRO$ZjWni<?}z3f|~*-5N2}XykA8Qj(q6(a5A>!0@ay4X6SW)c`(>;yk_{ck7qp(
zSkZDqAQo9Ve)%WJ|JOSW?P1UeWOui&mUEX?2Ve1+jLfCmvr1)L5u7Y~d;5(QP_kSs
z=<5$R^_w`08Si+L=Be6zdAbz!X5zXQI_;?NBeA6mHk$7@c*D3)_*kf*TF`Rh#0#?!
zW&VpZ?eWgxV&XakcRxMev}mVeP}rEOazb6Cob;}}_eHHZrCYb->0H#yxD2k`Ri&74
zgnNv&>Ebtefn#GLgPw{zr`Jj<v^TLf+ZB!M=8@r)#Q3090o2ldciyd0KX{%E)5NwQ
z1i$EwmwN(akZC;sSIIykP6botcFr9AE)Q@*H_#MvFo7<5z8<Pg6^y8{QO^lM1n^i!
z`UCIG52zjN(+utWqXYbpgOdvt;~Q%tBhYl4d|<g`p-#&eQlD<Frd|K4Kb=SEbbm%$
zO+$lsL~h0ub}*?Ezore6{i#$w#aB9glNZxCZH4y!xUw!kVqzm_{;@tJlISZu4jqx_
z=e&_mE}d)rla4%B1ybCnbDpGL#kO-wCs1viIt5R!r%`<$l^BHW-L~v7MrpM6vYfq!
z#i9J`8?XY*Yi+5(Q~_)m*<@J<U%>rA=3tdZHWGGJ(a|X*@8q?#iRv$+c&(*0Ow)8i
zM6&e@H$m(zR`79EQ{{I`?<5m;_)Yf+hE2EWQi3W160vgJa;xr}#5hPqId4k(u~r~A
z6_rz0C((Y*G?kIlQ@h&TFQU}rwsyYG@|`dl*a(6tz(yEoLfgssJNyLrFvC4m9L{{j
zsavbDleP-(5(@N~U@oIo+QM1Ka$S_B!mI1%_!ld`HBk1|XOP*A=V;seYiGD1ORDDh
zmUic|bmmy;_N<ip)iglQWpwRfmZ!aiFP)Y|8!K^Xz57Z$OO+{IR`0RVL^XQKf*}1%
zUo<bOPhzcv0(SZG?5=mAx=H~W4hHQuxf^t@q2G8H!(^Sdkn3v@X1$CIfwBXuPac#f
zScy`XcdznHnK^cTRnuBaHi8*|5_w!fl-q?HV%!(rotTpG1yby787b&_uD2C+l#SVn
zYI^fcG)Xzv`eqxYa_`wvPuuga#}_~!wQ>!{$!L1R*w{pVK&4ap)Rm_iTT1_O0Gm0@
zn$arw$yVE!rY`H13^0RA%{>8b09s$+#Er4vO}MwBpU4Z?*mMYr^*qqQ(VgXot*NFQ
z+)|8PVf4eyD~>}|;YdI0>aKL13iz$s!jv=40czjB;go@(W&}rlGWU>^E&%Kzy~GEa
z3qA{Pp<2;nJvSOnsD2qTgVD3KbTmHXqv40$6#;MB6*K}&{CBD}9L3+`V!Vs$xvffW
zPjYx^Njpr`bFY`d#VzS#4qxh;o5MPyj%q-AOXm5tt6;u7QAxJD?CLwSwiOYyy-fk{
zUWp9M0a9Ux$t*>N?4mRoq9=@K9Yu5r!|_YCf64yE2nm}9ZP^AqjQFZt70bsUIavg(
z6|*pZplPIYXg9H&Be^HECzE4S<t7RTg`%2`F+0YjuS(hWK^9)ON=olPd4vaes8_n}
z#vfI+trAq8o6f&P@7V1LcqO5o#`y3smWc{`-`!iMy@UM0-Q~44BPnaGN`3ZXx5AD6
z&S)|_uc<q_F_}t<d@-e-tbVe<`~OA<P@6n<QY_$PfuS;t1E(&88muhs2mhalwSk6V
z4^5P4ANtmiT3NI@gVL7c)YKqe96a%Bb?8$%W^{a%Jqx}uQCY`uU1=FD`TXfGNrEA<
z52MKqf&=aU5Z?z3sPqpl)U-f{*%_Z(GXw~cZBR$d7blUFaW1&nLW<9~c3wrk{_=`7
ze_tESsP8PJlbDDgRWH?uQ_DNLt1S^(Q{vW=;{~a7>nNAHPSy%_l2fORDpK_+RVDf<
z2S7h{;l5jv@dE9Fs>P<7(Phf5^|$-JBpj-Kjg6<tm|=6USD8a;*37p|@u}wHW9wDX
zyM~zb4X2ZGwrWEU@-Ey3#C0@exI!?P5=)aCH~F6U%)Wo@ERa`B!AZzLo*)Lwr)5oe
z=TCNf-61lJX9rzsWE$CNzr0&cl1`&)Ks`H{y4I{j3dwO`eC|R|>L&<aZK_#o9R)J~
z?32@0*LL<B3VvZG>zA8dx?HgVF5T8D(H>r}Ays+MY`HX1pKS48FVO;|Z*i<He&pXC
zx9&db)f1%$waa8o5%EFJ(oOg+(^hQRg#7FtpFSH#TVZiw(nth4KD)_^(@P{z&bEyf
zV||oT?Dv7;wG!PI^m(VB!kD^4A9AT3SQF+^9DZB<-59Zz!TQ%V!L@1c7Y2;!;3B0j
zHibFiZt&EO1%J<y{0cSrSCI`d{!i7ef<Iqz)tiUk0vbF3&`3Mv3n>2$XuxPg#N3z;
z4tnn_u%*Dd?5=+D9M5C`-75y}Q}a*8rr4Kt#q1>Al9==xufJ!4pPbP1f2c_C7ia!5
zgz_74h>_C1x8aEy>E;Z)L+U^{>i7oi8fJG+K2pw3H7ah#Hja3yE5BpU@w{VY?#jmg
zQ;5yTyS4iKShHMU!epieFvb@K%qbeYjrILF3oObfrb1K*7YT`!^R7F8&}U{1w(<N^
zo-UgnhPV@Xz0v11cMe5aSuQK7(_dvB#v_1RELkO>UwMs}O><%Ki~8Iq?PVZ?<8<iN
zM^R3UDPPQ|Qf-;BZ^m)Uhzt}fG(Vb#|KmXbY>|x=kHY7#&HPb^y_Mw8vl&dXp-@#@
zg2}kDVDBWnp}mBZLpE>SH(4rRKTI%rjnO}o{h}`Y+nVOHGV?-tZH8D8q|@7dE4w?E
zei(wc6mwv8a9)<iPwYZd_vJ&pG-(>;vqf&7iQFITXoy=81Yqs2(7s$MBGuB{Y!BN+
zlv|xfvy0SV4&QfkXp~-`@kM4eaAFZmM{pWmA6jhEJu&6=TyMe}^D9pCpb8h@{FY2f
z{tY{)W`HZuazCUa4{j>iDSUg`4FR+e?`(%X(Jbe70w$flFjtkijMyngdUBa6O~a;h
z1uxx4)#B%f(Mbi$G~c}Y29cF3o;%HBRs$mXVNlf-Pj%K;Xu6+)M__l;a(mMupGi^;
zGZfwZL?r}{T(0{P?>`${kje27>o>?!cEvbRcrI9op(=q&$d{do__ZHjdfMM!FDtFS
zT^|pRzG1AXQ1q{&!BzKXD1)th*+=v3u~F!!%K9wY+5Ox7E5E`xLiVO=(moI5dht5Z
zcDV6{lhwK9XXmHwt844SU>+{pJ*o|O2nqJ*kg1T)6*#3qrP8kSehTF;9t4eVu{f~l
zCTb#u!`P*oXS3^va7f1T7qk_xSOT!oTY9cLJDJqPy`Xu;Pi|f0?e@RK=$qqRq}on$
zeD29ZyxlUin9yvQgIh~)ChbiVo(UexVK>uQ?k$?ruZ_|c30>5Mo%-2NFL~hw2Vd7U
z6w}%0(OM|INCv;*HoG0t3#=tKTDSKY{kSGw!o30h=zTwV<~8Q*2-*4mn>R42@LsBs
z)Te$m`U5L1ab!ACNm^;fEjCsdJoy>VHyE^;@e<}fi=p1KHXqUe$!AtTlLGEGJSCT?
zKQSUO@THoXn&$HI$KSjDRVEM4!n$*;)`twQM|JPt#Ejl6QpG^JrQ)KAY!)M)+k9P!
zS`s7~4a<oLq$WODJ+of<)Ev2ZX$Ts|c?bS>5LN!CFHSevr`EJxk*n?FKC9jfjZWfm
z^pqbA*}AWZaLW6yhnU_w*_D7u6}sgC#}T}di%pl{tgujBg=3${3u>LI1o0`N=|%un
zQtT}4CMvDu!~dhT*8j{SKv!XCXh>PWZ~4`~Hyn&D_L{TRHSY_Bwkwe!Ws_NdYeLlr
z%cih|le^I7FOZ-3WmIYh#WSZJo--=Sbn<myip|V>J#wa2d45y9AxxYv6t7xCri5Oz
zvb>q^zQ|FIUN6Kkn<(fX0Vy&XAYIZhr#=t#Nq*&osoOQalN5ivYhyJF>otL)BD5Q5
z?sSPqixC12MjwMTM3R*fjo{pkemi_SqC`aVDa}3{tNwC0%j@z^8<&JcQbCE0a(w*B
z$OxH6h}ZAn?<kH2e`vYa7Gk=-8l)0)$w+9v^D3B_S{KDIFhpwP4TI#6l3s08;IB_O
z2CO?u_NgxnE93t(XKip}%u;*0yi7`<Vfk!8G_XQ<CQ33M5*LRtY+~|d=w*AhReM~7
z@rGfNO|jtzbziLm3JMq1>y#Oal-i6Gp8HOpqkS~uREVCq0ao!ENiT*HAcev+b}v92
z1(qwtV*9nMOyz1{LwVsp43oxJ7-zvkQ64`k#Zl06gf571i|6%A^}XJR86`<<JF|PL
za9#;i42;^AtQR;fEgdHZ5?Sm<u!c)i3;R=H^gp>K-88WtFSOh=MwSP=jh}Knm0#c2
z8i|BFeKgA5SXWo9e5BbbXV9V;DZU`nY-0_Yd~qMv33TI%aRcF{TjwLAuOh9o3RT~*
zi`@Rs^E97V7$3}_1*H5U7fUJw7V(i`FaFT8Xfc#ziCKjx_8cC0_TjSZ#4{;`SV&ap
zg`Oz+=OomM0KHQN-w;9H4(~S{PZFL4*1QNY4$#+FrT8anBKDM!(}nan-FCk0o(KOh
zqhG@-Tsv*yh)_cZCmrDpq4fK>SeQGs+(Z^Tqx4|yWqv+1u)~OnU9R4JXTA2@+{R(Y
z@=*Ol)KCPai^$>kmb+&gypH`wodgOgB`O41<f~yawS`OiZ#7EH>(pFkeWr;2WNU!-
z&y9W_uZo%?!H(GnBRLozG^g7V235>U2s)4)CZ5?Z*mngCEyAz*Z%h<VCz1$BMLP}W
zC3RsHOY6L(QEBE59ZA2tLx);0j&6GQ@HoHQ2Zry(^ZdFeDb*l@A7lUdU_{SKW0)eI
z{V^Ny3KKa!uL9cn$^E+*RuMk1X{OBm(whrq?h%8jOjN2@nGCl|((c}YSavBi9YVb)
z8kqV7JA-z@D!yZIR~==&qgURw+}7^o&nPN-(<R7wyZ)W-lqTH_b>QQ_qssT=<!tfD
zVv5-~<Y7PeH_%-z7tx<lxbPT~>z{l91S8eR8k%MG3&g7YW-3fq4KYTyo(#l_e>`e2
zVC5U*=R{ENjfY0NDaP+lj+V|iSy(BivsYauh6!V{znOV}&74!HMVP7y*+!-x@Qpj|
zxvf<5JzNUF??KGt2^u~x5}_Dp_8quw+jF@5;X*Wo@$6Z<4eP!Z=9yXZ9dNn1{!0?q
z^%$e1)<lAfr4rq@x$chZAL=oke+fMP2hI2@()pBdDO_+H*(6VOe#K>ez!{Orm`28=
z5b%-wQ{evp>*+fDq5l84bhfgxv(CsUBO)Q2$SNFV9348l{A_n8oXm?P<b;rrIGb>`
zv(74!(HTd|CY$_D_<kOrKj8Cze_pTG^Ywhqz$+liYaMWIQ27TdIhsC?^*^<EO>%Yd
zk<s>eLM{3-W@viJN-6F{G%3_GZv<~_(zO2Vjp&q~lUQlzB5E@Xrf~o(rI}i--$<kG
z-4S<rHh`3?#aHRpB>#N)73BPjsUe?<q{0mqpF~2lzo%oebp{7z?NpdWFCSa=pAVl1
z3G0)Uef91$rDo>?^G<FU|5y@Ng3wY%=%wf!&bI7OT(zjEh%r<6{96negFEdjzQlA6
z9)ucyf}cd$OmE%3I?(Ur(Phu&!>^Ij>^qAs36$p;ME>RP``V<WQCK7+f3%F^3|BJZ
zbw)t1p8V&2a`NgaB2)9Uy@7?X0cHEVdUV8d%)@BI9{V?TnR@aLohl01ZYJ-`F78~L
z7!jrWO}rOCebAmAQsqM6md0LBnFM}oB+(Z&HPRdyRa&9`PJp=M=55U>CTbrlYd&R3
zD56Gl+!LCYEX6-mP{a$?`fYB`dX&iagOwP0f&{lx$q>FjKSzU;hrBe+IE_ouzCJVl
zvcPZ-AH(cLbdm{UQ(27zV~KX?Xs+nGTfYyB77MyKZa~KM6_0VwWtNlQ(sqZx&Z1Q}
zBYj4{JsR05?KI@u2OJV+<<O`t33vXJ;2^sNRUBblXWUkT0yB*cmS?Ku$p5K6B6!A;
zp99sJV-F!gHuMFVdHx)=Oa@@|rP}-IK$(wCV<jx|Z&}X40-i{0Zz@Pg%ts6@LBcGZ
zO)M;~G$UU74uvsa+SN3QJ#7zHGXkt?-d2}TsG5T28ijj)-j#Oln&xSm-<PMzZ}et{
zZrch73hsnOCdvrQ5=?2sRh&TYPPVCM(lK^SF~G|sll5bDP`3JZdR+%AA$Lm!l(OwC
zbxUc*)W)ld0#?yn4Nct@s_7=gINJRw8m*;pZ_s5rn*Um>LYalh8usUOKv`tBcFSc-
z1}};4y&OP3QpGw%vQJD#)RnZLjNc#<<AK2d?$i4dvZF>S3>IpwzM3e+#T6Wk@?vY`
zHD2XM#pYc|+Mx}$qIGORWO3D(GdsMimqyNNqA#J?*2_*?Dd(F3gGUEfh^+<#Imrom
z$;^T3;+^UC=iMOmE^YHas{&O)HTYLgpYkmS#Jch-(!u=te|jJh-Rfvlo#wU#F*~TF
z-Z7A^mu}=Jp~+g|dlM^*ALGBOV$G1MbM9S6l#|^QB953)RXrk({n8%Mcv}C|)aNr|
zwtLh1EG#+-SyQtE^n6D$Y37xxO$iVU)vjt~32x*zOo~o8EYoVMfeRXBR}p>+!J_8m
z4?L&`3ZY$JR#m*zITc&=?{Y$VCSEwPaIf!m$2wJJJji{|wliMk5r8qd*zk4@J^Hu3
zQUlAnB#TRMcK&5aD$4x~JWdp>6|ASxrjm`XKbb|?+X7+v)a;vg$r0PL?5dzOD|-TL
zqVywe>YwLo{>+)5M|X&xQ0u(oP&jQ(@7Xv4lTtvI!jKvLPEAqUrirVcV~a9Rce!Sr
zbOE+kGrvE_^bs`@p~~<SU_7Moo`o{RtCG9*thGvwdZgCidaPwp@}DgAp+;atYn`;x
zx7Ydcq~T!lH>AELfWFUE%IibU+f);qfQm@%7)*^MXn98Ah7J_$B=eSOuPv9VnBc7f
z8NPR%pAaJXL>Xk^BFs<*^6CJ4O&cY_30^n`e?UF?Q0`|g6<Dmtf~gf&!UvxUl#lKp
z<hig*nwbL^CCuNkUA4|3^{ty#xM5xS6t$fH%gs@cPu0s-l?~0Vb*rQ~`mq|O3qi`r
zisCS8ITw-VvJbPLT}U615-}zyP+X_4OlP&ZFGAM*=v1c?0q2f>Bs^&@S+u+_?#~t#
zC{|n|EhtMMI*x|@_A)laZv?Jp?<Q?8h_nezOOZA-ZZ<i6MxibgT-0b7*C)-IEN5cT
zrQ-<U4CkT4a^+C`7I$`lN|_%^+-BECol+v1&APYfGRP-MU~lGb_`xKD#Tt69yZg0m
zxE<g?hX-7Z0U;0d>r?r_4y(;9+>HBck4U!S_zk%)VCRvsp3XmdMWzG98?b{fo(;$z
zT@Wd^d8Cs(Oc6gA;7_+eksJApk9FUNW=zXOd86Mgq({8`B*p4|jeKt|0iplvyR8x9
z=2hFUM}~NXfX>ZOk#*K>(xP0HoR~{6)o^zgoEziPfW_J(Own-aHEynHl~(xwN*zPy
z<GAK7yW6;T=Kk5I;JgyTEJDIbq||NW`a_z=K5TtJMX=}MJ7K_e0KAw99kIfqlhMe|
zAh-ybv4N?-G+S*0ICgEav%~W96NdIloPrkYwkrPdb;TiYYYhMT9{bWpZ8TNdy_3UA
zYLlssp_|drvY+3q16~z#x1n7be35zm#n(nXHt<3r@||eQ77=nt=)7mlShi~qSC33~
zK2+ZH#7JxF@Xz+aUp!{|xFlSkHw>~v@z%k#gufi{l?CY?VE-iGbx23P5lO)Lb7Av4
zullx_E<M4`5E!Oh`M4LRpSya$H0e-ic#mJ1GMP^{EvI*DUHGE2rj$zBN5y{}FiNB{
z73y~(ul_zs*=Fc%g)05Rg=xlel3|r6U1_nfNjmx$IwQjbOsgRS_AAE}!GKJzb0%`<
zdf01P%YiudR=C8L-)K|`<lkvYq*;jES#oXgIYi_e{0;xZ`#-DQ*rP6aee1cCW34<D
z{WFb8szdI?<4hKb&yxC(!eRmt&W<RYEv<4#t|KY$8!ibaXcBQQ@UlnXc40x$FA`fN
zJ4`Jy4Ic(ny#vAX3Jt}EoOOcb!#52VU{^wlG><uJO_K5}BmPWJ{lVfI#VJZhkNa<>
zFZq6-1FV{Bfz~_E0mFjjY#zMii(^ZoS`OrTU)cJd%%x?;E(}%^G7{IP6HsH*<e9N3
zox%Bqmv9y%SzvC2OY^-16a1bk+|UoE<QRYKBmbgz)5t7!*zv$?Z#i~d8dx2vN4ZjD
z+bSOQmEKgkrovy);3P#!@@t*`j#t_3sK?m0*vh@D$H2C{?wfn0;FF>@p9OijhCQe!
z<OFa0Ce)XP@*%eDrEn?4a6%{bJB#x(SyU6~<flVnH#H7M7E!}yGfJeL5{!ARBJuyM
zpGg-WL5m7jgGmj=`{}~M6NV1f&9!agH`Y1?u&*uO@qOIGLn1#j$q7lprUv4!bwvkr
z8-Q|!$btepQs3C3&Nu_f_xQteb>C1Df?O_oEdH2i6(&IyMv8=8OWI7S@c!0Qh`&zC
zcOTbWN-$9ae&v{UCA>DXc2mDloW*XKCs^~woViOlg{bS{HP#6#x6}4(xN#u4wbXuC
z$jJs;(45pgXWAQmIN$PBc1vxXK?lGcs*+kQul?iT;0`5=kjXS4Z#j8~>e&a918+w?
z8KW`Rr2*^2G|U-VV-q?w)JvQ#R65Fl4c}I^=W$(d*wZE#gxFXlE4`@HGQ2-ED%EK1
zMfpNY%G>1-TISF&z4^r5bruf)GeeUFv|Ik)K&_SJl}uNpCms~VhHq$?y9M26LodAC
zJyEc?>X~o^oZ?fL!1k8m)Oh!Pd2IEvRn{8DoShCBvb7vNqfnv>1y2q)V;zz@$#0tW
z`QagMm31Ck3XVbLy56%m!{JNY6T)^#z;#lRs83DfUl5Ojd`E-)B`<SrOpGO%5`7t^
zia4ml{XSpVudvC#-eMkiaTc``cs?v?P8NU=OH?eeE719Io>kyBC!<&k;EWp4guGzh
zy>{_R`(qt9rn8AHnLIR;e(#a<O(R}`I8PG`l@DC@0uq%GN&W^$PT?{fU}b_aF5FjQ
zWTBwR?OB=LXiAT5_<UpT)31lA(IFG`&53J4j%g(I=OzL1_jzej5kqDUwa}+&96Oz$
zMn^IO53VZSggA@{K3_rr*LyZpTfPIw><q>`3o;&^|Ex$_Wo1pH)NkEYN`LyD-q%Aw
zQ3o7Bbjv}%_ORW*9K;vXPo&0`vA@8_UtImV(}j()Hu8#EZJ)Q9l01MVhXwKEI)e~G
zEWIz9op&IsAsT#izuhXjIz!oD%vN#+!pSVEeT+711siE3g<J<dlVSlOael!10#(M{
zsUqFtThtGc9iFhY8EDnm4e<S*FJu_Vrs@kFbmQ;fF>FcYYCQ^zslfPw&W4lKQL9~q
zMs;{P0LBf&G6f|aKSk+U4fh!ozJdO6{qEmO;ZAXS$I;PhgKq_=0L;<CDa_)zZ*qM5
z)w;(@KF)sG>mYa_w=gyvEMH@#WqDb*DsonO^?<i{E5ip&I{J=MF_bjV$S4qqi^p2i
zC^#9joBkkp00V?+5tBWIF;-}%g|!_4#44oBCX&6KuVJ@&rYc%WS{~)=mUh;!D#}BC
zjcFHtYkBJp=UPm~-Ep7qlAp>>0i{=D5>e#Sp%(fhD}rt7GHb8Z$eo<jqp=|;4y!vf
z&I{?d#F80|F1VG!T@)>{{Nei&$FcjC+1S_juXMtJvEwdr`0p<2Cc{%OHmQ1x-*QCi
z<~x7VP!!QuVl<g<mc2Jx&Z_jhoykAdAJr&HEobw;_R4T)h%8#!f~L8jgmV%Z&(&n7
z9E}mv#vWe6)!(AGCqGP4hTM&g@Ns`oPQ<O%D!R!oUYS@<wkOHkG2J<TtaR~r_$^-Y
zfejXE5iGc5M(})J>`ZOUN)gqjTV|uD-f8uai3?!ym6!X|A0AvPvV&~+vJ0mWLkfvZ
zD%kz$?~*s%lq;#bfqMLV#cA?ht<>m;wzVL|gTok^wlBVEB#PX3OtbDx9-T9<zvp&a
So8bi+>Cw|R(kjxh3;jP}`u&ms

literal 0
HcmV?d00001

diff --git a/api/core/model_runtime/docs/en_US/images/index/image-3.png b/api/core/model_runtime/docs/en_US/images/index/image-3.png
new file mode 100644
index 0000000000000000000000000000000000000000..bf0b9a7f47fddfcb7969e4ea05e9d07b800fd8d9
GIT binary patch
literal 44778
zcmeFYcQl;e*Eg(-h!(*&I!Q<fB3c+mjTVC-L{B2xFnSwphy)>ei%#_Dy^WS8h~B~&
zL@?SIj5=d@$M^T#&wJlbvex_0`{y%j&CIo~bDeYc*?XV8_xbF7z17#%pu5a|nS_Ld
zPD@kGfP{o%hlGUm^(89eofge`PZE+VKu1+oeJxd0UVRT&J4a_*5)#d~DJIlU4f|Mg
z%nkS`F7fi-+82nNATdyJAT4C;QP(7Uef0%z>Ce%Jl+&WDHTmWIQCy6G7xs^xf3ndE
z^|w8(5q33DnhshCK;c*C&(Azp743(z&8bLmG&yRC(O|N8LDA%IZ>a8Tloa0A(j&dj
z8$$1vH~gDT^QX9Y0?BOm3FdH{YX6PT<m0M?;j?qF8QF7E9+I0>OJ2&PiVRESB#Tc1
z-!?Fke8_5kZ*)7s=4VLkHGxaJfwDtJOG2``CQAyJb6-E8{~|&XKM9q06d++w;~T!a
zWUoROV-=bGl`P>O5;^7<U2u3H=@&;vgbEMM7mpxg9D`gsY+Q$DVtMyIb*+c!<r~y*
zFW+p1-#NFJJ!@WB>8E<}^mB2}ExPEW;p<aqJKf;doaL%>H+18o!k=<1G4O|e;(3vr
zaf|km5w1Y}l#IUO`bgMM({9?Rke`>97~Nfr3ZwLf8Q-|^g$q+;-xXRGP^w_q4ScTb
z2Jg2okW|!Vp{*Q__xLGv#;Nx)QaH}rIyLFL@p{+LRgWjQ_v-G(D)+w_MT$hLX+36i
zQra?#XXeS@yKb8G^)35OLjTh8iMa}=;I#-Up@Z5FR~$97`n&E{KYP~s^V?k}8SOVm
zEOxhE-`Nan+<``&{<5eUP!R{XD``BvC+G1@l<c{FjmP#+iTnrRRz+B4gMa=h;fyxC
z)haFoo3nV&A%q&_^AWxsWfJo>>GH2vMjOAcCw_WQVVmRUE-fanp3cxop+-i3iA;w#
zOoVGW%$XOw>9`hZ(x1cFO>PiR^Sb--*DLQdjrSveiF02IQ{=n)>Q#?>Xp|M-?N_JM
z%EnMHitADH{9hwEcPN(R8A&~UHR_0K%l{)nBOm^5jb{{Ev-`ZWwWbe$uAtZ*Z2vR<
zt&w<LXrvA<L{R+wKWbNQgfpVuROc9udT#s@Cy{2$dJ{nR<9EB_FPi-9op`tG5v>*}
zWg~vVoq@-B%o`D<;Dj?95jUsO^;qhqPJJ$+-~KIpZeNhkIHk84ZnUeW7WA82v8?LN
z!ZLdpd{tL8+>7&)`I1SZ(8D*Rwxs5$pDZ_?)dKr1#;G?4uWy~m0#Okgw2QJ!_Il31
zCLTa^_Lm_WjT)-%FR?Z;PlbJqVyR80ZN6@>N3xcD7o^^ZJY#ZQ9O@S~U9{MoL!UmJ
zF}a;}7p6h3+xareC^`Mf_m`i=Ud8vZV6(%;Ewd~wNaJDF$<@^|wNV(V57W5pOe{4?
zXD=R)XMmwT#gW%cGXu{eIz5~29<S?@FV69G>X;QSyPbrREL`<EF1${{{EXE3SD-}*
z^$sa1??11@9O6j~y8Ye<v09)E?$PLw*DK$-P5r1_yN1=7B(uk-hIWgbZt>CsnxT;B
z#rrMv+uf7nlvW{DJM@9-jIUphFjBn;dCzYgNxt-|hEeh6KX28R;;-GL>r?v@c_lsk
zhjM>{?RT@6%m!gw59#>?6R?^94_au5*Ny4~z1!g*3*<`0!7=n2rnfm_l$b}{9R%c<
zMAT6qqDLkjq}rG(HS|8Pjcz&!;>ggSNs7bQpWD5^@$J)tr;@2rw$BWlsf@#)Jez#V
zHAR8JKB-{d>;W#RWra$(YqY%jyrY&Cr}xEvk#@K7*ayz-+Ecy6uw!+~5vqQV{{bVM
z#^RHMioo-$`N{gqKeZodKA_7j68!|=Pbk%B;Pd}vFaG8|C`;Qy(?VkoB(Ll&2;@IZ
z(7Z{WNZyxr4YZVSS*xf}sxZ6occE~h=oob!`%Lnd3O!#*;@9P?9+Xh(lYdSwpWHc-
zc$oC!;cl^-Aw}|?<gdvh$pZbiwAof9o<xIH4~hiz>&xoP%FDbb#wXe*WWZqXD%c8K
z3+8V$1(#1AmS!6U<Q+b?G)G!x7XXbdi#s)gGI2Bq<{PdKk{MDWA*tHJg(4p&M|4NG
z$AJ#_B(o%2*vVqj*;CjL*i9s@#_o6t%Vh#{taPgLFeTO>jvNdmE2U&J<Oi>3q+~=$
z0wnPb|Crrq_|Ul5zzRl>$W<TcWEMgyxhh0Pa>_eQ@yUlbsJ_)abNeUo8|OFHZ;t9|
z^pAVhBg&I14Xf%Om4Hf2Ow6Z2GFrZME=Ajqk-7>x!)9^iY30pNU32H@I`mZEMbt!G
zin#lafy{D${1Hy|fPDWA&KB2xP`wGe|BRi2WBH%EQV#c8>?1j5Ili&$-nVe|o2{`g
z9n#OuKh3vQJ)-M?yk?)__<9{AHS0V9p0Ia*!3n-AZe<hL-_ol#e6#x|;^mt@@gx_+
z@Uq;p?_O`#-N#WnmI9gS0~UkM&kr-Mr@u+>`yrj)m>wkEEUqUhCcXd(9)1Pxsrz2@
z^!cdssFNdDw%;SewsHj`9K|vqGyqPSG%Y`xlAh|D!d7rp<j$-i+#pd9Fl2jn%$WcY
zoHa&R)!&${-SXI)nz3xsKqJxhTf%5D**tn~dj1G+di3W@pW)$eBc3p9x*7St1hC3Z
z$pPKQJfpmX+`(@CvU47O8#_gxhw0B(x3;r3;a^XNE}@xz-OyH-jPXOiz<ly8^J-ny
z_M~&qUY}jV^$+%d)|%$g#emO2$$0rAyQ6CE9BzX+xi}ZSN5hiCLc{c_^upnd@!mtx
zi!5tW<C~Gw8m<f&6->}P?BoG%`FLvwDm`Xb=(e1i5EnQY1ntZW1Q6Z_I|g$Q<OqYO
z(6h-i2J*XP6r|?lId$ltGh5kPX`w@*W8Gn)7}@!(^(<!Pr}_^1ruybJK5j2|W*6~%
zt9<EvjeH&X+xdp7das#2@qbDU_jZeE4RoPvmYI_X^!I!8+Fvd3%WPys!ePSL_u|KE
zr8|!YG>|cMQCBsj)WRHG_V?BT#HGcVZRmTi+St4hw#oex6Pd~CE1kon$QP_}+Wxxx
z^;a6-s}N?}w}V>yg|%8zT9~A;#I#$!vD`p^+*<UV@H-!pzZn^T4K}7GI>mczjAPf?
zH-SxFTQTm^zXjp|it}!P+s6gWYs`Mkhb2s7iQ_(nLBAeiKSG1n=k`DDTLB}wqRv$q
zd7tvFvEnYL(d614+ua@TOI^=wm9Ga3D3<e=YkXm<Wa;RqADE9P00rgj&2?ce9^2dU
z9#&@ae6>vvH0Bar6BT8<ESGriq<HD>=d^gdsFqY|Yb3ezkIBf>QgYa`xouOO^-G4R
z(<q^6ppJG%1FUZ9<?iy-okK%b9jF<n!70KhPzVObO4d8pmngg+%IS2Zx_=``9jg<a
z?rHBccsaW<NZQ+HXlvn7)}z!Ecy^L}iDT97oT(=d;VSMCq)ff6p}{6+7*2=XJ^63~
zp|xPEFx{!t(f7%pYlVL}9Z)&@MB5d6?0ry8h`!I-D027|9!%r@)~_$7=<HFm*tnT7
zd~0;?@xZ3ZJ(J{0zjlvL{a6+k77pXbCVj@<Rk0SxmTw+|iBi^6c9k7g*Pphw_m_f|
zC-L5|yiKO4E2<IC;Q4`{OJY96$jH2L`B>dlRrOP6&Z6O@FSSs_9%e%@uAg*sI+?0Q
zd_l-H_d;CDb}Pzeo;TRIi<k+QqvvIPP%D<o?d+SJo6_hhgk>l7RstH@DFea81To<=
zj_MDmxoYI}z2n_$D!9vX4HgjblOr?l$9@LiKGpY3(i`vjtQ*nHB@Qrv8Xc`I*-Y4Y
z#vSm80q1>HJU*aauJ&mRGG}12*n>9A9yvcktVfQchF)jMdZz{*9?LFxY77PBDtd)_
z=l}Z9`xR$=G!@p>ejZ#iSWtPEh${LvV3n3RtI@>Mz64uYp3Ez^uX3-pgmZg$5Y(4q
zGP!I3a;}$MfbA&(o0I4dh+NE%wnr`IK6+TIuR#U)N!U+9ZTolub}cNAx{~2H&ph-C
zY=6nU=eL-pXYNRjk-Sup8TQA@Ou}~j_P}tjJjS07y4zZ^uUDy9ZZ+E}zhlyEUD;6(
z&~Bw=eb8lf{B3ctsoBLf6aNCY-8B$!bP_Z}9marAtOMo{l26yyQCilT*2b_#{OrES
z5(0S!KMSVO>o+{T{H!Mb`gIccJQaxuB?;AB?&IFuN3hGP$pOZcWal)(N0FDYBzg2_
zUA>3hU77SGht8o0u~I!ATBR#=ArxXEZEf^Tp&c17tnbM1&rv(;oMz;t(VxZKIm%L8
zCSb3Be}e2DM!`t)=L1zm6Ystt9w?k`pK95GKqNxM>q{gQr0gV=#4A$b7f8zSzpvFv
zZ<CP!d!LMiB*Kw|;(tA(OZ>n1#1p?4eg5-LmPhtqPf+aSk^k2<>FbL}Jzs`CC;rj6
zX_^vGGOV{QzNA_PH;HE)B@!()Wg|b*EhM$ywTB$tyJ^=hJ(T-kMEk1ac@G(XR1`n!
z-H;F0lt1R>y-=k980+gRe~IGyI#28K)k^WPa+EfPLM734Ue>POxAl?euWgUkQxx<Q
zBp105KC=~ID72D!^CktO5(()aKbqR14TbW;x|jcKx_G>Tgj7D5>F+&Xkdpan1$sV^
z`eV?Go|hJP0U;!0mw5lrAEm3bjNOlxGC!B`{hv)2k1LVHE&QF9#HS%7R#z7|b?N@v
zPUd$?|Hot(BgTIrqu5D2UUdWiNer=_vHQ30Ul^lA5_*%BHze(gOoXB8-`Yb+xVy;z
zI^xY*5>oihPT4H>zbv1GNX#+U-$bFr`<C+3;;r4mUlV_2IB!TV=5GuoA=A36L?ZUq
zmA)kWukAz;QvY>C&HpRH|NljJKk;~O$p`E-U6qt2VqUoFgV0h}_s<X9!(vyuZauoC
zAY#cbS<2QBJh>s2n4K-+4W3REdh(H?QkX%)AEJ1A9KF1<qE((MQV5=`*mIlBy78wL
z=|86wvY?Zvvp~;n@H)1daGac^Eqt0kj#E5QIRxfqHz*W8%0EEMdDU(L?;~#Bhy1L0
z94oha4Ue6McSZleMOSwwM2m5W#0~m2jreq(Nt?)V!rC|y>sM%evjBo~VhRBasnZLA
z%_E(=7W#>s(I}{!^Pk(CcZNdfC7rx4v(OU{!=79-fSAE-RuZ%0laal*NyYtWUl6^S
z@tjVFzqjpBiK~{2OI6^0AIb+S9nsJvMev%@GbfLonaPZjpFmmvNYLPc&7NhA>^hWL
zcjd9>PjyEpr;1vN%FG3pSR(6t(-WlY2L|?VQS#!({C`@GI|BOo<bK;$orm859tZC;
zkrfrzT|K&D=1mr6JX53QRm^QV<BQJ=3jGhjK4#~#MOd2qxQ|TpaBdr9+rfOGmcSXo
z^{tmedmM5FFN=9pz~y_pZ+ooERWJQ)&G0_#jEqAKzK$y|m_45YY}X>Z9bko=)3ref
z_~exX7U+u#wLgu7fldfXPHR59!}v3kmTQi!RC^Zp^p4DHU)YXpt5RF?TIy{&r_u4a
zHwK=qOnc}ySKWW9Z{5CC-F24Ajbf-7uO7;Hj-MF#9yy2}={#|1*}g2>Jk6VaA1%I)
zz;im*ZVoA<*yONN2+V*g!*ES>wG5s$t)IvufXvj5KUq|ElQA>$u`JstyS*#?G})KM
zI*5>MA%#;1V63ZX+k7@{$b$<+J2Yc~8<$*ji$UxG<8t|pSRbd_Xv?P0^oW35m-NH0
zs4-E16fJmlBh_^PMbqmYFVXIkpk18pCMiYgG*K#=9jo~5igRsw!yIJ#MZM_ertl%f
z%%8kr)MN_1%+#HU@3r<gQegOLzKS8={$?(yN$~gH+8Z^A-B-`Dv{5gcEDY*x1iC2#
z&}V)9{dJ_Z<cA9-BXZ3k)4<p^aF(+;VfiG<X{kKSw1rf}xXNM$;(@nJD%8!Moe%sR
zPN$|Z!zBxT*Q_H*+)a0;EudN>_wD{>46okM=OgnQ0AZ)D>kZz$`EgL)vhC$hrM`-+
zJ2Nvg2GuTHtt;V9h1$j}qi1WU>mMI~w#60?j&)NUd07@a9)1kdOZLvdjr_g^e8Xn#
zudoqxY4IWGYxtiILP>{GNR6c0F-iTt&;AOl%gi3j{a^EUd~kEy_CHrYIIcU|<wb^j
z?ELB-Sc092Fzr|cG$b<2L&Y6(^%U_2ZMke~3%VvP0u^#XXo4Y6Euzg<t{&g15h?wA
zQqoxyCcF7g)Tiy{`4yg7>%BexxO)KfFF`ogtyf`xk$gp8q95XZM?}Q0WxlSZb7?5y
z;4c(@ylfOnk`ucm=b+>oz>w>=0mWDyu1~5SBJK5DgFSU)v!{yd={P2iQX?{-wt~IC
zJX=T=X?^p`cKe^R?rihoTo>rjl;=)$v>k3k_>+wLVBZYCF9^QCZC&kj3Yyl&1hdVk
zZgjKz587u1*JVwV6-YY5Z6r;#Of9U91iz{Og`~^>l62mwfStSRdN`9VMoGCd7u8#$
zC)5OMZ(bniw`C(1&{y|c&w~R70M&~2X|}GjHwClDKI>Xbzlh{q*mVFX48QY79DVxb
zgKI4*)<T;#ZtWi;re&noFH^PD&^|gDHu3CCVoHli%?&;ooZ1Q4or$<5)4cezeY-9x
z5Q}Cw)SyO>utV8@00<39&RxC<b)eMEXs_VA8h@&{fmWUh<xazJ+YC?8DzZ!KNZ}@g
zLD5K#8!>NQ%*?!<Ed+4kl9hIN<GLEb1KS;>)N^UJq?>VF=-B$iU<X^OoOt})SNg^f
zxc@$3v$b&0t<fjzQ8erMZ-GarWno@F&W<J0-G;A0&p#{8vzKN{ueeBja;7D^fpchY
zphG`JnR#=Q{K-D;eF%>EgVf$WYD~B)iIgyuW`*)jVr}2vf?x!QmSRqJ=k1>{gVD9n
z%PidiE0fKak<-Cp>fd`YU3X(RSR{FZRvJGcSZ2lhDP6m;`ACB>2T?zhxvD4S#wpwO
zxdPX(ryG>`^xUp}h?ZHHSj&{%l!ETRwO1>#95aJ`4&P4_dUBa8+$B4Bd&DhZ^b8w*
zWlI>-DP!)^33(NG{(0DAAvppaM<r9;vj3q*ipBd}%`$iOp~|1^75bp`hGx;x+KoZ}
zxD=wjfy)FmJ`#lXP;}rJIQhhYN2R{!m=3(%jLlK%d|(SrWtq>{$>|956%qVWx$Db4
z@^N%v+E8!6Ky+rQc}bHF99x_>Xk2Z|A;}}>>E8@g4s9RTmDN(e!gIU58O+q+SZRhx
zSH7=XJK!~2^zNOPY?V!b?BRQ#y^T%G_ZAM<G{v(lK{*3VO%Gek;42_AG)VcnMA%<A
z%Nu%`xjW#vw!~R!V=qB<B^4nsYc5{{6FXU%UVzk(?)uLCMjkdaNNUNDR(>#WCrk`x
z9yMo8-e5c)W3mg#@icxMA>5!#{@S(kpwp-_<JP2344gp`H-WIzka*$Z(B^siRF+UN
z-?Z-31w?6MS3#;WwODf<UeCV7A|L;QjYefD2Jd{U02DKz|H68wi*@o7^-35*ukhUk
zb1pbKjT+8${L|Ox@zW5E!u;Nf@G>rQyXL6TWZ1=uq}pRer}g05s7L>G)Z4D=wop|u
z#Ogw5_2437Rq?_*jqM~m5J%eD`{?r%lkHAas0yCb{g}_LJ38Y{YBON(p6O0+v-F~g
zLy>FfREB5JZz)M~J2h>y8+!Py=;W(GK?y@)D}S*DA89XW87%&mO*M`?Lf<*qn51bd
zVP_l{D7x^g47xB2<4QBGhTs^-HrzIzXU100d*gO`de2Kx=4*x@*})rH2CW{pGcE1~
zoKvfdc{UqA=<VX)Q41hNIqK_e_kWrfnKgd&=&K<ebLi#TYJ$)^ilDMNESxai<cBGu
zf-Ejk?Y57qWHC>Cx(+s{5yP72fv0_JJ}Fa|CrHIoMa;P!$MaSiLDqrn6P|Yxrcry#
z%zqwQN+ZcqbQlE{Q^YJZa*+5CpTn!DKwPS%l+>>#*SeBy7-kS<o7+J?o^WMeoE(r9
zq^k(f-l|@JaXQu;h*@_9he<A=mD<l~hvn~TaJxIajXM|@>*{;lOMz<?-W__pz0(5_
zi;7J(AG;R4DeRcz9QUU}7A7jWdDj`W)m2W9<pnF4m!tI!3qbfcw9~%GcO`cHiAABo
zzf@=}VjBC?)1PNHw4r@xpn|d|g}>)22w6nuT?h5l9jO~hr{j?Jj}AUT`<9u@w1nq1
zdu+a*dgauLsLGH24XmE#$uL;QsZ_^6|I{pAQwoDlsvMOC05qd=rfH!wV1~rXfexN+
z)^vy~7e>d7E+SP81t`#+uA7G+f@Z$npOhi(8_5Vc6KF`-%aDQPXSwaoyfL;O!E?4A
ze;`fF$uV}wHrMaF;`~@l=b8B9Y1B}d-k<a4@n0?sq?7L`3!^@~vp4Spz#5s>`&i~f
ztbzz=mC%mZ-ljCUMkkh8WBeE`h`7E;D2bQ_=s{&lNoa@W9mIaU?@Y=xw0t4xH`F|H
z=WCi(S{Z|aD0WpWK0t8ps<b8wea(*l@id!&6YF1&x-9B>Diht<1rS7q!;E~mYaY~f
z)YI!>chu?AgfK1XM_`7U$bk*u8I_!S1^v>lQ(qu$pxoW<8@d|V9e&GFdx-PZSuFhd
zR*J_jN>qArN9GC~sQSY@$RW*%WT!SlpijB-2vNLAgOr`0X3OBg{%s%8sJY0bgH701
z4je)-Rc{&mdH{4kk>bHwaA~LT%9Y~Knu1t`hm!YFe>fIh2l+0SihI0B+qT--o)FD0
zw9fTJJ8oWTjniq@pgP3ZqP!)|ATI!i;fM}&(h2T?%wKr^;gV6LK*Q*q4qC(Ius64F
z8PAjGyKw$}+heVVdNvr{FtafPkV;Ck2MT2C0j8(EjCa4M`Xb6cr+0`H^t*|eo3k)&
z4i1u>iJH>)bR>k+)%g#y%a||QPgdL=<;_y8H0w|Y9`7)Z7IzH~2F`S8SGBL+zb={A
zZ1mBn&%6pB$04CT%Qd5pqho)bNvqf58?_c^HQtIGju*|26Pu84uBaLYxC{IlA%<SQ
zssAf;WE*binSR>s1$DzQOW#9n2UTO+`eKS(9UbDGJJ-dw`$uK^Ib1tX(kb-lxa8-m
zxadu4OzrjI{v9ThfX(0H?gMnwn${8|Z;JJRgs};B_yVmw4!h}ZW`7;J%R5u~RrKuZ
zn_{VBnUx-m3+KRUHobmq^KxyGb~!$bzGjnX6W{1YQvOAjlvnd;DqClc=4A$H=N&8C
zP&pX_%}DuT{T*GHScZ(NP@JN?sU1+k@^H2wh45sQJwRsOBW)RlLY#nRGPmqMoQ<-(
zS54)umJJ4r@yHqhGYJ`UFbw1?6~-%fwh`Q4ly)VxwJVU%#6tRqRf*{I{wxRPdxn$~
zxy0&%nssiPIP>4Ax>o8YwIlhxI84TYJpjGZ<hYgW7^gL~$|?JMKgkx4-qavC<%QOl
z<YwQBQ`mi{lPmX_UGma9igtIKCP#J-OXk^93EG(&qUSPb1gBbloR;+ad3(klqp<w?
zx7<&s%QJAJdg7tE9=<n=sC&77zxDT2?YuaVO;+D@{sITF=;wW++!0z-SWRU53kcI+
zO9FBj*d6>m@wZk3Ya#-*v0zih_@^;f%Ht&BTEe|DafW{h2N^Ykh-g_EEUs1M&rLxR
z>Oll>syP?rb-DgBFz*2o7AiVZSiSr=nurvW!aV}954`_2@B|wX!onTB7yirXl<{^i
z5sT{ozl$Y?m;YaqeVP2*Gj5;^g+leNv~XS{N~v^>EQJW$tWj7f{ttXcUx+B^@XFFW
zu3fkR!K#*-*vgx~D+e=ubT7`D@qY-3*FRKy_02~WV_Djctq>K`hoRcR38bBCWiwYV
zW_UmrN@AJ!&C-px+91E1y9XYH4d_Yv@7RaZI)#u0$uLg~mKwdcw^xw|GgW4+_G;OB
zL-h1UUZJlREtMBVMFzdQ%xZM9FD*aWUtq1``R_VLjOt2BtlV})iD$)zZ<cxC$Z+Qb
z%-N5mbCDlul#d>!Rx-`BXx(I-ivH5;aY1~jCK-Eix!J?QU{FFaotl|+*-bn;QDCyp
zqNXabu*i|II?q-9+^g`nfPZUndQxVl6KuaVAuZP)Y$u4j`<;O}xuyuI2yR@cXa=R^
z@87l-hjG^z6-`B{wzhz>ZjxO}mIyWS1Ad8yf1=p2*cl1}#{jaeY9#gM2?Ex@lk{hn
z@Zi1&qC~bXQb?Bi9_R3L(U-n_<?kr%hXOCRE@tDs@X3=UKtT_WCe5MLxjH$@y@x~*
z3neSrskgnfvwNSHQ(l&xl{Sp!lf}^56IoRztZ$XWu-i$_!XJWq<%3%%^TnNPZ3RL%
zGyRx71OHvN8yOb^qijLt)F<If|ITuy*_uBUGM{xL6SOl_<mDObpHubinQ3k=KnzIS
zsG)m8d9XC<Giip?pd6<&Y2UIsnx!XIS?hL58%Zct10=mJ6;bc>^WZ?9tM0-Ez>pFd
zYr{o`*GiS`U1`gGpIui57K+O%6wz%q8CwhdARnhJftU^BME1!a(YDG1qUXgr|6)AX
zZsc`bt3bqyMVl-SV50Yy5v|a@UPVeVhqeHT@rJ|G*p$MV!#<`j<6^$jkbb5k{`BV=
zUbgSq`6#xcT4yTMiM(lgtaNU?@B%FFDsIW@Feau;be1|9vwreCO}kX}03xsTwL@RR
zfXRDx+KuMg)H^C+zO%lg=d7#!%V`g%GSAQ!W59Gdu0+0e8PN6yZaeVg!r=6mcURNy
zEjNla{RC<trHo_<%XC{6LCZx5c9$ykoGQJSi+5#Y^y*|Fbp(VXsQt&gW!G;jSN`p5
z@KA}|R@S`fXXnR8p27;DCpzbTws;x2;RvG>^pYH-Ql32R&EHtQ(VROR-L!0Q^O~C`
z+kU_Ye$$&RI*wh|{vVKi7s^&TLWY5*36C%Q*u5ed>D#w%R}+mM5Qr>+F*^UCr@-t#
zto5+-kVfuxzrXMY1{TjeW6S6uoli%3ttKmkY>k=YP^wYqAKaltCTvqAuAJzzy4m8r
zjkBsANcxpvq0rX79mI*%VEe`sp}ta=6q@<3>)dWniAs}qo0PDmHt<X&TFSA~L8PFu
zf<AdBRFuIhTX@`C&PpqD5FAHKfmyos!~H^G5^2`h?wAQU%~aTWncm7_u)U?%N%W{n
zexWa+z}JfH$D*J<lAK%}>(hNiyFG9Mh`B7+z~OHu0J0<K|Emi$tRpV84R?LC1{_4n
z8-M7kjSQ`nb0s?E`p!D1X87v9xrl=EwkR8Prs_;Do?*`R1Ku>tI=$G}YZU#s6Go)z
zx8U+*+Qrs1AYnoXgR6IQtb$`Lm3MSA2gepHw_xcH_%GDKfK(*9QHMAaSft+a1T8j>
zl&(FU2%K}MpG9%eTv!<1RtgK^QW{fd-fGM9T)hPXUMyyvqJQH=zpXxI(CQ9bw;^jK
zxSHIU;+5aY-gYzPQ@Ll9LGvtr_b$v=a$`tppP6)fp>0#wmFfEmR(E>gjNq|RJMWz<
zSBWHqD0$F05GQC5FtHO{x?*WXoFpH_ebLZIc`?bJL=2Mq4zB5gYd|QwABW6JUr86M
zMRBb3qkjPfO;J)9v3pZ0GBkrVbWfh!zC5%iOQiL1<ofk1cGnneVlR-9FDl(Bqh@x!
z*hgHxRq{Mp6uJM-OnVb|qCdq>nNV2=O!{XTlVP@=?_~z-Bzz8ZQ8=su6pwmIQw?#F
z5g8HL7$RLXdEYE4b*5=+-YgdJrp;oZ^u>D*Px792FT?hNM0!un|4^h&dZ8p!%l+Ga
zIi&+(C{_6sS!v=@YtUqGgMrTY>#hhU1EG+TYq;Q&M(Kv=E09+Q_@!5Abp-RmxfK@#
z(%$ug@oy<kZ5vXNX5rNR9$XXrhk_ecBMx#L0p_uh7b`OBPMNIQiappR^m;;D<O!Sa
zFH8U<ksl_;?R9|GA<-uLq*qKLN2w*s@Icd+3KHk$Z+HA!DbsqdU5LJl@_!AV32U}W
z5=Pkgz772y&GqY^kH(57kvk-0yNn~F4pOak0B&fDO`E($j*ILC+epaz1Q+|kCO(1-
z*S~GQGZ6T&;%ZhTB}LXKwfFMt>??M^m<?8)-q`pX{~ni!thZLD?%!4#1i;V$m}yrf
zbajRLRhVQh(pc;8BFI`2`**lr35~Rf0f`)_`@cDgPzbY|ZKH5|GuolA!?&Me=T$oL
zQUgDSGz$e9Y$W`iem+N*$R?7|I@X=EQ8OPW?2J;k=wTyf-^}vah{FB=4%hdukL0Zj
z5iYl!;erl8tbgyhL0#}nvSj~i9-P9$6it-UO45~HP%BAJyCZHa%@u5NY>jW4g6mQF
z4w3%51=NTuaGHS(>{PyK6Go`|JZHVQeZlR2e@Yq2h?`)%{aduvn`7h(Ka*14t4sRD
zCh;7Dw*SM;tA!?CUR-~aGxd}AU)B_+D&<LAfi0O${wKzyCLC`-3SW#ql*ys9`po$+
ztx%@CyJBL#QP}?;0)`&b6uOa!1v_wF|1V<xU$Tg-QKKQ|Xp<#R{t`hGHQ+Ka#3@p;
zVf*iE`QP2I-il8sefPaY=~<9ZXW$N6ARTKSdwS>_d)F#e&`E}ZYT;*<S%&1tT!G3A
zTkTEbbdQO%Cx!mTMX$2c|FaFWLgDh*yHA^~rd+?LZh<my3ff06uddEIf%Np!e*C!7
z_whvz&1BuncJ|3zA~BdngFUv;=!>o@6n?<hA@`y$f**Y%jWqCGPc}%xbywG9Qyxe%
z{1TzA+A3ENhaET-+55B=lW6zC*D|(32pLx?B*O05e$}u2`$ZyLGqSrFlxcg?KUoJp
z$cQ{~<kf@K)yqH;lT}X)Hzvz&J^5%9+rlIKuN3>MNrof#0x%CeiR|FzCR2O8l+V8U
z-dCYsmpjNt^}W<Xu`#7ws=fUZ)b8ZV<8-O0c|K*QgNN7wv944i#-OJk8|s{T*VOJ+
zF4UmQN}=z&rk=z6+aFt9=c)atl(a>}gjL-%cct@GTZP>q`XVc+mU9`(7ox;Td1l^b
zky`z9WktWk|4G^NvjvmE{zRb^iHA<}j$@0+;z8Luae(<$<pS^Kv};4Cyh!xsWTglZ
zxyQ!#Q2eVMCv1x+`eztElZ$lG7qgH<nOasl*_^31fbU$C0VPI--Zn13cg!Z$He+`}
z$Fa7TZ;SvZ-CKH8Ex2VAoQp5nYCTV=J0e9d!@ACy>I3az`{{wWnY6kWwmpv;0LO2_
zG_(oo?E@px)o#&O_n@fP_~kF6$d$Us(=(^R$Gi5j4QLt<^=4q5$K=NAEb?@s`97s+
zF#ej0W%KW8C*m*tRREc!RS<i+nRjDR2F*nKVs_3O-Oc-!C5W}Awm1Pgl8xH^_y3kR
zgzkxSWAg(S6o<HpQj9Dio{S!B1Y-kY_q)7PAoLvN&O{)W+c3`3c)cYST@=0!(;BZ~
zGI=`{^+;{(Ebm2jgYly-<r~Y(;{pARHm-NP{4A>-2U5H3^XZlvcqJspxrt!cBJdnL
zeAUz*_YF5U;V++$ovCvhRdnx9|I8nv1gDgC{*el$Wt~~AF*SYivcxXNwl|#4Ii}qK
zJ2eY!i`y>m9Vu?~*HMtB2_4Lo8R$9}Atic={&b=edH8&OV0ocLJ;Za_5=B+xvie;8
z?4do|=G}%kc?BQS$uH$5&9QR<M}voTlBHe<%cK`UZ)pJYCe|>)fmD&^lSYLPpig@u
zq6P2XDQ)_&DNW?9n^yyG5FyH>3xK-UM^3dtouBBQ`_n?EDxN+g+i7iu8hEy|*FC}t
zG}&1?R;xIztc&kIB@*ZZ2QS%XuC{YXJHNvT+HbNk7IicwZu<@|Q`=UP$2KVTal_kA
z`)lW0S5X>k*P8zcY2wmt>iy@zeqEl(!AwnWu*Uab>c+IpUcF0Zy;sOC-K<s;QsmJ#
zAG(D>3L(=WlGb&ts0ku?E{0V8NlOgS8%Tl{zpfv~r8fzP_5gg_*uRX+bGH{)nJ147
zC&_sySCx3DqU<{Fju$KPh2f48vwgs+Bc72D?W#L=V}0M>pjZ5oDY{8B-IaU(@HB*Q
zl#rGKGm^{kD^}>}H+Nc3IcQBJp4HUEYStERS?yG56@-ssp8iPnuXq#vaH_&mFe8HM
z;6*mQ)3g@rIo5h20zFz;J|`P&F@bR140oEUkh1<>Q4^u%wNq1w>hzct!dM)9Z89iI
zjFav<&V@k+w^jzor?-ZlPYqNfR^+OY;ug{`F5O@SZ~~b$UxXS3104pkeKfcQs2k_4
z_qkq_tb&q`b-!RH_M@RkS(m51^-eujzZ2_Kgf7JICg|OL8M48SDTF#$)pGuPakfCa
zav-x1voVpG5&YhQT^6h}L2kzB>1BSA2=*}vavv?W9=%yf_?4);+5GhDtsuev``JAa
zk1~$164tflK1f?Z$2o^yx6H3EXUwn$L2CG+rHn7EzWOG@JI&{VDaa!Qeh1IWgHpxb
z+_^zf%rwg4x_IET5$PYqJ-J8Y5V34F@0c|GlnOJKvLSw+m1~0?f%gYKi#C|L>pA_|
zew~8e$+M>UH!-Wg>xXi@7pk;(xe$j{K6eQ^IA}%>bNT9}qH=RY@#uP!y?Mr=D;Jj5
zfM}-0JH=-!8YK$`Qx7F2tA`@&wW7b(wx%ou9~BU@)8DS<4&ix%`Eksk*<WYvaFY3j
z;^E>cNNkK<a~>oXuwIdO^1&_`ziTL=ZIi8%d67pdmhS8FX_bC7*(1bXWS^6H3ukBq
zcIgCVVb{%{%Z#S^q?!Dri+Z~D3YwqUva&fCnM|}Fnl7Z3piT8m<a0dE9bl?(2=8K>
zfrB$2mzNurai+n)S$GodACj0+`s(tTs+R4+9L0`Z7d{a!+=yn1_y#1u2(QFhG>cGS
zQw{CliLc5)+#9?#_AHf{V|mdPGWBfUy;!QRTx=+Qih7%^K}6!;nCKzx7l92!FK!kv
z`dzb(z^T8~u+z|jsK5yrrp+F?S8{@7^Q+I_gtPE5KSb9Dh564<IXdoQ!o>{?g$M%o
zQt#M%dXYz4Ku_F=$u4Bn4_pTWunCmpU7%5nG<=Eun4OW<RXp&}N?w>rQ#o`+n7lw^
z^Lwq?&^O`rTB*2{@~Ozx8HWfe@!^l7v2?c2&gaDyMXBFqD-9rhSO|PXH?v1uIEGcm
z+n_yz<7(Cls@eHIi?;SpeHy3Z_J2)c_4GFDMXd6WSV$#IR#I3#Wse%Dh|^~624IfC
z9&<g#RF+9znaH^}t$V9!GkcL`rfiTgc`@1caHC-%D34q|)Jwi8D8IPoB3T<7><>=U
zy?@XNQ1zt&Ec`kif9TY*zH2A%Vf_>0l`s8yGJ^~XP0)Y_ZDup95ED_C<idkLmn<wg
z-rMq%=vIAPW`+Djpa2hcAut)vz6jr3g<A%YM03|J6CnOn_zsvO&9}uY*d_ZMyCY4l
zm8r_GCw3W78AJ<~JFQVI^r_5v-u#@8Xsf!EtgLb>%sRg1I3VR3e+Pde8iyUpVXmg?
zG)d5H8OZF*b|%g8WP!<YlY07Do)JC8a{&%e^QB6i?lK>=nKE#8GL&9(x1{WrWxX9}
z1{rvg(CQtZn@4x!$J*KX`(S)Ss&^y!Ns(#2z*zScs%wz3vz2{lEfLIiZ0WduFdg{Z
zK&qdl!U`jgrTnRQerCT{A=bV%C%+MdJgGEopPI5TZS_czbejEg)&ev@O8){*SDD+h
zIk8<0u68?8l9|AsBE0fPzsM&^<vQ$*EI2iqHaOH8RawkTRaDm>9GPplEWGDB4k-0+
z1T(>?z1zE12b+&dTi2Uw9h1H>HPiWK-&M_n(;5|`R0q*bNdVV&E8yvxAztP*fG6;#
znYzXXxsmz%ALjv~cTL-V068);>-ti}e{Bb}?12W}^T2Rl<rz#60`0#Jrvrn}f9sjL
z&I|iU4CkXc)T(Y(quN~v{+S(Jfs<z83+Ko6FAGbeeFm6ucWHe)6H?bPUvPAxN8Q>6
zTKm}%uZfY#1EQJG<)_N!32fKe`@pGFr4{DY3RM!BDS+;X{In^#GGkN1?lH$_dViyH
zvd+zMxjPkLJ7|an)`9}T-dEU15L7!>?+-F6S-U*N!=406p#U84aqK2S__Wz=d+wmV
zbk+fv#|hkHZ$J4SP2C6a)vX(6v-Zk=*S<>+J(-kwSaL(*6#b3wd@aoVXJk3NNM^w;
zhSOt&bRG^{O6bjLs;@L_s>#OSroVccSsz-M5sQ5ENidbo3u&;lyD3#AK22hUE&w7j
z#4AfVm61jUB5F<wF;x2<oR|@@+89hwAB8)m7dQ09hg$`{fgg`;9vZO{4#M7bZredO
zX4tl(vjAN#_BjI6mI19I*wy!sKWSJh3<c!0K-#>3RswVGhP6F4M~bIUtKh3t9B@zk
zM#&+q%+&HI+|&)?c`iMBz82Fw$I_(tla%1!SNcdy=>9prSSO8`apv%X_@Cy7iN;5V
z?_=#*e9Xw*rZes5_BC>^m{p5;*l1XbB8TKe99m&^`PT;v51#kVKKvYhm?kX7v0Zt%
z5C&U4zui0^94C+aGHz9};qp1Ns>^Ih@yJ<AMy%jh;$g}~PUzFD6%5I;H;?BTC^C6h
z+ukXUb(ejxe7!s@-@<OH>tKlb4SjVNxlQjo?TiJ4ir{xLiWj%_M{^i(Z}#)4d2VCi
zGKf^bY=d#v+>Rh&&+!z^Y`dWvhi5}Likw$>E+y8iG<OA>6BEiFeYOQ*ZC&Yy-AEr2
zI^PkQ1%JJ77K)wvvc_G(ma+%Sx~l_(ftwO9yB`Eeb9!&xIb`(b=8BW;PxhYfW3vuM
z(>p013>*;)G=#r)NKm3*+U;{aj63?tJpA*?>=(Z&&$x#NMcQk%?=?t5C$2@p@Lwmq
zXak;j21i%X6m4^g`k@p{sSdp|9SpVhVh`WDwEqV9#H$0gpV?z__iD^JcRg_h5?gW*
zmw^4p8ftT*87#Pv^Hl_o^O>OcWqABraBl|nDzWOPPR!>l;bnpBYD0wm_pLZ!eddPd
z(}Y*PqnyH42+18O2`kq7pB9lvx+&2heZzD8{}}6vg?NZw-M1+-`?n8l($WS1Oi<Re
zh+w|oW5jyAMCuYntJ~cN%Muc{-HQo(D<%uy-<EiZU4xGk>^n}w_qTl$oaf;)Tva0_
zE}n<7gv%Z8*TdNZH|DV_5-%{bD){wRU!%`c#)r3Ar2R@Zr5^Z>-ZJ^gCX&qnYCD^)
zZDwtgM!GB&mW|a<^X|so<8JGk(&oJHY?Qy4?NTB&d{AT*7M5VpdF=sS&^zE<VT1{M
z&>IuErZ~@HVl5_v^lyeG$W(@oV)68E9Rfb}*=L>|u?8UU*oil|4~QmJR80@SlBgx<
zIP?oA$_;s%P$ME5*C1Ta>|QA<mgdV(taSOs5OqRbIu2<-WOlOA(-Qajfmu`T41sLa
ztS`UMdg|SE8Q^U{wzkUP*75q`#hR7(sc+fk<x!`htB-1wAa>1NigA>*Y+BPU9ki4f
zF9b1Br!w21&@;vBi)TG5>j4nP`YADzmyY*;8`8|PQ1|*%2)$QwJE*?wvs>Faw{R9K
z8d5T7r|LBS2Ec|?Z;mOJg)~4T21;6NXOr!e&aoiKTk|b)kkhBhxDg7+vg;>1eI?d4
z5nBt*_Fr5v(F2ClsAcB_cbPdC87@2U-NF*K(>8F*-Y>R;v|w_L?^3JqS?*EwCJBPe
zwY1f<-L!YaI=JQ;DHE$DkRW(zd>G|7xPA(?dvEugYnI0`li{QsBPIo+wlI4JHg-s|
z0%hxMt+8T>=$6*qcu0l;_#^vswJpD_wt;~at~mLomG^R&fAJNws&@euKwae^H7A7Q
zh-Q*>`4|rfb@;$g7r6ewkfXdfaBn89X(~;0`ha~Iu(YG<-K0M@`#fW4Yxe6@{D@dS
z@MQmy`ER|<@r~hUbcvvJ#{I0Zv_?;Vy8TyXf;ny#cbwuRK=y;Y4_7%G)w8zC#S&(-
zq#Npe=v*9$fI2s_Eh-jXWsBR2Q>+)W%MB9mt)$08dyz-b9@R?sX*_(}!wB&ig8h9o
zpDSWmtYAm(`rKEvN2M~T1k@o4UksGp2?-kfzRh9#&2V^tc~v8cisRsv&}h?esI@>#
z3*L8xeA5-=ywTn~2kIo8X_$XDu(rSuYg8ib?+<ys%@5TCu^j@e{SUV6aFB{y)|q0P
z6DCey9q2l!i-G(yvBISyJi8$k7kR&y`?D!CZ(FYYXScP<ODY{rPA{~XBAgcn9o)9=
z8{)SL;uM>s8}J##Dk)JsroH*pb(?4J^oPlE*i7zEyT%UDe}smA_>D*}_~P-32ekYm
ziUEfY22x$+^2USlRO%gs?TFkhHcgj+O0DVH<WM>tS;9bd*TCR!Ow$T{^<ckVEMrOy
z)dC6w@&xbS0l<_oo`s)B(K@za;Z2VsoqCZ=3aT#G6?>#P=x4=$yraTMmafG@iT0;o
zb@k*LB#rJleIGwD27Vm+DpE7b>KVV7({3HSv~C=T`2cEzNRq-Gu(8q$Bk27&Ke|Q7
zF?Q#)n18Oez0HO#m7(A(_PNlHGL{)MmHE5G5|o)|`O<6|t~k_giD<T7eAzg6^I(9O
zjB=|MGZ_lCAmHV{I5Gf#=sPbx#6dT0{yyX_hfC~S++y59;BA29+&r*GO7r85!872I
z7^cDw0tqgLC1<(KiO+iw6Z5=K%Kd>$ptlad&1)7`qOQ&IthwwKcZ8IF?~SB!i<E!5
z6(k<8H%blf+&M(piy_WuS_S-86QyM)*@V;Rf5LbqKDw|-7~tr9MtzHOCpAgskxIFS
zH^Lb_+7WhH_PO34IXyqEAFgPVKb)0Pn;w@7K&*Lg+hgZO3qi7&HiCb3G!>-ZtIo00
z+qtC5=bYw9H^Hzy>vjNorr)Jn?Ah7k)LOZE{~~G7mK;HYJk>tiJ*olY_mQo8&)6mV
z_@H?KWWq$*xH<8C;nXNB6df!7V~Udo*<6U7eeJg@!GP!_%ys(5LW`~GD=AB~tP2UG
z@~}abgI0WOcgG0iXu;K0wi%y3w)u*W$!U5W6YB)F%e<et+AX@?$YqC%P(W8;R}dJT
zdpj%BFS8uAIm)hABjWTZ`s+i43oO&KBc0r#t#IaqlGS|Dw%y`D&&FT`=a^Z8wR~Dm
z=1JWV<tzG6*f~Q6=;pzr@T!vRyQnOgZmQk===fecc@FpPD<TUy>Ecq(hUSABNx)nI
zP|M_gQBVVb@}hXbBmT}N!-Iw}%VOJSg(NvHc|i_%YVWbp=>rP}r|G_J#my9%D7wY6
zfywFlm(&L%*|8ER3GY3wL#RE&Cq?3UC`6A~GR<IPE;#KHDH?*Js)!Wma%hf`#yIR<
zX^uHG&0o0ZnBqa0>Qu!ydFXEvUP(`qWF9^$DLXnrhT$Nzv-hH|SK6-P+A|%j7%d`D
zXO;(X{&b(Xf+w^NIqDA4^vz-#XAqT=qnQ8(#7&dxqLQ+i$E;f9o+&{=;JariYmDUS
z;N^Wpx6BXPVxP`(Eo*unsQidv?v&6kbJ@5U8j3_SrA#%j_DkMRD=T}kQt^ijCKht&
zWab-GJE9!o4f?I~z`}ty%4~<C$#13%eAY{HVctn^fN}Cm`qzQ`pHv18XcPOGzYU}-
z;AXuS0`VP01i08jPyYJxRc-4tmv-BCY;aHNS@($Wrt=PN6lE;&rC19RJ!&sEsR7A<
zShi2S<HPX+v1y(GY!s}>+5<ysn8$jCum%MsX3it1b-a7A;|Sy#t$hIfYRg7Rys5l@
zf+q9BuAZY=5%wc#o}IPW4bM$bL5H`Ciqn=}r#&1Zw`w(plp{=t6<K`bELPOya#)T@
zBOjJ+=E}WPNJPQ^qc`3}y|GD0Wi)O&pZlEIfGYu9H&M67T^nf^H*pDU42P})!@%=I
zne*=r0Ue9k=B4C&^3TZ3D-T_4*n$XyM@-~?3Any-u1226gty7;QnNk5zdoQolDiNq
z{^)UGzAi-1s*3+M4g?;g=LO*&vcJ;}^DXF|Uz)r$0X!obEEtbl@5)6X-37RN%B?2i
z*1Fz+Cz@Jukoj4o7!Rg%H)~oZCw9;4*CNy5Q(Vkc<S)v`+aHFWZH5mF=Ft(TV;Vfo
zBvdr&Vv_ec<RG648)vC3y;>pkg^f4)i#U3%?#QY9Ub!Jwm+26%uMr!S9DFR!C_k=v
z)xz!mSd)GB=tHlXXg9}4E_qb3RhdBVS8L6&qK@?yzX7=rdqM#v_z8?|F~|nY?_dxB
zX)9Xal4z>*DNwj{eu1OAU;Dn}9l)3|7+NfC=`gmZm+1wTEfG`q-khRAeu<6WU}c@8
z&XErC7AtcJI4UDtZY^v)w(wZ#7;*yDXo~6MT_NE`h@f+V-q>f!-Ld_F>ht<sT-;QA
ztYnLUL*5voPv>-vR(pf?S)mu)v8}C=6drUqFR}H@+siM_YhNSHtI&)s&w2W=CbChr
zsr<2O?7eR;-B+8tf^pQ9V|L#sKU&=i@;{0X8f;bD+rtb?5|6>nA*oE;w6PzCjHu(C
z0$%bmmLS#xUT?}NQzt}99*0qV+IKxaC!9slP*HX2UrN)<RHZ!|qHpdjy-3~q1y!`&
z7n?hb;N%P@kw^NlVeckYhV8Ghu_ciCO~<xR`mp8U*V3@QNFVZ``8!lZT$2AXV|TQs
z;T{X($)lvn&FJbkGQQ8Nt~*N0`6Hz>3ZcCKM;<8&?vCMion9XatOSd3rXyi7`R7ir
z9AWBmxQgOCNiNNUN{>8&IFZS8Bz~4RpNh=&Ucn4Y8WqEq4Vfio_O#t+^M;thd8wMe
z_$c230aUl0ewufrVqgTsshC<M)k;Tj{v#ezlKr85!VOR5G8T-pR%wi*Pdl7<@&(8)
zz$G0QUj24~BZr*gYYsy3ZG*>Z#COHmMwAk#yhd_0m5ZKQzbZFniaTw{&8=31HUBge
z72;(UeSlJY)?zl@22@aJv$rfqeuAZc=p8Y8G5gxYsojx|HpqR#_B?&G!*Yz>!L{os
zG&AN{7M0oUsVCzGa`kSr9k4vO;XKy8m}AE%aDM3W8PgQ}ya0PvT<+3=cXrHxLY$)Z
zUu|LZjGZD{_J8Y9N^v8xeU-F?IapWFC#s$O*<MRP|Gs5FJ@R4pfky6;nNw}p$>?Hv
zm#O7iZMp_;3WWjsM1QVh)DCyF;F;+lA_Ft4^(c}lZ}dHq%gbwOY<{u%x*Q2O$50Y$
z?q+Ll7|TSCm^1@YQz5fgOz>N$r_t3bxv+C)1!`aCX3NL9?P&H^&RDyBUz6X6mHfu4
z15>q&Bu&)SNYN;x38tqg8p)45mL!-#v3d|mb$GM&V`gHYp@r2|n{^sSt^mj7?YiS1
zp$Fn(!+qyv?Pu*8GhoD;W>!<1s!~r2r;kXdogpbc{|TeqILW8s=MQE%4@Mo6bh^|n
zA&Zy`2;Zqb=L)WPp{oOCW_?X^L34dTcEKMMEXM`YDIDC|BbJn@TXYez`}wwSe6R**
zX%MZ&*jcs(I7mqu^odx>d;ngsbbXcau;ZV00<k6`beZw{j~_KX?VwCgotT0zr|wly
z$1(Wtgv5?3;5oluelc_EBDdiT+krHMShjJdSp4ao47OIq^9&4nkNDnIha;T_Yf6R`
zgi|_1=4e<cx1<eL$R6T9Cs!<R(p22gkVrr-cA;jAR3G<I#(##4pTEND*zX>^93wD5
zkCdER9byy+*M0h)d_Y(POcud^gp<EC)ut(&#VUvG;#N8p9S7yOSd5d-tX+TJJC1g?
zw942sIWm9I>GxLPX)&U$FpMSU9!BgJn5G@_;iMwj_)f;)hk3h^hiQF=apD`!6VIF@
zl_*&vhpi7;uGK%O*Ez8bRyOR@ydrn|t9!&t7Qod?Zs82lyOKPEfXD0a#`T)GbQWY*
zo&(hI62pUarcQ%3PSy01K_cs|4)rrwe}*Q)_<CoO;5@rgV`t0n^??u{<8sjBgSaqV
z%eKCY+VIb_oEFPfU+DtfdqScif+({}%i*#H$I&U`%O|g&wt6Uw02JoK={V5O58}9C
zV~TF{QVT}GF5pP``7c4A6rY9wFHObAXF<dScIYzm_xqv5q8(7(=zgtnCD>R+S06i8
z2sQcA(7mt0Ru-P_|ASzY)4QZjsItwlBfe7M9C2l9(~dR>Jm@GgP;SR^48~fgv1TxR
z&Xj^I5%1&3VP`2Kpury7fzFzO8khS8Z1}E;T~{`c+h|wN-C{!H_UT~EXD)oexXiQ1
z>kWpwR|%tby|Y^I*n6iNo2~)vYnK|%3C(KU@2{<yOq&xkZg&LY!NU<5h{<@QIhxfg
z&Fp;!I1YIBj=*o`#I1-uy$jHchuwgkln><kEeD@(<T;w0ZF~}lq0Km@h8TVMp=3y-
zj4v54gazb|6RtSc4h$6|!e@QHVA0^I_<6fzb1pBzc`a&k+)ja>vy~ZZY{zaBKOPl*
zbmB&=V$ux$f7pA^aJKvQ54f}_sye8mN{3ODwpMGa(i%Y#YJ{p<v1`PhMO!sWQAzDp
zBh(CHYpD_|W{6o^tl0DUrPqDk_kFd!kLTs{;`zVHksL{W-+g}1_4%A6XM5E{q=DIY
z+T!E&A0|#q0PeCEL_htrfo?2SSld((9Cj63s%s2At^%kgfpJL6#*)`irRYt;{)K>0
z8^S%j?p_I&>$A*}W@l2Q0PLU^2*p)#AwerEf#)+82EA?0s9R6qU5OXxT#Qh4lxEw@
zqqG-r!v;$_<(1Wxf+9hp$X+w?i4@8e!%92ndY(!<W*{{na+tg~*KANNuOA3iZamX)
z;)Un-0)lpg@%Ef$K2F`BJRM{XA%Y1JJ-rXCJCkcr?m|KXFGc70m2j=Xp)PY`SefYB
z0x6WCjL5C51F;|#ah3wnx$g81Nn0qZz=6Hp_Ccd9YpxjvAyO!a-sK$J%T&a85G5|;
zXA^0Ns-9a15vkvG9@_LI`FMNcj;^UoNnu@Q^CheaA6T1=IKPSnh4Nk#z0aXU!P`KT
zLj;B2UZ#`UJ3)!;6fmj$=o~PcEF#d0)M^I7`@~#S3=cgdn1oV{;Xtg?PJ9N7$|#$=
z$|Gz=oBZI!it}~*gRAlnx%-`KvoEE12SClJN$D(S)RLbQxEt8~%@XS#@9Dm?wAK<d
zg!9dW8(e|O6s!5fqZD40{TO0T=rgNSL%rs*N+c~+YeCJG+KvMup4L!-nj($Po0fsM
zgy^Lt<$<^ezn~jzRe|Zv;U#xx3oDpvTyoOLfxudLGB#zwg`Mq+!!#CKO;8MHxwfli
z=)`Uxau?JlRw^e<uu;sU0p<Hj`ZkZ-vHC^_6t~}1&|j=9HZU_LXBWH1HT}b8HTJeY
zDV^gpz7JOl#FiQQOGFYavVl+)+;C*IO^`VPnZ^c97BKXfsiv>~x<^2v#E^m@0jscA
zsg!rS^?UVO?WVqi?U)8lBVtENPIsQkW}aKHS6q_mG;YNhU&BbO^u1=XsJ>9ivqWT>
z?a?2HuYYy@w(>YA5y|qgXQr5Pr-4%hQVK1l>)YAu$J&-gZ_(~0uxkrtmBW3;*Pb$y
zPbmSJJK2X`on&(*M-m|Y!0nnFSVGC(!$2=_k8&;u`<%*YNNvw4nv|)6O#|KT^lK65
z9#sE5lR%ibTg1Y%dRx|r4!{e6Yn+d}sKqrpXLj3dV2ZkBv^A3Y#2Fr4AV~Mj4V_~F
zjuLoy!3)8FNhy83SkYe)DSA8>$DR;gHkc<-Go9_ueHD32kPnwvGFo?bgnjnmrOOw)
zmbtZV8U;ePR6TA@ghn4-+8Ekfuq)49NT&Lm3jkvvyxNYiAxPNE-QWUJJ8iK;FeBgg
zK88;DS^1GsnEH*=Xb{?RWhDy<;6#S2;t5X60FW$p^ib?)q$A!u-~;?V;*#n|VQ|x^
zV?i9M&vK<!qg|v_sHA{{&M^OW!gds3Oy8prUKv+R#@-PgX_-oKuf&<HwhIl07X)e<
z`f80&$LX``?=ioa8|$MgSCZhs4q8w`<t+W6b&>B(lFWouZx$XOx^xXmFCqHrZC|FL
zzSD!~zEt!<e#=N9{-w`quW#o`F@sqrUt`(Ftw;10zWSC365t5S?p@Tm_JF?=$T&w+
z(cMaX(d6zbJlsFx)@-<Z{oeQwyE4=5M>DB`a;3PLTUMno)(WQvZxb<tnrA^MFLh&$
z9)oXyG`+6z7;q26wk?Go1*)KiMIRrEV167Zp>rLTsJ>%m0VNxU^<?kti>-xpEU{Hq
z4(xE%JT7D6*Eh8EJdoI1<7>X4e|eb#q8f?-yx%KD_iC4)vVL2?a0}SmCv6y$scql-
zGFGx1h|HUp6n@tysy^`8!_#BLmUKT*DUbQ%hMmc!0ps;W-QnIhY)6fy(Hj9&*6r!t
z)l}cMQ6NfRgjK^_O@xb9GmAC7RTP^xcb^#FQ*}xUIGmq8IxKzZBXP7~?}TZsAKY{+
zsM>zHjo_<L&9NgGLu<XaqEN=3IwW6yybUFur}fwf)ij08s>~5itLGiHh-E*JhOzX%
zx<i0FX$-WW(q7IQXe`}(;DZ(HTuC_~xYpX(KFhnt9c<n7gY7M8!#eqUxIhQ>AM)uU
z3)$?Na|NsDrl;hIo$2mP*Q^{`kEZUYOFF0P5>Ul?-9M-<N68$*xtrOX-ZKlQ)OZ0A
z5Z5nSx)O`)S`sP%bc`}#r2Mw-K+TiV&5*wL5`fD=f$A1+R8kAbM&T4hz`X_AJbH(+
z)C(?;lt^1-s>kyHXpUw9tY55XST~%TMYZ=VeG+0P3gul+jWm;*hO1Pc9_?7{()CDm
ziY*Zl(OPd!hK(R>_Y&z)_E%%AY6Bt=GaEM`p14I}jU1W7bU9VzV5;D3iV`BIqAnr7
zf7%9cq^Oo0$$H24O`GPsBShs!QpPt^LY`jse}5i`$;v+#n;8QT6ltnx+}GVp6$h~k
zRgiY+N$s6=Y=O-^<h8a$XJz_(8}HS+ry~Px!pv`4a}7&;9Xgx{r8&AuK)$^N1X8rR
z*rg5;_SM`oTWnXklbow82R9qlYrnnCN<<xaJsgnA4}Ay`S9LD2XIqy~Dn6iuTveC3
z4Zb7f28rO4q<WbAky_*Gm+jzE%a2#2A_}-7;Bc4MkKl3B%mwLk!fdZycCV@9I1*X`
z+PK3zYZG#1O#uSf=>!0q_lOh~^4fpR9EF#BPA|THm)Y6GzM}4}YVNhAAQh>0X^1#H
zF1J&ryOAP&-`mVaH!WC~mSfn_CBEk5Twp;S_PJ1ZA{Sd-syF&t8~YW5#S+^2xzN@~
zIq#`6#E`I>wqdWK6Y<W*_0q+Oo)7F;Wa_K7t8J{`m)t8U-7(E9C2Si+<Q{iP&&pc3
z4%lf?+R4f>-!(g&Baw1Inlpbi1TOZvHGaNO#L5V6QPph=yU=qw{C!C5vKeC?8h~V^
zrll{?!Vh0^5e(E&)CVsfJ{0nrU(N1b{gPj{DPMjpMxxJ@9z-+WCmXjwJD)Jw+ydaK
zOvSKF;H{$hO>9?3X<r!S@x}~N>`Gk*5k7)CeP&}8@GhSoOFE5n9xhG$5j(`~Au=AQ
z)8ww0B-0_a6N!-#c7xeP2u*dXP$HN*BN`Kfq~jhMNj@1~j_4EaPwGtX_zKwk%#|P9
zPiSg#^d{0P;M4g8*_UEolb|T;BE|vbHPWS$HH6sPsm7|-XH)`hVCr77OR-cmY?`C;
z`y!h(wi;go3j2-hj>9Tn`#e66L;Cx;C(9KB5MIqMkDt{lBhD`PxwD1KU?^8gxVD~P
zHI-UDN!R_@i3z41$9vj!9pe^#kp({ZOTDj3h8l_=5>ItiO}r~Her=KD+aip0SDx-L
z9y3c3;a#dZ@R0cS66LV)l5#H5cm4{IZugoIcTcqHin4EIV&5PovBdnNjMdl;VYeA$
z_Juij;-RP@$X>yFz}V*iZ?{+M9^ai=-<6JUlVx1W0B^w(+^sS9yzee_Y&ngfLb|1+
z3|BxMDtik~B9}HOrh_cl%MNzu#u!=Zuswv~!?NsEQ=!CCA}7HTzw`z0)kQk^2b-5H
zZg4#+xYjLTxSu3!)pILzCb_xC)FS@fW08U)HLEMe{OK7P551N3)h@){kC^$ciC>;g
z3xD5lE_FjUqC#of>j2W3;@nfqC%``KO~zQ#-6o7HOsQJmpM24bkFMr``?7jl^~5#f
z^%p0M7eUswiZQBwOFBfew|%KAS=OrOi+rY4kmJ{mlSv<m11d)0q&>^EFcty_z4W@9
zMx>f`_^jnJ0RFT*Zds{rDKzJtR6Kh9v+ZjBVY0cR$F|J*(gNNCy;6^`eN*U%1<MK$
zX7pHo=C`IWR{RtLO%&8meXBQoQGPkh$hg@Q*Af$FtcrP|9{KtWkfzfS($<z*d$;=q
zaX-<bwQ>63estfrjjf~D`1l)IOK5qn5ee|+kCzP(=QejK8eWYDHk{UE*j4Q$G-jL+
zU*IUHxhtQ4$y+`J=Zbr~k$UC8hDl_6b0aSMaIVw#U|}D7xF#-C?63nQPdRT4tnUG;
zbwu{H7buC;Fk}lGoC*N+Y22pZjc3u>$#$6YVD{hya?z)`N7oIaO(nM2H)ZPspYLYq
zQKNe*LdzZ<8R)jmZ@4?Q6zgmVyN?y&ZWnlGjoz2qAHc2w1QJvUsPHD|t})tl-%+B?
zF<wiv5Ah12NAGT6FY__1Lp`6fGD{{$TpCJ&e5ih7l%CcNW%q<jds)QpH07IhNxam{
z#rTOFyU>wd|9x+nZk>=NB76CEBF>^yeYEZL-u+ZE30t2<a&Hei!VsHX#Nz^^WEoD#
zX79)4A8(iZ&$_ydWEj^T4XmTJHw71#ytW@aH8|lB*0JddP+ggK$v6nzxdVz!joMKX
zt3P&w;mnFF&-j;0<Z&r&5wd~@86bpctEA8;keyv(&v3qCTL56p7!<oE?A+%7x!&=A
z@77nN@*bSGrf0i*67n2wVg7-I5vQI|D++UTe{Nxc2=9&?;Lv9i(i{)xDxdLX=7aj$
zuI(=Ss+jGpF>2=axptR#g+E?i2uDOmU!%qNG2<3dCXJLu)%NLzD9}|Imz8kL^he>j
zPbG(OkB{69CD)+t4Z;f8_1Yw96sGB50bk<ty0mUx3l02>^;i#X<2v958L+HL@6=B7
z@Ipt7?`ZTu9<_<Hy|R3r4MgeR+8`l6W}n*cn#u7;$F34p6)3!=E#ouE9ssiAy}JJc
zDfaLBSq`(cc^y0oYyIX(6n!@5wIw|^B1a<S^7_E4rFibOnVmq`?#-hLJo<YVVMvj&
z+zF*_tnK`5dt6?y1mwFU0=h4vA+nd&@x0jG|0d{I2b-sXG^?Glv3{vd;mr^pS7Te=
zh3@VQ|3h-B5AQY_L}>ef=+9ZgsCx90Cw4=35Xb21znm1PzsxNtz4vOp!l7K3PZM$V
zOcn9U?botVw3bhiR+DTXU*n?zzo8_e({m~<ftWINNdgf76F@N%bR%<x)FTu}=PZi{
z^hFFP1mC_8;NxrE3}cv@mP1%Ms8|!{)UBdxRH37)Em0+dOIjUOPt|9lOpZ?h(0SSf
z_~yn;08HON@!%RKUGUQLT`cDnKfgn%Z^Zf|&t#FpQm&%IFML+a?BcHstJWsUf$)MX
z|A|ljOtfFGhNp<EHmMq%FP<*J*|@4E!5yhb=FDbaX^@z2wA7cb;|5E7CfP&NM;)ZL
z8u-@ytc-HOlLkv=*01UJ;;atE9cD99F@cx8#`0f^R^ME0vBS$*^BQ^+KqS<-LSn&!
zn8VW}qq7K?3a3t7Cdrrv;_7;*MS$?bw3;2O^jV$)@@%8y9e5_>Yt*-%#qK&pCXnr_
z48b~QDo@SU9E_ISd+1!EC-yW*6k%J{xOrmv(yb)Nvj7oHi7!`NiHr<kOTG4@0ooNg
zCp53+KG8BW3?&Bj^mfneUmlS>f6^vbHWm8hy#k%2MI*6Z=_k*_1Hs5oGxF;e3`sYs
z$G$d~c0OMAFvpA1+oT-V`I6g#zR%r84kVx^pE|*-e5|AE&;mL)Uw1(OEQV=-emGm1
z3)d%OX>-DelxC+;5=7yG@;_6HQKCn^2hMeGp3f57A;*)lZ#n=uqP9maUm=|J=!ut4
zj9L62R42!n9qXuA@p3rlMM@X@N_!oEXza&-fziS{oBI1DDk(?A*%LX1Mn-swrExr&
zh?-APbDK()m-N6S4cxADDOLQe=+6Z4R~3GgBTxiaaI0>9*>Q5qr|s+?`FI-vjH+_A
zc<Cn>U50ZD&jfBge`-9}dHASes_gaJ0lf?bT~FbHfCLE@9cO2ez!icx+pPqMP!F<1
z`V5pQ2^zxzrugKqO9^xRADQdufqeL!P?it81R8sT<V7JF>hpSBZyu8R#p@@lkn2BF
zn;|}1uP^kZewS`As@f`=?UfQK8yoH>p8PO&PN>+Ulz*`O)Z6n`$H<Ou$18G-kS~D3
zIyqJ12JNnvq`99XUuAftO&9w7lm86t$Ie*)Qx+iL!NX5lxbctY(JxutUT>6DSjjGZ
z5!KtwI#tB20`wekKr?ju89uEG`j&9WJ#OWs%xEY}Wr?fsSUXknxZa*BE-95<l3jDX
z>4hHY7*JDmE0;?l{y-8$`j(u;Lv{v{_g_$jf2a4K;mHQls-xZ%N|Ex%0XylRFVqYG
z;8fP8=;uQH=TCpVOHK!jW@X-{&0qfX*T?G`0J@gEu;q_bbpBBQSpQf-l=^kSQL#n<
zB?xp=lX|<=(swXHo4cjupS>{Sh-@IMs#`Gb<)wIxc44K-Y^{Uy|Af!JIRk(`D9m8{
z_uXU4E{n?~S;(J71%ZD)?__xqE(X6UdE<Y>Nt7v8)`P)03|>ku)(ll2ZW=wll`xD7
zDkf3H@h1A!zTJ$&^QM@2=COabGDu=tN-<vjQt^y*ruP4Sy3_xDx<d(HMTf|IF|qcC
zNPgT5VQVinz{>v0U%+amD`SFb2WF^R)rhIrnHdy0V^cdrOHFr<hm+t^9A||7!ta<<
zop$0NHPOb!|8o@{o%`zxB{e>QtamTSB)3v^TU0H>!XPmIlQ?e))KhT*_)ko$Y;q6z
z;T`?fB>;mc8;OY_WJtSxV8ddugzhTIL^>0<C!}1w>oI@9VmhKp@S(|grskjcp#jmK
z&3F!l#}0dD86Fd?L)N2}qs;T|mZvL<N)gv~CX*z0T*w5WEeCupgFs@b0z)SWh6d6%
zw;n{AmB&u`b5?SdwMhMN0xlzPb%gEdOFyapI4+RuD_ZLzTVn8qckR+m%O%~UADnx3
z4;XahB@!~~{RTKo^_bvF6w_GGR<*2cm~gg!MbxD#gYcPs9Fswn)n9o2DET^LQR{f5
z8R}8;C|cJ2@?CWjs>mY08Pj&Mr(vjw6J|i)eE4%WI-Ud4ied*3E$)9TDt2G-dKD+o
zOhS=9<Bpu2kzZ`Da(<{Bf6S<UYC$C_Q&`JW-g-PkX-JATz;*N>o_pE1IaI&yl`jd_
zSH#qLN3Yd9uB5H6NTM?$$WyhV2ye@X=m3x`p2;^WDkgG9o2S2Bq5zJ$ENQ5s*37_A
z2Z5nFHaobldhw{eA(1r64xW79OH8rJ7mdvI6yj}ldSx}Ti!>bjsxEV!tg2~40Y|79
zY_-@$+!M~&5dH~f{H`X8XotoxM#c8(o1>2<il+6>x%HNJn(B_Qd}$rjTrrD~U?+i9
z0YHFGT?|HF$2vaY(--2E$K5|jz@qnR$$5Zqlo%RVj4980YSWJTmbQ)1nJYO=&$7N5
z?tE`2@{*MyDtQ|JCH}<fn=j8fN&C=(jBdnA4`{TD^gLj;`I_+_zRb(9UnhzIx)>lQ
zFi1S0clu)h@%7ijdz3$q^4$MBCXr^0am7+ASC$~b_Ep39gpg_yvjy9`o4vd<Lzq7M
z?{SEvVO1v_xo{>)8os2qEY`ajSjyj*ZdOd{2vq=VcPdmmm3*oK&l7&6ogHOmy)!jE
zefCRenotO71E(@}#tGewE~PAs?J|pTlYSo~xkI|C#wh^I=b6LB33N#mbLq8}O@>jg
zbFR*?d%M}TsZO9I=H2M%Vui8~(~q5)LM8X@z=NOljA@Qr)k3<{u~(RjJh2R67L~L=
z<JLT;WJs5+<2u`WRoF574x6ZQsNyqnH|Da4Yiwj1zQV6&pd%F>ZjgW)9&>x#bH#Kz
z=lV#icFuR0(g_#HB_`5MpiaKKg%%C%hQnB@T5~kN`V?x#_5-jN{*4w^;QNy&f1+Zy
zxTeV6XlCgrD9GJ1i3UOq$&@QEoKW8aGIrM5n)Qk-6`rokh14vBTNU|+3J)S30YsL8
z5>#8Lbem?v%TDO8U6Fd`*0NENoqP?y(L2Ff!SB)bGt({XM+~QQQbIb-i!I=A7>Ly}
zj*3~-Zw5MUb*zJeFl@mvXouz64XMurW%h;a^t?A|m+Rnuykn1TDHZ}2TO*ueeHpg5
zwcAI=S2CE_=c5uUs0mE>q_F^r{@{D5Y>^E6xz~s_GD+`&dAfdCBaT;DB5Nh0M)_6y
zW@hmRDlrc{qM1nBrcTGd%+n^SoyvX$XQLAc^M4voJIq<M37G{AR^4kB&bWE8=f3gv
z4KHVh>vYp+J82nBNx&eUc)vsMd>MVL+K~~=_~YzjKzAvcB?`d##&*txof{(EB^gXS
z1sx&X#*XWrna_`@3mV&3?)AvftVbO0{|K}s@CxPT&bhN4ry36=%qnG<ec7Dfp#q)_
zpeT_&!ya*<7hf*l>;7cqqp^~oJear8D6BzInHK=0vvn>py=W5(BuP?^Grs<+L<KeT
zfQN{i7%*_u!&vtI)w&?5brWaLOdFte8N)Mzf<R!!YSayfQcSi&`q+JIk4s`oZ#6zL
z`zFJOHJjRX9<<zTx%P<(FL=&?1QJXkXYt8}?xJ$Fq;y^LqWL7mhOc^xVq&0r{>Kvf
z%CY*k{+Y%8XB$g{4fOh|Xsam``j-QTnPkf{AJz0No<?!D{=EKyV{u66ZhPvazl0Wa
zaMp7Nz`WnF_z6P*tXOq2y@7besP6__omNRFM`FA6!xENnSoI_48uuGQg_6<37WJ$w
zg4Xl*Nf*%@rjtHoSig@;X@XYfs)NXDfms<8iiqB`7Wk6B5|y$s$f|DZg`d**?-$v9
z3KB5><#>Rb{(_pdre&wVnJ?HKSzq@`@3E{gN`}*8j3>5e%AS#Q27pef9?Z2*CYj*s
z-BbZEC-fzR3$T829}jR{(xPb18MINT@-!rE!GEhQK$?Hh$+0=x6{=O|5sC#G>Ej8r
zl{gs7z+VVd8?-EkMpm;U8M=<+jKI^ZB)DN`yEiUq|DUK!4h}{xzpO1B#p-ZLU!9N&
z;YGi*jy0}=#jP?SpZh0Z^SvU6hLHO47yj&0!Pd&T3neW!E3fiTEGtj8kzmoJz3}rT
za-N(K0Qiat-F;;dH{h8kvby*&te+ylH3)&hfwoCIOqN60KVZ6q<rD4V59ly(=5As|
zPAsQBFcTu_futXhUOqXHGXnWjOP^&Tvs%gPD+6$@6!mCXIw=%85N`w8`bv8uz%idP
z(RXhwIU)aw!_1ZmNzwez6L|PFOpd$xyy<*uv?6|V3H$S3T{PR){Xt9eodVh~|4HS<
zQ3R-kXAr3Y?pF|ZNC-?04RW#oSI)X;TL1l2**fmXafEsI$(4YX0<yn{{htZ&lf6To
z&WUYt5cuzD`Oo)fNL052fj<Gf|L88M<59|h2Kg^$h5s7KzsBVM6bOITUvB%kApbSz
ze@(nLpqfU_yE@$YW4k8+ZWr3kr>B2w5@^Oh{T!h4YNG<9<^DbW|CvTv4iatgXDX*Z
zKKFVIFl*_RY##x9+^<~%`b7c)21tGQBmXc*Ho&pW*-IFE{`*jYZk`0li<uVde`0|D
zy}kaPM<q^R7PIZBxPRHme|<0e8R*ng`sqJD_YsNTM9D?sF7rQ@_3x1f0Hy<l=dF<6
zngoUiDC_{LIA^H-7j6AtBmcjp{d-3LZ)yK?&i?Pz{#!2m->LogRGj?(nLw5eIyHub
zpw)Ep3^@(Teo$K5B(!#P#Q7c>Tvdq^5d29d?urLU3uurkn-l<H@fv<hW{}u{q4OoI
zRs{p0px7keouoXyV$sCGKT>&TkeH+d0WH(j&Ly&BI5&)%>)^tsxeq{VNrLyM6K+Fj
zkoNs4p^&Y2gIw>d8iT`L^xV!GNUBxk$P>0|D_?!lMhzkC)b5Qec3(;z@V>xr$S-8<
zr67QQk+;9ozn@Ha2=T5i5z1sbT_J3FPv}qg-8+gqR|}L6L}GY@YMTJ|^oD|4Dq=G~
z9vIFT-*)WQt^g7X4i&zt_2=s9P6JGtmW1*M>78M_(do!O%?xEm^kNp0ta))Us{hcF
z5C(e=z{7@2UBwSC&bis_J?r4^rerzMi+jzIXYkn}r43Q5akxzs+;P~nC<5@?)%f|o
z?3)1p^|y9wWHUf1+^L3=n$WOk{mXgq%VqfH!lzC$$%kq6$lT`PMRT(Q08ijCSTxnW
z`#F5%MM9?;oRXfElY|b%?snV&KoDKJVJp)c7Z>`FThl6cQZO)dBEETP?cn29MpwAR
z`{!HthiXirLmP`y0|xmRU%~1LaeL|R2i27>RmVR1A^=Wl&@LT+xegHJ!snRLX6uWz
zV!i{|7-A;%m^;ItsP4W3xwJwLSS%4TzH(3U9<_=@&IGtuPQGE}3*2NEpCz9INC(k2
z+plt`hyW9XQ>9A`fU3ut`5tVb*6;fICHT1Xoyyii$3B1l<<QmAsju_t5PI2{!8!k(
zIY1}KBI<{lxr_{O&3>F#Dbgc5NkF}Q2Mmr0={=)PBBvGG+X_%9vRthN_;+8Qeu#<z
z2t>zq3Q4H%|I&_r>Z%&#g%KBH<ZDQjiDE`R-zA6zHHT{XEZ%7eTY~)*5cEc2J+Cc`
z2LM-b+Y`3!9zgU_BNI>tp^(KQHrIiXJ?GXE<{TCDVvp*4f5D8o=~@K<q^;PjG?)UY
z7d@I6u(=@s!qk&E&J~N0{oy2JT$bxPP$O3kO8=TK(Mn}s%OL^xOQ4+31o<{>y7zxJ
z3H%IlDji&<&Fo850b>30T-SUKHz%`FMUDY3<G;12zu(C!Fnu~hR#oA<&j}kP`c?Of
zH{_{^jfl(5_tTzut!R8t6JRtv(bf0i5GQP<fdq1fD=-8L9i3RZF}BY~(m{O`woep=
zJiH5Pck&Ir<}3?>$MVsO*CRklKPVWkCBP{e4?wo|7d_1eKv>WjhXN@WJz8M6feJM0
znSZaauP^lduxMqCes9h@QB#sr?3ey&q?`Z<$JMQ4Fm6!u6xKZ;+UMCSz$~h9htERU
z)wK+{<3b{(jLG9oJ?PrOg42Rqr$LAef9j<<fHHp(OqIm5kY0}x8v#(1Kqe6hUq3Qu
zDKxU6Xv}rU;ydy^lA6s~J|e{mNLnZY=(j^`9<L2@s>HU480;}g)=YM-Z8yH7+Wg?K
z%)so~NJy=J?0e+S*%2f4>plQf>7KKsnFOfkTL!w~hR|g;J2YLg?=qdxNmp*dhQQt3
z*f!a-bW>xE5%eqyy>{AdGM@2=W72LjUz3s22a76mK1RvD5|D9dtd-y5VRpoTE&fiz
zp!evtFTJYRx<Ne)=fa`)Ka%(gD(YFybhfaE3r&XJ+1j}}U>r;m)7OM$aq;#{Fx|!R
z4Zq!tpD}N_i2{}AL6ZqZXDI@WUM7ZY&TNe@s#kw?QjInRiSr^U&p_t82+kE7WuC)%
z!*`c>?Zt<`u-a1wggm_5&Qm(>_2l=O9E@7Y_klb>hakD*hod9jb#vg}9d?=c7FCbs
zYS~98Z=`SCe?)fwp&^5b3z$K{aHDo%KJ~EQg|X}FQA0p}D2*@Q=SR!bSER07zkVg=
z<+{<M0-@-|oy2%2rz8n=#Pj>+iF@}7`rC+-D5<TQ<%BAuof@B+j~x{f`&ryaEW=h|
z)cskoj<Mn<sFqg$pw3M=RcmzhAnHi=`{#o4^6MuFse8eJ=oRtWs!xlT8WxJ&eV3(-
zg?-(3@1f?rM|UzSh?@o=6O)v9X(zQ?%rBcZu;~|Sy|yhfLjj5hkWlyGSP1=z@llYN
zs%V6yv+w7KA)Yns-M9Cq{e5N5x9#-8dm2Oh^!t+Kyot@Vl9M{V%ji+z1o1N1vWuEe
zEFP?4Az|;%dr_NJ@>7!M)K5>(FXn{b?ZT5Q4PyE+<~0W$#e!TX+<bP}32u&Eh0FWm
zN8(dR=IOGI;VBues+@(no_^)P+(}Um1W*GCzw5i-C#Bd6WQ8u~p^}ceN6}jyk~;?2
z^2gV<YRp=v^;}>stRX%((!D09vQ$+hRv%0%(gli`D$<D&Evn5g`>fw)s;ovkyhaLt
zw7l%QXd)2~fx%AKv?6)W`5t0@mGo1XE_K#C-kvFtZ}mnQRMQJCn>Dv+nVIl;s=-D#
z^pI+U=IwB%Jl0M@$Km>?6!w2O1s7?)@YZ2JxQr*yb{QYEzb(RxaKB|oI4rPx{J4$H
zw3!}w$(TaFENH;I`+i#|#=?8gaQZ?Ht_@{iUm^X;E7!KUp9sP`G!}Q-;mhy^Ii0zc
z$GNRp)Z0Tsd@bY~!$c+b7n#mk1c2)@3xV9iwJ5f$Ch__5GSDH|;wf^;FbEUd<WiyW
zK|_#R4y9^;rSsBQrEq~Bx;;2@E|(q47UJ@oknxkvrv^Ihxf3JSimw&)nKCfn91@Io
zwH-RKVB}M}?^^$zrY8Q5OYo4qQ2nl5Zv5%T&Fyu&*|ildIt`2;xWZU1aFPofIo;@;
zVCy!)%ew_n*@Hs1j)uI=2Rh~Ii{4c2v}$GGT}wMWLr2AYy_d|Kh+)xz@^9J2#NCpX
zK+8r_7<%6oTPz^Eoe_n9Fw|cq$l8!srKUb3E{yJvfCG0{r|qf^p~PX@0yui<FxLGr
zew0e==-pBIs3C4LYRsufjHQ4ev30;RnDcZ{V!6h`XJ?@qKeJGfz`)<Jo(a$-OphDy
zUpFZ}PDRh@)xH^Mo|GU2^LL-1SFc!I-uIcu<-C@X_8z*)jBeZYPAy<rSca+Z-!oZ-
zh@QDAAW(MRarg{>lGF^F(^mgs*C`gaik*$Lqs#nT_C3h0T;rp%!3t}XUa?or4W3sR
z(>ZE4QvBhU-J(hJh3f{XML2!G?XA&68Dm;5eGOC>K<iQ7a^7ciEMAGZCrntTMHOkH
zKw&zE3#LP~amyS~pk6l=N!T4duf((>wPd%O5caGtzDR4nHS;z&s|~iP{R1N3>KP5V
z5%W=M6l9YkP~5c{=2`p0eDt6d+o7n->7|1ju0I~F&bzw#c2J3XZToxmsHlSI7c@RN
z5L%!fKV4>E4+3i5oDVOcN9vb4Y$-0!w0SmL4om;3Dw1rrEI)ZwwWrJ-ncbGar8hF3
z&9-uWx=)@wy}Z4;g`rC-$71^C*=fL`3DySBj|&^mZOt4V@nH+)(KKu?hBn!zWe$n`
zQ=0%`jYGqAMx1PnLu+h-bBrrZ?}&$|PQ^hg?cqjq#^tW5t`3y<T2ne%Vp6rv(C7<n
z>2$n>@g0`q%L`k>Q%(JQyJAO!3`051wbhgMZLI}rjv>|1iSpTT{LE>>5d87Akjho+
zB+qdX@~U)Y{{N;_`?)s$<uqTe)}fDP=+U<b`5mJs06%h+LjmKfZ1hZj%^C#}(R^gN
zxhxa;`7oJKIe)ycMOU4tzq(j{???T|CV%V~y5?{;74aHFYYC;`%IM1yKEqx|A@m<A
zT^2sr`y8|?PWEKeu}Rhq@5Fz}sXjc|{jk6txBO=BE4tRi$ZIF;B{AIR;@#weY@i~8
zv*Y&;l~@6K@^8;s4;Jemw$0P`v9Q6uB_9gZVO{%;gz7PObMqVLizGdIZi!h1_l2<u
z-4|3E%93`Erd>(&J~Sg@*?l`0E-+ljhX?)P#1Ef%Azg=UEB5(VCg?rhp1!r!72-1=
zd$fOeWMFsDV<iOf%hG$6w2f(;>dP`nzQqT&?)jX8SRDWME={o)3KcVgxttj*wZeaw
z+^(b7A|!q`98aTP5x(&i9NkpGTgD>twqj{oVz{jKh>(`45Nb}nm$A1}AvnT%NvJZR
zX>-q)K2C>f#JhjyE$^j9Ld~*QW1FW%%DWkZK6McO)t>PPca>hlUM@FePC6Ir*wj|R
zHPj+_?sNmyug3^*s!~w^gRwRV6a$yi6*Af;h`ZcsY0BRtMt8s=Xc%{OHQY}Q#Awn|
zS+bi@={jeVkC5#0buO@$F|=1p<X3)j;lN>iw(x1!_g-sNZ02jmfco!6Vj_eAFP1og
zVeR;>w&+?BykNyZiCyoVC#4QuQDfd_)%RQ%%Mp?Ho3-Nfg5Im|AGW+~%<Wz-WjdOO
zwbG;&p?D#_rHzye+?CTEo-y`{IVQ-+(E2l6_upcG!<GCVI{?W(0Uimvbo=@g=AeDc
z=Y1LalWWfh=eedZEjK2*6}!V9Gf`@5k@FAkV4%4N+SKLHxy2VD+vQw7cb#V)IypD%
zsSY{k(iz0>Rn#6%mppy(X2^zVP}R2kC7+?LA^3cbN5NHyj}RN;PX%LK7{BctQx}`g
zg&jd~drm`0F?gWphc{)qMhH!T0mLnzJge^V$Ie=5h3l!H;#}VC+&0&g=3wi=qKe0`
zet`rD_(-)!W_3x(A+Z*}8(QfqI@uJiibm>$7fw!PXR{L4nVm3Ss7>(}K7a7uI$n6C
z;1@SXIeKJ2%7<zxnJdO<9TBSx^5dUmRf%912aEe{a3x=dkJ9#3v%(3jg!7P<X+HIb
zOfh?Tw^zhn7M5dWw${&E@lmuAy!<chV`aQB?Kz0UT@kgBeTvXw<`L85JylgzZ!L!M
z^t8mT1aEq_yQXp0wLWU1{OwqrIZj4i^$pZ_(Vbpu?ZZx&`ekQp;r!u1nLXWNmjzd+
zjg6!iLIvT>zJc!Z>Jjnv$M+OA^DEhz1l>nVXu=}m=2T$>T=a|RR*%g`73~F|bPMty
zPvP4kXbiVq&4^(8wz$;<qrkS-#^f9}3q*(cBKHUH@H2i{KV4U1XsAkC+aR$`@Mr;M
z*Qa;$@|{n$J9uB2A6%a`;=1Az4+1MTGds$atE_E4z$Q0z5uUZX3fL^zGu`f}sB=!+
z-`?e2<^zxIxc-=Z>@+qecXXRzoq#I^bUzcEZ3CZ-%S>D&d&iz9CRoCBo@hR}2F?WP
zE>Vx7e0}GwtEO{SGt+;(ub9Lpaa5^jOdkKuH2DJN*d$AAGT56cU>B=(-&Ys-K%{T5
zSgykCZ-<P6PPPctd<u2Px99KTmyZTOoh!gy&2nJrg(?_z(!F(hBq_<GZ~p+bL=5@9
zyRlhYEwNzh3q>FJ96F&ast<w&G@><5&5rhaZxnab+DJbsZU?EZ&Tt>i$MzN&cL>|`
z+#2tBz7O1L??1n#|7{`tWIs|oV;`HFmRZXkrHlrD%2^5@vM)W)drE!R(0RV6et-3~
zF+9F_I)2Y*<$ST%6jZk`OSv*GaW5^GMZ$6aLASTyj|K{-ky{wTMW?|s!BHZ;Kzf_&
zp6bv|Z{G~}?XKo?qUcRHb$;b-lPm21eWL%q=g*yZ#x!Q7DuPMP0hnvN_q00pi89`e
z3<>;2rlk-I;Z~@qZO^xqX{6X8l)KWcet%qD9nsA0d3SbgXKu*uwwNIdEE;le!FIkk
zH_mZXv8)nfY>mwf0|ioU2rW&DT?7<tgj;IAS=_F|4P%KC`-Pf^y-wLUK>Wr_d2Utq
z?zr!mx6PWEPi*sP+P44X|Ncb<@xSBG;+pD*wfxoU{8uQUm*HD3JnO68AC4_B<9;Vj
zuDG54)6ma+Y9K%12L9hw8^=~p{jo`nAc~VKIV=COh+^F&^Dpj)zdk0@(4H9skN?lI
zlzz=Zzu49P8X@Baj=J22|6L_A!s74c{Kq)`c}Zt9=zkY&yxj4N(etm5U?84UK=Z%L
zU`i(bvPA#-m;*$0j2QoSRZ2I>KUcHVHUR`0@Bgmzx#C9okAeN&`u_(D*pt^Vvssvk
zo5MD}fB$|UNh|Y5&kZM>Kqvj5N`{W@)AoFN+&IfEBiQj{W`TRuNBU^gfUPDam@fNS
zf@VUI3FPtP*Ee_t@=tGtI{l{V0l7feG&G7cZ}5B*a~JIibqaOygH1etN4I=(X=0zM
zuJpI$I@w+(9-dEDP{s?|2j`{KEiSemQwPJE)ft$pyZ<zWIW>McE6{%4S7stci-1|?
z2;2Xn)P&02^^N-oaMb;}{UbS&Z?6GAcQ*aOnfGm#zHnaFP~ZKbTij%aaNO?!xqH0H
z*_X6GFrtiN3Tak)Y(=M?06Cu|822-q>>qpG@hlJB>YIrd6-(l#=f|uX-jrt)7kWfh
zkVww|xw&73`Q5o}Cgo@a*yqk#@wTD@5yJhyR~U0VqtSOh?iupQ2{#1=PDeMp3dP?Z
zCCi~h?%3i5Xl2Ewtnk!{i6>H3hWx(|`wSJCne=y`7u8c<e!80XHz4^kUkurQ``s5^
zj?;tcr^u_cR=oo%{vgUYnkkf(U&!pW931OVd-|ta^~^Vrf7He2GJv*~5HA04c6fhp
zGUwMB0GwEF$sND>oLsphO=<IwY5doUX@E%*YCd7S>i7Qwhv(I6*>>HruRo+;#Q(O$
zz~k=m%KF8niN&`5%<cPS`oGZuXLYzd!jR&(@dM%4aGEn?Q?p@HMSp7dg}mP}L9wN4
zx+Lt|Z(A!35XMYnO?D%djd`QxU%g6}>nOH2zwdmH<_>Mk$9rP3wX%w(pDMN7C_bu_
z12l)Oye%cv_`9^`#jPY|lu}0CP%$j2tK#9%;qj2RLeu87(2m{We6y{@h_^Oak%ZRG
z1aw&bks<E&d2akpmXM^^wq^J7J)ixTU6m6PQBL=QXp8I}9Se!r9phKoO<M#ROV`0Y
ze&640$-$F7ZRXd?vL(qVjEEQb9qZ)|c3m(c=M{BF>g6A5))ln}zuXK}jZmqCS5<w<
zbNxC`kNMuapz7n3SmC%9lOk2@iFPE45=XjYG8MyEru$gA&4!iaTHd2(;DDW9VH<L4
z)>PcZ2H{w%&C0J|)Q|^3we;i~G^>_BE5E_d|EP5C-pzj4EV?I{e$E1ivYi@Xm+}~{
zx@Q}Xj!2Lw6AGkScWv#M>JLxwY|C>G9loL%anVvAWs|t>V1ObB6GtEueURJG&sJSk
zR^5c|n>`0^hcT#OuC$P$WhnR%?Hn;vix!@f@|oM16Od#ZSiXZTbqaAd24^q4dGn?%
zv;YKaz9jJ?(h8UUI4DBKY3VA{duTIR+?Ftt%iNQ3)8{z3j-4_B`t6d{=!n$0cBjq#
zt+vOkHS>cBl2)pj@`a=fsef!MdB3gOGA9ry8~Pldc8it5A)T<B28B%!2r5)OhIU2!
zKuc5eMIWP0j<QxRIJdLf9MHw`DCFF-Y-a=wq0n9dRc1{y@sW{;O*BY5X!AR(kj<Gp
zI0go7yI0&voUJ+)%h`IxnO6OoW)Awx1H`RfgE+(Cdry1>?8|U<4?!uGt<xCI%TQt;
zq23k1Z!1+m`uL#NNAEz)5^FbqcqZ!Ywl{2r1=PJMsj6a`j$PmuHzrsYD`Hb1)>_5p
zHM!f`6bbMpb%h}~y%^;<1gJ0Caic2QKT#&gg^K>NDAj2;(U8+@g-P_>`w-;*(*$Od
zMq`l!{EaFt>)tyXQm(8k06FnLro{gaaLgih9zQ9zGhGQ|@_rkmQXi+&jfk)jlP%2*
zV>OqY_z1f|1%MO#ZVr38RDZl@ky7!Zt@Gu4`dcc#p{M%;Cgx2aVVHJ^Cq~I2M#Oen
z!hE9q?BG&juj}$n2%~cXp+75xwyz|uLB+*?9R!9Bmh5iKe>f|@mo=;niRxt$$+JJ+
z&yWp=)H=cDU}82JGIzOAllKla%S)Yi=ErzyJWB#{ER5W)J;@i7t}`fi3(Ebn&Sy|&
zX|71eIwk|Z%38K#8yLhqvTf8wWqUF6m<Cc0tEkxTDPW5a#!%1X_^lzCi9(k(uawe{
zHLw;G)erBPU2%7F->$y@@L>DGX`!JZji_H;+p?m7m>A?Ss|W+g^(^HB-eYvm?Km!L
zn{CvGuxi@Dp$dGgk=n|=;2gn6#fGLZwi2)+wC2sDYhyw3)JW^0$_!*MpLa>mfpaJu
zQ5`g%(7*=fsb1}~VOr#ot$^-7N`roSBfJyFeEVD5M|7vnXyH&e684SX$2X>8*i|?0
zZRU60Fs)BIS!%Y}46VJf&-3JLb~OeLZhA#qb4{&UAAU5K5b@5@Y-!HJ72Px?Ww!d)
zhkoS6-8cyn?MpiS0b0>e5p_<ABIBlb{!p?{wd$4~Wcd4Fndj%e+@XajZvmH9fKrTi
zSfK)-c;)IC8TxL=XB)R_IpEq8Zu}`p6@+h#F1748lZsx`fVoiA>`HVNpSMHUP~_X^
z;IzIPRrZxSc@AL?RF8eIGwAz8CH|2Q=x3ZG20;b@tN}CsYB$2CY$%xi)4U*sE>kL|
z&=j^4+2UK<Aaw8&kuwP&_%0c?k>Oa^!+_iV&JX?&R%luUN;=42fNta7>n%A6pP^<K
z>fIX;+>2*2PMFSF(U$AOyIO<=b7gOWP+s{+mh~7@mKh96PxRJ$N<_ODV$88I6S?#x
zj74N6gxUzqsH3fIb9#`P(M!TkKEKzxZ!aR#oye7`7u~c(Y`SStnFcMQ4eN@_VH~il
zWn{Ma9Lm;Opp+)W7z5{-Y~0B3U;p-KDtQU%C0=50Dn7_Q%7XJ`wPQhicE@%*^?Opd
zx!63Ki5m<&fmn#|DtCw(e4gz&9q7b{4kEZ3UIPWCmEChp<(XYWE@4~rg*+j<1~vo9
ziJsHKT<MARa8U;p(fq{9eclsWAb+RKzI+g$7D|B3h>QD|%EuJ~k^8=x6+XQ)bGtOO
zv|6rK>n4#$Q1HS>kje!v9ve+JH^+W8(Yv2MQd4Mqm$e1Ii?%^&%g8-QL_SypD;^F<
zLqf=J1ic739jC)|;|GhF$E(q=)j7EtTs|6GTHvH?S{tA$EkQ)v<~|wtVc4a<QJ0|i
zh2dg-YKsm6AiSveSlWU$Hm9T;gjYbt1jUH^f>OtZ_V}b7dbN>@U(SzvCLe3jUCV~^
z;*`;IT;Wp}=%$2qQp3Jbr5`(Xr9tU|oHqSRpFIzfrpM8%{DM^GQJxb$m7C`lK`eAp
z<iSwGRxH)llk+ayX9!<fqL(C`KbEDe9yRQtpk2aKl`F1H6V(mELC+8E)njZ-7o}($
z8X34V2*&PC=AT_1j((hf|1IO{<iMQKo>Ny%HhffZftF^6TRi6cKmAPpcd1uu8<neu
z!fb=UIm~*D8Uz7Wv73?zD>JWZt~`w`e1hw#QU^C@(PkapmI~SnR54C)#@-Nn-3a!n
z(>qNDPBwA9fp+9Hm{=CO`Ez&8b`5pG({Z`b<-xej?T9HOMsuDL4P9k058eD|?y%_G
zg-<GlIVr-0K~0%jOc3x+0!T|t1>B}E#A|SNYkapGit-SR>}f-emziKi(*%lGh&R3H
zy*Lx{0gtdy40PQ4V9!z)V~GO0fH!-6eUjc(L8pfuC_aqpZZ+%b<lGk+btUV)7z7kw
z7lj{FsE?~$3Px}(X{_j@)YwpJrklOUAM+I}TV{Jk9q-6H&G`DSO9qCqh&oITB#3+S
zPNwuI?0$RKy<lsN_Rd|{swivRbjvBz#uA0xCuZM=X#3?^eC|~Hp31BIn})K#ykhD`
z$xyYuqJ|fU4s$Fs8&YFp`II|1=*)?H+uC`Z#>v)gy!p1)EJK2&d`cKQ5vh{pD)9d9
ziyt4X@HrI{_7pxjby`@Lo@}jF8(fu?$;XnF3f@aBGOnNE#k~`0!x#+Lmj*GG{3UOM
zDX!O4*X|$JjhT_L*vlDvXqv%k%~GHWVvIa*!!+(3t@?6HH#yz9#~XxFC+S4eagcK(
zy^>qq+WjH}f?oD5%3C_7d3Xm*`O}YbwIj9AJ}u@4i2ER!*wNMV+yVI=;$jqrm7{Q;
ziHf)^uC~;iNy~=d=OC~h^hL!E;3-E~R-sl-u-6o+789foh8OG%YE{{de7rq0>TxOF
z)=qPrOq`bl5%h4vz!B2QGbpuMfH0kBlyC=2@s;Mu-o{+~x(~l`X`c;Hm-VsSP^Vm6
z*1o>&Of#%U(Q}K<VTkt&7@TJ><!NYhfuA?wt$eHEDJ5@Jq;9{NNKm)%KW}Za6W3+^
zV2Me`!?entcHP;!N%JU}UhOamHcC-k$QEcbQ9d5yq2sx^q<+$qrKcdy!HeQW&MvjC
zg_UH{<FfC0WeNU5!#zRTDWUlpkfFZm7_m%r&gW+mHF4x<GS{kdlx=Cs{V1|H<LQPv
zXF1iI5(4ms%9iuZ1K+FNGqplQ8BZ_wcn^exi|bNwD7fAaCO4>mZ;SO_i@q_7>#=nm
z3h!Rsb3O&8n3veyTw1C`K+W;dWuaCw63l@O1(sSIxt=(jYj*Y2E_GKg<Kccij!y1F
zmLFFAT-3Vk=HYc~ni(&bG;(Pf*A#LAiqADW8lclO>{t<s%3Y;CwIp#iXXdc{t<iVT
zof`(!9-_8&Q&N}(EekYc<J2ah7Y}LN4KoN_8lCsZTLFRJ(+Kts=Y4a{lVm@<vDot@
zd@f&xWvt@(pmC)rJ2^ByQGS%u00Q3-%g=OLs~`EHWNO%N26o)FxaXxle0u*E+vFAU
z7}<z>%?5nCp6U|Kd8h(;A#s0WX$77wGQ2xLTmhK`(o2@8_CmzfT@qCLyu`ZAZ}5KC
z$DLW8r%roIPwi42pRbqlI-vg0Tac~^p|Xk3j#X$5z>3^7-Y^L=Pw&0!o1=zxbGa70
z>4d-ENS4@;_qdU&TuUP+xu8ljMsGQB>Ub5XY+e)pdIOs+Kg7YK>_RseyWqh<nR8=t
zy|^+H_i)GpI&RlMLlqsQaBb1eExW0P*sP1{gb^-!w7|pXFbt79Il|y%*Bh+#D}D)5
zZdQ7wV;mBFh6G%OzvDrr3nrUC|JrUf&5eT3rX0}WuMUh^ISNB4YqJXj$@=3t<U=>?
zJ6t+mRX$Zus^<7~3H(XSZoM;|(PMmB>l@1cD1I;+h!)$X%}b5y4=6oaH=MJ#PNplx
zmag!Z#&o%e;bFhVG?RBvo*7GM_RPpHW{CzNeGV!WCxgMgnfbvq!P+;O*g`Q$#g&vx
zIEaUowBPDNLjeoA7546wVE-Omgopw4hr>+svstsurNGIl`5$Q3_FtauI_UNWC^K^K
z!d5SvX0$p7W{Q@W<v*-Af)g^wG)<X#dzAZ}y^J4c+oi>A%*NUX2|a(xUOX9NRaTTc
zfNAo?gH5CD;t2P>b(LC_hBAunn;<sh2Dr@C#qhz~=e>stts_{)1jPNnO_=4j%`=?#
zF$T+ILTT&{4o*=8&kHQ_jW(E8m6n`;woNVUtQuzKCJOR5EXrK&-j2@c)c%}*n=(zP
z9;!(Lp1$!J(<4v}G2KYZOxA%o_pa|hY%a&1D#`hxqT7|_EuvLuBg|zYDs;)3{TBf&
zy9D_860$t=G#=nJy;(&{N<3P95APat8{U@uV5U84(Aplmz*o7hXX?DMIMNF~Ip>s9
z))#D3Q>oJXfTdM61`MUSwXO#O)G@+vKxcuL*<4gkQV)0qPMu9fq$k9#AuEq|1wy12
zMLyvin-*b^yT#Lubj-QEsZ=^{aru~!L^}uD`TTgOp#+6U+Tvn4yy^k|sz@1}-m{xc
zW>7px3I6E$)Vxb(HUXEngr<3^NT+hBno+dXY*d`@JS={sR{+=ubCt3o+(#OnObJz2
ztJD=n_}X-$L+O|A5R9K_U_bG}jKU<vQWP@@#gGDCDa?f`<f41856vSg+|nr9P*#%V
z^<Egiakc0Zmu}ZlBa>#$#leHNi#~9ryNJv)`YEoNIlVUFUqyd$1XNrD1nA|Zj0SYP
z(*;U7b-<1M)`!7phngeRc){Y5lu0v~HbQ}EM6Fk8eqOIjr`9Op{bc=(OY<&PR*a7`
zCxdg#8lulz%0P;Jdh!#5`>N;rx$Qp82FQhxWp?HlAqs-EdZR0sb9%G2hJws#s_wXm
zA;0Z>QtljOrBIn?SwJjv5o8v?rsU8IS}ycPCg^q7xq*@P9({=_r4=K_z#)0lC95Fb
z%idaXKkAQrpqFxL_VVwGh1E0J&XG$>SuBX!P&Wl)YiLTem9{qXV5;SspDiLnRf(A`
zxr6&K-jWS`&Oy;^E-yc93Ab<|x8dec=H923d&Amf?xG;BoSvTNm&U(KOC9A@=2@1S
zZ(mx6RhuoeuUsu|Yd6oE8n3+>^ufbuVX(G2H~&)n=W1yid=0kE;B!w<r<k$W6*Eqo
z3O2o8TtbmVz-Ef^3WY6}j3=-;30z9;xs`!m4Zkd5KJrnNn;@2>&(|x;g2=yfJiCND
z>bV(O6lwFM_?`)<!3B5gL5%z6l&KY5q+k7XXRq3mF~t4z6r4FPH7(Db#_fw*5G$Q`
zpeRHoSp?g?Z$X+gkRC6_rzuFgQz-uf55O?+9Fn4UV&WIo4QFOCVyo(JtEPH$2gclc
z*PmvWA(K}_8=%XEwuQWIsq_~p#6JU9_oqM>o~BY@d?c3f-Gd<qcU{D96OwWoZryc^
zuBpu@SNCn}{(RRxZc?aO#Vf0YL(~ojU9$}`1Q}GLup>^R`*%YRoz}s<DMgxopH@$K
zKe{}UpOzr3(6Lw)W8DziTCLmrv}iD`bUYYjU^sgLZxw{B?5SQUrnMg~!FWsCEklYt
z(*%ku#h^5-B6jS*`E{uQC;o&}a$%`C2J-&Q2zX*V`Y@~ZPDIk&lbpoLjV2{(MD_E=
z>@-nlZN}&VN}@@u=ttXT7Of|aYy5m!=W6)Zt^nrU-q(b{Quy2Zy;d|1!o_(Fyg|Y<
zIAOlIk2|3d&W{K^=OZ#PSHtacxPsWQ@VkJfSS=XF8DiP%0%r6a&-^TWDH58HFMnl>
z&%WK+B$D?EnC{}~;`S46<7v!JJ>j|2$De?nPgp2piYM8jO;6dZw?@3{)mn8E=Pt1D
z=Z&!*fVQnW%l+ajc^=aHZhw2MTYnKE(7$q>QNQ2~Dxr*~q-4TA!c!n^PeF!HyEU`r
zuAX0(MO#E#*5E6!NiI|aEC{XOJ#>~4SjIQvV#C?9m(VTn%>vT`mGMP!&t>YBs2jiJ
zVUlWP8FAuaLD$At)KcE@+BW#Vs;pB}?siejO3smDlaGeHz1S@?rzcW^cgQsuvO`BC
zMr10TWw?~%vNVuC{K15#HJ0CiV15Bo@;rleH1Fz6id}eDyei{o%V5YTd}v-QNQ5Rz
zSnhsYJ0U=cbzjs2{CTbsv$D3bCyF$njM4Ym$;1pn7*mcGjTdk;mOnZV%@qgTys(51
zzWr(}le%b*+;#b9H-1R6cHU6MK0oVXyVc{M6*Dm}!n)|bd24w{&qN*frp^KT{L+X)
z$ET(YTGYyq^W#%2RBAC!Ik(>G&_L0<W!4S_iFm^SanBJ?R|5&H+yvDJc#mG(P)^6n
z6Bqr=WT8s~(0Cl67RAelM7;NTv}yq9<U2Rb`*xG;d69g(!RIdt=x{#WQ3a#(eg98;
z*ZtQ-w#5}?)hNnJ6GBy&5)dpD5s{+80|mq+V4(>jH8cst2q^jxMTsCHLg*#HB%zZ?
zLWc#Vw+JaTK}smHN;QISa9>;SPuSlwpUj=P_slKle9!lOuVOi(_eN_E(PK&E{yWKU
zh4j9{&xo?6Xn}BKWSm>q?kbu5jq~RbFy&*iWSthlC*67<h#;;7-%W5kF{U7C8|yl3
z!dI)j+LEvF#e=|+FJD~=`J@o;$~?$B)+~HdvB1-F$MBL_tTD|Cc{#V)#aW#{W4D-E
zVEE<gw0DQo&1@rkN$gQWOMb<ERqn*Bg+CpUf);>G1z8)1g&I$5fV2aSt05mUU#Z*@
zs^@v7OZ;rFH^)(Vi#3fIewQ5u9KSw!rzCKMskF>k$$eVNiieb+*qZ!+tn17GDUl2R
zz`}$YV{uwqswP_Y;+4Dszj4dXC&pc^l{V})1TpH<=(~f~WP%W!dL85<Out3+<1)lL
zseyrF=^kbBS+368O83rqp3iiiTQYUxW<j7Cjostcs4OSoQb%AbBOu|bb^l<UnL#-W
z5x$2zLi6)TN>+$``WiqNe-BWshP&p>4l>O{qPZBA^^+g}503kRd7G9$mk>iWQ>3*2
zNb!oS$uf~e*aoaJtY5+}SNn9dot|<5(ZIUnG96*tWW(YCOxG23NO0Lhq(dn1dY~lc
z#<IJ7t7HOT>Q)~WQiFK_;1VT)%NLenAu(XELSvsZAmtf|+yfe56uB%z5j1)}_G&`n
zS)o8z&vKvcbH6||tXbkFNOUZ$T;LNtdWpLTm>l-EJEbQx6x*kYcOU*^3SpMR>!`MB
z&(+D%Tbf&PyX9!_EM}LtL=TJ$_>S13ac{V5f;)8E3F>Iv=*Ho7hK#f;^&S0eB!aD+
zVbTZ9&VA5RxZ7Gh4yurS*bYUZL^9;RcGN?G{qskz)r`lbG!xqbE2Xq9nIT^TJLhfP
zR(%&vbB^9g<mKe`x^uaSd3=|O+rQ?$;m%&XKi;?e*CbE99hg(xFS8Y$(H#5Thvz$8
zYw)}wk$rC{O51NfXKKGVVYjzSacrtmAKW1gQr_g+?V`35mL%+-iBGL<7bcn#4pB@|
zZ8}d|=W60yP75I0Rz#)Z+SIHunOws#O~|*gzQQ#S=$^2q_x?)36I1R4KGOZ<kuL#7
z%>U+zc_kxEh?G9lK`I#`3Cs@aoQz1>DK|*t5m3&IN?YZs9L^N5@4bS+Vl5P)srA3{
zm0{k8Ner3cSIjSo95`bIEF~~ByIUx<<O$5PQ^_>=JxR0T^E|;j6GN3spNMVnn$ju-
z5L8wgnMAt|vLvtgzFj3N$v*$3PJCbWMhb_v0+?XAO;umvmpT*F<JnivK(9{cc!h~4
z@TqAa#a1YHrWR#O=9<3|X}<x`8*~b}X^eBkR)}o+nVr%)Rr|a01Fux>!Q6;OFwDjR
zm%9&mUH7E+W(5Rk=_$Jc!CEiPJX1X^tTnRp(j-PMIwj2%#3<yrTgC>^ZiVS^DR(X7
zaH`L`>bG;pfR6m*c*7=F#mTz#)9i|qoL`uQAG<WdUJRR_tRmCk;oGhq>0Nz)PP&tj
zSJ6p0HfR0{$#2w~G~>eWBvzE0ZmB1sT!@Y^QTnz*E)aR`4hS>0E5~1RftCPR_!vM!
z@GEXuKtVN<?HFzBcMG<-cJSR!T-`@H&7Ju!{a_XB<CGj**UC&T%-d}GjZXq6)iUE3
zLaqQVfoUSC-b?}(SQIz6I$*CC+`b}vA8Hz7A92mR52k_6I?X2h(W|}}p8)$R&nj%`
z=|2jGyB@D;ljbEP4e5tIzQs@WJ{yOKeF0HjJ-3}kxQKd9uEegzy3O__yfrkl)kOMY
zmf!wrEtd^)r!0SaMZ5O?$wk?;;P-iKS55YoqnrrO2{2R;33Xo!D$;Eb$@nFXjbJ?9
zEqdg3vxq3b2gd>WpfbOGEhhY9fa-Ohky2~IH4&9wZM~(>h^4E=3<8m~CxBM%-4mwW
zO`=ICw13yN;gp_jGOd`)ay@OP%Vv4^KRU7<b3=0AuSwlwBC^%ZY0#=>tA`aRy7FwX
zklR0%b8Cl_8cN{Qj{du*jq30N8FT0Q?*;2%=P^e`uboWY`5ZM49<S9s-wcLi;pU<E
zhUzc<CZ!=g1zTq&W5TWuucD>8Z?y&1M86__b|(oW)H)@GQ7snjgyfzT_Fo@NF0Ce8
z9l(sfJ=V=Dy3mBGbbS}$a8<Kk3b`xCL{wcqd7m&6l}&hZz`qibd8;qp(RsBkeQBob
zWxiyOE=*OGjNM~Cq%ts}vse1m^dbp4)uKzDNaK{TRJ*w$B!XAsh|#{n;jtSh&|goB
z?X1#9<%!>-AEr`~ag3~)0>+C4Zt||I6G?%z0y?s}MW|d@w8uX-kc;IM^L~%7RJ}4t
zK<>5<9<LkS%x!j5(g$-%{Q<2vS87<1N>wv`g<T!v9m912j~FN9F8haTlJV@TD%WZ8
zr^cBE^CFZL)ffmgAeR4jh+dcrWEf%g{&(EIv27>#)k7{n(aHuv@r=>{19phcwjA2S
zukR_SF_?N4iq;EX<kgbhVvK5x@2%B5DAEEJvCVCx#MyjzNF8E<!LeFHmbhv58w4c2
z;n&oMcfawaPuK5pGTQ5Z;+1<m@QL!Nk4!Zh7fUwl4P>5nXrFuuy37`WzwXsW7LMgN
z238@(E=XC(5ojEvZSdd}AWxX}PdscdI;9B3S`60mHJUsu+G!$Pq>al%Wz4u&1bt-5
z)+3MRMqOq+qv3p=9=)a#+5^W4h2Mn{fQL^BkJM%y8^<h(p`^EKtw>TCF3b`>s0tyW
z{xEZGa*gr=buX`Zo&(}=CwZiqQd^wR_cC4I^m*?y%s`u)#0B%DGtSs06`A7->}2On
zPDj+TUK6MC{cbE8l5t$1C10qg&ux+dKab-(9d@V^@3CBq_?4h<awE!6N9JG{wUZZs
zn!bY86zujDwjtXUpP@Y=!aUw*ez<XUS-<glivNi(UX+pc+?64ZPnD~;8ryV@M1t^?
zmw7~6qE!LW@F@S&Q+gU<{{Xght0cRyY5wjm$^ss&T?(S~5w4_N$|;@KNmi7BbH(kC
z@U=%H)3UiVFjH0c$jC+h@f=&}+vTiS@8QDEM|&nlV7=<Y25(rM-o(N$=}uozU^peI
zMjZRXK5uUygia8^3M-D4p(^6eEKR)EpxtR|hqQ{S+If9qv{l<cl~tsi8YT9G-^9u{
zu)W?T2#-FVfsrDr#BUF1pi#XJNR6+v`4Az7b6NUKMV4}8NRb=pJKg-cP^OMIQ|r?N
z?B@FmHHyDfuI@<8CNbt)7&!kOalXZ&<BQb58r9coKV)zImnzNi2eX2v3T7eIG_;3x
zljjRAkV|kd4}EgW5ydKrcDu2SKv>jUWwu?(l{Irp4lW3I$aEt<44TVu5)&0o7;BZl
zZw*%KZoKUtKJ%<}_=(E3IAnb8elw)qFoCHR40q@$JPXwt$)&+?B`0W)&K|C^J#Q~!
zW1f7Hb+I%y7$gpBfk3FRq+!hjL&zw80ZOHr7CVFmcek1!E+|#uyg<IW?uXI&ZWkH}
zaD(7l12>&&CZ(ly1615{vcL2!Wzw@DpS~UNSfkIg=q}VuIVTK$g6D(D258c8{wCT)
zVnbc`iklHDjKVGi4nMzU(eEKMv|?I6v1IeM9U`r<)tG{cm{BD~6&10PD|nnth2aq;
zpZsSoIzbmVOq@GMU#&xe<tYo(+Xe5Us3yqgx#?qLQ}PHFbyDwJubswXA}$*!DxSP0
zzSi}XaxkyYsyU1v4U?Ql-U5a$YoyY)0SCz*%yVK=EJ{G=Yg`trTs4VJ7q_Wvl<vu=
zvKx084x`6E+tsi}pE$Kpr07kNg9H;D;gIwA27&UAU!UY99JNi&;HCRqV7qh+A0W7&
zvLZOR%-tUd@Qd1}st+&7@i85~f%zPLv(cInH)vEm!wqqB+=mpjV^w$*s4^>uoxwdT
zXoIQa&<3D6pYQk}_NsAf&)qZldu?68#$;Al&BF4ogIy>YyKLF(F$L;wL4@W&Oz?8>
zS!h?2I_UAcT<6D@lcq?sp?uDID;Jiply7it^<`Bq)vaag@eC-ncdcTU!e?&sZvwSJ
zS2&a-l4GyN8?a(nrhU`5WBdU->GJ<B@_+lN9|9zK7<;SygyREhDM7)tzsy&r4Dyq*
zp_1!MT_?2Jc|+Rr2UTSB5g0ytag6iU54`{H+y4Da))?fw`2AaXY!kVpOb&Fnpi9xj
zjm6fz{I`~&iTmc<dvc;PsJNdpULK!aJbS<KHg3~%LH!BTI?rZ?taJOr${p<#%jB{v
z13#(!F(L37F<&OFLPFZ?9b48^%4{~bVd$B@WKl|@1r^-U(9W^^*>iS|zWi|yNMH{?
zZ_t{cz2M+?Y{Ojq<1#DWLUK&46(mgWyKS%)iTP&vSEu5~O=a?eaV-j>TkVpQ-lnJC
zhn3s#4t>OKwrwR}6ZYjwecJCMu$d$h@`@lHFA=^Y&SMi@Ep!>8_y!KLu0-)5Q@bR9
z%m+WTS1M!V^SocurdkT>Y<1RY!hGm)6d6)VBHI$91l7zfPWDD@GPCxfOA-Sj4S8|S
zk=>M!Ew(M64b78IIpb0kk7KibV(p!X;LxsG5@v6-X!DkKy%!4RqVeFKobS=rn~lCD
z6sDNEOZkKPnC-l1{i$l9N5EV}gy}`gV+MBf!$W^ja`}_7<RixqaeuGQ&%aU#4oi*a
z!S}}vdZOQCX4bDI9{)|V2j2eSz@qA+Wcj+V(}q^d5brloyDch{>SXoo^;Q<YG-4sh
z#b&SzO_PV~H(7qYdmfdyD(>*B+r-FU|1ZZ298HWFoNwiZ_x!jl+%H@Glrb~r__MMA
z9vn?*x>KUs#yWy!^+_OS))7tr3Hc732)Obf!q7i^VEl&y%Z5e@DuWBum3}W((VvXc
zRN6XpM%8KfA$qm3&h?<OikcM_|MVw&kp6D!f9(ACSN(Gk|J{NA>=EEtT@#}#l4P}u
TKZ9(JEn7^^o1e=wba?PDUY&?8

literal 0
HcmV?d00001

diff --git a/api/core/model_runtime/docs/en_US/images/index/image.png b/api/core/model_runtime/docs/en_US/images/index/image.png
new file mode 100644
index 0000000000000000000000000000000000000000..eb63d107e1c385495f3ecf7a751582099873c0c5
GIT binary patch
literal 267979
zcmd43WmsHW5;lr!un;_W2p)pFd$8c{?(WhMG!Wd~Ex5Y|g1dWgcc+ot<jl;S`+X;8
z_<4Dr-Q9Gry?gDdT2*hoRja>#mKH@qz(oK9149xQ6Osc1gHr(mgC2x~0lh<*_}v!_
z>=loxpx|e5K|!L=cGgCw7KUJ8VqasGVO8Y&Fw@lKh@s$!h{!J92Tg*>37CNAV0DOy
zLHNDaBPy636NaAofLWedL=udT%Bg28YcY$3NY&q{Qci6x$2a4->AnrzT7G)8-{LhM
zPF05iyMs>?iV87<h#>zE^V1)OK{P*yL0kqLhsYb*CSznBOKg^&J`!xc{r33k2<F1y
zX-c;Aa^&&JObtR8oDl3C%$fr~I4{Z?B-k%SkFV9JU|A`32?`XE`m^3)Z{EY5daw^G
ztWmK`E3a`Qr2DZUchQ1HOu2HIz6ZmNCmx|$GZsJ!)e1`OhKR%l<3QJIYi{-c?=nSQ
z5Fmu_vh!5BL*YmSOh^(=Zk#f}R@i+&Acg&k@UEem^2wO}v2Js-A4X5*M_w8kQb_a&
z&h(*?w3px8BEdyc>G0qH6}&YR5}!^&y_h62L<xnvY>|5i<YJsr-&xgm#9;4PxHVK;
zD}|h3nGsZf8{z<Js8kxNjrV-TD5oB}{5H+~#@S4~(in&(BN2A9RF7|Ez6Vi<J8H*8
z_bTnS4ZpTiyh{+VRT5z6QV60A5fYb0HRn4}h(IUIJi}2<>Hdm)64}4DajP!ymi$d1
z4Ao^t)+<x7l>RoFG98`P*`G9Mte^a^F^tIkDEED9PF#cUS2fB91n4<!`9xJda@gs7
zfYAM1Zg(`xn8`-3m3z)FhmD>8HbkDRfu0JmsF8q2wLL`aL`@N_9NHa?u=+(|ZyhJ9
zGXcsl&Bd05j!PsFr4>pD0vQfMlE{}9f5X>;$ZX$q$49w84YeIoE&|@KUAFtxH!-D)
zpjCQ;H@>{YufKfhu=NSnBBuCqKf|x&>HvikyiC#^^!5a5jSCgrZnZ{|{u38AEj(Ai
zw;jSU*YZ={)`s#v;1f4*yO;57#8(CSN}nLfJ8yFO1Z<&KqyeahHiC;N*Bzv*^k6Jl
zDgN$AgX>MatMHkrClNNOqvG|<{0bxwlmoIE=zD<$W|5Ejv^M4iyJ4_vt)KC!*4^rf
zZMv3p-m)B_kTwa%W%sLVVQk6F1Cl#XoCP;UZSyi>h-1Kl4@XFu511EW6*p=gD|q@f
zCSdo6a1L(SdA0-h5Pz|+8OvDk)Dm)rq;?JKi<ZN*cZKPPs!%&AguyD$=y&ho24Uv+
zlEdx=J)&9v8t$i7{iU(Lcz7>7r%aJT0}zFjZZ!y2h)GoJHTXgIC87`GJT-t`GeuJa
zJOZE{Q&z@W5qu1jHFKAmd=3lN+5-dvQ2^VW;gIDs=sb^st@gDvH@lx9e=QQXN~+~-
z*xdSnt-N-)$-x0b*8#U!_0aH!Jpl(N!uInui2##pcS-QX)Yz8$2rmg)#ZO5AE7AU`
z9Mb|Uxx=X(@c<I(7aSY>uy@EWhI-_q_NfVIEpM$8WDgNkKfh5_7(MR<5~U!>wJ+tU
zyzj8T3av%Fd56>|)D`q9F<_9tKhm&Q%>Z4__dpn#m^|`ajMEO$)!TuzEK-Id;CnVl
z0lir$vZyKrUML^>sIAF+4m4Vk?W~Z|DHG;K^b%2-EUdA86Y@KV!yjOI0lT_J38X(e
z*;JU~f(>=#EMSxZ6m_Ol@TZ}U&lQW&_d9sjgi?GMZAI(9{5TOx376?I{)Kp2bK}%Z
zVBJx+hH|2HkLM%uk%SGk8D68)MBu&d>&%$X{Ij3f#MqG1b3b%)l0+7WRuj8*8q@nH
zNTqzz5YrG{l;YyIAm<^uiWGYX83oxF|3+#p5<xsShdC!TXFZ2H=fgN`CGK3zfdDdb
zepL6yYddIH*jwydgj>p6M&W2Z;nO@Jd8io5nC=+bnD_l;pRhI=6+_GfFLU32t}3i5
zEGl%IoS1ByWHmE0+cMKKt1u&}Q8g=?x++MOch9(z)l^^7O3mg`(#&fW^Gv>jzf|9|
zHepI)ruB~dM4dyMH8m<db~M3b@{uWpsU8<1G!Zuz_Yzl`No$<aftn+kCrwMTEaNy|
zJL}p+j;Vy1J&9`wCn+{5kcpECSdFblTAfvMR*h+PILc9WDVdyOS%P0oJDOJ1q6&<;
zB8B-`u49Ak@$>CZ%%7$r@yN10B7sHGCGw?J68Td3%F609Qmo?6l~%b&vP;t3k|S#2
zMe#*-D%R=CNG&pg-vY}6;R0!}<ybfRBd+fRFCj1d?+ovnF3a`-3_7?_cpKO>%qAb}
zjf3#!@qXe;GiX@5%$FM%41Z3~yw5ZgyhduV^uwLQ>&B5{p0}7Zn>4o2duv8RuchzN
zU*97%@~-{ef`NY@eYBN)Kw)}euY>=t?ZmdE=KJKt0gWLG-K!*=ME}ITL6*dtL{FAF
zdKo4<`V~vBkuPQ)mA&OEx?>h&=B8%s{dP%)C7YJi!59Nn17@*Pszuk+EYp3{=f!x%
z>2o^^HkQGbW|l|u;}#E=<nu}kT2-X;6$f?)({q}&qK8X|RR`3EbnF?(1jr<Tj>v~U
z;C?g*d<|4Y*|$+}HsHi$pXT7P8MhC1pt3cyabsV!bJ;t|{V{_4v~_Tlvftc&I}CS-
zwo3X*gelbJQ17_YxzM3u>ytf_ZR+m)&Rzddhty8(@?jq556>7N*R|1g89^F>TsTL#
zm5js)(+JfFavU;sKuv_>aL6x=9p;JspczqXlw*No&t<?Z+ug>^!HFx&xKWPHMqFgL
z$B?IMYqAIDLxPv77v2NM!_d9!<J2PxBn<=<xH@E7<>BnyLFz%g&#=#UyRXkN`*O-|
z3Oc{aXOqvWpViBqZ1hg%e*uZNh!crxh+8s`GUWwj{Lnf{I^zNyZ9*G7tdQzh7a2X=
zTqOM5ggm<Dg90P3BFB64WOoWqWCui-LMww`i!uxOnpj<&?YPsk(4*@k_q^8E*Q3@?
z?+Oh{#&l*$L*pg(61{KoYxnDhcYbY&ZuoUb{3549oLT%h+BYhm%sGsJ$L($>gff6K
zE9R$yoSEF-^kgf2hrUwSF77^0t;0d6EzA1*a8BN38;_%#Z1f#;7xb%qwDG73ryS2!
z;q&jVp1X?|-!HUy0^5S01W<`oh<7mW5aQv}jc$x+23+EHlN-3II0tx(NQy+e&`K~`
z`jH2gBOZ9jIgHh%0akWLM_hJVYRkkGwQQn`QPxqs>}{HBM3k`((ids<c6wz@0*y6b
zl-SCnDg}^$4RyoXN^Jv_;QL^z5FW`-N!5VLX@k>^Y04{kOi5R@w{rIj3LaDdvvZ~@
z)2e*#gyFPSQy2zPPmy!UkVJc9ry+#Y8c!BSr{RMYi4=*r*yhw|u6)zd(?wN#!iU#)
z5|)J`h1J#iX(P=@=cl(>x0Z++SjDO*C6b?=G8Y?~v+f539y<}+!fqTd%N{})Fl%V9
zRGPiuZNIwoh2}m=)X`0-DK#IAoyiXDD}PjuDRF7C>+C<ru)@Gol2z_gax4wgSgQYN
zHx$LJJ#AFdqJ^W<(9~aG#y<sg{NkuQ4O?8cpwpb`@gqMpE0mSh-|Bl=ZE5MO<f2CP
zZC_jttua9VL8)rW#{71=Y@us`qpru&y708Pa89>c-<DR5y6AA3-D!JMlfMaf|Ls1@
z;q-!LE9^n!p=&Fv<#DJd8Zha)>S_kRoa3`&gl%~-L1DVw%0k4{=v>11tQW=oXZvWG
ziH;xAfpz%woN}%T*E?(allu1ImxOdY%gzFJS=$C%eWF9?NyjO;o_lH%77`1)K@;1<
zektsZah_K<>?`)7!|v(44nB^Vt64qWcS_gOzHLoUUgbmCC67_txjzTA;*;k^YYCgy
z0Gk_A8AZmWwq=^l1dc5aB5R?^`1+h2)(BQSO|kC#Q-@g#>BobO67}j%GUr;|p4q@D
z!0bas(?s_9j&BBR3CcR*vTGOMV$HT=oles}eKbvhi`i4{d~kCv@+jhHV5EoZ*p1lr
zv?2dOri8ahYrd81M7dqNq$S(ENlRS&vQ6ve=dYpKIxFjBpx)h4+kpG^t>+x9FUkUM
zB~RKz%>C}}wz#&Kwi2KQIDbLAwy^Zr{OAQQ(=UIIpi`cSg9FyQ3<E|B4F<EAF55$K
z4L}f#aaV$dc!H<C4njBw%Rqi?>$z%gOGXB}vhZ1;E07^X<a>qW4MpeO*oa)~(~_j8
zP032K2x}pEpOhAl{1{4kox;2EaE{ZfxYRzf4FJzv_7MCKMWY8wC@c(B#Eqn+z^Fj4
z;lQB4alxQLufRbE4>;a`UW<TJfI<F#9|8<4&=d^nUt^>}&(D7mpyT<Rzn&p7ApSi9
zP9+2K->;zupNAfm6S{();BCZI?ZLn>$(|2zak+OVU|@V;;zIliF5m}CusUzMTW(HQ
zb|;f*ub17sVSMRfCOE+*Q8EMGyoPWkKK74@2MFL6`ki<~2_R;|z@?%#b|irP{Q0wG
z4Py3}*5m-d%ROo8Xjj@f>L`bEmvh+E+1SX~n3p=MX=!B$%IhI{Wf~qH914{W?0-KH
zediOIe2ws;v7pPn1&5%y5q`;qL4)<+P<7BCgY;fD6K^m?FW8sO!v`AzN~H@d3;TcH
z-?Ipa+P|#4Y$P9102nwXy*Gu(3q%JR0$2Xa`DG(fgNXUOg<lZ`qrFUcATi^=Y~=rM
z#N2xj1mzD6Jzk>D+DF$LUpz?D2M3cnQCw8CACy19PV5Z^0f+6=y!iI*KWkZyC}U*Y
z+{`SevQqBqaDk4~`6x!S$+<2gH>^b>5Wc*Z8D;wJdY>E`T9^d<Ntm>}^SA>O0*>hU
z`v3T#M1%?_W@MCi-aj%Duhrtl0gmN<br4IdUKjL1oWzt+Co(#f))ROYx!mGz&;46b
z9GQHouEOM8=F&L-WI<x3e4~SJWaY-ETw~+n`nQJCdxtffqTEN{SaX*^(l~SlgG1is
zv(YaJzQ9_DJfR&FLS(Z=k|JrYdgmSYQwxR0uPxM;&%ebT&wIIJVr~rF*sjBsnzTu@
zN4~(GI1>46Z7U>GKJwpP98L8Nb2&67%EfDKFJ&7$<M3FhxH%>*ll)KV!4ZvuLoqP=
zGPs=R^o`I(IPT9ldou<<y<r%jrdlss8fJ!lwPXB+I-u5`j*aVosN%PxQq?nzM+l-4
z6Df<7%8r;&O&NvZIxteVoTJc%|F@jPAbp~uN*T;HdYN`i^%@fr4M{%9ahq-+k}Bui
zDP{rw*%tnqfE1rEO#l-9JY^oD_$?X@0wSWpM(=9sp>xFf3ueGaL<{9G>c`XKo-3@a
zKG7G0QjX2xAZ9)s?)d_(pz;y*K|BCaA`T7?-Vx?{Cd=%Cg@8i>C1hoFfe5w^FY=lu
zazu=b6h8uyEGwlOvGOdug`K2pU*WP5EqU(pzDNfG8TIulpBkMG%eRxFz%d==$RB{0
zhvhpd*e^L9Up52q^ho09h<`fyE*FA^sa)V`b93|4N~<TwJ2rCv|FtARYTy&aprGiD
zU;EIVNfd&NX%fi*j)|+HEC2K;iKIuX@)E<0kcRMdySCb~f<>t8wO}i{Qay)Jnk9IH
zfjO@j_X0bBBZ?4#=<YUxLbR^r$Mo@LWd(siN2n074h;YI^&}CI%LS&_>U@DO(W+Pk
z-$(4251LIZRzomqOdrHUQk>w+N85uz7o(cacE7In&W?W3oKRIW)W!Y45KF%EW18Zc
zjOEGu2_b$zvBP@v#)VZ?^nb4qJ}@69^wBx$BuSQDlF&RP(<(HYHTcrciw6hA??Wkt
zUUY>?%--SQurgnkQIgjAL$x9)2%0sd($8OCF=8RU3}lQe6V0X=KbgKZh`ay`*@9Be
zIWB|Ye$o8FsZmXiCJ9YZ4I(GOLgJv5Cv(A*9{hE_zQC@)AmCq%ox1qsK&w>0t0&>#
zVDf#@S-TVwo&1>Mh~;8E`$Fk2Q8`G?Vi4l+<vJI^Vv`%c<J>>Sp=YC)^m&8dO|lA(
z38mb1ub2VxlEntcL>BRd9}^y_oTxslut)~yMJIL9M(BY~LIj;ea-<DT^^yhU&Bp~n
z`NkkJ0u1pB^;*N;B>#(S#RxG+EJC;llIAszhHN|u#ZYU)!3*LFXb6vw&`)$Rgf8@I
zs&<k{Y$)X$uaVi8;2)wx=+`*Pqs#o5w8-UR6aw>4pKM+v7NQ6TAE&o{S>526Uxjop
zf`!b+Uj!lwNk?U2Y)oXmg+<wKCvP3I$X~XY<M}v4^~Mf<3$!ssLEuFKHGr=cj1aDZ
zq=|Aost$RP`NHLV=c|IC!EPZ@q$7QJ$=75=!&jeDVA{f}yrZtqDh-vXkAC2=ZG6dX
zRX6cf>LKSK=_4~NXhtIc%j)@rd~yer`i<3VISe81u4C!5MMaA1Zn6z4?Jm{wE$(N>
z7_1H`>Xz@2@C+YeLsAacItAEQZ?YFFYqbhX3nOJ%09ypLfG3`)L>ypA?-f4hW<2Yo
ztTgAQ3%kdb!wOH)per52b{S<tByrUUTeqdQa;^0WW!X^T{?-uB!y9;*^)Nce>yYY7
z0HCnq;WLZXU6(5o4#VdUTu;SfiqTQ5+e5CEdQ-S_^{41_)wk#khcaz(Sy`PqZ$m`>
zVZrmgyOKxBj;R)lDyMEY!PF(P_`Q}K%zS2+T_m(z52bp3RY!-1<LaEQNtE?195-)I
z0{fwjZy{({JetJ`Tz2U5Eq-}F(nHd;&elrBeNdm$yB0)FqjGI%K)YVig&-?!y>hEN
z7+!cJYH`1IsLL*CzBt>jbb9EqJ1=qWZ@K0r;eFb5WN;Y;%fI{1yuWxp@$CdG1cO=`
z*e}C-+2seVGC4inpMKmd)89Xk+S<$t@OkUGA-TJdUssiSK#0d;6@{hc(ak;BbW9ha
zAI{UcWCZ<3T<dd%Iod%6;lr{cjE7HA3Kqh`rpag)7)kao56347jz5;!_<g>lPQT}a
zv**b2L#5+B*3nhzlKbN2_wQ8NiI{INF~#`#Th)!1?n2Lc)F0QZSe-X}c<<BRI{X4Z
zyK24`5_K76vR#|SgvD1-Q0Si+Y29(k1_6J!tAit}y6xaFOxw9`tlmTC<$b61LgPm=
zWv>T0=O%XQh1$c9Gu!9&&bteB;bo;TSr>;(1Myhq*>#00*`jqSQxt!WFd5wtmT!Qg
zYmdyvJqh5Lmh^7I#_*-1u<EL+szVJtlYe2qcJj~i^3?Hw2VCtm=k>&Gtv8;+pVS9`
z0LuVCHSMtm<J1pWNwRdWQy-77nieDROM%Bl>Th-EotK-eIOlt!WjK`1&d**iX==Mm
z#Fs_@cpjSR=;lR2ak@K`9hf357i!(Bk~GpCyL9dR60d}!!@O;KG_P4z+X@9z-*4Te
z-+wQp3hi&T?e4Za``W|OPY(HKRS6+fi$@6Ky?tloEle^-QZuFch5EuL1ew(9(7A1s
zN9^lHh4YG+>s=xi35M`svR%6n4zHDD7lZYdZ#WmRqb`b1^Ja-EN&@@wg5B`#^hNAn
z2sR0GXRIYTacj*cf803TGIo<1FAuTeR;^$newImhQDt|kD6nd-B&W-$&aLnmEzM*8
ziS={?J&N#Ex##wW#5Nt-5a3FRXU{i+v9<Mb(my-q-FKrfrzqmvCNp3m!>=_?t8d?*
za$j?AXbkc$dNf(gmiOJ(+62H$!pSVuS;zN0-deagO&)fB>8yvKc|5(DkX;ZH0tb2W
z&pAjS+)Px<CnCZJgS<lr%I_ShkC!JtA_W;aQGN(3lBC|;+NyVSG`uM%wi{B^wm;iy
zIZvZ&@7yxaWNlv?WR#Bc&69Sz65j}a_F;KL>Ge`(-mnv@^>2&+9MmJ6AUJ$Q+V3T1
zCJnrW8S-IwX*w|OLS*IT*MSc%_kG5~fu>JoMQxOciTv6)c)wk1sxn{x)95ZY6?M^D
zzw2$lj$W}!jR=7!%^Mo$qq%&bh~V8Jp;c_1)JMO(vV>ipVfqQL*^8ekb(`T&&Vf~h
zJ~(yD9@(^?Za1q&fcYHkhcW9zE0b!4rO3BC@CsAvbJ2wb1?5G<a{1e5d1q&Kg+=BG
z)1>4*ggR>_<})&EM-+VnLp90pngkzBvlh~%r)<aJd+HaKW3?AuqOBD|fyWJs!6CuT
zii?+Z%U;S%854OSbX>|t>9)a327d%#5$_@D7QNCZCIs`O;y{O~QrU_pAhQ|hWm3IW
z)aGbfbtUDrnFFKx&NboNdr)9Qno*rk_q?MVUOaDmT%tXq+)p{fH8^WAFSTIkU5hZ4
z&&?oOS&~uKa@Q}lZfd}5Yg~21;daZZTHs+M>myUv*kn2FBE3#<#G_XrH%<jKs9)7J
zS;b~$2^hO~2P+PxZ7*2{VLT^OK0y%fS4XkH$49ZQ$|>{dudPf&2|PnsI+3ZVn@1D5
z5?Icf%FWqi{^3fiMG1O%5^<#wPtA4hhf_K5>jtx<9cQaqWM|RoPt!vv+U(bzqdR!3
zw@$#Ufi_F!)J=VqC0Z?BZQX~Txmf9Rx2~>{!LQS>;AY!89`i<+j>($k!v{Tq-0Yh<
z`VX;#ajkW5gf)9P=wWdthb(5^ul|(n`NW`?`@BAdQFZa9mOC;SZx)n=q?tFq2<}C}
z7+;*-easAe`wnowG_slBc9VLrjE~!|*=DIRUK8SRScKJ(U$8QGnK$sEp&fqsWQ$g-
zS+e$Uc6G>X`Ys?enM@}jVMd{`ZLsdRKh}J)vT>e6doHPF{~a&yg}3Qk|5Pgj%WtY?
zUP-3`-ogE{rY^<T2!?U9KHHdmzeuSB!_zgIcA{C`x^00niA}e7kV7w^E@?-BREFpB
zOY(MVnjqtG6R+J#54w>-9+^ZOClKS!M&Ddj+eYnJ6b91cge-gW?%;%s6ALvJb;}%O
zg_fo!Ek~+&`f^K-^Kn|qUNnqYOl&Nxnf~l3zci2oA}$Z`090f*o7N58qrEdJv{s9i
zm*E(mgcma{d^&D9GI`2lHS_xiVWN&kQd`*gYyL#2ghMGEWs#%L-KW3=i8RXpt*2Q;
z7@meM?PYRZTk~q**$QuShpS6ed^~jzzIzU>Q=EFOA3@6OV5#nkQ0;Ar>ay+igRFDg
zBY9h}+Lb*c7yIvYh{zR!oqevAk_6;eTG1~uL??H--+v=qf9tXHi@&mqYZP!>!%lnj
zEzAgwKqu1&Nrq<v#F8$%R96}fD1U9W7$3t_i?(W8*Sc=D<OGpXwMFonA~+vOj`Vk8
zKA>=)$xt%Nho?C8!`pQmVu>s&4>0o?J|n7Sjh>^WPbAX|754EOkC)~oajuNaH=AEV
z+E#~x>mL`}v@k*8>&fNh!`KS@>pyv7J~M<_9R!dR;#qi`XCl8)XH6lku&ORd4b4VQ
z5{0}={oDl>Vi2YknR70dY;*}eS7vn+d4Dzoh#`F1ac3|%;nQAOadmEODgnp7Bl!&!
z9FW0N1J;tYo4VNNf`dO9$3VDm!Aj!NDQi4pG*N~>oU1ywkd`#x-Kl4hatTYU|NIuE
z{j>&HT5GUw+eaiii8IX=CZvrHBz(wMe+nRc3dd83h>j*zRYjU8s*g=l85^Ve(viWr
zDdOttkFJr~)XWReodiR~C*JFh*Rt1AyC&zoYn4o85LXueGo%&5C<lMdQs4Q(vUQ0b
z!<X7(W!`X3!9bLnW0InsBoV|)xn3X~Gvl&YlrMU24=vLeKlu8GY?jnLt#1)>o98tV
zSZ<!vwFW6|4743?X-ja;T!jM~1*c1&X6Kv_c4W^Esp-yZ&`n8o0tm=|i{+h?&Lb@0
z9j?cUg|^MI1tmrA_2&N13Q<Dhc#q9yk=s=qnZz;5G?%^G6AS02j0`>j`A>e8%M2Ze
z9F{?O1WQeQ7rrR3FcQ+2Wzmq6IOY;tJ27D(aGYvX2S#`nsOlE4X5L>@Qc=ZJyzlMI
ztJ!CrRW#nEo%n{&>2%xQlZf`NvML!%Aroom(&aF;LHeWzp!p!hUg_l)JC-d{f4F~v
ze>Ywz&^aM`DPJ&q-=Pa0$C6I=C$j;AScPCo=!8+8H57alaHc|AQqG62FkH2)>2Z>;
zKo-aE&MJhgL(KwZO^6Go-B$S5s~#8LQK-u8j2~`OShxG6>h8O@T{=%ifwhMsBWc_Q
z3t{-fTOf&&P!HDWH91h))rNDs#l2-&*D=-%2TY{jTh6m&(ef@;cGIf0+Qr_lpBgV7
zs!;p;L!X6Ge+W|ywfpVz*`S_bxi+n*4Pe01x!>{0GnCWqOG&v!H&p|$HSoY_a<dbK
zH&`tXTC?fRz)Y)Kq`au8+346i3y!S<(v5-bk=o=|u#jkJE+wYkEat+-JdwwfeBQde
zuQ^A+yOg@6Nap3V>ohzadb_UAE4IuLBy!%}S>bVwmv@)@=#Igtb#f2=hjsTNAi{dz
zsA$3x(Aop#-b@y+<<Oeqde>*ADkW92c(>h@TA3ZrwRoFcEZ;F(?KZ@5`{>A9lT7o+
zY-C)l9Y%8Y7edpd{7GrPnvFl1v_P@1WusKQC8iD<^=yXJNMk7SRq9itmP6Qu*lvMp
zV4ri_InrE(eg6IGR$FgW?ttf1tJ|*sb2g!!wv@nrjym=WHu)e3y^-5>4J!GNW+ep}
z_8fADmY(*vI#O6|avFNFd7f>+`#C8y><A3OlfB<=fj-N@@z=ZB&A;jAd2k5Xp^h}&
zt&BaI#F{?>kA47X>){uPh96ZLk7(tqaZT9ltVke%iTNw&f01Z@PT}NH5-9AqTlNjw
zK<UcCiuT^O_fhHTbQ}-y5AJ3BG(Gr?15HixGwxRM<T4krJx!StKkpyI>p}>$Cp-ZF
zq3_?nZ!A{HIM-UYk-l-s#KZNj(3>*5?WkEXWx`LCPbB=~sy8JcRfzCUll%^&-1S9z
zj{>s-d3n%o1$+5al6Y+JN?+l)K3jR`t8!8xL4}-;>qmJdwXGtjES!cliTt!*1LszQ
z30@IzKT@{BYH~L^o*(sdJj;d;7jDqiTx(M6@$fi2DWe%9wEx`Bo@}~kOj?$~A;z!g
z>9)AcVX{}Z^-u-x8Nw|BMoNYU!mMO*udas|e*W@-Ono%hieJ-X6rS2#USrXuD=Rp`
zr{m;Qe=mu>JupG3Zjr>KTdWxiZPZwHk}<wXt=zQP$S@|qqCDCaAf0};p1!;Tqimw+
z$YG%NqnegO@!KQvYz{x`j0TX@)!ypIg?{a~KQ3iHHw2Rie;7Hxd<VGFQCS+Hc{ULu
zhf$kSOzXNnILuWPT!9uOCO5FLwJNc7e0w{)jh-2ROzV?xI>&_9eLT+8niylFuzaDq
z_}9ggr$^CRi@7~X6gheIeWi4+idFE4uy?lCsjN+g6-qoGQT)BG23DG;jvCHoY*BQN
zj=B(ji?9gF>2R6!$>m^I#B#^?yz<PB{0WP7`JL!nuL5;Z(dzn*tn<Y~6Y6m8(`v4h
z6ZEm*rF+gtw>BEB%E2X`gHc~H#wu59oD1tG^-Gj1`#~m*c>um*M>0bd?)+dc9NYHA
zalCdsp4S;~rSk}n^%RTdOe9IO_*2@|(bSK`JHdRjv5nOC27^11S@KS%<K`E;#_Apu
zZZz~00WK5u71=GxQl&FVZKoqVPf?4G$B?aV+>OrB+KVTG&#87YuhSsKGp7)V2e>9*
zF(Crf-#X2;dIm~jv;Hg@_<f6i&*dBvXN;f-Fp{v1VOn908@>q%Ajh+sQ5BDAm+^8a
zIhe0musr%oo*s{a`N6L4@RPW>^6ePPlLUxOSmJ8mr6jbTH3-rB5cmpbV{+GMp5ns?
zv&zTEak<;G?iGv5u)348GVS%HZ}pF(4*)cLe_y}nSwvv_z!G>2BITUa7G$G)=!v`=
zEfl;UZiOM4z1Z8g?i)7@6|E`XbZW4b-!Lt*X!yXZQ7xq9mYZpb^sW9N>bym9O{?u8
z%$+nX#4^J?{)5w$Dgh`W4sso!;&8Y`&^#6pxBpQ3URW3_91<^9LO$>CBVed@R&`kW
z*Ar)vypv1xBmR}NY81{ZmZZln&M*F9uXQInCi40jM_P6)_L%TDH!-5jzle$igdSBG
z4Y_H1cXtXSd15oT=nS^aT)8yfAMab47!=4ynPeDVTahh)A9lMwBC}f`0ldR~cYbDW
zzi5{#Efg;GM=ngv+Q)fCo8q|jxDBh+*x>wtiDcI^u5iJfwVgzD!q;JIjQZJqySJas
z<Jo6FzAx5j5P7tlS;>m)p{hUIj}Mln75IH|F>N2kqM1gNZbv_e)K6=d?AA@>IX=j^
zI+^?^RbdIlbgdUj^0n?G1Y^o1;~V1Zn;^dx58!&No>;t`@l<xZ+8H~@W4^m+FN`)k
zYIY_)a_|4z^XcPGa-rnUZ#Xy=r9&EZkNOQAjN?&5*B=0nMZ7MocqLDF&UH6ulV_2z
zM;DuiApdEF^@ars?RzLv`6vSQ;3OfeWZJK97MX{$Xd)LJdzWL?AWy9QYa^x)J@wuN
zFw4?%N!PXkF<0T4e}1;*J`>J+B&J&D<y*{b#KKJOJ#qBo^`D3bB1foBU+Mz|;%De2
z&)u6fBw8^_XO*;#`Blsc#6;klTjE9f(yJ{{0Y%YA1Jxm6ufmny(gUJzud482xb>hJ
zeg`<Ucc0bO$(@%UoSj$Zj8n`AA^-<g`~4X{kSRKiwnkJz<CP5C9j#?w<_bzn<1CFh
z;F6g>la5~yJltJYD0)bM4DsZ?Dg`a}J5`avxL-hFelrHZ<DsU~dc|uDjU>%|mCP0J
zD6l)0-IT%zWU8CjWkik}A7^k@R|X}<!ugeKyIuMow}sWGx7!PgM-G%`*d2O{v$9sL
zr5*-(kE39S%*h;vQOys-?$Wns7)48NA)O*?2|8bllA^s%9*%phy`BKAntITQH&ip$
zpjECk%z5o_dnyCKFHGS2V`}#XE9i~JDpF2)2Zqxwt;DIkM9~o$b32JzyHNFgb(DmS
zag}}L>qWhTEeECKQq6}SRXy|OJ;Rz6-#>>vuiQ}xpR+Rt)fS`Fs{F73o5{#PW`y$5
zPHV5jsJn(fA!&|G+lWpO$-3Dz^H{dv+xB=H9u+xuUm%$pl~d>tfs-l_4YS;GYGe`<
zx+)lW13_aXng0InlR|&*Nw`a7kW*Dv6&~|>Uw(RntLNQ?!_4{lnGhsK9F6+XhN(@T
zFjm@nvH55a$jW3%si>&P$jQ~0J@)aP0h)<yt*+phzOYCU>5n_W*Mt2rL&?0coJWl-
zGg`%=*4nNO2F6BxnOhtdi`DE(j{Sc`IdY$glx+Nw@}4DRdM<ww^NuN4cs*@*5D)yE
zPElSCl7=L`T5l)Y6R>x{k0qMux}8t(vCkFNe&hkAiLLHomsKtzecNgw);Jv?4#Lvx
zO_b|D!<=T;HC<P`m4&q5{RW9wSiPWzfJ8=Q+q_A(+Z=si!Y$EHAVfkw5MTC9VPnJd
zY!|am5|a6-F<)2VRb6`(!%rxfxAgXn<f0o@Benfl3$?(rEK7^Iy9V{=NMT?*8-!$~
zJNqe|U@1p?r8nEePtlPz5CxA!hIg5*U8ke7JIeY|DGUeSH8}Ema0mB%aOd^P{d{gz
za`JjDmZjmumds)nkEdJHexu#5NpUN0P(0qDAOL||2D3>U?G_|FR(8^F&+Kk8U{E%Y
zlOM}UB!yaja0feysl78fD^%dDaNC)0#gBk}|KqV$G3lR*FgcNY#lH~Tss6(5xkGiA
za^^VfPDC|#JI7L-P%1Z^?Mh;NZL#Wd<BHNuD<WPu738_a#fy}q4f`X!+ph>Gq#1+#
zSnHp8nxXR3q2xV5yo0fl<PBlC<r91wi|oNym#vnfKSs}WyZ<7~s|9}5>l^t)&`7hf
zm{HrWc!@rqYyA)e7GOy6tGc=lnx-;s&~vrx4KF%xtcpI(DNLXZT)6PI61P%P40Ei%
zX}F=@7PiP|PsDwtmp?c<M)slc#DaBH@99ylVY8M~v#hcbVT_VB33p5%7?#@l<N9ds
zM;*6m7N9@ona?AJm5J3Agz2*t9f_jX7fGW0EW*)2O4r8GQ+JHUh;nIJXC&xL{pc}Z
zP0DgQW$iJ;Dka(V$>5JQvz%tW<_?T2A&ZaxNTuZx$=3T777cvk5Huv=VrVu2cW}p6
z8;z?q96hzj<UGI&`Hc?IG#hCZLK}P#?K|G&Y!YNw)40D3>xo@8B=|ipr>ND_Uy|Ow
zhn1Wb6lJ~c&U;ZDk|vNHf3FZ+M9CyJtq(0b9%EzpVinLA)@z5(i~CBGg`7%ds>d)Y
z22hL!*~XKBPtB~_1RgiG&T1&mq9gsdRc)-7ds82UgSJrcID{MwP?6q`RohLOk++jj
zQ%jWIO)IBz0oO0t?GL|KjrhS2B@=WlCI05j{?76HNK@)SzPs2!No1L8&g*h0cRXlq
ztahb$wpvGUOXo#3;R+U}l0lEE1`ENQE7ByF+@H3L(ES0^WbCzeD%u2&nRL6J{vhH7
zSbvIMThqG*`i8TwSk=b%gb{$4>Ppws56quQ=;&CN_oTTwD@CfDvKW4YLJbIx%Fe|6
z^hJXuQlsGrG}8MC*&Sr{I*$)@ZCix&9SftX#%X4cyltn|q(1^#?jI^Hffq+)V|nSI
z8QN`!X<C*0Y;`wj%gRPtW_1s+ZZ*?}#M9Vc6Ee7Dq&Te6AJxJLXgD8E*<IGmG`WWJ
z-WI=N%I3j$?Rv#J(v$^`zMNy}7Oj;{m(IJ;H;%>#6BxVDsLs3ceju)_GTH7Tl97xN
z+o4XqDQm);3%J-u%2H*?Nh&4mM~wb=QUtvxChr4c^$vFDyb<sWbk7$I=h(cmG#eue
zd^tcZ5a8pOl;dI;k!SN+bcJ@`upTdi7eS2E*|obBBlw%lK&i&&N7$8E?T0O+{ABf^
z`zv1HloZ3Iw|HwIKNH(pQc>&lcX?u+EJcrlZ=h&V-Ovt-dR7jDo776oZnHf0KZ#qt
zo3EF(P4ar`Kfxdeap%Us5DFpk!SU9^JerPv$pgiNUa2-<VpZA!;X>VU%(MS?PM~h{
zGR-Cm=<tQ@buPPWT=dvawOZm`KEuKpZ7chR%`-VvMGX>bT~SsC1~l~TQLX0fcp>@x
zRKb8Ns2)bm`CPzrB4Igd9O2DUB$@xs*Z*xT)MpZ{{!Tr<ww@8!Ko^o$ZBY#G02{YE
zOA7g&QT;b1LJCS!6(QuLAhe|MJ9|@R5p@<F5l&l?>~fUcI9hGW2^99;!%kulj#O^-
z2;({#)t6r0W8NPu-`-!e)4c7tE{kjRbO)X#RoTu7WAzq{g|<O2Dc}_;mBzY1J=(}C
zQvB}KG9mSupXYFW^pAgc&kuy&9*R33VHXTQAtV6+f!Wg5ES8T#rFTatn+q%T8eu&B
z-|`|7W9q;x?Y44r4d0)fT(?FtO0yjm6p(V@8;XRh)aunVU-a)=J$X4#Sf1tc-pkq9
z+b=eksJERwluS{oPpXL4jJ7QdiPN>*1(dmg*a2%HMB2fT5&JJ}9vIQ}vZ|`FUDKwx
ze=sQ!%S{16F90pM;X5}5AmvfahZkSTr<N_1YBXp&-CAfZ{h->|iKUb2Z=Afk7@xe_
z&8t%1tE|lWn+Y&H^Aoup>QiZPN}ZgXW-`W7wIcW&`4<JOtOKgJoi!#Btb=H^=#yQd
zF!ntg_Uqz_RODe$&9#sb9O9by2Bq#t-I8rMC?{LD`x!4CwH}s8CM@i4H3w>k+sut$
zv8h{EO}st=S}`9EWQTJfTJ=;hoR|CN0@_+n6&&{zl$8gbvDDdVfYcF>=HMGOti+72
zym5?pO_RrflGjK~W7XZ?kR4~ZQF;|u?!GrIB-?K{>^Cwv4T`NdX_f+4v~SMTD^Ii2
z@2S?TWVo;D?rt|pFo6t#uL#~5r#MB?(M|u%;D=11!oBBs@<s8S^6hxF!tz%^=I?oe
z(P}<3C>4jOG3xI{*|3a~NleSmxfcr!BV5;dUB3`Y>J4`7*lqT5jJRXY@Y^j0Q?E;{
zZ*ETPYJa<IIhkL1$xk%)Tdn97`q~Y@XxR4$>BL1ueh^MqNtSCVDdjDCt_U9SG+iG=
zwsD^noO2v*tZxPnw5`RIXf|mPxLv^4Y1u1RFKr$1dd&W+e~L4WXtxA3DD!AEcg@rg
zKApI?@}YsM4kFUhwHoJ#WqsFx8~M7M&6%xUA6V%)!F|yPx6PWmMFvgwb|{9UrK*wm
znKV2Q2zu?=o$Qbl^>n+66cwMme;}u%b#(NdpK_Ng_^Ug~pKDz?fqV}{7#w;I5Jr0@
zb@s!Pnaok?5Jo14Y+>l4u^ZRS?4kt)1u<l4u1klqcbA8xGf%&8ao952q7_wDDa(>z
zxd84bvXs>3E~t9@3!9)qG^%H+)s_#h`&FTw1wjlGEB;5*FAm3A<Hbu3!@v_*^&vCw
z{@=POc}+n<V^aMzh!fb$ZtO|9=YX!w0HAij{SHcCi)C#@oSc-hujH8ZL12J!=;tTx
z>(Z93PdMy$381=bz%=LnH~EdP`DxBY2g@xzn!Hm`$<Mj_aBUb=RC;{895|+x=(+|D
zlSzjB1fGhnbE@qQQ&ayRb009D4frlJ*~|(kaPmEpK5$6e3o=ZhPSNrp>If30G7g<v
z&g0>DOHQYo;e&EPl(?eO`|pfTKsiR?ATQEesq^4BD)e1k@7PcMvNuo3(=Kbyt8q33
zwC{*^a@NC7xedn6Gwpj86wq0|{4~vfODB(<U=XsdW@h8h+Agp#>AGn-ZY>q)b!&+%
z@f*YnA*8t6kDYafl!=8OEp5a^heu}{4aG#2YU&YWDUT4GwH#|v<|jNIw=GD9R)88Y
z61`3oZu{d$9Ip?b^CE{%`XeXx;vM>RDNxVbMtIv(!~JF$Fr_WgINdert?f<b>Cw?7
z=Wz<-Ca4%6|B8|OCjfB0z%ckNLpS<S^FV1_(aU|J#BMM$kaMoZ?J@$u<q}^Kbz*q4
zGI&tj1>O8=XJzJ=8$MlR>z8jH=zEh>PwX@K|CAhoL%pWV@D)zTANgugIb{j$bvl5D
zm)2g_$o`c<PWS?i#)H%(Ax$#bDPtm}c`*g*;=N<isCBQ_{Z!E+vZhxbbk+)m*X<Ph
zS^JOhkZZhZE3&HbY<1T2lYNVGFw=k?ja>{3jI&<ev?u`2(#fH^^}}<xb9}O|+3~G*
zKL)6zxov4LujP~hZkww%n(d`!?rhnflP(?T{(kC|sjw?&K^ZfUOROCEHlG2iw0>@M
zMhxYsc3SFTaOl-pQ6Q4eWdoRiEyE&W4fxsWrx3NVIh#+I0k00RxMh-mej}jd7z&gM
z!K91*ivF{vmrv7iBi|(3*<!X{6$a^G#R}YabIZB0mW8QA#mY)6WO$5Y7ewW8#7T7v
zck_rQkNksO8o&*@Z}UE~^Y3M&cjm)?W%(l92}#q5vN#wmeFKLX8Ado$0O}$;51I_Q
z8rmPvRSDy<%RA@WZ#)L{3}*7o??!mZv&h-_PxC$<FmBZA8;i$nFBG%0v)lCvZ}+V-
z<_|nK4V?4dj5_NvGc#{3)D0zH{g~mSX8klEfC8aD_p0f4#(@ovzeGF^hXQI(HH{5#
z6=yfu+fb^BJV<q{4(^NoDatJ=Zz${hoi4~=Z^`&11AKXiMSDa_C&DxyQ<8Xki+m9L
zkBK2iE#F&tdlN~1YJI}_u5u&@8i(F6WCCuq?cB!pS3RC%{@XVYM93T`V8(9Y%G!Vd
z)6HH|<Y-l+`t!c%_iUs}LgsGI72S10FBzWY^1g3!r(c?Rwk`Ab>RLhBsS!-7<3sH+
z@kFhNISJ)0B<zLErzWQ0XZ`;vmHO+0(MafdqFSObo$x$AbPi#$jYz628I&x1`+T^<
z@Q$4Vc|RscG>4mOf5H!Okol5v_xH_t-$iH_8ZQx5tDIUu^^IQzjug+0ZyZ^CqG=>K
zq*`>t|I}v55$+9gC~2kmX(at`h%>1oA!+1HlOkBKJF>oNBkxc2!PZmI^v3@{NnVCh
zILHEZwg<P0{ja-tzQ>#xZ+!MF9+7sdJO#q_9Z8Y5j%tczX9bugW>Kks)XRT&_GduC
z=LlNxk3-UMj9zXi{P#zH3;3QP4N)qz!_7C$ymzoFJ2wQs)AK(fo&S756vt=QgZ+-Z
zJqtc%(#bSy{msjIFhiim*_4QM!7*vmP;xi3t*dv`D$xI@il2p3UBNdy`BpbsmQ$X*
z5M3dn94w^3SOWY-=MqJ1fp)Lt+(PF7U!n)12oIl0gz_NgR}mXcHc)@|m;WVS6MsHU
z*bs0rGf<P_53}xLq^3V6u>Wn(eCQB`1o{zkLM!U3Z$1KNYATVO{?kO&3Dudc33-=B
zDuS>gTGUge{mqE-A9l(32qETJ74yG_q)Fl{1|Uc6jgd>GfD8ZYjz^~<aNn)oj}W%+
z1}F<5B$WGLIN>aBivj&+bn*V9*Z<o}UQ>T26wItUp4e6Ll*@+`FaB8a?`E`oi~^L1
z0(;ROyB}0tNVDHCg>=szh;fu;vzbZaXL0{rNqd$XHJ`VC?0>q7DQR`|i6P#{C}yh)
z4t&n=$1_;vhFr?z@#-wZm@g!LGyPL(HlPJq`J06MlRxl-uO4|rtedLypX@J)R)>^C
zQi@YcW$+gqp6;rU0n=NOEiWWR)qmU-qCn&*QvQn{wk`5s!b25r|EV}@sDjU!Cafa;
zHkpO*2eK_%&j*EnXwvoBLKJ%I-*`Ke(?tLeU%n$oL}{zI1N9<<WrJ{}JoE4KgHo}P
zdeOoM^QlHPS$%~Evhwb!+?O@GkETI@TCR2GBd(f@-~N*c{M*cZF3`m`ReV9+-8n0U
zSv@aWO5S`1p#5tkO%X3_rLBiZJi?{^_oeyN5hkaE+0^*D?&-+nx*ikLpyhIXDKmx0
z|2?Pw+j4i9AS?-WWS=BSVz;GsnUvH2wX-Fo1xoVd3tn3W^(8-|kGE=>gbxv7zWqEF
zO2j#V2*AYX0S{M(zUbdy{eJ`=K6m(8<1(N3twdWbOE}9TD$ZQV^Q_G!dA0iZ`Gua8
z``!>&aqhUOYEck0<DW@FKj3Qq)zU~*0?kyPr8LssOg6&`y!#$rYa@VlVJYMe^U*|m
z517qu8+{JnxvSgcG;)aYpBw)_zTF#b^jGk_+BQs{fx!-~^7i|pk-XYkmNiQmb#dpu
z(bVjmdED(htC;bdD&LNso}e}iGbbkS7P6I_86`ltVB1WgL#sGKxmd1z4&A?QFya$L
zW}V<2<LF61U$ecP9_}E@1J*7iUsH<jF@EZ9l@sR~XN$<v1PTqy?yV3fIi9Lj{9aTf
z+8r3ES~I>7w&i+$sS+8sT7vaGdbNJR^N>Mp;fh1pOnspWxR6_iPm3a>HZncMuu!V$
zwlDe}uO#ircp-EO_LyGfKMwA{m6#z2H_00^%@<<AkNK5tWRzC^=|SS5HKhSkZKZkn
z39*D_alpJ+LWs}=(&y?8QF|8mB|FPsiz!vMy|uv!A~w}|IqVC31zgLil?bY!-Hioo
zjrEnF4|`Pfp~DBG2!EtkCO1Txgl0_T9TQrf+bxcQ-EQA9+|5`jH3pi2z-vV4T$ce$
z$Va#NfLg-6z&gv4<0ane^tRSIjeXW?{6?p%i_U#hO)0e%0>+qZ+0)1Md5>ec>xrBJ
z8&UNwOIB=)g9FQO$G?pi98uoO;f&6NL$rgV6c|#_swSH5gfd7M)5$G8nBn3k0Cxbr
zgZ9TCri?Mh3lsWG9+AbaJwf?|ll&mNl_PD>ddzcmC$`b}*Tqf9eNMK2HR8L)ElcBn
zTGl-bd}Dp$v&CMfL@-TUWIeFpMV;*Egwi5~s2OuoyY$#&J9KaxcrAYZW1+^;!fS^O
zHwf*H$mraX`=PzyVbYEkk*@CY?>GL8%0dfV|72LCW80<WJU*Nm^=u7rH$g(qbT>y@
zkSb%Bk7UzcW4PQqMil9vSf84e_1fQOUSi0_*UTt%@7Q%XDn!$Xl-9ez`1evZ$>ZDk
zA+aOx=XpbjfLrinFgA~Z-3ahJ@o^d=$WC0csBAnevs|!Z&EL=T+(XGuKRVc*0W?-w
z209CgmW-lNG0m-;X*QK2?ZS#ljm^Wwp8dz)|M&I<pD@~t>+y@$#A}{Jt2&lBpmnPg
z7uwq+V#zdePK6fPfHDjgR55ChcN*Wb8&mz4V8x8V@lbyuA=~h}RG)PT9NzG<)l`Z*
zJ6o#r^11Q`HS7=FSQuT~@_sW-c8j^YT7q;_DvteNUPF4vi;`5j@CcnH`D4vAO!4oH
zpqbCb#@}9BYGi&f{5;N;I7Qt$4QRFO=7uudbhtE9y5R1_p(!R$nO2)LpUwa^ukH;N
za@7j)mhfM)<EZw2;L7xYx*O~C8ZDQo)1ZDwv@3>!n@g|{DbpNi2h8Vm;SsB;eVXUZ
zM}qyiCv+S;jgqi#F10kWv|?dSn0mI;B>(0Aa4x_9<qLdcleZ1W$>`oDd7SM!pbyPE
z+vo(aWN8j97i6BVwq&9@F-Ta5hi?Mxn#L(Y%?p91#-h7L^p`tvU;rmq$#~3kod6$5
zh4~5U?5<)aG<%(T0U7?@a=tXs?)CMAmI$Y!sJ+@Witccc7dbh0r-Q<|#kk{8cnpZj
z*W9zJC2TBhnqwh7lp*-rAc?Qo?jAmrrk?|}z`@x<ltepfGfjixV#2^iGv#%s(R2f%
z7L_{lPNRv5J}dR~`<U@xT9DIf=lWd6oI3asWoZ9m*SX)Btq|XVS0TX!3<<8Bkj+BJ
zn_zLPrkXTU-)~;#o&qxy5&Mx?BHWZWx_j#B&IMtA^@<(p7xrwW4Qw(YoXU^SNM*V=
zJQjaj;~5=;hInt&nb$afhi7w%<0d2JOU^V0G|!}clA%R}!J9vgGfQ0dotT{DZV#Xn
z@ZS@;n^PbpnIcE=Q2@C$5{xng&%es%L_^Svb*GJb1VPSsaEKTm(*-;T`33`{MgybE
z?=dxiCr-`$Z)r_FSi}ePI^iv5U|a~sinN%YrN&23+=#m1r;9Rr33~vi2!Eyl*0ZDj
zb}JTN59Wz9P<CCieBs99a#`c%q2XtZ0DHwzjXwZ>U_n~`n(VMe$lCM&7AmTkgwHGV
z52e`>Y-5XCuP%OkB8jf^d}xo^xvJqlmmL&R=vr|Gt`4zp-%Zp!*v(ITxqDMIvU+*w
z^#@!?)abp9sd4e?lI(Q3@*As8R-yN|gw#w32ucdjHf{v1H;uo0CqUp2U#1-f!vxh2
z>7AHEyLT-XoC)xK&<|9zkD^&ohk(z%G<&|RYU6-~utQ}NP9csW&`%Bg^;-Z;hk+ti
zv$d8D&V`0|2r?{m&4(v$e^4V1G!u3w42KDr@Qr8nsAt_)pHRKjOfyVDC|?o+s!n7c
zR+!aq1Gs?(_4gi*=pg>vzx}rp`hVZ^fkH5NS5BMs$Cily?`a%4V5BMzy#xK?&uBYU
zCd1s;htZMm#~TQvlq%_%@8+OiK(U~n_9AXdOQMTacM>nBP-bYt5+kA({LZsP>6mL6
z#rQqlfqh-4kBag&JWNhrwH%eQWeoSoV07czC(7IYiYVr+HUk4=@u90FxmuIlm|;~S
zpH42yh8u;xgQX?*mHycM9EW9(!Uw%3gK%sTXmWg7pXRsu#Ro$TauqIxPz=6NjJhcs
z(!4@l#pYC=d*X%5^~2Q@>UPN78ee>2IBy++*PrGlP$Uk+M#zBeI8p4Ye0>yqB#w%T
za==};LXQW^$l!-4&%4dn2V*PMShppU`RQKo+wP{y47Z0<$7{@$6RpBLw{Y?$bY(N`
z-yVmD&sayWTepzWJpjojS~4znCq%u@7w}T4tXps6ns3(H41WZKj@L+h9ppFz5#@_j
zsg=oAjjD09>hBE=H$jXwb|N?M6%MOcihQln5bo`Q=k)-nDLd!qd6TT19LbC}AoXH@
zMw#c)#rOWQ%sF|p2gFj6N24UrC~Md@Y;0|T8f3W0N4TGw4_4~5K^2J)qHlTR!?=&I
zt?QO9uUFFPta0{HFli#VCWdtx^9Bs#wYToC>vcK=T#p*C45q#-T0P_`UEBk&kI8v~
zb24Nm7LEm-mE(5CPgqY6*0(#^xp_-!>=&cE48TFg&6exsc-9pUd7h_xM1A#lf%r}Z
zDY_{kBy6mX8vXx`o&J-xiRHAXaRXDg8e`v(SI#9^dLF~_+V7>fomDWHZ#mq~43rJ@
z^$?Od&95yS-6R*Qw~ZCrgL;~K$+vj6+8!SQs+Q`kt#6l|Ti<u+TpVw=ogB|>#?CC;
z?Xh0>fcj)Vkf+-y*gKmmKHzg+vjMS^w_y(xFItVaKRFJ!XtuQB@!(5|+63JzWRRxd
zZ2wNvM*E;HP#k|O+Jbh|Cdy^7)IfZygmsAh(&zh*bOK88V3u4l*;$NlF!}qSx54Eu
zm2hBa<iTrn3!Q#u!vz0^FI(CayETIafb(4Ic80xyYA&5E#A@q)0Yg2tX>-E)9#z}g
zA+KTN%oW6?5n2%ZunwqGZVRch79<>`uO{^2sAB3Fdvuvcylh;}w_Mm&^;>6}WO^t#
zme`@WkQA`yN#&T2h-Q+}&a@w+QP;bcN8HciN5lPCVSG$5*J>GPN`JB9dl+AWXV%&(
zxlnBVW1&{Q$6h0!GmUV1_DV3MI(2W4irhsXSZnz{U%IlgQYu2M0v&F3ZfWpA#r%s<
zza(#+M5yYM#Q^49gL%9I<Nrh0S3p&@b#22Dl#~YP?v(CsK~hS(L%O@WySowTZjg}f
zmJsRgM*fZ7`+e_!U%lTLdoa#9?3in=IiEGxtldeFO!h_hF2cO(!wsWqwM#g3Sy6Nj
zP>g3YgEY}BL!Zof0XE-YlDoguE;BGFX>e1u!RUM{oXl*ZyUj*s=d=8|^0#|n{t3u^
z^J2KIP1Om6Su3wwY}fesc$=dY{CERurNOx1@A|799tCgt>6)+AZ$4)}wRgDsqH)E8
zvoN|J8yicj)qL(|i%`D959V4f>&U$|6f1PscYQM#OV7??w^DamG26M!7Ct_mhgN^#
zOnxA|_JKW%t#wK9@y(7!;Uf$<`i>_&$Fra7Epe8m?CwykCAP}=vSrf3ZH~UaA6`uT
zg>^12Ju#nhNCX!7{6mw(&4jdkt8M046koGzG^Vs})_VBsTs|j0^&~cRmtn@CcD2c(
zHgH%7pX@KD3NnM|na^4ElNiGrW>8Bo1{_U|dFp9uB0a3eTt;~7hD?SjPihMVP_y-%
z7g+D643gTW5_dEg66$#hnKJM(3y!j1;h$~F(MvzH)vaWdWZN>Bn>J^onrh7oSqcjo
z+}@c-n(xi)Z6CEhC!f2tcfS4>ICy)3KL}`(FrTc+#O(qbEkf<5Myy+8$+FlTz&tJ1
zdDz{%K`&^PT@kvQW0idKX7~$ss(vOWq)@pq5DwVrj(r;^mn(YXP+nAI1tPZd3t4J*
zOl?z2j^<GM0>^|b>aVI*>6Q=V&kMYopPy%R_{zQo%o(NcbA%Xcy}kifl(b5V1GV|8
z3)^HSqp~nOo&`+l-0Fu%FNB1rimU%o`oBJTJD^ltE$>%-;yRu9XxS8a?=(ryXgmgH
z)I1jkavj-M*eJi^T~F8>$j8W_21q&E5Lq~0w!Ay9rYCwAoU--Rk>v_=kx|vtb)|+j
z@_;AdAdsz`sWh<6xsOW$M8mnvvJnP6^(htMQ8o|{#*x=Y5xY~coS%XG%_%v2+`aWf
zeqECqvkejPB6nDMX2ECI__V!3Z&Wgc6<sq1-(YYC0k8{jgn|xedImmbI-xU`9l_cC
ziC~C2gNqxr&ZWXOsmSb)*=`!MN<$O+#u>egegk`vmW+Et)QZDQF~ScXBO@93@R=yg
zcG=l&JQ0N@wjUNPD75xwulo9-YQLHH^_LyxZkkQ-?0YmVm22TAvmOutIiSrS7HTA_
z)AHOWU6y${#hCc8hJ?0`>JD%Qiw#n(6BnjsgLy_}?-qx?U;>qHOyDSJhAN#|ady<v
zC*S*0jBmUT+Z;s1wp7}58)-CK=K9r+J3$=V9mR&{AqQ30xsmz@O&4!7-$*v~C$`z8
zpvaY@UHFbWp?J63&c3ycz|ODAQAek_?tn<ZIe<{rH!zqNs`zamfUIGNv~kPf@dcZ8
zANrw(hGS-fQ>SF!1F%B;e8b&MWz9@#da<On@7bQ*lt5tB2Zz}{#&(ppC0%^c@^R}y
zd|rC9Hb2mzTl2+{`y?4xQeIab?-5D5XmzT&<&~i$`v@*nMG#7e=}(J&hj2{pQM+$K
zW{1s~tf&jR)OeR<$kzrpU9dcpE5vWpMBBID7DTs;e`;1!&ZFhtfG=sNRh@IU$TsZI
z4NW`FXoMf0QB-aei2<_-wK1DqR}24X6GA5mtq`GrUSKk0ljC6=iB>8@VccZ=F_)m7
zpG+!=z;zfESA>kpP5aT?@Z2A#er0*V-*6z*``L^BI7)>yR@WqPYgdm}m&>TvdVS)-
z$#l6zMJmtXa8?uG@p+|u3J}l??Lq{6pG(fhlv$`2J!E0_=oTox+X+G(OgGTciC*8@
z;J640g)taP8lM1)%lcotV{_^k=^Gew7(Bc<kCi|n=TvY!KU&dpx!A{?arlh6t%xTS
z2p_$hVp^E)4cIL9gF|t+p9A3I%QTu(^fm{-$PwqsWbz^{?#$Mg4{@$pks}I;eIIki
zvu`x{Zb(T%!R(FsJ+H!p!=~Fx3)?qrDGbvP8I{SWFCaXd9~pOCcU>HJoY+nS+hq{z
zx0{-Nj4{`Ul!a4|3m*i&Q`{pd6b9?5Qq0pzOaq4>K9TE7H~1kDWM+5F`p00lN3=3P
z1vBm4W#YiINN!5bQ~$$UZ}dhEjuDPIx!iR4QI4h4JM0fiv4bDmA7gtrT=46yExK&?
z?TwG_{ARenk&F8t5FuXo&|AW}A5@|RB0M4t-mm!EDsbz@UOd(|+_7s7eAw=)@kOw4
z7#xIu5W}v7|CKI^ScE}AU2Zkg;>q^4{dKxHP?bUUX|%4cMt&gCGKxp0m9}<cW8xqN
zg~WA=A0;W8l4a3BV!jKl>ha<RnPMQ(A~BsV*}aKt>wU9hd))A&xT)ho!cgY}spEtS
z^i*V&-0f{DU{nS}3~fVmp!B_29ugX^z%qn@kG<qOX;8FOF<aOnZVVibSzRT3>#@|<
zDu28FOupGYbRCCTfstG~Hf*hIg#V?ztApf)*pcM$;X#b;{s-vwb&fmXTe<fN<Ukps
zLkKesF&q8tkk%_0O{j7$1qCzSxwtp+$uB?qK__K-WuanVa7>#H+0!ybZ(y}#NjxJw
zJr(L(s&~vxM(^I2QzlL(V<762%8ZUbz^@b#vl~Z{M(b5voQb6MCtDSI_R2K<dkP~{
zm*1t_GX>^X!hCn{6ctleL*%!T7Baa};<6QNgeAvsi_DF!#O&?kO%$HV4^>JtNIV$5
z(Nb8zNMFmdd-SN!>>)eOUD($$yw!Y<efuy&DR?SXA?k!}TIfyYLL<$=(O95acJ%7Q
ze?}4hXG1@x1Zs9XvVA$-3S{wc=9NA=<$gW)bWC>D<2W6rri-K7#QRQLze3&YaCCB!
z-Hd5+d9Dzd_-qJhg*bQlIWnO?H<E8p>SFJc(?xt@{}&{ojl0>$iN?NdTC)Qe9Ubql
zPM;Cf7Nv$B>W`q7at|i*84Q>7mA;g=)29#V<*6+$XT{WSwSIRP?MG<ue2-6sn1O4W
zOBlhAE#YMR_F7;mLsTo+t&rQrUGmrZCGHG<gnt-!^udQS>6i@-dRN&tWqhx|ayvco
zCj0#e-xhqt=Ix*zS5hXuvqG=GVHHCsB<5Qy5qF@#JG)Mh)(^`jq7Vt{!u!QS1|GLI
z-liF>!#dp3WIqK>LaD?gX?1}lW7=xSn04Aimj{F|IOUESassxq!iu_43KBg@yS1|+
z!z8qeu{x35S*v|K9sMi>c9OEwgbCMacxGEytZ3mP)Wvbt@qw}Xhqu@Z814qC<8&<&
z&#2DPsC(YiD{;mB)&~%U2)>n3(D+HET}nLlH5MkTUByreheib7r4@?0+`b!To7T#*
zg;$T2g2!C;i+*nfG7hbds$uD5X3gGkJi@c(2mT)q1F3Jb{GaQ<;WAQ~Jd~G|yaoKL
zEou3!FGt`twIot4YBr7pQZrP@d0v5|r=NG(F8t+*fdXej7OF5i%k5x-fywJ_X)Jq9
ztTeUzu|;YNF?zvVOFSL7k#JO)C3PJ815A?k3raeZt@^b(wm)pk!NA))II3l=na;C+
zKyGCa;?t<pC=3~+VKF1o?i68&;l37nnQ2|YlrCPbxwM32?s;~$2MBBp&-;Xx0BdsE
zJpxz4F+ZTioAj#e-y8U1T3NFK|DwfOF;nSo+kj+Vl5MF^NbdU3uB^$ecaqa7@kEBV
z!MpJIfmQYkVWX^X(v!<_N2D5~vx#N(d+>NV>-zr`ix(&Ljw$Q0*_UJbekcz#9+Tp8
z3Mp^f1Tw0!1iT)JuxM2Niu?dh<gtbz<4Pp^T6mr_KepL}Wom;^<s^d8r5KK)%ym3=
z36Nn)unMK^7sP%QC8F9)%lH&~>io@Ec*7bP+hh0UVLw95m~kU7{qucdz~Q7TC;wZ^
z{Fi&J5b4V{v(Z5CUr|B+0T>A8hR*xZ2o~k{a_?4VIgRXKilHG@AV8a+Lfdu~h9j2b
zTRmdq%GJrIvH6gSOY20W>Fn)7(w-Gch_3vQFI4c{E<!W76REN#nQJXvA7NJl-57=-
z$}ot<!pjH!4<h_`tPEGv3yFsFo^%u9S7b@t-J4%YmhtpdIPKdeRi4eXVLv9%7pZZi
zxZO`&c`kzv<THd(W(9{|`fX&2Mx=feKBh&58jd_m;&Ew#aJ`7fN=Mh9eT%XvM~tZ<
zNy4t31<`zI71{YOB{xt6SwJ}U*!6GP4(PHbUGyEOWA>3A^6o!R%infNqNr3#lOz$c
za1Gbs0=^8nBy!cyFT}GtAaqG6Am20v|1qM@p97KB>4Gx->YzrFzZ??UTlmRaBQCx0
zm*d479ajI6BB6_zOk#NXLB94-0Y$rVu*RU^Idl>d70el_<33`4Ca)ZbY}@B=dHpM7
z@;ALt{a!kxcuI4RW@b`cQ#pAUi-#enG-EYkkwL*vKD_9Zn!np{(PA(h+4}M{iJwSJ
z=6@JBExvJU7Pk+FXhxMZY|k*X3;kgj7m7UDb_H+!xrM;~!ZK;;<y7)29Fu#H!cueS
zd+$Hx1QVQq1kOk+)od-Ro0`<Bs<8{W-`+<XzfSR5Fq<$F5G05OcuXQ6Tld#aAqx!z
z4!vpcrq?S2d1y3x!QH?REEFsHD-JaJx_DMaM>1o#s~3M=KY}4AzLM$HVeGCs7s*F0
z%j);>ik_81c~xql8|RqlbXvi@hQx^92RRVMWB3gXWlN;IL=4`cful<>MogFfft+{{
zI0^$&)KG?o$T=D@WGF0&N*q0lszC94<9jNl(wL?y#$sxECSq+jI(rtDnDP9N5KRuR
ziB%DcaeQO{(BD$ri{G#yLQRfx1MOW*W*;*lG=aGi3MWs{f*FiPsNum|1BjHT(2)cl
z%wJJef_O;ACR7Q-M_&2dTxGW=5h^})qUk+0adVU6se=`20j6D%O)n4R{dwwm8Q-?w
zCpqxHK!!>gx2vvZ9EL)lmN#ufR8omr@gAYOkP#VX*<l&`-E>Gpf>c#wc&y`+MM;!*
zsi^WCjx>wsD+j}r<o2W}V_01L^x^TzI_Y}M7wBIG{9d2a-xH9NivjHRkI0KkR@E^S
z9NGCwU~YH0vB_JRVy5D+`3#UqbU-4<b7}(rx`#OxA`MkST5v9@D?fhm*K=T;i`2+)
zk~93cjX8n(>l?{lusLyy=Q&mkzu9F1Su%}Y5ouA0sl@pkXjPV;jky9*A~v3s%ne;?
zA9-?>NoJd?>)v{a7`2|JvHdc6am5Kfhv^fxx`<~u|4MEBOD6wg3+t_j2sAy5;XErZ
zZJBUB*3!@*zS)G;!IXL*JvOND{{5CzPY$uN@xb6PdM!C&6dVA!G1-!-|J2NXT18_!
zkTzA-!**!kDEde#O@h8)Hnu}keVcFel9Q7m-q#$n3Hwr&eEgVSRWgu_ESx%S9cI4{
z<3p_$G+<#34Ri^>K#BivV*l5x`|}eJ>EC|#m<3%>#Gs&}sPv8t9NeMCf3v<9FB+n%
znvs+|&_sY}J}OZKAR2j3!tDPd+RICZz_<Xie{_&V0F3igFy47u9LP$Zn%*HH%G-q6
z7Dw~u=lAJ^V}cPMYmowjoh)MxjB5G>G+oU>bK3vEQ>MD9v%J9sn9RlBI<&aTB7j54
zOArvpa|&twpCkP>+<(L={$m#Z!I)pC#~}j(m{0d2{a=zL{`bOvMnV4XyMK`?ULM2z
zC95$cPmKyRn(U%|d_es`iBU@ncmK#p!qAXZ{}+N~28n&<-z6eM34&^CYvqAf7(j}m
zypj^x-Bsd1+rypyY?+3<oMyBjPySz_`Md(alqq)B*hYJjZGI2{N(WScH^osCOHtsr
zzWppF*lmpP^EYC*K*iRnoN+}@o%!NA?Ab!CZDD!X-<}*OK-fAhfA*x~4@4eAq6CWB
z$$=6uYtfQzS@ORAe+u@`wJ3_j+sDU^>2{9*liMkZsj9~))14}n&2ufxDLPMRg65a{
znUDkVlLZY%&Zn9L0z?V^XVlMM;%v7hof5|&I{#J}yUDXU(TPRFfEtp3{s4IFVY6r6
z=qKZHKVuwOzOabe<DK*jg-?)T_uQM%_r>6sXkuYsnm?quMZE%lAetP@78*)SdH&R1
zRe!QqI@lyh>6gHXD}ZMR=kNIEF8(A&H-LsF3fQVk)*mi?`!6@wFPS2N-o{83jl?2t
zbvchh`=~Vd!~HJCjwIzN=YlhdpJ1%>ZO~-D*DvAmwnN>XEzgunWer~qv1@TU|6OBQ
z0$NDPCi_^SLVO~?RUk9?W6}f=sjN?sQ02yEC+ql;V4>^Swm*}>ydhxW@OjlmlY^Z9
z(AA@+wPWtJ-cVdC@#2pT93+5daVBpdpp!R`7P_PX*oxS4KRZhi_(SwY;F@i^;Uo$2
z<orJS!;aS*sQl=^hW^gA5^UT1By~&>LFLBJl~NQ*JBXy|6kC=`i;GPZYfDSjcXxNO
zHpr=ROG?y$CP!*ZX?3~uPc$M*Yu{g=0M$=N#y#u5-LrV{W5b*uOp1|7r_2S2t;;?=
zrwN)tLcB?%QsrSd?49@NInrE{ed_)9ehmGI>v6_|4hf)Xz?R*QZ+#i@K--izK<(YS
zkL}x7VR(^hy)8QD^Wd%g0SINF0j|&p*WDDl&NR>|E!B1<HZtK;OvX}kvcp>m75w*8
zgbspW-f^fvpBSJ8S#pefuEKAH(oV@6=pf8NtV*L%Ra1>@wfsb)J~+zzP9VY7WP&3R
zER;m6Ri4SEdn?oO%@;Yq$8-2xTG;fs6#-o$9a>4{eB5TdNv|&~YB<vk=!t3kFr>|p
z?G3vm*(9lHQ3jZS(`kApCdE5i9(iRf$%WwfndT2&J9k@JA7T+OXk#f~_ywS)-i^4a
zwm3P*-Ys_VGYFNntO+(hZALhwI*z}6#%#=arXln8cMaaM^q{xJ(c{tLt6VO2BkXn4
z;DKowgYnjW9xGAfLvP>+%w>P#=2LMa(1Rl*Di?=mPc~Uwv@j^{G6jih#<QiLa-e@C
z;dO2x*WMY1$GTGHD+8^1!MOo>rkrjc5oxisghQYOpUd{R5eYe4-%>;kqw|_dGP6!t
zx%e}hx4b$D_XpmdoHv%ku37<bcm$-R8fk3$(?|t8CoI}*YoCnvs)e7@$evUM4u_qA
z_EwVox4CEAr6AuI2r&~?@;CZt<mvHVXFrn?fz&pjXd^1LMf)4_mDEdas~jeI|K}PX
zzFrrnJ|5nSRIPq4ct!>TbG9C)5b1b1U+vhw|Fv;(iEkz(VkW!1U;^;96QMg&(j_D_
z74)kEe({}o&sHi74Xc7MoY#slFt9*F9;Vsy6%x)<NFYC#GlO41z&i|!rR011E2H0<
z>ijQIUvU|R0Y&`pFONd+X|kSWFcv<YDUAAs0q8T4I3$591LQG~Xia7dJs$wJx9ArX
zseJQ9maVujR7tPhOW^RP_)M49a=L;H-o2Zya@vtBq|yGNm(HzITHQ7#Rr^gEXm@sD
zA*#M&<ikirdY}x{L*(ToId5#lSDDxfWd$x}&qx{B>}ef3vR-^eNMn>sC|4m~blkTq
zz$<3>91M>$HtRg^PDJ~W>LcKtg%H96;qq&*2O;6E(y&AR-u{a0NRlJo@P&ls8ilqk
zcv1oga35--EW0Qgz$kJfWZ^6}iaeBylvi6|SQAR5>ye%bpZSv3cq9oZ!5FhimvR5V
zKwQ-+j4GfIj)^oVIN9D%DNQNgw;l2nRif}I-$!t+$x_l;Kpj)i(#EGOa8jOz(a9+*
zlRsv*o`Hqs>@FNj&_f`bLP>15UJ((rQV<apdHJ9rilAQ9`^L-2p;GHYQiq^|?*u_n
zh`3m0Uu8sTKDrt*HZwI>yR*$NwQSw>CNCbj8_w^KxL>6`yoi)t;$)#@r}#1^lTvgy
zJTxCrvn<lK9lBzCyRwjO5mZ5ZVqNJn9rXf4AnOA~LP8=dHD5()+@nGT<puWp<C@pI
zYGXi7cUqX!UjC)o;T?4s3_lnIf5&T3)fG(NLRQ@oqE|<k<6-=vRBS{>mirw7_CN}2
z3B5+SO6>?ild)_&>pKdqnN`Z%722gDH;yefT4$h;Z&p4EQ4S8DHHwIsIPW#b{TDv$
zy0KgM_r*rDsYQn2B>~R!u5xNlJJ*(r77Gg7(VykH${ZD%6B0tA5kLg1-ywW_eTxAE
z;BAAE!iRXG=y!)CnYS8?*+gUY76ltyy*{X;L;wszk`5dRjf4R)Tj2f%63p8lVP#42
z+m23uNIMJls&xNnv54{D*us3bmzWck?$n%4*F^;xJi*<5193J&FR8n}2>{_q<KZT0
zyW^;@$`+b<2Z)-CsM(*reI?}UlQEOe`zEFp^Cf96c9)|iMMULWn^=N{e7bv;Dmc!U
z-2*?&1S4lK#aWvxi9#ZRfoBp&B_)a9dpv&z4t)Flam)j{|4^=0YZ@0va)f=o48x%J
zbfQ%Rf7D#6{|y`XzUIyL!-f?OEJit_Q?YD9bVjrQ^`m+@N%6bT-scD?l7<0@{1xn<
zrv!a`DQIYhqS6uZ#oY3~isyT>S>(6Ah%oNmn`p!ZIMpHRImc7~hEQ<k(_>^+RrSvP
zJnZ>W<XKUEl2D`tE{l#3TzIK^MUWtFg5O8YMmOo*MM)-9Z!oB925=9{@AO(U6<)x%
zKS~pwx9N1r#0g3tB^$Xi4ANmUb>|Q;CHx({AOM7%+N%-bWoCUky2LE;U85V^Zc66&
z*&ga+=c*yFVNn-qEew^6mVGu1ozswrS`@>a*EOZ*FE#Frmu`nsxDT%`J$Ujfh<AOw
zyueIV+FwqWzklO*GbfVDdK48>Z@wbrdWDzM>A1WmYmAH@SOkglb-twi$GD%cTx6sf
zz|j@5jy0G%s&^$&SE{m3KvG)w{u6D#J0v;5J$e%n@n!lVGYLE{`&^+Io~Gnj8HA{O
zy_e3Sh2c;!lzc7xwkKpG&)EhlN{!_s=H}NX13lkDD0Be`NJz|q4;rZJ8>#P#k>B@6
zVW?8c``vE~BfP}_Kvu?}h>UTcU97RUX*asMNWo3?RaUXU&g3%--4ikV()>uuR+{SS
zX-!bG1Yn>)9lVlRlJQQTho(ot;FBAE3Kt&%n6*FtSK${4U<l|hJ5CEqHU{JKfe?>e
zW(M12G1C^s&Nj4(OMY&S@v0ZE-OT_?C{fVnFqYhW5~JZD^bC?7Ajesv@Ov=vCI%(%
zn*)SkCg&v4wma{hztJSe3gs0P#BptWI6FdLsNM21%wUt3XE7lGh<3~Lx($6NB^3r`
zhPrbX`U^vT$5`OpqPXjChVxYhI}5dD$nZ{89ibpVKvtmm)xTLfPF=d-ES}UY;hUlo
zdfx@W#R<dGRLH(p-;nGMBsH7LsuRlQT%%IX<5PYl#!a}biPs`6Lnb9-zx-(T(%$Ak
zt?E;m+)DsJav`R=U<{zqs1#-Cte^iu@t$V%83>J9i5ZH<iDST4zz^=kH#L;hi(i6_
zAO{Z;y8MR|J+(7E2#Ut-I)iH$scSyp(7*sC6j!}6Ee$6ymAt!8&JFF8?D|;rcSH?#
zDP(y5jyos{tsE&Mr8YO`Nt(Pr3D^_QlE5B|b^&YOklKq4mWkxc6FEpAZ$TBLR}`q|
z7041`pipF|#Yhs|!ZQrI)KA<D1YTHE<>m=ms(!b|GkM;syhiG1cw}$Wd8y-jLK%qA
zgtw7--bfYYB$~67D-D+zeqR=s4(#ZnVnOxsx<aWSDT9KA6>>s6YIjc}{G$@H3wwR`
z^YagsQ!-3Y<q87egy0!4L7~tU6+S#lL64s$8Ni`vBv*%5&R5=sew;y3>F()4#&>a0
z31UI@7TH6WE>fFzYBl}x6Y_lb_})<D<mS?%5&jv+`8xukySux`{Le7#l(X&$U@b1~
zPY6&T__G?IQ3F8VOjg}X8Y}}G!w|gv9lyR95)ABE*<(e*e^Sa7@+>!rs=An>A_}t}
zihqMyIZ%GC<Rs(?pV@Ou^qK7}IN<)`P!*0aucOA_K(@R?g?7Zph5*6hH|rcG9pXN?
zD)^ti=G5rS41bBL51?Lycn6*}Bmxp9=FXb9F(74s(#_Ryh@2crIV-Cu4TaD2-3MKM
zlE2RF5e5UnBKWM--ht)k6{z8Uo7ngmaO!>+Ip!al2yfCFdR!rEx>QxH8NKA3FDc@5
zAWG(G)pU?>$U2LRg{4~*o0w?S&<hFW<>iP}p{(9n9T60CT-LA0`bSd0Ko*3nU#j_>
zs;ppb>_9(hUL^z}syIQ>I5z7{vJMQ4)VcbhkjtRp&(H`%!K$OFsO)KJIGoQ(EL1o*
za}Mrv5;F}(U2Ua!Qc6Mbm#PjLh8GDuEX<V14r9Ihlwhm#_AT8~L$qjPW24=^-w+J}
z!<ZqLA|T5YoS=vxYA2=a?3oq55R1)M(7(spC(P?ehrn^ad=c%>wUbN#gtTDbj>v%B
zk<KiA3j<~I7~q+Odef&*fIwQVDN{hLTsFVtYHEsgqBCqUM^Qhcq)}&k#e3`coz))w
z-J7u4yy&Or^MWEmsGSNrJFjCE-5lYFegCv@d2K<lpY0<M!HzB=oWuR%5jXh0A;;*5
z;h-GC%SzS0RO+~JIw9a)B^K$r09ab=l$RhX6hWeYL;5yV<t$Wx9{v^>hdS|USc8uk
zoF4|(Y)3B3A63kW65J^BK|udoJ{#&J2pFw5vJ@a0d<{1zf)L{^$ux9OXoSM&r8XZ#
zir@0{gT+(g<}3Hm06J~j9atE)HPK{%QheukJ@67<p;C~esZ~p+Sh`pqNwHGUIdCDD
z5fBg~D0b?y<Dm~@)PJPH%c~s4>r2<{5y2*4bAf?=i2S4>6Z*uuJCG&xqk8c5F`1rs
zkMzNN4i-|2_pNF^3g+Wbf1btv0(=}&#BJkeQ!oC)Umno0X0{ll^S?-I8FELpm@;$@
z&rhnonxXi$ztP2Pzj%5mfIDuTl$~t&Z~ngFFG=~2zurUyUPe(OV3<O`Oq4kizyOpM
zF8ZaZ+Tmbrb1J#MjQ?@PKbHnaq>gHF$7o?-DShI}$deSR+86^S^`Er;zkiq21<01u
zwO~PqdAjV7e3h=p$-|@Vjyw?WUlxRx0I%i@@TR@wyjibK;U~2K7Ksor2y|ZiW(L(K
znvd~7hqpxE%RE{F;S7H%^M5}b_$5oMLt_-QG#$B&6a6>kGXPCG;V%b%-PaDfVGlN}
z`F1>=gz?`<iU3H0Zu*E{{>%^hm*MBc0-%4(?WMNO6Gv9vA--Ppi0p`QxUNXPCjK|Z
zBFkn#QW+`VIVkg$jXXQ}^;F51Pix6YYp>+LQ1Of8|NNUqg1E2G#r!wg7ee!^9x`(>
zL;Q0#|A(GTlBm>F@?kClf({1Hi(10TPFTnDZ>v#7PsTzyME4y^yU@v9$h}g2+Jk+k
zN+J^kL4{QS*20q*@k5iYzSQyKx~0gzuKe&nxBTaaV`|WS>DW7O;KXC<A{9753C|$9
zUFE%vKh5_chJ<GEb5QtwJW?K!fun5=2Bvk{j*S81=`P2%pcDjBk2WP5gg+}+2P-%f
z$p<DucmOSQNeECeoSknbAVB_OUE51JueSWML*u<@>M3U^wXJ?prbVz?!Y)|es+mDw
z(Sh?X>`f!ix{{KmAB|qtFhBaDQUnWZA8nx5x3*v_MesJZsxx6#BuK%$MI>6!RM`HE
zCjUTJFChpO<7jF>bVNrEzhO%%A?<H$^~kW`Pg8V6>V<&0>}tut^iL-Kg#MpAc@A<4
z*oJomw*`KxNaferMr?7MmmadDdVV)78Dc^Rc^@g}x73e@Ay_KxOax!Z&|}BMKm#y+
z5&|SVNN|FW#2e=G3`8NG*3++D=EiT8|49e*%S`_LLyuiS_cX8hAG9aD(V8b_yme&J
zo@Rp?;<Q`TI%-bQ`m*)SBqlPi0DzL@mMDQU{phA4``qMQ{69x=Oa!8V8{Bvdx~kb1
zOja>Qlq-_efGF`p&q!emo!0&r`1tY0D-)cvIQ>|QLIo2TfEFTv%JdewOQ8VvSSa>}
zsOn-c!gj$BBdR~`#7M!>qbj3-_ZfM0`J^kTcDVX$wJ}#v{C-eR()9{lnOm-5Vnz>E
zun!=E)CWu$opg=$O(^QyRh>&@?Q*o8c|_&8LD&mmxc6ROQSsGdJ}Y<EJYy)&kI*u*
zLVzLvK`sGn7qzFF|9*S$?vTq60z~Q$7D{uBxBlpe4T9x5s=P*G%MXSZ?e}j>H<yKf
zy2-u2*D2l?BiG>!1SF9Y|AFP39K1ivl-CyY+?mu^hEZrn=2J~CFvvdW^^GmAWhmO8
z(CF}$uE*B@*<Fakc?i`{K-z$?Yuj;}i1kmL>^T2V%=pJ=H$N~4EMA*j25H1&F_5kh
z$BF`7=R<TNLK!qbx`mqITW|q&ofi~}fW%TDL0t2UC(NHgii)-Ce0(4GM^gfrA8>Hj
z4ETk!uyF)gZD@cb4{-u5r^<>g_Ys6dO`-${Z~yScG*Vq30-7LraSC8>LLmpiE(O)j
z*7XkxfBSU^nxDjQq+82|prXQteBSkqQbs%s#gh#I?n$aGq?Sy(j7{s@P_-rG8-Dl@
zPc;Mi=aoxVyd%TN?~>cB24X6{F5y;~_h}~P-tb@QdgFgw7#vL870}k%urN3qt<m`<
zz)cWo?vXWHp|OWvU;d1?hyWHkDQzp)ql7-`yEpQa&cN0uKNg+Q|9z)h2%B!#yd4aJ
zqOJ>7LX1BM*pqF@0dm^h#$l@g21(%oa=y9g;e#IPH~<j3!iyQ9+!0nk3LZf5D+sA~
zC`#yISCZ6LvJn3=>QV@T5elXIS2)04`#x*Mn-=5v6bUeiRT|_Pooe3``2NIr24vZG
z<%Kf?NXt*64h=sU{CgFvusj>jFpAWMetP7x1b-2Yw#Snt1zV4T_nyW}fc1%~MNkcU
zt;kwYin7)}&-=yAfBf|j2O9-@I4;nY0RF)Wy&71^igcq62}MKmwTe3^m?rAiy<S1H
zpK1mpAc<Q18(9p1B_%`*`x7!Sz=H<w_dHHb0hU?GWS3WfJ)Z|2@k6?!-;(hcHhQkW
zmA?GrB;1F3QICSs&G$ZVJrYGu0|P?E0N>}pz;GT7hxPW~n#(cz#T-{?h`IMjPeAgg
z!z~6Z1N?T~{PWY^Se;~b9pSwO4%#q}IKVQ5E|I@H9*p2_9v#%+f4mES2NpQDP>tvd
zgI_#c76dN0Y(z<V@x<+G3UFoh{l<wIF>9TLe=zxfzslx7H=Ih1#n+KyaiagUfC*Mv
z0Xb-ZC*}M(-}vOKguxBN7T>VO`n1@SergAB_kSK<Df}}dxw|A~(-0rHVQX$^<dzbj
z%_17IgTcQ0g`A~NRwNZ3s?X1hokGCcrkp}(ZXutv_BRIw9jJe$JZ~snqKBnrG%VD}
zm)G_qgWRt(Q9zw!CM16$a@ID~oNOt^k}w1Ef06Y!p$(#|Yqna5uI1wDyz{jY_pK}$
z(6@KEy|+&@)KLx%SlF}m7@<FHsZr#p_|1fNB<F$rOe1Cg<v#<z2PnBYmSUsuUH8`Z
zx7K~jpp+(v>ttQ84Y7#sQIcn+@5T<$lK<lMD&7-U1NEXWV(d60W8Q8_e!sB5SJMB!
zfJH$cxos-hrbx@p>?N7V_;_wB9jDb!$cqUbN{b;Ljbwvs;d<I~9a-_kN=}Jayy>#{
z@ffS_a5Ot&xQjw?=v`$gO=SS=sXH|E%dFjZVP;?Y2{}F9(!s*OWTl++GUm7D8sQy%
zjpydHrHmX9dm;a@n{ii|mZHgDNVh)j{CfH3;|+6>8H9fTC#T-jl?zCu)Xxi+zuZ$q
z+vk4;<^Q*_9T3GcYGYzNH6_3y*PI=-!9-SGc>Ckd<xz>y#!>8^>~DCv{JOSHq{Ftk
zs;#G!+>p!N0t=Ue*kKRWyk`?`cZt=fMZoxEwCsag1D!^SYHkYQ`Koz-oC($lAB{9y
z5MscJd`QYPbo77&Zhv-WLZQSgbiB`YF=0NuxfFdw<{&XT2fL}XslRN+$Vc~`04cR<
z7KPn1^Un-eQ#@U!vmfgX#{-sBYfjtYOjp>$o;x62^lWWP--Yc(LcwOSiHTv~&bu=Q
zdeEtB$r6mOSH2ElG@`4j3MTWgPd|luU6J*DSZ=&&&$hsGl64VbW6Y=av~{D3Ggx4%
z2^<x@AT7$l2h6Nge>ff`{Eb5_;GL3psJp)+U}pBC>Ql186t9!@1-khQPRjIzpcSdB
z&MsFo<L;8!*q#}mOvK0anKldVeXZnqSE`~ZvG!}j{r~eA;GYsS&O(xq8_sf?kQnQR
z_%Z^hyIRG;CJV=YA$V)te}`9(s8|X6=B8#k!W@-sR8642y)?!Y4FXm6TsDaM0|68I
z_xY@7$g*EJ<QC#Jl29!(+}<rUqo|%ulO}>LHo88M>pKJqFPWahzO|8h6jJ~_`XVtl
zTsJ7}gI@ZZE|e~4!UXMRr|E{`1j_DblDvujq3&+AVxj)JxsQ3r1*=7^w^q#gnxh!Y
zeCu^I>)P}vjn0tIvjuHZtjk;;j6xgDm#Njop!BRi|M=d>T=6lEW<q+6p)*gW8A`2+
zPzZ~gyxxoTu)pB#FfIyD8nUL)+_L~$JyFy5)wH2D27x%cAC~rPx|&6oN}N*1{e7hn
ztztShX%}~976VPyMsGbhFLvkO#cmm|eVsp>s@yYAF^pYy3Kr#K6pY<E>`wT!xbtmX
z#-vGu`RD}{4YP36Z^34axWd?!llBLNP!Pp>6w-vMQ}oyya`XQ9Gm)F~=>9cM{vaBk
z;@NHSlKBDa%ATI!NfGg=+%7YbZ$sUBQ`@eD6=TgZzkGcFKzR~*rOV(qriN^VESwNd
zebiu|mW0Yd!>$g3L|-=L@;3@`HVex@m^xFhqx4NMm*4s_n^?XzPTs!6fU1c8O)#T%
zwn_|ENQ8^-$5J}2bLMIZZfXFux4*q5T7gDYB214NMB;_TRx%>3rF*Ji<4i5yTH><W
z_~)F9of^~7{ov1b%e!{NRGrLgGvbtH!N<$i*Ky;-Ax#;zW-rkgw}6UB(}l9Px=;}4
z1+tQ|O+mR4aZpvt<r;-Z2;nU6_r{PSCo%^&<*N(AhX%f;TLUeXQ?7fP9qbnimD#X>
zHj>j#P#Ah>2r$0|MjAwwa{<nW6CKXVLYRKPFFrhquRP$DlU`ETWiM>Kp2xO65FACT
zRU6C+$$f7dgkbT$I^R57WOJFM>3a9P>-9~y#cRK-5W?0C?)zC<i0gSGSA1`8FFxc|
zeA=ze%^+f#{i}5pKH}-a+w1QoOU+~+uG-n^QC7J4I5;E`h%m??&R0;Vkid>Q#HR|o
zH;Iu>^t7#`3#NyiS^Sfqw#0|*^_#m!p;(3zb>2veye(J+WKnCdz$<MoE22Po7#=yr
zAZQ|V)C&KAhU)^VA5{<UUP=1h!J<6WmTo2WL0K3@YRJ@PmyPZC`QsP+1k*OB=q~E{
zf8v#W2?G2!0Zq0LXJ=<SC*)cL=<M6`JXhg4XrV?5)7?(3y7e3R`T3Lk9I*Slk$Y62
zQ1ae(an~zcX0yaJUDRnxf4#nT;c)|+(7i(}JFald3orSly-@&Hvg1GgNZ6eJ*f-2<
z#m`rs$wUVA8_RB4%Bu%0W(%e=ne3o2xUW5c36{=pl~6VWUU2H_1ZwxJ;XaqCm4#!i
zmB={)3YyU<f4?@z>jfG@gtM|DT*NVfi=Fw_NwwfIiN80$1VKkvNNFj}z)0n!w3CF_
z+B(+Fr^FS$Q-5(%F&yD}U3!M3uC6YE8mkz}lzADz?zmG$laTJA7CEXz>IT8o7Wbe@
zwrOTe1THN4%dezwai4g^tG>D<<_cUH&>tM4%Bl$i5egENpjo;FTN@+1H0nf0!xXPH
zFfg&UNr56I;#Q_8uuxE#&KO)|hC_RS6j(0{fi9RP>Ub5`DtF&d>g9lgT*>XnQi3G)
zz{RzhFKAkHmidg_2EGfP?r?@G4TNS;^VVJKOGx=E;hkhPx#u5!Mm1TNVKraOPBPg$
zA+^2OH}9S@g#||(dI5DD?%1V{EScnHWK&E}zIhlb&71n;(_EgGp7kzGOM}_==Nk1O
z8r_`1?V09`G(o)#rwEI$;w*8IxMzqL2TMy)ms36Cx#Y&Phwcvco@?EdA1{zm*4#z4
zPksbuPA(pK=&K-vDn#IMNsMx{^hPO9VWj?86g~6QlG^(*FKjfDkvlcTYyyKu6<2t*
z>=A!^^_kux<TgW_`@ij?JLrlAA0gjY8KbARWU)rlHkOcr>PRN<)O8+<g!^rk$Ly_x
zz(^DcAaHsK{jWlrT{$M+1%EOlE6@%a;Yo>ORZ2%POd8JaUWTV>DT3BF+L9GDO!!8M
zPschgC?x2&<*$9>Vdy4Cx$As$>U0xB9;HsG(w{w4iX+hwL)=Y`^i2m9%XE=An!X&@
zmR=Jy_duihBVZg&&M(IL9PZ<64G1=WXqhPxjf0Us@Vp!A@>1v1^31b&<~x})StJSF
z!*Y*!8WvnzBNN}i@JLVT{8$YAcgr1gNK>T|_nUPG+Akkk)R(_X6rl?1Jr_~MuG=2$
ze$kST9hR-4A~n{i`igcXNuqCLV6V|_i9+!UA}Ds}EG#u~Rv|nSaemUTsnoF<ssz)L
zPD&PiULs6kQCL4Gj>FbaCseoI9tI=P&G=w)MSka+!&w7K2b|d>=o+oDf8}ygkml@L
zd9POb)dv(p!P$mOm+28RLlV}%djWtbfr(%NUq<wN_(Zu&v(uzjD=Z(16cd40c!!Vo
zrrPsXQhqmu^$Rlos<a5!6<pg=O^Q}wp~t;{KyuZbX!!2rm%u}X7>9InnH#^6^k!`(
z$8L8=g+X*_f7f>I@CL!+FFv6s-$g^|-}YYV=MPP8skV=hUy|h*(M&YDT%>3so&IJg
z(P0LKvk*reOzmXL<K_QyK=p%c$)tOqs3ZthnLMMBFF0naJgd}QOoSczBukO|5MKKT
z?S8KQIH>IfpORHP(}J%TAIg`35%e{fo4!t((x9*)uc9t^wlsfAn!HtF#vHE+CPtEM
zktPy-Bt8g)i!gE*HYr?<YxXDK=j!v65bC+J?p$9<4j!gUw@wf|x=O{TIhxRCQarls
zJYLV>6_`#<&NSzV1sfT?*AN2kW4P#BNc7QM+8a`0WTwBb-*3x%=d|666})?pMdIpb
zpYx}7wi9$J(^i1RQmM(RX>^+2cK>!<tS;msWwOOq`C}nL#MBZVCqmfiRgNza4z%mW
zST%8>fw#}6syg__R?}!Q_@p=A7ZTn%sDk35X!F16PxycC=RN4!30*Pph6)u)(n~K{
zs1MmlWrTmIS_NwK5tUzgt%4xFO05pWe-vas@l!2#M8M<6_dMpq;xOc<Rcq;z{W02}
zYVviGesU|bWiJ~3{`lCcDPIQRv(rG&#D*tTVt}n4q1rJkH8zTL>~eAm*C!=G)Xpv_
z*AIY|cfR~0Gx?nXJtB&hv(kW!3|q*V6t&HD+(dq;lI-<xSmyRk;tF#*%*1c(>{TX+
zmlj{Lg|MsqqXeZ?gQ)ZH11OC)9KR81G@A#xcB8$B*<Rzygtdg$rHIAP?)OCzK?t~l
z$#e$S$sv%1gU}yEJ8e@|EVhE#f+tH_gszw)DpAZkhKdm>Hc5`j^5=Qk8SB3&mn$T|
z-)sg$qAB3zk3`dCrG(gNj%Mvoj)97U!%w)Tt;fbqV<#)szXgBG@o~6SodORz?soTV
zg!mLcMU+LbNHOn3Vec?cU9C>evs!XA_IY5F?5S!IqH<?dlmJiL?QPn&-U;8rYwQqb
zw0hF?k=%Ge_RoS5yj`#X8COu3jMi_1<gpYHkO)=7aHee^4@Th@9Aq~?t)I0jVu#Yk
z-4(A>XKS(#Up>@X>PM1Fa%5VXBJb8A((Za#A7DeIg>kN7hf;T`8p0Z!ZTDr8UeaW-
zUaG%#o`5*fc`c4CuioO6JU8gCYrFL*5DfipJ02rbsp_jkMEEChzM(M~*gEqSkpd7b
zI#w94-s*hS^0=38(JJ~Oi<<CZxg|Vc>9HsqPiyG@7G-y;GMzB8)uzI#8@G3vbQ6jf
zo6Y?7p2dAN%J%5S8A+L;e+k^n;*+7Xa-bb}BK}+F*n#-)Xhs!&om~SzSoZRw9yIPl
zSB1ip{u2KV`;2$I6#wlzqR<3+x+T#$M{8I59uj)M_1^5acD;L}S>ba{a9BO2;J;rf
z2$v-(wgP#CR#sIvrKvTiTg-YY0JtF6gCEde!gf8k78(DFFVaC2N>fE4?xzi{dnCkw
zXhjU@>{qDCbd5%2B4LZYk@@Yb=_*aA+ENT!G>_MO(oURFJci6An1Tq)WUi|XIWXzj
zE1_4Ln=s_4D>QLJiA6@m6!BclNZ+`S*N3}ui4=o~uyi6{YN)*uvaGlY^(8l8{VH>%
z)LTXMGKjaE_oM7WKNAO`pc92A<0a1U44TS)GZ&De68mVS91fpjeHA2&Pu8m`Jf#aA
zt1C32gxhL^^cL<p$N_1!A&|iqghOqyo@8V^yx=(u)1J&bT_G@vN2(GvmU~3N>3Z0b
zeHfVOCNW(?Hg@cK#JcXhM#RJv+sl2DraD{R!kz)doyQ~6UAGtPZIKctR>!`PY)n}3
zdfb!5Bb(~=SbSv&&{%6Z;BuJM@_ZDEr+Nn@d+bt9?{nI!L`3o>37d0WzW&(m_$Y8|
zxGBSY8;xXWup8#ccReL_-nFs+F3<AlaJM7|Z1Yej5;2r<E~&pJlH6<;Rb=h!4Q!L`
z;n!jwX_p7m5{>2nCw4`CLZU-rAjtk-zMz(PuNIL4mxR1!CIyXIi|-3pi?9j>as|hl
zop_85@12F6y}+X7S%2gn`89Nn&%*G*u2;}8AW}d>0I6AdR|qd*b3WWB`q^V#VI|vJ
zSY)8m`5g)9`&nz1HEac(Joo~!6-@{HzOSk&F@DM*qgLf(8}^o<HSq8Ak>###{wA0>
zqFoKC9^3kK`FZHF$rJ~Hd5lE&eM0_Suy0~-*Xw}>cdvF`tkFOcCwQGY)N#`8mu0f#
zavqeK?1^F~$yM9leQ%0LTt5}5Q6|Mq4tsKZI$Sw-Sx$j#$}nw|P7UxkeKBrzQ1`*N
zgNT{L%i_S;t$Nvkp^u<K`^3L#$WE&DoR6#8<ohy>t|b;d9K8_Qz;pRwr|*0k=vGBN
z7?Svbia8Oh1P86MyQ^?O1ky?g_1gz7qph8f2iKoh<_a|lUJB`$TDJZmVzHc2iPfmo
zb=E~dV!LY4h~@k7NG7kYZm>K5U?|<jg>vN15DqNWvv#WG7Uozsv;34tp0Rv+z$-@0
z2Fyt{UH5oCzEcsmnH>wv=S&>Mt58!euP_WZJ@5TEI@vut?EjKp<{tmzXg0M3%EG12
zaRoeu)jdEJ$#p;Jy!Vaajrsfftb<hLN#M20`Bh}$*jU`^N$)PTdk>fH>|H%CHjIjO
ze4CmKe(NbNCnmtNeJ)u>4I^x<7>!s)3TI;97K<{<9VguHS(-(2;u{5aLZf`CNSHQs
zrl;%Ct>%=FQ3o?Ea!vq6lP3JhWV3X;Yu+KH6NvRAAOY+9{F<2EhiD`uJS@m1HKL9o
zrfQo!S44CrFnaU^e!>+>P_KqsVM=Z^7xcLWwm>biPu}MmEp(|0qY%x5F8GbK!lYsX
zAqm?|YGI0CHRz1y>GxOI0B+%<^GaoAN{+%!x=4vj43eYEsxa9nAx*mMWBMZ0_49qr
zdNX^Px0i4sxKoCAieHGIm_HdIWb-8y-Cl^9kZSWsKKunCOWiU4%g$^k&kEW6FIBTg
z1B|8iAECC!3*FgCVIomYx!gGWIs(onCt#V^qHc+#f@t@qXu=M+b?=EE`)&d64sEQY
z|BH!+=G-DTIL};ywK3ul#De(l3_Z2})={;Lx7=c{pj1c++@IB|<qBhLhB9CQF94&m
zPC`)levBR&(Xzt)Uh<6Wre4cK7$Mw6UV(SLO|C}Guhn=VFPm%yH<<Rl<AhkIhsg|D
zC_lSfFxz5LpxD?w<Qo#pR|RMK(LG`1ysX0oK$Qp**800@w~=dMrydGyVpn^{PaQWJ
zoGbb%8hJZ;dVRyadUE1VyXVtr@ZW9JB>7!RNrcg=o<Ue|2A52RoOT97>f`YWV(_Ms
zHKFiQDUOztM4}N5tESPfqfc-c5_F@z)n>IM@@5$oeF=5V>HSsR`6#SaS=@&~ODigJ
zI5U~a6<zhX6-r*c`)-btLF0&o;3&yu>#O8YsS4a{1NE?X<!Cf9>^5Y9yqS)^1*6ij
z+=^BiHYz^(=fhp_GV|@OVS)$-VWEQe46g5*VwlrZ`E?E$6l-j(Z`9fZtU9K#Fhb}`
znWICnrlGR&K7G#5^+$YOu`!4RzoYh$sk?}QT4Gdt!i_b<5TaMNpV$K{K-8|Q`kiOV
zRZ&TR%6$HF;KDAt%eCy5Y7_chV7DMNx^hLO7$W*mMX<Zp?FWmJ&OxY9O0vf&K{uJF
zAR&-nKmG6qg={>sdNav>rqzr*h1(=YwT<)p<>KvFL!Velb>3d{5w1kx+f0^0M^a*9
z;#fS-Q|D?r;HA=HOHFpngI9HM=w5D&7v@VQfEDMXRHmk<ba_T^8!SRvz<QqHJ@wH@
zdGbSK-VIAkjEztO5zCFQj$REYJBh8f7k-c9ASSklY-H%O-OHH??f~M1j0}PO(@}*@
z8dtM3xm*2mmcyD4-xjE7hwp8Zxcd0D{%OC-;y~=yAw;(gN{&!<Z>oaf6A|JL)UtD#
z(v&q^pKQr{k~m1O|JbOuiY9ceUbSwp{Lx;70`Kjgn<R%1A&CmU;VeTTc#6L0t~^J5
zl`}`?<NW}<e|&4^C}C3zR-Rk#xJ1RcGf-DuN!?-9iB6-ftevtW^++zd7oi$wAXkG}
zbxzx)4coy8llACg*NP&H-XoFiF5ZTeJ1C~ku1t$3sYDSfCu|=S0tKIChqOQ<;RGp1
zBYx4@thPb0c2mU`LD)*PlNc{jEfH<r@@b-`tA`s~2_e6|z8_n3)?$4!e(xrREqDP?
z(zTxNK~U7Wb<c>TLUnNxgrVbe)>S#H9Yc;1@{l5PRg$2b8BO`);^0u-c^r)5NE*Hj
zM6>NR9J=!52XU$a7M*Jv;78ggKkm)D+S$Y8OJ-S(!q$?>lq4jqd;HLTey>pDzGb0S
z?~$9G3B2$L@#<m5jQ-WBdaKjbgUqm?udqmb&SSD)e}mcB!Bm+!b?E1GZs!ya<Sxis
z>!66<aI9BV9zX2y(%G#CqRQS?b1v~Xd7U43N9}ZrJh;>bJ6pd(QE=V7nKl02l#hx@
z`54}6w*w<R22I(eNwDm8ATqfba9*mlHQA2MhCSgn-l*ql@?&?%NXuzh3WvdA=yUhg
zk1qc==MOT~CL=8(Seu{Tv&6FV*gY~VG2riOtcy>)Hc2vHbo>|?TUP&JQ9>qccI07)
zyiYcK_ucUBX#h`PX)=!^5-zfZy!4S+Y%l!wa<;F3VZ=+pX!;*2Di!Us57eK8G}Nm_
z?XmbQYDng;5Bs!CB3|ho{qU7X^N=1+zMwr(lMbJ~e2A<z`CLG2cl||9Qd&B;<d~<J
zw5y|H+hBX?86|67j){Ww0?Izivm#`moUz8P$J8X8+tc0&H^cEjjU^9}Wu|MYsEJ8k
z6L`Cb2U2-0>EL7BA;)8?k)8(*a-l|Q0`vWHv%MC1z+0>T0<|`NewVMM_d!|5L(aW@
zV=v2z&Vr8BPx_e~5f0^dWNefP8RNKj1nDC1#umR>j{&w&mOj!KYP)&7(grYk)xoQf
zX{!Br*KcUCT_pEDUI*z|?J5OF^gcXt$DPa6sMe6M)a>OAVV)|tp%IB8Nk#lZP9g2}
zgI!<-5&qu_;_~JjkB`Dt+hSh7fx?|ieKf}o!0b|>urE#DDw8N|6_Zwk^M^Nu?)2&G
z4m5r#hOG5Eh#exhxWh-tDlZRP1PTM@bG()7<*C(Sm7JZ8rv327F5_Ee<P3^AV#rh?
zwI~H)8jB*L%?=`H|KU1c4kAUVS0`?4s4fVNDKTOkx~d_Fr(+V|u|D<%+s~gs`u5Q7
zIhr0}1io-K6-SMy)%!~G26ZiVD=;zJ<V+rZ4J%`D>&#L-2i4$^9SEVbi>~&DE8gla
z%r=EfCen$}x11(eBx{ldV#;A#etqk-V`881W;V6{!)*HAqGg*P8g;_#`7j99tWM{<
zry(d*4gsO?<-y{s6!V3ach%Wi?V7kQo5imKY?SGGc`q4gRT`!$vdhoyFTXn9uI7&J
z#A6s(l(*fK={(aa{=6l_=avfiwTh~lXU%3UvsO0grq&@tG8-%Sh0U4TA4SsR$}Ep&
z#TE#T#vK_bu{U=m_oaOiEVxcKjJ7Ms5ljTD?<@85Euc{I5zt90z=X{K=ZCKpBs%P)
z%G=S_^MFH~2HH@F>uU|GQYG5xwud#L&}=c_<)KEbWP#;6^aKxw4a&Bejh4U~-b9nO
z%Di~7j_Pu{lStYbn3nrMU%jR`>AAT5sh$r}9S^nq(TD~nj=NKni<pDObp#ZgO9b)y
zQ|^#!E@GHf5|+mS&*ipn9>sz5o^492n^ZDq`10K<Q>$(;KGW9skCvV!mwrqK-}}3d
zpnoQ^|7RrNSOm1+sk>9SU+MGNFDY$)Qds9l$j;8x-MDz_8!64UwCkuArHUvYeBY`H
zK~4-!+qO-ptZi_2;ewfw{o(9u(K;a(v-gBRo&k_Zg^ExEg&^<dsG?K;HFPZ%TJ(|)
zRDx^Z`98(7tWdexijz@hnY%n*wx!;3+A}#$je=1Y7Ho|gQm%HOv*t{^Zk^-XxrpkA
zj%5PYgK3LZE|)}?nwPnNbCYBgnMUf>)bQ&g+*%Yt8W<WZC<F}C=0dSN5*3E|!nl%B
z6hR2h_KZsLwTDd0Z8QlY=*g<YBYGi&Tmx{Sg5$F6I)-nZK0O#iXWOdt)IAxw^D!Mc
zqG$(hGSDL|-+i}5QxHA#Fxiy%l<SX#T8UO=BR#cN#0Sv#8m)RLUdas{9pF+rH#cOM
z;)EW(hJ(NfR`)_+?VWwIGa)z8yQTMi=>M_zmSJ&a+tzS`yGw$*OM<(*ySrN;1PJc#
z?!n#No#5^oG!Wd~-%a=Jd#wBP{r-Q?!;gY0s&=hC*IZM^9OJt+1@>OwQhmj3;z4Wl
zFm{n7#v_18_(>`Ym7uZYN`H%a3sg^K8dC3IF{XWlCE<<VcE@(bIqKxyQrR-eJ9+Aj
zL*{?o-e;bsdF?qUwQH#h^1yO%8Mit=1tW>upV$>FBH??Slfq*yWRz<)Nwl(CTiifL
z!gIH*S*0A++gC$@Gg#f`VEDuDdoKt=w=}!n4~ot%(rI5<ZayLM7C5dxEF60*)Cfnm
z`|e^BR=)%K87oJ4cvo@BjRF;ClZNDGyQh;_<A^R-GtSfPN&GE0#|7_o-K~`2a;$&K
zT)XL#wLc@c-;E;f6I$uiu=(|3@Qx(qQzLA6u07h=7M*J7SCr}0=(Ns3PSW#wK-*wk
z3kY)mTOGH8NUk6a3OU;;2FU>CSwESuY^=d>Zjd@*@(Nu*i?A0oni=g2;}N3_;W84p
zUHT(+ebRHAy{oBh?;~rB`FfAp$`>On=u<{WjDZEi$I{8M;%oar4gxtNu-X$6VHO&Q
zDMByTEKqQDI^>DtiRKw}EKW&+J$pQNo)ZY?ZsOjMaP?XPyMW~fnRa<V76cm{0yo9a
zycHj=k=-Z8j{sm$#ecS-WBIT$V(J~Rq^68PBVeDTAz+2X?Wk*hZ|=+y-k2BHb>XW}
z;e3QdrOFX?4U62zgWR5*N#I)yV=HjcTs@E}BY;$;3Q?zwj`zN~7Am&ey6^(s25Fy5
zR##Wm|JQmIQMVC9xMctq7=|g`cCa)P^3Mu;udNM%GZ$MWKTdt`-shhpC+I&!+lsHB
z7;O}&;}$Obr|=t?;rqLt6G~+ag7mQYI1xhG=(<lKhfyf$haH9BH#TRERyrMR-(^^j
z>A*kVKj1S)LXOewC-86wd6tc3u>iRKmcz%pPz=^?=Zm=+IvqDH6g4aK4T^;vFYbEx
zerwm_rZNOqIlahtG-My@`ZL_p$I@H$QfQ%7bv%`I%?wl)2g5}BpECq$*=GhT2^|%-
z7bFnsYLxB*Gd%3Xw6czOfOeyN`2I!E;u6DkxnAUJ%SZb=Ue)b=%^YOlh<*j;n)^j?
zokakHBOtlW1fa9yf=e~p5_G<uCLmF@<r}>p`}ty%Ew~AjN9bz${>vK0z3C`^PdDAW
ze+M_*VtH%!3r|grGZ<<6Ik{2c2xC>S(?;-i)E;#aInTmOswn+}Onw|l8lUQN-t^&C
zNBad{Sf2Bc3a~c5>tlt>dij;Fn95(@W=iqta{@FQQS;|`b@_<xeXU8n-wGB$RcWE&
zV&DDZy5v1+S^_t!m;e+uA+H6rw-<4ryG!?%i`&>dt5J(Ug<2@s?0N?Iw@e#d3^8vE
z&KKS}MB6`I`7jNlf|CbahUCt{3ykio4XuAhH$0hILo%ZQ!*JUQ-78y65Kk1^jqsCb
zg76mJhT*@wrx-5(K^tzKf^SYV^?}xrf|jR%`YY+;F1*i$gy@j}p|)X`#ac|8`V4w^
za0(EA5*6xs`$JI)q}~jIKw6I*pdfrSh!hTWW!aCQXikWcEvYw}a>lJdcX|5!E*>HL
zd*P=XxqD^5&+<}rG^DVmyo7JE{7h}Ow!~3)XZC6o1_Cv0@b{<m>bva<IU8<NL9gc&
zZ;oK+fA)tb2xxsSgsoFwdVG0|b);+#JMfg9F;s7PjVc|EXLS7h1`0e`t?GBXHfqK@
z@>q3U`S&e)+8pH*$8d8Gu3oHTx`UG<&Sp?^nADjql;WqzWz(_>GExgE6;Y0sYIE_r
z?L>X=9n5ahpK8y?I~)X~D;@o3hj$dzl74QCIJFrM#pMHodkR8+hh-L1oQe*;=!a#1
zG~24ZT>(N2CL+Rcw+^n9pe>C~bIh-&_N3$1oQ!I;ma(x)RgTp*Q~a~mtFX3VuE)*~
z#VX9+3$eo)0#a2bM;#{0?0$p3H;8#a8SMQK+as)N*klb)TeEzmby|_z-Io~1uwF@3
z=;#S&X=NF^g;c~<cfr_5u>UYe5O2+NJ#XO+JFzQ4v38X_RrhEaZqWl+-0}(yvP;xH
z(;p@Wzli7($2OZQEyAF7%tx0HB67{Y(@825R9A`&xf>%%Hb#=V{0=j#j07rJ*S6=y
z8ivO-w4?p<RsNiq@;qcwPKIf1l$Tb+`kY45K@0&<&@!kUQYtYfrge#ax;^=ikIeT1
zQ6a{ldqsrjF7Yz$+PUGq<y$4=roF9bKaohUYn?N7DN$ArkE&BO{ZQ~NQroC}sepQ9
zT8h7XVR4W;&0-Y)%kfZt!Etc*!g&HXqJ_7xQf`ve2BuHQHVCN=#(VX1@;lIONq-=M
z+(}Q+;p-oNe<FMUaqSN&KPOKs<L8JB2}^@Utw<F$6sO?Mhzvfj2n)BNuZFzZ9&FJi
zA^44-FVgEeFIP3EXXh;c9RI>I_s}XwO~QLB+Zi=9^*gGLt7y?L<sG2Y&@e6<oVGI-
z;hc^`)q^KL4AeXGt+hFt?VL*7%hb41HR{bQ3IR@5Nt=!bWcSBK)c~$-gXL@OA<{E0
z^i-fumE7sWdZ_DsW}X-F>k;TiSWHQ5HktOJ;D=Nd#J#xQ+hW~ZN-`bH-$~Tl^EmB7
z(W=*nKp<4gPIUVC%Jy=btrRz@Q$!l}%kH~)ud~o>7NV#wez=}+g_v4>f-I7ZeEa%=
zF{kGrKX;aG&P!dly=NcRd2chMWLH6<)khTH)Zrny?}=|H!c8D>NJD)6tZy;h@l8BK
z%3X7446|xjC*{LnZAjenr#aS?d9pxr=e@gvLYo)Do*ThWYcFh?esS*wfmw`wE3JQr
zn|jG2H9A<4(7t?YQya)4V=~={_nG$eNXw$G6nnON4-CmzpK5fiQszp#>{*qUL=A!V
zvSoAKvmAiH2F!~D(bSL0l3$z1_jc{n@Vs3^joqr)^C!ZPwU7J*=?l!`0Qh%uY1Yf$
z4FM}?a;tdi>GgIl^R(1J@oaCl4fJj$y)L|&86p_Xi%KO{Whe-#Pir4PLBUObX{+V_
zq!k=j>EKy^iW5x=-*(<1)%-H2prq@Rrf1um%BTQm;Ad_(z1y^rSES_k+&5+(|G}>p
z%c4Y;*6%A$46D`+EdTIPcBS!(TnMU9c#5(QseJ!E?h?GZD3@<xzk)8%sgz-5o`s>9
zvmVd)KL}f;nz^!lMS!w{jHa*<t@_Q@RG2>IJ{)+oHJ5Mo#)Y#uiG4WQnaZO!Nqtr>
z?va>nPkDX|wA<Hm8odMqlL8Er?pLvAiVR(+kKy-VHWTmCE9=WN#oTx?Qx_A$STA)m
z*41_qB;6Ndc|EVC!W(^-{rtdS^?RH&+Ry20r0p;;8ZCgt(!>Dz*lwb8ndGCt`3sGT
zljr$?{oyU;prKE*+gK~_A&OchpTWkDlOh*~Mc;sC#P(;yL6~V8O74e(3MSypLkSi2
zxI2@iL<@-N-$5mpL3SnG@RY0{QNA9er0mgQlnUpzcDuWKz_RsM2n0tCMBwF*Eie;I
zE`78sQ}746IiG$-`@k+@`=_4<e)swiI=1qV>U^bF{#yQ<V5R%_I((-p)MY#vMhGVC
z8JxPyk7AAs0l$IGgO7uk13Qxe#>RM@0`o5@n=sch^0x=|KHX{Vdo<Os0Q=MsEE`SG
zZHEG$yQGaE?;KrjQ4ox#rz|E2?9YX{rPr~#-z)x|j!<TfpWRJYJvgVwcUCg@3{Y2<
zHLE^`Hq!T6kT&W{PPfWXtPpp*I!y@(#6{+$<Mh_GPcy=pjp>RY;@lDOh@%s7=)<BE
z{Bj7>A(>*h*bBWYia8aqQL4KMUihLRiy3Y)@T7#{1CuwtJM}guO7JYHsz`3D%e`g?
z8l$7p0*Aqm+ourG^?P6=2g?$$iJb7?B7}AkA-3h%WE`hJUx+Zw@0>z5H92o`e-zvK
z@Iujjba^mzG(8=D(v<RcCCcf(=%b2j^6fdJ!Y)in{12stv;2AmHKev<5dUQ&1bTf7
z9E_hp3Oj8TIOSQfoT-^HZ#{S+Sa>Vgqpu<rE=*u<UJvg4;(B2mx(Rb^H}*p~Uwayq
zI6!Wi41V-HJ}jLm4BFNGBXfEztsUgUtY_IYZUDH@newLnFq>T?)RBQjBE2r*wnN9~
zGlpq{Vjzta`C)pgW}rd%F$;$Ke7!Ayv33{nOb!|1-)pVFaPY~`Xfy)n61_-t0a$q_
zTnZEgjmMpZYb}kZx--{3|KyBd7lSS>RSrAOXqBZ<O9yt{3uHS>{1A*l?zwxJbj-;+
z-=<pWAe!Gs7v#{uz8)?~rZyZ-!1bnozFREK2|UM<m{!-{?Dii9WMfan`!|UDFN+Bf
z!5-7S9voMm@RA=e7h2qtX7)lx&jTi-8QFV4ph?#zk<~N1+$rW-KKV&kypzxt9qZbh
zMWq%{-F=_f#?B$4Xn&`H=iWpRANK(>MIgEm4^!&qidI-ZOGI8cD1t|?hc=IbF%`Wp
z;V45}UEvm`yi35h{peFmA>x#7$TmNWq=#l2KYLT(-K9Auzj1b$Wn-Hi7-`yIo?W7G
zB#~*a^L4R3){-c-Ow$)`-Ra08ihT1UBnu^C5mXN$?xrnl@PLnO9nykj^Ux03qUA43
z3F2;@uLjL0y}nh%$mIBpuf$%DVk-tw|C1=pcd3|^OiUcX^y9^Y?u@WXg=mCxp9MSS
z@nX9Q*0PB(B()pog=q<`hi;``z=~bLEXOgG)k<~KY<aOt{9THQge2gIHP%ayRtiqQ
ziAmV}79@j7sddXsttS`lc8X+<Qksn=Ot6tLdScEn`MJk&;;B?BIq;7o50W&dCV%Fu
zo0vOR^2X%y$~=568|_mxJ?$tOHbs1`X#l#d^SSKuO$x%PG(!hY9+N*U=GYnj!wE=2
zUph;1>7qyKtGW{k!O=v?8AzSxs4L%B*leF4BZ?~2%@<7hzt2$kHA&p9=wC>VcM!6U
zpUvZ18nrBrNi=-hqv{8-XPMC`5L)b2Pb*cB;AoZsYI#8L+uT7p>H-y4`q-_0{mkTE
z&CXuLvJ2`9-$9|1&comiBMx4sO+~r=GA+0LvhmIqAjOAPM8i4%u3D#ofq}nM_7$2;
z%9_xN)|%D+MKNClCySLELvAJvv7VxHodLjHRCcRJ2dp+m{1P5$MlM>65M*bv6vo=#
z#&&IL5hbZw(0dxAp^R-~l}dB)6$v`Un(NMFQg-(HaGl@Q>_n?eS_zd|AO3@wWg<eL
z_Nyk#kqkp~Yh*A?Zh4s!EGB%<+6OFWSG2Y#_Yi)*sw+JfW+XhylzC>Ch6U#yijWOk
zi{MK9hsbRh8_d4Q^ZcJB)2>3=b^Z{L6L_&aCD6!{dkT-EU5p83X+EY9{1rf*yjMbR
zcrJ~}{9Ng^sLK=yum)g*VTeYiO^l2xG(t*~lB238pRXa1ryj6yem;e&de@TzWYp=u
z=(PkuZ8~N(m>c5MFUC4SdsG-O5gQHHLow<1zD`#<ldrgGCS%g;)KgXJoxoOAq0@e~
z+um1Rsdij@iM#zgGZsXjTt9+)NWc5p=!AwQ^K<Xm{)i_~Un=CUi9m21#*@93v^5Jl
zy9oN8mihMI!&@|BZvbt(H%{M+i0RMg(?$mYHe^ak>s%W(?Kv)HT7W)tl_eJ9{<)N<
znk|6@Z&gA%?=t}ov!5_VUrm%%ki8UrKhQ`^IF!%vs)vmh4qYa!rszwA4uE*uK`xi&
zroOo&^H^$nmEswcA<3=RyIiye+6BVj`5lNsPCw|0Hk-=xm-#7Rq>=!M7(u0z4`T0k
z@1l3!!sxw}rE(ZiKp2{OZ7c~A62}x@Eugf!z$8WNA}54_Bj78T3P2jO(?4|=3a`jf
zjzHWD8hDpM))9dj(@b9!!S<y+!>V5Fwe(&$8Ca?jFIbNww)D7Rl+Rytmda9!%jt9A
zUz^!>AVs*feVCQ&@rg}Q&W2y5V>x88#t`uXms2u&<6bkMwl*;_7Mh1WqP<&lgXjuI
z#q8JU3yxJ^#s|d$vZwKd#~79#PgDl`1Tjz)fO9f~fKQEsl+N-cv!dwPfQ2;r?DVg`
zxyEOcg5RId0u&?6^0h{N0>Pz8>BT=o#TKNyFP1Ncd42_PXfMK5E|gm>zaLWx39CGI
z4_4kyue95tZ?0P&R=osA86uc8mY_~--jPhCJNU(!YA)+}tmk?B1H6Wk#X@e}APK!@
z2!rA9tW!U$$og7&gKSE9GSIsJKiae2L_`yo_F!fAF>Oj{&vMczny2f1Oq@k(jIjq$
z(Y?Oa1Dp4{!xZW`{&hyO7=0>4+I*7&T3vzP$+$J*U02(oKw!B&9<3%6C=}Z6U!lR3
zQeGL~+s_}_?vMM~YhTv&ms(DcMjvfi!reZ<<G!h7Avr&f_nZa-mfW#)tcS|e6zgZu
zZ3a50{_ThgZOQz)7prP6efg;pSJRNNuu2U`e}eV0SL>MLG-RMBJrq51`biHryGoZI
zHXZQugSpPuVB8Fd%g<$_6<eD&N}uKi78mEn97!;LFmSVOIj#0|(T-B7f;&rw-QI^`
zF#})semmw>B74sH!L3tV3wl%zWjJ*jE3@f|s9Go6eK$P!4iag_Rlw5~y|+<Umg{h-
zBzd>!A$IN5waR5eCzE4)BQzOa3)1}cJB5I`#_}U#ipd8=atMc-lgOpIIJXogM*nL(
zW*F+YqIhb<WPb(7*KwH}q()ofc$sf(+v#Igms2I8m^D~{>TZGGEt^K2OIbe6$B#M^
zjt5iGbXpTPA)To$j|Z-Gb(YgHL+VhhSu?PR4Rkzm`-`sWXHPT+Cpg#dsJnEXwrTHO
z?v-4NAGFd+X>!)babp2?Kk)3a>-c!%XV$*O_C{^%bjyn>$3w0(TOz9<6RWEuX#JE`
zu1}+&+yCCmTbBlN;l`q5*$;kx<PhYzaPT6j){De&(@w&$CY|%3-w$e~yAf58YG_wb
zKi3_W=29I<r!i=TwG|$3N1(s@;f>=g6%QGXD_}I)%G;{O28Rj{96*O;NXk?(BD`NJ
zkklVV0)=;%x*UH<bZsV9Dx(&g-HXXYe|IGb*=ee?RGE)cm(Xf<^Pghek~`nuX+4Ba
z>{r%w)1uKvyHYLJDHvmEv0keu1;a2=irS!lwTO65Wj5Jst*oRkWf<BN#=Y()k?`@u
zkdt@R=VT&Lw-ZCz-m6F5MF#58AF4dkx{DmX%cazCq(!7LyJYN6kVv7ZCMnTh8wK4a
zq5`1YqP=R0Z23{7{lzf~tSY81&KBXRc0=~Kc;b>oP1Y!XG@N=vwG**>Xa7Uw96>|&
z3~>D8+6knaYt?!wf_NB2l4~vl3)+MCy?pt`Vn6`eD3ZD6rEjm}=CrN*oJfr==z5K7
zUnV%Yv>L9tO3-ezt9NfZhx0w@wfZ_MoCK*@qT5g(H+zzx(-WBfY2g<Cs>7}^A3$>T
zD+XSFVscKGs$|<1CQO>B0qQxEE8p>=e;gDId+IjEK_E2FZZ^+2xF0ne2)mEtAaZ{g
z`K3^QvLH>H!a6xqqFInYc4U^_`q{S76xYT=@(M=R&}I8uBF!ehPRrM{nGK7OAx&RA
zQPr2Sb9&bWyCLH<(%G4BO}yUO9lH$1QY`{lf5iEJ;CtL6_{Jv<d}X=rwVuKs-IO)F
zHYd9)vF?%&mrkbkw#sp^o;{d^9gnDSLs((In#vR0Ls!TzQv#G4b)UQ$Af34|Tz`wS
zGAW=cGdH_7979c$_XLol`9l**r{IRiy&^N%Xt%eaBHQz~_lIVSN#&+RuJTI9JMO*d
zbQnDGm~U?K_Q|HMuu;<|w`_M&(lIpypdJp7@g2<}1VO+;g1i>U0`5S(MIqx~Clbth
zO+GC^VyJ#rx7Ab<debX3*d1J*w{3T$qF*aD(j*!B!GMrI3BCZXY8zzltV`aW&#5@+
z$&{m46)t<31IsgOBpcKckLX7iL4D~Xnaz@F5S5llWuZco08$%Mb{Nz6bmF<0lkUSY
zzu$!3k7?={6|13KNTGo!7^*7q;rlEl5x-@%N;+Q@gKg(lc8(=`Dr@QaFvR?lZdh>5
z?JxJ14b1D((t_qY`s%meSNEwyPU+f~Tnaw)&1uLTvx8h)uew@8JkQ!mN!2FLLr$Z!
zn9p=F4O=WBCx-c+99AH!Rf-1o_i<P3W$?1qh#6de0~dJng-<bG#%lHq>+C8l!)z~P
zcHg3Tp62qi!$rDo>NwX+vGq$VxpXFw{Z72Bx+V98dk&V4F^+eTl|>8FJsVZUQ>D-r
zsEqDk9>0*KY-0+G2*YNniSRIevu;t=ZB#_NsAdjGkMi|Lp7-A+CR=ir7tFq$^)Bi(
znEC!aZ@#FEJqsAR{KH;;fAE;hH!j~D>+i&KH{Z}L2@d&)l>KEe)_*7#tN>Iv1ssF-
zNteL^4^Lb<H9%~YW+4YW5N?nwL8uCm`xuStTF3TTuAB++V;IT(y7QqH2qMPyo9!K@
z2(Fy!0@<0j-Z6aZyJxYN2R5O*-pG#fT2`b~iew=OcoNy$CC~MZkHCxp3VuSV<jyrv
zg~e+zYMAcftjPa$TNNp1l?S7+s)KBCe?i|dA4mKar2Tb95x?28Z8;DbpQe`0g)Gqp
z+6Ji(<=Zlv<4yH4)CsCvVK_Lb8=D1=gyE*3KQ?0V%|_1<e~+j<b|}QE$B5Q$r+(3K
zi!eowx>spRLD?O187@()^RBC7VcI*Xakv`1&{dU}h`1UyA4Jts*@w@Ypg?LTXAjow
za3nOW`fTjb>4Epd>3E`@Svj~w%5*%ffh|z~=7)HfKs3wUlPoTRZnfSdL+Dd6Y_om)
zps{x^eT`=Uv*nAjBeiye^{kW|;d>%&#5PxgCl*J-GR+Qo#YomBUFWM$Dc5I4((M)G
zD4TWGpw8UlqLGRk8XWNqws)#%mc~^G_%vJc%g5Xgtr~h+{n0y~rxVDVAJ`IT1RFJt
zDn8uZ-9d~qv_4NphGi0DUGF5N98wB>TK;;dK>k{p`Pbnt(geOp(XG#fp6;o1q0~Kg
zn&MW3K?C}N@4^>)Fkg(Ld&kE~73)k0XU=rg&5&t@FeX<TBYyU6hV-70h1$a4V_qaf
zQsIoB;Y|X>QE$MkP#fL0vD{D3R26zD@jb)=|KoMa@{Nx$t80ppIS+E$)Y4^qnN7Az
zOu2Ab6z`n`d(%1MCqS)4v}WgtQ2FNv66KP<X@K6X)~M+Se}vNgCmcw-%<G+Q^`)sM
zsK$-xhD<d1R(;tuopiZRt(#J!>f5vRdXAA!R0(R9C}((HLz|<{S&(AFILA76OUh0s
zk*C)z40|oxRH^-<1;?R3?S;14liB%B8Jjz5MYp9Z<92$CJ@YGn)x=>@u%EALPKUcC
zCH~F=0@msB@J3j2Z<Wv&=p1lifYIC<zTbwv@d>dNOV)h2j(bPtJofEm*E6)w66U5$
zUnAY=5P1|+c)A@Chf4qu@0d4guviSUP^u^%yb#K^OpIt=sirVJ^U0uO64JH08?3F^
z(<YQ+V7^}HT19)f+$Ss3)@~Pe^4X}Ht5DTUD{E^1a)tF&XVOLn5ZqW+CSv?&>jl;O
z7;%^=w8Ioe7Graf*0nAQD8^5AGimb`TyITP(38RffA2RFB&Mhee!gM046A=mms!T7
z`ibr`hUEL6%M1u1!8gF!>LY%-Yhq*mHoz;#?>BFp_5IsgqSyv&zS|{uIIcS{v(1ti
zN0QuE83cC@cQRb=uL>Z2I;0k}r6N0H+L_`fgX39hac8%?kQ78e++IF@BpiFgLCih<
zc~r-y>981TvP8{mt6fxl0}kX_oM3@B(O>z0sO!95P}ZeqB6*U8Sgms<@Kb7P>?k-n
zHP5L&nyr882G=rAIa{I0!js7yOIyNb`Qq!-g9^^NnC5tyBU|7Ou*~5z9aQ;1?M>t>
z+ts+)3_ESavyQeBo#Lox8Z^s~VxL#YIz{NXUuf4DbDdod*V?WqBzE&8IcS8vKP1c&
zz92KvK}knDC{014Z_B^TznLu1oy>L2PeI4!t(`4~>Mr$`fq?z$@r_wjv0H~3&)J^q
z>4PDbzh9DPZky8?!;0<vlgS2?j@z;LE7~EYoh!PsC%R4c(dInxa*6uIl;tow*RXjL
z2IG{%vX<o6MqRl&d&Z?e>=w-AFi^7?FP^nzHvHOJL`$1=(ME;{{cjywq*F+c6+8tY
zjJ@LRN@|0;;M&Wd?CdWy=~gLE5-iHb-u?q8A4p<Qu+%Yytdhsm8G`C*qsG6S46*ew
z>IUy({?3#K<z3F|m<h8zJb7v35qJM`8G5q2NG>XwPV?pBeD~me4cul3-KO)$Q0j%v
zO@qB?$bgA?M$fFD<TBHyx+A&$A`zJb15ia;C%oo+eK)fMVv8-#{OM=#5o@UW1nVbY
zBN-wO4<S`Nzh8VLawn#6WY(<x2p)Q=Gs^l<hRll!%xIVO$}t|7R{$cmlr$D=F3?aC
z?}p3!p5l+6&)%0uOy*qWboU+$=rh)s`1%hJNByLRPUSKO$ty5Tnw4i@)9T)$%%Bk8
z8vzM)aj>*JzeEYgo|_-8r#^`!r8b}3L^p9zcX--bh&H+V_R(eQiRtEs_jM2|<qHD(
z`4aZYRNzTR-r&u4k6{jeX0&1YN6qff$i4|Dh_3SH3z_LqhS8#dkvMsjiUoV;IE?#m
z6M-x_ScbI@Rrj%2GD~76)}X5s5HRBZ9`Hl_Xt_T7d^2<_Pg0eH@U1KPM+f(TPbb(M
zYAgPt73`(`IfkSDXf!fIXjy#6_&h>xlhV0cacGoaqnu<f7;%H}O$P|u3Ml8FZ&Q-j
zUJM3)!g}B$H17~%asJY`SSv@bq6iXJ!{IU7w2XD=mE+?5&uvfbU5+TCaD_owd;9>$
zQN^g~R6TJ<gG|uvPmxXrGhT4BEs;+Ct38D%4HbLX2X>Vy%;uc{4vZi8yB&le7jFsI
z>oEN$PZd?Y@yR<?EZ%y9=T}1+2E$@47Bv(J&m8Y~OxIwsq2ZEL%ChXP>5=VtmVrbd
z6^Bm^+9_&SnetUounsyM>xzl;;nu!9nvG8=(;^P0A#1&B-$KsOPW=^N+PW`4JMID?
zBJ;A{hE!F%n4xaD0r3ChGlrq}>lw683U*3bIA9I-qVd!dtT^S=+VyxO#g6k9mL`S8
ze7D}&YPnsqf&Q9lBF_$z(n@6fnDt;zo`KbhMl?2FXrn3iR@og0yQRLYB(?&xaO3Sj
zo*J0p&Xg=5DCjT@FqZBT6K3zuJ;>C2(P4B|-!&nm@lyw-AAiS|nE@qO6^22tU6f4O
zkxtfn9YHp(;M2=ZjE`qsw2<X?y?VM`ds8H;dx1m>9P`3%3x<*0_XNFtDzyd0n)`6=
zHN3QHwP6W#FVm$RKy!n{gZngAz1<w*CG}{nqJhkHQ*wL)<}B~#u<0Vj4TFaJ<Fl}I
zdLEc7<=#q7YM#N?e|C^i8{5r;lRV!j+c?51+u#XUc1RXn@eXw~|Ek0-{R)Mo&$$3+
z4tr;MJ?I1TOgpGfSRU277JBX9LHn_u^{C=Q*<&GTeL0dvCk^VlY2Z0+l@=r^u*fRV
zOvSo^&o+-aXu5d}pYCn2pSWDTvpj$)P#3*hZKjKYHK5B2I#Z~xjdc<9cv!Bxb{rFA
zz8AtA)4!GBktS<7XPp_1h>!<^KJ@;6J=BPBOpb<2$|FJ}%I2{j``x=-G~j&pkmP%I
zE?T2Qg#Cj8>V+^WHP&+800ED~N~J<V%eaP1mBAymMUFM@?ag57sh}vAg!{vqAb`%Y
zp7r%##qsfTS&HpjCA*z}&cEIrcqnoxG09&{VG31B$FV%F^QE?hg1>&ZPJjKpHBVB3
zJXUTxR%#R6(bzPc0?o|rDr7qcad{$H(f(LkGn!`2x{bTTetvK^kKe)YQ1>SA?frdd
zk&G*(z70Xg<Ej5`9!UEshr8h<(!Ykkoh8sDHtHgSFSL0FsR<Gnt#{92%{CIXw;sJZ
zWVcA&Ft%mlp(A{BvHO*$KGPE1)aH438NCsq&A65jY})TPVJbY@03xn+dN8#KbRelN
z*Oi?NqULZg6O;{iPwMs~AmUy7u?*HBdv&nvo9=o4QSf$gShYf)ax$?fjm=qPs9340
z_42t7+988K<bI@<aEf4E^fisnNe#~Xo#*0~)f*TWQ2hF}epdKAa2GH>D;07co-rEP
zv@U(T!@BotquC#2xlC*B6UsREvgBnIW|wHqI@gh32f!vD!9#;D=(QS!-V+H4fFhe1
zbbUL4@81OHRa~B<9l8I7xYv4JTWjC^l{=YEr&S7=2SvqW3EwxX5#f06S@Dp#u<Tb<
z^=8}u1=}>GEY;)dqoI<tpTh^{YqHb6msQr$>qUEim~HdfDfSs4Nk%C1W<9|iedx+p
z>&=pZrRIHHZ(}-m*h7GY72&bRmws?!ur*7~X1PwP`Z8!U1I%;>2H<_U!JX;|4cQzn
zPCx5FMY#Lkiu>`Jit%Rty#gD$=u#Hlzn4RAVZ0AsXtO{z!hI0id3a9@&G(g4{U>0D
zhZOr=iD44E$I*J^cbf#hW)SQIw}~`K6&G>2eqgs;3J&?EGP`4np<uI#qY1f64YRQh
z@YJmHgNVjNOqbZW^YK1%sltGF3&OWqz5!qr2?BzbYk{{-)E&A{4aVfONW2M_lciZC
z^}Ev-CLuvk^CJldxKwTtr5<GQsfnhAnD6|`OhMtaZfW0+KMT#+RO$Z@YY)M?I>_6%
zzwm{eK|3U_@@KAorRKhK&SxAz0r7i+L^yCUEmUo&&m~D`a7ZDs3JMF$90-CE_y3k0
zFHDp2fq5Eng}vLi4@iUffxh%|ZE@VMD5S$Sf$`@%N5(V#!NyNcPHe^|CKknN`?5us
zKS#`G8XO%-%`YDBH5Kg-+)h`qzuGKaV1`EKw+cJ{aOhCi9XSIalnMY4IIsFc+GpY{
z|MH6wtW=|gVoZY#N%^1%NF<el)ReO4CJz8#$Mmx#a`(gm+w9j(N0L9B{LP?lak|I3
zJQh=BhwwO*%Z93e;8n+uq6Zb-oBhRjmNHF|jz^Y^c4Pyh=@0Msn_uS|$GM-_l<8aX
z_Zckb>lfI%)BE)!QF%vh!_-Kwd_R9T!*UVk#zcTV?>r@ZMf<{afXF45R9Maxidd>V
z({(q|<N+{)WT1XhMmMMBh?eE3<8%!j8L&197xW)+5HwWptxv;#m@prmyye89`~~Nl
zi0wCHZP8Y)F+z|D1vyqQ#b^B%HrLem>|JCdE>}Xo`<>D^f`Y(~e@oy}R}&9CRu9rI
zHI=&P%obyjFH+~d(DkP7eTB?=0hEYVOV!%@RC^21Sd12ID?h4mKdn>+u6mp>T2OQc
zhPjv}nG~-5qR$UcZJ74*XnTeB*ssf>;Bwlgkab)Ufge1|?LGEN23#{2lfEmzU?vn8
z$*{zx+5`c61SqjEJiZWD!M#RR6fhtWF0d3K>~{z62noB0+O?gvt9nxEUN~B<Rv!`2
zfF2h*$eXs)cP_YONltA80|S}P9rSZ05*hB75nFp+tj@f*X3fG`{?>-kg90#`1~6^)
zII<JSKXo5lzH<cAlfMoeEzSZl5^&rxxZX1y<yfa_d>!>(uG)d*`S#Ic#r|MQo}-++
zU;+?!1(>QY;-Zs<H}v%dTB9iZLG-=DXtAFLpf<lN8&)9U6dNBV1@6rhG1@<O3v-Yo
zVAP&=r0y%D26OYCe*dWX%lvxib|kt@^i+Q+dQEe&%)r0^B>A~TaY>V-X5Dm`8#K?C
z{j`nU_1Ncf*mrk#N=dphr~<`Q{?edG_dichdX&a?hu|%7A8t`FsAyj|gv?66B{VvO
zUX)b732y1^y*MN_t4Cod+Us9pN4a0P$I+(dC6A?(vb1D|(&^bcPNPJaw-hqShbho)
zuaVw9>wFq#e1ixdY0ckB4OQ8ohap*6hpcaU*Mz$a5Zy|cB7DNJI{#j}wS^r=>fzR)
zKdo|RwTA!@w5G<bw(WSG8cm8>6g-b_eD8-FEQaoz=wxT#<gT5#^mxpRd(D4d*^yu!
zV|ZcFJ+!LjKwrt9ehY}yLDD$u69p_9IfgR58<1gH7MdN8glcwZDocj*rEBrC?^JJj
zM<2CdZnBQ`_TyOh(p!$Bs{6QkV(<X10VmAcv;-PW;e)wh{Po$Z6=&@h6=}z>f$dgE
zLBVu_ZcG+xYa*lBl)s|z57WBk5z&PiIfrwcLJR@mpu)G2gkSPsX=EW@qlSO^0ZyP8
zM;cNT-gJZk%utl<hc{;>`@1aNcESgm_NvJXY<?(fAqij@sV1fq$?l6&B5g}K!UUl=
z%6POidGL2yx>mOrw{8B7MD^u{>DUhBAyTjAT*<H3U+x}&MN_HA>GGI+`JJTwcp(cu
zgJ~|yoqDOtl0J>YPUc)qzD7iu!{v~q?7_4K01)|0S^6bv)_)*e4GIlqzKl<`ghK|Y
zihT=9_n7qNu~x0IiNEv@N^~_+uN#RHCoMX+?cOt`GL3A>r_T?j{k$f-If4&!0hX4S
z6>aC9Jvty#;osptCmXz0+BUu)E5uACf0qbz+vQ2&btS{)Z1nedw&vib7eM{Ny!+K+
zvNd=mi;XF5pJahL&>V*^&0^qB$;c?)f|c9o65W+X5&1&#gT2seKvB~KQ_qN@VEIpE
z&JRyCu2D^-RU~BH1`rD1XFY-`^3;gJ4Z>V7o163U)9pwP<P(t$oQ@9(YGnn4z4E0;
zqTHJPe7ZPN{DTX?Q{jciqmZ0yt(sQ5U4-Mp6NuNX8dd19X3C76R36sm(q``RzbZ^-
z5sU71_YO-i?~Uk-(&;o4X*;$wfdZW94M~hJ-+3+;ew*c7@9OT>#?bi|9Ogp}TlY%g
z25=LmdR^+MS0g*wU%igF-p_`bU8OkK@1Dv7!qGLBT+alW>OsUPeyJ?iKFByYMrQLs
z1;#?VP>;n3Lq|wRNG@%OdGFf8JxEYbt-8OTOXwBC3=o_z!*O;Qgs!$VO9VD|0C^EW
zi7*sR$>UeU$>Gq-<KW(QkW@YCviQF=Hm?Trzdsb-fIZUXdGTH25n|AfRW(H9oHlYG
z!~$rh&fs&#z?uiU-9v92!0CLIvVA7V=mmGJMKbj*fQ>HKqW3d6@RMUCEdPx*pHCiT
zeWIyf_FF>0Aw{b~<?Tji7x8O9jIu(oMFs8sdwmvC1oP!~X;_V`c^68q0YA*X_xo08
z8v+nN<}@X0K3nw`*9BA~9BE*c-)&y**(q=|OU038peph{r4R`RY1VJWDeZ;5{Ko9t
z+1-r`hk+kajV@g5&oKZH4ZiV?k?~ri)=wwx#KXE@nUcggeqR8Ld)8V_R?s>0AtJTk
zT7^iu3nfwt{*HP0*Glbcy|z;sWZn_x1l>QKR82{~=_I!ND`j296oKEZ+%h|1wO(v`
z2s#f|N!SMh3!%nWZbJi*)R$%bcU8*6`BPUuyejX`?=KGx8f7FUS+y>Na{{hNyoCWK
z=QIun6&Uzx_Z=M#>TbWwgJtbUC;Fu4mB#pM+^AkGX{`AIg*+kvf%y2no%;b1*|k-#
zM6IgTUgah5VH>ECKEE<+!#XN%FZKz*8~vW`thb?5{L^jx-~0Lcz<ZM{s?|04REHmJ
z<x(ZY{A-D50V+9{@^1}@PEDw-?QO|bawnen=Zv<8H<j5O*gVfx^p~rguuw!W5{da<
zUOq$=b9&*pIl+^ylk^`gn8^Cdt1^35E&2H03JZaQ2}MyU@1<$ipE#!5Ihxokue5x2
zyg6QZab6ovW4HS?lDt-G$I3bG9s&K?Y`WMVcgfGLZ=i<0ovdwSh5&2*v~rLn`r9{}
zW$AiYT6pZ4h7+IwfH4|Wy6BHdd)FmJI;G8G^@E652HU9mvS$bwz}>2A;Hms{m0<^<
zb<wHk2Vm*j_U71O3lQJ$(lnp6xu%(cGSnK}9Jf`wZBsd7!$K)313GIVO<(Tw4uxXn
zxl_#rLXfku-hvgk@cA@}0>cr_FKC~i|5IX!ZyV-N_n9}fE&CYSFt;4KGLyjT4ah8@
zmqpCsU`4)0|0I8ZO_$OPI6amx^GE4cJb>V4BoJoq<KWUsJH|E%gvyX-Plu3@lMi-2
zx8<pfVol{d*k(@$noVYgsnDhjCg|1=0_y9QAdLIs20k{K>e4llGR!e1eH*$WKp`3^
z<<!nnOJTr)&^-F9RfK?iy!r9qm==AQ6f)NKQ`3N?vb<7Qe+b|=Ki;o<p=^__hgTf&
z9EjIGA6-B&eaO%>jtu$vD(<9@cG~<B0Cbg4Si&ty`gZLTc=8&L4?W&~f_3Ae7J<V*
zp2qxhv8gz+O2cbng`L~Um>+CqwfR;RM&5gjJhq~liWB?81xfuM-#YliKc?OvnyQ2>
zk9VZi=*(S@d9)HE6?Z=S27a0jxH|A_9{yOjT;rGlA5$2ef>oPA-*K-}ZTE$Qhi*&?
zc1j$J8SdK{@oq#u12wf~yLO`x1pMJomc=T=hFtu?D5Kb?Z1w3ro)`FEzkW%G^dT>+
z_|kvV3*X#dkr~$4zgY#XpKq?SOb|TDax9uIQrcm7IRHQ2TYji`k#0M+ZS}0%(K&i3
zU3GJ5boV#zddVcTmw*UGiMs64kGXRmCkhpYmaqc!ruHc^-+MgzQWxid?))Z<2CUmZ
z+k~7mNHXWVM67<gjhX5y1senKtlS+40A--IMOy7bCS`%E{zyqVN!}=s%4EJdD1n^Y
z)YIQDd9+jq!h)`fKM@53Mx@mlyGGg#sh<2@G9(iD;sxzh(Z^fBn+=}lXWzg$=)Zj8
z?-#&F3L90I9Yg;nQ1mzTvP>*MzZAJ!SXKtgwK^IpLIM?d4*~}8%XG27(bLOp1~lc6
zyV2*FZi|A!)cyE`fqxD9^0p0=fmtk0*GL9X{`)VAe*}eq(jC7#2sJaQ2mFOzbm^@(
z%M2hkbP3CL!{=V`x;OrLz5eM}zB7<V$~LFekk#(GLbJ17Uc9#$#*KY4WTCl@42MFB
zz^7)0z>=i6;w4iT-wGSu{>A&e0YN82LjCDnnvtO^r{BND*A5cVW=HGRP#w@Y324yS
zqJ0%73m2A-9TmXmP<UNgGjy>JH*BLna&EZ(oBpgH84~7>GgxYBD)tgWEli7-;fYpg
zCB7FzNg1D89ruMHm+h!Yv=ttQ1xskFTl5b=N|5!$nG7jht=6!|*bxk=5bbT+Gx6Nx
z7^V%8<&|Sc_nT#j_pp$!>T`d-qM5}ofN_Y+Sf?B9Y>62d2q<%hQ*m3@&1cX1`0%lT
zLshbgNV~r&deQEKT|4Cwk@zb5##;jELzVt-Ci53>3zfs?)(W9)Bx#}&k(-Z}M9+GT
z^D(S7?`z()Z($q8cpZfUCKdrw^&3S$KPh3szqq!4Ty&<9Hyf&4bkh1=x@%uFF)6Ew
z<pMa4T^oc8Ed`XHBA}G@*<uRwuWtJv_MqhIotsON`l7=~X=Gxo>>3#YB4^1wUkTF&
zfd<x13y2Xm3h(`Q+XF7CxX#N&Moj#gYn}I7x_Fc@_#>0qXMYV#dDm)Hr8h3jTz=%N
z(#%}8GJhuK_m8ji`iO$wyQc??^23WG3F9))?fB?u>`WY5u<=qzWz5bFW{?bQb~1Hs
zE(v+ak|GiXB5-;P3C8pMsrK1F$q{BQ!IF@YrR-4|NJaUZ?!&-?h{;Gq_4t7WM@Et_
zY*Kg%FclK9vZ_|w*~4JS>iPfnDNyBX5=viT#et=aN;3OX(Xv05@wW{Yr_G0d+n=l#
zLc{1U{8gqL9_q6X$j~abD;TPnR8V3rNj^BViJ-O<U_ek6<ZXYhPy5gB$OB?)q7gL{
zXM_rtnT+{mgj)9e{5&s{j~qEGPMD8Txu>T`OiQbD5Cp|khs?|0Ty;o5Y}pWy7TtVz
zx`6%{cKDA~fl}rJeY*+N+r)#jj`=wOe?#KEAch5J_&2`lj~hlL53Gz>D^(aE@b>3B
z>5o_b_AHSE@baK$95{&o+sps|cmJ2=`^O6Ze_8iT3)r3>rK2;(`!~Q=0TC<w$L;^?
zN9zk<P!((}9b!LH|Iar#Qv$Xpsw-PUw?ogDypt#i#^Emyihp_n$M?+{dcPNDRtt(A
zau#NQ{@-s3C?(^!WY9%Js`LvB(3Md60=R$Y<6bwXD7Bse^?;sjty6Rh{#%S_*H7=>
z|NZ~*p`!(bZUgE{tyXym$t4<8$ngKT4gcj)W;1Z&$**{VuzCLd&hV~;?nyaB|2Tq(
z!ZI$l9|SCUGZtx>@li88sa2UvDnfy&n2wIeEH)eQZ`zH2ecu3kkw7A71Vq)T+9}&Z
zB6*7gd7okztG_;pT%5y~zvDJ24C66Oay`a!`}j1?9y6C!Cfl}5(2W5Ed|TI?FLVu0
z$5`eo#r}`2wl%E}U2w^6XU<cV{B2O!wBP^UDF~pR8dX4*(Uujb1JMYw%P3Q$!Tkt3
zzi7b~mJ`sLloLN#kYVZ+@cnR^x16u#*i?oHuyrc+iG|?`TkE8|+sxGl=jP`Q@Z#S^
z{%{&KGM{WUAIN#!=vS`ZeK00$xUqIpi2S*oi?`BRqW1So5h2sb1rZ*8#RbdVjIcmG
z1-RXyPU<aOt*k|=S~CKFj8UZx2t_w4hTa)V7T6Oh%t$VFWjxPan|gPWXH?9c<mA3&
zA`((|@$#l~1LvKm_vbgBJo+F)bc{R8Jh@pX3r&(LkF_deZ0>SmM`<Fuv}z<+3;ch-
zOx=V?m<P@XSP|{J#0;+&Re4}g@IW!5wZV%XgrE#cRT~Z=zYvB_Hti3gCe`%(<qTnD
z!uZ5oh-i-~jXm%jhEWYb7b}nX*@=Wg=yIKJ!G3vJJB~_WRuIJcDj9obaBFT1Td7uK
zZ7AkG4fVIamO`2@!;WK|U(#~;x|kX`@cpzoN9D3e%Y4Zhg)2AASzzJYC-(R*x4{&g
zOsh}J5$8l3556A$pv7jfrc`X8Rw<8udO*BC%uy!|fIp_^Em!+=uv}+ODxJb~Y5x7c
zMRvbOrAU+@V5&kx@f0t%z@B;TLwE73Mfmi|1eUwdr1^6`2IUDqjn28R8_$iUxmt(L
zK%mC$6N7rq^-kuIYQ@<Mc_fa2m#`cm)!&8y)!<O&g@@E~RsH0h*#>Qr+$k~-S;C;C
z5>zaovp-WIpYXw)+{awgM}F#eIHpjQNk>zde>q7HG+Kyy%iyG!5w(;!L&5zs2$~EU
zuTcJnM<60Toi*y!KY6o153~7-fG-#)f#v}Jdaot-rJkz3KJ}r9=JTV=oQ)}LNPWy+
zZaqk=k3QEakfDVZjS>_P@Ht%e<V*K>o@L0;C`(};YDMJy^EdxTxWLDZ8H5}h8XlkY
zu7<Bj1<-dD+TH%jppM4MMgE4kOt57$MabSADB9>nBY)^?ZFNi5XRw3Anh>>^FMnUT
z(qS-A<SwlH-y+!G9ZzN}xPi2uuoaCr`t}h{(Nn+3erNGPqe#Zb5$-F|^dyV?c8VdT
z4Xes4f5qT1JoXtCOU{gY<a8I)3^CXFLttX8020n}cIoPTu(T;AF$2Fc9OiGGWF`X`
zQza1{UBr6GP-kSOVN4g14_yFbT(-G@)I?7#b`PhdzoQ9k*}OTQ)#u(=#k7@At|xr(
zj)GvcGsVh^9*gzddO`rKIksGTDpzM^>FshW_1EJiVtnb0ZEc5{VZ(m7L=kvwoU7-<
zpnojspFfhP&_hw)RU*(xLLh;uIFbK`@)kqXTV=Jy?6V()441+?rVU)B4IEA158nu2
zPG~S!uVj`(Ty*ch!!%t*xccB7UuJ<C&jb-3*9XAw57KuWM*;HZHzsk(@uaZQUcK?L
z<V{B*nrvUTf@sw11(<vTMM80<KL#eexSuyA8{`6~mni4}&)=AxP$-{N*lrI#mYDvt
zZX(QD&Ndi(<Yy=#$3*D;vMUHm^Xljr{^xMGfilNkS~3Q*W7Th7aPYOK<qh!Erx^fu
z&%kX}4oC!>`!Afq(b20So2A@nqhzlTppMeCNR>dZODZld{>5#6jOUZ2#(N&CC=wE!
zmB=axM>8+*KhstIBcK^lB><YY++I5ydG0QIbHR5CMz(7RH(a|k7W*AaKRlj~zFk$<
zJ6`|9{mp%O=eRP2Yx+a`VSUJY`~JSTG`kCCG?h92XA3xl=!oX2b|<|sP>-n=T&XfF
zfUlf*+8NO#czU>uxxGDoR6Aa5kp=wKrvjWJ8qA14Q<VRwbolj_BjJLGvNj0Vf8#Vo
z7V#DgoRYo{?#bl-wf#$6OH1CUm}ROSejlh*)gM(p&^+nqS*@OjUY^WTo8gNp-W_$5
zo?pAZaHeQ4LjJAq1$Gp^_h~N7>0V6yXQ*9!{H(a45seM*_7&G_2>|5LR`O%J_0cUj
z5;kmmFK$jprBsX97veJ7aW#X^7xFsZL>jB*0r#N$)riPUs8Igi+X7#;pNh|`w+be}
z_GK?)Bw>$nC$6c#7U9U~av+OG(T~j?d|Bles9)uGpq#fI7wZc4DLFo#7CP8_k=6yr
z^BdRG<1oY1^x_v1M#;W!A8B@<MKznuFI4uN6>f4?(w_cHRP@K)NBRt$dY>t&Y`UNW
z*p`q>wLAFnoWJ;*edvulpl?_Key=)Rsrhsnl*Pa?@)mB)=|VangxmE-{K@6&8&$^5
z<s}?ocIQRY@iJ*k%gyRFJA-$qeG(QH5`3vSWdlP)F@Swaq^i(H)Wcp2{_m&$b==g6
z4zO+K6@CLxgX!V4^oI~IgLc?QJb7ZC5b$MYiV^^&#d9_&HcP|UdAE&hl;sVlu|Hd8
zxc2lixgLr|S@b-s5T#4ta4=KvjRbms4YS}ugKh~^pk}^SK3Zk}z5Q>t>Tix+4T#RE
z?jZH=3xpMYq?(S;qd=*@>01&6KuEKxe`p1zWEv0v{QxXe@u%|TA>vMhwRX_@7rnO_
znhn+$RwXCcZqaG6h6RweK8zk3bus`_ck5e%`#dF2yJoEvSp@!1McpL{K>gN@>&edi
zQ?)99y>F@gW%Ya6>c1~hrjqx9g62f#4=|?NuE7sk$<zX)8pu|Bz2GtnZ^F7u-!*4I
z82Gh3*1DuIS<mnRKssSi^7cFHU>YWG<O@ui7))JfO${bXs?k&yTg24ZSZV;{9Rh5c
zk{@Re2iuKf=I5)T86glNPKofT^+S==s;;WwH2pnbN`q7QDns<k^}`d24~yZUi8QPZ
z)Q1a?nhiGd8Yrl!PbKRAm+S8o^@79T!HIA9LeqslRyfU3Tk{qhAX~C8K5RK!HmabA
zjERpYwQJWRFqFc8FK}xy3pO$~mi+oP0%%Dni6p;?OzIYcg70)X_MaE>CAV%mlK>*q
zqHo+^{T`pX$hVgLCG%`fV!v5$?ZBtz8V1%MH{bY}6gb^>b%?oXeRHfkTxt_tb@psF
zTNkQGUpcB*wwxRGvB6y4es-5lP;^Pxdzk&W^5v&8&qIyRpC38@VnKW#09)}XQj$wE
zsr)R#214_#{c&x$=EO!y&JRNU@?d2ekL4pFzkPA6-gZBF$@WzPZya&5y7g!O`e`^!
zz->e)>Uz$bbWf~9!S~#H>_vV#%RTFAu6~QX+J&BiZa8_vJFraFC{RGvbQp6%$}$(5
zxt)D6o&dcQmJrWd7Jq_4U#FNqQ>rc)fUr+qaD75sR7A-i$xr(C`P3Vb2p}_v5JGD$
zNVY;8iz7z2<p31q7!EGzjc+RzNl8hY76gx1Yh+iQZJKFxn<QlRC#Cx$aY?WCXV1)e
zJ${(6JMC6-osD>I|FAV3xW;Wcyhe>U*_<A86<)SsO8zicwnlK#p>74rZBIiigHJ;_
zD84F9Iw%F2-9M-5A`&o~@NqiqDQ7QZ@eGltu41N4aoH4K<^SZfzyO~`lD*rO?Lz~|
z!jnEA;>7WTx?q+xIIelz4YwhR``%w|1*|=bG}|A)tNjGjWu)3~xtwplzdzqk#b03g
z*$0O>KzFt|Y%_>}>Q-fUqk}D%1>fRuM6#34Ez#_FKvXAX5;7$bKBM%WrLkjF=j`<;
z0b(Qu%WJOpEJj|?&ulrhD~l0-Zz6K=uW5x>X!=~1wwfq%X~!gSz|du~9m%(8e|x6)
zjAPiI`GX?z4GU-Utjpz1n+B+All+@MC7*4l2||N+!lJ9lDh*zqeW!A5zk7Kw2MLT<
zln;7M&RESq8v=-hC4YtJC?tB&hYX^p`^%VKNslNsJGcMS+0zAPaNKlr0^Y?Emsg(I
z*%A5<x|I5`>Que4MceUe=_PKJd^zA*kt@!aOS$N1$pO0?>rzg^lk`)Ggh?1s(ds5&
zUkw#JtggM*X|T({JM!?TlGE5(Sqs|u?tZQR7a>QssNSzq_2a=^Oma|wJ|Z0xd<Okk
ze(8oLba|XRb4@^f`Qd1c>eYJr-h=(}Z0r2=>7?VC^{QJA^u=PlZHX13>z^7FL#o$;
zd{pCP8zU51KWU0OV(Q~}*r!<kFUGH|ow>!1U3|az(c`~HyxtO+1EZsM{h4bL2m?g?
z5U}%s;6NRl<ZT;kvY_j+JO@OZs`8W?n^n(iyP+yE!2ADFHVN{Q`DBf-#og8<qSe#u
zLgo(z1y>TqZyP0sO3+C(RSxy@?{inALbtc5w$X`+HvPz!AhaRXno7?T@A$WYz*3(5
z80YtzqqhP$0xky)dsCA_iwZ4JVTCoTkgw(Zddx6cV>?wS51MiGx_d@Oq>5A<B1lL{
z@2pB-s9mgwJs-u-FD~-z3L;`+Vzkf$M&JEqQkSj7_o7oK3`%QYn9NBcL2p-7fL?-V
z9dQ+JdC#^4Bya2<9_bHM4n0i&leJ($?b2Huab(b>HDAR8ZVbVt?*ah#M(~<PeREk_
z@H+Ah2(MdiXZS=_ZDfQK-Bl1?Qvr|7n&`B9f<YU-_OwCgd{@^1x>TmoCKC{jLthY?
zLeO=8ca!rHElJm+!;{R;ske4sN-=yFzr(hR{a2x1<`=Nblba@I=XS+lrT<6STgOHH
zwOhl2fP@H&poD^;bST{*2ug!=E1l9HF@#760@Bjm-93OvcQbSi4TE&^?9tzS&Uv46
zd(Qj(<3}AG<~w`uE7rBv+I{leX3sDIYIbjc=jopfWO&z~jv8O(=F>n17!OaY<z9}*
zR_E%nE}jrvQFT(%8{^RM*C%yqCXGb5!^5gd=egMii_F^vnhFo`!(w8T^;GBFtJHK{
z;4<gD`X}<H6UDSRe*I)l?hRX$6|Z+1uIDO`!f<HLEDyB{diM7pVVVDbQp<l0Eecff
z+zxNF;*BT2W%IDerYm`!aEJd`R+E|+#ovG0k$!YX1DpIF&-s1-N`_3!fjil+ewPu+
zF4#RM2X?3Xx;svCf&v09%!t{B*Di;oJ7b%l#!PG6ySCq^iRWsMd9)=s`%WKvRMl=y
zwlB1-@3^QF0cIQBYNAKY4BW5PF2lJj(-5Xz2wlAA{40aYU^tVBvttQye8H1`dHri(
zsdA?H$xXwbyHB3&D8{U+E;dd4wRMw+qXXpJha!;PgSgus^;bzNgsJlO44UujpZCdG
z7u<2i%kcGYzCd-QZCNl4`28&iSKto_W8#Wqss!YX3**pCh#XJVIVAx$wvJOqylY=f
z<<5%cHRBxfg${G9ec`nA3}NMm7T;dm1*^i6?qMM}8Bilq{ra^a;Tahb@$~rW@J`Lg
zJ%_!mk!W~8H*;9<qsh~+{T_y6#&&QY(<0c0YY#LdvWvfQsIATMZ_&XkmqqxS*l44#
zA&>XT_vdJ8{-o>+WRptwAFwTyGxb<2$jggpX7SDbb8Lj!^;@#Y8y6%Q&)j1!=CqGt
z4hWY5Zl@tQCtK^bN413SoyYS#!?}hyPU;WX>%=tLk17jIhue1Q6tnC)c`_hVV}ipU
zxewd-+}r%$v;LYcNo@@I943)w)Jwu@wro07qj(FnB`G7Ksz;}hyuu`Z#VC}VQ9d!|
zUOc;finge)m*DUB?HJ1<`{h0NXVerMsm~Wvi%nlH9DRDoCh^9m`Jp?c^q)mw$H~`3
zb_W;4t^s{Y8nw1_LXRUs+YCFayE<?+9m)362Bt~H5UYt|uY(O(3p3v}nCMS=`h?4F
zJudDwRU}4KUC+%u<zney9q7AH&C+;OxhB;Og`CfDUb~z~<fs-E*3G6%P{BOrlP|80
z8mk>@<+B_uYxd=;PDV=pUXBTiVd5Jo-0h0=__}LQjQK`2Me*~An9K?K`|2-j<?8R;
z#Npk~QU9zXz1q8w80b<3*DMZqu592!gl47|1J)NTaA4`ch&ue%PJ?1tSh&G?aDpts
z`ISf293xB}Hq?QpyH<(DR`41eddQ?wTbCV6{o9om_uWC7CkRnbu?{8nKS%E0l<$~j
z!ik;OkH{Z9SkYN$4ntdPxi^I`Li3^8M_03M<HOgdByE45O}G`i+zh6*1KkWcN(OEn
zk*||Jjxp=HHKXx861ULGXSW)ACA%*4)_tb3LE3xMQOo#raUaj^NPV|gq-K7|e45ud
zAlO6TBI07T<{Ah?WmRkAp9|Xu++kAuP_5Ay_k4^w8lDtqRj;wPA9y{V#JfG}nIm!*
zKcskeu%o>3UhlHhR=Evqo;QErJg<s8Pxq^?aHp>kql2zLMiD)Wd2@7jZGS@7t4!g2
zBN=$DIdhq}>tT@7%u5}?QxTpNS5+qgKALvs%@GP25)S)KZzh1LB~QHjIz3(Q`3@rs
z57C63B|DdJ7XRbLM=Kn|(|#wkq#sl?G^&r?w}dJ^1d485UtKSk=6xvDcCXuY9kfe)
ze#J=4W>yW_0&pdRgkikyo11BO|CacLN3fUU7>QoIuvsCj$Wkki`+6hL406ijjNBAn
zZwNnbCg9DaG+fTU7DOTH^HtSxd+TzzwUCbixwbfpW?f&*^ZvkB(*A4jlm;<j8KbVw
zYBcTRcv`7x^(_?mdBcO^Dmw@p8(_?h`)x`&npwfxAEv8QzWnH|>TW;lPi^g99-G<;
zBNQsABxV)-=o-7G<siF004-otubqZoAqvl5v1ynTx1QLwy-Ap?NH9V{gxb2J+)P9L
zTk4|7MaRa*mi9!yOEmZ~6sy?3Ew78`cj2LA>IH6d=&CP*+Vn4UX6b%YJU<Ry$ilPP
zv|kvcNenlVAw3^CEn>X>MRQH>)SyxjJGAuqgX3kOvu@oXLH{F$?a9i@&CX7V4-K+x
zsjz49d)-A-@vW?ln=H}$D;`uKf>-%HmqT_Ixar@|wx~eZd98jrOj6bVAQ7UpF<!(_
z)OZS&hWt_<{7Q6o*7C?B(!B62NE&J%X1_JiXNh1`S~=y>pZ^0U{hP;2KEwK0s`m4q
z!Cjx|#2bOM_1W?FewTxzL??3x>L@hUqYhPF7d7Nkfh%Ie`WIH>@9I1siRPbd?Xy@)
z;o4e%VvwBTc=k+;nfsdTONiVz7l+NLvFQ(U2CRDZ#|yWe*GH>fMJJ@Hvt>?OM?QT<
zh;Gk4b<kiM-e$WWq}q);uJfd9wI}@i`ZW8Z*e!$qa*O?ZXg-|p%35(in!MISEFtM}
z;NN4>4b(w8fZN;asP<tapDsB7dwF@e3n-=yvbwn4xMvbuVfwE-&#7q_Hq5Oe@n)Sn
z1<Gh1*XRas4s~kdRVK$g-Z|M9Xy&YOm}AnQloQxb)h2gcc~b=~_c=S<=+eKws*@|b
zc?<6}|GemHl6{TbyH*}Ri7DXHWzgnzIxy7$xHXT}4{pt@Zdp+7h2L%Q>sZgi4O>rg
zWA>;+=$2ls2C;6I)v&_P7!r&G5rod}g13zRQmuo}KZ(5@V_Lbiw3xen`g-#V!MnQ!
zs!{tQ$T*iBhB~a{-hy%a;@2@(*Uf9Up2U_4Rk^4nm4b?4J^qt{r5)$iwl>LlEN<Ey
z6Oz)fuC|~!cs-7Q7Z!*4*f%L=-lFsslqq&k{S|;dhlMfbP^Yy<>oQqSv73+m)OIAg
zx}2Y3y{hlt8hNa$?OsxDI@|?Y-Ul==8ONPz!_7Nn+{@uLq6~-A{bT}KeGgwq1B+u1
z<bKVtN(zUZ4MYYAHhQASd;Ql^>c7@l@&f4CuX-$CMzP;EoX-Yb#sE&g840id-Gk=F
zHzf+V!V?>X4B7=w!z4a<is|Wnx>$nN8BBkcf+Nq%s>60rRZaKny4}km_@Y7A$L&Tr
zsZAB8u949qfRs>$2oV(bdjdw=hlEL#LvM_uaHh@ZNzbpIK{|x74thJ$iFTGl)hXjd
zb7H+FFF;C%M(UUoLfm!yjPCxMqw%Mvo5cB!iSy(KWvu&c&{$^Xb=LgO^x)i|%d}zR
zI#!<rNTt8@b`e}nMZw`UYzrFKrt5D!3D}zRXfi&3$;Q#XRig{5eUauXCGrXJ+AS}t
z`6f+Ul7Kisao`=HR6|7wIzh;W<ZxyXJcj1_*=trdcwyc3696pOc1M_}gGvxM%gr_{
z4+TQXbPo8}(TMq|t#d=Cr}^1-E*%>W?E4RB9G`Uvfs(#&%uFQ+m0+>X^*E8D05xx8
ztbl>Uejy@&UugAvJ(6+vUzg)Q|Co$wQ0xy)^8S8~@DOe>3eLIj@A2Tp(R+WVtg{yV
zKo8GgSTde${7{*(T;2Du_+_PEzdLbxS=8C%Ir%#Z8m>iCHF8dlLXGE*W6sB1dMDON
zkld-Rqep!3lW+{!uulEEsB+errzBNdX5*_pZmPB92zBN@_&(46@<m`_m@h7AM=trw
zgYm8G&Ea9TbeY&ywzJEs%+4khNZ0guK$1%L{BT>Ob7Se&ZtE)fo2n(=Xu}6noMfx2
zE8eC5hMp!LgqGQV@>5{yVK|Gfu3#%aJ$5#O#r)ieu~6CD{|jWN&GVeK0?=J|t)@~%
z(zV0j$l}q}riUWSZNlB6qv^cBMR3Quo6G0%HbBxl*f~EO!*@)^`sb_steYH)^>rX0
zA0jR8&uOV4;qLC<^is!lNdQ)3t}5K1!c9zmkKo=XS=hqX2vl2>$GivAcobR$O1siB
zJGYGsRaVyEYGt7U9frfrbE64sE+S?VqA#+hk1q1MA=>?Ojjx7JPm@SD|NY8=H4J5#
zCw+`xvt(e90w(WwYYCbDl)$uHUe(pyr`wUyQQwIdgTZC+ivAJo{<>9`(Ty|5@+C6t
z)jx!q^e&vEzwA6Pt6Azlj+kUuqYmmbeI(UeSYk7*!n2gWfAg$yuQYJkiA!5iFWpMo
zBA*@e-id<3>VxMd^nb;Ua;t=GavJlhBqIe~Q5ka&;QG72KmOoT>&uGh=#RYN)QL>D
z@E+YUVY>|PM@41vO%i+sdIo|4N8#&P6bCRrrC0P66TbD5{*z(YKt2)4M+L3Uo}P%Q
zs{>W1{Vln<3K+NG$r^$4j*qIrQYO>h&on$V+*E5P>{5B~*_H>C{s&Y*_&$<vDE1&+
z@psej|H11V(ZDv3w(9mmch>Fvrl!z(eHpwAwr_B}BT(wRuTpcJC?pj(#&q-OSJ8Kl
zOQ!2w&qDyjn}&}Kk2PM8mEU8tMB5@0X>|&W>Oi=AWaaCW^;JTTcZ52zX6m{WLoP)R
zCX#xuhM*X?c@%Zr7bJLUR73Dx4%xqa-p6q+{(C74iY0$S|DcJ`7tjA`i)d9SgoJcK
zl|~%%v6-+^W>xkPJmuCkHJhMYuhEwQVV~x#5CU8+|K(DvYn0Q=ZI94FeHLqPhQ7yq
zl7+PwdVE_yf4(|JjI_fi^Ju0Va32rSz1o2u>wY+YBPi77`?mlAw%H(+Fh<~7O;r5v
zoiEIR4`7;b=l#aetgHP`J&EU$0!xBf2NB4!#9kxV(NeR!9nrvOoM>0l3&sXGB*oUl
zhpMoz__ME^ot<sD1!n4UCk%Tj7a|V6-sZ91O)k?|bd|I%HTm}03E7ai{eq9DD#q3C
z&$iFTEf+b>Qjcrf#Q@e<^<nMx+?8$JZITJq!rC1bzDHeYa{LTuLyQj|=sthhmdL~U
z?ku#6SIB9e@vl7@fV-5ieolK_-$ZYB5l3H~6Z?Aq{>fNPo>;#WzaIx_{nXLcx`a?-
zSiF6Hi{5V{i`QGUdY~p}uZWW}N6c<s@|K;S&RkdY)ir{Yf4;5UpvoaCwiKn!^dmk7
z5!>{Ib8m!=rE8xBYIc)`ETJ=6Htf8|p5*G~?Kn=$P@x%zBsX63;9(_STw+>ey8yJT
zmYX{6+m<z_V?uRP-TDhCYz3W!(b@)eS1!#W;SNheAKv*N|LPN1l|H>h6o!`vZ#U%K
zjN;&4g01R?-MTteYpPVBk_#s-YJBUG<Q6ugG6)VS#lJnITEDr{(TCj}v%1A4q&@l>
zp0z&&>Y;&4M+*(|)9$mD_0ipvv#z>PAe4BOA}OJf4$;|WLP!ep>BMge^Yi?s#!0&(
z=4rJF@6!Y8bL{R<*DpoaBIra!gj4^Cj-eF0KZg}(=COMvfAx_n!jq1#pHWncv`=Qr
zFgP1dD(=n3u}FegKDA)p7I^;n;{>inXID?|!0^!5d$MD=0>!fCd+l>`v~xXiPp+L8
zlQJ+=*95_^DN1PCwjuv~ov&u8lJ1DVLZ|zyoKUp)Ej(wHo<@++cf8d9KJmKYm3^PW
zTTDquMHX2yV#+k}dB37{jTlS3O_W0GJ0@Ph(r|$>b+U7P{5rfjwSu4T(6F<!FBEI3
zKddc?)ORe@G{Xzm3VNARf3cytw`dn4ktUA21$`ZCAnD={7_4s38+|#<M!rQv7-;6L
z+FwRYWt{(N?<A)3h1u<yyDQ~&duY*J$Nh@gny%1%eAYhmf&A~E8519T)*D!afYdF6
z*75GHHZdb+aW?h!0=TIb?BeHML#Wgj$8-e(C`4C!`Tst}djhDmxgP};Q%Mz!Pf{70
zKGBKfZhCaa|Nla6F+VfnliNy>Q2uAV@VDQQRS4_2dx5(xzpEz3%<|~6+u1iO9IFU8
zeSMMjh=72~#-=0{_GEF1;3>0;IUQz1BbbQ5t$ObMHIe-DLNW=C$vZI>l_)?;R?+v_
zjjzQ>H+d%s#d_QMf+f^K0?9B<e~QZVLsXjO84{00y%YDpz2^-t{kt7~eWA$lP-eh#
zIvN4~w|A^8KcyIB9s~z)&KaS0a-cBqgvCVa-&D1g-j$TJ4SPyWO{Sc*^;wt}P)t6j
zz9Ri6D*t`6C4WU#hry_;DW`Taya;pX=+KlS2A`|h;Ytq*V?#1B@4o^3WzGT07L3yS
zZ{F(Z70+ag&VNVO3#qxq`}cSK57cA#?yf=ME}H5{ekdb{dHOxA1*-drgt(RRnjB-P
zC`@K@FO&Jf@P-yo`RuQF$v;2F%Z%Zhac1#=>C5sp@9X9CbTe}W1)9P(v6n>ZspFoX
zgsZ_|1+$9w@_)#z|DcCNU=Do`QeQOu313UWMhM%iV8yKH|LrXZ2YRqzJT*rbii&S@
z3q%bQs`U%UJsF829SBp8t3ls@sDStUZ%dr~3EkGtZtE^bxQK;?G&*@gOHR$CxlJKD
zJ?3XABuQLi2=6Zy@qg@aKoSR)w)wa3^UfW*+b`)+no67s#h%^v5*5z@yH~;GJrC3W
zwhOX9dI8Rn>%hs$Nq`Xd>jUr(=#2kQ9|5pr%$p7<B9GJNC-}xw_r!mMrpbUOFsVz8
z|KA@W){~|_b?8pNR|8r#k<LR#Uor4si=PkutA_m_(j-{XS7qQi2i+Waxc;~2{JoPf
z1HnIk|2`cl|M-u6`q$t28+o0W2-cWzS5<6sR%b0VY^e}2rvnw!S?d4%ss4FY_yoBt
z8`obPuAwPNU)u+78*iVb-Uz(Mm$+_PWGNo<&@3)kP+#p!UBq}B_!c|+S-w?QhY_R$
zeOdRyLCf1L$L&25>`P>DQ(#Cw*7!Zj1Mdpt$v@`k1pZCXgVPot_z{LBBv3WT2@VQK
z%E}g!@!NN3<2gKe#-sjLQSlS!9MA~yA06O0PFXR>Hbpmj(<=~fBUx!3gkw$ExVe}-
z$*G@h<XCQ34mjaf()J4nf+$pKe5_*%D~;hxEoxL}1e4sSNJ?te&T3Gc*iHH&-g1Xc
z_fv<si={ca?ir_9`oqmphfP4KKRtAM#_>$7ErgVCgBd=#Q{Z$>Lr2$XyOKQ|e8zq2
z&f^FW8znedkWV(xBL7w%{TI=d#_v>-f_2ia_;Jo0s}i&Ko63LdFo3iO5D%gs=(r5u
zdAC_K1>a(O{l>x~f32*hJ#i0BG{Vr{_%`|5t+K+ccrs+sw$04PRmC_nMC6ry7tGaA
znc)X=Mu7zP7jDxOcpKM-3{nZeXv?A1bi(oZ5BlnybZhgA=*G2mwyyeKgsedJqja}=
zPgy+964))&giIc|lqK+P;}Q=b?kC*{)TVweD%DWy@M%X?BLl-?8NPapU9;LGCV#5<
zMKZKJpw?ldBm1tXq`Au05!wFBFuOJDk||Nx?u6p7^ic?Z&Bh*1hI)9LU>%Z@AQ-nF
zDW|QjdUm}2%u=I@aq~ktQpa+C*L@(XWf$%3eZwMJ?0nkiQvn+cHXUR@l!u?qib^7^
zAV1%!wm@ofm#TILqWf$jwM$C{;}2f{pQW>K#XS??E`($W+##2Z20qf=!3lK+d#$U=
z3^!$oyaMi1G^J8j1oA{V34g{=F_9OuX};fFYdy(}h%FKEX0lNE`on_gUyR&lN1@Ex
zKGW?C#rC7UUo=T^afd&XA-$)I=AFIW;qOJy&Mn40y~YYOz6<HNUS)5W{d%Y}W;O9N
zPo*eY*2p3yszB2tGU!w@gszElL_cz5r_J1hsz2^sge&<+>B5uVr#v};@pbxSznR;k
znUa{l*s5f>XpLje6jc@DU>m-(+VD+})m<JyQ)ZYSg+})8<aV&!X525NDoi!gQWXCJ
z8c^DKENO-_hY<p`6sl1xmyZEf#bVVSPE7omR{w2Jc{V9vtvxWmzG#m#nC!OdmdhLO
z3WzjcZjGLc$YZXP|4;|lND>rEIP_uEgbt0(xU~?5egD2zrmbo=o=YsR8WjLJzta)D
z-^gSWz0e}d<VX^B&&hxa^)3Z~0iWrj(QJOcLlr;lY*Oq49V?t$ljTcoxBc||hx4Ry
z62}kOLz}a2UV@4~YcK4(fE$JfbvR25t%N(*dII0^n?NI(A3XP~iey)&S$MB4oEpk~
zte~2dB$$i3rKM$ta5?kKJnusDrZ+30((0uaVguV(k(Rx;bQ)ZsFuo0N0bps6))R8K
ziyW--kJ^0#t`WB%Et7svRZRWa%Hy$nynQIt-$1!NRT2E%RMG8qI|(Z0XZm?kEroZi
zQ)@Sc<$MFaN{1Q>t#-#2pqCm(cYFvB+I|!D=AB@Ye26%Zs0b&!;H+6G(-}Ed`c{Z(
zaNJuTVHI~N4~ZOEuKyuC|B%2U<t0&K^1FuDeBGf>vzi^9sk7B}&uT(-zv+HIqzx1A
zR0AQ|EB9G))z!pzwR0Z)2M>&|sy@b_-p<xoZCVu_t=0KT1Qboa!t8E>O1P%T#WlQ}
ztxl!v?g|&9?pS0xx3K;JOM5nZaWz?MXqWWipe7*-2u3B^k1V%sO|RGXi*D~<NOcE4
zS_v8QG0f_WNNGQ0KAX16G8F<$4#b8h9NORt<POOqV#>$}@u@HZLv+plNvszo83=M<
z!SI6%CAUP&<PUZnQuAN5n}L8RH&vinB^53ED1H3P3}nMWj7^5!YVs9(bxL(AG3$jl
zet@~=<HOfIx7zp+a(FrB#(g>RQ@YP>v<B>k=do~8v&}PQksNkFbz<V3Q#KXL(o=`p
zMrygjpJi|u@_`Ij>`9=tP;cIbtvy@HplifJ_pM0A`bG2oyKX4NV3m}3P#Ur_<Af{h
z_m90~Rt`tIP`^a-ug@tPOh~+-UZWgCV>J2ZL(`<}%|J=lAtFR$YulsbP~)uJmxzL_
z$U~Xmc@s!C#m1zJH1t#dY?lxbsXwp5&q03Y={m%1KA^j%kdG81Pi;=J11q$M&Yrwj
zD%HoP=wtBR)1T@HXJEGA5J61ZGg5x{UtR!QmKDlpPH2~(>}@h+;-YLts+b-375kx2
z`JA?~Omh2+FOXScLoq`!dab&gth0x;zrKh3JuRk^4QK)=*HF!#e}>w3;rAM)WH2@p
z$#Q;OPwFc<k8-m15j?rJft$JJf6(TG9k)tFb+`3u!snYsQ@b_J4%eRnf$<=-lgXU;
z##jy##fbv^L|ARZ`O%cqC5-@;D7$=$O&UfReiB6x$(WNj{?WD&ZR1t#jAfk@!ae*2
zWp^g;5P+shzaQaFZ!f3~l+5m5U*W%11vdkFSvoB^FMX<H*(6`|1C07FSXiv*fv2`-
z170YQt6EDxlh|ij7E>j|%to<zG&0xA+8bQi485`JM}>MoUq>k~j3CfBn8C^)IIdbH
z?8sl#AEexseu$iC$XB4VuPrhV3}iMFwa9gaV`UMcE1M*XrK71E7_Yvm#45Cwj3<r@
z?(eL=`z3}mzGM%HaVnvsTsBPa*VU{Z@AKoDv3dB9sitC;zLA#p?--une?rNE;|)Z*
zwD+=elkgPmvA7A*@OKGLoR^H8mkL4*_A&^z%aSK6%)2%?HsQm-u*_V&ymp3PaY)t1
za5o1sb0d&Hi2F{-cd}$bCJ(2UNI*Lh)vVGR)o|X3xjwTnM)%641%d6e4hL@Oz+!!t
zgUM<zdx5BC+-!RRtrji96Z?9}(8#cB)o%4%$qJBh>vkGtxY$ic1~AktX17d_vxKx*
zBqXS5?B2sFA%4^6A?@&xDlOv;$SP#A+BC5K;>=Z6k*iCF&Ddn!wM&B4=Hg*peVYN|
z`{`-4BTs|}Ct+|_dgYYNWLk|b^YeD@vI#ogfGJ%nRC3odu^Mn6YWICK*5YeA*5tX)
zm&5Y?y2bKP^=hS!80iPHre>dLje6oY^CyoX=Xh_<D+}Ofb2=M}{h$<i73FyqV#kyy
zH1ATXVlf-TGHgWHv67l^UMb^eRAQ^wt|noAm!WsF{Y|gZey`cwoTeq;)o;9xR$JKk
zJ#V}MMdJ?Bw~EH)?jeWMQg_Wk@;mHJmK8Bk>s&4S84mhD7;EMIke;nWr2SiDq&`$(
zn>fpz;OVSeTC8WlP!SgsG8UF%u0fP_6Q7FP9623o5om>)XX+2#;JOsGd;9j&r53GJ
zBYZTf-)}E0Ll?x~Ce`Fzh~P<ZA_ve5a(V%xa;IpBpt;;;5}pthmkyiyWb76J)|X$Y
zFsgXl6YJVClQ{>?V%{tTGs)62psIj2p%Jf69D-=p8HaGUh;|?e8!OQ4$Q|7@Jj*JZ
zIbx*1|6L^x5GLBrz!zi2^!1xY9?v!t%yzfp>fU4(4<R$-653lHS&V(82mzBzc;%e-
z(oqTu<}v?nwmo99RaSX$&>)Z7?qHo0$9pQ<`h9P0QifPrreHH+?fO89z2+*pR>~A$
z7WP8z)w|18XDj#FWcb?&looOhFXAE-m|D?gK;C^jS!K>tVK#O!D<-?PBjK>DxL%Q}
z=;UVji6pklgx>q?_-OL&+2U7qyY%2M4=pB=-AjV(?CcB^VyH8vR)YmCCC@~i=M-hj
zQOE`PP6~Qhy{OH{ED!Pw0ZEEdqDiQla*lsnvy4M7k+m_)42;eXOvoA1T-v3gRT9|`
zWd#tw%ZAr1$tY$O=07K*XLb8G7)s5gpgkwds!^9)SnmdetNdo^zM1#;gv&B<6d1zM
z)z@11d(|>+oMUq+p9t;^kC0!EzLx{5Mp24(;E}X>Kyh~LMYi9#z%d{*OhyyO5^SPv
zNF+#3ub8Ay<tgm&z<;^lZ^PXZ$kuYbg<CKVU)_8(%Q-x=c|dyYwk%}ZV&y?alQ_@|
z9TCp@?lJ9#bpj}_Oka#52fpr4NoO9olDkchnaZ~vHYNb-rI>C&C25aPfd;1?5&VsK
zX5DlOM6<4!PZ4r^4$U5-=C6Yc!dsIra#(uaOw!aCOy<4bPFdV2M6>El_HjVmJXS(G
z<-=y_UuNapuzCIM8j*+S$hmp;rh7$3tW0w4KGE^@K32`a4=hfNKAuX!Hy0m%v62#J
z?Tc-a)5P2E?!E_#p~Nzm5yVWLV^kn@0)hQC>4%zOHQaFK1YWcm9jy#mAy&7EoW17P
zY=G-7`|W<wTEpVe;29HjtX$Ss_3lNP)P>5z28U3$Vt>QaOPAw_eVTYfsE^w6+qc&i
zqM(pq!e^}3>+Wls59p-2XYdBEU&}%H{422lh_?8rs3azX8Oo_W+EFwteFtu$qBm#R
zcekT}s4mwAj;6lG6`nAU4f&Qe+^O)pQKZ)q&eE5ZlrHzkc0?sTz#G)iCHCp}>ylsn
zjs&`rh0`H|Xu;n|z|m9jdg6TsV>uB7eEg&P<k$V|r<W0Plg4(jknI|9KdLta2Huvr
zH$(aRG9wvj#{h}F>A<2uBR!8rA(k!&Rc*7zX{~r=K#8rR;}{Kxr&9EXC#Z9M|89Vd
zs=(4jFpUZQNkY5ui1BWY=x9e3*X-4sp<%Y=tupbB)K;O1(}+T@ySBVC{^H#UFWmf{
zaVuZmj8*?3!B^hggELB3(BNuz>ke5Im>$<|hE^t{5b5?^_kr(vr|0eyT)EZiAyawT
z9cPg*+erTFxDF&oI^oFoLhq9zl8rj(Tq;Qqki<?e?^{x4lztaOf{_#7gFyO9LA?Tp
zI58=SW>}Enc?<)8<Cn2F0imS*5)|fQK03sAkl_7CAQdXei3wOQIkxSAs3Ur*17mAm
zUz@&7JoSF5<s<hXb@-FX3ic<46D*_$0Hvt#&6(0xX_x`jKJ|DlBBIIaT>0mlQ@3nP
z!(ikX_cJp-T@qeO$oD9)q`0W<PMa1<F9Z&IJTg|lw8!<_>ouH%UhZ{~BD!-qHn-`_
zFwWRsCs5zQdtBPT2|8hmC>)6IG2roXb}aTaUUnP(T+?UGzJ){4*TKDLn~|A033m%$
z5C0_Wa1XRZyxQBg+~byjO#x8zC*VFW+Xk#V16J1DVv1=Ag7(SfrlTT27e+{`^ulye
zP2Hqabxnr9pa0cJT&<I5Yl^5}`iBNy*$1x=3+Fb558~kp2Q4e{W#V0%RvxrB&l2<~
zQKkASQ`GsRu8KuKb*}Vg?)lc(GnuQ8-4)c5*0KB-KkdMjzPon>^kxD`sy7ErIEtul
z24=0b%5EGA>=gXvQE&?fgnFQdhF%u4%J)~KtU+pi?h6XIQE}li^|gkRtMRgJo=pJ_
z#3;cnxT*>7$(F_bjeY|puGXQYVCjCrtpIvSR8$waKKhQzW0M9|8fp~_$0vg1xlMRj
zqOWdvnbFC^lJG|FC4>W==b*8)$H@%G&<}&yYsZCSB%QjNzM!|Pwli8f0*DKwqOPkY
zx1!yP>VJM9w`+SI=d9P(Q{xyLsyh2oZpFPuxUhB$6;L6eQ+~{3&zp&5C1m&b2^gO%
z>hL||jAHBwMn?=XfCA${N7K~^&Lh?lplmKiwQ#U-xhz%qjl?I^>?%FhXQ^+>s&sCb
z&y8CYda^+pV>@FAoPr*ob{Tj-<TI+S3+x*nDHR*klHKveCpuV2$8az(Gy+{f-76u?
z0ERgxEqWlB#G07SH>Glt&GiiOA-p_4Dt(FT@^uK6(a8NO1BfS@WE0+Md0ZV%UPZ+f
zMgTD#J)Xc>{N?j}^%es3wK*_x-Fs;DiA$0`sXqph=}w>XXbN;hdgvErswNt91$BGN
z`-WnsJ`NiAkmA?B3x#YgU?H+80E=M;FTEH-a3|2eQq*mmu(DsP+dns{ew~+e7iMH8
zz=S1D_27ff8B#72tS*e`RR;<H#OSYIHZL33oqBFz4WT6Ay_C;Jk}xRWB$o=3Q~G<R
zfS81ivo`naz<KEE@)qP1L%gvB$wfw^=-GqIxVG<Xvf+T8Pd}{VtXUbyqKcTUlQTne
z=BzN?XK{3bL}i?t{@P(H49n~-ySYG)M*jvNa@e1kcb?BGdLm)0^nxjhG?{cY?f?u`
z=_H=^?XGjR5AYX^BRT4dpbRj|y4sZ`7%4Sm<FV|xk(9VO_RHY?K(ZfEZ>(hdo$Ykc
zk{X$X;%(o55S#JRs)@l}gMr3gsLyzANQKk#RLsbI1<PQ-l#jKnY^(fX9FEmq^U)mC
zE}D!djJP-l_yK*GdCvy#1Y`HQ%~>39b{PHGhTv0TQqyL*)dPzohK7dSsR)naobN;|
zr_GkD4b~!6OVZd1{bb6shZ`v{1|xUQ{>!64cLuRQck@BIlytvfu6;Yj--a;ky{^zN
zJEOX8hmcBDk=V!E`eKU1;H5wD2+=7u2Jb9Q67cXJa6NjK_2BzlFaNRi<)eaK1onG!
zYuWeKVs2Wk0_8a85(7_=nWh!>jO%X0>&=TXmws9L3u;~GB-SK6IFs}>$Qw;1SW087
zEhRy*4R?i;>5-I51*DS5-ekN@7Jcv6*M9h?FH7l`Q&Fu(EhENuB(x<azwnMu$0P)5
zV;ddptJL`}lXTbTO2+fgVfP$N=EgFH=DZW`b+gg=eLAKk`i*k!tlbs0r1&{`Ncu{b
z<bH~SM%O)y1%tc+VO*xkXvO%e_(SV@a|$15eRl|hN;eKm<Fn&SI48o)iRT$lrz~kI
zmwE41^4$+}u@?7`ep0KBU97{d_OhGvl`J2g03M$-PuTk`(o?-zuAwwnRi2u0tV-uh
ztoQ+*^s}QKV*6ucUN`Dp8_HQ;BIEt|#t3qyJRn%lfR~juwJsN5dq~O|Bb5160)S`f
zfOoaGS($JIw>Wl33=~=p+`zF+dr3sl9%2(`Y-9v;Pw*FZx`_kwjSsr?XP-6i8aJM>
zNERFwLiATb8@;dX38BTe0htVN&=I}e?nlx9j^Dz{T9LvfP`mpe3#Hgajs^QO?gKB3
zCHS(-m3+!vK<00#_Eo0mY?`H~PmS^BRt}udel=bo2`?B8S?LKI;3yreiW-<d=LH@p
zRw%1DO=s~NlX)zmFK?c8EM^FG&5^r_)W)uh7D6vXlAhEb&R#FaA)-3C!%f}0$8-%x
zV-HJ#Bct<tBA>;L5;^^{R9ziM$vRG^1hI4YOXJpa9GL7Q1m>%_8*)~Z^8vgs@6W=h
z;|*5@dJWvHB{lpisXywam2E&>Zz)wr8ol&ph}Zu(9^JlIk3J_+wKNoPS=sjO<)z!5
zyqEZ8jVF^B+Kjp-lI&~Ee1V^x^n1BWhNPuR0VG^%0K=}!DX8(%WnU*VH9{qhF9k{$
zQ@~6_mSMfI9HBzjzuMj&du+~x7=n9Jhnmp7n!p;2i?D*TpM|p|s%=0sICf>(3P2GE
z5GsBZJLtuvbj_Zn>CA+wmZasYn;>&MN4d0`q64^T#%8bUdqdDGN#icw_|3<8_03Zj
zEXUg6)ZVs$CSvBY&Ut64?|FlVhuuanT%c%IlOk3xTzE=9)BxGdB)g#>JT5ri;3wcO
zINsU26R@<O;s(4~kch>?z3jDqKkJ}4cku5vu-4evaia!oC5s-1SwJ!K#TwJ|;R7ix
zD^5yt?LR=VbMJC-AE>&z(N~k0aG>Y2g)G-Lk?|vGaWnnrNfqW?)p-S)mDbUAJ<+KN
zKSnaavJW!OpX<J5Pcr?S6o(<aW`0IfX1V*>qxkc2?Xj&*Rt5@j+Uq-!)^a$DkosQg
z*llt=0k}4t*~5x*@uJ&#HPJK;UcY$`zzFgytEh{UVUj^e-$TKlu4Vmu3|944-VE%T
zt%lL@#;m%Gq2kmxeF3T=yO~0c5k$hd3#+OJQE=AmCfQX#nqJSTo|iz6&uUP3<Lq&9
z{6`Vhee8=H*ba<mL5|zT`CrJk9Ns#0a@+nihB7ojFVD}6)=$iuDb=e!q)9^!KJ~w6
zPs0FoK1wNJea3@EyZvj<A5PmvVskEnCHe}~fgL0qX}`14fNMF1n#rK6)7%pFR#k+)
z-kuD0Z*{SUKFry(jZ63hhD<+{y26N2!bk6#S4EevwVsP0mx)q{d_WxdfH8ovL*!1U
zxu;K564!?)z&f3Tqg=|CAZD+DU0=7L)6`x!XIx1tq;70)-?x|5>dFZPh3L#DtY)9T
zzxg00++6}jSK|}I<&ePQ$4y`V8adaw>*&zHCbJen-d9dOOUw{|0|${$<U(@Rq%Gx)
zp*(T({?+hLgD=(FQRXwb<I*=?eLdCxk^V)0hF5z6C<6G87dPLIOUEqevTOQas7t$^
zT9Q5-Y|wmgO}%plGY*OfltxQnTO9_v#=5C^pBZXbpV^57#)k~jZGrC$D<tgFJq~u6
z3u3v#0Yu`W8d>eOW?d`o3b%}dwD5I^dh=1ovx)1tYzvUy7y@Q%_s;1)ES18AIv?u;
z+ey+XgYiOvp=5XMsv2HyB3Z+5aa>E1EF+mf!HIZFz;+W($KpPvXYWWAuH0cJ(bd2K
zZhP!%zq7bh(>d7s9~%7HTh|ehyb!<HsX)|~A1`dTb*1uS0zUsHoZ`b!`x+)|oT67v
z67;#-8T<7Ie-F$-PT@W_Tb+eiG?XC!0$=Oh6&KDn21I;}r@&h~@z=Uml}$fEh*NZe
z&yuvtY3vHm%=`U%^Myr2TeE9H&L%z3l^eGA&7}6DB%}@pZ?X61l;7rnZ7eMYT=_zt
zvJH>z8|rp935^PassIz(0h$qV#uOA_(oxeKTwBkk6d@jNKuu@~?y<DYgfX)J*ajt&
z?PmPt>UTO}jViMzpyhMSb({6Oy<t9#_h_7_ce+CnQt4d8&KUoxrLNDzxgB@`lsYR2
zw<z{sjB^g+VP>tBX7UV0D;~5gRjFz{eTM##+wxF=YRRAfY9#P&qs~n7aM5j^iX!Cj
z<gv>E&*^2a>bU`IcPA?y4n)WjNs6+p!$BjJy}#`Sazm_oc-E@s&kuj~8Y~>-!Cq>_
znBQlN3?KGInv$uZ1RXd8FuSPi8zP=Sp0CpNZtmmvmi#6u`?I4UTn}bNT^jq?j`ym6
zoYxqaFT_(nZEPaA-@_$LxlQBs3z<;7{;bWdN=)4#yW6YZ7dB%Of7cK{nC7SR5?r0C
z&$EQU@p%{5`4-7#!+FtUkTEdtr<^z5DVHapD>BByo0XY~CT!}P0A1k$sUKR}-?LMH
zNnI5r=t1JNf>~GSvHJZK7QMrHZTuEUjmxGs&`$5|H<EkEzU_VBTS3PCaRR!JhpMa%
zRw>sVoyy<>0Xg{s@t=cQnkU^Y&^a#)a;aO0?QG3N4`9oQs7N<9_-=K>%nY8I&6U(A
zqBb}p*4r?2Z=IRsU_pray6{8eru%iS-Z+4TXk_NafkEGLZZ`{X>iV2)NfR$HiJ`~8
z_fsjxcH2;JJ?zuYymHOes!XYKH;ZlaPn!?hNFn@t@AJA%9&o?TnCq}%e2hqXb!zid
ztNQ&=`_0Lm2^*p+7U0<-e-{;f#Y@Yp<~XLQAU`!ZAT~#=oA3qp?~acGzt-~@v@bI*
zc*kh{gy;i4vzqHn0$v}}Cm6K@h>+<Tj>C;N5bx^b9NfX<YuB^KPXzhRJ0Xknv@3Gl
znQ0#*bw&P#7b>XyH&l)`oIhz_%`e4#nCTe4RoA_~LLBY1ZUiL*dG1VXnMu>VqW=uI
z7(daCH5CD}dg;ZbCHSTlQFqsmufJwO5PDf<;}%4q8sA%zGk?XVV}g{vIE_21pN~F3
z_cFQ1eMSQtOmC_o3vZ%loAEUodm}3w1@OjNh-vw>zFY$Z(k+0Qx>vq96@~B`a`*@c
z1}brB2jWf!i7s0zW%CB})tAz?Q%5e@J(m)1J&x3<FzXnN)>YFR_;j`CHkfXgp4Zu9
zgYXXNUVPRqwLNFL3!5&oI0{o?ii6S1v_=30wM^XogsD<;)ydj$93jR8NW<JM+%2kD
z)8+0W=wYD|5}FonIwc=<LC`;+lRa{0d`)zMSuK1U1q>Ex#s35&S3H*b9KPctJ;w~>
zyz&1UnPcV&>*zq^N*YRlU#UK8m-%4M)}d6jcds2KuJy^jE;)OD)nhw0ZFUR!xyoS{
zT0TWrMvXo11OAdah*6gAPSBSx6|hNN*Ug(#D%U(t(<bUAwn5XE$A0j_#!D$cd!%xj
zS{KEr?d}sn=+MN{iJeWU9v5oIo6X+M9s=~>Y>{XykfaTPsnL!<IjEq&K7PTzcAxM=
z8n&9rcx9v(ziIs!qjQy=dcb)meDh#^qNKn?1t($H^aoE`;3FRlC6i2FKC2|X)VZ+^
zK16>p3KkR2O`L$47ahL3Rolhn_h`sGw7wOzt7PB_J9OPiuoX?)FRVUrBdt)?DEeh%
zW83Mol;ONjz4t@PYO*G;kC?CYCxqj>-f2kQR>U*eYqK=+&wXyHyooTo#wcim9Cqwj
zCrmEHWA<~I!A3b(y2lH;<oHTtT4=@Dr6^qN<)H+zTV4lsdbnFE%Kg8oEt9!XE^R+#
z1#d5(X9&d_i;E3+B4-{uto^g)%W*;)TSD)k6a_^;hGq;32U3T>td&y&6$fjvhgFVY
zqS)b@dwiUwPNUXLt{58(BxZor*=4-<c0th<99vc89eoo^?B_wu<5qlbVeZrUsRVe!
zUQ_2&$-@C%Cz~?5N2vM-aV3V<8ZN_xOt-1xEF$*FOY(zF^Kh)FkGpenbzB|6+f$M{
zIywWMsJ+%HC^+HNUn{s&gYq@6mds+#hDi6?H4EA9+I=x3Iol)5+g5rQi%S(RR%VCv
zwu05T7HQdE@f`6dlz&Jedp(s1;FSbQfOy*XF$k>6nr?gV2s(jwBiUVm2fEuLL&~IG
z`Bruv){<ak=oND=c~cPx{~VD`MxBF185C5-8hvCAvB%VaFqj;McJL(LeJSQo64gdc
zv>>usajIP7W$F9x_IH$u6dW94%v`$Ik3q~OeYzkZIkOV6-<CuNG@z-vJ)@>@QVz!0
zKK+Yo1zEQ9oy*(OH<>>`7bH&|7w)Y?_P#i}ZclQ+`-Ege52%R4;v=4dE9f1QbWOd1
zNpm;Eqo%iVbRH+mp0l6Ldz5d=`E_zGAc);XQI#a0-^@#2OsLW8a)_QflASq_lldk_
zcWyqH)~h7}S63;}p|OIVmk4~G?Xa`kU%p8ooU$~cf6SMW_4MgS_hoyH(Do$vK_l;n
z0|<k<klA9Ens@cn`yuVG%Ivh94l~}CiJO&(jxpYm1@<wKHTy&z=mhP&$~)(YGv`3}
z3j^X4e2XiZ-lxB=+LI<N?ve1@&RBWqQyqWWCYMaW;GK48m)Tq4yAFTdJ3_}>;b&Bq
ztyo7(vAnx*JI0vRE8b}*X2&lySpU*jX6F#xRh-;=@B+ZO$T~^>4il9e`BUWeI7pMD
z1Md*96roS$?O6iT6<!b_V+)vkShKJM5?TJYeNR!3de%A<79dOKQ7+bqzuo8&PK_J5
z%`7N5%O99aSlm#3=IN_MfxiIwEJ}OmLx9ghjaiZIunQXAuY|x`f}MpRUg3nkAY`w@
zJ~Y(kRTQ%>ovue&N8ePs9i0sD^x`rRUUfWM%(H@fhZqGIEmW@r1RBm=-Q2q<ABX=W
zsO?-1q^b5o<Gpe@+~`0QHi|s<FpA|rSYxwo;Rw%gZ>UAd%4{YCh*GGc%VF^p5u{Pe
z*48#-$|eDIQr7iw+p|`lUI(%S;jup{=1F3BS|B<l4@|OyFch^MVCbMzi`+Y#BJXNv
zGm6W9D5g#e)hHh5a~REfjxEo3(IYLB3KL*&4Uam-A)0CmSD^BnbU=6v^&VoX*iYWS
zwAMbke9ObsL8jJxlyc|4$LaqoYzrqNksZoFN%f&*KVU^L$%tq8$P+{5vqpT=SNSQw
zp&joK<1WViU7^nO>-D*;Xf5m6YR|9OvF;|oCMng6XpUKfa(x}RB8Zf!#Cz<h?~uvP
zwwr^g_{CLb9{ff4!Xmj*ZD<{$ezb2jy_qZ|rUwtmD@*Oh9}ngeXopCZ)%Cp?)-_2Y
zxm?j@UwU0vS#XgD8M}5n7J`g9zymtNpN<+pF~^LBmsnMh&CmLDj(5UT7fiX#I(s!4
zrJ*2ajGX`D8++bw0CBXxsW4*q?pkr0rx1GPlCm{hrfS=BHRo3Klb0I;*ej2g@wjxU
zILz*6c-vO~(TIRL0MrbB`$fK1cz$?o=r~2}JG%YZ*P%?rz!o0ws4s@{v~_oaAl1N)
zprNB<*b`|65{iLRF}A@i%)SxwR)+#FBGFriT*JQ0T|U(KX-QZeyNNBG9*w@Zo81M}
z9eV3{4Aa7>2`l=$IhPiS0y?e)><UHC>sn*d<k-|7qYmH21r8W8qDmUyUKNGcO3m;(
zxcMs7slYOa)@BRJ3=ECW4-|%wjL-g+i3zosk9{WVLp5ody4_!2Hn<#d&lV_hAxc`?
z+hOzAmd`)*So0nFPfeJo81*h@Y5p>6LuQu0KJ%X!gz=G1Cdse#B0*evNi>^L`ia5*
zwzjmVn0g=v?*YQq_D$Vrq4uz`EDbK>+SGY^^0^~o?3u~sPrEj>g5#pw6$#hwvzx6!
z&2=&o-6Zb@tDNMLs9pLpoYw^*kRv^$!VEJ+*u}`pt%O~QsHiV=G|p~$eCE`amTJyX
zhk-*Smg{R|Ujae7>M9zsn_k$+n_#Sdul3)`GzvZRz5QNeug<5@iJ^2rf9V(tt*__j
zn+1Xr(vJxitka`!WTG@U%+Y$)9+c3uH~p?wf;Bw*{n9e^U;k6Q_`gNDlrnu8k*KN6
z$8=?;JzA)?91?vau%l6Y2a}^iB<sF<T^}2VKlA240zcp{Ja!qzPy}&&C3vynzx-go
z5%o^`Y9Q2oGJ*lo&qrCZQZ-U?Q*rszuwuGCj(Y><>R~bq_-PdP&`L{74f~-tORt-n
z_59Mjh|aotdTggg3lud;M)LGCo1_>Wg*bmNijp=80NZye^q2oxBzY+68TurW9LVMI
zE@)IJ@h!cU@OH&oMKT4xEJ~v1XT>tqGs^5#)Ned!$dVxl{jvLIas7d*cntc1lQM4k
z!1yyNMfl1W!#Jayo?EwVZb9&?trw12FAknO*}0o&#OJib*-Uo|+$3)f=ESE~Y)zC%
zaazLYIn2_3v)_kDYC-?|MRr^4<qfYj{D~6tkS$o<oInf-$J_MGQ-mPk6^7j-c*Uk_
zGB_WQzXf307&^aUwx<G@{@|y(+IXKiWJsi8d?5d$cu;~MpYV!f`Sw3mWby?X&^J93
z;*tY=*k^+mR>qR8IKX(hfe08K(>2K{adWJ1UY9wMdn(*FztP^v9yX9i(eQu|4@Xi#
zKEPkC_?_yOntaKFSH=1*pB}j)vL{@deNpKHJlWS9!mn21^toK4>z2c&wV9)@>z!(t
z%Gl}}TN{>-rmX&afb4YC*owtv8$cG&y8E$%Bss;>G>%Xmcr`GY&Kc_RER5_gCpZax
zg%$S1oEL4Cs>;C%%&kt1`|AA<&wHzcI!4IdqbfGBt!#ONl3bhNL&c`XHEToW@nSEW
z8;`GE8w`9SSnkTRD!shCYG;HMjSKStC6#)WyQuV$mn_&HE~sZpdyP2PU^ED1Moin2
z<q;~qak2UL9WrXM(rv-fW3ktWUgs@Hb8sMK`t$kzc%XMN0g?_}&|?cZ)kGfMpW;o<
zYxI0HG@Oryr0$}$&+<GFqc6VU#R^FF3gX<He@wS16wut`!l_oLYi92p-_6k<N)auW
zv8+p#ymr4O0iZyOW2)EMG_lH*M9*kw_Pm~*DV@Nm#5)&WCsA?*H>7pEUvpuqp{5s%
zddAfI0D`zEDYbw}!s?v^FHA~(Z<D#zUsM^WGdlS}!NYw*)dZR@^uXwe1wmHUp$vy^
zqf)yiaZY`)>KPM9oEh<|QBM0cd`rctOp0#_2~6Fp`Dj4tOdoS22`_T%b4lI5&cS=s
zoE#{kx6&9s@m_@9;&rY`J1@UKU`+Rk5oXL6ry(hJ_vKCR{myz`U6je{)ZzlwFF7qb
zLr$2@<0c)&3`<Sq8x#527aU!+T!ya;&#A>P9_o+Ivo!X78_gS^*l{^rdz|+{YKw+r
zlxs4cU_Rv~Q<Ai^3<;+N#r~Bxg&&#AR}*nOzz<PkMuD(8mtcMC>h6x7tT2<DQt7oc
z3MhT-V^lG~S<VEWCRl|!aNFo(7~oU*Ty}l=75)lO@FE?UePlseU9Fpy$AcY?mD_Qd
zrPfD}t*6>fA84;v)wpxo!^-@&YK~vpo}Zjwj=k`{f%=Q#TSpR}VMAXl0!BZTXvy~S
zTi@zPwI+gDlga|j_I`gqDipvDH8<lF$~RSONlO2M$hzl{Y^?^w2nm!q4_K&0Xyw07
z+Ysj0nDQ6&&4hP6G1N`S`{(56)b9@;x7^vL5e*?>w5mT$ly*52Bev;VAF6fe{=Osn
zl=pl-t$*6cg{-)UPkWF+)<pmjG*sxi&5VGa#vhQ{=g%~l@s(;R9YBYKz*v^0*x*!>
zt2)(rxNX5oO+?0WvoV8`l*^K6ThyDGQu_!2$!659q}(!+BjZ%sc_hTVbp|h}wppSk
zcwRNeNw7H|JhXFOd$>6R86^#okjS_eF<UQ%bCwOpazOPRkKXUOl25{Ui&|u~c{W5O
zWLHk{+Wnl+7L_SNg~-|WwGoNu-Lo@C_by3O9rB2NxsasR>XLA8yUC)H#y-zY80P--
zm0{6-OH5>r-Bw|vOV2ct@QulOrTflRC5mIbQSXpU9^bzlAdsO%83KP22K|>NNRegm
zS}{e+9Zf%+K`1j%rBOD0h7Qzv&|`0*J;|>^J8(s2#wA);K>X~H?KeK%>7UII7m0I6
zOAEq=lJuPI7P_;8az%TMBvgi+HG#aZ;3Boras=0$%UmJ2b*zHjyw+~OB~-1tbNXP*
zX5gKT?*(m+3{76!mdA?Zs({Oh@_7D}GZO2Q=XDob3pCx?QhmpHEw{m;Pig{#(dqXu
zDT%cf$Z{%A5kJVL{EkMe6>b_Ux7J6$U=wZKvRFK*^*Dx1dlOGIvNRJGVh~{dED}!j
z5L`nPC<<PTKzCwxi|bD}hD%p>>e>)k@B9C?VcueztbCScGgwjCZ&+Mmf+8@nO8s|o
zhUc-{HMTo-?rI})CwWnbln;CCrhjinWes7!?T<TY1uxhJpcv%eo@~d7t*C~qvt51L
z6ra=BxV+xmM2P7wL{Nv!u41S$YSh1~br5<Q-+QrfAhgBZc)Dn0bp1lvUM;+db^lm>
z#dHhqSaPRsu-J69iGJt8+)A1VLa3?{ll3c8=JKftAugrO8Ma2scOThev3G>Vr%_BL
z=f_0Vcs-z<{)ieIWYjx(SZN!cK;m1aKx}3@g+s(t0vRpP?&g<yTX4G|qQ3JxX(4UN
z+vvVWQu(4j9XMp~P#K}87uu)&w;v4WwY!HRJc;fKQ>3DPIXF#@p)agEdn-ha>z~;8
zF0}rz@0?Ru&70{g&cZ@*)>)$V(n_wew&y=8tPH|z6($+%p6g1>{z|SZ(5&@5kB^}q
zs6S9gm6~!=&IQ*pQ1CMP8{rIw#VE*(Bz_WRN29chrXq0_kNYR~;d6LJ3kF?QdrYq4
zW{n1Pyio4%!N<M78|8!ZpsvhobD9u^pw+T0d^Kl7Eu9H3l3ph%J<y=D^XM{E;ceWw
zj>8xG_34xeGceUmc;q6+Ai%fYtT)_V^+Ea6*NE<DUj2(ODt5klRr6!zhT4kWXo2Rk
zjaQ4s6_(rjGiO1m5KhY>+Izic6@&JxZMIiD#wSer)KD~DN0Z+|2;mEyr$eiv+Zr5(
zC#EzQV&sq+T4HCI181O*%Et|LP)LfD;Q!N$NHTrXDTsX<b8X3n$7OoM6UsU?{K{`k
zMn(qE)kjjHxWC>0$BmX8P#M@Q#Mo4UMn_zU69*}XDgK`Q$Le2r5z_*uLlNG{r0v<7
zb2C@!8X}DmutG<qx8vcw_p5&j4ts0RJRr!z2hAMmp!Kk?yt99od!+sS>v&qBtBX7`
zo@IIG#ZcE4-uIAOWK6M7M5bKUVXs9#q_uoq3$Hvqb$+txwg(%x-M2hGDHx|&KEUF}
zRRNteMWcfk8!(2!(|Q`qv$S*mxU&0r0`dUsko^c02v3DBP6F?%wwv;+Mcp<|IKK4U
zag{p0JUU{=<`r09f3UrG^Jq>L&{7|j-@yUf=?avw`-YdkL5{CC$MZ-JAB_nD5AJh|
zQbq4-VoJ*H^n0W4f%{BYhIr7%*vs~`Yr8RveX$A0bpupXpq*b@O%vr|(k*#=#i8yC
ztb&;mWUMt}Pd_nEr%thG0)JX5g~N@Qp;ALT(~;*FDT-tH>gGZ$;94NsASbh@Bm+^z
zMxI|EBTxyFt1V8khmA!|r!ti%rg+<{Y=)`Nmyc^5HZu$ZSB@fMH_jo=IRsmf;gw!+
z4nyIG*p7F~Q+HM=*<PvH?Z@O_ejfI_svGP7Ka_oST$I}yu2O=8C;|c^hyv0|cZWzR
zNT+l+44onfN=i2pN_Wp7Aky6}4Kj2y#9f1XpR@0|=bU}+KmKN%Z<w{-^{%zve4YVn
znV1BY(cJN_*9XpLh>@Hhl~zaRg0DZj9fV(ERLq>AV{h=759ijDo6ayAPV2R{v!c=0
z2`is^e9_K}nQ|ZP-aV*N+Su9#dRnl`!7JUmhposiHrXT*VziVyd{~wqUBk&U8b>)y
zJ$4{h^vwJ0B=&@VN-yv+RweY1&19ybwB<~@ztID4W9W%v4Bhl>w`S$oCUjJ_d*In#
zn0jPFBo!}Q@281+@i}gV?4_LL_^(d+9IoJfn+I2&HQ+M+#F)l)rqneX44cKWpB4^@
ztjd(EMj0MZl#7vlhmJKG;$}pQd8wINdejK2j;_2Vd?^%Nx!yYjQ)DV;GfyO}g^EHQ
z&6X}1#9~Lj0((bpt-<D<)_EJoiH-8=H`JU*8%l@=Ljt@C5@Fn1IAnS|hl{r?CuTVd
zLVDx2JIj&I^wCz=pubSdGQ18(9NR6@&{n^=OJ;e{$wK}RGR@a7uj{r`!@4=*z6hnG
z2$2swwI5NzJ5JZt3cc@?T(^IyX&vr)wE3&Yoi+s|+xaHH8n~}bJ9uy9evx1A=FZ92
zuIpj>HXo(X@M(!bwY~4rjY+b1@&W*xic%1rGd_`1!+fai)Be58z9GT#A|}ns_lMf8
zID~g@EoI0Q0fgK0zRixvlZMXKMC({p9q-8*Q-?#Fu@b3vIW8TeBG*B$<cF%dLOtjF
zcE4~)_=PtcRx0oK6T;3;`DzZ|X<FI_loh~k!Dm%}BnWC$?Dvi=7O){wYwXv_kJ<w%
zyf?Kib9(s@jy0#jAt4)=AdD5WbHu+0v8>&Zc+?Xa!opwoJkb}MU3lJjkK}0Ifdx@8
zKxSnirAyr$%GY6216gjkjCoOYeyb!zQpYhASx3KVzQv<fxqYi<Z}Y}d4Y6)CF!pMl
z9Oo+&i(&`m1$7j=x|{SQw13RltG=8}w5gSZ9q;nh%qKi^9z2?H6FI~>{-AB{r&Gwj
zE_4FVqv~F7@GUVY3{CW|l5h@3#}t2wEW9=Sw(W&DQxHVVletg#r;D*4cWJV7+n!C|
zY9mm9{sgl+IQ2f~h((8bxfj`ltjNW77T(|Bs-MklrD=N;c<3;Bb{xfG*wwudqSZ>u
zna?8i!0QVZ45m2HrBpPNk&1icf&E+UOho_IPt@*;O`4N-v<&wU45?X%6*amw8=5Un
z-U9E>rtSdUBSAY$?2D||!b!2;49cYbgWnc$?vrr+cj*23p=)ait`A%5cOSE`%wjmb
zJ8`m7HpF)8B1OrYt;o)NJUJ_&s3dm{ySWgpLjTPo`q@^DILuSyj}A>hZ4<}ISO1X8
zvC1D^)#a8HDoc4!`@$J$tXX%-sn{|_SHoEyRZZBO0--fcW&jZM!Bkk(ZfR{rr!>!r
zJ3{cr;gdr1_v`!EW3$<p)(9bZpV-v35V0=e1RK}^47zjJD3#it<_k-drsg!ImvvjF
zK3Sfmuv9+J;CO8-!EGgxC;zDTDSciy!O#p#DlPy1+wR<*EE37ZcB!q95#~N-P`&+J
zuXALhGAn_vlvDi6xPD?k40*7I#$Ln5jdSRjK9c)Ho679A2VWYUi6o@)u^gga>HPUw
z-aZ!rRq)z+G+u5NLM|xHMvKOvA%%L)j}R9xj_pYIWE-j~p85rkQ6`o-_H&pJ^m(7J
zl7c@im?uq`pd+Nu61(=|M+t;?_<V*U?)dH5S*J~l;B>ya0P}rUr@o(f-BX!dxWtK+
z!7n9RDV#%06k}a`8=*%+RHrvp3TrNmU)_NE-cw_@T`Ff|!O^l@89t2F&FPeG4z(L~
zM+^nHJ=%S%)XQ|oW8O0D5(9Q|=<d2+VIub<yieq*yeG(nbfGBk>@|{=7f&0q>=KcW
zY;`$EbyUrsuWG~P*DE}KbgAz~)Og)6{=BVY@#6#87O~bGXR(2(|CXO8#+o@U_u$V^
z9?%XJ6*f%^%Bvl@jt)zX8>m?w$4Eqk7+|TVQI1<2FOAQ3v)Ik*qv|~E2uQWkMFQk)
z12D|^xz--~F$9Yl{K5J05RDz5vgNEG?!j4a3%+K_z3XW6$i7Y&9s>A{-}d*jJ8{q_
z_1Cb;i5@?7kSNw3;?>|vniFl>;8PQ#W*U4wx039T;5u=PhY0v=9BF!ws^v6S*<x`?
zb+%G2#bKH5VX7C?R|V}<09(9ZT{EddZkBQ^Crx*D@Iq->$9{*${_y1UB1V}m)Xgzg
z!+2^XvSZ41Z|T?ikD6iY>dMvbM-Zb=G}#y2-Q>*QCIdpT$B)ud4&Nnn3GC2Ek&e^3
zDai#g_OrUYoKte?aYG)C7ig7<=~U&033H_S%1N*!n8C13_4yRr+?(-H{uYnJZ6)!*
zcYT(GZc4f#*&jz^W1<xJRq%xJ8?f$tWtzR*E+b=4U*~w4-g+G|?&d3$^fTBX5U)hx
zo<;c2BPH)j5;n(bd{1pXzk4ZOXI>4xblvyu2G&;UxVqR~Y_`)b4ka5lKO2AQvpS$j
z#Ur+#9*#YADrmeh^D~$r5I5Q%v0qBJ(r~hQ!+ilBn3{cJMZoRQ`&McSz#FL9{rF1T
z<_pCuZq5V96<oW=nJZ!#l4zC6Q}xD0eXkE!Qk)Xbiq;>LT1|4gu0?Y#N$d?|G*Z@H
z&Ri4nbVGDGV6*9)a)yiDc)*y<=N!Jqc3DjFV*khd+Zz<v;|JVqi?JBNraVXR!Nq;0
zchUlE>|I4Paf+S?U+ueXcg6^MI|>#GvB&Wk@SmeyUm@MlDl@Q?mDiP@Em<906GMH%
zj@e1a%NU`=?lSix3fhQO*f)0~&wXJ!ZV@)Um|C-Esdu0{>tPd!%X6L|vRDzDQr=9!
zGwD9d;UZd;nv>*`1chZK)-kwmmgE(%E*uG+`!TSG=O@-+aI%W|pB)_Cukk!Xb59<H
z>3E)#M}1^*DV}BPh3Zf8t$&Ehv`L+zxmRC8zWeKk9i%$-A>>2DEMCa`bd}y9QoSkU
z6Fu?ZbPDmkIkthBH%DkX|8R>yJq-Sre3SslHTw12#j}6d0x#)(v-0+ub7#F%w{1$z
zrfpFefkNk#<0wkol$!bgYLjbrOJA`4hweBSEaO5>c3K~-$;TRrrrKWY6^(Poi4CT+
zurD;bcZN5jF1?p#8Jm+SWECMG*gD!st&;z_j|>wq*B%wxId{20L_W}1*PPFU3!Lvv
zjvh*Lor@~_tEm*}7^J2-Z~{anD9p=@zxcATM<1!aq3G;tvNDYCZrI&JDsS>qgv)5+
zY|FGT<g_nQ(XHU7%Kne>&RzlIcdTEnRkvyog##I`Y&_#=T-*|YgWJY0S8GQ%o$E(|
z3#Xsgb?m9qb`85AEmi3=GszsgN#IxRi|R*9KYgVp6LsAl=?20`-G=n4LcMyrSyTBm
z9KQr3HeuuBelL_ND69ttDePBY)}5c<5@43XBY9l6InX{bN@h6o)4zXIo0&L?YpQUN
zg7l^R)HA6zk@HYTRGg`3KAMXZfmTCvw>Is3$(cD0b~UlUJYJ(?bnE0&`Bi*68|dK&
z_L;8{lW#W-U?dRPvgfA#WXUt&%zt3#_h`tZNdV@M?sPQaz{PX&)?jJBj2VY_KX9bq
zn@^tJlkas1$>w)DH%r7;Rn>i`DKl2jcUU7XTc{)^?`~po$i6QFC3>%o|I}W_-`cB$
z=(`rYvwXUYBR%9fYwe}}cE)90fa_bHL7?;Q4+e#rbq-9+)mMR?(m5%#q{nH#=DiL3
zwKzgd$1mTydAe_8(fZ+0XPle%_vXAmnx#WMJt)~b1;m(@rByW}km;f#?d;KSFRv46
zyy?PKZ97>>db9p3j5(Qa!YB0Hox(g;&g=5?9aP4lrBgDSsGO7oqib-R!<F}eTP3Lw
z>t@-1_@KywNo3|$lr~Z9H@luNI;Ryb%eM^Ay-&rBO$B7uvu_~xU%qOU9ic5}H!2i7
zm^ez)DTP!CoSU0n`_9c~Tpz<G&nM!TSW47AU^@yW*jT<5N5<VF$8F>8HYnl$7>|4V
zDdO$82=rDlS5dtaN)BYUAcWjrJZ$Oio}r_}UEHKE%Or#9fjV}<Tvnxmaz_z0*N<CM
zy~l^$_BO|fbt(o=Uvnx{@R$P<%vluPMgJJJnJ3Pt_NSrceFcyyRsAR_@@G#B0Sm`x
zW_o*Vp(lFJzusOUnkl#E9b^=eKXDkD?J<lvN)b4W0#I^QZT51*7u)9xOW^b=U-wEB
z90cqWuMb#YaO^uni4KtIXFqTYZAm9;qzwRm(&VBz+xGw>&pQ)X#cJg)@5wyU8c`Y#
zb^V_|aMi0%@{oM<GjOmH*2^yPNWNxn;*}SGtG9=qWc!lc5@2mtf1Q0~QCL65Qr%=G
zF4T~GEd(8#m?uQ;rv&Vbe}jv!2qddHC$v{&O@HJCQrx#uTHxb)mF#p{DB>^wRfVZL
z-f=chwU_<trDv+crYH^mbG<*+><d!OBH#`S+1(k<2-gtf0yV=6Nf$^*tF=#bVZ)i?
zrEtmUtjFjyT%C<w@+ZE$8O`9N0nK*Wp!KvBRM$I^gcXoF?d|gt=6~Su)jqwcoWN<N
zk2=qO558`b^DftT@(4qfn~>v4K`Kedc2qB{G!<)|%M#~=D*{9H!vpug#8(DESo4Ev
zx3*?c<?H^6(e_dqsZjJL2@8Lq3kPFrIkmK~E;aXd5q#;~=uYnya%;|Pw$cw1<l<83
zf$Duq9-(cL)K)ycYj5T{lCMp>@}YXlI$MG0h;K11TyB-XIw3>5O!PzbF@55uRw^r=
zvNgGT!BCMZ08%>B*Oc(*uC&^txE!zB*z!L1BionJqILORH^b*S^-L@iw4cvNY&u$}
zY+*TTZ|@Tvg$P!x+?C@K(%ar~R%zgO4$fvBxfU|p6;-b>q@y%w!E+I}YjmUJ<2EWd
z$AdD2b4}E8zFA+Gl)geEf?lkJ6S^f2k{KMVk7DOjF=?_xVvS5p>IQuYc*_GF2%gy{
zhgxji?e?3yyW&=~mmvLc2)7XGoEbmr#%6LsN1Jf>P~+khL4ha~VsCz_s2It&m5hn=
zc&8B2Ou(MOwJI8Y-PkmA7efsWn`9PAw3(A0Zx~<`ANC6fnZ6b|13kQT;^+j}g|Y&l
zd6`K4pthj}XpXifPI1w{;z!%J+2lV7biv;4+H>iFNcp^uFxwROb<>s&_(s(fCDiQ{
zg}OZDf_x4KcewbCr`cewQC~NQi>CL{rAZGzLH-=}@lBNygjYa=$qh~2snVn=EjvNo
zqz?j(zqaZWM)g8AN-LMkWM#o+h0BOd=;*aj&9`CuTtmS>XRB_Vs9%`lPhS~%z9_4w
zZBOpx=6@@>s<J{JH0H_*F}Rv9di0+iHSn(u@Yd@RmRoa6AVQt{pAG1`ZQkbD4u7-}
z8$g>XC_3YfkVurCOToI;33@EYv|T*rr=8<MO5tIGOHObslL<;Z+@b2FY>O9`%YJ)-
ziNbGttuDm^abqE~9`d5o+#PLg+cG?tKbsY5xjqE1KesuK{&c4S0lUN|(JOH@dlPfh
zhSzyx@Z^qaj?)Lz9C)RL313a$)DWnm`iQk%J=tS{FI9E9G4~otXfv-RY_q=})pUB(
zX;5e&@F#!#{f`%pNhiAAXZ7E;nK$3L7XQSu@oy|8T08ypX<Kle*9r3hJlkrun}Iw-
zyi8)1pu^2TA+|uR=7s8H2FZQF^CpNxmpT#LbHWDJE1<oC!;~>d&R02Dl_r78<-9h~
zUIbIJ`w{BHn|y@VJ(mxe;RC(GOHAq3dLy~YgH{@;h0Yxm-WMCP!Fwl*%$RXw7QNH{
z7;38}VRbv^HFMdM8$`eIS<dT6^C>*LW&90LQnhZGrl_Ock)-Swq<F#mj2EZ>^Q{@`
z1nMTukwVRt9`nOQ?K&c3276U_oNeg>qgJ^Z2naoPG9SH}PxVo@vU)yPXVNh)Vk_t3
zvCzeS9_Y3db)-nD44E8JA5GU@UAHavw6AoSN&X<+Q*X>fRo&G87w}fl!SEI`nn{ms
z2mfsjJ&qF|d^d-#=5e;?OrZD51@$vGMgf_8Kg{5j^K@1jzy5%CDrAoEaCyi`H8e&^
zfp-LtEaN;?;A~N!P5y3GSfZ@P8*of3W(t`Mpsw4JQWKI}FVsjA05e0(O790UA4MVS
zHU7b0>p0W$3+KLM=pk9|8NXF+;=ozo^&5e?#<5(cFNGFQFJl-TMmeyL6?rBz<(*%+
zxz)1KP1h0`&vRd9FigP-xf0W&=B>4Aai4;|GP{T2F)`-;45g(GT?B{{S5{@jK!&H(
z=kExPFY0B<Ap&rKXHlKXkvE!OBH~LX>yujb*5UKU><RX?+%k*WqPzwE3p)@xGZsN*
zT)yC&wa%Ul>_|%|1YO6nOHA$USqNyOdyi9f_<S(IPlEg&9+PDHkE9&o?}5&zpnAU0
zV$)f*STq`aKT@+(%TPUj4=3CRi#psi^%f~(R^#YQMm!zx-fW21yw^XHaqSjX4jR>5
z{CdYAr>Di;h%<x7Wi`Y2xoeiCu@vsI9?hQoDe1Lhv|<5;-A?pw$DV@iG($9L^gk<W
z?!2Ppoi(J+xZC<ibkz^j(LA2Un~CxMO{$?c0*HPUWCY96O!=$`dm?58H^+3ggueB@
zXSFPs!nok>G28A0S=5FMVx;RPm(CX1#OHEEhRBls=AT#q&3;L4KhSq4_h&b<ZwcAv
z=>GZ{IWd^XQ3Tz8OKvXKP%q%v_S6*wU=1%f5qIm>z9Y7tIG?MpXw%y$Tzt#k81|&M
zP%nM-O*HWJSCxXNG69uUPV1bL`CfDzOsDAkUdz%A6x{?Px?bl{ZH_cuiZutlTQ(fo
zz~t>gXO;aqbhx2jYJ~s_7;!raIt(i}omDPj)TtL2H8On}21=i*8TmNOO>@@chW6z9
z--t8wU&@+Pw$&Lw7_e|PmP{|gvzh;Nd7=4>E9rgNl<N_OF?Z)a^mBWCiQ9n!jhbx$
z;G_w-|L#%J&s#Qi5pjpc1>1Nzck(yl8!kyKOX}cX9^nyqt`*bIdL4cCxNPn<zL&2s
zeGn$Flb^3vqWB=uC9LiP78+Ouuw~5?ak!KDPHu2lam7DKai8NLdy*gTrQ_~Uio<vK
zq+xIkdhsimIMs8=1`7EB7QJ-2JHbsgfWs4pb?4Hsb({RO?ns*U{v*H!|EyVYn4^)j
zB1>f>!TyIj@%Xx}8g9(zNM4WgjiX8yC{ZZ%fSH8vG}peoE&C7e1sTqK?d3ba75DmL
zVFB|GPlr-CO*4nx6+t%Jz3ZbaVO->8=s)G9lN>HW1Uw|CCZ-7pW9P?F-8VL8eAjPJ
zRADfz;@{6=YPjH3k{4RnPxKV6f^Pnj*}$hLyP&_{s}sBalw5vCp09J?Y|13XL-^eT
zp(#8>dOYK1t>=d%uU}7dUIai{G|o~t>I?!i*4}bdi_84%dsqwKZbZ#5leefb+mD7}
zc~SWYXWjntN<hr<_%n#}zVOF1e1N&~t%3!NN14&PH{}u3eS>1Z&*?e733u2gPypH}
zvCrhbM76Ivm%HKNgd${^CAO+;@IWQw-gsKl%-i{<KfenZdln-zU=yDXYUPDi9G^t!
zEv42pR^3wJQtZ^Iu(S@`?W6p`(oL|Y+0{^@Csvcsk!qIeePX3H?YuVCZ<zS&iAkz!
z@VU)mwCZeRzk#Q3*s0?QlR*)~vte&F@>f{C+=O3lrrzR(x7^!<G}AS+6q|?y)|1(f
z34hfQ7!&|FtHTaCWSsfBzt)TD?63$(hu~<cro(1RLeaIxnBTu;knFSfF-;gX2~>><
zm-(NPo7K4up%rbFOfqLQ!vD$=n42rSwri!LOda~0^uibI4!3QQBNmq?+&gb~5fnTa
zHCZGs6`|KeMrrEl@)mSDCXZm{n5H7v(_Kk0lE>llV10L9rig>9;LM-uj2<l>06)ZL
zUcPCqX(Wu*@hDu)Gf8b6K>U2@Rm9c<8?zG0K^<M^<l{19w1J+MYi{l4wFbdnyfE3c
z#?N{muxw_|WHnJvKhQjo)wuyYi4ehn6}fm1;N?9BS<<PE1Da{jfA)}vhuTEWj!PdH
zH~P#kEu@?ikU?lRM=5rQD54Km&h05x*5G1^m<G{r3`I1CRL`A?I+tDnh(pdnc8t>g
z`+2KL(Odml>hIZQT0j~jSuRs1Hgt1;@kp?=xv3hPpMb*atj46rdmxdzDY3t?A*PLw
zrc5zE1S7*#g__B2VW&x^ka>4JH&y2JBwyrSV{k&DZbKu6(p><<sr=P~H7Lpb(9BDr
zGWHW)bJ2BU{4Z}&dN_Fdg(#4DUwGZZX*!7uphhrjD9aMLC*Te@iCsH=DPpSeX80G!
z;c1<v>l!L0(kIzC>KatINvc(1KWq*;rmC{POCx`ubEq;`#5hdIqiC`j{ZUb{^rdy+
zUtKwT6HwdueH`lz<oEszsuceC;W5cT*|u4}?djk{VR@KqfuoR3wC9>*b+aG$h6Qp#
zJ2V{FqyxLT5ofP9aX&q{ahx)6KA3qvoECV4R*+=(M+}7GU~M$fx~5Zm{~NSw98S3D
z$11zHLOhl(A&A(<MPwFmrgJsc8??rzkK0>iG?IH?oK#tTmdlGNH5)WOIS*531yTIn
z=CFpg#{`BM)K9Z^Q3+DRLG0debnTWW)tGy;O`&V8dTZ|Jw@7er)f~iuLeJwnePn}f
zduky%VxP5>1j`j@`NYZ*AF2+$S;ZKpJv960sM!<4Lmjq0dDc2P(NGWK%-Wxoz;c2(
ze16trv)npWKa6ZI-EP>tAp=7GJ9Uy4xo)1j8l51Z7FoCDK4vu;-}mjigX9N$<B<BT
zC9LDV^~=#&StiAm0?15kPO5uN+pZd8--kOrwko98?7u(D=@1K(jl3mz+I4?RoW$k$
z>B(G<*+Ym?lYf|W&h$~Ne4SIQ`<C}(Qz(bUXl*BC2d$tQJ6@R57cIAA#sA`g;ORt@
z#Vk2QfU>jusql<Fyzk*AygetZ*!S+HA2~sSr)gCo6n+3%-7!HcWVVCV(8js$L}p3+
zTHhL1)#j^g+cjmmr_VZ~%|6dql{nZ_{|$#&*@ekb6=|9L8Dx6>?W|B*n(_KOOzHf+
z_WnQ8P%+=I0uA2@$UaxYefLTl@j>Ebf3K*UAa68|<!oeZ6PUaY75F{E9H%WHt_KH@
zF1frs`0g{VuqrOi2MM-y<<`@zi-l(9J9Y+zF=Ricm{|n(f~0d`VVaP+Bwgz5wUGvs
zUJfps&?=`H?-hxiYy$}A5SgLI;go^u$-e*S(gc}x1n_Tl(*dYJJei?P>`CcsNBCF|
zhH{eaFV`OCx-Eb>wkp3a_r<R2boT!1BdPA{{a7u>(1N4UP|Jsaul8m__Ouna4Dr9%
z`eGdX8p}BxEtoa^;Zj6B4AIJ?f*NRccR$<^3pphDn1PIZ7QOU@mmru~({xi`=6!|%
zLgJK2*FECrm;Sfzl3y&W=z2xP>biI0;_;uhHG)vCP{a8>u4U5x^rHPWvu<{~A-kOR
zbX2?;XqQPxLdE)$KorTsM!V9K)w@)$488&`{8X>A>m(jS($4W1_dy_0C_};8qBsQO
zLH*81o%OjMK|+=(GBOiqx3KZcK!IOLo@b4F%_s#vq6h7T9ZTxC8`u}i8I<OKJdyQo
z^w6Prwtsgr6{B>uQ<J?-JQZh@?2E6&;0&KHVXe;K!O><-L%GGI#=db*b}MAyXdRzH
zj&$qp7}_4@<0N4!Pr<to`@`|NJRDM&AntM3eTT(OK5Kg};K@ER`Zc5rcXFG{Xt5u=
zP1_M6b)SCvXnN8ZZNQ=%1lcx68usTi8f5Ai#AE@fi0FRShWA-JvHe6mamQRBE<UvK
z{9;9ScYwjBW1qzP?HF%uND)jw^Xv^uX~R>or)w+l@kh*2s=6Mc1BxeZJhi*!e{!8g
zYx`^mcX(lhYzj48C+2wb2WTS}OCL-<H;FZ^mfwU`>3fE$?u1EsFa{6OHU}gMot!~U
zYbJ{UD{V`$p~VVkzgWp(83-@;PKj8ZN3))nQD>$wv}uIT1?&72+hYD4txP9Tcaq%s
zi~|3K>6(eP7y67i7H44{B!3x{dDbfQpK`3m_me4~wirsp22yx-Y>EyhC$@<|-Q!r^
zJ4oatr(r`&9Or`AjCA)}%4RHU&kGSo^N#{1L6S_p#_H^R`Nx#|;d}Td!ii0n6B1?9
zfatHR1JyF16h15CR@4RP8@T7Yzi*E0=W~zpI_Vf5s8rt5H;?s<t}Kv13<!Q!-mI;9
zS||TSaQ{^2)peqyDIOK(I{R{{Z#GvyN++{6@B0y~F6+!jmZJLYxBF!5LxKYL8GXK)
zt)Jx3oZ?Pqb{=5NLl;`xvs9<_^430zuNY^PCC}&$<&1EsJyPS@f)6hBs+Ua*2e0V3
zDDN*ZEm+rUY*r(#bE$o=W1bj1J=y!wsW5(l>dzoAxT97xUy9v@+qC7k`Ob@3*J}-Z
z{tVGCD|dR6@wCQeear%tm@Q0L^ycG<6)`*7wTV1tyQ25V#9@+|-_X8WzC=+gh>%IL
z+p78Fuw&{sp@y}ix#mBQI)wyrEl5P%Y+=0TzM#xmEShIZsr|E9j&v$Z1XtH{&+}P<
zxK}h@)b2Ja<i+fDsqt&r-EY}P3{MBw1^!7YL0K;RbM_$JqBOe{Q+!Ka%ILP9&~JAH
zRI$e1*HrR4;nTd(h$B<>WiC*ZKm<$n+kMpa-qVhHcSfjQ4S*8^1S9z8HWQ5?y=n(N
zFK{zl_M&q@)Ty^)wFENOLLg`B3w!r~V+=#n)ELXA|C`UZpMXI6xdMw?aog0;>0`L@
z<*9pU0bl17d>~z+iH;ngSK843Is~yDqHg|l-IQMz%jGN^rJ$&oi(B+)9au5dB%=Em
zgOf!<kC4U<*GJd@aKj~=<po>uvUbm8iY7VnHU~Rldmx%LubBn_#!6ezURlkiP+u}>
zKLqrep{sbv5=2SKT2oy@?MPIVhR7(TrWtCee9UW<$vupO1D@g>kEbU_HC|m-tQ(^w
zUUIm=;;&iGjj;9sY;$abp!8;UGm)iDfZ&0%|NVNW{p^&PsdYdb11aJ2oG(s~G(>r_
z$=+yxdbhM{N>6by)yfZ#rDlrQXAVRj*JH&NY{(wTC2~g4rn)t<1c|2B!*)(Pmx{#@
z<JWFX)7ajNyBk(E`dzikP<1O-=__4h!^o-`tNE%0s81>FGBkampb$&647s)<y>2yG
z$rGsbI7kBZspF(w)r|Jy)auRwUv-t1`*4<_u|0bHmzyZ>cTd7&R6O)jb+{DxmRwwa
zoOz~8uubq^Y(qEc<Yi={S}AIz9=NPxx$c@>uBWODTq}AN`ym{}z93b0G$=v@NlU(4
zsEDIzx~=-hqoJ^@=ao0fvlr&Edn?7Z(XDg#W-M|cyv2o(WXOa&R9=R`&9eYdM)6<n
zKjLf?o%HB3EL@-dT-W{f?%0jJf4Vn<*Y{}IsT<KEtsBlm{dONH`J!<~ynd`*=~_Ib
zigI~7o^WPp;u0}?X$h6x3n2SlZ(fTsvu?vJ5dC@<I&y>ExC*=aVx85-{Ty+U(|~76
z^~Tt<m5C<6ImMMQ&Th~F&|IlX3aT7S%nqs=euulPFt<lLY-+KPRW3?p_40B=E|oB>
zOY*x12(}z_?G<&<rh2Va&Ek@AhKI0l%k=Q<y(XjlS;gAc6`9q!xFg+5jP+u5xU3Ge
zUfA+v+H6s35;mB|klguv)Ck*&Z@925b)7FT*daqx&udOH^$}$c!HD8N?bEKr_Vhbd
zU-)V;a(ObzC1P`Lt&f7YKA0dVo$SF7;@YGnAg`L%J~x`c;aPq==fySYxHQYwKp-G=
z7wwM7&;6ar>E<|F`N~z?prIjl2Lhh<dQc{zF{@N)on<d;IWu2}b-%MA3|ZLTI8RLz
zB?lWSRA$phx`d#i)!1Q=R$U{JY`VWo=M~=W&2h)(A9AJNafBkwQ=VJT<bQQ3dl+3V
zHd$p8Yt*Sav2G2}9}G<hJm#S4DpQBNnXRBH%&<G5LJ{!Sn2k8D6jwP-bho!hfH+Rs
zIhXa+b7xku<kyDk`iIn$@F)QCm`IGZX?wUa&{-5qESKxaba{62nVhIqeZ_N>vbp7w
ziDNvma^f0px4eW%8n>?%O^~ciH)(u}GQ$zCAVr6q+9rm*!=7PQyK#x}1}_=@Xojm2
zS(ozWI|1BP4liG<$D_SBpxI_VH-JLW=LqKuQp{_cbljk*t@|QUHuFuru4zh&zSOdn
zh<Z8=^gc|!`6h0*e#5ed5twZFl|C0Abp|*X;XmljTDD-Rzcy>27Z`r<phuWm@H!Pf
z5eyhjS5DxhiZ~4lAMWS=jP$lQmf`@a@2XqpInH(R%7snr^|miJxWxI}L;-TthA9p%
z>j^aM%#UH8Zahli9=#{tFG3Z)0EfeU$$_}_qNnO)Jr!TODOkmHOeq_=W-lnmmoW@L
zDPB@#=iZD<&bxuH28dnuSM!y;XMsv81sVplpysAL4PZDark<U<A8KhTR6#eE*WrWh
z&b}+Hpq%y-M8{j@U8?*0L}hU$CV)sOqJPJio4D^AI5VANT9zd%FvQR5o=PRett%dM
zm7-s>p03n(RKxGqN^c~b;A|wkxWsfE3*)Wuq+2~7{lp&|5%CZSZd;=tPI{VIYTVqX
zsva@}XRx?5+9<2ua8!#_)p4Ixsj*x<u^(O{Cj5HQtS|_&*`5<x=VFBCRQaH`H(jOd
zzAruXw-x`(wGmm#gmx+2etNf$TT@@f$`f+5|BxVXGfc6^0&O#<u{K&YKoNpk$RNdU
zTiTAMuR<)LoAsITz>gqY;*4g$4<tH#pf<XL>$LV2tI1}A{o-6;$1^Dk=F{(%t}mu}
zD<lRt%wOzt&3NfPU~3KpMQhcp9Pwt>!sg+=*+NCS4c74!ed%&A#4eQ8N;BqTiU(Y<
z=HkaB$czUDOx4G@25Yx!UF@vrE}c2Z--@r)A?tgn;&pOiaAI}31z?+*5b=qsz>EK)
zg+eNuu1*sZ*M&)XuSBo^4xr?$9OC-gIG7i3v~Q{69Gb6Pr2}YnO>gN0qMMBv()uyb
zj|v66sH#Gy^U3tm#lD;gxQD{aAgW~5)n<b<C%+QE?gEzE>Q<-k;bPNfGyuNA5cFca
z;LO1I)3pE*pitpp*J9sD7ZjY^kuEEr9w8wKnuxaX`fvA0QV*L1g+~%DSOtZD6M$`1
zbni|{X`&LxnqPdnKD-oZAf2jB&rT1@ZnN$@-+w5IqVVHp)Dp7%c5|MTX2Y|abeMXO
zUpcRlsK@p$?<YSDWlOfTE-ssKdQE2>n+!6}r`JtBl0B0R4`~yBjsC#nSegW~K23dS
zH`zVfxjiV)Sn_}n|H((;e8q37-|-*%;&}+71wRb~j?r79py3s<Re_)|4xHO}C{}!-
zVAaR4OlcI}J-+iMD17WZ@iXlo`Vds&gCU6{NGYas>X;~d;heaLxMf3RZxbn(TlP?Z
zZRxET>*tsXD#Mlhs|^^G3mzZh5PJd&WJLgWu_-ETeIr4OQz3~TXx<SU37!vEBewI!
z?iErxeK4f7S-pNe{sfU)>NtM0Dhx1<B+4qew-+yjMxxekEijvapvDUp(6zDFJqBL>
z5n@aGg}GXsgr@p<`NTLV{5yZqeZt=>Vnk#*%cyM4T@a(?>>+nz?iM4PnEa@Wc&Tk*
z_@va#0%ldE;>2d`-fRfqcI9fFUPPfDy@(|xq3Q6uYH$q?M_F4N2EaO!1BN*mrDl6X
z<E2h@mc1R3Nl9A$U4$C!Q4$&&xrUjEy>EY!DCQaC_OlQG6xO%jfXUj<rXXlqH>1y*
zo3B-&kpPnGV-E}I^@Y|5d^?!I6pV!Y;h^6Ck2JGzIT~hRP>f|um5h%>ODgIivO)r^
zLiOD~pZ6luY;Xo<NO&?*oA85bd@y8(o{cl^gX6dqbLgqqMETeL(o#4x;=ORz-Jg5u
ziYc7V|LilTe1QUk6TUJ}GG@^7X?;(BcDESy&!7E=SBwl^{w3oL-bU+ud+a7Z@~nam
zivcu;G^J?Z;rv77zeo3P-cqWd(BZ_AGW@!^9tdEsKIa(xxk~@>E7yF*@t;de$7DV(
zReju4!?<}5uo}n*I{nLd{_WelkA!U{o^fMU244$7<4eDf_ph_~f4?PciInt5hx&a{
zABp-TtV3iD^Iuy#|7Eag1DK$c+Du;gyYr`OAwjPO@c&RS;_c9m7~PkL&SO`dGnByC
zu_!;9AfMJT`**Z_l!a(r#A0Zw--k)wkN1FHWj1r(t2Gs-jPuz#)G~H^^SWiZAYibu
zQjY}r6T&T6E=BzTqnoP%8jm+jYtwvs-v#rn?|(q<TH2pldC3+K7tQE`n8Cn7lW~nV
z3|C5{hRHo@zA`}`BDi)nIWVmD_I6|eue8?GXKeDE0>C|&L;G&}>bIY3T|ZA|pliQ;
zRPEK>L}ns0N+eHPh$4jq9egc+eT!qlePiTvzlD3|Wug4cj_aHUohHxlm<9MLfE)h%
zm%2YYpW2>&FxD;qh*YGPg4b6xHni$GM@coWGMle6L1S!Y+_&Jx#Z>`IrHEFk**UD^
z*=Z~rGV?!3Cxp?^>bd3;INBhZNoAR~6WVRajyxY>ElePR8H=i;ga_^9P8WK;N1<|e
z+VM=Wk;_J@kJ3{yQR6I?KSznn<H_xb@wJKSN(S7wZ3~BcD){7`L#D4*7gP1ycg5A0
zl&2suQ4T}B34wf3-9a`wpYiD3x>A7LG@m`0xNQ7E+LXx+t$VMZd0Njo5@ybY1OLjm
znrK_+^TEVMr^cMSd1Fnb)b6$Yub50HCmx*Q1JTI-AYaozgz-@@i8EO+Niccoch3KN
zs*gi~bK&$(RUrI|SJQ_<uc1hR)%PyU=O)ydPt5n}{h++pOSFo4_1CWb*4Wo^K7bI1
z#e%A?qJsb`kmK*)r5VHG&C(Ow!xZG7U#uk7iWCn1btvaLhen7_*}+7Cw=Lr0+abF8
zSoH0)IS;Hv6BI`Fil^RPQuwji$njy`T;<Rx8rDd%fS@I^415~Ak$OwQR1LZHzefw_
z3sKDf`sojsHP2=7(NbNT<&?O9J=6U_s9yOwITB?;%!=5M3!X7Ba%&i`^UCAQO}PfK
z&Yj^|3A%bq|MDB6HSyxjDq`WOI%^(vyCAv!lmOyqjus>VM5Vw0j`6p35)TpzMy+q-
z7TXrCz|Qw|);sG~CY&ciSssGsj*iLulUMp8F8!@Wa-C!Hh|lX2??gXi>aG5REGOiG
zE25NKcmxf34gR^+-y(t9%5g2OmJk#p_{@A7oF}W2OZC~Tka_d6<%9R$QYXpO^F5Ls
z0-!QqT&_cD)+<1XV)uwemCb#lm0tRliOr$lE_aOtmK{B7od;q8;_p7GAO{?D)ASbv
zDhYT+#&2&>iW|u5dI-L~rvD77;*BVT2PZEZ1Lcl>uh`>ypFKIfCW<FFe)Ba8^JA?N
zn83&!PB+O03)ICTX6g*<>Q44sd7QuCL#`8CeT$GPP0W9l1B|9%#G;Y_m>mLZj^!z8
zBb9j@n5Av)OgKLr4ncbsaxlS!wuRlu=bZCF3Zx`5n*#iIn&s6nDxpSKTOUjiQ<c)$
z`yilBYUq!4#YXh{uM18Zx^#4>cO8T8!)!v?^K|2C=6$O$OP_Ch<K2G%)*sv{s;d=a
zbZ#F<EGm%Nw4v3Sd@y5CU&eT6YdAMcI=IFx7WCST^8_FW-dCDkohxZ@6gRmMi!!h=
zG|1PRR>bmK)@~t(ppcd!MA0-U^N+o)4(eT=gpf}D5p2DEU5<LIQzzonCrg{5tb}KY
zVe2K@xvC7s0+#Lc;gh;O#?|3LE`yjzIrCm8ClW7ebYmLH7&V4Uu2)PD1EbjPls0<H
z&6+`CN6O7Dp*s?Aw=0B8+)Xw^-g}s;!Zs;u9O`WM);MZ2fUpRu6&jG+#n#v5!~+us
zVf<akb<KD8<}t<|lMRC<g|aWG-km&5J`zDiNLC1uN?l#GAAP)T=QaB;u2_Tzu!Kl>
zd6lx;d}7Htn5OCVNTJg#JIGER8`^IGbZKHoa?xOGhJENJ$sh)5mWynQ^-r}o`q6O0
z?28m`jq4*_y(E^sa2D(co#q38j+oJSpJZ*VRDhv#rmvH}Ev9^EEc<iV;j;lj<ittP
zK!=OOAk1R<CiG&xf+t4IAB}Y91pi4BKl0;eC_9HIFQ47q|ASL{9?1~6vf|PanA-U%
ztYLdnCYfTWA?&+mTEGR0y557$L6$@V_3~MdK|AvoFG{9~;&@%Nt)&lwt=>i>W%Cw|
zJ)mAEc3-;zf7oo#+X0slF#|IzOk&f^t3T9!$Edzu2V46HBe^t|sY{MBDmrW*NLMdq
zFGzjNk(bD+Z)?CJ8Ys_({7YI`aqb4Ct`@$G+M1LOF^ir1H7Be#>TmP{HuLeY?ep8U
z`kT;HGcz+MV)1TgjM{kefJDnyKfmy~HXqv~?b6aNrNL#$Bxr_4APS6!gCs8d`8g*5
zVk+9awy?KJFOc>W|7^GyvhORbuh%b_8BiMRlcaeAv3H>CoSnRk?!|8X9b`q?RmH%r
zLYu%P(wbw>6Com>s48JGQI}|F;9F8suQC6a<oN?@X(n20?{SdVn<SkAjrUe^C#%??
zD9*CRd?-kN{o;7f@#MIy$Z(${_O52wm9!)<SLL4R`Xtm0(!zX#i9Xsd%{OW;K~wWU
zqj^?cT^82`ip@n0L;P<e_=|86vpTCwDxAF8FLUB(==E%|JRxCU8rLo>+aUEMnkpc_
znjrIZD^sx9Kk*avi07a;zoc?QF-2NjUxcsg;ye0wzfQjeSv|Td|1~CtemGZos+h9J
z%VqY&W83^G^3oMoNGtUe6j$&=61yS4mVnH3?YKf%RRJQVBnIxPfj?cn2r;SJ(*ZQ~
z@9zD&+<RTbuu}<f7&!~CW~B`mOVor5;*b&9ikv@NERAe-vKTFl&FYMq6#iK2WZqfb
zA4`mQoDyns)H14+r8}$Rg~R5RN^NFlRw}cC=hYdHM8lhx7Pj$N<4~ME=zMuTppqJd
zC)@U!>+oWfQwoM)kjp+#WG3|!xx&p0O7WpV|24uX<L1~Nk&P>x%^pQA%BFfe!{HFX
zqWM;*E}0Sov~Id9mm5J#auA;k{21-K2!>A)l}F3E{Y+|CM*2$~Uv`ISIkrZ^Qxg0l
z04wbxzy>Eq!F*gQY;^4L+|WA|7f1t~pK73s`?k!6;y#DTrbm@?pmV{rEFmvrTaqhw
zSDr0Y^BEqk;0fFU&`Br+PnS8eshBmN9%Op?*8<~TvUHL;IrCNX>7nuf@$jJ6`E;gK
z5AaDKd+UGzeKuA{;E?)TlR^LqFwdo~Lwj8P8O8U%azpKM-18JDL(X?AT^~iIz(gwV
zS3+XCG?J+HR5LqDo#;xfWr8g`Ijz1gG?pkJ_YO6FU#FFE1<F^|M+(k`v9wmCD;>;C
zyP0n9SdN)Eh9Lyq#Cqu@RP5iHs(Mbehh2+;nup0efS1yxG&GNG&gt8IVmuXfhBCYD
zNm!z=4_H1h9|!xKYx7GNohMky`)?%84lZ0`o_GXZI(&ZsssNS<zjT3Kan)pjs}oO{
z(hIp^QXSqLeaFhTopDOXL!W0N1t>&BhfBEMF~6)#XJ=%)KPWD8+#G7;SC6&SS^g??
zE~?*(0FB_Ewz4m%Mz!oj;+++2o_ifGK}s*ZrZb%j60FGpY0Wc|XPhr4ISZzYu2Ql)
zjQEUgRCLmTSp8%!Zq1CdXd0&PLioSNC$S(LyJD?3%DUa6NZO3)m~o1p#!6ll&I#W`
zfeCZwWNf)vh)%$rRErvUEu&cS!fm+mO*D$e36pD@#x1t$9)uZ}^%4IOHJ9?~y&>gy
zb%|bS0`eb9R<Jw7f0Sqym_7}_J9-t9f0uwpes|sd<)vVTI`cdT#Tq{kS?0Sjoos#@
zr_IwL^uy{TINJWl*VfYtR>+Jf-)yObQp3_!t16htJ;U_ofYok+bJ5rS42p`f6G!~B
zb%leJWlgGc5rnwKl=B%c1L?r7oIqLj`$gyG=@h)Bk*DtO!GfFC7rVs^cefwM$_4V)
zEKs=G8~=|-;+y*jxMAq27Py6&IFXoBSyUv9>`E=Ix@an8drDpQ<f4X)=*|9xVT(=F
zGO?Ze@!qv^J;$Cgo*k#>w}i?%Cn~I%lKCnfMy1^5JBzYk0A81@@gBLegbR>)9;j7z
zkY+Kt7O4CAYDUlSF>y-!n4Q$^OQjD|w%qk6vOT%J&L+bN(P_~6ujHC`hC<qv4Q8Ou
zxSybT&vLGgXfoRRC3+*_5?2m7a>xBc*nhsFPXtZjgdRWG9Riei%Pk)HW9-`*Z@BB(
z1~3m9J(4=U@zONy9fk_GSfK35_TCqhQnfN`D5#?AnAVkhlNz18!E7<$LVE%M5S?Xi
z$(pTv&ulJ4$BWIPK(i3emVU$9FOIy<!eqW<ra=;`qa$_i&J&a5`6~I;<qLxmA;1ws
z%{9B@Gh5rOS=qMStr^S+q2DOcG$FB96EcZG>PMO=TD=HSE$V4zIfgQz)*Amncv@U;
zT=wM2b2~P*W9Lt@goO4@t9P%X-4Xs48z5TL0E>MP$uibNEdB~e;C{YM4O=T<o%z2#
z>ECZ(69QBi`fXW+AJy`tz>|%>2fn_<J!fwNX~rqByA5w|q>CPoRg|rEf|&hL35Xd;
z_V9Yfm9^n`?$(^`uFHb@#&o9&k8SBU;qM_(`6~1A`sY3^BL$O&18hZOxlEdzdz0d2
zrlz(nuf*`8yKpWaKKDxH(7cFe(>;Aq-7w49Bb!|7lq3zh8Ls)>B<v%DkUq_p$T^@h
zzipz!rebOvP<X(^0PfVn&sW2L%aGP+0m~2}g3;mHVrymZp{c_4=6;{0|FW{a^hhET
z^b|*6)xXH}d&rLZ<L!HxbA>1EcGDW|$QD><4t1VWjE%#<RIex&*r^V{!}m*D7i01`
zIcP8SYW_i$cVN5-Xcm|e&XY2Df~!7M>rqikhAqL$_QFTdkbDEca7l@8gVKY`53M}0
zK0MhxUaRN=Tri%@IE7Da{aNQX)W(qZFr^N<$Ee2a!7x_-_J+lqw1Bd$9;fm6m4ETZ
z(;QK1YFsxlPPtSf8p^uL#3f=j#+oi|IIBn?R71|z0Y2}6qcfJIj)wpIHSn?G4F)aV
zZq+W-)Bd{dpk)O3jWQU`Uyad_HsZ@)lajchE*>f4&+3(Y5qN(fC&%NorOHcAA(cla
zrbvx$CB-#k*4`L-1$8j~JU3u@Iwl&{d9qNN=A*k$zvpT^3lgOBXi4j)#RA(p{{96f
zSFS(OkN!(kMi>Pvj+T_erabD+#Dn<v)z8Nx);?_o-&xg)U3dwM>l-?cxsspSMOqb#
z)p61+h@}P@ru4lL71z!F<cxWz?$ulG#g&y#-OH{h&^gHLe@h^9sGvx#MkB@hL~h)>
z!FpiN$MYP8v3vuP<3R=O+Lgp5`&4ViiQL8X{ys;sTn)P0T|LAQ|6<nrWC)~-^F74p
zNP~<8@!jjM<NR~jufu)bSGRQzrq{LLtR%zRtzU^C#Ht9HuM`C(QC(l^+S@WRhvn`n
zF^>=TpViBX5T%zYkLh6#N@u>KvOBD*KAM`+l9rQ;2IY!vVGRKM+iTC?a;9E_x6+0o
zk;`gla(~5)ZK<eX<(O&G;E1_&^g-R2(ZtALFO(kCoJTDL;&F<-wcO1^gpuHZ4rmVh
z61ywSRu=Ig^HH8-bolUp3HZ-}BPX*%m3GU+qF*><(Qvl7Y^T%)E(J$ki^)zADeU(|
zQR$L0#N4=j@s$KmMI{psjOcF7u*oOQiJVQaH{PEx`9k@jbf3^N5df05B}NIA$3J%4
zYzX@R-4Oh$UiNNv-wOd>GLAd8S+j@i{fZ}1EYcx|TvU;GKL=#9pq;Y-5Z-qTTP&S2
zu#}6CS+|fA0AjJ2GI=2pWczz{u<O4)T>pCDQc?kmjtCo8&DFk#{5T}H2VM+gB2V7h
z`bIon0h1Y&y#yQGrE!iNZO_qIHZ*D12^^D2l7`+po*T6?QfvL5Htn6gWr-hQZkex(
zQb<a%3W`{AOvB$*fi#s;AuzQLehnWmJLdof&nZ0|$iv^x?0*y|_&_)l-5G~GGEy{L
zBGYUEDuBW`565wO8h;P;mWCLo+M-Z*lqJq^QVnO938AB-vyP$#$e!=0o|sKICVrjY
zj|z+KA)g$l?^MYz-)di;SDaB461NQ)ER+n1pT<((#opbxC*8v>-AGRpZz-5C5_n}A
z{$>08b#;G_^Db@L5jj@r6Zt1QBe(T#x2_D4-FuZXowPf*vNX=WSjl#ItU2T29-I%5
ztc+9p&I?lF*DzGyPCdQJfV7N9Kax<WbUrGoer5=rhO9Ayw1HyV!SA5=zq$JUxjOf+
z`!U{spSvTEyv$9ypS;nf!G23b>)w)f;R*Lp_u{rXo!G<WnJ;HE)zdX8o!SDN+5-ja
zoAy2=H!}gs<Zh2UMQ4Zdg_t{cvgK^b5->IP&4<sHtE{JKcxreyOOxPQ5Z<Ksq{Dfg
z(Xq+8mcv39XPSbDU3U`K#_jL}KxRi`PtNeq0sV(pzQSOjC=z!dDIku^UzMP)-_fdj
zcN}0ZPCQpfeaK#|-}GHtVqyq=32Op#L~m38Lg*QwRM^liRQLT@v*>6Neo&P?_E54O
zTeH$~Iho0smxDuDgVRRCGC4~|H3;M7r`JVel4?i`_1~Qz|NK8ROnk%J*CNx(fIHLw
ze)wB3K!1$osiqoBH%-(4Zq1_zQv1iquGhNdUF`3P6;zL=h5TDe#BOxHlEQIJbln-Q
zx=PplX@&f!`MMCdb@s*%efs1wD|c9QAJ~VVP+1}q1zmcp?<5Yw?`B~|-s91C`T8ri
z8g=PQ-#46xkr62mN&~uZv{kDGyA3nb{yFCV^eS31P57};gK%V26n!^y+2<^RpUlo{
z_4Jv}KVMQgV@BNUdH(7lzT>~rPX50#`d%UoYu5(>6n!1L`Y5<YJbGkjJvG@6amuIF
zqI}70HZa*ZZaSLt!T*5>R`?hN9i1i`o%$g<V4-f!Q4U0l-`yd0zLlI$IAcWcAmQ;N
zmV28JwAvZEEE4(e<1`vud&?}5AAr^+=$6<=BNHO7m(2SBlr)jkCL%e6(Fl26|JU6|
z3HsDvj704;A?Rj|^9b2!uPtziLDj!k7GKBbLwUmfl<FyE6B*+inxbhk&mPejBhl^2
z2uzN0>+^Tv3s?eYV`UHBYmORe{R-E|U{-xgjDJUhd}Z~IAbnzuslimPG||Abdmugn
zxaLQd9vvG3cMwn*;1SL<&CH3H7pQ%b?uF@>$$|3p&WskB#p=&DZfqPDP8P;~*xyfE
z_hTAzru(}o`Uo4{{X?ZXs@LHT&%lr0^(P2XCdHh?LCx6eajf>#bs5JM$#e7P@>2=w
zYqHjc#C9iEO^k$+vc8<CvRWFo|M+XzNUy>Fh3AgxU~K@hMN?OECZ2SdEo(icE+$vo
zs!TP^6WV<h0{HJD7uNwqOW*46Q9s@-Cmg2&_rX`2Q)gC{Pc+3%7z*%v8jo#{_<mdE
zbgWa)^`OtYK_4q~3#W3OiX4f@YGca?R%NNe7OEeUbF7iEaa(*2uER=mylK6X7(sde
z)1=&?fA6{c)mrSJ)1zX1{58p{*p(j(_~SQB#QYiev&1~m7<|W{I0}y8Q`YhqTNfXy
zA=3~4eL|G@+Sggt#7C>BG-$tNPz{OguNjtMmQ2{#nf*GoY6I9#fAf7G;UV;!>`i_+
zM6V#BMmO3^;2!|!+TWZLU^bvahg3%dl#rq@-dXpsXaC#Z0Tu<N9-G?t=Fs9>5r7Yg
zN-?q=fRp_X-~0c`y@7#KQ3Lxb^Rd~DzsmeSPU(&c$mhQmsJMUce;Ep<_zUp(bLoLU
zBk%uRx{vdLvxntw&hzwd%@w4?gFb=BBO)`&|EH&)t0857IApK)KgbYx4s_sgDr#xI
z|LN)QVwA5~?uP$^Q~-F5>j`>fVxobLYh-LJ3$R=i@ud)|B+k_Ba6{0#2z*ExbMs)=
zEa?}e%vqH9o?sjqHnO+=ZXAH3g}s=HfkCF!!9K58)DTY_RgwKE^(@-g#kkn#m9J%F
zqWAY5QJAR2WEolj=<zEYyuAOo%zu91yNe0vFxx6O)UC6FKkTy;;D;NtvShws1%<5Q
zi8gG$iwQo__s3JB4Nz6>Ww&1#OV+;5i&4qu9LMyv1?DfN)O+5)8@P`!9XcS?m9v$o
zoPL4IkZWRm=k==>&3eNePbE<4_LT+7ge$EljZv8@Cxl~Aq>EIT#*7)o(xy<_m#lp;
z{v7Fl(cUvOk9Y=;MnVn?!{?wP0;_PFAh_!ThUS+7Qj2~cqKX_qW`-1U`inpDzy9Mr
z^wJ1@w+o*LxV%Kroz#dF9U~@vr=R9p_Y5Pb-SD1ZiX+Qio`(#8{K$V-_19*;`T30f
zlL%f$%tpczL{8d>dEw#z-=9#w-klVl9Ay#K;AU(b?eygfDg=NKhPPy7^oY(Mf}()=
z6wT1TO~C6jh681EACkQZ0L!0kCT|XDZB%ftpb-A=Cqr(%u(fpx$CsGpx^~?q5&CQ2
zN+3NF=x`KmnlH#5?WWJe_;-6W2#Q3s-(ct9;5bYk{G1APVAc}>sXGvIdlBjUcdMM%
zfem^^)KK5u_*(6KGZcK8-dbUmryr=`^K!P1D~^c&wj#m-SjSBa^!e_*hnTj^{P$!f
z#lb=bJsm*ilm4SN{NsbirC^%xJ)C(s|8AOBS4B_|c>MO8%-jFd)6Wf&2iFPQ-tvDL
z3MKw+-=Uy#(MSKo!-Q1DwmWymE3N7rH^-PwFF}JPh>M-2tvkN4w=V_U0W6EB(Rm|p
zwNw)bLCfCDgkM7vl$%Jewoa)D?~f$>f4P(9u(}YC^eMzog38_ZtrT=O;G+d??;vt6
zuQ_dI6b)<2(nQYP><15GdCDYaT4$VfBaDpHw|@#Obd-`zQsQ$VEwApcdQyM4`qz91
zuuP4E!SUAK(GdxF@H9ECryips$Y1A5f)lklx5)S2>A-I9l=nv6ArgY04+<K9)EnM>
zPTm}Y6?p2FIVZ1%_z6>iV=)5LB*NhDE_7ddy1sNzvYOq2y~TdoRATau1?VloC|okG
zPCF?I8fj^1FPgfo5p|6F9%*%O5KE11Fz$`0ukULPtd8#f{?OB1x(M~KBxJMx=(f7y
zI~o$(=QeBJwl;O}g<e=p`<6EqliatKmPb9sov@P&6W67KcIN~N%-7C96X+2nhL^wd
zHUD{-X)myW-IuQA)<15)g&H<V8Pc7|Rrdg(q&o)QFq^H6F1H?@(pClKN5JpH!u{le
zm=3d4?v6_bwA+IJhqAv8h-&TL$6*OUL_!cefRqS`lrTfLfGDXT(t`BRjdUr3w7?KU
zBOpjO3@~(ecXtilFu#qS^M0T3oagiYCr<5s?>pAI)^%O$?7$RZK>mSaJr;He9V2t6
zdDV6k@E%xX44wJ~`2SP516G}%zuU01&yEPE!y>RM0{YdRMg<JiXi+#;f{E=n2H6ya
zKy*`V9bFx^7`%W3k5NUD7br0-Y>s7~(o32d2u9L`>^J}P4E={m{OWO%7oilj^>_Fz
za?IA;7e_0pj(|0GYcqGiKVtBn00JPE0$QS(Zw_0rUjYxqVo(fQvGGv=;6>I(mhr%7
z#phBr-`Ky3>l6c?l}PsNn>c*9E^EvHFrIx3Xad6kyLsR2hTE9)YQRqLL0Dju;(u-h
zgc3Rn%>YGD#!^~LtY&8X@_2(`s&bp{8l83BJLTD~q{H7}1b8*uRq}RVGS+J|BsQ0;
zC7vx8pR>co^bPi+0g)@gDZJi%T<MeDY3xMkD{+0lv5+h5W43E59#AL(P>|j1{OiKc
zKP)`pRZ$M~v9Ynekxf2H6T9`%ZZ?^zu=r<2;35gYu9YMce!d-_rV?@qbWU@(8%aw`
zs~7;9VTDdPZftgA!%i+k5K~R`LqlK28E$&DKLH+!fspOy&wfY$^s{f&yw-oE8L@Xr
zg}njaqJyG8z)5*PN+h!)zhwXzeqIdofyfL;LjkM}su%p=QAI_CgSE8udnm5Cwzj(6
zhmcQzb1?)Na-Hzwb0`)W;(-KTP#~W;Rw@S=Zs*fo8+HKlKY)Z&AKuLN%LC!>heC9S
zwD(2;NP%Mcp(c<vZ5PX5H(N;hY^3elVCiHH%1Ct0wwn$^qsyPpMy#|-CAyQ8kH*gc
z<;<$nZtU$UN8MUlY$Bbxo9`?8uN|Es3X}qWNpb%E5qu==O~Q8jq+w$VR63-yJho#P
zBi}UTdK|Rfug{XA$9xh1?EjHVvIU8M{~domUr{DZJLAyoO|w&rb#{jzP{Z%RM_cOW
zB~SI1Sp*tCvvqcM?r7XMSq{ic_f>DQ!6dp4Kt2jrKX~xJX6qYeukQHqfEp;PHWR8_
z`_awQQ=7p#jLjWd6M$YI<9=pPxdZ`fCxA<(f$AI6;auDwRFQ-)z}JV-R-ldG|Gach
zsU)D@o6=y<@~%*}2K7GnVAR7Bezes_T_NuGU>rrmvr}IL>20(6e5=C#!bhykh!#Ck
z8Myp*=jBn$|9Ldh_+1AKRmXK1HYZ8qkNEi1946Aji%kclm%FqdvsWJl4+3`Yq?cuH
zYU0m|rYeELm42R(175%4jwGf$rsI3aHT@y%WF?yJCk_C(lEb*op@2I_CSZ>FEj|4n
z-xt6I9%zgc`R!50HO_i5oHvvIy3}%p?QnfK3^;8N_PfHV=R)U+PW|Q9bdbS(uW#S(
z2^|?zHF)0)(BCNC>aQB>R)jm`7&}N#{LOfYHUhzK;Zi}yyHF9MOFFINNoDm?V?&Tq
zoP9HnP0JLqBtpdf#D8Vd{ZL@hwp*~z^*(HqjTtguu7}T_`9%u_2pN5qL!@uLA@$lp
zF=w(!HO{{p()-H`08njTFWAr<=%<Dcs<`8ipuup5C^e5Rd}oPk;n(JHN!bC=B{J-=
z`W(MtqlU{~SFIF)1+l4IKe~1Cc-=N&(rXTi^1;S0IY*t|pNUBGCn{4jdssB%yy8}|
zH(p>kS@+5PFt6syaLrt{r+Vl1p9Rj}-EO29;2Tm?43OUQP;9E3+ctpZ%Hj!)*WCsM
zK6&!Nyp>^GBi*&HOGMqwai!mVL=A{&d?XF{H8z>AC$nhey%*BA-A7}m9=>VoZo?^Q
zw>-U8yk5&rE!AFv@W=<F0M#~d%nF_!ol~110uuD17*4yfL;>_?rItNUGBY1;h@7l~
zoa!asn)2Dz*7YMpeKzgX+>Jxq0jE0&A0|@NaUzTANg@!o)VN-D5dfTGo3b{bcXRuY
z3g8dOW!jpN=wY7WayZ(P)9{mP_?z93um}BzN*-&WOKf$@c@5CT*o}8Pya!Q`SNo->
z`Xc=Sm%iubggHBG?V*)hzLSx0n9hc(ujs?;$=GTXrT3G9_2Mxao^ZK)=hild`B>qQ
zeBQh2=apMCJzBZMm9iaxN*PMdl{Epmv~HaB;sluU+f}6nPWz;eHDF@E&pjf!g`f=9
z8F!Qh=F&Te%y2a<RF1Q0zG*?Mrlw}o#KlvwpUeO*9LEXW8FpHkB3Rtq#|;^RcgJiG
z7E}XOY~G^1E=Q5_0Zi5Md;hA0{+E4LG!FOTVnY@oBxuqqy5=&A#_2Jq8)G<pxwCKA
zht!O}t`47172wmQ%-j_@o`>!AF>1${JGTPDhR{GtafOXnLx3fLh}E&Gk#zy`eh|6&
z$VTb(R`@d^HB~dyY>A}#>yYEb_EnZpG{~_=10jCSy$vQNVN6<hGX&;gS5i9d$(-*-
z7ovN`g?SJ)`Dr<7PKSQb2Cvx4o7E0pVmfka?%wBDq`&&wXRA==xfVMBY5#}=n|;^h
zotK51U0q1p%tKbj)|kSzzdGA*NGv&LG}vQeiMw7cmARjjo1`oMAO-wHu1#Ys!($mP
z<H=2ETBjTX!z>ucHHuU|bhI$oCn%A!zI!MGWpgz2sX5*nkjY9*yQ$T3-KX()Are4z
zj^{6pJQlg=G$a>BZHA}GZ>iffjKU?5dQJht`3{S77}r}=!Pt}Zo;Z4Tgz%WVBS0m=
z9EP|BJ+St*GoE?NhBzWc)ZWQpWwfe~VPCrh;zd$7NFUO~lG?@oB`lu`pnuN>TWKah
zQv}kyNUp0*%bm=3@_W^N&W?c&s^_Frp7aK!<vWEGi~YDgUje>+<fw0YBDLfBWq+Qf
z0@)}siD*OVXD6}{(HS4-b}3}_Joj!#9%!+$uWZ~2X`}VjZq(`=x4_V0RqtVN{IzMF
zQGbo81MVfgnoKx@Kn1d!v$~rv&3psepZ^}ii2ETDh&v-FEGv}~Jl5>Xs^>BiA7?T#
zZM)|Cb1}SH3@V#0`VqT1_*j($GLswRdbl=GScLPvSM<wE1;k^OcLHz8c`WYPt=^@4
zyJQ}ISkIIZ?-T|i+jBdjR+16oe&5=x`v0Wb|1w2^r5W*B6oC-RH*#F_c)?Dr`Fg@I
zz_`*@kEb?83--2O4^^F}F(Pyw5o}3fuvXfgiMSl2X-nDDyI*|2-=SQ(G!17xp5|cF
z!Qkc!Ee#EQ#_c7yr%(tZbP#X<`WzY>`kYAdme<GLy7Y?UX%}R5TVW;YSQFB`e-_T7
z=B$9|SXt|utQe@-XMyn;b+0`EYP(QABzyi0vnk>8j9|)1jv=Afnhz>3?Cp^OTePpH
zuEUx#V>~nD));mB1hN~i(qL<3giiY;qC)KX^MIrm%B$d)UsBYWIXGT3%shR&%&KYV
z&C+~+@#!7}0|3&anmSY@$w?3Ql9$;VX?<)`sd_{K?9l}sE=xG2ggD6&dTN<fOntGs
z(k}?l-169jp%m1M9~0P=TXFOpL3$ei7erV2$3j9=SK=RfYbQMogfUDZ&wDJ7;MDC=
zh@)uO`&82?`Mr^q1Zxc8XOyv6ZXaa)T~WK1Sd_8)-r%#NH4;0u{O%dlu9Inc2iTkc
z%|7%^TW9Mnu0Z#{SrxODw<q0N?l^j=doR?u7`CmKQ0Vu1*1V^_Cj(j%wfI9sy*Y@A
z?Z>cL1eCQ1kOyCI?-3x*6B$$+rp!uYUr~Nj|6}Y0D*z0?mf49M%K_RTz@5m;Tt?&p
zL*~Magj4bO+l7x=o_y{9s;Ds{97`_koSKuA|DwHJ+9y3hK}OELhi~wsg?OT{Jf|4J
zNRejCYM76Cds+~zEY?oiO`P(bC@Q9CMSmgPoO2<a0t2dCs&_aZRA@AFor~eNe860f
z_B1s3FxRAl9=u+Cyi(z&?X*by@}*oD5Ivj|%u~CgCcZ)w7+Qi8ty)$a%dYlCzO`Pw
zGk@>RZm0HzNiQOPs?$i;ywljKG46P|JtXlW&1|&1NeGiwX)nH|#^Q>qJeH$DYn5kr
z0CgTY<?A4_*VNNqi*|YO{EhojqrcML7RsqUm)E=<Z|?$#Uv20_P)o?gT_oxa_yav1
zozNmL>E9&_B}~1=FX_&ctcqKbjz^y3Q!j!poTqK~$2dT@244d=xW4+y8egVJg_rdW
zA0cDwGZ{ZYMeIV=O&+cdkBcJhi3azbE0@&U`~d4F{9>1#oiBrYCs%>!_~}Zz0uMp!
z7ONaN&+R`J%u<zF9QIIDr<%2UvFp8H8FB36sM7$blzr~>f^p=4qD2CK--L478Zo1S
zGDPU@vm0KO^u0jN=yXwFIqJi2e-^kJ<Z4chK(ej%RsVPtTWlWNZ$7#Tu~}T@fU}HK
zySW@koKjaUgIw1NBW-3FMUY2)F=SxRIi8-Yu_FOD{(&E;KX{6Ku(&5}N6fb?`{Z5s
zCuX9I<kD|t*l&)$yISeXdueFsD;sA6?=nm#V-j{lwaDgceg$;`|LDl;An@2Ljvo>A
z%kwy#nKHNnPTAbBo48SxQ$OThl9*!?P!^z3b$f?QyL`S_WkgsRE@97HLIo(zbmb-A
zW)`GMVQR?!DS`W4V05c76UYecFY5X0h~=%C=IB}$M|y-UO&ASgiiJT2XL8}2WUIWE
zh5_S(IV8*@6>#lJUAZ(Ld@x>E83`%rc*GXDXmXxoXNCKcOkx<6glVr@&(!8@HaT@K
zZ)66iG5`ZR#E)vFC&7Jq`o<yk!Hzh=3vi6!#fTo@RlOc13LDN@#7*?5O*oi&cM;{X
zmsp8P?H4Hv5y%^JE`~rt<4C)vu5^fvdg2Ml0$=NL;yGbFuG)}VXV2xuLh|>?id;1H
zv;vDhi}GyM!%b57v;a;&q|X$Ku=4+w{%!QiQs@Z7X{n@aQAGiz$`7mb8b5H{^bonC
z<KbO6iUI6wMLg=YTJZPIIXvo!PWu${u;h6I6yo`}upq|T4Zdo77c<(W0f@`4ycaF_
zTbspH1M$?S5%JTd?+Xs!D&S4_a=S)bJ}TQ<{L||A_hX!MJ;o>f+mj_cv#VD-D^ZJ^
z6}CrdWYkob%?;eGDe$)Kq`YwpdV2aovnBQ8r`(ofv@7hIPtj<-$QUQZ68v?>7>f^D
zD)#RiwB%0hwQtyNcGWT_Ud3OuAUxF7mx_0?wNBdrVi{+T;8gbcPAR>NO2-i;+*~ER
zImn>%dolwh%An8GW+IRYt&Yge+5R|p8x$WAaqsE0(T$%pXhO8E!P#Punmu0y&rbuW
z3)`MnU`LJ{OF|+&%0;&?PmFxn2KvdVJJ|?W9zR9912S|U3_d?oK|GG7d^S;YiM=$C
zcupX2xfkH^l<Xt<#2S<pCZ*infPZw+OY_xn#<^{9$8;_{$hf||uh6y6%;_w~$oy4f
z*3gDrdT{36WRZf*jIhua2QR%OIJHH={Ukn#j7<<;SVz|FWQ>3^SjUzjv=iTPgQSA_
z(mxBkAV7~+&+8>SCa-9H>L=hBzqrf44SoPQ3?Wd(dd`+gEcp7(o2$V>s-p_q`;)1T
zP&j1?xS$R5&8kv!tX9e=Cx~a$dI?Ow|6`xB@WZZ)B(9X_cPJlKEHjKQC<q$(u)K59
zD-}GLWz4G5(%MyYuW>3om#F2IPvm3nCZ0e0SXL%i^FI3#1A{X>OVDle1m|>4Au)6B
ztosY2d_2x~|0R}d??GJ!Q4gLEmWv|$8zSwzO2NAAi~hol`}^9$%sCn*2W$I@W95(^
z9#H~$F%(z_NYfQA4WEJSntg5_TM6g0q3mbsh7v`FyjweAsonZ%(<PHH^Pyp37Siq!
z*p9C4nD(1Ib(F^<C-jNQ<TrMKeVl3t`Cf>?4I<#YSx%wDVXOm4=QFfV&5p+af)&sg
zh&#wC$2Q}rRKKHJ%SATy#mOe^TxkR6?+;BsZ7)n)lbVpDW@N?iA!xu<OgxGII$)OX
zGMR?*7|(C<*)H7;PoCc=aF=yI8?@Z&@^BkJ0iudZgJ)Sz$I9cAdn8L^%+Es}JQ?eX
z9lsGk-|<#k$@9~(C?^QCI1RT5Hv&3qS!jS>8EY$R;M%6{)SV*_)h+|HPfI8H^$P7a
z`r9%P42Fe;<gMi#4H(>R6-l*mGxAFB(iD0biyfn7$^CKs`*TdC9H-v$%~=*!Ue66L
ze)gT=3FOTO3r00N%4&j0IRIQJfp6wGqPJAwGV|?TVdHL+Yjq)oGgzd)$?5Csx4KbR
zOkE^Mb9h<FJ;V!ft}v`9*ITwb9qaK}$W5)IRB%PTCgZY+o=}PLM<Cw3gG!93!7WlR
zHipU;gVjKja<U2F>3Dpeu!+17!K;~>(C0zqz7)-WGpyfR(9d@A4Hej#;ks0$3#1<A
zuJ;Dt!QSDLqYp6w7?{1uaBXR&;}(m6BeH~@F?zm@T9|<xGihtfab>)pW?CnZoHy`0
zx6@p`<XE`@j8Z^fM+Bndja|GJK8lS`1;yW*7(X?=>`PO^b_6!lL`3=xTu1Q|aM)0i
zm+{jt9s&**-&jBI#VEMF;S^K|5m>N0by}b_(=xh)1PYQc^rj~Iy@?rAlc7>;2V7r4
zci5b#<3}EXF41lE$KH5{hqc0Hl!6d#wacN^ijN$NcE$6njXvZ}jm4uJ)snaFhZ8HC
z7v^QMw&!$mVAFAlgsU;s*`0wIIc11}fkC1En)K;_w)V9QykqJkR*KNoD9>W*@iZBD
zwv+M?!)Ktom%9q=CxIu~II4YxZ;vOr2DY@T^gjt6q~&$wCu1y?^;`^YIWiQRwIM|I
z@c7*4zim!dX(}q77)9tc{0+rbljMc8bOh&q^*49ynAlhwdo-mHIiTI%T1gV^t9bND
zGGI!Rj}9L|cv_WuHcK>)u#{vIoqh0fdv5|VJ@QevH13LV!MoWxQe|xYgOhdx+$N2L
zYL%`fEZQZx<LE4aSYoU6P|EuCDZ5me&$XtErupX8!N`4s;{><8Ak;BEusbA_8l@U=
z07XV7a+ci%Mvu9KWGJ(=w7d;4ipI0g$^pm+tzspWV4Rk`bH_hqirC^qEvHbr;|z>Z
zeVT4SzZd|LGX>J^Hy$|hkA8$k^qUdk-e-?Opca%hh#Ea6qo}Xq9>sdxKwVPjov`o4
zj1<x}$T4N=m`{5*r9T)f@RZd4a6+MIb`^HXL0pw=-hFPnRbJivh>59}i!tL<H8;OQ
zTOkSOd(EK^?KN)W-=1zXgo)^F%PteTfFf*Z6(|3DylM8)_<?#a;dgl;kjqjk!YWOR
zVTxlRtvroS2Y)=0LqNW^nH2PNS<D@b)Y3{+w>8PAOcXS9g(KuCg!Mh~@>LC`KTo=E
zfcIrK$3<)y6%rh{+3_f#Mr~FCy2dGj9IsE&v}-hJnn)0<C5%YLQ>Beql5geTAJ1mr
zz8!E)S+c7rTkCpVJtr*=Xv6REx}62OB0Um|u{}_X+x8eN2h_77XLWj<`?(737oTJ%
z!w1Z8N5+SvZ(uVjHRF0**{Au;fvx5;wJ*45pEd_lu>ry4-r9&P5s>MW#5@77&NVm%
zxI(dEr`rVWLf)+VM^i36*-2=m<c+&CA*fdT>ir`Ppnt)1gcyeHQW(4Y+ii~>nY~t(
z8hUyL(^iDrJl)=~0CK{bM=pMGs#stO2sp`j9KHcAt)2S!V&ptOG5;ON^WkQ3cHxV#
zd-q6T*aMG==SwQ62Gpq}wn`$k12zg&j$UhH=3ZJjx`kym?GF;s%I`zbciSvDjB*8a
z(b-LYQ1aR}Lecv!fe3kv>Umji{3Q4oR<`}L<E*?Ri+^{{doLrED>?}h-mE+&`(}eA
zn!*<Dh>WUnbB<P5d!0jcnDmgCH%##^*YP^cR)4o&8EwP;*gpgPtcr5xB5@DqD?fv9
zJ75bRclQK!04uLCx|Wzh?-Q1Q>(M}B=;mg+uv8uzu6p;pjP{3D?x@Lqo9#)b#3gd8
zB}|siawlog(=SI`-cgWBz2Q5=V)0L+*tMZR>)TT^B_qJq4ANns$@Vz3MK<Xf5Hh<i
zh4sh)0p#XrSi|R5Mv-mUp)Sxp0f=%SXcK3h;@VK2X=k3KnF)?T%w?d^-~_Li2XNFe
zJVG4qofd^F#Zu%bedW~M2CNC7H-%%z0aHqhSTXnuRG&=5A_69+H~*2Nj@#gT8@}QI
zm@F8`x}W%1%9z|<0~90$zk^wBcsV}_jfmE%5s>93W08I|<9ZxBL{pR?J=hKTQNdk}
z<;-e)>rDm_3=U<gVN5u?9DUw?Tf*YUQ(JFX`28_Sir63g!yXlI8$1PoQa`=@es%VW
zTdt}I{_!Iw8OY}QOSrH$0GhkVp^?(3rExl55y9??mx-9~osf47TdgV9bh=vtxb)mJ
z$Vhr5UjF0nt&fsd-caWB>L<6hw->@<?Kk8&^~Qs?4HXjjSw+s9iuSfF1?Lym?tC!q
z#<rQ$O_(&YD0S9wety71U%kNrD78x1t*+ykg~WIH*n1V&B4h`i38$w7<|u`PlSI!Y
z>P{_V43=1)9;nD=O;z!W+9X0sax=m#E{nkz;9KGz>?Wny=726@3@_h!>WNgsMc+d5
z+vBy7SwesWWr<9%87O|qjr;LRz4|a%VYI;Gv;owgB3atwJZDIC(%G=hec5t-fLtNh
z_BP20+>X%t?$Ayj0g=8MG)I+n{WN)ZAid7T2m9z4e?O+CV&1U?tDZ%*{0)NwIkG4Y
z5IFrdA{4!OAXa$2UB%q{8jed}O+)~tRkcTR`=H)+Xz|W~YWc?*m0D5Fb!PuLuN-;_
z)F>;q@0zF)qLx`nbX%O8H&pT3Z<QvX9C8h68VB>Vul=K1^Xt6Eh<)+fOVCIu%FICY
zfDr0+>Ryt!&iP~+Dm2p+(QXt5ZRjne(J0-D7%=J{H&9{-3a4leIU;k+0jbD-`Qr0A
zWW4KQJB0m(Ls7cTNcI}1*^ASlQW>Yyqh<~Jo2tfYp;&mFn^)JJ)U@SoA8wFyS*#8q
zSrkADunnbwIhOc>>%Q`4w~C*7-Hj#D{n&P^uRza*iJ2Lfcjx{)S4sIeL9*UF%7Ccs
z{L3>qt!f*-Q5w)>pe+ZK&R=xo7KPK#{N#y3)N)ZRC_GV{YWIhhMj0}~aM@A6Jl%fF
z*$%l5k^pH`UCdK|GYYZjf0B^~H#>~oqY)}gC1*Hr%vO3PDdTNS*p_&;m(l}0q#h}n
zIe~ATZ{Dvlnc>oTs3}+Di&xah#0A(0BMEip=-U5=_Qs1I%O{4sT)clTwtwelDU1eJ
zD#LJ^tGvG8txKWGV4$tr%o8A^nk7#1x;NcP+)WEX41ixf1R$T=yVR*-_s_E0L<3kg
z9KQ^gk5DLAJ3H))U_z$&rN(1nzMn&=!nroM+Z%UN(VIUATpjhBPOBwCrdkFmT4FXy
zh+WTq@TZEAoD)2N9#WLGwhwJAO`hU9D1nS*!ze8O*a)YkL-x}{#>-zTJCfulTz$3F
zQQR81o@mzr1yPqKheM}dC=Kdx3;Bulb22j(2f`F_?!zc<A@0L`fx)s}zJdmS*BW_2
zdNGnZv%i58HzZ3w!OeoN`t-f#Ay&)H&7_otl(xMh50sd|@nT4go|{jQ@-ZK>D{{)^
zXsqctmWU>7yqJk+RFN(=NW4g!@FwNUz@^_F&4UVFgM;@fbg$r=uk<Z1q9P&yo|JC|
z>W$D?|3-$%tXfp0RzaF7AaN=UwA)WAEXbgJMGQcJgC9-kOEbmpDG2Rj!<(;BKnmmJ
zCE>VDk}z+&s<da<k?0iM!6u}3%(rHvT*vFd_UsdR7Ojfr+)Odj5Z5h4gN4DiciLSH
z7Vn!1`aGr&<GYBba@0#A*4t^4vCo3Bk~(9US>^c83Q|9JB%!%2!4(AQ;{_adfQ~Q)
z_vQFMbUl8pCD-BkC>r%PZ|aZywxX5piJFoP9)H6hJC;%*@!&DRZcHtnGHrI)RJ{k~
zGN4M?#=STfPrU6CdOR)E0ma^<J~-Mu#z1FJmrZyq2Lm5bKhf(%@}r>P11XAVq*oi*
zun}4Ek}>sCGdA_^RND;vuM>F+qvEHh?CwTUj$8dN8RQ;NIApWMJ<Sz-^{V4zCyz&-
zTs$YoJpY~F?kYFD&X|&poetc44Z_==Vca-~On~YlmjIJ6VzbvxA)m06jiHw*^EOIT
zfxbD$%eTFmv;5jkUSAVSNtC+q_6#$3uta#h?pCfCOS`T{zgmbhhtSl~F|?-e!=M+V
z<q#OW?Ull=$<8Nk)?MqJ-q!$TT>=kwOaTxsn~j6E>0I)7Dii70JF7D-hNFJPsXm{B
zU!0fLg$VRlUUhKUHYZeLfAeqi+wF$$Pd3HUi(8IDYCos<psc2e;G3hn#79#wMG+2;
zHX|(~IzHA|88~50_aE2&WVml8oLrS*;Ja>LH0||^bbwHkE<G$xsD@8A?U+`SyjCv~
zPNq})3&i1`_7?)vF6Y!&T4?)GY)djN+dk%~AE*<hQCSK!<`Kmf5BDvZ{$#h4eeD91
z9G1LW_byU!tX|5>27V|W8|%_<%&&8(R*{P#1Q?KCIsWF1Wd-{7h08wuG|kTZgv}a*
zb)8F^E78?Ny0D8|RxQ-T*T5DHn8TQ%nlt!}Cv{9{HYcqdr>A*Wt5hEcka&$&_pxGl
zoz4QQKt>5*iwOGTDbN6x*5+ijg2VQ-skYeLiu7DAh6eA_O+&dQkD9}!{4sj>Q<w<Q
zXiPwkDcGJCgC%!OCo8`Yi&J7iTBZZg`=30`JWm}91}#sX1~}{`_qf+HmM?UU+3t-x
zdOXoSySVHxh7cQ-v#6CyzjHS)6};%Z?En5+hcu{5!{NnCi2t2dJXAy}TnzAOqFKJ#
z-rl~rf^dCXbBtJ85_LXWWl*>}f!mzyyskJcJ=Z@E*5Z7>C+oCk+0(w0?4!Hv3RtH;
z=3}U38F!d(%nTJj2e5FP7=YsBBMbQcv?#<yz_-i0)`CxjyCSyxYwvVXrHI)M#b`0j
zca!~l)#Cq$)^ac=@mzsG{c?n!hVLG8=PCA`E1X@<M@9?4#)WhMb@xmg13kU8jm=l5
zm>Z<+xr5u1nNlvmnNG^?ke}qQEplG9s#Gs;Td2ruAQ5-<0to<k;UZT{^iBs91QTCr
z$#s(Dj@P;(?m`5@pzP{}{rF@c6XQmh44p4QF&b$=R1S2)P?iI+sk^e^xVUQVk`SNr
zfk^Oc44a6}W_eJG>A(}l*g4MYjH}o#@?zs~%m&qhJW?~-I_ugy7rE4Gj+rGuXK^)-
z7NcRic|bh{WJTfHm&+jx9bFGyf@9_8LJ;Q_a!eWY9Mi5#c{Uwe)MYU&=`fuDwX!J|
z0@S(DH-t9fTL>gG^W2f?2_tewE^2Ro>EaXa>zt^t1dINHn^6ASC3|j(7p~tWx%y{-
zuJ>v?b*6yr5_j8El<k7~O7|z>mNg!^Qq=ywf5<e+E>Z-^9dqpq@ZT|s*BNV)*NvCU
z0t3aX2=1I2oj4(Mds!NXZL1NE@(pX=Ai2|9=9WEI`O<J11(_fH&V9Ldtr2Hd(kgeT
zqA9FPCciNHb6xg9&EelX5UmQNQj@Fp<oQk;_!`}*kb&l*Lcp|eK_c<pR7wp~#_JE%
zfD``dF6y#bA7#wUBmaQ)?BMJf@$6oRM>W&JbAl+B?daF6EzQpXiK)eH<I^PV%Ok$i
zOIwd3FXHak)2KF=CF~H70Y<-MjgfB7l^{e15B0fMr;}xF)!CWA5f@}$)oC#xF_L51
z(YN&eJ7>wOg&!gOurFtqdZ?pUgL)pu?Xj+BULle~HQZPxPCsbV^&?7w)L+J7b5hS|
z_T7hF?NWB;A70?B6M->@Xlj#-rOzTOoqPYpemN8c*c<9qt<$nZ>doR4E7TX!#-sto
z#rK?MiH}Exks(qwJ4{Hk+G3@L5+fr-v{zkl1(ZQ#y!Lx(H{*@|3LkKf1-bPhnD`5i
z?kG*TqFWA&Rsp{u;7Jque!4!5^ZjP4-^w3dH{=eBou{71SPumC-EnNO1CN2Wt2ek$
z&qcJOT=x%7NQn=udn3@xFH0eukoxHFK3tP{oIr_?gK-B?Gm3$GpTP{%P8wUbp*RcA
zv4ui{DJMGiJ<cM5yib<u5{ZiJ(kPn<ydUd7-=&gRh5+5hz3r2$0gGpjsxrNjhmao9
zx?Tln7l}Q3ok#e7);CCFY0W6%kjx%PUAP7O6d13e+~4p5e*WJpBtW11z9cg}C5ZN8
zN#GzY=1K9RR!8ddESERq$47A6t*YRldo;-%QZO4L`TbmxslkA}+_$>@?WUOo=5(J4
zf}M*ReDCA~(Jha8IPU#@z+oW!`$$fX7fiMkr#hL>S^r>#Y+IG!fuLF0NRdmFRf9Jh
zg@AK(m*(!*HudB!4ej%kVhc*Gn6M;@Z-BeadbX!wZiSP&6V2PVZ=uGdpDZs|qSDI0
zt8P~=yoHPvI=yKT?#BG=cCw9Fq$iF$96?-)Cjf5jJTw~sHpB$7tY*sEdSm~}2Mh0E
zZWb9V&enW5+aH|WID^+9H~UUkH|1bsLSWu~yv$VZICYHiSqUs%?yFLU!bH<c|CgNa
zFCNG6SDam4-6_#N`v+Dd5LW<sK*xyjSuw?8rqR)Osuem@oOpjWDfx#vFeQcDM8ywX
z7u)5tY-%Q6x>$1oJqP3%yZXBLMD3)oykp}r6~he)OCl2>$0II(Y9is(>plHoV7W?a
z@>M8WcV)LrDD7Q182+&gLer*!bfp238343zmnr=arDEVbU2l*s_`(q~_D~>e=sIhY
zk$rld^XiX=UU9l_-g+tj5{jrk$-3Ap^MIajCos94ANZMQ+yUxXqjv!U8scI%N*MbB
z0QM%0gvVn5ihl)?u52R8F+MQRu203B2k*m<$cS;sOPK_TiR(z|?Doj>I}~PxtjY?a
zTIHAd|HR_fRk&VtHHPE|Nip<sv$t2KQ2BPU#b`v=aU*I-QYJtG?q1t2#t%NPXHU4W
z2&JEcgHPuCTTmdKULflGQ7nY?&C;X5#-}WR*;(+`9M?VLb3KR_JgKdo)>3vlJK0q#
zv^jJ>onJu`R0efK;hLd#!!ryTmbU;MHtAS_X-Ki<gnFg@+G1|1nzZxr>JyafA?4{%
zkw@6H5WK^04kOJnkIH`DFw-UJZ|Efdx?HRu0#GYud8j(6g!Z%zRrh;{8enR9*WBu3
zcy&tFhQP*91`vn$M0II6jw($T9eaW~GS?@H+rx+%TIX`cs9PyPmQd2J)DG^uZCjU4
z0pFD6WpI?KB#JbQ&ztfR!vPdl8s%QSMCUm5gilNw4QxIxA-rGf2J|dpyFSMRkE1+B
z?QPEwzbZ^SGJ27|^`ZWU1^sKKiyq#3B)Hp1;AZ;p&3m{?qX#k0ytYYQPF3h^8Aeg9
zfWyx0UhyQ^8pIuAqbHmA(OvUbl26H|noF{&&0xFZWk5BV&-vh^P<3{;?3#JIBTzN-
zgJQ=Bc=%89T6c0qvh7xGjTb50#{KdPdeFOc9n=A+(Kj)}bjKmTf!Isk@HS`b#Zl`R
zr;b6t1u{?1+aEko&*_LDRp;<&OFai6#0+YDWomkmN$Fx(m-KhXA7H~GIlJX(2+v4a
zvOZiI=*&s?toed=CGIv10woXcQ?GQYI8+r;0Z_E3s4^xXN6aczpUf`$(585-xrx+9
zX|IzcV_i)xA%3_m5O&T`ZI@;l)O%#SHj4nnQZ!mf+M&CDyr!HepI*_@IN#{c^jYmd
z&@<AvMZ8?NfagJ`HJ0f{eIK$Clm-uE7|8kJQO^GJY4hhpc@>22KUB`j{pnxl(BD6>
z#xwx<v36491DrK}pFsS-!SPZj`Th*&i@MM|2ddY=BJ-2@o1EkSedPLINjB&>krdsl
ze_R*5u-J$`|8n>G$Lk~1K*!O~_shX0ZR#V^xqFjH{C{2@@KfIdZ|ssp_I(K|0GRU4
z_VbVbVF&#y=Fye*d`Qb-Uy;JA3TR$Lb^lKahu65#qqhO!Tn-5c11H}vMyAQ@7qRef
z54!6Y8ofmoAC}GQ&G9QA{ZWw>NuqacL-mgz`T6Rgz%}(?ihkwe|KWa2q-px}^Ijwg
zzmVG<UUvYHZ)JCfe~ME6yaGZ6bih?{VCX>TJ`kb`MgOus`)ibbKl(}1=s?vhtN00X
zlP55ZIIO~WCH@c7h@UN#=4-UEd++>PU>22E7sve{)N{3mL@%TlkIoMZb8{}OKAFlS
zT>f8%{FA7nV#=dOq16xB*bJk?ff;xQKj8Ow^Y8b1ixKd}<c&X3f9@&is-@74_3xAV
z-{1Ue<;?Qq5NcY-CvCJ^06`7ieeZ_<wW5IU^rJ-cX4YL%R~`vtP#N(Z{Yx4+SXCRz
z{15<$mz5dP3=C|V>{bJQC+K{&kB)<blalwXQKAoK0!S?kYnG2YsD9oeK*c$o%JaCt
zI)hVucCBl?telQx{Yf4Tyd4pv^JRWsk%^7TzYjE<|KC%~YxKSql|#YI&>L5|@UL+!
zOiVAo8e0_i{ru{AZBdUIMg}6)yC@a@6Ea|pPJVj#>)EddioSncfRjs;pVF;Kh>!W-
z1GjT)6@%kU8gMKv9=z8>#PIXC0oN-G=5`QS(}0>aTO<kK4^pTb_)A~`4D4z+;muD?
zDM;{d=9gC`YZ5`LvD(N%@x4N|!7Z7lzH2c~AMTq7^gfcu*hE4af+3G;A136nt^6E9
z!XorDk%gO|BvT-akl(Gq0@@^kb^z4p6eG?28X?gZI$Cb>fF2mQSG1v_H0mb{CeYg;
z1$shZDt92MKk<4R@N+j4HOGxG%B3k!Db2$;|BW^z3oym#T=fBwt!3&L?%ZHtWUR1V
z;$gqsR*zi?d?rLmwT5N%{cw`<Ozw?C)y#V5&j~)LhgQ$HvD>sV_;Z-n(><X#X>BE^
zs<{6$>HXI`APf?93YD}5i<AAOj^%g+Jor0WEchhX4}#|E=Iwwlk|5RI-z@@N4N=5G
zAJX}+ni`=9aC?nCC5y5@->s`DipqeOmp^Lm@oxCd{SjJW9>A%n(yhTCe!WU_1XSVu
zE=^m>er{xq{-V~_khnOUjNdU7XK(GqhtSpiSs5*02QGV4TfF<{N~6KY%+<Wh#765P
z`9M=z{+f^ZA8TvpZZFdH08UN*=a$-71MElz9(a5h*KhUWC49_EfBoU-zx(j5C$)|u
zVrr|u6!0|5*P5;hnP#+~dWyzWQ3ym$FoAUCSA)8LKe%)-PqsgY{HfO8r3Eh9Ra2__
zr?oOrT9xhqzG+vcb?!IH4BDF~+l-I!aDUy2-?hM2ciqN1|L5b}ePZ?LS=_EOgpd8d
zZdFti_(_7l`^l5U%|8&H53UUm|Kn&OeC~;Kub<7o^>g&R{<xM=`?8UK{@-2+*WXlf
zZJxDNUZnEh*Ylgr(2w@|c2;$*zyH@|{Kr-O{PUH-HNYDz;PR)f<bOZke|_Bl|6h8j
zrU2X^@1HN_P%`P;w?UyHpsv*yr2v2E<wk;W@%b#<$Pm4BNu<`Yrf}i%tq~9`&MO04
znHXDU6lEG?CgR8Zp*~+;*6q-`lq{R<&%ZW;INS$<uChnlB9Dx^6GFy7tWTam0WzPA
zTFj})!D9^wBqmP4Lu`+2Nvh)c$@{21cGzXqC+atNG79m+%+8xvTL5Nfb86$|;TXAc
zs41VoPiI1ryH@Z($X)T^A}<Xfxs5Fl7)3`#QQHA#FP~C6s<dY{I|?*=>}8~-vlFRM
zk<5J_?mssE=t+RnFm~F%_4sRD{QIOhd{*m9kTKJ=<`{N%X$T-o5z$dcL(PtMhxLM3
z3SaM#a(w9H)&KRxP1Hp(h@d%|mZ-ky$VjGvjKI7wt#!N|AqKyO%@@vxs}KGCYYGxI
zN|Df<iQ^Z*x!W0Qt(O+^9#uzI=+N+%!rJvX8*+DK#N!b{#4<AuVbvE*$?*+gx7o`u
z@0^U4t%i_v*y5wX;{@rq<E7GbanTR<ny!zyt=1F=QE|K&FKH_nI6b|&H$6#z#P6`)
z`_`ABt>kiKZ($V3$c)Fx;r_dNzS2LST|O4uJ7EijW<$>KO<(JVG|7EKv6fF8gB!c!
z6V&6h3sNw*CCqJ^yxl!z3h#FdJS&}LC=l50mG)YHrdzp|@<!qOysnepkt@8*U+?qF
zqVyo<>dQ|uBcr0ETGp+#!u=(-pQSD7`fs1{iP#)`T;-w+o#<HMz0a=Ae9FN4Tm%kE
zm-8!Tg8pXF1H_d?^yM+el{)Urt1}aafD=Kj?kCrJ*|4BLe(j2))x^9XR=-n(X@v8n
zE5v5w!b&%bYk<neom=_687o}UyqZVeR`;cf_>E;`Wbk-8wYV+&q<H|-lBm3C8pv8E
zt<SAV5^>o$VcTxDJ{EjS1iW_gI~6+<(-4JzvhW6vdC~bxd3k7Qjz*<aQT+o1vD#Fs
z24fjYodF+4u{VC;@kpJ!2frc{sQgtf#2WG7@v*Jnq*dJgbZ*AGLIt5_ertITZ}7XU
zjI3;^fF;bt&0dQb_aV^p@$2H^qkX4p-9p2N(XO?_OtXm!7X*c7*bvQrtT*$ZsE7x-
zto7Wpgg8Vr<9+sbocm6~9=WC4qjY+*R>9ntwi!UD1{-zXnr4R^BLBI+|6Kpy8PU5P
z!^n-3X9)V&Gk`%m+^@79{<W+>l8jauA>dUT_O5#~QxGn>I1U5+VlLDE$Md97*{Unk
zgBDX+pW=u6iEyK`p%jdP2TqkD^UY7NWD|I$Nm<nB0ZbbDFnma7yPky8zCYSnCB$TM
z9B*Rs*1FiG%UCV8)@vV`;zpkVaECaU^@o^mAQ>O;;w#pQlpO2S?0gil^@aXv4Fy-H
zO$gKCbLIk39|CT(>FwOP2#-Ed?<j4Lk*yIN##-ION>=W0|Dt?@05dWk6F(4h6T=_%
zWE&38qph-PG@sVpCQnLcg{$fH?N>ufjb_9W^@dQ8bE~>NKiDA6Lq^ehfe$d8xKY)+
zjf-2y@}{r+YT}mhxFJT~&;L5L|6_x-QWixN4P*_zBq3wXWqY3`eC>8Xoy)oMD-ZLO
z678_r<mRy!=iVo102%d6G{vUrv~0W4e)d%l)w9r1ks0=WC2(uNV;ECe^Qpp`sck|%
zPsY8_<RqZ^9s^9`ROC}}RdaGf%O*e2kCSKS5C`QbqG@JXwSl({V+bMXVt5sk?oQl%
zC;SsY0PWLb1WK0yg%%6kWXA0rZ=?n~F<nX;zd*>y*7_^-1RRUh=bI+t3K^$sbOxf>
z?_+M1x}AyOS1OpxlMx9l9MiG0&vec9rrYBp(nh#{cs-Gq7DlmtS|6G9?^nFM7s1oj
zJ8Ry6br_y7B`}fOlz!pQHd06hbn+>6TLIdvX?W8M3CR(4mE>%f4n;KX68N_AjEHN8
z#X|dVhA(7_R*e*e!0qoaYz}(7+#Ra6Jlk{@z+KMCWbE}NZ1LkUoWd#?;QAng-x<tZ
zx9-)cu^2?s9lG#cb_c!nxCS>xMq)v~PPN+G#yfS(Iq;PJUW$Ug;6sa7MU2KeV&=v4
z9V3-uH@dI5%Eh#&)5qvzV>?L$x=!dw4A>9SE5pY`L15WGKyQ}W-x<g-b-H_m#UCTy
zZ%r9~U+ZhYZlgeUmZ-6SF~k*=avM<T4&Ut?4%xZXc;Y&7#jy|VKeKV7QFhBY2~8WL
ziyu+)nw1TeTZ=mibbmBde$2)e7QkRt54p*3>-~9mb$2${a{5gnjJsg}oIs1U3}B|R
zz>gkSwVWz^i!NlK6e!+B;`7Z^h-*N`PfVVF1xT^&+to{G;(C{yma8SoiGUbWMrz+i
zozB+0nw5)7GM0x~#33H&e13;mV(uU&sJsE7XVQQg;`4%WN(N1rF}c^cHKGfj;N0!M
zLrO{tr#CMal+A~lcgvtyxJ^cD-Is?gMxUIK$#SFJ)e>;Lr79^CYaIG@08H~((R6}e
ziW3B+-;0elzllk_PKaq*A714`zJYFE$(emCQT;0m=3owujV0$y8Lm)6#%U_kS^7+E
z4(Z&$+g=?`ORXrFm>z55sy03Hm19uVk7a@<^E8*xzm>S-(nCM$WL)csC8ycR?7^??
zL&y+BJ!dPZF0FeKs@b<k9$PIY<9dgxw<S(|HVBePa%l+u;<Wj_u5{3cD$2rgzI@I-
z^$E`p1Ex5NJ!khE+jfwoF<&4ZN!=so3*we{cy<6w@^jbo(1F5=9<E-B3wee+UC(D{
z+B-rc&g{;y<?J}p^y6*>RG+sOnXKNH-|^{bPaqR=u=H$N0~8rSeAuW`=Pkoy@(?PH
zRy-R(<WYFuyHyOC>i6IhNE}E}Yc&v4DG!b(-%Jv_)4ie$5|e#8(uucy%bJD%YU`M^
zQsy&3w|z!6@%q;R&WQ;U2lc!XTb4?T@5_OGOH7#;?2?Zee~U@aYkSfJoFJN~L5mZw
z@~{Se?Z<-{UJMN4byG#@H-(h&Xn3901GUn53K}B4ZdH0cjc1g5i_5BDbFm0Xgq^>C
zi&OCeRaa@Z#Lu(e_)uxM4kx(WX^$5^4X0+~pyy+I!AvasbXq@Uhnv-D=jtMjo7ra|
zFOB`3)4rP#_pFvB_lBGSD@n4#2?50<yN+x*K5I<M7$(4hW@HR3fYS7_kSRmdpFac1
zpOl$1q&5O`cI+W}v}f3Es#}lCj2B%Bz33MiY1dq@aQ|mL1#S~RVyG_4^7gON9v`TG
zn7Hzhn%pFtzB10(-^X?3e9XJCc4*xmZtRPbW2vgBXu=6Jcs<UgH21&FDmQN~B%4l%
z7YL|)2(_TboX_;%+8*H`uTtWNd&023n~hfvtbj(#UyGCJT}?)6?Q-FjYO$i(t|^|L
zmU0-F7Z-PxlnNFJb-hiI6A=NW2NP7h#M|lX(J_(!VoXZ(H&L7#@Bq$i4^O#^uQW}&
z0dv*slqVnES#J`cE)){gc{Afv@>y7}n=l1sMYP{?x$&A!@I5*h8Vl=i4%dQV-L&Hx
zBd(IO%t{1;G6(zZ(cx)ujon&daXZ8!L<55p?DL6RC?qft%xAUpJu}9!<laXu)m`zT
z+rZhWC>)LGoE{vZ=%{e3l&NH*dmWR`lOJDdy^jwYs*edd(z5KdoT?F6-N9ur7Ka%U
z#^_2X8H<X6<t$G&_}1<5%H;E?w7z-FPzl3a54#A=)`&A~x6X1LIQuX~JL803Dk^%T
zY;8A!g;KDt<b34$9a(x(UcXXZg*x=<4cx%2o(>{6Z2Mv9rsKm{wP6XGxHAlv`N(Zw
zu`t{wERO$x-vso9H(uV*w;|oEQ{-K|9z0$+-pRDvl?Oxi+#%<iV@Pgy)@wwMMsfdT
zQ93iER~j)rl`|xOWH0s-j9oGS?2N_oT9(b>i?efj4j$yUIBZ61L(hk;KR;<+PZ|Ku
z%K~!!R5gRR;S3cXwF<*jkSV&*GfO3G2!zy}94zrSKpMvkd(;^6>_9+P{o*zzs#(Cj
zCkpiEd%bATB_*4dm)9BK5~y7s=MOuuAM*bG3_%oagD+Ek%re)tGUOBYE68S^{_yH)
z2J1Xaj2+p$D{ngfEi4eEM&3W57|(W|OV4?bUY;NW*hMb_jO>Z*+W<e0J4v@SsfM#h
zf1?*RDvotjQb#^#5s?wIM@KHpp|_41eyST|XEsyWUq_-&)r51q&u5Tgu&JqOIj%zn
zb`mG*M^-H#D<%$0fI)Ba*=&9kPzUxRVXVcY;w3ONnqpNl{bO^|RkgX5w?Mw`GYrFn
zSQBUn5G2kc%KP^P9JXwISiv{w-&lQ`%C8Wvi^Wpl9TNlJSTL5XAjT8zy+M}2Cq}TI
zR7X1(i#55*tO{3jXkzuBLrjiIeZS&FYyw?XpLG$CjY=)M=8v)s_sn{_3vKajPI-*J
zQ<=(xzU`a6)%7f;j(!}h*HWMMV`}ye-t!nUbVz+X!o%~^ROo5)xs^A#SV82od>a|-
zSc*q;$r=Zr*L@&h&reeW<9^v!J9<Y9NB^7fvjshE*`;;y){M%qUP3vqJ>Whu&uK&S
zYs1ro)qY(voucS<F5>B8*O?nHrJ5RdUKpJ?gDqIaPTz5`cww4oDAl_#*iTp&tHeMj
zAo1K6%3#{mrPS<C+?c#Vu@=V#|BPh;93c}}+?W%V<-*nPx6Pg9VP#uTuZrrbsrMO5
za!H>J9Y^7)P$2wlg;wuisT$|kmG_Bna)Ud3!z)CKz>7zaKk~OQ7)<@4GJPt5N%bVf
zDH=xVYVUrZ4xTxAV1JRzO1|mZ$%`%xEJ?sNZ>n?TD>={#?GBuY)uq5>YdJ~J4i+4P
z^^{kQ66^yQRV>ATyAu(QDV`LX-=T;z?amWSVq^vlMC|*yQe6oQ(|o09#*1PJcdLip
zlq||AHigkK%h7~kA&f=4&iA3?J$}&W7{rG(n3JRr8*pTjfg3(KdXRs#KGIB|Sc)L&
z!|!sRD<<uNjrT-ejggyp2iJ353i9RfONaAL%>Re&@Gq#rEBgLZ?fcM}&z1ECw}Ve@
zvsV}aa6$-m!|LqUvoMB)5;J@d)p+|h{a|&tO@5?_T3uSqa&q(ZlJqYx0Mk;Oc>Lkw
zgF654rIm$A8Uk*%A;AJ<gy)WSLR{RVvKdY_wk2&+R;9i{!LB@>s*=HvSn9^he(l^V
z>BWWtkDsbBbjKrf3hzC5O3&l`cAa`6c1jFGCKhb*nB0yqd+jJb@R2%cs%74Goo}08
z^5xR#gsq(6M<O*dZ!?Q(h;FZSpTl~Mv`m0(+ZVLEDjcYgyQND@OC3wz&m21oh%+QQ
zQUP5_-PVA=J*>3th!DS9w2O);3!)QbD^!g}sgB>&_8?5Ry8{Av^%ki;h&$aUF;I&;
zHNGY_st&jXaf{K#&PiY-f^%jogC<;kmy0_~V_3mS42m13(jKcaFsy!K93u0x(?kyY
zs(1(Yh}b*qv271E0d>*DsO!<-bCj!b=&j${Rik0%-Zti*`uOqDe&u+}n~PhTm!-V!
z-Tt6I?F+pg-#kb!flgVCmxnEl5yYKtoC+`b_LXLTD(Wd7e8WTMeRRmd<9~jzsSAiH
z^}MT*oMMFX4w!D_sGUCI(qix6A&O4^JRl7sc8uHFx#+#6XC5mgDnl!qIPXF?4gzPr
zWVM`+c(ZEIEr#-{*4cgHvhTXG?RK|F=maU@!(3nTN#<?EO8U}cfcv=-bO1C&?a)+W
zlk5;Sn=b9BCZo1EuY<02UA<h8Y8~Ek`4+=8m9q5m!1uEJb)AS(H`nb_)~$&OBahmY
z5{lAsBKFwGItaO6a^AWzx!E^4t+?4QFDB&VUXrpZhewA#20D<0`ASu57HKX<>%`HK
zB&uk=#!&T*-Q%VlF9~)QnD&B|A#IETWSwGJCmdlSN?KDjKrR6vm}3CHc;OTL(7+9+
z3=jHS11Z2g;q1}~Oz@(qNUSP>0*!Rn+2Aio@Zl}Ms=S{5#}vgb(7uz<4FYtn#>6x|
z16P;Ur#w`0n0=N{d$i`xYVdjtIz-bTEJST4botpBF}<WPhpn-_wQne-T4Unl7`Do{
zAv3``z%cH3X+6@J$!!JD#P+$r*K56!BbMjJ=^{^BdRc6s|7{`(3-5*B8~Ieu&&g`s
zai*y}S0x9Z2f+Cah?1u*1L?}R8dv%OOltiV;sR}*cc>mGk?{mmF%!J9E=?s9k4vXl
zxkhZ3c;ogSk>BI^?qO-<cTUAt`%J{h<%qHng)tvE;EB;C!C~2VJ5d&B=3yVX+~sZL
zW95)8#w9=YM@A1y3l#A{#CqJp)Ys1#_D%{9e~-J1MlmvdB`Ze0bZb=a>dw2a%H%|k
zE_KB*GsbDXUXhW{_|(S^ZRxB6TZxXyL{>b`z8+JI{gvHxik7p$k(dUlkdDGg3bW8&
z$})MJn3oJXlKj8N9;DEh)L4jc3jG>P0q5Zd@7~Dc{0g*Qptl6WX1}DV_%PeKdY>w2
zTrG86gH#Uug9Kd;geIN9A@zexhh~uD<GSSpf6Jt-$ZFAO#ml^S=OT5GMs(RmKfr-z
zB`;6Tky-gchJKJ<I^8-fi|(v6L2=8P6}$BPz5~1x$``}y5RFIP3!5L0V<FFE5b!I*
zqa0M2juz^{ngD37@)mJV`{ASUY(P@C-=t!B0mr>yz#>>R$KU{uN@56uHT*n=Z!(V3
zsopzwRyqPmx&Ti>8lOtouxuJ(NuXA0QZ4j3jX}V7pgqs%TL4)Nl|R-u-ig7j!wt%+
zZ(oWOP>w>)m)Q6+U)!FBnH6e3;2D{OklPXEMn*)aGvn>>>Uyih(J&zSf&MV1iDPma
zj?Qq5bgqr(aBe=d%kctaEiJ{pIfPX}n|jK~Gq;j05v#`T`_ZQNH^+@%5>|7Mga7zu
z+fsV%?Dg*;<zIh_x&amB6k$z+?z_M6;gH(C1xACuelCmU+dU~JF?Sr4O4Yd^EIkLG
z*8+LRX`uT6Ccz>siLeWOR$s}=8_#Zbe#vyJYcb`CA=S=U%uHYzrc94T{A3!?)`7Td
zJBep>Bzs2J2M4^fpgytapQ_)sXU`oj@ue-=dU?`mtYpQr$50@a;%FM0kN{X&+2t=i
zK8sakNsfgs_XggdPOoE`zCo&$zLG3PZiy)Coz5PSeHbg(Kw)OZRa+0|BAyNR7{d!)
zv-GwH^0>xt5@ekDd8?)J3Bnt>#ZDVTh4(Ft=ot@sKHB2b`5r~RpjW-%NpxhIF}$&h
zR8etrf*F40wJ7`^VbsZMAQCmbRF}@RN*kNrT!+qb04fC#6}W$M5c&8Wu2?zAA`fGP
zVQ23K$J4T?0wKK?h|^acxsjfyE8prsE}bQewGX3--2NKbe=BMDMZ_W}54>2^U9VZd
z+9Tmbzjk$+Y|rVcBP<nT-y3d<Q!WjaVt_y#SCL$lCtHo(jzUc>hSRWdHtF0%zdf)~
z9T0Zv2R+ym4j&T55Gb|>(AxffSCEwP(}pv-2}zJRfc7f;;;(CDF7$HZ$G)z0A=b9^
zoDnpLgU2s3xO|+J9pPYNXn3|FZ=Pl^kLy?gO7>!u-|-(<K}}aG%kT_%6>ZkmrE=qb
z1|h4AVte{V!+Z?tzAAHncz-pooX*mLX;e^=jMR_Z=@KsJ#C&C(FAnr@e>4y=v%}rB
zSKziHY#UdtzpzYch*0SWUo9t6%5B^o?lIQ&wygc)Ug}dKECe4%_4RGcCPKtpjAB^s
zU7p#M)jb7Bpj>*y2CK8~DF!mdB_K7W;AqZA)9GH1r#;Dy#^MM_+nTCmy3KkWSqqe6
zbSUshZO|pW&8^1T506Ie;p^)WH`d>yyMxOmWKuWB9D5Vs;bL^rK1EyC=2M0>y|a{E
zhGC7UNBrPlo$vm0C0Hrfy52OYS+BAfi^$uK&WybWEQ$Y*v8#@&GTpigh=eGLz|e?<
zbhng*bazXGba#q~C`h-2q;z+efOK={?(Y1aqjSHxchqtIncrcAbKdjh-fOSD7OCAB
z%lC4Roj9+B^E%3j%`)xiyIVfD4}vR@BhGPC4HA3BZ#AffR1{=Wvy-!fdun}1K4I#~
zaBL^Fn%?so<eEp>H%UN4ne|8?+3{gDGn!XKaYkU87iHFnS!rxPPA|w65JydWw&L~E
z+i8;cD4l((Tt@C5ox$C?)0ecI*zizt7!_r>T(3JxLDFVb$K3nn(e0XwT(i9l8J0hq
zJmg%C2?`tVoQt?d;O@&CQt|}aFHPJXvl*Npc)&28Ymi?K*tcG{(<ut&?yQVZt|W=a
zGmRffA<4E1peW@`nsTK{VG-C~O2<22C8<^-zUw1=mz=B<yUEvPta2Yve3K9<kQ5YK
zJYr$3UB(H?Hl<i@(z27zB)O@D$~4mfi$!QfL7wi-^^xrrUMM-SeD7;HXMwiEwLZb(
z>Yy^?#4rO#Ec)y^YK2do2m6C*3ficFMVbzwn*y*@Od0&0kQKxCKP$fjthE8|fUhWP
z+6x|sPs<d#Wz<FjK7B|0W@2uU2I3iXpy37t#3lnqSvaVSjBea1n{?&(d_Q%g)^az*
z%~=c6B=ZiLXQPXcW~Z)Et0%x<t2$v>6#egZ!R$;`r!K8tF;qnuq#+!JT}R#bS*WQH
z+W<;=3{9bnk0xji4X)K}mLQL?pL|+Q#A0NcNoj@)UXie%UYyZmah%5O0T+8r>{%`O
zDvQ~eaS^KNWG%#i4gp;qZA-b4dY};BPs08-7Y!FTS^cwtd1Ep0>N6fp(8H0(&mZdc
ztW$%nZENzXm6~&c3Vq5wp4$nu75Ty=P-)dPcoc9>H^X5WTLzTHgIYt{#{SW%)XUe%
z>G2$Hk_;CNrfi*Y#{xFET}82=k*DT;I}C%Xa4wU@@SuUW6~>uiz2#ZnqDGLz>F2#0
zI9)o)Sj=&<kGky-_vvg}i8-@U^^1+kBQ3F*n9u^`Oa#Q5QMF%;o&2nF?8)nS1B(hq
zSf|$KMnF{Et6(YKI`xrsyPDOe$9XLX-_Q_8^j?OWyVm^j7UM7Xr`~84IDcyIwUVJ(
zT!!@HOs%ir8E&~%Nx;$`ye>4z&j@7k*cD}!^PGA!+>enI1OH&fx(m*-U9V8NP$&K!
zJ~!Q<WtkSE6q`h+4BSg+8?7Rn6$v|(FUp0kIqwJ<2Mt&LkZa}lp)IDSYjf+AK%0E`
z2RCdM+oy)R^Uv>v=Wrf<*=eZQQaqR@9S?a|HDH^q3bLko+gsHZkb;qA=i7Ja@M7?E
zue}~OV2sr=l6uzj)a!t<gt_AT50uUr8&fX42k5!0rxRQ+RqIiyF>Sb1U&sXGEHGN1
zczzgB2$F*hihJ4rNv-XT-$<<nC41ds@@0{ttSmNaIhqyh&q-T}_y}PhWSgf9mH2Ts
z)d@h{oY~R$FfB;#_R~T0;?i>NTjehXSfcBUe2JrJ%X!?cxzcE<2PtHZU!EOfUhTeq
z7>ffn4T*Hwt|eKs!}`;~(F|b$WHhxterCo2cU4&)Jsa?ZGdr`b9ROj++LXNgS)2oP
zRigfb@zVBPk&dNn`8vj0ifwS_Saqr!AtH4B@gURqHFDRUw5%+VpSVg8O00wMbuCw8
zZfPEH-s<5}A4a5hv$|+DW_P|w!R53PWm<m@)uHe8H5lzJBnFUiegjOTgbEt<d=ekl
z&=}L3W*(#Fis%HRj&w&ze#2#T(aiyDrnxTS**m?~^iwymH%Hdv=T1%8bW3dgJ;Yer
zWaNVtnmkk#mYFS+SYi#}`jmX&vHCyj&#=1$nFX|KCrdEq0J+%iswXS9WM~MV#-s+-
z8AFZW(SHurA)Fu?d7*y0T4j7jY&wrL3VY!sUcJ0LJNO`i_8dY6pBF`Y$h=z|DmkGd
zmC$&xOt~%WAs0{`nR$(8Hc+{7Hn@=~XSmSZF1Li(gv@=o^l9?${P%!0gWQW!U+LQE
zv~|PDoc!LTmW-8E^Q0!zB2V$EgkCx`v&S`h^oGQH99Gah*XnLxL|QniXWm=i#*(8J
zqwRfUpAmZb%uuz=<OM7Zl7w_}#M=xLwG#Ux=slI2i$$L{MK=yFD>VLbYVae2RoK@n
z?ckahHC=ixUl>?h`1P^->K~vy=}lXHnE9Sz3|Z5~C3n*PE#g^k8VNa*?(?ZsF_xxR
z5$|Qty7UrLiM_+?MoSFqaG7pmPbr#uKl|Lz^yxLrvSadOE6V)q+joXc@hL}43oo(1
zCSi#mwAL9|{=sN1Eq{F#ZP)q!CggCtSJ(Y8=bq^?$gG<TNO!f$N|#Cm#L)H)AE|TN
zE|k|Caipn59z9~za4ZWp_EusNR_0*@X$Db!V?YzrkK_n&kOz7WBP6wie`B(-FVm{h
zv>jy1?e{w`y16(Uv#;%A(g^eMIlZ+_xMd)epgME|*$l|ugnC_`kFIL-#4wxT9+3BF
z>#lwW@e)77kcqSSI;G(@si2)KcL^lw!^frR=M-<Ww7;!uIxwi1ni$?AmF(=HwPX}@
zvLKX-q3g$*t99v=E|34{PyZNx?7<^I9&qT?UzxM|^m<hH4u<KB%4A@4m1RXu0q~-$
z&`w5#`_fT~eKLzjU|RX237(-LY&)vVC{-i70XvdLBVvLunrb+Edd{3E#`<oI$h^>e
z2<YZZD`)9aOv;gIqH?LmrppJ(Gfq^#PM44kl~*+D0XB%{4N(dxm3I9!djO$+W8BzS
zDbHA+H679yLZ_-Js0R9L#IboA;Sv_wZ|_HrdBz_v(S*ySEr@c4mJln(T*c<`85|_U
zTQ&%@Qx`q{Xb<gyQcvQmadWm0V4K3~<w*DK(wv1mXS@7)hy!L~7f7tO*p(L|E+Gzc
z8lDYLyL)Wuusdu$Yp&U*FR5XTk)Irw{VI{FEmhd9mftvt#q-cg4bnA`e}*(<Fv+37
z5oO!+jVb1JMrldE9kMZ6!b4mSpcU3d+NDMRh(?4Vu*_-9K<{7C(sreO)^GqkKO4!T
z1A`s^xAlmIZe2%-d=FyYC)xOwnjgW{c;#ZD7uT>l_!qx~bihAbnNCvOYBUB!>F&9i
zcPPi221m<g*sd!~%O~+p%Nj)_uI1Vl)Z0mJpdPA%78CmwE8Gl(S)KU}6{I*#LRD*u
z`^QKAp@w>mqe-NWi5CfGF*S251)=s8_z4Et)-iG+bJgr=454_$^t&ehNEWDX4O}g|
z+sJ8?Nr-J-u=SEj1TSD!Z3$vZcr8A=#2Re$u0J#zD8P|0MkLdXjJU`!9k78|h_;(f
zRd%8j_SN4<A?h<FbjA-QefQzsGHqE~X|o<(AMXQdw>d3*fuQG%kpcs6#f9#qXG(ds
zPoYYP(R&d%tM`>!dXy$l3hB45E}JhMh%6P^kHbOEuD1DJyf{rb)rPE_6l#QNkcX;=
z2LCEHLKwV#bJ{|By$NfipS_fYH5!9jgF1Doy=Qg_vG1U`_|0?ci3zrt%Pm)jiLY3p
zlQ@#*;1K;5^vZwogh$?BX>bdUnZmutDw$w3as#=>X)|nh$6U_m*1_vKkX|JsEtA$c
zypt?5i`Rw4u;s*7&UahZyne(&$!|Z<ahuVRH6=Q+SfI6`sm-(uEz0t2gFTdQEzAi1
z9jsqRq|^1?5yD$27rpZXX{@KrKF^x6;q7ZFcsLwPUozMccZMvkZ`dFH0kCFe1MygG
z*7Q<r;iPlir}eCi;qa8W49kgQ<eKi68a8q~^fRbTY<!I9Y!a}PaV$E*9<iq{Qpm~Y
zx-jObwtI6+4vsrBtWH*}LPaPKBpfx4-#aYD;Ot`W1RJ$+XmU(vxnLUZ-=AjAcyU<S
zwWNyY&}B!0Vh~MdA0>tA<4#H%TSzIUn9Uf?yJyz=?v|LM2yhnYLh3Id%8>WrDkc`w
zhm51B32OBfZ)tFL0Q~(Emg-AgrgJ|zbA$CGE>gC&>O=XLW@pi2d0*V;a8s}i@70v@
zxOA8P$<zChz@@_nsJZL5;RiyE<93TB4d<tmQ3&=35QnW`_HAlQ%=awh9gwJ4@mbYE
zCkE@au-)0$a8#VyvL+Ve>K`9H)_#QdZ59tD2^kqDBUH+V(MDt+jx$+4voK;cAC!(^
zen~|weuJ0MMo!UUNoatncNSjGM7J&1&3fXg#`Z+ih@G6AT&j~j8WuyM1yd{%Cn+tD
z*09;cQlPOya4H`oyHNz3#mnj3aLz0hM2=z$GX@d7`_wIT_6ynxu@_tQqLN11?e8+@
z7hyxO=Jyt1RtK`Os&1jES2%UBja6Oe17Rr|I<(0u+Ye3X4Yy4esjKGjmc`dm?7Bo}
znW*I$)+2D@RgN|CUH~kI&xhuK&5ma_HxpjzA$<GM2S~b0FCnT!8y{(P1A0t&vQkiY
zf4d$R;&2R$u^!&xSvHHlf~csdX)AKgu%xGNX#3;Qvy!27uSY0PM@agusx3N_j8H}F
zO61aeI%MS1njuVtpBYrxs@EC5dB1}L@b8IvGgKnkZby_xrB<labaEA61+;AFtt6^D
za1Xw6U!BYmYm*M}Intt+$_G^3u-B3x=1rVJU^4nZo-Uz+&{dl#l`?I$9%jPna9On)
z5lgB$aKuE(uh1dv=n$jNAQ~2BiZ;c7g<U<XXWV-SAx46~6VFJ<k0u@h8G@7<>w~=|
zRHO_HOC^~Ce9BN5J)W;~bfdUz7ru-huJ^-|Bl{ApOO~R6Gfmh@Wz)sP)=*bsG$p=O
zZT7jV_5RlDDosoj?%v%#CX^L~AzL-j&I5qJa3D7Txp5n8znPmTTfO@TBTO&78A+K*
zS~O{?Bh#?hoR2r=XpqZLG=>ZKac~fd+Tl8lM9@&QEme8uM*(BWuvBO&zDK3}eR^to
z<?^Qy&FiN^5T;(?fCeTr>yi|S3_F}?LTqM~Nd45%_GDRU8Ek<(FZc`FoD;_TCRGPN
zWoLo0KZGuQ4##j{6gwsv>pai4I<&tgmnF#qZS??CBOmEGbup~;w9G%v^_7PhJ?n;H
zS9wg%i;0y-SNsBu9O4x(Y<cU-_+gnpbCx%b)tW<i${STM)-*<T@4+B4A2}bs+<9C3
zxj!xT4%QTlUZRD{!|g)L0qiG<eB@VN<m74U?`ut}j4ea2D}-FcRmVxjhf8d3qfzxe
z+SoZVQ}eI=1<waOTP;k<wOQfzE}bEq&>ey&zG<x|;LhoP+Ce7f^@}>>|1u}MV$cAd
zuw?paYHxQwec|FXVCw`eZR))KswQ=G5Ru2E-`hp4(5aC@o?{ze%%G{M$Hnh?91V*X
z?$4OnKhii6ptN4cg!RIMmdNRYu^t=?j~T}8`1Y^Dk+%uTM<66hDPQBsv>Q{kzfov^
zJed9moAok*KJNIFmlphydo+-EGGbDp>`c6^C$UWC56$U2|L2!~d>GytY&g-*TwxKB
z0;h0MK+06Fux@!OE|K+)*BdJiFug@Y{6pg6^fbvs!ZK=@;7l~Z%7|etb$LQfZFj79
zRefHpQYIOWS9R18k;7?rT}(c|c(7VjXwoIb>9IeE7!nhslg`M%;5s}MjqK!9ny1jY
z#NakAz}a>0e~10YN6Wy0!B&=0R(>MbLSi@XsdMs#iBJ$$u0#BTXn~XeLlF>ErKG0b
zH@e?6!mM~V9(q=%?(OYOa-H550m8UKfcmgbV=-M?(j`#fJZVrNt4Io1reeqD@so>F
zh$s`59o~9|tcNr@P`CI0=GOl3qlrplMusvC=EKTp3Q$-U=jDCdUj2c}Ur#fme~7#&
znpaf3UbFy6|3K=)b<nDTquUa9xn?R#(F2(Ka1S0+8Co$TE@<q*DM#^)2FF}I+w9M*
z!7FWrQ+;KENfTz6*Spvi3)}-c{C}m0`kytBTL$_Hf)|wFl#H|mi%8!SXYr+>zyP-W
zniBtJF^s3`=ge^v1LP7#8*CV6Xt#qz`<dDCU?yWJS<40ANDOaZvAbI(<b{({<ibwh
zJf9M~$80hrA%3zs-rzf5yA&?@vg%Q3_uD_=oBw?B_W_c=S$jTf<H>vZ@JWxljC&DJ
z7GEU_JYh!i84t3_XD<ws>`iOF;0!JluNc%s0bz*@Mbjd&bbBIFE(zeeu`yaRa>Qab
zF`9M*ySJ{)wr)^Fwe;&<CpdtXqoaK3eL$F{Xc1OgwgC5z3)IyiBk%5D<lchDLcH0C
zE7y82yV!n~QSx3u5hst4Fve$uA-tIt&olJ-idWxX9~6JV+Inmjx;P8u<*LPQ4Rg24
z-w{rR6LVyWP*nB2gAq;|;O-01;`T%@w^XQD!-B9A{96p_nD&J0vqAt8iNW7pd29wL
zMyhD&jHDIo+0xk91L%Kyx5l|^YeYmuN9EhSN1!Oi?EC#ZDj@$5_C>3`z-f`q<i{mN
zSa@UlQh-GV4rZ5BT3OjhOwni+X}Q&-I$l@fDM5*ShwlnAkT{!!ndr(im#Nzf4-W^N
zj@tF3rNu=^CC}}<xWTsq6w5EC{KZZq7k!+>7RHt}L+~d#M4q)sl#OlEen9>hLKs`N
z5%vI0FUf4BUVWQ1>RV&`8+xx4;w#a>0<qTFpL7h#An!~{4a(mUAf$m~auJlV^5=^D
z`tDshTe>m_Vw-xN*LS(Q`pLMp!2`(fm<{H_DW+Da&60+i-(9nXh**mc8*gRt*ICWO
zC1T4U`asvVaQkfDH&PRHWZI89p_E23y$Fl1)$w7TLPHtvW!j@u%qhCW_DmY1`*YwZ
z6Bko{dZzD^gSLcSCk=K=xMzq!7!x8GBl0Nt0eVuDNw!wkkY=n6wTY>rY55db12X)P
z)%An&XY12i;YEOy!llRZB@Ao89QE?~;3i|bKj@P@n!QL3vG9Ikzd2^H98ZAFM4#wk
z0o9m3*qqb==GJDI#`&%7AT8;cf(`Mgs6Rqkn|=<KZoGoiw|0&<d|t2Mf*;D=3;Y=p
zimC;379Zq(%n#Jf&K?86=A#!kbLxe=_9Z(Y>81e_BI>p7U@PJRsDUktM#I4YliCzN
zE7*VXTYx!)=>6oHBqlnTX22ZR@c8MA3552qUq4)3I**^)7MV^&z1ZvZAqYp*?P6fe
z+e{;$CmDk?e2330U~L^E29W(peSUa3{AQwESa5*7vGNLgSg725Lb9;)lfI@KH*5D1
z36o&G$}zz~UbGzbmEW$!J_pZbS#@ySD}TNmZANAcyE+A}?S0rVJ2lA&oD}hiZ@T^J
z1RuBcGBu0xvwJT_Kiurkm7wf%zb>d$_o3r<pWX!G84=y0O|MZFbPe3MYhgoPc(eaG
z!?Yia3Nej^lIPxzyB2r|#r?$GQM&FH=$*Oj_5)*kC^pj(PI>f-{GP1K$Yg;-uHYsf
zqDvs_%`3&c2FjR-2#QM^ms==Ny^|)SNa@CDZ*w&`UgKuXgKlR-Je2H}fvl=;I%nYQ
zz(q)jejs}x<$$X${FBa?wHi<NwolWCpc^Rh3Ng%2X|QfTH^A^~Nk)JYNokP(jeyZw
zcU&%LJj#oUl#-tM3Z*=tNp30Z$2-HT{6Vky&>7yn|1bvT#K3KNQu#3s<@=u_tPG&6
zk|N9Ihsao2I>*8M5abe~j=(H7pl9llzhBS#+WxH`KBxrxNY>WFm#e>bKNUm!`Seou
zdFJfSh^#FC9w!redM0>svwfOgU2)+2Pek&D=jPSxpt_eGSl`~V+28Z(o|A!44Cp&=
zYsa#}{_#zCFOXIo*v>Qp-P6Hx{&ppLod6+u30F;}ZY1c>gn@AVvU0>E>>_Ui#8S&s
zhy%Od$^FE<QAdlAgnjHOVW@yM=w=e_zTp*xG3v3CuMU&~Yn*LSobdsZD;!u}Bz6|d
z`7}Y<_7xY+?E{Y@6%6<~F~-8Nx|QM9;-Jz0R){6<_eX_8TH!Gt=zl;m1g91j5fP&l
z9TwwpM+*cmwJ;-zKXWcjti1K$!2<y;k)<kZYS@slFukQL7h2PM;HzZB)o*c~FB;Z4
z4(A^91$_<ad;A*CbcRgoQfO}v$dqv60ZG&j2mH&oj5s@2C)Zb6yjltz?z;;PS%ui^
zgwlpkk~3qZH|(EP!a^{*E`{v|Nwhz^ul`JJVxoztsU?t*kryp!Ybh{k)Wc%a#Iqm1
zgMTkLz=X=LHEzE#t~+fvjRz?!86cDV7y>P5t>RBsjvj-o;VlS(n|;8T>{_C;-~REk
z`QLjxBNAq^$}S?$<8*&7cKTYOIIJxeNE$5DF0^wwe9c#JsD4s$>D*q7LsqU~(RC$V
zYCdgTL|PU9f)oh0kpb<IqQ~wBMOvLV<zn>47j_vATX##hC+}f=S+}Sm!xOJ>Y0`46
zKcZGIl-0agUVl;AK?JAs94!6O3PoJ<9~b%GvO82lDBsuh6bMx(a9wiZ=mkecE(e0p
z#9IwNMJ+P0KNE1`RRA!tWy^$y{bY9y$7VwsI8#Z$QY=;1)u&6r0ylWA>HXCr#tW#J
zC$!%T`KSG6e_Q52+j9q#n}ncSjf&ACddVP9x$H$&z(CGRLx)NRv~NP?AA>&{oQsNz
zP2RXmjX8I~Ny)aqy1FWfxsA{awWg$ZS!J{Q-RSeqeD;KYHbK8{P^Jqx(f|;!z0nT;
zd1=n1K1Xm^oZhk&5Gn?y)Zs$!3yG<j6)I!IF~^zCTL8$t=p{9w3w)_Y9cJ>Y_s%%&
z`yR?YOK};IY<N9Z3uMQ{Kx0m-@c`+gwtOIvP6e(Njam*|oZk<*z)Yj>)6-7m>kvw%
z4%di4Y<hv$cRIpHrd*Y#&MlLwOqbq~heSoKbgv8C3+>Zc-5V<1&hxmAbn7u6mapAk
zQQA1z3tykOfdaWd3)WIK!HZX4%$C^Frr5aW*x(7JF`?SS0ZP8Q&;EBM{v^(BM0kjf
z-VQ{ZV1*7g`i723l=RYjvAU?}-8bHN7F1N!=uJf3FSOjXc2E%&eaktS>{PmNw7v*T
zsi4|h5Ip1TyMt!^sBaXY?RJM9Rug}l<sW{Owzvf}jiHtTq&|Y|4V5UdZd7SfvBbW<
zg8f;BtN<se>R~V>oeV=c{SsgHMWZN93}iu@lA@I{i_gmPXw6iobgvtl?8ndNzXZ=G
zlL!IL<ETBt_?m!;)xb?irT+5?NmA4|e9-6FNw)g(!vp+n{eydtseW$Y+CDE}rzOu7
z5<lz@8t`xqc9!dEHU<dHm#%`QscR@0v`>N{g=Xnrojhti$~Y4u9Yn4U$2p;vRO1VX
zaLSH+QGqJZ&phh4n)%&ldh#0GL~-s_D$YqMe;Oc|vYT*z&_;w((CM*%zWfQhjSx;p
z5Nu%O3RzsT->xQU#GBM8vdZ$r0g<_dFYV}3IV$Zo7m9ojaY{!Y5R69$2X7To4St<C
z72~H)Ch#i909}{4w6tX3I|adcc;Ruh1zY=h&;8<rmD39q;Z+T5VdTc6CxrZb@)O9<
z<5NqnhljU}&CD)U)UrxU)Q%P9QTcdfUVBpU9Zkj36<zlnBQ9HRiJCV`lp`kijdXD>
zv-dc4+e|Gly<TwLKK{wA^zHWUR`LRNC;r^_n^TlPJty4z!nI<q@V@;SDl>a{IEfau
zs)^IuP|joO7*@o9wfYAB7bTzW!-PwKk?F7(WGDFT&aZz=XS%}Mx5@#SStQ^9kGOJk
z8PD04__!qbh{5_q(0s<%a3nEcR`ULyAI~+{%9T8kv<!6XWGDq%{R*A=Yi{}AK*GD&
zl*zbIC=)4cYMKpr%Zk7rodSo&ELdsURm((PDl`Y;(1E#+F<yf+oJR)jAnBLep@1Rt
zu98+eg?`k<NEGAme6PLs1qdlA)vgX3I}5FcR9A;neVB|VdJWRb=S$ZY`s$vN_yy^8
z@LXReX8}8LVuwlhn&y_4O1l<n@$$98xtP&Si|Ru@<xQhkyn^=uUXiR$-uipf_J@ym
zb`QqP1Az0!*bU^Ij0jbQTzaqj913sB#H)?H2R}z|<xto=EeB4vlqTJT35P0jbT}%o
zll^X9*eu|Vwt4=r%ms!_QhDa3SIz#R3~o!f=Q|cBb1%2QphpIM(B!Hd7stGU_~dZP
z1;9as@L<BBzJDFAcI>Rbd;g3B800$5Et^0kb@$9WcoR~9cR(7@ReFjXDGD@|AZaQC
zLb4?$=$!Ar6zH}swkS;n8xQa>o*zVh<l+!3jAjuK<Ky$H0=$6sx{KBvT9a8Xf{RA;
zVYiy|w&v!CB{&e&H>zN(Cm;3G{oWZl(b22aO!doi0uKriQo6a1oEBkg#zR2L&Pn_(
zLHog&N;(Xd!x+NqctuT@Vu9@`h0|tEdomw}%Q0SlXC?RQ6ePti3j>&ST-8WvzaEH2
zW1G#!Jicg8PYB>HPKn7By*h~{z$`<C+<zkqeW<-cx+lLIAzs3#UKm|5UuJH)>79_k
z0;}JcE{wiJcMaggqraclUuPk;+Lu!VT}1IFi7mIx)H)_^1?Z2z>oH68g?|CW%A3v5
zR(<2*CWIYCJP^WS!DYNC)aCVKQ2s4+{p&woh4;&yFO14`<GtrA5$E-GLl*mU#BTQ`
z`b!HIY2Lval!8B6Ckd$4d}Oya81m{>Bjiia)lnO%^5emesab~hANnf_lx#u&;C;Li
zFPXn{lm2$HgCFms3FifqM&7|fwfETF-R%b*5(!7D&-re6F~S(`)x~A%ey33W^@P8A
zRS&w)eQ)~+{&r};-MEBNU_=_^GyWCkc@yp_Xg+tOQ~upKa1*Xuh;Yf&-eCF-k|i}!
zhlrt&=M7{{_`i;%U%cXd1*fBfWU;=yj46=uyplZAQ+HcEAY@C*1?2T$FhyTG|Mq(&
zzJjr`@-5T0v9S>=C+m7lGBe6KLL__He-AwT4)#w%i@&V~B4Shm0z;RM?sk9aAX_=v
zn{87S7$lPkM+mQ=LI7cR4uA3Ve^n!R5pT`Sqix`)7g{$*e@=d-(R)D!_st)9WR9v*
zR|(D)<N`ah+P8i+i~k=EJ*nYUH+@G;+6=i1sIMw#fulw8Fo6}Cur(MD;fSE-0>sR#
z$iI7=ZWrJ%NZhA(9wRa<uQ=?qEu(qEBxlJaZ^YA^Hw79B7aOr3l$l?W_B|J*0oLO8
zvE0qg&6O|p^bN&Q0#zn%?(grD(9+hG8yJ>o^qdPDEU6qebpP`>=s!-3Mw1Oew%8%y
zbbm`2Q263#Rl$dHP&HsYw5HL|FB}2!;ZmBvw4tq>Jq)%qz5Y0rN?8MIezINXQh%jB
zBbME2Zz1Q2%0V_CN|pU~csM9xfx|$!6GOtxqV94Yjlpz<a1~Njh4S#kd6ktRUyGLU
zo@~2jh}88tiLrED?mGg`!Y?vchyI4Zt}D=bHNQcyG@!ZlUfS0Wc$^NMH4c-w{#)Kh
zyY1;V3j<?X9gpjrzNAD^#I;cx@l*T=hy3>-cl-0|=y*Sj@*tDahu~aSR5;0!kO=BT
zml7q9JccURs{-k<BoY#KnZb|VyebG`8|L;Z?>xB!(4+Qt#iRGyvlBO&W*QMI-$W1w
zWM`X>73;@lGw`8fvss-8r%5!}ERKR?BH)$dK5S|tU1!tPr5|WwPV1N>p<FnM%>$0?
zYkWOJbqr)=N<d+k@bX-9r4A^+2G!|~m`7VK{*iEIsaSjsEK$+NibdvMGdS{Y6G2Wa
zCJN2`a5-!RD(%g;>&UYYZ!X29i^q~{QL9tHo}s&4y|5F0{Er*V>kTq732DqKg^u_K
z!P!A{Y`70`VcI;rSp-i>N0V(Au;r7mri*NBks=!Fk>W7~7?n`fWjN%Hkf2&~J8|7A
zl-DyzUCQ)8`qSYpD%nn>6U{bZfQHLaM|Zr7UYXOBc_E7_q;vpY{z()({vA3Ma-jc5
z@*aozL)0H`YLrxn{<|SB)B1apRjA#s=0*0t+R(OdUmc`OxSgW_>%jfS*MhE-)pk{r
z;~jcCPd;aB7B!fgj~C0xba|UYs;)h|YWwmWf)XKRX452EvGpH9;FMPe@vmKv1dXS%
ztukS$`uk<O=+D0#MB9jaibxSa5@Ayx{P)EF?W{oMoryzOLAho_6A#u;%m2bcKsfAq
zgY3yn+?|XE_)PSVrrc*@jrk~MTizGS5!DlQ(7bVAo0>r)%Wp({YWa~lan9f4sYQ{p
z(FfmQPDr&|_|RUgg$MD=5sR=^lyxy5_(Q#&dhSBQZSJv(mw6W}=MH{PO?93nKaBdq
zkUp3b`sf~)W6;HRxWj^f`p3xaD8c|qym9teA8x2pK7N_1$xtFBzC$5vL^M5clbD;P
zJCmZ%=F-`2VXv>rduzXC!nehsd}k{vHj=F8^4PA=s3kTRZ+mi~C@wP6!rO>Nv^H9Y
zL_SlEm7n`ur#7BKAYFv~GZX7T5!=2i`cR<RKc4`>n7ugH3bG;JbZEcO2R-dO=exC}
zDZNs~7fkS6e?^9;KZjr=Fd!iO<b-XX?k1IZl`IyGnJ?RI5FeZ{<S^cZlZ>{V7BHlr
zQDIblx;sd!#sf}?=!Zf_ex8z|DoI|b*mlcZlDlJ=$fllxG8*G|Fi6dU%78Op`(~N?
z<GkwKkL?D#g%eimC8SOKw=K{T7A=5RIZ*OD8$8Whxt9jcl+XfQ@7uyBWwhoRMN$h3
z3nD@}2Xis-lUG~ka{BXWIeP<s_#Y?~WW7uKIyO1&JEjv(Exv3{p=`U+byZ>8p0?ea
zqI{1%1QZTM|9R`|-S6e~WZUPPNNL_<$A{kSUw}36eU+X!N2TWcFM69ZX8GT4j2Pt5
z$@v+%7$y)oyqRfDk*z%^ixMu(&ff40Zf<SuKp9bgG$boW!p_dFeGcT!fej*e{vmLQ
z3BfaP4gov<wD^3nX_|3gp=G@CeQfEdtr)f=gT<9jg8dbx&(|;<C`T*rNF{;u`sG-O
zOOSZ=DR#oyoFdbi<?HBfejll=%1!4?tujfgs?p6Lo-;8KI`a3TYJqq>k9e+5AE!>n
zU;eNM{{Oc6|I1YpIqa&N`jtLKzPZ~AOTqHny9NclZk;wUAApygS!C1)!gR^_J;jR7
z7!=)&sM)Pbhe!KgWcupgs^!*osF_6`Z$M}Xq!P?z$>YaT2GT$v2Xs?5oYNS-^herY
zj=B4oa;$$wBa`gt%&Q{<phyVfe4FsHBxamlg6)sd)eoywx)#I&VR;Ujff$nc0wS;s
zqRgmu@u1G{UJHbY5A5R^KNIpycjGLU1yo_<U_J(?Jo=arPOLIAD~@lGP5gMtLgPn`
zIE7=U`K26|`o>J<3X`0SPAPZC1g31Qp3J*v)%F_gFsk164^phxOY!#dmbHQ>s-HUT
zR5^$S7)W>*Cw~J+e;l~~9$Uh3=*`?uxq$Myca@s7#039HxaRzIjT||4?cvr##>JM4
z4=$2tv$Xu?HY{J&qz{VmfS(<m-0OEXF@BeE7BEkz+&$jjS#fz$0Q5pK;RB55o)ZVa
zD}arJ42YR>6|0q$Bqb){!uC>HKMc{2qqz74L>ufnkq6Q_YAlg<I4tPgZ=;X+(;It}
z&%>0k7|Ih4wuY!@EJ0LE19D=D<0(6vdX{vZ*<}5~oppV|(!SDFQ0cD<TXCB5&eI3L
z|6a}c7Q)NAi)vRi?fYX00Vm~br@QyIg5Rk!Kb~?>CKLSU%jny=?DYm37}9rqabN<M
zO6bib%eA-CDpa;ecP5v@hN)xO1N5)rWGE+b`3Y+E2bRoa8EalM84uPw_!ydP*IqJ=
z=;b~6{65{uACu?wz^20WfKCKx#k`z$ZD-vi0=^-)bK_+wv*!oCYeD_vRX|t?ov*p|
zPjYVWu(<`8IDHKY%dZ;}Aa;0#<^%OeISy}2@$6Qm4VgbDyaXwCb~Y98BMxqU1?P}m
zd37@VF??5-mU(%!F^6`w(N0N=hFX;=oXh37bAx?fe<ErYpwWL6#(KNP6D@4OP(gPj
z2Xp%_8W{;Is}w^JKC;tOLw>+H>^bE;?A`|(gR+3WpxeYh!DioDtok13@3RmDIOu_7
z^gw}Gvi=q0tvCBr>FID*@M^){l(t{H2MKN;5rmx`u5+Y6D&mf%gxGI{aM&?Sa9sNW
z#in-mq680i>xB%pp<FY{DqI2rGO-A^$H3t*WLvzNPe<41Hl>lafa%8AE#-^d0uO+^
zMsNLrf3^5&%R0k%o$Ot>yPGq-X?HjIop-`n1`L*Q`ZH}L{P4nEG0ZT5e7AlJFM0Bw
zfKC0%!meADgAZe!#q_9MISJNevR|))PR`#*D8Pss1h$2Ckh?0Un;gdag`z<5p;7Wt
z^893hOAwB>6)~e6cN7DE>$_@!&*`bYPvdpl0z31rb{wIneThk0=cU6Wk;lcl+}>Jj
zs30u1xBllZMId|fp{>3|FS)2WH^<yGY|W<CJ3orIUY9^KwTW*9-i4%{O+3|Fb8~a9
zmH9Uxax?o<l1ZTYN9^i8$57n)CN=lZr@%}cm=r6?(<YvBP_T7bL2;q(!<DyM8zet9
zSj0Xx7{XSm)Co6|767MGieOQz{RZ|zZ=sVM+M9eaXZv;2^{H#^%9J$4@FdK`qV=TO
zl>;Ikx0R}!v1O)Xq#hnibvF;e^%Z%(U8`_)b>LyD-S%Y`^Hy<tXD1^Tm8%Z{40be9
z{4j9?W2%3dgpiO#Rg7xMTY1K4vbB%hjWVB{nZYKJdky^4`>VvhETS32+~<-)`rT~F
zOM4Z4mDl3ysg%7Jdm{kJ(diR)KG6sq*~VCvl>NpiHKgkRA6ve_Vfv=iSw-Z6cWI7b
zI&DRw24!ZvOy_g89^KVF1%|D{`wZ7vBzr@1sn+MzBceKQZonHZH#ftHh)7y1PA-GS
zF^4e^K3E*{@+Bmv(?f?;rvafk@kevTon2jwV<c~Tw5>KytPsQP8dp27*@12&M~MZQ
zSY*x}&ic6%e>B-o@W5&!BKtQGo3sp^4n+qqmf<LL0ZJ&;qQq07ciV^rZgttqgj|Y_
z+aJZ9?hPfU+MnH<RDwkW(3?(?^vYQfe)aPZ35|^YT8)>dj%TL8GQOnxV4M>~>jVHh
z<Ih+!)Bw`-4zZW>W7pNBJlDGpZVUnoV>bfk7?p9EquBuHC^uJNyE>sc<$C-W*dFN>
z+$0{afX~t!r1<(qX3nPo_IDEzXLGY_O)66zJh{N^Vceg|iENPg@lf`CeLSP!C{)^b
zj$*WU>rXO1&ZP6t?D|;!duo*lkuhk<8OonY<wmy?gN}ior^kiyG<p}K!KvKKmJZcR
zpXm!<3}84p?6p}#B+w=%C07Rtc3V6+ak<0DGZnHb&wZ<UQwMIbyjjTd71QUfLk^gf
zXmi{ZybLgV{!NMSpN`r$k~illllLp*x-#C@eR5@~DYEdt1x&h@Nv6m~?_%pE1&JI>
zh84!hrf4O6Y`BYOKiR<loVWqbSTB0!HN>KZMf+_aplFEj@wwLz61^ZG6y`@+DZk!Q
z6G+F}jmX|G<;1Xvrx^jrA!sDD0)BL(C6K`(%UFdRRl&8vcFPfdYLHKN_BLGVoqc8R
z&y%|6{~m=4uf0H*YysfEEI_!yWAHbt#*?=g`9lNIc%HOK+jCDrk(Q498iQO*(e;Qf
z{lPbnQjr384C<p8y5LlbO}1=tn0sE1n1vaB^k{)sT1h<Gg3uZd$-1?c*`A8_wOz*^
z;3b?DX@Jc^nHZpJuojfk-0XRF*8m&M%}B+fP}In{H!+t=y;_}Sqi;;o_`y~&DNuR}
ziB9MW&x>?>%vEJq>UQhA(0S5XZL(hPR?`3671&=zBmScW@RRk#d+Q-x0lHp(hUZ95
z)FZ#EKiud!3)u@{rV<h2r*!+>QhS~z$@v%&s1BQr60Z+iyxq|5DJ&baS?Zj=I`uB5
z=Ogwwn{2Xo{y<J}Thbq>X-h)wdFuNB-a!I%i(v5qQv|^XM^eyS^P`z!s;Fpimy&DI
zIK2D@#6kM%d@q*aFmz0?^msbM3lH;Xg~+0YQ&2L-20@`Vu1xssc$0kB$|@;FI_NKW
z=I5_us8E#zgwCJfv+z?20pd@~jqs5r5av|*_22L>q8|4<%Djou7t*4$AF;VNJBU^2
zya|uxw%P&(7D!Dfg<1o#Xgu8ZRB#aH#0_Por3t6ARVhc**-1U06-itiD0$q?pfzjx
zSaZIStB@s+IMz~mh@%gj{J>qI1j*zyuePs-?*3#xEYOa4w<4j4&xhy1+`Qg|ev`ye
zfIfp$$GgPkQp0WcbvkhHb>}|)-2ww&Q@{)0doq^ytND8~`6!&*duYER%*Oe_D~d<6
z$SLM%l7aGNn8@*zK^nL5tSL}Gj}-=q;zGrB4b45bYjY8)yhpL6rbPpO#b!2X@kRV5
zCJIg`e#e|yTl>Q~oC~YVn${yt8YdS4w@QsS$C-V$tT#px8B1AHG-4zAYs8*I(g1L7
zeXt_qv>$(u<ngC8&i8LTZbwJA=?yR2=!3ljJ-5g2{DIaQ6wLb7eG^g=0utEM4FIz*
z_vgEFg;sn@lH#KLU34l<52qt>gmjt6p1OxfCny}IF4x7Blxo*$aK!#Vi)c)#e7?QC
zJv^LrU!cT9VUq@6)*sqFItm5ivQtb~T*9Bu9@oyJsm~s`>>nzLI3F@O?X6yaDMZg&
zsJWLm=J|Tf*jN(nxIz?g2^GXp9F1lK0ow<XTkh-QcegXWsP18W2tPQ{apFD3%AK+<
zPOm?f4%kgN55Og`G|}J_1O7tz^7VNoE`u982`$g`$8FASsc=7!!wdjWXqq#gn}C4e
z<|R!SNKC&l_S8_Rbum4W?8{VN@+c9=EeFO;frvf783TDHAdgHjj*je6t(T85{kgH!
z)|<6s4eq1Ul|HaG4NV~?YX>3M5uL+?bB<fCTvY_H!ma7gcu_<2sn~Y<yw)bG`eYok
zw}G5D?&h)AbPn@6PV=z@heI`Cq9!G1+9CFES}^1*V-M#TMxV~MSKPD#J+6xYM{rxS
zT^ty*S)6Q_Gpj3`V&jv|f|q<`-KC(b*OW%R*MSq-bz%|Wil&})6ZAf-$PVbuMcHl(
zph?fgh6}*5lg4-k)Lm^(@9kx6zx|lL1*{3TA$Kr0$4im^Kt}(~w<birgz+7tO!kO-
zmv0Nr>AI=sTvi7OgYUk|XFcMHCi(RGp8tts4PGFYQvoiX<%D8OORMlFAr!NG-+?r%
zPiNga!9AIo3JV`O%|rP87)c&I;x^*-fhUNQxX~CM85DLO^Iqvq6f3I&*&>5V5cSY{
zqj!Z^jb(`GjrMwRM^jdtG3&NYk}|f5ph+hQPScU8vyY4K_6dWldDCpIDEoZUW<<hz
zZ$U&YPsS9K)joKUR^9q7>`Rg88I4XnJVBc(ujv)VbG6&&UBp!bv%5_sgoNa}KA~_u
z5;`_x2B~pz;Kc1RvaCgqCu-z@G>l<~2__hFk{&(3;k9yqr}p|a<L9~Ys_jGf__522
zsq2z%$n?&Ori&w;Uj&{M0rg>-1VGB4fa8vEf`vKdt-<)ICMMd#SL3vUk|bsXp9$8W
z$f(E7OHbBq`dSLD?)YUzWt?l)!+{sKu7-RAYw2$<rTamY1TZe{VGE@fyeLi*VMi3;
zD1YXsf#DW`cH8e)W!Aq^Uy4^Z8qm?_7~&5J5%bwzKI8a*T;Z?Z`THw2ty>7WY#zB!
zZ(=O>lq`w=Gk*K+uZf-j<#ITBN?s*6WqHqlX4HT2H_2k2GjCTa@fYc1PWIRk+Wy6p
z@g~Y5G{0$XX%!WJr;-?_1eR%>QY1(r{yFZ4{_B$*z-i5X^PVI>x!4`ujR%H`;tXBy
z;&+<@<nH<uiC*Y21>Py>`W?83@gToGjC^zRC$!GLzgJjQ{c0xfLAm&C*{w4fC*FTs
zsk`B2jn=#Ia7o*+Q2s+s^27#ZuNNpA^I`@h{o|*i%MfyNcjMX5c8S%zUnRel#Dk})
z&T$pu{<%73lk1b-HwBBCr<iZJdU_HDex{}UlPP!$RYg}O!v3Ey<|m=cw?Lr&6c8<2
zPOp$=F8;*=0Q^5?uM-`$-v7*AcXWNN+wFx|g{(!$&`Xj}iV~~6N%6%XI(zWo<UYQC
zfc0Fn2t?x9vuC4ezza9T+Z)CoAf72HDY7W@?dez4*N$hms<t;)w?%6z?L>#}aegD;
z{ru5+PZ2F0(k`;uAAJ8dFOY$2Q`O%3G!N<dF#$qlPkHC_4&9(DRZ!ALu8i-PyESYv
zO6&$7Sd6kA=lJRAUR45mU8r$ixxNx4S8YU+!L1|q#pe9<%9mjA<8cGS<r52op9)F+
z8I31Pdy+p(-btn-hGLyw`$~kM0)c;Zy1nSNecfDhlR$jFrd@7vPAo4BK&yzQ(UlL7
zjD+5XL{wK-hipA_`3HnFqaFstkoA~BN#Eicd2TPQi?g7;jeXh*GW#BlvJOF?bFnOp
zu7#PTCYHoGsh~>@`>k<mwNlg2KHa=XX3C`q=g&yTyi`9Qx&Op`51}=);HReFYnX`_
za7><!IO@9%oV23T+?bWHM&rc=E5#2Z-{aG;N>yvs#+`@ba@wmKe7#TyuIEdox<8Rr
z|F0EDxChOoro!;QFQq5p_2YIp6PJJ@4vjC&>5zv8`WMo9u4>;^UT~;gwyK2&+{tHi
zzjy^&5G9VbCTRggCKgKiKE=Qp#UhfslT0%1Vf+=*qeq&+IAxP%Q+Kt{f|!R#>+E35
z_~Sk?Cuf_vm_!Z*Jv}|Qo1+pFf<XGro1h?I71?-I4Y7k9;Z^Ci1;!`Kr7)Q)YXQYA
zz411u*osb2u!3GnRhoca;Q#)8$)&|;xe7~cHa@jPY0y$vO!;P~%_Ce+yVQReOx_l>
z8QKf<?HGEUZ;L^c!g6nc{P=53%nhGp(Y%k7dL0W&X7&8<*M@vpxLk?l@GtkLT?=*C
zP}(%@GcV?53EmjA?9gsj-OZ1tf33;Hi|?^fwikcy{$wAM$~xAqYD4U_^O}p~+I?1E
z?f7WDr*p>Za$<AJSEK4!I2~9MDr_THbw1y&o13H>F#H-Z&9zq^f7V=;#CPeuH<MDe
zI2@aIs0dk%)O3=&nhx^Nj~vpTZ6w)S;7JI*Jij+uS!G@)a<v0V7+>7GdJw3wIbPD?
zqKpfrdj8jh{%4zizcC2G;(<e5{niC>!%G+e<lYUNOY^HmdTrer{rP-uW!)ZpK2*=b
zH{x(BHoLbw;%Jv{o9$iQCg|*6q|`qPW92s}>di(ko7&Y-DYO;jC5Q~F7(~fGnfWC3
z3|?9M@<h&InlkT?d7Aac?4H`Nb@N(CdbbKF{*<94_ljeeP22m(nI$mK`^rr@9rHWj
zN@_4Y#ni_{nb*DZ_mlj;rcT-#FfK&RTk*gDPME)^B&4rYf>Nc(zNV|yayM0}{c)X}
z@@oy(;1wp^Sqv23lUe>cYadDu2NPOWb(d7z&3>8TT*woPI^Y#|<(g|sKVUrZbl2Uf
zO)pVcT>Q?8dqZQMS$v@J#Ny*rvGlviUN}Q9Q=*NZIlnQ8?&W<`Hg#J6Zr{*e^!~Nv
zl#|C>wNw?xxWY9^iS?mf8E2ODaeP@6;rm*9(GVw6wN|;FYV9JUrw0SmF{{t8JMzV7
ztRc>rK+B0$(;@s|ZAfEnu=_)$9YeK5ce}NW{jwe}?sqBApHE%4Dkwe{k*53!D-e<G
z986ymN4~FXhsRtA!*rP1hlQMexsNF}J;jy!JctGeLSXs_1tB|0qc}RAmXvMUqXar`
zcU_CBsxntieR9F9FuBmTo|JGHBSelrn+N?_cfLHBwaxWH>EYS;2nh+<kgw=rowO`@
zdNvwrpS_Pk1&qYcDj~Knjz89H4HA8Nem4}@VUKUdgwLh<A6J+qxL(|^*ovW?IP`)&
z6ShF3vRY(DGn<S*zT#zZJr8egxy7?QqUEkggU`@A$)k$#wlE-qp5gAV*C{H379q<;
zTJfJm4X;9smBE4<*PZSF7S314%5v#$6m=(EpBS7s4*Ltv2~WGWuZglovX5D;BBxzl
z3ac*qJwAfjkL!5#%JH-eF~$-Uo;%}PJeS+9FJcoxF`<`p(N7X>POEyr#}W3)fny98
ze`g6=>nWTPo7)}yQK?#&>3I9Y(Smtksi<fe+OyxRmKW49gM_CiW6w?2@K5-lJOWs&
z6wUfNAHFTt<xo#}6)FujInuhz<ds3pD?vuJi%o)_4$nur%{byoo7Oh_dP^I$8#Gv0
zB`Hotq*qmtL+qh}4O`+$kcZDs1O0FV**tT+YZHeWLBJV0Z``uiXiI84iYS{t|B}0Q
zK*~>xJ!2lj>XB46QD<QR5W}SO_TkyE@T_6B0PY@ulkBPlMTF07B<sh$gLX~)>U3zt
z)PDP6C&%bJWsSE25!CBM>uNE_+n_h{aJB|n00u1(b)eT>THa#Ya+-N=zxHl-UI<d@
zbNgNegbDSN#`4ux<+Zo@Nb(qQpq_!~VKGwz5Hn@oQMp%YJy|ku=J^_Ir+UQrDgFQl
zXH=92z<br!VY<d8uj-=y*n_C<>QWcf(i*T5hN;zRWm~~H>rkYTt5#;utHC7yN*A0H
zdOtZQc%2X(dE(H}&>o-dkl%Ws>`>|DA(wFZRg_oj3gu#0Y<$YXK@oWL79y><gtC`4
z9zWp{am)LhP1flYt_YBl)Noy%@59m|0-`xd&}K|lHevg|`l!E7Z+1dz;_H2mSIe4?
zV|f1XqK*OZ<gc6^A)8}GJyqC8dr>k?lcfaW?2Hm$+6D|N50&26ebpF-X4dD~evGw%
zsd2i2wi<o`vyb!imx<HN1v;Gv84X>~#f58o=fLxi3o(Aft3KYIz-_{N<j*b=baCGF
zOUGRvDFAq21{ehPl5<Obd`d!2-hNXBbPYb1Z}0B5XB1;HAEc8nB3ifm98B377~wE|
zEj;D8?snQ_;ZA{#y0^7JZqE}>@#$q`Ah+9~c;0O8^}_Se6tU&I(5Yi=1qG5b3y$?e
z01%c1g;Oe3bqZ1&mEKF2&y;OZS-|?$`blI!iYiA!DeX;;PVdnx%XmlD$?gkv5@8c9
z{+=-<;=)ZiFT?h&GV09hEOI(bn5JJL*eRpPt_+uiem&KxHA|MW2GB%T%<2|&Gc1H8
z`Ai%Zs+KRU6|mGl^Wj>eRVL0+Wsam>cGth>u#bWUb5||BJKC5Fn>kyHM)%GiuQGQe
zY3ER65r>B})vEO>ik?USd5kA5o1V9Rwd%c)0LzEE*eFrZl*-0WJ_;Fg){8+<_!Zyk
z-EZI2s~_p>wMoInnuA9Ki05RY>G$`Es`}ocOIrG}WGj^gn&pi>2GNmjq}|$k=9eRo
zMQyF4O-@2c>2?^n=iwl@_LG7>X94gwMyI{vXbhX7DqFSld4Fuua3X_ZrzJ+0Q@2&*
zB_@-LZ=e(bh?WkfU5D5G5fS?<<Xp-Uj!C8LB7LY=o+equ5dM+=Fx;uhxQX#snE>zN
zgI-=_!?h0`b_Q`0Y#@?*^+^Ed&y~$LGoF=a83gT&5ClQ>LaqxY)6UEYDQXjWO#8F+
z+g2T~DBCy!tS_mL-@wbBM1uG?7`gHw(!Az%M|qa{?zDQc9m-nyRm`q^0wB=$NTMfG
zf|G~vaQ&9#W?#{BRzO+$XqG1z@bvI2vi;b(;I+@0kDt47Te>b4%}jHZg-C=&)v0RP
zDeUyMKWB-*nv*l159L<u!3XgKddQ^hVtM_{eTT=$zYGJyl@b#z*FJc|6!Eg>!9@`*
zR98;KQ9;LomkIe`WlCkd)L6er(rC$_8}@lejl(o4SEXISfCny*ySk!Vm7=EmrkP5u
zPPhvNkMj<<g9>JDJe}lH_qC{|+oi@>p?<8o^JxiiwS2NVP?f&0yx4w`R~tcgbw%rt
zK;XKK?@&>UtQ1J_*yFsdD0$%3iY<<6Zq<@_{hKl^lrN5lE$R9@B=e>v1{){i;x+V+
z(mcjP$<d~hrTkZyfKIHpid0nt+;ADJn_H%6h)P|%hUT-ElUq7X6C9>46J43~D?#;{
zpNU`4U`d9vbTz?!DeyVY)!8zSb&<V*DV^=lEQ!9=xTTYz^yn_z>4FDnO<|AF*Bawn
zv;pjZ{M7z?hE5m)b4HNgZgDLIJ+c{WjxbBt(?8z*x;2B#$XxRI>G9X=_7;A4;8mw`
ze%hJf{@CScqv+D;L(9B}ORMRCpT+4EtV8>&tZu#Pok{E~RX^nMx7<vOujOd)@Hw6&
z^Pk)AL4(`y5%J<KJLf=cIop-N#}+kbam2)7T^=eA(9jlr4X;T@rYs{lkEC_2y4Sn?
zS;y-!4dV@exhwaTfYAB=`Z%NDms7hMb~WfPdOg9+P{>Ma-`aTxEUb?wjG4SqPKiP{
zUn~_nz<5)}m4aQva`FuJ2@`5loe>K%isi|e6#l{VLT+1IBvr<28tD5upR#R&OE4Cj
zK>6s~0DIofX;BTxXNP-(7&L299!3l{UoSTv$JR7K2x(}#HQv8xKfmr+VImz!uO;U>
z?Moplk_F8hf|{4p1AAY#-4#ultgY;)HnAtge+dd@l*3@_HP}8vKRyfBTcG#2nEzzR
zMKA7k;&2~r2gW)*VYeqxhs9CLbh65!Y+g!4lA_(I_nT2=;zO9J<7rEK+T5CR&Fzus
z$)4>?t@Y|l6atp77PN{AGdscJDm4c2rsKs5fcBMI)Bd2ZSmZIN<Xoq@nwx(FrAVc_
z_)qn`@nuthE^bKJLk%9oAUy*pRgJigP-{(3PZ;D3-R;PPJ>p4wT^&P(y2~^#^L#So
zIQgOCP?J6FiEK}PJZz*9;nKt}N4b!@>ajmaPEVL1WpLI!w~9XI-(1G;wMa_YytFrj
z0(1!E^|p_^n7(0b<HuT1kPSFx4*6~m)mkC^H_AqGZe;QYUjKg_<cP?mKRL%7tPVAC
zi$Y$yy2de?N>78A&sDaSxTqx#ta1Yk+E##XvRmAGfu-v&SJ7LN50MG9yC12(qQ5+N
z$Aas%!*hi#Y4PWM^go%=4?)q4>q9z3%)NcLBAU3MJK?8sz+myKQl!nlU>ZMR^>Phu
zAsP%@TVz4FD{lbBb+M@N{zbV1a9(Z+D9ewF0o~cue`lA->xBH!pg{Zchdq1Rp!xgS
zGWZ?1{}kl?^+hR)L2U*7Y>(dYR|ELhUGHUt_NB8YsPqy2M+7J<{H8$VucFAmJ^+=s
z9-WTy)k8Y5H=Qyn7ywP9QVQAs7k~GL=q3?W@`{dt-V*>uyb6~5@Av%Gw6Q(rg%4oF
z2-gR_tR2=Q?7vlB04GYu{U?NtL?*d{De%uc1yhJuf7Vj@_Yd?MfDuQPJ=H`4sB7(i
zdQJZ;S^bfDgb?@gCWCU^Iwk7vJU}w7WI6s!Z2lVHpS|+xAVt+0lgwa9=&3x66(Eo_
zp8Q;)Vf!)O`QyI=`Unf5Sk5rxT#YerOPFzy-sEdWFZ3-w?&w9DKy2=2nxIPC3^j*b
z3~Os^P&n904Ucw`|12*Od~#B6rq4C3+a@I}YF*ge5?q{`owG{tvuFID|CjL&X2LA6
zKkKs>`Z3RmjniD_vd?X$llLxf5leghwip-ZqutdyEfip}+%E!nIzsd*-R{PO$5Y?C
zUXz0I3%Cy#3qs;UA47?H(6h*;9Q4F==t@0(e0~p?BB;sc&c9fxqJ+{{nQ#<0RaFwa
z;f0OzU&*PKQN<zYftvJO{N3hgnr4HFonnV0F3d41?&#fUJ$-$VDi-SaS;#W5*2H?6
zALPnn=yk9@Fkq;rlGiN!j!OU(ZryUV@^a;Yeh44Ny$QE0a3JMn`|)1?GFbWRAJEl7
z_Wq>Ovmp&O8kCI^5|Oqo<F%39hP9C_v|q!kWJB^*NGJb~v9}J3a&7m96$u3gkQ9(m
zR8Ufyp<8LBq*1zaKw^dl0SQUzZWxfRAw;A*hVB?<2&E;zn`b?H@3r?{@B4oIHOB!Q
zF!x;Nd7beK{*fkyB=)^|cX4wVA2QN?@9NI&F6+^Hzk3y5Il>c5qd$L6zCDW|w<;=%
zNjIF=(?pwD=teg;d(X<rJIBmHyCU|Fz0}vgbX`4;-9-$1ZjdQDw~+h)@wj?#bXv#^
zOn^u8BDp3zxBUUN=`yGq;E2Ww|Bl*Jn2X{LjuCCWUS@Km1d;tYSXBv`y9}aSyxTA^
zok_*H%Rl|>i5ft9;C{E3HfWp${Rv<^HMKvJfLcDT6j824wO)qiYL$_Mc8uWt@0;X*
zhKl$zK;`y{JbSTuhGQ=PC$-HCtTG(6u&|E)&PuUPYWfXeW$Dy@h@W#DARn!BKZ%xt
zZHHRqY7fD=3VrsX?Y>V}0Q=?XY7adDrFlN^&yEo8d5FRAmSiyDF4JPEz6mCU-;Nc6
zF+Z(-|9|Yo{vPkZ_F##3DQ#xe7MLcC{U)j$E+rX^d!n53TEt<x(Eu16Q`)(ML9adH
zIu#OtP3oq+?7(3KQ5}F6taO|InYn{3Wh+&ob9!)_LL#W}{~h&%4mZ@Ir8ArSdj$f1
z4+a}gbON4M3k*?nE@LZMrinJ6V=Bi^ZelscihiroVhHwUR_vM)gh9sAHo1i>lfDJ;
z1>GInGYFN90V7s`?FX8Ej*x{=37G>hu)$!0nc5g3d&Hy;+y0-#*7K74h{D$UbdGLG
z++o6ZY5pHsg;|n;qgCA5#igH71t>Z5GBd^T!z_pL#r+<tlLO5+uw^6xm?w1iNGMBc
z$gq2FM06_o+4t{WsRsmFZx1*x*9y{p&C=mW6~}iN4ysK7!r>-In9T=&zjdF9p<`#2
zm4Yy7i!a^#45uhoO?|iF;JL@Cwoi;F#75x=ZS~m8p!I##(}Jf8tFl?#1{!Qn#2HPy
zI&BLL5y`-iyY#D;?Z1>j{?AtoFy5wq_XQw5P75}C*}D+NsNH$X!})9xyJ+oe$m?2w
zr=sxoEr}n+=CsYSekwb#i<1=j@d`1`)U4!>tNoE$0B}pm(_w9vAmdxgHglmdez3(#
zK@Sy{0;b#RS6=q<KnfAcUZ8Y~4F;h-WH+1YO{e*9m{(Net=`;kay7?pg_E`fyt1LD
z5}3d)MtuYdw>!1!|EP8Q9}|BD-?r(yrWTvsJ?c94`C)6h*rdeK##|gf6j+le<z-SF
zxc%0Wed8YiTuzfa+Rlo<<bjOv^!f9+Vih+@hDX4G`mXk}gHPEp&^+s|79eC+rj+@k
z@PU*!7IX7#z)m>?fWxx%8^5o7CtZ1;2K@T1B41dsX7N`Z*`N_OAwUOiPB?;$z%Y_d
zgdLBFE`e><4co!knzWxN7WW7fJZG2;ns{5%J7Usx$ISp?AK)zvtv%fHdf&$P61cYg
zEXg<}8lDT3azgYCueJ`Ibl?0_sm$zhTj`C%*l5?!nejs>Cns}6KT+RpU{{4j08n@U
z&3#)hTtFu2*|VU`u)F#yAjJ`Je5l5o7tsK6b4EOzlrK83?b7ihHlYmV-BX}IS?*A4
zVWEDcrtgHs$y5Jf6iRw0*=8~vVm<>(!+NUmjfzcKg=r;Y`N_EG@;I_A_QCLWO`X`?
z!UHW!bPiQjOSGU1rDq8=frhV;g-Fz~KxKX!u2GWk5-_$QM46AQnB8FB-jqM^)Q$B`
zE4(>xgMQ*I_wu`YMqDj&qlEA-9{mIDq5>;9W+6Wh**cMW#T@~&<b3>iaE>!V=7t;h
z>j87WzS=|mFS@)sF#}{2aoq%#6*ATnFP}5MPIx2)>&319>L|daBb}a17jXlM+2^{;
zC-MnUvy`b)4SbnAu@YJ^SPhzh3!~}1mm;4O-1Yc(YJ8CN#B^&vem@}Nf*oaLz-8+D
z`4Q|(6I+f}<2L6s@1a-=wL$56+`Z#>mRPL2^+&pOv^Pz-I1j#Q(xFrG%xC}qI*I@J
ziWrJ@=Jyc#^y!Zj6~)*D(K9$PqvX=u-L<3_tN?k44O=^Ih}RM89ex07<<mjzw$7v_
z0(`75H|ael`N3`<5IL;3Is{V(RF_M!J&x_8OmV{*LS7<^!8JBLr+sp`kJPJR0_q6@
zG?<Apwq%jjUte<UYs1J{v=m}O$+o9j%Emt5w2utM*gQ5tgR})hWdCs4`JpI=iMj!0
z;uMuars#nujU41TQknYLhb?i+yQO>g1|zPW1%G+Kh-(H12C97YkT$B%p8XM~Ts~xE
z1Fpqf8HM2tcXTAlHpPTe!B8rrl{q+4)eN}{tu&pMpKiT2E_`l(=?g<h2QUM;3Tc4Z
zDY7;7m{Z`xU~@SjqdnLH!Hn021O0T(ZHmV0Qo(~<qdmIk2E>cFk)!}1hFApatnaj_
z)HcO~$y2uT^*@!lVI;#uy{W?ScJ}$t<@Fz_2a}m~@?pQM@-;8-7EOz<2jmCML?;S`
z`FW8<9BAIx@-G>}Wi$-)!i1P%jdDFT2j(S?!D|^F--|fV9`nHiGIs1$c0h!2xjKr`
z_!S(y3G&pRLR5{Hr9E(Nau);n-_-;n-F|_XG{An4Tg?hg=P*)RW1Vaz>9+H2LLsp{
zrfk;jb!&@A^T4~CWScFS$?1aAPOj$k>LK?3zTs6RZ(q*V!W@H3bDss~2x7&*X5i=V
zN>_R>OOs#7yr)28laHh3#Wq;Oy!XnnxtFC)?u4bgCXLH3LD70b7-d=DEj(6c1Qa7p
z+pHh!gdksFk)feqew*AW!2F`Km=4o){-HSX%2M3H0aBE3LNCA!NK+FuI;4jZM-q`x
z%>nL_IsmodK6+7a0p<_}7%s`7k!7USL@$B6&d=pV=9d~7oTRkAZ%(%VbbMBDfb;9D
zXnmHuLNf*|&0ePYDywBZdzJO;xlf)@JHDf&b-9lX(&oVm_bmAm>%$K`z`XW}JUk3n
z^^|hJ_?7i3v0j-U>6ITz^3e@bibGxW_q_J3pj}Ob`+1>4$Qgb9+~f_HYY}ch|7CTD
zTmAERsg8)Hk<N22fyb{nn`9CrDZ6I?oLz5q(sLQTgwLWO6cTvKHxA{-*p-SD45IUG
z3jeUtEj-0K<EOjSBLb?NL9yFuSh@sb;~7mhdjOYoWiBF+QJe3kBk5o%zD)+Ce@amL
z-2@NGkPU2U%jm#^yL_ggAH!P`whTYFd<h(kgLW{k6|xnV3hM4JSrmXd+I)pOB%i}$
zx3xS7F$6*yzAvEty_<RcLd5Y`^thtRiB?lzS5!uH9D`cB-Ar|bT|N*yNPS3yM$z|F
zZFXuPy%3XQon0lkcpA18J|qtKrwdzv<@TkXilleykDvP<$SOd^4*TU--`K2o1;v@E
zH2)&CZkm%ls^PkUmq(VPfg24Vk>M8{4cn4ZyDl3wOHf?Rx*7zq%<&~7zH{UR`J@;_
z<MNfSrQX#k?>uoI=<6c|Sr%I4uXGp2vjYUpSjHr|8^RRVM`L;bfX(iFZG62Y?EQ3X
zH`b8xBQ61`r7-#Jv&=>R3#=kF5hd5_E2t;7ZNbehRRD=4C>Uh5I(Q@?@DG06zT#`x
zhYq4|-T>{Ga$Gv`Czzi<ihjj|TN2*s;ZoGMBYm%enIFA`w-2z4tR$b~+^ZNC;q{fI
zrw+^_VpHY}J}8CE$Rfk7b9SDnx$b7aV&ov4)`Z`Ke_XFQu~qH3b29hDrA424qByUY
zjbHy+)b-Bd4Gy(oGcrZ7HZ(hc^@7L|HT*!2oVD#kQ&O<BLgV7$<7q*`{X-Xy=_UO6
zs~z;wXcm;lz_kysbY3r!>03kp&Y1^O+}z;%B+Q)GeSJLxY~>wJkNzo4oJodT{{R)o
z>-+Sk986Axe>u{NPv5AOxz)ZunFH9c?#a<iSHGJT8y!A$L+!+6KQY|5C~%w+u<Jv6
zbiwEMTN4-f_z7Ir{JiXA0l4CBR!nS+?YzsimlEu{fN4D+QyvHS7}w8+>z%p{coqTp
zKY%qKhO6w2SfdH$p5}4&nA|+beShgWnrUcIN2An!T)t@tycxG%m8(VQ{sLY`cN4v0
z>@j6E-*uMAf!K_aUTq(L$s|xK^nL^Sko6atR6flA)wccDt9W(s>tj-feEk1F6^nOL
zFfeJ7>5HYLeOEll0?Jn<u1>`RHMPDk{fZtI`(69p8P|`GFvu?<bj@9gj4=-!!8aJa
ztG&6GHvnIis{<U5KQ-<*6r$TZ7c77%lr3>5aE;6<5m2yw(Q(<BK9dDudVc7^o$fI4
zpjM_S4IhBgKmPI)%6%W;$+f$`T<JArs8iusG#kQhJ6A2En`BIRvy*gWznKLOb-ovT
z{=ZYPuejjdUv(MI;v2+&y}7smz)Yd{wqUB%)Rq{ZE&Kh7Q`3Zb8H~Jos+F{SlaEzo
zgAJ1I&xZxcKilCrqxJ8h4FJSE4}iwXxaCnX-{6G<&Q4;ixJ45eh4+0%5jJvE!kJOk
zmJ2^XF~8in7O`b4qh%Hi=k02g)w#A7LKmLz?(?%CyKNkl5@HJ~=YW$}NX3pzgks}B
z97hfWW_492aVEeR`3;2@<}+?~HkraW`};s}s5nFh{b%j*pIu{I!7ZQj7b<rD{@#Ba
zxM#aM1-K(GByA}ONKIdL3d<(-Q!L}kTN-zgs)+2-)s+_D3wrFw0dR7~4Yqfz01v3r
zkGyW4rycjr$ASe%JhBguMG(RIUPtTBzoW0XATzce7kk$U<KxAEa1KBet0>Pc(5~~c
z4%|Cb5leh8g$LZh&Yu>fkLd-TI<yMOXqD=+0iHq8zZ0We&mG;U#oXdd=|rAipCYf<
zdcP$mf2KI!&L5<m6;%3obA$G3W&&UOQC;o-bpQ9N__kigW$?;LE_str0sLnaqDhtB
zck<lgkBs*)2Opmnmvvsq8UYh2v^!0N20ad6*D-nE7k1;57iYi?svLeH$x|;GE-K_l
z(YOmr9`?m&<1^{p+NP}i?rk++*<gn$P3%1Qw%WCR8I`Y)LD2LsiY7p*5i98!A1r*v
zzv!3;Vg|q-L9Az<LHG9+aVkHoTMGht(Y*OCB=ZJATn@X|F4E)%rr^%ufN=UtOZ9<t
zCrkuszw2}L^?3e;I=i2dbk8-5bFuAIZmurXNvw+)gURoQ|A1*v6}o+O?6SKp$c42T
z&=wq_w~hDpNWZ+*JoOMW&$v`g@3lfD=(hb)9HE|dG6}n^!b_X*wHT1558Xd$z2YO8
zAI8JlR0KLt7g_qVHszoAOx==tjeP}oI3#z<{K#`;@4q-$!WU`5l@>k@$-qtPWt=)i
zBTE=3N(yjh*VGF}oB+hLDg6SdOqVYfU}b)_CF5<L5bfg~sBLm+9l_SBjypZgHd>%I
zbvh!?S$8IV=XC0r4^Wwx$0%9;D{1`WE9oX*^ZeKH@-YB&&SU-P<SF+-p+0y!n*^k{
zmlC87=%E251Nfq4#?h1#4S#i?*U3&D{nyEOzB1jZ#yfaKp`y%pHE{j1_r$e>xl1!Q
zGbJd_hHWV@7nSPRLV5f5OMlU|K2`U-F+yyz2~AWl(fT5B$CI+jsEg1p+-}t_I_~Rh
z#~{i;vwuk9|24?FS&35p@;SV0T2I<EBuVRC?I_CV{nBr;5zv=&N0fOU5Gtc``fYqN
zC(3p5q}8BG8OS;FjoB3JRA5!y<ECN0KVz#ZeOlbSJ)abCdmImzt=KZW&l(i@sD7n<
z^$p)!@JcC5;dakzzM3>`>Zinh*?7~>z;M!RDZq+C<lS3V60Qb|tcW4EAdY{0j(;1r
zIvU=y%m-9%91Dc6)~firSEjz1?Eaze>31!_0~T;a0Q0tcUW=`!xarZLKdyZ9NjuN1
z2DiE8h;JB>w@!-2+QR8C*4#xQH-AgR@dmmvj5F?w&s3O=1xoq9LqCE4O`Hno(tfz_
zr7?3`_`ERb`s`Sx&1!H}=C!89Mj+O<qZQZytTJ@P`{LG~>LsQmZsIzkTpueVssgG2
z>&b16-=D>UZm^!;4Z*no;adHFe#@x<1N##|ylW%}z%F%E&#NT^=&AWj_w2`l>Q4ol
zU`tDnlddO700QXE9Ky<q%fNDEn-Tz4#?0C$8zTdt)6I~OW+9K_d7gg7$ACsfmGOY>
zJP>05S`PwnB_33QXox?V-;ALCpgH;euczeZyfAzcH{W>W@q5BOH<Mw}!Un(}<;vA5
z?_Y&v+$a#t*_p@@#ZFDVb2p%z`TCgx020+gZZbU-uE}k16S_P5{y~6F1V~S<Kar?l
zkU2l7d)@p|2M3*otq?T>am3C1dR6ni;$M$M+OMsawh3U1Y8Mx1Rz2uc<IE)?dMqR)
z2k8660i4?b4#HROcQO(aIphW|0s$GC7~LBAcjsgisMF9`zGvyz9PBbX>dAJ%FgfOs
zdg2-T;0a}Jhg@Xxrvrd-7pi!M0MHkJ@o@#z;At{UiX1WrC@uhmiTGm8x01X$bbf`D
z0-lHUKr=2`{1xd0Ik-po4?rMdHmvFC>HVE#$n6fR+5Eh+aIv&BNo}60ADOoicV^ts
z<I1?ldu%|xwq&pHTHwIFbp`<b0K&YxwZ;T2Ejl||lt?Wa>kKmNm&G>=nxo!V_J4g6
zy{Zg=G{**qzWI|3f~UMSVV|kB_(pydD4X^CWSi}L^m?Yo>H)C`!g>JrfZ)3<4G*XN
zVdpc^RW_gqBH?Z|R#g0~FJdqK?Hu!G5bzSrMF?5X9`BCm0Alv4+psKZZ-i}p>0<o^
z?!)^uqbgzeukTLC2^@Le7fDEfA)4Og;6irxLrDuv&q^DP!F)EOs-8zEQ_Rpt-;Jzf
z_#dNLcHE=A7e{DQ>>c~AK@l#W5w0e??l!}X#QbE5S+!7rFvj|>S)Y&&{3LXPh+y?+
z+y#_*W+>|S{{jq)cajei1-fEJC9a#TZ;*#jW%Kb*SuyePYyht!1~8vLw{_8q8K`~?
z@RMZW)cn1$C1nhPer5FPC`+c`<mR3<IEGDIT~QROe+I~-ykh}8C-%;XDSY-QWuX<V
zN?(BW1;ibeAy4q<e-#$uIs~?}Eqx1d$kl4g$@+f6x*=18Q*+Jtfe=iB(jCBfPwd*m
z!y{XVR9>B@Bp@L-_eX_W#$~4@f4!vpFipvup3uhucf&9m+#;4EtQ6{^e*nDg!>0?V
zy_I7ki>5PHA=<}q9e(P`^3~nrc&b+12urK6;v}f(EC(=Dsxhz}mQQ=~WT0+mYQ46v
zFdkMLV&;QAm?WOxW1RJ@On1VC50~(-|Mb7rLD;0jx58%~8YbsmuLH!~6Co1U!37dT
z*Iw&(8E4sA_K4rs2k+~@-+y?ub3>OJGxy#w+JVyph+qK5SlqO0@X)Ep7>l);fmQq6
zP4e5$Gx@iwUR+umm^5@dF2U@f@$#db(sUxa!D<6!=@+f-#~ZL3jpH3L%h`!??OFEC
zjg6kvhVb>3cBnkBQtva$qxhjs_81W_qStQ9k^RXB@%QOxTBp9n<+jJHmnad-zrQ74
zH@LISq14!lo473k6>V%W9&v`gtj*O@`f>$r46&^Ph_g*f1@g?+Gw~wl_-v3c;!7EL
zX&7!aiig>A=deNc$eK@2*7G5<x%8)W>odwJYyZYnq-A2H9U}#?#{QSn-K^V+V9l5W
zz<}b^{;<@Fc!Kf(pS$T}@g0hiq|f5XW*hnm<rL+d89dWgA&y(m?&K!_VI+7ae7k%e
zwH&Zr{c1rZe0JBa&v^@{d9XQHVej2|y5Pn&>mHP6jL*vWN=0M*njj1v^4}yxh-tB?
z4>O}273`Il7bU}NuG3~6nBk%GSAfn-$V8;<T@A9Ehuj<gA@$!W?X`Hp`c#F79%@1H
z`yHi_av6uu<?|KO?1^g%^MFc&z8RZ<ssBrW9beTU^4{LldgKw|AV@~<eG5;ac9&@^
z^F?{Ia``uCmwnTl02bWio%dWik78BKVOHGY{TJSTD5cc#V$Ia~4o|Jt_Dha9!2^mu
z`24x;t+r9}lDv{6BqxcC1KNJ{g_h!*H*XR)lxP~}(PSJ8D~et#Wdya}`)ftU6Sd0R
zp{Q3i0%Orr<gFY49NW=Bp1;s&*1^V9i??ts{7<-9`MovAO4pBGTIa#$#c?f^ZZF1e
z5$(OwVMqAlz7jyb+H@6{+VTpYOUF}grL>N!3A<a-JvjKg2S#d*J$Ry0xU;TSZUQMy
zTPocbus0L<xr;b2OKWLGF$G>IIL%(pRvChC8+PQmu=E%<xly)ELNa(rRsY*0*)NPO
zQ#}9}7WX9{P)w6reaJ^!PE;lhy#auPy~@9{;?LZFI`)<hZ3K;d=DVu_zWE&HpREV}
zRdX}ziU$yB!J1l7vITW$rFzF%PoeScd(`h7*#Sj1g*k>>o2S%snM4jF+a$^BOA2GK
z%`Gbq>wN7x08FaP&2xWuQC;-h`Ab^%+A5XN)X@wTFJeBhX0BYD|Bo@6HKp^zd@$SS
zq$2XcF2_q5TI2Md#^pKYo#`@1I%&J>pnE6*C$Oqy7mc!U_ePnJt<i)M?%nIk5$v_K
zuiZyE&mF(NDCj?AvzfGW)_TUsXkeX33H$ETcl=$Y3nPnoPzzhts!O|;^iP`1wTcB0
z7BX;*5}K-NbNUaX6t4y)DEUvf;#fA->nF>!=y&7Vbbc2XSb`pC#po{w3z-&QV^Ya}
zEs#*su=M0F=p8~k?=3i^d5skcgo{~a?sW3mzPxDJU2JL^FG+cDIG#V0jKOV=XtjM8
zMCdl!Mh{XQe=EZF3tCyfBjHyl%^;8Cw#%MMs2GB*fXy@0KI{0qHmaW^4xm5&>}<;H
zs=RgNLe$7uR2tH0#yGb)rebE+qaB_wvkOqq{v}0Og1Li7P+&n=@+yM3P#P!&G_u2c
zivFtoh$8|0s_gR&i|gaT0nY+Mu{^zG5(oR_b4n2I@KU4}Te5uLM#Lq5qP0O4zqg6M
zjO<q{egKJNOaIO3+wn#7CdsxikFj{b^T>4Nr~-+*Qg}zlRC2Vjp*h-7%5G?^frU(U
zZ`Zxig;c4Jt)>ELok_1is96IZ(%?vi0?B$yV-5?KqI6!FM*gwamyJP5PbfGHo(ZYo
z#iC9*s2!gVn{^h~0JKr3B0Z*#O=@5X9eXcFrixpKGu4k=St@3;bjIR3GnB&oB^6WK
zhp~<CrPMQy9r;fx^S<tQ+R2j`XTioqFTb01Ge3&jEl(0cPr_eJ^t(Py+K@dS6!9{{
z)Ec;@Qd@G3>GLaXsz_uC+#Wd0mbl#>2D?6^1U1&%t`TRT4Mo7Sx7ZxkW7ZeLaJ7N<
zg4wDIE^?fof6ziP2wR_)y#q?!rUb`}7p~Y@v_J+<yb2RWj9R^hVD8cFSn^@BySx2Z
zSmfeuNnA=kKdk{|?SJ(Mz$i4Nts?7=OaW%9)+Di?H|hQP$&QuEHlk;2?uE20VJq|6
zV&o<HcR=fd$mgo*8eYn%nuUY42h%`}z5n&^5?3`<92A?igmz-|=Nr3)wHbxJ^mJ!&
zrv8*kOQzN^WLh(vVh1?6{c311`oaE~e=7pvL6F#(#MJWjx`yV_tNj>1w^kMNL0)xd
zQ+W|mMb2*1>|YchuCMRkPFY0*Z<ytK6aAp0)}8dD=K%|I$SrAC>KKA1_xqOW8bQiQ
z+3NQnaZ-^V?}}L#FpRKL<-Gs|`$-1`mW{MhK9dX<^p+(1ev!EJ9`ZVv1Q;m3D5f%m
z1iSKce333d+v?mMQw48-vam8|3PHQRq+s$2R;TRQjClUB;~sqYo4&J}SVqe9*_r)l
zeM>c;pRlrc-|tnXwXYt5fJ;z}NnUvv6ETfm<z3%*vF%S{WEL64PCwQ{%XP@AMsXdU
zGCh;sCqC4azwN+xvt`3Rd^>mSOIs)oMAxCkcF)+f?x^C@lmGsKz<sagcmJ)D`soUk
zV{%@*;oz?{ut-$b4LB@t@D>;JXvKyehkFq0yx}V~QfYJ(jeS;E&&}kRE%3NjTk9qz
zV&d1E|D6)Ed>0<N1Ij77p|T(>bxDWC2DwA?8d&l-9(A)g$JY`md%U-)4=C7O#&&E}
z#T8;j+Vyy517Rcv3J3WhnSy$N6_l%FQocIrd!MOZ5~=<+#`b4Gv+%(K48t=3T{qXg
z$Cv@*TLeCsrC{Y9Jv>Dav8i^(OGM9J961Yerj%mPvwkR9aAlU4#-9(l4btTa=;q|c
zk+0_u=XI1GQY*JnI|pPtB*E^Mt1jCoq8H=J&=n1Ko#Qftaif`^k<`?jn*L+Evo$m_
zAbH(mS^?DV=x{N@o%TPp0Gef`aEdHB!*O+OUr9W>>AY>LqUwdt*-#jnOV_YZ))U3a
zMw}(nM#6vxU&){=k@XIC8tQ0bFKaFLX^~NRtw+TA)YT!cuEuk$BV;L#?#k-wc8w#E
ztEWVrO<OcBmen=5B|BeagvJ!89FIxy2EQGhe*ioevaRTOe-##9rmbR}Al_L&^LEDj
zdp=}#jUGVcbAK)$YEKS<j{=nGbG4GEWeOgRj^?UCGx~XxC4S=W<Z6YYqHB|^?8f!N
zMKR;DXLMS0GIxduAAxQ2m5b^2&*cNE_6O)KoCj?BUQBO5cD=<=<IHD>y=MD;m5lJ(
z0X`Q5xIE&i%JA!7$k%>Q1ZMmtwQ=?s{8XiL<<a7`j!_iefCnm0_6(X$v{`>;r%<_j
z;XU5M)qfuD^TNKV7Aa;+JJ?^H6nH42?O+(<N#vq<^l|W{;`$Y0kM>NIjMXb{H)oFq
zPsuUE{9~~}iFx-`by>@CT1M2uo&|_?{G>;<oCriXii|o#mKm_Dy^=ig>DT8N=p{4p
zPqqHKCk3UClootHcv(SJ=V`t4zVXj$)-{=Eo=7AvtM#Ji?wQ;hXj-t|_%g5&w5YU9
zI*FkM#>GRi#+L$O>di>qR;DAU<j{*8bI-Pht)t-#+A~CMd5vSG8sMh-Tu)&hT(hlu
z$eU#{xdJtq!XKjC^jh?oM-M%|q(i<PA9}EfkeL;{l-u_bKmR*<%&`UVW+}S9Q3f70
zt=<8Yf&U2UNS&YmrmZtzin>~Vao+U(Mi{s+#`J2z6}f|?{NoimdwTdiu!Q}f9KZ4E
zryf0aSpkV(7~ps)BZF}MO!Kd7>US3(MgM~mb{VmwcFnv{u8qt@2&pTgA_AAC&T;J2
zXL<phkMp^Qg$Zac7a50bl#&vnXO#?*^lI4D>eqE(o${9~{`5nAYQ91ypy%i}*Bnmq
zr&b-w(Zfr!tAtomgwPoXP0?^rR$Fbk()>y_wWr1mAVxayQu-+m@`@Uq=ywN13%Jwh
zIvaE{Ax!}Cmp_k`1iygKX@>l$6#0hR(sIOQ%u1s9AlF6<BqE!rY_|gnYcSWPKom`v
z9pG&Il&|rH^BT*R7cJPf?(e_9J*%LbC2`-gN;4cclgxv(&t<jdX_u4zI7)P)_^d_t
z;(Tjz3(HEi{8#Komi-6owq9pa)E683w;tgOPbZ6v8{Moo%WC^P88xdAmR6IwSx_Z5
zot9oiXfD6<gv<W;G+(#GtW`zmNq*&|pbYmX)Wd2^<5=@3r?7?k6bGy@9!#ju2u(EL
zL3z(p-2pI2b(5!?m+RW!vLM*z7*wYBG_=#@b<qkC#e8E@Ni(b4IvI#29-ArEM+{<L
z+E(uXUSzf{V$}7QG9Mn<TdY?v&H0WatJDu`GIcCdW8BYVO)B<S3Y4tUY4}T-z-gb<
z(P=FWTfhjlicTe9xOv)g`#k&}3H5M6)6iov&T^~kJ@T}K*Yvusjp>n;)s;E|-flcB
zgEnt4HAYMXfhWA#)A3p<ODt(;J@<H=&PFh!i{1`_`1eOorT_@StTC1a{ifjf5dJ)T
zucT?<ke<3?9a};0TyV%UcB-d;^}6cKELokwklF9ic=MO}w6{0oEn&E>7-zqslZb`a
z+8kY(1%HzA#yI~_f^N-xvnel|q*)F)&qyMXcL?Utt$fS#adfL|+GsK7aMP&bgdfV`
zY<PZVm@7flxI3qh6m{*tb9PkW_uO-JAqlI0Ytf2(N>Oc{-gRlwC)K0epyFc#f^gvy
zaz+akD>Cutl~+H6>RubIf^1#S)L2C70vX1V1nt5+7aX{pp+`0MD!C*cFJv$BYnJ4`
zz80PLzv`c;^<+D2w9L%EhKyVuiHH|a;Ztm%F3553h~5Y<PQpda&q{EzZ$Dkg7K;Vs
zpe#>}tXiV9l6)2SoDLsG^uIbYoa$^!8LH%`QRo1!MOXCGuD3pLbhX8mOaZma6-QCK
z=qg8(bOQX{jxu?^*c}dj^_N=@n`O=c@9Ib)<pJr0)YOlYi+2`GD*o19xBX$h5#`>s
z!t7lFe&sSg<%plyvM7zm3tYeR?>#@?O=^XHtP9dzuo7jKlmM{83~ejk&FIvBQ!`*#
zp0Ph0JVx_OW~2H2FWtgX=v0qz=o!ZL=Q0^WkaK5@BTe)Co)xVHFk6@h@A-2p$(9jQ
zQK7%4bTWp3s#{f2urs-9RsPmA*pYbzuDvTi|975>tX#8>d^^!5lwg@Y;xP)uo)Mf*
zeyLHg&^Dg((DRO*!|umBtcUrX;x1zf;S0^vu(QNeKo>p9H>B-Wc-h9!6qvOEy{GzI
z!?$fg)2VPzQR>nxnE;iUT6@E_mIWG78NS2W!i>*u_E#v55YI=K8nzYLhMrX~HPqyo
zj5}K+U}6IN<W25YdsWjZo6=%_kG661cQg5;>!`fgz%@O9eD##sHAhN+jdR#Sw|m$^
z^+bxUa<N;%g2yC=@8IAMm-SSpZ^|H!vw2*`H6VtMtsJ&Lczw#<-+9Y<k%7PWF|+?1
z!;_GevXLK!ubfN>t8}WKUq4NWNQu}OwvFTsN!i|*iU@rb3V@LPS3)WA!|E6k85j!n
zzqfH7__Xi}^Qs^<RNRLS`wtJj>5NRYwMC9JCYp*%LOftcUbY2Q>BfWAPg5DaV6pLm
zGGDBxr=m<}?Y95)KF`V?il^7l^1v5W+p?|_F|9&X<q*Dxn#c1pj)}t+Ph_SY=bX+k
z=sAo_zAuiuN|7O*VEdK4n=a;=KGB!;-Ll0np9o<G?xp(kx)O}x#7MdEXYP%RN8Wh#
ze>^8%F%!R7PCk0c%(-VMbl6+Oq@s-ah1@*x3n^~ZBh!E^Bo+KLk5=2D)E*b4XDdoz
znX;<_S%$OV%X@ithD(Si6Wmo_N^7yvUkcMdn>Ci0AJbPGjwgVt8h6LVT4OPi^Hl!w
zlk4#1uZ#4WK*#`593m~$3zdRK)t{H?X_wDznnj7)DJKcOxU1#bDkhC##*EFWlY}_!
z&uVYr%!jOo?RrHCfBT-wXW@BjK&#DRpYAV82fLWgMFtEe!t=(I(+o?h#NZ-6>HZ?*
z#+&pZdT<YXieX|ie8Le60o^zWg)Um{>K4&+@UiJ_J6)CAhD0;(2Ul8X^p}F+_n>>y
z;k6IEIi`^>-uPP?h-dihjX|;(JE~h?d(m0fI;z10?g;A?{NT50z6GKgjnBz%75fql
z9Q2V+O79di{~9vmu<EvGbI7pz+dx`3wk`52o@*T9!~PHS;s|*zVlAu7KcldZt#1DP
zppaJ!=uGtSjt@&imQ1$YZ#qA3N&tDZIK<iYmm7mP{%r_hlUTlc#fxHfk%{!|23dnr
z!aI;cmRadarigfR&-11Um2)Dxaj~r{)V;aZg$4n)2JrTfkc;xKr`5sgs`5@|ULRd{
zf{pKfI>UZfI3_JK{e<Vx7u-DYMZoRw+c%ZspSxevidf0<S=DB|Akm%VHG5(*V9y`F
z>a8X4h9`si{Fz2GDkS(U()d_v4I8<SexR2-@CMev>7OBX)1~W2FY$F9RZdG>wnS6X
zq}%z>-EZ-%8fB2p+@oMm`7;y@p{Vx$%TDzig{yvXha44q+~o(<QgR!f4-LCxRvOg0
zL8W|GVzbN(>>h~FGVtwejQD+<0`2-XGcfGpqM&9XbQ3%r2a=ZHY=pVZM4CggUx*+o
z35AUVI0viK20dSjz?lVJpXV;crPsQ=Q~7*Sdp<M%O#7H|epM@^MV{xp2atmLKG?=?
zFiKCtFu(J%FuQ2C1ARJPlhsz%0<~TGda3`?@lCtw%iii7FuST%{=mcB$niuYJyKIX
zRoeL|SA$~L>I`yF1CNP3Br`GC#hHqOjBGML^Nf22;dMd!ioIU)Hrq@-7CssCMkUGC
zL8h^=n0i2MiTDeoM2Yg^tDMMJy}p^3K1UId_F*l2EY(f;q7!1H45TZ<tP3~M#Pzjw
z^1GkBmySBYjMrS`X)oRnU@<M*zJ0j^RqV>%*{POtF9gIydl|Un`oglkB7+5H-Y42D
zl8JM<%mik_BPT#cIaIwLm+*|36$1O7=kCeZ<PEjv^V=;@q?*)ZMLL7<mIsZ7>rcdR
zv@J8{1t%|+Hw{l{%dVqAnW|36(w0P5;qs{^2{t6G^#g_p)9v0)PP;u!OFjDLw!#ck
z0}8B8LnL|;CpZS<e@2jd;Ku(|gQ*2es2UyyE2lEuOH12sK~zEHh=~hsk<PnMd=tCu
zuA$aY{w+4|k~nkeWQVCgnjfo@LoyJA`CGag*6434Y`}iyV-f2+2I)iSwU!iVgb-GJ
zlg3Gt6NtAG{P?4vSlDS2Dg*A*aB)_QY);~wp-#Zshlr6ySlx^BSXhYD2``mOMJtC)
z)koA^W9h@nF<7ys0F7CJ&Sh^(`}z7%X?0SPGRVir!?Tm%eI{H~iv3h_v9HmAIa5aF
zJahYnwTjGnneRh#Xw;dGCiOnmUFtQTziqUtCH*Wid|oXV4hfy}aG$Rg@zq0PJqFpm
z(9@cHrwwOW4>TW{UejeRlktA;HatIx)U!u)uo~`vGO$QaGQKV=dT*^c*BX#LCU!mo
zY2VklK3_?^W-H8gyQrOO6?|JZ<c#fnV_c#y#JElL7Y0I!IWv+R)U@3Dxo0Bn!!A_4
z1^q(1HK4FM_i6v9%*%(oAHGtVao@^COB<P7f82wqX&XSi8{R(~&sqV02P|G5O@`AI
z{AATu3-JYW#XZh^EThugNciQ!@x$@0!$K&RHZ7F{+#rK%R?2l2{iC=<xYk?dqwT87
zLvlA?X{1`n#r<;)>)|gBLXqQ6d^2EV0g7I_YG<sU`ek;>&fy3t@Jr6LWg~p^f?<Oy
zM}}KSIXdniZpkg=8u2CY_c#jOGPDMs`1(@1Pg!PaKkKWR--C2Fayg1GGWfPb`3?{3
zM-Q0J<ubtjt6<`ZKY4?^J2~lO$dbkB{2j!)2tuUW9!cJMM~^nV5qG{3pNk5zty<;S
zlOquwGy`4EL++67g(mUo>!CF?!GA1^VRg+#hAw7eqk;=Nd%O3=q@d@A^6z9is2WK4
z^#k=LNem%MEK6-e!&w-Gv<`M=#k5-Xw(k?_cyLVNj{RO9bldR>%xN!trv`(?dUl9T
zIy77>7n*TfIXig_A;x$!Gx~)c%SotSMy68woIZn$pSM$<%~>z*>FK8pNJt5x4)#R*
ztUG*q#dI?L2qecnolX)=`^<8U?`SL6-X?(&$75OXGzWv5YK-t*4_Y0@|M(1Lo3*<K
zwUkc?a_HfNF$*Z<l|%Oiqh=$8+-DY17DlsTy>6)`AaWeFr$W1^qX*d7?^MjX9A!?P
znFTu29$70amY+qctBJoekV%T&=ualm8aXOB|BVQzzt<u~#+-dCIdruv5Q0uty%pKC
z;W{+7M*&B9)7uPCikyC2Dr-x#e2%NlzpFXzdx*%-&a&pQ{Yuj-wJ6kxXH!Uk^S%h>
z)r)hV+44IMvDNhW8>T>H#RA4lZgZ+(i#vY_bfWW?wiQ)XxTmk}3PMd8+zQ#S@wq5{
zea5=w)d8~Ghl6C}>x>R{Tx4!Pfl*nty1UxFV354c3Fe?~kqu^NRzL;>Gye`B{04_m
zlAWE#4yE86nB#WjZk=n&AZL2`h#7U!=Tx6+JN_<SqZ{7#?byNmgw1pK3RE=iebuWF
zuuM>n*bS-SO0i~+b{R@`P<|-%@)YVFPkVM;JG)2d&w99rrG(i;mx!7L6VZHV*^k(p
zX|<WaHWyH#6j)v;@Psb4%SFI6z*AZB_Vv4%BCYp=Sb{flHOlD}Zhz@T`~9RGrr(dh
znuoZuJgb6{>1X=-)@S?P0?^5L5pUsL<PhI;KN2U%$_B*J$Xq>_vWpELHiC?e-6A{r
z(TYajQ*A|kJ<4(3xX@`YxjOmcvVr4AgTC*gq7zI|#ktJ&jM*kQm`P1-^K*KzrnqGB
zG7r}@jtDJ!Fan$i#B&&yP_|535IR$R^Ai{gj|k<E@x+d%jG|e4b6Dtj>q9HE$tM=H
ztrpYQpNZ}rvCOm5Jhb)bCd^=2N)P7i2;al+Za<sNh067XJSUtmn@9XeJmhC^zOsP5
z^e^nH#TI|txf)hRzy*3@jI<tos9_O51KSHQ$k$68w0Al!&h)Ea!q<NP2};t26Ttjc
zvi&j4NyX77-G{dH;BC`TvgoV(a7T=FlZ#!!+WZLzf%%`bD|n=rrUk?WPutSS*-R`|
z!Jr&tw{c55Se9$K*)D@eyJ!eHSd+w3j*f#!kY0&K)MUIuzS-kqJ?*2h(WU(ca}1h|
z^vAj&)P<lLZ-o<y2`?2V`SVSv%!Y7p2exPr)=2sMAfIQ<R}p=6w|Pwc$VsXa)zUCT
zc%C`B08=SqoZ^76Y2!Z9ChK0u)2{#1=wsV-($28@qXl|CzrCc1oXWk|6*B!Y?t$rX
zf2ah-Elt&56Q&N9#C7|BEC!rnzm@mCTft%-|D93F{O*Y-l=+j);=#GCt`0elSHluM
zGIBCDb~o)TT>-FrsLIdKu=xhwvY1(hlU)>4t|CrsCPGuP+VCk))oF({Nqv{ln3G31
zbi#G$b>m6C^AM>-yY`7tFHQ0A^ZD0ZZp9#7mv(X()Qrlbn=m8W#C<xndajisc*-lR
z33drt-!R(LWzzBO9RIp3;tFcIMKFEVPHrzGH~X1tSLn^OTJ~)U_Pd%#Iek1T)68TS
z*{kZu_h`fY1NGHZ%~Hd$l;Z9wPQT*$+>3F56I4btEGk6O2ns)<{!x>MJcHq0@crdO
z`xlrsllR_zDxS7fVQ6I%zeO+zj(T{p3&)ew$DPayz)QALfByCN>d)t<cv_i3{>6!6
zm#I3<8h+E~w!ge7Q<gu7=lp)*nBp(4aDxuXOBX!;_qYi6uI;>e0Lyl4<k_7D7M%RX
zF;vH;8>^_$nz(@s?eiYAWj$_dPCeX>+9V0g21wq-4ZD7Knu&a1R{QDj%d_#J4H_9}
z6Jf^V+Jg>qoX^a^Bvm%iZ#2b-zfTkmVZu3D=}I@i@VT*t<RV49m(TcwJ{cWWZ;6D{
z5yC^5?26meRnoUP2n&vgG7LhB)jdMEO`h}EH)6da+s^hS5*c_z8@>Q3mKI_$Ne)lr
zH6M56UisQ?Ubf?_Ycf=)V72=alZxrFh49H_P$S3o;D@{Mr)!9ma%~X=23&HO>!9x|
zGbh{{v3y))fJ~}^b!-IQBJlZ0jI=6(kgpvse^Gw(r*KRVUda7!Q-@0Tcv}f($n#W!
zJx+f*%9WL;I#6DS=5C%bvo@J(?s-Aj+w;1^m{Jp+qBDb{l7P91o=KazUytB+khMmp
zrOV7Hx&=o@^zfwnnA7o9^scHhS~?Q*+HDpx{o+vcOuBVrqDDXM1pWh3*knd+Eqh`;
zUl(`THuM>=ZQ_`vT0D5omyhq|Jt4(rId2uKtmuWhU&n?xTl689!(zHzw{L`FI&j`G
zGc&-eYa~cuZ6^9@fUo7uWz=<Sp#L@5(=6k9%cVE^ddYPY)Q#DXnPWgHxyz~g(^gf<
zx$3pF_Xaf#KxtoCiPhSfET$hTWy?&f>lb0b$jW-mTv>_s{K=}nAJ`2=tc9@+;R|Ph
z;j$)1bd+y$y+bt>Se*R0Pf=WOo>i$;-=^jxK!FqT^t@%Xu_`~!bl2$YonxBKLz<JO
zq``Qs=$QDQr}v-EQXKq^{?(1=M=&JDD186Y-siW>X7Z+1i{(#RG4S2wyOgre<BnG1
zlGF!27m*HxZEE9n^jml9#?zEvH(t_;R!ZJkJiF$zGgdE+6ptnNam}04;~oI|w}zg$
z`n|$c)vcs?`r0bQ{F&QV3mdWV$F9o_2c*eONL-@A4&Bx6CR=>=DRJ`n1cP(vuh+LE
zzxRT?SA&yOx#g91!_pbF-0&cgECaRaPgQs?LiOt4+IhN0SBJ$6v4QDxS{pdmJvwDV
z!;z3ZO2|0sjW*8HA0i^VXgBm!mX@bCPL*`|GujY;9*P-_E&N>7=MT(Oe+hwXuvqmm
zNv-Ijtp|LJfiSYt%t1xk4hovN<?N*zho~uyBTxC6yyL|^k>6Knj7@;zLnpnHD2&0Z
zAm5Yxw>mMl^zdRZ)kM}&J6uGft4K`oQwHL@&*(b|Ww<5%mFMP%liJCWSviO#T?Mkd
zmXZyBp*tvs<@`(UiZAHC^9kOcvrWZmwbX|6Ift!hI!vXJ3K=FRf_BBL8k|~{*~;(c
z3)c67ja3aG**46?B1+}itbDzLxs8pRhgG-nyQaKaX+AI&#<gF&syIDkC(ePd+dm{K
zE008(b{eSbw#?(Y{5g>yNbw&1y#WUr&nymo1IQ010^eaUKJHu9#NCj_YgCi$=2%%d
zxG0R?GPV>BJ~vUI_IZJcdHS)x1_>`^6+q5d;Ed$W&?j0+V=&-C%)(afZ;6=C2KEhJ
z^P6TXDBvgCCnCH=P$Eg1+Om(KLyapIpPUa4lUZjIjpW*2IOVm$;IZU5Ck<^>I$dVj
z!;NQ`GCHuh%O3_R#e2b0Zzfx?)Z=L7q;wmO4h1!dO^e@V42c)>>HG~BCvNj#xK&~!
zl}lu&LzCfREp(Lx%|sBXQuu=gTlWiqmMTM;x)lD^ZWIQ5CxWXMw17z`JhsEqTT*?j
zFF59VJ+t@@<(x$XX|%yJhqG*D^{hD43wuX2{2@tq$EhfGmee@|z|_TETt2)XU-HC4
zRU17ay5}mDZs?D3g3#U16-m3b9dD&#3nQ^eaEZ{A=caX7D*C_P)H|DIN70%XrZK{%
z>$lPv2tveE#?ZEE=u5f8^1I--WNVHa7qK1I=<Xu+(5>SRd@5z;h<doiWua|Fn%K~k
zr59ucC6~klWWWZ{&tKsf)bCB1g3I@Jv-{p6w2h;@X0#>9@$ahXSO|$}j<@F$e$Xak
zUQTp9Rym4qc?Fc<4&htlSV6TN?=l!cc$rMU;^)m`XwNoKSI})pH6RM>cdsYiSOpO@
z&g*J)EQTyX78-#Jyc#sfq}8<5ID>-yR4aUq4y6%*F*WV4fD*U8X&blJ79?S3HQ&Q@
zsSagR>hAo068I0Zib9(P`plT)&5w(sbq&8XyF^CKw{P0R<>toVfwz;nA@n|c75LiS
z?V*u;xANiHi*ov$1Mqn-b=-I3n+sHkgqWQude$5#e{Qy2-QlA|)~eu{af`Qo@di)k
zZJv)o-)Q-C4JP}B8j2K$a4TqOXEdDK5$Cbse~>R=?wD36Z=oU_%iA*(Q&6qs<GQ^6
z)F$N5-#W~H0R&8?fF976c@+1r7LZL^JZ%9jFkbKF`N42Cnxo1j=v>BU?j$4gFtLH@
zNjQFZ%;awj|8X@~{K(Z~uKu{EwoI7-C+<|m9#f=1TlW)3Th5y=4E%cj;U4<oL!a-3
zx&y^I%Su8&g8a1(508Fj-SIzAFZmEd1;)iuuUCGHV}rW4u)uc-jEsP3fpxm)aggzj
zpn5)b`@@m^@#X6af^yu|7MrOE+XVC_${}iJXX#;$54EIhCxov7I<(<7G1+z-YNrx^
zuhIgApqJLL4tS+MGW<*ggL?gTGbMf4{`GN?pWF#Ii&o1c*Mzbv34Zp~V6cy-rz<8V
zdP9u*X;x&ZY3EKxf_r5WS8@&ygvr>1J_qZ|Mh2MeF6?ax0zSdCv-$ePj=?6>U~8ME
z(x*}E#X)%Rx@y6ye*?Fjh0Q+GAWdbNzP|R!p{dk;U{v84BtI9+-OHR!51RhAwd3yp
zna}G4_K+?=SoGsk*7`=Z;P2ak?`0e$XU9+c!@%b`?NmhxB`q|;5{tKccHP;R6shj;
z*veL;>Ww@u%L1pP#A-(SQ$%>*r!Sz=mxyLF&$T&(K8N_z&(64(=ed0<j2ZoKPzdt0
z*{!r0qH7`urHa3RIv*!*64@;ye-~VO6=21o@g<#31{?ipV`>mW1ErK#XTXWN{9!k1
zw3`o6YOs0kiD&>v95o8c)Kd5~^(w9>PLwx4Fs*gm3b-`8I_xWl-wR;jG0IRd`|fx<
z8fM@FjvexCr5pR7=ZPm6NG=gs@+p6PW39qBSpK6|-((=uhRa8H%v?=GE$NQt>{f1M
zT$PGgo|XYbV|$^B1Jyq)35s9FVm6<B%&M<e5enuy952_~O_<o!?BHefIPIw*m7rbg
z0XyX@QClRi>e1QCPRk7bsuUitu@h19T3nIye~BMA@`~9c7)$x3EmIj5Yj0`zeT{^f
zOF@^b(!faX0iPbx2W5kgM3&2p15Dd_i}f<Nt^!A7)A(jMb}|Sm1Z;(pG-C1?qGl7o
z`RRH`@{Wpf(}Ul?$Z<*hn3xbYt8z>ezZ&bZ>yz)$jEm?TqI<yJy%HGtEehm0U3p|s
z)nynpLkW~%5qf)xZnY-!<dAN@<EsWc8KcX=rGOBUD%dIHvijocB0;pwrbcCtuHBLJ
z>~sD2pEK4b#TV0}vU&5`jzM9DNC)rP`?SIgvnf5($8Ier5~H^WD5G=J%30Lks+>zV
z!!*cH$1yv@kdBZ-P?{W$Law^ol(DcoQ3U3ea!vD(NX@22L!p-jO|FE&^70l<BjCOa
z9dfru%-Pscp?DAfWGRvV>cv^g;xxr>Rps-orjcvb61cu|7D}iNo}GozIJ`_%PPl{f
zr7?T2)pkKB{#9@M$xbv&5`~%0Ol#|z?(nd|E_Fn0PL`7X8}LO~N%T7Mhb`-TY%P2-
zZ?3@wnFHI?x$2h}n(KP=2B}UFqrJUH{SyDt82DwB*ikb=*#b$Wn9yUvU>SG)Kk=-?
zpRmd{oeE{8bN@ZU){XJK>dqKY4%rZPgUpoF6LKnv9n7JZUaYq+Ht{_4f&LDN3Xytu
zdSu%Qc{i?;NsnV|3W(qSsCv&-#1(vBc*b1)^;;goY(03S801N=`UT-xeWX~;$h^6%
zqe*s&y3e0678_oX0m9Yg-W<19C?kLC<8t&=iXg;)%OQYv@T)er^d~RLx?#7}JiYOr
z+TS~D5}7}(Q<U0#8U^7X9G?0WG0oW_LdZ(*2HWhMX_+=KQz}FbWr4BoTbDiLmYSB=
z_vkYRIdudSx)eKPBwwnfWpa<W5f-Do3rYEPMtE~B)t6*z!?6<eyGk*r&$Gec&V!ml
z(!<g*k~^`k2B#UVwsc-RUQ&sJX_eF7Io0H}36G~`N4LhcYg0_u2Wn_KWahVr<=~Gh
zr)kMPlOS7wNm%`sD5ke|M`+HpSNl5j3I-(=E6^33aK2t>em|CwJ3Ho{)P$_JrGQtq
zB-YxOELh+q{Lx{FBY#a5eNoxjC&~GJ!F|qRsyw!A58OI%rifYCz-#?uXNh7@++v>h
z{kp)X9>Rj|cc-$L6Ta5$Guy>VzA$qdv!2!Iy$E;;a4ifQ-A6h$cFz-&n40(W#WFvX
zT|yak0Cjrx`q$GD$;GIqpP;U6hE~%g7ask|UP>d7o5G?d^_fCPbP=;S<1_?*JRkkO
zSO~g*%HXRs23DaSWzp$5<c=<--T0e2(C|3cLIz!m=a?<WleUIWzQ1bQv9_*dn%Yyc
z{}X|gX@5J<1vxc$!U64Pj&LnrNoRKVnbW>hN?E*@z8!&ir^Xb)^)eM<T)Q<4H<-F8
zo+zJw&t-To9hcBz(nXSrLO-N_bc)5~b#XSKc;fz*olt2=@RQe?hq+B*kqCGxX>a6k
zfp%2puGZ`~6nN6Le@0v6F(y5F(ILZHE5x8S9L!&!Ed2PfS1>_^rOBQ9TrcEKO5QM)
zf2MazE4Efi)3^Ubf!JH9$<HY^T3nDI4%jG(wd?Fn<VjK;r#g(L6}9*spn@$1fm}Q<
z{kj$b)-{Okq87Rz;YBT`^E^=E&Up~q3*#;W%(9|T);PNNLpIcV=`{tJj|X4=ghW~P
zP%KzDM@D5B(`Xg8UkeKMu>AwD<~EJ+Gd3K;cF^28aaXVr(fV!XtBYV|Ocj^Y#Y3T7
zsk$XzYR8e`<EomLri6)fdQ)OMqubJOW|z0XQ*c0fV2~4h`V{yBCqknc-=xUrTymxK
zq#AyZx01G6oY=o7hYmtG^s5hlPi@OwkzQW$8QdvKAfg8kYl2#gcx=_Hnd?_t@QTH{
z&RF-(D?DG5OpI~ifQk!Ue1do4M2n@)Ic<WDufME^N<kk%G(s|4uIjpIo^%dq7`Iuc
zG%!<!aXmzd(y-xs{t6(y5RbPsxJJMgC>3*-?wf^#;g%M*U>%Zqzt}5yB4w~$aodDQ
zWIxX1qGyw2x2Z9qiT>X9m-EW5bU4Il_yVZFCYc$O7L{ms8k{>@DO_opbJVLXTj-nD
z%Sp>pc24l)#;!^Z3r(g<<r0wCA0lalMwu0QZ-$diIbJygFp$N}J?Nl5XE@iLEjoUW
z?^C~;-gTUz+~X{ZhAbFnaQFV!D_XMOrp%>1*^AkjDVVG^u-F@vu@Qnr*8RqQ_$2qY
zt<`@>M!vnPp{aHK%4lL-V#EEzB=1u{n@N$HMF8MYA)Tm9Yxgq@f2?RsEx%-^bUbzt
zT{uF&75~B*wUiSuaY=aV@|)3@{&l`SDXH|IM&I_=D@avU+y;frtcU+B9rI6@fbIW3
z&b|UHs;zq;5EN;pMAD?CyG6RYhLV<U=@L*-Lb|)VyOr+ll%Yc!W`O@eyx+a|llysp
z|NA`NXJ(i=XP>p#UVH7e-t{hUZ}e-N2{d9`i`t`BwRA5Dp?IpH_g$)qN68cDi{*5B
zi`S>0a(~l)U%z-OMU3WI7l<bc<~lU=gO4!V=xx0(X*!zwqqkEeF&b9azMo-@Evy}p
z&M*rSaXj6OlnwIC$Jc|A5i0dTmG)f>IG$y}XQ{)q!&xJiF*CexD#UNs9Z5e+X2B=W
zgvnv&ioq$7)5}&?)5=Y~lh4Q1!C@jSA!Jl2%XPtG$lK63)^~fspnNxmfMGpyKFc=E
zUy1bFV|<(8E^fY=F^!yRL9ESrgH_?tfH{4a^yTVP;b_-1%Q0!j^-U@onu;cIHPfIr
zr$Zy=CEJ59l?fgKaTy^dD<B<QHp&xGfw5qExY{Fd^*q6<h^abdiGif(_@}WgE8NIM
zoD&s<=~~@_YRQ9;E`cZqkW6AAdn)8;3me60{*JdHB2k2qV3QgWx12tqi~a~5zLE{P
zY|ObSiF^_&%n+LepK<nF-)x(0pPk;jI?_rK^qB;TO)im*H#~Mz?w>}O4f_zt)@ltC
zKr#m9p6}$b#sO}~ig)fzxQJ<xBx0)jxmuSIZ>s9u7()kgj^gw8LQ!F4+!ctOQ1sH$
zn(9Q4iN}iaTohrvDbKpAsL<9;>>>G(i8Is~FQ?G@J4~C+nL56=b5Kk9x$5>*ZQYYS
zWkbVwJwQm@7I?R(kEZRS_Z56PkNOLw3K}(5F5*m?CMa!shl3N$qO=*FQ<uV0p1Hgn
z1F-$!6A*nVC$pN{eCS9f(pqtpI)NOm61NKi0qQ^d9iF$l2utXs%rma$WDf-AG?Cbc
zT#vVgo*<qKFl9RxBcBbOznoLbU!4wkNJquVg7iumGWr}+wT1`1NOJljY2J>7Ahhmi
zlD){yKN(-U=$FrM?4%@w-@fFUFzC9@kkyx+ucTg);n-4m+ETD){{%rX6Ke`aCQc0(
zCCjPS2kE=7;MGDin~S>6S6`Q?SR=0Ph=d)@JwLP0rbm*C*5Ey&=qNs5Iwvan>I^iT
zf2`U2wC$Qph8OwXxS5<V4|JRX?84z%o1k}f8us;Lx15`inQ_SJC9yR;xqKiK{bo#N
zX5zM7<)Xb$r9+laQY5nxXcsGu9X6aBnyN;<`fiN76#@TU*yg+SRU^d+dJrLHiM}`r
zM+>yNBR%Zw{F@P`XNIPNP}4&|g;;b1N$oBSqc(O&4(M8B?V*$$KI|Nf&jFC3jmn*2
zk>qm)gToZ5TtxP>&}<Gf0^7J0Kff<F2a~?y{=FF+kyO+@o5Ydp&ap;u@r8t}NSxAY
z>$Lt(50xQl0nu{$;`Dom{#_XJ@n~Zgl`eZ59fBJMOT@2{Kr$;ul-2p`o!rR*5J&_k
z4rqm4(YEoc?QjmmW@T^-s6@pj?(O1(l72e^iW9-nNw>=frF3aUZeycH-HYsNGt9%V
zQ{#I398)IJLOnrTLt7T?>#sIva%xa}Rqok5rGOD6bk`9g#^07KHm1uBx`T-n)Q`EA
zzpp8Ez9HoCsIzgw4^DP^7pK}a)cY0BmZ@|s+%q82(24Xmlc+w)4(CZ8%gAiOIY0zt
zfG<5jX{fY6(vg#3DC7R+TS$OZ<d%J@vBx&-Hbd3E%g%?3$fSp|)b{v;T%fT6xX-$Y
zeWKGV0>=iI^*Kq8L5&IZ5GV=VOhhOX@5Fui(C;FASk*l18X3G)xbA5?GRNxcM<N~$
zmdE$65_uzh1xxbM&6BoezCF?gPy)zISjTxq9_n<I&Mlm<xR<ll_T)jWeJBe7J71QB
z@YH$oGwX>ND|dt~qSW$9un{CFy27_R&iLazPY;$N*7q5DT`!+rR0$j#>5)xnF-RM!
zRr?%ZQlFVrF4QG=J=~~E_>>k{*CyZ(M+wrWBb+-RXKL}!%J+PZ)t=?#nXiU$OeQQZ
z<!@w&p(M%as*9n#FuU=v-`<bg>5OnG%Rp(OyzP$v1_aN-bl>mKIwxLh<W+N7-_4yW
zq{XAl^oeQ6NObuQ-0C^+_NWb%z<BBI(+mHG@8$emKhJIbaPSAI8~Da=*xgfVYTeup
zrD2?owo!c|-UDkJaxfZl#Px}tI6?KhC*hL|-pE1ChH*2V`>YG=IhX=`_ty{bOgL6S
zyk6`iAcK5am*WaT;cSS9>6p~BJW}2g8H%kV=pxUDEuYfe2xC<~Y*EYY$H^n#X-)Sk
zX~W)@x~lISVy%jl^Ft<{=mc+M8R#LhI-I>LOc9iNUDM0IrD+gT)PilQrWY|oYWs#v
z&Wq9&!;d|B`g>*XtF#6sCnS0qAu+~8fvwUK{@lt1K~T;5s{semip>1Mp&V;b0jUxL
z)zSikZjK2~eYwgm9y3U>G&9i2)CjD5`QBCqj`69#)B5<9|ELma1mF&grKpX9-<Qnd
z!UKosXSNNUG;fnHg(okPyw1j|UItBku>-G(7$9K13Y2T2U$?PsTVMh2r10<z8QGHc
zCU8+Bl1bl*Y(Re)4^I8whLB)ArVW}`x1Ln2hR8@uGgxU)v!9t{E{$_>ZQU`@a})<W
zc;;L7J2Eb8HMw_@F4(t**YrDDy1dYuFk3&O+(MOC8VlH#bCt<RF<+rEUN^cVsB#jV
z=C16TF4$RLtCoi8Ussw%+?MGGkXSMGLw(a9vNND{do27iTT)N&scbp2eV|nrHWrgZ
z^LLMem=z6do1`QVmNU3`sZwt}R*AHF{nZWv5{wM*$9z)LG2&u_uJEn=598u|S>e8v
zQy<9f-lx*z2PLB|ty&sg`k)i)0#$ks^*j2{2hdZ8>V^8kOw>udq1Y&lm%~O)`hc$q
ziO$z{b9%$i=YWKGwh`y=Rn0Vaf5d6iye%PLV|{(NR!oC*(yE8uOjO0;_V`D9sVQyg
z(6m+>B%CN#ZSmc@rM|IJH;GnLO%E?53i-svJIs*`st0uri+5ArxJczt>mkVm1*QiA
znWXABg*-^}%HN?~Wu8w;1IqK}V4>Zru1`W*1IwBMOpW94M+V9g6W__>aV!R?Y(QR^
z);eW}S1lH7NbOA40Ts(IV}@GqamX+uA+Z||{fH8ZWys#X1XDcp6rYo*Mw>B^jWS=P
zw<dop!AJQ<(}S<`_F$TI#%-6|o^kVk#b_{=@X7=eQLr1Lg%Z<mb>6>}9P6$brC=^-
z_|&P9b_Oc-YXOtQ2q(__;G2nG(kzPDWp}=cAw#_UzNL8ga+~O!!(V;=A}PO`G+z1q
z>AU#(YvK&Xe0R4?2?MdLb>HJNVVx$&U`PY_bgL(ZgQZw;!?9*epTaP|xcl~<Ol$c#
zxRkA-Fez@i{V#Sd$0Bjyb!UJ5cNSY~fye8+N(ObF6d=+DwQY?<|J}K5SCR!)%dY7h
zR|D3K1G(v`@iC_@+WVb$NdcH+mPr|^JE|wr(wE(*C7VNiYF?xt55J|cqhBGp`W=!p
z3c_HQ!p7#1Jjt+F7vrbp+eQeFsZful2xr;Nc)mJqqNPQy#mCww1Rs^_Tzotwy<==T
zQ9Yw?p1Hj$O&UlHUpyUOQ?VsCNU_fP!D(G7eLKr~gkz{YC%_HcR)fb-?p&&z-;eBl
zQ&_`H-P6@nd~Lazz)!lv(tb7wq?tpy&6NXYVeiJAOb8JQGKW-bXDrw1Vwa8>l02}?
zOXqgBooU)nl3;Gzx(ji*$ueqRHre(Qx!J;a#D9)(c!3?qfa+F!*$A7WtuQPzLNIWQ
z6g`MJE2NSFB$njfk|zogzFjBdIsPbzuNQ&@9<;1ZMuC3_exy2Vl-Uh`hLT-BDi>w$
zMU#BYh+CNCGzVmCKQNCZ&uvT&7m6QwsT&(&T{dPj%*(?cKaUU*>KZ9phckG=zEf&}
zjR1Bjbg~b!%@~qd)7+_UAlqFJLIn@4e3?@mf2;fWMPhdYT5rh;6??2q_HwYqJ9F%(
z6@3~kIL$W2b2V$5nw3^u&u#k78|~tA<TP8zPUF@O4fC6FD*HSi%Fe$p4$NG2HhPbO
zpq%D3>xnG~m1+vC*|LA<$#Dqr?3$Dxd-H<qnVIqfH!k%dBu=?0LQ~%NkQav=!%T!4
z$mO!ut~)0UIppoauzY{8f*EqcqTZQ|?rMhMdixg4Nf@8IzNv|R)5B5hlc!-IQm#=P
zMxU2(bq<$~aK9kRa$~3{!vis#t0|~&6;qYDU+Aa6_}_=v&)PBjn^;GHPdGHJiAx&J
zNyEo<KcQ_m48)uTOM9(ut~=SjCV2TAz4jCmWW!!Pkcw@L6V#O4>PdbLUR15)t+6vL
z88q5}dp6RMsi`%hGFWDup+)AAe<ZSK<CY|U+edCw8sB_8L!@HDazeeWRuZ5c{m|pu
zKk<==yIzhbB&2FAeN(a8Ny%}l5gl94%7VOTrZ@$1Y!}ir)Ossm8#5I$IyB;l#Gv3S
zUTH;$`OyF3qXR>a7xRf;!@2J`8AP{)5^0y!Wg8`C+u8c1g*R}9rEfGTUprn-CAPg7
zR=}EGv_~Z8Z>%yH-7XzvgIZi(s7krzm~Lg1%Z;%qSa`#!ChK))E})|2OUF!_pEIVC
z<vC9c=A#-V%Zw?|`f*DQ1V99548@l^kp@8mV+W2-GdZsqvUIHzu(Z;W-5SXwDJIIX
zz6}6n@6t+Rf%KqN&a#T*0YjPozzgjHC+&J{!jb?Ug#oVQe8?*s7Z>6+(xK%2@HN|q
zJtj$}6vF%huowgLNIg<zAlpSh&x&YzcQL%Xjuq~{xS_C3=$rl}IenZ6Os5|2*Ghx8
zk1O+6#HWrb=7z^S8%*6)J!0}#kKz7SxvwjgSX#>19-SF*k(_J=*)bq`(HC#F7mgMh
zai<B}uWOJkktE0zsE;b97InJzF-mD+xtOvt1V=ukGOn*Agw>fmUQcD+FI7RkFT?!t
z4CTp!d8wC6acQ%D%&u>T^sdi_{KjmVb^ixd6$Lz-U6djn&m4l83VV#gL?gYT2PVEY
z((Kj^v5aQog}sd_kh2ne&b-Dus9s3{=*RC0jOFj`AW6Q*v0c=#II1PV7utPiQ`#JS
z%uciJE)Ucy!LfhFE(^<Mc(ZsyXD5E7&UYKb_H&M$NF3{{ZBjWsS9xA3M`JP~1cEI7
zIlbf$qg|~KJOF_J!1++hgSVNoO^6sa3sE}!xLXGLEQ_nV&0S$pDU+PSL<1wWKgv>@
zUp&!5#}L-k%IO(i_VbDRg9+*SSaJl9v^7V3NHXcmTr)rGa>vu>ZgG3OIY<mC-(s|Q
z>_*wyx^9prbmgv*j<9obQE#Br(dxX}>Xp$Gh;-^4R^2#bz{(8X0`MXAlf9*KiW{%I
zctK7*huJ7F9mOz=W86hE;m>w^NA11l{q94m;`B-;E9Wy?S&>jx<dh0u>6nBtTZ?J!
zbJ+3>B)oAjBSdxb;+=uJJE6SE75+Ci!wKyEIa-r^6^_k~ro<)9D6Ofmu!%dut0#Tw
zr9J|Rw>9*TWyp=y1~c7wI>Gfrid~oh9Ai0-;poACB5^Ptiq8_0ZatS`oi!9D5_P>B
zI*B$@clcnVm=4=+Vu3Ha?SxUVg<e1`dwJljw&0$$MM{YREoPbn0@`@SYG|IoN~%(z
zSx*gv$zK|md2r&8mGKkej$a;4%{5`eBqo%neHBFiY-q^uNxg2rhb=E7BcqK*l+CO_
ziKHe2_G@)!&2V+<QaybN;H|jhBm(|Y8|cA_wF<O!8$f?{39p~P$+m*_W<Ax?c3Uoc
z**1T=6PJj^DiZi33;g6H54*<OZRqqk-q97U*-u(F4^CWZU$-RGbg~M5Wdct5d$(1;
zoochJErem=PG}e6;iQTTsI9dk78<6J#<%>2ffdFh{T^9qi)SgNv9T9SsKY|gIhzR@
zaXt!tmjYAIi*G%Op;<2{PC^NbGBf)x#$j-L)H`DKV>a#p1l{=wkB<IVd#fJ<-h0XG
z0po2q9yIv5mX!bv&O?#3F)<@=JJh=I=jq@-{84}5k@@qppe!KO6XhrKd%Ei@@u<dK
zWu5;&FYY0TK}JSq_)gL22C`6I5?uI`DXAB+2uIfhe;z)_7+M1s{Hi|zMon<gCpAs)
z82=ple_nmd1&FaaCgfTIu2T^~SGjRKOPT9XoJZ?DGLsf1tEKiqieG8M`k`>pt&Gz?
zSf}gD@iN1&_H_6>HocZsKd(pg%lgmw!tLw7ij%h&31AqD<zm`=aD8FCyZtdC_i)3m
z!-a+f9?i|otwxbRGaf*8f9TiGSJCPK(-k9xT<gyDAKLwZL91w|-n;(d!{^e{TH=;6
zGA$3aP5=A1z%M*vG5hD74=RZNu<WJIfWcP1VEcx6trxEnD<i3J9IPnVZlDn-!J0Sc
zeuRJtJQME*CQ0*)onAIrp#8Tf8_CtZJi;mZ(<SibK;k(5;f@#M=D!Yrgd(xwaMhgq
zhsWVDO3n+H0_E6o&RLmbTYmpz1geF+Mj~{Vt7c6rf$BGOf{b;Cqdce}`Z0dq4Qw(J
zFM-)OXE>_8?NBS;AHt?qPc$_uB8&*B91V9RnsEDN6oz$hXHx+et^7Yob`59XpjkYZ
zy*v)>YVB+;);-l)QZqRXQ^=ArTMc>-ujHQSjl8<%Q=aw=%zMfwr=T#>9OyH1!`M_K
zcx^2N?y88YjTx;N0@LU56pm9o3O@a7@AJmTZBjGM0_agJWBC<4Bd|D@H8r~)dQF=M
z`;v6<YVyd0oHp&76D4Cd8<WohJ~~{<xEC1<2MEcUV(VJA%PT5^p%=wozv>eV+;hZi
zX)qEO@6QKXhGQ=GM8P?=N5#+zPFa!nz<y!zct%~`<($zrQNY2WYNFD}ACKF_qPk{s
z^>ACWwwv?p3j^5c@<KuGO#)oTl@3$O@Lpn=D=EI|L?@#6MN=sK!mpL^t3rAxK9L52
zbVbD;;G~YkTA*EZXcK-UEnKLMloVbS3FA+|Y0h%KSaBDR(g!^c4I~`?^(8E}YdA83
z%If)uW@WNS!1mXXf~d=yLpuihBNXdVPqw?kRcf;n`b5qr9~`%YEM<bLRy`RoC&em~
zqK2MFKSSrzk9W08i(+$;t<I6TvUKetu;(qcm6`pwRS1}UDG*EBJ3_ul{&#-Soi_*A
z$BO;^kv9%2HWmF_J+8A!{Vx%aR=T4gI}>yEF*{FHvUGwK*M>5r3e#O58JRBkFl}^K
z)bH!;&q9db`R31m=;C&*ci5>r_DZ}<-Oq`jPWC>z8cY6;oP3l;{G_sB32UZ?4tmHy
zfw}omiGjU!oBDB#?FLH}UE8NpLXVLZ^Ggq_5}h`ubW<av=02bcQa5srCS8gUQ3&l^
zff5L0;)6;kU!W=4)j1gr^hS!zh<BLnj%NFsJ||+iKp6ZD*s>nVkbG?es49~_Relej
zE8W%+U~YeTEC|%gY8s!egg*JjY<Pr|CN^}CZ5YaECy~pvBE|5jxRorJGiCKH#~6>k
z3;Tv`fWsn1nRbThR(#luD)^i}k60zH;z|WYVO^9;jY5)jp6XGa%4iRzSG3%^yWJpS
zwmvXndHe>lb+0@u?8B<ykYVYe09~rbAV|Ti`m|yp9s5C%Ry}GeEB&Sq{);cf7cF{u
z()!2<22`x3cDqaHG|9|v7fTy0cKF7xn=6abfz%N=J;u$z^?S}Gljg}Lsf5auEKs$<
zVts^XZESj!??g=NOG&8{F_I1H!?Tu*GIML9z8in^f?7K=$dSJIAsphjNcZ}U!$7-p
z=^uvV=Kql#!!LT?vNETcNI3&2uw{tL1eJwaJS!LP$QZ`MJqO8PcvB)ZR<40<o7D7j
zD$j&zeR7sO>4)v)P;lwzS6#uEE^8%V9=Go?B<+h`mj_Yfi1J&zY8%YaN&L5~`9(Gk
z2T0?Y(WzV*^Qv9b!f0y0HRM@4PKJhbukGuPsQ7x{w0{5g8E>q6vC{Apx|bli?k+yS
zI$O=BlDWTBA(KyT!j4{ndsf;_(`hEdCI7xh%krtech}n_5@enh@A3sW8Ypuy*G!EP
z`hFS>)XSk|TdKG{x5byT2WU^)y{RI{Fj)&n|AN(?PjEw5wMwWWKPvyF%P;2f@9Jax
zEmiFgE&!lcNq~{=gNIN13IMJrt9keu7^E#xM#SbmjRj9qsQOWf^2xqeITQsKohoO<
zuGF=Bma_dNPd=5~8CyxOey+tbuY&_^)+tRLv^pkk_iO;U53RLu-{5FtXlm8>Z$9VW
zMDj<F^IJ4;QjZ2-hdFGo1w(=+kNe~3q{r39mH;Bz-+TQ{Bu0Z_)LeQ4b%TbEl)d5H
zY^muKGjyU>FNd6(nu)8%9(|>18G;@fLz>_)3)oZUBeQ{<h-m+~-Su7`gbnQFsHm^y
zl%@Anfj)RW&W`FwNT`SRdNh*Mt-H>7Kj_wM`rv(!AVtG2p4A+lU#o*~GeVAcqzp3(
z3;Thlnrzy$<4b}M&MRdBca(d)ZVK(#vN{I4zFv<(J=gL3@~(?Ird$0vQHd?2##TY3
zD%UzA0K6y0{kfP3`(`cvi(lZu3bce@l;5)^g2iF8((pwmlsA@|1(oEF)>h(O&^}Tn
zeKF)cDhDH|-nBHDZ#>&`dG1tVIRt)3F;i_vH+KPMy1Sojj`hlNv2rQ7gu#`kJNj^K
zigM`?ijAekcM}c$-}Pht3D95LWmt8a9bTDnsj5r2JwTTab8WMp*gaVC+-3u)n&{e;
zCh~1rr>Q@vRypg}S%6f200=RodUi~hR$`fG%q#`>TLbY!WLzlzlHdQ2oA!7AbK*Wp
z<zuza)ER_;0<MAt7!*?-i7PgcacLNkA0tEUildqqGtss%%ZHO(YTg}NnwOhQ;B;@#
z6-Z6L2LG}p{C%c@TUJp6?vr*5b9!Q<q_k=n;9Id-RpW9Oc0V|OO3pF^oq8&E-@y1m
zP>h+Rxnb}ZrjpC;7WCto0@r<~B2>2}HI~+1dxnD3N!zAxK~1;1nKWw$dnDhg7wrj@
z7>{kgbBL-Zr=(<%BmLB8!?gapr{$%@0#E~#lckReld<E^=7a7Q8s-_)-wmiKEgc3O
zMoBRmiamSwRb*_g)*)x2xQzXY;Gifuj?MabFpNKzAk}XU8Hr`cWQ8yBD1}uAste!&
zaejl+MB4jZIR?$MhLz}gsnS(v*_8Lg1?DPIa4okCm4it51#Qv<T?^=Z1wTEQHB@6;
z>2~-u%8dH^C&6W<$9j+T04+646Ilo?0mfgy7R}S$?@5cNA^StEqQ>nQo@pVQswrCp
z>B{?K0S(h5OaIWl{;w-y{=Iv#+;~Sc6}E29{AA`89ldFYepHGHwvrINa%#<!D0u0s
z4X3(Czbrn$yl3*zsA3||*uc7@tyfI7Av?R1fZLgMZBHBr3kT=EW%%6(w!iu~KRs`Z
zcIm{-Oj=M%^}g&(q(+wLh0WV+8)y`_LcEcSIZf1$z9ojs00k4{5(0*9%ma2-g4<84
zDP3FrY;Ms)Z9z`v`WWN$qa8;Z9BLH^u=keQK8Z{mT$$#af`Y7)S-=<QIb@lYH;A*r
z)qs@cNEB$y4CP63{l-88^6*e31D>2A;b4LP4gxF=P-tg>^TM(H6@xfR!4sI4t+VDQ
zEgz|Xxdn_)wnSOp{R+<9abZlrt&3b-T<^+tyX4jfABd1w$OHbzXnoluJuZb4r~$Y3
zWJFk+0*k8A+{HP#;R13zFj0Gh+xyoU`C*w7!-0iUCLC?&14jXJf2ZM<i!&?z2pOF4
z+N|0^-4>%Y+P2Bc4P6cueQ|P>ePAB7A+9x2qxRrliPJHW^}?wEv*12o%&oIyCLy50
zso>~%b4ID)Q2D9@#NUn2FK-hM@0J?5_{KU3Z5o94OOtenPSf)=K7PoXw27U6GG(#<
zaH9;E^=Zqob8x)3SdTmG`9oudQvY8J!LR!dZwSEDJ)k?>!IIjDh>DQB=!0Zg=JgS0
zmYyf^r7g(E&k)AjlX9BAj1aZnV#@$h2A${I_no>hBXZPrpmlCby}MLgcX#q%v_|7i
z>_7leAsrn{EN>5S^Jy>N_LMcf#C@>f`FI*I=$TYFO2l|CpFZOEMje!vf$>=<lWxH-
zo59kHAVkwGAhcUrR+^)|+`FvTIy+p8E1XcTIf}Hm-xQC;J=X7=t+OoAS;?C!?6BGF
ze`KeU9R)esky7wp{`HA}6$(6{-QW7XEe;iwB4f-hObdmOiu(iU<pBGD6urv;dgt+=
zJisrZ>~W;_Wd?Ihb}h6MAH3@7(!G*jI}F&RqEUUNx|8AoDRv;IGP%)~cRbf{sxxi!
zFDx@6Nsz)-Y_W5qu$7UZz0b|Lu4)ap0hK|T?EPX-*~S*eW3T(*Q?)@WH$v*6LK2Cv
zs^+b(DuX_TnHoE$w6tetip?9LA4sDaaATusX|}J5GGg@X-Ur)s9*((w1+YdCo9xl-
z5`#W&>#akcT&4uPe<P7s2#AI6{k}CiAqbvKkb%H{x;0?*JIU-XHUx^#J~8pIKYiao
zh!BW{jnk@MzPn*m4FJ9L0ZdVH3FfRlYjzx)_NL)bxoH+TMfRLtu5B9uOLOScyNDv=
zB<8;;_g{AXpDRJNjJFAuWm<Ve+JgeqG`Ue;;aM;89L}q9ecTgUDbDX-5?@AbLDRpW
zL+-64f5h&zN?LezMc>X_4R>+Q5@#zrg1&PY-0FGw_}Hyye=3H>-b|Nz_GcN&e@#Dc
zJtAkJuE<+51Ox;@p1WV|C`Y30Q(u<8X?8r4C_YR+oQ8;R)EvJ3(ND7-H2+~)%7+r+
zcy3kVTv(aU#_q6=oetdS>h#W3Pw#Q{0OP~Fxs*#IM)9ysL^7>r1PQVnfl@MtcYaQv
zUaWBa8gaS79{|b#VEldXnN^Tq^e1;PxBL7Hi2;1P7=~WA{?+vXIn4kHKfKi<`&+~F
zV{RHTw3-BN80g<!m}|94#UK+civ9W<9b*(uOkT98bmimo7?66IMPIf;n1zuST^R2o
zmy>7c-42LcNO6T4IQ-<eZ%Z7FXYa@W`1_o_LoK%aGoa^#j3(RYzpyy%=pNY=Iy!aY
zVJ@itsD)%Ea$IEaO)5#EIq^U!S0TX_jp+Bbmqf9#Rh57ngh%y8=N9#wVzn=$`uda@
zzy+{lb<ytq{sM^?aA9Fg9yq3xw06Sh&(w~hM^436kLU`Ie<P(I@@1@choh!IBbFo(
zO&zG{yln-{gg=N4bMxP{RNF=-z++F%?)>TW<*nOT7ygTzPGGP)SAG`9nQ%AZGX-`B
zPf47bzd$nnYX$x1Z4d_>yvn!UDnmc6M~%tkp%XtVF#gvac{PwQbb})@n2r(wi=(G~
zQ}2TR+~5x#T)PYWCSFdgK)_u<jK^z}=l6QgTY^lGpKEgDqIKt?5+K3W!!+C9Yrn@R
ziL{=$UfMbFGjqTqwol1*dHmknMr2ImP3w6rBk)l`?la3;IS?F|HDmVq<uQJdOHd!2
z!Yk!(TWil6ADc^2-}x6m0r0{jDpbNc1H;=AMO<)H(J!NZ25bKi)-QLt{sXT+4xqvU
z3EX!7W(|zNqbVtZCtnotr~sP(82LTpZ`zjqcvxOCEg{6L*P%f{Pc)wc;)`JDu)N7=
z{MTpSf?)39u-VWT7gNG3>v{(Z{u=|Bc@GGgeiEn`zo!*UWAJ(DHy+iMVQa_ppm<DW
z1qSIF3jw1Gf4cQw1G(|7#VdE%mwP+1fEX@&$%X#0H}jwJ@1KU#Ei~L_2s+%t)C#th
zefaxcfWQ-bI0N~n$8)`8ZPzVTey2VvS1dOhsf_P`_r)Iyarzu^t?T1t0{%_cQ&{-o
z7r@s?MT{_hZ|(re?p?8PICG>B{AA(y2NnEdpt|b7*K<P#Nq)bKfuDSf0le^2J3%(0
z->=`RNxo_@phB9)_wT;V;*spkLGmx8(k|f95~h%zy!bPR#i*d5kWPVX&**!c)wrz&
z>}=#zCOUq8NLx}x<R&RqSnv-uuj~$s{#ApuKOX(YSN(nnSjsXQ8fHL_&+seK1e*;)
zDj99s;O#7eVxCq4#%Esz(o0K2dHpvrhC~UkwsMb_$G?5N#76+>0TDeADDsV>m<-36
za(90n<B-%|`e^@Uogt-{I0-hh)^ZS=v@01Rh)vf^sdF+s`t#>S$FdGl#MCFi{IiP=
z|2}X+k447ww*R&g0m`VXmOxE0J-}ei_%SEH5eq?fHA7t`>=b|$kn}IMkOZZ5hK7fJ
z$!q0=e@NAz-bCI1Kc({1Y<cq9g8kO#{m{0iqJe++K*1!TaV$vXL3tk)u=*c6A;Eom
z1dLTDbMUuYNvWzwiya~|Dspmiz-4(_THa`K78<CPi4SL4aK+t|xBSZFmy>P%&f@dt
z;E>VKu&mFH6_UU!G0vos7)Nz!{NJvCEe(F><xAnY1SBU$-w;M`G{#l+!m3b3Z%}>P
zvAbK?K4;w({M&(GBFoDYAeCo|BbiQOF@7JNi0#%ZSnNfhK-3ijl+Q$D5bbwstkluN
z&#%RUPsGxe)Gg<Z(QTM3P(Q;aTuiqda8TJh_+q8}TLnIV^F>#>jZaNM@#gGP$yazM
z`xjb#Bx-=F>WY1~`<*#Y96^>>P(Zkif<EXR;aOt8l6-pfj(X*9>)r#0+G4&Dc7MSa
z;+1fg`eKOP%%yAuv|Qj;GxhNS2la`=!q<qBa(tix2IAEDk-clAOpP!4*tj8kT(V|;
zoWb;|MHd=jcIHT|+Y5EB<A+KOinqFN-ZMOK$fXH21|HjU?bt<I2is?qPN2UBB0W3^
zn?Dtn364~pa|Go9^NW+yiJpdhl_2nNp!r#W#rcOJNiplZO{J-d9eyA^(mTHbgo;O*
z6B`l7wjn!iK{7QJ=ZC%h`U}Z-hE$K!A~aFQPA|~SM3oJe(t8$)3(hxe?{*ysUTlK|
zH6bXRFA%ZVwsHfbiJ_}2`ZH&r%0><Tmyh-BDqvA%%RP>NZ5>{EXr7-PKYwBTQuz_R
z<!=9@{i#cc9BV~oDr>BM^8+62Sf03i2^+%8^WcweYs}cga*6GJD;kJ!RFcu`ea>rL
zbBvw+|0{UvF^xsm*)#LH;ChM;<KUisW#YTH?(SwWk)nSwEjQS#TzZm3Qp2wrmNlFT
zH+DWX*1Sbd3rOa;yH&zGre<3OtM{@c!Hk7cDUaD6x@?~g6!|YWKmQ*(L&5?NtkX~X
zjT+b16oYb?BPRIq42=IAbca}HN=ttq>0oPU3UpW{^>zD;1ka=amh$%=a~B@vi@ltJ
z7~}dg>V3C{Qxy{z{hQB-Kc<;?|0+=8d}Q)7ksb;sM(jA8tJW}pLd2NZTWwxjHOg58
zjbh*ULOor(OjzR3rC4G<=bNt7N2EXRev$41v9nY#?#ao`WyEKxH)XD?`x;tK&D4fM
z{D$3rT{%Z18;`@3*%!4u4oHx}LlSkUDUD;^l+`NFCh<G%S5eZikbWdI?d9AVoR8eD
zar;mOnf(ghWLR1GN@s;}^OEd8fx~AuMe`vcvh403|FkuJT0byn6B5R3IvKe=Q~gxU
zq`aYqVXask=U{VyI%IuVfMn%Lm{oGkXxGL@>m&Q@@%9X^G<0Efs;6wOGF>uU?c!w2
zqN>4+m4UwKZ3n&e61tnd2IyQ=_XAyox>fM$Sb@4}wY}7Ze?A;fz3mKo_l9|{i1KAF
z131e3d_(Cz_?grG<c4v#z4^(9D!1=mW!a#^38^TzQ-J;bdH?IZ)VtuAORG<(y~l0@
z%F~Whb3pfaTLs#G=xo0T4}|ubpx;&+&~RL!DJ#)gA+2NCM=dd~*cUxjqc1s9)W%+b
zr{=^j>{mDHFB2?t^f-4`c<Q;AY?Wa>K=%zp#5h{&e{RIQWsRlem_4Z>k|?u=tAV%f
zpgm1H-EBDshpXk>q8gUrv23nPwJn?Xc!nM$_Ttjn1VGTu?X@*T`ty9RBFXWd73u=Y
zvZDjl&!SYDu(R#^T+ky`_InxdO-pS6hAD>Ly4}6;d89o1{Q_5mi|=BZ1q{j6`A2Gt
z#eJ@4m<wi<?gCL<H+G@}CYYB3Rxh25ni@5G=4^$d=P`t3UR+FD+i|89=?km=c+h;C
zDgTt+K9s7>EPgbJB!~uX%|&)^*^mwGdikpcHD(l4J{?!nD^{+|v<tsuPi8+;LiUlf
ze$Q`R`}uV7>ISgdX^9)4)(e|rG#BqrdkQXwsH#``z*A$cBOi$zObShwvq!)~2@S0T
z?DF?UN4@|t-zX{2_rQhPiIVh~1n92gexr*U9gFMnBnjCo2xLeTDy2-yUS!axvH!h`
z+rVx!Q<B`ib<Mhm#_fW*iQ9Q)>hiS@8Xi_|^3vHh>cj`XfNlO4^<#x(xUu&7mBt0N
zXM(IfH#8Uh;SL0eYn>D}fpadQY)d|KmB$x%dJ-<}3bL}Zv&%XomjojVK<5*?2H%QW
zyx4uwNr}0?nCtd#;(-(L&*XO*fV^i@@~O4H&%VB70#^4By^_?1G)6lvtPf(xe~v2O
z;u<}Hv=YBJ6m0LtxRT%`xJh@>XKsEMY%Cl&B*ifT{Vr9}Hky;kYVzWISYe@fqyCU7
zC5aUjMSJnqdhHo8UCD~37PHB?Za9_8c&2GXZ`q|Q|APl~_Ptd5ZI^Wuh9=x$m~EfR
zTKqwVz4i0-DZBx?0;aXl^^K-Qe;&v9)3v|`10*CQ&*Mp7>t6PrSo$5P76;2Yv|-DV
zhBig$lkg}=qld?}+i<_4VGf^QC?IYEZ-}<Qo%O-ZKH+=p)sd2#(-j&GBiFr>OS-AK
z86)D(Dg4o?GAH|!pxMGlw|Wvi&Rg_^CKj?8$bl4|ne_^nzn4O1SC`jV&5qh@92yr!
z?f7B>w^Pmf5L%kIEteRE3s+~H1=65v07<(hY?^{K9(la$(J3ui6VGeC=(m=Klufym
zFB--}v2Y>&nCS_Qjsgz|=-jr+!TSlPy3JN~Yr{n;r!6~XXe)EBZCneb&xtfI#CJ8C
z2rdsxh~G6~=gR_U*wdLy=x|-$VRn+_g~d}C-TB6po<LU%^#`B@isQAJE+~|zVP`g4
zfu^27as~1#kZ7ivf4ba7=znODC~wpKOPdjF{i6=K9#Wb*GU(JsHc-K<M96Go!5Y``
z&ga5{N6%F3lj}4D&8`Ni9KfpPZ-Iq24!1`bLL_8yJz}wd)8}8P&fa*4@wg_Pc1XEl
z*H$f~0}EnYn-#j}Rsm&mcIx<w`)<e*voBW}uUFCCyX)b-UQT*Ct1lKq6;kFU%W-*p
zyFNb9ewua;`+D5lhT;kAbSKZ)s`F<ld0)34bbyQ660*xqzDB%p4NhQpcsb}EWZ6cH
zn-B3sEv)>eV@s^2%oh0kb6f%^`zy4+Kmq@HbStI;5d8anY+NRXWn*Ix^dhBYc3TF{
z&)#)0$|%b}sGU2Vg@CgR6r%$OIU;2tBs0=6YcB;RXn^K>bI$9<{+`E{mud}`C8v;?
z<2(M~$pv1ZoB~UUI<n#@4>jF5LNZ+}qE}?25XJ)pgc%-30Gwh!f*Jkm2)EHz&9P9f
zHjaH`4)OK|*YkuP86jIl5chZ)NB3<@z&1!U)PHDrAoRW5+>IsHwwsQ>WvvpR;R`^6
zqbfB4deHE>f6XbV!&4+=+8eiEH+$W^8e9a6GgUsa*^+PpwJ-VBv|egyX<F$(8Lzhd
zeE;H5J*K)3n?)npsgLp>{NH)x&DpxK>AegNaRr4mo+<Gt#@Ua=EUnYFJGb;Z_yfJW
zdf9Z$@^Zhljeo?E;ywpSub@}V+^e<QnEdJti5Y_*lH#eMK<P)s&(>L-g+z=Nl{yYg
z*zthl*ctS=C`f|*uW}5FD%uIxp3E*hx1R5Xw7Ylb+>WXtx3*kO)=7&!=zE91mxJy3
z&N;fBw||&ka=w+#W&Oh9j8k^o`qar0*do4xGj>e^rO@M%<+7JL6C`mNX;xTaR<+`h
zLfFR$C1C7jK&_19;ouvpPogxr`FLn)27yFIxqR3+8w10<t5YFkLA}PlR;)GaQTr`B
z0ij|!!+LtL?pW>)g}a(<AMv#H#bb1doEFzCG>TZTY|%MD2Y_*X*R}XuvEe|Kl|w;Y
zK|TfxUiGSH8TDaz(cBMjBu{WYG$g~D-wwJy)_@N5m!PEVxrBnl8?}HXyG7F&BIii9
zJ|5N+Ge_4M5S*JkRY(pR?nn=n!!xN^>U1=(idkvM*K8PY!<V$*9wv{C*Z0LcT~tUw
zn-d~E2=uxi<CmXnI<=bYSZ8iLG*`73eJ<3-W)KISeeU9N{)uS;Xsx=rDKxI7`}*?Y
z7TDaafhCs1(Prk-@Gd3&KSZzcWE0hE=oz+82AI7B4i^fFTc{y=P0)$-T4*Ak<t8md
zCmJfD(t-U@poT^IX?qp2xPX9wbXsAdgtEfchLAvulgYUJK1qB|sa_Xz$|FI>;RB19
zdr$IFIIA^TH-)FIz>oT60SC*F+`#YrI<r~^=k4tcE=jm}cRG5=L%Y7~rvFvcy8lA9
z_9*YrqXWn=)heo1u;#{d8Ph+`jLVXf^vl8NR;zkrTiGpu8Ti0boK*@I$u?|+rr}{t
zrBGmW`#0Dl+;z5+m=~y^Iqtd^7g@%mvHHEQK$o5m8Sz<?pE8zRAfvenxvu=+%f*NW
zRHd;(+J}_E|AbR{iYe{>l?QZDWbvIx!`hTJ?}MkOC&}u<am59-wyU~1x5JvZ{;SWA
z(lv`(glRO+ga=0ehtXiWa^qrrd4DBV++83Sg+VR)N;kGX7ByV!hU~l!ZX7<m^6LEM
z@o@#7p1s9#Gjgw8h;6-O6z3z5>#>VJK(vK4`#yX{sr~`l44pRgyI@0Vj?LOMd3Wmf
z`Qo;2Ah9go7eAAdgRfWc_Vp=48Q8e>yH56#IHvRV6sd=Lro!QPi>5Jp{QM53=5xIO
zVnro~F_XYJrBvNveGMDj!(<+{(Bf06Ql`P=+yP)1efHZv65&Z2I4useB`~r;+oW3b
zX&Mev78tLfp*IC5cwG>%4iXb=#J(o2Q{iN-yB!g~5K(xhe33H;rOR`af0>wuIQ-4t
zh}@~$_4;VhkAzTdyYU7~H)$_oYKPu_R-F*RdEyyvTHKme^M|%Ou=n3A0rs7q=dAB?
zQ1|6T)jk*CH}MenLRN*cE7tVID|EhRt>W%c7F4&Uf8^N2K`20qtb`JOHt><A_2vkm
z2GA!7RqI`4lhXVZ7pG{}%_13*mY4=H#L!)XyHlAv_^82FO%4V=IeqFDNe)|w(T{kW
z5$qOc?Cqnixvd-?7}7g_E7ISLlyMEq6>qp~da)mpaM2m-HjvSLghk6>zMZ?v^*_YH
z$k<*=&V)WkUb4HXGIZY+x{wyag@HSBQ7jbCxCVVSw!{TTtq$OgsLxS~FaNL;Ara~(
zW{|G9yKRt;uE9Q=L=)_a^M!^UPwN+Ga8F%#Qxak#G#zHphUi-KlUKDBIV(ET&D9><
z#wL5PuqO-|bK5WmTXNL;cs>bDTRbS?(eLHju(9smeFXj*p}}N7?Vw5fQKVT?aMt;_
z+B`BWM+0enrs0Iox)&iL3L&G1j+DR+L@PPRb<Cx8;X3BBpqfy>IA~FwggJLI8PUD?
ztjlvrDX87Zz-2hrZ5t0LhFt0vhM<44#b=IQiFaz6cG+2+e=u`&vDE5jAA;A=DLR@@
zM%h-w%+cEV$vnoF$aT60O(Tr%8)WWOznOFPW@UnbJH&pV%>Y?niOZYnxx(IMcZyoi
zs_c6$@DGDM_N+Ll4zm+aXX_lJA9>wLEz64AsY_7gqIeW!ezmtu;Vr-YvA66g^2Y6S
zrtSh~+dFNn+wbVWCq+#BW~tWmDS|{gX6Fr>(IuwRUY#z-G<RK-H+mzvC!m|8a}1`F
zPoLMwKL5lsvnOV~7rWk#1DwIjo9^deD(*~WR8(z#M(+6}LN5g*ya%9%QVq<2V~YWA
z&3V4LFUVsl0AV*-ov#!)jApUmb=%LFyBq{TKlFL?;WS*fXnu749Hw~z<>~%%cj+^=
zq;Wm5R<Mb8pSnZUX@$A#SHL4A&1rVNd5hvr)PS?z^0IrFuM<KT?QkmB=dD7$UF&4I
zTOrHjcvh>TJui^`DV5_@`|(8zc%ekzw3xsnoCF90im{4qKDpZPd2=wyDEuz{>i32s
zqG(*I*jt(mE|S=VANwtNKK3(d^{b?}?GyoTH_W_tbslny@+DS}mrmpNJp%gyHMaQ|
zi(%=hsZX_|oERL35v_OP+!)8)j)ltLp1i0#d6Ld9_Ju#cn%qRri3jUl(Qdw454xL`
zHSISM<<s^r0>o@>(0v}4Q50h4r`o|B%so^T0W6}*kH9^l9i&ki#A>=`g|H+QP&x;#
zuXq`KWo=evY)6sWba@X_K9b~P;XJ>cTzP6+hek->PByPI+HC}e?)TFAq5Z%!-(;75
zWbX#2oi1gsV8=~fHf^8SM*;e!{#^BnCg@s$Jp#X!xwDf$z#n4Z)pJ<lR>B47(-~rH
zd#g(Fr*JZ9ixyNCMJAiJ9P!c*a4834u6O5fPOP^|)#t~!zJI~=LIPajC)f@@&?P`+
z)a6Bt<kU@tojB$(FSU-hxtN)znXPx{`G(TOA`(LyGBclBkc@DXp-z{Z?o4WVG3)Op
z!isp&ufJJo^ti?6SY79I9Nino5mWEt_za$AxXg?{Vr$N=v37S+3>;@V<wdqtTy82s
z!vX5FZ2QEh&Tjw4a6k8kcG4skue27<{a^Y68`h*$$CTN#o<K?weP4x;JY2>1;<~Cl
z#q8$ygvocCw(I!fw$)>f%s>7BrPxWuaCGTskH7YP@7m&gn80Ijn_z|}vY`zv#(62w
zjG1_IGR^`eEiDa<f4PI44#XGk?x&b<t5gnC0s9{<Bsnk^Tb%%MK~ph@*`2_%@{MHt
zNxAHkp6Tzi`H9QDgdB{6ID^K_Aq`a-W3Rii55Me@wh;a-BrNp2E+p*sI6p$1>g;@F
zlVZfE+91R;WEe^DbGm~daD*I0dIxfI$(ogN=Yw(CTjk7|d`);gGMxavx+CN{V_W*f
zKgGZM$>99MISI-kqh}J@?q6k?y}k>XYP|!9C%UjB)I-0|IaHX;uCp{T!sQ79L1RVh
zY&PkCSegHUZnO(|5OU1s*eWjt5Su=iBEapw1fnKajxz$!fM3P5oiE$Y=BfVT3Fc@0
z&mZ!ad~iPW0&s}atF%&mf_A{~D|=Mu0g4}S*82Tk?w)RLhisJ!?LWBV(`#!va~TY^
z+=l91YP22%*V(NG{gpbN)-~72*cVmiyYJaihTBofMsxXk^vM->yfOhHb8M31x~z`M
z3!uOzxXZxT<@dM(SwsS-ZVc<W<U0~ss2@+DVG+@ic806Q1(2juEdUP5#Pn^y%An^e
zTn7O7ln)mv!2pRiGg9&R#dfcj>P|uZx}5$=;Uh5R+=<ysEg$1<1j)+dEDhqfyD}-n
zVQ6-lpYih{j@b&z0lbm`+94S5qT~MBN%5~IxjzG=gP0HK4c*xs&U6{fdr|N~19a`2
z0~JyQw2eS-2oL?q(7XCj0Fn+ktK*q))Bvx|@DBj<+qD!vfQFd%HYx!Ifa3d-#E(5;
z_W20IkRAy4M}Le){aw}p2JX`n3jp2r4%HYjz@LVFjI&}-;L3*<<iqUnE7pjmx6Y2D
zvrSg{4t)dE4zS39Kr^45mku>6njUtH=FU0S7z1aA+$;a^Xt-$UCX@EJ&v<KjBYFvY
zV&&4?Xg)@%$;;~k3KvoMY9EN3Nk5SexlGe1lKa~)ZgSmzeijehmgsiKzxz6GEbwhh
zitgX5$d6g!1O&*5^8BBnmp}6j9zKXzjJkTGJ=riTk8@%f^?|(8JpIkdh`&yK@l$DO
z>6tuD6_v0w9HZxQuveLy^bI*;4_&|^cp9SpFHqn~yxQ8fxF*0c#>ZHeOAW4fHqPIq
zOpF64B4}c|(dYe}Fm*+)z@vt1^GW<FZ-Vvdu`Q0aLAur4hxdf963p~?f8{p&hs)t1
zVPRw6F(x46(VdpY#$)C?yp7lnI-=MHF($Gp3Xt#tMhG75ADyN@r|u9JK=^2h`uTg_
z=ALe7xIit^uY-@R>h<VFK=3fRT1XNSlfPR>9*Q*5s;X~>%L>Gi?sRsk)d|VLwyT0~
zDrNmAHx7}cKN}&GEh*R3_Rb|x6N^}E=$XUUgk7u{%r!w74O540mdwr~LCI->_6I-K
zpR4%4Ul~$)eE$4ePVaKcZC<PYyQYlgcqxFjHhob3IOkhfn1W!7Cz-XksLIs+C;zA1
zr=!Q<=ldKW0T0yTOrq+Sd@!bB{t_aVMdTC|kH?eamEL<&0RMG2@jvuTV=sJpbr*yt
z?88@jWu?FYR@nc8=%0fBhgZB{xb<C{(k85BJ4Amk-wPh6V_4!bg=enNkAmt}J9QD`
zmecn+0=CA<C4JweLCKMkYl0dFBQA?~H7A?5;ksmT?vsdVYT0Mq#gL6#$04=bWZkY)
znh4iDj0(Oup8oFS$PgXOI4B(LuwUc+&0)D7vVA^V^Z^{l#$s{{E;iA9SZ&Gg6`VK1
zsjU2$Vv-Lgtl=8Kk{sL+G5&`U|Eb%&A9?$!kY!b6O*~;3_g}U{Up>8V6dsPF{_5Rl
zKX?6XM0mLSDBvh6625&MIAvwKXV@>-?f3yrHJi62|A!dUpZedg0*eZd7_G<ehs5E8
za0sCHLg4r4vltHkAzb(0O40rbcn*3?#Z}DT-s<Y6x}h*)?{tT4`th#!@kw0TuJ2K{
zr^&|q0?{-DeHnMiZxRUC3k*c~S8kpT4);;krTRCIPfV*|f8J2{Jzxs)=<nHFXJtH&
z;G-pI1<0_1rCFynn0jy8GY%4Y{ZEtqFUI@cyl@kj3re6tK;Y|O;KTicr(w4OxaLK?
zrr;0WtNB|Vt?5B`^GOPu@5psv{I>{r8-3x<ydRvCleI@AH^UJJ1U>qBQ+EKX;QgfO
zXQU7BdE(qVLqn7a^H;jcW2-mizk$*uSH9=K^e!gD6{~!iD7?l^yKBO?D=kMv=@R@9
zPW}CU&DIV4XvF;8kzD#;W9z#$?ylN`JVOJ0c7G&r=6Qf=g0f_MM7%QFBwYx}QIye}
zj`1qoSk*W9tKVXI#l-wOL_2W}tjreiXn&ksBmqktj5*Xz@?#q*7U7LR7|{H3<9tn!
zW1toz>KUx8G5wsSr$7gR$K<G_*TP$M0VImi<$1u(VSfVzY4Am&is)g<&l*%TWOz6>
z&bMx8aR3W{FJ;OPyV?E{_B@RP3&}vcZ;k#2)#H5(d!8%AvwzJI@GTu?0sA7x{;x6+
zLwK1a9DWpQc2jM|Q_OjW1YLT9D=<Ek27|=Qd*^u+4Dy@=`K{~Q3wng(1WQM5Vo$Kz
zAKYL&4W)Q__6Pa^1k06et|H}i_Pzwi!XRJvkU#nP03O2cHnCM$O>J)Y`D!Rk0zjzn
z!?d}hBPc)pVjd#u?med;`u&Hk*G=Pn+N>I@!e(jnqk#55?(dJidiCAaui-vidFk$R
z@Zu&2cvMj4tjF-mC+3oP_kPSK4$`Y=0L;;Bl=25%)t^Seoi_viWHUofT>fgP&d)Hq
zmt<-So*(RKMJn+0Qs`Vi0J{ec&SG~pTT08XOP1FF0k}Yk?C`aNLQ--CVA(uPvj6c_
zw#SV$6KwPYD42hBB7Ufbm8^T8tCiIx9wUISD&&pt(731Nf}vS|3$bu_xs;ylckVxc
zR;$M#sZ6FBc3O#tS3U?}u>U<$2xGU3m(j_)?@m-p%ag{(zN86BHz~z`?XsyNBqX0B
z3@#E+6x{*Ukfgr#FTAl;Y7Z@<z&WR`>VoYSl;beY;~#MD|LtP!dh^mol6t^QP&(5^
zZJ~I=_}f{qdjGja{NpQLCpdVe(u}17Yhdb)x$#i1SBeoPtKDKL0lO9G1t?ckEv9Zp
zT24;5H=cv0r-wSO-uCGrRhYtvteo1|n(<_gvYL|l_T02aSZQnytD8*e&iAz9_0Pj&
zEMJJX_v!<dL0=JMrjGoMmWQLPlb8N67~qOcQrTR<5c3vv`T^bgs*d^Up{-PSZ{62$
zK2<O>G78??ldaMjIpk^{=<g@HBZ1*t6%2&_RrVG~Gp4Fm+&3q!r4M+dcdGW&Lm?L5
z#ziKIO(P__j=)Z4Tf2^gJL@aV!U~7a)a<`E=jb+{3=nEW<Tg_lY=@rT2^3W^`jA__
zz)ns@C9GJc)^<qbNJBkF9Z#bp6j(TAYm2%!T(YsnDM`R1h(jH#6KKB@IAlPfyX9Xn
z{ybmPvFsxOH>0*SsbqXGE=%2;zL`SjU=)0IR&6e{e*$VuPyls%<-MRWh{SFG{3*+4
z*Rrq2X0?Ncx{wvzisiHf1%BF(?WitvS>c|jH*V;t$1t)^Xl;G&#EFGe!8R6kD|&=a
z45(=|$nLT>*0l4YJn;M!Q_-<jf*n`QER(j6OR=Thv$(q&LL*Ez>^kcE%GAb2d`4rY
zeoC>cuP=fw#iu5sR`;xc?}=)<RGgt^4W#XZoPjB3b@=>l68~pr`^_)p6|?=1ygrXE
z{bOd^77n&Z6J@whguLn*?`{9-Xmi2qZ3(avM}(Y<qcMYCusGWij)qGw^~y{gb{j7U
zXSL78)V2n3RV|0xwe2_v0@|YwW0upg$2#ZvZn2v@sc-I~jA%#RVCje|SxM`FSK_Pz
zKdV}9P5VAlUDlvCWoB;_bgI--I>BDbhx&reIDF5dvsGd$w#QDht4maOtql3|=)g0j
zQj}FqGx~t1m&*&?RjYN7V@i8SXz2i!)Z1GMNJz4MNXdNUy~WfH4tx-QX+L#gvBSB`
zHF6I7H#Qp*YfhR3Ot7-O&V6-Of}%}K3l`(Im*L?YJ|_@w4zsFp-IA5YfNXf*J~50@
zWrS&mAZHwdrtr&5b4&>X{>u_vzwL*>MSrNO>l1Wp5SlJxL3qsryVb*CP4m-atzyEe
zhc7iQ=ydFu-X{W8D(m%awDi~AK_e}Q`PThjn?JRn7zf12hdmPU#-wju9E?aE-NF}v
z{5^c&iEj<Mm#xF6B~r<$p@*O%=Z=>l9m#_6&iZjK<Irwo{N?hP2oVRBtcnk46(i<f
z2aR=me0r?b57@GD0cefp6!kv1dvkZPSH8ddfv?R6Y&lU;KfS?sMcaDSU;HaryW-yl
zKAX%hFc-ENCmKqGugkzOz`I|Ac`PI$5eJcWImpzuG^nY_@nNTtR4F!$|A@~pII%Un
z@kEM~CYn&d|Al(8Smc4ghgZ>C5HT<FL6umNyQT~_Y?VnRTRaEtL*-IUTGF#8X9{uE
zn|B$A*;iDz>z7(1Khjlq_o0O(33i2*q<7@v$!(TjEO4z5eVpod4UIlN3Kdz}4BVEx
zb{YQ5LarjiqUQGZIg6+c;GRWUO94)(!Hd<q*K4fqabnY*-dlYEW|?!eJ0|MpYSCPL
zw~)`0bLT3*qQY5h7c`-sEm23SzYgj!De4{Oad4PG?vYJu!d7MEZ}Cf0>e}C6o-nx+
zSfF8ieCXkvz5M?u`^tbQx2|nzDN#}y6+tAWn?X`35$RH?LAnK`L?i_SiJ?2Bqy&Z*
z>6$@$=p4Gcz8m8_o}=gWeE%2*?$~>;zSgy%Qd_f1AtGkQ8p}e~lE%@qg4}c|l}6sJ
z&`A~)+nwVfOFJ_Kkj&^T2z$lK1Xbb^t;iDu1xHdMBDmNFhCYMjWQ}4I18NTV44^07
zH4=&><bw90z-T7eDy_g6MMqCBaTuVbpgU81IPz-V*wZr2w3a4f&>4S~CsEFxiGb}l
zp0Iw&Y)%;!^}uF$x)c|+P=yfTxu;WAx~H?Ox|(3zT>ePK3@scGei=Z6-5_d8-*mT$
zefnFhl2<Q=J=0|(jpWAIc%Gft<)7GrFn!;Mf&6<bM?CQ@78dzK+Lv7oE}7Z+r&y_M
z7?<_AuY$aVG87XY*ZR|V)G!~#R6nUaa6Ha49(wcX6Y}8|=Xx-4p%aULQ!)w~1$Q!m
zRVn!Lr#A!a!3#~+;r<e|$VFS+m}>E;eGLw$Nb-jLXcy`gbcMu7R!zDXY>A`vmAqx+
z{#Q}FwWQ&uhP{5^48Klor?%GidivqBh;}VBT|GpFI-BnMh${`gO&wmCa{Leu1|{>(
zmy<!vP`rDehe``J#EHT?A95;sHYPW~b1Aeo>XIFc(xbNo`LZ!I1`(AWU}@aW_sVKo
zQ#58)2F<HzEAewWwAkBA>I@QC1h#AWUL-MI8Ewb0$sX#Z{2or+athF*x^h0l-!3Cc
zr{WXhZLW53B+fhu1-cf-u;>X`q?z>FDZIG<;BgYbxdoUf1%6I^I(~JV`h-f$jbq0s
zFWEtm5LP!NmlP`8N**o_`f}owiR{5~it$$d!3996H@0Y;pm8+CY-#;@g^#^VLF}Nz
zrI=o(G>E$%eIQnzcq~O6GbeF;LbSZk?lAN;Sw_t@r_p@L<IxhDp01)K?Yds^L|gP0
z^$IeKNdlAL9bhhIbMN%ul;xI_R&O&C+VeM#^JA}6koHIOO&Pl0CSTXGGtd}(v<?qX
zk@G|rT^SRw=2S*rZGJ;eGH%{cgBkcdu*;_m`H)3%7jJDNVWgjhnd0IXWq1ck*HL9E
zG;U4T3)?`ZH5Z~nG`77xgaf~6#Z4IyS$#whOP73Y3Z3L;mES-4Rmd4b6UU*;2HTVA
zqD|Yx;R_E!7;oRZRH+)SJFn)4kS%o|nO**BV`n}9L79N?0p53pssc~Zt4=hcOGSi$
z4PQ_Ah-g+ycso2S2v6sR7Cn-V#jLajMLi#UO!HB&T^Py!Sw(&q-3o3_q}ms?oP3X{
z$&bxOSzDX^h74ZmNxvuFy!M7*a9G%N0hi@>lGptQST16E+ita)%`VLAzq96l>P-m@
zWn^e<lo~@gmg^cU$88P=SM%oh6ps(7cTy#{Kc&Cj#{iZHG`_596bWi=mpe|DJT9f+
zPci6qwpfdSd|_-?EzWG#IbFY@gh*G4^VZfIHR5v+`Bb@9u8f02cs4(mUyUV&-=iE%
zBW8rACwORWxL4!NG_E+#NRgFlr~cpp-AS9A!ee=a(W5Y)+_f*b5vsi9vISo*FqYm+
z2Va&mawzQ-P1)+L)pw5awtHA%mC@TrmtAS8%cLY)srZbIU8UH4TJ0ckU^H&qyG(vT
z=;R+t^~XmKfXVhVLf^oU`@3fk^YV1(I-XVkh|oF=Jp_~?q246RojX8z4Y1iug#8PP
z^#NNEcL=7BBBC0((@}_^EU%42tlI$amQh7(&jX!DNUEX&sN`!*e56KG5Dim^=s{KI
z`^&uTm%yOLPY1=b0^xZQ#r!OBd<aZ4ojY##<tfMFuNNQtVXBNcx8sB#Oi4jFEFZPi
zlR$+7h=@}x5(VwL(4Ai`OSrC84<q`Wd+;c<?9C7HPs202p;2Hp{OaZr0j}jRrlrmz
z76k<b&NHCV9uosx7u1z4TIW8!q+YRN?;yd2%MH0>3cv8qu%6`OOGx4>DG$uK-;+Xj
zN-~Pcw1A4=ek8#$->e|tSQ`$aVMMlQtJm734v`uw#?FXP73Xbzu$(5%=74BdNv(;v
zEq><px;-Y?^)c9DtDoA~oN<{ye)>=g{8D7DyZTcVnT#vJk)w8}mBvR|8s{lzvt&6x
z<oFigr5!nT;Nkre3w<?woJv3{rz4ig<y(6Xr~o&YkKpCbwbIB6^{&vAa!;QzBrFC$
z*1*y(V!>BsTYQU1sld+B0`tPkJOCgaF&M))fdpN<$H|GuCr(;cpk-NfMbew!zKZ&d
zCT+<Q+9<|cvZHz=1HUygx-4+_+D7mwmf=_&K&OL1FmZd)gfj${)}zZ6#9P&NuHk=x
z`R=tBXoF*6Hzk7;`7O$L8XDp>EsEKwcr@kgD~M0vSUWNGDLGQ}vfex;V2-vR$Cm^p
z!ToEN<RZ4war`9s97Cmh4=PbeDdY)x<x;`$WtL_(1pT6Szv~pl;SS23F_)M|L&UC#
z;6->Y&4v_EJyNV9mx~j0QH(Oo&dzqYbknWx0)!}I+&=)b$b6)v6SFXg20|&02@o}K
zj-m5W-;o;y4|%L#L?QJbqoVqKNajUz&Y;%}HvLAR=e2wpmM-2EGl9Z7YkYuq;^>wU
z`Iw_yw0yvv2pXUDs-t0}cs^ws!fj)oa(g%0_`-f?gA%=P!FZ%cgD#((g>E%BH*XtV
zc%cL>Y)#1evoSXkFxZ#^V7U1ee1=LdHlbuMI`z&3_-YZ=26ZnB357<AoC3JXPf`lu
z#2JRPy#`R@%RN#c)pv`CbKlmae4aE!s^^4gd$l9z`e2R3pS!hmpkrk{Y;Lt#Jvu4+
zI=SHUq(?egC2|Ac$Z*4Ur{__s5VywOG+<dTZ1mbuj!fXVu?O*JJ^*CRG717uBB@S{
zv@rH3azoN4N<H<)4g!1$ZlqLH205Q`8vfIsJ+B(E__|QT2MDJXJ8$NaH-&SDg$&vf
z)vdY7_1EUv_nwRfYMb*y45t<Q-9OSKa++3AK33Kw(Z#%_-F&z9_k{(zqjLmzjX-Mi
zH#L%`Faa0^fXn5}#HLy55JB9@*b7~U1_wt?-1!74A0JC;bdd}>&W@NnL&gQxi}Ws|
zRxjz|mM=OxJMW5rG^T>3Kq%s{5E32U&Y5~IggT=VPYY^}lTLzF2b$IInd;LX3$=T9
za2vqPud(MM3}0S}QGhi~UMd#(JTC9eJj`s?Z)My3(RwU54TTDDNF&)b>AGWPA1*FY
z8aA(<FkCfw=BmIbk|}<HPp?+PaYf*kcV`c4jGK6rjYbXqhU;yJTMs3@K=N83S;zCz
z<-t<@d~M-cg-$`DThlD8Ukdag+iU7$8NqfFs^O5k<BVF{Cw2&h_VkjHK>-W2(2q4A
z5%`vHBeSL;ifa^p$x>YLV2)sw$mm%|ZiCk<_T)tPz2^a~T~_CkPoQei7ehXqwsT)}
z{fR`JE&>aG3I@V;=mik>H8P`3TrVWhp2bZW(%bAh|62yZMGH_J38Wd{8HhL^fxMZ4
zD%Y~8-vCzkVQM`F#pTc}Fvv5qDT86p%i%?JQBPPj$bWyUbr%S-I&85pu$7h?j+hnt
zy}F1`Ue~JFV`Ess?<|Ex0yf1|e<y@*)Gg30I-lSpPuBI(##V-8O)h(9Le`iAkT(T4
zU2^Iy_rrw{(JH5x*q9IymX#xh3Henk*9LhH`)qvOQ%TxIB?>WKhNL=ebuP;|L2BKw
z27!EXiMSNGd=@Q!8D5K_Nj9YvFZ8Vf=%?(dnhfT;yTuY&Ic$*RdM=9IWUm`zMRrPN
zA3m{!pQKf%kn6@3^`e_S_hycphUT^}aFIN8%WgV7fzN7WDAHIdB?>r=Wit>phxlXX
zTPuPw!&FSVhIr!)#s<zE?@IM(Pbw>GAG^C~d|kquCDcw97S(qAjCwrlQseDiwK4*C
zz?lBOPqdJqw|Pm+qs4!-pGi&_+mq`t_<8jtE_<E_iILDl7w5&H9Y>)dT8gWoxm_d7
zCTTp6=`?A>s0YkZu`W>u3aq$A7t~W6diGas-xn~WDL0#V9oZV*RI%6Vc?G`;R(HWI
zYYqREHM7GL14OXM?(NR354J-*wsXtQM9(jGW=$BvZ*dQaSIsUSsJA83cywXnnm>2W
z87Z04S~GP5(xh^#Sxuu#9}6F^lJH)Z@Ydy?kSO*Eq7}NG0VP^a-hz*ES}dZ+R^Od>
zPQ+G>7O)6Iq|w;DWZ+(Z3gl5%EB5hM!VS_+D2z~*!Xj)8WQJWAollNt&^7qwoRAxU
zFNKu*#A*6z5Kus4bog#!jLhDcJ7rrr233z#m?<WJn$GWy+hwExke7Dpij4fGK>)7g
zleyzz{DN`_bI)k2y)eg`Vxlx2?RBe~`Tl4InmhGq4h**>xv=<$h~uypw3MSXVD5Z8
zyp1|h!PL5UKQVD-DBAOOHtZETLN4mo%tWX=wUytwVKL#Wog7HXSD<r!iO;eWLeG7p
z#i#A|hOs2fm@z02-|-e3TM-H>qXMSiZsbLBSgj8Cv`-nBBaqKis@wEJAf~DCVJ}6j
zDTvD-%!_Yz0c7UsQ&yU9k&|OQ4s%}~_26d!jwTR}TpGrG4ij2({mj%TRVaR|%Vyr8
zC8<CC_9CTLK~C<o&Qs3kunU<kvFd9N$u<~cWA3gdXC`{UM}x=Lp@||+*P8fh@)LLC
z_Y^sz=QE$&A^m-}vjr7cw4%ph$m`z%jRnH2oSYEse3<VcK0|w5#TMD9Hj<FGY^<bX
zbk7ppk86=yRrM&6O>aVE)^d-pJ;+d7z$mgA>7PQd7w@|Z)sa*lk#NN~e(rEY1Mj$8
zPz9CCPY@GIw|oaN0-*NyrO@c(IFmfYD%|>PP1Lu=YuJ*L9dI_cjCwxb;d~~E=)j|R
zK82oGPS$J@L?NJ)A0r|%hlCqsnv(`cdmCr=R)s|BR)X|2YFjZyg=;OtH6x9Dv=yRK
z{3BOQKD6*nQy4|!z>O5QPjsG)bxZi-6?Lu+yTYpHbIDiO33?{5KfrJ53^POHf>rEK
z<EGGMz@hh!AKp}?cC#pOV+4pYoxN|_i)m(O=P5Y84{VM<SlHj}8dxsR3x0+i5tN#b
z|2I78cE*Lo`@0#udtdX}prjCt?c_31&w||dAIWh^JK}G4*L=8>p&rvJ&P76hyNj=B
z%lqKo5<N6OGgwtszer9H08*uWDf5%O&YNTi^i{3$0BM#IfSuvl(sCH-s=uW<()$z$
zVFyXK#g9HY7Y~>4Hn12a%V~jOZS@V!@~_`)IJyr%lD7plC)E=x_Fu;D&*Cw-YWfl%
zfDte@E5=2V>AUyoxow5r<h^>MCR;)#7%>sNz<~QXPV>u@l9QR(a7dxCdGiV|v=RC$
zh?LQ<Zo#$8(vU5|=%p%m-QD)!(IX=u8I<-OD*f!);_@AQ?hVY8L2nz6ch6B(<LaWz
zZ?9UGnj)Bdd@#B)@KfLw0~@A}gu3r<DrfW9CFFvwU)Ly=B99m#Z;8^vqR2v@NIc~V
zq&KTfopxczUxek-Qm{->YZsY0w9;jkkUg5}UFgbjI!L^>Vt8?(OQf9iUxZ%vTR}ef
zTaJA~#Vevc|LruR5&~;|Nc3}Rvr=om2-b@C*1)*CMC>HedumaxbXUAAp*Nj#mzw|S
z(kM9cgLPYP>-0(yDH)qu#eBS-u5+L77bTIhdfk}2Tq^Xg>s9s}FV2<jE-%!+JwAbj
zU}aRB*8S|jf8K%w9~fl_Zo<<&q&KJO0z?@TkDvUqTI}hF!%`bB!ExEzhzaf7+OQp{
zDTF#=w+`a7W!#QkyzRFucQlW#At`_1B`N+TSsC~&Kewl>x2k<b(1Tlsk@!Pvydp~h
z*{hC4Z^EYfb4kW-8WQ#uls{OY*~K)*<3^FsRPgkP#3f_Cskz>Lu$4DZiLX0$i=`#%
zupNceyYnGEwCcVU34Rvr!WC9%9PE`%Dx$g+v+%jMQ5eR%5`-+H(&EVCxO1lboi*nl
z?6Ut%M8aZ$5yB&}!y~v~*WL^7G7Ky0aD9}Z(fQ&k`W0lf-tcjZFpx)ZM!xOajP*?n
zWlYdpVtX<P(4J+jkpiZMcCA5s&yb8aL4U7^e|&Ku=-P@%TEFjnmviT{52xvQ@z7M`
z1nagB$^OM;NKW#Yc!@wv!}Gja#b-C&XOSD$ZMYfZdN<~uJL&P<__Q;i%WBYR%k4DQ
z1C{g(Pre2&;z`Tt;3gL@L0)xS7JZEB-!<sw)*V76+G|<1=ZonISQId&OmMus*65KQ
zxyZ9}ZL5}0uhrdCiRO`&{WmAR!A-Qbd(~S*dbHao1)4hB&8^0hygiL?#1lH$T=!^K
zKSYckX+=GT&-%E$fA#5!v!lh*QE?r5GQHMfM&_zg<@b>Gr)t+%tL}-Dwnk(j{KJ#=
zvcy5!UIM(~@Mg7wx73wd#RT+b{hG$fACesWUmGwg4&8ikH`zM2eH-P2@uTM_)O`1y
zO|xCym&__VImw06xd`5k28NlNk3m&#cj=dyY;A2xalN*G#NFW2-)B3*vQ%i%JQmD$
zXNWF^;LKsUHPRj~ZNES;FTM>V_bo5`2WuX^C&$W=4=_f};)CGYlZV)o;a{(?gjYo`
zlPh{k0hQaEx>9Jq_RjdMf>DYeTDF;p1l<AwQ(W%AqcQ(bA+K`H9BHE)AQgxi3Z5g#
zL>AprgjWdL&%7#Ls#)b8x*NCW=L##n)A>c8+LvvIP#>7K*sUaLh3eNolsF6pdrU2*
zLp>UUT32t9G3U%yl(JVGZBS}|TnW1oHSX)8IBq7QY=0agd>Ut#zo#*WjkoCU`mc@d
zp|YMg?-FKP|A=v2q>|Dr8mqe#JJ5Qq`KMO-eD0o$jhyiT0qd+$UOnL<{xJiQ&Yi~~
zF>-U7Wdd^H2mO!^{6mly7rs1%xN~gMuoku?^M;Bc29^Rhbmm{vJQVLve+amY*RTj=
zib(;*t8H1X2#_8Kncs~f&@8p}7q%?206(Z(t;Xt%z8_{x0k0$M)JkfM6}dL_CXywF
z+2i@PCc?3P78D$AJQPepME^iV+l)&dS#;-Fid<F)2SHoB{G~D8JzC-G&BAirNP$ik
zSVRgrPYM6Y!hOn(0Wc;Gz9TGMHZBF_a<=e(?Ph?f?V-Gn=2&W*i|K_1zex#mOgvq|
z>HTc3sj9VYn!}{=%H#IS#c2R>nSsJ%Dei`r_JS(oaUww_!LnwhxIz?=;u&G}qU&zJ
zLby?sGcDM8V^-a$780eL@g^+gdUm-ryu^9rW9Gbr^E>xhBk0gdil!1b!WVNI44+jy
zT}N^`W)q1d&~|XmceM`%;NC$YFq&IS+H%Taj)#HXwY^8WY84C^IW8_zE+46sRU3WO
zTy2b<ZEy|#9kKPfK1pH(%DBjTy2Iaz550I%O*$Kf=KHJY$#UK|#u|O1)KLc5ifNM)
zxCm@Yk=~|mbOmxE{OS4m^`f4ACteKEpP2aa%ln7y^@X%vKV6tQ;N9{PeWZUfg1HBs
zl_MP%Wx(OJW`^KXHy4}nLJ9nKl26-a1JkOic-9-V#o&&$?nXsV7y1|s>e_cpFwJ`V
z``hD&z91eAm{Tf3L`$ca@nWc_8KxgT)rt2c=u}Hi_o%gv^?9<<r{0RY#H+a!CI6r<
zAl-t?3Q6VDu+x=SdHjA$69^OZ^dI#KX6jnP?2<)a>|ttvcP}vt`DJQV#-7K5G_KFy
zfa&Dy?26+A-MUUc>!(3vxcM1tbc!eL!<Pvn=0x(Pyd9^|oVXFMF`r~frKD`-qi_JJ
z?xi>`P33LI8e|+pOwPVb_A)@nf4EnNJq53$#Lx97$hR5p-Ro6ok{{pGWqr0{n0BvU
z8iww~atHE2I>cNz?53vfcPsf%b#(z%<UHz7`iopb^LA4EHNX|6H(gs=N+p<O@9Q0$
zy&0xT#bcZ0cJVy{M1N%aJ(%NFG5ss!4474|YoM_!j1K!G@+@t`U}PEl@UbS>ZSc9t
z<aF$irM#$YI_yyMq+nh;M6nxfV6NPEG={v6N%1V+ZC<t3;S4A`DNKlShkJ_}>bb1E
zAb(bLIZf*|0p;}pmgpNXJD@;g4j3LPbENJHCN@$6>^4y#b;JiVT(%C_@~mvz&K@$v
zf(Knv+`vpmdb^njcWqvei3cIR(1mj`Tx1YUl0q=?$mR04zu{kb)&%gKqAF6a<@CE*
z%!|e*gF?)jUOl=`2ZMf!cY2{n3B?Nv?rVLWghIAL(%hllY)o6RA?Umiv62E2oodcR
z0ofTV6hm%T4LFxsmAk(^&lz<^?783*Mst}#JKs+lR)FJgTHyj&k^3DF*?@4&ZoWM`
zc||^!Q`xvXjxbzYqPzL3!!@KhqGl%}y{MhVsK|9ZIU00p{`Bzu4nq*-4cxh|0Otbd
zC3N96ajV*r;Jvs#3KQ*lX|A|;5OX%Dc^hEMD@i_2mbCw)=?^r1@p-LIYVB6(LHx)I
ztmMwZcHI@^0K#zKCZ*~7o$%5I8}!TT5YJPD;p?<BY00xu8$CB(#n)e0fd~NF*!-Q{
zY+nvUd!LryzNA6iwC!|vae;4qM|<w@@la*l%P${{iDAAIhadF3Z^!u_cR9~HbYxxG
z6*C5C6j=rhXm*>(PB3fP+F|iMFc>|MqBX{aO$nEac@*}rktLIJ1k;2hF!RcB^GkUX
zbFVizd*+YN3@l@fYUV%bd3A?DMWzNY{(T>F++aqxSF0R+s`G{yJtme!cIrQtfj@k*
z;lk`x+^7^2HKp}*?OytbsRvfW%mhF7Q6Xiox}V^FDpk?X!Zg)MIu$oC%a<R}o@Mii
z1(GahQlfskJy@Jl&rDIC<Uq|nnkZml>l*b@AagCEtZMQjZ*w)4Wra<)6W^qxj0G_<
zv7a%oGP|y{ly~5^8@$`__}1#@B6f<aN$tK!SHJ;oV$milk{fy+RqVVaQpT_xyEjoK
zz$N8j{u(PBe%GUz^6}w>;q{GvhmSiVX=0<S>K)P74^zeY0NW=|OR>xzdFTY<N({Ox
zn%1D*!8x#`l+rU+A-N;)TVgCw2u^f22hVu!)<Jr8lz+K3Cp}SIU+SUe7k+>Nh)dP}
z8~VX3feh%O=Y<fy9IbhaDFTN3PNv^%oGVSlqY(zZhFtKhXSkYDya&S?=8I0}-CY%9
zG3kmAIUqoM`gAw^<Z%*9oKPj=BriKBJdZ)eQ$;$aj(|0$^+|L7N!ODjLCa-XC$Xz|
zZTmK<GcEA&-6zGO4+acISt&GZ=UE5Zzc6q~ySaIHaBAYjrFb!)Cx#=V_@8ho@;C1|
z49{%XPR)=nrPZQ8uO*+#n>RPJ4m3`;4q#KU?bhWJs9eqIdZO#*_9+Q>C37sZTA*(N
zN7wjz4Vd&|=gZJmsc~-#`6Mx2rbG92<GX-$ba`Vu2TQEEHGF0tl+1qECGx|~`dfs6
zjP&#`hp|&VgR=*f8kZr}a&+Pvfx@KW#ti&BMOx3b2Wo1!=R5g&cPKYvu&3Dd%5GD4
z4)bnLZt9lhEvvc4L|p4LqMx$T(9>gh5`M3)QvS6lzf%!6(v~JhA+;i1lRLFx=H9e9
z(X@0(2?{PWRk<ykA(*6dYJzp3BhjD$K8xF<$-O?=rLES|`o+X(8F}T7f*C0~zM>y{
z!|O|td^M+}v2Rk<imB<D1lheYcgga`At@--tpOj?#a1GuaOYbNa8(`2N$`(hAQb%7
zbz?j5n#*rdi08h$8|#E2W<Ln`<OL)w7S*fZ#cWQ@LYnr?7?wtaB)3KBp5FnBh}P7&
z+1x9>&bF5rxaK1Bu?1tigI_JMY=P(A_;^>zh_c(>_+fd|vV+ap>Czo5*RtIE;}R$3
z2$M!j1q(N|?Qq54p;YV@lB#m8IEftd$`)!D;mtHolyBV8gip~Vbji6^2WOeXsDe$q
zM;AUz7aiG8J$!%uz{)6F-4X6gAs$MC?h*jyNPV7R_0*YiEtSk=5B_IM=;Zv%{%0Pa
zs>R<^Zt$&q-hsr2R}i5k{ad{9H?vmF@HpXUspr}ld)V%9b%k;w0F;Jj9*<D{?Uei|
zQvEGALPv>lu?XlITGC6z{(wxNRsRq8t(mL~;-nQk8$ice{|hdyzg_RQ`6@5&bA~|a
z4m_)X3xRlvbkLusqW(TJ2ym(n0ZfF3yT@rcPUsqG|ABJo922ER*E14VY|tIy1+cU+
zb9M6nR$TtR5H5;IQfU=K$Ko4#K$)~AUJLW9Y6?|A^#akwLgn;-ca4XaCyEn^KV+R#
zDjHURid&z;?*>Ulc0b)*1@1iZn(JGzv>kJy)^Lx4hwodW<Zp}Pq6+}_T;k&>&rg`!
zZP?vcQi+FqGo8RT{N*Sng`?IBpR!kkKYH|=dQ6cbaD^hvHsm?X6vpZWCdKX1u*bb4
z!9HiI_*xNf^c&m|0M3P;$#vzCo0R+#!iCDAzSs?yEAG_z`j*Embzi8&y4eiA-mQhg
z>oG?(k6U``FaHNTayIt*kkT`~-gALErn_uvx!b^TRpM)V1>Lv0GaZZ!)yq>>+}&M&
zB*yi3g0!9F={}d5_<b&dt$bbX<(^Nsuhqobqq4l<OUL^Ly4A+O!D)H=HqTDxH}w*r
zQR2m1xcuw$rV|j`#mi47%QPA4K3Q?KeQTvGQln#MSNeK^QX<*rUpEt=ssB6R6@Yy5
ztsU1w&*t7;`HDw2ECeY(L8-Mj**N~8PJf2h)Lntq;hf}tn-UOOiOMRev2VGF{b7Cd
z9p|^{Gv;Ks*lu$y*%on_eC)n)@(WS5*9Zx$+i{!gJ8st!_XEi9Y%cL<(5o%ViKB##
z@-=Z%P0<khUxypns?bsBD9c6y=igoW4R9(y<Al~<?{C0z1@I}43CU5<GoGGj=r8Mq
znV9CtY|c7azQJk#CJbV@NHUXcF6tBWzh)ks5>1><p8Nnn*D<0V;yRn&drM3_b7Uxq
z_ZxJ_dV#0=>(s<qq>12PsPN1SO6JD`1wlCH3aL&8RmunBVJPjp2e+;#TMx?IF!+JC
z{z<EWc5#Vk=407g(6Mt{PO}Zsznww*b3Soj$w%KS25T7Gu>?|5&JV#aS+BqKJ1~uu
zwJR=(-a%~t^#VViUDy^N)SzufuxJk42=uZ(X=v@>$QJ9n25Z3soI_3n8YITI^l&@?
z7a!bG;Ke=p+rYG6HP&{*1zfM41Rkf~i22Va<(vF_v7AG?<rss13!wgIV#*V=c1~L5
z+CTCiNXCI16g>iJS1j@T(DOs^%f){AT{UCqxNF&z;vfGMv<l#!wdtuYK@x$yZD&tU
z&vn*{c6j<*tPWLRO@VXO+D#CXh>AV8pp1hm!VYXxH>KT=rTyC}``KW$JOSWnU~}_#
z84vh<NqV^K@nzw4?~IAk2sy>BD?HQ)We572(6!Qhqt%9i5!or!RKJeT`Q@cXuYjN7
z+;fvijm=Ie&#FU5CywsGJqVMT^I?PYBIN0w5PcvFd)*X@uY(PlhzOgy4uD&c4`X#+
zflPqHD>8`C;n!XJhbi?U#-NCek++r#ufC7Qd|&SV#rLvs4rK>?c{m8?go)MkEl|9o
z1S^}MX=<fqec*r_QAs2rP@P9Dy)gfHaTN5Y=s=$VFjZyNsPUnm+I?%Opa(9F+=F*5
z`ZNMDOR_{Tyb{joiy0r8{)z<u$K?T|_%!kZ%m_#vdY?_Dq{NjX^Sp9w%tED7kyH;Z
z%bs^^ERd%Ek<7ov)T^mU8s2;aWw}^SQoX14z>G)TH69HIb>W;JW;CVbk5*lPISE1(
zbN7kRX-`;KoC7k)OK46I%dilcqq)B3258K+5l+sZPsi_$4bsTVn%{6O3qriPiCX=_
z$ue{9SwSNN_t9n9sPk6|OQZhNC0d+;OFYHL!A&7Ju%k&MxIpD<!giR1JJOSR4(GFf
zO8x_h_OAl8T?3-;F}I3IOdLPrn-9LT+oe<Y5(GW(tBU$F2`w-`tE8kv11%D3p|b3y
z!CHsZ_|7Y#BY<-ix3=@v^uJd9zORRuD3RZ5_3fdRib5IBiuK1IAf$li>iHyl<Bwu|
zE-Mozz+WE=R|et#<52(pW6yO!mbI>$qW}Kue;l`C7NB(}(Il$ycl|@ZEGI|{ee+8X
z&YNGp_(i5KZbR$(Erimtndk%a6XwUxYqLA)a;xqpd_cP(fEA#BvcqL^dq!yd%wJ?r
zeeHR;e5@eTIn4aOzwQg>56<moo8gIP!@+C&tzUqUtxQ%=FM&$<#iN?7YA}c7ZSPu(
zzWWtRyRn3NJ;XToj8scS28b^)<~E3<AG6gFUUN@;()33;s3E>8Xc+n~Dr%@uWNSpw
zdA+&z-QjC#{on}kYoXWx3s=i4%_M?9?j<Mym8dBsqWic(X2zs;KW}-sSjlmDgjwvQ
zv0&D1&AIZ$=9=i`FT+GiBJ2e<AHUEX4f>u+<KW^NbS){KxzAbzr`t|UHQsi9GTP+G
zuIVtnnbsa!;(CqoO3Q_FOZb=&`PFUsbuWI}E*Oc|X3@inwdyqvfG6Uu8`h=v*wqS&
zO>c2>F-LoOJJmk;tVq)v@|!A*N<jiq=JPs@&e=IuA2Pe+1m@L*TzBJ;Zq>jz+AFsG
zy&j7Yp0Ngtj|I9VfDz-pP1SOKu8WK^kc@w{WuYJt+U>Eet}<BwdX4Ru6(uGD70gfr
zy0NH4ZAWNueQsEExiTw7bEpAj)-N=NV+8JOCf)IS0Q|Amb-$3exn_Sz=`$5Ta8RPJ
zWj_?wBpu1-kMfj}`Md$A%{2APpSB}(37}!3jnF@tQ`%g0s@4SN(6F5xMYShXW6k*Y
zbu*?5xXtAvm5Pd<9+X;oO40Yk<!Z%FqhNogI@<RgW4`g4ptRX>d5m%VsQWBHp!*W)
z(7CI_+eP8_J!$-};^!4dng#Ti)NHKKYIu8`^?S7Wj-OdVz(q>L#B+fFobbMaLS8X|
zJJCroP)zR$dupOGcC=b{?-o_jQFBOQnEu)6mCC&K&>&hdX~#7Ix)9NgAnina;6MYS
zJZ=)b@h4ICTtf%=|L7@@>j+=k(srxkBM;$Qw9vHz!!gz3m5@v2OvGVFwPz{<&yiWw
zwZ;+AoX#)f1RbPLXU+By1{Ci$>W{`)0ZL@Y@uTMQ!qkz`Lnx+`Yuk)}$~@3#V;_RC
z`IJkeoN@5nb#xY`!2LE&@OQ@Huix29Kc3x@n>q2=iducX=CN*KyWe6{NgFIV<12PR
z%UAxivB=EA+9!dE=h>@P#SJxWr6sZT2_+!mKDNqul=?BhOGH(RqSF>mXlfIJ;a6bI
zI=Ef{n3vOdZRy{pu6`C0QL=K|UA%+3>b0U&9gem8;hIlNbhsfmYtFnA+q*`OmNx{7
zjoWm~#_9X|qDDYCo%S|}_Yu2Qn3^wk<@BTQDfQpZ%^AD0oDx`FkTx~V78zynUux2C
z7e#fy^4R}!Z3q1yxzPbSyJ!)+nB$Y?adrdgkaus`Yp<W3EQyt00*3p%b*kO!-29?{
zDy@g8(KhamVWWM%CbV)OWL2rmz4h5ZNz^2Y{9v|#CTw1BkZ#Sj8)1lq)mu_Rm5p9<
z#XZz{1m@F@(dN>anZ=0zQgwYo8<5f$LZ$zP#QdLyuoX{APnQCa5$Q!mf(+$Phr-Ne
z*jFobad2@6jWaGrBLjd(xR(#LjXYkCs*U(!(sjSlC-9;1Z~(hZaKZ;)AW#1EC1=W#
z*Yjds^eqkHYhyufXMlS=Su0>9`9}ejKs_R$iagt24KZp<!**l}m7#If8rz+#G@ANc
z$XkB>iVvm?IO%gnC&F+#gO$@`m8G>k_D22Vgm>?Vn{`r9sDLVINppsL0x|(t=K88*
zLPhXP&v8IMtRLZViT=sySTFU&lm!b|BmG`Xj5)RK$dR?r<soL<^<6>6kXeb<lcloe
zpff&#6HUDrrhy}F^S1>YkJFaNTtSzp1=#dY=9T2!Kdd~Y&FoIFD@f#il3O`SRV)dh
zHQDeeYe*(^R}tApkN>m=6HZ=-hnSQs$<tNFlR8Xnx5IjR2%X(K109i3Y<;rs?#p|t
zIq%u-85pFR50-02vTK_!Eu5MmoDeJfVgOs$3=M^tV^AeEj?c)-%W~K6G70!hU{3Ns
zrmAKl&n@^U3V<Al^!5#DTi`NT+3mSGFsQypE0?beb_$8-H{w#0Pv9O~EzsNxD~lQJ
zpA}EgF0QvEgG{_pPL(cq_%zvn^R9ix6&zG|fGq#h8dt~u|Bl!1f2u1(UDA6o6s+J)
z3y|PFEu(dv5^$=*ozq>g6rxm@PrN=mYvi~$+gZ@Ixs;;G0;G#?nRKb~_bunJ3EA$B
zI9Iwg?8wnyHFx{0GY)$kvQ*7<apon+oGn`C>K%jMG~xc^DPGW7w;5*hznCefZhaUY
zao3yHaczjba(B77Ig&*W9EIyo`wZ|gfce-jG*6bbhSqzSH;0tmKId(w4AmaxYcJOr
z7d*G!!CkGm8Aw)+fZ81IGgXQuc{q32j2_ILh`j?8x@*;YXhx^=CA!czsCko8LxZk|
zK|d>`-yVyZtU5V%D+iex>v~nkBZ3g&1)9p8Ig(XJ3yb*5Jj!B|&JRaCJf*hFgZCg}
zP9H}->RE^gFZr(Q9+L3^S`7a9jevUByP~3pv7J+&8t*#;oogxN>?L@Vyq{}a&R~u_
zTRx}Uyu2<-g>OLp=h+Ce%;0}5%KdmkP(BV2UGg5i3ic?+wvOho3`&3gg>v;B?eVsU
zVX`KLq&HQK5$8Q(GO~8l@RhsmKnvc=(v1JKRMakP!dF52W9v(2yBXM|fo;3OfE-;Y
z<NWQec7U>w;A8&jMpNZlD7DaDo09vf*<N(9%~qRHSH=+(JSj{Wnv8Y(Hd~9CkW=OS
zAc#);-0Js3Qoqq4{&=|GENN&lD#g`1Zj1EJMoW$rj>=0t)G7>0q1|qasr~6gEE~$2
zs;ntuQSEW<5N3OUE4pfF<>j)GfLnPO5!E^(pzg;;n<C}SdF~(qxTP?s>XSDN90sAp
zq{KRwHxD0L=EKvaGa&AtqqNoYLV@<3;RYdjQan_aDWAb^X|Fo<n$5n*RmUo+so0Bt
z`OVX;tCn@Hz2>n*V^CaD$+OL<=-QTV6O9u{)=!TR5Fv+0<0oaJ`FC7(M{3VpIuK*5
zIckl%jcmVnu=b@U&ePkzgcs2}%BH8qINp71X4Ai--)GSxLOY=L!l(R%cQ06|n+<i8
zBr7++PyryRN~%<FSUlV!V;iRPFY4)ewWF_5sT_0nS}Or&XA#{Oqe0lmc4zu37?ruX
zdDt6vtG7#0td2Qx0xH)^cN`v?7wH37IN`aqd^XmH>grZY&(?j2a^EhOBftda0Cc-3
z@6xjV7+RiFP3!UT<9X(z4|C8R$`R!u50B~dbh;XlPCvAAFz@g>@E|D5D>45(UPAOH
ztUiTE_R*i>Dim~1<;N`(;wz5rlvHrbMs&u&033st%q4ApSKztC*ze`s-D|ej#tS;L
z4Vw{<W-g$Tvt9073!98jbNxTg$j@0V{qxj()Zqx_zj|{&e^GQ7@Pl)s83g|YK>LrI
zx-kJBqyD1ZZ^m2yE%khPJpxpxp4z%v&p|dM;sF_S{f|@>K=(Y|Ww+e%k&&>7lRb${
zY9MYt=3r<FB|on=H3`FT|7ZohKpa=y&?ztZx^cuTQ+<tn9fM!$nWO;D6@;I8`;S?!
zqK9;1V&J1uVjLr*vlEzU^B4qIEXe<k;p_7&+#%RT{doodp^U$Nd_fnQsQ*|74!cTd
zl8r-)$87dG>1z%lFn2+iy7T#Y?BgDQr2D-{{4e$~3<Pno(*fKhA1fVpc7}Q^#<#!R
z84!R~B+-Q`0vGAYzNh_<9s1>C=oTuF;7SSE<q5wlVPMeZx=re-l~Xai^#7NdLFtTp
z@Q})_PG1E93YB87qU0{Wn^&3zaMie=h+tV==~M{!Ml)&o<@P@v&9BN0q=XI3o0JVa
z(07vvcUz~uB$HN>Cb_Bc@p|42(Bhub{t`Tg1GwrwDZ9kzF6XBcAlXy+)20LYmXuF-
zi*=;x&^0Ba@?mtI08IxI%dODk;}<Vkq5GHMNdeaUEDG0jU*4q9Nd9`0m&{PVHluh#
zAj|cgPT_Z3`M=*iB!gj{iWAAh0gNLU$HxNXNkkm#Vy5#}Ht7SMen5Z*Dg9gg<{oB1
zLR>^l|D5^oy8@8Ah~JkBP`K=AtatH8)-xs0@4>?>FnN-1InQtmda%3v1>E-hw7asS
zmH^QQ9AB92-}(yv_q2k{NnRJki_K+$7WIN88lZ?22GV12c>ROK{x_E9Z(9X9j|~~R
zaNPb|#yJRB#_OpW-d+DYy#L2fAO=85z4b4?`<>|cub}FG&;RpJbgAdvDp?Cj--}-V
z$Cb}tE7bp7S4`!?82@x^zTUUx*10%76GY<uNgTi4*YH2<k*}}7c^^2!K{1yE|J#73
z|DR=Vu>|Uel}Aczh>J620-yd>DS7FUfB_1J*RL5&&2fO3JW`haI%_9uPft$ugz@1;
z`cSga#~imX4gdxu4Z6lgAHtRozBfwoSvQaSrH85*>0N_g&l*Zevh|h97CX}_yC~9i
zR}maWU>a`2k~*#?vg4W%w*)v#Jyh#xC{9qxGy+zgt`cM4!*4mER<fG0?)qDt=AXCo
zCrja_cll!M<A9i9%<KXyy66<LDj}%PYew*60MVs!H~B*^#2NdMKM)<sIcpZq#3lvS
zupBA9n)Of{Xr~?Ey(|yMB1%SKQ7t|E!crOhYG#sWCq<mBKwpi;s<J{dBX|L^Cqo-*
zEq8tnye^@0vdR{Fui8S1P>*Zf7(=uE4Sk-bopksx4?920DYkToFaRe#iEdoDh2Xv(
z4YP11e|0&0WQ1kX`ZJ~mH@9?g0659#^j%bh?3ue!^<Zqjw%gJG&*m26k`}>-h%2ZR
z_Q_tPhg)6R_h${Pn7aMk;n-iFrsHzU2YL;NSJ-SdAUW-fp{u{Ii-R1o!TkY&syw?W
zHM>>0C4<NB29pAoDg4D;bG%r6d=6FUj2M#cdb2)TTI^Kv#`l98>0sjDS}QdtjVYbN
z6{|7iV%S$15lij45UpT56v=+<AXcQgZ=ywh-jNSeKq19mV0hf5t9a4ANdcSsU!=Vj
zha2L%MMCx-m}O!7{;J?}PdsD+!IHN-wQYZh+`RO>?EL+C`KU{9f!>Pcg(3%i+0~P>
z`!AVYC<NcE0Js4`6G5kMULqA4M<I#fhxKt+AQk-=<<2h(zzdHU)<HscJ14!OLJCN~
z8V-hf@ygxBbc|rTc^MW&dVWDcU%@EBOc7OXyT<WR{hO+#ogxmn!z<TN;<@65;w8RD
zi$GBcU30ZFDqP9ahdmR>e*R6#{jgS`3~Xc(ouYLk&=t#>=enI4jw<M+80ynyszjno
z`f4H|p=ag5(P7p8(hp9h33fnf9P=awVg6q1I3JuJp~~|@206btljFU4m?oR#Jn!hN
zc5VI>%~1Q)U2#{aj$hZ!uWVj)d+3|;y8#!Ly!UOVfHK9u^Xaq@qv%SfWPBWRGy&BX
z$uE$eATvC<q6EtpbZ$#VMeHAhJB%(<N0|h8cHNZoF>Eu35dh~cX%GEkK2l_9tcD1M
zw3@Z+<oEJOmx=!D8Rhcx_WOYVUpZZ7)9QTvqRwL6Gn+Yn>gkijKKSD}`w}Ti$hEpS
z<L}N`QW<&*-~=pLj%1ddV^A$@R{teEEpP({)bw_C0F3m#t{@?GOf5Bvc$IjA_@&jz
z;n$=G)5FN4^hygGx4DbRxW$o&U!Tzy^xvp_ylP0nU@)Ouu9jgD^~^lh5mabItP@o*
zQ?N|1S#V&KwcE<#-5fXiGy5`HokTqp;KtftPl)Fk#!F{m_q`()u7>|3x9-LO8ze5;
zV`IC$Ts$>e68VD*K-79Vfy8>QAF{MMVzHF~yn&E~wS_SbM;Szbefzu|Et~m+BB;kS
z#@_-MJ^1g2uZ|2As8<K40t$z0`pIPrjJOJiIjF^(F4`eG?nC#5W0KwuQj|GfZR3Ei
zT;aMz0HKnZ<<N5y6w)Dnv)|EOor_k{T{=SlLH8nISv7g}D8Yq}Jgx!VXEU2pjLa!(
zQL82Or4XjN5#zZk81uRE2FAsyr%Bv#9R@U$>s&y>?|>Bm79tY6gu85zX1<WdDY8>%
zzH;gOvhlS=7M*eq@I5&M7gehx@T>@v(9?6>=8thPUv_zFK6NI2YipU4_goGBDm+LP
z6q=|#CXa$KVQRb4k+5Jj7l*--+;G$m3&!NE^Ta6N6ocr&s9-ddV)0fcKn!i4LCIPE
zY`a0~7~HO^dNf`nx;yQ=OMpkQOP=<PUHpe0_`Opkaytp&o+x~oMk3M{xA6vm-)yKz
zV*QHq_~m6>bSE|+(z)nIqAfU_1XW!F$nXSM?Y}}4e)Hx(twz!SI=7Hdg{D~~PLb+%
zj2r1FFiVUjh6Vb64p=oz1-ZM!Rvf?aHN1<Q8Nal4i`wJ;_-&=qn#?HkuiH3N4g@g*
zr<jlT|K+6tT^J|kq0FYHI%N5VnhVgq0i1b(0)2@&CZug6Oc<Nt&CdX(yyts5$rp);
zr4b~FEJehm6=e_z!jBr}g1=uj)a%>{n@EG*{QW+I=tfa}L#U!8d_yuumZ!|-<V;QP
zwGAv_yp@%#IefEr51@-W=wK9MIXRS~eJAv0Wz^|`VuIie<^ugGfUqF(EXt~dS-T9e
zLhH6>X4@E)y?OvNVYQA`W6d^EM&1_}rM0Q#5GmiNaM}yMBjLSSGZ4|{M4KPc7fWG;
z^sIWfR(|r}*6f8h;wPIkqh(Rhy~EEZs+%MI(oR5K;`_-=;=|Z4=pO5x)BE;PGXR*V
z_$K>Kw~p<DfNOuZ*CNd8<F)P@G|tNl@2oa!#VYliqJr-oK#jL>C&k<jw2KZQrMtsU
zHG2k|YiEb#5J97q(X$z`)12JV^`g<!0$O*K48>rY0kpTWn|K)G`?~~&Tag1Zu42a(
zBWG{e{f*0$w&Ut%;0L_rch??-(uPi=ZOzDee9@m%V3s||AP|1UEOu&VaettCWv+jM
z#!dUO&Ras-i0`MAt`pdeH1c<I-`-3%sxE*au~^|`F%h%9r9RK>QAW<msLZF$8js3b
zdo?xkW6owveBwQQr?maNE$Z&h_l%V~qG(SWlI(c^+SkAP-_<PZ3*Bz2IJiRY(FrCg
zOSK19iBYEsEF|8Tc-F42Kz`GC;7F8B|E$f#s5J$H(`JgF59rC%5#C>6kG*j6@ya1e
z8OB?M0`p>^EBDs8VZl{_r{(i3K=;QwoUhcF+SKyOK|XfaepfT5<AU13J<+k}n@bWm
zVUCL{B8O2Q6E4PdX4VLB&$^yHQScr)JBoQ%Fpc_S3+eO#_xt&k-EZsAa>0v;<Vton
zU!qnt=ZnW-O0>|S?lYGZ#YGBj!-imnBm8&P^e(lJVo${QA_teI3%*ztar<Z)w%3IA
z@|4J$7Vh0B1h2T(Ix6o{;Wz6qu6oSro*4D<6fP~h?`uaCxx0pIXEIYdu`xxm$5<h@
zjk9^SvS~>pNJ|}LWyE^5YVQ#vA#-2gQsX5y;{&%>G<aA1-Dt;-pqv$JtDH`81gFVr
zr&~(?wGa~hfoi|M4kDl}Tpm2`d(&>zRKm2mzc_E$+*Q4+Xa$|W1ObSvT?b@nv{_v{
zil_nhrX^BrpSzn#qHgHC&~O$&%YeqhtuZxC>t4_SeF{DPQtSSudYdTmFzk41{!H)m
zP}l0MNrJXX)J6E|@kX~&t4Kj*AAl<6ug&8Hm=kI}mc-{jyg=5}a*C6RYzY;+hEt8-
z;;uDet0;o04w}0Y^mG$er8A98xxbziSv4nM=?<^D<gQK~_R02!cW2nO?8uyq^;`6+
zT#Mi31@si<H0OFDuFSVBL8^G^GE}~bv}y3Z7d5ayrIcfIwjtZ&-o=A*DREycZ=7>O
zt?JRhb)~A_#{{PDJ44(m+gt_HS1O}SJ$+i{M6Nl%Fep*TL^(0`&saM?T2pOBUMr+^
zaht$5vfX`ZS4sPcJ0{6%pRh)FMoOJHrh^7fpz274)9w}i1ae1I-#94nbpFk*44P_2
z><MJ9JQ9uZdeOA?t(sQHJs#`HiB_Zn)-JAnw8Bj9f#0J($)mgzxF^<1AZB=0vy!NY
ztg+JKeyskA<Kkv(Ta59+GC__}4rXCAK{@wxgHE&tK|3Y~=&huBk`2J>nBGTYud2VQ
zYwB(Glj)OHorv+VlLDrK2rpx5u;@cuXzYfEOHC3fAKMJ9@U{BjMlfM$@UJ}Ax2isH
zfdKRlRcEPZN{fq&E7?{ad(E~dJ&ot-d`Z!DxJOCzDo}|mT85e#B17xP+DBKgJoe=T
zT#nO5^?@>>X5q2c9u=j?jdk_1sNr#=`2O01oR6AxG@~#&{llS~nfGIX&Q8-0@)V7!
ze!Lu1-%XNNF7ZU|vm)DJBtZRm&RdD$sCx-DTHxBjnQDP^={+```!x&}MZGc4_SR#N
zSX$8SmMKl26wO<L+BBS|JUqyW;UOPR2l<*<$kX;3hJzDaxZ^a!1#Z~t$L^%7zD#r2
zM~2znTL>@I<2HaEOq1-!?Xqe*cT=aQh?8o$8}?hBtV%p_ZrCJr-ol3$9&Z;49(1ie
zlSBiJxu|Rp-dLiX*%DJ^txM@Q9Dm-e&Nh69j1+yQ`;1pI%uRbv$EbMZWMPXpOVZ+O
zKjIo8Hww_-FbqD*u4jmGyFraY8WXugtzUI3lf49Uh&cKb2yj7M55d-reOx#zJ%YF<
zCA7CvQHaOr0_-tNwnr|S{zMAu^%0d~ef<koE!$en8qw}deA(zNkG;2hjgJ8Cx<qR<
zT1*1PD*Mm9OhmIX<0mo4E6pkO!~`$45-oQ9E9{K@jp6$%4I%fHpCnU>+1!5ODb;v@
zyO}e7(mGaFQt}zo7+rRfjEg{|aLaKk9WN6}hO<j0m*CTrJrFSv;*t4YAu`ZE_2uRy
z-$!?wnQn&AX-Fr0yg;S88D5-ME0pQNy7QNpQUKb_<Q>y8{~?7TxL{D7&my?)^P&P=
z{<5#4)C{G~_@xR9o5GHm!9$Np{hESWJZe}B_0vr3uKl?JgnQ~|Ay0i_O%tb`*jq8H
zTj5q~^CF|GyLQ?XvmK436>B?2wtesVu{zodZij7#fu;Qo@xdIPm^1*B43&Qya2o<S
zp!WO0s}*tdDI~|}^X$jqIDhvCo8dBMvv9IhM@EHR9M%}7y4L<lDmnXc!)8oKZ>s+B
z(&=V4R*re7k8#lRiD`c^j^mQ-RB>Xh)9D*%BUT4hrbhm`qy1n}VZ4^5$d8zjl4g!8
z-4dCfEUg<U^V#3lR%gm%?kTPIl~!6usMl$(PS@8e37?HYHxP=}KA0wHH_)VC94-H3
z&0(ruDw*AFa0AAX_U_mq5`fGLFPu@qnt)o=MdyBiwnqW{res_$A;7;AK9w?PM8mH8
z{w+{$48VLSe^pG{eZ5XyA*0q4C^7&MLCs!`gfUP8<QxAW9}qx@HEgC9=zF?{1r*1R
zrAhTgl%(+(T@!H!B`DAK^WA%Db9eSS&Rm2KHSa2obR9f>S+*iG#`}4+Pd+a3NV&AK
z`>#>1Fm=>CC6mMVSR-WL9nai4b_JNk3+aH9<_2T(&_OF-c^c{<><=(HEJm<6<+J`?
z%$0Y|@)zz(sTRBKp9w0F`a7~M^fKvQGRV^8+9{qxBfvL&zOSz>b`g&GY)O0Obx1<^
zW{>be{5^vAUby}NM06X6^j;UkJNPtqX}iNip9r`gP~M8UdAIm}+#R`Kq>}T!I?P~x
z2t%BUi<n5i-h_0DXdW~}mc}Dnmb+ZC*hbhgQoX`Lc?m1PrBi~zwVAZ|k#8hH%&bye
z5EZnbp1$^IUV}hnEWT*Tdg<upM8K--?s4hZ1P*?e#{O8bh=<DRPvJX=5u;{vXU$QA
zOHAw*fc>$LA_bUA3&jatCX*YQN0tiWL>fl>-?X(oJ7_sOsj6Nbt97qD?h_iL2j(IA
zWp7pb(uM=Yp~B__{jC(&)xH5oDB^6RE+oyPwx1nS`0#VF<4~B=>F5!Waiul8VQM>S
zjI(mFq)U5OdppD>WMTK%)Mqo>siaHk*m=JF$_Bv7bk?qY-{_r5wz4jR>Jhi4#uU3*
z$w0rIHf4Hp_DT2s;mQQNAI_7fI*|V;z(joeyq#P(9^<>ITFPpBtn*nkcOd_{%9GQ%
zA(z;~x+4^c_oE_KsLVCbopm3bJWffw1x=HT*TkHT6Nt*)=0d@ZZw-dkvOM<2v&Vj`
zP}%tqhfBNGc^fgmWZ~#5KI@fMPtsS~S+oM;mKEo6puZhsaYM7|P1(s8j3Kk02={pd
zG=n0u=ZJ$E&AG{}*5@`&hF@f7B@()8Gi$3#ithrx(M+Pd?LE@i*?VzxSMV42H`&>S
z5a78tdn3wo)kbXys63#-K=;|Zl~OS67YJFQ=J!nIGKOCuWW22~BN;E5-?JwWA?5W4
zb?F5SPLNltzkmZ%FCwlGl@PwG>W%pr`DRXnt@%#PisaFuUSDA*A^n$lQHDi{+s#wn
z%nvIfp+6SwJo*c(3t9-bNP4xquv~$|=l!_`&3DxD87Yp{!R7$1b-_V5?O7QB$kI98
zpcQ4oILM=wTLVyQ{g%&%Ts31?ED%m*0vTAMI%f&5ngs?$P)<fyPebrI#JS_mOg^<F
z$zLU>41!lM6AkpRE4)CmYk3s(vnL2|-A_H<o5yO$+Y)kmKbxmtPIoj_Qx*1%viC2j
zJ(XAQnc3%xF!k@Qr)-WV+g4JeW{m`GZ!Lmla`K&6eqLE@xQx@uS<Ttx5!{F^jyHZl
zUqpp{?@aWd=z3X<IsTc8l}Dya&j4yM`fK<0nP!iCp6M(nG?tF2*cmb<Nfm$hiG=<g
zm1)YuX()1&rkIhG-VHfMZ24A|w!p8pcXc3|g<Naws4viR-47U3wck;B*nWd@vGXCd
z!Y(XcE<gq615%$Nlzp5JBzm}Vzw(h^216g{5^Ex7{^4aYu;{D9Lo9^*MTZGVVC-=s
z`!Ljl10wR8AVB(hl_LlJo;z!d``$(Y{@XQC9kYRO2Tlmm_de}$psjN|eje?klNq>R
z<+49!As9Yzla}A4%p`{8y4b-sU1grftDrtKg1&&R0u}w@c$P%%TUOQCj~HSI>Z_!p
zvYppMN>eoNp3b_R%;GYH8dff(MxGt@+d3mU^VXWx_k3rT&a`g`b%b$i*)QXEZj$k_
z-3sG#Jv38_5sUO#k9+k7u0B3Kj&9D4*gLqxh~{6hOS9FP6;hjC)mgQ$Lou^aC&9C+
zO=Osc!luPNC~+1%0{GmTf|)#_%?VY<GVk5C^<sl9UNZM21O2ssg*PIs)?eDvIgPhv
z_jq#>b(tBJ;%)Zg-!O$U^2gfFAhG%Cp01sGuWp#CFhCK|jn<HA{qqukzR1za24kYT
zI0a?wOkm=Xq<pAUht%?RMP1ZshDS#aUG-?P$5Gvdty4c~C*(YK;<28pm6O-eU%AP9
z7|_#K45s06Km4!Ni^ww05*;U_B?*HQ(4=20G>M>4<zTL+jJB4GoIdU*%l~8VF9V`n
zyZ(Pz5D+940TB>H1d&oYq+1%44(aZa7(yfj=^i?yq+{q7>23xXKw{`_2L30y_rBcM
zeeIw7dH=j)6waABYOUkD*7~fvcE?&Z>$ic7DQU2mwq4=F;q2WpoGTIa4f=0BVi&8H
z6)5DHfepPn<BEYxilsQT(N=fEwQknQ!x9;(e7gRn?a}RQtI|)m4uiu@VzBJVX?Xm#
zGvayp4&NT!B<{Vv>m!E~5s$BU)t_ppLeRB2P8h^kmagyGLQkM*SsKEr#=&<OivS`5
z%KX`l!PIEY7&hqQ>0_?4p<L~brqPZ?qCWhI>#T1Vzckhvkim|xCTlS{bGr_`zxO*0
zJ;7ECe2V@)O3!;{#`eY4xB8=sdMC}pshMD3Bz9HwXKx4JJFIM#l3g9U823-IRPbm#
z=Q(#f?dkh?T(`mtE9=?lap~I!3m`LS>`Tp9Cc4U!fKOXA3PI1ham*LAAX5fg0~SwH
zq&q8HV>p{1s?*T)k00HR6kV!w<M8J_-j5;du0FOZ9g~GVc)K^8D0Hr>&#bL!fh=6t
zhVg0OEiRc+n}&l10x`l9ScgQZF7C-z<ZclJ-;S(=zcAlv|9l0tICeaGM9@sGUPKK#
zfapM1y8wvW(+|G{7zE1ENp0aK#4mFip-G1%?0#6q6D1l%m3>SsS7DFbu%3Gs#dhNI
zUoJ3i57vL-DZnpKuLx^@Sfgdtc&>YTmO$$?!vT|v*4Tb0Z`FAEtg(jM{`B)o*PJ&2
z2qW(;p{Dab@7{5QVVZJDJ`Rf0b~M#I^DEbc@qQ-_X~I|5gSx_O8d-#7^#`@E3;y=i
zJ>q?k;2AerULrHM*3pK>E~rCg7<1$Jk-BrlX#-WefpLDmgK6yCas9`m$Hghpmi^p+
z-l$Wbd9UfH<>ZNu{pm^uqaLJ}%sLjek-SZ7^)p$0?NcJ=$V1j5#57{QI?XPa*n{Nc
zz2a{)IFR-#aj&Er-E)6mt<U0fe+HwPG2_6w=U@qpAHgDHV&vDnQdqG$ZR2-!lE2O#
z+ZA}&@O^~Ip(WbDo~Zj4HKs_;0@fWnm}u|xBOxP%78TL-cK6-EK@|R%ZZeo0J|a2n
zdFoCxt1>{80xeYY*ugg#$KOucg)ky-3cW(xhc1LP{#iDdDUv&Z>4Q-Z-Q8Z!^sIjO
z@=-=Px!_35ke;YA+`cf|$o<7|XM2l*-ps3t5*g}udOQv@i}4LiW#u~U@%eWPz~7)0
zAw1fBrSTN3!|l*3Q`%f3&b!vTN$t#$OO(9^nFG?~yIV6-V31Jda4I@>@@XZpv2#bC
zWa$LT4PNuJTyGk%VFncEqus>LoB%O)AzYu?Qq$26f4@y59ide>A5YIy9)y->67~&`
z<swjN%aa9)UC7ylvSLCQY(?GbXGH08^u55zt+%1Hs55ciU@^y%JP9;@P+dBA+j9Tn
zG2Lb8r~|phFEV@+gm4E<&YgLeuKMkUDYNIL5Xb!tscf8xWbJTJ|EiAJtH6tuZ*){_
zY@*ce?468TQ=cQtrnakP)?<b$q{JHyy;(Xt=lPk?#ta+Ikdu0x(PRfL-hU)0>(e{G
zIPE`8$F&~pcP)aA3^6`qCnjL>EO(8kIbc~NpixFyU%#w}Ny3fwK_8y&F6(bv>L-AE
z0tp()AHeyKwq)3Y^l~g!|H^nB5dj1na8Ah|Jp4ndonca)A3>*qFAJ1qXc#7|letFL
zGcL<gsPz<u-k8Q<UGd9Po5P01fJGb3pvSpy6liSdReF0~m<a^B@0)wWcSwq6;pd9q
zCWUUIHQ;va%;noBUumzde{Gam_LFAG^d}_yGOEG&f*3b3UD#~eRd!T3h4b)GjpgmM
zv#ye}39eRF>zipukHe|K%{tLgL{{0esrxljrPC(MMK5yY52q1^e#Dan;t>ctx<M?m
z1p~N*L_j1K%Qk~+gJTsg!x`%6Ve(r-fD%|6$hz-?RxX{S_4GUrbSvhIoGB&E7SY8L
zTq?z?qM7)N=WLp-x7UWU0oHj*J+aU%1@0zxwkV%`%zE<ZH2|7^F-Sq=I5O0V<+7hc
zY-ZR)(D0TN`?(!5{XuszN8_F$+aPSFmMQ1HhbZ)+1@a4=Mx**emo_Xx`|t4$sYz2u
zF9y=N_vM0N&l-E^qIi33Xyvq)NQAPup-O*VA^>H=+mazHvh0l=<cyCW-hdh8p2C0g
zJZ<u+Uz{`%zs!{kyX7pUXCIgbhIrmv07)S>qtg5s%F#Tja+wkZc0hh0V!Jg-<DbvH
zRWD~SmWcXjoC`CyKy^uaPZJjzT(Y~o*PnG(y*u9xK1WqEx#1ZalGd1DGFiS*W}05j
zC3JJWm4NKke!Vp)<+!|W7NN8lQ%1*!zg|io`RH-uz?B$|)<e}Zr+rZ2NWzW8(g}mb
zk^XFrn^q{cWBY%lB#ykMIM5Mdvkcz+Q_59Fj_hK=*{F33?3JElU3Oo6UhFJ7;9JHR
zVO(FP!$Lo8&CJ8uswPsZ<+yfn@_{Kx$Pr@u{zjsu)r)2vf9(qGWqMk|-sJgufg&U{
z@8-LGPTu|8Sbi2*n$N8Ku#AIe6_g@N-I`qFb+AcuWVl*T_)uczsXJO|!EGWp=_zUL
zd+{F1RxeHj)_NuyK6a9cabj4oG>YVx9t)61A{6&0&RPofZGed5$~pp=M5n{PEEe2X
zmGQi+Qm=Aqla64I+pr65Am)b1e5|zrXzp-%PTBafs)Zme#0)wXa94eI&R?*B_k5OV
z&kQPVe8mQ$u1e3c`{fKM=0z!NJZuPXcKfQUZ{g)@wR5Th>&o2I<#)JD9n3hOwUB&H
zdu(-fY^XcF9!T7HX-XcgtY=t%8@|GcxIi>uc*GeY@7?U?^3RewSJoX;fA`E>45K&j
zeqsKU`~tvb5kPn{6{6?hR?1W@$?1%?)ME&)n}l^>_bVGDVpWoL|0zXygMlO#1$;Kh
z&t$pOWE7zJ%JvKeU9afMs!<hs?|qxE0dqk+;Z94rbn91EXVVwV2{WmAd~DqTkyQ2^
zQg2PoN@b@_nFUS!g3o6=Vs7@dy)!-u^p|$}U^9=$UDvU~ztoMg|D@}}u;GSRp;27$
zen`i2%gn%PwPywxTY?>Cf|xqb`f=v-2S97F!r|_nix&7Nb@t<Pa~NcIztV~lyuj~E
zFYceE0Q2bVVq@d0s`&>GE?`N1*D4}rWTg|&VgB+GK0j=FnjX)L+`aIIfb0!XePoBm
znG+_TtX7dA;e)pdnu>#2!yJeiT#lC^XqBA8KAh*dz$xuywny5zLk?aNlA>}nAG-C&
zAp8gir8c05Va>GS+>A&_pxF*ifp2A96@orXt%Pj3SABF`=jiy}Uq^Yq$*bwN#Wmp0
zcyLvRXT=j^t>GPMg<;z+#Os&_#Ej-U=`;e+3`WY~#$wqok1my6E+Yq812`Su&3YiW
zJ!n?aUg*)fceX>fOy`|yXa#Bn3$n~ZF!SILqT+2E0<{pJRiBiAPD0O}bM1_Sfy0VC
z+dDSpZ!mX=EJnpfetM*D6ENUm1g`!aO5|1qy)CJ&taV4eOmDszVeM$!z>JdezJ**K
z%ZAjkR8zJYSILmql*eyWXzUc<Q`d<#XAcOT{LE}B=1ux?gy?m;XU7_>)R%hwD~wLc
zz?=9?Ke^hA-%>e^o}T={*U6NS^E@>aBPK@nP*UR((3ZuwNL~-vj`QHj8=Mzq9UJR|
z#~08MGZo$jQSIt9!{DnqSqw5~y!2adm=sB};bC;q1ih?xRO@tPUB(Mw3z%ItF>gkX
zhANto2<JmsMs7v>F26K49j;re%}C@M6(vuCtgWA}?!EBAD9&Nho%&CnRZd7#@uQoc
zEumvu)G!bJUL2$a#vN&(;cNFD|4%U-u+xFE*NYvkV&}d73B<>jD5o1i)*IieGH}AX
znRv&G);2kGo@9t^tya?87;T8Ebl0H<ArrAQ2BRFXG~$@9-a{l760^|Ji7wk;6d7?!
zSjX%06`4;)C_PK)S@9%ef8G6HUrVAkrlH@<aB27Wiu=(pj*A)Hhm-kw@9$x0wA51i
zHl7eLi4fz3bz4~4Be3p|UP3pJDVKZ2VIEzH9FN#uSUX`XIq$sTh{GJ3xCg{`+573l
zfAp#u5I}v{SXW(l=2H0b#3|&{6`xMSdBFhM1NeL*H+TJP=o&%U_O?e+6-DEV?yL8%
znPdZ)2WgCfMNm!3CwEwLTO~dvMA7i~(mLFvA2DbNJuMVB1jQTsbqt8-4At7}1>dB9
zB?DvfbO^Fk4&BXgQ}2fAtOVjXrniDkp7)sQuU)<L<!0P2Wuy-HN*j8?;;?)d`?Vyy
znh5jI!Fq8=p!5AGg`0|tRUn~Td7PHp(-lshW=HTj9vnpd$8w8xKCvmLwZCpp=oEMb
z$TVc0AFv*dz&5oN6vCxXEFyTwIc+-x&(Tm-0bP!@Sbe}<wolpR5YRA|Zjo0N@oY*{
z#o?-RN1;D)MLpOra<1x!eq(p!;TfI2F^{5ik1%?&2?IN3;6?yPV^3t~BdZ%v*;GpU
z>hv={HGYW(+>A8$OZ$dtH(i!WnIz7y)9(gLb7^FJVd_`$Tbsh3IHP|M2=H>ky<Bi<
zX|EJ{W>M5%8&u{fm|NVtiJW#d72tjI`yU<5OmdK2y*=J;8&sd9^;$g_EbBQ*O(R_U
zesX|`XB>cSeg1Lu!ec!>gJmh-$LA*Yx7jjDBgUOZ^R`-YV|5l66Se9z$_M(3R-No8
zA`EAaa#0Y5yBl%E+*iv$6Cs&azV%|_@t3QZA?O;pfdpW+ubn40fY+R<74)upCrA3!
zLERhhj0Gvc-WgwrA$o2xGp?^QTPNDE2GhhG6<FnK#r;JRUSzr$C1Pz8xvR))ZM2)V
za^WB@O5L?2J9LE{R3DJ#QT1YFM<m}h_V<~L0Xdv?vHTLDFL_#QHG_8Qx_QquB7PoC
z7w8r}yDcw{LqczA`Ucndbp8X=^rq8#W#q<mb~ewp^vjcuvc1a_xunUkeuE<P&QdMH
zwG{_!kAVntzF6KPj}TzDp16A)Myl-%IWlaV)iRnqyQNj#Y?NP&`I6rOz5C{6g|^*Z
zm%)`_TA7w>9?Q>c=O!oor=vVO(t`p)CsUuSD`rBx$i{tv__*U=GQdcQfR4SwMP4NH
zlAm#zMK_pv|ID~_am*h4%F&06a&@ZnR%W+uYm4WjG1d$##sTpmq5_sttjP4Xgpaz#
zl3Zsq`j#7_4qDmXINP2HG+fo6sZCl{bYlJKz`E(|qAYqZ&r7UH#|1$Vi#he~lOzZ3
zFKr}QxR#V*_bX!gYt+6WvZLz<bP^DwM!*Hswrd~XPINdIjgTx-x$)~7jHMGHL|-hM
zbE+k?<A>A9JctsV=q&PaLq54?H1|tuBxmc_3)mpBVgf9wVTKM_bjy`Z<KcO%Cp!+p
z({_T5WP`}X9}I(CU9>u{j6gOr?5AJ;DufT9?qS|G*j+Fzb#HriEMHi=q|E~wvL#Ss
zd<fcm;#@E5e)YYd8HP1ESUVo#(Ou8t>npl9l|Wt<B+EB1;XK7tGKS&+_qmHee$e5W
zLNol8x5)68Vg7}^LU^w<m%)lbg5d6{xkg6`#|OT~2by;Z;Wn9@8orI;^V6*?(rZqz
zu@u?v*tRb|WfW#_`F{32^sc!J(6)O+GKKx0TS#c-vL`w{YXDhug0ixIDmP#2nl$mT
zZ1(pX=5F^!q!P6pO1&3dZh_!D9!i%KZE9WKV3w53>o_PL{G_EmlkR+(o&DjeG`(i?
zO=W{euWgqn`xKMDvh~vv=K>QFx5MN9uv7#VJ$PoWCL5&ey}o9q-5zv^gKanNa=H5I
zJWf}fRo_2La+JxA74Li9W1vIz@u-!!&x0efILb()dlUKOlFLjD8OO4bAqDsbh!{ne
z33sl;kHl7@-Gw~LC@(}xGw_e4B11ho@Oz!?mdO%ChMRHV(FT0dUpsL<?{x%IKb2^_
zZ$3k>UC3J@NhRY~kg;+ea3^<f9``go*J%44mxe6Qb1m*l$uRya#)zwiXnlYZ@6FO<
zmp!6FehiPKzxMT4_X9<IEPkb6u!?wZY>*Dlb0_W%Wc0_Ek3T7OIDKWTFtO4JlZih5
zCWoc&xt@gExi2@%L`!>_GM^Zz<Blk>nA+%mIsuWP8_Xw43oJCB%G~R}<H?{-YNixp
zDU5`$IGvJVc*50rOsr+?b@J|2!vSr-*y?6ig`8n05pFpI32#Ok(q3t8Nt-<bk#sQW
zln8=osd24@Ih@D8UWX4Pl@X0^p#^mvwf&(uXd>+^X9@em@r+T_s-6Yw0WH?VY{6U-
z*LsAvA64okQeVAuBsM#Il2$>xKF!ztNXZ+!YG9rxq{hm}-Nu;iI83&_H<qhe3l9~)
z?TMixse5Zc@EZ0gFX9z-oBNB5)^{d@Or{6#PiGNdWU=!cw;Mb7=bG8??g(!_1ToRP
zH8m&;t*afyK}xS+3zDG2e~G7BC+rKCr##<CrmUBlYdd=8oC1^XtFg~8Sq>qRCd356
z$#|DqQKzg{ZK(ZUM^sb2h8@N|tjny{JZtP_SN+~{WrdTMbza?{7+E#<E-tm^4MWv#
zepaa-=p(k-yM6=uLGibPUbhjA3sGlV--nkk_#C4>*|Cjb(Y)qw{j9z$y|niy_@xUd
zaAC87zc8-$#_4Og(k|_Z$8rT|r14k}s=i>6b4$c9=sc-}?H6l^`#egtz@Pf2)L%~>
ztL*@(osA#hd8y-q;5Am3<y|$ge?;lvb=Am7OqU^s<GAhz<~+Zk`=Bc%)M=4mh>#Au
zEK}i#%@l#%YGAgv%`KhhWZR#M{9YMo#5ppQp`U(j2}wPN(T<k`Er;!oeu`fL2y{FH
z1^&XpX|pzt!NF;%v!7RefyZ}F4zk^=I}L6O1kHi1_H*e#I_jkCFGsQ3@R*2ri`5Rg
zKP?+2DIG9+Mt<yJ@Ld+e?q!tMdBRh<21z-2`vsoCy+nv#RXpjwJM;~OsoXMLz93B=
zY^)4=!vDBnjXUkE)<(3X{-)SC_I51&t=Lc|5>*bHoUkt`6ANf9oVl+YNw%x^#Mu2w
zO8YSacObgZ+y$F*emSklmJn~ArG6ov%s_n2a8r8L)^DBX*+I^YyLZGyK25NxUAoz_
z5OdJV8TByjor9DflSLS`<D^~nU{&H~lY6f_p0>ua5s^_EER47}-ni4dmEJy&HNL=0
zQ?Z&+Nw!^%_9oF2W|YvTH!x8#`(_k8CVhW}T+&OlVj$<Q`zwV!4Kl)M^)U>wqtpRm
zeVZ<~{cvw3-^B{=TfMU;P3^|C1|g#UHDlEQEFwYU&d$XSqH;>WKf&6w339y5L*9k6
zIn{yD{&0Ex5?9G{w>!jkCk})dQ*jpaCrPs)_vY=DCnq9}o<HW=8zU+c@gK?0hvP+q
zPj|g|-8mG>*;1DH8nP=$_5Cm><K7_X69v!n7WQrBL4-fGart%toi*~pB#MLREMfcU
zHtWn!coYR8s)H&9%~M_f7>vE?ie3aV?a>77q;|3O<ZGpnH|U>g3qE5s<W!xn#8&&-
zT_T#*P@h2p++USF>`_xRV0o2E-QGFG*0vRfmDKYbdp;2DrioZ!Q-nrmT(%bUNdRA-
z25a0R7Rzf<5J7#d(OH@0V>9i>=eOUe3}^bO^A$TQb}Y>%Ar2hiAo50QFg&9)UGRxY
z%7Ro;f+4cL!wCWQtAV*B`6Wtw@o+v$+H)+4J0SSXE2;4~hbPMp>)Rr4DFTL-bG|!|
zt?rkNm+$nQ$*gZD8{t&Xf1;+2l*nBqijiD_V5E1`(T#5E8(MO$vvw@d+8>V&lfPl|
zlBgVSy3N4W>7mQ9_(!v@9|o!7-hIy06zP;x|7&mNU~N##bz?%b#&MHgtJWc+2Mbp0
z2L7zqVjMZm+Z%rzt7+FJ<IzlTJbh`sI#itjl8$HB;yYg;(YQ)$PmVO|1$TMh+Ss(Z
z+FReQ8Kz0G$*d)J)+N9XilvWTSkRM{mW~n)iVfuWE>`tOGS4Jpl<nStfCmkycEY=z
zO;VN*21~8ZoV7#rL`Aa`icVW=wN&BB4^#?B0L|e*&6>571BM#6_>|q`arMgtze=1x
zpad4?Nbl((6(>hfawI}%e=_e+FJX@6d55<9%rjwUNr<p^W;_!XveYe>zFeh3{<4pY
z%dwJWSXV(yJYX*`z?rTfN=>R^18LXQs5XGL@ye@>Xa@vBn-)+!G#wnW5*>@TySc2B
zN{%fe;xX5btR9n5E|$M&KHu9?$&O8Uq04{8Rm$XnTp8?pnzd>>#9F^~buer{JNhZ3
zmLV`Jb+jpPEv9BG!zhK~rl?~-@_#ZPP|^kGPV{ji@k4oEE)*IVu5=KL&n~Oz_#;i4
zpWr!9M#m=6U=lHgHxe>Hx~=N>1tSmBV*!s;+v8k6qzB9kD{$Kgzg0UNF01J>vsr34
z3Qk>{k$D?nG?Qw<<4rE@^p&^QnGoSd)ft3uzcku!|E0p7?Au;OtwutA^sZg^zFMi1
z(qWkjJQ%wOHqE@~A=Qh0gZt&vKirwkCjwz@M-ec{dBM5$WToZRB*{-t`NiX6w<0Q+
z{g{}cVH62Y@7vL2iL*)MdF2*&nfyNxe-m1FwoZB5&bK)YS>wOi$Hm%;6Er{5FRK;N
zqAkzAy@N%}!pXKFy5pHLD*{=JlI)jl0O`0W;`7i;@W`RzY&Seh7^BDHW)(4K$?Wi@
z7Bziicv^E;lEq-&uQ12V?N53CkDOR+EeacnhP;q$(@5!S4S8>|tY|84T+6Y9)M@9^
zBf&5;yZh$x^izGN8uhV4D24b#_HSq^CZF>+EKC7gL}q8usL=XZ_}s|@D#pyB2Y<-C
zg^+un_mI^nCDo8V$J4u;d{VEBmQTqi-=mHCbY8O3Dk|gnRWybXdIvblh1yHVzY!~%
z*S2|2ZxAH^Q`MVDMt(xwrl==$Xnhx&3^Ia^;^&-oy*ixAVE^N?F(Q<z;xZ@50M4;f
z>V2Q)jUOJhh67DgV-PZg*c^>!_*|tteC)iB9Y2>})K_L?P&H_<8&D44A>c?FWgS}k
zqet$a$+{0ddl@Ceht0*^Hgaic?Y_~rzk+lclt{t&O8d2ig~8=kqlypfR;XPNo4e`6
z42kLSj7RSF&hZDm_f`ipzLZUDS8w;#KjCTTK+)+Y>_v>N_8^@HM@`qgbapu>o9b-H
z@cU>~oI27dU@IAqVYFj7LC;1@$-WZxs%q|beOh`>mGpuP!bbRm)eETkQIO5&4yx1v
z>3ixe6}$V1LX!o?5k(PWER(#A7oP>R<Kz)(q%&f1KNXwQLX@8KxCZ03^G8qEQ5Uuv
z&Xf1}Rm?oR4DU{eBoc>0J>z*uPHGL5->u`2at=6AF|ZwIRkLkB(fE-yjB@*(j_1vB
zUvN=g!?`+namfth+@Vga3zUS#)uBnb=f5+=2X;`E3$?9r(_3*0NG_XgP;^n()?VU1
zsAb>aQyttv&Ekhb=E+4ELf*D~m76K}E`6p9Iw7!Eqcu@;@omcUfX2t$KvBo+m%nJh
zCPRz?Uynl38zr(WJ#V~F0%`_7^oZNba6>;j<iN{%nNML;vRyI7zuZ*kR%JQbDKaPf
z*o5`n&H*}m`frf|KMWcQ<~!@>*xKC@egIoYRG820;EzUje+Iq&TmgTm$v<9F<e|Kh
zl0uZe6MwS;mo5NG)N%;ONq)IL{gSKlUy0=c`qyl&7m0=;cm7Y(_^)sA<9z_JEh>(l
zFlqv5j753WM=Sob-PnIU^6xK;c~QH*V&Rwa@-#K93j-0Avwu-U3HZ}ui~=7-^AX5+
zZ||5JNBsjP$~)&yKoB^2ApDFlC&-?RT!pA$<T!lDLk5Mjl1)LM1Q@i`dCaip^!b$5
zrpLgCmjB@L4+EJRg0ipxBI=PM_NfgK$tWUTndV-pVdu4o0E$E`FVbB!`Yc&f$bR^Y
zBPjyJcL)#tYm?NeoAh{^%QXJ?LjyqGA3mVCg|)RkDe){0K0r^hUr$`0dhTGdc@D~z
zA+aDyLQaIjU;eDl$gxI3L(jFIw}kNlmTfkX_rI6&&shZARf$*9<>KnEUr$SfRPB;w
z*;@g<El9liwA>q+DolMWGnBmSNs?B4wXcP!Lq?z-uPe>$7!O60o<>j^|Ns7|CalP`
zosvhY`<S;$dS7rrZaVfA*cY1HU-+Dzc<7i7qd~>8Z>+F$_hZ)oEX+!+0rI$Ct9&~9
zPm;pleiHbAoSWNu*LaQa9^prpD3K2diUmmMaIR7NFDt73j+WgyNVIp!vkMgTBnA3&
zlYdsg0859}9jK#taLKiZ{(qL=&$a#vb)m<1w7mQ^w6>tyMPW*+=8Y24O5z&Vn7tM>
zBK!f3_@@Iz!w7@{68aE{H`~?f&q68SBtS1VIavvN{6{jwe=T_Lr-U6nHpujQp9c(h
zQos6nC3hE8pB}wUwAUmOciKZTbsWU96bZkv!Wt-xru9e2AO;wPk-K5Ne})wP`<Y!R
zck%I`u(Uc4x=S66b>)>7B4Kc@DE1d5f3JDQ%R}9@sz0=z$2B$ChwRFDnE12O2smTd
zflV9`yx*blR|?4ANB15kt7XI^Ny(3juV2$Aup5ip%B`wXA-m!nYOVoetTDcnLSxX?
zrxm$Yqc#Qo_TxYPdbrc|!;|bg!~gN&<t8z-ySug#u{uDB^}UpA0J#CA#Q%7BL6KcI
zmrSM_n$G-qtTm|e%ZhCOqd%{JdQx7EeR}Kec7Ha7{I8P-?=v#@Cz0-DnHh>8M-p}^
zU0sE!qZljH+wkfk`(+drcy)evnjrFVcI&H`HnuG9>5+9GG}WO2|E_NmhGtXxXE&o!
zKr~JVzkU5z&eMM_7PDtziHTGN8tnG|m}-r-K^C1|;R_29VD?U@L;BPevHtL6I%8El
zUVm4{^;#@=G?kprOya!I8c7WfbX-^PAaS)i<B!1+*ubFrHv7o-pO`xW2e(beeAo-X
zQp!ilos(c14g3Ag$8DO$ax#*TW}=TDF{p=8kJ~Ln%T(3X&2DP5#!)t=%HMve1~?x>
znWmASET<TH-|wsiGc2Z6+Al`;+gGkOOsXBC$dX4gUcF4a^b5|_w$1?%mxIsb71fPq
zjRQB_WF#eXqoSiF)ykdNU!4;epAird(O_WgQYXBA??NT{p13CHeG>#Y+VwP^mR*<K
z{ZF$YAiy+|o&7nEBFMe7yStKTubycCW*$pz63T&{jJkTBfl=nzdKkqUvSjTlehZPj
zx?RI$_Y+V<6d2PkYLf}insJBh1@jwOK8oN+f{?4#)LZBx0QpB5AgxQJVa#QW$O^!a
z*q(nLF^GRws@8P-2^Xz`qM{;tc@7zahByUa;86Y<RIw&3P^QLe9Yfm-Hj?mYVA2p$
zuq}y-ijs)0#%q0Ya3+rn%c_4oHjR`J7DgetJUx+5ncI29IJu)?S<43QUHHb5(6K~g
z)Wy+A*m2+@%4}^~lCP;lIa$3dgfB>R=*r9*moDj==wOmW0ji<uO68Me2Ez0PnReq2
z^)17k$S)}5O-dgvEfH01ujkTAQd3ZLqHVQBx}=F2SB)4U)3PJciBf-m9geLzxS?d4
zOA9r)U-ZBc9IPh<7`ZskVA6l+mCkB_Bx9dsRH_E|c`xt=*;Vi=EG_p);f>Z$H`hYP
z`+bWLzNUf+&<~Pe)zsAJ{&VCqGdJat=aX1PS)m8z%$>d6W!{9<>|vjKP$a3xoro+#
z1tgHrb}>6!g#KZw|1<rm%g7Btuzv0~nb|Xz$4PU6V0*tAMIVL{<=_evfuJcRyc}HM
z1YQ%4X8(sX?XM#)b?|mn+#`AiRXhgQd44uu`ZtG&?$mDtf`q8)D_-vfv@n6_>iEB4
zWd^IKrltxU>^0zE3fZ(l&fomNZSvl_hq89vov`L#5IarEk-=PP#W@oXJzdq+GaM(h
zZ*9e`b5R7f9dTop)aoG9>m5(q8T{=K{kx;`<D@DMMhy%K+IR>MD%{<t*=KobmFz3V
z7ZL&DW)9u;u)K<J`rn)Bmk}PsyhjwP>a$1%6TK9}EhPhxn=6_Y{*1-l_pfQVVsiBU
zKVn_~bvFL(B~&{#Gjn8%^~DQY=yoFS*(A>jG9k*^bqRmXzo34a93UGS8jg9*3qoK;
z0C*%ZXmcF@h}0agR$r%xmww`+{~urEZ7s%03mpWyH3?8R?dO}{mdajxmH1*0e$SNw
z9_y`ojY*B^XKeo~o92I;nm?w%@fy3C^n3{Ym&nXNPu15_ATUzm?!@&Eg`59%*yZ4R
zf8{py75jgD-9IM}Qr%S4*dNi-LP=cOe#03iD%utiowV5}&K({Dezv-5`shoMTAB7k
zmp#<ltsbS&NNwI)|NM3#6g@RyVpRERN&mPp|35x_e$zp`X9s3con@%?Y1gRTh^{^r
zAD=cMgZ8t~m-Co@IepFKtghr-8*>O@D@E1UR0)*^la9V^ZI9NA@w`=u-UfuJ^0$(b
z|8A(hW(uq;Gf=Uxv_6E6M|y^3Cwt25?P2Qesd7C9ga|Jz!a*BC_|4xvhC*J(yHJa}
z6MJ0q8B_gPNKbsFfKhKU6H~((B+>1N*6whL=&a&sk_gse?b!qAzI55;r{dz8w65<1
zJQU*cd~Qs3%PC6O<@16e8L{n3l?h#mXER~Z=px?0EGpRi8}6n&zIV?j2QJTzT<Q+J
zM0(XK)exRFJV|MZFvcYA6Sm4H-5hn|D<PYoS3+hk#(WdhDm{D_mzSWU=41KxUkbEp
zWgQx4he|*^YUXR6yNAxRZGyu<3qh(TN(o<7Y*Aq9{OtY>cN0$t;BNNuAy7IyJK2!#
zxa<vnU_zKhM@M%qJn7ynO<#yeJ)G^*s&$H(ahmL8bX|y7G1#oZo*i{Mo57_<V>s^*
zRyko9PWyq~oQQaj?8mr!Q!h&pK#&Yl$dM%d7Jm7SvDjo+(ZzgP+-{7xUFzJ)^0Vo%
zbUcS=n)}fbhpwq|ks93@x{BB)+X!~?Q(9VV6$TTR#IL6e=i_bqpS7T4PLui7n2R5?
zb&fRpv5>J9uD|-bG5D_$(f_%0S<SqGOwrDpxcvIqTb^*!>c#hEnF*bI;ZccC!MrWB
zTwEGSlIQvBLJaoLK4*XCj$}T{kNlid)zQbK)9`B6V_To6pf${4bx2{kPva>*-}zKQ
zKMHaQ0NoVLPtgs*{mZ8OxfB0uv+BvFI({Kc2SS(b&zo@?ZLhA9BZB(*mC!aIK_EC1
z5g|cmhFrtbpi6DDP@S?@2NNY<^3^NEwyL%jYNdUcFSjqs03D&N&&|_eYh*U5x4YLl
z+ihb)#$o20Ou~$z?d~Np{WO6z1)ebS?h6Nc(%)a!2TV|oCf(vRcS47!zk?2PV-0pA
zmYf=@GDVU>^Vk4%bEG2;D3DcwSi|}ZvN=1)hR;HI57QTymnTnc_#w;7HD88?L28U$
zuezoUZNI8i5lVk8ygt|5thj#5g$}Uw_NYAw$L!~6eYn(wgpfA+h*MgTgMIH$7C@&#
z%)32c^&L&4I-M4RoFiIsW_3zSOJ{xnI0xXM4Oi^as$IxcSLm>j1c2StB0Om&@d{5)
zwo-|G1+VdjGKUJ}qdyms$WN3E`5)x_zn`%WJ~uJ)97cS(bo6%EDE5&)ck!#cJbE{t
z24oD`jjgtK2NeZJ@?R#>)gEjz5QXs06!IUWB1Bl^q%Yh{$&3~&DraLmVV6%=l0l{A
zRx?AE!ZeV$Ijro=EGlSc=};gR;qER*pIP#q05!k}>m7~Q{g(bXa37_stBZ=|u@<{^
z-yUY+eJ+b)Qu~qSh{ULS%q*gb9}78wk6>8o0b2SzJ{t}k6<ue+2~@4+1gbIgTDn9P
zmcv~rh9}={?i$rf1Q_pp5MWu>zizWW!IMq;M@Gy4+$ui|9V?}GBOznV74K!;0ep58
z2X$}vYtiE3_s}S-8ScrsK>2$QNYlLUHcDS<Wg(cAoVH;q3*ViiKYVy6DJ69X6}ic|
zd{xoGxL%^q;jCpV`H_dr8j?YQw|;HE&xn_Q%YmfGLS+>Y>TYR_?eJYmVss0gP|CVG
z%bs?G=?{%#k=xC3WuK5;RUPw(=koVPJ;o(t=#D9h0-C~h!K7lkc`L6p?&81qKU1xD
zN~#4N7bZ=wDmFe6LNB(_jRU5@qg(&qOabt5U0hDtRfR(BcTz7Sz!8*y1>Ng)R@zK6
zHeQj?OTPjgYd{R71O>e>I4x^7*zDvq<W{au><g1{E=t}u!?x0a^Sx#mfQD^3C^qr^
zu|a_$80cL5LwNoBI{fw2NVUbdI6JpE_F`)i>P~NyGCTxVIJtf;Za$hk*Pj>=baI*u
ze8aBa7BW*iak}ks(y7rayS_sa*?2xEn;kb7e6$Izb_@)n^Ftq;7ki5SwdQ(rdiZhu
z7JeF_=1rb;X%PxAIL{8&bhBwZ*e7{sy>FgX5BGiTt9O$yGV)78xSUy+j@#%E2Az5`
zM6j8U^Nenk;qa6?&7kqrZ_l<;JxW`+Yp?CmpjCLDMFA}43;SkPzsJk>f9I6n@V<|y
zYFPT3Ub}SqVJG?z2<LL&xzuux#@kqIsC^h2Bh`-l_wU~g;d3)f-IB2GQ2yR9lA`*N
z0_M>ka+itwJVldqd{=uK&NIclw`zvDOWaSN@xby;vtV{fE<1z9Pk0$d8DSC8_~pr{
z=s@f~a3|gGUqLuo6}?kDE=PIy1aCT*4S(mqoYr!hL&}(5Ezq!^F-@BIY)#IS0;K!K
zFlf-$9>d!DJk9+uXc-wH!vK&VcG4cy{Gj#-D#^?|WXT>BoWR~)spN6D_g&g&nFM8E
zQiY+TJw#_upFZ6w)O+BKMhEC#Xl`q8&fj@BtjgXI5sC>Qq0q(x3>)}Vb;f7TEx7Pf
zJP|BR(8vT2+wNV5K?H!>jvl3k`o@eEs86ks)jHlGn-caR_lQ5O#Ud^Ka#8;-{?nwV
z^~a!~*z5dAlJDQ3HD})^9^;0qm92CG?%pP&JvGjRaVE<uA1*O#6Pf}1{oy)C7kSxA
z&3c?bYTC6EAm2}l_^lHl2JqDG)Vn#w0-3S2!v3$ZW(M`!3+;w0V%1_5S-!R}GpJFX
zJ!PtmF)kdK7jaMx9~rqv%;g@y+SwC!S-VBg?H)jg-36a6si)T{dDX|cq>3LmC=?MH
z(WSA=YY4V<$O+fZBKKZu8qCPNb{Tf*BT;>B1PMG^CP^2Vo^ggLDMLeICEw!Prhngk
z7sNxdqGguVd)s=(`zGk>bhyXU;3w}0ajn!po>Y2sI5U3c^=-omspzT(fN;hQX5576
z{X^ehJ<%UJyf841EY9t>JanHncXw5y=dZCZ7UG)F6$>@F4>a5yV+j(lL}fQ#nxIy%
zv`~g-QBHZZev@IBpnS~BYk;tBOriD_6^)1sd+Ud~%3!H}3wPy?op4!kYN~t|;-%8j
z!|H4e!(Q-VDI1fDqRdgy5l{KydE*!1plkD7^T2%`#r%WD7P~qwKF|uhq=4NTtcF`4
zK5L+-C$t!?U%ZT~<lzw7)o_W{kQ3`qfWmlGuYBH48s`i^IJx7|r%h1+2qz^n)sJRa
z|7;|#Eydu|fjq;YSG7ynAvxuG0nxOyoj}M{{j(O2j&dO0=#v*O6u|73cgvqm@1+LP
z3s7{Sx}R#L*F`$~)@Ay676O4O6jcj3(9yqZk%E0KUZB~6;Z(odH9yRpg&bgxA=!NA
zzg@I(cLRcQGfXFLwN{%<tS~)rlk*Z~O(r<|Z3F>Zk%F<$S*@|WJYVpLpOAZVf@g)2
zFyy0t01oip>EHhahcf`l`vVH>$;86K!hKG#p9^|Oy*=ZDWrcxtlk42m2mQZC3woo`
zhQ-CbJ=aowJwkC#j+x}Nkz{`dv;FG}5r4og4BGzHNavrY`>U6Lo?ZlR;Kh4NTfe*j
z?^d(t{y$ipTtwJe`wpyYfIJ;=_AhuoscGm`U)v?^6w^1YthBX-SDkvUVVw4<EAIi(
z1vp0(V!`FVRTzS`DV{tLJs4^*QtV)QYI5)pdRw4iRCDtXpmFAyBk90jYgN8-`}uWV
zD^c+9)PgnS|2Y`t4F@o37=R8Ik)T7trLk{Gqf|YoQ$!=>Lq}>}LE|P`Z^tq94Wrnk
zJUjX3wQQR%Me&4<p@v(;`d6#{cmF@fSHR9LA}Zd~<<$46#naaIV`lbe$(OClBSLKN
zhKGij2XzHm?R{g)M+_9)ZX+LFcRL0B)$R13r-IehJ8IE@MN>2pqjU#bD^2hQpzr~n
zA=Sr^A5k<$b~r?Qo&h@0=VA%}x4Wpy@L~#ZHBbcpAiY#g1lUg^Uq$|jCG?-qpL&Mu
z+RWdKqI3_?$ZiJ({-AWqBgQ5uqXdAd%Zz)ZdY_Og3a5Kb5!iDzBbf}p)$S+&!S(_W
zY{3*adh6s39zFrY)Mti<+yUsHxsk!Sl#GnBckeF95Qd@bKS!d+s#PQtS{u9G(du(0
zw7=cZc)4C^$W(E<!io`(XWqL4$;&dxyMrUdXTQ`A{twCQFWE{Y<9a8&Q-<UJ_N}h}
zMkE^ebp^+N3%mUPKmD(P{x2!*|Cf;wXeA5?GM_+X9}5EO7$v;;KkN0MfAXHfz|&UX
zX*8Jz-v}~!`7gM|K-g6f^J{o>x?oD9d=kLjjJn<Z&%69TK75Yvja`C@rl27ijN*VH
zeBsJ}@kb9zZ}@y*T`VKY+eT6Spr}jyZ&n{0!4LLh4z2cqmAIGaK{AQ!zfdX%ZlK;>
zsJ;igA*T_owVbakDT4ajJ$pT@CL6$DvmU{2g#6!TcK_!aruLyM$Q%W7e!G>+(r;o0
z;MPpINe=!Du>GH3|If4mLyo68m@23985r?&#fQq2|FmL$SwML4UQBMH29CL-JAlAz
z@^0nR!+()CAd(=EDB)grD){kV-^YQ^0)0jCD7kwZ+U4rD>CaMEzc|VC`?(U3MyI0s
zDy2y}NvF49plpW+M4jKaKKuK)dk<X8-DxVtKYQl>O-(W}02qR-$@s}OO-LEmx#z6~
zRLt2CZ>B<l3m$Zf>IwB2sM6#|^26J6jMuwoumX+a&2f&S9~qGHE687z3!={-x+`9)
zmy4j`z*oe!`S(tOm`|bPpOxB=hU#2~A8b|@=SW!9e}5FeyG~EcYNG;3ljvqE>zzuA
zzOk^F^o6^tCHmv3hiqy)BV|RMRmap4uIhOXtJ7)JG685$F1AaE4EpLCGoAalA`%l{
zXE^0eJ#9=CuxdaEo-RXkOSg081xGzkc8OfJK35tn<*}7&G$y?tQ>l2{IYeMLjtGUC
z@zywFs38iR`C`Ul7fs14IGa%4{8k_ww1Y9d52IP!XKIXAaKm=gVG^FSm;uur^X+6_
zT()ZxGrtqac{Zx;H0tQwMWh3w(LaK9Th+c>wR72OtIWKvy1n*Kp&Io@N%>lpg6ica
zgpgSx9Zx1X{^Od=&slNfeIXESTicRpJ?)Gv>4Zmaq4)1PX>12ZMBH_;zQ~R+o65ED
zoVCnj0bL2p1vNNhkul5P6{;(Gon0WYKGt^rDhFg}ywY*4yVELVclYre<{vme8##Kz
zFnnZPb?9+ASk7+&ZwlLPeN!>7b=96?zXXV>uWBKdnQCR!vu^~4?N=D0+LK+?fOj;%
z&?X&uiuq%gK;oOyMS<obt7NljTYESa(j0kfemgeMY96e(94mxZZ~S#l)dhef-T96V
zR)bhn*Y-4A`nf3JYLc#1O#PC<6ZO;&{cnt30bwFl6J)Z3g(O03*ZcNOvrcDCd&WFU
z{`|A<D~~6^3C76i5i?A^Mm-lw2Q6B4#6hLNO^R(63h2a^P}{7ed&3?%K-y%PB^kC~
zKsmie!>hf${Z{31Tt`=h9Vz>gV&F=vyS^H@m`SxdR-q)HB<jf&ZP=~S6&Sr65fejL
zu1)lEzlDU^vl(_>)H=FT^M#R_C|&4QZd6RvD-REErI~G)RzI1&X7LmNXPCRVwA2mU
z3q95y#rX65%r(Y6&k?NVFFczmiR)n!TK8VtL9C+Rx3m-^(5IT{2ZD@TJ<K(L$_aFm
z2_>l0#6xe(C&jZDjU!i)ZwP?)O8YUdbiU-7J+=YOc^LPCt$m7sCf(T88&`cVkZgHS
ztn^kY$Z`4h5<v`{hhH-c=C%+*V|34vbk3%7n{8!jUlb{vJyeiaE$LAs+B$8OYltG3
z5%|_W>=_HSs%b&XL0q;c20#Zej~`4B7Urwhj}_*1Z@fH2G^Bt005LG^JUb*YOmcz0
z=t(YXx(wlLpY-sDl-6Q6jCj>_y{dBwU7yU*Fa+wPSv(UgJ#`Lw`KHf{HMrh?%;oRD
zhu~SN<TvVm-*^e}q)P;y?H+CWHuWCz(siS}Gbs+QyjCQ!9+oEn8FWLnai+hVGizF2
zbAhfjAvI(`PLgBx2b^E(yyPp9<z6BJvGaByYbYA<(i)u70Um|eB2`0Y=SAI&%`$CG
zH|Glxo&x@4W@i?aH<z}{7(<zoM)DP>`yglD!*wW)NfP5g&>5X_A!OZYYna6#Ngk8j
z<H9$9(Pqt;tbV<!G#v@=c*aFbK3d1<LO0eFEfi4X&kBGngP1It1PKpq!Dz$)X1X^K
zwPXBNWobImcWw7WvPuM({Z#~X%awDH+EthBbrueG*T4-zhV%H1`(W&h)YRuW*!5*v
z-5b@ILWSdA(I9c`eW)xHr*2|h<xxw#ru{sm#;|$su3B-O>J)2eY}P7)voLTVOO88X
z=vl?!Xej5?MJLuNnkJjR_gDAhyG0!zNUy120hIuXv$yaPT|`wbk5o!}v^DAJ#$qMZ
zUnOF}<Eao%(osgl1fbi{J7s!HXNt85`e8rfvp3fTVp=`Ip>z>I{nkf|Ub@Jh@^`Bf
z)TNe`rsEpM%kTB#YcnT~Dx|wIh1XG5XSf89;7awM7=x5!Iu=Hs9NmXPa_Wi_?7Iwl
z#q94=KGo_uRd!?W6*Xk`RQq^gcr;p;amu@s$X+m9Z7CG(tRz<4os(w=Z$6I?IwV;P
zRReGfG}=+?MVI{NQ}-r1m#Le~J5n#HG20wmHVbe?eDVP0!1!$icl`KV`F8e&1(X{R
zr9ZvFUsSj2@wJFpruV$Fcl@eM_6*<j@!@I1UiT;1Qg9_d)+7OlX<$R9c5A5Jh!(7;
zoy|{&TZ&aJc-nF`3-YOT4Z<*pvqF9>Owju&O|(qHWGqxds@BQUAU4VE^v!r9Zepqs
zsDG>x_W(3ENi-W(Vw|$rP-H6L>w<~Da;OD!w(IhtKj78zAwP6p4ACguGfyG@b>Yf1
zxm!p-_jL2`DwVfB&z#S}%UPsL&w-OC{m1nJg0|}Da8!a`)xlv+6WdZ{xG0fUp&;Rz
zs==7fyV=!>38}U$a0_VKEoB1ru|F<h{f+FUNI9D7c&vk#k4CmRc^_(n-X(W}Q$?9b
zeV<`=mGPTAI8`oqCTC4<=2m}`{k7|fp1rST8JeH^O1YA{IyFCH?+eTNaegy)WArU*
zdM${MgTxK%LDnt(e(p}KgHRT^sM-3O+Wo$-md^Z?*AXT)H<}_ecatus`givWJtvJ4
zVs+fc>x0y`&2TR63P4A1++~=5&z6PQ;A!}Dz&hEfiB)@c#zR-%l)E~8cqb19O_kH}
zAYHxa;Y_01@hOaB#{Ae32w%pfVS;PBlRn1kc*+5x&FvoL_0vy0kk{&sD%mJp?(YIg
z8?9}Wqezl_!>gr&3?~>R_nO+8S2qu*b<STyE_VF4tq@8$646p2GRII@5q0qbB-YMi
zz}va)OUij^-o`ZBAY|s{u=6}jLp+<qD7x!2u5kiI$bcY<+g2i#b5hDG8V-t;j)5Nk
z<uOnRdQ76TjE{Z8Wq*06l@G;=yz^o)-gggHX=jLt8Hj(N&iXNyPF6Y=am%+)`uj{f
zSmv|cnq`AMkf8N_8|xCZGv>Lz?(Sl%XW-3oov~#WUtcA{=uXiMje?Z}KqS1L_*>as
zb4htb6bs|>0mSuQg@WR<3kF$(;~Cd7cx{)g`Hu3jRgzDPigWX7HYpwM%ZpDYsZrWH
z4HrZ9z2-rlctH&wq;3bz)ERP8#8<mSN8!=YkNwF?8_94ab?nRU#Te%fH1HHL_1VH*
zjpHwk{BP$Ck3)!bh_l~V#(eSsxj{`4FH${qXak>PiW#c8Orz1o4~!pA+Aph|xmH38
zdHgNX_-0kk4yFbhWQ~n76Vl?qAB|E6=oNV4DEzva$Vm3;DOgy9Yy3M3Gn-kx4JQLf
zm+Q~ho7z0~!7noMxM#5sK#b3ugjlTLFjQ9$R-bR8D=Cbj{D*;Xw|X4ILtt$&#|Otu
zVu{Z}s69E}9Hg~%rYnpiWy~U8{34WjsR%T(d<?}!oK9<muKvAH`KNE~MIw@R&l~BU
z>m4C-Stm@H-U4r;3Wn|AtDY>-xD(C;-Y)iR+8IX)3<l#QCm`1d);H~|Zp0W!AeDIf
z92}*(TOy*AxQ<8cnq6XKZ`1WHIAQ4tK51&3r2QojTNYJ%Z*B^OM4JS<GBoys%~KAM
zXCH?)nq&(jIG4UY7DjtZ>e-VfNR)Lz8KLe7FZR)Ve)PI=xt!nC+a`OKSLCL23#w{Q
z(s68WVH$bMC`GXYv!z?mEO8>d#CB{|l-1f)k1=jFsSa#)m9U-F>2Z#ATnHHki(OSz
zeuJ_iYUVP(hZg6K)Ds*cj|+$LS%-=;4fv7}&L!+&s-JK0_Mj!gXSOu#FO(|GYlo$~
zoHrvcwH-I72W#ng3Rd_uuf1TG4OxVQW#XCRZay*N!*@I?2Q8Ov?I*XyUyokaTU!8x
z2^`W0(tDb_qX$BR6xi8RG;NZhI2aKc^3+n&erOE#(%<~QTTju4UY314SXJEh6dO0C
z_Afa$l1$Dm60L-lfZj*NVYbxtefmZ*ynL?{gTp$U=yK+@I>&yaeBpW3Maf3$91c{K
z3^6e0m(7=tZZoifI|EQAPq*Km=YU{QT4rQ~1=5lBxVXF@LWYxYJ#+3O8y9fVNR3AY
z7$xTfrGo;?b2bl7MUQN+L}d)V7KYzagU=lGJ->B!B^QL&xO%layS#8Jj^D6{nYDK$
zxqUXC=7B8rHdT4<Ok#yCZeiM9Hil=*Wdu!+0Nf*Bs_UmYluz(+`EYIKn5HY!+B1q*
zI&2DDv2H04>+Go`b0&B&&)`a0IlCKH+mRqBNMyzXMitWOXXBhZ$-s47U;D6Mvt3!7
zY{YK;;>xq@CU46oW$~44{n<O6!MFFUv#3Dip3`-5PvRftwV+M~-{<bYhB%yM^{Z&Q
z)jE&KuDdnOKK4gpxbo;?Vy6)DUx3%z_op64GXU`b;eoXj8H(_jM{Wm|{WG;rF;_fS
zSP8tMKKrxxIkb=1wmD{M97@+Y;(H!|wS+y%0qI@0tN`v7jP@qvKKsPj{I9h}hfh*r
zt0u)RO)4yh6+BbHibz-1hUw*z4w%2*&deC+3<}Xl-xr>DYlaO<6M75YPg_%gY+yzv
zyRFHwSP&qZa*4iQvQDDRx_G|V$|8z$+N_LB8Vd?^o;mia5Af*U9Ixs1upWvtCoVrP
z++t1|(B)6m6;CAS!@I|w5*=+xw4T&A-0eR<fa=O&U6ua5%D>AdZMD6Tg^2G^PZqZK
zdRdklhm1tKfONP`Dh66>cDMkK&wfZ(8^`nIa3V4i!BlF7=5x$<G2$e8Q|<`nEK+6l
z#;w|;m>-A{S-cHIOT;28t<@oMB`U^GOh+QQ6$kP>=p;&~Jw8l?lMw-7h#zIh!e2~o
zh~=cWHgY;A;6m+F3x3pTyonw)ACIrCCz9~A7e5Zrb^DNyAB@Sw|7eCuN;HI+@q52$
zn;k4c!}3Opxy~XK^Hltk#K)X-rnsi}xDETLH%*j3B=qf(9~m8r>BxHzPw<Dz#eP4M
ztY~$EGuL}vjhtw@v>B8RG$cBELflUUFdH99X<yR!V;2te+^5-2S@W385RQC#VO0n~
z6RcUpKP6~V5It18Q<q9ix5d%E9Mx0K3Dog7FeYdBJp*FsRi_)$Kq=28%H0!n4qJ-7
z3Jm7zC0|mvx_&vJWO(9*9mmy`_-q4^A}Z?T%Sdj$e6#HRyLF!72kv{lt)u-G8PG;u
zW>*vW03}xZ_G4<gP9N$WG5@>Od%G4c>oG=mZn?tEvp7hCAzxZaO(ylUhmV-myb|g0
zoYGFyCMgmr7<=#DQ{{Fx+h!cvAX?|Juv-z9dB=jvpamK7P=FUW?_r3?S-fDFc4D=c
zuIbdqSM}TGk_ZfXmHAn%2uXd)v;Ibrs7~x5o2HIu;`q=|sp4Q07lSM>cH$l(DEK;C
zXS2<{H_m<+)p%-L*^=pX7Uni2T-sbfEyv>3_>6lIY<bFh`#h@Qn_HQQ$r7yBT;K3B
zV5gozb+qz_?Hw0&^)J9f$92|?$A+?8=byfBRv&$l?<EFpj~!z$xMAhLgKZm?S|qv-
zDI9?$uO1GhewAY6?jB1x8<tL>mGTkAnXqQ6UcYR&cn-VcAA-|Ws>d<+>z=A0c%Gu=
z_!N)9J6^yX$_nL>M!J$K!rs3RXZux1`G7YdaDxF#X$Zj~*)?@x$FNr6J6n|E`QprZ
zOZ#M;pic|h$L=a96EurNmIoJwVD|C$YE>9YL0+COBy3%@<2OQvU7NRnN;TV9HyUl1
z?}*eOaLwf`Paz9`_XgR7Ypi2Y@{HvTKkbXbH10m$m3xu%^IUHJE*z2Xb&9b05f>y!
z_gV2WQiw}MnCe`%pN=_I07Zwj^>I0-J>tF?5<8>$C%N8O3GAr28fj_(xJDM_DzPF<
zw$aw|F-0cE2Y0%?Al-Da!n1R^a9KZ;dm`)NYKSyNJ9D&$RQ=d7a;Bwv;ES4A9g%60
z-vQ3NiVU7#3|XzS?Y2=r)6+1f#yzcjPHSuXfzjq08%0K{QS$9+t>_=<NP2Dkr@rMj
zy4c6;;`VukTQj?iI1YOGq)|U#SXxi$R2a8}G~d#>RD2qaqU!$gvp_Oq|5UNmu}JlC
z3$+!nzxr!+3ec@M9Ubx#t5UA;sJNfCii&4TWkWFRVspO`?KPf!7Qi81!`uOCJ0)3y
zEbPy*r3uMAzG~c+dFh1~Z1Vr{^_5X|WJ}vvkl^mF!GpVNu;4BU?he5r1a}Ya?(Q61
zgS#DqySslUnR&lEGwaS>tABK#UZ=bE?yA~VPwlFLMnJ^Cebw#2H>kF8EXLOXWq*Nk
zu|o(YL|H9nR4028I0e*W(l-lOS+_{y`g*k4CN_Neoiq!yZ|ATbx27MpoBXJN54G6X
ztv%s;c4SNXqh{I3;tkunViqnlgBbhPfl>DZ{2iaf*nV!r!?K!#0@6&Jo8zYXo`%Tw
zSKj25KqI^TrTpBz*VWFPt~U=p>Zfyp55bU->W`!2oZz2$=s)Id`$X~U8>^1uytljc
zCOvekB{W-m^_mrL0D|};R(kD;1||}yeS^=w!Z=;Vi#M3n1uxHUo9mti+TB#6o7aPD
z4wlvNV#^Fp?wY!ukdllbiLL*o0KaN|m1AnZb6vgL8v9Nl|50GNP2OFm2mz&~Ltssv
zEv)C_VqSAKDQ?x=(x&1EEwa|cb_!|<V<_}y+fCvrzri38qy?c)w%0?|lPy&GjY;{`
zJ!nV8Vs;KKe8I1*wnt#ez@Aa(g!TVrd;GG`W;BFqmA^;O)u=?5=kzDO|JC>6y{7{=
z+h3fc5;VKOH}@lXSe53JpZI~^w}!3FCU)PAYt*>5(^wUvhp3F(fSM15xSP66cjnLm
zombl+A3d>>TH*aQQZz@oHngsp{gRA~%q6~_C=&~h_~W^ZTgiEx0$;Zr^+}m)hO-nR
zA{A`gVW;aS4*3o`0I%yQs_2iV(-Cndq?2_1Uf@tEHXHlhMrA<Qs>-c9su`2=Z9^+j
zV!JGl^DabAEdVX<>(8DjKviw)(UkTJ4RLp`<|&`0kfe!33#+9+(MmEb6<gMpUr=Vk
zyXk_F=q|{ohP#1e>Xu790{c-+>_PrluPJCFzUP%gmvp_onM=%roC>ust2!FfPA^(*
zy9!K72UTF84ChfUKUN@50;s|}KmqswG~ewmb=CAzFwN^<%nF?c{772EdLW=(-bMDo
z8X~N``%2xCQU(*HGbhHj(52n5leC!4>$E6uvVe@32dj=moIy736-&)S&~HhV8h~z4
z&0RPHt-I&q#RGHHIlX{bXu9AHb0b{3S4BCWkwRKL^uVsPibW_O%4%yi`GWu^2`2E{
zQW&)654~7gTO7kxlcggvTp$xySl{E?LyfwVMH~q>VS)`LeO<?iWt2eIwdipvW773;
zA`L$esMr~%NifOuZYN2S-Q$Ic%l$^_sZ~{?$-tygc?%7LHXEntvx-q2tQ9LWU36_n
zjdcn~`Agpei=5);4<ep-YYL`oV~!>!1&AFJKdV{gx%)eI;!|8QeD19e^KfIYuBFzL
zOO&1NNWBneHV}^{8Y){(kXm*8x2_1x<{}PlTILx`_oE)%^(S9sU#}JJ^Z1G;mIX%<
zE}mp;o5NC)Widn{f#@p<v)}K!jJ{7b%<kBslN5pzrM#h-a_F`X8XEI*Z<%H9wB50-
zUAaFhBe*xb;d9lab~y!vrO%TVvQ^2C#OBi-UD}%L(NV{#2D+WjWGNAH^gg~0tNoGj
zEu;cR;r@bit=ooh_c;Buh;qk0%I#YMVcuiCYejsgv-;q-%o_D8oHz01S6WotZU*^(
zl4*iJ`BDF<B9@19H7&0Ny1o3+Nh^|WfJY3TK!EG`#1a`-S_zmoS6ZsKW}qf=ZIVN+
zOeVkjx**@~vCvw(`oTnGKyzLe25--Pvp&gPU7YWE5=y@zeDw!2#hAu%HpfjtFS1T0
zI9>|32&>#to3esq;hn9w>ci=r`p%EX5yT1{9;2+$_;rNzZ0y#noE<Rb@po^idi)(>
zY+|HOUrX#^Rpq&Em=~uVQDq^C*l;MStd<DUW<?HvZ9Y7*s4?5?D;y;~*kdNS#C~D`
zJipflC99u8%j`8Qt>wy|NfkZsKj5;y*6m6a_JGp&b%JQ^hp98=S$f`D#dlHRORuH7
zbBC?2e^_}#N_9n+%(>88yrCmI4IgNp@e|Lq9VzH1lpoi~!9}w*{91tRC|OLpHkI5I
zWCsf4TC#~o(+=EP6?@C*w0g=*0o43+ml3bsaEOTpvgsg#Y4WWm;G_4=K|zitH5xkY
zlEoy2UfL)A{#1G-$<uVa8hyO2Y)}Qu<CA6*<(re!OvP^P!iMS0SDhHd&ARt5i|iux
z<(hnKog!f-Z{ZNCCP)<oI7jLU1o$+0>LMEl>W!R32t$oMpft?%mpGyt!Lh>Tp-jI|
zUN7gGQKk6kUPMK=&Sg;}s!Y9F%HQCPLS$9?;O*@|tp%(!d1iMo($kOgY|R`l<@*m&
z{3QOFVEMJG?hE8b+QWLr_3d-_rTnD_!|PM2?eQDu_D-K0(9X_6eeX?93YDgYwvp@l
z=do9vV_l=d%ZrPn+J<suZBKjBj(x4CvZ!0k<H(L3OW#aoHvfPQiG%kRRk_M3?3Np{
zb9b6S*LeCwG(}yQJhDFm5${i;6=T?rK#AtgYi{G5p5IvkR(-Uro(ZR3E1GjNP)lb0
z7^A*9!tub3HsU=N#JP!}4qG&Xb=A*!e|0Sm{!6lHL#{8h`V0km0Pwka5Q@u;_T~Vq
zeZlyI{cUQ->-PLwEp~@94kBXzL>M~|6m|J@u~~<k(*?;hBW1jZ*tYCecs;i83r{tS
z<QoLKrNw_YVvJYS33fT)*#|<NYD|TNZw@ov5wMxWuca&F#b+mz0*POf5h*4ckhtyi
z7uSq7CtAOF+yWJm@M@!uAL({9*avfDc9C3?+({3HeKMFdH8pFdbigbYYnu0>Y47jf
z<hYtzqJ$>O6S`K7k406Rrke%5v)@VI1y$drIT9~tl?N5KC-}kM<Mw3)<U?))VKS~5
zYp`?Jsy<)6kc$VYfZ&<k8k13X?w1A8eDO;Y8{PJw5B*KZ$8CCTZ?QYG7R6nbolRX^
zGv%s6p*8j<D1qAv_3KuiOVW}7d^e>Pt<w+so?W<e3&p2=fV*L?IMUu%Br{W8T|OqA
z{njr)dU8f16#o;$;^t28b;!lfXh{p5LAw>^v~qIKAGc3cF|jwDzlW53z;{H|C-yV$
zt&j$F2nnch@G1t?XEEVkL_l|#h{iCVXKAW%)kb_M8vprrHOa(zB1bg6jt==_HGKDz
z*(@6n!4%!J)`4CJH|^cl^4i<ga6);GaB|7K!JgBdVGukPtFNKPl+4xbG_TuM^rUmr
zV7Tfz8f&c=#ikq?_uRi2!zvct;k{?+*LZa_DJReV>Pq$wee|#lYn0{+6e4*n7(Jr^
zp*Ax{$M;%{kAT|anqluJzURg)1T@XXMPYC(9;1Lnr#R;vo<4k<LU7WQZPAgKpooUb
zTtXK!YD->g*u}gK6JA`NEB3*6<v$K>i#DMAD5%wqgzSNIBC}CWnqKXddoS$E(VUwT
z&&d4NtqB%s3Op~cao4#UuK-pfjlAcZIXVtSxiLEpycG_IJCxQI;}I-}fpT@X(g!)q
zpIeqD8t<HE%4qPO_TuwL)P^_mM~)v8Hrqs7IZSYp6&r@blYQ>qs1UeMc%M9Gq<HJS
zv}T>EKHT~%$Nu=J@w?^W{}+#}jyDH8aXwzRv46HB`U3|hAp;*3QH=Zfw$Of64U(+{
z-)Md19G{lAcP7(hft2j3yT!yDR@LcISwY|fk=F(#?@BKb)uErB6&(FN75+D9crN2w
zI2*fR(7w@>tiZ^cPMvtq&0}P14p>`>H-#C~15Lg^;YXXYx*+kuvPHeWjJOIu9gSJI
z-^Y#i(Uli&PcwB?Co?+XAR%f@$P$cqoCWH*j`qc-Y~Xi`o%YYcD?dyKMbDPb4}nS+
z8ZBh_=Vk878@f`nr`e5({@lUq&E~-8v~KVsm;h8Zv$oah3xz`$?h$sUa#)|kS53gD
zx`WG(2k8weE%$4%*=>-=k&2A`Oc&Mf)du9NjI}b2Ac8HJ3V}ADtJ%k}PRJedQZl!G
ztL2N`JyaNv$2_dv0(s*##+J!^+ydhQY@*_?86hrQsEB!IMn;i&i#)}(q$953?L6Wy
z*P=}5RxF37pBKMlriuJ~KK0zqYQ#nAgX5Sal?Ul=f31!T#XY1<v;#1<$3bCro457D
zjE6HET_ByVP1+j@r#yN%C9!6Yw2%MbO5STiioJiB1njGPHEnf0JknD^Qq}a&Y!hc~
z5>o<IR5I3&yT1NB-qqR)Y9J%LD_VOBVvmKNNQetXWdv4Ve*_{E8kLqYFQTV`!Fr{;
z9Y-Lq$R(wcX4RRU^>5i0F72aTrsRn0I}*D+iGu=E?$-y5FbVobu2=__CD0Y-<^WM`
zM?ascbfD>Jc23ksaY)6Aju*z(gG}s9^pemu7HS$f&<e4kbLYBad7+beN(X!~Goc*)
znI-Q3lJWjj_}8X>{pf&3@4RUqr{P+rj(zCTaDs*){HCH)%Q26%Mt)G1b_IL%#}Pep
z=w;G|HsJ_)HAJr{@7-s(8-9s3<$)tW(yGbo1EM!LR%;@Rso4auC#v1x(Cj8*<VQ1N
zlG~8l#$VlJSdx#(5HUVIipac$(tk1aQHr$vmR9n$8u`!b3$WdR&p(xU9fw0CKPV~X
zfD(igRZBgwzqWycgP^a5{V~(0heAS$&CrptOL(veiL0^MPFaKw{(EniqX~IXxU?SI
zl8rpjBnDPi{ZwZ`Cir$7$8Kvq_jt9=Jw}m9N1oGl!dq|~yU|UqwZd@FBj|M^>`Sej
z!B;Fo=&jyO<Rm~6eIs@g(97p@uWMW@Z*gG=<_<)m$+i@Yf|vg16DhqWv+6bA+qZ9z
z+kn@{IvPiF^>K)xz376BI35-gc3?ngwmb1t4VTb3B)wB58{-ovm(b}3R$}Y%+8<{`
ztqkRZX;t1b$xk=xF>QJRRslD+IIIYgMKuO);u3FlU(ci7IP9Ha;2L}-5Wyrg5etJB
z`C`L?tXe;1yaPz!d#NnA05;z5C>qq7?Ll+98Rzsw?}^rm=s2mQq&4a9b*Q)aYnA;N
zy(Kh!R9lCXgGilkseQp()aod&^>&m`eqx#DEE`^@vJ6nxN-4u`byBDk#&v%RLMYeR
zojDh<METGZt3pJBYVo|1Pf5iaX{(L6qfwu&(s5t^v4e)VMr?d@P`6ZYCDa7gZg_gU
z-G;i=WuJxzO3^0{bLYl`sNN4&UB{&YamwL<XFR1@95VaiyDX1t0^0O>lAhOq+Olot
z`o`!&marYqQ_mx@Hx(1$rb5T(n9`*w#Ljw&o_{;peM)u?in>!oU*_u7%V2VTwlfQ*
ze6ze1)MpPY4CLR%C<r_+S(kt!@h48k=v02<^`RZ==F*(WU(%GcfWfz3!<yLpmksXv
z_dk2apIhpFRst*ujk1GYhw)EC8n)w0=-GVFkuD!gb~O((ye}rbal_MGQrNA@f(Y9H
zh~H+w!lXhwr~%I>)HYa@)|<NH4TMXLaqnS{2&I*0{+?!OL5cF9ouS1()c@B#jiBP9
z{H{l|yeX2#emM0*zNlYTQc85DzhYxxEx`wX!fUB**JF}4?4~$~rd^Yg1T=tGZV>)1
z{YmlDV3<d{yNhc{nBL%YYk9l4a>3ylA-X`S(am*r^Ja58Zb44Fwm1{B1zg)l6%`e$
zQ#|5ZAa#!6B21!~<Kt`ZIMs4c1swxjLIghNvc*!O-pXYB5lGNu+sU7z{coI8?~bFm
zy$we!qeBuofkaH^#U1d#6vfSlQl<uQniiXUIFZRuoDgBGVnRP1O!ZE=Ud+0a#>%|B
z7gk7#Y|M}Wp(Z9z1Nu#26|D!ad8(yY-iur-3E=2>PdTn~Fo_`J9m7jZDeZ`?c9$`_
zt?j8Q6|@&XC?&-=u`>sk%R<|!2+`{s-L3wNn9C{VOKu(XhZRqkf|aW@rq<5wD3g)a
zM)o)?xv!82T1F|_opO^xu%l&J!YJ$vct+V>=non-En)>@jIp?vVaFglHY%up`-6>e
za+;Uj*{Q1Bt@f-74__!zChIo|)obT=R;LefPCii~RLb9!`;S1yWbD7eHBFRZKWH>I
z&y3ov&$wn>KdpFqqnmQ<Y5G%iCur-RjpMAB7cRFu>9U&ZdCRvSipXJOVI|MEaI;*_
zYr1?#q&}1KsYDjM0$XhN5yUqr@Y5y_M2xlVNTpKtNih!+x@aiG^#V1%*Z03c8f@Dv
z-<4H(@{A=6mA>6Hg0}-z4GaF{e{3gLI$y>rO6Fw|lyX)PWz+)N&SMfB<7hP-Kt(qN
z!gJ6@IkGgp(?B&OD56rWq?b-*82Lst>5Gdocdw6K1zl=MnlKN{yWuX(0HWeMc}nO)
zg?(_cr9%~mmtKAQ<G0G*4L$G86G{1Pc_=<eNTeWsab(ftaxT3r^Fs!t-ZC<{XQZu7
z4lhmI5>p57l=!4GS<}G<pCcVBYNn>xi@(~~u{d!#v)g?nmR~fhVL~-nH#_@J<v9?9
zbNvqGMLYXph0gk~yV5#cQ15(hfRI9wxx=qMytQuXQ?xQ_4P{n~HE9ggR%zB-tT^b_
zojA<6-UF!KXMo35LIMH;9IG({I|{?E5nXW08e`H(!`v0!6?IcPi|KbhmF^{lg^eBs
z9xf&>E;rgYR<7#p?yA@OlSgXTu6q0Bf-G?K(97@P)+=s>lZuduJ=fR^MsVAZTdm?q
zN!zIvGK8=*0NwO)=y2mNqdt$Ov9?OQ`EZ*9#ZqruWJM-N8jX+lyBo{12wQK8U}VvN
zPxw!AR>!O{2y(WonR?yhGC1y*DI+5B%88j-BDsRE8<sQE_)1Kr5ot0Sj`#ibH`JQ|
zP;1S+<a$Ex5+aKN;|p<peSo{>W&Pa-laCA>VcB-_n0Wk|@(c76aOc_FBn*2rHzNaB
z$M+b(2b)0gJpbw9x^X$ZyYuit$s`Jm%($|>LxMjVUHHdB!TTj0!uqv3*^UYA4!+A3
zItr#ninS^xwOE*o+Cu1kHXznkj&h0lB>nDf!O6HafwTyj+SN7ZM5FhjC@B`3=Xrxo
z%OzdzPy-#JoY{vE#)i`<<LmO}L6`Vri7n^VH}p`S`D4VU?oRb@ix}8W?qq@mzS!$d
zHo(X1p*<oG6gRU=Tts|XR3lMFOVx!5hgsfiu3(pTdYfMHP(o@mHBUw|xu1>wuuq=E
z8z%KtIFiM_`)9-X#POhq;eK@j#qw93E=5+V2)WbB21lOI;1ToI^vX+9Rp#$Ry1Q+)
z3hEH>A;XS~eY>@1w3q>6Va4*>)(KYt?ZO)hg>N{o8rM0rFJO{bRgL6t9DMSQJQ`Po
zD=YLwYvh7GkshQkG)wy&75JaqO578dy7zIWBO<?(BcFi0@b7$@=@=&Iu4DU8S>6yK
zccYfWHv+Yfi_N8)#-qvn)AUoIlDaNR1;UAOBCmQeJDk>w6=adHj*B(hCy(Bh?#?|h
zK=HK~;4$q(ukP(jBn4S$=1kvL?nR(XtW^)0qMF!X_H@H>t5h%fjS(hT!+9e6MNI(-
zJm&7@DU{%TR}hf`y87TL2CCt~%rIH5N35cz6Hz1y7{uR>k5UlH>Va<;Sv>v0V155`
z#1n$M2{9t?QYoZl^;03EP&#xwpSFB}L$G<=Z$DV>BnF&&K@vfKpt&2!_!3Bg;9X$b
z?jAtZ9r=1!NnC6#zUB8+Kd+NrasVSV7ZldQzgVy}qb#3?uApIb!r+6auW!Hj<^Vj-
zDw7mBVe3-@A5K|u1xo;a@`E#Q(e#Y7N8!EPNyn-8^-md%kUL@OMH5jUnJ-K4l?jM_
zm&jfAS{`p2CkS1iwIbAa_cswweF}LJY|T%XU75?7mUfoc9`);YpV{e{@(p+Mh*%CB
z8gs82<H#Kb8w<BnZ+SvGMW%QwJHof{j~8S4I?=mA^0r4J3sX}3v>65=O{8n&!mdec
zhqs79QKa_({_@KwBEa2=d&`Y=G0cMZfdUaWb0cI71HB#)C!1P5$Q<34`(jh@ddd{2
znYm-L{$nQI^Ah+3M_(3kwj58<($l}H=>AjFzf#PND@L58Pio+Dj_}Y$ETPG8Z5>B0
z^E)F^)va(Z&JYQ!6JHp%Q*05|XpoY&xyBpJB?)3;&oNztY-<PZmmU3q2`j9ME%J~e
zAt=KOO~-z7>Pe8B<}(X;ehXvl8$6ZBs3~MQZ(G27dQa$efIeS~k|QGSy6}A^XYUw0
zExlU%;`nvh=SIeuJky+v#8HN<W=hd_i`A1&6650WqmJio4UOCby>$N`=a7rsl)@s_
z_O|-kmG){<yJv$;3adw;CKJyfELOlvJJ}@d)-%a|jf}<@!;lw6T~+Z554CSBX#Bml
zXsBBrYtKm6yuxDqE{dt%OvV><8xn<Af@?1=$_<u451<z&|Kn3BTl$+6K>I7xL9PC4
z1Zl$6sjNB-4H!7e-wr=Mbg6Q+t;P2%et+Rk5T}OoK>&6cxEs#AkJWYD!FBCkG-*Tq
zK1K%}ZE{cfTLszIwTp^t-aQox<`AJ-d{9#8#ksIv_PO5}r8gM7P<7Rv+@^`a+|q6n
ze7>)g>WC<f;~Pv;Vep`(m|kwG+`VY*-6q{aSKu#v_=G7YMS<i7e7$V-DJ9x@dgSbo
zLL{*6NpU&0QCqhB!E4jt&SlOJw};hb@PQdcN<b_@ig-htx}8156aPLe^SNkDp7#Y{
z3yh^Ai?(LZ8;^SNN~nqR_PXjh%_n}jtEh8-Zj&2{@^Q#K00fJUZP~`}k{@rw#9c60
zxP=IbeWOPU5EZ<_6SOHSG(l>_UG|V#1G=sb*WCuXkRWUJRl@4^M_5lS?kG0An?!aR
z{Xv#NFcQO9wtqO!IiZDxT}Ko#U75L;yM0+|);C2vEtd0H3IhriW95DxMc2WCeRj49
za4L9~Nm%ElA1*d{-pLtpE;ziL{A906%x>3y>f?^_+WLOAFkPprZ3#o;N;8K)4;S<2
z#%H2w+f23wkNdGokA2BZ!oJ(NX3`M7bL+^u?6{d(=`hi)2gXQfC1w7JA@s_?{S!%%
z&_>t~HRAY_9YGs3r<ax9`*9hm-lwhVQ#M$_$1CmX{6_-ji)gV7Y!vOgORiim9us3x
z7jtlp7aYPDwvr7|6Q5aiPn_aughtz*zgnT5*S31+HavY7unvtYHq3xsJo)J!n<s}F
zawkW2Ln{%>K=UnPB`yTLKUMAk7I8Ll+^lw@N1D{9Qpzbp+m`cW|0sD{Iwwu)8=55=
zYPY1@NXcSt!E}kKUJaHT&p1Z2*f+liR)JC6SNz1aH?dfZ-&55nu)7y*OV-!6P1D@b
zoSmGDOP0Ca`yr9}O<ML3K6sgYPv!#uCksQMNtM%JRiQRP|NRaPA_N-gXGsUTV9#3`
zf)E<}fII$A``zFt;%sN{UnGuQea@8YkNej&G3?F*c8hGe)oT{kpPxQHuW@hmX7@<$
zftt4|F54j3ZLo@VI6wD|1I4Bzlda~$&;8k?6X<Owq&tpB9#f*NxNF=FkxDiOG1UA5
z8b+Mr8M}57{;w85&?A~|_qz*`0{1x<LU}J9$;z{oNCI=tKFkZmwJ5Kx_Q(w=REM_5
zboCNju-El*<kBKEv=^T$^Yy1@r%lB9$#nW~xBUc86^u~ar(;j)3q-F<kz~=JB^5$9
zn!PIqZ&j`z-ZM3@KVdR)#ZO;vk!CNMUh~3U&WPpM0^Xp-ifuaL+A^H01E)6ZB-1+V
zO*y?z<8oD!pMzlEiRqB;Lt@XT^eHbUi3LYrEzms=XKI>pYhIfyv}p-2T?08;;ho6H
zf|BF>>k26)Dy>5&2$2K!HczFnsrL(fhL@9qBPc?gNGF*n`tGyKtsY7D)6iEM&F`6Q
z;T_ftNhvk%StSo?jQs-e!lQ^Un#r{JhIR|BG%1**FSsd}{hKIJkypG2n53Z8$Q<r^
z<xX#(q8}kon^QW*mzkvUpir%bdLG_fMmk##|G=J{2d&e3)~Xc9Gn8l|?+Q&bJT#qc
zU)s3MH!`JWzA&k2_I~m1R~%fUBU?gW7mmIM-en)^2(q@XFbNt?HC!a0#)~<f=T~~k
zV6OIx0J^B$)_^XL?R5q()h2b+ixo0tT6fZ=YZ@|#T-`Cj_(}9B3L^?UKn3T8uYXsK
zzCw85uT$rB2TgxdrQj$uln?{Pu#b$-o@sBemfuoTMIYl*42kl&Rh`6B7(KMU>^rDl
zU6o+p<5$7C@6U4Xm9E_(wfC*XV}bkT0R->qL+m;;M6ywaK$Y^{pCg=I;=t$C>;o`9
z#3M-04WiQKLV36K8^&#!hDDxl4wM9C)QJ0vy)ieFS_|yvp?5=}Teg*og~__eq9vB@
z7=1dA>d<d?cN~Rw-?LRHa$F4Uw2TeXk2Ok7bohWayJ1PpEOI527dpN&CEhj?o-E*F
z-HuzoabJ)6FyGpu3I%O#m&G8z_F#<s3}|gtRdZH-Qx57SuQ8}bjNXr<{OAK{4Dx>3
z)w4Kc6BAR)4#+k2yzC%~boc~|kyz~&IUeSm<Vg&0R?gMjC5mJ|?-9J3LX{!jeoB1w
z*au1~pAOi1(W~yNV(ET+&@}GaAFfqWWRNW!LdTtV8wtM{(1^}y#H(KF`wX|VL@;u9
z0FQ62d6>}M?CGtQ`!@KCaL$`%%6reY?FQb2CK0k_&z&9oALUUi2O5{g`1uMoMWO=Z
z+g<mBt}iKiZq`552+`ccBqPnOt;tcB2@Mnj^Ayf@zdG~6l)fr)B^B{rBvbBxy&U^+
zsH8u5$4M>{s3c4FF@Aj2U7Iam)@!xevU15rJO^A;KluQ`I5`-cvPH3IJoqcQS4=Qr
z=Id3&j!eVSaJ=_GOuhGWD~u`|R`WdIVHMKpHS;C0`0cO|7}Vbm5^`{{s=;8GfWKLb
z%ongaPdAQ^cvn49&!V%u#`ja<*9wBpxB>Ut#Ut`MIXU=dqya~+J{<6186NAZnCl`$
zzI#D8OWCjxTf)*6#qGJustGW02?_hl+*=2ZRo4_-ujEnua>`T+d7!*_I#)-NM|4~%
zBb3Vh7FS*AW$t)SnA(NzNgnx)$LJ`ENs+Bj2MS+gu+p`V2-OsY(;6-(I2?|S4mcX)
zl354V7RtRJ9((3SVz~hp;)v~S??n_Ukd!a8al9VY`L81I5hle&t=~Lvz1kW1z3!`!
zeI0O1g?9^y7U>03i|JfY1aAo-|Ge-9jZEThC8|Z`SziDsc@bMKNuhbDfQO$|sh^3)
zh=UdI%I&HZ-EkQr@vua=S|V6u9NcCU9^0MBAd|PolHoS3{3`JJ6n>tieWXtBjI9w|
zSvppdU8sjlmpR^nS#YWFp%I&C-e)?PhU`XOx<QNEQTjZ`lk2)GgZfp^$MIFPE}>Cr
zN?Htg?P;w1%I{WJ^x||pMU0CCIhIl`okdz4P-SarRPV}*-8F|T;E}uZS&U4R?`|I8
z*c*;M8VpAk+qzi+>ckNnwBESlPHn60fO&m;J-tGoBiQH?mheg-DS{G7hER9wOl%uN
zW<Eo+xmI!#XXSS_ACx(hB<jMqSr_No$Z$%0=rP#nPo^5IEzne8FIIS{;aS|5ojVY|
zVQpHF!asD?s&kSM{B`Tadk3Afh^O&oOApg}KQ?SEcQQBh)y>R>fckIT0|wEP2nL1K
zb2FE{`L`R)(Z1iJlkCnOgnNa;esH6KZ^jZpVOerLe1oMFYqA4!n|_ZNtmkX9*ghsU
zBL0#-rfHwTlK`@$D)NTVsf%*Os_$N+ybBueW85#wt<!tgk_(yUlQ7<1PafKuM13a3
zG#YnjfEc>z^m+Bea=#!RPd(m4#|_I<B~Stqfr}n|k!IQ1@q&rpZwEa2Z;)dz$Hj;n
z(IwpDcAwo|e20naeI4cW@@>(WF^GtNuVLR^FuHsXex^Uw((vFM$<bYz%vx~%xZ#9p
zo?F3)=_QOW*tOgzRZ?D_jh@6fwWt|2W$G5K(`RVGzhcsH$|`<7xD!u4_$c3k1mM@5
zetkB8EZ;8Y)8fwQ?P0oh%nP99xfR=7e|^Fgei#KrW-q_Y$SAa#1dAQzK%%}4j%rEq
zBh+!XqI}*LOr60ZcMn4xxs;}C;7V<ryPHSmo^N!~ilXGI!Yxfcb;Lbv*T)xa6r7M5
zuVih|tBO=4sNm`dwdrWd5R1-^3*T&JhmM0!Qcvwh)wL1KXejys$Ju+^!;}w&1x~zK
z4U`-eT{_?F0anV+*C2j<v5GN6nRBXZ_s?)U!KCCv$G=Aa*<y}S>C&#iOWy0n2W2dF
z1|8-kM!*(?BX6EoWXoBD<G9Kzqs^CBUg;dgt(4yRU4d)VmdB2@>1~4dJuLy_qR6kh
z{8jVM?i<A8#R992YtK%7#@NwT9y2p+fy)Z71B#=d76Qxjfao2Whr)sHmio@^C}F%c
zl&Il+7kAUcN8Oj>A6S#d6I91L@-B`Po%5nzzwGp}c`#uIJhgA+dsWAnQPivk^6xX4
z9^HFadbqnktKk|DBT1T#w#d=?Y<#o7Dpt@L|2jBxvHkMA56Bc3TTi^2QyxiT$gVaV
zpkd{Hg^`?yWr3w%86+kChnht9n?azA!~CcI3`7J+)#U5f_8W8|(Tf9Wi5=|3x`S*m
zY14Fj7EJ&pzt@t=QmrfTs{14V!FuzmYvwlTLRs>(O10HKs3~D#oQ`%MrV>qeu%()w
zMN3RD#sx|LvEPGosoqbRxXNOfGs(z?ROI(ety`YcLDM0@;&nJZ#VgXD>qfir)RW{_
za|9PzHsE3J#MmxI)&#N|m}1HW%GL%GY!yc!RoqD4#Ry(uQAgl5K-gQ%r%)r8>#d`l
z*UWx!cE*%U-utLX(QC5OL{(aR*bJ+du@zLzlA+t0I_f>-KWACl{52189q?&W?V9Kg
z6m!b+9P(#ru2e3NRYCS0h{Bg)>Ue6Q^%%g!zIR8V*B;{H--~CEc`CrRCEjfeHTRiU
zmq^{Up(YFea;AM^MeXPm^4e4#^RnFzx75KT0izkLILdkxz)H9O9c~}FFn!6)qAW{W
zkTaAk3+&iDWRe32E&2@t-b7`pE@bi0GIJ{EwMkJQR}z)3&UHECsW6pGPdfU$eXg=~
znB9g)CSJbhF%L>Er1kp}gW0d)UP!k2WQJALZj7%hQSEE^xR)iw5%sOvsTC#4mEdnQ
zS&ZNlyV^Kenv=%AA1}L3E4~oK%`7Y!yhU|6U6l0kxO%7jg>oBycU(1wRDu%V{F=|?
z`IYGTary%S6$G_gh}i-EX-lZlVHO<|;A6=)<<Rao{+#GK>J4OR;>SS`b0W#f=wL3v
z)<@1de?jb*tDm_h%NNZg(!~CrgfegEwj4j|wLAd;_p{*L@T%X}2l_{s<6*^nNrb&O
zkAdw9n#%!pOXA_=RL@T*Yb7%G{LXW@h0#{Z6aY?fp&W6$sWnW~^qkUU0j7;(?y~*L
zD5EH|U*_gqhY^q9ectBKyAQJlP*@k-^deaXBvRUqrz)di8jSNQeWdOOO&w2T=%1pV
zi9W=~%K|dXj}MN`^>o|&SBPZBDEC=P`NxF+VIy?HgTvCdef)~}H{b|?6#^}QEJ+HZ
z8R;^7I`PuKUDL>C&u71HcQW&ybQ$Ef4AK@<HIT0rc`dt{795iUE{4O(sMy@EXs&<S
zKnSwc6;L5PKBvDvg|gEfx0zp=)nf*Z#WSx{r&oQ?rJ$eFG#_wtG>vONX7fkDz7Vpv
zb_k0?FmdWWho5a6-TsvKZ3$aAG?>cB3axr`r8Z)ZXlbAqZZm;IG>nDfwN5!P|9q;!
zI}$Yboc7Vj=@O}6&rb)GL!49Ufa=JtPe~N0@=}8jm!N#Wmx8m1+Dhf_2-ADgqZY)}
z#hW6$QqMSTz1<2fT{u2@O>YS6o#PMkyxYCua$zWxXdx+I@lci89$L<;WNF9J?8DLS
z7l;RqowPm&<Q~wOEw-%`9-tY2x{1$n-?lVEZbvCNeFn7{;zdp1bC!7kXxfvByv5x#
z=-+Q;rPL5^5MHP+)V!8`dDiLOpiSqw*S_Z|i3r9L`$D-4tEud3d$XO?7BMDpiGdli
zo-hwo`doOBa%HVp=6vA>?3?^Tk?#C3^g)Mom#p`0W}s62a}?)%S)mt!wNL}+{%#SJ
z8fgGls5GHv?6GVaDodGh@>uRuGs#zE51!W*)+5b>mfG6Xyw~pMeth7lb!@0fR^uB~
zQ1a>q2&l$6^95uG)^58dvKymmU~Jxo)-%cP#!<<Ayfy2;GE58IA0BbQ-W>E;7}Eoy
zH&Zpg&g(uRUAvuNobP=vjvI2<7bm$hV?hgPmqR5VI#of1cpgXgF~O2?!bW8y{RhzF
zE5ryz+wpvSdiRPKTAmNJsH^g^8*6^5t<Hxu0KTkYB#B8KC|({3^&8pzr~xrIuFu3^
zT+1nVXg#LFzq||B(tJ*LJ#l+Nl-NwQZa8Ry`t63(KtfzFGcf9xH&WiaRIzLMmw%Z^
zx@dhz0Oz<B6zpzhitUTg0v-k#QgF^EIT8)?ZRB3N+gq6nEt`<{hUPo~6m+!R%?7ON
z>Z=zLeO|IIo4+_dQdbA$cQ7!1Mkt(VJ0*eFU;07j_T4NB=Q9?Sf*4p$B6*-}Du=jN
zs-76-g^<JCkA1fE%d5ue<7?2rtDIlE)GW-o%NpV)wV9n9&F)2j#FoT%CsCcm>esuf
z2J4mA7xjq@Hbx^P(qtcMULW1=XF1J=Hr`~s>PlgNhs2VIyc3k}m}KgP6<oeFd)00@
z&Rcvs)jz;y8QKkmT@W*JAb?*#l5WsktT7HsCj7~uXrxVRDMne#>_%4hEqR9$z_i56
zBXtsHMMx_0)-v`2)<-_9O8gt{97X1uU+B}Tz~!Zf_ya@cz?5w(&cM>V&Y0`Tl1wW%
z-fUa5;}2`Iy3EA6o)65_^JJ0Hd&-7x+BlWcKp+10_78T{R;#g3L@PgTHFC%`8wujd
zyD1(9zrh&}b)<*g)Z0&UV9H5XB+(~Vdi(6N5d5KQLM2JMP)pVLYlp2qdfOtPs9l^c
zgS!pkxdRP&s?wWgYm<|cypKmi|Abu5<bK{`D=pKPjO#xJ@%X&R)R^LD&JKz6e7kFu
zyr5Tr@yI1w5iwW?N0a^K>}2V{X}m$w>;FEgheenkTVnc{PVJ`VL#I}jRZ_Cyxja@J
zH3<0^#Kw*aB?9&Oruiv*S4VvNg4M>NsQwxCdl!-TU#;0~8ItbqTh-Ex_b@cbqtL+w
zcM4;_XkN_lX`W@puS>tHZ&V$ig!L6_MvES&B1-j?z{%u&(5Qf99BUeuGl*eqW<nCJ
zgZRzO%(8HU#F%P|Vgf;v^Ybu%;sSo%*#A+tTOmS}@RSmqLNqbdX?BqRf%eH`g=jEn
z7n(>DX>hc-p#5Sbg*tGCbLBmdx~JC-SROb8Y$br&-R8Z)eJ5gX-2R|noO)tt{Y%W>
z{J_7f{NX4e7~9c+-;&l73r5HIxw^@H=%&#orY&fDVPZcGR>&X*okIoU_pA+a`X$cI
zdA5EN{5z!kmnjAFLnE=P%Fl;#_hbAOedzoN+q+3t7aRl25G}z7T3#eh+6M-IF+)}!
ztdkv7SY4Gi<t(og^xw^}zfwtp4N<c*4+Lh&sLuRC;{0MMNJPSD;2F+jb17^Ux_ICC
ziFcvq{z-&5Nm!Qh4ddbQF*yDIppu1bVtQJx<|gUAu_o^i4eZ};F*62Ktagx)(R+-u
zpawZu`oKg%Etp?#tO3Ei8S&OtkGI5wCSM)?$8$f48LV|Y`yU(|%oD)Oa?$<@!U$nO
zRBI5#(AO~2HhzYsA3l!!%=ha9F^9vkah~=Oll9!aWFP*O<^FbVKuNN81gtfUVu#=T
zb>pK86iE$Dt9FXL5Q>ITufND|a{NA{bGcthRi}!1{O#X=4-dT|vD=c;25;*Iu`3vC
zJP~*>{fE;Gj1ox{4UACwr$1n)Eir@3ol)*tQOf@@&?&O2Pn!i(2Pp2px%3SMSd2LM
zdSbLCEd7T)lPtR5lvY5|dRoOV%+E`vrxR|6LTvuQNB;n^eqH*Z3TNf!ray>~jvHRr
z27+4ZA^c{NL8#B?Ut+Yy9Q&xi{KTt6a{2!lZ5h;DxpvYlH1z)%Y`)CTa&}9ZM#nwe
zU!%nc75Yj;v`Z{yECU7+R?}XOgY)ZW1bqm!zKyeN(H>W@J?{UA;xF#~%Moq>5u~N<
zC0>csRmAKBBAA5&WtRpE&WwKeYz$BIYiEv?Edw<4FNw{UBhelNfNDH`q{m`77S3}0
zA2gan8OfbSaIVRY)an~)<JT>3;{B3@Kzu(a46L5xM7I+rD69;{E|lN$L9zz5LZ!sx
zGhHBeZ-e&oKLq=&l)-Z#;&7|pW-aG-qHSXj`|q)uLP%i1f2+nByPYUMMbT)A5f=D8
z%FbwXGjsFoxkP_aAhOe6rSBi-B)oq9Fvt^KkQ(D7@^7*J7+FukFBIT<EPL(OHM0mT
zsB%bTOKYpdpxJxbf8`;Z45*eCp8fL1IfU`C-1%RB{Rjq8{iO{|Z{48?PRQW>j4Rmj
zA5qNjP%GV#9Zwe{GxPJNOY2bot}U{hU<a30;hEa@7DyY}1tkrQelfKWH|kG95wr!$
zytI=dJa{N9V^^8?B)>?b`2->G5?!Uxmf`cl9V*sM$;ah{_^*$bAOhE{)vmWtUFtlx
z^e_*&fhV>Myeat|2k@Js_C;=<0!Y5N!}<jX>Q(&;4)~!ebk1)LLNW-x80CP{{IQ6N
zy8oT}w+cdY2N|9R@5T}DHQ}!|*w$X!>frEy(Rn5chD!f5fk)JUqJa?g_)YKku!LbT
z93f&W8#263G!ITrXkn4L-u$Tu|I56eNo19l#=TS@nA`SimR;1le9y;K)%nrTb_K%$
zn*MD$67T@=#Z?24jyg&FEcE+h)<MzIwl!v2vygAA;d{MIg)u40j0m41cA@?2Bngq&
zLLm|G35^TBe!Uo=`%<9)=$@fzz&&Z{nMgmcENB@~ArJb}pvQ|!f#H8cAbo;h*VQBe
z*unp~-etfI<_%YdI<79&&n&qIV=-unB_!NA@8EL$Rto=M$v-YhBtc3xey&1Wfq}m2
zEcFSN?&%?r{}*$|(}Rmu(JX4mSq_|u(W*NX!Tf#=GX+!~j_3=OxwUyVf9vhkP$7Nc
zJLLidX`cz2ze)bzv;Y)Dc{#&XKfY~A_Q7tz+#V*mA9z3PpLC;W{N(>b-+okrs`C|b
z9riB2pgbUfr+1nOo2_h@&z0ny3K>Z}x38-RCI8o#_;!-OIXO8^999B^*X;lodGlNJ
zy#C)A7IIY<D&<7e{v@cAnG(#ekGb7`kK@;)5VRp`)-^1&B^+@gMM8~=+v%T|>=xG6
z6qo>4XtQ&3Q?ZmXW=Bn+y-H~z*wnwi?SDOfCi@1{MAp?$g0`0zRWjwxEr!=~Xoj)I
z^1F){)N{?(#Pc3wN&d-PH0BT}8u3@h!n?oCFE+TrZ-Xi&k9(g54$bw^BCB%9K83z}
zcz7_oKA0_0XEhzmGAxiQ?C+PTU6FMC^!8s!0D~Y8^eww}c6OfZ*lDk`yd`p6sX?M<
zf_rCagP2nxhwl7&A#c~Mnbw7$$?N|+t4*lBo!t=#7k?U>B0^LjwaPy6H*%h+6j1B;
z_l`xkY9HSmEy}dU#dQ@FP^GXq(?;O31|A$(6qnS5(n5fHu>Lci_n(EzuZB80c;Fpc
z#7A4(;H72Wz0Q|sr@H4Nyhc68xUd<wla0`L<WfM5WH|UAi)j2ikg+p-)&E)Jw;07k
ziFH=Y9yMR}(8gBHDU5K6U0TFaLqx<6c7L3)oD8g30|Lk5`dqxsLwTG6Yt)}_u#)VU
zH>a&1xR(YDEAb=xjx_(3IvO(Y(&#LlQ68umJ1ezo8F9aaH}xvRb89NiW$n1m%5IC#
zT#E%3#)6kT9Y+}k;D65hJvnGeebMMf|ILX)M6iRsbgADvfwC4gi7L9$NZMu^?&pPG
zDo}0pmDFGQ(6shm76&RSA^Q%3ZBYJ?1^zQdf9M@QQqrvX{5H0Vx%jO|vHhjD^Ch1%
zJCG7e(lSmD$jV8Qc+8RP{*MAm39#x(nWMlg-anXeEF0q5NS63G&8cG#q3rA1p?Tvu
z9(lJGKIP<wNl8I;EnS#@doMW3V1RFCeUYrCj=XWK>1Yx=$8$Q7Wk>xQk3MnwaG4v|
zFG~(|_^r!>mj88cm`R^s*Ekg5g+o!;NzHT09pZO3(%xRlHyqpTTZb&7MxWU7E}fHY
zmOxS*=jBL0jcF~Jyzr2`vQ)D9@7j+A?X301sKR=2ZaojRmVPlu=P)UTvODz3xjxT)
zsq2Kgd}95xZ@cIkK?%1B;@P$tJ2UINJ0w3T%x{HikcZZZMM{1>X1Ma3am1jE&~mBG
zl3P$<p=gMrT2q?tC*$(sy9HvXyeGDostxsx916wIfe$5K<}m?kH;0~ox?2CmH~?AF
z?(IVgx{l8y;++PG$aX2A8l>W%cTElxD|LsLGLqs)qp*pq!2Q-GNU#(>vf-_(hl3mf
zEhg9dU7nRdwYryKSElVtavX$@#>Ub5su)|xfEb}H94L}18ffFX-(QtH+E?}knvN}6
zFuDu^cEA!<WZ~su2Ax%xa~9e9`-7+qmB*4zr#0HzN|3Bb6u2{&Xn%!riFLaFzw5&$
z^r{MmW^L}I@7>Uv(6wcX=9cX?1(j}ZnA~U&?TU4(@y#sbmoIB|3)|cU@ev?T%RAAa
zLS=-$VLm@UV?>IDE!*KXP!X%wR1llSeE6m>_Nk<zBrhce50+v0%sXi*8bp3Q@7I@K
zdbJalk_7!db+*5DM?b|jlB2k)Rr(5UhckY^m7>Co&a#v|?RD==u}=%JjpS+|W23x^
zvz|cYT=h;5vFBeGioc#?1o?VB(<G^l){}-wb)O7Wy+<gsRB5RYSUTLdtbe=PfRe=K
zaJimX>9q+9u`N$R5c#pFaxI>1c21?5BrBAhNks`P2HT<K*&Y*)V)$zU?De%|yd7ot
zMncOHUP(p8@DMevpD#8@6@STw`fV(L`8}aWH>U-`B?sdfK@{{M&N;3YLp@gS^6IAE
z4B;ACkQGxW7UP;N^>QRLtj<19bUYEUB680X1N2k}9E!ft{`a*345E)Dx?#@9mbsq7
znCaY!;n1MXjJ<O%#aN$ix<9t^;4pk^GGmvjqH1a1(W|am7C^ADdgY+w2Wp?Uha9A-
zgQ`Z3_DS4JmrxATtqDLTi%J^Mc5sL`ga7tjws<944MkBw3lDO6=NniA3eegnL^b%^
z;X{=aqT=_*M7r}+E4sPbVwl#bi>L=I0hgE#p=RnZoy8nD$_~F9-*XM+b=aNR9vc>q
zN^@4MiXp%5D=u|7^9HrEQX%h)Mt40N*F$d=>XpdL$+3^7bArMeJ4#x0PM_Re?cV(N
z#Yg~_A}N?}{;W1-eYRL{SE+-QE$g_q@_J=|3E1Fqq|uuHYU!c><Z-5+yKsDeNLV(K
z;EUSatwC&URg|v<?1p&v3MFPWx$touRrsN}wNL?0YwHp|hW=t4k0^wss<ZV^>%|m;
z<=qekTXaZ)DpYhiDYXxe0&!o^^~#mGn_lI}^<<rN#=_;c7PUw6m;!H2b48r_lezO-
zyHEQZZz#egL-I~vmT%2_mLeeC4P^(ftpsCK^+<0%r?=D>w?L+~n(Xo9-Egi{*+33i
zo==Hi<@-G5STbPzl}Z0|E>w&*ai~<{<G$ZxY)&&e$3fD_lAcLbVRu$}+*>X%FKMYP
z?>6V~*z7zyqDM9F{mgw2T01cOg~A)FqOFA~gmn}VHjLhCAjF6ZzHx(3Bxx{shxCLx
z^T7Zr0#o!4^OS}i!X09-TA#)3g2!*F!O!=yvPIV0Nq62x8W86--;n&&{P3>lYo+If
zWnD}kCd+Q3bdx>X7@`@E<VASAjp<h5Fv7w<N1%SO^%vF;%7PI$O7ns3Hky%WpEgG_
zp8ygtXXyXkKK+o^>B~+t@(`~)U*E2rywzN8{Dv07LVQ3&&b~B#czjNNc*-GBN4<9I
zCkE7TtdNfnKuaRgS{S7{=S!o82E5~aQH+N#6vRE%&!?GpYvrB9f7)>iuuXCsRwf5i
zHita+(+b<x@^$TT#Hr>>`Ktou9~rICH5*-^XXvI8LUbml90q&I-qT)rE^;LWV>}2d
z=f?iP^<|Av`noYJuWb$pj9FJslFqOZkCf<Wm{Etbu(r-Wwa@$)Bqs|4j;n$_pUIia
zg?Bk6v3;eQu+%h&ol}>3lwNxeF08s8Igu>wtB@v2^d9EAluAT?M@0pFKU*4fcJc^+
zw1*#PG$3&k7cl<=ib61i2uPVRN>^WeX*gd(E|gB|K4n>-Zj_jxmJG3rDVDP?Jme+1
z?o}gRl$x(4COk|M-U(z+jg6s$mI#YpAF@Vu3$U+S(irdSuVS{y0}{k$(oVl9{XA-1
zJ?OK@zoe}Ex@!AV3JZT8VP%X$6~^?xYa0;odvi_hRjxIADEII^o6h{2;3!Y0GQeD$
z`@pEIeXn%uy{Z+~UW<uDAbP2faiZ3VeVOroq@i0yGUP){DftW}cobC5WN=Nq)LU>X
z#q}7immlAntCI3g#b4TbrlqJqa!-L9JpJ)<e{zzzI5D!3AA|XDg3jqylnQ!!vC3Y$
z<6+3wafccSHoo{X-cHN%13F7*8}r5Gp4<FSyOOaGu;)dm(g!rAW2b0C2};(M3)0LV
zLNrmO=2#B2tc{0lb?_${5RjL?5>j~;?B<mrz&Cr}jYmywpkp5*{L4ZO@}4})7YA7a
z`STNHk0^KQ=Q&swxkKrGV1=XW>Jyju{gS@JW@2qier{q}-Bu<Ye<4yhvE29D9WBFO
z2K5IUel92bR-GP3Re_q1vCoci48HNx`32;`v;PLrK+dGN5F3g&8xD)&1ue@&!$T1d
z6}mzYoOw!wHIx$7JMC7zW2+Ow$2b$iWcovnso`3kEhXwO8T~$9j?>TY#XinX?$y@`
zQ|;M%pSx1r<;PXK1=;6wUi?&2bi91EkEHLeUcU3T8x@nCjeC$l2<K8qSGb7HlxUUs
zhLEdT9jCO4ZT$V|(-uCZLx<y~yzE%HuN~5V)4v~aAmrgO{mq6B-9K{hs@n%tU*)(L
zdF|yU**sQb`VavmxAV;xd_m|%@cB|f-tY&$P{NYb?8QTm_bSgqt=p1duO%a=8K1>|
zn^I4(IODywH`-Lj$xtYB(y8u7r_AQp8|2FPEpg`7_li+wFgLlP`o~nbm8%EZRNc6T
z$Td{}fZ#h<Del;U%?t9bZdfdNim#rEUELvE3XEth&bzI>&lS;p+({)?*k%?v{kY1-
zhNIALt8{(Sp%x~pQ0TU29M~KXxC7#c>cM-Plr+r+DfB(|H#^Ax858*fYVwut)KFt)
zlG*QVq1w*ZUp!(Bp{tplJ@%-VVVvKy-~tIT`h?|=1tOz8n_82~)4qvoD262bU#`ag
zM-^cqSa658qat(;rPWstTDEx+6|9b>x9wfX1neZzMjLN-69O9Mz0R~0ma(t6?C#!}
z%ZsN}xg{bBi1nc$S7ni*<V{jD6tuvh?{zAVD~iYS7Cc~W-llz`P6`>OP4Vu+Ras^W
zT$;DZX%ee6un$;~K$W9Wl&B#T&H1z?ljyL^teoU|$SanKe<N*bjy=Rj@Lx<((6VO`
z47V_!-sUnO!_R3i9w3I}ET!jpU}4i>bKjy@U|8YoSe{hOL<8w(1y!~3J>ui}usu`g
zfO}pYg+8wMpFpDDS9H4xp}fZZ^Gbq7xEM<T1=xI9fK%>`Wx6fDdv*2Fp`-4&`;76n
z2gbgrFS#1d@|Pk8o><58w~HtK`l^VCY==3u#THXWp^j#4VpXvlU3lKrr0~pWv#M<F
z=hbYxv5xQX4#?j^&w9nR<U})XVQ#3rc*fi~#ab?UQ!Hi;CpvCoel%99IK=gueNpnR
z6Z{vU&bostvy^^Vv1MwTiDSn&<O!|FjT8WF2^dfwt^wPejVNp|_#@s02Jwc1L;)}Q
z&q*_of@_lNy-29csHK)Hq6)?8689wrw$|rn@9{kFB<0@3?0*kin<^%+-Qto=Wxm1M
zO6Q-sAvrLM|7fd?ACbY4Y@+Xuq&hQSTlKXMC&3EGAY0s-QjTv^jAy;yvN(1;%4)eU
zhACE{spCD&eCoH}`;=U*@0DDcBR(xU?qs|DJ}u#f#zmgG%V)G4MA#4?U^v|W2xa^O
zK4AU<YovufW$cd|i%m>c0(+KZSXVBPa@;~66P&I*tlkl4yY(m2gDeb?kCX3iI>GeY
z1v7y%5=pSOL`V)hw?RIU9=5t<XYyVp)cVj{DYHBknB>wcQD|j2>*5`$_Ue5QVogG>
z0RTQN!{ctIB1=pB;33s<7Ek~_F1Hjrz6zxDY5Rt2GY2(1o!eV@SVB@lz?I-~hVJON
zW2w$T)6C3CuX<={mUYE>d*{Ezc@N34U1g&Mb$=PjI|<W}n&J-h6w%*D!_@a@QnH0b
zT_XAKg4$_EW>D7l<<Nf4nfmY|gfl5pE%G=o(Lm>b%cRhl=RvL(WW@eqQ(**w(9yp2
zaJ64w)L;Q6dgudbFXSTFo%Q@EkU_tm3B3;iou`;?XymTGyyRzfWCO!^ZWa2FD9d?%
zD6t~x|6}hf|Ki$WwOfi)T#FQ!BE{X^-Q9~@ahKv0FYfN{6nA$i?k>e;uy<(BdGC>)
z+kfElVSY2r?CiagtgIyKNtUN+UoAeiV&>O4xR%P;56Rln)0g(u7oAT6oDfPfo<{7U
zsay$cNelV<*N9>QgeXTxN?5KUpsx{41NHbo(R(_)ouX@5=%)$N@}lP84UHhlyXHlN
zxVzEUm5oXK*q3c)-BCiihrlgc_Byv&rG7bCn~+Zk^T)J~%v9L8kK)xBL<eJG>HwCv
zuds>e@8`eXo+fcC9086<4}WG~{NaHX$$zw&Px?m^e-oG&>JDMS-{ARdv|~bI>9Is2
zVtF-CkV=!A+Ll*Cj6PqMR<X-ApODtrIAYB;P>WT7fpE_iaqWXt4C<r~#|kg=#Sb}P
z)*{P<LMmrDHfzd#6=7NuTYNez(fDal{oVYEH2o--AJaX6SIoClBb=;{YyCdzh#MJ|
zGC<~9(j$;TPe)AKSVZ+(2`oIpr1d$&Se^RFcFDqUCfNCQ;tq=_qLceDraiu{$O93D
zdcjW2_tfUQeD1}xnW?6R-dxQMkY9L{VktZTwt&{<`Po8jvjcm&-i|5EgywI<b^0Er
z&OOIH=QUZ{qcysaAcUXVlFVl0S+FxErBpw_u^<J>87ap(A6;gb*2eDN!hDn}$l0P<
z0z`jR90yjupbxWKhDE<Q<4G?DX2E4=>d-&EdQMzS5aNw4E;&?1^hY6)^mPp~wbKup
z^ACKX&MoDGI88J3$s@HY<uZ4#(AsaD;c!n4;nu9P3wQfoZ+jLcyo=<_w<N#{?rb%h
zXqsr>-@xE}i&Tik9Y#-4@O2A;k-IsuL-R;si6QtyiY;_W!*}?p=IeCwviv|YU4JLk
zTOo~XOv=jDC&kKUL50-lh>r4Tu>#Bj`hr-G9Bk+N)3^ZKtbn)9Q`L`iCEPTlHgqf?
z4z=q?Y20@vqXZ#OEn7-eufD6@!uhmyba^BHkw5&O<dHWL0SyMC-y{uPiD=H8kqTV@
zW@pI|t#;X3pHI5@fXt@ct2y8z?f~)(mv($wzb60c{KVS`7zh!q6<Dq?g9ieymK@HD
zXXCW&xHO&T{K6d8M0|mJ--@a%nD#(l1@a5~^D6lOr|<Uc3`D$onSiiG4ss%=Od_Q!
zh{gptV(}KA8U2RKStl-awU}597>x;;uA5F{Tsd&&0<6|hl{rA6T6Mf2k0eDZE*G)>
z28F1`Ad$jTC9NtZKgt<XNaWMr&W{pAVtoI(0EKl*s6wA)JbzC<zpf;@@w!G<!vy~(
zppm*q2oDL2Fnu}hBZ+7>O3w@~K}1B8MBVsFCixls-kOd9Y9X5kO1zh1eP)tNAQ%-4
z*R`m7(65psuUWm_V2D1xi72Zf(bw&+Nswrvi)mmTQs@n-Cy;lqW3qqL**C>y@oBTh
zqtU%rF*P?&KR$lzvcI=ekaLOPzWeI?TR2T&TwWa&B(F9-m1QQoAA3*0vwbcjINg)!
z-hD(y-&D(08$%wHklu~ftdJgB1Vk^N-)qg?;VlQa(2br!9P{&7Z17TMMo{r0q4L5R
z$%@uw36J0Ar3jomvqwN^X-PBc@0)B-I9I!l@`_#f8@IzM18o*{H(#>?`y(nej^$21
zdefeWoL{tvz&H9kW;;X<HAK}D&;C2ZKnjMJR9_?bJcwh^`(FvSMI2sOkP}JIZp&$l
zE~@5WHULhicdhA_B3{dZmrOM*$S_$f^$5N?VzOG;-qU*XBt&_6i*nr)%JfH0_^+1D
zAM0gaO-8YPxkU7mIsNUW=C*Fia0zZ7FiChKtOJB`TwGFn#vVUx;TmsjmsERnqBZWs
zwdN&p-E#rrsF#zs4=Y;BGV5k6Wt^YE*b17ODN66rvxFJp`p%<mE{q{K&+iA{6lvJC
zvz9JI7!(KcA&|X%PL{Vx@}NIINctjVpRYJKwj)*&$J7XBHy?>!ZNE0j%G^LU^_#n#
zg)`ch)za_j0gj(!Np69gcHY?XkrV>M5ju(j+D6PJRu);?GQPD|%hJ&wLh}+=rJx<E
z3mybA-8rXdOekx<?`Eo~?C)c>5x_X)`C=W)?@`>m1-l8&qAK2ft$u4lq3jxq5qstG
zad2rlKxu`A`n@-mn9kZ_ZXh+<2>VC_Lc`Cn=x?tw{-B2k(k=e^vtQ{H2DZ878}wB_
zzU-->oP$@z8j7Wx9#ul$!ULY_%fK`>*~<;^KvP|meuiZ^YPz#3f`4ML1<;z1v9X07
zaek0U66?ibXD=B#S|3|u+K0U?SVvpFQq5Be`+|R@Hd$Q|T@Z8h^f*;DbNF({D7h9J
zB;=(+1nqMeL#uI&0)ToB7X(T0kmEB@VB(Lkq=zSktYUMvFwqkRx(L)8&N4}zuJTTW
zkl%O)d*(5Gp-cfpKXKFq1>h^myd+TBjL;DUZS~m@l6RV6Fty6huo>r?zv_djPRYdH
z^~q9~ND#;%6wYt7K~bDhlK0J05iy2F{D46wXx;7sav$8y8U}_7M*E#muPbA$=~9Vc
zk;tg~7S({eJrT1ozc?G?ns8<zZ!z#irxO%dL0V^MY@xFR0jAS#WG}drhV`O`OYc@3
zT;J#y-RqhJcWND8pPQ)7-n8ks8|PlbI5OPxximD^`>i!$lk?gma%~TdZl=bep$q3T
zbQqYuuhk{d)|kbfM=-z2g##->EDyw3FJ-9{3<{D<h9$Pb4EqhydMhd2`;AUz>e5Cl
z+R!1Li1B=a;XO~5niNH-VF>sx#yX|56=N9pfv`|TEh0=T8B4}gMl-gQa)XS4#s{O;
zbfx9}xzxv-Ho_}Pz_rzVS~bU9qX2;T-PEi!Fw1JPRS!(8i##XRoo2_3et31@z|3{D
zqyirN7`-5T-zMQK9d90~${ZjLjZ;5jbk#oTQ-i(g9XrO(r(z0qCy^)%jCD8pS8i5<
zAMsh+6UEq?JkV>bmPFl!=g9{60s#HUpXHcX8*38C%~W(`mHG8QyyZ_BzEg?ASNH9^
z>pUSqM}0Yhu;p7{Qkp*C4-V~kapH&3Q&3ouN~kN5?k{VvFufr3kb!YI8@`6a`?*9M
z$sGqJ43uob&6g2--C#EwjYb6wbVf_GB@kBN+o~W2z8*vo3-28gcOxwI<08NiP<ttQ
z7kee3LmN+|Urs{xdfcK<761|H7lZUKBqeYMKaq!P#si!a9Cr{$3y{=LGE4U+Yw$XH
z!auFwGZFti&;}UymU3Qvy~E9f*Yu>4OyLL;3b)K2R<E7ZmoEO8V_TjRv$+*J!oKrb
zUI%f5y2V-Y%EG1fWG{TeK&vXI+GA3w@geN4snG}Ld5aSWySSQ3*MFX{G#8)-1oFE9
zfK)*=IOHw~WGc<rf+sEWp|3hto)JDUlV#z^OWJwl2?NQ8=P7Z3PF1E0ok*}&5R^@Q
zuSN%k$z4_Y0^I?O{l$u;Oiuhl9A^JQOw)IEy9oA2+#>qnNm-?4y$o`@#;>}`>n&M^
zA}Pq<A#M>0H|1{zwbwgW6%@84t`|FSHJVWeq+{_*o*R9%4#Bvrag?1McL=r%C`cq+
z8ipH9zpABS>~)+NnwzdJBkeQ9zan<8A~|l3OK7eM=Bk@M-`L?r<p0(MI$!cY)&cJ9
zzGs`}fQsgJ9Z>gf&AYjDvIa@adPSK{p!Z~+`m~}Ose!MK@bv_1_S1$a^<)jDFxF&@
zvPNzX@y4Wy4pw6q%ltybbsi^mGkED=SAsv@>tG;>q<wV7^oqN+VZxgIs6}l{C4M!+
zbaXvs)G9OwSe?Z^<W2tqg>A_$w(>MSs({|=IEZtRAng|T!S*7k5O`AjE+U(~;~@I9
zoajLESv4@tUBZ<U?y=j95+pQbZ+rd{@R0UcIAVwul|Y|24lFs?TkB1wJF)=*=KO&Z
z=VNOp#*bX$>nJY?5xO80x&~Z{rI#LpW2-bKUqllG*sR610O_QfEj=dR#r?VJFH4fo
zoaMtek0R+C6z^N`H^PsB5XU386C1VrHOI-0Yi&_=TsI%KTJx<E_CgkNK`=(DMDbAh
zURgX9e(ZW5AH`&CdEV(DU@HW+Ld_xQ9ob9DbQZn<CX014G~KXZSwY4cMut=0jr_`%
z$U;?X#GMizlF?MD(?`f7S9e&8y%hmTSlYxD6~PjSJF*GHEWA49QD;SS*X)$KcW*M2
zR*Ce+5Z{#QjeiQ8(LvYY=<@mOM_E};dP*vbQ!GlL)pvy%)wL39unA#7*pj1BM{&E}
zxtzNh5-Q(%Xc2s*6J6tuQ`0#weEPjdx7;G@Cs5^Qn&i*dFAUzilm@$(W#$`$A9mDI
zduwQCnJVp52O;EVKe@(x@$@6c(|Aw@jjh4Ma}?LE+$_U;ilsf?z*Zmn67Lh^zzGxB
zEl3{&@t@cEkKzhJ4ZS0ouOQTRlU5j8rid~>OS|9mEA*O`UNPtc6zD0s#)Iq=T@THp
zDhcm%m{Saz(V;*GQh?FjmPY_B*B?fYBZTNP>Gdn2FOs*Vq)Kz3C|XD0c06=+4@>dA
zP`UaG#E~Y8*c+rF6e?tq`8wlZ0pT04{3;@d*`cj25g)W%-b!-ag0i6$E2+CE%+8)V
z;BT<V3CaYK6gn@22kTHxD=Lgv?xO6Igeg=+PEXgoEKKC||Da(9C)}8yty3NhJUe3@
zsECb*=$GA@C|mjHYqgeztd$`jY`Q{5DqilFrtu^th0CRL^TuG;>T&D{t}EJAcOrO-
zcoMG*RYWTa;8}!3v|&o}!9WYV&<$P0^8s&&6+vU0>&8@jo#7ss!c`~SP!5AyY+>@`
zxa<(Z#rPnieRPy4^oa#A%~G@{9SrB=m$<ia>NuPKXh85B4Y!pOM(h|uH!sJ8enMwg
zPTHK0GjpyyH8*pb0|Y+f>Wn~k(p*U-F}TA#KJHM*v|DgD$35QK;wnc{W>#Q4!{p=F
zgU?5}_=70AtoNt-u~XJLFV0;(fX8PrPlkKCz`LL29u%a`nlH&oTa~FQXj=(1h;&8a
zf@t`7?vT7q{zB`00Xb*LA@lkt$Oq36WR65Qfy1>V=?I>BQhML2rusw9VznrRsHRf|
zqM1e9u=ap$EPVpT4q#xt(JnajWnl@%&D(wpJ$FAuE#L5uo)jQ23$fS}KP0<DOS0L}
zhO%3dz|aETp+U)K=Y?mw;|E{)8w<^{B)+lC45jCr@@O}}yd0O$HzgL%#aNKGNUEDi
z^(@Zaj$6*0fn9iq0WC9maZ5RZA*>JIv)gnmrpEf6lI^k5EdTN7Nx-|nwgH(pk%pXC
zzDJx?Z)zHO!@)i;d!<{_0nuAiL-!~Y60%qi`rRbKD+|)dcf}@Lvo^?Yztv%k^MMsu
zeUICmTZrXI5inSr39Ef`I8M$M35hvFw2naZuYe4Qj1mRup@S;pxBh<Laz=y4ZKgiY
zBwzK$!r&o5zSh3l+|1KI?$Dz;Ws+7MH+dfq9(2I3?Nl6XY#8^e7Oc^2<pyhI7p>hF
zLh<`ep#K1yR9kqT6M{Yw6Q?PXf^AbDH>*o5njIoe&8D3a(zlNFbjp3PK3Ha(t@{X`
z!as4Do!`=;rP}43WlV&C0c`8lu1B<F@fr0w4)+iAlCZ24438-77Z{v$z@@W^?l0g!
zn}|&X_7#_R;n&({v7@*yzSv%RaH_Grrpb@;1rVyS;j4ebs1s&_5yenVZnq&Iry%Qd
zz4n{``fhPn9BR3a8ZA`5z9`8>Z6QCl4m#oEb`T0@MqFV}SSHDc*+PBHRjIo-CrN&4
z=h86OIT72{&8>p6B7Z>G)!16^%-F<?d$rv$im5tBl7pT9Zo#EG+1Wl?un;?$+#G*Z
zQXpm_1zCrB#378LtbrY4>+JfX5OO!6P4MC37QkcOzlw-);7e%0-3hSuZ}#D>`4;M4
zhj%q!iwesVi&|TmWK&77R8nnDCbRis;s81?ACspxz>OT>cV&b<*_#Q9vS^uWhkEV4
zK~<NVwCh@Gt7<~{^d<;~o~E)=b{#sTB<0iX^2Cu!m|U!BG)z}`cU(70-rt=2pV>T`
zL#W-F_uQ{e)kw26p@JyxKjYNm$&`Y|QYH&)13E~Jfekx$x0JQ>w4WY%ZS(1kB(mUv
zndHVD<?G3L`y?QA{nqYWm^*)<to{O|hUuwqL%PU9$m8-W_vGYglL)i>O*4Zoh%jz#
zHXNGKu7bLJiPUu5TvWo#Fon;c|6B*@L9_(W=&~jFAUID44G6iK=7UP48pB#*^=lo9
zYqwn9!&ecz_z~R}ytO=+KZaHs%0J>W=)eh+%|9XwO<+ti6$;M{^o<tBR=fmh+6q{_
zg9_OhTi(&i`y^ea7)j;R{vtGuImwjtbB?|gSET8_>8`P7yc!&nyrV4zBc*vJr>zRv
z_@YoG;~lG#U~;Tvss!o=sl2o%b!e$YHva?f{L;b*F=b&oMkst2*$i&pa_12Wmz~!g
zMzm#|T@)EQur}YEwlbm)szSI5V+aozmZmDXx8FOP%u1mnwd>(8sNQWG>=1u%Z_R;`
zuAGi!s*n_65+nNqr}7i6^5d0^8AQrV-lJ!?#zTlbyaReKuK4;nF9`$|6w5O+P072u
z8YR}RAJR+bn3cB7;q(#x*$d!vVNoGAp?qKKHO2ep2q_j1+NHZ9?y>e*B{sKBGRAUh
z*ZF3nw1Gqho2G^q%TNczO|l*lJelKXM8`Gz=hNrpCr>J5AFk>rM5LPnYUDDBNJ;~d
zT<(Oi28k(&B+A!aQgWpmlb0!p{nL5TWpX1$cn<+pHvHr&8i*30WaN<)H~Gh;==QkA
z8HN6)Egvcb4-knh^YdTJ)RO{3tRn?>m)FA_adil+-nhrMInbrr(_?MwUJA11YAEAS
zXrjP0ABJ>Ru5@0)=waRiDq^N>?xc?sSm^Qsuf-Zm%?Oro7Ky@mq$ZW!ZV8owy8%37
zPb1mHig18BAip{3z&f6NT;5E#Pv)e?(mPW8E{RH(>SIA#fmBA_Q+b^Qx>Iq}I^9>O
z1}I5z41!m`$np8pdP|r|Hoy&rkAAKNLsB*{pX^MBP+#Fkrg`ZdMOtog_a3nJv@B@5
zGbE?(eSaBAy4O5gnmt#QM`7bP`9c62|H(s3z1d&O;|P{5sLlZnG&aiZv?MGGQ9ZiY
z-fFIv-*A}Ip;i?IxGCF)w(9aP=0hPya8ETs!wH<P3L0ei$0dJbg;kU;ZK^J^=nhAB
z&AU2uR!X~~t2=P3IofWuz7@?7tMqpAx?YZU9f>4FBB)-yUv@7+$uCUzf44zqcLV9y
zi4T2V`!--%MSAHDequ^KJkJG=eU`;J>RoaqBnG%Je-l2Kci{^#lCZ3Y?+koX-JP<g
z9v;j!qe*;0`z+I(vtrF=O0r~)-a@?uz?RCN;T(_!Rlec-1W$Pq%+UZeS>C@JZ4gR6
zA$K@IvL~2Mx=*HJ0iIh-prVqcQx7G5h)54CXrjk-`e#>-=M?8XQlFs(nKtLif`if>
z5~}iM2>QKOn>8sO2$bl8<8f{h_Z2UjWA|P)VA%dbd*y+aYWs+v>nid(2ln)l`zch*
zWeaxs;|;<5LQ~r8P3`2!((nba@Qhp$9K!O_ru;HO)pMkHZ`+w8RV^JJ&&RJHjCAW6
z4Sv>Khw%d>zM0}BQ3-x`vJwtoMgTKS`oNq#SoM*U_hR1Dq72`qB<~2qwz}@9Tlu|x
z)6H;q&u$35x{e7LF^GRMuiIr!&G{9!EGE4x=4CB*j!J&`dL+iQWGYV#BjePxTVQ4e
znQ%#d7LOtEbeKi96HCX!=W=c(?f}=0<XOB%;DZzK*~jpjZietMh;dG;+EoDT_ub71
zyM3Vc3kq|;jngV19_Nlnausflp$D*7gE{s~02q`#u3WCnsxmhF2^MOxkZSPNaSBY)
zKLn2P>;o%z)81|#@!8E++GEWvj2U`_PcTz=z`1Th?I{D|DWAx08JdT7W87U4PTwu0
zJBdm?d~r<XWUVo)YyW{4{~u7u50Cq`5lF*`d)3S&4B2baqLRu}sqv6j^X0QOLUk-M
z2yZ!3gmrr*L~#+em^!u6tr{iKqEvTLoUe_I1yQr=B?VboA%MHPz3w>WN!PNkiCS)3
z_V&#VOC(_i6<wx<BXyQbCfmcwT*DmU0tVyLB8s(4c3fe2GE$%uPsA~keKcr4oD}aU
zV!Y_LWj<g`mni)m7w?c;yhoKJ(Of#VYTPwtXD4Sp4h~$lTu@m3Vh|x>370EAMf>{F
zcXSYwucL$t?DhHM*bOBNnp_MOpse<x4HT+95*1_zzX(X#a8*$pUGTP6rNqu1n8(~-
zI}!l###5YCl9lHWlf3WQ4eiNqz9cL0*rN(QzrqI|ewkqtZ8QlrnUwGN9d3TPCB%V2
zaq{L?yud~R8OOthu>1OO%|f(ZSWIy^gLbGyQh{0QDqyyRM<B4GvM;Mo7$pee@k&Ik
ztUFJ+m0(2W=5{oDy7$jrqrb8qxHQhx#B1~%6tooA*qj*@9e!!Z!_{(k`3O#kfUyR;
ztWniYga-m6%I<KK@rkNr->o|#uQ2PfH^C@b!m3nN+VvVo29Y8xg+AKhaFKtM>uKPH
zw_kAp_rqaGp$lvXbFNg1d1LSyTZz<-V~Oah$_QvVq7>_j=I0+y@_?^bAY~KLc{p>z
z4Kb;wB@p@1x@b<fb<k+pD`7ua^_dxDt(jxJ2~5~WyKBLxZyQhGqoqdl{AjrGD1DiX
z7Z|EyXsE(8gFm3LbILV8WPKcRr?;T2Dqx=JdqxW(4$2;E(QN*yBtuJqw~!*(yD8j^
zhb8>|5aN6i6@Jp;gx*Y_@rx^WO@>cT_1x7Tp)6Xhxfqw6s8T#TNsSeZ-(7OleKHlE
z9-~!GmEIW9<n40kSs)t3%6Y+TmO&ishlEs4Mxb5^gP^2klTeshH#BzH9Ec>~i=PSl
zFfmSBEqyfjR*`Q<JzUGY_)2Q0(LIlj+F7gTup8d|7kPs8DaiT~_Pu)zK+(Yftv(%^
z)E7~uh3qWz_w=;^z)~D<UPPzI9A!m!L~hD>X5|#_vu~5Rqrhfl%%hk6nYXqD6&1Xi
zEsX#www>b1>DyQf_WYd7awd7<Ifv{^nItyPg$5(Jh%DFxl>Gzwya^^GNTBosb!d@}
z>3-tOor&;}3*7{)P7Z`gj47jr6EA$FFJ%{u5~%mt);VzXk-3%9KBm0eDjJrsKxrJh
zu~p;2stgYHU8GsVqdI58?{}4Lk0jJ<suAW6A;TmIS6mO$Q6%<*!_Bc)rp9rLiJ8GJ
z$jq(KqAiB%Eoj7-V9dfQsLJ}3&uzY0wIX}^B4xaHZRZ1%>_O!ll>M-Y>e}~-PP+)9
zWya|bu6M>(!t^ytxY;ADS@}j!`1^Nv=U;ZiW%1C~QpK!7<1Q*yM15Z?$$AB%&<X0?
zi~zk=`=kBrdIGA+kNA&=dQh5_hDsIwft=MP?(7JAS-9Gbu&u(bc5r*q&D2s{B^wBl
z{&8$l9uYqw{6BrxKVQ3vR*i<qv>(o<bG4*5dLE|EDS#)vvLgq$-S$s`5AdW<L1C|w
zp#lTNBFoa%d=vtYZ`ICN7K=@aTPe<feg$o9W*rRg7iDFsE=ZiG;XA-x^p%?dBgv#V
zr>_w)%38&xh7W*K(p1{P5f1kT=l7(;!w^%K7+Re*dAN&LwuNOI`Q6EEY+sMM>%qNs
zMS-h|uJgi}$&a*v8VH?De*jrq87^>D$;cvAE=+UL%YS=i^davcNRg{2zJW1oYCad;
zl{Cri#`^4lQ5u0X2X&uJ-CsmZCc-oLzQc)eC{av&*%MZfpHpFpCDjj^5k+qch)+eG
zmM+ZlZ*wvvzh)b`d$K?N#;g0-`PJdj7?bk=!xV_K(*J}yrK~A_hs5jR?}$_Y4Q?Vw
zDRm@WnsuouTtKWKK1r@W)p|}GW^EI=AATX1PcJ0Qo<&()?pqWM7ku;SX)pozdJ<53
zwr+$716kT~M(1);)>Df?Ky%Ya$~}l2JGi?~@m{%kzPdZm6dwfQU0rR3_2Yf@qsJCZ
zS`Pe5@;m|mz6jWnIJVkD&K=wP;kX)YIqAD7+aqZ=(Z^gW^$7qmQxpr6M546G=zMNU
z<V|Jlpi?FB9>3pb`h>@`I$@J>bh+1?5BityEan%c?QMRR`~J`7;j`!Pk!aQRZSm|V
zhW|<<9sG6lJ;3oig__&)F%ll6)31%9ti{CPa&wC`{~Y28P@mw~zc5Na3Tz!QY|E-z
z>Ew;1V&`~0w%mL-J%eY}RFj&zX2DkqjA06ki?h+S)FiQ(-spwq^ipU0`@5hHXI9dQ
z957kp2RY%JtP5x@)DWLiP!l1Lkv{M6OfGe)JR6SmKu`maX!w^t+T1Q3xr77;Se2>9
zIZ#yDQ2{||r9}&Uy-qb{AK7!aR|!fG>XWl9W6j@mtOo~!PD=uJ&&oX5(|V2Xv{Q5Z
zzBl-G;4FO!>?IrU{B|9nPgxi!)A@yD-^wfpT1jm-3|De`1Cwmz(lxwV{<;ONP|F5s
z`UFeIg-n2+(8hFbQnbhsC#NS!(ym9ecg$CX_{d{}=AlFvyqjb+m%rrT-AqLqc|A^+
zV`H6=^tm&2F(1TN?IJI}F6g8>b7cZB3sN5#s>U(rtE`*nXw;R-{T3!I>-+s`z=xLU
zI7m8y*SgX(kGphKndMjfX$M)g!@Z~8t(?i*LV!(Ts8<K|25tn6)_NP=xWBTeIzV)H
znZ7e-xE6sZ=G|q+>>fu=j6N3dEbOc4VfZs%;P-*2=%ZSj#>i156>YjRpH24fHJ|Lj
z9u8Ai#`>EW10k_3*F=zNYHI8vvNS~!X*`nI-m<%re9zV3f4DcSn_X7n2X_9*N%?v;
z(T$!uIQaD?n_YdVhwam+0VR(n+pVo1yL)%&t<2{aK&{|bPfP5bl+wxteG_GKbAqV$
zRX1AI9~l9o6c`j?lsdva{HW_f!ox#Oa0`U{cuI)O0-V&mhADPAK~x-&R1Bs?@F{WO
zr)Junx5nMZUB0Vt*kCApRU`O2C0Vag5^av(pqL({A9=d_=cc-V*dMaB%^pqJdet4X
z*!k`Q`zD#zmYC;f=`5iH3umHiavvY`+GfKj@he|m%PJ~*^%U>)LEZPdGh3mb4J|7b
z29p@`U6SoD+r=+0BIJhkdF+4S6%x!3nSZ;)KmO=2<Jc9w4xM4%HIINO@<@!f{Cz!+
zb=?Q@HSjnH0f99Vya+c$QkL5@^3pL*M<mr<x;t4wwaEcmuftKTR;m4hmU(E25;(MQ
z)0E%e+d}+wP=GF>l$k<Hk3OQ(=040iw04*7YY;F(ypF)z-r<bd6%Apd<k<N}{o{(N
z-KV>O9fhwpS|V28cfWA54S7&*^`S-?up`bXW!KjFk8m9@U`~JIA$_kYQ0X2>glISv
z&lGcY%+530`aOkm2t#}tJM>#3W<!`cBVG8Nu5_G+-CN02242Im9Wo3Am2XgME-cSG
z*(}S97jK?pxdC7;!uEYIssRBhH8&p&aQi9Dgbzk>6}R|mD<^m5)cI6ZjESHU!N%|K
zG{YAr2-;qWSszJ~^GE==`Gn;MS2uZ46)T@!n(CsyunQSNOeQsDJ!D{-%<3nEL+=fu
zIIW<l%#RxLlDKJim5L4y_c(Jrnq;E-P48GLC&b>Wy865FYdlH#DY=Jr!KwQEAic>?
z)n=FK)#qEoDBsQUoG8V{jX(R`gCoB4lbYb)Adkr#40f-EhEDE6M9Sgy1-0yeHuLO`
zYoZHcY61*SQJ><AUW<dGHIN*{*Jziu97-4zMf#n~FZeoLzGDAX1`iCxV<!INewm;G
z{-v!2HM(xM&itTBaXu&rVIYsYJC6G%;NWIZ8J!~1*M3(nck*jlLsUm3q~e*wpg7uS
zUJ$a}g0n^y9cN+j<vA}XI&i03cn{b=G#v<I$z2THrtWtKqOG%1WvHjm&^L3X624p|
z&^eAs0A^YrLMl;D{iI-2yw<*y4cdTuukod?FRVO<9nBU#m14om7hi&N5z*s%%%vIj
z@xxa|7~g*6*Hr8kd-pRH>Al9=24GDMe|vX|zJH)BIe`1|lW3U5yr&Utt~bamDt^06
z?zQG4?{w2+7N4+xAOuCe?yX*8^*s^W&Aw3L$9;lCmjS)`>j^bz91cGH?UD2#@ek)K
zAgCyNP!;u={%lu=nVn{Ja<em<Q+9`TnA}!ORyY&Fob0s|*<oaicC%wQ`lK=&QL0m{
zwL!J54MD&nNpe4@JA`pXta($aOq$KNXLVwQRs^85XN*<aoNj4cI1?K}XwkdX#*ZBC
ziRZF+SjA;P&%+rn|IhNkzvbIj!@aa1E2b*B7-|e$xrPNbXumNXr3}__fAs|7d0QZ>
z9UUl?k$Xb><HT?Jz1!9ZaSD}?L-_pZ6hHcIM<%Cwt_j-S?AtCr!VurG52jkUZRb7l
zO^&{m2>8TV!Mxv&OQ7Q91^N?c<2?kr8JC>JgKx_?JE3vEx8jX9oLc0W2eN@~Unmet
z`f%;AALKsp+DH^+29|}$!;eu<AiXkJFN4NUg$0qtmKH~&Ie`xXPL6UydRUrMGGN&}
zvvkSz-Qt8$yJ_vw(FPUYe!j;HL-6JG#IccfI-YPtJcALFAW3+x`#6oova1)|)4F2t
z85$v~28kNR0WPDG#b^AI2<#32%7#kj(&7o2_ZaI8D)_$5ceiJ8)ZAB(Yd@eU|Mb7{
zq=5(o;K>Lr%Q0r+E=$8!4{kMO3&VBXN9cd<g`e=Z0Vq49Isa^-2<)7gu=={DkB`kD
z<z$PzmgZgscl<od*=CzYmJvTYDQS*yCp!uW6(rxs@g4#8-4kiteIaTL(|J)g29Aq6
z6;;bx%9P_OO-;)^iQoczf%qndH)sgCsqtad2APbubmX$f{(MWE%QX%`u)w>w`&q{+
zB|TJ-0nX}?@oex!L|T4rampGNq69kxry||OGG8OCQ!|ia(}E1ByM@gE3(>y`WuV&8
zm-nK*u_qQL&h1KYezdMtqsN!+4;{h-u0?HkKquqb&oF>`tyG;s>hlRv&6ve9nnhkA
z%Fw6i;WbNaC=KC2A(%dm51-b{x#CY%0y>eq-<1~J>{`x9aI|W@a<%Q)aOSEb0iU2S
zyCca9F%MYO1bxGb7o<*8-=;EDy(Vb)3Cb=PG(dq`o*WecKSW_;rB|yNAeL($k0p2z
zQ4KD^FQvIBPTBa=3ICgpzX0(grs>Optg|n4vr{X}gY8kqzhjQ~$awz;G;;G3aP#ao
z9TNBFWx;JaV2Dx}m%c+NbS!TS`>Uj0pX1k-Gv(bG4coFg9zEfaPJK|nJ=15qaFlCb
zho;h}A$m=qiNMS6<)Y3`HZd$qs<+sWF+Sq@C@_BAZL$vxr6^$D&(B)uzZCMM1AxA<
zAwe?7h~o`<-629w;6EU$MS5;u487-FtLXDu(Ncu$qNa9?4<^z({}GP1byLRrZ$9Rq
z8Ci^W(56!QEv!iUBXgP-nCU0JTbv;kS#;1!M3EBi4^;xr8Bn{78+>)Io(P|@pM*PF
z`q|GTltBTZTeCq$ew_(%dsamH`4&yh080%hIlrqk3%44^nyHmh0_aCGMX0wb<#-c*
zEFt7!EftZO8E>I|>yvGfn`!6IdxcHzgD3M{zw>53V(cgtQVD*(4omeQ0z-47=^>rf
z_;5c8+iy=^aCIXmPX0NIPWhCotH>rt>WgNww3%5U;)e+ypR|8S>E}nnKrb!Q!QKd%
z8YV@u)>{mEH!mI%khA5Onh_SyAW8%pB@+6%EXpTP-)17Wu+6k1k!!M1f<JM}cx9l_
zmk=R0!*Ny#Jp0O85ZrtzICZ-$<D6uXzKFwoo3L)JH&)-A#3{hoRdw1;!E3exL}=!}
z?ww>6<!3&Sg;20efLTEhkVD>fX7l+XW4H5}59g!0z4wG0lztc2AEY5JcXTFO8d?y1
z&meZhB-FCG53pgg9M@<O`gK<c?=#Z$+0M=Q@Gqm6Y_6VXHE_n)A!Z0<;NCi8PM}z!
zXEOsL{CqrBv)6heY~DhHQ9w}{I;+!?{fO9HK>U(lY`Ly~NKx4~`^b9Xm|o9W4`?hJ
z0*7sXMN!2nJhGKP8cK(#X07@~5QP>^`6MJ~vDi7V!kS=Y0%G~CZ#0tpiL`!;RUpBp
z#>i)m&YlP{<8jbCSu`Ql0lwnTPv7tovf;+3XG<is{%iJZ38u>&Xx(b1RmoFt&Sy4L
zJy&`_Dky=+ELaoCL0+Hid3yni0Q>^TX<ojR2AYHuUnrrU!fZewyAa5iL1w+{Le301
zbbPuo6gNJt?8Xk^6M&&i6nR^d-`*E$8+3{b(JSvD_nrXK!U{Xotb%rbaGWJ>9(8}Q
zz7~coJ5NsdP|iZVdG{^G)`XSwzdAYn+=vGzXk|V#mhhGIs?MRiNRH2=jqs++fRa-b
zeK}}(44fy)mN{xGqklQ~R^XfX#C69UQUl&<8j#(O2^ru&Oo10XNOwE}s*zr=EHaC^
z&*UIPSS8Z~z|-B2L^7f6lys9Dkq$!-;?@M6Be1ExZ;}tPVjB|e9?9FEG1w2kou&!0
z?E(YUKe+=z2jeo-wy25ER0K)-O#TpQaW?4U@(EBl&<Xzw59zr87!EXfCf802|4#e!
zCG(yq?F<r3KU51Z4>BY;d46M|oaU839~)LL0pi%u*4w^j9uRLgpqELI#;^wS!~V8K
z@>Gxrbtmr5K@^Zy%!OUKY&kt-ef8uKkE1fY^+7b^>oM-bOcpXT$Bsgh>BDUV!~l>5
zyAXt;ZqTltvb)rJPZEkFI4~~lQq8q0zho(iX#dx;fC6ctt~{~TeQZWMyAWLCuVU_R
zY_*A$Ur-=R5r-t2QP%3Y=|gtb=T}1$r>NfmA%YiS1>$c<zM{1T|KnL-fSQ9|fWC_y
zgCDd2cW$EmpzhH~g}la>!w0xp*s*fEGpjxZX4ZFB9{p-97x!tL^JP32gD*M|q*Yds
zXKY`R$iX6t7(20H`dti=#nBg_$~epb<Our9i%JS|ZbNp)2M(6oK|(e{3GROpr2jCW
z7(6^>Ww2yqJVLY!<UH4FNT3=}BAdRzQr7*WgS_KZE<@!Ih{w{VGg-WK=wY>Qgf4<y
z`5ghzFbfBQKT6=q$X+JUpz<5we=3(637-DIy9P_R(_7`mt0nSF(+2!!=O-_FsuC5>
zx4}#yj7-(9K={%d8^zrFDW}kt;2*QbF{WpyWnU~PZC=76g_DZZf<*#jlYP@7)Vn4o
z5F@+@V1XRiGW1XCf}}DMk$!KJ?JL16{BnJM1<w5eb`T-a(WIwf3dk`3`b}TZl`x4%
z(w?#RTS+>Ca3xJGOpuI)B&wpaqO74IaTgkE3#WhrVyMtN>_{q=*SdaPbg_M?1m?UT
z-h9B$2AMJjYv?~j2yEob>5U0p@ySx%(V7g9zR-jChiL!lExiB@7Hn&IuT<<%a`c4{
zxA>u*6<DU}#c={(8K0R}R}LwhnLCz=VG;g?sj!H=z`stXy%vvfD-3>#1-~4F?aKH^
z5XIvKMDLsCgTYxwH)b58FXR0ehw&m{Ue;ZfYPfAxNflgCr0M%oBUlsS!9N}xn5u%}
zMWPej2g_2TDIaV=)rs{v65&OXZ=MrjytYP(z-v`$^RZCIR>h}(0-CWtZh5Dl>T@E>
z^N2t$Nc(Bhw+!WO51D!BXMw@LhdL1GHHoS84$i9nq6%d$LwZY)$M*&AGE;VLV^fdc
zxL2D9KCq%#q+t?+^x5KkVeytjRsB|Bm<Y+|rkm$y_G{zt!rKj@ubLzV{>kr(cJWTz
znCk@oN1+;x+cOf??HJYNM;8l2={B};%rHGLvnLpvDu2RgK5H-OUy}ruzWQmQlh>TO
z?7_$XPr>$@&@*__p?6IC$MS;@oTjfceWqY%CFEEOsw52GQ6&Gm66JY^Ru@VqrLf#>
z4d;g@=omjY&#As2XZvAWf$gVepx$($sp@24{HLb)dwJQzT0tfXNB<#S83tf33k8-D
z^XKA<LHUwX_S2YNIg#u?I;Q?~LV>TVPgdVYS~m)hLA|v_Yl*OcfBNR-6`>WX$h9P6
z2vG<i_}T3E&B?5`+pY2RMvtktx*h}BlcKM@8FZmY%!kp?Aj+IpEX+k~4f-dV?!?M2
z3ilNY6k-iTFP(V{>lQC<wUNFa$GJjjH1k}!FHOf~gMxrUBH+<*gR<=a)6P*ZUIIT$
zK;DEb`1;ZR_=?Bt$Nv!Ue!Q9GYjd=l<<m@gzRJ1;tXRjUM40E{`O{dGS8H2M7YXt+
zi9de+#~be=q?6`F#nXO?XJY3XJYL-#xIDw8JYGCsVGKMoLegsCOs?yApP>GJ9gxxo
zKD;|POOgR4k{@K_r9nU_w<nEFD4stup!7WnN|y`rlf(b}%I9xAXJ=?K0)9M2dMGHU
zO7*zT0Xgu>dC_zLwHFN6DuHS{p=w7~ioc5XzbWL(rUUM(O(@l5ffS6jy>FJ9^?9Kn
z%T;7L3+SJT%18sr1j10Q5<DVGItez9JEr}WSwHWq1|(9szf<~5e{~<sS6QQZO@o5h
zt6KvL@dzKtRc}?q{70BgH~9L%hlKC$@iE@YP(KFyljT3DNP-o%!kfv1FOl$gwG~6(
ztSqZ;rN#_^stiF_i9NGhrxXU|0pExVj8etWytGw!Uifc<15%0S^v?X=_@T^!5fMn4
zhh+8ZorK<+8+5q}i+%y=^KEE`K|&u1*sgu3DH)iCJIh%9Dap2EG~<h<>150L9w3eM
z5>PkC)-iu|uWmjp*@r*I5^v_|9V%O%rQcsxp}N=d<Kh1KD9<DF76R#4zEUbT24&z*
zvmY7#V8-}!)G}0K(|b=dMUooGf8oGiNO+s!ud?5nWpJI_Ck7=r>ofjKYyP2sf7dyn
zh!S`fYqKZG(cT0||8t0c2+&#*blnk{@F_Yq1F3jHfvQvS0zxtFnQpo}*u8g&d#~2$
zzcuEEAS@woN<Az;cS=<Ad38(VlSls`sX^#%fCy8hqCx07jr1Smn%GChMl1As5#q;I
z2@rf9u?5B0T!Mfq(G$zRhV(-(p0x)><cU~Bdu%lGrdXM{O>d@OQRD}+K4_&=%&iRb
z{)KaYQ>&p=ukYJNqX4yOgR3pI{-U@a&pDLNv$#@HH25?;idT{O>mhpasN#8$;<hF^
zQkuV{r-pHLy9|52cwU7kLJ$kSrS#W=Yb%C=ERAQ}H`9uI5n{##iuu#Fe30Ua=pLFQ
zPxx{Ff6EfG)VR!%$4k&Pv~KyCb39nkkd?-ryt9VCmR-Ef%XU<HDUKa-1RdHPe}PXw
zSl4y|^W_tz`4iJay%^EY8~=l)-d9K>EoSH%<68WwmkzQH>pukt%!-20_UXwgD`Vi`
zh<ga3I~e^O?LRJm=IxLdkCBnl?%{<M{m6@>CNep%?j{V9(4Y64Bt>{yogE)X?dbtk
zlSnI!`UQdKiHr}^)fRk1O-J&?0%WMy|7R%A9dsA>i`Yjzm3gW|F^zG(Vt3Aq<o{sW
z6l7NkPgGm5Q`h$?9H8ZMg7|Sa9xR}}#rv!tKm?;k)T;t`qg+R(8Qf^RG0Q*uOIeR`
zf!NpwXV)PYDX5fWx`Vggrb3q+FN=x&;C*>Im{@z=94?07`upAX8P;D51URMvkeo}2
z++<1c|N6!MC@Ie!B=B(cw2B$|-~KjT@a0Q1s*Zl4|GD}<*iju0oK;tE%s7dEDf++8
z3pEr3opFL>^FQX-U%&dh!2a(+|Bs{pr=$NpKc1(e8cMwFOEerD94ZGvrr*wIo%fKR
z!@@@VK1i5hZEJXSfBW{W+CfJ5w==Qr5c)Md{HUd0mCQmiWO-ki8g1>1ER_$s*oyzT
z`u(}iY=3TBUDauzx{dcb8+LZ0BL$&6vy4O6K@o599Qmb5{EG!?<3q1^I-T%2SWkv2
z$Eas1UyZO2{0|fK=M5pNqj{cqo_r3DUxTT9p60Q)n_47qvH?u(b=+D0hiC8?CZF@K
z-uuBwGFG$1rZa7)7j%4J83z(#Fuo*$dw38<86+~n{Bn%=mn(kFK5JMoU{}G8u~nH3
zw4u&=uQ~W~V~KdK=CPGo0dmp5Te@G2$h(2WdVeUPVJ6$eWHkA8Ig8!NYPFpd{;zNL
zSC$<@C$rgnbf8Gincp_>>RuhwVR1OTp+wC0|HW<oK8$!cZ<bkZH$wxq;6z))Ar&t{
zVc}jy`>T^kLgg6k55Lzh2e49y(kD}W*bkL;^k_OZjG@PnIJ&!MGAuep5k<ORe!4*a
zUsC^mGnB}ayb*5uU^C*0|JIhSjo=fP-Hr9`!NK5r_utFTRt9h6sF(H8*`~Ulu(ulM
zyc=BfFF)Mv__YfC?rLub)cF2}hlJ0f@dHkh;$Gbf<i&Bd%UtBZ(!sTZrLPBuw*L-3
z{<$vJ`-9j>agQ9K<6P>!44weu9c)S3G7+bOkV!jR9#$B{*5CY{<G*lJX$C=T-gA%a
z)pgn&y`oI~wm>9Qjm`B@AVQG6DH8D0_55%3Ip@OzSMF-K&s8cgCScMjbp*^qs4(Dv
zVeR#<&=f(|iTXFs|NbBEHz<L&*EVq6Z|9O9Un?<_25T`Sl?3MN>x+h-lCoOa*%3kY
zO!&qWtWOZ2Lx6W2sawL&YPU9>Wl;QHKYBplYwW_fULV1_KfLm@?`n7y4acC5z&@v<
z(Fa#9$3TDAAaU@WMJCPTN|nWWb+tOSh~~Fc(^c?B#(Nt8F72q@4Y@9k+j_9M^HKX3
z^o>4(i$;c&nw6Q2^99=IG4>~88P;c01=fRrOWg;6y*#x7-0!Jpw2f!=c0^_gcLMF0
zWFt!)HxQCk%miC*jA`TcJz^g(*k!>tdn>mc3;#`9M!C`V7Hgup1;dS=g^shWL4M(#
z3s{{=_lSt!%|1#5?|Rg}y~(`JplH(0lM|836k!b={5_3NUcviEyaVm%uxCT0rEfIb
z_1N&mmZvH7KF5I56zs|&8J}L3YMR6h#%ZF@6ssbas@GE5IXVW6jEtPNr+$rmKWUOf
z`MYtBeQt}PQ0~|o?A+T&L8sPy?T!xUyoEvU7#oujyqY*hHSra^^>_{Bbp`Sk7b9B&
z^Gh_plt99m&)R^<kQCqyg+C=_7wQ$Nnp_{%SdLJ?&wRWhQT5kYUymAetis;HlKO9@
z{b|&A2*&P*Y&Qo!q=neVD2MiuhVFFtZ$w}09_|iFiKjL61)U|k9cvk`U)2{PnKn;O
zi@Q%z6qGuqKum8bHN6%kK4sJN&%Q^>VWudg793`9Qa=l?bganq-4BV!e`JVZ8BX5n
z<0=xjll<qj{l_dk!jBlnGD;f;g4JZ16DYV-aH^(Qp?4-Jw!e$Bi=zxr?}QYmE>Lq9
zkrlbs?Zr^<$Fj>#6{ko7<Mc1;$+$x#^hU?G-s1jru$5%pY$)i7-bU0yT%K8lq|)ds
z9V$jP&9H+~Z$Ff|e{yV)MNECp7W^;D(Z+$$Hws(F#GO1W&OE|-i|q(i<Z{fBBwv;z
zu`l@;w6_5~N8I4j!1iYDA(x9Z*o6F6Lu~2-*q~EcsB>JuccONs$-!b*Yf)n6-GIP@
z0kBOMp>oG<ON)Va+++D(t$H+%??5{_$l>ncYyB)6iU>Qq4am5tnAp1f{-~B0w?+-u
zCyRAEoe5Pte9Fqo-<nK1iB6GoJ|iy#Z9*3o)I5gvH0188v|>vCM!-5kpqQh8S|vl1
z()uiyyRgiTv*}**+XD>%B@2P0**Zg#QMhBz&g6#1y@Bx3i%>V`Xw)dKv<2IZVFA(X
zB^J*|J}2PW6BFm2&TBimmxVEs_MU}7>pC7+-6JRCq@I#hTz41nojtGzW<HdrAVhlm
zEG#w=8`6MEJKsp4ZoA^TVb_|^W6YF%EhyMNTt1h)G2NLpd4pxCii<{doJiD1prq--
zUdA+*S?_JXe81jx{F_CitqF|`4AbCzEZTvXJ4a)H%_NqR7ZY0+5fX7{u{TX=k$S1f
zQoeaLc;8OdefGX8U3$3I<V+^|gLq-I<$$rM(Q&h@t4g)8joP|Vny2D>Ryc{4GS#^`
zIifxLSUfl<;LM&s@Pzw%bqG_|boJrpuuN37|Dig$y7*pQsgz`~#@%d>K-BeQb#21g
zW+F&p!>9r_#~}w+<hVx~;&%&;F%b=IF9+@SMY4yRs(XzqqRB~<24wo}y}2$Tw<uQ4
zF+2+9q`R=H<s2Vz+qIk>8&j^4=2_0lhPy^6a@@_qQa3us(nKwwHA=$9*{ILfmswq!
zoWyzSvf6U;7=RnFK`P~soHycIW2S`8gijz{j7ghH#4H2vcAaW{ym)JEVol?AKew~C
zShwO&n&w3LTJtS*Ui8Ne*^4ku$E67x8tZq5_Y0}z^<PmAX|-Gg0ZCSB$c?V#X}xxg
zpR^T3mD2w!xAf<90;Ux0mYa0W+{g(USoacd_p_7dbpna_ozFa)G8Z*m?WY9|mGXcm
z$LHc$>3=e<B-q~EsRmc+df88)9PDy8J+g$~N+6g>;wn557!st>*-5w+5~uCVNvS%a
z*)r0MzU6vKgm#(YdJMNvx1`n5dX<~hLH%^vaI&4Y0^1vg3xC(jisJ9@-<h&d>p#0P
zFDn*TlaXw;zB!dZzb0+<IWMVRlk2XVn*+J=tNOS1Ifbn`AXp698d=D{vDWe2AYk;1
z(}>_;qH0Il1-tVgMc=buqE9o!ke8splEEsE9Mo89UCb&-W-8Vv2a}Ya7mSPM@t-sr
zky2Z1u-zFjkxJo;E`DLt;FiMOUq)iHQHVvZ<XG=yLZh*JFiJ`-D5$XfrZo|xKQeM{
zY?fIg!jwRAPgi&sH~)PrA?4l?)F{`3fs<LUtEg9Z``Gj}s*B6l&O|ILiAu>Yh<7nK
zW|l)rL-9_joD`lU)PmC20;|He9`vNJTKP@7^n#-N|C_x7nnGJ~9X-^5H}~jAMR`@x
zg~5-SPsLQQuh0s&%h`L&%nC*JjTNMnW~YR+8gmQcXiUNixug<X$a#C8?zNv>51F4H
zKTaRbP@lKIq`tqOE5F)i6xGeK;225eL?0npu<y!Be*g4du<WpL;Jvt~p*e${y<HEI
z9&JNhmYsX&K#Z;;(l=R`SPCNN5TWR%uMXc~9L<-I8rk&UAPZ|F#h)ImG{rC99J^op
zLg3cd-P6pt@2%_HtQVDax?eA&y4V((NDz^yu|DT(|JO-4Ng6SnwIe<x8PL0M34(<Q
z#{VT`5bUkNRu{Q&@6OiXb<cuq*$|~><E2Q$$zs1#lI?U3y!dS)+ZT=kW$Dtcf)piX
zj+#YXwK~Xz;$w`P2r%xh>aD&60|Yo6t|-fm(`Z~z?nR*SFUUUyRkiM3et_uEc5p*U
z_1HfxUyHrmIug;Vq?Fuo@)gPs|3t&YM$4CyLBE8Vt#H%zJ%0TN2REBJR;!N4jT9=g
z4_8@8sME^hIsid=f+ZR>h~q9A_lwrZy?8P=ZW*1nzwK{y7H_PY#{-Tnt-zG45ql~^
zn%-ks!HbWsOiNQpc3Rb(ubFPp&;Y7vfu7c0`@vvYLm5=%KK$vb$a-ml*e{gZ^?2dr
zV5Q{?*IgWXDtwWpifa7v#H!2<_tW*P-{}4lhsV9aO(*V(1a9l&Hv3xZ(?J=DQYFw3
zdSkqK3-`U{S{8HjL@es+wD<A-qW#VcgEV$p{09c8+(1MI+CBofo4!5~r=!riPU}k8
zfpE_Yf!(7c$w}pVu}pf0M+-+5k97l=yVLKSNNChr3UJGT%;;3d&?~Mtk0&FpxVd{=
zj~5{)W3R3aLyW%!54P&wo=;2;RL^my{VrZhHx6ShlS5TN#Ep4{dA8RxdH}{tlwc~S
z92cF_r&P9kICMN^K(&(pKI?v!Tx$PnjeMYKzBuFzXBY%E7-l$&So57%_-(LNc+Osx
zBFWiS8k;@a{o~e?gcX%mHQD_IDYqDOe;!#1wOSL){l;h-6m=^g4>+X}mYuIVtsnWc
zk#0_SL>ym>Ms<5jqBmoBHnc2PQDbwta8Ru}Qnz5HIs*dM{9`D{p`W(f)=rioaC5GY
zTg=qISsyfBaB!;-Dwq1IN8c4Gmxw8qvtV&O=^52K4<}k5_{+@)gB)n0T)ZQ}5r5qH
z-+dflsSkBTT0p9tRnc~mI(eQBUF&j_r@6YDQc6z#vXCLeWUt(#YCLn<u>^f+k(hji
zkKlCGn^q2YfVtBtyeE0daYPfsDCeC*HyXj7ai|qC5?0kZU~3J1^ogMfkY?pCZ#3cy
zJS0#Z$$T8$<Kh!(N2XxP)>Rm>VfXdHF?*fgAb(t^aGJ+k@b_BKz~8AkJ?0_XAI{e>
zABuo5iWT(oIXXHzeeaa+$Ptf=N^8F`!eb3DGE%^0Uw*)AvYxcfsLR_a^dSn?YTIx8
zuWgk-Z2^xfjK=ihM_y}QdmRDH&d5hok5B&lbfyG!mH1fUX=KJiv3go=QUb(7dbavQ
zU2*J=ex7#^xE<D&p~4&<iN)`2&gU7UYxVXfmQT*p9$3s5t;G|s+S&=%LhBrS$0>3U
z`@xB{y3ZkwTbVg7H#bb?*CN+WCMe1RtVYwMQ|*0$F!Rm*14v+yCHU+C={_+-OqTGB
zXxGaiYE+otR+c+7g~owHbV58`nj~_!^v!vG1Y`ScVwVaA=m?!wZ4o93;8Np_#5j_0
zS9IM=D|I_j3P=_U*khUcP^ek1-uJT*Qm6>@%^&fIb<!{}U@t!oy6S#!`DB=r=W*jy
zPxq1daLE;K(c{iy>uygJ`*4vQ?R(3U4O>Jw*O|_)s8aH`W=GFJUNozg2g{qa=<u8c
zUjwjgUbO!1SSqaOyOgyW<`pRrHh+H4r|xpot%34~Fe+WES#LYk;N#Te%_?>~QVKar
zQ4g263x8tqsL|Brv4sU>G*b01*}@aC@Z10fx76a)3etkRHHcQMSQ3+}EI`h{@!hXC
zHU1PT(9%zmBmBR@&ODk8Y>(rrqm)70X_U6<N@){x6iI6>GPShQ)ly4nwHS@Hh^5j}
zORcR8seMW*jjaiSHd0%WRuHwslGGBRgb-U0UV7d+Z{B<J&U62~|J`%$@Av!O&-dQ@
zV|Py<UJ^Yf=@-xZ5)S&lpS~JEs=9pJ4H;f@VNU<#4!Nb9WW7NBjpKC2m=OY<?n3vs
zXoRB|8p}gEHo5czI^Eot8NJ0apUV(dHcj=7mm1j!rK#jR91aw^_3TpPRlHcMaJHdN
zAf6nOHPSZjc<b_S>0#XkcHJr9Q9=T-6h7CiYEp$9-bX#fiS2@Vc-m(%$f)@{@|UY5
zrlvU37Fad2xg{lJqFp4|taNURN*dXS+~jhNnMD$%roa5t75jfl>%RG)j>n8k>#V{-
zzI0cYrP$p2*a6!^PA|MZS7aWz#W1QT)WTZb@N33$_V4C<?HUZW7F?Z|z8`9E2xMIQ
zJ$Q7P4V>%soJ@WgfI=yL5K!0MYo}Ayq?rP$!z&!&q^=;ZrLi3{{F0IM<X}ru&gG7C
zIU84-Qo!0?12%6xHdbDpY-yZBY~EcJSfIhoEnebEqZcAnEA&W&{--3H6|b$a9v4=^
zPG_H$QSRamrYy+1ScPMpd@EIH*=(DF{8NF&F{+uXfixw4hg8WfYZt$1{?uG`vdImc
z+4wouAQq5Lh9q@>tjbr3Omea(=GxaB`ESpGLeaP6TRSuhoEqx_S<^alFWQlU@2z&N
z9F=;sbLU9?9OJ-+PmMZO#WqKdpchAkC1VSasJLUF&w9f8Ea@ChO|ZcogEtuB@g}<2
zsF@d?q-vevsMj7{x!f=Cusxe&`F#z*w$!83eI-}AUz0C<q7B1FUn-T^L48JVLLW^$
z=>Nd4D%Z5lI&s7ptNteEGZujj`-I?guEefgvwUe#6z2TadBHvFf7^zCiIgY7qCLEh
zC0YX0OS25frbs0XMSx{aSn1dr!mc=6si;&jE6OaMcqV6QwPiw~8{j0$<>jm8M>||D
z5vF1(bm4=TSdl1jyIvsDqKXfdPt1ZUFa+Q7gni6_RZ2U4nO~h|BO=x72C}lf7}dvz
zs^(J{b?GQWfryC2x|zwvo8c`^G1;@(+y?b*3#*Dk`{yScTAr9UfN6_4q09H6p^G+9
z-$fJDL<_uNC2ChRt%EVk(3HpvSf#b(Bj)c>NG>(Xm2VREN!ArE`9}Y>uz!ovf9-9K
zVyByB&F<)Q!{~5b|2IUQ?5Hj(nV9x5>i%~yByVIq<b2PE3q=y;sPR|g$sePdBY`ov
zOQ9Q62c`(?vvuc+?Smc#I|4yxH8c)0*14Dl4Rc5wK!?WB&z2xt2s=J@GLJGs*M~dP
zhT`7swG9H5k@a=?O?iNOt{3Hc>H&O{dDx@{CB7ao-(SnBa#@O}9m142y>l_c=fe2S
zX{`?CfHDFhu`?~CD=y^R3o;jsC3Ax!GWb1O2NikSVetDlPUbtryE{`K^aj&%zbk$r
zLhK5UrS|xG!+!B&E@3J(2ld_iTs7)*LmxLSBQqNPEwFOEffF-CmPniq(Mwub(!rMl
z-Mz~X@44nx6Af0$t$nOyfqj}|urkISbipNC7#oi)5jU?};-PfO@fz10Nj;LDk8IL{
z;68krV{p^5*4fxyEFLx15Gge!U@&GEcr&l5I#@kR1D<;U<85qidTlG2d*$Bq?bb3;
zVi;x_OtaK9&CK!0{@okqOj<grOpDwv$%|)#EAs=QNNI*izp<!`1Q;$kMP-v8$?h)E
z4i3TE`#;jt(;LYN;}a?}Q$Xf8drdpHMMTvDThwOFo5rp673>+j9&b_h2dg_XEBuj2
zR2$Fsu?Ywxfk;EO{(d$?q{0ehtIefa2h;eL8fTWY?E<A^M(m3|D96ekpj7lo=?uPl
z8+Efow8*e4llNK^<3bU)YQo>%U0SGim!_Sd?6yCm>w$vbX$Mzpyhc+3&N@;HQYzyx
zHhACu%FZ#Np?W8%4Q}tB3vF-#+YaTT{z%G!0vyp}ob;GfgSus_)!@Etp>R)cGX6SI
zV(mEdxefEi7{N3gC_xgv)QCl8W$HgG+od;+6=>n6t4?1zy9vTw@Kjy+p$3LSR@g`T
zGFwu$*Kc6x(%b0wmwlq`)B}Cga5#-q{B2j;0&nuy@)J^}F}&4gKYnE->$8eM`zh*Y
zh_CV(ZqHYX>u1UWio)*?VsJ$@-NDlCYJsq~Fo1O*_m0!I>!_{9B+TPTFG`6`O7Y`&
zXLf(>C(LCi?zyA*p>j;X<zo@i<SmA=5YlFNR7pPD`Kwj=qt@(s4ea7_l>c;Z0js4x
zt<VEp?C3DCpa}fZ#uyssPxiY(14d2O2wKtuDv<XqOxs&zjg!JQ7L!w$)MaO*^GEl;
z5WCexg=@teM_4M9UHfQ+95OjSXar455s8DE@-KqwQ>@-i&XMH<IE(}OtCQW59nnXs
z)rZhby;mMS*46A%<Ol~i7fE2NKP9m1q}A&d-jutKm5gM+Cr0~2%64Y~vsfQ`F3O!W
zbNXbB1|enf$iY-a<W$L_YSgd7tipD$YtPVr3X5wYN~fuub~v3^`(b9jmGxq#!tG*0
zIsi&}mi3>c{_jamc;)AYL_uq8kxW!!c3#f>x7gz2ZLTEw`0Ix}^3E0vRN3Dm1m9M0
z+8bh9F52{v=*Fd8?MHvR+K+iq<}=>k7CTDxtWnin;0m0D&06!<pK%5%JsJ#1+|X5D
zb}1LTVqiXp_TzfT82B7YMg~5RvC|%w^R%Qj?`RW;9fm%9ZNJ&uo<ZZuKb#jL5LG66
zwSK+ZG;2l-B+YL6Gj-Sbx_aQq0yjmI8M(Q+=w}A^s>zJSv^lOm8wKHOwh5vqVrpmU
zp}v@5_4a$#r>GMo!uyh+)c2>@L4H4hb4xzYy`F|qq)^gJJAz*vQB6fCICb4Q-tH0?
zAr}Fx+xXpjAqa}iUz{kmC0*DM+?CD0xR2iGTYTxxP#0Zcmc*uNJ|87?kR{QBmCD1~
zik`p)7IoO23;Q`B26tN*wMBA&4g<5N9oA{1o;Ecnif>nH&<E~<H>MH-8@lc))7Ia8
zNdcw4VfUz8h0f!#US?i_=@WVI&4WK&Kw$%F%yeS;oo3WA)3t|^<bo;k*&Bt|u5TA}
zEuxW6WX;g{AWmrCTQ-^3@ja^#gg#RbxRK!QmY=mWV9>qds`~Ln!1zL^H_W>^nZfX8
zFPFG?HYGq9SG)Ui#k77L;B}WLjYmR~hH`#7e^}~$vPi5zTvjPary>7UeW%V5iM>+q
zH7*=n6W>Miof`@ZSyPD6R+E+~U}r$QerB$}7pL*s!jR@wd!OMvo9^~y3s~bs<dBWb
zM5||F%1L9qj{Y_=_Sldo>K_686sIEjHYVw|v`wKMbN`7`-<4ET^giRPHQ~b#DFY&L
zxrI>$aUPl6#F7l5-O^&@ia*%QA$mzp_cHz{zWjnUWg4JdB&G^c)~A%Lb<*ho&q0*O
z+BqS#npvN4f(i+Nci%G7_Q@+y5gUPCk{y|xA9L=Zq)Z7)sfpY#-S8;5zCc5Ftu^8m
z!RhJNW>@qNmu1p{b|Z{m<w@V~X-a{Hj=RZ|G!DuIs>=u&gMtthpHA3CA;0!Ve0VJe
zdjy8`|LyAFe`(5!1jD=<maY-WDfZ<l1M94gb1A%9j>=1T+3CR0bh0Duw|YQ#*K+!@
h3*@_&N(iHYy(NxrwrA!9<p4y4!whC=QflNB{twG=DTM$4

literal 0
HcmV?d00001

diff --git a/api/core/model_runtime/docs/en_US/predefined_model_scale_out.md b/api/core/model_runtime/docs/en_US/predefined_model_scale_out.md
new file mode 100644
index 0000000000000..3e16257452c7a
--- /dev/null
+++ b/api/core/model_runtime/docs/en_US/predefined_model_scale_out.md
@@ -0,0 +1,173 @@
+## Predefined Model Integration
+
+After completing the vendor integration, the next step is to integrate the models from the vendor.
+
+First, we need to determine the type of model to be integrated and create the corresponding model type `module` under the respective vendor's directory.
+
+Currently supported model types are:
+
+- `llm` Text Generation Model
+- `text_embedding` Text Embedding Model
+- `rerank` Rerank Model
+- `speech2text` Speech-to-Text
+- `tts` Text-to-Speech
+- `moderation` Moderation
+
+Continuing with `Anthropic` as an example, `Anthropic` only supports LLM, so create a `module` named `llm` under `model_providers.anthropic`.
+
+For predefined models, we first need to create a YAML file named after the model under the `llm` `module`, such as `claude-2.1.yaml`.
+
+### Prepare Model YAML
+
+```yaml
+model: claude-2.1  # Model identifier
+# Display name of the model, which can be set to en_US English or zh_Hans Chinese. If zh_Hans is not set, it will default to en_US.
+# This can also be omitted, in which case the model identifier will be used as the label
+label:
+  en_US: claude-2.1
+model_type: llm  # Model type, claude-2.1 is an LLM
+features:  # Supported features, agent-thought supports Agent reasoning, vision supports image understanding
+- agent-thought
+model_properties:  # Model properties
+  mode: chat  # LLM mode, complete for text completion models, chat for conversation models
+  context_size: 200000  # Maximum context size
+parameter_rules:  # Parameter rules for the model call; only LLM requires this
+- name: temperature  # Parameter variable name
+  # Five default configuration templates are provided: temperature/top_p/max_tokens/presence_penalty/frequency_penalty
+  # The template variable name can be set directly in use_template, which will use the default configuration in entities.defaults.PARAMETER_RULE_TEMPLATE
+  # Additional configuration parameters will override the default configuration if set
+  use_template: temperature
+- name: top_p
+  use_template: top_p
+- name: top_k
+  label:  # Display name of the parameter
+    zh_Hans: 取样数量
+    en_US: Top k
+  type: int  # Parameter type, supports float/int/string/boolean
+  help:  # Help information, describing the parameter's function
+    zh_Hans: 仅从每个后续标记的前 K 个选项中采样。
+    en_US: Only sample from the top K options for each subsequent token.
+  required: false  # Whether the parameter is mandatory; can be omitted
+- name: max_tokens_to_sample
+  use_template: max_tokens
+  default: 4096  # Default value of the parameter
+  min: 1  # Minimum value of the parameter, applicable to float/int only
+  max: 4096  # Maximum value of the parameter, applicable to float/int only
+pricing:  # Pricing information
+  input: '8.00'  # Input unit price, i.e., prompt price
+  output: '24.00'  # Output unit price, i.e., response content price
+  unit: '0.000001'  # Price unit, meaning the above prices are per 100K
+  currency: USD  # Price currency
+```
+
+It is recommended to prepare all model configurations before starting the implementation of the model code.
+
+You can also refer to the YAML configuration information under the corresponding model type directories of other vendors in the `model_providers` directory. For the complete YAML rules, refer to: [Schema](schema.md#aimodelentity).
+
+### Implement the Model Call Code
+
+Next, create a Python file named `llm.py` under the `llm` `module` to write the implementation code.
+
+Create an Anthropic LLM class named `AnthropicLargeLanguageModel` (or any other name), inheriting from the `__base.large_language_model.LargeLanguageModel` base class, and implement the following methods:
+
+- LLM Call
+
+Implement the core method for calling the LLM, supporting both streaming and synchronous responses.
+
+```python
+  def _invoke(self, model: str, credentials: dict,
+              prompt_messages: list[PromptMessage], model_parameters: dict,
+              tools: Optional[list[PromptMessageTool]] = None, stop: Optional[list[str]] = None,
+              stream: bool = True, user: Optional[str] = None) \
+          -> Union[LLMResult, Generator]:
+      """
+      Invoke large language model
+  
+      :param model: model name
+      :param credentials: model credentials
+      :param prompt_messages: prompt messages
+      :param model_parameters: model parameters
+      :param tools: tools for tool calling
+      :param stop: stop words
+      :param stream: is stream response
+      :param user: unique user id
+      :return: full response or stream response chunk generator result
+      """
+```
+
+Ensure to use two functions for returning data, one for synchronous returns and the other for streaming returns, because Python identifies functions containing the `yield` keyword as generator functions, fixing the return type to `Generator`. Thus, synchronous and streaming returns need to be implemented separately, as shown below (note that the example uses simplified parameters, for actual implementation follow the above parameter list):
+
+```python
+  def _invoke(self, stream: bool, **kwargs) \
+          -> Union[LLMResult, Generator]:
+      if stream:
+            return self._handle_stream_response(**kwargs)
+      return self._handle_sync_response(**kwargs)
+
+  def _handle_stream_response(self, **kwargs) -> Generator:
+      for chunk in response:
+            yield chunk
+  def _handle_sync_response(self, **kwargs) -> LLMResult:
+      return LLMResult(**response)
+```
+
+- Pre-compute Input Tokens
+
+If the model does not provide an interface to precompute tokens, return 0 directly.
+
+```python
+  def get_num_tokens(self, model: str, credentials: dict, prompt_messages: list[PromptMessage],
+                     tools: Optional[list[PromptMessageTool]] = None) -> int:
+      """
+      Get number of tokens for given prompt messages
+
+      :param model: model name
+      :param credentials: model credentials
+      :param prompt_messages: prompt messages
+      :param tools: tools for tool calling
+      :return:
+      """
+```
+
+- Validate Model Credentials
+
+Similar to vendor credential validation, but specific to a single model.
+
+```python
+  def validate_credentials(self, model: str, credentials: dict) -> None:
+      """
+      Validate model credentials
+  
+      :param model: model name
+      :param credentials: model credentials
+      :return:
+      """
+```
+
+- Map Invoke Errors
+
+When a model call fails, map it to a specific `InvokeError` type as required by Runtime, allowing Dify to handle different errors accordingly.
+
+Runtime Errors:
+
+- `InvokeConnectionError` Connection error
+
+- `InvokeServerUnavailableError` Service provider unavailable
+- `InvokeRateLimitError` Rate limit reached
+- `InvokeAuthorizationError` Authorization failed
+- `InvokeBadRequestError` Parameter error
+
+```python
+  @property
+  def _invoke_error_mapping(self) -> dict[type[InvokeError], list[type[Exception]]]:
+      """
+      Map model invoke error to unified error
+      The key is the error type thrown to the caller
+      The value is the error type thrown by the model,
+      which needs to be converted into a unified error type for the caller.
+  
+      :return: Invoke error mapping
+      """
+```
+
+For interface method explanations, see: [Interfaces](./interfaces.md). For detailed implementation, refer to: [llm.py](https://github.com/langgenius/dify-runtime/blob/main/lib/model_providers/anthropic/llm/llm.py).
\ No newline at end of file
diff --git a/api/core/model_runtime/docs/en_US/provider_scale_out.md b/api/core/model_runtime/docs/en_US/provider_scale_out.md
index ba356c5cab63d..07be5811d3013 100644
--- a/api/core/model_runtime/docs/en_US/provider_scale_out.md
+++ b/api/core/model_runtime/docs/en_US/provider_scale_out.md
@@ -58,7 +58,7 @@ provider_credential_schema:  # Provider credential rules, as Anthropic only supp
       en_US: Enter your API URL
 ```
 
-You can also refer to the YAML configuration information under other provider directories in `model_providers`. The complete YAML rules are available at: [Schema](schema.md#Provider).
+You can also refer to the YAML configuration information under other provider directories in `model_providers`. The complete YAML rules are available at: [Schema](schema.md#provider).
 
 ### Implementing Provider Code
 
diff --git a/api/core/model_runtime/docs/zh_Hans/provider_scale_out.md b/api/core/model_runtime/docs/zh_Hans/provider_scale_out.md
index b34544c789fa7..78aad8876f4b8 100644
--- a/api/core/model_runtime/docs/zh_Hans/provider_scale_out.md
+++ b/api/core/model_runtime/docs/zh_Hans/provider_scale_out.md
@@ -117,7 +117,7 @@ model_credential_schema:
       en_US: Enter your API Base
 ```
 
-也可以参考  `model_providers` 目录下其他供应商目录下的 YAML 配置信息，完整的 YAML 规则见：[Schema](schema.md#Provider)。
+也可以参考  `model_providers` 目录下其他供应商目录下的 YAML 配置信息，完整的 YAML 规则见：[Schema](schema.md#provider)。
 
 #### 实现供应商代码
 

From fb49413a41fd22828fe306aadd2e2fea2648f6ec Mon Sep 17 00:00:00 2001
From: zhuhao <37029601+hwzhuhao@users.noreply.github.com>
Date: Sun, 29 Sep 2024 16:55:59 +0800
Subject: [PATCH 107/235] feat: add voyage ai as a new model provider (#8747)

---
 .../model_providers/_position.yaml            |   1 +
 .../model_providers/voyage/__init__.py        |   0
 .../voyage/_assets/icon_l_en.svg              |  21 +++
 .../voyage/_assets/icon_s_en.svg              |   8 +
 .../model_providers/voyage/rerank/__init__.py |   0
 .../voyage/rerank/rerank-1.yaml               |   4 +
 .../voyage/rerank/rerank-lite-1.yaml          |   4 +
 .../model_providers/voyage/rerank/rerank.py   | 123 +++++++++++++
 .../voyage/text_embedding/__init__.py         |   0
 .../voyage/text_embedding/text_embedding.py   | 172 ++++++++++++++++++
 .../voyage/text_embedding/voyage-3-lite.yaml  |   8 +
 .../voyage/text_embedding/voyage-3.yaml       |   8 +
 .../model_providers/voyage/voyage.py          |  28 +++
 .../model_providers/voyage/voyage.yaml        |  31 ++++
 api/pyproject.toml                            |   1 +
 .../model_runtime/voyage/__init__.py          |   0
 .../model_runtime/voyage/test_provider.py     |  25 +++
 .../model_runtime/voyage/test_rerank.py       |  92 ++++++++++
 .../voyage/test_text_embedding.py             |  70 +++++++
 dev/pytest/pytest_model_runtime.sh            |   3 +-
 20 files changed, 598 insertions(+), 1 deletion(-)
 create mode 100644 api/core/model_runtime/model_providers/voyage/__init__.py
 create mode 100644 api/core/model_runtime/model_providers/voyage/_assets/icon_l_en.svg
 create mode 100644 api/core/model_runtime/model_providers/voyage/_assets/icon_s_en.svg
 create mode 100644 api/core/model_runtime/model_providers/voyage/rerank/__init__.py
 create mode 100644 api/core/model_runtime/model_providers/voyage/rerank/rerank-1.yaml
 create mode 100644 api/core/model_runtime/model_providers/voyage/rerank/rerank-lite-1.yaml
 create mode 100644 api/core/model_runtime/model_providers/voyage/rerank/rerank.py
 create mode 100644 api/core/model_runtime/model_providers/voyage/text_embedding/__init__.py
 create mode 100644 api/core/model_runtime/model_providers/voyage/text_embedding/text_embedding.py
 create mode 100644 api/core/model_runtime/model_providers/voyage/text_embedding/voyage-3-lite.yaml
 create mode 100644 api/core/model_runtime/model_providers/voyage/text_embedding/voyage-3.yaml
 create mode 100644 api/core/model_runtime/model_providers/voyage/voyage.py
 create mode 100644 api/core/model_runtime/model_providers/voyage/voyage.yaml
 create mode 100644 api/tests/integration_tests/model_runtime/voyage/__init__.py
 create mode 100644 api/tests/integration_tests/model_runtime/voyage/test_provider.py
 create mode 100644 api/tests/integration_tests/model_runtime/voyage/test_rerank.py
 create mode 100644 api/tests/integration_tests/model_runtime/voyage/test_text_embedding.py

diff --git a/api/core/model_runtime/model_providers/_position.yaml b/api/core/model_runtime/model_providers/_position.yaml
index 80db22ea84fe6..89fccef6598fd 100644
--- a/api/core/model_runtime/model_providers/_position.yaml
+++ b/api/core/model_runtime/model_providers/_position.yaml
@@ -40,3 +40,4 @@
 - fireworks
 - mixedbread
 - nomic
+- voyage
diff --git a/api/core/model_runtime/model_providers/voyage/__init__.py b/api/core/model_runtime/model_providers/voyage/__init__.py
new file mode 100644
index 0000000000000..e69de29bb2d1d
diff --git a/api/core/model_runtime/model_providers/voyage/_assets/icon_l_en.svg b/api/core/model_runtime/model_providers/voyage/_assets/icon_l_en.svg
new file mode 100644
index 0000000000000..a961f5e4355ee
--- /dev/null
+++ b/api/core/model_runtime/model_providers/voyage/_assets/icon_l_en.svg
@@ -0,0 +1,21 @@
+<svg version="1.0" xmlns="http://www.w3.org/2000/svg" width="100.000000pt" height="19.000000pt" viewBox="0 0 300.000000 57.000000" preserveAspectRatio="xMidYMid meet"><g transform="translate(0.000000,57.000000) scale(0.100000,-0.100000)" fill="#000000" stroke="none"><path d="M2505 368 c-38 -84 -86 -188 -106 -230 l-38 -78 27 0 c24 0 30 7 55
+    75 l28 75 100 0 100 0 25 -55 c13 -31 24 -64 24 -75 0 -17 7 -20 44 -20 l43 0
+    -37 73 c-20 39 -68 143 -106 229 -38 87 -74 158 -80 158 -5 0 -41 -69 -79
+    -152z m110 -30 c22 -51 41 -95 42 -98 2 -3 -36 -6 -83 -7 -76 -1 -85 0 -81 15
+    12 40 72 182 77 182 3 0 24 -41 45 -92z"/><path d="M63 493 c19 -61 197 -438 209 -440 10 -2 147 282 216 449 2 4 -10 8
+    -27 8 -23 0 -31 -5 -31 -17 0 -16 -142 -365 -146 -360 -8 11 -144 329 -149
+    350 -6 23 -12 27 -42 27 -29 0 -34 -3 -30 -17z"/><path d="M2855 285 l0 -225 30 0 30 0 0 225 0 225 -30 0 -30 0 0 -225z"/><path d="M588 380 c-55 -30 -82 -74 -86 -145 -3 -50 0 -66 20 -95 39 -58 82
+    -80 153 -80 68 0 110 21 149 73 32 43 30 150 -3 196 -47 66 -158 90 -233 51z
+    m133 -16 c59 -30 89 -156 54 -224 -45 -87 -162 -78 -201 16 -18 44 -18 128 1
+    164 28 55 90 73 146 44z"/><path d="M935 303 l76 -98 -7 -72 -6 -73 33 0 34 0 -3 78 -4 77 71 93 c65 85
+    68 92 46 92 -15 0 -29 -9 -36 -22 -18 -33 -90 -128 -98 -128 -6 1 -67 85 -88
+    122 -8 15 -24 23 -53 25 l-41 4 76 -98z"/><path d="M1257 230 c-82 -169 -83 -170 -57 -170 17 0 27 6 27 15 0 8 7 31 17
+    52 l17 38 79 0 78 1 16 -34 c9 -18 16 -42 16 -52 0 -17 7 -20 41 -20 22 0 39
+    3 37 8 -2 4 -39 80 -83 170 -43 89 -84 162 -92 162 -7 0 -50 -76 -96 -170z
+    m90 -38 c-33 -2 -61 -1 -63 1 -2 2 10 34 26 71 l31 68 33 -68 33 -69 -60 -3z"/><path d="M1665 386 c-37 -16 -84 -63 -97 -96 -13 -35 -12 -104 2 -132 49 -94
+    182 -134 280 -83 24 12 29 22 32 64 3 49 3 49 -30 53 l-33 4 3 -45 c4 -61 -5
+    -71 -60 -71 -93 0 -142 57 -142 164 0 44 5 60 25 85 47 55 136 65 184 20 30
+    -28 35 -20 11 19 -19 31 -22 32 -82 32 -35 -1 -76 -7 -93 -14z"/><path d="M1955 230 l0 -170 91 0 c76 0 93 3 98 16 4 9 5 18 4 20 -2 1 -31 -1
+    -66 -5 -34 -4 -64 -5 -67 -3 -3 3 -5 36 -5 73 l0 68 55 -6 c49 -5 55 -4 55 13
+    0 17 -6 19 -55 16 l-55 -4 0 61 0 61 64 0 c48 0 65 4 70 15 4 13 -10 15 -92
+    15 l-97 0 0 -170z"/></g></svg>
\ No newline at end of file
diff --git a/api/core/model_runtime/model_providers/voyage/_assets/icon_s_en.svg b/api/core/model_runtime/model_providers/voyage/_assets/icon_s_en.svg
new file mode 100644
index 0000000000000..2c4e121dd71f0
--- /dev/null
+++ b/api/core/model_runtime/model_providers/voyage/_assets/icon_s_en.svg
@@ -0,0 +1,8 @@
+<?xml version="1.0" encoding="UTF-8"?>
+<svg width="64px" height="64px" viewBox="0 0 64 64" version="1.1" xmlns="http://www.w3.org/2000/svg" xmlns:xlink="http://www.w3.org/1999/xlink">
+    <title>voyage</title>
+    <g id="voyage" stroke="none" stroke-width="1" fill="none" fill-rule="evenodd">
+        <rect id="矩形" fill="#333333" x="0" y="0" width="64" height="64" rx="12"></rect>
+        <path d="M12.1128004,51.4376727 C13.8950799,45.8316747 30.5922254,11.1847688 31.7178757,11.0009656 C32.6559176,10.8171624 45.5070913,36.9172188 51.9795803,52.2647871 C52.1671887,52.6323936 51.0415384,53 49.4468672,53 C47.2893709,53 46.5389374,52.540492 46.5389374,51.4376727 C46.5389374,49.967247 33.2187427,17.8935861 32.8435259,18.3530942 C32.0930924,19.3640118 19.3357228,48.5887229 18.8667019,50.5186566 C18.3038768,52.6323936 17.7410516,53 14.926926,53 C12.2066045,53 11.7375836,52.7242952 12.1128004,51.4376727 Z" id="路径" fill="#FFFFFF" transform="translate(32, 32) scale(1, -1) translate(-32, -32)"></path>
+    </g>
+</svg>
\ No newline at end of file
diff --git a/api/core/model_runtime/model_providers/voyage/rerank/__init__.py b/api/core/model_runtime/model_providers/voyage/rerank/__init__.py
new file mode 100644
index 0000000000000..e69de29bb2d1d
diff --git a/api/core/model_runtime/model_providers/voyage/rerank/rerank-1.yaml b/api/core/model_runtime/model_providers/voyage/rerank/rerank-1.yaml
new file mode 100644
index 0000000000000..9c894eda85203
--- /dev/null
+++ b/api/core/model_runtime/model_providers/voyage/rerank/rerank-1.yaml
@@ -0,0 +1,4 @@
+model: rerank-1
+model_type: rerank
+model_properties:
+  context_size: 8000
diff --git a/api/core/model_runtime/model_providers/voyage/rerank/rerank-lite-1.yaml b/api/core/model_runtime/model_providers/voyage/rerank/rerank-lite-1.yaml
new file mode 100644
index 0000000000000..b052d6f00028c
--- /dev/null
+++ b/api/core/model_runtime/model_providers/voyage/rerank/rerank-lite-1.yaml
@@ -0,0 +1,4 @@
+model: rerank-lite-1
+model_type: rerank
+model_properties:
+  context_size: 4000
diff --git a/api/core/model_runtime/model_providers/voyage/rerank/rerank.py b/api/core/model_runtime/model_providers/voyage/rerank/rerank.py
new file mode 100644
index 0000000000000..33fdebbb45ef3
--- /dev/null
+++ b/api/core/model_runtime/model_providers/voyage/rerank/rerank.py
@@ -0,0 +1,123 @@
+from typing import Optional
+
+import httpx
+
+from core.model_runtime.entities.common_entities import I18nObject
+from core.model_runtime.entities.model_entities import AIModelEntity, FetchFrom, ModelPropertyKey, ModelType
+from core.model_runtime.entities.rerank_entities import RerankDocument, RerankResult
+from core.model_runtime.errors.invoke import (
+    InvokeAuthorizationError,
+    InvokeBadRequestError,
+    InvokeConnectionError,
+    InvokeError,
+    InvokeRateLimitError,
+    InvokeServerUnavailableError,
+)
+from core.model_runtime.errors.validate import CredentialsValidateFailedError
+from core.model_runtime.model_providers.__base.rerank_model import RerankModel
+
+
+class VoyageRerankModel(RerankModel):
+    """
+    Model class for Voyage rerank model.
+    """
+
+    def _invoke(
+        self,
+        model: str,
+        credentials: dict,
+        query: str,
+        docs: list[str],
+        score_threshold: Optional[float] = None,
+        top_n: Optional[int] = None,
+        user: Optional[str] = None,
+    ) -> RerankResult:
+        """
+        Invoke rerank model
+        :param model: model name
+        :param credentials: model credentials
+        :param query: search query
+        :param docs: docs for reranking
+        :param score_threshold: score threshold
+        :param top_n: top n documents to return
+        :param user: unique user id
+        :return: rerank result
+        """
+        if len(docs) == 0:
+            return RerankResult(model=model, docs=[])
+
+        base_url = credentials.get("base_url", "https://api.voyageai.com/v1")
+        base_url = base_url.removesuffix("/")
+
+        try:
+            response = httpx.post(
+                base_url + "/rerank",
+                json={"model": model, "query": query, "documents": docs, "top_k": top_n, "return_documents": True},
+                headers={"Authorization": f"Bearer {credentials.get('api_key')}", "Content-Type": "application/json"},
+            )
+            response.raise_for_status()
+            results = response.json()
+
+            rerank_documents = []
+            for result in results["data"]:
+                rerank_document = RerankDocument(
+                    index=result["index"],
+                    text=result["document"],
+                    score=result["relevance_score"],
+                )
+                if score_threshold is None or result["relevance_score"] >= score_threshold:
+                    rerank_documents.append(rerank_document)
+
+            return RerankResult(model=model, docs=rerank_documents)
+        except httpx.HTTPStatusError as e:
+            raise InvokeServerUnavailableError(str(e))
+
+    def validate_credentials(self, model: str, credentials: dict) -> None:
+        """
+        Validate model credentials
+        :param model: model name
+        :param credentials: model credentials
+        :return:
+        """
+        try:
+            self._invoke(
+                model=model,
+                credentials=credentials,
+                query="What is the capital of the United States?",
+                docs=[
+                    "Carson City is the capital city of the American state of Nevada. At the 2010 United States "
+                    "Census, Carson City had a population of 55,274.",
+                    "The Commonwealth of the Northern Mariana Islands is a group of islands in the Pacific Ocean that "
+                    "are a political division controlled by the United States. Its capital is Saipan.",
+                ],
+                score_threshold=0.8,
+            )
+        except Exception as ex:
+            raise CredentialsValidateFailedError(str(ex))
+
+    @property
+    def _invoke_error_mapping(self) -> dict[type[InvokeError], list[type[Exception]]]:
+        """
+        Map model invoke error to unified error
+        """
+        return {
+            InvokeConnectionError: [httpx.ConnectError],
+            InvokeServerUnavailableError: [httpx.RemoteProtocolError],
+            InvokeRateLimitError: [],
+            InvokeAuthorizationError: [httpx.HTTPStatusError],
+            InvokeBadRequestError: [httpx.RequestError],
+        }
+
+    def get_customizable_model_schema(self, model: str, credentials: dict) -> AIModelEntity:
+        """
+        generate custom model entities from credentials
+        """
+        entity = AIModelEntity(
+            model=model,
+            label=I18nObject(en_US=model),
+            model_type=ModelType.RERANK,
+            fetch_from=FetchFrom.CUSTOMIZABLE_MODEL,
+            model_properties={ModelPropertyKey.CONTEXT_SIZE: int(credentials.get("context_size", "8000"))},
+        )
+
+        return entity
diff --git a/api/core/model_runtime/model_providers/voyage/text_embedding/__init__.py b/api/core/model_runtime/model_providers/voyage/text_embedding/__init__.py
new file mode 100644
index 0000000000000..e69de29bb2d1d
diff --git a/api/core/model_runtime/model_providers/voyage/text_embedding/text_embedding.py b/api/core/model_runtime/model_providers/voyage/text_embedding/text_embedding.py
new file mode 100644
index 0000000000000..a8a4d3c15bbb1
--- /dev/null
+++ b/api/core/model_runtime/model_providers/voyage/text_embedding/text_embedding.py
@@ -0,0 +1,172 @@
+import time
+from json import JSONDecodeError, dumps
+from typing import Optional
+
+import requests
+
+from core.embedding.embedding_constant import EmbeddingInputType
+from core.model_runtime.entities.common_entities import I18nObject
+from core.model_runtime.entities.model_entities import AIModelEntity, FetchFrom, ModelPropertyKey, ModelType, PriceType
+from core.model_runtime.entities.text_embedding_entities import EmbeddingUsage, TextEmbeddingResult
+from core.model_runtime.errors.invoke import (
+    InvokeAuthorizationError,
+    InvokeBadRequestError,
+    InvokeConnectionError,
+    InvokeError,
+    InvokeRateLimitError,
+    InvokeServerUnavailableError,
+)
+from core.model_runtime.errors.validate import CredentialsValidateFailedError
+from core.model_runtime.model_providers.__base.text_embedding_model import TextEmbeddingModel
+
+
+class VoyageTextEmbeddingModel(TextEmbeddingModel):
+    """
+    Model class for Voyage text embedding model.
+    """
+
+    api_base: str = "https://api.voyageai.com/v1"
+
+    def _invoke(
+        self,
+        model: str,
+        credentials: dict,
+        texts: list[str],
+        user: Optional[str] = None,
+        input_type: EmbeddingInputType = EmbeddingInputType.DOCUMENT,
+    ) -> TextEmbeddingResult:
+        """
+        Invoke text embedding model
+
+        :param model: model name
+        :param credentials: model credentials
+        :param texts: texts to embed
+        :param user: unique user id
+        :param input_type: input type
+        :return: embeddings result
+        """
+        api_key = credentials["api_key"]
+        if not api_key:
+            raise CredentialsValidateFailedError("api_key is required")
+
+        base_url = credentials.get("base_url", self.api_base)
+        base_url = base_url.removesuffix("/")
+
+        url = base_url + "/embeddings"
+        headers = {"Authorization": "Bearer " + api_key, "Content-Type": "application/json"}
+        voyage_input_type = "null"
+        if input_type is not None:
+            voyage_input_type = input_type.value
+        data = {"model": model, "input": texts, "input_type": voyage_input_type}
+
+        try:
+            response = requests.post(url, headers=headers, data=dumps(data))
+        except Exception as e:
+            raise InvokeConnectionError(str(e))
+
+        if response.status_code != 200:
+            try:
+                resp = response.json()
+                msg = resp["detail"]
+                if response.status_code == 401:
+                    raise InvokeAuthorizationError(msg)
+                elif response.status_code == 429:
+                    raise InvokeRateLimitError(msg)
+                elif response.status_code == 500:
+                    raise InvokeServerUnavailableError(msg)
+                else:
+                    raise InvokeBadRequestError(msg)
+            except JSONDecodeError as e:
+                raise InvokeServerUnavailableError(
+                    f"Failed to convert response to json: {e} with text: {response.text}"
+                )
+
+        try:
+            resp = response.json()
+            embeddings = resp["data"]
+            usage = resp["usage"]
+        except Exception as e:
+            raise InvokeServerUnavailableError(f"Failed to convert response to json: {e} with text: {response.text}")
+
+        usage = self._calc_response_usage(model=model, credentials=credentials, tokens=usage["total_tokens"])
+
+        result = TextEmbeddingResult(
+            model=model, embeddings=[[float(data) for data in x["embedding"]] for x in embeddings], usage=usage
+        )
+
+        return result
+
+    def get_num_tokens(self, model: str, credentials: dict, texts: list[str]) -> int:
+        """
+        Get number of tokens for given prompt messages
+
+        :param model: model name
+        :param credentials: model credentials
+        :param texts: texts to embed
+        :return:
+        """
+        return sum(self._get_num_tokens_by_gpt2(text) for text in texts)
+
+    def validate_credentials(self, model: str, credentials: dict) -> None:
+        """
+        Validate model credentials
+
+        :param model: model name
+        :param credentials: model credentials
+        :return:
+        """
+        try:
+            self._invoke(model=model, credentials=credentials, texts=["ping"])
+        except Exception as e:
+            raise CredentialsValidateFailedError(f"Credentials validation failed: {e}")
+
+    @property
+    def _invoke_error_mapping(self) -> dict[type[InvokeError], list[type[Exception]]]:
+        return {
+            InvokeConnectionError: [InvokeConnectionError],
+            InvokeServerUnavailableError: [InvokeServerUnavailableError],
+            InvokeRateLimitError: [InvokeRateLimitError],
+            InvokeAuthorizationError: [InvokeAuthorizationError],
+            InvokeBadRequestError: [KeyError, InvokeBadRequestError],
+        }
+
+    def _calc_response_usage(self, model: str, credentials: dict, tokens: int) -> EmbeddingUsage:
+        """
+        Calculate response usage
+
+        :param model: model name
+        :param credentials: model credentials
+        :param tokens: input tokens
+        :return: usage
+        """
+        # get input price info
+        input_price_info = self.get_price(
+            model=model, credentials=credentials, price_type=PriceType.INPUT, tokens=tokens
+        )
+
+        # transform usage
+        usage = EmbeddingUsage(
+            tokens=tokens,
+            total_tokens=tokens,
+            unit_price=input_price_info.unit_price,
+            price_unit=input_price_info.unit,
+            total_price=input_price_info.total_amount,
+            currency=input_price_info.currency,
+            latency=time.perf_counter() - self.started_at,
+        )
+
+        return usage
+
+    def get_customizable_model_schema(self, model: str, credentials: dict) -> AIModelEntity:
+        """
+        generate custom model entities from credentials
+        """
+        entity = AIModelEntity(
+            model=model,
+            label=I18nObject(en_US=model),
+            model_type=ModelType.TEXT_EMBEDDING,
+            fetch_from=FetchFrom.CUSTOMIZABLE_MODEL,
+            model_properties={ModelPropertyKey.CONTEXT_SIZE: int(credentials.get("context_size"))},
+        )
+
+        return entity
diff --git a/api/core/model_runtime/model_providers/voyage/text_embedding/voyage-3-lite.yaml b/api/core/model_runtime/model_providers/voyage/text_embedding/voyage-3-lite.yaml
new file mode 100644
index 0000000000000..a06bb7639feac
--- /dev/null
+++ b/api/core/model_runtime/model_providers/voyage/text_embedding/voyage-3-lite.yaml
@@ -0,0 +1,8 @@
+model: voyage-3-lite
+model_type: text-embedding
+model_properties:
+  context_size: 32000
+pricing:
+  input: '0.00002'
+  unit: '0.001'
+  currency: USD
diff --git a/api/core/model_runtime/model_providers/voyage/text_embedding/voyage-3.yaml b/api/core/model_runtime/model_providers/voyage/text_embedding/voyage-3.yaml
new file mode 100644
index 0000000000000..117afbcaf3c80
--- /dev/null
+++ b/api/core/model_runtime/model_providers/voyage/text_embedding/voyage-3.yaml
@@ -0,0 +1,8 @@
+model: voyage-3
+model_type: text-embedding
+model_properties:
+  context_size: 32000
+pricing:
+  input: '0.00006'
+  unit: '0.001'
+  currency: USD
diff --git a/api/core/model_runtime/model_providers/voyage/voyage.py b/api/core/model_runtime/model_providers/voyage/voyage.py
new file mode 100644
index 0000000000000..3e33b45e110d5
--- /dev/null
+++ b/api/core/model_runtime/model_providers/voyage/voyage.py
@@ -0,0 +1,28 @@
+import logging
+
+from core.model_runtime.entities.model_entities import ModelType
+from core.model_runtime.errors.validate import CredentialsValidateFailedError
+from core.model_runtime.model_providers.__base.model_provider import ModelProvider
+
+logger = logging.getLogger(__name__)
+
+
+class VoyageProvider(ModelProvider):
+    def validate_provider_credentials(self, credentials: dict) -> None:
+        """
+        Validate provider credentials
+        if validate failed, raise exception
+
+        :param credentials: provider credentials, credentials form defined in `provider_credential_schema`.
+        """
+        try:
+            model_instance = self.get_model_instance(ModelType.TEXT_EMBEDDING)
+
+            # Use `voyage-3` model for validate,
+            # no matter what model you pass in, text completion model or chat model
+            model_instance.validate_credentials(model="voyage-3", credentials=credentials)
+        except CredentialsValidateFailedError as ex:
+            raise ex
+        except Exception as ex:
+            logger.exception(f"{self.get_provider_schema().provider} credentials validate failed")
+            raise ex
diff --git a/api/core/model_runtime/model_providers/voyage/voyage.yaml b/api/core/model_runtime/model_providers/voyage/voyage.yaml
new file mode 100644
index 0000000000000..c64707800eebe
--- /dev/null
+++ b/api/core/model_runtime/model_providers/voyage/voyage.yaml
@@ -0,0 +1,31 @@
+provider: voyage
+label:
+  en_US: Voyage
+description:
+  en_US: Embedding and Rerank Model Supported
+icon_small:
+  en_US: icon_s_en.svg
+icon_large:
+  en_US: icon_l_en.svg
+background: "#EFFDFD"
+help:
+  title:
+    en_US: Get your API key from Voyage AI
+    zh_Hans: 从 Voyage 获取 API Key
+  url:
+    en_US: https://dash.voyageai.com/
+supported_model_types:
+  - text-embedding
+  - rerank
+configurate_methods:
+  - predefined-model
+provider_credential_schema:
+  credential_form_schemas:
+    - variable: api_key
+      label:
+        en_US: API Key
+      type: secret-input
+      required: true
+      placeholder:
+        zh_Hans: 在此输入您的 API Key
+        en_US: Enter your API Key
diff --git a/api/pyproject.toml b/api/pyproject.toml
index 64b35621b2ec6..e737761f3b2c0 100644
--- a/api/pyproject.toml
+++ b/api/pyproject.toml
@@ -123,6 +123,7 @@ FIRECRAWL_API_KEY = "fc-"
 TEI_EMBEDDING_SERVER_URL = "http://a.abc.com:11451"
 TEI_RERANK_SERVER_URL = "http://a.abc.com:11451"
 MIXEDBREAD_API_KEY = "mk-aaaaaaaaaaaaaaaaaaaa"
+VOYAGE_API_KEY = "va-aaaaaaaaaaaaaaaaaaaa"
 
 [tool.poetry]
 name = "dify-api"
diff --git a/api/tests/integration_tests/model_runtime/voyage/__init__.py b/api/tests/integration_tests/model_runtime/voyage/__init__.py
new file mode 100644
index 0000000000000..e69de29bb2d1d
diff --git a/api/tests/integration_tests/model_runtime/voyage/test_provider.py b/api/tests/integration_tests/model_runtime/voyage/test_provider.py
new file mode 100644
index 0000000000000..08978c88a961e
--- /dev/null
+++ b/api/tests/integration_tests/model_runtime/voyage/test_provider.py
@@ -0,0 +1,25 @@
+import os
+from unittest.mock import Mock, patch
+
+import pytest
+
+from core.model_runtime.errors.validate import CredentialsValidateFailedError
+from core.model_runtime.model_providers.voyage.voyage import VoyageProvider
+
+
+def test_validate_provider_credentials():
+    provider = VoyageProvider()
+
+    with pytest.raises(CredentialsValidateFailedError):
+        provider.validate_provider_credentials(credentials={"api_key": "hahahaha"})
+    with patch("requests.post") as mock_post:
+        mock_response = Mock()
+        mock_response.json.return_value = {
+            "object": "list",
+            "data": [{"object": "embedding", "embedding": [0.23333 for _ in range(1024)], "index": 0}],
+            "model": "voyage-3",
+            "usage": {"total_tokens": 1},
+        }
+        mock_response.status_code = 200
+        mock_post.return_value = mock_response
+        provider.validate_provider_credentials(credentials={"api_key": os.environ.get("VOYAGE_API_KEY")})
diff --git a/api/tests/integration_tests/model_runtime/voyage/test_rerank.py b/api/tests/integration_tests/model_runtime/voyage/test_rerank.py
new file mode 100644
index 0000000000000..e97a9e4c811c8
--- /dev/null
+++ b/api/tests/integration_tests/model_runtime/voyage/test_rerank.py
@@ -0,0 +1,92 @@
+import os
+from unittest.mock import Mock, patch
+
+import pytest
+
+from core.model_runtime.entities.rerank_entities import RerankResult
+from core.model_runtime.errors.validate import CredentialsValidateFailedError
+from core.model_runtime.model_providers.voyage.rerank.rerank import VoyageRerankModel
+
+
+def test_validate_credentials():
+    model = VoyageRerankModel()
+
+    with pytest.raises(CredentialsValidateFailedError):
+        model.validate_credentials(
+            model="rerank-lite-1",
+            credentials={"api_key": "invalid_key"},
+        )
+    with patch("httpx.post") as mock_post:
+        mock_response = Mock()
+        mock_response.json.return_value = {
+            "object": "list",
+            "data": [
+                {
+                    "relevance_score": 0.546875,
+                    "index": 0,
+                    "document": "Carson City is the capital city of the American state of Nevada. At the 2010 United "
+                    "States Census, Carson City had a population of 55,274.",
+                },
+                {
+                    "relevance_score": 0.4765625,
+                    "index": 1,
+                    "document": "The Commonwealth of the Northern Mariana Islands is a group of islands in the "
+                    "Pacific Ocean that are a political division controlled by the United States. Its "
+                    "capital is Saipan.",
+                },
+            ],
+            "model": "rerank-lite-1",
+            "usage": {"total_tokens": 96},
+        }
+        mock_response.status_code = 200
+        mock_post.return_value = mock_response
+        model.validate_credentials(
+            model="rerank-lite-1",
+            credentials={
+                "api_key": os.environ.get("VOYAGE_API_KEY"),
+            },
+        )
+
+
+def test_invoke_model():
+    model = VoyageRerankModel()
+    with patch("httpx.post") as mock_post:
+        mock_response = Mock()
+        mock_response.json.return_value = {
+            "object": "list",
+            "data": [
+                {
+                    "relevance_score": 0.84375,
+                    "index": 0,
+                    "document": "Kasumi is a girl name of Japanese origin meaning mist.",
+                },
+                {
+                    "relevance_score": 0.4765625,
+                    "index": 1,
+                    "document": "Her music is a kawaii bass, a mix of future bass, pop, and kawaii music and she "
+                    "leads a team named PopiParty.",
+                },
+            ],
+            "model": "rerank-lite-1",
+            "usage": {"total_tokens": 59},
+        }
+        mock_response.status_code = 200
+        mock_post.return_value = mock_response
+        result = model.invoke(
+            model="rerank-lite-1",
+            credentials={
+                "api_key": os.environ.get("VOYAGE_API_KEY"),
+            },
+            query="Who is Kasumi?",
+            docs=[
+                "Kasumi is a girl name of Japanese origin meaning mist.",
+                "Her music is a kawaii bass, a mix of future bass, pop, and kawaii music and she leads a team named "
+                "PopiParty.",
+            ],
+            score_threshold=0.5,
+        )
+
+        assert isinstance(result, RerankResult)
+        assert len(result.docs) == 1
+        assert result.docs[0].index == 0
+        assert result.docs[0].score >= 0.5
diff --git a/api/tests/integration_tests/model_runtime/voyage/test_text_embedding.py b/api/tests/integration_tests/model_runtime/voyage/test_text_embedding.py
new file mode 100644
index 0000000000000..75719672a9ecc
--- /dev/null
+++ b/api/tests/integration_tests/model_runtime/voyage/test_text_embedding.py
@@ -0,0 +1,70 @@
+import os
+from unittest.mock import Mock, patch
+
+import pytest
+
+from core.model_runtime.entities.text_embedding_entities import TextEmbeddingResult
+from core.model_runtime.errors.validate import CredentialsValidateFailedError
+from core.model_runtime.model_providers.voyage.text_embedding.text_embedding import VoyageTextEmbeddingModel
+
+
+def test_validate_credentials():
+    model = VoyageTextEmbeddingModel()
+
+    with pytest.raises(CredentialsValidateFailedError):
+        model.validate_credentials(model="voyage-3", credentials={"api_key": "invalid_key"})
+    with patch("requests.post") as mock_post:
+        mock_response = Mock()
+        mock_response.json.return_value = {
+            "object": "list",
+            "data": [{"object": "embedding", "embedding": [0.23333 for _ in range(1024)], "index": 0}],
+            "model": "voyage-3",
+            "usage": {"total_tokens": 1},
+        }
+        mock_response.status_code = 200
+        mock_post.return_value = mock_response
+        model.validate_credentials(model="voyage-3", credentials={"api_key": os.environ.get("VOYAGE_API_KEY")})
+
+
+def test_invoke_model():
+    model = VoyageTextEmbeddingModel()
+
+    with patch("requests.post") as mock_post:
+        mock_response = Mock()
+        mock_response.json.return_value = {
+            "object": "list",
+            "data": [
+                {"object": "embedding", "embedding": [0.23333 for _ in range(1024)], "index": 0},
+                {"object": "embedding", "embedding": [0.23333 for _ in range(1024)], "index": 1},
+            ],
+            "model": "voyage-3",
+            "usage": {"total_tokens": 2},
+        }
+        mock_response.status_code = 200
+        mock_post.return_value = mock_response
+        result = model.invoke(
+            model="voyage-3",
+            credentials={
+                "api_key": os.environ.get("VOYAGE_API_KEY"),
+            },
+            texts=["hello", "world"],
+            user="abc-123",
+        )
+
+        assert isinstance(result, TextEmbeddingResult)
+        assert len(result.embeddings) == 2
+        assert result.usage.total_tokens == 2
+
+
+def test_get_num_tokens():
+    model = VoyageTextEmbeddingModel()
+
+    num_tokens = model.get_num_tokens(
+        model="voyage-3",
+        credentials={
+            "api_key": os.environ.get("VOYAGE_API_KEY"),
+        },
+        texts=["ping"],
+    )
+
+    assert num_tokens == 1
diff --git a/dev/pytest/pytest_model_runtime.sh b/dev/pytest/pytest_model_runtime.sh
index b60ff64fdcd90..63891eb9f8d13 100755
--- a/dev/pytest/pytest_model_runtime.sh
+++ b/dev/pytest/pytest_model_runtime.sh
@@ -9,4 +9,5 @@ pytest api/tests/integration_tests/model_runtime/anthropic \
   api/tests/integration_tests/model_runtime/upstage \
   api/tests/integration_tests/model_runtime/fireworks \
   api/tests/integration_tests/model_runtime/nomic \
-  api/tests/integration_tests/model_runtime/mixedbread
+  api/tests/integration_tests/model_runtime/mixedbread \
+  api/tests/integration_tests/model_runtime/voyage
\ No newline at end of file

From 369e1e6f58814929c7a236a482189f2ff4605570 Mon Sep 17 00:00:00 2001
From: Zhaofeng Miao <522856232@qq.com>
Date: Mon, 30 Sep 2024 09:57:19 +0800
Subject: [PATCH 108/235] feat(website-crawl): add jina reader as additional
 alternative for website crawling (#8761)

---
 api/controllers/console/datasets/website.py   |   6 +-
 api/core/rag/extractor/extract_processor.py   |  10 +
 .../rag/extractor/jina_reader_extractor.py    |  35 +++
 api/services/auth/api_key_auth_factory.py     |   3 +
 api/services/auth/jina.py                     |  44 ++++
 api/services/website_service.py               | 100 ++++++++
 .../datasets/create/assets/jina.png           | Bin 0 -> 2773 bytes
 web/app/components/datasets/create/index.tsx  |  12 +-
 .../datasets/create/step-one/index.tsx        |  11 +-
 .../datasets/create/step-two/index.tsx        |  11 +-
 .../base/checkbox-with-label.tsx              |  11 +
 .../crawled-result-item.tsx                   |   0
 .../{firecrawl => base}/crawled-result.tsx    |   2 +-
 .../website/{firecrawl => base}/crawling.tsx  |   0
 .../{firecrawl => }/base/error-message.tsx    |   0
 .../website/{firecrawl => }/base/field.tsx    |   0
 .../website/{firecrawl => }/base/input.tsx    |   0
 .../{firecrawl => base}/mock-crawl-result.ts  |   0
 .../{firecrawl => }/base/options-wrap.tsx     |   0
 .../{firecrawl => }/base/url-input.tsx        |   0
 .../create/website/firecrawl/index.tsx        |  10 +-
 .../create/website/firecrawl/options.tsx      |   4 +-
 .../datasets/create/website/index.module.css  |   6 +
 .../datasets/create/website/index.tsx         | 100 ++++++--
 .../create/website/jina-reader/header.tsx     |  42 ++++
 .../create/website/jina-reader/index.tsx      | 232 ++++++++++++++++++
 .../create/website/jina-reader/options.tsx    |  59 +++++
 .../datasets/create/website/no-data.tsx       |  45 +++-
 .../config-firecrawl-modal.tsx                |   2 +-
 .../config-jina-reader-modal.tsx              | 140 +++++++++++
 .../data-source-website/index.tsx             |  51 ++--
 .../data-source-page/index.tsx                |   4 +-
 .../data-source-page/panel/index.tsx          |   5 +-
 web/i18n/en-US/dataset-creation.ts            |  15 +-
 web/i18n/zh-Hans/dataset-creation.ts          |  15 +-
 web/models/common.ts                          |   1 +
 web/models/datasets.ts                        |   1 +
 web/service/datasets.ts                       |  25 +-
 38 files changed, 927 insertions(+), 75 deletions(-)
 create mode 100644 api/core/rag/extractor/jina_reader_extractor.py
 create mode 100644 api/services/auth/jina.py
 create mode 100644 web/app/components/datasets/create/assets/jina.png
 rename web/app/components/datasets/create/website/{firecrawl => }/base/checkbox-with-label.tsx (72%)
 rename web/app/components/datasets/create/website/{firecrawl => base}/crawled-result-item.tsx (100%)
 rename web/app/components/datasets/create/website/{firecrawl => base}/crawled-result.tsx (97%)
 rename web/app/components/datasets/create/website/{firecrawl => base}/crawling.tsx (100%)
 rename web/app/components/datasets/create/website/{firecrawl => }/base/error-message.tsx (100%)
 rename web/app/components/datasets/create/website/{firecrawl => }/base/field.tsx (100%)
 rename web/app/components/datasets/create/website/{firecrawl => }/base/input.tsx (100%)
 rename web/app/components/datasets/create/website/{firecrawl => base}/mock-crawl-result.ts (100%)
 rename web/app/components/datasets/create/website/{firecrawl => }/base/options-wrap.tsx (100%)
 rename web/app/components/datasets/create/website/{firecrawl => }/base/url-input.tsx (100%)
 create mode 100644 web/app/components/datasets/create/website/index.module.css
 create mode 100644 web/app/components/datasets/create/website/jina-reader/header.tsx
 create mode 100644 web/app/components/datasets/create/website/jina-reader/index.tsx
 create mode 100644 web/app/components/datasets/create/website/jina-reader/options.tsx
 create mode 100644 web/app/components/header/account-setting/data-source-page/data-source-website/config-jina-reader-modal.tsx

diff --git a/api/controllers/console/datasets/website.py b/api/controllers/console/datasets/website.py
index cb54f1aacbc2a..e80ce17c6866a 100644
--- a/api/controllers/console/datasets/website.py
+++ b/api/controllers/console/datasets/website.py
@@ -14,7 +14,9 @@ class WebsiteCrawlApi(Resource):
     @account_initialization_required
     def post(self):
         parser = reqparse.RequestParser()
-        parser.add_argument("provider", type=str, choices=["firecrawl"], required=True, nullable=True, location="json")
+        parser.add_argument(
+            "provider", type=str, choices=["firecrawl", "jinareader"], required=True, nullable=True, location="json"
+        )
         parser.add_argument("url", type=str, required=True, nullable=True, location="json")
         parser.add_argument("options", type=dict, required=True, nullable=True, location="json")
         args = parser.parse_args()
@@ -33,7 +35,7 @@ class WebsiteCrawlStatusApi(Resource):
     @account_initialization_required
     def get(self, job_id: str):
         parser = reqparse.RequestParser()
-        parser.add_argument("provider", type=str, choices=["firecrawl"], required=True, location="args")
+        parser.add_argument("provider", type=str, choices=["firecrawl", "jinareader"], required=True, location="args")
         args = parser.parse_args()
         # get crawl status
         try:
diff --git a/api/core/rag/extractor/extract_processor.py b/api/core/rag/extractor/extract_processor.py
index 0ffc89b214c2d..9048138511b12 100644
--- a/api/core/rag/extractor/extract_processor.py
+++ b/api/core/rag/extractor/extract_processor.py
@@ -12,6 +12,7 @@
 from core.rag.extractor.excel_extractor import ExcelExtractor
 from core.rag.extractor.firecrawl.firecrawl_web_extractor import FirecrawlWebExtractor
 from core.rag.extractor.html_extractor import HtmlExtractor
+from core.rag.extractor.jina_reader_extractor import JinaReaderWebExtractor
 from core.rag.extractor.markdown_extractor import MarkdownExtractor
 from core.rag.extractor.notion_extractor import NotionExtractor
 from core.rag.extractor.pdf_extractor import PdfExtractor
@@ -171,6 +172,15 @@ def extract(
                     only_main_content=extract_setting.website_info.only_main_content,
                 )
                 return extractor.extract()
+            elif extract_setting.website_info.provider == "jinareader":
+                extractor = JinaReaderWebExtractor(
+                    url=extract_setting.website_info.url,
+                    job_id=extract_setting.website_info.job_id,
+                    tenant_id=extract_setting.website_info.tenant_id,
+                    mode=extract_setting.website_info.mode,
+                    only_main_content=extract_setting.website_info.only_main_content,
+                )
+                return extractor.extract()
             else:
                 raise ValueError(f"Unsupported website provider: {extract_setting.website_info.provider}")
         else:
diff --git a/api/core/rag/extractor/jina_reader_extractor.py b/api/core/rag/extractor/jina_reader_extractor.py
new file mode 100644
index 0000000000000..5b780af126b30
--- /dev/null
+++ b/api/core/rag/extractor/jina_reader_extractor.py
@@ -0,0 +1,35 @@
+from core.rag.extractor.extractor_base import BaseExtractor
+from core.rag.models.document import Document
+from services.website_service import WebsiteService
+
+
+class JinaReaderWebExtractor(BaseExtractor):
+    """
+    Crawl and scrape websites and return content in clean llm-ready markdown.
+    """
+
+    def __init__(self, url: str, job_id: str, tenant_id: str, mode: str = "crawl", only_main_content: bool = False):
+        """Initialize with url, api_key, base_url and mode."""
+        self._url = url
+        self.job_id = job_id
+        self.tenant_id = tenant_id
+        self.mode = mode
+        self.only_main_content = only_main_content
+
+    def extract(self) -> list[Document]:
+        """Extract content from the URL."""
+        documents = []
+        if self.mode == "crawl":
+            crawl_data = WebsiteService.get_crawl_url_data(self.job_id, "jinareader", self._url, self.tenant_id)
+            if crawl_data is None:
+                return []
+            document = Document(
+                page_content=crawl_data.get("content", ""),
+                metadata={
+                    "source_url": crawl_data.get("url"),
+                    "description": crawl_data.get("description"),
+                    "title": crawl_data.get("title"),
+                },
+            )
+            documents.append(document)
+        return documents
diff --git a/api/services/auth/api_key_auth_factory.py b/api/services/auth/api_key_auth_factory.py
index ae5b953b47f58..36387e9c2efdb 100644
--- a/api/services/auth/api_key_auth_factory.py
+++ b/api/services/auth/api_key_auth_factory.py
@@ -1,10 +1,13 @@
 from services.auth.firecrawl import FirecrawlAuth
+from services.auth.jina import JinaAuth
 
 
 class ApiKeyAuthFactory:
     def __init__(self, provider: str, credentials: dict):
         if provider == "firecrawl":
             self.auth = FirecrawlAuth(credentials)
+        elif provider == "jinareader":
+            self.auth = JinaAuth(credentials)
         else:
             raise ValueError("Invalid provider")
 
diff --git a/api/services/auth/jina.py b/api/services/auth/jina.py
new file mode 100644
index 0000000000000..de898a1f94b76
--- /dev/null
+++ b/api/services/auth/jina.py
@@ -0,0 +1,44 @@
+import json
+
+import requests
+
+from services.auth.api_key_auth_base import ApiKeyAuthBase
+
+
+class JinaAuth(ApiKeyAuthBase):
+    def __init__(self, credentials: dict):
+        super().__init__(credentials)
+        auth_type = credentials.get("auth_type")
+        if auth_type != "bearer":
+            raise ValueError("Invalid auth type, Jina Reader auth type must be Bearer")
+        self.api_key = credentials.get("config").get("api_key", None)
+
+        if not self.api_key:
+            raise ValueError("No API key provided")
+
+    def validate_credentials(self):
+        headers = self._prepare_headers()
+        options = {
+            "url": "https://example.com",
+        }
+        response = self._post_request("https://r.jina.ai", options, headers)
+        if response.status_code == 200:
+            return True
+        else:
+            self._handle_error(response)
+
+    def _prepare_headers(self):
+        return {"Content-Type": "application/json", "Authorization": f"Bearer {self.api_key}"}
+
+    def _post_request(self, url, data, headers):
+        return requests.post(url, headers=headers, json=data)
+
+    def _handle_error(self, response):
+        if response.status_code in {402, 409, 500}:
+            error_message = response.json().get("error", "Unknown error occurred")
+            raise Exception(f"Failed to authorize. Status code: {response.status_code}. Error: {error_message}")
+        else:
+            if response.text:
+                error_message = json.loads(response.text).get("error", "Unknown error occurred")
+                raise Exception(f"Failed to authorize. Status code: {response.status_code}. Error: {error_message}")
+            raise Exception(f"Unexpected error occurred while trying to authorize. Status code: {response.status_code}")
diff --git a/api/services/website_service.py b/api/services/website_service.py
index fea605cf30b41..be01815720ab5 100644
--- a/api/services/website_service.py
+++ b/api/services/website_service.py
@@ -1,6 +1,7 @@
 import datetime
 import json
 
+import requests
 from flask_login import current_user
 
 from core.helper import encrypter
@@ -65,6 +66,35 @@ def crawl_url(cls, args: dict) -> dict:
             time = str(datetime.datetime.now().timestamp())
             redis_client.setex(website_crawl_time_cache_key, 3600, time)
             return {"status": "active", "job_id": job_id}
+        elif provider == "jinareader":
+            api_key = encrypter.decrypt_token(
+                tenant_id=current_user.current_tenant_id, token=credentials.get("config").get("api_key")
+            )
+            crawl_sub_pages = options.get("crawl_sub_pages", False)
+            if not crawl_sub_pages:
+                response = requests.get(
+                    f"https://r.jina.ai/{url}",
+                    headers={"Accept": "application/json", "Authorization": f"Bearer {api_key}"},
+                )
+                if response.json().get("code") != 200:
+                    raise ValueError("Failed to crawl")
+                return {"status": "active", "data": response.json().get("data")}
+            else:
+                response = requests.post(
+                    "https://adaptivecrawl-kir3wx7b3a-uc.a.run.app",
+                    json={
+                        "url": url,
+                        "maxPages": options.get("limit", 1),
+                        "useSitemap": options.get("use_sitemap", True),
+                    },
+                    headers={
+                        "Content-Type": "application/json",
+                        "Authorization": f"Bearer {api_key}",
+                    },
+                )
+                if response.json().get("code") != 200:
+                    raise ValueError("Failed to crawl")
+                return {"status": "active", "job_id": response.json().get("data", {}).get("taskId")}
         else:
             raise ValueError("Invalid provider")
 
@@ -93,6 +123,42 @@ def get_crawl_status(cls, job_id: str, provider: str) -> dict:
                     time_consuming = abs(end_time - float(start_time))
                     crawl_status_data["time_consuming"] = f"{time_consuming:.2f}"
                     redis_client.delete(website_crawl_time_cache_key)
+        elif provider == "jinareader":
+            api_key = encrypter.decrypt_token(
+                tenant_id=current_user.current_tenant_id, token=credentials.get("config").get("api_key")
+            )
+            response = requests.post(
+                "https://adaptivecrawlstatus-kir3wx7b3a-uc.a.run.app",
+                headers={"Content-Type": "application/json", "Authorization": f"Bearer {api_key}"},
+                json={"taskId": job_id},
+            )
+            data = response.json().get("data", {})
+            crawl_status_data = {
+                "status": data.get("status", "active"),
+                "job_id": job_id,
+                "total": len(data.get("urls", [])),
+                "current": len(data.get("processed", [])) + len(data.get("failed", [])),
+                "data": [],
+                "time_consuming": data.get("duration", 0) / 1000,
+            }
+
+            if crawl_status_data["status"] == "completed":
+                response = requests.post(
+                    "https://adaptivecrawlstatus-kir3wx7b3a-uc.a.run.app",
+                    headers={"Content-Type": "application/json", "Authorization": f"Bearer {api_key}"},
+                    json={"taskId": job_id, "urls": list(data.get("processed", {}).keys())},
+                )
+                data = response.json().get("data", {})
+                formatted_data = [
+                    {
+                        "title": item.get("data", {}).get("title"),
+                        "source_url": item.get("data", {}).get("url"),
+                        "description": item.get("data", {}).get("description"),
+                        "markdown": item.get("data", {}).get("content"),
+                    }
+                    for item in data.get("processed", {}).values()
+                ]
+                crawl_status_data["data"] = formatted_data
         else:
             raise ValueError("Invalid provider")
         return crawl_status_data
@@ -119,6 +185,40 @@ def get_crawl_url_data(cls, job_id: str, provider: str, url: str, tenant_id: str
                     if item.get("source_url") == url:
                         return item
             return None
+        elif provider == "jinareader":
+            file_key = "website_files/" + job_id + ".txt"
+            if storage.exists(file_key):
+                data = storage.load_once(file_key)
+                if data:
+                    data = json.loads(data.decode("utf-8"))
+            elif not job_id:
+                response = requests.get(
+                    f"https://r.jina.ai/{url}",
+                    headers={"Accept": "application/json", "Authorization": f"Bearer {api_key}"},
+                )
+                if response.json().get("code") != 200:
+                    raise ValueError("Failed to crawl")
+                return response.json().get("data")
+            else:
+                api_key = encrypter.decrypt_token(tenant_id=tenant_id, token=credentials.get("config").get("api_key"))
+                response = requests.post(
+                    "https://adaptivecrawlstatus-kir3wx7b3a-uc.a.run.app",
+                    headers={"Content-Type": "application/json", "Authorization": f"Bearer {api_key}"},
+                    json={"taskId": job_id},
+                )
+                data = response.json().get("data", {})
+                if data.get("status") != "completed":
+                    raise ValueError("Crawl job is not completed")
+
+                response = requests.post(
+                    "https://adaptivecrawlstatus-kir3wx7b3a-uc.a.run.app",
+                    headers={"Content-Type": "application/json", "Authorization": f"Bearer {api_key}"},
+                    json={"taskId": job_id, "urls": list(data.get("processed", {}).keys())},
+                )
+                data = response.json().get("data", {})
+                for item in data.get("processed", {}).values():
+                    if item.get("data", {}).get("url") == url:
+                        return item.get("data", {})
         else:
             raise ValueError("Invalid provider")
 
diff --git a/web/app/components/datasets/create/assets/jina.png b/web/app/components/datasets/create/assets/jina.png
new file mode 100644
index 0000000000000000000000000000000000000000..b4beeafdfb127115d0e29ad3728654310dfc389a
GIT binary patch
literal 2773
zcmXX|c{~&T|9|gbi^k^I9K(h*K91$em4*q0mJXD(ijqvONNHws7CMxpD5>O#s8vLc
zkK7fNPq~eVTyth>_^!wA_s9GBe!d>B$LsNWzmCWI@l2w*I-n$!Bme*?Cr7HA=w|&r
zNO2K1SRZc(KqA|TYIER1_DqhGpPY*v%ToIr$tLKk(#mCbw&WEFw$k>cN2*>@s{MN}
z((1=F^LULM%RF`!0)wI2i-`p{>Q#)en|37aN^x!a9obl&e`@`wbHrX3_w&DZkkmJJ
zzOQETi`}1{jr8?QZ!GlO_bMh_tiTfvZ0P&o1f;GK6bZc=0@!tVn%4<s6M8HWn24C4
zs)+fiCyGGpuc?zM0zCo^xv;R19Kcn$b+UPUd_1%%vmAp6gQ>_zWvnM`D;e7p&*&<^
z80hL67nub1o0XKpS+?UVGkqESzSq|>Zr`3?SO{4kZa`=#O)O0J7%@j%FxDZXj`HXU
zesDfl7)w0;F}t<6xY)d<HhQDt4BMi9XskWsv}6pk;cn-GtT^7-552LzRz;p0s*R4c
zQpKT?8M|SZPn2g0Bn^W{s1K?-vh8c<UHo*9SFin9>HE0-sQ=lsjYC&)1`-38Yg8t-
z*UYA8kiBedVib<qltpCe7JOeE?}(eVCU)|{fS$o|Zo1x)g>yqSik81DA)wQg(R&Nf
zAQXFh`w7w5D1dYY+`fJL-vT%Vp1?X(au@^lA##g~uBfvD6VPBbHUu<ca?ekF{iTT>
z0~j%0^EEa2|GVN@cq+@EgNt#C1GLSQyQU05$Nimg!JAmfmsU_pt*kT1W%RugetULI
z4iaV=3xc-Zg3;hS0$1UYACjb_Ga_mVcGqJ~GTWSjfuI?n_vKj!451Bh&kfxJiH1PM
zB8+EZVsZ;M5srlLTqjIu!Elnkem)XCW=Cv<Q@D_c?s5%#775l_xPL2@FbRaKiw>F|
z#<B`Fz#?MS8Bi5j`bXcO!J8Z#P=ULHpy45J!e{)|B%k03Y&**9zyTx}_(%Z#k|@X(
z4a&;wfHstB4qck~!~uyGpxnbb)_pfr14-$)w`0@aJjLz?P|9l~Bv`RM_cww+`1b!I
zetv1zuW@wBg`^U}BvuI;LL*kEt5eGEZmSnuFz591YPD+=H8jB5?`6p?fTqoLO4EAf
zvco!hZWx<0jg=8bm9ko-&B5OL)-Mfikg=>V_iW>UZW_K>M_YSXR?C_9<;mXhFZnLt
zzc*jQ`Sv{sJf=hP=FLk`oSc#v{bdla{jS0$ihoC(SVrSj+~PxrYYTr?_<g=TpIb8w
zjtNqfw+jZ}URupI52bFx^_(f9(3ALCR*@lo*#@w=(N@_{S;<gJIE^EY5+o}c2`WR!
zKOf><H?Xa1SNz;6n&D~;3?WFwXy763;;he~KX3fRiUZXh7kwPIX@KiFhQ4ib%&D&>
znhp(YNn?d`MgvWd`}G}d(#+pWXRI&G(Uo5X%zijdMDS(@s+c*?RDrP;o@*665?i8c
zVW|9F9WCIM#a$8=7%n~?;YcTn0n5hQ|F}=5=E?))a`)8E%fB^m?U{M21AmBQ@@M*9
zH`=BiLwdA8%$ULprYaOhN5YY(WeFp_7fazbU{x@$p|(BHOD1tyh2=5Q^P16@j|0Nt
zmuJ68#Ep_w9uad}G<+7kQ|fJ{fVaxHeAE_@JAde@$MJt2H?WX+xl5wbxLc7PNJyQg
zB7Etq3}g=R#5L#6yu5zPw)y3$ZzI|F``#tS$${jHLHFusUg*HUF>{{da^LH7DV6@H
zhLN5Zr;iSCKiDEc@9*J;gdS(;DgvD0SO#z$1wtwmuWS^z3u_BH2NBY$@3n+~U=*pf
z{G}+*&KUksEp8E$brB&+1jyAq`>SjOKwVs(Jg5P=j^aix{zw)-L4<*2PH1u&TianY
z8%Dp#-)kb{RVnpP>Unzz))1frJxGgvF_&SiNA=TbPm`0_Z}g77;eI65*@}bIjj3<X
zI!uL^V3g8)`FP9lfx*%bws01HJz=xrO@?4a94PtE_E$EoFnc_Lq|x}ba6_yl23#pk
z?vP>~tqNCMnH<;5d{`GR^CYO%77naC*kY0gVx!LidR64&DUv#C^Utp(M^wh!JsrXI
z*D*LOR$Lh}mvETtM>IU^NlX(X#-TKit3gj}%5;C`T~f4Cg`ikl%Qqb{p}V`~KPc&<
zgE@10ugc0Uo!%`o456B7KxWCUr1%iS{`L%mnYsJ^W>_d7-z+)&a<?!4*#^i>Ewd~1
zd~xaloS@HG4;ZYjY#UV50}>Jvwy3)4o)-sKn0)iPx)_+9r^EPv#qO!?`S0m~JkLSB
zdaedVP-+f6&ddAP%+42@TN)X9iUbNQrz75nd`YkXgd6M?nx{>gSlW}5&8g-}*A*qe
zq;>SV&?}aC4zx5`hW>EA$<~sL??fq^c7p^6%fpsGP!%y=G=2#IQbd2u8_g{)*svdu
zoYL@@Fh-jJcS^K|_#rPYV8tsLol3^UFFZFwx++;J-c*FMtcLf0gQjCa=Y=3S$w4RI
zI2Sp5o|KyPboU=U4xk)4bnxK(zK6U1D|~?kK5?dj{h2;KvSKKSstjiY;a;bFeoyfI
z+76Ctr5Hv3sh6i`;Rm|6_eN)Zeju{YB~0DSLfYOC8dehld^n7unt^T~bc)1X_sPk%
z^i!ZSoJN9Odnq;%6%AT4bn7+kJ9mXn2}M2cdpr1!F<c(II8+_kkRYvA+OWC~uBV8P
zd5<Q1(wneS*z>6%>$xA649Hdw-j*J#4OBZRNf_RbVD%Tf=2-Y{hBHlF$^Znck$3i#
zpLF=zsFw_YDanc{G;4)ez;T7ViaSs5ZvtIfI*(XU>%t!Ip7vMODTlxuq^#{nRRo`F
z!}?eo_VecV=g*|T$b3bRHFLZ*ZEIu9VLYFs6Q5-Sd|*;47G2ib_yakBTs9rm{er|P
zlw@5vk=@K^y9Ls;QT1I43;)3g%9wP$;BP_tldx9kP|A?>*!Li-WhJ*Pbuij%FHjj9
z8;criyB&UDw>$ZTHy|j>V9CFxd(wR54(@L%6w9M8z>Csyzpup}NEZVk+Lo36hyaRc
z`BL7BKX_jDxE$~}>ci-T+~mRl(#~a0S)xzT4Fj)=I>crLy5%SSg*GoJ0mpU54bVyK
zn~cRDANKTCWwcu~H$ezHjfKj#YD=9s18+@5TuNd*l}tYKZ@PX4%h#<yJ9*cKSEtM6
z!03>JN_L)yi%UvoU_9rJNE+;Y9z%5Uo&)qWJHw?!(OZ(tVNZz@hwA1)=f@n&)$a0>
zjD$jo;`eqX^nzBHucxR}?{NBFq_5P`Z9Ax`9Xo(45p8(BKuYSQrh-g<fM|o8+uPdS
zMM345R?+L#W}J$E9>a23#r6m7LxsB?(gpejM#wMErB$tX5wramF>-FfvPf(QZf@E0
zE?>qmWPh4jS996u$IZa)QheGC-iBRkXn*C@Xn~iOTk^IWb#Jn+C3JjUwj%o1Z;*Z|
zkm(jr+A^rT(_yh)Q9nZ@yEIzvI#22WS6YSM6gE~h9b%d!2fB2jA|XeDqTyH+1Ri5=
z{#W?PIHEn4uoMv}iI}mph#6z3B9Q%S$~&VyxPQI71Lo9>)ES#@8FWvI<SKBoccqrv
H(%Jt5@CMbo

literal 0
HcmV?d00001

diff --git a/web/app/components/datasets/create/index.tsx b/web/app/components/datasets/create/index.tsx
index 12c6284d882c5..98098445c7695 100644
--- a/web/app/components/datasets/create/index.tsx
+++ b/web/app/components/datasets/create/index.tsx
@@ -11,7 +11,7 @@ import { DataSourceType } from '@/models/datasets'
 import type { CrawlOptions, CrawlResultItem, DataSet, FileItem, createDocumentResponse } from '@/models/datasets'
 import { fetchDataSource } from '@/service/common'
 import { fetchDatasetDetail } from '@/service/datasets'
-import type { NotionPage } from '@/models/common'
+import { DataSourceProvider, type NotionPage } from '@/models/common'
 import { useModalContext } from '@/context/modal-context'
 import { useDefaultModel } from '@/app/components/header/account-setting/model-provider-page/hooks'
 
@@ -26,6 +26,7 @@ const DEFAULT_CRAWL_OPTIONS: CrawlOptions = {
   excludes: '',
   limit: 10,
   max_depth: '',
+  use_sitemap: true,
 }
 
 const DatasetUpdateForm = ({ datasetId }: DatasetUpdateFormProps) => {
@@ -51,7 +52,8 @@ const DatasetUpdateForm = ({ datasetId }: DatasetUpdateFormProps) => {
   const updateFileList = (preparedFiles: FileItem[]) => {
     setFiles(preparedFiles)
   }
-  const [fireCrawlJobId, setFireCrawlJobId] = useState('')
+  const [websiteCrawlProvider, setWebsiteCrawlProvider] = useState<DataSourceProvider>(DataSourceProvider.fireCrawl)
+  const [websiteCrawlJobId, setWebsiteCrawlJobId] = useState('')
 
   const updateFile = (fileItem: FileItem, progress: number, list: FileItem[]) => {
     const targetIndex = list.findIndex(file => file.fileID === fileItem.fileID)
@@ -137,7 +139,8 @@ const DatasetUpdateForm = ({ datasetId }: DatasetUpdateFormProps) => {
             onStepChange={nextStep}
             websitePages={websitePages}
             updateWebsitePages={setWebsitePages}
-            onFireCrawlJobIdChange={setFireCrawlJobId}
+            onWebsiteCrawlProviderChange={setWebsiteCrawlProvider}
+            onWebsiteCrawlJobIdChange={setWebsiteCrawlJobId}
             crawlOptions={crawlOptions}
             onCrawlOptionsChange={setCrawlOptions}
           />
@@ -151,7 +154,8 @@ const DatasetUpdateForm = ({ datasetId }: DatasetUpdateFormProps) => {
           files={fileList.map(file => file.file)}
           notionPages={notionPages}
           websitePages={websitePages}
-          fireCrawlJobId={fireCrawlJobId}
+          websiteCrawlProvider={websiteCrawlProvider}
+          websiteCrawlJobId={websiteCrawlJobId}
           onStepChange={changeStep}
           updateIndexingTypeCache={updateIndexingTypeCache}
           updateResultCache={updateResultCache}
diff --git a/web/app/components/datasets/create/step-one/index.tsx b/web/app/components/datasets/create/step-one/index.tsx
index c2d77f4cecdcc..643932e9ae21d 100644
--- a/web/app/components/datasets/create/step-one/index.tsx
+++ b/web/app/components/datasets/create/step-one/index.tsx
@@ -10,7 +10,7 @@ import WebsitePreview from '../website/preview'
 import s from './index.module.css'
 import cn from '@/utils/classnames'
 import type { CrawlOptions, CrawlResultItem, FileItem } from '@/models/datasets'
-import type { NotionPage } from '@/models/common'
+import type { DataSourceProvider, NotionPage } from '@/models/common'
 import { DataSourceType } from '@/models/datasets'
 import Button from '@/app/components/base/button'
 import { NotionPageSelector } from '@/app/components/base/notion-page-selector'
@@ -33,7 +33,8 @@ type IStepOneProps = {
   changeType: (type: DataSourceType) => void
   websitePages?: CrawlResultItem[]
   updateWebsitePages: (value: CrawlResultItem[]) => void
-  onFireCrawlJobIdChange: (jobId: string) => void
+  onWebsiteCrawlProviderChange: (provider: DataSourceProvider) => void
+  onWebsiteCrawlJobIdChange: (jobId: string) => void
   crawlOptions: CrawlOptions
   onCrawlOptionsChange: (payload: CrawlOptions) => void
 }
@@ -69,7 +70,8 @@ const StepOne = ({
   updateNotionPages,
   websitePages = [],
   updateWebsitePages,
-  onFireCrawlJobIdChange,
+  onWebsiteCrawlProviderChange,
+  onWebsiteCrawlJobIdChange,
   crawlOptions,
   onCrawlOptionsChange,
 }: IStepOneProps) => {
@@ -229,7 +231,8 @@ const StepOne = ({
                   onPreview={setCurrentWebsite}
                   checkedCrawlResult={websitePages}
                   onCheckedCrawlResultChange={updateWebsitePages}
-                  onJobIdChange={onFireCrawlJobIdChange}
+                  onCrawlProviderChange={onWebsiteCrawlProviderChange}
+                  onJobIdChange={onWebsiteCrawlJobIdChange}
                   crawlOptions={crawlOptions}
                   onCrawlOptionsChange={onCrawlOptionsChange}
                 />
diff --git a/web/app/components/datasets/create/step-two/index.tsx b/web/app/components/datasets/create/step-two/index.tsx
index f4fc58ee2a2d8..ee01652de8dd2 100644
--- a/web/app/components/datasets/create/step-two/index.tsx
+++ b/web/app/components/datasets/create/step-two/index.tsx
@@ -33,6 +33,7 @@ import { ensureRerankModelSelected, isReRankModelSelected } from '@/app/componen
 import Toast from '@/app/components/base/toast'
 import { formatNumber } from '@/utils/format'
 import type { NotionPage } from '@/models/common'
+import { DataSourceProvider } from '@/models/common'
 import { DataSourceType, DocForm } from '@/models/datasets'
 import NotionIcon from '@/app/components/base/notion-icon'
 import Switch from '@/app/components/base/switch'
@@ -63,7 +64,8 @@ type StepTwoProps = {
   notionPages?: NotionPage[]
   websitePages?: CrawlResultItem[]
   crawlOptions?: CrawlOptions
-  fireCrawlJobId?: string
+  websiteCrawlProvider?: DataSourceProvider
+  websiteCrawlJobId?: string
   onStepChange?: (delta: number) => void
   updateIndexingTypeCache?: (type: string) => void
   updateResultCache?: (res: createDocumentResponse) => void
@@ -94,7 +96,8 @@ const StepTwo = ({
   notionPages = [],
   websitePages = [],
   crawlOptions,
-  fireCrawlJobId = '',
+  websiteCrawlProvider = DataSourceProvider.fireCrawl,
+  websiteCrawlJobId = '',
   onStepChange,
   updateIndexingTypeCache,
   updateResultCache,
@@ -260,8 +263,8 @@ const StepTwo = ({
 
   const getWebsiteInfo = () => {
     return {
-      provider: 'firecrawl',
-      job_id: fireCrawlJobId,
+      provider: websiteCrawlProvider,
+      job_id: websiteCrawlJobId,
       urls: websitePages.map(page => page.source_url),
       only_main_content: crawlOptions?.only_main_content,
     }
diff --git a/web/app/components/datasets/create/website/firecrawl/base/checkbox-with-label.tsx b/web/app/components/datasets/create/website/base/checkbox-with-label.tsx
similarity index 72%
rename from web/app/components/datasets/create/website/firecrawl/base/checkbox-with-label.tsx
rename to web/app/components/datasets/create/website/base/checkbox-with-label.tsx
index 5c574ebe3e619..25d40fe0763da 100644
--- a/web/app/components/datasets/create/website/firecrawl/base/checkbox-with-label.tsx
+++ b/web/app/components/datasets/create/website/base/checkbox-with-label.tsx
@@ -3,6 +3,7 @@ import type { FC } from 'react'
 import React from 'react'
 import cn from '@/utils/classnames'
 import Checkbox from '@/app/components/base/checkbox'
+import Tooltip from '@/app/components/base/tooltip'
 
 type Props = {
   className?: string
@@ -10,6 +11,7 @@ type Props = {
   onChange: (isChecked: boolean) => void
   label: string
   labelClassName?: string
+  tooltip?: string
 }
 
 const CheckboxWithLabel: FC<Props> = ({
@@ -18,11 +20,20 @@ const CheckboxWithLabel: FC<Props> = ({
   onChange,
   label,
   labelClassName,
+  tooltip,
 }) => {
   return (
     <label className={cn(className, 'flex items-center h-7 space-x-2')}>
       <Checkbox checked={isChecked} onCheck={() => onChange(!isChecked)} />
       <div className={cn(labelClassName, 'text-sm font-normal text-gray-800')}>{label}</div>
+      {tooltip && (
+        <Tooltip
+          popupContent={
+            <div className='w-[200px]'>{tooltip}</div>
+          }
+          triggerClassName='ml-0.5 w-4 h-4'
+        />
+      )}
     </label>
   )
 }
diff --git a/web/app/components/datasets/create/website/firecrawl/crawled-result-item.tsx b/web/app/components/datasets/create/website/base/crawled-result-item.tsx
similarity index 100%
rename from web/app/components/datasets/create/website/firecrawl/crawled-result-item.tsx
rename to web/app/components/datasets/create/website/base/crawled-result-item.tsx
diff --git a/web/app/components/datasets/create/website/firecrawl/crawled-result.tsx b/web/app/components/datasets/create/website/base/crawled-result.tsx
similarity index 97%
rename from web/app/components/datasets/create/website/firecrawl/crawled-result.tsx
rename to web/app/components/datasets/create/website/base/crawled-result.tsx
index 2bd51e4d731a9..d5c8d1b80a5a1 100644
--- a/web/app/components/datasets/create/website/firecrawl/crawled-result.tsx
+++ b/web/app/components/datasets/create/website/base/crawled-result.tsx
@@ -2,7 +2,7 @@
 import type { FC } from 'react'
 import React, { useCallback } from 'react'
 import { useTranslation } from 'react-i18next'
-import CheckboxWithLabel from './base/checkbox-with-label'
+import CheckboxWithLabel from './checkbox-with-label'
 import CrawledResultItem from './crawled-result-item'
 import cn from '@/utils/classnames'
 import type { CrawlResultItem } from '@/models/datasets'
diff --git a/web/app/components/datasets/create/website/firecrawl/crawling.tsx b/web/app/components/datasets/create/website/base/crawling.tsx
similarity index 100%
rename from web/app/components/datasets/create/website/firecrawl/crawling.tsx
rename to web/app/components/datasets/create/website/base/crawling.tsx
diff --git a/web/app/components/datasets/create/website/firecrawl/base/error-message.tsx b/web/app/components/datasets/create/website/base/error-message.tsx
similarity index 100%
rename from web/app/components/datasets/create/website/firecrawl/base/error-message.tsx
rename to web/app/components/datasets/create/website/base/error-message.tsx
diff --git a/web/app/components/datasets/create/website/firecrawl/base/field.tsx b/web/app/components/datasets/create/website/base/field.tsx
similarity index 100%
rename from web/app/components/datasets/create/website/firecrawl/base/field.tsx
rename to web/app/components/datasets/create/website/base/field.tsx
diff --git a/web/app/components/datasets/create/website/firecrawl/base/input.tsx b/web/app/components/datasets/create/website/base/input.tsx
similarity index 100%
rename from web/app/components/datasets/create/website/firecrawl/base/input.tsx
rename to web/app/components/datasets/create/website/base/input.tsx
diff --git a/web/app/components/datasets/create/website/firecrawl/mock-crawl-result.ts b/web/app/components/datasets/create/website/base/mock-crawl-result.ts
similarity index 100%
rename from web/app/components/datasets/create/website/firecrawl/mock-crawl-result.ts
rename to web/app/components/datasets/create/website/base/mock-crawl-result.ts
diff --git a/web/app/components/datasets/create/website/firecrawl/base/options-wrap.tsx b/web/app/components/datasets/create/website/base/options-wrap.tsx
similarity index 100%
rename from web/app/components/datasets/create/website/firecrawl/base/options-wrap.tsx
rename to web/app/components/datasets/create/website/base/options-wrap.tsx
diff --git a/web/app/components/datasets/create/website/firecrawl/base/url-input.tsx b/web/app/components/datasets/create/website/base/url-input.tsx
similarity index 100%
rename from web/app/components/datasets/create/website/firecrawl/base/url-input.tsx
rename to web/app/components/datasets/create/website/base/url-input.tsx
diff --git a/web/app/components/datasets/create/website/firecrawl/index.tsx b/web/app/components/datasets/create/website/firecrawl/index.tsx
index de4f8bb129344..aa4dffc174315 100644
--- a/web/app/components/datasets/create/website/firecrawl/index.tsx
+++ b/web/app/components/datasets/create/website/firecrawl/index.tsx
@@ -2,13 +2,13 @@
 import type { FC } from 'react'
 import React, { useCallback, useEffect, useState } from 'react'
 import { useTranslation } from 'react-i18next'
+import UrlInput from '../base/url-input'
+import OptionsWrap from '../base/options-wrap'
+import CrawledResult from '../base/crawled-result'
+import Crawling from '../base/crawling'
+import ErrorMessage from '../base/error-message'
 import Header from './header'
-import UrlInput from './base/url-input'
-import OptionsWrap from './base/options-wrap'
 import Options from './options'
-import CrawledResult from './crawled-result'
-import Crawling from './crawling'
-import ErrorMessage from './base/error-message'
 import cn from '@/utils/classnames'
 import { useModalContext } from '@/context/modal-context'
 import type { CrawlOptions, CrawlResultItem } from '@/models/datasets'
diff --git a/web/app/components/datasets/create/website/firecrawl/options.tsx b/web/app/components/datasets/create/website/firecrawl/options.tsx
index 20cc4f073fe43..8cc2c6757c961 100644
--- a/web/app/components/datasets/create/website/firecrawl/options.tsx
+++ b/web/app/components/datasets/create/website/firecrawl/options.tsx
@@ -2,8 +2,8 @@
 import type { FC } from 'react'
 import React, { useCallback } from 'react'
 import { useTranslation } from 'react-i18next'
-import CheckboxWithLabel from './base/checkbox-with-label'
-import Field from './base/field'
+import CheckboxWithLabel from '../base/checkbox-with-label'
+import Field from '../base/field'
 import cn from '@/utils/classnames'
 import type { CrawlOptions } from '@/models/datasets'
 
diff --git a/web/app/components/datasets/create/website/index.module.css b/web/app/components/datasets/create/website/index.module.css
new file mode 100644
index 0000000000000..abaab4bea4b7a
--- /dev/null
+++ b/web/app/components/datasets/create/website/index.module.css
@@ -0,0 +1,6 @@
+.jinaLogo {
+  @apply w-4 h-4 bg-center bg-no-repeat inline-block;
+  background-color: #F5FAFF;
+  background-image: url(../assets/jina.png);
+  background-size: 16px;
+}
diff --git a/web/app/components/datasets/create/website/index.tsx b/web/app/components/datasets/create/website/index.tsx
index e06fbb4a1210b..58b7f5f2fd77b 100644
--- a/web/app/components/datasets/create/website/index.tsx
+++ b/web/app/components/datasets/create/website/index.tsx
@@ -1,8 +1,12 @@
 'use client'
 import type { FC } from 'react'
 import React, { useCallback, useEffect, useState } from 'react'
+import { useTranslation } from 'react-i18next'
+import s from './index.module.css'
 import NoData from './no-data'
 import Firecrawl from './firecrawl'
+import JinaReader from './jina-reader'
+import cn from '@/utils/classnames'
 import { useModalContext } from '@/context/modal-context'
 import type { CrawlOptions, CrawlResultItem } from '@/models/datasets'
 import { fetchDataSources } from '@/service/datasets'
@@ -12,6 +16,7 @@ type Props = {
   onPreview: (payload: CrawlResultItem) => void
   checkedCrawlResult: CrawlResultItem[]
   onCheckedCrawlResultChange: (payload: CrawlResultItem[]) => void
+  onCrawlProviderChange: (provider: DataSourceProvider) => void
   onJobIdChange: (jobId: string) => void
   crawlOptions: CrawlOptions
   onCrawlOptionsChange: (payload: CrawlOptions) => void
@@ -21,17 +26,32 @@ const Website: FC<Props> = ({
   onPreview,
   checkedCrawlResult,
   onCheckedCrawlResultChange,
+  onCrawlProviderChange,
   onJobIdChange,
   crawlOptions,
   onCrawlOptionsChange,
 }) => {
+  const { t } = useTranslation()
   const { setShowAccountSettingModal } = useModalContext()
   const [isLoaded, setIsLoaded] = useState(false)
-  const [isSetFirecrawlApiKey, setIsSetFirecrawlApiKey] = useState(false)
+  const [selectedProvider, setSelectedProvider] = useState<DataSourceProvider>(DataSourceProvider.jinaReader)
+  const [sources, setSources] = useState<DataSourceItem[]>([])
+
+  useEffect(() => {
+    onCrawlProviderChange(selectedProvider)
+  }, [selectedProvider, onCrawlProviderChange])
+
   const checkSetApiKey = useCallback(async () => {
     const res = await fetchDataSources() as any
-    const isFirecrawlSet = res.sources.some((item: DataSourceItem) => item.provider === DataSourceProvider.fireCrawl)
-    setIsSetFirecrawlApiKey(isFirecrawlSet)
+    setSources(res.sources)
+
+    // If users have configured one of the providers, select it.
+    const availableProviders = res.sources.filter((item: DataSourceItem) =>
+      [DataSourceProvider.jinaReader, DataSourceProvider.fireCrawl].includes(item.provider),
+    )
+
+    if (availableProviders.length > 0)
+      setSelectedProvider(availableProviders[0].provider)
   }, [])
 
   useEffect(() => {
@@ -52,20 +72,66 @@ const Website: FC<Props> = ({
 
   return (
     <div>
-      {isSetFirecrawlApiKey
-        ? (
-          <Firecrawl
-            onPreview={onPreview}
-            checkedCrawlResult={checkedCrawlResult}
-            onCheckedCrawlResultChange={onCheckedCrawlResultChange}
-            onJobIdChange={onJobIdChange}
-            crawlOptions={crawlOptions}
-            onCrawlOptionsChange={onCrawlOptionsChange}
-          />
-        )
-        : (
-          <NoData onConfig={handleOnConfig} />
-        )}
+      <div className="mb-4">
+        <div className="font-medium text-gray-700 mb-2 h-6">
+          {t('datasetCreation.stepOne.website.chooseProvider')}
+        </div>
+        <div className="flex space-x-2">
+          <button
+            className={`px-4 py-2 text-sm font-medium rounded-md flex items-center justify-center ${
+              selectedProvider === DataSourceProvider.jinaReader
+                ? 'bg-primary-50 text-primary-600'
+                : 'bg-gray-100 text-gray-600 hover:bg-gray-200'
+            }`}
+            onClick={() => setSelectedProvider(DataSourceProvider.jinaReader)}
+          >
+            <span className={cn(s.jinaLogo, 'mr-2')} />
+            <span>Jina Reader</span>
+          </button>
+          <button
+            className={`px-4 py-2 text-sm font-medium rounded-md ${
+              selectedProvider === DataSourceProvider.fireCrawl
+                ? 'bg-primary-50 text-primary-600'
+                : 'bg-gray-100 text-gray-600 hover:bg-gray-200'
+            }`}
+            onClick={() => setSelectedProvider(DataSourceProvider.fireCrawl)}
+          >
+            🔥 Firecrawl
+          </button>
+        </div>
+      </div>
+
+      {
+        selectedProvider === DataSourceProvider.fireCrawl
+          ? sources.find(source => source.provider === DataSourceProvider.fireCrawl)
+            ? (
+              <Firecrawl
+                onPreview={onPreview}
+                checkedCrawlResult={checkedCrawlResult}
+                onCheckedCrawlResultChange={onCheckedCrawlResultChange}
+                onJobIdChange={onJobIdChange}
+                crawlOptions={crawlOptions}
+                onCrawlOptionsChange={onCrawlOptionsChange}
+              />
+            )
+            : (
+              <NoData onConfig={handleOnConfig} provider={selectedProvider} />
+            )
+          : sources.find(source => source.provider === DataSourceProvider.jinaReader)
+            ? (
+              <JinaReader
+                onPreview={onPreview}
+                checkedCrawlResult={checkedCrawlResult}
+                onCheckedCrawlResultChange={onCheckedCrawlResultChange}
+                onJobIdChange={onJobIdChange}
+                crawlOptions={crawlOptions}
+                onCrawlOptionsChange={onCrawlOptionsChange}
+              />
+            )
+            : (
+              <NoData onConfig={handleOnConfig} provider={selectedProvider} />
+            )
+      }
     </div>
   )
 }
diff --git a/web/app/components/datasets/create/website/jina-reader/header.tsx b/web/app/components/datasets/create/website/jina-reader/header.tsx
new file mode 100644
index 0000000000000..85014a30ee2b1
--- /dev/null
+++ b/web/app/components/datasets/create/website/jina-reader/header.tsx
@@ -0,0 +1,42 @@
+'use client'
+import type { FC } from 'react'
+import React from 'react'
+import { useTranslation } from 'react-i18next'
+import { Settings01 } from '@/app/components/base/icons/src/vender/line/general'
+import { BookOpen01 } from '@/app/components/base/icons/src/vender/line/education'
+
+const I18N_PREFIX = 'datasetCreation.stepOne.website'
+
+type Props = {
+  onSetting: () => void
+}
+
+const Header: FC<Props> = ({
+  onSetting,
+}) => {
+  const { t } = useTranslation()
+
+  return (
+    <div className='flex h-6 items-center justify-between'>
+      <div className='flex items-center'>
+        <div className='text-base font-medium text-gray-700'>{t(`${I18N_PREFIX}.jinaReaderTitle`)}</div>
+        <div className='ml-2 mr-1 w-px h-3.5 bg-gray-200'></div>
+        <div
+          className='p-1 rounded-md hover:bg-black/5 cursor-pointer'
+          onClick={onSetting}
+        >
+          <Settings01 className='w-3.5 h-3.5 text-gray-500' />
+        </div>
+      </div>
+      <a
+        href='https://jina.ai/reader'
+        target='_blank' rel='noopener noreferrer'
+        className='flex items-center text-xs text-primary-600'
+      >
+        <BookOpen01 className='mr-1 w-3.5 h-3.5 text-primary-600' />
+        {t(`${I18N_PREFIX}.jinaReaderDoc`)}
+      </a>
+    </div>
+  )
+}
+export default React.memo(Header)
diff --git a/web/app/components/datasets/create/website/jina-reader/index.tsx b/web/app/components/datasets/create/website/jina-reader/index.tsx
new file mode 100644
index 0000000000000..51d77d712140b
--- /dev/null
+++ b/web/app/components/datasets/create/website/jina-reader/index.tsx
@@ -0,0 +1,232 @@
+'use client'
+import type { FC } from 'react'
+import React, { useCallback, useEffect, useState } from 'react'
+import { useTranslation } from 'react-i18next'
+import UrlInput from '../base/url-input'
+import OptionsWrap from '../base/options-wrap'
+import CrawledResult from '../base/crawled-result'
+import Crawling from '../base/crawling'
+import ErrorMessage from '../base/error-message'
+import Header from './header'
+import Options from './options'
+import cn from '@/utils/classnames'
+import { useModalContext } from '@/context/modal-context'
+import Toast from '@/app/components/base/toast'
+import { checkJinaReaderTaskStatus, createJinaReaderTask } from '@/service/datasets'
+import { sleep } from '@/utils'
+import type { CrawlOptions, CrawlResultItem } from '@/models/datasets'
+
+const ERROR_I18N_PREFIX = 'common.errorMsg'
+const I18N_PREFIX = 'datasetCreation.stepOne.website'
+
+type Props = {
+  onPreview: (payload: CrawlResultItem) => void
+  checkedCrawlResult: CrawlResultItem[]
+  onCheckedCrawlResultChange: (payload: CrawlResultItem[]) => void
+  onJobIdChange: (jobId: string) => void
+  crawlOptions: CrawlOptions
+  onCrawlOptionsChange: (payload: CrawlOptions) => void
+}
+
+enum Step {
+  init = 'init',
+  running = 'running',
+  finished = 'finished',
+}
+
+const JinaReader: FC<Props> = ({
+  onPreview,
+  checkedCrawlResult,
+  onCheckedCrawlResultChange,
+  onJobIdChange,
+  crawlOptions,
+  onCrawlOptionsChange,
+}) => {
+  const { t } = useTranslation()
+  const [step, setStep] = useState<Step>(Step.init)
+  const [controlFoldOptions, setControlFoldOptions] = useState<number>(0)
+  useEffect(() => {
+    if (step !== Step.init)
+      setControlFoldOptions(Date.now())
+  }, [step])
+  const { setShowAccountSettingModal } = useModalContext()
+  const handleSetting = useCallback(() => {
+    setShowAccountSettingModal({
+      payload: 'data-source',
+    })
+  }, [setShowAccountSettingModal])
+
+  const checkValid = useCallback((url: string) => {
+    let errorMsg = ''
+    if (!url) {
+      errorMsg = t(`${ERROR_I18N_PREFIX}.fieldRequired`, {
+        field: 'url',
+      })
+    }
+
+    if (!errorMsg && !((url.startsWith('http://') || url.startsWith('https://'))))
+      errorMsg = t(`${ERROR_I18N_PREFIX}.urlError`)
+
+    if (!errorMsg && (crawlOptions.limit === null || crawlOptions.limit === undefined || crawlOptions.limit === '')) {
+      errorMsg = t(`${ERROR_I18N_PREFIX}.fieldRequired`, {
+        field: t(`${I18N_PREFIX}.limit`),
+      })
+    }
+
+    return {
+      isValid: !errorMsg,
+      errorMsg,
+    }
+  }, [crawlOptions, t])
+
+  const isInit = step === Step.init
+  const isCrawlFinished = step === Step.finished
+  const isRunning = step === Step.running
+  const [crawlResult, setCrawlResult] = useState<{
+    current: number
+    total: number
+    data: CrawlResultItem[]
+    time_consuming: number | string
+  } | undefined>(undefined)
+  const [crawlErrorMessage, setCrawlErrorMessage] = useState('')
+  const showError = isCrawlFinished && crawlErrorMessage
+
+  const waitForCrawlFinished = useCallback(async (jobId: string) => {
+    try {
+      const res = await checkJinaReaderTaskStatus(jobId) as any
+      console.log('res', res)
+      if (res.status === 'completed') {
+        return {
+          isError: false,
+          data: {
+            ...res,
+            total: Math.min(res.total, parseFloat(crawlOptions.limit as string)),
+          },
+        }
+      }
+      if (res.status === 'failed' || !res.status) {
+        return {
+          isError: true,
+          errorMessage: res.message,
+          data: {
+            data: [],
+          },
+        }
+      }
+      // update the progress
+      setCrawlResult({
+        ...res,
+        total: Math.min(res.total, parseFloat(crawlOptions.limit as string)),
+      })
+      onCheckedCrawlResultChange(res.data || []) // default select the crawl result
+      await sleep(2500)
+      return await waitForCrawlFinished(jobId)
+    }
+    catch (e: any) {
+      const errorBody = await e.json()
+      return {
+        isError: true,
+        errorMessage: errorBody.message,
+        data: {
+          data: [],
+        },
+      }
+    }
+  }, [crawlOptions.limit])
+
+  const handleRun = useCallback(async (url: string) => {
+    const { isValid, errorMsg } = checkValid(url)
+    if (!isValid) {
+      Toast.notify({
+        message: errorMsg!,
+        type: 'error',
+      })
+      return
+    }
+    setStep(Step.running)
+    try {
+      const startTime = Date.now()
+      const res = await createJinaReaderTask({
+        url,
+        options: crawlOptions,
+      }) as any
+
+      if (res.data) {
+        const data = {
+          current: 1,
+          total: 1,
+          data: [{
+            title: res.data.title,
+            markdown: res.data.content,
+            description: res.data.description,
+            source_url: res.data.url,
+          }],
+          time_consuming: (Date.now() - startTime) / 1000,
+        }
+        setCrawlResult(data)
+        onCheckedCrawlResultChange(data.data || [])
+        setCrawlErrorMessage('')
+      }
+      else if (res.job_id) {
+        const jobId = res.job_id
+        onJobIdChange(jobId)
+        const { isError, data, errorMessage } = await waitForCrawlFinished(jobId)
+        if (isError) {
+          setCrawlErrorMessage(errorMessage || t(`${I18N_PREFIX}.unknownError`))
+        }
+        else {
+          setCrawlResult(data)
+          onCheckedCrawlResultChange(data.data || []) // default select the crawl result
+          setCrawlErrorMessage('')
+        }
+      }
+    }
+    catch (e) {
+      setCrawlErrorMessage(t(`${I18N_PREFIX}.unknownError`)!)
+      console.log(e)
+    }
+    finally {
+      setStep(Step.finished)
+    }
+  }, [checkValid, crawlOptions, onJobIdChange, t, waitForCrawlFinished])
+
+  return (
+    <div>
+      <Header onSetting={handleSetting} />
+      <div className={cn('mt-2 p-4 pb-0 rounded-xl border border-gray-200')}>
+        <UrlInput onRun={handleRun} isRunning={isRunning} />
+        <OptionsWrap
+          className={cn('mt-4')}
+          controlFoldOptions={controlFoldOptions}
+        >
+          <Options className='mt-2' payload={crawlOptions} onChange={onCrawlOptionsChange} />
+        </OptionsWrap>
+
+        {!isInit && (
+          <div className='mt-3 relative left-[-16px] w-[calc(100%_+_32px)] rounded-b-xl'>
+            {isRunning
+              && <Crawling
+                className='mt-2'
+                crawledNum={crawlResult?.current || 0}
+                totalNum={crawlResult?.total || parseFloat(crawlOptions.limit as string) || 0}
+              />}
+            {showError && (
+              <ErrorMessage className='rounded-b-xl' title={t(`${I18N_PREFIX}.exceptionErrorTitle`)} errorMsg={crawlErrorMessage} />
+            )}
+            {isCrawlFinished && !showError
+              && <CrawledResult
+                className='mb-2'
+                list={crawlResult?.data || []}
+                checkedList={checkedCrawlResult}
+                onSelectedChange={onCheckedCrawlResultChange}
+                onPreview={onPreview}
+                usedTime={parseFloat(crawlResult?.time_consuming as string) || 0}
+              />
+            }
+          </div>
+        )}
+      </div>
+    </div>
+  )
+}
+export default React.memo(JinaReader)
diff --git a/web/app/components/datasets/create/website/jina-reader/options.tsx b/web/app/components/datasets/create/website/jina-reader/options.tsx
new file mode 100644
index 0000000000000..52cfaa8b3b40f
--- /dev/null
+++ b/web/app/components/datasets/create/website/jina-reader/options.tsx
@@ -0,0 +1,59 @@
+'use client'
+import type { FC } from 'react'
+import React, { useCallback } from 'react'
+import { useTranslation } from 'react-i18next'
+import CheckboxWithLabel from '../base/checkbox-with-label'
+import Field from '../base/field'
+import cn from '@/utils/classnames'
+import type { CrawlOptions } from '@/models/datasets'
+
+const I18N_PREFIX = 'datasetCreation.stepOne.website'
+
+type Props = {
+  className?: string
+  payload: CrawlOptions
+  onChange: (payload: CrawlOptions) => void
+}
+
+const Options: FC<Props> = ({
+  className = '',
+  payload,
+  onChange,
+}) => {
+  const { t } = useTranslation()
+
+  const handleChange = useCallback((key: keyof CrawlOptions) => {
+    return (value: any) => {
+      onChange({
+        ...payload,
+        [key]: value,
+      })
+    }
+  }, [payload, onChange])
+  return (
+    <div className={cn(className, ' space-y-2')}>
+      <CheckboxWithLabel
+        label={t(`${I18N_PREFIX}.crawlSubPage`)}
+        isChecked={payload.crawl_sub_pages}
+        onChange={handleChange('crawl_sub_pages')}
+      />
+      <CheckboxWithLabel
+        label={t(`${I18N_PREFIX}.useSitemap`)}
+        isChecked={payload.use_sitemap}
+        onChange={handleChange('use_sitemap')}
+        tooltip={t(`${I18N_PREFIX}.useSitemapTooltip`) as string}
+      />
+      <div className='flex justify-between space-x-4'>
+        <Field
+          className='grow shrink-0'
+          label={t(`${I18N_PREFIX}.limit`)}
+          value={payload.limit}
+          onChange={handleChange('limit')}
+          isNumber
+          isRequired
+        />
+      </div>
+    </div>
+  )
+}
+export default React.memo(Options)
diff --git a/web/app/components/datasets/create/website/no-data.tsx b/web/app/components/datasets/create/website/no-data.tsx
index 13e5ee7dfbd50..8a508a48c6bb8 100644
--- a/web/app/components/datasets/create/website/no-data.tsx
+++ b/web/app/components/datasets/create/website/no-data.tsx
@@ -2,35 +2,56 @@
 import type { FC } from 'react'
 import React from 'react'
 import { useTranslation } from 'react-i18next'
+import s from './index.module.css'
 import { Icon3Dots } from '@/app/components/base/icons/src/vender/line/others'
 import Button from '@/app/components/base/button'
+import { DataSourceProvider } from '@/models/common'
 
 const I18N_PREFIX = 'datasetCreation.stepOne.website'
 
 type Props = {
   onConfig: () => void
+  provider: DataSourceProvider
 }
 
 const NoData: FC<Props> = ({
   onConfig,
+  provider,
 }) => {
   const { t } = useTranslation()
 
+  const providerConfig = {
+    [DataSourceProvider.jinaReader]: {
+      emoji: <span className={s.jinaLogo} />,
+      title: t(`${I18N_PREFIX}.jinaReaderNotConfigured`),
+      description: t(`${I18N_PREFIX}.jinaReaderNotConfiguredDescription`),
+    },
+    [DataSourceProvider.fireCrawl]: {
+      emoji: '🔥',
+      title: t(`${I18N_PREFIX}.fireCrawlNotConfigured`),
+      description: t(`${I18N_PREFIX}.fireCrawlNotConfiguredDescription`),
+    },
+  }
+
+  const currentProvider = providerConfig[provider]
+
   return (
-    <div className='max-w-[640px] p-6 rounded-2xl bg-gray-50'>
-      <div className='flex w-11 h-11 items-center justify-center bg-gray-50 rounded-xl border-[0.5px] border-gray-100 shadow-lg'>
-        🔥
-      </div>
-      <div className='my-2'>
-        <span className='text-gray-700 font-semibold'>{t(`${I18N_PREFIX}.fireCrawlNotConfigured`)}<Icon3Dots className='inline relative -top-3 -left-1.5' /></span>
-        <div className='mt-1 pb-3 text-gray-500 text-[13px] font-normal'>
-          {t(`${I18N_PREFIX}.fireCrawlNotConfiguredDescription`)}
+    <>
+      <div className='max-w-[640px] p-6 rounded-2xl bg-gray-50 mt-4'>
+        <div className='flex w-11 h-11 items-center justify-center bg-gray-50 rounded-xl border-[0.5px] border-gray-100 shadow-lg'>
+          {currentProvider.emoji}
+        </div>
+        <div className='my-2'>
+          <span className='text-gray-700 font-semibold'>{currentProvider.title}<Icon3Dots className='inline relative -top-3 -left-1.5' /></span>
+          <div className='mt-1 pb-3 text-gray-500 text-[13px] font-normal'>
+            {currentProvider.description}
+          </div>
         </div>
+        <Button variant='primary' onClick={onConfig}>
+          {t(`${I18N_PREFIX}.configure`)}
+        </Button>
       </div>
-      <Button variant='primary' onClick={onConfig}>
-        {t(`${I18N_PREFIX}.configure`)}
-      </Button>
-    </div>
+    </>
   )
 }
 export default React.memo(NoData)
diff --git a/web/app/components/header/account-setting/data-source-page/data-source-website/config-firecrawl-modal.tsx b/web/app/components/header/account-setting/data-source-page/data-source-website/config-firecrawl-modal.tsx
index d68fc79b0d5a7..a4a8b9b63722b 100644
--- a/web/app/components/header/account-setting/data-source-page/data-source-website/config-firecrawl-modal.tsx
+++ b/web/app/components/header/account-setting/data-source-page/data-source-website/config-firecrawl-modal.tsx
@@ -9,7 +9,7 @@ import {
 import { Lock01 } from '@/app/components/base/icons/src/vender/solid/security'
 import Button from '@/app/components/base/button'
 import type { FirecrawlConfig } from '@/models/common'
-import Field from '@/app/components/datasets/create/website/firecrawl/base/field'
+import Field from '@/app/components/datasets/create/website/base/field'
 import Toast from '@/app/components/base/toast'
 import { createDataSourceApiKeyBinding } from '@/service/datasets'
 import { LinkExternal02 } from '@/app/components/base/icons/src/vender/line/general'
diff --git a/web/app/components/header/account-setting/data-source-page/data-source-website/config-jina-reader-modal.tsx b/web/app/components/header/account-setting/data-source-page/data-source-website/config-jina-reader-modal.tsx
new file mode 100644
index 0000000000000..c6d6ad02565cb
--- /dev/null
+++ b/web/app/components/header/account-setting/data-source-page/data-source-website/config-jina-reader-modal.tsx
@@ -0,0 +1,140 @@
+'use client'
+import type { FC } from 'react'
+import React, { useCallback, useState } from 'react'
+import { useTranslation } from 'react-i18next'
+import {
+  PortalToFollowElem,
+  PortalToFollowElemContent,
+} from '@/app/components/base/portal-to-follow-elem'
+import { Lock01 } from '@/app/components/base/icons/src/vender/solid/security'
+import Button from '@/app/components/base/button'
+import { DataSourceProvider } from '@/models/common'
+import Field from '@/app/components/datasets/create/website/base/field'
+import Toast from '@/app/components/base/toast'
+import { createDataSourceApiKeyBinding } from '@/service/datasets'
+import { LinkExternal02 } from '@/app/components/base/icons/src/vender/line/general'
+type Props = {
+  onCancel: () => void
+  onSaved: () => void
+}
+
+const I18N_PREFIX = 'datasetCreation.jinaReader'
+
+const ConfigJinaReaderModal: FC<Props> = ({
+  onCancel,
+  onSaved,
+}) => {
+  const { t } = useTranslation()
+  const [isSaving, setIsSaving] = useState(false)
+  const [apiKey, setApiKey] = useState('')
+
+  const handleSave = useCallback(async () => {
+    if (isSaving)
+      return
+    let errorMsg = ''
+    if (!errorMsg) {
+      if (!apiKey) {
+        errorMsg = t('common.errorMsg.fieldRequired', {
+          field: 'API Key',
+        })
+      }
+    }
+
+    if (errorMsg) {
+      Toast.notify({
+        type: 'error',
+        message: errorMsg,
+      })
+      return
+    }
+    const postData = {
+      category: 'website',
+      provider: DataSourceProvider.jinaReader,
+      credentials: {
+        auth_type: 'bearer',
+        config: {
+          api_key: apiKey,
+        },
+      },
+    }
+    try {
+      setIsSaving(true)
+      await createDataSourceApiKeyBinding(postData)
+      Toast.notify({
+        type: 'success',
+        message: t('common.api.success'),
+      })
+    }
+    finally {
+      setIsSaving(false)
+    }
+
+    onSaved()
+  }, [apiKey, onSaved, t, isSaving])
+
+  return (
+    <PortalToFollowElem open>
+      <PortalToFollowElemContent className='w-full h-full z-[60]'>
+        <div className='fixed inset-0 flex items-center justify-center bg-black/[.25]'>
+          <div className='mx-2 w-[640px] max-h-[calc(100vh-120px)] bg-white shadow-xl rounded-2xl overflow-y-auto'>
+            <div className='px-8 pt-8'>
+              <div className='flex justify-between items-center mb-4'>
+                <div className='text-xl font-semibold text-gray-900'>{t(`${I18N_PREFIX}.configJinaReader`)}</div>
+              </div>
+
+              <div className='space-y-4'>
+                <Field
+                  label='API Key'
+                  labelClassName='!text-sm'
+                  isRequired
+                  value={apiKey}
+                  onChange={(value: string | number) => setApiKey(value as string)}
+                  placeholder={t(`${I18N_PREFIX}.apiKeyPlaceholder`)!}
+                />
+              </div>
+              <div className='my-8 flex justify-between items-center h-8'>
+                <a className='flex items-center space-x-1 leading-[18px] text-xs font-normal text-[#155EEF]' target='_blank' href='https://jina.ai/reader/'>
+                  <span>{t(`${I18N_PREFIX}.getApiKeyLinkText`)}</span>
+                  <LinkExternal02 className='w-3 h-3' />
+                </a>
+                <div className='flex'>
+                  <Button
+                    size='large'
+                    className='mr-2'
+                    onClick={onCancel}
+                  >
+                    {t('common.operation.cancel')}
+                  </Button>
+                  <Button
+                    variant='primary'
+                    size='large'
+                    onClick={handleSave}
+                    loading={isSaving}
+                  >
+                    {t('common.operation.save')}
+                  </Button>
+                </div>
+
+              </div>
+            </div>
+            <div className='border-t-[0.5px] border-t-black/5'>
+              <div className='flex justify-center items-center py-3 bg-gray-50 text-xs text-gray-500'>
+                <Lock01 className='mr-1 w-3 h-3 text-gray-500' />
+                {t('common.modelProvider.encrypted.front')}
+                <a
+                  className='text-primary-600 mx-1'
+                  target='_blank' rel='noopener noreferrer'
+                  href='https://pycryptodome.readthedocs.io/en/latest/src/cipher/oaep.html'
+                >
+                  PKCS1_OAEP
+                </a>
+                {t('common.modelProvider.encrypted.back')}
+              </div>
+            </div>
+          </div>
+        </div>
+      </PortalToFollowElemContent>
+    </PortalToFollowElem>
+  )
+}
+export default React.memo(ConfigJinaReaderModal)
diff --git a/web/app/components/header/account-setting/data-source-page/data-source-website/index.tsx b/web/app/components/header/account-setting/data-source-page/data-source-website/index.tsx
index 21f7660ef1dd1..628510c5dd387 100644
--- a/web/app/components/header/account-setting/data-source-page/data-source-website/index.tsx
+++ b/web/app/components/header/account-setting/data-source-page/data-source-website/index.tsx
@@ -2,11 +2,12 @@
 import type { FC } from 'react'
 import React, { useCallback, useEffect, useState } from 'react'
 import { useTranslation } from 'react-i18next'
-import { useBoolean } from 'ahooks'
 import Panel from '../panel'
 import { DataSourceType } from '../panel/types'
 import ConfigFirecrawlModal from './config-firecrawl-modal'
+import ConfigJinaReaderModal from './config-jina-reader-modal'
 import cn from '@/utils/classnames'
+import s from '@/app/components/datasets/create/website/index.module.css'
 import { fetchDataSources, removeDataSourceApiKeyBinding } from '@/service/datasets'
 
 import type {
@@ -19,9 +20,11 @@ import {
 } from '@/models/common'
 import Toast from '@/app/components/base/toast'
 
-type Props = {}
+type Props = {
+  provider: DataSourceProvider
+}
 
-const DataSourceWebsite: FC<Props> = () => {
+const DataSourceWebsite: FC<Props> = ({ provider }) => {
   const { t } = useTranslation()
   const { isCurrentWorkspaceManager } = useAppContext()
   const [sources, setSources] = useState<DataSourceItem[]>([])
@@ -36,22 +39,26 @@ const DataSourceWebsite: FC<Props> = () => {
     // eslint-disable-next-line react-hooks/exhaustive-deps
   }, [])
 
-  const [isShowConfig, {
-    setTrue: showConfig,
-    setFalse: hideConfig,
-  }] = useBoolean(false)
+  const [configTarget, setConfigTarget] = useState<DataSourceProvider | null>(null)
+  const showConfig = useCallback((provider: DataSourceProvider) => {
+    setConfigTarget(provider)
+  }, [setConfigTarget])
+
+  const hideConfig = useCallback(() => {
+    setConfigTarget(null)
+  }, [setConfigTarget])
 
   const handleAdded = useCallback(() => {
     checkSetApiKey()
     hideConfig()
   }, [checkSetApiKey, hideConfig])
 
-  const getIdByProvider = (provider: string): string | undefined => {
+  const getIdByProvider = (provider: DataSourceProvider): string | undefined => {
     const source = sources.find(item => item.provider === provider)
     return source?.id
   }
 
-  const handleRemove = useCallback((provider: string) => {
+  const handleRemove = useCallback((provider: DataSourceProvider) => {
     return async () => {
       const dataSourceId = getIdByProvider(provider)
       if (dataSourceId) {
@@ -69,22 +76,34 @@ const DataSourceWebsite: FC<Props> = () => {
     <>
       <Panel
         type={DataSourceType.website}
-        isConfigured={sources.length > 0}
-        onConfigure={showConfig}
+        provider={provider}
+        isConfigured={sources.find(item => item.provider === provider) !== undefined}
+        onConfigure={() => showConfig(provider)}
         readOnly={!isCurrentWorkspaceManager}
-        configuredList={sources.map(item => ({
+        configuredList={sources.filter(item => item.provider === provider).map(item => ({
           id: item.id,
           logo: ({ className }: { className: string }) => (
-            <div className={cn(className, 'flex items-center justify-center w-5 h-5 bg-white border border-gray-100 text-xs font-medium text-gray-500 rounded ml-3')}>🔥</div>
+            item.provider === DataSourceProvider.fireCrawl
+              ? (
+                <div className={cn(className, 'flex items-center justify-center w-5 h-5 bg-white border border-gray-100 text-xs font-medium text-gray-500 rounded ml-3')}>🔥</div>
+              )
+              : (
+                <div className={cn(className, 'flex items-center justify-center w-5 h-5 bg-white border border-gray-100 text-xs font-medium text-gray-500 rounded ml-3')}>
+                  <span className={s.jinaLogo} />
+                </div>
+              )
           ),
-          name: 'Firecrawl',
+          name: item.provider === DataSourceProvider.fireCrawl ? 'Firecrawl' : 'Jina Reader',
           isActive: true,
         }))}
-        onRemove={handleRemove(DataSourceProvider.fireCrawl)}
+        onRemove={handleRemove(provider)}
       />
-      {isShowConfig && (
+      {configTarget === DataSourceProvider.fireCrawl && (
         <ConfigFirecrawlModal onSaved={handleAdded} onCancel={hideConfig} />
       )}
+      {configTarget === DataSourceProvider.jinaReader && (
+        <ConfigJinaReaderModal onSaved={handleAdded} onCancel={hideConfig} />
+      )}
     </>
 
   )
diff --git a/web/app/components/header/account-setting/data-source-page/index.tsx b/web/app/components/header/account-setting/data-source-page/index.tsx
index ede83152b223e..c3da977ca4e20 100644
--- a/web/app/components/header/account-setting/data-source-page/index.tsx
+++ b/web/app/components/header/account-setting/data-source-page/index.tsx
@@ -3,6 +3,7 @@ import { useTranslation } from 'react-i18next'
 import DataSourceNotion from './data-source-notion'
 import DataSourceWebsite from './data-source-website'
 import { fetchDataSource } from '@/service/common'
+import { DataSourceProvider } from '@/models/common'
 
 export default function DataSourcePage() {
   const { t } = useTranslation()
@@ -13,7 +14,8 @@ export default function DataSourcePage() {
     <div className='mb-8'>
       <div className='mb-2 text-sm font-medium text-gray-900'>{t('common.dataSource.add')}</div>
       <DataSourceNotion workspaces={notionWorkspaces} />
-      <DataSourceWebsite />
+      <DataSourceWebsite provider={DataSourceProvider.jinaReader} />
+      <DataSourceWebsite provider={DataSourceProvider.fireCrawl} />
     </div>
   )
 }
diff --git a/web/app/components/header/account-setting/data-source-page/panel/index.tsx b/web/app/components/header/account-setting/data-source-page/panel/index.tsx
index 988aedcaf7476..4a810020b440e 100644
--- a/web/app/components/header/account-setting/data-source-page/panel/index.tsx
+++ b/web/app/components/header/account-setting/data-source-page/panel/index.tsx
@@ -8,10 +8,12 @@ import ConfigItem from './config-item'
 
 import s from './style.module.css'
 import { DataSourceType } from './types'
+import { DataSourceProvider } from '@/models/common'
 import cn from '@/utils/classnames'
 
 type Props = {
   type: DataSourceType
+  provider: DataSourceProvider
   isConfigured: boolean
   onConfigure: () => void
   readOnly: boolean
@@ -25,6 +27,7 @@ type Props = {
 
 const Panel: FC<Props> = ({
   type,
+  provider,
   isConfigured,
   onConfigure,
   readOnly,
@@ -46,7 +49,7 @@ const Panel: FC<Props> = ({
             <div className='text-sm font-medium text-gray-800'>{t(`common.dataSource.${type}.title`)}</div>
             {isWebsite && (
               <div className='ml-1 leading-[18px] px-1.5 rounded-md bg-white border border-gray-100 text-xs font-medium text-gray-700'>
-                <span className='text-gray-500'>{t('common.dataSource.website.with')}</span> 🔥 Firecrawl
+                <span className='text-gray-500'>{t('common.dataSource.website.with')}</span> { provider === DataSourceProvider.fireCrawl ? '🔥 Firecrawl' : 'Jina Reader'}
               </div>
             )}
           </div>
diff --git a/web/i18n/en-US/dataset-creation.ts b/web/i18n/en-US/dataset-creation.ts
index 32f9d596ca6ac..1849b12757f9d 100644
--- a/web/i18n/en-US/dataset-creation.ts
+++ b/web/i18n/en-US/dataset-creation.ts
@@ -16,6 +16,11 @@ const translation = {
     apiKeyPlaceholder: 'API key from firecrawl.dev',
     getApiKeyLinkText: 'Get your API key from firecrawl.dev',
   },
+  jinaReader: {
+    configJinaReader: 'Configure Jina Reader',
+    apiKeyPlaceholder: 'API key from jina.ai',
+    getApiKeyLinkText: 'Get your free API key at jina.ai',
+  },
   stepOne: {
     filePreview: 'File Preview',
     pagePreview: 'Page Preview',
@@ -56,13 +61,21 @@ const translation = {
       failed: 'Creation failed',
     },
     website: {
+      chooseProvider: 'Select a provider',
       fireCrawlNotConfigured: 'Firecrawl is not configured',
       fireCrawlNotConfiguredDescription: 'Configure Firecrawl with API key to use it.',
+      jinaReaderNotConfigured: 'Jina Reader is not configured',
+      jinaReaderNotConfiguredDescription: 'Set up Jina Reader by entering your free API key for access.',
       configure: 'Configure',
       run: 'Run',
       firecrawlTitle: 'Extract web content with 🔥Firecrawl',
       firecrawlDoc: 'Firecrawl docs',
       firecrawlDocLink: 'https://docs.dify.ai/guides/knowledge-base/sync-from-website',
+      jinaReaderTitle: 'Convert the entire site to Markdown',
+      jinaReaderDoc: 'Learn more about Jina Reader',
+      jinaReaderDocLink: 'https://jina.ai/reader',
+      useSitemap: 'Use sitemap',
+      useSitemapTooltip: 'Follow the sitemap to crawl the site. If not, Jina Reader will crawl iteratively based on page relevance, yielding fewer but higher-quality pages.',
       options: 'Options',
       crawlSubPage: 'Crawl sub-pages',
       limit: 'Limit',
@@ -70,7 +83,7 @@ const translation = {
       excludePaths: 'Exclude paths',
       includeOnlyPaths: 'Include only paths',
       extractOnlyMainContent: 'Extract only main content (no headers, navs, footers, etc.)',
-      exceptionErrorTitle: 'An exception occurred while running Firecrawl job:',
+      exceptionErrorTitle: 'An exception occurred while running crawling job:',
       unknownError: 'Unknown error',
       totalPageScraped: 'Total pages scraped:',
       selectAll: 'Select All',
diff --git a/web/i18n/zh-Hans/dataset-creation.ts b/web/i18n/zh-Hans/dataset-creation.ts
index 78f5170791877..4f6786a1919b0 100644
--- a/web/i18n/zh-Hans/dataset-creation.ts
+++ b/web/i18n/zh-Hans/dataset-creation.ts
@@ -16,6 +16,11 @@ const translation = {
     apiKeyPlaceholder: '从 firecrawl.dev 获取 API Key',
     getApiKeyLinkText: '从 firecrawl.dev 获取您的 API Key',
   },
+  jinaReader: {
+    configJinaReader: '配置 Jina Reader',
+    apiKeyPlaceholder: '从 jina.ai 获取 API Key',
+    getApiKeyLinkText: '从 jina.ai 获取您的免费 API Key',
+  },
   stepOne: {
     filePreview: '文件预览',
     pagePreview: '页面预览',
@@ -56,13 +61,21 @@ const translation = {
       failed: '创建失败',
     },
     website: {
+      chooseProvider: '选择工具',
       fireCrawlNotConfigured: 'Firecrawl 未配置',
       fireCrawlNotConfiguredDescription: '请配置 Firecrawl 的 API 密钥以使用它。',
+      jinaReaderNotConfigured: 'Jina Reader 未配置',
+      jinaReaderNotConfiguredDescription: '请配置 Jina Reader 的免费 API 密钥以访问它。',
       configure: '配置',
       run: '运行',
       firecrawlTitle: '使用 🔥Firecrawl 提取网页内容',
       firecrawlDoc: 'Firecrawl 文档',
       firecrawlDocLink: 'https://docs.dify.ai/v/zh-hans/guides/knowledge-base/sync-from-website',
+      jinaReaderTitle: '将整个站点内容转换为 Markdown 格式',
+      jinaReaderDoc: '了解更多关于 Jina Reader',
+      jinaReaderDocLink: 'https://jina.ai/reader',
+      useSitemap: '使用 sitemap',
+      useSitemapTooltip: '根据 sitemap 爬取站点。否则，Jina Reader 将基于页面相关性迭代爬取，抓取较少的页面，但质量更高。',
       options: '选项',
       crawlSubPage: '爬取子页面',
       limit: '限制数量',
@@ -70,7 +83,7 @@ const translation = {
       excludePaths: '排除路径',
       includeOnlyPaths: '仅包含路径',
       extractOnlyMainContent: '仅提取主要内容（无标题、导航、页脚等）',
-      exceptionErrorTitle: '运行 Firecrawl 时发生异常:',
+      exceptionErrorTitle: '运行时发生异常:',
       unknownError: '未知错误',
       totalPageScraped: '抓取页面总数:',
       selectAll: '全选',
diff --git a/web/models/common.ts b/web/models/common.ts
index 78f09bee09649..204e89ed9b6b7 100644
--- a/web/models/common.ts
+++ b/web/models/common.ts
@@ -177,6 +177,7 @@ export enum DataSourceCategory {
 }
 export enum DataSourceProvider {
   fireCrawl = 'firecrawl',
+  jinaReader = 'jinareader',
 }
 
 export type FirecrawlConfig = {
diff --git a/web/models/datasets.ts b/web/models/datasets.ts
index 23d1fe6136b85..9358f6fcb9fb4 100644
--- a/web/models/datasets.ts
+++ b/web/models/datasets.ts
@@ -49,6 +49,7 @@ export type CrawlOptions = {
   excludes: string
   limit: number | string
   max_depth: number | string
+  use_sitemap: boolean
 }
 
 export type CrawlResultItem = {
diff --git a/web/service/datasets.ts b/web/service/datasets.ts
index 4ca269a7d6e21..689ac7c4035f1 100644
--- a/web/service/datasets.ts
+++ b/web/service/datasets.ts
@@ -23,7 +23,7 @@ import type {
   SegmentsResponse,
   createDocumentResponse,
 } from '@/models/datasets'
-import type { CommonResponse, DataSourceNotionWorkspace } from '@/models/common'
+import { type CommonResponse, type DataSourceNotionWorkspace, DataSourceProvider } from '@/models/common'
 import type {
   ApiKeysListResponse,
   CreateApiKeyResponse,
@@ -253,7 +253,7 @@ export const createFirecrawlTask: Fetcher<CommonResponse, Record<string, any>> =
   return post<CommonResponse>('website/crawl', {
     body: {
       ...body,
-      provider: 'firecrawl',
+      provider: DataSourceProvider.fireCrawl,
     },
   })
 }
@@ -261,7 +261,26 @@ export const createFirecrawlTask: Fetcher<CommonResponse, Record<string, any>> =
 export const checkFirecrawlTaskStatus: Fetcher<CommonResponse, string> = (jobId: string) => {
   return get<CommonResponse>(`website/crawl/status/${jobId}`, {
     params: {
-      provider: 'firecrawl',
+      provider: DataSourceProvider.fireCrawl,
+    },
+  }, {
+    silent: true,
+  })
+}
+
+export const createJinaReaderTask: Fetcher<CommonResponse, Record<string, any>> = (body) => {
+  return post<CommonResponse>('website/crawl', {
+    body: {
+      ...body,
+      provider: DataSourceProvider.jinaReader,
+    },
+  })
+}
+
+export const checkJinaReaderTaskStatus: Fetcher<CommonResponse, string> = (jobId: string) => {
+  return get<CommonResponse>(`website/crawl/status/${jobId}`, {
+    params: {
+      provider: 'jinareader',
     },
   }, {
     silent: true,

From 3af65b2f452ae765cde4916507350b8c59ea2aea Mon Sep 17 00:00:00 2001
From: -LAN- <laipz8200@outlook.com>
Date: Mon, 30 Sep 2024 11:12:26 +0800
Subject: [PATCH 109/235] feat(api): add version comparison logic (#8902)

---
 api/controllers/console/version.py            | 49 +++++++++++++++++--
 .../controllers/test_compare_versions.py      | 38 ++++++++++++++
 2 files changed, 83 insertions(+), 4 deletions(-)
 create mode 100644 api/tests/unit_tests/controllers/test_compare_versions.py

diff --git a/api/controllers/console/version.py b/api/controllers/console/version.py
index 76adbfe6a9d6c..deda1a0d02730 100644
--- a/api/controllers/console/version.py
+++ b/api/controllers/console/version.py
@@ -38,11 +38,52 @@ def get(self):
             return result
 
         content = json.loads(response.content)
-        result["version"] = content["version"]
-        result["release_date"] = content["releaseDate"]
-        result["release_notes"] = content["releaseNotes"]
-        result["can_auto_update"] = content["canAutoUpdate"]
+        if _has_new_version(latest_version=content["version"], current_version=f"{args.get('current_version')}"):
+            result["version"] = content["version"]
+            result["release_date"] = content["releaseDate"]
+            result["release_notes"] = content["releaseNotes"]
+            result["can_auto_update"] = content["canAutoUpdate"]
         return result
 
 
+def _has_new_version(*, latest_version: str, current_version: str) -> bool:
+    def parse_version(version: str) -> tuple:
+        # Split version into parts and pre-release suffix if any
+        parts = version.split("-")
+        version_parts = parts[0].split(".")
+        pre_release = parts[1] if len(parts) > 1 else None
+
+        # Validate version format
+        if len(version_parts) != 3:
+            raise ValueError(f"Invalid version format: {version}")
+
+        try:
+            # Convert version parts to integers
+            major, minor, patch = map(int, version_parts)
+            return (major, minor, patch, pre_release)
+        except ValueError:
+            raise ValueError(f"Invalid version format: {version}")
+
+    latest = parse_version(latest_version)
+    current = parse_version(current_version)
+
+    # Compare major, minor, and patch versions
+    for latest_part, current_part in zip(latest[:3], current[:3]):
+        if latest_part > current_part:
+            return True
+        elif latest_part < current_part:
+            return False
+
+    # If versions are equal, check pre-release suffixes
+    if latest[3] is None and current[3] is not None:
+        return True
+    elif latest[3] is not None and current[3] is None:
+        return False
+    elif latest[3] is not None and current[3] is not None:
+        # Simple string comparison for pre-release versions
+        return latest[3] > current[3]
+
+    return False
+
+
 api.add_resource(VersionApi, "/version")
diff --git a/api/tests/unit_tests/controllers/test_compare_versions.py b/api/tests/unit_tests/controllers/test_compare_versions.py
new file mode 100644
index 0000000000000..87902b6d44e66
--- /dev/null
+++ b/api/tests/unit_tests/controllers/test_compare_versions.py
@@ -0,0 +1,38 @@
+import pytest
+
+from controllers.console.version import _has_new_version
+
+
+@pytest.mark.parametrize(
+    ("latest_version", "current_version", "expected"),
+    [
+        ("1.0.1", "1.0.0", True),
+        ("1.1.0", "1.0.0", True),
+        ("2.0.0", "1.9.9", True),
+        ("1.0.0", "1.0.0", False),
+        ("1.0.0", "1.0.1", False),
+        ("1.0.0", "2.0.0", False),
+        ("1.0.1", "1.0.0-beta", True),
+        ("1.0.0", "1.0.0-alpha", True),
+        ("1.0.0-beta", "1.0.0-alpha", True),
+        ("1.0.0", "1.0.0-rc1", True),
+        ("1.0.0", "0.9.9", True),
+        ("1.0.0", "1.0.0-dev", True),
+    ],
+)
+def test_has_new_version(latest_version, current_version, expected):
+    assert _has_new_version(latest_version=latest_version, current_version=current_version) == expected
+
+
+def test_has_new_version_invalid_input():
+    with pytest.raises(ValueError):
+        _has_new_version(latest_version="1.0", current_version="1.0.0")
+
+    with pytest.raises(ValueError):
+        _has_new_version(latest_version="1.0.0", current_version="1.0")
+
+    with pytest.raises(ValueError):
+        _has_new_version(latest_version="invalid", current_version="1.0.0")
+
+    with pytest.raises(ValueError):
+        _has_new_version(latest_version="1.0.0", current_version="invalid")

From ada9d408ac66bc8775d0a2c72568c43666946b1a Mon Sep 17 00:00:00 2001
From: -LAN- <laipz8200@outlook.com>
Date: Mon, 30 Sep 2024 12:48:58 +0800
Subject: [PATCH 110/235] refactor(api/variables): VariableError as a
 ValueError. (#8554)

---
 api/core/app/segments/exc.py | 2 +-
 1 file changed, 1 insertion(+), 1 deletion(-)

diff --git a/api/core/app/segments/exc.py b/api/core/app/segments/exc.py
index d15d6d500ffa4..5cf67c3baccac 100644
--- a/api/core/app/segments/exc.py
+++ b/api/core/app/segments/exc.py
@@ -1,2 +1,2 @@
-class VariableError(Exception):
+class VariableError(ValueError):
     pass

From 503561f464c50a9a8469614603407f8accfd025f Mon Sep 17 00:00:00 2001
From: zhuhao <37029601+hwzhuhao@users.noreply.github.com>
Date: Mon, 30 Sep 2024 12:52:18 +0800
Subject: [PATCH 111/235] fix: fix the data validation consistency issue in
 keyword content review (#8908)

---
 api/core/moderation/keywords/keywords.py | 8 ++++++--
 1 file changed, 6 insertions(+), 2 deletions(-)

diff --git a/api/core/moderation/keywords/keywords.py b/api/core/moderation/keywords/keywords.py
index dc6a7ec5644a7..4846da8f93076 100644
--- a/api/core/moderation/keywords/keywords.py
+++ b/api/core/moderation/keywords/keywords.py
@@ -18,8 +18,12 @@ def validate_config(cls, tenant_id: str, config: dict) -> None:
         if not config.get("keywords"):
             raise ValueError("keywords is required")
 
-        if len(config.get("keywords")) > 1000:
-            raise ValueError("keywords length must be less than 1000")
+        if len(config.get("keywords")) > 10000:
+            raise ValueError("keywords length must be less than 10000")
+
+        keywords_row_len = config["keywords"].split("\n")
+        if len(keywords_row_len) > 100:
+            raise ValueError("the number of rows for the keywords must be less than 100")
 
     def moderation_for_inputs(self, inputs: dict, query: str = "") -> ModerationInputsResult:
         flagged = False

From 77aef9ff1db220dfbab5da6e48061051b98b7cbd Mon Sep 17 00:00:00 2001
From: zhuhao <37029601+hwzhuhao@users.noreply.github.com>
Date: Mon, 30 Sep 2024 12:55:01 +0800
Subject: [PATCH 112/235] refactor: optimize the calculation of rerank
 threshold and the logic for forbidden characters in model_uid (#8879)

---
 .../model_providers/xinference/llm/llm.py           |  3 ++-
 .../model_providers/xinference/rerank/rerank.py     |  8 +++-----
 .../xinference/speech2text/speech2text.py           |  3 ++-
 .../xinference/text_embedding/text_embedding.py     |  4 ++--
 .../model_providers/xinference/tts/tts.py           |  4 ++--
 .../model_providers/xinference/xinference_helper.py | 13 +++++++++++++
 6 files changed, 24 insertions(+), 11 deletions(-)

diff --git a/api/core/model_runtime/model_providers/xinference/llm/llm.py b/api/core/model_runtime/model_providers/xinference/llm/llm.py
index 286640079b02a..0c9d08679ad04 100644
--- a/api/core/model_runtime/model_providers/xinference/llm/llm.py
+++ b/api/core/model_runtime/model_providers/xinference/llm/llm.py
@@ -59,6 +59,7 @@
 from core.model_runtime.model_providers.xinference.xinference_helper import (
     XinferenceHelper,
     XinferenceModelExtraParameter,
+    validate_model_uid,
 )
 from core.model_runtime.utils import helper
 
@@ -114,7 +115,7 @@ def validate_credentials(self, model: str, credentials: dict) -> None:
         }
         """
         try:
-            if "/" in credentials["model_uid"] or "?" in credentials["model_uid"] or "#" in credentials["model_uid"]:
+            if not validate_model_uid(credentials):
                 raise CredentialsValidateFailedError("model_uid should not contain /, ?, or #")
 
             extra_param = XinferenceHelper.get_xinference_extra_parameter(
diff --git a/api/core/model_runtime/model_providers/xinference/rerank/rerank.py b/api/core/model_runtime/model_providers/xinference/rerank/rerank.py
index 8f18bc42d2339..6368cd76dc97b 100644
--- a/api/core/model_runtime/model_providers/xinference/rerank/rerank.py
+++ b/api/core/model_runtime/model_providers/xinference/rerank/rerank.py
@@ -15,6 +15,7 @@
 )
 from core.model_runtime.errors.validate import CredentialsValidateFailedError
 from core.model_runtime.model_providers.__base.rerank_model import RerankModel
+from core.model_runtime.model_providers.xinference.xinference_helper import validate_model_uid
 
 
 class XinferenceRerankModel(RerankModel):
@@ -77,10 +78,7 @@ def _invoke(
             )
 
             # score threshold check
-            if score_threshold is not None:
-                if result["relevance_score"] >= score_threshold:
-                    rerank_documents.append(rerank_document)
-            else:
+            if score_threshold is None or result["relevance_score"] >= score_threshold:
                 rerank_documents.append(rerank_document)
 
         return RerankResult(model=model, docs=rerank_documents)
@@ -94,7 +92,7 @@ def validate_credentials(self, model: str, credentials: dict) -> None:
         :return:
         """
         try:
-            if "/" in credentials["model_uid"] or "?" in credentials["model_uid"] or "#" in credentials["model_uid"]:
+            if not validate_model_uid(credentials):
                 raise CredentialsValidateFailedError("model_uid should not contain /, ?, or #")
 
             credentials["server_url"] = credentials["server_url"].removesuffix("/")
diff --git a/api/core/model_runtime/model_providers/xinference/speech2text/speech2text.py b/api/core/model_runtime/model_providers/xinference/speech2text/speech2text.py
index a6c5b8a0a571e..c5ad38391185a 100644
--- a/api/core/model_runtime/model_providers/xinference/speech2text/speech2text.py
+++ b/api/core/model_runtime/model_providers/xinference/speech2text/speech2text.py
@@ -14,6 +14,7 @@
 )
 from core.model_runtime.errors.validate import CredentialsValidateFailedError
 from core.model_runtime.model_providers.__base.speech2text_model import Speech2TextModel
+from core.model_runtime.model_providers.xinference.xinference_helper import validate_model_uid
 
 
 class XinferenceSpeech2TextModel(Speech2TextModel):
@@ -42,7 +43,7 @@ def validate_credentials(self, model: str, credentials: dict) -> None:
         :return:
         """
         try:
-            if "/" in credentials["model_uid"] or "?" in credentials["model_uid"] or "#" in credentials["model_uid"]:
+            if not validate_model_uid(credentials):
                 raise CredentialsValidateFailedError("model_uid should not contain /, ?, or #")
 
             credentials["server_url"] = credentials["server_url"].removesuffix("/")
diff --git a/api/core/model_runtime/model_providers/xinference/text_embedding/text_embedding.py b/api/core/model_runtime/model_providers/xinference/text_embedding/text_embedding.py
index 16272391320d5..ddc21b365c1b7 100644
--- a/api/core/model_runtime/model_providers/xinference/text_embedding/text_embedding.py
+++ b/api/core/model_runtime/model_providers/xinference/text_embedding/text_embedding.py
@@ -17,7 +17,7 @@
 )
 from core.model_runtime.errors.validate import CredentialsValidateFailedError
 from core.model_runtime.model_providers.__base.text_embedding_model import TextEmbeddingModel
-from core.model_runtime.model_providers.xinference.xinference_helper import XinferenceHelper
+from core.model_runtime.model_providers.xinference.xinference_helper import XinferenceHelper, validate_model_uid
 
 
 class XinferenceTextEmbeddingModel(TextEmbeddingModel):
@@ -110,7 +110,7 @@ def validate_credentials(self, model: str, credentials: dict) -> None:
         :return:
         """
         try:
-            if "/" in credentials["model_uid"] or "?" in credentials["model_uid"] or "#" in credentials["model_uid"]:
+            if not validate_model_uid(credentials):
                 raise CredentialsValidateFailedError("model_uid should not contain /, ?, or #")
 
             server_url = credentials["server_url"]
diff --git a/api/core/model_runtime/model_providers/xinference/tts/tts.py b/api/core/model_runtime/model_providers/xinference/tts/tts.py
index 81dbe397d2f10..6290e8551d802 100644
--- a/api/core/model_runtime/model_providers/xinference/tts/tts.py
+++ b/api/core/model_runtime/model_providers/xinference/tts/tts.py
@@ -15,7 +15,7 @@
 )
 from core.model_runtime.errors.validate import CredentialsValidateFailedError
 from core.model_runtime.model_providers.__base.tts_model import TTSModel
-from core.model_runtime.model_providers.xinference.xinference_helper import XinferenceHelper
+from core.model_runtime.model_providers.xinference.xinference_helper import XinferenceHelper, validate_model_uid
 
 
 class XinferenceText2SpeechModel(TTSModel):
@@ -70,7 +70,7 @@ def validate_credentials(self, model: str, credentials: dict) -> None:
         :return:
         """
         try:
-            if "/" in credentials["model_uid"] or "?" in credentials["model_uid"] or "#" in credentials["model_uid"]:
+            if not validate_model_uid(credentials):
                 raise CredentialsValidateFailedError("model_uid should not contain /, ?, or #")
 
             credentials["server_url"] = credentials["server_url"].removesuffix("/")
diff --git a/api/core/model_runtime/model_providers/xinference/xinference_helper.py b/api/core/model_runtime/model_providers/xinference/xinference_helper.py
index 619ee1492a927..baa3ccbe8adbc 100644
--- a/api/core/model_runtime/model_providers/xinference/xinference_helper.py
+++ b/api/core/model_runtime/model_providers/xinference/xinference_helper.py
@@ -132,3 +132,16 @@ def _get_xinference_extra_parameter(server_url: str, model_uid: str, api_key: st
             context_length=context_length,
             model_family=model_family,
         )
+
+
+def validate_model_uid(credentials: dict) -> bool:
+    """
+    Validate the model_uid within the credentials dictionary to ensure it does not
+    contain forbidden characters ("/", "?", "#").
+
+    param credentials: model credentials
+    :return: True if the model_uid does not contain forbidden characters ("/", "?", "#"), else False.
+    """
+    forbidden_characters = ["/", "?", "#"]
+    model_uid = credentials.get("model_uid", "")
+    return not any(char in forbidden_characters for char in model_uid)

From 9d221a5e19f8613fa1a47984649378b22a0fd66e Mon Sep 17 00:00:00 2001
From: Jyong <76649700+JohnJyong@users.noreply.github.com>
Date: Mon, 30 Sep 2024 15:38:43 +0800
Subject: [PATCH 113/235] external knowledge api (#8913)

Co-authored-by: Yi <yxiaoisme@gmail.com>
---
 api/controllers/console/__init__.py           |   11 +-
 api/controllers/console/datasets/datasets.py  |   54 +-
 api/controllers/console/datasets/external.py  |  239 +++
 .../console/datasets/hit_testing.py           |    2 +
 .../service_api/dataset/dataset.py            |   27 +-
 .../index_tool_callback_handler.py            |    2 +-
 api/core/rag/datasource/retrieval_service.py  |   14 +
 api/core/rag/entities/context_entities.py     |   10 +
 api/core/rag/models/document.py               |    2 +
 api/core/rag/rerank/rerank_model.py           |   17 +-
 api/core/rag/retrieval/dataset_retrieval.py   |  300 ++--
 .../knowledge_retrieval_node.py               |   44 +-
 api/fields/dataset_fields.py                  |   13 +
 api/fields/external_dataset_fields.py         |   11 +
 ...-6af6a521a53e_update_retrieval_resource.py |   48 +
 ...434-33f5fac87f29_external_knowledge_api.py |   73 +
 ...fca025d3b60f_add_dataset_retrival_model.py |    2 +-
 api/models/dataset.py                         |  106 ++
 api/models/model.py                           |    6 +-
 api/poetry.lock                               | 1542 +++++++++--------
 api/pyproject.toml                            |    1 +
 api/schedule/clean_unused_messages_task.py    |   92 +
 api/services/dataset_service.py               |  191 +-
 .../external_knowledge_entities.py            |   26 +
 api/services/external_knowledge_service.py    |  274 +++
 api/services/hit_testing_service.py           |   68 +-
 api/tasks/external_document_indexing_task.py  |   93 +
 .../[datasetId]/layout.tsx                    |   26 +-
 web/app/(commonLayout)/datasets/Container.tsx |   20 +-
 .../(commonLayout)/datasets/DatasetCard.tsx   |   29 +-
 .../datasets/NewDatasetCard.tsx               |   27 +-
 .../(commonLayout)/datasets/connect/page.tsx  |    8 +
 web/app/(commonLayout)/datasets/layout.tsx    |   14 +
 web/app/(commonLayout)/datasets/page.tsx      |    4 +-
 web/app/(commonLayout)/datasets/store.ts      |   11 +
 web/app/components/app-sidebar/basic.tsx      |    7 +-
 web/app/components/app-sidebar/index.tsx      |    4 +-
 .../dataset-config/card-item/item.tsx         |   10 +-
 .../params-config/config-content.tsx          |   20 +-
 .../dataset-config/params-config/index.tsx    |   15 +-
 .../dataset-config/select-dataset/index.tsx   |    7 +-
 .../dataset-config/settings-modal/index.tsx   |  148 +-
 .../components/base/corner-label/index.tsx    |   21 +
 .../solid/development/api-connection-mod.svg  |    5 +
 .../assets/vender/solid/shapes/corner.svg     |    3 +
 .../solid/development/ApiConnectionMod.json   |   38 +
 .../solid/development/ApiConnectionMod.tsx    |   16 +
 .../src/vender/solid/development/index.ts     |    1 +
 .../icons/src/vender/solid/shapes/Corner.json |   27 +
 .../icons/src/vender/solid/shapes/Corner.tsx  |   16 +
 .../icons/src/vender/solid/shapes/index.ts    |    1 +
 web/app/components/base/modal/index.css       |    2 +-
 web/app/components/base/param-item/index.tsx  |    1 +
 web/app/components/base/select/index.tsx      |    2 +-
 .../detail/completed/SegmentCard.tsx          |   20 +-
 .../datasets/external-api/declarations.ts     |   16 +
 .../external-api/external-api-modal/Form.tsx  |   90 +
 .../external-api/external-api-modal/index.tsx |  218 +++
 .../external-api/external-api-panel/index.tsx |   90 +
 .../external-knowledge-api-card/index.tsx     |  151 ++
 .../connector/index.tsx                       |   36 +
 .../create/ExternalApiSelect.tsx              |  110 ++
 .../create/ExternalApiSelection.tsx           |   96 +
 .../create/InfoPanel.tsx                      |   33 +
 .../create/KnowledgeBaseInfo.tsx              |   65 +
 .../create/RetrievalSettings.tsx              |   67 +
 .../create/declarations.ts                    |   12 +
 .../external-knowledge-base/create/index.tsx  |  128 ++
 .../datasets/hit-testing/hit-detail.tsx       |   10 +-
 .../components/datasets/hit-testing/index.tsx |  111 +-
 .../modify-external-retrieval-modal.tsx       |   71 +
 .../datasets/hit-testing/textarea.tsx         |   91 +-
 .../datasets/rename-modal/index.tsx           |   15 +-
 .../datasets/settings/form/index.tsx          |  122 +-
 .../components/header/dataset-nav/index.tsx   |    2 +-
 .../components/dataset-item.tsx               |   58 +-
 .../nodes/knowledge-retrieval/use-config.ts   |   10 +-
 .../nodes/knowledge-retrieval/utils.ts        |   31 +-
 web/context/external-api-panel-context.tsx    |   28 +
 .../external-knowledge-api-context.tsx        |   46 +
 web/context/modal-context.tsx                 |   39 +
 web/i18n/en-US/dataset-hit-testing.ts         |    4 +-
 web/i18n/en-US/dataset-settings.ts            |   11 +-
 web/i18n/en-US/dataset.ts                     |   74 +
 web/i18n/zh-Hans/dataset-hit-testing.ts       |    4 +-
 web/i18n/zh-Hans/dataset-settings.ts          |    6 +-
 web/i18n/zh-Hans/dataset.ts                   |   74 +
 web/i18n/zh-Hant/dataset-hit-testing.ts       |    1 +
 web/models/datasets.ts                        |   81 +
 web/service/datasets.ts                       |   42 +-
 90 files changed, 4633 insertions(+), 1181 deletions(-)
 create mode 100644 api/controllers/console/datasets/external.py
 create mode 100644 api/core/rag/entities/context_entities.py
 create mode 100644 api/fields/external_dataset_fields.py
 create mode 100644 api/migrations/versions/2024_09_24_0922-6af6a521a53e_update_retrieval_resource.py
 create mode 100644 api/migrations/versions/2024_09_25_0434-33f5fac87f29_external_knowledge_api.py
 create mode 100644 api/schedule/clean_unused_messages_task.py
 create mode 100644 api/services/entities/external_knowledge_entities/external_knowledge_entities.py
 create mode 100644 api/services/external_knowledge_service.py
 create mode 100644 api/tasks/external_document_indexing_task.py
 create mode 100644 web/app/(commonLayout)/datasets/connect/page.tsx
 create mode 100644 web/app/(commonLayout)/datasets/layout.tsx
 create mode 100644 web/app/(commonLayout)/datasets/store.ts
 create mode 100644 web/app/components/base/corner-label/index.tsx
 create mode 100644 web/app/components/base/icons/assets/vender/solid/development/api-connection-mod.svg
 create mode 100644 web/app/components/base/icons/assets/vender/solid/shapes/corner.svg
 create mode 100644 web/app/components/base/icons/src/vender/solid/development/ApiConnectionMod.json
 create mode 100644 web/app/components/base/icons/src/vender/solid/development/ApiConnectionMod.tsx
 create mode 100644 web/app/components/base/icons/src/vender/solid/shapes/Corner.json
 create mode 100644 web/app/components/base/icons/src/vender/solid/shapes/Corner.tsx
 create mode 100644 web/app/components/datasets/external-api/declarations.ts
 create mode 100644 web/app/components/datasets/external-api/external-api-modal/Form.tsx
 create mode 100644 web/app/components/datasets/external-api/external-api-modal/index.tsx
 create mode 100644 web/app/components/datasets/external-api/external-api-panel/index.tsx
 create mode 100644 web/app/components/datasets/external-api/external-knowledge-api-card/index.tsx
 create mode 100644 web/app/components/datasets/external-knowledge-base/connector/index.tsx
 create mode 100644 web/app/components/datasets/external-knowledge-base/create/ExternalApiSelect.tsx
 create mode 100644 web/app/components/datasets/external-knowledge-base/create/ExternalApiSelection.tsx
 create mode 100644 web/app/components/datasets/external-knowledge-base/create/InfoPanel.tsx
 create mode 100644 web/app/components/datasets/external-knowledge-base/create/KnowledgeBaseInfo.tsx
 create mode 100644 web/app/components/datasets/external-knowledge-base/create/RetrievalSettings.tsx
 create mode 100644 web/app/components/datasets/external-knowledge-base/create/declarations.ts
 create mode 100644 web/app/components/datasets/external-knowledge-base/create/index.tsx
 create mode 100644 web/app/components/datasets/hit-testing/modify-external-retrieval-modal.tsx
 create mode 100644 web/context/external-api-panel-context.tsx
 create mode 100644 web/context/external-knowledge-api-context.tsx

diff --git a/api/controllers/console/__init__.py b/api/controllers/console/__init__.py
index eb7c1464d3972..c7282fcf14531 100644
--- a/api/controllers/console/__init__.py
+++ b/api/controllers/console/__init__.py
@@ -37,7 +37,16 @@
 from .billing import billing
 
 # Import datasets controllers
-from .datasets import data_source, datasets, datasets_document, datasets_segments, file, hit_testing, website
+from .datasets import (
+    data_source,
+    datasets,
+    datasets_document,
+    datasets_segments,
+    external,
+    file,
+    hit_testing,
+    website,
+)
 
 # Import explore controllers
 from .explore import (
diff --git a/api/controllers/console/datasets/datasets.py b/api/controllers/console/datasets/datasets.py
index 5a763b3457c69..9561fd8b70e4b 100644
--- a/api/controllers/console/datasets/datasets.py
+++ b/api/controllers/console/datasets/datasets.py
@@ -49,7 +49,7 @@ def get(self):
         page = request.args.get("page", default=1, type=int)
         limit = request.args.get("limit", default=20, type=int)
         ids = request.args.getlist("ids")
-        provider = request.args.get("provider", default="vendor")
+        # provider = request.args.get("provider", default="vendor")
         search = request.args.get("keyword", default=None, type=str)
         tag_ids = request.args.getlist("tag_ids")
 
@@ -57,7 +57,7 @@ def get(self):
             datasets, total = DatasetService.get_datasets_by_ids(ids, current_user.current_tenant_id)
         else:
             datasets, total = DatasetService.get_datasets(
-                page, limit, provider, current_user.current_tenant_id, current_user, search, tag_ids
+                page, limit, current_user.current_tenant_id, current_user, search, tag_ids
             )
 
         # check embedding setting
@@ -110,6 +110,26 @@ def post(self):
             nullable=True,
             help="Invalid indexing technique.",
         )
+        parser.add_argument(
+            "external_knowledge_api_id",
+            type=str,
+            nullable=True,
+            required=False,
+        )
+        parser.add_argument(
+            "provider",
+            type=str,
+            nullable=True,
+            choices=Dataset.PROVIDER_LIST,
+            required=False,
+            default="vendor",
+        )
+        parser.add_argument(
+            "external_knowledge_id",
+            type=str,
+            nullable=True,
+            required=False,
+        )
         args = parser.parse_args()
 
         # The role of the current user in the ta table must be admin, owner, or editor, or dataset_operator
@@ -123,6 +143,9 @@ def post(self):
                 indexing_technique=args["indexing_technique"],
                 account=current_user,
                 permission=DatasetPermissionEnum.ONLY_ME,
+                provider=args["provider"],
+                external_knowledge_api_id=args["external_knowledge_api_id"],
+                external_knowledge_id=args["external_knowledge_id"],
             )
         except services.errors.dataset.DatasetNameDuplicateError:
             raise DatasetNameDuplicateError()
@@ -211,6 +234,33 @@ def patch(self, dataset_id):
         )
         parser.add_argument("retrieval_model", type=dict, location="json", help="Invalid retrieval model.")
         parser.add_argument("partial_member_list", type=list, location="json", help="Invalid parent user list.")
+
+        parser.add_argument(
+            "external_retrieval_model",
+            type=dict,
+            required=False,
+            nullable=True,
+            location="json",
+            help="Invalid external retrieval model.",
+        )
+
+        parser.add_argument(
+            "external_knowledge_id",
+            type=str,
+            required=False,
+            nullable=True,
+            location="json",
+            help="Invalid external knowledge id.",
+        )
+
+        parser.add_argument(
+            "external_knowledge_api_id",
+            type=str,
+            required=False,
+            nullable=True,
+            location="json",
+            help="Invalid external knowledge api id.",
+        )
         args = parser.parse_args()
         data = request.get_json()
 
diff --git a/api/controllers/console/datasets/external.py b/api/controllers/console/datasets/external.py
new file mode 100644
index 0000000000000..dcef979360ba6
--- /dev/null
+++ b/api/controllers/console/datasets/external.py
@@ -0,0 +1,239 @@
+from flask import request
+from flask_login import current_user
+from flask_restful import Resource, marshal, reqparse
+from werkzeug.exceptions import Forbidden, InternalServerError, NotFound
+
+import services
+from controllers.console import api
+from controllers.console.datasets.error import DatasetNameDuplicateError
+from controllers.console.setup import setup_required
+from controllers.console.wraps import account_initialization_required
+from fields.dataset_fields import dataset_detail_fields
+from libs.login import login_required
+from services.dataset_service import DatasetService
+from services.external_knowledge_service import ExternalDatasetService
+from services.hit_testing_service import HitTestingService
+
+
+def _validate_name(name):
+    if not name or len(name) < 1 or len(name) > 100:
+        raise ValueError("Name must be between 1 to 100 characters.")
+    return name
+
+
+def _validate_description_length(description):
+    if description and len(description) > 400:
+        raise ValueError("Description cannot exceed 400 characters.")
+    return description
+
+
+class ExternalApiTemplateListApi(Resource):
+    @setup_required
+    @login_required
+    @account_initialization_required
+    def get(self):
+        page = request.args.get("page", default=1, type=int)
+        limit = request.args.get("limit", default=20, type=int)
+        search = request.args.get("keyword", default=None, type=str)
+
+        external_knowledge_apis, total = ExternalDatasetService.get_external_knowledge_apis(
+            page, limit, current_user.current_tenant_id, search
+        )
+        response = {
+            "data": [item.to_dict() for item in external_knowledge_apis],
+            "has_more": len(external_knowledge_apis) == limit,
+            "limit": limit,
+            "total": total,
+            "page": page,
+        }
+        return response, 200
+
+    @setup_required
+    @login_required
+    @account_initialization_required
+    def post(self):
+        parser = reqparse.RequestParser()
+        parser.add_argument(
+            "name",
+            nullable=False,
+            required=True,
+            help="Name is required. Name must be between 1 to 100 characters.",
+            type=_validate_name,
+        )
+        parser.add_argument(
+            "settings",
+            type=dict,
+            location="json",
+            nullable=False,
+            required=True,
+        )
+        args = parser.parse_args()
+
+        ExternalDatasetService.validate_api_list(args["settings"])
+
+        # The role of the current user in the ta table must be admin, owner, or editor, or dataset_operator
+        if not current_user.is_dataset_editor:
+            raise Forbidden()
+
+        try:
+            external_knowledge_api = ExternalDatasetService.create_external_knowledge_api(
+                tenant_id=current_user.current_tenant_id, user_id=current_user.id, args=args
+            )
+        except services.errors.dataset.DatasetNameDuplicateError:
+            raise DatasetNameDuplicateError()
+
+        return external_knowledge_api.to_dict(), 201
+
+
+class ExternalApiTemplateApi(Resource):
+    @setup_required
+    @login_required
+    @account_initialization_required
+    def get(self, external_knowledge_api_id):
+        external_knowledge_api_id = str(external_knowledge_api_id)
+        external_knowledge_api = ExternalDatasetService.get_external_knowledge_api(external_knowledge_api_id)
+        if external_knowledge_api is None:
+            raise NotFound("API template not found.")
+
+        return external_knowledge_api.to_dict(), 200
+
+    @setup_required
+    @login_required
+    @account_initialization_required
+    def patch(self, external_knowledge_api_id):
+        external_knowledge_api_id = str(external_knowledge_api_id)
+
+        parser = reqparse.RequestParser()
+        parser.add_argument(
+            "name",
+            nullable=False,
+            required=True,
+            help="type is required. Name must be between 1 to 100 characters.",
+            type=_validate_name,
+        )
+        parser.add_argument(
+            "settings",
+            type=dict,
+            location="json",
+            nullable=False,
+            required=True,
+        )
+        args = parser.parse_args()
+        ExternalDatasetService.validate_api_list(args["settings"])
+
+        external_knowledge_api = ExternalDatasetService.update_external_knowledge_api(
+            tenant_id=current_user.current_tenant_id,
+            user_id=current_user.id,
+            external_knowledge_api_id=external_knowledge_api_id,
+            args=args,
+        )
+
+        return external_knowledge_api.to_dict(), 200
+
+    @setup_required
+    @login_required
+    @account_initialization_required
+    def delete(self, external_knowledge_api_id):
+        external_knowledge_api_id = str(external_knowledge_api_id)
+
+        # The role of the current user in the ta table must be admin, owner, or editor
+        if not current_user.is_editor or current_user.is_dataset_operator:
+            raise Forbidden()
+
+        ExternalDatasetService.delete_external_knowledge_api(current_user.current_tenant_id, external_knowledge_api_id)
+        return {"result": "success"}, 200
+
+
+class ExternalApiUseCheckApi(Resource):
+    @setup_required
+    @login_required
+    @account_initialization_required
+    def get(self, external_knowledge_api_id):
+        external_knowledge_api_id = str(external_knowledge_api_id)
+
+        external_knowledge_api_is_using, count = ExternalDatasetService.external_knowledge_api_use_check(
+            external_knowledge_api_id
+        )
+        return {"is_using": external_knowledge_api_is_using, "count": count}, 200
+
+
+class ExternalDatasetCreateApi(Resource):
+    @setup_required
+    @login_required
+    @account_initialization_required
+    def post(self):
+        # The role of the current user in the ta table must be admin, owner, or editor
+        if not current_user.is_editor:
+            raise Forbidden()
+
+        parser = reqparse.RequestParser()
+        parser.add_argument("external_knowledge_api_id", type=str, required=True, nullable=False, location="json")
+        parser.add_argument("external_knowledge_id", type=str, required=True, nullable=False, location="json")
+        parser.add_argument(
+            "name",
+            nullable=False,
+            required=True,
+            help="name is required. Name must be between 1 to 100 characters.",
+            type=_validate_name,
+        )
+        parser.add_argument("description", type=str, required=False, nullable=True, location="json")
+        parser.add_argument("external_retrieval_model", type=dict, required=False, location="json")
+
+        args = parser.parse_args()
+
+        # The role of the current user in the ta table must be admin, owner, or editor, or dataset_operator
+        if not current_user.is_dataset_editor:
+            raise Forbidden()
+
+        try:
+            dataset = ExternalDatasetService.create_external_dataset(
+                tenant_id=current_user.current_tenant_id,
+                user_id=current_user.id,
+                args=args,
+            )
+        except services.errors.dataset.DatasetNameDuplicateError:
+            raise DatasetNameDuplicateError()
+
+        return marshal(dataset, dataset_detail_fields), 201
+
+
+class ExternalKnowledgeHitTestingApi(Resource):
+    @setup_required
+    @login_required
+    @account_initialization_required
+    def post(self, dataset_id):
+        dataset_id_str = str(dataset_id)
+        dataset = DatasetService.get_dataset(dataset_id_str)
+        if dataset is None:
+            raise NotFound("Dataset not found.")
+
+        try:
+            DatasetService.check_dataset_permission(dataset, current_user)
+        except services.errors.account.NoPermissionError as e:
+            raise Forbidden(str(e))
+
+        parser = reqparse.RequestParser()
+        parser.add_argument("query", type=str, location="json")
+        parser.add_argument("external_retrieval_model", type=dict, required=False, location="json")
+        args = parser.parse_args()
+
+        HitTestingService.hit_testing_args_check(args)
+
+        try:
+            response = HitTestingService.external_retrieve(
+                dataset=dataset,
+                query=args["query"],
+                account=current_user,
+                external_retrieval_model=args["external_retrieval_model"],
+            )
+
+            return response
+        except Exception as e:
+            raise InternalServerError(str(e))
+
+
+api.add_resource(ExternalKnowledgeHitTestingApi, "/datasets/<uuid:dataset_id>/external-hit-testing")
+api.add_resource(ExternalDatasetCreateApi, "/datasets/external")
+api.add_resource(ExternalApiTemplateListApi, "/datasets/external-knowledge-api")
+api.add_resource(ExternalApiTemplateApi, "/datasets/external-knowledge-api/<uuid:external_knowledge_api_id>")
+api.add_resource(ExternalApiUseCheckApi, "/datasets/external-knowledge-api/<uuid:external_knowledge_api_id>/use-check")
diff --git a/api/controllers/console/datasets/hit_testing.py b/api/controllers/console/datasets/hit_testing.py
index 0b4a7be986e16..6e6d8c0bd76f2 100644
--- a/api/controllers/console/datasets/hit_testing.py
+++ b/api/controllers/console/datasets/hit_testing.py
@@ -47,6 +47,7 @@ def post(self, dataset_id):
         parser = reqparse.RequestParser()
         parser.add_argument("query", type=str, location="json")
         parser.add_argument("retrieval_model", type=dict, required=False, location="json")
+        parser.add_argument("external_retrieval_model", type=dict, required=False, location="json")
         args = parser.parse_args()
 
         HitTestingService.hit_testing_args_check(args)
@@ -57,6 +58,7 @@ def post(self, dataset_id):
                 query=args["query"],
                 account=current_user,
                 retrieval_model=args["retrieval_model"],
+                external_retrieval_model=args["external_retrieval_model"],
                 limit=10,
             )
 
diff --git a/api/controllers/service_api/dataset/dataset.py b/api/controllers/service_api/dataset/dataset.py
index c2c0672a0346d..f076cff6c8770 100644
--- a/api/controllers/service_api/dataset/dataset.py
+++ b/api/controllers/service_api/dataset/dataset.py
@@ -28,11 +28,11 @@ def get(self, tenant_id):
 
         page = request.args.get("page", default=1, type=int)
         limit = request.args.get("limit", default=20, type=int)
-        provider = request.args.get("provider", default="vendor")
+        # provider = request.args.get("provider", default="vendor")
         search = request.args.get("keyword", default=None, type=str)
         tag_ids = request.args.getlist("tag_ids")
 
-        datasets, total = DatasetService.get_datasets(page, limit, provider, tenant_id, current_user, search, tag_ids)
+        datasets, total = DatasetService.get_datasets(page, limit, tenant_id, current_user, search, tag_ids)
         # check embedding setting
         provider_manager = ProviderManager()
         configurations = provider_manager.get_configurations(tenant_id=current_user.current_tenant_id)
@@ -82,6 +82,26 @@ def post(self, tenant_id):
             required=False,
             nullable=False,
         )
+        parser.add_argument(
+            "external_knowledge_api_id",
+            type=str,
+            nullable=True,
+            required=False,
+            default="_validate_name",
+        )
+        parser.add_argument(
+            "provider",
+            type=str,
+            nullable=True,
+            required=False,
+            default="vendor",
+        )
+        parser.add_argument(
+            "external_knowledge_id",
+            type=str,
+            nullable=True,
+            required=False,
+        )
         args = parser.parse_args()
 
         try:
@@ -91,6 +111,9 @@ def post(self, tenant_id):
                 indexing_technique=args["indexing_technique"],
                 account=current_user,
                 permission=args["permission"],
+                provider=args["provider"],
+                external_knowledge_api_id=args["external_knowledge_api_id"],
+                external_knowledge_id=args["external_knowledge_id"],
             )
         except services.errors.dataset.DatasetNameDuplicateError:
             raise DatasetNameDuplicateError()
diff --git a/api/core/callback_handler/index_tool_callback_handler.py b/api/core/callback_handler/index_tool_callback_handler.py
index 7cf472d984a38..23dc092895c8c 100644
--- a/api/core/callback_handler/index_tool_callback_handler.py
+++ b/api/core/callback_handler/index_tool_callback_handler.py
@@ -59,7 +59,7 @@ def return_retriever_resource_info(self, resource: list):
             for item in resource:
                 dataset_retriever_resource = DatasetRetrieverResource(
                     message_id=self._message_id,
-                    position=item.get("position"),
+                    position=item.get("position") or 0,
                     dataset_id=item.get("dataset_id"),
                     dataset_name=item.get("dataset_name"),
                     document_id=item.get("document_id"),
diff --git a/api/core/rag/datasource/retrieval_service.py b/api/core/rag/datasource/retrieval_service.py
index afac1bf30086e..d3fd0c672ad27 100644
--- a/api/core/rag/datasource/retrieval_service.py
+++ b/api/core/rag/datasource/retrieval_service.py
@@ -10,6 +10,7 @@
 from core.rag.retrieval.retrieval_methods import RetrievalMethod
 from extensions.ext_database import db
 from models.dataset import Dataset
+from services.external_knowledge_service import ExternalDatasetService
 
 default_retrieval_model = {
     "search_method": RetrievalMethod.SEMANTIC_SEARCH.value,
@@ -34,6 +35,9 @@ def retrieve(
         weights: Optional[dict] = None,
     ):
         dataset = db.session.query(Dataset).filter(Dataset.id == dataset_id).first()
+        if not dataset:
+            return []
+
         if not dataset or dataset.available_document_count == 0 or dataset.available_segment_count == 0:
             return []
         all_documents = []
@@ -108,6 +112,16 @@ def retrieve(
             )
         return all_documents
 
+    @classmethod
+    def external_retrieve(cls, dataset_id: str, query: str, external_retrieval_model: Optional[dict] = None):
+        dataset = db.session.query(Dataset).filter(Dataset.id == dataset_id).first()
+        if not dataset:
+            return []
+        all_documents = ExternalDatasetService.fetch_external_knowledge_retrieval(
+            dataset.tenant_id, dataset_id, query, external_retrieval_model
+        )
+        return all_documents
+
     @classmethod
     def keyword_search(
         cls, flask_app: Flask, dataset_id: str, query: str, top_k: int, all_documents: list, exceptions: list
diff --git a/api/core/rag/entities/context_entities.py b/api/core/rag/entities/context_entities.py
new file mode 100644
index 0000000000000..dde3beccf6e46
--- /dev/null
+++ b/api/core/rag/entities/context_entities.py
@@ -0,0 +1,10 @@
+from pydantic import BaseModel
+
+
+class DocumentContext(BaseModel):
+    """
+    Model class for document context.
+    """
+
+    content: str
+    score: float
diff --git a/api/core/rag/models/document.py b/api/core/rag/models/document.py
index 0ff1fdb81cb87..1e9aaa24f04c9 100644
--- a/api/core/rag/models/document.py
+++ b/api/core/rag/models/document.py
@@ -17,6 +17,8 @@ class Document(BaseModel):
     """
     metadata: Optional[dict] = Field(default_factory=dict)
 
+    provider: Optional[str] = "dify"
+
 
 class BaseDocumentTransformer(ABC):
     """Abstract base class for document transformation systems.
diff --git a/api/core/rag/rerank/rerank_model.py b/api/core/rag/rerank/rerank_model.py
index 6356ff87ab8ef..27f86aed34a81 100644
--- a/api/core/rag/rerank/rerank_model.py
+++ b/api/core/rag/rerank/rerank_model.py
@@ -28,11 +28,16 @@ def run(
         docs = []
         doc_id = []
         unique_documents = []
-        for document in documents:
+        dify_documents = [item for item in documents if item.provider == "dify"]
+        external_documents = [item for item in documents if item.provider == "external"]
+        for document in dify_documents:
             if document.metadata["doc_id"] not in doc_id:
                 doc_id.append(document.metadata["doc_id"])
                 docs.append(document.page_content)
                 unique_documents.append(document)
+        for document in external_documents:
+            docs.append(document.page_content)
+            unique_documents.append(document)
 
         documents = unique_documents
 
@@ -46,14 +51,10 @@ def run(
             # format document
             rerank_document = Document(
                 page_content=result.text,
-                metadata={
-                    "doc_id": documents[result.index].metadata["doc_id"],
-                    "doc_hash": documents[result.index].metadata["doc_hash"],
-                    "document_id": documents[result.index].metadata["document_id"],
-                    "dataset_id": documents[result.index].metadata["dataset_id"],
-                    "score": result.score,
-                },
+                metadata=documents[result.index].metadata,
+                provider=documents[result.index].provider,
             )
+            rerank_document.metadata["score"] = result.score
             rerank_documents.append(rerank_document)
 
         return rerank_documents
diff --git a/api/core/rag/retrieval/dataset_retrieval.py b/api/core/rag/retrieval/dataset_retrieval.py
index 4603957d68599..ae61ba7112243 100644
--- a/api/core/rag/retrieval/dataset_retrieval.py
+++ b/api/core/rag/retrieval/dataset_retrieval.py
@@ -20,6 +20,7 @@
 from core.rag.data_post_processor.data_post_processor import DataPostProcessor
 from core.rag.datasource.keyword.jieba.jieba_keyword_table_handler import JiebaKeywordTableHandler
 from core.rag.datasource.retrieval_service import RetrievalService
+from core.rag.entities.context_entities import DocumentContext
 from core.rag.models.document import Document
 from core.rag.retrieval.retrieval_methods import RetrievalMethod
 from core.rag.retrieval.router.multi_dataset_function_call_router import FunctionCallMultiDatasetRouter
@@ -30,6 +31,7 @@
 from extensions.ext_database import db
 from models.dataset import Dataset, DatasetQuery, DocumentSegment
 from models.dataset import Document as DatasetDocument
+from services.external_knowledge_service import ExternalDatasetService
 
 default_retrieval_model = {
     "search_method": RetrievalMethod.SEMANTIC_SEARCH.value,
@@ -110,7 +112,7 @@ def retrieve(
                 continue
 
             # pass if dataset is not available
-            if dataset and dataset.available_document_count == 0:
+            if dataset and dataset.available_document_count == 0 and dataset.provider != "external":
                 continue
 
             available_datasets.append(dataset)
@@ -146,69 +148,93 @@ def retrieve(
                 message_id,
             )
 
-        document_score_list = {}
-        for item in all_documents:
-            if item.metadata.get("score"):
-                document_score_list[item.metadata["doc_id"]] = item.metadata["score"]
-
+        dify_documents = [item for item in all_documents if item.provider == "dify"]
+        external_documents = [item for item in all_documents if item.provider == "external"]
         document_context_list = []
-        index_node_ids = [document.metadata["doc_id"] for document in all_documents]
-        segments = DocumentSegment.query.filter(
-            DocumentSegment.dataset_id.in_(dataset_ids),
-            DocumentSegment.completed_at.isnot(None),
-            DocumentSegment.status == "completed",
-            DocumentSegment.enabled == True,
-            DocumentSegment.index_node_id.in_(index_node_ids),
-        ).all()
-
-        if segments:
-            index_node_id_to_position = {id: position for position, id in enumerate(index_node_ids)}
-            sorted_segments = sorted(
-                segments, key=lambda segment: index_node_id_to_position.get(segment.index_node_id, float("inf"))
-            )
-            for segment in sorted_segments:
-                if segment.answer:
-                    document_context_list.append(f"question:{segment.get_sign_content()} answer:{segment.answer}")
-                else:
-                    document_context_list.append(segment.get_sign_content())
-            if show_retrieve_source:
-                context_list = []
-                resource_number = 1
+        retrieval_resource_list = []
+        # deal with external documents
+        for item in external_documents:
+            document_context_list.append(DocumentContext(content=item.page_content, score=item.metadata.get("score")))
+            source = {
+                "dataset_id": item.metadata.get("dataset_id"),
+                "dataset_name": item.metadata.get("dataset_name"),
+                "document_name": item.metadata.get("title"),
+                "data_source_type": "external",
+                "retriever_from": invoke_from.to_source(),
+                "score": item.metadata.get("score"),
+                "content": item.page_content,
+            }
+            retrieval_resource_list.append(source)
+        document_score_list = {}
+        # deal with dify documents
+        if dify_documents:
+            for item in dify_documents:
+                if item.metadata.get("score"):
+                    document_score_list[item.metadata["doc_id"]] = item.metadata["score"]
+
+            index_node_ids = [document.metadata["doc_id"] for document in dify_documents]
+            segments = DocumentSegment.query.filter(
+                DocumentSegment.dataset_id.in_(dataset_ids),
+                DocumentSegment.status == "completed",
+                DocumentSegment.enabled == True,
+                DocumentSegment.index_node_id.in_(index_node_ids),
+            ).all()
+
+            if segments:
+                index_node_id_to_position = {id: position for position, id in enumerate(index_node_ids)}
+                sorted_segments = sorted(
+                    segments, key=lambda segment: index_node_id_to_position.get(segment.index_node_id, float("inf"))
+                )
                 for segment in sorted_segments:
-                    dataset = Dataset.query.filter_by(id=segment.dataset_id).first()
-                    document = DatasetDocument.query.filter(
-                        DatasetDocument.id == segment.document_id,
-                        DatasetDocument.enabled == True,
-                        DatasetDocument.archived == False,
-                    ).first()
-                    if dataset and document:
-                        source = {
-                            "position": resource_number,
-                            "dataset_id": dataset.id,
-                            "dataset_name": dataset.name,
-                            "document_id": document.id,
-                            "document_name": document.name,
-                            "data_source_type": document.data_source_type,
-                            "segment_id": segment.id,
-                            "retriever_from": invoke_from.to_source(),
-                            "score": document_score_list.get(segment.index_node_id, None),
-                        }
-
-                        if invoke_from.to_source() == "dev":
-                            source["hit_count"] = segment.hit_count
-                            source["word_count"] = segment.word_count
-                            source["segment_position"] = segment.position
-                            source["index_node_hash"] = segment.index_node_hash
-                        if segment.answer:
-                            source["content"] = f"question:{segment.content} \nanswer:{segment.answer}"
-                        else:
-                            source["content"] = segment.content
-                        context_list.append(source)
-                    resource_number += 1
-                if hit_callback:
-                    hit_callback.return_retriever_resource_info(context_list)
-
-            return str("\n".join(document_context_list))
+                    if segment.answer:
+                        document_context_list.append(
+                            DocumentContext(
+                                content=f"question:{segment.get_sign_content()} answer:{segment.answer}",
+                                score=document_score_list.get(segment.index_node_id, None),
+                            )
+                        )
+                    else:
+                        document_context_list.append(
+                            DocumentContext(
+                                content=segment.get_sign_content(),
+                                score=document_score_list.get(segment.index_node_id, None),
+                            )
+                        )
+                if show_retrieve_source:
+                    for segment in sorted_segments:
+                        dataset = Dataset.query.filter_by(id=segment.dataset_id).first()
+                        document = DatasetDocument.query.filter(
+                            DatasetDocument.id == segment.document_id,
+                            DatasetDocument.enabled == True,
+                            DatasetDocument.archived == False,
+                        ).first()
+                        if dataset and document:
+                            source = {
+                                "dataset_id": dataset.id,
+                                "dataset_name": dataset.name,
+                                "document_id": document.id,
+                                "document_name": document.name,
+                                "data_source_type": document.data_source_type,
+                                "segment_id": segment.id,
+                                "retriever_from": invoke_from.to_source(),
+                                "score": document_score_list.get(segment.index_node_id, None),
+                            }
+
+                            if invoke_from.to_source() == "dev":
+                                source["hit_count"] = segment.hit_count
+                                source["word_count"] = segment.word_count
+                                source["segment_position"] = segment.position
+                                source["index_node_hash"] = segment.index_node_hash
+                            if segment.answer:
+                                source["content"] = f"question:{segment.content} \nanswer:{segment.answer}"
+                            else:
+                                source["content"] = segment.content
+                            retrieval_resource_list.append(source)
+        if hit_callback and retrieval_resource_list:
+            hit_callback.return_retriever_resource_info(retrieval_resource_list)
+        if document_context_list:
+            document_context_list = sorted(document_context_list, key=lambda x: x.score, reverse=True)
+            return str("\n".join([document_context.content for document_context in document_context_list]))
         return ""
 
     def single_retrieve(
@@ -256,36 +282,58 @@ def single_retrieve(
             # get retrieval model config
             dataset = db.session.query(Dataset).filter(Dataset.id == dataset_id).first()
             if dataset:
-                retrieval_model_config = dataset.retrieval_model or default_retrieval_model
-
-                # get top k
-                top_k = retrieval_model_config["top_k"]
-                # get retrieval method
-                if dataset.indexing_technique == "economy":
-                    retrieval_method = "keyword_search"
-                else:
-                    retrieval_method = retrieval_model_config["search_method"]
-                # get reranking model
-                reranking_model = (
-                    retrieval_model_config["reranking_model"] if retrieval_model_config["reranking_enable"] else None
-                )
-                # get score threshold
-                score_threshold = 0.0
-                score_threshold_enabled = retrieval_model_config.get("score_threshold_enabled")
-                if score_threshold_enabled:
-                    score_threshold = retrieval_model_config.get("score_threshold")
-
-                with measure_time() as timer:
-                    results = RetrievalService.retrieve(
-                        retrieval_method=retrieval_method,
-                        dataset_id=dataset.id,
+                results = []
+                if dataset.provider == "external":
+                    external_documents = ExternalDatasetService.fetch_external_knowledge_retrieval(
+                        tenant_id=dataset.tenant_id,
+                        dataset_id=dataset_id,
                         query=query,
-                        top_k=top_k,
-                        score_threshold=score_threshold,
-                        reranking_model=reranking_model,
-                        reranking_mode=retrieval_model_config.get("reranking_mode", "reranking_model"),
-                        weights=retrieval_model_config.get("weights", None),
+                        external_retrieval_parameters=dataset.retrieval_model,
+                    )
+                    for external_document in external_documents:
+                        document = Document(
+                            page_content=external_document.get("content"),
+                            metadata=external_document.get("metadata"),
+                            provider="external",
+                        )
+                        document.metadata["score"] = external_document.get("score")
+                        document.metadata["title"] = external_document.get("title")
+                        document.metadata["dataset_id"] = dataset_id
+                        document.metadata["dataset_name"] = dataset.name
+                        results.append(document)
+                else:
+                    retrieval_model_config = dataset.retrieval_model or default_retrieval_model
+
+                    # get top k
+                    top_k = retrieval_model_config["top_k"]
+                    # get retrieval method
+                    if dataset.indexing_technique == "economy":
+                        retrieval_method = "keyword_search"
+                    else:
+                        retrieval_method = retrieval_model_config["search_method"]
+                    # get reranking model
+                    reranking_model = (
+                        retrieval_model_config["reranking_model"]
+                        if retrieval_model_config["reranking_enable"]
+                        else None
                     )
+                    # get score threshold
+                    score_threshold = 0.0
+                    score_threshold_enabled = retrieval_model_config.get("score_threshold_enabled")
+                    if score_threshold_enabled:
+                        score_threshold = retrieval_model_config.get("score_threshold")
+
+                    with measure_time() as timer:
+                        results = RetrievalService.retrieve(
+                            retrieval_method=retrieval_method,
+                            dataset_id=dataset.id,
+                            query=query,
+                            top_k=top_k,
+                            score_threshold=score_threshold,
+                            reranking_model=reranking_model,
+                            reranking_mode=retrieval_model_config.get("reranking_mode", "reranking_model"),
+                            weights=retrieval_model_config.get("weights", None),
+                        )
                 self._on_query(query, [dataset_id], app_id, user_from, user_id)
 
                 if results:
@@ -356,7 +404,8 @@ def _on_retrieval_end(
         self, documents: list[Document], message_id: Optional[str] = None, timer: Optional[dict] = None
     ) -> None:
         """Handle retrieval end."""
-        for document in documents:
+        dify_documents = [document for document in documents if document.provider == "dify"]
+        for document in dify_documents:
             query = db.session.query(DocumentSegment).filter(
                 DocumentSegment.index_node_id == document.metadata["doc_id"]
             )
@@ -409,35 +458,54 @@ def _retriever(self, flask_app: Flask, dataset_id: str, query: str, top_k: int,
             if not dataset:
                 return []
 
-            # get retrieval model , if the model is not setting , using default
-            retrieval_model = dataset.retrieval_model or default_retrieval_model
-
-            if dataset.indexing_technique == "economy":
-                # use keyword table query
-                documents = RetrievalService.retrieve(
-                    retrieval_method="keyword_search", dataset_id=dataset.id, query=query, top_k=top_k
+            if dataset.provider == "external":
+                external_documents = ExternalDatasetService.fetch_external_knowledge_retrieval(
+                    tenant_id=dataset.tenant_id,
+                    dataset_id=dataset_id,
+                    query=query,
+                    external_retrieval_parameters=dataset.retrieval_model,
                 )
-                if documents:
-                    all_documents.extend(documents)
+                for external_document in external_documents:
+                    document = Document(
+                        page_content=external_document.get("content"),
+                        metadata=external_document.get("metadata"),
+                        provider="external",
+                    )
+                    document.metadata["score"] = external_document.get("score")
+                    document.metadata["title"] = external_document.get("title")
+                    document.metadata["dataset_id"] = dataset_id
+                    document.metadata["dataset_name"] = dataset.name
+                    all_documents.append(document)
             else:
-                if top_k > 0:
-                    # retrieval source
+                # get retrieval model , if the model is not setting , using default
+                retrieval_model = dataset.retrieval_model or default_retrieval_model
+
+                if dataset.indexing_technique == "economy":
+                    # use keyword table query
                     documents = RetrievalService.retrieve(
-                        retrieval_method=retrieval_model["search_method"],
-                        dataset_id=dataset.id,
-                        query=query,
-                        top_k=retrieval_model.get("top_k") or 2,
-                        score_threshold=retrieval_model.get("score_threshold", 0.0)
-                        if retrieval_model["score_threshold_enabled"]
-                        else 0.0,
-                        reranking_model=retrieval_model.get("reranking_model", None)
-                        if retrieval_model["reranking_enable"]
-                        else None,
-                        reranking_mode=retrieval_model.get("reranking_mode") or "reranking_model",
-                        weights=retrieval_model.get("weights", None),
+                        retrieval_method="keyword_search", dataset_id=dataset.id, query=query, top_k=top_k
                     )
-
-                    all_documents.extend(documents)
+                    if documents:
+                        all_documents.extend(documents)
+                else:
+                    if top_k > 0:
+                        # retrieval source
+                        documents = RetrievalService.retrieve(
+                            retrieval_method=retrieval_model["search_method"],
+                            dataset_id=dataset.id,
+                            query=query,
+                            top_k=retrieval_model.get("top_k") or 2,
+                            score_threshold=retrieval_model.get("score_threshold", 0.0)
+                            if retrieval_model["score_threshold_enabled"]
+                            else 0.0,
+                            reranking_model=retrieval_model.get("reranking_model", None)
+                            if retrieval_model["reranking_enable"]
+                            else None,
+                            reranking_mode=retrieval_model.get("reranking_mode") or "reranking_model",
+                            weights=retrieval_model.get("weights", None),
+                        )
+
+                        all_documents.extend(documents)
 
     def to_dataset_retriever_tool(
         self,
diff --git a/api/core/workflow/nodes/knowledge_retrieval/knowledge_retrieval_node.py b/api/core/workflow/nodes/knowledge_retrieval/knowledge_retrieval_node.py
index af55688a52c9b..c08dc143a6f32 100644
--- a/api/core/workflow/nodes/knowledge_retrieval/knowledge_retrieval_node.py
+++ b/api/core/workflow/nodes/knowledge_retrieval/knowledge_retrieval_node.py
@@ -156,16 +156,34 @@ def _fetch_dataset_retriever(self, node_data: KnowledgeRetrievalNodeData, query:
                 weights,
                 node_data.multiple_retrieval_config.reranking_enable,
             )
-
-        context_list = []
-        if all_documents:
+        dify_documents = [item for item in all_documents if item.provider == "dify"]
+        external_documents = [item for item in all_documents if item.provider == "external"]
+        retrieval_resource_list = []
+        # deal with external documents
+        for item in external_documents:
+            source = {
+                "metadata": {
+                    "_source": "knowledge",
+                    "dataset_id": item.metadata.get("dataset_id"),
+                    "dataset_name": item.metadata.get("dataset_name"),
+                    "document_name": item.metadata.get("title"),
+                    "data_source_type": "external",
+                    "retriever_from": "workflow",
+                    "score": item.metadata.get("score"),
+                },
+                "title": item.metadata.get("title"),
+                "content": item.page_content,
+            }
+            retrieval_resource_list.append(source)
+        document_score_list = {}
+        # deal with dify documents
+        if dify_documents:
             document_score_list = {}
-            page_number_list = {}
-            for item in all_documents:
+            for item in dify_documents:
                 if item.metadata.get("score"):
                     document_score_list[item.metadata["doc_id"]] = item.metadata["score"]
 
-            index_node_ids = [document.metadata["doc_id"] for document in all_documents]
+            index_node_ids = [document.metadata["doc_id"] for document in dify_documents]
             segments = DocumentSegment.query.filter(
                 DocumentSegment.dataset_id.in_(dataset_ids),
                 DocumentSegment.completed_at.isnot(None),
@@ -186,13 +204,10 @@ def _fetch_dataset_retriever(self, node_data: KnowledgeRetrievalNodeData, query:
                         Document.enabled == True,
                         Document.archived == False,
                     ).first()
-
-                    resource_number = 1
                     if dataset and document:
                         source = {
                             "metadata": {
                                 "_source": "knowledge",
-                                "position": resource_number,
                                 "dataset_id": dataset.id,
                                 "dataset_name": dataset.name,
                                 "document_id": document.id,
@@ -212,9 +227,14 @@ def _fetch_dataset_retriever(self, node_data: KnowledgeRetrievalNodeData, query:
                             source["content"] = f"question:{segment.get_sign_content()} \nanswer:{segment.answer}"
                         else:
                             source["content"] = segment.get_sign_content()
-                        context_list.append(source)
-                        resource_number += 1
-        return context_list
+                        retrieval_resource_list.append(source)
+        if retrieval_resource_list:
+            retrieval_resource_list = sorted(retrieval_resource_list, key=lambda x: x.get("score"), reverse=True)
+            position = 1
+            for item in retrieval_resource_list:
+                item["metadata"]["position"] = position
+                position += 1
+        return retrieval_resource_list
 
     @classmethod
     def _extract_variable_selector_to_variable_mapping(
diff --git a/api/fields/dataset_fields.py b/api/fields/dataset_fields.py
index 9cf8da7acdc98..b32423f10c9dd 100644
--- a/api/fields/dataset_fields.py
+++ b/api/fields/dataset_fields.py
@@ -38,9 +38,20 @@
     "score_threshold_enabled": fields.Boolean,
     "score_threshold": fields.Float,
 }
+external_retrieval_model_fields = {
+    "top_k": fields.Integer,
+    "score_threshold": fields.Float,
+}
 
 tag_fields = {"id": fields.String, "name": fields.String, "type": fields.String}
 
+external_knowledge_info_fields = {
+    "external_knowledge_id": fields.String,
+    "external_knowledge_api_id": fields.String,
+    "external_knowledge_api_name": fields.String,
+    "external_knowledge_api_endpoint": fields.String,
+}
+
 dataset_detail_fields = {
     "id": fields.String,
     "name": fields.String,
@@ -61,6 +72,8 @@
     "embedding_available": fields.Boolean,
     "retrieval_model_dict": fields.Nested(dataset_retrieval_model_fields),
     "tags": fields.List(fields.Nested(tag_fields)),
+    "external_knowledge_info": fields.Nested(external_knowledge_info_fields),
+    "external_retrieval_model": fields.Nested(external_retrieval_model_fields, allow_null=True),
 }
 
 dataset_query_detail_fields = {
diff --git a/api/fields/external_dataset_fields.py b/api/fields/external_dataset_fields.py
new file mode 100644
index 0000000000000..2281460fe2214
--- /dev/null
+++ b/api/fields/external_dataset_fields.py
@@ -0,0 +1,11 @@
+from flask_restful import fields
+
+from libs.helper import TimestampField
+
+external_knowledge_api_query_detail_fields = {
+    "id": fields.String,
+    "name": fields.String,
+    "setting": fields.String,
+    "created_by": fields.String,
+    "created_at": TimestampField,
+}
diff --git a/api/migrations/versions/2024_09_24_0922-6af6a521a53e_update_retrieval_resource.py b/api/migrations/versions/2024_09_24_0922-6af6a521a53e_update_retrieval_resource.py
new file mode 100644
index 0000000000000..5337b340db769
--- /dev/null
+++ b/api/migrations/versions/2024_09_24_0922-6af6a521a53e_update_retrieval_resource.py
@@ -0,0 +1,48 @@
+"""update-retrieval-resource
+
+Revision ID: 6af6a521a53e
+Revises: ec3df697ebbb
+Create Date: 2024-09-24 09:22:43.570120
+
+"""
+from alembic import op
+import models as models
+import sqlalchemy as sa
+from sqlalchemy.dialects import postgresql
+
+# revision identifiers, used by Alembic.
+revision = '6af6a521a53e'
+down_revision = 'd57ba9ebb251'
+branch_labels = None
+depends_on = None
+
+
+def upgrade():
+    # ### commands auto generated by Alembic - please adjust! ###
+    with op.batch_alter_table('dataset_retriever_resources', schema=None) as batch_op:
+        batch_op.alter_column('document_id',
+               existing_type=sa.UUID(),
+               nullable=True)
+        batch_op.alter_column('data_source_type',
+               existing_type=sa.TEXT(),
+               nullable=True)
+        batch_op.alter_column('segment_id',
+               existing_type=sa.UUID(),
+               nullable=True)
+    # ### end Alembic commands ###
+
+
+def downgrade():
+    # ### commands auto generated by Alembic - please adjust! ###
+    with op.batch_alter_table('dataset_retriever_resources', schema=None) as batch_op:
+        batch_op.alter_column('segment_id',
+               existing_type=sa.UUID(),
+               nullable=False)
+        batch_op.alter_column('data_source_type',
+               existing_type=sa.TEXT(),
+               nullable=False)
+        batch_op.alter_column('document_id',
+               existing_type=sa.UUID(),
+               nullable=False)
+
+    # ### end Alembic commands ###
diff --git a/api/migrations/versions/2024_09_25_0434-33f5fac87f29_external_knowledge_api.py b/api/migrations/versions/2024_09_25_0434-33f5fac87f29_external_knowledge_api.py
new file mode 100644
index 0000000000000..3cb76e72c1ebb
--- /dev/null
+++ b/api/migrations/versions/2024_09_25_0434-33f5fac87f29_external_knowledge_api.py
@@ -0,0 +1,73 @@
+"""external_knowledge_api
+
+Revision ID: 33f5fac87f29
+Revises: 6af6a521a53e
+Create Date: 2024-09-25 04:34:57.249436
+
+"""
+from alembic import op
+import models as models
+import sqlalchemy as sa
+from sqlalchemy.dialects import postgresql
+
+# revision identifiers, used by Alembic.
+revision = '33f5fac87f29'
+down_revision = '6af6a521a53e'
+branch_labels = None
+depends_on = None
+
+
+def upgrade():
+    # ### commands auto generated by Alembic - please adjust! ###
+    op.create_table('external_knowledge_apis',
+    sa.Column('id', models.types.StringUUID(), server_default=sa.text('uuid_generate_v4()'), nullable=False),
+    sa.Column('name', sa.String(length=255), nullable=False),
+    sa.Column('description', sa.String(length=255), nullable=False),
+    sa.Column('tenant_id', models.types.StringUUID(), nullable=False),
+    sa.Column('settings', sa.Text(), nullable=True),
+    sa.Column('created_by', models.types.StringUUID(), nullable=False),
+    sa.Column('created_at', sa.DateTime(), server_default=sa.text('CURRENT_TIMESTAMP(0)'), nullable=False),
+    sa.Column('updated_by', models.types.StringUUID(), nullable=True),
+    sa.Column('updated_at', sa.DateTime(), server_default=sa.text('CURRENT_TIMESTAMP(0)'), nullable=False),
+    sa.PrimaryKeyConstraint('id', name='external_knowledge_apis_pkey')
+    )
+    with op.batch_alter_table('external_knowledge_apis', schema=None) as batch_op:
+        batch_op.create_index('external_knowledge_apis_name_idx', ['name'], unique=False)
+        batch_op.create_index('external_knowledge_apis_tenant_idx', ['tenant_id'], unique=False)
+
+    op.create_table('external_knowledge_bindings',
+    sa.Column('id', models.types.StringUUID(), server_default=sa.text('uuid_generate_v4()'), nullable=False),
+    sa.Column('tenant_id', models.types.StringUUID(), nullable=False),
+    sa.Column('external_knowledge_api_id', models.types.StringUUID(), nullable=False),
+    sa.Column('dataset_id', models.types.StringUUID(), nullable=False),
+    sa.Column('external_knowledge_id', sa.Text(), nullable=False),
+    sa.Column('created_by', models.types.StringUUID(), nullable=False),
+    sa.Column('created_at', sa.DateTime(), server_default=sa.text('CURRENT_TIMESTAMP(0)'), nullable=False),
+    sa.Column('updated_by', models.types.StringUUID(), nullable=True),
+    sa.Column('updated_at', sa.DateTime(), server_default=sa.text('CURRENT_TIMESTAMP(0)'), nullable=False),
+    sa.PrimaryKeyConstraint('id', name='external_knowledge_bindings_pkey')
+    )
+    with op.batch_alter_table('external_knowledge_bindings', schema=None) as batch_op:
+        batch_op.create_index('external_knowledge_bindings_dataset_idx', ['dataset_id'], unique=False)
+        batch_op.create_index('external_knowledge_bindings_external_knowledge_api_idx', ['external_knowledge_api_id'], unique=False)
+        batch_op.create_index('external_knowledge_bindings_external_knowledge_idx', ['external_knowledge_id'], unique=False)
+        batch_op.create_index('external_knowledge_bindings_tenant_idx', ['tenant_id'], unique=False)
+
+    # ### end Alembic commands ###
+
+
+def downgrade():
+    # ### commands auto generated by Alembic - please adjust! ###
+    with op.batch_alter_table('external_knowledge_bindings', schema=None) as batch_op:
+        batch_op.drop_index('external_knowledge_bindings_tenant_idx')
+        batch_op.drop_index('external_knowledge_bindings_external_knowledge_idx')
+        batch_op.drop_index('external_knowledge_bindings_external_knowledge_api_idx')
+        batch_op.drop_index('external_knowledge_bindings_dataset_idx')
+
+    op.drop_table('external_knowledge_bindings')
+    with op.batch_alter_table('external_knowledge_apis', schema=None) as batch_op:
+        batch_op.drop_index('external_knowledge_apis_tenant_idx')
+        batch_op.drop_index('external_knowledge_apis_name_idx')
+
+    op.drop_table('external_knowledge_apis')
+    # ### end Alembic commands ###
diff --git a/api/migrations/versions/fca025d3b60f_add_dataset_retrival_model.py b/api/migrations/versions/fca025d3b60f_add_dataset_retrival_model.py
index 1f8250c3eb42a..52495be60a62e 100644
--- a/api/migrations/versions/fca025d3b60f_add_dataset_retrival_model.py
+++ b/api/migrations/versions/fca025d3b60f_add_dataset_retrival_model.py
@@ -1,4 +1,4 @@
-"""add-dataset-retrival-model
+"""add-dataset-retrieval-model
 
 Revision ID: fca025d3b60f
 Revises: b3a09c049e8e
diff --git a/api/models/dataset.py b/api/models/dataset.py
index a2d2a3454d493..4224ee5e9cadd 100644
--- a/api/models/dataset.py
+++ b/api/models/dataset.py
@@ -38,6 +38,7 @@ class Dataset(db.Model):
     )
 
     INDEXING_TECHNIQUE_LIST = ["high_quality", "economy", None]
+    PROVIDER_LIST = ["vendor", "external", None]
 
     id = db.Column(StringUUID, server_default=db.text("uuid_generate_v4()"))
     tenant_id = db.Column(StringUUID, nullable=False)
@@ -71,6 +72,14 @@ def dataset_keyword_table(self):
     def index_struct_dict(self):
         return json.loads(self.index_struct) if self.index_struct else None
 
+    @property
+    def external_retrieval_model(self):
+        default_retrieval_model = {
+            "top_k": 2,
+            "score_threshold": 0.0,
+        }
+        return self.retrieval_model or default_retrieval_model
+
     @property
     def created_by_account(self):
         return db.session.get(Account, self.created_by)
@@ -162,6 +171,29 @@ def tags(self):
 
         return tags or []
 
+    @property
+    def external_knowledge_info(self):
+        if self.provider != "external":
+            return None
+        external_knowledge_binding = (
+            db.session.query(ExternalKnowledgeBindings).filter(ExternalKnowledgeBindings.dataset_id == self.id).first()
+        )
+        if not external_knowledge_binding:
+            return None
+        external_knowledge_api = (
+            db.session.query(ExternalKnowledgeApis)
+            .filter(ExternalKnowledgeApis.id == external_knowledge_binding.external_knowledge_api_id)
+            .first()
+        )
+        if not external_knowledge_api:
+            return None
+        return {
+            "external_knowledge_id": external_knowledge_binding.external_knowledge_id,
+            "external_knowledge_api_id": external_knowledge_api.id,
+            "external_knowledge_api_name": external_knowledge_api.name,
+            "external_knowledge_api_endpoint": json.loads(external_knowledge_api.settings).get("endpoint", ""),
+        }
+
     @staticmethod
     def gen_collection_name_by_id(dataset_id: str) -> str:
         normalized_dataset_id = dataset_id.replace("-", "_")
@@ -687,3 +719,77 @@ class DatasetPermission(db.Model):
     tenant_id = db.Column(StringUUID, nullable=False)
     has_permission = db.Column(db.Boolean, nullable=False, server_default=db.text("true"))
     created_at = db.Column(db.DateTime, nullable=False, server_default=db.text("CURRENT_TIMESTAMP(0)"))
+
+
+class ExternalKnowledgeApis(db.Model):
+    __tablename__ = "external_knowledge_apis"
+    __table_args__ = (
+        db.PrimaryKeyConstraint("id", name="external_knowledge_apis_pkey"),
+        db.Index("external_knowledge_apis_tenant_idx", "tenant_id"),
+        db.Index("external_knowledge_apis_name_idx", "name"),
+    )
+
+    id = db.Column(StringUUID, nullable=False, server_default=db.text("uuid_generate_v4()"))
+    name = db.Column(db.String(255), nullable=False)
+    description = db.Column(db.String(255), nullable=False)
+    tenant_id = db.Column(StringUUID, nullable=False)
+    settings = db.Column(db.Text, nullable=True)
+    created_by = db.Column(StringUUID, nullable=False)
+    created_at = db.Column(db.DateTime, nullable=False, server_default=db.text("CURRENT_TIMESTAMP(0)"))
+    updated_by = db.Column(StringUUID, nullable=True)
+    updated_at = db.Column(db.DateTime, nullable=False, server_default=db.text("CURRENT_TIMESTAMP(0)"))
+
+    def to_dict(self):
+        return {
+            "id": self.id,
+            "tenant_id": self.tenant_id,
+            "name": self.name,
+            "description": self.description,
+            "settings": self.settings_dict,
+            "dataset_bindings": self.dataset_bindings,
+            "created_by": self.created_by,
+            "created_at": self.created_at.isoformat(),
+        }
+
+    @property
+    def settings_dict(self):
+        try:
+            return json.loads(self.settings) if self.settings else None
+        except JSONDecodeError:
+            return None
+
+    @property
+    def dataset_bindings(self):
+        external_knowledge_bindings = (
+            db.session.query(ExternalKnowledgeBindings)
+            .filter(ExternalKnowledgeBindings.external_knowledge_api_id == self.id)
+            .all()
+        )
+        dataset_ids = [binding.dataset_id for binding in external_knowledge_bindings]
+        datasets = db.session.query(Dataset).filter(Dataset.id.in_(dataset_ids)).all()
+        dataset_bindings = []
+        for dataset in datasets:
+            dataset_bindings.append({"id": dataset.id, "name": dataset.name})
+
+        return dataset_bindings
+
+
+class ExternalKnowledgeBindings(db.Model):
+    __tablename__ = "external_knowledge_bindings"
+    __table_args__ = (
+        db.PrimaryKeyConstraint("id", name="external_knowledge_bindings_pkey"),
+        db.Index("external_knowledge_bindings_tenant_idx", "tenant_id"),
+        db.Index("external_knowledge_bindings_dataset_idx", "dataset_id"),
+        db.Index("external_knowledge_bindings_external_knowledge_idx", "external_knowledge_id"),
+        db.Index("external_knowledge_bindings_external_knowledge_api_idx", "external_knowledge_api_id"),
+    )
+
+    id = db.Column(StringUUID, nullable=False, server_default=db.text("uuid_generate_v4()"))
+    tenant_id = db.Column(StringUUID, nullable=False)
+    external_knowledge_api_id = db.Column(StringUUID, nullable=False)
+    dataset_id = db.Column(StringUUID, nullable=False)
+    external_knowledge_id = db.Column(db.Text, nullable=False)
+    created_by = db.Column(StringUUID, nullable=False)
+    created_at = db.Column(db.DateTime, nullable=False, server_default=db.text("CURRENT_TIMESTAMP(0)"))
+    updated_by = db.Column(StringUUID, nullable=True)
+    updated_at = db.Column(db.DateTime, nullable=False, server_default=db.text("CURRENT_TIMESTAMP(0)"))
diff --git a/api/models/model.py b/api/models/model.py
index 53940a5a16cc8..0ac93343217d4 100644
--- a/api/models/model.py
+++ b/api/models/model.py
@@ -1423,10 +1423,10 @@ class DatasetRetrieverResource(db.Model):
     position = db.Column(db.Integer, nullable=False)
     dataset_id = db.Column(StringUUID, nullable=False)
     dataset_name = db.Column(db.Text, nullable=False)
-    document_id = db.Column(StringUUID, nullable=False)
+    document_id = db.Column(StringUUID, nullable=True)
     document_name = db.Column(db.Text, nullable=False)
-    data_source_type = db.Column(db.Text, nullable=False)
-    segment_id = db.Column(StringUUID, nullable=False)
+    data_source_type = db.Column(db.Text, nullable=True)
+    segment_id = db.Column(StringUUID, nullable=True)
     score = db.Column(db.Float, nullable=True)
     content = db.Column(db.Text, nullable=False)
     hit_count = db.Column(db.Integer, nullable=True)
diff --git a/api/poetry.lock b/api/poetry.lock
index 85c68cd75f129..4fb8d54f11852 100644
--- a/api/poetry.lock
+++ b/api/poetry.lock
@@ -2,13 +2,13 @@
 
 [[package]]
 name = "aiohappyeyeballs"
-version = "2.4.0"
+version = "2.4.2"
 description = "Happy Eyeballs for asyncio"
 optional = false
 python-versions = ">=3.8"
 files = [
-    {file = "aiohappyeyeballs-2.4.0-py3-none-any.whl", hash = "sha256:7ce92076e249169a13c2f49320d1967425eaf1f407522d707d59cac7628d62bd"},
-    {file = "aiohappyeyeballs-2.4.0.tar.gz", hash = "sha256:55a1714f084e63d49639800f95716da97a1f173d46a16dfcfda0016abb93b6b2"},
+    {file = "aiohappyeyeballs-2.4.2-py3-none-any.whl", hash = "sha256:8522691d9a154ba1145b157d6d5c15e5c692527ce6a53c5e5f9876977f6dab2f"},
+    {file = "aiohappyeyeballs-2.4.2.tar.gz", hash = "sha256:4ca893e6c5c1f5bf3888b04cb5a3bee24995398efef6e0b9f747b5e89d84fd74"},
 ]
 
 [[package]]
@@ -153,13 +153,13 @@ frozenlist = ">=1.1.0"
 
 [[package]]
 name = "alembic"
-version = "1.13.2"
+version = "1.13.3"
 description = "A database migration tool for SQLAlchemy."
 optional = false
 python-versions = ">=3.8"
 files = [
-    {file = "alembic-1.13.2-py3-none-any.whl", hash = "sha256:6b8733129a6224a9a711e17c99b08462dbf7cc9670ba8f2e2ae9af860ceb1953"},
-    {file = "alembic-1.13.2.tar.gz", hash = "sha256:1ff0ae32975f4fd96028c39ed9bb3c867fe3af956bd7bb37343b54c9fe7445ef"},
+    {file = "alembic-1.13.3-py3-none-any.whl", hash = "sha256:908e905976d15235fae59c9ac42c4c5b75cfcefe3d27c0fbf7ae15a37715d80e"},
+    {file = "alembic-1.13.3.tar.gz", hash = "sha256:203503117415561e203aa14541740643a611f641517f0209fcae63e9fa09f1a2"},
 ]
 
 [package.dependencies]
@@ -293,13 +293,13 @@ alibabacloud-tea = "*"
 
 [[package]]
 name = "alibabacloud-tea"
-version = "0.3.9"
+version = "0.3.10"
 description = "The tea module of alibabaCloud Python SDK."
 optional = false
 python-versions = ">=3.6"
 files = [
-    {file = "alibabacloud-tea-0.3.9.tar.gz", hash = "sha256:a9689770003fa9313d1995812f9fe36a2be315e5cdfc8d58de0d96808219ced9"},
-    {file = "alibabacloud_tea-0.3.9-py3-none-any.whl", hash = "sha256:402fd2a92e6729f228d8c0300b182f80019edce19d83afa497aeb15fd7947f9a"},
+    {file = "alibabacloud-tea-0.3.10.tar.gz", hash = "sha256:bcf972416af5d8b5e671078c2ec20296dbc792e85e68acd685730a0a016afd2a"},
+    {file = "alibabacloud_tea-0.3.10-py3-none-any.whl", hash = "sha256:9136f302a3baea8a1528f500bf5d47c3727b827a09b5c14b283ca53578e30082"},
 ]
 
 [package.dependencies]
@@ -321,17 +321,17 @@ alibabacloud-tea = ">=0.0.1"
 
 [[package]]
 name = "alibabacloud-tea-openapi"
-version = "0.3.11"
+version = "0.3.12"
 description = "Alibaba Cloud openapi SDK Library for Python"
 optional = false
 python-versions = ">=3.6"
 files = [
-    {file = "alibabacloud_tea_openapi-0.3.11.tar.gz", hash = "sha256:3f5cace1b1aeb8a64587574097403cfd066b86ee4c3c9abde587f9abfcad38de"},
+    {file = "alibabacloud_tea_openapi-0.3.12.tar.gz", hash = "sha256:2e14809f357438e62c1ef4976a7655110dd54a75bbfa7d905fa3798355cfd974"},
 ]
 
 [package.dependencies]
-alibabacloud_credentials = ">=0.3.1,<1.0.0"
-alibabacloud_gateway_spi = ">=0.0.1,<1.0.0"
+alibabacloud_credentials = ">=0.3.5,<1.0.0"
+alibabacloud_gateway_spi = ">=0.0.2,<1.0.0"
 alibabacloud_openapi_util = ">=0.2.1,<1.0.0"
 alibabacloud_tea_util = ">=0.3.13,<1.0.0"
 alibabacloud_tea_xml = ">=0.0.2,<1.0.0"
@@ -455,13 +455,13 @@ vertex = ["google-auth (>=2,<3)"]
 
 [[package]]
 name = "anyio"
-version = "4.4.0"
+version = "4.6.0"
 description = "High level compatibility layer for multiple asynchronous event loop implementations"
 optional = false
-python-versions = ">=3.8"
+python-versions = ">=3.9"
 files = [
-    {file = "anyio-4.4.0-py3-none-any.whl", hash = "sha256:c1b2d8f46a8a812513012e1107cb0e68c17159a7a594208005a57dc776e1bdc7"},
-    {file = "anyio-4.4.0.tar.gz", hash = "sha256:5aadc6a1bbb7cdb0bede386cac5e2940f5e2ff3aa20277e991cf028e0585ce94"},
+    {file = "anyio-4.6.0-py3-none-any.whl", hash = "sha256:c7d2e9d63e31599eeb636c8c5c03a7e108d73b345f064f1c19fdc87b79036a9a"},
+    {file = "anyio-4.6.0.tar.gz", hash = "sha256:137b4559cbb034c477165047febb6ff83f390fc3b20bf181c1fc0a728cb8beeb"},
 ]
 
 [package.dependencies]
@@ -471,9 +471,9 @@ sniffio = ">=1.1"
 typing-extensions = {version = ">=4.1", markers = "python_version < \"3.11\""}
 
 [package.extras]
-doc = ["Sphinx (>=7)", "packaging", "sphinx-autodoc-typehints (>=1.2.0)", "sphinx-rtd-theme"]
-test = ["anyio[trio]", "coverage[toml] (>=7)", "exceptiongroup (>=1.2.0)", "hypothesis (>=4.0)", "psutil (>=5.9)", "pytest (>=7.0)", "pytest-mock (>=3.6.1)", "trustme", "uvloop (>=0.17)"]
-trio = ["trio (>=0.23)"]
+doc = ["Sphinx (>=7.4,<8.0)", "packaging", "sphinx-autodoc-typehints (>=1.2.0)", "sphinx-rtd-theme"]
+test = ["anyio[trio]", "coverage[toml] (>=7)", "exceptiongroup (>=1.2.0)", "hypothesis (>=4.0)", "psutil (>=5.9)", "pytest (>=7.0)", "pytest-mock (>=3.6.1)", "trustme", "uvloop (>=0.21.0b1)"]
+trio = ["trio (>=0.26.1)"]
 
 [[package]]
 name = "arxiv"
@@ -698,23 +698,23 @@ msrest = ">=0.6.21"
 
 [[package]]
 name = "azure-storage-file-share"
-version = "12.17.0"
+version = "12.18.0"
 description = "Microsoft Azure Azure File Share Storage Client Library for Python"
 optional = false
 python-versions = ">=3.8"
 files = [
-    {file = "azure-storage-file-share-12.17.0.tar.gz", hash = "sha256:f7b2c6cfc1b7cb80097a53b1ed2efa9e545b49a291430d369cdb49fafbc841d6"},
-    {file = "azure_storage_file_share-12.17.0-py3-none-any.whl", hash = "sha256:c4652759a9d529bf08881bb53275bf38774bb643746b849d27c47118f9cf923d"},
+    {file = "azure_storage_file_share-12.18.0-py3-none-any.whl", hash = "sha256:23ca35206a0cb8af0decd9d1363d0ad8ab31584a5e55cfc64528d7192ff748a9"},
+    {file = "azure_storage_file_share-12.18.0.tar.gz", hash = "sha256:0a81daee5e13598accde3c73b1aeccc6edfdcec5e957bf40150c0622fb95dc76"},
 ]
 
 [package.dependencies]
-azure-core = ">=1.28.0"
+azure-core = ">=1.30.0"
 cryptography = ">=2.1.4"
 isodate = ">=0.6.1"
 typing-extensions = ">=4.6.0"
 
 [package.extras]
-aio = ["azure-core[aio] (>=1.28.0)"]
+aio = ["azure-core[aio] (>=1.30.0)"]
 
 [[package]]
 name = "backoff"
@@ -787,13 +787,13 @@ lxml = ["lxml"]
 
 [[package]]
 name = "billiard"
-version = "4.2.0"
+version = "4.2.1"
 description = "Python multiprocessing fork with improvements and bugfixes"
 optional = false
 python-versions = ">=3.7"
 files = [
-    {file = "billiard-4.2.0-py3-none-any.whl", hash = "sha256:07aa978b308f334ff8282bd4a746e681b3513db5c9a514cbdd810cbbdc19714d"},
-    {file = "billiard-4.2.0.tar.gz", hash = "sha256:9a3c3184cb275aa17a732f93f65b20c525d3d9f253722d26a82194803ade5a2c"},
+    {file = "billiard-4.2.1-py3-none-any.whl", hash = "sha256:40b59a4ac8806ba2c2369ea98d876bc6108b051c227baffd928c644d15d8f3cb"},
+    {file = "billiard-4.2.1.tar.gz", hash = "sha256:12b641b0c539073fc8d3f5b8b7be998956665c4233c7c1fcd66a7e677c4fb36f"},
 ]
 
 [[package]]
@@ -828,13 +828,13 @@ crt = ["botocore[crt] (>=1.21.0,<2.0a0)"]
 
 [[package]]
 name = "botocore"
-version = "1.35.19"
+version = "1.35.29"
 description = "Low-level, data-driven core of boto 3."
 optional = false
 python-versions = ">=3.8"
 files = [
-    {file = "botocore-1.35.19-py3-none-any.whl", hash = "sha256:c83f7f0cacfe7c19b109b363ebfa8736e570d24922f16ed371681f58ebab44a9"},
-    {file = "botocore-1.35.19.tar.gz", hash = "sha256:42d6d8db7250cbd7899f786f9861e02cab17dc238f64d6acb976098ed9809625"},
+    {file = "botocore-1.35.29-py3-none-any.whl", hash = "sha256:f8e3ae0d84214eff3fb69cb4dc51cea6c43d3bde82027a94d00c52b941d6c3d5"},
+    {file = "botocore-1.35.29.tar.gz", hash = "sha256:4ed28ab03675bb008a290c452c5ddd7aaa5d4e3fa1912aadbdf93057ee84362b"},
 ]
 
 [package.dependencies]
@@ -2077,13 +2077,13 @@ tokenizer = ["tiktoken"]
 
 [[package]]
 name = "dataclass-wizard"
-version = "0.22.3"
+version = "0.23.0"
 description = "Marshal dataclasses to/from JSON. Use field properties with initial values. Construct a dataclass schema with JSON input."
 optional = false
 python-versions = "*"
 files = [
-    {file = "dataclass-wizard-0.22.3.tar.gz", hash = "sha256:4c46591782265058f1148cfd1f54a3a91221e63986fdd04c9d59f4ced61f4424"},
-    {file = "dataclass_wizard-0.22.3-py2.py3-none-any.whl", hash = "sha256:63751203e54b9b9349212cc185331da73c1adc99c51312575eb73bb5c00c1962"},
+    {file = "dataclass-wizard-0.23.0.tar.gz", hash = "sha256:da29ec19846d46a1eef0692ba7c59c8a86ecd3a9eaddc0511cfc7485ad6d9c50"},
+    {file = "dataclass_wizard-0.23.0-py2.py3-none-any.whl", hash = "sha256:50207dec6d36494421366b49b7a9ba6a4d831e2650c0af25cb4c057103d4a97c"},
 ]
 
 [package.extras]
@@ -2164,13 +2164,13 @@ dev = ["PyTest", "PyTest-Cov", "bump2version (<1)", "sphinx (<2)", "tox"]
 
 [[package]]
 name = "dill"
-version = "0.3.8"
+version = "0.3.9"
 description = "serialize all of Python"
 optional = false
 python-versions = ">=3.8"
 files = [
-    {file = "dill-0.3.8-py3-none-any.whl", hash = "sha256:c36ca9ffb54365bdd2f8eb3eff7d2a21237f8452b57ace88b1ac615b7e815bd7"},
-    {file = "dill-0.3.8.tar.gz", hash = "sha256:3ebe3c479ad625c4553aca177444d89b486b1d84982eeacded644afc0cf797ca"},
+    {file = "dill-0.3.9-py3-none-any.whl", hash = "sha256:468dff3b89520b474c0397703366b7b95eebe6303f108adf9b19da1f702be87a"},
+    {file = "dill-0.3.9.tar.gz", hash = "sha256:81aa267dddf68cbfe8029c42ca9ec6a4ab3b22371d1c450abc54422577b4512c"},
 ]
 
 [package.extras]
@@ -2241,78 +2241,94 @@ typing_extensions = ">=4.0,<5.0"
 
 [[package]]
 name = "duckdb"
-version = "1.1.0"
+version = "1.1.1"
 description = "DuckDB in-process database"
 optional = false
 python-versions = ">=3.7.0"
 files = [
-    {file = "duckdb-1.1.0-cp310-cp310-macosx_12_0_arm64.whl", hash = "sha256:5e4cbc408e6e41146dea89b9044dae7356e353db0c96b183e5583ee02bc6ae5d"},
-    {file = "duckdb-1.1.0-cp310-cp310-macosx_12_0_universal2.whl", hash = "sha256:6370ae27ec8167ccfbefb94f58ad9fdc7bac142399960549d6d367f233189868"},
-    {file = "duckdb-1.1.0-cp310-cp310-macosx_12_0_x86_64.whl", hash = "sha256:4e1c3414f7fd01f4810dc8b335deffc91933a159282d65fef11c1286bc0ded04"},
-    {file = "duckdb-1.1.0-cp310-cp310-manylinux_2_17_aarch64.manylinux2014_aarch64.whl", hash = "sha256:c6bc2a58689adf5520303c5f68b065b9f980bd31f1366c541b8c7490abaf55cd"},
-    {file = "duckdb-1.1.0-cp310-cp310-manylinux_2_17_x86_64.manylinux2014_x86_64.whl", hash = "sha256:d02be208d2885ca085d4c852b911493b8cdac9d6eae893259da32bd72a437c25"},
-    {file = "duckdb-1.1.0-cp310-cp310-manylinux_2_24_aarch64.manylinux_2_28_aarch64.whl", hash = "sha256:655df442ceebfc6f3fd6c8766e04b60d44dddedfa90275d794f9fab2d3180879"},
-    {file = "duckdb-1.1.0-cp310-cp310-musllinux_1_2_x86_64.whl", hash = "sha256:6e183729bb64be7798ccbfda6283ebf423c869268c25af2b56929e48f763be2f"},
-    {file = "duckdb-1.1.0-cp310-cp310-win_amd64.whl", hash = "sha256:61fb838da51e07ceb0222c4406b059b90e10efcc453c19a3650b73c0112138c4"},
-    {file = "duckdb-1.1.0-cp311-cp311-macosx_12_0_arm64.whl", hash = "sha256:7807e2f0d3344668e433f0dc1f54bfaddd410589611393e9a7ed56f8dec9514f"},
-    {file = "duckdb-1.1.0-cp311-cp311-macosx_12_0_universal2.whl", hash = "sha256:3da30b7b466f710d52caa1fdc3ef0bf4176ad7f115953cd9f8b0fbf0f723778f"},
-    {file = "duckdb-1.1.0-cp311-cp311-macosx_12_0_x86_64.whl", hash = "sha256:b9b6a77ef0183f561b1fc2945fcc762a71570ffd33fea4e3a855d413ed596fe4"},
-    {file = "duckdb-1.1.0-cp311-cp311-manylinux_2_17_aarch64.manylinux2014_aarch64.whl", hash = "sha256:16243e66a9fd0e64ee265f2634d137adc6593f54ddf3ef55cb8a29e1decf6e54"},
-    {file = "duckdb-1.1.0-cp311-cp311-manylinux_2_17_x86_64.manylinux2014_x86_64.whl", hash = "sha256:42b910a149e00f40a1766dc74fa309d4255b912a5d2fdcc387287658048650f6"},
-    {file = "duckdb-1.1.0-cp311-cp311-manylinux_2_24_aarch64.manylinux_2_28_aarch64.whl", hash = "sha256:47849d546dc4238c0f20e95fe53b621aa5b08684e68fff91fd84a7092be91a17"},
-    {file = "duckdb-1.1.0-cp311-cp311-musllinux_1_2_x86_64.whl", hash = "sha256:11ec967b67159361ceade34095796a8d19368ea5c30cad988f44896b082b0816"},
-    {file = "duckdb-1.1.0-cp311-cp311-win_amd64.whl", hash = "sha256:510b5885ed6c267b9c0e1e7c6138fdffc2dd6f934a5a95b76da85da127213338"},
-    {file = "duckdb-1.1.0-cp312-cp312-macosx_12_0_arm64.whl", hash = "sha256:657bc7ac64d5faf069a782ae73afac51ef30ae2e5d0e09ce6a09d03db84ab35e"},
-    {file = "duckdb-1.1.0-cp312-cp312-macosx_12_0_universal2.whl", hash = "sha256:89f3de8cba57d19b41cd3c47dd06d979bd2a2ffead115480e37afbe72b02896d"},
-    {file = "duckdb-1.1.0-cp312-cp312-macosx_12_0_x86_64.whl", hash = "sha256:f6486323ab20656d22ffa8f3c6e109dde30d0b327b7c831f22ebcfe747f97fb0"},
-    {file = "duckdb-1.1.0-cp312-cp312-manylinux_2_17_aarch64.manylinux2014_aarch64.whl", hash = "sha256:78a4510f82431ee3f14db689fe8727a4a9062c8f2fbb3bcfe3bfad3c1a198004"},
-    {file = "duckdb-1.1.0-cp312-cp312-manylinux_2_17_x86_64.manylinux2014_x86_64.whl", hash = "sha256:64bf2a6e23840d662bd2ac09206a9bd4fa657418884d69e5c352d4456dc70b3c"},
-    {file = "duckdb-1.1.0-cp312-cp312-manylinux_2_24_aarch64.manylinux_2_28_aarch64.whl", hash = "sha256:23fc9aa0af74e3803ed90c8d98280fd5bcac8c940592bf6288e8fd60fb051d00"},
-    {file = "duckdb-1.1.0-cp312-cp312-musllinux_1_2_x86_64.whl", hash = "sha256:1f3aea31341ce400640dd522e4399b941f66df17e39884f446638fe958d6117c"},
-    {file = "duckdb-1.1.0-cp312-cp312-win_amd64.whl", hash = "sha256:3db4ab31c20de4edaef152930836b38e7662cd71370748fdf2c38ba9cf854dc4"},
-    {file = "duckdb-1.1.0-cp37-cp37m-manylinux_2_17_aarch64.manylinux2014_aarch64.whl", hash = "sha256:e3b6b4fe1edfe35f64f403a9f0ab75258cee35abd964356893ee37424174b7e4"},
-    {file = "duckdb-1.1.0-cp37-cp37m-manylinux_2_17_x86_64.manylinux2014_x86_64.whl", hash = "sha256:aad02f50d5a2020822d1638fc1a9bcf082056f11d2e15ccfc1c1ed4d0f85a3be"},
-    {file = "duckdb-1.1.0-cp37-cp37m-manylinux_2_24_aarch64.manylinux_2_28_aarch64.whl", hash = "sha256:eb66e9e7391801928ea134dcab12d2e4c97f2ce0391c603a3e480bbb15830bc8"},
-    {file = "duckdb-1.1.0-cp37-cp37m-musllinux_1_2_x86_64.whl", hash = "sha256:069fb7bca459e31edb32a61f0eea95d7a8a766bef7b8318072563abf8e939593"},
-    {file = "duckdb-1.1.0-cp37-cp37m-win_amd64.whl", hash = "sha256:e39f9b7b62e64e10d421ff04480290a70129c38067d1a4f600e9212b10542c5a"},
-    {file = "duckdb-1.1.0-cp38-cp38-macosx_12_0_arm64.whl", hash = "sha256:55ef98bcc7ba745752607f1b926e8d9b7ce32c42c423bbad10c44820aefe23a7"},
-    {file = "duckdb-1.1.0-cp38-cp38-macosx_12_0_universal2.whl", hash = "sha256:e2a08175e43b865c1e9611efd18cacd29ddd69093de442b1ebdf312071df7719"},
-    {file = "duckdb-1.1.0-cp38-cp38-macosx_12_0_x86_64.whl", hash = "sha256:0e3644b1f034012d82b9baa12a7ea306fe71dc6623731b28c753c4a617ff9499"},
-    {file = "duckdb-1.1.0-cp38-cp38-manylinux_2_17_aarch64.manylinux2014_aarch64.whl", hash = "sha256:211a33c1ddb5cc609f75eb43772b0b03b45d2fa89bec107e4715267ca907806a"},
-    {file = "duckdb-1.1.0-cp38-cp38-manylinux_2_17_x86_64.manylinux2014_x86_64.whl", hash = "sha256:8e74b6f8a5145abbf7e6c1a2a61f0adbcd493c19b358f524ec9a3cebdf362abb"},
-    {file = "duckdb-1.1.0-cp38-cp38-manylinux_2_24_aarch64.manylinux_2_28_aarch64.whl", hash = "sha256:58f1633dd2c5af5088ae2d119418e200855d0699d84f2fae9d46d30f404bcead"},
-    {file = "duckdb-1.1.0-cp38-cp38-musllinux_1_2_x86_64.whl", hash = "sha256:d18caea926b1e301c29b140418fca697aad728129e269b4f82c2795a184549e1"},
-    {file = "duckdb-1.1.0-cp38-cp38-win_amd64.whl", hash = "sha256:cd9fb1408942411ad360f8414bc3fbf0091c396ca903d947a10f2e31324d5cbd"},
-    {file = "duckdb-1.1.0-cp39-cp39-macosx_12_0_arm64.whl", hash = "sha256:bd11bc899cebf5ff936d1276a2dfb7b7db08aba3bcc42924afeafc2163bddb43"},
-    {file = "duckdb-1.1.0-cp39-cp39-macosx_12_0_universal2.whl", hash = "sha256:53825a63193c582a78c152ea53de8d145744ddbeea18f452625a82ebc33eb14a"},
-    {file = "duckdb-1.1.0-cp39-cp39-macosx_12_0_x86_64.whl", hash = "sha256:29dc18087de47563b3859a6b98bbed96e1c96ce5db829646dc3b16a916997e7d"},
-    {file = "duckdb-1.1.0-cp39-cp39-manylinux_2_17_aarch64.manylinux2014_aarch64.whl", hash = "sha256:ecb19319883564237a7a03a104dbe7f445e73519bb67108fcab3d19b6b91fe30"},
-    {file = "duckdb-1.1.0-cp39-cp39-manylinux_2_17_x86_64.manylinux2014_x86_64.whl", hash = "sha256:aac2fcabe2d5072c252d0b3087365f431de812d8199705089fb073e4d039d19c"},
-    {file = "duckdb-1.1.0-cp39-cp39-manylinux_2_24_aarch64.manylinux_2_28_aarch64.whl", hash = "sha256:d89eaaa5df8a57e7d2bc1f4c46493bb1fee319a00155f2015810ad2ace6570ae"},
-    {file = "duckdb-1.1.0-cp39-cp39-musllinux_1_2_x86_64.whl", hash = "sha256:d86a6926313913cd2cc7e08816d3e7f72ba340adf2959279b1a80058be6526d9"},
-    {file = "duckdb-1.1.0-cp39-cp39-win_amd64.whl", hash = "sha256:d8333f3e85fa2a0f1c222b752c2bd42ea875235ff88492f7bcbb6867d0f644eb"},
-    {file = "duckdb-1.1.0.tar.gz", hash = "sha256:b4d4c12b1f98732151bd31377753e0da1a20f6423016d2d097d2e31953ec7c23"},
+    {file = "duckdb-1.1.1-cp310-cp310-macosx_12_0_arm64.whl", hash = "sha256:e310610b692d30aa7f1f40d7878b26978a5b191f23fa8fa082bd17092c67c2fd"},
+    {file = "duckdb-1.1.1-cp310-cp310-macosx_12_0_universal2.whl", hash = "sha256:7acc97c3cc995850a4fa59dfa6ce713d7ea187c9696632161aa09d898f001a2b"},
+    {file = "duckdb-1.1.1-cp310-cp310-macosx_12_0_x86_64.whl", hash = "sha256:c0a09d78daea0de7ddf3d6d1113e80ceed8c15537e93f8efaad53024ffbde245"},
+    {file = "duckdb-1.1.1-cp310-cp310-manylinux_2_17_aarch64.manylinux2014_aarch64.whl", hash = "sha256:50c3b1667b0c73cb076b1b1f8fa0fd88fcef5c2bbb2b9acdef79e2eae429c248"},
+    {file = "duckdb-1.1.1-cp310-cp310-manylinux_2_17_x86_64.manylinux2014_x86_64.whl", hash = "sha256:1499a9b159d4675ea46786b7ebdbabd8287c62b6b116ccfd529112318d47184e"},
+    {file = "duckdb-1.1.1-cp310-cp310-manylinux_2_24_aarch64.manylinux_2_28_aarch64.whl", hash = "sha256:876deda2ce97f4a9005a9ac862f0ebee9e5956d51d589a24955802ca91726d49"},
+    {file = "duckdb-1.1.1-cp310-cp310-musllinux_1_2_x86_64.whl", hash = "sha256:40be901b38c709076f699b0c2f42a0c5663a496647eba350530e3a77f46a239b"},
+    {file = "duckdb-1.1.1-cp310-cp310-win_amd64.whl", hash = "sha256:5cb7642c5b21b8165b60029c274fc931c7c29cae3124b9a95ed73d050dd23584"},
+    {file = "duckdb-1.1.1-cp311-cp311-macosx_12_0_arm64.whl", hash = "sha256:959716b65cf1c94fc117ac9c9692eea0bd64ae53bc8ab6538d459087b474dbeb"},
+    {file = "duckdb-1.1.1-cp311-cp311-macosx_12_0_universal2.whl", hash = "sha256:6ff3c52ce0f8d25478155eb01de043ad0a25badbd10e684a2cd74363f1b86cde"},
+    {file = "duckdb-1.1.1-cp311-cp311-macosx_12_0_x86_64.whl", hash = "sha256:430294cf11ce866d3b726cf4530462316e20b773fed3cf2de3cf63eb89650da6"},
+    {file = "duckdb-1.1.1-cp311-cp311-manylinux_2_17_aarch64.manylinux2014_aarch64.whl", hash = "sha256:dc9d48f772fafeea52568a0568cd11314cd79a10214069f3700dbcb31ebdf511"},
+    {file = "duckdb-1.1.1-cp311-cp311-manylinux_2_17_x86_64.manylinux2014_x86_64.whl", hash = "sha256:572095739024d9a5aa2dd8336c289af6a624c203004213e49b7e2469275e940f"},
+    {file = "duckdb-1.1.1-cp311-cp311-manylinux_2_24_aarch64.manylinux_2_28_aarch64.whl", hash = "sha256:660d9baf637b9a15e1ba74bbe02d3b4a20d82e8cbbd7d0712e0d59e3e9d6efea"},
+    {file = "duckdb-1.1.1-cp311-cp311-musllinux_1_2_x86_64.whl", hash = "sha256:b91973605c8a30a38c4381a27895e7768cb3caa6700b2534ab76cc6b72cac390"},
+    {file = "duckdb-1.1.1-cp311-cp311-win_amd64.whl", hash = "sha256:f57c9e070cecf42d379145a75f325ec57fb1d410d6ff6592b5a28c2ff2b5792c"},
+    {file = "duckdb-1.1.1-cp312-cp312-macosx_12_0_arm64.whl", hash = "sha256:926a99b81c50b9a4a43ca26dcb781f934d35e773d22913548396601ab8d44c12"},
+    {file = "duckdb-1.1.1-cp312-cp312-macosx_12_0_universal2.whl", hash = "sha256:55a2632d27b5a965f1d9fc74b03383e80a3f8e3dc9596807dfb02c8db08cfcb7"},
+    {file = "duckdb-1.1.1-cp312-cp312-macosx_12_0_x86_64.whl", hash = "sha256:8d8174fe47caf48d830dc477a45cedc8c970722df09dc1456bddc760ff6ccf68"},
+    {file = "duckdb-1.1.1-cp312-cp312-manylinux_2_17_aarch64.manylinux2014_aarch64.whl", hash = "sha256:9ad84023399002222fa8d5264a8dc2083053027910df728da92cabb07494a489"},
+    {file = "duckdb-1.1.1-cp312-cp312-manylinux_2_17_x86_64.manylinux2014_x86_64.whl", hash = "sha256:1c8adbc8b37444424c72043288f1521c860555a4f151ee4b744e6125f5d05729"},
+    {file = "duckdb-1.1.1-cp312-cp312-manylinux_2_24_aarch64.manylinux_2_28_aarch64.whl", hash = "sha256:550524c1b423eeb7ca0fdf1c2e6d29e723d7ec7cfab3050b9feb55a620ae927f"},
+    {file = "duckdb-1.1.1-cp312-cp312-musllinux_1_2_x86_64.whl", hash = "sha256:4064243e4d3f445975b78773677de0ccbe924f9c7058a7c2cfedb24bba2ba939"},
+    {file = "duckdb-1.1.1-cp312-cp312-win_amd64.whl", hash = "sha256:4f64516dc62dd0fcbb9785c5bc7532a4fca3e6016bbcc92a2b235aa972c631f6"},
+    {file = "duckdb-1.1.1-cp313-cp313-macosx_12_0_arm64.whl", hash = "sha256:4bf75a64c927470b6618496adcfbf0f316ef09d46a44cfe8e38b78e9ff40c8a0"},
+    {file = "duckdb-1.1.1-cp313-cp313-macosx_12_0_universal2.whl", hash = "sha256:5c8cd6fd7107299b9a243836cd8163e4c08d6228f18cbee4ed9f535f53300096"},
+    {file = "duckdb-1.1.1-cp313-cp313-macosx_12_0_x86_64.whl", hash = "sha256:fc81c02b4d73533a438a9bbae19499531d85b752233c905facc4df41bbde043c"},
+    {file = "duckdb-1.1.1-cp313-cp313-manylinux_2_17_aarch64.manylinux2014_aarch64.whl", hash = "sha256:baff4014caf6553b624a296e4db2926602670bd9be6e0fc75f3e970b085631b0"},
+    {file = "duckdb-1.1.1-cp313-cp313-manylinux_2_17_x86_64.manylinux2014_x86_64.whl", hash = "sha256:e21b75a9a60f10b5b5033138c317d929018c92f355fadae5949b310a9179e0a7"},
+    {file = "duckdb-1.1.1-cp313-cp313-manylinux_2_24_aarch64.manylinux_2_28_aarch64.whl", hash = "sha256:8220f039c5ea06dc126232464ab9b77197f80ae53d4611b0a41f73c54f6f3931"},
+    {file = "duckdb-1.1.1-cp313-cp313-musllinux_1_2_x86_64.whl", hash = "sha256:07384414ceae585d4106a7dc154331ae42f45390ed675ec81e3d01f2252a6b01"},
+    {file = "duckdb-1.1.1-cp313-cp313-win_amd64.whl", hash = "sha256:82776b3999e71a962db0bdc3f0258407ef41453f63eb47c33da29b644f8eb530"},
+    {file = "duckdb-1.1.1-cp37-cp37m-manylinux_2_17_aarch64.manylinux2014_aarch64.whl", hash = "sha256:35d4323655be4053fb90d47e85222c93fd56aea0e8ab0ac44bd8f7249ba85697"},
+    {file = "duckdb-1.1.1-cp37-cp37m-manylinux_2_17_x86_64.manylinux2014_x86_64.whl", hash = "sha256:990d0799e0f543a4369413dc6caf7782cbbab49955c08c28ac56d5dab5ccef11"},
+    {file = "duckdb-1.1.1-cp37-cp37m-manylinux_2_24_aarch64.manylinux_2_28_aarch64.whl", hash = "sha256:5ef3ba36b317abe000f502702eaaefdd8c3651a25aa0ad409f9487b286e2fb28"},
+    {file = "duckdb-1.1.1-cp37-cp37m-win_amd64.whl", hash = "sha256:2c6e513a572967cd2bab0f20ce265f8eaf95ea7b554eecf1c233717c38569abc"},
+    {file = "duckdb-1.1.1-cp38-cp38-macosx_12_0_arm64.whl", hash = "sha256:567471cb964a0e54a7874c578e81af7b6ab474676ae6469ae1c33c2353f76fb1"},
+    {file = "duckdb-1.1.1-cp38-cp38-macosx_12_0_universal2.whl", hash = "sha256:a41d8eb4dc538d17660b78f2f4ecd0ba29666a396453bb71d6f4972bf2b3959e"},
+    {file = "duckdb-1.1.1-cp38-cp38-macosx_12_0_x86_64.whl", hash = "sha256:31be0b9bc1909fb60abda7cd30615fe0224d1e451160d79e8e0313d6205417b0"},
+    {file = "duckdb-1.1.1-cp38-cp38-manylinux_2_17_aarch64.manylinux2014_aarch64.whl", hash = "sha256:541fb49da108e080d4f2984d2fdabaee36d65967a33642f8bce03373b29952f0"},
+    {file = "duckdb-1.1.1-cp38-cp38-manylinux_2_17_x86_64.manylinux2014_x86_64.whl", hash = "sha256:e1c54f836dac5eddbe369fa654811e979bb07688638a52d1c006172feb5b75a5"},
+    {file = "duckdb-1.1.1-cp38-cp38-manylinux_2_24_aarch64.manylinux_2_28_aarch64.whl", hash = "sha256:afb97970ee72e554b507c6f2e40b356bdbf8fc1f466e7c4d1797183eb66c0809"},
+    {file = "duckdb-1.1.1-cp38-cp38-win_amd64.whl", hash = "sha256:a2cdcb68247f02017a35a0b617ceb1d36a02a7c0588d7e2ed91c9a4e9f14c3f6"},
+    {file = "duckdb-1.1.1-cp39-cp39-macosx_12_0_arm64.whl", hash = "sha256:36d71969cb98d10dc2391d8755921258d197995cc8c69e6c82fc377c2f71940a"},
+    {file = "duckdb-1.1.1-cp39-cp39-macosx_12_0_universal2.whl", hash = "sha256:3693f464409379a21aff4e35b5f67eb6c96fc402649d9ffddbda4ee9ee9ba9b6"},
+    {file = "duckdb-1.1.1-cp39-cp39-macosx_12_0_x86_64.whl", hash = "sha256:06ca7f4ca785cc86e9f9aa23d16b67b82dc454b14c396b2e0ff4c09698c7838e"},
+    {file = "duckdb-1.1.1-cp39-cp39-manylinux_2_17_aarch64.manylinux2014_aarch64.whl", hash = "sha256:5ed92f3229bf70897a742e7648f648aa8b0c81a7489072aec5515c5635f3303c"},
+    {file = "duckdb-1.1.1-cp39-cp39-manylinux_2_17_x86_64.manylinux2014_x86_64.whl", hash = "sha256:a80ebf52c03f81265b67720abc06a5c7770d08df82b30cabbe266012bd526229"},
+    {file = "duckdb-1.1.1-cp39-cp39-manylinux_2_24_aarch64.manylinux_2_28_aarch64.whl", hash = "sha256:402a42b992227ebb371a48681ce71b6d1c0661385454b269e6aa379f77a8a83a"},
+    {file = "duckdb-1.1.1-cp39-cp39-musllinux_1_2_x86_64.whl", hash = "sha256:a182d3cbf2e352aaddf392887331bbac460c473cbd55c65d6b6121ef7b43f174"},
+    {file = "duckdb-1.1.1-cp39-cp39-win_amd64.whl", hash = "sha256:fafc7d1ec4401787597a5f983d4ef8a9b0638f31e1674a458c57383911166f27"},
+    {file = "duckdb-1.1.1.tar.gz", hash = "sha256:74fb07c1334a73e0ead1b0a03646d349921dac655762d916c8e45194c8218d30"},
 ]
 
 [[package]]
 name = "duckduckgo-search"
-version = "6.2.12"
+version = "6.2.13"
 description = "Search for words, documents, images, news, maps and text translation using the DuckDuckGo.com search engine."
 optional = false
 python-versions = ">=3.8"
 files = [
-    {file = "duckduckgo_search-6.2.12-py3-none-any.whl", hash = "sha256:0d379c1f845b632a41553efb13d571788f19ad289229e641a27b5710d92097a6"},
-    {file = "duckduckgo_search-6.2.12.tar.gz", hash = "sha256:04f9f1459763668d268344c7a32d943173d0e060dad53a5c2df4b4d3ca9a74cf"},
+    {file = "duckduckgo_search-6.2.13-py3-none-any.whl", hash = "sha256:a6618fb2744fa1d081b1bf2e47ef8051de993276a15c20f4e879a150726472de"},
+    {file = "duckduckgo_search-6.2.13.tar.gz", hash = "sha256:f89a9782f0f47d18c01a761c83453d0aef7a4335d1b6466fc47709652d5ca791"},
 ]
 
 [package.dependencies]
 click = ">=8.1.7"
-primp = ">=0.6.2"
+primp = ">=0.6.3"
 
 [package.extras]
 dev = ["mypy (>=1.11.1)", "pytest (>=8.3.1)", "pytest-asyncio (>=0.23.8)", "ruff (>=0.6.1)"]
 lxml = ["lxml (>=5.2.2)"]
 
+[[package]]
+name = "durationpy"
+version = "0.7"
+description = "Module for converting between datetime.timedelta and Go's Duration strings."
+optional = false
+python-versions = "*"
+files = [
+    {file = "durationpy-0.7.tar.gz", hash = "sha256:8447c43df4f1a0b434e70c15a38d77f5c9bd17284bfc1ff1d430f233d5083732"},
+]
+
 [[package]]
 name = "elastic-transport"
 version = "8.15.0"
@@ -2356,18 +2372,15 @@ vectorstore-mmr = ["numpy (>=1)", "simsimd (>=3)"]
 
 [[package]]
 name = "emoji"
-version = "2.12.1"
+version = "2.13.2"
 description = "Emoji for Python"
 optional = false
 python-versions = ">=3.7"
 files = [
-    {file = "emoji-2.12.1-py3-none-any.whl", hash = "sha256:a00d62173bdadc2510967a381810101624a2f0986145b8da0cffa42e29430235"},
-    {file = "emoji-2.12.1.tar.gz", hash = "sha256:4aa0488817691aa58d83764b6c209f8a27c0b3ab3f89d1b8dceca1a62e4973eb"},
+    {file = "emoji-2.13.2-py3-none-any.whl", hash = "sha256:ef6f2ee63b245e934c763b1a9a0637713955aa3d9e322432e036bb60559de4d6"},
+    {file = "emoji-2.13.2.tar.gz", hash = "sha256:f95d10d96c5f21299ed2c4b32511611ba890b8c07f5f2bf5b04d5d3eee91fd19"},
 ]
 
-[package.dependencies]
-typing-extensions = ">=4.7.0"
-
 [package.extras]
 dev = ["coverage", "pytest (>=7.4.4)"]
 
@@ -2432,13 +2445,13 @@ test = ["pytest (>=6)"]
 
 [[package]]
 name = "fastapi"
-version = "0.114.2"
+version = "0.115.0"
 description = "FastAPI framework, high performance, easy to learn, fast to code, ready for production"
 optional = false
 python-versions = ">=3.8"
 files = [
-    {file = "fastapi-0.114.2-py3-none-any.whl", hash = "sha256:44474a22913057b1acb973ab90f4b671ba5200482e7622816d79105dcece1ac5"},
-    {file = "fastapi-0.114.2.tar.gz", hash = "sha256:0adb148b62edb09e8c6eeefa3ea934e8f276dabc038c5a82989ea6346050c3da"},
+    {file = "fastapi-0.115.0-py3-none-any.whl", hash = "sha256:17ea427674467486e997206a5ab25760f6b09e069f099b96f5b55a32fb6f1631"},
+    {file = "fastapi-0.115.0.tar.gz", hash = "sha256:f93b4ca3529a8ebc6fc3fcf710e5efa8de3df9b41570958abf1d97d843138004"},
 ]
 
 [package.dependencies]
@@ -2527,18 +2540,18 @@ sgmllib3k = "*"
 
 [[package]]
 name = "filelock"
-version = "3.16.0"
+version = "3.16.1"
 description = "A platform independent file lock."
 optional = false
 python-versions = ">=3.8"
 files = [
-    {file = "filelock-3.16.0-py3-none-any.whl", hash = "sha256:f6ed4c963184f4c84dd5557ce8fece759a3724b37b80c6c4f20a2f63a4dc6609"},
-    {file = "filelock-3.16.0.tar.gz", hash = "sha256:81de9eb8453c769b63369f87f11131a7ab04e367f8d97ad39dc230daa07e3bec"},
+    {file = "filelock-3.16.1-py3-none-any.whl", hash = "sha256:2082e5703d51fbf98ea75855d9d5527e33d8ff23099bec374a134febee6946b0"},
+    {file = "filelock-3.16.1.tar.gz", hash = "sha256:c249fbfcd5db47e5e2d6d62198e565475ee65e4831e2561c8e313fa7eb961435"},
 ]
 
 [package.extras]
-docs = ["furo (>=2024.8.6)", "sphinx (>=8.0.2)", "sphinx-autodoc-typehints (>=2.4)"]
-testing = ["covdefaults (>=2.3)", "coverage (>=7.6.1)", "diff-cover (>=9.1.1)", "pytest (>=8.3.2)", "pytest-asyncio (>=0.24)", "pytest-cov (>=5)", "pytest-mock (>=3.14)", "pytest-timeout (>=2.3.1)", "virtualenv (>=20.26.3)"]
+docs = ["furo (>=2024.8.6)", "sphinx (>=8.0.2)", "sphinx-autodoc-typehints (>=2.4.1)"]
+testing = ["covdefaults (>=2.3)", "coverage (>=7.6.1)", "diff-cover (>=9.2)", "pytest (>=8.3.3)", "pytest-asyncio (>=0.24)", "pytest-cov (>=5)", "pytest-mock (>=3.14)", "pytest-timeout (>=2.3.1)", "virtualenv (>=20.26.4)"]
 typing = ["typing-extensions (>=4.12.2)"]
 
 [[package]]
@@ -2701,53 +2714,59 @@ files = [
 
 [[package]]
 name = "fonttools"
-version = "4.53.1"
+version = "4.54.1"
 description = "Tools to manipulate font files"
 optional = false
 python-versions = ">=3.8"
 files = [
-    {file = "fonttools-4.53.1-cp310-cp310-macosx_10_9_universal2.whl", hash = "sha256:0679a30b59d74b6242909945429dbddb08496935b82f91ea9bf6ad240ec23397"},
-    {file = "fonttools-4.53.1-cp310-cp310-macosx_11_0_arm64.whl", hash = "sha256:e8bf06b94694251861ba7fdeea15c8ec0967f84c3d4143ae9daf42bbc7717fe3"},
-    {file = "fonttools-4.53.1-cp310-cp310-manylinux_2_17_aarch64.manylinux2014_aarch64.whl", hash = "sha256:b96cd370a61f4d083c9c0053bf634279b094308d52fdc2dd9a22d8372fdd590d"},
-    {file = "fonttools-4.53.1-cp310-cp310-manylinux_2_17_x86_64.manylinux2014_x86_64.whl", hash = "sha256:a1c7c5aa18dd3b17995898b4a9b5929d69ef6ae2af5b96d585ff4005033d82f0"},
-    {file = "fonttools-4.53.1-cp310-cp310-musllinux_1_2_aarch64.whl", hash = "sha256:e013aae589c1c12505da64a7d8d023e584987e51e62006e1bb30d72f26522c41"},
-    {file = "fonttools-4.53.1-cp310-cp310-musllinux_1_2_x86_64.whl", hash = "sha256:9efd176f874cb6402e607e4cc9b4a9cd584d82fc34a4b0c811970b32ba62501f"},
-    {file = "fonttools-4.53.1-cp310-cp310-win32.whl", hash = "sha256:c8696544c964500aa9439efb6761947393b70b17ef4e82d73277413f291260a4"},
-    {file = "fonttools-4.53.1-cp310-cp310-win_amd64.whl", hash = "sha256:8959a59de5af6d2bec27489e98ef25a397cfa1774b375d5787509c06659b3671"},
-    {file = "fonttools-4.53.1-cp311-cp311-macosx_10_9_universal2.whl", hash = "sha256:da33440b1413bad53a8674393c5d29ce64d8c1a15ef8a77c642ffd900d07bfe1"},
-    {file = "fonttools-4.53.1-cp311-cp311-macosx_11_0_arm64.whl", hash = "sha256:5ff7e5e9bad94e3a70c5cd2fa27f20b9bb9385e10cddab567b85ce5d306ea923"},
-    {file = "fonttools-4.53.1-cp311-cp311-manylinux_2_17_aarch64.manylinux2014_aarch64.whl", hash = "sha256:c6e7170d675d12eac12ad1a981d90f118c06cf680b42a2d74c6c931e54b50719"},
-    {file = "fonttools-4.53.1-cp311-cp311-manylinux_2_17_x86_64.manylinux2014_x86_64.whl", hash = "sha256:bee32ea8765e859670c4447b0817514ca79054463b6b79784b08a8df3a4d78e3"},
-    {file = "fonttools-4.53.1-cp311-cp311-musllinux_1_2_aarch64.whl", hash = "sha256:6e08f572625a1ee682115223eabebc4c6a2035a6917eac6f60350aba297ccadb"},
-    {file = "fonttools-4.53.1-cp311-cp311-musllinux_1_2_x86_64.whl", hash = "sha256:b21952c092ffd827504de7e66b62aba26fdb5f9d1e435c52477e6486e9d128b2"},
-    {file = "fonttools-4.53.1-cp311-cp311-win32.whl", hash = "sha256:9dfdae43b7996af46ff9da520998a32b105c7f098aeea06b2226b30e74fbba88"},
-    {file = "fonttools-4.53.1-cp311-cp311-win_amd64.whl", hash = "sha256:d4d0096cb1ac7a77b3b41cd78c9b6bc4a400550e21dc7a92f2b5ab53ed74eb02"},
-    {file = "fonttools-4.53.1-cp312-cp312-macosx_10_9_universal2.whl", hash = "sha256:d92d3c2a1b39631a6131c2fa25b5406855f97969b068e7e08413325bc0afba58"},
-    {file = "fonttools-4.53.1-cp312-cp312-macosx_11_0_arm64.whl", hash = "sha256:3b3c8ebafbee8d9002bd8f1195d09ed2bd9ff134ddec37ee8f6a6375e6a4f0e8"},
-    {file = "fonttools-4.53.1-cp312-cp312-manylinux_2_17_aarch64.manylinux2014_aarch64.whl", hash = "sha256:32f029c095ad66c425b0ee85553d0dc326d45d7059dbc227330fc29b43e8ba60"},
-    {file = "fonttools-4.53.1-cp312-cp312-manylinux_2_5_x86_64.manylinux1_x86_64.manylinux_2_17_x86_64.manylinux2014_x86_64.whl", hash = "sha256:10f5e6c3510b79ea27bb1ebfcc67048cde9ec67afa87c7dd7efa5c700491ac7f"},
-    {file = "fonttools-4.53.1-cp312-cp312-musllinux_1_2_aarch64.whl", hash = "sha256:f677ce218976496a587ab17140da141557beb91d2a5c1a14212c994093f2eae2"},
-    {file = "fonttools-4.53.1-cp312-cp312-musllinux_1_2_x86_64.whl", hash = "sha256:9e6ceba2a01b448e36754983d376064730690401da1dd104ddb543519470a15f"},
-    {file = "fonttools-4.53.1-cp312-cp312-win32.whl", hash = "sha256:791b31ebbc05197d7aa096bbc7bd76d591f05905d2fd908bf103af4488e60670"},
-    {file = "fonttools-4.53.1-cp312-cp312-win_amd64.whl", hash = "sha256:6ed170b5e17da0264b9f6fae86073be3db15fa1bd74061c8331022bca6d09bab"},
-    {file = "fonttools-4.53.1-cp38-cp38-macosx_10_9_universal2.whl", hash = "sha256:c818c058404eb2bba05e728d38049438afd649e3c409796723dfc17cd3f08749"},
-    {file = "fonttools-4.53.1-cp38-cp38-macosx_11_0_arm64.whl", hash = "sha256:651390c3b26b0c7d1f4407cad281ee7a5a85a31a110cbac5269de72a51551ba2"},
-    {file = "fonttools-4.53.1-cp38-cp38-manylinux_2_17_aarch64.manylinux2014_aarch64.whl", hash = "sha256:e54f1bba2f655924c1138bbc7fa91abd61f45c68bd65ab5ed985942712864bbb"},
-    {file = "fonttools-4.53.1-cp38-cp38-manylinux_2_17_x86_64.manylinux2014_x86_64.whl", hash = "sha256:c9cd19cf4fe0595ebdd1d4915882b9440c3a6d30b008f3cc7587c1da7b95be5f"},
-    {file = "fonttools-4.53.1-cp38-cp38-musllinux_1_2_aarch64.whl", hash = "sha256:2af40ae9cdcb204fc1d8f26b190aa16534fcd4f0df756268df674a270eab575d"},
-    {file = "fonttools-4.53.1-cp38-cp38-musllinux_1_2_x86_64.whl", hash = "sha256:35250099b0cfb32d799fb5d6c651220a642fe2e3c7d2560490e6f1d3f9ae9169"},
-    {file = "fonttools-4.53.1-cp38-cp38-win32.whl", hash = "sha256:f08df60fbd8d289152079a65da4e66a447efc1d5d5a4d3f299cdd39e3b2e4a7d"},
-    {file = "fonttools-4.53.1-cp38-cp38-win_amd64.whl", hash = "sha256:7b6b35e52ddc8fb0db562133894e6ef5b4e54e1283dff606fda3eed938c36fc8"},
-    {file = "fonttools-4.53.1-cp39-cp39-macosx_10_9_universal2.whl", hash = "sha256:75a157d8d26c06e64ace9df037ee93a4938a4606a38cb7ffaf6635e60e253b7a"},
-    {file = "fonttools-4.53.1-cp39-cp39-macosx_11_0_arm64.whl", hash = "sha256:4824c198f714ab5559c5be10fd1adf876712aa7989882a4ec887bf1ef3e00e31"},
-    {file = "fonttools-4.53.1-cp39-cp39-manylinux_2_17_aarch64.manylinux2014_aarch64.whl", hash = "sha256:becc5d7cb89c7b7afa8321b6bb3dbee0eec2b57855c90b3e9bf5fb816671fa7c"},
-    {file = "fonttools-4.53.1-cp39-cp39-manylinux_2_17_x86_64.manylinux2014_x86_64.whl", hash = "sha256:84ec3fb43befb54be490147b4a922b5314e16372a643004f182babee9f9c3407"},
-    {file = "fonttools-4.53.1-cp39-cp39-musllinux_1_2_aarch64.whl", hash = "sha256:73379d3ffdeecb376640cd8ed03e9d2d0e568c9d1a4e9b16504a834ebadc2dfb"},
-    {file = "fonttools-4.53.1-cp39-cp39-musllinux_1_2_x86_64.whl", hash = "sha256:02569e9a810f9d11f4ae82c391ebc6fb5730d95a0657d24d754ed7763fb2d122"},
-    {file = "fonttools-4.53.1-cp39-cp39-win32.whl", hash = "sha256:aae7bd54187e8bf7fd69f8ab87b2885253d3575163ad4d669a262fe97f0136cb"},
-    {file = "fonttools-4.53.1-cp39-cp39-win_amd64.whl", hash = "sha256:e5b708073ea3d684235648786f5f6153a48dc8762cdfe5563c57e80787c29fbb"},
-    {file = "fonttools-4.53.1-py3-none-any.whl", hash = "sha256:f1f8758a2ad110bd6432203a344269f445a2907dc24ef6bccfd0ac4e14e0d71d"},
-    {file = "fonttools-4.53.1.tar.gz", hash = "sha256:e128778a8e9bc11159ce5447f76766cefbd876f44bd79aff030287254e4752c4"},
+    {file = "fonttools-4.54.1-cp310-cp310-macosx_10_9_universal2.whl", hash = "sha256:7ed7ee041ff7b34cc62f07545e55e1468808691dddfd315d51dd82a6b37ddef2"},
+    {file = "fonttools-4.54.1-cp310-cp310-macosx_11_0_arm64.whl", hash = "sha256:41bb0b250c8132b2fcac148e2e9198e62ff06f3cc472065dff839327945c5882"},
+    {file = "fonttools-4.54.1-cp310-cp310-manylinux_2_17_aarch64.manylinux2014_aarch64.whl", hash = "sha256:7965af9b67dd546e52afcf2e38641b5be956d68c425bef2158e95af11d229f10"},
+    {file = "fonttools-4.54.1-cp310-cp310-manylinux_2_17_x86_64.manylinux2014_x86_64.whl", hash = "sha256:278913a168f90d53378c20c23b80f4e599dca62fbffae4cc620c8eed476b723e"},
+    {file = "fonttools-4.54.1-cp310-cp310-musllinux_1_2_aarch64.whl", hash = "sha256:0e88e3018ac809b9662615072dcd6b84dca4c2d991c6d66e1970a112503bba7e"},
+    {file = "fonttools-4.54.1-cp310-cp310-musllinux_1_2_x86_64.whl", hash = "sha256:4aa4817f0031206e637d1e685251ac61be64d1adef111060df84fdcbc6ab6c44"},
+    {file = "fonttools-4.54.1-cp310-cp310-win32.whl", hash = "sha256:7e3b7d44e18c085fd8c16dcc6f1ad6c61b71ff463636fcb13df7b1b818bd0c02"},
+    {file = "fonttools-4.54.1-cp310-cp310-win_amd64.whl", hash = "sha256:dd9cc95b8d6e27d01e1e1f1fae8559ef3c02c76317da650a19047f249acd519d"},
+    {file = "fonttools-4.54.1-cp311-cp311-macosx_10_9_universal2.whl", hash = "sha256:5419771b64248484299fa77689d4f3aeed643ea6630b2ea750eeab219588ba20"},
+    {file = "fonttools-4.54.1-cp311-cp311-macosx_11_0_arm64.whl", hash = "sha256:301540e89cf4ce89d462eb23a89464fef50915255ece765d10eee8b2bf9d75b2"},
+    {file = "fonttools-4.54.1-cp311-cp311-manylinux_2_17_aarch64.manylinux2014_aarch64.whl", hash = "sha256:76ae5091547e74e7efecc3cbf8e75200bc92daaeb88e5433c5e3e95ea8ce5aa7"},
+    {file = "fonttools-4.54.1-cp311-cp311-manylinux_2_17_x86_64.manylinux2014_x86_64.whl", hash = "sha256:82834962b3d7c5ca98cb56001c33cf20eb110ecf442725dc5fdf36d16ed1ab07"},
+    {file = "fonttools-4.54.1-cp311-cp311-musllinux_1_2_aarch64.whl", hash = "sha256:d26732ae002cc3d2ecab04897bb02ae3f11f06dd7575d1df46acd2f7c012a8d8"},
+    {file = "fonttools-4.54.1-cp311-cp311-musllinux_1_2_x86_64.whl", hash = "sha256:58974b4987b2a71ee08ade1e7f47f410c367cdfc5a94fabd599c88165f56213a"},
+    {file = "fonttools-4.54.1-cp311-cp311-win32.whl", hash = "sha256:ab774fa225238986218a463f3fe151e04d8c25d7de09df7f0f5fce27b1243dbc"},
+    {file = "fonttools-4.54.1-cp311-cp311-win_amd64.whl", hash = "sha256:07e005dc454eee1cc60105d6a29593459a06321c21897f769a281ff2d08939f6"},
+    {file = "fonttools-4.54.1-cp312-cp312-macosx_10_13_universal2.whl", hash = "sha256:54471032f7cb5fca694b5f1a0aaeba4af6e10ae989df408e0216f7fd6cdc405d"},
+    {file = "fonttools-4.54.1-cp312-cp312-macosx_11_0_arm64.whl", hash = "sha256:8fa92cb248e573daab8d032919623cc309c005086d743afb014c836636166f08"},
+    {file = "fonttools-4.54.1-cp312-cp312-manylinux_2_17_aarch64.manylinux2014_aarch64.whl", hash = "sha256:0a911591200114969befa7f2cb74ac148bce5a91df5645443371aba6d222e263"},
+    {file = "fonttools-4.54.1-cp312-cp312-manylinux_2_5_x86_64.manylinux1_x86_64.manylinux_2_17_x86_64.manylinux2014_x86_64.whl", hash = "sha256:93d458c8a6a354dc8b48fc78d66d2a8a90b941f7fec30e94c7ad9982b1fa6bab"},
+    {file = "fonttools-4.54.1-cp312-cp312-musllinux_1_2_aarch64.whl", hash = "sha256:5eb2474a7c5be8a5331146758debb2669bf5635c021aee00fd7c353558fc659d"},
+    {file = "fonttools-4.54.1-cp312-cp312-musllinux_1_2_x86_64.whl", hash = "sha256:c9c563351ddc230725c4bdf7d9e1e92cbe6ae8553942bd1fb2b2ff0884e8b714"},
+    {file = "fonttools-4.54.1-cp312-cp312-win32.whl", hash = "sha256:fdb062893fd6d47b527d39346e0c5578b7957dcea6d6a3b6794569370013d9ac"},
+    {file = "fonttools-4.54.1-cp312-cp312-win_amd64.whl", hash = "sha256:e4564cf40cebcb53f3dc825e85910bf54835e8a8b6880d59e5159f0f325e637e"},
+    {file = "fonttools-4.54.1-cp313-cp313-macosx_10_13_universal2.whl", hash = "sha256:6e37561751b017cf5c40fce0d90fd9e8274716de327ec4ffb0df957160be3bff"},
+    {file = "fonttools-4.54.1-cp313-cp313-macosx_11_0_arm64.whl", hash = "sha256:357cacb988a18aace66e5e55fe1247f2ee706e01debc4b1a20d77400354cddeb"},
+    {file = "fonttools-4.54.1-cp313-cp313-manylinux_2_5_x86_64.manylinux1_x86_64.manylinux_2_17_x86_64.manylinux2014_x86_64.whl", hash = "sha256:f8e953cc0bddc2beaf3a3c3b5dd9ab7554677da72dfaf46951e193c9653e515a"},
+    {file = "fonttools-4.54.1-cp313-cp313-musllinux_1_2_x86_64.whl", hash = "sha256:58d29b9a294573d8319f16f2f79e42428ba9b6480442fa1836e4eb89c4d9d61c"},
+    {file = "fonttools-4.54.1-cp313-cp313-win32.whl", hash = "sha256:9ef1b167e22709b46bf8168368b7b5d3efeaaa746c6d39661c1b4405b6352e58"},
+    {file = "fonttools-4.54.1-cp313-cp313-win_amd64.whl", hash = "sha256:262705b1663f18c04250bd1242b0515d3bbae177bee7752be67c979b7d47f43d"},
+    {file = "fonttools-4.54.1-cp38-cp38-macosx_10_9_universal2.whl", hash = "sha256:ed2f80ca07025551636c555dec2b755dd005e2ea8fbeb99fc5cdff319b70b23b"},
+    {file = "fonttools-4.54.1-cp38-cp38-macosx_11_0_arm64.whl", hash = "sha256:9dc080e5a1c3b2656caff2ac2633d009b3a9ff7b5e93d0452f40cd76d3da3b3c"},
+    {file = "fonttools-4.54.1-cp38-cp38-manylinux_2_17_aarch64.manylinux2014_aarch64.whl", hash = "sha256:1d152d1be65652fc65e695e5619e0aa0982295a95a9b29b52b85775243c06556"},
+    {file = "fonttools-4.54.1-cp38-cp38-manylinux_2_17_x86_64.manylinux2014_x86_64.whl", hash = "sha256:8583e563df41fdecef31b793b4dd3af8a9caa03397be648945ad32717a92885b"},
+    {file = "fonttools-4.54.1-cp38-cp38-musllinux_1_2_aarch64.whl", hash = "sha256:0d1d353ef198c422515a3e974a1e8d5b304cd54a4c2eebcae708e37cd9eeffb1"},
+    {file = "fonttools-4.54.1-cp38-cp38-musllinux_1_2_x86_64.whl", hash = "sha256:fda582236fee135d4daeca056c8c88ec5f6f6d88a004a79b84a02547c8f57386"},
+    {file = "fonttools-4.54.1-cp38-cp38-win32.whl", hash = "sha256:e7d82b9e56716ed32574ee106cabca80992e6bbdcf25a88d97d21f73a0aae664"},
+    {file = "fonttools-4.54.1-cp38-cp38-win_amd64.whl", hash = "sha256:ada215fd079e23e060157aab12eba0d66704316547f334eee9ff26f8c0d7b8ab"},
+    {file = "fonttools-4.54.1-cp39-cp39-macosx_10_9_universal2.whl", hash = "sha256:f5b8a096e649768c2f4233f947cf9737f8dbf8728b90e2771e2497c6e3d21d13"},
+    {file = "fonttools-4.54.1-cp39-cp39-macosx_11_0_arm64.whl", hash = "sha256:4e10d2e0a12e18f4e2dd031e1bf7c3d7017be5c8dbe524d07706179f355c5dac"},
+    {file = "fonttools-4.54.1-cp39-cp39-manylinux_2_17_aarch64.manylinux2014_aarch64.whl", hash = "sha256:31c32d7d4b0958600eac75eaf524b7b7cb68d3a8c196635252b7a2c30d80e986"},
+    {file = "fonttools-4.54.1-cp39-cp39-manylinux_2_17_x86_64.manylinux2014_x86_64.whl", hash = "sha256:c39287f5c8f4a0c5a55daf9eaf9ccd223ea59eed3f6d467133cc727d7b943a55"},
+    {file = "fonttools-4.54.1-cp39-cp39-musllinux_1_2_aarch64.whl", hash = "sha256:a7a310c6e0471602fe3bf8efaf193d396ea561486aeaa7adc1f132e02d30c4b9"},
+    {file = "fonttools-4.54.1-cp39-cp39-musllinux_1_2_x86_64.whl", hash = "sha256:d3b659d1029946f4ff9b6183984578041b520ce0f8fb7078bb37ec7445806b33"},
+    {file = "fonttools-4.54.1-cp39-cp39-win32.whl", hash = "sha256:e96bc94c8cda58f577277d4a71f51c8e2129b8b36fd05adece6320dd3d57de8a"},
+    {file = "fonttools-4.54.1-cp39-cp39-win_amd64.whl", hash = "sha256:e8a4b261c1ef91e7188a30571be6ad98d1c6d9fa2427244c545e2fa0a2494dd7"},
+    {file = "fonttools-4.54.1-py3-none-any.whl", hash = "sha256:37cddd62d83dc4f72f7c3f3c2bcf2697e89a30efb152079896544a93907733bd"},
+    {file = "fonttools-4.54.1.tar.gz", hash = "sha256:957f669d4922f92c171ba01bef7f29410668db09f6c02111e22b2bce446f3285"},
 ]
 
 [package.extras]
@@ -3211,30 +3230,30 @@ xai = ["tensorflow (>=2.3.0,<3.0.0dev)"]
 
 [[package]]
 name = "google-cloud-bigquery"
-version = "3.25.0"
+version = "3.26.0"
 description = "Google BigQuery API client library"
 optional = false
 python-versions = ">=3.7"
 files = [
-    {file = "google-cloud-bigquery-3.25.0.tar.gz", hash = "sha256:5b2aff3205a854481117436836ae1403f11f2594e6810a98886afd57eda28509"},
-    {file = "google_cloud_bigquery-3.25.0-py2.py3-none-any.whl", hash = "sha256:7f0c371bc74d2a7fb74dacbc00ac0f90c8c2bec2289b51dd6685a275873b1ce9"},
+    {file = "google_cloud_bigquery-3.26.0-py2.py3-none-any.whl", hash = "sha256:e0e9ad28afa67a18696e624cbccab284bf2c0a3f6eeb9eeb0426c69b943793a8"},
+    {file = "google_cloud_bigquery-3.26.0.tar.gz", hash = "sha256:edbdc788beea659e04c0af7fe4dcd6d9155344b98951a0d5055bd2f15da4ba23"},
 ]
 
 [package.dependencies]
-google-api-core = {version = ">=1.34.1,<2.0.dev0 || >=2.11.dev0,<3.0.0dev", extras = ["grpc"]}
+google-api-core = {version = ">=2.11.1,<3.0.0dev", extras = ["grpc"]}
 google-auth = ">=2.14.1,<3.0.0dev"
-google-cloud-core = ">=1.6.0,<3.0.0dev"
-google-resumable-media = ">=0.6.0,<3.0dev"
+google-cloud-core = ">=2.4.1,<3.0.0dev"
+google-resumable-media = ">=2.0.0,<3.0dev"
 packaging = ">=20.0.0"
-python-dateutil = ">=2.7.2,<3.0dev"
+python-dateutil = ">=2.7.3,<3.0dev"
 requests = ">=2.21.0,<3.0.0dev"
 
 [package.extras]
-all = ["Shapely (>=1.8.4,<3.0.0dev)", "db-dtypes (>=0.3.0,<2.0.0dev)", "geopandas (>=0.9.0,<1.0dev)", "google-cloud-bigquery-storage (>=2.6.0,<3.0.0dev)", "grpcio (>=1.47.0,<2.0dev)", "grpcio (>=1.49.1,<2.0dev)", "importlib-metadata (>=1.0.0)", "ipykernel (>=6.0.0)", "ipython (>=7.23.1,!=8.1.0)", "ipywidgets (>=7.7.0)", "opentelemetry-api (>=1.1.0)", "opentelemetry-instrumentation (>=0.20b0)", "opentelemetry-sdk (>=1.1.0)", "pandas (>=1.1.0)", "proto-plus (>=1.15.0,<2.0.0dev)", "protobuf (>=3.19.5,!=3.20.0,!=3.20.1,!=4.21.0,!=4.21.1,!=4.21.2,!=4.21.3,!=4.21.4,!=4.21.5,<5.0.0dev)", "pyarrow (>=3.0.0)", "tqdm (>=4.7.4,<5.0.0dev)"]
-bigquery-v2 = ["proto-plus (>=1.15.0,<2.0.0dev)", "protobuf (>=3.19.5,!=3.20.0,!=3.20.1,!=4.21.0,!=4.21.1,!=4.21.2,!=4.21.3,!=4.21.4,!=4.21.5,<5.0.0dev)"]
+all = ["Shapely (>=1.8.4,<3.0.0dev)", "bigquery-magics (>=0.1.0)", "db-dtypes (>=0.3.0,<2.0.0dev)", "geopandas (>=0.9.0,<1.0dev)", "google-cloud-bigquery-storage (>=2.6.0,<3.0.0dev)", "grpcio (>=1.47.0,<2.0dev)", "grpcio (>=1.49.1,<2.0dev)", "importlib-metadata (>=1.0.0)", "ipykernel (>=6.0.0)", "ipywidgets (>=7.7.0)", "opentelemetry-api (>=1.1.0)", "opentelemetry-instrumentation (>=0.20b0)", "opentelemetry-sdk (>=1.1.0)", "pandas (>=1.1.0)", "proto-plus (>=1.22.3,<2.0.0dev)", "protobuf (>=3.20.2,!=4.21.0,!=4.21.1,!=4.21.2,!=4.21.3,!=4.21.4,!=4.21.5,<6.0.0dev)", "pyarrow (>=3.0.0)", "tqdm (>=4.7.4,<5.0.0dev)"]
+bigquery-v2 = ["proto-plus (>=1.22.3,<2.0.0dev)", "protobuf (>=3.20.2,!=4.21.0,!=4.21.1,!=4.21.2,!=4.21.3,!=4.21.4,!=4.21.5,<6.0.0dev)"]
 bqstorage = ["google-cloud-bigquery-storage (>=2.6.0,<3.0.0dev)", "grpcio (>=1.47.0,<2.0dev)", "grpcio (>=1.49.1,<2.0dev)", "pyarrow (>=3.0.0)"]
 geopandas = ["Shapely (>=1.8.4,<3.0.0dev)", "geopandas (>=0.9.0,<1.0dev)"]
-ipython = ["ipykernel (>=6.0.0)", "ipython (>=7.23.1,!=8.1.0)"]
+ipython = ["bigquery-magics (>=0.1.0)"]
 ipywidgets = ["ipykernel (>=6.0.0)", "ipywidgets (>=7.7.0)"]
 opentelemetry = ["opentelemetry-api (>=1.1.0)", "opentelemetry-instrumentation (>=0.20b0)", "opentelemetry-sdk (>=1.1.0)"]
 pandas = ["db-dtypes (>=0.3.0,<2.0.0dev)", "importlib-metadata (>=1.0.0)", "pandas (>=1.1.0)", "pyarrow (>=3.0.0)"]
@@ -3413,77 +3432,84 @@ grpc = ["grpcio (>=1.44.0,<2.0.0.dev0)"]
 
 [[package]]
 name = "greenlet"
-version = "3.1.0"
+version = "3.1.1"
 description = "Lightweight in-process concurrent programming"
 optional = false
 python-versions = ">=3.7"
 files = [
-    {file = "greenlet-3.1.0-cp310-cp310-macosx_10_9_universal2.whl", hash = "sha256:a814dc3100e8a046ff48faeaa909e80cdb358411a3d6dd5293158425c684eda8"},
-    {file = "greenlet-3.1.0-cp310-cp310-manylinux_2_17_aarch64.manylinux2014_aarch64.whl", hash = "sha256:a771dc64fa44ebe58d65768d869fcfb9060169d203446c1d446e844b62bdfdca"},
-    {file = "greenlet-3.1.0-cp310-cp310-manylinux_2_17_ppc64le.manylinux2014_ppc64le.whl", hash = "sha256:0e49a65d25d7350cca2da15aac31b6f67a43d867448babf997fe83c7505f57bc"},
-    {file = "greenlet-3.1.0-cp310-cp310-manylinux_2_17_s390x.manylinux2014_s390x.whl", hash = "sha256:2cd8518eade968bc52262d8c46727cfc0826ff4d552cf0430b8d65aaf50bb91d"},
-    {file = "greenlet-3.1.0-cp310-cp310-manylinux_2_17_x86_64.manylinux2014_x86_64.whl", hash = "sha256:76dc19e660baea5c38e949455c1181bc018893f25372d10ffe24b3ed7341fb25"},
-    {file = "greenlet-3.1.0-cp310-cp310-manylinux_2_24_x86_64.manylinux_2_28_x86_64.whl", hash = "sha256:c0a5b1c22c82831f56f2f7ad9bbe4948879762fe0d59833a4a71f16e5fa0f682"},
-    {file = "greenlet-3.1.0-cp310-cp310-musllinux_1_1_aarch64.whl", hash = "sha256:2651dfb006f391bcb240635079a68a261b227a10a08af6349cba834a2141efa1"},
-    {file = "greenlet-3.1.0-cp310-cp310-musllinux_1_1_x86_64.whl", hash = "sha256:3e7e6ef1737a819819b1163116ad4b48d06cfdd40352d813bb14436024fcda99"},
-    {file = "greenlet-3.1.0-cp310-cp310-win_amd64.whl", hash = "sha256:ffb08f2a1e59d38c7b8b9ac8083c9c8b9875f0955b1e9b9b9a965607a51f8e54"},
-    {file = "greenlet-3.1.0-cp311-cp311-macosx_10_9_universal2.whl", hash = "sha256:9730929375021ec90f6447bff4f7f5508faef1c02f399a1953870cdb78e0c345"},
-    {file = "greenlet-3.1.0-cp311-cp311-manylinux_2_17_aarch64.manylinux2014_aarch64.whl", hash = "sha256:713d450cf8e61854de9420fb7eea8ad228df4e27e7d4ed465de98c955d2b3fa6"},
-    {file = "greenlet-3.1.0-cp311-cp311-manylinux_2_17_ppc64le.manylinux2014_ppc64le.whl", hash = "sha256:4c3446937be153718250fe421da548f973124189f18fe4575a0510b5c928f0cc"},
-    {file = "greenlet-3.1.0-cp311-cp311-manylinux_2_17_s390x.manylinux2014_s390x.whl", hash = "sha256:1ddc7bcedeb47187be74208bc652d63d6b20cb24f4e596bd356092d8000da6d6"},
-    {file = "greenlet-3.1.0-cp311-cp311-manylinux_2_17_x86_64.manylinux2014_x86_64.whl", hash = "sha256:44151d7b81b9391ed759a2f2865bbe623ef00d648fed59363be2bbbd5154656f"},
-    {file = "greenlet-3.1.0-cp311-cp311-manylinux_2_24_x86_64.manylinux_2_28_x86_64.whl", hash = "sha256:6cea1cca3be76c9483282dc7760ea1cc08a6ecec1f0b6ca0a94ea0d17432da19"},
-    {file = "greenlet-3.1.0-cp311-cp311-musllinux_1_1_aarch64.whl", hash = "sha256:619935a44f414274a2c08c9e74611965650b730eb4efe4b2270f91df5e4adf9a"},
-    {file = "greenlet-3.1.0-cp311-cp311-musllinux_1_1_x86_64.whl", hash = "sha256:221169d31cada333a0c7fd087b957c8f431c1dba202c3a58cf5a3583ed973e9b"},
-    {file = "greenlet-3.1.0-cp311-cp311-win_amd64.whl", hash = "sha256:01059afb9b178606b4b6e92c3e710ea1635597c3537e44da69f4531e111dd5e9"},
-    {file = "greenlet-3.1.0-cp312-cp312-macosx_11_0_universal2.whl", hash = "sha256:24fc216ec7c8be9becba8b64a98a78f9cd057fd2dc75ae952ca94ed8a893bf27"},
-    {file = "greenlet-3.1.0-cp312-cp312-manylinux_2_17_aarch64.manylinux2014_aarch64.whl", hash = "sha256:3d07c28b85b350564bdff9f51c1c5007dfb2f389385d1bc23288de51134ca303"},
-    {file = "greenlet-3.1.0-cp312-cp312-manylinux_2_17_ppc64le.manylinux2014_ppc64le.whl", hash = "sha256:243a223c96a4246f8a30ea470c440fe9db1f5e444941ee3c3cd79df119b8eebf"},
-    {file = "greenlet-3.1.0-cp312-cp312-manylinux_2_17_s390x.manylinux2014_s390x.whl", hash = "sha256:26811df4dc81271033a7836bc20d12cd30938e6bd2e9437f56fa03da81b0f8fc"},
-    {file = "greenlet-3.1.0-cp312-cp312-manylinux_2_17_x86_64.manylinux2014_x86_64.whl", hash = "sha256:c9d86401550b09a55410f32ceb5fe7efcd998bd2dad9e82521713cb148a4a15f"},
-    {file = "greenlet-3.1.0-cp312-cp312-manylinux_2_24_x86_64.manylinux_2_28_x86_64.whl", hash = "sha256:26d9c1c4f1748ccac0bae1dbb465fb1a795a75aba8af8ca871503019f4285e2a"},
-    {file = "greenlet-3.1.0-cp312-cp312-musllinux_1_1_aarch64.whl", hash = "sha256:cd468ec62257bb4544989402b19d795d2305eccb06cde5da0eb739b63dc04665"},
-    {file = "greenlet-3.1.0-cp312-cp312-musllinux_1_1_x86_64.whl", hash = "sha256:a53dfe8f82b715319e9953330fa5c8708b610d48b5c59f1316337302af5c0811"},
-    {file = "greenlet-3.1.0-cp312-cp312-win_amd64.whl", hash = "sha256:28fe80a3eb673b2d5cc3b12eea468a5e5f4603c26aa34d88bf61bba82ceb2f9b"},
-    {file = "greenlet-3.1.0-cp313-cp313-macosx_11_0_universal2.whl", hash = "sha256:76b3e3976d2a452cba7aa9e453498ac72240d43030fdc6d538a72b87eaff52fd"},
-    {file = "greenlet-3.1.0-cp313-cp313-manylinux_2_17_aarch64.manylinux2014_aarch64.whl", hash = "sha256:655b21ffd37a96b1e78cc48bf254f5ea4b5b85efaf9e9e2a526b3c9309d660ca"},
-    {file = "greenlet-3.1.0-cp313-cp313-manylinux_2_17_ppc64le.manylinux2014_ppc64le.whl", hash = "sha256:c6f4c2027689093775fd58ca2388d58789009116844432d920e9147f91acbe64"},
-    {file = "greenlet-3.1.0-cp313-cp313-manylinux_2_17_s390x.manylinux2014_s390x.whl", hash = "sha256:76e5064fd8e94c3f74d9fd69b02d99e3cdb8fc286ed49a1f10b256e59d0d3a0b"},
-    {file = "greenlet-3.1.0-cp313-cp313-manylinux_2_17_x86_64.manylinux2014_x86_64.whl", hash = "sha256:6a4bf607f690f7987ab3291406e012cd8591a4f77aa54f29b890f9c331e84989"},
-    {file = "greenlet-3.1.0-cp313-cp313-manylinux_2_24_x86_64.manylinux_2_28_x86_64.whl", hash = "sha256:037d9ac99540ace9424cb9ea89f0accfaff4316f149520b4ae293eebc5bded17"},
-    {file = "greenlet-3.1.0-cp313-cp313-musllinux_1_1_aarch64.whl", hash = "sha256:90b5bbf05fe3d3ef697103850c2ce3374558f6fe40fd57c9fac1bf14903f50a5"},
-    {file = "greenlet-3.1.0-cp313-cp313-musllinux_1_1_x86_64.whl", hash = "sha256:726377bd60081172685c0ff46afbc600d064f01053190e4450857483c4d44484"},
-    {file = "greenlet-3.1.0-cp313-cp313-win_amd64.whl", hash = "sha256:d46d5069e2eeda111d6f71970e341f4bd9aeeee92074e649ae263b834286ecc0"},
-    {file = "greenlet-3.1.0-cp37-cp37m-manylinux_2_17_aarch64.manylinux2014_aarch64.whl", hash = "sha256:81eeec4403a7d7684b5812a8aaa626fa23b7d0848edb3a28d2eb3220daddcbd0"},
-    {file = "greenlet-3.1.0-cp37-cp37m-manylinux_2_17_ppc64le.manylinux2014_ppc64le.whl", hash = "sha256:4a3dae7492d16e85ea6045fd11cb8e782b63eac8c8d520c3a92c02ac4573b0a6"},
-    {file = "greenlet-3.1.0-cp37-cp37m-manylinux_2_17_s390x.manylinux2014_s390x.whl", hash = "sha256:4b5ea3664eed571779403858d7cd0a9b0ebf50d57d2cdeafc7748e09ef8cd81a"},
-    {file = "greenlet-3.1.0-cp37-cp37m-manylinux_2_17_x86_64.manylinux2014_x86_64.whl", hash = "sha256:a22f4e26400f7f48faef2d69c20dc055a1f3043d330923f9abe08ea0aecc44df"},
-    {file = "greenlet-3.1.0-cp37-cp37m-manylinux_2_24_x86_64.manylinux_2_28_x86_64.whl", hash = "sha256:13ff8c8e54a10472ce3b2a2da007f915175192f18e6495bad50486e87c7f6637"},
-    {file = "greenlet-3.1.0-cp37-cp37m-musllinux_1_1_aarch64.whl", hash = "sha256:f9671e7282d8c6fcabc32c0fb8d7c0ea8894ae85cee89c9aadc2d7129e1a9954"},
-    {file = "greenlet-3.1.0-cp37-cp37m-musllinux_1_1_x86_64.whl", hash = "sha256:184258372ae9e1e9bddce6f187967f2e08ecd16906557c4320e3ba88a93438c3"},
-    {file = "greenlet-3.1.0-cp37-cp37m-win32.whl", hash = "sha256:a0409bc18a9f85321399c29baf93545152d74a49d92f2f55302f122007cfda00"},
-    {file = "greenlet-3.1.0-cp37-cp37m-win_amd64.whl", hash = "sha256:9eb4a1d7399b9f3c7ac68ae6baa6be5f9195d1d08c9ddc45ad559aa6b556bce6"},
-    {file = "greenlet-3.1.0-cp38-cp38-macosx_11_0_universal2.whl", hash = "sha256:a8870983af660798dc1b529e1fd6f1cefd94e45135a32e58bd70edd694540f33"},
-    {file = "greenlet-3.1.0-cp38-cp38-manylinux_2_17_aarch64.manylinux2014_aarch64.whl", hash = "sha256:cfcfb73aed40f550a57ea904629bdaf2e562c68fa1164fa4588e752af6efdc3f"},
-    {file = "greenlet-3.1.0-cp38-cp38-manylinux_2_17_ppc64le.manylinux2014_ppc64le.whl", hash = "sha256:f9482c2ed414781c0af0b35d9d575226da6b728bd1a720668fa05837184965b7"},
-    {file = "greenlet-3.1.0-cp38-cp38-manylinux_2_17_s390x.manylinux2014_s390x.whl", hash = "sha256:d58ec349e0c2c0bc6669bf2cd4982d2f93bf067860d23a0ea1fe677b0f0b1e09"},
-    {file = "greenlet-3.1.0-cp38-cp38-manylinux_2_17_x86_64.manylinux2014_x86_64.whl", hash = "sha256:dd65695a8df1233309b701dec2539cc4b11e97d4fcc0f4185b4a12ce54db0491"},
-    {file = "greenlet-3.1.0-cp38-cp38-manylinux_2_24_x86_64.manylinux_2_28_x86_64.whl", hash = "sha256:665b21e95bc0fce5cab03b2e1d90ba9c66c510f1bb5fdc864f3a377d0f553f6b"},
-    {file = "greenlet-3.1.0-cp38-cp38-musllinux_1_1_aarch64.whl", hash = "sha256:d3c59a06c2c28a81a026ff11fbf012081ea34fb9b7052f2ed0366e14896f0a1d"},
-    {file = "greenlet-3.1.0-cp38-cp38-musllinux_1_1_x86_64.whl", hash = "sha256:5415b9494ff6240b09af06b91a375731febe0090218e2898d2b85f9b92abcda0"},
-    {file = "greenlet-3.1.0-cp38-cp38-win32.whl", hash = "sha256:1544b8dd090b494c55e60c4ff46e238be44fdc472d2589e943c241e0169bcea2"},
-    {file = "greenlet-3.1.0-cp38-cp38-win_amd64.whl", hash = "sha256:7f346d24d74c00b6730440f5eb8ec3fe5774ca8d1c9574e8e57c8671bb51b910"},
-    {file = "greenlet-3.1.0-cp39-cp39-macosx_10_9_universal2.whl", hash = "sha256:db1b3ccb93488328c74e97ff888604a8b95ae4f35f4f56677ca57a4fc3a4220b"},
-    {file = "greenlet-3.1.0-cp39-cp39-manylinux_2_17_aarch64.manylinux2014_aarch64.whl", hash = "sha256:44cd313629ded43bb3b98737bba2f3e2c2c8679b55ea29ed73daea6b755fe8e7"},
-    {file = "greenlet-3.1.0-cp39-cp39-manylinux_2_17_ppc64le.manylinux2014_ppc64le.whl", hash = "sha256:fad7a051e07f64e297e6e8399b4d6a3bdcad3d7297409e9a06ef8cbccff4f501"},
-    {file = "greenlet-3.1.0-cp39-cp39-manylinux_2_17_s390x.manylinux2014_s390x.whl", hash = "sha256:c3967dcc1cd2ea61b08b0b276659242cbce5caca39e7cbc02408222fb9e6ff39"},
-    {file = "greenlet-3.1.0-cp39-cp39-manylinux_2_17_x86_64.manylinux2014_x86_64.whl", hash = "sha256:d45b75b0f3fd8d99f62eb7908cfa6d727b7ed190737dec7fe46d993da550b81a"},
-    {file = "greenlet-3.1.0-cp39-cp39-manylinux_2_24_x86_64.manylinux_2_28_x86_64.whl", hash = "sha256:2d004db911ed7b6218ec5c5bfe4cf70ae8aa2223dffbb5b3c69e342bb253cb28"},
-    {file = "greenlet-3.1.0-cp39-cp39-musllinux_1_1_aarch64.whl", hash = "sha256:b9505a0c8579899057cbefd4ec34d865ab99852baf1ff33a9481eb3924e2da0b"},
-    {file = "greenlet-3.1.0-cp39-cp39-musllinux_1_1_x86_64.whl", hash = "sha256:5fd6e94593f6f9714dbad1aaba734b5ec04593374fa6638df61592055868f8b8"},
-    {file = "greenlet-3.1.0-cp39-cp39-win32.whl", hash = "sha256:d0dd943282231480aad5f50f89bdf26690c995e8ff555f26d8a5b9887b559bcc"},
-    {file = "greenlet-3.1.0-cp39-cp39-win_amd64.whl", hash = "sha256:ac0adfdb3a21dc2a24ed728b61e72440d297d0fd3a577389df566651fcd08f97"},
-    {file = "greenlet-3.1.0.tar.gz", hash = "sha256:b395121e9bbe8d02a750886f108d540abe66075e61e22f7353d9acb0b81be0f0"},
+    {file = "greenlet-3.1.1-cp310-cp310-macosx_11_0_universal2.whl", hash = "sha256:0bbae94a29c9e5c7e4a2b7f0aae5c17e8e90acbfd3bf6270eeba60c39fce3563"},
+    {file = "greenlet-3.1.1-cp310-cp310-manylinux_2_17_aarch64.manylinux2014_aarch64.whl", hash = "sha256:0fde093fb93f35ca72a556cf72c92ea3ebfda3d79fc35bb19fbe685853869a83"},
+    {file = "greenlet-3.1.1-cp310-cp310-manylinux_2_17_ppc64le.manylinux2014_ppc64le.whl", hash = "sha256:36b89d13c49216cadb828db8dfa6ce86bbbc476a82d3a6c397f0efae0525bdd0"},
+    {file = "greenlet-3.1.1-cp310-cp310-manylinux_2_17_s390x.manylinux2014_s390x.whl", hash = "sha256:94b6150a85e1b33b40b1464a3f9988dcc5251d6ed06842abff82e42632fac120"},
+    {file = "greenlet-3.1.1-cp310-cp310-manylinux_2_17_x86_64.manylinux2014_x86_64.whl", hash = "sha256:93147c513fac16385d1036b7e5b102c7fbbdb163d556b791f0f11eada7ba65dc"},
+    {file = "greenlet-3.1.1-cp310-cp310-manylinux_2_24_x86_64.manylinux_2_28_x86_64.whl", hash = "sha256:da7a9bff22ce038e19bf62c4dd1ec8391062878710ded0a845bcf47cc0200617"},
+    {file = "greenlet-3.1.1-cp310-cp310-musllinux_1_1_aarch64.whl", hash = "sha256:b2795058c23988728eec1f36a4e5e4ebad22f8320c85f3587b539b9ac84128d7"},
+    {file = "greenlet-3.1.1-cp310-cp310-musllinux_1_1_x86_64.whl", hash = "sha256:ed10eac5830befbdd0c32f83e8aa6288361597550ba669b04c48f0f9a2c843c6"},
+    {file = "greenlet-3.1.1-cp310-cp310-win_amd64.whl", hash = "sha256:77c386de38a60d1dfb8e55b8c1101d68c79dfdd25c7095d51fec2dd800892b80"},
+    {file = "greenlet-3.1.1-cp311-cp311-macosx_11_0_universal2.whl", hash = "sha256:e4d333e558953648ca09d64f13e6d8f0523fa705f51cae3f03b5983489958c70"},
+    {file = "greenlet-3.1.1-cp311-cp311-manylinux_2_17_aarch64.manylinux2014_aarch64.whl", hash = "sha256:09fc016b73c94e98e29af67ab7b9a879c307c6731a2c9da0db5a7d9b7edd1159"},
+    {file = "greenlet-3.1.1-cp311-cp311-manylinux_2_17_ppc64le.manylinux2014_ppc64le.whl", hash = "sha256:d5e975ca70269d66d17dd995dafc06f1b06e8cb1ec1e9ed54c1d1e4a7c4cf26e"},
+    {file = "greenlet-3.1.1-cp311-cp311-manylinux_2_17_s390x.manylinux2014_s390x.whl", hash = "sha256:3b2813dc3de8c1ee3f924e4d4227999285fd335d1bcc0d2be6dc3f1f6a318ec1"},
+    {file = "greenlet-3.1.1-cp311-cp311-manylinux_2_17_x86_64.manylinux2014_x86_64.whl", hash = "sha256:e347b3bfcf985a05e8c0b7d462ba6f15b1ee1c909e2dcad795e49e91b152c383"},
+    {file = "greenlet-3.1.1-cp311-cp311-manylinux_2_24_x86_64.manylinux_2_28_x86_64.whl", hash = "sha256:9e8f8c9cb53cdac7ba9793c276acd90168f416b9ce36799b9b885790f8ad6c0a"},
+    {file = "greenlet-3.1.1-cp311-cp311-musllinux_1_1_aarch64.whl", hash = "sha256:62ee94988d6b4722ce0028644418d93a52429e977d742ca2ccbe1c4f4a792511"},
+    {file = "greenlet-3.1.1-cp311-cp311-musllinux_1_1_x86_64.whl", hash = "sha256:1776fd7f989fc6b8d8c8cb8da1f6b82c5814957264d1f6cf818d475ec2bf6395"},
+    {file = "greenlet-3.1.1-cp311-cp311-win_amd64.whl", hash = "sha256:48ca08c771c268a768087b408658e216133aecd835c0ded47ce955381105ba39"},
+    {file = "greenlet-3.1.1-cp312-cp312-macosx_11_0_universal2.whl", hash = "sha256:4afe7ea89de619adc868e087b4d2359282058479d7cfb94970adf4b55284574d"},
+    {file = "greenlet-3.1.1-cp312-cp312-manylinux_2_17_aarch64.manylinux2014_aarch64.whl", hash = "sha256:f406b22b7c9a9b4f8aa9d2ab13d6ae0ac3e85c9a809bd590ad53fed2bf70dc79"},
+    {file = "greenlet-3.1.1-cp312-cp312-manylinux_2_17_ppc64le.manylinux2014_ppc64le.whl", hash = "sha256:c3a701fe5a9695b238503ce5bbe8218e03c3bcccf7e204e455e7462d770268aa"},
+    {file = "greenlet-3.1.1-cp312-cp312-manylinux_2_17_s390x.manylinux2014_s390x.whl", hash = "sha256:2846930c65b47d70b9d178e89c7e1a69c95c1f68ea5aa0a58646b7a96df12441"},
+    {file = "greenlet-3.1.1-cp312-cp312-manylinux_2_17_x86_64.manylinux2014_x86_64.whl", hash = "sha256:99cfaa2110534e2cf3ba31a7abcac9d328d1d9f1b95beede58294a60348fba36"},
+    {file = "greenlet-3.1.1-cp312-cp312-manylinux_2_24_x86_64.manylinux_2_28_x86_64.whl", hash = "sha256:1443279c19fca463fc33e65ef2a935a5b09bb90f978beab37729e1c3c6c25fe9"},
+    {file = "greenlet-3.1.1-cp312-cp312-musllinux_1_1_aarch64.whl", hash = "sha256:b7cede291382a78f7bb5f04a529cb18e068dd29e0fb27376074b6d0317bf4dd0"},
+    {file = "greenlet-3.1.1-cp312-cp312-musllinux_1_1_x86_64.whl", hash = "sha256:23f20bb60ae298d7d8656c6ec6db134bca379ecefadb0b19ce6f19d1f232a942"},
+    {file = "greenlet-3.1.1-cp312-cp312-win_amd64.whl", hash = "sha256:7124e16b4c55d417577c2077be379514321916d5790fa287c9ed6f23bd2ffd01"},
+    {file = "greenlet-3.1.1-cp313-cp313-macosx_11_0_universal2.whl", hash = "sha256:05175c27cb459dcfc05d026c4232f9de8913ed006d42713cb8a5137bd49375f1"},
+    {file = "greenlet-3.1.1-cp313-cp313-manylinux_2_17_aarch64.manylinux2014_aarch64.whl", hash = "sha256:935e943ec47c4afab8965954bf49bfa639c05d4ccf9ef6e924188f762145c0ff"},
+    {file = "greenlet-3.1.1-cp313-cp313-manylinux_2_17_ppc64le.manylinux2014_ppc64le.whl", hash = "sha256:667a9706c970cb552ede35aee17339a18e8f2a87a51fba2ed39ceeeb1004798a"},
+    {file = "greenlet-3.1.1-cp313-cp313-manylinux_2_17_s390x.manylinux2014_s390x.whl", hash = "sha256:b8a678974d1f3aa55f6cc34dc480169d58f2e6d8958895d68845fa4ab566509e"},
+    {file = "greenlet-3.1.1-cp313-cp313-manylinux_2_17_x86_64.manylinux2014_x86_64.whl", hash = "sha256:efc0f674aa41b92da8c49e0346318c6075d734994c3c4e4430b1c3f853e498e4"},
+    {file = "greenlet-3.1.1-cp313-cp313-manylinux_2_24_x86_64.manylinux_2_28_x86_64.whl", hash = "sha256:0153404a4bb921f0ff1abeb5ce8a5131da56b953eda6e14b88dc6bbc04d2049e"},
+    {file = "greenlet-3.1.1-cp313-cp313-musllinux_1_1_aarch64.whl", hash = "sha256:275f72decf9932639c1c6dd1013a1bc266438eb32710016a1c742df5da6e60a1"},
+    {file = "greenlet-3.1.1-cp313-cp313-musllinux_1_1_x86_64.whl", hash = "sha256:c4aab7f6381f38a4b42f269057aee279ab0fc7bf2e929e3d4abfae97b682a12c"},
+    {file = "greenlet-3.1.1-cp313-cp313-win_amd64.whl", hash = "sha256:b42703b1cf69f2aa1df7d1030b9d77d3e584a70755674d60e710f0af570f3761"},
+    {file = "greenlet-3.1.1-cp313-cp313t-manylinux_2_17_aarch64.manylinux2014_aarch64.whl", hash = "sha256:f1695e76146579f8c06c1509c7ce4dfe0706f49c6831a817ac04eebb2fd02011"},
+    {file = "greenlet-3.1.1-cp313-cp313t-manylinux_2_17_ppc64le.manylinux2014_ppc64le.whl", hash = "sha256:7876452af029456b3f3549b696bb36a06db7c90747740c5302f74a9e9fa14b13"},
+    {file = "greenlet-3.1.1-cp313-cp313t-manylinux_2_17_s390x.manylinux2014_s390x.whl", hash = "sha256:4ead44c85f8ab905852d3de8d86f6f8baf77109f9da589cb4fa142bd3b57b475"},
+    {file = "greenlet-3.1.1-cp313-cp313t-manylinux_2_17_x86_64.manylinux2014_x86_64.whl", hash = "sha256:8320f64b777d00dd7ccdade271eaf0cad6636343293a25074cc5566160e4de7b"},
+    {file = "greenlet-3.1.1-cp313-cp313t-manylinux_2_24_x86_64.manylinux_2_28_x86_64.whl", hash = "sha256:6510bf84a6b643dabba74d3049ead221257603a253d0a9873f55f6a59a65f822"},
+    {file = "greenlet-3.1.1-cp313-cp313t-musllinux_1_1_aarch64.whl", hash = "sha256:04b013dc07c96f83134b1e99888e7a79979f1a247e2a9f59697fa14b5862ed01"},
+    {file = "greenlet-3.1.1-cp313-cp313t-musllinux_1_1_x86_64.whl", hash = "sha256:411f015496fec93c1c8cd4e5238da364e1da7a124bcb293f085bf2860c32c6f6"},
+    {file = "greenlet-3.1.1-cp37-cp37m-manylinux_2_17_aarch64.manylinux2014_aarch64.whl", hash = "sha256:47da355d8687fd65240c364c90a31569a133b7b60de111c255ef5b606f2ae291"},
+    {file = "greenlet-3.1.1-cp37-cp37m-manylinux_2_17_ppc64le.manylinux2014_ppc64le.whl", hash = "sha256:98884ecf2ffb7d7fe6bd517e8eb99d31ff7855a840fa6d0d63cd07c037f6a981"},
+    {file = "greenlet-3.1.1-cp37-cp37m-manylinux_2_17_s390x.manylinux2014_s390x.whl", hash = "sha256:f1d4aeb8891338e60d1ab6127af1fe45def5259def8094b9c7e34690c8858803"},
+    {file = "greenlet-3.1.1-cp37-cp37m-manylinux_2_17_x86_64.manylinux2014_x86_64.whl", hash = "sha256:db32b5348615a04b82240cc67983cb315309e88d444a288934ee6ceaebcad6cc"},
+    {file = "greenlet-3.1.1-cp37-cp37m-manylinux_2_24_x86_64.manylinux_2_28_x86_64.whl", hash = "sha256:dcc62f31eae24de7f8dce72134c8651c58000d3b1868e01392baea7c32c247de"},
+    {file = "greenlet-3.1.1-cp37-cp37m-musllinux_1_1_aarch64.whl", hash = "sha256:1d3755bcb2e02de341c55b4fca7a745a24a9e7212ac953f6b3a48d117d7257aa"},
+    {file = "greenlet-3.1.1-cp37-cp37m-musllinux_1_1_x86_64.whl", hash = "sha256:b8da394b34370874b4572676f36acabac172602abf054cbc4ac910219f3340af"},
+    {file = "greenlet-3.1.1-cp37-cp37m-win32.whl", hash = "sha256:a0dfc6c143b519113354e780a50381508139b07d2177cb6ad6a08278ec655798"},
+    {file = "greenlet-3.1.1-cp37-cp37m-win_amd64.whl", hash = "sha256:54558ea205654b50c438029505def3834e80f0869a70fb15b871c29b4575ddef"},
+    {file = "greenlet-3.1.1-cp38-cp38-macosx_11_0_universal2.whl", hash = "sha256:346bed03fe47414091be4ad44786d1bd8bef0c3fcad6ed3dee074a032ab408a9"},
+    {file = "greenlet-3.1.1-cp38-cp38-manylinux_2_17_aarch64.manylinux2014_aarch64.whl", hash = "sha256:dfc59d69fc48664bc693842bd57acfdd490acafda1ab52c7836e3fc75c90a111"},
+    {file = "greenlet-3.1.1-cp38-cp38-manylinux_2_17_ppc64le.manylinux2014_ppc64le.whl", hash = "sha256:d21e10da6ec19b457b82636209cbe2331ff4306b54d06fa04b7c138ba18c8a81"},
+    {file = "greenlet-3.1.1-cp38-cp38-manylinux_2_17_s390x.manylinux2014_s390x.whl", hash = "sha256:37b9de5a96111fc15418819ab4c4432e4f3c2ede61e660b1e33971eba26ef9ba"},
+    {file = "greenlet-3.1.1-cp38-cp38-manylinux_2_17_x86_64.manylinux2014_x86_64.whl", hash = "sha256:6ef9ea3f137e5711f0dbe5f9263e8c009b7069d8a1acea822bd5e9dae0ae49c8"},
+    {file = "greenlet-3.1.1-cp38-cp38-manylinux_2_24_x86_64.manylinux_2_28_x86_64.whl", hash = "sha256:85f3ff71e2e60bd4b4932a043fbbe0f499e263c628390b285cb599154a3b03b1"},
+    {file = "greenlet-3.1.1-cp38-cp38-musllinux_1_1_aarch64.whl", hash = "sha256:95ffcf719966dd7c453f908e208e14cde192e09fde6c7186c8f1896ef778d8cd"},
+    {file = "greenlet-3.1.1-cp38-cp38-musllinux_1_1_x86_64.whl", hash = "sha256:03a088b9de532cbfe2ba2034b2b85e82df37874681e8c470d6fb2f8c04d7e4b7"},
+    {file = "greenlet-3.1.1-cp38-cp38-win32.whl", hash = "sha256:8b8b36671f10ba80e159378df9c4f15c14098c4fd73a36b9ad715f057272fbef"},
+    {file = "greenlet-3.1.1-cp38-cp38-win_amd64.whl", hash = "sha256:7017b2be767b9d43cc31416aba48aab0d2309ee31b4dbf10a1d38fb7972bdf9d"},
+    {file = "greenlet-3.1.1-cp39-cp39-macosx_11_0_universal2.whl", hash = "sha256:396979749bd95f018296af156201d6211240e7a23090f50a8d5d18c370084dc3"},
+    {file = "greenlet-3.1.1-cp39-cp39-manylinux_2_17_aarch64.manylinux2014_aarch64.whl", hash = "sha256:ca9d0ff5ad43e785350894d97e13633a66e2b50000e8a183a50a88d834752d42"},
+    {file = "greenlet-3.1.1-cp39-cp39-manylinux_2_17_ppc64le.manylinux2014_ppc64le.whl", hash = "sha256:f6ff3b14f2df4c41660a7dec01045a045653998784bf8cfcb5a525bdffffbc8f"},
+    {file = "greenlet-3.1.1-cp39-cp39-manylinux_2_17_s390x.manylinux2014_s390x.whl", hash = "sha256:94ebba31df2aa506d7b14866fed00ac141a867e63143fe5bca82a8e503b36437"},
+    {file = "greenlet-3.1.1-cp39-cp39-manylinux_2_17_x86_64.manylinux2014_x86_64.whl", hash = "sha256:73aaad12ac0ff500f62cebed98d8789198ea0e6f233421059fa68a5aa7220145"},
+    {file = "greenlet-3.1.1-cp39-cp39-manylinux_2_24_x86_64.manylinux_2_28_x86_64.whl", hash = "sha256:63e4844797b975b9af3a3fb8f7866ff08775f5426925e1e0bbcfe7932059a12c"},
+    {file = "greenlet-3.1.1-cp39-cp39-musllinux_1_1_aarch64.whl", hash = "sha256:7939aa3ca7d2a1593596e7ac6d59391ff30281ef280d8632fa03d81f7c5f955e"},
+    {file = "greenlet-3.1.1-cp39-cp39-musllinux_1_1_x86_64.whl", hash = "sha256:d0028e725ee18175c6e422797c407874da24381ce0690d6b9396c204c7f7276e"},
+    {file = "greenlet-3.1.1-cp39-cp39-win32.whl", hash = "sha256:5e06afd14cbaf9e00899fae69b24a32f2196c19de08fcb9f4779dd4f004e5e7c"},
+    {file = "greenlet-3.1.1-cp39-cp39-win_amd64.whl", hash = "sha256:3319aa75e0e0639bc15ff54ca327e8dc7a6fe404003496e3c6925cd3142e0e22"},
+    {file = "greenlet-3.1.1.tar.gz", hash = "sha256:4ce3ac6cdb6adf7946475d7ef31777c26d94bccc377e070a7986bd2d5c515467"},
 ]
 
 [package.extras]
@@ -3508,61 +3534,70 @@ protobuf = ">=3.20.2,<4.21.1 || >4.21.1,<4.21.2 || >4.21.2,<4.21.3 || >4.21.3,<4
 
 [[package]]
 name = "grpcio"
-version = "1.66.1"
+version = "1.66.2"
 description = "HTTP/2-based RPC framework"
 optional = false
 python-versions = ">=3.8"
 files = [
-    {file = "grpcio-1.66.1-cp310-cp310-linux_armv7l.whl", hash = "sha256:4877ba180591acdf127afe21ec1c7ff8a5ecf0fe2600f0d3c50e8c4a1cbc6492"},
-    {file = "grpcio-1.66.1-cp310-cp310-macosx_12_0_universal2.whl", hash = "sha256:3750c5a00bd644c75f4507f77a804d0189d97a107eb1481945a0cf3af3e7a5ac"},
-    {file = "grpcio-1.66.1-cp310-cp310-manylinux_2_17_aarch64.whl", hash = "sha256:a013c5fbb12bfb5f927444b477a26f1080755a931d5d362e6a9a720ca7dbae60"},
-    {file = "grpcio-1.66.1-cp310-cp310-manylinux_2_17_i686.manylinux2014_i686.whl", hash = "sha256:b1b24c23d51a1e8790b25514157d43f0a4dce1ac12b3f0b8e9f66a5e2c4c132f"},
-    {file = "grpcio-1.66.1-cp310-cp310-manylinux_2_17_x86_64.manylinux2014_x86_64.whl", hash = "sha256:b7ffb8ea674d68de4cac6f57d2498fef477cef582f1fa849e9f844863af50083"},
-    {file = "grpcio-1.66.1-cp310-cp310-musllinux_1_1_i686.whl", hash = "sha256:307b1d538140f19ccbd3aed7a93d8f71103c5d525f3c96f8616111614b14bf2a"},
-    {file = "grpcio-1.66.1-cp310-cp310-musllinux_1_1_x86_64.whl", hash = "sha256:1c17ebcec157cfb8dd445890a03e20caf6209a5bd4ac5b040ae9dbc59eef091d"},
-    {file = "grpcio-1.66.1-cp310-cp310-win32.whl", hash = "sha256:ef82d361ed5849d34cf09105d00b94b6728d289d6b9235513cb2fcc79f7c432c"},
-    {file = "grpcio-1.66.1-cp310-cp310-win_amd64.whl", hash = "sha256:292a846b92cdcd40ecca46e694997dd6b9be6c4c01a94a0dfb3fcb75d20da858"},
-    {file = "grpcio-1.66.1-cp311-cp311-linux_armv7l.whl", hash = "sha256:c30aeceeaff11cd5ddbc348f37c58bcb96da8d5aa93fed78ab329de5f37a0d7a"},
-    {file = "grpcio-1.66.1-cp311-cp311-macosx_10_9_universal2.whl", hash = "sha256:8a1e224ce6f740dbb6b24c58f885422deebd7eb724aff0671a847f8951857c26"},
-    {file = "grpcio-1.66.1-cp311-cp311-manylinux_2_17_aarch64.whl", hash = "sha256:a66fe4dc35d2330c185cfbb42959f57ad36f257e0cc4557d11d9f0a3f14311df"},
-    {file = "grpcio-1.66.1-cp311-cp311-manylinux_2_17_i686.manylinux2014_i686.whl", hash = "sha256:e3ba04659e4fce609de2658fe4dbf7d6ed21987a94460f5f92df7579fd5d0e22"},
-    {file = "grpcio-1.66.1-cp311-cp311-manylinux_2_17_x86_64.manylinux2014_x86_64.whl", hash = "sha256:4573608e23f7e091acfbe3e84ac2045680b69751d8d67685ffa193a4429fedb1"},
-    {file = "grpcio-1.66.1-cp311-cp311-musllinux_1_1_i686.whl", hash = "sha256:7e06aa1f764ec8265b19d8f00140b8c4b6ca179a6dc67aa9413867c47e1fb04e"},
-    {file = "grpcio-1.66.1-cp311-cp311-musllinux_1_1_x86_64.whl", hash = "sha256:3885f037eb11f1cacc41f207b705f38a44b69478086f40608959bf5ad85826dd"},
-    {file = "grpcio-1.66.1-cp311-cp311-win32.whl", hash = "sha256:97ae7edd3f3f91480e48ede5d3e7d431ad6005bfdbd65c1b56913799ec79e791"},
-    {file = "grpcio-1.66.1-cp311-cp311-win_amd64.whl", hash = "sha256:cfd349de4158d797db2bd82d2020554a121674e98fbe6b15328456b3bf2495bb"},
-    {file = "grpcio-1.66.1-cp312-cp312-linux_armv7l.whl", hash = "sha256:a92c4f58c01c77205df6ff999faa008540475c39b835277fb8883b11cada127a"},
-    {file = "grpcio-1.66.1-cp312-cp312-macosx_10_9_universal2.whl", hash = "sha256:fdb14bad0835914f325349ed34a51940bc2ad965142eb3090081593c6e347be9"},
-    {file = "grpcio-1.66.1-cp312-cp312-manylinux_2_17_aarch64.whl", hash = "sha256:f03a5884c56256e08fd9e262e11b5cfacf1af96e2ce78dc095d2c41ccae2c80d"},
-    {file = "grpcio-1.66.1-cp312-cp312-manylinux_2_17_i686.manylinux2014_i686.whl", hash = "sha256:2ca2559692d8e7e245d456877a85ee41525f3ed425aa97eb7a70fc9a79df91a0"},
-    {file = "grpcio-1.66.1-cp312-cp312-manylinux_2_17_x86_64.manylinux2014_x86_64.whl", hash = "sha256:84ca1be089fb4446490dd1135828bd42a7c7f8421e74fa581611f7afdf7ab761"},
-    {file = "grpcio-1.66.1-cp312-cp312-musllinux_1_1_i686.whl", hash = "sha256:d639c939ad7c440c7b2819a28d559179a4508783f7e5b991166f8d7a34b52815"},
-    {file = "grpcio-1.66.1-cp312-cp312-musllinux_1_1_x86_64.whl", hash = "sha256:b9feb4e5ec8dc2d15709f4d5fc367794d69277f5d680baf1910fc9915c633524"},
-    {file = "grpcio-1.66.1-cp312-cp312-win32.whl", hash = "sha256:7101db1bd4cd9b880294dec41a93fcdce465bdbb602cd8dc5bd2d6362b618759"},
-    {file = "grpcio-1.66.1-cp312-cp312-win_amd64.whl", hash = "sha256:b0aa03d240b5539648d996cc60438f128c7f46050989e35b25f5c18286c86734"},
-    {file = "grpcio-1.66.1-cp38-cp38-linux_armv7l.whl", hash = "sha256:ecfe735e7a59e5a98208447293ff8580e9db1e890e232b8b292dc8bd15afc0d2"},
-    {file = "grpcio-1.66.1-cp38-cp38-macosx_10_9_universal2.whl", hash = "sha256:4825a3aa5648010842e1c9d35a082187746aa0cdbf1b7a2a930595a94fb10fce"},
-    {file = "grpcio-1.66.1-cp38-cp38-manylinux_2_17_aarch64.whl", hash = "sha256:f517fd7259fe823ef3bd21e508b653d5492e706e9f0ef82c16ce3347a8a5620c"},
-    {file = "grpcio-1.66.1-cp38-cp38-manylinux_2_17_i686.manylinux2014_i686.whl", hash = "sha256:f1fe60d0772831d96d263b53d83fb9a3d050a94b0e94b6d004a5ad111faa5b5b"},
-    {file = "grpcio-1.66.1-cp38-cp38-manylinux_2_17_x86_64.manylinux2014_x86_64.whl", hash = "sha256:31a049daa428f928f21090403e5d18ea02670e3d5d172581670be006100db9ef"},
-    {file = "grpcio-1.66.1-cp38-cp38-musllinux_1_1_i686.whl", hash = "sha256:6f914386e52cbdeb5d2a7ce3bf1fdfacbe9d818dd81b6099a05b741aaf3848bb"},
-    {file = "grpcio-1.66.1-cp38-cp38-musllinux_1_1_x86_64.whl", hash = "sha256:bff2096bdba686019fb32d2dde45b95981f0d1490e054400f70fc9a8af34b49d"},
-    {file = "grpcio-1.66.1-cp38-cp38-win32.whl", hash = "sha256:aa8ba945c96e73de29d25331b26f3e416e0c0f621e984a3ebdb2d0d0b596a3b3"},
-    {file = "grpcio-1.66.1-cp38-cp38-win_amd64.whl", hash = "sha256:161d5c535c2bdf61b95080e7f0f017a1dfcb812bf54093e71e5562b16225b4ce"},
-    {file = "grpcio-1.66.1-cp39-cp39-linux_armv7l.whl", hash = "sha256:d0cd7050397b3609ea51727b1811e663ffda8bda39c6a5bb69525ef12414b503"},
-    {file = "grpcio-1.66.1-cp39-cp39-macosx_10_9_universal2.whl", hash = "sha256:0e6c9b42ded5d02b6b1fea3a25f036a2236eeb75d0579bfd43c0018c88bf0a3e"},
-    {file = "grpcio-1.66.1-cp39-cp39-manylinux_2_17_aarch64.whl", hash = "sha256:c9f80f9fad93a8cf71c7f161778ba47fd730d13a343a46258065c4deb4b550c0"},
-    {file = "grpcio-1.66.1-cp39-cp39-manylinux_2_17_i686.manylinux2014_i686.whl", hash = "sha256:5dd67ed9da78e5121efc5c510f0122a972216808d6de70953a740560c572eb44"},
-    {file = "grpcio-1.66.1-cp39-cp39-manylinux_2_17_x86_64.manylinux2014_x86_64.whl", hash = "sha256:48b0d92d45ce3be2084b92fb5bae2f64c208fea8ceed7fccf6a7b524d3c4942e"},
-    {file = "grpcio-1.66.1-cp39-cp39-musllinux_1_1_i686.whl", hash = "sha256:4d813316d1a752be6f5c4360c49f55b06d4fe212d7df03253dfdae90c8a402bb"},
-    {file = "grpcio-1.66.1-cp39-cp39-musllinux_1_1_x86_64.whl", hash = "sha256:9c9bebc6627873ec27a70fc800f6083a13c70b23a5564788754b9ee52c5aef6c"},
-    {file = "grpcio-1.66.1-cp39-cp39-win32.whl", hash = "sha256:30a1c2cf9390c894c90bbc70147f2372130ad189cffef161f0432d0157973f45"},
-    {file = "grpcio-1.66.1-cp39-cp39-win_amd64.whl", hash = "sha256:17663598aadbedc3cacd7bbde432f541c8e07d2496564e22b214b22c7523dac8"},
-    {file = "grpcio-1.66.1.tar.gz", hash = "sha256:35334f9c9745add3e357e3372756fd32d925bd52c41da97f4dfdafbde0bf0ee2"},
-]
-
-[package.extras]
-protobuf = ["grpcio-tools (>=1.66.1)"]
+    {file = "grpcio-1.66.2-cp310-cp310-linux_armv7l.whl", hash = "sha256:fe96281713168a3270878255983d2cb1a97e034325c8c2c25169a69289d3ecfa"},
+    {file = "grpcio-1.66.2-cp310-cp310-macosx_12_0_universal2.whl", hash = "sha256:73fc8f8b9b5c4a03e802b3cd0c18b2b06b410d3c1dcbef989fdeb943bd44aff7"},
+    {file = "grpcio-1.66.2-cp310-cp310-manylinux_2_17_aarch64.whl", hash = "sha256:03b0b307ba26fae695e067b94cbb014e27390f8bc5ac7a3a39b7723fed085604"},
+    {file = "grpcio-1.66.2-cp310-cp310-manylinux_2_17_i686.manylinux2014_i686.whl", hash = "sha256:7d69ce1f324dc2d71e40c9261d3fdbe7d4c9d60f332069ff9b2a4d8a257c7b2b"},
+    {file = "grpcio-1.66.2-cp310-cp310-manylinux_2_17_x86_64.manylinux2014_x86_64.whl", hash = "sha256:05bc2ceadc2529ab0b227b1310d249d95d9001cd106aa4d31e8871ad3c428d73"},
+    {file = "grpcio-1.66.2-cp310-cp310-musllinux_1_1_i686.whl", hash = "sha256:8ac475e8da31484efa25abb774674d837b343afb78bb3bcdef10f81a93e3d6bf"},
+    {file = "grpcio-1.66.2-cp310-cp310-musllinux_1_1_x86_64.whl", hash = "sha256:0be4e0490c28da5377283861bed2941d1d20ec017ca397a5df4394d1c31a9b50"},
+    {file = "grpcio-1.66.2-cp310-cp310-win32.whl", hash = "sha256:4e504572433f4e72b12394977679161d495c4c9581ba34a88d843eaf0f2fbd39"},
+    {file = "grpcio-1.66.2-cp310-cp310-win_amd64.whl", hash = "sha256:2018b053aa15782db2541ca01a7edb56a0bf18c77efed975392583725974b249"},
+    {file = "grpcio-1.66.2-cp311-cp311-linux_armv7l.whl", hash = "sha256:2335c58560a9e92ac58ff2bc5649952f9b37d0735608242973c7a8b94a6437d8"},
+    {file = "grpcio-1.66.2-cp311-cp311-macosx_10_9_universal2.whl", hash = "sha256:45a3d462826f4868b442a6b8fdbe8b87b45eb4f5b5308168c156b21eca43f61c"},
+    {file = "grpcio-1.66.2-cp311-cp311-manylinux_2_17_aarch64.whl", hash = "sha256:a9539f01cb04950fd4b5ab458e64a15f84c2acc273670072abe49a3f29bbad54"},
+    {file = "grpcio-1.66.2-cp311-cp311-manylinux_2_17_i686.manylinux2014_i686.whl", hash = "sha256:ce89f5876662f146d4c1f695dda29d4433a5d01c8681fbd2539afff535da14d4"},
+    {file = "grpcio-1.66.2-cp311-cp311-manylinux_2_17_x86_64.manylinux2014_x86_64.whl", hash = "sha256:d25a14af966438cddf498b2e338f88d1c9706f3493b1d73b93f695c99c5f0e2a"},
+    {file = "grpcio-1.66.2-cp311-cp311-musllinux_1_1_i686.whl", hash = "sha256:6001e575b8bbd89eee11960bb640b6da6ae110cf08113a075f1e2051cc596cae"},
+    {file = "grpcio-1.66.2-cp311-cp311-musllinux_1_1_x86_64.whl", hash = "sha256:4ea1d062c9230278793820146c95d038dc0f468cbdd172eec3363e42ff1c7d01"},
+    {file = "grpcio-1.66.2-cp311-cp311-win32.whl", hash = "sha256:38b68498ff579a3b1ee8f93a05eb48dc2595795f2f62716e797dc24774c1aaa8"},
+    {file = "grpcio-1.66.2-cp311-cp311-win_amd64.whl", hash = "sha256:6851de821249340bdb100df5eacfecfc4e6075fa85c6df7ee0eb213170ec8e5d"},
+    {file = "grpcio-1.66.2-cp312-cp312-linux_armv7l.whl", hash = "sha256:802d84fd3d50614170649853d121baaaa305de7b65b3e01759247e768d691ddf"},
+    {file = "grpcio-1.66.2-cp312-cp312-macosx_10_9_universal2.whl", hash = "sha256:80fd702ba7e432994df208f27514280b4b5c6843e12a48759c9255679ad38db8"},
+    {file = "grpcio-1.66.2-cp312-cp312-manylinux_2_17_aarch64.whl", hash = "sha256:12fda97ffae55e6526825daf25ad0fa37483685952b5d0f910d6405c87e3adb6"},
+    {file = "grpcio-1.66.2-cp312-cp312-manylinux_2_17_i686.manylinux2014_i686.whl", hash = "sha256:950da58d7d80abd0ea68757769c9db0a95b31163e53e5bb60438d263f4bed7b7"},
+    {file = "grpcio-1.66.2-cp312-cp312-manylinux_2_17_x86_64.manylinux2014_x86_64.whl", hash = "sha256:e636ce23273683b00410f1971d209bf3689238cf5538d960adc3cdfe80dd0dbd"},
+    {file = "grpcio-1.66.2-cp312-cp312-musllinux_1_1_i686.whl", hash = "sha256:a917d26e0fe980b0ac7bfcc1a3c4ad6a9a4612c911d33efb55ed7833c749b0ee"},
+    {file = "grpcio-1.66.2-cp312-cp312-musllinux_1_1_x86_64.whl", hash = "sha256:49f0ca7ae850f59f828a723a9064cadbed90f1ece179d375966546499b8a2c9c"},
+    {file = "grpcio-1.66.2-cp312-cp312-win32.whl", hash = "sha256:31fd163105464797a72d901a06472860845ac157389e10f12631025b3e4d0453"},
+    {file = "grpcio-1.66.2-cp312-cp312-win_amd64.whl", hash = "sha256:ff1f7882e56c40b0d33c4922c15dfa30612f05fb785074a012f7cda74d1c3679"},
+    {file = "grpcio-1.66.2-cp313-cp313-linux_armv7l.whl", hash = "sha256:3b00efc473b20d8bf83e0e1ae661b98951ca56111feb9b9611df8efc4fe5d55d"},
+    {file = "grpcio-1.66.2-cp313-cp313-macosx_10_13_universal2.whl", hash = "sha256:1caa38fb22a8578ab8393da99d4b8641e3a80abc8fd52646f1ecc92bcb8dee34"},
+    {file = "grpcio-1.66.2-cp313-cp313-manylinux_2_17_aarch64.whl", hash = "sha256:c408f5ef75cfffa113cacd8b0c0e3611cbfd47701ca3cdc090594109b9fcbaed"},
+    {file = "grpcio-1.66.2-cp313-cp313-manylinux_2_17_i686.manylinux2014_i686.whl", hash = "sha256:c806852deaedee9ce8280fe98955c9103f62912a5b2d5ee7e3eaa284a6d8d8e7"},
+    {file = "grpcio-1.66.2-cp313-cp313-manylinux_2_17_x86_64.manylinux2014_x86_64.whl", hash = "sha256:f145cc21836c332c67baa6fc81099d1d27e266401565bf481948010d6ea32d46"},
+    {file = "grpcio-1.66.2-cp313-cp313-musllinux_1_1_i686.whl", hash = "sha256:73e3b425c1e155730273f73e419de3074aa5c5e936771ee0e4af0814631fb30a"},
+    {file = "grpcio-1.66.2-cp313-cp313-musllinux_1_1_x86_64.whl", hash = "sha256:9c509a4f78114cbc5f0740eb3d7a74985fd2eff022971bc9bc31f8bc93e66a3b"},
+    {file = "grpcio-1.66.2-cp313-cp313-win32.whl", hash = "sha256:20657d6b8cfed7db5e11b62ff7dfe2e12064ea78e93f1434d61888834bc86d75"},
+    {file = "grpcio-1.66.2-cp313-cp313-win_amd64.whl", hash = "sha256:fb70487c95786e345af5e854ffec8cb8cc781bcc5df7930c4fbb7feaa72e1cdf"},
+    {file = "grpcio-1.66.2-cp38-cp38-linux_armv7l.whl", hash = "sha256:a18e20d8321c6400185b4263e27982488cb5cdd62da69147087a76a24ef4e7e3"},
+    {file = "grpcio-1.66.2-cp38-cp38-macosx_10_9_universal2.whl", hash = "sha256:02697eb4a5cbe5a9639f57323b4c37bcb3ab2d48cec5da3dc2f13334d72790dd"},
+    {file = "grpcio-1.66.2-cp38-cp38-manylinux_2_17_aarch64.whl", hash = "sha256:99a641995a6bc4287a6315989ee591ff58507aa1cbe4c2e70d88411c4dcc0839"},
+    {file = "grpcio-1.66.2-cp38-cp38-manylinux_2_17_i686.manylinux2014_i686.whl", hash = "sha256:3ed71e81782966ffead60268bbda31ea3f725ebf8aa73634d5dda44f2cf3fb9c"},
+    {file = "grpcio-1.66.2-cp38-cp38-manylinux_2_17_x86_64.manylinux2014_x86_64.whl", hash = "sha256:bbd27c24a4cc5e195a7f56cfd9312e366d5d61b86e36d46bbe538457ea6eb8dd"},
+    {file = "grpcio-1.66.2-cp38-cp38-musllinux_1_1_i686.whl", hash = "sha256:d9a9724a156c8ec6a379869b23ba3323b7ea3600851c91489b871e375f710bc8"},
+    {file = "grpcio-1.66.2-cp38-cp38-musllinux_1_1_x86_64.whl", hash = "sha256:d8d4732cc5052e92cea2f78b233c2e2a52998ac40cd651f40e398893ad0d06ec"},
+    {file = "grpcio-1.66.2-cp38-cp38-win32.whl", hash = "sha256:7b2c86457145ce14c38e5bf6bdc19ef88e66c5fee2c3d83285c5aef026ba93b3"},
+    {file = "grpcio-1.66.2-cp38-cp38-win_amd64.whl", hash = "sha256:e88264caad6d8d00e7913996030bac8ad5f26b7411495848cc218bd3a9040b6c"},
+    {file = "grpcio-1.66.2-cp39-cp39-linux_armv7l.whl", hash = "sha256:c400ba5675b67025c8a9f48aa846f12a39cf0c44df5cd060e23fda5b30e9359d"},
+    {file = "grpcio-1.66.2-cp39-cp39-macosx_10_9_universal2.whl", hash = "sha256:66a0cd8ba6512b401d7ed46bb03f4ee455839957f28b8d61e7708056a806ba6a"},
+    {file = "grpcio-1.66.2-cp39-cp39-manylinux_2_17_aarch64.whl", hash = "sha256:06de8ec0bd71be123eec15b0e0d457474931c2c407869b6c349bd9bed4adbac3"},
+    {file = "grpcio-1.66.2-cp39-cp39-manylinux_2_17_i686.manylinux2014_i686.whl", hash = "sha256:fb57870449dfcfac428afbb5a877829fcb0d6db9d9baa1148705739e9083880e"},
+    {file = "grpcio-1.66.2-cp39-cp39-manylinux_2_17_x86_64.manylinux2014_x86_64.whl", hash = "sha256:b672abf90a964bfde2d0ecbce30f2329a47498ba75ce6f4da35a2f4532b7acbc"},
+    {file = "grpcio-1.66.2-cp39-cp39-musllinux_1_1_i686.whl", hash = "sha256:ad2efdbe90c73b0434cbe64ed372e12414ad03c06262279b104a029d1889d13e"},
+    {file = "grpcio-1.66.2-cp39-cp39-musllinux_1_1_x86_64.whl", hash = "sha256:9c3a99c519f4638e700e9e3f83952e27e2ea10873eecd7935823dab0c1c9250e"},
+    {file = "grpcio-1.66.2-cp39-cp39-win32.whl", hash = "sha256:78fa51ebc2d9242c0fc5db0feecc57a9943303b46664ad89921f5079e2e4ada7"},
+    {file = "grpcio-1.66.2-cp39-cp39-win_amd64.whl", hash = "sha256:728bdf36a186e7f51da73be7f8d09457a03061be848718d0edf000e709418987"},
+    {file = "grpcio-1.66.2.tar.gz", hash = "sha256:563588c587b75c34b928bc428548e5b00ea38c46972181a4d8b75ba7e3f24231"},
+]
+
+[package.extras]
+protobuf = ["grpcio-tools (>=1.66.2)"]
 
 [[package]]
 name = "grpcio-status"
@@ -3993,13 +4028,13 @@ files = [
 
 [[package]]
 name = "idna"
-version = "3.9"
+version = "3.10"
 description = "Internationalized Domain Names in Applications (IDNA)"
 optional = false
 python-versions = ">=3.6"
 files = [
-    {file = "idna-3.9-py3-none-any.whl", hash = "sha256:69297d5da0cc9281c77efffb4e730254dd45943f45bbfb461de5991713989b1e"},
-    {file = "idna-3.9.tar.gz", hash = "sha256:e5c5dafde284f26e9e0f28f6ea2d6400abd5ca099864a67f576f3981c6476124"},
+    {file = "idna-3.10-py3-none-any.whl", hash = "sha256:946d195a0d259cbba61165e88e65941f16e9b36ea6ddb97f00452bae8b1287d3"},
+    {file = "idna-3.10.tar.gz", hash = "sha256:12f65c9b470abda6dc35cf8e63cc574b1c52b11df2c86030af0ac09b01b13ea9"},
 ]
 
 [package.extras]
@@ -4341,17 +4376,18 @@ files = [
 
 [[package]]
 name = "kombu"
-version = "5.4.1"
+version = "5.4.2"
 description = "Messaging library for Python."
 optional = false
 python-versions = ">=3.8"
 files = [
-    {file = "kombu-5.4.1-py3-none-any.whl", hash = "sha256:621d365f234e4c089596f3a2510f1ade07026efc28caca426161d8f458786cab"},
-    {file = "kombu-5.4.1.tar.gz", hash = "sha256:1c05178826dab811f8cab5b0a154d42a7a33d8bcdde9fa3d7b4582e43c3c03db"},
+    {file = "kombu-5.4.2-py3-none-any.whl", hash = "sha256:14212f5ccf022fc0a70453bb025a1dcc32782a588c49ea866884047d66e14763"},
+    {file = "kombu-5.4.2.tar.gz", hash = "sha256:eef572dd2fd9fc614b37580e3caeafdd5af46c1eff31e7fba89138cdb406f2cf"},
 ]
 
 [package.dependencies]
 amqp = ">=5.1.1,<6.0.0"
+tzdata = {version = "*", markers = "python_version >= \"3.9\""}
 vine = "5.1.0"
 
 [package.extras]
@@ -4373,17 +4409,18 @@ zookeeper = ["kazoo (>=2.8.0)"]
 
 [[package]]
 name = "kubernetes"
-version = "30.1.0"
+version = "31.0.0"
 description = "Kubernetes python client"
 optional = false
 python-versions = ">=3.6"
 files = [
-    {file = "kubernetes-30.1.0-py2.py3-none-any.whl", hash = "sha256:e212e8b7579031dd2e512168b617373bc1e03888d41ac4e04039240a292d478d"},
-    {file = "kubernetes-30.1.0.tar.gz", hash = "sha256:41e4c77af9f28e7a6c314e3bd06a8c6229ddd787cad684e0ab9f69b498e98ebc"},
+    {file = "kubernetes-31.0.0-py2.py3-none-any.whl", hash = "sha256:bf141e2d380c8520eada8b351f4e319ffee9636328c137aa432bc486ca1200e1"},
+    {file = "kubernetes-31.0.0.tar.gz", hash = "sha256:28945de906c8c259c1ebe62703b56a03b714049372196f854105afe4e6d014c0"},
 ]
 
 [package.dependencies]
 certifi = ">=14.05.14"
+durationpy = ">=0.7"
 google-auth = ">=1.0.1"
 oauthlib = ">=3.2.2"
 python-dateutil = ">=2.5.3"
@@ -4413,13 +4450,13 @@ six = "*"
 
 [[package]]
 name = "langfuse"
-version = "2.48.1"
+version = "2.51.2"
 description = "A client library for accessing langfuse"
 optional = false
 python-versions = "<4.0,>=3.8.1"
 files = [
-    {file = "langfuse-2.48.1-py3-none-any.whl", hash = "sha256:8661070b6d94ba1d7da92c054f3110b6ecf4489d6e8204a4080f934f3f49ebf2"},
-    {file = "langfuse-2.48.1.tar.gz", hash = "sha256:b8117d90babec6be1bc3303b42e0b71848531eae44118e6e0123d03e7961d0fc"},
+    {file = "langfuse-2.51.2-py3-none-any.whl", hash = "sha256:7aab94a9452cda4587a2cd4917e455da1afd7f8a2696688742130e2f2d23ca59"},
+    {file = "langfuse-2.51.2.tar.gz", hash = "sha256:0982b108ab4c02947f682e442b0796b7a73825d31eeace1771575f6454b8f79a"},
 ]
 
 [package.dependencies]
@@ -4438,13 +4475,13 @@ openai = ["openai (>=0.27.8)"]
 
 [[package]]
 name = "langsmith"
-version = "0.1.120"
+version = "0.1.129"
 description = "Client library to connect to the LangSmith LLM Tracing and Evaluation Platform."
 optional = false
 python-versions = "<4.0,>=3.8.1"
 files = [
-    {file = "langsmith-0.1.120-py3-none-any.whl", hash = "sha256:54d2785e301646c0988e0a69ebe4d976488c87b41928b358cb153b6ddd8db62b"},
-    {file = "langsmith-0.1.120.tar.gz", hash = "sha256:25499ca187b41bd89d784b272b97a8d76f60e0e21bdf20336e8a2aa6a9b23ac9"},
+    {file = "langsmith-0.1.129-py3-none-any.whl", hash = "sha256:31393fbbb17d6be5b99b9b22d530450094fab23c6c37281a6a6efb2143d05347"},
+    {file = "langsmith-0.1.129.tar.gz", hash = "sha256:6c3ba66471bef41b9f87da247cc0b493268b3f54656f73648a256a205261b6a0"},
 ]
 
 [package.dependencies]
@@ -4948,95 +4985,116 @@ tqdm = "*"
 
 [[package]]
 name = "mmh3"
-version = "4.1.0"
+version = "5.0.1"
 description = "Python extension for MurmurHash (MurmurHash3), a set of fast and robust hash functions."
 optional = false
-python-versions = "*"
+python-versions = ">=3.8"
 files = [
-    {file = "mmh3-4.1.0-cp310-cp310-macosx_10_9_universal2.whl", hash = "sha256:be5ac76a8b0cd8095784e51e4c1c9c318c19edcd1709a06eb14979c8d850c31a"},
-    {file = "mmh3-4.1.0-cp310-cp310-macosx_10_9_x86_64.whl", hash = "sha256:98a49121afdfab67cd80e912b36404139d7deceb6773a83620137aaa0da5714c"},
-    {file = "mmh3-4.1.0-cp310-cp310-macosx_11_0_arm64.whl", hash = "sha256:5259ac0535874366e7d1a5423ef746e0d36a9e3c14509ce6511614bdc5a7ef5b"},
-    {file = "mmh3-4.1.0-cp310-cp310-manylinux_2_17_aarch64.manylinux2014_aarch64.whl", hash = "sha256:c5950827ca0453a2be357696da509ab39646044e3fa15cad364eb65d78797437"},
-    {file = "mmh3-4.1.0-cp310-cp310-manylinux_2_17_ppc64le.manylinux2014_ppc64le.whl", hash = "sha256:1dd0f652ae99585b9dd26de458e5f08571522f0402155809fd1dc8852a613a39"},
-    {file = "mmh3-4.1.0-cp310-cp310-manylinux_2_17_s390x.manylinux2014_s390x.whl", hash = "sha256:99d25548070942fab1e4a6f04d1626d67e66d0b81ed6571ecfca511f3edf07e6"},
-    {file = "mmh3-4.1.0-cp310-cp310-manylinux_2_5_i686.manylinux1_i686.manylinux_2_17_i686.manylinux2014_i686.whl", hash = "sha256:53db8d9bad3cb66c8f35cbc894f336273f63489ce4ac416634932e3cbe79eb5b"},
-    {file = "mmh3-4.1.0-cp310-cp310-manylinux_2_5_x86_64.manylinux1_x86_64.manylinux_2_17_x86_64.manylinux2014_x86_64.whl", hash = "sha256:75da0f615eb55295a437264cc0b736753f830b09d102aa4c2a7d719bc445ec05"},
-    {file = "mmh3-4.1.0-cp310-cp310-musllinux_1_1_aarch64.whl", hash = "sha256:b926b07fd678ea84b3a2afc1fa22ce50aeb627839c44382f3d0291e945621e1a"},
-    {file = "mmh3-4.1.0-cp310-cp310-musllinux_1_1_i686.whl", hash = "sha256:c5b053334f9b0af8559d6da9dc72cef0a65b325ebb3e630c680012323c950bb6"},
-    {file = "mmh3-4.1.0-cp310-cp310-musllinux_1_1_ppc64le.whl", hash = "sha256:5bf33dc43cd6de2cb86e0aa73a1cc6530f557854bbbe5d59f41ef6de2e353d7b"},
-    {file = "mmh3-4.1.0-cp310-cp310-musllinux_1_1_s390x.whl", hash = "sha256:fa7eacd2b830727ba3dd65a365bed8a5c992ecd0c8348cf39a05cc77d22f4970"},
-    {file = "mmh3-4.1.0-cp310-cp310-musllinux_1_1_x86_64.whl", hash = "sha256:42dfd6742b9e3eec599f85270617debfa0bbb913c545bb980c8a4fa7b2d047da"},
-    {file = "mmh3-4.1.0-cp310-cp310-win32.whl", hash = "sha256:2974ad343f0d39dcc88e93ee6afa96cedc35a9883bc067febd7ff736e207fa47"},
-    {file = "mmh3-4.1.0-cp310-cp310-win_amd64.whl", hash = "sha256:74699a8984ded645c1a24d6078351a056f5a5f1fe5838870412a68ac5e28d865"},
-    {file = "mmh3-4.1.0-cp310-cp310-win_arm64.whl", hash = "sha256:f0dc874cedc23d46fc488a987faa6ad08ffa79e44fb08e3cd4d4cf2877c00a00"},
-    {file = "mmh3-4.1.0-cp311-cp311-macosx_10_9_universal2.whl", hash = "sha256:3280a463855b0eae64b681cd5b9ddd9464b73f81151e87bb7c91a811d25619e6"},
-    {file = "mmh3-4.1.0-cp311-cp311-macosx_10_9_x86_64.whl", hash = "sha256:97ac57c6c3301769e757d444fa7c973ceb002cb66534b39cbab5e38de61cd896"},
-    {file = "mmh3-4.1.0-cp311-cp311-macosx_11_0_arm64.whl", hash = "sha256:a7b6502cdb4dbd880244818ab363c8770a48cdccecf6d729ade0241b736b5ec0"},
-    {file = "mmh3-4.1.0-cp311-cp311-manylinux_2_17_aarch64.manylinux2014_aarch64.whl", hash = "sha256:52ba2da04671a9621580ddabf72f06f0e72c1c9c3b7b608849b58b11080d8f14"},
-    {file = "mmh3-4.1.0-cp311-cp311-manylinux_2_17_ppc64le.manylinux2014_ppc64le.whl", hash = "sha256:5a5fef4c4ecc782e6e43fbeab09cff1bac82c998a1773d3a5ee6a3605cde343e"},
-    {file = "mmh3-4.1.0-cp311-cp311-manylinux_2_17_s390x.manylinux2014_s390x.whl", hash = "sha256:5135358a7e00991f73b88cdc8eda5203bf9de22120d10a834c5761dbeb07dd13"},
-    {file = "mmh3-4.1.0-cp311-cp311-manylinux_2_5_i686.manylinux1_i686.manylinux_2_17_i686.manylinux2014_i686.whl", hash = "sha256:cff9ae76a54f7c6fe0167c9c4028c12c1f6de52d68a31d11b6790bb2ae685560"},
-    {file = "mmh3-4.1.0-cp311-cp311-manylinux_2_5_x86_64.manylinux1_x86_64.manylinux_2_17_x86_64.manylinux2014_x86_64.whl", hash = "sha256:f6f02576a4d106d7830ca90278868bf0983554dd69183b7bbe09f2fcd51cf54f"},
-    {file = "mmh3-4.1.0-cp311-cp311-musllinux_1_1_aarch64.whl", hash = "sha256:073d57425a23721730d3ff5485e2da489dd3c90b04e86243dd7211f889898106"},
-    {file = "mmh3-4.1.0-cp311-cp311-musllinux_1_1_i686.whl", hash = "sha256:71e32ddec7f573a1a0feb8d2cf2af474c50ec21e7a8263026e8d3b4b629805db"},
-    {file = "mmh3-4.1.0-cp311-cp311-musllinux_1_1_ppc64le.whl", hash = "sha256:7cbb20b29d57e76a58b40fd8b13a9130db495a12d678d651b459bf61c0714cea"},
-    {file = "mmh3-4.1.0-cp311-cp311-musllinux_1_1_s390x.whl", hash = "sha256:a42ad267e131d7847076bb7e31050f6c4378cd38e8f1bf7a0edd32f30224d5c9"},
-    {file = "mmh3-4.1.0-cp311-cp311-musllinux_1_1_x86_64.whl", hash = "sha256:4a013979fc9390abadc445ea2527426a0e7a4495c19b74589204f9b71bcaafeb"},
-    {file = "mmh3-4.1.0-cp311-cp311-win32.whl", hash = "sha256:1d3b1cdad7c71b7b88966301789a478af142bddcb3a2bee563f7a7d40519a00f"},
-    {file = "mmh3-4.1.0-cp311-cp311-win_amd64.whl", hash = "sha256:0dc6dc32eb03727467da8e17deffe004fbb65e8b5ee2b502d36250d7a3f4e2ec"},
-    {file = "mmh3-4.1.0-cp311-cp311-win_arm64.whl", hash = "sha256:9ae3a5c1b32dda121c7dc26f9597ef7b01b4c56a98319a7fe86c35b8bc459ae6"},
-    {file = "mmh3-4.1.0-cp312-cp312-macosx_10_9_universal2.whl", hash = "sha256:0033d60c7939168ef65ddc396611077a7268bde024f2c23bdc283a19123f9e9c"},
-    {file = "mmh3-4.1.0-cp312-cp312-macosx_10_9_x86_64.whl", hash = "sha256:d6af3e2287644b2b08b5924ed3a88c97b87b44ad08e79ca9f93d3470a54a41c5"},
-    {file = "mmh3-4.1.0-cp312-cp312-macosx_11_0_arm64.whl", hash = "sha256:d82eb4defa245e02bb0b0dc4f1e7ee284f8d212633389c91f7fba99ba993f0a2"},
-    {file = "mmh3-4.1.0-cp312-cp312-manylinux_2_17_aarch64.manylinux2014_aarch64.whl", hash = "sha256:ba245e94b8d54765e14c2d7b6214e832557e7856d5183bc522e17884cab2f45d"},
-    {file = "mmh3-4.1.0-cp312-cp312-manylinux_2_17_ppc64le.manylinux2014_ppc64le.whl", hash = "sha256:bb04e2feeabaad6231e89cd43b3d01a4403579aa792c9ab6fdeef45cc58d4ec0"},
-    {file = "mmh3-4.1.0-cp312-cp312-manylinux_2_17_s390x.manylinux2014_s390x.whl", hash = "sha256:1e3b1a27def545ce11e36158ba5d5390cdbc300cfe456a942cc89d649cf7e3b2"},
-    {file = "mmh3-4.1.0-cp312-cp312-manylinux_2_5_i686.manylinux1_i686.manylinux_2_17_i686.manylinux2014_i686.whl", hash = "sha256:ce0ab79ff736d7044e5e9b3bfe73958a55f79a4ae672e6213e92492ad5e734d5"},
-    {file = "mmh3-4.1.0-cp312-cp312-manylinux_2_5_x86_64.manylinux1_x86_64.manylinux_2_17_x86_64.manylinux2014_x86_64.whl", hash = "sha256:3b02268be6e0a8eeb8a924d7db85f28e47344f35c438c1e149878bb1c47b1cd3"},
-    {file = "mmh3-4.1.0-cp312-cp312-musllinux_1_1_aarch64.whl", hash = "sha256:deb887f5fcdaf57cf646b1e062d56b06ef2f23421c80885fce18b37143cba828"},
-    {file = "mmh3-4.1.0-cp312-cp312-musllinux_1_1_i686.whl", hash = "sha256:99dd564e9e2b512eb117bd0cbf0f79a50c45d961c2a02402787d581cec5448d5"},
-    {file = "mmh3-4.1.0-cp312-cp312-musllinux_1_1_ppc64le.whl", hash = "sha256:08373082dfaa38fe97aa78753d1efd21a1969e51079056ff552e687764eafdfe"},
-    {file = "mmh3-4.1.0-cp312-cp312-musllinux_1_1_s390x.whl", hash = "sha256:54b9c6a2ea571b714e4fe28d3e4e2db37abfd03c787a58074ea21ee9a8fd1740"},
-    {file = "mmh3-4.1.0-cp312-cp312-musllinux_1_1_x86_64.whl", hash = "sha256:a7b1edf24c69e3513f879722b97ca85e52f9032f24a52284746877f6a7304086"},
-    {file = "mmh3-4.1.0-cp312-cp312-win32.whl", hash = "sha256:411da64b951f635e1e2284b71d81a5a83580cea24994b328f8910d40bed67276"},
-    {file = "mmh3-4.1.0-cp312-cp312-win_amd64.whl", hash = "sha256:bebc3ecb6ba18292e3d40c8712482b4477abd6981c2ebf0e60869bd90f8ac3a9"},
-    {file = "mmh3-4.1.0-cp312-cp312-win_arm64.whl", hash = "sha256:168473dd608ade6a8d2ba069600b35199a9af837d96177d3088ca91f2b3798e3"},
-    {file = "mmh3-4.1.0-cp38-cp38-macosx_10_9_universal2.whl", hash = "sha256:372f4b7e1dcde175507640679a2a8790185bb71f3640fc28a4690f73da986a3b"},
-    {file = "mmh3-4.1.0-cp38-cp38-macosx_10_9_x86_64.whl", hash = "sha256:438584b97f6fe13e944faf590c90fc127682b57ae969f73334040d9fa1c7ffa5"},
-    {file = "mmh3-4.1.0-cp38-cp38-macosx_11_0_arm64.whl", hash = "sha256:6e27931b232fc676675fac8641c6ec6b596daa64d82170e8597f5a5b8bdcd3b6"},
-    {file = "mmh3-4.1.0-cp38-cp38-manylinux_2_17_aarch64.manylinux2014_aarch64.whl", hash = "sha256:571a92bad859d7b0330e47cfd1850b76c39b615a8d8e7aa5853c1f971fd0c4b1"},
-    {file = "mmh3-4.1.0-cp38-cp38-manylinux_2_17_ppc64le.manylinux2014_ppc64le.whl", hash = "sha256:4a69d6afe3190fa08f9e3a58e5145549f71f1f3fff27bd0800313426929c7068"},
-    {file = "mmh3-4.1.0-cp38-cp38-manylinux_2_17_s390x.manylinux2014_s390x.whl", hash = "sha256:afb127be0be946b7630220908dbea0cee0d9d3c583fa9114a07156f98566dc28"},
-    {file = "mmh3-4.1.0-cp38-cp38-manylinux_2_5_i686.manylinux1_i686.manylinux_2_17_i686.manylinux2014_i686.whl", hash = "sha256:940d86522f36348ef1a494cbf7248ab3f4a1638b84b59e6c9e90408bd11ad729"},
-    {file = "mmh3-4.1.0-cp38-cp38-manylinux_2_5_x86_64.manylinux1_x86_64.manylinux_2_17_x86_64.manylinux2014_x86_64.whl", hash = "sha256:b3dcccc4935686619a8e3d1f7b6e97e3bd89a4a796247930ee97d35ea1a39341"},
-    {file = "mmh3-4.1.0-cp38-cp38-musllinux_1_1_aarch64.whl", hash = "sha256:01bb9b90d61854dfc2407c5e5192bfb47222d74f29d140cb2dd2a69f2353f7cc"},
-    {file = "mmh3-4.1.0-cp38-cp38-musllinux_1_1_i686.whl", hash = "sha256:bcb1b8b951a2c0b0fb8a5426c62a22557e2ffc52539e0a7cc46eb667b5d606a9"},
-    {file = "mmh3-4.1.0-cp38-cp38-musllinux_1_1_ppc64le.whl", hash = "sha256:6477a05d5e5ab3168e82e8b106e316210ac954134f46ec529356607900aea82a"},
-    {file = "mmh3-4.1.0-cp38-cp38-musllinux_1_1_s390x.whl", hash = "sha256:da5892287e5bea6977364b15712a2573c16d134bc5fdcdd4cf460006cf849278"},
-    {file = "mmh3-4.1.0-cp38-cp38-musllinux_1_1_x86_64.whl", hash = "sha256:99180d7fd2327a6fffbaff270f760576839dc6ee66d045fa3a450f3490fda7f5"},
-    {file = "mmh3-4.1.0-cp38-cp38-win32.whl", hash = "sha256:9b0d4f3949913a9f9a8fb1bb4cc6ecd52879730aab5ff8c5a3d8f5b593594b73"},
-    {file = "mmh3-4.1.0-cp38-cp38-win_amd64.whl", hash = "sha256:598c352da1d945108aee0c3c3cfdd0e9b3edef74108f53b49d481d3990402169"},
-    {file = "mmh3-4.1.0-cp39-cp39-macosx_10_9_universal2.whl", hash = "sha256:475d6d1445dd080f18f0f766277e1237fa2914e5fe3307a3b2a3044f30892103"},
-    {file = "mmh3-4.1.0-cp39-cp39-macosx_10_9_x86_64.whl", hash = "sha256:5ca07c41e6a2880991431ac717c2a049056fff497651a76e26fc22224e8b5732"},
-    {file = "mmh3-4.1.0-cp39-cp39-macosx_11_0_arm64.whl", hash = "sha256:0ebe052fef4bbe30c0548d12ee46d09f1b69035ca5208a7075e55adfe091be44"},
-    {file = "mmh3-4.1.0-cp39-cp39-manylinux_2_17_aarch64.manylinux2014_aarch64.whl", hash = "sha256:eaefd42e85afb70f2b855a011f7b4d8a3c7e19c3f2681fa13118e4d8627378c5"},
-    {file = "mmh3-4.1.0-cp39-cp39-manylinux_2_17_ppc64le.manylinux2014_ppc64le.whl", hash = "sha256:ac0ae43caae5a47afe1b63a1ae3f0986dde54b5fb2d6c29786adbfb8edc9edfb"},
-    {file = "mmh3-4.1.0-cp39-cp39-manylinux_2_17_s390x.manylinux2014_s390x.whl", hash = "sha256:6218666f74c8c013c221e7f5f8a693ac9cf68e5ac9a03f2373b32d77c48904de"},
-    {file = "mmh3-4.1.0-cp39-cp39-manylinux_2_5_i686.manylinux1_i686.manylinux_2_17_i686.manylinux2014_i686.whl", hash = "sha256:ac59294a536ba447b5037f62d8367d7d93b696f80671c2c45645fa9f1109413c"},
-    {file = "mmh3-4.1.0-cp39-cp39-manylinux_2_5_x86_64.manylinux1_x86_64.manylinux_2_17_x86_64.manylinux2014_x86_64.whl", hash = "sha256:086844830fcd1e5c84fec7017ea1ee8491487cfc877847d96f86f68881569d2e"},
-    {file = "mmh3-4.1.0-cp39-cp39-musllinux_1_1_aarch64.whl", hash = "sha256:e42b38fad664f56f77f6fbca22d08450f2464baa68acdbf24841bf900eb98e87"},
-    {file = "mmh3-4.1.0-cp39-cp39-musllinux_1_1_i686.whl", hash = "sha256:d08b790a63a9a1cde3b5d7d733ed97d4eb884bfbc92f075a091652d6bfd7709a"},
-    {file = "mmh3-4.1.0-cp39-cp39-musllinux_1_1_ppc64le.whl", hash = "sha256:73ea4cc55e8aea28c86799ecacebca09e5f86500414870a8abaedfcbaf74d288"},
-    {file = "mmh3-4.1.0-cp39-cp39-musllinux_1_1_s390x.whl", hash = "sha256:f90938ff137130e47bcec8dc1f4ceb02f10178c766e2ef58a9f657ff1f62d124"},
-    {file = "mmh3-4.1.0-cp39-cp39-musllinux_1_1_x86_64.whl", hash = "sha256:aa1f13e94b8631c8cd53259250556edcf1de71738936b60febba95750d9632bd"},
-    {file = "mmh3-4.1.0-cp39-cp39-win32.whl", hash = "sha256:a3b680b471c181490cf82da2142029edb4298e1bdfcb67c76922dedef789868d"},
-    {file = "mmh3-4.1.0-cp39-cp39-win_amd64.whl", hash = "sha256:fefef92e9c544a8dbc08f77a8d1b6d48006a750c4375bbcd5ff8199d761e263b"},
-    {file = "mmh3-4.1.0-cp39-cp39-win_arm64.whl", hash = "sha256:8e2c1f6a2b41723a4f82bd5a762a777836d29d664fc0095f17910bea0adfd4a6"},
-    {file = "mmh3-4.1.0.tar.gz", hash = "sha256:a1cf25348b9acd229dda464a094d6170f47d2850a1fcb762a3b6172d2ce6ca4a"},
-]
-
-[package.extras]
-test = ["mypy (>=1.0)", "pytest (>=7.0.0)"]
+    {file = "mmh3-5.0.1-cp310-cp310-macosx_10_9_universal2.whl", hash = "sha256:f0a4b4bf05778ed77d820d6e7d0e9bd6beb0c01af10e1ce9233f5d2f814fcafa"},
+    {file = "mmh3-5.0.1-cp310-cp310-macosx_10_9_x86_64.whl", hash = "sha256:ac7a391039aeab95810c2d020b69a94eb6b4b37d4e2374831e92db3a0cdf71c6"},
+    {file = "mmh3-5.0.1-cp310-cp310-macosx_11_0_arm64.whl", hash = "sha256:3a2583b5521ca49756d8d8bceba80627a9cc295f255dcab4e3df7ccc2f09679a"},
+    {file = "mmh3-5.0.1-cp310-cp310-manylinux_2_17_aarch64.manylinux2014_aarch64.whl", hash = "sha256:081a8423fe53c1ac94f87165f3e4c500125d343410c1a0c5f1703e898a3ef038"},
+    {file = "mmh3-5.0.1-cp310-cp310-manylinux_2_17_ppc64le.manylinux2014_ppc64le.whl", hash = "sha256:b8b4d72713799755dc8954a7d36d5c20a6c8de7b233c82404d122c7c7c1707cc"},
+    {file = "mmh3-5.0.1-cp310-cp310-manylinux_2_17_s390x.manylinux2014_s390x.whl", hash = "sha256:389a6fd51efc76d3182d36ec306448559c1244f11227d2bb771bdd0e6cc91321"},
+    {file = "mmh3-5.0.1-cp310-cp310-manylinux_2_5_i686.manylinux1_i686.manylinux_2_17_i686.manylinux2014_i686.whl", hash = "sha256:39f4128edaa074bff721b1d31a72508cba4d2887ee7867f22082e1fe9d4edea0"},
+    {file = "mmh3-5.0.1-cp310-cp310-manylinux_2_5_x86_64.manylinux1_x86_64.manylinux_2_17_x86_64.manylinux2014_x86_64.whl", hash = "sha256:1d5d23a94d91aabba3386b3769048d5f4210fdfef80393fece2f34ba5a7b466c"},
+    {file = "mmh3-5.0.1-cp310-cp310-musllinux_1_2_aarch64.whl", hash = "sha256:16347d038361f8b8f24fd2b7ef378c9b68ddee9f7706e46269b6e0d322814713"},
+    {file = "mmh3-5.0.1-cp310-cp310-musllinux_1_2_i686.whl", hash = "sha256:6e299408565af7d61f2d20a5ffdd77cf2ed902460fe4e6726839d59ba4b72316"},
+    {file = "mmh3-5.0.1-cp310-cp310-musllinux_1_2_ppc64le.whl", hash = "sha256:42050af21ddfc5445ee5a66e73a8fc758c71790305e3ee9e4a85a8e69e810f94"},
+    {file = "mmh3-5.0.1-cp310-cp310-musllinux_1_2_s390x.whl", hash = "sha256:2ae9b1f5ef27ec54659920f0404b7ceb39966e28867c461bfe83a05e8d18ddb0"},
+    {file = "mmh3-5.0.1-cp310-cp310-musllinux_1_2_x86_64.whl", hash = "sha256:50c2495a02045f3047d71d4ae9cdd7a15efc0bcbb7ff17a18346834a8e2d1d19"},
+    {file = "mmh3-5.0.1-cp310-cp310-win32.whl", hash = "sha256:c028fa77cddf351ca13b4a56d43c1775652cde0764cadb39120b68f02a23ecf6"},
+    {file = "mmh3-5.0.1-cp310-cp310-win_amd64.whl", hash = "sha256:c5e741e421ec14400c4aae30890515c201f518403bdef29ae1e00d375bb4bbb5"},
+    {file = "mmh3-5.0.1-cp310-cp310-win_arm64.whl", hash = "sha256:b17156d56fabc73dbf41bca677ceb6faed435cc8544f6566d72ea77d8a17e9d0"},
+    {file = "mmh3-5.0.1-cp311-cp311-macosx_10_9_universal2.whl", hash = "sha256:9a6d5a9b1b923f1643559ba1fc0bf7a5076c90cbb558878d3bf3641ce458f25d"},
+    {file = "mmh3-5.0.1-cp311-cp311-macosx_10_9_x86_64.whl", hash = "sha256:3349b968be555f7334bbcce839da98f50e1e80b1c615d8e2aa847ea4a964a012"},
+    {file = "mmh3-5.0.1-cp311-cp311-macosx_11_0_arm64.whl", hash = "sha256:1bd3c94b110e55db02ab9b605029f48a2f7f677c6e58c09d44e42402d438b7e1"},
+    {file = "mmh3-5.0.1-cp311-cp311-manylinux_2_17_aarch64.manylinux2014_aarch64.whl", hash = "sha256:d47ba84d48608f79adbb10bb09986b6dc33eeda5c2d1bd75d00820081b73bde9"},
+    {file = "mmh3-5.0.1-cp311-cp311-manylinux_2_17_ppc64le.manylinux2014_ppc64le.whl", hash = "sha256:c0217987a8b8525c8d9170f66d036dec4ab45cfbd53d47e8d76125791ceb155e"},
+    {file = "mmh3-5.0.1-cp311-cp311-manylinux_2_17_s390x.manylinux2014_s390x.whl", hash = "sha256:b2797063a34e78d1b61639a98b0edec1c856fa86ab80c7ec859f1796d10ba429"},
+    {file = "mmh3-5.0.1-cp311-cp311-manylinux_2_5_i686.manylinux1_i686.manylinux_2_17_i686.manylinux2014_i686.whl", hash = "sha256:8bba16340adcbd47853a2fbe5afdb397549e8f2e79324ff1dced69a3f8afe7c3"},
+    {file = "mmh3-5.0.1-cp311-cp311-manylinux_2_5_x86_64.manylinux1_x86_64.manylinux_2_17_x86_64.manylinux2014_x86_64.whl", hash = "sha256:282797957c9f60b51b9d768a602c25f579420cc9af46feb77d457a27823d270a"},
+    {file = "mmh3-5.0.1-cp311-cp311-musllinux_1_2_aarch64.whl", hash = "sha256:e4fb670c29e63f954f9e7a2cdcd57b36a854c2538f579ef62681ccbaa1de2b69"},
+    {file = "mmh3-5.0.1-cp311-cp311-musllinux_1_2_i686.whl", hash = "sha256:8ee7d85438dc6aff328e19ab052086a3c29e8a9b632998a49e5c4b0034e9e8d6"},
+    {file = "mmh3-5.0.1-cp311-cp311-musllinux_1_2_ppc64le.whl", hash = "sha256:b7fb5db231f3092444bc13901e6a8d299667126b00636ffbad4a7b45e1051e2f"},
+    {file = "mmh3-5.0.1-cp311-cp311-musllinux_1_2_s390x.whl", hash = "sha256:c100dd441703da5ec136b1d9003ed4a041d8a1136234c9acd887499796df6ad8"},
+    {file = "mmh3-5.0.1-cp311-cp311-musllinux_1_2_x86_64.whl", hash = "sha256:71f3b765138260fd7a7a2dba0ea5727dabcd18c1f80323c9cfef97a7e86e01d0"},
+    {file = "mmh3-5.0.1-cp311-cp311-win32.whl", hash = "sha256:9a76518336247fd17689ce3ae5b16883fd86a490947d46a0193d47fb913e26e3"},
+    {file = "mmh3-5.0.1-cp311-cp311-win_amd64.whl", hash = "sha256:336bc4df2e44271f1c302d289cc3d78bd52d3eed8d306c7e4bff8361a12bf148"},
+    {file = "mmh3-5.0.1-cp311-cp311-win_arm64.whl", hash = "sha256:af6522722fbbc5999aa66f7244d0986767a46f1fb05accc5200f75b72428a508"},
+    {file = "mmh3-5.0.1-cp312-cp312-macosx_10_13_universal2.whl", hash = "sha256:f2730bb263ed9c388e8860438b057a53e3cc701134a6ea140f90443c4c11aa40"},
+    {file = "mmh3-5.0.1-cp312-cp312-macosx_10_13_x86_64.whl", hash = "sha256:6246927bc293f6d56724536400b85fb85f5be26101fa77d5f97dd5e2a4c69bf2"},
+    {file = "mmh3-5.0.1-cp312-cp312-macosx_11_0_arm64.whl", hash = "sha256:fbca322519a6e6e25b6abf43e940e1667cf8ea12510e07fb4919b48a0cd1c411"},
+    {file = "mmh3-5.0.1-cp312-cp312-manylinux_2_17_aarch64.manylinux2014_aarch64.whl", hash = "sha256:eae8c19903ed8a1724ad9e67e86f15d198a7a1271a4f9be83d47e38f312ed672"},
+    {file = "mmh3-5.0.1-cp312-cp312-manylinux_2_17_ppc64le.manylinux2014_ppc64le.whl", hash = "sha256:a09fd6cc72c07c0c07c3357714234b646d78052487c4a3bd5f7f6e08408cff60"},
+    {file = "mmh3-5.0.1-cp312-cp312-manylinux_2_17_s390x.manylinux2014_s390x.whl", hash = "sha256:2ff8551fee7ae3b11c5d986b6347ade0dccaadd4670ffdb2b944dee120ffcc84"},
+    {file = "mmh3-5.0.1-cp312-cp312-manylinux_2_5_i686.manylinux1_i686.manylinux_2_17_i686.manylinux2014_i686.whl", hash = "sha256:e39694c73a5a20c8bf36dfd8676ed351e5234d55751ba4f7562d85449b21ef3f"},
+    {file = "mmh3-5.0.1-cp312-cp312-manylinux_2_5_x86_64.manylinux1_x86_64.manylinux_2_17_x86_64.manylinux2014_x86_64.whl", hash = "sha256:eba6001989a92f72a89c7cf382fda831678bd780707a66b4f8ca90239fdf2123"},
+    {file = "mmh3-5.0.1-cp312-cp312-musllinux_1_2_aarch64.whl", hash = "sha256:0771f90c9911811cc606a5c7b7b58f33501c9ee896ed68a6ac22c7d55878ecc0"},
+    {file = "mmh3-5.0.1-cp312-cp312-musllinux_1_2_i686.whl", hash = "sha256:09b31ed0c0c0920363e96641fac4efde65b1ab62b8df86293142f35a254e72b4"},
+    {file = "mmh3-5.0.1-cp312-cp312-musllinux_1_2_ppc64le.whl", hash = "sha256:5cf4a8deda0235312db12075331cb417c4ba163770edfe789bde71d08a24b692"},
+    {file = "mmh3-5.0.1-cp312-cp312-musllinux_1_2_s390x.whl", hash = "sha256:41f7090a95185ef20ac018581a99337f0cbc84a2135171ee3290a9c0d9519585"},
+    {file = "mmh3-5.0.1-cp312-cp312-musllinux_1_2_x86_64.whl", hash = "sha256:b97b5b368fb7ff22194ec5854f5b12d8de9ab67a0f304728c7f16e5d12135b76"},
+    {file = "mmh3-5.0.1-cp312-cp312-win32.whl", hash = "sha256:842516acf04da546f94fad52db125ee619ccbdcada179da51c326a22c4578cb9"},
+    {file = "mmh3-5.0.1-cp312-cp312-win_amd64.whl", hash = "sha256:d963be0dbfd9fca209c17172f6110787ebf78934af25e3694fe2ba40e55c1e2b"},
+    {file = "mmh3-5.0.1-cp312-cp312-win_arm64.whl", hash = "sha256:a5da292ceeed8ce8e32b68847261a462d30fd7b478c3f55daae841404f433c15"},
+    {file = "mmh3-5.0.1-cp313-cp313-macosx_10_13_universal2.whl", hash = "sha256:673e3f1c8d4231d6fb0271484ee34cb7146a6499fc0df80788adb56fd76842da"},
+    {file = "mmh3-5.0.1-cp313-cp313-macosx_10_13_x86_64.whl", hash = "sha256:f795a306bd16a52ad578b663462cc8e95500b3925d64118ae63453485d67282b"},
+    {file = "mmh3-5.0.1-cp313-cp313-macosx_11_0_arm64.whl", hash = "sha256:5ed57a5e28e502a1d60436cc25c76c3a5ba57545f250f2969af231dc1221e0a5"},
+    {file = "mmh3-5.0.1-cp313-cp313-manylinux_2_17_aarch64.manylinux2014_aarch64.whl", hash = "sha256:632c28e7612e909dbb6cbe2fe496201ada4695b7715584005689c5dc038e59ad"},
+    {file = "mmh3-5.0.1-cp313-cp313-manylinux_2_17_ppc64le.manylinux2014_ppc64le.whl", hash = "sha256:53fd6bd525a5985e391c43384672d9d6b317fcb36726447347c7fc75bfed34ec"},
+    {file = "mmh3-5.0.1-cp313-cp313-manylinux_2_17_s390x.manylinux2014_s390x.whl", hash = "sha256:dceacf6b0b961a0e499836af3aa62d60633265607aef551b2a3e3c48cdaa5edd"},
+    {file = "mmh3-5.0.1-cp313-cp313-manylinux_2_5_i686.manylinux1_i686.manylinux_2_17_i686.manylinux2014_i686.whl", hash = "sha256:8f0738d478fdfb5d920f6aff5452c78f2c35b0eff72caa2a97dfe38e82f93da2"},
+    {file = "mmh3-5.0.1-cp313-cp313-manylinux_2_5_x86_64.manylinux1_x86_64.manylinux_2_17_x86_64.manylinux2014_x86_64.whl", hash = "sha256:8e70285e7391ab88b872e5bef632bad16b9d99a6d3ca0590656a4753d55988af"},
+    {file = "mmh3-5.0.1-cp313-cp313-musllinux_1_2_aarch64.whl", hash = "sha256:27e5fc6360aa6b828546a4318da1a7da6bf6e5474ccb053c3a6aa8ef19ff97bd"},
+    {file = "mmh3-5.0.1-cp313-cp313-musllinux_1_2_i686.whl", hash = "sha256:7989530c3c1e2c17bf5a0ec2bba09fd19819078ba90beedabb1c3885f5040b0d"},
+    {file = "mmh3-5.0.1-cp313-cp313-musllinux_1_2_ppc64le.whl", hash = "sha256:cdad7bee649950da7ecd3cbbbd12fb81f1161072ecbdb5acfa0018338c5cb9cf"},
+    {file = "mmh3-5.0.1-cp313-cp313-musllinux_1_2_s390x.whl", hash = "sha256:e143b8f184c1bb58cecd85ab4a4fd6dc65a2d71aee74157392c3fddac2a4a331"},
+    {file = "mmh3-5.0.1-cp313-cp313-musllinux_1_2_x86_64.whl", hash = "sha256:e5eb12e886f3646dd636f16b76eb23fc0c27e8ff3c1ae73d4391e50ef60b40f6"},
+    {file = "mmh3-5.0.1-cp313-cp313-win32.whl", hash = "sha256:16e6dddfa98e1c2d021268e72c78951234186deb4df6630e984ac82df63d0a5d"},
+    {file = "mmh3-5.0.1-cp313-cp313-win_amd64.whl", hash = "sha256:d3ffb792d70b8c4a2382af3598dad6ae0c5bd9cee5b7ffcc99aa2f5fd2c1bf70"},
+    {file = "mmh3-5.0.1-cp313-cp313-win_arm64.whl", hash = "sha256:122fa9ec148383f9124292962bda745f192b47bfd470b2af5fe7bb3982b17896"},
+    {file = "mmh3-5.0.1-cp38-cp38-macosx_10_9_universal2.whl", hash = "sha256:b12bad8c75e6ff5d67319794fb6a5e8c713826c818d47f850ad08b4aa06960c6"},
+    {file = "mmh3-5.0.1-cp38-cp38-macosx_10_9_x86_64.whl", hash = "sha256:e5bbb066538c1048d542246fc347bb7994bdda29a3aea61c22f9f8b57111ce69"},
+    {file = "mmh3-5.0.1-cp38-cp38-macosx_11_0_arm64.whl", hash = "sha256:eee6134273f64e2a106827cc8fd77e70cc7239a285006fc6ab4977d59b015af2"},
+    {file = "mmh3-5.0.1-cp38-cp38-manylinux_2_17_aarch64.manylinux2014_aarch64.whl", hash = "sha256:d04d9aa19d48e4c7bbec9cabc2c4dccc6ff3b2402f856d5bf0de03e10f167b5b"},
+    {file = "mmh3-5.0.1-cp38-cp38-manylinux_2_17_ppc64le.manylinux2014_ppc64le.whl", hash = "sha256:79f37da1eed034d06567a69a7988456345c7f29e49192831c3975b464493b16e"},
+    {file = "mmh3-5.0.1-cp38-cp38-manylinux_2_17_s390x.manylinux2014_s390x.whl", hash = "sha256:242f77666743337aa828a2bf2da71b6ba79623ee7f93edb11e009f69237c8561"},
+    {file = "mmh3-5.0.1-cp38-cp38-manylinux_2_5_i686.manylinux1_i686.manylinux_2_17_i686.manylinux2014_i686.whl", hash = "sha256:ffd943fff690463945f6441a2465555b3146deaadf6a5e88f2590d14c655d71b"},
+    {file = "mmh3-5.0.1-cp38-cp38-manylinux_2_5_x86_64.manylinux1_x86_64.manylinux_2_17_x86_64.manylinux2014_x86_64.whl", hash = "sha256:565b15f8d7df43acb791ff5a360795c20bfa68bca8b352509e0fbabd06cc48cd"},
+    {file = "mmh3-5.0.1-cp38-cp38-musllinux_1_2_aarch64.whl", hash = "sha256:fc6aafb867c2030df98ac7760ff76b500359252867985f357bd387739f3d5287"},
+    {file = "mmh3-5.0.1-cp38-cp38-musllinux_1_2_i686.whl", hash = "sha256:32898170644d45aa27c974ab0d067809c066205110f5c6d09f47d9ece6978bfe"},
+    {file = "mmh3-5.0.1-cp38-cp38-musllinux_1_2_ppc64le.whl", hash = "sha256:42865567838d2193eb64e0ef571f678bf361a254fcdef0c5c8e73243217829bd"},
+    {file = "mmh3-5.0.1-cp38-cp38-musllinux_1_2_s390x.whl", hash = "sha256:5ff5c1f301c4a8b6916498969c0fcc7e3dbc56b4bfce5cfe3fe31f3f4609e5ae"},
+    {file = "mmh3-5.0.1-cp38-cp38-musllinux_1_2_x86_64.whl", hash = "sha256:be74c2dda8a6f44a504450aa2c3507f8067a159201586fc01dd41ab80efc350f"},
+    {file = "mmh3-5.0.1-cp38-cp38-win32.whl", hash = "sha256:5610a842621ff76c04b20b29cf5f809b131f241a19d4937971ba77dc99a7f330"},
+    {file = "mmh3-5.0.1-cp38-cp38-win_amd64.whl", hash = "sha256:de15739ac50776fe8aa1ef13f1be46a6ee1fbd45f6d0651084097eb2be0a5aa4"},
+    {file = "mmh3-5.0.1-cp39-cp39-macosx_10_9_universal2.whl", hash = "sha256:48e84cf3cc7e8c41bc07de72299a73b92d9e3cde51d97851420055b1484995f7"},
+    {file = "mmh3-5.0.1-cp39-cp39-macosx_10_9_x86_64.whl", hash = "sha256:6dd9dc28c2d168c49928195c2e29b96f9582a5d07bd690a28aede4cc07b0e696"},
+    {file = "mmh3-5.0.1-cp39-cp39-macosx_11_0_arm64.whl", hash = "sha256:2771a1c56a3d4bdad990309cff5d0a8051f29c8ec752d001f97d6392194ae880"},
+    {file = "mmh3-5.0.1-cp39-cp39-manylinux_2_17_aarch64.manylinux2014_aarch64.whl", hash = "sha256:c5ff2a8322ba40951a84411550352fba1073ce1c1d1213bb7530f09aed7f8caf"},
+    {file = "mmh3-5.0.1-cp39-cp39-manylinux_2_17_ppc64le.manylinux2014_ppc64le.whl", hash = "sha256:a16bd3ec90682c9e0a343e6bd4c778c09947c8c5395cdb9e5d9b82b2559efbca"},
+    {file = "mmh3-5.0.1-cp39-cp39-manylinux_2_17_s390x.manylinux2014_s390x.whl", hash = "sha256:d45733a78d68b5b05ff4a823aea51fa664df1d3bf4929b152ff4fd6dea2dd69b"},
+    {file = "mmh3-5.0.1-cp39-cp39-manylinux_2_5_i686.manylinux1_i686.manylinux_2_17_i686.manylinux2014_i686.whl", hash = "sha256:904285e83cedebc8873b0838ed54c20f7344120be26e2ca5a907ab007a18a7a0"},
+    {file = "mmh3-5.0.1-cp39-cp39-manylinux_2_5_x86_64.manylinux1_x86_64.manylinux_2_17_x86_64.manylinux2014_x86_64.whl", hash = "sha256:ac4aeb1784e43df728034d0ed72e4b2648db1a69fef48fa58e810e13230ae5ff"},
+    {file = "mmh3-5.0.1-cp39-cp39-musllinux_1_2_aarch64.whl", hash = "sha256:cb3d4f751a0b8b4c8d06ef1c085216c8fddcc8b8c8d72445976b5167a40c6d1e"},
+    {file = "mmh3-5.0.1-cp39-cp39-musllinux_1_2_i686.whl", hash = "sha256:8021851935600e60c42122ed1176399d7692df338d606195cd599d228a04c1c6"},
+    {file = "mmh3-5.0.1-cp39-cp39-musllinux_1_2_ppc64le.whl", hash = "sha256:6182d5924a5efc451900f864cbb021d7e8ad5d524816ca17304a0f663bc09bb5"},
+    {file = "mmh3-5.0.1-cp39-cp39-musllinux_1_2_s390x.whl", hash = "sha256:5f30b834552a4f79c92e3d266336fb87fd92ce1d36dc6813d3e151035890abbd"},
+    {file = "mmh3-5.0.1-cp39-cp39-musllinux_1_2_x86_64.whl", hash = "sha256:cd4383f35e915e06d077df27e04ffd3be7513ec6a9de2d31f430393f67e192a7"},
+    {file = "mmh3-5.0.1-cp39-cp39-win32.whl", hash = "sha256:1455fb6b42665a97db8fc66e89a861e52b567bce27ed054c47877183f86ea6e3"},
+    {file = "mmh3-5.0.1-cp39-cp39-win_amd64.whl", hash = "sha256:9e26a0f4eb9855a143f5938a53592fa14c2d3b25801c2106886ab6c173982780"},
+    {file = "mmh3-5.0.1-cp39-cp39-win_arm64.whl", hash = "sha256:0d0a35a69abdad7549c4030a714bb4ad07902edb3bbe61e1bbc403ded5d678be"},
+    {file = "mmh3-5.0.1.tar.gz", hash = "sha256:7dab080061aeb31a6069a181f27c473a1f67933854e36a3464931f2716508896"},
+]
+
+[package.extras]
+benchmark = ["pymmh3 (==0.0.5)", "pyperf (==2.7.0)", "xxhash (==3.5.0)"]
+docs = ["myst-parser (==4.0.0)", "shibuya (==2024.8.30)", "sphinx (==8.0.2)", "sphinx-copybutton (==0.5.2)"]
+lint = ["black (==24.8.0)", "clang-format (==18.1.8)", "isort (==5.13.2)", "pylint (==3.2.7)"]
+plot = ["matplotlib (==3.9.2)", "pandas (==2.2.2)"]
+test = ["pytest (==8.3.3)", "pytest-sugar (==1.0.0)"]
+type = ["mypy (==1.11.2)"]
 
 [[package]]
 name = "mock"
@@ -5539,13 +5597,13 @@ signedtoken = ["cryptography (>=3.0.0)", "pyjwt (>=2.0.0,<3)"]
 
 [[package]]
 name = "oci"
-version = "2.133.0"
+version = "2.135.0"
 description = "Oracle Cloud Infrastructure Python SDK"
 optional = false
 python-versions = "*"
 files = [
-    {file = "oci-2.133.0-py3-none-any.whl", hash = "sha256:9706365481ca538c89b3a15e6b5c246801eccb06be831a7f21c40f2a2ee310a7"},
-    {file = "oci-2.133.0.tar.gz", hash = "sha256:800418025bb98f587c65bbf89c6b6d61ef0f2249e0698d73439baf3251640b7f"},
+    {file = "oci-2.135.0-py3-none-any.whl", hash = "sha256:c01f1d103ed034fa7ca2bceb297bf00e6f6c456d14a46b35ee9007b25f3ea397"},
+    {file = "oci-2.135.0.tar.gz", hash = "sha256:6e28e6595264705d8fd0719045ffc4b23170e7fd2cd76a1c3aa25e4cdaa5883a"},
 ]
 
 [package.dependencies]
@@ -6059,40 +6117,53 @@ files = [
 
 [[package]]
 name = "pandas"
-version = "2.2.2"
+version = "2.2.3"
 description = "Powerful data structures for data analysis, time series, and statistics"
 optional = false
 python-versions = ">=3.9"
 files = [
-    {file = "pandas-2.2.2-cp310-cp310-macosx_10_9_x86_64.whl", hash = "sha256:90c6fca2acf139569e74e8781709dccb6fe25940488755716d1d354d6bc58bce"},
-    {file = "pandas-2.2.2-cp310-cp310-macosx_11_0_arm64.whl", hash = "sha256:c7adfc142dac335d8c1e0dcbd37eb8617eac386596eb9e1a1b77791cf2498238"},
-    {file = "pandas-2.2.2-cp310-cp310-manylinux_2_17_aarch64.manylinux2014_aarch64.whl", hash = "sha256:4abfe0be0d7221be4f12552995e58723c7422c80a659da13ca382697de830c08"},
-    {file = "pandas-2.2.2-cp310-cp310-manylinux_2_17_x86_64.manylinux2014_x86_64.whl", hash = "sha256:8635c16bf3d99040fdf3ca3db669a7250ddf49c55dc4aa8fe0ae0fa8d6dcc1f0"},
-    {file = "pandas-2.2.2-cp310-cp310-musllinux_1_1_aarch64.whl", hash = "sha256:40ae1dffb3967a52203105a077415a86044a2bea011b5f321c6aa64b379a3f51"},
-    {file = "pandas-2.2.2-cp310-cp310-musllinux_1_1_x86_64.whl", hash = "sha256:8e5a0b00e1e56a842f922e7fae8ae4077aee4af0acb5ae3622bd4b4c30aedf99"},
-    {file = "pandas-2.2.2-cp310-cp310-win_amd64.whl", hash = "sha256:ddf818e4e6c7c6f4f7c8a12709696d193976b591cc7dc50588d3d1a6b5dc8772"},
-    {file = "pandas-2.2.2-cp311-cp311-macosx_10_9_x86_64.whl", hash = "sha256:696039430f7a562b74fa45f540aca068ea85fa34c244d0deee539cb6d70aa288"},
-    {file = "pandas-2.2.2-cp311-cp311-macosx_11_0_arm64.whl", hash = "sha256:8e90497254aacacbc4ea6ae5e7a8cd75629d6ad2b30025a4a8b09aa4faf55151"},
-    {file = "pandas-2.2.2-cp311-cp311-manylinux_2_17_aarch64.manylinux2014_aarch64.whl", hash = "sha256:58b84b91b0b9f4bafac2a0ac55002280c094dfc6402402332c0913a59654ab2b"},
-    {file = "pandas-2.2.2-cp311-cp311-manylinux_2_17_x86_64.manylinux2014_x86_64.whl", hash = "sha256:6d2123dc9ad6a814bcdea0f099885276b31b24f7edf40f6cdbc0912672e22eee"},
-    {file = "pandas-2.2.2-cp311-cp311-musllinux_1_1_aarch64.whl", hash = "sha256:2925720037f06e89af896c70bca73459d7e6a4be96f9de79e2d440bd499fe0db"},
-    {file = "pandas-2.2.2-cp311-cp311-musllinux_1_1_x86_64.whl", hash = "sha256:0cace394b6ea70c01ca1595f839cf193df35d1575986e484ad35c4aeae7266c1"},
-    {file = "pandas-2.2.2-cp311-cp311-win_amd64.whl", hash = "sha256:873d13d177501a28b2756375d59816c365e42ed8417b41665f346289adc68d24"},
-    {file = "pandas-2.2.2-cp312-cp312-macosx_10_9_x86_64.whl", hash = "sha256:9dfde2a0ddef507a631dc9dc4af6a9489d5e2e740e226ad426a05cabfbd7c8ef"},
-    {file = "pandas-2.2.2-cp312-cp312-macosx_11_0_arm64.whl", hash = "sha256:e9b79011ff7a0f4b1d6da6a61aa1aa604fb312d6647de5bad20013682d1429ce"},
-    {file = "pandas-2.2.2-cp312-cp312-manylinux_2_17_aarch64.manylinux2014_aarch64.whl", hash = "sha256:1cb51fe389360f3b5a4d57dbd2848a5f033350336ca3b340d1c53a1fad33bcad"},
-    {file = "pandas-2.2.2-cp312-cp312-manylinux_2_17_x86_64.manylinux2014_x86_64.whl", hash = "sha256:eee3a87076c0756de40b05c5e9a6069c035ba43e8dd71c379e68cab2c20f16ad"},
-    {file = "pandas-2.2.2-cp312-cp312-musllinux_1_1_aarch64.whl", hash = "sha256:3e374f59e440d4ab45ca2fffde54b81ac3834cf5ae2cdfa69c90bc03bde04d76"},
-    {file = "pandas-2.2.2-cp312-cp312-musllinux_1_1_x86_64.whl", hash = "sha256:43498c0bdb43d55cb162cdc8c06fac328ccb5d2eabe3cadeb3529ae6f0517c32"},
-    {file = "pandas-2.2.2-cp312-cp312-win_amd64.whl", hash = "sha256:d187d355ecec3629624fccb01d104da7d7f391db0311145817525281e2804d23"},
-    {file = "pandas-2.2.2-cp39-cp39-macosx_10_9_x86_64.whl", hash = "sha256:0ca6377b8fca51815f382bd0b697a0814c8bda55115678cbc94c30aacbb6eff2"},
-    {file = "pandas-2.2.2-cp39-cp39-macosx_11_0_arm64.whl", hash = "sha256:9057e6aa78a584bc93a13f0a9bf7e753a5e9770a30b4d758b8d5f2a62a9433cd"},
-    {file = "pandas-2.2.2-cp39-cp39-manylinux_2_17_aarch64.manylinux2014_aarch64.whl", hash = "sha256:001910ad31abc7bf06f49dcc903755d2f7f3a9186c0c040b827e522e9cef0863"},
-    {file = "pandas-2.2.2-cp39-cp39-manylinux_2_17_x86_64.manylinux2014_x86_64.whl", hash = "sha256:66b479b0bd07204e37583c191535505410daa8df638fd8e75ae1b383851fe921"},
-    {file = "pandas-2.2.2-cp39-cp39-musllinux_1_1_aarch64.whl", hash = "sha256:a77e9d1c386196879aa5eb712e77461aaee433e54c68cf253053a73b7e49c33a"},
-    {file = "pandas-2.2.2-cp39-cp39-musllinux_1_1_x86_64.whl", hash = "sha256:92fd6b027924a7e178ac202cfbe25e53368db90d56872d20ffae94b96c7acc57"},
-    {file = "pandas-2.2.2-cp39-cp39-win_amd64.whl", hash = "sha256:640cef9aa381b60e296db324337a554aeeb883ead99dc8f6c18e81a93942f5f4"},
-    {file = "pandas-2.2.2.tar.gz", hash = "sha256:9e79019aba43cb4fda9e4d983f8e88ca0373adbb697ae9c6c43093218de28b54"},
+    {file = "pandas-2.2.3-cp310-cp310-macosx_10_9_x86_64.whl", hash = "sha256:1948ddde24197a0f7add2bdc4ca83bf2b1ef84a1bc8ccffd95eda17fd836ecb5"},
+    {file = "pandas-2.2.3-cp310-cp310-macosx_11_0_arm64.whl", hash = "sha256:381175499d3802cde0eabbaf6324cce0c4f5d52ca6f8c377c29ad442f50f6348"},
+    {file = "pandas-2.2.3-cp310-cp310-manylinux2014_aarch64.manylinux_2_17_aarch64.whl", hash = "sha256:d9c45366def9a3dd85a6454c0e7908f2b3b8e9c138f5dc38fed7ce720d8453ed"},
+    {file = "pandas-2.2.3-cp310-cp310-manylinux_2_17_x86_64.manylinux2014_x86_64.whl", hash = "sha256:86976a1c5b25ae3f8ccae3a5306e443569ee3c3faf444dfd0f41cda24667ad57"},
+    {file = "pandas-2.2.3-cp310-cp310-musllinux_1_2_aarch64.whl", hash = "sha256:b8661b0238a69d7aafe156b7fa86c44b881387509653fdf857bebc5e4008ad42"},
+    {file = "pandas-2.2.3-cp310-cp310-musllinux_1_2_x86_64.whl", hash = "sha256:37e0aced3e8f539eccf2e099f65cdb9c8aa85109b0be6e93e2baff94264bdc6f"},
+    {file = "pandas-2.2.3-cp310-cp310-win_amd64.whl", hash = "sha256:56534ce0746a58afaf7942ba4863e0ef81c9c50d3f0ae93e9497d6a41a057645"},
+    {file = "pandas-2.2.3-cp311-cp311-macosx_10_9_x86_64.whl", hash = "sha256:66108071e1b935240e74525006034333f98bcdb87ea116de573a6a0dccb6c039"},
+    {file = "pandas-2.2.3-cp311-cp311-macosx_11_0_arm64.whl", hash = "sha256:7c2875855b0ff77b2a64a0365e24455d9990730d6431b9e0ee18ad8acee13dbd"},
+    {file = "pandas-2.2.3-cp311-cp311-manylinux2014_aarch64.manylinux_2_17_aarch64.whl", hash = "sha256:cd8d0c3be0515c12fed0bdbae072551c8b54b7192c7b1fda0ba56059a0179698"},
+    {file = "pandas-2.2.3-cp311-cp311-manylinux_2_17_x86_64.manylinux2014_x86_64.whl", hash = "sha256:c124333816c3a9b03fbeef3a9f230ba9a737e9e5bb4060aa2107a86cc0a497fc"},
+    {file = "pandas-2.2.3-cp311-cp311-musllinux_1_2_aarch64.whl", hash = "sha256:63cc132e40a2e084cf01adf0775b15ac515ba905d7dcca47e9a251819c575ef3"},
+    {file = "pandas-2.2.3-cp311-cp311-musllinux_1_2_x86_64.whl", hash = "sha256:29401dbfa9ad77319367d36940cd8a0b3a11aba16063e39632d98b0e931ddf32"},
+    {file = "pandas-2.2.3-cp311-cp311-win_amd64.whl", hash = "sha256:3fc6873a41186404dad67245896a6e440baacc92f5b716ccd1bc9ed2995ab2c5"},
+    {file = "pandas-2.2.3-cp312-cp312-macosx_10_9_x86_64.whl", hash = "sha256:b1d432e8d08679a40e2a6d8b2f9770a5c21793a6f9f47fdd52c5ce1948a5a8a9"},
+    {file = "pandas-2.2.3-cp312-cp312-macosx_11_0_arm64.whl", hash = "sha256:a5a1595fe639f5988ba6a8e5bc9649af3baf26df3998a0abe56c02609392e0a4"},
+    {file = "pandas-2.2.3-cp312-cp312-manylinux2014_aarch64.manylinux_2_17_aarch64.whl", hash = "sha256:5de54125a92bb4d1c051c0659e6fcb75256bf799a732a87184e5ea503965bce3"},
+    {file = "pandas-2.2.3-cp312-cp312-manylinux_2_17_x86_64.manylinux2014_x86_64.whl", hash = "sha256:fffb8ae78d8af97f849404f21411c95062db1496aeb3e56f146f0355c9989319"},
+    {file = "pandas-2.2.3-cp312-cp312-musllinux_1_2_aarch64.whl", hash = "sha256:6dfcb5ee8d4d50c06a51c2fffa6cff6272098ad6540aed1a76d15fb9318194d8"},
+    {file = "pandas-2.2.3-cp312-cp312-musllinux_1_2_x86_64.whl", hash = "sha256:062309c1b9ea12a50e8ce661145c6aab431b1e99530d3cd60640e255778bd43a"},
+    {file = "pandas-2.2.3-cp312-cp312-win_amd64.whl", hash = "sha256:59ef3764d0fe818125a5097d2ae867ca3fa64df032331b7e0917cf5d7bf66b13"},
+    {file = "pandas-2.2.3-cp313-cp313-macosx_10_13_x86_64.whl", hash = "sha256:f00d1345d84d8c86a63e476bb4955e46458b304b9575dcf71102b5c705320015"},
+    {file = "pandas-2.2.3-cp313-cp313-macosx_11_0_arm64.whl", hash = "sha256:3508d914817e153ad359d7e069d752cdd736a247c322d932eb89e6bc84217f28"},
+    {file = "pandas-2.2.3-cp313-cp313-manylinux2014_aarch64.manylinux_2_17_aarch64.whl", hash = "sha256:22a9d949bfc9a502d320aa04e5d02feab689d61da4e7764b62c30b991c42c5f0"},
+    {file = "pandas-2.2.3-cp313-cp313-manylinux_2_17_x86_64.manylinux2014_x86_64.whl", hash = "sha256:f3a255b2c19987fbbe62a9dfd6cff7ff2aa9ccab3fc75218fd4b7530f01efa24"},
+    {file = "pandas-2.2.3-cp313-cp313-musllinux_1_2_aarch64.whl", hash = "sha256:800250ecdadb6d9c78eae4990da62743b857b470883fa27f652db8bdde7f6659"},
+    {file = "pandas-2.2.3-cp313-cp313-musllinux_1_2_x86_64.whl", hash = "sha256:6374c452ff3ec675a8f46fd9ab25c4ad0ba590b71cf0656f8b6daa5202bca3fb"},
+    {file = "pandas-2.2.3-cp313-cp313-win_amd64.whl", hash = "sha256:61c5ad4043f791b61dd4752191d9f07f0ae412515d59ba8f005832a532f8736d"},
+    {file = "pandas-2.2.3-cp313-cp313t-macosx_10_13_x86_64.whl", hash = "sha256:3b71f27954685ee685317063bf13c7709a7ba74fc996b84fc6821c59b0f06468"},
+    {file = "pandas-2.2.3-cp313-cp313t-macosx_11_0_arm64.whl", hash = "sha256:38cf8125c40dae9d5acc10fa66af8ea6fdf760b2714ee482ca691fc66e6fcb18"},
+    {file = "pandas-2.2.3-cp313-cp313t-manylinux2014_aarch64.manylinux_2_17_aarch64.whl", hash = "sha256:ba96630bc17c875161df3818780af30e43be9b166ce51c9a18c1feae342906c2"},
+    {file = "pandas-2.2.3-cp313-cp313t-manylinux_2_17_x86_64.manylinux2014_x86_64.whl", hash = "sha256:1db71525a1538b30142094edb9adc10be3f3e176748cd7acc2240c2f2e5aa3a4"},
+    {file = "pandas-2.2.3-cp313-cp313t-musllinux_1_2_aarch64.whl", hash = "sha256:15c0e1e02e93116177d29ff83e8b1619c93ddc9c49083f237d4312337a61165d"},
+    {file = "pandas-2.2.3-cp313-cp313t-musllinux_1_2_x86_64.whl", hash = "sha256:ad5b65698ab28ed8d7f18790a0dc58005c7629f227be9ecc1072aa74c0c1d43a"},
+    {file = "pandas-2.2.3-cp39-cp39-macosx_10_9_x86_64.whl", hash = "sha256:bc6b93f9b966093cb0fd62ff1a7e4c09e6d546ad7c1de191767baffc57628f39"},
+    {file = "pandas-2.2.3-cp39-cp39-macosx_11_0_arm64.whl", hash = "sha256:5dbca4c1acd72e8eeef4753eeca07de9b1db4f398669d5994086f788a5d7cc30"},
+    {file = "pandas-2.2.3-cp39-cp39-manylinux2014_aarch64.manylinux_2_17_aarch64.whl", hash = "sha256:8cd6d7cc958a3910f934ea8dbdf17b2364827bb4dafc38ce6eef6bb3d65ff09c"},
+    {file = "pandas-2.2.3-cp39-cp39-manylinux_2_17_x86_64.manylinux2014_x86_64.whl", hash = "sha256:99df71520d25fade9db7c1076ac94eb994f4d2673ef2aa2e86ee039b6746d20c"},
+    {file = "pandas-2.2.3-cp39-cp39-musllinux_1_2_aarch64.whl", hash = "sha256:31d0ced62d4ea3e231a9f228366919a5ea0b07440d9d4dac345376fd8e1477ea"},
+    {file = "pandas-2.2.3-cp39-cp39-musllinux_1_2_x86_64.whl", hash = "sha256:7eee9e7cea6adf3e3d24e304ac6b8300646e2a5d1cd3a3c2abed9101b0846761"},
+    {file = "pandas-2.2.3-cp39-cp39-win_amd64.whl", hash = "sha256:4850ba03528b6dd51d6c5d273c46f183f39a9baf3f0143e566b89450965b105e"},
+    {file = "pandas-2.2.3.tar.gz", hash = "sha256:4f18ba62b61d7e192368b84517265a99b4d7ee8912f8708660fb4a366cc82667"},
 ]
 
 [package.dependencies]
@@ -6300,13 +6371,13 @@ xmp = ["defusedxml"]
 
 [[package]]
 name = "platformdirs"
-version = "4.3.3"
+version = "4.3.6"
 description = "A small Python package for determining appropriate platform-specific dirs, e.g. a `user data dir`."
 optional = false
 python-versions = ">=3.8"
 files = [
-    {file = "platformdirs-4.3.3-py3-none-any.whl", hash = "sha256:50a5450e2e84f44539718293cbb1da0a0885c9d14adf21b77bae4e66fc99d9b5"},
-    {file = "platformdirs-4.3.3.tar.gz", hash = "sha256:d4e0b7d8ec176b341fb03cb11ca12d0276faa8c485f9cd218f613840463fc2c0"},
+    {file = "platformdirs-4.3.6-py3-none-any.whl", hash = "sha256:73e575e1408ab8103900836b97580d5307456908a03e92031bab39e4554cc3fb"},
+    {file = "platformdirs-4.3.6.tar.gz", hash = "sha256:357fb2acbc885b0419afd3ce3ed34564c13c9b95c89360cd9563f73aa5e2b907"},
 ]
 
 [package.extras]
@@ -6376,13 +6447,13 @@ tests = ["pytest (>=5.4.1)", "pytest-cov (>=2.8.1)", "pytest-mypy (>=0.8.0)", "p
 
 [[package]]
 name = "posthog"
-version = "3.6.5"
+version = "3.6.6"
 description = "Integrate PostHog into any python application."
 optional = false
 python-versions = "*"
 files = [
-    {file = "posthog-3.6.5-py2.py3-none-any.whl", hash = "sha256:f8b7c573826b061a1d22c9495169c38ebe83a1df2729f49c7129a9c23a02acf6"},
-    {file = "posthog-3.6.5.tar.gz", hash = "sha256:7fd3ca809e15476c35f75d18cd6bba31395daf0a17b75242965c469fb6292510"},
+    {file = "posthog-3.6.6-py2.py3-none-any.whl", hash = "sha256:38834fd7f0732582a20d4eb4674c8d5c088e464d14d1b3f8c176e389aecaa4ef"},
+    {file = "posthog-3.6.6.tar.gz", hash = "sha256:1e04783293117109189ad7048f3eedbe21caff0e39bee5e2d47a93dd790fefac"},
 ]
 
 [package.dependencies]
@@ -6399,44 +6470,44 @@ test = ["coverage", "django", "flake8", "freezegun (==0.3.15)", "mock (>=2.0.0)"
 
 [[package]]
 name = "pox"
-version = "0.3.4"
+version = "0.3.5"
 description = "utilities for filesystem exploration and automated builds"
 optional = false
 python-versions = ">=3.8"
 files = [
-    {file = "pox-0.3.4-py3-none-any.whl", hash = "sha256:651b8ae8a7b341b7bfd267f67f63106daeb9805f1ac11f323d5280d2da93fdb6"},
-    {file = "pox-0.3.4.tar.gz", hash = "sha256:16e6eca84f1bec3828210b06b052adf04cf2ab20c22fd6fbef5f78320c9a6fed"},
+    {file = "pox-0.3.5-py3-none-any.whl", hash = "sha256:9e82bcc9e578b43e80a99cad80f0d8f44f4d424f0ee4ee8d4db27260a6aa365a"},
+    {file = "pox-0.3.5.tar.gz", hash = "sha256:8120ee4c94e950e6e0483e050a4f0e56076e590ba0a9add19524c254bd23c2d1"},
 ]
 
 [[package]]
 name = "ppft"
-version = "1.7.6.8"
+version = "1.7.6.9"
 description = "distributed and parallel Python"
 optional = false
 python-versions = ">=3.8"
 files = [
-    {file = "ppft-1.7.6.8-py3-none-any.whl", hash = "sha256:de2dd4b1b080923dd9627fbdea52649fd741c752fce4f3cf37e26f785df23d9b"},
-    {file = "ppft-1.7.6.8.tar.gz", hash = "sha256:76a429a7d7b74c4d743f6dba8351e58d62b6432ed65df9fe204790160dab996d"},
+    {file = "ppft-1.7.6.9-py3-none-any.whl", hash = "sha256:dab36548db5ca3055067fbe6b1a17db5fee29f3c366c579a9a27cebb52ed96f0"},
+    {file = "ppft-1.7.6.9.tar.gz", hash = "sha256:73161c67474ea9d81d04bcdad166d399cff3f084d5d2dc21ebdd46c075bbc265"},
 ]
 
 [package.extras]
-dill = ["dill (>=0.3.8)"]
+dill = ["dill (>=0.3.9)"]
 
 [[package]]
 name = "primp"
-version = "0.6.2"
+version = "0.6.3"
 description = "HTTP client that can impersonate web browsers, mimicking their headers and `TLS/JA3/JA4/HTTP2` fingerprints"
 optional = false
 python-versions = ">=3.8"
 files = [
-    {file = "primp-0.6.2-cp38-abi3-macosx_10_12_x86_64.whl", hash = "sha256:4a35d441462a55d9a9525bf170e2ffd2fcb3db6039b23e802859fa22c18cdd51"},
-    {file = "primp-0.6.2-cp38-abi3-macosx_11_0_arm64.whl", hash = "sha256:f67ccade95bdbca3cf9b96b93aa53f9617d85ddbf988da4e9c523aa785fd2d54"},
-    {file = "primp-0.6.2-cp38-abi3-manylinux_2_17_x86_64.manylinux2014_x86_64.whl", hash = "sha256:8074b93befaf36567e4cf3d4a1a8cd6ab9cc6e4dd4ff710650678daa405aee71"},
-    {file = "primp-0.6.2-cp38-abi3-manylinux_2_34_aarch64.whl", hash = "sha256:7d3e2a3f8c6262e9b883651b79c4ff2b7677a76f47293a139f541c9ea333ce3b"},
-    {file = "primp-0.6.2-cp38-abi3-musllinux_1_2_aarch64.whl", hash = "sha256:a460ea389371c6d04839b4b50b5805d99da8ebe281a2e8b534d27377c6d44f0e"},
-    {file = "primp-0.6.2-cp38-abi3-musllinux_1_2_x86_64.whl", hash = "sha256:5b6b27e89d3c05c811aff0e4fde7a36d6957b15b3112f4ce28b6b99e8ca1e725"},
-    {file = "primp-0.6.2-cp38-abi3-win_amd64.whl", hash = "sha256:1006a40a85f88a4c5222094813a1ebc01f85a63e9a33d2c443288c0720bed321"},
-    {file = "primp-0.6.2.tar.gz", hash = "sha256:5a96a6b65195a8a989157e67d23bd171c49be238654e02bdf1b1fda36cbcc068"},
+    {file = "primp-0.6.3-cp38-abi3-macosx_10_12_x86_64.whl", hash = "sha256:bdbe6a7cdaaf5c9ed863432a941f4a75bd4c6ff626cbc8d32fc232793c70ba06"},
+    {file = "primp-0.6.3-cp38-abi3-macosx_11_0_arm64.whl", hash = "sha256:eeb53eb987bdcbcd85740633470255cab887d921df713ffa12a36a13366c9cdb"},
+    {file = "primp-0.6.3-cp38-abi3-manylinux_2_17_x86_64.manylinux2014_x86_64.whl", hash = "sha256:78da53d3c92a8e3f05bd3286ac76c291f1b6fe5e08ea63b7ba92b0f9141800bb"},
+    {file = "primp-0.6.3-cp38-abi3-manylinux_2_34_aarch64.whl", hash = "sha256:86337b44deecdac752bd8112909987fc9fa9b894f30191c80a164dc8f895da53"},
+    {file = "primp-0.6.3-cp38-abi3-musllinux_1_2_aarch64.whl", hash = "sha256:d3cd9a22b97f3eae42b2a5fb99f00480daf4cd6d9b139e05b0ffb03f7cc037f3"},
+    {file = "primp-0.6.3-cp38-abi3-musllinux_1_2_x86_64.whl", hash = "sha256:7732bec917e2d3c48a31cdb92e1250f4ad6203a1aa4f802bd9abd84f2286a1e0"},
+    {file = "primp-0.6.3-cp38-abi3-win_amd64.whl", hash = "sha256:1e4113c34b86c676ae321af185f03a372caef3ee009f1682c2d62e30ec87348c"},
+    {file = "primp-0.6.3.tar.gz", hash = "sha256:17d30ebe26864defad5232dbbe1372e80483940012356e1f68846bb182282039"},
 ]
 
 [package.extras]
@@ -6444,13 +6515,13 @@ dev = ["certifi", "pytest (>=8.1.1)"]
 
 [[package]]
 name = "prompt-toolkit"
-version = "3.0.47"
+version = "3.0.48"
 description = "Library for building powerful interactive command lines in Python"
 optional = false
 python-versions = ">=3.7.0"
 files = [
-    {file = "prompt_toolkit-3.0.47-py3-none-any.whl", hash = "sha256:0d7bfa67001d5e39d02c224b663abc33687405033a8c422d0d675a5a13361d10"},
-    {file = "prompt_toolkit-3.0.47.tar.gz", hash = "sha256:1e1b29cb58080b1e69f207c893a1a7bf16d127a5c30c9d17a25a5d77792e5360"},
+    {file = "prompt_toolkit-3.0.48-py3-none-any.whl", hash = "sha256:f49a827f90062e411f1ce1f854f2aedb3c23353244f8108b89283587397ac10e"},
+    {file = "prompt_toolkit-3.0.48.tar.gz", hash = "sha256:d6623ab0477a80df74e646bdbc93621143f5caf104206aa29294d53de1a03d90"},
 ]
 
 [package.dependencies]
@@ -6475,22 +6546,22 @@ testing = ["google-api-core (>=1.31.5)"]
 
 [[package]]
 name = "protobuf"
-version = "4.25.4"
+version = "4.25.5"
 description = ""
 optional = false
 python-versions = ">=3.8"
 files = [
-    {file = "protobuf-4.25.4-cp310-abi3-win32.whl", hash = "sha256:db9fd45183e1a67722cafa5c1da3e85c6492a5383f127c86c4c4aa4845867dc4"},
-    {file = "protobuf-4.25.4-cp310-abi3-win_amd64.whl", hash = "sha256:ba3d8504116a921af46499471c63a85260c1a5fc23333154a427a310e015d26d"},
-    {file = "protobuf-4.25.4-cp37-abi3-macosx_10_9_universal2.whl", hash = "sha256:eecd41bfc0e4b1bd3fa7909ed93dd14dd5567b98c941d6c1ad08fdcab3d6884b"},
-    {file = "protobuf-4.25.4-cp37-abi3-manylinux2014_aarch64.whl", hash = "sha256:4c8a70fdcb995dcf6c8966cfa3a29101916f7225e9afe3ced4395359955d3835"},
-    {file = "protobuf-4.25.4-cp37-abi3-manylinux2014_x86_64.whl", hash = "sha256:3319e073562e2515c6ddc643eb92ce20809f5d8f10fead3332f71c63be6a7040"},
-    {file = "protobuf-4.25.4-cp38-cp38-win32.whl", hash = "sha256:7e372cbbda66a63ebca18f8ffaa6948455dfecc4e9c1029312f6c2edcd86c4e1"},
-    {file = "protobuf-4.25.4-cp38-cp38-win_amd64.whl", hash = "sha256:051e97ce9fa6067a4546e75cb14f90cf0232dcb3e3d508c448b8d0e4265b61c1"},
-    {file = "protobuf-4.25.4-cp39-cp39-win32.whl", hash = "sha256:90bf6fd378494eb698805bbbe7afe6c5d12c8e17fca817a646cd6a1818c696ca"},
-    {file = "protobuf-4.25.4-cp39-cp39-win_amd64.whl", hash = "sha256:ac79a48d6b99dfed2729ccccee547b34a1d3d63289c71cef056653a846a2240f"},
-    {file = "protobuf-4.25.4-py3-none-any.whl", hash = "sha256:bfbebc1c8e4793cfd58589acfb8a1026be0003e852b9da7db5a4285bde996978"},
-    {file = "protobuf-4.25.4.tar.gz", hash = "sha256:0dc4a62cc4052a036ee2204d26fe4d835c62827c855c8a03f29fe6da146b380d"},
+    {file = "protobuf-4.25.5-cp310-abi3-win32.whl", hash = "sha256:5e61fd921603f58d2f5acb2806a929b4675f8874ff5f330b7d6f7e2e784bbcd8"},
+    {file = "protobuf-4.25.5-cp310-abi3-win_amd64.whl", hash = "sha256:4be0571adcbe712b282a330c6e89eae24281344429ae95c6d85e79e84780f5ea"},
+    {file = "protobuf-4.25.5-cp37-abi3-macosx_10_9_universal2.whl", hash = "sha256:b2fde3d805354df675ea4c7c6338c1aecd254dfc9925e88c6d31a2bcb97eb173"},
+    {file = "protobuf-4.25.5-cp37-abi3-manylinux2014_aarch64.whl", hash = "sha256:919ad92d9b0310070f8356c24b855c98df2b8bd207ebc1c0c6fcc9ab1e007f3d"},
+    {file = "protobuf-4.25.5-cp37-abi3-manylinux2014_x86_64.whl", hash = "sha256:fe14e16c22be926d3abfcb500e60cab068baf10b542b8c858fa27e098123e331"},
+    {file = "protobuf-4.25.5-cp38-cp38-win32.whl", hash = "sha256:98d8d8aa50de6a2747efd9cceba361c9034050ecce3e09136f90de37ddba66e1"},
+    {file = "protobuf-4.25.5-cp38-cp38-win_amd64.whl", hash = "sha256:b0234dd5a03049e4ddd94b93400b67803c823cfc405689688f59b34e0742381a"},
+    {file = "protobuf-4.25.5-cp39-cp39-win32.whl", hash = "sha256:abe32aad8561aa7cc94fc7ba4fdef646e576983edb94a73381b03c53728a626f"},
+    {file = "protobuf-4.25.5-cp39-cp39-win_amd64.whl", hash = "sha256:7a183f592dc80aa7c8da7ad9e55091c4ffc9497b3054452d629bb85fa27c2a45"},
+    {file = "protobuf-4.25.5-py3-none-any.whl", hash = "sha256:0aebecb809cae990f8129ada5ca273d9d670b76d9bfc9b1809f0a9c02b7dbf41"},
+    {file = "protobuf-4.25.5.tar.gz", hash = "sha256:7f8249476b4a9473645db7f8ab42b02fe1488cbe5fb72fddd445e0665afd8584"},
 ]
 
 [[package]]
@@ -6644,19 +6715,6 @@ files = [
     {file = "pyarrow-17.0.0-cp312-cp312-win_amd64.whl", hash = "sha256:392bc9feabc647338e6c89267635e111d71edad5fcffba204425a7c8d13610d7"},
     {file = "pyarrow-17.0.0-cp38-cp38-macosx_10_15_x86_64.whl", hash = "sha256:af5ff82a04b2171415f1410cff7ebb79861afc5dae50be73ce06d6e870615204"},
     {file = "pyarrow-17.0.0-cp38-cp38-macosx_11_0_arm64.whl", hash = "sha256:edca18eaca89cd6382dfbcff3dd2d87633433043650c07375d095cd3517561d8"},
-    {file = "pyarrow-17.0.0-cp38-cp38-manylinux_2_17_aarch64.manylinux2014_aarch64.whl", hash = "sha256:7c7916bff914ac5d4a8fe25b7a25e432ff921e72f6f2b7547d1e325c1ad9d155"},
-    {file = "pyarrow-17.0.0-cp38-cp38-manylinux_2_17_x86_64.manylinux2014_x86_64.whl", hash = "sha256:f553ca691b9e94b202ff741bdd40f6ccb70cdd5fbf65c187af132f1317de6145"},
-    {file = "pyarrow-17.0.0-cp38-cp38-manylinux_2_28_aarch64.whl", hash = "sha256:0cdb0e627c86c373205a2f94a510ac4376fdc523f8bb36beab2e7f204416163c"},
-    {file = "pyarrow-17.0.0-cp38-cp38-manylinux_2_28_x86_64.whl", hash = "sha256:d7d192305d9d8bc9082d10f361fc70a73590a4c65cf31c3e6926cd72b76bc35c"},
-    {file = "pyarrow-17.0.0-cp38-cp38-win_amd64.whl", hash = "sha256:02dae06ce212d8b3244dd3e7d12d9c4d3046945a5933d28026598e9dbbda1fca"},
-    {file = "pyarrow-17.0.0-cp39-cp39-macosx_10_15_x86_64.whl", hash = "sha256:13d7a460b412f31e4c0efa1148e1d29bdf18ad1411eb6757d38f8fbdcc8645fb"},
-    {file = "pyarrow-17.0.0-cp39-cp39-macosx_11_0_arm64.whl", hash = "sha256:9b564a51fbccfab5a04a80453e5ac6c9954a9c5ef2890d1bcf63741909c3f8df"},
-    {file = "pyarrow-17.0.0-cp39-cp39-manylinux_2_17_aarch64.manylinux2014_aarch64.whl", hash = "sha256:32503827abbc5aadedfa235f5ece8c4f8f8b0a3cf01066bc8d29de7539532687"},
-    {file = "pyarrow-17.0.0-cp39-cp39-manylinux_2_17_x86_64.manylinux2014_x86_64.whl", hash = "sha256:a155acc7f154b9ffcc85497509bcd0d43efb80d6f733b0dc3bb14e281f131c8b"},
-    {file = "pyarrow-17.0.0-cp39-cp39-manylinux_2_28_aarch64.whl", hash = "sha256:dec8d129254d0188a49f8a1fc99e0560dc1b85f60af729f47de4046015f9b0a5"},
-    {file = "pyarrow-17.0.0-cp39-cp39-manylinux_2_28_x86_64.whl", hash = "sha256:a48ddf5c3c6a6c505904545c25a4ae13646ae1f8ba703c4df4a1bfe4f4006bda"},
-    {file = "pyarrow-17.0.0-cp39-cp39-win_amd64.whl", hash = "sha256:42bf93249a083aca230ba7e2786c5f673507fa97bbd9725a1e2754715151a204"},
-    {file = "pyarrow-17.0.0.tar.gz", hash = "sha256:4beca9521ed2c0921c1023e68d097d0299b62c362639ea315572a58f3f50fd28"},
 ]
 
 [package.dependencies]
@@ -6957,13 +7015,13 @@ tests = ["coverage[toml] (==5.0.4)", "pytest (>=6.0.0,<7.0.0)"]
 
 [[package]]
 name = "pymilvus"
-version = "2.4.6"
+version = "2.4.7"
 description = "Python Sdk for Milvus"
 optional = false
 python-versions = ">=3.8"
 files = [
-    {file = "pymilvus-2.4.6-py3-none-any.whl", hash = "sha256:b4c43472edc313b845d313be50610e19054e6954b2c5c3b515565c596c2d3d97"},
-    {file = "pymilvus-2.4.6.tar.gz", hash = "sha256:6ac3eb91c92cc01bbe444fe83f895f02d7b2546d96ac67998630bf31ac074d66"},
+    {file = "pymilvus-2.4.7-py3-none-any.whl", hash = "sha256:1e5d377bd40fa7eb459d3958dbd96201758f5cf997d41eb3d2d169d0b7fa462e"},
+    {file = "pymilvus-2.4.7.tar.gz", hash = "sha256:9ef460b940782a42e1b7b8ae0da03d8cc02d9d80044d13f4b689a7c935ec7aa7"},
 ]
 
 [package.dependencies]
@@ -7082,28 +7140,28 @@ files = [
 
 [[package]]
 name = "pyproject-hooks"
-version = "1.1.0"
+version = "1.2.0"
 description = "Wrappers to call pyproject.toml-based build backend hooks."
 optional = false
 python-versions = ">=3.7"
 files = [
-    {file = "pyproject_hooks-1.1.0-py3-none-any.whl", hash = "sha256:7ceeefe9aec63a1064c18d939bdc3adf2d8aa1988a510afec15151578b232aa2"},
-    {file = "pyproject_hooks-1.1.0.tar.gz", hash = "sha256:4b37730834edbd6bd37f26ece6b44802fb1c1ee2ece0e54ddff8bfc06db86965"},
+    {file = "pyproject_hooks-1.2.0-py3-none-any.whl", hash = "sha256:9e5c6bfa8dcc30091c74b0cf803c81fdd29d94f01992a7707bc97babb1141913"},
+    {file = "pyproject_hooks-1.2.0.tar.gz", hash = "sha256:1e859bd5c40fae9448642dd871adf459e5e2084186e8d2c2a79a824c970da1f8"},
 ]
 
 [[package]]
 name = "pyreadline3"
-version = "3.5.2"
+version = "3.5.4"
 description = "A python implementation of GNU readline."
 optional = false
 python-versions = ">=3.8"
 files = [
-    {file = "pyreadline3-3.5.2-py3-none-any.whl", hash = "sha256:a87d56791e2965b2b187e2ea33dcf664600842c997c0623c95cf8ef07db83de9"},
-    {file = "pyreadline3-3.5.2.tar.gz", hash = "sha256:ba82292e52c5a3bb256b291af0c40b457c1e8699cac9a873abbcaac8aef3a1bb"},
+    {file = "pyreadline3-3.5.4-py3-none-any.whl", hash = "sha256:eaf8e6cc3c49bcccf145fc6067ba8643d1df34d604a1ec0eccbf7a18e6d3fae6"},
+    {file = "pyreadline3-3.5.4.tar.gz", hash = "sha256:8d57d53039a1c75adba8e50dd3d992b28143480816187ea5efbd5c78e6c885b7"},
 ]
 
 [package.extras]
-dev = ["build", "flake8", "pytest", "twine"]
+dev = ["build", "flake8", "mypy", "pytest", "twine"]
 
 [[package]]
 name = "pytest"
@@ -7149,21 +7207,21 @@ histogram = ["pygal", "pygaljs"]
 
 [[package]]
 name = "pytest-env"
-version = "1.1.4"
+version = "1.1.5"
 description = "pytest plugin that allows you to add environment variables."
 optional = false
 python-versions = ">=3.8"
 files = [
-    {file = "pytest_env-1.1.4-py3-none-any.whl", hash = "sha256:a4212056d4d440febef311a98fdca56c31256d58fb453d103cba4e8a532b721d"},
-    {file = "pytest_env-1.1.4.tar.gz", hash = "sha256:86653658da8f11c6844975db955746c458a9c09f1e64957603161e2ff93f5133"},
+    {file = "pytest_env-1.1.5-py3-none-any.whl", hash = "sha256:ce90cf8772878515c24b31cd97c7fa1f4481cd68d588419fd45f10ecaee6bc30"},
+    {file = "pytest_env-1.1.5.tar.gz", hash = "sha256:91209840aa0e43385073ac464a554ad2947cc2fd663a9debf88d03b01e0cc1cf"},
 ]
 
 [package.dependencies]
-pytest = ">=8.3.2"
+pytest = ">=8.3.3"
 tomli = {version = ">=2.0.1", markers = "python_version < \"3.11\""}
 
 [package.extras]
-test = ["covdefaults (>=2.3)", "coverage (>=7.6.1)", "pytest-mock (>=3.14)"]
+testing = ["covdefaults (>=2.3)", "coverage (>=7.6.1)", "pytest-mock (>=3.14)"]
 
 [[package]]
 name = "pytest-mock"
@@ -7550,123 +7608,103 @@ dev = ["pytest"]
 
 [[package]]
 name = "rapidfuzz"
-version = "3.9.7"
+version = "3.10.0"
 description = "rapid fuzzy string matching"
 optional = false
-python-versions = ">=3.8"
+python-versions = ">=3.9"
 files = [
-    {file = "rapidfuzz-3.9.7-cp310-cp310-macosx_10_9_x86_64.whl", hash = "sha256:ccf68e30b80e903f2309f90a438dbd640dd98e878eeb5ad361a288051ee5b75c"},
-    {file = "rapidfuzz-3.9.7-cp310-cp310-macosx_11_0_arm64.whl", hash = "sha256:696a79018ef989bf1c9abd9005841cee18005ccad4748bad8a4c274c47b6241a"},
-    {file = "rapidfuzz-3.9.7-cp310-cp310-manylinux_2_17_aarch64.manylinux2014_aarch64.whl", hash = "sha256:c4eebf6c93af0ae866c22b403a84747580bb5c10f0d7b51c82a87f25405d4dcb"},
-    {file = "rapidfuzz-3.9.7-cp310-cp310-manylinux_2_17_i686.manylinux2014_i686.whl", hash = "sha256:0e9125377fa3d21a8abd4fbdbcf1c27be73e8b1850f0b61b5b711364bf3b59db"},
-    {file = "rapidfuzz-3.9.7-cp310-cp310-manylinux_2_17_ppc64le.manylinux2014_ppc64le.whl", hash = "sha256:c12d180b17a22d107c8747de9c68d0b9c1d15dcda5445ff9bf9f4ccfb67c3e16"},
-    {file = "rapidfuzz-3.9.7-cp310-cp310-manylinux_2_17_s390x.manylinux2014_s390x.whl", hash = "sha256:c1318d42610c26dcd68bd3279a1bf9e3605377260867c9a8ed22eafc1bd93a7c"},
-    {file = "rapidfuzz-3.9.7-cp310-cp310-manylinux_2_17_x86_64.manylinux2014_x86_64.whl", hash = "sha256:dd5fa6e3c6e0333051c1f3a49f0807b3366f4131c8d6ac8c3e05fd0d0ce3755c"},
-    {file = "rapidfuzz-3.9.7-cp310-cp310-musllinux_1_2_aarch64.whl", hash = "sha256:fcf79b686962d7bec458a0babc904cb4fa319808805e036b9d5a531ee6b9b835"},
-    {file = "rapidfuzz-3.9.7-cp310-cp310-musllinux_1_2_i686.whl", hash = "sha256:8b01153c7466d0bad48fba77a303d5a768e66f24b763853469f47220b3de4661"},
-    {file = "rapidfuzz-3.9.7-cp310-cp310-musllinux_1_2_ppc64le.whl", hash = "sha256:94baaeea0b4f8632a6da69348b1e741043eba18d4e3088d674d3f76586b6223d"},
-    {file = "rapidfuzz-3.9.7-cp310-cp310-musllinux_1_2_s390x.whl", hash = "sha256:6c5b32875646cb7f60c193ade99b2e4b124f19583492115293cd00f6fb198b17"},
-    {file = "rapidfuzz-3.9.7-cp310-cp310-musllinux_1_2_x86_64.whl", hash = "sha256:110b6294396bc0a447648627479c9320f095c2034c0537f687592e0f58622638"},
-    {file = "rapidfuzz-3.9.7-cp310-cp310-win32.whl", hash = "sha256:3445a35c4c8d288f2b2011eb61bce1227c633ce85a3154e727170f37c0266bb2"},
-    {file = "rapidfuzz-3.9.7-cp310-cp310-win_amd64.whl", hash = "sha256:0d1415a732ee75e74a90af12020b77a0b396b36c60afae1bde3208a78cd2c9fc"},
-    {file = "rapidfuzz-3.9.7-cp310-cp310-win_arm64.whl", hash = "sha256:836f4d88b8bd0fff2ebe815dcaab8aa6c8d07d1d566a7e21dd137cf6fe11ed5b"},
-    {file = "rapidfuzz-3.9.7-cp311-cp311-macosx_10_9_x86_64.whl", hash = "sha256:d098ce6162eb5e48fceb0745455bc950af059df6113eec83e916c129fca11408"},
-    {file = "rapidfuzz-3.9.7-cp311-cp311-macosx_11_0_arm64.whl", hash = "sha256:048d55d36c02c6685a2b2741688503c3d15149694506655b6169dcfd3b6c2585"},
-    {file = "rapidfuzz-3.9.7-cp311-cp311-manylinux_2_17_aarch64.manylinux2014_aarch64.whl", hash = "sha256:c33211cfff9aec425bb1bfedaf94afcf337063aa273754f22779d6dadebef4c2"},
-    {file = "rapidfuzz-3.9.7-cp311-cp311-manylinux_2_17_i686.manylinux2014_i686.whl", hash = "sha256:e6d9db2fa4e9be171e9bb31cf2d2575574774966b43f5b951062bb2e67885852"},
-    {file = "rapidfuzz-3.9.7-cp311-cp311-manylinux_2_17_ppc64le.manylinux2014_ppc64le.whl", hash = "sha256:d4e049d5ad61448c9a020d1061eba20944c4887d720c4069724beb6ea1692507"},
-    {file = "rapidfuzz-3.9.7-cp311-cp311-manylinux_2_17_s390x.manylinux2014_s390x.whl", hash = "sha256:cfa74aac64c85898b93d9c80bb935a96bf64985e28d4ee0f1a3d1f3bf11a5106"},
-    {file = "rapidfuzz-3.9.7-cp311-cp311-manylinux_2_17_x86_64.manylinux2014_x86_64.whl", hash = "sha256:965693c2e9efd425b0f059f5be50ef830129f82892fa1858e220e424d9d0160f"},
-    {file = "rapidfuzz-3.9.7-cp311-cp311-musllinux_1_2_aarch64.whl", hash = "sha256:8501000a5eb8037c4b56857724797fe5a8b01853c363de91c8d0d0ad56bef319"},
-    {file = "rapidfuzz-3.9.7-cp311-cp311-musllinux_1_2_i686.whl", hash = "sha256:8d92c552c6b7577402afdd547dcf5d31ea6c8ae31ad03f78226e055cfa37f3c6"},
-    {file = "rapidfuzz-3.9.7-cp311-cp311-musllinux_1_2_ppc64le.whl", hash = "sha256:1ee2086f490cb501d86b7e386c1eb4e3a0ccbb0c99067089efaa8c79012c8952"},
-    {file = "rapidfuzz-3.9.7-cp311-cp311-musllinux_1_2_s390x.whl", hash = "sha256:1de91e7fd7f525e10ea79a6e62c559d1b0278ec097ad83d9da378b6fab65a265"},
-    {file = "rapidfuzz-3.9.7-cp311-cp311-musllinux_1_2_x86_64.whl", hash = "sha256:a4da514d13f4433e16960a17f05b67e0af30ac771719c9a9fb877e5004f74477"},
-    {file = "rapidfuzz-3.9.7-cp311-cp311-win32.whl", hash = "sha256:a40184c67db8252593ec518e17fb8a6e86d7259dc9f2d6c0bf4ff4db8cf1ad4b"},
-    {file = "rapidfuzz-3.9.7-cp311-cp311-win_amd64.whl", hash = "sha256:c4f28f1930b09a2c300357d8465b388cecb7e8b2f454a5d5425561710b7fd07f"},
-    {file = "rapidfuzz-3.9.7-cp311-cp311-win_arm64.whl", hash = "sha256:675b75412a943bb83f1f53e2e54fd18c80ef15ed642dc6eb0382d1949419d904"},
-    {file = "rapidfuzz-3.9.7-cp312-cp312-macosx_10_9_x86_64.whl", hash = "sha256:1ef6a1a8f0b12f8722f595f15c62950c9a02d5abc64742561299ffd49f6c6944"},
-    {file = "rapidfuzz-3.9.7-cp312-cp312-macosx_11_0_arm64.whl", hash = "sha256:32532af1d70c6ec02ea5ac7ee2766dfff7c8ae8c761abfe8da9e527314e634e8"},
-    {file = "rapidfuzz-3.9.7-cp312-cp312-manylinux_2_17_aarch64.manylinux2014_aarch64.whl", hash = "sha256:ae1a38bade755aa9dd95a81cda949e1bf9cd92b79341ccc5e2189c9e7bdfc5ec"},
-    {file = "rapidfuzz-3.9.7-cp312-cp312-manylinux_2_17_i686.manylinux2014_i686.whl", hash = "sha256:d73ee2df41224c87336448d279b5b6a3a75f36e41dd3dcf538c0c9cce36360d8"},
-    {file = "rapidfuzz-3.9.7-cp312-cp312-manylinux_2_17_ppc64le.manylinux2014_ppc64le.whl", hash = "sha256:be3a1fc3e2ab3bdf93dc0c83c00acca8afd2a80602297d96cf4a0ba028333cdf"},
-    {file = "rapidfuzz-3.9.7-cp312-cp312-manylinux_2_17_s390x.manylinux2014_s390x.whl", hash = "sha256:603f48f621272a448ff58bb556feb4371252a02156593303391f5c3281dfaeac"},
-    {file = "rapidfuzz-3.9.7-cp312-cp312-manylinux_2_17_x86_64.manylinux2014_x86_64.whl", hash = "sha256:268f8e1ca50fc61c0736f3fe9d47891424adf62d96ed30196f30f4bd8216b41f"},
-    {file = "rapidfuzz-3.9.7-cp312-cp312-musllinux_1_2_aarch64.whl", hash = "sha256:5f8bf3f0d02935751d8660abda6044821a861f6229f7d359f98bcdcc7e66c39b"},
-    {file = "rapidfuzz-3.9.7-cp312-cp312-musllinux_1_2_i686.whl", hash = "sha256:b997ff3b39d4cee9fb025d6c46b0a24bd67595ce5a5b652a97fb3a9d60beb651"},
-    {file = "rapidfuzz-3.9.7-cp312-cp312-musllinux_1_2_ppc64le.whl", hash = "sha256:ca66676c8ef6557f9b81c5b2b519097817a7c776a6599b8d6fcc3e16edd216fe"},
-    {file = "rapidfuzz-3.9.7-cp312-cp312-musllinux_1_2_s390x.whl", hash = "sha256:35d3044cb635ca6b1b2b7b67b3597bd19f34f1753b129eb6d2ae04cf98cd3945"},
-    {file = "rapidfuzz-3.9.7-cp312-cp312-musllinux_1_2_x86_64.whl", hash = "sha256:5a93c9e60904cb76e7aefef67afffb8b37c4894f81415ed513db090f29d01101"},
-    {file = "rapidfuzz-3.9.7-cp312-cp312-win32.whl", hash = "sha256:579d107102c0725f7c79b4e79f16d3cf4d7c9208f29c66b064fa1fd4641d5155"},
-    {file = "rapidfuzz-3.9.7-cp312-cp312-win_amd64.whl", hash = "sha256:953b3780765c8846866faf891ee4290f6a41a6dacf4fbcd3926f78c9de412ca6"},
-    {file = "rapidfuzz-3.9.7-cp312-cp312-win_arm64.whl", hash = "sha256:7c20c1474b068c4bd45bf2fd0ad548df284f74e9a14a68b06746c56e3aa8eb70"},
-    {file = "rapidfuzz-3.9.7-cp313-cp313-macosx_10_13_x86_64.whl", hash = "sha256:fde81b1da9a947f931711febe2e2bee694e891f6d3e6aa6bc02c1884702aea19"},
-    {file = "rapidfuzz-3.9.7-cp313-cp313-macosx_11_0_arm64.whl", hash = "sha256:47e92c155a14f44511ea8ebcc6bc1535a1fe8d0a7d67ad3cc47ba61606df7bcf"},
-    {file = "rapidfuzz-3.9.7-cp313-cp313-manylinux_2_17_aarch64.manylinux2014_aarch64.whl", hash = "sha256:8772b745668260c5c4d069c678bbaa68812e6c69830f3771eaad521af7bc17f8"},
-    {file = "rapidfuzz-3.9.7-cp313-cp313-manylinux_2_17_i686.manylinux2014_i686.whl", hash = "sha256:578302828dd97ee2ba507d2f71d62164e28d2fc7bc73aad0d2d1d2afc021a5d5"},
-    {file = "rapidfuzz-3.9.7-cp313-cp313-manylinux_2_17_ppc64le.manylinux2014_ppc64le.whl", hash = "sha256:fc3e6081069eea61593f1d6839029da53d00c8c9b205c5534853eaa3f031085c"},
-    {file = "rapidfuzz-3.9.7-cp313-cp313-manylinux_2_17_s390x.manylinux2014_s390x.whl", hash = "sha256:0b1c2d504eddf97bc0f2eba422c8915576dbf025062ceaca2d68aecd66324ad9"},
-    {file = "rapidfuzz-3.9.7-cp313-cp313-manylinux_2_17_x86_64.manylinux2014_x86_64.whl", hash = "sha256:6fb76e5a21034f0307c51c5a2fc08856f698c53a4c593b17d291f7d6e9d09ca3"},
-    {file = "rapidfuzz-3.9.7-cp313-cp313-musllinux_1_2_aarch64.whl", hash = "sha256:d4ba2318ef670ce505f42881a5d2af70f948124646947341a3c6ccb33cd70369"},
-    {file = "rapidfuzz-3.9.7-cp313-cp313-musllinux_1_2_i686.whl", hash = "sha256:057bb03f39e285047d7e9412e01ecf31bb2d42b9466a5409d715d587460dd59b"},
-    {file = "rapidfuzz-3.9.7-cp313-cp313-musllinux_1_2_ppc64le.whl", hash = "sha256:a8feac9006d5c9758438906f093befffc4290de75663dbb2098461df7c7d28dd"},
-    {file = "rapidfuzz-3.9.7-cp313-cp313-musllinux_1_2_s390x.whl", hash = "sha256:95b8292383e717e10455f2c917df45032b611141e43d1adf70f71b1566136b11"},
-    {file = "rapidfuzz-3.9.7-cp313-cp313-musllinux_1_2_x86_64.whl", hash = "sha256:e9fbf659537d246086d0297628b3795dc3e4a384101ecc01e5791c827b8d7345"},
-    {file = "rapidfuzz-3.9.7-cp313-cp313-win32.whl", hash = "sha256:1dc516ac6d32027be2b0196bedf6d977ac26debd09ca182376322ad620460feb"},
-    {file = "rapidfuzz-3.9.7-cp313-cp313-win_amd64.whl", hash = "sha256:b4f86e09d3064dca0b014cd48688964036a904a2d28048f00c8f4640796d06a8"},
-    {file = "rapidfuzz-3.9.7-cp313-cp313-win_arm64.whl", hash = "sha256:19c64d8ddb2940b42a4567b23f1681af77f50a5ff6c9b8e85daba079c210716e"},
-    {file = "rapidfuzz-3.9.7-cp38-cp38-macosx_10_9_x86_64.whl", hash = "sha256:fbda3dd68d8b28ccb20ffb6f756fefd9b5ba570a772bedd7643ed441f5793308"},
-    {file = "rapidfuzz-3.9.7-cp38-cp38-macosx_11_0_arm64.whl", hash = "sha256:2379e0b2578ad3ac7004f223251550f08bca873ff76c169b09410ec562ad78d8"},
-    {file = "rapidfuzz-3.9.7-cp38-cp38-manylinux_2_17_aarch64.manylinux2014_aarch64.whl", hash = "sha256:5d1eff95362f993b0276fd3839aee48625b09aac8938bb0c23b40d219cba5dc5"},
-    {file = "rapidfuzz-3.9.7-cp38-cp38-manylinux_2_17_i686.manylinux2014_i686.whl", hash = "sha256:cd9360e30041690912525a210e48a897b49b230768cc8af1c702e5395690464f"},
-    {file = "rapidfuzz-3.9.7-cp38-cp38-manylinux_2_17_ppc64le.manylinux2014_ppc64le.whl", hash = "sha256:a93cd834b3c315ab437f0565ee3a2f42dd33768dc885ccbabf9710b131cf70d2"},
-    {file = "rapidfuzz-3.9.7-cp38-cp38-manylinux_2_17_s390x.manylinux2014_s390x.whl", hash = "sha256:4ff196996240db7075f62c7bc4506f40a3c80cd4ae3ab0e79ac6892283a90859"},
-    {file = "rapidfuzz-3.9.7-cp38-cp38-manylinux_2_17_x86_64.manylinux2014_x86_64.whl", hash = "sha256:948dcee7aaa1cd14358b2a7ef08bf0be42bf89049c3a906669874a715fc2c937"},
-    {file = "rapidfuzz-3.9.7-cp38-cp38-musllinux_1_2_aarch64.whl", hash = "sha256:d95751f505a301af1aaf086c19f34536056d6c8efa91b2240de532a3db57b543"},
-    {file = "rapidfuzz-3.9.7-cp38-cp38-musllinux_1_2_i686.whl", hash = "sha256:90db86fa196eecf96cb6db09f1083912ea945c50c57188039392d810d0b784e1"},
-    {file = "rapidfuzz-3.9.7-cp38-cp38-musllinux_1_2_ppc64le.whl", hash = "sha256:3171653212218a162540a3c8eb8ae7d3dcc8548540b69eaecaf3b47c14d89c90"},
-    {file = "rapidfuzz-3.9.7-cp38-cp38-musllinux_1_2_s390x.whl", hash = "sha256:36dd6e820379c37a1ffefc8a52b648758e867cd9d78ee5b5dc0c9a6a10145378"},
-    {file = "rapidfuzz-3.9.7-cp38-cp38-musllinux_1_2_x86_64.whl", hash = "sha256:7b702de95666a1f7d5c6b47eacadfe2d2794af3742d63d2134767d13e5d1c713"},
-    {file = "rapidfuzz-3.9.7-cp38-cp38-win32.whl", hash = "sha256:9030e7238c0df51aed5c9c5ed8eee2bdd47a2ae788e562c1454af2851c3d1906"},
-    {file = "rapidfuzz-3.9.7-cp38-cp38-win_amd64.whl", hash = "sha256:f847fb0fbfb72482b1c05c59cbb275c58a55b73708a7f77a83f8035ee3c86497"},
-    {file = "rapidfuzz-3.9.7-cp39-cp39-macosx_10_9_x86_64.whl", hash = "sha256:97f2ce529d2a70a60c290f6ab269a2bbf1d3b47b9724dccc84339b85f7afb044"},
-    {file = "rapidfuzz-3.9.7-cp39-cp39-macosx_11_0_arm64.whl", hash = "sha256:e2957fdad10bb83b1982b02deb3604a3f6911a5e545f518b59c741086f92d152"},
-    {file = "rapidfuzz-3.9.7-cp39-cp39-manylinux_2_17_aarch64.manylinux2014_aarch64.whl", hash = "sha256:5d5262383634626eb45c536017204b8163a03bc43bda880cf1bdd7885db9a163"},
-    {file = "rapidfuzz-3.9.7-cp39-cp39-manylinux_2_17_i686.manylinux2014_i686.whl", hash = "sha256:364587827d7cbd41afa0782adc2d2d19e3f07d355b0750a02a8e33ad27a9c368"},
-    {file = "rapidfuzz-3.9.7-cp39-cp39-manylinux_2_17_ppc64le.manylinux2014_ppc64le.whl", hash = "sha256:ecc24af7f905f3d6efb371a01680116ffea8d64e266618fb9ad1602a9b4f7934"},
-    {file = "rapidfuzz-3.9.7-cp39-cp39-manylinux_2_17_s390x.manylinux2014_s390x.whl", hash = "sha256:9dc86aa6b29d174713c5f4caac35ffb7f232e3e649113e8d13812b35ab078228"},
-    {file = "rapidfuzz-3.9.7-cp39-cp39-manylinux_2_17_x86_64.manylinux2014_x86_64.whl", hash = "sha256:e3dcfbe7266e74a707173a12a7b355a531f2dcfbdb32f09468e664330da14874"},
-    {file = "rapidfuzz-3.9.7-cp39-cp39-musllinux_1_2_aarch64.whl", hash = "sha256:b23806fbdd6b510ba9ac93bb72d503066263b0fba44b71b835be9f063a84025f"},
-    {file = "rapidfuzz-3.9.7-cp39-cp39-musllinux_1_2_i686.whl", hash = "sha256:5551d68264c1bb6943f542da83a4dc8940ede52c5847ef158698799cc28d14f5"},
-    {file = "rapidfuzz-3.9.7-cp39-cp39-musllinux_1_2_ppc64le.whl", hash = "sha256:13d8675a1fa7e2b19650ca7ef9a6ec01391d4bb12ab9e0793e8eb024538b4a34"},
-    {file = "rapidfuzz-3.9.7-cp39-cp39-musllinux_1_2_s390x.whl", hash = "sha256:9b6a5de507b9be6de688dae40143b656f7a93b10995fb8bd90deb555e7875c60"},
-    {file = "rapidfuzz-3.9.7-cp39-cp39-musllinux_1_2_x86_64.whl", hash = "sha256:111a20a3c090cf244d9406e60500b6c34b2375ba3a5009e2b38fd806fe38e337"},
-    {file = "rapidfuzz-3.9.7-cp39-cp39-win32.whl", hash = "sha256:22589c0b8ccc6c391ce7f776c93a8c92c96ab8d34e1a19f1bd2b12a235332632"},
-    {file = "rapidfuzz-3.9.7-cp39-cp39-win_amd64.whl", hash = "sha256:6f83221db5755b8f34222e40607d87f1176a8d5d4dbda4a55a0f0b67d588a69c"},
-    {file = "rapidfuzz-3.9.7-cp39-cp39-win_arm64.whl", hash = "sha256:3665b92e788578c3bb334bd5b5fa7ee1a84bafd68be438e3110861d1578c63a0"},
-    {file = "rapidfuzz-3.9.7-pp310-pypy310_pp73-macosx_10_15_x86_64.whl", hash = "sha256:d7df9c2194c7ec930b33c991c55dbd0c10951bd25800c0b7a7b571994ebbced5"},
-    {file = "rapidfuzz-3.9.7-pp310-pypy310_pp73-macosx_11_0_arm64.whl", hash = "sha256:68bd888eafd07b09585dcc8bc2716c5ecdb7eed62827470664d25588982b2873"},
-    {file = "rapidfuzz-3.9.7-pp310-pypy310_pp73-manylinux_2_17_aarch64.manylinux2014_aarch64.whl", hash = "sha256:d1230e0f9026851a6a432beaa0ce575dda7b39fe689b576f99a0704fbb81fc9c"},
-    {file = "rapidfuzz-3.9.7-pp310-pypy310_pp73-manylinux_2_17_i686.manylinux2014_i686.whl", hash = "sha256:a3b36e1c61b796ae1777f3e9e11fd39898b09d351c9384baf6e3b7e6191d8ced"},
-    {file = "rapidfuzz-3.9.7-pp310-pypy310_pp73-manylinux_2_17_x86_64.manylinux2014_x86_64.whl", hash = "sha256:9dba13d86806fcf3fe9c9919f58575e0090eadfb89c058bde02bcc7ab24e4548"},
-    {file = "rapidfuzz-3.9.7-pp310-pypy310_pp73-win_amd64.whl", hash = "sha256:1f1a33e84056b7892c721d84475d3bde49a145126bc4c6efe0d6d0d59cb31c29"},
-    {file = "rapidfuzz-3.9.7-pp38-pypy38_pp73-macosx_10_9_x86_64.whl", hash = "sha256:3492c7a42b7fa9f0051d7fcce9893e95ed91c97c9ec7fb64346f3e070dd318ed"},
-    {file = "rapidfuzz-3.9.7-pp38-pypy38_pp73-macosx_11_0_arm64.whl", hash = "sha256:ece45eb2af8b00f90d10f7419322e8804bd42fb1129026f9bfe712c37508b514"},
-    {file = "rapidfuzz-3.9.7-pp38-pypy38_pp73-manylinux_2_17_aarch64.manylinux2014_aarch64.whl", hash = "sha256:9dcd14cf4876f04b488f6e54a7abd3e9b31db5f5a6aba0ce90659917aaa8c088"},
-    {file = "rapidfuzz-3.9.7-pp38-pypy38_pp73-manylinux_2_17_i686.manylinux2014_i686.whl", hash = "sha256:521c58c72ed8a612b25cda378ff10dee17e6deb4ee99a070b723519a345527b9"},
-    {file = "rapidfuzz-3.9.7-pp38-pypy38_pp73-manylinux_2_17_x86_64.manylinux2014_x86_64.whl", hash = "sha256:18669bb6cdf7d40738526d37e550df09ba065b5a7560f3d802287988b6cb63cf"},
-    {file = "rapidfuzz-3.9.7-pp38-pypy38_pp73-win_amd64.whl", hash = "sha256:7abe2dbae81120a64bb4f8d3fcafe9122f328c9f86d7f327f174187a5af4ed86"},
-    {file = "rapidfuzz-3.9.7-pp39-pypy39_pp73-macosx_10_15_x86_64.whl", hash = "sha256:a3c0783910911f4f24655826d007c9f4360f08107410952c01ee3df98c713eb2"},
-    {file = "rapidfuzz-3.9.7-pp39-pypy39_pp73-macosx_11_0_arm64.whl", hash = "sha256:03126f9a040ff21d2a110610bfd6b93b79377ce8b4121edcb791d61b7df6eec5"},
-    {file = "rapidfuzz-3.9.7-pp39-pypy39_pp73-manylinux_2_17_aarch64.manylinux2014_aarch64.whl", hash = "sha256:591908240f4085e2ade5b685c6e8346e2ed44932cffeaac2fb32ddac95b55c7f"},
-    {file = "rapidfuzz-3.9.7-pp39-pypy39_pp73-manylinux_2_17_i686.manylinux2014_i686.whl", hash = "sha256:e9012d86c6397edbc9da4ac0132de7f8ee9d6ce857f4194d5684c4ddbcdd1c5c"},
-    {file = "rapidfuzz-3.9.7-pp39-pypy39_pp73-manylinux_2_17_x86_64.manylinux2014_x86_64.whl", hash = "sha256:df596ddd3db38aa513d4c0995611267b3946e7cbe5a8761b50e9306dfec720ee"},
-    {file = "rapidfuzz-3.9.7-pp39-pypy39_pp73-win_amd64.whl", hash = "sha256:3ed5adb752f4308fcc8f4fb6f8eb7aa4082f9d12676fda0a74fa5564242a8107"},
-    {file = "rapidfuzz-3.9.7.tar.gz", hash = "sha256:f1c7296534c1afb6f495aa95871f14ccdc197c6db42965854e483100df313030"},
-]
-
-[package.extras]
-full = ["numpy"]
+    {file = "rapidfuzz-3.10.0-cp310-cp310-macosx_10_9_x86_64.whl", hash = "sha256:884453860de029380dded8f3c1918af2d8eb5adf8010261645c7e5c88c2b5428"},
+    {file = "rapidfuzz-3.10.0-cp310-cp310-macosx_11_0_arm64.whl", hash = "sha256:718c9bd369288aca5fa929df6dbf66fdbe9768d90940a940c0b5cdc96ade4309"},
+    {file = "rapidfuzz-3.10.0-cp310-cp310-manylinux_2_17_aarch64.manylinux2014_aarch64.whl", hash = "sha256:a68e3724b7dab761c01816aaa64b0903734d999d5589daf97c14ef5cc0629a8e"},
+    {file = "rapidfuzz-3.10.0-cp310-cp310-manylinux_2_17_i686.manylinux2014_i686.whl", hash = "sha256:1af60988d47534246d9525f77288fdd9de652608a4842815d9018570b959acc6"},
+    {file = "rapidfuzz-3.10.0-cp310-cp310-manylinux_2_17_ppc64le.manylinux2014_ppc64le.whl", hash = "sha256:3084161fc3e963056232ef8d937449a2943852e07101f5a136c8f3cfa4119217"},
+    {file = "rapidfuzz-3.10.0-cp310-cp310-manylinux_2_17_s390x.manylinux2014_s390x.whl", hash = "sha256:6cd67d3d017296d98ff505529104299f78433e4b8af31b55003d901a62bbebe9"},
+    {file = "rapidfuzz-3.10.0-cp310-cp310-manylinux_2_17_x86_64.manylinux2014_x86_64.whl", hash = "sha256:b11a127ac590fc991e8a02c2d7e1ac86e8141c92f78546f18b5c904064a0552c"},
+    {file = "rapidfuzz-3.10.0-cp310-cp310-musllinux_1_2_aarch64.whl", hash = "sha256:aadce42147fc09dcef1afa892485311e824c050352e1aa6e47f56b9b27af4cf0"},
+    {file = "rapidfuzz-3.10.0-cp310-cp310-musllinux_1_2_i686.whl", hash = "sha256:b54853c2371bf0e38d67da379519deb6fbe70055efb32f6607081641af3dc752"},
+    {file = "rapidfuzz-3.10.0-cp310-cp310-musllinux_1_2_ppc64le.whl", hash = "sha256:ce19887268e90ee81a3957eef5e46a70ecc000713796639f83828b950343f49e"},
+    {file = "rapidfuzz-3.10.0-cp310-cp310-musllinux_1_2_s390x.whl", hash = "sha256:f39a2a5ded23b9b9194ec45740dce57177b80f86c6d8eba953d3ff1a25c97766"},
+    {file = "rapidfuzz-3.10.0-cp310-cp310-musllinux_1_2_x86_64.whl", hash = "sha256:0ec338d5f4ad8d9339a88a08db5c23e7f7a52c2b2a10510c48a0cef1fb3f0ddc"},
+    {file = "rapidfuzz-3.10.0-cp310-cp310-win32.whl", hash = "sha256:56fd15ea8f4c948864fa5ebd9261c67cf7b89a1c517a0caef4df75446a7af18c"},
+    {file = "rapidfuzz-3.10.0-cp310-cp310-win_amd64.whl", hash = "sha256:43dfc5e733808962a822ff6d9c29f3039a3cfb3620706f5953e17cfe4496724c"},
+    {file = "rapidfuzz-3.10.0-cp310-cp310-win_arm64.whl", hash = "sha256:ae7966f205b5a7fde93b44ca8fed37c1c8539328d7f179b1197de34eceaceb5f"},
+    {file = "rapidfuzz-3.10.0-cp311-cp311-macosx_10_9_x86_64.whl", hash = "sha256:bb0013795b40db5cf361e6f21ee7cda09627cf294977149b50e217d7fe9a2f03"},
+    {file = "rapidfuzz-3.10.0-cp311-cp311-macosx_11_0_arm64.whl", hash = "sha256:69ef5b363afff7150a1fbe788007e307b9802a2eb6ad92ed51ab94e6ad2674c6"},
+    {file = "rapidfuzz-3.10.0-cp311-cp311-manylinux_2_17_aarch64.manylinux2014_aarch64.whl", hash = "sha256:c582c46b1bb0b19f1a5f4c1312f1b640c21d78c371a6615c34025b16ee56369b"},
+    {file = "rapidfuzz-3.10.0-cp311-cp311-manylinux_2_17_i686.manylinux2014_i686.whl", hash = "sha256:288f6f6e7410cacb115fb851f3f18bf0e4231eb3f6cb5bd1cec0e7b25c4d039d"},
+    {file = "rapidfuzz-3.10.0-cp311-cp311-manylinux_2_17_ppc64le.manylinux2014_ppc64le.whl", hash = "sha256:c9e29a13d2fd9be3e7d8c26c7ef4ba60b5bc7efbc9dbdf24454c7e9ebba31768"},
+    {file = "rapidfuzz-3.10.0-cp311-cp311-manylinux_2_17_s390x.manylinux2014_s390x.whl", hash = "sha256:ea2da0459b951ee461bd4e02b8904890bd1c4263999d291c5cd01e6620177ad4"},
+    {file = "rapidfuzz-3.10.0-cp311-cp311-manylinux_2_17_x86_64.manylinux2014_x86_64.whl", hash = "sha256:457827ba82261aa2ae6ac06a46d0043ab12ba7216b82d87ae1434ec0f29736d6"},
+    {file = "rapidfuzz-3.10.0-cp311-cp311-musllinux_1_2_aarch64.whl", hash = "sha256:5d350864269d56f51ab81ab750c9259ae5cad3152c0680baef143dcec92206a1"},
+    {file = "rapidfuzz-3.10.0-cp311-cp311-musllinux_1_2_i686.whl", hash = "sha256:a9b8f51e08c3f983d857c3889930af9ddecc768453822076683664772d87e374"},
+    {file = "rapidfuzz-3.10.0-cp311-cp311-musllinux_1_2_ppc64le.whl", hash = "sha256:7f3a6aa6e70fc27e4ff5c479f13cc9fc26a56347610f5f8b50396a0d344c5f55"},
+    {file = "rapidfuzz-3.10.0-cp311-cp311-musllinux_1_2_s390x.whl", hash = "sha256:803f255f10d63420979b1909ef976e7d30dec42025c9b067fc1d2040cc365a7e"},
+    {file = "rapidfuzz-3.10.0-cp311-cp311-musllinux_1_2_x86_64.whl", hash = "sha256:2026651761bf83a0f31495cc0f70840d5c0d54388f41316e3f9cb51bd85e49a5"},
+    {file = "rapidfuzz-3.10.0-cp311-cp311-win32.whl", hash = "sha256:4df75b3ebbb8cfdb9bf8b213b168620b88fd92d0c16a8bc9f9234630b282db59"},
+    {file = "rapidfuzz-3.10.0-cp311-cp311-win_amd64.whl", hash = "sha256:f9f0bbfb6787b97c51516f3ccf97737d504db5d239ad44527673b81f598b84ab"},
+    {file = "rapidfuzz-3.10.0-cp311-cp311-win_arm64.whl", hash = "sha256:10fdad800441b9c97d471a937ba7d42625f1b530db05e572f1cb7d401d95c893"},
+    {file = "rapidfuzz-3.10.0-cp312-cp312-macosx_10_13_x86_64.whl", hash = "sha256:7dc87073ba3a40dd65591a2100aa71602107443bf10770579ff9c8a3242edb94"},
+    {file = "rapidfuzz-3.10.0-cp312-cp312-macosx_11_0_arm64.whl", hash = "sha256:a425a0a868cf8e9c6e93e1cda4b758cdfd314bb9a4fc916c5742c934e3613480"},
+    {file = "rapidfuzz-3.10.0-cp312-cp312-manylinux_2_17_aarch64.manylinux2014_aarch64.whl", hash = "sha256:a86d5d1d75e61df060c1e56596b6b0a4422a929dff19cc3dbfd5eee762c86b61"},
+    {file = "rapidfuzz-3.10.0-cp312-cp312-manylinux_2_17_i686.manylinux2014_i686.whl", hash = "sha256:34f213d59219a9c3ca14e94a825f585811a68ac56b4118b4dc388b5b14afc108"},
+    {file = "rapidfuzz-3.10.0-cp312-cp312-manylinux_2_17_ppc64le.manylinux2014_ppc64le.whl", hash = "sha256:96ad46f5f56f70fab2be9e5f3165a21be58d633b90bf6e67fc52a856695e4bcf"},
+    {file = "rapidfuzz-3.10.0-cp312-cp312-manylinux_2_17_s390x.manylinux2014_s390x.whl", hash = "sha256:9178277f72d144a6c7704d7ae7fa15b7b86f0f0796f0e1049c7b4ef748a662ef"},
+    {file = "rapidfuzz-3.10.0-cp312-cp312-manylinux_2_17_x86_64.manylinux2014_x86_64.whl", hash = "sha256:76a35e9e19a7c883c422ffa378e9a04bc98cb3b29648c5831596401298ee51e6"},
+    {file = "rapidfuzz-3.10.0-cp312-cp312-musllinux_1_2_aarch64.whl", hash = "sha256:8a6405d34c394c65e4f73a1d300c001f304f08e529d2ed6413b46ee3037956eb"},
+    {file = "rapidfuzz-3.10.0-cp312-cp312-musllinux_1_2_i686.whl", hash = "sha256:bd393683129f446a75d8634306aed7e377627098a1286ff3af2a4f1736742820"},
+    {file = "rapidfuzz-3.10.0-cp312-cp312-musllinux_1_2_ppc64le.whl", hash = "sha256:b0445fa9880ead81f5a7d0efc0b9c977a947d8052c43519aceeaf56eabaf6843"},
+    {file = "rapidfuzz-3.10.0-cp312-cp312-musllinux_1_2_s390x.whl", hash = "sha256:c50bc308fa29767ed8f53a8d33b7633a9e14718ced038ed89d41b886e301da32"},
+    {file = "rapidfuzz-3.10.0-cp312-cp312-musllinux_1_2_x86_64.whl", hash = "sha256:e89605afebbd2d4b045bccfdc12a14b16fe8ccbae05f64b4b4c64a97dad1c891"},
+    {file = "rapidfuzz-3.10.0-cp312-cp312-win32.whl", hash = "sha256:2db9187f3acf3cd33424ecdbaad75414c298ecd1513470df7bda885dcb68cc15"},
+    {file = "rapidfuzz-3.10.0-cp312-cp312-win_amd64.whl", hash = "sha256:50e3d0c72ea15391ba9531ead7f2068a67c5b18a6a365fef3127583aaadd1725"},
+    {file = "rapidfuzz-3.10.0-cp312-cp312-win_arm64.whl", hash = "sha256:9eac95b4278bd53115903d89118a2c908398ee8bdfd977ae844f1bd2b02b917c"},
+    {file = "rapidfuzz-3.10.0-cp313-cp313-macosx_10_13_x86_64.whl", hash = "sha256:fe5231e8afd069c742ac5b4f96344a0fe4aff52df8e53ef87faebf77f827822c"},
+    {file = "rapidfuzz-3.10.0-cp313-cp313-macosx_11_0_arm64.whl", hash = "sha256:886882367dbc985f5736356105798f2ae6e794e671fc605476cbe2e73838a9bb"},
+    {file = "rapidfuzz-3.10.0-cp313-cp313-manylinux_2_17_aarch64.manylinux2014_aarch64.whl", hash = "sha256:b33e13e537e3afd1627d421a142a12bbbe601543558a391a6fae593356842f6e"},
+    {file = "rapidfuzz-3.10.0-cp313-cp313-manylinux_2_17_i686.manylinux2014_i686.whl", hash = "sha256:094c26116d55bf9c53abd840d08422f20da78ec4c4723e5024322321caedca48"},
+    {file = "rapidfuzz-3.10.0-cp313-cp313-manylinux_2_17_ppc64le.manylinux2014_ppc64le.whl", hash = "sha256:545fc04f2d592e4350f59deb0818886c1b444ffba3bec535b4fbb97191aaf769"},
+    {file = "rapidfuzz-3.10.0-cp313-cp313-manylinux_2_17_s390x.manylinux2014_s390x.whl", hash = "sha256:916a6abf3632e592b937c3d04c00a6efadd8fd30539cdcd4e6e4d92be7ca5d90"},
+    {file = "rapidfuzz-3.10.0-cp313-cp313-manylinux_2_17_x86_64.manylinux2014_x86_64.whl", hash = "sha256:fb6ec40cef63b1922083d33bfef2f91fc0b0bc07b5b09bfee0b0f1717d558292"},
+    {file = "rapidfuzz-3.10.0-cp313-cp313-musllinux_1_2_aarch64.whl", hash = "sha256:c77a7330dd15c7eb5fd3631dc646fc96327f98db8181138766bd14d3e905f0ba"},
+    {file = "rapidfuzz-3.10.0-cp313-cp313-musllinux_1_2_i686.whl", hash = "sha256:949b5e9eeaa4ecb4c7e9c2a4689dddce60929dd1ff9c76a889cdbabe8bbf2171"},
+    {file = "rapidfuzz-3.10.0-cp313-cp313-musllinux_1_2_ppc64le.whl", hash = "sha256:b5363932a5aab67010ae1a6205c567d1ef256fb333bc23c27582481606be480c"},
+    {file = "rapidfuzz-3.10.0-cp313-cp313-musllinux_1_2_s390x.whl", hash = "sha256:5dd6eec15b13329abe66cc241b484002ecb0e17d694491c944a22410a6a9e5e2"},
+    {file = "rapidfuzz-3.10.0-cp313-cp313-musllinux_1_2_x86_64.whl", hash = "sha256:79e7f98525b60b3c14524e0a4e1fedf7654657b6e02eb25f1be897ab097706f3"},
+    {file = "rapidfuzz-3.10.0-cp313-cp313-win32.whl", hash = "sha256:d29d1b9857c65f8cb3a29270732e1591b9bacf89de9d13fa764f79f07d8f1fd2"},
+    {file = "rapidfuzz-3.10.0-cp313-cp313-win_amd64.whl", hash = "sha256:fa9720e56663cc3649d62b4b5f3145e94b8f5611e8a8e1b46507777249d46aad"},
+    {file = "rapidfuzz-3.10.0-cp313-cp313-win_arm64.whl", hash = "sha256:eda4c661e68dddd56c8fbfe1ca35e40dd2afd973f7ebb1605f4d151edc63dff8"},
+    {file = "rapidfuzz-3.10.0-cp39-cp39-macosx_10_9_x86_64.whl", hash = "sha256:cffbc50e0767396ed483900900dd58ce4351bc0d40e64bced8694bd41864cc71"},
+    {file = "rapidfuzz-3.10.0-cp39-cp39-macosx_11_0_arm64.whl", hash = "sha256:c038b9939da3035afb6cb2f465f18163e8f070aba0482923ecff9443def67178"},
+    {file = "rapidfuzz-3.10.0-cp39-cp39-manylinux_2_17_aarch64.manylinux2014_aarch64.whl", hash = "sha256:ca366c2e2a54e2f663f4529b189fdeb6e14d419b1c78b754ec1744f3c01070d4"},
+    {file = "rapidfuzz-3.10.0-cp39-cp39-manylinux_2_17_i686.manylinux2014_i686.whl", hash = "sha256:7c4c82b1689b23b1b5e6a603164ed2be41b6f6de292a698b98ba2381e889eb9d"},
+    {file = "rapidfuzz-3.10.0-cp39-cp39-manylinux_2_17_ppc64le.manylinux2014_ppc64le.whl", hash = "sha256:98f6ebe28831a482981ecfeedc8237047878424ad0c1add2c7f366ba44a20452"},
+    {file = "rapidfuzz-3.10.0-cp39-cp39-manylinux_2_17_s390x.manylinux2014_s390x.whl", hash = "sha256:4bd1a7676ee2a4c8e2f7f2550bece994f9f89e58afb96088964145a83af7408b"},
+    {file = "rapidfuzz-3.10.0-cp39-cp39-manylinux_2_17_x86_64.manylinux2014_x86_64.whl", hash = "sha256:ec9139baa3f85b65adc700eafa03ed04995ca8533dd56c924f0e458ffec044ab"},
+    {file = "rapidfuzz-3.10.0-cp39-cp39-musllinux_1_2_aarch64.whl", hash = "sha256:26de93e6495078b6af4c4d93a42ca067b16cc0e95699526c82ab7d1025b4d3bf"},
+    {file = "rapidfuzz-3.10.0-cp39-cp39-musllinux_1_2_i686.whl", hash = "sha256:f3a0bda83c18195c361b5500377d0767749f128564ca95b42c8849fd475bb327"},
+    {file = "rapidfuzz-3.10.0-cp39-cp39-musllinux_1_2_ppc64le.whl", hash = "sha256:63e4c175cbce8c3adc22dca5e6154588ae673f6c55374d156f3dac732c88d7de"},
+    {file = "rapidfuzz-3.10.0-cp39-cp39-musllinux_1_2_s390x.whl", hash = "sha256:4dd3d8443970eaa02ab5ae45ce584b061f2799cd9f7e875190e2617440c1f9d4"},
+    {file = "rapidfuzz-3.10.0-cp39-cp39-musllinux_1_2_x86_64.whl", hash = "sha256:e5ddb2388610799fc46abe389600625058f2a73867e63e20107c5ad5ffa57c47"},
+    {file = "rapidfuzz-3.10.0-cp39-cp39-win32.whl", hash = "sha256:2e9be5d05cd960914024412b5406fb75a82f8562f45912ff86255acbfdbfb78e"},
+    {file = "rapidfuzz-3.10.0-cp39-cp39-win_amd64.whl", hash = "sha256:47aca565a39c9a6067927871973ca827023e8b65ba6c5747f4c228c8d7ddc04f"},
+    {file = "rapidfuzz-3.10.0-cp39-cp39-win_arm64.whl", hash = "sha256:b0732343cdc4273b5921268026dd7266f75466eb21873cb7635a200d9d9c3fac"},
+    {file = "rapidfuzz-3.10.0-pp310-pypy310_pp73-macosx_10_15_x86_64.whl", hash = "sha256:f744b5eb1469bf92dd143d36570d2bdbbdc88fe5cb0b5405e53dd34f479cbd8a"},
+    {file = "rapidfuzz-3.10.0-pp310-pypy310_pp73-macosx_11_0_arm64.whl", hash = "sha256:b67cc21a14327a0eb0f47bc3d7e59ec08031c7c55220ece672f9476e7a8068d3"},
+    {file = "rapidfuzz-3.10.0-pp310-pypy310_pp73-manylinux_2_17_aarch64.manylinux2014_aarch64.whl", hash = "sha256:2fe5783676f0afba4a522c80b15e99dbf4e393c149ab610308a8ef1f04c6bcc8"},
+    {file = "rapidfuzz-3.10.0-pp310-pypy310_pp73-manylinux_2_17_i686.manylinux2014_i686.whl", hash = "sha256:d4688862f957c8629d557d084f20b2d803f8738b6c4066802a0b1cc472e088d9"},
+    {file = "rapidfuzz-3.10.0-pp310-pypy310_pp73-manylinux_2_17_x86_64.manylinux2014_x86_64.whl", hash = "sha256:20bd153aacc244e4c907d772c703fea82754c4db14f8aa64d75ff81b7b8ab92d"},
+    {file = "rapidfuzz-3.10.0-pp310-pypy310_pp73-win_amd64.whl", hash = "sha256:50484d563f8bfa723c74c944b0bb15b9e054db9c889348c8c307abcbee75ab92"},
+    {file = "rapidfuzz-3.10.0-pp39-pypy39_pp73-macosx_10_15_x86_64.whl", hash = "sha256:5897242d455461f2c5b82d7397b29341fd11e85bf3608a522177071044784ee8"},
+    {file = "rapidfuzz-3.10.0-pp39-pypy39_pp73-macosx_11_0_arm64.whl", hash = "sha256:116c71a81e046ba56551d8ab68067ca7034d94b617545316d460a452c5c3c289"},
+    {file = "rapidfuzz-3.10.0-pp39-pypy39_pp73-manylinux_2_17_aarch64.manylinux2014_aarch64.whl", hash = "sha256:f0a547e4350d1fa32624d3eab51eff8cf329f4cae110b4ea0402486b1da8be40"},
+    {file = "rapidfuzz-3.10.0-pp39-pypy39_pp73-manylinux_2_17_i686.manylinux2014_i686.whl", hash = "sha256:399b9b79ccfcf50ca3bad7692bc098bb8eade88d7d5e15773b7f866c91156d0c"},
+    {file = "rapidfuzz-3.10.0-pp39-pypy39_pp73-manylinux_2_17_x86_64.manylinux2014_x86_64.whl", hash = "sha256:7947a425d1be3e744707ee58c6cb318b93a56e08f080722dcc0347e0b7a1bb9a"},
+    {file = "rapidfuzz-3.10.0-pp39-pypy39_pp73-win_amd64.whl", hash = "sha256:94c48b4a2a4b1d22246f48e2b11cae01ec7d23f0c9123f8bb822839ad79d0a88"},
+    {file = "rapidfuzz-3.10.0.tar.gz", hash = "sha256:6b62af27e65bb39276a66533655a2fa3c60a487b03935721c45b7809527979be"},
+]
+
+[package.extras]
+all = ["numpy"]
 
 [[package]]
 name = "readabilipy"
@@ -8293,13 +8331,13 @@ test = ["accelerate (>=0.24.1,<=0.27.0)", "apache-airflow (==2.9.3)", "apache-ai
 
 [[package]]
 name = "sagemaker-core"
-version = "1.0.4"
+version = "1.0.9"
 description = "An python package for sagemaker core functionalities"
 optional = false
 python-versions = ">=3.8"
 files = [
-    {file = "sagemaker_core-1.0.4-py3-none-any.whl", hash = "sha256:bf71d988dbda03a3cd1557524f2fab4f19d89e54bd38fc7f05bbbcf580715f95"},
-    {file = "sagemaker_core-1.0.4.tar.gz", hash = "sha256:203f4eb9d0d2a0e6ba80d79ba8c28b8ea27c94d04f6d9ff01c2fd55b95615c78"},
+    {file = "sagemaker_core-1.0.9-py3-none-any.whl", hash = "sha256:7a22c46cf93594f8d44e3523d4ba98407911f3530af68a8ffdde5082d3b26fa3"},
+    {file = "sagemaker_core-1.0.9.tar.gz", hash = "sha256:664115faf797412553fb81b97a4777e78e51dfd4454c32edb2c8371bf203c535"},
 ]
 
 [package.dependencies]
@@ -8473,18 +8511,18 @@ tornado = ["tornado (>=5)"]
 
 [[package]]
 name = "setuptools"
-version = "74.1.2"
+version = "75.1.0"
 description = "Easily download, build, install, upgrade, and uninstall Python packages"
 optional = false
 python-versions = ">=3.8"
 files = [
-    {file = "setuptools-74.1.2-py3-none-any.whl", hash = "sha256:5f4c08aa4d3ebcb57a50c33b1b07e94315d7fc7230f7115e47fc99776c8ce308"},
-    {file = "setuptools-74.1.2.tar.gz", hash = "sha256:95b40ed940a1c67eb70fc099094bd6e99c6ee7c23aa2306f4d2697ba7916f9c6"},
+    {file = "setuptools-75.1.0-py3-none-any.whl", hash = "sha256:35ab7fd3bcd95e6b7fd704e4a1539513edad446c097797f2985e0e4b960772f2"},
+    {file = "setuptools-75.1.0.tar.gz", hash = "sha256:d59a21b17a275fb872a9c3dae73963160ae079f1049ed956880cd7c09b120538"},
 ]
 
 [package.extras]
 check = ["pytest-checkdocs (>=2.4)", "pytest-ruff (>=0.2.1)", "ruff (>=0.5.2)"]
-core = ["importlib-metadata (>=6)", "importlib-resources (>=5.10.2)", "jaraco.text (>=3.7)", "more-itertools (>=8.8)", "packaging (>=24)", "platformdirs (>=2.6.2)", "tomli (>=2.0.1)", "wheel (>=0.43.0)"]
+core = ["importlib-metadata (>=6)", "importlib-resources (>=5.10.2)", "jaraco.collections", "jaraco.functools", "jaraco.text (>=3.7)", "more-itertools", "more-itertools (>=8.8)", "packaging", "packaging (>=24)", "platformdirs (>=2.6.2)", "tomli (>=2.0.1)", "wheel (>=0.43.0)"]
 cover = ["pytest-cov"]
 doc = ["furo", "jaraco.packaging (>=9.3)", "jaraco.tidelift (>=1.4)", "pygments-github-lexers (==0.0.5)", "pyproject-hooks (!=1.1)", "rst.linker (>=1.9)", "sphinx (>=3.5)", "sphinx-favicon", "sphinx-inline-tabs", "sphinx-lint", "sphinx-notfound-page (>=1,<2)", "sphinx-reredirects", "sphinxcontrib-towncrier", "towncrier (<24.7)"]
 enabler = ["pytest-enabler (>=2.2)"]
@@ -8644,60 +8682,60 @@ files = [
 
 [[package]]
 name = "sqlalchemy"
-version = "2.0.34"
+version = "2.0.35"
 description = "Database Abstraction Library"
 optional = false
 python-versions = ">=3.7"
 files = [
-    {file = "SQLAlchemy-2.0.34-cp310-cp310-macosx_10_9_x86_64.whl", hash = "sha256:95d0b2cf8791ab5fb9e3aa3d9a79a0d5d51f55b6357eecf532a120ba3b5524db"},
-    {file = "SQLAlchemy-2.0.34-cp310-cp310-macosx_11_0_arm64.whl", hash = "sha256:243f92596f4fd4c8bd30ab8e8dd5965afe226363d75cab2468f2c707f64cd83b"},
-    {file = "SQLAlchemy-2.0.34-cp310-cp310-manylinux_2_17_aarch64.manylinux2014_aarch64.whl", hash = "sha256:9ea54f7300553af0a2a7235e9b85f4204e1fc21848f917a3213b0e0818de9a24"},
-    {file = "SQLAlchemy-2.0.34-cp310-cp310-manylinux_2_17_x86_64.manylinux2014_x86_64.whl", hash = "sha256:173f5f122d2e1bff8fbd9f7811b7942bead1f5e9f371cdf9e670b327e6703ebd"},
-    {file = "SQLAlchemy-2.0.34-cp310-cp310-musllinux_1_2_aarch64.whl", hash = "sha256:196958cde924a00488e3e83ff917be3b73cd4ed8352bbc0f2989333176d1c54d"},
-    {file = "SQLAlchemy-2.0.34-cp310-cp310-musllinux_1_2_x86_64.whl", hash = "sha256:bd90c221ed4e60ac9d476db967f436cfcecbd4ef744537c0f2d5291439848768"},
-    {file = "SQLAlchemy-2.0.34-cp310-cp310-win32.whl", hash = "sha256:3166dfff2d16fe9be3241ee60ece6fcb01cf8e74dd7c5e0b64f8e19fab44911b"},
-    {file = "SQLAlchemy-2.0.34-cp310-cp310-win_amd64.whl", hash = "sha256:6831a78bbd3c40f909b3e5233f87341f12d0b34a58f14115c9e94b4cdaf726d3"},
-    {file = "SQLAlchemy-2.0.34-cp311-cp311-macosx_10_9_x86_64.whl", hash = "sha256:c7db3db284a0edaebe87f8f6642c2b2c27ed85c3e70064b84d1c9e4ec06d5d84"},
-    {file = "SQLAlchemy-2.0.34-cp311-cp311-macosx_11_0_arm64.whl", hash = "sha256:430093fce0efc7941d911d34f75a70084f12f6ca5c15d19595c18753edb7c33b"},
-    {file = "SQLAlchemy-2.0.34-cp311-cp311-manylinux_2_17_aarch64.manylinux2014_aarch64.whl", hash = "sha256:79cb400c360c7c210097b147c16a9e4c14688a6402445ac848f296ade6283bbc"},
-    {file = "SQLAlchemy-2.0.34-cp311-cp311-manylinux_2_17_x86_64.manylinux2014_x86_64.whl", hash = "sha256:fb1b30f31a36c7f3fee848391ff77eebdd3af5750bf95fbf9b8b5323edfdb4ec"},
-    {file = "SQLAlchemy-2.0.34-cp311-cp311-musllinux_1_2_aarch64.whl", hash = "sha256:8fddde2368e777ea2a4891a3fb4341e910a056be0bb15303bf1b92f073b80c02"},
-    {file = "SQLAlchemy-2.0.34-cp311-cp311-musllinux_1_2_x86_64.whl", hash = "sha256:80bd73ea335203b125cf1d8e50fef06be709619eb6ab9e7b891ea34b5baa2287"},
-    {file = "SQLAlchemy-2.0.34-cp311-cp311-win32.whl", hash = "sha256:6daeb8382d0df526372abd9cb795c992e18eed25ef2c43afe518c73f8cccb721"},
-    {file = "SQLAlchemy-2.0.34-cp311-cp311-win_amd64.whl", hash = "sha256:5bc08e75ed11693ecb648b7a0a4ed80da6d10845e44be0c98c03f2f880b68ff4"},
-    {file = "SQLAlchemy-2.0.34-cp312-cp312-macosx_10_9_x86_64.whl", hash = "sha256:53e68b091492c8ed2bd0141e00ad3089bcc6bf0e6ec4142ad6505b4afe64163e"},
-    {file = "SQLAlchemy-2.0.34-cp312-cp312-macosx_11_0_arm64.whl", hash = "sha256:bcd18441a49499bf5528deaa9dee1f5c01ca491fc2791b13604e8f972877f812"},
-    {file = "SQLAlchemy-2.0.34-cp312-cp312-manylinux_2_17_aarch64.manylinux2014_aarch64.whl", hash = "sha256:165bbe0b376541092bf49542bd9827b048357f4623486096fc9aaa6d4e7c59a2"},
-    {file = "SQLAlchemy-2.0.34-cp312-cp312-manylinux_2_17_x86_64.manylinux2014_x86_64.whl", hash = "sha256:c3330415cd387d2b88600e8e26b510d0370db9b7eaf984354a43e19c40df2e2b"},
-    {file = "SQLAlchemy-2.0.34-cp312-cp312-musllinux_1_2_aarch64.whl", hash = "sha256:97b850f73f8abbffb66ccbab6e55a195a0eb655e5dc74624d15cff4bfb35bd74"},
-    {file = "SQLAlchemy-2.0.34-cp312-cp312-musllinux_1_2_x86_64.whl", hash = "sha256:7cee4c6917857fd6121ed84f56d1dc78eb1d0e87f845ab5a568aba73e78adf83"},
-    {file = "SQLAlchemy-2.0.34-cp312-cp312-win32.whl", hash = "sha256:fbb034f565ecbe6c530dff948239377ba859420d146d5f62f0271407ffb8c580"},
-    {file = "SQLAlchemy-2.0.34-cp312-cp312-win_amd64.whl", hash = "sha256:707c8f44931a4facd4149b52b75b80544a8d824162602b8cd2fe788207307f9a"},
-    {file = "SQLAlchemy-2.0.34-cp37-cp37m-macosx_10_9_x86_64.whl", hash = "sha256:24af3dc43568f3780b7e1e57c49b41d98b2d940c1fd2e62d65d3928b6f95f021"},
-    {file = "SQLAlchemy-2.0.34-cp37-cp37m-manylinux_2_17_aarch64.manylinux2014_aarch64.whl", hash = "sha256:e60ed6ef0a35c6b76b7640fe452d0e47acc832ccbb8475de549a5cc5f90c2c06"},
-    {file = "SQLAlchemy-2.0.34-cp37-cp37m-manylinux_2_17_x86_64.manylinux2014_x86_64.whl", hash = "sha256:413c85cd0177c23e32dee6898c67a5f49296640041d98fddb2c40888fe4daa2e"},
-    {file = "SQLAlchemy-2.0.34-cp37-cp37m-musllinux_1_2_aarch64.whl", hash = "sha256:25691f4adfb9d5e796fd48bf1432272f95f4bbe5f89c475a788f31232ea6afba"},
-    {file = "SQLAlchemy-2.0.34-cp37-cp37m-musllinux_1_2_x86_64.whl", hash = "sha256:526ce723265643dbc4c7efb54f56648cc30e7abe20f387d763364b3ce7506c82"},
-    {file = "SQLAlchemy-2.0.34-cp37-cp37m-win32.whl", hash = "sha256:13be2cc683b76977a700948411a94c67ad8faf542fa7da2a4b167f2244781cf3"},
-    {file = "SQLAlchemy-2.0.34-cp37-cp37m-win_amd64.whl", hash = "sha256:e54ef33ea80d464c3dcfe881eb00ad5921b60f8115ea1a30d781653edc2fd6a2"},
-    {file = "SQLAlchemy-2.0.34-cp38-cp38-macosx_10_9_x86_64.whl", hash = "sha256:43f28005141165edd11fbbf1541c920bd29e167b8bbc1fb410d4fe2269c1667a"},
-    {file = "SQLAlchemy-2.0.34-cp38-cp38-macosx_11_0_arm64.whl", hash = "sha256:b68094b165a9e930aedef90725a8fcfafe9ef95370cbb54abc0464062dbf808f"},
-    {file = "SQLAlchemy-2.0.34-cp38-cp38-manylinux_2_17_aarch64.manylinux2014_aarch64.whl", hash = "sha256:6a1e03db964e9d32f112bae36f0cc1dcd1988d096cfd75d6a588a3c3def9ab2b"},
-    {file = "SQLAlchemy-2.0.34-cp38-cp38-manylinux_2_17_x86_64.manylinux2014_x86_64.whl", hash = "sha256:203d46bddeaa7982f9c3cc693e5bc93db476ab5de9d4b4640d5c99ff219bee8c"},
-    {file = "SQLAlchemy-2.0.34-cp38-cp38-musllinux_1_2_aarch64.whl", hash = "sha256:ae92bebca3b1e6bd203494e5ef919a60fb6dfe4d9a47ed2453211d3bd451b9f5"},
-    {file = "SQLAlchemy-2.0.34-cp38-cp38-musllinux_1_2_x86_64.whl", hash = "sha256:9661268415f450c95f72f0ac1217cc6f10256f860eed85c2ae32e75b60278ad8"},
-    {file = "SQLAlchemy-2.0.34-cp38-cp38-win32.whl", hash = "sha256:895184dfef8708e15f7516bd930bda7e50ead069280d2ce09ba11781b630a434"},
-    {file = "SQLAlchemy-2.0.34-cp38-cp38-win_amd64.whl", hash = "sha256:6e7cde3a2221aa89247944cafb1b26616380e30c63e37ed19ff0bba5e968688d"},
-    {file = "SQLAlchemy-2.0.34-cp39-cp39-macosx_10_9_x86_64.whl", hash = "sha256:dbcdf987f3aceef9763b6d7b1fd3e4ee210ddd26cac421d78b3c206d07b2700b"},
-    {file = "SQLAlchemy-2.0.34-cp39-cp39-macosx_11_0_arm64.whl", hash = "sha256:ce119fc4ce0d64124d37f66a6f2a584fddc3c5001755f8a49f1ca0a177ef9796"},
-    {file = "SQLAlchemy-2.0.34-cp39-cp39-manylinux_2_17_aarch64.manylinux2014_aarch64.whl", hash = "sha256:a17d8fac6df9835d8e2b4c5523666e7051d0897a93756518a1fe101c7f47f2f0"},
-    {file = "SQLAlchemy-2.0.34-cp39-cp39-manylinux_2_17_x86_64.manylinux2014_x86_64.whl", hash = "sha256:9ebc11c54c6ecdd07bb4efbfa1554538982f5432dfb8456958b6d46b9f834bb7"},
-    {file = "SQLAlchemy-2.0.34-cp39-cp39-musllinux_1_2_aarch64.whl", hash = "sha256:2e6965346fc1491a566e019a4a1d3dfc081ce7ac1a736536367ca305da6472a8"},
-    {file = "SQLAlchemy-2.0.34-cp39-cp39-musllinux_1_2_x86_64.whl", hash = "sha256:220574e78ad986aea8e81ac68821e47ea9202b7e44f251b7ed8c66d9ae3f4278"},
-    {file = "SQLAlchemy-2.0.34-cp39-cp39-win32.whl", hash = "sha256:b75b00083e7fe6621ce13cfce9d4469c4774e55e8e9d38c305b37f13cf1e874c"},
-    {file = "SQLAlchemy-2.0.34-cp39-cp39-win_amd64.whl", hash = "sha256:c29d03e0adf3cc1a8c3ec62d176824972ae29b67a66cbb18daff3062acc6faa8"},
-    {file = "SQLAlchemy-2.0.34-py3-none-any.whl", hash = "sha256:7286c353ee6475613d8beff83167374006c6b3e3f0e6491bfe8ca610eb1dec0f"},
-    {file = "sqlalchemy-2.0.34.tar.gz", hash = "sha256:10d8f36990dd929690666679b0f42235c159a7051534adb135728ee52828dd22"},
+    {file = "SQLAlchemy-2.0.35-cp310-cp310-macosx_10_9_x86_64.whl", hash = "sha256:67219632be22f14750f0d1c70e62f204ba69d28f62fd6432ba05ab295853de9b"},
+    {file = "SQLAlchemy-2.0.35-cp310-cp310-macosx_11_0_arm64.whl", hash = "sha256:4668bd8faf7e5b71c0319407b608f278f279668f358857dbfd10ef1954ac9f90"},
+    {file = "SQLAlchemy-2.0.35-cp310-cp310-manylinux_2_17_aarch64.manylinux2014_aarch64.whl", hash = "sha256:cb8bea573863762bbf45d1e13f87c2d2fd32cee2dbd50d050f83f87429c9e1ea"},
+    {file = "SQLAlchemy-2.0.35-cp310-cp310-manylinux_2_17_x86_64.manylinux2014_x86_64.whl", hash = "sha256:f552023710d4b93d8fb29a91fadf97de89c5926c6bd758897875435f2a939f33"},
+    {file = "SQLAlchemy-2.0.35-cp310-cp310-musllinux_1_2_aarch64.whl", hash = "sha256:016b2e665f778f13d3c438651dd4de244214b527a275e0acf1d44c05bc6026a9"},
+    {file = "SQLAlchemy-2.0.35-cp310-cp310-musllinux_1_2_x86_64.whl", hash = "sha256:7befc148de64b6060937231cbff8d01ccf0bfd75aa26383ffdf8d82b12ec04ff"},
+    {file = "SQLAlchemy-2.0.35-cp310-cp310-win32.whl", hash = "sha256:22b83aed390e3099584b839b93f80a0f4a95ee7f48270c97c90acd40ee646f0b"},
+    {file = "SQLAlchemy-2.0.35-cp310-cp310-win_amd64.whl", hash = "sha256:a29762cd3d116585278ffb2e5b8cc311fb095ea278b96feef28d0b423154858e"},
+    {file = "SQLAlchemy-2.0.35-cp311-cp311-macosx_10_9_x86_64.whl", hash = "sha256:e21f66748ab725ade40fa7af8ec8b5019c68ab00b929f6643e1b1af461eddb60"},
+    {file = "SQLAlchemy-2.0.35-cp311-cp311-macosx_11_0_arm64.whl", hash = "sha256:8a6219108a15fc6d24de499d0d515c7235c617b2540d97116b663dade1a54d62"},
+    {file = "SQLAlchemy-2.0.35-cp311-cp311-manylinux_2_17_aarch64.manylinux2014_aarch64.whl", hash = "sha256:042622a5306c23b972192283f4e22372da3b8ddf5f7aac1cc5d9c9b222ab3ff6"},
+    {file = "SQLAlchemy-2.0.35-cp311-cp311-manylinux_2_17_x86_64.manylinux2014_x86_64.whl", hash = "sha256:627dee0c280eea91aed87b20a1f849e9ae2fe719d52cbf847c0e0ea34464b3f7"},
+    {file = "SQLAlchemy-2.0.35-cp311-cp311-musllinux_1_2_aarch64.whl", hash = "sha256:4fdcd72a789c1c31ed242fd8c1bcd9ea186a98ee8e5408a50e610edfef980d71"},
+    {file = "SQLAlchemy-2.0.35-cp311-cp311-musllinux_1_2_x86_64.whl", hash = "sha256:89b64cd8898a3a6f642db4eb7b26d1b28a497d4022eccd7717ca066823e9fb01"},
+    {file = "SQLAlchemy-2.0.35-cp311-cp311-win32.whl", hash = "sha256:6a93c5a0dfe8d34951e8a6f499a9479ffb9258123551fa007fc708ae2ac2bc5e"},
+    {file = "SQLAlchemy-2.0.35-cp311-cp311-win_amd64.whl", hash = "sha256:c68fe3fcde03920c46697585620135b4ecfdfc1ed23e75cc2c2ae9f8502c10b8"},
+    {file = "SQLAlchemy-2.0.35-cp312-cp312-macosx_10_9_x86_64.whl", hash = "sha256:eb60b026d8ad0c97917cb81d3662d0b39b8ff1335e3fabb24984c6acd0c900a2"},
+    {file = "SQLAlchemy-2.0.35-cp312-cp312-macosx_11_0_arm64.whl", hash = "sha256:6921ee01caf375363be5e9ae70d08ce7ca9d7e0e8983183080211a062d299468"},
+    {file = "SQLAlchemy-2.0.35-cp312-cp312-manylinux_2_17_aarch64.manylinux2014_aarch64.whl", hash = "sha256:8cdf1a0dbe5ced887a9b127da4ffd7354e9c1a3b9bb330dce84df6b70ccb3a8d"},
+    {file = "SQLAlchemy-2.0.35-cp312-cp312-manylinux_2_17_x86_64.manylinux2014_x86_64.whl", hash = "sha256:93a71c8601e823236ac0e5d087e4f397874a421017b3318fd92c0b14acf2b6db"},
+    {file = "SQLAlchemy-2.0.35-cp312-cp312-musllinux_1_2_aarch64.whl", hash = "sha256:e04b622bb8a88f10e439084486f2f6349bf4d50605ac3e445869c7ea5cf0fa8c"},
+    {file = "SQLAlchemy-2.0.35-cp312-cp312-musllinux_1_2_x86_64.whl", hash = "sha256:1b56961e2d31389aaadf4906d453859f35302b4eb818d34a26fab72596076bb8"},
+    {file = "SQLAlchemy-2.0.35-cp312-cp312-win32.whl", hash = "sha256:0f9f3f9a3763b9c4deb8c5d09c4cc52ffe49f9876af41cc1b2ad0138878453cf"},
+    {file = "SQLAlchemy-2.0.35-cp312-cp312-win_amd64.whl", hash = "sha256:25b0f63e7fcc2a6290cb5f7f5b4fc4047843504983a28856ce9b35d8f7de03cc"},
+    {file = "SQLAlchemy-2.0.35-cp37-cp37m-macosx_10_9_x86_64.whl", hash = "sha256:f021d334f2ca692523aaf7bbf7592ceff70c8594fad853416a81d66b35e3abf9"},
+    {file = "SQLAlchemy-2.0.35-cp37-cp37m-manylinux_2_17_aarch64.manylinux2014_aarch64.whl", hash = "sha256:05c3f58cf91683102f2f0265c0db3bd3892e9eedabe059720492dbaa4f922da1"},
+    {file = "SQLAlchemy-2.0.35-cp37-cp37m-manylinux_2_17_x86_64.manylinux2014_x86_64.whl", hash = "sha256:032d979ce77a6c2432653322ba4cbeabf5a6837f704d16fa38b5a05d8e21fa00"},
+    {file = "SQLAlchemy-2.0.35-cp37-cp37m-musllinux_1_2_aarch64.whl", hash = "sha256:2e795c2f7d7249b75bb5f479b432a51b59041580d20599d4e112b5f2046437a3"},
+    {file = "SQLAlchemy-2.0.35-cp37-cp37m-musllinux_1_2_x86_64.whl", hash = "sha256:cc32b2990fc34380ec2f6195f33a76b6cdaa9eecf09f0c9404b74fc120aef36f"},
+    {file = "SQLAlchemy-2.0.35-cp37-cp37m-win32.whl", hash = "sha256:9509c4123491d0e63fb5e16199e09f8e262066e58903e84615c301dde8fa2e87"},
+    {file = "SQLAlchemy-2.0.35-cp37-cp37m-win_amd64.whl", hash = "sha256:3655af10ebcc0f1e4e06c5900bb33e080d6a1fa4228f502121f28a3b1753cde5"},
+    {file = "SQLAlchemy-2.0.35-cp38-cp38-macosx_10_9_x86_64.whl", hash = "sha256:4c31943b61ed8fdd63dfd12ccc919f2bf95eefca133767db6fbbd15da62078ec"},
+    {file = "SQLAlchemy-2.0.35-cp38-cp38-macosx_11_0_arm64.whl", hash = "sha256:a62dd5d7cc8626a3634208df458c5fe4f21200d96a74d122c83bc2015b333bc1"},
+    {file = "SQLAlchemy-2.0.35-cp38-cp38-manylinux_2_17_aarch64.manylinux2014_aarch64.whl", hash = "sha256:0630774b0977804fba4b6bbea6852ab56c14965a2b0c7fc7282c5f7d90a1ae72"},
+    {file = "SQLAlchemy-2.0.35-cp38-cp38-manylinux_2_17_x86_64.manylinux2014_x86_64.whl", hash = "sha256:8d625eddf7efeba2abfd9c014a22c0f6b3796e0ffb48f5d5ab106568ef01ff5a"},
+    {file = "SQLAlchemy-2.0.35-cp38-cp38-musllinux_1_2_aarch64.whl", hash = "sha256:ada603db10bb865bbe591939de854faf2c60f43c9b763e90f653224138f910d9"},
+    {file = "SQLAlchemy-2.0.35-cp38-cp38-musllinux_1_2_x86_64.whl", hash = "sha256:c41411e192f8d3ea39ea70e0fae48762cd11a2244e03751a98bd3c0ca9a4e936"},
+    {file = "SQLAlchemy-2.0.35-cp38-cp38-win32.whl", hash = "sha256:d299797d75cd747e7797b1b41817111406b8b10a4f88b6e8fe5b5e59598b43b0"},
+    {file = "SQLAlchemy-2.0.35-cp38-cp38-win_amd64.whl", hash = "sha256:0375a141e1c0878103eb3d719eb6d5aa444b490c96f3fedab8471c7f6ffe70ee"},
+    {file = "SQLAlchemy-2.0.35-cp39-cp39-macosx_10_9_x86_64.whl", hash = "sha256:ccae5de2a0140d8be6838c331604f91d6fafd0735dbdcee1ac78fc8fbaba76b4"},
+    {file = "SQLAlchemy-2.0.35-cp39-cp39-macosx_11_0_arm64.whl", hash = "sha256:2a275a806f73e849e1c309ac11108ea1a14cd7058577aba962cd7190e27c9e3c"},
+    {file = "SQLAlchemy-2.0.35-cp39-cp39-manylinux_2_17_aarch64.manylinux2014_aarch64.whl", hash = "sha256:732e026240cdd1c1b2e3ac515c7a23820430ed94292ce33806a95869c46bd139"},
+    {file = "SQLAlchemy-2.0.35-cp39-cp39-manylinux_2_17_x86_64.manylinux2014_x86_64.whl", hash = "sha256:890da8cd1941fa3dab28c5bac3b9da8502e7e366f895b3b8e500896f12f94d11"},
+    {file = "SQLAlchemy-2.0.35-cp39-cp39-musllinux_1_2_aarch64.whl", hash = "sha256:c0d8326269dbf944b9201911b0d9f3dc524d64779a07518199a58384c3d37a44"},
+    {file = "SQLAlchemy-2.0.35-cp39-cp39-musllinux_1_2_x86_64.whl", hash = "sha256:b76d63495b0508ab9fc23f8152bac63205d2a704cd009a2b0722f4c8e0cba8e0"},
+    {file = "SQLAlchemy-2.0.35-cp39-cp39-win32.whl", hash = "sha256:69683e02e8a9de37f17985905a5eca18ad651bf592314b4d3d799029797d0eb3"},
+    {file = "SQLAlchemy-2.0.35-cp39-cp39-win_amd64.whl", hash = "sha256:aee110e4ef3c528f3abbc3c2018c121e708938adeeff9006428dd7c8555e9b3f"},
+    {file = "SQLAlchemy-2.0.35-py3-none-any.whl", hash = "sha256:2ab3f0336c0387662ce6221ad30ab3a5e6499aab01b9790879b6578fd9b8faa1"},
+    {file = "sqlalchemy-2.0.35.tar.gz", hash = "sha256:e11d7ea4d24f0a262bccf9a7cd6284c976c5369dac21db237cff59586045ab9f"},
 ]
 
 [package.dependencies]
@@ -8746,13 +8784,13 @@ doc = ["sphinx"]
 
 [[package]]
 name = "starlette"
-version = "0.38.5"
+version = "0.38.6"
 description = "The little ASGI library that shines."
 optional = false
 python-versions = ">=3.8"
 files = [
-    {file = "starlette-0.38.5-py3-none-any.whl", hash = "sha256:632f420a9d13e3ee2a6f18f437b0a9f1faecb0bc42e1942aa2ea0e379a4c4206"},
-    {file = "starlette-0.38.5.tar.gz", hash = "sha256:04a92830a9b6eb1442c766199d62260c3d4dc9c4f9188360626b1e0273cb7077"},
+    {file = "starlette-0.38.6-py3-none-any.whl", hash = "sha256:4517a1409e2e73ee4951214ba012052b9e16f60e90d73cfb06192c19203bbb05"},
+    {file = "starlette-0.38.6.tar.gz", hash = "sha256:863a1588f5574e70a821dadefb41e4881ea451a47a3cd1b4df359d4ffefe5ead"},
 ]
 
 [package.dependencies]
@@ -8777,13 +8815,13 @@ python-dateutil = ">=2.6.0"
 
 [[package]]
 name = "sympy"
-version = "1.13.2"
+version = "1.13.3"
 description = "Computer algebra system (CAS) in Python"
 optional = false
 python-versions = ">=3.8"
 files = [
-    {file = "sympy-1.13.2-py3-none-any.whl", hash = "sha256:c51d75517712f1aed280d4ce58506a4a88d635d6b5dd48b39102a7ae1f3fcfe9"},
-    {file = "sympy-1.13.2.tar.gz", hash = "sha256:401449d84d07be9d0c7a46a64bd54fe097667d5e7181bfe67ec777be9e01cb13"},
+    {file = "sympy-1.13.3-py3-none-any.whl", hash = "sha256:54612cf55a62755ee71824ce692986f23c88ffa77207b30c1368eda4a7060f73"},
+    {file = "sympy-1.13.3.tar.gz", hash = "sha256:b27fd2c6530e0ab39e275fc9b683895367e51d5da91baa8d3d64db2565fec4d9"},
 ]
 
 [package.dependencies]
@@ -8849,13 +8887,13 @@ test = ["pytest", "tornado (>=4.5)", "typeguard"]
 
 [[package]]
 name = "tencentcloud-sdk-python-common"
-version = "3.0.1231"
+version = "3.0.1243"
 description = "Tencent Cloud Common SDK for Python"
 optional = false
 python-versions = "*"
 files = [
-    {file = "tencentcloud-sdk-python-common-3.0.1231.tar.gz", hash = "sha256:22aa281ca2eac511e1615b2953da7c4a0bec87cf93a05a7a15dbb61b23a215ee"},
-    {file = "tencentcloud_sdk_python_common-3.0.1231-py2.py3-none-any.whl", hash = "sha256:bd0f7c4df4b156ec35c8731afa1f498043c7e1cd5d2feb595ee441fdb45a061e"},
+    {file = "tencentcloud-sdk-python-common-3.0.1243.tar.gz", hash = "sha256:4008126b5506543ddc33ea53ef25abe19b4fa025006ec7751b3d5b3c046438b6"},
+    {file = "tencentcloud_sdk_python_common-3.0.1243-py2.py3-none-any.whl", hash = "sha256:ed5d2040627aeed2df1fe5f9d65b9987ef842222d79fb13136cc5866aa5ac50c"},
 ]
 
 [package.dependencies]
@@ -8863,17 +8901,17 @@ requests = ">=2.16.0"
 
 [[package]]
 name = "tencentcloud-sdk-python-hunyuan"
-version = "3.0.1231"
+version = "3.0.1243"
 description = "Tencent Cloud Hunyuan SDK for Python"
 optional = false
 python-versions = "*"
 files = [
-    {file = "tencentcloud-sdk-python-hunyuan-3.0.1231.tar.gz", hash = "sha256:6da12f418f14305b3a6b7bb29b6d95bf4038a6b66b81c0e03b8dafc4f6df99ca"},
-    {file = "tencentcloud_sdk_python_hunyuan-3.0.1231-py2.py3-none-any.whl", hash = "sha256:21ba28f69c34c15e20900be3f2c06376fcaf7e58265f939833c55631f2348792"},
+    {file = "tencentcloud-sdk-python-hunyuan-3.0.1243.tar.gz", hash = "sha256:2caad419f24106574f05faf69d162060689e03452832fda890a28eb8814199bc"},
+    {file = "tencentcloud_sdk_python_hunyuan-3.0.1243-py2.py3-none-any.whl", hash = "sha256:b5a5c0dfa8b6de68c2c8747bf34b880c7fd2a1cbd59efc372e076260228ce231"},
 ]
 
 [package.dependencies]
-tencentcloud-sdk-python-common = "3.0.1231"
+tencentcloud-sdk-python-common = "3.0.1243"
 
 [[package]]
 name = "threadpoolctl"
@@ -9316,13 +9354,13 @@ typing-extensions = ">=3.7.4"
 
 [[package]]
 name = "tzdata"
-version = "2024.1"
+version = "2024.2"
 description = "Provider of IANA time zone data"
 optional = false
 python-versions = ">=2"
 files = [
-    {file = "tzdata-2024.1-py2.py3-none-any.whl", hash = "sha256:9068bc196136463f5245e51efda838afa15aaeca9903f49050dfa2679db4d252"},
-    {file = "tzdata-2024.1.tar.gz", hash = "sha256:2674120f8d891909751c38abcdfd386ac0a5a1127954fbc332af6b5ceae07efd"},
+    {file = "tzdata-2024.2-py2.py3-none-any.whl", hash = "sha256:a48093786cdcde33cad18c2555e8532f34422074448fbc874186f0abd79565cd"},
+    {file = "tzdata-2024.2.tar.gz", hash = "sha256:7d85cc416e9382e69095b7bdf4afd9e3880418a2413feec7069d533d6b4e31cc"},
 ]
 
 [[package]]
@@ -9542,13 +9580,13 @@ zstd = ["zstandard (>=0.18.0)"]
 
 [[package]]
 name = "uvicorn"
-version = "0.30.6"
+version = "0.31.0"
 description = "The lightning-fast ASGI server."
 optional = false
 python-versions = ">=3.8"
 files = [
-    {file = "uvicorn-0.30.6-py3-none-any.whl", hash = "sha256:65fd46fe3fda5bdc1b03b94eb634923ff18cd35b2f084813ea79d1f103f711b5"},
-    {file = "uvicorn-0.30.6.tar.gz", hash = "sha256:4b15decdda1e72be08209e860a1e10e92439ad5b97cf44cc945fcbee66fc5788"},
+    {file = "uvicorn-0.31.0-py3-none-any.whl", hash = "sha256:cac7be4dd4d891c363cd942160a7b02e69150dcbc7a36be04d5f4af4b17c8ced"},
+    {file = "uvicorn-0.31.0.tar.gz", hash = "sha256:13bc21373d103859f68fe739608e2eb054a816dea79189bc3ca08ea89a275906"},
 ]
 
 [package.dependencies]
@@ -9685,12 +9723,12 @@ files = [
 
 [[package]]
 name = "volcengine-python-sdk"
-version = "1.0.101"
+version = "1.0.103"
 description = "Volcengine SDK for Python"
 optional = false
 python-versions = "*"
 files = [
-    {file = "volcengine-python-sdk-1.0.101.tar.gz", hash = "sha256:1b76e71a6dcf3d5be1b2c058e7d281359e6cca2cc920ffe2567d3115beea1d02"},
+    {file = "volcengine-python-sdk-1.0.103.tar.gz", hash = "sha256:49fa8572802724972e1cb47a7e692b184b055f41b09099358c1a0fad1d146af5"},
 ]
 
 [package.dependencies]
@@ -9860,97 +9898,97 @@ test = ["websockets"]
 
 [[package]]
 name = "websockets"
-version = "13.0.1"
+version = "13.1"
 description = "An implementation of the WebSocket Protocol (RFC 6455 & 7692)"
 optional = false
 python-versions = ">=3.8"
 files = [
-    {file = "websockets-13.0.1-cp310-cp310-macosx_10_9_universal2.whl", hash = "sha256:1841c9082a3ba4a05ea824cf6d99570a6a2d8849ef0db16e9c826acb28089e8f"},
-    {file = "websockets-13.0.1-cp310-cp310-macosx_10_9_x86_64.whl", hash = "sha256:c5870b4a11b77e4caa3937142b650fbbc0914a3e07a0cf3131f35c0587489c1c"},
-    {file = "websockets-13.0.1-cp310-cp310-macosx_11_0_arm64.whl", hash = "sha256:f1d3d1f2eb79fe7b0fb02e599b2bf76a7619c79300fc55f0b5e2d382881d4f7f"},
-    {file = "websockets-13.0.1-cp310-cp310-manylinux_2_17_aarch64.manylinux2014_aarch64.whl", hash = "sha256:15c7d62ee071fa94a2fc52c2b472fed4af258d43f9030479d9c4a2de885fd543"},
-    {file = "websockets-13.0.1-cp310-cp310-manylinux_2_5_i686.manylinux1_i686.manylinux_2_17_i686.manylinux2014_i686.whl", hash = "sha256:6724b554b70d6195ba19650fef5759ef11346f946c07dbbe390e039bcaa7cc3d"},
-    {file = "websockets-13.0.1-cp310-cp310-manylinux_2_5_x86_64.manylinux1_x86_64.manylinux_2_17_x86_64.manylinux2014_x86_64.whl", hash = "sha256:56a952fa2ae57a42ba7951e6b2605e08a24801a4931b5644dfc68939e041bc7f"},
-    {file = "websockets-13.0.1-cp310-cp310-musllinux_1_2_aarch64.whl", hash = "sha256:17118647c0ea14796364299e942c330d72acc4b248e07e639d34b75067b3cdd8"},
-    {file = "websockets-13.0.1-cp310-cp310-musllinux_1_2_i686.whl", hash = "sha256:64a11aae1de4c178fa653b07d90f2fb1a2ed31919a5ea2361a38760192e1858b"},
-    {file = "websockets-13.0.1-cp310-cp310-musllinux_1_2_x86_64.whl", hash = "sha256:0617fd0b1d14309c7eab6ba5deae8a7179959861846cbc5cb528a7531c249448"},
-    {file = "websockets-13.0.1-cp310-cp310-win32.whl", hash = "sha256:11f9976ecbc530248cf162e359a92f37b7b282de88d1d194f2167b5e7ad80ce3"},
-    {file = "websockets-13.0.1-cp310-cp310-win_amd64.whl", hash = "sha256:c3c493d0e5141ec055a7d6809a28ac2b88d5b878bb22df8c621ebe79a61123d0"},
-    {file = "websockets-13.0.1-cp311-cp311-macosx_10_9_universal2.whl", hash = "sha256:699ba9dd6a926f82a277063603fc8d586b89f4cb128efc353b749b641fcddda7"},
-    {file = "websockets-13.0.1-cp311-cp311-macosx_10_9_x86_64.whl", hash = "sha256:cf2fae6d85e5dc384bf846f8243ddaa9197f3a1a70044f59399af001fd1f51d4"},
-    {file = "websockets-13.0.1-cp311-cp311-macosx_11_0_arm64.whl", hash = "sha256:52aed6ef21a0f1a2a5e310fb5c42d7555e9c5855476bbd7173c3aa3d8a0302f2"},
-    {file = "websockets-13.0.1-cp311-cp311-manylinux_2_17_aarch64.manylinux2014_aarch64.whl", hash = "sha256:8eb2b9a318542153674c6e377eb8cb9ca0fc011c04475110d3477862f15d29f0"},
-    {file = "websockets-13.0.1-cp311-cp311-manylinux_2_5_i686.manylinux1_i686.manylinux_2_17_i686.manylinux2014_i686.whl", hash = "sha256:5df891c86fe68b2c38da55b7aea7095beca105933c697d719f3f45f4220a5e0e"},
-    {file = "websockets-13.0.1-cp311-cp311-manylinux_2_5_x86_64.manylinux1_x86_64.manylinux_2_17_x86_64.manylinux2014_x86_64.whl", hash = "sha256:fac2d146ff30d9dd2fcf917e5d147db037a5c573f0446c564f16f1f94cf87462"},
-    {file = "websockets-13.0.1-cp311-cp311-musllinux_1_2_aarch64.whl", hash = "sha256:b8ac5b46fd798bbbf2ac6620e0437c36a202b08e1f827832c4bf050da081b501"},
-    {file = "websockets-13.0.1-cp311-cp311-musllinux_1_2_i686.whl", hash = "sha256:46af561eba6f9b0848b2c9d2427086cabadf14e0abdd9fde9d72d447df268418"},
-    {file = "websockets-13.0.1-cp311-cp311-musllinux_1_2_x86_64.whl", hash = "sha256:b5a06d7f60bc2fc378a333978470dfc4e1415ee52f5f0fce4f7853eb10c1e9df"},
-    {file = "websockets-13.0.1-cp311-cp311-win32.whl", hash = "sha256:556e70e4f69be1082e6ef26dcb70efcd08d1850f5d6c5f4f2bcb4e397e68f01f"},
-    {file = "websockets-13.0.1-cp311-cp311-win_amd64.whl", hash = "sha256:67494e95d6565bf395476e9d040037ff69c8b3fa356a886b21d8422ad86ae075"},
-    {file = "websockets-13.0.1-cp312-cp312-macosx_10_9_universal2.whl", hash = "sha256:f9c9e258e3d5efe199ec23903f5da0eeaad58cf6fccb3547b74fd4750e5ac47a"},
-    {file = "websockets-13.0.1-cp312-cp312-macosx_10_9_x86_64.whl", hash = "sha256:6b41a1b3b561f1cba8321fb32987552a024a8f67f0d05f06fcf29f0090a1b956"},
-    {file = "websockets-13.0.1-cp312-cp312-macosx_11_0_arm64.whl", hash = "sha256:f73e676a46b0fe9426612ce8caeca54c9073191a77c3e9d5c94697aef99296af"},
-    {file = "websockets-13.0.1-cp312-cp312-manylinux_2_17_aarch64.manylinux2014_aarch64.whl", hash = "sha256:1f613289f4a94142f914aafad6c6c87903de78eae1e140fa769a7385fb232fdf"},
-    {file = "websockets-13.0.1-cp312-cp312-manylinux_2_5_i686.manylinux1_i686.manylinux_2_17_i686.manylinux2014_i686.whl", hash = "sha256:0f52504023b1480d458adf496dc1c9e9811df4ba4752f0bc1f89ae92f4f07d0c"},
-    {file = "websockets-13.0.1-cp312-cp312-manylinux_2_5_x86_64.manylinux1_x86_64.manylinux_2_17_x86_64.manylinux2014_x86_64.whl", hash = "sha256:139add0f98206cb74109faf3611b7783ceafc928529c62b389917a037d4cfdf4"},
-    {file = "websockets-13.0.1-cp312-cp312-musllinux_1_2_aarch64.whl", hash = "sha256:47236c13be337ef36546004ce8c5580f4b1150d9538b27bf8a5ad8edf23ccfab"},
-    {file = "websockets-13.0.1-cp312-cp312-musllinux_1_2_i686.whl", hash = "sha256:c44ca9ade59b2e376612df34e837013e2b273e6c92d7ed6636d0556b6f4db93d"},
-    {file = "websockets-13.0.1-cp312-cp312-musllinux_1_2_x86_64.whl", hash = "sha256:9bbc525f4be3e51b89b2a700f5746c2a6907d2e2ef4513a8daafc98198b92237"},
-    {file = "websockets-13.0.1-cp312-cp312-win32.whl", hash = "sha256:3624fd8664f2577cf8de996db3250662e259bfbc870dd8ebdcf5d7c6ac0b5185"},
-    {file = "websockets-13.0.1-cp312-cp312-win_amd64.whl", hash = "sha256:0513c727fb8adffa6d9bf4a4463b2bade0186cbd8c3604ae5540fae18a90cb99"},
-    {file = "websockets-13.0.1-cp313-cp313-macosx_10_13_universal2.whl", hash = "sha256:1ee4cc030a4bdab482a37462dbf3ffb7e09334d01dd37d1063be1136a0d825fa"},
-    {file = "websockets-13.0.1-cp313-cp313-macosx_10_13_x86_64.whl", hash = "sha256:dbb0b697cc0655719522406c059eae233abaa3243821cfdfab1215d02ac10231"},
-    {file = "websockets-13.0.1-cp313-cp313-macosx_11_0_arm64.whl", hash = "sha256:acbebec8cb3d4df6e2488fbf34702cbc37fc39ac7abf9449392cefb3305562e9"},
-    {file = "websockets-13.0.1-cp313-cp313-manylinux_2_17_aarch64.manylinux2014_aarch64.whl", hash = "sha256:63848cdb6fcc0bf09d4a155464c46c64ffdb5807ede4fb251da2c2692559ce75"},
-    {file = "websockets-13.0.1-cp313-cp313-manylinux_2_5_i686.manylinux1_i686.manylinux_2_17_i686.manylinux2014_i686.whl", hash = "sha256:872afa52a9f4c414d6955c365b6588bc4401272c629ff8321a55f44e3f62b553"},
-    {file = "websockets-13.0.1-cp313-cp313-manylinux_2_5_x86_64.manylinux1_x86_64.manylinux_2_17_x86_64.manylinux2014_x86_64.whl", hash = "sha256:05e70fec7c54aad4d71eae8e8cab50525e899791fc389ec6f77b95312e4e9920"},
-    {file = "websockets-13.0.1-cp313-cp313-musllinux_1_2_aarch64.whl", hash = "sha256:e82db3756ccb66266504f5a3de05ac6b32f287faacff72462612120074103329"},
-    {file = "websockets-13.0.1-cp313-cp313-musllinux_1_2_i686.whl", hash = "sha256:4e85f46ce287f5c52438bb3703d86162263afccf034a5ef13dbe4318e98d86e7"},
-    {file = "websockets-13.0.1-cp313-cp313-musllinux_1_2_x86_64.whl", hash = "sha256:f3fea72e4e6edb983908f0db373ae0732b275628901d909c382aae3b592589f2"},
-    {file = "websockets-13.0.1-cp313-cp313-win32.whl", hash = "sha256:254ecf35572fca01a9f789a1d0f543898e222f7b69ecd7d5381d8d8047627bdb"},
-    {file = "websockets-13.0.1-cp313-cp313-win_amd64.whl", hash = "sha256:ca48914cdd9f2ccd94deab5bcb5ac98025a5ddce98881e5cce762854a5de330b"},
-    {file = "websockets-13.0.1-cp38-cp38-macosx_10_9_universal2.whl", hash = "sha256:b74593e9acf18ea5469c3edaa6b27fa7ecf97b30e9dabd5a94c4c940637ab96e"},
-    {file = "websockets-13.0.1-cp38-cp38-macosx_10_9_x86_64.whl", hash = "sha256:132511bfd42e77d152c919147078460c88a795af16b50e42a0bd14f0ad71ddd2"},
-    {file = "websockets-13.0.1-cp38-cp38-macosx_11_0_arm64.whl", hash = "sha256:165bedf13556f985a2aa064309baa01462aa79bf6112fbd068ae38993a0e1f1b"},
-    {file = "websockets-13.0.1-cp38-cp38-manylinux_2_17_aarch64.manylinux2014_aarch64.whl", hash = "sha256:e801ca2f448850685417d723ec70298feff3ce4ff687c6f20922c7474b4746ae"},
-    {file = "websockets-13.0.1-cp38-cp38-manylinux_2_5_i686.manylinux1_i686.manylinux_2_17_i686.manylinux2014_i686.whl", hash = "sha256:30d3a1f041360f029765d8704eae606781e673e8918e6b2c792e0775de51352f"},
-    {file = "websockets-13.0.1-cp38-cp38-manylinux_2_5_x86_64.manylinux1_x86_64.manylinux_2_17_x86_64.manylinux2014_x86_64.whl", hash = "sha256:67648f5e50231b5a7f6d83b32f9c525e319f0ddc841be0de64f24928cd75a603"},
-    {file = "websockets-13.0.1-cp38-cp38-musllinux_1_2_aarch64.whl", hash = "sha256:4f0426d51c8f0926a4879390f53c7f5a855e42d68df95fff6032c82c888b5f36"},
-    {file = "websockets-13.0.1-cp38-cp38-musllinux_1_2_i686.whl", hash = "sha256:ef48e4137e8799998a343706531e656fdec6797b80efd029117edacb74b0a10a"},
-    {file = "websockets-13.0.1-cp38-cp38-musllinux_1_2_x86_64.whl", hash = "sha256:249aab278810bee585cd0d4de2f08cfd67eed4fc75bde623be163798ed4db2eb"},
-    {file = "websockets-13.0.1-cp38-cp38-win32.whl", hash = "sha256:06c0a667e466fcb56a0886d924b5f29a7f0886199102f0a0e1c60a02a3751cb4"},
-    {file = "websockets-13.0.1-cp38-cp38-win_amd64.whl", hash = "sha256:1f3cf6d6ec1142412d4535adabc6bd72a63f5f148c43fe559f06298bc21953c9"},
-    {file = "websockets-13.0.1-cp39-cp39-macosx_10_9_universal2.whl", hash = "sha256:1fa082ea38d5de51dd409434edc27c0dcbd5fed2b09b9be982deb6f0508d25bc"},
-    {file = "websockets-13.0.1-cp39-cp39-macosx_10_9_x86_64.whl", hash = "sha256:4a365bcb7be554e6e1f9f3ed64016e67e2fa03d7b027a33e436aecf194febb63"},
-    {file = "websockets-13.0.1-cp39-cp39-macosx_11_0_arm64.whl", hash = "sha256:10a0dc7242215d794fb1918f69c6bb235f1f627aaf19e77f05336d147fce7c37"},
-    {file = "websockets-13.0.1-cp39-cp39-manylinux_2_17_aarch64.manylinux2014_aarch64.whl", hash = "sha256:59197afd478545b1f73367620407b0083303569c5f2d043afe5363676f2697c9"},
-    {file = "websockets-13.0.1-cp39-cp39-manylinux_2_5_i686.manylinux1_i686.manylinux_2_17_i686.manylinux2014_i686.whl", hash = "sha256:7d20516990d8ad557b5abeb48127b8b779b0b7e6771a265fa3e91767596d7d97"},
-    {file = "websockets-13.0.1-cp39-cp39-manylinux_2_5_x86_64.manylinux1_x86_64.manylinux_2_17_x86_64.manylinux2014_x86_64.whl", hash = "sha256:a1a2e272d067030048e1fe41aa1ec8cfbbaabce733b3d634304fa2b19e5c897f"},
-    {file = "websockets-13.0.1-cp39-cp39-musllinux_1_2_aarch64.whl", hash = "sha256:ad327ac80ba7ee61da85383ca8822ff808ab5ada0e4a030d66703cc025b021c4"},
-    {file = "websockets-13.0.1-cp39-cp39-musllinux_1_2_i686.whl", hash = "sha256:518f90e6dd089d34eaade01101fd8a990921c3ba18ebbe9b0165b46ebff947f0"},
-    {file = "websockets-13.0.1-cp39-cp39-musllinux_1_2_x86_64.whl", hash = "sha256:68264802399aed6fe9652e89761031acc734fc4c653137a5911c2bfa995d6d6d"},
-    {file = "websockets-13.0.1-cp39-cp39-win32.whl", hash = "sha256:a5dc0c42ded1557cc7c3f0240b24129aefbad88af4f09346164349391dea8e58"},
-    {file = "websockets-13.0.1-cp39-cp39-win_amd64.whl", hash = "sha256:b448a0690ef43db5ef31b3a0d9aea79043882b4632cfc3eaab20105edecf6097"},
-    {file = "websockets-13.0.1-pp310-pypy310_pp73-macosx_10_15_x86_64.whl", hash = "sha256:faef9ec6354fe4f9a2c0bbb52fb1ff852effc897e2a4501e25eb3a47cb0a4f89"},
-    {file = "websockets-13.0.1-pp310-pypy310_pp73-macosx_11_0_arm64.whl", hash = "sha256:03d3f9ba172e0a53e37fa4e636b86cc60c3ab2cfee4935e66ed1d7acaa4625ad"},
-    {file = "websockets-13.0.1-pp310-pypy310_pp73-manylinux_2_17_aarch64.manylinux2014_aarch64.whl", hash = "sha256:d450f5a7a35662a9b91a64aefa852f0c0308ee256122f5218a42f1d13577d71e"},
-    {file = "websockets-13.0.1-pp310-pypy310_pp73-manylinux_2_5_i686.manylinux1_i686.manylinux_2_17_i686.manylinux2014_i686.whl", hash = "sha256:3f55b36d17ac50aa8a171b771e15fbe1561217510c8768af3d546f56c7576cdc"},
-    {file = "websockets-13.0.1-pp310-pypy310_pp73-manylinux_2_5_x86_64.manylinux1_x86_64.manylinux_2_17_x86_64.manylinux2014_x86_64.whl", hash = "sha256:14b9c006cac63772b31abbcd3e3abb6228233eec966bf062e89e7fa7ae0b7333"},
-    {file = "websockets-13.0.1-pp310-pypy310_pp73-win_amd64.whl", hash = "sha256:b79915a1179a91f6c5f04ece1e592e2e8a6bd245a0e45d12fd56b2b59e559a32"},
-    {file = "websockets-13.0.1-pp38-pypy38_pp73-macosx_10_9_x86_64.whl", hash = "sha256:f40de079779acbcdbb6ed4c65af9f018f8b77c5ec4e17a4b737c05c2db554491"},
-    {file = "websockets-13.0.1-pp38-pypy38_pp73-macosx_11_0_arm64.whl", hash = "sha256:80e4ba642fc87fa532bac07e5ed7e19d56940b6af6a8c61d4429be48718a380f"},
-    {file = "websockets-13.0.1-pp38-pypy38_pp73-manylinux_2_17_aarch64.manylinux2014_aarch64.whl", hash = "sha256:2a02b0161c43cc9e0232711eff846569fad6ec836a7acab16b3cf97b2344c060"},
-    {file = "websockets-13.0.1-pp38-pypy38_pp73-manylinux_2_5_i686.manylinux1_i686.manylinux_2_17_i686.manylinux2014_i686.whl", hash = "sha256:6aa74a45d4cdc028561a7d6ab3272c8b3018e23723100b12e58be9dfa5a24491"},
-    {file = "websockets-13.0.1-pp38-pypy38_pp73-manylinux_2_5_x86_64.manylinux1_x86_64.manylinux_2_17_x86_64.manylinux2014_x86_64.whl", hash = "sha256:00fd961943b6c10ee6f0b1130753e50ac5dcd906130dcd77b0003c3ab797d026"},
-    {file = "websockets-13.0.1-pp38-pypy38_pp73-win_amd64.whl", hash = "sha256:d93572720d781331fb10d3da9ca1067817d84ad1e7c31466e9f5e59965618096"},
-    {file = "websockets-13.0.1-pp39-pypy39_pp73-macosx_10_15_x86_64.whl", hash = "sha256:71e6e5a3a3728886caee9ab8752e8113670936a193284be9d6ad2176a137f376"},
-    {file = "websockets-13.0.1-pp39-pypy39_pp73-macosx_11_0_arm64.whl", hash = "sha256:c4a6343e3b0714e80da0b0893543bf9a5b5fa71b846ae640e56e9abc6fbc4c83"},
-    {file = "websockets-13.0.1-pp39-pypy39_pp73-manylinux_2_17_aarch64.manylinux2014_aarch64.whl", hash = "sha256:1a678532018e435396e37422a95e3ab87f75028ac79570ad11f5bf23cd2a7d8c"},
-    {file = "websockets-13.0.1-pp39-pypy39_pp73-manylinux_2_5_i686.manylinux1_i686.manylinux_2_17_i686.manylinux2014_i686.whl", hash = "sha256:d6716c087e4aa0b9260c4e579bb82e068f84faddb9bfba9906cb87726fa2e870"},
-    {file = "websockets-13.0.1-pp39-pypy39_pp73-manylinux_2_5_x86_64.manylinux1_x86_64.manylinux_2_17_x86_64.manylinux2014_x86_64.whl", hash = "sha256:e33505534f3f673270dd67f81e73550b11de5b538c56fe04435d63c02c3f26b5"},
-    {file = "websockets-13.0.1-pp39-pypy39_pp73-win_amd64.whl", hash = "sha256:acab3539a027a85d568c2573291e864333ec9d912675107d6efceb7e2be5d980"},
-    {file = "websockets-13.0.1-py3-none-any.whl", hash = "sha256:b80f0c51681c517604152eb6a572f5a9378f877763231fddb883ba2f968e8817"},
-    {file = "websockets-13.0.1.tar.gz", hash = "sha256:4d6ece65099411cfd9a48d13701d7438d9c34f479046b34c50ff60bb8834e43e"},
+    {file = "websockets-13.1-cp310-cp310-macosx_10_9_universal2.whl", hash = "sha256:f48c749857f8fb598fb890a75f540e3221d0976ed0bf879cf3c7eef34151acee"},
+    {file = "websockets-13.1-cp310-cp310-macosx_10_9_x86_64.whl", hash = "sha256:c7e72ce6bda6fb9409cc1e8164dd41d7c91466fb599eb047cfda72fe758a34a7"},
+    {file = "websockets-13.1-cp310-cp310-macosx_11_0_arm64.whl", hash = "sha256:f779498eeec470295a2b1a5d97aa1bc9814ecd25e1eb637bd9d1c73a327387f6"},
+    {file = "websockets-13.1-cp310-cp310-manylinux_2_17_aarch64.manylinux2014_aarch64.whl", hash = "sha256:4676df3fe46956fbb0437d8800cd5f2b6d41143b6e7e842e60554398432cf29b"},
+    {file = "websockets-13.1-cp310-cp310-manylinux_2_5_i686.manylinux1_i686.manylinux_2_17_i686.manylinux2014_i686.whl", hash = "sha256:a7affedeb43a70351bb811dadf49493c9cfd1ed94c9c70095fd177e9cc1541fa"},
+    {file = "websockets-13.1-cp310-cp310-manylinux_2_5_x86_64.manylinux1_x86_64.manylinux_2_17_x86_64.manylinux2014_x86_64.whl", hash = "sha256:1971e62d2caa443e57588e1d82d15f663b29ff9dfe7446d9964a4b6f12c1e700"},
+    {file = "websockets-13.1-cp310-cp310-musllinux_1_2_aarch64.whl", hash = "sha256:5f2e75431f8dc4a47f31565a6e1355fb4f2ecaa99d6b89737527ea917066e26c"},
+    {file = "websockets-13.1-cp310-cp310-musllinux_1_2_i686.whl", hash = "sha256:58cf7e75dbf7e566088b07e36ea2e3e2bd5676e22216e4cad108d4df4a7402a0"},
+    {file = "websockets-13.1-cp310-cp310-musllinux_1_2_x86_64.whl", hash = "sha256:c90d6dec6be2c7d03378a574de87af9b1efea77d0c52a8301dd831ece938452f"},
+    {file = "websockets-13.1-cp310-cp310-win32.whl", hash = "sha256:730f42125ccb14602f455155084f978bd9e8e57e89b569b4d7f0f0c17a448ffe"},
+    {file = "websockets-13.1-cp310-cp310-win_amd64.whl", hash = "sha256:5993260f483d05a9737073be197371940c01b257cc45ae3f1d5d7adb371b266a"},
+    {file = "websockets-13.1-cp311-cp311-macosx_10_9_universal2.whl", hash = "sha256:61fc0dfcda609cda0fc9fe7977694c0c59cf9d749fbb17f4e9483929e3c48a19"},
+    {file = "websockets-13.1-cp311-cp311-macosx_10_9_x86_64.whl", hash = "sha256:ceec59f59d092c5007e815def4ebb80c2de330e9588e101cf8bd94c143ec78a5"},
+    {file = "websockets-13.1-cp311-cp311-macosx_11_0_arm64.whl", hash = "sha256:c1dca61c6db1166c48b95198c0b7d9c990b30c756fc2923cc66f68d17dc558fd"},
+    {file = "websockets-13.1-cp311-cp311-manylinux_2_17_aarch64.manylinux2014_aarch64.whl", hash = "sha256:308e20f22c2c77f3f39caca508e765f8725020b84aa963474e18c59accbf4c02"},
+    {file = "websockets-13.1-cp311-cp311-manylinux_2_5_i686.manylinux1_i686.manylinux_2_17_i686.manylinux2014_i686.whl", hash = "sha256:62d516c325e6540e8a57b94abefc3459d7dab8ce52ac75c96cad5549e187e3a7"},
+    {file = "websockets-13.1-cp311-cp311-manylinux_2_5_x86_64.manylinux1_x86_64.manylinux_2_17_x86_64.manylinux2014_x86_64.whl", hash = "sha256:87c6e35319b46b99e168eb98472d6c7d8634ee37750d7693656dc766395df096"},
+    {file = "websockets-13.1-cp311-cp311-musllinux_1_2_aarch64.whl", hash = "sha256:5f9fee94ebafbc3117c30be1844ed01a3b177bb6e39088bc6b2fa1dc15572084"},
+    {file = "websockets-13.1-cp311-cp311-musllinux_1_2_i686.whl", hash = "sha256:7c1e90228c2f5cdde263253fa5db63e6653f1c00e7ec64108065a0b9713fa1b3"},
+    {file = "websockets-13.1-cp311-cp311-musllinux_1_2_x86_64.whl", hash = "sha256:6548f29b0e401eea2b967b2fdc1c7c7b5ebb3eeb470ed23a54cd45ef078a0db9"},
+    {file = "websockets-13.1-cp311-cp311-win32.whl", hash = "sha256:c11d4d16e133f6df8916cc5b7e3e96ee4c44c936717d684a94f48f82edb7c92f"},
+    {file = "websockets-13.1-cp311-cp311-win_amd64.whl", hash = "sha256:d04f13a1d75cb2b8382bdc16ae6fa58c97337253826dfe136195b7f89f661557"},
+    {file = "websockets-13.1-cp312-cp312-macosx_10_9_universal2.whl", hash = "sha256:9d75baf00138f80b48f1eac72ad1535aac0b6461265a0bcad391fc5aba875cfc"},
+    {file = "websockets-13.1-cp312-cp312-macosx_10_9_x86_64.whl", hash = "sha256:9b6f347deb3dcfbfde1c20baa21c2ac0751afaa73e64e5b693bb2b848efeaa49"},
+    {file = "websockets-13.1-cp312-cp312-macosx_11_0_arm64.whl", hash = "sha256:de58647e3f9c42f13f90ac7e5f58900c80a39019848c5547bc691693098ae1bd"},
+    {file = "websockets-13.1-cp312-cp312-manylinux_2_17_aarch64.manylinux2014_aarch64.whl", hash = "sha256:a1b54689e38d1279a51d11e3467dd2f3a50f5f2e879012ce8f2d6943f00e83f0"},
+    {file = "websockets-13.1-cp312-cp312-manylinux_2_5_i686.manylinux1_i686.manylinux_2_17_i686.manylinux2014_i686.whl", hash = "sha256:cf1781ef73c073e6b0f90af841aaf98501f975d306bbf6221683dd594ccc52b6"},
+    {file = "websockets-13.1-cp312-cp312-manylinux_2_5_x86_64.manylinux1_x86_64.manylinux_2_17_x86_64.manylinux2014_x86_64.whl", hash = "sha256:8d23b88b9388ed85c6faf0e74d8dec4f4d3baf3ecf20a65a47b836d56260d4b9"},
+    {file = "websockets-13.1-cp312-cp312-musllinux_1_2_aarch64.whl", hash = "sha256:3c78383585f47ccb0fcf186dcb8a43f5438bd7d8f47d69e0b56f71bf431a0a68"},
+    {file = "websockets-13.1-cp312-cp312-musllinux_1_2_i686.whl", hash = "sha256:d6d300f8ec35c24025ceb9b9019ae9040c1ab2f01cddc2bcc0b518af31c75c14"},
+    {file = "websockets-13.1-cp312-cp312-musllinux_1_2_x86_64.whl", hash = "sha256:a9dcaf8b0cc72a392760bb8755922c03e17a5a54e08cca58e8b74f6902b433cf"},
+    {file = "websockets-13.1-cp312-cp312-win32.whl", hash = "sha256:2f85cf4f2a1ba8f602298a853cec8526c2ca42a9a4b947ec236eaedb8f2dc80c"},
+    {file = "websockets-13.1-cp312-cp312-win_amd64.whl", hash = "sha256:38377f8b0cdeee97c552d20cf1865695fcd56aba155ad1b4ca8779a5b6ef4ac3"},
+    {file = "websockets-13.1-cp313-cp313-macosx_10_13_universal2.whl", hash = "sha256:a9ab1e71d3d2e54a0aa646ab6d4eebfaa5f416fe78dfe4da2839525dc5d765c6"},
+    {file = "websockets-13.1-cp313-cp313-macosx_10_13_x86_64.whl", hash = "sha256:b9d7439d7fab4dce00570bb906875734df13d9faa4b48e261c440a5fec6d9708"},
+    {file = "websockets-13.1-cp313-cp313-macosx_11_0_arm64.whl", hash = "sha256:327b74e915cf13c5931334c61e1a41040e365d380f812513a255aa804b183418"},
+    {file = "websockets-13.1-cp313-cp313-manylinux_2_17_aarch64.manylinux2014_aarch64.whl", hash = "sha256:325b1ccdbf5e5725fdcb1b0e9ad4d2545056479d0eee392c291c1bf76206435a"},
+    {file = "websockets-13.1-cp313-cp313-manylinux_2_5_i686.manylinux1_i686.manylinux_2_17_i686.manylinux2014_i686.whl", hash = "sha256:346bee67a65f189e0e33f520f253d5147ab76ae42493804319b5716e46dddf0f"},
+    {file = "websockets-13.1-cp313-cp313-manylinux_2_5_x86_64.manylinux1_x86_64.manylinux_2_17_x86_64.manylinux2014_x86_64.whl", hash = "sha256:91a0fa841646320ec0d3accdff5b757b06e2e5c86ba32af2e0815c96c7a603c5"},
+    {file = "websockets-13.1-cp313-cp313-musllinux_1_2_aarch64.whl", hash = "sha256:18503d2c5f3943e93819238bf20df71982d193f73dcecd26c94514f417f6b135"},
+    {file = "websockets-13.1-cp313-cp313-musllinux_1_2_i686.whl", hash = "sha256:a9cd1af7e18e5221d2878378fbc287a14cd527fdd5939ed56a18df8a31136bb2"},
+    {file = "websockets-13.1-cp313-cp313-musllinux_1_2_x86_64.whl", hash = "sha256:70c5be9f416aa72aab7a2a76c90ae0a4fe2755c1816c153c1a2bcc3333ce4ce6"},
+    {file = "websockets-13.1-cp313-cp313-win32.whl", hash = "sha256:624459daabeb310d3815b276c1adef475b3e6804abaf2d9d2c061c319f7f187d"},
+    {file = "websockets-13.1-cp313-cp313-win_amd64.whl", hash = "sha256:c518e84bb59c2baae725accd355c8dc517b4a3ed8db88b4bc93c78dae2974bf2"},
+    {file = "websockets-13.1-cp38-cp38-macosx_10_9_universal2.whl", hash = "sha256:c7934fd0e920e70468e676fe7f1b7261c1efa0d6c037c6722278ca0228ad9d0d"},
+    {file = "websockets-13.1-cp38-cp38-macosx_10_9_x86_64.whl", hash = "sha256:149e622dc48c10ccc3d2760e5f36753db9cacf3ad7bc7bbbfd7d9c819e286f23"},
+    {file = "websockets-13.1-cp38-cp38-macosx_11_0_arm64.whl", hash = "sha256:a569eb1b05d72f9bce2ebd28a1ce2054311b66677fcd46cf36204ad23acead8c"},
+    {file = "websockets-13.1-cp38-cp38-manylinux_2_17_aarch64.manylinux2014_aarch64.whl", hash = "sha256:95df24ca1e1bd93bbca51d94dd049a984609687cb2fb08a7f2c56ac84e9816ea"},
+    {file = "websockets-13.1-cp38-cp38-manylinux_2_5_i686.manylinux1_i686.manylinux_2_17_i686.manylinux2014_i686.whl", hash = "sha256:d8dbb1bf0c0a4ae8b40bdc9be7f644e2f3fb4e8a9aca7145bfa510d4a374eeb7"},
+    {file = "websockets-13.1-cp38-cp38-manylinux_2_5_x86_64.manylinux1_x86_64.manylinux_2_17_x86_64.manylinux2014_x86_64.whl", hash = "sha256:035233b7531fb92a76beefcbf479504db8c72eb3bff41da55aecce3a0f729e54"},
+    {file = "websockets-13.1-cp38-cp38-musllinux_1_2_aarch64.whl", hash = "sha256:e4450fc83a3df53dec45922b576e91e94f5578d06436871dce3a6be38e40f5db"},
+    {file = "websockets-13.1-cp38-cp38-musllinux_1_2_i686.whl", hash = "sha256:463e1c6ec853202dd3657f156123d6b4dad0c546ea2e2e38be2b3f7c5b8e7295"},
+    {file = "websockets-13.1-cp38-cp38-musllinux_1_2_x86_64.whl", hash = "sha256:6d6855bbe70119872c05107e38fbc7f96b1d8cb047d95c2c50869a46c65a8e96"},
+    {file = "websockets-13.1-cp38-cp38-win32.whl", hash = "sha256:204e5107f43095012b00f1451374693267adbb832d29966a01ecc4ce1db26faf"},
+    {file = "websockets-13.1-cp38-cp38-win_amd64.whl", hash = "sha256:485307243237328c022bc908b90e4457d0daa8b5cf4b3723fd3c4a8012fce4c6"},
+    {file = "websockets-13.1-cp39-cp39-macosx_10_9_universal2.whl", hash = "sha256:9b37c184f8b976f0c0a231a5f3d6efe10807d41ccbe4488df8c74174805eea7d"},
+    {file = "websockets-13.1-cp39-cp39-macosx_10_9_x86_64.whl", hash = "sha256:163e7277e1a0bd9fb3c8842a71661ad19c6aa7bb3d6678dc7f89b17fbcc4aeb7"},
+    {file = "websockets-13.1-cp39-cp39-macosx_11_0_arm64.whl", hash = "sha256:4b889dbd1342820cc210ba44307cf75ae5f2f96226c0038094455a96e64fb07a"},
+    {file = "websockets-13.1-cp39-cp39-manylinux_2_17_aarch64.manylinux2014_aarch64.whl", hash = "sha256:586a356928692c1fed0eca68b4d1c2cbbd1ca2acf2ac7e7ebd3b9052582deefa"},
+    {file = "websockets-13.1-cp39-cp39-manylinux_2_5_i686.manylinux1_i686.manylinux_2_17_i686.manylinux2014_i686.whl", hash = "sha256:7bd6abf1e070a6b72bfeb71049d6ad286852e285f146682bf30d0296f5fbadfa"},
+    {file = "websockets-13.1-cp39-cp39-manylinux_2_5_x86_64.manylinux1_x86_64.manylinux_2_17_x86_64.manylinux2014_x86_64.whl", hash = "sha256:6d2aad13a200e5934f5a6767492fb07151e1de1d6079c003ab31e1823733ae79"},
+    {file = "websockets-13.1-cp39-cp39-musllinux_1_2_aarch64.whl", hash = "sha256:df01aea34b6e9e33572c35cd16bae5a47785e7d5c8cb2b54b2acdb9678315a17"},
+    {file = "websockets-13.1-cp39-cp39-musllinux_1_2_i686.whl", hash = "sha256:e54affdeb21026329fb0744ad187cf812f7d3c2aa702a5edb562b325191fcab6"},
+    {file = "websockets-13.1-cp39-cp39-musllinux_1_2_x86_64.whl", hash = "sha256:9ef8aa8bdbac47f4968a5d66462a2a0935d044bf35c0e5a8af152d58516dbeb5"},
+    {file = "websockets-13.1-cp39-cp39-win32.whl", hash = "sha256:deeb929efe52bed518f6eb2ddc00cc496366a14c726005726ad62c2dd9017a3c"},
+    {file = "websockets-13.1-cp39-cp39-win_amd64.whl", hash = "sha256:7c65ffa900e7cc958cd088b9a9157a8141c991f8c53d11087e6fb7277a03f81d"},
+    {file = "websockets-13.1-pp310-pypy310_pp73-macosx_10_15_x86_64.whl", hash = "sha256:5dd6da9bec02735931fccec99d97c29f47cc61f644264eb995ad6c0c27667238"},
+    {file = "websockets-13.1-pp310-pypy310_pp73-macosx_11_0_arm64.whl", hash = "sha256:2510c09d8e8df777177ee3d40cd35450dc169a81e747455cc4197e63f7e7bfe5"},
+    {file = "websockets-13.1-pp310-pypy310_pp73-manylinux_2_17_aarch64.manylinux2014_aarch64.whl", hash = "sha256:f1c3cf67185543730888b20682fb186fc8d0fa6f07ccc3ef4390831ab4b388d9"},
+    {file = "websockets-13.1-pp310-pypy310_pp73-manylinux_2_5_i686.manylinux1_i686.manylinux_2_17_i686.manylinux2014_i686.whl", hash = "sha256:bcc03c8b72267e97b49149e4863d57c2d77f13fae12066622dc78fe322490fe6"},
+    {file = "websockets-13.1-pp310-pypy310_pp73-manylinux_2_5_x86_64.manylinux1_x86_64.manylinux_2_17_x86_64.manylinux2014_x86_64.whl", hash = "sha256:004280a140f220c812e65f36944a9ca92d766b6cc4560be652a0a3883a79ed8a"},
+    {file = "websockets-13.1-pp310-pypy310_pp73-win_amd64.whl", hash = "sha256:e2620453c075abeb0daa949a292e19f56de518988e079c36478bacf9546ced23"},
+    {file = "websockets-13.1-pp38-pypy38_pp73-macosx_10_9_x86_64.whl", hash = "sha256:9156c45750b37337f7b0b00e6248991a047be4aa44554c9886fe6bdd605aab3b"},
+    {file = "websockets-13.1-pp38-pypy38_pp73-macosx_11_0_arm64.whl", hash = "sha256:80c421e07973a89fbdd93e6f2003c17d20b69010458d3a8e37fb47874bd67d51"},
+    {file = "websockets-13.1-pp38-pypy38_pp73-manylinux_2_17_aarch64.manylinux2014_aarch64.whl", hash = "sha256:82d0ba76371769d6a4e56f7e83bb8e81846d17a6190971e38b5de108bde9b0d7"},
+    {file = "websockets-13.1-pp38-pypy38_pp73-manylinux_2_5_i686.manylinux1_i686.manylinux_2_17_i686.manylinux2014_i686.whl", hash = "sha256:e9875a0143f07d74dc5e1ded1c4581f0d9f7ab86c78994e2ed9e95050073c94d"},
+    {file = "websockets-13.1-pp38-pypy38_pp73-manylinux_2_5_x86_64.manylinux1_x86_64.manylinux_2_17_x86_64.manylinux2014_x86_64.whl", hash = "sha256:a11e38ad8922c7961447f35c7b17bffa15de4d17c70abd07bfbe12d6faa3e027"},
+    {file = "websockets-13.1-pp38-pypy38_pp73-win_amd64.whl", hash = "sha256:4059f790b6ae8768471cddb65d3c4fe4792b0ab48e154c9f0a04cefaabcd5978"},
+    {file = "websockets-13.1-pp39-pypy39_pp73-macosx_10_15_x86_64.whl", hash = "sha256:25c35bf84bf7c7369d247f0b8cfa157f989862c49104c5cf85cb5436a641d93e"},
+    {file = "websockets-13.1-pp39-pypy39_pp73-macosx_11_0_arm64.whl", hash = "sha256:83f91d8a9bb404b8c2c41a707ac7f7f75b9442a0a876df295de27251a856ad09"},
+    {file = "websockets-13.1-pp39-pypy39_pp73-manylinux_2_17_aarch64.manylinux2014_aarch64.whl", hash = "sha256:7a43cfdcddd07f4ca2b1afb459824dd3c6d53a51410636a2c7fc97b9a8cf4842"},
+    {file = "websockets-13.1-pp39-pypy39_pp73-manylinux_2_5_i686.manylinux1_i686.manylinux_2_17_i686.manylinux2014_i686.whl", hash = "sha256:48a2ef1381632a2f0cb4efeff34efa97901c9fbc118e01951ad7cfc10601a9bb"},
+    {file = "websockets-13.1-pp39-pypy39_pp73-manylinux_2_5_x86_64.manylinux1_x86_64.manylinux_2_17_x86_64.manylinux2014_x86_64.whl", hash = "sha256:459bf774c754c35dbb487360b12c5727adab887f1622b8aed5755880a21c4a20"},
+    {file = "websockets-13.1-pp39-pypy39_pp73-win_amd64.whl", hash = "sha256:95858ca14a9f6fa8413d29e0a585b31b278388aa775b8a81fa24830123874678"},
+    {file = "websockets-13.1-py3-none-any.whl", hash = "sha256:a9a396a6ad26130cdae92ae10c36af09d9bfe6cafe69670fd3b6da9b07b4044f"},
+    {file = "websockets-13.1.tar.gz", hash = "sha256:a3b3366087c1bc0a2795111edcadddb8b3b59509d5db5d7ea3fdd69f954a8878"},
 ]
 
 [[package]]
@@ -10501,4 +10539,4 @@ cffi = ["cffi (>=1.11)"]
 [metadata]
 lock-version = "2.0"
 python-versions = ">=3.10,<3.13"
-content-hash = "c4580c22e2b220c8c80dbc3f765060a09e14874ed29b690c13a533bf0365e789"
+content-hash = "74a22b763f5f6f1414e843f1a508ff48210d32f9523b696e5dd443489deaad46"
diff --git a/api/pyproject.toml b/api/pyproject.toml
index e737761f3b2c0..fe9e64907b4c3 100644
--- a/api/pyproject.toml
+++ b/api/pyproject.toml
@@ -221,6 +221,7 @@ volcengine-python-sdk = {extras = ["ark"], version = "^1.0.98"}
 oci = "^2.133.0"
 tos = "^2.7.1"
 nomic = "^3.1.2"
+validators = "0.21.0"
 [tool.poetry.group.indriect.dependencies]
 kaleido = "0.2.1"
 rank-bm25 = "~0.2.2"
diff --git a/api/schedule/clean_unused_messages_task.py b/api/schedule/clean_unused_messages_task.py
new file mode 100644
index 0000000000000..85e6a58a0e90c
--- /dev/null
+++ b/api/schedule/clean_unused_messages_task.py
@@ -0,0 +1,92 @@
+import datetime
+import time
+
+import click
+from sqlalchemy import func
+from werkzeug.exceptions import NotFound
+
+import app
+from configs import dify_config
+from core.rag.index_processor.index_processor_factory import IndexProcessorFactory
+from extensions.ext_database import db
+from models.dataset import Dataset, DatasetQuery, Document
+
+
+@app.celery.task(queue="dataset")
+def clean_unused_message_task():
+    click.echo(click.style("Start clean unused messages .", fg="green"))
+    clean_days = int(dify_config.CLEAN_DAY_SETTING)
+    start_at = time.perf_counter()
+    thirty_days_ago = datetime.datetime.now() - datetime.timedelta(days=clean_days)
+    page = 1
+    while True:
+        try:
+            # Subquery for counting new documents
+            document_subquery_new = (
+                db.session.query(Document.dataset_id, func.count(Document.id).label("document_count"))
+                .filter(
+                    Document.indexing_status == "completed",
+                    Document.enabled == True,
+                    Document.archived == False,
+                    Document.updated_at > thirty_days_ago,
+                )
+                .group_by(Document.dataset_id)
+                .subquery()
+            )
+
+            # Subquery for counting old documents
+            document_subquery_old = (
+                db.session.query(Document.dataset_id, func.count(Document.id).label("document_count"))
+                .filter(
+                    Document.indexing_status == "completed",
+                    Document.enabled == True,
+                    Document.archived == False,
+                    Document.updated_at < thirty_days_ago,
+                )
+                .group_by(Document.dataset_id)
+                .subquery()
+            )
+
+            # Main query with join and filter
+            datasets = (
+                db.session.query(Dataset)
+                .outerjoin(document_subquery_new, Dataset.id == document_subquery_new.c.dataset_id)
+                .outerjoin(document_subquery_old, Dataset.id == document_subquery_old.c.dataset_id)
+                .filter(
+                    Dataset.created_at < thirty_days_ago,
+                    func.coalesce(document_subquery_new.c.document_count, 0) == 0,
+                    func.coalesce(document_subquery_old.c.document_count, 0) > 0,
+                )
+                .order_by(Dataset.created_at.desc())
+                .paginate(page=page, per_page=50)
+            )
+
+        except NotFound:
+            break
+        if datasets.items is None or len(datasets.items) == 0:
+            break
+        page += 1
+        for dataset in datasets:
+            dataset_query = (
+                db.session.query(DatasetQuery)
+                .filter(DatasetQuery.created_at > thirty_days_ago, DatasetQuery.dataset_id == dataset.id)
+                .all()
+            )
+            if not dataset_query or len(dataset_query) == 0:
+                try:
+                    # remove index
+                    index_processor = IndexProcessorFactory(dataset.doc_form).init_index_processor()
+                    index_processor.clean(dataset, None)
+
+                    # update document
+                    update_params = {Document.enabled: False}
+
+                    Document.query.filter_by(dataset_id=dataset.id).update(update_params)
+                    db.session.commit()
+                    click.echo(click.style("Cleaned unused dataset {} from db success!".format(dataset.id), fg="green"))
+                except Exception as e:
+                    click.echo(
+                        click.style("clean dataset index error: {} {}".format(e.__class__.__name__, str(e)), fg="red")
+                    )
+    end_at = time.perf_counter()
+    click.echo(click.style("Cleaned unused dataset from db success latency: {}".format(end_at - start_at), fg="green"))
diff --git a/api/services/dataset_service.py b/api/services/dataset_service.py
index e96f06ed40fa4..c165def6d5c1e 100644
--- a/api/services/dataset_service.py
+++ b/api/services/dataset_service.py
@@ -32,6 +32,7 @@
     DatasetQuery,
     Document,
     DocumentSegment,
+    ExternalKnowledgeBindings,
 )
 from models.model import UploadFile
 from models.source import DataSourceOauthBinding
@@ -39,6 +40,7 @@
 from services.errors.dataset import DatasetNameDuplicateError
 from services.errors.document import DocumentIndexingError
 from services.errors.file import FileNotExistsError
+from services.external_knowledge_service import ExternalDatasetService
 from services.feature_service import FeatureModel, FeatureService
 from services.tag_service import TagService
 from services.vector_service import VectorService
@@ -56,10 +58,8 @@
 
 class DatasetService:
     @staticmethod
-    def get_datasets(page, per_page, provider="vendor", tenant_id=None, user=None, search=None, tag_ids=None):
-        query = Dataset.query.filter(Dataset.provider == provider, Dataset.tenant_id == tenant_id).order_by(
-            Dataset.created_at.desc()
-        )
+    def get_datasets(page, per_page, tenant_id=None, user=None, search=None, tag_ids=None):
+        query = Dataset.query.filter(Dataset.tenant_id == tenant_id).order_by(Dataset.created_at.desc())
 
         if user:
             # get permitted dataset ids
@@ -137,7 +137,14 @@ def get_datasets_by_ids(ids, tenant_id):
 
     @staticmethod
     def create_empty_dataset(
-        tenant_id: str, name: str, indexing_technique: Optional[str], account: Account, permission: Optional[str] = None
+        tenant_id: str,
+        name: str,
+        indexing_technique: Optional[str],
+        account: Account,
+        permission: Optional[str] = None,
+        provider: str = "vendor",
+        external_knowledge_api_id: Optional[str] = None,
+        external_knowledge_id: Optional[str] = None,
     ):
         # check if dataset name already exists
         if Dataset.query.filter_by(name=name, tenant_id=tenant_id).first():
@@ -156,12 +163,28 @@ def create_empty_dataset(
         dataset.embedding_model_provider = embedding_model.provider if embedding_model else None
         dataset.embedding_model = embedding_model.model if embedding_model else None
         dataset.permission = permission or DatasetPermissionEnum.ONLY_ME
+        dataset.provider = provider
         db.session.add(dataset)
+        db.session.flush()
+
+        if provider == "external" and external_knowledge_api_id:
+            external_knowledge_api = ExternalDatasetService.get_external_knowledge_api(external_knowledge_api_id)
+            if not external_knowledge_api:
+                raise ValueError("External API template not found.")
+            external_knowledge_binding = ExternalKnowledgeBindings(
+                tenant_id=tenant_id,
+                dataset_id=dataset.id,
+                external_knowledge_api_id=external_knowledge_api_id,
+                external_knowledge_id=external_knowledge_id,
+                created_by=account.id,
+            )
+            db.session.add(external_knowledge_binding)
+
         db.session.commit()
         return dataset
 
     @staticmethod
-    def get_dataset(dataset_id):
+    def get_dataset(dataset_id) -> Dataset:
         return Dataset.query.filter_by(id=dataset_id).first()
 
     @staticmethod
@@ -202,81 +225,103 @@ def check_embedding_model_setting(tenant_id: str, embedding_model_provider: str,
 
     @staticmethod
     def update_dataset(dataset_id, data, user):
-        data.pop("partial_member_list", None)
-        filtered_data = {k: v for k, v in data.items() if v is not None or k == "description"}
         dataset = DatasetService.get_dataset(dataset_id)
+
         DatasetService.check_dataset_permission(dataset, user)
-        action = None
-        if dataset.indexing_technique != data["indexing_technique"]:
-            # if update indexing_technique
-            if data["indexing_technique"] == "economy":
-                action = "remove"
-                filtered_data["embedding_model"] = None
-                filtered_data["embedding_model_provider"] = None
-                filtered_data["collection_binding_id"] = None
-            elif data["indexing_technique"] == "high_quality":
-                action = "add"
-                # get embedding model setting
-                try:
-                    model_manager = ModelManager()
-                    embedding_model = model_manager.get_model_instance(
-                        tenant_id=current_user.current_tenant_id,
-                        provider=data["embedding_model_provider"],
-                        model_type=ModelType.TEXT_EMBEDDING,
-                        model=data["embedding_model"],
-                    )
-                    filtered_data["embedding_model"] = embedding_model.model
-                    filtered_data["embedding_model_provider"] = embedding_model.provider
-                    dataset_collection_binding = DatasetCollectionBindingService.get_dataset_collection_binding(
-                        embedding_model.provider, embedding_model.model
-                    )
-                    filtered_data["collection_binding_id"] = dataset_collection_binding.id
-                except LLMBadRequestError:
-                    raise ValueError(
-                        "No Embedding Model available. Please configure a valid provider "
-                        "in the Settings -> Model Provider."
-                    )
-                except ProviderTokenNotInitError as ex:
-                    raise ValueError(ex.description)
-        else:
+        if dataset.provider == "external":
+            dataset.retrieval_model = data.get("external_retrieval_model", None)
+            dataset.name = data.get("name", dataset.name)
+            dataset.description = data.get("description", "")
+            external_knowledge_id = data.get("external_knowledge_id", None)
+            db.session.add(dataset)
+            if not external_knowledge_id:
+                raise ValueError("External knowledge id is required.")
+            external_knowledge_api_id = data.get("external_knowledge_api_id", None)
+            if not external_knowledge_api_id:
+                raise ValueError("External knowledge api id is required.")
+            external_knowledge_binding = ExternalKnowledgeBindings.query.filter_by(dataset_id=dataset_id).first()
             if (
-                data["embedding_model_provider"] != dataset.embedding_model_provider
-                or data["embedding_model"] != dataset.embedding_model
+                external_knowledge_binding.external_knowledge_id != external_knowledge_id
+                or external_knowledge_binding.external_knowledge_api_id != external_knowledge_api_id
             ):
-                action = "update"
-                try:
-                    model_manager = ModelManager()
-                    embedding_model = model_manager.get_model_instance(
-                        tenant_id=current_user.current_tenant_id,
-                        provider=data["embedding_model_provider"],
-                        model_type=ModelType.TEXT_EMBEDDING,
-                        model=data["embedding_model"],
-                    )
-                    filtered_data["embedding_model"] = embedding_model.model
-                    filtered_data["embedding_model_provider"] = embedding_model.provider
-                    dataset_collection_binding = DatasetCollectionBindingService.get_dataset_collection_binding(
-                        embedding_model.provider, embedding_model.model
-                    )
-                    filtered_data["collection_binding_id"] = dataset_collection_binding.id
-                except LLMBadRequestError:
-                    raise ValueError(
-                        "No Embedding Model available. Please configure a valid provider "
-                        "in the Settings -> Model Provider."
-                    )
-                except ProviderTokenNotInitError as ex:
-                    raise ValueError(ex.description)
+                external_knowledge_binding.external_knowledge_id = external_knowledge_id
+                external_knowledge_binding.external_knowledge_api_id = external_knowledge_api_id
+                db.session.add(external_knowledge_binding)
+            db.session.commit()
+        else:
+            data.pop("partial_member_list", None)
+            filtered_data = {k: v for k, v in data.items() if v is not None or k == "description"}
+            action = None
+            if dataset.indexing_technique != data["indexing_technique"]:
+                # if update indexing_technique
+                if data["indexing_technique"] == "economy":
+                    action = "remove"
+                    filtered_data["embedding_model"] = None
+                    filtered_data["embedding_model_provider"] = None
+                    filtered_data["collection_binding_id"] = None
+                elif data["indexing_technique"] == "high_quality":
+                    action = "add"
+                    # get embedding model setting
+                    try:
+                        model_manager = ModelManager()
+                        embedding_model = model_manager.get_model_instance(
+                            tenant_id=current_user.current_tenant_id,
+                            provider=data["embedding_model_provider"],
+                            model_type=ModelType.TEXT_EMBEDDING,
+                            model=data["embedding_model"],
+                        )
+                        filtered_data["embedding_model"] = embedding_model.model
+                        filtered_data["embedding_model_provider"] = embedding_model.provider
+                        dataset_collection_binding = DatasetCollectionBindingService.get_dataset_collection_binding(
+                            embedding_model.provider, embedding_model.model
+                        )
+                        filtered_data["collection_binding_id"] = dataset_collection_binding.id
+                    except LLMBadRequestError:
+                        raise ValueError(
+                            "No Embedding Model available. Please configure a valid provider "
+                            "in the Settings -> Model Provider."
+                        )
+                    except ProviderTokenNotInitError as ex:
+                        raise ValueError(ex.description)
+            else:
+                if (
+                    data["embedding_model_provider"] != dataset.embedding_model_provider
+                    or data["embedding_model"] != dataset.embedding_model
+                ):
+                    action = "update"
+                    try:
+                        model_manager = ModelManager()
+                        embedding_model = model_manager.get_model_instance(
+                            tenant_id=current_user.current_tenant_id,
+                            provider=data["embedding_model_provider"],
+                            model_type=ModelType.TEXT_EMBEDDING,
+                            model=data["embedding_model"],
+                        )
+                        filtered_data["embedding_model"] = embedding_model.model
+                        filtered_data["embedding_model_provider"] = embedding_model.provider
+                        dataset_collection_binding = DatasetCollectionBindingService.get_dataset_collection_binding(
+                            embedding_model.provider, embedding_model.model
+                        )
+                        filtered_data["collection_binding_id"] = dataset_collection_binding.id
+                    except LLMBadRequestError:
+                        raise ValueError(
+                            "No Embedding Model available. Please configure a valid provider "
+                            "in the Settings -> Model Provider."
+                        )
+                    except ProviderTokenNotInitError as ex:
+                        raise ValueError(ex.description)
 
-        filtered_data["updated_by"] = user.id
-        filtered_data["updated_at"] = datetime.datetime.now()
+            filtered_data["updated_by"] = user.id
+            filtered_data["updated_at"] = datetime.datetime.now()
 
-        # update Retrieval model
-        filtered_data["retrieval_model"] = data["retrieval_model"]
+            # update Retrieval model
+            filtered_data["retrieval_model"] = data["retrieval_model"]
 
-        dataset.query.filter_by(id=dataset_id).update(filtered_data)
+            dataset.query.filter_by(id=dataset_id).update(filtered_data)
 
-        db.session.commit()
-        if action:
-            deal_dataset_vector_index_task.delay(dataset_id, action)
+            db.session.commit()
+            if action:
+                deal_dataset_vector_index_task.delay(dataset_id, action)
         return dataset
 
     @staticmethod
diff --git a/api/services/entities/external_knowledge_entities/external_knowledge_entities.py b/api/services/entities/external_knowledge_entities/external_knowledge_entities.py
new file mode 100644
index 0000000000000..4545f385eb989
--- /dev/null
+++ b/api/services/entities/external_knowledge_entities/external_knowledge_entities.py
@@ -0,0 +1,26 @@
+from typing import Literal, Optional, Union
+
+from pydantic import BaseModel
+
+
+class AuthorizationConfig(BaseModel):
+    type: Literal[None, "basic", "bearer", "custom"]
+    api_key: Union[None, str] = None
+    header: Union[None, str] = None
+
+
+class Authorization(BaseModel):
+    type: Literal["no-auth", "api-key"]
+    config: Optional[AuthorizationConfig] = None
+
+
+class ProcessStatusSetting(BaseModel):
+    request_method: str
+    url: str
+
+
+class ExternalKnowledgeApiSetting(BaseModel):
+    url: str
+    request_method: str
+    headers: Optional[dict] = None
+    params: Optional[dict] = None
diff --git a/api/services/external_knowledge_service.py b/api/services/external_knowledge_service.py
new file mode 100644
index 0000000000000..4efdf8d7dbe75
--- /dev/null
+++ b/api/services/external_knowledge_service.py
@@ -0,0 +1,274 @@
+import json
+from copy import deepcopy
+from datetime import datetime, timezone
+from typing import Any, Optional, Union
+
+import httpx
+import validators
+
+# from tasks.external_document_indexing_task import external_document_indexing_task
+from core.helper import ssrf_proxy
+from extensions.ext_database import db
+from models.dataset import (
+    Dataset,
+    ExternalKnowledgeApis,
+    ExternalKnowledgeBindings,
+)
+from services.entities.external_knowledge_entities.external_knowledge_entities import (
+    Authorization,
+    ExternalKnowledgeApiSetting,
+)
+from services.errors.dataset import DatasetNameDuplicateError
+
+
+class ExternalDatasetService:
+    @staticmethod
+    def get_external_knowledge_apis(page, per_page, tenant_id, search=None) -> tuple[list[ExternalKnowledgeApis], int]:
+        query = ExternalKnowledgeApis.query.filter(ExternalKnowledgeApis.tenant_id == tenant_id).order_by(
+            ExternalKnowledgeApis.created_at.desc()
+        )
+        if search:
+            query = query.filter(ExternalKnowledgeApis.name.ilike(f"%{search}%"))
+
+        external_knowledge_apis = query.paginate(page=page, per_page=per_page, max_per_page=100, error_out=False)
+
+        return external_knowledge_apis.items, external_knowledge_apis.total
+
+    @classmethod
+    def validate_api_list(cls, api_settings: dict):
+        if not api_settings:
+            raise ValueError("api list is empty")
+        if "endpoint" not in api_settings and not api_settings["endpoint"]:
+            raise ValueError("endpoint is required")
+        if "api_key" not in api_settings and not api_settings["api_key"]:
+            raise ValueError("api_key is required")
+
+    @staticmethod
+    def create_external_knowledge_api(tenant_id: str, user_id: str, args: dict) -> ExternalKnowledgeApis:
+        ExternalDatasetService.check_endpoint_and_api_key(args.get("settings"))
+        external_knowledge_api = ExternalKnowledgeApis(
+            tenant_id=tenant_id,
+            created_by=user_id,
+            updated_by=user_id,
+            name=args.get("name"),
+            description=args.get("description", ""),
+            settings=json.dumps(args.get("settings"), ensure_ascii=False),
+        )
+
+        db.session.add(external_knowledge_api)
+        db.session.commit()
+        return external_knowledge_api
+
+    @staticmethod
+    def check_endpoint_and_api_key(settings: dict):
+        if "endpoint" not in settings or not settings["endpoint"]:
+            raise ValueError("endpoint is required")
+        if "api_key" not in settings or not settings["api_key"]:
+            raise ValueError("api_key is required")
+
+        endpoint = f"{settings['endpoint']}/retrieval"
+        api_key = settings["api_key"]
+        if not validators.url(endpoint):
+            raise ValueError(f"invalid endpoint: {endpoint}")
+        try:
+            response = httpx.post(endpoint, headers={"Authorization": f"Bearer {api_key}"})
+        except Exception as e:
+            raise ValueError(f"failed to connect to the endpoint: {endpoint}")
+        if response.status_code == 502:
+            raise ValueError(f"Bad Gateway: failed to connect to the endpoint: {endpoint}")
+        if response.status_code == 404:
+            raise ValueError(f"Not Found: failed to connect to the endpoint: {endpoint}")
+        if response.status_code == 403:
+            raise ValueError(f"Forbidden: Authorization failed with api_key: {api_key}")
+
+    @staticmethod
+    def get_external_knowledge_api(external_knowledge_api_id: str) -> ExternalKnowledgeApis:
+        return ExternalKnowledgeApis.query.filter_by(id=external_knowledge_api_id).first()
+
+    @staticmethod
+    def update_external_knowledge_api(tenant_id, user_id, external_knowledge_api_id, args) -> ExternalKnowledgeApis:
+        external_knowledge_api = ExternalKnowledgeApis.query.filter_by(
+            id=external_knowledge_api_id, tenant_id=tenant_id
+        ).first()
+        if external_knowledge_api is None:
+            raise ValueError("api template not found")
+
+        external_knowledge_api.name = args.get("name")
+        external_knowledge_api.description = args.get("description", "")
+        external_knowledge_api.settings = json.dumps(args.get("settings"), ensure_ascii=False)
+        external_knowledge_api.updated_by = user_id
+        external_knowledge_api.updated_at = datetime.now(timezone.utc).replace(tzinfo=None)
+        db.session.commit()
+
+        return external_knowledge_api
+
+    @staticmethod
+    def delete_external_knowledge_api(tenant_id: str, external_knowledge_api_id: str):
+        external_knowledge_api = ExternalKnowledgeApis.query.filter_by(
+            id=external_knowledge_api_id, tenant_id=tenant_id
+        ).first()
+        if external_knowledge_api is None:
+            raise ValueError("api template not found")
+
+        db.session.delete(external_knowledge_api)
+        db.session.commit()
+
+    @staticmethod
+    def external_knowledge_api_use_check(external_knowledge_api_id: str) -> tuple[bool, int]:
+        count = ExternalKnowledgeBindings.query.filter_by(external_knowledge_api_id=external_knowledge_api_id).count()
+        if count > 0:
+            return True, count
+        return False, 0
+
+    @staticmethod
+    def get_external_knowledge_binding_with_dataset_id(tenant_id: str, dataset_id: str) -> ExternalKnowledgeBindings:
+        external_knowledge_binding = ExternalKnowledgeBindings.query.filter_by(
+            dataset_id=dataset_id, tenant_id=tenant_id
+        ).first()
+        if not external_knowledge_binding:
+            raise ValueError("external knowledge binding not found")
+        return external_knowledge_binding
+
+    @staticmethod
+    def document_create_args_validate(tenant_id: str, external_knowledge_api_id: str, process_parameter: dict):
+        external_knowledge_api = ExternalKnowledgeApis.query.filter_by(
+            id=external_knowledge_api_id, tenant_id=tenant_id
+        ).first()
+        if external_knowledge_api is None:
+            raise ValueError("api template not found")
+        settings = json.loads(external_knowledge_api.settings)
+        for setting in settings:
+            custom_parameters = setting.get("document_process_setting")
+            if custom_parameters:
+                for parameter in custom_parameters:
+                    if parameter.get("required", False) and not process_parameter.get(parameter.get("name")):
+                        raise ValueError(f'{parameter.get("name")} is required')
+
+    @staticmethod
+    def process_external_api(
+        settings: ExternalKnowledgeApiSetting, files: Union[None, dict[str, Any]]
+    ) -> httpx.Response:
+        """
+        do http request depending on api bundle
+        """
+
+        kwargs = {
+            "url": settings.url,
+            "headers": settings.headers,
+            "follow_redirects": True,
+        }
+
+        response = getattr(ssrf_proxy, settings.request_method)(data=json.dumps(settings.params), files=files, **kwargs)
+
+        return response
+
+    @staticmethod
+    def assembling_headers(authorization: Authorization, headers: Optional[dict] = None) -> dict[str, Any]:
+        authorization = deepcopy(authorization)
+        if headers:
+            headers = deepcopy(headers)
+        else:
+            headers = {}
+        if authorization.type == "api-key":
+            if authorization.config is None:
+                raise ValueError("authorization config is required")
+
+            if authorization.config.api_key is None:
+                raise ValueError("api_key is required")
+
+            if not authorization.config.header:
+                authorization.config.header = "Authorization"
+
+            if authorization.config.type == "bearer":
+                headers[authorization.config.header] = f"Bearer {authorization.config.api_key}"
+            elif authorization.config.type == "basic":
+                headers[authorization.config.header] = f"Basic {authorization.config.api_key}"
+            elif authorization.config.type == "custom":
+                headers[authorization.config.header] = authorization.config.api_key
+
+        return headers
+
+    @staticmethod
+    def get_external_knowledge_api_settings(settings: dict) -> ExternalKnowledgeApiSetting:
+        return ExternalKnowledgeApiSetting.parse_obj(settings)
+
+    @staticmethod
+    def create_external_dataset(tenant_id: str, user_id: str, args: dict) -> Dataset:
+        # check if dataset name already exists
+        if Dataset.query.filter_by(name=args.get("name"), tenant_id=tenant_id).first():
+            raise DatasetNameDuplicateError(f"Dataset with name {args.get('name')} already exists.")
+        external_knowledge_api = ExternalKnowledgeApis.query.filter_by(
+            id=args.get("external_knowledge_api_id"), tenant_id=tenant_id
+        ).first()
+
+        if external_knowledge_api is None:
+            raise ValueError("api template not found")
+
+        dataset = Dataset(
+            tenant_id=tenant_id,
+            name=args.get("name"),
+            description=args.get("description", ""),
+            provider="external",
+            retrieval_model=args.get("external_retrieval_model"),
+            created_by=user_id,
+        )
+
+        db.session.add(dataset)
+        db.session.flush()
+
+        external_knowledge_binding = ExternalKnowledgeBindings(
+            tenant_id=tenant_id,
+            dataset_id=dataset.id,
+            external_knowledge_api_id=args.get("external_knowledge_api_id"),
+            external_knowledge_id=args.get("external_knowledge_id"),
+            created_by=user_id,
+        )
+        db.session.add(external_knowledge_binding)
+
+        db.session.commit()
+
+        return dataset
+
+    @staticmethod
+    def fetch_external_knowledge_retrieval(
+        tenant_id: str, dataset_id: str, query: str, external_retrieval_parameters: dict
+    ) -> list:
+        external_knowledge_binding = ExternalKnowledgeBindings.query.filter_by(
+            dataset_id=dataset_id, tenant_id=tenant_id
+        ).first()
+        if not external_knowledge_binding:
+            raise ValueError("external knowledge binding not found")
+
+        external_knowledge_api = ExternalKnowledgeApis.query.filter_by(
+            id=external_knowledge_binding.external_knowledge_api_id
+        ).first()
+        if not external_knowledge_api:
+            raise ValueError("external api template not found")
+
+        settings = json.loads(external_knowledge_api.settings)
+        headers = {"Content-Type": "application/json"}
+        if settings.get("api_key"):
+            headers["Authorization"] = f"Bearer {settings.get('api_key')}"
+        score_threshold_enabled = external_retrieval_parameters.get("score_threshold_enabled") or False
+        score_threshold = external_retrieval_parameters.get("score_threshold", 0.0) if score_threshold_enabled else 0.0
+        request_params = {
+            "retrieval_setting": {
+                "top_k": external_retrieval_parameters.get("top_k"),
+                "score_threshold": score_threshold,
+            },
+            "query": query,
+            "knowledge_id": external_knowledge_binding.external_knowledge_id,
+        }
+
+        external_knowledge_api_setting = {
+            "url": f"{settings.get('endpoint')}/retrieval",
+            "request_method": "post",
+            "headers": headers,
+            "params": request_params,
+        }
+        response = ExternalDatasetService.process_external_api(
+            ExternalKnowledgeApiSetting(**external_knowledge_api_setting), None
+        )
+        if response.status_code == 200:
+            return response.json().get("records", [])
+        return []
diff --git a/api/services/hit_testing_service.py b/api/services/hit_testing_service.py
index 3dafafd5b46f6..7957b4dc82dfd 100644
--- a/api/services/hit_testing_service.py
+++ b/api/services/hit_testing_service.py
@@ -19,7 +19,15 @@
 
 class HitTestingService:
     @classmethod
-    def retrieve(cls, dataset: Dataset, query: str, account: Account, retrieval_model: dict, limit: int = 10) -> dict:
+    def retrieve(
+        cls,
+        dataset: Dataset,
+        query: str,
+        account: Account,
+        retrieval_model: dict,
+        external_retrieval_model: dict,
+        limit: int = 10,
+    ) -> dict:
         if dataset.available_document_count == 0 or dataset.available_segment_count == 0:
             return {
                 "query": {
@@ -62,10 +70,44 @@ def retrieve(cls, dataset: Dataset, query: str, account: Account, retrieval_mode
 
         return cls.compact_retrieve_response(dataset, query, all_documents)
 
+    @classmethod
+    def external_retrieve(
+        cls,
+        dataset: Dataset,
+        query: str,
+        account: Account,
+        external_retrieval_model: dict,
+    ) -> dict:
+        if dataset.provider != "external":
+            return {
+                "query": {"content": query},
+                "records": [],
+            }
+
+        start = time.perf_counter()
+
+        all_documents = RetrievalService.external_retrieve(
+            dataset_id=dataset.id,
+            query=cls.escape_query_for_search(query),
+            external_retrieval_model=external_retrieval_model,
+        )
+
+        end = time.perf_counter()
+        logging.debug(f"External knowledge hit testing retrieve in {end - start:0.4f} seconds")
+
+        dataset_query = DatasetQuery(
+            dataset_id=dataset.id, content=query, source="hit_testing", created_by_role="account", created_by=account.id
+        )
+
+        db.session.add(dataset_query)
+        db.session.commit()
+
+        return cls.compact_external_retrieve_response(dataset, query, all_documents)
+
     @classmethod
     def compact_retrieve_response(cls, dataset: Dataset, query: str, documents: list[Document]):
-        i = 0
         records = []
+
         for document in documents:
             index_node_id = document.metadata["doc_id"]
 
@@ -81,7 +123,6 @@ def compact_retrieve_response(cls, dataset: Dataset, query: str, documents: list
             )
 
             if not segment:
-                i += 1
                 continue
 
             record = {
@@ -91,8 +132,6 @@ def compact_retrieve_response(cls, dataset: Dataset, query: str, documents: list
 
             records.append(record)
 
-            i += 1
-
         return {
             "query": {
                 "content": query,
@@ -100,6 +139,25 @@ def compact_retrieve_response(cls, dataset: Dataset, query: str, documents: list
             "records": records,
         }
 
+    @classmethod
+    def compact_external_retrieve_response(cls, dataset: Dataset, query: str, documents: list):
+        records = []
+        if dataset.provider == "external":
+            for document in documents:
+                record = {
+                    "content": document.get("content", None),
+                    "title": document.get("title", None),
+                    "score": document.get("score", None),
+                    "metadata": document.get("metadata", None),
+                }
+                records.append(record)
+            return {
+                "query": {
+                    "content": query,
+                },
+                "records": records,
+            }
+
     @classmethod
     def hit_testing_args_check(cls, args):
         query = args["query"]
diff --git a/api/tasks/external_document_indexing_task.py b/api/tasks/external_document_indexing_task.py
new file mode 100644
index 0000000000000..6fc719ae8d085
--- /dev/null
+++ b/api/tasks/external_document_indexing_task.py
@@ -0,0 +1,93 @@
+import json
+import logging
+import time
+
+import click
+from celery import shared_task
+
+from core.indexing_runner import DocumentIsPausedException
+from extensions.ext_database import db
+from extensions.ext_storage import storage
+from models.dataset import Dataset, ExternalKnowledgeApis
+from models.model import UploadFile
+from services.external_knowledge_service import ExternalDatasetService
+
+
+@shared_task(queue="dataset")
+def external_document_indexing_task(
+    dataset_id: str, external_knowledge_api_id: str, data_source: dict, process_parameter: dict
+):
+    """
+    Async process document
+    :param dataset_id:
+    :param external_knowledge_api_id:
+    :param data_source:
+    :param process_parameter:
+    Usage: external_document_indexing_task.delay(dataset_id, document_id)
+    """
+    start_at = time.perf_counter()
+
+    dataset = db.session.query(Dataset).filter(Dataset.id == dataset_id).first()
+    if not dataset:
+        logging.info(
+            click.style("Processed external dataset: {} failed, dataset not exit.".format(dataset_id), fg="red")
+        )
+        return
+
+    # get external api template
+    external_knowledge_api = (
+        db.session.query(ExternalKnowledgeApis)
+        .filter(
+            ExternalKnowledgeApis.id == external_knowledge_api_id, ExternalKnowledgeApis.tenant_id == dataset.tenant_id
+        )
+        .first()
+    )
+
+    if not external_knowledge_api:
+        logging.info(
+            click.style(
+                "Processed external dataset: {} failed, api template: {} not exit.".format(
+                    dataset_id, external_knowledge_api_id
+                ),
+                fg="red",
+            )
+        )
+        return
+    files = {}
+    if data_source["type"] == "upload_file":
+        upload_file_list = data_source["info_list"]["file_info_list"]["file_ids"]
+        for file_id in upload_file_list:
+            file = (
+                db.session.query(UploadFile)
+                .filter(UploadFile.tenant_id == dataset.tenant_id, UploadFile.id == file_id)
+                .first()
+            )
+            if file:
+                files[file.id] = (file.name, storage.load_once(file.key), file.mime_type)
+    try:
+        settings = ExternalDatasetService.get_external_knowledge_api_settings(
+            json.loads(external_knowledge_api.settings)
+        )
+        # assemble headers
+        headers = ExternalDatasetService.assembling_headers(settings.authorization, settings.headers)
+
+        # do http request
+        response = ExternalDatasetService.process_external_api(settings, headers, process_parameter, files)
+        job_id = response.json().get("job_id")
+        if job_id:
+            # save job_id to dataset
+            dataset.job_id = job_id
+            db.session.commit()
+
+        end_at = time.perf_counter()
+        logging.info(
+            click.style(
+                "Processed external dataset: {} successful, latency: {}".format(dataset.id, end_at - start_at),
+                fg="green",
+            )
+        )
+    except DocumentIsPausedException as ex:
+        logging.info(click.style(str(ex), fg="yellow"))
+
+    except Exception:
+        pass
diff --git a/web/app/(commonLayout)/datasets/(datasetDetailLayout)/[datasetId]/layout.tsx b/web/app/(commonLayout)/datasets/(datasetDetailLayout)/[datasetId]/layout.tsx
index e691cc05f6326..a58027bcd12e0 100644
--- a/web/app/(commonLayout)/datasets/(datasetDetailLayout)/[datasetId]/layout.tsx
+++ b/web/app/(commonLayout)/datasets/(datasetDetailLayout)/[datasetId]/layout.tsx
@@ -1,6 +1,6 @@
 'use client'
 import type { FC, SVGProps } from 'react'
-import React, { useEffect } from 'react'
+import React, { useEffect, useMemo } from 'react'
 import { usePathname } from 'next/navigation'
 import useSWR from 'swr'
 import { useTranslation } from 'react-i18next'
@@ -203,12 +203,23 @@ const DatasetDetailLayout: FC<IAppDetailLayoutProps> = (props) => {
     datasetId,
   }, apiParams => fetchDatasetRelatedApps(apiParams.datasetId))
 
-  const navigation = [
-    { name: t('common.datasetMenus.documents'), href: `/datasets/${datasetId}/documents`, icon: DocumentTextIcon, selectedIcon: DocumentTextSolidIcon },
-    { name: t('common.datasetMenus.hitTesting'), href: `/datasets/${datasetId}/hitTesting`, icon: TargetIcon, selectedIcon: TargetSolidIcon },
-    // { name: 'api & webhook', href: `/datasets/${datasetId}/api`, icon: CommandLineIcon, selectedIcon: CommandLineSolidIcon },
-    { name: t('common.datasetMenus.settings'), href: `/datasets/${datasetId}/settings`, icon: Cog8ToothIcon, selectedIcon: Cog8ToothSolidIcon },
-  ]
+  const navigation = useMemo(() => {
+    const baseNavigation = [
+      { name: t('common.datasetMenus.hitTesting'), href: `/datasets/${datasetId}/hitTesting`, icon: TargetIcon, selectedIcon: TargetSolidIcon },
+      // { name: 'api & webhook', href: `/datasets/${datasetId}/api`, icon: CommandLineIcon, selectedIcon: CommandLineSolidIcon },
+      { name: t('common.datasetMenus.settings'), href: `/datasets/${datasetId}/settings`, icon: Cog8ToothIcon, selectedIcon: Cog8ToothSolidIcon },
+    ]
+
+    if (datasetRes?.provider !== 'external') {
+      baseNavigation.unshift({
+        name: t('common.datasetMenus.documents'),
+        href: `/datasets/${datasetId}/documents`,
+        icon: DocumentTextIcon,
+        selectedIcon: DocumentTextSolidIcon,
+      })
+    }
+    return baseNavigation
+  }, [datasetRes?.provider, datasetId, t])
 
   useEffect(() => {
     if (datasetRes)
@@ -233,6 +244,7 @@ const DatasetDetailLayout: FC<IAppDetailLayoutProps> = (props) => {
         icon={datasetRes?.icon || 'https://static.dify.ai/images/dataset-default-icon.png'}
         icon_background={datasetRes?.icon_background || '#F5F5F5'}
         desc={datasetRes?.description || '--'}
+        isExternal={datasetRes?.provider === 'external'}
         navigation={navigation}
         extraInfo={!isCurrentWorkspaceDatasetOperator ? mode => <ExtraInfo isMobile={mode === 'collapse'} relatedApps={relatedApps} /> : undefined}
         iconType={datasetRes?.data_source_type === DataSourceType.NOTION ? 'notion' : 'dataset'}
diff --git a/web/app/(commonLayout)/datasets/Container.tsx b/web/app/(commonLayout)/datasets/Container.tsx
index f532ca416fc24..e350a85354373 100644
--- a/web/app/(commonLayout)/datasets/Container.tsx
+++ b/web/app/(commonLayout)/datasets/Container.tsx
@@ -8,6 +8,7 @@ import { useDebounceFn } from 'ahooks'
 import useSWR from 'swr'
 
 // Components
+import ExternalAPIPanel from '../../components/datasets/external-api/external-api-panel'
 import Datasets from './Datasets'
 import DatasetFooter from './DatasetFooter'
 import ApiServer from './ApiServer'
@@ -16,6 +17,8 @@ import TabSliderNew from '@/app/components/base/tab-slider-new'
 import SearchInput from '@/app/components/base/search-input'
 import TagManagementModal from '@/app/components/base/tag-management'
 import TagFilter from '@/app/components/base/tag-management/filter'
+import Button from '@/app/components/base/button'
+import { ApiConnectionMod } from '@/app/components/base/icons/src/vender/solid/development'
 
 // Services
 import { fetchDatasetApiBaseUrl } from '@/service/datasets'
@@ -24,12 +27,14 @@ import { fetchDatasetApiBaseUrl } from '@/service/datasets'
 import { useTabSearchParams } from '@/hooks/use-tab-searchparams'
 import { useStore as useTagStore } from '@/app/components/base/tag-management/store'
 import { useAppContext } from '@/context/app-context'
+import { useExternalApiPanel } from '@/context/external-api-panel-context'
 
 const Container = () => {
   const { t } = useTranslation()
   const router = useRouter()
   const { currentWorkspace } = useAppContext()
   const showTagManagementModal = useTagStore(s => s.showTagManagementModal)
+  const { showExternalApiPanel, setShowExternalApiPanel } = useExternalApiPanel()
 
   const options = useMemo(() => {
     return [
@@ -66,7 +71,7 @@ const Container = () => {
   useEffect(() => {
     if (currentWorkspace.role === 'normal')
       return router.replace('/apps')
-  }, [currentWorkspace])
+  }, [currentWorkspace, router])
 
   return (
     <div ref={containerRef} className='grow relative flex flex-col bg-gray-100 overflow-y-auto'>
@@ -80,11 +85,18 @@ const Container = () => {
           <div className='flex items-center gap-2'>
             <TagFilter type='knowledge' value={tagFilterValue} onChange={handleTagsChange} />
             <SearchInput className='w-[200px]' value={keywords} onChange={handleKeywordsChange} />
+            <div className="w-[1px] h-4 bg-divider-regular" />
+            <Button
+              className='gap-0.5 shadows-shadow-xs'
+              onClick={() => setShowExternalApiPanel(true)}
+            >
+              <ApiConnectionMod className='w-4 h-4 text-components-button-secondary-text' />
+              <div className='flex px-0.5 justify-center items-center gap-1 text-components-button-secondary-text system-sm-medium'>{t('dataset.externalAPIPanelTitle')}</div>
+            </Button>
           </div>
         )}
         {activeTab === 'api' && data && <ApiServer apiBaseUrl={data.api_base_url || ''} />}
       </div>
-
       {activeTab === 'dataset' && (
         <>
           <Datasets containerRef={containerRef} tags={tagIDs} keywords={searchKeywords} />
@@ -94,10 +106,10 @@ const Container = () => {
           )}
         </>
       )}
-
       {activeTab === 'api' && data && <Doc apiBaseUrl={data.api_base_url || ''} />}
-    </div>
 
+      {showExternalApiPanel && <ExternalAPIPanel onClose={() => setShowExternalApiPanel(false)} />}
+    </div>
   )
 }
 
diff --git a/web/app/(commonLayout)/datasets/DatasetCard.tsx b/web/app/(commonLayout)/datasets/DatasetCard.tsx
index f66c6bccf421f..e8ccddbcb78d3 100644
--- a/web/app/(commonLayout)/datasets/DatasetCard.tsx
+++ b/web/app/(commonLayout)/datasets/DatasetCard.tsx
@@ -18,6 +18,7 @@ import Divider from '@/app/components/base/divider'
 import RenameDatasetModal from '@/app/components/datasets/rename-modal'
 import type { Tag } from '@/app/components/base/tag-management/constant'
 import TagSelector from '@/app/components/base/tag-management/selector'
+import CornerLabel from '@/app/components/base/corner-label'
 import { useAppContext } from '@/context/app-context'
 
 export type DatasetCardProps = {
@@ -32,6 +33,7 @@ const DatasetCard = ({
   const { t } = useTranslation()
   const { notify } = useContext(ToastContext)
   const { push } = useRouter()
+  const EXTERNAL_PROVIDER = 'external' as const
 
   const { isCurrentWorkspaceDatasetOperator } = useAppContext()
   const [tags, setTags] = useState<Tag[]>(dataset.tags)
@@ -39,6 +41,7 @@ const DatasetCard = ({
   const [showRenameModal, setShowRenameModal] = useState(false)
   const [showConfirmDelete, setShowConfirmDelete] = useState(false)
   const [confirmMessage, setConfirmMessage] = useState<string>('')
+  const isExternalProvider = (provider: string): boolean => provider === EXTERNAL_PROVIDER
   const detectIsUsedByApp = useCallback(async () => {
     try {
       const { is_using: isUsedByApp } = await checkIsUsedInApp(dataset.id)
@@ -108,13 +111,16 @@ const DatasetCard = ({
   return (
     <>
       <div
-        className='group col-span-1 bg-white border-2 border-solid border-transparent rounded-xl shadow-sm min-h-[160px] flex flex-col transition-all duration-200 ease-in-out cursor-pointer hover:shadow-lg'
+        className='group relative col-span-1 bg-white border-[0.5px] border-solid border-transparent rounded-xl shadow-sm min-h-[160px] flex flex-col transition-all duration-200 ease-in-out cursor-pointer hover:shadow-lg'
         data-disable-nprogress={true}
         onClick={(e) => {
           e.preventDefault()
-          push(`/datasets/${dataset.id}/documents`)
+          isExternalProvider(dataset.provider)
+            ? push(`/datasets/${dataset.id}/hitTesting`)
+            : push(`/datasets/${dataset.id}/documents`)
         }}
       >
+        {isExternalProvider(dataset.provider) && <CornerLabel label='External' className='absolute right-0' labelClassName='rounded-tr-xl' />}
         <div className='flex pt-[14px] px-[14px] pb-3 h-[66px] items-center gap-3 grow-0 shrink-0'>
           <div className={cn(
             'shrink-0 flex items-center justify-center p-2.5 bg-[#F5F8FF] rounded-md border-[0.5px] border-[#E0EAFF]',
@@ -136,13 +142,20 @@ const DatasetCard = ({
             <div className='flex items-center mt-[1px] text-xs leading-[18px] text-gray-500'>
               <div
                 className={cn('truncate', (!dataset.embedding_available || !dataset.document_count) && 'opacity-50')}
-                title={`${dataset.document_count}${t('dataset.documentCount')} · ${Math.round(dataset.word_count / 1000)}${t('dataset.wordCount')} · ${dataset.app_count}${t('dataset.appCount')}`}
+                title={dataset.provider === 'external' ? `${dataset.app_count}${t('dataset.appCount')}` : `${dataset.document_count}${t('dataset.documentCount')} · ${Math.round(dataset.word_count / 1000)}${t('dataset.wordCount')} · ${dataset.app_count}${t('dataset.appCount')}`}
               >
-                <span>{dataset.document_count}{t('dataset.documentCount')}</span>
-                <span className='shrink-0 mx-0.5 w-1 text-gray-400'>·</span>
-                <span>{Math.round(dataset.word_count / 1000)}{t('dataset.wordCount')}</span>
-                <span className='shrink-0 mx-0.5 w-1 text-gray-400'>·</span>
-                <span>{dataset.app_count}{t('dataset.appCount')}</span>
+                {dataset.provider === 'external'
+                  ? <>
+                    <span>{dataset.app_count}{t('dataset.appCount')}</span>
+                  </>
+                  : <>
+                    <span>{dataset.document_count}{t('dataset.documentCount')}</span>
+                    <span className='shrink-0 mx-0.5 w-1 text-gray-400'>·</span>
+                    <span>{Math.round(dataset.word_count / 1000)}{t('dataset.wordCount')}</span>
+                    <span className='shrink-0 mx-0.5 w-1 text-gray-400'>·</span>
+                    <span>{dataset.app_count}{t('dataset.appCount')}</span>
+                  </>
+                }
               </div>
             </div>
           </div>
diff --git a/web/app/(commonLayout)/datasets/NewDatasetCard.tsx b/web/app/(commonLayout)/datasets/NewDatasetCard.tsx
index f76efa57696d3..5dd244ad41006 100644
--- a/web/app/(commonLayout)/datasets/NewDatasetCard.tsx
+++ b/web/app/(commonLayout)/datasets/NewDatasetCard.tsx
@@ -4,21 +4,32 @@ import { forwardRef } from 'react'
 import { useTranslation } from 'react-i18next'
 import {
   RiAddLine,
+  RiArrowRightLine,
 } from '@remixicon/react'
 
 const CreateAppCard = forwardRef<HTMLAnchorElement>((_, ref) => {
   const { t } = useTranslation()
 
   return (
-    <a ref={ref} className='group flex flex-col col-span-1 bg-gray-200 border-[0.5px] border-black/5 rounded-xl min-h-[160px] transition-all duration-200 ease-in-out cursor-pointer hover:bg-white hover:shadow-lg' href='/datasets/create'>
-      <div className='shrink-0 flex items-center p-4 pb-3'>
-        <div className='w-10 h-10 flex items-center justify-center border border-gray-200 bg-gray-100 rounded-lg'>
-          <RiAddLine className='w-4 h-4 text-gray-500'/>
+    <div className='flex flex-col bg-background-default-dimm border-[0.5px] border-components-panel-border rounded-xl
+      min-h-[160px] transition-all duration-200 ease-in-out'
+    >
+      <a ref={ref} className='group flex flex-grow items-start p-4 cursor-pointer' href='/datasets/create'>
+        <div className='flex items-center gap-3'>
+          <div className='w-10 h-10 p-2 flex items-center justify-center border border-dashed border-divider-regular rounded-lg
+            bg-background-default-lighter group-hover:border-solid group-hover:border-effects-highlight group-hover:bg-background-default-dodge'
+          >
+            <RiAddLine className='w-4 h-4 text-text-tertiary group-hover:text-text-accent'/>
+          </div>
+          <div className='system-md-semibold text-text-secondary group-hover:text-text-accent'>{t('dataset.createDataset')}</div>
         </div>
-        <div className='ml-3 text-sm font-semibold leading-5 text-gray-800 group-hover:text-primary-600'>{t('dataset.createDataset')}</div>
-      </div>
-      <div className='mb-1 px-4 text-xs leading-normal text-gray-500 line-clamp-4'>{t('dataset.createDatasetIntro')}</div>
-    </a>
+      </a>
+      <div className='p-4 pt-0 text-text-tertiary system-xs-regular'>{t('dataset.createDatasetIntro')}</div>
+      <a className='group flex p-4 items-center gap-1 border-t-[0.5px] border-divider-subtle rounded-b-xl cursor-pointer' href='/datasets/connect'>
+        <div className='system-xs-medium text-text-tertiary group-hover:text-text-accent'>{t('dataset.connectDataset')}</div>
+        <RiArrowRightLine className='w-3.5 h-3.5 text-text-tertiary group-hover:text-text-accent' />
+      </a>
+    </div>
   )
 })
 
diff --git a/web/app/(commonLayout)/datasets/connect/page.tsx b/web/app/(commonLayout)/datasets/connect/page.tsx
new file mode 100644
index 0000000000000..724c506a7f4eb
--- /dev/null
+++ b/web/app/(commonLayout)/datasets/connect/page.tsx
@@ -0,0 +1,8 @@
+import React from 'react'
+import ExternalKnowledgeBaseConnector from '@/app/components/datasets/external-knowledge-base/connector'
+
+const ExternalKnowledgeBaseCreation = () => {
+  return <ExternalKnowledgeBaseConnector />
+}
+
+export default ExternalKnowledgeBaseCreation
diff --git a/web/app/(commonLayout)/datasets/layout.tsx b/web/app/(commonLayout)/datasets/layout.tsx
new file mode 100644
index 0000000000000..aecb537aa6d63
--- /dev/null
+++ b/web/app/(commonLayout)/datasets/layout.tsx
@@ -0,0 +1,14 @@
+'use client'
+
+import { ExternalApiPanelProvider } from '@/context/external-api-panel-context'
+import { ExternalKnowledgeApiProvider } from '@/context/external-knowledge-api-context'
+
+export default function DatasetsLayout({ children }: { children: React.ReactNode }) {
+  return (
+    <ExternalKnowledgeApiProvider>
+      <ExternalApiPanelProvider>
+        {children}
+      </ExternalApiPanelProvider>
+    </ExternalKnowledgeApiProvider>
+  )
+}
diff --git a/web/app/(commonLayout)/datasets/page.tsx b/web/app/(commonLayout)/datasets/page.tsx
index 5aa11aa275070..096a1b8979210 100644
--- a/web/app/(commonLayout)/datasets/page.tsx
+++ b/web/app/(commonLayout)/datasets/page.tsx
@@ -1,9 +1,7 @@
 import Container from './Container'
 
 const AppList = async () => {
-  return (
-    <Container />
-  )
+  return <Container />
 }
 
 export const metadata = {
diff --git a/web/app/(commonLayout)/datasets/store.ts b/web/app/(commonLayout)/datasets/store.ts
new file mode 100644
index 0000000000000..40b7b15594809
--- /dev/null
+++ b/web/app/(commonLayout)/datasets/store.ts
@@ -0,0 +1,11 @@
+import { create } from 'zustand'
+
+type DatasetStore = {
+  showExternalApiPanel: boolean
+  setShowExternalApiPanel: (show: boolean) => void
+}
+
+export const useDatasetStore = create<DatasetStore>(set => ({
+  showExternalApiPanel: false,
+  setShowExternalApiPanel: show => set({ showExternalApiPanel: show }),
+}))
diff --git a/web/app/components/app-sidebar/basic.tsx b/web/app/components/app-sidebar/basic.tsx
index c939cb7bb3a5a..51fc10721eb7a 100644
--- a/web/app/components/app-sidebar/basic.tsx
+++ b/web/app/components/app-sidebar/basic.tsx
@@ -1,4 +1,5 @@
 import React from 'react'
+import { useTranslation } from 'react-i18next'
 import AppIcon from '../base/app-icon'
 import Tooltip from '@/app/components/base/tooltip'
 
@@ -6,6 +7,7 @@ export type IAppBasicProps = {
   iconType?: 'app' | 'api' | 'dataset' | 'webapp' | 'notion'
   icon?: string
   icon_background?: string | null
+  isExternal?: boolean
   name: string
   type: string | React.ReactNode
   hoverTip?: string
@@ -52,7 +54,9 @@ const ICON_MAP = {
   notion: <AppIcon innerIcon={NotionSvg} className='!border-[0.5px] !border-indigo-100 !bg-white' />,
 }
 
-export default function AppBasic({ icon, icon_background, name, type, hoverTip, textStyle, mode = 'expand', iconType = 'app' }: IAppBasicProps) {
+export default function AppBasic({ icon, icon_background, name, isExternal, type, hoverTip, textStyle, mode = 'expand', iconType = 'app' }: IAppBasicProps) {
+  const { t } = useTranslation()
+
   return (
     <div className="flex items-start p-1">
       {icon && icon_background && iconType === 'app' && (
@@ -83,6 +87,7 @@ export default function AppBasic({ icon, icon_background, name, type, hoverTip,
           }
         </div>
         <div className={`text-xs font-normal text-gray-500 group-hover:text-gray-700 break-all ${textStyle?.extra ?? ''}`}>{type}</div>
+        <div className='text-text-tertiary system-2xs-medium-uppercase'>{isExternal ? t('dataset.externalTag') : ''}</div>
       </div>}
     </div>
   )
diff --git a/web/app/components/app-sidebar/index.tsx b/web/app/components/app-sidebar/index.tsx
index 5d5d407dc0f90..5ee063ad646f3 100644
--- a/web/app/components/app-sidebar/index.tsx
+++ b/web/app/components/app-sidebar/index.tsx
@@ -15,6 +15,7 @@ export type IAppDetailNavProps = {
   iconType?: 'app' | 'dataset' | 'notion'
   title: string
   desc: string
+  isExternal?: boolean
   icon: string
   icon_background: string
   navigation: Array<{
@@ -26,7 +27,7 @@ export type IAppDetailNavProps = {
   extraInfo?: (modeState: string) => React.ReactNode
 }
 
-const AppDetailNav = ({ title, desc, icon, icon_background, navigation, extraInfo, iconType = 'app' }: IAppDetailNavProps) => {
+const AppDetailNav = ({ title, desc, isExternal, icon, icon_background, navigation, extraInfo, iconType = 'app' }: IAppDetailNavProps) => {
   const { appSidebarExpand, setAppSiderbarExpand } = useAppStore(useShallow(state => ({
     appSidebarExpand: state.appSidebarExpand,
     setAppSiderbarExpand: state.setAppSiderbarExpand,
@@ -70,6 +71,7 @@ const AppDetailNav = ({ title, desc, icon, icon_background, navigation, extraInf
             icon_background={icon_background}
             name={title}
             type={desc}
+            isExternal={isExternal}
           />
         )}
       </div>
diff --git a/web/app/components/app/configuration/dataset-config/card-item/item.tsx b/web/app/components/app/configuration/dataset-config/card-item/item.tsx
index 184a598e1081c..9c5e6fa785ee4 100644
--- a/web/app/components/app/configuration/dataset-config/card-item/item.tsx
+++ b/web/app/components/app/configuration/dataset-config/card-item/item.tsx
@@ -5,6 +5,7 @@ import {
   RiDeleteBinLine,
   RiEditLine,
 } from '@remixicon/react'
+import { useTranslation } from 'react-i18next'
 import SettingsModal from '../settings-modal'
 import type { DataSet } from '@/models/datasets'
 import { DataSourceType } from '@/models/datasets'
@@ -33,6 +34,7 @@ const Item: FC<ItemProps> = ({
   const isMobile = media === MediaType.mobile
   const [showSettingsModal, setShowSettingsModal] = useState(false)
   const { formatIndexingTechniqueAndMethod } = useKnowledge()
+  const { t } = useTranslation()
 
   const handleSave = (newDataset: DataSet) => {
     onSave(newDataset)
@@ -65,9 +67,11 @@ const Item: FC<ItemProps> = ({
       <div className='grow'>
         <div className='flex items-center h-[18px]'>
           <div className='grow text-[13px] font-medium text-gray-800 truncate' title={config.name}>{config.name}</div>
-          <Badge
-            text={formatIndexingTechniqueAndMethod(config.indexing_technique, config.retrieval_model_dict?.search_method)}
-          />
+          {config.provider === 'external'
+            ? <Badge text={t('dataset.externalTag')}></Badge>
+            : <Badge
+              text={formatIndexingTechniqueAndMethod(config.indexing_technique, config.retrieval_model_dict?.search_method)}
+            />}
         </div>
       </div>
       <div className='hidden rounded-lg group-hover:flex items-center justify-end absolute right-0 top-0 bottom-0 pr-2 w-[124px] bg-gradient-to-r from-white/50 to-white to-50%'>
diff --git a/web/app/components/app/configuration/dataset-config/params-config/config-content.tsx b/web/app/components/app/configuration/dataset-config/params-config/config-content.tsx
index 91cae54bb8e7e..f99496b14fa8f 100644
--- a/web/app/components/app/configuration/dataset-config/params-config/config-content.tsx
+++ b/web/app/components/app/configuration/dataset-config/params-config/config-content.tsx
@@ -174,6 +174,20 @@ const ConfigContent: FC<Props> = ({
               </div>
             )
           }
+          {
+            selectedDatasetsMode.mixtureInternalAndExternal && (
+              <div className='mt-4 system-xs-medium text-text-warning'>
+                {t('dataset.mixtureInternalAndExternalTip')}
+              </div>
+            )
+          }
+          {
+            selectedDatasetsMode.allExternal && (
+              <div className='mt-4 system-xs-medium text-text-warning'>
+                {t('dataset.allExternalTip')}
+              </div>
+            )
+          }
           {
             selectedDatasetsMode.mixtureHighQualityAndEconomic
             && (
@@ -229,15 +243,15 @@ const ConfigContent: FC<Props> = ({
                       />
                     )
                   }
-                  <div className='ml-2 leading-[32px] text-[13px] font-medium text-gray-900'>{t('common.modelProvider.rerankModel.key')}</div>
+                  <div className='leading-[32px] text-text-secondary system-sm-semibold'>{t('common.modelProvider.rerankModel.key')}</div>
                   <Tooltip
                     popupContent={
                       <div className="w-[200px]">
                         {t('common.modelProvider.rerankModel.tip')}
                       </div>
                     }
-                    popupClassName='ml-0.5'
-                    triggerClassName='ml-0.5 w-3.5 h-3.5'
+                    popupClassName='ml-1'
+                    triggerClassName='ml-1 w-4 h-4'
                   />
                 </div>
                 <div>
diff --git a/web/app/components/app/configuration/dataset-config/params-config/index.tsx b/web/app/components/app/configuration/dataset-config/params-config/index.tsx
index 656cbfea6564f..2d3df0b03990a 100644
--- a/web/app/components/app/configuration/dataset-config/params-config/index.tsx
+++ b/web/app/components/app/configuration/dataset-config/params-config/index.tsx
@@ -39,13 +39,26 @@ const ParamsConfig = ({
   useEffect(() => {
     const {
       allEconomic,
+      allHighQuality,
+      allHighQualityFullTextSearch,
+      allHighQualityVectorSearch,
+      allExternal,
+      mixtureHighQualityAndEconomic,
+      inconsistentEmbeddingModel,
+      mixtureInternalAndExternal,
     } = getSelectedDatasetsMode(selectedDatasets)
     const { datasets, retrieval_model, score_threshold_enabled, ...restConfigs } = datasetConfigs
     let rerankEnable = restConfigs.reranking_enable
 
-    if (allEconomic && !restConfigs.reranking_model?.reranking_provider_name && rerankEnable === undefined)
+    if ((allEconomic && !restConfigs.reranking_model?.reranking_provider_name && rerankEnable === undefined) || allExternal)
       rerankEnable = false
 
+    if (allEconomic || allHighQuality || allHighQualityFullTextSearch || allHighQualityVectorSearch || (allExternal && selectedDatasets.length === 1))
+      setRerankSettingModalOpen(false)
+
+    if (mixtureHighQualityAndEconomic || inconsistentEmbeddingModel || mixtureInternalAndExternal || (allExternal && selectedDatasets.length > 1))
+      setRerankSettingModalOpen(true)
+
     setTempDataSetConfigs({
       ...getMultipleRetrievalConfig({
         top_k: restConfigs.top_k,
diff --git a/web/app/components/app/configuration/dataset-config/select-dataset/index.tsx b/web/app/components/app/configuration/dataset-config/select-dataset/index.tsx
index 4493755ba0138..0d94e599b4f2c 100644
--- a/web/app/components/app/configuration/dataset-config/select-dataset/index.tsx
+++ b/web/app/components/app/configuration/dataset-config/select-dataset/index.tsx
@@ -47,7 +47,7 @@ const SelectDataSet: FC<ISelectDataSetProps> = ({
         const { data, has_more } = await fetchDatasets({ url: '/datasets', params: { page } })
         setPage(getPage() + 1)
         setIsNoMore(!has_more)
-        const newList = [...(datasets || []), ...data.filter(item => item.indexing_technique)]
+        const newList = [...(datasets || []), ...data.filter(item => item.indexing_technique || item.provider === 'external')]
         setDataSets(newList)
         setLoaded(true)
         if (!selected.find(item => !item.name))
@@ -145,6 +145,11 @@ const SelectDataSet: FC<ISelectDataSetProps> = ({
                     />
                   )
                 }
+                {
+                  item.provider === 'external' && (
+                    <Badge text={t('dataset.externalTag')} />
+                  )
+                }
               </div>
             ))}
           </div>
diff --git a/web/app/components/app/configuration/dataset-config/settings-modal/index.tsx b/web/app/components/app/configuration/dataset-config/settings-modal/index.tsx
index e538c347d95c9..7b3b4abc0fb1e 100644
--- a/web/app/components/app/configuration/dataset-config/settings-modal/index.tsx
+++ b/web/app/components/app/configuration/dataset-config/settings-modal/index.tsx
@@ -5,8 +5,10 @@ import { useTranslation } from 'react-i18next'
 import { isEqual } from 'lodash-es'
 import { RiCloseLine } from '@remixicon/react'
 import { BookOpenIcon } from '@heroicons/react/24/outline'
+import { ApiConnectionMod } from '@/app/components/base/icons/src/vender/solid/development'
 import cn from '@/utils/classnames'
 import IndexMethodRadio from '@/app/components/datasets/settings/index-method-radio'
+import Divider from '@/app/components/base/divider'
 import Button from '@/app/components/base/button'
 import type { DataSet } from '@/models/datasets'
 import { useToastContext } from '@/app/components/base/toast'
@@ -14,6 +16,7 @@ import { updateDatasetSetting } from '@/service/datasets'
 import { useAppContext } from '@/context/app-context'
 import { useModalContext } from '@/context/modal-context'
 import type { RetrievalConfig } from '@/types/app'
+import RetrievalSettings from '@/app/components/datasets/external-knowledge-base/create/RetrievalSettings'
 import RetrievalMethodConfig from '@/app/components/datasets/common/retrieval-method-config'
 import EconomicalRetrievalMethodConfig from '@/app/components/datasets/common/economical-retrieval-method-config'
 import { ensureRerankModelSelected, isReRankModelSelected } from '@/app/components/datasets/common/check-rerank-model'
@@ -56,6 +59,9 @@ const SettingsModal: FC<SettingsModalProps> = ({
   const { t } = useTranslation()
   const { notify } = useToastContext()
   const ref = useRef(null)
+  const [topK, setTopK] = useState(currentDataset?.external_retrieval_model.top_k ?? 2)
+  const [scoreThreshold, setScoreThreshold] = useState(currentDataset?.external_retrieval_model.score_threshold ?? 0.5)
+  const [scoreThresholdEnabled, setScoreThresholdEnabled] = useState(currentDataset?.external_retrieval_model.score_threshold_enabled ?? false)
 
   const { setShowAccountSettingModal } = useModalContext()
   const [loading, setLoading] = useState(false)
@@ -73,6 +79,15 @@ const SettingsModal: FC<SettingsModalProps> = ({
   const [isHideChangedTip, setIsHideChangedTip] = useState(false)
   const isRetrievalChanged = !isEqual(retrievalConfig, localeCurrentDataset?.retrieval_model_dict) || indexMethod !== localeCurrentDataset?.indexing_technique
 
+  const handleSettingsChange = (data: { top_k?: number; score_threshold?: number; score_threshold_enabled?: boolean }) => {
+    if (data.top_k !== undefined)
+      setTopK(data.top_k)
+    if (data.score_threshold !== undefined)
+      setScoreThreshold(data.score_threshold)
+    if (data.score_threshold_enabled !== undefined)
+      setScoreThresholdEnabled(data.score_threshold_enabled)
+  }
+
   const handleSave = async () => {
     if (loading)
       return
@@ -107,10 +122,17 @@ const SettingsModal: FC<SettingsModalProps> = ({
           description,
           permission,
           indexing_technique: indexMethod,
+          external_retrieval_model: {
+            top_k: topK,
+            score_threshold: scoreThreshold,
+            score_threshold_enabled: scoreThresholdEnabled,
+          },
           retrieval_model: {
             ...postRetrievalConfig,
             score_threshold: postRetrievalConfig.score_threshold_enabled ? postRetrievalConfig.score_threshold : 0,
           },
+          external_knowledge_id: currentDataset!.external_knowledge_info.external_knowledge_id,
+          external_knowledge_api_id: currentDataset!.external_knowledge_info.external_knowledge_api_id,
           embedding_model: localeCurrentDataset.embedding_model,
           embedding_model_provider: localeCurrentDataset.embedding_model_provider,
         },
@@ -178,7 +200,7 @@ const SettingsModal: FC<SettingsModalProps> = ({
       }}>
         <div className={cn(rowClass, 'items-center')}>
           <div className={labelClass}>
-            {t('datasetSettings.form.name')}
+            <div className='text-text-secondary system-sm-semibold'>{t('datasetSettings.form.name')}</div>
           </div>
           <input
             value={localeCurrentDataset.name}
@@ -189,7 +211,7 @@ const SettingsModal: FC<SettingsModalProps> = ({
         </div>
         <div className={cn(rowClass)}>
           <div className={labelClass}>
-            {t('datasetSettings.form.desc')}
+            <div className='text-text-secondary system-sm-semibold'>{t('datasetSettings.form.desc')}</div>
           </div>
           <div className='w-full'>
             <textarea
@@ -206,7 +228,7 @@ const SettingsModal: FC<SettingsModalProps> = ({
         </div>
         <div className={rowClass}>
           <div className={labelClass}>
-            <div>{t('datasetSettings.form.permissions')}</div>
+            <div className='text-text-secondary system-sm-semibold'>{t('datasetSettings.form.permissions')}</div>
           </div>
           <div className='w-full'>
             <PermissionSelector
@@ -219,24 +241,25 @@ const SettingsModal: FC<SettingsModalProps> = ({
             />
           </div>
         </div>
-        <div className="w-full h-0 border-b-[0.5px] border-b-gray-200 my-2"></div>
-        <div className={cn(rowClass)}>
-          <div className={labelClass}>
-            {t('datasetSettings.form.indexMethod')}
-          </div>
-          <div className='grow'>
-            <IndexMethodRadio
-              disable={!localeCurrentDataset?.embedding_available}
-              value={indexMethod}
-              onChange={v => setIndexMethod(v!)}
-              itemClassName='sm:!w-[280px]'
-            />
+        {currentDataset && currentDataset.indexing_technique && (
+          <div className={cn(rowClass)}>
+            <div className={labelClass}>
+              <div className='text-text-secondary system-sm-semibold'>{t('datasetSettings.form.indexMethod')}</div>
+            </div>
+            <div className='grow'>
+              <IndexMethodRadio
+                disable={!localeCurrentDataset?.embedding_available}
+                value={indexMethod}
+                onChange={v => setIndexMethod(v!)}
+                itemClassName='sm:!w-[280px]'
+              />
+            </div>
           </div>
-        </div>
+        )}
         {indexMethod === 'high_quality' && (
           <div className={cn(rowClass)}>
             <div className={labelClass}>
-              {t('datasetSettings.form.embeddingModel')}
+              <div className='text-text-secondary system-sm-semibold'>{t('datasetSettings.form.embeddingModel')}</div>
             </div>
             <div className='w-full'>
               <div className='w-full h-9 rounded-lg bg-gray-100 opacity-60'>
@@ -258,32 +281,75 @@ const SettingsModal: FC<SettingsModalProps> = ({
         )}
 
         {/* Retrieval Method Config */}
-        <div className={rowClass}>
-          <div className={cn(labelClass, 'w-auto min-w-[168px]')}>
-            <div>
-              <div>{t('datasetSettings.form.retrievalSetting.title')}</div>
-              <div className='leading-[18px] text-xs font-normal text-gray-500'>
-                <a target='_blank' rel='noopener noreferrer' href='https://docs.dify.ai/guides/knowledge-base/create-knowledge-and-upload-documents#id-4-retrieval-settings' className='text-[#155eef]'>{t('datasetSettings.form.retrievalSetting.learnMore')}</a>
-                {t('datasetSettings.form.retrievalSetting.description')}
+        {currentDataset?.provider === 'external'
+          ? <>
+            <div className={rowClass}><Divider/></div>
+            <div className={rowClass}>
+              <div className={labelClass}>
+                <div className='text-text-secondary system-sm-semibold'>{t('datasetSettings.form.retrievalSetting.title')}</div>
               </div>
+              <RetrievalSettings
+                topK={topK}
+                scoreThreshold={scoreThreshold}
+                scoreThresholdEnabled={scoreThresholdEnabled}
+                onChange={handleSettingsChange}
+                isInRetrievalSetting={true}
+              />
             </div>
-          </div>
-          <div>
-            {indexMethod === 'high_quality'
-              ? (
-                <RetrievalMethodConfig
-                  value={retrievalConfig}
-                  onChange={setRetrievalConfig}
-                />
-              )
-              : (
-                <EconomicalRetrievalMethodConfig
-                  value={retrievalConfig}
-                  onChange={setRetrievalConfig}
-                />
-              )}
-          </div>
-        </div>
+            <div className={rowClass}><Divider/></div>
+            <div className={rowClass}>
+              <div className={labelClass}>
+                <div className='text-text-secondary system-sm-semibold'>{t('datasetSettings.form.externalKnowledgeAPI')}</div>
+              </div>
+              <div className='w-full max-w-[480px]'>
+                <div className='flex h-full px-3 py-2 items-center gap-1 rounded-lg bg-components-input-bg-normal'>
+                  <ApiConnectionMod className='w-4 h-4 text-text-secondary' />
+                  <div className='overflow-hidden text-text-secondary text-ellipsis system-sm-medium'>
+                    {currentDataset?.external_knowledge_info.external_knowledge_api_name}
+                  </div>
+                  <div className='text-text-tertiary system-xs-regular'>·</div>
+                  <div className='text-text-tertiary system-xs-regular'>{currentDataset?.external_knowledge_info.external_knowledge_api_endpoint}</div>
+                </div>
+              </div>
+            </div>
+            <div className={rowClass}>
+              <div className={labelClass}>
+                <div className='text-text-secondary system-sm-semibold'>{t('datasetSettings.form.externalKnowledgeID')}</div>
+              </div>
+              <div className='w-full max-w-[480px]'>
+                <div className='flex h-full px-3 py-2 items-center gap-1 rounded-lg bg-components-input-bg-normal'>
+                  <div className='text-text-tertiary system-xs-regular'>{currentDataset?.external_knowledge_info.external_knowledge_id}</div>
+                </div>
+              </div>
+            </div>
+            <div className={rowClass}><Divider/></div>
+          </>
+          : <div className={rowClass}>
+            <div className={cn(labelClass, 'w-auto min-w-[168px]')}>
+              <div>
+                <div className='text-text-secondary system-sm-semibold'>{t('datasetSettings.form.retrievalSetting.title')}</div>
+                <div className='leading-[18px] text-xs font-normal text-gray-500'>
+                  <a target='_blank' rel='noopener noreferrer' href='https://docs.dify.ai/guides/knowledge-base/create-knowledge-and-upload-documents#id-4-retrieval-settings' className='text-[#155eef]'>{t('datasetSettings.form.retrievalSetting.learnMore')}</a>
+                  {t('datasetSettings.form.retrievalSetting.description')}
+                </div>
+              </div>
+            </div>
+            <div>
+              {indexMethod === 'high_quality'
+                ? (
+                  <RetrievalMethodConfig
+                    value={retrievalConfig}
+                    onChange={setRetrievalConfig}
+                  />
+                )
+                : (
+                  <EconomicalRetrievalMethodConfig
+                    value={retrievalConfig}
+                    onChange={setRetrievalConfig}
+                  />
+                )}
+            </div>
+          </div>}
       </div>
       {isRetrievalChanged && !isHideChangedTip && (
         <div className='absolute z-10 left-[30px] right-[30px] bottom-[76px] flex h-10 items-center px-3 rounded-lg border border-[#FEF0C7] bg-[#FFFAEB] shadow-lg justify-between'>
diff --git a/web/app/components/base/corner-label/index.tsx b/web/app/components/base/corner-label/index.tsx
new file mode 100644
index 0000000000000..0ad33ce8a650a
--- /dev/null
+++ b/web/app/components/base/corner-label/index.tsx
@@ -0,0 +1,21 @@
+import { Corner } from '../icons/src/vender/solid/shapes'
+import cn from '@/utils/classnames'
+
+type CornerLabelProps = {
+  label: string
+  className?: string
+  labelClassName?: string
+}
+
+const CornerLabel: React.FC<CornerLabelProps> = ({ label, className, labelClassName }) => {
+  return (
+    <div className={cn('group/corner-label inline-flex items-start', className)}>
+      <Corner className='w-[13px] h-5 text-background-section group-hover/corner-label:text-background-section-burn' />
+      <div className={cn('flex py-1 pr-2 items-center gap-0.5 bg-background-section group-hover/corner-label:bg-background-section-burn', labelClassName)}>
+        <div className='text-text-tertiary system-2xs-medium-uppercase'>{label}</div>
+      </div>
+    </div>
+  )
+}
+
+export default CornerLabel
diff --git a/web/app/components/base/icons/assets/vender/solid/development/api-connection-mod.svg b/web/app/components/base/icons/assets/vender/solid/development/api-connection-mod.svg
new file mode 100644
index 0000000000000..9e4b0c81ec489
--- /dev/null
+++ b/web/app/components/base/icons/assets/vender/solid/development/api-connection-mod.svg
@@ -0,0 +1,5 @@
+<svg width="16" height="16" viewBox="0 0 16 16" fill="none" xmlns="http://www.w3.org/2000/svg">
+<g id="Icon L">
+<path id="Vector" fill-rule="evenodd" clip-rule="evenodd" d="M7.99996 3.33333C5.42263 3.33333 3.33329 5.42267 3.33329 8C3.33329 10.5773 5.42263 12.6667 7.99996 12.6667C9.72643 12.6667 11.2348 11.7295 12.0427 10.3329C12.227 10.0141 12.6349 9.90523 12.9536 10.0896C13.2723 10.274 13.3812 10.6818 13.1968 11.0005C12.1604 12.7921 10.2216 14 7.99996 14C4.91159 14 2.36821 11.6666 2.03658 8.66667H1.33329C0.965103 8.66667 0.666626 8.36819 0.666626 8C0.666626 7.63181 0.965103 7.33333 1.33329 7.33333H2.03658C2.36821 4.33337 4.91159 2 7.99996 2C10.2216 2 12.1604 3.20785 13.1968 4.99952C13.3812 5.31823 13.2723 5.72605 12.9536 5.91041C12.6349 6.09477 12.227 5.98585 12.0427 5.66714C11.2348 4.27054 9.72643 3.33333 7.99996 3.33333ZM7.99996 6C6.89539 6 5.99996 6.89543 5.99996 8C5.99996 9.10455 6.89539 10 7.99996 10C9.1045 10 9.99996 9.10454 9.99996 8C9.99996 6.89543 9.10451 6 7.99996 6ZM4.66663 8C4.66663 6.15905 6.15901 4.66667 7.99996 4.66667C9.61257 4.66667 10.9578 5.81184 11.2666 7.33333H14.6666C15.0348 7.33333 15.3333 7.63181 15.3333 8C15.3333 8.36819 15.0348 8.66667 14.6666 8.66667H11.2666C10.9578 10.1881 9.61257 11.3333 7.99996 11.3333C6.159 11.3333 4.66663 9.84092 4.66663 8Z" fill="#354052"/>
+</g>
+</svg>
diff --git a/web/app/components/base/icons/assets/vender/solid/shapes/corner.svg b/web/app/components/base/icons/assets/vender/solid/shapes/corner.svg
new file mode 100644
index 0000000000000..9b360e4be79cf
--- /dev/null
+++ b/web/app/components/base/icons/assets/vender/solid/shapes/corner.svg
@@ -0,0 +1,3 @@
+<svg width="13" height="20" viewBox="0 0 13 20" fill="none" xmlns="http://www.w3.org/2000/svg">
+<path id="Shape" d="M0 0H13V20C9.98017 20 7.26458 18.1615 6.14305 15.3576L0 0Z" fill="#F9FAFB"/>
+</svg>
diff --git a/web/app/components/base/icons/src/vender/solid/development/ApiConnectionMod.json b/web/app/components/base/icons/src/vender/solid/development/ApiConnectionMod.json
new file mode 100644
index 0000000000000..e8ebcc7448d03
--- /dev/null
+++ b/web/app/components/base/icons/src/vender/solid/development/ApiConnectionMod.json
@@ -0,0 +1,38 @@
+{
+	"icon": {
+		"type": "element",
+		"isRootNode": true,
+		"name": "svg",
+		"attributes": {
+			"width": "16",
+			"height": "16",
+			"viewBox": "0 0 16 16",
+			"fill": "none",
+			"xmlns": "http://www.w3.org/2000/svg"
+		},
+		"children": [
+			{
+				"type": "element",
+				"name": "g",
+				"attributes": {
+					"id": "Icon L"
+				},
+				"children": [
+					{
+						"type": "element",
+						"name": "path",
+						"attributes": {
+							"id": "Vector",
+							"fill-rule": "evenodd",
+							"clip-rule": "evenodd",
+							"d": "M7.99996 3.33333C5.42263 3.33333 3.33329 5.42267 3.33329 8C3.33329 10.5773 5.42263 12.6667 7.99996 12.6667C9.72643 12.6667 11.2348 11.7295 12.0427 10.3329C12.227 10.0141 12.6349 9.90523 12.9536 10.0896C13.2723 10.274 13.3812 10.6818 13.1968 11.0005C12.1604 12.7921 10.2216 14 7.99996 14C4.91159 14 2.36821 11.6666 2.03658 8.66667H1.33329C0.965103 8.66667 0.666626 8.36819 0.666626 8C0.666626 7.63181 0.965103 7.33333 1.33329 7.33333H2.03658C2.36821 4.33337 4.91159 2 7.99996 2C10.2216 2 12.1604 3.20785 13.1968 4.99952C13.3812 5.31823 13.2723 5.72605 12.9536 5.91041C12.6349 6.09477 12.227 5.98585 12.0427 5.66714C11.2348 4.27054 9.72643 3.33333 7.99996 3.33333ZM7.99996 6C6.89539 6 5.99996 6.89543 5.99996 8C5.99996 9.10455 6.89539 10 7.99996 10C9.1045 10 9.99996 9.10454 9.99996 8C9.99996 6.89543 9.10451 6 7.99996 6ZM4.66663 8C4.66663 6.15905 6.15901 4.66667 7.99996 4.66667C9.61257 4.66667 10.9578 5.81184 11.2666 7.33333H14.6666C15.0348 7.33333 15.3333 7.63181 15.3333 8C15.3333 8.36819 15.0348 8.66667 14.6666 8.66667H11.2666C10.9578 10.1881 9.61257 11.3333 7.99996 11.3333C6.159 11.3333 4.66663 9.84092 4.66663 8Z",
+							"fill": "currentColor"
+						},
+						"children": []
+					}
+				]
+			}
+		]
+	},
+	"name": "ApiConnectionMod"
+}
\ No newline at end of file
diff --git a/web/app/components/base/icons/src/vender/solid/development/ApiConnectionMod.tsx b/web/app/components/base/icons/src/vender/solid/development/ApiConnectionMod.tsx
new file mode 100644
index 0000000000000..f88431a237b38
--- /dev/null
+++ b/web/app/components/base/icons/src/vender/solid/development/ApiConnectionMod.tsx
@@ -0,0 +1,16 @@
+// GENERATE BY script
+// DON NOT EDIT IT MANUALLY
+
+import * as React from 'react'
+import data from './ApiConnectionMod.json'
+import IconBase from '@/app/components/base/icons/IconBase'
+import type { IconBaseProps, IconData } from '@/app/components/base/icons/IconBase'
+
+const Icon = React.forwardRef<React.MutableRefObject<SVGElement>, Omit<IconBaseProps, 'data'>>((
+  props,
+  ref,
+) => <IconBase {...props} ref={ref} data={data as IconData} />)
+
+Icon.displayName = 'ApiConnectionMod'
+
+export default Icon
diff --git a/web/app/components/base/icons/src/vender/solid/development/index.ts b/web/app/components/base/icons/src/vender/solid/development/index.ts
index 7fe781e2cec7c..159a1cbc0b284 100644
--- a/web/app/components/base/icons/src/vender/solid/development/index.ts
+++ b/web/app/components/base/icons/src/vender/solid/development/index.ts
@@ -1,3 +1,4 @@
+export { default as ApiConnectionMod } from './ApiConnectionMod'
 export { default as ApiConnection } from './ApiConnection'
 export { default as BarChartSquare02 } from './BarChartSquare02'
 export { default as Container } from './Container'
diff --git a/web/app/components/base/icons/src/vender/solid/shapes/Corner.json b/web/app/components/base/icons/src/vender/solid/shapes/Corner.json
new file mode 100644
index 0000000000000..2f35483a66133
--- /dev/null
+++ b/web/app/components/base/icons/src/vender/solid/shapes/Corner.json
@@ -0,0 +1,27 @@
+{
+	"icon": {
+		"type": "element",
+		"isRootNode": true,
+		"name": "svg",
+		"attributes": {
+			"width": "13",
+			"height": "20",
+			"viewBox": "0 0 13 20",
+			"fill": "none",
+			"xmlns": "http://www.w3.org/2000/svg"
+		},
+		"children": [
+			{
+				"type": "element",
+				"name": "path",
+				"attributes": {
+					"id": "Shape",
+					"d": "M0 0H13V20C9.98017 20 7.26458 18.1615 6.14305 15.3576L0 0Z",
+					"fill": "currentColor"
+				},
+				"children": []
+			}
+		]
+	},
+	"name": "Corner"
+}
\ No newline at end of file
diff --git a/web/app/components/base/icons/src/vender/solid/shapes/Corner.tsx b/web/app/components/base/icons/src/vender/solid/shapes/Corner.tsx
new file mode 100644
index 0000000000000..0edeb2cda15f1
--- /dev/null
+++ b/web/app/components/base/icons/src/vender/solid/shapes/Corner.tsx
@@ -0,0 +1,16 @@
+// GENERATE BY script
+// DON NOT EDIT IT MANUALLY
+
+import * as React from 'react'
+import data from './Corner.json'
+import IconBase from '@/app/components/base/icons/IconBase'
+import type { IconBaseProps, IconData } from '@/app/components/base/icons/IconBase'
+
+const Icon = React.forwardRef<React.MutableRefObject<SVGElement>, Omit<IconBaseProps, 'data'>>((
+  props,
+  ref,
+) => <IconBase {...props} ref={ref} data={data as IconData} />)
+
+Icon.displayName = 'Corner'
+
+export default Icon
diff --git a/web/app/components/base/icons/src/vender/solid/shapes/index.ts b/web/app/components/base/icons/src/vender/solid/shapes/index.ts
index 3122e776b520a..2768e3949a313 100644
--- a/web/app/components/base/icons/src/vender/solid/shapes/index.ts
+++ b/web/app/components/base/icons/src/vender/solid/shapes/index.ts
@@ -1,2 +1,3 @@
+export { default as Corner } from './Corner'
 export { default as Star04 } from './Star04'
 export { default as Star06 } from './Star06'
diff --git a/web/app/components/base/modal/index.css b/web/app/components/base/modal/index.css
index 799b375168d01..727a9455d7fbc 100644
--- a/web/app/components/base/modal/index.css
+++ b/web/app/components/base/modal/index.css
@@ -3,5 +3,5 @@
 }
 
 .modal-panel {
-  @apply w-full max-w-md transform rounded-2xl bg-white p-6 text-left align-middle shadow-xl transition-all;
+  @apply w-full max-w-[480px] transform rounded-2xl bg-white p-6 text-left align-middle shadow-xl transition-all;
 }
diff --git a/web/app/components/base/param-item/index.tsx b/web/app/components/base/param-item/index.tsx
index 318b0fb0e3b82..49acc8148455e 100644
--- a/web/app/components/base/param-item/index.tsx
+++ b/web/app/components/base/param-item/index.tsx
@@ -37,6 +37,7 @@ const ParamItem: FC<Props> = ({ className, id, name, noTooltip, tip, step = 0.1,
           <span className="mx-1 text-gray-900 text-[13px] leading-[18px] font-medium">{name}</span>
           {!noTooltip && (
             <Tooltip
+              triggerClassName='w-4 h-4 shrink-0'
               popupContent={<div className="w-[200px]">{tip}</div>}
             />
           )}
diff --git a/web/app/components/base/select/index.tsx b/web/app/components/base/select/index.tsx
index dee983690b3c4..e821cb154516b 100644
--- a/web/app/components/base/select/index.tsx
+++ b/web/app/components/base/select/index.tsx
@@ -87,7 +87,7 @@ const Select: FC<ISelectProps> = ({
         <div className='group text-gray-800'>
           {allowSearch
             ? <Combobox.Input
-              className={`w-full rounded-lg border-0 ${bgClassName} py-1.5 pl-3 pr-10 shadow-sm sm:text-sm sm:leading-6 focus-visible:outline-none focus-visible:bg-gray-200 group-hover:bg-gray-200 cursor-not-allowed`}
+              className={`w-full rounded-lg border-0 ${bgClassName} py-1.5 pl-3 pr-10 shadow-sm sm:text-sm sm:leading-6 focus-visible:outline-none focus-visible:bg-gray-200 group-hover:bg-gray-200 ${disabled ? 'cursor-not-allowed' : 'cursor-pointer'}`}
               onChange={(event) => {
                 if (!disabled)
                   setQuery(event.target.value)
diff --git a/web/app/components/datasets/documents/detail/completed/SegmentCard.tsx b/web/app/components/datasets/documents/detail/completed/SegmentCard.tsx
index c65b244f6d315..5b76acc9360c6 100644
--- a/web/app/components/datasets/documents/detail/completed/SegmentCard.tsx
+++ b/web/app/components/datasets/documents/detail/completed/SegmentCard.tsx
@@ -36,6 +36,12 @@ export type UsageScene = 'doc' | 'hitTesting'
 type ISegmentCardProps = {
   loading: boolean
   detail?: SegmentDetailModel & { document: { name: string } }
+  contentExternal?: string
+  refSource?: {
+    title: string
+    uri: string
+  }
+  isExternal?: boolean
   score?: number
   onClick?: () => void
   onChangeSwitch?: (segId: string, enabled: boolean) => Promise<void>
@@ -48,6 +54,9 @@ type ISegmentCardProps = {
 
 const SegmentCard: FC<ISegmentCardProps> = ({
   detail = {},
+  contentExternal,
+  isExternal,
+  refSource,
   score,
   onClick,
   onChangeSwitch,
@@ -88,6 +97,9 @@ const SegmentCard: FC<ISegmentCardProps> = ({
       )
     }
 
+    if (contentExternal)
+      return contentExternal
+
     return content
   }
 
@@ -199,16 +211,16 @@ const SegmentCard: FC<ISegmentCardProps> = ({
               </div>
               <div className={cn('w-full bg-gray-50 group-hover:bg-white')}>
                 <Divider />
-                <div className="relative flex items-center w-full">
+                <div className="relative flex items-center w-full pb-1">
                   <DocumentTitle
-                    name={detail?.document?.name || ''}
-                    extension={(detail?.document?.name || '').split('.').pop() || 'txt'}
+                    name={detail?.document?.name || refSource?.title || ''}
+                    extension={(detail?.document?.name || refSource?.title || '').split('.').pop() || 'txt'}
                     wrapperCls='w-full'
                     iconCls="!h-4 !w-4 !bg-contain"
                     textCls="text-xs text-gray-700 !font-normal overflow-hidden whitespace-nowrap text-ellipsis"
                   />
                   <div className={cn(s.chartLinkText, 'group-hover:inline-flex')}>
-                    {t('datasetHitTesting.viewChart')}
+                    {isExternal ? t('datasetHitTesting.viewDetail') : t('datasetHitTesting.viewChart')}
                     <ArrowUpRightIcon className="w-3 h-3 ml-1 stroke-current stroke-2" />
                   </div>
                 </div>
diff --git a/web/app/components/datasets/external-api/declarations.ts b/web/app/components/datasets/external-api/declarations.ts
new file mode 100644
index 0000000000000..ded736d1677da
--- /dev/null
+++ b/web/app/components/datasets/external-api/declarations.ts
@@ -0,0 +1,16 @@
+export type CreateExternalAPIReq = {
+  name: string
+  settings: {
+    endpoint: string
+    api_key: string
+  }
+}
+
+export type FormSchema = {
+  variable: string
+  type: 'text' | 'secret'
+  label: {
+    [key: string]: string
+  }
+  required: boolean
+}
diff --git a/web/app/components/datasets/external-api/external-api-modal/Form.tsx b/web/app/components/datasets/external-api/external-api-modal/Form.tsx
new file mode 100644
index 0000000000000..ada01493fe8e4
--- /dev/null
+++ b/web/app/components/datasets/external-api/external-api-modal/Form.tsx
@@ -0,0 +1,90 @@
+import React, { useState } from 'react'
+import type { FC } from 'react'
+import { useTranslation } from 'react-i18next'
+import { RiBookOpenLine } from '@remixicon/react'
+import type { CreateExternalAPIReq, FormSchema } from '../declarations'
+import Input from '@/app/components/base/input'
+import cn from '@/utils/classnames'
+
+type FormProps = {
+  className?: string
+  itemClassName?: string
+  fieldLabelClassName?: string
+  value: CreateExternalAPIReq
+  onChange: (val: CreateExternalAPIReq) => void
+  formSchemas: FormSchema[]
+  inputClassName?: string
+}
+
+const Form: FC<FormProps> = React.memo(({
+  className,
+  itemClassName,
+  fieldLabelClassName,
+  value,
+  onChange,
+  formSchemas,
+  inputClassName,
+}) => {
+  const { t, i18n } = useTranslation()
+  const [changeKey, setChangeKey] = useState('')
+
+  const handleFormChange = (key: string, val: string) => {
+    setChangeKey(key)
+    if (key === 'name') {
+      onChange({ ...value, [key]: val })
+    }
+    else {
+      onChange({
+        ...value,
+        settings: {
+          ...value.settings,
+          [key]: val,
+        },
+      })
+    }
+  }
+
+  const renderField = (formSchema: FormSchema) => {
+    const { variable, type, label, required } = formSchema
+    const fieldValue = variable === 'name' ? value[variable] : (value.settings[variable as keyof typeof value.settings] || '')
+
+    return (
+      <div key={variable} className={cn(itemClassName, 'flex flex-col items-start gap-1 self-stretch')}>
+        <div className="flex justify-between items-center w-full">
+          <label className={cn(fieldLabelClassName, 'text-text-secondary system-sm-semibold')} htmlFor={variable}>
+            {label[i18n.language] || label.en_US}
+            {required && <span className='ml-1 text-red-500'>*</span>}
+          </label>
+          {variable === 'endpoint' && (
+            <a
+              href={'https://docs.dify.ai/guides/knowledge-base/external-knowledge-api-documentation' || '/'}
+              target='_blank'
+              rel='noopener noreferrer'
+              className='text-text-accent body-xs-regular flex items-center'
+            >
+              <RiBookOpenLine className='w-3 h-3 text-text-accent mr-1' />
+              {t('dataset.externalAPIPanelDocumentation')}
+            </a>
+          )}
+        </div>
+        <Input
+          type={type === 'secret' ? 'password' : 'text'}
+          id={variable}
+          name={variable}
+          value={fieldValue}
+          onChange={val => handleFormChange(variable, val.target.value)}
+          required={required}
+          className={cn(inputClassName)}
+        />
+      </div>
+    )
+  }
+
+  return (
+    <form className={cn('flex flex-col justify-center items-start gap-4 self-stretch', className)}>
+      {formSchemas.map(formSchema => renderField(formSchema))}
+    </form>
+  )
+})
+
+export default Form
diff --git a/web/app/components/datasets/external-api/external-api-modal/index.tsx b/web/app/components/datasets/external-api/external-api-modal/index.tsx
new file mode 100644
index 0000000000000..340d147a505bd
--- /dev/null
+++ b/web/app/components/datasets/external-api/external-api-modal/index.tsx
@@ -0,0 +1,218 @@
+import type { FC } from 'react'
+import {
+  memo,
+  useEffect,
+  useState,
+} from 'react'
+import { useTranslation } from 'react-i18next'
+import {
+  RiBook2Line,
+  RiCloseLine,
+  RiInformation2Line,
+  RiLock2Fill,
+} from '@remixicon/react'
+import type { CreateExternalAPIReq, FormSchema } from '../declarations'
+import Form from './Form'
+import ActionButton from '@/app/components/base/action-button'
+import Confirm from '@/app/components/base/confirm'
+import {
+  PortalToFollowElem,
+  PortalToFollowElemContent,
+} from '@/app/components/base/portal-to-follow-elem'
+import { createExternalAPI } from '@/service/datasets'
+import { useToastContext } from '@/app/components/base/toast'
+import Button from '@/app/components/base/button'
+import Tooltip from '@/app/components/base/tooltip'
+
+type AddExternalAPIModalProps = {
+  data?: CreateExternalAPIReq
+  onSave: (formValue: CreateExternalAPIReq) => void
+  onCancel: () => void
+  onEdit?: (formValue: CreateExternalAPIReq) => Promise<void>
+  datasetBindings?: { id: string; name: string }[]
+  isEditMode: boolean
+}
+
+const formSchemas: FormSchema[] = [
+  {
+    variable: 'name',
+    type: 'text',
+    label: {
+      en_US: 'Name',
+    },
+    required: true,
+  },
+  {
+    variable: 'endpoint',
+    type: 'text',
+    label: {
+      en_US: 'API Endpoint',
+    },
+    required: true,
+  },
+  {
+    variable: 'api_key',
+    type: 'secret',
+    label: {
+      en_US: 'API Key',
+    },
+    required: true,
+  },
+]
+
+const AddExternalAPIModal: FC<AddExternalAPIModalProps> = ({ data, onSave, onCancel, datasetBindings, isEditMode, onEdit }) => {
+  const { t } = useTranslation()
+  const { notify } = useToastContext()
+  const [loading, setLoading] = useState(false)
+  const [showConfirm, setShowConfirm] = useState(false)
+  const [formData, setFormData] = useState<CreateExternalAPIReq>({ name: '', settings: { endpoint: '', api_key: '' } })
+
+  useEffect(() => {
+    if (isEditMode && data)
+      setFormData(data)
+  }, [isEditMode, data])
+
+  const hasEmptyInputs = Object.values(formData).some(value =>
+    typeof value === 'string' ? value.trim() === '' : Object.values(value).some(v => v.trim() === ''),
+  )
+  const handleDataChange = (val: CreateExternalAPIReq) => {
+    setFormData(val)
+  }
+
+  const handleSave = async () => {
+    if (formData && formData.settings.api_key && formData.settings.api_key?.length < 5) {
+      notify({ type: 'error', message: t('common.apiBasedExtension.modal.apiKey.lengthError') })
+      setLoading(false)
+      return
+    }
+    try {
+      setLoading(true)
+      if (isEditMode && onEdit) {
+        await onEdit(
+          {
+            ...formData,
+            settings: { ...formData.settings, api_key: formData.settings.api_key ? '[__HIDDEN__]' : formData.settings.api_key },
+          },
+        )
+        notify({ type: 'success', message: 'External API updated successfully' })
+      }
+      else {
+        const res = await createExternalAPI({ body: formData })
+        if (res && res.id) {
+          notify({ type: 'success', message: 'External API saved successfully' })
+          onSave(res)
+        }
+      }
+      onCancel()
+    }
+    catch (error) {
+      console.error('Error saving/updating external API:', error)
+      notify({ type: 'error', message: 'Failed to save/update External API' })
+    }
+    finally {
+      setLoading(false)
+    }
+  }
+
+  return (
+    <PortalToFollowElem open>
+      <PortalToFollowElemContent className='w-full h-full z-[60]'>
+        <div className='fixed inset-0 flex items-center justify-center bg-black/[.25]'>
+          <div className='flex relative w-[480px] flex-col items-start bg-components-panel-bg rounded-2xl border-[0.5px] border-components-panel-border shadows-shadow-xl'>
+            <div className='flex flex-col pt-6 pl-6 pb-3 pr-14 items-start gap-2 self-stretch'>
+              <div className='self-stretch text-text-primary title-2xl-semi-bold flex-grow'>
+                {
+                  isEditMode ? t('dataset.editExternalAPIFormTitle') : t('dataset.createExternalAPI')
+                }
+              </div>
+              {isEditMode && (datasetBindings?.length ?? 0) > 0 && (
+                <div className='text-text-tertiary system-xs-regular flex items-center'>
+                  {t('dataset.editExternalAPIFormWarning.front')}
+                  <span className='text-text-accent cursor-pointer flex items-center'>
+                    &nbsp;{datasetBindings?.length} {t('dataset.editExternalAPIFormWarning.end')}&nbsp;
+                    <Tooltip
+                      popupClassName='flex items-center self-stretch w-[320px]'
+                      popupContent={
+                        <div className='p-1'>
+                          <div className='flex pt-1 pb-0.5 pl-2 pr-3 items-start self-stretch'>
+                            <div className='text-text-tertiary system-xs-medium-uppercase'>{`${datasetBindings?.length} ${t('dataset.editExternalAPITooltipTitle')}`}</div>
+                          </div>
+                          {datasetBindings?.map(binding => (
+                            <div key={binding.id} className='flex px-2 py-1 items-center gap-1 self-stretch'>
+                              <RiBook2Line className='w-4 h-4 text-text-secondary' />
+                              <div className='text-text-secondary system-sm-medium'>{binding.name}</div>
+                            </div>
+                          ))}
+                        </div>
+                      }
+                      asChild={false}
+                      position='bottom'
+                    >
+                      <RiInformation2Line className='w-3.5 h-3.5' />
+                    </Tooltip>
+                  </span>
+                </div>
+              )}
+            </div>
+            <ActionButton className='absolute top-5 right-5' onClick={onCancel}>
+              <RiCloseLine className='w-[18px] h-[18px] text-text-tertiary flex-shrink-0' />
+            </ActionButton>
+            <Form
+              value={formData}
+              onChange={handleDataChange}
+              formSchemas={formSchemas}
+              className='flex px-6 py-3 flex-col justify-center items-start gap-4 self-stretch'
+            />
+            <div className='flex p-6 pt-5 justify-end items-center gap-2 self-stretch'>
+              <Button type='button' variant='secondary' onClick={onCancel}>
+                {t('dataset.externalAPIForm.cancel')}
+              </Button>
+              <Button
+                type='submit'
+                variant='primary'
+                onClick={() => {
+                  if (isEditMode && (datasetBindings?.length ?? 0) > 0)
+                    setShowConfirm(true)
+                  else if (isEditMode && onEdit)
+                    onEdit(formData)
+
+                  else
+                    handleSave()
+                }}
+                disabled={hasEmptyInputs || loading}
+              >
+                {t('dataset.externalAPIForm.save')}
+              </Button>
+            </div>
+            <div className='flex px-2 py-3 justify-center items-center gap-1 self-stretch rounded-b-2xl
+              border-t-[0.5px] border-divider-subtle bg-background-soft text-text-tertiary system-xs-regular'
+            >
+              <RiLock2Fill className='w-3 h-3 text-text-quaternary' />
+              {t('dataset.externalAPIForm.encrypted.front')}
+              <a
+                className='text-text-accent'
+                target='_blank' rel='noopener noreferrer'
+                href='https://pycryptodome.readthedocs.io/en/latest/src/cipher/oaep.html'
+              >
+                PKCS1_OAEP
+              </a>
+              {t('dataset.externalAPIForm.encrypted.end')}
+            </div>
+          </div>
+          {showConfirm && (datasetBindings?.length ?? 0) > 0 && (
+            <Confirm
+              isShow={showConfirm}
+              type='warning'
+              title='Warning'
+              content={`${t('dataset.editExternalAPIConfirmWarningContent.front')} ${datasetBindings?.length} ${t('dataset.editExternalAPIConfirmWarningContent.end')}`}
+              onCancel={() => setShowConfirm(false)}
+              onConfirm={handleSave}
+            />
+          )}
+        </div>
+      </PortalToFollowElemContent>
+    </PortalToFollowElem>
+  )
+}
+
+export default memo(AddExternalAPIModal)
diff --git a/web/app/components/datasets/external-api/external-api-panel/index.tsx b/web/app/components/datasets/external-api/external-api-panel/index.tsx
new file mode 100644
index 0000000000000..044c008b12a60
--- /dev/null
+++ b/web/app/components/datasets/external-api/external-api-panel/index.tsx
@@ -0,0 +1,90 @@
+import React from 'react'
+import {
+  RiAddLine,
+  RiBookOpenLine,
+  RiCloseLine,
+} from '@remixicon/react'
+import { useTranslation } from 'react-i18next'
+import ExternalKnowledgeAPICard from '../external-knowledge-api-card'
+import cn from '@/utils/classnames'
+import { useExternalKnowledgeApi } from '@/context/external-knowledge-api-context'
+import ActionButton from '@/app/components/base/action-button'
+import Button from '@/app/components/base/button'
+import Loading from '@/app/components/base/loading'
+import { useModalContext } from '@/context/modal-context'
+
+type ExternalAPIPanelProps = {
+  onClose: () => void
+}
+
+const ExternalAPIPanel: React.FC<ExternalAPIPanelProps> = ({ onClose }) => {
+  const { t } = useTranslation()
+  const { setShowExternalKnowledgeAPIModal } = useModalContext()
+  const { externalKnowledgeApiList, mutateExternalKnowledgeApis, isLoading } = useExternalKnowledgeApi()
+
+  const handleOpenExternalAPIModal = () => {
+    setShowExternalKnowledgeAPIModal({
+      payload: { name: '', settings: { endpoint: '', api_key: '' } },
+      datasetBindings: [],
+      onSaveCallback: () => {
+        mutateExternalKnowledgeApis()
+      },
+      onCancelCallback: () => {
+        mutateExternalKnowledgeApis()
+      },
+      isEditMode: false,
+    })
+  }
+
+  return (
+    <div
+      tabIndex={-1}
+      className={cn('absolute top-14 right-0 bottom-2 flex z-10 outline-none')}
+    >
+      <div
+        className={cn(
+          'relative flex flex-col w-[420px] bg-components-panel-bg-alt rounded-l-2xl h-full border border-components-panel-border',
+        )}
+      >
+        <div className='flex items-start self-stretch p-4 pb-0'>
+          <div className='flex flex-col items-start gap-1 flex-grow'>
+            <div className='self-stretch text-text-primary system-xl-semibold'>{t('dataset.externalAPIPanelTitle')}</div>
+            <div className='self-stretch text-text-tertiary body-xs-regular'>{t('dataset.externalAPIPanelDescription')}</div>
+            <a className='flex justify-center items-center gap-1 self-stretch cursor-pointer' href='https://docs.dify.ai/guides/knowledge-base/external-knowledge-api-documentation' target='_blank'>
+              <RiBookOpenLine className='w-3 h-3 text-text-accent' />
+              <div className='flex-grow text-text-accent body-xs-regular'>{t('dataset.externalAPIPanelDocumentation')}</div>
+            </a>
+          </div>
+          <div className='flex items-center'>
+            <ActionButton onClick={() => onClose()}>
+              <RiCloseLine className='w-4 h-4 text-text-tertiary' />
+            </ActionButton>
+          </div>
+        </div>
+        <div className='flex px-4 py-3 flex-col justify-center items-start gap-2 self-stretch'>
+          <Button
+            variant={'primary'}
+            className='flex justify-center items-center px-3 py-2 gap-0.5'
+            onClick={handleOpenExternalAPIModal}
+          >
+            <RiAddLine className='w-4 h-4 text-components-button-primary-text' />
+            <div className='text-components-button-primary-text system-sm-medium'>{t('dataset.createExternalAPI')}</div>
+          </Button>
+        </div>
+        <div className='flex py-0 px-4 flex-col items-start gap-1 flex-grow self-stretch'>
+          {isLoading
+            ? (
+              <Loading />
+            )
+            : (
+              externalKnowledgeApiList.map(api => (
+                <ExternalKnowledgeAPICard key={api.id} api={api} />
+              ))
+            )}
+        </div>
+      </div>
+    </div>
+  )
+}
+
+export default ExternalAPIPanel
diff --git a/web/app/components/datasets/external-api/external-knowledge-api-card/index.tsx b/web/app/components/datasets/external-api/external-knowledge-api-card/index.tsx
new file mode 100644
index 0000000000000..603b4fe7cb228
--- /dev/null
+++ b/web/app/components/datasets/external-api/external-knowledge-api-card/index.tsx
@@ -0,0 +1,151 @@
+import React, { useState } from 'react'
+import { useTranslation } from 'react-i18next'
+import {
+  RiDeleteBinLine,
+  RiEditLine,
+} from '@remixicon/react'
+import type { CreateExternalAPIReq } from '../declarations'
+import type { ExternalAPIItem } from '@/models/datasets'
+import { checkUsageExternalAPI, deleteExternalAPI, fetchExternalAPI, updateExternalAPI } from '@/service/datasets'
+import { ApiConnectionMod } from '@/app/components/base/icons/src/vender/solid/development'
+import { useExternalKnowledgeApi } from '@/context/external-knowledge-api-context'
+import { useModalContext } from '@/context/modal-context'
+import ActionButton from '@/app/components/base/action-button'
+import Confirm from '@/app/components/base/confirm'
+
+type ExternalKnowledgeAPICardProps = {
+  api: ExternalAPIItem
+}
+
+const ExternalKnowledgeAPICard: React.FC<ExternalKnowledgeAPICardProps> = ({ api }) => {
+  const { setShowExternalKnowledgeAPIModal } = useModalContext()
+  const [showConfirm, setShowConfirm] = useState(false)
+  const [isHovered, setIsHovered] = useState(false)
+  const [usageCount, setUsageCount] = useState(0)
+  const { mutateExternalKnowledgeApis } = useExternalKnowledgeApi()
+
+  const { t } = useTranslation()
+
+  const handleEditClick = async () => {
+    try {
+      const response = await fetchExternalAPI({ apiTemplateId: api.id })
+      const formValue: CreateExternalAPIReq = {
+        name: response.name,
+        settings: {
+          endpoint: response.settings.endpoint,
+          api_key: response.settings.api_key,
+        },
+      }
+
+      setShowExternalKnowledgeAPIModal({
+        payload: formValue,
+        onSaveCallback: () => {
+          mutateExternalKnowledgeApis()
+        },
+        onCancelCallback: () => {
+          mutateExternalKnowledgeApis()
+        },
+        isEditMode: true,
+        datasetBindings: response.dataset_bindings,
+        onEditCallback: async (updatedData: CreateExternalAPIReq) => {
+          try {
+            await updateExternalAPI({
+              apiTemplateId: api.id,
+              body: {
+                ...response,
+                name: updatedData.name,
+                settings: {
+                  ...response.settings,
+                  endpoint: updatedData.settings.endpoint,
+                  api_key: updatedData.settings.api_key,
+                },
+              },
+            })
+            mutateExternalKnowledgeApis()
+          }
+          catch (error) {
+            console.error('Error updating external knowledge API:', error)
+          }
+        },
+      })
+    }
+    catch (error) {
+      console.error('Error fetching external knowledge API data:', error)
+    }
+  }
+
+  const handleDeleteClick = async () => {
+    try {
+      const usage = await checkUsageExternalAPI({ apiTemplateId: api.id })
+      if (usage.is_using)
+        setUsageCount(usage.count)
+
+      setShowConfirm(true)
+    }
+    catch (error) {
+      console.error('Error checking external API usage:', error)
+    }
+  }
+
+  const handleConfirmDelete = async () => {
+    try {
+      const response = await deleteExternalAPI({ apiTemplateId: api.id })
+      if (response && response.result === 'success') {
+        setShowConfirm(false)
+        mutateExternalKnowledgeApis()
+      }
+      else {
+        console.error('Failed to delete external API')
+      }
+    }
+    catch (error) {
+      console.error('Error deleting external knowledge API:', error)
+    }
+  }
+
+  return (
+    <>
+      <div className={`flex p-2 pl-3 items-start self-stretch rounded-lg border-[0.5px] 
+        border-components-panel-border-subtle bg-components-panel-on-panel-item-bg 
+        shadows-shadow-xs ${isHovered ? 'bg-state-destructive-hover border-state-destructive-border' : ''}`}
+      >
+        <div className='flex py-1 flex-col justify-center items-start gap-1.5 flex-grow'>
+          <div className='flex items-center gap-1 self-stretch text-text-secondary'>
+            <ApiConnectionMod className='w-4 h-4' />
+            <div className='system-sm-medium'>{api.name}</div>
+          </div>
+          <div className='self-stretch text-text-tertiary system-xs-regular'>{api.settings.endpoint}</div>
+        </div>
+        <div className='flex items-start gap-1'>
+          <ActionButton onClick={handleEditClick}>
+            <RiEditLine className='w-4 h-4 text-text-tertiary hover:text-text-secondary' />
+          </ActionButton>
+          <ActionButton
+            className='hover:bg-state-destructive-hover'
+            onClick={handleDeleteClick}
+            onMouseEnter={() => setIsHovered(true)}
+            onMouseLeave={() => setIsHovered(false)}
+          >
+            <RiDeleteBinLine className='w-4 h-4 text-text-tertiary hover:text-text-destructive' />
+          </ActionButton>
+        </div>
+      </div>
+      {showConfirm && (
+        <Confirm
+          isShow={showConfirm}
+          title={`${t('dataset.deleteExternalAPIConfirmWarningContent.title.front')} ${api.name}${t('dataset.deleteExternalAPIConfirmWarningContent.title.end')}`}
+          content={
+            usageCount > 0
+              ? `${t('dataset.deleteExternalAPIConfirmWarningContent.content.front')} ${usageCount} ${t('dataset.deleteExternalAPIConfirmWarningContent.content.end')}`
+              : t('dataset.deleteExternalAPIConfirmWarningContent.noConnectionContent')
+          }
+          type='warning'
+          onConfirm={handleConfirmDelete}
+          onCancel={() => setShowConfirm(false)}
+        />
+      )}
+    </>
+  )
+}
+
+export default ExternalKnowledgeAPICard
diff --git a/web/app/components/datasets/external-knowledge-base/connector/index.tsx b/web/app/components/datasets/external-knowledge-base/connector/index.tsx
new file mode 100644
index 0000000000000..33f57d0b47dbf
--- /dev/null
+++ b/web/app/components/datasets/external-knowledge-base/connector/index.tsx
@@ -0,0 +1,36 @@
+'use client'
+
+import React, { useState } from 'react'
+import { useRouter } from 'next/navigation'
+import { useToastContext } from '@/app/components/base/toast'
+import ExternalKnowledgeBaseCreate from '@/app/components/datasets/external-knowledge-base/create'
+import type { CreateKnowledgeBaseReq } from '@/app/components/datasets/external-knowledge-base/create/declarations'
+import { createExternalKnowledgeBase } from '@/service/datasets'
+
+const ExternalKnowledgeBaseConnector = () => {
+  const { notify } = useToastContext()
+  const [loading, setLoading] = useState(false)
+  const router = useRouter()
+
+  const handleConnect = async (formValue: CreateKnowledgeBaseReq) => {
+    try {
+      setLoading(true)
+      const result = await createExternalKnowledgeBase({ body: formValue })
+      if (result && result.id) {
+        notify({ type: 'success', message: 'External Knowledge Base Connected Successfully' })
+        router.back()
+      }
+      else { throw new Error('Failed to create external knowledge base') }
+    }
+    catch (error) {
+      console.error('Error creating external knowledge base:', error)
+      notify({ type: 'error', message: 'Failed to connect External Knowledge Base' })
+    }
+    setLoading(false)
+  }
+  return (
+    <ExternalKnowledgeBaseCreate onConnect={handleConnect} loading={loading} />
+  )
+}
+
+export default ExternalKnowledgeBaseConnector
diff --git a/web/app/components/datasets/external-knowledge-base/create/ExternalApiSelect.tsx b/web/app/components/datasets/external-knowledge-base/create/ExternalApiSelect.tsx
new file mode 100644
index 0000000000000..a6a46479a4eb4
--- /dev/null
+++ b/web/app/components/datasets/external-knowledge-base/create/ExternalApiSelect.tsx
@@ -0,0 +1,110 @@
+import React, { useEffect, useState } from 'react'
+import {
+  RiAddLine,
+  RiArrowDownSLine,
+} from '@remixicon/react'
+import { useTranslation } from 'react-i18next'
+import { useRouter } from 'next/navigation'
+import { ApiConnectionMod } from '@/app/components/base/icons/src/vender/solid/development'
+import { useModalContext } from '@/context/modal-context'
+import { useExternalKnowledgeApi } from '@/context/external-knowledge-api-context'
+
+type ApiItem = {
+  value: string
+  name: string
+  url: string
+}
+
+type ExternalApiSelectProps = {
+  items: ApiItem[]
+  value?: string
+  onSelect: (item: ApiItem) => void
+}
+
+const ExternalApiSelect: React.FC<ExternalApiSelectProps> = ({ items, value, onSelect }) => {
+  const { t } = useTranslation()
+  const [isOpen, setIsOpen] = useState(false)
+  const [selectedItem, setSelectedItem] = useState<ApiItem | null>(
+    items.find(item => item.value === value) || null,
+  )
+  const { setShowExternalKnowledgeAPIModal } = useModalContext()
+  const { mutateExternalKnowledgeApis } = useExternalKnowledgeApi()
+  const router = useRouter()
+
+  useEffect(() => {
+    const newSelectedItem = items.find(item => item.value === value) || null
+    setSelectedItem(newSelectedItem)
+  }, [value, items])
+
+  const handleAddNewAPI = () => {
+    setShowExternalKnowledgeAPIModal({
+      payload: { name: '', settings: { endpoint: '', api_key: '' } },
+      onSaveCallback: async () => {
+        mutateExternalKnowledgeApis()
+        router.refresh()
+      },
+      onCancelCallback: () => {
+        mutateExternalKnowledgeApis()
+      },
+      isEditMode: false,
+    })
+  }
+
+  const handleSelect = (item: ApiItem) => {
+    setSelectedItem(item)
+    onSelect(item)
+    setIsOpen(false)
+  }
+
+  return (
+    <div className="relative w-full">
+      <div
+        className={`flex items-center justify-between cursor-pointer px-2 py-1 gap-0.5 self-stretch rounded-lg 
+        bg-components-input-bg-normal hover:bg-state-base-hover-alt ${isOpen && 'bg-state-base-hover-alt'}`}
+        onClick={() => setIsOpen(!isOpen)}
+      >
+        {selectedItem
+          ? (
+            <div className="flex p-1 items-center gap-2 self-stretch rounded-lg">
+              <ApiConnectionMod className='text-text-secondary w-4 h-4' />
+              <div className='flex items-center flex-grow'>
+                <span className='text-components-input-text-filled text-ellipsis system-sm-regular overflow-hidden'>{selectedItem.name}</span>
+              </div>
+            </div>
+          )
+          : (
+            <span className='text-components-input-text-placeholder system-sm-regular'>{t('dataset.selectExternalKnowledgeAPI.placeholder')}</span>
+          )}
+        <RiArrowDownSLine className={`w-4 h-4 text-text-quaternary transition-transform ${isOpen ? 'text-text-secondary' : ''}`} />
+      </div>
+      {isOpen && (
+        <div className="absolute z-10 w-full mt-1 bg-components-panel-bg-blur border rounded-xl shadow-lg">
+          {items.map(item => (
+            <div
+              key={item.value}
+              className="flex p-1 items-center cursor-pointer"
+              onClick={() => handleSelect(item)}
+            >
+              <div className="flex p-2 items-center gap-2 self-stretch rounded-lg hover:bg-state-base-hover w-full">
+                <ApiConnectionMod className='text-text-secondary w-4 h-4' />
+                <span className='text-text-secondary text-ellipsis system-sm-medium overflow-hidden flex-grow'>{item.name}</span>
+                <span className='text-text-tertiary overflow-hidden text-right text-ellipsis system-xs-regular'>{item.url}</span>
+              </div>
+            </div>
+          ))}
+          <div className='flex p-1 flex-col items-start self-stretch'>
+            <div
+              className='flex p-2 items-center gap-2 self-stretch rounded-lg cursor-pointer hover:bg-state-base-hover'
+              onClick={handleAddNewAPI}
+            >
+              <RiAddLine className='text-text-secondary w-4 h-4' />
+              <span className='flex-grow overflow-hidden text-text-secondary text-ellipsis system-sm-medium'>{t('dataset.createNewExternalAPI')}</span>
+            </div>
+          </div>
+        </div>
+      )}
+    </div>
+  )
+}
+
+export default ExternalApiSelect
diff --git a/web/app/components/datasets/external-knowledge-base/create/ExternalApiSelection.tsx b/web/app/components/datasets/external-knowledge-base/create/ExternalApiSelection.tsx
new file mode 100644
index 0000000000000..c910d9b2a7e7e
--- /dev/null
+++ b/web/app/components/datasets/external-knowledge-base/create/ExternalApiSelection.tsx
@@ -0,0 +1,96 @@
+'use client'
+
+import React, { useEffect, useState } from 'react'
+import { useTranslation } from 'react-i18next'
+import { RiAddLine } from '@remixicon/react'
+import { useRouter } from 'next/navigation'
+import ExternalApiSelect from './ExternalApiSelect'
+import Input from '@/app/components/base/input'
+import Button from '@/app/components/base/button'
+import { useModalContext } from '@/context/modal-context'
+import { useExternalKnowledgeApi } from '@/context/external-knowledge-api-context'
+
+type ExternalApiSelectionProps = {
+  external_knowledge_api_id: string
+  external_knowledge_id: string
+  onChange: (data: { external_knowledge_api_id?: string; external_knowledge_id?: string }) => void
+}
+
+const ExternalApiSelection: React.FC<ExternalApiSelectionProps> = ({ external_knowledge_api_id, external_knowledge_id, onChange }) => {
+  const { t } = useTranslation()
+  const router = useRouter()
+  const { externalKnowledgeApiList } = useExternalKnowledgeApi()
+  const [selectedApiId, setSelectedApiId] = useState(external_knowledge_api_id)
+  const { setShowExternalKnowledgeAPIModal } = useModalContext()
+  const { mutateExternalKnowledgeApis } = useExternalKnowledgeApi()
+
+  const apiItems = externalKnowledgeApiList.map(api => ({
+    value: api.id,
+    name: api.name,
+    url: api.settings.endpoint,
+  }))
+
+  useEffect(() => {
+    if (apiItems.length > 0) {
+      const newSelectedId = external_knowledge_api_id || apiItems[0].value
+      setSelectedApiId(newSelectedId)
+      if (newSelectedId !== external_knowledge_api_id)
+        onChange({ external_knowledge_api_id: newSelectedId, external_knowledge_id })
+    }
+  }, [apiItems, external_knowledge_api_id, external_knowledge_id, onChange])
+
+  const handleAddNewAPI = () => {
+    setShowExternalKnowledgeAPIModal({
+      payload: { name: '', settings: { endpoint: '', api_key: '' } },
+      onSaveCallback: async () => {
+        mutateExternalKnowledgeApis()
+        router.refresh()
+      },
+      onCancelCallback: () => {
+        mutateExternalKnowledgeApis()
+      },
+      isEditMode: false,
+    })
+  }
+
+  useEffect(() => {
+    if (!external_knowledge_api_id && apiItems.length > 0)
+      onChange({ external_knowledge_api_id: apiItems[0].value, external_knowledge_id })
+  }, [])
+
+  return (
+    <form className='flex flex-col gap-4 self-stretch'>
+      <div className='flex flex-col gap-1 self-stretch'>
+        <div className='flex flex-col self-stretch'>
+          <label className='text-text-secondary system-sm-semibold'>{t('dataset.externalAPIPanelTitle')}</label>
+        </div>
+        {apiItems.length > 0
+          ? <ExternalApiSelect
+            items={apiItems}
+            value={selectedApiId}
+            onSelect={(e) => {
+              setSelectedApiId(e.value)
+              onChange({ external_knowledge_api_id: e.value, external_knowledge_id })
+            }}
+          />
+          : <Button variant={'tertiary'} onClick={handleAddNewAPI} className='justify-start gap-0.5'>
+            <RiAddLine className='w-4 h-4 text-text-tertiary' />
+            <span className='text-text-tertiary system-sm-regular'>{t('dataset.noExternalKnowledge')}</span>
+          </Button>
+        }
+      </div>
+      <div className='flex flex-col gap-1 self-stretch'>
+        <div className='flex flex-col self-stretch'>
+          <label className='text-text-secondary system-sm-semibold'>{t('dataset.externalKnowledgeId')}</label>
+        </div>
+        <Input
+          value={external_knowledge_id}
+          onChange={e => onChange({ external_knowledge_id: e.target.value, external_knowledge_api_id })}
+          placeholder={t('dataset.externalKnowledgeIdPlaceholder') ?? ''}
+        />
+      </div>
+    </form>
+  )
+}
+
+export default ExternalApiSelection
diff --git a/web/app/components/datasets/external-knowledge-base/create/InfoPanel.tsx b/web/app/components/datasets/external-knowledge-base/create/InfoPanel.tsx
new file mode 100644
index 0000000000000..bd32683c8579c
--- /dev/null
+++ b/web/app/components/datasets/external-knowledge-base/create/InfoPanel.tsx
@@ -0,0 +1,33 @@
+import { RiBookOpenLine } from '@remixicon/react'
+import { useTranslation } from 'react-i18next'
+
+const InfoPanel = () => {
+  const { t } = useTranslation()
+
+  return (
+    <div className='flex w-[360px] pt-[108px] pb-2 pr-8 flex-col items-start'>
+      <div className='flex min-w-[240px] w-full p-6 flex-col items-start gap-3 self-stretch rounded-xl bg-background-section'>
+        <div className='flex p-1 w-10 h-10 justify-center items-center gap-2 flex-grow self-stretch rounded-lg border-[0.5px] border-components-card-border bg-components-card-bg'>
+          <RiBookOpenLine className='w-5 h-5 text-text-accent' />
+        </div>
+        <p className='flex flex-col items-start gap-2 self-stretch'>
+          <span className='self-stretch text-text-secondary system-xl-semibold'>
+            {t('dataset.connectDatasetIntro.title')}
+          </span>
+          <span className='text-text-tertiary system-sm-regular'>
+            {t('dataset.connectDatasetIntro.content.front')}
+            <a className='text-text-accent system-sm-regular ml-1' href='https://docs.dify.ai/guides/knowledge-base/external-knowledge-api-documentation' target='_blank' rel="noopener noreferrer">
+              {t('dataset.connectDatasetIntro.content.link')}
+            </a>
+            {t('dataset.connectDatasetIntro.content.end')}
+          </span>
+          <a className='self-stretch text-text-accent system-sm-regular' href='https://docs.dify.ai/guides/knowledge-base/connect-external-knowledge' target='_blank' rel="noopener noreferrer">
+            {t('dataset.connectDatasetIntro.learnMore')}
+          </a>
+        </p>
+      </div>
+    </div>
+  )
+}
+
+export default InfoPanel
diff --git a/web/app/components/datasets/external-knowledge-base/create/KnowledgeBaseInfo.tsx b/web/app/components/datasets/external-knowledge-base/create/KnowledgeBaseInfo.tsx
new file mode 100644
index 0000000000000..1a2e83d2d6681
--- /dev/null
+++ b/web/app/components/datasets/external-knowledge-base/create/KnowledgeBaseInfo.tsx
@@ -0,0 +1,65 @@
+import React from 'react'
+import { RiBookOpenLine } from '@remixicon/react'
+import { useTranslation } from 'react-i18next'
+import Input from '@/app/components/base/input'
+
+type KnowledgeBaseInfoProps = {
+  name: string
+  description?: string
+  onChange: (data: { name?: string; description?: string }) => void
+}
+
+const KnowledgeBaseInfo: React.FC<KnowledgeBaseInfoProps> = ({ name, description, onChange }) => {
+  const { t } = useTranslation()
+
+  const handleNameChange = (e: React.ChangeEvent<HTMLInputElement>) => {
+    onChange({ name: e.target.value })
+  }
+
+  const handleDescriptionChange = (e: React.ChangeEvent<HTMLTextAreaElement>) => {
+    onChange({ description: e.target.value })
+  }
+
+  return (
+    <form className='flex flex-col gap-4 self-stretch'>
+      <div className='flex flex-col gap-4 self-stretch'>
+        <div className='flex flex-col gap-1 self-stretch'>
+          <div className='flex flex-col justify-center self-stretch'>
+            <label className='text-text-secondary system-sm-semibold'>{t('dataset.externalKnowledgeName')}</label>
+          </div>
+          <Input
+            value={name}
+            onChange={handleNameChange}
+            placeholder={t('dataset.externalKnowledgeNamePlaceholder') ?? ''}
+          />
+        </div>
+        <div className='flex flex-col gap-1 self-stretch'>
+          <div className='flex flex-col justify-center self-stretch'>
+            <label className='text-text-secondary system-sm-semibold'>{t('dataset.externalKnowledgeDescription')}</label>
+          </div>
+          <div className='flex flex-col gap-1 self-stretch'>
+            <textarea
+              value={description}
+              onChange={ e => handleDescriptionChange(e)}
+              placeholder={t('dataset.externalKnowledgeDescriptionPlaceholder') ?? ''}
+              className={`flex h-20 py-2 p-3 self-stretch items-start rounded-lg bg-components-input-bg-normal ${description ? 'text-components-input-text-filled' : 'text-components-input-text-placeholder'} system-sm-regular`}
+            />
+            <a
+              className='flex py-0.5 gap-1 self-stretch'
+              href='https://docs.dify.ai/features/datasets#how-to-write-a-good-dataset-description'
+              target="_blank"
+              rel="noopener noreferrer"
+            >
+              <div className='flex p-0.5 items-center gap-2'>
+                <RiBookOpenLine className='w-3 h-3 text-text-tertiary' />
+              </div>
+              <div className='flex-grow text-text-tertiary body-xs-regular'>{t('dataset.learnHowToWriteGoodKnowledgeDescription')}</div>
+            </a>
+          </div>
+        </div>
+      </div>
+    </form>
+  )
+}
+
+export default KnowledgeBaseInfo
diff --git a/web/app/components/datasets/external-knowledge-base/create/RetrievalSettings.tsx b/web/app/components/datasets/external-knowledge-base/create/RetrievalSettings.tsx
new file mode 100644
index 0000000000000..d501dde271e54
--- /dev/null
+++ b/web/app/components/datasets/external-knowledge-base/create/RetrievalSettings.tsx
@@ -0,0 +1,67 @@
+import type { FC } from 'react'
+import React from 'react'
+import { useTranslation } from 'react-i18next'
+import TopKItem from '@/app/components/base/param-item/top-k-item'
+import ScoreThresholdItem from '@/app/components/base/param-item/score-threshold-item'
+import cn from '@/utils/classnames'
+
+type RetrievalSettingsProps = {
+  topK: number
+  scoreThreshold: number
+  scoreThresholdEnabled: boolean
+  isInHitTesting?: boolean
+  isInRetrievalSetting?: boolean
+  onChange: (data: { top_k?: number; score_threshold?: number; score_threshold_enabled?: boolean }) => void
+}
+
+const RetrievalSettings: FC<RetrievalSettingsProps> = ({
+  topK,
+  scoreThreshold,
+  scoreThresholdEnabled,
+  onChange,
+  isInHitTesting = false,
+  isInRetrievalSetting = false,
+}) => {
+  const { t } = useTranslation()
+
+  const handleScoreThresholdChange = (enabled: boolean) => {
+    onChange({ score_threshold_enabled: enabled })
+  }
+
+  return (
+    <div className={cn('flex flex-col gap-2 self-stretch', isInRetrievalSetting && 'w-full max-w-[480px]')}>
+      {!isInHitTesting && !isInRetrievalSetting && <div className='flex h-7 pt-1 flex-col gap-2 self-stretch'>
+        <label className='text-text-secondary system-sm-semibold'>{t('dataset.retrievalSettings')}</label>
+      </div>}
+      <div className={cn(
+        'flex gap-4 self-stretch',
+        {
+          'flex-col': isInHitTesting,
+          'flex-row': isInRetrievalSetting,
+          'flex-col sm:flex-row': !isInHitTesting && !isInRetrievalSetting,
+        },
+      )}>
+        <div className='flex flex-col gap-1 flex-grow'>
+          <TopKItem
+            className='grow'
+            value={topK}
+            onChange={(_key, v) => onChange({ top_k: v })}
+            enable={true}
+          />
+        </div>
+        <div className='flex flex-col gap-1 flex-grow'>
+          <ScoreThresholdItem
+            className='grow'
+            value={scoreThreshold}
+            onChange={(_key, v) => onChange({ score_threshold: v })}
+            enable={scoreThresholdEnabled}
+            hasSwitch={true}
+            onSwitchChange={(_key, v) => handleScoreThresholdChange(v)}
+          />
+        </div>
+      </div>
+    </div>
+  )
+}
+
+export default RetrievalSettings
diff --git a/web/app/components/datasets/external-knowledge-base/create/declarations.ts b/web/app/components/datasets/external-knowledge-base/create/declarations.ts
new file mode 100644
index 0000000000000..271caf33dfbbe
--- /dev/null
+++ b/web/app/components/datasets/external-knowledge-base/create/declarations.ts
@@ -0,0 +1,12 @@
+export type CreateKnowledgeBaseReq = {
+  name: string
+  description?: string
+  external_knowledge_api_id: string
+  provider: 'external'
+  external_knowledge_id: string
+  external_retrieval_model: {
+    top_k: number
+    score_threshold: number
+    score_threshold_enabled: boolean
+  }
+}
diff --git a/web/app/components/datasets/external-knowledge-base/create/index.tsx b/web/app/components/datasets/external-knowledge-base/create/index.tsx
new file mode 100644
index 0000000000000..921050e218d20
--- /dev/null
+++ b/web/app/components/datasets/external-knowledge-base/create/index.tsx
@@ -0,0 +1,128 @@
+'use client'
+
+import { useCallback, useState } from 'react'
+import { useRouter } from 'next/navigation'
+import { RiArrowLeftLine, RiArrowRightLine } from '@remixicon/react'
+import { useTranslation } from 'react-i18next'
+import KnowledgeBaseInfo from './KnowledgeBaseInfo'
+import ExternalApiSelection from './ExternalApiSelection'
+import RetrievalSettings from './RetrievalSettings'
+import InfoPanel from './InfoPanel'
+import type { CreateKnowledgeBaseReq } from './declarations'
+import Divider from '@/app/components/base/divider'
+import Button from '@/app/components/base/button'
+
+type ExternalKnowledgeBaseCreateProps = {
+  onConnect: (formValue: CreateKnowledgeBaseReq) => void
+  loading: boolean
+}
+
+const ExternalKnowledgeBaseCreate: React.FC<ExternalKnowledgeBaseCreateProps> = ({ onConnect, loading }) => {
+  const { t } = useTranslation()
+  const router = useRouter()
+  const [formData, setFormData] = useState<CreateKnowledgeBaseReq>({
+    name: '',
+    description: '',
+    external_knowledge_api_id: '',
+    external_knowledge_id: '',
+    external_retrieval_model: {
+      top_k: 2,
+      score_threshold: 0.5,
+      score_threshold_enabled: false,
+    },
+    provider: 'external',
+
+  })
+
+  const navBackHandle = useCallback(() => {
+    router.replace('/datasets')
+  }, [router])
+
+  const handleFormChange = (newData: CreateKnowledgeBaseReq) => {
+    setFormData(newData)
+  }
+
+  const isFormValid = formData.name.trim() !== ''
+    && formData.external_knowledge_api_id !== ''
+    && formData.external_knowledge_id !== ''
+    && formData.external_retrieval_model.top_k !== undefined
+    && formData.external_retrieval_model.score_threshold !== undefined
+
+  return (
+    <div className='flex flex-col flex-grow self-stretch rounded-t-2xl border-t border-effects-highlight bg-components-panel-bg'>
+      <div className='flex justify-center flex-grow self-stretch'>
+        <div className='flex w-full max-w-[960px] px-14 py-0 flex-col items-center'>
+          <div className='flex w-full max-w-[640px] pt-6 pb-8 flex-col grow items-center gap-4'>
+            <div className='relative flex flex-col py-2 items-center gap-[2px] self-stretch'>
+              <div className='flex-grow text-text-primary system-xl-semibold self-stretch'>{t('dataset.connectDataset')}</div>
+              <p className='text-text-tertiary system-sm-regular'>
+                <span>{t('dataset.connectHelper.helper1')}</span>
+                <span className='text-text-secondary system-sm-medium'>{t('dataset.connectHelper.helper2')}</span>
+                <span>{t('dataset.connectHelper.helper3')}</span>
+                <a className='self-stretch text-text-accent system-sm-regular' href='https://docs.dify.ai/guides/knowledge-base/connect-external-knowledge' target='_blank' rel="noopener noreferrer">
+                  {t('dataset.connectHelper.helper4')}
+                </a>
+                <span>{t('dataset.connectHelper.helper5')} </span>
+              </p>
+              <Button
+                className='flex w-8 h-8 p-2 items-center justify-center absolute left-[-44px] top-1 rounded-full'
+                variant='tertiary'
+                onClick={navBackHandle}
+              >
+                <RiArrowLeftLine className='w-4 h-4 text-text-tertiary' />
+              </Button>
+            </div>
+            <KnowledgeBaseInfo
+              name={formData.name}
+              description={formData.description ?? ''}
+              onChange={data => handleFormChange({
+                ...formData,
+                ...data,
+              })}
+            />
+            <Divider />
+            <ExternalApiSelection
+              external_knowledge_api_id={formData.external_knowledge_api_id}
+              external_knowledge_id={formData.external_knowledge_id}
+              onChange={data => handleFormChange({
+                ...formData,
+                ...data,
+              })}
+            />
+            <RetrievalSettings
+              topK={formData.external_retrieval_model.top_k}
+              scoreThreshold={formData.external_retrieval_model.score_threshold}
+              scoreThresholdEnabled={formData.external_retrieval_model.score_threshold_enabled}
+              onChange={data => handleFormChange({
+                ...formData,
+                external_retrieval_model: {
+                  ...formData.external_retrieval_model,
+                  ...data,
+                },
+              })}
+            />
+            <div className='flex py-2 justify-end items-center gap-2 self-stretch'>
+              <Button variant='secondary' onClick={navBackHandle}>
+                <div className='text-components-button-secondary-text system-sm-medium'>{t('dataset.externalKnowledgeForm.cancel')}</div>
+              </Button>
+              <Button
+                variant='primary'
+                onClick={() => {
+                  onConnect(formData)
+                }}
+                disabled={!isFormValid}
+                loading={loading}
+              >
+                <div className='text-components-button-primary-text system-sm-medium'>{t('dataset.externalKnowledgeForm.connect')}</div>
+                <RiArrowRightLine className='w-4 h-4 text-components-button-primary-text' />
+              </Button>
+            </div>
+          </div>
+        </div>
+        <InfoPanel />
+      </div>
+    </div>
+  )
+}
+
+export default ExternalKnowledgeBaseCreate
diff --git a/web/app/components/datasets/hit-testing/hit-detail.tsx b/web/app/components/datasets/hit-testing/hit-detail.tsx
index 70e43176d9da2..066e2238c8a0a 100644
--- a/web/app/components/datasets/hit-testing/hit-detail.tsx
+++ b/web/app/components/datasets/hit-testing/hit-detail.tsx
@@ -26,12 +26,15 @@ const HitDetail: FC<IHitDetailProps> = ({ segInfo }) => {
       )
     }
 
-    return segInfo?.content
+    return <div className='mb-4 text-md text-gray-800 h-full'>{segInfo?.content}</div>
   }
 
   return (
-    <div className='overflow-x-auto'>
-      <div className="bg-gray-25 p-6">
+    segInfo?.id === 'external'
+      ? <div className='w-full overflow-x-auto px-2'>
+        <div className={s.segModalContent}>{renderContent()}</div>
+      </div>
+      : <div className='overflow-x-auto'>
         <div className="flex items-center">
           <SegmentIndexTag
             positionId={segInfo?.position || ''}
@@ -59,7 +62,6 @@ const HitDetail: FC<IHitDetailProps> = ({ segInfo }) => {
             })}
         </div>
       </div>
-    </div>
   )
 }
 
diff --git a/web/app/components/datasets/hit-testing/index.tsx b/web/app/components/datasets/hit-testing/index.tsx
index 505cd98fa7de5..ce47f2bfa6c8c 100644
--- a/web/app/components/datasets/hit-testing/index.tsx
+++ b/web/app/components/datasets/hit-testing/index.tsx
@@ -13,7 +13,7 @@ import s from './style.module.css'
 import HitDetail from './hit-detail'
 import ModifyRetrievalModal from './modify-retrieval-modal'
 import cn from '@/utils/classnames'
-import type { HitTestingResponse, HitTesting as HitTestingType } from '@/models/datasets'
+import type { ExternalKnowledgeBaseHitTestingResponse, ExternalKnowledgeBaseHitTesting as ExternalKnowledgeBaseHitTestingType, HitTestingResponse, HitTesting as HitTestingType } from '@/models/datasets'
 import Loading from '@/app/components/base/loading'
 import Modal from '@/app/components/base/modal'
 import Drawer from '@/app/components/base/drawer'
@@ -49,8 +49,10 @@ const HitTesting: FC<Props> = ({ datasetId }: Props) => {
   const isMobile = media === MediaType.mobile
 
   const [hitResult, setHitResult] = useState<HitTestingResponse | undefined>() // 初始化记录为空数组
+  const [externalHitResult, setExternalHitResult] = useState<ExternalKnowledgeBaseHitTestingResponse | undefined>()
   const [submitLoading, setSubmitLoading] = useState(false)
   const [currParagraph, setCurrParagraph] = useState<{ paraInfo?: HitTestingType; showModal: boolean }>({ showModal: false })
+  const [externalCurrParagraph, setExternalCurrParagraph] = useState<{ paraInfo?: ExternalKnowledgeBaseHitTestingType; showModal: boolean }>({ showModal: false })
   const [text, setText] = useState('')
 
   const [currPage, setCurrPage] = React.useState<number>(0)
@@ -66,12 +68,52 @@ const HitTesting: FC<Props> = ({ datasetId }: Props) => {
     setCurrParagraph({ paraInfo: detail, showModal: true })
   }
 
+  const onClickExternalCard = (detail: ExternalKnowledgeBaseHitTestingType) => {
+    setExternalCurrParagraph({ paraInfo: detail, showModal: true })
+  }
   const { dataset: currentDataset } = useContext(DatasetDetailContext)
+  const isExternal = currentDataset?.provider === 'external'
 
   const [retrievalConfig, setRetrievalConfig] = useState(currentDataset?.retrieval_model_dict as RetrievalConfig)
   const [isShowModifyRetrievalModal, setIsShowModifyRetrievalModal] = useState(false)
   const [isShowRightPanel, { setTrue: showRightPanel, setFalse: hideRightPanel, set: setShowRightPanel }] = useBoolean(!isMobile)
 
+  const renderHitResults = (results: any[], onClickCard: (record: any) => void) => (
+    <>
+      <div className='text-gray-600 font-semibold mb-4'>{t('datasetHitTesting.hit.title')}</div>
+      <div className='overflow-auto flex-1'>
+        <div className={s.cardWrapper}>
+          {results.map((record, idx) => (
+            <SegmentCard
+              key={idx}
+              loading={false}
+              refSource= {{
+                title: record.title,
+                uri: record.metadata ? record.metadata['x-amz-bedrock-kb-source-uri'] : '',
+              }}
+              isExternal={isExternal}
+              detail={record.segment}
+              contentExternal={record.content}
+              score={record.score}
+              scene='hitTesting'
+              className='h-[216px] mb-4'
+              onClick={() => onClickCard(record)}
+            />
+          ))}
+        </div>
+      </div>
+    </>
+  )
+
+  const renderEmptyState = () => (
+    <div className='h-full flex flex-col justify-center items-center'>
+      <div className={cn(docStyle.commonIcon, docStyle.targetIcon, '!bg-gray-200 !h-14 !w-14')} />
+      <div className='text-gray-300 text-[13px] mt-3'>
+        {t('datasetHitTesting.hit.emptyTip')}
+      </div>
+    </div>
+  )
+
   useEffect(() => {
     setShowRightPanel(!isMobile)
   }, [isMobile, setShowRightPanel])
@@ -86,12 +128,14 @@ const HitTesting: FC<Props> = ({ datasetId }: Props) => {
         <Textarea
           datasetId={datasetId}
           setHitResult={setHitResult}
+          setExternalHitResult={setExternalHitResult}
           onSubmit={showRightPanel}
           onUpdateList={recordsMutate}
           loading={submitLoading}
           setLoading={setSubmitLoading}
           setText={setText}
           text={text}
+          isExternal={isExternal}
           onClickRetrievalMethod={() => setIsShowModifyRetrievalModal(true)}
           retrievalConfig={retrievalConfig}
           isEconomy={currentDataset?.indexing_technique === 'economy'}
@@ -159,47 +203,42 @@ const HitTesting: FC<Props> = ({ datasetId }: Props) => {
                 className='h-[216px]'
               />
             </div>
-            : !hitResult?.records.length
-              ? (
-                <div className='h-full flex flex-col justify-center items-center'>
-                  <div className={cn(docStyle.commonIcon, docStyle.targetIcon, '!bg-gray-200 !h-14 !w-14')} />
-                  <div className='text-gray-300 text-[13px] mt-3'>
-                    {t('datasetHitTesting.hit.emptyTip')}
-                  </div>
-                </div>
-              )
-              : (
-                <>
-                  <div className='text-gray-600 font-semibold mb-4'>{t('datasetHitTesting.hit.title')}</div>
-                  <div className='overflow-auto flex-1'>
-                    <div className={s.cardWrapper}>
-                      {hitResult?.records.map((record, idx) => {
-                        return <SegmentCard
-                          key={idx}
-                          loading={false}
-                          detail={record.segment as any}
-                          score={record.score}
-                          scene='hitTesting'
-                          className='h-[216px] mb-4'
-                          onClick={() => onClickCard(record as any)}
-                        />
-                      })}
-                    </div>
-                  </div>
-                </>
-              )
+            : (
+              (() => {
+                if (!hitResult?.records.length && !externalHitResult?.records.length)
+                  return renderEmptyState()
+
+                if (hitResult?.records.length)
+                  return renderHitResults(hitResult.records, onClickCard)
+
+                return renderHitResults(externalHitResult?.records || [], onClickExternalCard)
+              })()
+            )
           }
         </div>
       </FloatRightContainer>
       <Modal
-        className='w-[520px] p-0'
+        className={isExternal ? 'py-10 px-8' : 'w-full'}
         closable
-        onClose={() => setCurrParagraph({ showModal: false })}
-        isShow={currParagraph.showModal}
+        onClose={() => {
+          setCurrParagraph({ showModal: false })
+          setExternalCurrParagraph({ showModal: false })
+        }}
+        isShow={currParagraph.showModal || externalCurrParagraph.showModal}
       >
-        {currParagraph.showModal && <HitDetail
-          segInfo={currParagraph.paraInfo?.segment}
-        />}
+        {currParagraph.showModal && (
+          <HitDetail
+            segInfo={currParagraph.paraInfo?.segment}
+          />
+        )}
+        {externalCurrParagraph.showModal && (
+          <HitDetail
+            segInfo={{
+              id: 'external',
+              content: externalCurrParagraph.paraInfo?.content,
+            }}
+          />
+        )}
       </Modal>
       <Drawer isOpen={isShowModifyRetrievalModal} onClose={() => setIsShowModifyRetrievalModal(false)} footer={null} mask={isMobile} panelClassname='mt-16 mx-2 sm:mr-2 mb-3 !p-0 !max-w-[640px] rounded-xl'>
         <ModifyRetrievalModal
diff --git a/web/app/components/datasets/hit-testing/modify-external-retrieval-modal.tsx b/web/app/components/datasets/hit-testing/modify-external-retrieval-modal.tsx
new file mode 100644
index 0000000000000..fd99c698bbe9a
--- /dev/null
+++ b/web/app/components/datasets/hit-testing/modify-external-retrieval-modal.tsx
@@ -0,0 +1,71 @@
+import { useState } from 'react'
+import {
+  RiCloseLine,
+} from '@remixicon/react'
+import { useTranslation } from 'react-i18next'
+import RetrievalSettings from '../external-knowledge-base/create/RetrievalSettings'
+import Button from '@/app/components/base/button'
+import ActionButton from '@/app/components/base/action-button'
+
+type ModifyExternalRetrievalModalProps = {
+  onClose: () => void
+  onSave: (data: { top_k: number; score_threshold: number; score_threshold_enabled: boolean }) => void
+  initialTopK: number
+  initialScoreThreshold: number
+  initialScoreThresholdEnabled: boolean
+}
+
+const ModifyExternalRetrievalModal: React.FC<ModifyExternalRetrievalModalProps> = ({
+  onClose,
+  onSave,
+  initialTopK,
+  initialScoreThreshold,
+  initialScoreThresholdEnabled,
+}) => {
+  const { t } = useTranslation()
+  const [topK, setTopK] = useState(initialTopK)
+  const [scoreThreshold, setScoreThreshold] = useState(initialScoreThreshold)
+  const [scoreThresholdEnabled, setScoreThresholdEnabled] = useState(initialScoreThresholdEnabled)
+
+  const handleSettingsChange = (data: { top_k?: number; score_threshold?: number; score_threshold_enabled?: boolean }) => {
+    if (data.top_k !== undefined)
+      setTopK(data.top_k)
+    if (data.score_threshold !== undefined)
+      setScoreThreshold(data.score_threshold)
+    if (data.score_threshold_enabled !== undefined)
+      setScoreThresholdEnabled(data.score_threshold_enabled)
+  }
+
+  const handleSave = () => {
+    onSave({ top_k: topK, score_threshold: scoreThreshold, score_threshold_enabled: scoreThresholdEnabled })
+    onClose()
+  }
+
+  return (
+    <div className='absolute z-10 top-[36px] right-[14px] flex w-[320px] flex-col items-start rounded-2xl border-[0.5px]
+      border-components-panel-border bg-components-panel-bg shadows-shadow-2xl'
+    >
+      <div className='flex p-4 pb-2 items-center justify-between self-stretch'>
+        <div className='text-text-primary system-xl-semibold flex-grow'>{t('datasetHitTesting.settingTitle')}</div>
+        <ActionButton className='ml-auto' onClick={onClose}>
+          <RiCloseLine className='w-4 h-4 flex-shrink-0' />
+        </ActionButton>
+      </div>
+      <div className='flex p-4 pt-2 flex-col justify-center items-start gap-4 self-stretch'>
+        <RetrievalSettings
+          topK={topK}
+          scoreThreshold={scoreThreshold}
+          scoreThresholdEnabled={scoreThresholdEnabled}
+          onChange={handleSettingsChange}
+          isInHitTesting={true}
+        />
+      </div>
+      <div className='flex p-4 pt-2 justify-end items-end gap-1 w-full'>
+        <Button className='flex-shrink-0 min-w-[72px]' onClick={onClose}>{t('common.operation.cancel')}</Button>
+        <Button variant='primary' className='flex-shrink-0 min-w-[72px]' onClick={handleSave}>{t('common.operation.save')}</Button>
+      </div>
+    </div>
+  )
+}
+
+export default ModifyExternalRetrievalModal
diff --git a/web/app/components/datasets/hit-testing/textarea.tsx b/web/app/components/datasets/hit-testing/textarea.tsx
index 59f346db4a79f..9f8b55f6c6af7 100644
--- a/web/app/components/datasets/hit-testing/textarea.tsx
+++ b/web/app/components/datasets/hit-testing/textarea.tsx
@@ -1,12 +1,17 @@
+import React, { useState } from 'react'
 import { useTranslation } from 'react-i18next'
+import {
+  RiEqualizer2Line,
+} from '@remixicon/react'
 import Button from '../../base/button'
 import Tag from '../../base/tag'
 import { getIcon } from '../common/retrieval-method-info'
 import s from './style.module.css'
+import ModifyExternalRetrievalModal from './modify-external-retrieval-modal'
 import Tooltip from '@/app/components/base/tooltip'
 import cn from '@/utils/classnames'
-import type { HitTestingResponse } from '@/models/datasets'
-import { hitTesting } from '@/service/datasets'
+import type { ExternalKnowledgeBaseHitTestingResponse, HitTestingResponse } from '@/models/datasets'
+import { externalKnowledgeBaseHitTesting, hitTesting } from '@/service/datasets'
 import { asyncRunSafe } from '@/utils'
 import { RETRIEVE_METHOD, type RetrievalConfig } from '@/types/app'
 
@@ -14,10 +19,12 @@ type TextAreaWithButtonIProps = {
   datasetId: string
   onUpdateList: () => void
   setHitResult: (res: HitTestingResponse) => void
+  setExternalHitResult: (res: ExternalKnowledgeBaseHitTestingResponse) => void
   loading: boolean
   setLoading: (v: boolean) => void
   text: string
   setText: (v: string) => void
+  isExternal?: boolean
   onClickRetrievalMethod: () => void
   retrievalConfig: RetrievalConfig
   isEconomy: boolean
@@ -28,16 +35,29 @@ const TextAreaWithButton = ({
   datasetId,
   onUpdateList,
   setHitResult,
+  setExternalHitResult,
   setLoading,
   loading,
   text,
   setText,
+  isExternal = false,
   onClickRetrievalMethod,
   retrievalConfig,
   isEconomy,
   onSubmit: _onSubmit,
 }: TextAreaWithButtonIProps) => {
   const { t } = useTranslation()
+  const [isSettingsOpen, setIsSettingsOpen] = useState(false)
+  const [externalRetrievalSettings, setExternalRetrievalSettings] = useState({
+    top_k: 2,
+    score_threshold: 0.5,
+    score_threshold_enabled: false,
+  })
+
+  const handleSaveExternalRetrievalSettings = (data: { top_k: number; score_threshold: number; score_threshold_enabled: boolean }) => {
+    setExternalRetrievalSettings(data)
+    setIsSettingsOpen(false)
+  }
 
   function handleTextChange(event: any) {
     setText(event.target.value)
@@ -63,28 +83,70 @@ const TextAreaWithButton = ({
     _onSubmit && _onSubmit()
   }
 
+  const externalRetrievalTestingOnSubmit = async () => {
+    const [e, res] = await asyncRunSafe<ExternalKnowledgeBaseHitTestingResponse>(
+      externalKnowledgeBaseHitTesting({
+        datasetId,
+        query: text,
+        external_retrieval_model: {
+          top_k: externalRetrievalSettings.top_k,
+          score_threshold: externalRetrievalSettings.score_threshold,
+          score_threshold_enabled: externalRetrievalSettings.score_threshold_enabled,
+        },
+      }) as Promise<ExternalKnowledgeBaseHitTestingResponse>,
+    )
+    if (!e) {
+      setExternalHitResult(res)
+      onUpdateList?.()
+    }
+    setLoading(false)
+  }
+
   const retrievalMethod = isEconomy ? RETRIEVE_METHOD.invertedIndex : retrievalConfig.search_method
   const Icon = getIcon(retrievalMethod)
   return (
     <>
       <div className={s.wrapper}>
-        <div className='pt-2 rounded-tl-xl rounded-tr-xl bg-[#EEF4FF]'>
+        <div className='relative pt-2 rounded-tl-xl rounded-tr-xl bg-[#EEF4FF]'>
           <div className="px-4 pb-2 flex justify-between h-8 items-center">
             <span className="text-gray-800 font-semibold text-sm">
               {t('datasetHitTesting.input.title')}
             </span>
-            <Tooltip
-              popupContent={t('dataset.retrieval.changeRetrievalMethod')}
-            >
-              <div
-                onClick={onClickRetrievalMethod}
-                className='flex px-2 h-7 items-center space-x-1 bg-white hover:bg-[#ECE9FE] rounded-md shadow-sm cursor-pointer text-[#6927DA]'
+            {isExternal
+              ? <Button
+                variant='secondary'
+                size='small'
+                onClick={() => setIsSettingsOpen(!isSettingsOpen)}
               >
-                <Icon className='w-3.5 h-3.5'></Icon>
-                <div className='text-xs font-medium'>{t(`dataset.retrieval.${retrievalMethod}.title`)}</div>
-              </div>
-            </Tooltip>
+                <RiEqualizer2Line className='text-components-button-secondary-text w-3.5 h-3.5' />
+                <div className='flex px-[3px] justify-center items-center gap-1'>
+                  <span className='text-components-button-secondary-text system-xs-medium'>{t('datasetHitTesting.settingTitle')}</span>
+                </div>
+              </Button>
+              : <Tooltip
+                popupContent={t('dataset.retrieval.changeRetrievalMethod')}
+              >
+                <div
+                  onClick={onClickRetrievalMethod}
+                  className='flex px-2 h-7 items-center space-x-1 bg-white hover:bg-[#ECE9FE] rounded-md shadow-sm cursor-pointer text-[#6927DA]'
+                >
+                  <Icon className='w-3.5 h-3.5'></Icon>
+                  <div className='text-xs font-medium'>{t(`dataset.retrieval.${retrievalMethod}.title`)}</div>
+                </div>
+              </Tooltip>
+            }
           </div>
+          {
+            isSettingsOpen && (
+              <ModifyExternalRetrievalModal
+                onClose={() => setIsSettingsOpen(false)}
+                onSave={handleSaveExternalRetrievalSettings}
+                initialTopK={externalRetrievalSettings.top_k}
+                initialScoreThreshold={externalRetrievalSettings.score_threshold}
+                initialScoreThresholdEnabled={externalRetrievalSettings.score_threshold_enabled}
+              />
+            )
+          }
           <div className='h-2 rounded-tl-xl rounded-tr-xl bg-white'></div>
         </div>
         <div className='px-4 pb-11'>
@@ -122,7 +184,7 @@ const TextAreaWithButton = ({
 
             <div>
               <Button
-                onClick={onSubmit}
+                onClick={isExternal ? externalRetrievalTestingOnSubmit : onSubmit}
                 variant="primary"
                 loading={loading}
                 disabled={(!text?.length || text?.length > 200)}
@@ -132,7 +194,6 @@ const TextAreaWithButton = ({
             </div>
           </div>
         </div>
-
       </div>
     </>
   )
diff --git a/web/app/components/datasets/rename-modal/index.tsx b/web/app/components/datasets/rename-modal/index.tsx
index a7e9e6e335edd..7e5fb228f78f7 100644
--- a/web/app/components/datasets/rename-modal/index.tsx
+++ b/web/app/components/datasets/rename-modal/index.tsx
@@ -26,6 +26,8 @@ const RenameDatasetModal = ({ show, dataset, onSuccess, onClose }: RenameDataset
   const [loading, setLoading] = useState(false)
   const [name, setName] = useState<string>(dataset.name)
   const [description, setDescription] = useState<string>(dataset.description)
+  const [externalKnowledgeId, setExternalKnowledgeId] = useState<string>(dataset.external_knowledge_info.external_knowledge_id)
+  const [externalKnowledgeApiId, setExternalKnowledgeApiId] = useState<string>(dataset.external_knowledge_info.external_knowledge_api_id)
 
   const onConfirm: MouseEventHandler = async () => {
     if (!name.trim()) {
@@ -34,12 +36,17 @@ const RenameDatasetModal = ({ show, dataset, onSuccess, onClose }: RenameDataset
     }
     try {
       setLoading(true)
+      const body: Partial<DataSet> & { external_knowledge_id?: string; external_knowledge_api_id?: string } = {
+        name,
+        description,
+      }
+      if (externalKnowledgeId && externalKnowledgeApiId) {
+        body.external_knowledge_id = externalKnowledgeId
+        body.external_knowledge_api_id = externalKnowledgeApiId
+      }
       await updateDatasetSetting({
         datasetId: dataset.id,
-        body: {
-          name,
-          description,
-        },
+        body,
       })
       notify({ type: 'success', message: t('common.actionMsg.modifiedSuccessfully') })
       if (onSuccess)
diff --git a/web/app/components/datasets/settings/form/index.tsx b/web/app/components/datasets/settings/form/index.tsx
index 15b8abc242e1a..ed0865f87144a 100644
--- a/web/app/components/datasets/settings/form/index.tsx
+++ b/web/app/components/datasets/settings/form/index.tsx
@@ -8,11 +8,14 @@ import { useSWRConfig } from 'swr'
 import { unstable_serialize } from 'swr/infinite'
 import PermissionSelector from '../permission-selector'
 import IndexMethodRadio from '../index-method-radio'
+import RetrievalSettings from '../../external-knowledge-base/create/RetrievalSettings'
 import cn from '@/utils/classnames'
 import RetrievalMethodConfig from '@/app/components/datasets/common/retrieval-method-config'
 import EconomicalRetrievalMethodConfig from '@/app/components/datasets/common/economical-retrieval-method-config'
 import { ToastContext } from '@/app/components/base/toast'
 import Button from '@/app/components/base/button'
+import Divider from '@/app/components/base/divider'
+import { ApiConnectionMod } from '@/app/components/base/icons/src/vender/solid/development'
 import { updateDatasetSetting } from '@/service/datasets'
 import type { DataSetListResponse } from '@/models/datasets'
 import DatasetDetailContext from '@/context/dataset-detail'
@@ -55,6 +58,9 @@ const Form = () => {
   const [name, setName] = useState(currentDataset?.name ?? '')
   const [description, setDescription] = useState(currentDataset?.description ?? '')
   const [permission, setPermission] = useState(currentDataset?.permission)
+  const [topK, setTopK] = useState(currentDataset?.external_retrieval_model.top_k ?? 2)
+  const [scoreThreshold, setScoreThreshold] = useState(currentDataset?.external_retrieval_model.score_threshold ?? 0.5)
+  const [scoreThresholdEnabled, setScoreThresholdEnabled] = useState(currentDataset?.external_retrieval_model.score_threshold_enabled ?? false)
   const [selectedMemberIDs, setSelectedMemberIDs] = useState<string[]>(currentDataset?.partial_member_list || [])
   const [memberList, setMemberList] = useState<Member[]>([])
   const [indexMethod, setIndexMethod] = useState(currentDataset?.indexing_technique)
@@ -85,6 +91,15 @@ const Form = () => {
       setMemberList(accounts)
   }
 
+  const handleSettingsChange = (data: { top_k?: number; score_threshold?: number; score_threshold_enabled?: boolean }) => {
+    if (data.top_k !== undefined)
+      setTopK(data.top_k)
+    if (data.score_threshold !== undefined)
+      setScoreThreshold(data.score_threshold)
+    if (data.score_threshold_enabled !== undefined)
+      setScoreThresholdEnabled(data.score_threshold_enabled)
+  }
+
   useMount(() => {
     getMembers()
   })
@@ -126,10 +141,17 @@ const Form = () => {
           description,
           permission,
           indexing_technique: indexMethod,
+          external_retrieval_model: {
+            top_k: topK,
+            score_threshold: scoreThreshold,
+            score_threshold_enabled: scoreThresholdEnabled,
+          },
           retrieval_model: {
             ...postRetrievalConfig,
             score_threshold: postRetrievalConfig.score_threshold_enabled ? postRetrievalConfig.score_threshold : 0,
           },
+          external_knowledge_id: currentDataset!.external_knowledge_info.external_knowledge_id,
+          external_knowledge_api_id: currentDataset!.external_knowledge_info.external_knowledge_api_id,
           embedding_model: embeddingModel.model,
           embedding_model_provider: embeddingModel.provider,
         },
@@ -161,7 +183,7 @@ const Form = () => {
     <div className='w-full sm:w-[800px] p-4 sm:px-16 sm:py-6'>
       <div className={rowClass}>
         <div className={labelClass}>
-          <div>{t('datasetSettings.form.name')}</div>
+          <div className='text-text-secondary system-sm-semibold'>{t('datasetSettings.form.name')}</div>
         </div>
         <div className='w-full max-w-[480px]'>
           <input
@@ -174,7 +196,7 @@ const Form = () => {
       </div>
       <div className={rowClass}>
         <div className={labelClass}>
-          <div>{t('datasetSettings.form.desc')}</div>
+          <div className='text-text-secondary system-sm-semibold'>{t('datasetSettings.form.desc')}</div>
         </div>
         <div className='w-full max-w-[480px]'>
           <textarea
@@ -192,7 +214,7 @@ const Form = () => {
       </div>
       <div className={rowClass}>
         <div className={labelClass}>
-          <div>{t('datasetSettings.form.permissions')}</div>
+          <div className='text-text-secondary system-sm-semibold'>{t('datasetSettings.form.permissions')}</div>
         </div>
         <div className='w-full sm:w-[480px]'>
           <PermissionSelector
@@ -210,7 +232,7 @@ const Form = () => {
           <div className='w-full h-0 border-b-[0.5px] border-b-gray-200 my-2' />
           <div className={rowClass}>
             <div className={labelClass}>
-              <div>{t('datasetSettings.form.indexMethod')}</div>
+              <div className='text-text-secondary system-sm-semibold'>{t('datasetSettings.form.indexMethod')}</div>
             </div>
             <div className='w-full sm:w-[480px]'>
               <IndexMethodRadio
@@ -225,7 +247,7 @@ const Form = () => {
       {indexMethod === 'high_quality' && (
         <div className={rowClass}>
           <div className={labelClass}>
-            <div>{t('datasetSettings.form.embeddingModel')}</div>
+            <div className='text-text-secondary system-sm-semibold'>{t('datasetSettings.form.embeddingModel')}</div>
           </div>
           <div className='w-[480px]'>
             <ModelSelector
@@ -240,32 +262,76 @@ const Form = () => {
         </div>
       )}
       {/* Retrieval Method Config */}
-      <div className={rowClass}>
-        <div className={labelClass}>
-          <div>
-            <div>{t('datasetSettings.form.retrievalSetting.title')}</div>
-            <div className='leading-[18px] text-xs font-normal text-gray-500'>
-              <a target='_blank' rel='noopener noreferrer' href='https://docs.dify.ai/guides/knowledge-base/create-knowledge-and-upload-documents#id-4-retrieval-settings' className='text-[#155eef]'>{t('datasetSettings.form.retrievalSetting.learnMore')}</a>
-              {t('datasetSettings.form.retrievalSetting.description')}
+      {currentDataset?.provider === 'external'
+        ? <>
+          <div className={rowClass}><Divider/></div>
+          <div className={rowClass}>
+            <div className={labelClass}>
+              <div className='text-text-secondary system-sm-semibold'>{t('datasetSettings.form.retrievalSetting.title')}</div>
             </div>
+            <RetrievalSettings
+              topK={topK}
+              scoreThreshold={scoreThreshold}
+              scoreThresholdEnabled={scoreThresholdEnabled}
+              onChange={handleSettingsChange}
+              isInRetrievalSetting={true}
+            />
+          </div>
+          <div className={rowClass}><Divider/></div>
+          <div className={rowClass}>
+            <div className={labelClass}>
+              <div className='text-text-secondary system-sm-semibold'>{t('datasetSettings.form.externalKnowledgeAPI')}</div>
+            </div>
+            <div className='w-full max-w-[480px]'>
+              <div className='flex h-full px-3 py-2 items-center gap-1 rounded-lg bg-components-input-bg-normal'>
+                <ApiConnectionMod className='w-4 h-4 text-text-secondary' />
+                <div className='overflow-hidden text-text-secondary text-ellipsis system-sm-medium'>
+                  {currentDataset?.external_knowledge_info.external_knowledge_api_name}
+                </div>
+                <div className='text-text-tertiary system-xs-regular'>·</div>
+                <div className='text-text-tertiary system-xs-regular'>{currentDataset?.external_knowledge_info.external_knowledge_api_endpoint}</div>
+              </div>
+            </div>
+          </div>
+          <div className={rowClass}>
+            <div className={labelClass}>
+              <div className='text-text-secondary system-sm-semibold'>{t('datasetSettings.form.externalKnowledgeID')}</div>
+            </div>
+            <div className='w-full max-w-[480px]'>
+              <div className='flex h-full px-3 py-2 items-center gap-1 rounded-lg bg-components-input-bg-normal'>
+                <div className='text-text-tertiary system-xs-regular'>{currentDataset?.external_knowledge_info.external_knowledge_id}</div>
+              </div>
+            </div>
+          </div>
+          <div className={rowClass}><Divider/></div>
+        </>
+        : <div className={rowClass}>
+          <div className={labelClass}>
+            <div>
+              <div className='text-text-secondary system-sm-semibold'>{t('datasetSettings.form.retrievalSetting.title')}</div>
+              <div className='leading-[18px] text-xs font-normal text-gray-500'>
+                <a target='_blank' rel='noopener noreferrer' href='https://docs.dify.ai/guides/knowledge-base/create-knowledge-and-upload-documents#id-4-retrieval-settings' className='text-[#155eef]'>{t('datasetSettings.form.retrievalSetting.learnMore')}</a>
+                {t('datasetSettings.form.retrievalSetting.description')}
+              </div>
+            </div>
+          </div>
+          <div className='w-[480px]'>
+            {indexMethod === 'high_quality'
+              ? (
+                <RetrievalMethodConfig
+                  value={retrievalConfig}
+                  onChange={setRetrievalConfig}
+                />
+              )
+              : (
+                <EconomicalRetrievalMethodConfig
+                  value={retrievalConfig}
+                  onChange={setRetrievalConfig}
+                />
+              )}
           </div>
         </div>
-        <div className='w-[480px]'>
-          {indexMethod === 'high_quality'
-            ? (
-              <RetrievalMethodConfig
-                value={retrievalConfig}
-                onChange={setRetrievalConfig}
-              />
-            )
-            : (
-              <EconomicalRetrievalMethodConfig
-                value={retrievalConfig}
-                onChange={setRetrievalConfig}
-              />
-            )}
-        </div>
-      </div>
+      }
       <div className={rowClass}>
         <div className={labelClass} />
         <div className='w-[480px]'>
diff --git a/web/app/components/header/dataset-nav/index.tsx b/web/app/components/header/dataset-nav/index.tsx
index abf76608a8066..6205468ce458e 100644
--- a/web/app/components/header/dataset-nav/index.tsx
+++ b/web/app/components/header/dataset-nav/index.tsx
@@ -51,7 +51,7 @@ const DatasetNav = () => {
       navs={datasetItems.map(dataset => ({
         id: dataset.id,
         name: dataset.name,
-        link: `/datasets/${dataset.id}/documents`,
+        link: dataset.provider === 'external' ? `/datasets/${dataset.id}/hitTesting` : `/datasets/${dataset.id}/documents`,
         icon: dataset.icon,
         icon_background: dataset.icon_background,
       })) as NavItem[]}
diff --git a/web/app/components/workflow/nodes/knowledge-retrieval/components/dataset-item.tsx b/web/app/components/workflow/nodes/knowledge-retrieval/components/dataset-item.tsx
index 0369322783d9c..caf44d10bcc22 100644
--- a/web/app/components/workflow/nodes/knowledge-retrieval/components/dataset-item.tsx
+++ b/web/app/components/workflow/nodes/knowledge-retrieval/components/dataset-item.tsx
@@ -1,13 +1,15 @@
 'use client'
 import type { FC } from 'react'
-import React, { useCallback } from 'react'
+import React, { useCallback, useState } from 'react'
 import { useBoolean } from 'ahooks'
 import {
   RiDeleteBinLine,
   RiEditLine,
 } from '@remixicon/react'
+import { useTranslation } from 'react-i18next'
 import type { DataSet } from '@/models/datasets'
 import { DataSourceType } from '@/models/datasets'
+import ActionButton, { ActionButtonState } from '@/app/components/base/action-button'
 import FileIcon from '@/app/components/base/file-icon'
 import { Folder } from '@/app/components/base/icons/src/vender/solid/files'
 import SettingsModal from '@/app/components/app/configuration/dataset-config/settings-modal'
@@ -30,8 +32,10 @@ const DatasetItem: FC<Props> = ({
   readonly,
 }) => {
   const media = useBreakpoints()
+  const { t } = useTranslation()
   const isMobile = media === MediaType.mobile
   const { formatIndexingTechniqueAndMethod } = useKnowledge()
+  const [isDeleteHovered, setIsDeleteHovered] = useState(false)
 
   const [isShowSettingsModal, {
     setTrue: showSettingsModal,
@@ -43,8 +47,18 @@ const DatasetItem: FC<Props> = ({
     hideSettingsModal()
   }, [hideSettingsModal, onChange])
 
+  const handleRemove = useCallback((e: React.MouseEvent) => {
+    e.stopPropagation()
+    onRemove()
+  }, [onRemove])
+
   return (
-    <div className='flex items-center h-10 justify-between rounded-xl px-2 bg-white border border-gray-200  cursor-pointer group/dataset-item'>
+    <div className={`flex items-center h-10 justify-between rounded-xl px-2 border-[0.5px] 
+      border-components-panel-border-subtle cursor-pointer group/dataset-item 
+      ${isDeleteHovered
+      ? 'bg-state-destructive-hover border-state-destructive-border'
+      : 'bg-components-panel-on-panel-item-bg hover:bg-components-panel-on-panel-item-bg-hover'
+    }`}>
       <div className='w-0 grow flex items-center space-x-1.5'>
         {
           payload.data_source_type === DataSourceType.NOTION
@@ -61,24 +75,36 @@ const DatasetItem: FC<Props> = ({
       </div>
       {!readonly && (
         <div className='hidden group-hover/dataset-item:flex shrink-0 ml-2  items-center space-x-1'>
-          <div
-            className='flex items-center justify-center w-6 h-6 hover:bg-black/5 rounded-md cursor-pointer'
-            onClick={showSettingsModal}
+          <ActionButton
+            onClick={(e) => {
+              e.stopPropagation()
+              showSettingsModal()
+            }}
           >
-            <RiEditLine className='w-4 h-4 text-gray-500' />
-          </div>
-          <div
-            className='flex items-center justify-center w-6 h-6 hover:bg-black/5 rounded-md cursor-pointer'
-            onClick={onRemove}
+            <RiEditLine className='w-4 h-4 flex-shrink-0 text-text-tertiary' />
+          </ActionButton>
+          <ActionButton
+            onClick={handleRemove}
+            state={ActionButtonState.Destructive}
+            onMouseEnter={() => setIsDeleteHovered(true)}
+            onMouseLeave={() => setIsDeleteHovered(false)}
           >
-            <RiDeleteBinLine className='w-4 h-4 text-gray-500' />
-          </div>
+            <RiDeleteBinLine className={`w-4 h-4 flex-shrink-0 ${isDeleteHovered ? 'text-text-destructive' : 'text-text-tertiary'}`} />
+          </ActionButton>
         </div>
       )}
-      <Badge
-        className='group-hover/dataset-item:hidden shrink-0'
-        text={formatIndexingTechniqueAndMethod(payload.indexing_technique, payload.retrieval_model_dict?.search_method)}
-      />
+      {
+        payload.indexing_technique && <Badge
+          className='group-hover/dataset-item:hidden shrink-0'
+          text={formatIndexingTechniqueAndMethod(payload.indexing_technique, payload.retrieval_model_dict?.search_method)}
+        />
+      }
+      {
+        payload.provider === 'external' && <Badge
+          className='group-hover/dataset-item:hidden shrink-0'
+          text={t('dataset.externalTag')}
+        />
+      }
 
       {isShowSettingsModal && (
         <Drawer isOpen={isShowSettingsModal} onClose={hideSettingsModal} footer={null} mask={isMobile} panelClassname='mt-16 mx-2 sm:mr-2 mb-3 !p-0 !max-w-[640px] rounded-xl'>
diff --git a/web/app/components/workflow/nodes/knowledge-retrieval/use-config.ts b/web/app/components/workflow/nodes/knowledge-retrieval/use-config.ts
index 6d856003d4110..e83a5d97b5afd 100644
--- a/web/app/components/workflow/nodes/knowledge-retrieval/use-config.ts
+++ b/web/app/components/workflow/nodes/knowledge-retrieval/use-config.ts
@@ -207,9 +207,11 @@ const useConfig = (id: string, payload: KnowledgeRetrievalNodeType) => {
 
   const handleOnDatasetsChange = useCallback((newDatasets: DataSet[]) => {
     const {
-      allEconomic,
       mixtureHighQualityAndEconomic,
+      mixtureInternalAndExternal,
       inconsistentEmbeddingModel,
+      allInternal,
+      allExternal,
     } = getSelectedDatasetsMode(newDatasets)
     const newInputs = produce(inputs, (draft) => {
       draft.dataset_ids = newDatasets.map(d => d.id)
@@ -222,7 +224,11 @@ const useConfig = (id: string, payload: KnowledgeRetrievalNodeType) => {
     setInputs(newInputs)
     setSelectedDatasets(newDatasets)
 
-    if (allEconomic || mixtureHighQualityAndEconomic || inconsistentEmbeddingModel)
+    if (
+      (allInternal && (mixtureHighQualityAndEconomic || inconsistentEmbeddingModel))
+      || mixtureInternalAndExternal
+      || (allExternal && newDatasets.length > 1)
+    )
       setRerankModelOpen(true)
   }, [inputs, setInputs, payload.retrieval_mode])
 
diff --git a/web/app/components/workflow/nodes/knowledge-retrieval/utils.ts b/web/app/components/workflow/nodes/knowledge-retrieval/utils.ts
index 89ae9b47640c5..85ae6c4c96bcf 100644
--- a/web/app/components/workflow/nodes/knowledge-retrieval/utils.ts
+++ b/web/app/components/workflow/nodes/knowledge-retrieval/utils.ts
@@ -21,6 +21,9 @@ export const getSelectedDatasetsMode = (datasets: DataSet[]) => {
   let allHighQualityFullTextSearch = true
   let allEconomic = true
   let mixtureHighQualityAndEconomic = true
+  let allExternal = true
+  let allInternal = true
+  let mixtureInternalAndExternal = true
   let inconsistentEmbeddingModel = false
   if (!datasets.length) {
     allHighQuality = false
@@ -29,6 +32,9 @@ export const getSelectedDatasetsMode = (datasets: DataSet[]) => {
     allEconomic = false
     mixtureHighQualityAndEconomic = false
     inconsistentEmbeddingModel = false
+    allExternal = false
+    allInternal = false
+    mixtureInternalAndExternal = false
   }
   datasets.forEach((dataset) => {
     if (dataset.indexing_technique === 'economy') {
@@ -45,8 +51,21 @@ export const getSelectedDatasetsMode = (datasets: DataSet[]) => {
       if (dataset.retrieval_model_dict.search_method !== RETRIEVE_METHOD.fullText)
         allHighQualityFullTextSearch = false
     }
+    if (dataset.provider !== 'external') {
+      allExternal = false
+    }
+    else {
+      allInternal = false
+      allHighQuality = false
+      allHighQualityVectorSearch = false
+      allHighQualityFullTextSearch = false
+      mixtureHighQualityAndEconomic = false
+    }
   })
 
+  if (allExternal || allInternal)
+    mixtureInternalAndExternal = false
+
   if (allHighQuality || allEconomic)
     mixtureHighQualityAndEconomic = false
 
@@ -59,6 +78,9 @@ export const getSelectedDatasetsMode = (datasets: DataSet[]) => {
     allHighQualityFullTextSearch,
     allEconomic,
     mixtureHighQualityAndEconomic,
+    allInternal,
+    allExternal,
+    mixtureInternalAndExternal,
     inconsistentEmbeddingModel,
   } as SelectedDatasetsMode
 }
@@ -70,6 +92,9 @@ export const getMultipleRetrievalConfig = (multipleRetrievalConfig: MultipleRetr
     allHighQualityFullTextSearch,
     allEconomic,
     mixtureHighQualityAndEconomic,
+    allInternal,
+    allExternal,
+    mixtureInternalAndExternal,
     inconsistentEmbeddingModel,
   } = getSelectedDatasetsMode(selectedDatasets)
 
@@ -91,13 +116,13 @@ export const getMultipleRetrievalConfig = (multipleRetrievalConfig: MultipleRetr
     reranking_enable: allEconomic ? reranking_enable : true,
   }
 
-  if (allEconomic || mixtureHighQualityAndEconomic || inconsistentEmbeddingModel)
+  if (allEconomic || mixtureHighQualityAndEconomic || inconsistentEmbeddingModel || allExternal || mixtureInternalAndExternal)
     result.reranking_mode = RerankingModeEnum.RerankingModel
 
-  if (allHighQuality && !inconsistentEmbeddingModel && reranking_mode === undefined)
+  if (allHighQuality && !inconsistentEmbeddingModel && reranking_mode === undefined && allInternal)
     result.reranking_mode = RerankingModeEnum.WeightedScore
 
-  if (allHighQuality && !inconsistentEmbeddingModel && (reranking_mode === RerankingModeEnum.WeightedScore || reranking_mode === undefined) && !weights) {
+  if (allHighQuality && !inconsistentEmbeddingModel && (reranking_mode === RerankingModeEnum.WeightedScore || reranking_mode === undefined) && allInternal && !weights) {
     result.weights = {
       vector_setting: {
         vector_weight: allHighQualityVectorSearch
diff --git a/web/context/external-api-panel-context.tsx b/web/context/external-api-panel-context.tsx
new file mode 100644
index 0000000000000..05ae5c45c1baa
--- /dev/null
+++ b/web/context/external-api-panel-context.tsx
@@ -0,0 +1,28 @@
+'use client'
+
+import React, { createContext, useContext, useState } from 'react'
+
+type ExternalApiPanelContextType = {
+  showExternalApiPanel: boolean
+  setShowExternalApiPanel: (show: boolean) => void
+}
+
+const ExternalApiPanelContext = createContext<ExternalApiPanelContextType | undefined>(undefined)
+
+export const ExternalApiPanelProvider: React.FC<{ children: React.ReactNode }> = ({ children }) => {
+  const [showExternalApiPanel, setShowExternalApiPanel] = useState(false)
+
+  return (
+    <ExternalApiPanelContext.Provider value={{ showExternalApiPanel, setShowExternalApiPanel }}>
+      {children}
+    </ExternalApiPanelContext.Provider>
+  )
+}
+
+export const useExternalApiPanel = () => {
+  const context = useContext(ExternalApiPanelContext)
+  if (context === undefined)
+    throw new Error('useExternalApiPanel must be used within an ExternalApiPanelProvider')
+
+  return context
+}
diff --git a/web/context/external-knowledge-api-context.tsx b/web/context/external-knowledge-api-context.tsx
new file mode 100644
index 0000000000000..5f2d2ff3937b1
--- /dev/null
+++ b/web/context/external-knowledge-api-context.tsx
@@ -0,0 +1,46 @@
+'use client'
+
+import { createContext, useContext, useMemo } from 'react'
+import type { FC, ReactNode } from 'react'
+import useSWR from 'swr'
+import type { ExternalAPIItem, ExternalAPIListResponse } from '@/models/datasets'
+import { fetchExternalAPIList } from '@/service/datasets'
+
+type ExternalKnowledgeApiContextType = {
+  externalKnowledgeApiList: ExternalAPIItem[]
+  mutateExternalKnowledgeApis: () => Promise<ExternalAPIListResponse | undefined>
+  isLoading: boolean
+}
+
+const ExternalKnowledgeApiContext = createContext<ExternalKnowledgeApiContextType | undefined>(undefined)
+
+export type ExternalKnowledgeApiProviderProps = {
+  children: ReactNode
+}
+
+export const ExternalKnowledgeApiProvider: FC<ExternalKnowledgeApiProviderProps> = ({ children }) => {
+  const { data, mutate: mutateExternalKnowledgeApis, isLoading } = useSWR<ExternalAPIListResponse>(
+    { url: '/datasets/external-knowledge-api' },
+    fetchExternalAPIList,
+  )
+
+  const contextValue = useMemo<ExternalKnowledgeApiContextType>(() => ({
+    externalKnowledgeApiList: data?.data || [],
+    mutateExternalKnowledgeApis,
+    isLoading,
+  }), [data, mutateExternalKnowledgeApis, isLoading])
+
+  return (
+    <ExternalKnowledgeApiContext.Provider value={contextValue}>
+      {children}
+    </ExternalKnowledgeApiContext.Provider>
+  )
+}
+
+export const useExternalKnowledgeApi = () => {
+  const context = useContext(ExternalKnowledgeApiContext)
+  if (context === undefined)
+    throw new Error('useExternalKnowledgeApi must be used within a ExternalKnowledgeApiProvider')
+
+  return context
+}
diff --git a/web/context/modal-context.tsx b/web/context/modal-context.tsx
index 3547c45aacff8..727268a29a5ca 100644
--- a/web/context/modal-context.tsx
+++ b/web/context/modal-context.tsx
@@ -10,6 +10,7 @@ import ModerationSettingModal from '@/app/components/app/configuration/toolbox/m
 import ExternalDataToolModal from '@/app/components/app/configuration/tools/external-data-tool-modal'
 import AnnotationFullModal from '@/app/components/billing/annotation-full/modal'
 import ModelModal from '@/app/components/header/account-setting/model-provider-page/model-modal'
+import ExternalAPIModal from '@/app/components/datasets/external-api/external-api-modal'
 import type {
   ConfigurationMethodEnum,
   CustomConfigurationModelFixedFields,
@@ -23,6 +24,7 @@ import type {
   ApiBasedExtension,
   ExternalDataTool,
 } from '@/models/common'
+import type { CreateExternalAPIReq } from '@/app/components/datasets/external-api/declarations'
 import ModelLoadBalancingEntryModal from '@/app/components/header/account-setting/model-provider-page/model-modal/model-load-balancing-entry-modal'
 import type { ModelLoadBalancingModalProps } from '@/app/components/header/account-setting/model-provider-page/provider-added-card/model-load-balancing-modal'
 import ModelLoadBalancingModal from '@/app/components/header/account-setting/model-provider-page/provider-added-card/model-load-balancing-modal'
@@ -32,7 +34,10 @@ export type ModalState<T> = {
   onCancelCallback?: () => void
   onSaveCallback?: (newPayload: T) => void
   onRemoveCallback?: (newPayload: T) => void
+  onEditCallback?: (newPayload: T) => void
   onValidateBeforeSaveCallback?: (newPayload: T) => boolean
+  isEditMode?: boolean
+  datasetBindings?: { id: string; name: string }[]
 }
 
 export type ModelModalType = {
@@ -52,6 +57,7 @@ export type ModalContextState = {
   setShowPricingModal: () => void
   setShowAnnotationFullModal: () => void
   setShowModelModal: Dispatch<SetStateAction<ModalState<ModelModalType> | null>>
+  setShowExternalKnowledgeAPIModal: Dispatch<SetStateAction<ModalState<CreateExternalAPIReq> | null>>
   setShowModelLoadBalancingModal: Dispatch<SetStateAction<ModelLoadBalancingModalProps | null>>
   setShowModelLoadBalancingEntryModal: Dispatch<SetStateAction<ModalState<LoadBalancingEntryModalType> | null>>
 }
@@ -63,6 +69,7 @@ const ModalContext = createContext<ModalContextState>({
   setShowPricingModal: () => { },
   setShowAnnotationFullModal: () => { },
   setShowModelModal: () => { },
+  setShowExternalKnowledgeAPIModal: () => { },
   setShowModelLoadBalancingModal: () => { },
   setShowModelLoadBalancingEntryModal: () => { },
 })
@@ -86,6 +93,7 @@ export const ModalContextProvider = ({
   const [showModerationSettingModal, setShowModerationSettingModal] = useState<ModalState<ModerationConfig> | null>(null)
   const [showExternalDataToolModal, setShowExternalDataToolModal] = useState<ModalState<ExternalDataTool> | null>(null)
   const [showModelModal, setShowModelModal] = useState<ModalState<ModelModalType> | null>(null)
+  const [showExternalKnowledgeAPIModal, setShowExternalKnowledgeAPIModal] = useState<ModalState<CreateExternalAPIReq> | null>(null)
   const [showModelLoadBalancingModal, setShowModelLoadBalancingModal] = useState<ModelLoadBalancingModalProps | null>(null)
   const [showModelLoadBalancingEntryModal, setShowModelLoadBalancingEntryModal] = useState<ModalState<LoadBalancingEntryModalType> | null>(null)
   const searchParams = useSearchParams()
@@ -122,6 +130,24 @@ export const ModalContextProvider = ({
     setShowModelModal(null)
   }, [showModelModal])
 
+  const handleCancelExternalApiModal = useCallback(() => {
+    setShowExternalKnowledgeAPIModal(null)
+    if (showExternalKnowledgeAPIModal?.onCancelCallback)
+      showExternalKnowledgeAPIModal.onCancelCallback()
+  }, [showExternalKnowledgeAPIModal])
+
+  const handleSaveExternalApiModal = useCallback(async (updatedFormValue: CreateExternalAPIReq) => {
+    if (showExternalKnowledgeAPIModal?.onSaveCallback)
+      showExternalKnowledgeAPIModal.onSaveCallback(updatedFormValue)
+    setShowExternalKnowledgeAPIModal(null)
+  }, [showExternalKnowledgeAPIModal])
+
+  const handleEditExternalApiModal = useCallback(async (updatedFormValue: CreateExternalAPIReq) => {
+    if (showExternalKnowledgeAPIModal?.onEditCallback)
+      showExternalKnowledgeAPIModal.onEditCallback(updatedFormValue)
+    setShowExternalKnowledgeAPIModal(null)
+  }, [showExternalKnowledgeAPIModal])
+
   const handleCancelModelLoadBalancingEntryModal = useCallback(() => {
     showModelLoadBalancingEntryModal?.onCancelCallback?.()
     setShowModelLoadBalancingEntryModal(null)
@@ -173,6 +199,7 @@ export const ModalContextProvider = ({
       setShowPricingModal: () => setShowPricingModal(true),
       setShowAnnotationFullModal: () => setShowAnnotationFullModal(true),
       setShowModelModal,
+      setShowExternalKnowledgeAPIModal,
       setShowModelLoadBalancingModal,
       setShowModelLoadBalancingEntryModal,
     }}>
@@ -245,6 +272,18 @@ export const ModalContextProvider = ({
             />
           )
         }
+        {
+          !!showExternalKnowledgeAPIModal && (
+            <ExternalAPIModal
+              data={showExternalKnowledgeAPIModal.payload}
+              datasetBindings={showExternalKnowledgeAPIModal.datasetBindings ?? []}
+              onSave={handleSaveExternalApiModal}
+              onCancel={handleCancelExternalApiModal}
+              onEdit={handleEditExternalApiModal}
+              isEditMode={showExternalKnowledgeAPIModal.isEditMode ?? false}
+            />
+          )
+        }
         {
           Boolean(showModelLoadBalancingModal) && (
             <ModelLoadBalancingModal {...showModelLoadBalancingModal!} />
diff --git a/web/i18n/en-US/dataset-hit-testing.ts b/web/i18n/en-US/dataset-hit-testing.ts
index 4f34391ca048a..6dbfa47fee7b6 100644
--- a/web/i18n/en-US/dataset-hit-testing.ts
+++ b/web/i18n/en-US/dataset-hit-testing.ts
@@ -1,6 +1,7 @@
 const translation = {
   title: 'Retrieval Testing',
-  desc: 'Test the hitting effect of the Knowledge based on the given query text.',
+  settingTitle: 'Retrieval Setting',
+  desc: 'Test the hitting effect of the Knowledge based on the given query text',
   dateTimeFormat: 'MM/DD/YYYY hh:mm A',
   recents: 'Recents',
   table: {
@@ -23,6 +24,7 @@ const translation = {
   },
   noRecentTip: 'No recent query results here',
   viewChart: 'View VECTOR CHART',
+  viewDetail: 'View Detail',
 }
 
 export default translation
diff --git a/web/i18n/en-US/dataset-settings.ts b/web/i18n/en-US/dataset-settings.ts
index 8ca24596f8c6a..aa435c8e0e8ec 100644
--- a/web/i18n/en-US/dataset-settings.ts
+++ b/web/i18n/en-US/dataset-settings.ts
@@ -1,13 +1,13 @@
 const translation = {
   title: 'Knowledge settings',
-  desc: 'Here you can modify the properties and working methods of the Knowledge.',
+  desc: 'Here you can modify the properties and retrieval settings of this Knowledge.',
   form: {
     name: 'Knowledge Name',
     namePlaceholder: 'Please enter the Knowledge name',
     nameError: 'Name cannot be empty',
-    desc: 'Knowledge description',
+    desc: 'Knowledge Description',
     descInfo: 'Please write a clear textual description to outline the content of the Knowledge. This description will be used as a basis for matching when selecting from multiple Knowledge for inference.',
-    descPlaceholder: 'Describe what is in this Knowledge. A detailed description allows AI to access the content of the Knowledge in a timely manner. If empty, Dify will use the default hit strategy.',
+    descPlaceholder: 'Describe what\'s in this Knowledge (optional)',
     descWrite: 'Learn how to write a good Knowledge description.',
     permissions: 'Permissions',
     permissionsOnlyMe: 'Only me',
@@ -23,11 +23,14 @@ const translation = {
     embeddingModelTip: 'Change the embedded model, please go to ',
     embeddingModelTipLink: 'Settings',
     retrievalSetting: {
-      title: 'Retrieval setting',
+      title: 'Retrieval Setting',
       learnMore: 'Learn more',
       description: ' about retrieval method.',
       longDescription: ' about retrieval method, you can change this at any time in the Knowledge settings.',
     },
+    externalKnowledgeAPI: 'External Knowledge API',
+    externalKnowledgeID: 'External Knowledge ID',
+    retrievalSettings: 'Retrieval Settings',
     save: 'Save',
   },
 }
diff --git a/web/i18n/en-US/dataset.ts b/web/i18n/en-US/dataset.ts
index a15efe5dc0753..e89ea47c262d6 100644
--- a/web/i18n/en-US/dataset.ts
+++ b/web/i18n/en-US/dataset.ts
@@ -1,9 +1,65 @@
 const translation = {
   knowledge: 'Knowledge',
+  externalTag: 'External',
+  externalAPI: 'External API',
+  externalAPIPanelTitle: 'External Knowledge API',
+  externalKnowledgeId: 'External Knowledge ID',
+  externalKnowledgeName: 'External Knowledge Name',
+  externalKnowledgeDescription: 'Knowledge Description',
+  externalKnowledgeIdPlaceholder: 'Please enter the Knowledge ID',
+  externalKnowledgeNamePlaceholder: 'Please enter the name of the knowledge base',
+  externalKnowledgeDescriptionPlaceholder: 'Describe what\'s in this Knowledge Base (optional)',
+  learnHowToWriteGoodKnowledgeDescription: 'Learn how to write a good knowledge description',
+  externalAPIPanelDescription: 'The external knowledge API is used to connect to a knowledge base outside of Dify and retrieve knowledge from that knowledge base.',
+  externalAPIPanelDocumentation: 'Learn how to create an External Knowledge API',
   documentCount: ' docs',
   wordCount: ' k words',
   appCount: ' linked apps',
   createDataset: 'Create Knowledge',
+  createNewExternalAPI: 'Create a new External Knowledge API',
+  noExternalKnowledge: 'There is no External Knowledge API yet, click here to create',
+  createExternalAPI: 'Add an External Knowledge API',
+  editExternalAPIFormTitle: 'Edit the External Knowledge API',
+  editExternalAPITooltipTitle: 'LINKED KNOWLEDGE',
+  editExternalAPIConfirmWarningContent: {
+    front: 'This External Knowledge API is linked to',
+    end: 'external knowledge, and this modification will be applied to all of them. Are you sure you want to save this change?',
+  },
+  editExternalAPIFormWarning: {
+    front: 'This External API is linked to',
+    end: 'external knowledge',
+  },
+  deleteExternalAPIConfirmWarningContent: {
+    title: {
+      front: 'Delete',
+      end: '?',
+    },
+    content: {
+      front: 'This External Knowledge API is linked to',
+      end: 'external knowledge. Deleting this API will invalidate all of them. Are you sure you want to delete this API?',
+    },
+    noConnectionContent: 'Are you sure to delete this API?',
+  },
+  selectExternalKnowledgeAPI: {
+    placeholder: 'Choose an External Knowledge API',
+  },
+  connectDataset: 'Connect to an External Knowledge Base',
+  connectDatasetIntro: {
+    title: 'How to Connect to an External Knowledge Base',
+    content: {
+      front: 'To connect to an external knowledge base, you need to create an external API first. Please read carefully and refer to',
+      link: 'Learn how to create an external API',
+      end: '. Then find the corresponding knowledge ID and fill it in the form on the left. If all the information is correct, it will automatically jump to the retrieval test in the knowledge base after clicking the connect button.',
+    },
+    learnMore: 'Learn More',
+  },
+  connectHelper: {
+    helper1: 'Connect to external knowledge bases via API and knowledge base ID. Currently, ',
+    helper2: 'only the retrieval functionality is supported',
+    helper3: '. We strongly recommend that you ',
+    helper4: 'read the help documentation',
+    helper5: ' carefully before using this feature.',
+  },
   createDatasetIntro: 'Import your own text data or write data in real-time via Webhook for LLM context enhancement.',
   deleteDatasetConfirmTitle: 'Delete this Knowledge?',
   deleteDatasetConfirmContent:
@@ -22,6 +78,22 @@ const translation = {
   unavailableTip: 'Embedding model is not available, the default embedding model needs to be configured',
   datasets: 'KNOWLEDGE',
   datasetsApi: 'API ACCESS',
+  externalKnowledgeForm: {
+    connect: 'Connect',
+    cancel: 'Cancel',
+  },
+  externalAPIForm: {
+    name: 'Name',
+    endpoint: 'API Endpoint',
+    apiKey: 'API Key',
+    save: 'Save',
+    cancel: 'Cancel',
+    edit: 'Edit',
+    encrypted: {
+      front: 'Your API Token will be encrypted and stored using',
+      end: 'technology.',
+    },
+  },
   retrieval: {
     semantic_search: {
       title: 'Vector Search',
@@ -58,6 +130,8 @@ const translation = {
   defaultRetrievalTip: 'Multi-path retrieval is used by default. Knowledge is retrieved from multiple knowledge bases and then re-ranked.',
   mixtureHighQualityAndEconomicTip: 'The Rerank model is required for mixture of high quality and economical knowledge bases.',
   inconsistentEmbeddingModelTip: 'The Rerank model is required if the Embedding models of the selected knowledge bases are inconsistent.',
+  mixtureInternalAndExternalTip: 'The Rerank model is required for mixture of internal and  external knowledge.',
+  allExternalTip: 'When using external knowledge only, the user can choose whether to enable the Rerank model. If not enabled, retrieved chunks will be sorted based on scores. When the retrieval strategies of different knowledge bases are inconsistent, it will be inaccurate.',
   retrievalSettings: 'Retrieval Setting',
   rerankSettings: 'Rerank Setting',
   weightedScore: {
diff --git a/web/i18n/zh-Hans/dataset-hit-testing.ts b/web/i18n/zh-Hans/dataset-hit-testing.ts
index a30f794d4745f..09cfdc2824ce6 100644
--- a/web/i18n/zh-Hans/dataset-hit-testing.ts
+++ b/web/i18n/zh-Hans/dataset-hit-testing.ts
@@ -1,6 +1,7 @@
 const translation = {
   title: '召回测试',
-  desc: '基于给定的查询文本测试知识库的召回效果。',
+  settingTitle: '召回设置',
+  desc: '基于给定的查询文本测试知识库的召回效果',
   dateTimeFormat: 'YYYY-MM-DD HH:mm',
   recents: '最近查询',
   table: {
@@ -23,6 +24,7 @@ const translation = {
   },
   noRecentTip: '最近无查询结果',
   viewChart: '查看向量图表',
+  viewDetail: '查看详情',
 }
 
 export default translation
diff --git a/web/i18n/zh-Hans/dataset-settings.ts b/web/i18n/zh-Hans/dataset-settings.ts
index 8fd1cc0971903..f0f5190412013 100644
--- a/web/i18n/zh-Hans/dataset-settings.ts
+++ b/web/i18n/zh-Hans/dataset-settings.ts
@@ -1,13 +1,13 @@
 const translation = {
   title: '知识库设置',
-  desc: '在这里您可以修改知识库的工作方式以及其它设置。',
+  desc: '在这里，您可以修改此知识库的属性和检索设置',
   form: {
     name: '知识库名称',
     namePlaceholder: '请输入知识库名称',
     nameError: '名称不能为空',
     desc: '知识库描述',
     descInfo: '请写出清楚的文字描述来概述知识库的内容。当从多个知识库中进行选择匹配时，该描述将用作匹配的基础。',
-    descPlaceholder: '描述这个知识库中的内容。详细的描述可以让 AI 及时访问知识库的内容。如果为空，Dify 将使用默认的命中策略。',
+    descPlaceholder: '请描述这个知识库包含的内容（可选）',
     descWrite: '了解如何编写更好的知识库描述。',
     permissions: '可见权限',
     permissionsOnlyMe: '只有我',
@@ -28,6 +28,8 @@ const translation = {
       description: '关于检索方法。',
       longDescription: '关于检索方法，您可以随时在知识库设置中更改此设置。',
     },
+    externalKnowledgeAPI: '外部知识 API',
+    externalKnowledgeID: '外部知识库 ID',
     save: '保存',
   },
 }
diff --git a/web/i18n/zh-Hans/dataset.ts b/web/i18n/zh-Hans/dataset.ts
index 013830af6fce6..d057af0c16437 100644
--- a/web/i18n/zh-Hans/dataset.ts
+++ b/web/i18n/zh-Hans/dataset.ts
@@ -1,9 +1,62 @@
 const translation = {
   knowledge: '知识库',
+  externalTag: '外部',
+  externalAPI: '外部 API',
+  externalAPIPanelTitle: '外部知识库 API',
+  externalKnowledgeId: '外部知识库 ID',
+  externalKnowledgeName: '外部知识库名称',
+  externalKnowledgeDescription: '知识库描述',
+  externalKnowledgeIdPlaceholder: '请输入外部知识库 ID',
+  externalKnowledgeNamePlaceholder: '请输入外部知识库名称',
+  externalKnowledgeDescriptionPlaceholder: '描述知识库内容（可选）',
+  learnHowToWriteGoodKnowledgeDescription: '了解如何编写良好的知识库描述',
+  externalAPIPanelDescription: '外部知识库 API 用于连接到 Dify 之外的知识库并从中检索知识。',
+  externalAPIPanelDocumentation: '了解如何创建外部知识库 API',
   documentCount: ' 文档',
   wordCount: ' 千字符',
   appCount: ' 关联应用',
   createDataset: '创建知识库',
+  noExternalKnowledge: '还没有外部知识库 API，点击此处创建',
+  createExternalAPI: '添加外部知识库 API',
+  createNewExternalAPI: '创建新的外部知识库API',
+  editExternalAPIFormTitle: '编辑外部知识库 API',
+  editExternalAPITooltipTitle: '个关联知识库',
+  editExternalAPIConfirmWarningContent: {
+    front: '此外部知识库 API 已链接到',
+    end: '个外部知识库，此修改将应用于所有这些知识库。您确定要保存此更改吗？',
+  },
+  editExternalAPIFormWarning: {
+    front: '此外部 API 已链接到',
+    end: '外部知识库',
+  },
+  deleteExternalAPIConfirmWarningContent: {
+    title: {
+      front: '删除',
+      end: '？',
+    },
+    content: {
+      front: '此外部知识库 API 已链接到',
+      end: '个外部知识库。删除此 API 将使所有这些知识库失效。您确定要删除此 API 吗？',
+    },
+    noConnectionContent: '您确定要删除此 API 吗？',
+  },
+  connectDatasetIntro: {
+    title: '如何连接到外部知识库',
+    content: {
+      front: '要连接到外部知识库，您需要先创建一个外部 API。请仔细阅读并参考',
+      link: '了解如何创建外部 API',
+      end: '。然后找到相应的知识库 ID 并填写在左侧表单中。如果所有信息正确，点击连接按钮后将自动跳转到知识库中的检索测试。',
+    },
+    learnMore: '了解更多',
+  },
+  connectHelper: {
+    helper1: '通过 API 和知识库 ID 连接到外部知识库。目前，',
+    helper2: '仅支持检索功能',
+    helper3: '。我们强烈建议您在使用此功能之前',
+    helper4: '仔细阅读帮助文档',
+    helper5: '。',
+  },
+  connectDataset: '连接外部知识库',
   createDatasetIntro: '导入您自己的文本数据或通过 Webhook 实时写入数据以增强 LLM 的上下文。',
   deleteDatasetConfirmTitle: '要删除知识库吗？',
   deleteDatasetConfirmContent:
@@ -11,6 +64,9 @@ const translation = {
   datasetUsedByApp: '某些应用正在使用该知识库。应用将无法再使用该知识库,所有的提示配置和日志将被永久删除。',
   datasetDeleted: '知识库已删除',
   datasetDeleteFailed: '删除知识库失败',
+  selectExternalKnowledgeAPI: {
+    placeholder: '选择一个外部知识 API',
+  },
   didYouKnow: '你知道吗？',
   intro1: '知识库可以被集成到 Dify 应用中',
   intro2: '作为上下文',
@@ -22,6 +78,22 @@ const translation = {
   unavailableTip: '由于 embedding 模型不可用，需要配置默认 embedding 模型',
   datasets: '知识库',
   datasetsApi: 'API',
+  externalKnowledgeForm: {
+    connect: '连接',
+    cancel: '取消',
+  },
+  externalAPIForm: {
+    name: '名称',
+    endpoint: 'API 端点',
+    apiKey: 'API 密钥',
+    save: '保存',
+    cancel: '取消',
+    edit: '编辑',
+    encrypted: {
+      front: '您的 API Token 将使用',
+      end: '加密并存储。',
+    },
+  },
   retrieval: {
     semantic_search: {
       title: '向量检索',
@@ -58,6 +130,8 @@ const translation = {
   defaultRetrievalTip: '默认情况下使用多路召回。从多个知识库中检索知识，然后重新排序。',
   mixtureHighQualityAndEconomicTip: '混合使用高质量和经济型知识库需要配置 Rerank 模型。',
   inconsistentEmbeddingModelTip: '当所选知识库配置的 Embedding 模型不一致时，需要配置 Rerank 模型。',
+  mixtureInternalAndExternalTip: '混合使用内部和外部知识时需要配置 Rerank 模型。',
+  allExternalTip: '仅使用外部知识时，用户可以选择是否启用 Rerank 模型。如果不启用，检索到的文本块将根据分数排序。当不同知识库的检索策略不一致时，结果可能不准确。',
   retrievalSettings: '召回设置',
   rerankSettings: 'Rerank 设置',
   weightedScore: {
diff --git a/web/i18n/zh-Hant/dataset-hit-testing.ts b/web/i18n/zh-Hant/dataset-hit-testing.ts
index a1e84112efd13..aed73497de5ad 100644
--- a/web/i18n/zh-Hant/dataset-hit-testing.ts
+++ b/web/i18n/zh-Hant/dataset-hit-testing.ts
@@ -23,6 +23,7 @@ const translation = {
   },
   noRecentTip: '最近無查詢結果',
   viewChart: '查看向量圖表',
+  viewDetail: '查看詳情',
 }
 
 export default translation
diff --git a/web/models/datasets.ts b/web/models/datasets.ts
index 9358f6fcb9fb4..81a750968b00f 100644
--- a/web/models/datasets.ts
+++ b/web/models/datasets.ts
@@ -25,6 +25,7 @@ export type DataSet = {
   app_count: number
   document_count: number
   word_count: number
+  provider: string
   embedding_model: string
   embedding_model_provider: string
   embedding_available: boolean
@@ -32,6 +33,58 @@ export type DataSet = {
   retrieval_model: RetrievalConfig
   tags: Tag[]
   partial_member_list?: any[]
+  external_knowledge_info: {
+    external_knowledge_id: string
+    external_knowledge_api_id: string
+    external_knowledge_api_name: string
+    external_knowledge_api_endpoint: string
+  }
+  external_retrieval_model: {
+    top_k: number
+    score_threshold: number
+    score_threshold_enabled: boolean
+  }
+}
+
+export type ExternalAPIItem = {
+  id: string
+  tenant_id: string
+  name: string
+  description: string
+  settings: {
+    endpoint: string
+    api_key: string
+  }
+  dataset_bindings: { id: string; name: string }[]
+  created_by: string
+  created_at: string
+}
+
+export type ExternalKnowledgeItem = {
+  id: string
+  name: string
+  description: string | null
+  provider: 'external'
+  permission: DatasetPermission
+  data_source_type: null
+  indexing_technique: null
+  app_count: number
+  document_count: number
+  word_count: number
+  created_by: string
+  created_at: string
+  updated_by: string
+  updated_at: string
+  tags: Tag[]
+}
+
+export type ExternalAPIDeleteResponse = {
+  result: 'success' | 'error'
+}
+
+export type ExternalAPIUsage = {
+  is_using: boolean
+  count: number
 }
 
 export type CustomFile = File & {
@@ -73,6 +126,14 @@ export type DataSetListResponse = {
   total: number
 }
 
+export type ExternalAPIListResponse = {
+  data: ExternalAPIItem[]
+  has_more: boolean
+  limit: number
+  page: number
+  total: number
+}
+
 export type QA = {
   question: string
   answer: string
@@ -385,6 +446,16 @@ export type HitTesting = {
   tsne_position: TsnePosition
 }
 
+export type ExternalKnowledgeBaseHitTesting = {
+  content: string
+  title: string
+  score: number
+  metadata: {
+    'x-amz-bedrock-kb-source-uri': string
+    'x-amz-bedrock-kb-data-source-id': string
+  }
+}
+
 export type Segment = {
   id: string
   document: Document
@@ -425,6 +496,13 @@ export type HitTestingResponse = {
   records: Array<HitTesting>
 }
 
+export type ExternalKnowledgeBaseHitTestingResponse = {
+  query: {
+    content: string
+  }
+  records: Array<ExternalKnowledgeBaseHitTesting>
+}
+
 export type RelatedApp = {
   id: string
   name: string
@@ -462,6 +540,9 @@ export type SelectedDatasetsMode = {
   allHighQualityFullTextSearch: boolean
   allEconomic: boolean
   mixtureHighQualityAndEconomic: boolean
+  allInternal: boolean
+  allExternal: boolean
+  mixtureInternalAndExternal: boolean
   inconsistentEmbeddingModel: boolean
 }
 
diff --git a/web/service/datasets.ts b/web/service/datasets.ts
index 689ac7c4035f1..b3bb19b9f3781 100644
--- a/web/service/datasets.ts
+++ b/web/service/datasets.ts
@@ -8,6 +8,12 @@ import type {
   DocumentDetailResponse,
   DocumentListResponse,
   ErrorDocsResponse,
+  ExternalAPIDeleteResponse,
+  ExternalAPIItem,
+  ExternalAPIListResponse,
+  ExternalAPIUsage,
+  ExternalKnowledgeBaseHitTestingResponse,
+  ExternalKnowledgeItem,
   FileIndexingEstimateResponse,
   HitTestingRecordsResponse,
   HitTestingResponse,
@@ -23,7 +29,9 @@ import type {
   SegmentsResponse,
   createDocumentResponse,
 } from '@/models/datasets'
-import { type CommonResponse, type DataSourceNotionWorkspace, DataSourceProvider } from '@/models/common'
+import type { CreateKnowledgeBaseReq } from '@/app/components/datasets/external-knowledge-base/create/declarations'
+import type { CreateExternalAPIReq } from '@/app/components/datasets/external-api/declarations.ts'
+import type { CommonResponse, DataSourceNotionWorkspace } from '@/models/common'
 import type {
   ApiKeysListResponse,
   CreateApiKeyResponse,
@@ -82,6 +90,34 @@ export const deleteDataset: Fetcher<DataSet, string> = (datasetID) => {
   return del<DataSet>(`/datasets/${datasetID}`)
 }
 
+export const fetchExternalAPIList: Fetcher<ExternalAPIListResponse, { url: string }> = ({ url }) => {
+  return get<ExternalAPIListResponse>(url)
+}
+
+export const fetchExternalAPI: Fetcher<ExternalAPIItem, { apiTemplateId: string }> = ({ apiTemplateId }) => {
+  return get<ExternalAPIItem>(`/datasets/external-knowledge-api/${apiTemplateId}`)
+}
+
+export const updateExternalAPI: Fetcher<ExternalAPIItem, { apiTemplateId: string; body: ExternalAPIItem }> = ({ apiTemplateId, body }) => {
+  return patch<ExternalAPIItem>(`/datasets/external-knowledge-api/${apiTemplateId}`, { body })
+}
+
+export const deleteExternalAPI: Fetcher<ExternalAPIDeleteResponse, { apiTemplateId: string }> = ({ apiTemplateId }) => {
+  return del<ExternalAPIDeleteResponse>(`/datasets/external-knowledge-api/${apiTemplateId}`)
+}
+
+export const checkUsageExternalAPI: Fetcher<ExternalAPIUsage, { apiTemplateId: string }> = ({ apiTemplateId }) => {
+  return get<ExternalAPIUsage>(`/datasets/external-knowledge-api/${apiTemplateId}/use-check`)
+}
+
+export const createExternalAPI: Fetcher<ExternalAPIItem, { body: CreateExternalAPIReq }> = ({ body }) => {
+  return post<ExternalAPIItem>('/datasets/external-knowledge-api', { body })
+}
+
+export const createExternalKnowledgeBase: Fetcher<ExternalKnowledgeItem, { body: CreateKnowledgeBaseReq }> = ({ body }) => {
+  return post<ExternalKnowledgeItem>('/datasets/external', { body })
+}
+
 export const fetchDefaultProcessRule: Fetcher<ProcessRuleResponse, { url: string }> = ({ url }) => {
   return get<ProcessRuleResponse>(url)
 }
@@ -209,6 +245,10 @@ export const hitTesting: Fetcher<HitTestingResponse, { datasetId: string; queryT
   return post<HitTestingResponse>(`/datasets/${datasetId}/hit-testing`, { body: { query: queryText, retrieval_model } })
 }
 
+export const externalKnowledgeBaseHitTesting: Fetcher<ExternalKnowledgeBaseHitTestingResponse, { datasetId: string; query: string; external_retrieval_model: { top_k: number; score_threshold: number; score_threshold_enabled: boolean } }> = ({ datasetId, query, external_retrieval_model }) => {
+  return post<ExternalKnowledgeBaseHitTestingResponse>(`/datasets/${datasetId}/external-hit-testing`, { body: { query, external_retrieval_model } })
+}
+
 export const fetchTestingRecords: Fetcher<HitTestingRecordsResponse, { datasetId: string; params: { page: number; limit: number } }> = ({ datasetId, params }) => {
   return get<HitTestingRecordsResponse>(`/datasets/${datasetId}/queries`, { params })
 }

From 13046709a93494c3e4b273c60cedc1cd2fdb8f18 Mon Sep 17 00:00:00 2001
From: Yi Xiao <54782454+YIXIAO0@users.noreply.github.com>
Date: Mon, 30 Sep 2024 16:04:51 +0800
Subject: [PATCH 114/235] fix: line in iteration node is not straight (#8918)

---
 web/app/components/workflow/nodes/iteration-start/index.tsx | 2 +-
 1 file changed, 1 insertion(+), 1 deletion(-)

diff --git a/web/app/components/workflow/nodes/iteration-start/index.tsx b/web/app/components/workflow/nodes/iteration-start/index.tsx
index 9d7ac1f905008..a765f827331a3 100644
--- a/web/app/components/workflow/nodes/iteration-start/index.tsx
+++ b/web/app/components/workflow/nodes/iteration-start/index.tsx
@@ -9,7 +9,7 @@ const IterationStartNode = ({ id, data }: NodeProps) => {
   const { t } = useTranslation()
 
   return (
-    <div className='group flex nodrag items-center justify-center w-11 h-11 rounded-2xl border border-workflow-block-border bg-white'>
+    <div className='group flex nodrag items-center justify-center w-11 h-11 mt-1 rounded-2xl border border-workflow-block-border bg-white'>
       <Tooltip popupContent={t('workflow.blocks.iteration-start')} asChild={false}>
         <div className='flex items-center justify-center w-6 h-6 rounded-full border-[0.5px] border-components-panel-border-subtle bg-util-colors-blue-brand-blue-brand-500'>
           <RiHome5Fill className='w-3 h-3 text-text-primary-on-surface' />

From 4585cffce1af694639662cff9436e7225acfdcae Mon Sep 17 00:00:00 2001
From: Aurelius Huang <aureliusshu@gmail.com>
Date: Mon, 30 Sep 2024 16:32:23 +0800
Subject: [PATCH 115/235] fix: Compatible with special characters in pg
 full-text search. (#8921)

Co-authored-by: Aurelius Huang <cm.huang@aftership.com>
---
 api/core/rag/datasource/vdb/pgvector/pgvector.py | 2 +-
 1 file changed, 1 insertion(+), 1 deletion(-)

diff --git a/api/core/rag/datasource/vdb/pgvector/pgvector.py b/api/core/rag/datasource/vdb/pgvector/pgvector.py
index d90707ebcd5cb..25a10a1e48090 100644
--- a/api/core/rag/datasource/vdb/pgvector/pgvector.py
+++ b/api/core/rag/datasource/vdb/pgvector/pgvector.py
@@ -166,7 +166,7 @@ def search_by_full_text(self, query: str, **kwargs: Any) -> list[Document]:
 
         with self._get_cursor() as cur:
             cur.execute(
-                f"""SELECT meta, text, ts_rank(to_tsvector(coalesce(text, '')), to_tsquery(%s)) AS score
+                f"""SELECT meta, text, ts_rank(to_tsvector(coalesce(text, '')), plainto_tsquery(%s)) AS score
                 FROM {self.table_name}
                 WHERE to_tsvector(text) @@ plainto_tsquery(%s)
                 ORDER BY score DESC

From 824a71388a655529a5679653dadebb3117b0e126 Mon Sep 17 00:00:00 2001
From: "github-actions[bot]"
 <41898282+github-actions[bot]@users.noreply.github.com>
Date: Mon, 30 Sep 2024 16:35:00 +0800
Subject: [PATCH 116/235] chore: translate i18n files (#8917)

Co-authored-by: JohnJyong <76649700+JohnJyong@users.noreply.github.com>
Co-authored-by: crazywoola <100913391+crazywoola@users.noreply.github.com>
---
 web/i18n/de-DE/app-api.ts               |  1 +
 web/i18n/de-DE/dataset-creation.ts      | 13 +++++
 web/i18n/de-DE/dataset-hit-testing.ts   |  2 +
 web/i18n/de-DE/dataset-settings.ts      |  3 +
 web/i18n/de-DE/dataset.ts               | 74 +++++++++++++++++++++++++
 web/i18n/es-ES/app-api.ts               |  1 +
 web/i18n/es-ES/dataset-creation.ts      | 13 +++++
 web/i18n/es-ES/dataset-hit-testing.ts   |  2 +
 web/i18n/es-ES/dataset-settings.ts      |  3 +
 web/i18n/es-ES/dataset.ts               | 74 +++++++++++++++++++++++++
 web/i18n/fa-IR/app-api.ts               |  1 +
 web/i18n/fa-IR/dataset-creation.ts      | 13 +++++
 web/i18n/fa-IR/dataset-hit-testing.ts   |  2 +
 web/i18n/fa-IR/dataset-settings.ts      |  3 +
 web/i18n/fa-IR/dataset.ts               | 74 +++++++++++++++++++++++++
 web/i18n/fr-FR/app-api.ts               |  1 +
 web/i18n/fr-FR/dataset-creation.ts      | 13 +++++
 web/i18n/fr-FR/dataset-hit-testing.ts   |  2 +
 web/i18n/fr-FR/dataset-settings.ts      |  3 +
 web/i18n/fr-FR/dataset.ts               | 74 +++++++++++++++++++++++++
 web/i18n/hi-IN/app-api.ts               |  1 +
 web/i18n/hi-IN/dataset-creation.ts      | 13 +++++
 web/i18n/hi-IN/dataset-hit-testing.ts   |  2 +
 web/i18n/hi-IN/dataset-settings.ts      |  3 +
 web/i18n/hi-IN/dataset.ts               | 74 +++++++++++++++++++++++++
 web/i18n/it-IT/app-api.ts               |  1 +
 web/i18n/it-IT/dataset-creation.ts      | 13 +++++
 web/i18n/it-IT/dataset-hit-testing.ts   |  2 +
 web/i18n/it-IT/dataset-settings.ts      |  3 +
 web/i18n/it-IT/dataset.ts               | 74 +++++++++++++++++++++++++
 web/i18n/ja-JP/app-api.ts               |  1 +
 web/i18n/ja-JP/dataset-creation.ts      | 13 +++++
 web/i18n/ja-JP/dataset-hit-testing.ts   |  2 +
 web/i18n/ja-JP/dataset-settings.ts      |  3 +
 web/i18n/ja-JP/dataset.ts               | 74 +++++++++++++++++++++++++
 web/i18n/ko-KR/app-api.ts               |  1 +
 web/i18n/ko-KR/dataset-creation.ts      | 13 +++++
 web/i18n/ko-KR/dataset-hit-testing.ts   |  2 +
 web/i18n/ko-KR/dataset-settings.ts      |  3 +
 web/i18n/ko-KR/dataset.ts               | 74 +++++++++++++++++++++++++
 web/i18n/pl-PL/app-api.ts               |  1 +
 web/i18n/pl-PL/dataset-creation.ts      | 13 +++++
 web/i18n/pl-PL/dataset-hit-testing.ts   |  2 +
 web/i18n/pl-PL/dataset-settings.ts      |  3 +
 web/i18n/pl-PL/dataset.ts               | 74 +++++++++++++++++++++++++
 web/i18n/pt-BR/app-api.ts               |  1 +
 web/i18n/pt-BR/dataset-creation.ts      | 13 +++++
 web/i18n/pt-BR/dataset-hit-testing.ts   |  2 +
 web/i18n/pt-BR/dataset-settings.ts      |  3 +
 web/i18n/pt-BR/dataset.ts               | 74 +++++++++++++++++++++++++
 web/i18n/ro-RO/app-api.ts               |  1 +
 web/i18n/ro-RO/dataset-creation.ts      | 13 +++++
 web/i18n/ro-RO/dataset-hit-testing.ts   |  2 +
 web/i18n/ro-RO/dataset-settings.ts      |  3 +
 web/i18n/ro-RO/dataset.ts               | 74 +++++++++++++++++++++++++
 web/i18n/ru-RU/app-api.ts               |  1 +
 web/i18n/ru-RU/dataset-creation.ts      | 13 +++++
 web/i18n/ru-RU/dataset-hit-testing.ts   |  2 +
 web/i18n/ru-RU/dataset-settings.ts      |  3 +
 web/i18n/ru-RU/dataset.ts               | 74 +++++++++++++++++++++++++
 web/i18n/tr-TR/app-api.ts               |  1 +
 web/i18n/tr-TR/dataset-creation.ts      | 13 +++++
 web/i18n/tr-TR/dataset-hit-testing.ts   |  2 +
 web/i18n/tr-TR/dataset-settings.ts      |  3 +
 web/i18n/tr-TR/dataset.ts               | 74 +++++++++++++++++++++++++
 web/i18n/uk-UA/app-api.ts               |  1 +
 web/i18n/uk-UA/dataset-creation.ts      | 13 +++++
 web/i18n/uk-UA/dataset-hit-testing.ts   |  2 +
 web/i18n/uk-UA/dataset-settings.ts      |  3 +
 web/i18n/uk-UA/dataset.ts               | 74 +++++++++++++++++++++++++
 web/i18n/vi-VN/app-api.ts               |  1 +
 web/i18n/vi-VN/dataset-creation.ts      | 13 +++++
 web/i18n/vi-VN/dataset-hit-testing.ts   |  2 +
 web/i18n/vi-VN/dataset-settings.ts      |  3 +
 web/i18n/vi-VN/dataset.ts               | 74 +++++++++++++++++++++++++
 web/i18n/zh-Hans/dataset-settings.ts    |  1 +
 web/i18n/zh-Hant/app-api.ts             |  1 +
 web/i18n/zh-Hant/dataset-creation.ts    | 13 +++++
 web/i18n/zh-Hant/dataset-hit-testing.ts |  1 +
 web/i18n/zh-Hant/dataset-settings.ts    |  3 +
 web/i18n/zh-Hant/dataset.ts             | 74 +++++++++++++++++++++++++
 81 files changed, 1488 insertions(+)

diff --git a/web/i18n/de-DE/app-api.ts b/web/i18n/de-DE/app-api.ts
index 84e0989095d8b..4b27322170ede 100644
--- a/web/i18n/de-DE/app-api.ts
+++ b/web/i18n/de-DE/app-api.ts
@@ -78,6 +78,7 @@ const translation = {
     query: 'Anfrage',
   },
   loading: 'Laden',
+  regenerate: 'Erneuern',
 }
 
 export default translation
diff --git a/web/i18n/de-DE/dataset-creation.ts b/web/i18n/de-DE/dataset-creation.ts
index 5251cc94e846a..334f6d43bdd7b 100644
--- a/web/i18n/de-DE/dataset-creation.ts
+++ b/web/i18n/de-DE/dataset-creation.ts
@@ -73,6 +73,14 @@ const translation = {
       maxDepthTooltip: 'Maximale Tiefe für das Crawlen relativ zur eingegebenen URL. Tiefe 0 kratzt nur die Seite der eingegebenen URL, Tiefe 1 kratzt die URL und alles nach der eingegebenen URL + ein / und so weiter.',
       crawlSubPage: 'Unterseiten crawlen',
       scrapTimeInfo: 'Insgesamt {{{total}} Seiten innerhalb von {{time}}s gescrapt',
+      jinaReaderDocLink: 'https://jina.ai/reader',
+      jinaReaderTitle: 'Konvertieren Sie die gesamte Website in Markdown',
+      useSitemap: 'Sitemap verwenden',
+      chooseProvider: 'Wählen Sie einen Anbieter',
+      jinaReaderNotConfigured: 'Jina Reader ist nicht konfiguriert',
+      jinaReaderNotConfiguredDescription: 'Richten Sie Jina Reader ein, indem Sie Ihren kostenlosen API-Schlüssel für den Zugriff eingeben.',
+      useSitemapTooltip: 'Folgen Sie der Sitemap, um die Website zu crawlen. Ist dies nicht der Fall, crawlt Jina Reader iterativ basierend auf der Seitenrelevanz, sodass weniger, aber qualitativ hochwertigere Seiten angezeigt werden.',
+      jinaReaderDoc: 'Erfahre mehr über Jina Reader',
     },
   },
   stepTwo: {
@@ -157,6 +165,11 @@ const translation = {
     configFirecrawl: 'Konfigurieren von 🔥Firecrawl',
     getApiKeyLinkText: 'Holen Sie sich Ihren API-Schlüssel von firecrawl.dev',
   },
+  jinaReader: {
+    configJinaReader: 'Jina Reader konfigurieren',
+    apiKeyPlaceholder: 'API-Schlüssel von jina.ai',
+    getApiKeyLinkText: 'Holen Sie sich Ihren kostenlosen API-Schlüssel bei jina.ai',
+  },
 }
 
 export default translation
diff --git a/web/i18n/de-DE/dataset-hit-testing.ts b/web/i18n/de-DE/dataset-hit-testing.ts
index baf88016b3c87..eb9e85238cd7d 100644
--- a/web/i18n/de-DE/dataset-hit-testing.ts
+++ b/web/i18n/de-DE/dataset-hit-testing.ts
@@ -23,6 +23,8 @@ const translation = {
   },
   noRecentTip: 'Keine kürzlichen Abfrageergebnisse hier',
   viewChart: 'VEKTORDIAGRAMM ansehen',
+  viewDetail: 'Im Detail sehen',
+  settingTitle: 'Einstellung für den Abruf',
 }
 
 export default translation
diff --git a/web/i18n/de-DE/dataset-settings.ts b/web/i18n/de-DE/dataset-settings.ts
index b29e7780750a7..7c07f5b6e4b88 100644
--- a/web/i18n/de-DE/dataset-settings.ts
+++ b/web/i18n/de-DE/dataset-settings.ts
@@ -29,6 +29,9 @@ const translation = {
     save: 'Speichern',
     permissionsInvitedMembers: 'Teilweise Teammitglieder',
     me: '(Sie)',
+    externalKnowledgeID: 'ID für externes Wissen',
+    externalKnowledgeAPI: 'API für externes Wissen',
+    retrievalSettings: 'Einstellungen für den Abruf',
   },
 }
 
diff --git a/web/i18n/de-DE/dataset.ts b/web/i18n/de-DE/dataset.ts
index 6462e56dedad5..86ac623a6f129 100644
--- a/web/i18n/de-DE/dataset.ts
+++ b/web/i18n/de-DE/dataset.ts
@@ -72,6 +72,80 @@ const translation = {
   nTo1RetrievalLegacyLink: 'Mehr erfahren',
   nTo1RetrievalLegacyLinkText: 'N-zu-1-Abruf wird im September offiziell eingestellt.',
   defaultRetrievalTip: 'Standardmäßig wird der Multi-Path-Abruf verwendet. Das Wissen wird aus mehreren Wissensdatenbanken abgerufen und dann neu eingestuft.',
+  editExternalAPIConfirmWarningContent: {
+    end: 'externes Wissen, und diese Modifikation wird auf alle angewendet. Sind Sie sicher, dass Sie diese Änderung speichern möchten?',
+    front: 'Diese External Knowledge API ist verknüpft mit',
+  },
+  editExternalAPIFormWarning: {
+    front: 'Diese externe API ist verknüpft mit',
+    end: 'externes Wissen',
+  },
+  deleteExternalAPIConfirmWarningContent: {
+    title: {
+      front: 'Löschen',
+      end: '?',
+    },
+    content: {
+      front: 'Diese External Knowledge API ist verknüpft mit',
+      end: 'externes Wissen. Wenn Sie diese API löschen, werden alle ungültig. Sind Sie sicher, dass Sie diese API löschen möchten?',
+    },
+    noConnectionContent: 'Sind Sie sicher, dass Sie diese API löschen möchten?',
+  },
+  selectExternalKnowledgeAPI: {
+    placeholder: 'Auswählen einer externen Wissens-API',
+  },
+  connectDatasetIntro: {
+    content: {
+      front: 'Um eine Verbindung zu einer externen Wissensdatenbank herzustellen, müssen Sie zuerst eine externe API erstellen. Bitte lesen Sie diese sorgfältig durch und beziehen Sie sich auf',
+      link: 'Erfahren Sie, wie Sie eine externe API erstellen',
+      end: '. Suchen Sie dann die entsprechende Wissens-ID und füllen Sie diese in das Formular links aus. Wenn alle Informationen korrekt sind, wird nach dem Klicken auf die Schaltfläche "Verbinden" automatisch zum Abruftest in der Wissensdatenbank gesprungen.',
+    },
+    learnMore: 'Weitere Informationen',
+    title: 'So stellen Sie eine Verbindung zu einer externen Wissensdatenbank her',
+  },
+  connectHelper: {
+    helper3: '. Wir empfehlen Ihnen dringend,',
+    helper2: 'Es wird nur die Retrieval-Funktionalität unterstützt',
+    helper5: 'bevor Sie diese Funktion verwenden.',
+    helper4: 'Lesen Sie die Hilfedokumentation',
+    helper1: 'Verbinden Sie sich mit externen Wissensdatenbanken über API und Wissensdatenbank-ID.',
+  },
+  externalKnowledgeForm: {
+    connect: 'Verbinden',
+    cancel: 'Abbrechen',
+  },
+  externalAPIForm: {
+    encrypted: {
+      front: 'Ihr API-Token wird verschlüsselt und gespeichert mit',
+      end: 'Technologie.',
+    },
+    save: 'Retten',
+    cancel: 'Abbrechen',
+    endpoint: 'API-Endpunkt',
+    name: 'Name',
+    edit: 'Redigieren',
+    apiKey: 'API-Schlüssel',
+  },
+  externalTag: 'Äußerlich',
+  createExternalAPI: 'Hinzufügen einer externen Knowledge-API',
+  externalAPIPanelDescription: 'Die API für externes Wissen wird verwendet, um eine Verbindung zu einer Wissensdatenbank außerhalb von Dify herzustellen und Wissen aus dieser Wissensdatenbank abzurufen.',
+  createNewExternalAPI: 'Erstellen einer neuen API für externes Wissen',
+  externalKnowledgeDescriptionPlaceholder: 'Beschreiben Sie, was in dieser Wissensdatenbank enthalten ist (optional)',
+  externalAPIPanelDocumentation: 'Erfahren Sie, wie Sie eine API für externes Wissen erstellen',
+  externalAPIPanelTitle: 'API für externes Wissen',
+  learnHowToWriteGoodKnowledgeDescription: 'Erfahren Sie, wie Sie eine gute Wissensbeschreibung schreiben',
+  editExternalAPITooltipTitle: 'VERKNÜPFTES WISSEN',
+  externalKnowledgeIdPlaceholder: 'Bitte geben Sie die Knowledge ID ein',
+  connectDataset: 'Herstellen einer Verbindung mit einer externen Wissensdatenbank',
+  externalAPI: 'Externe API',
+  externalKnowledgeName: 'Name des externen Wissens',
+  allExternalTip: 'Wenn nur externes Wissen verwendet wird, kann der Benutzer auswählen, ob das Rerank-Modell aktiviert werden soll. Wenn diese Option nicht aktiviert ist, werden die abgerufenen Blöcke basierend auf den Punktzahlen sortiert. Wenn die Abrufstrategien verschiedener Wissensdatenbanken inkonsistent sind, ist dies ungenau.',
+  externalKnowledgeDescription: 'Wissen Beschreibung',
+  noExternalKnowledge: 'Es gibt noch keine External Knowledge API, klicken Sie hier, um zu erstellen',
+  externalKnowledgeNamePlaceholder: 'Bitte geben Sie den Namen der Wissensdatenbank ein.',
+  mixtureInternalAndExternalTip: 'Das Rerank-Modell ist für die Mischung von internem und externem Wissen erforderlich.',
+  externalKnowledgeId: 'ID für externes Wissen',
+  editExternalAPIFormTitle: 'Bearbeiten der API für externes Wissen',
 }
 
 export default translation
diff --git a/web/i18n/es-ES/app-api.ts b/web/i18n/es-ES/app-api.ts
index 1474e5fb1d592..9bfb488236876 100644
--- a/web/i18n/es-ES/app-api.ts
+++ b/web/i18n/es-ES/app-api.ts
@@ -78,6 +78,7 @@ const translation = {
     pathParams: 'Parámetros de ruta',
     query: 'Consulta',
   },
+  regenerate: 'Regenerar',
 }
 
 export default translation
diff --git a/web/i18n/es-ES/dataset-creation.ts b/web/i18n/es-ES/dataset-creation.ts
index e093632bc4fc2..913c230b2bc5b 100644
--- a/web/i18n/es-ES/dataset-creation.ts
+++ b/web/i18n/es-ES/dataset-creation.ts
@@ -78,6 +78,14 @@ const translation = {
       scrapTimeInfo: 'Se extrajeron {{total}} páginas en total en {{time}}s',
       preview: 'Vista previa',
       maxDepthTooltip: 'Profundidad máxima para rastrear en relación con la URL ingresada. La profundidad 0 solo extrae la página de la URL ingresada, la profundidad 1 extrae la URL y todo lo después de la URL ingresada + una /, y así sucesivamente.',
+      jinaReaderDocLink: 'https://jina.ai/reader',
+      jinaReaderNotConfigured: 'Jina Reader no está configurado',
+      useSitemap: 'Usar el mapa del sitio',
+      jinaReaderTitle: 'Convertir todo el sitio a Markdown',
+      jinaReaderNotConfiguredDescription: 'Configura Jina Reader introduciendo tu clave API gratuita para acceder.',
+      useSitemapTooltip: 'Siga el mapa del sitio para rastrear el sitio. De lo contrario, Jina Reader rastreará de forma iterativa en función de la relevancia de la página, lo que producirá menos páginas pero de mayor calidad.',
+      chooseProvider: 'Seleccione un proveedor',
+      jinaReaderDoc: 'Más información sobre Jina Reader',
     },
   },
   stepTwo: {
@@ -157,6 +165,11 @@ const translation = {
     modelButtonConfirm: 'Confirmar',
     modelButtonCancel: 'Cancelar',
   },
+  jinaReader: {
+    configJinaReader: 'Configurar Jina Reader',
+    apiKeyPlaceholder: 'Clave de API de jina.ai',
+    getApiKeyLinkText: 'Obtén tu clave API gratuita en jina.ai',
+  },
 }
 
 export default translation
diff --git a/web/i18n/es-ES/dataset-hit-testing.ts b/web/i18n/es-ES/dataset-hit-testing.ts
index 4ebdd03b9d7e4..29e2af1a64727 100644
--- a/web/i18n/es-ES/dataset-hit-testing.ts
+++ b/web/i18n/es-ES/dataset-hit-testing.ts
@@ -23,6 +23,8 @@ const translation = {
   },
   noRecentTip: 'No hay resultados de consulta recientes aquí',
   viewChart: 'Ver GRÁFICO VECTORIAL',
+  viewDetail: 'Ver Detalle',
+  settingTitle: 'Configuración de recuperación',
 }
 
 export default translation
diff --git a/web/i18n/es-ES/dataset-settings.ts b/web/i18n/es-ES/dataset-settings.ts
index 984b3783769a8..87f5da10108b5 100644
--- a/web/i18n/es-ES/dataset-settings.ts
+++ b/web/i18n/es-ES/dataset-settings.ts
@@ -29,6 +29,9 @@ const translation = {
       longDescription: ' sobre el método de recuperación, puedes cambiar esto en cualquier momento en la configuración del conjunto de datos.',
     },
     save: 'Guardar',
+    retrievalSettings: 'Configuración de recuperación',
+    externalKnowledgeID: 'ID de conocimiento externo',
+    externalKnowledgeAPI: 'API de conocimiento externo',
   },
 }
 
diff --git a/web/i18n/es-ES/dataset.ts b/web/i18n/es-ES/dataset.ts
index 4eefb621d2c96..4133eb59f8397 100644
--- a/web/i18n/es-ES/dataset.ts
+++ b/web/i18n/es-ES/dataset.ts
@@ -72,6 +72,80 @@ const translation = {
   nTo1RetrievalLegacyLink: 'Más información',
   nTo1RetrievalLegacyLinkText: 'La recuperación N-a-1 será oficialmente obsoleta en septiembre.',
   defaultRetrievalTip: 'De forma predeterminada, se utiliza la recuperación de varias rutas. El conocimiento se recupera de múltiples bases de conocimiento y luego se vuelve a clasificar.',
+  editExternalAPIConfirmWarningContent: {
+    front: 'Esta API de conocimiento externo está vinculada a',
+    end: 'conocimiento externo, y esta modificación se aplicará a todos ellos. ¿Estás seguro de que quieres guardar este cambio?',
+  },
+  editExternalAPIFormWarning: {
+    end: 'Conocimiento externo',
+    front: 'Esta API externa está vinculada a',
+  },
+  deleteExternalAPIConfirmWarningContent: {
+    title: {
+      end: '?',
+      front: 'Borrar',
+    },
+    content: {
+      end: 'conocimiento externo. Al eliminar esta API, se invalidarán todos ellos. ¿Estás seguro de que quieres eliminar esta API?',
+      front: 'Esta API de conocimiento externo está vinculada a',
+    },
+    noConnectionContent: '¿Está seguro de eliminar esta API?',
+  },
+  selectExternalKnowledgeAPI: {
+    placeholder: 'Elegir una API de conocimiento externa',
+  },
+  connectDatasetIntro: {
+    content: {
+      link: 'Más información sobre cómo crear una API externa',
+      front: 'Para conectarse a una base de conocimientos externa, primero debe crear una API externa. Por favor, lea atentamente y consulte',
+      end: '. A continuación, busque el ID de conocimiento correspondiente y rellénelo en el formulario de la izquierda. Si toda la información es correcta, saltará automáticamente a la prueba de recuperación en la base de conocimientos después de hacer clic en el botón conectar.',
+    },
+    learnMore: 'Aprende más',
+    title: 'Cómo conectarse a una base de conocimientos externa',
+  },
+  connectHelper: {
+    helper5: 'con cuidado antes de usar esta función.',
+    helper2: 'Solo se admite la funcionalidad de recuperación',
+    helper1: 'Conéctese a bases de conocimiento externas a través de la API y el ID de la base de conocimiento. Actualmente,',
+    helper3: '. Le recomendamos encarecidamente que',
+    helper4: 'Leer la documentación de ayuda',
+  },
+  externalKnowledgeForm: {
+    connect: 'Conectar',
+    cancel: 'Cancelar',
+  },
+  externalAPIForm: {
+    encrypted: {
+      front: 'Su token de API se cifrará y almacenará mediante',
+      end: 'Tecnología.',
+    },
+    cancel: 'Cancelar',
+    apiKey: 'Clave de API',
+    save: 'Salvar',
+    edit: 'Editar',
+    name: 'Nombre',
+    endpoint: 'Punto de conexión de API',
+  },
+  externalTag: 'Externo',
+  externalKnowledgeDescriptionPlaceholder: 'Describa lo que hay en esta base de conocimientos (opcional)',
+  externalKnowledgeNamePlaceholder: 'Introduzca el nombre de la base de conocimientos',
+  noExternalKnowledge: 'Todavía no hay una API de conocimiento externo, haga clic aquí para crear',
+  editExternalAPIFormTitle: 'Editar la API de conocimiento externo',
+  externalKnowledgeName: 'Nombre del conocimiento externo',
+  allExternalTip: 'Al usar solo conocimiento externo, el usuario puede elegir si desea habilitar el modelo Rerank. Si no se habilita, los fragmentos recuperados se ordenarán en función de las puntuaciones. Cuando las estrategias de recuperación de diferentes bases de conocimiento son inconsistentes, serán inexactas.',
+  createExternalAPI: 'Adición de una API de conocimiento externa',
+  externalKnowledgeId: 'ID de conocimiento externo',
+  connectDataset: 'Conéctese a una base de conocimientos externa',
+  createNewExternalAPI: 'Creación de una nueva API de conocimiento externo',
+  editExternalAPITooltipTitle: 'CONOCIMIENTO VINCULADO',
+  externalAPIPanelTitle: 'API de conocimiento externo',
+  externalKnowledgeDescription: 'Descripción del conocimiento',
+  externalAPIPanelDescription: 'La API de conocimiento externo se utiliza para conectarse a una base de conocimiento fuera de Dify y recuperar conocimiento de esa base de conocimiento.',
+  externalAPI: 'API externa',
+  externalKnowledgeIdPlaceholder: 'Introduzca el ID de conocimiento',
+  learnHowToWriteGoodKnowledgeDescription: 'Aprende a escribir una buena descripción del conocimiento',
+  externalAPIPanelDocumentation: 'Más información sobre cómo crear una API de conocimiento externo',
+  mixtureInternalAndExternalTip: 'El modelo de Rerank es necesario para la mezcla de conocimiento interno y externo.',
 }
 
 export default translation
diff --git a/web/i18n/fa-IR/app-api.ts b/web/i18n/fa-IR/app-api.ts
index 7f65481fcff36..e9c23b5483868 100644
--- a/web/i18n/fa-IR/app-api.ts
+++ b/web/i18n/fa-IR/app-api.ts
@@ -78,6 +78,7 @@ const translation = {
     pathParams: 'پارامترهای مسیر',
     query: 'پرس‌وجو',
   },
+  regenerate: 'بازسازی',
 }
 
 export default translation
diff --git a/web/i18n/fa-IR/dataset-creation.ts b/web/i18n/fa-IR/dataset-creation.ts
index 40dae4b02d826..48882f45cf18c 100644
--- a/web/i18n/fa-IR/dataset-creation.ts
+++ b/web/i18n/fa-IR/dataset-creation.ts
@@ -78,6 +78,14 @@ const translation = {
       scrapTimeInfo: 'در مجموع {{total}} صفحه در {{time}} ثانیه خراشیده شد',
       preview: 'پیشنمایش',
       maxDepthTooltip: 'حداکثر عمق برای خزش نسبت به URL وارد شده. عمق 0 فقط صفحه URL وارد شده را خراش میدهد، عمق 1 URL و همه چیز بعد از URL وارد شده + یک / را خراش میدهد، و غیره.',
+      jinaReaderDocLink: 'https://jina.ai/reader',
+      chooseProvider: 'یک ارائه دهنده را انتخاب کنید',
+      jinaReaderTitle: 'کل سایت را به Markdown تبدیل کنید',
+      jinaReaderNotConfigured: 'Jina Reader پیکربندی نشده است',
+      jinaReaderDoc: 'درباره Jina Reader بیشتر بدانید',
+      useSitemap: 'از نقشه سایت استفاده کنید',
+      jinaReaderNotConfiguredDescription: 'با وارد کردن کلید API رایگان خود برای دسترسی، Jina Reader را راه اندازی کنید.',
+      useSitemapTooltip: 'نقشه سایت را دنبال کنید تا سایت را بخزید. در غیر این صورت، Jina Reader بر اساس ارتباط صفحه به صورت تکراری می خزد و صفحات کمتر اما با کیفیت بالاتر را به دست می آورد.',
     },
   },
   stepTwo: {
@@ -157,6 +165,11 @@ const translation = {
     modelButtonConfirm: 'تأیید',
     modelButtonCancel: 'لغو',
   },
+  jinaReader: {
+    configJinaReader: 'پیکربندی Jina Reader',
+    apiKeyPlaceholder: 'کلید API از jina.ai',
+    getApiKeyLinkText: 'کلید API رایگان خود را در jina.ai دریافت کنید',
+  },
 }
 
 export default translation
diff --git a/web/i18n/fa-IR/dataset-hit-testing.ts b/web/i18n/fa-IR/dataset-hit-testing.ts
index 75b97c7abc0ec..8c78a1a21a5f9 100644
--- a/web/i18n/fa-IR/dataset-hit-testing.ts
+++ b/web/i18n/fa-IR/dataset-hit-testing.ts
@@ -23,6 +23,8 @@ const translation = {
   },
   noRecentTip: 'اینجا نتیجه پرسش اخیر وجود ندارد',
   viewChart: 'مشاهده نمودار بُرداری',
+  settingTitle: 'تنظیمات بازیابی',
+  viewDetail: 'نمایش جزئیات',
 }
 
 export default translation
diff --git a/web/i18n/fa-IR/dataset-settings.ts b/web/i18n/fa-IR/dataset-settings.ts
index a9c9bd8110e92..f7efa7f9977fb 100644
--- a/web/i18n/fa-IR/dataset-settings.ts
+++ b/web/i18n/fa-IR/dataset-settings.ts
@@ -29,6 +29,9 @@ const translation = {
       longDescription: ' درباره روش بازیابی، می‌توانید در هر زمانی در تنظیمات دانش این را تغییر دهید.',
     },
     save: 'ذخیره',
+    externalKnowledgeAPI: 'API دانش خارجی',
+    retrievalSettings: 'تنظیمات بازیابی',
+    externalKnowledgeID: 'شناسه دانش خارجی',
   },
 }
 
diff --git a/web/i18n/fa-IR/dataset.ts b/web/i18n/fa-IR/dataset.ts
index e3b9d70e07361..107e8dd0977d4 100644
--- a/web/i18n/fa-IR/dataset.ts
+++ b/web/i18n/fa-IR/dataset.ts
@@ -72,6 +72,80 @@ const translation = {
   nTo1RetrievalLegacyLink: 'بیشتر بدانید',
   nTo1RetrievalLegacyLinkText: ' بازیابی N-to-1 از سپتامبر به طور رسمی منسوخ خواهد شد.',
   defaultRetrievalTip: 'بازیابی چند مسیره به طور پیش فرض استفاده می شود. دانش از چندین پایگاه دانش بازیابی می شود و سپس دوباره رتبه بندی می شود.',
+  editExternalAPIConfirmWarningContent: {
+    front: 'این API دانش خارجی به',
+    end: 'دانش خارجی ، و این اصلاح برای همه آنها اعمال خواهد شد. آیا مطمئن هستید که می خواهید این تغییر را ذخیره کنید؟',
+  },
+  editExternalAPIFormWarning: {
+    front: 'این API خارجی به',
+    end: 'دانش بیرونی',
+  },
+  deleteExternalAPIConfirmWarningContent: {
+    title: {
+      front: 'حذف',
+      end: '?',
+    },
+    content: {
+      front: 'این API دانش خارجی به',
+      end: 'دانش بیرونی. حذف این API همه آنها را باطل می کند. آیا مطمئن هستید که می خواهید این API را حذف کنید؟',
+    },
+    noConnectionContent: 'آیا مطمئن هستید که این API را حذف خواهید کرد؟',
+  },
+  selectExternalKnowledgeAPI: {
+    placeholder: 'یک API دانش خارجی را انتخاب کنید',
+  },
+  connectDatasetIntro: {
+    content: {
+      link: 'یادگیری نحوه ایجاد یک API خارجی',
+      front: 'برای اتصال به یک پایگاه دانش خارجی، ابتدا باید یک API خارجی ایجاد کنید. لطفا با دقت بخوانید و به',
+      end: '. سپس شناسه دانش مربوطه را پیدا کرده و آن را در فرم سمت چپ پر کنید. اگر تمام اطلاعات صحیح باشد، پس از کلیک بر روی دکمه اتصال، به طور خودکار به آزمون بازیابی در پایگاه دانش می رود.',
+    },
+    learnMore: 'بیشتر بدانید',
+    title: 'چگونه به یک پایگاه دانش خارجی متصل شویم؟',
+  },
+  connectHelper: {
+    helper5: 'قبل از استفاده از این ویژگی با دقت',
+    helper3: '. اکیدا توصیه می کنیم که',
+    helper2: 'فقط قابلیت بازیابی پشتیبانی می شود',
+    helper4: 'مستندات راهنما را بخوانید',
+    helper1: 'از طریق API و شناسه پایگاه دانش به پایگاه های دانش خارجی متصل شوید. در حال حاضر،',
+  },
+  externalKnowledgeForm: {
+    cancel: 'لغو',
+    connect: 'اتصال',
+  },
+  externalAPIForm: {
+    encrypted: {
+      front: 'توکن API شما رمزگذاری و با استفاده از',
+      end: 'فناوری.',
+    },
+    apiKey: 'کلید API',
+    edit: 'ویرایش',
+    save: 'ذخیره',
+    cancel: 'لغو',
+    endpoint: 'نقطه پایانی API',
+    name: 'نام',
+  },
+  editExternalAPITooltipTitle: 'دانش مرتبط',
+  externalKnowledgeNamePlaceholder: 'لطفا نام پایگاه دانش را وارد کنید',
+  externalAPIPanelDocumentation: 'یادگیری نحوه ایجاد یک API دانش خارجی',
+  externalKnowledgeDescriptionPlaceholder: 'آنچه در این پایگاه دانش وجود دارد را توضیح دهید (اختیاری)',
+  externalKnowledgeDescription: 'توضیحات دانش',
+  externalTag: 'خارجی',
+  externalKnowledgeIdPlaceholder: 'لطفا شناسه دانش را وارد کنید',
+  noExternalKnowledge: 'هنوز هیچ API دانش خارجی وجود ندارد، برای ایجاد اینجا را کلیک کنید',
+  externalAPIPanelTitle: 'API دانش خارجی',
+  connectDataset: 'اتصال به یک پایگاه دانش خارجی',
+  externalKnowledgeId: 'شناسه دانش خارجی',
+  externalAPI: 'API خارجی',
+  externalKnowledgeName: 'نام دانش خارجی',
+  createExternalAPI: 'افزودن یک API دانش خارجی',
+  createNewExternalAPI: 'ایجاد یک API دانش خارجی جدید',
+  learnHowToWriteGoodKnowledgeDescription: 'یاد بگیرید که چگونه یک توضیحات دانش خوب بنویسید.',
+  editExternalAPIFormTitle: 'ویرایش API دانش خارجی',
+  externalAPIPanelDescription: 'API دانش خارجی برای اتصال به یک پایگاه دانش خارج از Dify و بازیابی دانش از آن پایگاه دانش استفاده می شود.',
+  allExternalTip: 'هنگامی که فقط از دانش خارجی استفاده می کنید، کاربر می تواند انتخاب کند که آیا مدل Rerank را فعال کند یا خیر. اگر فعال نباشد، تکه های بازیابی شده بر اساس امتیازات مرتب می شوند. هنگامی که استراتژی های بازیابی پایگاه های دانش مختلف متناقض باشد، نادرست خواهد بود.',
+  mixtureInternalAndExternalTip: 'مدل Rerank برای آمیختگی دانش درونی و بیرونی مورد نیاز است.',
 }
 
 export default translation
diff --git a/web/i18n/fr-FR/app-api.ts b/web/i18n/fr-FR/app-api.ts
index af3f752f98228..aad2b0394baa0 100644
--- a/web/i18n/fr-FR/app-api.ts
+++ b/web/i18n/fr-FR/app-api.ts
@@ -78,6 +78,7 @@ const translation = {
     query: 'Requête',
   },
   loading: 'Chargement',
+  regenerate: 'Régénérer',
 }
 
 export default translation
diff --git a/web/i18n/fr-FR/dataset-creation.ts b/web/i18n/fr-FR/dataset-creation.ts
index 38066885d0c6e..302e0796cbbeb 100644
--- a/web/i18n/fr-FR/dataset-creation.ts
+++ b/web/i18n/fr-FR/dataset-creation.ts
@@ -73,6 +73,14 @@ const translation = {
       extractOnlyMainContent: 'Extraire uniquement le contenu principal (pas d’en-têtes, de navigations, de pieds de page, etc.)',
       excludePaths: 'Exclure les chemins d’accès',
       maxDepthTooltip: 'Profondeur maximale à explorer par rapport à l’URL saisie. La profondeur 0 gratte simplement la page de l’URL saisie, la profondeur 1 récupère l’URL et tout ce qui suit l’URL saisie + un /, et ainsi de suite.',
+      jinaReaderDocLink: 'https://jina.ai/reader',
+      jinaReaderDoc: 'En savoir plus sur Jina Reader',
+      useSitemapTooltip: 'Suivez le plan du site pour explorer le site. Si ce n’est pas le cas, Jina Reader explorera de manière itérative en fonction de la pertinence de la page, produisant des pages moins nombreuses mais de meilleure qualité.',
+      jinaReaderNotConfiguredDescription: 'Configurez Jina Reader en saisissant votre clé API gratuite pour y accéder.',
+      useSitemap: 'Utiliser le sitemap',
+      jinaReaderNotConfigured: 'Jina Reader n’est pas configuré',
+      chooseProvider: 'Sélectionnez un fournisseur',
+      jinaReaderTitle: 'Convertir l’intégralité du site en Markdown',
     },
   },
   stepTwo: {
@@ -157,6 +165,11 @@ const translation = {
     configFirecrawl: 'Configurer 🔥Firecrawl',
     getApiKeyLinkText: 'Obtenez votre clé API auprès de firecrawl.dev',
   },
+  jinaReader: {
+    getApiKeyLinkText: 'Obtenez votre clé API gratuite chez jina.ai',
+    apiKeyPlaceholder: 'Clé API de jina.ai',
+    configJinaReader: 'Configurer Jina Reader',
+  },
 }
 
 export default translation
diff --git a/web/i18n/fr-FR/dataset-hit-testing.ts b/web/i18n/fr-FR/dataset-hit-testing.ts
index 1e527bcf245b5..3fa2abbab1f45 100644
--- a/web/i18n/fr-FR/dataset-hit-testing.ts
+++ b/web/i18n/fr-FR/dataset-hit-testing.ts
@@ -23,6 +23,8 @@ const translation = {
   },
   noRecentTip: 'Aucun résultat de requête récent ici',
   viewChart: 'Voir GRAPHIQUE VECTORIEL',
+  settingTitle: 'Réglage de récupération',
+  viewDetail: 'Voir les détails',
 }
 
 export default translation
diff --git a/web/i18n/fr-FR/dataset-settings.ts b/web/i18n/fr-FR/dataset-settings.ts
index 9b1f44f54fa19..f7c5f4601a6d5 100644
--- a/web/i18n/fr-FR/dataset-settings.ts
+++ b/web/i18n/fr-FR/dataset-settings.ts
@@ -29,6 +29,9 @@ const translation = {
     save: 'Enregistrer',
     me: '(Vous)',
     permissionsInvitedMembers: 'Membres partiels de l’équipe',
+    retrievalSettings: 'Paramètres de récupération',
+    externalKnowledgeAPI: 'API de connaissances externes',
+    externalKnowledgeID: 'Identification des connaissances externes',
   },
 }
 
diff --git a/web/i18n/fr-FR/dataset.ts b/web/i18n/fr-FR/dataset.ts
index 9c8df9f79d91f..54a73115a2f9a 100644
--- a/web/i18n/fr-FR/dataset.ts
+++ b/web/i18n/fr-FR/dataset.ts
@@ -72,6 +72,80 @@ const translation = {
   nTo1RetrievalLegacyLink: 'En savoir plus',
   nTo1RetrievalLegacyLinkText: 'La récupération N-à-1 sera officiellement obsolète en septembre.',
   defaultRetrievalTip: 'La récupération à chemins multiples est utilisée par défaut. Les connaissances sont extraites de plusieurs bases de connaissances, puis reclassées.',
+  editExternalAPIConfirmWarningContent: {
+    front: 'Cette API de connaissances externes est liée à',
+    end: 'connaissances externes, et cette modification sera appliquée à tous. Êtes-vous sûr de vouloir enregistrer cette modification ?',
+  },
+  editExternalAPIFormWarning: {
+    end: 'Connaissances externes',
+    front: 'Cette API externe est liée à',
+  },
+  deleteExternalAPIConfirmWarningContent: {
+    title: {
+      end: '?',
+      front: 'Supprimer',
+    },
+    content: {
+      front: 'Cette API de connaissances externes est liée à',
+      end: 'connaissances externes. La suppression de cette API les invalidera toutes. Êtes-vous sûr de vouloir supprimer cette API ?',
+    },
+    noConnectionContent: 'Êtes-vous sûr de supprimer cette API ?',
+  },
+  selectExternalKnowledgeAPI: {
+    placeholder: 'Choisir une API de connaissances externe',
+  },
+  connectDatasetIntro: {
+    content: {
+      link: 'Découvrez comment créer une API externe',
+      end: '. Trouvez ensuite l’ID de connaissances correspondant et remplissez-le dans le formulaire sur la gauche. Si toutes les informations sont correctes, il passera automatiquement au test de récupération dans la base de connaissances après avoir cliqué sur le bouton de connexion.',
+      front: 'Pour vous connecter à une base de connaissances externe, vous devez d’abord créer une API externe. Veuillez lire attentivement et vous référer à',
+    },
+    learnMore: 'Pour en savoir plus',
+    title: 'Comment se connecter à une base de connaissances externe',
+  },
+  connectHelper: {
+    helper2: 'Seule la fonctionnalité de récupération est prise en charge',
+    helper3: '. Nous vous recommandons vivement de',
+    helper4: 'Lire la documentation d’aide',
+    helper5: 'soigneusement avant d’utiliser cette fonctionnalité.',
+    helper1: 'Connectez-vous à des bases de connaissances externes via l’API et l’ID de base de connaissances.',
+  },
+  externalKnowledgeForm: {
+    cancel: 'Annuler',
+    connect: 'Relier',
+  },
+  externalAPIForm: {
+    encrypted: {
+      end: 'Technologie.',
+      front: 'Votre jeton API sera chiffré et stocké à l’aide de',
+    },
+    name: 'Nom',
+    apiKey: 'Clé API',
+    save: 'Sauvegarder',
+    cancel: 'Annuler',
+    edit: 'Éditer',
+    endpoint: 'Point de terminaison de l’API',
+  },
+  externalKnowledgeName: 'Nom de la connaissance externe',
+  mixtureInternalAndExternalTip: 'Le modèle Rerank est nécessaire pour mélanger les connaissances internes et externes.',
+  externalKnowledgeIdPlaceholder: 'Entrez l’ID de connaissances',
+  createExternalAPI: 'Ajouter une API de connaissances externe',
+  externalKnowledgeNamePlaceholder: 'Entrez le nom de la base de connaissances',
+  allExternalTip: 'Lorsqu’il utilise uniquement des connaissances externes, l’utilisateur peut choisir d’activer ou non le modèle Rerank. S’il n’est pas activé, les morceaux récupérés seront triés en fonction des scores. Lorsque les stratégies de récupération des différentes bases de connaissances sont incohérentes, elles seront inexactes.',
+  externalAPI: 'API externe',
+  editExternalAPIFormTitle: 'Modifier l’API de connaissances externes',
+  externalTag: 'Externe',
+  editExternalAPITooltipTitle: 'CONNAISSANCES ASSOCIÉES',
+  connectDataset: 'Se connecter à une base de connaissances externe',
+  externalKnowledgeDescription: 'Description des connaissances',
+  externalAPIPanelDocumentation: 'Découvrez comment créer une API de connaissances externe',
+  createNewExternalAPI: 'Créer une API de connaissances externe',
+  externalKnowledgeDescriptionPlaceholder: 'Décrivez le contenu de cette base de connaissances (facultatif)',
+  externalAPIPanelDescription: 'L’API de connaissances externe est utilisée pour se connecter à une base de connaissances en dehors de Dify et récupérer des connaissances de cette base de connaissances.',
+  externalKnowledgeId: 'Identification des connaissances externes',
+  externalAPIPanelTitle: 'API de connaissances externes',
+  noExternalKnowledge: 'Il n’y a pas encore d’API de connaissances externes, cliquez ici pour créer',
+  learnHowToWriteGoodKnowledgeDescription: 'Apprenez à rédiger une bonne description des connaissances',
 }
 
 export default translation
diff --git a/web/i18n/hi-IN/app-api.ts b/web/i18n/hi-IN/app-api.ts
index b983fd0ccb7d8..d50b6c026620f 100644
--- a/web/i18n/hi-IN/app-api.ts
+++ b/web/i18n/hi-IN/app-api.ts
@@ -78,6 +78,7 @@ const translation = {
     pathParams: 'पथ पैरामीटर',
     query: 'प्रश्न',
   },
+  regenerate: 'पुनर्जन्म',
 }
 
 export default translation
diff --git a/web/i18n/hi-IN/dataset-creation.ts b/web/i18n/hi-IN/dataset-creation.ts
index 19c396081a37e..36986d0f8606c 100644
--- a/web/i18n/hi-IN/dataset-creation.ts
+++ b/web/i18n/hi-IN/dataset-creation.ts
@@ -83,6 +83,14 @@ const translation = {
       preview: 'पूर्वावलोकन',
       maxDepthTooltip:
         'प्रविष्ट URL के सापेक्ष क्रॉल करने की अधिकतम गहराई। गहराई 0 केवल प्रविष्ट url का पृष्ठ स्क्रैप करता है, गहराई 1 url और प्रविष्टURL + एक / के बाद सब कुछ स्क्रैप करता है, और इसी तरह।',
+      jinaReaderDocLink: 'https://jina.ai/reader',
+      jinaReaderDoc: 'जीना रीडर के बारे में अधिक जानें',
+      useSitemapTooltip: 'साइट क्रॉल करने के लिए साइटमैप का अनुसरण करें. यदि नहीं, तो जीना रीडर पृष्ठ प्रासंगिकता के आधार पर पुनरावृत्त रूप से क्रॉल करेगा, कम लेकिन उच्च-गुणवत्ता वाले पृष्ठों की उपज देगा।',
+      useSitemap: 'साइटमैप का इस्तेमाल करना',
+      jinaReaderNotConfigured: 'जीना रीडर कॉन्फ़िगर नहीं किया गया है',
+      chooseProvider: 'एक प्रदाता का चयन करें',
+      jinaReaderTitle: 'पूरी साइट को मार्कडाउन में बदलें',
+      jinaReaderNotConfiguredDescription: 'एक्सेस के लिए अपनी मुफ्त एपीआई कुंजी दर्ज करके जीना रीडर सेट करें।',
     },
   },
   stepTwo: {
@@ -177,6 +185,11 @@ const translation = {
     modelButtonConfirm: 'पुष्टि करें',
     modelButtonCancel: 'रद्द करें',
   },
+  jinaReader: {
+    configJinaReader: 'जीना रीडर कॉन्फ़िगर करें',
+    apiKeyPlaceholder: 'jina.ai से एपीआई कुंजी',
+    getApiKeyLinkText: 'jina.ai पर अपनी निःशुल्क एपीआई कुंजी प्राप्त करें',
+  },
 }
 
 export default translation
diff --git a/web/i18n/hi-IN/dataset-hit-testing.ts b/web/i18n/hi-IN/dataset-hit-testing.ts
index 229e7a2d8e0d9..e15d57884d527 100644
--- a/web/i18n/hi-IN/dataset-hit-testing.ts
+++ b/web/i18n/hi-IN/dataset-hit-testing.ts
@@ -23,6 +23,8 @@ const translation = {
   },
   noRecentTip: 'यहां कोई हाल के प्रश्न परिणाम नहीं हैं',
   viewChart: 'वेक्टर चार्ट देखें',
+  viewDetail: 'विस्तार से देखें',
+  settingTitle: 'पुनर्प्राप्ति सेटिंग',
 }
 
 export default translation
diff --git a/web/i18n/hi-IN/dataset-settings.ts b/web/i18n/hi-IN/dataset-settings.ts
index 129643dd85d93..ea825417c1d7c 100644
--- a/web/i18n/hi-IN/dataset-settings.ts
+++ b/web/i18n/hi-IN/dataset-settings.ts
@@ -34,6 +34,9 @@ const translation = {
     save: 'सेवना',
     me: '(आप)',
     permissionsInvitedMembers: 'आंशिक टीम के सदस्य',
+    externalKnowledgeID: 'बाहरी ज्ञान ID',
+    externalKnowledgeAPI: 'बाहरी ज्ञान एपीआई',
+    retrievalSettings: 'पुनर्प्राप्ति सेटिंग्स',
   },
 }
 
diff --git a/web/i18n/hi-IN/dataset.ts b/web/i18n/hi-IN/dataset.ts
index afd8cd277c319..b0b8519fd7d1a 100644
--- a/web/i18n/hi-IN/dataset.ts
+++ b/web/i18n/hi-IN/dataset.ts
@@ -79,6 +79,80 @@ const translation = {
   nTo1RetrievalLegacyLink: 'और जानें',
   nTo1RetrievalLegacyLinkText: 'N-से-1 पुनर्प्राप्ति सितंबर में आधिकारिक तौर पर बंद कर दी जाएगी।',
   defaultRetrievalTip: 'मल्टी-पाथ रिट्रीवल का उपयोग डिफ़ॉल्ट रूप से किया जाता है। ज्ञान को कई ज्ञान आधारों से पुनर्प्राप्त किया जाता है और फिर फिर से रैंक किया जाता है।',
+  editExternalAPIConfirmWarningContent: {
+    end: 'बाहरी ज्ञान, और यह संशोधन उन सभी पर लागू किया जाएगा। क्या आप वाकई यह परिवर्तन सहेजना चाहते हैं?',
+    front: 'यह बाहरी ज्ञान API इससे जुड़ा हुआ है',
+  },
+  editExternalAPIFormWarning: {
+    end: 'बाहरी ज्ञान',
+    front: 'यह बाहरी एपीआई किससे जुड़ा हुआ है',
+  },
+  deleteExternalAPIConfirmWarningContent: {
+    title: {
+      front: 'मिटाना',
+      end: '?',
+    },
+    content: {
+      front: 'यह बाहरी ज्ञान API इससे जुड़ा हुआ है',
+      end: 'बाहरी ज्ञान। इस एपीआई को हटाने से वे सभी अमान्य हो जाएंगे। क्या आप वाकई इस API को हटाना चाहते हैं?',
+    },
+    noConnectionContent: 'क्या आप वाकई इस API को हटा देंगे?',
+  },
+  selectExternalKnowledgeAPI: {
+    placeholder: 'एक बाहरी ज्ञान API चुनें',
+  },
+  connectDatasetIntro: {
+    content: {
+      link: 'बाहरी API बनाने का तरीका जानें',
+      front: 'बाहरी ज्ञानकोष से कनेक्ट करने के लिए, आपको पहले एक बाहरी API बनाना होगा। कृपया काळजीपूर्वक वाचा आणि संदर्भ घ्या',
+      end: '. फिर संबंधित ज्ञान आईडी ढूंढें और इसे बाईं ओर के फॉर्म में भरें। यदि सभी जानकारी सही है, तो यह कनेक्ट बटन पर क्लिक करने के बाद स्वचालित रूप से नॉलेज बेस में पुनर्प्राप्ति परीक्षण पर कूद जाएगा।',
+    },
+    learnMore: 'और जानो',
+    title: 'किसी बाहरी ज्ञानकोष से कनेक्ट करने के लिए कैसे करें',
+  },
+  connectHelper: {
+    helper5: 'इस सुविधा का उपयोग करने से पहले सावधानी से।',
+    helper2: 'केवल पुनर्प्राप्ति कार्यक्षमता समर्थित है',
+    helper3: '. हम दृढ़ता से अनुशंसा करते हैं कि आप',
+    helper4: 'सहायता दस्तावेज़ पढ़ें',
+    helper1: 'API और नॉलेज बेस ID के माध्यम से बाहरी नॉलेज बेस से कनेक्ट करें. वर्तमान में,',
+  },
+  externalKnowledgeForm: {
+    connect: 'जोड़ना',
+    cancel: 'रद्द करना',
+  },
+  externalAPIForm: {
+    encrypted: {
+      end: 'टेक्‍नोलॉजी।',
+      front: 'आपका एपीआई टोकन एन्क्रिप्ट किया जाएगा और इसका उपयोग करके संग्रहीत किया जाएगा',
+    },
+    apiKey: 'एपीआई कुंजी',
+    name: 'नाम',
+    cancel: 'रद्द करना',
+    save: 'रक्षा कर',
+    edit: 'संपादन करना',
+    endpoint: 'एपीआई समापन बिंदु',
+  },
+  externalAPI: 'बाहरी एपीआई',
+  externalAPIPanelTitle: 'बाहरी ज्ञान एपीआई',
+  allExternalTip: 'केवल बाहरी ज्ञान का उपयोग करते समय, उपयोगकर्ता यह चुन सकता है कि Rerank मॉडल को सक्षम करना है या नहीं। यदि सक्षम नहीं है, तो पुनर्प्राप्त किए गए विखंडू स्कोर के आधार पर क्रमबद्ध किए जाएंगे। जब विभिन्न ज्ञान आधारों की पुनर्प्राप्ति रणनीतियाँ असंगत होती हैं, तो यह गलत होगी।',
+  externalKnowledgeName: 'बाहरी ज्ञान का नाम',
+  connectDataset: 'किसी बाह्य ज्ञानकोष से कनेक्ट करना',
+  mixtureInternalAndExternalTip: 'आंतरिक और बाहरी ज्ञान के मिश्रण के लिए रीरैंक मॉडल की आवश्यकता होती है।',
+  externalTag: 'बाह्य',
+  externalAPIPanelDescription: 'बाहरी ज्ञान API का उपयोग Dify के बाहर नॉलेज बेस से कनेक्ट करने और उस नॉलेज बेस से ज्ञान प्राप्त करने के लिए किया जाता है।',
+  externalKnowledgeDescription: 'ज्ञान विवरण',
+  createExternalAPI: 'कोई बाहरी नॉलेज API जोड़ना',
+  externalKnowledgeIdPlaceholder: 'कृपया नॉलेज आईडी दर्ज करें',
+  editExternalAPITooltipTitle: 'लिंक किया गया ज्ञान',
+  externalAPIPanelDocumentation: 'बाहरी नॉलेज API बनाने का तरीका जानें',
+  editExternalAPIFormTitle: 'बाह्य ज्ञान API संपादित करें',
+  externalKnowledgeNamePlaceholder: 'कृपया नॉलेज बेस का नाम दर्ज करें',
+  externalKnowledgeId: 'बाहरी ज्ञान ID',
+  externalKnowledgeDescriptionPlaceholder: 'वर्णन करें कि इस ज्ञानकोष में क्या है (वैकल्पिक)',
+  noExternalKnowledge: 'अभी तक कोई बाहरी ज्ञान एपीआई नहीं है, बनाने के लिए यहां क्लिक करें',
+  createNewExternalAPI: 'एक नया बाहरी नॉलेज API बनाएँ',
+  learnHowToWriteGoodKnowledgeDescription: 'एक अच्छा ज्ञान विवरण लिखना सीखें',
 }
 
 export default translation
diff --git a/web/i18n/it-IT/app-api.ts b/web/i18n/it-IT/app-api.ts
index 48aceb5c48f12..2028ae42c0d4d 100644
--- a/web/i18n/it-IT/app-api.ts
+++ b/web/i18n/it-IT/app-api.ts
@@ -99,6 +99,7 @@ const translation = {
     pathParams: 'Parametri del Percorso',
     query: 'Query',
   },
+  regenerate: 'Rigenerare',
 }
 
 export default translation
diff --git a/web/i18n/it-IT/dataset-creation.ts b/web/i18n/it-IT/dataset-creation.ts
index 46889b80a60c8..f46101257c69b 100644
--- a/web/i18n/it-IT/dataset-creation.ts
+++ b/web/i18n/it-IT/dataset-creation.ts
@@ -85,6 +85,14 @@ const translation = {
       preview: 'Anteprima',
       maxDepthTooltip:
         'Profondità massima da eseguire rispetto all\'URL inserito. La profondità 0 estrae solo la pagina dell\'URL inserito, la profondità 1 estrae l\'URL e tutto ciò che segue l\'URL inserito + uno /, e così via.',
+      jinaReaderDocLink: 'https://jina.ai/reader',
+      useSitemapTooltip: 'Segui la sitemap per eseguire la scansione del sito. In caso contrario, Jina Reader eseguirà la scansione in modo iterativo in base alla pertinenza della pagina, producendo meno pagine ma di qualità superiore.',
+      jinaReaderNotConfigured: 'Jina Reader non è configurato',
+      jinaReaderDoc: 'Scopri di più su Jina Reader',
+      jinaReaderTitle: 'Converti l\'intero sito in Markdown',
+      jinaReaderNotConfiguredDescription: 'Configura Jina Reader inserendo la tua chiave API gratuita per l\'accesso.',
+      useSitemap: 'Usa la mappa del sito',
+      chooseProvider: 'Seleziona un fornitore',
     },
   },
   stepTwo: {
@@ -180,6 +188,11 @@ const translation = {
     modelButtonConfirm: 'Conferma',
     modelButtonCancel: 'Annulla',
   },
+  jinaReader: {
+    getApiKeyLinkText: 'Ottieni la tua chiave API gratuita su jina.ai',
+    apiKeyPlaceholder: 'Chiave API da jina.ai',
+    configJinaReader: 'Configura Jina Reader',
+  },
 }
 
 export default translation
diff --git a/web/i18n/it-IT/dataset-hit-testing.ts b/web/i18n/it-IT/dataset-hit-testing.ts
index ef5ad0ecc6a5b..e8ca919220906 100644
--- a/web/i18n/it-IT/dataset-hit-testing.ts
+++ b/web/i18n/it-IT/dataset-hit-testing.ts
@@ -24,6 +24,8 @@ const translation = {
   },
   noRecentTip: 'Nessun risultato di query recente qui',
   viewChart: 'Visualizza GRAFICO VETTORIALE',
+  settingTitle: 'Impostazione di recupero',
+  viewDetail: 'vedi dettagli',
 }
 
 export default translation
diff --git a/web/i18n/it-IT/dataset-settings.ts b/web/i18n/it-IT/dataset-settings.ts
index d88417962c040..9d4b6218e5c62 100644
--- a/web/i18n/it-IT/dataset-settings.ts
+++ b/web/i18n/it-IT/dataset-settings.ts
@@ -34,6 +34,9 @@ const translation = {
         ' sul metodo di recupero, puoi cambiare questo in qualsiasi momento nelle impostazioni della Conoscenza.',
     },
     save: 'Salva',
+    retrievalSettings: 'Impostazioni di recupero',
+    externalKnowledgeID: 'ID conoscenza esterna',
+    externalKnowledgeAPI: 'API di conoscenza esterna',
   },
 }
 
diff --git a/web/i18n/it-IT/dataset.ts b/web/i18n/it-IT/dataset.ts
index fc79c9b4c74c0..22d1700d1422e 100644
--- a/web/i18n/it-IT/dataset.ts
+++ b/web/i18n/it-IT/dataset.ts
@@ -79,6 +79,80 @@ const translation = {
   nTo1RetrievalLegacyLink: 'Scopri di più',
   nTo1RetrievalLegacyLinkText: 'Il recupero N-a-1 sarà ufficialmente deprecato a settembre.',
   defaultRetrievalTip: 'Per impostazione predefinita, il recupero a percorsi multipli viene utilizzato. Le informazioni vengono recuperate da più knowledge base e quindi riclassificate.',
+  editExternalAPIConfirmWarningContent: {
+    end: 'conoscenza esterna, e questa modifica sarà applicata a tutti loro. Sei sicuro di voler salvare questa modifica?',
+    front: 'Questa API della conoscenza esterna è collegata a',
+  },
+  editExternalAPIFormWarning: {
+    end: 'Conoscenze esterne',
+    front: 'Questa API esterna è collegata a',
+  },
+  deleteExternalAPIConfirmWarningContent: {
+    title: {
+      front: 'Cancellare',
+      end: '?',
+    },
+    content: {
+      end: 'conoscenze esterne. L\'eliminazione di questa API invaliderà tutte. Si è certi di voler eliminare questa API?',
+      front: 'Questa API della conoscenza esterna è collegata a',
+    },
+    noConnectionContent: 'Sei sicuro di eliminare questa API?',
+  },
+  selectExternalKnowledgeAPI: {
+    placeholder: 'Scegliere un\'API di conoscenza esterna',
+  },
+  connectDatasetIntro: {
+    content: {
+      front: 'Per connettersi a una knowledge base esterna, è necessario prima creare un\'API esterna. Si prega di leggere attentamente e fare riferimento a',
+      end: '. Quindi trova l\'ID conoscenza corrispondente e compilalo nel modulo a sinistra. Se tutte le informazioni sono corrette, passerà automaticamente al test di recupero nella knowledge base dopo aver fatto clic sul pulsante di connessione.',
+      link: 'Scopri come creare un\'API esterna',
+    },
+    title: 'Come connettersi a una Knowledge Base esterna',
+    learnMore: 'Ulteriori informazioni',
+  },
+  connectHelper: {
+    helper2: 'È supportata solo la funzionalità di recupero',
+    helper4: 'Leggi la documentazione della Guida',
+    helper3: '. Ti consigliamo vivamente di',
+    helper1: 'Connettiti a knowledge base esterne tramite API e ID knowledge base. Attualmente,',
+    helper5: 'prima di utilizzare questa funzione.',
+  },
+  externalKnowledgeForm: {
+    cancel: 'Annulla',
+    connect: 'Connettersi',
+  },
+  externalAPIForm: {
+    encrypted: {
+      end: 'Tecnologia.',
+      front: 'Il tuo token API verrà crittografato e archiviato utilizzando',
+    },
+    apiKey: 'Chiave API',
+    name: 'Nome',
+    cancel: 'Annulla',
+    edit: 'Redigere',
+    save: 'Salvare',
+    endpoint: 'API Endpoint',
+  },
+  externalAPIPanelDescription: 'L\'API di conoscenza esterna viene utilizzata per connettersi a una knowledge base esterna a Dify e recuperare le informazioni da tale knowledge base.',
+  createExternalAPI: 'Aggiungere un\'API di conoscenza esterna',
+  learnHowToWriteGoodKnowledgeDescription: 'Impara a scrivere una buona descrizione della conoscenza',
+  externalKnowledgeName: 'Nome della conoscenza esterna',
+  externalAPIPanelTitle: 'API di conoscenza esterna',
+  externalAPI: 'API esterna',
+  createNewExternalAPI: 'Creare una nuova API della conoscenza esterna',
+  editExternalAPIFormTitle: 'Modificare l\'API della conoscenza esterna',
+  externalKnowledgeId: 'ID conoscenza esterna',
+  externalKnowledgeIdPlaceholder: 'Inserisci l\'ID conoscenza',
+  externalAPIPanelDocumentation: 'Scopri come creare un\'API della Knowledge Base esterna',
+  connectDataset: 'Connettiti a una Knowledge Base esterna',
+  mixtureInternalAndExternalTip: 'Il modello Rerank è necessario per la combinazione di conoscenze interne ed esterne.',
+  editExternalAPITooltipTitle: 'CONOSCENZA COLLEGATA',
+  externalTag: 'Esterno',
+  externalKnowledgeNamePlaceholder: 'Inserisci il nome della knowledge base',
+  externalKnowledgeDescription: 'Descrizione della conoscenza',
+  allExternalTip: 'Quando si utilizzano solo conoscenze esterne, l\'utente può scegliere se abilitare il modello Rerank. Se non è abilitato, i blocchi recuperati verranno ordinati in base ai punteggi. Quando le strategie di recupero di diverse basi di conoscenza sono incoerenti, saranno imprecise.',
+  externalKnowledgeDescriptionPlaceholder: 'Descrivi cosa c\'è in questa Knowledge Base (facoltativo)',
+  noExternalKnowledge: 'Non esiste ancora un\'API di conoscenza esterna, fai clic qui per creare',
 }
 
 export default translation
diff --git a/web/i18n/ja-JP/app-api.ts b/web/i18n/ja-JP/app-api.ts
index 721504f9f30a1..9472f1ea285ea 100644
--- a/web/i18n/ja-JP/app-api.ts
+++ b/web/i18n/ja-JP/app-api.ts
@@ -78,6 +78,7 @@ const translation = {
     pathParams: 'パスパラメータ',
     query: 'クエリ',
   },
+  regenerate: '再生',
 }
 
 export default translation
diff --git a/web/i18n/ja-JP/dataset-creation.ts b/web/i18n/ja-JP/dataset-creation.ts
index d11a0c94e9bba..570876c4b4c6e 100644
--- a/web/i18n/ja-JP/dataset-creation.ts
+++ b/web/i18n/ja-JP/dataset-creation.ts
@@ -78,6 +78,14 @@ const translation = {
       scrapTimeInfo: '{{time}} 秒以内に合計 {{total}} ページをスクレイピングしました',
       preview: 'プレビュー',
       maxDepthTooltip: '入力されたURLを基にしたクローリング作業での設定可能な最大深度について説明します。深度0は入力されたURL自体のページを対象としたスクレイピングを意味します。深度1では、元のURLの直下にあるページ（URLに続く最初の"/"以降の内容）もスクレイピングの対象になります。この深度は指定した数値まで増加させることができ、それに応じてスクレイピングの範囲も広がっていきます。',
+      jinaReaderDocLink: 'https://jina.ai/reader',
+      useSitemap: 'サイトマップを使用する',
+      jinaReaderNotConfigured: 'Jina Reader が設定されていません',
+      jinaReaderDoc: 'Jina Readerの詳細',
+      jinaReaderTitle: 'サイト全体をMarkdownに変換する',
+      chooseProvider: 'プロバイダーを選択する',
+      jinaReaderNotConfiguredDescription: '無料のAPIキーを入力してJina Readerを設定します。',
+      useSitemapTooltip: 'サイトマップに沿ってサイトをクロールします。そうでない場合、Jina Readerはページの関連性に基づいて繰り返しクロールし、ページ数は少なくなりますが、高品質のページが得られます。',
     },
   },
   stepTwo: {
@@ -157,6 +165,11 @@ const translation = {
     modelButtonConfirm: '確認',
     modelButtonCancel: 'キャンセル',
   },
+  jinaReader: {
+    getApiKeyLinkText: '無料のAPIキーを jina.ai で取得',
+    apiKeyPlaceholder: 'jina.ai からの API キー',
+    configJinaReader: 'Jina Readerの設定',
+  },
 }
 
 export default translation
diff --git a/web/i18n/ja-JP/dataset-hit-testing.ts b/web/i18n/ja-JP/dataset-hit-testing.ts
index abfd65ddd7afd..e2bc807969cb1 100644
--- a/web/i18n/ja-JP/dataset-hit-testing.ts
+++ b/web/i18n/ja-JP/dataset-hit-testing.ts
@@ -23,6 +23,8 @@ const translation = {
   },
   noRecentTip: '最近のクエリ結果はありません。',
   viewChart: 'ベクトルチャートを表示',
+  settingTitle: '取得設定',
+  viewDetail: '詳細を表示',
 }
 
 export default translation
diff --git a/web/i18n/ja-JP/dataset-settings.ts b/web/i18n/ja-JP/dataset-settings.ts
index d7913cb54ab3d..1eb3dabb746b5 100644
--- a/web/i18n/ja-JP/dataset-settings.ts
+++ b/web/i18n/ja-JP/dataset-settings.ts
@@ -29,6 +29,9 @@ const translation = {
       longDescription: ' 検索方法についての詳細については、いつでもナレッジの設定で変更できます。',
     },
     save: '保存',
+    externalKnowledgeID: '外部ナレッジID',
+    retrievalSettings: '取得設定',
+    externalKnowledgeAPI: '外部ナレッジAPI',
   },
 }
 
diff --git a/web/i18n/ja-JP/dataset.ts b/web/i18n/ja-JP/dataset.ts
index a765473b7e2f2..d995509a3fed4 100644
--- a/web/i18n/ja-JP/dataset.ts
+++ b/web/i18n/ja-JP/dataset.ts
@@ -72,6 +72,80 @@ const translation = {
   nTo1RetrievalLegacyLink: '詳細を見る',
   nTo1RetrievalLegacyLinkText: ' N-to-1 retrievalは9月に正式に廃止されます。',
   defaultRetrievalTip: 'デフォルトでは、マルチパス取得が使用されます。ナレッジは複数のナレッジ ベースから取得され、再ランク付けされます。',
+  editExternalAPIConfirmWarningContent: {
+    front: 'この外部ナレッジAPIは、',
+    end: '外部の知識、そしてこの変更はそれらすべてに適用されます。この変更を保存してもよろしいですか?',
+  },
+  editExternalAPIFormWarning: {
+    end: '外部の知識',
+    front: 'この外部APIはにリンクされています',
+  },
+  deleteExternalAPIConfirmWarningContent: {
+    title: {
+      end: '?',
+      front: '削除',
+    },
+    content: {
+      front: 'この外部ナレッジAPIは、',
+      end: '外部の知識。このAPIを削除すると、それらすべてが無効になります。この API を削除してもよろしいですか ?',
+    },
+    noConnectionContent: 'この API を削除してもよろしいですか ?',
+  },
+  selectExternalKnowledgeAPI: {
+    placeholder: '外部ナレッジ API を選択する',
+  },
+  connectDatasetIntro: {
+    content: {
+      link: '外部 API の作成方法を学ぶ',
+      front: '外部ナレッジ ベースに接続するには、まず外部 API を作成する必要があります。よくお読みになり、以下を参照してください。',
+      end: '.次に、対応するナレッジIDを見つけて、左側のフォームに入力します。すべての情報が正しい場合は、接続ボタンをクリックした後、ナレッジベースの検索テストに自動的にジャンプします。',
+    },
+    title: '外部ナレッジベースに接続する方法',
+    learnMore: '詳細情報',
+  },
+  connectHelper: {
+    helper2: '取得機能のみがサポートされています',
+    helper3: '.次のことを強くお勧めします。',
+    helper4: 'ヘルプドキュメントを読む',
+    helper5: 'この機能を使用する前に慎重に。',
+    helper1: 'APIとナレッジベースIDを介して外部ナレッジベースに接続します。',
+  },
+  externalKnowledgeForm: {
+    cancel: 'キャンセル',
+    connect: '繋ぐ',
+  },
+  externalAPIForm: {
+    encrypted: {
+      front: 'APIトークンは暗号化され、',
+      end: 'テクノロジー。',
+    },
+    apiKey: 'APIキー',
+    name: '名前',
+    edit: '編集',
+    save: 'セーブ',
+    cancel: 'キャンセル',
+    endpoint: 'API エンドポイント',
+  },
+  externalTag: '外',
+  editExternalAPITooltipTitle: 'リンクされた知識',
+  externalKnowledgeName: '外部ナレッジ名',
+  externalAPI: '外部 API',
+  externalAPIPanelDocumentation: 'External Knowledge API の作成方法を学ぶ',
+  editExternalAPIFormTitle: '外部ナレッジ API の編集',
+  externalAPIPanelTitle: '外部ナレッジAPI',
+  externalKnowledgeId: '外部ナレッジID',
+  connectDataset: '外部ナレッジベースへの接続',
+  externalKnowledgeIdPlaceholder: 'ナレッジIDを入力してください',
+  createNewExternalAPI: '新しい外部ナレッジ API を作成する',
+  noExternalKnowledge: 'External Knowledge APIはまだありませんので、こちらをクリックして作成してください',
+  mixtureInternalAndExternalTip: '再ランク付けモデルは、内部知識と外部知識の混合に必要です。',
+  learnHowToWriteGoodKnowledgeDescription: '良い知識の説明を書く方法を学ぶ',
+  externalKnowledgeNamePlaceholder: 'ナレッジベースの名前を入力してください',
+  externalKnowledgeDescription: 'ナレッジの説明',
+  createExternalAPI: '外部ナレッジ API を追加する',
+  externalKnowledgeDescriptionPlaceholder: 'このナレッジベースの内容を説明する(オプション)',
+  allExternalTip: '外部ナレッジのみを使用する場合、ユーザーは Rerank モデルを有効にするかどうかを選択できます。有効にしない場合、取得されたチャンクはスコアに基づいて並べ替えられます。異なるナレッジベースの検索戦略に一貫性がない場合、不正確になります。',
+  externalAPIPanelDescription: '外部ナレッジAPIは、Difyの外部のナレッジベースに接続し、そのナレッジベースからナレッジを取得するために使用されます。',
 }
 
 export default translation
diff --git a/web/i18n/ko-KR/app-api.ts b/web/i18n/ko-KR/app-api.ts
index 4708fac7dd22d..d518a9fa27c8f 100644
--- a/web/i18n/ko-KR/app-api.ts
+++ b/web/i18n/ko-KR/app-api.ts
@@ -80,6 +80,7 @@ const translation = {
     pathParams: '경로 매개변수',
     query: '쿼리',
   },
+  regenerate: '재생성',
 }
 
 export default translation
diff --git a/web/i18n/ko-KR/dataset-creation.ts b/web/i18n/ko-KR/dataset-creation.ts
index 6a4126d8dbe52..69d6cf7a55b82 100644
--- a/web/i18n/ko-KR/dataset-creation.ts
+++ b/web/i18n/ko-KR/dataset-creation.ts
@@ -73,6 +73,14 @@ const translation = {
       fireCrawlNotConfiguredDescription: 'API 키로 Firecrawl을 구성하여 사용합니다.',
       extractOnlyMainContent: '기본 콘텐츠만 추출합니다(머리글, 탐색, 바닥글 등 없음).',
       maxDepthTooltip: '입력한 URL을 기준으로 크롤링할 최대 수준입니다. 깊이 0은 입력 된 url의 페이지를 긁어 내고, 깊이 1은 url과 enteredURL + one / 이후의 모든 것을 긁어 모으는 식입니다.',
+      chooseProvider: '제공자 선택',
+      jinaReaderDocLink: 'https://jina.ai/reader',
+      useSitemap: '사이트맵 사용',
+      jinaReaderNotConfiguredDescription: '액세스를 위해 무료 API 키를 입력하여 Jina Reader를 설정합니다.',
+      jinaReaderDoc: 'Jina Reader에 대해 자세히 알아보기',
+      jinaReaderTitle: '전체 사이트를 Markdown으로 변환',
+      jinaReaderNotConfigured: 'Jina Reader가 구성되지 않았습니다.',
+      useSitemapTooltip: '사이트맵을 따라 사이트를 크롤링합니다. 그렇지 않은 경우 Jina Reader는 페이지 관련성에 따라 반복적으로 크롤링하여 더 적지만 더 높은 품질의 페이지를 생성합니다.',
     },
   },
   stepTwo: {
@@ -158,6 +166,11 @@ const translation = {
     apiKeyPlaceholder: 'firecrawl.dev 의 API 키',
     configFirecrawl: 'Firecrawl 구성 🔥',
   },
+  jinaReader: {
+    apiKeyPlaceholder: 'jina.ai 의 API 키',
+    getApiKeyLinkText: 'jina.ai 에서 무료 API 키 받기',
+    configJinaReader: 'Jina Reader 구성',
+  },
 }
 
 export default translation
diff --git a/web/i18n/ko-KR/dataset-hit-testing.ts b/web/i18n/ko-KR/dataset-hit-testing.ts
index df0dbe5208389..a2cbc55bbbd4e 100644
--- a/web/i18n/ko-KR/dataset-hit-testing.ts
+++ b/web/i18n/ko-KR/dataset-hit-testing.ts
@@ -23,6 +23,8 @@ const translation = {
   },
   noRecentTip: '최근 쿼리 결과가 없습니다.',
   viewChart: '벡터 차트 보기',
+  settingTitle: '검색 설정',
+  viewDetail: '자세히보기',
 }
 
 export default translation
diff --git a/web/i18n/ko-KR/dataset-settings.ts b/web/i18n/ko-KR/dataset-settings.ts
index ef451ee866506..11fd010eecd72 100644
--- a/web/i18n/ko-KR/dataset-settings.ts
+++ b/web/i18n/ko-KR/dataset-settings.ts
@@ -29,6 +29,9 @@ const translation = {
     save: '저장',
     permissionsInvitedMembers: '부분 팀 구성원',
     me: '(당신)',
+    externalKnowledgeAPI: '외부 지식 API',
+    externalKnowledgeID: '외부 지식 ID',
+    retrievalSettings: '검색 설정',
   },
 }
 
diff --git a/web/i18n/ko-KR/dataset.ts b/web/i18n/ko-KR/dataset.ts
index 9fd0dd16a7a64..1683b5fb7db85 100644
--- a/web/i18n/ko-KR/dataset.ts
+++ b/web/i18n/ko-KR/dataset.ts
@@ -71,6 +71,80 @@ const translation = {
   nTo1RetrievalLegacyLink: '자세히 알아보기',
   nTo1RetrievalLegacyLinkText: 'N-대-1 검색은 9월에 공식적으로 더 이상 사용되지 않습니다.',
   defaultRetrievalTip: '다중 경로 검색이 기본적으로 사용됩니다. 지식은 여러 기술 자료에서 검색된 다음 순위가 다시 매겨집니다.',
+  editExternalAPIConfirmWarningContent: {
+    front: '이 외부 지식 API는 다음에 연결됩니다.',
+    end: '외부 지식, 그리고 이 수정 사항은 그들 모두에게 적용될 것입니다. 이 변경 사항을 저장하시겠습니까?',
+  },
+  editExternalAPIFormWarning: {
+    end: '외부 지식',
+    front: '이 외부 API는 다음에 연결됩니다.',
+  },
+  deleteExternalAPIConfirmWarningContent: {
+    title: {
+      front: '삭제하다',
+      end: '?',
+    },
+    content: {
+      front: '이 외부 지식 API는 다음에 연결됩니다.',
+      end: '외부 지식. 이 API를 삭제하면 모두 무효화됩니다. 이 API를 삭제하시겠습니까?',
+    },
+    noConnectionContent: '이 API를 삭제하시겠습니까?',
+  },
+  selectExternalKnowledgeAPI: {
+    placeholder: '외부 지식 API 선택',
+  },
+  connectDatasetIntro: {
+    content: {
+      link: '외부 API를 만드는 방법 알아보기',
+      end: '. 그런 다음 해당 기술 ID를 찾아 왼쪽 양식에 입력합니다. 모든 정보가 올바르면 연결 단추를 클릭한 후 기술 자료의 검색 테스트로 자동으로 이동합니다.',
+      front: '외부 기술 자료에 연결하려면 먼저 외부 API를 만들어야 합니다. 주의 깊게 읽고 참조하십시오.',
+    },
+    learnMore: '더 알아보세요',
+    title: '외부 기술 자료에 연결하는 방법',
+  },
+  connectHelper: {
+    helper1: 'API 및 기술 자료 ID를 통해 외부 기술 자료에 연결합니다. 현재,',
+    helper4: '도움말 문서 읽기',
+    helper2: '검색 기능만 지원됩니다',
+    helper5: '이 기능을 사용하기 전에 주의하십시오.',
+    helper3: '. 다음을 강력히 권장합니다.',
+  },
+  externalKnowledgeForm: {
+    cancel: '취소',
+    connect: '연결하다',
+  },
+  externalAPIForm: {
+    encrypted: {
+      end: '기술.',
+      front: 'API 토큰은 다음을 사용하여 암호화되고 저장됩니다.',
+    },
+    save: '구해내다',
+    name: '이름',
+    endpoint: 'API 엔드포인트',
+    edit: '편집하다',
+    cancel: '취소',
+    apiKey: 'API 키',
+  },
+  editExternalAPITooltipTitle: '연결된 지식',
+  externalAPIPanelTitle: '외부 지식 API',
+  externalKnowledgeDescription: '지식 설명',
+  externalAPI: '외부 API',
+  externalKnowledgeName: '외부 지식 이름',
+  createExternalAPI: '외부 지식 API 추가',
+  externalTag: '외부',
+  editExternalAPIFormTitle: '외부 지식 API 편집',
+  externalKnowledgeNamePlaceholder: '기술 자료의 이름을 입력하십시오.',
+  externalAPIPanelDocumentation: '외부 지식 API를 만드는 방법 알아보기',
+  createNewExternalAPI: '새 외부 지식 API 만들기',
+  mixtureInternalAndExternalTip: '리랭크 모델은 내부 및 외부 지식의 혼합에 필요합니다.',
+  connectDataset: '외부 기술 자료에 연결',
+  learnHowToWriteGoodKnowledgeDescription: '적절한 지식 설명을 작성하는 방법 알아보기',
+  externalKnowledgeDescriptionPlaceholder: '이 기술 자료의 내용 설명(선택 사항)',
+  externalKnowledgeId: '외부 지식 ID',
+  externalKnowledgeIdPlaceholder: '지식 ID를 입력하십시오.',
+  allExternalTip: '외부 지식만 사용하는 경우 사용자는 리랭크 모델을 사용할지 여부를 선택할 수 있습니다. 활성화하지 않으면 검색된 청크가 점수에 따라 정렬됩니다. 서로 다른 기술 자료의 검색 전략이 일관되지 않으면 부정확합니다.',
+  externalAPIPanelDescription: '외부 지식 API는 Dify 외부의 기술 자료에 연결하고 해당 기술 자료에서 지식을 검색하는 데 사용됩니다.',
+  noExternalKnowledge: '아직 외부 지식 API가 없으므로 여기를 클릭하여 생성하십시오.',
 }
 
 export default translation
diff --git a/web/i18n/pl-PL/app-api.ts b/web/i18n/pl-PL/app-api.ts
index 05cad56fe2a6e..79ad3246fc066 100644
--- a/web/i18n/pl-PL/app-api.ts
+++ b/web/i18n/pl-PL/app-api.ts
@@ -97,6 +97,7 @@ const translation = {
     pathParams: 'Parametry ścieżki',
     query: 'Zapytanie',
   },
+  regenerate: 'Ponownie wygenerować',
 }
 
 export default translation
diff --git a/web/i18n/pl-PL/dataset-creation.ts b/web/i18n/pl-PL/dataset-creation.ts
index f5b36e62eec26..fa08dcd2f5298 100644
--- a/web/i18n/pl-PL/dataset-creation.ts
+++ b/web/i18n/pl-PL/dataset-creation.ts
@@ -74,6 +74,14 @@ const translation = {
       firecrawlTitle: 'Wyodrębnij zawartość internetową za pomocą 🔥Firecrawl',
       fireCrawlNotConfigured: 'Firecrawl nie jest skonfigurowany',
       maxDepthTooltip: 'Maksymalna głębokość przeszukiwania względem wprowadzonego adresu URL. Głębokość 0 po prostu zeskrobuje stronę z wprowadzonego adresu URL, głębokość 1 zeskrobuje adres URL i wszystko po wprowadzeniuURL+ jeden / i tak dalej.',
+      useSitemap: 'Użyj mapy witryny',
+      useSitemapTooltip: 'Postępuj zgodnie z mapą witryny, aby zindeksować witrynę. Jeśli nie, Jina Reader będzie indeksować iteracyjnie w oparciu o trafność strony, dając mniej stron, ale o wyższej jakości.',
+      chooseProvider: 'Wybierz dostawcę',
+      jinaReaderDocLink: 'https://jina.ai/reader',
+      jinaReaderNotConfigured: 'Czytnik Jina nie jest skonfigurowany',
+      jinaReaderDoc: 'Dowiedz się więcej o Jina Reader',
+      jinaReaderTitle: 'Konwertowanie całej witryny na język Markdown',
+      jinaReaderNotConfiguredDescription: 'Skonfiguruj Jina Reader, wprowadzając bezpłatny klucz API, aby uzyskać dostęp.',
     },
   },
   stepTwo: {
@@ -173,6 +181,11 @@ const translation = {
     configFirecrawl: 'Konfiguracja 🔥Firecrawla',
     getApiKeyLinkText: 'Pobierz klucz API z firecrawl.dev',
   },
+  jinaReader: {
+    getApiKeyLinkText: 'Odbierz darmowy klucz API na jina.ai',
+    apiKeyPlaceholder: 'Klucz API od jina.ai',
+    configJinaReader: 'Konfiguracja czytnika Jina',
+  },
 }
 
 export default translation
diff --git a/web/i18n/pl-PL/dataset-hit-testing.ts b/web/i18n/pl-PL/dataset-hit-testing.ts
index 25c6babc552b8..e9509a0f05372 100644
--- a/web/i18n/pl-PL/dataset-hit-testing.ts
+++ b/web/i18n/pl-PL/dataset-hit-testing.ts
@@ -23,6 +23,8 @@ const translation = {
   },
   noRecentTip: 'Brak ostatnich wyników zapytań tutaj',
   viewChart: 'Zobacz WYKRES WEKTOROWY',
+  settingTitle: 'Ustawienie pobierania',
+  viewDetail: 'Pokaż szczegóły',
 }
 
 export default translation
diff --git a/web/i18n/pl-PL/dataset-settings.ts b/web/i18n/pl-PL/dataset-settings.ts
index 3f069ab9b0ce2..0ed428d407978 100644
--- a/web/i18n/pl-PL/dataset-settings.ts
+++ b/web/i18n/pl-PL/dataset-settings.ts
@@ -34,6 +34,9 @@ const translation = {
     save: 'Zapisz',
     permissionsInvitedMembers: 'Częściowi członkowie zespołu',
     me: '(Ty)',
+    externalKnowledgeAPI: 'Interfejs API wiedzy zewnętrznej',
+    retrievalSettings: 'Ustawienia pobierania',
+    externalKnowledgeID: 'Zewnętrzny identyfikator wiedzy',
   },
 }
 
diff --git a/web/i18n/pl-PL/dataset.ts b/web/i18n/pl-PL/dataset.ts
index efd8b75c95cb7..4538b26ae2a9b 100644
--- a/web/i18n/pl-PL/dataset.ts
+++ b/web/i18n/pl-PL/dataset.ts
@@ -78,6 +78,80 @@ const translation = {
   nTo1RetrievalLegacyLink: 'Dowiedz się więcej',
   nTo1RetrievalLegacyLinkText: 'Wyszukiwanie N-do-1 zostanie oficjalnie wycofane we wrześniu.',
   defaultRetrievalTip: 'Pobieranie wielu ścieżek jest używane domyślnie. Wiedza jest pobierana z wielu baz wiedzy, a następnie ponownie klasyfikowana.',
+  editExternalAPIConfirmWarningContent: {
+    end: 'wiedzy zewnętrznej, a ta modyfikacja zostanie zastosowana do nich wszystkich. Czy na pewno chcesz zapisać tę zmianę?',
+    front: 'Ten interfejs API wiedzy zewnętrznej jest połączony z',
+  },
+  editExternalAPIFormWarning: {
+    front: 'Ten zewnętrzny interfejs API jest powiązany z',
+    end: 'Wiedza zewnętrzna',
+  },
+  deleteExternalAPIConfirmWarningContent: {
+    title: {
+      end: '?',
+      front: 'Usunąć',
+    },
+    content: {
+      front: 'Ten interfejs API wiedzy zewnętrznej jest połączony z',
+      end: 'wiedza zewnętrzna. Usunięcie tego interfejsu API spowoduje unieważnienie ich wszystkich. Czy na pewno chcesz usunąć ten interfejs API?',
+    },
+    noConnectionContent: 'Czy na pewno chcesz usunąć ten interfejs API?',
+  },
+  selectExternalKnowledgeAPI: {
+    placeholder: 'Wybieranie interfejsu API wiedzy zewnętrznej',
+  },
+  connectDatasetIntro: {
+    content: {
+      front: 'Aby nawiązać połączenie z zewnętrzną bazą wiedzy, należy najpierw utworzyć zewnętrzny interfejs API. Przeczytaj uważnie i zapoznaj się z',
+      link: 'Dowiedz się, jak utworzyć zewnętrzny interfejs API',
+      end: '. Następnie znajdź odpowiedni identyfikator wiedzy i wypełnij go w formularzu po lewej stronie. Jeśli wszystkie informacje są poprawne, po kliknięciu przycisku połączenia automatycznie przejdzie do testu wyszukiwania w bazie wiedzy.',
+    },
+    learnMore: 'Dowiedz się więcej',
+    title: 'Jak połączyć się z zewnętrzną bazą wiedzy',
+  },
+  connectHelper: {
+    helper1: 'Połącz się z zewnętrznymi bazami wiedzy za pośrednictwem interfejsu API i identyfikatora bazy wiedzy. Obecnie',
+    helper3: '. Zdecydowanie zalecamy, aby',
+    helper5: 'ostrożnie przed użyciem tej funkcji.',
+    helper4: 'Zapoznaj się z dokumentacją pomocy',
+    helper2: 'Obsługiwana jest tylko funkcja pobierania',
+  },
+  externalKnowledgeForm: {
+    connect: 'Połączyć',
+    cancel: 'Anuluj',
+  },
+  externalAPIForm: {
+    encrypted: {
+      end: 'Technologia.',
+      front: 'Twój token API zostanie zaszyfrowany i będzie przechowywany za pomocą',
+    },
+    edit: 'Redagować',
+    save: 'Zapisać',
+    name: 'Nazwa',
+    apiKey: 'Klucz API',
+    cancel: 'Anuluj',
+    endpoint: 'Punkt końcowy interfejsu API',
+  },
+  externalAPIPanelDocumentation: 'Dowiedz się, jak utworzyć interfejs API wiedzy zewnętrznej',
+  noExternalKnowledge: 'Nie ma jeszcze interfejsu API wiedzy zewnętrznej, kliknij tutaj, aby utworzyć',
+  createExternalAPI: 'Dodawanie interfejsu API wiedzy zewnętrznej',
+  connectDataset: 'Nawiązywanie połączenia z zewnętrzną bazą wiedzy',
+  editExternalAPITooltipTitle: 'POWIĄZANA WIEDZA',
+  externalKnowledgeId: 'Zewnętrzny identyfikator wiedzy',
+  externalAPIPanelTitle: 'Interfejs API wiedzy zewnętrznej',
+  externalKnowledgeName: 'Nazwa wiedzy zewnętrznej',
+  externalKnowledgeIdPlaceholder: 'Podaj identyfikator wiedzy',
+  createNewExternalAPI: 'Tworzenie nowego interfejsu API wiedzy zewnętrznej',
+  externalKnowledgeDescription: 'Opis wiedzy',
+  externalKnowledgeDescriptionPlaceholder: 'Opisz, co znajduje się w tej bazie wiedzy (opcjonalnie)',
+  allExternalTip: 'W przypadku korzystania tylko z wiedzy zewnętrznej użytkownik może zdecydować, czy chce włączyć model Rerank. Jeśli ta opcja nie jest włączona, pobrane fragmenty będą sortowane na podstawie wyników. Gdy strategie wyszukiwania z różnych baz wiedzy są niespójne, będzie to niedokładne.',
+  editExternalAPIFormTitle: 'Edytowanie interfejsu API wiedzy zewnętrznej',
+  mixtureInternalAndExternalTip: 'Model Rerank jest wymagany do połączenia wiedzy wewnętrznej i zewnętrznej.',
+  externalAPI: 'Zewnętrzny interfejs API',
+  externalTag: 'Zewnętrzny',
+  learnHowToWriteGoodKnowledgeDescription: 'Dowiedz się, jak napisać dobry opis wiedzy',
+  externalKnowledgeNamePlaceholder: 'Podaj nazwę bazy wiedzy',
+  externalAPIPanelDescription: 'Interfejs API wiedzy zewnętrznej służy do łączenia się z bazą wiedzy poza Dify i pobierania wiedzy z tej bazy wiedzy.',
 }
 
 export default translation
diff --git a/web/i18n/pt-BR/app-api.ts b/web/i18n/pt-BR/app-api.ts
index 7bbd25695de34..707c7514ee748 100644
--- a/web/i18n/pt-BR/app-api.ts
+++ b/web/i18n/pt-BR/app-api.ts
@@ -78,6 +78,7 @@ const translation = {
   loading: 'Carregamento',
   pause: 'Pausa',
   playing: 'Jogar',
+  regenerate: 'Regenerar',
 }
 
 export default translation
diff --git a/web/i18n/pt-BR/dataset-creation.ts b/web/i18n/pt-BR/dataset-creation.ts
index 511f0d5bcbd37..c9aeff197106c 100644
--- a/web/i18n/pt-BR/dataset-creation.ts
+++ b/web/i18n/pt-BR/dataset-creation.ts
@@ -73,6 +73,14 @@ const translation = {
       totalPageScraped: 'Total de páginas raspadas:',
       excludePaths: 'Excluir caminhos',
       extractOnlyMainContent: 'Extraia apenas o conteúdo principal (sem cabeçalhos, navs, rodapés, etc.)',
+      jinaReaderNotConfiguredDescription: 'Configure o Jina Reader inserindo sua chave de API gratuita para acesso.',
+      jinaReaderDoc: 'Saiba mais sobre o Jina Reader',
+      chooseProvider: 'Selecione um provedor',
+      jinaReaderNotConfigured: 'Jina Reader não está configurado',
+      jinaReaderDocLink: 'https://jina.ai/reader',
+      useSitemap: 'Usar o mapa do site',
+      useSitemapTooltip: 'Siga o mapa do site para rastrear o site. Caso contrário, o Jina Reader rastreará iterativamente com base na relevância da página, produzindo menos páginas, mas de maior qualidade.',
+      jinaReaderTitle: 'Converter todo o site em Markdown',
     },
   },
   stepTwo: {
@@ -157,6 +165,11 @@ const translation = {
     configFirecrawl: 'Configurar 🔥o Firecrawl',
     getApiKeyLinkText: 'Obtenha sua chave de API do firecrawl.dev',
   },
+  jinaReader: {
+    getApiKeyLinkText: 'Obtenha sua chave de API gratuita em jina.ai',
+    configJinaReader: 'Configurar o Jina Reader',
+    apiKeyPlaceholder: 'Chave de API do jina.ai',
+  },
 }
 
 export default translation
diff --git a/web/i18n/pt-BR/dataset-hit-testing.ts b/web/i18n/pt-BR/dataset-hit-testing.ts
index 11d1ddf87ef35..9e65c1cb0df23 100644
--- a/web/i18n/pt-BR/dataset-hit-testing.ts
+++ b/web/i18n/pt-BR/dataset-hit-testing.ts
@@ -23,6 +23,8 @@ const translation = {
   },
   noRecentTip: 'Nenhum resultado de consulta recente aqui',
   viewChart: 'Ver GRÁFICO DE VETORES',
+  viewDetail: 'Ver detalhes',
+  settingTitle: 'Configuração de recuperação',
 }
 
 export default translation
diff --git a/web/i18n/pt-BR/dataset-settings.ts b/web/i18n/pt-BR/dataset-settings.ts
index cfedbff337420..380a7e7f9d4ab 100644
--- a/web/i18n/pt-BR/dataset-settings.ts
+++ b/web/i18n/pt-BR/dataset-settings.ts
@@ -29,6 +29,9 @@ const translation = {
     save: 'Salvar',
     permissionsInvitedMembers: 'Membros parciais da equipe',
     me: '(Você)',
+    retrievalSettings: 'Configurações de recuperação',
+    externalKnowledgeID: 'ID de conhecimento externo',
+    externalKnowledgeAPI: 'API de conhecimento externo',
   },
 }
 
diff --git a/web/i18n/pt-BR/dataset.ts b/web/i18n/pt-BR/dataset.ts
index 3feeb99a8b4b8..fa0e74e399957 100644
--- a/web/i18n/pt-BR/dataset.ts
+++ b/web/i18n/pt-BR/dataset.ts
@@ -72,6 +72,80 @@ const translation = {
   nTo1RetrievalLegacyLinkText: 'A recuperação N-para-1 será oficialmente descontinuada em setembro.',
   intro6: 'como um plug-in de índice ChatGPT autônomo para publicar',
   defaultRetrievalTip: 'A recuperação de vários caminhos é usada por padrão. O conhecimento é recuperado de várias bases de dados de conhecimento e, em seguida, reclassificado.',
+  editExternalAPIConfirmWarningContent: {
+    front: 'Esta API de conhecimento externo está vinculada a',
+    end: 'conhecimento externo, e essa modificação será aplicada a todos eles. Tem certeza de que deseja salvar essa alteração?',
+  },
+  editExternalAPIFormWarning: {
+    end: 'Conhecimento externo',
+    front: 'Esta API externa está vinculada a',
+  },
+  deleteExternalAPIConfirmWarningContent: {
+    title: {
+      front: 'Excluir',
+      end: '?',
+    },
+    content: {
+      end: 'conhecimento externo. A exclusão dessa API invalidará todos eles. Tem certeza de que deseja excluir esta API?',
+      front: 'Esta API de conhecimento externo está vinculada a',
+    },
+    noConnectionContent: 'Tem certeza de que deseja excluir essa API?',
+  },
+  selectExternalKnowledgeAPI: {
+    placeholder: 'Escolher uma API de conhecimento externa',
+  },
+  connectDatasetIntro: {
+    content: {
+      front: 'Para se conectar a uma base de dados de conhecimento externa, você precisa primeiro criar uma API externa. Por favor, leia com atenção e consulte',
+      link: 'Saiba como criar uma API externa',
+      end: '. Em seguida, encontre o ID de conhecimento correspondente e preencha-o no formulário à esquerda. Se todas as informações estiverem corretas, ele pulará automaticamente para o teste de recuperação na base de conhecimento depois de clicar no botão conectar.',
+    },
+    learnMore: 'Saiba Mais',
+    title: 'Como se conectar a uma base de conhecimento externa',
+  },
+  connectHelper: {
+    helper3: '. Recomendamos fortemente que você',
+    helper5: 'cuidadosamente antes de usar esse recurso.',
+    helper2: 'apenas a funcionalidade de recuperação é suportada',
+    helper4: 'Leia a documentação de ajuda',
+    helper1: 'Conecte-se a bases de conhecimento externas por meio da API e do ID da base de conhecimento. Atualmente,',
+  },
+  externalKnowledgeForm: {
+    cancel: 'Cancelar',
+    connect: 'Ligar',
+  },
+  externalAPIForm: {
+    encrypted: {
+      front: 'Seu token de API será criptografado e armazenado usando',
+      end: 'Tecnologia.',
+    },
+    name: 'Nome',
+    apiKey: 'Chave de API',
+    cancel: 'Cancelar',
+    save: 'Salvar',
+    edit: 'Editar',
+    endpoint: 'API Endpoint',
+  },
+  externalAPI: 'API externa',
+  editExternalAPITooltipTitle: 'CONHECIMENTO VINCULADO',
+  noExternalKnowledge: 'Ainda não existe uma API de conhecimento externo, clique aqui para criar',
+  externalAPIPanelDescription: 'A API de conhecimento externo é usada para se conectar a uma base de conhecimento fora do Dify e recuperar o conhecimento dessa base de conhecimento.',
+  externalKnowledgeIdPlaceholder: 'Insira o ID de conhecimento',
+  externalKnowledgeDescriptionPlaceholder: 'Descreva o que há nesta Base de Dados de Conhecimento (opcional)',
+  connectDataset: 'Conectar-se a uma base de conhecimento externa',
+  createNewExternalAPI: 'Criar uma nova API de conhecimento externo',
+  allExternalTip: 'Ao usar apenas conhecimento externo, o usuário pode escolher se deseja habilitar o modelo de reclassificação. Se não estiver ativado, os blocos recuperados serão classificados com base nas pontuações. Quando as estratégias de recuperação de diferentes bases de conhecimento são inconsistentes, elas serão imprecisas.',
+  externalTag: 'Externo',
+  externalKnowledgeName: 'Nome do Conhecimento Externo',
+  externalKnowledgeId: 'ID de conhecimento externo',
+  externalAPIPanelTitle: 'API de conhecimento externo',
+  externalKnowledgeNamePlaceholder: 'Insira o nome da base de conhecimento',
+  createExternalAPI: 'Adicionar uma API de conhecimento externo',
+  editExternalAPIFormTitle: 'Editar a API de conhecimento externo',
+  mixtureInternalAndExternalTip: 'O modelo de Reclassificação é necessário para a mistura de conhecimento interno e externo.',
+  learnHowToWriteGoodKnowledgeDescription: 'Aprenda a escrever uma boa descrição de conhecimento',
+  externalAPIPanelDocumentation: 'Saiba como criar uma API de conhecimento externo',
+  externalKnowledgeDescription: 'Descrição do Conhecimento',
 }
 
 export default translation
diff --git a/web/i18n/ro-RO/app-api.ts b/web/i18n/ro-RO/app-api.ts
index e6a52ade426f1..e1fa8e1b06e29 100644
--- a/web/i18n/ro-RO/app-api.ts
+++ b/web/i18n/ro-RO/app-api.ts
@@ -78,6 +78,7 @@ const translation = {
     pathParams: 'Parametrii căii',
     query: 'Interogare',
   },
+  regenerate: 'Regenera',
 }
 
 export default translation
diff --git a/web/i18n/ro-RO/dataset-creation.ts b/web/i18n/ro-RO/dataset-creation.ts
index 4ea0b0475827d..2505348e7097c 100644
--- a/web/i18n/ro-RO/dataset-creation.ts
+++ b/web/i18n/ro-RO/dataset-creation.ts
@@ -73,6 +73,14 @@ const translation = {
       includeOnlyPaths: 'Includeți numai căi',
       totalPageScraped: 'Total pagini răzuite:',
       maxDepthTooltip: 'Adâncimea maximă de accesat cu crawlere în raport cu adresa URL introdusă. Adâncimea 0 doar răzuiește pagina URL-ului introdus, adâncimea 1 răzuiește url-ul și tot ceea ce după ce a introdusURL + un / și așa mai departe.',
+      jinaReaderDocLink: 'https://jina.ai/reader',
+      chooseProvider: 'Selectați un furnizor',
+      jinaReaderNotConfiguredDescription: 'Configurați Jina Reader introducând cheia API gratuită pentru acces.',
+      useSitemap: 'Utilizarea hărții site-ului',
+      jinaReaderDoc: 'Aflați mai multe despre Jina Reader',
+      jinaReaderTitle: 'Convertiți întregul site în Markdown',
+      jinaReaderNotConfigured: 'Jina Reader nu este configurat',
+      useSitemapTooltip: 'Urmați harta site-ului pentru a accesa cu crawlere site-ul. Dacă nu, Jina Reader va accesa cu crawlere iterativ în funcție de relevanța paginii, producând mai puține pagini, dar de calitate superioară.',
     },
   },
   stepTwo: {
@@ -157,6 +165,11 @@ const translation = {
     getApiKeyLinkText: 'Obțineți cheia API de la firecrawl.dev',
     apiKeyPlaceholder: 'Cheie API de la firecrawl.dev',
   },
+  jinaReader: {
+    configJinaReader: 'Configurați Jina Reader',
+    apiKeyPlaceholder: 'Cheie API de la jina.ai',
+    getApiKeyLinkText: 'Obțineți cheia API gratuită la jina.ai',
+  },
 }
 
 export default translation
diff --git a/web/i18n/ro-RO/dataset-hit-testing.ts b/web/i18n/ro-RO/dataset-hit-testing.ts
index 5dc5d70221950..9bf934420a137 100644
--- a/web/i18n/ro-RO/dataset-hit-testing.ts
+++ b/web/i18n/ro-RO/dataset-hit-testing.ts
@@ -23,6 +23,8 @@ const translation = {
   },
   noRecentTip: 'Nu există rezultate de interogare recente aici',
   viewChart: 'Vizualizați GRAFICUL VECTORIAL',
+  settingTitle: 'Setare de recuperare',
+  viewDetail: 'Vezi detalii',
 }
 
 export default translation
diff --git a/web/i18n/ro-RO/dataset-settings.ts b/web/i18n/ro-RO/dataset-settings.ts
index 54780f9c16e80..0452d568c6bd8 100644
--- a/web/i18n/ro-RO/dataset-settings.ts
+++ b/web/i18n/ro-RO/dataset-settings.ts
@@ -29,6 +29,9 @@ const translation = {
     save: 'Salvare',
     permissionsInvitedMembers: 'Membri parțiali ai echipei',
     me: '(Tu)',
+    externalKnowledgeID: 'ID de cunoștințe extern',
+    externalKnowledgeAPI: 'API de cunoștințe externe',
+    retrievalSettings: 'Setări de recuperare',
   },
 }
 
diff --git a/web/i18n/ro-RO/dataset.ts b/web/i18n/ro-RO/dataset.ts
index baf54f68e6600..8e35331983f1e 100644
--- a/web/i18n/ro-RO/dataset.ts
+++ b/web/i18n/ro-RO/dataset.ts
@@ -72,6 +72,80 @@ const translation = {
   nTo1RetrievalLegacyLink: 'Află mai multe',
   nTo1RetrievalLegacyLinkText: 'Recuperarea N-la-1 va fi oficial depreciată în septembrie.',
   defaultRetrievalTip: 'Recuperarea pe mai multe căi este utilizată în mod implicit. Cunoștințele sunt preluate din mai multe baze de cunoștințe și apoi reclasificate.',
+  editExternalAPIConfirmWarningContent: {
+    front: 'Acest API de cunoștințe externe este legat de',
+    end: 'cunoștințe externe, iar această modificare va fi aplicată tuturor. Sunteți sigur că doriți să salvați această modificare?',
+  },
+  editExternalAPIFormWarning: {
+    front: 'Acest API extern este legat de',
+    end: 'cunoștințe externe',
+  },
+  deleteExternalAPIConfirmWarningContent: {
+    title: {
+      front: 'Șterge',
+      end: '?',
+    },
+    content: {
+      front: 'Acest API de cunoștințe externe este legat de',
+      end: 'cunoștințe externe. Ștergerea acestui API le va invalida pe toate. Sunteți sigur că doriți să ștergeți acest API?',
+    },
+    noConnectionContent: 'Sunteți sigur că ștergeți acest API?',
+  },
+  selectExternalKnowledgeAPI: {
+    placeholder: 'Alegeți un API de cunoștințe extern',
+  },
+  connectDatasetIntro: {
+    content: {
+      end: '. Apoi găsiți ID-ul de cunoștințe corespunzător și completați-l în formularul din stânga. Dacă toate informațiile sunt corecte, va sări automat la testul de recuperare din baza de cunoștințe după ce faceți clic pe butonul de conectare.',
+      link: 'Aflați cum să creați un API extern',
+      front: 'Pentru a vă conecta la o bază de cunoștințe externă, trebuie mai întâi să creați un API extern. Vă rugăm să citiți cu atenție și să consultați',
+    },
+    title: 'Cum să vă conectați la o bază de cunoștințe externă',
+    learnMore: 'Află mai multe',
+  },
+  connectHelper: {
+    helper2: 'este acceptată doar funcționalitatea de recuperare',
+    helper5: 'Cu atenție înainte de a utiliza această caracteristică.',
+    helper3: '. Vă recomandăm cu tărie să',
+    helper4: 'Citiți documentația de ajutor',
+    helper1: 'Conectați-vă la baze de cunoștințe externe prin API și ID-ul bazei de cunoștințe. În prezent,',
+  },
+  externalKnowledgeForm: {
+    connect: 'Conecta',
+    cancel: 'Anula',
+  },
+  externalAPIForm: {
+    encrypted: {
+      end: 'Tehnologie.',
+      front: 'Tokenul API va fi criptat și stocat folosind',
+    },
+    edit: 'Editare',
+    endpoint: 'Punct final API',
+    apiKey: 'Cheie API',
+    name: 'Nume',
+    save: 'Salva',
+    cancel: 'Anula',
+  },
+  editExternalAPIFormTitle: 'Editarea API-ului de cunoștințe externe',
+  externalTag: 'Extern',
+  createExternalAPI: 'Adăugarea unui API de cunoștințe extern',
+  connectDataset: 'Conectați-vă la o bază de cunoștințe externă',
+  externalKnowledgeDescriptionPlaceholder: 'Descrieți ce este în această bază de cunoștințe (opțional)',
+  externalAPI: 'API extern',
+  learnHowToWriteGoodKnowledgeDescription: 'Aflați cum să scrieți o descriere bună a cunoștințelor',
+  externalAPIPanelTitle: 'API de cunoștințe externe',
+  allExternalTip: 'Când utilizează numai cunoștințe externe, utilizatorul poate alege dacă să activeze modelul Rerank. Dacă nu este activată, bucățile preluate vor fi sortate pe baza scorurilor. Când strategiile de recuperare a diferitelor baze de cunoștințe sunt inconsistente, acestea vor fi inexacte.',
+  externalKnowledgeNamePlaceholder: 'Vă rugăm să introduceți numele bazei de cunoștințe',
+  externalAPIPanelDocumentation: 'Aflați cum să creați un API de cunoștințe externe',
+  externalKnowledgeName: 'Nume cunoștințe externe',
+  externalKnowledgeDescription: 'Descrierea cunoștințelor',
+  externalKnowledgeIdPlaceholder: 'Vă rugăm să introduceți ID-ul de cunoștințe',
+  noExternalKnowledge: 'Nu există încă un API de cunoștințe externe, faceți clic aici pentru a crea',
+  externalKnowledgeId: 'ID de cunoștințe extern',
+  editExternalAPITooltipTitle: 'CUNOȘTINȚE LEGATE',
+  mixtureInternalAndExternalTip: 'Modelul Rerank este necesar pentru amestecul de cunoștințe interne și externe.',
+  externalAPIPanelDescription: 'API-ul de cunoștințe externe este utilizat pentru a se conecta la o bază de cunoștințe din afara Dify și pentru a prelua cunoștințe din acea bază de cunoștințe.',
+  createNewExternalAPI: 'Creați un nou API de cunoștințe externe',
 }
 
 export default translation
diff --git a/web/i18n/ru-RU/app-api.ts b/web/i18n/ru-RU/app-api.ts
index 0c56d7a5013bc..e6fe9021fecf4 100644
--- a/web/i18n/ru-RU/app-api.ts
+++ b/web/i18n/ru-RU/app-api.ts
@@ -78,6 +78,7 @@ const translation = {
     pathParams: 'Параметры пути',
     query: 'Запрос',
   },
+  regenerate: 'Регенерировать',
 }
 
 export default translation
diff --git a/web/i18n/ru-RU/dataset-creation.ts b/web/i18n/ru-RU/dataset-creation.ts
index c97daeeeceac7..ffa7f961d0fc6 100644
--- a/web/i18n/ru-RU/dataset-creation.ts
+++ b/web/i18n/ru-RU/dataset-creation.ts
@@ -78,6 +78,14 @@ const translation = {
       scrapTimeInfo: 'Всего просканировано {{total}} страниц за {{time}} секунд',
       preview: 'Предварительный просмотр',
       maxDepthTooltip: 'Максимальная глубина сканирования относительно введенного URL. Глубина 0 сканирует только страницу введенного URL, глубина 1 сканирует URL и все, что находится после введенного URL + один /, и так далее.',
+      jinaReaderNotConfiguredDescription: 'Настройте Jina Reader, введя свой бесплатный ключ API для доступа.',
+      jinaReaderDocLink: 'https://jina.ai/reader',
+      useSitemap: 'Использовать карту сайта',
+      chooseProvider: 'Выберите провайдера',
+      jinaReaderNotConfigured: 'Jina Reader не настроен',
+      jinaReaderDoc: 'Узнайте больше о Jina Reader',
+      jinaReaderTitle: 'Конвертируйте весь сайт в Markdown',
+      useSitemapTooltip: 'Следуйте карте сайта, чтобы просканировать сайт. Если нет, Jina Reader будет сканировать итеративно в зависимости от релевантности страницы, выдавая меньшее количество страниц, но более высокого качества.',
     },
   },
   stepTwo: {
@@ -157,6 +165,11 @@ const translation = {
     modelButtonConfirm: 'Подтвердить',
     modelButtonCancel: 'Отмена',
   },
+  jinaReader: {
+    getApiKeyLinkText: 'Получите бесплатный ключ API в jina.ai',
+    configJinaReader: 'Настройка Jina Reader',
+    apiKeyPlaceholder: 'Ключ API от jina.ai',
+  },
 }
 
 export default translation
diff --git a/web/i18n/ru-RU/dataset-hit-testing.ts b/web/i18n/ru-RU/dataset-hit-testing.ts
index 0d3a14a676d07..c8aeff7d5f9e8 100644
--- a/web/i18n/ru-RU/dataset-hit-testing.ts
+++ b/web/i18n/ru-RU/dataset-hit-testing.ts
@@ -23,6 +23,8 @@ const translation = {
   },
   noRecentTip: 'Здесь нет результатов недавних запросов',
   viewChart: 'Посмотреть ВЕКТОРНУЮ ДИАГРАММУ',
+  viewDetail: 'Подробнее',
+  settingTitle: 'Настройка извлечения',
 }
 
 export default translation
diff --git a/web/i18n/ru-RU/dataset-settings.ts b/web/i18n/ru-RU/dataset-settings.ts
index f732562d23752..c862d9ca468b7 100644
--- a/web/i18n/ru-RU/dataset-settings.ts
+++ b/web/i18n/ru-RU/dataset-settings.ts
@@ -29,6 +29,9 @@ const translation = {
       longDescription: ' о методе поиска, вы можете изменить это в любое время в настройках базы знаний.',
     },
     save: 'Сохранить',
+    externalKnowledgeAPI: 'API внешних знаний',
+    retrievalSettings: 'Настройки извлечения',
+    externalKnowledgeID: 'Внешний идентификатор базы знаний',
   },
 }
 
diff --git a/web/i18n/ru-RU/dataset.ts b/web/i18n/ru-RU/dataset.ts
index 4a7590ad6055b..cdff5061389f9 100644
--- a/web/i18n/ru-RU/dataset.ts
+++ b/web/i18n/ru-RU/dataset.ts
@@ -72,6 +72,80 @@ const translation = {
   nTo1RetrievalLegacyLink: 'Узнать больше',
   nTo1RetrievalLegacyLinkText: ' Поиск N-к-1 будет официально прекращен в сентябре.',
   defaultRetrievalTip: 'По умолчанию используется многоканальная проверка. Знания извлекаются из нескольких баз знаний, а затем повторно ранжируются.',
+  editExternalAPIConfirmWarningContent: {
+    end: 'внешних знаний, и эта модификация будет применена ко всем им. Вы уверены, что хотите сохранить это изменение?',
+    front: 'Этот API внешних знаний связан с',
+  },
+  editExternalAPIFormWarning: {
+    end: 'Внешние знания',
+    front: 'Этот внешний API связан с',
+  },
+  deleteExternalAPIConfirmWarningContent: {
+    title: {
+      end: '?',
+      front: 'Удалить',
+    },
+    content: {
+      front: 'Этот API внешних знаний связан с',
+      end: 'внешнее знание. Удаление этого API сделает их все недействительными. Вы уверены, что хотите удалить этот API?',
+    },
+    noConnectionContent: 'Вы уверены, что удалите этот API?',
+  },
+  selectExternalKnowledgeAPI: {
+    placeholder: 'Выбор API внешних знаний',
+  },
+  connectDatasetIntro: {
+    content: {
+      link: 'Узнайте, как создать внешний API',
+      front: 'Чтобы подключиться к внешней базе знаний, необходимо сначала создать внешний API. Пожалуйста, внимательно прочтите и обратитесь к',
+      end: '. Затем найдите соответствующий идентификатор знания и заполните его в форме слева. Если вся информация верна, он автоматически перейдет к тесту извлечения в базе знаний после нажатия кнопки подключения.',
+    },
+    learnMore: 'Подробнее',
+    title: 'Как подключиться к внешней базе знаний',
+  },
+  connectHelper: {
+    helper2: 'Поддерживается только функция извлечения',
+    helper3: '. Мы настоятельно рекомендуем вам',
+    helper4: 'Ознакомьтесь с справочной документацией',
+    helper5: 'Будьте внимательны перед использованием этой функции.',
+    helper1: 'Подключение к внешним базам знаний через API и идентификатор базы знаний. В настоящее время',
+  },
+  externalKnowledgeForm: {
+    connect: 'Соединять',
+    cancel: 'Отмена',
+  },
+  externalAPIForm: {
+    encrypted: {
+      end: 'Технологии.',
+      front: 'Ваш токен API будет зашифрован и сохранен с помощью',
+    },
+    cancel: 'Отмена',
+    endpoint: 'Конечная точка API',
+    save: 'Спасать',
+    edit: 'Редактировать',
+    name: 'Имя',
+    apiKey: 'Ключ API',
+  },
+  externalKnowledgeNamePlaceholder: 'Пожалуйста, введите название базы знаний',
+  externalTag: 'Внешний',
+  learnHowToWriteGoodKnowledgeDescription: 'Узнайте, как написать хорошее описание знаний',
+  externalAPIPanelTitle: 'API внешних знаний',
+  externalKnowledgeDescription: 'Описание знаний',
+  editExternalAPITooltipTitle: 'СВЯЗАННЫЕ ЗНАНИЯ',
+  externalKnowledgeName: 'Имя внешнего базы знаний',
+  createExternalAPI: 'Добавление API внешних знаний',
+  externalKnowledgeIdPlaceholder: 'Пожалуйста, введите идентификатор знаний',
+  externalKnowledgeDescriptionPlaceholder: 'Опишите, что входит в эту базу знаний (необязательно)',
+  noExternalKnowledge: 'У нас еще нет External Knowledge API, нажмите здесь, чтобы создать',
+  externalAPI: 'Внешний API',
+  externalKnowledgeId: 'Внешний идентификатор базы знаний',
+  createNewExternalAPI: 'Создание нового API внешних знаний',
+  editExternalAPIFormTitle: 'Редактирование API внешних знаний',
+  connectDataset: 'Подключение к внешней базе знаний',
+  mixtureInternalAndExternalTip: 'Модель Rerank необходима для смешивания внутренних и внешних знаний.',
+  allExternalTip: 'При использовании только внешних знаний пользователь может выбрать, следует ли включать модель повторного ранжирования. Если этот параметр не включен, полученные фрагменты будут сортироваться на основе баллов. Когда стратегии извлечения из разных баз знаний несовместимы, они будут неточными.',
+  externalAPIPanelDocumentation: 'Узнайте, как создать API внешних знаний',
+  externalAPIPanelDescription: 'Внешний API базы знаний используется для подключения к базе знаний за пределами Dify и извлечения знаний из этой базы знаний.',
 }
 
 export default translation
diff --git a/web/i18n/tr-TR/app-api.ts b/web/i18n/tr-TR/app-api.ts
index 9a64de546b26e..adb00ad062830 100644
--- a/web/i18n/tr-TR/app-api.ts
+++ b/web/i18n/tr-TR/app-api.ts
@@ -78,6 +78,7 @@ const translation = {
     pathParams: 'Path Params',
     query: 'Query',
   },
+  regenerate: 'Yenilemek',
 }
 
 export default translation
diff --git a/web/i18n/tr-TR/dataset-creation.ts b/web/i18n/tr-TR/dataset-creation.ts
index c29e3045b8cf1..90ccecf285621 100644
--- a/web/i18n/tr-TR/dataset-creation.ts
+++ b/web/i18n/tr-TR/dataset-creation.ts
@@ -78,6 +78,14 @@ const translation = {
       scrapTimeInfo: 'Toplam {{total}} sayfa {{time}}s içinde kazındı',
       preview: 'Önizleme',
       maxDepthTooltip: 'Girilen URL\'ye göre tarama için maksimum derinlik. Derinlik 0 sadece girilen url sayfasını kazır, derinlik 1 url ve girilen URL + bir / \'dan sonraki her şeyi kazır ve böyle devam eder.',
+      jinaReaderTitle: 'Tüm siteyi Markdown\'a dönüştürün',
+      useSitemap: 'Site haritasını kullan',
+      useSitemapTooltip: 'Siteyi taramak için site haritasını takip edin. Aksi takdirde, Jina Reader sayfa alaka düzeyine göre yinelemeli olarak tarar ve daha az ancak daha yüksek kaliteli sayfalar verir.',
+      jinaReaderNotConfiguredDescription: 'Erişim için ücretsiz API anahtarınızı girerek Jina Reader\'ı kurun.',
+      chooseProvider: 'Bir sağlayıcı seçin',
+      jinaReaderDoc: 'Jina Reader hakkında daha fazla bilgi edinin',
+      jinaReaderNotConfigured: 'Jina Reader yapılandırılmadı',
+      jinaReaderDocLink: 'https://jina.ai/reader',
     },
   },
   stepTwo: {
@@ -157,6 +165,11 @@ const translation = {
     modelButtonConfirm: 'Onayla',
     modelButtonCancel: 'İptal',
   },
+  jinaReader: {
+    apiKeyPlaceholder: 'jina.ai\'dan API anahtarı',
+    configJinaReader: 'Jina Reader\'ı Yapılandırma',
+    getApiKeyLinkText: 'Ücretsiz API anahtarınızı hemen jina.ai alın',
+  },
 }
 
 export default translation
diff --git a/web/i18n/tr-TR/dataset-hit-testing.ts b/web/i18n/tr-TR/dataset-hit-testing.ts
index 1635e3e331adf..545e0f074caf9 100644
--- a/web/i18n/tr-TR/dataset-hit-testing.ts
+++ b/web/i18n/tr-TR/dataset-hit-testing.ts
@@ -23,6 +23,8 @@ const translation = {
   },
   noRecentTip: 'Burada son sorgu sonuçları yok',
   viewChart: 'VEKTÖR GRAFİĞİNİ GÖRÜNTÜLE',
+  viewDetail: 'ayrıntılara bakın',
+  settingTitle: 'Alma Ayarı',
 }
 
 export default translation
diff --git a/web/i18n/tr-TR/dataset-settings.ts b/web/i18n/tr-TR/dataset-settings.ts
index 090602e8907a8..9bd716c063d12 100644
--- a/web/i18n/tr-TR/dataset-settings.ts
+++ b/web/i18n/tr-TR/dataset-settings.ts
@@ -29,6 +29,9 @@ const translation = {
       longDescription: ' geri alım yöntemi hakkında, bunu Bilgi ayarlarında istediğiniz zaman değiştirebilirsiniz.',
     },
     save: 'Kaydet',
+    retrievalSettings: 'Alma Ayarları',
+    externalKnowledgeAPI: 'Harici Bilgi API\'si',
+    externalKnowledgeID: 'Harici Bilgi Kimliği',
   },
 }
 
diff --git a/web/i18n/tr-TR/dataset.ts b/web/i18n/tr-TR/dataset.ts
index 5e55e071c5664..611917be7dabd 100644
--- a/web/i18n/tr-TR/dataset.ts
+++ b/web/i18n/tr-TR/dataset.ts
@@ -72,6 +72,80 @@ const translation = {
   nTo1RetrievalLegacyLink: 'Daha fazla bilgi edin',
   nTo1RetrievalLegacyLinkText: 'N-1 geri alma Eylül ayında resmi olarak kullanımdan kaldırılacaktır.',
   defaultRetrievalTip: 'Varsayılan olarak çok alma kullanılır. Bilgi, birden fazla bilgi tabanından alınır ve ardından yeniden sıralanır.',
+  editExternalAPIConfirmWarningContent: {
+    front: 'Bu Harici Bilgi API\'si aşağıdakilerle bağlantılıdır',
+    end: 'Dışsal bilgi ve bu değişiklik hepsine uygulanacaktır. Bu değişikliği kaydetmek istediğinizden emin misiniz?',
+  },
+  editExternalAPIFormWarning: {
+    end: 'Dış bilgi',
+    front: 'Bu Harici API aşağıdakilere bağlıdır:',
+  },
+  deleteExternalAPIConfirmWarningContent: {
+    title: {
+      front: 'Silmek',
+      end: '?',
+    },
+    content: {
+      front: 'Bu Harici Bilgi API\'si aşağıdakilerle bağlantılıdır',
+      end: 'dış bilgi. Bu API\'yi silmek hepsini geçersiz kılacaktır. Bu API\'yi silmek istediğinizden emin misiniz?',
+    },
+    noConnectionContent: 'Bu API\'yi sildiğinizden emin misiniz?',
+  },
+  selectExternalKnowledgeAPI: {
+    placeholder: 'Bir Harici Bilgi API\'si seçin',
+  },
+  connectDatasetIntro: {
+    content: {
+      front: 'Harici bir bilgi bankasına bağlanmak için önce harici bir API oluşturmanız gerekir. Lütfen dikkatlice okuyun ve bakınız',
+      end: '. Ardından ilgili bilgi kimliğini bulun ve soldaki forma doldurun. Tüm bilgiler doğruysa, bağlan düğmesine tıkladıktan sonra otomatik olarak bilgi tabanındaki alma testine atlayacaktır.',
+      link: 'Harici API oluşturmayı öğrenin',
+    },
+    title: 'Harici bir bilgi bankasına nasıl bağlanılır',
+    learnMore: 'Daha fazla bilgi edinin',
+  },
+  connectHelper: {
+    helper2: 'Yalnızca alma işlevi desteklenir',
+    helper5: 'Bu özelliği kullanmadan önce dikkatlice kullanın.',
+    helper3: '. Şunları yapmanızı şiddetle tavsiye ederiz',
+    helper1: 'API ve bilgi bankası kimliği aracılığıyla harici bilgi bankalarına bağlanın. Şu anda,',
+    helper4: 'Yardım belgelerini okuyun',
+  },
+  externalKnowledgeForm: {
+    connect: 'Bağlamak',
+    cancel: 'İptal',
+  },
+  externalAPIForm: {
+    encrypted: {
+      end: 'Teknoloji.',
+      front: 'API Token\'ınız kullanılarak şifrelenecek ve saklanacaktır.',
+    },
+    save: 'Kurtarmak',
+    cancel: 'İptal',
+    endpoint: 'API Uç Noktası',
+    edit: 'Düzenlemek',
+    name: 'Ad',
+    apiKey: 'API Anahtarı',
+  },
+  externalTag: 'Dış',
+  externalAPI: 'Harici API',
+  externalKnowledgeDescription: 'Bilgi Açıklaması',
+  externalAPIPanelDescription: 'Harici bilgi API\'si, Dify dışındaki bir bilgi bankasına bağlanmak ve bu bilgi bankasından bilgi almak için kullanılır.',
+  externalKnowledgeDescriptionPlaceholder: 'Bu Bilgi Bankası\'nda neler olduğunu açıklayın (isteğe bağlı)',
+  externalAPIPanelDocumentation: 'External Knowledge API\'nin nasıl oluşturulacağını öğrenin',
+  mixtureInternalAndExternalTip: 'Rerank modeli, iç ve dış bilgilerin karışımı için gereklidir.',
+  externalKnowledgeName: 'Dış Bilgi Adı',
+  connectDataset: 'Harici bir bilgi bankasına bağlanın',
+  editExternalAPITooltipTitle: 'BAĞLANTILI BILGI',
+  externalAPIPanelTitle: 'Harici Bilgi API\'si',
+  editExternalAPIFormTitle: 'External Knowledge API\'yi düzenleme',
+  externalKnowledgeIdPlaceholder: 'Lütfen Bilgi Kimliğini girin',
+  learnHowToWriteGoodKnowledgeDescription: 'İyi bir bilgi açıklamasının nasıl yazılacağını öğrenin',
+  externalKnowledgeNamePlaceholder: 'Lütfen bilgi bankasının adını giriniz',
+  noExternalKnowledge: 'Henüz Harici Bilgi API\'si yok, oluşturmak için buraya tıklayın',
+  allExternalTip: 'Yalnızca harici bilgileri kullanırken, kullanıcı Rerank modelinin etkinleştirilip etkinleştirilmeyeceğini seçebilir. Etkinleştirilmezse, alınan parçalar puanlara göre sıralanır. Farklı bilgi tabanlarının erişim stratejileri tutarsız olduğunda, yanlış olacaktır.',
+  externalKnowledgeId: 'Harici Bilgi Kimliği',
+  createExternalAPI: 'Harici bilgi API\'si ekleme',
+  createNewExternalAPI: 'Yeni bir External Knowledge API oluşturma',
 }
 
 export default translation
diff --git a/web/i18n/uk-UA/app-api.ts b/web/i18n/uk-UA/app-api.ts
index 6f46e9a9b4240..e7dd34bf4acff 100644
--- a/web/i18n/uk-UA/app-api.ts
+++ b/web/i18n/uk-UA/app-api.ts
@@ -78,6 +78,7 @@ const translation = {
     parametersApiTip: 'Отримання налаштованих вхідних параметрів, включаючи імена змінних, імена полів, типи та значення за замовчуванням. Зазвичай використовується для відображення цих полів у формі або заповнення значень за замовчуванням після завантаження клієнта.',
   },
   loading: 'Завантаження',
+  regenerate: 'Відновити',
 }
 
 export default translation
diff --git a/web/i18n/uk-UA/dataset-creation.ts b/web/i18n/uk-UA/dataset-creation.ts
index 5b2c9503cf1c4..57f51971cb880 100644
--- a/web/i18n/uk-UA/dataset-creation.ts
+++ b/web/i18n/uk-UA/dataset-creation.ts
@@ -73,6 +73,14 @@ const translation = {
       fireCrawlNotConfiguredDescription: 'Налаштуйте Firecrawl за допомогою ключа API, щоб використовувати його.',
       extractOnlyMainContent: 'Витягуйте лише основний контент (без заголовків, навігаторів, нижніх колонтитулів тощо)',
       maxDepthTooltip: 'Максимальна глибина для сканування щодо введеної URL-адреси. Глибина 0 просто зішкрібає сторінку введеного url, глибина 1 шкребе url і все після введеногоURL + один /, і так далі.',
+      jinaReaderDocLink: 'https://jina.ai/reader',
+      chooseProvider: 'Оберіть провайдера',
+      jinaReaderNotConfiguredDescription: 'Налаштуйте Jina Reader, ввівши безкоштовний API-ключ для доступу.',
+      jinaReaderDoc: 'Дізнайтеся більше про Jina Reader',
+      useSitemapTooltip: 'Дотримуйтесь карти сайту, щоб просканувати сайт. Якщо ні, Jina Reader скануватиметься ітеративно залежно від релевантності сторінки, отримуючи менше, але якісніших сторінок.',
+      jinaReaderNotConfigured: 'Jina Reader не налаштована',
+      jinaReaderTitle: 'Перетворіть весь сайт на Markdown',
+      useSitemap: 'Використовуйте карту сайту',
     },
   },
   stepTwo: {
@@ -157,6 +165,11 @@ const translation = {
     configFirecrawl: 'Налаштування 🔥Firecrawl',
     apiKeyPlaceholder: 'Ключ API від firecrawl.dev',
   },
+  jinaReader: {
+    apiKeyPlaceholder: 'Ключ API від jina.ai',
+    configJinaReader: 'Налаштування Jina Reader',
+    getApiKeyLinkText: 'Отримайте безкоштовний API-ключ за адресою jina.ai',
+  },
 }
 
 export default translation
diff --git a/web/i18n/uk-UA/dataset-hit-testing.ts b/web/i18n/uk-UA/dataset-hit-testing.ts
index 252d8c8a4d270..d348e56a7096e 100644
--- a/web/i18n/uk-UA/dataset-hit-testing.ts
+++ b/web/i18n/uk-UA/dataset-hit-testing.ts
@@ -23,6 +23,8 @@ const translation = {
   },
   noRecentTip: 'Тут немає результатів останніх запитів',
   viewChart: 'ПЕРЕГЛЯНУТИ ВЕКТОРНУ ДІАГРАМУ',
+  settingTitle: 'Налаштування отримання',
+  viewDetail: 'Переглянути деталі',
 }
 
 export default translation
diff --git a/web/i18n/uk-UA/dataset-settings.ts b/web/i18n/uk-UA/dataset-settings.ts
index 85e80902a7ee3..4532ed4addea4 100644
--- a/web/i18n/uk-UA/dataset-settings.ts
+++ b/web/i18n/uk-UA/dataset-settings.ts
@@ -29,6 +29,9 @@ const translation = {
     save: 'Зберегти',
     me: '(Ви)',
     permissionsInvitedMembers: 'Часткові члени команди',
+    externalKnowledgeAPI: 'API зовнішніх знань',
+    externalKnowledgeID: 'Зовнішній ідентифікатор знань',
+    retrievalSettings: 'Налаштування отримання',
   },
 }
 
diff --git a/web/i18n/uk-UA/dataset.ts b/web/i18n/uk-UA/dataset.ts
index 1bf6786976066..3e1349a617dfc 100644
--- a/web/i18n/uk-UA/dataset.ts
+++ b/web/i18n/uk-UA/dataset.ts
@@ -73,6 +73,80 @@ const translation = {
   nTo1RetrievalLegacyLink: 'Дізнатися більше',
   nTo1RetrievalLegacyLinkText: 'N-до-1 пошук буде офіційно застарілим у вересні.',
   defaultRetrievalTip: 'За замовчуванням використовується отримання кількома шляхами. Знання витягуються з кількох баз знань, а потім заново ранжуються.',
+  editExternalAPIConfirmWarningContent: {
+    front: 'Цей API зовнішніх знань пов\'язаний з',
+    end: 'зовнішні знання, і ця модифікація буде застосована до всіх них. Ви впевнені, що хочете зберегти цю зміну?',
+  },
+  editExternalAPIFormWarning: {
+    end: 'Зовнішні знання',
+    front: 'Цей зовнішній API пов\'язаний з',
+  },
+  deleteExternalAPIConfirmWarningContent: {
+    title: {
+      front: 'Видалити',
+      end: '?',
+    },
+    content: {
+      front: 'Цей API зовнішніх знань пов\'язаний з',
+      end: 'зовнішні знання. Видалення цього API призведе до втрати чинності всіх з них. Ви впевнені, що хочете видалити цей API?',
+    },
+    noConnectionContent: 'Ви впевнені, що видалили цей API?',
+  },
+  selectExternalKnowledgeAPI: {
+    placeholder: 'Виберіть API зовнішніх знань',
+  },
+  connectDatasetIntro: {
+    content: {
+      link: 'Дізнайтеся, як створити зовнішній API',
+      front: 'Щоб підключитися до зовнішньої бази знань, спочатку потрібно створити зовнішній API. Будь ласка, уважно прочитайте та зверніться до',
+      end: '. Потім знайдіть відповідний ідентифікатор знань і заповніть його у формі зліва. Якщо вся інформація вірна, він автоматично перейде до пошукового тесту в базі знань після натискання кнопки підключення.',
+    },
+    title: 'Як підключитися до зовнішньої бази знань',
+    learnMore: 'Дізнатися більше',
+  },
+  connectHelper: {
+    helper2: 'Підтримується лише функція пошуку',
+    helper4: 'Прочитайте довідкову документацію',
+    helper3: '. Ми настійно рекомендуємо вам',
+    helper5: 'Уважно перед використанням цієї функції.',
+    helper1: 'Підключайтеся до зовнішніх баз знань через API та ідентифікатор бази знань. Наразі',
+  },
+  externalKnowledgeForm: {
+    cancel: 'Скасувати',
+    connect: 'Підключатися',
+  },
+  externalAPIForm: {
+    encrypted: {
+      end: 'Технології.',
+      front: 'Ваш токен API буде зашифровано та збережено за допомогою',
+    },
+    name: 'Ім\'я',
+    edit: 'Редагувати',
+    apiKey: 'Ключ API',
+    cancel: 'Скасувати',
+    save: 'Рятувати',
+    endpoint: 'Кінцева точка API',
+  },
+  externalKnowledgeDescriptionPlaceholder: 'Опишіть, що міститься в цій базі знань (необов\'язково)',
+  externalAPIPanelDocumentation: 'Дізнайтеся, як створити API зовнішніх знань',
+  externalAPIPanelTitle: 'API зовнішніх знань',
+  externalKnowledgeDescription: 'Опис знань',
+  noExternalKnowledge: 'API зовнішніх знань поки що не існує, натисніть тут, щоб створити',
+  externalTag: 'Зовнішній',
+  externalKnowledgeId: 'Зовнішній ідентифікатор знань',
+  externalAPI: 'Зовнішній API',
+  createExternalAPI: 'Додавання зовнішнього API знань',
+  createNewExternalAPI: 'Створення нового API зовнішніх знань',
+  externalKnowledgeName: 'Зовнішнє найменування знань',
+  externalKnowledgeNamePlaceholder: 'Будь ласка, введіть назву бази знань',
+  editExternalAPITooltipTitle: 'ПОВ\'ЯЗАНІ ЗНАННЯ',
+  editExternalAPIFormTitle: 'Редагування API зовнішніх знань',
+  connectDataset: 'Підключення до зовнішньої бази знань',
+  learnHowToWriteGoodKnowledgeDescription: 'Навчіться писати хороший опис знань',
+  allExternalTip: 'При використанні тільки зовнішніх знань користувач може вибрати, чи включати модель Rerank. Якщо цей параметр не увімкнено, отримані фрагменти будуть відсортовані на основі оцінок. Коли стратегії пошуку різних баз знань несумісні, це буде неточно.',
+  mixtureInternalAndExternalTip: 'Модель Rerank необхідна для поєднання внутрішніх і зовнішніх знань.',
+  externalKnowledgeIdPlaceholder: 'Будь ласка, введіть Knowledge ID',
+  externalAPIPanelDescription: 'API зовнішніх знань використовується для підключення до бази знань за межами Dify і отримання знань із цієї бази знань.',
 }
 
 export default translation
diff --git a/web/i18n/vi-VN/app-api.ts b/web/i18n/vi-VN/app-api.ts
index 2d4ee90e3c9d0..93f1fb0197319 100644
--- a/web/i18n/vi-VN/app-api.ts
+++ b/web/i18n/vi-VN/app-api.ts
@@ -78,6 +78,7 @@ const translation = {
     query: 'Truy vấn',
   },
   loading: 'Tải',
+  regenerate: 'Tái tạo',
 }
 
 export default translation
diff --git a/web/i18n/vi-VN/dataset-creation.ts b/web/i18n/vi-VN/dataset-creation.ts
index af49575b906c4..a2e87d667edc0 100644
--- a/web/i18n/vi-VN/dataset-creation.ts
+++ b/web/i18n/vi-VN/dataset-creation.ts
@@ -73,6 +73,14 @@ const translation = {
       resetAll: 'Đặt lại tất cả',
       crawlSubPage: 'Thu thập dữ liệu các trang phụ',
       maxDepthTooltip: 'Độ sâu tối đa cần thu thập dữ liệu so với URL đã nhập. Độ sâu 0 chỉ cần cạo trang của url đã nhập, độ sâu 1 cạo url và mọi thứ sau khi nhậpURL + một /, v.v.',
+      jinaReaderTitle: 'Chuyển đổi toàn bộ trang web thành Markdown',
+      jinaReaderDoc: 'Tìm hiểu thêm về Jina Reader',
+      useSitemap: 'Sử dụng sơ đồ trang web',
+      chooseProvider: 'Chọn nhà cung cấp',
+      jinaReaderDocLink: 'https://jina.ai/reader',
+      jinaReaderNotConfigured: 'Jina Reader không được cấu hình',
+      jinaReaderNotConfiguredDescription: 'Thiết lập Jina Reader bằng cách nhập khóa API miễn phí của bạn để truy cập.',
+      useSitemapTooltip: 'Thực hiện theo sơ đồ trang web để thu thập dữ liệu trang web. Nếu không, Jina Reader sẽ thu thập dữ liệu lặp đi lặp lại dựa trên mức độ liên quan của trang, mang lại ít trang hơn nhưng chất lượng cao hơn.',
     },
   },
   stepTwo: {
@@ -157,6 +165,11 @@ const translation = {
     configFirecrawl: 'Định cấu hình 🔥Firecrawl',
     apiKeyPlaceholder: 'Khóa API từ firecrawl.dev',
   },
+  jinaReader: {
+    getApiKeyLinkText: 'Nhận khóa API miễn phí của bạn tại jina.ai',
+    configJinaReader: 'Định cấu hình Jina Reader',
+    apiKeyPlaceholder: 'Khóa API từ jina.ai',
+  },
 }
 
 export default translation
diff --git a/web/i18n/vi-VN/dataset-hit-testing.ts b/web/i18n/vi-VN/dataset-hit-testing.ts
index a512888250610..7972eba288687 100644
--- a/web/i18n/vi-VN/dataset-hit-testing.ts
+++ b/web/i18n/vi-VN/dataset-hit-testing.ts
@@ -23,6 +23,8 @@ const translation = {
   },
   noRecentTip: 'Không có kết quả truy vấn gần đây',
   viewChart: 'Xem BIỂU ĐỒ VECTOR',
+  settingTitle: 'Cài đặt truy xuất',
+  viewDetail: 'Xem chi tiết',
 }
 
 export default translation
diff --git a/web/i18n/vi-VN/dataset-settings.ts b/web/i18n/vi-VN/dataset-settings.ts
index cc68bea7ae3d8..48ff2f038d6e1 100644
--- a/web/i18n/vi-VN/dataset-settings.ts
+++ b/web/i18n/vi-VN/dataset-settings.ts
@@ -29,6 +29,9 @@ const translation = {
     save: 'Lưu',
     permissionsInvitedMembers: 'Thành viên một phần trong nhóm',
     me: '(Bạn)',
+    externalKnowledgeAPI: 'API kiến thức bên ngoài',
+    retrievalSettings: 'Cài đặt truy xuất',
+    externalKnowledgeID: 'ID kiến thức bên ngoài',
   },
 }
 
diff --git a/web/i18n/vi-VN/dataset.ts b/web/i18n/vi-VN/dataset.ts
index a2b9f8d087115..eb7ef414f3e17 100644
--- a/web/i18n/vi-VN/dataset.ts
+++ b/web/i18n/vi-VN/dataset.ts
@@ -72,6 +72,80 @@ const translation = {
   nTo1RetrievalLegacyLink: 'Tìm hiểu thêm',
   nTo1RetrievalLegacyLinkText: 'Truy xuất N-đến-1 sẽ chính thức bị loại bỏ vào tháng 9.',
   defaultRetrievalTip: 'Truy xuất nhiều đường dẫn được sử dụng theo mặc định. Kiến thức được lấy từ nhiều cơ sở kiến thức và sau đó được xếp hạng lại.',
+  editExternalAPIConfirmWarningContent: {
+    front: 'API Kiến thức Bên ngoài này được liên kết với',
+    end: 'kiến thức bên ngoài, và sửa đổi này sẽ được áp dụng cho tất cả chúng. Bạn có chắc chắn muốn lưu thay đổi này không?',
+  },
+  editExternalAPIFormWarning: {
+    front: 'API bên ngoài này được liên kết với',
+    end: 'Kiến thức bên ngoài',
+  },
+  deleteExternalAPIConfirmWarningContent: {
+    title: {
+      end: '?',
+      front: 'Xóa',
+    },
+    content: {
+      front: 'API Kiến thức Bên ngoài này được liên kết với',
+      end: 'kiến thức bên ngoài. Xóa API này sẽ làm mất hiệu lực tất cả chúng. Bạn có chắc chắn muốn xóa API này không?',
+    },
+    noConnectionContent: 'Bạn có chắc chắn xóa API này không?',
+  },
+  selectExternalKnowledgeAPI: {
+    placeholder: 'Chọn một API kiến thức bên ngoài',
+  },
+  connectDatasetIntro: {
+    content: {
+      end: '. Sau đó tìm ID kiến thức tương ứng và điền vào biểu mẫu bên trái. Nếu tất cả thông tin là chính xác, nó sẽ tự động chuyển đến bài kiểm tra truy xuất trong cơ sở kiến thức sau khi nhấp vào nút kết nối.',
+      front: 'Để kết nối với cơ sở kiến thức bên ngoài, trước tiên bạn cần tạo API bên ngoài. Vui lòng đọc kỹ và tham khảo',
+      link: 'Tìm hiểu cách tạo API bên ngoài',
+    },
+    learnMore: 'Tìm hiểu thêm',
+    title: 'Cách kết nối với cơ sở kiến thức bên ngoài',
+  },
+  connectHelper: {
+    helper3: '. Chúng tôi thực sự khuyên bạn nên',
+    helper4: 'Đọc tài liệu trợ giúp',
+    helper1: 'Kết nối với cơ sở kiến thức bên ngoài thông qua API và ID cơ sở kiến thức. Hiện tại,',
+    helper2: 'Chỉ hỗ trợ chức năng truy xuất',
+    helper5: 'cẩn thận trước khi sử dụng tính năng này.',
+  },
+  externalKnowledgeForm: {
+    cancel: 'Hủy',
+    connect: 'Kết nối',
+  },
+  externalAPIForm: {
+    encrypted: {
+      end: 'Công nghệ.',
+      front: 'Mã thông báo API của bạn sẽ được mã hóa và lưu trữ bằng cách sử dụng',
+    },
+    apiKey: 'Khóa API',
+    endpoint: 'Điểm cuối API',
+    edit: 'Biên tập',
+    cancel: 'Hủy',
+    name: 'Tên',
+    save: 'Cứu',
+  },
+  learnHowToWriteGoodKnowledgeDescription: 'Tìm hiểu cách viết mô tả kiến thức tốt',
+  noExternalKnowledge: 'Chưa có API Kiến thức Bên ngoài, hãy nhấp vào đây để tạo',
+  connectDataset: 'Kết nối với cơ sở kiến thức bên ngoài',
+  externalTag: 'Ngoài',
+  externalAPIPanelTitle: 'API kiến thức bên ngoài',
+  editExternalAPITooltipTitle: 'KIẾN THỨC LIÊN KẾT',
+  externalKnowledgeNamePlaceholder: 'Vui lòng nhập tên của cơ sở kiến thức',
+  createExternalAPI: 'Thêm API kiến thức bên ngoài',
+  externalKnowledgeDescription: 'Mô tả kiến thức',
+  externalAPIPanelDocumentation: 'Tìm hiểu cách tạo API Kiến thức Bên ngoài',
+  allExternalTip: 'Khi chỉ sử dụng kiến thức bên ngoài, người dùng có thể chọn có bật mô hình Xếp hạng lại hay không. Nếu không được bật, các đoạn được truy xuất sẽ được sắp xếp dựa trên điểm số. Khi các chiến lược truy xuất của các cơ sở kiến thức khác nhau không nhất quán, nó sẽ không chính xác.',
+  editExternalAPIFormTitle: 'Chỉnh sửa API Kiến thức Bên ngoài',
+  externalKnowledgeId: 'ID kiến thức bên ngoài',
+  mixtureInternalAndExternalTip: 'Mô hình Rerank là cần thiết cho sự kết hợp giữa kiến thức bên trong và bên ngoài.',
+  externalAPI: 'API bên ngoài',
+  externalKnowledgeDescriptionPlaceholder: 'Mô tả nội dung trong Cơ sở Kiến thức này (tùy chọn)',
+  externalKnowledgeName: 'Tên kiến thức bên ngoài',
+  externalKnowledgeIdPlaceholder: 'Vui lòng nhập ID kiến thức',
+  createNewExternalAPI: 'Tạo API Kiến thức Bên ngoài mới',
+  externalAPIPanelDescription: 'API kiến thức bên ngoài được sử dụng để kết nối với cơ sở kiến thức bên ngoài Dify và truy xuất kiến thức từ cơ sở kiến thức đó.',
 }
 
 export default translation
diff --git a/web/i18n/zh-Hans/dataset-settings.ts b/web/i18n/zh-Hans/dataset-settings.ts
index f0f5190412013..423703e04ca7e 100644
--- a/web/i18n/zh-Hans/dataset-settings.ts
+++ b/web/i18n/zh-Hans/dataset-settings.ts
@@ -31,6 +31,7 @@ const translation = {
     externalKnowledgeAPI: '外部知识 API',
     externalKnowledgeID: '外部知识库 ID',
     save: '保存',
+    retrievalSettings: '检索设置',
   },
 }
 
diff --git a/web/i18n/zh-Hant/app-api.ts b/web/i18n/zh-Hant/app-api.ts
index 63a977837872e..0bbd7ffa84fc8 100644
--- a/web/i18n/zh-Hant/app-api.ts
+++ b/web/i18n/zh-Hant/app-api.ts
@@ -78,6 +78,7 @@ const translation = {
     pathParams: 'Path Params',
     query: 'Query',
   },
+  regenerate: '重新生成',
 }
 
 export default translation
diff --git a/web/i18n/zh-Hant/dataset-creation.ts b/web/i18n/zh-Hant/dataset-creation.ts
index 73a57db6a06c4..265731d1f8b6a 100644
--- a/web/i18n/zh-Hant/dataset-creation.ts
+++ b/web/i18n/zh-Hant/dataset-creation.ts
@@ -73,6 +73,14 @@ const translation = {
       unknownError: '未知錯誤',
       fireCrawlNotConfigured: '未配置 Firecrawl',
       maxDepthTooltip: '相對於輸入的 URL 的最大爬網深度。深度 0 只是抓取輸入的 url 的頁面，深度 1 抓取 url 以及 enteredURL + 1 / 之後的所有內容，依此類推。',
+      jinaReaderNotConfigured: '未配置 Jina Reader',
+      jinaReaderNotConfiguredDescription: '通過輸入您的免費 API 金鑰進行訪問來設置 Jina Reader。',
+      chooseProvider: '選擇供應商',
+      jinaReaderDoc: '了解有關 Jina Reader 的更多資訊',
+      jinaReaderTitle: '將整個網站轉換為 Markdown',
+      useSitemap: '使用網站地圖',
+      jinaReaderDocLink: 'https://jina.ai/reader',
+      useSitemapTooltip: '按照網站地圖對網站進行爬網。否則，Jina Reader 將根據頁面相關性反覆運算抓取，從而生成更少但品質更高的頁面。',
     },
   },
   stepTwo: {
@@ -157,6 +165,11 @@ const translation = {
     apiKeyPlaceholder: '來自 firecrawl.dev 的 API 金鑰',
     getApiKeyLinkText: '從 firecrawl.dev 獲取 API 金鑰',
   },
+  jinaReader: {
+    configJinaReader: '配置 Jina Reader',
+    getApiKeyLinkText: '在 jina.ai 獲取您的免費 API 金鑰',
+    apiKeyPlaceholder: '來自 jina.ai 的 API 金鑰',
+  },
 }
 
 export default translation
diff --git a/web/i18n/zh-Hant/dataset-hit-testing.ts b/web/i18n/zh-Hant/dataset-hit-testing.ts
index aed73497de5ad..72ca8f8d48630 100644
--- a/web/i18n/zh-Hant/dataset-hit-testing.ts
+++ b/web/i18n/zh-Hant/dataset-hit-testing.ts
@@ -24,6 +24,7 @@ const translation = {
   noRecentTip: '最近無查詢結果',
   viewChart: '查看向量圖表',
   viewDetail: '查看詳情',
+  settingTitle: '檢索設置',
 }
 
 export default translation
diff --git a/web/i18n/zh-Hant/dataset-settings.ts b/web/i18n/zh-Hant/dataset-settings.ts
index f34d1d4acccd7..a528befc5dd51 100644
--- a/web/i18n/zh-Hant/dataset-settings.ts
+++ b/web/i18n/zh-Hant/dataset-settings.ts
@@ -29,6 +29,9 @@ const translation = {
     save: '儲存',
     permissionsInvitedMembers: '部分團隊成員',
     me: '（您）',
+    externalKnowledgeID: '外部知識ID',
+    externalKnowledgeAPI: '外部知識 API',
+    retrievalSettings: '檢索設置',
   },
 }
 
diff --git a/web/i18n/zh-Hant/dataset.ts b/web/i18n/zh-Hant/dataset.ts
index 1888a28631a3c..30dfd6b98cceb 100644
--- a/web/i18n/zh-Hant/dataset.ts
+++ b/web/i18n/zh-Hant/dataset.ts
@@ -72,6 +72,80 @@ const translation = {
   nTo1RetrievalLegacyLink: '了解更多',
   nTo1RetrievalLegacyLinkText: 'N對1檢索將於9月正式棄用。',
   defaultRetrievalTip: '默認情況下，使用多路徑檢索。從多個知識庫中檢索知識，然後重新排名。',
+  editExternalAPIConfirmWarningContent: {
+    end: 'external knowledge，並且此修改將應用於所有這些 Knowledge。是否確實要保存此更改？',
+    front: '此外部知識 API 連結到',
+  },
+  editExternalAPIFormWarning: {
+    end: '外部知識',
+    front: '此外部 API 連結到',
+  },
+  deleteExternalAPIConfirmWarningContent: {
+    title: {
+      end: '?',
+      front: '刪除',
+    },
+    content: {
+      front: '此外部知識 API 連結到',
+      end: '外部知識。刪除此 API 將使所有這些 API 失效。是否確實要刪除此 API？',
+    },
+    noConnectionContent: '您確定要刪除此 API 嗎？',
+  },
+  selectExternalKnowledgeAPI: {
+    placeholder: '選擇外部知識 API',
+  },
+  connectDatasetIntro: {
+    content: {
+      link: '瞭解如何創建外部 API',
+      front: '要連接到外部知識庫，您需要先創建外部 API。請仔細閱讀並參考',
+      end: '.然後找到對應的知識ID並在左側的表單中填寫。如果資訊全部正確，點擊連接按鈕后，會自動跳轉到知識庫中的檢索測試。',
+    },
+    title: '如何連接到外部知識庫',
+    learnMore: '瞭解更多資訊',
+  },
+  connectHelper: {
+    helper5: '使用此功能前請小心。',
+    helper3: '.我們強烈建議您',
+    helper2: '僅支援檢索功能',
+    helper4: '閱讀幫助文件',
+    helper1: '通過 API 和知識庫 ID 連接到外部知識庫。目前，',
+  },
+  externalKnowledgeForm: {
+    connect: '連接',
+    cancel: '取消',
+  },
+  externalAPIForm: {
+    encrypted: {
+      end: '科技。',
+      front: '您的 API 令牌將使用',
+    },
+    save: '救',
+    cancel: '取消',
+    name: '名字',
+    apiKey: 'API 金鑰',
+    edit: '編輯',
+    endpoint: 'API 終端節點',
+  },
+  externalTag: '外部',
+  externalKnowledgeDescription: '知識描述',
+  mixtureInternalAndExternalTip: 'Rerank 模型是內部和外部知識的混合所必需的。',
+  connectDataset: '連接到外部知識庫',
+  learnHowToWriteGoodKnowledgeDescription: '瞭解如何編寫良好的知識描述',
+  createExternalAPI: '添加外部知識 API',
+  externalAPIPanelTitle: '外部知識 API',
+  createNewExternalAPI: '創建新的外部知識 API',
+  externalKnowledgeDescriptionPlaceholder: '描述此知識庫中的內容（選擇）',
+  allExternalTip: '僅使用外部知識時，用戶可以選擇是否啟用 Rerank 模型。如果未啟用，則檢索到的數據塊將根據分數進行排序。當不同知識庫的檢索策略不一致時，就會不準確。',
+  externalKnowledgeIdPlaceholder: '請輸入 Knowledge ID',
+  editExternalAPIFormTitle: '編輯外部知識 API',
+  externalKnowledgeId: '外部知識ID',
+  externalAPIPanelDescription: '外部知識 API 用於連接到 Dify 外部的知識庫，並從該知識庫中檢索知識。',
+  externalAPI: '外部 API',
+  editExternalAPITooltipTitle: '關聯知識',
+  externalKnowledgeName: '外部知識名稱',
+  externalAPIPanelDocumentation: '瞭解如何創建外部知識 API',
+  externalKnowledgeNamePlaceholder: '請輸入知識庫的名稱',
+  noExternalKnowledge: '目前還沒有外部知識 API，按兩下此處創建',
 }
 
 export default translation

From fa837b2dfdb0cd4804fcccc78f4f1d909be82ad1 Mon Sep 17 00:00:00 2001
From: zhuhao <37029601+hwzhuhao@users.noreply.github.com>
Date: Mon, 30 Sep 2024 17:14:13 +0800
Subject: [PATCH 117/235] fix: fix the issue with the system model
 configuration update (#8923)

---
 api/controllers/console/workspace/models.py | 5 +++--
 1 file changed, 3 insertions(+), 2 deletions(-)

diff --git a/api/controllers/console/workspace/models.py b/api/controllers/console/workspace/models.py
index dc88f6b8127fb..3138a260b3842 100644
--- a/api/controllers/console/workspace/models.py
+++ b/api/controllers/console/workspace/models.py
@@ -72,8 +72,9 @@ def post(self):
                     provider=model_setting["provider"],
                     model=model_setting["model"],
                 )
-            except Exception:
-                logging.warning(f"{model_setting['model_type']} save error")
+            except Exception as ex:
+                logging.exception(f"{model_setting['model_type']} save error: {ex}")
+                raise ex
 
         return {"result": "success"}
 

From 1af4ca344ede6f2ee8d97d0171241f0995200c5e Mon Sep 17 00:00:00 2001
From: KVOJJJin <jzongcode@gmail.com>
Date: Mon, 30 Sep 2024 17:18:47 +0800
Subject: [PATCH 118/235] Feat: add debounce for search in logs (#8924)

---
 web/app/components/app/annotation/index.tsx   |  8 +++++---
 web/app/components/app/log/index.tsx          | 14 ++++++++------
 web/app/components/app/workflow-log/index.tsx | 10 ++++++----
 3 files changed, 19 insertions(+), 13 deletions(-)

diff --git a/web/app/components/app/annotation/index.tsx b/web/app/components/app/annotation/index.tsx
index c66aaef6ce491..0bdd12c550512 100644
--- a/web/app/components/app/annotation/index.tsx
+++ b/web/app/components/app/annotation/index.tsx
@@ -3,6 +3,7 @@ import type { FC } from 'react'
 import React, { useEffect, useState } from 'react'
 import { useTranslation } from 'react-i18next'
 import { Pagination } from 'react-headless-pagination'
+import { useDebounce } from 'ahooks'
 import { ArrowLeftIcon, ArrowRightIcon } from '@heroicons/react/24/outline'
 import Toast from '../../base/toast'
 import Filter from './filter'
@@ -67,10 +68,11 @@ const Annotation: FC<Props> = ({
 
   const [queryParams, setQueryParams] = useState<QueryParam>({})
   const [currPage, setCurrPage] = React.useState<number>(0)
+  const debouncedQueryParams = useDebounce(queryParams, { wait: 500 })
   const query = {
     page: currPage + 1,
     limit: APP_PAGE_LIMIT,
-    keyword: queryParams.keyword || '',
+    keyword: debouncedQueryParams.keyword || '',
   }
 
   const [controlUpdateList, setControlUpdateList] = useState(Date.now())
@@ -232,8 +234,8 @@ const Annotation: FC<Props> = ({
             middlePagesSiblingCount={1}
             setCurrentPage={setCurrPage}
             totalPages={Math.ceil(total / APP_PAGE_LIMIT)}
-            truncatableClassName="w-8 px-0.5 text-center"
-            truncatableText="..."
+            truncableClassName="w-8 px-0.5 text-center"
+            truncableText="..."
           >
             <Pagination.PrevButton
               disabled={currPage === 0}
diff --git a/web/app/components/app/log/index.tsx b/web/app/components/app/log/index.tsx
index 77fcb6fcf94ab..7c0e5881ed4a4 100644
--- a/web/app/components/app/log/index.tsx
+++ b/web/app/components/app/log/index.tsx
@@ -4,6 +4,7 @@ import React, { useState } from 'react'
 import useSWR from 'swr'
 import { usePathname } from 'next/navigation'
 import { Pagination } from 'react-headless-pagination'
+import { useDebounce } from 'ahooks'
 import { omit } from 'lodash-es'
 import dayjs from 'dayjs'
 import { ArrowLeftIcon, ArrowRightIcon } from '@heroicons/react/24/outline'
@@ -59,6 +60,7 @@ const Logs: FC<ILogsProps> = ({ appDetail }) => {
     sort_by: '-created_at',
   })
   const [currPage, setCurrPage] = React.useState<number>(0)
+  const debouncedQueryParams = useDebounce(queryParams, { wait: 500 })
 
   // Get the app type first
   const isChatMode = appDetail.mode !== 'completion'
@@ -66,14 +68,14 @@ const Logs: FC<ILogsProps> = ({ appDetail }) => {
   const query = {
     page: currPage + 1,
     limit: APP_PAGE_LIMIT,
-    ...(queryParams.period !== 'all'
+    ...(debouncedQueryParams.period !== 'all'
       ? {
-        start: dayjs().subtract(queryParams.period as number, 'day').startOf('day').format('YYYY-MM-DD HH:mm'),
+        start: dayjs().subtract(debouncedQueryParams.period as number, 'day').startOf('day').format('YYYY-MM-DD HH:mm'),
         end: dayjs().endOf('day').format('YYYY-MM-DD HH:mm'),
       }
       : {}),
-    ...(isChatMode ? { sort_by: queryParams.sort_by } : {}),
-    ...omit(queryParams, ['period']),
+    ...(isChatMode ? { sort_by: debouncedQueryParams.sort_by } : {}),
+    ...omit(debouncedQueryParams, ['period']),
   }
 
   const getWebAppType = (appType: AppMode) => {
@@ -119,8 +121,8 @@ const Logs: FC<ILogsProps> = ({ appDetail }) => {
             middlePagesSiblingCount={1}
             setCurrentPage={setCurrPage}
             totalPages={Math.ceil(total / APP_PAGE_LIMIT)}
-            truncatableClassName="w-8 px-0.5 text-center"
-            truncatableText="..."
+            truncableClassName="w-8 px-0.5 text-center"
+            truncableText="..."
           >
             <Pagination.PrevButton
               disabled={currPage === 0}
diff --git a/web/app/components/app/workflow-log/index.tsx b/web/app/components/app/workflow-log/index.tsx
index b7e20f0a0b5cf..ff746d73858e6 100644
--- a/web/app/components/app/workflow-log/index.tsx
+++ b/web/app/components/app/workflow-log/index.tsx
@@ -4,6 +4,7 @@ import React, { useState } from 'react'
 import useSWR from 'swr'
 import { usePathname } from 'next/navigation'
 import { Pagination } from 'react-headless-pagination'
+import { useDebounce } from 'ahooks'
 import { ArrowLeftIcon, ArrowRightIcon } from '@heroicons/react/24/outline'
 import { Trans, useTranslation } from 'react-i18next'
 import Link from 'next/link'
@@ -51,12 +52,13 @@ const Logs: FC<ILogsProps> = ({ appDetail }) => {
   const { t } = useTranslation()
   const [queryParams, setQueryParams] = useState<QueryParam>({ status: 'all' })
   const [currPage, setCurrPage] = React.useState<number>(0)
+  const debouncedQueryParams = useDebounce(queryParams, { wait: 500 })
 
   const query = {
     page: currPage + 1,
     limit: APP_PAGE_LIMIT,
-    ...(queryParams.status !== 'all' ? { status: queryParams.status } : {}),
-    ...(queryParams.keyword ? { keyword: queryParams.keyword } : {}),
+    ...(debouncedQueryParams.status !== 'all' ? { status: debouncedQueryParams.status } : {}),
+    ...(debouncedQueryParams.keyword ? { keyword: debouncedQueryParams.keyword } : {}),
   }
 
   const getWebAppType = (appType: AppMode) => {
@@ -93,8 +95,8 @@ const Logs: FC<ILogsProps> = ({ appDetail }) => {
             middlePagesSiblingCount={1}
             setCurrentPage={setCurrPage}
             totalPages={Math.ceil(total / APP_PAGE_LIMIT)}
-            truncatableClassName="w-8 px-0.5 text-center"
-            truncatableText="..."
+            truncableClassName="w-8 px-0.5 text-center"
+            truncableText="..."
           >
             <Pagination.PrevButton
               disabled={currPage === 0}

From d2ce4960f121783ff970c9ffe7183a7bdb1b9e01 Mon Sep 17 00:00:00 2001
From: -LAN- <laipz8200@outlook.com>
Date: Mon, 30 Sep 2024 18:33:20 +0800
Subject: [PATCH 119/235] chore(versioning): bump version to 0.9.0 (#8911)

---
 api/configs/packaging/__init__.py | 2 +-
 docker-legacy/docker-compose.yaml | 6 +++---
 docker/docker-compose.yaml        | 6 +++---
 web/package.json                  | 2 +-
 4 files changed, 8 insertions(+), 8 deletions(-)

diff --git a/api/configs/packaging/__init__.py b/api/configs/packaging/__init__.py
index c7526601225d8..2a6fb0a4117db 100644
--- a/api/configs/packaging/__init__.py
+++ b/api/configs/packaging/__init__.py
@@ -9,7 +9,7 @@ class PackagingInfo(BaseSettings):
 
     CURRENT_VERSION: str = Field(
         description="Dify version",
-        default="0.8.3",
+        default="0.9.0",
     )
 
     COMMIT_SHA: str = Field(
diff --git a/docker-legacy/docker-compose.yaml b/docker-legacy/docker-compose.yaml
index 1636bb6a214bd..435af75827651 100644
--- a/docker-legacy/docker-compose.yaml
+++ b/docker-legacy/docker-compose.yaml
@@ -2,7 +2,7 @@ version: '3'
 services:
   # API service
   api:
-    image: langgenius/dify-api:0.8.3
+    image: langgenius/dify-api:0.9.0
     restart: always
     environment:
       # Startup mode, 'api' starts the API server.
@@ -227,7 +227,7 @@ services:
   # worker service
   # The Celery worker for processing the queue.
   worker:
-    image: langgenius/dify-api:0.8.3
+    image: langgenius/dify-api:0.9.0
     restart: always
     environment:
       CONSOLE_WEB_URL: ''
@@ -396,7 +396,7 @@ services:
 
   # Frontend web application.
   web:
-    image: langgenius/dify-web:0.8.3
+    image: langgenius/dify-web:0.9.0
     restart: always
     environment:
       # The base URL of console application api server, refers to the Console base URL of WEB service if console domain is
diff --git a/docker/docker-compose.yaml b/docker/docker-compose.yaml
index 95e271a0e9e4f..07f1130e67376 100644
--- a/docker/docker-compose.yaml
+++ b/docker/docker-compose.yaml
@@ -213,7 +213,7 @@ x-shared-env: &shared-api-worker-env
 services:
   # API service
   api:
-    image: langgenius/dify-api:0.8.3
+    image: langgenius/dify-api:0.9.0
     restart: always
     environment:
       # Use the shared environment variables.
@@ -233,7 +233,7 @@ services:
   # worker service
   # The Celery worker for processing the queue.
   worker:
-    image: langgenius/dify-api:0.8.3
+    image: langgenius/dify-api:0.9.0
     restart: always
     environment:
       # Use the shared environment variables.
@@ -252,7 +252,7 @@ services:
 
   # Frontend web application.
   web:
-    image: langgenius/dify-web:0.8.3
+    image: langgenius/dify-web:0.9.0
     restart: always
     environment:
       CONSOLE_API_URL: ${CONSOLE_API_URL:-}
diff --git a/web/package.json b/web/package.json
index b775d87184a9e..7ed8174acce9c 100644
--- a/web/package.json
+++ b/web/package.json
@@ -1,6 +1,6 @@
 {
   "name": "dify-web",
-  "version": "0.8.3",
+  "version": "0.9.0",
   "private": true,
   "engines": {
     "node": ">=18.17.0"

From a651b73db0b85c1ed7493f373000ddfbea52b909 Mon Sep 17 00:00:00 2001
From: Jyong <76649700+JohnJyong@users.noreply.github.com>
Date: Mon, 30 Sep 2024 21:17:12 +0800
Subject: [PATCH 120/235] original dataset update issue (#8935)

---
 api/services/dataset_service.py | 2 ++
 1 file changed, 2 insertions(+)

diff --git a/api/services/dataset_service.py b/api/services/dataset_service.py
index c165def6d5c1e..91d9254c6731b 100644
--- a/api/services/dataset_service.py
+++ b/api/services/dataset_service.py
@@ -250,6 +250,8 @@ def update_dataset(dataset_id, data, user):
             db.session.commit()
         else:
             data.pop("partial_member_list", None)
+            data.pop("external_knowledge_api_id", None)
+            data.pop("external_knowledge_id", None)
             filtered_data = {k: v for k, v in data.items() if v is not None or k == "description"}
             action = None
             if dataset.indexing_technique != data["indexing_technique"]:

From 7873e455bb1027f484b09a1303cb019a10e048bf Mon Sep 17 00:00:00 2001
From: NFish <douxc512@gmail.com>
Date: Mon, 30 Sep 2024 21:27:11 +0800
Subject: [PATCH 121/235] fix: Fix the error when importing web pages using
 jina (#8937)

---
 web/service/datasets.ts | 3 ++-
 1 file changed, 2 insertions(+), 1 deletion(-)

diff --git a/web/service/datasets.ts b/web/service/datasets.ts
index b3bb19b9f3781..90411efd4e19f 100644
--- a/web/service/datasets.ts
+++ b/web/service/datasets.ts
@@ -30,8 +30,9 @@ import type {
   createDocumentResponse,
 } from '@/models/datasets'
 import type { CreateKnowledgeBaseReq } from '@/app/components/datasets/external-knowledge-base/create/declarations'
-import type { CreateExternalAPIReq } from '@/app/components/datasets/external-api/declarations.ts'
+import type { CreateExternalAPIReq } from '@/app/components/datasets/external-api/declarations'
 import type { CommonResponse, DataSourceNotionWorkspace } from '@/models/common'
+import { DataSourceProvider } from '@/models/common'
 import type {
   ApiKeysListResponse,
   CreateApiKeyResponse,

From 3069b5cf57e6363f83d123168b6b7af26e048dcf Mon Sep 17 00:00:00 2001
From: Jyong <76649700+JohnJyong@users.noreply.github.com>
Date: Mon, 30 Sep 2024 22:01:32 +0800
Subject: [PATCH 122/235] original dataset update remove unuseful parameters
 (#8939)

---
 api/services/dataset_service.py | 1 +
 1 file changed, 1 insertion(+)

diff --git a/api/services/dataset_service.py b/api/services/dataset_service.py
index 91d9254c6731b..b8f80a9f77e2a 100644
--- a/api/services/dataset_service.py
+++ b/api/services/dataset_service.py
@@ -252,6 +252,7 @@ def update_dataset(dataset_id, data, user):
             data.pop("partial_member_list", None)
             data.pop("external_knowledge_api_id", None)
             data.pop("external_knowledge_id", None)
+            data.pop("external_retrieval_model", None)
             filtered_data = {k: v for k, v in data.items() if v is not None or k == "description"}
             action = None
             if dataset.indexing_technique != data["indexing_technique"]:

From d6a44e99903cd3ad4cafd802c1822c0ca6687877 Mon Sep 17 00:00:00 2001
From: Yi Xiao <54782454+YIXIAO0@users.noreply.github.com>
Date: Mon, 30 Sep 2024 22:10:27 +0800
Subject: [PATCH 123/235] fix: request params for internal dataset (#8940)

---
 .../dataset-config/settings-modal/index.tsx    | 18 ++++++++++--------
 .../datasets/settings/form/index.tsx           | 16 +++++++++-------
 2 files changed, 19 insertions(+), 15 deletions(-)

diff --git a/web/app/components/app/configuration/dataset-config/settings-modal/index.tsx b/web/app/components/app/configuration/dataset-config/settings-modal/index.tsx
index 7b3b4abc0fb1e..ee211325cf68e 100644
--- a/web/app/components/app/configuration/dataset-config/settings-modal/index.tsx
+++ b/web/app/components/app/configuration/dataset-config/settings-modal/index.tsx
@@ -59,10 +59,10 @@ const SettingsModal: FC<SettingsModalProps> = ({
   const { t } = useTranslation()
   const { notify } = useToastContext()
   const ref = useRef(null)
+  const isExternal = currentDataset.provider === 'external'
   const [topK, setTopK] = useState(currentDataset?.external_retrieval_model.top_k ?? 2)
   const [scoreThreshold, setScoreThreshold] = useState(currentDataset?.external_retrieval_model.score_threshold ?? 0.5)
   const [scoreThresholdEnabled, setScoreThresholdEnabled] = useState(currentDataset?.external_retrieval_model.score_threshold_enabled ?? false)
-
   const { setShowAccountSettingModal } = useModalContext()
   const [loading, setLoading] = useState(false)
   const { isCurrentWorkspaceDatasetOperator } = useAppContext()
@@ -122,19 +122,21 @@ const SettingsModal: FC<SettingsModalProps> = ({
           description,
           permission,
           indexing_technique: indexMethod,
-          external_retrieval_model: {
-            top_k: topK,
-            score_threshold: scoreThreshold,
-            score_threshold_enabled: scoreThresholdEnabled,
-          },
           retrieval_model: {
             ...postRetrievalConfig,
             score_threshold: postRetrievalConfig.score_threshold_enabled ? postRetrievalConfig.score_threshold : 0,
           },
-          external_knowledge_id: currentDataset!.external_knowledge_info.external_knowledge_id,
-          external_knowledge_api_id: currentDataset!.external_knowledge_info.external_knowledge_api_id,
           embedding_model: localeCurrentDataset.embedding_model,
           embedding_model_provider: localeCurrentDataset.embedding_model_provider,
+          ...(isExternal && {
+            external_knowledge_id: currentDataset!.external_knowledge_info.external_knowledge_id,
+            external_knowledge_api_id: currentDataset!.external_knowledge_info.external_knowledge_api_id,
+            external_retrieval_model: {
+              top_k: topK,
+              score_threshold: scoreThreshold,
+              score_threshold_enabled: scoreThresholdEnabled,
+            },
+          }),
         },
       } as any
       if (permission === 'partial_members') {
diff --git a/web/app/components/datasets/settings/form/index.tsx b/web/app/components/datasets/settings/form/index.tsx
index ed0865f87144a..a3b6e79c3bd14 100644
--- a/web/app/components/datasets/settings/form/index.tsx
+++ b/web/app/components/datasets/settings/form/index.tsx
@@ -141,19 +141,21 @@ const Form = () => {
           description,
           permission,
           indexing_technique: indexMethod,
-          external_retrieval_model: {
-            top_k: topK,
-            score_threshold: scoreThreshold,
-            score_threshold_enabled: scoreThresholdEnabled,
-          },
           retrieval_model: {
             ...postRetrievalConfig,
             score_threshold: postRetrievalConfig.score_threshold_enabled ? postRetrievalConfig.score_threshold : 0,
           },
-          external_knowledge_id: currentDataset!.external_knowledge_info.external_knowledge_id,
-          external_knowledge_api_id: currentDataset!.external_knowledge_info.external_knowledge_api_id,
           embedding_model: embeddingModel.model,
           embedding_model_provider: embeddingModel.provider,
+          ...(currentDataset!.provider === 'external' && {
+            external_knowledge_id: currentDataset!.external_knowledge_info.external_knowledge_id,
+            external_knowledge_api_id: currentDataset!.external_knowledge_info.external_knowledge_api_id,
+            external_retrieval_model: {
+              top_k: topK,
+              score_threshold: scoreThreshold,
+              score_threshold_enabled: scoreThresholdEnabled,
+            },
+          }),
         },
       } as any
       if (permission === 'partial_members') {

From 730d3a6d7ca2badc70c21c7476adf9bf560f2c21 Mon Sep 17 00:00:00 2001
From: -LAN- <laipz8200@outlook.com>
Date: Mon, 30 Sep 2024 22:13:38 +0800
Subject: [PATCH 124/235] chore(version): bump to 0.9.1 (#8938)

---
 api/configs/packaging/__init__.py | 2 +-
 docker-legacy/docker-compose.yaml | 6 +++---
 docker/docker-compose.yaml        | 6 +++---
 web/package.json                  | 2 +-
 4 files changed, 8 insertions(+), 8 deletions(-)

diff --git a/api/configs/packaging/__init__.py b/api/configs/packaging/__init__.py
index 2a6fb0a4117db..c311a989b418c 100644
--- a/api/configs/packaging/__init__.py
+++ b/api/configs/packaging/__init__.py
@@ -9,7 +9,7 @@ class PackagingInfo(BaseSettings):
 
     CURRENT_VERSION: str = Field(
         description="Dify version",
-        default="0.9.0",
+        default="0.9.1",
     )
 
     COMMIT_SHA: str = Field(
diff --git a/docker-legacy/docker-compose.yaml b/docker-legacy/docker-compose.yaml
index 435af75827651..3f230b47ab9f1 100644
--- a/docker-legacy/docker-compose.yaml
+++ b/docker-legacy/docker-compose.yaml
@@ -2,7 +2,7 @@ version: '3'
 services:
   # API service
   api:
-    image: langgenius/dify-api:0.9.0
+    image: langgenius/dify-api:0.9.1
     restart: always
     environment:
       # Startup mode, 'api' starts the API server.
@@ -227,7 +227,7 @@ services:
   # worker service
   # The Celery worker for processing the queue.
   worker:
-    image: langgenius/dify-api:0.9.0
+    image: langgenius/dify-api:0.9.1
     restart: always
     environment:
       CONSOLE_WEB_URL: ''
@@ -396,7 +396,7 @@ services:
 
   # Frontend web application.
   web:
-    image: langgenius/dify-web:0.9.0
+    image: langgenius/dify-web:0.9.1
     restart: always
     environment:
       # The base URL of console application api server, refers to the Console base URL of WEB service if console domain is
diff --git a/docker/docker-compose.yaml b/docker/docker-compose.yaml
index 07f1130e67376..ddb6ded206012 100644
--- a/docker/docker-compose.yaml
+++ b/docker/docker-compose.yaml
@@ -213,7 +213,7 @@ x-shared-env: &shared-api-worker-env
 services:
   # API service
   api:
-    image: langgenius/dify-api:0.9.0
+    image: langgenius/dify-api:0.9.1
     restart: always
     environment:
       # Use the shared environment variables.
@@ -233,7 +233,7 @@ services:
   # worker service
   # The Celery worker for processing the queue.
   worker:
-    image: langgenius/dify-api:0.9.0
+    image: langgenius/dify-api:0.9.1
     restart: always
     environment:
       # Use the shared environment variables.
@@ -252,7 +252,7 @@ services:
 
   # Frontend web application.
   web:
-    image: langgenius/dify-web:0.9.0
+    image: langgenius/dify-web:0.9.1
     restart: always
     environment:
       CONSOLE_API_URL: ${CONSOLE_API_URL:-}
diff --git a/web/package.json b/web/package.json
index 7ed8174acce9c..8237623b7c065 100644
--- a/web/package.json
+++ b/web/package.json
@@ -1,6 +1,6 @@
 {
   "name": "dify-web",
-  "version": "0.9.0",
+  "version": "0.9.1",
   "private": true,
   "engines": {
     "node": ">=18.17.0"

From 7850a28ec8022cbeaa117ad1bb12c34ba9ada975 Mon Sep 17 00:00:00 2001
From: -LAN- <laipz8200@outlook.com>
Date: Mon, 30 Sep 2024 22:53:32 +0800
Subject: [PATCH 125/235] Revert "chore(version): bump to 0.9.1" (#8944)

---
 api/configs/packaging/__init__.py | 2 +-
 docker-legacy/docker-compose.yaml | 6 +++---
 docker/docker-compose.yaml        | 6 +++---
 web/package.json                  | 2 +-
 4 files changed, 8 insertions(+), 8 deletions(-)

diff --git a/api/configs/packaging/__init__.py b/api/configs/packaging/__init__.py
index c311a989b418c..2a6fb0a4117db 100644
--- a/api/configs/packaging/__init__.py
+++ b/api/configs/packaging/__init__.py
@@ -9,7 +9,7 @@ class PackagingInfo(BaseSettings):
 
     CURRENT_VERSION: str = Field(
         description="Dify version",
-        default="0.9.1",
+        default="0.9.0",
     )
 
     COMMIT_SHA: str = Field(
diff --git a/docker-legacy/docker-compose.yaml b/docker-legacy/docker-compose.yaml
index 3f230b47ab9f1..435af75827651 100644
--- a/docker-legacy/docker-compose.yaml
+++ b/docker-legacy/docker-compose.yaml
@@ -2,7 +2,7 @@ version: '3'
 services:
   # API service
   api:
-    image: langgenius/dify-api:0.9.1
+    image: langgenius/dify-api:0.9.0
     restart: always
     environment:
       # Startup mode, 'api' starts the API server.
@@ -227,7 +227,7 @@ services:
   # worker service
   # The Celery worker for processing the queue.
   worker:
-    image: langgenius/dify-api:0.9.1
+    image: langgenius/dify-api:0.9.0
     restart: always
     environment:
       CONSOLE_WEB_URL: ''
@@ -396,7 +396,7 @@ services:
 
   # Frontend web application.
   web:
-    image: langgenius/dify-web:0.9.1
+    image: langgenius/dify-web:0.9.0
     restart: always
     environment:
       # The base URL of console application api server, refers to the Console base URL of WEB service if console domain is
diff --git a/docker/docker-compose.yaml b/docker/docker-compose.yaml
index ddb6ded206012..07f1130e67376 100644
--- a/docker/docker-compose.yaml
+++ b/docker/docker-compose.yaml
@@ -213,7 +213,7 @@ x-shared-env: &shared-api-worker-env
 services:
   # API service
   api:
-    image: langgenius/dify-api:0.9.1
+    image: langgenius/dify-api:0.9.0
     restart: always
     environment:
       # Use the shared environment variables.
@@ -233,7 +233,7 @@ services:
   # worker service
   # The Celery worker for processing the queue.
   worker:
-    image: langgenius/dify-api:0.9.1
+    image: langgenius/dify-api:0.9.0
     restart: always
     environment:
       # Use the shared environment variables.
@@ -252,7 +252,7 @@ services:
 
   # Frontend web application.
   web:
-    image: langgenius/dify-web:0.9.1
+    image: langgenius/dify-web:0.9.0
     restart: always
     environment:
       CONSOLE_API_URL: ${CONSOLE_API_URL:-}
diff --git a/web/package.json b/web/package.json
index 8237623b7c065..7ed8174acce9c 100644
--- a/web/package.json
+++ b/web/package.json
@@ -1,6 +1,6 @@
 {
   "name": "dify-web",
-  "version": "0.9.1",
+  "version": "0.9.0",
   "private": true,
   "engines": {
     "node": ">=18.17.0"

From 625e4c4c72860516469599dc9ab1ab1c9341c360 Mon Sep 17 00:00:00 2001
From: Jyong <76649700+JohnJyong@users.noreply.github.com>
Date: Mon, 30 Sep 2024 23:07:04 +0800
Subject: [PATCH 126/235] fix multiple retrieval in knowledge node (#8942)

---
 .../nodes/knowledge_retrieval/knowledge_retrieval_node.py     | 4 +++-
 api/services/website_service.py                               | 4 ++--
 2 files changed, 5 insertions(+), 3 deletions(-)

diff --git a/api/core/workflow/nodes/knowledge_retrieval/knowledge_retrieval_node.py b/api/core/workflow/nodes/knowledge_retrieval/knowledge_retrieval_node.py
index c08dc143a6f32..0caf99a963822 100644
--- a/api/core/workflow/nodes/knowledge_retrieval/knowledge_retrieval_node.py
+++ b/api/core/workflow/nodes/knowledge_retrieval/knowledge_retrieval_node.py
@@ -229,7 +229,9 @@ def _fetch_dataset_retriever(self, node_data: KnowledgeRetrievalNodeData, query:
                             source["content"] = segment.get_sign_content()
                         retrieval_resource_list.append(source)
         if retrieval_resource_list:
-            retrieval_resource_list = sorted(retrieval_resource_list, key=lambda x: x.get("score"), reverse=True)
+            retrieval_resource_list = sorted(
+                retrieval_resource_list, key=lambda x: x.get("metadata").get("score"), reverse=True
+            )
             position = 1
             for item in retrieval_resource_list:
                 item["metadata"]["position"] = position
diff --git a/api/services/website_service.py b/api/services/website_service.py
index be01815720ab5..13cc9c679adb9 100644
--- a/api/services/website_service.py
+++ b/api/services/website_service.py
@@ -166,6 +166,8 @@ def get_crawl_status(cls, job_id: str, provider: str) -> dict:
     @classmethod
     def get_crawl_url_data(cls, job_id: str, provider: str, url: str, tenant_id: str) -> dict | None:
         credentials = ApiKeyAuthService.get_auth_credentials(tenant_id, "website", provider)
+        # decrypt api_key
+        api_key = encrypter.decrypt_token(tenant_id=tenant_id, token=credentials.get("config").get("api_key"))
         if provider == "firecrawl":
             file_key = "website_files/" + job_id + ".txt"
             if storage.exists(file_key):
@@ -173,8 +175,6 @@ def get_crawl_url_data(cls, job_id: str, provider: str, url: str, tenant_id: str
                 if data:
                     data = json.loads(data.decode("utf-8"))
             else:
-                # decrypt api_key
-                api_key = encrypter.decrypt_token(tenant_id=tenant_id, token=credentials.get("config").get("api_key"))
                 firecrawl_app = FirecrawlApp(api_key=api_key, base_url=credentials.get("config").get("base_url", None))
                 result = firecrawl_app.check_crawl_status(job_id)
                 if result.get("status") != "completed":

From 1f5cc071f8e3541df5d16b258059b157a7b60b73 Mon Sep 17 00:00:00 2001
From: -LAN- <laipz8200@outlook.com>
Date: Mon, 30 Sep 2024 23:22:21 +0800
Subject: [PATCH 127/235] chore(version): bump to 0.9.1 (#8945)

---
 api/configs/packaging/__init__.py | 2 +-
 docker-legacy/docker-compose.yaml | 6 +++---
 docker/docker-compose.yaml        | 6 +++---
 web/package.json                  | 2 +-
 4 files changed, 8 insertions(+), 8 deletions(-)

diff --git a/api/configs/packaging/__init__.py b/api/configs/packaging/__init__.py
index 2a6fb0a4117db..c311a989b418c 100644
--- a/api/configs/packaging/__init__.py
+++ b/api/configs/packaging/__init__.py
@@ -9,7 +9,7 @@ class PackagingInfo(BaseSettings):
 
     CURRENT_VERSION: str = Field(
         description="Dify version",
-        default="0.9.0",
+        default="0.9.1",
     )
 
     COMMIT_SHA: str = Field(
diff --git a/docker-legacy/docker-compose.yaml b/docker-legacy/docker-compose.yaml
index 435af75827651..3f230b47ab9f1 100644
--- a/docker-legacy/docker-compose.yaml
+++ b/docker-legacy/docker-compose.yaml
@@ -2,7 +2,7 @@ version: '3'
 services:
   # API service
   api:
-    image: langgenius/dify-api:0.9.0
+    image: langgenius/dify-api:0.9.1
     restart: always
     environment:
       # Startup mode, 'api' starts the API server.
@@ -227,7 +227,7 @@ services:
   # worker service
   # The Celery worker for processing the queue.
   worker:
-    image: langgenius/dify-api:0.9.0
+    image: langgenius/dify-api:0.9.1
     restart: always
     environment:
       CONSOLE_WEB_URL: ''
@@ -396,7 +396,7 @@ services:
 
   # Frontend web application.
   web:
-    image: langgenius/dify-web:0.9.0
+    image: langgenius/dify-web:0.9.1
     restart: always
     environment:
       # The base URL of console application api server, refers to the Console base URL of WEB service if console domain is
diff --git a/docker/docker-compose.yaml b/docker/docker-compose.yaml
index 07f1130e67376..ddb6ded206012 100644
--- a/docker/docker-compose.yaml
+++ b/docker/docker-compose.yaml
@@ -213,7 +213,7 @@ x-shared-env: &shared-api-worker-env
 services:
   # API service
   api:
-    image: langgenius/dify-api:0.9.0
+    image: langgenius/dify-api:0.9.1
     restart: always
     environment:
       # Use the shared environment variables.
@@ -233,7 +233,7 @@ services:
   # worker service
   # The Celery worker for processing the queue.
   worker:
-    image: langgenius/dify-api:0.9.0
+    image: langgenius/dify-api:0.9.1
     restart: always
     environment:
       # Use the shared environment variables.
@@ -252,7 +252,7 @@ services:
 
   # Frontend web application.
   web:
-    image: langgenius/dify-web:0.9.0
+    image: langgenius/dify-web:0.9.1
     restart: always
     environment:
       CONSOLE_API_URL: ${CONSOLE_API_URL:-}
diff --git a/web/package.json b/web/package.json
index 7ed8174acce9c..8237623b7c065 100644
--- a/web/package.json
+++ b/web/package.json
@@ -1,6 +1,6 @@
 {
   "name": "dify-web",
-  "version": "0.9.0",
+  "version": "0.9.1",
   "private": true,
   "engines": {
     "node": ">=18.17.0"

From cdd7e55a88ef0e8cf62099df8d1fd08582cc62ca Mon Sep 17 00:00:00 2001
From: CXwudi <cx2010017@gmail.com>
Date: Mon, 30 Sep 2024 21:14:21 -0400
Subject: [PATCH 128/235] chore: add missing models from Voyage (#8950)

---
 .../model_providers/voyage/rerank/_position.yaml          | 4 ++++
 .../model_providers/voyage/rerank/rerank-2.yaml           | 4 ++++
 .../model_providers/voyage/rerank/rerank-lite-2.yaml      | 4 ++++
 .../model_providers/voyage/text_embedding/_position.yaml  | 6 ++++++
 .../voyage/text_embedding/voyage-code-2.yaml              | 8 ++++++++
 .../voyage/text_embedding/voyage-finance-2.yaml           | 8 ++++++++
 .../voyage/text_embedding/voyage-law-2.yaml               | 8 ++++++++
 .../voyage/text_embedding/voyage-multilingual-2.yaml      | 8 ++++++++
 8 files changed, 50 insertions(+)
 create mode 100644 api/core/model_runtime/model_providers/voyage/rerank/_position.yaml
 create mode 100644 api/core/model_runtime/model_providers/voyage/rerank/rerank-2.yaml
 create mode 100644 api/core/model_runtime/model_providers/voyage/rerank/rerank-lite-2.yaml
 create mode 100644 api/core/model_runtime/model_providers/voyage/text_embedding/_position.yaml
 create mode 100644 api/core/model_runtime/model_providers/voyage/text_embedding/voyage-code-2.yaml
 create mode 100644 api/core/model_runtime/model_providers/voyage/text_embedding/voyage-finance-2.yaml
 create mode 100644 api/core/model_runtime/model_providers/voyage/text_embedding/voyage-law-2.yaml
 create mode 100644 api/core/model_runtime/model_providers/voyage/text_embedding/voyage-multilingual-2.yaml

diff --git a/api/core/model_runtime/model_providers/voyage/rerank/_position.yaml b/api/core/model_runtime/model_providers/voyage/rerank/_position.yaml
new file mode 100644
index 0000000000000..32afefbe04780
--- /dev/null
+++ b/api/core/model_runtime/model_providers/voyage/rerank/_position.yaml
@@ -0,0 +1,4 @@
+- rerank-2
+- rerank-lite-2
+- rerank-1
+- rerank-lite-1
diff --git a/api/core/model_runtime/model_providers/voyage/rerank/rerank-2.yaml b/api/core/model_runtime/model_providers/voyage/rerank/rerank-2.yaml
new file mode 100644
index 0000000000000..b760d3c41894a
--- /dev/null
+++ b/api/core/model_runtime/model_providers/voyage/rerank/rerank-2.yaml
@@ -0,0 +1,4 @@
+model: rerank-2
+model_type: rerank
+model_properties:
+  context_size: 16000
diff --git a/api/core/model_runtime/model_providers/voyage/rerank/rerank-lite-2.yaml b/api/core/model_runtime/model_providers/voyage/rerank/rerank-lite-2.yaml
new file mode 100644
index 0000000000000..b6fa37a25bf18
--- /dev/null
+++ b/api/core/model_runtime/model_providers/voyage/rerank/rerank-lite-2.yaml
@@ -0,0 +1,4 @@
+model: rerank-lite-2
+model_type: rerank
+model_properties:
+  context_size: 8000
diff --git a/api/core/model_runtime/model_providers/voyage/text_embedding/_position.yaml b/api/core/model_runtime/model_providers/voyage/text_embedding/_position.yaml
new file mode 100644
index 0000000000000..595663990f594
--- /dev/null
+++ b/api/core/model_runtime/model_providers/voyage/text_embedding/_position.yaml
@@ -0,0 +1,6 @@
+- voyage-3
+- voyage-3-lite
+- voyage-finance-2
+- voyage-multilingual-2
+- voyage-law-2
+- voyage-code-2
diff --git a/api/core/model_runtime/model_providers/voyage/text_embedding/voyage-code-2.yaml b/api/core/model_runtime/model_providers/voyage/text_embedding/voyage-code-2.yaml
new file mode 100644
index 0000000000000..693669c82c7ba
--- /dev/null
+++ b/api/core/model_runtime/model_providers/voyage/text_embedding/voyage-code-2.yaml
@@ -0,0 +1,8 @@
+model: voyage-code-2
+model_type: text-embedding
+model_properties:
+  context_size: 16000
+pricing:
+  input: '0.00012'
+  unit: '0.001'
+  currency: USD
diff --git a/api/core/model_runtime/model_providers/voyage/text_embedding/voyage-finance-2.yaml b/api/core/model_runtime/model_providers/voyage/text_embedding/voyage-finance-2.yaml
new file mode 100644
index 0000000000000..555e11002ade3
--- /dev/null
+++ b/api/core/model_runtime/model_providers/voyage/text_embedding/voyage-finance-2.yaml
@@ -0,0 +1,8 @@
+model: voyage-finance-2
+model_type: text-embedding
+model_properties:
+  context_size: 32000
+pricing:
+  input: '0.00012'
+  unit: '0.001'
+  currency: USD
diff --git a/api/core/model_runtime/model_providers/voyage/text_embedding/voyage-law-2.yaml b/api/core/model_runtime/model_providers/voyage/text_embedding/voyage-law-2.yaml
new file mode 100644
index 0000000000000..032693286f6e8
--- /dev/null
+++ b/api/core/model_runtime/model_providers/voyage/text_embedding/voyage-law-2.yaml
@@ -0,0 +1,8 @@
+model: voyage-law-2
+model_type: text-embedding
+model_properties:
+  context_size: 16000
+pricing:
+  input: '0.00012'
+  unit: '0.001'
+  currency: USD
diff --git a/api/core/model_runtime/model_providers/voyage/text_embedding/voyage-multilingual-2.yaml b/api/core/model_runtime/model_providers/voyage/text_embedding/voyage-multilingual-2.yaml
new file mode 100644
index 0000000000000..9ecf4d5009833
--- /dev/null
+++ b/api/core/model_runtime/model_providers/voyage/text_embedding/voyage-multilingual-2.yaml
@@ -0,0 +1,8 @@
+model: voyage-multilingual-2
+model_type: text-embedding
+model_properties:
+  context_size: 32000
+pricing:
+  input: '0.00012'
+  unit: '0.001'
+  currency: USD

From 0d84221b2ca8790a2b1085bdf77cee84c0045ab8 Mon Sep 17 00:00:00 2001
From: CXwudi <cx2010017@gmail.com>
Date: Mon, 30 Sep 2024 21:14:36 -0400
Subject: [PATCH 129/235] chore: sort Gemini models (#8951)

---
 .../model_providers/google/llm/_position.yaml     | 15 +++++++++++++++
 1 file changed, 15 insertions(+)
 create mode 100644 api/core/model_runtime/model_providers/google/llm/_position.yaml

diff --git a/api/core/model_runtime/model_providers/google/llm/_position.yaml b/api/core/model_runtime/model_providers/google/llm/_position.yaml
new file mode 100644
index 0000000000000..63b9ca3a292e7
--- /dev/null
+++ b/api/core/model_runtime/model_providers/google/llm/_position.yaml
@@ -0,0 +1,15 @@
+- gemini-1.5-pro
+- gemini-1.5-pro-latest
+- gemini-1.5-pro-001
+- gemini-1.5-pro-002
+- gemini-1.5-pro-exp-0801
+- gemini-1.5-pro-exp-0827
+- gemini-1.5-flash
+- gemini-1.5-flash-latest
+- gemini-1.5-flash-001
+- gemini-1.5-flash-002
+- gemini-1.5-flash-exp-0827
+- gemini-1.5-flash-8b-exp-0827
+- gemini-1.5-flash-8b-exp-0924
+- gemini-pro
+- gemini-pro-vision

From 2deaece7e2ffecc119cc2a9dba6433bb63b37713 Mon Sep 17 00:00:00 2001
From: omr <145922434+y-omr@users.noreply.github.com>
Date: Tue, 1 Oct 2024 10:14:49 +0900
Subject: [PATCH 130/235] refactor: remove unnecessary comment (#8949)

---
 api/core/callback_handler/index_tool_callback_handler.py | 1 -
 1 file changed, 1 deletion(-)

diff --git a/api/core/callback_handler/index_tool_callback_handler.py b/api/core/callback_handler/index_tool_callback_handler.py
index 23dc092895c8c..1481578630f63 100644
--- a/api/core/callback_handler/index_tool_callback_handler.py
+++ b/api/core/callback_handler/index_tool_callback_handler.py
@@ -44,7 +44,6 @@ def on_tool_end(self, documents: list[Document]) -> None:
                 DocumentSegment.index_node_id == document.metadata["doc_id"]
             )
 
-            # if 'dataset_id' in document.metadata:
             if "dataset_id" in document.metadata:
                 query = query.filter(DocumentSegment.dataset_id == document.metadata["dataset_id"])
 

From c2d606d587ca385106ea8439779ce6d8e65f3a35 Mon Sep 17 00:00:00 2001
From: ice yao <yao3690093@gmail.com>
Date: Tue, 1 Oct 2024 09:15:18 +0800
Subject: [PATCH 131/235] chore: remove unexecuted scripts to avoid ambiguity
 (#8946)

---
 dev/pytest/pytest_vdb.sh | 3 +--
 1 file changed, 1 insertion(+), 2 deletions(-)

diff --git a/dev/pytest/pytest_vdb.sh b/dev/pytest/pytest_vdb.sh
index 0b23200dc33b0..bad809cbfdb90 100755
--- a/dev/pytest/pytest_vdb.sh
+++ b/dev/pytest/pytest_vdb.sh
@@ -7,5 +7,4 @@ pytest api/tests/integration_tests/vdb/chroma \
   api/tests/integration_tests/vdb/pgvector \
   api/tests/integration_tests/vdb/qdrant \
   api/tests/integration_tests/vdb/weaviate \
-  api/tests/integration_tests/vdb/elasticsearch \
-  api/tests/integration_tests/vdb/test_vector_store.py
\ No newline at end of file
+  api/tests/integration_tests/vdb/elasticsearch
\ No newline at end of file

From 824a0dd63ec30f94eacf64fddbbdd367ff5ad48e Mon Sep 17 00:00:00 2001
From: zhuhao <37029601+hwzhuhao@users.noreply.github.com>
Date: Tue, 1 Oct 2024 10:55:51 +0800
Subject: [PATCH 132/235] feat: add qwen2.5-72b and llama3.2 for openrouter
 (#8956)

---
 .../openrouter/llm/_position.yaml             |  5 +++
 .../llm/llama-3.2-11b-vision-instruct.yaml    | 45 +++++++++++++++++++
 .../openrouter/llm/llama-3.2-1b-instruct.yaml | 45 +++++++++++++++++++
 .../openrouter/llm/llama-3.2-3b-instruct.yaml | 45 +++++++++++++++++++
 .../llm/llama-3.2-90b-vision-instruct.yaml    | 45 +++++++++++++++++++
 .../openrouter/llm/qwen2.5-72b-instruct.yaml  | 30 +++++++++++++
 6 files changed, 215 insertions(+)
 create mode 100644 api/core/model_runtime/model_providers/openrouter/llm/llama-3.2-11b-vision-instruct.yaml
 create mode 100644 api/core/model_runtime/model_providers/openrouter/llm/llama-3.2-1b-instruct.yaml
 create mode 100644 api/core/model_runtime/model_providers/openrouter/llm/llama-3.2-3b-instruct.yaml
 create mode 100644 api/core/model_runtime/model_providers/openrouter/llm/llama-3.2-90b-vision-instruct.yaml
 create mode 100644 api/core/model_runtime/model_providers/openrouter/llm/qwen2.5-72b-instruct.yaml

diff --git a/api/core/model_runtime/model_providers/openrouter/llm/_position.yaml b/api/core/model_runtime/model_providers/openrouter/llm/_position.yaml
index d9497b76b894c..5a25c84c34c0c 100644
--- a/api/core/model_runtime/model_providers/openrouter/llm/_position.yaml
+++ b/api/core/model_runtime/model_providers/openrouter/llm/_position.yaml
@@ -14,6 +14,10 @@
 - google/gemini-pro
 - cohere/command-r-plus
 - cohere/command-r
+- meta-llama/llama-3.2-1b-instruct
+- meta-llama/llama-3.2-3b-instruct
+- meta-llama/llama-3.2-11b-vision-instruct
+- meta-llama/llama-3.2-90b-vision-instruct
 - meta-llama/llama-3.1-405b-instruct
 - meta-llama/llama-3.1-70b-instruct
 - meta-llama/llama-3.1-8b-instruct
@@ -22,6 +26,7 @@
 - mistralai/mixtral-8x22b-instruct
 - mistralai/mixtral-8x7b-instruct
 - mistralai/mistral-7b-instruct
+- qwen/qwen-2.5-72b-instruct
 - qwen/qwen-2-72b-instruct
 - deepseek/deepseek-chat
 - deepseek/deepseek-coder
diff --git a/api/core/model_runtime/model_providers/openrouter/llm/llama-3.2-11b-vision-instruct.yaml b/api/core/model_runtime/model_providers/openrouter/llm/llama-3.2-11b-vision-instruct.yaml
new file mode 100644
index 0000000000000..235156997f4b3
--- /dev/null
+++ b/api/core/model_runtime/model_providers/openrouter/llm/llama-3.2-11b-vision-instruct.yaml
@@ -0,0 +1,45 @@
+model: meta-llama/llama-3.2-11b-vision-instruct
+label:
+  zh_Hans: llama-3.2-11b-vision-instruct
+  en_US: llama-3.2-11b-vision-instruct
+model_type: llm
+features:
+  - agent-thought
+model_properties:
+  mode: chat
+  context_size: 131072
+parameter_rules:
+  - name: temperature
+    use_template: temperature
+  - name: top_p
+    use_template: top_p
+  - name: top_k
+    label:
+      zh_Hans: 取样数量
+      en_US: Top k
+    type: int
+    help:
+      zh_Hans: 仅从每个后续标记的前 K 个选项中采样。
+      en_US: Only sample from the top K options for each subsequent token.
+  - name: max_tokens
+    use_template: max_tokens
+  - name: context_length_exceeded_behavior
+    default: None
+    label:
+      zh_Hans: 上下文长度超出行为
+      en_US: Context Length Exceeded Behavior
+    help:
+      zh_Hans: 上下文长度超出行为
+      en_US: Context Length Exceeded Behavior
+    type: string
+    options:
+      - None
+      - truncate
+      - error
+  - name: response_format
+    use_template: response_format
+pricing:
+  input: '0.055'
+  output: '0.055'
+  unit: '0.000001'
+  currency: USD
diff --git a/api/core/model_runtime/model_providers/openrouter/llm/llama-3.2-1b-instruct.yaml b/api/core/model_runtime/model_providers/openrouter/llm/llama-3.2-1b-instruct.yaml
new file mode 100644
index 0000000000000..657ef16835f42
--- /dev/null
+++ b/api/core/model_runtime/model_providers/openrouter/llm/llama-3.2-1b-instruct.yaml
@@ -0,0 +1,45 @@
+model: meta-llama/llama-3.2-1b-instruct
+label:
+  zh_Hans: llama-3.2-1b-instruct
+  en_US: llama-3.2-1b-instruct
+model_type: llm
+features:
+  - agent-thought
+model_properties:
+  mode: chat
+  context_size: 131072
+parameter_rules:
+  - name: temperature
+    use_template: temperature
+  - name: top_p
+    use_template: top_p
+  - name: top_k
+    label:
+      zh_Hans: 取样数量
+      en_US: Top k
+    type: int
+    help:
+      zh_Hans: 仅从每个后续标记的前 K 个选项中采样。
+      en_US: Only sample from the top K options for each subsequent token.
+  - name: max_tokens
+    use_template: max_tokens
+  - name: context_length_exceeded_behavior
+    default: None
+    label:
+      zh_Hans: 上下文长度超出行为
+      en_US: Context Length Exceeded Behavior
+    help:
+      zh_Hans: 上下文长度超出行为
+      en_US: Context Length Exceeded Behavior
+    type: string
+    options:
+      - None
+      - truncate
+      - error
+  - name: response_format
+    use_template: response_format
+pricing:
+  input: '0.01'
+  output: '0.02'
+  unit: '0.000001'
+  currency: USD
diff --git a/api/core/model_runtime/model_providers/openrouter/llm/llama-3.2-3b-instruct.yaml b/api/core/model_runtime/model_providers/openrouter/llm/llama-3.2-3b-instruct.yaml
new file mode 100644
index 0000000000000..7f6e24e591aac
--- /dev/null
+++ b/api/core/model_runtime/model_providers/openrouter/llm/llama-3.2-3b-instruct.yaml
@@ -0,0 +1,45 @@
+model: meta-llama/llama-3.2-3b-instruct
+label:
+  zh_Hans: llama-3.2-3b-instruct
+  en_US: llama-3.2-3b-instruct
+model_type: llm
+features:
+  - agent-thought
+model_properties:
+  mode: chat
+  context_size: 131072
+parameter_rules:
+  - name: temperature
+    use_template: temperature
+  - name: top_p
+    use_template: top_p
+  - name: top_k
+    label:
+      zh_Hans: 取样数量
+      en_US: Top k
+    type: int
+    help:
+      zh_Hans: 仅从每个后续标记的前 K 个选项中采样。
+      en_US: Only sample from the top K options for each subsequent token.
+  - name: max_tokens
+    use_template: max_tokens
+  - name: context_length_exceeded_behavior
+    default: None
+    label:
+      zh_Hans: 上下文长度超出行为
+      en_US: Context Length Exceeded Behavior
+    help:
+      zh_Hans: 上下文长度超出行为
+      en_US: Context Length Exceeded Behavior
+    type: string
+    options:
+      - None
+      - truncate
+      - error
+  - name: response_format
+    use_template: response_format
+pricing:
+  input: '0.03'
+  output: '0.05'
+  unit: '0.000001'
+  currency: USD
diff --git a/api/core/model_runtime/model_providers/openrouter/llm/llama-3.2-90b-vision-instruct.yaml b/api/core/model_runtime/model_providers/openrouter/llm/llama-3.2-90b-vision-instruct.yaml
new file mode 100644
index 0000000000000..5d597f00a2d1b
--- /dev/null
+++ b/api/core/model_runtime/model_providers/openrouter/llm/llama-3.2-90b-vision-instruct.yaml
@@ -0,0 +1,45 @@
+model: meta-llama/llama-3.2-90b-vision-instruct
+label:
+  zh_Hans: llama-3.2-90b-vision-instruct
+  en_US: llama-3.2-90b-vision-instruct
+model_type: llm
+features:
+  - agent-thought
+model_properties:
+  mode: chat
+  context_size: 131072
+parameter_rules:
+  - name: temperature
+    use_template: temperature
+  - name: top_p
+    use_template: top_p
+  - name: top_k
+    label:
+      zh_Hans: 取样数量
+      en_US: Top k
+    type: int
+    help:
+      zh_Hans: 仅从每个后续标记的前 K 个选项中采样。
+      en_US: Only sample from the top K options for each subsequent token.
+  - name: max_tokens
+    use_template: max_tokens
+  - name: context_length_exceeded_behavior
+    default: None
+    label:
+      zh_Hans: 上下文长度超出行为
+      en_US: Context Length Exceeded Behavior
+    help:
+      zh_Hans: 上下文长度超出行为
+      en_US: Context Length Exceeded Behavior
+    type: string
+    options:
+      - None
+      - truncate
+      - error
+  - name: response_format
+    use_template: response_format
+pricing:
+  input: '0.35'
+  output: '0.4'
+  unit: '0.000001'
+  currency: USD
diff --git a/api/core/model_runtime/model_providers/openrouter/llm/qwen2.5-72b-instruct.yaml b/api/core/model_runtime/model_providers/openrouter/llm/qwen2.5-72b-instruct.yaml
new file mode 100644
index 0000000000000..f141a40a00c9c
--- /dev/null
+++ b/api/core/model_runtime/model_providers/openrouter/llm/qwen2.5-72b-instruct.yaml
@@ -0,0 +1,30 @@
+model: qwen/qwen-2.5-72b-instruct
+label:
+  en_US: qwen-2.5-72b-instruct
+model_type: llm
+features:
+  - agent-thought
+model_properties:
+  mode: chat
+  context_size: 131072
+parameter_rules:
+  - name: temperature
+    use_template: temperature
+  - name: max_tokens
+    use_template: max_tokens
+    type: int
+    default: 512
+    min: 1
+    max: 8192
+    help:
+      zh_Hans: 指定生成结果长度的上限。如果生成结果截断，可以调大该参数。
+      en_US: Specifies the upper limit on the length of generated results. If the generated results are truncated, you can increase this parameter.
+  - name: top_p
+    use_template: top_p
+  - name: frequency_penalty
+    use_template: frequency_penalty
+pricing:
+  input: "0.35"
+  output: "0.4"
+  unit: "0.000001"
+  currency: USD

From 24ba9fdf6c990ffa5d38463d38181c5415b418e0 Mon Sep 17 00:00:00 2001
From: zhuhao <37029601+hwzhuhao@users.noreply.github.com>
Date: Tue, 1 Oct 2024 10:55:54 +0800
Subject: [PATCH 133/235] feat: enhance stepfun image generation tool (#8954)

---
 .../provider/builtin/stepfun/tools/image.py   | 12 ++---
 .../provider/builtin/stepfun/tools/image.yaml | 50 ++++++++++---------
 2 files changed, 33 insertions(+), 29 deletions(-)

diff --git a/api/core/tools/provider/builtin/stepfun/tools/image.py b/api/core/tools/provider/builtin/stepfun/tools/image.py
index eb55dae5183c4..61cc14fac6ca9 100644
--- a/api/core/tools/provider/builtin/stepfun/tools/image.py
+++ b/api/core/tools/provider/builtin/stepfun/tools/image.py
@@ -32,16 +32,17 @@ def _invoke(
         prompt = tool_parameters.get("prompt", "")
         if not prompt:
             return self.create_text_message("Please input prompt")
-
+        if len(prompt) > 1024:
+            return self.create_text_message("The prompt length should less than 1024")
         seed = tool_parameters.get("seed", 0)
         if seed > 0:
             extra_body["seed"] = seed
-        steps = tool_parameters.get("steps", 0)
+        steps = tool_parameters.get("steps", 50)
         if steps > 0:
             extra_body["steps"] = steps
-        negative_prompt = tool_parameters.get("negative_prompt", "")
-        if negative_prompt:
-            extra_body["negative_prompt"] = negative_prompt
+        cfg_scale = tool_parameters.get("cfg_scale", 7.5)
+        if cfg_scale > 0:
+            extra_body["cfg_scale"] = cfg_scale
 
         # call openapi stepfun model
         response = client.images.generate(
@@ -51,7 +52,6 @@ def _invoke(
             n=tool_parameters.get("n", 1),
             extra_body=extra_body,
         )
-        print(response)
 
         result = []
         for image in response.data:
diff --git a/api/core/tools/provider/builtin/stepfun/tools/image.yaml b/api/core/tools/provider/builtin/stepfun/tools/image.yaml
index 8d7c9b6586b6c..dfda6ed191484 100644
--- a/api/core/tools/provider/builtin/stepfun/tools/image.yaml
+++ b/api/core/tools/provider/builtin/stepfun/tools/image.yaml
@@ -33,9 +33,9 @@ parameters:
     type: select
     required: false
     human_description:
-      en_US: used for selecting the image size
-      zh_Hans: 用于选择图像大小
-      pt_BR: used for selecting the image size
+      en_US: The size of the generated image
+      zh_Hans: 生成的图片大小
+      pt_BR: The size of the generated image
     label:
       en_US: Image size
       zh_Hans: 图像大小
@@ -77,17 +77,17 @@ parameters:
     type: number
     required: true
     human_description:
-      en_US: used for selecting the number of images
-      zh_Hans: 用于选择图像数量
-      pt_BR: used for selecting the number of images
+      en_US: Number of generated images, now only one image can be generated at a time
+      zh_Hans: 生成的图像数量，当前仅支持每次生成一张图片
+      pt_BR: Number of generated images, now only one image can be generated at a time
     label:
-      en_US: Number of images
-      zh_Hans: 图像数量
-      pt_BR: Number of images
+      en_US: Number of generated images
+      zh_Hans: 生成的图像数量
+      pt_BR: Number of generated images
     form: form
     default: 1
     min: 1
-    max: 10
+    max: 1
   - name: seed
     type: number
     required: false
@@ -109,21 +109,25 @@ parameters:
       zh_Hans: Steps
       pt_BR: Steps
     human_description:
-      en_US: Steps
-      zh_Hans: Steps
-      pt_BR: Steps
+      en_US: Steps, now support integers between 1 and 100
+      zh_Hans: Steps, 当前支持 1～100 之间整数
+      pt_BR: Steps, now support integers between 1 and 100
     form: form
-    default: 10
-  - name: negative_prompt
-    type: string
+    default: 50
+    min: 1
+    max: 100
+  - name: cfg_scale
+    type: number
     required: false
     label:
-      en_US: Negative prompt
-      zh_Hans: Negative prompt
-      pt_BR: Negative prompt
+      en_US: classifier-free guidance scale
+      zh_Hans: classifier-free guidance scale
+      pt_BR: classifier-free guidance scale
     human_description:
-      en_US: Negative prompt
-      zh_Hans: Negative prompt
-      pt_BR: Negative prompt
+      en_US: classifier-free guidance scale
+      zh_Hans: classifier-free guidance scale
+      pt_BR: classifier-free guidance scale
     form: form
-    default: (worst quality:1.3), (nsfw), low quality
+    default: 7.5
+    min: 1
+    max: 10

From 5f8a27074ebe5797b30ce44e8fc6252adf4bef51 Mon Sep 17 00:00:00 2001
From: Hash Brown <hi@xzd.me>
Date: Tue, 1 Oct 2024 12:10:36 +0800
Subject: [PATCH 134/235] fix: chat API is not bringing the
 conversation/session history (#8965)

---
 api/controllers/service_api/app/completion.py | 2 ++
 1 file changed, 2 insertions(+)

diff --git a/api/controllers/service_api/app/completion.py b/api/controllers/service_api/app/completion.py
index 8d8e356c4cb94..5c3601cf23627 100644
--- a/api/controllers/service_api/app/completion.py
+++ b/api/controllers/service_api/app/completion.py
@@ -4,6 +4,7 @@
 from werkzeug.exceptions import InternalServerError, NotFound
 
 import services
+from constants import UUID_NIL
 from controllers.service_api import api
 from controllers.service_api.app.error import (
     AppUnavailableError,
@@ -107,6 +108,7 @@ def post(self, app_model: App, end_user: EndUser):
         parser.add_argument("conversation_id", type=uuid_value, location="json")
         parser.add_argument("retriever_from", type=str, required=False, default="dev", location="json")
         parser.add_argument("auto_generate_name", type=bool, required=False, default=True, location="json")
+        parser.add_argument("parent_message_id", type=uuid_value, required=False, default=UUID_NIL, location="json")
 
         args = parser.parse_args()
 

From 5366820a2f2c3d356a24f0296b42bcbdee4e2212 Mon Sep 17 00:00:00 2001
From: omr <145922434+y-omr@users.noreply.github.com>
Date: Wed, 2 Oct 2024 13:54:22 +0900
Subject: [PATCH 135/235] fix: corrected typo (#8979)

---
 README_JA.md | 2 +-
 1 file changed, 1 insertion(+), 1 deletion(-)

diff --git a/README_JA.md b/README_JA.md
index e6a8621e7baae..98966d6c38447 100644
--- a/README_JA.md
+++ b/README_JA.md
@@ -68,7 +68,7 @@ DifyはオープンソースのLLMアプリケーション開発プラットフ
   プロンプトの作成、モデルパフォーマンスの比較が行え、チャットベースのアプリに音声合成などの機能も追加できます。
 
 **4. RAGパイプライン**: 
-  ドキュメントの取り込みから検索までをカバーする広範なRAG機能ができます。ほかにもPDF、PPT、その他の一般的なドキュメントフォーマットからのテキスト抽出のサーポイントも提供します。
+  ドキュメントの取り込みから検索までをカバーする広範なRAG機能ができます。ほかにもPDF、PPT、その他の一般的なドキュメントフォーマットからのテキスト抽出のサポートも提供します。
 
 **5. エージェント機能**: 
   LLM Function CallingやReActに基づくエージェントの定義が可能で、AIエージェント用のプリビルトまたはカスタムツールを追加できます。Difyには、Google検索、DALL·E、Stable Diffusion、WolframAlphaなどのAIエージェント用の50以上の組み込みツールが提供します。

From 415d27c8bfb8998c373244cc33aa67df44a4368a Mon Sep 17 00:00:00 2001
From: ice yao <yao3690093@gmail.com>
Date: Wed, 2 Oct 2024 13:18:35 +0800
Subject: [PATCH 136/235] feat(Tools): add discord incoming webhook for sending
 messages (#7852)

---
 .../provider/builtin/discord/_assets/icon.svg |  7 +++
 .../tools/provider/builtin/discord/discord.py |  9 ++++
 .../provider/builtin/discord/discord.yaml     | 16 ++++++
 .../builtin/discord/tools/discord_webhook.py  | 49 +++++++++++++++++++
 .../discord/tools/discord_webhook.yaml        | 40 +++++++++++++++
 5 files changed, 121 insertions(+)
 create mode 100644 api/core/tools/provider/builtin/discord/_assets/icon.svg
 create mode 100644 api/core/tools/provider/builtin/discord/discord.py
 create mode 100644 api/core/tools/provider/builtin/discord/discord.yaml
 create mode 100644 api/core/tools/provider/builtin/discord/tools/discord_webhook.py
 create mode 100644 api/core/tools/provider/builtin/discord/tools/discord_webhook.yaml

diff --git a/api/core/tools/provider/builtin/discord/_assets/icon.svg b/api/core/tools/provider/builtin/discord/_assets/icon.svg
new file mode 100644
index 0000000000000..177a0591f9cb0
--- /dev/null
+++ b/api/core/tools/provider/builtin/discord/_assets/icon.svg
@@ -0,0 +1,7 @@
+<!DOCTYPE svg PUBLIC "-//W3C//DTD SVG 1.1//EN" "http://www.w3.org/Graphics/SVG/1.1/DTD/svg11.dtd">
+
<!-- Uploaded to: SVG Repo, www.svgrepo.com, Transformed by: SVG Repo Mixer Tools -->
+<svg width="80px" height="80px" viewBox="0 -28.5 256 256" version="1.1" xmlns="http://www.w3.org/2000/svg" xmlns:xlink="http://www.w3.org/1999/xlink" preserveAspectRatio="xMidYMid" fill="#000000">
+
<g id="SVGRepo_bgCarrier" stroke-width="0"/>
+
<g id="SVGRepo_tracerCarrier" stroke-linecap="round" stroke-linejoin="round"/>
+
<g id="SVGRepo_iconCarrier"> <g> <path d="M216.856339,16.5966031 C200.285002,8.84328665 182.566144,3.2084988 164.041564,0 C161.766523,4.11318106 159.108624,9.64549908 157.276099,14.0464379 C137.583995,11.0849896 118.072967,11.0849896 98.7430163,14.0464379 C96.9108417,9.64549908 94.1925838,4.11318106 91.8971895,0 C73.3526068,3.2084988 55.6133949,8.86399117 39.0420583,16.6376612 C5.61752293,67.146514 -3.4433191,116.400813 1.08711069,164.955721 C23.2560196,181.510915 44.7403634,191.567697 65.8621325,198.148576 C71.0772151,190.971126 75.7283628,183.341335 79.7352139,175.300261 C72.104019,172.400575 64.7949724,168.822202 57.8887866,164.667963 C59.7209612,163.310589 61.5131304,161.891452 63.2445898,160.431257 C105.36741,180.133187 151.134928,180.133187 192.754523,160.431257 C194.506336,161.891452 196.298154,163.310589 198.110326,164.667963 C191.183787,168.842556 183.854737,172.420929 176.223542,175.320965 C180.230393,183.341335 184.861538,190.991831 190.096624,198.16893 C211.238746,191.588051 232.743023,181.531619 254.911949,164.955721 C260.227747,108.668201 245.831087,59.8662432 216.856339,16.5966031 Z M85.4738752,135.09489 C72.8290281,135.09489 62.4592217,123.290155 62.4592217,108.914901 C62.4592217,94.5396472 72.607595,82.7145587 85.4738752,82.7145587 C98.3405064,82.7145587 108.709962,94.5189427 108.488529,108.914901 C108.508531,123.290155 98.3405064,135.09489 85.4738752,135.09489 Z M170.525237,135.09489 C157.88039,135.09489 147.510584,123.290155 147.510584,108.914901 C147.510584,94.5396472 157.658606,82.7145587 170.525237,82.7145587 C183.391518,82.7145587 193.761324,94.5189427 193.539891,108.914901 C193.539891,123.290155 183.391518,135.09489 170.525237,135.09489 Z" fill="#5865F2" fill-rule="nonzero"> </path> </g> </g>
+
</svg>
\ No newline at end of file
diff --git a/api/core/tools/provider/builtin/discord/discord.py b/api/core/tools/provider/builtin/discord/discord.py
new file mode 100644
index 0000000000000..c94824b591cd9
--- /dev/null
+++ b/api/core/tools/provider/builtin/discord/discord.py
@@ -0,0 +1,9 @@
+from typing import Any
+
+from core.tools.provider.builtin.discord.tools.discord_webhook import DiscordWebhookTool
+from core.tools.provider.builtin_tool_provider import BuiltinToolProviderController
+
+
+class DiscordProvider(BuiltinToolProviderController):
+    def _validate_credentials(self, credentials: dict[str, Any]) -> None:
+        DiscordWebhookTool()
diff --git a/api/core/tools/provider/builtin/discord/discord.yaml b/api/core/tools/provider/builtin/discord/discord.yaml
new file mode 100644
index 0000000000000..18b249b5229a0
--- /dev/null
+++ b/api/core/tools/provider/builtin/discord/discord.yaml
@@ -0,0 +1,16 @@
+identity:
+  author: Ice Yao
+  name: discord
+  label:
+    en_US: Discord
+    zh_Hans: Discord
+    pt_BR: Discord
+  description:
+    en_US: Discord Webhook
+    zh_Hans: Discord Webhook
+    pt_BR: Discord Webhook
+  icon: icon.svg
+  tags:
+    - social
+    - productivity
+credentials_for_provider:
diff --git a/api/core/tools/provider/builtin/discord/tools/discord_webhook.py b/api/core/tools/provider/builtin/discord/tools/discord_webhook.py
new file mode 100644
index 0000000000000..dc148994f31e5
--- /dev/null
+++ b/api/core/tools/provider/builtin/discord/tools/discord_webhook.py
@@ -0,0 +1,49 @@
+from typing import Any, Union
+
+import httpx
+
+from core.tools.entities.tool_entities import ToolInvokeMessage
+from core.tools.tool.builtin_tool import BuiltinTool
+
+
+class DiscordWebhookTool(BuiltinTool):
+    def _invoke(self, user_id: str, tool_parameters: dict[str, Any]
+                ) -> Union[ToolInvokeMessage, list[ToolInvokeMessage]]:
+        """
+            Incoming Webhooks
+            API Document:
+                https://discord.com/developers/docs/resources/webhook#execute-webhook
+        """
+
+        content = tool_parameters.get('content', '')
+        if not content:
+            return self.create_text_message('Invalid parameter content')
+
+        webhook_url = tool_parameters.get('webhook_url', '')
+
+        if not webhook_url.startswith('https://discord.com/api/webhooks/'):
+            return self.create_text_message(
+                f'Invalid parameter webhook_url ${webhook_url}, \
+                    not a valid Discord webhook URL')
+
+        headers = {
+            'Content-Type': 'application/json',
+        }
+        params = {}
+        payload = {
+            "content": content,
+        }
+
+        try:
+            res = httpx.post(webhook_url, headers=headers,
+                             params=params, json=payload)
+            if res.is_success:
+                return self.create_text_message(
+                    "Text message was sent successfully")
+            else:
+                return self.create_text_message(
+                    f"Failed to send the text message, \
+                        status code: {res.status_code}, response: {res.text}")
+        except Exception as e:
+            return self.create_text_message(
+                "Failed to send message through webhook. {}".format(e))
diff --git a/api/core/tools/provider/builtin/discord/tools/discord_webhook.yaml b/api/core/tools/provider/builtin/discord/tools/discord_webhook.yaml
new file mode 100644
index 0000000000000..bb3fa43f244d8
--- /dev/null
+++ b/api/core/tools/provider/builtin/discord/tools/discord_webhook.yaml
@@ -0,0 +1,40 @@
+identity:
+  name: discord_webhook
+  author: Ice Yao
+  label:
+    en_US: Incoming Webhook to send message
+    zh_Hans: 通过入站Webhook发送消息
+    pt_BR: Incoming Webhook to send message
+  icon: icon.svg
+description:
+  human:
+    en_US: Sending a message on Discord via the Incoming Webhook
+    zh_Hans: 通过入站Webhook在Discord上发送消息
+    pt_BR: Sending a message on Discord via the Incoming Webhook
+  llm: A tool for sending messages to a chat on Discord.
+parameters:
+  - name: webhook_url
+    type: string
+    required: true
+    label:
+      en_US: Discord Incoming Webhook url
+      zh_Hans: Discord入站Webhook的url
+      pt_BR: Discord Incoming Webhook url
+    human_description:
+      en_US: Discord Incoming Webhook url
+      zh_Hans: Discord入站Webhook的url
+      pt_BR: Discord Incoming Webhook url
+    form: form
+  - name: content
+    type: string
+    required: true
+    label:
+      en_US: content
+      zh_Hans: 消息内容
+      pt_BR: content
+    human_description:
+      en_US: Content to sent to the channel or person.
+      zh_Hans: 消息内容文本
+      pt_BR: Content to sent to the channel or person.
+    llm_description: Content of the message
+    form: llm

From 4373777871b33daa13ae55c52a06f8e76764cbfc Mon Sep 17 00:00:00 2001
From: zhaoyi233 <11971297+zhao85@users.noreply.github.com>
Date: Thu, 3 Oct 2024 10:20:56 +0800
Subject: [PATCH 137/235] Update json_in_md_parser.py (#8983)

Co-authored-by: crazywoola <427733928@qq.com>
---
 .../builtin/discord/tools/discord_webhook.py  | 38 +++++++++----------
 api/libs/json_in_md_parser.py                 | 37 +++++++++---------
 2 files changed, 39 insertions(+), 36 deletions(-)

diff --git a/api/core/tools/provider/builtin/discord/tools/discord_webhook.py b/api/core/tools/provider/builtin/discord/tools/discord_webhook.py
index dc148994f31e5..7fdf791abad3f 100644
--- a/api/core/tools/provider/builtin/discord/tools/discord_webhook.py
+++ b/api/core/tools/provider/builtin/discord/tools/discord_webhook.py
@@ -7,27 +7,29 @@
 
 
 class DiscordWebhookTool(BuiltinTool):
-    def _invoke(self, user_id: str, tool_parameters: dict[str, Any]
-                ) -> Union[ToolInvokeMessage, list[ToolInvokeMessage]]:
+    def _invoke(
+        self, user_id: str, tool_parameters: dict[str, Any]
+    ) -> Union[ToolInvokeMessage, list[ToolInvokeMessage]]:
         """
-            Incoming Webhooks
-            API Document:
-                https://discord.com/developers/docs/resources/webhook#execute-webhook
+        Incoming Webhooks
+        API Document:
+            https://discord.com/developers/docs/resources/webhook#execute-webhook
         """
 
-        content = tool_parameters.get('content', '')
+        content = tool_parameters.get("content", "")
         if not content:
-            return self.create_text_message('Invalid parameter content')
+            return self.create_text_message("Invalid parameter content")
 
-        webhook_url = tool_parameters.get('webhook_url', '')
+        webhook_url = tool_parameters.get("webhook_url", "")
 
-        if not webhook_url.startswith('https://discord.com/api/webhooks/'):
+        if not webhook_url.startswith("https://discord.com/api/webhooks/"):
             return self.create_text_message(
-                f'Invalid parameter webhook_url ${webhook_url}, \
-                    not a valid Discord webhook URL')
+                f"Invalid parameter webhook_url ${webhook_url}, \
+                    not a valid Discord webhook URL"
+            )
 
         headers = {
-            'Content-Type': 'application/json',
+            "Content-Type": "application/json",
         }
         params = {}
         payload = {
@@ -35,15 +37,13 @@ def _invoke(self, user_id: str, tool_parameters: dict[str, Any]
         }
 
         try:
-            res = httpx.post(webhook_url, headers=headers,
-                             params=params, json=payload)
+            res = httpx.post(webhook_url, headers=headers, params=params, json=payload)
             if res.is_success:
-                return self.create_text_message(
-                    "Text message was sent successfully")
+                return self.create_text_message("Text message was sent successfully")
             else:
                 return self.create_text_message(
                     f"Failed to send the text message, \
-                        status code: {res.status_code}, response: {res.text}")
+                        status code: {res.status_code}, response: {res.text}"
+                )
         except Exception as e:
-            return self.create_text_message(
-                "Failed to send message through webhook. {}".format(e))
+            return self.create_text_message("Failed to send message through webhook. {}".format(e))
diff --git a/api/libs/json_in_md_parser.py b/api/libs/json_in_md_parser.py
index 185ff3f95eba3..9131408817bbe 100644
--- a/api/libs/json_in_md_parser.py
+++ b/api/libs/json_in_md_parser.py
@@ -4,25 +4,28 @@
 
 
 def parse_json_markdown(json_string: str) -> dict:
-    # Remove the triple backticks if present
+    # Get json from the backticks/braces
     json_string = json_string.strip()
-    start_index = json_string.find("```json")
-    end_index = json_string.find("```", start_index + len("```json"))
-
-    if start_index != -1 and end_index != -1:
-        extracted_content = json_string[start_index + len("```json") : end_index].strip()
-
-        # Parse the JSON string into a Python dictionary
-        parsed = json.loads(extracted_content)
-    elif start_index != -1 and end_index == -1 and json_string.endswith("``"):
-        end_index = json_string.find("``", start_index + len("```json"))
-        extracted_content = json_string[start_index + len("```json") : end_index].strip()
-
-        # Parse the JSON string into a Python dictionary
+    starts = ["```json", "```", "``", "`", "{"]
+    ends = ["```", "``", "`", "}"]
+    end_index = -1
+    for s in starts:
+        start_index = json_string.find(s)
+        if start_index != -1:
+            if json_string[start_index] != "{":
+                start_index += len(s)
+            break
+    if start_index != -1:
+        for e in ends:
+            end_index = json_string.rfind(e, start_index)
+            if end_index != -1:
+                if json_string[end_index] == "}":
+                    end_index += 1
+                break
+    if start_index != -1 and end_index != -1 and start_index < end_index:
+        extracted_content = json_string[start_index:end_index].strip()
+        print("content:", extracted_content, start_index, end_index)
         parsed = json.loads(extracted_content)
-    elif json_string.startswith("{"):
-        # Parse the JSON string into a Python dictionary
-        parsed = json.loads(json_string)
     else:
         raise Exception("Could not find JSON block in the output.")
 

From dc5839b6bb560dbc7e65571581311edba2161612 Mon Sep 17 00:00:00 2001
From: Giannis Kepas <kepas.giannis@gmail.com>
Date: Thu, 3 Oct 2024 10:17:28 +0300
Subject: [PATCH 138/235] feat: Update AWS Bedrock supported regions (#8992)

---
 .../model_providers/bedrock/bedrock.yaml      | 38 ++++++++++++++++---
 1 file changed, 33 insertions(+), 5 deletions(-)

diff --git a/api/core/model_runtime/model_providers/bedrock/bedrock.yaml b/api/core/model_runtime/model_providers/bedrock/bedrock.yaml
index c540ee23b3167..952f968b9d0f2 100644
--- a/api/core/model_runtime/model_providers/bedrock/bedrock.yaml
+++ b/api/core/model_runtime/model_providers/bedrock/bedrock.yaml
@@ -50,34 +50,62 @@ provider_credential_schema:
           label:
             en_US: US East (N. Virginia)
             zh_Hans: 美国东部 (弗吉尼亚北部)
+        - value: us-east-2
+          label:
+            en_US: US East (Ohio)
+            zh_Hans: 美国东部 (弗吉尼亚北部)
         - value: us-west-2
           label:
             en_US: US West (Oregon)
             zh_Hans: 美国西部 (俄勒冈州)
+        - value: ap-south-1
+          label:
+            en_US: Asia Pacific (Mumbai)
+            zh_Hans: 亚太地区（孟买）
         - value: ap-southeast-1
           label:
             en_US: Asia Pacific (Singapore)
             zh_Hans: 亚太地区 (新加坡)
+        - value: ap-southeast-2
+          label:
+            en_US: Asia Pacific (Sydney)
+            zh_Hans: 亚太地区 (悉尼)
         - value: ap-northeast-1
           label:
             en_US: Asia Pacific (Tokyo)
             zh_Hans: 亚太地区 (东京)
+        - value: ap-northeast-2
+          label:
+            en_US: Asia Pacific (Seoul)
+            zh_Hans: 亚太地区（首尔）
+        - value: ca-central-1
+          label:
+            en_US: Canada (Central)
+            zh_Hans: 加拿大（中部）
         - value: eu-central-1
           label:
             en_US: Europe (Frankfurt)
             zh_Hans: 欧洲 (法兰克福)
+        - value: eu-west-1
+          label:
+            en_US: Europe (Ireland)
+            zh_Hans: 欧洲（爱尔兰）
         - value: eu-west-2
           label:
-            en_US: Eu west London (London)
+            en_US: Europe (London)
             zh_Hans: 欧洲西部 (伦敦)
+        - value: eu-west-3
+          label:
+            en_US: Europe (Paris)
+            zh_Hans: 欧洲（巴黎）
+        - value: sa-east-1
+          label:
+            en_US: South America (São Paulo)
+            zh_Hans: 南美洲（圣保罗）
         - value: us-gov-west-1
           label:
             en_US: AWS GovCloud (US-West)
             zh_Hans: AWS GovCloud (US-West)
-        - value: ap-southeast-2
-          label:
-            en_US: Asia Pacific (Sydney)
-            zh_Hans: 亚太地区 (悉尼)
     - variable: model_for_validation
       required: false
       label:

From 9815a0911bca9519469a5f215392487503febbd4 Mon Sep 17 00:00:00 2001
From: crazywoola <100913391+crazywoola@users.noreply.github.com>
Date: Thu, 3 Oct 2024 21:53:11 +0800
Subject: [PATCH 139/235] fix: tools description is missing (#8999)

---
 web/i18n/language.ts | 6 ++++--
 1 file changed, 4 insertions(+), 2 deletions(-)

diff --git a/web/i18n/language.ts b/web/i18n/language.ts
index c2e23cc19a638..fde69328bd922 100644
--- a/web/i18n/language.ts
+++ b/web/i18n/language.ts
@@ -31,8 +31,10 @@ export const languages = data.languages
 export const LanguagesSupported = languages.filter(item => item.supported).map(item => item.value)
 
 export const getLanguage = (locale: string) => {
-  const supportedLocale = LanguagesSupported.find(lang => lang.startsWith(locale.split('-')[0]))
-  return (supportedLocale || LanguagesSupported[0]).replace('-', '_')
+  if (locale === 'zh-Hans')
+    return locale.replace('-', '_')
+
+  return LanguagesSupported[0].replace('-', '_')
 }
 
 export const NOTICE_I18N = {

From e1e2d0b3643bc4f17566f4f7f066f2c1bf0ba7e2 Mon Sep 17 00:00:00 2001
From: zhuhao <37029601+hwzhuhao@users.noreply.github.com>
Date: Mon, 7 Oct 2024 11:06:08 +0800
Subject: [PATCH 140/235] fix: failed to open links to images generated by QR
 code tool when using Huawei OBS (#9034)

---
 api/extensions/storage/huawei_storage.py | 3 ++-
 1 file changed, 2 insertions(+), 1 deletion(-)

diff --git a/api/extensions/storage/huawei_storage.py b/api/extensions/storage/huawei_storage.py
index 269a008fba3f2..b6ce10bd1f7a9 100644
--- a/api/extensions/storage/huawei_storage.py
+++ b/api/extensions/storage/huawei_storage.py
@@ -29,7 +29,8 @@ def load_once(self, filename: str) -> bytes:
     def load_stream(self, filename: str) -> Generator:
         def generate(filename: str = filename) -> Generator:
             response = self.client.getObject(bucketName=self.bucket_name, objectKey=filename)["body"].response
-            yield from response.read(4096)
+            while chunk := response.read(4096):
+                yield chunk
 
         return generate()
 

From fcfa1252a052455f2b199e0be555d9098ed28cf2 Mon Sep 17 00:00:00 2001
From: zg0d233 <11971297+zhao85@users.noreply.github.com>
Date: Mon, 7 Oct 2024 11:06:38 +0800
Subject: [PATCH 141/235] fix bug when adding openai or openai-compatible stt
 model instance (#9006)

---
 .../agent/output_parser/cot_output_parser.py    |  2 +-
 .../openai/speech2text/speech2text.py           | 17 +++++++++++++++++
 .../speech2text/speech2text.py                  | 17 +++++++++++++++++
 api/core/tools/tool/workflow_tool.py            | 11 +++++++----
 4 files changed, 42 insertions(+), 5 deletions(-)

diff --git a/api/core/agent/output_parser/cot_output_parser.py b/api/core/agent/output_parser/cot_output_parser.py
index d04e38777a54a..99876b2f5eb3e 100644
--- a/api/core/agent/output_parser/cot_output_parser.py
+++ b/api/core/agent/output_parser/cot_output_parser.py
@@ -14,7 +14,7 @@ def handle_react_stream_output(
     ) -> Generator[Union[str, AgentScratchpadUnit.Action], None, None]:
         def parse_action(json_str):
             try:
-                action = json.loads(json_str)
+                action = json.loads(json_str, strict=False)
                 action_name = None
                 action_input = None
 
diff --git a/api/core/model_runtime/model_providers/openai/speech2text/speech2text.py b/api/core/model_runtime/model_providers/openai/speech2text/speech2text.py
index 18f97e45f33bd..0d54d2ea9a9dd 100644
--- a/api/core/model_runtime/model_providers/openai/speech2text/speech2text.py
+++ b/api/core/model_runtime/model_providers/openai/speech2text/speech2text.py
@@ -2,6 +2,8 @@
 
 from openai import OpenAI
 
+from core.model_runtime.entities.common_entities import I18nObject
+from core.model_runtime.entities.model_entities import AIModelEntity, FetchFrom, ModelType
 from core.model_runtime.errors.validate import CredentialsValidateFailedError
 from core.model_runtime.model_providers.__base.speech2text_model import Speech2TextModel
 from core.model_runtime.model_providers.openai._common import _CommonOpenAI
@@ -58,3 +60,18 @@ def _speech2text_invoke(self, model: str, credentials: dict, file: IO[bytes]) ->
         response = client.audio.transcriptions.create(model=model, file=file)
 
         return response.text
+
+    def get_customizable_model_schema(self, model: str, credentials: dict) -> AIModelEntity | None:
+        """
+        used to define customizable model schema
+        """
+        entity = AIModelEntity(
+            model=model,
+            label=I18nObject(en_US=model),
+            fetch_from=FetchFrom.CUSTOMIZABLE_MODEL,
+            model_type=ModelType.SPEECH2TEXT,
+            model_properties={},
+            parameter_rules=[],
+        )
+
+        return entity
diff --git a/api/core/model_runtime/model_providers/openai_api_compatible/speech2text/speech2text.py b/api/core/model_runtime/model_providers/openai_api_compatible/speech2text/speech2text.py
index 405096578cdd5..cef77cc9419d1 100644
--- a/api/core/model_runtime/model_providers/openai_api_compatible/speech2text/speech2text.py
+++ b/api/core/model_runtime/model_providers/openai_api_compatible/speech2text/speech2text.py
@@ -3,6 +3,8 @@
 
 import requests
 
+from core.model_runtime.entities.common_entities import I18nObject
+from core.model_runtime.entities.model_entities import AIModelEntity, FetchFrom, ModelType
 from core.model_runtime.errors.invoke import InvokeBadRequestError
 from core.model_runtime.errors.validate import CredentialsValidateFailedError
 from core.model_runtime.model_providers.__base.speech2text_model import Speech2TextModel
@@ -59,3 +61,18 @@ def validate_credentials(self, model: str, credentials: dict) -> None:
                 self._invoke(model, credentials, audio_file)
         except Exception as ex:
             raise CredentialsValidateFailedError(str(ex))
+
+    def get_customizable_model_schema(self, model: str, credentials: dict) -> AIModelEntity | None:
+        """
+        used to define customizable model schema
+        """
+        entity = AIModelEntity(
+            model=model,
+            label=I18nObject(en_US=model),
+            fetch_from=FetchFrom.CUSTOMIZABLE_MODEL,
+            model_type=ModelType.SPEECH2TEXT,
+            model_properties={},
+            parameter_rules=[],
+        )
+
+        return entity
diff --git a/api/core/tools/tool/workflow_tool.py b/api/core/tools/tool/workflow_tool.py
index ad0c7fc631fd0..a885b8784f27f 100644
--- a/api/core/tools/tool/workflow_tool.py
+++ b/api/core/tools/tool/workflow_tool.py
@@ -68,10 +68,13 @@ def _invoke(
 
         result = []
 
-        outputs = data.get("outputs", {})
-        outputs, files = self._extract_files(outputs)
-        for file in files:
-            result.append(self.create_file_var_message(file))
+        outputs = data.get("outputs")
+        if outputs == None:
+            outputs = {}
+        else:
+            outputs, files = self._extract_files(outputs)
+            for file in files:
+                result.append(self.create_file_var_message(file))
 
         result.append(self.create_text_message(json.dumps(outputs, ensure_ascii=False)))
         result.append(self.create_json_message(outputs))

From 4480b469a68f3a14e7b6dc52792e12996c321c80 Mon Sep 17 00:00:00 2001
From: Bowen Liang <liangbowen@gf.com.cn>
Date: Mon, 7 Oct 2024 11:07:34 +0800
Subject: [PATCH 142/235] chore: fix the yanked dependency vesion
 aiohappyeyeballs 2.4.2 (#9020)

---
 api/poetry.lock | 6 +++---
 1 file changed, 3 insertions(+), 3 deletions(-)

diff --git a/api/poetry.lock b/api/poetry.lock
index 4fb8d54f11852..96c2979729460 100644
--- a/api/poetry.lock
+++ b/api/poetry.lock
@@ -2,13 +2,13 @@
 
 [[package]]
 name = "aiohappyeyeballs"
-version = "2.4.2"
+version = "2.4.3"
 description = "Happy Eyeballs for asyncio"
 optional = false
 python-versions = ">=3.8"
 files = [
-    {file = "aiohappyeyeballs-2.4.2-py3-none-any.whl", hash = "sha256:8522691d9a154ba1145b157d6d5c15e5c692527ce6a53c5e5f9876977f6dab2f"},
-    {file = "aiohappyeyeballs-2.4.2.tar.gz", hash = "sha256:4ca893e6c5c1f5bf3888b04cb5a3bee24995398efef6e0b9f747b5e89d84fd74"},
+    {file = "aiohappyeyeballs-2.4.3-py3-none-any.whl", hash = "sha256:8a7a83727b2756f394ab2895ea0765a0a8c475e3c71e98d43d76f22b4b435572"},
+    {file = "aiohappyeyeballs-2.4.3.tar.gz", hash = "sha256:75cf88a15106a5002a8eb1dab212525c00d1f4c0fa96e551c9fbe6f09a621586"},
 ]
 
 [[package]]

From 959a81a41b2e7d2a3f7a1f1d92961cc35c4727b3 Mon Sep 17 00:00:00 2001
From: zhuhao <37029601+hwzhuhao@users.noreply.github.com>
Date: Mon, 7 Oct 2024 11:08:06 +0800
Subject: [PATCH 143/235] refactor: remove the duplicate definitions across
 different modules (#9022)

---
 api/controllers/files/error.py         | 7 +++++++
 api/controllers/files/image_preview.py | 8 +-------
 api/controllers/files/tool_files.py    | 8 +-------
 3 files changed, 9 insertions(+), 14 deletions(-)
 create mode 100644 api/controllers/files/error.py

diff --git a/api/controllers/files/error.py b/api/controllers/files/error.py
new file mode 100644
index 0000000000000..a7ce4cd6f793e
--- /dev/null
+++ b/api/controllers/files/error.py
@@ -0,0 +1,7 @@
+from libs.exception import BaseHTTPException
+
+
+class UnsupportedFileTypeError(BaseHTTPException):
+    error_code = "unsupported_file_type"
+    description = "File type not allowed."
+    code = 415
diff --git a/api/controllers/files/image_preview.py b/api/controllers/files/image_preview.py
index 2432285d9365a..a56c1c332d2c1 100644
--- a/api/controllers/files/image_preview.py
+++ b/api/controllers/files/image_preview.py
@@ -4,7 +4,7 @@
 
 import services
 from controllers.files import api
-from libs.exception import BaseHTTPException
+from controllers.files.error import UnsupportedFileTypeError
 from services.account_service import TenantService
 from services.file_service import FileService
 
@@ -50,9 +50,3 @@ def get(self, workspace_id):
 
 api.add_resource(ImagePreviewApi, "/files/<uuid:file_id>/image-preview")
 api.add_resource(WorkspaceWebappLogoApi, "/files/workspaces/<uuid:workspace_id>/webapp-logo")
-
-
-class UnsupportedFileTypeError(BaseHTTPException):
-    error_code = "unsupported_file_type"
-    description = "File type not allowed."
-    code = 415
diff --git a/api/controllers/files/tool_files.py b/api/controllers/files/tool_files.py
index 38ac0815daa10..406cd4221493c 100644
--- a/api/controllers/files/tool_files.py
+++ b/api/controllers/files/tool_files.py
@@ -3,8 +3,8 @@
 from werkzeug.exceptions import Forbidden, NotFound
 
 from controllers.files import api
+from controllers.files.error import UnsupportedFileTypeError
 from core.tools.tool_file_manager import ToolFileManager
-from libs.exception import BaseHTTPException
 
 
 class ToolFilePreviewApi(Resource):
@@ -43,9 +43,3 @@ def get(self, file_id, extension):
 
 
 api.add_resource(ToolFilePreviewApi, "/files/tools/<uuid:file_id>.<string:extension>")
-
-
-class UnsupportedFileTypeError(BaseHTTPException):
-    error_code = "unsupported_file_type"
-    description = "File type not allowed."
-    code = 415

From 2571b0c4e36b7ce7a3ada94aac751ad20f319e1f Mon Sep 17 00:00:00 2001
From: zhuhao <37029601+hwzhuhao@users.noreply.github.com>
Date: Mon, 7 Oct 2024 11:09:27 +0800
Subject: [PATCH 144/235] feat: add baidu obs storage (#9024)

---
 api/.env.example                              |   8 +-
 api/configs/middleware/__init__.py            |   8 +-
 .../storage/baidu_obs_storage_config.py       |  29 ++
 api/extensions/ext_storage.py                 |   3 +
 api/extensions/storage/baidu_storage.py       |  60 +++
 api/poetry.lock                               | 342 ++++++++++--------
 api/pyproject.toml                            |   1 +
 7 files changed, 304 insertions(+), 147 deletions(-)
 create mode 100644 api/configs/middleware/storage/baidu_obs_storage_config.py
 create mode 100644 api/extensions/storage/baidu_storage.py

diff --git a/api/.env.example b/api/.env.example
index 7e5c636ff8975..fb0fc045b0850 100644
--- a/api/.env.example
+++ b/api/.env.example
@@ -39,7 +39,7 @@ DB_DATABASE=dify
 
 # Storage configuration
 # use for store upload files, private keys...
-# storage type: local, s3, azure-blob, google-storage, tencent-cos, huawei-obs, volcengine-tos
+# storage type: local, s3, azure-blob, google-storage, tencent-cos, huawei-obs, volcengine-tos, baidu-obs
 STORAGE_TYPE=local
 STORAGE_LOCAL_PATH=storage
 S3_USE_AWS_MANAGED_IAM=false
@@ -79,6 +79,12 @@ HUAWEI_OBS_SECRET_KEY=your-secret-key
 HUAWEI_OBS_ACCESS_KEY=your-access-key
 HUAWEI_OBS_SERVER=your-server-url
 
+# Baidu OBS Storage Configuration
+BAIDU_OBS_BUCKET_NAME=your-bucket-name
+BAIDU_OBS_SECRET_KEY=your-secret-key
+BAIDU_OBS_ACCESS_KEY=your-access-key
+BAIDU_OBS_ENDPOINT=your-server-url
+
 # OCI Storage configuration
 OCI_ENDPOINT=your-endpoint
 OCI_BUCKET_NAME=your-bucket-name
diff --git a/api/configs/middleware/__init__.py b/api/configs/middleware/__init__.py
index 6ad216c1913f9..8626236856ddb 100644
--- a/api/configs/middleware/__init__.py
+++ b/api/configs/middleware/__init__.py
@@ -8,6 +8,7 @@
 from configs.middleware.storage.aliyun_oss_storage_config import AliyunOSSStorageConfig
 from configs.middleware.storage.amazon_s3_storage_config import S3StorageConfig
 from configs.middleware.storage.azure_blob_storage_config import AzureBlobStorageConfig
+from configs.middleware.storage.baidu_obs_storage_config import BaiduOBSStorageConfig
 from configs.middleware.storage.google_cloud_storage_config import GoogleCloudStorageConfig
 from configs.middleware.storage.huawei_obs_storage_config import HuaweiCloudOBSStorageConfig
 from configs.middleware.storage.oci_storage_config import OCIStorageConfig
@@ -200,12 +201,13 @@ class MiddlewareConfig(
     StorageConfig,
     AliyunOSSStorageConfig,
     AzureBlobStorageConfig,
+    BaiduOBSStorageConfig,
     GoogleCloudStorageConfig,
-    TencentCloudCOSStorageConfig,
     HuaweiCloudOBSStorageConfig,
-    VolcengineTOSStorageConfig,
-    S3StorageConfig,
     OCIStorageConfig,
+    S3StorageConfig,
+    TencentCloudCOSStorageConfig,
+    VolcengineTOSStorageConfig,
     # configs of vdb and vdb providers
     VectorStoreConfig,
     AnalyticdbConfig,
diff --git a/api/configs/middleware/storage/baidu_obs_storage_config.py b/api/configs/middleware/storage/baidu_obs_storage_config.py
new file mode 100644
index 0000000000000..c511628a1514a
--- /dev/null
+++ b/api/configs/middleware/storage/baidu_obs_storage_config.py
@@ -0,0 +1,29 @@
+from typing import Optional
+
+from pydantic import BaseModel, Field
+
+
+class BaiduOBSStorageConfig(BaseModel):
+    """
+    Configuration settings for Baidu Object Storage Service (OBS)
+    """
+
+    BAIDU_OBS_BUCKET_NAME: Optional[str] = Field(
+        description="Name of the Baidu OBS bucket to store and retrieve objects (e.g., 'my-obs-bucket')",
+        default=None,
+    )
+
+    BAIDU_OBS_ACCESS_KEY: Optional[str] = Field(
+        description="Access Key ID for authenticating with Baidu OBS",
+        default=None,
+    )
+
+    BAIDU_OBS_SECRET_KEY: Optional[str] = Field(
+        description="Secret Access Key for authenticating with Baidu OBS",
+        default=None,
+    )
+
+    BAIDU_OBS_ENDPOINT: Optional[str] = Field(
+        description="URL of the Baidu OSS endpoint for your chosen region (e.g., 'https://.bj.bcebos.com')",
+        default=None,
+    )
diff --git a/api/extensions/ext_storage.py b/api/extensions/ext_storage.py
index 1e6530f6f4745..863d7a9227e49 100644
--- a/api/extensions/ext_storage.py
+++ b/api/extensions/ext_storage.py
@@ -6,6 +6,7 @@
 
 from extensions.storage.aliyun_storage import AliyunStorage
 from extensions.storage.azure_storage import AzureStorage
+from extensions.storage.baidu_storage import BaiduStorage
 from extensions.storage.google_storage import GoogleStorage
 from extensions.storage.huawei_storage import HuaweiStorage
 from extensions.storage.local_storage import LocalStorage
@@ -35,6 +36,8 @@ def init_app(self, app: Flask):
             self.storage_runner = OCIStorage(app=app)
         elif storage_type == "huawei-obs":
             self.storage_runner = HuaweiStorage(app=app)
+        elif storage_type == "baidu-obs":
+            self.storage_runner = BaiduStorage(app=app)
         elif storage_type == "volcengine-tos":
             self.storage_runner = VolcengineStorage(app=app)
         else:
diff --git a/api/extensions/storage/baidu_storage.py b/api/extensions/storage/baidu_storage.py
new file mode 100644
index 0000000000000..691ac7a58bb5d
--- /dev/null
+++ b/api/extensions/storage/baidu_storage.py
@@ -0,0 +1,60 @@
+import base64
+import hashlib
+from collections.abc import Generator
+
+from baidubce.auth.bce_credentials import BceCredentials
+from baidubce.bce_client_configuration import BceClientConfiguration
+from baidubce.services.bos.bos_client import BosClient
+from flask import Flask
+
+from extensions.storage.base_storage import BaseStorage
+
+
+class BaiduStorage(BaseStorage):
+    """Implementation for baidu obs storage."""
+
+    def __init__(self, app: Flask):
+        super().__init__(app)
+        app_config = self.app.config
+        self.bucket_name = app_config.get("BAIDU_OBS_BUCKET_NAME")
+        client_config = BceClientConfiguration(
+            credentials=BceCredentials(
+                access_key_id=app_config.get("BAIDU_OBS_ACCESS_KEY"),
+                secret_access_key=app_config.get("BAIDU_OBS_SECRET_KEY"),
+            ),
+            endpoint=app_config.get("BAIDU_OBS_ENDPOINT"),
+        )
+
+        self.client = BosClient(config=client_config)
+
+    def save(self, filename, data):
+        md5 = hashlib.md5()
+        md5.update(data)
+        content_md5 = base64.standard_b64encode(md5.digest())
+        self.client.put_object(
+            bucket_name=self.bucket_name, key=filename, data=data, content_length=len(data), content_md5=content_md5
+        )
+
+    def load_once(self, filename: str) -> bytes:
+        response = self.client.get_object(bucket_name=self.bucket_name, key=filename)
+        return response.data.read()
+
+    def load_stream(self, filename: str) -> Generator:
+        def generate(filename: str = filename) -> Generator:
+            response = self.client.get_object(bucket_name=self.bucket_name, key=filename).data
+            while chunk := response.read(4096):
+                yield chunk
+
+        return generate()
+
+    def download(self, filename, target_filepath):
+        self.client.get_object_to_file(bucket_name=self.bucket_name, key=filename, file_name=target_filepath)
+
+    def exists(self, filename):
+        res = self.client.get_object_meta_data(bucket_name=self.bucket_name, key=filename)
+        if res is None:
+            return False
+        return True
+
+    def delete(self, filename):
+        self.client.delete_object(bucket_name=self.bucket_name, key=filename)
diff --git a/api/poetry.lock b/api/poetry.lock
index 96c2979729460..7ea8e49941364 100644
--- a/api/poetry.lock
+++ b/api/poetry.lock
@@ -727,6 +727,22 @@ files = [
     {file = "backoff-2.2.1.tar.gz", hash = "sha256:03f829f5bb1923180821643f8753b0502c3b682293992485b0eef2807afa5cba"},
 ]
 
+[[package]]
+name = "bce-python-sdk"
+version = "0.9.23"
+description = "BCE SDK for python"
+optional = false
+python-versions = ">=2.7, !=3.0.*, !=3.1.*, !=3.2.*, <4"
+files = [
+    {file = "bce_python_sdk-0.9.23-py3-none-any.whl", hash = "sha256:8debe21a040e00060f6044877d594765ed7b18bc765c6bf16b878bca864140a3"},
+    {file = "bce_python_sdk-0.9.23.tar.gz", hash = "sha256:19739fed5cd0725356fc5ffa2acbdd8fb23f2a81edb91db21a03174551d0cf41"},
+]
+
+[package.dependencies]
+future = ">=0.6.0"
+pycryptodome = ">=3.8.0"
+six = ">=1.4.0"
+
 [[package]]
 name = "bcrypt"
 version = "4.2.0"
@@ -828,13 +844,13 @@ crt = ["botocore[crt] (>=1.21.0,<2.0a0)"]
 
 [[package]]
 name = "botocore"
-version = "1.35.29"
+version = "1.35.34"
 description = "Low-level, data-driven core of boto 3."
 optional = false
-python-versions = ">=3.8"
+python-versions = ">= 3.8"
 files = [
-    {file = "botocore-1.35.29-py3-none-any.whl", hash = "sha256:f8e3ae0d84214eff3fb69cb4dc51cea6c43d3bde82027a94d00c52b941d6c3d5"},
-    {file = "botocore-1.35.29.tar.gz", hash = "sha256:4ed28ab03675bb008a290c452c5ddd7aaa5d4e3fa1912aadbdf93057ee84362b"},
+    {file = "botocore-1.35.34-py3-none-any.whl", hash = "sha256:ccb0fe397b11b81c9abc0c87029d17298e17bf658d8db5c0c5a551a12a207e7a"},
+    {file = "botocore-1.35.34.tar.gz", hash = "sha256:789b6501a3bb4a9591c1fe10da200cc315c1fa5df5ada19c720d8ef06439b3e3"},
 ]
 
 [package.dependencies]
@@ -843,7 +859,7 @@ python-dateutil = ">=2.1,<3.0.0"
 urllib3 = {version = ">=1.25.4,<2.2.0 || >2.2.0,<3", markers = "python_version >= \"3.10\""}
 
 [package.extras]
-crt = ["awscrt (==0.21.5)"]
+crt = ["awscrt (==0.22.0)"]
 
 [[package]]
 name = "bottleneck"
@@ -1049,13 +1065,13 @@ beautifulsoup4 = "*"
 
 [[package]]
 name = "build"
-version = "1.2.2"
+version = "1.2.2.post1"
 description = "A simple, correct Python build frontend"
 optional = false
-python-versions = ">=3.8"
+python-versions = ">= 3.8"
 files = [
-    {file = "build-1.2.2-py3-none-any.whl", hash = "sha256:277ccc71619d98afdd841a0e96ac9fe1593b823af481d3b0cea748e8894e0613"},
-    {file = "build-1.2.2.tar.gz", hash = "sha256:119b2fb462adef986483438377a13b2f42064a2a3a4161f24a0cca698a07ac8c"},
+    {file = "build-1.2.2.post1-py3-none-any.whl", hash = "sha256:1d61c0887fa860c01971625baae8bdd338e517b836a2f70dd1f7aa3a6b2fc5b5"},
+    {file = "build-1.2.2.post1.tar.gz", hash = "sha256:b36993e92ca9375a219c99e606a122ff365a760a2d4bba0caa09bd5278b608b7"},
 ]
 
 [package.dependencies]
@@ -2321,12 +2337,13 @@ lxml = ["lxml (>=5.2.2)"]
 
 [[package]]
 name = "durationpy"
-version = "0.7"
+version = "0.9"
 description = "Module for converting between datetime.timedelta and Go's Duration strings."
 optional = false
 python-versions = "*"
 files = [
-    {file = "durationpy-0.7.tar.gz", hash = "sha256:8447c43df4f1a0b434e70c15a38d77f5c9bd17284bfc1ff1d430f233d5083732"},
+    {file = "durationpy-0.9-py3-none-any.whl", hash = "sha256:e65359a7af5cedad07fb77a2dd3f390f8eb0b74cb845589fa6c057086834dd38"},
+    {file = "durationpy-0.9.tar.gz", hash = "sha256:fd3feb0a69a0057d582ef643c355c40d2fa1c942191f914d12203b1a01ac722a"},
 ]
 
 [[package]]
@@ -2372,13 +2389,13 @@ vectorstore-mmr = ["numpy (>=1)", "simsimd (>=3)"]
 
 [[package]]
 name = "emoji"
-version = "2.13.2"
+version = "2.14.0"
 description = "Emoji for Python"
 optional = false
 python-versions = ">=3.7"
 files = [
-    {file = "emoji-2.13.2-py3-none-any.whl", hash = "sha256:ef6f2ee63b245e934c763b1a9a0637713955aa3d9e322432e036bb60559de4d6"},
-    {file = "emoji-2.13.2.tar.gz", hash = "sha256:f95d10d96c5f21299ed2c4b32511611ba890b8c07f5f2bf5b04d5d3eee91fd19"},
+    {file = "emoji-2.14.0-py3-none-any.whl", hash = "sha256:fcc936bf374b1aec67dda5303ae99710ba88cc9cdce2d1a71c5f2204e6d78799"},
+    {file = "emoji-2.14.0.tar.gz", hash = "sha256:f68ac28915a2221667cddb3e6c589303c3c6954c6c5af6fefaec7f9bdf72fdca"},
 ]
 
 [package.extras]
@@ -2785,44 +2802,49 @@ woff = ["brotli (>=1.0.1)", "brotlicffi (>=0.8.0)", "zopfli (>=0.1.4)"]
 
 [[package]]
 name = "frozendict"
-version = "2.4.4"
+version = "2.4.5"
 description = "A simple immutable dictionary"
 optional = false
 python-versions = ">=3.6"
 files = [
-    {file = "frozendict-2.4.4-cp310-cp310-macosx_11_0_arm64.whl", hash = "sha256:4a59578d47b3949437519b5c39a016a6116b9e787bb19289e333faae81462e59"},
-    {file = "frozendict-2.4.4-cp310-cp310-manylinux_2_17_aarch64.manylinux2014_aarch64.whl", hash = "sha256:12a342e439aef28ccec533f0253ea53d75fe9102bd6ea928ff530e76eac38906"},
-    {file = "frozendict-2.4.4-cp310-cp310-manylinux_2_17_x86_64.manylinux2014_x86_64.whl", hash = "sha256:7f79c26dff10ce11dad3b3627c89bb2e87b9dd5958c2b24325f16a23019b8b94"},
-    {file = "frozendict-2.4.4-cp310-cp310-musllinux_1_1_aarch64.whl", hash = "sha256:2bd009cf4fc47972838a91e9b83654dc9a095dc4f2bb3a37c3f3124c8a364543"},
-    {file = "frozendict-2.4.4-cp310-cp310-musllinux_1_1_x86_64.whl", hash = "sha256:87ebcde21565a14fe039672c25550060d6f6d88cf1f339beac094c3b10004eb0"},
-    {file = "frozendict-2.4.4-cp310-cp310-win_amd64.whl", hash = "sha256:fefeb700bc7eb8b4c2dc48704e4221860d254c8989fb53488540bc44e44a1ac2"},
-    {file = "frozendict-2.4.4-cp310-cp310-win_arm64.whl", hash = "sha256:4297d694eb600efa429769125a6f910ec02b85606f22f178bafbee309e7d3ec7"},
-    {file = "frozendict-2.4.4-cp36-cp36m-manylinux_2_17_aarch64.manylinux2014_aarch64.whl", hash = "sha256:812ab17522ba13637826e65454115a914c2da538356e85f43ecea069813e4b33"},
-    {file = "frozendict-2.4.4-cp36-cp36m-manylinux_2_17_x86_64.manylinux2014_x86_64.whl", hash = "sha256:7fee9420475bb6ff357000092aa9990c2f6182b2bab15764330f4ad7de2eae49"},
-    {file = "frozendict-2.4.4-cp36-cp36m-musllinux_1_1_aarch64.whl", hash = "sha256:3148062675536724502c6344d7c485dd4667fdf7980ca9bd05e338ccc0c4471e"},
-    {file = "frozendict-2.4.4-cp36-cp36m-musllinux_1_1_x86_64.whl", hash = "sha256:78c94991944dd33c5376f720228e5b252ee67faf3bac50ef381adc9e51e90d9d"},
-    {file = "frozendict-2.4.4-cp36-cp36m-win_amd64.whl", hash = "sha256:1697793b5f62b416c0fc1d94638ec91ed3aa4ab277f6affa3a95216ecb3af170"},
-    {file = "frozendict-2.4.4-cp37-cp37m-manylinux_2_17_aarch64.manylinux2014_aarch64.whl", hash = "sha256:199a4d32194f3afed6258de7e317054155bc9519252b568d9cfffde7e4d834e5"},
-    {file = "frozendict-2.4.4-cp37-cp37m-manylinux_2_17_x86_64.manylinux2014_x86_64.whl", hash = "sha256:85375ec6e979e6373bffb4f54576a68bf7497c350861d20686ccae38aab69c0a"},
-    {file = "frozendict-2.4.4-cp37-cp37m-musllinux_1_1_aarch64.whl", hash = "sha256:2d8536e068d6bf281f23fa835ac07747fb0f8851879dd189e9709f9567408b4d"},
-    {file = "frozendict-2.4.4-cp37-cp37m-musllinux_1_1_x86_64.whl", hash = "sha256:259528ba6b56fa051bc996f1c4d8b57e30d6dd3bc2f27441891b04babc4b5e73"},
-    {file = "frozendict-2.4.4-cp37-cp37m-win_amd64.whl", hash = "sha256:07c3a5dee8bbb84cba770e273cdbf2c87c8e035903af8f781292d72583416801"},
-    {file = "frozendict-2.4.4-cp38-cp38-macosx_11_0_arm64.whl", hash = "sha256:6874fec816b37b6eb5795b00e0574cba261bf59723e2de607a195d5edaff0786"},
-    {file = "frozendict-2.4.4-cp38-cp38-manylinux_2_17_aarch64.manylinux2014_aarch64.whl", hash = "sha256:c8f92425686323a950337da4b75b4c17a3327b831df8c881df24038d560640d4"},
-    {file = "frozendict-2.4.4-cp38-cp38-manylinux_2_17_x86_64.manylinux2014_x86_64.whl", hash = "sha256:5d58d9a8d9e49662c6dafbea5e641f97decdb3d6ccd76e55e79818415362ba25"},
-    {file = "frozendict-2.4.4-cp38-cp38-musllinux_1_1_aarch64.whl", hash = "sha256:93a7b19afb429cbf99d56faf436b45ef2fa8fe9aca89c49eb1610c3bd85f1760"},
-    {file = "frozendict-2.4.4-cp38-cp38-musllinux_1_1_x86_64.whl", hash = "sha256:2b70b431e3a72d410a2cdf1497b3aba2f553635e0c0f657ce311d841bf8273b6"},
-    {file = "frozendict-2.4.4-cp38-cp38-win_amd64.whl", hash = "sha256:e1b941132d79ce72d562a13341d38fc217bc1ee24d8c35a20d754e79ff99e038"},
-    {file = "frozendict-2.4.4-cp39-cp39-macosx_11_0_arm64.whl", hash = "sha256:dc2228874eacae390e63fd4f2bb513b3144066a977dc192163c9f6c7f6de6474"},
-    {file = "frozendict-2.4.4-cp39-cp39-manylinux_2_17_aarch64.manylinux2014_aarch64.whl", hash = "sha256:63aa49f1919af7d45fb8fd5dec4c0859bc09f46880bd6297c79bb2db2969b63d"},
-    {file = "frozendict-2.4.4-cp39-cp39-manylinux_2_17_x86_64.manylinux2014_x86_64.whl", hash = "sha256:c6bf9260018d653f3cab9bd147bd8592bf98a5c6e338be0491ced3c196c034a3"},
-    {file = "frozendict-2.4.4-cp39-cp39-musllinux_1_1_aarch64.whl", hash = "sha256:6eb716e6a6d693c03b1d53280a1947716129f5ef9bcdd061db5c17dea44b80fe"},
-    {file = "frozendict-2.4.4-cp39-cp39-musllinux_1_1_x86_64.whl", hash = "sha256:d13b4310db337f4d2103867c5a05090b22bc4d50ca842093779ef541ea9c9eea"},
-    {file = "frozendict-2.4.4-cp39-cp39-win_amd64.whl", hash = "sha256:b3b967d5065872e27b06f785a80c0ed0a45d1f7c9b85223da05358e734d858ca"},
-    {file = "frozendict-2.4.4-cp39-cp39-win_arm64.whl", hash = "sha256:4ae8d05c8d0b6134bfb6bfb369d5fa0c4df21eabb5ca7f645af95fdc6689678e"},
-    {file = "frozendict-2.4.4-py311-none-any.whl", hash = "sha256:705efca8d74d3facbb6ace80ab3afdd28eb8a237bfb4063ed89996b024bc443d"},
-    {file = "frozendict-2.4.4-py312-none-any.whl", hash = "sha256:d9647563e76adb05b7cde2172403123380871360a114f546b4ae1704510801e5"},
-    {file = "frozendict-2.4.4.tar.gz", hash = "sha256:3f7c031b26e4ee6a3f786ceb5e3abf1181c4ade92dce1f847da26ea2c96008c7"},
+    {file = "frozendict-2.4.5-cp310-cp310-macosx_10_9_x86_64.whl", hash = "sha256:1f364cfe5ef97523a4434e9f458bb4821594d3531d898621e5acae43463dcb5e"},
+    {file = "frozendict-2.4.5-cp310-cp310-macosx_11_0_arm64.whl", hash = "sha256:4f073c926b1f88fa85ed85222101f61f6c4b2180c95d1528ca6ecc7cef835442"},
+    {file = "frozendict-2.4.5-cp310-cp310-manylinux_2_17_aarch64.manylinux2014_aarch64.whl", hash = "sha256:8582f26ca862bb1e40ed790d934adf6afcfc904b0425dcfe01aed2103bed27fb"},
+    {file = "frozendict-2.4.5-cp310-cp310-manylinux_2_17_x86_64.manylinux2014_x86_64.whl", hash = "sha256:4b3eea1678607174c468fe4e3b903625bccfb3215ae2b0138220dc1f6ef71f37"},
+    {file = "frozendict-2.4.5-cp310-cp310-musllinux_1_2_aarch64.whl", hash = "sha256:3af1b09bad761d5500b096b757c346591b5dfdc8443aa9642c2c02b4885dc09e"},
+    {file = "frozendict-2.4.5-cp310-cp310-musllinux_1_2_x86_64.whl", hash = "sha256:76c7af5f9db9ae01531edaf38fd3e8faae1bb6b94abbf8fa5bd0326f52e777df"},
+    {file = "frozendict-2.4.5-cp310-cp310-win_amd64.whl", hash = "sha256:a5448639058157ccd7f26ba83e441066c6ae515beb68b9b99d5b2dbb0bb36b19"},
+    {file = "frozendict-2.4.5-cp310-cp310-win_arm64.whl", hash = "sha256:24953a1cfe344415e7557413e493e21fa9c4cecbc13284b6f334837aa5601c9f"},
+    {file = "frozendict-2.4.5-cp36-cp36m-macosx_10_9_x86_64.whl", hash = "sha256:754924d53b1fd2dae7b15f9c86b0610334a840ae728cc717f24aa040fc94136f"},
+    {file = "frozendict-2.4.5-cp36-cp36m-manylinux_2_17_aarch64.manylinux2014_aarch64.whl", hash = "sha256:95bf788d65e3a50d9deef5c68e568c59acf473fea8a7a4d7e405a7e63ac85cdd"},
+    {file = "frozendict-2.4.5-cp36-cp36m-manylinux_2_17_x86_64.manylinux2014_x86_64.whl", hash = "sha256:35730ff269b8a6eb07c3d91d5fd6a7f5c32e08bef665fe1ef51012bdb2702196"},
+    {file = "frozendict-2.4.5-cp36-cp36m-musllinux_1_2_aarch64.whl", hash = "sha256:eb58e0c41bc6bb60a8b0852d50da8836380fb78311608dc26791e3e9aa25bcfe"},
+    {file = "frozendict-2.4.5-cp36-cp36m-musllinux_1_2_x86_64.whl", hash = "sha256:0483776fa25f0c3f9ff75a89b5276a582231f83baea85b091dfefde99b95ac5c"},
+    {file = "frozendict-2.4.5-cp36-cp36m-win_amd64.whl", hash = "sha256:a967730a115cb8b5d028e0d07cd680c04324e913ee3d1caef9e039ca740343c4"},
+    {file = "frozendict-2.4.5-cp37-cp37m-macosx_10_9_x86_64.whl", hash = "sha256:6d10ad91ded20fae8737fddcfed1b6358207f95292a335f96c4cd365d97f5e30"},
+    {file = "frozendict-2.4.5-cp37-cp37m-manylinux_2_17_aarch64.manylinux2014_aarch64.whl", hash = "sha256:3702438d81deab127963ab41dca1738a0fbf11038a50a25c5f814943a2d25de5"},
+    {file = "frozendict-2.4.5-cp37-cp37m-manylinux_2_17_x86_64.manylinux2014_x86_64.whl", hash = "sha256:de1126d6dd39d7939be388161b973ffadb9c6e97f8f9fdcb52ba790828621191"},
+    {file = "frozendict-2.4.5-cp37-cp37m-musllinux_1_2_aarch64.whl", hash = "sha256:f8d677a6066ca3289181028fccabf77413addddaa7a6629485d52aeaa4f4899c"},
+    {file = "frozendict-2.4.5-cp37-cp37m-musllinux_1_2_x86_64.whl", hash = "sha256:f2ff03ed21657cde7ea04fffc76f0c1736100a03671e4682722685e63486cd96"},
+    {file = "frozendict-2.4.5-cp37-cp37m-win_amd64.whl", hash = "sha256:14755127b988b428f95f11fa626374e247f8a40316697218f1e2f44c107c5027"},
+    {file = "frozendict-2.4.5-cp38-cp38-macosx_10_9_x86_64.whl", hash = "sha256:bbf9dcb69a44519308c775bcf72b8417c3102f415f77df19b8eddb97b8f88d78"},
+    {file = "frozendict-2.4.5-cp38-cp38-macosx_11_0_arm64.whl", hash = "sha256:ee3e6f4013446d17c580ce0bac41a1f81371c21dd6e34a85c0f26bae94fdfd9b"},
+    {file = "frozendict-2.4.5-cp38-cp38-manylinux_2_17_aarch64.manylinux2014_aarch64.whl", hash = "sha256:2b51e0aae859fc8d56138eac4be121a76647fa66ea620af8a62d9d6938729441"},
+    {file = "frozendict-2.4.5-cp38-cp38-manylinux_2_17_x86_64.manylinux2014_x86_64.whl", hash = "sha256:af4bdffecff350a68f1a966dec84459d3ea8e2606c14fb3ebe937b8c6c9e9129"},
+    {file = "frozendict-2.4.5-cp38-cp38-musllinux_1_2_aarch64.whl", hash = "sha256:f1385852cb05a31ae76c972caa3a07679414b01a664848bde2bae71c9910f045"},
+    {file = "frozendict-2.4.5-cp38-cp38-musllinux_1_2_x86_64.whl", hash = "sha256:e591124818f2d5d8f2d284a7b78604f4180e73d770c33252370edccff5969293"},
+    {file = "frozendict-2.4.5-cp38-cp38-win_amd64.whl", hash = "sha256:05786d9cff61ee95149c15cdfc10cf4b4cb10e1eab1d4648b0c99ed58fedb86d"},
+    {file = "frozendict-2.4.5-cp39-cp39-macosx_10_9_x86_64.whl", hash = "sha256:98b1483980c47bb74ef777ab748031a69eb41c43ef44494e9db79111638fdb84"},
+    {file = "frozendict-2.4.5-cp39-cp39-macosx_11_0_arm64.whl", hash = "sha256:08e8d4ebc950916b5cdeedcb697cf893c5903d9cb8ab27dffe41072a08e1cfc1"},
+    {file = "frozendict-2.4.5-cp39-cp39-manylinux_2_17_aarch64.manylinux2014_aarch64.whl", hash = "sha256:73ff80ce3f95c3ec60b38ef4ddf5cb8e20ea3edce2f56e89af7c4c45013684bf"},
+    {file = "frozendict-2.4.5-cp39-cp39-manylinux_2_17_x86_64.manylinux2014_x86_64.whl", hash = "sha256:dee64c44e1a146171bdd2bf4501210f273a5ab73cc2e924a01f81b14e4612f94"},
+    {file = "frozendict-2.4.5-cp39-cp39-musllinux_1_2_aarch64.whl", hash = "sha256:a81d80ded07f0e34a3fa2b78235a48259b886ad20a96da5526a9404d192c6eb3"},
+    {file = "frozendict-2.4.5-cp39-cp39-musllinux_1_2_x86_64.whl", hash = "sha256:043bce9f5e7e8df8313d0c961b2da3346c0b002311d88c74cadae2b1f8fd2904"},
+    {file = "frozendict-2.4.5-cp39-cp39-win_amd64.whl", hash = "sha256:d2b2a9c809ebb54c8c0fe2acc38d45299f3e9a9d90061773468896221cab1eaa"},
+    {file = "frozendict-2.4.5-cp39-cp39-win_arm64.whl", hash = "sha256:20cf1db30ca0c8f76a05ec1256132f505cf6f93eb382a5961c738377037b5b9d"},
+    {file = "frozendict-2.4.5-py311-none-any.whl", hash = "sha256:6be054ba76e8a49c6846a7369db3eaaa0593c29a644026c25923f13fdea06483"},
+    {file = "frozendict-2.4.5-py312-none-any.whl", hash = "sha256:b8481d83a7219e9e14c719113ea2d8321d7840af35af58c72b3864b7123e7de3"},
+    {file = "frozendict-2.4.5.tar.gz", hash = "sha256:fd7add309789595c044c0155a0bddfa9d20c77f65de1e33a14aa3033b936ef63"},
 ]
 
 [[package]]
@@ -2950,6 +2972,17 @@ test-downstream = ["aiobotocore (>=2.5.4,<3.0.0)", "dask-expr", "dask[dataframe,
 test-full = ["adlfs", "aiohttp (!=4.0.0a0,!=4.0.0a1)", "cloudpickle", "dask", "distributed", "dropbox", "dropboxdrivefs", "fastparquet", "fusepy", "gcsfs", "jinja2", "kerchunk", "libarchive-c", "lz4", "notebook", "numpy", "ocifs", "pandas", "panel", "paramiko", "pyarrow", "pyarrow (>=1)", "pyftpdlib", "pygit2", "pytest", "pytest-asyncio (!=0.22.0)", "pytest-benchmark", "pytest-cov", "pytest-mock", "pytest-recording", "pytest-rerunfailures", "python-snappy", "requests", "smbprotocol", "tqdm", "urllib3", "zarr", "zstandard"]
 tqdm = ["tqdm"]
 
+[[package]]
+name = "future"
+version = "1.0.0"
+description = "Clean single-source support for Python 3 and 2"
+optional = false
+python-versions = ">=2.6, !=3.0.*, !=3.1.*, !=3.2.*"
+files = [
+    {file = "future-1.0.0-py3-none-any.whl", hash = "sha256:929292d34f5872e70396626ef385ec22355a1fae8ad29e1a734c3e43f9fbc216"},
+    {file = "future-1.0.0.tar.gz", hash = "sha256:bd2968309307861edae1458a4f8a4f3598c03be43b97521076aebf5d94c07b05"},
+]
+
 [[package]]
 name = "gevent"
 version = "23.9.1"
@@ -3861,13 +3894,13 @@ lxml = ["lxml"]
 
 [[package]]
 name = "httpcore"
-version = "1.0.5"
+version = "1.0.6"
 description = "A minimal low-level HTTP client."
 optional = false
 python-versions = ">=3.8"
 files = [
-    {file = "httpcore-1.0.5-py3-none-any.whl", hash = "sha256:421f18bac248b25d310f3cacd198d55b8e6125c107797b609ff9b7a6ba7991b5"},
-    {file = "httpcore-1.0.5.tar.gz", hash = "sha256:34a38e2f9291467ee3b44e89dd52615370e152954ba21721378a87b2960f7a61"},
+    {file = "httpcore-1.0.6-py3-none-any.whl", hash = "sha256:27b59625743b85577a8c0e10e55b50b5368a4f2cfe8cc7bcfa9cf00829c2682f"},
+    {file = "httpcore-1.0.6.tar.gz", hash = "sha256:73f6dbd6eb8c21bbf7ef8efad555481853f5f6acdeaff1edb0694289269ee17f"},
 ]
 
 [package.dependencies]
@@ -3878,7 +3911,7 @@ h11 = ">=0.13,<0.15"
 asyncio = ["anyio (>=4.0,<5.0)"]
 http2 = ["h2 (>=3,<5)"]
 socks = ["socksio (==1.*)"]
-trio = ["trio (>=0.22.0,<0.26.0)"]
+trio = ["trio (>=0.22.0,<1.0)"]
 
 [[package]]
 name = "httplib2"
@@ -4450,13 +4483,13 @@ six = "*"
 
 [[package]]
 name = "langfuse"
-version = "2.51.2"
+version = "2.51.3"
 description = "A client library for accessing langfuse"
 optional = false
-python-versions = "<4.0,>=3.8.1"
+python-versions = ">=3.8.1,<4.0"
 files = [
-    {file = "langfuse-2.51.2-py3-none-any.whl", hash = "sha256:7aab94a9452cda4587a2cd4917e455da1afd7f8a2696688742130e2f2d23ca59"},
-    {file = "langfuse-2.51.2.tar.gz", hash = "sha256:0982b108ab4c02947f682e442b0796b7a73825d31eeace1771575f6454b8f79a"},
+    {file = "langfuse-2.51.3-py3-none-any.whl", hash = "sha256:32aba050123656ec0c165583e1d33243cc7a14e7b8498ed3c9de808aa90306f1"},
+    {file = "langfuse-2.51.3.tar.gz", hash = "sha256:ccd2109556ee232db717abfb751ee8a3139f074db7d3e06a4f1a756c349fc5ba"},
 ]
 
 [package.dependencies]
@@ -4475,13 +4508,13 @@ openai = ["openai (>=0.27.8)"]
 
 [[package]]
 name = "langsmith"
-version = "0.1.129"
+version = "0.1.131"
 description = "Client library to connect to the LangSmith LLM Tracing and Evaluation Platform."
 optional = false
-python-versions = "<4.0,>=3.8.1"
+python-versions = ">=3.8.1,<4.0"
 files = [
-    {file = "langsmith-0.1.129-py3-none-any.whl", hash = "sha256:31393fbbb17d6be5b99b9b22d530450094fab23c6c37281a6a6efb2143d05347"},
-    {file = "langsmith-0.1.129.tar.gz", hash = "sha256:6c3ba66471bef41b9f87da247cc0b493268b3f54656f73648a256a205261b6a0"},
+    {file = "langsmith-0.1.131-py3-none-any.whl", hash = "sha256:80c106b1c42307195cc0bb3a596472c41ef91b79d15bcee9938307800336c563"},
+    {file = "langsmith-0.1.131.tar.gz", hash = "sha256:626101a3bf3ca481e5110d5155ace8aa066e4e9cc2fa7d96c8290ade0fbff797"},
 ]
 
 [package.dependencies]
@@ -4492,6 +4525,7 @@ pydantic = [
     {version = ">=2.7.4,<3.0.0", markers = "python_full_version >= \"3.12.4\""},
 ]
 requests = ">=2,<3"
+requests-toolbelt = ">=1.0.0,<2.0.0"
 
 [[package]]
 name = "llvmlite"
@@ -5318,27 +5352,31 @@ typing-extensions = {version = ">=4.1.0", markers = "python_version < \"3.11\""}
 
 [[package]]
 name = "multiprocess"
-version = "0.70.16"
+version = "0.70.17"
 description = "better multiprocessing and multithreading in Python"
 optional = false
 python-versions = ">=3.8"
 files = [
-    {file = "multiprocess-0.70.16-pp310-pypy310_pp73-macosx_10_13_x86_64.whl", hash = "sha256:476887be10e2f59ff183c006af746cb6f1fd0eadcfd4ef49e605cbe2659920ee"},
-    {file = "multiprocess-0.70.16-pp310-pypy310_pp73-manylinux_2_28_x86_64.whl", hash = "sha256:d951bed82c8f73929ac82c61f01a7b5ce8f3e5ef40f5b52553b4f547ce2b08ec"},
-    {file = "multiprocess-0.70.16-pp38-pypy38_pp73-macosx_10_9_x86_64.whl", hash = "sha256:37b55f71c07e2d741374998c043b9520b626a8dddc8b3129222ca4f1a06ef67a"},
-    {file = "multiprocess-0.70.16-pp38-pypy38_pp73-manylinux_2_28_x86_64.whl", hash = "sha256:ba8c31889abf4511c7308a8c52bb4a30b9d590e7f58523302ba00237702ca054"},
-    {file = "multiprocess-0.70.16-pp39-pypy39_pp73-macosx_10_13_x86_64.whl", hash = "sha256:0dfd078c306e08d46d7a8d06fb120313d87aa43af60d66da43ffff40b44d2f41"},
-    {file = "multiprocess-0.70.16-pp39-pypy39_pp73-manylinux_2_28_x86_64.whl", hash = "sha256:e7b9d0f307cd9bd50851afaac0dba2cb6c44449efff697df7c7645f7d3f2be3a"},
-    {file = "multiprocess-0.70.16-py310-none-any.whl", hash = "sha256:c4a9944c67bd49f823687463660a2d6daae94c289adff97e0f9d696ba6371d02"},
-    {file = "multiprocess-0.70.16-py311-none-any.whl", hash = "sha256:af4cabb0dac72abfb1e794fa7855c325fd2b55a10a44628a3c1ad3311c04127a"},
-    {file = "multiprocess-0.70.16-py312-none-any.whl", hash = "sha256:fc0544c531920dde3b00c29863377f87e1632601092ea2daca74e4beb40faa2e"},
-    {file = "multiprocess-0.70.16-py38-none-any.whl", hash = "sha256:a71d82033454891091a226dfc319d0cfa8019a4e888ef9ca910372a446de4435"},
-    {file = "multiprocess-0.70.16-py39-none-any.whl", hash = "sha256:a0bafd3ae1b732eac64be2e72038231c1ba97724b60b09400d68f229fcc2fbf3"},
-    {file = "multiprocess-0.70.16.tar.gz", hash = "sha256:161af703d4652a0e1410be6abccecde4a7ddffd19341be0a7011b94aeb171ac1"},
+    {file = "multiprocess-0.70.17-pp310-pypy310_pp73-macosx_10_15_x86_64.whl", hash = "sha256:7ddb24e5bcdb64e90ec5543a1f05a39463068b6d3b804aa3f2a4e16ec28562d6"},
+    {file = "multiprocess-0.70.17-pp310-pypy310_pp73-macosx_11_0_arm64.whl", hash = "sha256:d729f55198a3579f6879766a6d9b72b42d4b320c0dcb7844afb774d75b573c62"},
+    {file = "multiprocess-0.70.17-pp310-pypy310_pp73-manylinux_2_28_x86_64.whl", hash = "sha256:c2c82d0375baed8d8dd0d8c38eb87c5ae9c471f8e384ad203a36f095ee860f67"},
+    {file = "multiprocess-0.70.17-pp38-pypy38_pp73-macosx_10_9_arm64.whl", hash = "sha256:a22a6b1a482b80eab53078418bb0f7025e4f7d93cc8e1f36481477a023884861"},
+    {file = "multiprocess-0.70.17-pp38-pypy38_pp73-macosx_10_9_x86_64.whl", hash = "sha256:349525099a0c9ac5936f0488b5ee73199098dac3ac899d81d326d238f9fd3ccd"},
+    {file = "multiprocess-0.70.17-pp38-pypy38_pp73-manylinux_2_28_x86_64.whl", hash = "sha256:27b8409c02b5dd89d336107c101dfbd1530a2cd4fd425fc27dcb7adb6e0b47bf"},
+    {file = "multiprocess-0.70.17-pp39-pypy39_pp73-macosx_10_13_arm64.whl", hash = "sha256:2ea0939b0f4760a16a548942c65c76ff5afd81fbf1083c56ae75e21faf92e426"},
+    {file = "multiprocess-0.70.17-pp39-pypy39_pp73-macosx_10_13_x86_64.whl", hash = "sha256:2b12e081df87ab755190e227341b2c3b17ee6587e9c82fecddcbe6aa812cd7f7"},
+    {file = "multiprocess-0.70.17-pp39-pypy39_pp73-manylinux_2_28_x86_64.whl", hash = "sha256:a0f01cd9d079af7a8296f521dc03859d1a414d14c1e2b6e676ef789333421c95"},
+    {file = "multiprocess-0.70.17-py310-none-any.whl", hash = "sha256:38357ca266b51a2e22841b755d9a91e4bb7b937979a54d411677111716c32744"},
+    {file = "multiprocess-0.70.17-py311-none-any.whl", hash = "sha256:2884701445d0177aec5bd5f6ee0df296773e4fb65b11903b94c613fb46cfb7d1"},
+    {file = "multiprocess-0.70.17-py312-none-any.whl", hash = "sha256:2818af14c52446b9617d1b0755fa70ca2f77c28b25ed97bdaa2c69a22c47b46c"},
+    {file = "multiprocess-0.70.17-py313-none-any.whl", hash = "sha256:20c28ca19079a6c879258103a6d60b94d4ffe2d9da07dda93fb1c8bc6243f522"},
+    {file = "multiprocess-0.70.17-py38-none-any.whl", hash = "sha256:1d52f068357acd1e5bbc670b273ef8f81d57863235d9fbf9314751886e141968"},
+    {file = "multiprocess-0.70.17-py39-none-any.whl", hash = "sha256:c3feb874ba574fbccfb335980020c1ac631fbf2a3f7bee4e2042ede62558a021"},
+    {file = "multiprocess-0.70.17.tar.gz", hash = "sha256:4ae2f11a3416809ebc9a48abfc8b14ecce0652a0944731a1493a3c1ba44ff57a"},
 ]
 
 [package.dependencies]
-dill = ">=0.3.8"
+dill = ">=0.3.9"
 
 [[package]]
 name = "multitasking"
@@ -5597,13 +5635,13 @@ signedtoken = ["cryptography (>=3.0.0)", "pyjwt (>=2.0.0,<3)"]
 
 [[package]]
 name = "oci"
-version = "2.135.0"
+version = "2.135.1"
 description = "Oracle Cloud Infrastructure Python SDK"
 optional = false
 python-versions = "*"
 files = [
-    {file = "oci-2.135.0-py3-none-any.whl", hash = "sha256:c01f1d103ed034fa7ca2bceb297bf00e6f6c456d14a46b35ee9007b25f3ea397"},
-    {file = "oci-2.135.0.tar.gz", hash = "sha256:6e28e6595264705d8fd0719045ffc4b23170e7fd2cd76a1c3aa25e4cdaa5883a"},
+    {file = "oci-2.135.1-py3-none-any.whl", hash = "sha256:249e24638faa2058b2fa9ab4917ff230f2a8b3b97cd11a261ce6c7f97c3ed2ff"},
+    {file = "oci-2.135.1.tar.gz", hash = "sha256:bea8aaba19d3fed186ceaeedba6d46f1cdb3c6c7d93b39ab225a15cbd63f9ce8"},
 ]
 
 [package.dependencies]
@@ -6212,20 +6250,20 @@ xml = ["lxml (>=4.9.2)"]
 
 [[package]]
 name = "pathos"
-version = "0.3.2"
+version = "0.3.3"
 description = "parallel graph management and execution in heterogeneous computing"
 optional = false
 python-versions = ">=3.8"
 files = [
-    {file = "pathos-0.3.2-py3-none-any.whl", hash = "sha256:d669275e6eb4b3fbcd2846d7a6d1bba315fe23add0c614445ba1408d8b38bafe"},
-    {file = "pathos-0.3.2.tar.gz", hash = "sha256:4f2a42bc1e10ccf0fe71961e7145fc1437018b6b21bd93b2446abc3983e49a7a"},
+    {file = "pathos-0.3.3-py3-none-any.whl", hash = "sha256:e04616c6448608ad1f809360be22e3f2078d949a36a81e6991da6c2dd1f82513"},
+    {file = "pathos-0.3.3.tar.gz", hash = "sha256:dcb2a5f321aa34ca541c1c1861011ea49df357bb908379c21dd5741f666e0a58"},
 ]
 
 [package.dependencies]
-dill = ">=0.3.8"
-multiprocess = ">=0.70.16"
-pox = ">=0.3.4"
-ppft = ">=1.7.6.8"
+dill = ">=0.3.9"
+multiprocess = ">=0.70.17"
+pox = ">=0.3.5"
+ppft = ">=1.7.6.9"
 
 [[package]]
 name = "peewee"
@@ -6447,13 +6485,13 @@ tests = ["pytest (>=5.4.1)", "pytest-cov (>=2.8.1)", "pytest-mypy (>=0.8.0)", "p
 
 [[package]]
 name = "posthog"
-version = "3.6.6"
+version = "3.7.0"
 description = "Integrate PostHog into any python application."
 optional = false
 python-versions = "*"
 files = [
-    {file = "posthog-3.6.6-py2.py3-none-any.whl", hash = "sha256:38834fd7f0732582a20d4eb4674c8d5c088e464d14d1b3f8c176e389aecaa4ef"},
-    {file = "posthog-3.6.6.tar.gz", hash = "sha256:1e04783293117109189ad7048f3eedbe21caff0e39bee5e2d47a93dd790fefac"},
+    {file = "posthog-3.7.0-py2.py3-none-any.whl", hash = "sha256:3555161c3a9557b5666f96d8e1f17f410ea0f07db56e399e336a1656d4e5c722"},
+    {file = "posthog-3.7.0.tar.gz", hash = "sha256:b095d4354ba23f8b346ab5daed8ecfc5108772f922006982dfe8b2d29ebc6e0e"},
 ]
 
 [package.dependencies]
@@ -6715,6 +6753,19 @@ files = [
     {file = "pyarrow-17.0.0-cp312-cp312-win_amd64.whl", hash = "sha256:392bc9feabc647338e6c89267635e111d71edad5fcffba204425a7c8d13610d7"},
     {file = "pyarrow-17.0.0-cp38-cp38-macosx_10_15_x86_64.whl", hash = "sha256:af5ff82a04b2171415f1410cff7ebb79861afc5dae50be73ce06d6e870615204"},
     {file = "pyarrow-17.0.0-cp38-cp38-macosx_11_0_arm64.whl", hash = "sha256:edca18eaca89cd6382dfbcff3dd2d87633433043650c07375d095cd3517561d8"},
+    {file = "pyarrow-17.0.0-cp38-cp38-manylinux_2_17_aarch64.manylinux2014_aarch64.whl", hash = "sha256:7c7916bff914ac5d4a8fe25b7a25e432ff921e72f6f2b7547d1e325c1ad9d155"},
+    {file = "pyarrow-17.0.0-cp38-cp38-manylinux_2_17_x86_64.manylinux2014_x86_64.whl", hash = "sha256:f553ca691b9e94b202ff741bdd40f6ccb70cdd5fbf65c187af132f1317de6145"},
+    {file = "pyarrow-17.0.0-cp38-cp38-manylinux_2_28_aarch64.whl", hash = "sha256:0cdb0e627c86c373205a2f94a510ac4376fdc523f8bb36beab2e7f204416163c"},
+    {file = "pyarrow-17.0.0-cp38-cp38-manylinux_2_28_x86_64.whl", hash = "sha256:d7d192305d9d8bc9082d10f361fc70a73590a4c65cf31c3e6926cd72b76bc35c"},
+    {file = "pyarrow-17.0.0-cp38-cp38-win_amd64.whl", hash = "sha256:02dae06ce212d8b3244dd3e7d12d9c4d3046945a5933d28026598e9dbbda1fca"},
+    {file = "pyarrow-17.0.0-cp39-cp39-macosx_10_15_x86_64.whl", hash = "sha256:13d7a460b412f31e4c0efa1148e1d29bdf18ad1411eb6757d38f8fbdcc8645fb"},
+    {file = "pyarrow-17.0.0-cp39-cp39-macosx_11_0_arm64.whl", hash = "sha256:9b564a51fbccfab5a04a80453e5ac6c9954a9c5ef2890d1bcf63741909c3f8df"},
+    {file = "pyarrow-17.0.0-cp39-cp39-manylinux_2_17_aarch64.manylinux2014_aarch64.whl", hash = "sha256:32503827abbc5aadedfa235f5ece8c4f8f8b0a3cf01066bc8d29de7539532687"},
+    {file = "pyarrow-17.0.0-cp39-cp39-manylinux_2_17_x86_64.manylinux2014_x86_64.whl", hash = "sha256:a155acc7f154b9ffcc85497509bcd0d43efb80d6f733b0dc3bb14e281f131c8b"},
+    {file = "pyarrow-17.0.0-cp39-cp39-manylinux_2_28_aarch64.whl", hash = "sha256:dec8d129254d0188a49f8a1fc99e0560dc1b85f60af729f47de4046015f9b0a5"},
+    {file = "pyarrow-17.0.0-cp39-cp39-manylinux_2_28_x86_64.whl", hash = "sha256:a48ddf5c3c6a6c505904545c25a4ae13646ae1f8ba703c4df4a1bfe4f4006bda"},
+    {file = "pyarrow-17.0.0-cp39-cp39-win_amd64.whl", hash = "sha256:42bf93249a083aca230ba7e2786c5f673507fa97bbd9725a1e2754715151a204"},
+    {file = "pyarrow-17.0.0.tar.gz", hash = "sha256:4beca9521ed2c0921c1023e68d097d0299b62c362639ea315572a58f3f50fd28"},
 ]
 
 [package.dependencies]
@@ -7446,25 +7497,29 @@ files = [
 
 [[package]]
 name = "pywin32"
-version = "306"
+version = "307"
 description = "Python for Window Extensions"
 optional = false
 python-versions = "*"
 files = [
-    {file = "pywin32-306-cp310-cp310-win32.whl", hash = "sha256:06d3420a5155ba65f0b72f2699b5bacf3109f36acbe8923765c22938a69dfc8d"},
-    {file = "pywin32-306-cp310-cp310-win_amd64.whl", hash = "sha256:84f4471dbca1887ea3803d8848a1616429ac94a4a8d05f4bc9c5dcfd42ca99c8"},
-    {file = "pywin32-306-cp311-cp311-win32.whl", hash = "sha256:e65028133d15b64d2ed8f06dd9fbc268352478d4f9289e69c190ecd6818b6407"},
-    {file = "pywin32-306-cp311-cp311-win_amd64.whl", hash = "sha256:a7639f51c184c0272e93f244eb24dafca9b1855707d94c192d4a0b4c01e1100e"},
-    {file = "pywin32-306-cp311-cp311-win_arm64.whl", hash = "sha256:70dba0c913d19f942a2db25217d9a1b726c278f483a919f1abfed79c9cf64d3a"},
-    {file = "pywin32-306-cp312-cp312-win32.whl", hash = "sha256:383229d515657f4e3ed1343da8be101000562bf514591ff383ae940cad65458b"},
-    {file = "pywin32-306-cp312-cp312-win_amd64.whl", hash = "sha256:37257794c1ad39ee9be652da0462dc2e394c8159dfd913a8a4e8eb6fd346da0e"},
-    {file = "pywin32-306-cp312-cp312-win_arm64.whl", hash = "sha256:5821ec52f6d321aa59e2db7e0a35b997de60c201943557d108af9d4ae1ec7040"},
-    {file = "pywin32-306-cp37-cp37m-win32.whl", hash = "sha256:1c73ea9a0d2283d889001998059f5eaaba3b6238f767c9cf2833b13e6a685f65"},
-    {file = "pywin32-306-cp37-cp37m-win_amd64.whl", hash = "sha256:72c5f621542d7bdd4fdb716227be0dd3f8565c11b280be6315b06ace35487d36"},
-    {file = "pywin32-306-cp38-cp38-win32.whl", hash = "sha256:e4c092e2589b5cf0d365849e73e02c391c1349958c5ac3e9d5ccb9a28e017b3a"},
-    {file = "pywin32-306-cp38-cp38-win_amd64.whl", hash = "sha256:e8ac1ae3601bee6ca9f7cb4b5363bf1c0badb935ef243c4733ff9a393b1690c0"},
-    {file = "pywin32-306-cp39-cp39-win32.whl", hash = "sha256:e25fd5b485b55ac9c057f67d94bc203f3f6595078d1fb3b458c9c28b7153a802"},
-    {file = "pywin32-306-cp39-cp39-win_amd64.whl", hash = "sha256:39b61c15272833b5c329a2989999dcae836b1eed650252ab1b7bfbe1d59f30f4"},
+    {file = "pywin32-307-cp310-cp310-win32.whl", hash = "sha256:f8f25d893c1e1ce2d685ef6d0a481e87c6f510d0f3f117932781f412e0eba31b"},
+    {file = "pywin32-307-cp310-cp310-win_amd64.whl", hash = "sha256:36e650c5e5e6b29b5d317385b02d20803ddbac5d1031e1f88d20d76676dd103d"},
+    {file = "pywin32-307-cp310-cp310-win_arm64.whl", hash = "sha256:0c12d61e0274e0c62acee79e3e503c312426ddd0e8d4899c626cddc1cafe0ff4"},
+    {file = "pywin32-307-cp311-cp311-win32.whl", hash = "sha256:fec5d27cc893178fab299de911b8e4d12c5954e1baf83e8a664311e56a272b75"},
+    {file = "pywin32-307-cp311-cp311-win_amd64.whl", hash = "sha256:987a86971753ed7fdd52a7fb5747aba955b2c7fbbc3d8b76ec850358c1cc28c3"},
+    {file = "pywin32-307-cp311-cp311-win_arm64.whl", hash = "sha256:fd436897c186a2e693cd0437386ed79f989f4d13d6f353f8787ecbb0ae719398"},
+    {file = "pywin32-307-cp312-cp312-win32.whl", hash = "sha256:07649ec6b01712f36debf39fc94f3d696a46579e852f60157a729ac039df0815"},
+    {file = "pywin32-307-cp312-cp312-win_amd64.whl", hash = "sha256:00d047992bb5dcf79f8b9b7c81f72e0130f9fe4b22df613f755ab1cc021d8347"},
+    {file = "pywin32-307-cp312-cp312-win_arm64.whl", hash = "sha256:b53658acbfc6a8241d72cc09e9d1d666be4e6c99376bc59e26cdb6223c4554d2"},
+    {file = "pywin32-307-cp313-cp313-win32.whl", hash = "sha256:ea4d56e48dc1ab2aa0a5e3c0741ad6e926529510516db7a3b6981a1ae74405e5"},
+    {file = "pywin32-307-cp313-cp313-win_amd64.whl", hash = "sha256:576d09813eaf4c8168d0bfd66fb7cb3b15a61041cf41598c2db4a4583bf832d2"},
+    {file = "pywin32-307-cp313-cp313-win_arm64.whl", hash = "sha256:b30c9bdbffda6a260beb2919f918daced23d32c79109412c2085cbc513338a0a"},
+    {file = "pywin32-307-cp37-cp37m-win32.whl", hash = "sha256:5101472f5180c647d4525a0ed289ec723a26231550dbfd369ec19d5faf60e511"},
+    {file = "pywin32-307-cp37-cp37m-win_amd64.whl", hash = "sha256:05de55a7c110478dc4b202230e98af5e0720855360d2b31a44bb4e296d795fba"},
+    {file = "pywin32-307-cp38-cp38-win32.whl", hash = "sha256:13d059fb7f10792542082f5731d5d3d9645320fc38814759313e5ee97c3fac01"},
+    {file = "pywin32-307-cp38-cp38-win_amd64.whl", hash = "sha256:7e0b2f93769d450a98ac7a31a087e07b126b6d571e8b4386a5762eb85325270b"},
+    {file = "pywin32-307-cp39-cp39-win32.whl", hash = "sha256:55ee87f2f8c294e72ad9d4261ca423022310a6e79fb314a8ca76ab3f493854c6"},
+    {file = "pywin32-307-cp39-cp39-win_amd64.whl", hash = "sha256:e9d5202922e74985b037c9ef46778335c102b74b95cec70f629453dbe7235d87"},
 ]
 
 [[package]]
@@ -7968,18 +8023,19 @@ requests = "2.31.0"
 
 [[package]]
 name = "rich"
-version = "13.8.1"
+version = "13.9.2"
 description = "Render rich text, tables, progress bars, syntax highlighting, markdown and more to the terminal"
 optional = false
-python-versions = ">=3.7.0"
+python-versions = ">=3.8.0"
 files = [
-    {file = "rich-13.8.1-py3-none-any.whl", hash = "sha256:1760a3c0848469b97b558fc61c85233e3dafb69c7a071b4d60c38099d3cd4c06"},
-    {file = "rich-13.8.1.tar.gz", hash = "sha256:8260cda28e3db6bf04d2d1ef4dbc03ba80a824c88b0e7668a0f23126a424844a"},
+    {file = "rich-13.9.2-py3-none-any.whl", hash = "sha256:8c82a3d3f8dcfe9e734771313e606b39d8247bb6b826e196f4914b333b743cf1"},
+    {file = "rich-13.9.2.tar.gz", hash = "sha256:51a2c62057461aaf7152b4d611168f93a9fc73068f8ded2790f29fe2b5366d0c"},
 ]
 
 [package.dependencies]
 markdown-it-py = ">=2.2.0"
 pygments = ">=2.13.0,<3.0.0"
+typing-extensions = {version = ">=4.0.0,<5.0", markers = "python_version < \"3.11\""}
 
 [package.extras]
 jupyter = ["ipywidgets (>=7.5.1,<9)"]
@@ -8112,29 +8168,29 @@ pyasn1 = ">=0.1.3"
 
 [[package]]
 name = "ruff"
-version = "0.6.8"
+version = "0.6.9"
 description = "An extremely fast Python linter and code formatter, written in Rust."
 optional = false
 python-versions = ">=3.7"
 files = [
-    {file = "ruff-0.6.8-py3-none-linux_armv6l.whl", hash = "sha256:77944bca110ff0a43b768f05a529fecd0706aac7bcce36d7f1eeb4cbfca5f0f2"},
-    {file = "ruff-0.6.8-py3-none-macosx_10_12_x86_64.whl", hash = "sha256:27b87e1801e786cd6ede4ada3faa5e254ce774de835e6723fd94551464c56b8c"},
-    {file = "ruff-0.6.8-py3-none-macosx_11_0_arm64.whl", hash = "sha256:cd48f945da2a6334f1793d7f701725a76ba93bf3d73c36f6b21fb04d5338dcf5"},
-    {file = "ruff-0.6.8-py3-none-manylinux_2_17_aarch64.manylinux2014_aarch64.whl", hash = "sha256:677e03c00f37c66cea033274295a983c7c546edea5043d0c798833adf4cf4c6f"},
-    {file = "ruff-0.6.8-py3-none-manylinux_2_17_armv7l.manylinux2014_armv7l.whl", hash = "sha256:9f1476236b3eacfacfc0f66aa9e6cd39f2a624cb73ea99189556015f27c0bdeb"},
-    {file = "ruff-0.6.8-py3-none-manylinux_2_17_i686.manylinux2014_i686.whl", hash = "sha256:6f5a2f17c7d32991169195d52a04c95b256378bbf0de8cb98478351eb70d526f"},
-    {file = "ruff-0.6.8-py3-none-manylinux_2_17_ppc64.manylinux2014_ppc64.whl", hash = "sha256:5fd0d4b7b1457c49e435ee1e437900ced9b35cb8dc5178921dfb7d98d65a08d0"},
-    {file = "ruff-0.6.8-py3-none-manylinux_2_17_ppc64le.manylinux2014_ppc64le.whl", hash = "sha256:f8034b19b993e9601f2ddf2c517451e17a6ab5cdb1c13fdff50c1442a7171d87"},
-    {file = "ruff-0.6.8-py3-none-manylinux_2_17_s390x.manylinux2014_s390x.whl", hash = "sha256:6cfb227b932ba8ef6e56c9f875d987973cd5e35bc5d05f5abf045af78ad8e098"},
-    {file = "ruff-0.6.8-py3-none-manylinux_2_17_x86_64.manylinux2014_x86_64.whl", hash = "sha256:6ef0411eccfc3909269fed47c61ffebdcb84a04504bafa6b6df9b85c27e813b0"},
-    {file = "ruff-0.6.8-py3-none-musllinux_1_2_aarch64.whl", hash = "sha256:007dee844738c3d2e6c24ab5bc7d43c99ba3e1943bd2d95d598582e9c1b27750"},
-    {file = "ruff-0.6.8-py3-none-musllinux_1_2_armv7l.whl", hash = "sha256:ce60058d3cdd8490e5e5471ef086b3f1e90ab872b548814e35930e21d848c9ce"},
-    {file = "ruff-0.6.8-py3-none-musllinux_1_2_i686.whl", hash = "sha256:1085c455d1b3fdb8021ad534379c60353b81ba079712bce7a900e834859182fa"},
-    {file = "ruff-0.6.8-py3-none-musllinux_1_2_x86_64.whl", hash = "sha256:70edf6a93b19481affd287d696d9e311388d808671bc209fb8907b46a8c3af44"},
-    {file = "ruff-0.6.8-py3-none-win32.whl", hash = "sha256:792213f7be25316f9b46b854df80a77e0da87ec66691e8f012f887b4a671ab5a"},
-    {file = "ruff-0.6.8-py3-none-win_amd64.whl", hash = "sha256:ec0517dc0f37cad14a5319ba7bba6e7e339d03fbf967a6d69b0907d61be7a263"},
-    {file = "ruff-0.6.8-py3-none-win_arm64.whl", hash = "sha256:8d3bb2e3fbb9875172119021a13eed38849e762499e3cfde9588e4b4d70968dc"},
-    {file = "ruff-0.6.8.tar.gz", hash = "sha256:a5bf44b1aa0adaf6d9d20f86162b34f7c593bfedabc51239953e446aefc8ce18"},
+    {file = "ruff-0.6.9-py3-none-linux_armv6l.whl", hash = "sha256:064df58d84ccc0ac0fcd63bc3090b251d90e2a372558c0f057c3f75ed73e1ccd"},
+    {file = "ruff-0.6.9-py3-none-macosx_10_12_x86_64.whl", hash = "sha256:140d4b5c9f5fc7a7b074908a78ab8d384dd7f6510402267bc76c37195c02a7ec"},
+    {file = "ruff-0.6.9-py3-none-macosx_11_0_arm64.whl", hash = "sha256:53fd8ca5e82bdee8da7f506d7b03a261f24cd43d090ea9db9a1dc59d9313914c"},
+    {file = "ruff-0.6.9-py3-none-manylinux_2_17_aarch64.manylinux2014_aarch64.whl", hash = "sha256:645d7d8761f915e48a00d4ecc3686969761df69fb561dd914a773c1a8266e14e"},
+    {file = "ruff-0.6.9-py3-none-manylinux_2_17_armv7l.manylinux2014_armv7l.whl", hash = "sha256:eae02b700763e3847595b9d2891488989cac00214da7f845f4bcf2989007d577"},
+    {file = "ruff-0.6.9-py3-none-manylinux_2_17_i686.manylinux2014_i686.whl", hash = "sha256:7d5ccc9e58112441de8ad4b29dcb7a86dc25c5f770e3c06a9d57e0e5eba48829"},
+    {file = "ruff-0.6.9-py3-none-manylinux_2_17_ppc64.manylinux2014_ppc64.whl", hash = "sha256:417b81aa1c9b60b2f8edc463c58363075412866ae4e2b9ab0f690dc1e87ac1b5"},
+    {file = "ruff-0.6.9-py3-none-manylinux_2_17_ppc64le.manylinux2014_ppc64le.whl", hash = "sha256:3c866b631f5fbce896a74a6e4383407ba7507b815ccc52bcedabb6810fdb3ef7"},
+    {file = "ruff-0.6.9-py3-none-manylinux_2_17_s390x.manylinux2014_s390x.whl", hash = "sha256:7b118afbb3202f5911486ad52da86d1d52305b59e7ef2031cea3425142b97d6f"},
+    {file = "ruff-0.6.9-py3-none-manylinux_2_17_x86_64.manylinux2014_x86_64.whl", hash = "sha256:a67267654edc23c97335586774790cde402fb6bbdb3c2314f1fc087dee320bfa"},
+    {file = "ruff-0.6.9-py3-none-musllinux_1_2_aarch64.whl", hash = "sha256:3ef0cc774b00fec123f635ce5c547dac263f6ee9fb9cc83437c5904183b55ceb"},
+    {file = "ruff-0.6.9-py3-none-musllinux_1_2_armv7l.whl", hash = "sha256:12edd2af0c60fa61ff31cefb90aef4288ac4d372b4962c2864aeea3a1a2460c0"},
+    {file = "ruff-0.6.9-py3-none-musllinux_1_2_i686.whl", hash = "sha256:55bb01caeaf3a60b2b2bba07308a02fca6ab56233302406ed5245180a05c5625"},
+    {file = "ruff-0.6.9-py3-none-musllinux_1_2_x86_64.whl", hash = "sha256:925d26471fa24b0ce5a6cdfab1bb526fb4159952385f386bdcc643813d472039"},
+    {file = "ruff-0.6.9-py3-none-win32.whl", hash = "sha256:eb61ec9bdb2506cffd492e05ac40e5bc6284873aceb605503d8494180d6fc84d"},
+    {file = "ruff-0.6.9-py3-none-win_amd64.whl", hash = "sha256:785d31851c1ae91f45b3d8fe23b8ae4b5170089021fbb42402d811135f0b7117"},
+    {file = "ruff-0.6.9-py3-none-win_arm64.whl", hash = "sha256:a9641e31476d601f83cd602608739a0840e348bda93fec9f1ee816f8b6798b93"},
+    {file = "ruff-0.6.9.tar.gz", hash = "sha256:b076ef717a8e5bc819514ee1d602bbdca5b4420ae13a9cf61a0c0a4f53a2baa2"},
 ]
 
 [[package]]
@@ -8331,13 +8387,13 @@ test = ["accelerate (>=0.24.1,<=0.27.0)", "apache-airflow (==2.9.3)", "apache-ai
 
 [[package]]
 name = "sagemaker-core"
-version = "1.0.9"
+version = "1.0.10"
 description = "An python package for sagemaker core functionalities"
 optional = false
 python-versions = ">=3.8"
 files = [
-    {file = "sagemaker_core-1.0.9-py3-none-any.whl", hash = "sha256:7a22c46cf93594f8d44e3523d4ba98407911f3530af68a8ffdde5082d3b26fa3"},
-    {file = "sagemaker_core-1.0.9.tar.gz", hash = "sha256:664115faf797412553fb81b97a4777e78e51dfd4454c32edb2c8371bf203c535"},
+    {file = "sagemaker_core-1.0.10-py3-none-any.whl", hash = "sha256:0bdcf6a467db988919cc6b6d0077f74871ee24c24adf7f759f9cb98460e08953"},
+    {file = "sagemaker_core-1.0.10.tar.gz", hash = "sha256:6d34a9b6dc5e17e8bfffd1d0650726865779c92b3b8f1b59fc15d42061a0dd29"},
 ]
 
 [package.dependencies]
@@ -8887,13 +8943,13 @@ test = ["pytest", "tornado (>=4.5)", "typeguard"]
 
 [[package]]
 name = "tencentcloud-sdk-python-common"
-version = "3.0.1243"
+version = "3.0.1244"
 description = "Tencent Cloud Common SDK for Python"
 optional = false
 python-versions = "*"
 files = [
-    {file = "tencentcloud-sdk-python-common-3.0.1243.tar.gz", hash = "sha256:4008126b5506543ddc33ea53ef25abe19b4fa025006ec7751b3d5b3c046438b6"},
-    {file = "tencentcloud_sdk_python_common-3.0.1243-py2.py3-none-any.whl", hash = "sha256:ed5d2040627aeed2df1fe5f9d65b9987ef842222d79fb13136cc5866aa5ac50c"},
+    {file = "tencentcloud-sdk-python-common-3.0.1244.tar.gz", hash = "sha256:51e895a4f380ead51de5ebcef990cfda17cf466b130b8d1be7a95c11158a66e2"},
+    {file = "tencentcloud_sdk_python_common-3.0.1244-py2.py3-none-any.whl", hash = "sha256:a3275496740d7876386166e83331eab7ad66e70fe4216c96893e9697fdeb9848"},
 ]
 
 [package.dependencies]
@@ -8901,17 +8957,17 @@ requests = ">=2.16.0"
 
 [[package]]
 name = "tencentcloud-sdk-python-hunyuan"
-version = "3.0.1243"
+version = "3.0.1244"
 description = "Tencent Cloud Hunyuan SDK for Python"
 optional = false
 python-versions = "*"
 files = [
-    {file = "tencentcloud-sdk-python-hunyuan-3.0.1243.tar.gz", hash = "sha256:2caad419f24106574f05faf69d162060689e03452832fda890a28eb8814199bc"},
-    {file = "tencentcloud_sdk_python_hunyuan-3.0.1243-py2.py3-none-any.whl", hash = "sha256:b5a5c0dfa8b6de68c2c8747bf34b880c7fd2a1cbd59efc372e076260228ce231"},
+    {file = "tencentcloud-sdk-python-hunyuan-3.0.1244.tar.gz", hash = "sha256:b43f6848ac464cb244f6b1a17c97b00cd36ea40e91a4fba9f7b8ac1e94ae600a"},
+    {file = "tencentcloud_sdk_python_hunyuan-3.0.1244-py2.py3-none-any.whl", hash = "sha256:c2679f49d129dd5ec905cbac97884f6a872e50300ec7ba1ab4fe654d1458213f"},
 ]
 
 [package.dependencies]
-tencentcloud-sdk-python-common = "3.0.1243"
+tencentcloud-sdk-python-common = "3.0.1244"
 
 [[package]]
 name = "threadpoolctl"
@@ -9164,13 +9220,13 @@ files = [
 
 [[package]]
 name = "tomli"
-version = "2.0.1"
+version = "2.0.2"
 description = "A lil' TOML parser"
 optional = false
-python-versions = ">=3.7"
+python-versions = ">=3.8"
 files = [
-    {file = "tomli-2.0.1-py3-none-any.whl", hash = "sha256:939de3e7a6161af0c887ef91b7d41a53e7c5a1ca976325f429cb46ea9bc30ecc"},
-    {file = "tomli-2.0.1.tar.gz", hash = "sha256:de526c12914f0c550d15924c62d72abc48d6fe7364aa87328337a31007fe8a4f"},
+    {file = "tomli-2.0.2-py3-none-any.whl", hash = "sha256:2ebe24485c53d303f690b0ec092806a085f07af5a5aa1464f3931eec36caaa38"},
+    {file = "tomli-2.0.2.tar.gz", hash = "sha256:d46d457a85337051c36524bc5349dd91b1877838e2979ac5ced3e710ed8a60ed"},
 ]
 
 [[package]]
@@ -10293,13 +10349,13 @@ multidict = ">=4.0"
 
 [[package]]
 name = "yfinance"
-version = "0.2.43"
+version = "0.2.44"
 description = "Download market data from Yahoo! Finance API"
 optional = false
 python-versions = "*"
 files = [
-    {file = "yfinance-0.2.43-py2.py3-none-any.whl", hash = "sha256:11b4f5515b17450bd3bdcdc26b299aeeaea7ff9cb63d0fa0a865f460c0c7618f"},
-    {file = "yfinance-0.2.43.tar.gz", hash = "sha256:32404597f325a2a2c2708aceb8d552088dd26891ac0e6018f6c5f3f2f61055f0"},
+    {file = "yfinance-0.2.44-py2.py3-none-any.whl", hash = "sha256:fdc18791662f286539f7a08dccd7e8191b1ca509814f7b0faac264623bebe8a8"},
+    {file = "yfinance-0.2.44.tar.gz", hash = "sha256:532ad1644ee9cf4024ec0d9cade0cc073664ec0d140cc6c22a0cce8a9118b523"},
 ]
 
 [package.dependencies]
@@ -10539,4 +10595,4 @@ cffi = ["cffi (>=1.11)"]
 [metadata]
 lock-version = "2.0"
 python-versions = ">=3.10,<3.13"
-content-hash = "74a22b763f5f6f1414e843f1a508ff48210d32f9523b696e5dd443489deaad46"
+content-hash = "eff91cb29dc285a313ddf0ed6fbbc9c9972d955d97781f56dfd9d3bbde5fba4a"
diff --git a/api/pyproject.toml b/api/pyproject.toml
index fe9e64907b4c3..f177838bdbc9b 100644
--- a/api/pyproject.toml
+++ b/api/pyproject.toml
@@ -148,6 +148,7 @@ chardet = "~5.1.0"
 cohere = "~5.2.4"
 cos-python-sdk-v5 = "1.9.30"
 esdk-obs-python = "3.24.6.1"
+bce-python-sdk = "~0.9.23"
 dashscope = { version = "~1.17.0", extras = ["tokenizer"] }
 flask = "~3.0.1"
 flask-compress = "~1.14"

From 2ab8bc679fb3f558c989745965b5299d229a2ef9 Mon Sep 17 00:00:00 2001
From: aiscrm <30600079+aiscrm@users.noreply.github.com>
Date: Mon, 7 Oct 2024 18:03:30 +0800
Subject: [PATCH 145/235] fix: Missing model information in llm span of
 Langfuse #9029 (#9030)

Co-authored-by: corel <corelchen@qq.com>
---
 api/core/ops/langfuse_trace/langfuse_trace.py | 11 ++++++++++-
 1 file changed, 10 insertions(+), 1 deletion(-)

diff --git a/api/core/ops/langfuse_trace/langfuse_trace.py b/api/core/ops/langfuse_trace/langfuse_trace.py
index 6aefbec9aa626..171e34f8cb48f 100644
--- a/api/core/ops/langfuse_trace/langfuse_trace.py
+++ b/api/core/ops/langfuse_trace/langfuse_trace.py
@@ -159,6 +159,16 @@ def workflow_trace(self, trace_info: WorkflowTraceInfo):
                     "status": status,
                 }
             )
+            process_data = json.loads(node_execution.process_data) if node_execution.process_data else {}
+            model_provider = process_data.get("model_provider", None)
+            model_name = process_data.get("model_name", None)
+            if model_provider is not None and model_name is not None:
+                metadata.update(
+                    {
+                        "model_provider": model_provider,
+                        "model_name": model_name,
+                    }
+                )
 
             # add span
             if trace_info.message_id:
@@ -191,7 +201,6 @@ def workflow_trace(self, trace_info: WorkflowTraceInfo):
 
             self.add_span(langfuse_span_data=span_data)
 
-            process_data = json.loads(node_execution.process_data) if node_execution.process_data else {}
             if process_data and process_data.get("model_mode") == "chat":
                 total_token = metadata.get("total_tokens", 0)
                 # add generation

From f45042aa8efef635056f7ccdfdc18d76ed24015e Mon Sep 17 00:00:00 2001
From: =?UTF-8?q?=E5=91=86=E8=90=8C=E9=97=B7=E6=B2=B9=E7=93=B6?=
 <253605712@qq.com>
Date: Mon, 7 Oct 2024 22:13:41 +0800
Subject: [PATCH 146/235] fix:ddg ratelimit 202 (#9047)

---
 api/poetry.lock    | 18 +++++++++---------
 api/pyproject.toml |  2 +-
 2 files changed, 10 insertions(+), 10 deletions(-)

diff --git a/api/poetry.lock b/api/poetry.lock
index 7ea8e49941364..c24ec81639b45 100644
--- a/api/poetry.lock
+++ b/api/poetry.lock
@@ -732,7 +732,7 @@ name = "bce-python-sdk"
 version = "0.9.23"
 description = "BCE SDK for python"
 optional = false
-python-versions = ">=2.7, !=3.0.*, !=3.1.*, !=3.2.*, <4"
+python-versions = "!=3.0.*,!=3.1.*,!=3.2.*,<4,>=2.7"
 files = [
     {file = "bce_python_sdk-0.9.23-py3-none-any.whl", hash = "sha256:8debe21a040e00060f6044877d594765ed7b18bc765c6bf16b878bca864140a3"},
     {file = "bce_python_sdk-0.9.23.tar.gz", hash = "sha256:19739fed5cd0725356fc5ffa2acbdd8fb23f2a81edb91db21a03174551d0cf41"},
@@ -847,7 +847,7 @@ name = "botocore"
 version = "1.35.34"
 description = "Low-level, data-driven core of boto 3."
 optional = false
-python-versions = ">= 3.8"
+python-versions = ">=3.8"
 files = [
     {file = "botocore-1.35.34-py3-none-any.whl", hash = "sha256:ccb0fe397b11b81c9abc0c87029d17298e17bf658d8db5c0c5a551a12a207e7a"},
     {file = "botocore-1.35.34.tar.gz", hash = "sha256:789b6501a3bb4a9591c1fe10da200cc315c1fa5df5ada19c720d8ef06439b3e3"},
@@ -1068,7 +1068,7 @@ name = "build"
 version = "1.2.2.post1"
 description = "A simple, correct Python build frontend"
 optional = false
-python-versions = ">= 3.8"
+python-versions = ">=3.8"
 files = [
     {file = "build-1.2.2.post1-py3-none-any.whl", hash = "sha256:1d61c0887fa860c01971625baae8bdd338e517b836a2f70dd1f7aa3a6b2fc5b5"},
     {file = "build-1.2.2.post1.tar.gz", hash = "sha256:b36993e92ca9375a219c99e606a122ff365a760a2d4bba0caa09bd5278b608b7"},
@@ -2318,13 +2318,13 @@ files = [
 
 [[package]]
 name = "duckduckgo-search"
-version = "6.2.13"
+version = "6.3.0"
 description = "Search for words, documents, images, news, maps and text translation using the DuckDuckGo.com search engine."
 optional = false
 python-versions = ">=3.8"
 files = [
-    {file = "duckduckgo_search-6.2.13-py3-none-any.whl", hash = "sha256:a6618fb2744fa1d081b1bf2e47ef8051de993276a15c20f4e879a150726472de"},
-    {file = "duckduckgo_search-6.2.13.tar.gz", hash = "sha256:f89a9782f0f47d18c01a761c83453d0aef7a4335d1b6466fc47709652d5ca791"},
+    {file = "duckduckgo_search-6.3.0-py3-none-any.whl", hash = "sha256:9a231a7b325226811cf7d35a240f3f501e718ae10a1aa0a638cabc80e129dfe7"},
+    {file = "duckduckgo_search-6.3.0.tar.gz", hash = "sha256:e9f56955569325a7d9cacda2488ca78bf6629a459e74415892bee560b664f5eb"},
 ]
 
 [package.dependencies]
@@ -4486,7 +4486,7 @@ name = "langfuse"
 version = "2.51.3"
 description = "A client library for accessing langfuse"
 optional = false
-python-versions = ">=3.8.1,<4.0"
+python-versions = "<4.0,>=3.8.1"
 files = [
     {file = "langfuse-2.51.3-py3-none-any.whl", hash = "sha256:32aba050123656ec0c165583e1d33243cc7a14e7b8498ed3c9de808aa90306f1"},
     {file = "langfuse-2.51.3.tar.gz", hash = "sha256:ccd2109556ee232db717abfb751ee8a3139f074db7d3e06a4f1a756c349fc5ba"},
@@ -4511,7 +4511,7 @@ name = "langsmith"
 version = "0.1.131"
 description = "Client library to connect to the LangSmith LLM Tracing and Evaluation Platform."
 optional = false
-python-versions = ">=3.8.1,<4.0"
+python-versions = "<4.0,>=3.8.1"
 files = [
     {file = "langsmith-0.1.131-py3-none-any.whl", hash = "sha256:80c106b1c42307195cc0bb3a596472c41ef91b79d15bcee9938307800336c563"},
     {file = "langsmith-0.1.131.tar.gz", hash = "sha256:626101a3bf3ca481e5110d5155ace8aa066e4e9cc2fa7d96c8290ade0fbff797"},
@@ -10595,4 +10595,4 @@ cffi = ["cffi (>=1.11)"]
 [metadata]
 lock-version = "2.0"
 python-versions = ">=3.10,<3.13"
-content-hash = "eff91cb29dc285a313ddf0ed6fbbc9c9972d955d97781f56dfd9d3bbde5fba4a"
+content-hash = "dbcbcd7addb7c35c67f074fce00f6e25ec1e225498e5b9de3ef2fcd6af749bf6"
diff --git a/api/pyproject.toml b/api/pyproject.toml
index f177838bdbc9b..15674004e5084 100644
--- a/api/pyproject.toml
+++ b/api/pyproject.toml
@@ -237,7 +237,7 @@ arxiv = "2.1.0"
 cloudscraper = "1.2.71"
 matplotlib = "~3.8.2"
 newspaper3k = "0.2.8"
-duckduckgo-search = "^6.2.6"
+duckduckgo-search = "~6.3.0"
 jsonpath-ng = "1.6.1"
 numexpr = "~2.9.0"
 opensearch-py = "2.4.0"

From b933c9d2063769e577de8fa4169db4e310b060bf Mon Sep 17 00:00:00 2001
From: Bowen Liang <liangbowen@gf.com.cn>
Date: Tue, 8 Oct 2024 08:40:29 +0800
Subject: [PATCH 147/235] chore: move testing env variables from pyproject.toml
 to pytest.ini (#9019)

---
 api/pyproject.toml | 29 -----------------------------
 api/pytest.ini     | 29 +++++++++++++++++++++++++++++
 2 files changed, 29 insertions(+), 29 deletions(-)
 create mode 100644 api/pytest.ini

diff --git a/api/pyproject.toml b/api/pyproject.toml
index 15674004e5084..0266edc4d3f8a 100644
--- a/api/pyproject.toml
+++ b/api/pyproject.toml
@@ -96,35 +96,6 @@ ignore = [
 exclude = [
 ]
 
-[tool.pytest_env]
-OPENAI_API_KEY = "sk-IamNotARealKeyJustForMockTestKawaiiiiiiiiii"
-UPSTAGE_API_KEY = "up-aaaaaaaaaaaaaaaaaaaa"
-FIREWORKS_API_KEY = "fw_aaaaaaaaaaaaaaaaaaaa"
-NOMIC_API_KEY = "nk-aaaaaaaaaaaaaaaaaaaa"
-AZURE_OPENAI_API_BASE = "https://difyai-openai.openai.azure.com"
-AZURE_OPENAI_API_KEY = "xxxxb1707exxxxxxxxxxaaxxxxxf94"
-ANTHROPIC_API_KEY = "sk-ant-api11-IamNotARealKeyJustForMockTestKawaiiiiiiiiii-NotBaka-ASkksz"
-CHATGLM_API_BASE = "http://a.abc.com:11451"
-XINFERENCE_SERVER_URL = "http://a.abc.com:11451"
-XINFERENCE_GENERATION_MODEL_UID = "generate"
-XINFERENCE_CHAT_MODEL_UID = "chat"
-XINFERENCE_EMBEDDINGS_MODEL_UID = "embedding"
-XINFERENCE_RERANK_MODEL_UID = "rerank"
-GOOGLE_API_KEY = "abcdefghijklmnopqrstuvwxyz"
-HUGGINGFACE_API_KEY = "hf-awuwuwuwuwuwuwuwuwuwuwuwuwuwuwuwuwu"
-HUGGINGFACE_TEXT_GEN_ENDPOINT_URL = "a"
-HUGGINGFACE_TEXT2TEXT_GEN_ENDPOINT_URL = "b"
-HUGGINGFACE_EMBEDDINGS_ENDPOINT_URL = "c"
-MOCK_SWITCH = "true"
-CODE_MAX_STRING_LENGTH = "80000"
-CODE_EXECUTION_ENDPOINT = "http://127.0.0.1:8194"
-CODE_EXECUTION_API_KEY = "dify-sandbox"
-FIRECRAWL_API_KEY = "fc-"
-TEI_EMBEDDING_SERVER_URL = "http://a.abc.com:11451"
-TEI_RERANK_SERVER_URL = "http://a.abc.com:11451"
-MIXEDBREAD_API_KEY = "mk-aaaaaaaaaaaaaaaaaaaa"
-VOYAGE_API_KEY = "va-aaaaaaaaaaaaaaaaaaaa"
-
 [tool.poetry]
 name = "dify-api"
 package-mode = false
diff --git a/api/pytest.ini b/api/pytest.ini
new file mode 100644
index 0000000000000..dcca08e2e5cb9
--- /dev/null
+++ b/api/pytest.ini
@@ -0,0 +1,29 @@
+[pytest]
+env =
+    ANTHROPIC_API_KEY = sk-ant-api11-IamNotARealKeyJustForMockTestKawaiiiiiiiiii-NotBaka-ASkksz
+    AZURE_OPENAI_API_BASE = https://difyai-openai.openai.azure.com
+    AZURE_OPENAI_API_KEY = xxxxb1707exxxxxxxxxxaaxxxxxf94
+    CHATGLM_API_BASE = http://a.abc.com:11451
+    CODE_EXECUTION_API_KEY = dify-sandbox
+    CODE_EXECUTION_ENDPOINT = http://127.0.0.1:8194
+    CODE_MAX_STRING_LENGTH = 80000
+    FIRECRAWL_API_KEY = fc-
+    FIREWORKS_API_KEY = fw_aaaaaaaaaaaaaaaaaaaa
+    GOOGLE_API_KEY = abcdefghijklmnopqrstuvwxyz
+    HUGGINGFACE_API_KEY = hf-awuwuwuwuwuwuwuwuwuwuwuwuwuwuwuwuwu
+    HUGGINGFACE_EMBEDDINGS_ENDPOINT_URL = c
+    HUGGINGFACE_TEXT2TEXT_GEN_ENDPOINT_URL = b
+    HUGGINGFACE_TEXT_GEN_ENDPOINT_URL = a
+    MIXEDBREAD_API_KEY = mk-aaaaaaaaaaaaaaaaaaaa
+    MOCK_SWITCH = true
+    NOMIC_API_KEY = nk-aaaaaaaaaaaaaaaaaaaa
+    OPENAI_API_KEY = sk-IamNotARealKeyJustForMockTestKawaiiiiiiiiii
+    TEI_EMBEDDING_SERVER_URL = http://a.abc.com:11451
+    TEI_RERANK_SERVER_URL = http://a.abc.com:11451
+    UPSTAGE_API_KEY = up-aaaaaaaaaaaaaaaaaaaa
+    VOYAGE_API_KEY = va-aaaaaaaaaaaaaaaaaaaa
+    XINFERENCE_CHAT_MODEL_UID = chat
+    XINFERENCE_EMBEDDINGS_MODEL_UID = embedding
+    XINFERENCE_GENERATION_MODEL_UID = generate
+    XINFERENCE_RERANK_MODEL_UID = rerank
+    XINFERENCE_SERVER_URL = http://a.abc.com:11451

From a8b4d1ac2a17ec57b945e35b1dd59cc6e8fba759 Mon Sep 17 00:00:00 2001
From: =?UTF-8?q?Sergio=20Sacrist=C3=A1n?= <extra@sacristan.com.es>
Date: Tue, 8 Oct 2024 05:12:21 +0200
Subject: [PATCH 148/235] feat: Improvement- use non root user for Web
 container (#8928)

---
 web/Dockerfile | 14 ++++++++++----
 1 file changed, 10 insertions(+), 4 deletions(-)

diff --git a/web/Dockerfile b/web/Dockerfile
index 48bdb2301ad20..29f7675f4af7e 100644
--- a/web/Dockerfile
+++ b/web/Dockerfile
@@ -46,21 +46,27 @@ ENV TZ=UTC
 RUN ln -s /usr/share/zoneinfo/${TZ} /etc/localtime \
     && echo ${TZ} > /etc/timezone
 
-# global runtime packages
-RUN yarn global add pm2 \
-    && yarn cache clean
 
 WORKDIR /app/web
 COPY --from=builder /app/web/public ./public
 COPY --from=builder /app/web/.next/standalone ./
 COPY --from=builder /app/web/.next/static ./.next/static
 
-
 COPY docker/pm2.json ./pm2.json
 COPY docker/entrypoint.sh ./entrypoint.sh
 
+
+# global runtime packages
+RUN yarn global add pm2 \
+    && yarn cache clean \
+    && mkdir /.pm2 \
+    && chown -R 1001:0 /.pm2 /app/web \
+    && chmod -R g=u /.pm2 /app/web
+
+
 ARG COMMIT_SHA
 ENV COMMIT_SHA=${COMMIT_SHA}
 
+USER 1001
 EXPOSE 3000
 ENTRYPOINT ["/bin/sh", "./entrypoint.sh"]

From 7c0b159a81f099c5089c96ba77b32d592e5787b8 Mon Sep 17 00:00:00 2001
From: Bowen Liang <liangbowen@gf.com.cn>
Date: Tue, 8 Oct 2024 11:13:11 +0800
Subject: [PATCH 149/235] chore: removing unused imports in tests (#9049)

---
 api/app.py                                           | 12 ++++++------
 api/poetry.lock                                      |  2 +-
 api/pyproject.toml                                   | 11 +++++++----
 .../model_runtime/__mock/anthropic.py                |  2 +-
 .../model_runtime/__mock/nomic_embeddings.py         |  2 +-
 .../model_runtime/__mock/openai_chat.py              |  4 +---
 .../model_runtime/__mock/openai_embeddings.py        |  1 -
 .../model_runtime/__mock/openai_remote.py            |  1 -
 .../model_runtime/__mock/xinference.py               |  1 -
 .../model_runtime/azure_ai_studio/test_llm.py        |  3 ---
 .../model_runtime/chatglm/test_llm.py                |  1 -
 .../model_runtime/huggingface_tei/test_rerank.py     |  3 +--
 .../model_runtime/localai/test_llm.py                |  2 --
 .../model_runtime/nomic/test_provider.py             |  1 -
 .../model_runtime/novita/test_llm.py                 |  1 -
 .../integration_tests/model_runtime/oci/test_llm.py  |  1 -
 .../model_runtime/openai/test_llm.py                 |  1 -
 .../model_runtime/openrouter/test_llm.py             |  1 -
 .../model_runtime/sagemaker/test_provider.py         |  2 --
 .../model_runtime/sagemaker/test_text_embedding.py   |  2 --
 .../model_runtime/stepfun/test_llm.py                |  4 +---
 .../model_runtime/togetherai/test_llm.py             |  1 -
 .../model_runtime/upstage/test_llm.py                |  3 +--
 .../model_runtime/xinference/test_llm.py             |  2 --
 .../integration_tests/vdb/pgvector/test_pgvector.py  |  1 -
 api/tests/integration_tests/vdb/test_vector_store.py |  1 -
 .../workflow/nodes/__mock/code_executor.py           |  2 +-
 .../workflow/nodes/test_parameter_extractor.py       |  1 -
 .../unit_tests/core/app/segments/test_variables.py   |  1 -
 .../core/rag/extractor/firecrawl/test_firecrawl.py   |  3 ---
 30 files changed, 21 insertions(+), 52 deletions(-)

diff --git a/api/app.py b/api/app.py
index 1b58beee15819..0662d94e1b4c8 100644
--- a/api/app.py
+++ b/api/app.py
@@ -26,7 +26,7 @@
 from configs import dify_config
 
 # DO NOT REMOVE BELOW
-from events import event_handlers
+from events import event_handlers  # noqa: F401
 from extensions import (
     ext_celery,
     ext_code_based_extension,
@@ -45,7 +45,7 @@
 from libs.passport import PassportService
 
 # TODO: Find a way to avoid importing models here
-from models import account, dataset, model, source, task, tool, tools, web
+from models import account, dataset, model, source, task, tool, tools, web  # noqa: F401
 from services.account_service import AccountService
 
 # DO NOT REMOVE ABOVE
@@ -181,10 +181,10 @@ def load_user_from_request(request_from_flask_login):
     decoded = PassportService().verify(auth_token)
     user_id = decoded.get("user_id")
 
-    account = AccountService.load_logged_in_account(account_id=user_id, token=auth_token)
-    if account:
-        contexts.tenant_id.set(account.current_tenant_id)
-    return account
+    logged_in_account = AccountService.load_logged_in_account(account_id=user_id, token=auth_token)
+    if logged_in_account:
+        contexts.tenant_id.set(logged_in_account.current_tenant_id)
+    return logged_in_account
 
 
 @login_manager.unauthorized_handler
diff --git a/api/poetry.lock b/api/poetry.lock
index c24ec81639b45..4b9fdd15263c3 100644
--- a/api/poetry.lock
+++ b/api/poetry.lock
@@ -10595,4 +10595,4 @@ cffi = ["cffi (>=1.11)"]
 [metadata]
 lock-version = "2.0"
 python-versions = ">=3.10,<3.13"
-content-hash = "dbcbcd7addb7c35c67f074fce00f6e25ec1e225498e5b9de3ef2fcd6af749bf6"
+content-hash = "d29d0c4ce384ef94fe65f0a2a145898bd1a117d4fd59c217d15bbb8993f4ce4e"
diff --git a/api/pyproject.toml b/api/pyproject.toml
index 0266edc4d3f8a..1c12cc3a3692c 100644
--- a/api/pyproject.toml
+++ b/api/pyproject.toml
@@ -74,8 +74,6 @@ ignore = [
 
 [tool.ruff.lint.per-file-ignores]
 "app.py" = [
-    "F401", # unused-import
-    "F811", # redefined-while-unused
 ]
 "__init__.py" = [
     "F401", # unused-import
@@ -88,10 +86,15 @@ ignore = [
     "N803", # invalid-argument-name
 ]
 "tests/*" = [
-    "F401", # unused-import
     "F811", # redefined-while-unused
 ]
 
+[tool.ruff.lint.pyflakes]
+allowed-unused-imports=[
+    "_pytest.monkeypatch",
+    "tests.integration_tests",
+]
+
 [tool.ruff.format]
 exclude = [
 ]
@@ -260,4 +263,4 @@ optional = true
 
 [tool.poetry.group.lint.dependencies]
 dotenv-linter = "~0.5.0"
-ruff = "~0.6.8"
+ruff = "~0.6.9"
diff --git a/api/tests/integration_tests/model_runtime/__mock/anthropic.py b/api/tests/integration_tests/model_runtime/__mock/anthropic.py
index 79a3dc03941c9..5092af4f13b2f 100644
--- a/api/tests/integration_tests/model_runtime/__mock/anthropic.py
+++ b/api/tests/integration_tests/model_runtime/__mock/anthropic.py
@@ -5,7 +5,7 @@
 import anthropic
 import pytest
 from _pytest.monkeypatch import MonkeyPatch
-from anthropic import Anthropic, Stream
+from anthropic import Stream
 from anthropic.resources import Messages
 from anthropic.types import (
     ContentBlock,
diff --git a/api/tests/integration_tests/model_runtime/__mock/nomic_embeddings.py b/api/tests/integration_tests/model_runtime/__mock/nomic_embeddings.py
index 281e866e45c2e..6a25398cbf069 100644
--- a/api/tests/integration_tests/model_runtime/__mock/nomic_embeddings.py
+++ b/api/tests/integration_tests/model_runtime/__mock/nomic_embeddings.py
@@ -1,6 +1,6 @@
 import os
 from collections.abc import Callable
-from typing import Any, Literal, Union
+from typing import Any, Literal
 
 import pytest
 
diff --git a/api/tests/integration_tests/model_runtime/__mock/openai_chat.py b/api/tests/integration_tests/model_runtime/__mock/openai_chat.py
index 439f7d56e9b5d..1dc5df766748e 100644
--- a/api/tests/integration_tests/model_runtime/__mock/openai_chat.py
+++ b/api/tests/integration_tests/model_runtime/__mock/openai_chat.py
@@ -1,6 +1,6 @@
 import re
 from collections.abc import Generator
-from json import dumps, loads
+from json import dumps
 from time import time
 
 # import monkeypatch
@@ -11,11 +11,9 @@
 from openai.resources.chat.completions import Completions
 from openai.types import Completion as CompletionMessage
 from openai.types.chat import (
-    ChatCompletion,
     ChatCompletionChunk,
     ChatCompletionMessageParam,
     ChatCompletionMessageToolCall,
-    ChatCompletionToolChoiceOptionParam,
     ChatCompletionToolParam,
     completion_create_params,
 )
diff --git a/api/tests/integration_tests/model_runtime/__mock/openai_embeddings.py b/api/tests/integration_tests/model_runtime/__mock/openai_embeddings.py
index e27b9891f5c8a..3cc1fa9ff1db4 100644
--- a/api/tests/integration_tests/model_runtime/__mock/openai_embeddings.py
+++ b/api/tests/integration_tests/model_runtime/__mock/openai_embeddings.py
@@ -1,7 +1,6 @@
 import re
 from typing import Any, Literal, Union
 
-from openai import OpenAI
 from openai._types import NOT_GIVEN, NotGiven
 from openai.resources.embeddings import Embeddings
 from openai.types.create_embedding_response import CreateEmbeddingResponse, Usage
diff --git a/api/tests/integration_tests/model_runtime/__mock/openai_remote.py b/api/tests/integration_tests/model_runtime/__mock/openai_remote.py
index cb8f249543878..704dbad5d288c 100644
--- a/api/tests/integration_tests/model_runtime/__mock/openai_remote.py
+++ b/api/tests/integration_tests/model_runtime/__mock/openai_remote.py
@@ -1,6 +1,5 @@
 from time import time
 
-from openai.resources.models import Models
 from openai.types.model import Model
 
 
diff --git a/api/tests/integration_tests/model_runtime/__mock/xinference.py b/api/tests/integration_tests/model_runtime/__mock/xinference.py
index 8deb50635f3d5..5f7dad50c10f1 100644
--- a/api/tests/integration_tests/model_runtime/__mock/xinference.py
+++ b/api/tests/integration_tests/model_runtime/__mock/xinference.py
@@ -5,7 +5,6 @@
 import pytest
 from _pytest.monkeypatch import MonkeyPatch
 from requests import Response
-from requests.exceptions import ConnectionError
 from requests.sessions import Session
 from xinference_client.client.restful.restful_client import (
     Client,
diff --git a/api/tests/integration_tests/model_runtime/azure_ai_studio/test_llm.py b/api/tests/integration_tests/model_runtime/azure_ai_studio/test_llm.py
index 8655b43d8f0b3..85a4f7734dc47 100644
--- a/api/tests/integration_tests/model_runtime/azure_ai_studio/test_llm.py
+++ b/api/tests/integration_tests/model_runtime/azure_ai_studio/test_llm.py
@@ -6,10 +6,7 @@
 from core.model_runtime.entities.llm_entities import LLMResult, LLMResultChunk, LLMResultChunkDelta
 from core.model_runtime.entities.message_entities import (
     AssistantPromptMessage,
-    ImagePromptMessageContent,
-    PromptMessageTool,
     SystemPromptMessage,
-    TextPromptMessageContent,
     UserPromptMessage,
 )
 from core.model_runtime.errors.validate import CredentialsValidateFailedError
diff --git a/api/tests/integration_tests/model_runtime/chatglm/test_llm.py b/api/tests/integration_tests/model_runtime/chatglm/test_llm.py
index 418e88874d157..a7c5229e05141 100644
--- a/api/tests/integration_tests/model_runtime/chatglm/test_llm.py
+++ b/api/tests/integration_tests/model_runtime/chatglm/test_llm.py
@@ -8,7 +8,6 @@
     AssistantPromptMessage,
     PromptMessageTool,
     SystemPromptMessage,
-    TextPromptMessageContent,
     UserPromptMessage,
 )
 from core.model_runtime.entities.model_entities import AIModelEntity
diff --git a/api/tests/integration_tests/model_runtime/huggingface_tei/test_rerank.py b/api/tests/integration_tests/model_runtime/huggingface_tei/test_rerank.py
index 45370d9fba41b..cd1c20dd026f7 100644
--- a/api/tests/integration_tests/model_runtime/huggingface_tei/test_rerank.py
+++ b/api/tests/integration_tests/model_runtime/huggingface_tei/test_rerank.py
@@ -2,8 +2,7 @@
 
 import pytest
 
-from core.model_runtime.entities.rerank_entities import RerankDocument, RerankResult
-from core.model_runtime.entities.text_embedding_entities import TextEmbeddingResult
+from core.model_runtime.entities.rerank_entities import RerankResult
 from core.model_runtime.errors.validate import CredentialsValidateFailedError
 from core.model_runtime.model_providers.huggingface_tei.rerank.rerank import (
     HuggingfaceTeiRerankModel,
diff --git a/api/tests/integration_tests/model_runtime/localai/test_llm.py b/api/tests/integration_tests/model_runtime/localai/test_llm.py
index aa5436c34fc7d..51e899fd5186c 100644
--- a/api/tests/integration_tests/model_runtime/localai/test_llm.py
+++ b/api/tests/integration_tests/model_runtime/localai/test_llm.py
@@ -8,10 +8,8 @@
     AssistantPromptMessage,
     PromptMessageTool,
     SystemPromptMessage,
-    TextPromptMessageContent,
     UserPromptMessage,
 )
-from core.model_runtime.entities.model_entities import ParameterRule
 from core.model_runtime.errors.validate import CredentialsValidateFailedError
 from core.model_runtime.model_providers.localai.llm.llm import LocalAILanguageModel
 
diff --git a/api/tests/integration_tests/model_runtime/nomic/test_provider.py b/api/tests/integration_tests/model_runtime/nomic/test_provider.py
index 6cad400c06955..ece4bb920080b 100644
--- a/api/tests/integration_tests/model_runtime/nomic/test_provider.py
+++ b/api/tests/integration_tests/model_runtime/nomic/test_provider.py
@@ -4,7 +4,6 @@
 
 from core.model_runtime.errors.validate import CredentialsValidateFailedError
 from core.model_runtime.model_providers.nomic.nomic import NomicAtlasProvider
-from core.model_runtime.model_providers.nomic.text_embedding.text_embedding import NomicTextEmbeddingModel
 from tests.integration_tests.model_runtime.__mock.nomic_embeddings import setup_nomic_mock
 
 
diff --git a/api/tests/integration_tests/model_runtime/novita/test_llm.py b/api/tests/integration_tests/model_runtime/novita/test_llm.py
index 35fa0dc1904f7..9f92679cd5ccf 100644
--- a/api/tests/integration_tests/model_runtime/novita/test_llm.py
+++ b/api/tests/integration_tests/model_runtime/novita/test_llm.py
@@ -6,7 +6,6 @@
 from core.model_runtime.entities.llm_entities import LLMResult, LLMResultChunk, LLMResultChunkDelta
 from core.model_runtime.entities.message_entities import (
     AssistantPromptMessage,
-    PromptMessageTool,
     SystemPromptMessage,
     UserPromptMessage,
 )
diff --git a/api/tests/integration_tests/model_runtime/oci/test_llm.py b/api/tests/integration_tests/model_runtime/oci/test_llm.py
index 531f26a32e657..bd5d27eb0f2a0 100644
--- a/api/tests/integration_tests/model_runtime/oci/test_llm.py
+++ b/api/tests/integration_tests/model_runtime/oci/test_llm.py
@@ -8,7 +8,6 @@
     AssistantPromptMessage,
     PromptMessageTool,
     SystemPromptMessage,
-    TextPromptMessageContent,
     UserPromptMessage,
 )
 from core.model_runtime.errors.validate import CredentialsValidateFailedError
diff --git a/api/tests/integration_tests/model_runtime/openai/test_llm.py b/api/tests/integration_tests/model_runtime/openai/test_llm.py
index 3b3ea9ec800cb..41c99f68756eb 100644
--- a/api/tests/integration_tests/model_runtime/openai/test_llm.py
+++ b/api/tests/integration_tests/model_runtime/openai/test_llm.py
@@ -14,7 +14,6 @@
 )
 from core.model_runtime.entities.model_entities import AIModelEntity, ModelType
 from core.model_runtime.errors.validate import CredentialsValidateFailedError
-from core.model_runtime.model_providers.__base.large_language_model import LargeLanguageModel
 from core.model_runtime.model_providers.openai.llm.llm import OpenAILargeLanguageModel
 
 """FOR MOCK FIXTURES, DO NOT REMOVE"""
diff --git a/api/tests/integration_tests/model_runtime/openrouter/test_llm.py b/api/tests/integration_tests/model_runtime/openrouter/test_llm.py
index ce4876a73a740..1b0cc6bf4b8e7 100644
--- a/api/tests/integration_tests/model_runtime/openrouter/test_llm.py
+++ b/api/tests/integration_tests/model_runtime/openrouter/test_llm.py
@@ -6,7 +6,6 @@
 from core.model_runtime.entities.llm_entities import LLMResult, LLMResultChunk, LLMResultChunkDelta
 from core.model_runtime.entities.message_entities import (
     AssistantPromptMessage,
-    PromptMessageTool,
     SystemPromptMessage,
     UserPromptMessage,
 )
diff --git a/api/tests/integration_tests/model_runtime/sagemaker/test_provider.py b/api/tests/integration_tests/model_runtime/sagemaker/test_provider.py
index 9f0b439d6c32a..41de2a17fda04 100644
--- a/api/tests/integration_tests/model_runtime/sagemaker/test_provider.py
+++ b/api/tests/integration_tests/model_runtime/sagemaker/test_provider.py
@@ -1,5 +1,3 @@
-import os
-
 import pytest
 
 from core.model_runtime.errors.validate import CredentialsValidateFailedError
diff --git a/api/tests/integration_tests/model_runtime/sagemaker/test_text_embedding.py b/api/tests/integration_tests/model_runtime/sagemaker/test_text_embedding.py
index e4e404c7a86ae..f77601eea2c26 100644
--- a/api/tests/integration_tests/model_runtime/sagemaker/test_text_embedding.py
+++ b/api/tests/integration_tests/model_runtime/sagemaker/test_text_embedding.py
@@ -1,5 +1,3 @@
-import os
-
 import pytest
 
 from core.model_runtime.entities.text_embedding_entities import TextEmbeddingResult
diff --git a/api/tests/integration_tests/model_runtime/stepfun/test_llm.py b/api/tests/integration_tests/model_runtime/stepfun/test_llm.py
index c03b1bae1f157..f9afca6f5945b 100644
--- a/api/tests/integration_tests/model_runtime/stepfun/test_llm.py
+++ b/api/tests/integration_tests/model_runtime/stepfun/test_llm.py
@@ -6,13 +6,11 @@
 from core.model_runtime.entities.llm_entities import LLMResult, LLMResultChunk, LLMResultChunkDelta
 from core.model_runtime.entities.message_entities import (
     AssistantPromptMessage,
-    ImagePromptMessageContent,
     PromptMessageTool,
     SystemPromptMessage,
-    TextPromptMessageContent,
     UserPromptMessage,
 )
-from core.model_runtime.entities.model_entities import AIModelEntity, ModelType
+from core.model_runtime.entities.model_entities import AIModelEntity
 from core.model_runtime.errors.validate import CredentialsValidateFailedError
 from core.model_runtime.model_providers.stepfun.llm.llm import StepfunLargeLanguageModel
 
diff --git a/api/tests/integration_tests/model_runtime/togetherai/test_llm.py b/api/tests/integration_tests/model_runtime/togetherai/test_llm.py
index 06ebc2a82dc75..5787e1bf6a8d9 100644
--- a/api/tests/integration_tests/model_runtime/togetherai/test_llm.py
+++ b/api/tests/integration_tests/model_runtime/togetherai/test_llm.py
@@ -6,7 +6,6 @@
 from core.model_runtime.entities.llm_entities import LLMResult, LLMResultChunk, LLMResultChunkDelta
 from core.model_runtime.entities.message_entities import (
     AssistantPromptMessage,
-    PromptMessageTool,
     SystemPromptMessage,
     UserPromptMessage,
 )
diff --git a/api/tests/integration_tests/model_runtime/upstage/test_llm.py b/api/tests/integration_tests/model_runtime/upstage/test_llm.py
index bc7517acbe260..0f39e902f3380 100644
--- a/api/tests/integration_tests/model_runtime/upstage/test_llm.py
+++ b/api/tests/integration_tests/model_runtime/upstage/test_llm.py
@@ -10,9 +10,8 @@
     SystemPromptMessage,
     UserPromptMessage,
 )
-from core.model_runtime.entities.model_entities import AIModelEntity, ModelType
+from core.model_runtime.entities.model_entities import AIModelEntity
 from core.model_runtime.errors.validate import CredentialsValidateFailedError
-from core.model_runtime.model_providers.__base.large_language_model import LargeLanguageModel
 from core.model_runtime.model_providers.upstage.llm.llm import UpstageLargeLanguageModel
 
 """FOR MOCK FIXTURES, DO NOT REMOVE"""
diff --git a/api/tests/integration_tests/model_runtime/xinference/test_llm.py b/api/tests/integration_tests/model_runtime/xinference/test_llm.py
index fb5e03855d212..5e4cde36389b5 100644
--- a/api/tests/integration_tests/model_runtime/xinference/test_llm.py
+++ b/api/tests/integration_tests/model_runtime/xinference/test_llm.py
@@ -8,10 +8,8 @@
     AssistantPromptMessage,
     PromptMessageTool,
     SystemPromptMessage,
-    TextPromptMessageContent,
     UserPromptMessage,
 )
-from core.model_runtime.entities.model_entities import AIModelEntity
 from core.model_runtime.errors.validate import CredentialsValidateFailedError
 from core.model_runtime.model_providers.xinference.llm.llm import XinferenceAILargeLanguageModel
 
diff --git a/api/tests/integration_tests/vdb/pgvector/test_pgvector.py b/api/tests/integration_tests/vdb/pgvector/test_pgvector.py
index 72efdc2780eca..3d2cfde5d1dc4 100644
--- a/api/tests/integration_tests/vdb/pgvector/test_pgvector.py
+++ b/api/tests/integration_tests/vdb/pgvector/test_pgvector.py
@@ -1,5 +1,4 @@
 from core.rag.datasource.vdb.pgvector.pgvector import PGVector, PGVectorConfig
-from core.rag.models.document import Document
 from tests.integration_tests.vdb.test_vector_store import (
     AbstractVectorTest,
     get_example_text,
diff --git a/api/tests/integration_tests/vdb/test_vector_store.py b/api/tests/integration_tests/vdb/test_vector_store.py
index a11cd225b3ba3..50519e2052cd8 100644
--- a/api/tests/integration_tests/vdb/test_vector_store.py
+++ b/api/tests/integration_tests/vdb/test_vector_store.py
@@ -1,4 +1,3 @@
-import random
 import uuid
 from unittest.mock import MagicMock
 
diff --git a/api/tests/integration_tests/workflow/nodes/__mock/code_executor.py b/api/tests/integration_tests/workflow/nodes/__mock/code_executor.py
index 6fb8c86b82a13..30414811ea798 100644
--- a/api/tests/integration_tests/workflow/nodes/__mock/code_executor.py
+++ b/api/tests/integration_tests/workflow/nodes/__mock/code_executor.py
@@ -1,5 +1,5 @@
 import os
-from typing import Literal, Optional
+from typing import Literal
 
 import pytest
 from _pytest.monkeypatch import MonkeyPatch
diff --git a/api/tests/integration_tests/workflow/nodes/test_parameter_extractor.py b/api/tests/integration_tests/workflow/nodes/test_parameter_extractor.py
index 88435c4022737..4c695f744336d 100644
--- a/api/tests/integration_tests/workflow/nodes/test_parameter_extractor.py
+++ b/api/tests/integration_tests/workflow/nodes/test_parameter_extractor.py
@@ -1,4 +1,3 @@
-import json
 import os
 import time
 import uuid
diff --git a/api/tests/unit_tests/core/app/segments/test_variables.py b/api/tests/unit_tests/core/app/segments/test_variables.py
index b3f0ae626cf24..6179675cde7cc 100644
--- a/api/tests/unit_tests/core/app/segments/test_variables.py
+++ b/api/tests/unit_tests/core/app/segments/test_variables.py
@@ -2,7 +2,6 @@
 from pydantic import ValidationError
 
 from core.app.segments import (
-    ArrayAnyVariable,
     FloatVariable,
     IntegerVariable,
     ObjectVariable,
diff --git a/api/tests/unit_tests/core/rag/extractor/firecrawl/test_firecrawl.py b/api/tests/unit_tests/core/rag/extractor/firecrawl/test_firecrawl.py
index d5a1d8f436c75..8fcdf2e8e5310 100644
--- a/api/tests/unit_tests/core/rag/extractor/firecrawl/test_firecrawl.py
+++ b/api/tests/unit_tests/core/rag/extractor/firecrawl/test_firecrawl.py
@@ -1,9 +1,6 @@
 import os
-from unittest import mock
 
 from core.rag.extractor.firecrawl.firecrawl_app import FirecrawlApp
-from core.rag.extractor.firecrawl.firecrawl_web_extractor import FirecrawlWebExtractor
-from core.rag.models.document import Document
 from tests.unit_tests.core.rag.extractor.test_notion_extractor import _mock_response
 
 

From 4abca8614f2d24413946a3d1688720b1b304b668 Mon Sep 17 00:00:00 2001
From: NFish <douxc512@gmail.com>
Date: Tue, 8 Oct 2024 11:48:22 +0800
Subject: [PATCH 150/235] fix: Ignore the error toast notification if the
 status is 401 and isPublicAPI is true (#9062)

---
 web/service/base.ts | 8 +++-----
 1 file changed, 3 insertions(+), 5 deletions(-)

diff --git a/web/service/base.ts b/web/service/base.ts
index 83389d8be80fa..fbdd5c1fd3584 100644
--- a/web/service/base.ts
+++ b/web/service/base.ts
@@ -359,9 +359,6 @@ const baseFetch = <T>(
               case 401: {
                 if (isPublicAPI) {
                   return bodyJson.then((data: ResponseError) => {
-                    if (!silent)
-                      Toast.notify({ type: 'error', message: data.message })
-
                     if (data.code === 'web_sso_auth_required')
                       requiredWebSSOLogin()
 
@@ -536,8 +533,6 @@ export const ssePost = (
     .then((res) => {
       if (!/^(2|3)\d{2}$/.test(String(res.status))) {
         res.json().then((data: any) => {
-          Toast.notify({ type: 'error', message: data.message || 'Server Error' })
-
           if (isPublicAPI) {
             if (data.code === 'web_sso_auth_required')
               requiredWebSSOLogin()
@@ -546,7 +541,10 @@ export const ssePost = (
               removeAccessToken()
               globalThis.location.reload()
             }
+            if (res.status === 401)
+              return
           }
+          Toast.notify({ type: 'error', message: data.message || 'Server Error' })
         })
         onError?.('Server Error')
         return

From 896998ef3fc28628a70d07e7ca483b253f3bd436 Mon Sep 17 00:00:00 2001
From: Bowen Liang <liangbowen@gf.com.cn>
Date: Tue, 8 Oct 2024 15:11:45 +0800
Subject: [PATCH 151/235] chore: refine python dependency list and check
 dependencies in order (#9061)

---
 .github/workflows/api-tests.yml               |  5 +-
 api/poetry.lock                               |  2 +-
 api/pyproject.toml                            | 73 ++++++++++---------
 .../artifact_tests/dependencies/__init__.py   |  0
 .../dependencies/test_dependencies_sorted.py  | 61 ++++++++++++++++
 dev/pytest/pytest_artifacts.sh                |  4 +
 dev/sync-poetry                               |  3 +
 7 files changed, 112 insertions(+), 36 deletions(-)
 create mode 100644 api/tests/artifact_tests/dependencies/__init__.py
 create mode 100644 api/tests/artifact_tests/dependencies/test_dependencies_sorted.py
 create mode 100755 dev/pytest/pytest_artifacts.sh

diff --git a/.github/workflows/api-tests.yml b/.github/workflows/api-tests.yml
index 7c632f8a34d56..b1cf41a226178 100644
--- a/.github/workflows/api-tests.yml
+++ b/.github/workflows/api-tests.yml
@@ -39,7 +39,7 @@ jobs:
             api/pyproject.toml
             api/poetry.lock
 
-      - name: Poetry check
+      - name: Check Poetry lockfile
         run: |
           poetry check -C api --lock
           poetry show -C api
@@ -47,6 +47,9 @@ jobs:
       - name: Install dependencies
         run: poetry install -C api --with dev
 
+      - name: Check dependencies in pyproject.toml
+        run: poetry run -C api bash dev/pytest/pytest_artifacts.sh
+
       - name: Run Unit tests
         run: poetry run -C api bash dev/pytest/pytest_unit_tests.sh
 
diff --git a/api/poetry.lock b/api/poetry.lock
index 4b9fdd15263c3..cf03f8c0f5b80 100644
--- a/api/poetry.lock
+++ b/api/poetry.lock
@@ -10595,4 +10595,4 @@ cffi = ["cffi (>=1.11)"]
 [metadata]
 lock-version = "2.0"
 python-versions = ">=3.10,<3.13"
-content-hash = "d29d0c4ce384ef94fe65f0a2a145898bd1a117d4fd59c217d15bbb8993f4ce4e"
+content-hash = "fd183812f910faf4e840267501c571db5d758ad6eb328d106ba6f79a0322a555"
diff --git a/api/pyproject.toml b/api/pyproject.toml
index 1c12cc3a3692c..839df9ea6713e 100644
--- a/api/pyproject.toml
+++ b/api/pyproject.toml
@@ -104,25 +104,23 @@ name = "dify-api"
 package-mode = false
 
 ############################################################
-# Main dependencies
+# [ Main ] Dependency group
 ############################################################
 
 [tool.poetry.dependencies]
 anthropic = "~0.23.1"
 authlib = "1.3.1"
+azure-ai-inference = "~1.0.0b3"
+azure-ai-ml = "~1.20.0"
 azure-identity = "1.16.1"
 azure-storage-blob = "12.13.0"
 beautifulsoup4 = "4.12.2"
 boto3 = "1.35.17"
-sagemaker = "2.231.0"
 bs4 = "~0.0.1"
 cachetools = "~5.3.0"
 celery = "~5.3.6"
 chardet = "~5.1.0"
 cohere = "~5.2.4"
-cos-python-sdk-v5 = "1.9.30"
-esdk-obs-python = "3.24.6.1"
-bce-python-sdk = "~0.9.23"
 dashscope = { version = "~1.17.0", extras = ["tokenizer"] }
 flask = "~3.0.1"
 flask-compress = "~1.14"
@@ -130,7 +128,7 @@ flask-cors = "~4.0.0"
 flask-login = "~0.6.3"
 flask-migrate = "~4.0.5"
 flask-restful = "~0.3.10"
-Flask-SQLAlchemy = "~3.1.1"
+flask-sqlalchemy = "~3.1.1"
 gevent = "~23.9.1"
 gmpy2 = "~2.2.1"
 google-ai-generativelanguage = "0.6.9"
@@ -139,22 +137,22 @@ google-api-python-client = "2.90.0"
 google-auth = "2.29.0"
 google-auth-httplib2 = "0.2.0"
 google-cloud-aiplatform = "1.49.0"
-google-cloud-storage = "2.16.0"
 google-generativeai = "0.8.1"
 googleapis-common-protos = "1.63.0"
 gunicorn = "~22.0.0"
 httpx = { version = "~0.27.0", extras = ["socks"] }
 huggingface-hub = "~0.16.4"
 jieba = "0.42.1"
-langfuse = "^2.48.0"
-langsmith = "^0.1.77"
+langfuse = "~2.51.3"
+langsmith = "~0.1.77"
 mailchimp-transactional = "~1.0.50"
 markdown = "~3.5.1"
-novita-client = "^0.5.7"
+nomic = "~3.1.2"
+novita-client = "~0.5.7"
 numpy = "~1.26.4"
+oci = "~2.135.1"
 openai = "~1.29.0"
 openpyxl = "~3.1.5"
-oss2 = "2.18.5"
 pandas = { version = "~2.2.2", extras = ["performance", "excel"] }
 psycopg2-binary = "~2.9.6"
 pycryptodome = "3.19.1"
@@ -171,7 +169,8 @@ readabilipy = "0.2.0"
 redis = { version = "~5.0.3", extras = ["hiredis"] }
 replicate = "~0.22.0"
 resend = "~0.7.0"
-scikit-learn = "^1.5.1"
+sagemaker = "2.231.0"
+scikit-learn = "~1.5.1"
 sentry-sdk = { version = "~1.44.1", extras = ["flask"] }
 sqlalchemy = "~2.0.29"
 tencentcloud-sdk-python-hunyuan = "~3.0.1158"
@@ -179,6 +178,8 @@ tiktoken = "~0.7.0"
 tokenizers = "~0.15.0"
 transformers = "~4.35.0"
 unstructured = { version = "~0.10.27", extras = ["docx", "epub", "md", "msg", "ppt", "pptx"] }
+validators = "0.21.0"
+volcengine-python-sdk = {extras = ["ark"], version = "~1.0.98"}
 websocket-client = "~1.7.0"
 werkzeug = "~3.0.1"
 xinference-client = "0.15.2"
@@ -187,32 +188,26 @@ zhipuai = "1.0.7"
 # Before adding new dependency, consider place it in alphabet order (a-z) and suitable group.
 
 ############################################################
+# [ Indirect ] dependency group
 # Related transparent dependencies with pinned version
 # required by main implementations
 ############################################################
-azure-ai-ml = "^1.19.0"
-azure-ai-inference = "^1.0.0b3"
-volcengine-python-sdk = {extras = ["ark"], version = "^1.0.98"}
-oci = "^2.133.0"
-tos = "^2.7.1"
-nomic = "^3.1.2"
-validators = "0.21.0"
-[tool.poetry.group.indriect.dependencies]
+[tool.poetry.group.indirect.dependencies]
 kaleido = "0.2.1"
 rank-bm25 = "~0.2.2"
 safetensors = "~0.4.3"
 
 ############################################################
-# Tool dependencies required by tool implementations
+# [ Tools ] dependency group
 ############################################################
-
-[tool.poetry.group.tool.dependencies]
+[tool.poetry.group.tools.dependencies]
 arxiv = "2.1.0"
 cloudscraper = "1.2.71"
-matplotlib = "~3.8.2"
-newspaper3k = "0.2.8"
 duckduckgo-search = "~6.3.0"
 jsonpath-ng = "1.6.1"
+matplotlib = "~3.8.2"
+newspaper3k = "0.2.8"
+nltk = "3.8.1"
 numexpr = "~2.9.0"
 opensearch-py = "2.4.0"
 qrcode = "~7.4.2"
@@ -220,11 +215,23 @@ twilio = "~9.0.4"
 vanna = { version = "0.5.5", extras = ["postgres", "mysql", "clickhouse", "duckdb"] }
 wikipedia = "1.4.0"
 yfinance = "~0.2.40"
-nltk = "3.8.1"
+
 ############################################################
-# VDB dependencies required by vector store clients
+# [ Storage ] dependency group
+# Required for storage clients
 ############################################################
+[tool.poetry.group.storage.dependencies]
+bce-python-sdk = "~0.9.23"
+cos-python-sdk-v5 = "1.9.30"
+esdk-obs-python = "3.24.6.1"
+google-cloud-storage = "2.16.0"
+oss2 = "2.18.5"
+tos = "~2.7.1"
 
+############################################################
+# [ VDB ] dependency group
+# Required by vector store clients
+############################################################
 [tool.poetry.group.vdb.dependencies]
 alibabacloud_gpdb20160503 = "~3.8.0"
 alibabacloud_tea_openapi = "~0.3.9"
@@ -235,18 +242,17 @@ oracledb = "~2.2.1"
 pgvecto-rs = { version = "~0.2.1", extras = ['sqlalchemy'] }
 pgvector = "0.2.5"
 pymilvus = "~2.4.4"
+qdrant-client = "1.7.3"
 tcvectordb = "1.3.2"
 tidb-vector = "0.0.9"
-qdrant-client = "1.7.3"
 weaviate-client = "~3.21.0"
 
 ############################################################
-# Dev dependencies for running tests
+# [ Dev ] dependency group
+# Required for development and running tests
 ############################################################
-
 [tool.poetry.group.dev]
 optional = true
-
 [tool.poetry.group.dev.dependencies]
 coverage = "~7.2.4"
 pytest = "~8.3.2"
@@ -255,12 +261,11 @@ pytest-env = "~1.1.3"
 pytest-mock = "~3.14.0"
 
 ############################################################
-# Lint dependencies for code style linting
+# [ Lint ] dependency group
+# Required for code style linting
 ############################################################
-
 [tool.poetry.group.lint]
 optional = true
-
 [tool.poetry.group.lint.dependencies]
 dotenv-linter = "~0.5.0"
 ruff = "~0.6.9"
diff --git a/api/tests/artifact_tests/dependencies/__init__.py b/api/tests/artifact_tests/dependencies/__init__.py
new file mode 100644
index 0000000000000..e69de29bb2d1d
diff --git a/api/tests/artifact_tests/dependencies/test_dependencies_sorted.py b/api/tests/artifact_tests/dependencies/test_dependencies_sorted.py
new file mode 100644
index 0000000000000..518cee1a3ad2d
--- /dev/null
+++ b/api/tests/artifact_tests/dependencies/test_dependencies_sorted.py
@@ -0,0 +1,61 @@
+from typing import Any
+
+import toml
+
+ALL_DEPENDENCY_GROUP_NAMES = [
+    # default main group
+    "",
+    # required groups
+    "indirect",
+    "storage",
+    "tools",
+    "vdb",
+    # optional groups
+    "dev",
+    "lint",
+]
+
+
+def load_api_poetry_configs() -> dict[str, Any]:
+    pyproject_toml = toml.load("api/pyproject.toml")
+    return pyproject_toml.get("tool").get("poetry")
+
+
+def load_dependency_groups() -> dict[str, dict[str, dict[str, Any]]]:
+    poetry_configs = load_api_poetry_configs()
+    group_name_to_dependencies = {
+        group_name: (poetry_configs.get("group").get(group_name) if group_name else poetry_configs).get("dependencies")
+        for group_name in ALL_DEPENDENCY_GROUP_NAMES
+    }
+    return group_name_to_dependencies
+
+
+def test_group_dependencies_sorted():
+    for group_name, dependencies in load_dependency_groups().items():
+        dependency_names = list(dependencies.keys())
+        expected_dependency_names = sorted(set(dependency_names))
+        section = f"tool.poetry.group.{group_name}.dependencies" if group_name else "tool.poetry.dependencies"
+        assert expected_dependency_names == dependency_names, (
+            f"Dependencies in group {group_name} are not sorted. "
+            f"Check and fix [{section}] section in pyproject.toml file"
+        )
+
+
+def test_group_dependencies_version_operator():
+    for group_name, dependencies in load_dependency_groups().items():
+        for dependency_name, specification in dependencies.items():
+            version_spec = specification if isinstance(specification, str) else specification.get("version")
+            assert not version_spec.startswith("^"), (
+                f"'^' is not allowed in dependency version," f" but found in '{dependency_name} = {version_spec}'"
+            )
+
+
+def test_duplicated_dependency_crossing_groups():
+    all_dependency_names: list[str] = []
+    for dependencies in load_dependency_groups().values():
+        dependency_names = list(dependencies.keys())
+        all_dependency_names.extend(dependency_names)
+    expected_all_dependency_names = set(all_dependency_names)
+    assert sorted(expected_all_dependency_names) == sorted(
+        all_dependency_names
+    ), "Duplicated dependencies crossing groups are found"
diff --git a/dev/pytest/pytest_artifacts.sh b/dev/pytest/pytest_artifacts.sh
new file mode 100755
index 0000000000000..d52acb2273846
--- /dev/null
+++ b/dev/pytest/pytest_artifacts.sh
@@ -0,0 +1,4 @@
+#!/bin/bash
+set -x
+
+pytest api/tests/artifact_tests/
diff --git a/dev/sync-poetry b/dev/sync-poetry
index 2dd4dd4fc3109..23d5d79e900aa 100755
--- a/dev/sync-poetry
+++ b/dev/sync-poetry
@@ -11,5 +11,8 @@ poetry check -C api --lock
 if [ $? -ne 0 ]; then
     # update poetry.lock
     # refreshing lockfile only without updating locked versions
+    echo "poetry.lock is outdated, refreshing without updating locked versions ..."
     poetry lock -C api --no-update
+else
+  echo "poetry.lock is ready."
 fi

From 966e65bb66207b7cc538719d34c669019321e63d Mon Sep 17 00:00:00 2001
From: =?UTF-8?q?=E9=9D=9E=E6=B3=95=E6=93=8D=E4=BD=9C?= <hjlarry@163.com>
Date: Tue, 8 Oct 2024 15:36:31 +0800
Subject: [PATCH 152/235] fix: zhipu ai web_search not work (#9058)

---
 .../model_runtime/model_providers/zhipuai/llm/llm.py   | 10 ++++++++++
 1 file changed, 10 insertions(+)

diff --git a/api/core/model_runtime/model_providers/zhipuai/llm/llm.py b/api/core/model_runtime/model_providers/zhipuai/llm/llm.py
index ea331701abb78..e0c4980523075 100644
--- a/api/core/model_runtime/model_providers/zhipuai/llm/llm.py
+++ b/api/core/model_runtime/model_providers/zhipuai/llm/llm.py
@@ -223,6 +223,16 @@ def _generate(
                 else:
                     new_prompt_messages.append(copy_prompt_message)
 
+        # zhipuai moved web_search param to tools
+        if "web_search" in model_parameters:
+            enable_web_search = model_parameters.get("web_search")
+            model_parameters.pop("web_search")
+            web_search_params = {"type": "web_search", "web_search": {"enable": enable_web_search}}
+            if "tools" in model_parameters:
+                model_parameters["tools"].append(web_search_params)
+            else:
+                model_parameters["tools"] = [web_search_params]
+
         if model in {"glm-4v", "glm-4v-plus"}:
             params = self._construct_glm_4v_parameter(model, new_prompt_messages, model_parameters)
         else:

From 0564e8a284314e6f4664fa718260446743da56d6 Mon Sep 17 00:00:00 2001
From: Sota Oizumi <60801230+sotazum@users.noreply.github.com>
Date: Tue, 8 Oct 2024 16:39:47 +0900
Subject: [PATCH 153/235] docs: Add Google Cloud Terraform to README (#9065)

---
 README.md    | 6 +++++-
 README_AR.md | 5 ++++-
 README_CN.md | 6 +++++-
 README_ES.md | 5 ++++-
 README_FR.md | 5 ++++-
 README_JA.md | 7 +++++--
 README_KL.md | 7 +++++--
 README_KR.md | 7 +++++--
 README_TR.md | 8 ++++++--
 README_VI.md | 6 +++++-
 10 files changed, 48 insertions(+), 14 deletions(-)

diff --git a/README.md b/README.md
index 1c49c415fe09a..ac4a4e5839352 100644
--- a/README.md
+++ b/README.md
@@ -196,10 +196,14 @@ If you'd like to configure a highly-available setup, there are community-contrib
 
 #### Using Terraform for Deployment
 
+Deploy Dify to Cloud Platform with a single click using [terraform](https://www.terraform.io/)
+
 ##### Azure Global
-Deploy Dify to Azure with a single click using [terraform](https://www.terraform.io/).
 - [Azure Terraform by @nikawang](https://github.com/nikawang/dify-azure-terraform)
 
+##### Google Cloud
+- [Google Cloud Terraform by @sotazum](https://github.com/DeNA/dify-google-cloud-terraform)
+
 ## Contributing
 
 For those who'd like to contribute code, see our [Contribution Guide](https://github.com/langgenius/dify/blob/main/CONTRIBUTING.md). 
diff --git a/README_AR.md b/README_AR.md
index 10d572cc49a83..42dbd9edff7b0 100644
--- a/README_AR.md
+++ b/README_AR.md
@@ -179,10 +179,13 @@ docker compose up -d
 
 #### استخدام Terraform للتوزيع
 
+انشر Dify إلى منصة السحابة بنقرة واحدة باستخدام [terraform](https://www.terraform.io/)
+
 ##### Azure Global
-استخدم [terraform](https://www.terraform.io/) لنشر Dify على Azure بنقرة واحدة.
 - [Azure Terraform بواسطة @nikawang](https://github.com/nikawang/dify-azure-terraform)
 
+##### Google Cloud
+- [Google Cloud Terraform بواسطة @sotazum](https://github.com/DeNA/dify-google-cloud-terraform)
 
 ## المساهمة
 
diff --git a/README_CN.md b/README_CN.md
index 32551fcc31393..afeca3147b37b 100644
--- a/README_CN.md
+++ b/README_CN.md
@@ -202,10 +202,14 @@ docker compose up -d
 
 #### 使用 Terraform 部署
 
+使用 [terraform](https://www.terraform.io/) 一键将 Dify 部署到云平台
+
 ##### Azure Global
-使用 [terraform](https://www.terraform.io/) 一键部署 Dify 到 Azure。
 - [Azure Terraform by @nikawang](https://github.com/nikawang/dify-azure-terraform)
 
+##### Google Cloud
+- [Google Cloud Terraform by @sotazum](https://github.com/DeNA/dify-google-cloud-terraform)
+
 ## Star History
 
 [![Star History Chart](https://api.star-history.com/svg?repos=langgenius/dify&type=Date)](https://star-history.com/#langgenius/dify&Date)
diff --git a/README_ES.md b/README_ES.md
index 2ae044b32883b..a6296aa93d7f9 100644
--- a/README_ES.md
+++ b/README_ES.md
@@ -204,10 +204,13 @@ Si desea configurar una configuración de alta disponibilidad, la comunidad prop
 
 #### Uso de Terraform para el despliegue
 
+Despliega Dify en una plataforma en la nube con un solo clic utilizando [terraform](https://www.terraform.io/)
+
 ##### Azure Global
-Utiliza [terraform](https://www.terraform.io/) para desplegar Dify en Azure con un solo clic.
 - [Azure Terraform por @nikawang](https://github.com/nikawang/dify-azure-terraform)
 
+##### Google Cloud
+- [Google Cloud Terraform por @sotazum](https://github.com/DeNA/dify-google-cloud-terraform)
 
 ## Contribuir
 
diff --git a/README_FR.md b/README_FR.md
index 681d596749c9e..da8b7baa3e725 100644
--- a/README_FR.md
+++ b/README_FR.md
@@ -202,10 +202,13 @@ Si vous souhaitez configurer une configuration haute disponibilité, la communau
 
 #### Utilisation de Terraform pour le déploiement
 
+Déployez Dify sur une plateforme cloud en un clic en utilisant [terraform](https://www.terraform.io/)
+
 ##### Azure Global
-Utilisez [terraform](https://www.terraform.io/) pour déployer Dify sur Azure en un clic.
 - [Azure Terraform par @nikawang](https://github.com/nikawang/dify-azure-terraform)
 
+##### Google Cloud
+- [Google Cloud Terraform par @sotazum](https://github.com/DeNA/dify-google-cloud-terraform)
 
 ## Contribuer
 
diff --git a/README_JA.md b/README_JA.md
index 98966d6c38447..8c6076362bec9 100644
--- a/README_JA.md
+++ b/README_JA.md
@@ -201,10 +201,13 @@ docker compose up -d
 
 #### Terraformを使用したデプロイ
 
+[terraform](https://www.terraform.io/) を使用して、ワンクリックでDifyをクラウドプラットフォームにデプロイします
+
 ##### Azure Global
-[terraform](https://www.terraform.io/) を使用して、AzureにDifyをワンクリックでデプロイします。
-- [nikawangのAzure Terraform](https://github.com/nikawang/dify-azure-terraform)
+- [@nikawangによるAzure Terraform](https://github.com/nikawang/dify-azure-terraform)
 
+##### Google Cloud
+- [@sotazumによるGoogle Cloud Terraform](https://github.com/DeNA/dify-google-cloud-terraform)
 
 ## 貢献
 
diff --git a/README_KL.md b/README_KL.md
index 04620d42bbec8..ce2f78570bdc9 100644
--- a/README_KL.md
+++ b/README_KL.md
@@ -202,10 +202,13 @@ If you'd like to configure a highly-available setup, there are community-contrib
 
 #### Terraform atorlugu pilersitsineq
 
+wa'logh nIqHom neH ghun deployment toy'wI' [terraform](https://www.terraform.io/) lo'laH.
+
 ##### Azure Global
-Atoruk [terraform](https://www.terraform.io/) Dify-mik Azure-mut ataatsikkut ikkussuilluarlugu.
-- [Azure Terraform atorlugu @nikawang](https://github.com/nikawang/dify-azure-terraform)
+- [Azure Terraform mung @nikawang](https://github.com/nikawang/dify-azure-terraform)
 
+##### Google Cloud
+- [Google Cloud Terraform qachlot @sotazum](https://github.com/DeNA/dify-google-cloud-terraform)
 
 ## Contributing
 
diff --git a/README_KR.md b/README_KR.md
index a5f3bc68d04d7..87409bfb8c29d 100644
--- a/README_KR.md
+++ b/README_KR.md
@@ -39,7 +39,6 @@
   <a href="./README_AR.md"><img alt="README بالعربية" src="https://img.shields.io/badge/العربية-d9d9d9"></a>
   <a href="./README_TR.md"><img alt="Türkçe README" src="https://img.shields.io/badge/Türkçe-d9d9d9"></a>
   <a href="./README_VI.md"><img alt="README Tiếng Việt" src="https://img.shields.io/badge/Ti%E1%BA%BFng%20Vi%E1%BB%87t-d9d9d9"></a>
-
 </p>
 
 
@@ -195,10 +194,14 @@ Dify를 Kubernetes에 배포하고 프리미엄 스케일링 설정을 구성했
 
 #### Terraform을 사용한 배포
 
+[terraform](https://www.terraform.io/)을 사용하여 단 한 번의 클릭으로 Dify를 클라우드 플랫폼에 배포하십시오
+
 ##### Azure Global
-[terraform](https://www.terraform.io/)을 사용하여 Azure에 Dify를 원클릭으로 배포하세요.
 - [nikawang의 Azure Terraform](https://github.com/nikawang/dify-azure-terraform)
 
+##### Google Cloud
+- [sotazum의 Google Cloud Terraform](https://github.com/DeNA/dify-google-cloud-terraform)
+
 ## 기여
 
 코드에 기여하고 싶은 분들은 [기여 가이드](https://github.com/langgenius/dify/blob/main/CONTRIBUTING.md)를 참조하세요.
diff --git a/README_TR.md b/README_TR.md
index 54b6db3f82371..08080a8c7bf75 100644
--- a/README_TR.md
+++ b/README_TR.md
@@ -200,9 +200,13 @@ Yüksek kullanılabilirliğe sahip bir kurulum yapılandırmak isterseniz, Dify'
 
 #### Dağıtım için Terraform Kullanımı
 
+Dify'ı bulut platformuna tek tıklamayla dağıtın [terraform](https://www.terraform.io/) kullanarak
+
 ##### Azure Global
-[Terraform](https://www.terraform.io/) kullanarak Dify'ı Azure'a tek tıklamayla dağıtın.
-- [@nikawang tarafından Azure Terraform](https://github.com/nikawang/dify-azure-terraform)
+- [Azure Terraform tarafından @nikawang](https://github.com/nikawang/dify-azure-terraform)
+
+##### Google Cloud
+- [Google Cloud Terraform tarafından @sotazum](https://github.com/DeNA/dify-google-cloud-terraform)
 
 ## Katkıda Bulunma
 
diff --git a/README_VI.md b/README_VI.md
index 6d4035eceb06d..d67f024c7955b 100644
--- a/README_VI.md
+++ b/README_VI.md
@@ -196,10 +196,14 @@ Nếu bạn muốn cấu hình một cài đặt có độ sẵn sàng cao, có
 
 #### Sử dụng Terraform để Triển khai
 
+Triển khai Dify lên nền tảng đám mây với một cú nhấp chuột bằng cách sử dụng [terraform](https://www.terraform.io/)
+
 ##### Azure Global
-Triển khai Dify lên Azure chỉ với một cú nhấp chuột bằng cách sử dụng [terraform](https://www.terraform.io/).
 - [Azure Terraform bởi @nikawang](https://github.com/nikawang/dify-azure-terraform)
 
+##### Google Cloud
+- [Google Cloud Terraform bởi @sotazum](https://github.com/DeNA/dify-google-cloud-terraform)
+
 ## Đóng góp
 
 Đối với những người muốn đóng góp mã, xem [Hướng dẫn Đóng góp](https://github.com/langgenius/dify/blob/main/CONTRIBUTING.md) của chúng tôi. 

From e741ee2f45f8d60e789ceb5bec05f25ebccaf124 Mon Sep 17 00:00:00 2001
From: Infinitnet <6189915+infinitnet@users.noreply.github.com>
Date: Tue, 8 Oct 2024 10:06:47 +0200
Subject: [PATCH 154/235] Correct max_tokens for OpenRouter Sonnet 3.5 (#9068)

---
 .../model_providers/openrouter/llm/claude-3-5-sonnet.yaml     | 4 ++--
 1 file changed, 2 insertions(+), 2 deletions(-)

diff --git a/api/core/model_runtime/model_providers/openrouter/llm/claude-3-5-sonnet.yaml b/api/core/model_runtime/model_providers/openrouter/llm/claude-3-5-sonnet.yaml
index 40558854e2a7b..e829048e55e0d 100644
--- a/api/core/model_runtime/model_providers/openrouter/llm/claude-3-5-sonnet.yaml
+++ b/api/core/model_runtime/model_providers/openrouter/llm/claude-3-5-sonnet.yaml
@@ -27,9 +27,9 @@ parameter_rules:
   - name: max_tokens
     use_template: max_tokens
     required: true
-    default: 4096
+    default: 8192
     min: 1
-    max: 4096
+    max: 8192
   - name: response_format
     use_template: response_format
 pricing:

From 8204e0e14adbacff155855fcb810488dcc9c84bb Mon Sep 17 00:00:00 2001
From: gaocarri <gaocarri@gmail.com>
Date: Tue, 8 Oct 2024 17:25:33 +0800
Subject: [PATCH 155/235] fix: Count exception occurs when searching
 conversations (#8754)

Co-authored-by: zheng.gao <zheng.gao@amh-group.com>
---
 api/controllers/console/app/conversation.py | 1 +
 1 file changed, 1 insertion(+)

diff --git a/api/controllers/console/app/conversation.py b/api/controllers/console/app/conversation.py
index df7bd352af296..c1e16b3b9b65a 100644
--- a/api/controllers/console/app/conversation.py
+++ b/api/controllers/console/app/conversation.py
@@ -188,6 +188,7 @@ def get(self, app_model):
                         subquery.c.from_end_user_session_id.ilike(keyword_filter),
                     ),
                 )
+                .group_by(Conversation.id)
             )
 
         account = current_user

From 5213650fed92e3b693dc997534d406ce47c8e7e5 Mon Sep 17 00:00:00 2001
From: zhuhao <37029601+hwzhuhao@users.noreply.github.com>
Date: Tue, 8 Oct 2024 19:07:53 +0800
Subject: [PATCH 156/235] feat: update the xinf tool's API key to optional
 (#9073)

---
 .../builtin/xinference/tools/stable_diffusion.py  | 15 +++++++++------
 .../provider/builtin/xinference/xinference.py     | 14 ++++++++++----
 .../provider/builtin/xinference/xinference.yaml   |  2 +-
 3 files changed, 20 insertions(+), 11 deletions(-)

diff --git a/api/core/tools/provider/builtin/xinference/tools/stable_diffusion.py b/api/core/tools/provider/builtin/xinference/tools/stable_diffusion.py
index 847f2730f240c..a44d3b730a84f 100644
--- a/api/core/tools/provider/builtin/xinference/tools/stable_diffusion.py
+++ b/api/core/tools/provider/builtin/xinference/tools/stable_diffusion.py
@@ -104,14 +104,15 @@ def _invoke(
         model = self.runtime.credentials.get("model", None)
         if not model:
             return self.create_text_message("Please input model")
-
+        api_key = self.runtime.credentials.get("api_key") or "abc"
+        headers = {"Authorization": f"Bearer {api_key}"}
         # set model
         try:
             url = str(URL(base_url) / "sdapi" / "v1" / "options")
             response = post(
                 url,
                 json={"sd_model_checkpoint": model},
-                headers={"Authorization": f"Bearer {self.runtime.credentials['api_key']}"},
+                headers=headers,
             )
             if response.status_code != 200:
                 raise ToolProviderCredentialValidationError("Failed to set model, please tell user to set model")
@@ -257,14 +258,15 @@ def img2img(
             draw_options["prompt"] = f"{lora},{prompt}"
         else:
             draw_options["prompt"] = prompt
-
+        api_key = self.runtime.credentials.get("api_key") or "abc"
+        headers = {"Authorization": f"Bearer {api_key}"}
         try:
             url = str(URL(base_url) / "sdapi" / "v1" / "img2img")
             response = post(
                 url,
                 json=draw_options,
                 timeout=120,
-                headers={"Authorization": f"Bearer {self.runtime.credentials['api_key']}"},
+                headers=headers,
             )
             if response.status_code != 200:
                 return self.create_text_message("Failed to generate image")
@@ -298,14 +300,15 @@ def text2img(
         else:
             draw_options["prompt"] = prompt
         draw_options["override_settings"]["sd_model_checkpoint"] = model
-
+        api_key = self.runtime.credentials.get("api_key") or "abc"
+        headers = {"Authorization": f"Bearer {api_key}"}
         try:
             url = str(URL(base_url) / "sdapi" / "v1" / "txt2img")
             response = post(
                 url,
                 json=draw_options,
                 timeout=120,
-                headers={"Authorization": f"Bearer {self.runtime.credentials['api_key']}"},
+                headers=headers,
             )
             if response.status_code != 200:
                 return self.create_text_message("Failed to generate image")
diff --git a/api/core/tools/provider/builtin/xinference/xinference.py b/api/core/tools/provider/builtin/xinference/xinference.py
index 7c2428cc00534..9692e4060e8a8 100644
--- a/api/core/tools/provider/builtin/xinference/xinference.py
+++ b/api/core/tools/provider/builtin/xinference/xinference.py
@@ -6,12 +6,18 @@
 
 class XinferenceProvider(BuiltinToolProviderController):
     def _validate_credentials(self, credentials: dict) -> None:
-        base_url = credentials.get("base_url")
-        api_key = credentials.get("api_key")
-        model = credentials.get("model")
+        base_url = credentials.get("base_url", "").removesuffix("/")
+        api_key = credentials.get("api_key", "")
+        if not api_key:
+            api_key = "abc"
+            credentials["api_key"] = api_key
+        model = credentials.get("model", "")
+        if not base_url or not model:
+            raise ToolProviderCredentialValidationError("Xinference base_url and model is required")
+        headers = {"Authorization": f"Bearer {api_key}"}
         res = requests.post(
             f"{base_url}/sdapi/v1/options",
-            headers={"Authorization": f"Bearer {api_key}"},
+            headers=headers,
             json={"sd_model_checkpoint": model},
         )
         if res.status_code != 200:
diff --git a/api/core/tools/provider/builtin/xinference/xinference.yaml b/api/core/tools/provider/builtin/xinference/xinference.yaml
index 19aaf5cbd1398..b0c02b9cbcb01 100644
--- a/api/core/tools/provider/builtin/xinference/xinference.yaml
+++ b/api/core/tools/provider/builtin/xinference/xinference.yaml
@@ -31,7 +31,7 @@ credentials_for_provider:
       zh_Hans: 请输入你的模型名称
   api_key:
     type: secret-input
-    required: true
+    required: false
     label:
       en_US: API Key
       zh_Hans: Xinference 服务器的 API Key

From 7121afdd4426648b99055b7041b5e99bc7b1ad3a Mon Sep 17 00:00:00 2001
From: Dongsheng Zhao <41723543+Blackman99@users.noreply.github.com>
Date: Tue, 8 Oct 2024 19:58:03 +0800
Subject: [PATCH 157/235] fix: #8969 (#9076)

---
 .../base/chat/embedded-chatbot/theme/theme-context.ts           | 2 +-
 1 file changed, 1 insertion(+), 1 deletion(-)

diff --git a/web/app/components/base/chat/embedded-chatbot/theme/theme-context.ts b/web/app/components/base/chat/embedded-chatbot/theme/theme-context.ts
index fd424cd046eea..7a814e252125c 100644
--- a/web/app/components/base/chat/embedded-chatbot/theme/theme-context.ts
+++ b/web/app/components/base/chat/embedded-chatbot/theme/theme-context.ts
@@ -26,7 +26,7 @@ export class Theme {
     if (this.chatColorTheme !== null && this.chatColorTheme !== '') {
       this.primaryColor = this.chatColorTheme ?? '#1C64F2'
       this.backgroundHeaderColorStyle = `backgroundColor: ${this.primaryColor}`
-      this.backgroundButtonDefaultColorStyle = `backgroundColor: ${this.primaryColor}`
+      this.backgroundButtonDefaultColorStyle = `backgroundColor: ${this.primaryColor}; color: ${colorFontOnHeaderStyle};`
       this.roundedBackgroundColorStyle = `backgroundColor: ${hexToRGBA(this.primaryColor, 0.05)}`
       this.chatBubbleColorStyle = `backgroundColor: ${hexToRGBA(this.primaryColor, 0.15)}`
       this.chatBubbleColor = `${hexToRGBA(this.primaryColor, 0.15)}`

From 5b7de7705ebf1eba29f891ff2e6d7a22063a6284 Mon Sep 17 00:00:00 2001
From: zhuhao <37029601+hwzhuhao@users.noreply.github.com>
Date: Tue, 8 Oct 2024 23:25:11 +0800
Subject: [PATCH 158/235] feat: add timestamp conversion and timestamp
 retrieval for time tool (#9085)

---
 .../time/tools/localtime_to_timestamp.py      | 44 +++++++++++++++++++
 .../time/tools/localtime_to_timestamp.yaml    | 33 ++++++++++++++
 .../time/tools/timestamp_to_localtime.py      | 44 +++++++++++++++++++
 .../time/tools/timestamp_to_localtime.yaml    | 33 ++++++++++++++
 4 files changed, 154 insertions(+)
 create mode 100644 api/core/tools/provider/builtin/time/tools/localtime_to_timestamp.py
 create mode 100644 api/core/tools/provider/builtin/time/tools/localtime_to_timestamp.yaml
 create mode 100644 api/core/tools/provider/builtin/time/tools/timestamp_to_localtime.py
 create mode 100644 api/core/tools/provider/builtin/time/tools/timestamp_to_localtime.yaml

diff --git a/api/core/tools/provider/builtin/time/tools/localtime_to_timestamp.py b/api/core/tools/provider/builtin/time/tools/localtime_to_timestamp.py
new file mode 100644
index 0000000000000..e16b732d0242d
--- /dev/null
+++ b/api/core/tools/provider/builtin/time/tools/localtime_to_timestamp.py
@@ -0,0 +1,44 @@
+from datetime import datetime
+from typing import Any, Union
+
+import pytz
+
+from core.tools.entities.tool_entities import ToolInvokeMessage
+from core.tools.errors import ToolInvokeError
+from core.tools.tool.builtin_tool import BuiltinTool
+
+
+class LocaltimeToTimestampTool(BuiltinTool):
+    def _invoke(
+        self,
+        user_id: str,
+        tool_parameters: dict[str, Any],
+    ) -> Union[ToolInvokeMessage, list[ToolInvokeMessage]]:
+        """
+        Convert localtime to timestamp
+        """
+        localtime = tool_parameters.get("localtime")
+        timezone = tool_parameters.get("timezone", "Asia/Shanghai")
+        if not timezone:
+            timezone = None
+        time_format = "%Y-%m-%d %H:%M:%S"
+
+        timestamp = self.localtime_to_timestamp(localtime, time_format, timezone)
+        if not timestamp:
+            return self.create_text_message(f"Invalid localtime: {localtime}")
+
+        return self.create_text_message(f"{timestamp}")
+
+    @staticmethod
+    def localtime_to_timestamp(localtime: str, time_format: str, local_tz=None) -> int | None:
+        try:
+            if local_tz is None:
+                local_tz = datetime.now().astimezone().tzinfo
+            if isinstance(local_tz, str):
+                local_tz = pytz.timezone(local_tz)
+            local_time = datetime.strptime(localtime, time_format)
+            localtime = local_tz.localize(local_time)
+            timestamp = int(localtime.timestamp())
+            return timestamp
+        except Exception as e:
+            raise ToolInvokeError(str(e))
diff --git a/api/core/tools/provider/builtin/time/tools/localtime_to_timestamp.yaml b/api/core/tools/provider/builtin/time/tools/localtime_to_timestamp.yaml
new file mode 100644
index 0000000000000..6a3b90595fd3f
--- /dev/null
+++ b/api/core/tools/provider/builtin/time/tools/localtime_to_timestamp.yaml
@@ -0,0 +1,33 @@
+identity:
+  name: localtime_to_timestamp
+  author: zhuhao
+  label:
+    en_US: localtime to timestamp
+    zh_Hans: 获取时间戳
+description:
+  human:
+    en_US: A tool for localtime convert to timestamp
+    zh_Hans: 获取时间戳
+  llm: A tool for localtime convert to timestamp
+parameters:
+  - name: localtime
+    type: string
+    required: true
+    form: llm
+    label:
+      en_US: localtime
+      zh_Hans: 本地时间
+    human_description:
+      en_US: localtime, such as 2024-1-1 0:0:0
+      zh_Hans: 本地时间, 比如2024-1-1 0:0:0
+  - name: timezone
+    type: string
+    required: false
+    form: llm
+    label:
+      en_US: Timezone
+      zh_Hans: 时区
+    human_description:
+      en_US: Timezone, such as Asia/Shanghai
+      zh_Hans: 时区, 比如Asia/Shanghai
+    default: Asia/Shanghai
diff --git a/api/core/tools/provider/builtin/time/tools/timestamp_to_localtime.py b/api/core/tools/provider/builtin/time/tools/timestamp_to_localtime.py
new file mode 100644
index 0000000000000..bcdd34fd4ec54
--- /dev/null
+++ b/api/core/tools/provider/builtin/time/tools/timestamp_to_localtime.py
@@ -0,0 +1,44 @@
+from datetime import datetime
+from typing import Any, Union
+
+import pytz
+
+from core.tools.entities.tool_entities import ToolInvokeMessage
+from core.tools.errors import ToolInvokeError
+from core.tools.tool.builtin_tool import BuiltinTool
+
+
+class TimestampToLocaltimeTool(BuiltinTool):
+    def _invoke(
+        self,
+        user_id: str,
+        tool_parameters: dict[str, Any],
+    ) -> Union[ToolInvokeMessage, list[ToolInvokeMessage]]:
+        """
+        Convert timestamp to localtime
+        """
+        timestamp = tool_parameters.get("timestamp")
+        timezone = tool_parameters.get("timezone", "Asia/Shanghai")
+        if not timezone:
+            timezone = None
+        time_format = "%Y-%m-%d %H:%M:%S"
+
+        locatime = self.timestamp_to_localtime(timestamp, timezone)
+        if not locatime:
+            return self.create_text_message(f"Invalid timestamp: {timestamp}")
+
+        localtime_format = locatime.strftime(time_format)
+
+        return self.create_text_message(f"{localtime_format}")
+
+    @staticmethod
+    def timestamp_to_localtime(timestamp: int, local_tz=None) -> datetime | None:
+        try:
+            if local_tz is None:
+                local_tz = datetime.now().astimezone().tzinfo
+            if isinstance(local_tz, str):
+                local_tz = pytz.timezone(local_tz)
+            local_time = datetime.fromtimestamp(timestamp, local_tz)
+            return local_time
+        except Exception as e:
+            raise ToolInvokeError(str(e))
diff --git a/api/core/tools/provider/builtin/time/tools/timestamp_to_localtime.yaml b/api/core/tools/provider/builtin/time/tools/timestamp_to_localtime.yaml
new file mode 100644
index 0000000000000..3794e717b4dc8
--- /dev/null
+++ b/api/core/tools/provider/builtin/time/tools/timestamp_to_localtime.yaml
@@ -0,0 +1,33 @@
+identity:
+  name: timestamp_to_localtime
+  author: zhuhao
+  label:
+    en_US: Timestamp to localtime
+    zh_Hans: 时间戳转换
+description:
+  human:
+    en_US: A tool for timestamp convert to localtime
+    zh_Hans: 时间戳转换
+  llm: A tool for timestamp convert to localtime
+parameters:
+  - name: timestamp
+    type: number
+    required: true
+    form: llm
+    label:
+      en_US: Timestamp
+      zh_Hans: 时间戳
+    human_description:
+      en_US: Timestamp
+      zh_Hans: 时间戳
+  - name: timezone
+    type: string
+    required: false
+    form: llm
+    label:
+      en_US: Timezone
+      zh_Hans: 时区
+    human_description:
+      en_US: Timezone, such as Asia/Shanghai
+      zh_Hans: 时区, 比如Asia/Shanghai
+    default: Asia/Shanghai

From 3a0734d94cbcd0da4cc106006e5dcfbe2113c13c Mon Sep 17 00:00:00 2001
From: crazywoola <100913391+crazywoola@users.noreply.github.com>
Date: Wed, 9 Oct 2024 01:00:10 +0800
Subject: [PATCH 159/235] Feat/9081 add support for llamaguard through groq
 provider (#9083)

---
 .../model_providers/groq/llm/_position.yaml   |  1 +
 .../groq/llm/llama-guard-3-8b.yaml            | 25 +++++++++++++++++++
 2 files changed, 26 insertions(+)
 create mode 100644 api/core/model_runtime/model_providers/groq/llm/llama-guard-3-8b.yaml

diff --git a/api/core/model_runtime/model_providers/groq/llm/_position.yaml b/api/core/model_runtime/model_providers/groq/llm/_position.yaml
index be115ca920df0..0613b19f87ee5 100644
--- a/api/core/model_runtime/model_providers/groq/llm/_position.yaml
+++ b/api/core/model_runtime/model_providers/groq/llm/_position.yaml
@@ -5,3 +5,4 @@
 - llama3-8b-8192
 - mixtral-8x7b-32768
 - llama2-70b-4096
+- llama-guard-3-8b
diff --git a/api/core/model_runtime/model_providers/groq/llm/llama-guard-3-8b.yaml b/api/core/model_runtime/model_providers/groq/llm/llama-guard-3-8b.yaml
new file mode 100644
index 0000000000000..03779ccc66f63
--- /dev/null
+++ b/api/core/model_runtime/model_providers/groq/llm/llama-guard-3-8b.yaml
@@ -0,0 +1,25 @@
+model: llama-guard-3-8b
+label:
+  zh_Hans: Llama-Guard-3-8B
+  en_US: Llama-Guard-3-8B
+model_type: llm
+features:
+  - agent-thought
+model_properties:
+  mode: chat
+  context_size: 8192
+parameter_rules:
+  - name: temperature
+    use_template: temperature
+  - name: top_p
+    use_template: top_p
+  - name: max_tokens
+    use_template: max_tokens
+    default: 512
+    min: 1
+    max: 8192
+pricing:
+  input: '0.20'
+  output: '0.20'
+  unit: '0.000001'
+  currency: USD

From 0540995e5c9f14f9b5ad19ba794dc5f9a8671142 Mon Sep 17 00:00:00 2001
From: Pika <codingme+git@icloud.com>
Date: Wed, 9 Oct 2024 10:22:20 +0800
Subject: [PATCH 160/235] fix: prompt-editor regex.lastIndex needed to reset
 (#9097)

Co-authored-by: Chen(MAC) <chenchen404@outlook.com>
---
 ...kflow-variable-block-replacement-block.tsx |  5 +++--
 web/config/index.ts                           | 22 ++++++++++---------
 2 files changed, 15 insertions(+), 12 deletions(-)

diff --git a/web/app/components/base/prompt-editor/plugins/workflow-variable-block/workflow-variable-block-replacement-block.tsx b/web/app/components/base/prompt-editor/plugins/workflow-variable-block/workflow-variable-block-replacement-block.tsx
index c88b1661ad17f..0a80ecc220548 100644
--- a/web/app/components/base/prompt-editor/plugins/workflow-variable-block/workflow-variable-block-replacement-block.tsx
+++ b/web/app/components/base/prompt-editor/plugins/workflow-variable-block/workflow-variable-block-replacement-block.tsx
@@ -12,7 +12,7 @@ import type { WorkflowVariableBlockType } from '../../types'
 import { CustomTextNode } from '../custom-text/node'
 import { $createWorkflowVariableBlockNode } from './node'
 import { WorkflowVariableBlockNode } from './index'
-import { VAR_REGEX as REGEX } from '@/config'
+import { VAR_REGEX as REGEX, resetReg } from '@/config'
 
 const WorkflowVariableBlockReplacementBlock = ({
   workflowNodesMap,
@@ -48,11 +48,12 @@ const WorkflowVariableBlockReplacementBlock = ({
   }, [])
 
   const transformListener = useCallback((textNode: any) => {
+    resetReg()
     return decoratorTransform(textNode, getMatch, createWorkflowVariableBlockNode)
   }, [createWorkflowVariableBlockNode, getMatch])
 
   useEffect(() => {
-    REGEX.lastIndex = 0
+    resetReg()
     return mergeRegister(
       editor.registerNodeTransform(CustomTextNode, transformListener),
     )
diff --git a/web/config/index.ts b/web/config/index.ts
index 21fc2f211c9c3..9eb4889441bb4 100644
--- a/web/config/index.ts
+++ b/web/config/index.ts
@@ -158,28 +158,28 @@ export const DEFAULT_AGENT_SETTING = {
 }
 
 export const DEFAULT_AGENT_PROMPT = {
-  chat: `Respond to the human as helpfully and accurately as possible. 
+  chat: `Respond to the human as helpfully and accurately as possible.
 
   {{instruction}}
-  
+
   You have access to the following tools:
-  
+
   {{tools}}
-  
+
   Use a json blob to specify a tool by providing an {{TOOL_NAME_KEY}} key (tool name) and an {{ACTION_INPUT_KEY}} key (tool input).
   Valid "{{TOOL_NAME_KEY}}" values: "Final Answer" or {{tool_names}}
-  
+
   Provide only ONE action per $JSON_BLOB, as shown:
-  
+
   \`\`\`
   {
     "{{TOOL_NAME_KEY}}": $TOOL_NAME,
     "{{ACTION_INPUT_KEY}}": $ACTION_INPUT
   }
   \`\`\`
-  
+
   Follow this format:
-  
+
   Question: input question to answer
   Thought: consider previous and subsequent steps
   Action:
@@ -196,10 +196,10 @@ export const DEFAULT_AGENT_PROMPT = {
     "{{ACTION_INPUT_KEY}}": "Final response to human"
   }
   \`\`\`
-  
+
   Begin! Reminder to ALWAYS respond with a valid json blob of a single action. Use tools if necessary. Respond directly if appropriate. Format is Action:\`\`\`$JSON_BLOB\`\`\`then Observation:.`,
   completion: `
-  Respond to the human as helpfully and accurately as possible. 
+  Respond to the human as helpfully and accurately as possible.
 
 {{instruction}}
 
@@ -246,6 +246,8 @@ Thought: {{agent_scratchpad}}
 
 export const VAR_REGEX = /\{\{(#[a-zA-Z0-9_-]{1,50}(\.[a-zA-Z_][a-zA-Z0-9_]{0,29}){1,10}#)\}\}/gi
 
+export const resetReg = () => VAR_REGEX.lastIndex = 0
+
 export let textGenerationTimeoutMs = 60000
 
 if (process.env.NEXT_PUBLIC_TEXT_GENERATION_TIMEOUT_MS && process.env.NEXT_PUBLIC_TEXT_GENERATION_TIMEOUT_MS !== '')

From 57994e4a244c7087cf6a9e282c20e0edc784cef5 Mon Sep 17 00:00:00 2001
From: ybalbert001 <120714773+ybalbert001@users.noreply.github.com>
Date: Wed, 9 Oct 2024 10:23:21 +0800
Subject: [PATCH 161/235] fix s3 presign url check problem, support two
 versions(v2,v4) (#9093)

Co-authored-by: Yuanbo Li <ybalbert@amazon.com>
---
 api/core/file/message_file_parser.py | 36 +++++++++++++++++++++-------
 1 file changed, 27 insertions(+), 9 deletions(-)

diff --git a/api/core/file/message_file_parser.py b/api/core/file/message_file_parser.py
index 83059b216ee66..641686bd7cd13 100644
--- a/api/core/file/message_file_parser.py
+++ b/api/core/file/message_file_parser.py
@@ -198,16 +198,34 @@ def is_s3_presigned_url(url):
                     if "amazonaws.com" not in parsed_url.netloc:
                         return False
                     query_params = parse_qs(parsed_url.query)
-                    required_params = ["Signature", "Expires"]
-                    for param in required_params:
-                        if param not in query_params:
+
+                    def check_presign_v2(query_params):
+                        required_params = ["Signature", "Expires"]
+                        for param in required_params:
+                            if param not in query_params:
+                                return False
+                        if not query_params["Expires"][0].isdigit():
                             return False
-                    if not query_params["Expires"][0].isdigit():
-                        return False
-                    signature = query_params["Signature"][0]
-                    if not re.match(r"^[A-Za-z0-9+/]+={0,2}$", signature):
-                        return False
-                    return True
+                        signature = query_params["Signature"][0]
+                        if not re.match(r"^[A-Za-z0-9+/]+={0,2}$", signature):
+                            return False
+
+                        return True
+
+                    def check_presign_v4(query_params):
+                        required_params = ["X-Amz-Signature", "X-Amz-Expires"]
+                        for param in required_params:
+                            if param not in query_params:
+                                return False
+                        if not query_params["X-Amz-Expires"][0].isdigit():
+                            return False
+                        signature = query_params["X-Amz-Signature"][0]
+                        if not re.match(r"^[A-Za-z0-9+/]+={0,2}$", signature):
+                            return False
+
+                        return True
+
+                    return check_presign_v4(query_params) or check_presign_v2(query_params)
                 except Exception:
                     return False
 

From f8af9c6ad01587a133f1729eaaf0cf142f5ceabd Mon Sep 17 00:00:00 2001
From: SebastjanPrachovskij
 <86522260+SebastjanPrachovskij@users.noreply.github.com>
Date: Wed, 9 Oct 2024 05:24:09 +0300
Subject: [PATCH 162/235] refactor: Update countries & languages list for
 SearchApi engines (#9090)

---
 .../builtin/searchapi/tools/google.yaml       | 1526 ++++++++++++++++-
 .../builtin/searchapi/tools/google_jobs.yaml  | 1043 ++++++++++-
 .../builtin/searchapi/tools/google_news.yaml  | 1526 ++++++++++++++++-
 3 files changed, 3950 insertions(+), 145 deletions(-)

diff --git a/api/core/tools/provider/builtin/searchapi/tools/google.yaml b/api/core/tools/provider/builtin/searchapi/tools/google.yaml
index b69a0e1d3e706..0dc1b6672436c 100644
--- a/api/core/tools/provider/builtin/searchapi/tools/google.yaml
+++ b/api/core/tools/provider/builtin/searchapi/tools/google.yaml
@@ -65,206 +65,1206 @@ parameters:
     form: form
     default: US
     options:
+      - value: AF
+        label:
+          en_US: Afghanistan
+          zh_Hans: 阿富汗
+          pt_BR: Afeganistão
+      - value: AL
+        label:
+          en_US: Albania
+          zh_Hans: 阿尔巴尼亚
+          pt_BR: Albânia
+      - value: DZ
+        label:
+          en_US: Algeria
+          zh_Hans: 阿尔及利亚
+          pt_BR: Argélia
+      - value: AS
+        label:
+          en_US: American Samoa
+          zh_Hans: 美属萨摩亚
+          pt_BR: Samoa Americana
+      - value: AD
+        label:
+          en_US: Andorra
+          zh_Hans: 安道尔
+          pt_BR: Andorra
+      - value: AO
+        label:
+          en_US: Angola
+          zh_Hans: 安哥拉
+          pt_BR: Angola
+      - value: AI
+        label:
+          en_US: Anguilla
+          zh_Hans: 安圭拉
+          pt_BR: Anguilla
+      - value: AQ
+        label:
+          en_US: Antarctica
+          zh_Hans: 南极洲
+          pt_BR: Antártica
+      - value: AG
+        label:
+          en_US: Antigua and Barbuda
+          zh_Hans: 安提瓜和巴布达
+          pt_BR: Antígua e Barbuda
       - value: AR
         label:
           en_US: Argentina
           zh_Hans: 阿根廷
           pt_BR: Argentina
+      - value: AM
+        label:
+          en_US: Armenia
+          zh_Hans: 亚美尼亚
+          pt_BR: Armênia
+      - value: AW
+        label:
+          en_US: Aruba
+          zh_Hans: 阿鲁巴
+          pt_BR: Aruba
       - value: AU
         label:
           en_US: Australia
           zh_Hans: 澳大利亚
-          pt_BR: Australia
+          pt_BR: Austrália
       - value: AT
         label:
           en_US: Austria
           zh_Hans: 奥地利
-          pt_BR: Austria
+          pt_BR: Áustria
+      - value: AZ
+        label:
+          en_US: Azerbaijan
+          zh_Hans: 阿塞拜疆
+          pt_BR: Azerbaijão
+      - value: BS
+        label:
+          en_US: Bahamas
+          zh_Hans: 巴哈马
+          pt_BR: Bahamas
+      - value: BH
+        label:
+          en_US: Bahrain
+          zh_Hans: 巴林
+          pt_BR: Bahrein
+      - value: BD
+        label:
+          en_US: Bangladesh
+          zh_Hans: 孟加拉国
+          pt_BR: Bangladesh
+      - value: BB
+        label:
+          en_US: Barbados
+          zh_Hans: 巴巴多斯
+          pt_BR: Barbados
+      - value: BY
+        label:
+          en_US: Belarus
+          zh_Hans: 白俄罗斯
+          pt_BR: Bielorrússia
       - value: BE
         label:
           en_US: Belgium
           zh_Hans: 比利时
-          pt_BR: Belgium
+          pt_BR: Bélgica
+      - value: BZ
+        label:
+          en_US: Belize
+          zh_Hans: 伯利兹
+          pt_BR: Belize
+      - value: BJ
+        label:
+          en_US: Benin
+          zh_Hans: 贝宁
+          pt_BR: Benim
+      - value: BM
+        label:
+          en_US: Bermuda
+          zh_Hans: 百慕大
+          pt_BR: Bermudas
+      - value: BT
+        label:
+          en_US: Bhutan
+          zh_Hans: 不丹
+          pt_BR: Butão
+      - value: BO
+        label:
+          en_US: Bolivia
+          zh_Hans: 玻利维亚
+          pt_BR: Bolívia
+      - value: BA
+        label:
+          en_US: Bosnia and Herzegovina
+          zh_Hans: 波斯尼亚和黑塞哥维那
+          pt_BR: Bósnia e Herzegovina
+      - value: BW
+        label:
+          en_US: Botswana
+          zh_Hans: 博茨瓦纳
+          pt_BR: Botsuana
+      - value: BV
+        label:
+          en_US: Bouvet Island
+          zh_Hans: 布韦岛
+          pt_BR: Ilha Bouvet
       - value: BR
         label:
           en_US: Brazil
           zh_Hans: 巴西
-          pt_BR: Brazil
+          pt_BR: Brasil
+      - value: IO
+        label:
+          en_US: British Indian Ocean Territory
+          zh_Hans: 英属印度洋领地
+          pt_BR: Território Britânico do Oceano Índico
+      - value: BN
+        label:
+          en_US: Brunei Darussalam
+          zh_Hans: 文莱
+          pt_BR: Brunei Darussalam
+      - value: BG
+        label:
+          en_US: Bulgaria
+          zh_Hans: 保加利亚
+          pt_BR: Bulgária
+      - value: BF
+        label:
+          en_US: Burkina Faso
+          zh_Hans: 布基纳法索
+          pt_BR: Burkina Faso
+      - value: BI
+        label:
+          en_US: Burundi
+          zh_Hans: 布隆迪
+          pt_BR: Burundi
+      - value: KH
+        label:
+          en_US: Cambodia
+          zh_Hans: 柬埔寨
+          pt_BR: Camboja
+      - value: CM
+        label:
+          en_US: Cameroon
+          zh_Hans: 喀麦隆
+          pt_BR: Camarões
       - value: CA
         label:
           en_US: Canada
           zh_Hans: 加拿大
-          pt_BR: Canada
+          pt_BR: Canadá
+      - value: CV
+        label:
+          en_US: Cape Verde
+          zh_Hans: 佛得角
+          pt_BR: Cabo Verde
+      - value: KY
+        label:
+          en_US: Cayman Islands
+          zh_Hans: 开曼群岛
+          pt_BR: Ilhas Cayman
+      - value: CF
+        label:
+          en_US: Central African Republic
+          zh_Hans: 中非共和国
+          pt_BR: República Centro-Africana
+      - value: TD
+        label:
+          en_US: Chad
+          zh_Hans: 乍得
+          pt_BR: Chade
       - value: CL
         label:
           en_US: Chile
           zh_Hans: 智利
           pt_BR: Chile
-      - value: CO
-        label:
-          en_US: Colombia
-          zh_Hans: 哥伦比亚
-          pt_BR: Colombia
       - value: CN
         label:
           en_US: China
           zh_Hans: 中国
           pt_BR: China
+      - value: CX
+        label:
+          en_US: Christmas Island
+          zh_Hans: 圣诞岛
+          pt_BR: Ilha do Natal
+      - value: CC
+        label:
+          en_US: Cocos (Keeling) Islands
+          zh_Hans: 科科斯（基林）群岛
+          pt_BR: Ilhas Cocos (Keeling)
+      - value: CO
+        label:
+          en_US: Colombia
+          zh_Hans: 哥伦比亚
+          pt_BR: Colômbia
+      - value: KM
+        label:
+          en_US: Comoros
+          zh_Hans: 科摩罗
+          pt_BR: Comores
+      - value: CG
+        label:
+          en_US: Congo
+          zh_Hans: 刚果
+          pt_BR: Congo
+      - value: CD
+        label:
+          en_US: Congo, the Democratic Republic of the
+          zh_Hans: 刚果民主共和国
+          pt_BR: Congo, República Democrática do
+      - value: CK
+        label:
+          en_US: Cook Islands
+          zh_Hans: 库克群岛
+          pt_BR: Ilhas Cook
+      - value: CR
+        label:
+          en_US: Costa Rica
+          zh_Hans: 哥斯达黎加
+          pt_BR: Costa Rica
+      - value: CI
+        label:
+          en_US: Cote D'ivoire
+          zh_Hans: 科特迪瓦
+          pt_BR: Costa do Marfim
+      - value: HR
+        label:
+          en_US: Croatia
+          zh_Hans: 克罗地亚
+          pt_BR: Croácia
+      - value: CU
+        label:
+          en_US: Cuba
+          zh_Hans: 古巴
+          pt_BR: Cuba
+      - value: CY
+        label:
+          en_US: Cyprus
+          zh_Hans: 塞浦路斯
+          pt_BR: Chipre
       - value: CZ
         label:
           en_US: Czech Republic
           zh_Hans: 捷克共和国
-          pt_BR: Czech Republic
+          pt_BR: República Tcheca
       - value: DK
         label:
           en_US: Denmark
           zh_Hans: 丹麦
-          pt_BR: Denmark
+          pt_BR: Dinamarca
+      - value: DJ
+        label:
+          en_US: Djibouti
+          zh_Hans: 吉布提
+          pt_BR: Djibuti
+      - value: DM
+        label:
+          en_US: Dominica
+          zh_Hans: 多米尼克
+          pt_BR: Dominica
+      - value: DO
+        label:
+          en_US: Dominican Republic
+          zh_Hans: 多米尼加共和国
+          pt_BR: República Dominicana
+      - value: EC
+        label:
+          en_US: Ecuador
+          zh_Hans: 厄瓜多尔
+          pt_BR: Equador
+      - value: EG
+        label:
+          en_US: Egypt
+          zh_Hans: 埃及
+          pt_BR: Egito
+      - value: SV
+        label:
+          en_US: El Salvador
+          zh_Hans: 萨尔瓦多
+          pt_BR: El Salvador
+      - value: GQ
+        label:
+          en_US: Equatorial Guinea
+          zh_Hans: 赤道几内亚
+          pt_BR: Guiné Equatorial
+      - value: ER
+        label:
+          en_US: Eritrea
+          zh_Hans: 厄立特里亚
+          pt_BR: Eritreia
+      - value: EE
+        label:
+          en_US: Estonia
+          zh_Hans: 爱沙尼亚
+          pt_BR: Estônia
+      - value: ET
+        label:
+          en_US: Ethiopia
+          zh_Hans: 埃塞俄比亚
+          pt_BR: Etiópia
+      - value: FK
+        label:
+          en_US: Falkland Islands (Malvinas)
+          zh_Hans: 福克兰群岛（马尔维纳斯）
+          pt_BR: Ilhas Falkland (Malvinas)
+      - value: FO
+        label:
+          en_US: Faroe Islands
+          zh_Hans: 法罗群岛
+          pt_BR: Ilhas Faroe
+      - value: FJ
+        label:
+          en_US: Fiji
+          zh_Hans: 斐济
+          pt_BR: Fiji
       - value: FI
         label:
           en_US: Finland
           zh_Hans: 芬兰
-          pt_BR: Finland
+          pt_BR: Finlândia
       - value: FR
         label:
           en_US: France
           zh_Hans: 法国
-          pt_BR: France
+          pt_BR: França
+      - value: GF
+        label:
+          en_US: French Guiana
+          zh_Hans: 法属圭亚那
+          pt_BR: Guiana Francesa
+      - value: PF
+        label:
+          en_US: French Polynesia
+          zh_Hans: 法属波利尼西亚
+          pt_BR: Polinésia Francesa
+      - value: TF
+        label:
+          en_US: French Southern Territories
+          zh_Hans: 法属南部领地
+          pt_BR: Territórios Franceses do Sul
+      - value: GA
+        label:
+          en_US: Gabon
+          zh_Hans: 加蓬
+          pt_BR: Gabão
+      - value: GM
+        label:
+          en_US: Gambia
+          zh_Hans: 冈比亚
+          pt_BR: Gâmbia
+      - value: GE
+        label:
+          en_US: Georgia
+          zh_Hans: 格鲁吉亚
+          pt_BR: Geórgia
       - value: DE
         label:
           en_US: Germany
           zh_Hans: 德国
-          pt_BR: Germany
+          pt_BR: Alemanha
+      - value: GH
+        label:
+          en_US: Ghana
+          zh_Hans: 加纳
+          pt_BR: Gana
+      - value: GI
+        label:
+          en_US: Gibraltar
+          zh_Hans: 直布罗陀
+          pt_BR: Gibraltar
+      - value: GR
+        label:
+          en_US: Greece
+          zh_Hans: 希腊
+          pt_BR: Grécia
+      - value: GL
+        label:
+          en_US: Greenland
+          zh_Hans: 格陵兰
+          pt_BR: Groenlândia
+      - value: GD
+        label:
+          en_US: Grenada
+          zh_Hans: 格林纳达
+          pt_BR: Granada
+      - value: GP
+        label:
+          en_US: Guadeloupe
+          zh_Hans: 瓜德罗普
+          pt_BR: Guadalupe
+      - value: GU
+        label:
+          en_US: Guam
+          zh_Hans: 关岛
+          pt_BR: Guam
+      - value: GT
+        label:
+          en_US: Guatemala
+          zh_Hans: 危地马拉
+          pt_BR: Guatemala
+      - value: GN
+        label:
+          en_US: Guinea
+          zh_Hans: 几内亚
+          pt_BR: Guiné
+      - value: GW
+        label:
+          en_US: Guinea-Bissau
+          zh_Hans: 几内亚比绍
+          pt_BR: Guiné-Bissau
+      - value: GY
+        label:
+          en_US: Guyana
+          zh_Hans: 圭亚那
+          pt_BR: Guiana
+      - value: HT
+        label:
+          en_US: Haiti
+          zh_Hans: 海地
+          pt_BR: Haiti
+      - value: HM
+        label:
+          en_US: Heard Island and McDonald Islands
+          zh_Hans: 赫德岛和麦克唐纳群岛
+          pt_BR: Ilha Heard e Ilhas McDonald
+      - value: VA
+        label:
+          en_US: Holy See (Vatican City State)
+          zh_Hans: 教廷（梵蒂冈城国）
+          pt_BR: Santa Sé (Estado da Cidade do Vaticano)
+      - value: HN
+        label:
+          en_US: Honduras
+          zh_Hans: 洪都拉斯
+          pt_BR: Honduras
       - value: HK
         label:
           en_US: Hong Kong
           zh_Hans: 香港
           pt_BR: Hong Kong
+      - value: HU
+        label:
+          en_US: Hungary
+          zh_Hans: 匈牙利
+          pt_BR: Hungria
+      - value: IS
+        label:
+          en_US: Iceland
+          zh_Hans: 冰岛
+          pt_BR: Islândia
       - value: IN
         label:
           en_US: India
           zh_Hans: 印度
-          pt_BR: India
+          pt_BR: Índia
       - value: ID
         label:
           en_US: Indonesia
           zh_Hans: 印度尼西亚
-          pt_BR: Indonesia
+          pt_BR: Indonésia
+      - value: IR
+        label:
+          en_US: Iran, Islamic Republic of
+          zh_Hans: 伊朗
+          pt_BR: Irã
+      - value: IQ
+        label:
+          en_US: Iraq
+          zh_Hans: 伊拉克
+          pt_BR: Iraque
+      - value: IE
+        label:
+          en_US: Ireland
+          zh_Hans: 爱尔兰
+          pt_BR: Irlanda
+      - value: IL
+        label:
+          en_US: Israel
+          zh_Hans: 以色列
+          pt_BR: Israel
       - value: IT
         label:
           en_US: Italy
           zh_Hans: 意大利
-          pt_BR: Italy
+          pt_BR: Itália
+      - value: JM
+        label:
+          en_US: Jamaica
+          zh_Hans: 牙买加
+          pt_BR: Jamaica
       - value: JP
         label:
           en_US: Japan
           zh_Hans: 日本
-          pt_BR: Japan
+          pt_BR: Japão
+      - value: JO
+        label:
+          en_US: Jordan
+          zh_Hans: 约旦
+          pt_BR: Jordânia
+      - value: KZ
+        label:
+          en_US: Kazakhstan
+          zh_Hans: 哈萨克斯坦
+          pt_BR: Cazaquistão
+      - value: KE
+        label:
+          en_US: Kenya
+          zh_Hans: 肯尼亚
+          pt_BR: Quênia
+      - value: KI
+        label:
+          en_US: Kiribati
+          zh_Hans: 基里巴斯
+          pt_BR: Kiribati
+      - value: KP
+        label:
+          en_US: Korea, Democratic People's Republic of
+          zh_Hans: 朝鲜
+          pt_BR: Coreia, República Democrática Popular da
       - value: KR
         label:
-          en_US: Korea
+          en_US: Korea, Republic of
           zh_Hans: 韩国
-          pt_BR: Korea
+          pt_BR: Coreia, República da
+      - value: KW
+        label:
+          en_US: Kuwait
+          zh_Hans: 科威特
+          pt_BR: Kuwait
+      - value: KG
+        label:
+          en_US: Kyrgyzstan
+          zh_Hans: 吉尔吉斯斯坦
+          pt_BR: Quirguistão
+      - value: LA
+        label:
+          en_US: Lao People's Democratic Republic
+          zh_Hans: 老挝
+          pt_BR: República Democrática Popular do Laos
+      - value: LV
+        label:
+          en_US: Latvia
+          zh_Hans: 拉脱维亚
+          pt_BR: Letônia
+      - value: LB
+        label:
+          en_US: Lebanon
+          zh_Hans: 黎巴嫩
+          pt_BR: Líbano
+      - value: LS
+        label:
+          en_US: Lesotho
+          zh_Hans: 莱索托
+          pt_BR: Lesoto
+      - value: LR
+        label:
+          en_US: Liberia
+          zh_Hans: 利比里亚
+          pt_BR: Libéria
+      - value: LY
+        label:
+          en_US: Libyan Arab Jamahiriya
+          zh_Hans: 利比亚
+          pt_BR: Líbia
+      - value: LI
+        label:
+          en_US: Liechtenstein
+          zh_Hans: 列支敦士登
+          pt_BR: Liechtenstein
+      - value: LT
+        label:
+          en_US: Lithuania
+          zh_Hans: 立陶宛
+          pt_BR: Lituânia
+      - value: LU
+        label:
+          en_US: Luxembourg
+          zh_Hans: 卢森堡
+          pt_BR: Luxemburgo
+      - value: MO
+        label:
+          en_US: Macao
+          zh_Hans: 澳门
+          pt_BR: Macau
+      - value: MK
+        label:
+          en_US: Macedonia, the Former Yugosalv Republic of
+          zh_Hans: 前南斯拉夫马其顿共和国
+          pt_BR: Macedônia, Ex-República Iugoslava da
+      - value: MG
+        label:
+          en_US: Madagascar
+          zh_Hans: 马达加斯加
+          pt_BR: Madagascar
+      - value: MW
+        label:
+          en_US: Malawi
+          zh_Hans: 马拉维
+          pt_BR: Malaui
       - value: MY
         label:
           en_US: Malaysia
           zh_Hans: 马来西亚
-          pt_BR: Malaysia
+          pt_BR: Malásia
+      - value: MV
+        label:
+          en_US: Maldives
+          zh_Hans: 马尔代夫
+          pt_BR: Maldivas
+      - value: ML
+        label:
+          en_US: Mali
+          zh_Hans: 马里
+          pt_BR: Mali
+      - value: MT
+        label:
+          en_US: Malta
+          zh_Hans: 马耳他
+          pt_BR: Malta
+      - value: MH
+        label:
+          en_US: Marshall Islands
+          zh_Hans: 马绍尔群岛
+          pt_BR: Ilhas Marshall
+      - value: MQ
+        label:
+          en_US: Martinique
+          zh_Hans: 马提尼克
+          pt_BR: Martinica
+      - value: MR
+        label:
+          en_US: Mauritania
+          zh_Hans: 毛里塔尼亚
+          pt_BR: Mauritânia
+      - value: MU
+        label:
+          en_US: Mauritius
+          zh_Hans: 毛里求斯
+          pt_BR: Maurício
+      - value: YT
+        label:
+          en_US: Mayotte
+          zh_Hans: 马约特
+          pt_BR: Mayotte
       - value: MX
         label:
           en_US: Mexico
           zh_Hans: 墨西哥
-          pt_BR: Mexico
+          pt_BR: México
+      - value: FM
+        label:
+          en_US: Micronesia, Federated States of
+          zh_Hans: 密克罗尼西亚联邦
+          pt_BR: Micronésia, Estados Federados da
+      - value: MD
+        label:
+          en_US: Moldova, Republic of
+          zh_Hans: 摩尔多瓦共和国
+          pt_BR: Moldávia, República da
+      - value: MC
+        label:
+          en_US: Monaco
+          zh_Hans: 摩纳哥
+          pt_BR: Mônaco
+      - value: MN
+        label:
+          en_US: Mongolia
+          zh_Hans: 蒙古
+          pt_BR: Mongólia
+      - value: MS
+        label:
+          en_US: Montserrat
+          zh_Hans: 蒙特塞拉特
+          pt_BR: Montserrat
+      - value: MA
+        label:
+          en_US: Morocco
+          zh_Hans: 摩洛哥
+          pt_BR: Marrocos
+      - value: MZ
+        label:
+          en_US: Mozambique
+          zh_Hans: 莫桑比克
+          pt_BR: Moçambique
+      - value: MM
+        label:
+          en_US: Myanmar
+          zh_Hans: 缅甸
+          pt_BR: Mianmar
+      - value: NA
+        label:
+          en_US: Namibia
+          zh_Hans: 纳米比亚
+          pt_BR: Namíbia
+      - value: NR
+        label:
+          en_US: Nauru
+          zh_Hans: 瑙鲁
+          pt_BR: Nauru
+      - value: NP
+        label:
+          en_US: Nepal
+          zh_Hans: 尼泊尔
+          pt_BR: Nepal
       - value: NL
         label:
           en_US: Netherlands
           zh_Hans: 荷兰
-          pt_BR: Netherlands
+          pt_BR: Países Baixos
+      - value: AN
+        label:
+          en_US: Netherlands Antilles
+          zh_Hans: 荷属安的列斯
+          pt_BR: Antilhas Holandesas
+      - value: NC
+        label:
+          en_US: New Caledonia
+          zh_Hans: 新喀里多尼亚
+          pt_BR: Nova Caledônia
       - value: NZ
         label:
           en_US: New Zealand
           zh_Hans: 新西兰
-          pt_BR: New Zealand
-      - value: 'NO'
+          pt_BR: Nova Zelândia
+      - value: NI
+        label:
+          en_US: Nicaragua
+          zh_Hans: 尼加拉瓜
+          pt_BR: Nicarágua
+      - value: NE
+        label:
+          en_US: Niger
+          zh_Hans: 尼日尔
+          pt_BR: Níger
+      - value: NG
+        label:
+          en_US: Nigeria
+          zh_Hans: 尼日利亚
+          pt_BR: Nigéria
+      - value: NU
+        label:
+          en_US: Niue
+          zh_Hans: 纽埃
+          pt_BR: Niue
+      - value: NF
+        label:
+          en_US: Norfolk Island
+          zh_Hans: 诺福克岛
+          pt_BR: Ilha Norfolk
+      - value: MP
+        label:
+          en_US: Northern Mariana Islands
+          zh_Hans: 北马里亚纳群岛
+          pt_BR: Ilhas Marianas do Norte
+      - value: "NO"
         label:
           en_US: Norway
           zh_Hans: 挪威
-          pt_BR: Norway
+          pt_BR: Noruega
+      - value: OM
+        label:
+          en_US: Oman
+          zh_Hans: 阿曼
+          pt_BR: Omã
+      - value: PK
+        label:
+          en_US: Pakistan
+          zh_Hans: 巴基斯坦
+          pt_BR: Paquistão
+      - value: PW
+        label:
+          en_US: Palau
+          zh_Hans: 帕劳
+          pt_BR: Palau
+      - value: PS
+        label:
+          en_US: Palestinian Territory, Occupied
+          zh_Hans: 巴勒斯坦领土
+          pt_BR: Palestina, Território Ocupado
+      - value: PA
+        label:
+          en_US: Panama
+          zh_Hans: 巴拿马
+          pt_BR: Panamá
+      - value: PG
+        label:
+          en_US: Papua New Guinea
+          zh_Hans: 巴布亚新几内亚
+          pt_BR: Papua Nova Guiné
+      - value: PY
+        label:
+          en_US: Paraguay
+          zh_Hans: 巴拉圭
+          pt_BR: Paraguai
+      - value: PE
+        label:
+          en_US: Peru
+          zh_Hans: 秘鲁
+          pt_BR: Peru
       - value: PH
         label:
           en_US: Philippines
           zh_Hans: 菲律宾
-          pt_BR: Philippines
+          pt_BR: Filipinas
+      - value: PN
+        label:
+          en_US: Pitcairn
+          zh_Hans: 皮特凯恩岛
+          pt_BR: Pitcairn
       - value: PL
         label:
           en_US: Poland
           zh_Hans: 波兰
-          pt_BR: Poland
+          pt_BR: Polônia
       - value: PT
         label:
           en_US: Portugal
           zh_Hans: 葡萄牙
           pt_BR: Portugal
+      - value: PR
+        label:
+          en_US: Puerto Rico
+          zh_Hans: 波多黎各
+          pt_BR: Porto Rico
+      - value: QA
+        label:
+          en_US: Qatar
+          zh_Hans: 卡塔尔
+          pt_BR: Catar
+      - value: RE
+        label:
+          en_US: Reunion
+          zh_Hans: 留尼旺
+          pt_BR: Reunião
+      - value: RO
+        label:
+          en_US: Romania
+          zh_Hans: 罗马尼亚
+          pt_BR: Romênia
       - value: RU
         label:
-          en_US: Russia
-          zh_Hans: 俄罗斯
-          pt_BR: Russia
+          en_US: Russian Federation
+          zh_Hans: 俄罗斯联邦
+          pt_BR: Rússia
+      - value: RW
+        label:
+          en_US: Rwanda
+          zh_Hans: 卢旺达
+          pt_BR: Ruanda
+      - value: SH
+        label:
+          en_US: Saint Helena
+          zh_Hans: 圣赫勒拿
+          pt_BR: Santa Helena
+      - value: KN
+        label:
+          en_US: Saint Kitts and Nevis
+          zh_Hans: 圣基茨和尼维斯
+          pt_BR: São Cristóvão e Nevis
+      - value: LC
+        label:
+          en_US: Saint Lucia
+          zh_Hans: 圣卢西亚
+          pt_BR: Santa Lúcia
+      - value: PM
+        label:
+          en_US: Saint Pierre and Miquelon
+          zh_Hans: 圣皮埃尔和密克隆
+          pt_BR: São Pedro e Miquelon
+      - value: VC
+        label:
+          en_US: Saint Vincent and the Grenadines
+          zh_Hans: 圣文森特和格林纳丁斯
+          pt_BR: São Vicente e Granadinas
+      - value: WS
+        label:
+          en_US: Samoa
+          zh_Hans: 萨摩亚
+          pt_BR: Samoa
+      - value: SM
+        label:
+          en_US: San Marino
+          zh_Hans: 圣马力诺
+          pt_BR: San Marino
+      - value: ST
+        label:
+          en_US: Sao Tome and Principe
+          zh_Hans: 圣多美和普林西比
+          pt_BR: São Tomé e Príncipe
       - value: SA
         label:
           en_US: Saudi Arabia
           zh_Hans: 沙特阿拉伯
-          pt_BR: Saudi Arabia
+          pt_BR: Arábia Saudita
+      - value: SN
+        label:
+          en_US: Senegal
+          zh_Hans: 塞内加尔
+          pt_BR: Senegal
+      - value: RS
+        label:
+          en_US: Serbia and Montenegro
+          zh_Hans: 塞尔维亚和黑山
+          pt_BR: Sérvia e Montenegro
+      - value: SC
+        label:
+          en_US: Seychelles
+          zh_Hans: 塞舌尔
+          pt_BR: Seicheles
+      - value: SL
+        label:
+          en_US: Sierra Leone
+          zh_Hans: 塞拉利昂
+          pt_BR: Serra Leoa
       - value: SG
         label:
           en_US: Singapore
           zh_Hans: 新加坡
-          pt_BR: Singapore
+          pt_BR: Singapura
+      - value: SK
+        label:
+          en_US: Slovakia
+          zh_Hans: 斯洛伐克
+          pt_BR: Eslováquia
+      - value: SI
+        label:
+          en_US: Slovenia
+          zh_Hans: 斯洛文尼亚
+          pt_BR: Eslovênia
+      - value: SB
+        label:
+          en_US: Solomon Islands
+          zh_Hans: 所罗门群岛
+          pt_BR: Ilhas Salomão
+      - value: SO
+        label:
+          en_US: Somalia
+          zh_Hans: 索马里
+          pt_BR: Somália
       - value: ZA
         label:
           en_US: South Africa
           zh_Hans: 南非
-          pt_BR: South Africa
+          pt_BR: África do Sul
+      - value: GS
+        label:
+          en_US: South Georgia and the South Sandwich Islands
+          zh_Hans: 南乔治亚和南桑威奇群岛
+          pt_BR: Geórgia do Sul e Ilhas Sandwich do Sul
       - value: ES
         label:
           en_US: Spain
           zh_Hans: 西班牙
-          pt_BR: Spain
+          pt_BR: Espanha
+      - value: LK
+        label:
+          en_US: Sri Lanka
+          zh_Hans: 斯里兰卡
+          pt_BR: Sri Lanka
+      - value: SD
+        label:
+          en_US: Sudan
+          zh_Hans: 苏丹
+          pt_BR: Sudão
+      - value: SR
+        label:
+          en_US: Suriname
+          zh_Hans: 苏里南
+          pt_BR: Suriname
+      - value: SJ
+        label:
+          en_US: Svalbard and Jan Mayen
+          zh_Hans: 斯瓦尔巴特和扬马延岛
+          pt_BR: Svalbard e Jan Mayen
+      - value: SZ
+        label:
+          en_US: Swaziland
+          zh_Hans: 斯威士兰
+          pt_BR: Essuatíni
       - value: SE
         label:
           en_US: Sweden
           zh_Hans: 瑞典
-          pt_BR: Sweden
+          pt_BR: Suécia
       - value: CH
         label:
           en_US: Switzerland
           zh_Hans: 瑞士
-          pt_BR: Switzerland
+          pt_BR: Suíça
+      - value: SY
+        label:
+          en_US: Syrian Arab Republic
+          zh_Hans: 叙利亚
+          pt_BR: Síria
       - value: TW
         label:
-          en_US: Taiwan
+          en_US: Taiwan, Province of China
           zh_Hans: 台湾
           pt_BR: Taiwan
+      - value: TJ
+        label:
+          en_US: Tajikistan
+          zh_Hans: 塔吉克斯坦
+          pt_BR: Tajiquistão
+      - value: TZ
+        label:
+          en_US: Tanzania, United Republic of
+          zh_Hans: 坦桑尼亚联合共和国
+          pt_BR: Tanzânia
       - value: TH
         label:
           en_US: Thailand
           zh_Hans: 泰国
-          pt_BR: Thailand
+          pt_BR: Tailândia
+      - value: TL
+        label:
+          en_US: Timor-Leste
+          zh_Hans: 东帝汶
+          pt_BR: Timor-Leste
+      - value: TG
+        label:
+          en_US: Togo
+          zh_Hans: 多哥
+          pt_BR: Togo
+      - value: TK
+        label:
+          en_US: Tokelau
+          zh_Hans: 托克劳
+          pt_BR: Toquelau
+      - value: TO
+        label:
+          en_US: Tonga
+          zh_Hans: 汤加
+          pt_BR: Tonga
+      - value: TT
+        label:
+          en_US: Trinidad and Tobago
+          zh_Hans: 特立尼达和多巴哥
+          pt_BR: Trindade e Tobago
+      - value: TN
+        label:
+          en_US: Tunisia
+          zh_Hans: 突尼斯
+          pt_BR: Tunísia
       - value: TR
         label:
           en_US: Turkey
           zh_Hans: 土耳其
-          pt_BR: Turkey
+          pt_BR: Turquia
+      - value: TM
+        label:
+          en_US: Turkmenistan
+          zh_Hans: 土库曼斯坦
+          pt_BR: Turcomenistão
+      - value: TC
+        label:
+          en_US: Turks and Caicos Islands
+          zh_Hans: 特克斯和凯科斯群岛
+          pt_BR: Ilhas Turks e Caicos
+      - value: TV
+        label:
+          en_US: Tuvalu
+          zh_Hans: 图瓦卢
+          pt_BR: Tuvalu
+      - value: UG
+        label:
+          en_US: Uganda
+          zh_Hans: 乌干达
+          pt_BR: Uganda
+      - value: UA
+        label:
+          en_US: Ukraine
+          zh_Hans: 乌克兰
+          pt_BR: Ucrânia
+      - value: AE
+        label:
+          en_US: United Arab Emirates
+          zh_Hans: 阿联酋
+          pt_BR: Emirados Árabes Unidos
+      - value: UK
+        label:
+          en_US: United Kingdom
+          zh_Hans: 英国
+          pt_BR: Reino Unido
       - value: GB
         label:
           en_US: United Kingdom
           zh_Hans: 英国
-          pt_BR: United Kingdom
+          pt_BR: Reino Unido
       - value: US
         label:
           en_US: United States
           zh_Hans: 美国
-          pt_BR: United States
+          pt_BR: Estados Unidos
+      - value: UM
+        label:
+          en_US: United States Minor Outlying Islands
+          zh_Hans: 美国本土外小岛屿
+          pt_BR: Ilhas Menores Distantes dos Estados Unidos
+      - value: UY
+        label:
+          en_US: Uruguay
+          zh_Hans: 乌拉圭
+          pt_BR: Uruguai
+      - value: UZ
+        label:
+          en_US: Uzbekistan
+          zh_Hans: 乌兹别克斯坦
+          pt_BR: Uzbequistão
+      - value: VU
+        label:
+          en_US: Vanuatu
+          zh_Hans: 瓦努阿图
+          pt_BR: Vanuatu
+      - value: VE
+        label:
+          en_US: Venezuela
+          zh_Hans: 委内瑞拉
+          pt_BR: Venezuela
+      - value: VN
+        label:
+          en_US: Viet Nam
+          zh_Hans: 越南
+          pt_BR: Vietnã
+      - value: VG
+        label:
+          en_US: Virgin Islands, British
+          zh_Hans: 英属维尔京群岛
+          pt_BR: Ilhas Virgens Britânicas
+      - value: VI
+        label:
+          en_US: Virgin Islands, U.S.
+          zh_Hans: 美属维尔京群岛
+          pt_BR: Ilhas Virgens dos EUA
+      - value: WF
+        label:
+          en_US: Wallis and Futuna
+          zh_Hans: 瓦利斯和富图纳群岛
+          pt_BR: Wallis e Futuna
+      - value: EH
+        label:
+          en_US: Western Sahara
+          zh_Hans: 西撒哈拉
+          pt_BR: Saara Ocidental
+      - value: YE
+        label:
+          en_US: Yemen
+          zh_Hans: 也门
+          pt_BR: Iémen
+      - value: ZM
+        label:
+          en_US: Zambia
+          zh_Hans: 赞比亚
+          pt_BR: Zâmbia
+      - value: ZW
+        label:
+          en_US: Zimbabwe
+          zh_Hans: 津巴布韦
+          pt_BR: Zimbábue
   - name: hl
     type: select
     label:
@@ -277,18 +1277,94 @@ parameters:
     default: en
     form: form
     options:
+      - value: af
+        label:
+          en_US: Afrikaans
+          zh_Hans: 南非语
+      - value: ak
+        label:
+          en_US: Akan
+          zh_Hans: 阿坎语
+      - value: sq
+        label:
+          en_US: Albanian
+          zh_Hans: 阿尔巴尼亚语
+      - value: ws
+        label:
+          en_US: Samoa
+          zh_Hans: 萨摩亚语
+      - value: am
+        label:
+          en_US: Amharic
+          zh_Hans: 阿姆哈拉语
       - value: ar
         label:
           en_US: Arabic
           zh_Hans: 阿拉伯语
+      - value: hy
+        label:
+          en_US: Armenian
+          zh_Hans: 亚美尼亚语
+      - value: az
+        label:
+          en_US: Azerbaijani
+          zh_Hans: 阿塞拜疆语
+      - value: eu
+        label:
+          en_US: Basque
+          zh_Hans: 巴斯克语
+      - value: be
+        label:
+          en_US: Belarusian
+          zh_Hans: 白俄罗斯语
+      - value: bem
+        label:
+          en_US: Bemba
+          zh_Hans: 班巴语
+      - value: bn
+        label:
+          en_US: Bengali
+          zh_Hans: 孟加拉语
+      - value: bh
+        label:
+          en_US: Bihari
+          zh_Hans: 比哈尔语
+      - value: xx-bork
+        label:
+          en_US: Bork, bork, bork!
+          zh_Hans: 博克语
+      - value: bs
+        label:
+          en_US: Bosnian
+          zh_Hans: 波斯尼亚语
+      - value: br
+        label:
+          en_US: Breton
+          zh_Hans: 布列塔尼语
       - value: bg
         label:
           en_US: Bulgarian
           zh_Hans: 保加利亚语
+      - value: bt
+        label:
+          en_US: Bhutanese
+          zh_Hans: 不丹语
+      - value: km
+        label:
+          en_US: Cambodian
+          zh_Hans: 高棉语
       - value: ca
         label:
           en_US: Catalan
           zh_Hans: 加泰罗尼亚语
+      - value: chr
+        label:
+          en_US: Cherokee
+          zh_Hans: 切罗基语
+      - value: ny
+        label:
+          en_US: Chichewa
+          zh_Hans: 齐切瓦语
       - value: zh-cn
         label:
           en_US: Chinese (Simplified)
@@ -297,6 +1373,14 @@ parameters:
         label:
           en_US: Chinese (Traditional)
           zh_Hans: 中文（繁体）
+      - value: co
+        label:
+          en_US: Corsican
+          zh_Hans: 科西嘉语
+      - value: hr
+        label:
+          en_US: Croatian
+          zh_Hans: 克罗地亚语
       - value: cs
         label:
           en_US: Czech
@@ -309,14 +1393,34 @@ parameters:
         label:
           en_US: Dutch
           zh_Hans: 荷兰语
+      - value: xx-elmer
+        label:
+          en_US: Elmer Fudd
+          zh_Hans: 艾尔默福德语
       - value: en
         label:
           en_US: English
           zh_Hans: 英语
+      - value: eo
+        label:
+          en_US: Esperanto
+          zh_Hans: 世界语
       - value: et
         label:
           en_US: Estonian
           zh_Hans: 爱沙尼亚语
+      - value: ee
+        label:
+          en_US: Ewe
+          zh_Hans: 埃维语
+      - value: fo
+        label:
+          en_US: Faroese
+          zh_Hans: 法罗语
+      - value: tl
+        label:
+          en_US: Filipino
+          zh_Hans: 菲律宾语
       - value: fi
         label:
           en_US: Finnish
@@ -325,6 +1429,22 @@ parameters:
         label:
           en_US: French
           zh_Hans: 法语
+      - value: fy
+        label:
+          en_US: Frisian
+          zh_Hans: 弗里西亚语
+      - value: gaa
+        label:
+          en_US: Ga
+          zh_Hans: 加语
+      - value: gl
+        label:
+          en_US: Galician
+          zh_Hans: 加利西亚语
+      - value: ka
+        label:
+          en_US: Georgian
+          zh_Hans: 格鲁吉亚语
       - value: de
         label:
           en_US: German
@@ -333,6 +1453,34 @@ parameters:
         label:
           en_US: Greek
           zh_Hans: 希腊语
+      - value: kl
+        label:
+          en_US: Greenlandic
+          zh_Hans: 格陵兰语
+      - value: gn
+        label:
+          en_US: Guarani
+          zh_Hans: 瓜拉尼语
+      - value: gu
+        label:
+          en_US: Gujarati
+          zh_Hans: 古吉拉特语
+      - value: xx-hacker
+        label:
+          en_US: Hacker
+          zh_Hans: 黑客语
+      - value: ht
+        label:
+          en_US: Haitian Creole
+          zh_Hans: 海地克里奥尔语
+      - value: ha
+        label:
+          en_US: Hausa
+          zh_Hans: 豪萨语
+      - value: haw
+        label:
+          en_US: Hawaiian
+          zh_Hans: 夏威夷语
       - value: iw
         label:
           en_US: Hebrew
@@ -345,10 +1493,26 @@ parameters:
         label:
           en_US: Hungarian
           zh_Hans: 匈牙利语
+      - value: is
+        label:
+          en_US: Icelandic
+          zh_Hans: 冰岛语
+      - value: ig
+        label:
+          en_US: Igbo
+          zh_Hans: 伊博语
       - value: id
         label:
           en_US: Indonesian
           zh_Hans: 印尼语
+      - value: ia
+        label:
+          en_US: Interlingua
+          zh_Hans: 国际语
+      - value: ga
+        label:
+          en_US: Irish
+          zh_Hans: 爱尔兰语
       - value: it
         label:
           en_US: Italian
@@ -357,22 +1521,94 @@ parameters:
         label:
           en_US: Japanese
           zh_Hans: 日语
+      - value: jw
+        label:
+          en_US: Javanese
+          zh_Hans: 爪哇语
       - value: kn
         label:
           en_US: Kannada
           zh_Hans: 卡纳达语
+      - value: kk
+        label:
+          en_US: Kazakh
+          zh_Hans: 哈萨克语
+      - value: rw
+        label:
+          en_US: Kinyarwanda
+          zh_Hans: 基尼亚卢旺达语
+      - value: rn
+        label:
+          en_US: Kirundi
+          zh_Hans: 基隆迪语
+      - value: xx-klingon
+        label:
+          en_US: Klingon
+          zh_Hans: 克林贡语
+      - value: kg
+        label:
+          en_US: Kongo
+          zh_Hans: 刚果语
       - value: ko
         label:
           en_US: Korean
           zh_Hans: 韩语
+      - value: kri
+        label:
+          en_US: Krio (Sierra Leone)
+          zh_Hans: 塞拉利昂克里奥尔语
+      - value: ku
+        label:
+          en_US: Kurdish
+          zh_Hans: 库尔德语
+      - value: ckb
+        label:
+          en_US: Kurdish (Soranî)
+          zh_Hans: 库尔德语（索拉尼）
+      - value: ky
+        label:
+          en_US: Kyrgyz
+          zh_Hans: 吉尔吉斯语
+      - value: lo
+        label:
+          en_US: Laothian
+          zh_Hans: 老挝语
+      - value: la
+        label:
+          en_US: Latin
+          zh_Hans: 拉丁语
       - value: lv
         label:
           en_US: Latvian
           zh_Hans: 拉脱维亚语
+      - value: ln
+        label:
+          en_US: Lingala
+          zh_Hans: 林加拉语
       - value: lt
         label:
           en_US: Lithuanian
           zh_Hans: 立陶宛语
+      - value: loz
+        label:
+          en_US: Lozi
+          zh_Hans: 洛齐语
+      - value: lg
+        label:
+          en_US: Luganda
+          zh_Hans: 卢干达语
+      - value: ach
+        label:
+          en_US: Luo
+          zh_Hans: 卢奥语
+      - value: mk
+        label:
+          en_US: Macedonian
+          zh_Hans: 马其顿语
+      - value: mg
+        label:
+          en_US: Malagasy
+          zh_Hans: 马尔加什语
       - value: my
         label:
           en_US: Malay
@@ -381,18 +1617,90 @@ parameters:
         label:
           en_US: Malayalam
           zh_Hans: 马拉雅拉姆语
+      - value: mt
+        label:
+          en_US: Maltese
+          zh_Hans: 马耳他语
+      - value: mv
+        label:
+          en_US: Maldives
+          zh_Hans: 马尔代夫语
+      - value: mi
+        label:
+          en_US: Maori
+          zh_Hans: 毛利语
       - value: mr
         label:
           en_US: Marathi
           zh_Hans: 马拉地语
+      - value: mfe
+        label:
+          en_US: Mauritian Creole
+          zh_Hans: 毛里求斯克里奥尔语
+      - value: mo
+        label:
+          en_US: Moldavian
+          zh_Hans: 摩尔达维亚语
+      - value: mn
+        label:
+          en_US: Mongolian
+          zh_Hans: 蒙古语
+      - value: sr-me
+        label:
+          en_US: Montenegrin
+          zh_Hans: 黑山语
+      - value: ne
+        label:
+          en_US: Nepali
+          zh_Hans: 尼泊尔语
+      - value: pcm
+        label:
+          en_US: Nigerian Pidgin
+          zh_Hans: 尼日利亚皮钦语
+      - value: nso
+        label:
+          en_US: Northern Sotho
+          zh_Hans: 北索托语
       - value: "no"
         label:
           en_US: Norwegian
           zh_Hans: 挪威语
+      - value: nn
+        label:
+          en_US: Norwegian (Nynorsk)
+          zh_Hans: 挪威语（尼诺斯克语）
+      - value: oc
+        label:
+          en_US: Occitan
+          zh_Hans: 奥克语
+      - value: or
+        label:
+          en_US: Oriya
+          zh_Hans: 奥里亚语
+      - value: om
+        label:
+          en_US: Oromo
+          zh_Hans: 奥罗莫语
+      - value: ps
+        label:
+          en_US: Pashto
+          zh_Hans: 普什图语
+      - value: fa
+        label:
+          en_US: Persian
+          zh_Hans: 波斯语
+      - value: xx-pirate
+        label:
+          en_US: Pirate
+          zh_Hans: 海盗语
       - value: pl
         label:
           en_US: Polish
           zh_Hans: 波兰语
+      - value: pt
+        label:
+          en_US: Portuguese
+          zh_Hans: 葡萄牙语
       - value: pt-br
         label:
           en_US: Portuguese (Brazil)
@@ -405,18 +1713,62 @@ parameters:
         label:
           en_US: Punjabi
           zh_Hans: 旁遮普语
+      - value: qu
+        label:
+          en_US: Quechua
+          zh_Hans: 克丘亚语
       - value: ro
         label:
           en_US: Romanian
           zh_Hans: 罗马尼亚语
+      - value: rm
+        label:
+          en_US: Romansh
+          zh_Hans: 罗曼什语
+      - value: nyn
+        label:
+          en_US: Runyakitara
+          zh_Hans: 卢尼亚基塔拉语
       - value: ru
         label:
           en_US: Russian
           zh_Hans: 俄语
+      - value: gd
+        label:
+          en_US: Scots Gaelic
+          zh_Hans: 苏格兰盖尔语
       - value: sr
         label:
           en_US: Serbian
           zh_Hans: 塞尔维亚语
+      - value: sh
+        label:
+          en_US: Serbo-Croatian
+          zh_Hans: 塞尔维亚-克罗地亚语
+      - value: st
+        label:
+          en_US: Sesotho
+          zh_Hans: 塞索托语
+      - value: tn
+        label:
+          en_US: Setswana
+          zh_Hans: 塞茨瓦纳语
+      - value: crs
+        label:
+          en_US: Seychellois Creole
+          zh_Hans: 塞舌尔克里奥尔语
+      - value: sn
+        label:
+          en_US: Shona
+          zh_Hans: 绍纳语
+      - value: sd
+        label:
+          en_US: Sindhi
+          zh_Hans: 信德语
+      - value: si
+        label:
+          en_US: Sinhalese
+          zh_Hans: 僧伽罗语
       - value: sk
         label:
           en_US: Slovak
@@ -425,18 +1777,42 @@ parameters:
         label:
           en_US: Slovenian
           zh_Hans: 斯洛文尼亚语
+      - value: so
+        label:
+          en_US: Somali
+          zh_Hans: 索马里语
       - value: es
         label:
           en_US: Spanish
           zh_Hans: 西班牙语
+      - value: es-419
+        label:
+          en_US: Spanish (Latin American)
+          zh_Hans: 西班牙语（拉丁美洲）
+      - value: su
+        label:
+          en_US: Sundanese
+          zh_Hans: 巽他语
+      - value: sw
+        label:
+          en_US: Swahili
+          zh_Hans: 斯瓦希里语
       - value: sv
         label:
           en_US: Swedish
           zh_Hans: 瑞典语
+      - value: tg
+        label:
+          en_US: Tajik
+          zh_Hans: 塔吉克语
       - value: ta
         label:
           en_US: Tamil
           zh_Hans: 泰米尔语
+      - value: tt
+        label:
+          en_US: Tatar
+          zh_Hans: 鞑靼语
       - value: te
         label:
           en_US: Telugu
@@ -445,18 +1821,82 @@ parameters:
         label:
           en_US: Thai
           zh_Hans: 泰语
+      - value: ti
+        label:
+          en_US: Tigrinya
+          zh_Hans: 提格利尼亚语
+      - value: to
+        label:
+          en_US: Tonga
+          zh_Hans: 汤加语
+      - value: lua
+        label:
+          en_US: Tshiluba
+          zh_Hans: 卢巴语
+      - value: tum
+        label:
+          en_US: Tumbuka
+          zh_Hans: 图布卡语
       - value: tr
         label:
           en_US: Turkish
           zh_Hans: 土耳其语
+      - value: tk
+        label:
+          en_US: Turkmen
+          zh_Hans: 土库曼语
+      - value: tw
+        label:
+          en_US: Twi
+          zh_Hans: 契维语
+      - value: ug
+        label:
+          en_US: Uighur
+          zh_Hans: 维吾尔语
       - value: uk
         label:
           en_US: Ukrainian
           zh_Hans: 乌克兰语
+      - value: ur
+        label:
+          en_US: Urdu
+          zh_Hans: 乌尔都语
+      - value: uz
+        label:
+          en_US: Uzbek
+          zh_Hans: 乌兹别克语
+      - value: vu
+        label:
+          en_US: Vanuatu
+          zh_Hans: 瓦努阿图语
       - value: vi
         label:
           en_US: Vietnamese
           zh_Hans: 越南语
+      - value: cy
+        label:
+          en_US: Welsh
+          zh_Hans: 威尔士语
+      - value: wo
+        label:
+          en_US: Wolof
+          zh_Hans: 沃洛夫语
+      - value: xh
+        label:
+          en_US: Xhosa
+          zh_Hans: 科萨语
+      - value: yi
+        label:
+          en_US: Yiddish
+          zh_Hans: 意第绪语
+      - value: yo
+        label:
+          en_US: Yoruba
+          zh_Hans: 约鲁巴语
+      - value: zu
+        label:
+          en_US: Zulu
+          zh_Hans: 祖鲁语
   - name: google_domain
     type: string
     required: false
diff --git a/api/core/tools/provider/builtin/searchapi/tools/google_jobs.yaml b/api/core/tools/provider/builtin/searchapi/tools/google_jobs.yaml
index 9033bc0f8784c..3e00e20fbd6e3 100644
--- a/api/core/tools/provider/builtin/searchapi/tools/google_jobs.yaml
+++ b/api/core/tools/provider/builtin/searchapi/tools/google_jobs.yaml
@@ -65,36 +65,141 @@ parameters:
     form: form
     default: US
     options:
-      - value: AR
-        label:
-          en_US: Argentina
-          zh_Hans: 阿根廷
-          pt_BR: Argentina
-      - value: AU
-        label:
-          en_US: Australia
-          zh_Hans: 澳大利亚
-          pt_BR: Australia
+      - value: DZ
+        label:
+          en_US: Algeria
+          zh_Hans: 阿尔及利亚
+          pt_BR: Algeria
+      - value: AS
+        label:
+          en_US: American Samoa
+          zh_Hans: 美属萨摩亚
+          pt_BR: American Samoa
+      - value: AO
+        label:
+          en_US: Angola
+          zh_Hans: 安哥拉
+          pt_BR: Angola
+      - value: AI
+        label:
+          en_US: Anguilla
+          zh_Hans: 安圭拉
+          pt_BR: Anguilla
+      - value: AG
+        label:
+          en_US: Antigua and Barbuda
+          zh_Hans: 安提瓜和巴布达
+          pt_BR: Antigua and Barbuda
+      - value: AW
+        label:
+          en_US: Aruba
+          zh_Hans: 阿鲁巴
+          pt_BR: Aruba
       - value: AT
         label:
           en_US: Austria
           zh_Hans: 奥地利
           pt_BR: Austria
+      - value: BS
+        label:
+          en_US: Bahamas
+          zh_Hans: 巴哈马
+          pt_BR: Bahamas
+      - value: BH
+        label:
+          en_US: Bahrain
+          zh_Hans: 巴林
+          pt_BR: Bahrain
+      - value: BD
+        label:
+          en_US: Bangladesh
+          zh_Hans: 孟加拉国
+          pt_BR: Bangladesh
+      - value: BY
+        label:
+          en_US: Belarus
+          zh_Hans: 白俄罗斯
+          pt_BR: Belarus
       - value: BE
         label:
           en_US: Belgium
           zh_Hans: 比利时
           pt_BR: Belgium
+      - value: BZ
+        label:
+          en_US: Belize
+          zh_Hans: 伯利兹
+          pt_BR: Belize
+      - value: BJ
+        label:
+          en_US: Benin
+          zh_Hans: 贝宁
+          pt_BR: Benin
+      - value: BM
+        label:
+          en_US: Bermuda
+          zh_Hans: 百慕大
+          pt_BR: Bermuda
+      - value: BO
+        label:
+          en_US: Bolivia
+          zh_Hans: 玻利维亚
+          pt_BR: Bolivia
+      - value: BW
+        label:
+          en_US: Botswana
+          zh_Hans: 博茨瓦纳
+          pt_BR: Botswana
       - value: BR
         label:
           en_US: Brazil
           zh_Hans: 巴西
           pt_BR: Brazil
+      - value: IO
+        label:
+          en_US: British Indian Ocean Territory
+          zh_Hans: 英属印度洋领地
+          pt_BR: British Indian Ocean Territory
+      - value: BF
+        label:
+          en_US: Burkina Faso
+          zh_Hans: 布基纳法索
+          pt_BR: Burkina Faso
+      - value: BI
+        label:
+          en_US: Burundi
+          zh_Hans: 布隆迪
+          pt_BR: Burundi
+      - value: CM
+        label:
+          en_US: Cameroon
+          zh_Hans: 喀麦隆
+          pt_BR: Cameroon
       - value: CA
         label:
           en_US: Canada
           zh_Hans: 加拿大
           pt_BR: Canada
+      - value: CV
+        label:
+          en_US: Cape Verde
+          zh_Hans: 佛得角
+          pt_BR: Cape Verde
+      - value: KY
+        label:
+          en_US: Cayman Islands
+          zh_Hans: 开曼群岛
+          pt_BR: Cayman Islands
+      - value: CF
+        label:
+          en_US: Central African Republic
+          zh_Hans: 中非共和国
+          pt_BR: Central African Republic
+      - value: TD
+        label:
+          en_US: Chad
+          zh_Hans: 乍得
+          pt_BR: Chad
       - value: CL
         label:
           en_US: Chile
@@ -105,36 +210,141 @@ parameters:
           en_US: Colombia
           zh_Hans: 哥伦比亚
           pt_BR: Colombia
-      - value: CN
-        label:
-          en_US: China
-          zh_Hans: 中国
-          pt_BR: China
-      - value: CZ
-        label:
-          en_US: Czech Republic
-          zh_Hans: 捷克共和国
-          pt_BR: Czech Republic
+      - value: CD
+        label:
+          en_US: Congo, the Democratic Republic of the
+          zh_Hans: 刚果民主共和国
+          pt_BR: Congo, the Democratic Republic of the
+      - value: CR
+        label:
+          en_US: Costa Rica
+          zh_Hans: 哥斯达黎加
+          pt_BR: Costa Rica
+      - value: CI
+        label:
+          en_US: Cote D'ivoire
+          zh_Hans: 科特迪瓦
+          pt_BR: Cote D'ivoire
+      - value: CU
+        label:
+          en_US: Cuba
+          zh_Hans: 古巴
+          pt_BR: Cuba
       - value: DK
         label:
           en_US: Denmark
           zh_Hans: 丹麦
           pt_BR: Denmark
-      - value: FI
-        label:
-          en_US: Finland
-          zh_Hans: 芬兰
-          pt_BR: Finland
+      - value: DJ
+        label:
+          en_US: Djibouti
+          zh_Hans: 吉布提
+          pt_BR: Djibouti
+      - value: DM
+        label:
+          en_US: Dominica
+          zh_Hans: 多米尼克
+          pt_BR: Dominica
+      - value: DO
+        label:
+          en_US: Dominican Republic
+          zh_Hans: 多米尼加共和国
+          pt_BR: Dominican Republic
+      - value: EC
+        label:
+          en_US: Ecuador
+          zh_Hans: 厄瓜多尔
+          pt_BR: Ecuador
+      - value: EG
+        label:
+          en_US: Egypt
+          zh_Hans: 埃及
+          pt_BR: Egypt
+      - value: SV
+        label:
+          en_US: El Salvador
+          zh_Hans: 萨尔瓦多
+          pt_BR: El Salvador
+      - value: ET
+        label:
+          en_US: Ethiopia
+          zh_Hans: 埃塞俄比亚
+          pt_BR: Ethiopia
+      - value: FK
+        label:
+          en_US: Falkland Islands (Malvinas)
+          zh_Hans: 福克兰群岛（马尔维纳斯）
+          pt_BR: Falkland Islands (Malvinas)
       - value: FR
         label:
           en_US: France
           zh_Hans: 法国
           pt_BR: France
+      - value: GF
+        label:
+          en_US: French Guiana
+          zh_Hans: 法属圭亚那
+          pt_BR: French Guiana
+      - value: PF
+        label:
+          en_US: French Polynesia
+          zh_Hans: 法属波利尼西亚
+          pt_BR: French Polynesia
+      - value: TF
+        label:
+          en_US: French Southern Territories
+          zh_Hans: 法属南部领地
+          pt_BR: French Southern Territories
+      - value: GA
+        label:
+          en_US: Gabon
+          zh_Hans: 加蓬
+          pt_BR: Gabon
+      - value: GM
+        label:
+          en_US: Gambia
+          zh_Hans: 冈比亚
+          pt_BR: Gambia
       - value: DE
         label:
           en_US: Germany
           zh_Hans: 德国
           pt_BR: Germany
+      - value: GH
+        label:
+          en_US: Ghana
+          zh_Hans: 加纳
+          pt_BR: Ghana
+      - value: GR
+        label:
+          en_US: Greece
+          zh_Hans: 希腊
+          pt_BR: Greece
+      - value: GP
+        label:
+          en_US: Guadeloupe
+          zh_Hans: 瓜德罗普
+          pt_BR: Guadeloupe
+      - value: GT
+        label:
+          en_US: Guatemala
+          zh_Hans: 危地马拉
+          pt_BR: Guatemala
+      - value: GY
+        label:
+          en_US: Guyana
+          zh_Hans: 圭亚那
+          pt_BR: Guyana
+      - value: HT
+        label:
+          en_US: Haiti
+          zh_Hans: 海地
+          pt_BR: Haiti
+      - value: HN
+        label:
+          en_US: Honduras
+          zh_Hans: 洪都拉斯
+          pt_BR: Honduras
       - value: HK
         label:
           en_US: Hong Kong
@@ -150,91 +360,291 @@ parameters:
           en_US: Indonesia
           zh_Hans: 印度尼西亚
           pt_BR: Indonesia
+      - value: IQ
+        label:
+          en_US: Iraq
+          zh_Hans: 伊拉克
+          pt_BR: Iraq
       - value: IT
         label:
           en_US: Italy
           zh_Hans: 意大利
           pt_BR: Italy
+      - value: JM
+        label:
+          en_US: Jamaica
+          zh_Hans: 牙买加
+          pt_BR: Jamaica
       - value: JP
         label:
           en_US: Japan
           zh_Hans: 日本
           pt_BR: Japan
-      - value: KR
-        label:
-          en_US: Korea
-          zh_Hans: 韩国
-          pt_BR: Korea
+      - value: JO
+        label:
+          en_US: Jordan
+          zh_Hans: 约旦
+          pt_BR: Jordan
+      - value: KZ
+        label:
+          en_US: Kazakhstan
+          zh_Hans: 哈萨克斯坦
+          pt_BR: Kazakhstan
+      - value: KE
+        label:
+          en_US: Kenya
+          zh_Hans: 肯尼亚
+          pt_BR: Kenya
+      - value: KW
+        label:
+          en_US: Kuwait
+          zh_Hans: 科威特
+          pt_BR: Kuwait
+      - value: KG
+        label:
+          en_US: Kyrgyzstan
+          zh_Hans: 吉尔吉斯斯坦
+          pt_BR: Kyrgyzstan
+      - value: LB
+        label:
+          en_US: Lebanon
+          zh_Hans: 黎巴嫩
+          pt_BR: Lebanon
+      - value: LS
+        label:
+          en_US: Lesotho
+          zh_Hans: 莱索托
+          pt_BR: Lesotho
+      - value: LY
+        label:
+          en_US: Libyan Arab Jamahiriya
+          zh_Hans: 利比亚
+          pt_BR: Libyan Arab Jamahiriya
+      - value: MG
+        label:
+          en_US: Madagascar
+          zh_Hans: 马达加斯加
+          pt_BR: Madagascar
+      - value: MW
+        label:
+          en_US: Malawi
+          zh_Hans: 马拉维
+          pt_BR: Malawi
       - value: MY
         label:
           en_US: Malaysia
           zh_Hans: 马来西亚
           pt_BR: Malaysia
+      - value: ML
+        label:
+          en_US: Mali
+          zh_Hans: 马里
+          pt_BR: Mali
+      - value: MQ
+        label:
+          en_US: Martinique
+          zh_Hans: 马提尼克
+          pt_BR: Martinique
+      - value: MU
+        label:
+          en_US: Mauritius
+          zh_Hans: 毛里求斯
+          pt_BR: Mauritius
+      - value: YT
+        label:
+          en_US: Mayotte
+          zh_Hans: 马约特
+          pt_BR: Mayotte
       - value: MX
         label:
           en_US: Mexico
           zh_Hans: 墨西哥
           pt_BR: Mexico
+      - value: MS
+        label:
+          en_US: Montserrat
+          zh_Hans: 蒙特塞拉特
+          pt_BR: Montserrat
+      - value: MA
+        label:
+          en_US: Morocco
+          zh_Hans: 摩洛哥
+          pt_BR: Morocco
+      - value: MZ
+        label:
+          en_US: Mozambique
+          zh_Hans: 莫桑比克
+          pt_BR: Mozambique
+      - value: NA
+        label:
+          en_US: Namibia
+          zh_Hans: 纳米比亚
+          pt_BR: Namibia
       - value: NL
         label:
           en_US: Netherlands
           zh_Hans: 荷兰
           pt_BR: Netherlands
-      - value: NZ
-        label:
-          en_US: New Zealand
-          zh_Hans: 新西兰
-          pt_BR: New Zealand
-      - value: 'NO'
-        label:
-          en_US: Norway
-          zh_Hans: 挪威
-          pt_BR: Norway
+      - value: NC
+        label:
+          en_US: New Caledonia
+          zh_Hans: 新喀里多尼亚
+          pt_BR: New Caledonia
+      - value: NI
+        label:
+          en_US: Nicaragua
+          zh_Hans: 尼加拉瓜
+          pt_BR: Nicaragua
+      - value: NE
+        label:
+          en_US: Niger
+          zh_Hans: 尼日尔
+          pt_BR: Niger
+      - value: NG
+        label:
+          en_US: Nigeria
+          zh_Hans: 尼日利亚
+          pt_BR: Nigeria
+      - value: OM
+        label:
+          en_US: Oman
+          zh_Hans: 阿曼
+          pt_BR: Oman
+      - value: PK
+        label:
+          en_US: Pakistan
+          zh_Hans: 巴基斯坦
+          pt_BR: Pakistan
+      - value: PS
+        label:
+          en_US: Palestinian Territory, Occupied
+          zh_Hans: 巴勒斯坦领土
+          pt_BR: Palestinian Territory, Occupied
+      - value: PA
+        label:
+          en_US: Panama
+          zh_Hans: 巴拿马
+          pt_BR: Panama
+      - value: PY
+        label:
+          en_US: Paraguay
+          zh_Hans: 巴拉圭
+          pt_BR: Paraguay
+      - value: PE
+        label:
+          en_US: Peru
+          zh_Hans: 秘鲁
+          pt_BR: Peru
       - value: PH
         label:
           en_US: Philippines
           zh_Hans: 菲律宾
           pt_BR: Philippines
-      - value: PL
-        label:
-          en_US: Poland
-          zh_Hans: 波兰
-          pt_BR: Poland
       - value: PT
         label:
           en_US: Portugal
           zh_Hans: 葡萄牙
           pt_BR: Portugal
+      - value: PR
+        label:
+          en_US: Puerto Rico
+          zh_Hans: 波多黎各
+          pt_BR: Puerto Rico
+      - value: QA
+        label:
+          en_US: Qatar
+          zh_Hans: 卡塔尔
+          pt_BR: Qatar
+      - value: RE
+        label:
+          en_US: Reunion
+          zh_Hans: 留尼旺
+          pt_BR: Reunion
       - value: RU
         label:
-          en_US: Russia
-          zh_Hans: 俄罗斯
-          pt_BR: Russia
+          en_US: Russian Federation
+          zh_Hans: 俄罗斯联邦
+          pt_BR: Russian Federation
+      - value: RW
+        label:
+          en_US: Rwanda
+          zh_Hans: 卢旺达
+          pt_BR: Rwanda
+      - value: SH
+        label:
+          en_US: Saint Helena
+          zh_Hans: 圣赫勒拿
+          pt_BR: Saint Helena
+      - value: PM
+        label:
+          en_US: Saint Pierre and Miquelon
+          zh_Hans: 圣皮埃尔和密克隆
+          pt_BR: Saint Pierre and Miquelon
+      - value: VC
+        label:
+          en_US: Saint Vincent and the Grenadines
+          zh_Hans: 圣文森特和格林纳丁斯
+          pt_BR: Saint Vincent and the Grenadines
+      - value: ST
+        label:
+          en_US: Sao Tome and Principe
+          zh_Hans: 圣多美和普林西比
+          pt_BR: Sao Tome and Principe
       - value: SA
         label:
           en_US: Saudi Arabia
           zh_Hans: 沙特阿拉伯
           pt_BR: Saudi Arabia
+      - value: SN
+        label:
+          en_US: Senegal
+          zh_Hans: 塞内加尔
+          pt_BR: Senegal
+      - value: SC
+        label:
+          en_US: Seychelles
+          zh_Hans: 塞舌尔
+          pt_BR: Seychelles
+      - value: SL
+        label:
+          en_US: Sierra Leone
+          zh_Hans: 塞拉利昂
+          pt_BR: Sierra Leone
       - value: SG
         label:
           en_US: Singapore
           zh_Hans: 新加坡
           pt_BR: Singapore
+      - value: SO
+        label:
+          en_US: Somalia
+          zh_Hans: 索马里
+          pt_BR: Somalia
       - value: ZA
         label:
           en_US: South Africa
           zh_Hans: 南非
           pt_BR: South Africa
+      - value: GS
+        label:
+          en_US: South Georgia and the South Sandwich Islands
+          zh_Hans: 南乔治亚和南桑威奇群岛
+          pt_BR: South Georgia and the South Sandwich Islands
       - value: ES
         label:
           en_US: Spain
           zh_Hans: 西班牙
           pt_BR: Spain
-      - value: SE
+      - value: LK
+        label:
+          en_US: Sri Lanka
+          zh_Hans: 斯里兰卡
+          pt_BR: Sri Lanka
+      - value: SR
         label:
-          en_US: Sweden
-          zh_Hans: 瑞典
-          pt_BR: Sweden
+          en_US: Suriname
+          zh_Hans: 苏里南
+          pt_BR: Suriname
       - value: CH
         label:
           en_US: Switzerland
@@ -242,19 +652,54 @@ parameters:
           pt_BR: Switzerland
       - value: TW
         label:
-          en_US: Taiwan
-          zh_Hans: 台湾
-          pt_BR: Taiwan
+          en_US: Taiwan, Province of China
+          zh_Hans: 中国台湾省
+          pt_BR: Taiwan, Province of China
+      - value: TZ
+        label:
+          en_US: Tanzania, United Republic of
+          zh_Hans: 坦桑尼亚联合共和国
+          pt_BR: Tanzania, United Republic of
       - value: TH
         label:
           en_US: Thailand
           zh_Hans: 泰国
           pt_BR: Thailand
-      - value: TR
+      - value: TG
+        label:
+          en_US: Togo
+          zh_Hans: 多哥
+          pt_BR: Togo
+      - value: TT
+        label:
+          en_US: Trinidad and Tobago
+          zh_Hans: 特立尼达和多巴哥
+          pt_BR: Trinidad and Tobago
+      - value: TN
+        label:
+          en_US: Tunisia
+          zh_Hans: 突尼斯
+          pt_BR: Tunisia
+      - value: TC
+        label:
+          en_US: Turks and Caicos Islands
+          zh_Hans: 特克斯和凯科斯群岛
+          pt_BR: Turks and Caicos Islands
+      - value: UG
+        label:
+          en_US: Uganda
+          zh_Hans: 乌干达
+          pt_BR: Uganda
+      - value: AE
+        label:
+          en_US: United Arab Emirates
+          zh_Hans: 阿联酋
+          pt_BR: United Arab Emirates
+      - value: UK
         label:
-          en_US: Turkey
-          zh_Hans: 土耳其
-          pt_BR: Turkey
+          en_US: United Kingdom
+          zh_Hans: 英国
+          pt_BR: United Kingdom
       - value: GB
         label:
           en_US: United Kingdom
@@ -265,6 +710,46 @@ parameters:
           en_US: United States
           zh_Hans: 美国
           pt_BR: United States
+      - value: UY
+        label:
+          en_US: Uruguay
+          zh_Hans: 乌拉圭
+          pt_BR: Uruguay
+      - value: UZ
+        label:
+          en_US: Uzbekistan
+          zh_Hans: 乌兹别克斯坦
+          pt_BR: Uzbekistan
+      - value: VE
+        label:
+          en_US: Venezuela
+          zh_Hans: 委内瑞拉
+          pt_BR: Venezuela
+      - value: VN
+        label:
+          en_US: Viet Nam
+          zh_Hans: 越南
+          pt_BR: Viet Nam
+      - value: VG
+        label:
+          en_US: Virgin Islands, British
+          zh_Hans: 英属维尔京群岛
+          pt_BR: Virgin Islands, British
+      - value: VI
+        label:
+          en_US: Virgin Islands, U.S.
+          zh_Hans: 美属维尔京群岛
+          pt_BR: Virgin Islands, U.S.
+      - value: ZM
+        label:
+          en_US: Zambia
+          zh_Hans: 赞比亚
+          pt_BR: Zambia
+      - value: ZW
+        label:
+          en_US: Zimbabwe
+          zh_Hans: 津巴布韦
+          pt_BR: Zimbabwe
   - name: hl
     type: select
     label:
@@ -277,18 +762,94 @@ parameters:
     default: en
     form: form
     options:
+      - value: af
+        label:
+          en_US: Afrikaans
+          zh_Hans: 南非语
+      - value: ak
+        label:
+          en_US: Akan
+          zh_Hans: 阿坎语
+      - value: sq
+        label:
+          en_US: Albanian
+          zh_Hans: 阿尔巴尼亚语
+      - value: ws
+        label:
+          en_US: Samoa
+          zh_Hans: 萨摩亚语
+      - value: am
+        label:
+          en_US: Amharic
+          zh_Hans: 阿姆哈拉语
       - value: ar
         label:
           en_US: Arabic
           zh_Hans: 阿拉伯语
+      - value: hy
+        label:
+          en_US: Armenian
+          zh_Hans: 亚美尼亚语
+      - value: az
+        label:
+          en_US: Azerbaijani
+          zh_Hans: 阿塞拜疆语
+      - value: eu
+        label:
+          en_US: Basque
+          zh_Hans: 巴斯克语
+      - value: be
+        label:
+          en_US: Belarusian
+          zh_Hans: 白俄罗斯语
+      - value: bem
+        label:
+          en_US: Bemba
+          zh_Hans: 班巴语
+      - value: bn
+        label:
+          en_US: Bengali
+          zh_Hans: 孟加拉语
+      - value: bh
+        label:
+          en_US: Bihari
+          zh_Hans: 比哈尔语
+      - value: xx-bork
+        label:
+          en_US: Bork, bork, bork!
+          zh_Hans: 博克语
+      - value: bs
+        label:
+          en_US: Bosnian
+          zh_Hans: 波斯尼亚语
+      - value: br
+        label:
+          en_US: Breton
+          zh_Hans: 布列塔尼语
       - value: bg
         label:
           en_US: Bulgarian
           zh_Hans: 保加利亚语
+      - value: bt
+        label:
+          en_US: Bhutanese
+          zh_Hans: 不丹语
+      - value: km
+        label:
+          en_US: Cambodian
+          zh_Hans: 高棉语
       - value: ca
         label:
           en_US: Catalan
           zh_Hans: 加泰罗尼亚语
+      - value: chr
+        label:
+          en_US: Cherokee
+          zh_Hans: 切罗基语
+      - value: ny
+        label:
+          en_US: Chichewa
+          zh_Hans: 齐切瓦语
       - value: zh-cn
         label:
           en_US: Chinese (Simplified)
@@ -297,6 +858,14 @@ parameters:
         label:
           en_US: Chinese (Traditional)
           zh_Hans: 中文（繁体）
+      - value: co
+        label:
+          en_US: Corsican
+          zh_Hans: 科西嘉语
+      - value: hr
+        label:
+          en_US: Croatian
+          zh_Hans: 克罗地亚语
       - value: cs
         label:
           en_US: Czech
@@ -309,14 +878,34 @@ parameters:
         label:
           en_US: Dutch
           zh_Hans: 荷兰语
+      - value: xx-elmer
+        label:
+          en_US: Elmer Fudd
+          zh_Hans: 艾尔默福德语
       - value: en
         label:
           en_US: English
           zh_Hans: 英语
+      - value: eo
+        label:
+          en_US: Esperanto
+          zh_Hans: 世界语
       - value: et
         label:
           en_US: Estonian
           zh_Hans: 爱沙尼亚语
+      - value: ee
+        label:
+          en_US: Ewe
+          zh_Hans: 埃维语
+      - value: fo
+        label:
+          en_US: Faroese
+          zh_Hans: 法罗语
+      - value: tl
+        label:
+          en_US: Filipino
+          zh_Hans: 菲律宾语
       - value: fi
         label:
           en_US: Finnish
@@ -325,6 +914,22 @@ parameters:
         label:
           en_US: French
           zh_Hans: 法语
+      - value: fy
+        label:
+          en_US: Frisian
+          zh_Hans: 弗里西亚语
+      - value: gaa
+        label:
+          en_US: Ga
+          zh_Hans: 加语
+      - value: gl
+        label:
+          en_US: Galician
+          zh_Hans: 加利西亚语
+      - value: ka
+        label:
+          en_US: Georgian
+          zh_Hans: 格鲁吉亚语
       - value: de
         label:
           en_US: German
@@ -333,6 +938,34 @@ parameters:
         label:
           en_US: Greek
           zh_Hans: 希腊语
+      - value: kl
+        label:
+          en_US: Greenlandic
+          zh_Hans: 格陵兰语
+      - value: gn
+        label:
+          en_US: Guarani
+          zh_Hans: 瓜拉尼语
+      - value: gu
+        label:
+          en_US: Gujarati
+          zh_Hans: 古吉拉特语
+      - value: xx-hacker
+        label:
+          en_US: Hacker
+          zh_Hans: 黑客语
+      - value: ht
+        label:
+          en_US: Haitian Creole
+          zh_Hans: 海地克里奥尔语
+      - value: ha
+        label:
+          en_US: Hausa
+          zh_Hans: 豪萨语
+      - value: haw
+        label:
+          en_US: Hawaiian
+          zh_Hans: 夏威夷语
       - value: iw
         label:
           en_US: Hebrew
@@ -345,10 +978,26 @@ parameters:
         label:
           en_US: Hungarian
           zh_Hans: 匈牙利语
+      - value: is
+        label:
+          en_US: Icelandic
+          zh_Hans: 冰岛语
+      - value: ig
+        label:
+          en_US: Igbo
+          zh_Hans: 伊博语
       - value: id
         label:
           en_US: Indonesian
           zh_Hans: 印尼语
+      - value: ia
+        label:
+          en_US: Interlingua
+          zh_Hans: 国际语
+      - value: ga
+        label:
+          en_US: Irish
+          zh_Hans: 爱尔兰语
       - value: it
         label:
           en_US: Italian
@@ -357,22 +1006,94 @@ parameters:
         label:
           en_US: Japanese
           zh_Hans: 日语
+      - value: jw
+        label:
+          en_US: Javanese
+          zh_Hans: 爪哇语
       - value: kn
         label:
           en_US: Kannada
           zh_Hans: 卡纳达语
+      - value: kk
+        label:
+          en_US: Kazakh
+          zh_Hans: 哈萨克语
+      - value: rw
+        label:
+          en_US: Kinyarwanda
+          zh_Hans: 基尼亚卢旺达语
+      - value: rn
+        label:
+          en_US: Kirundi
+          zh_Hans: 基隆迪语
+      - value: xx-klingon
+        label:
+          en_US: Klingon
+          zh_Hans: 克林贡语
+      - value: kg
+        label:
+          en_US: Kongo
+          zh_Hans: 刚果语
       - value: ko
         label:
           en_US: Korean
           zh_Hans: 韩语
+      - value: kri
+        label:
+          en_US: Krio (Sierra Leone)
+          zh_Hans: 塞拉利昂克里奥尔语
+      - value: ku
+        label:
+          en_US: Kurdish
+          zh_Hans: 库尔德语
+      - value: ckb
+        label:
+          en_US: Kurdish (Soranî)
+          zh_Hans: 库尔德语（索拉尼）
+      - value: ky
+        label:
+          en_US: Kyrgyz
+          zh_Hans: 吉尔吉斯语
+      - value: lo
+        label:
+          en_US: Laothian
+          zh_Hans: 老挝语
+      - value: la
+        label:
+          en_US: Latin
+          zh_Hans: 拉丁语
       - value: lv
         label:
           en_US: Latvian
           zh_Hans: 拉脱维亚语
+      - value: ln
+        label:
+          en_US: Lingala
+          zh_Hans: 林加拉语
       - value: lt
         label:
           en_US: Lithuanian
           zh_Hans: 立陶宛语
+      - value: loz
+        label:
+          en_US: Lozi
+          zh_Hans: 洛齐语
+      - value: lg
+        label:
+          en_US: Luganda
+          zh_Hans: 卢干达语
+      - value: ach
+        label:
+          en_US: Luo
+          zh_Hans: 卢奥语
+      - value: mk
+        label:
+          en_US: Macedonian
+          zh_Hans: 马其顿语
+      - value: mg
+        label:
+          en_US: Malagasy
+          zh_Hans: 马尔加什语
       - value: my
         label:
           en_US: Malay
@@ -381,18 +1102,90 @@ parameters:
         label:
           en_US: Malayalam
           zh_Hans: 马拉雅拉姆语
+      - value: mt
+        label:
+          en_US: Maltese
+          zh_Hans: 马耳他语
+      - value: mv
+        label:
+          en_US: Maldives
+          zh_Hans: 马尔代夫语
+      - value: mi
+        label:
+          en_US: Maori
+          zh_Hans: 毛利语
       - value: mr
         label:
           en_US: Marathi
           zh_Hans: 马拉地语
+      - value: mfe
+        label:
+          en_US: Mauritian Creole
+          zh_Hans: 毛里求斯克里奥尔语
+      - value: mo
+        label:
+          en_US: Moldavian
+          zh_Hans: 摩尔达维亚语
+      - value: mn
+        label:
+          en_US: Mongolian
+          zh_Hans: 蒙古语
+      - value: sr-me
+        label:
+          en_US: Montenegrin
+          zh_Hans: 黑山语
+      - value: ne
+        label:
+          en_US: Nepali
+          zh_Hans: 尼泊尔语
+      - value: pcm
+        label:
+          en_US: Nigerian Pidgin
+          zh_Hans: 尼日利亚皮钦语
+      - value: nso
+        label:
+          en_US: Northern Sotho
+          zh_Hans: 北索托语
       - value: "no"
         label:
           en_US: Norwegian
           zh_Hans: 挪威语
+      - value: nn
+        label:
+          en_US: Norwegian (Nynorsk)
+          zh_Hans: 挪威语（尼诺斯克语）
+      - value: oc
+        label:
+          en_US: Occitan
+          zh_Hans: 奥克语
+      - value: or
+        label:
+          en_US: Oriya
+          zh_Hans: 奥里亚语
+      - value: om
+        label:
+          en_US: Oromo
+          zh_Hans: 奥罗莫语
+      - value: ps
+        label:
+          en_US: Pashto
+          zh_Hans: 普什图语
+      - value: fa
+        label:
+          en_US: Persian
+          zh_Hans: 波斯语
+      - value: xx-pirate
+        label:
+          en_US: Pirate
+          zh_Hans: 海盗语
       - value: pl
         label:
           en_US: Polish
           zh_Hans: 波兰语
+      - value: pt
+        label:
+          en_US: Portuguese
+          zh_Hans: 葡萄牙语
       - value: pt-br
         label:
           en_US: Portuguese (Brazil)
@@ -405,18 +1198,62 @@ parameters:
         label:
           en_US: Punjabi
           zh_Hans: 旁遮普语
+      - value: qu
+        label:
+          en_US: Quechua
+          zh_Hans: 克丘亚语
       - value: ro
         label:
           en_US: Romanian
           zh_Hans: 罗马尼亚语
+      - value: rm
+        label:
+          en_US: Romansh
+          zh_Hans: 罗曼什语
+      - value: nyn
+        label:
+          en_US: Runyakitara
+          zh_Hans: 卢尼亚基塔拉语
       - value: ru
         label:
           en_US: Russian
           zh_Hans: 俄语
+      - value: gd
+        label:
+          en_US: Scots Gaelic
+          zh_Hans: 苏格兰盖尔语
       - value: sr
         label:
           en_US: Serbian
           zh_Hans: 塞尔维亚语
+      - value: sh
+        label:
+          en_US: Serbo-Croatian
+          zh_Hans: 塞尔维亚-克罗地亚语
+      - value: st
+        label:
+          en_US: Sesotho
+          zh_Hans: 塞索托语
+      - value: tn
+        label:
+          en_US: Setswana
+          zh_Hans: 塞茨瓦纳语
+      - value: crs
+        label:
+          en_US: Seychellois Creole
+          zh_Hans: 塞舌尔克里奥尔语
+      - value: sn
+        label:
+          en_US: Shona
+          zh_Hans: 绍纳语
+      - value: sd
+        label:
+          en_US: Sindhi
+          zh_Hans: 信德语
+      - value: si
+        label:
+          en_US: Sinhalese
+          zh_Hans: 僧伽罗语
       - value: sk
         label:
           en_US: Slovak
@@ -425,18 +1262,42 @@ parameters:
         label:
           en_US: Slovenian
           zh_Hans: 斯洛文尼亚语
+      - value: so
+        label:
+          en_US: Somali
+          zh_Hans: 索马里语
       - value: es
         label:
           en_US: Spanish
           zh_Hans: 西班牙语
+      - value: es-419
+        label:
+          en_US: Spanish (Latin American)
+          zh_Hans: 西班牙语（拉丁美洲）
+      - value: su
+        label:
+          en_US: Sundanese
+          zh_Hans: 巽他语
+      - value: sw
+        label:
+          en_US: Swahili
+          zh_Hans: 斯瓦希里语
       - value: sv
         label:
           en_US: Swedish
           zh_Hans: 瑞典语
+      - value: tg
+        label:
+          en_US: Tajik
+          zh_Hans: 塔吉克语
       - value: ta
         label:
           en_US: Tamil
           zh_Hans: 泰米尔语
+      - value: tt
+        label:
+          en_US: Tatar
+          zh_Hans: 鞑靼语
       - value: te
         label:
           en_US: Telugu
@@ -445,18 +1306,82 @@ parameters:
         label:
           en_US: Thai
           zh_Hans: 泰语
+      - value: ti
+        label:
+          en_US: Tigrinya
+          zh_Hans: 提格利尼亚语
+      - value: to
+        label:
+          en_US: Tonga
+          zh_Hans: 汤加语
+      - value: lua
+        label:
+          en_US: Tshiluba
+          zh_Hans: 卢巴语
+      - value: tum
+        label:
+          en_US: Tumbuka
+          zh_Hans: 图布卡语
       - value: tr
         label:
           en_US: Turkish
           zh_Hans: 土耳其语
+      - value: tk
+        label:
+          en_US: Turkmen
+          zh_Hans: 土库曼语
+      - value: tw
+        label:
+          en_US: Twi
+          zh_Hans: 契维语
+      - value: ug
+        label:
+          en_US: Uighur
+          zh_Hans: 维吾尔语
       - value: uk
         label:
           en_US: Ukrainian
           zh_Hans: 乌克兰语
+      - value: ur
+        label:
+          en_US: Urdu
+          zh_Hans: 乌尔都语
+      - value: uz
+        label:
+          en_US: Uzbek
+          zh_Hans: 乌兹别克语
+      - value: vu
+        label:
+          en_US: Vanuatu
+          zh_Hans: 瓦努阿图语
       - value: vi
         label:
           en_US: Vietnamese
           zh_Hans: 越南语
+      - value: cy
+        label:
+          en_US: Welsh
+          zh_Hans: 威尔士语
+      - value: wo
+        label:
+          en_US: Wolof
+          zh_Hans: 沃洛夫语
+      - value: xh
+        label:
+          en_US: Xhosa
+          zh_Hans: 科萨语
+      - value: yi
+        label:
+          en_US: Yiddish
+          zh_Hans: 意第绪语
+      - value: yo
+        label:
+          en_US: Yoruba
+          zh_Hans: 约鲁巴语
+      - value: zu
+        label:
+          en_US: Zulu
+          zh_Hans: 祖鲁语
   - name: is_remote
     type: select
     label:
diff --git a/api/core/tools/provider/builtin/searchapi/tools/google_news.yaml b/api/core/tools/provider/builtin/searchapi/tools/google_news.yaml
index cbb0edf982959..ff34af34cc9f5 100644
--- a/api/core/tools/provider/builtin/searchapi/tools/google_news.yaml
+++ b/api/core/tools/provider/builtin/searchapi/tools/google_news.yaml
@@ -65,206 +65,1206 @@ parameters:
     form: form
     default: US
     options:
+      - value: AF
+        label:
+          en_US: Afghanistan
+          zh_Hans: 阿富汗
+          pt_BR: Afeganistão
+      - value: AL
+        label:
+          en_US: Albania
+          zh_Hans: 阿尔巴尼亚
+          pt_BR: Albânia
+      - value: DZ
+        label:
+          en_US: Algeria
+          zh_Hans: 阿尔及利亚
+          pt_BR: Argélia
+      - value: AS
+        label:
+          en_US: American Samoa
+          zh_Hans: 美属萨摩亚
+          pt_BR: Samoa Americana
+      - value: AD
+        label:
+          en_US: Andorra
+          zh_Hans: 安道尔
+          pt_BR: Andorra
+      - value: AO
+        label:
+          en_US: Angola
+          zh_Hans: 安哥拉
+          pt_BR: Angola
+      - value: AI
+        label:
+          en_US: Anguilla
+          zh_Hans: 安圭拉
+          pt_BR: Anguilla
+      - value: AQ
+        label:
+          en_US: Antarctica
+          zh_Hans: 南极洲
+          pt_BR: Antártica
+      - value: AG
+        label:
+          en_US: Antigua and Barbuda
+          zh_Hans: 安提瓜和巴布达
+          pt_BR: Antígua e Barbuda
       - value: AR
         label:
           en_US: Argentina
           zh_Hans: 阿根廷
           pt_BR: Argentina
+      - value: AM
+        label:
+          en_US: Armenia
+          zh_Hans: 亚美尼亚
+          pt_BR: Armênia
+      - value: AW
+        label:
+          en_US: Aruba
+          zh_Hans: 阿鲁巴
+          pt_BR: Aruba
       - value: AU
         label:
           en_US: Australia
           zh_Hans: 澳大利亚
-          pt_BR: Australia
+          pt_BR: Austrália
       - value: AT
         label:
           en_US: Austria
           zh_Hans: 奥地利
-          pt_BR: Austria
+          pt_BR: Áustria
+      - value: AZ
+        label:
+          en_US: Azerbaijan
+          zh_Hans: 阿塞拜疆
+          pt_BR: Azerbaijão
+      - value: BS
+        label:
+          en_US: Bahamas
+          zh_Hans: 巴哈马
+          pt_BR: Bahamas
+      - value: BH
+        label:
+          en_US: Bahrain
+          zh_Hans: 巴林
+          pt_BR: Bahrein
+      - value: BD
+        label:
+          en_US: Bangladesh
+          zh_Hans: 孟加拉国
+          pt_BR: Bangladesh
+      - value: BB
+        label:
+          en_US: Barbados
+          zh_Hans: 巴巴多斯
+          pt_BR: Barbados
+      - value: BY
+        label:
+          en_US: Belarus
+          zh_Hans: 白俄罗斯
+          pt_BR: Bielorrússia
       - value: BE
         label:
           en_US: Belgium
           zh_Hans: 比利时
-          pt_BR: Belgium
+          pt_BR: Bélgica
+      - value: BZ
+        label:
+          en_US: Belize
+          zh_Hans: 伯利兹
+          pt_BR: Belize
+      - value: BJ
+        label:
+          en_US: Benin
+          zh_Hans: 贝宁
+          pt_BR: Benim
+      - value: BM
+        label:
+          en_US: Bermuda
+          zh_Hans: 百慕大
+          pt_BR: Bermudas
+      - value: BT
+        label:
+          en_US: Bhutan
+          zh_Hans: 不丹
+          pt_BR: Butão
+      - value: BO
+        label:
+          en_US: Bolivia
+          zh_Hans: 玻利维亚
+          pt_BR: Bolívia
+      - value: BA
+        label:
+          en_US: Bosnia and Herzegovina
+          zh_Hans: 波斯尼亚和黑塞哥维那
+          pt_BR: Bósnia e Herzegovina
+      - value: BW
+        label:
+          en_US: Botswana
+          zh_Hans: 博茨瓦纳
+          pt_BR: Botsuana
+      - value: BV
+        label:
+          en_US: Bouvet Island
+          zh_Hans: 布韦岛
+          pt_BR: Ilha Bouvet
       - value: BR
         label:
           en_US: Brazil
           zh_Hans: 巴西
-          pt_BR: Brazil
+          pt_BR: Brasil
+      - value: IO
+        label:
+          en_US: British Indian Ocean Territory
+          zh_Hans: 英属印度洋领地
+          pt_BR: Território Britânico do Oceano Índico
+      - value: BN
+        label:
+          en_US: Brunei Darussalam
+          zh_Hans: 文莱
+          pt_BR: Brunei Darussalam
+      - value: BG
+        label:
+          en_US: Bulgaria
+          zh_Hans: 保加利亚
+          pt_BR: Bulgária
+      - value: BF
+        label:
+          en_US: Burkina Faso
+          zh_Hans: 布基纳法索
+          pt_BR: Burkina Faso
+      - value: BI
+        label:
+          en_US: Burundi
+          zh_Hans: 布隆迪
+          pt_BR: Burundi
+      - value: KH
+        label:
+          en_US: Cambodia
+          zh_Hans: 柬埔寨
+          pt_BR: Camboja
+      - value: CM
+        label:
+          en_US: Cameroon
+          zh_Hans: 喀麦隆
+          pt_BR: Camarões
       - value: CA
         label:
           en_US: Canada
           zh_Hans: 加拿大
-          pt_BR: Canada
+          pt_BR: Canadá
+      - value: CV
+        label:
+          en_US: Cape Verde
+          zh_Hans: 佛得角
+          pt_BR: Cabo Verde
+      - value: KY
+        label:
+          en_US: Cayman Islands
+          zh_Hans: 开曼群岛
+          pt_BR: Ilhas Cayman
+      - value: CF
+        label:
+          en_US: Central African Republic
+          zh_Hans: 中非共和国
+          pt_BR: República Centro-Africana
+      - value: TD
+        label:
+          en_US: Chad
+          zh_Hans: 乍得
+          pt_BR: Chade
       - value: CL
         label:
           en_US: Chile
           zh_Hans: 智利
           pt_BR: Chile
-      - value: CO
-        label:
-          en_US: Colombia
-          zh_Hans: 哥伦比亚
-          pt_BR: Colombia
       - value: CN
         label:
           en_US: China
           zh_Hans: 中国
           pt_BR: China
+      - value: CX
+        label:
+          en_US: Christmas Island
+          zh_Hans: 圣诞岛
+          pt_BR: Ilha do Natal
+      - value: CC
+        label:
+          en_US: Cocos (Keeling) Islands
+          zh_Hans: 科科斯（基林）群岛
+          pt_BR: Ilhas Cocos (Keeling)
+      - value: CO
+        label:
+          en_US: Colombia
+          zh_Hans: 哥伦比亚
+          pt_BR: Colômbia
+      - value: KM
+        label:
+          en_US: Comoros
+          zh_Hans: 科摩罗
+          pt_BR: Comores
+      - value: CG
+        label:
+          en_US: Congo
+          zh_Hans: 刚果
+          pt_BR: Congo
+      - value: CD
+        label:
+          en_US: Congo, the Democratic Republic of the
+          zh_Hans: 刚果民主共和国
+          pt_BR: Congo, República Democrática do
+      - value: CK
+        label:
+          en_US: Cook Islands
+          zh_Hans: 库克群岛
+          pt_BR: Ilhas Cook
+      - value: CR
+        label:
+          en_US: Costa Rica
+          zh_Hans: 哥斯达黎加
+          pt_BR: Costa Rica
+      - value: CI
+        label:
+          en_US: Cote D'ivoire
+          zh_Hans: 科特迪瓦
+          pt_BR: Costa do Marfim
+      - value: HR
+        label:
+          en_US: Croatia
+          zh_Hans: 克罗地亚
+          pt_BR: Croácia
+      - value: CU
+        label:
+          en_US: Cuba
+          zh_Hans: 古巴
+          pt_BR: Cuba
+      - value: CY
+        label:
+          en_US: Cyprus
+          zh_Hans: 塞浦路斯
+          pt_BR: Chipre
       - value: CZ
         label:
           en_US: Czech Republic
           zh_Hans: 捷克共和国
-          pt_BR: Czech Republic
+          pt_BR: República Tcheca
       - value: DK
         label:
           en_US: Denmark
           zh_Hans: 丹麦
-          pt_BR: Denmark
+          pt_BR: Dinamarca
+      - value: DJ
+        label:
+          en_US: Djibouti
+          zh_Hans: 吉布提
+          pt_BR: Djibuti
+      - value: DM
+        label:
+          en_US: Dominica
+          zh_Hans: 多米尼克
+          pt_BR: Dominica
+      - value: DO
+        label:
+          en_US: Dominican Republic
+          zh_Hans: 多米尼加共和国
+          pt_BR: República Dominicana
+      - value: EC
+        label:
+          en_US: Ecuador
+          zh_Hans: 厄瓜多尔
+          pt_BR: Equador
+      - value: EG
+        label:
+          en_US: Egypt
+          zh_Hans: 埃及
+          pt_BR: Egito
+      - value: SV
+        label:
+          en_US: El Salvador
+          zh_Hans: 萨尔瓦多
+          pt_BR: El Salvador
+      - value: GQ
+        label:
+          en_US: Equatorial Guinea
+          zh_Hans: 赤道几内亚
+          pt_BR: Guiné Equatorial
+      - value: ER
+        label:
+          en_US: Eritrea
+          zh_Hans: 厄立特里亚
+          pt_BR: Eritreia
+      - value: EE
+        label:
+          en_US: Estonia
+          zh_Hans: 爱沙尼亚
+          pt_BR: Estônia
+      - value: ET
+        label:
+          en_US: Ethiopia
+          zh_Hans: 埃塞俄比亚
+          pt_BR: Etiópia
+      - value: FK
+        label:
+          en_US: Falkland Islands (Malvinas)
+          zh_Hans: 福克兰群岛（马尔维纳斯）
+          pt_BR: Ilhas Falkland (Malvinas)
+      - value: FO
+        label:
+          en_US: Faroe Islands
+          zh_Hans: 法罗群岛
+          pt_BR: Ilhas Faroe
+      - value: FJ
+        label:
+          en_US: Fiji
+          zh_Hans: 斐济
+          pt_BR: Fiji
       - value: FI
         label:
           en_US: Finland
           zh_Hans: 芬兰
-          pt_BR: Finland
+          pt_BR: Finlândia
       - value: FR
         label:
           en_US: France
           zh_Hans: 法国
-          pt_BR: France
+          pt_BR: França
+      - value: GF
+        label:
+          en_US: French Guiana
+          zh_Hans: 法属圭亚那
+          pt_BR: Guiana Francesa
+      - value: PF
+        label:
+          en_US: French Polynesia
+          zh_Hans: 法属波利尼西亚
+          pt_BR: Polinésia Francesa
+      - value: TF
+        label:
+          en_US: French Southern Territories
+          zh_Hans: 法属南部领地
+          pt_BR: Territórios Franceses do Sul
+      - value: GA
+        label:
+          en_US: Gabon
+          zh_Hans: 加蓬
+          pt_BR: Gabão
+      - value: GM
+        label:
+          en_US: Gambia
+          zh_Hans: 冈比亚
+          pt_BR: Gâmbia
+      - value: GE
+        label:
+          en_US: Georgia
+          zh_Hans: 格鲁吉亚
+          pt_BR: Geórgia
       - value: DE
         label:
           en_US: Germany
           zh_Hans: 德国
-          pt_BR: Germany
+          pt_BR: Alemanha
+      - value: GH
+        label:
+          en_US: Ghana
+          zh_Hans: 加纳
+          pt_BR: Gana
+      - value: GI
+        label:
+          en_US: Gibraltar
+          zh_Hans: 直布罗陀
+          pt_BR: Gibraltar
+      - value: GR
+        label:
+          en_US: Greece
+          zh_Hans: 希腊
+          pt_BR: Grécia
+      - value: GL
+        label:
+          en_US: Greenland
+          zh_Hans: 格陵兰
+          pt_BR: Groenlândia
+      - value: GD
+        label:
+          en_US: Grenada
+          zh_Hans: 格林纳达
+          pt_BR: Granada
+      - value: GP
+        label:
+          en_US: Guadeloupe
+          zh_Hans: 瓜德罗普
+          pt_BR: Guadalupe
+      - value: GU
+        label:
+          en_US: Guam
+          zh_Hans: 关岛
+          pt_BR: Guam
+      - value: GT
+        label:
+          en_US: Guatemala
+          zh_Hans: 危地马拉
+          pt_BR: Guatemala
+      - value: GN
+        label:
+          en_US: Guinea
+          zh_Hans: 几内亚
+          pt_BR: Guiné
+      - value: GW
+        label:
+          en_US: Guinea-Bissau
+          zh_Hans: 几内亚比绍
+          pt_BR: Guiné-Bissau
+      - value: GY
+        label:
+          en_US: Guyana
+          zh_Hans: 圭亚那
+          pt_BR: Guiana
+      - value: HT
+        label:
+          en_US: Haiti
+          zh_Hans: 海地
+          pt_BR: Haiti
+      - value: HM
+        label:
+          en_US: Heard Island and McDonald Islands
+          zh_Hans: 赫德岛和麦克唐纳群岛
+          pt_BR: Ilha Heard e Ilhas McDonald
+      - value: VA
+        label:
+          en_US: Holy See (Vatican City State)
+          zh_Hans: 教廷（梵蒂冈城国）
+          pt_BR: Santa Sé (Estado da Cidade do Vaticano)
+      - value: HN
+        label:
+          en_US: Honduras
+          zh_Hans: 洪都拉斯
+          pt_BR: Honduras
       - value: HK
         label:
           en_US: Hong Kong
           zh_Hans: 香港
           pt_BR: Hong Kong
+      - value: HU
+        label:
+          en_US: Hungary
+          zh_Hans: 匈牙利
+          pt_BR: Hungria
+      - value: IS
+        label:
+          en_US: Iceland
+          zh_Hans: 冰岛
+          pt_BR: Islândia
       - value: IN
         label:
           en_US: India
           zh_Hans: 印度
-          pt_BR: India
+          pt_BR: Índia
       - value: ID
         label:
           en_US: Indonesia
           zh_Hans: 印度尼西亚
-          pt_BR: Indonesia
+          pt_BR: Indonésia
+      - value: IR
+        label:
+          en_US: Iran, Islamic Republic of
+          zh_Hans: 伊朗
+          pt_BR: Irã
+      - value: IQ
+        label:
+          en_US: Iraq
+          zh_Hans: 伊拉克
+          pt_BR: Iraque
+      - value: IE
+        label:
+          en_US: Ireland
+          zh_Hans: 爱尔兰
+          pt_BR: Irlanda
+      - value: IL
+        label:
+          en_US: Israel
+          zh_Hans: 以色列
+          pt_BR: Israel
       - value: IT
         label:
           en_US: Italy
           zh_Hans: 意大利
-          pt_BR: Italy
+          pt_BR: Itália
+      - value: JM
+        label:
+          en_US: Jamaica
+          zh_Hans: 牙买加
+          pt_BR: Jamaica
       - value: JP
         label:
           en_US: Japan
           zh_Hans: 日本
-          pt_BR: Japan
+          pt_BR: Japão
+      - value: JO
+        label:
+          en_US: Jordan
+          zh_Hans: 约旦
+          pt_BR: Jordânia
+      - value: KZ
+        label:
+          en_US: Kazakhstan
+          zh_Hans: 哈萨克斯坦
+          pt_BR: Cazaquistão
+      - value: KE
+        label:
+          en_US: Kenya
+          zh_Hans: 肯尼亚
+          pt_BR: Quênia
+      - value: KI
+        label:
+          en_US: Kiribati
+          zh_Hans: 基里巴斯
+          pt_BR: Kiribati
+      - value: KP
+        label:
+          en_US: Korea, Democratic People's Republic of
+          zh_Hans: 朝鲜
+          pt_BR: Coreia, República Democrática Popular da
       - value: KR
         label:
-          en_US: Korea
+          en_US: Korea, Republic of
           zh_Hans: 韩国
-          pt_BR: Korea
+          pt_BR: Coreia, República da
+      - value: KW
+        label:
+          en_US: Kuwait
+          zh_Hans: 科威特
+          pt_BR: Kuwait
+      - value: KG
+        label:
+          en_US: Kyrgyzstan
+          zh_Hans: 吉尔吉斯斯坦
+          pt_BR: Quirguistão
+      - value: LA
+        label:
+          en_US: Lao People's Democratic Republic
+          zh_Hans: 老挝
+          pt_BR: República Democrática Popular do Laos
+      - value: LV
+        label:
+          en_US: Latvia
+          zh_Hans: 拉脱维亚
+          pt_BR: Letônia
+      - value: LB
+        label:
+          en_US: Lebanon
+          zh_Hans: 黎巴嫩
+          pt_BR: Líbano
+      - value: LS
+        label:
+          en_US: Lesotho
+          zh_Hans: 莱索托
+          pt_BR: Lesoto
+      - value: LR
+        label:
+          en_US: Liberia
+          zh_Hans: 利比里亚
+          pt_BR: Libéria
+      - value: LY
+        label:
+          en_US: Libyan Arab Jamahiriya
+          zh_Hans: 利比亚
+          pt_BR: Líbia
+      - value: LI
+        label:
+          en_US: Liechtenstein
+          zh_Hans: 列支敦士登
+          pt_BR: Liechtenstein
+      - value: LT
+        label:
+          en_US: Lithuania
+          zh_Hans: 立陶宛
+          pt_BR: Lituânia
+      - value: LU
+        label:
+          en_US: Luxembourg
+          zh_Hans: 卢森堡
+          pt_BR: Luxemburgo
+      - value: MO
+        label:
+          en_US: Macao
+          zh_Hans: 澳门
+          pt_BR: Macau
+      - value: MK
+        label:
+          en_US: Macedonia, the Former Yugosalv Republic of
+          zh_Hans: 前南斯拉夫马其顿共和国
+          pt_BR: Macedônia, Ex-República Iugoslava da
+      - value: MG
+        label:
+          en_US: Madagascar
+          zh_Hans: 马达加斯加
+          pt_BR: Madagascar
+      - value: MW
+        label:
+          en_US: Malawi
+          zh_Hans: 马拉维
+          pt_BR: Malaui
       - value: MY
         label:
           en_US: Malaysia
           zh_Hans: 马来西亚
-          pt_BR: Malaysia
+          pt_BR: Malásia
+      - value: MV
+        label:
+          en_US: Maldives
+          zh_Hans: 马尔代夫
+          pt_BR: Maldivas
+      - value: ML
+        label:
+          en_US: Mali
+          zh_Hans: 马里
+          pt_BR: Mali
+      - value: MT
+        label:
+          en_US: Malta
+          zh_Hans: 马耳他
+          pt_BR: Malta
+      - value: MH
+        label:
+          en_US: Marshall Islands
+          zh_Hans: 马绍尔群岛
+          pt_BR: Ilhas Marshall
+      - value: MQ
+        label:
+          en_US: Martinique
+          zh_Hans: 马提尼克
+          pt_BR: Martinica
+      - value: MR
+        label:
+          en_US: Mauritania
+          zh_Hans: 毛里塔尼亚
+          pt_BR: Mauritânia
+      - value: MU
+        label:
+          en_US: Mauritius
+          zh_Hans: 毛里求斯
+          pt_BR: Maurício
+      - value: YT
+        label:
+          en_US: Mayotte
+          zh_Hans: 马约特
+          pt_BR: Mayotte
       - value: MX
         label:
           en_US: Mexico
           zh_Hans: 墨西哥
-          pt_BR: Mexico
+          pt_BR: México
+      - value: FM
+        label:
+          en_US: Micronesia, Federated States of
+          zh_Hans: 密克罗尼西亚联邦
+          pt_BR: Micronésia, Estados Federados da
+      - value: MD
+        label:
+          en_US: Moldova, Republic of
+          zh_Hans: 摩尔多瓦共和国
+          pt_BR: Moldávia, República da
+      - value: MC
+        label:
+          en_US: Monaco
+          zh_Hans: 摩纳哥
+          pt_BR: Mônaco
+      - value: MN
+        label:
+          en_US: Mongolia
+          zh_Hans: 蒙古
+          pt_BR: Mongólia
+      - value: MS
+        label:
+          en_US: Montserrat
+          zh_Hans: 蒙特塞拉特
+          pt_BR: Montserrat
+      - value: MA
+        label:
+          en_US: Morocco
+          zh_Hans: 摩洛哥
+          pt_BR: Marrocos
+      - value: MZ
+        label:
+          en_US: Mozambique
+          zh_Hans: 莫桑比克
+          pt_BR: Moçambique
+      - value: MM
+        label:
+          en_US: Myanmar
+          zh_Hans: 缅甸
+          pt_BR: Mianmar
+      - value: NA
+        label:
+          en_US: Namibia
+          zh_Hans: 纳米比亚
+          pt_BR: Namíbia
+      - value: NR
+        label:
+          en_US: Nauru
+          zh_Hans: 瑙鲁
+          pt_BR: Nauru
+      - value: NP
+        label:
+          en_US: Nepal
+          zh_Hans: 尼泊尔
+          pt_BR: Nepal
       - value: NL
         label:
           en_US: Netherlands
           zh_Hans: 荷兰
-          pt_BR: Netherlands
+          pt_BR: Países Baixos
+      - value: AN
+        label:
+          en_US: Netherlands Antilles
+          zh_Hans: 荷属安的列斯
+          pt_BR: Antilhas Holandesas
+      - value: NC
+        label:
+          en_US: New Caledonia
+          zh_Hans: 新喀里多尼亚
+          pt_BR: Nova Caledônia
       - value: NZ
         label:
           en_US: New Zealand
           zh_Hans: 新西兰
-          pt_BR: New Zealand
-      - value: 'NO'
+          pt_BR: Nova Zelândia
+      - value: NI
+        label:
+          en_US: Nicaragua
+          zh_Hans: 尼加拉瓜
+          pt_BR: Nicarágua
+      - value: NE
+        label:
+          en_US: Niger
+          zh_Hans: 尼日尔
+          pt_BR: Níger
+      - value: NG
+        label:
+          en_US: Nigeria
+          zh_Hans: 尼日利亚
+          pt_BR: Nigéria
+      - value: NU
+        label:
+          en_US: Niue
+          zh_Hans: 纽埃
+          pt_BR: Niue
+      - value: NF
+        label:
+          en_US: Norfolk Island
+          zh_Hans: 诺福克岛
+          pt_BR: Ilha Norfolk
+      - value: MP
+        label:
+          en_US: Northern Mariana Islands
+          zh_Hans: 北马里亚纳群岛
+          pt_BR: Ilhas Marianas do Norte
+      - value: "NO"
         label:
           en_US: Norway
           zh_Hans: 挪威
-          pt_BR: Norway
+          pt_BR: Noruega
+      - value: OM
+        label:
+          en_US: Oman
+          zh_Hans: 阿曼
+          pt_BR: Omã
+      - value: PK
+        label:
+          en_US: Pakistan
+          zh_Hans: 巴基斯坦
+          pt_BR: Paquistão
+      - value: PW
+        label:
+          en_US: Palau
+          zh_Hans: 帕劳
+          pt_BR: Palau
+      - value: PS
+        label:
+          en_US: Palestinian Territory, Occupied
+          zh_Hans: 巴勒斯坦领土
+          pt_BR: Palestina, Território Ocupado
+      - value: PA
+        label:
+          en_US: Panama
+          zh_Hans: 巴拿马
+          pt_BR: Panamá
+      - value: PG
+        label:
+          en_US: Papua New Guinea
+          zh_Hans: 巴布亚新几内亚
+          pt_BR: Papua Nova Guiné
+      - value: PY
+        label:
+          en_US: Paraguay
+          zh_Hans: 巴拉圭
+          pt_BR: Paraguai
+      - value: PE
+        label:
+          en_US: Peru
+          zh_Hans: 秘鲁
+          pt_BR: Peru
       - value: PH
         label:
           en_US: Philippines
           zh_Hans: 菲律宾
-          pt_BR: Philippines
+          pt_BR: Filipinas
+      - value: PN
+        label:
+          en_US: Pitcairn
+          zh_Hans: 皮特凯恩岛
+          pt_BR: Pitcairn
       - value: PL
         label:
           en_US: Poland
           zh_Hans: 波兰
-          pt_BR: Poland
+          pt_BR: Polônia
       - value: PT
         label:
           en_US: Portugal
           zh_Hans: 葡萄牙
           pt_BR: Portugal
+      - value: PR
+        label:
+          en_US: Puerto Rico
+          zh_Hans: 波多黎各
+          pt_BR: Porto Rico
+      - value: QA
+        label:
+          en_US: Qatar
+          zh_Hans: 卡塔尔
+          pt_BR: Catar
+      - value: RE
+        label:
+          en_US: Reunion
+          zh_Hans: 留尼旺
+          pt_BR: Reunião
+      - value: RO
+        label:
+          en_US: Romania
+          zh_Hans: 罗马尼亚
+          pt_BR: Romênia
       - value: RU
         label:
-          en_US: Russia
-          zh_Hans: 俄罗斯
-          pt_BR: Russia
+          en_US: Russian Federation
+          zh_Hans: 俄罗斯联邦
+          pt_BR: Rússia
+      - value: RW
+        label:
+          en_US: Rwanda
+          zh_Hans: 卢旺达
+          pt_BR: Ruanda
+      - value: SH
+        label:
+          en_US: Saint Helena
+          zh_Hans: 圣赫勒拿
+          pt_BR: Santa Helena
+      - value: KN
+        label:
+          en_US: Saint Kitts and Nevis
+          zh_Hans: 圣基茨和尼维斯
+          pt_BR: São Cristóvão e Nevis
+      - value: LC
+        label:
+          en_US: Saint Lucia
+          zh_Hans: 圣卢西亚
+          pt_BR: Santa Lúcia
+      - value: PM
+        label:
+          en_US: Saint Pierre and Miquelon
+          zh_Hans: 圣皮埃尔和密克隆
+          pt_BR: São Pedro e Miquelon
+      - value: VC
+        label:
+          en_US: Saint Vincent and the Grenadines
+          zh_Hans: 圣文森特和格林纳丁斯
+          pt_BR: São Vicente e Granadinas
+      - value: WS
+        label:
+          en_US: Samoa
+          zh_Hans: 萨摩亚
+          pt_BR: Samoa
+      - value: SM
+        label:
+          en_US: San Marino
+          zh_Hans: 圣马力诺
+          pt_BR: San Marino
+      - value: ST
+        label:
+          en_US: Sao Tome and Principe
+          zh_Hans: 圣多美和普林西比
+          pt_BR: São Tomé e Príncipe
       - value: SA
         label:
           en_US: Saudi Arabia
           zh_Hans: 沙特阿拉伯
-          pt_BR: Saudi Arabia
+          pt_BR: Arábia Saudita
+      - value: SN
+        label:
+          en_US: Senegal
+          zh_Hans: 塞内加尔
+          pt_BR: Senegal
+      - value: RS
+        label:
+          en_US: Serbia and Montenegro
+          zh_Hans: 塞尔维亚和黑山
+          pt_BR: Sérvia e Montenegro
+      - value: SC
+        label:
+          en_US: Seychelles
+          zh_Hans: 塞舌尔
+          pt_BR: Seicheles
+      - value: SL
+        label:
+          en_US: Sierra Leone
+          zh_Hans: 塞拉利昂
+          pt_BR: Serra Leoa
       - value: SG
         label:
           en_US: Singapore
           zh_Hans: 新加坡
-          pt_BR: Singapore
+          pt_BR: Singapura
+      - value: SK
+        label:
+          en_US: Slovakia
+          zh_Hans: 斯洛伐克
+          pt_BR: Eslováquia
+      - value: SI
+        label:
+          en_US: Slovenia
+          zh_Hans: 斯洛文尼亚
+          pt_BR: Eslovênia
+      - value: SB
+        label:
+          en_US: Solomon Islands
+          zh_Hans: 所罗门群岛
+          pt_BR: Ilhas Salomão
+      - value: SO
+        label:
+          en_US: Somalia
+          zh_Hans: 索马里
+          pt_BR: Somália
       - value: ZA
         label:
           en_US: South Africa
           zh_Hans: 南非
-          pt_BR: South Africa
+          pt_BR: África do Sul
+      - value: GS
+        label:
+          en_US: South Georgia and the South Sandwich Islands
+          zh_Hans: 南乔治亚和南桑威奇群岛
+          pt_BR: Geórgia do Sul e Ilhas Sandwich do Sul
       - value: ES
         label:
           en_US: Spain
           zh_Hans: 西班牙
-          pt_BR: Spain
+          pt_BR: Espanha
+      - value: LK
+        label:
+          en_US: Sri Lanka
+          zh_Hans: 斯里兰卡
+          pt_BR: Sri Lanka
+      - value: SD
+        label:
+          en_US: Sudan
+          zh_Hans: 苏丹
+          pt_BR: Sudão
+      - value: SR
+        label:
+          en_US: Suriname
+          zh_Hans: 苏里南
+          pt_BR: Suriname
+      - value: SJ
+        label:
+          en_US: Svalbard and Jan Mayen
+          zh_Hans: 斯瓦尔巴特和扬马延岛
+          pt_BR: Svalbard e Jan Mayen
+      - value: SZ
+        label:
+          en_US: Swaziland
+          zh_Hans: 斯威士兰
+          pt_BR: Essuatíni
       - value: SE
         label:
           en_US: Sweden
           zh_Hans: 瑞典
-          pt_BR: Sweden
+          pt_BR: Suécia
       - value: CH
         label:
           en_US: Switzerland
           zh_Hans: 瑞士
-          pt_BR: Switzerland
+          pt_BR: Suíça
+      - value: SY
+        label:
+          en_US: Syrian Arab Republic
+          zh_Hans: 叙利亚
+          pt_BR: Síria
       - value: TW
         label:
-          en_US: Taiwan
+          en_US: Taiwan, Province of China
           zh_Hans: 台湾
           pt_BR: Taiwan
+      - value: TJ
+        label:
+          en_US: Tajikistan
+          zh_Hans: 塔吉克斯坦
+          pt_BR: Tajiquistão
+      - value: TZ
+        label:
+          en_US: Tanzania, United Republic of
+          zh_Hans: 坦桑尼亚联合共和国
+          pt_BR: Tanzânia
       - value: TH
         label:
           en_US: Thailand
           zh_Hans: 泰国
-          pt_BR: Thailand
+          pt_BR: Tailândia
+      - value: TL
+        label:
+          en_US: Timor-Leste
+          zh_Hans: 东帝汶
+          pt_BR: Timor-Leste
+      - value: TG
+        label:
+          en_US: Togo
+          zh_Hans: 多哥
+          pt_BR: Togo
+      - value: TK
+        label:
+          en_US: Tokelau
+          zh_Hans: 托克劳
+          pt_BR: Toquelau
+      - value: TO
+        label:
+          en_US: Tonga
+          zh_Hans: 汤加
+          pt_BR: Tonga
+      - value: TT
+        label:
+          en_US: Trinidad and Tobago
+          zh_Hans: 特立尼达和多巴哥
+          pt_BR: Trindade e Tobago
+      - value: TN
+        label:
+          en_US: Tunisia
+          zh_Hans: 突尼斯
+          pt_BR: Tunísia
       - value: TR
         label:
           en_US: Turkey
           zh_Hans: 土耳其
-          pt_BR: Turkey
+          pt_BR: Turquia
+      - value: TM
+        label:
+          en_US: Turkmenistan
+          zh_Hans: 土库曼斯坦
+          pt_BR: Turcomenistão
+      - value: TC
+        label:
+          en_US: Turks and Caicos Islands
+          zh_Hans: 特克斯和凯科斯群岛
+          pt_BR: Ilhas Turks e Caicos
+      - value: TV
+        label:
+          en_US: Tuvalu
+          zh_Hans: 图瓦卢
+          pt_BR: Tuvalu
+      - value: UG
+        label:
+          en_US: Uganda
+          zh_Hans: 乌干达
+          pt_BR: Uganda
+      - value: UA
+        label:
+          en_US: Ukraine
+          zh_Hans: 乌克兰
+          pt_BR: Ucrânia
+      - value: AE
+        label:
+          en_US: United Arab Emirates
+          zh_Hans: 阿联酋
+          pt_BR: Emirados Árabes Unidos
+      - value: UK
+        label:
+          en_US: United Kingdom
+          zh_Hans: 英国
+          pt_BR: Reino Unido
       - value: GB
         label:
           en_US: United Kingdom
           zh_Hans: 英国
-          pt_BR: United Kingdom
+          pt_BR: Reino Unido
       - value: US
         label:
           en_US: United States
           zh_Hans: 美国
-          pt_BR: United States
+          pt_BR: Estados Unidos
+      - value: UM
+        label:
+          en_US: United States Minor Outlying Islands
+          zh_Hans: 美国本土外小岛屿
+          pt_BR: Ilhas Menores Distantes dos Estados Unidos
+      - value: UY
+        label:
+          en_US: Uruguay
+          zh_Hans: 乌拉圭
+          pt_BR: Uruguai
+      - value: UZ
+        label:
+          en_US: Uzbekistan
+          zh_Hans: 乌兹别克斯坦
+          pt_BR: Uzbequistão
+      - value: VU
+        label:
+          en_US: Vanuatu
+          zh_Hans: 瓦努阿图
+          pt_BR: Vanuatu
+      - value: VE
+        label:
+          en_US: Venezuela
+          zh_Hans: 委内瑞拉
+          pt_BR: Venezuela
+      - value: VN
+        label:
+          en_US: Viet Nam
+          zh_Hans: 越南
+          pt_BR: Vietnã
+      - value: VG
+        label:
+          en_US: Virgin Islands, British
+          zh_Hans: 英属维尔京群岛
+          pt_BR: Ilhas Virgens Britânicas
+      - value: VI
+        label:
+          en_US: Virgin Islands, U.S.
+          zh_Hans: 美属维尔京群岛
+          pt_BR: Ilhas Virgens dos EUA
+      - value: WF
+        label:
+          en_US: Wallis and Futuna
+          zh_Hans: 瓦利斯和富图纳群岛
+          pt_BR: Wallis e Futuna
+      - value: EH
+        label:
+          en_US: Western Sahara
+          zh_Hans: 西撒哈拉
+          pt_BR: Saara Ocidental
+      - value: YE
+        label:
+          en_US: Yemen
+          zh_Hans: 也门
+          pt_BR: Iémen
+      - value: ZM
+        label:
+          en_US: Zambia
+          zh_Hans: 赞比亚
+          pt_BR: Zâmbia
+      - value: ZW
+        label:
+          en_US: Zimbabwe
+          zh_Hans: 津巴布韦
+          pt_BR: Zimbábue
   - name: hl
     type: select
     label:
@@ -277,18 +1277,94 @@ parameters:
     default: en
     form: form
     options:
+      - value: af
+        label:
+          en_US: Afrikaans
+          zh_Hans: 南非语
+      - value: ak
+        label:
+          en_US: Akan
+          zh_Hans: 阿坎语
+      - value: sq
+        label:
+          en_US: Albanian
+          zh_Hans: 阿尔巴尼亚语
+      - value: ws
+        label:
+          en_US: Samoa
+          zh_Hans: 萨摩亚语
+      - value: am
+        label:
+          en_US: Amharic
+          zh_Hans: 阿姆哈拉语
       - value: ar
         label:
           en_US: Arabic
           zh_Hans: 阿拉伯语
+      - value: hy
+        label:
+          en_US: Armenian
+          zh_Hans: 亚美尼亚语
+      - value: az
+        label:
+          en_US: Azerbaijani
+          zh_Hans: 阿塞拜疆语
+      - value: eu
+        label:
+          en_US: Basque
+          zh_Hans: 巴斯克语
+      - value: be
+        label:
+          en_US: Belarusian
+          zh_Hans: 白俄罗斯语
+      - value: bem
+        label:
+          en_US: Bemba
+          zh_Hans: 班巴语
+      - value: bn
+        label:
+          en_US: Bengali
+          zh_Hans: 孟加拉语
+      - value: bh
+        label:
+          en_US: Bihari
+          zh_Hans: 比哈尔语
+      - value: xx-bork
+        label:
+          en_US: Bork, bork, bork!
+          zh_Hans: 博克语
+      - value: bs
+        label:
+          en_US: Bosnian
+          zh_Hans: 波斯尼亚语
+      - value: br
+        label:
+          en_US: Breton
+          zh_Hans: 布列塔尼语
       - value: bg
         label:
           en_US: Bulgarian
           zh_Hans: 保加利亚语
+      - value: bt
+        label:
+          en_US: Bhutanese
+          zh_Hans: 不丹语
+      - value: km
+        label:
+          en_US: Cambodian
+          zh_Hans: 高棉语
       - value: ca
         label:
           en_US: Catalan
           zh_Hans: 加泰罗尼亚语
+      - value: chr
+        label:
+          en_US: Cherokee
+          zh_Hans: 切罗基语
+      - value: ny
+        label:
+          en_US: Chichewa
+          zh_Hans: 齐切瓦语
       - value: zh-cn
         label:
           en_US: Chinese (Simplified)
@@ -297,6 +1373,14 @@ parameters:
         label:
           en_US: Chinese (Traditional)
           zh_Hans: 中文（繁体）
+      - value: co
+        label:
+          en_US: Corsican
+          zh_Hans: 科西嘉语
+      - value: hr
+        label:
+          en_US: Croatian
+          zh_Hans: 克罗地亚语
       - value: cs
         label:
           en_US: Czech
@@ -309,14 +1393,34 @@ parameters:
         label:
           en_US: Dutch
           zh_Hans: 荷兰语
+      - value: xx-elmer
+        label:
+          en_US: Elmer Fudd
+          zh_Hans: 艾尔默福德语
       - value: en
         label:
           en_US: English
           zh_Hans: 英语
+      - value: eo
+        label:
+          en_US: Esperanto
+          zh_Hans: 世界语
       - value: et
         label:
           en_US: Estonian
           zh_Hans: 爱沙尼亚语
+      - value: ee
+        label:
+          en_US: Ewe
+          zh_Hans: 埃维语
+      - value: fo
+        label:
+          en_US: Faroese
+          zh_Hans: 法罗语
+      - value: tl
+        label:
+          en_US: Filipino
+          zh_Hans: 菲律宾语
       - value: fi
         label:
           en_US: Finnish
@@ -325,6 +1429,22 @@ parameters:
         label:
           en_US: French
           zh_Hans: 法语
+      - value: fy
+        label:
+          en_US: Frisian
+          zh_Hans: 弗里西亚语
+      - value: gaa
+        label:
+          en_US: Ga
+          zh_Hans: 加语
+      - value: gl
+        label:
+          en_US: Galician
+          zh_Hans: 加利西亚语
+      - value: ka
+        label:
+          en_US: Georgian
+          zh_Hans: 格鲁吉亚语
       - value: de
         label:
           en_US: German
@@ -333,6 +1453,34 @@ parameters:
         label:
           en_US: Greek
           zh_Hans: 希腊语
+      - value: kl
+        label:
+          en_US: Greenlandic
+          zh_Hans: 格陵兰语
+      - value: gn
+        label:
+          en_US: Guarani
+          zh_Hans: 瓜拉尼语
+      - value: gu
+        label:
+          en_US: Gujarati
+          zh_Hans: 古吉拉特语
+      - value: xx-hacker
+        label:
+          en_US: Hacker
+          zh_Hans: 黑客语
+      - value: ht
+        label:
+          en_US: Haitian Creole
+          zh_Hans: 海地克里奥尔语
+      - value: ha
+        label:
+          en_US: Hausa
+          zh_Hans: 豪萨语
+      - value: haw
+        label:
+          en_US: Hawaiian
+          zh_Hans: 夏威夷语
       - value: iw
         label:
           en_US: Hebrew
@@ -345,10 +1493,26 @@ parameters:
         label:
           en_US: Hungarian
           zh_Hans: 匈牙利语
+      - value: is
+        label:
+          en_US: Icelandic
+          zh_Hans: 冰岛语
+      - value: ig
+        label:
+          en_US: Igbo
+          zh_Hans: 伊博语
       - value: id
         label:
           en_US: Indonesian
           zh_Hans: 印尼语
+      - value: ia
+        label:
+          en_US: Interlingua
+          zh_Hans: 国际语
+      - value: ga
+        label:
+          en_US: Irish
+          zh_Hans: 爱尔兰语
       - value: it
         label:
           en_US: Italian
@@ -357,22 +1521,94 @@ parameters:
         label:
           en_US: Japanese
           zh_Hans: 日语
+      - value: jw
+        label:
+          en_US: Javanese
+          zh_Hans: 爪哇语
       - value: kn
         label:
           en_US: Kannada
           zh_Hans: 卡纳达语
+      - value: kk
+        label:
+          en_US: Kazakh
+          zh_Hans: 哈萨克语
+      - value: rw
+        label:
+          en_US: Kinyarwanda
+          zh_Hans: 基尼亚卢旺达语
+      - value: rn
+        label:
+          en_US: Kirundi
+          zh_Hans: 基隆迪语
+      - value: xx-klingon
+        label:
+          en_US: Klingon
+          zh_Hans: 克林贡语
+      - value: kg
+        label:
+          en_US: Kongo
+          zh_Hans: 刚果语
       - value: ko
         label:
           en_US: Korean
           zh_Hans: 韩语
+      - value: kri
+        label:
+          en_US: Krio (Sierra Leone)
+          zh_Hans: 塞拉利昂克里奥尔语
+      - value: ku
+        label:
+          en_US: Kurdish
+          zh_Hans: 库尔德语
+      - value: ckb
+        label:
+          en_US: Kurdish (Soranî)
+          zh_Hans: 库尔德语（索拉尼）
+      - value: ky
+        label:
+          en_US: Kyrgyz
+          zh_Hans: 吉尔吉斯语
+      - value: lo
+        label:
+          en_US: Laothian
+          zh_Hans: 老挝语
+      - value: la
+        label:
+          en_US: Latin
+          zh_Hans: 拉丁语
       - value: lv
         label:
           en_US: Latvian
           zh_Hans: 拉脱维亚语
+      - value: ln
+        label:
+          en_US: Lingala
+          zh_Hans: 林加拉语
       - value: lt
         label:
           en_US: Lithuanian
           zh_Hans: 立陶宛语
+      - value: loz
+        label:
+          en_US: Lozi
+          zh_Hans: 洛齐语
+      - value: lg
+        label:
+          en_US: Luganda
+          zh_Hans: 卢干达语
+      - value: ach
+        label:
+          en_US: Luo
+          zh_Hans: 卢奥语
+      - value: mk
+        label:
+          en_US: Macedonian
+          zh_Hans: 马其顿语
+      - value: mg
+        label:
+          en_US: Malagasy
+          zh_Hans: 马尔加什语
       - value: my
         label:
           en_US: Malay
@@ -381,18 +1617,90 @@ parameters:
         label:
           en_US: Malayalam
           zh_Hans: 马拉雅拉姆语
+      - value: mt
+        label:
+          en_US: Maltese
+          zh_Hans: 马耳他语
+      - value: mv
+        label:
+          en_US: Maldives
+          zh_Hans: 马尔代夫语
+      - value: mi
+        label:
+          en_US: Maori
+          zh_Hans: 毛利语
       - value: mr
         label:
           en_US: Marathi
           zh_Hans: 马拉地语
+      - value: mfe
+        label:
+          en_US: Mauritian Creole
+          zh_Hans: 毛里求斯克里奥尔语
+      - value: mo
+        label:
+          en_US: Moldavian
+          zh_Hans: 摩尔达维亚语
+      - value: mn
+        label:
+          en_US: Mongolian
+          zh_Hans: 蒙古语
+      - value: sr-me
+        label:
+          en_US: Montenegrin
+          zh_Hans: 黑山语
+      - value: ne
+        label:
+          en_US: Nepali
+          zh_Hans: 尼泊尔语
+      - value: pcm
+        label:
+          en_US: Nigerian Pidgin
+          zh_Hans: 尼日利亚皮钦语
+      - value: nso
+        label:
+          en_US: Northern Sotho
+          zh_Hans: 北索托语
       - value: "no"
         label:
           en_US: Norwegian
           zh_Hans: 挪威语
+      - value: nn
+        label:
+          en_US: Norwegian (Nynorsk)
+          zh_Hans: 挪威语（尼诺斯克语）
+      - value: oc
+        label:
+          en_US: Occitan
+          zh_Hans: 奥克语
+      - value: or
+        label:
+          en_US: Oriya
+          zh_Hans: 奥里亚语
+      - value: om
+        label:
+          en_US: Oromo
+          zh_Hans: 奥罗莫语
+      - value: ps
+        label:
+          en_US: Pashto
+          zh_Hans: 普什图语
+      - value: fa
+        label:
+          en_US: Persian
+          zh_Hans: 波斯语
+      - value: xx-pirate
+        label:
+          en_US: Pirate
+          zh_Hans: 海盗语
       - value: pl
         label:
           en_US: Polish
           zh_Hans: 波兰语
+      - value: pt
+        label:
+          en_US: Portuguese
+          zh_Hans: 葡萄牙语
       - value: pt-br
         label:
           en_US: Portuguese (Brazil)
@@ -405,18 +1713,62 @@ parameters:
         label:
           en_US: Punjabi
           zh_Hans: 旁遮普语
+      - value: qu
+        label:
+          en_US: Quechua
+          zh_Hans: 克丘亚语
       - value: ro
         label:
           en_US: Romanian
           zh_Hans: 罗马尼亚语
+      - value: rm
+        label:
+          en_US: Romansh
+          zh_Hans: 罗曼什语
+      - value: nyn
+        label:
+          en_US: Runyakitara
+          zh_Hans: 卢尼亚基塔拉语
       - value: ru
         label:
           en_US: Russian
           zh_Hans: 俄语
+      - value: gd
+        label:
+          en_US: Scots Gaelic
+          zh_Hans: 苏格兰盖尔语
       - value: sr
         label:
           en_US: Serbian
           zh_Hans: 塞尔维亚语
+      - value: sh
+        label:
+          en_US: Serbo-Croatian
+          zh_Hans: 塞尔维亚-克罗地亚语
+      - value: st
+        label:
+          en_US: Sesotho
+          zh_Hans: 塞索托语
+      - value: tn
+        label:
+          en_US: Setswana
+          zh_Hans: 塞茨瓦纳语
+      - value: crs
+        label:
+          en_US: Seychellois Creole
+          zh_Hans: 塞舌尔克里奥尔语
+      - value: sn
+        label:
+          en_US: Shona
+          zh_Hans: 绍纳语
+      - value: sd
+        label:
+          en_US: Sindhi
+          zh_Hans: 信德语
+      - value: si
+        label:
+          en_US: Sinhalese
+          zh_Hans: 僧伽罗语
       - value: sk
         label:
           en_US: Slovak
@@ -425,18 +1777,42 @@ parameters:
         label:
           en_US: Slovenian
           zh_Hans: 斯洛文尼亚语
+      - value: so
+        label:
+          en_US: Somali
+          zh_Hans: 索马里语
       - value: es
         label:
           en_US: Spanish
           zh_Hans: 西班牙语
+      - value: es-419
+        label:
+          en_US: Spanish (Latin American)
+          zh_Hans: 西班牙语（拉丁美洲）
+      - value: su
+        label:
+          en_US: Sundanese
+          zh_Hans: 巽他语
+      - value: sw
+        label:
+          en_US: Swahili
+          zh_Hans: 斯瓦希里语
       - value: sv
         label:
           en_US: Swedish
           zh_Hans: 瑞典语
+      - value: tg
+        label:
+          en_US: Tajik
+          zh_Hans: 塔吉克语
       - value: ta
         label:
           en_US: Tamil
           zh_Hans: 泰米尔语
+      - value: tt
+        label:
+          en_US: Tatar
+          zh_Hans: 鞑靼语
       - value: te
         label:
           en_US: Telugu
@@ -445,18 +1821,82 @@ parameters:
         label:
           en_US: Thai
           zh_Hans: 泰语
+      - value: ti
+        label:
+          en_US: Tigrinya
+          zh_Hans: 提格利尼亚语
+      - value: to
+        label:
+          en_US: Tonga
+          zh_Hans: 汤加语
+      - value: lua
+        label:
+          en_US: Tshiluba
+          zh_Hans: 卢巴语
+      - value: tum
+        label:
+          en_US: Tumbuka
+          zh_Hans: 图布卡语
       - value: tr
         label:
           en_US: Turkish
           zh_Hans: 土耳其语
+      - value: tk
+        label:
+          en_US: Turkmen
+          zh_Hans: 土库曼语
+      - value: tw
+        label:
+          en_US: Twi
+          zh_Hans: 契维语
+      - value: ug
+        label:
+          en_US: Uighur
+          zh_Hans: 维吾尔语
       - value: uk
         label:
           en_US: Ukrainian
           zh_Hans: 乌克兰语
+      - value: ur
+        label:
+          en_US: Urdu
+          zh_Hans: 乌尔都语
+      - value: uz
+        label:
+          en_US: Uzbek
+          zh_Hans: 乌兹别克语
+      - value: vu
+        label:
+          en_US: Vanuatu
+          zh_Hans: 瓦努阿图语
       - value: vi
         label:
           en_US: Vietnamese
           zh_Hans: 越南语
+      - value: cy
+        label:
+          en_US: Welsh
+          zh_Hans: 威尔士语
+      - value: wo
+        label:
+          en_US: Wolof
+          zh_Hans: 沃洛夫语
+      - value: xh
+        label:
+          en_US: Xhosa
+          zh_Hans: 科萨语
+      - value: yi
+        label:
+          en_US: Yiddish
+          zh_Hans: 意第绪语
+      - value: yo
+        label:
+          en_US: Yoruba
+          zh_Hans: 约鲁巴语
+      - value: zu
+        label:
+          en_US: Zulu
+          zh_Hans: 祖鲁语
   - name: google_domain
     type: string
     required: false

From 5e250403d30fe8492ffe828ac601befb262e1cf2 Mon Sep 17 00:00:00 2001
From: zhuhao <37029601+hwzhuhao@users.noreply.github.com>
Date: Wed, 9 Oct 2024 11:04:26 +0800
Subject: [PATCH 163/235] chore: add the missing tool names to the _position
 file (#9099)

---
 api/core/tools/provider/_position.yaml | 66 +++++++++++++++++++-------
 1 file changed, 48 insertions(+), 18 deletions(-)

diff --git a/api/core/tools/provider/_position.yaml b/api/core/tools/provider/_position.yaml
index e0a8e7511e59b..6bab9a09d8804 100644
--- a/api/core/tools/provider/_position.yaml
+++ b/api/core/tools/provider/_position.yaml
@@ -5,31 +5,41 @@
 - searchapi
 - serper
 - searxng
+- websearch
+- tavily
+- stackexchange
+- pubmed
+- arxiv
+- aws
+- nominatim
+- devdocs
+- spider
+- firecrawl
+- brave
+- crossref
+- jina
+- webscraper
 - dalle
 - azuredalle
 - stability
-- wikipedia
-- nominatim
-- yahoo
-- alphavantage
-- arxiv
-- pubmed
 - stablediffusion
-- webscraper
-- jina
+- cogview
+- comfyui
+- getimgai
+- siliconflow
+- spark
+- stepfun
+- xinference
+- alphavantage
+- yahoo
+- openweather
+- gaode
 - aippt
-- youtube
-- code
-- wolframalpha
-- maths
-- github
 - chart
-- time
-- vectorizer
-- gaode
-- wecom
-- qrcode
+- youtube
+- did
 - dingtalk
+- discord
 - feishu
 - feishu_base
 - feishu_document
@@ -39,4 +49,24 @@
 - feishu_calendar
 - feishu_spreadsheet
 - slack
+- twilio
+- wecom
+- wikipedia
+- code
+- wolframalpha
+- maths
+- github
+- gitlab
+- time
+- vectorizer
+- qrcode
 - tianditu
+- google_translate
+- hap
+- json_process
+- judge0ce
+- novitaai
+- onebot
+- regex
+- trello
+- vanna

From 4be1aa516c3d48d0b1f8973d2827a732ab552f28 Mon Sep 17 00:00:00 2001
From: zhuhao <37029601+hwzhuhao@users.noreply.github.com>
Date: Wed, 9 Oct 2024 12:53:03 +0800
Subject: [PATCH 164/235] feat: add tags for vanna tools (#9102)

---
 api/core/tools/provider/builtin/vanna/vanna.yaml | 3 +++
 1 file changed, 3 insertions(+)

diff --git a/api/core/tools/provider/builtin/vanna/vanna.yaml b/api/core/tools/provider/builtin/vanna/vanna.yaml
index b29fa103e1d8c..7f953be17224f 100644
--- a/api/core/tools/provider/builtin/vanna/vanna.yaml
+++ b/api/core/tools/provider/builtin/vanna/vanna.yaml
@@ -8,6 +8,9 @@ identity:
     en_US: The fastest way to get actionable insights from your database just by asking questions.
     zh_Hans: 一个基于大模型和RAG的Text2SQL工具。
   icon: icon.png
+  tags:
+    - utilities
+    - productivity
 credentials_for_provider:
   api_key:
     type: secret-input

From 5fcd614186217673c5f3e6de054cecb30330a85c Mon Sep 17 00:00:00 2001
From: zhuhao <37029601+hwzhuhao@users.noreply.github.com>
Date: Wed, 9 Oct 2024 12:53:55 +0800
Subject: [PATCH 165/235] feat: output the execution results of tool should
 only in debug mode (#9104)

---
 .../agent_tool_callback_handler.py            | 41 +++++++++++--------
 1 file changed, 23 insertions(+), 18 deletions(-)

diff --git a/api/core/callback_handler/agent_tool_callback_handler.py b/api/core/callback_handler/agent_tool_callback_handler.py
index 99e992fd89063..d826edf6a0fc1 100644
--- a/api/core/callback_handler/agent_tool_callback_handler.py
+++ b/api/core/callback_handler/agent_tool_callback_handler.py
@@ -1,9 +1,9 @@
-import os
 from collections.abc import Mapping, Sequence
 from typing import Any, Optional, TextIO, Union
 
 from pydantic import BaseModel
 
+from configs import dify_config
 from core.ops.entities.trace_entity import TraceTaskName
 from core.ops.ops_trace_manager import TraceQueueManager, TraceTask
 from core.tools.entities.tool_entities import ToolInvokeMessage
@@ -50,7 +50,8 @@ def on_tool_start(
         tool_inputs: Mapping[str, Any],
     ) -> None:
         """Do nothing."""
-        print_text("\n[on_tool_start] ToolCall:" + tool_name + "\n" + str(tool_inputs) + "\n", color=self.color)
+        if dify_config.DEBUG:
+            print_text("\n[on_tool_start] ToolCall:" + tool_name + "\n" + str(tool_inputs) + "\n", color=self.color)
 
     def on_tool_end(
         self,
@@ -62,11 +63,12 @@ def on_tool_end(
         trace_manager: Optional[TraceQueueManager] = None,
     ) -> None:
         """If not the final action, print out observation."""
-        print_text("\n[on_tool_end]\n", color=self.color)
-        print_text("Tool: " + tool_name + "\n", color=self.color)
-        print_text("Inputs: " + str(tool_inputs) + "\n", color=self.color)
-        print_text("Outputs: " + str(tool_outputs)[:1000] + "\n", color=self.color)
-        print_text("\n")
+        if dify_config.DEBUG:
+            print_text("\n[on_tool_end]\n", color=self.color)
+            print_text("Tool: " + tool_name + "\n", color=self.color)
+            print_text("Inputs: " + str(tool_inputs) + "\n", color=self.color)
+            print_text("Outputs: " + str(tool_outputs)[:1000] + "\n", color=self.color)
+            print_text("\n")
 
         if trace_manager:
             trace_manager.add_trace_task(
@@ -82,30 +84,33 @@ def on_tool_end(
 
     def on_tool_error(self, error: Union[Exception, KeyboardInterrupt], **kwargs: Any) -> None:
         """Do nothing."""
-        print_text("\n[on_tool_error] Error: " + str(error) + "\n", color="red")
+        if dify_config.DEBUG:
+            print_text("\n[on_tool_error] Error: " + str(error) + "\n", color="red")
 
     def on_agent_start(self, thought: str) -> None:
         """Run on agent start."""
-        if thought:
-            print_text(
-                "\n[on_agent_start] \nCurrent Loop: " + str(self.current_loop) + "\nThought: " + thought + "\n",
-                color=self.color,
-            )
-        else:
-            print_text("\n[on_agent_start] \nCurrent Loop: " + str(self.current_loop) + "\n", color=self.color)
+        if dify_config.DEBUG:
+            if thought:
+                print_text(
+                    "\n[on_agent_start] \nCurrent Loop: " + str(self.current_loop) + "\nThought: " + thought + "\n",
+                    color=self.color,
+                )
+            else:
+                print_text("\n[on_agent_start] \nCurrent Loop: " + str(self.current_loop) + "\n", color=self.color)
 
     def on_agent_finish(self, color: Optional[str] = None, **kwargs: Any) -> None:
         """Run on agent end."""
-        print_text("\n[on_agent_finish]\n Loop: " + str(self.current_loop) + "\n", color=self.color)
+        if dify_config.DEBUG:
+            print_text("\n[on_agent_finish]\n Loop: " + str(self.current_loop) + "\n", color=self.color)
 
         self.current_loop += 1
 
     @property
     def ignore_agent(self) -> bool:
         """Whether to ignore agent callbacks."""
-        return not os.environ.get("DEBUG") or os.environ.get("DEBUG").lower() != "true"
+        return not dify_config.DEBUG
 
     @property
     def ignore_chat_model(self) -> bool:
         """Whether to ignore chat model callbacks."""
-        return not os.environ.get("DEBUG") or os.environ.get("DEBUG").lower() != "true"
+        return not dify_config.DEBUG

From b360feb4c19af52faf64c27fd2183bfabfc917ca Mon Sep 17 00:00:00 2001
From: Bowen Liang <liangbowen@gf.com.cn>
Date: Wed, 9 Oct 2024 14:15:27 +0800
Subject: [PATCH 166/235] refactor: introduce storage factory and speed up api
 startup by importing storage client on demand (#9086)

---
 api/extensions/ext_storage.py                 | 80 ++++++++++++-------
 ...liyun_storage.py => aliyun_oss_storage.py} |  4 +-
 .../{s3_storage.py => aws_s3_storage.py}      |  4 +-
 ...azure_storage.py => azure_blob_storage.py} |  4 +-
 ...{baidu_storage.py => baidu_obs_storage.py} |  4 +-
 ...gle_storage.py => google_cloud_storage.py} |  4 +-
 ...uawei_storage.py => huawei_obs_storage.py} |  4 +-
 .../{local_storage.py => local_fs_storage.py} |  4 +-
 .../{oci_storage.py => oracle_oci_storage.py} |  4 +-
 api/extensions/storage/storage_type.py        | 14 ++++
 ...cent_storage.py => tencent_cos_storage.py} |  4 +-
 ...e_storage.py => volcengine_tos_storage.py} |  2 +-
 api/poetry.lock                               |  2 +-
 api/pyproject.toml                            |  2 +-
 14 files changed, 85 insertions(+), 51 deletions(-)
 rename api/extensions/storage/{aliyun_storage.py => aliyun_oss_storage.py} (96%)
 rename api/extensions/storage/{s3_storage.py => aws_s3_storage.py} (97%)
 rename api/extensions/storage/{azure_storage.py => azure_blob_storage.py} (97%)
 rename api/extensions/storage/{baidu_storage.py => baidu_obs_storage.py} (96%)
 rename api/extensions/storage/{google_storage.py => google_cloud_storage.py} (96%)
 rename api/extensions/storage/{huawei_storage.py => huawei_obs_storage.py} (95%)
 rename api/extensions/storage/{local_storage.py => local_fs_storage.py} (96%)
 rename api/extensions/storage/{oci_storage.py => oracle_oci_storage.py} (96%)
 create mode 100644 api/extensions/storage/storage_type.py
 rename api/extensions/storage/{tencent_storage.py => tencent_cos_storage.py} (94%)
 rename api/extensions/storage/{volcengine_storage.py => volcengine_tos_storage.py} (97%)

diff --git a/api/extensions/ext_storage.py b/api/extensions/ext_storage.py
index 863d7a9227e49..be57b633bed4a 100644
--- a/api/extensions/ext_storage.py
+++ b/api/extensions/ext_storage.py
@@ -4,16 +4,9 @@
 
 from flask import Flask
 
-from extensions.storage.aliyun_storage import AliyunStorage
-from extensions.storage.azure_storage import AzureStorage
-from extensions.storage.baidu_storage import BaiduStorage
-from extensions.storage.google_storage import GoogleStorage
-from extensions.storage.huawei_storage import HuaweiStorage
-from extensions.storage.local_storage import LocalStorage
-from extensions.storage.oci_storage import OCIStorage
-from extensions.storage.s3_storage import S3Storage
-from extensions.storage.tencent_storage import TencentStorage
-from extensions.storage.volcengine_storage import VolcengineStorage
+from configs import dify_config
+from extensions.storage.base_storage import BaseStorage
+from extensions.storage.storage_type import StorageType
 
 
 class Storage:
@@ -21,27 +14,52 @@ def __init__(self):
         self.storage_runner = None
 
     def init_app(self, app: Flask):
-        storage_type = app.config.get("STORAGE_TYPE")
-        if storage_type == "s3":
-            self.storage_runner = S3Storage(app=app)
-        elif storage_type == "azure-blob":
-            self.storage_runner = AzureStorage(app=app)
-        elif storage_type == "aliyun-oss":
-            self.storage_runner = AliyunStorage(app=app)
-        elif storage_type == "google-storage":
-            self.storage_runner = GoogleStorage(app=app)
-        elif storage_type == "tencent-cos":
-            self.storage_runner = TencentStorage(app=app)
-        elif storage_type == "oci-storage":
-            self.storage_runner = OCIStorage(app=app)
-        elif storage_type == "huawei-obs":
-            self.storage_runner = HuaweiStorage(app=app)
-        elif storage_type == "baidu-obs":
-            self.storage_runner = BaiduStorage(app=app)
-        elif storage_type == "volcengine-tos":
-            self.storage_runner = VolcengineStorage(app=app)
-        else:
-            self.storage_runner = LocalStorage(app=app)
+        storage_factory = self.get_storage_factory(dify_config.STORAGE_TYPE)
+        self.storage_runner = storage_factory(app=app)
+
+    @staticmethod
+    def get_storage_factory(storage_type: str) -> type[BaseStorage]:
+        match storage_type:
+            case StorageType.S3:
+                from extensions.storage.aws_s3_storage import AwsS3Storage
+
+                return AwsS3Storage
+            case StorageType.AZURE_BLOB:
+                from extensions.storage.azure_blob_storage import AzureBlobStorage
+
+                return AzureBlobStorage
+            case StorageType.ALIYUN_OSS:
+                from extensions.storage.aliyun_oss_storage import AliyunOssStorage
+
+                return AliyunOssStorage
+            case StorageType.GOOGLE_STORAGE:
+                from extensions.storage.google_cloud_storage import GoogleCloudStorage
+
+                return GoogleCloudStorage
+            case StorageType.TENCENT_COS:
+                from extensions.storage.tencent_cos_storage import TencentCosStorage
+
+                return TencentCosStorage
+            case StorageType.OCI_STORAGE:
+                from extensions.storage.oracle_oci_storage import OracleOCIStorage
+
+                return OracleOCIStorage
+            case StorageType.HUAWEI_OBS:
+                from extensions.storage.huawei_obs_storage import HuaweiObsStorage
+
+                return HuaweiObsStorage
+            case StorageType.BAIDU_OBS:
+                from extensions.storage.baidu_obs_storage import BaiduObsStorage
+
+                return BaiduObsStorage
+            case StorageType.VOLCENGINE_TOS:
+                from extensions.storage.volcengine_tos_storage import VolcengineTosStorage
+
+                return VolcengineTosStorage
+            case StorageType.LOCAL | _:
+                from extensions.storage.local_fs_storage import LocalFsStorage
+
+                return LocalFsStorage
 
     def save(self, filename, data):
         try:
diff --git a/api/extensions/storage/aliyun_storage.py b/api/extensions/storage/aliyun_oss_storage.py
similarity index 96%
rename from api/extensions/storage/aliyun_storage.py
rename to api/extensions/storage/aliyun_oss_storage.py
index 2677912aa97bf..53bd399d6d2d5 100644
--- a/api/extensions/storage/aliyun_storage.py
+++ b/api/extensions/storage/aliyun_oss_storage.py
@@ -7,8 +7,8 @@
 from extensions.storage.base_storage import BaseStorage
 
 
-class AliyunStorage(BaseStorage):
-    """Implementation for aliyun storage."""
+class AliyunOssStorage(BaseStorage):
+    """Implementation for Aliyun OSS storage."""
 
     def __init__(self, app: Flask):
         super().__init__(app)
diff --git a/api/extensions/storage/s3_storage.py b/api/extensions/storage/aws_s3_storage.py
similarity index 97%
rename from api/extensions/storage/s3_storage.py
rename to api/extensions/storage/aws_s3_storage.py
index 0858be3af6ad0..fede683aa7bb1 100644
--- a/api/extensions/storage/s3_storage.py
+++ b/api/extensions/storage/aws_s3_storage.py
@@ -9,8 +9,8 @@
 from extensions.storage.base_storage import BaseStorage
 
 
-class S3Storage(BaseStorage):
-    """Implementation for s3 storage."""
+class AwsS3Storage(BaseStorage):
+    """Implementation for Amazon Web Services S3 storage."""
 
     def __init__(self, app: Flask):
         super().__init__(app)
diff --git a/api/extensions/storage/azure_storage.py b/api/extensions/storage/azure_blob_storage.py
similarity index 97%
rename from api/extensions/storage/azure_storage.py
rename to api/extensions/storage/azure_blob_storage.py
index ca8cbb9188b5c..daea660a490ec 100644
--- a/api/extensions/storage/azure_storage.py
+++ b/api/extensions/storage/azure_blob_storage.py
@@ -8,8 +8,8 @@
 from extensions.storage.base_storage import BaseStorage
 
 
-class AzureStorage(BaseStorage):
-    """Implementation for azure storage."""
+class AzureBlobStorage(BaseStorage):
+    """Implementation for Azure Blob storage."""
 
     def __init__(self, app: Flask):
         super().__init__(app)
diff --git a/api/extensions/storage/baidu_storage.py b/api/extensions/storage/baidu_obs_storage.py
similarity index 96%
rename from api/extensions/storage/baidu_storage.py
rename to api/extensions/storage/baidu_obs_storage.py
index 691ac7a58bb5d..c5acff4a9d1c1 100644
--- a/api/extensions/storage/baidu_storage.py
+++ b/api/extensions/storage/baidu_obs_storage.py
@@ -10,8 +10,8 @@
 from extensions.storage.base_storage import BaseStorage
 
 
-class BaiduStorage(BaseStorage):
-    """Implementation for baidu obs storage."""
+class BaiduObsStorage(BaseStorage):
+    """Implementation for Baidu OBS storage."""
 
     def __init__(self, app: Flask):
         super().__init__(app)
diff --git a/api/extensions/storage/google_storage.py b/api/extensions/storage/google_cloud_storage.py
similarity index 96%
rename from api/extensions/storage/google_storage.py
rename to api/extensions/storage/google_cloud_storage.py
index c42f946fa8e71..d9c74b8d40f25 100644
--- a/api/extensions/storage/google_storage.py
+++ b/api/extensions/storage/google_cloud_storage.py
@@ -10,8 +10,8 @@
 from extensions.storage.base_storage import BaseStorage
 
 
-class GoogleStorage(BaseStorage):
-    """Implementation for google storage."""
+class GoogleCloudStorage(BaseStorage):
+    """Implementation for Google Cloud storage."""
 
     def __init__(self, app: Flask):
         super().__init__(app)
diff --git a/api/extensions/storage/huawei_storage.py b/api/extensions/storage/huawei_obs_storage.py
similarity index 95%
rename from api/extensions/storage/huawei_storage.py
rename to api/extensions/storage/huawei_obs_storage.py
index b6ce10bd1f7a9..dd243d400127b 100644
--- a/api/extensions/storage/huawei_storage.py
+++ b/api/extensions/storage/huawei_obs_storage.py
@@ -6,8 +6,8 @@
 from extensions.storage.base_storage import BaseStorage
 
 
-class HuaweiStorage(BaseStorage):
-    """Implementation for huawei obs storage."""
+class HuaweiObsStorage(BaseStorage):
+    """Implementation for Huawei OBS storage."""
 
     def __init__(self, app: Flask):
         super().__init__(app)
diff --git a/api/extensions/storage/local_storage.py b/api/extensions/storage/local_fs_storage.py
similarity index 96%
rename from api/extensions/storage/local_storage.py
rename to api/extensions/storage/local_fs_storage.py
index f833ae85dc54e..9308c4d18019a 100644
--- a/api/extensions/storage/local_storage.py
+++ b/api/extensions/storage/local_fs_storage.py
@@ -8,8 +8,8 @@
 from extensions.storage.base_storage import BaseStorage
 
 
-class LocalStorage(BaseStorage):
-    """Implementation for local storage."""
+class LocalFsStorage(BaseStorage):
+    """Implementation for local filesystem storage."""
 
     def __init__(self, app: Flask):
         super().__init__(app)
diff --git a/api/extensions/storage/oci_storage.py b/api/extensions/storage/oracle_oci_storage.py
similarity index 96%
rename from api/extensions/storage/oci_storage.py
rename to api/extensions/storage/oracle_oci_storage.py
index e32fa0a0ae78a..6934583567584 100644
--- a/api/extensions/storage/oci_storage.py
+++ b/api/extensions/storage/oracle_oci_storage.py
@@ -8,7 +8,9 @@
 from extensions.storage.base_storage import BaseStorage
 
 
-class OCIStorage(BaseStorage):
+class OracleOCIStorage(BaseStorage):
+    """Implementation for Oracle OCI storage."""
+
     def __init__(self, app: Flask):
         super().__init__(app)
         app_config = self.app.config
diff --git a/api/extensions/storage/storage_type.py b/api/extensions/storage/storage_type.py
new file mode 100644
index 0000000000000..e494a520a20b0
--- /dev/null
+++ b/api/extensions/storage/storage_type.py
@@ -0,0 +1,14 @@
+from enum import Enum
+
+
+class StorageType(str, Enum):
+    ALIYUN_OSS = "aliyun-oss"
+    AZURE_BLOB = "azure-blob"
+    BAIDU_OBS = "baidu-obs"
+    GOOGLE_STORAGE = "google-storage"
+    HUAWEI_OBS = "huawei-obs"
+    LOCAL = "local"
+    OCI_STORAGE = "oci-storage"
+    S3 = "s3"
+    TENCENT_COS = "tencent-cos"
+    VOLCENGINE_TOS = "volcengine-tos"
diff --git a/api/extensions/storage/tencent_storage.py b/api/extensions/storage/tencent_cos_storage.py
similarity index 94%
rename from api/extensions/storage/tencent_storage.py
rename to api/extensions/storage/tencent_cos_storage.py
index 1d499cd3bcea1..c529dce7ad8f6 100644
--- a/api/extensions/storage/tencent_storage.py
+++ b/api/extensions/storage/tencent_cos_storage.py
@@ -6,8 +6,8 @@
 from extensions.storage.base_storage import BaseStorage
 
 
-class TencentStorage(BaseStorage):
-    """Implementation for tencent cos storage."""
+class TencentCosStorage(BaseStorage):
+    """Implementation for Tencent Cloud COS storage."""
 
     def __init__(self, app: Flask):
         super().__init__(app)
diff --git a/api/extensions/storage/volcengine_storage.py b/api/extensions/storage/volcengine_tos_storage.py
similarity index 97%
rename from api/extensions/storage/volcengine_storage.py
rename to api/extensions/storage/volcengine_tos_storage.py
index f74ad2ee6dc03..1bedcf24c2ea7 100644
--- a/api/extensions/storage/volcengine_storage.py
+++ b/api/extensions/storage/volcengine_tos_storage.py
@@ -6,7 +6,7 @@
 from extensions.storage.base_storage import BaseStorage
 
 
-class VolcengineStorage(BaseStorage):
+class VolcengineTosStorage(BaseStorage):
     """Implementation for Volcengine TOS storage."""
 
     def __init__(self, app: Flask):
diff --git a/api/poetry.lock b/api/poetry.lock
index cf03f8c0f5b80..6b96350ffdb14 100644
--- a/api/poetry.lock
+++ b/api/poetry.lock
@@ -10595,4 +10595,4 @@ cffi = ["cffi (>=1.11)"]
 [metadata]
 lock-version = "2.0"
 python-versions = ">=3.10,<3.13"
-content-hash = "fd183812f910faf4e840267501c571db5d758ad6eb328d106ba6f79a0322a555"
+content-hash = "34ba8efcc67da342036ef075b693f59fdc67d246f40b857c9c1bd6f80c7283bd"
diff --git a/api/pyproject.toml b/api/pyproject.toml
index 839df9ea6713e..957ae10a647d9 100644
--- a/api/pyproject.toml
+++ b/api/pyproject.toml
@@ -113,7 +113,6 @@ authlib = "1.3.1"
 azure-ai-inference = "~1.0.0b3"
 azure-ai-ml = "~1.20.0"
 azure-identity = "1.16.1"
-azure-storage-blob = "12.13.0"
 beautifulsoup4 = "4.12.2"
 boto3 = "1.35.17"
 bs4 = "~0.0.1"
@@ -221,6 +220,7 @@ yfinance = "~0.2.40"
 # Required for storage clients
 ############################################################
 [tool.poetry.group.storage.dependencies]
+azure-storage-blob = "12.13.0"
 bce-python-sdk = "~0.9.23"
 cos-python-sdk-v5 = "1.9.30"
 esdk-obs-python = "3.24.6.1"

From 240b66d7373f11db378d5b3a67594e5ea1f2fc66 Mon Sep 17 00:00:00 2001
From: Bowen Liang <liangbowen@gf.com.cn>
Date: Wed, 9 Oct 2024 14:36:43 +0800
Subject: [PATCH 167/235] chore: avoid implicit optional in type annotations of
 method (#8727)

---
 api/core/agent/cot_agent_runner.py                 |  2 +-
 api/core/agent/cot_chat_agent_runner.py            |  2 +-
 api/core/agent/cot_completion_agent_runner.py      |  3 ++-
 api/core/agent/fc_agent_runner.py                  |  6 +++---
 api/core/indexing_runner.py                        |  4 ++--
 .../model_providers/anthropic/llm/llm.py           |  2 +-
 .../model_providers/bedrock/llm/llm.py             |  2 +-
 .../model_providers/fireworks/llm/llm.py           |  2 +-
 .../model_providers/openai/llm/llm.py              |  2 +-
 .../openai_api_compatible/llm/llm.py               |  2 +-
 .../model_providers/sagemaker/tts/tts.py           |  2 +-
 .../model_providers/wenxin/llm/llm.py              |  2 +-
 .../api_resource/assistant/assistant.py            | 14 +++++++-------
 .../zhipuai/zhipuai_sdk/api_resource/files.py      | 10 +++++-----
 .../api_resource/knowledge/document/document.py    |  8 ++++----
 .../zhipuai_sdk/api_resource/videos/videos.py      |  8 ++++----
 api/core/rag/datasource/vdb/relyt/relyt_vector.py  |  2 +-
 api/core/rag/datasource/vdb/vector_factory.py      |  6 +++---
 api/core/rag/extractor/extract_processor.py        |  4 ++--
 .../unstructured/unstructured_epub_extractor.py    |  3 ++-
 api/core/tools/entities/tool_entities.py           |  2 +-
 .../provider/builtin/aws/tools/sagemaker_tts.py    |  4 ++--
 api/core/tools/tool/builtin_tool.py                |  4 +++-
 api/core/tools/tool/tool.py                        |  2 +-
 api/core/tools/tool_manager.py                     |  4 ++--
 api/core/tools/utils/feishu_api_utils.py           |  9 ++++++++-
 api/core/tools/utils/parser.py                     | 11 ++++++-----
 api/core/tools/utils/web_reader_tool.py            |  3 ++-
 api/libs/helper.py                                 |  2 +-
 api/pyproject.toml                                 |  1 +
 api/services/account_service.py                    |  2 +-
 api/services/errors/base.py                        |  5 ++++-
 api/services/tag_service.py                        |  3 ++-
 api/services/tools/api_tools_manage_service.py     |  3 ++-
 api/services/tools/tools_transform_service.py      | 10 +++++-----
 .../tools/workflow_tools_manage_service.py         |  5 +++--
 .../integration_tests/vdb/__mock/tcvectordb.py     |  6 +++---
 37 files changed, 91 insertions(+), 71 deletions(-)

diff --git a/api/core/agent/cot_agent_runner.py b/api/core/agent/cot_agent_runner.py
index ebe04bf26000f..d98ba5a3fad84 100644
--- a/api/core/agent/cot_agent_runner.py
+++ b/api/core/agent/cot_agent_runner.py
@@ -369,7 +369,7 @@ def _format_assistant_message(self, agent_scratchpad: list[AgentScratchpadUnit])
         return message
 
     def _organize_historic_prompt_messages(
-        self, current_session_messages: list[PromptMessage] = None
+        self, current_session_messages: Optional[list[PromptMessage]] = None
     ) -> list[PromptMessage]:
         """
         organize historic prompt messages
diff --git a/api/core/agent/cot_chat_agent_runner.py b/api/core/agent/cot_chat_agent_runner.py
index bdec6b7ed15d7..5e16373fff028 100644
--- a/api/core/agent/cot_chat_agent_runner.py
+++ b/api/core/agent/cot_chat_agent_runner.py
@@ -27,7 +27,7 @@ def _organize_system_prompt(self) -> SystemPromptMessage:
 
         return SystemPromptMessage(content=system_prompt)
 
-    def _organize_user_query(self, query, prompt_messages: list[PromptMessage] = None) -> list[PromptMessage]:
+    def _organize_user_query(self, query, prompt_messages: list[PromptMessage]) -> list[PromptMessage]:
         """
         Organize user query
         """
diff --git a/api/core/agent/cot_completion_agent_runner.py b/api/core/agent/cot_completion_agent_runner.py
index 9dab956f9a459..0563090537e62 100644
--- a/api/core/agent/cot_completion_agent_runner.py
+++ b/api/core/agent/cot_completion_agent_runner.py
@@ -1,4 +1,5 @@
 import json
+from typing import Optional
 
 from core.agent.cot_agent_runner import CotAgentRunner
 from core.model_runtime.entities.message_entities import AssistantPromptMessage, PromptMessage, UserPromptMessage
@@ -21,7 +22,7 @@ def _organize_instruction_prompt(self) -> str:
 
         return system_prompt
 
-    def _organize_historic_prompt(self, current_session_messages: list[PromptMessage] = None) -> str:
+    def _organize_historic_prompt(self, current_session_messages: Optional[list[PromptMessage]] = None) -> str:
         """
         Organize historic prompt
         """
diff --git a/api/core/agent/fc_agent_runner.py b/api/core/agent/fc_agent_runner.py
index 13164e0bfca21..7b22025582ae4 100644
--- a/api/core/agent/fc_agent_runner.py
+++ b/api/core/agent/fc_agent_runner.py
@@ -2,7 +2,7 @@
 import logging
 from collections.abc import Generator
 from copy import deepcopy
-from typing import Any, Union
+from typing import Any, Optional, Union
 
 from core.agent.base_agent_runner import BaseAgentRunner
 from core.app.apps.base_app_queue_manager import PublishFrom
@@ -370,7 +370,7 @@ def extract_blocking_tool_calls(self, llm_result: LLMResult) -> Union[None, list
         return tool_calls
 
     def _init_system_message(
-        self, prompt_template: str, prompt_messages: list[PromptMessage] = None
+        self, prompt_template: str, prompt_messages: Optional[list[PromptMessage]] = None
     ) -> list[PromptMessage]:
         """
         Initialize system message
@@ -385,7 +385,7 @@ def _init_system_message(
 
         return prompt_messages
 
-    def _organize_user_query(self, query, prompt_messages: list[PromptMessage] = None) -> list[PromptMessage]:
+    def _organize_user_query(self, query, prompt_messages: list[PromptMessage]) -> list[PromptMessage]:
         """
         Organize user query
         """
diff --git a/api/core/indexing_runner.py b/api/core/indexing_runner.py
index af20df41b1f28..8df26172b76d4 100644
--- a/api/core/indexing_runner.py
+++ b/api/core/indexing_runner.py
@@ -211,9 +211,9 @@ def indexing_estimate(
         tenant_id: str,
         extract_settings: list[ExtractSetting],
         tmp_processing_rule: dict,
-        doc_form: str = None,
+        doc_form: Optional[str] = None,
         doc_language: str = "English",
-        dataset_id: str = None,
+        dataset_id: Optional[str] = None,
         indexing_technique: str = "economy",
     ) -> dict:
         """
diff --git a/api/core/model_runtime/model_providers/anthropic/llm/llm.py b/api/core/model_runtime/model_providers/anthropic/llm/llm.py
index 46e1b415b8125..3a5a42ba05b44 100644
--- a/api/core/model_runtime/model_providers/anthropic/llm/llm.py
+++ b/api/core/model_runtime/model_providers/anthropic/llm/llm.py
@@ -169,7 +169,7 @@ def _code_block_mode_wrapper(
         stop: Optional[list[str]] = None,
         stream: bool = True,
         user: Optional[str] = None,
-        callbacks: list[Callback] = None,
+        callbacks: Optional[list[Callback]] = None,
     ) -> Union[LLMResult, Generator]:
         """
         Code block mode wrapper for invoking large language model
diff --git a/api/core/model_runtime/model_providers/bedrock/llm/llm.py b/api/core/model_runtime/model_providers/bedrock/llm/llm.py
index d1961784f2e84..ff0403ee474d0 100644
--- a/api/core/model_runtime/model_providers/bedrock/llm/llm.py
+++ b/api/core/model_runtime/model_providers/bedrock/llm/llm.py
@@ -92,7 +92,7 @@ def _code_block_mode_wrapper(
         stop: Optional[list[str]] = None,
         stream: bool = True,
         user: Optional[str] = None,
-        callbacks: list[Callback] = None,
+        callbacks: Optional[list[Callback]] = None,
     ) -> Union[LLMResult, Generator]:
         """
         Code block mode wrapper for invoking large language model
diff --git a/api/core/model_runtime/model_providers/fireworks/llm/llm.py b/api/core/model_runtime/model_providers/fireworks/llm/llm.py
index 2dcf1adba6451..24aad9c4d3366 100644
--- a/api/core/model_runtime/model_providers/fireworks/llm/llm.py
+++ b/api/core/model_runtime/model_providers/fireworks/llm/llm.py
@@ -511,7 +511,7 @@ def _num_tokens_from_messages(
         model: str,
         messages: list[PromptMessage],
         tools: Optional[list[PromptMessageTool]] = None,
-        credentials: dict = None,
+        credentials: Optional[dict] = None,
     ) -> int:
         """
         Approximate num tokens with GPT2 tokenizer.
diff --git a/api/core/model_runtime/model_providers/openai/llm/llm.py b/api/core/model_runtime/model_providers/openai/llm/llm.py
index d42fce528a8f3..1ac3837ad3c1b 100644
--- a/api/core/model_runtime/model_providers/openai/llm/llm.py
+++ b/api/core/model_runtime/model_providers/openai/llm/llm.py
@@ -111,7 +111,7 @@ def _code_block_mode_wrapper(
         stop: Optional[list[str]] = None,
         stream: bool = True,
         user: Optional[str] = None,
-        callbacks: list[Callback] = None,
+        callbacks: Optional[list[Callback]] = None,
     ) -> Union[LLMResult, Generator]:
         """
         Code block mode wrapper for invoking large language model
diff --git a/api/core/model_runtime/model_providers/openai_api_compatible/llm/llm.py b/api/core/model_runtime/model_providers/openai_api_compatible/llm/llm.py
index c2ffe653c8b70..356ac56b1e4b2 100644
--- a/api/core/model_runtime/model_providers/openai_api_compatible/llm/llm.py
+++ b/api/core/model_runtime/model_providers/openai_api_compatible/llm/llm.py
@@ -688,7 +688,7 @@ def _num_tokens_from_messages(
         model: str,
         messages: list[PromptMessage],
         tools: Optional[list[PromptMessageTool]] = None,
-        credentials: dict = None,
+        credentials: Optional[dict] = None,
     ) -> int:
         """
         Approximate num tokens with GPT2 tokenizer.
diff --git a/api/core/model_runtime/model_providers/sagemaker/tts/tts.py b/api/core/model_runtime/model_providers/sagemaker/tts/tts.py
index a22bd6dd6ef4f..9652e1d650587 100644
--- a/api/core/model_runtime/model_providers/sagemaker/tts/tts.py
+++ b/api/core/model_runtime/model_providers/sagemaker/tts/tts.py
@@ -77,7 +77,7 @@ def validate_credentials(self, model: str, credentials: dict) -> None:
         """
         pass
 
-    def _detect_lang_code(self, content: str, map_dict: dict = None):
+    def _detect_lang_code(self, content: str, map_dict: Optional[dict] = None):
         map_dict = {"zh": "<|zh|>", "en": "<|en|>", "ja": "<|jp|>", "zh-TW": "<|yue|>", "ko": "<|ko|>"}
 
         response = self.comprehend_client.detect_dominant_language(Text=content)
diff --git a/api/core/model_runtime/model_providers/wenxin/llm/llm.py b/api/core/model_runtime/model_providers/wenxin/llm/llm.py
index f7c160b6b47b7..952cbb33f4b02 100644
--- a/api/core/model_runtime/model_providers/wenxin/llm/llm.py
+++ b/api/core/model_runtime/model_providers/wenxin/llm/llm.py
@@ -64,7 +64,7 @@ def _code_block_mode_wrapper(
         stop: Optional[list[str]] = None,
         stream: bool = True,
         user: Optional[str] = None,
-        callbacks: list[Callback] = None,
+        callbacks: Optional[list[Callback]] = None,
     ) -> Union[LLMResult, Generator]:
         """
         Code block mode wrapper for invoking large language model
diff --git a/api/core/model_runtime/model_providers/zhipuai/zhipuai_sdk/api_resource/assistant/assistant.py b/api/core/model_runtime/model_providers/zhipuai/zhipuai_sdk/api_resource/assistant/assistant.py
index f772340a82c4b..c29b05749847c 100644
--- a/api/core/model_runtime/model_providers/zhipuai/zhipuai_sdk/api_resource/assistant/assistant.py
+++ b/api/core/model_runtime/model_providers/zhipuai/zhipuai_sdk/api_resource/assistant/assistant.py
@@ -41,8 +41,8 @@ def conversation(
         conversation_id: Optional[str] = None,
         attachments: Optional[list[assistant_create_params.AssistantAttachments]] = None,
         metadata: dict | None = None,
-        request_id: str = None,
-        user_id: str = None,
+        request_id: Optional[str] = None,
+        user_id: Optional[str] = None,
         extra_headers: Headers | None = None,
         extra_body: Body | None = None,
         timeout: float | httpx.Timeout | None | NotGiven = NOT_GIVEN,
@@ -72,9 +72,9 @@ def conversation(
     def query_support(
         self,
         *,
-        assistant_id_list: list[str] = None,
-        request_id: str = None,
-        user_id: str = None,
+        assistant_id_list: Optional[list[str]] = None,
+        request_id: Optional[str] = None,
+        user_id: Optional[str] = None,
         extra_headers: Headers | None = None,
         extra_body: Body | None = None,
         timeout: float | httpx.Timeout | None | NotGiven = NOT_GIVEN,
@@ -99,8 +99,8 @@ def query_conversation_usage(
         page: int = 1,
         page_size: int = 10,
         *,
-        request_id: str = None,
-        user_id: str = None,
+        request_id: Optional[str] = None,
+        user_id: Optional[str] = None,
         extra_headers: Headers | None = None,
         extra_body: Body | None = None,
         timeout: float | httpx.Timeout | None | NotGiven = NOT_GIVEN,
diff --git a/api/core/model_runtime/model_providers/zhipuai/zhipuai_sdk/api_resource/files.py b/api/core/model_runtime/model_providers/zhipuai/zhipuai_sdk/api_resource/files.py
index ba9de75b7ef09..c723f6f66e41c 100644
--- a/api/core/model_runtime/model_providers/zhipuai/zhipuai_sdk/api_resource/files.py
+++ b/api/core/model_runtime/model_providers/zhipuai/zhipuai_sdk/api_resource/files.py
@@ -1,7 +1,7 @@
 from __future__ import annotations
 
 from collections.abc import Mapping
-from typing import TYPE_CHECKING, Literal, cast
+from typing import TYPE_CHECKING, Literal, Optional, cast
 
 import httpx
 
@@ -34,11 +34,11 @@ def __init__(self, client: ZhipuAI) -> None:
     def create(
         self,
         *,
-        file: FileTypes = None,
-        upload_detail: list[UploadDetail] = None,
+        file: Optional[FileTypes] = None,
+        upload_detail: Optional[list[UploadDetail]] = None,
         purpose: Literal["fine-tune", "retrieval", "batch"],
-        knowledge_id: str = None,
-        sentence_size: int = None,
+        knowledge_id: Optional[str] = None,
+        sentence_size: Optional[int] = None,
         extra_headers: Headers | None = None,
         extra_body: Body | None = None,
         timeout: float | httpx.Timeout | None | NotGiven = NOT_GIVEN,
diff --git a/api/core/model_runtime/model_providers/zhipuai/zhipuai_sdk/api_resource/knowledge/document/document.py b/api/core/model_runtime/model_providers/zhipuai/zhipuai_sdk/api_resource/knowledge/document/document.py
index 2c4066d893034..492c49da6636c 100644
--- a/api/core/model_runtime/model_providers/zhipuai/zhipuai_sdk/api_resource/knowledge/document/document.py
+++ b/api/core/model_runtime/model_providers/zhipuai/zhipuai_sdk/api_resource/knowledge/document/document.py
@@ -34,12 +34,12 @@ def __init__(self, client: ZhipuAI) -> None:
     def create(
         self,
         *,
-        file: FileTypes = None,
+        file: Optional[FileTypes] = None,
         custom_separator: Optional[list[str]] = None,
-        upload_detail: list[UploadDetail] = None,
+        upload_detail: Optional[list[UploadDetail]] = None,
         purpose: Literal["retrieval"],
-        knowledge_id: str = None,
-        sentence_size: int = None,
+        knowledge_id: Optional[str] = None,
+        sentence_size: Optional[int] = None,
         extra_headers: Headers | None = None,
         extra_body: Body | None = None,
         timeout: float | httpx.Timeout | None | NotGiven = NOT_GIVEN,
diff --git a/api/core/model_runtime/model_providers/zhipuai/zhipuai_sdk/api_resource/videos/videos.py b/api/core/model_runtime/model_providers/zhipuai/zhipuai_sdk/api_resource/videos/videos.py
index f1f1c08036a66..71c8316602a08 100644
--- a/api/core/model_runtime/model_providers/zhipuai/zhipuai_sdk/api_resource/videos/videos.py
+++ b/api/core/model_runtime/model_providers/zhipuai/zhipuai_sdk/api_resource/videos/videos.py
@@ -31,11 +31,11 @@ def generations(
         self,
         model: str,
         *,
-        prompt: str = None,
-        image_url: str = None,
+        prompt: Optional[str] = None,
+        image_url: Optional[str] = None,
         sensitive_word_check: Optional[SensitiveWordCheckRequest] | NotGiven = NOT_GIVEN,
-        request_id: str = None,
-        user_id: str = None,
+        request_id: Optional[str] = None,
+        user_id: Optional[str] = None,
         extra_headers: Headers | None = None,
         extra_body: Body | None = None,
         timeout: float | httpx.Timeout | None | NotGiven = NOT_GIVEN,
diff --git a/api/core/rag/datasource/vdb/relyt/relyt_vector.py b/api/core/rag/datasource/vdb/relyt/relyt_vector.py
index f47f75718a744..254956970f6af 100644
--- a/api/core/rag/datasource/vdb/relyt/relyt_vector.py
+++ b/api/core/rag/datasource/vdb/relyt/relyt_vector.py
@@ -162,7 +162,7 @@ def get_ids_by_metadata_field(self, key: str, value: str):
         else:
             return None
 
-    def delete_by_uuids(self, ids: list[str] = None):
+    def delete_by_uuids(self, ids: Optional[list[str]] = None):
         """Delete by vector IDs.
 
         Args:
diff --git a/api/core/rag/datasource/vdb/vector_factory.py b/api/core/rag/datasource/vdb/vector_factory.py
index ca90233b7f344..943b23870cc5c 100644
--- a/api/core/rag/datasource/vdb/vector_factory.py
+++ b/api/core/rag/datasource/vdb/vector_factory.py
@@ -1,5 +1,5 @@
 from abc import ABC, abstractmethod
-from typing import Any
+from typing import Any, Optional
 
 from configs import dify_config
 from core.embedding.cached_embedding import CacheEmbedding
@@ -25,7 +25,7 @@ def gen_index_struct_dict(vector_type: VectorType, collection_name: str) -> dict
 
 
 class Vector:
-    def __init__(self, dataset: Dataset, attributes: list = None):
+    def __init__(self, dataset: Dataset, attributes: Optional[list] = None):
         if attributes is None:
             attributes = ["doc_id", "dataset_id", "document_id", "doc_hash"]
         self._dataset = dataset
@@ -106,7 +106,7 @@ def get_vector_factory(vector_type: str) -> type[AbstractVectorFactory]:
             case _:
                 raise ValueError(f"Vector store {vector_type} is not supported.")
 
-    def create(self, texts: list = None, **kwargs):
+    def create(self, texts: Optional[list] = None, **kwargs):
         if texts:
             embeddings = self._embeddings.embed_documents([document.page_content for document in texts])
             self._vector_processor.create(texts=texts, embeddings=embeddings, **kwargs)
diff --git a/api/core/rag/extractor/extract_processor.py b/api/core/rag/extractor/extract_processor.py
index 9048138511b12..706a42b7354c5 100644
--- a/api/core/rag/extractor/extract_processor.py
+++ b/api/core/rag/extractor/extract_processor.py
@@ -1,7 +1,7 @@
 import re
 import tempfile
 from pathlib import Path
-from typing import Union
+from typing import Optional, Union
 from urllib.parse import unquote
 
 from configs import dify_config
@@ -84,7 +84,7 @@ def load_from_url(cls, url: str, return_text: bool = False) -> Union[list[Docume
 
     @classmethod
     def extract(
-        cls, extract_setting: ExtractSetting, is_automatic: bool = False, file_path: str = None
+        cls, extract_setting: ExtractSetting, is_automatic: bool = False, file_path: Optional[str] = None
     ) -> list[Document]:
         if extract_setting.datasource_type == DatasourceType.FILE.value:
             with tempfile.TemporaryDirectory() as temp_dir:
diff --git a/api/core/rag/extractor/unstructured/unstructured_epub_extractor.py b/api/core/rag/extractor/unstructured/unstructured_epub_extractor.py
index fa50fa76b2498..a41ed3a5585a5 100644
--- a/api/core/rag/extractor/unstructured/unstructured_epub_extractor.py
+++ b/api/core/rag/extractor/unstructured/unstructured_epub_extractor.py
@@ -1,4 +1,5 @@
 import logging
+from typing import Optional
 
 from core.rag.extractor.extractor_base import BaseExtractor
 from core.rag.models.document import Document
@@ -17,7 +18,7 @@ class UnstructuredEpubExtractor(BaseExtractor):
     def __init__(
         self,
         file_path: str,
-        api_url: str = None,
+        api_url: Optional[str] = None,
     ):
         """Initialize with file path."""
         self._file_path = file_path
diff --git a/api/core/tools/entities/tool_entities.py b/api/core/tools/entities/tool_entities.py
index 02b8b35be7d62..9962b559fa5dd 100644
--- a/api/core/tools/entities/tool_entities.py
+++ b/api/core/tools/entities/tool_entities.py
@@ -341,7 +341,7 @@ def set_text(self, tool_name: str, name: str, value: str) -> None:
 
         self.pool.append(variable)
 
-    def set_file(self, tool_name: str, value: str, name: str = None) -> None:
+    def set_file(self, tool_name: str, value: str, name: Optional[str] = None) -> None:
         """
         set an image variable
 
diff --git a/api/core/tools/provider/builtin/aws/tools/sagemaker_tts.py b/api/core/tools/provider/builtin/aws/tools/sagemaker_tts.py
index bceeaab7453d9..1fafe09b4d96b 100644
--- a/api/core/tools/provider/builtin/aws/tools/sagemaker_tts.py
+++ b/api/core/tools/provider/builtin/aws/tools/sagemaker_tts.py
@@ -1,6 +1,6 @@
 import json
 from enum import Enum
-from typing import Any, Union
+from typing import Any, Optional, Union
 
 import boto3
 
@@ -21,7 +21,7 @@ class SageMakerTTSTool(BuiltinTool):
     s3_client: Any = None
     comprehend_client: Any = None
 
-    def _detect_lang_code(self, content: str, map_dict: dict = None):
+    def _detect_lang_code(self, content: str, map_dict: Optional[dict] = None):
         map_dict = {"zh": "<|zh|>", "en": "<|en|>", "ja": "<|jp|>", "zh-TW": "<|yue|>", "ko": "<|ko|>"}
 
         response = self.comprehend_client.detect_dominant_language(Text=content)
diff --git a/api/core/tools/tool/builtin_tool.py b/api/core/tools/tool/builtin_tool.py
index 8edaf7c0e6a5f..e2a81ed0a36ed 100644
--- a/api/core/tools/tool/builtin_tool.py
+++ b/api/core/tools/tool/builtin_tool.py
@@ -1,3 +1,5 @@
+from typing import Optional
+
 from core.model_runtime.entities.llm_entities import LLMResult
 from core.model_runtime.entities.message_entities import PromptMessage, SystemPromptMessage, UserPromptMessage
 from core.tools.entities.tool_entities import ToolProviderType
@@ -124,7 +126,7 @@ def summarize(content: str) -> str:
 
         return result
 
-    def get_url(self, url: str, user_agent: str = None) -> str:
+    def get_url(self, url: str, user_agent: Optional[str] = None) -> str:
         """
         get url
         """
diff --git a/api/core/tools/tool/tool.py b/api/core/tools/tool/tool.py
index d9e9a0faad88f..cb4ab51ceb199 100644
--- a/api/core/tools/tool/tool.py
+++ b/api/core/tools/tool/tool.py
@@ -318,7 +318,7 @@ def create_text_message(self, text: str, save_as: str = "") -> ToolInvokeMessage
         """
         return ToolInvokeMessage(type=ToolInvokeMessage.MessageType.TEXT, message=text, save_as=save_as)
 
-    def create_blob_message(self, blob: bytes, meta: dict = None, save_as: str = "") -> ToolInvokeMessage:
+    def create_blob_message(self, blob: bytes, meta: Optional[dict] = None, save_as: str = "") -> ToolInvokeMessage:
         """
         create a blob message
 
diff --git a/api/core/tools/tool_manager.py b/api/core/tools/tool_manager.py
index a3303797e16db..ed66dd1357c69 100644
--- a/api/core/tools/tool_manager.py
+++ b/api/core/tools/tool_manager.py
@@ -4,7 +4,7 @@
 from collections.abc import Generator
 from os import listdir, path
 from threading import Lock
-from typing import Any, Union
+from typing import Any, Optional, Union
 
 from configs import dify_config
 from core.agent.entities import AgentToolEntity
@@ -72,7 +72,7 @@ def get_builtin_tool(cls, provider: str, tool_name: str) -> BuiltinTool:
 
     @classmethod
     def get_tool(
-        cls, provider_type: str, provider_id: str, tool_name: str, tenant_id: str = None
+        cls, provider_type: str, provider_id: str, tool_name: str, tenant_id: Optional[str] = None
     ) -> Union[BuiltinTool, ApiTool]:
         """
         get the tool
diff --git a/api/core/tools/utils/feishu_api_utils.py b/api/core/tools/utils/feishu_api_utils.py
index ce1fd7dc19e80..245b296d18206 100644
--- a/api/core/tools/utils/feishu_api_utils.py
+++ b/api/core/tools/utils/feishu_api_utils.py
@@ -1,3 +1,5 @@
+from typing import Optional
+
 import httpx
 
 from core.tools.errors import ToolProviderCredentialValidationError
@@ -32,7 +34,12 @@ def tenant_access_token(self):
         return res.get("tenant_access_token")
 
     def _send_request(
-        self, url: str, method: str = "post", require_token: bool = True, payload: dict = None, params: dict = None
+        self,
+        url: str,
+        method: str = "post",
+        require_token: bool = True,
+        payload: Optional[dict] = None,
+        params: Optional[dict] = None,
     ):
         headers = {
             "Content-Type": "application/json",
diff --git a/api/core/tools/utils/parser.py b/api/core/tools/utils/parser.py
index 9ead4f8e5cf47..0d801d36c4dcd 100644
--- a/api/core/tools/utils/parser.py
+++ b/api/core/tools/utils/parser.py
@@ -3,6 +3,7 @@
 from json import dumps as json_dumps
 from json import loads as json_loads
 from json.decoder import JSONDecodeError
+from typing import Optional
 
 from requests import get
 from yaml import YAMLError, safe_load
@@ -16,7 +17,7 @@
 class ApiBasedToolSchemaParser:
     @staticmethod
     def parse_openapi_to_tool_bundle(
-        openapi: dict, extra_info: dict = None, warning: dict = None
+        openapi: dict, extra_info: Optional[dict], warning: Optional[dict]
     ) -> list[ApiToolBundle]:
         warning = warning if warning is not None else {}
         extra_info = extra_info if extra_info is not None else {}
@@ -174,7 +175,7 @@ def _get_tool_parameter_type(parameter: dict) -> ToolParameter.ToolParameterType
 
     @staticmethod
     def parse_openapi_yaml_to_tool_bundle(
-        yaml: str, extra_info: dict = None, warning: dict = None
+        yaml: str, extra_info: Optional[dict], warning: Optional[dict]
     ) -> list[ApiToolBundle]:
         """
         parse openapi yaml to tool bundle
@@ -191,7 +192,7 @@ def parse_openapi_yaml_to_tool_bundle(
         return ApiBasedToolSchemaParser.parse_openapi_to_tool_bundle(openapi, extra_info=extra_info, warning=warning)
 
     @staticmethod
-    def parse_swagger_to_openapi(swagger: dict, extra_info: dict = None, warning: dict = None) -> dict:
+    def parse_swagger_to_openapi(swagger: dict, extra_info: Optional[dict], warning: Optional[dict]) -> dict:
         """
         parse swagger to openapi
 
@@ -253,7 +254,7 @@ def parse_swagger_to_openapi(swagger: dict, extra_info: dict = None, warning: di
 
     @staticmethod
     def parse_openai_plugin_json_to_tool_bundle(
-        json: str, extra_info: dict = None, warning: dict = None
+        json: str, extra_info: Optional[dict], warning: Optional[dict]
     ) -> list[ApiToolBundle]:
         """
         parse openapi plugin yaml to tool bundle
@@ -287,7 +288,7 @@ def parse_openai_plugin_json_to_tool_bundle(
 
     @staticmethod
     def auto_parse_to_tool_bundle(
-        content: str, extra_info: dict = None, warning: dict = None
+        content: str, extra_info: Optional[dict], warning: Optional[dict]
     ) -> tuple[list[ApiToolBundle], str]:
         """
         auto parse to tool bundle
diff --git a/api/core/tools/utils/web_reader_tool.py b/api/core/tools/utils/web_reader_tool.py
index 1ced7d0488e3f..5807d61b9409a 100644
--- a/api/core/tools/utils/web_reader_tool.py
+++ b/api/core/tools/utils/web_reader_tool.py
@@ -9,6 +9,7 @@
 import unicodedata
 from contextlib import contextmanager
 from pathlib import Path
+from typing import Optional
 from urllib.parse import unquote
 
 import chardet
@@ -36,7 +37,7 @@ def page_result(text: str, cursor: int, max_length: int) -> str:
     return text[cursor : cursor + max_length]
 
 
-def get_url(url: str, user_agent: str = None) -> str:
+def get_url(url: str, user_agent: Optional[str] = None) -> str:
     """Fetch URL and return the contents as a string."""
     headers = {
         "User-Agent": "Mozilla/5.0 (Windows NT 10.0; Win64; x64) AppleWebKit/537.36 (KHTML, like Gecko)"
diff --git a/api/libs/helper.py b/api/libs/helper.py
index d664ef1ae7d66..9c3a1ff04d1f8 100644
--- a/api/libs/helper.py
+++ b/api/libs/helper.py
@@ -189,7 +189,7 @@ def generate() -> Generator:
 
 class TokenManager:
     @classmethod
-    def generate_token(cls, account: Account, token_type: str, additional_data: dict = None) -> str:
+    def generate_token(cls, account: Account, token_type: str, additional_data: Optional[dict] = None) -> str:
         old_token = cls._get_current_token_for_account(account.id, token_type)
         if old_token:
             if isinstance(old_token, bytes):
diff --git a/api/pyproject.toml b/api/pyproject.toml
index 957ae10a647d9..48418974d47de 100644
--- a/api/pyproject.toml
+++ b/api/pyproject.toml
@@ -28,6 +28,7 @@ select = [
     "PLR0402", # manual-from-import
     "PLR1711", # useless-return
     "PLR1714", # repeated-equality-comparison
+    "RUF013", # implicit-optional
     "RUF019", # unnecessary-key-check
     "RUF100", # unused-noqa
     "RUF101", # redirected-noqa
diff --git a/api/services/account_service.py b/api/services/account_service.py
index 66ff5d2b7ce7e..c3d3354a1d83f 100644
--- a/api/services/account_service.py
+++ b/api/services/account_service.py
@@ -321,7 +321,7 @@ def get_current_tenant_by_account(account: Account):
         return tenant
 
     @staticmethod
-    def switch_tenant(account: Account, tenant_id: int = None) -> None:
+    def switch_tenant(account: Account, tenant_id: Optional[int] = None) -> None:
         """Switch the current workspace for the account"""
 
         # Ensure tenant_id is provided
diff --git a/api/services/errors/base.py b/api/services/errors/base.py
index 1fed71cf9e380..4d39f956b8c93 100644
--- a/api/services/errors/base.py
+++ b/api/services/errors/base.py
@@ -1,3 +1,6 @@
+from typing import Optional
+
+
 class BaseServiceError(Exception):
-    def __init__(self, description: str = None):
+    def __init__(self, description: Optional[str] = None):
         self.description = description
diff --git a/api/services/tag_service.py b/api/services/tag_service.py
index 5e2851cd8febc..a374bdcf002be 100644
--- a/api/services/tag_service.py
+++ b/api/services/tag_service.py
@@ -1,4 +1,5 @@
 import uuid
+from typing import Optional
 
 from flask_login import current_user
 from sqlalchemy import func
@@ -11,7 +12,7 @@
 
 class TagService:
     @staticmethod
-    def get_tags(tag_type: str, current_tenant_id: str, keyword: str = None) -> list:
+    def get_tags(tag_type: str, current_tenant_id: str, keyword: Optional[str] = None) -> list:
         query = (
             db.session.query(Tag.id, Tag.type, Tag.name, func.count(TagBinding.id).label("binding_count"))
             .outerjoin(TagBinding, Tag.id == TagBinding.tag_id)
diff --git a/api/services/tools/api_tools_manage_service.py b/api/services/tools/api_tools_manage_service.py
index 6f6074f59640d..9f5298a506752 100644
--- a/api/services/tools/api_tools_manage_service.py
+++ b/api/services/tools/api_tools_manage_service.py
@@ -1,5 +1,6 @@
 import json
 import logging
+from typing import Optional
 
 from httpx import get
 
@@ -79,7 +80,7 @@ def parser_api_schema(schema: str) -> list[ApiToolBundle]:
             raise ValueError(f"invalid schema: {str(e)}")
 
     @staticmethod
-    def convert_schema_to_tool_bundles(schema: str, extra_info: dict = None) -> list[ApiToolBundle]:
+    def convert_schema_to_tool_bundles(schema: str, extra_info: Optional[dict] = None) -> list[ApiToolBundle]:
         """
         convert schema to tool bundles
 
diff --git a/api/services/tools/tools_transform_service.py b/api/services/tools/tools_transform_service.py
index 2bc48c41850de..4af73d5063054 100644
--- a/api/services/tools/tools_transform_service.py
+++ b/api/services/tools/tools_transform_service.py
@@ -144,7 +144,7 @@ def workflow_provider_to_controller(db_provider: WorkflowToolProvider) -> Workfl
 
     @staticmethod
     def workflow_provider_to_user_provider(
-        provider_controller: WorkflowToolProviderController, labels: list[str] = None
+        provider_controller: WorkflowToolProviderController, labels: Optional[list[str]] = None
     ):
         """
         convert provider controller to user provider
@@ -174,7 +174,7 @@ def api_provider_to_user_provider(
         provider_controller: ApiToolProviderController,
         db_provider: ApiToolProvider,
         decrypt_credentials: bool = True,
-        labels: list[str] = None,
+        labels: Optional[list[str]] = None,
     ) -> UserToolProvider:
         """
         convert provider controller to user provider
@@ -223,9 +223,9 @@ def api_provider_to_user_provider(
     @staticmethod
     def tool_to_user_tool(
         tool: Union[ApiToolBundle, WorkflowTool, Tool],
-        credentials: dict = None,
-        tenant_id: str = None,
-        labels: list[str] = None,
+        credentials: Optional[dict] = None,
+        tenant_id: Optional[str] = None,
+        labels: Optional[list[str]] = None,
     ) -> UserTool:
         """
         convert tool to user tool
diff --git a/api/services/tools/workflow_tools_manage_service.py b/api/services/tools/workflow_tools_manage_service.py
index 3830e75339eeb..5868ef37559ba 100644
--- a/api/services/tools/workflow_tools_manage_service.py
+++ b/api/services/tools/workflow_tools_manage_service.py
@@ -1,5 +1,6 @@
 import json
 from datetime import datetime
+from typing import Optional
 
 from sqlalchemy import or_
 
@@ -32,7 +33,7 @@ def create_workflow_tool(
         description: str,
         parameters: list[dict],
         privacy_policy: str = "",
-        labels: list[str] = None,
+        labels: Optional[list[str]] = None,
     ) -> dict:
         """
         Create a workflow tool.
@@ -106,7 +107,7 @@ def update_workflow_tool(
         description: str,
         parameters: list[dict],
         privacy_policy: str = "",
-        labels: list[str] = None,
+        labels: Optional[list[str]] = None,
     ) -> dict:
         """
         Update a workflow tool.
diff --git a/api/tests/integration_tests/vdb/__mock/tcvectordb.py b/api/tests/integration_tests/vdb/__mock/tcvectordb.py
index 53c9b3cae319f..61d6ed16560c0 100644
--- a/api/tests/integration_tests/vdb/__mock/tcvectordb.py
+++ b/api/tests/integration_tests/vdb/__mock/tcvectordb.py
@@ -48,7 +48,7 @@ def create_collection(
         description: str,
         index: Index,
         embedding: Embedding = None,
-        timeout: float = None,
+        timeout: Optional[float] = None,
     ) -> Collection:
         return Collection(
             self,
@@ -97,9 +97,9 @@ def collection_query(
 
     def collection_delete(
         self,
-        document_ids: list[str] = None,
+        document_ids: Optional[list[str]] = None,
         filter: Filter = None,
-        timeout: float = None,
+        timeout: Optional[float] = None,
     ):
         return {"code": 0, "msg": "operation success"}
 

From c0b71f8286e90f575b1c1eb8d0cbd081595cab26 Mon Sep 17 00:00:00 2001
From: kurokobo <kuro664@gmail.com>
Date: Wed, 9 Oct 2024 15:42:30 +0900
Subject: [PATCH 168/235] feat: respect x-* headers for redirections (#9054)

---
 api/.env.example                |  3 +++
 api/app.py                      |  2 ++
 api/configs/feature/__init__.py |  6 ++++++
 api/extensions/ext_proxy_fix.py | 10 ++++++++++
 4 files changed, 21 insertions(+)
 create mode 100644 api/extensions/ext_proxy_fix.py

diff --git a/api/.env.example b/api/.env.example
index fb0fc045b0850..71f0e5db8f8b9 100644
--- a/api/.env.example
+++ b/api/.env.example
@@ -271,6 +271,9 @@ HTTP_REQUEST_MAX_WRITE_TIMEOUT=600
 HTTP_REQUEST_NODE_MAX_BINARY_SIZE=10485760
 HTTP_REQUEST_NODE_MAX_TEXT_SIZE=1048576
 
+# Respect X-* headers to redirect clients
+RESPECT_XFORWARD_HEADERS_ENABLED=false
+
 # Log file path
 LOG_FILE=
 
diff --git a/api/app.py b/api/app.py
index 0662d94e1b4c8..a251ef5f0f72c 100644
--- a/api/app.py
+++ b/api/app.py
@@ -36,6 +36,7 @@
     ext_login,
     ext_mail,
     ext_migrate,
+    ext_proxy_fix,
     ext_redis,
     ext_sentry,
     ext_storage,
@@ -156,6 +157,7 @@ def initialize_extensions(app):
     ext_mail.init_app(app)
     ext_hosting_provider.init_app(app)
     ext_sentry.init_app(app)
+    ext_proxy_fix.init_app(app)
 
 
 # Flask-Login configuration
diff --git a/api/configs/feature/__init__.py b/api/configs/feature/__init__.py
index 9218d529cc09e..93dbc1367f394 100644
--- a/api/configs/feature/__init__.py
+++ b/api/configs/feature/__init__.py
@@ -247,6 +247,12 @@ def WEB_API_CORS_ALLOW_ORIGINS(self) -> list[str]:
         default=None,
     )
 
+    RESPECT_XFORWARD_HEADERS_ENABLED: bool = Field(
+        description="Enable or disable the X-Forwarded-For Proxy Fix middleware from Werkzeug"
+        " to respect X-* headers to redirect clients",
+        default=False,
+    )
+
 
 class InnerAPIConfig(BaseSettings):
     """
diff --git a/api/extensions/ext_proxy_fix.py b/api/extensions/ext_proxy_fix.py
new file mode 100644
index 0000000000000..c106a4384a156
--- /dev/null
+++ b/api/extensions/ext_proxy_fix.py
@@ -0,0 +1,10 @@
+from flask import Flask
+
+from configs import dify_config
+
+
+def init_app(app: Flask):
+    if dify_config.RESPECT_XFORWARD_HEADERS_ENABLED:
+        from werkzeug.middleware.proxy_fix import ProxyFix
+
+        app.wsgi_app = ProxyFix(app.wsgi_app)

From 55679b43898d2eb3f71a2fd5fea08c3c0a3e5926 Mon Sep 17 00:00:00 2001
From: "Charlie.Wei" <luowei@cvte.com>
Date: Wed, 9 Oct 2024 16:15:03 +0800
Subject: [PATCH 169/235] =?UTF-8?q?azure=20add=20o1-mini=E3=80=81o1-previe?=
 =?UTF-8?q?w=20models=20(#9088)?=
MIME-Version: 1.0
Content-Type: text/plain; charset=UTF-8
Content-Transfer-Encoding: 8bit

Co-authored-by: crazywoola <100913391+crazywoola@users.noreply.github.com>
---
 .../model_providers/azure_openai/_constant.py |  75 ++++++++++-
 .../azure_openai/azure_openai.yaml            |  12 ++
 .../model_providers/azure_openai/llm/llm.py   | 122 ++++++++++++++++--
 3 files changed, 196 insertions(+), 13 deletions(-)

diff --git a/api/core/model_runtime/model_providers/azure_openai/_constant.py b/api/core/model_runtime/model_providers/azure_openai/_constant.py
index 0dada70cc5b7f..baa542139629d 100644
--- a/api/core/model_runtime/model_providers/azure_openai/_constant.py
+++ b/api/core/model_runtime/model_providers/azure_openai/_constant.py
@@ -1081,8 +1081,81 @@ class AzureBaseModel(BaseModel):
             ),
         ),
     ),
+    AzureBaseModel(
+        base_model_name="o1-preview",
+        entity=AIModelEntity(
+            model="fake-deployment-name",
+            label=I18nObject(
+                en_US="fake-deployment-name-label",
+            ),
+            model_type=ModelType.LLM,
+            features=[
+                ModelFeature.AGENT_THOUGHT,
+            ],
+            fetch_from=FetchFrom.CUSTOMIZABLE_MODEL,
+            model_properties={
+                ModelPropertyKey.MODE: LLMMode.CHAT.value,
+                ModelPropertyKey.CONTEXT_SIZE: 128000,
+            },
+            parameter_rules=[
+                ParameterRule(
+                    name="response_format",
+                    label=I18nObject(zh_Hans="回复格式", en_US="response_format"),
+                    type="string",
+                    help=I18nObject(
+                        zh_Hans="指定模型必须输出的格式", en_US="specifying the format that the model must output"
+                    ),
+                    required=False,
+                    options=["text", "json_object"],
+                ),
+                _get_max_tokens(default=512, min_val=1, max_val=32768),
+            ],
+            pricing=PriceConfig(
+                input=15.00,
+                output=60.00,
+                unit=0.000001,
+                currency="USD",
+            ),
+        ),
+    ),
+    AzureBaseModel(
+        base_model_name="o1-mini",
+        entity=AIModelEntity(
+            model="fake-deployment-name",
+            label=I18nObject(
+                en_US="fake-deployment-name-label",
+            ),
+            model_type=ModelType.LLM,
+            features=[
+                ModelFeature.AGENT_THOUGHT,
+            ],
+            fetch_from=FetchFrom.CUSTOMIZABLE_MODEL,
+            model_properties={
+                ModelPropertyKey.MODE: LLMMode.CHAT.value,
+                ModelPropertyKey.CONTEXT_SIZE: 128000,
+            },
+            parameter_rules=[
+                ParameterRule(
+                    name="response_format",
+                    label=I18nObject(zh_Hans="回复格式", en_US="response_format"),
+                    type="string",
+                    help=I18nObject(
+                        zh_Hans="指定模型必须输出的格式", en_US="specifying the format that the model must output"
+                    ),
+                    required=False,
+                    options=["text", "json_object"],
+                ),
+                _get_max_tokens(default=512, min_val=1, max_val=65536),
+            ],
+            pricing=PriceConfig(
+                input=3.00,
+                output=12.00,
+                unit=0.000001,
+                currency="USD",
+            ),
+        ),
+    ),
 ]
-
 EMBEDDING_BASE_MODELS = [
     AzureBaseModel(
         base_model_name="text-embedding-ada-002",
diff --git a/api/core/model_runtime/model_providers/azure_openai/azure_openai.yaml b/api/core/model_runtime/model_providers/azure_openai/azure_openai.yaml
index 867f9fec42000..cb87bfdf6b9b7 100644
--- a/api/core/model_runtime/model_providers/azure_openai/azure_openai.yaml
+++ b/api/core/model_runtime/model_providers/azure_openai/azure_openai.yaml
@@ -120,6 +120,18 @@ model_credential_schema:
           show_on:
             - variable: __model_type
               value: llm
+        - label:
+            en_US: o1-mini
+          value: o1-mini
+          show_on:
+            - variable: __model_type
+              value: llm
+        - label:
+            en_US: o1-preview
+          value: o1-preview
+          show_on:
+            - variable: __model_type
+              value: llm
         - label:
             en_US: gpt-4o-mini
           value: gpt-4o-mini
diff --git a/api/core/model_runtime/model_providers/azure_openai/llm/llm.py b/api/core/model_runtime/model_providers/azure_openai/llm/llm.py
index f0033ea051b50..f63c80e58e215 100644
--- a/api/core/model_runtime/model_providers/azure_openai/llm/llm.py
+++ b/api/core/model_runtime/model_providers/azure_openai/llm/llm.py
@@ -312,20 +312,118 @@ def _chat_generate(
         if user:
             extra_model_kwargs["user"] = user
 
-        # chat model
-        messages = [self._convert_prompt_message_to_dict(m) for m in prompt_messages]
-        response = client.chat.completions.create(
-            messages=messages,
-            model=model,
-            stream=stream,
-            **model_parameters,
-            **extra_model_kwargs,
+            # clear illegal prompt messages
+            prompt_messages = self._clear_illegal_prompt_messages(model, prompt_messages)
+
+            block_as_stream = False
+            if model.startswith("o1"):
+                if stream:
+                    block_as_stream = True
+                    stream = False
+
+                    if "stream_options" in extra_model_kwargs:
+                        del extra_model_kwargs["stream_options"]
+
+                if "stop" in extra_model_kwargs:
+                    del extra_model_kwargs["stop"]
+
+            # chat model
+            response = client.chat.completions.create(
+                messages=[self._convert_prompt_message_to_dict(m) for m in prompt_messages],
+                model=model,
+                stream=stream,
+                **model_parameters,
+                **extra_model_kwargs,
+            )
+
+            if stream:
+                return self._handle_chat_generate_stream_response(model, credentials, response, prompt_messages, tools)
+
+            block_result = self._handle_chat_generate_response(model, credentials, response, prompt_messages, tools)
+
+            if block_as_stream:
+                return self._handle_chat_block_as_stream_response(block_result, prompt_messages, stop)
+
+            return block_result
+
+    def _handle_chat_block_as_stream_response(
+        self,
+        block_result: LLMResult,
+        prompt_messages: list[PromptMessage],
+        stop: Optional[list[str]] = None,
+    ) -> Generator[LLMResultChunk, None, None]:
+        """
+        Handle llm chat response
+
+        :param model: model name
+        :param credentials: credentials
+        :param response: response
+        :param prompt_messages: prompt messages
+        :param tools: tools for tool calling
+        :param stop: stop words
+        :return: llm response chunk generator
+        """
+        text = block_result.message.content
+        text = cast(str, text)
+
+        if stop:
+            text = self.enforce_stop_tokens(text, stop)
+
+        yield LLMResultChunk(
+            model=block_result.model,
+            prompt_messages=prompt_messages,
+            system_fingerprint=block_result.system_fingerprint,
+            delta=LLMResultChunkDelta(
+                index=0,
+                message=AssistantPromptMessage(content=text),
+                finish_reason="stop",
+                usage=block_result.usage,
+            ),
         )
 
-        if stream:
-            return self._handle_chat_generate_stream_response(model, credentials, response, prompt_messages, tools)
+    def _clear_illegal_prompt_messages(self, model: str, prompt_messages: list[PromptMessage]) -> list[PromptMessage]:
+        """
+        Clear illegal prompt messages for OpenAI API
+
+        :param model: model name
+        :param prompt_messages: prompt messages
+        :return: cleaned prompt messages
+        """
+        checklist = ["gpt-4-turbo", "gpt-4-turbo-2024-04-09"]
+
+        if model in checklist:
+            # count how many user messages are there
+            user_message_count = len([m for m in prompt_messages if isinstance(m, UserPromptMessage)])
+            if user_message_count > 1:
+                for prompt_message in prompt_messages:
+                    if isinstance(prompt_message, UserPromptMessage):
+                        if isinstance(prompt_message.content, list):
+                            prompt_message.content = "\n".join(
+                                [
+                                    item.data
+                                    if item.type == PromptMessageContentType.TEXT
+                                    else "[IMAGE]"
+                                    if item.type == PromptMessageContentType.IMAGE
+                                    else ""
+                                    for item in prompt_message.content
+                                ]
+                            )
+
+        if model.startswith("o1"):
+            system_message_count = len([m for m in prompt_messages if isinstance(m, SystemPromptMessage)])
+            if system_message_count > 0:
+                new_prompt_messages = []
+                for prompt_message in prompt_messages:
+                    if isinstance(prompt_message, SystemPromptMessage):
+                        prompt_message = UserPromptMessage(
+                            content=prompt_message.content,
+                            name=prompt_message.name,
+                        )
+
+                    new_prompt_messages.append(prompt_message)
+                prompt_messages = new_prompt_messages
 
-        return self._handle_chat_generate_response(model, credentials, response, prompt_messages, tools)
+        return prompt_messages
 
     def _handle_chat_generate_response(
         self,
@@ -560,7 +658,7 @@ def _num_tokens_from_messages(
             tokens_per_message = 4
             # if there's a name, the role is omitted
             tokens_per_name = -1
-        elif model.startswith("gpt-35-turbo") or model.startswith("gpt-4"):
+        elif model.startswith("gpt-35-turbo") or model.startswith("gpt-4") or model.startswith("o1"):
             tokens_per_message = 3
             tokens_per_name = 1
         else:

From ec0aa32cd4e39568c95d056aeafea9218975f128 Mon Sep 17 00:00:00 2001
From: Alter-xyz <88554920+alterxyz@users.noreply.github.com>
Date: Wed, 9 Oct 2024 04:56:09 -0400
Subject: [PATCH 170/235] Add customizable size for chatbot embed with default
 values (#9115)

---
 web/app/components/app/overview/embedded/index.tsx | 4 ++++
 1 file changed, 4 insertions(+)

diff --git a/web/app/components/app/overview/embedded/index.tsx b/web/app/components/app/overview/embedded/index.tsx
index fa023192c9a55..b71a3c3fdf9d2 100644
--- a/web/app/components/app/overview/embedded/index.tsx
+++ b/web/app/components/app/overview/embedded/index.tsx
@@ -52,6 +52,10 @@ const OPTION_MAP = {
   #dify-chatbot-bubble-button {
     background-color: ${primaryColor} !important;
   }
+  #dify-chatbot-bubble-window {
+    width: 24rem !important;
+    height: 40rem !important;
+  }
 </style>`,
   },
   chromePlugin: {

From cbd3ffe0562bc8f2ce819a0e253ac7c9d1c69b54 Mon Sep 17 00:00:00 2001
From: Yi Xiao <54782454+YIXIAO0@users.noreply.github.com>
Date: Wed, 9 Oct 2024 18:26:38 +0800
Subject: [PATCH 171/235] Fix the margin for the rerank switch in retrieval
 setting (#9124)

---
 .../dataset-config/params-config/config-content.tsx             | 2 +-
 1 file changed, 1 insertion(+), 1 deletion(-)

diff --git a/web/app/components/app/configuration/dataset-config/params-config/config-content.tsx b/web/app/components/app/configuration/dataset-config/params-config/config-content.tsx
index f99496b14fa8f..7f83a14d58acc 100644
--- a/web/app/components/app/configuration/dataset-config/params-config/config-content.tsx
+++ b/web/app/components/app/configuration/dataset-config/params-config/config-content.tsx
@@ -243,7 +243,7 @@ const ConfigContent: FC<Props> = ({
                       />
                     )
                   }
-                  <div className='leading-[32px] text-text-secondary system-sm-semibold'>{t('common.modelProvider.rerankModel.key')}</div>
+                  <div className='leading-[32px] ml-1 text-text-secondary system-sm-semibold'>{t('common.modelProvider.rerankModel.key')}</div>
                   <Tooltip
                     popupContent={
                       <div className="w-[200px]">

From 7a1da2409d222257912adf16277dc34112819e46 Mon Sep 17 00:00:00 2001
From: pinsily <13160724868@163.com>
Date: Wed, 9 Oct 2024 21:09:36 +0800
Subject: [PATCH 172/235] add generate conversation name error log (#9117)

---
 api/core/app/task_pipeline/message_cycle_manage.py | 4 ++--
 1 file changed, 2 insertions(+), 2 deletions(-)

diff --git a/api/core/app/task_pipeline/message_cycle_manage.py b/api/core/app/task_pipeline/message_cycle_manage.py
index 5872e00740b61..3a1d1b227f62f 100644
--- a/api/core/app/task_pipeline/message_cycle_manage.py
+++ b/api/core/app/task_pipeline/message_cycle_manage.py
@@ -82,8 +82,8 @@ def _generate_conversation_name_worker(self, flask_app: Flask, conversation_id:
                 try:
                     name = LLMGenerator.generate_conversation_name(app_model.tenant_id, query)
                     conversation.name = name
-                except:
-                    pass
+                except Exception as e:
+                    logging.exception(f"generate conversation name failed: {e}")
 
                 db.session.merge(conversation)
                 db.session.commit()

From 511ffa46989399c3211d4564df5be1b2dd69d1d7 Mon Sep 17 00:00:00 2001
From: Hash Brown <hi@xzd.me>
Date: Wed, 9 Oct 2024 21:10:55 +0800
Subject: [PATCH 173/235] =?UTF-8?q?fix:=20follow-up=20(suggested=20questio?=
 =?UTF-8?q?ns)=20does=20not=20refer=20to=20the=20most=20recen=E2=80=A6=20(?=
 =?UTF-8?q?#9122)?=
MIME-Version: 1.0
Content-Type: text/plain; charset=UTF-8
Content-Transfer-Encoding: 8bit

---
 api/core/memory/token_buffer_memory.py | 6 +++++-
 1 file changed, 5 insertions(+), 1 deletion(-)

diff --git a/api/core/memory/token_buffer_memory.py b/api/core/memory/token_buffer_memory.py
index 60b36c50f00d3..72da3b0c6f251 100644
--- a/api/core/memory/token_buffer_memory.py
+++ b/api/core/memory/token_buffer_memory.py
@@ -58,7 +58,11 @@ def get_history_prompt_messages(
         # instead of all messages from the conversation, we only need to extract messages
         # that belong to the thread of last message
         thread_messages = extract_thread_messages(messages)
-        thread_messages.pop(0)
+
+        # for newly created message, its answer is temporarily empty, we don't need to add it to memory
+        if thread_messages and not thread_messages[-1].answer:
+            thread_messages.pop()
+
         messages = list(reversed(thread_messages))
 
         message_file_parser = MessageFileParser(tenant_id=app_record.tenant_id, app_id=app_record.id)

From cb35e84f512eb9ce1c23417d718130ce2223c4ef Mon Sep 17 00:00:00 2001
From: =?UTF-8?q?=E9=9D=9E=E6=B3=95=E6=93=8D=E4=BD=9C?= <hjlarry@163.com>
Date: Wed, 9 Oct 2024 21:11:13 +0800
Subject: [PATCH 174/235] fix: docker env typo (#9119)

---
 docker/docker-compose.yaml | 2 +-
 1 file changed, 1 insertion(+), 1 deletion(-)

diff --git a/docker/docker-compose.yaml b/docker/docker-compose.yaml
index ddb6ded206012..c1a76d7f88216 100644
--- a/docker/docker-compose.yaml
+++ b/docker/docker-compose.yaml
@@ -204,7 +204,7 @@ x-shared-env: &shared-api-worker-env
   CODE_MAX_NUMBER_ARRAY_LENGTH: ${CODE_MAX_NUMBER_ARRAY_LENGTH:-1000}
   WORKFLOW_MAX_EXECUTION_STEPS: ${WORKFLOW_MAX_EXECUTION_STEPS:-500}
   WORKFLOW_MAX_EXECUTION_TIME: ${WORKFLOW_MAX_EXECUTION_TIME:-1200}
-  WORKFLOW_CALL_MAX_DEPTH: ${WORKFLOW_MAX_EXECUTION_TIME:-5}
+  WORKFLOW_CALL_MAX_DEPTH: ${WORKFLOW_CALL_MAX_DEPTH:-5}
   SSRF_PROXY_HTTP_URL: ${SSRF_PROXY_HTTP_URL:-http://ssrf_proxy:3128}
   SSRF_PROXY_HTTPS_URL: ${SSRF_PROXY_HTTPS_URL:-http://ssrf_proxy:3128}
   HTTP_REQUEST_NODE_MAX_BINARY_SIZE: ${HTTP_REQUEST_NODE_MAX_BINARY_SIZE:-10485760}

From b279d1904085d0265c4b0a0ef5843bc11b22c837 Mon Sep 17 00:00:00 2001
From: Bowen Liang <liangbowen@gf.com.cn>
Date: Wed, 9 Oct 2024 21:12:50 +0800
Subject: [PATCH 175/235] chore: avoid star imports usage (#9123)

---
 api/pyproject.toml              |  2 --
 api/services/account_service.py | 13 ++++++++++++-
 2 files changed, 12 insertions(+), 3 deletions(-)

diff --git a/api/pyproject.toml b/api/pyproject.toml
index 48418974d47de..956d4dbd11960 100644
--- a/api/pyproject.toml
+++ b/api/pyproject.toml
@@ -45,8 +45,6 @@ ignore = [
     "E721", # type-comparison
     "E722", # bare-except
     "E731", # lambda-assignment
-    "F403", # undefined-local-with-import-star
-    "F405", # undefined-local-with-import-star-usage
     "F821", # undefined-name
     "F841", # unused-variable
     "FURB113", # repeated-append
diff --git a/api/services/account_service.py b/api/services/account_service.py
index c3d3354a1d83f..05b505f8a6218 100644
--- a/api/services/account_service.py
+++ b/api/services/account_service.py
@@ -1,4 +1,5 @@
 import base64
+import json
 import logging
 import secrets
 import uuid
@@ -12,12 +13,22 @@
 from configs import dify_config
 from constants.languages import language_timezone_mapping, languages
 from events.tenant_event import tenant_was_created
+from extensions.ext_database import db
 from extensions.ext_redis import redis_client
 from libs.helper import RateLimiter, TokenManager
 from libs.passport import PassportService
 from libs.password import compare_password, hash_password, valid_password
 from libs.rsa import generate_key_pair
-from models.account import *
+from models.account import (
+    Account,
+    AccountIntegrate,
+    AccountStatus,
+    Tenant,
+    TenantAccountJoin,
+    TenantAccountJoinRole,
+    TenantAccountRole,
+    TenantStatus,
+)
 from models.model import DifySetup
 from services.errors.account import (
     AccountAlreadyInTenantError,

From 499cc570821dafd26facbae1cab5192a5c6c4a3d Mon Sep 17 00:00:00 2001
From: =?UTF-8?q?=E9=9D=9E=E6=B3=95=E6=93=8D=E4=BD=9C?= <hjlarry@163.com>
Date: Thu, 10 Oct 2024 10:07:21 +0800
Subject: [PATCH 176/235] fix: response_format of model_parameters will not be
 removed (#9148)

---
 .../model_providers/__base/large_language_model.py              | 2 +-
 1 file changed, 1 insertion(+), 1 deletion(-)

diff --git a/api/core/model_runtime/model_providers/__base/large_language_model.py b/api/core/model_runtime/model_providers/__base/large_language_model.py
index ba88cc1f384fc..0027411a6e4e5 100644
--- a/api/core/model_runtime/model_providers/__base/large_language_model.py
+++ b/api/core/model_runtime/model_providers/__base/large_language_model.py
@@ -94,7 +94,7 @@ def invoke(
         )
 
         try:
-            if "response_format" in model_parameters:
+            if "response_format" in model_parameters and model_parameters["response_format"] in {"JSON", "XML"}:
                 result = self._code_block_mode_wrapper(
                     model=model,
                     credentials=credentials,

From 060897b25ba93db90499d42fd5b0fbdc36b96687 Mon Sep 17 00:00:00 2001
From: =?UTF-8?q?=E5=91=86=E8=90=8C=E9=97=B7=E6=B2=B9=E7=93=B6?=
 <253605712@qq.com>
Date: Thu, 10 Oct 2024 10:07:49 +0800
Subject: [PATCH 177/235] chore:add azure openai api version 2024-09-01-preview
 (#9141)

---
 .../model_providers/azure_openai/azure_openai.yaml             | 3 +++
 1 file changed, 3 insertions(+)

diff --git a/api/core/model_runtime/model_providers/azure_openai/azure_openai.yaml b/api/core/model_runtime/model_providers/azure_openai/azure_openai.yaml
index cb87bfdf6b9b7..093f57c51e2eb 100644
--- a/api/core/model_runtime/model_providers/azure_openai/azure_openai.yaml
+++ b/api/core/model_runtime/model_providers/azure_openai/azure_openai.yaml
@@ -53,6 +53,9 @@ model_credential_schema:
       type: select
       required: true
       options:
+        - label:
+            en_US: 2024-09-01-preview
+          value: 2024-09-01-preview
         - label:
             en_US: 2024-08-01-preview
           value: 2024-08-01-preview

From 37d5c166ca1a43a2270fe41e1acbc2d1609c6215 Mon Sep 17 00:00:00 2001
From: -LAN- <laipz8200@outlook.com>
Date: Thu, 10 Oct 2024 10:16:24 +0800
Subject: [PATCH 178/235] fix(migrations): correct parent_message_id for
 service-api records (#9132)

---
 ...744d88ed6_fix_wrong_service_api_history.py | 48 +++++++++++++++++++
 1 file changed, 48 insertions(+)
 create mode 100644 api/migrations/versions/2024_10_09_1329-d8e744d88ed6_fix_wrong_service_api_history.py

diff --git a/api/migrations/versions/2024_10_09_1329-d8e744d88ed6_fix_wrong_service_api_history.py b/api/migrations/versions/2024_10_09_1329-d8e744d88ed6_fix_wrong_service_api_history.py
new file mode 100644
index 0000000000000..b3b8dfa7d4f2f
--- /dev/null
+++ b/api/migrations/versions/2024_10_09_1329-d8e744d88ed6_fix_wrong_service_api_history.py
@@ -0,0 +1,48 @@
+"""fix wrong service-api history
+
+Revision ID: d8e744d88ed6
+Revises: 33f5fac87f29
+Create Date: 2024-10-09 13:29:23.548498
+
+"""
+from alembic import op
+from constants import UUID_NIL
+import models as models
+import sqlalchemy as sa
+
+
+# revision identifiers, used by Alembic.
+revision = 'd8e744d88ed6'
+down_revision = '33f5fac87f29'
+branch_labels = None
+depends_on = None
+
+# (UTC) release date of v0.9.0
+v0_9_0_release_date= '2024-09-29 12:00:00'
+
+def upgrade():
+    # ### commands auto generated by Alembic - please adjust! ###
+    sql = f"""UPDATE
+    public.messages
+SET
+    parent_message_id = '{UUID_NIL}'
+WHERE
+    invoke_from = 'service-api'
+    AND parent_message_id IS NULL
+    AND created_at >= '{v0_9_0_release_date}';"""
+    op.execute(sql)
+    # ### end Alembic commands ###
+
+
+def downgrade():
+    # ### commands auto generated by Alembic - please adjust! ###
+    sql = f"""UPDATE
+    public.messages
+SET
+    parent_message_id = NULL
+WHERE
+    invoke_from = 'service-api'
+    AND parent_message_id = '{UUID_NIL}'
+    AND created_at >= '{v0_9_0_release_date}';"""
+    op.execute(sql)
+    # ### end Alembic commands ###

From 2024a6c941fd618fa53349d09e2d69adb7acf2d9 Mon Sep 17 00:00:00 2001
From: luckylhb90 <luckylhb90@gmail.com>
Date: Thu, 10 Oct 2024 05:16:42 +0300
Subject: [PATCH 179/235] fix: vertex ai remote url error(Error: not enough
 values to unpack) (#9134)

Co-authored-by: hobo.l <hobo.l@binance.com>
---
 .../model_providers/vertex_ai/llm/llm.py            | 13 ++++++++++---
 1 file changed, 10 insertions(+), 3 deletions(-)

diff --git a/api/core/model_runtime/model_providers/vertex_ai/llm/llm.py b/api/core/model_runtime/model_providers/vertex_ai/llm/llm.py
index 1dd785d545408..1469de605525e 100644
--- a/api/core/model_runtime/model_providers/vertex_ai/llm/llm.py
+++ b/api/core/model_runtime/model_providers/vertex_ai/llm/llm.py
@@ -7,6 +7,7 @@
 from typing import Optional, Union, cast
 
 import google.auth.transport.requests
+import requests
 import vertexai.generative_models as glm
 from anthropic import AnthropicVertex, Stream
 from anthropic.types import (
@@ -653,9 +654,15 @@ def _format_message_to_glm_content(self, message: PromptMessage) -> glm.Content:
                     if c.type == PromptMessageContentType.TEXT:
                         parts.append(glm.Part.from_text(c.data))
                     else:
-                        metadata, data = c.data.split(",", 1)
-                        mime_type = metadata.split(";", 1)[0].split(":")[1]
-                        parts.append(glm.Part.from_data(mime_type=mime_type, data=data))
+                        message_content = cast(ImagePromptMessageContent, c)
+                        if not message_content.data.startswith("data:"):
+                            url_arr = message_content.data.split(".")
+                            mime_type = f"image/{url_arr[-1]}"
+                            parts.append(glm.Part.from_uri(mime_type=mime_type, uri=message_content.data))
+                        else:
+                            metadata, data = c.data.split(",", 1)
+                            mime_type = metadata.split(";", 1)[0].split(":")[1]
+                            parts.append(glm.Part.from_data(mime_type=mime_type, data=data))
                 glm_content = glm.Content(role="user", parts=parts)
             return glm_content
         elif isinstance(message, AssistantPromptMessage):

From 783a6b866bfaed59721389ed4c92245270a65507 Mon Sep 17 00:00:00 2001
From: crazywoola <100913391+crazywoola@users.noreply.github.com>
Date: Thu, 10 Oct 2024 10:25:45 +0800
Subject: [PATCH 180/235] fix: #9076 (#9150)

---
 .../base/chat/embedded-chatbot/theme/theme-context.ts           | 2 +-
 1 file changed, 1 insertion(+), 1 deletion(-)

diff --git a/web/app/components/base/chat/embedded-chatbot/theme/theme-context.ts b/web/app/components/base/chat/embedded-chatbot/theme/theme-context.ts
index 7a814e252125c..49994d93e2f1f 100644
--- a/web/app/components/base/chat/embedded-chatbot/theme/theme-context.ts
+++ b/web/app/components/base/chat/embedded-chatbot/theme/theme-context.ts
@@ -26,7 +26,7 @@ export class Theme {
     if (this.chatColorTheme !== null && this.chatColorTheme !== '') {
       this.primaryColor = this.chatColorTheme ?? '#1C64F2'
       this.backgroundHeaderColorStyle = `backgroundColor: ${this.primaryColor}`
-      this.backgroundButtonDefaultColorStyle = `backgroundColor: ${this.primaryColor}; color: ${colorFontOnHeaderStyle};`
+      this.backgroundButtonDefaultColorStyle = `backgroundColor: ${this.primaryColor}; color: ${this.colorFontOnHeaderStyle};`
       this.roundedBackgroundColorStyle = `backgroundColor: ${hexToRGBA(this.primaryColor, 0.05)}`
       this.chatBubbleColorStyle = `backgroundColor: ${hexToRGBA(this.primaryColor, 0.15)}`
       this.chatBubbleColor = `${hexToRGBA(this.primaryColor, 0.15)}`

From 62051d51716d7249ee9eb8ee7c1f1f9b9f2ecf35 Mon Sep 17 00:00:00 2001
From: ronaksingh27 <102898752+ronaksingh27@users.noreply.github.com>
Date: Thu, 10 Oct 2024 08:04:25 +0530
Subject: [PATCH 181/235] Corrected type annotation to "Any" from "any" all
 files in "model_providers" folder (#9135)

---
 .../model_runtime/model_providers/__base/tts_model.py     | 4 ++--
 .../model_runtime/model_providers/azure_openai/tts/tts.py | 6 +++---
 .../model_runtime/model_providers/fishaudio/tts/tts.py    | 8 ++++----
 api/core/model_runtime/model_providers/openai/tts/tts.py  | 6 +++---
 .../model_runtime/model_providers/sagemaker/tts/tts.py    | 4 ++--
 api/core/model_runtime/model_providers/tongyi/tts/tts.py  | 6 +++---
 .../model_runtime/model_providers/xinference/tts/tts.py   | 6 +++---
 7 files changed, 20 insertions(+), 20 deletions(-)

diff --git a/api/core/model_runtime/model_providers/__base/tts_model.py b/api/core/model_runtime/model_providers/__base/tts_model.py
index 70be9322a75d5..862ec29dafab8 100644
--- a/api/core/model_runtime/model_providers/__base/tts_model.py
+++ b/api/core/model_runtime/model_providers/__base/tts_model.py
@@ -1,7 +1,7 @@
 import logging
 import re
 from abc import abstractmethod
-from typing import Optional
+from typing import Any, Optional
 
 from pydantic import ConfigDict
 
@@ -88,7 +88,7 @@ def get_tts_model_voices(self, model: str, credentials: dict, language: Optional
             else:
                 return [{"name": d["name"], "value": d["mode"]} for d in voices]
 
-    def _get_model_default_voice(self, model: str, credentials: dict) -> any:
+    def _get_model_default_voice(self, model: str, credentials: dict) -> Any:
         """
         Get voice for given tts model
 
diff --git a/api/core/model_runtime/model_providers/azure_openai/tts/tts.py b/api/core/model_runtime/model_providers/azure_openai/tts/tts.py
index af178703a0695..133cc9f76e072 100644
--- a/api/core/model_runtime/model_providers/azure_openai/tts/tts.py
+++ b/api/core/model_runtime/model_providers/azure_openai/tts/tts.py
@@ -1,6 +1,6 @@
 import concurrent.futures
 import copy
-from typing import Optional
+from typing import Any, Optional
 
 from openai import AzureOpenAI
 
@@ -19,7 +19,7 @@ class AzureOpenAIText2SpeechModel(_CommonAzureOpenAI, TTSModel):
 
     def _invoke(
         self, model: str, tenant_id: str, credentials: dict, content_text: str, voice: str, user: Optional[str] = None
-    ) -> any:
+    ) -> Any:
         """
         _invoke text2speech model
 
@@ -56,7 +56,7 @@ def validate_credentials(self, model: str, credentials: dict) -> None:
         except Exception as ex:
             raise CredentialsValidateFailedError(str(ex))
 
-    def _tts_invoke_streaming(self, model: str, credentials: dict, content_text: str, voice: str) -> any:
+    def _tts_invoke_streaming(self, model: str, credentials: dict, content_text: str, voice: str) -> Any:
         """
         _tts_invoke_streaming text2speech model
         :param model: model name
diff --git a/api/core/model_runtime/model_providers/fishaudio/tts/tts.py b/api/core/model_runtime/model_providers/fishaudio/tts/tts.py
index 895a7a914c97d..e518d7b95b6e3 100644
--- a/api/core/model_runtime/model_providers/fishaudio/tts/tts.py
+++ b/api/core/model_runtime/model_providers/fishaudio/tts/tts.py
@@ -1,4 +1,4 @@
-from typing import Optional
+from typing import Any, Optional
 
 import httpx
 
@@ -46,7 +46,7 @@ def _invoke(
         content_text: str,
         voice: str,
         user: Optional[str] = None,
-    ) -> any:
+    ) -> Any:
         """
         Invoke text2speech model
 
@@ -87,7 +87,7 @@ def validate_credentials(self, credentials: dict, user: Optional[str] = None) ->
         except Exception as ex:
             raise CredentialsValidateFailedError(str(ex))
 
-    def _tts_invoke_streaming(self, model: str, credentials: dict, content_text: str, voice: str) -> any:
+    def _tts_invoke_streaming(self, model: str, credentials: dict, content_text: str, voice: str) -> Any:
         """
         Invoke streaming text2speech model
         :param model: model name
@@ -112,7 +112,7 @@ def _tts_invoke_streaming(self, model: str, credentials: dict, content_text: str
         except Exception as ex:
             raise InvokeBadRequestError(str(ex))
 
-    def _tts_invoke_streaming_sentence(self, credentials: dict, content_text: str, voice: Optional[str] = None) -> any:
+    def _tts_invoke_streaming_sentence(self, credentials: dict, content_text: str, voice: Optional[str] = None) -> Any:
         """
         Invoke streaming text2speech model
 
diff --git a/api/core/model_runtime/model_providers/openai/tts/tts.py b/api/core/model_runtime/model_providers/openai/tts/tts.py
index a14c91639b88f..2e57b95944c6b 100644
--- a/api/core/model_runtime/model_providers/openai/tts/tts.py
+++ b/api/core/model_runtime/model_providers/openai/tts/tts.py
@@ -1,5 +1,5 @@
 import concurrent.futures
-from typing import Optional
+from typing import Any, Optional
 
 from openai import OpenAI
 
@@ -16,7 +16,7 @@ class OpenAIText2SpeechModel(_CommonOpenAI, TTSModel):
 
     def _invoke(
         self, model: str, tenant_id: str, credentials: dict, content_text: str, voice: str, user: Optional[str] = None
-    ) -> any:
+    ) -> Any:
         """
         _invoke text2speech model
 
@@ -55,7 +55,7 @@ def validate_credentials(self, model: str, credentials: dict, user: Optional[str
         except Exception as ex:
             raise CredentialsValidateFailedError(str(ex))
 
-    def _tts_invoke_streaming(self, model: str, credentials: dict, content_text: str, voice: str) -> any:
+    def _tts_invoke_streaming(self, model: str, credentials: dict, content_text: str, voice: str) -> Any:
         """
         _tts_invoke_streaming text2speech model
 
diff --git a/api/core/model_runtime/model_providers/sagemaker/tts/tts.py b/api/core/model_runtime/model_providers/sagemaker/tts/tts.py
index 9652e1d650587..1a5afd18f996b 100644
--- a/api/core/model_runtime/model_providers/sagemaker/tts/tts.py
+++ b/api/core/model_runtime/model_providers/sagemaker/tts/tts.py
@@ -192,7 +192,7 @@ def _invoke_error_mapping(self) -> dict[type[InvokeError], list[type[Exception]]
             InvokeBadRequestError: [InvokeBadRequestError, KeyError, ValueError],
         }
 
-    def _get_model_default_voice(self, model: str, credentials: dict) -> any:
+    def _get_model_default_voice(self, model: str, credentials: dict) -> Any:
         return ""
 
     def _get_model_word_limit(self, model: str, credentials: dict) -> int:
@@ -225,7 +225,7 @@ def _invoke_sagemaker(self, payload: dict, endpoint: str):
         json_obj = json.loads(json_str)
         return json_obj
 
-    def _tts_invoke_streaming(self, model_type: str, payload: dict, sagemaker_endpoint: str) -> any:
+    def _tts_invoke_streaming(self, model_type: str, payload: dict, sagemaker_endpoint: str) -> Any:
         """
         _tts_invoke_streaming text2speech model
 
diff --git a/api/core/model_runtime/model_providers/tongyi/tts/tts.py b/api/core/model_runtime/model_providers/tongyi/tts/tts.py
index 48a38897a87e9..ca3b9fbc1c3c0 100644
--- a/api/core/model_runtime/model_providers/tongyi/tts/tts.py
+++ b/api/core/model_runtime/model_providers/tongyi/tts/tts.py
@@ -1,6 +1,6 @@
 import threading
 from queue import Queue
-from typing import Optional
+from typing import Any, Optional
 
 import dashscope
 from dashscope import SpeechSynthesizer
@@ -20,7 +20,7 @@ class TongyiText2SpeechModel(_CommonTongyi, TTSModel):
 
     def _invoke(
         self, model: str, tenant_id: str, credentials: dict, content_text: str, voice: str, user: Optional[str] = None
-    ) -> any:
+    ) -> Any:
         """
         _invoke text2speech model
 
@@ -58,7 +58,7 @@ def validate_credentials(self, model: str, credentials: dict, user: Optional[str
         except Exception as ex:
             raise CredentialsValidateFailedError(str(ex))
 
-    def _tts_invoke_streaming(self, model: str, credentials: dict, content_text: str, voice: str) -> any:
+    def _tts_invoke_streaming(self, model: str, credentials: dict, content_text: str, voice: str) -> Any:
         """
         _tts_invoke_streaming text2speech model
 
diff --git a/api/core/model_runtime/model_providers/xinference/tts/tts.py b/api/core/model_runtime/model_providers/xinference/tts/tts.py
index 6290e8551d802..3f46b50c3386d 100644
--- a/api/core/model_runtime/model_providers/xinference/tts/tts.py
+++ b/api/core/model_runtime/model_providers/xinference/tts/tts.py
@@ -1,5 +1,5 @@
 import concurrent.futures
-from typing import Optional
+from typing import Any, Optional
 
 from xinference_client.client.restful.restful_client import RESTfulAudioModelHandle
 
@@ -166,7 +166,7 @@ def get_tts_model_voices(self, model: str, credentials: dict, language: Optional
 
         return self.model_voices["__default"]["all"]
 
-    def _get_model_default_voice(self, model: str, credentials: dict) -> any:
+    def _get_model_default_voice(self, model: str, credentials: dict) -> Any:
         return ""
 
     def _get_model_word_limit(self, model: str, credentials: dict) -> int:
@@ -178,7 +178,7 @@ def _get_model_audio_type(self, model: str, credentials: dict) -> str:
     def _get_model_workers_limit(self, model: str, credentials: dict) -> int:
         return 5
 
-    def _tts_invoke_streaming(self, model: str, credentials: dict, content_text: str, voice: str) -> any:
+    def _tts_invoke_streaming(self, model: str, credentials: dict, content_text: str, voice: str) -> Any:
         """
         _tts_invoke_streaming text2speech model
 

From e947103b6d67ffb074532ea96c12a13928a7d82b Mon Sep 17 00:00:00 2001
From: Joe <79627742+ZhouhaoJiang@users.noreply.github.com>
Date: Thu, 10 Oct 2024 11:15:52 +0800
Subject: [PATCH 182/235] Feat/add workflow sys params (#9108)

Co-authored-by: Joel <iamjoel007@gmail.com>
---
 .../app/apps/advanced_chat/app_generator.py   |  2 ++
 api/core/app/apps/advanced_chat/app_runner.py |  3 ++
 .../advanced_chat/generate_task_pipeline.py   |  4 +++
 api/core/app/apps/workflow/app_generator.py   |  2 ++
 api/core/app/apps/workflow/app_runner.py      |  3 ++
 .../apps/workflow/generate_task_pipeline.py   |  3 ++
 api/core/app/entities/app_invoke_entities.py  |  2 ++
 .../task_pipeline/workflow_cycle_manage.py    |  3 ++
 api/core/workflow/enums.py                    |  3 ++
 web/app/components/workflow/index.tsx         |  6 ++--
 .../nodes/_base/components/variable/utils.ts  | 13 ++++++++
 .../components/workflow/nodes/start/panel.tsx | 33 +++++++++++++++++++
 12 files changed, 74 insertions(+), 3 deletions(-)

diff --git a/api/core/app/apps/advanced_chat/app_generator.py b/api/core/app/apps/advanced_chat/app_generator.py
index 445ef6d0ab19f..0d183596f3951 100644
--- a/api/core/app/apps/advanced_chat/app_generator.py
+++ b/api/core/app/apps/advanced_chat/app_generator.py
@@ -113,6 +113,7 @@ def generate(
             # always enable retriever resource in debugger mode
             app_config.additional_features.show_retrieve_source = True
 
+        workflow_run_id = str(uuid.uuid4())
         # init application generate entity
         application_generate_entity = AdvancedChatAppGenerateEntity(
             task_id=str(uuid.uuid4()),
@@ -127,6 +128,7 @@ def generate(
             invoke_from=invoke_from,
             extras=extras,
             trace_manager=trace_manager,
+            workflow_run_id=workflow_run_id,
         )
         contexts.tenant_id.set(application_generate_entity.app_config.tenant_id)
 
diff --git a/api/core/app/apps/advanced_chat/app_runner.py b/api/core/app/apps/advanced_chat/app_runner.py
index 1bca1e1b71adc..1dcd051d15049 100644
--- a/api/core/app/apps/advanced_chat/app_runner.py
+++ b/api/core/app/apps/advanced_chat/app_runner.py
@@ -149,6 +149,9 @@ def run(self) -> None:
                 SystemVariableKey.CONVERSATION_ID: self.conversation.id,
                 SystemVariableKey.USER_ID: user_id,
                 SystemVariableKey.DIALOGUE_COUNT: conversation_dialogue_count,
+                SystemVariableKey.APP_ID: app_config.app_id,
+                SystemVariableKey.WORKFLOW_ID: app_config.workflow_id,
+                SystemVariableKey.WORKFLOW_RUN_ID: self.application_generate_entity.workflow_run_id,
             }
 
             # init variable pool
diff --git a/api/core/app/apps/advanced_chat/generate_task_pipeline.py b/api/core/app/apps/advanced_chat/generate_task_pipeline.py
index 897b6fd063c76..5ce25ebf2804c 100644
--- a/api/core/app/apps/advanced_chat/generate_task_pipeline.py
+++ b/api/core/app/apps/advanced_chat/generate_task_pipeline.py
@@ -107,6 +107,10 @@ def __init__(
             SystemVariableKey.FILES: application_generate_entity.files,
             SystemVariableKey.CONVERSATION_ID: conversation.id,
             SystemVariableKey.USER_ID: user_id,
+            SystemVariableKey.DIALOGUE_COUNT: conversation.dialogue_count,
+            SystemVariableKey.APP_ID: application_generate_entity.app_config.app_id,
+            SystemVariableKey.WORKFLOW_ID: workflow.id,
+            SystemVariableKey.WORKFLOW_RUN_ID: application_generate_entity.workflow_run_id,
         }
 
         self._task_state = WorkflowTaskState()
diff --git a/api/core/app/apps/workflow/app_generator.py b/api/core/app/apps/workflow/app_generator.py
index 57a77591a09cd..bd0ab53278015 100644
--- a/api/core/app/apps/workflow/app_generator.py
+++ b/api/core/app/apps/workflow/app_generator.py
@@ -99,6 +99,7 @@ def generate(
         user_id = user.id if isinstance(user, Account) else user.session_id
         trace_manager = TraceQueueManager(app_model.id, user_id)
 
+        workflow_run_id = str(uuid.uuid4())
         # init application generate entity
         application_generate_entity = WorkflowAppGenerateEntity(
             task_id=str(uuid.uuid4()),
@@ -110,6 +111,7 @@ def generate(
             invoke_from=invoke_from,
             call_depth=call_depth,
             trace_manager=trace_manager,
+            workflow_run_id=workflow_run_id,
         )
         contexts.tenant_id.set(application_generate_entity.app_config.tenant_id)
 
diff --git a/api/core/app/apps/workflow/app_runner.py b/api/core/app/apps/workflow/app_runner.py
index 22ec228fa79b1..378a4bb8bc127 100644
--- a/api/core/app/apps/workflow/app_runner.py
+++ b/api/core/app/apps/workflow/app_runner.py
@@ -90,6 +90,9 @@ def run(self) -> None:
             system_inputs = {
                 SystemVariableKey.FILES: files,
                 SystemVariableKey.USER_ID: user_id,
+                SystemVariableKey.APP_ID: app_config.app_id,
+                SystemVariableKey.WORKFLOW_ID: app_config.workflow_id,
+                SystemVariableKey.WORKFLOW_RUN_ID: self.application_generate_entity.workflow_run_id,
             }
 
             variable_pool = VariablePool(
diff --git a/api/core/app/apps/workflow/generate_task_pipeline.py b/api/core/app/apps/workflow/generate_task_pipeline.py
index 798847a5070b8..3afc5053673e3 100644
--- a/api/core/app/apps/workflow/generate_task_pipeline.py
+++ b/api/core/app/apps/workflow/generate_task_pipeline.py
@@ -97,6 +97,9 @@ def __init__(
         self._workflow_system_variables = {
             SystemVariableKey.FILES: application_generate_entity.files,
             SystemVariableKey.USER_ID: user_id,
+            SystemVariableKey.APP_ID: application_generate_entity.app_config.app_id,
+            SystemVariableKey.WORKFLOW_ID: workflow.id,
+            SystemVariableKey.WORKFLOW_RUN_ID: application_generate_entity.workflow_run_id,
         }
 
         self._task_state = WorkflowTaskState()
diff --git a/api/core/app/entities/app_invoke_entities.py b/api/core/app/entities/app_invoke_entities.py
index 87ca51ef1b1db..e757c8db7539a 100644
--- a/api/core/app/entities/app_invoke_entities.py
+++ b/api/core/app/entities/app_invoke_entities.py
@@ -152,6 +152,7 @@ class AdvancedChatAppGenerateEntity(AppGenerateEntity):
 
     conversation_id: Optional[str] = None
     parent_message_id: Optional[str] = None
+    workflow_run_id: Optional[str] = None
     query: str
 
     class SingleIterationRunEntity(BaseModel):
@@ -172,6 +173,7 @@ class WorkflowAppGenerateEntity(AppGenerateEntity):
 
     # app config
     app_config: WorkflowUIBasedAppConfig
+    workflow_run_id: Optional[str] = None
 
     class SingleIterationRunEntity(BaseModel):
         """
diff --git a/api/core/app/task_pipeline/workflow_cycle_manage.py b/api/core/app/task_pipeline/workflow_cycle_manage.py
index f10189798f85f..4fc587db77c14 100644
--- a/api/core/app/task_pipeline/workflow_cycle_manage.py
+++ b/api/core/app/task_pipeline/workflow_cycle_manage.py
@@ -85,6 +85,9 @@ def _handle_workflow_run_start(self) -> WorkflowRun:
 
         # init workflow run
         workflow_run = WorkflowRun()
+        workflow_run_id = self._workflow_system_variables[SystemVariableKey.WORKFLOW_RUN_ID]
+        if workflow_run_id:
+            workflow_run.id = workflow_run_id
         workflow_run.tenant_id = self._workflow.tenant_id
         workflow_run.app_id = self._workflow.app_id
         workflow_run.sequence_number = new_sequence_number
diff --git a/api/core/workflow/enums.py b/api/core/workflow/enums.py
index da65f6b1fb95e..213ed57f57096 100644
--- a/api/core/workflow/enums.py
+++ b/api/core/workflow/enums.py
@@ -11,3 +11,6 @@ class SystemVariableKey(str, Enum):
     CONVERSATION_ID = "conversation_id"
     USER_ID = "user_id"
     DIALOGUE_COUNT = "dialogue_count"
+    APP_ID = "app_id"
+    WORKFLOW_ID = "workflow_id"
+    WORKFLOW_RUN_ID = "workflow_run_id"
diff --git a/web/app/components/workflow/index.tsx b/web/app/components/workflow/index.tsx
index cdccd60a3b5a1..938ae679c3d26 100644
--- a/web/app/components/workflow/index.tsx
+++ b/web/app/components/workflow/index.tsx
@@ -405,9 +405,9 @@ const WorkflowWrap = memo(() => {
   const initialFeatures: FeaturesData = {
     file: {
       image: {
-        enabled: !!features.file_upload?.image.enabled,
-        number_limits: features.file_upload?.image.number_limits || 3,
-        transfer_methods: features.file_upload?.image.transfer_methods || ['local_file', 'remote_url'],
+        enabled: !!features.file_upload?.image?.enabled,
+        number_limits: features.file_upload?.image?.number_limits || 3,
+        transfer_methods: features.file_upload?.image?.transfer_methods || ['local_file', 'remote_url'],
       },
     },
     opening: {
diff --git a/web/app/components/workflow/nodes/_base/components/variable/utils.ts b/web/app/components/workflow/nodes/_base/components/variable/utils.ts
index 3deec09dc2c9a..89ba4e5cf9342 100644
--- a/web/app/components/workflow/nodes/_base/components/variable/utils.ts
+++ b/web/app/components/workflow/nodes/_base/components/variable/utils.ts
@@ -116,6 +116,19 @@ const formatItem = (
         variable: 'sys.files',
         type: VarType.arrayFile,
       })
+      res.vars.push({
+        variable: 'sys.app_id',
+        type: VarType.string,
+      })
+      res.vars.push({
+        variable: 'sys.workflow_id',
+        type: VarType.string,
+      })
+      res.vars.push({
+        variable: 'sys.workflow_run_id',
+        type: VarType.string,
+      })
+
       break
     }
 
diff --git a/web/app/components/workflow/nodes/start/panel.tsx b/web/app/components/workflow/nodes/start/panel.tsx
index ce86a34265b61..3a1eed5ff4622 100644
--- a/web/app/components/workflow/nodes/start/panel.tsx
+++ b/web/app/components/workflow/nodes/start/panel.tsx
@@ -121,6 +121,39 @@ const Panel: FC<NodePanelProps<StartNodeType>> = ({
                   </div>
                 }
               />
+              <VarItem
+                readonly
+                payload={{
+                  variable: 'sys.app_id',
+                } as any}
+                rightContent={
+                  <div className='text-xs font-normal text-gray-500'>
+                    String
+                  </div>
+                }
+              />
+              <VarItem
+                readonly
+                payload={{
+                  variable: 'sys.workflow_id',
+                } as any}
+                rightContent={
+                  <div className='text-xs font-normal text-gray-500'>
+                    String
+                  </div>
+                }
+              />
+              <VarItem
+                readonly
+                payload={{
+                  variable: 'sys.workflow_run_id',
+                } as any}
+                rightContent={
+                  <div className='text-xs font-normal text-gray-500'>
+                    String
+                  </div>
+                }
+              />
             </div>
 
           </>

From bffb0919ccd8ef6390049b5801ec2122d9ea02b7 Mon Sep 17 00:00:00 2001
From: dai <90462400+daisuke0926dev@users.noreply.github.com>
Date: Thu, 10 Oct 2024 15:13:20 +0900
Subject: [PATCH 183/235] fix: Twitter to X(Twitter) (#9157)

---
 README.md    | 4 ++--
 README_AR.md | 2 +-
 README_CN.md | 4 ++--
 README_ES.md | 4 ++--
 README_FR.md | 4 ++--
 README_JA.md | 4 ++--
 README_KL.md | 4 ++--
 README_KR.md | 2 +-
 README_TR.md | 4 ++--
 README_VI.md | 4 ++--
 10 files changed, 18 insertions(+), 18 deletions(-)

diff --git a/README.md b/README.md
index ac4a4e5839352..75094d39dbf60 100644
--- a/README.md
+++ b/README.md
@@ -17,7 +17,7 @@
             alt="chat on Discord"></a>
     <a href="https://twitter.com/intent/follow?screen_name=dify_ai" target="_blank">
         <img src="https://img.shields.io/twitter/follow/dify_ai?logo=X&color=%20%23f5f5f5"
-            alt="follow on Twitter"></a>
+            alt="follow on X(Twitter)"></a>
     <a href="https://hub.docker.com/u/langgenius" target="_blank">
         <img alt="Docker Pulls" src="https://img.shields.io/docker/pulls/langgenius/dify-web?labelColor=%20%23FDB062&color=%20%23f79009"></a>
     <a href="https://github.com/langgenius/dify/graphs/commit-activity" target="_blank">
@@ -223,7 +223,7 @@ At the same time, please consider supporting Dify by sharing it on social media
 * [Github Discussion](https://github.com/langgenius/dify/discussions). Best for: sharing feedback and asking questions.
 * [GitHub Issues](https://github.com/langgenius/dify/issues). Best for: bugs you encounter using Dify.AI, and feature proposals. See our [Contribution Guide](https://github.com/langgenius/dify/blob/main/CONTRIBUTING.md).
 * [Discord](https://discord.gg/FngNHpbcY7). Best for: sharing your applications and hanging out with the community.
-* [Twitter](https://twitter.com/dify_ai). Best for: sharing your applications and hanging out with the community.
+* [X(Twitter)](https://twitter.com/dify_ai). Best for: sharing your applications and hanging out with the community.
 
 ## Star history
 
diff --git a/README_AR.md b/README_AR.md
index 42dbd9edff7b0..e46ba7373840c 100644
--- a/README_AR.md
+++ b/README_AR.md
@@ -17,7 +17,7 @@
             alt="chat on Discord"></a>
     <a href="https://twitter.com/intent/follow?screen_name=dify_ai" target="_blank">
         <img src="https://img.shields.io/twitter/follow/dify_ai?logo=X&color=%20%23f5f5f5"
-            alt="follow on Twitter"></a>
+            alt="follow on X(Twitter)"></a>
     <a href="https://hub.docker.com/u/langgenius" target="_blank">
         <img alt="Docker Pulls" src="https://img.shields.io/docker/pulls/langgenius/dify-web?labelColor=%20%23FDB062&color=%20%23f79009"></a>
     <a href="https://github.com/langgenius/dify/graphs/commit-activity" target="_blank">
diff --git a/README_CN.md b/README_CN.md
index afeca3147b37b..4553524ce53b5 100644
--- a/README_CN.md
+++ b/README_CN.md
@@ -17,7 +17,7 @@
             alt="chat on Discord"></a>
     <a href="https://twitter.com/intent/follow?screen_name=dify_ai" target="_blank">
         <img src="https://img.shields.io/twitter/follow/dify_ai?logo=X&color=%20%23f5f5f5"
-            alt="follow on Twitter"></a>
+            alt="follow on X(Twitter)"></a>
     <a href="https://hub.docker.com/u/langgenius" target="_blank">
         <img alt="Docker Pulls" src="https://img.shields.io/docker/pulls/langgenius/dify-web?labelColor=%20%23FDB062&color=%20%23f79009"></a>
     <a href="https://github.com/langgenius/dify/graphs/commit-activity" target="_blank">
@@ -236,7 +236,7 @@ docker compose up -d
 - [GitHub Issues](https://github.com/langgenius/dify/issues)。👉：使用 Dify.AI 时遇到的错误和问题，请参阅[贡献指南](CONTRIBUTING.md)。
 - [电子邮件支持](mailto:hello@dify.ai?subject=[GitHub]Questions%20About%20Dify)。👉：关于使用 Dify.AI 的问题。
 - [Discord](https://discord.gg/FngNHpbcY7)。👉：分享您的应用程序并与社区交流。
-- [Twitter](https://twitter.com/dify_ai)。👉：分享您的应用程序并与社区交流。
+- [X(Twitter)](https://twitter.com/dify_ai)。👉：分享您的应用程序并与社区交流。
 - [商业许可](mailto:business@dify.ai?subject=[GitHub]Business%20License%20Inquiry)。👉：有关商业用途许可 Dify.AI 的商业咨询。
  - [微信]() 👉：扫描下方二维码，添加微信好友，备注 Dify，我们将邀请您加入 Dify 社区。  
 <img src="./images/wechat.png" alt="wechat" width="100"/>
diff --git a/README_ES.md b/README_ES.md
index a6296aa93d7f9..7da5ac7b61fb1 100644
--- a/README_ES.md
+++ b/README_ES.md
@@ -17,7 +17,7 @@
             alt="chat en Discord"></a>
     <a href="https://twitter.com/intent/follow?screen_name=dify_ai" target="_blank">
         <img src="https://img.shields.io/twitter/follow/dify_ai?logo=X&color=%20%23f5f5f5"
-            alt="seguir en Twitter"></a>
+            alt="seguir en X(Twitter)"></a>
     <a href="https://hub.docker.com/u/langgenius" target="_blank">
         <img alt="Descargas de Docker" src="https://img.shields.io/docker/pulls/langgenius/dify-web?labelColor=%20%23FDB062&color=%20%23f79009"></a>
     <a href="https://github.com/langgenius/dify/graphs/commit-activity" target="_blank">
@@ -231,7 +231,7 @@ Al mismo tiempo, considera apoyar a Dify compartiéndolo en redes sociales y en
 * [Discusión en GitHub](https://github.com/langgenius/dify/discussions). Lo mejor para: compartir comentarios y hacer preguntas.
 * [Reporte de problemas en GitHub](https://github.com/langgenius/dify/issues). Lo mejor para: errores que encuentres usando Dify.AI y propuestas de características. Consulta nuestra [Guía de contribución](https://github.com/langgenius/dify/blob/main/CONTRIBUTING.md).
 * [Discord](https://discord.gg/FngNHpbcY7). Lo mejor para: compartir tus aplicaciones y pasar el rato con la comunidad.
-* [Twitter](https://twitter.com/dify_ai). Lo mejor para: compartir tus aplicaciones y pasar el rato con la comunidad.
+* [X(Twitter)](https://twitter.com/dify_ai). Lo mejor para: compartir tus aplicaciones y pasar el rato con la comunidad.
 
 ## Historial de Estrellas
 
diff --git a/README_FR.md b/README_FR.md
index da8b7baa3e725..15f6f2650f8f2 100644
--- a/README_FR.md
+++ b/README_FR.md
@@ -17,7 +17,7 @@
             alt="chat sur Discord"></a>
     <a href="https://twitter.com/intent/follow?screen_name=dify_ai" target="_blank">
         <img src="https://img.shields.io/twitter/follow/dify_ai?logo=X&color=%20%23f5f5f5"
-            alt="suivre sur Twitter"></a>
+            alt="suivre sur X(Twitter)"></a>
     <a href="https://hub.docker.com/u/langgenius" target="_blank">
         <img alt="Tirages Docker" src="https://img.shields.io/docker/pulls/langgenius/dify-web?labelColor=%20%23FDB062&color=%20%23f79009"></a>
     <a href="https://github.com/langgenius/dify/graphs/commit-activity" target="_blank">
@@ -229,7 +229,7 @@ Dans le même temps, veuillez envisager de soutenir Dify en le partageant sur le
 * [Discussion GitHub](https://github.com/langgenius/dify/discussions). Meilleur pour: partager des commentaires et poser des questions.
 * [Problèmes GitHub](https://github.com/langgenius/dify/issues). Meilleur pour: les bogues que vous rencontrez en utilisant Dify.AI et les propositions de fonctionnalités. Consultez notre [Guide de contribution](https://github.com/langgenius/dify/blob/main/CONTRIBUTING.md).
 * [Discord](https://discord.gg/FngNHpbcY7). Meilleur pour: partager vos applications et passer du temps avec la communauté.
-* [Twitter](https://twitter.com/dify_ai). Meilleur pour: partager vos applications et passer du temps avec la communauté.
+* [X(Twitter)](https://twitter.com/dify_ai). Meilleur pour: partager vos applications et passer du temps avec la communauté.
 
 ## Historique des étoiles
 
diff --git a/README_JA.md b/README_JA.md
index 8c6076362bec9..a2e6b173f5bfc 100644
--- a/README_JA.md
+++ b/README_JA.md
@@ -17,7 +17,7 @@
             alt="Discordでチャット"></a>
     <a href="https://twitter.com/intent/follow?screen_name=dify_ai" target="_blank">
         <img src="https://img.shields.io/twitter/follow/dify_ai?logo=X&color=%20%23f5f5f5"
-            alt="Twitterでフォロー"></a>
+            alt="X(Twitter)でフォロー"></a>
     <a href="https://hub.docker.com/u/langgenius" target="_blank">
         <img alt="Docker Pulls" src="https://img.shields.io/docker/pulls/langgenius/dify-web?labelColor=%20%23FDB062&color=%20%23f79009"></a>
     <a href="https://github.com/langgenius/dify/graphs/commit-activity" target="_blank">
@@ -228,7 +228,7 @@ docker compose up -d
 * [Github Discussion](https://github.com/langgenius/dify/discussions). 主に: フィードバックの共有や質問。
 * [GitHub Issues](https://github.com/langgenius/dify/issues). 主に: Dify.AIを使用する際に発生するエラーや問題については、[貢献ガイド](CONTRIBUTING_JA.md)を参照してください
 * [Discord](https://discord.gg/FngNHpbcY7). 主に: アプリケーションの共有やコミュニティとの交流。
-* [Twitter](https://twitter.com/dify_ai). 主に: アプリケーションの共有やコミュニティとの交流。
+* [X(Twitter)](https://twitter.com/dify_ai). 主に: アプリケーションの共有やコミュニティとの交流。
 
 
 
diff --git a/README_KL.md b/README_KL.md
index ce2f78570bdc9..8f2affdce5ae5 100644
--- a/README_KL.md
+++ b/README_KL.md
@@ -17,7 +17,7 @@
             alt="chat on Discord"></a>
     <a href="https://twitter.com/intent/follow?screen_name=dify_ai" target="_blank">
         <img src="https://img.shields.io/twitter/follow/dify_ai?logo=X&color=%20%23f5f5f5"
-            alt="follow on Twitter"></a>
+            alt="follow on X(Twitter)"></a>
     <a href="https://hub.docker.com/u/langgenius" target="_blank">
         <img alt="Docker Pulls" src="https://img.shields.io/docker/pulls/langgenius/dify-web?labelColor=%20%23FDB062&color=%20%23f79009"></a>
     <a href="https://github.com/langgenius/dify/graphs/commit-activity" target="_blank">
@@ -231,7 +231,7 @@ At the same time, please consider supporting Dify by sharing it on social media
 ). Best for: sharing feedback and asking questions.
 * [GitHub Issues](https://github.com/langgenius/dify/issues). Best for: bugs you encounter using Dify.AI, and feature proposals. See our [Contribution Guide](https://github.com/langgenius/dify/blob/main/CONTRIBUTING.md).
 * [Discord](https://discord.gg/FngNHpbcY7). Best for: sharing your applications and hanging out with the community.
-* [Twitter](https://twitter.com/dify_ai). Best for: sharing your applications and hanging out with the community.
+* [X(Twitter)](https://twitter.com/dify_ai). Best for: sharing your applications and hanging out with the community.
 
 ## Star History
 
diff --git a/README_KR.md b/README_KR.md
index 87409bfb8c29d..6c3a9ed7f6ea0 100644
--- a/README_KR.md
+++ b/README_KR.md
@@ -17,7 +17,7 @@
             alt="chat on Discord"></a>
     <a href="https://twitter.com/intent/follow?screen_name=dify_ai" target="_blank">
         <img src="https://img.shields.io/twitter/follow/dify_ai?logo=X&color=%20%23f5f5f5"
-            alt="follow on Twitter"></a>
+            alt="follow on X(Twitter)"></a>
     <a href="https://hub.docker.com/u/langgenius" target="_blank">
         <img alt="Docker Pulls" src="https://img.shields.io/docker/pulls/langgenius/dify-web?labelColor=%20%23FDB062&color=%20%23f79009"></a>
     <a href="https://github.com/langgenius/dify/graphs/commit-activity" target="_blank">
diff --git a/README_TR.md b/README_TR.md
index 08080a8c7bf75..a75889e576063 100644
--- a/README_TR.md
+++ b/README_TR.md
@@ -17,7 +17,7 @@
             alt="Discord'da sohbet et"></a>
     <a href="https://twitter.com/intent/follow?screen_name=dify_ai" target="_blank">
         <img src="https://img.shields.io/twitter/follow/dify_ai?logo=X&color=%20%23f5f5f5"
-            alt="Twitter'da takip et"></a>
+            alt="X(Twitter)'da takip et"></a>
     <a href="https://hub.docker.com/u/langgenius" target="_blank">
         <img alt="Docker Çekmeleri" src="https://img.shields.io/docker/pulls/langgenius/dify-web?labelColor=%20%23FDB062&color=%20%23f79009"></a>
     <a href="https://github.com/langgenius/dify/graphs/commit-activity" target="_blank">
@@ -226,7 +226,7 @@ Aynı zamanda, lütfen Dify'ı sosyal medyada, etkinliklerde ve konferanslarda p
 * [Github Tartışmaları](https://github.com/langgenius/dify/discussions). En uygun: geri bildirim paylaşmak ve soru sormak için.
 * [GitHub Sorunları](https://github.com/langgenius/dify/issues). En uygun: Dify.AI kullanırken karşılaştığınız hatalar ve özellik önerileri için. [Katkı Kılavuzumuza](https://github.com/langgenius/dify/blob/main/CONTRIBUTING.md) bakın.
 * [Discord](https://discord.gg/FngNHpbcY7). En uygun: uygulamalarınızı paylaşmak ve toplulukla vakit geçirmek için.
-* [Twitter](https://twitter.com/dify_ai). En uygun: uygulamalarınızı paylaşmak ve toplulukla vakit geçirmek için.
+* [X(Twitter)](https://twitter.com/dify_ai). En uygun: uygulamalarınızı paylaşmak ve toplulukla vakit geçirmek için.
 
 ## Star history
 
diff --git a/README_VI.md b/README_VI.md
index d67f024c7955b..8d49e4976626b 100644
--- a/README_VI.md
+++ b/README_VI.md
@@ -17,7 +17,7 @@
             alt="chat trên Discord"></a>
     <a href="https://twitter.com/intent/follow?screen_name=dify_ai" target="_blank">
         <img src="https://img.shields.io/twitter/follow/dify_ai?logo=X&color=%20%23f5f5f5"
-            alt="theo dõi trên Twitter"></a>
+            alt="theo dõi trên X(Twitter)"></a>
     <a href="https://hub.docker.com/u/langgenius" target="_blank">
         <img alt="Docker Pulls" src="https://img.shields.io/docker/pulls/langgenius/dify-web?labelColor=%20%23FDB062&color=%20%23f79009"></a>
     <a href="https://github.com/langgenius/dify/graphs/commit-activity" target="_blank">
@@ -223,7 +223,7 @@ Triển khai Dify lên nền tảng đám mây với một cú nhấp chuột b
 * [Thảo luận GitHub](https://github.com/langgenius/dify/discussions). Tốt nhất cho: chia sẻ phản hồi và đặt câu hỏi.
 * [Vấn đề GitHub](https://github.com/langgenius/dify/issues). Tốt nhất cho: lỗi bạn gặp phải khi sử dụng Dify.AI và đề xuất tính năng. Xem [Hướng dẫn Đóng góp](https://github.com/langgenius/dify/blob/main/CONTRIBUTING.md) của chúng tôi.
 * [Discord](https://discord.gg/FngNHpbcY7). Tốt nhất cho: chia sẻ ứng dụng của bạn và giao lưu với cộng đồng.
-* [Twitter](https://twitter.com/dify_ai). Tốt nhất cho: chia sẻ ứng dụng của bạn và giao lưu với cộng đồng.
+* [X(Twitter)](https://twitter.com/dify_ai). Tốt nhất cho: chia sẻ ứng dụng của bạn và giao lưu với cộng đồng.
 
 ## Lịch sử Yêu thích
 

From fc60b554a1c3d1767608bb7d54a61b769ac49fa6 Mon Sep 17 00:00:00 2001
From: Ziyu Huang <ziyu4huang@gmail.com>
Date: Thu, 10 Oct 2024 15:18:07 +0800
Subject: [PATCH 184/235] Fixes #9159: Modify to make it works to llama.cpp
 rerank API (#9160)

---
 .../model_providers/jina/rerank/rerank.py            | 12 ++++++++++--
 .../model_providers/localai/rerank/rerank.py         | 12 ++++++++++--
 2 files changed, 20 insertions(+), 4 deletions(-)

diff --git a/api/core/model_runtime/model_providers/jina/rerank/rerank.py b/api/core/model_runtime/model_providers/jina/rerank/rerank.py
index 79ca68914f979..03502076517f7 100644
--- a/api/core/model_runtime/model_providers/jina/rerank/rerank.py
+++ b/api/core/model_runtime/model_providers/jina/rerank/rerank.py
@@ -61,11 +61,19 @@ def _invoke(
 
             rerank_documents = []
             for result in results["results"]:
+                index = result["index"]
+                if "document" in result:
+                    text = result["document"]["text"]
+                else:
+                    # llama.cpp rerank maynot return original documents
+                    text = docs[index]
+
                 rerank_document = RerankDocument(
-                    index=result["index"],
-                    text=result["document"]["text"],
+                    index=index,
+                    text=text,
                     score=result["relevance_score"],
                 )
+
                 if score_threshold is None or result["relevance_score"] >= score_threshold:
                     rerank_documents.append(rerank_document)
 
diff --git a/api/core/model_runtime/model_providers/localai/rerank/rerank.py b/api/core/model_runtime/model_providers/localai/rerank/rerank.py
index 2b0f53bc19e8e..075b44658dcc3 100644
--- a/api/core/model_runtime/model_providers/localai/rerank/rerank.py
+++ b/api/core/model_runtime/model_providers/localai/rerank/rerank.py
@@ -70,11 +70,19 @@ def _invoke(
 
             rerank_documents = []
             for result in results["results"]:
+                index = result["index"]
+                if "document" in result:
+                    text = result["document"]["text"]
+                else:
+                    # llama.cpp rerank maynot return original documents
+                    text = docs[index]
+
                 rerank_document = RerankDocument(
-                    index=result["index"],
-                    text=result["document"]["text"],
+                    index=index,
+                    text=text,
                     score=result["relevance_score"],
                 )
+
                 if score_threshold is None or result["relevance_score"] >= score_threshold:
                     rerank_documents.append(rerank_document)
 

From 6b6e94da0822f8aebc4d95bffd0b536ac2152579 Mon Sep 17 00:00:00 2001
From: "Charlie.Wei" <luowei@cvte.com>
Date: Thu, 10 Oct 2024 15:26:38 +0800
Subject: [PATCH 185/235] Fix code indentation errors (#9164)

---
 .../model_providers/azure_openai/llm/llm.py   | 56 +++++++++----------
 1 file changed, 28 insertions(+), 28 deletions(-)

diff --git a/api/core/model_runtime/model_providers/azure_openai/llm/llm.py b/api/core/model_runtime/model_providers/azure_openai/llm/llm.py
index f63c80e58e215..b9cc3bb6727b9 100644
--- a/api/core/model_runtime/model_providers/azure_openai/llm/llm.py
+++ b/api/core/model_runtime/model_providers/azure_openai/llm/llm.py
@@ -312,39 +312,39 @@ def _chat_generate(
         if user:
             extra_model_kwargs["user"] = user
 
-            # clear illegal prompt messages
-            prompt_messages = self._clear_illegal_prompt_messages(model, prompt_messages)
-
-            block_as_stream = False
-            if model.startswith("o1"):
-                if stream:
-                    block_as_stream = True
-                    stream = False
-
-                    if "stream_options" in extra_model_kwargs:
-                        del extra_model_kwargs["stream_options"]
-
-                if "stop" in extra_model_kwargs:
-                    del extra_model_kwargs["stop"]
-
-            # chat model
-            response = client.chat.completions.create(
-                messages=[self._convert_prompt_message_to_dict(m) for m in prompt_messages],
-                model=model,
-                stream=stream,
-                **model_parameters,
-                **extra_model_kwargs,
-            )
+        # clear illegal prompt messages
+        prompt_messages = self._clear_illegal_prompt_messages(model, prompt_messages)
 
+        block_as_stream = False
+        if model.startswith("o1"):
             if stream:
-                return self._handle_chat_generate_stream_response(model, credentials, response, prompt_messages, tools)
+                block_as_stream = True
+                stream = False
+
+                if "stream_options" in extra_model_kwargs:
+                    del extra_model_kwargs["stream_options"]
+
+            if "stop" in extra_model_kwargs:
+                del extra_model_kwargs["stop"]
+
+        # chat model
+        response = client.chat.completions.create(
+            messages=[self._convert_prompt_message_to_dict(m) for m in prompt_messages],
+            model=model,
+            stream=stream,
+            **model_parameters,
+            **extra_model_kwargs,
+        )
+
+        if stream:
+            return self._handle_chat_generate_stream_response(model, credentials, response, prompt_messages, tools)
 
-            block_result = self._handle_chat_generate_response(model, credentials, response, prompt_messages, tools)
+        block_result = self._handle_chat_generate_response(model, credentials, response, prompt_messages, tools)
 
-            if block_as_stream:
-                return self._handle_chat_block_as_stream_response(block_result, prompt_messages, stop)
+        if block_as_stream:
+            return self._handle_chat_block_as_stream_response(block_result, prompt_messages, stop)
 
-            return block_result
+        return block_result
 
     def _handle_chat_block_as_stream_response(
         self,

From 6689b592ff404bad10e2ca27a288b43b707b0208 Mon Sep 17 00:00:00 2001
From: Yi Xiao <54782454+YIXIAO0@users.noreply.github.com>
Date: Thu, 10 Oct 2024 15:48:26 +0800
Subject: [PATCH 186/235] fix: delete offline "how to write a good knowledge
 description" helper (#9165)

---
 .../create/KnowledgeBaseInfo.tsx                     | 12 ------------
 web/app/components/datasets/rename-modal/index.tsx   |  5 -----
 web/app/components/datasets/settings/form/index.tsx  |  5 -----
 3 files changed, 22 deletions(-)

diff --git a/web/app/components/datasets/external-knowledge-base/create/KnowledgeBaseInfo.tsx b/web/app/components/datasets/external-knowledge-base/create/KnowledgeBaseInfo.tsx
index 1a2e83d2d6681..fec526b8811f7 100644
--- a/web/app/components/datasets/external-knowledge-base/create/KnowledgeBaseInfo.tsx
+++ b/web/app/components/datasets/external-knowledge-base/create/KnowledgeBaseInfo.tsx
@@ -1,5 +1,4 @@
 import React from 'react'
-import { RiBookOpenLine } from '@remixicon/react'
 import { useTranslation } from 'react-i18next'
 import Input from '@/app/components/base/input'
 
@@ -44,17 +43,6 @@ const KnowledgeBaseInfo: React.FC<KnowledgeBaseInfoProps> = ({ name, description
               placeholder={t('dataset.externalKnowledgeDescriptionPlaceholder') ?? ''}
               className={`flex h-20 py-2 p-3 self-stretch items-start rounded-lg bg-components-input-bg-normal ${description ? 'text-components-input-text-filled' : 'text-components-input-text-placeholder'} system-sm-regular`}
             />
-            <a
-              className='flex py-0.5 gap-1 self-stretch'
-              href='https://docs.dify.ai/features/datasets#how-to-write-a-good-dataset-description'
-              target="_blank"
-              rel="noopener noreferrer"
-            >
-              <div className='flex p-0.5 items-center gap-2'>
-                <RiBookOpenLine className='w-3 h-3 text-text-tertiary' />
-              </div>
-              <div className='flex-grow text-text-tertiary body-xs-regular'>{t('dataset.learnHowToWriteGoodKnowledgeDescription')}</div>
-            </a>
           </div>
         </div>
       </div>
diff --git a/web/app/components/datasets/rename-modal/index.tsx b/web/app/components/datasets/rename-modal/index.tsx
index 7e5fb228f78f7..7d6008c087a5e 100644
--- a/web/app/components/datasets/rename-modal/index.tsx
+++ b/web/app/components/datasets/rename-modal/index.tsx
@@ -3,7 +3,6 @@
 import type { MouseEventHandler } from 'react'
 import { useState } from 'react'
 import { RiCloseLine } from '@remixicon/react'
-import { BookOpenIcon } from '@heroicons/react/24/outline'
 import { useContext } from 'use-context-selector'
 import { useTranslation } from 'react-i18next'
 import cn from '@/utils/classnames'
@@ -94,10 +93,6 @@ const RenameDatasetModal = ({ show, dataset, onSuccess, onClose }: RenameDataset
               className='block px-3 py-2 w-full h-[88px] rounded-lg bg-gray-100 text-sm outline-none appearance-none resize-none'
               placeholder={t('datasetSettings.form.descPlaceholder') || ''}
             />
-            <a className='mt-2 flex items-center h-[18px] px-3 text-xs text-gray-500 hover:text-primary-600' href="https://docs.dify.ai/features/datasets#how-to-write-a-good-dataset-description" target='_blank' rel='noopener noreferrer'>
-              <BookOpenIcon className='w-3 h-[18px] mr-1' />
-              {t('datasetSettings.form.descWrite')}
-            </a>
           </div>
         </div>
       </div>
diff --git a/web/app/components/datasets/settings/form/index.tsx b/web/app/components/datasets/settings/form/index.tsx
index a3b6e79c3bd14..0bdfa3c5c883f 100644
--- a/web/app/components/datasets/settings/form/index.tsx
+++ b/web/app/components/datasets/settings/form/index.tsx
@@ -2,7 +2,6 @@
 import { useState } from 'react'
 import { useMount } from 'ahooks'
 import { useContext } from 'use-context-selector'
-import { BookOpenIcon } from '@heroicons/react/24/outline'
 import { useTranslation } from 'react-i18next'
 import { useSWRConfig } from 'swr'
 import { unstable_serialize } from 'swr/infinite'
@@ -208,10 +207,6 @@ const Form = () => {
             value={description}
             onChange={e => setDescription(e.target.value)}
           />
-          <a className='flex items-center h-[18px] px-3 text-xs text-gray-500' href="https://docs.dify.ai/features/datasets#how-to-write-a-good-dataset-description" target='_blank' rel='noopener noreferrer'>
-            <BookOpenIcon className='w-3 h-[18px] mr-1' />
-            {t('datasetSettings.form.descWrite')}
-          </a>
         </div>
       </div>
       <div className={rowClass}>

From 477beae3bb0c288574e85705f6e06699cc7f3bc4 Mon Sep 17 00:00:00 2001
From: takatost <takatost@users.noreply.github.com>
Date: Thu, 10 Oct 2024 16:34:06 +0800
Subject: [PATCH 187/235] fix: missing usage of metadata in the chatflow app
 (#9167)

---
 api/core/app/apps/advanced_chat/generate_task_pipeline.py | 5 +++++
 1 file changed, 5 insertions(+)

diff --git a/api/core/app/apps/advanced_chat/generate_task_pipeline.py b/api/core/app/apps/advanced_chat/generate_task_pipeline.py
index 5ce25ebf2804c..6bf684f8e40a3 100644
--- a/api/core/app/apps/advanced_chat/generate_task_pipeline.py
+++ b/api/core/app/apps/advanced_chat/generate_task_pipeline.py
@@ -45,6 +45,7 @@
 from core.app.task_pipeline.based_generate_task_pipeline import BasedGenerateTaskPipeline
 from core.app.task_pipeline.message_cycle_manage import MessageCycleManage
 from core.app.task_pipeline.workflow_cycle_manage import WorkflowCycleManage
+from core.model_runtime.entities.llm_entities import LLMUsage
 from core.model_runtime.utils.encoders import jsonable_encoder
 from core.ops.ops_trace_manager import TraceQueueManager
 from core.workflow.enums import SystemVariableKey
@@ -509,6 +510,10 @@ def _save_message(self, graph_runtime_state: Optional[GraphRuntimeState] = None)
             self._message.total_price = usage.total_price
             self._message.currency = usage.currency
 
+            self._task_state.metadata["usage"] = jsonable_encoder(usage)
+        else:
+            self._task_state.metadata["usage"] = jsonable_encoder(LLMUsage.empty_usage())
+
         db.session.commit()
 
         message_was_created.send(

From 68107fe3554a7af28622fc379eaf7016a78115bb Mon Sep 17 00:00:00 2001
From: Jyong <76649700+JohnJyong@users.noreply.github.com>
Date: Thu, 10 Oct 2024 16:43:11 +0800
Subject: [PATCH 188/235] downgrade elastic to 8.14 (#9171)

---
 api/poetry.lock            | 530 ++++++++++++++++++-------------------
 api/pyproject.toml         |   2 +-
 docker/docker-compose.yaml |   4 +-
 3 files changed, 264 insertions(+), 272 deletions(-)

diff --git a/api/poetry.lock b/api/poetry.lock
index 6b96350ffdb14..a63200cacaea7 100644
--- a/api/poetry.lock
+++ b/api/poetry.lock
@@ -293,13 +293,13 @@ alibabacloud-tea = "*"
 
 [[package]]
 name = "alibabacloud-tea"
-version = "0.3.10"
+version = "0.4.0"
 description = "The tea module of alibabaCloud Python SDK."
 optional = false
-python-versions = ">=3.6"
+python-versions = ">=3.7"
 files = [
-    {file = "alibabacloud-tea-0.3.10.tar.gz", hash = "sha256:bcf972416af5d8b5e671078c2ec20296dbc792e85e68acd685730a0a016afd2a"},
-    {file = "alibabacloud_tea-0.3.10-py3-none-any.whl", hash = "sha256:9136f302a3baea8a1528f500bf5d47c3727b827a09b5c14b283ca53578e30082"},
+    {file = "alibabacloud-tea-0.4.0.tar.gz", hash = "sha256:bdf72d747723bab190331b3c8593109fe2807504469bc0147f78c8c4945ed396"},
+    {file = "alibabacloud_tea-0.4.0-py3-none-any.whl", hash = "sha256:59fae5765e6654f884e130233df6fb61ca0fbe01a29ed0755a1cf099a3d4d863"},
 ]
 
 [package.dependencies]
@@ -364,12 +364,12 @@ alibabacloud-tea = ">=0.0.1"
 
 [[package]]
 name = "aliyun-python-sdk-core"
-version = "2.15.2"
+version = "2.16.0"
 description = "The core module of Aliyun Python SDK."
 optional = false
-python-versions = "*"
+python-versions = ">=3.7"
 files = [
-    {file = "aliyun-python-sdk-core-2.15.2.tar.gz", hash = "sha256:54f66a53e193c61c5e16ea4505a0cab43543f8ad2ef22833f69c4d5e5151c17d"},
+    {file = "aliyun-python-sdk-core-2.16.0.tar.gz", hash = "sha256:651caad597eb39d4fad6cf85133dffe92837d53bdf62db9d8f37dab6508bb8f9"},
 ]
 
 [package.dependencies]
@@ -698,13 +698,13 @@ msrest = ">=0.6.21"
 
 [[package]]
 name = "azure-storage-file-share"
-version = "12.18.0"
+version = "12.19.0"
 description = "Microsoft Azure Azure File Share Storage Client Library for Python"
 optional = false
 python-versions = ">=3.8"
 files = [
-    {file = "azure_storage_file_share-12.18.0-py3-none-any.whl", hash = "sha256:23ca35206a0cb8af0decd9d1363d0ad8ab31584a5e55cfc64528d7192ff748a9"},
-    {file = "azure_storage_file_share-12.18.0.tar.gz", hash = "sha256:0a81daee5e13598accde3c73b1aeccc6edfdcec5e957bf40150c0622fb95dc76"},
+    {file = "azure_storage_file_share-12.19.0-py3-none-any.whl", hash = "sha256:eac6cf1a454aba58af4e6ba450b36d16aa1d0c49679fb64ea8756bb896698c5b"},
+    {file = "azure_storage_file_share-12.19.0.tar.gz", hash = "sha256:ea7a4174dc6c52f50ac8c30f228159fcc3675d1f8ba771b8d0efcbc310740278"},
 ]
 
 [package.dependencies]
@@ -844,13 +844,13 @@ crt = ["botocore[crt] (>=1.21.0,<2.0a0)"]
 
 [[package]]
 name = "botocore"
-version = "1.35.34"
+version = "1.35.37"
 description = "Low-level, data-driven core of boto 3."
 optional = false
 python-versions = ">=3.8"
 files = [
-    {file = "botocore-1.35.34-py3-none-any.whl", hash = "sha256:ccb0fe397b11b81c9abc0c87029d17298e17bf658d8db5c0c5a551a12a207e7a"},
-    {file = "botocore-1.35.34.tar.gz", hash = "sha256:789b6501a3bb4a9591c1fe10da200cc315c1fa5df5ada19c720d8ef06439b3e3"},
+    {file = "botocore-1.35.37-py3-none-any.whl", hash = "sha256:64f965d4ba7adb8d79ce044c3aef7356e05dd74753cf7e9115b80f477845d920"},
+    {file = "botocore-1.35.37.tar.gz", hash = "sha256:b2b4d29bafd95b698344f2f0577bb67064adbf1735d8a0e3c7473daa59c23ba6"},
 ]
 
 [package.dependencies]
@@ -1257,101 +1257,116 @@ files = [
 
 [[package]]
 name = "charset-normalizer"
-version = "3.3.2"
+version = "3.4.0"
 description = "The Real First Universal Charset Detector. Open, modern and actively maintained alternative to Chardet."
 optional = false
 python-versions = ">=3.7.0"
 files = [
-    {file = "charset-normalizer-3.3.2.tar.gz", hash = "sha256:f30c3cb33b24454a82faecaf01b19c18562b1e89558fb6c56de4d9118a032fd5"},
-    {file = "charset_normalizer-3.3.2-cp310-cp310-macosx_10_9_universal2.whl", hash = "sha256:25baf083bf6f6b341f4121c2f3c548875ee6f5339300e08be3f2b2ba1721cdd3"},
-    {file = "charset_normalizer-3.3.2-cp310-cp310-macosx_10_9_x86_64.whl", hash = "sha256:06435b539f889b1f6f4ac1758871aae42dc3a8c0e24ac9e60c2384973ad73027"},
-    {file = "charset_normalizer-3.3.2-cp310-cp310-macosx_11_0_arm64.whl", hash = "sha256:9063e24fdb1e498ab71cb7419e24622516c4a04476b17a2dab57e8baa30d6e03"},
-    {file = "charset_normalizer-3.3.2-cp310-cp310-manylinux_2_17_aarch64.manylinux2014_aarch64.whl", hash = "sha256:6897af51655e3691ff853668779c7bad41579facacf5fd7253b0133308cf000d"},
-    {file = "charset_normalizer-3.3.2-cp310-cp310-manylinux_2_17_ppc64le.manylinux2014_ppc64le.whl", hash = "sha256:1d3193f4a680c64b4b6a9115943538edb896edc190f0b222e73761716519268e"},
-    {file = "charset_normalizer-3.3.2-cp310-cp310-manylinux_2_17_s390x.manylinux2014_s390x.whl", hash = "sha256:cd70574b12bb8a4d2aaa0094515df2463cb429d8536cfb6c7ce983246983e5a6"},
-    {file = "charset_normalizer-3.3.2-cp310-cp310-manylinux_2_17_x86_64.manylinux2014_x86_64.whl", hash = "sha256:8465322196c8b4d7ab6d1e049e4c5cb460d0394da4a27d23cc242fbf0034b6b5"},
-    {file = "charset_normalizer-3.3.2-cp310-cp310-manylinux_2_5_i686.manylinux1_i686.manylinux_2_17_i686.manylinux2014_i686.whl", hash = "sha256:a9a8e9031d613fd2009c182b69c7b2c1ef8239a0efb1df3f7c8da66d5dd3d537"},
-    {file = "charset_normalizer-3.3.2-cp310-cp310-musllinux_1_1_aarch64.whl", hash = "sha256:beb58fe5cdb101e3a055192ac291b7a21e3b7ef4f67fa1d74e331a7f2124341c"},
-    {file = "charset_normalizer-3.3.2-cp310-cp310-musllinux_1_1_i686.whl", hash = "sha256:e06ed3eb3218bc64786f7db41917d4e686cc4856944f53d5bdf83a6884432e12"},
-    {file = "charset_normalizer-3.3.2-cp310-cp310-musllinux_1_1_ppc64le.whl", hash = "sha256:2e81c7b9c8979ce92ed306c249d46894776a909505d8f5a4ba55b14206e3222f"},
-    {file = "charset_normalizer-3.3.2-cp310-cp310-musllinux_1_1_s390x.whl", hash = "sha256:572c3763a264ba47b3cf708a44ce965d98555f618ca42c926a9c1616d8f34269"},
-    {file = "charset_normalizer-3.3.2-cp310-cp310-musllinux_1_1_x86_64.whl", hash = "sha256:fd1abc0d89e30cc4e02e4064dc67fcc51bd941eb395c502aac3ec19fab46b519"},
-    {file = "charset_normalizer-3.3.2-cp310-cp310-win32.whl", hash = "sha256:3d47fa203a7bd9c5b6cee4736ee84ca03b8ef23193c0d1ca99b5089f72645c73"},
-    {file = "charset_normalizer-3.3.2-cp310-cp310-win_amd64.whl", hash = "sha256:10955842570876604d404661fbccbc9c7e684caf432c09c715ec38fbae45ae09"},
-    {file = "charset_normalizer-3.3.2-cp311-cp311-macosx_10_9_universal2.whl", hash = "sha256:802fe99cca7457642125a8a88a084cef28ff0cf9407060f7b93dca5aa25480db"},
-    {file = "charset_normalizer-3.3.2-cp311-cp311-macosx_10_9_x86_64.whl", hash = "sha256:573f6eac48f4769d667c4442081b1794f52919e7edada77495aaed9236d13a96"},
-    {file = "charset_normalizer-3.3.2-cp311-cp311-macosx_11_0_arm64.whl", hash = "sha256:549a3a73da901d5bc3ce8d24e0600d1fa85524c10287f6004fbab87672bf3e1e"},
-    {file = "charset_normalizer-3.3.2-cp311-cp311-manylinux_2_17_aarch64.manylinux2014_aarch64.whl", hash = "sha256:f27273b60488abe721a075bcca6d7f3964f9f6f067c8c4c605743023d7d3944f"},
-    {file = "charset_normalizer-3.3.2-cp311-cp311-manylinux_2_17_ppc64le.manylinux2014_ppc64le.whl", hash = "sha256:1ceae2f17a9c33cb48e3263960dc5fc8005351ee19db217e9b1bb15d28c02574"},
-    {file = "charset_normalizer-3.3.2-cp311-cp311-manylinux_2_17_s390x.manylinux2014_s390x.whl", hash = "sha256:65f6f63034100ead094b8744b3b97965785388f308a64cf8d7c34f2f2e5be0c4"},
-    {file = "charset_normalizer-3.3.2-cp311-cp311-manylinux_2_17_x86_64.manylinux2014_x86_64.whl", hash = "sha256:753f10e867343b4511128c6ed8c82f7bec3bd026875576dfd88483c5c73b2fd8"},
-    {file = "charset_normalizer-3.3.2-cp311-cp311-manylinux_2_5_i686.manylinux1_i686.manylinux_2_17_i686.manylinux2014_i686.whl", hash = "sha256:4a78b2b446bd7c934f5dcedc588903fb2f5eec172f3d29e52a9096a43722adfc"},
-    {file = "charset_normalizer-3.3.2-cp311-cp311-musllinux_1_1_aarch64.whl", hash = "sha256:e537484df0d8f426ce2afb2d0f8e1c3d0b114b83f8850e5f2fbea0e797bd82ae"},
-    {file = "charset_normalizer-3.3.2-cp311-cp311-musllinux_1_1_i686.whl", hash = "sha256:eb6904c354526e758fda7167b33005998fb68c46fbc10e013ca97f21ca5c8887"},
-    {file = "charset_normalizer-3.3.2-cp311-cp311-musllinux_1_1_ppc64le.whl", hash = "sha256:deb6be0ac38ece9ba87dea880e438f25ca3eddfac8b002a2ec3d9183a454e8ae"},
-    {file = "charset_normalizer-3.3.2-cp311-cp311-musllinux_1_1_s390x.whl", hash = "sha256:4ab2fe47fae9e0f9dee8c04187ce5d09f48eabe611be8259444906793ab7cbce"},
-    {file = "charset_normalizer-3.3.2-cp311-cp311-musllinux_1_1_x86_64.whl", hash = "sha256:80402cd6ee291dcb72644d6eac93785fe2c8b9cb30893c1af5b8fdd753b9d40f"},
-    {file = "charset_normalizer-3.3.2-cp311-cp311-win32.whl", hash = "sha256:7cd13a2e3ddeed6913a65e66e94b51d80a041145a026c27e6bb76c31a853c6ab"},
-    {file = "charset_normalizer-3.3.2-cp311-cp311-win_amd64.whl", hash = "sha256:663946639d296df6a2bb2aa51b60a2454ca1cb29835324c640dafb5ff2131a77"},
-    {file = "charset_normalizer-3.3.2-cp312-cp312-macosx_10_9_universal2.whl", hash = "sha256:0b2b64d2bb6d3fb9112bafa732def486049e63de9618b5843bcdd081d8144cd8"},
-    {file = "charset_normalizer-3.3.2-cp312-cp312-macosx_10_9_x86_64.whl", hash = "sha256:ddbb2551d7e0102e7252db79ba445cdab71b26640817ab1e3e3648dad515003b"},
-    {file = "charset_normalizer-3.3.2-cp312-cp312-macosx_11_0_arm64.whl", hash = "sha256:55086ee1064215781fff39a1af09518bc9255b50d6333f2e4c74ca09fac6a8f6"},
-    {file = "charset_normalizer-3.3.2-cp312-cp312-manylinux_2_17_aarch64.manylinux2014_aarch64.whl", hash = "sha256:8f4a014bc36d3c57402e2977dada34f9c12300af536839dc38c0beab8878f38a"},
-    {file = "charset_normalizer-3.3.2-cp312-cp312-manylinux_2_17_ppc64le.manylinux2014_ppc64le.whl", hash = "sha256:a10af20b82360ab00827f916a6058451b723b4e65030c5a18577c8b2de5b3389"},
-    {file = "charset_normalizer-3.3.2-cp312-cp312-manylinux_2_17_s390x.manylinux2014_s390x.whl", hash = "sha256:8d756e44e94489e49571086ef83b2bb8ce311e730092d2c34ca8f7d925cb20aa"},
-    {file = "charset_normalizer-3.3.2-cp312-cp312-manylinux_2_17_x86_64.manylinux2014_x86_64.whl", hash = "sha256:90d558489962fd4918143277a773316e56c72da56ec7aa3dc3dbbe20fdfed15b"},
-    {file = "charset_normalizer-3.3.2-cp312-cp312-manylinux_2_5_i686.manylinux1_i686.manylinux_2_17_i686.manylinux2014_i686.whl", hash = "sha256:6ac7ffc7ad6d040517be39eb591cac5ff87416c2537df6ba3cba3bae290c0fed"},
-    {file = "charset_normalizer-3.3.2-cp312-cp312-musllinux_1_1_aarch64.whl", hash = "sha256:7ed9e526742851e8d5cc9e6cf41427dfc6068d4f5a3bb03659444b4cabf6bc26"},
-    {file = "charset_normalizer-3.3.2-cp312-cp312-musllinux_1_1_i686.whl", hash = "sha256:8bdb58ff7ba23002a4c5808d608e4e6c687175724f54a5dade5fa8c67b604e4d"},
-    {file = "charset_normalizer-3.3.2-cp312-cp312-musllinux_1_1_ppc64le.whl", hash = "sha256:6b3251890fff30ee142c44144871185dbe13b11bab478a88887a639655be1068"},
-    {file = "charset_normalizer-3.3.2-cp312-cp312-musllinux_1_1_s390x.whl", hash = "sha256:b4a23f61ce87adf89be746c8a8974fe1c823c891d8f86eb218bb957c924bb143"},
-    {file = "charset_normalizer-3.3.2-cp312-cp312-musllinux_1_1_x86_64.whl", hash = "sha256:efcb3f6676480691518c177e3b465bcddf57cea040302f9f4e6e191af91174d4"},
-    {file = "charset_normalizer-3.3.2-cp312-cp312-win32.whl", hash = "sha256:d965bba47ddeec8cd560687584e88cf699fd28f192ceb452d1d7ee807c5597b7"},
-    {file = "charset_normalizer-3.3.2-cp312-cp312-win_amd64.whl", hash = "sha256:96b02a3dc4381e5494fad39be677abcb5e6634bf7b4fa83a6dd3112607547001"},
-    {file = "charset_normalizer-3.3.2-cp37-cp37m-macosx_10_9_x86_64.whl", hash = "sha256:95f2a5796329323b8f0512e09dbb7a1860c46a39da62ecb2324f116fa8fdc85c"},
-    {file = "charset_normalizer-3.3.2-cp37-cp37m-manylinux_2_17_aarch64.manylinux2014_aarch64.whl", hash = "sha256:c002b4ffc0be611f0d9da932eb0f704fe2602a9a949d1f738e4c34c75b0863d5"},
-    {file = "charset_normalizer-3.3.2-cp37-cp37m-manylinux_2_17_ppc64le.manylinux2014_ppc64le.whl", hash = "sha256:a981a536974bbc7a512cf44ed14938cf01030a99e9b3a06dd59578882f06f985"},
-    {file = "charset_normalizer-3.3.2-cp37-cp37m-manylinux_2_17_s390x.manylinux2014_s390x.whl", hash = "sha256:3287761bc4ee9e33561a7e058c72ac0938c4f57fe49a09eae428fd88aafe7bb6"},
-    {file = "charset_normalizer-3.3.2-cp37-cp37m-manylinux_2_17_x86_64.manylinux2014_x86_64.whl", hash = "sha256:42cb296636fcc8b0644486d15c12376cb9fa75443e00fb25de0b8602e64c1714"},
-    {file = "charset_normalizer-3.3.2-cp37-cp37m-manylinux_2_5_i686.manylinux1_i686.manylinux_2_17_i686.manylinux2014_i686.whl", hash = "sha256:0a55554a2fa0d408816b3b5cedf0045f4b8e1a6065aec45849de2d6f3f8e9786"},
-    {file = "charset_normalizer-3.3.2-cp37-cp37m-musllinux_1_1_aarch64.whl", hash = "sha256:c083af607d2515612056a31f0a8d9e0fcb5876b7bfc0abad3ecd275bc4ebc2d5"},
-    {file = "charset_normalizer-3.3.2-cp37-cp37m-musllinux_1_1_i686.whl", hash = "sha256:87d1351268731db79e0f8e745d92493ee2841c974128ef629dc518b937d9194c"},
-    {file = "charset_normalizer-3.3.2-cp37-cp37m-musllinux_1_1_ppc64le.whl", hash = "sha256:bd8f7df7d12c2db9fab40bdd87a7c09b1530128315d047a086fa3ae3435cb3a8"},
-    {file = "charset_normalizer-3.3.2-cp37-cp37m-musllinux_1_1_s390x.whl", hash = "sha256:c180f51afb394e165eafe4ac2936a14bee3eb10debc9d9e4db8958fe36afe711"},
-    {file = "charset_normalizer-3.3.2-cp37-cp37m-musllinux_1_1_x86_64.whl", hash = "sha256:8c622a5fe39a48f78944a87d4fb8a53ee07344641b0562c540d840748571b811"},
-    {file = "charset_normalizer-3.3.2-cp37-cp37m-win32.whl", hash = "sha256:db364eca23f876da6f9e16c9da0df51aa4f104a972735574842618b8c6d999d4"},
-    {file = "charset_normalizer-3.3.2-cp37-cp37m-win_amd64.whl", hash = "sha256:86216b5cee4b06df986d214f664305142d9c76df9b6512be2738aa72a2048f99"},
-    {file = "charset_normalizer-3.3.2-cp38-cp38-macosx_10_9_universal2.whl", hash = "sha256:6463effa3186ea09411d50efc7d85360b38d5f09b870c48e4600f63af490e56a"},
-    {file = "charset_normalizer-3.3.2-cp38-cp38-macosx_10_9_x86_64.whl", hash = "sha256:6c4caeef8fa63d06bd437cd4bdcf3ffefe6738fb1b25951440d80dc7df8c03ac"},
-    {file = "charset_normalizer-3.3.2-cp38-cp38-macosx_11_0_arm64.whl", hash = "sha256:37e55c8e51c236f95b033f6fb391d7d7970ba5fe7ff453dad675e88cf303377a"},
-    {file = "charset_normalizer-3.3.2-cp38-cp38-manylinux_2_17_aarch64.manylinux2014_aarch64.whl", hash = "sha256:fb69256e180cb6c8a894fee62b3afebae785babc1ee98b81cdf68bbca1987f33"},
-    {file = "charset_normalizer-3.3.2-cp38-cp38-manylinux_2_17_ppc64le.manylinux2014_ppc64le.whl", hash = "sha256:ae5f4161f18c61806f411a13b0310bea87f987c7d2ecdbdaad0e94eb2e404238"},
-    {file = "charset_normalizer-3.3.2-cp38-cp38-manylinux_2_17_s390x.manylinux2014_s390x.whl", hash = "sha256:b2b0a0c0517616b6869869f8c581d4eb2dd83a4d79e0ebcb7d373ef9956aeb0a"},
-    {file = "charset_normalizer-3.3.2-cp38-cp38-manylinux_2_17_x86_64.manylinux2014_x86_64.whl", hash = "sha256:45485e01ff4d3630ec0d9617310448a8702f70e9c01906b0d0118bdf9d124cf2"},
-    {file = "charset_normalizer-3.3.2-cp38-cp38-manylinux_2_5_i686.manylinux1_i686.manylinux_2_17_i686.manylinux2014_i686.whl", hash = "sha256:eb00ed941194665c332bf8e078baf037d6c35d7c4f3102ea2d4f16ca94a26dc8"},
-    {file = "charset_normalizer-3.3.2-cp38-cp38-musllinux_1_1_aarch64.whl", hash = "sha256:2127566c664442652f024c837091890cb1942c30937add288223dc895793f898"},
-    {file = "charset_normalizer-3.3.2-cp38-cp38-musllinux_1_1_i686.whl", hash = "sha256:a50aebfa173e157099939b17f18600f72f84eed3049e743b68ad15bd69b6bf99"},
-    {file = "charset_normalizer-3.3.2-cp38-cp38-musllinux_1_1_ppc64le.whl", hash = "sha256:4d0d1650369165a14e14e1e47b372cfcb31d6ab44e6e33cb2d4e57265290044d"},
-    {file = "charset_normalizer-3.3.2-cp38-cp38-musllinux_1_1_s390x.whl", hash = "sha256:923c0c831b7cfcb071580d3f46c4baf50f174be571576556269530f4bbd79d04"},
-    {file = "charset_normalizer-3.3.2-cp38-cp38-musllinux_1_1_x86_64.whl", hash = "sha256:06a81e93cd441c56a9b65d8e1d043daeb97a3d0856d177d5c90ba85acb3db087"},
-    {file = "charset_normalizer-3.3.2-cp38-cp38-win32.whl", hash = "sha256:6ef1d82a3af9d3eecdba2321dc1b3c238245d890843e040e41e470ffa64c3e25"},
-    {file = "charset_normalizer-3.3.2-cp38-cp38-win_amd64.whl", hash = "sha256:eb8821e09e916165e160797a6c17edda0679379a4be5c716c260e836e122f54b"},
-    {file = "charset_normalizer-3.3.2-cp39-cp39-macosx_10_9_universal2.whl", hash = "sha256:c235ebd9baae02f1b77bcea61bce332cb4331dc3617d254df3323aa01ab47bd4"},
-    {file = "charset_normalizer-3.3.2-cp39-cp39-macosx_10_9_x86_64.whl", hash = "sha256:5b4c145409bef602a690e7cfad0a15a55c13320ff7a3ad7ca59c13bb8ba4d45d"},
-    {file = "charset_normalizer-3.3.2-cp39-cp39-macosx_11_0_arm64.whl", hash = "sha256:68d1f8a9e9e37c1223b656399be5d6b448dea850bed7d0f87a8311f1ff3dabb0"},
-    {file = "charset_normalizer-3.3.2-cp39-cp39-manylinux_2_17_aarch64.manylinux2014_aarch64.whl", hash = "sha256:22afcb9f253dac0696b5a4be4a1c0f8762f8239e21b99680099abd9b2b1b2269"},
-    {file = "charset_normalizer-3.3.2-cp39-cp39-manylinux_2_17_ppc64le.manylinux2014_ppc64le.whl", hash = "sha256:e27ad930a842b4c5eb8ac0016b0a54f5aebbe679340c26101df33424142c143c"},
-    {file = "charset_normalizer-3.3.2-cp39-cp39-manylinux_2_17_s390x.manylinux2014_s390x.whl", hash = "sha256:1f79682fbe303db92bc2b1136016a38a42e835d932bab5b3b1bfcfbf0640e519"},
-    {file = "charset_normalizer-3.3.2-cp39-cp39-manylinux_2_17_x86_64.manylinux2014_x86_64.whl", hash = "sha256:b261ccdec7821281dade748d088bb6e9b69e6d15b30652b74cbbac25e280b796"},
-    {file = "charset_normalizer-3.3.2-cp39-cp39-manylinux_2_5_i686.manylinux1_i686.manylinux_2_17_i686.manylinux2014_i686.whl", hash = "sha256:122c7fa62b130ed55f8f285bfd56d5f4b4a5b503609d181f9ad85e55c89f4185"},
-    {file = "charset_normalizer-3.3.2-cp39-cp39-musllinux_1_1_aarch64.whl", hash = "sha256:d0eccceffcb53201b5bfebb52600a5fb483a20b61da9dbc885f8b103cbe7598c"},
-    {file = "charset_normalizer-3.3.2-cp39-cp39-musllinux_1_1_i686.whl", hash = "sha256:9f96df6923e21816da7e0ad3fd47dd8f94b2a5ce594e00677c0013018b813458"},
-    {file = "charset_normalizer-3.3.2-cp39-cp39-musllinux_1_1_ppc64le.whl", hash = "sha256:7f04c839ed0b6b98b1a7501a002144b76c18fb1c1850c8b98d458ac269e26ed2"},
-    {file = "charset_normalizer-3.3.2-cp39-cp39-musllinux_1_1_s390x.whl", hash = "sha256:34d1c8da1e78d2e001f363791c98a272bb734000fcef47a491c1e3b0505657a8"},
-    {file = "charset_normalizer-3.3.2-cp39-cp39-musllinux_1_1_x86_64.whl", hash = "sha256:ff8fa367d09b717b2a17a052544193ad76cd49979c805768879cb63d9ca50561"},
-    {file = "charset_normalizer-3.3.2-cp39-cp39-win32.whl", hash = "sha256:aed38f6e4fb3f5d6bf81bfa990a07806be9d83cf7bacef998ab1a9bd660a581f"},
-    {file = "charset_normalizer-3.3.2-cp39-cp39-win_amd64.whl", hash = "sha256:b01b88d45a6fcb69667cd6d2f7a9aeb4bf53760d7fc536bf679ec94fe9f3ff3d"},
-    {file = "charset_normalizer-3.3.2-py3-none-any.whl", hash = "sha256:3e4d1f6587322d2788836a99c69062fbb091331ec940e02d12d179c1d53e25fc"},
+    {file = "charset_normalizer-3.4.0-cp310-cp310-macosx_10_9_universal2.whl", hash = "sha256:4f9fc98dad6c2eaa32fc3af1417d95b5e3d08aff968df0cd320066def971f9a6"},
+    {file = "charset_normalizer-3.4.0-cp310-cp310-macosx_10_9_x86_64.whl", hash = "sha256:0de7b687289d3c1b3e8660d0741874abe7888100efe14bd0f9fd7141bcbda92b"},
+    {file = "charset_normalizer-3.4.0-cp310-cp310-macosx_11_0_arm64.whl", hash = "sha256:5ed2e36c3e9b4f21dd9422f6893dec0abf2cca553af509b10cd630f878d3eb99"},
+    {file = "charset_normalizer-3.4.0-cp310-cp310-manylinux_2_17_aarch64.manylinux2014_aarch64.whl", hash = "sha256:40d3ff7fc90b98c637bda91c89d51264a3dcf210cade3a2c6f838c7268d7a4ca"},
+    {file = "charset_normalizer-3.4.0-cp310-cp310-manylinux_2_17_ppc64le.manylinux2014_ppc64le.whl", hash = "sha256:1110e22af8ca26b90bd6364fe4c763329b0ebf1ee213ba32b68c73de5752323d"},
+    {file = "charset_normalizer-3.4.0-cp310-cp310-manylinux_2_17_s390x.manylinux2014_s390x.whl", hash = "sha256:86f4e8cca779080f66ff4f191a685ced73d2f72d50216f7112185dc02b90b9b7"},
+    {file = "charset_normalizer-3.4.0-cp310-cp310-manylinux_2_17_x86_64.manylinux2014_x86_64.whl", hash = "sha256:7f683ddc7eedd742e2889d2bfb96d69573fde1d92fcb811979cdb7165bb9c7d3"},
+    {file = "charset_normalizer-3.4.0-cp310-cp310-manylinux_2_5_i686.manylinux1_i686.manylinux_2_17_i686.manylinux2014_i686.whl", hash = "sha256:27623ba66c183eca01bf9ff833875b459cad267aeeb044477fedac35e19ba907"},
+    {file = "charset_normalizer-3.4.0-cp310-cp310-musllinux_1_2_aarch64.whl", hash = "sha256:f606a1881d2663630ea5b8ce2efe2111740df4b687bd78b34a8131baa007f79b"},
+    {file = "charset_normalizer-3.4.0-cp310-cp310-musllinux_1_2_i686.whl", hash = "sha256:0b309d1747110feb25d7ed6b01afdec269c647d382c857ef4663bbe6ad95a912"},
+    {file = "charset_normalizer-3.4.0-cp310-cp310-musllinux_1_2_ppc64le.whl", hash = "sha256:136815f06a3ae311fae551c3df1f998a1ebd01ddd424aa5603a4336997629e95"},
+    {file = "charset_normalizer-3.4.0-cp310-cp310-musllinux_1_2_s390x.whl", hash = "sha256:14215b71a762336254351b00ec720a8e85cada43b987da5a042e4ce3e82bd68e"},
+    {file = "charset_normalizer-3.4.0-cp310-cp310-musllinux_1_2_x86_64.whl", hash = "sha256:79983512b108e4a164b9c8d34de3992f76d48cadc9554c9e60b43f308988aabe"},
+    {file = "charset_normalizer-3.4.0-cp310-cp310-win32.whl", hash = "sha256:c94057af19bc953643a33581844649a7fdab902624d2eb739738a30e2b3e60fc"},
+    {file = "charset_normalizer-3.4.0-cp310-cp310-win_amd64.whl", hash = "sha256:55f56e2ebd4e3bc50442fbc0888c9d8c94e4e06a933804e2af3e89e2f9c1c749"},
+    {file = "charset_normalizer-3.4.0-cp311-cp311-macosx_10_9_universal2.whl", hash = "sha256:0d99dd8ff461990f12d6e42c7347fd9ab2532fb70e9621ba520f9e8637161d7c"},
+    {file = "charset_normalizer-3.4.0-cp311-cp311-macosx_10_9_x86_64.whl", hash = "sha256:c57516e58fd17d03ebe67e181a4e4e2ccab1168f8c2976c6a334d4f819fe5944"},
+    {file = "charset_normalizer-3.4.0-cp311-cp311-macosx_11_0_arm64.whl", hash = "sha256:6dba5d19c4dfab08e58d5b36304b3f92f3bd5d42c1a3fa37b5ba5cdf6dfcbcee"},
+    {file = "charset_normalizer-3.4.0-cp311-cp311-manylinux_2_17_aarch64.manylinux2014_aarch64.whl", hash = "sha256:bf4475b82be41b07cc5e5ff94810e6a01f276e37c2d55571e3fe175e467a1a1c"},
+    {file = "charset_normalizer-3.4.0-cp311-cp311-manylinux_2_17_ppc64le.manylinux2014_ppc64le.whl", hash = "sha256:ce031db0408e487fd2775d745ce30a7cd2923667cf3b69d48d219f1d8f5ddeb6"},
+    {file = "charset_normalizer-3.4.0-cp311-cp311-manylinux_2_17_s390x.manylinux2014_s390x.whl", hash = "sha256:8ff4e7cdfdb1ab5698e675ca622e72d58a6fa2a8aa58195de0c0061288e6e3ea"},
+    {file = "charset_normalizer-3.4.0-cp311-cp311-manylinux_2_17_x86_64.manylinux2014_x86_64.whl", hash = "sha256:3710a9751938947e6327ea9f3ea6332a09bf0ba0c09cae9cb1f250bd1f1549bc"},
+    {file = "charset_normalizer-3.4.0-cp311-cp311-manylinux_2_5_i686.manylinux1_i686.manylinux_2_17_i686.manylinux2014_i686.whl", hash = "sha256:82357d85de703176b5587dbe6ade8ff67f9f69a41c0733cf2425378b49954de5"},
+    {file = "charset_normalizer-3.4.0-cp311-cp311-musllinux_1_2_aarch64.whl", hash = "sha256:47334db71978b23ebcf3c0f9f5ee98b8d65992b65c9c4f2d34c2eaf5bcaf0594"},
+    {file = "charset_normalizer-3.4.0-cp311-cp311-musllinux_1_2_i686.whl", hash = "sha256:8ce7fd6767a1cc5a92a639b391891bf1c268b03ec7e021c7d6d902285259685c"},
+    {file = "charset_normalizer-3.4.0-cp311-cp311-musllinux_1_2_ppc64le.whl", hash = "sha256:f1a2f519ae173b5b6a2c9d5fa3116ce16e48b3462c8b96dfdded11055e3d6365"},
+    {file = "charset_normalizer-3.4.0-cp311-cp311-musllinux_1_2_s390x.whl", hash = "sha256:63bc5c4ae26e4bc6be6469943b8253c0fd4e4186c43ad46e713ea61a0ba49129"},
+    {file = "charset_normalizer-3.4.0-cp311-cp311-musllinux_1_2_x86_64.whl", hash = "sha256:bcb4f8ea87d03bc51ad04add8ceaf9b0f085ac045ab4d74e73bbc2dc033f0236"},
+    {file = "charset_normalizer-3.4.0-cp311-cp311-win32.whl", hash = "sha256:9ae4ef0b3f6b41bad6366fb0ea4fc1d7ed051528e113a60fa2a65a9abb5b1d99"},
+    {file = "charset_normalizer-3.4.0-cp311-cp311-win_amd64.whl", hash = "sha256:cee4373f4d3ad28f1ab6290684d8e2ebdb9e7a1b74fdc39e4c211995f77bec27"},
+    {file = "charset_normalizer-3.4.0-cp312-cp312-macosx_10_13_universal2.whl", hash = "sha256:0713f3adb9d03d49d365b70b84775d0a0d18e4ab08d12bc46baa6132ba78aaf6"},
+    {file = "charset_normalizer-3.4.0-cp312-cp312-macosx_10_13_x86_64.whl", hash = "sha256:de7376c29d95d6719048c194a9cf1a1b0393fbe8488a22008610b0361d834ecf"},
+    {file = "charset_normalizer-3.4.0-cp312-cp312-macosx_11_0_arm64.whl", hash = "sha256:4a51b48f42d9358460b78725283f04bddaf44a9358197b889657deba38f329db"},
+    {file = "charset_normalizer-3.4.0-cp312-cp312-manylinux_2_17_aarch64.manylinux2014_aarch64.whl", hash = "sha256:b295729485b06c1a0683af02a9e42d2caa9db04a373dc38a6a58cdd1e8abddf1"},
+    {file = "charset_normalizer-3.4.0-cp312-cp312-manylinux_2_17_ppc64le.manylinux2014_ppc64le.whl", hash = "sha256:ee803480535c44e7f5ad00788526da7d85525cfefaf8acf8ab9a310000be4b03"},
+    {file = "charset_normalizer-3.4.0-cp312-cp312-manylinux_2_17_s390x.manylinux2014_s390x.whl", hash = "sha256:3d59d125ffbd6d552765510e3f31ed75ebac2c7470c7274195b9161a32350284"},
+    {file = "charset_normalizer-3.4.0-cp312-cp312-manylinux_2_17_x86_64.manylinux2014_x86_64.whl", hash = "sha256:8cda06946eac330cbe6598f77bb54e690b4ca93f593dee1568ad22b04f347c15"},
+    {file = "charset_normalizer-3.4.0-cp312-cp312-manylinux_2_5_i686.manylinux1_i686.manylinux_2_17_i686.manylinux2014_i686.whl", hash = "sha256:07afec21bbbbf8a5cc3651aa96b980afe2526e7f048fdfb7f1014d84acc8b6d8"},
+    {file = "charset_normalizer-3.4.0-cp312-cp312-musllinux_1_2_aarch64.whl", hash = "sha256:6b40e8d38afe634559e398cc32b1472f376a4099c75fe6299ae607e404c033b2"},
+    {file = "charset_normalizer-3.4.0-cp312-cp312-musllinux_1_2_i686.whl", hash = "sha256:b8dcd239c743aa2f9c22ce674a145e0a25cb1566c495928440a181ca1ccf6719"},
+    {file = "charset_normalizer-3.4.0-cp312-cp312-musllinux_1_2_ppc64le.whl", hash = "sha256:84450ba661fb96e9fd67629b93d2941c871ca86fc38d835d19d4225ff946a631"},
+    {file = "charset_normalizer-3.4.0-cp312-cp312-musllinux_1_2_s390x.whl", hash = "sha256:44aeb140295a2f0659e113b31cfe92c9061622cadbc9e2a2f7b8ef6b1e29ef4b"},
+    {file = "charset_normalizer-3.4.0-cp312-cp312-musllinux_1_2_x86_64.whl", hash = "sha256:1db4e7fefefd0f548d73e2e2e041f9df5c59e178b4c72fbac4cc6f535cfb1565"},
+    {file = "charset_normalizer-3.4.0-cp312-cp312-win32.whl", hash = "sha256:5726cf76c982532c1863fb64d8c6dd0e4c90b6ece9feb06c9f202417a31f7dd7"},
+    {file = "charset_normalizer-3.4.0-cp312-cp312-win_amd64.whl", hash = "sha256:b197e7094f232959f8f20541ead1d9862ac5ebea1d58e9849c1bf979255dfac9"},
+    {file = "charset_normalizer-3.4.0-cp313-cp313-macosx_10_13_universal2.whl", hash = "sha256:dd4eda173a9fcccb5f2e2bd2a9f423d180194b1bf17cf59e3269899235b2a114"},
+    {file = "charset_normalizer-3.4.0-cp313-cp313-macosx_10_13_x86_64.whl", hash = "sha256:e9e3c4c9e1ed40ea53acf11e2a386383c3304212c965773704e4603d589343ed"},
+    {file = "charset_normalizer-3.4.0-cp313-cp313-macosx_11_0_arm64.whl", hash = "sha256:92a7e36b000bf022ef3dbb9c46bfe2d52c047d5e3f3343f43204263c5addc250"},
+    {file = "charset_normalizer-3.4.0-cp313-cp313-manylinux_2_17_aarch64.manylinux2014_aarch64.whl", hash = "sha256:54b6a92d009cbe2fb11054ba694bc9e284dad30a26757b1e372a1fdddaf21920"},
+    {file = "charset_normalizer-3.4.0-cp313-cp313-manylinux_2_17_ppc64le.manylinux2014_ppc64le.whl", hash = "sha256:1ffd9493de4c922f2a38c2bf62b831dcec90ac673ed1ca182fe11b4d8e9f2a64"},
+    {file = "charset_normalizer-3.4.0-cp313-cp313-manylinux_2_17_s390x.manylinux2014_s390x.whl", hash = "sha256:35c404d74c2926d0287fbd63ed5d27eb911eb9e4a3bb2c6d294f3cfd4a9e0c23"},
+    {file = "charset_normalizer-3.4.0-cp313-cp313-manylinux_2_17_x86_64.manylinux2014_x86_64.whl", hash = "sha256:4796efc4faf6b53a18e3d46343535caed491776a22af773f366534056c4e1fbc"},
+    {file = "charset_normalizer-3.4.0-cp313-cp313-manylinux_2_5_i686.manylinux1_i686.manylinux_2_17_i686.manylinux2014_i686.whl", hash = "sha256:e7fdd52961feb4c96507aa649550ec2a0d527c086d284749b2f582f2d40a2e0d"},
+    {file = "charset_normalizer-3.4.0-cp313-cp313-musllinux_1_2_aarch64.whl", hash = "sha256:92db3c28b5b2a273346bebb24857fda45601aef6ae1c011c0a997106581e8a88"},
+    {file = "charset_normalizer-3.4.0-cp313-cp313-musllinux_1_2_i686.whl", hash = "sha256:ab973df98fc99ab39080bfb0eb3a925181454d7c3ac8a1e695fddfae696d9e90"},
+    {file = "charset_normalizer-3.4.0-cp313-cp313-musllinux_1_2_ppc64le.whl", hash = "sha256:4b67fdab07fdd3c10bb21edab3cbfe8cf5696f453afce75d815d9d7223fbe88b"},
+    {file = "charset_normalizer-3.4.0-cp313-cp313-musllinux_1_2_s390x.whl", hash = "sha256:aa41e526a5d4a9dfcfbab0716c7e8a1b215abd3f3df5a45cf18a12721d31cb5d"},
+    {file = "charset_normalizer-3.4.0-cp313-cp313-musllinux_1_2_x86_64.whl", hash = "sha256:ffc519621dce0c767e96b9c53f09c5d215578e10b02c285809f76509a3931482"},
+    {file = "charset_normalizer-3.4.0-cp313-cp313-win32.whl", hash = "sha256:f19c1585933c82098c2a520f8ec1227f20e339e33aca8fa6f956f6691b784e67"},
+    {file = "charset_normalizer-3.4.0-cp313-cp313-win_amd64.whl", hash = "sha256:707b82d19e65c9bd28b81dde95249b07bf9f5b90ebe1ef17d9b57473f8a64b7b"},
+    {file = "charset_normalizer-3.4.0-cp37-cp37m-macosx_10_9_x86_64.whl", hash = "sha256:dbe03226baf438ac4fda9e2d0715022fd579cb641c4cf639fa40d53b2fe6f3e2"},
+    {file = "charset_normalizer-3.4.0-cp37-cp37m-manylinux_2_17_aarch64.manylinux2014_aarch64.whl", hash = "sha256:dd9a8bd8900e65504a305bf8ae6fa9fbc66de94178c420791d0293702fce2df7"},
+    {file = "charset_normalizer-3.4.0-cp37-cp37m-manylinux_2_17_ppc64le.manylinux2014_ppc64le.whl", hash = "sha256:b8831399554b92b72af5932cdbbd4ddc55c55f631bb13ff8fe4e6536a06c5c51"},
+    {file = "charset_normalizer-3.4.0-cp37-cp37m-manylinux_2_17_s390x.manylinux2014_s390x.whl", hash = "sha256:a14969b8691f7998e74663b77b4c36c0337cb1df552da83d5c9004a93afdb574"},
+    {file = "charset_normalizer-3.4.0-cp37-cp37m-manylinux_2_17_x86_64.manylinux2014_x86_64.whl", hash = "sha256:dcaf7c1524c0542ee2fc82cc8ec337f7a9f7edee2532421ab200d2b920fc97cf"},
+    {file = "charset_normalizer-3.4.0-cp37-cp37m-manylinux_2_5_i686.manylinux1_i686.manylinux_2_17_i686.manylinux2014_i686.whl", hash = "sha256:425c5f215d0eecee9a56cdb703203dda90423247421bf0d67125add85d0c4455"},
+    {file = "charset_normalizer-3.4.0-cp37-cp37m-musllinux_1_2_aarch64.whl", hash = "sha256:d5b054862739d276e09928de37c79ddeec42a6e1bfc55863be96a36ba22926f6"},
+    {file = "charset_normalizer-3.4.0-cp37-cp37m-musllinux_1_2_i686.whl", hash = "sha256:f3e73a4255342d4eb26ef6df01e3962e73aa29baa3124a8e824c5d3364a65748"},
+    {file = "charset_normalizer-3.4.0-cp37-cp37m-musllinux_1_2_ppc64le.whl", hash = "sha256:2f6c34da58ea9c1a9515621f4d9ac379871a8f21168ba1b5e09d74250de5ad62"},
+    {file = "charset_normalizer-3.4.0-cp37-cp37m-musllinux_1_2_s390x.whl", hash = "sha256:f09cb5a7bbe1ecae6e87901a2eb23e0256bb524a79ccc53eb0b7629fbe7677c4"},
+    {file = "charset_normalizer-3.4.0-cp37-cp37m-musllinux_1_2_x86_64.whl", hash = "sha256:0099d79bdfcf5c1f0c2c72f91516702ebf8b0b8ddd8905f97a8aecf49712c621"},
+    {file = "charset_normalizer-3.4.0-cp37-cp37m-win32.whl", hash = "sha256:9c98230f5042f4945f957d006edccc2af1e03ed5e37ce7c373f00a5a4daa6149"},
+    {file = "charset_normalizer-3.4.0-cp37-cp37m-win_amd64.whl", hash = "sha256:62f60aebecfc7f4b82e3f639a7d1433a20ec32824db2199a11ad4f5e146ef5ee"},
+    {file = "charset_normalizer-3.4.0-cp38-cp38-macosx_10_9_universal2.whl", hash = "sha256:af73657b7a68211996527dbfeffbb0864e043d270580c5aef06dc4b659a4b578"},
+    {file = "charset_normalizer-3.4.0-cp38-cp38-macosx_10_9_x86_64.whl", hash = "sha256:cab5d0b79d987c67f3b9e9c53f54a61360422a5a0bc075f43cab5621d530c3b6"},
+    {file = "charset_normalizer-3.4.0-cp38-cp38-macosx_11_0_arm64.whl", hash = "sha256:9289fd5dddcf57bab41d044f1756550f9e7cf0c8e373b8cdf0ce8773dc4bd417"},
+    {file = "charset_normalizer-3.4.0-cp38-cp38-manylinux_2_17_aarch64.manylinux2014_aarch64.whl", hash = "sha256:6b493a043635eb376e50eedf7818f2f322eabbaa974e948bd8bdd29eb7ef2a51"},
+    {file = "charset_normalizer-3.4.0-cp38-cp38-manylinux_2_17_ppc64le.manylinux2014_ppc64le.whl", hash = "sha256:9fa2566ca27d67c86569e8c85297aaf413ffab85a8960500f12ea34ff98e4c41"},
+    {file = "charset_normalizer-3.4.0-cp38-cp38-manylinux_2_17_s390x.manylinux2014_s390x.whl", hash = "sha256:a8e538f46104c815be19c975572d74afb53f29650ea2025bbfaef359d2de2f7f"},
+    {file = "charset_normalizer-3.4.0-cp38-cp38-manylinux_2_17_x86_64.manylinux2014_x86_64.whl", hash = "sha256:6fd30dc99682dc2c603c2b315bded2799019cea829f8bf57dc6b61efde6611c8"},
+    {file = "charset_normalizer-3.4.0-cp38-cp38-manylinux_2_5_i686.manylinux1_i686.manylinux_2_17_i686.manylinux2014_i686.whl", hash = "sha256:2006769bd1640bdf4d5641c69a3d63b71b81445473cac5ded39740a226fa88ab"},
+    {file = "charset_normalizer-3.4.0-cp38-cp38-musllinux_1_2_aarch64.whl", hash = "sha256:dc15e99b2d8a656f8e666854404f1ba54765871104e50c8e9813af8a7db07f12"},
+    {file = "charset_normalizer-3.4.0-cp38-cp38-musllinux_1_2_i686.whl", hash = "sha256:ab2e5bef076f5a235c3774b4f4028a680432cded7cad37bba0fd90d64b187d19"},
+    {file = "charset_normalizer-3.4.0-cp38-cp38-musllinux_1_2_ppc64le.whl", hash = "sha256:4ec9dd88a5b71abfc74e9df5ebe7921c35cbb3b641181a531ca65cdb5e8e4dea"},
+    {file = "charset_normalizer-3.4.0-cp38-cp38-musllinux_1_2_s390x.whl", hash = "sha256:43193c5cda5d612f247172016c4bb71251c784d7a4d9314677186a838ad34858"},
+    {file = "charset_normalizer-3.4.0-cp38-cp38-musllinux_1_2_x86_64.whl", hash = "sha256:aa693779a8b50cd97570e5a0f343538a8dbd3e496fa5dcb87e29406ad0299654"},
+    {file = "charset_normalizer-3.4.0-cp38-cp38-win32.whl", hash = "sha256:7706f5850360ac01d80c89bcef1640683cc12ed87f42579dab6c5d3ed6888613"},
+    {file = "charset_normalizer-3.4.0-cp38-cp38-win_amd64.whl", hash = "sha256:c3e446d253bd88f6377260d07c895816ebf33ffffd56c1c792b13bff9c3e1ade"},
+    {file = "charset_normalizer-3.4.0-cp39-cp39-macosx_10_9_universal2.whl", hash = "sha256:980b4f289d1d90ca5efcf07958d3eb38ed9c0b7676bf2831a54d4f66f9c27dfa"},
+    {file = "charset_normalizer-3.4.0-cp39-cp39-macosx_10_9_x86_64.whl", hash = "sha256:f28f891ccd15c514a0981f3b9db9aa23d62fe1a99997512b0491d2ed323d229a"},
+    {file = "charset_normalizer-3.4.0-cp39-cp39-macosx_11_0_arm64.whl", hash = "sha256:a8aacce6e2e1edcb6ac625fb0f8c3a9570ccc7bfba1f63419b3769ccf6a00ed0"},
+    {file = "charset_normalizer-3.4.0-cp39-cp39-manylinux_2_17_aarch64.manylinux2014_aarch64.whl", hash = "sha256:bd7af3717683bea4c87acd8c0d3d5b44d56120b26fd3f8a692bdd2d5260c620a"},
+    {file = "charset_normalizer-3.4.0-cp39-cp39-manylinux_2_17_ppc64le.manylinux2014_ppc64le.whl", hash = "sha256:5ff2ed8194587faf56555927b3aa10e6fb69d931e33953943bc4f837dfee2242"},
+    {file = "charset_normalizer-3.4.0-cp39-cp39-manylinux_2_17_s390x.manylinux2014_s390x.whl", hash = "sha256:e91f541a85298cf35433bf66f3fab2a4a2cff05c127eeca4af174f6d497f0d4b"},
+    {file = "charset_normalizer-3.4.0-cp39-cp39-manylinux_2_17_x86_64.manylinux2014_x86_64.whl", hash = "sha256:309a7de0a0ff3040acaebb35ec45d18db4b28232f21998851cfa709eeff49d62"},
+    {file = "charset_normalizer-3.4.0-cp39-cp39-manylinux_2_5_i686.manylinux1_i686.manylinux_2_17_i686.manylinux2014_i686.whl", hash = "sha256:285e96d9d53422efc0d7a17c60e59f37fbf3dfa942073f666db4ac71e8d726d0"},
+    {file = "charset_normalizer-3.4.0-cp39-cp39-musllinux_1_2_aarch64.whl", hash = "sha256:5d447056e2ca60382d460a604b6302d8db69476fd2015c81e7c35417cfabe4cd"},
+    {file = "charset_normalizer-3.4.0-cp39-cp39-musllinux_1_2_i686.whl", hash = "sha256:20587d20f557fe189b7947d8e7ec5afa110ccf72a3128d61a2a387c3313f46be"},
+    {file = "charset_normalizer-3.4.0-cp39-cp39-musllinux_1_2_ppc64le.whl", hash = "sha256:130272c698667a982a5d0e626851ceff662565379baf0ff2cc58067b81d4f11d"},
+    {file = "charset_normalizer-3.4.0-cp39-cp39-musllinux_1_2_s390x.whl", hash = "sha256:ab22fbd9765e6954bc0bcff24c25ff71dcbfdb185fcdaca49e81bac68fe724d3"},
+    {file = "charset_normalizer-3.4.0-cp39-cp39-musllinux_1_2_x86_64.whl", hash = "sha256:7782afc9b6b42200f7362858f9e73b1f8316afb276d316336c0ec3bd73312742"},
+    {file = "charset_normalizer-3.4.0-cp39-cp39-win32.whl", hash = "sha256:2de62e8801ddfff069cd5c504ce3bc9672b23266597d4e4f50eda28846c322f2"},
+    {file = "charset_normalizer-3.4.0-cp39-cp39-win_amd64.whl", hash = "sha256:95c3c157765b031331dd4db3c775e58deaee050a3042fcad72cbc4189d7c8dca"},
+    {file = "charset_normalizer-3.4.0-py3-none-any.whl", hash = "sha256:fe9f97feb71aa9896b81973a7bbada8c49501dc73e58a10fcef6663af95e5079"},
+    {file = "charset_normalizer-3.4.0.tar.gz", hash = "sha256:223217c3d4f82c3ac5e29032b3f1c2eb0fb591b72161f86d93f5719079dae93e"},
 ]
 
 [[package]]
@@ -1995,43 +2010,38 @@ files = [
 
 [[package]]
 name = "cryptography"
-version = "42.0.8"
+version = "43.0.1"
 description = "cryptography is a package which provides cryptographic recipes and primitives to Python developers."
 optional = false
 python-versions = ">=3.7"
 files = [
-    {file = "cryptography-42.0.8-cp37-abi3-macosx_10_12_universal2.whl", hash = "sha256:81d8a521705787afe7a18d5bfb47ea9d9cc068206270aad0b96a725022e18d2e"},
-    {file = "cryptography-42.0.8-cp37-abi3-macosx_10_12_x86_64.whl", hash = "sha256:961e61cefdcb06e0c6d7e3a1b22ebe8b996eb2bf50614e89384be54c48c6b63d"},
-    {file = "cryptography-42.0.8-cp37-abi3-manylinux_2_17_aarch64.manylinux2014_aarch64.whl", hash = "sha256:e3ec3672626e1b9e55afd0df6d774ff0e953452886e06e0f1eb7eb0c832e8902"},
-    {file = "cryptography-42.0.8-cp37-abi3-manylinux_2_17_x86_64.manylinux2014_x86_64.whl", hash = "sha256:e599b53fd95357d92304510fb7bda8523ed1f79ca98dce2f43c115950aa78801"},
-    {file = "cryptography-42.0.8-cp37-abi3-manylinux_2_28_aarch64.whl", hash = "sha256:5226d5d21ab681f432a9c1cf8b658c0cb02533eece706b155e5fbd8a0cdd3949"},
-    {file = "cryptography-42.0.8-cp37-abi3-manylinux_2_28_x86_64.whl", hash = "sha256:6b7c4f03ce01afd3b76cf69a5455caa9cfa3de8c8f493e0d3ab7d20611c8dae9"},
-    {file = "cryptography-42.0.8-cp37-abi3-musllinux_1_1_aarch64.whl", hash = "sha256:2346b911eb349ab547076f47f2e035fc8ff2c02380a7cbbf8d87114fa0f1c583"},
-    {file = "cryptography-42.0.8-cp37-abi3-musllinux_1_1_x86_64.whl", hash = "sha256:ad803773e9df0b92e0a817d22fd8a3675493f690b96130a5e24f1b8fabbea9c7"},
-    {file = "cryptography-42.0.8-cp37-abi3-musllinux_1_2_aarch64.whl", hash = "sha256:2f66d9cd9147ee495a8374a45ca445819f8929a3efcd2e3df6428e46c3cbb10b"},
-    {file = "cryptography-42.0.8-cp37-abi3-musllinux_1_2_x86_64.whl", hash = "sha256:d45b940883a03e19e944456a558b67a41160e367a719833c53de6911cabba2b7"},
-    {file = "cryptography-42.0.8-cp37-abi3-win32.whl", hash = "sha256:a0c5b2b0585b6af82d7e385f55a8bc568abff8923af147ee3c07bd8b42cda8b2"},
-    {file = "cryptography-42.0.8-cp37-abi3-win_amd64.whl", hash = "sha256:57080dee41209e556a9a4ce60d229244f7a66ef52750f813bfbe18959770cfba"},
-    {file = "cryptography-42.0.8-cp39-abi3-macosx_10_12_universal2.whl", hash = "sha256:dea567d1b0e8bc5764b9443858b673b734100c2871dc93163f58c46a97a83d28"},
-    {file = "cryptography-42.0.8-cp39-abi3-manylinux_2_17_aarch64.manylinux2014_aarch64.whl", hash = "sha256:c4783183f7cb757b73b2ae9aed6599b96338eb957233c58ca8f49a49cc32fd5e"},
-    {file = "cryptography-42.0.8-cp39-abi3-manylinux_2_17_x86_64.manylinux2014_x86_64.whl", hash = "sha256:a0608251135d0e03111152e41f0cc2392d1e74e35703960d4190b2e0f4ca9c70"},
-    {file = "cryptography-42.0.8-cp39-abi3-manylinux_2_28_aarch64.whl", hash = "sha256:dc0fdf6787f37b1c6b08e6dfc892d9d068b5bdb671198c72072828b80bd5fe4c"},
-    {file = "cryptography-42.0.8-cp39-abi3-manylinux_2_28_x86_64.whl", hash = "sha256:9c0c1716c8447ee7dbf08d6db2e5c41c688544c61074b54fc4564196f55c25a7"},
-    {file = "cryptography-42.0.8-cp39-abi3-musllinux_1_1_aarch64.whl", hash = "sha256:fff12c88a672ab9c9c1cf7b0c80e3ad9e2ebd9d828d955c126be4fd3e5578c9e"},
-    {file = "cryptography-42.0.8-cp39-abi3-musllinux_1_1_x86_64.whl", hash = "sha256:cafb92b2bc622cd1aa6a1dce4b93307792633f4c5fe1f46c6b97cf67073ec961"},
-    {file = "cryptography-42.0.8-cp39-abi3-musllinux_1_2_aarch64.whl", hash = "sha256:31f721658a29331f895a5a54e7e82075554ccfb8b163a18719d342f5ffe5ecb1"},
-    {file = "cryptography-42.0.8-cp39-abi3-musllinux_1_2_x86_64.whl", hash = "sha256:b297f90c5723d04bcc8265fc2a0f86d4ea2e0f7ab4b6994459548d3a6b992a14"},
-    {file = "cryptography-42.0.8-cp39-abi3-win32.whl", hash = "sha256:2f88d197e66c65be5e42cd72e5c18afbfae3f741742070e3019ac8f4ac57262c"},
-    {file = "cryptography-42.0.8-cp39-abi3-win_amd64.whl", hash = "sha256:fa76fbb7596cc5839320000cdd5d0955313696d9511debab7ee7278fc8b5c84a"},
-    {file = "cryptography-42.0.8-pp310-pypy310_pp73-macosx_10_12_x86_64.whl", hash = "sha256:ba4f0a211697362e89ad822e667d8d340b4d8d55fae72cdd619389fb5912eefe"},
-    {file = "cryptography-42.0.8-pp310-pypy310_pp73-manylinux_2_28_aarch64.whl", hash = "sha256:81884c4d096c272f00aeb1f11cf62ccd39763581645b0812e99a91505fa48e0c"},
-    {file = "cryptography-42.0.8-pp310-pypy310_pp73-manylinux_2_28_x86_64.whl", hash = "sha256:c9bb2ae11bfbab395bdd072985abde58ea9860ed84e59dbc0463a5d0159f5b71"},
-    {file = "cryptography-42.0.8-pp310-pypy310_pp73-win_amd64.whl", hash = "sha256:7016f837e15b0a1c119d27ecd89b3515f01f90a8615ed5e9427e30d9cdbfed3d"},
-    {file = "cryptography-42.0.8-pp39-pypy39_pp73-macosx_10_12_x86_64.whl", hash = "sha256:5a94eccb2a81a309806027e1670a358b99b8fe8bfe9f8d329f27d72c094dde8c"},
-    {file = "cryptography-42.0.8-pp39-pypy39_pp73-manylinux_2_28_aarch64.whl", hash = "sha256:dec9b018df185f08483f294cae6ccac29e7a6e0678996587363dc352dc65c842"},
-    {file = "cryptography-42.0.8-pp39-pypy39_pp73-manylinux_2_28_x86_64.whl", hash = "sha256:343728aac38decfdeecf55ecab3264b015be68fc2816ca800db649607aeee648"},
-    {file = "cryptography-42.0.8-pp39-pypy39_pp73-win_amd64.whl", hash = "sha256:013629ae70b40af70c9a7a5db40abe5d9054e6f4380e50ce769947b73bf3caad"},
-    {file = "cryptography-42.0.8.tar.gz", hash = "sha256:8d09d05439ce7baa8e9e95b07ec5b6c886f548deb7e0f69ef25f64b3bce842f2"},
+    {file = "cryptography-43.0.1-cp37-abi3-macosx_10_9_universal2.whl", hash = "sha256:8385d98f6a3bf8bb2d65a73e17ed87a3ba84f6991c155691c51112075f9ffc5d"},
+    {file = "cryptography-43.0.1-cp37-abi3-manylinux_2_17_aarch64.manylinux2014_aarch64.whl", hash = "sha256:27e613d7077ac613e399270253259d9d53872aaf657471473ebfc9a52935c062"},
+    {file = "cryptography-43.0.1-cp37-abi3-manylinux_2_17_x86_64.manylinux2014_x86_64.whl", hash = "sha256:68aaecc4178e90719e95298515979814bda0cbada1256a4485414860bd7ab962"},
+    {file = "cryptography-43.0.1-cp37-abi3-manylinux_2_28_aarch64.whl", hash = "sha256:de41fd81a41e53267cb020bb3a7212861da53a7d39f863585d13ea11049cf277"},
+    {file = "cryptography-43.0.1-cp37-abi3-manylinux_2_28_x86_64.whl", hash = "sha256:f98bf604c82c416bc829e490c700ca1553eafdf2912a91e23a79d97d9801372a"},
+    {file = "cryptography-43.0.1-cp37-abi3-musllinux_1_2_aarch64.whl", hash = "sha256:61ec41068b7b74268fa86e3e9e12b9f0c21fcf65434571dbb13d954bceb08042"},
+    {file = "cryptography-43.0.1-cp37-abi3-musllinux_1_2_x86_64.whl", hash = "sha256:014f58110f53237ace6a408b5beb6c427b64e084eb451ef25a28308270086494"},
+    {file = "cryptography-43.0.1-cp37-abi3-win32.whl", hash = "sha256:2bd51274dcd59f09dd952afb696bf9c61a7a49dfc764c04dd33ef7a6b502a1e2"},
+    {file = "cryptography-43.0.1-cp37-abi3-win_amd64.whl", hash = "sha256:666ae11966643886c2987b3b721899d250855718d6d9ce41b521252a17985f4d"},
+    {file = "cryptography-43.0.1-cp39-abi3-macosx_10_9_universal2.whl", hash = "sha256:ac119bb76b9faa00f48128b7f5679e1d8d437365c5d26f1c2c3f0da4ce1b553d"},
+    {file = "cryptography-43.0.1-cp39-abi3-manylinux_2_17_aarch64.manylinux2014_aarch64.whl", hash = "sha256:1bbcce1a551e262dfbafb6e6252f1ae36a248e615ca44ba302df077a846a8806"},
+    {file = "cryptography-43.0.1-cp39-abi3-manylinux_2_17_x86_64.manylinux2014_x86_64.whl", hash = "sha256:58d4e9129985185a06d849aa6df265bdd5a74ca6e1b736a77959b498e0505b85"},
+    {file = "cryptography-43.0.1-cp39-abi3-manylinux_2_28_aarch64.whl", hash = "sha256:d03a475165f3134f773d1388aeb19c2d25ba88b6a9733c5c590b9ff7bbfa2e0c"},
+    {file = "cryptography-43.0.1-cp39-abi3-manylinux_2_28_x86_64.whl", hash = "sha256:511f4273808ab590912a93ddb4e3914dfd8a388fed883361b02dea3791f292e1"},
+    {file = "cryptography-43.0.1-cp39-abi3-musllinux_1_2_aarch64.whl", hash = "sha256:80eda8b3e173f0f247f711eef62be51b599b5d425c429b5d4ca6a05e9e856baa"},
+    {file = "cryptography-43.0.1-cp39-abi3-musllinux_1_2_x86_64.whl", hash = "sha256:38926c50cff6f533f8a2dae3d7f19541432610d114a70808f0926d5aaa7121e4"},
+    {file = "cryptography-43.0.1-cp39-abi3-win32.whl", hash = "sha256:a575913fb06e05e6b4b814d7f7468c2c660e8bb16d8d5a1faf9b33ccc569dd47"},
+    {file = "cryptography-43.0.1-cp39-abi3-win_amd64.whl", hash = "sha256:d75601ad10b059ec832e78823b348bfa1a59f6b8d545db3a24fd44362a1564cb"},
+    {file = "cryptography-43.0.1-pp310-pypy310_pp73-macosx_10_9_x86_64.whl", hash = "sha256:ea25acb556320250756e53f9e20a4177515f012c9eaea17eb7587a8c4d8ae034"},
+    {file = "cryptography-43.0.1-pp310-pypy310_pp73-manylinux_2_28_aarch64.whl", hash = "sha256:c1332724be35d23a854994ff0b66530119500b6053d0bd3363265f7e5e77288d"},
+    {file = "cryptography-43.0.1-pp310-pypy310_pp73-manylinux_2_28_x86_64.whl", hash = "sha256:fba1007b3ef89946dbbb515aeeb41e30203b004f0b4b00e5e16078b518563289"},
+    {file = "cryptography-43.0.1-pp310-pypy310_pp73-win_amd64.whl", hash = "sha256:5b43d1ea6b378b54a1dc99dd8a2b5be47658fe9a7ce0a58ff0b55f4b43ef2b84"},
+    {file = "cryptography-43.0.1-pp39-pypy39_pp73-macosx_10_9_x86_64.whl", hash = "sha256:88cce104c36870d70c49c7c8fd22885875d950d9ee6ab54df2745f83ba0dc365"},
+    {file = "cryptography-43.0.1-pp39-pypy39_pp73-manylinux_2_28_aarch64.whl", hash = "sha256:9d3cdb25fa98afdd3d0892d132b8d7139e2c087da1712041f6b762e4f807cc96"},
+    {file = "cryptography-43.0.1-pp39-pypy39_pp73-manylinux_2_28_x86_64.whl", hash = "sha256:e710bf40870f4db63c3d7d929aa9e09e4e7ee219e703f949ec4073b4294f6172"},
+    {file = "cryptography-43.0.1-pp39-pypy39_pp73-win_amd64.whl", hash = "sha256:7c05650fe8023c5ed0d46793d4b7d7e6cd9c04e68eabe5b0aeea836e37bdcec2"},
+    {file = "cryptography-43.0.1.tar.gz", hash = "sha256:203e92a75716d8cfb491dc47c79e17d0d9207ccffcbcb35f598fbe463ae3444d"},
 ]
 
 [package.dependencies]
@@ -2044,7 +2054,7 @@ nox = ["nox"]
 pep8test = ["check-sdist", "click", "mypy", "ruff"]
 sdist = ["build"]
 ssh = ["bcrypt (>=3.1.5)"]
-test = ["certifi", "pretend", "pytest (>=6.2.0)", "pytest-benchmark", "pytest-cov", "pytest-xdist"]
+test = ["certifi", "cryptography-vectors (==43.0.1)", "pretend", "pytest (>=6.2.0)", "pytest-benchmark", "pytest-cov", "pytest-xdist"]
 test-randomorder = ["pytest-randomly"]
 
 [[package]]
@@ -2348,13 +2358,13 @@ files = [
 
 [[package]]
 name = "elastic-transport"
-version = "8.15.0"
+version = "8.15.1"
 description = "Transport classes and utilities shared among Python Elastic client libraries"
 optional = false
 python-versions = ">=3.8"
 files = [
-    {file = "elastic_transport-8.15.0-py3-none-any.whl", hash = "sha256:d7080d1dada2b4eee69e7574f9c17a76b42f2895eff428e562f94b0360e158c0"},
-    {file = "elastic_transport-8.15.0.tar.gz", hash = "sha256:85d62558f9baafb0868c801233a59b235e61d7b4804c28c2fadaa866b6766233"},
+    {file = "elastic_transport-8.15.1-py3-none-any.whl", hash = "sha256:b5e82ff1679d8c7705a03fd85c7f6ef85d6689721762d41228dd312e34f331fc"},
+    {file = "elastic_transport-8.15.1.tar.gz", hash = "sha256:9cac4ab5cf9402668cf305ae0b7d93ddc0c7b61461d6d1027850db6da9cc5742"},
 ]
 
 [package.dependencies]
@@ -2362,17 +2372,17 @@ certifi = "*"
 urllib3 = ">=1.26.2,<3"
 
 [package.extras]
-develop = ["aiohttp", "furo", "httpx", "opentelemetry-api", "opentelemetry-sdk", "orjson", "pytest", "pytest-asyncio", "pytest-cov", "pytest-httpserver", "pytest-mock", "requests", "respx", "sphinx (>2)", "sphinx-autodoc-typehints", "trustme"]
+develop = ["aiohttp", "furo", "httpcore (<1.0.6)", "httpx", "opentelemetry-api", "opentelemetry-sdk", "orjson", "pytest", "pytest-asyncio", "pytest-cov", "pytest-httpserver", "pytest-mock", "requests", "respx", "sphinx (>2)", "sphinx-autodoc-typehints", "trustme"]
 
 [[package]]
 name = "elasticsearch"
-version = "8.15.1"
+version = "8.14.0"
 description = "Python client for Elasticsearch"
 optional = false
-python-versions = ">=3.8"
+python-versions = ">=3.7"
 files = [
-    {file = "elasticsearch-8.15.1-py3-none-any.whl", hash = "sha256:02a0476e98768a30d7926335fc0d305c04fdb928eea1354c6e6040d8c2814569"},
-    {file = "elasticsearch-8.15.1.tar.gz", hash = "sha256:40c0d312f8adf8bdc81795bc16a0b546ddf544cb1f90e829a244e4780c4dbfd8"},
+    {file = "elasticsearch-8.14.0-py3-none-any.whl", hash = "sha256:cef8ef70a81af027f3da74a4f7d9296b390c636903088439087b8262a468c130"},
+    {file = "elasticsearch-8.14.0.tar.gz", hash = "sha256:aa2490029dd96f4015b333c1827aa21fd6c0a4d223b00dfb0fe933b8d09a511b"},
 ]
 
 [package.dependencies]
@@ -2380,10 +2390,7 @@ elastic-transport = ">=8.13,<9"
 
 [package.extras]
 async = ["aiohttp (>=3,<4)"]
-dev = ["aiohttp", "black", "build", "coverage", "isort", "jinja2", "mapbox-vector-tile", "nox", "numpy", "orjson", "pandas", "pyarrow", "pytest", "pytest-asyncio", "pytest-cov", "python-dateutil", "pyyaml (>=5.4)", "requests (>=2,<3)", "simsimd", "twine", "unasync"]
-docs = ["sphinx", "sphinx-autodoc-typehints", "sphinx-rtd-theme (>=2.0)"]
 orjson = ["orjson (>=3)"]
-pyarrow = ["pyarrow (>=1)"]
 requests = ["requests (>=2.4.0,!=2.32.2,<3.0.0)"]
 vectorstore-mmr = ["numpy (>=1)", "simsimd (>=3)"]
 
@@ -4124,18 +4131,15 @@ files = [
 
 [[package]]
 name = "isodate"
-version = "0.6.1"
+version = "0.7.2"
 description = "An ISO 8601 date/time/duration parser and formatter"
 optional = false
-python-versions = "*"
+python-versions = ">=3.7"
 files = [
-    {file = "isodate-0.6.1-py2.py3-none-any.whl", hash = "sha256:0751eece944162659049d35f4f549ed815792b38793f07cf73381c1c87cbed96"},
-    {file = "isodate-0.6.1.tar.gz", hash = "sha256:48c5881de7e8b0a0d648cb024c8062dc84e7b840ed81e864c7614fd3c127bde9"},
+    {file = "isodate-0.7.2-py3-none-any.whl", hash = "sha256:28009937d8031054830160fce6d409ed342816b543597cece116d966c6d99e15"},
+    {file = "isodate-0.7.2.tar.gz", hash = "sha256:4cd1aa0f43ca76f4a6c6c0292a85f40b35ec2e43e315b59f06e6d32171a953e6"},
 ]
 
-[package.dependencies]
-six = "*"
-
 [[package]]
 name = "itsdangerous"
 version = "2.2.0"
@@ -4257,13 +4261,13 @@ format-nongpl = ["fqdn", "idna", "isoduration", "jsonpointer (>1.13)", "rfc3339-
 
 [[package]]
 name = "jsonschema-specifications"
-version = "2023.12.1"
+version = "2024.10.1"
 description = "The JSON Schema meta-schemas and vocabularies, exposed as a Registry"
 optional = false
-python-versions = ">=3.8"
+python-versions = ">=3.9"
 files = [
-    {file = "jsonschema_specifications-2023.12.1-py3-none-any.whl", hash = "sha256:87e4fdf3a94858b8a2ba2778d9ba57d8a9cafca7c7489c46ba0d30a8bc6a9c3c"},
-    {file = "jsonschema_specifications-2023.12.1.tar.gz", hash = "sha256:48a76787b3e70f5ed53f1160d2b81f586e4ca6d1548c5de7085d1682674764cc"},
+    {file = "jsonschema_specifications-2024.10.1-py3-none-any.whl", hash = "sha256:a09a0680616357d9a0ecf05c12ad234479f549239d0f5b55f3deea67475da9bf"},
+    {file = "jsonschema_specifications-2024.10.1.tar.gz", hash = "sha256:0f38b83639958ce1152d02a7f062902c41c8fd20d558b0c34344292d417ae272"},
 ]
 
 [package.dependencies]
@@ -4483,13 +4487,13 @@ six = "*"
 
 [[package]]
 name = "langfuse"
-version = "2.51.3"
+version = "2.51.5"
 description = "A client library for accessing langfuse"
 optional = false
 python-versions = "<4.0,>=3.8.1"
 files = [
-    {file = "langfuse-2.51.3-py3-none-any.whl", hash = "sha256:32aba050123656ec0c165583e1d33243cc7a14e7b8498ed3c9de808aa90306f1"},
-    {file = "langfuse-2.51.3.tar.gz", hash = "sha256:ccd2109556ee232db717abfb751ee8a3139f074db7d3e06a4f1a756c349fc5ba"},
+    {file = "langfuse-2.51.5-py3-none-any.whl", hash = "sha256:b95401ca710ef94b521afa6541933b6f93d7cfd4a97523c8fc75bca4d6d219fb"},
+    {file = "langfuse-2.51.5.tar.gz", hash = "sha256:55bc37b5c5d3ae133c1a95db09117cfb3117add110ba02ebbf2ce45ac4395c5b"},
 ]
 
 [package.dependencies]
@@ -4508,13 +4512,13 @@ openai = ["openai (>=0.27.8)"]
 
 [[package]]
 name = "langsmith"
-version = "0.1.131"
+version = "0.1.133"
 description = "Client library to connect to the LangSmith LLM Tracing and Evaluation Platform."
 optional = false
 python-versions = "<4.0,>=3.8.1"
 files = [
-    {file = "langsmith-0.1.131-py3-none-any.whl", hash = "sha256:80c106b1c42307195cc0bb3a596472c41ef91b79d15bcee9938307800336c563"},
-    {file = "langsmith-0.1.131.tar.gz", hash = "sha256:626101a3bf3ca481e5110d5155ace8aa066e4e9cc2fa7d96c8290ade0fbff797"},
+    {file = "langsmith-0.1.133-py3-none-any.whl", hash = "sha256:82e837a6039c483beadbe19c2ba7ebafbd402d3e8105234f5ef334425cff7b45"},
+    {file = "langsmith-0.1.133.tar.gz", hash = "sha256:7bfd8bef166b9a64ee540a11bee4aa7bf43b1d9229f95b0fc19086454955185d"},
 ]
 
 [package.dependencies]
@@ -4856,71 +4860,72 @@ testing = ["coverage", "pytest", "pytest-cov", "pytest-regressions"]
 
 [[package]]
 name = "markupsafe"
-version = "2.1.5"
+version = "3.0.1"
 description = "Safely add untrusted strings to HTML/XML markup."
 optional = false
-python-versions = ">=3.7"
+python-versions = ">=3.9"
 files = [
-    {file = "MarkupSafe-2.1.5-cp310-cp310-macosx_10_9_universal2.whl", hash = "sha256:a17a92de5231666cfbe003f0e4b9b3a7ae3afb1ec2845aadc2bacc93ff85febc"},
-    {file = "MarkupSafe-2.1.5-cp310-cp310-macosx_10_9_x86_64.whl", hash = "sha256:72b6be590cc35924b02c78ef34b467da4ba07e4e0f0454a2c5907f473fc50ce5"},
-    {file = "MarkupSafe-2.1.5-cp310-cp310-manylinux_2_17_aarch64.manylinux2014_aarch64.whl", hash = "sha256:e61659ba32cf2cf1481e575d0462554625196a1f2fc06a1c777d3f48e8865d46"},
-    {file = "MarkupSafe-2.1.5-cp310-cp310-manylinux_2_17_x86_64.manylinux2014_x86_64.whl", hash = "sha256:2174c595a0d73a3080ca3257b40096db99799265e1c27cc5a610743acd86d62f"},
-    {file = "MarkupSafe-2.1.5-cp310-cp310-manylinux_2_5_i686.manylinux1_i686.manylinux_2_17_i686.manylinux2014_i686.whl", hash = "sha256:ae2ad8ae6ebee9d2d94b17fb62763125f3f374c25618198f40cbb8b525411900"},
-    {file = "MarkupSafe-2.1.5-cp310-cp310-musllinux_1_1_aarch64.whl", hash = "sha256:075202fa5b72c86ad32dc7d0b56024ebdbcf2048c0ba09f1cde31bfdd57bcfff"},
-    {file = "MarkupSafe-2.1.5-cp310-cp310-musllinux_1_1_i686.whl", hash = "sha256:598e3276b64aff0e7b3451b72e94fa3c238d452e7ddcd893c3ab324717456bad"},
-    {file = "MarkupSafe-2.1.5-cp310-cp310-musllinux_1_1_x86_64.whl", hash = "sha256:fce659a462a1be54d2ffcacea5e3ba2d74daa74f30f5f143fe0c58636e355fdd"},
-    {file = "MarkupSafe-2.1.5-cp310-cp310-win32.whl", hash = "sha256:d9fad5155d72433c921b782e58892377c44bd6252b5af2f67f16b194987338a4"},
-    {file = "MarkupSafe-2.1.5-cp310-cp310-win_amd64.whl", hash = "sha256:bf50cd79a75d181c9181df03572cdce0fbb75cc353bc350712073108cba98de5"},
-    {file = "MarkupSafe-2.1.5-cp311-cp311-macosx_10_9_universal2.whl", hash = "sha256:629ddd2ca402ae6dbedfceeba9c46d5f7b2a61d9749597d4307f943ef198fc1f"},
-    {file = "MarkupSafe-2.1.5-cp311-cp311-macosx_10_9_x86_64.whl", hash = "sha256:5b7b716f97b52c5a14bffdf688f971b2d5ef4029127f1ad7a513973cfd818df2"},
-    {file = "MarkupSafe-2.1.5-cp311-cp311-manylinux_2_17_aarch64.manylinux2014_aarch64.whl", hash = "sha256:6ec585f69cec0aa07d945b20805be741395e28ac1627333b1c5b0105962ffced"},
-    {file = "MarkupSafe-2.1.5-cp311-cp311-manylinux_2_17_x86_64.manylinux2014_x86_64.whl", hash = "sha256:b91c037585eba9095565a3556f611e3cbfaa42ca1e865f7b8015fe5c7336d5a5"},
-    {file = "MarkupSafe-2.1.5-cp311-cp311-manylinux_2_5_i686.manylinux1_i686.manylinux_2_17_i686.manylinux2014_i686.whl", hash = "sha256:7502934a33b54030eaf1194c21c692a534196063db72176b0c4028e140f8f32c"},
-    {file = "MarkupSafe-2.1.5-cp311-cp311-musllinux_1_1_aarch64.whl", hash = "sha256:0e397ac966fdf721b2c528cf028494e86172b4feba51d65f81ffd65c63798f3f"},
-    {file = "MarkupSafe-2.1.5-cp311-cp311-musllinux_1_1_i686.whl", hash = "sha256:c061bb86a71b42465156a3ee7bd58c8c2ceacdbeb95d05a99893e08b8467359a"},
-    {file = "MarkupSafe-2.1.5-cp311-cp311-musllinux_1_1_x86_64.whl", hash = "sha256:3a57fdd7ce31c7ff06cdfbf31dafa96cc533c21e443d57f5b1ecc6cdc668ec7f"},
-    {file = "MarkupSafe-2.1.5-cp311-cp311-win32.whl", hash = "sha256:397081c1a0bfb5124355710fe79478cdbeb39626492b15d399526ae53422b906"},
-    {file = "MarkupSafe-2.1.5-cp311-cp311-win_amd64.whl", hash = "sha256:2b7c57a4dfc4f16f7142221afe5ba4e093e09e728ca65c51f5620c9aaeb9a617"},
-    {file = "MarkupSafe-2.1.5-cp312-cp312-macosx_10_9_universal2.whl", hash = "sha256:8dec4936e9c3100156f8a2dc89c4b88d5c435175ff03413b443469c7c8c5f4d1"},
-    {file = "MarkupSafe-2.1.5-cp312-cp312-macosx_10_9_x86_64.whl", hash = "sha256:3c6b973f22eb18a789b1460b4b91bf04ae3f0c4234a0a6aa6b0a92f6f7b951d4"},
-    {file = "MarkupSafe-2.1.5-cp312-cp312-manylinux_2_17_aarch64.manylinux2014_aarch64.whl", hash = "sha256:ac07bad82163452a6884fe8fa0963fb98c2346ba78d779ec06bd7a6262132aee"},
-    {file = "MarkupSafe-2.1.5-cp312-cp312-manylinux_2_17_x86_64.manylinux2014_x86_64.whl", hash = "sha256:f5dfb42c4604dddc8e4305050aa6deb084540643ed5804d7455b5df8fe16f5e5"},
-    {file = "MarkupSafe-2.1.5-cp312-cp312-manylinux_2_5_i686.manylinux1_i686.manylinux_2_17_i686.manylinux2014_i686.whl", hash = "sha256:ea3d8a3d18833cf4304cd2fc9cbb1efe188ca9b5efef2bdac7adc20594a0e46b"},
-    {file = "MarkupSafe-2.1.5-cp312-cp312-musllinux_1_1_aarch64.whl", hash = "sha256:d050b3361367a06d752db6ead6e7edeb0009be66bc3bae0ee9d97fb326badc2a"},
-    {file = "MarkupSafe-2.1.5-cp312-cp312-musllinux_1_1_i686.whl", hash = "sha256:bec0a414d016ac1a18862a519e54b2fd0fc8bbfd6890376898a6c0891dd82e9f"},
-    {file = "MarkupSafe-2.1.5-cp312-cp312-musllinux_1_1_x86_64.whl", hash = "sha256:58c98fee265677f63a4385256a6d7683ab1832f3ddd1e66fe948d5880c21a169"},
-    {file = "MarkupSafe-2.1.5-cp312-cp312-win32.whl", hash = "sha256:8590b4ae07a35970728874632fed7bd57b26b0102df2d2b233b6d9d82f6c62ad"},
-    {file = "MarkupSafe-2.1.5-cp312-cp312-win_amd64.whl", hash = "sha256:823b65d8706e32ad2df51ed89496147a42a2a6e01c13cfb6ffb8b1e92bc910bb"},
-    {file = "MarkupSafe-2.1.5-cp37-cp37m-macosx_10_9_x86_64.whl", hash = "sha256:c8b29db45f8fe46ad280a7294f5c3ec36dbac9491f2d1c17345be8e69cc5928f"},
-    {file = "MarkupSafe-2.1.5-cp37-cp37m-manylinux_2_17_aarch64.manylinux2014_aarch64.whl", hash = "sha256:ec6a563cff360b50eed26f13adc43e61bc0c04d94b8be985e6fb24b81f6dcfdf"},
-    {file = "MarkupSafe-2.1.5-cp37-cp37m-manylinux_2_17_x86_64.manylinux2014_x86_64.whl", hash = "sha256:a549b9c31bec33820e885335b451286e2969a2d9e24879f83fe904a5ce59d70a"},
-    {file = "MarkupSafe-2.1.5-cp37-cp37m-manylinux_2_5_i686.manylinux1_i686.manylinux_2_17_i686.manylinux2014_i686.whl", hash = "sha256:4f11aa001c540f62c6166c7726f71f7573b52c68c31f014c25cc7901deea0b52"},
-    {file = "MarkupSafe-2.1.5-cp37-cp37m-musllinux_1_1_aarch64.whl", hash = "sha256:7b2e5a267c855eea6b4283940daa6e88a285f5f2a67f2220203786dfa59b37e9"},
-    {file = "MarkupSafe-2.1.5-cp37-cp37m-musllinux_1_1_i686.whl", hash = "sha256:2d2d793e36e230fd32babe143b04cec8a8b3eb8a3122d2aceb4a371e6b09b8df"},
-    {file = "MarkupSafe-2.1.5-cp37-cp37m-musllinux_1_1_x86_64.whl", hash = "sha256:ce409136744f6521e39fd8e2a24c53fa18ad67aa5bc7c2cf83645cce5b5c4e50"},
-    {file = "MarkupSafe-2.1.5-cp37-cp37m-win32.whl", hash = "sha256:4096e9de5c6fdf43fb4f04c26fb114f61ef0bf2e5604b6ee3019d51b69e8c371"},
-    {file = "MarkupSafe-2.1.5-cp37-cp37m-win_amd64.whl", hash = "sha256:4275d846e41ecefa46e2015117a9f491e57a71ddd59bbead77e904dc02b1bed2"},
-    {file = "MarkupSafe-2.1.5-cp38-cp38-macosx_10_9_universal2.whl", hash = "sha256:656f7526c69fac7f600bd1f400991cc282b417d17539a1b228617081106feb4a"},
-    {file = "MarkupSafe-2.1.5-cp38-cp38-macosx_10_9_x86_64.whl", hash = "sha256:97cafb1f3cbcd3fd2b6fbfb99ae11cdb14deea0736fc2b0952ee177f2b813a46"},
-    {file = "MarkupSafe-2.1.5-cp38-cp38-manylinux_2_17_aarch64.manylinux2014_aarch64.whl", hash = "sha256:1f3fbcb7ef1f16e48246f704ab79d79da8a46891e2da03f8783a5b6fa41a9532"},
-    {file = "MarkupSafe-2.1.5-cp38-cp38-manylinux_2_17_x86_64.manylinux2014_x86_64.whl", hash = "sha256:fa9db3f79de01457b03d4f01b34cf91bc0048eb2c3846ff26f66687c2f6d16ab"},
-    {file = "MarkupSafe-2.1.5-cp38-cp38-manylinux_2_5_i686.manylinux1_i686.manylinux_2_17_i686.manylinux2014_i686.whl", hash = "sha256:ffee1f21e5ef0d712f9033568f8344d5da8cc2869dbd08d87c84656e6a2d2f68"},
-    {file = "MarkupSafe-2.1.5-cp38-cp38-musllinux_1_1_aarch64.whl", hash = "sha256:5dedb4db619ba5a2787a94d877bc8ffc0566f92a01c0ef214865e54ecc9ee5e0"},
-    {file = "MarkupSafe-2.1.5-cp38-cp38-musllinux_1_1_i686.whl", hash = "sha256:30b600cf0a7ac9234b2638fbc0fb6158ba5bdcdf46aeb631ead21248b9affbc4"},
-    {file = "MarkupSafe-2.1.5-cp38-cp38-musllinux_1_1_x86_64.whl", hash = "sha256:8dd717634f5a044f860435c1d8c16a270ddf0ef8588d4887037c5028b859b0c3"},
-    {file = "MarkupSafe-2.1.5-cp38-cp38-win32.whl", hash = "sha256:daa4ee5a243f0f20d528d939d06670a298dd39b1ad5f8a72a4275124a7819eff"},
-    {file = "MarkupSafe-2.1.5-cp38-cp38-win_amd64.whl", hash = "sha256:619bc166c4f2de5caa5a633b8b7326fbe98e0ccbfacabd87268a2b15ff73a029"},
-    {file = "MarkupSafe-2.1.5-cp39-cp39-macosx_10_9_universal2.whl", hash = "sha256:7a68b554d356a91cce1236aa7682dc01df0edba8d043fd1ce607c49dd3c1edcf"},
-    {file = "MarkupSafe-2.1.5-cp39-cp39-macosx_10_9_x86_64.whl", hash = "sha256:db0b55e0f3cc0be60c1f19efdde9a637c32740486004f20d1cff53c3c0ece4d2"},
-    {file = "MarkupSafe-2.1.5-cp39-cp39-manylinux_2_17_aarch64.manylinux2014_aarch64.whl", hash = "sha256:3e53af139f8579a6d5f7b76549125f0d94d7e630761a2111bc431fd820e163b8"},
-    {file = "MarkupSafe-2.1.5-cp39-cp39-manylinux_2_17_x86_64.manylinux2014_x86_64.whl", hash = "sha256:17b950fccb810b3293638215058e432159d2b71005c74371d784862b7e4683f3"},
-    {file = "MarkupSafe-2.1.5-cp39-cp39-manylinux_2_5_i686.manylinux1_i686.manylinux_2_17_i686.manylinux2014_i686.whl", hash = "sha256:4c31f53cdae6ecfa91a77820e8b151dba54ab528ba65dfd235c80b086d68a465"},
-    {file = "MarkupSafe-2.1.5-cp39-cp39-musllinux_1_1_aarch64.whl", hash = "sha256:bff1b4290a66b490a2f4719358c0cdcd9bafb6b8f061e45c7a2460866bf50c2e"},
-    {file = "MarkupSafe-2.1.5-cp39-cp39-musllinux_1_1_i686.whl", hash = "sha256:bc1667f8b83f48511b94671e0e441401371dfd0f0a795c7daa4a3cd1dde55bea"},
-    {file = "MarkupSafe-2.1.5-cp39-cp39-musllinux_1_1_x86_64.whl", hash = "sha256:5049256f536511ee3f7e1b3f87d1d1209d327e818e6ae1365e8653d7e3abb6a6"},
-    {file = "MarkupSafe-2.1.5-cp39-cp39-win32.whl", hash = "sha256:00e046b6dd71aa03a41079792f8473dc494d564611a8f89bbbd7cb93295ebdcf"},
-    {file = "MarkupSafe-2.1.5-cp39-cp39-win_amd64.whl", hash = "sha256:fa173ec60341d6bb97a89f5ea19c85c5643c1e7dedebc22f5181eb73573142c5"},
-    {file = "MarkupSafe-2.1.5.tar.gz", hash = "sha256:d283d37a890ba4c1ae73ffadf8046435c76e7bc2247bbb63c00bd1a709c6544b"},
+    {file = "MarkupSafe-3.0.1-cp310-cp310-macosx_10_9_universal2.whl", hash = "sha256:db842712984e91707437461930e6011e60b39136c7331e971952bb30465bc1a1"},
+    {file = "MarkupSafe-3.0.1-cp310-cp310-macosx_11_0_arm64.whl", hash = "sha256:3ffb4a8e7d46ed96ae48805746755fadd0909fea2306f93d5d8233ba23dda12a"},
+    {file = "MarkupSafe-3.0.1-cp310-cp310-manylinux_2_17_aarch64.manylinux2014_aarch64.whl", hash = "sha256:67c519635a4f64e495c50e3107d9b4075aec33634272b5db1cde839e07367589"},
+    {file = "MarkupSafe-3.0.1-cp310-cp310-manylinux_2_17_x86_64.manylinux2014_x86_64.whl", hash = "sha256:48488d999ed50ba8d38c581d67e496f955821dc183883550a6fbc7f1aefdc170"},
+    {file = "MarkupSafe-3.0.1-cp310-cp310-manylinux_2_5_i686.manylinux1_i686.manylinux_2_17_i686.manylinux2014_i686.whl", hash = "sha256:f31ae06f1328595d762c9a2bf29dafd8621c7d3adc130cbb46278079758779ca"},
+    {file = "MarkupSafe-3.0.1-cp310-cp310-musllinux_1_2_aarch64.whl", hash = "sha256:80fcbf3add8790caddfab6764bde258b5d09aefbe9169c183f88a7410f0f6dea"},
+    {file = "MarkupSafe-3.0.1-cp310-cp310-musllinux_1_2_i686.whl", hash = "sha256:3341c043c37d78cc5ae6e3e305e988532b072329639007fd408a476642a89fd6"},
+    {file = "MarkupSafe-3.0.1-cp310-cp310-musllinux_1_2_x86_64.whl", hash = "sha256:cb53e2a99df28eee3b5f4fea166020d3ef9116fdc5764bc5117486e6d1211b25"},
+    {file = "MarkupSafe-3.0.1-cp310-cp310-win32.whl", hash = "sha256:db15ce28e1e127a0013dfb8ac243a8e392db8c61eae113337536edb28bdc1f97"},
+    {file = "MarkupSafe-3.0.1-cp310-cp310-win_amd64.whl", hash = "sha256:4ffaaac913c3f7345579db4f33b0020db693f302ca5137f106060316761beea9"},
+    {file = "MarkupSafe-3.0.1-cp311-cp311-macosx_10_9_universal2.whl", hash = "sha256:26627785a54a947f6d7336ce5963569b5d75614619e75193bdb4e06e21d447ad"},
+    {file = "MarkupSafe-3.0.1-cp311-cp311-macosx_11_0_arm64.whl", hash = "sha256:b954093679d5750495725ea6f88409946d69cfb25ea7b4c846eef5044194f583"},
+    {file = "MarkupSafe-3.0.1-cp311-cp311-manylinux_2_17_aarch64.manylinux2014_aarch64.whl", hash = "sha256:973a371a55ce9ed333a3a0f8e0bcfae9e0d637711534bcb11e130af2ab9334e7"},
+    {file = "MarkupSafe-3.0.1-cp311-cp311-manylinux_2_17_x86_64.manylinux2014_x86_64.whl", hash = "sha256:244dbe463d5fb6d7ce161301a03a6fe744dac9072328ba9fc82289238582697b"},
+    {file = "MarkupSafe-3.0.1-cp311-cp311-manylinux_2_5_i686.manylinux1_i686.manylinux_2_17_i686.manylinux2014_i686.whl", hash = "sha256:d98e66a24497637dd31ccab090b34392dddb1f2f811c4b4cd80c230205c074a3"},
+    {file = "MarkupSafe-3.0.1-cp311-cp311-musllinux_1_2_aarch64.whl", hash = "sha256:ad91738f14eb8da0ff82f2acd0098b6257621410dcbd4df20aaa5b4233d75a50"},
+    {file = "MarkupSafe-3.0.1-cp311-cp311-musllinux_1_2_i686.whl", hash = "sha256:7044312a928a66a4c2a22644147bc61a199c1709712069a344a3fb5cfcf16915"},
+    {file = "MarkupSafe-3.0.1-cp311-cp311-musllinux_1_2_x86_64.whl", hash = "sha256:a4792d3b3a6dfafefdf8e937f14906a51bd27025a36f4b188728a73382231d91"},
+    {file = "MarkupSafe-3.0.1-cp311-cp311-win32.whl", hash = "sha256:fa7d686ed9883f3d664d39d5a8e74d3c5f63e603c2e3ff0abcba23eac6542635"},
+    {file = "MarkupSafe-3.0.1-cp311-cp311-win_amd64.whl", hash = "sha256:9ba25a71ebf05b9bb0e2ae99f8bc08a07ee8e98c612175087112656ca0f5c8bf"},
+    {file = "MarkupSafe-3.0.1-cp312-cp312-macosx_10_13_universal2.whl", hash = "sha256:8ae369e84466aa70f3154ee23c1451fda10a8ee1b63923ce76667e3077f2b0c4"},
+    {file = "MarkupSafe-3.0.1-cp312-cp312-macosx_11_0_arm64.whl", hash = "sha256:40f1e10d51c92859765522cbd79c5c8989f40f0419614bcdc5015e7b6bf97fc5"},
+    {file = "MarkupSafe-3.0.1-cp312-cp312-manylinux_2_17_aarch64.manylinux2014_aarch64.whl", hash = "sha256:5a4cb365cb49b750bdb60b846b0c0bc49ed62e59a76635095a179d440540c346"},
+    {file = "MarkupSafe-3.0.1-cp312-cp312-manylinux_2_17_x86_64.manylinux2014_x86_64.whl", hash = "sha256:ee3941769bd2522fe39222206f6dd97ae83c442a94c90f2b7a25d847d40f4729"},
+    {file = "MarkupSafe-3.0.1-cp312-cp312-manylinux_2_5_i686.manylinux1_i686.manylinux_2_17_i686.manylinux2014_i686.whl", hash = "sha256:62fada2c942702ef8952754abfc1a9f7658a4d5460fabe95ac7ec2cbe0d02abc"},
+    {file = "MarkupSafe-3.0.1-cp312-cp312-musllinux_1_2_aarch64.whl", hash = "sha256:4c2d64fdba74ad16138300815cfdc6ab2f4647e23ced81f59e940d7d4a1469d9"},
+    {file = "MarkupSafe-3.0.1-cp312-cp312-musllinux_1_2_i686.whl", hash = "sha256:fb532dd9900381d2e8f48172ddc5a59db4c445a11b9fab40b3b786da40d3b56b"},
+    {file = "MarkupSafe-3.0.1-cp312-cp312-musllinux_1_2_x86_64.whl", hash = "sha256:0f84af7e813784feb4d5e4ff7db633aba6c8ca64a833f61d8e4eade234ef0c38"},
+    {file = "MarkupSafe-3.0.1-cp312-cp312-win32.whl", hash = "sha256:cbf445eb5628981a80f54087f9acdbf84f9b7d862756110d172993b9a5ae81aa"},
+    {file = "MarkupSafe-3.0.1-cp312-cp312-win_amd64.whl", hash = "sha256:a10860e00ded1dd0a65b83e717af28845bb7bd16d8ace40fe5531491de76b79f"},
+    {file = "MarkupSafe-3.0.1-cp313-cp313-macosx_10_13_universal2.whl", hash = "sha256:e81c52638315ff4ac1b533d427f50bc0afc746deb949210bc85f05d4f15fd772"},
+    {file = "MarkupSafe-3.0.1-cp313-cp313-macosx_11_0_arm64.whl", hash = "sha256:312387403cd40699ab91d50735ea7a507b788091c416dd007eac54434aee51da"},
+    {file = "MarkupSafe-3.0.1-cp313-cp313-manylinux_2_17_aarch64.manylinux2014_aarch64.whl", hash = "sha256:2ae99f31f47d849758a687102afdd05bd3d3ff7dbab0a8f1587981b58a76152a"},
+    {file = "MarkupSafe-3.0.1-cp313-cp313-manylinux_2_17_x86_64.manylinux2014_x86_64.whl", hash = "sha256:c97ff7fedf56d86bae92fa0a646ce1a0ec7509a7578e1ed238731ba13aabcd1c"},
+    {file = "MarkupSafe-3.0.1-cp313-cp313-manylinux_2_5_i686.manylinux1_i686.manylinux_2_17_i686.manylinux2014_i686.whl", hash = "sha256:a7420ceda262dbb4b8d839a4ec63d61c261e4e77677ed7c66c99f4e7cb5030dd"},
+    {file = "MarkupSafe-3.0.1-cp313-cp313-musllinux_1_2_aarch64.whl", hash = "sha256:45d42d132cff577c92bfba536aefcfea7e26efb975bd455db4e6602f5c9f45e7"},
+    {file = "MarkupSafe-3.0.1-cp313-cp313-musllinux_1_2_i686.whl", hash = "sha256:4c8817557d0de9349109acb38b9dd570b03cc5014e8aabf1cbddc6e81005becd"},
+    {file = "MarkupSafe-3.0.1-cp313-cp313-musllinux_1_2_x86_64.whl", hash = "sha256:6a54c43d3ec4cf2a39f4387ad044221c66a376e58c0d0e971d47c475ba79c6b5"},
+    {file = "MarkupSafe-3.0.1-cp313-cp313-win32.whl", hash = "sha256:c91b394f7601438ff79a4b93d16be92f216adb57d813a78be4446fe0f6bc2d8c"},
+    {file = "MarkupSafe-3.0.1-cp313-cp313-win_amd64.whl", hash = "sha256:fe32482b37b4b00c7a52a07211b479653b7fe4f22b2e481b9a9b099d8a430f2f"},
+    {file = "MarkupSafe-3.0.1-cp313-cp313t-macosx_10_13_universal2.whl", hash = "sha256:17b2aea42a7280db02ac644db1d634ad47dcc96faf38ab304fe26ba2680d359a"},
+    {file = "MarkupSafe-3.0.1-cp313-cp313t-macosx_11_0_arm64.whl", hash = "sha256:852dc840f6d7c985603e60b5deaae1d89c56cb038b577f6b5b8c808c97580f1d"},
+    {file = "MarkupSafe-3.0.1-cp313-cp313t-manylinux_2_17_aarch64.manylinux2014_aarch64.whl", hash = "sha256:0778de17cff1acaeccc3ff30cd99a3fd5c50fc58ad3d6c0e0c4c58092b859396"},
+    {file = "MarkupSafe-3.0.1-cp313-cp313t-manylinux_2_17_x86_64.manylinux2014_x86_64.whl", hash = "sha256:800100d45176652ded796134277ecb13640c1a537cad3b8b53da45aa96330453"},
+    {file = "MarkupSafe-3.0.1-cp313-cp313t-manylinux_2_5_i686.manylinux1_i686.manylinux_2_17_i686.manylinux2014_i686.whl", hash = "sha256:d06b24c686a34c86c8c1fba923181eae6b10565e4d80bdd7bc1c8e2f11247aa4"},
+    {file = "MarkupSafe-3.0.1-cp313-cp313t-musllinux_1_2_aarch64.whl", hash = "sha256:33d1c36b90e570ba7785dacd1faaf091203d9942bc036118fab8110a401eb1a8"},
+    {file = "MarkupSafe-3.0.1-cp313-cp313t-musllinux_1_2_i686.whl", hash = "sha256:beeebf760a9c1f4c07ef6a53465e8cfa776ea6a2021eda0d0417ec41043fe984"},
+    {file = "MarkupSafe-3.0.1-cp313-cp313t-musllinux_1_2_x86_64.whl", hash = "sha256:bbde71a705f8e9e4c3e9e33db69341d040c827c7afa6789b14c6e16776074f5a"},
+    {file = "MarkupSafe-3.0.1-cp313-cp313t-win32.whl", hash = "sha256:82b5dba6eb1bcc29cc305a18a3c5365d2af06ee71b123216416f7e20d2a84e5b"},
+    {file = "MarkupSafe-3.0.1-cp313-cp313t-win_amd64.whl", hash = "sha256:730d86af59e0e43ce277bb83970530dd223bf7f2a838e086b50affa6ec5f9295"},
+    {file = "MarkupSafe-3.0.1-cp39-cp39-macosx_10_9_universal2.whl", hash = "sha256:4935dd7883f1d50e2ffecca0aa33dc1946a94c8f3fdafb8df5c330e48f71b132"},
+    {file = "MarkupSafe-3.0.1-cp39-cp39-macosx_11_0_arm64.whl", hash = "sha256:e9393357f19954248b00bed7c56f29a25c930593a77630c719653d51e7669c2a"},
+    {file = "MarkupSafe-3.0.1-cp39-cp39-manylinux_2_17_aarch64.manylinux2014_aarch64.whl", hash = "sha256:40621d60d0e58aa573b68ac5e2d6b20d44392878e0bfc159012a5787c4e35bc8"},
+    {file = "MarkupSafe-3.0.1-cp39-cp39-manylinux_2_17_x86_64.manylinux2014_x86_64.whl", hash = "sha256:f94190df587738280d544971500b9cafc9b950d32efcb1fba9ac10d84e6aa4e6"},
+    {file = "MarkupSafe-3.0.1-cp39-cp39-manylinux_2_5_i686.manylinux1_i686.manylinux_2_17_i686.manylinux2014_i686.whl", hash = "sha256:b6a387d61fe41cdf7ea95b38e9af11cfb1a63499af2759444b99185c4ab33f5b"},
+    {file = "MarkupSafe-3.0.1-cp39-cp39-musllinux_1_2_aarch64.whl", hash = "sha256:8ad4ad1429cd4f315f32ef263c1342166695fad76c100c5d979c45d5570ed58b"},
+    {file = "MarkupSafe-3.0.1-cp39-cp39-musllinux_1_2_i686.whl", hash = "sha256:e24bfe89c6ac4c31792793ad9f861b8f6dc4546ac6dc8f1c9083c7c4f2b335cd"},
+    {file = "MarkupSafe-3.0.1-cp39-cp39-musllinux_1_2_x86_64.whl", hash = "sha256:2a4b34a8d14649315c4bc26bbfa352663eb51d146e35eef231dd739d54a5430a"},
+    {file = "MarkupSafe-3.0.1-cp39-cp39-win32.whl", hash = "sha256:242d6860f1fd9191aef5fae22b51c5c19767f93fb9ead4d21924e0bcb17619d8"},
+    {file = "MarkupSafe-3.0.1-cp39-cp39-win_amd64.whl", hash = "sha256:93e8248d650e7e9d49e8251f883eed60ecbc0e8ffd6349e18550925e31bd029b"},
+    {file = "markupsafe-3.0.1.tar.gz", hash = "sha256:3e683ee4f5d0fa2dde4db77ed8dd8a876686e3fc417655c2ece9a90576905344"},
 ]
 
 [[package]]
@@ -5635,19 +5640,19 @@ signedtoken = ["cryptography (>=3.0.0)", "pyjwt (>=2.0.0,<3)"]
 
 [[package]]
 name = "oci"
-version = "2.135.1"
+version = "2.135.2"
 description = "Oracle Cloud Infrastructure Python SDK"
 optional = false
 python-versions = "*"
 files = [
-    {file = "oci-2.135.1-py3-none-any.whl", hash = "sha256:249e24638faa2058b2fa9ab4917ff230f2a8b3b97cd11a261ce6c7f97c3ed2ff"},
-    {file = "oci-2.135.1.tar.gz", hash = "sha256:bea8aaba19d3fed186ceaeedba6d46f1cdb3c6c7d93b39ab225a15cbd63f9ce8"},
+    {file = "oci-2.135.2-py3-none-any.whl", hash = "sha256:5213319244e1c7f108bcb417322f33f01f043fd9636d4063574039f5fdf4e4f7"},
+    {file = "oci-2.135.2.tar.gz", hash = "sha256:520f78983c5246eae80dd5ecfd05e3a565c8b98d02ef0c1b11ba1f61bcccb61d"},
 ]
 
 [package.dependencies]
 certifi = "*"
 circuitbreaker = {version = ">=1.3.1,<3.0.0", markers = "python_version >= \"3.7\""}
-cryptography = ">=3.2.1,<43.0.0"
+cryptography = ">=3.2.1,<46.0.0"
 pyOpenSSL = ">=17.5.0,<25.0.0"
 python-dateutil = ">=2.5.3,<3.0.0"
 pytz = ">=2016.10"
@@ -6277,13 +6282,13 @@ files = [
 
 [[package]]
 name = "pgvecto-rs"
-version = "0.2.1"
+version = "0.2.2"
 description = "Python binding for pgvecto.rs"
 optional = false
 python-versions = "<3.13,>=3.8"
 files = [
-    {file = "pgvecto_rs-0.2.1-py3-none-any.whl", hash = "sha256:b3ee2c465219469ad537b3efea2916477c6c576b3d6fd4298980d0733d12bb27"},
-    {file = "pgvecto_rs-0.2.1.tar.gz", hash = "sha256:07046eaad2c4f75745f76de9ba483541909f1c595aced8d3434224a4f933daca"},
+    {file = "pgvecto_rs-0.2.2-py3-none-any.whl", hash = "sha256:5f3f7f806813de408c45dc10a9eb418b986c4d7b7723e8fce9298f2f7d8fbbd5"},
+    {file = "pgvecto_rs-0.2.2.tar.gz", hash = "sha256:edaa913d1747152b1407cbdf6337d51ac852547b54953ef38997433be3a75a3b"},
 ]
 
 [package.dependencies]
@@ -6753,19 +6758,6 @@ files = [
     {file = "pyarrow-17.0.0-cp312-cp312-win_amd64.whl", hash = "sha256:392bc9feabc647338e6c89267635e111d71edad5fcffba204425a7c8d13610d7"},
     {file = "pyarrow-17.0.0-cp38-cp38-macosx_10_15_x86_64.whl", hash = "sha256:af5ff82a04b2171415f1410cff7ebb79861afc5dae50be73ce06d6e870615204"},
     {file = "pyarrow-17.0.0-cp38-cp38-macosx_11_0_arm64.whl", hash = "sha256:edca18eaca89cd6382dfbcff3dd2d87633433043650c07375d095cd3517561d8"},
-    {file = "pyarrow-17.0.0-cp38-cp38-manylinux_2_17_aarch64.manylinux2014_aarch64.whl", hash = "sha256:7c7916bff914ac5d4a8fe25b7a25e432ff921e72f6f2b7547d1e325c1ad9d155"},
-    {file = "pyarrow-17.0.0-cp38-cp38-manylinux_2_17_x86_64.manylinux2014_x86_64.whl", hash = "sha256:f553ca691b9e94b202ff741bdd40f6ccb70cdd5fbf65c187af132f1317de6145"},
-    {file = "pyarrow-17.0.0-cp38-cp38-manylinux_2_28_aarch64.whl", hash = "sha256:0cdb0e627c86c373205a2f94a510ac4376fdc523f8bb36beab2e7f204416163c"},
-    {file = "pyarrow-17.0.0-cp38-cp38-manylinux_2_28_x86_64.whl", hash = "sha256:d7d192305d9d8bc9082d10f361fc70a73590a4c65cf31c3e6926cd72b76bc35c"},
-    {file = "pyarrow-17.0.0-cp38-cp38-win_amd64.whl", hash = "sha256:02dae06ce212d8b3244dd3e7d12d9c4d3046945a5933d28026598e9dbbda1fca"},
-    {file = "pyarrow-17.0.0-cp39-cp39-macosx_10_15_x86_64.whl", hash = "sha256:13d7a460b412f31e4c0efa1148e1d29bdf18ad1411eb6757d38f8fbdcc8645fb"},
-    {file = "pyarrow-17.0.0-cp39-cp39-macosx_11_0_arm64.whl", hash = "sha256:9b564a51fbccfab5a04a80453e5ac6c9954a9c5ef2890d1bcf63741909c3f8df"},
-    {file = "pyarrow-17.0.0-cp39-cp39-manylinux_2_17_aarch64.manylinux2014_aarch64.whl", hash = "sha256:32503827abbc5aadedfa235f5ece8c4f8f8b0a3cf01066bc8d29de7539532687"},
-    {file = "pyarrow-17.0.0-cp39-cp39-manylinux_2_17_x86_64.manylinux2014_x86_64.whl", hash = "sha256:a155acc7f154b9ffcc85497509bcd0d43efb80d6f733b0dc3bb14e281f131c8b"},
-    {file = "pyarrow-17.0.0-cp39-cp39-manylinux_2_28_aarch64.whl", hash = "sha256:dec8d129254d0188a49f8a1fc99e0560dc1b85f60af729f47de4046015f9b0a5"},
-    {file = "pyarrow-17.0.0-cp39-cp39-manylinux_2_28_x86_64.whl", hash = "sha256:a48ddf5c3c6a6c505904545c25a4ae13646ae1f8ba703c4df4a1bfe4f4006bda"},
-    {file = "pyarrow-17.0.0-cp39-cp39-win_amd64.whl", hash = "sha256:42bf93249a083aca230ba7e2786c5f673507fa97bbd9725a1e2754715151a204"},
-    {file = "pyarrow-17.0.0.tar.gz", hash = "sha256:4beca9521ed2c0921c1023e68d097d0299b62c362639ea315572a58f3f50fd28"},
 ]
 
 [package.dependencies]
@@ -7124,13 +7116,13 @@ test = ["pretend", "pytest (>=3.0.1)", "pytest-rerunfailures"]
 
 [[package]]
 name = "pypandoc"
-version = "1.13"
+version = "1.14"
 description = "Thin wrapper for pandoc."
 optional = false
 python-versions = ">=3.6"
 files = [
-    {file = "pypandoc-1.13-py3-none-any.whl", hash = "sha256:4c7d71bf2f1ed122aac287113b5c4d537a33bbc3c1df5aed11a7d4a7ac074681"},
-    {file = "pypandoc-1.13.tar.gz", hash = "sha256:31652073c7960c2b03570bd1e94f602ca9bc3e70099df5ead4cea98ff5151c1e"},
+    {file = "pypandoc-1.14-py3-none-any.whl", hash = "sha256:1315c7ad7fac7236dacf69a05b521ed2c3f1d0177f70e9b92bfffce6c023df22"},
+    {file = "pypandoc-1.14.tar.gz", hash = "sha256:6b4c45f5f1b9fb5bb562079164806bdbbc3e837b5402bcf3f1139edc5730a197"},
 ]
 
 [[package]]
@@ -8195,13 +8187,13 @@ files = [
 
 [[package]]
 name = "s3transfer"
-version = "0.10.2"
+version = "0.10.3"
 description = "An Amazon S3 Transfer Manager"
 optional = false
 python-versions = ">=3.8"
 files = [
-    {file = "s3transfer-0.10.2-py3-none-any.whl", hash = "sha256:eca1c20de70a39daee580aef4986996620f365c4e0fda6a86100231d62f1bf69"},
-    {file = "s3transfer-0.10.2.tar.gz", hash = "sha256:0711534e9356d3cc692fdde846b4a1e4b0cb6519971860796e6bc4c7aea00ef6"},
+    {file = "s3transfer-0.10.3-py3-none-any.whl", hash = "sha256:263ed587a5803c6c708d3ce44dc4dfedaab4c1a32e8329bab818933d79ddcf5d"},
+    {file = "s3transfer-0.10.3.tar.gz", hash = "sha256:4f50ed74ab84d474ce614475e0b8d5047ff080810aac5d01ea25231cfc944b0c"},
 ]
 
 [package.dependencies]
@@ -8943,13 +8935,13 @@ test = ["pytest", "tornado (>=4.5)", "typeguard"]
 
 [[package]]
 name = "tencentcloud-sdk-python-common"
-version = "3.0.1244"
+version = "3.0.1246"
 description = "Tencent Cloud Common SDK for Python"
 optional = false
 python-versions = "*"
 files = [
-    {file = "tencentcloud-sdk-python-common-3.0.1244.tar.gz", hash = "sha256:51e895a4f380ead51de5ebcef990cfda17cf466b130b8d1be7a95c11158a66e2"},
-    {file = "tencentcloud_sdk_python_common-3.0.1244-py2.py3-none-any.whl", hash = "sha256:a3275496740d7876386166e83331eab7ad66e70fe4216c96893e9697fdeb9848"},
+    {file = "tencentcloud-sdk-python-common-3.0.1246.tar.gz", hash = "sha256:a724d53d5dd6c68beff9ae1c2eb7737889d67f7724819ec5189e878e5322940b"},
+    {file = "tencentcloud_sdk_python_common-3.0.1246-py2.py3-none-any.whl", hash = "sha256:e82a103cc4f1a8d07a83600604eba89e46899b005328b921dbcba9f73b3e819b"},
 ]
 
 [package.dependencies]
@@ -8957,17 +8949,17 @@ requests = ">=2.16.0"
 
 [[package]]
 name = "tencentcloud-sdk-python-hunyuan"
-version = "3.0.1244"
+version = "3.0.1246"
 description = "Tencent Cloud Hunyuan SDK for Python"
 optional = false
 python-versions = "*"
 files = [
-    {file = "tencentcloud-sdk-python-hunyuan-3.0.1244.tar.gz", hash = "sha256:b43f6848ac464cb244f6b1a17c97b00cd36ea40e91a4fba9f7b8ac1e94ae600a"},
-    {file = "tencentcloud_sdk_python_hunyuan-3.0.1244-py2.py3-none-any.whl", hash = "sha256:c2679f49d129dd5ec905cbac97884f6a872e50300ec7ba1ab4fe654d1458213f"},
+    {file = "tencentcloud-sdk-python-hunyuan-3.0.1246.tar.gz", hash = "sha256:099db4a80e7788e1ca09b8be45758e47a6c5976e304857f26621f958cca5b39b"},
+    {file = "tencentcloud_sdk_python_hunyuan-3.0.1246-py2.py3-none-any.whl", hash = "sha256:14ad0a6787bb579a9270e6fe6ee81131a286c2f1b56cac8d7de0056983b6548f"},
 ]
 
 [package.dependencies]
-tencentcloud-sdk-python-common = "3.0.1244"
+tencentcloud-sdk-python-common = "3.0.1246"
 
 [[package]]
 name = "threadpoolctl"
@@ -9636,13 +9628,13 @@ zstd = ["zstandard (>=0.18.0)"]
 
 [[package]]
 name = "uvicorn"
-version = "0.31.0"
+version = "0.31.1"
 description = "The lightning-fast ASGI server."
 optional = false
 python-versions = ">=3.8"
 files = [
-    {file = "uvicorn-0.31.0-py3-none-any.whl", hash = "sha256:cac7be4dd4d891c363cd942160a7b02e69150dcbc7a36be04d5f4af4b17c8ced"},
-    {file = "uvicorn-0.31.0.tar.gz", hash = "sha256:13bc21373d103859f68fe739608e2eb054a816dea79189bc3ca08ea89a275906"},
+    {file = "uvicorn-0.31.1-py3-none-any.whl", hash = "sha256:adc42d9cac80cf3e51af97c1851648066841e7cfb6993a4ca8de29ac1548ed41"},
+    {file = "uvicorn-0.31.1.tar.gz", hash = "sha256:f5167919867b161b7bcaf32646c6a94cdbd4c3aa2eb5c17d36bb9aa5cfd8c493"},
 ]
 
 [package.dependencies]
@@ -10233,13 +10225,13 @@ files = [
 
 [[package]]
 name = "xmltodict"
-version = "0.13.0"
+version = "0.14.1"
 description = "Makes working with XML feel like you are working with JSON"
 optional = false
-python-versions = ">=3.4"
+python-versions = ">=3.6"
 files = [
-    {file = "xmltodict-0.13.0-py2.py3-none-any.whl", hash = "sha256:aa89e8fd76320154a40d19a0df04a4695fb9dc5ba977cbb68ab3e4eb225e7852"},
-    {file = "xmltodict-0.13.0.tar.gz", hash = "sha256:341595a488e3e01a85a9d8911d8912fd922ede5fecc4dce437eb4b6c8d037e56"},
+    {file = "xmltodict-0.14.1-py2.py3-none-any.whl", hash = "sha256:3ef4a7b71c08f19047fcbea572e1d7f4207ab269da1565b5d40e9823d3894e63"},
+    {file = "xmltodict-0.14.1.tar.gz", hash = "sha256:338c8431e4fc554517651972d62f06958718f6262b04316917008e8fd677a6b0"},
 ]
 
 [[package]]
@@ -10595,4 +10587,4 @@ cffi = ["cffi (>=1.11)"]
 [metadata]
 lock-version = "2.0"
 python-versions = ">=3.10,<3.13"
-content-hash = "34ba8efcc67da342036ef075b693f59fdc67d246f40b857c9c1bd6f80c7283bd"
+content-hash = "259bffdcb68115a512695954dea8058bdc0bcc3770ce79106271ea94c6b5ed40"
diff --git a/api/pyproject.toml b/api/pyproject.toml
index 956d4dbd11960..b72cc44abb813 100644
--- a/api/pyproject.toml
+++ b/api/pyproject.toml
@@ -236,7 +236,7 @@ alibabacloud_gpdb20160503 = "~3.8.0"
 alibabacloud_tea_openapi = "~0.3.9"
 chromadb = "0.5.1"
 clickhouse-connect = "~0.7.16"
-elasticsearch = "~8.15.1"
+elasticsearch = "8.14.0"
 oracledb = "~2.2.1"
 pgvecto-rs = { version = "~0.2.1", extras = ['sqlalchemy'] }
 pgvector = "0.2.5"
diff --git a/docker/docker-compose.yaml b/docker/docker-compose.yaml
index c1a76d7f88216..316a6fb64e64c 100644
--- a/docker/docker-compose.yaml
+++ b/docker/docker-compose.yaml
@@ -630,7 +630,7 @@ services:
   # https://www.elastic.co/guide/en/elasticsearch/reference/current/settings.html
   # https://www.elastic.co/guide/en/elasticsearch/reference/current/docker.html#docker-prod-prerequisites
   elasticsearch:
-    image: docker.elastic.co/elasticsearch/elasticsearch:8.15.1
+    image: docker.elastic.co/elasticsearch/elasticsearch:8.14.3
     container_name: elasticsearch
     profiles:
       - elasticsearch
@@ -657,7 +657,7 @@ services:
   # https://www.elastic.co/guide/en/kibana/current/docker.html
   # https://www.elastic.co/guide/en/kibana/current/settings.html
   kibana:
-    image: docker.elastic.co/kibana/kibana:8.15.1
+    image: docker.elastic.co/kibana/kibana:8.14.3
     container_name: kibana
     profiles:
       - elasticsearch

From 75f0a5e36af9889c1173f26daa4e53c21bf09291 Mon Sep 17 00:00:00 2001
From: Sa Zhang <55871322+Nick17t@users.noreply.github.com>
Date: Thu, 10 Oct 2024 17:00:34 +0800
Subject: [PATCH 189/235] fix: Typo Correction in jina.yaml (#9161)

---
 api/core/tools/provider/builtin/jina/jina.yaml | 6 +++---
 1 file changed, 3 insertions(+), 3 deletions(-)

diff --git a/api/core/tools/provider/builtin/jina/jina.yaml b/api/core/tools/provider/builtin/jina/jina.yaml
index 346175c41fbe1..af3ca23ffaff4 100644
--- a/api/core/tools/provider/builtin/jina/jina.yaml
+++ b/api/core/tools/provider/builtin/jina/jina.yaml
@@ -6,9 +6,9 @@ identity:
     zh_Hans: Jina AI
     pt_BR: Jina AI
   description:
-    en_US: Convert any URL to an LLM-friendly input or perform searches on the web for grounding information. Experience improved output for your agent and RAG systems at no cost.
-    zh_Hans: 将任何URL转换为LLM易读的输入或在网页上搜索引擎上搜索引擎。
-    pt_BR: Converte qualquer URL em uma entrada LLm-fácil de ler ou realize pesquisas na web para obter informação de grounding. Tenha uma experiência melhor para seu agente e sistemas RAG sem custo.
+    en_US: Your Search Foundation, Supercharged!
+    zh_Hans: 您的搜索底座，从此不同!
+    pt_BR: Your Search Foundation, Supercharged!
   icon: icon.svg
   tags:
     - search

From be2f1e59f2c2984c4df7f274d1ac0edc67880de7 Mon Sep 17 00:00:00 2001
From: Xiaoguang Sun <sunxiaoguang@users.noreply.github.com>
Date: Thu, 10 Oct 2024 17:21:28 +0800
Subject: [PATCH 190/235] improve: Refresh updated_at field of
 DataSourceOauthBinding model (#9173)

Signed-off-by: Xiaoguang Sun <sunxiaoguang@gmail.com>
---
 api/libs/oauth_data_source.py | 4 ++++
 1 file changed, 4 insertions(+)

diff --git a/api/libs/oauth_data_source.py b/api/libs/oauth_data_source.py
index 05a73b09b7b8c..e747ea97ada4b 100644
--- a/api/libs/oauth_data_source.py
+++ b/api/libs/oauth_data_source.py
@@ -1,3 +1,4 @@
+import datetime
 import urllib.parse
 
 import requests
@@ -69,6 +70,7 @@ def get_access_token(self, code: str):
         if data_source_binding:
             data_source_binding.source_info = source_info
             data_source_binding.disabled = False
+            data_source_binding.updated_at = datetime.datetime.now(datetime.timezone.utc).replace(tzinfo=None)
             db.session.commit()
         else:
             new_data_source_binding = DataSourceOauthBinding(
@@ -104,6 +106,7 @@ def save_internal_access_token(self, access_token: str):
         if data_source_binding:
             data_source_binding.source_info = source_info
             data_source_binding.disabled = False
+            data_source_binding.updated_at = datetime.datetime.now(datetime.timezone.utc).replace(tzinfo=None)
             db.session.commit()
         else:
             new_data_source_binding = DataSourceOauthBinding(
@@ -138,6 +141,7 @@ def sync_data_source(self, binding_id: str):
             }
             data_source_binding.source_info = new_source_info
             data_source_binding.disabled = False
+            data_source_binding.updated_at = datetime.datetime.now(datetime.timezone.utc).replace(tzinfo=None)
             db.session.commit()
         else:
             raise ValueError("Data source binding not found")

From 05c1ef75c4086b20457ac68bd884e9b93d69085e Mon Sep 17 00:00:00 2001
From: zhuhao <37029601+hwzhuhao@users.noreply.github.com>
Date: Thu, 10 Oct 2024 17:39:26 +0800
Subject: [PATCH 191/235] feat: add allow_llm_to_see_data flag for vanna
 (#9156)

---
 .../provider/builtin/vanna/tools/vanna.py     |   7 +-
 .../provider/builtin/vanna/tools/vanna.yaml   |  12 +-
 api/poetry.lock                               | 276 ++++++------------
 api/pyproject.toml                            |   2 +-
 4 files changed, 102 insertions(+), 195 deletions(-)

diff --git a/api/core/tools/provider/builtin/vanna/tools/vanna.py b/api/core/tools/provider/builtin/vanna/tools/vanna.py
index c90d766e483aa..2443991d57ff6 100644
--- a/api/core/tools/provider/builtin/vanna/tools/vanna.py
+++ b/api/core/tools/provider/builtin/vanna/tools/vanna.py
@@ -111,9 +111,10 @@ def _invoke(
         # with "visualize" set to True (default behavior) leads to remote code execution.
         # Affected versions: <= 0.5.5
         #########################################################################################
-        generate_chart = False
-        # generate_chart = tool_parameters.get("generate_chart", True)
-        res = vn.ask(prompt, False, True, generate_chart)
+        allow_llm_to_see_data = tool_parameters.get("allow_llm_to_see_data", False)
+        res = vn.ask(
+            prompt, print_results=False, auto_train=True, visualize=False, allow_llm_to_see_data=allow_llm_to_see_data
+        )
 
         result = []
 
diff --git a/api/core/tools/provider/builtin/vanna/tools/vanna.yaml b/api/core/tools/provider/builtin/vanna/tools/vanna.yaml
index ae2eae94c4dbc..12ca8a862e966 100644
--- a/api/core/tools/provider/builtin/vanna/tools/vanna.yaml
+++ b/api/core/tools/provider/builtin/vanna/tools/vanna.yaml
@@ -200,14 +200,14 @@ parameters:
       en_US: If enabled, it will attempt to train on the metadata of that database
       zh_Hans: 是否自动从数据库获取元数据来训练
     form: form
-  - name: generate_chart
+  - name: allow_llm_to_see_data
     type: boolean
     required: false
-    default: True
+    default: false
     label:
-      en_US: Generate Charts
-      zh_Hans: 生成图表
+      en_US: Whether to allow the LLM to see the data
+      zh_Hans: 是否允许LLM查看数据
     human_description:
-      en_US: Generate Charts
-      zh_Hans: 是否生成图表
+      en_US: Whether to allow the LLM to see the data
+      zh_Hans: 是否允许LLM查看数据
     form: form
diff --git a/api/poetry.lock b/api/poetry.lock
index a63200cacaea7..b7421ca566929 100644
--- a/api/poetry.lock
+++ b/api/poetry.lock
@@ -1628,128 +1628,6 @@ pandas = ["pandas"]
 sqlalchemy = ["sqlalchemy (>1.3.21,<2.0)"]
 tzlocal = ["tzlocal (>=4.0)"]
 
-[[package]]
-name = "clickhouse-driver"
-version = "0.2.9"
-description = "Python driver with native interface for ClickHouse"
-optional = false
-python-versions = "<4,>=3.7"
-files = [
-    {file = "clickhouse-driver-0.2.9.tar.gz", hash = "sha256:050ea4870ead993910b39e7fae965dc1c347b2e8191dcd977cd4b385f9e19f87"},
-    {file = "clickhouse_driver-0.2.9-cp310-cp310-macosx_10_9_x86_64.whl", hash = "sha256:6ce04e9d0d0f39561f312d1ac1a8147bc9206e4267e1a23e20e0423ebac95534"},
-    {file = "clickhouse_driver-0.2.9-cp310-cp310-macosx_11_0_arm64.whl", hash = "sha256:7ae5c8931bf290b9d85582e7955b9aad7f19ff9954e48caa4f9a180ea4d01078"},
-    {file = "clickhouse_driver-0.2.9-cp310-cp310-manylinux_2_17_aarch64.manylinux2014_aarch64.whl", hash = "sha256:3e51792f3bd12c32cb15a907f12de3c9d264843f0bb33dce400e3966c9f09a3f"},
-    {file = "clickhouse_driver-0.2.9-cp310-cp310-manylinux_2_17_ppc64le.manylinux2014_ppc64le.whl", hash = "sha256:42fc546c31e4a04c97b749769335a679c9044dc693fa7a93e38c97fd6727173d"},
-    {file = "clickhouse_driver-0.2.9-cp310-cp310-manylinux_2_17_s390x.manylinux2014_s390x.whl", hash = "sha256:6a383a403d185185c64e49edd6a19b2ec973c5adcb8ebff7ed2fc539a2cc65a5"},
-    {file = "clickhouse_driver-0.2.9-cp310-cp310-manylinux_2_17_x86_64.manylinux2014_x86_64.whl", hash = "sha256:f05321a97e816afc75b3e4f9eda989848fecf14ecf1a91d0f22c04258123d1f7"},
-    {file = "clickhouse_driver-0.2.9-cp310-cp310-manylinux_2_5_i686.manylinux1_i686.manylinux_2_17_i686.manylinux2014_i686.whl", hash = "sha256:be47e793846aac28442b6b1c6554e0731b848a5a7759a54aa2489997354efe4a"},
-    {file = "clickhouse_driver-0.2.9-cp310-cp310-musllinux_1_1_aarch64.whl", hash = "sha256:780e42a215d1ae2f6d695d74dd6f087781fb2fa51c508b58f79e68c24c5364e0"},
-    {file = "clickhouse_driver-0.2.9-cp310-cp310-musllinux_1_1_i686.whl", hash = "sha256:9e28f1fe850675e173db586e9f1ac790e8f7edd507a4227cd54cd7445f8e75b6"},
-    {file = "clickhouse_driver-0.2.9-cp310-cp310-musllinux_1_1_ppc64le.whl", hash = "sha256:125aae7f1308d3083dadbb3c78f828ae492e060f13e4007a0cf53a8169ed7b39"},
-    {file = "clickhouse_driver-0.2.9-cp310-cp310-musllinux_1_1_s390x.whl", hash = "sha256:2f3c4fbb61e75c62a1ab93a1070d362de4cb5682f82833b2c12deccb3bae888d"},
-    {file = "clickhouse_driver-0.2.9-cp310-cp310-musllinux_1_1_x86_64.whl", hash = "sha256:0dc03196a84e32d23b88b665be69afae98f57426f5fdf203e16715b756757961"},
-    {file = "clickhouse_driver-0.2.9-cp310-cp310-win32.whl", hash = "sha256:25695d78a1d7ad6e221e800612eac08559f6182bf6dee0a220d08de7b612d993"},
-    {file = "clickhouse_driver-0.2.9-cp310-cp310-win_amd64.whl", hash = "sha256:367acac95398d721a0a2a6cf87e93638c5588b79498a9848676ce7f182540a6c"},
-    {file = "clickhouse_driver-0.2.9-cp311-cp311-macosx_10_9_x86_64.whl", hash = "sha256:5a7353a7a08eee3aa0001d8a5d771cb1f37e2acae1b48178002431f23892121a"},
-    {file = "clickhouse_driver-0.2.9-cp311-cp311-macosx_11_0_arm64.whl", hash = "sha256:6af1c6cbc3481205503ab72a34aa76d6519249c904aa3f7a84b31e7b435555be"},
-    {file = "clickhouse_driver-0.2.9-cp311-cp311-manylinux_2_17_aarch64.manylinux2014_aarch64.whl", hash = "sha256:48033803abd1100bfff6b9a1769d831b672cd3cda5147e0323b956fd1416d38d"},
-    {file = "clickhouse_driver-0.2.9-cp311-cp311-manylinux_2_17_ppc64le.manylinux2014_ppc64le.whl", hash = "sha256:1f202a58a540c85e47c31dabc8f84b6fe79dca5315c866450a538d58d6fa0571"},
-    {file = "clickhouse_driver-0.2.9-cp311-cp311-manylinux_2_17_s390x.manylinux2014_s390x.whl", hash = "sha256:e4df50fd84bfa4aa1eb7b52d48136066bfb64fabb7ceb62d4c318b45a296200b"},
-    {file = "clickhouse_driver-0.2.9-cp311-cp311-manylinux_2_17_x86_64.manylinux2014_x86_64.whl", hash = "sha256:433a650571a0d7766eb6f402e8f5930222997686c2ee01ded22f1d8fd46af9d4"},
-    {file = "clickhouse_driver-0.2.9-cp311-cp311-manylinux_2_5_i686.manylinux1_i686.manylinux_2_17_i686.manylinux2014_i686.whl", hash = "sha256:232ee260475611cbf7adb554b81db6b5790b36e634fe2164f4ffcd2ca3e63a71"},
-    {file = "clickhouse_driver-0.2.9-cp311-cp311-musllinux_1_1_aarch64.whl", hash = "sha256:09049f7e71f15c9c9a03f597f77fc1f7b61ababd155c06c0d9e64d1453d945d7"},
-    {file = "clickhouse_driver-0.2.9-cp311-cp311-musllinux_1_1_i686.whl", hash = "sha256:424153d1d5f5a807f596a48cc88119f9fb3213ca7e38f57b8d15dcc964dd91f7"},
-    {file = "clickhouse_driver-0.2.9-cp311-cp311-musllinux_1_1_ppc64le.whl", hash = "sha256:4f078fd1cf19c4ca63b8d1e0803df665310c8d5b644c5b02bf2465e8d6ef8f55"},
-    {file = "clickhouse_driver-0.2.9-cp311-cp311-musllinux_1_1_s390x.whl", hash = "sha256:f138d939e26e767537f891170b69a55a88038919f5c10d8865b67b8777fe4848"},
-    {file = "clickhouse_driver-0.2.9-cp311-cp311-musllinux_1_1_x86_64.whl", hash = "sha256:9aafabc7e32942f85dcb46f007f447ab69024831575df97cae28c6ed127654d1"},
-    {file = "clickhouse_driver-0.2.9-cp311-cp311-win32.whl", hash = "sha256:935e16ebf1a1998d8493979d858821a755503c9b8af572d9c450173d4b88868c"},
-    {file = "clickhouse_driver-0.2.9-cp311-cp311-win_amd64.whl", hash = "sha256:306b3102cba278b5dfec6f5f7dc8b78416c403901510475c74913345b56c9e42"},
-    {file = "clickhouse_driver-0.2.9-cp312-cp312-macosx_10_9_x86_64.whl", hash = "sha256:fcb2fd00e58650ae206a6d5dbc83117240e622471aa5124733fbf2805eb8bda0"},
-    {file = "clickhouse_driver-0.2.9-cp312-cp312-macosx_11_0_arm64.whl", hash = "sha256:b7a3e6b0a1eb218e3d870a94c76daaf65da46dca8f6888ea6542f94905c24d88"},
-    {file = "clickhouse_driver-0.2.9-cp312-cp312-manylinux_2_17_aarch64.manylinux2014_aarch64.whl", hash = "sha256:4a8d8e2888a857d8db3d98765a5ad23ab561241feaef68bbffc5a0bd9c142342"},
-    {file = "clickhouse_driver-0.2.9-cp312-cp312-manylinux_2_17_ppc64le.manylinux2014_ppc64le.whl", hash = "sha256:85d50c011467f5ff6772c4059345968b854b72e07a0219030b7c3f68419eb7f7"},
-    {file = "clickhouse_driver-0.2.9-cp312-cp312-manylinux_2_17_s390x.manylinux2014_s390x.whl", hash = "sha256:93b395c1370629ccce8fb3e14cd5be2646d227bd32018c21f753c543e9a7e96b"},
-    {file = "clickhouse_driver-0.2.9-cp312-cp312-manylinux_2_17_x86_64.manylinux2014_x86_64.whl", hash = "sha256:6dbcee870c60d9835e5dce1456ab6b9d807e6669246357f4b321ef747b90fa43"},
-    {file = "clickhouse_driver-0.2.9-cp312-cp312-manylinux_2_5_i686.manylinux1_i686.manylinux_2_17_i686.manylinux2014_i686.whl", hash = "sha256:fffa5a5f317b1ec92e406a30a008929054cf3164d2324a3c465d0a0330273bf8"},
-    {file = "clickhouse_driver-0.2.9-cp312-cp312-musllinux_1_1_aarch64.whl", hash = "sha256:476702740a279744badbd177ae1c4a2d089ec128bd676861219d1f92078e4530"},
-    {file = "clickhouse_driver-0.2.9-cp312-cp312-musllinux_1_1_i686.whl", hash = "sha256:5cd6d95fab5ff80e9dc9baedc9a926f62f74072d42d5804388d63b63bec0bb63"},
-    {file = "clickhouse_driver-0.2.9-cp312-cp312-musllinux_1_1_ppc64le.whl", hash = "sha256:05027d32d7cf3e46cb8d04f8c984745ae01bd1bc7b3579f9dadf9b3cca735697"},
-    {file = "clickhouse_driver-0.2.9-cp312-cp312-musllinux_1_1_s390x.whl", hash = "sha256:3d11831842250b4c1b26503a6e9c511fc03db096608b7c6af743818c421a3032"},
-    {file = "clickhouse_driver-0.2.9-cp312-cp312-musllinux_1_1_x86_64.whl", hash = "sha256:81b4b671b785ebb0b8aeabf2432e47072413d81db959eb8cfd8b6ab58c5799c6"},
-    {file = "clickhouse_driver-0.2.9-cp312-cp312-win32.whl", hash = "sha256:e893bd4e014877174a59e032b0e99809c95ec61328a0e6bd9352c74a2f6111a8"},
-    {file = "clickhouse_driver-0.2.9-cp312-cp312-win_amd64.whl", hash = "sha256:de6624e28eeffd01668803d28ae89e3d4e359b1bff8b60e4933e1cb3c6f86f18"},
-    {file = "clickhouse_driver-0.2.9-cp37-cp37m-macosx_10_9_x86_64.whl", hash = "sha256:909205324089a9ee59bee7ecbfa94595435118cca310fd62efdf13f225aa2965"},
-    {file = "clickhouse_driver-0.2.9-cp37-cp37m-manylinux_2_17_aarch64.manylinux2014_aarch64.whl", hash = "sha256:03f31d6e47dc2b0f367f598f5629147ed056d7216c1788e25190fcfbfa02e749"},
-    {file = "clickhouse_driver-0.2.9-cp37-cp37m-manylinux_2_17_ppc64le.manylinux2014_ppc64le.whl", hash = "sha256:ed84179914b2b7bb434c2322a6e7fd83daa681c97a050450511b66d917a129bb"},
-    {file = "clickhouse_driver-0.2.9-cp37-cp37m-manylinux_2_17_s390x.manylinux2014_s390x.whl", hash = "sha256:67d1bf63efb4ba14ae6c6da99622e4a549e68fc3ee14d859bf611d8e6a61b3fa"},
-    {file = "clickhouse_driver-0.2.9-cp37-cp37m-manylinux_2_17_x86_64.manylinux2014_x86_64.whl", hash = "sha256:9eed23ea41dd582d76f7a2ec7e09cbe5e9fec008f11a4799fa35ce44a3ebd283"},
-    {file = "clickhouse_driver-0.2.9-cp37-cp37m-manylinux_2_5_i686.manylinux1_i686.manylinux_2_17_i686.manylinux2014_i686.whl", hash = "sha256:a654291132766efa2703058317749d7c69b69f02d89bac75703eaf7f775e20da"},
-    {file = "clickhouse_driver-0.2.9-cp37-cp37m-musllinux_1_1_aarch64.whl", hash = "sha256:1c26c5ef16d0ef3cabc5bc03e827e01b0a4afb5b4eaf8850b7cf740cee04a1d4"},
-    {file = "clickhouse_driver-0.2.9-cp37-cp37m-musllinux_1_1_i686.whl", hash = "sha256:b57e83d7986d3cbda6096974a9510eb53cb33ad9072288c87c820ba5eee3370e"},
-    {file = "clickhouse_driver-0.2.9-cp37-cp37m-musllinux_1_1_ppc64le.whl", hash = "sha256:153cc03b36f22cbde55aa6a5bbe99072a025567a54c48b262eb0da15d8cd7c83"},
-    {file = "clickhouse_driver-0.2.9-cp37-cp37m-musllinux_1_1_s390x.whl", hash = "sha256:83a857d99192936091f495826ae97497cd1873af213b1e069d56369fb182ab8e"},
-    {file = "clickhouse_driver-0.2.9-cp37-cp37m-musllinux_1_1_x86_64.whl", hash = "sha256:bb05a9bb22cbe9ad187ad268f86adf7e60df6083331fe59c01571b7b725212dd"},
-    {file = "clickhouse_driver-0.2.9-cp37-cp37m-win32.whl", hash = "sha256:3e282c5c25e32d96ed151e5460d2bf4ecb805ea64449197dd918e84e768016df"},
-    {file = "clickhouse_driver-0.2.9-cp37-cp37m-win_amd64.whl", hash = "sha256:c46dccfb04a9afd61a1b0e60bfefceff917f76da2c863f9b36b39248496d5c77"},
-    {file = "clickhouse_driver-0.2.9-cp38-cp38-macosx_10_9_x86_64.whl", hash = "sha256:612ca9028c718f362c97f552e63d313cf1a70a616ef8532ddb0effdaf12ebef9"},
-    {file = "clickhouse_driver-0.2.9-cp38-cp38-macosx_11_0_arm64.whl", hash = "sha256:471b884d318e012f68d858476052742048918854f7dfe87d78e819f87a848ffb"},
-    {file = "clickhouse_driver-0.2.9-cp38-cp38-manylinux_2_17_aarch64.manylinux2014_aarch64.whl", hash = "sha256:58ee63c35e99da887eb035c8d6d9e64fd298a0efc1460395297dd5cc281a6912"},
-    {file = "clickhouse_driver-0.2.9-cp38-cp38-manylinux_2_17_ppc64le.manylinux2014_ppc64le.whl", hash = "sha256:0819bb63d2c5025a1fb9589f57ef82602687cef11081d6dfa6f2ce44606a1772"},
-    {file = "clickhouse_driver-0.2.9-cp38-cp38-manylinux_2_17_s390x.manylinux2014_s390x.whl", hash = "sha256:f6680ee18870bca1fbab1736c8203a965efaec119ab4c37821ad99add248ee08"},
-    {file = "clickhouse_driver-0.2.9-cp38-cp38-manylinux_2_17_x86_64.manylinux2014_x86_64.whl", hash = "sha256:713c498741b54debd3a10a5529e70b6ed85ca33c3e8629e24ae5cd8160b5a5f2"},
-    {file = "clickhouse_driver-0.2.9-cp38-cp38-manylinux_2_5_i686.manylinux1_i686.manylinux_2_17_i686.manylinux2014_i686.whl", hash = "sha256:730837b8f63941065c9c955c44286aef0987fb084ffb3f55bf1e4fe07df62269"},
-    {file = "clickhouse_driver-0.2.9-cp38-cp38-musllinux_1_1_aarch64.whl", hash = "sha256:9f4e38b2ea09214c8e7848a19391009a18c56a3640e1ba1a606b9e57aeb63404"},
-    {file = "clickhouse_driver-0.2.9-cp38-cp38-musllinux_1_1_i686.whl", hash = "sha256:457f1d6639e0345b717ae603c79bd087a35361ce68c1c308d154b80b841e5e7d"},
-    {file = "clickhouse_driver-0.2.9-cp38-cp38-musllinux_1_1_ppc64le.whl", hash = "sha256:49a55aeb8ea625a87965a96e361bbb1ad67d0931bfb2a575f899c1064e70c2da"},
-    {file = "clickhouse_driver-0.2.9-cp38-cp38-musllinux_1_1_s390x.whl", hash = "sha256:9230058d8c9b1a04079afae4650fb67745f0f1c39db335728f64d48bd2c19246"},
-    {file = "clickhouse_driver-0.2.9-cp38-cp38-musllinux_1_1_x86_64.whl", hash = "sha256:8798258bd556542dd9c6b8ebe62f9c5110c9dcdf97c57fb077e7b8b6d6da0826"},
-    {file = "clickhouse_driver-0.2.9-cp38-cp38-win32.whl", hash = "sha256:ce8e3f4be46bcc63555863f70ab0035202b082b37e6f16876ef50e7bc4b47056"},
-    {file = "clickhouse_driver-0.2.9-cp38-cp38-win_amd64.whl", hash = "sha256:2d982959ff628255808d895a67493f2dab0c3a9bfc65eeda0f00c8ae9962a1b3"},
-    {file = "clickhouse_driver-0.2.9-cp39-cp39-macosx_10_9_x86_64.whl", hash = "sha256:a46b227fab4420566ed24ee70d90076226d16fcf09c6ad4d428717efcf536446"},
-    {file = "clickhouse_driver-0.2.9-cp39-cp39-macosx_11_0_arm64.whl", hash = "sha256:7eaa2ce5ea08cf5fddebb8c274c450e102f329f9e6966b6cd85aa671c48e5552"},
-    {file = "clickhouse_driver-0.2.9-cp39-cp39-manylinux_2_17_aarch64.manylinux2014_aarch64.whl", hash = "sha256:f97f0083194d6e23b5ef6156ed0d5388c37847b298118199d7937ba26412a9e2"},
-    {file = "clickhouse_driver-0.2.9-cp39-cp39-manylinux_2_17_ppc64le.manylinux2014_ppc64le.whl", hash = "sha256:a6cab5cdbb0f8ee51d879d977b78f07068b585225ac656f3c081896c362e8f83"},
-    {file = "clickhouse_driver-0.2.9-cp39-cp39-manylinux_2_17_s390x.manylinux2014_s390x.whl", hash = "sha256:cdb1b011a53ee71539e9dc655f268b111bac484db300da92829ed59e910a8fd0"},
-    {file = "clickhouse_driver-0.2.9-cp39-cp39-manylinux_2_17_x86_64.manylinux2014_x86_64.whl", hash = "sha256:7bf51bb761b281d20910b4b689c699ef98027845467daa5bb5dfdb53bd6ee404"},
-    {file = "clickhouse_driver-0.2.9-cp39-cp39-manylinux_2_5_i686.manylinux1_i686.manylinux_2_17_i686.manylinux2014_i686.whl", hash = "sha256:b8ea462e3cebb121ff55002e9c8a9a0a3fd9b5bbbf688b4960f0a83c0172fb31"},
-    {file = "clickhouse_driver-0.2.9-cp39-cp39-musllinux_1_1_aarch64.whl", hash = "sha256:70bee21c245226ad0d637bf470472e2d487b86911b6d673a862127b934336ff4"},
-    {file = "clickhouse_driver-0.2.9-cp39-cp39-musllinux_1_1_i686.whl", hash = "sha256:253a3c223b944d691bf0abbd599f592ea3b36f0a71d2526833b1718f37eca5c2"},
-    {file = "clickhouse_driver-0.2.9-cp39-cp39-musllinux_1_1_ppc64le.whl", hash = "sha256:a6549b53fc5c403dc556cb39b2ae94d73f9b113daa00438a660bb1dd5380ae4d"},
-    {file = "clickhouse_driver-0.2.9-cp39-cp39-musllinux_1_1_s390x.whl", hash = "sha256:1c685cd4abe61af1c26279ff04b9f567eb4d6c1ec7fb265af7481b1f153043aa"},
-    {file = "clickhouse_driver-0.2.9-cp39-cp39-musllinux_1_1_x86_64.whl", hash = "sha256:7e25144219577491929d032a6c3ddd63c6cd7fa764af829a5637f798190d9b26"},
-    {file = "clickhouse_driver-0.2.9-cp39-cp39-win32.whl", hash = "sha256:0b9925610d25405a8e6d83ff4f54fc2456a121adb0155999972f5edd6ba3efc8"},
-    {file = "clickhouse_driver-0.2.9-cp39-cp39-win_amd64.whl", hash = "sha256:b243de483cfa02716053b0148d73558f4694f3c27b97fc1eaa97d7079563a14d"},
-    {file = "clickhouse_driver-0.2.9-pp310-pypy310_pp73-macosx_10_9_x86_64.whl", hash = "sha256:45a3d5b1d06750fd6a18c29b871494a2635670099ec7693e756a5885a4a70dbf"},
-    {file = "clickhouse_driver-0.2.9-pp310-pypy310_pp73-manylinux_2_17_aarch64.manylinux2014_aarch64.whl", hash = "sha256:8415ffebd6ca9eef3024763abc450f8659f1716d015bd563c537d01c7fbc3569"},
-    {file = "clickhouse_driver-0.2.9-pp310-pypy310_pp73-manylinux_2_17_x86_64.manylinux2014_x86_64.whl", hash = "sha256:ace48db993aa4bd31c42de0fa8d38c94ad47405916d6b61f7a7168a48fb52ac1"},
-    {file = "clickhouse_driver-0.2.9-pp310-pypy310_pp73-manylinux_2_5_i686.manylinux1_i686.manylinux_2_17_i686.manylinux2014_i686.whl", hash = "sha256:b07123334fe143bfe6fa4e3d4b732d647d5fd2cfb9ec7f2f76104b46fe9d20c6"},
-    {file = "clickhouse_driver-0.2.9-pp310-pypy310_pp73-win_amd64.whl", hash = "sha256:e2af3efa73d296420ce6362789f5b1febf75d4aa159a479393f01549115509d5"},
-    {file = "clickhouse_driver-0.2.9-pp37-pypy37_pp73-macosx_10_9_x86_64.whl", hash = "sha256:baf57eede88d07a1eb04352d26fc58a4d97991ca3d8840f7c5d48691dec9f251"},
-    {file = "clickhouse_driver-0.2.9-pp37-pypy37_pp73-manylinux_2_17_aarch64.manylinux2014_aarch64.whl", hash = "sha256:275d0ccdab9c3571bdb3e9acfab4497930aa584ff2766b035bb2f854deaf8b82"},
-    {file = "clickhouse_driver-0.2.9-pp37-pypy37_pp73-manylinux_2_17_x86_64.manylinux2014_x86_64.whl", hash = "sha256:293da77bfcac3168fb35b27c242f97c1a05502435c0686ecbb8e2e4abcb3de26"},
-    {file = "clickhouse_driver-0.2.9-pp37-pypy37_pp73-manylinux_2_5_i686.manylinux1_i686.manylinux_2_17_i686.manylinux2014_i686.whl", hash = "sha256:8d6c2e5830705e4eeef33070ca4d5a24dfa221f28f2f540e5e6842c26e70b10b"},
-    {file = "clickhouse_driver-0.2.9-pp37-pypy37_pp73-win_amd64.whl", hash = "sha256:11934bd78d97dd7e1a23a6222b5edd1e1b4d34e1ead5c846dc2b5c56fdc35ff5"},
-    {file = "clickhouse_driver-0.2.9-pp38-pypy38_pp73-macosx_10_9_x86_64.whl", hash = "sha256:b802b6f0fbdcc3ab81b87f09b694dde91ab049f44d1d2c08c3dc8ea9a5950cfa"},
-    {file = "clickhouse_driver-0.2.9-pp38-pypy38_pp73-manylinux_2_17_aarch64.manylinux2014_aarch64.whl", hash = "sha256:7af871c5315eb829ecf4533c790461ea8f73b3bfd5f533b0467e479fdf6ddcfd"},
-    {file = "clickhouse_driver-0.2.9-pp38-pypy38_pp73-manylinux_2_17_x86_64.manylinux2014_x86_64.whl", hash = "sha256:9d577dd4867b9e26cf60590e1f500990c8701a6e3cfbb9e644f4d0c0fb607028"},
-    {file = "clickhouse_driver-0.2.9-pp38-pypy38_pp73-manylinux_2_5_i686.manylinux1_i686.manylinux_2_17_i686.manylinux2014_i686.whl", hash = "sha256:2ed3dea2d1eca85fef5b8564ddd76dedb15a610c77d55d555b49d9f7c896b64b"},
-    {file = "clickhouse_driver-0.2.9-pp38-pypy38_pp73-win_amd64.whl", hash = "sha256:91ec96f2c48e5bdeac9eea43a9bc9cc19acb2d2c59df0a13d5520dfc32457605"},
-    {file = "clickhouse_driver-0.2.9-pp39-pypy39_pp73-macosx_10_9_x86_64.whl", hash = "sha256:7667ab423452754f36ba8fb41e006a46baace9c94e2aca2a745689b9f2753dfb"},
-    {file = "clickhouse_driver-0.2.9-pp39-pypy39_pp73-manylinux_2_17_aarch64.manylinux2014_aarch64.whl", hash = "sha256:653583b1f3b088d106f180d6f02c90917ecd669ec956b62903a05df4a7f44863"},
-    {file = "clickhouse_driver-0.2.9-pp39-pypy39_pp73-manylinux_2_17_x86_64.manylinux2014_x86_64.whl", hash = "sha256:7ef3dd0cbdf2f0171caab90389af0ede068ec802bf46c6a77f14e6edc86671bc"},
-    {file = "clickhouse_driver-0.2.9-pp39-pypy39_pp73-manylinux_2_5_i686.manylinux1_i686.manylinux_2_17_i686.manylinux2014_i686.whl", hash = "sha256:11b1833ee8ff8d5df39a34a895e060b57bd81e05ea68822bc60476daff4ce1c8"},
-    {file = "clickhouse_driver-0.2.9-pp39-pypy39_pp73-win_amd64.whl", hash = "sha256:8a3195639e6393b9d4aafe736036881ff86b6be5855d4bf7d9f5c31637181ec3"},
-]
-
-[package.dependencies]
-pytz = "*"
-tzlocal = "*"
-
-[package.extras]
-lz4 = ["clickhouse-cityhash (>=1.0.2.1)", "lz4", "lz4 (<=3.0.1)"]
-numpy = ["numpy (>=1.12.0)", "pandas (>=0.24.0)"]
-zstd = ["clickhouse-cityhash (>=1.0.2.1)", "zstd"]
-
 [[package]]
 name = "cloudpickle"
 version = "2.2.1"
@@ -2589,6 +2467,24 @@ files = [
     {file = "filetype-1.2.0.tar.gz", hash = "sha256:66b56cd6474bf41d8c54660347d37afcc3f7d1970648de365c102ef77548aadb"},
 ]
 
+[[package]]
+name = "flasgger"
+version = "0.9.7.1"
+description = "Extract swagger specs from your flask project"
+optional = false
+python-versions = "*"
+files = [
+    {file = "flasgger-0.9.7.1.tar.gz", hash = "sha256:ca098e10bfbb12f047acc6299cc70a33851943a746e550d86e65e60d4df245fb"},
+]
+
+[package.dependencies]
+Flask = ">=0.10"
+jsonschema = ">=3.0.1"
+mistune = "*"
+packaging = "*"
+PyYAML = ">=3.0"
+six = ">=1.10.0"
+
 [[package]]
 name = "flask"
 version = "3.0.3"
@@ -5022,6 +4918,17 @@ files = [
 [package.dependencies]
 tqdm = "*"
 
+[[package]]
+name = "mistune"
+version = "3.0.2"
+description = "A sane and fast Markdown parser with useful plugins and renderers"
+optional = false
+python-versions = ">=3.7"
+files = [
+    {file = "mistune-3.0.2-py3-none-any.whl", hash = "sha256:71481854c30fdbc938963d3605b72501f5c10a9320ecd412c121c163a1c7d205"},
+    {file = "mistune-3.0.2.tar.gz", hash = "sha256:fc7f93ded930c92394ef2cb6f04a8aabab4117a91449e72dcc8dfa646a508be8"},
+]
+
 [[package]]
 name = "mmh3"
 version = "5.0.1"
@@ -6758,6 +6665,19 @@ files = [
     {file = "pyarrow-17.0.0-cp312-cp312-win_amd64.whl", hash = "sha256:392bc9feabc647338e6c89267635e111d71edad5fcffba204425a7c8d13610d7"},
     {file = "pyarrow-17.0.0-cp38-cp38-macosx_10_15_x86_64.whl", hash = "sha256:af5ff82a04b2171415f1410cff7ebb79861afc5dae50be73ce06d6e870615204"},
     {file = "pyarrow-17.0.0-cp38-cp38-macosx_11_0_arm64.whl", hash = "sha256:edca18eaca89cd6382dfbcff3dd2d87633433043650c07375d095cd3517561d8"},
+    {file = "pyarrow-17.0.0-cp38-cp38-manylinux_2_17_aarch64.manylinux2014_aarch64.whl", hash = "sha256:7c7916bff914ac5d4a8fe25b7a25e432ff921e72f6f2b7547d1e325c1ad9d155"},
+    {file = "pyarrow-17.0.0-cp38-cp38-manylinux_2_17_x86_64.manylinux2014_x86_64.whl", hash = "sha256:f553ca691b9e94b202ff741bdd40f6ccb70cdd5fbf65c187af132f1317de6145"},
+    {file = "pyarrow-17.0.0-cp38-cp38-manylinux_2_28_aarch64.whl", hash = "sha256:0cdb0e627c86c373205a2f94a510ac4376fdc523f8bb36beab2e7f204416163c"},
+    {file = "pyarrow-17.0.0-cp38-cp38-manylinux_2_28_x86_64.whl", hash = "sha256:d7d192305d9d8bc9082d10f361fc70a73590a4c65cf31c3e6926cd72b76bc35c"},
+    {file = "pyarrow-17.0.0-cp38-cp38-win_amd64.whl", hash = "sha256:02dae06ce212d8b3244dd3e7d12d9c4d3046945a5933d28026598e9dbbda1fca"},
+    {file = "pyarrow-17.0.0-cp39-cp39-macosx_10_15_x86_64.whl", hash = "sha256:13d7a460b412f31e4c0efa1148e1d29bdf18ad1411eb6757d38f8fbdcc8645fb"},
+    {file = "pyarrow-17.0.0-cp39-cp39-macosx_11_0_arm64.whl", hash = "sha256:9b564a51fbccfab5a04a80453e5ac6c9954a9c5ef2890d1bcf63741909c3f8df"},
+    {file = "pyarrow-17.0.0-cp39-cp39-manylinux_2_17_aarch64.manylinux2014_aarch64.whl", hash = "sha256:32503827abbc5aadedfa235f5ece8c4f8f8b0a3cf01066bc8d29de7539532687"},
+    {file = "pyarrow-17.0.0-cp39-cp39-manylinux_2_17_x86_64.manylinux2014_x86_64.whl", hash = "sha256:a155acc7f154b9ffcc85497509bcd0d43efb80d6f733b0dc3bb14e281f131c8b"},
+    {file = "pyarrow-17.0.0-cp39-cp39-manylinux_2_28_aarch64.whl", hash = "sha256:dec8d129254d0188a49f8a1fc99e0560dc1b85f60af729f47de4046015f9b0a5"},
+    {file = "pyarrow-17.0.0-cp39-cp39-manylinux_2_28_x86_64.whl", hash = "sha256:a48ddf5c3c6a6c505904545c25a4ae13646ae1f8ba703c4df4a1bfe4f4006bda"},
+    {file = "pyarrow-17.0.0-cp39-cp39-win_amd64.whl", hash = "sha256:42bf93249a083aca230ba7e2786c5f673507fa97bbd9725a1e2754715151a204"},
+    {file = "pyarrow-17.0.0.tar.gz", hash = "sha256:4beca9521ed2c0921c1023e68d097d0299b62c362639ea315572a58f3f50fd28"},
 ]
 
 [package.dependencies]
@@ -9411,23 +9331,6 @@ files = [
     {file = "tzdata-2024.2.tar.gz", hash = "sha256:7d85cc416e9382e69095b7bdf4afd9e3880418a2413feec7069d533d6b4e31cc"},
 ]
 
-[[package]]
-name = "tzlocal"
-version = "5.2"
-description = "tzinfo object for the local timezone"
-optional = false
-python-versions = ">=3.8"
-files = [
-    {file = "tzlocal-5.2-py3-none-any.whl", hash = "sha256:49816ef2fe65ea8ac19d19aa7a1ae0551c834303d5014c6d5a62e4cbda8047b8"},
-    {file = "tzlocal-5.2.tar.gz", hash = "sha256:8d399205578f1a9342816409cc1e46a93ebd5755e39ea2d85334bea911bf0e6e"},
-]
-
-[package.dependencies]
-tzdata = {version = "*", markers = "platform_system == \"Windows\""}
-
-[package.extras]
-devenv = ["check-manifest", "pytest (>=4.3)", "pytest-cov", "pytest-mock (>=3.3)", "zest.releaser"]
-
 [[package]]
 name = "ujson"
 version = "5.10.0"
@@ -9709,19 +9612,20 @@ files = [
 
 [[package]]
 name = "vanna"
-version = "0.5.5"
+version = "0.7.3"
 description = "Generate SQL queries from natural language"
 optional = false
 python-versions = ">=3.9"
 files = [
-    {file = "vanna-0.5.5-py3-none-any.whl", hash = "sha256:e1a308b7127b9e98c2579c0e4178fc1475d891c498e4a0667cffa10df8891e73"},
-    {file = "vanna-0.5.5.tar.gz", hash = "sha256:7d9bf188a635bb75e4f8db15f0e6dbe72a426784779485f087b2df0ce175e664"},
+    {file = "vanna-0.7.3-py3-none-any.whl", hash = "sha256:82ba39e5d6c503d1c8cca60835ed401d20ec3a3da98d487f529901dcb30061d6"},
+    {file = "vanna-0.7.3.tar.gz", hash = "sha256:4590dd94d2fe180b4efc7a83c867b73144ef58794018910dc226857cfb703077"},
 ]
 
 [package.dependencies]
-clickhouse_driver = {version = "*", optional = true, markers = "extra == \"clickhouse\""}
+clickhouse_connect = {version = "*", optional = true, markers = "extra == \"clickhouse\""}
 db-dtypes = {version = "*", optional = true, markers = "extra == \"postgres\""}
 duckdb = {version = "*", optional = true, markers = "extra == \"duckdb\""}
+flasgger = "*"
 flask = "*"
 flask-sock = "*"
 kaleido = "*"
@@ -9735,17 +9639,20 @@ sqlparse = "*"
 tabulate = "*"
 
 [package.extras]
-all = ["PyMySQL", "anthropic", "chromadb", "db-dtypes", "duckdb", "fastembed", "google-cloud-aiplatform", "google-cloud-bigquery", "google-generativeai", "httpx", "marqo", "mistralai", "ollama", "openai", "opensearch-dsl", "opensearch-py", "pinecone-client", "psycopg2-binary", "qdrant-client", "snowflake-connector-python", "transformers", "zhipuai"]
+all = ["PyMySQL", "anthropic", "azure-common", "azure-identity", "azure-search-documents", "chromadb", "db-dtypes", "duckdb", "fastembed", "google-cloud-aiplatform", "google-cloud-bigquery", "google-generativeai", "httpx", "marqo", "mistralai (>=1.0.0)", "ollama", "openai", "opensearch-dsl", "opensearch-py", "pinecone-client", "psycopg2-binary", "pymilvus[model]", "qdrant-client", "qianfan", "snowflake-connector-python", "transformers", "weaviate-client", "zhipuai"]
 anthropic = ["anthropic"]
+azuresearch = ["azure-common", "azure-identity", "azure-search-documents", "fastembed"]
+bedrock = ["boto3", "botocore"]
 bigquery = ["google-cloud-bigquery"]
 chromadb = ["chromadb"]
-clickhouse = ["clickhouse_driver"]
+clickhouse = ["clickhouse_connect"]
 duckdb = ["duckdb"]
 gemini = ["google-generativeai"]
 google = ["google-cloud-aiplatform", "google-generativeai"]
 hf = ["transformers"]
 marqo = ["marqo"]
-mistralai = ["mistralai"]
+milvus = ["pymilvus[model]"]
+mistralai = ["mistralai (>=1.0.0)"]
 mysql = ["PyMySQL"]
 ollama = ["httpx", "ollama"]
 openai = ["openai"]
@@ -9753,9 +9660,11 @@ opensearch = ["opensearch-dsl", "opensearch-py"]
 pinecone = ["fastembed", "pinecone-client"]
 postgres = ["db-dtypes", "psycopg2-binary"]
 qdrant = ["fastembed", "qdrant-client"]
+qianfan = ["qianfan"]
 snowflake = ["snowflake-connector-python"]
 test = ["tox"]
 vllm = ["vllm"]
+weaviate = ["weaviate-client"]
 zhipuai = ["zhipuai"]
 
 [[package]]
@@ -10423,54 +10332,51 @@ test = ["zope.testrunner"]
 
 [[package]]
 name = "zope-interface"
-version = "7.0.3"
+version = "7.1.0"
 description = "Interfaces for Python"
 optional = false
 python-versions = ">=3.8"
 files = [
-    {file = "zope.interface-7.0.3-cp310-cp310-macosx_10_9_x86_64.whl", hash = "sha256:9b9369671a20b8d039b8e5a1a33abd12e089e319a3383b4cc0bf5c67bd05fe7b"},
-    {file = "zope.interface-7.0.3-cp310-cp310-macosx_11_0_arm64.whl", hash = "sha256:db6237e8fa91ea4f34d7e2d16d74741187e9105a63bbb5686c61fea04cdbacca"},
-    {file = "zope.interface-7.0.3-cp310-cp310-manylinux_2_17_aarch64.manylinux2014_aarch64.whl", hash = "sha256:53d678bb1c3b784edbfb0adeebfeea6bf479f54da082854406a8f295d36f8386"},
-    {file = "zope.interface-7.0.3-cp310-cp310-manylinux_2_5_i686.manylinux1_i686.manylinux_2_17_i686.manylinux2014_i686.whl", hash = "sha256:3aa8fcbb0d3c2be1bfd013a0f0acd636f6ed570c287743ae2bbd467ee967154d"},
-    {file = "zope.interface-7.0.3-cp310-cp310-manylinux_2_5_x86_64.manylinux1_x86_64.manylinux_2_17_x86_64.manylinux2014_x86_64.whl", hash = "sha256:6195c3c03fef9f87c0dbee0b3b6451df6e056322463cf35bca9a088e564a3c58"},
-    {file = "zope.interface-7.0.3-cp310-cp310-win_amd64.whl", hash = "sha256:11fa1382c3efb34abf16becff8cb214b0b2e3144057c90611621f2d186b7e1b7"},
-    {file = "zope.interface-7.0.3-cp311-cp311-macosx_10_9_x86_64.whl", hash = "sha256:af94e429f9d57b36e71ef4e6865182090648aada0cb2d397ae2b3f7fc478493a"},
-    {file = "zope.interface-7.0.3-cp311-cp311-macosx_11_0_arm64.whl", hash = "sha256:6dd647fcd765030638577fe6984284e0ebba1a1008244c8a38824be096e37fe3"},
-    {file = "zope.interface-7.0.3-cp311-cp311-manylinux_2_17_aarch64.manylinux2014_aarch64.whl", hash = "sha256:1bee1b722077d08721005e8da493ef3adf0b7908e0cd85cc7dc836ac117d6f32"},
-    {file = "zope.interface-7.0.3-cp311-cp311-manylinux_2_5_i686.manylinux1_i686.manylinux_2_17_i686.manylinux2014_i686.whl", hash = "sha256:2545d6d7aac425d528cd9bf0d9e55fcd47ab7fd15f41a64b1c4bf4c6b24946dc"},
-    {file = "zope.interface-7.0.3-cp311-cp311-manylinux_2_5_x86_64.manylinux1_x86_64.manylinux_2_17_x86_64.manylinux2014_x86_64.whl", hash = "sha256:6d04b11ea47c9c369d66340dbe51e9031df2a0de97d68f442305ed7625ad6493"},
-    {file = "zope.interface-7.0.3-cp311-cp311-win_amd64.whl", hash = "sha256:064ade95cb54c840647205987c7b557f75d2b2f7d1a84bfab4cf81822ef6e7d1"},
-    {file = "zope.interface-7.0.3-cp312-cp312-macosx_10_9_x86_64.whl", hash = "sha256:3fcdc76d0cde1c09c37b7c6b0f8beba2d857d8417b055d4f47df9c34ec518bdd"},
-    {file = "zope.interface-7.0.3-cp312-cp312-macosx_11_0_arm64.whl", hash = "sha256:3d4b91821305c8d8f6e6207639abcbdaf186db682e521af7855d0bea3047c8ca"},
-    {file = "zope.interface-7.0.3-cp312-cp312-manylinux_2_17_aarch64.manylinux2014_aarch64.whl", hash = "sha256:35062d93bc49bd9b191331c897a96155ffdad10744ab812485b6bad5b588d7e4"},
-    {file = "zope.interface-7.0.3-cp312-cp312-manylinux_2_5_i686.manylinux1_i686.manylinux_2_17_i686.manylinux2014_i686.whl", hash = "sha256:c96b3e6b0d4f6ddfec4e947130ec30bd2c7b19db6aa633777e46c8eecf1d6afd"},
-    {file = "zope.interface-7.0.3-cp312-cp312-manylinux_2_5_x86_64.manylinux1_x86_64.manylinux_2_17_x86_64.manylinux2014_x86_64.whl", hash = "sha256:7e0c151a6c204f3830237c59ee4770cc346868a7a1af6925e5e38650141a7f05"},
-    {file = "zope.interface-7.0.3-cp312-cp312-win_amd64.whl", hash = "sha256:3de1d553ce72868b77a7e9d598c9bff6d3816ad2b4cc81c04f9d8914603814f3"},
-    {file = "zope.interface-7.0.3-cp313-cp313-manylinux_2_17_aarch64.manylinux2014_aarch64.whl", hash = "sha256:ab985c566a99cc5f73bc2741d93f1ed24a2cc9da3890144d37b9582965aff996"},
-    {file = "zope.interface-7.0.3-cp313-cp313-manylinux_2_5_i686.manylinux1_i686.manylinux_2_17_i686.manylinux2014_i686.whl", hash = "sha256:d976fa7b5faf5396eb18ce6c132c98e05504b52b60784e3401f4ef0b2e66709b"},
-    {file = "zope.interface-7.0.3-cp313-cp313-manylinux_2_5_x86_64.manylinux1_x86_64.manylinux_2_17_x86_64.manylinux2014_x86_64.whl", hash = "sha256:21a207c6b2c58def5011768140861a73f5240f4f39800625072ba84e76c9da0b"},
-    {file = "zope.interface-7.0.3-cp38-cp38-macosx_10_9_x86_64.whl", hash = "sha256:382d31d1e68877061daaa6499468e9eb38eb7625d4369b1615ac08d3860fe896"},
-    {file = "zope.interface-7.0.3-cp38-cp38-macosx_11_0_arm64.whl", hash = "sha256:2c4316a30e216f51acbd9fb318aa5af2e362b716596d82cbb92f9101c8f8d2e7"},
-    {file = "zope.interface-7.0.3-cp38-cp38-manylinux_2_17_aarch64.manylinux2014_aarch64.whl", hash = "sha256:01e6e58078ad2799130c14a1d34ec89044ada0e1495329d72ee0407b9ae5100d"},
-    {file = "zope.interface-7.0.3-cp38-cp38-manylinux_2_5_i686.manylinux1_i686.manylinux_2_17_i686.manylinux2014_i686.whl", hash = "sha256:799ef7a444aebbad5a145c3b34bff012b54453cddbde3332d47ca07225792ea4"},
-    {file = "zope.interface-7.0.3-cp38-cp38-manylinux_2_5_x86_64.manylinux1_x86_64.manylinux_2_17_x86_64.manylinux2014_x86_64.whl", hash = "sha256:d3b7ce6d46fb0e60897d62d1ff370790ce50a57d40a651db91a3dde74f73b738"},
-    {file = "zope.interface-7.0.3-cp38-cp38-win_amd64.whl", hash = "sha256:f418c88f09c3ba159b95a9d1cfcdbe58f208443abb1f3109f4b9b12fd60b187c"},
-    {file = "zope.interface-7.0.3-cp39-cp39-macosx_10_9_x86_64.whl", hash = "sha256:84f8794bd59ca7d09d8fce43ae1b571be22f52748169d01a13d3ece8394d8b5b"},
-    {file = "zope.interface-7.0.3-cp39-cp39-macosx_11_0_arm64.whl", hash = "sha256:7d92920416f31786bc1b2f34cc4fc4263a35a407425319572cbf96b51e835cd3"},
-    {file = "zope.interface-7.0.3-cp39-cp39-manylinux_2_17_aarch64.manylinux2014_aarch64.whl", hash = "sha256:95e5913ec718010dc0e7c215d79a9683b4990e7026828eedfda5268e74e73e11"},
-    {file = "zope.interface-7.0.3-cp39-cp39-manylinux_2_5_i686.manylinux1_i686.manylinux_2_17_i686.manylinux2014_i686.whl", hash = "sha256:1eeeb92cb7d95c45e726e3c1afe7707919370addae7ed14f614e22217a536958"},
-    {file = "zope.interface-7.0.3-cp39-cp39-manylinux_2_5_x86_64.manylinux1_x86_64.manylinux_2_17_x86_64.manylinux2014_x86_64.whl", hash = "sha256:ecd32f30f40bfd8511b17666895831a51b532e93fc106bfa97f366589d3e4e0e"},
-    {file = "zope.interface-7.0.3-cp39-cp39-win_amd64.whl", hash = "sha256:5112c530fa8aa2108a3196b9c2f078f5738c1c37cfc716970edc0df0414acda8"},
-    {file = "zope.interface-7.0.3.tar.gz", hash = "sha256:cd2690d4b08ec9eaf47a85914fe513062b20da78d10d6d789a792c0b20307fb1"},
+    {file = "zope.interface-7.1.0-cp310-cp310-macosx_10_9_x86_64.whl", hash = "sha256:2bd9e9f366a5df08ebbdc159f8224904c1c5ce63893984abb76954e6fbe4381a"},
+    {file = "zope.interface-7.1.0-cp310-cp310-macosx_11_0_arm64.whl", hash = "sha256:661d5df403cd3c5b8699ac480fa7f58047a3253b029db690efa0c3cf209993ef"},
+    {file = "zope.interface-7.1.0-cp310-cp310-manylinux_2_5_i686.manylinux1_i686.manylinux_2_17_i686.manylinux2014_i686.whl", hash = "sha256:2b6a4924f5bad9fe21d99f66a07da60d75696a136162427951ec3cb223a5570d"},
+    {file = "zope.interface-7.1.0-cp310-cp310-manylinux_2_5_x86_64.manylinux1_x86_64.manylinux_2_17_x86_64.manylinux2014_x86_64.whl", hash = "sha256:80a3c00b35f6170be5454b45abe2719ea65919a2f09e8a6e7b1362312a872cd3"},
+    {file = "zope.interface-7.1.0-cp310-cp310-win_amd64.whl", hash = "sha256:b936d61dbe29572fd2cfe13e30b925e5383bed1aba867692670f5a2a2eb7b4e9"},
+    {file = "zope.interface-7.1.0-cp311-cp311-macosx_10_9_x86_64.whl", hash = "sha256:0ac20581fc6cd7c754f6dff0ae06fedb060fa0e9ea6309d8be8b2701d9ea51c4"},
+    {file = "zope.interface-7.1.0-cp311-cp311-macosx_11_0_arm64.whl", hash = "sha256:848b6fa92d7c8143646e64124ed46818a0049a24ecc517958c520081fd147685"},
+    {file = "zope.interface-7.1.0-cp311-cp311-manylinux_2_5_i686.manylinux1_i686.manylinux_2_17_i686.manylinux2014_i686.whl", hash = "sha256:3bcff5c09d0215f42ba64b49205a278e44413d9bf9fa688fd9e42bfe472b5f4f"},
+    {file = "zope.interface-7.1.0-cp311-cp311-manylinux_2_5_x86_64.manylinux1_x86_64.manylinux_2_17_x86_64.manylinux2014_x86_64.whl", hash = "sha256:07add15de0cc7e69917f7d286b64d54125c950aeb43efed7a5ea7172f000fbc1"},
+    {file = "zope.interface-7.1.0-cp311-cp311-win_amd64.whl", hash = "sha256:9940d5bc441f887c5f375ec62bcf7e7e495a2d5b1da97de1184a88fb567f06af"},
+    {file = "zope.interface-7.1.0-cp312-cp312-macosx_10_9_x86_64.whl", hash = "sha256:f245d039f72e6f802902375755846f5de1ee1e14c3e8736c078565599bcab621"},
+    {file = "zope.interface-7.1.0-cp312-cp312-macosx_11_0_arm64.whl", hash = "sha256:6159e767d224d8f18deff634a1d3722e68d27488c357f62ebeb5f3e2f5288b1f"},
+    {file = "zope.interface-7.1.0-cp312-cp312-manylinux_2_5_i686.manylinux1_i686.manylinux_2_17_i686.manylinux2014_i686.whl", hash = "sha256:ff115ef91c0eeac69cd92daeba36a9d8e14daee445b504eeea2b1c0b55821984"},
+    {file = "zope.interface-7.1.0-cp312-cp312-manylinux_2_5_x86_64.manylinux1_x86_64.manylinux_2_17_x86_64.manylinux2014_x86_64.whl", hash = "sha256:bec001798ab62c3fc5447162bf48496ae9fba02edc295a9e10a0b0c639a6452e"},
+    {file = "zope.interface-7.1.0-cp312-cp312-win_amd64.whl", hash = "sha256:124149e2d42067b9c6597f4dafdc7a0983d0163868f897b7bb5dc850b14f9a87"},
+    {file = "zope.interface-7.1.0-cp313-cp313-macosx_10_9_x86_64.whl", hash = "sha256:9733a9a0f94ef53d7aa64661811b20875b5bc6039034c6e42fb9732170130573"},
+    {file = "zope.interface-7.1.0-cp313-cp313-macosx_11_0_arm64.whl", hash = "sha256:5fcf379b875c610b5a41bc8a891841533f98de0520287d7f85e25386cd10d3e9"},
+    {file = "zope.interface-7.1.0-cp313-cp313-manylinux_2_5_i686.manylinux1_i686.manylinux_2_17_i686.manylinux2014_i686.whl", hash = "sha256:4af4a12b459a273b0b34679a5c3dc5e34c1847c3dd14a628aa0668e19e638ea2"},
+    {file = "zope.interface-7.1.0-cp313-cp313-manylinux_2_5_x86_64.manylinux1_x86_64.manylinux_2_17_x86_64.manylinux2014_x86_64.whl", hash = "sha256:a735f82d2e3ed47ca01a20dfc4c779b966b16352650a8036ab3955aad151ed8a"},
+    {file = "zope.interface-7.1.0-cp313-cp313-win_amd64.whl", hash = "sha256:5501e772aff595e3c54266bc1bfc5858e8f38974ce413a8f1044aae0f32a83a3"},
+    {file = "zope.interface-7.1.0-cp38-cp38-macosx_10_9_x86_64.whl", hash = "sha256:ec59fe53db7d32abb96c6d4efeed84aab4a7c38c62d7a901a9b20c09dd936e7a"},
+    {file = "zope.interface-7.1.0-cp38-cp38-macosx_11_0_arm64.whl", hash = "sha256:e53c291debef523b09e1fe3dffe5f35dde164f1c603d77f770b88a1da34b7ed6"},
+    {file = "zope.interface-7.1.0-cp38-cp38-manylinux_2_5_i686.manylinux1_i686.manylinux_2_17_i686.manylinux2014_i686.whl", hash = "sha256:4a00ead2e24c76436e1b457a5132d87f83858330f6c923640b7ef82d668525d1"},
+    {file = "zope.interface-7.1.0-cp38-cp38-manylinux_2_5_x86_64.manylinux1_x86_64.manylinux_2_17_x86_64.manylinux2014_x86_64.whl", hash = "sha256:5e28ea0bc4b084fc93a483877653a033062435317082cdc6388dec3438309faf"},
+    {file = "zope.interface-7.1.0-cp38-cp38-win_amd64.whl", hash = "sha256:27cfb5205d68b12682b6e55ab8424662d96e8ead19550aad0796b08dd2c9a45e"},
+    {file = "zope.interface-7.1.0-cp39-cp39-macosx_10_9_x86_64.whl", hash = "sha256:9e3e48f3dea21c147e1b10c132016cb79af1159facca9736d231694ef5a740a8"},
+    {file = "zope.interface-7.1.0-cp39-cp39-macosx_11_0_arm64.whl", hash = "sha256:a99240b1d02dc469f6afbe7da1bf617645e60290c272968f4e53feec18d7dce8"},
+    {file = "zope.interface-7.1.0-cp39-cp39-manylinux_2_5_i686.manylinux1_i686.manylinux_2_17_i686.manylinux2014_i686.whl", hash = "sha256:b7b25db127db3e6b597c5f74af60309c4ad65acd826f89609662f0dc33a54728"},
+    {file = "zope.interface-7.1.0-cp39-cp39-manylinux_2_5_x86_64.manylinux1_x86_64.manylinux_2_17_x86_64.manylinux2014_x86_64.whl", hash = "sha256:2a29ac607e970b5576547f0e3589ec156e04de17af42839eedcf478450687317"},
+    {file = "zope.interface-7.1.0-cp39-cp39-win_amd64.whl", hash = "sha256:a14c9decf0eb61e0892631271d500c1e306c7b6901c998c7035e194d9150fdd1"},
+    {file = "zope_interface-7.1.0.tar.gz", hash = "sha256:3f005869a1a05e368965adb2075f97f8ee9a26c61898a9e52a9764d93774f237"},
 ]
 
 [package.dependencies]
 setuptools = "*"
 
 [package.extras]
-docs = ["Sphinx", "repoze.sphinx.autointerface", "sphinx-rtd-theme"]
-test = ["coverage (>=5.0.3)", "zope.event", "zope.testing"]
-testing = ["coverage (>=5.0.3)", "zope.event", "zope.testing"]
+docs = ["Sphinx", "furo", "repoze.sphinx.autointerface"]
+test = ["coverage[toml]", "zope.event", "zope.testing"]
+testing = ["coverage[toml]", "zope.event", "zope.testing"]
 
 [[package]]
 name = "zstandard"
@@ -10587,4 +10493,4 @@ cffi = ["cffi (>=1.11)"]
 [metadata]
 lock-version = "2.0"
 python-versions = ">=3.10,<3.13"
-content-hash = "259bffdcb68115a512695954dea8058bdc0bcc3770ce79106271ea94c6b5ed40"
+content-hash = "d324192116c4b243e504d57f4605b79c46592a976201d903b16a910b71d84b57"
diff --git a/api/pyproject.toml b/api/pyproject.toml
index b72cc44abb813..11bcc255d7cd9 100644
--- a/api/pyproject.toml
+++ b/api/pyproject.toml
@@ -210,7 +210,7 @@ numexpr = "~2.9.0"
 opensearch-py = "2.4.0"
 qrcode = "~7.4.2"
 twilio = "~9.0.4"
-vanna = { version = "0.5.5", extras = ["postgres", "mysql", "clickhouse", "duckdb"] }
+vanna = { version = "0.7.3", extras = ["postgres", "mysql", "clickhouse", "duckdb"] }
 wikipedia = "1.4.0"
 yfinance = "~0.2.40"
 

From a34891851bd3dc28c443d4255a955a3bc857e6d0 Mon Sep 17 00:00:00 2001
From: -LAN- <laipz8200@outlook.com>
Date: Fri, 11 Oct 2024 10:32:27 +0800
Subject: [PATCH 192/235] fix(log list): prevent duplicate data fetch (#9190)

---
 web/app/components/app/log/list.tsx | 8 ++++++--
 1 file changed, 6 insertions(+), 2 deletions(-)

diff --git a/web/app/components/app/log/list.tsx b/web/app/components/app/log/list.tsx
index 149e877fa4ac2..7dff48c20a28e 100644
--- a/web/app/components/app/log/list.tsx
+++ b/web/app/components/app/log/list.tsx
@@ -299,10 +299,14 @@ function DetailPanel<T extends ChatConversationFullDetailResponse | CompletionCo
     }))
   }, [items])
 
+  const fetchInitiated = useRef(false)
+
   useEffect(() => {
-    if (appDetail?.id && detail.id && appDetail?.mode !== 'completion')
+    if (appDetail?.id && detail.id && appDetail?.mode !== 'completion' && !fetchInitiated.current) {
+      fetchInitiated.current = true
       fetchData()
-  }, [appDetail?.id, detail.id, appDetail?.mode])
+    }
+  }, [appDetail?.id, detail.id, appDetail?.mode, fetchData])
 
   const isChatMode = appDetail?.mode !== 'completion'
   const isAdvanced = appDetail?.mode === 'advanced-chat'

From cabdb4ef1779a139d464b0495b7ed9874818b81c Mon Sep 17 00:00:00 2001
From: Jyong <76649700+JohnJyong@users.noreply.github.com>
Date: Fri, 11 Oct 2024 10:32:42 +0800
Subject: [PATCH 193/235] fix retrieval resource positon missed (#9155)

Co-authored-by: Bowen Liang <liangbowen@gf.com.cn>
---
 api/core/rag/entities/context_entities.py   | 4 +++-
 api/core/rag/retrieval/dataset_retrieval.py | 3 +++
 2 files changed, 6 insertions(+), 1 deletion(-)

diff --git a/api/core/rag/entities/context_entities.py b/api/core/rag/entities/context_entities.py
index dde3beccf6e46..cd18ad081ff4f 100644
--- a/api/core/rag/entities/context_entities.py
+++ b/api/core/rag/entities/context_entities.py
@@ -1,3 +1,5 @@
+from typing import Optional
+
 from pydantic import BaseModel
 
 
@@ -7,4 +9,4 @@ class DocumentContext(BaseModel):
     """
 
     content: str
-    score: float
+    score: Optional[float] = None
diff --git a/api/core/rag/retrieval/dataset_retrieval.py b/api/core/rag/retrieval/dataset_retrieval.py
index ae61ba7112243..ae79b44158c84 100644
--- a/api/core/rag/retrieval/dataset_retrieval.py
+++ b/api/core/rag/retrieval/dataset_retrieval.py
@@ -231,6 +231,9 @@ def retrieve(
                                 source["content"] = segment.content
                             retrieval_resource_list.append(source)
         if hit_callback and retrieval_resource_list:
+            retrieval_resource_list = sorted(retrieval_resource_list, key=lambda x: x.get("score"), reverse=True)
+            for position, item in enumerate(retrieval_resource_list, start=1):
+                item["position"] = position
             hit_callback.return_retriever_resource_info(retrieval_resource_list)
         if document_context_list:
             document_context_list = sorted(document_context_list, key=lambda x: x.score, reverse=True)

From 5c76131d3dc1e4a5719cc828ba5a485e97024d90 Mon Sep 17 00:00:00 2001
From: Fei He <droxer.he@gmail.com>
Date: Fri, 11 Oct 2024 10:35:56 +0800
Subject: [PATCH 194/235] feat: add gte rerank for tongyi  (#9153)

---
 .../model_providers/tongyi/rerank/__init__.py |   0
 .../tongyi/rerank/_position.yaml              |   1 +
 .../tongyi/rerank/gte-rerank.yaml             |   4 +
 .../model_providers/tongyi/rerank/rerank.py   | 136 ++++++++++++++++++
 .../model_providers/tongyi/tongyi.yaml        |   1 +
 .../model_runtime/tongyi/test_rerank.py       |  40 ++++++
 6 files changed, 182 insertions(+)
 create mode 100644 api/core/model_runtime/model_providers/tongyi/rerank/__init__.py
 create mode 100644 api/core/model_runtime/model_providers/tongyi/rerank/_position.yaml
 create mode 100644 api/core/model_runtime/model_providers/tongyi/rerank/gte-rerank.yaml
 create mode 100644 api/core/model_runtime/model_providers/tongyi/rerank/rerank.py
 create mode 100644 api/tests/integration_tests/model_runtime/tongyi/test_rerank.py

diff --git a/api/core/model_runtime/model_providers/tongyi/rerank/__init__.py b/api/core/model_runtime/model_providers/tongyi/rerank/__init__.py
new file mode 100644
index 0000000000000..e69de29bb2d1d
diff --git a/api/core/model_runtime/model_providers/tongyi/rerank/_position.yaml b/api/core/model_runtime/model_providers/tongyi/rerank/_position.yaml
new file mode 100644
index 0000000000000..439afda99263a
--- /dev/null
+++ b/api/core/model_runtime/model_providers/tongyi/rerank/_position.yaml
@@ -0,0 +1 @@
+- gte-rerank
diff --git a/api/core/model_runtime/model_providers/tongyi/rerank/gte-rerank.yaml b/api/core/model_runtime/model_providers/tongyi/rerank/gte-rerank.yaml
new file mode 100644
index 0000000000000..44d51b9b0d9cd
--- /dev/null
+++ b/api/core/model_runtime/model_providers/tongyi/rerank/gte-rerank.yaml
@@ -0,0 +1,4 @@
+model: gte-rerank
+model_type: rerank
+model_properties:
+  context_size: 4000
diff --git a/api/core/model_runtime/model_providers/tongyi/rerank/rerank.py b/api/core/model_runtime/model_providers/tongyi/rerank/rerank.py
new file mode 100644
index 0000000000000..c9245bd82ddb0
--- /dev/null
+++ b/api/core/model_runtime/model_providers/tongyi/rerank/rerank.py
@@ -0,0 +1,136 @@
+from typing import Optional
+
+import dashscope
+from dashscope.common.error import (
+    AuthenticationError,
+    InvalidParameter,
+    RequestFailure,
+    ServiceUnavailableError,
+    UnsupportedHTTPMethod,
+    UnsupportedModel,
+)
+
+from core.model_runtime.entities.rerank_entities import RerankDocument, RerankResult
+from core.model_runtime.errors.invoke import (
+    InvokeAuthorizationError,
+    InvokeBadRequestError,
+    InvokeConnectionError,
+    InvokeError,
+    InvokeRateLimitError,
+    InvokeServerUnavailableError,
+)
+from core.model_runtime.errors.validate import CredentialsValidateFailedError
+from core.model_runtime.model_providers.__base.rerank_model import RerankModel
+
+
+class GTERerankModel(RerankModel):
+    """
+    Model class for GTE rerank model.
+    """
+
+    def _invoke(
+        self,
+        model: str,
+        credentials: dict,
+        query: str,
+        docs: list[str],
+        score_threshold: Optional[float] = None,
+        top_n: Optional[int] = None,
+        user: Optional[str] = None,
+    ) -> RerankResult:
+        """
+        Invoke rerank model
+
+        :param model: model name
+        :param credentials: model credentials
+        :param query: search query
+        :param docs: docs for reranking
+        :param score_threshold: score threshold
+        :param top_n: top n
+        :param user: unique user id
+        :return: rerank result
+        """
+        if len(docs) == 0:
+            return RerankResult(model=model, docs=docs)
+
+        # initialize client
+        dashscope.api_key = credentials["dashscope_api_key"]
+
+        response = dashscope.TextReRank.call(
+            query=query,
+            documents=docs,
+            model=model,
+            top_n=top_n,
+            return_documents=True,
+        )
+
+        rerank_documents = []
+        for _, result in enumerate(response.output.results):
+            # format document
+            rerank_document = RerankDocument(
+                index=result.index,
+                score=result.relevance_score,
+                text=result["document"]["text"],
+            )
+
+            # score threshold check
+            if score_threshold is not None:
+                if result.relevance_score >= score_threshold:
+                    rerank_documents.append(rerank_document)
+            else:
+                rerank_documents.append(rerank_document)
+
+        return RerankResult(model=model, docs=rerank_documents)
+
+    def validate_credentials(self, model: str, credentials: dict) -> None:
+        """
+        Validate model credentials
+
+        :param model: model name
+        :param credentials: model credentials
+        :return:
+        """
+        try:
+            self.invoke(
+                model=model,
+                credentials=credentials,
+                query="What is the capital of the United States?",
+                docs=[
+                    "Carson City is the capital city of the American state of Nevada. At the 2010 United States "
+                    "Census, Carson City had a population of 55,274.",
+                    "The Commonwealth of the Northern Mariana Islands is a group of islands in the Pacific Ocean that "
+                    "are a political division controlled by the United States. Its capital is Saipan.",
+                ],
+                score_threshold=0.8,
+            )
+        except Exception as ex:
+            print(ex)
+            raise CredentialsValidateFailedError(str(ex))
+
+    @property
+    def _invoke_error_mapping(self) -> dict[type[InvokeError], list[type[Exception]]]:
+        """
+        Map model invoke error to unified error
+        The key is the error type thrown to the caller
+        The value is the error type thrown by the model,
+        which needs to be converted into a unified error type for the caller.
+
+        :return: Invoke error mapping
+        """
+        return {
+            InvokeConnectionError: [
+                RequestFailure,
+            ],
+            InvokeServerUnavailableError: [
+                ServiceUnavailableError,
+            ],
+            InvokeRateLimitError: [],
+            InvokeAuthorizationError: [
+                AuthenticationError,
+            ],
+            InvokeBadRequestError: [
+                InvalidParameter,
+                UnsupportedModel,
+                UnsupportedHTTPMethod,
+            ],
+        }
diff --git a/api/core/model_runtime/model_providers/tongyi/tongyi.yaml b/api/core/model_runtime/model_providers/tongyi/tongyi.yaml
index 1a09c20fd93f6..6349c22714265 100644
--- a/api/core/model_runtime/model_providers/tongyi/tongyi.yaml
+++ b/api/core/model_runtime/model_providers/tongyi/tongyi.yaml
@@ -18,6 +18,7 @@ supported_model_types:
   - llm
   - tts
   - text-embedding
+  - rerank
 configurate_methods:
   - predefined-model
   - customizable-model
diff --git a/api/tests/integration_tests/model_runtime/tongyi/test_rerank.py b/api/tests/integration_tests/model_runtime/tongyi/test_rerank.py
new file mode 100644
index 0000000000000..2dcfb92c63fee
--- /dev/null
+++ b/api/tests/integration_tests/model_runtime/tongyi/test_rerank.py
@@ -0,0 +1,40 @@
+import os
+
+import dashscope
+import pytest
+
+from core.model_runtime.entities.rerank_entities import RerankResult
+from core.model_runtime.errors.validate import CredentialsValidateFailedError
+from core.model_runtime.model_providers.tongyi.rerank.rerank import GTERerankModel
+
+
+def test_validate_credentials():
+    model = GTERerankModel()
+
+    with pytest.raises(CredentialsValidateFailedError):
+        model.validate_credentials(model="get-rank", credentials={"dashscope_api_key": "invalid_key"})
+
+    model.validate_credentials(
+        model="get-rank", credentials={"dashscope_api_key": os.environ.get("TONGYI_DASHSCOPE_API_KEY")}
+    )
+
+
+def test_invoke_model():
+    model = GTERerankModel()
+
+    result = model.invoke(
+        model=dashscope.TextReRank.Models.gte_rerank,
+        credentials={"dashscope_api_key": os.environ.get("TONGYI_DASHSCOPE_API_KEY")},
+        query="什么是文本排序模型",
+        docs=[
+            "文本排序模型广泛用于搜索引擎和推荐系统中，它们根据文本相关性对候选文本进行排序",
+            "量子计算是计算科学的一个前沿领域",
+            "预训练语言模型的发展给文本排序模型带来了新的进展",
+        ],
+        score_threshold=0.7,
+    )
+
+    assert isinstance(result, RerankResult)
+    assert len(result.docs) == 1
+    assert result.docs[0].index == 0
+    assert result.docs[0].score >= 0.7

From 8bcad002dffcf0cb3c06f6240370d331dcad6949 Mon Sep 17 00:00:00 2001
From: Wu Tianwei <30284043+WTW0313@users.noreply.github.com>
Date: Fri, 11 Oct 2024 11:22:59 +0800
Subject: [PATCH 195/235] fix: add loading to 'delete' button & 'save' button
 (#9214)

---
 web/app/components/datasets/documents/list.tsx      | 6 ++++++
 web/app/components/datasets/settings/form/index.tsx | 2 ++
 2 files changed, 8 insertions(+)

diff --git a/web/app/components/datasets/documents/list.tsx b/web/app/components/datasets/documents/list.tsx
index 540474e7a5739..0e0eebb034df5 100644
--- a/web/app/components/datasets/documents/list.tsx
+++ b/web/app/components/datasets/documents/list.tsx
@@ -122,6 +122,7 @@ export const OperationAction: FC<{
 }> = ({ embeddingAvailable, datasetId, detail, onUpdate, scene = 'list', className = '' }) => {
   const { id, enabled = false, archived = false, data_source_type } = detail || {}
   const [showModal, setShowModal] = useState(false)
+  const [deleting, setDeleting] = useState(false)
   const { notify } = useContext(ToastContext)
   const { t } = useTranslation()
   const router = useRouter()
@@ -153,6 +154,7 @@ export const OperationAction: FC<{
         break
       default:
         opApi = deleteDocument
+        setDeleting(true)
         break
     }
     const [e] = await asyncRunSafe<CommonResponse>(opApi({ datasetId, documentId: id }) as Promise<CommonResponse>)
@@ -160,6 +162,8 @@ export const OperationAction: FC<{
       notify({ type: 'success', message: t('common.actionMsg.modifiedSuccessfully') })
     else
       notify({ type: 'error', message: t('common.actionMsg.modifiedUnsuccessfully') })
+    if (operationName === 'delete')
+      setDeleting(false)
     onUpdate(operationName)
   }
 
@@ -295,6 +299,8 @@ export const OperationAction: FC<{
     {showModal
       && <Confirm
         isShow={showModal}
+        isLoading={deleting}
+        isDisabled={deleting}
         title={t('datasetDocuments.list.delete.title')}
         content={t('datasetDocuments.list.delete.content')}
         confirmText={t('common.operation.sure')}
diff --git a/web/app/components/datasets/settings/form/index.tsx b/web/app/components/datasets/settings/form/index.tsx
index 0bdfa3c5c883f..fa8c8de62ec28 100644
--- a/web/app/components/datasets/settings/form/index.tsx
+++ b/web/app/components/datasets/settings/form/index.tsx
@@ -335,6 +335,8 @@ const Form = () => {
           <Button
             className='min-w-24'
             variant='primary'
+            loading={loading}
+            disabled={loading}
             onClick={handleSave}
           >
             {t('datasetSettings.form.save')}

From e0c8189f1ab6985598c93abc5374f21c5c89236d Mon Sep 17 00:00:00 2001
From: crazywoola <100913391+crazywoola@users.noreply.github.com>
Date: Fri, 11 Oct 2024 11:28:09 +0800
Subject: [PATCH 196/235] feat: add NotFound error for dataset service (#9215)

---
 api/services/dataset_service.py | 3 +++
 1 file changed, 3 insertions(+)

diff --git a/api/services/dataset_service.py b/api/services/dataset_service.py
index b8f80a9f77e2a..4f77fdc18cb43 100644
--- a/api/services/dataset_service.py
+++ b/api/services/dataset_service.py
@@ -8,6 +8,7 @@
 
 from flask_login import current_user
 from sqlalchemy import func
+from werkzeug.exceptions import NotFound
 
 from configs import dify_config
 from core.errors.error import LLMBadRequestError, ProviderTokenNotInitError
@@ -975,6 +976,8 @@ def update_document_with_dataset_id(
     ):
         DatasetService.check_dataset_model_setting(dataset)
         document = DocumentService.get_document(dataset.id, document_data["original_document_id"])
+        if document is None:
+            raise NotFound("Document not found")
         if document.display_status != "available":
             raise ValueError("Document is not available")
         # update document name

From 1e7ef46e9c8754c4953cdd597271efb31f81a223 Mon Sep 17 00:00:00 2001
From: Joe <79627742+ZhouhaoJiang@users.noreply.github.com>
Date: Fri, 11 Oct 2024 11:35:01 +0800
Subject: [PATCH 197/235] fix: update inner api proxies (#9174)

---
 api/services/enterprise/base.py | 7 ++++++-
 1 file changed, 6 insertions(+), 1 deletion(-)

diff --git a/api/services/enterprise/base.py b/api/services/enterprise/base.py
index ddee52164b746..7d4fdfd2d039d 100644
--- a/api/services/enterprise/base.py
+++ b/api/services/enterprise/base.py
@@ -7,11 +7,16 @@ class EnterpriseRequest:
     base_url = os.environ.get("ENTERPRISE_API_URL", "ENTERPRISE_API_URL")
     secret_key = os.environ.get("ENTERPRISE_API_SECRET_KEY", "ENTERPRISE_API_SECRET_KEY")
 
+    proxies = {
+        "http": None,
+        "https": None,
+    }
+
     @classmethod
     def send_request(cls, method, endpoint, json=None, params=None):
         headers = {"Content-Type": "application/json", "Enterprise-Api-Secret-Key": cls.secret_key}
 
         url = f"{cls.base_url}{endpoint}"
-        response = requests.request(method, url, json=json, params=params, headers=headers)
+        response = requests.request(method, url, json=json, params=params, headers=headers, proxies=cls.proxies)
 
         return response.json()

From ccc6723a8e43b129363f59357dfb4ca63cb91c26 Mon Sep 17 00:00:00 2001
From: massif-01 <massif0601@gmail.com>
Date: Fri, 11 Oct 2024 11:55:16 +0800
Subject: [PATCH 198/235] Update docker-compose.yaml (#9183)

Co-authored-by: crazywoola <427733928@qq.com>
---
 docker/docker-compose.yaml | 1 +
 1 file changed, 1 insertion(+)

diff --git a/docker/docker-compose.yaml b/docker/docker-compose.yaml
index 316a6fb64e64c..becc45339bb04 100644
--- a/docker/docker-compose.yaml
+++ b/docker/docker-compose.yaml
@@ -209,6 +209,7 @@ x-shared-env: &shared-api-worker-env
   SSRF_PROXY_HTTPS_URL: ${SSRF_PROXY_HTTPS_URL:-http://ssrf_proxy:3128}
   HTTP_REQUEST_NODE_MAX_BINARY_SIZE: ${HTTP_REQUEST_NODE_MAX_BINARY_SIZE:-10485760}
   HTTP_REQUEST_NODE_MAX_TEXT_SIZE: ${HTTP_REQUEST_NODE_MAX_TEXT_SIZE:-1048576}
+  APP_MAX_EXECUTION_TIME: ${APP_MAX_EXECUTION_TIME:-12000}
 
 services:
   # API service

From fe41e8bc18a0338de9cb6e9373826edf782894d9 Mon Sep 17 00:00:00 2001
From: AAEE86 <33052466+AAEE86@users.noreply.github.com>
Date: Fri, 11 Oct 2024 11:56:11 +0800
Subject: [PATCH 199/235] feat: add siliconflow custom add model interface
 (#8745)

---
 .../model_providers/siliconflow/llm/llm.py    | 62 ++++++++++++++++++-
 .../siliconflow/siliconflow.yaml              | 55 ++++++++++++++++
 2 files changed, 116 insertions(+), 1 deletion(-)

diff --git a/api/core/model_runtime/model_providers/siliconflow/llm/llm.py b/api/core/model_runtime/model_providers/siliconflow/llm/llm.py
index c1868b6ad02b8..4f8f4e0f61c70 100644
--- a/api/core/model_runtime/model_providers/siliconflow/llm/llm.py
+++ b/api/core/model_runtime/model_providers/siliconflow/llm/llm.py
@@ -1,8 +1,18 @@
 from collections.abc import Generator
 from typing import Optional, Union
 
-from core.model_runtime.entities.llm_entities import LLMResult
+from core.model_runtime.entities.common_entities import I18nObject
+from core.model_runtime.entities.llm_entities import LLMMode, LLMResult
 from core.model_runtime.entities.message_entities import PromptMessage, PromptMessageTool
+from core.model_runtime.entities.model_entities import (
+    AIModelEntity,
+    FetchFrom,
+    ModelFeature,
+    ModelPropertyKey,
+    ModelType,
+    ParameterRule,
+    ParameterType,
+)
 from core.model_runtime.model_providers.openai_api_compatible.llm.llm import OAIAPICompatLargeLanguageModel
 
 
@@ -29,3 +39,53 @@ def validate_credentials(self, model: str, credentials: dict) -> None:
     def _add_custom_parameters(cls, credentials: dict) -> None:
         credentials["mode"] = "chat"
         credentials["endpoint_url"] = "https://api.siliconflow.cn/v1"
+
+    def get_customizable_model_schema(self, model: str, credentials: dict) -> AIModelEntity | None:
+        return AIModelEntity(
+            model=model,
+            label=I18nObject(en_US=model, zh_Hans=model),
+            model_type=ModelType.LLM,
+            features=[ModelFeature.TOOL_CALL, ModelFeature.MULTI_TOOL_CALL, ModelFeature.STREAM_TOOL_CALL]
+            if credentials.get("function_calling_type") == "tool_call"
+            else [],
+            fetch_from=FetchFrom.CUSTOMIZABLE_MODEL,
+            model_properties={
+                ModelPropertyKey.CONTEXT_SIZE: int(credentials.get("context_size", 8000)),
+                ModelPropertyKey.MODE: LLMMode.CHAT.value,
+            },
+            parameter_rules=[
+                ParameterRule(
+                    name="temperature",
+                    use_template="temperature",
+                    label=I18nObject(en_US="Temperature", zh_Hans="温度"),
+                    type=ParameterType.FLOAT,
+                ),
+                ParameterRule(
+                    name="max_tokens",
+                    use_template="max_tokens",
+                    default=512,
+                    min=1,
+                    max=int(credentials.get("max_tokens", 1024)),
+                    label=I18nObject(en_US="Max Tokens", zh_Hans="最大标记"),
+                    type=ParameterType.INT,
+                ),
+                ParameterRule(
+                    name="top_p",
+                    use_template="top_p",
+                    label=I18nObject(en_US="Top P", zh_Hans="Top P"),
+                    type=ParameterType.FLOAT,
+                ),
+                ParameterRule(
+                    name="top_k",
+                    use_template="top_k",
+                    label=I18nObject(en_US="Top K", zh_Hans="Top K"),
+                    type=ParameterType.FLOAT,
+                ),
+                ParameterRule(
+                    name="frequency_penalty",
+                    use_template="frequency_penalty",
+                    label=I18nObject(en_US="Frequency Penalty", zh_Hans="重复惩罚"),
+                    type=ParameterType.FLOAT,
+                ),
+            ],
+        )
diff --git a/api/core/model_runtime/model_providers/siliconflow/siliconflow.yaml b/api/core/model_runtime/model_providers/siliconflow/siliconflow.yaml
index c46a891604c48..71f9a9238145c 100644
--- a/api/core/model_runtime/model_providers/siliconflow/siliconflow.yaml
+++ b/api/core/model_runtime/model_providers/siliconflow/siliconflow.yaml
@@ -20,6 +20,7 @@ supported_model_types:
   - speech2text
 configurate_methods:
   - predefined-model
+  - customizable-model
 provider_credential_schema:
   credential_form_schemas:
     - variable: api_key
@@ -30,3 +31,57 @@ provider_credential_schema:
       placeholder:
         zh_Hans: 在此输入您的 API Key
         en_US: Enter your API Key
+model_credential_schema:
+  model:
+    label:
+      en_US: Model Name
+      zh_Hans: 模型名称
+    placeholder:
+      en_US: Enter your model name
+      zh_Hans: 输入模型名称
+  credential_form_schemas:
+    - variable: api_key
+      label:
+        en_US: API Key
+      type: secret-input
+      required: true
+      placeholder:
+        zh_Hans: 在此输入您的 API Key
+        en_US: Enter your API Key
+    - variable: context_size
+      label:
+        zh_Hans: 模型上下文长度
+        en_US: Model context size
+      required: true
+      type: text-input
+      default: '4096'
+      placeholder:
+        zh_Hans: 在此输入您的模型上下文长度
+        en_US: Enter your Model context size
+    - variable: max_tokens
+      label:
+        zh_Hans: 最大 token 上限
+        en_US: Upper bound for max tokens
+      default: '4096'
+      type: text-input
+      show_on:
+        - variable: __model_type
+          value: llm
+    - variable: function_calling_type
+      label:
+        en_US: Function calling
+      type: select
+      required: false
+      default: no_call
+      options:
+        - value: no_call
+          label:
+            en_US: Not Support
+            zh_Hans: 不支持
+        - value: function_call
+          label:
+            en_US: Support
+            zh_Hans: 支持
+      show_on:
+        - variable: __model_type
+          value: llm

From 1c1e008dcf97400108b915b59a0b2bfe6f32d269 Mon Sep 17 00:00:00 2001
From: Hash Brown <hi@xzd.me>
Date: Fri, 11 Oct 2024 12:07:39 +0800
Subject: [PATCH 200/235] fix: reduce webapp icon displayed on browser tab
 flickering when page is loading (#9212)

---
 web/app/(shareLayout)/layout.tsx | 5 +++++
 web/hooks/use-app-favicon.ts     | 2 +-
 2 files changed, 6 insertions(+), 1 deletion(-)

diff --git a/web/app/(shareLayout)/layout.tsx b/web/app/(shareLayout)/layout.tsx
index 9c4632cd450df..259af2bc2dc84 100644
--- a/web/app/(shareLayout)/layout.tsx
+++ b/web/app/(shareLayout)/layout.tsx
@@ -1,7 +1,12 @@
 import React from 'react'
 import type { FC } from 'react'
+import type { Metadata } from 'next'
 import GA, { GaType } from '@/app/components/base/ga'
 
+export const metadata: Metadata = {
+  icons: 'data:,', // prevent browser from using default favicon
+}
+
 const Layout: FC<{
   children: React.ReactNode
 }> = ({ children }) => {
diff --git a/web/hooks/use-app-favicon.ts b/web/hooks/use-app-favicon.ts
index 1ff743928faae..e8a01733712a6 100644
--- a/web/hooks/use-app-favicon.ts
+++ b/web/hooks/use-app-favicon.ts
@@ -21,7 +21,7 @@ export function useAppFavicon(options: UseAppFaviconOptions) {
   } = options
 
   useAsyncEffect(async () => {
-    if (!enable)
+    if (!enable || (icon_type === 'image' && !icon_url) || (icon_type === 'emoji' && !icon))
       return
 
     const isValidImageIcon = icon_type === 'image' && icon_url

From d498f4e55e64f6918695e7381626d3a16b704344 Mon Sep 17 00:00:00 2001
From: zxhlyh <jasonapring2015@outlook.com>
Date: Fri, 11 Oct 2024 12:33:34 +0800
Subject: [PATCH 201/235] Chore/model icon url (#9218)

Co-authored-by: takatost <takatost@gmail.com>
Co-authored-by: jyong <718720800@qq.com>
---
 api/controllers/console/workspace/model_providers.py       | 7 +++----
 .../model-provider-page/model-icon/index.tsx               | 2 +-
 .../model-provider-page/provider-icon/index.tsx            | 2 +-
 3 files changed, 5 insertions(+), 6 deletions(-)

diff --git a/api/controllers/console/workspace/model_providers.py b/api/controllers/console/workspace/model_providers.py
index fe0bcf73384ed..9e8a53bbfbf2a 100644
--- a/api/controllers/console/workspace/model_providers.py
+++ b/api/controllers/console/workspace/model_providers.py
@@ -126,13 +126,12 @@ class ModelProviderIconApi(Resource):
     Get model provider icon
     """
 
-    @setup_required
-    @login_required
-    @account_initialization_required
     def get(self, provider: str, icon_type: str, lang: str):
         model_provider_service = ModelProviderService()
         icon, mimetype = model_provider_service.get_model_provider_icon(
-            provider=provider, icon_type=icon_type, lang=lang
+            provider=provider,
+            icon_type=icon_type,
+            lang=lang,
         )
 
         return send_file(io.BytesIO(icon), mimetype=mimetype)
diff --git a/web/app/components/header/account-setting/model-provider-page/model-icon/index.tsx b/web/app/components/header/account-setting/model-provider-page/model-icon/index.tsx
index a22ec16c25228..a16b101e6a57c 100644
--- a/web/app/components/header/account-setting/model-provider-page/model-icon/index.tsx
+++ b/web/app/components/header/account-setting/model-provider-page/model-icon/index.tsx
@@ -26,7 +26,7 @@ const ModelIcon: FC<ModelIconProps> = ({
     return (
       <img
         alt='model-icon'
-        src={`${provider.icon_small[language] || provider.icon_small.en_US}?_token=${localStorage.getItem('console_token')}`}
+        src={`${provider.icon_small[language] || provider.icon_small.en_US}`}
         className={`w-4 h-4 ${className}`}
       />
     )
diff --git a/web/app/components/header/account-setting/model-provider-page/provider-icon/index.tsx b/web/app/components/header/account-setting/model-provider-page/provider-icon/index.tsx
index c618b6f1a970d..768f2c2766d6f 100644
--- a/web/app/components/header/account-setting/model-provider-page/provider-icon/index.tsx
+++ b/web/app/components/header/account-setting/model-provider-page/provider-icon/index.tsx
@@ -16,7 +16,7 @@ const ProviderIcon: FC<ProviderIconProps> = ({
     return (
       <img
         alt='provider-icon'
-        src={`${provider.icon_large[language] || provider.icon_large.en_US}?_token=${localStorage.getItem('console_token')}`}
+        src={`${provider.icon_large[language] || provider.icon_large.en_US}`}
         className={`w-auto h-6 ${className}`}
       />
     )

From 80b62d50f5cc034c5df2dcd1f1b9c43da5b8289b Mon Sep 17 00:00:00 2001
From: Jyong <76649700+JohnJyong@users.noreply.github.com>
Date: Fri, 11 Oct 2024 16:04:23 +0800
Subject: [PATCH 202/235] Fix/add es num_candidates (#9225)

---
 .../rag/datasource/vdb/elasticsearch/elasticsearch_vector.py  | 4 +++-
 1 file changed, 3 insertions(+), 1 deletion(-)

diff --git a/api/core/rag/datasource/vdb/elasticsearch/elasticsearch_vector.py b/api/core/rag/datasource/vdb/elasticsearch/elasticsearch_vector.py
index 8d578551209e0..f585e12b2e99c 100644
--- a/api/core/rag/datasource/vdb/elasticsearch/elasticsearch_vector.py
+++ b/api/core/rag/datasource/vdb/elasticsearch/elasticsearch_vector.py
@@ -1,5 +1,6 @@
 import json
 import logging
+import math
 from typing import Any, Optional
 from urllib.parse import urlparse
 
@@ -112,7 +113,8 @@ def delete(self) -> None:
 
     def search_by_vector(self, query_vector: list[float], **kwargs: Any) -> list[Document]:
         top_k = kwargs.get("top_k", 10)
-        knn = {"field": Field.VECTOR.value, "query_vector": query_vector, "k": top_k}
+        num_candidates = math.ceil(top_k * 1.5)
+        knn = {"field": Field.VECTOR.value, "query_vector": query_vector, "k": top_k, "num_candidates": num_candidates}
 
         results = self._client.search(index=self._collection_name, knn=knn, size=top_k)
 

From 7c6ae96a095cbb1ee55e5693710c65ee32df4dbb Mon Sep 17 00:00:00 2001
From: Bowen Liang <liangbowen@gf.com.cn>
Date: Fri, 11 Oct 2024 16:08:36 +0800
Subject: [PATCH 203/235] chore: check all dependencies groups in
 pyproject.toml (#9224)

---
 .../dependencies/test_dependencies_sorted.py  | 40 +++++++------------
 1 file changed, 14 insertions(+), 26 deletions(-)

diff --git a/api/tests/artifact_tests/dependencies/test_dependencies_sorted.py b/api/tests/artifact_tests/dependencies/test_dependencies_sorted.py
index 518cee1a3ad2d..64f2884c4b828 100644
--- a/api/tests/artifact_tests/dependencies/test_dependencies_sorted.py
+++ b/api/tests/artifact_tests/dependencies/test_dependencies_sorted.py
@@ -2,36 +2,23 @@
 
 import toml
 
-ALL_DEPENDENCY_GROUP_NAMES = [
-    # default main group
-    "",
-    # required groups
-    "indirect",
-    "storage",
-    "tools",
-    "vdb",
-    # optional groups
-    "dev",
-    "lint",
-]
-
 
 def load_api_poetry_configs() -> dict[str, Any]:
     pyproject_toml = toml.load("api/pyproject.toml")
-    return pyproject_toml.get("tool").get("poetry")
+    return pyproject_toml["tool"]["poetry"]
 
 
-def load_dependency_groups() -> dict[str, dict[str, dict[str, Any]]]:
-    poetry_configs = load_api_poetry_configs()
-    group_name_to_dependencies = {
-        group_name: (poetry_configs.get("group").get(group_name) if group_name else poetry_configs).get("dependencies")
-        for group_name in ALL_DEPENDENCY_GROUP_NAMES
-    }
-    return group_name_to_dependencies
+def load_all_dependency_groups() -> dict[str, dict[str, dict[str, Any]]]:
+    configs = load_api_poetry_configs()
+    configs_by_group = {"main": configs}
+    for group_name in configs["group"]:
+        configs_by_group[group_name] = configs["group"][group_name]
+    dependencies_by_group = {group_name: base["dependencies"] for group_name, base in configs_by_group.items()}
+    return dependencies_by_group
 
 
 def test_group_dependencies_sorted():
-    for group_name, dependencies in load_dependency_groups().items():
+    for group_name, dependencies in load_all_dependency_groups().items():
         dependency_names = list(dependencies.keys())
         expected_dependency_names = sorted(set(dependency_names))
         section = f"tool.poetry.group.{group_name}.dependencies" if group_name else "tool.poetry.dependencies"
@@ -42,17 +29,18 @@ def test_group_dependencies_sorted():
 
 
 def test_group_dependencies_version_operator():
-    for group_name, dependencies in load_dependency_groups().items():
+    for group_name, dependencies in load_all_dependency_groups().items():
         for dependency_name, specification in dependencies.items():
-            version_spec = specification if isinstance(specification, str) else specification.get("version")
+            version_spec = specification if isinstance(specification, str) else specification["version"]
             assert not version_spec.startswith("^"), (
-                f"'^' is not allowed in dependency version," f" but found in '{dependency_name} = {version_spec}'"
+                f"Please replace '{dependency_name} = {version_spec}' with '{dependency_name} = ~{version_spec[1:]}' "
+                f"'^' operator is too wide and not allowed in the version specification."
             )
 
 
 def test_duplicated_dependency_crossing_groups():
     all_dependency_names: list[str] = []
-    for dependencies in load_dependency_groups().values():
+    for dependencies in load_all_dependency_groups().values():
         dependency_names = list(dependencies.keys())
         all_dependency_names.extend(dependency_names)
     expected_all_dependency_names = set(all_dependency_names)

From f4ce08211d3c614f6fbffb47d55e3a96678fb78d Mon Sep 17 00:00:00 2001
From: NFish <douxc512@gmail.com>
Date: Fri, 11 Oct 2024 16:14:56 +0800
Subject: [PATCH 204/235] feat: support csp (#9111)

Co-authored-by: Joel <iamjoel007@gmail.com>
---
 docker/.env.example                      |  4 +-
 docker/docker-compose.yaml               | 63 ++++++++++++--------
 web/.env.example                         |  3 +
 web/app/components/base/ga/index.tsx     | 11 +++-
 web/app/components/base/topbar/index.tsx | 16 -----
 web/app/layout.tsx                       |  2 -
 web/docker/entrypoint.sh                 |  1 +
 web/middleware.ts                        | 76 ++++++++++++++++++++++++
 web/package.json                         |  1 -
 web/yarn.lock                            | 33 +++++-----
 10 files changed, 148 insertions(+), 62 deletions(-)
 delete mode 100644 web/app/components/base/topbar/index.tsx
 create mode 100644 web/middleware.ts

diff --git a/docker/.env.example b/docker/.env.example
index eb05f7aa4f0b2..87d7709a1830a 100644
--- a/docker/.env.example
+++ b/docker/.env.example
@@ -797,4 +797,6 @@ POSITION_TOOL_EXCLUDES=
 # Example: POSITION_PROVIDER_PINS=openai,openllm
 POSITION_PROVIDER_PINS=
 POSITION_PROVIDER_INCLUDES=
-POSITION_PROVIDER_EXCLUDES=
\ No newline at end of file
+POSITION_PROVIDER_EXCLUDES=
+# CSP https://developer.mozilla.org/en-US/docs/Web/HTTP/CSP
+CSP_WHITELIST=
\ No newline at end of file
diff --git a/docker/docker-compose.yaml b/docker/docker-compose.yaml
index becc45339bb04..62d798a695967 100644
--- a/docker/docker-compose.yaml
+++ b/docker/docker-compose.yaml
@@ -261,6 +261,7 @@ services:
       SENTRY_DSN: ${WEB_SENTRY_DSN:-}
       NEXT_TELEMETRY_DISABLED: ${NEXT_TELEMETRY_DISABLED:-0}
       TEXT_GENERATION_TIMEOUT_MS: ${TEXT_GENERATION_TIMEOUT_MS:-60000}
+      CSP_WHITELIST: ${CSP_WHITELIST:-}
 
   # The postgres database.
   db:
@@ -280,7 +281,7 @@ services:
     volumes:
       - ./volumes/db/data:/var/lib/postgresql/data
     healthcheck:
-      test: [ "CMD", "pg_isready" ]
+      test: ['CMD', 'pg_isready']
       interval: 1s
       timeout: 3s
       retries: 30
@@ -295,7 +296,7 @@ services:
     # Set the redis password when startup redis server.
     command: redis-server --requirepass ${REDIS_PASSWORD:-difyai123456}
     healthcheck:
-      test: [ "CMD", "redis-cli", "ping" ]
+      test: ['CMD', 'redis-cli', 'ping']
 
   # The DifySandbox
   sandbox:
@@ -315,7 +316,7 @@ services:
     volumes:
       - ./volumes/sandbox/dependencies:/dependencies
     healthcheck:
-      test: [ "CMD", "curl", "-f", "http://localhost:8194/health" ]
+      test: ['CMD', 'curl', '-f', 'http://localhost:8194/health']
     networks:
       - ssrf_proxy_network
 
@@ -328,7 +329,12 @@ services:
     volumes:
       - ./ssrf_proxy/squid.conf.template:/etc/squid/squid.conf.template
       - ./ssrf_proxy/docker-entrypoint.sh:/docker-entrypoint-mount.sh
-    entrypoint: [ "sh", "-c", "cp /docker-entrypoint-mount.sh /docker-entrypoint.sh && sed -i 's/\r$$//' /docker-entrypoint.sh && chmod +x /docker-entrypoint.sh && /docker-entrypoint.sh" ]
+    entrypoint:
+      [
+        'sh',
+        '-c',
+        "cp /docker-entrypoint-mount.sh /docker-entrypoint.sh && sed -i 's/\r$$//' /docker-entrypoint.sh && chmod +x /docker-entrypoint.sh && /docker-entrypoint.sh",
+      ]
     environment:
       # pls clearly modify the squid env vars to fit your network environment.
       HTTP_PORT: ${SSRF_HTTP_PORT:-3128}
@@ -357,8 +363,8 @@ services:
       - CERTBOT_EMAIL=${CERTBOT_EMAIL}
       - CERTBOT_DOMAIN=${CERTBOT_DOMAIN}
       - CERTBOT_OPTIONS=${CERTBOT_OPTIONS:-}
-    entrypoint: [ "/docker-entrypoint.sh" ]
-    command: [ "tail", "-f", "/dev/null" ]
+    entrypoint: ['/docker-entrypoint.sh']
+    command: ['tail', '-f', '/dev/null']
 
   # The nginx reverse proxy.
   # used for reverse proxying the API service and Web service.
@@ -375,7 +381,12 @@ services:
       - ./volumes/certbot/conf/live:/etc/letsencrypt/live # cert dir (with certbot container)
       - ./volumes/certbot/conf:/etc/letsencrypt
       - ./volumes/certbot/www:/var/www/html
-    entrypoint: [ "sh", "-c", "cp /docker-entrypoint-mount.sh /docker-entrypoint.sh && sed -i 's/\r$$//' /docker-entrypoint.sh && chmod +x /docker-entrypoint.sh && /docker-entrypoint.sh" ]
+    entrypoint:
+      [
+        'sh',
+        '-c',
+        "cp /docker-entrypoint-mount.sh /docker-entrypoint.sh && sed -i 's/\r$$//' /docker-entrypoint.sh && chmod +x /docker-entrypoint.sh && /docker-entrypoint.sh",
+      ]
     environment:
       NGINX_SERVER_NAME: ${NGINX_SERVER_NAME:-_}
       NGINX_HTTPS_ENABLED: ${NGINX_HTTPS_ENABLED:-false}
@@ -397,14 +408,14 @@ services:
       - api
       - web
     ports:
-      - "${EXPOSE_NGINX_PORT:-80}:${NGINX_PORT:-80}"
-      - "${EXPOSE_NGINX_SSL_PORT:-443}:${NGINX_SSL_PORT:-443}"
+      - '${EXPOSE_NGINX_PORT:-80}:${NGINX_PORT:-80}'
+      - '${EXPOSE_NGINX_SSL_PORT:-443}:${NGINX_SSL_PORT:-443}'
 
   # The Weaviate vector store.
   weaviate:
     image: semitechnologies/weaviate:1.19.0
     profiles:
-      - ""
+      - ''
       - weaviate
     restart: always
     volumes:
@@ -453,7 +464,7 @@ services:
     volumes:
       - ./volumes/pgvector/data:/var/lib/postgresql/data
     healthcheck:
-      test: [ "CMD", "pg_isready" ]
+      test: ['CMD', 'pg_isready']
       interval: 1s
       timeout: 3s
       retries: 30
@@ -475,7 +486,7 @@ services:
     volumes:
       - ./volumes/pgvecto_rs/data:/var/lib/postgresql/data
     healthcheck:
-      test: [ "CMD", "pg_isready" ]
+      test: ['CMD', 'pg_isready']
       interval: 1s
       timeout: 3s
       retries: 30
@@ -523,7 +534,7 @@ services:
       - ./volumes/milvus/etcd:/etcd
     command: etcd -advertise-client-urls=http://127.0.0.1:2379 -listen-client-urls http://0.0.0.0:2379 --data-dir /etcd
     healthcheck:
-      test: [ "CMD", "etcdctl", "endpoint", "health" ]
+      test: ['CMD', 'etcdctl', 'endpoint', 'health']
       interval: 30s
       timeout: 20s
       retries: 3
@@ -542,7 +553,7 @@ services:
       - ./volumes/milvus/minio:/minio_data
     command: minio server /minio_data --console-address ":9001"
     healthcheck:
-      test: [ "CMD", "curl", "-f", "http://localhost:9000/minio/health/live" ]
+      test: ['CMD', 'curl', '-f', 'http://localhost:9000/minio/health/live']
       interval: 30s
       timeout: 20s
       retries: 3
@@ -554,7 +565,7 @@ services:
     image: milvusdb/milvus:v2.3.1
     profiles:
       - milvus
-    command: [ "milvus", "run", "standalone" ]
+    command: ['milvus', 'run', 'standalone']
     environment:
       ETCD_ENDPOINTS: ${ETCD_ENDPOINTS:-etcd:2379}
       MINIO_ADDRESS: ${MINIO_ADDRESS:-minio:9000}
@@ -562,7 +573,7 @@ services:
     volumes:
       - ./volumes/milvus/milvus:/var/lib/milvus
     healthcheck:
-      test: [ "CMD", "curl", "-f", "http://localhost:9091/healthz" ]
+      test: ['CMD', 'curl', '-f', 'http://localhost:9091/healthz']
       interval: 30s
       start_period: 90s
       timeout: 20s
@@ -644,13 +655,13 @@ services:
       node.name: dify-es0
       discovery.type: single-node
       xpack.license.self_generated.type: trial
-      xpack.security.enabled: "true"
-      xpack.security.enrollment.enabled: "false"
-      xpack.security.http.ssl.enabled: "false"
+      xpack.security.enabled: 'true'
+      xpack.security.enrollment.enabled: 'false'
+      xpack.security.http.ssl.enabled: 'false'
     ports:
       - ${ELASTICSEARCH_PORT:-9200}:9200
     healthcheck:
-      test: [ "CMD", "curl", "-s", "http://localhost:9200/_cluster/health?pretty" ]
+      test: ['CMD', 'curl', '-s', 'http://localhost:9200/_cluster/health?pretty']
       interval: 30s
       timeout: 10s
       retries: 50
@@ -668,17 +679,17 @@ services:
     environment:
       XPACK_ENCRYPTEDSAVEDOBJECTS_ENCRYPTIONKEY: d1a66dfd-c4d3-4a0a-8290-2abcb83ab3aa
       NO_PROXY: localhost,127.0.0.1,elasticsearch,kibana
-      XPACK_SECURITY_ENABLED: "true"
-      XPACK_SECURITY_ENROLLMENT_ENABLED: "false"
-      XPACK_SECURITY_HTTP_SSL_ENABLED: "false"
-      XPACK_FLEET_ISAIRGAPPED: "true"
+      XPACK_SECURITY_ENABLED: 'true'
+      XPACK_SECURITY_ENROLLMENT_ENABLED: 'false'
+      XPACK_SECURITY_HTTP_SSL_ENABLED: 'false'
+      XPACK_FLEET_ISAIRGAPPED: 'true'
       I18N_LOCALE: zh-CN
-      SERVER_PORT: "5601"
+      SERVER_PORT: '5601'
       ELASTICSEARCH_HOSTS: http://elasticsearch:9200
     ports:
       - ${KIBANA_PORT:-5601}:5601
     healthcheck:
-      test: [ "CMD-SHELL", "curl -s http://localhost:5601 >/dev/null || exit 1" ]
+      test: ['CMD-SHELL', 'curl -s http://localhost:5601 >/dev/null || exit 1']
       interval: 30s
       timeout: 10s
       retries: 3
diff --git a/web/.env.example b/web/.env.example
index 8e254082b385f..13ea01a2c7b12 100644
--- a/web/.env.example
+++ b/web/.env.example
@@ -22,3 +22,6 @@ NEXT_PUBLIC_UPLOAD_IMAGE_AS_ICON=false
 
 # The timeout for the text generation in millisecond
 NEXT_PUBLIC_TEXT_GENERATION_TIMEOUT_MS=60000
+
+# CSP https://developer.mozilla.org/en-US/docs/Web/HTTP/CSP
+NEXT_PUBLIC_CSP_WHITELIST=
diff --git a/web/app/components/base/ga/index.tsx b/web/app/components/base/ga/index.tsx
index ec0089ff70efc..219724113f681 100644
--- a/web/app/components/base/ga/index.tsx
+++ b/web/app/components/base/ga/index.tsx
@@ -1,6 +1,7 @@
 import type { FC } from 'react'
 import React from 'react'
 import Script from 'next/script'
+import { headers } from 'next/headers'
 import { IS_CE_EDITION } from '@/config'
 
 export enum GaType {
@@ -23,9 +24,16 @@ const GA: FC<IGAProps> = ({
   if (IS_CE_EDITION)
     return null
 
+  const nonce = process.env.NODE_ENV === 'production' ? headers().get('x-nonce') : ''
+
   return (
     <>
-      <Script strategy="beforeInteractive" async src={`https://www.googletagmanager.com/gtag/js?id=${gaIdMaps[gaType]}`}></Script>
+      <Script
+        strategy="beforeInteractive"
+        async
+        src={`https://www.googletagmanager.com/gtag/js?id=${gaIdMaps[gaType]}`}
+        nonce={nonce!}
+      ></Script>
       <Script
         id="ga-init"
         dangerouslySetInnerHTML={{
@@ -36,6 +44,7 @@ gtag('js', new Date());
 gtag('config', '${gaIdMaps[gaType]}');
           `,
         }}
+        nonce={nonce!}
       >
       </Script>
     </>
diff --git a/web/app/components/base/topbar/index.tsx b/web/app/components/base/topbar/index.tsx
deleted file mode 100644
index cf67456bd3423..0000000000000
--- a/web/app/components/base/topbar/index.tsx
+++ /dev/null
@@ -1,16 +0,0 @@
-'use client'
-
-import { AppProgressBar as ProgressBar } from 'next-nprogress-bar'
-
-const Topbar = () => {
-  return (
-    <>
-      <ProgressBar
-        height='2px'
-        color="#1C64F2FF"
-        options={{ showSpinner: false }}
-        shallowRouting />
-    </>)
-}
-
-export default Topbar
diff --git a/web/app/layout.tsx b/web/app/layout.tsx
index e9242edfade52..48e35c50e065d 100644
--- a/web/app/layout.tsx
+++ b/web/app/layout.tsx
@@ -2,7 +2,6 @@ import type { Viewport } from 'next'
 import I18nServer from './components/i18n-server'
 import BrowserInitor from './components/browser-initor'
 import SentryInitor from './components/sentry-initor'
-import Topbar from './components/base/topbar'
 import { getLocaleOnServer } from '@/i18n/server'
 import './styles/globals.css'
 import './styles/markdown.scss'
@@ -45,7 +44,6 @@ const LocaleLayout = ({
         data-public-site-about={process.env.NEXT_PUBLIC_SITE_ABOUT}
         data-public-text-generation-timeout-ms={process.env.NEXT_PUBLIC_TEXT_GENERATION_TIMEOUT_MS}
       >
-        <Topbar />
         <BrowserInitor>
           <SentryInitor>
             <I18nServer>{children}</I18nServer>
diff --git a/web/docker/entrypoint.sh b/web/docker/entrypoint.sh
index fc4a8f45bc8d7..ad4b17a476b02 100755
--- a/web/docker/entrypoint.sh
+++ b/web/docker/entrypoint.sh
@@ -22,5 +22,6 @@ export NEXT_PUBLIC_SITE_ABOUT=${SITE_ABOUT}
 export NEXT_TELEMETRY_DISABLED=${NEXT_TELEMETRY_DISABLED}
 
 export NEXT_PUBLIC_TEXT_GENERATION_TIMEOUT_MS=${TEXT_GENERATION_TIMEOUT_MS}
+export NEXT_PUBLIC_CSP_WHITELIST=${CSP_WHITELIST}
 
 pm2 start ./pm2.json --no-daemon
diff --git a/web/middleware.ts b/web/middleware.ts
new file mode 100644
index 0000000000000..0c5817445fd1a
--- /dev/null
+++ b/web/middleware.ts
@@ -0,0 +1,76 @@
+import type { NextRequest } from 'next/server'
+import { NextResponse } from 'next/server'
+
+const NECESSARY_DOMAIN = '*.sentry.io http://localhost:* http://127.0.0.1:* https://analytics.google.com https://googletagmanager.com https://api.github.com'
+
+export function middleware(request: NextRequest) {
+  const isWhiteListEnabled = !!process.env.NEXT_PUBLIC_CSP_WHITELIST && process.env.NODE_ENV === 'production'
+  if (!isWhiteListEnabled)
+    return NextResponse.next()
+
+  const whiteList = `${process.env.NEXT_PUBLIC_CSP_WHITELIST} ${NECESSARY_DOMAIN}`
+  const nonce = Buffer.from(crypto.randomUUID()).toString('base64')
+  const csp = `'nonce-${nonce}'`
+
+  const scheme_source = 'data: mediastream: blob: filesystem:'
+
+  const cspHeader = `
+    default-src 'self' ${scheme_source} ${csp} ${whiteList};
+    connect-src 'self' ${scheme_source} ${csp} ${whiteList};
+    script-src 'self' ${scheme_source} ${csp} ${whiteList};
+    style-src 'self' 'unsafe-inline' ${scheme_source} ${whiteList};
+    worker-src 'self' ${scheme_source} ${csp} ${whiteList};
+    media-src 'self' ${scheme_source} ${csp} ${whiteList};
+    img-src 'self' ${scheme_source} ${csp} ${whiteList};
+    font-src 'self';
+    object-src 'none';
+    base-uri 'self';
+    form-action 'self';
+    upgrade-insecure-requests;
+`
+  // Replace newline characters and spaces
+  const contentSecurityPolicyHeaderValue = cspHeader
+    .replace(/\s{2,}/g, ' ')
+    .trim()
+
+  const requestHeaders = new Headers(request.headers)
+  requestHeaders.set('x-nonce', nonce)
+
+  requestHeaders.set(
+    'Content-Security-Policy',
+    contentSecurityPolicyHeaderValue,
+  )
+
+  const response = NextResponse.next({
+    request: {
+      headers: requestHeaders,
+    },
+  })
+  response.headers.set(
+    'Content-Security-Policy',
+    contentSecurityPolicyHeaderValue,
+  )
+
+  return response
+}
+
+export const config = {
+  matcher: [
+    /*
+     * Match all request paths except for the ones starting with:
+     * - api (API routes)
+     * - _next/static (static files)
+     * - _next/image (image optimization files)
+     * - favicon.ico (favicon file)
+     */
+    {
+      // source: '/((?!api|_next/static|_next/image|favicon.ico).*)',
+      source: '/((?!_next/static|_next/image|favicon.ico).*)',
+      // source: '/(.*)',
+      // missing: [
+      //   { type: 'header', key: 'next-router-prefetch' },
+      //   { type: 'header', key: 'purpose', value: 'prefetch' },
+      // ],
+    },
+  ],
+}
diff --git a/web/package.json b/web/package.json
index 8237623b7c065..96b89c92315d9 100644
--- a/web/package.json
+++ b/web/package.json
@@ -62,7 +62,6 @@
     "mermaid": "10.4.0",
     "negotiator": "^0.6.3",
     "next": "^14.1.1",
-    "next-nprogress-bar": "^2.3.8",
     "pinyin-pro": "^3.23.0",
     "qrcode.react": "^3.1.0",
     "qs": "^6.11.1",
diff --git a/web/yarn.lock b/web/yarn.lock
index bec2059a47083..45870045a26ac 100644
--- a/web/yarn.lock
+++ b/web/yarn.lock
@@ -7278,13 +7278,6 @@ negotiator@^0.6.3:
   resolved "https://registry.npmjs.org/negotiator/-/negotiator-0.6.3.tgz"
   integrity sha512-+EUsqGPLsM+j/zdChZjsnX51g4XrHFOIXwfnCVPGlQk/k5giakcKsuxCObBRu6DSm9opw/O6slWbJdghQM4bBg==
 
-next-nprogress-bar@^2.3.8:
-  version "2.3.11"
-  resolved "https://registry.npmjs.org/next-nprogress-bar/-/next-nprogress-bar-2.3.11.tgz"
-  integrity sha512-OjSvsQwgSWa2qBMYO478QreGG9Jt82tr4wTQptmiyzNqqjzHCyKZNkhANnzPrjuFAoelIvmruJuakODofSnvTQ==
-  dependencies:
-    nprogress "^0.2.0"
-
 next@^14.1.1:
   version "14.2.4"
   resolved "https://registry.npmjs.org/next/-/next-14.2.4.tgz"
@@ -7367,11 +7360,6 @@ npm-run-path@^5.1.0:
   dependencies:
     path-key "^4.0.0"
 
-nprogress@^0.2.0:
-  version "0.2.0"
-  resolved "https://registry.npmjs.org/nprogress/-/nprogress-0.2.0.tgz"
-  integrity sha512-I19aIingLgR1fmhftnbWWO3dXc0hSxqHQHQb3H8m+K3TnEn/iSeTZZOyvKXWqQESMwuUVnatlCnZdLBZZt2VSA==
-
 nth-check@^2.0.1:
   version "2.1.1"
   resolved "https://registry.npmjs.org/nth-check/-/nth-check-2.1.1.tgz"
@@ -8816,7 +8804,14 @@ stringify-entities@^4.0.0:
     character-entities-html4 "^2.0.0"
     character-entities-legacy "^3.0.0"
 
-"strip-ansi-cjs@npm:strip-ansi@^6.0.1", strip-ansi@^6.0.0, strip-ansi@^6.0.1:
+"strip-ansi-cjs@npm:strip-ansi@^6.0.1":
+  version "6.0.1"
+  resolved "https://registry.npmjs.org/strip-ansi/-/strip-ansi-6.0.1.tgz"
+  integrity sha512-Y38VPSHcqkFrCpFnQ9vuSXmquuv5oXOKpGeT6aGrr3o3Gc9AlVa6JBfUSOCnbxGGZF+/0ooI7KrPuUSztUdU5A==
+  dependencies:
+    ansi-regex "^5.0.1"
+
+strip-ansi@^6.0.0, strip-ansi@^6.0.1:
   version "6.0.1"
   resolved "https://registry.npmjs.org/strip-ansi/-/strip-ansi-6.0.1.tgz"
   integrity sha512-Y38VPSHcqkFrCpFnQ9vuSXmquuv5oXOKpGeT6aGrr3o3Gc9AlVa6JBfUSOCnbxGGZF+/0ooI7KrPuUSztUdU5A==
@@ -9653,8 +9648,7 @@ word-wrap@^1.2.3:
   resolved "https://registry.npmjs.org/word-wrap/-/word-wrap-1.2.5.tgz"
   integrity sha512-BN22B5eaMMI9UMtjrGd5g5eCYPpCPDUy0FJXbYsaT5zYxjFOckS53SQDE3pWkVoWpHXVb3BrYcEN4Twa55B5cA==
 
-"wrap-ansi-cjs@npm:wrap-ansi@^7.0.0", wrap-ansi@^7.0.0:
-  name wrap-ansi-cjs
+"wrap-ansi-cjs@npm:wrap-ansi@^7.0.0":
   version "7.0.0"
   resolved "https://registry.npmjs.org/wrap-ansi/-/wrap-ansi-7.0.0.tgz"
   integrity sha512-YVGIj2kamLSTxw6NsZjoBxfSwsn0ycdesmc4p+Q21c5zPuZ1pl+NfxVdxPtdHvmNVOQ6XSYG4AUtyt/Fi7D16Q==
@@ -9672,6 +9666,15 @@ wrap-ansi@^6.2.0:
     string-width "^4.1.0"
     strip-ansi "^6.0.0"
 
+wrap-ansi@^7.0.0:
+  version "7.0.0"
+  resolved "https://registry.npmjs.org/wrap-ansi/-/wrap-ansi-7.0.0.tgz"
+  integrity sha512-YVGIj2kamLSTxw6NsZjoBxfSwsn0ycdesmc4p+Q21c5zPuZ1pl+NfxVdxPtdHvmNVOQ6XSYG4AUtyt/Fi7D16Q==
+  dependencies:
+    ansi-styles "^4.0.0"
+    string-width "^4.1.0"
+    strip-ansi "^6.0.0"
+
 wrap-ansi@^8.1.0:
   version "8.1.0"
   resolved "https://registry.npmjs.org/wrap-ansi/-/wrap-ansi-8.1.0.tgz"

From 82024a65cdcd1c4aae527a27ee6cdebdd987c220 Mon Sep 17 00:00:00 2001
From: "dependabot[bot]" <49699333+dependabot[bot]@users.noreply.github.com>
Date: Fri, 11 Oct 2024 16:56:32 +0800
Subject: [PATCH 205/235] chore(deps): bump micromatch from 4.0.5 to 4.0.8 in
 /web (#9234)

Signed-off-by: dependabot[bot] <support@github.com>
Co-authored-by: dependabot[bot] <49699333+dependabot[bot]@users.noreply.github.com>
---
 web/yarn.lock | 10 +++++-----
 1 file changed, 5 insertions(+), 5 deletions(-)

diff --git a/web/yarn.lock b/web/yarn.lock
index 45870045a26ac..5573849249ad3 100644
--- a/web/yarn.lock
+++ b/web/yarn.lock
@@ -2698,7 +2698,7 @@ brace-expansion@^2.0.1:
   dependencies:
     balanced-match "^1.0.0"
 
-braces@^3.0.2, braces@~3.0.2:
+braces@^3.0.3, braces@~3.0.2:
   version "3.0.3"
   resolved "https://registry.yarnpkg.com/braces/-/braces-3.0.3.tgz#490332f40919452272d55a8480adc0c441358789"
   integrity sha512-yQbXgO/OSZVD2IsiLlro+7Hf6Q18EJrKSEsdoMzKePKXct3gvD8oLcOQdIzGupr5Fj+EDe8gO/lxc1BzfMpxvA==
@@ -7164,11 +7164,11 @@ micromark@~2.11.0:
     parse-entities "^2.0.0"
 
 micromatch@^4.0.4, micromatch@^4.0.5:
-  version "4.0.5"
-  resolved "https://registry.npmjs.org/micromatch/-/micromatch-4.0.5.tgz"
-  integrity sha512-DMy+ERcEW2q8Z2Po+WNXuw3c5YaUSFjAO5GsJqfEl7UjvtIuFKO6ZrKvcItdy98dwFI2N1tg3zNIdKaQT+aNdA==
+  version "4.0.8"
+  resolved "https://registry.yarnpkg.com/micromatch/-/micromatch-4.0.8.tgz#d66fa18f3a47076789320b9b1af32bd86d9fa202"
+  integrity sha512-PXwfBhYu0hBCPw8Dn0E+WDYb7af3dSLVWKi3HGv84IdF4TyFoC0ysxFd0Goxw7nSv4T/PzEJQxsYsEiFCKo2BA==
   dependencies:
-    braces "^3.0.2"
+    braces "^3.0.3"
     picomatch "^2.3.1"
 
 mime-db@1.52.0:

From 3f1aa1f9e263f1134ba0a89c61a5b0dd14eac0ac Mon Sep 17 00:00:00 2001
From: "dependabot[bot]" <49699333+dependabot[bot]@users.noreply.github.com>
Date: Fri, 11 Oct 2024 16:57:07 +0800
Subject: [PATCH 206/235] chore(deps): bump dompurify from 3.0.5 to 3.1.7 in
 /web (#9232)

Signed-off-by: dependabot[bot] <support@github.com>
Co-authored-by: dependabot[bot] <49699333+dependabot[bot]@users.noreply.github.com>
---
 web/yarn.lock | 6 +++---
 1 file changed, 3 insertions(+), 3 deletions(-)

diff --git a/web/yarn.lock b/web/yarn.lock
index 5573849249ad3..5d693ff786d8f 100644
--- a/web/yarn.lock
+++ b/web/yarn.lock
@@ -3793,9 +3793,9 @@ domhandler@^5.0.2, domhandler@^5.0.3:
     domelementtype "^2.3.0"
 
 dompurify@^3.0.5:
-  version "3.0.5"
-  resolved "https://registry.npmjs.org/dompurify/-/dompurify-3.0.5.tgz"
-  integrity sha512-F9e6wPGtY+8KNMRAVfxeCOHU0/NPWMSENNq4pQctuXRqqdEPW7q3CrLbR5Nse044WwacyjHGOMlvNsBe1y6z9A==
+  version "3.1.7"
+  resolved "https://registry.yarnpkg.com/dompurify/-/dompurify-3.1.7.tgz#711a8c96479fb6ced93453732c160c3c72418a6a"
+  integrity sha512-VaTstWtsneJY8xzy7DekmYWEOZcmzIe3Qb3zPd4STve1OBTa+e+WmS1ITQec1fZYXI3HCsOZZiSMpG6oxoWMWQ==
 
 domutils@^3.0.1:
   version "3.1.0"

From d7b8e071dd0f23a0d0c239a44f7d6bb239840e1c Mon Sep 17 00:00:00 2001
From: AkisAya <akikevinsama@gmail.com>
Date: Fri, 11 Oct 2024 17:12:30 +0800
Subject: [PATCH 207/235] fix:#9222 create or update custum tools error (#9228)

---
 api/core/tools/utils/parser.py                 | 2 +-
 api/services/tools/api_tools_manage_service.py | 4 +++-
 2 files changed, 4 insertions(+), 2 deletions(-)

diff --git a/api/core/tools/utils/parser.py b/api/core/tools/utils/parser.py
index 0d801d36c4dcd..5867a11bb39da 100644
--- a/api/core/tools/utils/parser.py
+++ b/api/core/tools/utils/parser.py
@@ -288,7 +288,7 @@ def parse_openai_plugin_json_to_tool_bundle(
 
     @staticmethod
     def auto_parse_to_tool_bundle(
-        content: str, extra_info: Optional[dict], warning: Optional[dict]
+        content: str, extra_info: Optional[dict] = None, warning: Optional[dict] = None
     ) -> tuple[list[ApiToolBundle], str]:
         """
         auto parse to tool bundle
diff --git a/api/services/tools/api_tools_manage_service.py b/api/services/tools/api_tools_manage_service.py
index 9f5298a506752..257c6cf52bbb3 100644
--- a/api/services/tools/api_tools_manage_service.py
+++ b/api/services/tools/api_tools_manage_service.py
@@ -80,7 +80,9 @@ def parser_api_schema(schema: str) -> list[ApiToolBundle]:
             raise ValueError(f"invalid schema: {str(e)}")
 
     @staticmethod
-    def convert_schema_to_tool_bundles(schema: str, extra_info: Optional[dict] = None) -> list[ApiToolBundle]:
+    def convert_schema_to_tool_bundles(
+        schema: str, extra_info: Optional[dict] = None
+    ) -> tuple[list[ApiToolBundle], str]:
         """
         convert schema to tool bundles
 

From 44f6a536d24447a05d1ac169254a8e4fcf561d31 Mon Sep 17 00:00:00 2001
From: Yi Xiao <54782454+YIXIAO0@users.noreply.github.com>
Date: Fri, 11 Oct 2024 19:17:22 +0800
Subject: [PATCH 208/235] fix: reference issue in external knowledge api
 (#9240)

---
 web/app/components/base/chat/chat/citation/popup.tsx | 2 +-
 1 file changed, 1 insertion(+), 1 deletion(-)

diff --git a/web/app/components/base/chat/chat/citation/popup.tsx b/web/app/components/base/chat/chat/citation/popup.tsx
index b61bf623febe9..9b98329e70beb 100644
--- a/web/app/components/base/chat/chat/citation/popup.tsx
+++ b/web/app/components/base/chat/chat/citation/popup.tsx
@@ -100,7 +100,7 @@ const Popup: FC<PopupProps> = ({
                             />
                             <Tooltip
                               text={t('common.chat.citation.vectorHash')}
-                              data={source.index_node_hash.substring(0, 7)}
+                              data={source.index_node_hash?.substring(0, 7)}
                               icon={<BezierCurve03 className='mr-1 w-3 h-3' />}
                             />
                             {

From 42b02b3a5f6c833ac5e09fd9667920a34ebed518 Mon Sep 17 00:00:00 2001
From: Jyong <76649700+JohnJyong@users.noreply.github.com>
Date: Fri, 11 Oct 2024 19:21:03 +0800
Subject: [PATCH 209/235] Fix/agent external knowledge retrieval (#9241)

---
 api/configs/middleware/__init__.py            |  17 ++
 api/controllers/console/datasets/external.py  |  24 ++
 api/core/rag/retrieval/dataset_retrieval.py   |   2 +-
 .../dataset_retriever_tool.py                 | 218 +++++++++++-------
 .../knowledge_retrieval_node.py               |  14 +-
 api/services/dataset_service.py               |   1 +
 api/services/knowledge_service.py             |  45 ++++
 7 files changed, 227 insertions(+), 94 deletions(-)
 create mode 100644 api/services/knowledge_service.py

diff --git a/api/configs/middleware/__init__.py b/api/configs/middleware/__init__.py
index 8626236856ddb..5fec991d6e2d9 100644
--- a/api/configs/middleware/__init__.py
+++ b/api/configs/middleware/__init__.py
@@ -191,6 +191,22 @@ def BROKER_USE_SSL(self) -> bool:
         return self.CELERY_BROKER_URL.startswith("rediss://") if self.CELERY_BROKER_URL else False
 
 
+class InternalTestConfig(BaseSettings):
+    """
+    Configuration settings for Internal Test
+    """
+
+    AWS_SECRET_ACCESS_KEY: Optional[str] = Field(
+        description="Internal test AWS secret access key",
+        default=None,
+    )
+
+    AWS_ACCESS_KEY_ID: Optional[str] = Field(
+        description="Internal test AWS access key ID",
+        default=None,
+    )
+
+
 class MiddlewareConfig(
     # place the configs in alphabet order
     CeleryConfig,
@@ -224,5 +240,6 @@ class MiddlewareConfig(
     TiDBVectorConfig,
     WeaviateConfig,
     ElasticsearchConfig,
+    InternalTestConfig,
 ):
     pass
diff --git a/api/controllers/console/datasets/external.py b/api/controllers/console/datasets/external.py
index dcef979360ba6..2dc054cfbdf4a 100644
--- a/api/controllers/console/datasets/external.py
+++ b/api/controllers/console/datasets/external.py
@@ -13,6 +13,7 @@
 from services.dataset_service import DatasetService
 from services.external_knowledge_service import ExternalDatasetService
 from services.hit_testing_service import HitTestingService
+from services.knowledge_service import ExternalDatasetTestService
 
 
 def _validate_name(name):
@@ -232,8 +233,31 @@ def post(self, dataset_id):
             raise InternalServerError(str(e))
 
 
+class BedrockRetrievalApi(Resource):
+    # this api is only for internal testing
+    def post(self):
+        parser = reqparse.RequestParser()
+        parser.add_argument("retrieval_setting", nullable=False, required=True, type=dict, location="json")
+        parser.add_argument(
+            "query",
+            nullable=False,
+            required=True,
+            type=str,
+        )
+        parser.add_argument("knowledge_id", nullable=False, required=True, type=str)
+        args = parser.parse_args()
+
+        # Call the knowledge retrieval service
+        result = ExternalDatasetTestService.knowledge_retrieval(
+            args["retrieval_setting"], args["query"], args["knowledge_id"]
+        )
+        return result, 200
+
+
 api.add_resource(ExternalKnowledgeHitTestingApi, "/datasets/<uuid:dataset_id>/external-hit-testing")
 api.add_resource(ExternalDatasetCreateApi, "/datasets/external")
 api.add_resource(ExternalApiTemplateListApi, "/datasets/external-knowledge-api")
 api.add_resource(ExternalApiTemplateApi, "/datasets/external-knowledge-api/<uuid:external_knowledge_api_id>")
 api.add_resource(ExternalApiUseCheckApi, "/datasets/external-knowledge-api/<uuid:external_knowledge_api_id>/use-check")
+# this api is only for internal test
+api.add_resource(BedrockRetrievalApi, "/test/retrieval")
diff --git a/api/core/rag/retrieval/dataset_retrieval.py b/api/core/rag/retrieval/dataset_retrieval.py
index ae79b44158c84..633e41d5cf1ed 100644
--- a/api/core/rag/retrieval/dataset_retrieval.py
+++ b/api/core/rag/retrieval/dataset_retrieval.py
@@ -539,7 +539,7 @@ def to_dataset_retriever_tool(
                 continue
 
             # pass if dataset is not available
-            if dataset and dataset.available_document_count == 0:
+            if dataset and dataset.provider != "external" and dataset.available_document_count == 0:
                 continue
 
             available_datasets.append(dataset)
diff --git a/api/core/tools/tool/dataset_retriever/dataset_retriever_tool.py b/api/core/tools/tool/dataset_retriever/dataset_retriever_tool.py
index 8dc60408c93b4..987f94a35046e 100644
--- a/api/core/tools/tool/dataset_retriever/dataset_retriever_tool.py
+++ b/api/core/tools/tool/dataset_retriever/dataset_retriever_tool.py
@@ -1,10 +1,12 @@
 from pydantic import BaseModel, Field
 
 from core.rag.datasource.retrieval_service import RetrievalService
+from core.rag.models.document import Document as RetrievalDocument
 from core.rag.retrieval.retrieval_methods import RetrievalMethod
 from core.tools.tool.dataset_retriever.dataset_retriever_base_tool import DatasetRetrieverBaseTool
 from extensions.ext_database import db
 from models.dataset import Dataset, Document, DocumentSegment
+from services.external_knowledge_service import ExternalDatasetService
 
 default_retrieval_model = {
     "search_method": RetrievalMethod.SEMANTIC_SEARCH.value,
@@ -53,97 +55,137 @@ def _run(self, query: str) -> str:
 
         for hit_callback in self.hit_callbacks:
             hit_callback.on_query(query, dataset.id)
-
-        # get retrieval model , if the model is not setting , using default
-        retrieval_model = dataset.retrieval_model or default_retrieval_model
-        if dataset.indexing_technique == "economy":
-            # use keyword table query
-            documents = RetrievalService.retrieve(
-                retrieval_method="keyword_search", dataset_id=dataset.id, query=query, top_k=self.top_k
+        if dataset.provider == "external":
+            results = []
+            external_documents = ExternalDatasetService.fetch_external_knowledge_retrieval(
+                tenant_id=dataset.tenant_id,
+                dataset_id=dataset.id,
+                query=query,
+                external_retrieval_parameters=dataset.retrieval_model,
             )
-            return str("\n".join([document.page_content for document in documents]))
+            for external_document in external_documents:
+                document = RetrievalDocument(
+                    page_content=external_document.get("content"),
+                    metadata=external_document.get("metadata"),
+                    provider="external",
+                )
+                document.metadata["score"] = external_document.get("score")
+                document.metadata["title"] = external_document.get("title")
+                document.metadata["dataset_id"] = dataset.id
+                document.metadata["dataset_name"] = dataset.name
+                results.append(document)
+            # deal with external documents
+            context_list = []
+            for position, item in enumerate(results, start=1):
+                source = {
+                    "position": position,
+                    "dataset_id": item.metadata.get("dataset_id"),
+                    "dataset_name": item.metadata.get("dataset_name"),
+                    "document_name": item.metadata.get("title"),
+                    "data_source_type": "external",
+                    "retriever_from": self.retriever_from,
+                    "score": item.metadata.get("score"),
+                    "title": item.metadata.get("title"),
+                    "content": item.page_content,
+                }
+                context_list.append(source)
+            for hit_callback in self.hit_callbacks:
+                hit_callback.return_retriever_resource_info(context_list)
+
+            return str("\n".join([item.page_content for item in results]))
         else:
-            if self.top_k > 0:
-                # retrieval source
+            # get retrieval model , if the model is not setting , using default
+            retrieval_model = dataset.retrieval_model or default_retrieval_model
+            if dataset.indexing_technique == "economy":
+                # use keyword table query
                 documents = RetrievalService.retrieve(
-                    retrieval_method=retrieval_model.get("search_method", "semantic_search"),
-                    dataset_id=dataset.id,
-                    query=query,
-                    top_k=self.top_k,
-                    score_threshold=retrieval_model.get("score_threshold", 0.0)
-                    if retrieval_model["score_threshold_enabled"]
-                    else 0.0,
-                    reranking_model=retrieval_model.get("reranking_model", None)
-                    if retrieval_model["reranking_enable"]
-                    else None,
-                    reranking_mode=retrieval_model.get("reranking_mode") or "reranking_model",
-                    weights=retrieval_model.get("weights", None),
+                    retrieval_method="keyword_search", dataset_id=dataset.id, query=query, top_k=self.top_k
                 )
+                return str("\n".join([document.page_content for document in documents]))
             else:
-                documents = []
-
-            for hit_callback in self.hit_callbacks:
-                hit_callback.on_tool_end(documents)
-            document_score_list = {}
-            if dataset.indexing_technique != "economy":
-                for item in documents:
-                    if item.metadata.get("score"):
-                        document_score_list[item.metadata["doc_id"]] = item.metadata["score"]
-            document_context_list = []
-            index_node_ids = [document.metadata["doc_id"] for document in documents]
-            segments = DocumentSegment.query.filter(
-                DocumentSegment.dataset_id == self.dataset_id,
-                DocumentSegment.completed_at.isnot(None),
-                DocumentSegment.status == "completed",
-                DocumentSegment.enabled == True,
-                DocumentSegment.index_node_id.in_(index_node_ids),
-            ).all()
-
-            if segments:
-                index_node_id_to_position = {id: position for position, id in enumerate(index_node_ids)}
-                sorted_segments = sorted(
-                    segments, key=lambda segment: index_node_id_to_position.get(segment.index_node_id, float("inf"))
-                )
-                for segment in sorted_segments:
-                    if segment.answer:
-                        document_context_list.append(f"question:{segment.get_sign_content()} answer:{segment.answer}")
-                    else:
-                        document_context_list.append(segment.get_sign_content())
-                if self.return_resource:
-                    context_list = []
-                    resource_number = 1
+                if self.top_k > 0:
+                    # retrieval source
+                    documents = RetrievalService.retrieve(
+                        retrieval_method=retrieval_model.get("search_method", "semantic_search"),
+                        dataset_id=dataset.id,
+                        query=query,
+                        top_k=self.top_k,
+                        score_threshold=retrieval_model.get("score_threshold", 0.0)
+                        if retrieval_model["score_threshold_enabled"]
+                        else 0.0,
+                        reranking_model=retrieval_model.get("reranking_model", None)
+                        if retrieval_model["reranking_enable"]
+                        else None,
+                        reranking_mode=retrieval_model.get("reranking_mode") or "reranking_model",
+                        weights=retrieval_model.get("weights", None),
+                    )
+                else:
+                    documents = []
+
+                for hit_callback in self.hit_callbacks:
+                    hit_callback.on_tool_end(documents)
+                document_score_list = {}
+                if dataset.indexing_technique != "economy":
+                    for item in documents:
+                        if item.metadata.get("score"):
+                            document_score_list[item.metadata["doc_id"]] = item.metadata["score"]
+                document_context_list = []
+                index_node_ids = [document.metadata["doc_id"] for document in documents]
+                segments = DocumentSegment.query.filter(
+                    DocumentSegment.dataset_id == self.dataset_id,
+                    DocumentSegment.completed_at.isnot(None),
+                    DocumentSegment.status == "completed",
+                    DocumentSegment.enabled == True,
+                    DocumentSegment.index_node_id.in_(index_node_ids),
+                ).all()
+
+                if segments:
+                    index_node_id_to_position = {id: position for position, id in enumerate(index_node_ids)}
+                    sorted_segments = sorted(
+                        segments, key=lambda segment: index_node_id_to_position.get(segment.index_node_id, float("inf"))
+                    )
                     for segment in sorted_segments:
-                        context = {}
-                        document = Document.query.filter(
-                            Document.id == segment.document_id,
-                            Document.enabled == True,
-                            Document.archived == False,
-                        ).first()
-                        if dataset and document:
-                            source = {
-                                "position": resource_number,
-                                "dataset_id": dataset.id,
-                                "dataset_name": dataset.name,
-                                "document_id": document.id,
-                                "document_name": document.name,
-                                "data_source_type": document.data_source_type,
-                                "segment_id": segment.id,
-                                "retriever_from": self.retriever_from,
-                                "score": document_score_list.get(segment.index_node_id, None),
-                            }
-                            if self.retriever_from == "dev":
-                                source["hit_count"] = segment.hit_count
-                                source["word_count"] = segment.word_count
-                                source["segment_position"] = segment.position
-                                source["index_node_hash"] = segment.index_node_hash
-                            if segment.answer:
-                                source["content"] = f"question:{segment.content} \nanswer:{segment.answer}"
-                            else:
-                                source["content"] = segment.content
-                            context_list.append(source)
-                        resource_number += 1
-
-                    for hit_callback in self.hit_callbacks:
-                        hit_callback.return_retriever_resource_info(context_list)
-
-            return str("\n".join(document_context_list))
+                        if segment.answer:
+                            document_context_list.append(
+                                f"question:{segment.get_sign_content()} answer:{segment.answer}"
+                            )
+                        else:
+                            document_context_list.append(segment.get_sign_content())
+                    if self.return_resource:
+                        context_list = []
+                        resource_number = 1
+                        for segment in sorted_segments:
+                            context = {}
+                            document = Document.query.filter(
+                                Document.id == segment.document_id,
+                                Document.enabled == True,
+                                Document.archived == False,
+                            ).first()
+                            if dataset and document:
+                                source = {
+                                    "position": resource_number,
+                                    "dataset_id": dataset.id,
+                                    "dataset_name": dataset.name,
+                                    "document_id": document.id,
+                                    "document_name": document.name,
+                                    "data_source_type": document.data_source_type,
+                                    "segment_id": segment.id,
+                                    "retriever_from": self.retriever_from,
+                                    "score": document_score_list.get(segment.index_node_id, None),
+                                }
+                                if self.retriever_from == "dev":
+                                    source["hit_count"] = segment.hit_count
+                                    source["word_count"] = segment.word_count
+                                    source["segment_position"] = segment.position
+                                    source["index_node_hash"] = segment.index_node_hash
+                                if segment.answer:
+                                    source["content"] = f"question:{segment.content} \nanswer:{segment.answer}"
+                                else:
+                                    source["content"] = segment.content
+                                context_list.append(source)
+                            resource_number += 1
+
+                        for hit_callback in self.hit_callbacks:
+                            hit_callback.return_retriever_resource_info(context_list)
+
+                return str("\n".join(document_context_list))
diff --git a/api/core/workflow/nodes/knowledge_retrieval/knowledge_retrieval_node.py b/api/core/workflow/nodes/knowledge_retrieval/knowledge_retrieval_node.py
index 0caf99a963822..0b3e9bd6a888e 100644
--- a/api/core/workflow/nodes/knowledge_retrieval/knowledge_retrieval_node.py
+++ b/api/core/workflow/nodes/knowledge_retrieval/knowledge_retrieval_node.py
@@ -79,8 +79,9 @@ def _fetch_dataset_retriever(self, node_data: KnowledgeRetrievalNodeData, query:
 
         results = (
             db.session.query(Dataset)
-            .join(subquery, Dataset.id == subquery.c.dataset_id)
+            .outerjoin(subquery, Dataset.id == subquery.c.dataset_id)
             .filter(Dataset.tenant_id == self.tenant_id, Dataset.id.in_(dataset_ids))
+            .filter((subquery.c.available_document_count > 0) | (Dataset.provider == "external"))
             .all()
         )
 
@@ -121,10 +122,13 @@ def _fetch_dataset_retriever(self, node_data: KnowledgeRetrievalNodeData, query:
                 )
         elif node_data.retrieval_mode == DatasetRetrieveConfigEntity.RetrieveStrategy.MULTIPLE.value:
             if node_data.multiple_retrieval_config.reranking_mode == "reranking_model":
-                reranking_model = {
-                    "reranking_provider_name": node_data.multiple_retrieval_config.reranking_model.provider,
-                    "reranking_model_name": node_data.multiple_retrieval_config.reranking_model.model,
-                }
+                if node_data.multiple_retrieval_config.reranking_model:
+                    reranking_model = {
+                        "reranking_provider_name": node_data.multiple_retrieval_config.reranking_model.provider,
+                        "reranking_model_name": node_data.multiple_retrieval_config.reranking_model.model,
+                    }
+                else:
+                    reranking_model = None
                 weights = None
             elif node_data.multiple_retrieval_config.reranking_mode == "weighted_score":
                 reranking_model = None
diff --git a/api/services/dataset_service.py b/api/services/dataset_service.py
index 4f77fdc18cb43..ede8764086219 100644
--- a/api/services/dataset_service.py
+++ b/api/services/dataset_service.py
@@ -234,6 +234,7 @@ def update_dataset(dataset_id, data, user):
             dataset.name = data.get("name", dataset.name)
             dataset.description = data.get("description", "")
             external_knowledge_id = data.get("external_knowledge_id", None)
+            dataset.permission = data.get("permission")
             db.session.add(dataset)
             if not external_knowledge_id:
                 raise ValueError("External knowledge id is required.")
diff --git a/api/services/knowledge_service.py b/api/services/knowledge_service.py
new file mode 100644
index 0000000000000..02fe1d19bc42b
--- /dev/null
+++ b/api/services/knowledge_service.py
@@ -0,0 +1,45 @@
+import boto3
+
+from configs import dify_config
+
+
+class ExternalDatasetTestService:
+    # this service is only for internal testing
+    @staticmethod
+    def knowledge_retrieval(retrieval_setting: dict, query: str, knowledge_id: str):
+        # get bedrock client
+        client = boto3.client(
+            "bedrock-agent-runtime",
+            aws_secret_access_key=dify_config.AWS_SECRET_ACCESS_KEY,
+            aws_access_key_id=dify_config.AWS_ACCESS_KEY_ID,
+            # example: us-east-1
+            region_name="us-east-1",
+        )
+        # fetch external knowledge retrieval
+        response = client.retrieve(
+            knowledgeBaseId=knowledge_id,
+            retrievalConfiguration={
+                "vectorSearchConfiguration": {
+                    "numberOfResults": retrieval_setting.get("top_k"),
+                    "overrideSearchType": "HYBRID",
+                }
+            },
+            retrievalQuery={"text": query},
+        )
+        # parse response
+        results = []
+        if response.get("ResponseMetadata") and response.get("ResponseMetadata").get("HTTPStatusCode") == 200:
+            if response.get("retrievalResults"):
+                retrieval_results = response.get("retrievalResults")
+                for retrieval_result in retrieval_results:
+                    # filter out results with score less than threshold
+                    if retrieval_result.get("score") < retrieval_setting.get("score_threshold", 0.0):
+                        continue
+                    result = {
+                        "metadata": retrieval_result.get("metadata"),
+                        "score": retrieval_result.get("score"),
+                        "title": retrieval_result.get("metadata").get("x-amz-bedrock-kb-source-uri"),
+                        "content": retrieval_result.get("content").get("text"),
+                    }
+                    results.append(result)
+        return {"records": results}

From ea584e94bd5fc30aed78a506261c7ae987e2e9f1 Mon Sep 17 00:00:00 2001
From: Likename Haojie <likenamehaojie@hotmail.com>
Date: Fri, 11 Oct 2024 22:46:44 +0800
Subject: [PATCH 210/235] fix: dialog box cannot correctly display LaTeX
 formulas (#9242)

---
 web/app/components/base/markdown.tsx | 2 +-
 1 file changed, 1 insertion(+), 1 deletion(-)

diff --git a/web/app/components/base/markdown.tsx b/web/app/components/base/markdown.tsx
index 39a399cc9f73c..dbe408788253b 100644
--- a/web/app/components/base/markdown.tsx
+++ b/web/app/components/base/markdown.tsx
@@ -245,7 +245,7 @@ export function Markdown(props: { content: string; className?: string }) {
   return (
     <div className={cn(props.className, 'markdown-body')}>
       <ReactMarkdown
-        remarkPlugins={[[RemarkGfm, RemarkMath, { singleDollarTextMath: false }], RemarkBreaks]}
+        remarkPlugins={[RemarkGfm, RemarkMath, RemarkBreaks]}
         rehypePlugins={[
           RehypeKatex,
           RehypeRaw as any,

From 7a6970e570242f84dff5ec1b2be33696625ef414 Mon Sep 17 00:00:00 2001
From: zhuhao <37029601+hwzhuhao@users.noreply.github.com>
Date: Fri, 11 Oct 2024 22:48:57 +0800
Subject: [PATCH 211/235] feat: add supabase object storage (#9229)

---
 api/.env.example                              |   7 +-
 api/configs/middleware/__init__.py            |   2 +
 .../storage/supabase_storage_config.py        |  24 ++
 api/extensions/ext_storage.py                 |   4 +
 api/extensions/storage/storage_type.py        |   1 +
 api/extensions/storage/supabase_storage.py    |  60 ++++
 api/poetry.lock                               | 324 ++++++++++++------
 api/pyproject.toml                            |   1 +
 8 files changed, 318 insertions(+), 105 deletions(-)
 create mode 100644 api/configs/middleware/storage/supabase_storage_config.py
 create mode 100644 api/extensions/storage/supabase_storage.py

diff --git a/api/.env.example b/api/.env.example
index 71f0e5db8f8b9..7b5e4950c82dd 100644
--- a/api/.env.example
+++ b/api/.env.example
@@ -39,7 +39,7 @@ DB_DATABASE=dify
 
 # Storage configuration
 # use for store upload files, private keys...
-# storage type: local, s3, azure-blob, google-storage, tencent-cos, huawei-obs, volcengine-tos, baidu-obs
+# storage type: local, s3, azure-blob, google-storage, tencent-cos, huawei-obs, volcengine-tos, baidu-obs, supabase
 STORAGE_TYPE=local
 STORAGE_LOCAL_PATH=storage
 S3_USE_AWS_MANAGED_IAM=false
@@ -99,6 +99,11 @@ VOLCENGINE_TOS_ACCESS_KEY=your-access-key
 VOLCENGINE_TOS_SECRET_KEY=your-secret-key
 VOLCENGINE_TOS_REGION=your-region
 
+# Supabase Storage Configuration
+SUPABASE_BUCKET_NAME=your-bucket-name
+SUPABASE_API_KEY=your-access-key
+SUPABASE_URL=your-server-url
+
 # CORS configuration
 WEB_API_CORS_ALLOW_ORIGINS=http://127.0.0.1:3000,*
 CONSOLE_CORS_ALLOW_ORIGINS=http://127.0.0.1:3000,*
diff --git a/api/configs/middleware/__init__.py b/api/configs/middleware/__init__.py
index 5fec991d6e2d9..25f3df6dde41d 100644
--- a/api/configs/middleware/__init__.py
+++ b/api/configs/middleware/__init__.py
@@ -12,6 +12,7 @@
 from configs.middleware.storage.google_cloud_storage_config import GoogleCloudStorageConfig
 from configs.middleware.storage.huawei_obs_storage_config import HuaweiCloudOBSStorageConfig
 from configs.middleware.storage.oci_storage_config import OCIStorageConfig
+from configs.middleware.storage.supabase_storage_config import SupabaseStorageConfig
 from configs.middleware.storage.tencent_cos_storage_config import TencentCloudCOSStorageConfig
 from configs.middleware.storage.volcengine_tos_storage_config import VolcengineTOSStorageConfig
 from configs.middleware.vdb.analyticdb_config import AnalyticdbConfig
@@ -222,6 +223,7 @@ class MiddlewareConfig(
     HuaweiCloudOBSStorageConfig,
     OCIStorageConfig,
     S3StorageConfig,
+    SupabaseStorageConfig,
     TencentCloudCOSStorageConfig,
     VolcengineTOSStorageConfig,
     # configs of vdb and vdb providers
diff --git a/api/configs/middleware/storage/supabase_storage_config.py b/api/configs/middleware/storage/supabase_storage_config.py
new file mode 100644
index 0000000000000..a3e905b21c63e
--- /dev/null
+++ b/api/configs/middleware/storage/supabase_storage_config.py
@@ -0,0 +1,24 @@
+from typing import Optional
+
+from pydantic import BaseModel, Field
+
+
+class SupabaseStorageConfig(BaseModel):
+    """
+    Configuration settings for Supabase Object Storage Service
+    """
+
+    SUPABASE_BUCKET_NAME: Optional[str] = Field(
+        description="Name of the Supabase bucket to store and retrieve objects (e.g., 'dify-bucket')",
+        default=None,
+    )
+
+    SUPABASE_API_KEY: Optional[str] = Field(
+        description="API KEY for authenticating with Supabase",
+        default=None,
+    )
+
+    SUPABASE_URL: Optional[str] = Field(
+        description="URL of the Supabase",
+        default=None,
+    )
diff --git a/api/extensions/ext_storage.py b/api/extensions/ext_storage.py
index be57b633bed4a..f90629262d89d 100644
--- a/api/extensions/ext_storage.py
+++ b/api/extensions/ext_storage.py
@@ -56,6 +56,10 @@ def get_storage_factory(storage_type: str) -> type[BaseStorage]:
                 from extensions.storage.volcengine_tos_storage import VolcengineTosStorage
 
                 return VolcengineTosStorage
+            case StorageType.SUPBASE:
+                from extensions.storage.supabase_storage import SupabaseStorage
+
+                return SupabaseStorage
             case StorageType.LOCAL | _:
                 from extensions.storage.local_fs_storage import LocalFsStorage
 
diff --git a/api/extensions/storage/storage_type.py b/api/extensions/storage/storage_type.py
index e494a520a20b0..415bf251f6e28 100644
--- a/api/extensions/storage/storage_type.py
+++ b/api/extensions/storage/storage_type.py
@@ -12,3 +12,4 @@ class StorageType(str, Enum):
     S3 = "s3"
     TENCENT_COS = "tencent-cos"
     VOLCENGINE_TOS = "volcengine-tos"
+    SUPBASE = "supabase"
diff --git a/api/extensions/storage/supabase_storage.py b/api/extensions/storage/supabase_storage.py
new file mode 100644
index 0000000000000..1e399f87c8ba3
--- /dev/null
+++ b/api/extensions/storage/supabase_storage.py
@@ -0,0 +1,60 @@
+import io
+from collections.abc import Generator
+from pathlib import Path
+
+from flask import Flask
+from supabase import Client
+
+from extensions.storage.base_storage import BaseStorage
+
+
+class SupabaseStorage(BaseStorage):
+    """Implementation for supabase obs storage."""
+
+    def __init__(self, app: Flask):
+        super().__init__(app)
+        app_config = self.app.config
+        self.bucket_name = app_config.get("SUPABASE_BUCKET_NAME")
+        self.client = Client(
+            supabase_url=app_config.get("SUPABASE_URL"), supabase_key=app_config.get("SUPABASE_API_KEY")
+        )
+        self.create_bucket(
+            id=app_config.get("SUPABASE_BUCKET_NAME"), bucket_name=app_config.get("SUPABASE_BUCKET_NAME")
+        )
+
+    def create_bucket(self, id, bucket_name):
+        if not self.bucket_exists():
+            self.client.storage.create_bucket(id=id, name=bucket_name)
+
+    def save(self, filename, data):
+        self.client.storage.from_(self.bucket_name).upload(filename, data)
+
+    def load_once(self, filename: str) -> bytes:
+        content = self.client.storage.from_(self.bucket_name).download(filename)
+        return content
+
+    def load_stream(self, filename: str) -> Generator:
+        def generate(filename: str = filename) -> Generator:
+            result = self.client.storage.from_(self.bucket_name).download(filename)
+            byte_stream = io.BytesIO(result)
+            while chunk := byte_stream.read(4096):  # Read in chunks of 4KB
+                yield chunk
+
+        return generate()
+
+    def download(self, filename, target_filepath):
+        result = self.client.storage.from_(self.bucket_name).download(filename)
+        Path(result).write_bytes(result)
+
+    def exists(self, filename):
+        result = self.client.storage.from_(self.bucket_name).list(filename)
+        if result.count() > 0:
+            return True
+        return False
+
+    def delete(self, filename):
+        self.client.storage.from_(self.bucket_name).remove(filename)
+
+    def bucket_exists(self):
+        buckets = self.client.storage.list_buckets()
+        return any(bucket.name == self.bucket_name for bucket in buckets)
diff --git a/api/poetry.lock b/api/poetry.lock
index b7421ca566929..fd6b98191df02 100644
--- a/api/poetry.lock
+++ b/api/poetry.lock
@@ -844,13 +844,13 @@ crt = ["botocore[crt] (>=1.21.0,<2.0a0)"]
 
 [[package]]
 name = "botocore"
-version = "1.35.37"
+version = "1.35.38"
 description = "Low-level, data-driven core of boto 3."
 optional = false
 python-versions = ">=3.8"
 files = [
-    {file = "botocore-1.35.37-py3-none-any.whl", hash = "sha256:64f965d4ba7adb8d79ce044c3aef7356e05dd74753cf7e9115b80f477845d920"},
-    {file = "botocore-1.35.37.tar.gz", hash = "sha256:b2b4d29bafd95b698344f2f0577bb67064adbf1735d8a0e3c7473daa59c23ba6"},
+    {file = "botocore-1.35.38-py3-none-any.whl", hash = "sha256:2eb17d32fa2d3bb5d475132a83564d28e3acc2161534f24b75a54418a1d51359"},
+    {file = "botocore-1.35.38.tar.gz", hash = "sha256:55d9305c44e5ba29476df456120fa4fb919f03f066afa82f2ae400485e7465f4"},
 ]
 
 [package.dependencies]
@@ -2066,6 +2066,20 @@ wrapt = ">=1.10,<2"
 [package.extras]
 dev = ["PyTest", "PyTest-Cov", "bump2version (<1)", "sphinx (<2)", "tox"]
 
+[[package]]
+name = "deprecation"
+version = "2.1.0"
+description = "A library to handle automated deprecations"
+optional = false
+python-versions = "*"
+files = [
+    {file = "deprecation-2.1.0-py2.py3-none-any.whl", hash = "sha256:a10811591210e1fb0e768a8c25517cabeabcba6f0bf96564f8ff45189f90b14a"},
+    {file = "deprecation-2.1.0.tar.gz", hash = "sha256:72b3bde64e5d778694b0cf68178aed03d15e15477116add3fb773e581f9518ff"},
+]
+
+[package.dependencies]
+packaging = "*"
+
 [[package]]
 name = "dill"
 version = "0.3.9"
@@ -3366,6 +3380,21 @@ protobuf = ">=3.19.5,<3.20.0 || >3.20.0,<3.20.1 || >3.20.1,<4.21.1 || >4.21.1,<4
 [package.extras]
 grpc = ["grpcio (>=1.44.0,<2.0.0.dev0)"]
 
+[[package]]
+name = "gotrue"
+version = "2.9.2"
+description = "Python Client Library for Supabase Auth"
+optional = false
+python-versions = "<4.0,>=3.8"
+files = [
+    {file = "gotrue-2.9.2-py3-none-any.whl", hash = "sha256:fcd5279e8f1cc630f3ac35af5485fe39f8030b23906776920d2c32a4e308cff4"},
+    {file = "gotrue-2.9.2.tar.gz", hash = "sha256:57b3245e916c5efbf19a21b1181011a903c1276bb1df2d847558f2f24f29abb2"},
+]
+
+[package.dependencies]
+httpx = {version = ">=0.26,<0.28", extras = ["http2"]}
+pydantic = ">=1.10,<3"
+
 [[package]]
 name = "greenlet"
 version = "3.1.1"
@@ -4408,13 +4437,13 @@ openai = ["openai (>=0.27.8)"]
 
 [[package]]
 name = "langsmith"
-version = "0.1.133"
+version = "0.1.134"
 description = "Client library to connect to the LangSmith LLM Tracing and Evaluation Platform."
 optional = false
 python-versions = "<4.0,>=3.8.1"
 files = [
-    {file = "langsmith-0.1.133-py3-none-any.whl", hash = "sha256:82e837a6039c483beadbe19c2ba7ebafbd402d3e8105234f5ef334425cff7b45"},
-    {file = "langsmith-0.1.133.tar.gz", hash = "sha256:7bfd8bef166b9a64ee540a11bee4aa7bf43b1d9229f95b0fc19086454955185d"},
+    {file = "langsmith-0.1.134-py3-none-any.whl", hash = "sha256:ada98ad80ef38807725f32441a472da3dd28394010877751f48f458d3289da04"},
+    {file = "langsmith-0.1.134.tar.gz", hash = "sha256:23abee3b508875a0e63c602afafffc02442a19cfd88f9daae05b3e9054fd6b61"},
 ]
 
 [package.dependencies]
@@ -6395,6 +6424,23 @@ docs = ["sphinx (>=1.7.1)"]
 redis = ["redis"]
 tests = ["pytest (>=5.4.1)", "pytest-cov (>=2.8.1)", "pytest-mypy (>=0.8.0)", "pytest-timeout (>=2.1.0)", "redis", "sphinx (>=6.0.0)", "types-redis"]
 
+[[package]]
+name = "postgrest"
+version = "0.17.1"
+description = "PostgREST client for Python. This library provides an ORM interface to PostgREST."
+optional = false
+python-versions = "<4.0,>=3.8"
+files = [
+    {file = "postgrest-0.17.1-py3-none-any.whl", hash = "sha256:ec1d00dc8532fe5ffb342cfc7c4e610a1e0e2272eb14f78f9b2b61094f9be510"},
+    {file = "postgrest-0.17.1.tar.gz", hash = "sha256:e31d9977dbb80dc5f9fdd4d444014686606692dc4ddb9adc85639e56c6d54c92"},
+]
+
+[package.dependencies]
+deprecation = ">=2.1.0,<3.0.0"
+httpx = {version = ">=0.26,<0.28", extras = ["http2"]}
+pydantic = ">=1.9,<3.0"
+strenum = ">=0.4.9,<0.5.0"
+
 [[package]]
 name = "posthog"
 version = "3.7.0"
@@ -7695,6 +7741,23 @@ dev = ["coveralls", "m2r", "pycodestyle", "pyflakes", "pylint", "pytest", "pytes
 docs = ["m2r", "sphinx"]
 test = ["coveralls", "pycodestyle", "pyflakes", "pylint", "pytest", "pytest-benchmark", "pytest-cov"]
 
+[[package]]
+name = "realtime"
+version = "2.0.2"
+description = ""
+optional = false
+python-versions = "<4.0,>=3.9"
+files = [
+    {file = "realtime-2.0.2-py3-none-any.whl", hash = "sha256:2634c915bc38807f2013f21e8bcc4d2f79870dfd81460ddb9393883d0489928a"},
+    {file = "realtime-2.0.2.tar.gz", hash = "sha256:519da9325b3b8102139d51785013d592f6b2403d81fa21d838a0b0234723ed7d"},
+]
+
+[package.dependencies]
+aiohttp = ">=3.10.2,<4.0.0"
+python-dateutil = ">=2.8.1,<3.0.0"
+typing-extensions = ">=4.12.2,<5.0.0"
+websockets = ">=11,<13"
+
 [[package]]
 name = "redis"
 version = "5.0.8"
@@ -8578,19 +8641,20 @@ files = [
 
 [[package]]
 name = "simple-websocket"
-version = "1.0.0"
+version = "1.1.0"
 description = "Simple WebSocket server and client for Python"
 optional = false
 python-versions = ">=3.6"
 files = [
-    {file = "simple-websocket-1.0.0.tar.gz", hash = "sha256:17d2c72f4a2bd85174a97e3e4c88b01c40c3f81b7b648b0cc3ce1305968928c8"},
-    {file = "simple_websocket-1.0.0-py3-none-any.whl", hash = "sha256:1d5bf585e415eaa2083e2bcf02a3ecf91f9712e7b3e6b9fa0b461ad04e0837bc"},
+    {file = "simple_websocket-1.1.0-py3-none-any.whl", hash = "sha256:4af6069630a38ed6c561010f0e11a5bc0d4ca569b36306eb257cd9a192497c8c"},
+    {file = "simple_websocket-1.1.0.tar.gz", hash = "sha256:7939234e7aa067c534abdab3a9ed933ec9ce4691b0713c78acb195560aa52ae4"},
 ]
 
 [package.dependencies]
 wsproto = "*"
 
 [package.extras]
+dev = ["flake8", "pytest", "pytest-cov", "tox"]
 docs = ["sphinx"]
 
 [[package]]
@@ -8767,6 +8831,38 @@ anyio = ">=3.4.0,<5"
 [package.extras]
 full = ["httpx (>=0.22.0)", "itsdangerous", "jinja2", "python-multipart (>=0.0.7)", "pyyaml"]
 
+[[package]]
+name = "storage3"
+version = "0.8.1"
+description = "Supabase Storage client for Python."
+optional = false
+python-versions = "<4.0,>=3.8"
+files = [
+    {file = "storage3-0.8.1-py3-none-any.whl", hash = "sha256:0b21205f43eaf0d1dd33bde6c6d0612f88524b7865f017d2ae9827e3f63d9cdc"},
+    {file = "storage3-0.8.1.tar.gz", hash = "sha256:ea60b68b2221b3868ccc1a7f1294d57d0d9c51642cdc639d8115fe5d0adc8892"},
+]
+
+[package.dependencies]
+httpx = {version = ">=0.26,<0.28", extras = ["http2"]}
+python-dateutil = ">=2.8.2,<3.0.0"
+typing-extensions = ">=4.2.0,<5.0.0"
+
+[[package]]
+name = "strenum"
+version = "0.4.15"
+description = "An Enum that inherits from str."
+optional = false
+python-versions = "*"
+files = [
+    {file = "StrEnum-0.4.15-py3-none-any.whl", hash = "sha256:a30cda4af7cc6b5bf52c8055bc4bf4b2b6b14a93b574626da33df53cf7740659"},
+    {file = "StrEnum-0.4.15.tar.gz", hash = "sha256:878fb5ab705442070e4dd1929bb5e2249511c0bcf2b0eeacf3bcd80875c82eff"},
+]
+
+[package.extras]
+docs = ["myst-parser[linkify]", "sphinx", "sphinx-rtd-theme"]
+release = ["twine"]
+test = ["pylint", "pytest", "pytest-black", "pytest-cov", "pytest-pylint"]
+
 [[package]]
 name = "strictyaml"
 version = "1.7.3"
@@ -8781,6 +8877,40 @@ files = [
 [package.dependencies]
 python-dateutil = ">=2.6.0"
 
+[[package]]
+name = "supabase"
+version = "2.8.1"
+description = "Supabase client for Python."
+optional = false
+python-versions = "<4.0,>=3.9"
+files = [
+    {file = "supabase-2.8.1-py3-none-any.whl", hash = "sha256:dfa8bef89b54129093521d5bba2136ff765baf67cd76d8ad0aa4984d61a7815c"},
+    {file = "supabase-2.8.1.tar.gz", hash = "sha256:711c70e6acd9e2ff48ca0dc0b1bb70c01c25378cc5189ec9f5ed9655b30bc41d"},
+]
+
+[package.dependencies]
+gotrue = ">=2.7.0,<3.0.0"
+httpx = ">=0.24,<0.28"
+postgrest = ">=0.17.0,<0.18.0"
+realtime = ">=2.0.0,<3.0.0"
+storage3 = ">=0.8.0,<0.9.0"
+supafunc = ">=0.6.0,<0.7.0"
+typing-extensions = ">=4.12.2,<5.0.0"
+
+[[package]]
+name = "supafunc"
+version = "0.6.1"
+description = "Library for Supabase Functions"
+optional = false
+python-versions = "<4.0,>=3.8"
+files = [
+    {file = "supafunc-0.6.1-py3-none-any.whl", hash = "sha256:01aeeeb4bf429977664454a32c86418345140faf6d2e6eb0636d52e4547c5fbb"},
+    {file = "supafunc-0.6.1.tar.gz", hash = "sha256:3c8761e3999336ccdb7550498a395fd08afc8469382f55ea56f7f640e5a909aa"},
+]
+
+[package.dependencies]
+httpx = {version = ">=0.26,<0.28", extras = ["http2"]}
+
 [[package]]
 name = "sympy"
 version = "1.13.3"
@@ -8855,13 +8985,13 @@ test = ["pytest", "tornado (>=4.5)", "typeguard"]
 
 [[package]]
 name = "tencentcloud-sdk-python-common"
-version = "3.0.1246"
+version = "3.0.1247"
 description = "Tencent Cloud Common SDK for Python"
 optional = false
 python-versions = "*"
 files = [
-    {file = "tencentcloud-sdk-python-common-3.0.1246.tar.gz", hash = "sha256:a724d53d5dd6c68beff9ae1c2eb7737889d67f7724819ec5189e878e5322940b"},
-    {file = "tencentcloud_sdk_python_common-3.0.1246-py2.py3-none-any.whl", hash = "sha256:e82a103cc4f1a8d07a83600604eba89e46899b005328b921dbcba9f73b3e819b"},
+    {file = "tencentcloud-sdk-python-common-3.0.1247.tar.gz", hash = "sha256:1467ac3eaaa5b5d299570ba781903debc4be32dbb3f0f39929a357531ab89170"},
+    {file = "tencentcloud_sdk_python_common-3.0.1247-py2.py3-none-any.whl", hash = "sha256:9829d2299c85a2494d6d816247345e98abd2f936cd309e1f67847243f5235091"},
 ]
 
 [package.dependencies]
@@ -8869,17 +8999,17 @@ requests = ">=2.16.0"
 
 [[package]]
 name = "tencentcloud-sdk-python-hunyuan"
-version = "3.0.1246"
+version = "3.0.1247"
 description = "Tencent Cloud Hunyuan SDK for Python"
 optional = false
 python-versions = "*"
 files = [
-    {file = "tencentcloud-sdk-python-hunyuan-3.0.1246.tar.gz", hash = "sha256:099db4a80e7788e1ca09b8be45758e47a6c5976e304857f26621f958cca5b39b"},
-    {file = "tencentcloud_sdk_python_hunyuan-3.0.1246-py2.py3-none-any.whl", hash = "sha256:14ad0a6787bb579a9270e6fe6ee81131a286c2f1b56cac8d7de0056983b6548f"},
+    {file = "tencentcloud-sdk-python-hunyuan-3.0.1247.tar.gz", hash = "sha256:85b7332ec55f891a3b4d776e6b30ee2a44cc08c70b689615805aadff6e424fdd"},
+    {file = "tencentcloud_sdk_python_hunyuan-3.0.1247-py2.py3-none-any.whl", hash = "sha256:69fdb886616e53ce02e848e5a1a8b36922db731457b07365f230ffb0aa472b5b"},
 ]
 
 [package.dependencies]
-tencentcloud-sdk-python-common = "3.0.1246"
+tencentcloud-sdk-python-common = "3.0.1247"
 
 [[package]]
 name = "threadpoolctl"
@@ -9855,97 +9985,83 @@ test = ["websockets"]
 
 [[package]]
 name = "websockets"
-version = "13.1"
+version = "12.0"
 description = "An implementation of the WebSocket Protocol (RFC 6455 & 7692)"
 optional = false
 python-versions = ">=3.8"
 files = [
-    {file = "websockets-13.1-cp310-cp310-macosx_10_9_universal2.whl", hash = "sha256:f48c749857f8fb598fb890a75f540e3221d0976ed0bf879cf3c7eef34151acee"},
-    {file = "websockets-13.1-cp310-cp310-macosx_10_9_x86_64.whl", hash = "sha256:c7e72ce6bda6fb9409cc1e8164dd41d7c91466fb599eb047cfda72fe758a34a7"},
-    {file = "websockets-13.1-cp310-cp310-macosx_11_0_arm64.whl", hash = "sha256:f779498eeec470295a2b1a5d97aa1bc9814ecd25e1eb637bd9d1c73a327387f6"},
-    {file = "websockets-13.1-cp310-cp310-manylinux_2_17_aarch64.manylinux2014_aarch64.whl", hash = "sha256:4676df3fe46956fbb0437d8800cd5f2b6d41143b6e7e842e60554398432cf29b"},
-    {file = "websockets-13.1-cp310-cp310-manylinux_2_5_i686.manylinux1_i686.manylinux_2_17_i686.manylinux2014_i686.whl", hash = "sha256:a7affedeb43a70351bb811dadf49493c9cfd1ed94c9c70095fd177e9cc1541fa"},
-    {file = "websockets-13.1-cp310-cp310-manylinux_2_5_x86_64.manylinux1_x86_64.manylinux_2_17_x86_64.manylinux2014_x86_64.whl", hash = "sha256:1971e62d2caa443e57588e1d82d15f663b29ff9dfe7446d9964a4b6f12c1e700"},
-    {file = "websockets-13.1-cp310-cp310-musllinux_1_2_aarch64.whl", hash = "sha256:5f2e75431f8dc4a47f31565a6e1355fb4f2ecaa99d6b89737527ea917066e26c"},
-    {file = "websockets-13.1-cp310-cp310-musllinux_1_2_i686.whl", hash = "sha256:58cf7e75dbf7e566088b07e36ea2e3e2bd5676e22216e4cad108d4df4a7402a0"},
-    {file = "websockets-13.1-cp310-cp310-musllinux_1_2_x86_64.whl", hash = "sha256:c90d6dec6be2c7d03378a574de87af9b1efea77d0c52a8301dd831ece938452f"},
-    {file = "websockets-13.1-cp310-cp310-win32.whl", hash = "sha256:730f42125ccb14602f455155084f978bd9e8e57e89b569b4d7f0f0c17a448ffe"},
-    {file = "websockets-13.1-cp310-cp310-win_amd64.whl", hash = "sha256:5993260f483d05a9737073be197371940c01b257cc45ae3f1d5d7adb371b266a"},
-    {file = "websockets-13.1-cp311-cp311-macosx_10_9_universal2.whl", hash = "sha256:61fc0dfcda609cda0fc9fe7977694c0c59cf9d749fbb17f4e9483929e3c48a19"},
-    {file = "websockets-13.1-cp311-cp311-macosx_10_9_x86_64.whl", hash = "sha256:ceec59f59d092c5007e815def4ebb80c2de330e9588e101cf8bd94c143ec78a5"},
-    {file = "websockets-13.1-cp311-cp311-macosx_11_0_arm64.whl", hash = "sha256:c1dca61c6db1166c48b95198c0b7d9c990b30c756fc2923cc66f68d17dc558fd"},
-    {file = "websockets-13.1-cp311-cp311-manylinux_2_17_aarch64.manylinux2014_aarch64.whl", hash = "sha256:308e20f22c2c77f3f39caca508e765f8725020b84aa963474e18c59accbf4c02"},
-    {file = "websockets-13.1-cp311-cp311-manylinux_2_5_i686.manylinux1_i686.manylinux_2_17_i686.manylinux2014_i686.whl", hash = "sha256:62d516c325e6540e8a57b94abefc3459d7dab8ce52ac75c96cad5549e187e3a7"},
-    {file = "websockets-13.1-cp311-cp311-manylinux_2_5_x86_64.manylinux1_x86_64.manylinux_2_17_x86_64.manylinux2014_x86_64.whl", hash = "sha256:87c6e35319b46b99e168eb98472d6c7d8634ee37750d7693656dc766395df096"},
-    {file = "websockets-13.1-cp311-cp311-musllinux_1_2_aarch64.whl", hash = "sha256:5f9fee94ebafbc3117c30be1844ed01a3b177bb6e39088bc6b2fa1dc15572084"},
-    {file = "websockets-13.1-cp311-cp311-musllinux_1_2_i686.whl", hash = "sha256:7c1e90228c2f5cdde263253fa5db63e6653f1c00e7ec64108065a0b9713fa1b3"},
-    {file = "websockets-13.1-cp311-cp311-musllinux_1_2_x86_64.whl", hash = "sha256:6548f29b0e401eea2b967b2fdc1c7c7b5ebb3eeb470ed23a54cd45ef078a0db9"},
-    {file = "websockets-13.1-cp311-cp311-win32.whl", hash = "sha256:c11d4d16e133f6df8916cc5b7e3e96ee4c44c936717d684a94f48f82edb7c92f"},
-    {file = "websockets-13.1-cp311-cp311-win_amd64.whl", hash = "sha256:d04f13a1d75cb2b8382bdc16ae6fa58c97337253826dfe136195b7f89f661557"},
-    {file = "websockets-13.1-cp312-cp312-macosx_10_9_universal2.whl", hash = "sha256:9d75baf00138f80b48f1eac72ad1535aac0b6461265a0bcad391fc5aba875cfc"},
-    {file = "websockets-13.1-cp312-cp312-macosx_10_9_x86_64.whl", hash = "sha256:9b6f347deb3dcfbfde1c20baa21c2ac0751afaa73e64e5b693bb2b848efeaa49"},
-    {file = "websockets-13.1-cp312-cp312-macosx_11_0_arm64.whl", hash = "sha256:de58647e3f9c42f13f90ac7e5f58900c80a39019848c5547bc691693098ae1bd"},
-    {file = "websockets-13.1-cp312-cp312-manylinux_2_17_aarch64.manylinux2014_aarch64.whl", hash = "sha256:a1b54689e38d1279a51d11e3467dd2f3a50f5f2e879012ce8f2d6943f00e83f0"},
-    {file = "websockets-13.1-cp312-cp312-manylinux_2_5_i686.manylinux1_i686.manylinux_2_17_i686.manylinux2014_i686.whl", hash = "sha256:cf1781ef73c073e6b0f90af841aaf98501f975d306bbf6221683dd594ccc52b6"},
-    {file = "websockets-13.1-cp312-cp312-manylinux_2_5_x86_64.manylinux1_x86_64.manylinux_2_17_x86_64.manylinux2014_x86_64.whl", hash = "sha256:8d23b88b9388ed85c6faf0e74d8dec4f4d3baf3ecf20a65a47b836d56260d4b9"},
-    {file = "websockets-13.1-cp312-cp312-musllinux_1_2_aarch64.whl", hash = "sha256:3c78383585f47ccb0fcf186dcb8a43f5438bd7d8f47d69e0b56f71bf431a0a68"},
-    {file = "websockets-13.1-cp312-cp312-musllinux_1_2_i686.whl", hash = "sha256:d6d300f8ec35c24025ceb9b9019ae9040c1ab2f01cddc2bcc0b518af31c75c14"},
-    {file = "websockets-13.1-cp312-cp312-musllinux_1_2_x86_64.whl", hash = "sha256:a9dcaf8b0cc72a392760bb8755922c03e17a5a54e08cca58e8b74f6902b433cf"},
-    {file = "websockets-13.1-cp312-cp312-win32.whl", hash = "sha256:2f85cf4f2a1ba8f602298a853cec8526c2ca42a9a4b947ec236eaedb8f2dc80c"},
-    {file = "websockets-13.1-cp312-cp312-win_amd64.whl", hash = "sha256:38377f8b0cdeee97c552d20cf1865695fcd56aba155ad1b4ca8779a5b6ef4ac3"},
-    {file = "websockets-13.1-cp313-cp313-macosx_10_13_universal2.whl", hash = "sha256:a9ab1e71d3d2e54a0aa646ab6d4eebfaa5f416fe78dfe4da2839525dc5d765c6"},
-    {file = "websockets-13.1-cp313-cp313-macosx_10_13_x86_64.whl", hash = "sha256:b9d7439d7fab4dce00570bb906875734df13d9faa4b48e261c440a5fec6d9708"},
-    {file = "websockets-13.1-cp313-cp313-macosx_11_0_arm64.whl", hash = "sha256:327b74e915cf13c5931334c61e1a41040e365d380f812513a255aa804b183418"},
-    {file = "websockets-13.1-cp313-cp313-manylinux_2_17_aarch64.manylinux2014_aarch64.whl", hash = "sha256:325b1ccdbf5e5725fdcb1b0e9ad4d2545056479d0eee392c291c1bf76206435a"},
-    {file = "websockets-13.1-cp313-cp313-manylinux_2_5_i686.manylinux1_i686.manylinux_2_17_i686.manylinux2014_i686.whl", hash = "sha256:346bee67a65f189e0e33f520f253d5147ab76ae42493804319b5716e46dddf0f"},
-    {file = "websockets-13.1-cp313-cp313-manylinux_2_5_x86_64.manylinux1_x86_64.manylinux_2_17_x86_64.manylinux2014_x86_64.whl", hash = "sha256:91a0fa841646320ec0d3accdff5b757b06e2e5c86ba32af2e0815c96c7a603c5"},
-    {file = "websockets-13.1-cp313-cp313-musllinux_1_2_aarch64.whl", hash = "sha256:18503d2c5f3943e93819238bf20df71982d193f73dcecd26c94514f417f6b135"},
-    {file = "websockets-13.1-cp313-cp313-musllinux_1_2_i686.whl", hash = "sha256:a9cd1af7e18e5221d2878378fbc287a14cd527fdd5939ed56a18df8a31136bb2"},
-    {file = "websockets-13.1-cp313-cp313-musllinux_1_2_x86_64.whl", hash = "sha256:70c5be9f416aa72aab7a2a76c90ae0a4fe2755c1816c153c1a2bcc3333ce4ce6"},
-    {file = "websockets-13.1-cp313-cp313-win32.whl", hash = "sha256:624459daabeb310d3815b276c1adef475b3e6804abaf2d9d2c061c319f7f187d"},
-    {file = "websockets-13.1-cp313-cp313-win_amd64.whl", hash = "sha256:c518e84bb59c2baae725accd355c8dc517b4a3ed8db88b4bc93c78dae2974bf2"},
-    {file = "websockets-13.1-cp38-cp38-macosx_10_9_universal2.whl", hash = "sha256:c7934fd0e920e70468e676fe7f1b7261c1efa0d6c037c6722278ca0228ad9d0d"},
-    {file = "websockets-13.1-cp38-cp38-macosx_10_9_x86_64.whl", hash = "sha256:149e622dc48c10ccc3d2760e5f36753db9cacf3ad7bc7bbbfd7d9c819e286f23"},
-    {file = "websockets-13.1-cp38-cp38-macosx_11_0_arm64.whl", hash = "sha256:a569eb1b05d72f9bce2ebd28a1ce2054311b66677fcd46cf36204ad23acead8c"},
-    {file = "websockets-13.1-cp38-cp38-manylinux_2_17_aarch64.manylinux2014_aarch64.whl", hash = "sha256:95df24ca1e1bd93bbca51d94dd049a984609687cb2fb08a7f2c56ac84e9816ea"},
-    {file = "websockets-13.1-cp38-cp38-manylinux_2_5_i686.manylinux1_i686.manylinux_2_17_i686.manylinux2014_i686.whl", hash = "sha256:d8dbb1bf0c0a4ae8b40bdc9be7f644e2f3fb4e8a9aca7145bfa510d4a374eeb7"},
-    {file = "websockets-13.1-cp38-cp38-manylinux_2_5_x86_64.manylinux1_x86_64.manylinux_2_17_x86_64.manylinux2014_x86_64.whl", hash = "sha256:035233b7531fb92a76beefcbf479504db8c72eb3bff41da55aecce3a0f729e54"},
-    {file = "websockets-13.1-cp38-cp38-musllinux_1_2_aarch64.whl", hash = "sha256:e4450fc83a3df53dec45922b576e91e94f5578d06436871dce3a6be38e40f5db"},
-    {file = "websockets-13.1-cp38-cp38-musllinux_1_2_i686.whl", hash = "sha256:463e1c6ec853202dd3657f156123d6b4dad0c546ea2e2e38be2b3f7c5b8e7295"},
-    {file = "websockets-13.1-cp38-cp38-musllinux_1_2_x86_64.whl", hash = "sha256:6d6855bbe70119872c05107e38fbc7f96b1d8cb047d95c2c50869a46c65a8e96"},
-    {file = "websockets-13.1-cp38-cp38-win32.whl", hash = "sha256:204e5107f43095012b00f1451374693267adbb832d29966a01ecc4ce1db26faf"},
-    {file = "websockets-13.1-cp38-cp38-win_amd64.whl", hash = "sha256:485307243237328c022bc908b90e4457d0daa8b5cf4b3723fd3c4a8012fce4c6"},
-    {file = "websockets-13.1-cp39-cp39-macosx_10_9_universal2.whl", hash = "sha256:9b37c184f8b976f0c0a231a5f3d6efe10807d41ccbe4488df8c74174805eea7d"},
-    {file = "websockets-13.1-cp39-cp39-macosx_10_9_x86_64.whl", hash = "sha256:163e7277e1a0bd9fb3c8842a71661ad19c6aa7bb3d6678dc7f89b17fbcc4aeb7"},
-    {file = "websockets-13.1-cp39-cp39-macosx_11_0_arm64.whl", hash = "sha256:4b889dbd1342820cc210ba44307cf75ae5f2f96226c0038094455a96e64fb07a"},
-    {file = "websockets-13.1-cp39-cp39-manylinux_2_17_aarch64.manylinux2014_aarch64.whl", hash = "sha256:586a356928692c1fed0eca68b4d1c2cbbd1ca2acf2ac7e7ebd3b9052582deefa"},
-    {file = "websockets-13.1-cp39-cp39-manylinux_2_5_i686.manylinux1_i686.manylinux_2_17_i686.manylinux2014_i686.whl", hash = "sha256:7bd6abf1e070a6b72bfeb71049d6ad286852e285f146682bf30d0296f5fbadfa"},
-    {file = "websockets-13.1-cp39-cp39-manylinux_2_5_x86_64.manylinux1_x86_64.manylinux_2_17_x86_64.manylinux2014_x86_64.whl", hash = "sha256:6d2aad13a200e5934f5a6767492fb07151e1de1d6079c003ab31e1823733ae79"},
-    {file = "websockets-13.1-cp39-cp39-musllinux_1_2_aarch64.whl", hash = "sha256:df01aea34b6e9e33572c35cd16bae5a47785e7d5c8cb2b54b2acdb9678315a17"},
-    {file = "websockets-13.1-cp39-cp39-musllinux_1_2_i686.whl", hash = "sha256:e54affdeb21026329fb0744ad187cf812f7d3c2aa702a5edb562b325191fcab6"},
-    {file = "websockets-13.1-cp39-cp39-musllinux_1_2_x86_64.whl", hash = "sha256:9ef8aa8bdbac47f4968a5d66462a2a0935d044bf35c0e5a8af152d58516dbeb5"},
-    {file = "websockets-13.1-cp39-cp39-win32.whl", hash = "sha256:deeb929efe52bed518f6eb2ddc00cc496366a14c726005726ad62c2dd9017a3c"},
-    {file = "websockets-13.1-cp39-cp39-win_amd64.whl", hash = "sha256:7c65ffa900e7cc958cd088b9a9157a8141c991f8c53d11087e6fb7277a03f81d"},
-    {file = "websockets-13.1-pp310-pypy310_pp73-macosx_10_15_x86_64.whl", hash = "sha256:5dd6da9bec02735931fccec99d97c29f47cc61f644264eb995ad6c0c27667238"},
-    {file = "websockets-13.1-pp310-pypy310_pp73-macosx_11_0_arm64.whl", hash = "sha256:2510c09d8e8df777177ee3d40cd35450dc169a81e747455cc4197e63f7e7bfe5"},
-    {file = "websockets-13.1-pp310-pypy310_pp73-manylinux_2_17_aarch64.manylinux2014_aarch64.whl", hash = "sha256:f1c3cf67185543730888b20682fb186fc8d0fa6f07ccc3ef4390831ab4b388d9"},
-    {file = "websockets-13.1-pp310-pypy310_pp73-manylinux_2_5_i686.manylinux1_i686.manylinux_2_17_i686.manylinux2014_i686.whl", hash = "sha256:bcc03c8b72267e97b49149e4863d57c2d77f13fae12066622dc78fe322490fe6"},
-    {file = "websockets-13.1-pp310-pypy310_pp73-manylinux_2_5_x86_64.manylinux1_x86_64.manylinux_2_17_x86_64.manylinux2014_x86_64.whl", hash = "sha256:004280a140f220c812e65f36944a9ca92d766b6cc4560be652a0a3883a79ed8a"},
-    {file = "websockets-13.1-pp310-pypy310_pp73-win_amd64.whl", hash = "sha256:e2620453c075abeb0daa949a292e19f56de518988e079c36478bacf9546ced23"},
-    {file = "websockets-13.1-pp38-pypy38_pp73-macosx_10_9_x86_64.whl", hash = "sha256:9156c45750b37337f7b0b00e6248991a047be4aa44554c9886fe6bdd605aab3b"},
-    {file = "websockets-13.1-pp38-pypy38_pp73-macosx_11_0_arm64.whl", hash = "sha256:80c421e07973a89fbdd93e6f2003c17d20b69010458d3a8e37fb47874bd67d51"},
-    {file = "websockets-13.1-pp38-pypy38_pp73-manylinux_2_17_aarch64.manylinux2014_aarch64.whl", hash = "sha256:82d0ba76371769d6a4e56f7e83bb8e81846d17a6190971e38b5de108bde9b0d7"},
-    {file = "websockets-13.1-pp38-pypy38_pp73-manylinux_2_5_i686.manylinux1_i686.manylinux_2_17_i686.manylinux2014_i686.whl", hash = "sha256:e9875a0143f07d74dc5e1ded1c4581f0d9f7ab86c78994e2ed9e95050073c94d"},
-    {file = "websockets-13.1-pp38-pypy38_pp73-manylinux_2_5_x86_64.manylinux1_x86_64.manylinux_2_17_x86_64.manylinux2014_x86_64.whl", hash = "sha256:a11e38ad8922c7961447f35c7b17bffa15de4d17c70abd07bfbe12d6faa3e027"},
-    {file = "websockets-13.1-pp38-pypy38_pp73-win_amd64.whl", hash = "sha256:4059f790b6ae8768471cddb65d3c4fe4792b0ab48e154c9f0a04cefaabcd5978"},
-    {file = "websockets-13.1-pp39-pypy39_pp73-macosx_10_15_x86_64.whl", hash = "sha256:25c35bf84bf7c7369d247f0b8cfa157f989862c49104c5cf85cb5436a641d93e"},
-    {file = "websockets-13.1-pp39-pypy39_pp73-macosx_11_0_arm64.whl", hash = "sha256:83f91d8a9bb404b8c2c41a707ac7f7f75b9442a0a876df295de27251a856ad09"},
-    {file = "websockets-13.1-pp39-pypy39_pp73-manylinux_2_17_aarch64.manylinux2014_aarch64.whl", hash = "sha256:7a43cfdcddd07f4ca2b1afb459824dd3c6d53a51410636a2c7fc97b9a8cf4842"},
-    {file = "websockets-13.1-pp39-pypy39_pp73-manylinux_2_5_i686.manylinux1_i686.manylinux_2_17_i686.manylinux2014_i686.whl", hash = "sha256:48a2ef1381632a2f0cb4efeff34efa97901c9fbc118e01951ad7cfc10601a9bb"},
-    {file = "websockets-13.1-pp39-pypy39_pp73-manylinux_2_5_x86_64.manylinux1_x86_64.manylinux_2_17_x86_64.manylinux2014_x86_64.whl", hash = "sha256:459bf774c754c35dbb487360b12c5727adab887f1622b8aed5755880a21c4a20"},
-    {file = "websockets-13.1-pp39-pypy39_pp73-win_amd64.whl", hash = "sha256:95858ca14a9f6fa8413d29e0a585b31b278388aa775b8a81fa24830123874678"},
-    {file = "websockets-13.1-py3-none-any.whl", hash = "sha256:a9a396a6ad26130cdae92ae10c36af09d9bfe6cafe69670fd3b6da9b07b4044f"},
-    {file = "websockets-13.1.tar.gz", hash = "sha256:a3b3366087c1bc0a2795111edcadddb8b3b59509d5db5d7ea3fdd69f954a8878"},
+    {file = "websockets-12.0-cp310-cp310-macosx_10_9_universal2.whl", hash = "sha256:d554236b2a2006e0ce16315c16eaa0d628dab009c33b63ea03f41c6107958374"},
+    {file = "websockets-12.0-cp310-cp310-macosx_10_9_x86_64.whl", hash = "sha256:2d225bb6886591b1746b17c0573e29804619c8f755b5598d875bb4235ea639be"},
+    {file = "websockets-12.0-cp310-cp310-macosx_11_0_arm64.whl", hash = "sha256:eb809e816916a3b210bed3c82fb88eaf16e8afcf9c115ebb2bacede1797d2547"},
+    {file = "websockets-12.0-cp310-cp310-manylinux_2_17_aarch64.manylinux2014_aarch64.whl", hash = "sha256:c588f6abc13f78a67044c6b1273a99e1cf31038ad51815b3b016ce699f0d75c2"},
+    {file = "websockets-12.0-cp310-cp310-manylinux_2_5_i686.manylinux1_i686.manylinux_2_17_i686.manylinux2014_i686.whl", hash = "sha256:5aa9348186d79a5f232115ed3fa9020eab66d6c3437d72f9d2c8ac0c6858c558"},
+    {file = "websockets-12.0-cp310-cp310-manylinux_2_5_x86_64.manylinux1_x86_64.manylinux_2_17_x86_64.manylinux2014_x86_64.whl", hash = "sha256:6350b14a40c95ddd53e775dbdbbbc59b124a5c8ecd6fbb09c2e52029f7a9f480"},
+    {file = "websockets-12.0-cp310-cp310-musllinux_1_1_aarch64.whl", hash = "sha256:70ec754cc2a769bcd218ed8d7209055667b30860ffecb8633a834dde27d6307c"},
+    {file = "websockets-12.0-cp310-cp310-musllinux_1_1_i686.whl", hash = "sha256:6e96f5ed1b83a8ddb07909b45bd94833b0710f738115751cdaa9da1fb0cb66e8"},
+    {file = "websockets-12.0-cp310-cp310-musllinux_1_1_x86_64.whl", hash = "sha256:4d87be612cbef86f994178d5186add3d94e9f31cc3cb499a0482b866ec477603"},
+    {file = "websockets-12.0-cp310-cp310-win32.whl", hash = "sha256:befe90632d66caaf72e8b2ed4d7f02b348913813c8b0a32fae1cc5fe3730902f"},
+    {file = "websockets-12.0-cp310-cp310-win_amd64.whl", hash = "sha256:363f57ca8bc8576195d0540c648aa58ac18cf85b76ad5202b9f976918f4219cf"},
+    {file = "websockets-12.0-cp311-cp311-macosx_10_9_universal2.whl", hash = "sha256:5d873c7de42dea355d73f170be0f23788cf3fa9f7bed718fd2830eefedce01b4"},
+    {file = "websockets-12.0-cp311-cp311-macosx_10_9_x86_64.whl", hash = "sha256:3f61726cae9f65b872502ff3c1496abc93ffbe31b278455c418492016e2afc8f"},
+    {file = "websockets-12.0-cp311-cp311-macosx_11_0_arm64.whl", hash = "sha256:ed2fcf7a07334c77fc8a230755c2209223a7cc44fc27597729b8ef5425aa61a3"},
+    {file = "websockets-12.0-cp311-cp311-manylinux_2_17_aarch64.manylinux2014_aarch64.whl", hash = "sha256:8e332c210b14b57904869ca9f9bf4ca32f5427a03eeb625da9b616c85a3a506c"},
+    {file = "websockets-12.0-cp311-cp311-manylinux_2_5_i686.manylinux1_i686.manylinux_2_17_i686.manylinux2014_i686.whl", hash = "sha256:5693ef74233122f8ebab026817b1b37fe25c411ecfca084b29bc7d6efc548f45"},
+    {file = "websockets-12.0-cp311-cp311-manylinux_2_5_x86_64.manylinux1_x86_64.manylinux_2_17_x86_64.manylinux2014_x86_64.whl", hash = "sha256:6e9e7db18b4539a29cc5ad8c8b252738a30e2b13f033c2d6e9d0549b45841c04"},
+    {file = "websockets-12.0-cp311-cp311-musllinux_1_1_aarch64.whl", hash = "sha256:6e2df67b8014767d0f785baa98393725739287684b9f8d8a1001eb2839031447"},
+    {file = "websockets-12.0-cp311-cp311-musllinux_1_1_i686.whl", hash = "sha256:bea88d71630c5900690fcb03161ab18f8f244805c59e2e0dc4ffadae0a7ee0ca"},
+    {file = "websockets-12.0-cp311-cp311-musllinux_1_1_x86_64.whl", hash = "sha256:dff6cdf35e31d1315790149fee351f9e52978130cef6c87c4b6c9b3baf78bc53"},
+    {file = "websockets-12.0-cp311-cp311-win32.whl", hash = "sha256:3e3aa8c468af01d70332a382350ee95f6986db479ce7af14d5e81ec52aa2b402"},
+    {file = "websockets-12.0-cp311-cp311-win_amd64.whl", hash = "sha256:25eb766c8ad27da0f79420b2af4b85d29914ba0edf69f547cc4f06ca6f1d403b"},
+    {file = "websockets-12.0-cp312-cp312-macosx_10_9_universal2.whl", hash = "sha256:0e6e2711d5a8e6e482cacb927a49a3d432345dfe7dea8ace7b5790df5932e4df"},
+    {file = "websockets-12.0-cp312-cp312-macosx_10_9_x86_64.whl", hash = "sha256:dbcf72a37f0b3316e993e13ecf32f10c0e1259c28ffd0a85cee26e8549595fbc"},
+    {file = "websockets-12.0-cp312-cp312-macosx_11_0_arm64.whl", hash = "sha256:12743ab88ab2af1d17dd4acb4645677cb7063ef4db93abffbf164218a5d54c6b"},
+    {file = "websockets-12.0-cp312-cp312-manylinux_2_17_aarch64.manylinux2014_aarch64.whl", hash = "sha256:7b645f491f3c48d3f8a00d1fce07445fab7347fec54a3e65f0725d730d5b99cb"},
+    {file = "websockets-12.0-cp312-cp312-manylinux_2_5_i686.manylinux1_i686.manylinux_2_17_i686.manylinux2014_i686.whl", hash = "sha256:9893d1aa45a7f8b3bc4510f6ccf8db8c3b62120917af15e3de247f0780294b92"},
+    {file = "websockets-12.0-cp312-cp312-manylinux_2_5_x86_64.manylinux1_x86_64.manylinux_2_17_x86_64.manylinux2014_x86_64.whl", hash = "sha256:1f38a7b376117ef7aff996e737583172bdf535932c9ca021746573bce40165ed"},
+    {file = "websockets-12.0-cp312-cp312-musllinux_1_1_aarch64.whl", hash = "sha256:f764ba54e33daf20e167915edc443b6f88956f37fb606449b4a5b10ba42235a5"},
+    {file = "websockets-12.0-cp312-cp312-musllinux_1_1_i686.whl", hash = "sha256:1e4b3f8ea6a9cfa8be8484c9221ec0257508e3a1ec43c36acdefb2a9c3b00aa2"},
+    {file = "websockets-12.0-cp312-cp312-musllinux_1_1_x86_64.whl", hash = "sha256:9fdf06fd06c32205a07e47328ab49c40fc1407cdec801d698a7c41167ea45113"},
+    {file = "websockets-12.0-cp312-cp312-win32.whl", hash = "sha256:baa386875b70cbd81798fa9f71be689c1bf484f65fd6fb08d051a0ee4e79924d"},
+    {file = "websockets-12.0-cp312-cp312-win_amd64.whl", hash = "sha256:ae0a5da8f35a5be197f328d4727dbcfafa53d1824fac3d96cdd3a642fe09394f"},
+    {file = "websockets-12.0-cp38-cp38-macosx_10_9_universal2.whl", hash = "sha256:5f6ffe2c6598f7f7207eef9a1228b6f5c818f9f4d53ee920aacd35cec8110438"},
+    {file = "websockets-12.0-cp38-cp38-macosx_10_9_x86_64.whl", hash = "sha256:9edf3fc590cc2ec20dc9d7a45108b5bbaf21c0d89f9fd3fd1685e223771dc0b2"},
+    {file = "websockets-12.0-cp38-cp38-macosx_11_0_arm64.whl", hash = "sha256:8572132c7be52632201a35f5e08348137f658e5ffd21f51f94572ca6c05ea81d"},
+    {file = "websockets-12.0-cp38-cp38-manylinux_2_17_aarch64.manylinux2014_aarch64.whl", hash = "sha256:604428d1b87edbf02b233e2c207d7d528460fa978f9e391bd8aaf9c8311de137"},
+    {file = "websockets-12.0-cp38-cp38-manylinux_2_5_i686.manylinux1_i686.manylinux_2_17_i686.manylinux2014_i686.whl", hash = "sha256:1a9d160fd080c6285e202327aba140fc9a0d910b09e423afff4ae5cbbf1c7205"},
+    {file = "websockets-12.0-cp38-cp38-manylinux_2_5_x86_64.manylinux1_x86_64.manylinux_2_17_x86_64.manylinux2014_x86_64.whl", hash = "sha256:87b4aafed34653e465eb77b7c93ef058516cb5acf3eb21e42f33928616172def"},
+    {file = "websockets-12.0-cp38-cp38-musllinux_1_1_aarch64.whl", hash = "sha256:b2ee7288b85959797970114deae81ab41b731f19ebcd3bd499ae9ca0e3f1d2c8"},
+    {file = "websockets-12.0-cp38-cp38-musllinux_1_1_i686.whl", hash = "sha256:7fa3d25e81bfe6a89718e9791128398a50dec6d57faf23770787ff441d851967"},
+    {file = "websockets-12.0-cp38-cp38-musllinux_1_1_x86_64.whl", hash = "sha256:a571f035a47212288e3b3519944f6bf4ac7bc7553243e41eac50dd48552b6df7"},
+    {file = "websockets-12.0-cp38-cp38-win32.whl", hash = "sha256:3c6cc1360c10c17463aadd29dd3af332d4a1adaa8796f6b0e9f9df1fdb0bad62"},
+    {file = "websockets-12.0-cp38-cp38-win_amd64.whl", hash = "sha256:1bf386089178ea69d720f8db6199a0504a406209a0fc23e603b27b300fdd6892"},
+    {file = "websockets-12.0-cp39-cp39-macosx_10_9_universal2.whl", hash = "sha256:ab3d732ad50a4fbd04a4490ef08acd0517b6ae6b77eb967251f4c263011a990d"},
+    {file = "websockets-12.0-cp39-cp39-macosx_10_9_x86_64.whl", hash = "sha256:a1d9697f3337a89691e3bd8dc56dea45a6f6d975f92e7d5f773bc715c15dde28"},
+    {file = "websockets-12.0-cp39-cp39-macosx_11_0_arm64.whl", hash = "sha256:1df2fbd2c8a98d38a66f5238484405b8d1d16f929bb7a33ed73e4801222a6f53"},
+    {file = "websockets-12.0-cp39-cp39-manylinux_2_17_aarch64.manylinux2014_aarch64.whl", hash = "sha256:23509452b3bc38e3a057382c2e941d5ac2e01e251acce7adc74011d7d8de434c"},
+    {file = "websockets-12.0-cp39-cp39-manylinux_2_5_i686.manylinux1_i686.manylinux_2_17_i686.manylinux2014_i686.whl", hash = "sha256:2e5fc14ec6ea568200ea4ef46545073da81900a2b67b3e666f04adf53ad452ec"},
+    {file = "websockets-12.0-cp39-cp39-manylinux_2_5_x86_64.manylinux1_x86_64.manylinux_2_17_x86_64.manylinux2014_x86_64.whl", hash = "sha256:46e71dbbd12850224243f5d2aeec90f0aaa0f2dde5aeeb8fc8df21e04d99eff9"},
+    {file = "websockets-12.0-cp39-cp39-musllinux_1_1_aarch64.whl", hash = "sha256:b81f90dcc6c85a9b7f29873beb56c94c85d6f0dac2ea8b60d995bd18bf3e2aae"},
+    {file = "websockets-12.0-cp39-cp39-musllinux_1_1_i686.whl", hash = "sha256:a02413bc474feda2849c59ed2dfb2cddb4cd3d2f03a2fedec51d6e959d9b608b"},
+    {file = "websockets-12.0-cp39-cp39-musllinux_1_1_x86_64.whl", hash = "sha256:bbe6013f9f791944ed31ca08b077e26249309639313fff132bfbf3ba105673b9"},
+    {file = "websockets-12.0-cp39-cp39-win32.whl", hash = "sha256:cbe83a6bbdf207ff0541de01e11904827540aa069293696dd528a6640bd6a5f6"},
+    {file = "websockets-12.0-cp39-cp39-win_amd64.whl", hash = "sha256:fc4e7fa5414512b481a2483775a8e8be7803a35b30ca805afa4998a84f9fd9e8"},
+    {file = "websockets-12.0-pp310-pypy310_pp73-macosx_10_9_x86_64.whl", hash = "sha256:248d8e2446e13c1d4326e0a6a4e9629cb13a11195051a73acf414812700badbd"},
+    {file = "websockets-12.0-pp310-pypy310_pp73-manylinux_2_17_aarch64.manylinux2014_aarch64.whl", hash = "sha256:f44069528d45a933997a6fef143030d8ca8042f0dfaad753e2906398290e2870"},
+    {file = "websockets-12.0-pp310-pypy310_pp73-manylinux_2_5_i686.manylinux1_i686.manylinux_2_17_i686.manylinux2014_i686.whl", hash = "sha256:c4e37d36f0d19f0a4413d3e18c0d03d0c268ada2061868c1e6f5ab1a6d575077"},
+    {file = "websockets-12.0-pp310-pypy310_pp73-manylinux_2_5_x86_64.manylinux1_x86_64.manylinux_2_17_x86_64.manylinux2014_x86_64.whl", hash = "sha256:3d829f975fc2e527a3ef2f9c8f25e553eb7bc779c6665e8e1d52aa22800bb38b"},
+    {file = "websockets-12.0-pp310-pypy310_pp73-win_amd64.whl", hash = "sha256:2c71bd45a777433dd9113847af751aae36e448bc6b8c361a566cb043eda6ec30"},
+    {file = "websockets-12.0-pp38-pypy38_pp73-macosx_10_9_x86_64.whl", hash = "sha256:0bee75f400895aef54157b36ed6d3b308fcab62e5260703add87f44cee9c82a6"},
+    {file = "websockets-12.0-pp38-pypy38_pp73-manylinux_2_17_aarch64.manylinux2014_aarch64.whl", hash = "sha256:423fc1ed29f7512fceb727e2d2aecb952c46aa34895e9ed96071821309951123"},
+    {file = "websockets-12.0-pp38-pypy38_pp73-manylinux_2_5_i686.manylinux1_i686.manylinux_2_17_i686.manylinux2014_i686.whl", hash = "sha256:27a5e9964ef509016759f2ef3f2c1e13f403725a5e6a1775555994966a66e931"},
+    {file = "websockets-12.0-pp38-pypy38_pp73-manylinux_2_5_x86_64.manylinux1_x86_64.manylinux_2_17_x86_64.manylinux2014_x86_64.whl", hash = "sha256:c3181df4583c4d3994d31fb235dc681d2aaad744fbdbf94c4802485ececdecf2"},
+    {file = "websockets-12.0-pp38-pypy38_pp73-win_amd64.whl", hash = "sha256:b067cb952ce8bf40115f6c19f478dc71c5e719b7fbaa511359795dfd9d1a6468"},
+    {file = "websockets-12.0-pp39-pypy39_pp73-macosx_10_9_x86_64.whl", hash = "sha256:00700340c6c7ab788f176d118775202aadea7602c5cc6be6ae127761c16d6b0b"},
+    {file = "websockets-12.0-pp39-pypy39_pp73-manylinux_2_17_aarch64.manylinux2014_aarch64.whl", hash = "sha256:e469d01137942849cff40517c97a30a93ae79917752b34029f0ec72df6b46399"},
+    {file = "websockets-12.0-pp39-pypy39_pp73-manylinux_2_5_i686.manylinux1_i686.manylinux_2_17_i686.manylinux2014_i686.whl", hash = "sha256:ffefa1374cd508d633646d51a8e9277763a9b78ae71324183693959cf94635a7"},
+    {file = "websockets-12.0-pp39-pypy39_pp73-manylinux_2_5_x86_64.manylinux1_x86_64.manylinux_2_17_x86_64.manylinux2014_x86_64.whl", hash = "sha256:ba0cab91b3956dfa9f512147860783a1829a8d905ee218a9837c18f683239611"},
+    {file = "websockets-12.0-pp39-pypy39_pp73-win_amd64.whl", hash = "sha256:2cb388a5bfb56df4d9a406783b7f9dbefb888c09b71629351cc6b036e9259370"},
+    {file = "websockets-12.0-py3-none-any.whl", hash = "sha256:dc284bbc8d7c78a6c69e0c7325ab46ee5e40bb4d50e494d8131a07ef47500e9e"},
+    {file = "websockets-12.0.tar.gz", hash = "sha256:81df9cbcbb6c260de1e007e58c011bfebe2dafc8435107b0537f393dd38c8b1b"},
 ]
 
 [[package]]
@@ -10493,4 +10609,4 @@ cffi = ["cffi (>=1.11)"]
 [metadata]
 lock-version = "2.0"
 python-versions = ">=3.10,<3.13"
-content-hash = "d324192116c4b243e504d57f4605b79c46592a976201d903b16a910b71d84b57"
+content-hash = "b1152e5ef8d1980cf4ac7fd1ffee60123c582ab7bfddf8c2e281baa70e61c2d5"
diff --git a/api/pyproject.toml b/api/pyproject.toml
index 11bcc255d7cd9..da09932e8934e 100644
--- a/api/pyproject.toml
+++ b/api/pyproject.toml
@@ -225,6 +225,7 @@ cos-python-sdk-v5 = "1.9.30"
 esdk-obs-python = "3.24.6.1"
 google-cloud-storage = "2.16.0"
 oss2 = "2.18.5"
+supabase = "~2.8.1"
 tos = "~2.7.1"
 
 ############################################################

From 93af87a9e0c7010e81e22ea8488fd0fa2bd33913 Mon Sep 17 00:00:00 2001
From: crazywoola <100913391+crazywoola@users.noreply.github.com>
Date: Sat, 12 Oct 2024 09:28:45 +0800
Subject: [PATCH 212/235] fix: move exception to debug mode (#9258)

---
 api/core/app/task_pipeline/message_cycle_manage.py | 6 +++++-
 1 file changed, 5 insertions(+), 1 deletion(-)

diff --git a/api/core/app/task_pipeline/message_cycle_manage.py b/api/core/app/task_pipeline/message_cycle_manage.py
index 3a1d1b227f62f..236eebf0b85ff 100644
--- a/api/core/app/task_pipeline/message_cycle_manage.py
+++ b/api/core/app/task_pipeline/message_cycle_manage.py
@@ -1,8 +1,10 @@
+import logging
 from threading import Thread
 from typing import Optional, Union
 
 from flask import Flask, current_app
 
+from configs import dify_config
 from core.app.entities.app_invoke_entities import (
     AdvancedChatAppGenerateEntity,
     AgentChatAppGenerateEntity,
@@ -83,7 +85,9 @@ def _generate_conversation_name_worker(self, flask_app: Flask, conversation_id:
                     name = LLMGenerator.generate_conversation_name(app_model.tenant_id, query)
                     conversation.name = name
                 except Exception as e:
-                    logging.exception(f"generate conversation name failed: {e}")
+                    if dify_config.DEBUG:
+                        logging.exception(f"generate conversation name failed: {e}")
+                    pass
 
                 db.session.merge(conversation)
                 db.session.commit()

From 1206b1eb96d108c516f1171e938d0e5dc3fc1a80 Mon Sep 17 00:00:00 2001
From: NFish <douxc512@gmail.com>
Date: Sat, 12 Oct 2024 11:32:40 +0800
Subject: [PATCH 213/235] fix: add new domain to whitelist (#9265)

---
 web/middleware.ts | 2 +-
 1 file changed, 1 insertion(+), 1 deletion(-)

diff --git a/web/middleware.ts b/web/middleware.ts
index 0c5817445fd1a..e0f8f3782f78d 100644
--- a/web/middleware.ts
+++ b/web/middleware.ts
@@ -1,7 +1,7 @@
 import type { NextRequest } from 'next/server'
 import { NextResponse } from 'next/server'
 
-const NECESSARY_DOMAIN = '*.sentry.io http://localhost:* http://127.0.0.1:* https://analytics.google.com https://googletagmanager.com https://api.github.com'
+const NECESSARY_DOMAIN = '*.sentry.io http://localhost:* http://127.0.0.1:* https://analytics.google.com googletagmanager.com *.googletagmanager.com https://www.google-analytics.com https://api.github.com'
 
 export function middleware(request: NextRequest) {
   const isWhiteListEnabled = !!process.env.NEXT_PUBLIC_CSP_WHITELIST && process.env.NODE_ENV === 'production'

From d9773c963fceed9af750d46a2d04ddd88bc26f37 Mon Sep 17 00:00:00 2001
From: zhuhao <37029601+hwzhuhao@users.noreply.github.com>
Date: Sat, 12 Oct 2024 17:37:01 +0800
Subject: [PATCH 214/235] chore: fix the misclassification of the opensearch-py
 package (#9266)

---
 api/poetry.lock    | 32 +++++++++++++++++++-------------
 api/pyproject.toml |  2 +-
 2 files changed, 20 insertions(+), 14 deletions(-)

diff --git a/api/poetry.lock b/api/poetry.lock
index fd6b98191df02..52e0f3031151d 100644
--- a/api/poetry.lock
+++ b/api/poetry.lock
@@ -732,7 +732,7 @@ name = "bce-python-sdk"
 version = "0.9.23"
 description = "BCE SDK for python"
 optional = false
-python-versions = "!=3.0.*,!=3.1.*,!=3.2.*,<4,>=2.7"
+python-versions = ">=2.7, !=3.0.*, !=3.1.*, !=3.2.*, <4"
 files = [
     {file = "bce_python_sdk-0.9.23-py3-none-any.whl", hash = "sha256:8debe21a040e00060f6044877d594765ed7b18bc765c6bf16b878bca864140a3"},
     {file = "bce_python_sdk-0.9.23.tar.gz", hash = "sha256:19739fed5cd0725356fc5ffa2acbdd8fb23f2a81edb91db21a03174551d0cf41"},
@@ -847,7 +847,7 @@ name = "botocore"
 version = "1.35.38"
 description = "Low-level, data-driven core of boto 3."
 optional = false
-python-versions = ">=3.8"
+python-versions = ">= 3.8"
 files = [
     {file = "botocore-1.35.38-py3-none-any.whl", hash = "sha256:2eb17d32fa2d3bb5d475132a83564d28e3acc2161534f24b75a54418a1d51359"},
     {file = "botocore-1.35.38.tar.gz", hash = "sha256:55d9305c44e5ba29476df456120fa4fb919f03f066afa82f2ae400485e7465f4"},
@@ -1068,7 +1068,7 @@ name = "build"
 version = "1.2.2.post1"
 description = "A simple, correct Python build frontend"
 optional = false
-python-versions = ">=3.8"
+python-versions = ">= 3.8"
 files = [
     {file = "build-1.2.2.post1-py3-none-any.whl", hash = "sha256:1d61c0887fa860c01971625baae8bdd338e517b836a2f70dd1f7aa3a6b2fc5b5"},
     {file = "build-1.2.2.post1.tar.gz", hash = "sha256:b36993e92ca9375a219c99e606a122ff365a760a2d4bba0caa09bd5278b608b7"},
@@ -3385,7 +3385,7 @@ name = "gotrue"
 version = "2.9.2"
 description = "Python Client Library for Supabase Auth"
 optional = false
-python-versions = "<4.0,>=3.8"
+python-versions = ">=3.8,<4.0"
 files = [
     {file = "gotrue-2.9.2-py3-none-any.whl", hash = "sha256:fcd5279e8f1cc630f3ac35af5485fe39f8030b23906776920d2c32a4e308cff4"},
     {file = "gotrue-2.9.2.tar.gz", hash = "sha256:57b3245e916c5efbf19a21b1181011a903c1276bb1df2d847558f2f24f29abb2"},
@@ -4415,7 +4415,7 @@ name = "langfuse"
 version = "2.51.5"
 description = "A client library for accessing langfuse"
 optional = false
-python-versions = "<4.0,>=3.8.1"
+python-versions = ">=3.8.1,<4.0"
 files = [
     {file = "langfuse-2.51.5-py3-none-any.whl", hash = "sha256:b95401ca710ef94b521afa6541933b6f93d7cfd4a97523c8fc75bca4d6d219fb"},
     {file = "langfuse-2.51.5.tar.gz", hash = "sha256:55bc37b5c5d3ae133c1a95db09117cfb3117add110ba02ebbf2ce45ac4395c5b"},
@@ -4440,7 +4440,7 @@ name = "langsmith"
 version = "0.1.134"
 description = "Client library to connect to the LangSmith LLM Tracing and Evaluation Platform."
 optional = false
-python-versions = "<4.0,>=3.8.1"
+python-versions = ">=3.8.1,<4.0"
 files = [
     {file = "langsmith-0.1.134-py3-none-any.whl", hash = "sha256:ada98ad80ef38807725f32441a472da3dd28394010877751f48f458d3289da04"},
     {file = "langsmith-0.1.134.tar.gz", hash = "sha256:23abee3b508875a0e63c602afafffc02442a19cfd88f9daae05b3e9054fd6b61"},
@@ -6429,7 +6429,7 @@ name = "postgrest"
 version = "0.17.1"
 description = "PostgREST client for Python. This library provides an ORM interface to PostgREST."
 optional = false
-python-versions = "<4.0,>=3.8"
+python-versions = ">=3.8,<4.0"
 files = [
     {file = "postgrest-0.17.1-py3-none-any.whl", hash = "sha256:ec1d00dc8532fe5ffb342cfc7c4e610a1e0e2272eb14f78f9b2b61094f9be510"},
     {file = "postgrest-0.17.1.tar.gz", hash = "sha256:e31d9977dbb80dc5f9fdd4d444014686606692dc4ddb9adc85639e56c6d54c92"},
@@ -7746,7 +7746,7 @@ name = "realtime"
 version = "2.0.2"
 description = ""
 optional = false
-python-versions = "<4.0,>=3.9"
+python-versions = ">=3.9,<4.0"
 files = [
     {file = "realtime-2.0.2-py3-none-any.whl", hash = "sha256:2634c915bc38807f2013f21e8bcc4d2f79870dfd81460ddb9393883d0489928a"},
     {file = "realtime-2.0.2.tar.gz", hash = "sha256:519da9325b3b8102139d51785013d592f6b2403d81fa21d838a0b0234723ed7d"},
@@ -8173,7 +8173,7 @@ name = "s3transfer"
 version = "0.10.3"
 description = "An Amazon S3 Transfer Manager"
 optional = false
-python-versions = ">=3.8"
+python-versions = ">= 3.8"
 files = [
     {file = "s3transfer-0.10.3-py3-none-any.whl", hash = "sha256:263ed587a5803c6c708d3ce44dc4dfedaab4c1a32e8329bab818933d79ddcf5d"},
     {file = "s3transfer-0.10.3.tar.gz", hash = "sha256:4f50ed74ab84d474ce614475e0b8d5047ff080810aac5d01ea25231cfc944b0c"},
@@ -8836,7 +8836,7 @@ name = "storage3"
 version = "0.8.1"
 description = "Supabase Storage client for Python."
 optional = false
-python-versions = "<4.0,>=3.8"
+python-versions = ">=3.8,<4.0"
 files = [
     {file = "storage3-0.8.1-py3-none-any.whl", hash = "sha256:0b21205f43eaf0d1dd33bde6c6d0612f88524b7865f017d2ae9827e3f63d9cdc"},
     {file = "storage3-0.8.1.tar.gz", hash = "sha256:ea60b68b2221b3868ccc1a7f1294d57d0d9c51642cdc639d8115fe5d0adc8892"},
@@ -8882,7 +8882,7 @@ name = "supabase"
 version = "2.8.1"
 description = "Supabase client for Python."
 optional = false
-python-versions = "<4.0,>=3.9"
+python-versions = ">=3.9,<4.0"
 files = [
     {file = "supabase-2.8.1-py3-none-any.whl", hash = "sha256:dfa8bef89b54129093521d5bba2136ff765baf67cd76d8ad0aa4984d61a7815c"},
     {file = "supabase-2.8.1.tar.gz", hash = "sha256:711c70e6acd9e2ff48ca0dc0b1bb70c01c25378cc5189ec9f5ed9655b30bc41d"},
@@ -8902,7 +8902,7 @@ name = "supafunc"
 version = "0.6.1"
 description = "Library for Supabase Functions"
 optional = false
-python-versions = "<4.0,>=3.8"
+python-versions = ">=3.8,<4.0"
 files = [
     {file = "supafunc-0.6.1-py3-none-any.whl", hash = "sha256:01aeeeb4bf429977664454a32c86418345140faf6d2e6eb0636d52e4547c5fbb"},
     {file = "supafunc-0.6.1.tar.gz", hash = "sha256:3c8761e3999336ccdb7550498a395fd08afc8469382f55ea56f7f640e5a909aa"},
@@ -10455,31 +10455,37 @@ python-versions = ">=3.8"
 files = [
     {file = "zope.interface-7.1.0-cp310-cp310-macosx_10_9_x86_64.whl", hash = "sha256:2bd9e9f366a5df08ebbdc159f8224904c1c5ce63893984abb76954e6fbe4381a"},
     {file = "zope.interface-7.1.0-cp310-cp310-macosx_11_0_arm64.whl", hash = "sha256:661d5df403cd3c5b8699ac480fa7f58047a3253b029db690efa0c3cf209993ef"},
+    {file = "zope.interface-7.1.0-cp310-cp310-manylinux_2_17_aarch64.manylinux2014_aarch64.whl", hash = "sha256:91b6c30689cfd87c8f264acb2fc16ad6b3c72caba2aec1bf189314cf1a84ca33"},
     {file = "zope.interface-7.1.0-cp310-cp310-manylinux_2_5_i686.manylinux1_i686.manylinux_2_17_i686.manylinux2014_i686.whl", hash = "sha256:2b6a4924f5bad9fe21d99f66a07da60d75696a136162427951ec3cb223a5570d"},
     {file = "zope.interface-7.1.0-cp310-cp310-manylinux_2_5_x86_64.manylinux1_x86_64.manylinux_2_17_x86_64.manylinux2014_x86_64.whl", hash = "sha256:80a3c00b35f6170be5454b45abe2719ea65919a2f09e8a6e7b1362312a872cd3"},
     {file = "zope.interface-7.1.0-cp310-cp310-win_amd64.whl", hash = "sha256:b936d61dbe29572fd2cfe13e30b925e5383bed1aba867692670f5a2a2eb7b4e9"},
     {file = "zope.interface-7.1.0-cp311-cp311-macosx_10_9_x86_64.whl", hash = "sha256:0ac20581fc6cd7c754f6dff0ae06fedb060fa0e9ea6309d8be8b2701d9ea51c4"},
     {file = "zope.interface-7.1.0-cp311-cp311-macosx_11_0_arm64.whl", hash = "sha256:848b6fa92d7c8143646e64124ed46818a0049a24ecc517958c520081fd147685"},
+    {file = "zope.interface-7.1.0-cp311-cp311-manylinux_2_17_aarch64.manylinux2014_aarch64.whl", hash = "sha256:ec1ef1fdb6f014d5886b97e52b16d0f852364f447d2ab0f0c6027765777b6667"},
     {file = "zope.interface-7.1.0-cp311-cp311-manylinux_2_5_i686.manylinux1_i686.manylinux_2_17_i686.manylinux2014_i686.whl", hash = "sha256:3bcff5c09d0215f42ba64b49205a278e44413d9bf9fa688fd9e42bfe472b5f4f"},
     {file = "zope.interface-7.1.0-cp311-cp311-manylinux_2_5_x86_64.manylinux1_x86_64.manylinux_2_17_x86_64.manylinux2014_x86_64.whl", hash = "sha256:07add15de0cc7e69917f7d286b64d54125c950aeb43efed7a5ea7172f000fbc1"},
     {file = "zope.interface-7.1.0-cp311-cp311-win_amd64.whl", hash = "sha256:9940d5bc441f887c5f375ec62bcf7e7e495a2d5b1da97de1184a88fb567f06af"},
     {file = "zope.interface-7.1.0-cp312-cp312-macosx_10_9_x86_64.whl", hash = "sha256:f245d039f72e6f802902375755846f5de1ee1e14c3e8736c078565599bcab621"},
     {file = "zope.interface-7.1.0-cp312-cp312-macosx_11_0_arm64.whl", hash = "sha256:6159e767d224d8f18deff634a1d3722e68d27488c357f62ebeb5f3e2f5288b1f"},
+    {file = "zope.interface-7.1.0-cp312-cp312-manylinux_2_17_aarch64.manylinux2014_aarch64.whl", hash = "sha256:5e956b1fd7f3448dd5e00f273072e73e50dfafcb35e4227e6d5af208075593c9"},
     {file = "zope.interface-7.1.0-cp312-cp312-manylinux_2_5_i686.manylinux1_i686.manylinux_2_17_i686.manylinux2014_i686.whl", hash = "sha256:ff115ef91c0eeac69cd92daeba36a9d8e14daee445b504eeea2b1c0b55821984"},
     {file = "zope.interface-7.1.0-cp312-cp312-manylinux_2_5_x86_64.manylinux1_x86_64.manylinux_2_17_x86_64.manylinux2014_x86_64.whl", hash = "sha256:bec001798ab62c3fc5447162bf48496ae9fba02edc295a9e10a0b0c639a6452e"},
     {file = "zope.interface-7.1.0-cp312-cp312-win_amd64.whl", hash = "sha256:124149e2d42067b9c6597f4dafdc7a0983d0163868f897b7bb5dc850b14f9a87"},
     {file = "zope.interface-7.1.0-cp313-cp313-macosx_10_9_x86_64.whl", hash = "sha256:9733a9a0f94ef53d7aa64661811b20875b5bc6039034c6e42fb9732170130573"},
     {file = "zope.interface-7.1.0-cp313-cp313-macosx_11_0_arm64.whl", hash = "sha256:5fcf379b875c610b5a41bc8a891841533f98de0520287d7f85e25386cd10d3e9"},
+    {file = "zope.interface-7.1.0-cp313-cp313-manylinux_2_17_aarch64.manylinux2014_aarch64.whl", hash = "sha256:d0a45b5af9f72c805ee668d1479480ca85169312211bed6ed18c343e39307d5f"},
     {file = "zope.interface-7.1.0-cp313-cp313-manylinux_2_5_i686.manylinux1_i686.manylinux_2_17_i686.manylinux2014_i686.whl", hash = "sha256:4af4a12b459a273b0b34679a5c3dc5e34c1847c3dd14a628aa0668e19e638ea2"},
     {file = "zope.interface-7.1.0-cp313-cp313-manylinux_2_5_x86_64.manylinux1_x86_64.manylinux_2_17_x86_64.manylinux2014_x86_64.whl", hash = "sha256:a735f82d2e3ed47ca01a20dfc4c779b966b16352650a8036ab3955aad151ed8a"},
     {file = "zope.interface-7.1.0-cp313-cp313-win_amd64.whl", hash = "sha256:5501e772aff595e3c54266bc1bfc5858e8f38974ce413a8f1044aae0f32a83a3"},
     {file = "zope.interface-7.1.0-cp38-cp38-macosx_10_9_x86_64.whl", hash = "sha256:ec59fe53db7d32abb96c6d4efeed84aab4a7c38c62d7a901a9b20c09dd936e7a"},
     {file = "zope.interface-7.1.0-cp38-cp38-macosx_11_0_arm64.whl", hash = "sha256:e53c291debef523b09e1fe3dffe5f35dde164f1c603d77f770b88a1da34b7ed6"},
+    {file = "zope.interface-7.1.0-cp38-cp38-manylinux_2_17_aarch64.manylinux2014_aarch64.whl", hash = "sha256:711eebc77f2092c6a8b304bad0b81a6ce3cf5490b25574e7309fbc07d881e3af"},
     {file = "zope.interface-7.1.0-cp38-cp38-manylinux_2_5_i686.manylinux1_i686.manylinux_2_17_i686.manylinux2014_i686.whl", hash = "sha256:4a00ead2e24c76436e1b457a5132d87f83858330f6c923640b7ef82d668525d1"},
     {file = "zope.interface-7.1.0-cp38-cp38-manylinux_2_5_x86_64.manylinux1_x86_64.manylinux_2_17_x86_64.manylinux2014_x86_64.whl", hash = "sha256:5e28ea0bc4b084fc93a483877653a033062435317082cdc6388dec3438309faf"},
     {file = "zope.interface-7.1.0-cp38-cp38-win_amd64.whl", hash = "sha256:27cfb5205d68b12682b6e55ab8424662d96e8ead19550aad0796b08dd2c9a45e"},
     {file = "zope.interface-7.1.0-cp39-cp39-macosx_10_9_x86_64.whl", hash = "sha256:9e3e48f3dea21c147e1b10c132016cb79af1159facca9736d231694ef5a740a8"},
     {file = "zope.interface-7.1.0-cp39-cp39-macosx_11_0_arm64.whl", hash = "sha256:a99240b1d02dc469f6afbe7da1bf617645e60290c272968f4e53feec18d7dce8"},
+    {file = "zope.interface-7.1.0-cp39-cp39-manylinux_2_17_aarch64.manylinux2014_aarch64.whl", hash = "sha256:cc8a318162123eddbdf22fcc7b751288ce52e4ad096d3766ff1799244352449d"},
     {file = "zope.interface-7.1.0-cp39-cp39-manylinux_2_5_i686.manylinux1_i686.manylinux_2_17_i686.manylinux2014_i686.whl", hash = "sha256:b7b25db127db3e6b597c5f74af60309c4ad65acd826f89609662f0dc33a54728"},
     {file = "zope.interface-7.1.0-cp39-cp39-manylinux_2_5_x86_64.manylinux1_x86_64.manylinux_2_17_x86_64.manylinux2014_x86_64.whl", hash = "sha256:2a29ac607e970b5576547f0e3589ec156e04de17af42839eedcf478450687317"},
     {file = "zope.interface-7.1.0-cp39-cp39-win_amd64.whl", hash = "sha256:a14c9decf0eb61e0892631271d500c1e306c7b6901c998c7035e194d9150fdd1"},
@@ -10609,4 +10615,4 @@ cffi = ["cffi (>=1.11)"]
 [metadata]
 lock-version = "2.0"
 python-versions = ">=3.10,<3.13"
-content-hash = "b1152e5ef8d1980cf4ac7fd1ffee60123c582ab7bfddf8c2e281baa70e61c2d5"
+content-hash = "cc10ee218369eb5576d1e5ac8aeeb72e8927bbcb8bd1ac1594167c45aa9d9a21"
diff --git a/api/pyproject.toml b/api/pyproject.toml
index da09932e8934e..277d1690c7270 100644
--- a/api/pyproject.toml
+++ b/api/pyproject.toml
@@ -207,7 +207,6 @@ matplotlib = "~3.8.2"
 newspaper3k = "0.2.8"
 nltk = "3.8.1"
 numexpr = "~2.9.0"
-opensearch-py = "2.4.0"
 qrcode = "~7.4.2"
 twilio = "~9.0.4"
 vanna = { version = "0.7.3", extras = ["postgres", "mysql", "clickhouse", "duckdb"] }
@@ -238,6 +237,7 @@ alibabacloud_tea_openapi = "~0.3.9"
 chromadb = "0.5.1"
 clickhouse-connect = "~0.7.16"
 elasticsearch = "8.14.0"
+opensearch-py = "2.4.0"
 oracledb = "~2.2.1"
 pgvecto-rs = { version = "~0.2.1", extras = ['sqlalchemy'] }
 pgvector = "0.2.5"

From 29188e0562384d3c8e6cfbd0cf5c4ffd5d57e8aa Mon Sep 17 00:00:00 2001
From: takatost <takatost@users.noreply.github.com>
Date: Sat, 12 Oct 2024 17:48:59 +0800
Subject: [PATCH 215/235] chore: use cache instead of re-querying node record
 during workflow execution (#9280)

---
 .../advanced_chat/generate_task_pipeline.py   |  3 +++
 .../apps/workflow/generate_task_pipeline.py   |  3 +++
 .../task_pipeline/workflow_cycle_manage.py    | 21 +++++++------------
 3 files changed, 14 insertions(+), 13 deletions(-)

diff --git a/api/core/app/apps/advanced_chat/generate_task_pipeline.py b/api/core/app/apps/advanced_chat/generate_task_pipeline.py
index 6bf684f8e40a3..fd63c7787fa63 100644
--- a/api/core/app/apps/advanced_chat/generate_task_pipeline.py
+++ b/api/core/app/apps/advanced_chat/generate_task_pipeline.py
@@ -56,6 +56,7 @@
 from models.model import Conversation, EndUser, Message
 from models.workflow import (
     Workflow,
+    WorkflowNodeExecution,
     WorkflowRunStatus,
 )
 
@@ -72,6 +73,7 @@ class AdvancedChatAppGenerateTaskPipeline(BasedGenerateTaskPipeline, WorkflowCyc
     _workflow: Workflow
     _user: Union[Account, EndUser]
     _workflow_system_variables: dict[SystemVariableKey, Any]
+    _wip_workflow_node_executions: dict[str, WorkflowNodeExecution]
 
     def __init__(
         self,
@@ -115,6 +117,7 @@ def __init__(
         }
 
         self._task_state = WorkflowTaskState()
+        self._wip_workflow_node_executions = {}
 
         self._conversation_name_generate_thread = None
 
diff --git a/api/core/app/apps/workflow/generate_task_pipeline.py b/api/core/app/apps/workflow/generate_task_pipeline.py
index 3afc5053673e3..7c53556e43bc4 100644
--- a/api/core/app/apps/workflow/generate_task_pipeline.py
+++ b/api/core/app/apps/workflow/generate_task_pipeline.py
@@ -52,6 +52,7 @@
     Workflow,
     WorkflowAppLog,
     WorkflowAppLogCreatedFrom,
+    WorkflowNodeExecution,
     WorkflowRun,
     WorkflowRunStatus,
 )
@@ -69,6 +70,7 @@ class WorkflowAppGenerateTaskPipeline(BasedGenerateTaskPipeline, WorkflowCycleMa
     _task_state: WorkflowTaskState
     _application_generate_entity: WorkflowAppGenerateEntity
     _workflow_system_variables: dict[SystemVariableKey, Any]
+    _wip_workflow_node_executions: dict[str, WorkflowNodeExecution]
 
     def __init__(
         self,
@@ -103,6 +105,7 @@ def __init__(
         }
 
         self._task_state = WorkflowTaskState()
+        self._wip_workflow_node_executions = {}
 
     def process(self) -> Union[WorkflowAppBlockingResponse, Generator[WorkflowAppStreamResponse, None, None]]:
         """
diff --git a/api/core/app/task_pipeline/workflow_cycle_manage.py b/api/core/app/task_pipeline/workflow_cycle_manage.py
index 4fc587db77c14..f48ae9c01e6de 100644
--- a/api/core/app/task_pipeline/workflow_cycle_manage.py
+++ b/api/core/app/task_pipeline/workflow_cycle_manage.py
@@ -57,6 +57,7 @@ class WorkflowCycleManage:
     _user: Union[Account, EndUser]
     _task_state: WorkflowTaskState
     _workflow_system_variables: dict[SystemVariableKey, Any]
+    _wip_workflow_node_executions: dict[str, WorkflowNodeExecution]
 
     def _handle_workflow_run_start(self) -> WorkflowRun:
         max_sequence = (
@@ -251,6 +252,8 @@ def _handle_node_execution_start(
         db.session.refresh(workflow_node_execution)
         db.session.close()
 
+        self._wip_workflow_node_executions[workflow_node_execution.node_execution_id] = workflow_node_execution
+
         return workflow_node_execution
 
     def _handle_workflow_node_execution_success(self, event: QueueNodeSucceededEvent) -> WorkflowNodeExecution:
@@ -275,9 +278,10 @@ def _handle_workflow_node_execution_success(self, event: QueueNodeSucceededEvent
         workflow_node_execution.elapsed_time = (workflow_node_execution.finished_at - event.start_at).total_seconds()
 
         db.session.commit()
-        db.session.refresh(workflow_node_execution)
         db.session.close()
 
+        self._wip_workflow_node_executions.pop(workflow_node_execution.node_execution_id)
+
         return workflow_node_execution
 
     def _handle_workflow_node_execution_failed(self, event: QueueNodeFailedEvent) -> WorkflowNodeExecution:
@@ -300,9 +304,10 @@ def _handle_workflow_node_execution_failed(self, event: QueueNodeFailedEvent) ->
         workflow_node_execution.elapsed_time = (workflow_node_execution.finished_at - event.start_at).total_seconds()
 
         db.session.commit()
-        db.session.refresh(workflow_node_execution)
         db.session.close()
 
+        self._wip_workflow_node_executions.pop(workflow_node_execution.node_execution_id)
+
         return workflow_node_execution
 
     #################################################
@@ -678,17 +683,7 @@ def _refetch_workflow_node_execution(self, node_execution_id: str) -> WorkflowNo
         :param node_execution_id: workflow node execution id
         :return:
         """
-        workflow_node_execution = (
-            db.session.query(WorkflowNodeExecution)
-            .filter(
-                WorkflowNodeExecution.tenant_id == self._application_generate_entity.app_config.tenant_id,
-                WorkflowNodeExecution.app_id == self._application_generate_entity.app_config.app_id,
-                WorkflowNodeExecution.workflow_id == self._workflow.id,
-                WorkflowNodeExecution.triggered_from == WorkflowNodeExecutionTriggeredFrom.WORKFLOW_RUN.value,
-                WorkflowNodeExecution.node_execution_id == node_execution_id,
-            )
-            .first()
-        )
+        workflow_node_execution = self._wip_workflow_node_executions.get(node_execution_id)
 
         if not workflow_node_execution:
             raise Exception(f"Workflow node execution not found: {node_execution_id}")

From 23ce1fb1ba0baa523fe3928bf8fc2fe0dacf3f5f Mon Sep 17 00:00:00 2001
From: takatost <takatost@users.noreply.github.com>
Date: Sat, 12 Oct 2024 18:30:46 +0800
Subject: [PATCH 216/235] chore: optimize the trace ops slow queries on node
 executions. (#9282)

---
 .../task_pipeline/workflow_cycle_manage.py    | 55 ++++++++++++++-----
 api/core/ops/langfuse_trace/langfuse_trace.py | 41 ++++++++------
 .../ops/langsmith_trace/langsmith_trace.py    | 41 ++++++++------
 3 files changed, 92 insertions(+), 45 deletions(-)

diff --git a/api/core/app/task_pipeline/workflow_cycle_manage.py b/api/core/app/task_pipeline/workflow_cycle_manage.py
index f48ae9c01e6de..b8f5ac260340e 100644
--- a/api/core/app/task_pipeline/workflow_cycle_manage.py
+++ b/api/core/app/task_pipeline/workflow_cycle_manage.py
@@ -266,20 +266,35 @@ def _handle_workflow_node_execution_success(self, event: QueueNodeSucceededEvent
 
         inputs = WorkflowEntry.handle_special_values(event.inputs)
         outputs = WorkflowEntry.handle_special_values(event.outputs)
-
-        workflow_node_execution.status = WorkflowNodeExecutionStatus.SUCCEEDED.value
-        workflow_node_execution.inputs = json.dumps(inputs) if inputs else None
-        workflow_node_execution.process_data = json.dumps(event.process_data) if event.process_data else None
-        workflow_node_execution.outputs = json.dumps(outputs) if outputs else None
-        workflow_node_execution.execution_metadata = (
+        execution_metadata = (
             json.dumps(jsonable_encoder(event.execution_metadata)) if event.execution_metadata else None
         )
-        workflow_node_execution.finished_at = datetime.now(timezone.utc).replace(tzinfo=None)
-        workflow_node_execution.elapsed_time = (workflow_node_execution.finished_at - event.start_at).total_seconds()
+        finished_at = datetime.now(timezone.utc).replace(tzinfo=None)
+        elapsed_time = (finished_at - event.start_at).total_seconds()
+
+        db.session.query(WorkflowNodeExecution).filter(WorkflowNodeExecution.id == workflow_node_execution.id).update(
+            {
+                WorkflowNodeExecution.status: WorkflowNodeExecutionStatus.SUCCEEDED.value,
+                WorkflowNodeExecution.inputs: json.dumps(inputs) if inputs else None,
+                WorkflowNodeExecution.process_data: json.dumps(event.process_data) if event.process_data else None,
+                WorkflowNodeExecution.outputs: json.dumps(outputs) if outputs else None,
+                WorkflowNodeExecution.execution_metadata: execution_metadata,
+                WorkflowNodeExecution.finished_at: finished_at,
+                WorkflowNodeExecution.elapsed_time: elapsed_time,
+            }
+        )
 
         db.session.commit()
         db.session.close()
 
+        workflow_node_execution.status = WorkflowNodeExecutionStatus.SUCCEEDED.value
+        workflow_node_execution.inputs = json.dumps(inputs) if inputs else None
+        workflow_node_execution.process_data = json.dumps(event.process_data) if event.process_data else None
+        workflow_node_execution.outputs = json.dumps(outputs) if outputs else None
+        workflow_node_execution.execution_metadata = execution_metadata
+        workflow_node_execution.finished_at = finished_at
+        workflow_node_execution.elapsed_time = elapsed_time
+
         self._wip_workflow_node_executions.pop(workflow_node_execution.node_execution_id)
 
         return workflow_node_execution
@@ -294,17 +309,31 @@ def _handle_workflow_node_execution_failed(self, event: QueueNodeFailedEvent) ->
 
         inputs = WorkflowEntry.handle_special_values(event.inputs)
         outputs = WorkflowEntry.handle_special_values(event.outputs)
+        finished_at = datetime.now(timezone.utc).replace(tzinfo=None)
+        elapsed_time = (finished_at - event.start_at).total_seconds()
+
+        db.session.query(WorkflowNodeExecution).filter(WorkflowNodeExecution.id == workflow_node_execution.id).update(
+            {
+                WorkflowNodeExecution.status: WorkflowNodeExecutionStatus.FAILED.value,
+                WorkflowNodeExecution.error: event.error,
+                WorkflowNodeExecution.inputs: json.dumps(inputs) if inputs else None,
+                WorkflowNodeExecution.process_data: json.dumps(event.process_data) if event.process_data else None,
+                WorkflowNodeExecution.outputs: json.dumps(outputs) if outputs else None,
+                WorkflowNodeExecution.finished_at: finished_at,
+                WorkflowNodeExecution.elapsed_time: elapsed_time,
+            }
+        )
+
+        db.session.commit()
+        db.session.close()
 
         workflow_node_execution.status = WorkflowNodeExecutionStatus.FAILED.value
         workflow_node_execution.error = event.error
-        workflow_node_execution.finished_at = datetime.now(timezone.utc).replace(tzinfo=None)
         workflow_node_execution.inputs = json.dumps(inputs) if inputs else None
         workflow_node_execution.process_data = json.dumps(event.process_data) if event.process_data else None
         workflow_node_execution.outputs = json.dumps(outputs) if outputs else None
-        workflow_node_execution.elapsed_time = (workflow_node_execution.finished_at - event.start_at).total_seconds()
-
-        db.session.commit()
-        db.session.close()
+        workflow_node_execution.finished_at = finished_at
+        workflow_node_execution.elapsed_time = elapsed_time
 
         self._wip_workflow_node_executions.pop(workflow_node_execution.node_execution_id)
 
diff --git a/api/core/ops/langfuse_trace/langfuse_trace.py b/api/core/ops/langfuse_trace/langfuse_trace.py
index 171e34f8cb48f..0cba40c51a0d1 100644
--- a/api/core/ops/langfuse_trace/langfuse_trace.py
+++ b/api/core/ops/langfuse_trace/langfuse_trace.py
@@ -110,26 +110,35 @@ def workflow_trace(self, trace_info: WorkflowTraceInfo):
             self.add_trace(langfuse_trace_data=trace_data)
 
         # through workflow_run_id get all_nodes_execution
-        workflow_nodes_executions = (
-            db.session.query(
-                WorkflowNodeExecution.id,
-                WorkflowNodeExecution.tenant_id,
-                WorkflowNodeExecution.app_id,
-                WorkflowNodeExecution.title,
-                WorkflowNodeExecution.node_type,
-                WorkflowNodeExecution.status,
-                WorkflowNodeExecution.inputs,
-                WorkflowNodeExecution.outputs,
-                WorkflowNodeExecution.created_at,
-                WorkflowNodeExecution.elapsed_time,
-                WorkflowNodeExecution.process_data,
-                WorkflowNodeExecution.execution_metadata,
-            )
+        workflow_nodes_execution_id_records = (
+            db.session.query(WorkflowNodeExecution.id)
             .filter(WorkflowNodeExecution.workflow_run_id == trace_info.workflow_run_id)
             .all()
         )
 
-        for node_execution in workflow_nodes_executions:
+        for node_execution_id_record in workflow_nodes_execution_id_records:
+            node_execution = (
+                db.session.query(
+                    WorkflowNodeExecution.id,
+                    WorkflowNodeExecution.tenant_id,
+                    WorkflowNodeExecution.app_id,
+                    WorkflowNodeExecution.title,
+                    WorkflowNodeExecution.node_type,
+                    WorkflowNodeExecution.status,
+                    WorkflowNodeExecution.inputs,
+                    WorkflowNodeExecution.outputs,
+                    WorkflowNodeExecution.created_at,
+                    WorkflowNodeExecution.elapsed_time,
+                    WorkflowNodeExecution.process_data,
+                    WorkflowNodeExecution.execution_metadata,
+                )
+                .filter(WorkflowNodeExecution.id == node_execution_id_record.id)
+                .first()
+            )
+
+            if not node_execution:
+                continue
+
             node_execution_id = node_execution.id
             tenant_id = node_execution.tenant_id
             app_id = node_execution.app_id
diff --git a/api/core/ops/langsmith_trace/langsmith_trace.py b/api/core/ops/langsmith_trace/langsmith_trace.py
index 37cbea13fd873..ad450504057be 100644
--- a/api/core/ops/langsmith_trace/langsmith_trace.py
+++ b/api/core/ops/langsmith_trace/langsmith_trace.py
@@ -100,26 +100,35 @@ def workflow_trace(self, trace_info: WorkflowTraceInfo):
         self.add_run(langsmith_run)
 
         # through workflow_run_id get all_nodes_execution
-        workflow_nodes_executions = (
-            db.session.query(
-                WorkflowNodeExecution.id,
-                WorkflowNodeExecution.tenant_id,
-                WorkflowNodeExecution.app_id,
-                WorkflowNodeExecution.title,
-                WorkflowNodeExecution.node_type,
-                WorkflowNodeExecution.status,
-                WorkflowNodeExecution.inputs,
-                WorkflowNodeExecution.outputs,
-                WorkflowNodeExecution.created_at,
-                WorkflowNodeExecution.elapsed_time,
-                WorkflowNodeExecution.process_data,
-                WorkflowNodeExecution.execution_metadata,
-            )
+        workflow_nodes_execution_id_records = (
+            db.session.query(WorkflowNodeExecution.id)
             .filter(WorkflowNodeExecution.workflow_run_id == trace_info.workflow_run_id)
             .all()
         )
 
-        for node_execution in workflow_nodes_executions:
+        for node_execution_id_record in workflow_nodes_execution_id_records:
+            node_execution = (
+                db.session.query(
+                    WorkflowNodeExecution.id,
+                    WorkflowNodeExecution.tenant_id,
+                    WorkflowNodeExecution.app_id,
+                    WorkflowNodeExecution.title,
+                    WorkflowNodeExecution.node_type,
+                    WorkflowNodeExecution.status,
+                    WorkflowNodeExecution.inputs,
+                    WorkflowNodeExecution.outputs,
+                    WorkflowNodeExecution.created_at,
+                    WorkflowNodeExecution.elapsed_time,
+                    WorkflowNodeExecution.process_data,
+                    WorkflowNodeExecution.execution_metadata,
+                )
+                .filter(WorkflowNodeExecution.id == node_execution_id_record.id)
+                .first()
+            )
+
+            if not node_execution:
+                continue
+
             node_execution_id = node_execution.id
             tenant_id = node_execution.tenant_id
             app_id = node_execution.app_id

From c6b74daa0a87bd02bdd6e82c2cc00e45772a793f Mon Sep 17 00:00:00 2001
From: Garfield Dai <dai.hai@foxmail.com>
Date: Sat, 12 Oct 2024 18:47:59 +0800
Subject: [PATCH 217/235] Fix/s3 iam add region name (#7819)

---
 api/extensions/storage/aws_s3_storage.py | 10 +++++++++-
 1 file changed, 9 insertions(+), 1 deletion(-)

diff --git a/api/extensions/storage/aws_s3_storage.py b/api/extensions/storage/aws_s3_storage.py
index fede683aa7bb1..38f823763fa4f 100644
--- a/api/extensions/storage/aws_s3_storage.py
+++ b/api/extensions/storage/aws_s3_storage.py
@@ -1,3 +1,4 @@
+import logging
 from collections.abc import Generator
 from contextlib import closing
 
@@ -8,6 +9,8 @@
 
 from extensions.storage.base_storage import BaseStorage
 
+logger = logging.getLogger(__name__)
+
 
 class AwsS3Storage(BaseStorage):
     """Implementation for Amazon Web Services S3 storage."""
@@ -17,9 +20,14 @@ def __init__(self, app: Flask):
         app_config = self.app.config
         self.bucket_name = app_config.get("S3_BUCKET_NAME")
         if app_config.get("S3_USE_AWS_MANAGED_IAM"):
+            logger.info("Using AWS managed IAM role for S3")
+
             session = boto3.Session()
-            self.client = session.client("s3")
+            region_name = app_config.get("S3_REGION")
+            self.client = session.client(service_name="s3", region_name=region_name)
         else:
+            logger.info("Using ak and sk for S3")
+
             self.client = boto3.client(
                 "s3",
                 aws_secret_access_key=app_config.get("S3_SECRET_KEY"),

From 793205afc547264151ef6ac428f485422adf6daf Mon Sep 17 00:00:00 2001
From: Yi Xiao <54782454+YIXIAO0@users.noreply.github.com>
Date: Sat, 12 Oct 2024 21:24:43 +0800
Subject: [PATCH 218/235] Feat: rerank model verification in front end (#9271)

---
 .../params-config/config-content.tsx          | 49 +++++++++++++----
 .../common/retrieval-param-config/index.tsx   | 49 +++++++++++++----
 .../workflow/hooks/use-workflow-start-run.tsx | 55 +++++++++++++++++++
 .../components/workflow/hooks/use-workflow.ts | 28 ++++++++++
 web/i18n/en-US/workflow.ts                    |  1 +
 web/i18n/zh-Hans/workflow.ts                  |  1 +
 6 files changed, 159 insertions(+), 24 deletions(-)

diff --git a/web/app/components/app/configuration/dataset-config/params-config/config-content.tsx b/web/app/components/app/configuration/dataset-config/params-config/config-content.tsx
index 7f83a14d58acc..f556121518073 100644
--- a/web/app/components/app/configuration/dataset-config/params-config/config-content.tsx
+++ b/web/app/components/app/configuration/dataset-config/params-config/config-content.tsx
@@ -1,6 +1,6 @@
 'use client'
 
-import { memo, useEffect, useMemo } from 'react'
+import { memo, useCallback, useEffect, useMemo } from 'react'
 import type { FC } from 'react'
 import { useTranslation } from 'react-i18next'
 import WeightedScore from './weighted-score'
@@ -11,7 +11,7 @@ import type {
   DatasetConfigs,
 } from '@/models/debug'
 import ModelSelector from '@/app/components/header/account-setting/model-provider-page/model-selector'
-import { useModelListAndDefaultModelAndCurrentProviderAndModel } from '@/app/components/header/account-setting/model-provider-page/hooks'
+import { useCurrentProviderAndModel, useModelListAndDefaultModelAndCurrentProviderAndModel } from '@/app/components/header/account-setting/model-provider-page/hooks'
 import type { ModelConfig } from '@/app/components/workflow/types'
 import ModelParameterModal from '@/app/components/header/account-setting/model-provider-page/model-parameter-modal'
 import Tooltip from '@/app/components/base/tooltip'
@@ -23,6 +23,7 @@ import { RerankingModeEnum } from '@/models/datasets'
 import cn from '@/utils/classnames'
 import { useSelectedDatasetsMode } from '@/app/components/workflow/nodes/knowledge-retrieval/hooks'
 import Switch from '@/app/components/base/switch'
+import Toast from '@/app/components/base/toast'
 
 type Props = {
   datasetConfigs: DatasetConfigs
@@ -60,6 +61,24 @@ const ConfigContent: FC<Props> = ({
     modelList: rerankModelList,
     defaultModel: rerankDefaultModel,
   } = useModelListAndDefaultModelAndCurrentProviderAndModel(ModelTypeEnum.rerank)
+
+  const {
+    currentModel,
+  } = useCurrentProviderAndModel(
+    rerankModelList,
+    rerankDefaultModel
+      ? {
+        ...rerankDefaultModel,
+        provider: rerankDefaultModel.provider.provider,
+      }
+      : undefined,
+  )
+
+  const handleDisabledSwitchClick = useCallback(() => {
+    if (!currentModel)
+      Toast.notify({ type: 'error', message: t('workflow.errorMsg.rerankModelRequired') })
+  }, [currentModel, rerankDefaultModel, t])
+
   const rerankModel = (() => {
     if (datasetConfigs.reranking_model?.reranking_provider_name) {
       return {
@@ -231,16 +250,22 @@ const ConfigContent: FC<Props> = ({
                 <div className='flex items-center'>
                   {
                     selectedDatasetsMode.allEconomic && (
-                      <Switch
-                        size='md'
-                        defaultValue={showRerankModel}
-                        onChange={(v) => {
-                          onChange({
-                            ...datasetConfigs,
-                            reranking_enable: v,
-                          })
-                        }}
-                      />
+                      <div
+                        className='flex items-center'
+                        onClick={handleDisabledSwitchClick}
+                      >
+                        <Switch
+                          size='md'
+                          defaultValue={currentModel ? showRerankModel : false}
+                          disabled={!currentModel}
+                          onChange={(v) => {
+                            onChange({
+                              ...datasetConfigs,
+                              reranking_enable: v,
+                            })
+                          }}
+                        />
+                      </div>
                     )
                   }
                   <div className='leading-[32px] ml-1 text-text-secondary system-sm-semibold'>{t('common.modelProvider.rerankModel.key')}</div>
diff --git a/web/app/components/datasets/common/retrieval-param-config/index.tsx b/web/app/components/datasets/common/retrieval-param-config/index.tsx
index 323e47f3b4ae2..9d48d56a8dc51 100644
--- a/web/app/components/datasets/common/retrieval-param-config/index.tsx
+++ b/web/app/components/datasets/common/retrieval-param-config/index.tsx
@@ -1,6 +1,6 @@
 'use client'
 import type { FC } from 'react'
-import React from 'react'
+import React, { useCallback } from 'react'
 import { useTranslation } from 'react-i18next'
 
 import cn from '@/utils/classnames'
@@ -11,7 +11,7 @@ import Switch from '@/app/components/base/switch'
 import Tooltip from '@/app/components/base/tooltip'
 import type { RetrievalConfig } from '@/types/app'
 import ModelSelector from '@/app/components/header/account-setting/model-provider-page/model-selector'
-import { useModelListAndDefaultModel } from '@/app/components/header/account-setting/model-provider-page/hooks'
+import { useCurrentProviderAndModel, useModelListAndDefaultModel } from '@/app/components/header/account-setting/model-provider-page/hooks'
 import { ModelTypeEnum } from '@/app/components/header/account-setting/model-provider-page/declarations'
 import {
   DEFAULT_WEIGHTED_SCORE,
@@ -19,6 +19,7 @@ import {
   WeightedScoreEnum,
 } from '@/models/datasets'
 import WeightedScore from '@/app/components/app/configuration/dataset-config/params-config/weighted-score'
+import Toast from '@/app/components/base/toast'
 
 type Props = {
   type: RETRIEVE_METHOD
@@ -38,6 +39,24 @@ const RetrievalParamConfig: FC<Props> = ({
     defaultModel: rerankDefaultModel,
     modelList: rerankModelList,
   } = useModelListAndDefaultModel(ModelTypeEnum.rerank)
+
+  const {
+    currentModel,
+  } = useCurrentProviderAndModel(
+    rerankModelList,
+    rerankDefaultModel
+      ? {
+        ...rerankDefaultModel,
+        provider: rerankDefaultModel.provider.provider,
+      }
+      : undefined,
+  )
+
+  const handleDisabledSwitchClick = useCallback(() => {
+    if (!currentModel)
+      Toast.notify({ type: 'error', message: t('workflow.errorMsg.rerankModelRequired') })
+  }, [currentModel, rerankDefaultModel, t])
+
   const isHybridSearch = type === RETRIEVE_METHOD.hybrid
 
   const rerankModel = (() => {
@@ -99,16 +118,22 @@ const RetrievalParamConfig: FC<Props> = ({
         <div>
           <div className='flex h-8 items-center text-[13px] font-medium text-gray-900 space-x-2'>
             {canToggleRerankModalEnable && (
-              <Switch
-                size='md'
-                defaultValue={value.reranking_enable}
-                onChange={(v) => {
-                  onChange({
-                    ...value,
-                    reranking_enable: v,
-                  })
-                }}
-              />
+              <div
+                className='flex items-center'
+                onClick={handleDisabledSwitchClick}
+              >
+                <Switch
+                  size='md'
+                  defaultValue={currentModel ? value.reranking_enable : false}
+                  onChange={(v) => {
+                    onChange({
+                      ...value,
+                      reranking_enable: v,
+                    })
+                  }}
+                  disabled={!currentModel}
+                />
+              </div>
             )}
             <div className='flex items-center'>
               <span className='mr-0.5'>{t('common.modelProvider.rerankModel.key')}</span>
diff --git a/web/app/components/workflow/hooks/use-workflow-start-run.tsx b/web/app/components/workflow/hooks/use-workflow-start-run.tsx
index b2b1c69975658..77e959b573ba5 100644
--- a/web/app/components/workflow/hooks/use-workflow-start-run.tsx
+++ b/web/app/components/workflow/hooks/use-workflow-start-run.tsx
@@ -1,17 +1,25 @@
 import { useCallback } from 'react'
 import { useStoreApi } from 'reactflow'
+import { useTranslation } from 'react-i18next'
 import { useWorkflowStore } from '../store'
 import {
   BlockEnum,
   WorkflowRunningStatus,
 } from '../types'
+import type { KnowledgeRetrievalNodeType } from '../nodes/knowledge-retrieval/types'
+import type { Node } from '../types'
+import { useWorkflow } from './use-workflow'
 import {
   useIsChatMode,
   useNodesSyncDraft,
   useWorkflowInteractions,
   useWorkflowRun,
 } from './index'
+import { ModelTypeEnum } from '@/app/components/header/account-setting/model-provider-page/declarations'
+import { useCurrentProviderAndModel, useModelListAndDefaultModelAndCurrentProviderAndModel } from '@/app/components/header/account-setting/model-provider-page/hooks'
 import { useFeaturesStore } from '@/app/components/base/features/hooks'
+import KnowledgeRetrievalDefault from '@/app/components/workflow/nodes/knowledge-retrieval/default'
+import Toast from '@/app/components/base/toast'
 
 export const useWorkflowStartRun = () => {
   const store = useStoreApi()
@@ -20,7 +28,26 @@ export const useWorkflowStartRun = () => {
   const isChatMode = useIsChatMode()
   const { handleCancelDebugAndPreviewPanel } = useWorkflowInteractions()
   const { handleRun } = useWorkflowRun()
+  const { isFromStartNode } = useWorkflow()
   const { doSyncWorkflowDraft } = useNodesSyncDraft()
+  const { checkValid: checkKnowledgeRetrievalValid } = KnowledgeRetrievalDefault
+  const { t } = useTranslation()
+  const {
+    modelList: rerankModelList,
+    defaultModel: rerankDefaultModel,
+  } = useModelListAndDefaultModelAndCurrentProviderAndModel(ModelTypeEnum.rerank)
+
+  const {
+    currentModel,
+  } = useCurrentProviderAndModel(
+    rerankModelList,
+    rerankDefaultModel
+      ? {
+        ...rerankDefaultModel,
+        provider: rerankDefaultModel.provider.provider,
+      }
+      : undefined,
+  )
 
   const handleWorkflowStartRunInWorkflow = useCallback(async () => {
     const {
@@ -33,6 +60,9 @@ export const useWorkflowStartRun = () => {
     const { getNodes } = store.getState()
     const nodes = getNodes()
     const startNode = nodes.find(node => node.data.type === BlockEnum.Start)
+    const knowledgeRetrievalNodes = nodes.filter((node: Node<KnowledgeRetrievalNodeType>) =>
+      node.data.type === BlockEnum.KnowledgeRetrieval,
+    )
     const startVariables = startNode?.data.variables || []
     const fileSettings = featuresStore!.getState().features.file
     const {
@@ -42,6 +72,31 @@ export const useWorkflowStartRun = () => {
       setShowEnvPanel,
     } = workflowStore.getState()
 
+    if (knowledgeRetrievalNodes.length > 0) {
+      for (const node of knowledgeRetrievalNodes) {
+        if (isFromStartNode(node.id)) {
+          const res = checkKnowledgeRetrievalValid(node.data, t)
+          if (!res.isValid || !currentModel || !rerankDefaultModel) {
+            const errorMessage = res.errorMessage
+            if (errorMessage) {
+              Toast.notify({
+                type: 'error',
+                message: errorMessage,
+              })
+              return false
+            }
+            else {
+              Toast.notify({
+                type: 'error',
+                message: t('appDebug.datasetConfig.rerankModelRequired'),
+              })
+              return false
+            }
+          }
+        }
+      }
+    }
+
     setShowEnvPanel(false)
 
     if (showDebugAndPreviewPanel) {
diff --git a/web/app/components/workflow/hooks/use-workflow.ts b/web/app/components/workflow/hooks/use-workflow.ts
index b201b28b88d14..ec7ce66e5fdce 100644
--- a/web/app/components/workflow/hooks/use-workflow.ts
+++ b/web/app/components/workflow/hooks/use-workflow.ts
@@ -235,6 +235,33 @@ export const useWorkflow = () => {
     return nodes.filter(node => node.parentId === nodeId)
   }, [store])
 
+  const isFromStartNode = useCallback((nodeId: string) => {
+    const { getNodes } = store.getState()
+    const nodes = getNodes()
+    const currentNode = nodes.find(node => node.id === nodeId)
+
+    if (!currentNode)
+      return false
+
+    if (currentNode.data.type === BlockEnum.Start)
+      return true
+
+    const checkPreviousNodes = (node: Node) => {
+      const previousNodes = getBeforeNodeById(node.id)
+
+      for (const prevNode of previousNodes) {
+        if (prevNode.data.type === BlockEnum.Start)
+          return true
+        if (checkPreviousNodes(prevNode))
+          return true
+      }
+
+      return false
+    }
+
+    return checkPreviousNodes(currentNode)
+  }, [store, getBeforeNodeById])
+
   const handleOutVarRenameChange = useCallback((nodeId: string, oldValeSelector: ValueSelector, newVarSelector: ValueSelector) => {
     const { getNodes, setNodes } = store.getState()
     const afterNodes = getAfterNodesInSameBranch(nodeId)
@@ -389,6 +416,7 @@ export const useWorkflow = () => {
     checkParallelLimit,
     checkNestedParallelLimit,
     isValidConnection,
+    isFromStartNode,
     formatTimeFromNow,
     getNode,
     getBeforeNodeById,
diff --git a/web/i18n/en-US/workflow.ts b/web/i18n/en-US/workflow.ts
index a7e768911ffec..d5ab6eb72894f 100644
--- a/web/i18n/en-US/workflow.ts
+++ b/web/i18n/en-US/workflow.ts
@@ -172,6 +172,7 @@ const translation = {
   },
   errorMsg: {
     fieldRequired: '{{field}} is required',
+    rerankModelRequired: 'Before turning on the Rerank Model, please confirm that the model has been successfully configured in the settings.',
     authRequired: 'Authorization is required',
     invalidJson: '{{field}} is invalid JSON',
     fields: {
diff --git a/web/i18n/zh-Hans/workflow.ts b/web/i18n/zh-Hans/workflow.ts
index 3579ec5df3440..4959a87be7db7 100644
--- a/web/i18n/zh-Hans/workflow.ts
+++ b/web/i18n/zh-Hans/workflow.ts
@@ -172,6 +172,7 @@ const translation = {
   },
   errorMsg: {
     fieldRequired: '{{field}} 不能为空',
+    rerankModelRequired: '开启 Rerank 模型前，请务必确认模型已在设置中成功配置。',
     authRequired: '请先授权',
     invalidJson: '{{field}} 是非法的 JSON',
     fields: {

From 2ec6ffe478984539209725b754631565515da8bd Mon Sep 17 00:00:00 2001
From: Shili Cao <shilicaohust@163.com>
Date: Sat, 12 Oct 2024 23:24:17 +0800
Subject: [PATCH 219/235] feat:support baidu vector db (#9185)

---
 api/.env.example                              |   9 +
 api/commands.py                               |   8 +
 .../middleware/vdb/baidu_vector_config.py     |  45 +++
 api/controllers/console/datasets/datasets.py  |   2 +
 api/core/rag/datasource/vdb/baidu/__init__.py |   0
 .../rag/datasource/vdb/baidu/baidu_vector.py  | 272 ++++++++++++++++++
 api/core/rag/datasource/vdb/vector_factory.py |   4 +
 api/core/rag/datasource/vdb/vector_type.py    |   1 +
 api/poetry.lock                               |  47 ++-
 api/pyproject.toml                            |   1 +
 .../vdb/__mock/baiduvectordb.py               | 154 ++++++++++
 .../integration_tests/vdb/baidu/__init__.py   |   0
 .../integration_tests/vdb/baidu/test_baidu.py |  36 +++
 docker/.env.example                           |   9 +
 docker/docker-compose.yaml                    |   7 +
 15 files changed, 582 insertions(+), 13 deletions(-)
 create mode 100644 api/configs/middleware/vdb/baidu_vector_config.py
 create mode 100644 api/core/rag/datasource/vdb/baidu/__init__.py
 create mode 100644 api/core/rag/datasource/vdb/baidu/baidu_vector.py
 create mode 100644 api/tests/integration_tests/vdb/__mock/baiduvectordb.py
 create mode 100644 api/tests/integration_tests/vdb/baidu/__init__.py
 create mode 100644 api/tests/integration_tests/vdb/baidu/test_baidu.py

diff --git a/api/.env.example b/api/.env.example
index 7b5e4950c82dd..3f88fb3cdf5b0 100644
--- a/api/.env.example
+++ b/api/.env.example
@@ -208,6 +208,15 @@ OPENSEARCH_USER=admin
 OPENSEARCH_PASSWORD=admin
 OPENSEARCH_SECURE=true
 
+# Baidu configuration
+BAIDU_VECTOR_DB_ENDPOINT=http://127.0.0.1:5287
+BAIDU_VECTOR_DB_CONNECTION_TIMEOUT_MS=30000
+BAIDU_VECTOR_DB_ACCOUNT=root
+BAIDU_VECTOR_DB_API_KEY=dify
+BAIDU_VECTOR_DB_DATABASE=dify
+BAIDU_VECTOR_DB_SHARD=1
+BAIDU_VECTOR_DB_REPLICAS=3
+
 # Upload configuration
 UPLOAD_FILE_SIZE_LIMIT=15
 UPLOAD_FILE_BATCH_LIMIT=5
diff --git a/api/commands.py b/api/commands.py
index 7ef4aed7f7766..dbcd8a744d3a4 100644
--- a/api/commands.py
+++ b/api/commands.py
@@ -347,6 +347,14 @@ def migrate_knowledge_vector_database():
                     index_name = Dataset.gen_collection_name_by_id(dataset_id)
                     index_struct_dict = {"type": "elasticsearch", "vector_store": {"class_prefix": index_name}}
                     dataset.index_struct = json.dumps(index_struct_dict)
+                elif vector_type == VectorType.BAIDU:
+                    dataset_id = dataset.id
+                    collection_name = Dataset.gen_collection_name_by_id(dataset_id)
+                    index_struct_dict = {
+                        "type": VectorType.BAIDU,
+                        "vector_store": {"class_prefix": collection_name},
+                    }
+                    dataset.index_struct = json.dumps(index_struct_dict)
                 else:
                     raise ValueError(f"Vector store {vector_type} is not supported.")
 
diff --git a/api/configs/middleware/vdb/baidu_vector_config.py b/api/configs/middleware/vdb/baidu_vector_config.py
new file mode 100644
index 0000000000000..44742c2e2f434
--- /dev/null
+++ b/api/configs/middleware/vdb/baidu_vector_config.py
@@ -0,0 +1,45 @@
+from typing import Optional
+
+from pydantic import Field, NonNegativeInt, PositiveInt
+from pydantic_settings import BaseSettings
+
+
+class BaiduVectorDBConfig(BaseSettings):
+    """
+    Configuration settings for Baidu Vector Database
+    """
+
+    BAIDU_VECTOR_DB_ENDPOINT: Optional[str] = Field(
+        description="URL of the Baidu Vector Database service (e.g., 'http://vdb.bj.baidubce.com')",
+        default=None,
+    )
+
+    BAIDU_VECTOR_DB_CONNECTION_TIMEOUT_MS: PositiveInt = Field(
+        description="Timeout in milliseconds for Baidu Vector Database operations (default is 30000 milliseconds)",
+        default=30000,
+    )
+
+    BAIDU_VECTOR_DB_ACCOUNT: Optional[str] = Field(
+        description="Account for authenticating with the Baidu Vector Database",
+        default=None,
+    )
+
+    BAIDU_VECTOR_DB_API_KEY: Optional[str] = Field(
+        description="API key for authenticating with the Baidu Vector Database service",
+        default=None,
+    )
+
+    BAIDU_VECTOR_DB_DATABASE: Optional[str] = Field(
+        description="Name of the specific Baidu Vector Database to connect to",
+        default=None,
+    )
+
+    BAIDU_VECTOR_DB_SHARD: PositiveInt = Field(
+        description="Number of shards for the Baidu Vector Database (default is 1)",
+        default=1,
+    )
+
+    BAIDU_VECTOR_DB_REPLICAS: NonNegativeInt = Field(
+        description="Number of replicas for the Baidu Vector Database (default is 3)",
+        default=3,
+    )
diff --git a/api/controllers/console/datasets/datasets.py b/api/controllers/console/datasets/datasets.py
index 9561fd8b70e4b..102089bf071ac 100644
--- a/api/controllers/console/datasets/datasets.py
+++ b/api/controllers/console/datasets/datasets.py
@@ -617,6 +617,7 @@ def get(self):
                 | VectorType.CHROMA
                 | VectorType.TENCENT
                 | VectorType.PGVECTO_RS
+                | VectorType.BAIDU
             ):
                 return {"retrieval_method": [RetrievalMethod.SEMANTIC_SEARCH.value]}
             case (
@@ -653,6 +654,7 @@ def get(self, vector_type):
                 | VectorType.CHROMA
                 | VectorType.TENCENT
                 | VectorType.PGVECTO_RS
+                | VectorType.BAIDU
             ):
                 return {"retrieval_method": [RetrievalMethod.SEMANTIC_SEARCH.value]}
             case (
diff --git a/api/core/rag/datasource/vdb/baidu/__init__.py b/api/core/rag/datasource/vdb/baidu/__init__.py
new file mode 100644
index 0000000000000..e69de29bb2d1d
diff --git a/api/core/rag/datasource/vdb/baidu/baidu_vector.py b/api/core/rag/datasource/vdb/baidu/baidu_vector.py
new file mode 100644
index 0000000000000..543cfa67b3540
--- /dev/null
+++ b/api/core/rag/datasource/vdb/baidu/baidu_vector.py
@@ -0,0 +1,272 @@
+import json
+import time
+import uuid
+from typing import Any
+
+from pydantic import BaseModel, model_validator
+from pymochow import MochowClient
+from pymochow.auth.bce_credentials import BceCredentials
+from pymochow.configuration import Configuration
+from pymochow.model.enum import FieldType, IndexState, IndexType, MetricType, TableState
+from pymochow.model.schema import Field, HNSWParams, Schema, VectorIndex
+from pymochow.model.table import AnnSearch, HNSWSearchParams, Partition, Row
+
+from configs import dify_config
+from core.rag.datasource.entity.embedding import Embeddings
+from core.rag.datasource.vdb.vector_base import BaseVector
+from core.rag.datasource.vdb.vector_factory import AbstractVectorFactory
+from core.rag.datasource.vdb.vector_type import VectorType
+from core.rag.models.document import Document
+from extensions.ext_redis import redis_client
+from models.dataset import Dataset
+
+
+class BaiduConfig(BaseModel):
+    endpoint: str
+    connection_timeout_in_mills: int = 30 * 1000
+    account: str
+    api_key: str
+    database: str
+    index_type: str = "HNSW"
+    metric_type: str = "L2"
+    shard: int = 1
+    replicas: int = 3
+
+    @model_validator(mode="before")
+    @classmethod
+    def validate_config(cls, values: dict) -> dict:
+        if not values["endpoint"]:
+            raise ValueError("config BAIDU_VECTOR_DB_ENDPOINT is required")
+        if not values["account"]:
+            raise ValueError("config BAIDU_VECTOR_DB_ACCOUNT is required")
+        if not values["api_key"]:
+            raise ValueError("config BAIDU_VECTOR_DB_API_KEY is required")
+        if not values["database"]:
+            raise ValueError("config BAIDU_VECTOR_DB_DATABASE is required")
+        return values
+
+
+class BaiduVector(BaseVector):
+    field_id: str = "id"
+    field_vector: str = "vector"
+    field_text: str = "text"
+    field_metadata: str = "metadata"
+    field_app_id: str = "app_id"
+    field_annotation_id: str = "annotation_id"
+    index_vector: str = "vector_idx"
+
+    def __init__(self, collection_name: str, config: BaiduConfig):
+        super().__init__(collection_name)
+        self._client_config = config
+        self._client = self._init_client(config)
+        self._db = self._init_database()
+
+    def get_type(self) -> str:
+        return VectorType.BAIDU
+
+    def to_index_struct(self) -> dict:
+        return {"type": self.get_type(), "vector_store": {"class_prefix": self._collection_name}}
+
+    def create(self, texts: list[Document], embeddings: list[list[float]], **kwargs):
+        self._create_table(len(embeddings[0]))
+        self.add_texts(texts, embeddings)
+
+    def add_texts(self, documents: list[Document], embeddings: list[list[float]], **kwargs):
+        texts = [doc.page_content for doc in documents]
+        metadatas = [doc.metadata for doc in documents]
+        total_count = len(documents)
+        batch_size = 1000
+
+        # upsert texts and embeddings batch by batch
+        table = self._db.table(self._collection_name)
+        for start in range(0, total_count, batch_size):
+            end = min(start + batch_size, total_count)
+            rows = []
+            for i in range(start, end, 1):
+                row = Row(
+                    id=metadatas[i].get("doc_id", str(uuid.uuid4())),
+                    vector=embeddings[i],
+                    text=texts[i],
+                    metadata=json.dumps(metadatas[i]),
+                    app_id=metadatas[i].get("app_id", ""),
+                    annotation_id=metadatas[i].get("annotation_id", ""),
+                )
+                rows.append(row)
+            table.upsert(rows=rows)
+
+        # rebuild vector index after upsert finished
+        table.rebuild_index(self.index_vector)
+        while True:
+            time.sleep(1)
+            index = table.describe_index(self.index_vector)
+            if index.state == IndexState.NORMAL:
+                break
+
+    def text_exists(self, id: str) -> bool:
+        res = self._db.table(self._collection_name).query(primary_key={self.field_id: id})
+        if res and res.code == 0:
+            return True
+        return False
+
+    def delete_by_ids(self, ids: list[str]) -> None:
+        quoted_ids = [f"'{id}'" for id in ids]
+        self._db.table(self._collection_name).delete(filter=f"id IN({', '.join(quoted_ids)})")
+
+    def delete_by_metadata_field(self, key: str, value: str) -> None:
+        self._db.table(self._collection_name).delete(filter=f"{key} = '{value}'")
+
+    def search_by_vector(self, query_vector: list[float], **kwargs: Any) -> list[Document]:
+        anns = AnnSearch(
+            vector_field=self.field_vector,
+            vector_floats=query_vector,
+            params=HNSWSearchParams(ef=kwargs.get("ef", 10), limit=kwargs.get("top_k", 4)),
+        )
+        res = self._db.table(self._collection_name).search(
+            anns=anns,
+            projections=[self.field_id, self.field_text, self.field_metadata],
+            retrieve_vector=True,
+        )
+        score_threshold = float(kwargs.get("score_threshold") or 0.0)
+        return self._get_search_res(res, score_threshold)
+
+    def search_by_full_text(self, query: str, **kwargs: Any) -> list[Document]:
+        # baidu vector database doesn't support bm25 search on current version
+        return []
+
+    def _get_search_res(self, res, score_threshold):
+        docs = []
+        for row in res.rows:
+            row_data = row.get("row", {})
+            meta = row_data.get(self.field_metadata)
+            if meta is not None:
+                meta = json.loads(meta)
+            score = row.get("score", 0.0)
+            if score > score_threshold:
+                meta["score"] = score
+                doc = Document(page_content=row_data.get(self.field_text), metadata=meta)
+                docs.append(doc)
+
+        return docs
+
+    def delete(self) -> None:
+        self._db.drop_table(table_name=self._collection_name)
+
+    def _init_client(self, config) -> MochowClient:
+        config = Configuration(credentials=BceCredentials(config.account, config.api_key), endpoint=config.endpoint)
+        client = MochowClient(config)
+        return client
+
+    def _init_database(self):
+        exists = False
+        for db in self._client.list_databases():
+            if db.database_name == self._client_config.database:
+                exists = True
+                break
+        # Create database if not existed
+        if exists:
+            return self._client.database(self._client_config.database)
+        else:
+            return self._client.create_database(database_name=self._client_config.database)
+
+    def _table_existed(self) -> bool:
+        tables = self._db.list_table()
+        return any(table.table_name == self._collection_name for table in tables)
+
+    def _create_table(self, dimension: int) -> None:
+        # Try to grab distributed lock and create table
+        lock_name = "vector_indexing_lock_{}".format(self._collection_name)
+        with redis_client.lock(lock_name, timeout=20):
+            table_exist_cache_key = "vector_indexing_{}".format(self._collection_name)
+            if redis_client.get(table_exist_cache_key):
+                return
+
+            if self._table_existed():
+                return
+
+            self.delete()
+
+            # check IndexType and MetricType
+            index_type = None
+            for k, v in IndexType.__members__.items():
+                if k == self._client_config.index_type:
+                    index_type = v
+            if index_type is None:
+                raise ValueError("unsupported index_type")
+            metric_type = None
+            for k, v in MetricType.__members__.items():
+                if k == self._client_config.metric_type:
+                    metric_type = v
+            if metric_type is None:
+                raise ValueError("unsupported metric_type")
+
+            # Construct field schema
+            fields = []
+            fields.append(
+                Field(
+                    self.field_id,
+                    FieldType.STRING,
+                    primary_key=True,
+                    partition_key=True,
+                    auto_increment=False,
+                    not_null=True,
+                )
+            )
+            fields.append(Field(self.field_metadata, FieldType.STRING, not_null=True))
+            fields.append(Field(self.field_app_id, FieldType.STRING))
+            fields.append(Field(self.field_annotation_id, FieldType.STRING))
+            fields.append(Field(self.field_text, FieldType.TEXT, not_null=True))
+            fields.append(Field(self.field_vector, FieldType.FLOAT_VECTOR, not_null=True, dimension=dimension))
+
+            # Construct vector index params
+            indexes = []
+            indexes.append(
+                VectorIndex(
+                    index_name="vector_idx",
+                    index_type=index_type,
+                    field="vector",
+                    metric_type=metric_type,
+                    params=HNSWParams(m=16, efconstruction=200),
+                )
+            )
+
+            # Create table
+            self._db.create_table(
+                table_name=self._collection_name,
+                replication=self._client_config.replicas,
+                partition=Partition(partition_num=self._client_config.shard),
+                schema=Schema(fields=fields, indexes=indexes),
+                description="Table for Dify",
+            )
+
+            redis_client.set(table_exist_cache_key, 1, ex=3600)
+
+        # Wait for table created
+        while True:
+            time.sleep(1)
+            table = self._db.describe_table(self._collection_name)
+            if table.state == TableState.NORMAL:
+                break
+
+
+class BaiduVectorFactory(AbstractVectorFactory):
+    def init_vector(self, dataset: Dataset, attributes: list, embeddings: Embeddings) -> BaiduVector:
+        if dataset.index_struct_dict:
+            class_prefix: str = dataset.index_struct_dict["vector_store"]["class_prefix"]
+            collection_name = class_prefix.lower()
+        else:
+            dataset_id = dataset.id
+            collection_name = Dataset.gen_collection_name_by_id(dataset_id).lower()
+            dataset.index_struct = json.dumps(self.gen_index_struct_dict(VectorType.BAIDU, collection_name))
+
+        return BaiduVector(
+            collection_name=collection_name,
+            config=BaiduConfig(
+                endpoint=dify_config.BAIDU_VECTOR_DB_ENDPOINT,
+                connection_timeout_in_mills=dify_config.BAIDU_VECTOR_DB_CONNECTION_TIMEOUT_MS,
+                account=dify_config.BAIDU_VECTOR_DB_ACCOUNT,
+                api_key=dify_config.BAIDU_VECTOR_DB_API_KEY,
+                database=dify_config.BAIDU_VECTOR_DB_DATABASE,
+                shard=dify_config.BAIDU_VECTOR_DB_SHARD,
+                replicas=dify_config.BAIDU_VECTOR_DB_REPLICAS,
+            ),
+        )
diff --git a/api/core/rag/datasource/vdb/vector_factory.py b/api/core/rag/datasource/vdb/vector_factory.py
index 943b23870cc5c..1f4a4d44a23ee 100644
--- a/api/core/rag/datasource/vdb/vector_factory.py
+++ b/api/core/rag/datasource/vdb/vector_factory.py
@@ -103,6 +103,10 @@ def get_vector_factory(vector_type: str) -> type[AbstractVectorFactory]:
                 from core.rag.datasource.vdb.analyticdb.analyticdb_vector import AnalyticdbVectorFactory
 
                 return AnalyticdbVectorFactory
+            case VectorType.BAIDU:
+                from core.rag.datasource.vdb.baidu.baidu_vector import BaiduVectorFactory
+
+                return BaiduVectorFactory
             case _:
                 raise ValueError(f"Vector store {vector_type} is not supported.")
 
diff --git a/api/core/rag/datasource/vdb/vector_type.py b/api/core/rag/datasource/vdb/vector_type.py
index ba04ea879d9b4..996ff48615c90 100644
--- a/api/core/rag/datasource/vdb/vector_type.py
+++ b/api/core/rag/datasource/vdb/vector_type.py
@@ -16,3 +16,4 @@ class VectorType(str, Enum):
     TENCENT = "tencent"
     ORACLE = "oracle"
     ELASTICSEARCH = "elasticsearch"
+    BAIDU = "baidu"
diff --git a/api/poetry.lock b/api/poetry.lock
index 52e0f3031151d..6565db27ad572 100644
--- a/api/poetry.lock
+++ b/api/poetry.lock
@@ -732,7 +732,7 @@ name = "bce-python-sdk"
 version = "0.9.23"
 description = "BCE SDK for python"
 optional = false
-python-versions = ">=2.7, !=3.0.*, !=3.1.*, !=3.2.*, <4"
+python-versions = "!=3.0.*,!=3.1.*,!=3.2.*,<4,>=2.7"
 files = [
     {file = "bce_python_sdk-0.9.23-py3-none-any.whl", hash = "sha256:8debe21a040e00060f6044877d594765ed7b18bc765c6bf16b878bca864140a3"},
     {file = "bce_python_sdk-0.9.23.tar.gz", hash = "sha256:19739fed5cd0725356fc5ffa2acbdd8fb23f2a81edb91db21a03174551d0cf41"},
@@ -847,7 +847,7 @@ name = "botocore"
 version = "1.35.38"
 description = "Low-level, data-driven core of boto 3."
 optional = false
-python-versions = ">= 3.8"
+python-versions = ">=3.8"
 files = [
     {file = "botocore-1.35.38-py3-none-any.whl", hash = "sha256:2eb17d32fa2d3bb5d475132a83564d28e3acc2161534f24b75a54418a1d51359"},
     {file = "botocore-1.35.38.tar.gz", hash = "sha256:55d9305c44e5ba29476df456120fa4fb919f03f066afa82f2ae400485e7465f4"},
@@ -1068,7 +1068,7 @@ name = "build"
 version = "1.2.2.post1"
 description = "A simple, correct Python build frontend"
 optional = false
-python-versions = ">= 3.8"
+python-versions = ">=3.8"
 files = [
     {file = "build-1.2.2.post1-py3-none-any.whl", hash = "sha256:1d61c0887fa860c01971625baae8bdd338e517b836a2f70dd1f7aa3a6b2fc5b5"},
     {file = "build-1.2.2.post1.tar.gz", hash = "sha256:b36993e92ca9375a219c99e606a122ff365a760a2d4bba0caa09bd5278b608b7"},
@@ -3385,7 +3385,7 @@ name = "gotrue"
 version = "2.9.2"
 description = "Python Client Library for Supabase Auth"
 optional = false
-python-versions = ">=3.8,<4.0"
+python-versions = "<4.0,>=3.8"
 files = [
     {file = "gotrue-2.9.2-py3-none-any.whl", hash = "sha256:fcd5279e8f1cc630f3ac35af5485fe39f8030b23906776920d2c32a4e308cff4"},
     {file = "gotrue-2.9.2.tar.gz", hash = "sha256:57b3245e916c5efbf19a21b1181011a903c1276bb1df2d847558f2f24f29abb2"},
@@ -4415,7 +4415,7 @@ name = "langfuse"
 version = "2.51.5"
 description = "A client library for accessing langfuse"
 optional = false
-python-versions = ">=3.8.1,<4.0"
+python-versions = "<4.0,>=3.8.1"
 files = [
     {file = "langfuse-2.51.5-py3-none-any.whl", hash = "sha256:b95401ca710ef94b521afa6541933b6f93d7cfd4a97523c8fc75bca4d6d219fb"},
     {file = "langfuse-2.51.5.tar.gz", hash = "sha256:55bc37b5c5d3ae133c1a95db09117cfb3117add110ba02ebbf2ce45ac4395c5b"},
@@ -4440,7 +4440,7 @@ name = "langsmith"
 version = "0.1.134"
 description = "Client library to connect to the LangSmith LLM Tracing and Evaluation Platform."
 optional = false
-python-versions = ">=3.8.1,<4.0"
+python-versions = "<4.0,>=3.8.1"
 files = [
     {file = "langsmith-0.1.134-py3-none-any.whl", hash = "sha256:ada98ad80ef38807725f32441a472da3dd28394010877751f48f458d3289da04"},
     {file = "langsmith-0.1.134.tar.gz", hash = "sha256:23abee3b508875a0e63c602afafffc02442a19cfd88f9daae05b3e9054fd6b61"},
@@ -6429,7 +6429,7 @@ name = "postgrest"
 version = "0.17.1"
 description = "PostgREST client for Python. This library provides an ORM interface to PostgREST."
 optional = false
-python-versions = ">=3.8,<4.0"
+python-versions = "<4.0,>=3.8"
 files = [
     {file = "postgrest-0.17.1-py3-none-any.whl", hash = "sha256:ec1d00dc8532fe5ffb342cfc7c4e610a1e0e2272eb14f78f9b2b61094f9be510"},
     {file = "postgrest-0.17.1.tar.gz", hash = "sha256:e31d9977dbb80dc5f9fdd4d444014686606692dc4ddb9adc85639e56c6d54c92"},
@@ -7047,6 +7047,22 @@ bulk-writer = ["azure-storage-blob", "minio (>=7.0.0)", "pyarrow (>=12.0.0)", "r
 dev = ["black", "grpcio (==1.62.2)", "grpcio-testing (==1.62.2)", "grpcio-tools (==1.62.2)", "pytest (>=5.3.4)", "pytest-cov (>=2.8.1)", "pytest-timeout (>=1.3.4)", "ruff (>0.4.0)"]
 model = ["milvus-model (>=0.1.0)"]
 
+[[package]]
+name = "pymochow"
+version = "1.3.1"
+description = "Python SDK for mochow"
+optional = false
+python-versions = ">=3.7"
+files = [
+    {file = "pymochow-1.3.1-py3-none-any.whl", hash = "sha256:a7f3b34fd6ea5d1d8413650bb6678365aa148fc396ae945e4ccb4f2365a52327"},
+    {file = "pymochow-1.3.1.tar.gz", hash = "sha256:1693d10cd0bb7bce45327890a90adafb503155922ccc029acb257699a73a20ba"},
+]
+
+[package.dependencies]
+future = "*"
+orjson = "*"
+requests = "*"
+
 [[package]]
 name = "pymysql"
 version = "1.1.1"
@@ -7746,7 +7762,7 @@ name = "realtime"
 version = "2.0.2"
 description = ""
 optional = false
-python-versions = ">=3.9,<4.0"
+python-versions = "<4.0,>=3.9"
 files = [
     {file = "realtime-2.0.2-py3-none-any.whl", hash = "sha256:2634c915bc38807f2013f21e8bcc4d2f79870dfd81460ddb9393883d0489928a"},
     {file = "realtime-2.0.2.tar.gz", hash = "sha256:519da9325b3b8102139d51785013d592f6b2403d81fa21d838a0b0234723ed7d"},
@@ -8173,7 +8189,7 @@ name = "s3transfer"
 version = "0.10.3"
 description = "An Amazon S3 Transfer Manager"
 optional = false
-python-versions = ">= 3.8"
+python-versions = ">=3.8"
 files = [
     {file = "s3transfer-0.10.3-py3-none-any.whl", hash = "sha256:263ed587a5803c6c708d3ce44dc4dfedaab4c1a32e8329bab818933d79ddcf5d"},
     {file = "s3transfer-0.10.3.tar.gz", hash = "sha256:4f50ed74ab84d474ce614475e0b8d5047ff080810aac5d01ea25231cfc944b0c"},
@@ -8417,6 +8433,11 @@ files = [
     {file = "scikit_learn-1.5.2-cp312-cp312-manylinux_2_17_aarch64.manylinux2014_aarch64.whl", hash = "sha256:f60021ec1574e56632be2a36b946f8143bf4e5e6af4a06d85281adc22938e0dd"},
     {file = "scikit_learn-1.5.2-cp312-cp312-manylinux_2_17_x86_64.manylinux2014_x86_64.whl", hash = "sha256:394397841449853c2290a32050382edaec3da89e35b3e03d6cc966aebc6a8ae6"},
     {file = "scikit_learn-1.5.2-cp312-cp312-win_amd64.whl", hash = "sha256:57cc1786cfd6bd118220a92ede80270132aa353647684efa385a74244a41e3b1"},
+    {file = "scikit_learn-1.5.2-cp313-cp313-macosx_10_13_x86_64.whl", hash = "sha256:e9a702e2de732bbb20d3bad29ebd77fc05a6b427dc49964300340e4c9328b3f5"},
+    {file = "scikit_learn-1.5.2-cp313-cp313-macosx_12_0_arm64.whl", hash = "sha256:b0768ad641981f5d3a198430a1d31c3e044ed2e8a6f22166b4d546a5116d7908"},
+    {file = "scikit_learn-1.5.2-cp313-cp313-manylinux_2_17_aarch64.manylinux2014_aarch64.whl", hash = "sha256:178ddd0a5cb0044464fc1bfc4cca5b1833bfc7bb022d70b05db8530da4bb3dd3"},
+    {file = "scikit_learn-1.5.2-cp313-cp313-manylinux_2_17_x86_64.manylinux2014_x86_64.whl", hash = "sha256:f7284ade780084d94505632241bf78c44ab3b6f1e8ccab3d2af58e0e950f9c12"},
+    {file = "scikit_learn-1.5.2-cp313-cp313-win_amd64.whl", hash = "sha256:b7b0f9a0b1040830d38c39b91b3a44e1b643f4b36e36567b80b7c6bd2202a27f"},
     {file = "scikit_learn-1.5.2-cp39-cp39-macosx_10_9_x86_64.whl", hash = "sha256:757c7d514ddb00ae249832fe87100d9c73c6ea91423802872d9e74970a0e40b9"},
     {file = "scikit_learn-1.5.2-cp39-cp39-macosx_12_0_arm64.whl", hash = "sha256:52788f48b5d8bca5c0736c175fa6bdaab2ef00a8f536cda698db61bd89c551c1"},
     {file = "scikit_learn-1.5.2-cp39-cp39-manylinux_2_17_aarch64.manylinux2014_aarch64.whl", hash = "sha256:643964678f4b5fbdc95cbf8aec638acc7aa70f5f79ee2cdad1eec3df4ba6ead8"},
@@ -8836,7 +8857,7 @@ name = "storage3"
 version = "0.8.1"
 description = "Supabase Storage client for Python."
 optional = false
-python-versions = ">=3.8,<4.0"
+python-versions = "<4.0,>=3.8"
 files = [
     {file = "storage3-0.8.1-py3-none-any.whl", hash = "sha256:0b21205f43eaf0d1dd33bde6c6d0612f88524b7865f017d2ae9827e3f63d9cdc"},
     {file = "storage3-0.8.1.tar.gz", hash = "sha256:ea60b68b2221b3868ccc1a7f1294d57d0d9c51642cdc639d8115fe5d0adc8892"},
@@ -8882,7 +8903,7 @@ name = "supabase"
 version = "2.8.1"
 description = "Supabase client for Python."
 optional = false
-python-versions = ">=3.9,<4.0"
+python-versions = "<4.0,>=3.9"
 files = [
     {file = "supabase-2.8.1-py3-none-any.whl", hash = "sha256:dfa8bef89b54129093521d5bba2136ff765baf67cd76d8ad0aa4984d61a7815c"},
     {file = "supabase-2.8.1.tar.gz", hash = "sha256:711c70e6acd9e2ff48ca0dc0b1bb70c01c25378cc5189ec9f5ed9655b30bc41d"},
@@ -8902,7 +8923,7 @@ name = "supafunc"
 version = "0.6.1"
 description = "Library for Supabase Functions"
 optional = false
-python-versions = ">=3.8,<4.0"
+python-versions = "<4.0,>=3.8"
 files = [
     {file = "supafunc-0.6.1-py3-none-any.whl", hash = "sha256:01aeeeb4bf429977664454a32c86418345140faf6d2e6eb0636d52e4547c5fbb"},
     {file = "supafunc-0.6.1.tar.gz", hash = "sha256:3c8761e3999336ccdb7550498a395fd08afc8469382f55ea56f7f640e5a909aa"},
@@ -10615,4 +10636,4 @@ cffi = ["cffi (>=1.11)"]
 [metadata]
 lock-version = "2.0"
 python-versions = ">=3.10,<3.13"
-content-hash = "cc10ee218369eb5576d1e5ac8aeeb72e8927bbcb8bd1ac1594167c45aa9d9a21"
+content-hash = "375ac3a91760513924647e67376cb6018505ec61d967651b254c68af9808d774"
diff --git a/api/pyproject.toml b/api/pyproject.toml
index 277d1690c7270..594517771b34f 100644
--- a/api/pyproject.toml
+++ b/api/pyproject.toml
@@ -242,6 +242,7 @@ oracledb = "~2.2.1"
 pgvecto-rs = { version = "~0.2.1", extras = ['sqlalchemy'] }
 pgvector = "0.2.5"
 pymilvus = "~2.4.4"
+pymochow = "1.3.1"
 qdrant-client = "1.7.3"
 tcvectordb = "1.3.2"
 tidb-vector = "0.0.9"
diff --git a/api/tests/integration_tests/vdb/__mock/baiduvectordb.py b/api/tests/integration_tests/vdb/__mock/baiduvectordb.py
new file mode 100644
index 0000000000000..a8eaf42b7de1d
--- /dev/null
+++ b/api/tests/integration_tests/vdb/__mock/baiduvectordb.py
@@ -0,0 +1,154 @@
+import os
+
+import pytest
+from _pytest.monkeypatch import MonkeyPatch
+from pymochow import MochowClient
+from pymochow.model.database import Database
+from pymochow.model.enum import IndexState, IndexType, MetricType, ReadConsistency, TableState
+from pymochow.model.schema import HNSWParams, VectorIndex
+from pymochow.model.table import Table
+from requests.adapters import HTTPAdapter
+
+
+class MockBaiduVectorDBClass:
+    def mock_vector_db_client(
+        self,
+        config=None,
+        adapter: HTTPAdapter = None,
+    ):
+        self._conn = None
+        self._config = None
+
+    def list_databases(self, config=None) -> list[Database]:
+        return [
+            Database(
+                conn=self._conn,
+                database_name="dify",
+                config=self._config,
+            )
+        ]
+
+    def create_database(self, database_name: str, config=None) -> Database:
+        return Database(conn=self._conn, database_name=database_name, config=config)
+
+    def list_table(self, config=None) -> list[Table]:
+        return []
+
+    def drop_table(self, table_name: str, config=None):
+        return {"code": 0, "msg": "Success"}
+
+    def create_table(
+        self,
+        table_name: str,
+        replication: int,
+        partition: int,
+        schema,
+        enable_dynamic_field=False,
+        description: str = "",
+        config=None,
+    ) -> Table:
+        return Table(self, table_name, replication, partition, schema, enable_dynamic_field, description, config)
+
+    def describe_table(self, table_name: str, config=None) -> Table:
+        return Table(
+            self,
+            table_name,
+            3,
+            1,
+            None,
+            enable_dynamic_field=False,
+            description="table for dify",
+            config=config,
+            state=TableState.NORMAL,
+        )
+
+    def upsert(self, rows, config=None):
+        return {"code": 0, "msg": "operation success", "affectedCount": 1}
+
+    def rebuild_index(self, index_name: str, config=None):
+        return {"code": 0, "msg": "Success"}
+
+    def describe_index(self, index_name: str, config=None):
+        return VectorIndex(
+            index_name=index_name,
+            index_type=IndexType.HNSW,
+            field="vector",
+            metric_type=MetricType.L2,
+            params=HNSWParams(m=16, efconstruction=200),
+            auto_build=False,
+            state=IndexState.NORMAL,
+        )
+
+    def query(
+        self,
+        primary_key,
+        partition_key=None,
+        projections=None,
+        retrieve_vector=False,
+        read_consistency=ReadConsistency.EVENTUAL,
+        config=None,
+    ):
+        return {
+            "row": {
+                "id": "doc_id_001",
+                "vector": [0.23432432, 0.8923744, 0.89238432],
+                "text": "text",
+                "metadata": {"doc_id": "doc_id_001"},
+            },
+            "code": 0,
+            "msg": "Success",
+        }
+
+    def delete(self, primary_key=None, partition_key=None, filter=None, config=None):
+        return {"code": 0, "msg": "Success"}
+
+    def search(
+        self,
+        anns,
+        partition_key=None,
+        projections=None,
+        retrieve_vector=False,
+        read_consistency=ReadConsistency.EVENTUAL,
+        config=None,
+    ):
+        return {
+            "rows": [
+                {
+                    "row": {
+                        "id": "doc_id_001",
+                        "vector": [0.23432432, 0.8923744, 0.89238432],
+                        "text": "text",
+                        "metadata": {"doc_id": "doc_id_001"},
+                    },
+                    "distance": 0.1,
+                    "score": 0.5,
+                }
+            ],
+            "code": 0,
+            "msg": "Success",
+        }
+
+
+MOCK = os.getenv("MOCK_SWITCH", "false").lower() == "true"
+
+
+@pytest.fixture
+def setup_baiduvectordb_mock(request, monkeypatch: MonkeyPatch):
+    if MOCK:
+        monkeypatch.setattr(MochowClient, "__init__", MockBaiduVectorDBClass.mock_vector_db_client)
+        monkeypatch.setattr(MochowClient, "list_databases", MockBaiduVectorDBClass.list_databases)
+        monkeypatch.setattr(MochowClient, "create_database", MockBaiduVectorDBClass.create_database)
+        monkeypatch.setattr(Database, "table", MockBaiduVectorDBClass.describe_table)
+        monkeypatch.setattr(Database, "list_table", MockBaiduVectorDBClass.list_table)
+        monkeypatch.setattr(Database, "create_table", MockBaiduVectorDBClass.create_table)
+        monkeypatch.setattr(Database, "drop_table", MockBaiduVectorDBClass.drop_table)
+        monkeypatch.setattr(Database, "describe_table", MockBaiduVectorDBClass.describe_table)
+        monkeypatch.setattr(Table, "rebuild_index", MockBaiduVectorDBClass.rebuild_index)
+        monkeypatch.setattr(Table, "describe_index", MockBaiduVectorDBClass.describe_index)
+        monkeypatch.setattr(Table, "delete", MockBaiduVectorDBClass.delete)
+        monkeypatch.setattr(Table, "search", MockBaiduVectorDBClass.search)
+
+    yield
+
+    if MOCK:
+        monkeypatch.undo()
diff --git a/api/tests/integration_tests/vdb/baidu/__init__.py b/api/tests/integration_tests/vdb/baidu/__init__.py
new file mode 100644
index 0000000000000..e69de29bb2d1d
diff --git a/api/tests/integration_tests/vdb/baidu/test_baidu.py b/api/tests/integration_tests/vdb/baidu/test_baidu.py
new file mode 100644
index 0000000000000..01a7f8853ac36
--- /dev/null
+++ b/api/tests/integration_tests/vdb/baidu/test_baidu.py
@@ -0,0 +1,36 @@
+from unittest.mock import MagicMock
+
+from core.rag.datasource.vdb.baidu.baidu_vector import BaiduConfig, BaiduVector
+from tests.integration_tests.vdb.__mock.baiduvectordb import setup_baiduvectordb_mock
+from tests.integration_tests.vdb.test_vector_store import AbstractVectorTest, get_example_text, setup_mock_redis
+
+mock_client = MagicMock()
+mock_client.list_databases.return_value = [{"name": "test"}]
+
+
+class BaiduVectorTest(AbstractVectorTest):
+    def __init__(self):
+        super().__init__()
+        self.vector = BaiduVector(
+            "dify",
+            BaiduConfig(
+                endpoint="http://127.0.0.1:5287",
+                account="root",
+                api_key="dify",
+                database="dify",
+                shard=1,
+                replicas=3,
+            ),
+        )
+
+    def search_by_vector(self):
+        hits_by_vector = self.vector.search_by_vector(query_vector=self.example_embedding)
+        assert len(hits_by_vector) == 1
+
+    def search_by_full_text(self):
+        hits_by_full_text = self.vector.search_by_full_text(query=get_example_text())
+        assert len(hits_by_full_text) == 0
+
+
+def test_baidu_vector(setup_mock_redis, setup_baiduvectordb_mock):
+    BaiduVectorTest().run_all_tests()
diff --git a/docker/.env.example b/docker/.env.example
index 87d7709a1830a..c4eae46cb0215 100644
--- a/docker/.env.example
+++ b/docker/.env.example
@@ -462,6 +462,15 @@ ELASTICSEARCH_PORT=9200
 ELASTICSEARCH_USERNAME=elastic
 ELASTICSEARCH_PASSWORD=elastic
 
+# baidu vector configurations, only available when VECTOR_STORE is `baidu`
+BAIDU_VECTOR_DB_ENDPOINT=http://127.0.0.1:5287
+BAIDU_VECTOR_DB_CONNECTION_TIMEOUT_MS=30000
+BAIDU_VECTOR_DB_ACCOUNT=root
+BAIDU_VECTOR_DB_API_KEY=dify
+BAIDU_VECTOR_DB_DATABASE=dify
+BAIDU_VECTOR_DB_SHARD=1
+BAIDU_VECTOR_DB_REPLICAS=3
+
 # ------------------------------
 # Knowledge Configuration
 # ------------------------------
diff --git a/docker/docker-compose.yaml b/docker/docker-compose.yaml
index 62d798a695967..c046c17ef8f2b 100644
--- a/docker/docker-compose.yaml
+++ b/docker/docker-compose.yaml
@@ -165,6 +165,13 @@ x-shared-env: &shared-api-worker-env
   TENCENT_VECTOR_DB_DATABASE: ${TENCENT_VECTOR_DB_DATABASE:-dify}
   TENCENT_VECTOR_DB_SHARD: ${TENCENT_VECTOR_DB_SHARD:-1}
   TENCENT_VECTOR_DB_REPLICAS: ${TENCENT_VECTOR_DB_REPLICAS:-2}
+  BAIDU_VECTOR_DB_ENDPOINT: ${BAIDU_VECTOR_DB_ENDPOINT:-http://127.0.0.1:5287}
+  BAIDU_VECTOR_DB_CONNECTION_TIMEOUT_MS: ${BAIDU_VECTOR_DB_CONNECTION_TIMEOUT_MS:-30000}
+  BAIDU_VECTOR_DB_ACCOUNT: ${BAIDU_VECTOR_DB_ACCOUNT:-root}
+  BAIDU_VECTOR_DB_API_KEY: ${BAIDU_VECTOR_DB_API_KEY:-dify}
+  BAIDU_VECTOR_DB_DATABASE: ${BAIDU_VECTOR_DB_DATABASE:-dify}
+  BAIDU_VECTOR_DB_SHARD: ${BAIDU_VECTOR_DB_SHARD:-1}
+  BAIDU_VECTOR_DB_REPLICAS: ${BAIDU_VECTOR_DB_REPLICAS:-3}
   UPLOAD_FILE_SIZE_LIMIT: ${UPLOAD_FILE_SIZE_LIMIT:-15}
   UPLOAD_FILE_BATCH_LIMIT: ${UPLOAD_FILE_BATCH_LIMIT:-5}
   ETL_TYPE: ${ETL_TYPE:-dify}

From 70c5b23089584bcd700ddba3fa9d1ced4d0a57fd Mon Sep 17 00:00:00 2001
From: "github-actions[bot]"
 <41898282+github-actions[bot]@users.noreply.github.com>
Date: Sat, 12 Oct 2024 23:27:11 +0800
Subject: [PATCH 220/235] chore: translate i18n files (#9284)

Co-authored-by: YIXIAO0 <54782454+YIXIAO0@users.noreply.github.com>
Co-authored-by: crazywoola <100913391+crazywoola@users.noreply.github.com>
---
 web/i18n/de-DE/workflow.ts   | 1 +
 web/i18n/es-ES/workflow.ts   | 1 +
 web/i18n/fa-IR/workflow.ts   | 1 +
 web/i18n/fr-FR/workflow.ts   | 1 +
 web/i18n/hi-IN/workflow.ts   | 1 +
 web/i18n/it-IT/workflow.ts   | 1 +
 web/i18n/ja-JP/workflow.ts   | 1 +
 web/i18n/ko-KR/workflow.ts   | 1 +
 web/i18n/pl-PL/workflow.ts   | 1 +
 web/i18n/pt-BR/workflow.ts   | 1 +
 web/i18n/ro-RO/workflow.ts   | 1 +
 web/i18n/ru-RU/workflow.ts   | 1 +
 web/i18n/tr-TR/workflow.ts   | 1 +
 web/i18n/uk-UA/workflow.ts   | 1 +
 web/i18n/vi-VN/workflow.ts   | 1 +
 web/i18n/zh-Hant/workflow.ts | 1 +
 16 files changed, 16 insertions(+)

diff --git a/web/i18n/de-DE/workflow.ts b/web/i18n/de-DE/workflow.ts
index c01d0e6f99418..b6d0e8cde4ee2 100644
--- a/web/i18n/de-DE/workflow.ts
+++ b/web/i18n/de-DE/workflow.ts
@@ -182,6 +182,7 @@ const translation = {
       rerankModel: 'Neusortierungsmodell',
     },
     invalidVariable: 'Ungültige Variable',
+    rerankModelRequired: 'Bevor Sie das Rerank-Modell aktivieren, bestätigen Sie bitte, dass das Modell in den Einstellungen erfolgreich konfiguriert wurde.',
   },
   singleRun: {
     testRun: 'Testlauf ',
diff --git a/web/i18n/es-ES/workflow.ts b/web/i18n/es-ES/workflow.ts
index 2260631d0fa96..275149a056094 100644
--- a/web/i18n/es-ES/workflow.ts
+++ b/web/i18n/es-ES/workflow.ts
@@ -182,6 +182,7 @@ const translation = {
       rerankModel: 'Modelo de reordenamiento',
     },
     invalidVariable: 'Variable no válida',
+    rerankModelRequired: 'Antes de activar el modelo de reclasificación, confirme que el modelo se ha configurado correctamente en la configuración.',
   },
   singleRun: {
     testRun: 'Ejecución de prueba',
diff --git a/web/i18n/fa-IR/workflow.ts b/web/i18n/fa-IR/workflow.ts
index eb36dfdc88836..609f446b43e64 100644
--- a/web/i18n/fa-IR/workflow.ts
+++ b/web/i18n/fa-IR/workflow.ts
@@ -182,6 +182,7 @@ const translation = {
       rerankModel: 'مدل مجدد رتبه‌بندی',
     },
     invalidVariable: 'متغیر نامعتبر',
+    rerankModelRequired: 'قبل از روشن کردن Rerank Model، لطفا تأیید کنید که مدل با موفقیت در تنظیمات پیکربندی شده است.',
   },
   singleRun: {
     testRun: 'اجرای آزمایشی',
diff --git a/web/i18n/fr-FR/workflow.ts b/web/i18n/fr-FR/workflow.ts
index 878d25804e3b3..068c41b853d83 100644
--- a/web/i18n/fr-FR/workflow.ts
+++ b/web/i18n/fr-FR/workflow.ts
@@ -182,6 +182,7 @@ const translation = {
       rerankModel: 'Modèle de rerank',
     },
     invalidVariable: 'Variable invalide',
+    rerankModelRequired: 'Avant d’activer le modèle de reclassement, veuillez confirmer que le modèle a été correctement configuré dans les paramètres.',
   },
   singleRun: {
     testRun: 'Exécution de test',
diff --git a/web/i18n/hi-IN/workflow.ts b/web/i18n/hi-IN/workflow.ts
index ac356c206758f..e402200462d21 100644
--- a/web/i18n/hi-IN/workflow.ts
+++ b/web/i18n/hi-IN/workflow.ts
@@ -185,6 +185,7 @@ const translation = {
       rerankModel: 'पुनः रैंक मॉडल',
     },
     invalidVariable: 'अमान्य वेरिएबल',
+    rerankModelRequired: 'Rerank मॉडल चालू करने से पहले, कृपया पुष्टि करें कि मॉडल को सेटिंग्स में सफलतापूर्वक कॉन्फ़िगर किया गया है।',
   },
   singleRun: {
     testRun: 'परीक्षण रन',
diff --git a/web/i18n/it-IT/workflow.ts b/web/i18n/it-IT/workflow.ts
index 0427a45cd95f0..ce460ed252d7f 100644
--- a/web/i18n/it-IT/workflow.ts
+++ b/web/i18n/it-IT/workflow.ts
@@ -187,6 +187,7 @@ const translation = {
       rerankModel: 'Modello Rerank',
     },
     invalidVariable: 'Variabile non valida',
+    rerankModelRequired: 'Prima di attivare il modello di reranking, conferma che il modello è stato configurato correttamente nelle impostazioni.',
   },
   singleRun: {
     testRun: 'Esecuzione Test ',
diff --git a/web/i18n/ja-JP/workflow.ts b/web/i18n/ja-JP/workflow.ts
index 48c20196016b0..2906f7ef8c90e 100644
--- a/web/i18n/ja-JP/workflow.ts
+++ b/web/i18n/ja-JP/workflow.ts
@@ -182,6 +182,7 @@ const translation = {
       rerankModel: 'Rerankモデル',
     },
     invalidVariable: '無効な変数',
+    rerankModelRequired: 'モデルの再ランク付けをオンにする前に、モデルが設定で正常に構成されていることを確認してください。',
   },
   singleRun: {
     testRun: 'テスト実行',
diff --git a/web/i18n/ko-KR/workflow.ts b/web/i18n/ko-KR/workflow.ts
index 4a97943790903..99d5c47c0bffb 100644
--- a/web/i18n/ko-KR/workflow.ts
+++ b/web/i18n/ko-KR/workflow.ts
@@ -182,6 +182,7 @@ const translation = {
       rerankModel: '재정렬 모델',
     },
     invalidVariable: '잘못된 변수',
+    rerankModelRequired: 'Rerank Model을 켜기 전에 설정에서 모델이 성공적으로 구성되었는지 확인하십시오.',
   },
   singleRun: {
     testRun: '테스트 실행',
diff --git a/web/i18n/pl-PL/workflow.ts b/web/i18n/pl-PL/workflow.ts
index 41927668f7bcd..b26c429fb1939 100644
--- a/web/i18n/pl-PL/workflow.ts
+++ b/web/i18n/pl-PL/workflow.ts
@@ -182,6 +182,7 @@ const translation = {
       rerankModel: 'Model rerank',
     },
     invalidVariable: 'Nieprawidłowa zmienna',
+    rerankModelRequired: 'Przed włączeniem Rerank Model upewnij się, że model został pomyślnie skonfigurowany w ustawieniach.',
   },
   singleRun: {
     testRun: 'Testowe uruchomienie ',
diff --git a/web/i18n/pt-BR/workflow.ts b/web/i18n/pt-BR/workflow.ts
index 222fc788bfdfe..9092ccda3e41d 100644
--- a/web/i18n/pt-BR/workflow.ts
+++ b/web/i18n/pt-BR/workflow.ts
@@ -182,6 +182,7 @@ const translation = {
       rerankModel: 'Modelo de reordenação',
     },
     invalidVariable: 'Variável inválida',
+    rerankModelRequired: 'Antes de ativar o modelo de reclassificação, confirme se o modelo foi configurado com sucesso nas configurações.',
   },
   singleRun: {
     testRun: 'Execução de teste ',
diff --git a/web/i18n/ro-RO/workflow.ts b/web/i18n/ro-RO/workflow.ts
index ac4b718b072ab..bb66169da89f8 100644
--- a/web/i18n/ro-RO/workflow.ts
+++ b/web/i18n/ro-RO/workflow.ts
@@ -182,6 +182,7 @@ const translation = {
       rerankModel: 'Model de rerankare',
     },
     invalidVariable: 'Variabilă invalidă',
+    rerankModelRequired: 'Înainte de a activa modelul de reclasificare, vă rugăm să confirmați că modelul a fost configurat cu succes în setări.',
   },
   singleRun: {
     testRun: 'Rulare de test ',
diff --git a/web/i18n/ru-RU/workflow.ts b/web/i18n/ru-RU/workflow.ts
index 19318638957f0..5b2bc7e2902c5 100644
--- a/web/i18n/ru-RU/workflow.ts
+++ b/web/i18n/ru-RU/workflow.ts
@@ -182,6 +182,7 @@ const translation = {
       rerankModel: 'Модель переранжирования',
     },
     invalidVariable: 'Неверная переменная',
+    rerankModelRequired: 'Перед включением модели повторного ранжирования убедитесь, что модель успешно настроена в настройках.',
   },
   singleRun: {
     testRun: 'Тестовый запуск ',
diff --git a/web/i18n/tr-TR/workflow.ts b/web/i18n/tr-TR/workflow.ts
index a33a3724ad48e..8e1ce596309d8 100644
--- a/web/i18n/tr-TR/workflow.ts
+++ b/web/i18n/tr-TR/workflow.ts
@@ -182,6 +182,7 @@ const translation = {
       rerankModel: 'Yeniden Sıralama Modeli',
     },
     invalidVariable: 'Geçersiz değişken',
+    rerankModelRequired: 'Yeniden Sıralama Modelini açmadan önce, lütfen ayarlarda modelin başarıyla yapılandırıldığını onaylayın.',
   },
   singleRun: {
     testRun: 'Test Çalıştırma',
diff --git a/web/i18n/uk-UA/workflow.ts b/web/i18n/uk-UA/workflow.ts
index e1bea99bcd2a8..f7747541ccfca 100644
--- a/web/i18n/uk-UA/workflow.ts
+++ b/web/i18n/uk-UA/workflow.ts
@@ -182,6 +182,7 @@ const translation = {
       rerankModel: 'Модель повторного ранжування',
     },
     invalidVariable: 'Недійсна змінна',
+    rerankModelRequired: 'Перед увімкненням Rerank Model, будь ласка, підтвердьте, що модель успішно налаштована в налаштуваннях.',
   },
   singleRun: {
     testRun: 'Тестовий запуск',
diff --git a/web/i18n/vi-VN/workflow.ts b/web/i18n/vi-VN/workflow.ts
index 5e6f0e00631f3..aa9fbf865d6b3 100644
--- a/web/i18n/vi-VN/workflow.ts
+++ b/web/i18n/vi-VN/workflow.ts
@@ -182,6 +182,7 @@ const translation = {
       rerankModel: 'Mô hình xếp hạng lại',
     },
     invalidVariable: 'Biến không hợp lệ',
+    rerankModelRequired: 'Trước khi bật Mô hình xếp hạng lại, vui lòng xác nhận rằng mô hình đã được định cấu hình thành công trong cài đặt.',
   },
   singleRun: {
     testRun: 'Chạy thử nghiệm ',
diff --git a/web/i18n/zh-Hant/workflow.ts b/web/i18n/zh-Hant/workflow.ts
index 8e1b7529fe18d..35ed68c437ed5 100644
--- a/web/i18n/zh-Hant/workflow.ts
+++ b/web/i18n/zh-Hant/workflow.ts
@@ -182,6 +182,7 @@ const translation = {
       rerankModel: 'Rerank 模型',
     },
     invalidVariable: '無效的變量',
+    rerankModelRequired: '在開啟 Rerank 模型之前，請在設置中確認模型配置成功。',
   },
   singleRun: {
     testRun: '測試運行',

From dbfbc56de7e840b1e76a5a72d81dbd23c623f45a Mon Sep 17 00:00:00 2001
From: Wu Tianwei <30284043+WTW0313@users.noreply.github.com>
Date: Sat, 12 Oct 2024 23:40:38 +0800
Subject: [PATCH 221/235] feat: refresh-token (#9286)

Co-authored-by: NFish <douxc512@gmail.com>
---
 web/app/components/swr-initor.tsx | 24 ++++++--
 web/app/signin/normalForm.tsx     |  6 +-
 web/app/signin/userSSOForm.tsx    | 11 +++-
 web/hooks/use-refresh-token.ts    | 92 +++++++++++++++++++++++++++++++
 web/package.json                  |  1 +
 web/service/common.ts             | 17 +++++-
 web/utils/index.ts                | 18 ++++++
 web/yarn.lock                     |  5 ++
 8 files changed, 163 insertions(+), 11 deletions(-)
 create mode 100644 web/hooks/use-refresh-token.ts

diff --git a/web/app/components/swr-initor.tsx b/web/app/components/swr-initor.tsx
index afb4c58cbcda1..85e05499e6d1e 100644
--- a/web/app/components/swr-initor.tsx
+++ b/web/app/components/swr-initor.tsx
@@ -4,6 +4,7 @@ import { SWRConfig } from 'swr'
 import { useEffect, useState } from 'react'
 import type { ReactNode } from 'react'
 import { useRouter, useSearchParams } from 'next/navigation'
+import useRefreshToken from '@/hooks/use-refresh-token'
 
 type SwrInitorProps = {
   children: ReactNode
@@ -13,18 +14,31 @@ const SwrInitor = ({
 }: SwrInitorProps) => {
   const router = useRouter()
   const searchParams = useSearchParams()
-  const consoleToken = searchParams.get('console_token')
+  const consoleToken = searchParams.get('access_token')
+  const refreshToken = searchParams.get('refresh_token')
   const consoleTokenFromLocalStorage = localStorage?.getItem('console_token')
+  const refreshTokenFromLocalStorage = localStorage?.getItem('refresh_token')
   const [init, setInit] = useState(false)
+  const { getNewAccessToken } = useRefreshToken()
 
   useEffect(() => {
-    if (!(consoleToken || consoleTokenFromLocalStorage))
+    if (!(consoleToken || refreshToken || consoleTokenFromLocalStorage || refreshTokenFromLocalStorage)) {
       router.replace('/signin')
+      return
+    }
+    if (consoleTokenFromLocalStorage && refreshTokenFromLocalStorage)
+      getNewAccessToken(consoleTokenFromLocalStorage, refreshTokenFromLocalStorage)
 
-    if (consoleToken) {
-      localStorage?.setItem('console_token', consoleToken!)
-      router.replace('/apps', { forceOptimisticNavigation: false } as any)
+    if (consoleToken && refreshToken) {
+      localStorage.setItem('console_token', consoleToken)
+      localStorage.setItem('refresh_token', refreshToken)
+      getNewAccessToken(consoleToken, refreshToken).then(() => {
+        router.replace('/apps', { forceOptimisticNavigation: false } as any)
+      }).catch(() => {
+        router.replace('/signin')
+      })
     }
+
     setInit(true)
   }, [])
 
diff --git a/web/app/signin/normalForm.tsx b/web/app/signin/normalForm.tsx
index 816df8007d9e7..0ae4eb1f43210 100644
--- a/web/app/signin/normalForm.tsx
+++ b/web/app/signin/normalForm.tsx
@@ -11,6 +11,7 @@ import { IS_CE_EDITION, SUPPORT_MAIL_LOGIN, apiPrefix, emailRegex } from '@/conf
 import Button from '@/app/components/base/button'
 import { login, oauth } from '@/service/common'
 import { getPurifyHref } from '@/utils'
+import useRefreshToken from '@/hooks/use-refresh-token'
 
 type IState = {
   formValid: boolean
@@ -61,6 +62,7 @@ function reducer(state: IState, action: IAction) {
 
 const NormalForm = () => {
   const { t } = useTranslation()
+  const { getNewAccessToken } = useRefreshToken()
   const useEmailLogin = IS_CE_EDITION || SUPPORT_MAIL_LOGIN
 
   const router = useRouter()
@@ -95,7 +97,9 @@ const NormalForm = () => {
         },
       })
       if (res.result === 'success') {
-        localStorage.setItem('console_token', res.data)
+        localStorage.setItem('console_token', res.data.access_token)
+        localStorage.setItem('refresh_token', res.data.refresh_token)
+        getNewAccessToken(res.data.access_token, res.data.refresh_token)
         router.replace('/apps')
       }
       else {
diff --git a/web/app/signin/userSSOForm.tsx b/web/app/signin/userSSOForm.tsx
index 9cd889a0a5129..e4b61413bc0ef 100644
--- a/web/app/signin/userSSOForm.tsx
+++ b/web/app/signin/userSSOForm.tsx
@@ -7,6 +7,7 @@ import cn from '@/utils/classnames'
 import Toast from '@/app/components/base/toast'
 import { getUserOAuth2SSOUrl, getUserOIDCSSOUrl, getUserSAMLSSOUrl } from '@/service/sso'
 import Button from '@/app/components/base/button'
+import useRefreshToken from '@/hooks/use-refresh-token'
 
 type UserSSOFormProps = {
   protocol: string
@@ -15,8 +16,10 @@ type UserSSOFormProps = {
 const UserSSOForm: FC<UserSSOFormProps> = ({
   protocol,
 }) => {
+  const { getNewAccessToken } = useRefreshToken()
   const searchParams = useSearchParams()
-  const consoleToken = searchParams.get('console_token')
+  const consoleToken = searchParams.get('access_token')
+  const refreshToken = searchParams.get('refresh_token')
   const message = searchParams.get('message')
 
   const router = useRouter()
@@ -25,8 +28,10 @@ const UserSSOForm: FC<UserSSOFormProps> = ({
   const [isLoading, setIsLoading] = useState(false)
 
   useEffect(() => {
-    if (consoleToken) {
+    if (refreshToken && consoleToken) {
       localStorage.setItem('console_token', consoleToken)
+      localStorage.setItem('refresh_token', refreshToken)
+      getNewAccessToken(consoleToken, refreshToken)
       router.replace('/apps')
     }
 
@@ -36,7 +41,7 @@ const UserSSOForm: FC<UserSSOFormProps> = ({
         message,
       })
     }
-  }, [])
+  }, [consoleToken, refreshToken, message, router])
 
   const handleSSOLogin = () => {
     setIsLoading(true)
diff --git a/web/hooks/use-refresh-token.ts b/web/hooks/use-refresh-token.ts
new file mode 100644
index 0000000000000..3d8779636f4b4
--- /dev/null
+++ b/web/hooks/use-refresh-token.ts
@@ -0,0 +1,92 @@
+'use client'
+import { useCallback, useEffect, useRef } from 'react'
+import { jwtDecode } from 'jwt-decode'
+import dayjs from 'dayjs'
+import utc from 'dayjs/plugin/utc'
+import { useRouter } from 'next/navigation'
+import type { CommonResponse } from '@/models/common'
+import { fetchNewToken } from '@/service/common'
+import { fetchWithRetry } from '@/utils'
+
+dayjs.extend(utc)
+
+const useRefreshToken = () => {
+  const router = useRouter()
+  const timer = useRef<NodeJS.Timeout>()
+  const advanceTime = useRef<number>(5 * 60 * 1000)
+  const interval = useRef<number>(55 * 60 * 1000)
+
+  const getExpireTime = useCallback((token: string) => {
+    if (!token)
+      return 0
+    const decoded = jwtDecode(token)
+    return (decoded.exp || 0) * 1000
+  }, [])
+
+  const getCurrentTimeStamp = useCallback(() => {
+    return dayjs.utc().valueOf()
+  }, [])
+
+  const handleError = useCallback(() => {
+    localStorage?.removeItem('is_refreshing')
+    localStorage?.removeItem('console_token')
+    localStorage?.removeItem('refresh_token')
+    localStorage?.removeItem('last_refresh_time')
+    router.replace('/signin')
+  }, [])
+
+  const getNewAccessToken = useCallback(async (currentAccessToken: string, currentRefreshToken: string) => {
+    if (localStorage?.getItem('is_refreshing') === '1')
+      return null
+    const currentTokenExpireTime = getExpireTime(currentAccessToken)
+    let lastRefreshTime = parseInt(localStorage?.getItem('last_refresh_time') || '0')
+    lastRefreshTime = isNaN(lastRefreshTime) ? 0 : lastRefreshTime
+    if (getCurrentTimeStamp() + advanceTime.current > currentTokenExpireTime
+      && lastRefreshTime + interval.current < getCurrentTimeStamp()) {
+      localStorage?.setItem('is_refreshing', '1')
+      const [e, res] = await fetchWithRetry(fetchNewToken({
+        body: { refresh_token: currentRefreshToken },
+      }) as Promise<CommonResponse & { data: { access_token: string; refresh_token: string } }>)
+      if (e) {
+        handleError()
+        return e
+      }
+      const { access_token, refresh_token } = res.data
+      localStorage?.setItem('is_refreshing', '0')
+      localStorage?.setItem('last_refresh_time', getCurrentTimeStamp().toString())
+      localStorage?.setItem('console_token', access_token)
+      localStorage?.setItem('refresh_token', refresh_token)
+      const newTokenExpireTime = getExpireTime(access_token)
+      timer.current = setTimeout(() => {
+        const consoleTokenFromLocalStorage = localStorage?.getItem('console_token')
+        const refreshTokenFromLocalStorage = localStorage?.getItem('refresh_token')
+        if (consoleTokenFromLocalStorage && refreshTokenFromLocalStorage)
+          getNewAccessToken(consoleTokenFromLocalStorage, refreshTokenFromLocalStorage)
+      }, newTokenExpireTime - advanceTime.current - getCurrentTimeStamp())
+    }
+    else {
+      const newTokenExpireTime = getExpireTime(currentAccessToken)
+      timer.current = setTimeout(() => {
+        const consoleTokenFromLocalStorage = localStorage?.getItem('console_token')
+        const refreshTokenFromLocalStorage = localStorage?.getItem('refresh_token')
+        if (consoleTokenFromLocalStorage && refreshTokenFromLocalStorage)
+          getNewAccessToken(consoleTokenFromLocalStorage, refreshTokenFromLocalStorage)
+      }, newTokenExpireTime - advanceTime.current - getCurrentTimeStamp())
+    }
+    return null
+  }, [getExpireTime, getCurrentTimeStamp, handleError])
+
+  useEffect(() => {
+    return () => {
+      clearTimeout(timer.current)
+      localStorage?.removeItem('is_refreshing')
+      localStorage?.removeItem('last_refresh_time')
+    }
+  }, [])
+
+  return {
+    getNewAccessToken,
+  }
+}
+
+export default useRefreshToken
diff --git a/web/package.json b/web/package.json
index 96b89c92315d9..8a17997fe963c 100644
--- a/web/package.json
+++ b/web/package.json
@@ -55,6 +55,7 @@
     "immer": "^9.0.19",
     "js-audio-recorder": "^1.0.7",
     "js-cookie": "^3.0.1",
+    "jwt-decode": "^4.0.0",
     "katex": "^0.16.10",
     "lamejs": "^1.2.1",
     "lexical": "^0.16.0",
diff --git a/web/service/common.ts b/web/service/common.ts
index 3fbcde2a2ab3b..bd3dbca0aaedb 100644
--- a/web/service/common.ts
+++ b/web/service/common.ts
@@ -38,8 +38,21 @@ import type {
 import type { RETRIEVE_METHOD } from '@/types/app'
 import type { SystemFeatures } from '@/types/feature'
 
-export const login: Fetcher<CommonResponse & { data: string }, { url: string; body: Record<string, any> }> = ({ url, body }) => {
-  return post(url, { body }) as Promise<CommonResponse & { data: string }>
+type LoginSuccess = {
+  result: 'success'
+  data: { access_token: string;refresh_token: string }
+}
+type LoginFail = {
+  result: 'fail'
+  data: string
+}
+type LoginResponse = LoginSuccess | LoginFail
+export const login: Fetcher<LoginResponse, { url: string; body: Record<string, any> }> = ({ url, body }) => {
+  return post(url, { body }) as Promise<LoginResponse>
+}
+
+export const fetchNewToken: Fetcher<CommonResponse & { data: { access_token: string; refresh_token: string } }, { body: Record<string, any> }> = ({ body }) => {
+  return post('/refresh-token', { body }) as Promise<CommonResponse & { data: { access_token: string; refresh_token: string } }>
 }
 
 export const setup: Fetcher<CommonResponse, { body: Record<string, any> }> = ({ body }) => {
diff --git a/web/utils/index.ts b/web/utils/index.ts
index 8afd8afae70d6..7aa6fef0a88c0 100644
--- a/web/utils/index.ts
+++ b/web/utils/index.ts
@@ -39,3 +39,21 @@ export const getPurifyHref = (href: string) => {
 
   return escape(href)
 }
+
+export async function fetchWithRetry<T = any>(fn: Promise<T>, retries = 3): Promise<[Error] | [null, T]> {
+  const [error, res] = await asyncRunSafe(fn)
+  if (error) {
+    if (retries > 0) {
+      const res = await fetchWithRetry(fn, retries - 1)
+      return res
+    }
+    else {
+      if (error instanceof Error)
+        return [error]
+      return [new Error('unknown error')]
+    }
+  }
+  else {
+    return [null, res]
+  }
+}
diff --git a/web/yarn.lock b/web/yarn.lock
index 5d693ff786d8f..4121752dd7de0 100644
--- a/web/yarn.lock
+++ b/web/yarn.lock
@@ -6205,6 +6205,11 @@ jsonc-eslint-parser@^2.0.4, jsonc-eslint-parser@^2.1.0:
     array-includes "^3.1.5"
     object.assign "^4.1.3"
 
+jwt-decode@^4.0.0:
+  version "4.0.0"
+  resolved "https://registry.npmmirror.com/jwt-decode/-/jwt-decode-4.0.0.tgz#2270352425fd413785b2faf11f6e755c5151bd4b"
+  integrity sha512-+KJGIyHgkGuIq3IEBNftfhW/LfWhXUIY6OmyVWjliu5KH1y0fw7VQ8YndE2O4qZdMSd9SqbnC8GOcZEy0Om7sA==
+
 katex@^0.16.0, katex@^0.16.10:
   version "0.16.10"
   resolved "https://registry.npmjs.org/katex/-/katex-0.16.10.tgz"

From f73751843f9c88dea82821e6a109a0e6d225c4e5 Mon Sep 17 00:00:00 2001
From: -LAN- <laipz8200@outlook.com>
Date: Sat, 12 Oct 2024 23:46:30 +0800
Subject: [PATCH 222/235] Feat/implement-refresh-tokens (#9233)

---
 api/.env.example                      |  3 +
 api/app.py                            |  2 +-
 api/configs/feature/__init__.py       | 15 +++--
 api/controllers/console/auth/login.py | 23 +++++--
 api/controllers/console/auth/oauth.py | 11 ++-
 api/controllers/console/setup.py      |  4 +-
 api/libs/helper.py                    |  2 +-
 api/services/account_service.py       | 97 +++++++++++++++++++++------
 docker/.env.example                   |  3 +
 docker/docker-compose.yaml            |  1 +
 10 files changed, 123 insertions(+), 38 deletions(-)

diff --git a/api/.env.example b/api/.env.example
index 3f88fb3cdf5b0..468130b1628e9 100644
--- a/api/.env.example
+++ b/api/.env.example
@@ -20,6 +20,9 @@ FILES_URL=http://127.0.0.1:5001
 # The time in seconds after the signature is rejected
 FILES_ACCESS_TIMEOUT=300
 
+# Access token expiration time in minutes
+ACCESS_TOKEN_EXPIRE_MINUTES=60
+
 # celery configuration
 CELERY_BROKER_URL=redis://:difyai123456@localhost:6379/1
 
diff --git a/api/app.py b/api/app.py
index a251ef5f0f72c..52dd492225339 100644
--- a/api/app.py
+++ b/api/app.py
@@ -183,7 +183,7 @@ def load_user_from_request(request_from_flask_login):
     decoded = PassportService().verify(auth_token)
     user_id = decoded.get("user_id")
 
-    logged_in_account = AccountService.load_logged_in_account(account_id=user_id, token=auth_token)
+    logged_in_account = AccountService.load_logged_in_account(account_id=user_id)
     if logged_in_account:
         contexts.tenant_id.set(logged_in_account.current_tenant_id)
     return logged_in_account
diff --git a/api/configs/feature/__init__.py b/api/configs/feature/__init__.py
index 93dbc1367f394..a3334d16345e9 100644
--- a/api/configs/feature/__init__.py
+++ b/api/configs/feature/__init__.py
@@ -360,9 +360,9 @@ class WorkflowConfig(BaseSettings):
     )
 
 
-class OAuthConfig(BaseSettings):
+class AuthConfig(BaseSettings):
     """
-    Configuration for OAuth authentication
+    Configuration for authentication and OAuth
     """
 
     OAUTH_REDIRECT_PATH: str = Field(
@@ -371,7 +371,7 @@ class OAuthConfig(BaseSettings):
     )
 
     GITHUB_CLIENT_ID: Optional[str] = Field(
-        description="GitHub OAuth client secret",
+        description="GitHub OAuth client ID",
         default=None,
     )
 
@@ -390,6 +390,11 @@ class OAuthConfig(BaseSettings):
         default=None,
     )
 
+    ACCESS_TOKEN_EXPIRE_MINUTES: PositiveInt = Field(
+        description="Expiration time for access tokens in minutes",
+        default=60,
+    )
+
 
 class ModerationConfig(BaseSettings):
     """
@@ -607,6 +612,7 @@ def POSITION_TOOL_EXCLUDES_SET(self) -> set[str]:
 class FeatureConfig(
     # place the configs in alphabet order
     AppExecutionConfig,
+    AuthConfig,  # Changed from OAuthConfig to AuthConfig
     BillingConfig,
     CodeExecutionSandboxConfig,
     DataSetConfig,
@@ -621,14 +627,13 @@ class FeatureConfig(
     MailConfig,
     ModelLoadBalanceConfig,
     ModerationConfig,
-    OAuthConfig,
+    PositionConfig,
     RagEtlConfig,
     SecurityConfig,
     ToolConfig,
     UpdateConfig,
     WorkflowConfig,
     WorkspaceConfig,
-    PositionConfig,
     # hosted services config
     HostedServiceConfig,
     CeleryBeatConfig,
diff --git a/api/controllers/console/auth/login.py b/api/controllers/console/auth/login.py
index 62837af2b9b0e..18a7b2316660c 100644
--- a/api/controllers/console/auth/login.py
+++ b/api/controllers/console/auth/login.py
@@ -7,7 +7,7 @@
 import services
 from controllers.console import api
 from controllers.console.setup import setup_required
-from libs.helper import email, get_remote_ip
+from libs.helper import email, extract_remote_ip
 from libs.password import valid_password
 from models.account import Account
 from services.account_service import AccountService, TenantService
@@ -40,17 +40,16 @@ def post(self):
                 "data": "workspace not found, please contact system admin to invite you to join in a workspace",
             }
 
-        token = AccountService.login(account, ip_address=get_remote_ip(request))
+        token_pair = AccountService.login(account=account, ip_address=extract_remote_ip(request))
 
-        return {"result": "success", "data": token}
+        return {"result": "success", "data": token_pair.model_dump()}
 
 
 class LogoutApi(Resource):
     @setup_required
     def get(self):
         account = cast(Account, flask_login.current_user)
-        token = request.headers.get("Authorization", "").split(" ")[1]
-        AccountService.logout(account=account, token=token)
+        AccountService.logout(account=account)
         flask_login.logout_user()
         return {"result": "success"}
 
@@ -106,5 +105,19 @@ def get(self):
         return {"result": "success"}
 
 
+class RefreshTokenApi(Resource):
+    def post(self):
+        parser = reqparse.RequestParser()
+        parser.add_argument("refresh_token", type=str, required=True, location="json")
+        args = parser.parse_args()
+
+        try:
+            new_token_pair = AccountService.refresh_token(args["refresh_token"])
+            return {"result": "success", "data": new_token_pair.model_dump()}
+        except Exception as e:
+            return {"result": "fail", "data": str(e)}, 401
+
+
 api.add_resource(LoginApi, "/login")
 api.add_resource(LogoutApi, "/logout")
+api.add_resource(RefreshTokenApi, "/refresh-token")
diff --git a/api/controllers/console/auth/oauth.py b/api/controllers/console/auth/oauth.py
index ad0c0580aeaba..c5909b8c1092e 100644
--- a/api/controllers/console/auth/oauth.py
+++ b/api/controllers/console/auth/oauth.py
@@ -9,7 +9,7 @@
 from configs import dify_config
 from constants.languages import languages
 from extensions.ext_database import db
-from libs.helper import get_remote_ip
+from libs.helper import extract_remote_ip
 from libs.oauth import GitHubOAuth, GoogleOAuth, OAuthUserInfo
 from models.account import Account, AccountStatus
 from services.account_service import AccountService, RegisterService, TenantService
@@ -81,9 +81,14 @@ def get(self, provider: str):
 
         TenantService.create_owner_tenant_if_not_exist(account)
 
-        token = AccountService.login(account, ip_address=get_remote_ip(request))
+        token_pair = AccountService.login(
+            account=account,
+            ip_address=extract_remote_ip(request),
+        )
 
-        return redirect(f"{dify_config.CONSOLE_WEB_URL}?console_token={token}")
+        return redirect(
+            f"{dify_config.CONSOLE_WEB_URL}?access_token={token_pair.access_token}&refresh_token={token_pair.refresh_token}"
+        )
 
 
 def _get_account_by_openid_or_email(provider: str, user_info: OAuthUserInfo) -> Optional[Account]:
diff --git a/api/controllers/console/setup.py b/api/controllers/console/setup.py
index 46b4ef5d87a8a..15a4af118b05b 100644
--- a/api/controllers/console/setup.py
+++ b/api/controllers/console/setup.py
@@ -4,7 +4,7 @@
 from flask_restful import Resource, reqparse
 
 from configs import dify_config
-from libs.helper import StrLen, email, get_remote_ip
+from libs.helper import StrLen, email, extract_remote_ip
 from libs.password import valid_password
 from models.model import DifySetup
 from services.account_service import RegisterService, TenantService
@@ -46,7 +46,7 @@ def post(self):
 
         # setup
         RegisterService.setup(
-            email=args["email"], name=args["name"], password=args["password"], ip_address=get_remote_ip(request)
+            email=args["email"], name=args["name"], password=args["password"], ip_address=extract_remote_ip(request)
         )
 
         return {"result": "success"}, 201
diff --git a/api/libs/helper.py b/api/libs/helper.py
index 9c3a1ff04d1f8..d8a8e7f4118d8 100644
--- a/api/libs/helper.py
+++ b/api/libs/helper.py
@@ -162,7 +162,7 @@ def generate_string(n):
     return result
 
 
-def get_remote_ip(request) -> str:
+def extract_remote_ip(request) -> str:
     if request.headers.get("CF-Connecting-IP"):
         return request.headers.get("Cf-Connecting-Ip")
     elif request.headers.getlist("X-Forwarded-For"):
diff --git a/api/services/account_service.py b/api/services/account_service.py
index 05b505f8a6218..eda6011aef516 100644
--- a/api/services/account_service.py
+++ b/api/services/account_service.py
@@ -7,6 +7,7 @@
 from hashlib import sha256
 from typing import Any, Optional
 
+from pydantic import BaseModel
 from sqlalchemy import func
 from werkzeug.exceptions import Unauthorized
 
@@ -49,9 +50,39 @@
 from tasks.mail_reset_password_task import send_reset_password_mail_task
 
 
+class TokenPair(BaseModel):
+    access_token: str
+    refresh_token: str
+
+
+REFRESH_TOKEN_PREFIX = "refresh_token:"
+ACCOUNT_REFRESH_TOKEN_PREFIX = "account_refresh_token:"
+REFRESH_TOKEN_EXPIRY = timedelta(days=30)
+
+
 class AccountService:
     reset_password_rate_limiter = RateLimiter(prefix="reset_password_rate_limit", max_attempts=5, time_window=60 * 60)
 
+    @staticmethod
+    def _get_refresh_token_key(refresh_token: str) -> str:
+        return f"{REFRESH_TOKEN_PREFIX}{refresh_token}"
+
+    @staticmethod
+    def _get_account_refresh_token_key(account_id: str) -> str:
+        return f"{ACCOUNT_REFRESH_TOKEN_PREFIX}{account_id}"
+
+    @staticmethod
+    def _store_refresh_token(refresh_token: str, account_id: str) -> None:
+        redis_client.setex(AccountService._get_refresh_token_key(refresh_token), REFRESH_TOKEN_EXPIRY, account_id)
+        redis_client.setex(
+            AccountService._get_account_refresh_token_key(account_id), REFRESH_TOKEN_EXPIRY, refresh_token
+        )
+
+    @staticmethod
+    def _delete_refresh_token(refresh_token: str, account_id: str) -> None:
+        redis_client.delete(AccountService._get_refresh_token_key(refresh_token))
+        redis_client.delete(AccountService._get_account_refresh_token_key(account_id))
+
     @staticmethod
     def load_user(user_id: str) -> None | Account:
         account = Account.query.filter_by(id=user_id).first()
@@ -61,9 +92,7 @@ def load_user(user_id: str) -> None | Account:
         if account.status in {AccountStatus.BANNED.value, AccountStatus.CLOSED.value}:
             raise Unauthorized("Account is banned or closed.")
 
-        current_tenant: TenantAccountJoin = TenantAccountJoin.query.filter_by(
-            account_id=account.id, current=True
-        ).first()
+        current_tenant = TenantAccountJoin.query.filter_by(account_id=account.id, current=True).first()
         if current_tenant:
             account.current_tenant_id = current_tenant.tenant_id
         else:
@@ -84,10 +113,12 @@ def load_user(user_id: str) -> None | Account:
         return account
 
     @staticmethod
-    def get_account_jwt_token(account, *, exp: timedelta = timedelta(days=30)):
+    def get_account_jwt_token(account: Account) -> str:
+        exp_dt = datetime.now(timezone.utc) + timedelta(minutes=dify_config.ACCESS_TOKEN_EXPIRE_MINUTES)
+        exp = int(exp_dt.timestamp())
         payload = {
             "user_id": account.id,
-            "exp": datetime.now(timezone.utc).replace(tzinfo=None) + exp,
+            "exp": exp,
             "iss": dify_config.EDITION,
             "sub": "Console API Passport",
         }
@@ -213,7 +244,7 @@ def update_account(account, **kwargs):
         return account
 
     @staticmethod
-    def update_last_login(account: Account, *, ip_address: str) -> None:
+    def update_login_info(account: Account, *, ip_address: str) -> None:
         """Update last login time and ip"""
         account.last_login_at = datetime.now(timezone.utc).replace(tzinfo=None)
         account.last_login_ip = ip_address
@@ -221,22 +252,45 @@ def update_last_login(account: Account, *, ip_address: str) -> None:
         db.session.commit()
 
     @staticmethod
-    def login(account: Account, *, ip_address: Optional[str] = None):
+    def login(account: Account, *, ip_address: Optional[str] = None) -> TokenPair:
         if ip_address:
-            AccountService.update_last_login(account, ip_address=ip_address)
-        exp = timedelta(days=30)
-        token = AccountService.get_account_jwt_token(account, exp=exp)
-        redis_client.set(_get_login_cache_key(account_id=account.id, token=token), "1", ex=int(exp.total_seconds()))
-        return token
+            AccountService.update_login_info(account=account, ip_address=ip_address)
+
+        access_token = AccountService.get_account_jwt_token(account=account)
+        refresh_token = _generate_refresh_token()
+
+        AccountService._store_refresh_token(refresh_token, account.id)
+
+        return TokenPair(access_token=access_token, refresh_token=refresh_token)
 
     @staticmethod
-    def logout(*, account: Account, token: str):
-        redis_client.delete(_get_login_cache_key(account_id=account.id, token=token))
+    def logout(*, account: Account) -> None:
+        refresh_token = redis_client.get(AccountService._get_account_refresh_token_key(account.id))
+        if refresh_token:
+            AccountService._delete_refresh_token(refresh_token.decode("utf-8"), account.id)
 
     @staticmethod
-    def load_logged_in_account(*, account_id: str, token: str):
-        if not redis_client.get(_get_login_cache_key(account_id=account_id, token=token)):
-            return None
+    def refresh_token(refresh_token: str) -> TokenPair:
+        # Verify the refresh token
+        account_id = redis_client.get(AccountService._get_refresh_token_key(refresh_token))
+        if not account_id:
+            raise ValueError("Invalid refresh token")
+
+        account = AccountService.load_user(account_id.decode("utf-8"))
+        if not account:
+            raise ValueError("Invalid account")
+
+        # Generate new access token and refresh token
+        new_access_token = AccountService.get_account_jwt_token(account)
+        new_refresh_token = _generate_refresh_token()
+
+        AccountService._delete_refresh_token(refresh_token, account.id)
+        AccountService._store_refresh_token(new_refresh_token, account.id)
+
+        return TokenPair(access_token=new_access_token, refresh_token=new_refresh_token)
+
+    @staticmethod
+    def load_logged_in_account(*, account_id: str):
         return AccountService.load_user(account_id)
 
     @classmethod
@@ -258,10 +312,6 @@ def get_reset_password_data(cls, token: str) -> Optional[dict[str, Any]]:
         return TokenManager.get_token_data(token, "reset_password")
 
 
-def _get_login_cache_key(*, account_id: str, token: str):
-    return f"account_login:{account_id}:{token}"
-
-
 class TenantService:
     @staticmethod
     def create_tenant(name: str) -> Tenant:
@@ -698,3 +748,8 @@ def _get_invitation_by_token(cls, token: str, workspace_id: str, email: str) ->
 
             invitation = json.loads(data)
             return invitation
+
+
+def _generate_refresh_token(length: int = 64):
+    token = secrets.token_hex(length)
+    return token
diff --git a/docker/.env.example b/docker/.env.example
index c4eae46cb0215..c7b4f38d2ec11 100644
--- a/docker/.env.example
+++ b/docker/.env.example
@@ -91,6 +91,9 @@ MIGRATION_ENABLED=true
 # The default value is 300 seconds.
 FILES_ACCESS_TIMEOUT=300
 
+# Access token expiration time in minutes
+ACCESS_TOKEN_EXPIRE_MINUTES=60
+
 # The maximum number of active requests for the application, where 0 means unlimited, should be a non-negative integer.
 APP_MAX_ACTIVE_REQUESTS=0
 
diff --git a/docker/docker-compose.yaml b/docker/docker-compose.yaml
index c046c17ef8f2b..cb6ecba2791db 100644
--- a/docker/docker-compose.yaml
+++ b/docker/docker-compose.yaml
@@ -47,6 +47,7 @@ x-shared-env: &shared-api-worker-env
   REDIS_SENTINEL_SERVICE_NAME: ${REDIS_SENTINEL_SERVICE_NAME:-}
   REDIS_SENTINEL_USERNAME: ${REDIS_SENTINEL_USERNAME:-}
   REDIS_SENTINEL_PASSWORD: ${REDIS_SENTINEL_PASSWORD:-}
+  ACCESS_TOKEN_EXPIRE_MINUTES: ${ACCESS_TOKEN_EXPIRE_MINUTES:-60}
   REDIS_SENTINEL_SOCKET_TIMEOUT: ${REDIS_SENTINEL_SOCKET_TIMEOUT:-0.1}
   CELERY_BROKER_URL: ${CELERY_BROKER_URL:-redis://:difyai123456@redis:6379/1}
   BROKER_USE_SSL: ${BROKER_USE_SSL:-false}

From ea6734f55009e6a031ff19255a7bc630728818da Mon Sep 17 00:00:00 2001
From: NFish <douxc512@gmail.com>
Date: Sat, 12 Oct 2024 23:49:18 +0800
Subject: [PATCH 223/235] Feat/new account page (#9236)

---
 web/app/account/account-page/index.module.css |   9 +
 web/app/account/account-page/index.tsx        | 304 ++++++++++++++++++
 web/app/account/avatar.tsx                    |  94 ++++++
 web/app/account/header.tsx                    |  37 +++
 web/app/account/layout.tsx                    |  40 +++
 web/app/account/page.tsx                      |   7 +
 .../header/account-dropdown/index.tsx         |  11 +-
 .../header/account-setting/index.tsx          |  22 +-
 web/app/components/header/header-wrapper.tsx  |   2 +-
 web/i18n/en-US/common.ts                      |   5 +-
 web/i18n/zh-Hans/common.ts                    |   6 +-
 11 files changed, 511 insertions(+), 26 deletions(-)
 create mode 100644 web/app/account/account-page/index.module.css
 create mode 100644 web/app/account/account-page/index.tsx
 create mode 100644 web/app/account/avatar.tsx
 create mode 100644 web/app/account/header.tsx
 create mode 100644 web/app/account/layout.tsx
 create mode 100644 web/app/account/page.tsx

diff --git a/web/app/account/account-page/index.module.css b/web/app/account/account-page/index.module.css
new file mode 100644
index 0000000000000..949d1257e9820
--- /dev/null
+++ b/web/app/account/account-page/index.module.css
@@ -0,0 +1,9 @@
+.modal {
+  padding: 24px 32px !important;
+  width: 400px !important;
+}
+
+.bg {
+  background: linear-gradient(180deg, rgba(217, 45, 32, 0.05) 0%, rgba(217, 45, 32, 0.00) 24.02%), #F9FAFB;
+}
+
diff --git a/web/app/account/account-page/index.tsx b/web/app/account/account-page/index.tsx
new file mode 100644
index 0000000000000..53f7692e6c8f5
--- /dev/null
+++ b/web/app/account/account-page/index.tsx
@@ -0,0 +1,304 @@
+'use client'
+import { useState } from 'react'
+import { useTranslation } from 'react-i18next'
+
+import { useContext } from 'use-context-selector'
+import s from './index.module.css'
+import Collapse from '@/app/components/header/account-setting/collapse'
+import type { IItem } from '@/app/components/header/account-setting/collapse'
+import Modal from '@/app/components/base/modal'
+import Confirm from '@/app/components/base/confirm'
+import Button from '@/app/components/base/button'
+import { updateUserProfile } from '@/service/common'
+import { useAppContext } from '@/context/app-context'
+import { ToastContext } from '@/app/components/base/toast'
+import AppIcon from '@/app/components/base/app-icon'
+import Avatar from '@/app/components/base/avatar'
+import { IS_CE_EDITION } from '@/config'
+
+const titleClassName = `
+  text-sm font-medium text-gray-900
+`
+const descriptionClassName = `
+  mt-1 text-xs font-normal text-gray-500
+`
+const inputClassName = `
+  mt-2 w-full px-3 py-2 bg-gray-100 rounded
+  text-sm font-normal text-gray-800
+`
+
+const validPassword = /^(?=.*[a-zA-Z])(?=.*\d).{8,}$/
+
+export default function AccountPage() {
+  const { t } = useTranslation()
+  const { mutateUserProfile, userProfile, apps } = useAppContext()
+  const { notify } = useContext(ToastContext)
+  const [editNameModalVisible, setEditNameModalVisible] = useState(false)
+  const [editName, setEditName] = useState('')
+  const [editing, setEditing] = useState(false)
+  const [editPasswordModalVisible, setEditPasswordModalVisible] = useState(false)
+  const [currentPassword, setCurrentPassword] = useState('')
+  const [password, setPassword] = useState('')
+  const [confirmPassword, setConfirmPassword] = useState('')
+  const [showDeleteAccountModal, setShowDeleteAccountModal] = useState(false)
+
+  const handleEditName = () => {
+    setEditNameModalVisible(true)
+    setEditName(userProfile.name)
+  }
+  const handleSaveName = async () => {
+    try {
+      setEditing(true)
+      await updateUserProfile({ url: 'account/name', body: { name: editName } })
+      notify({ type: 'success', message: t('common.actionMsg.modifiedSuccessfully') })
+      mutateUserProfile()
+      setEditNameModalVisible(false)
+      setEditing(false)
+    }
+    catch (e) {
+      notify({ type: 'error', message: (e as Error).message })
+      setEditNameModalVisible(false)
+      setEditing(false)
+    }
+  }
+
+  const showErrorMessage = (message: string) => {
+    notify({
+      type: 'error',
+      message,
+    })
+  }
+  const valid = () => {
+    if (!password.trim()) {
+      showErrorMessage(t('login.error.passwordEmpty'))
+      return false
+    }
+    if (!validPassword.test(password)) {
+      showErrorMessage(t('login.error.passwordInvalid'))
+      return false
+    }
+    if (password !== confirmPassword) {
+      showErrorMessage(t('common.account.notEqual'))
+      return false
+    }
+
+    return true
+  }
+  const resetPasswordForm = () => {
+    setCurrentPassword('')
+    setPassword('')
+    setConfirmPassword('')
+  }
+  const handleSavePassword = async () => {
+    if (!valid())
+      return
+    try {
+      setEditing(true)
+      await updateUserProfile({
+        url: 'account/password',
+        body: {
+          password: currentPassword,
+          new_password: password,
+          repeat_new_password: confirmPassword,
+        },
+      })
+      notify({ type: 'success', message: t('common.actionMsg.modifiedSuccessfully') })
+      mutateUserProfile()
+      setEditPasswordModalVisible(false)
+      resetPasswordForm()
+      setEditing(false)
+    }
+    catch (e) {
+      notify({ type: 'error', message: (e as Error).message })
+      setEditPasswordModalVisible(false)
+      setEditing(false)
+    }
+  }
+
+  const renderAppItem = (item: IItem) => {
+    return (
+      <div className='flex px-3 py-1'>
+        <div className='mr-3'>
+          <AppIcon size='tiny' />
+        </div>
+        <div className='mt-[3px] text-xs font-medium text-gray-700 leading-[18px]'>{item.name}</div>
+      </div>
+    )
+  }
+
+  return (
+    <>
+      <div className='pt-2 pb-3'>
+        <h4 className='title-2xl-semi-bold text-primary'>{t('common.account.myAccount')}</h4>
+      </div>
+      <div className='mb-8 p-6 rounded-xl flex items-center bg-gradient-to-r from-background-gradient-bg-fill-chat-bg-2 to-background-gradient-bg-fill-chat-bg-1'>
+        <Avatar name={userProfile.name} size={64} />
+        <div className='ml-4'>
+          <p className='system-xl-semibold text-text-primary'>{userProfile.name}</p>
+          <p className='system-xs-regular text-text-tertiary'>{userProfile.email}</p>
+        </div>
+      </div>
+      <div className='mb-8'>
+        <div className={titleClassName}>{t('common.account.name')}</div>
+        <div className='flex items-center justify-between gap-2 w-full mt-2'>
+          <div className='flex-1 bg-gray-100 rounded-md p-2 system-sm-regular text-components-input-text-filled '>
+            <span className='pl-1'>{userProfile.name}</span>
+          </div>
+          <div className=' bg-gray-100 rounded-md py-2 px-3 cursor-pointer system-sm-medium text-components-input-text-filled' onClick={handleEditName}>
+            {t('common.operation.edit')}
+          </div>
+        </div>
+      </div>
+      <div className='mb-8'>
+        <div className={titleClassName}>{t('common.account.email')}</div>
+        <div className='flex items-center justify-between gap-2 w-full mt-2'>
+          <div className='flex-1 bg-gray-100 rounded-md p-2 system-sm-regular text-components-input-text-filled '>
+            <span className='pl-1'>{userProfile.email}</span>
+          </div>
+        </div>
+      </div>
+      {
+        IS_CE_EDITION && (
+          <div className='mb-8 flex justify-between'>
+            <div>
+              <div className='mb-1 text-sm font-medium text-gray-900'>{t('common.account.password')}</div>
+              <div className='mb-2 text-xs text-gray-500'>{t('common.account.passwordTip')}</div>
+            </div>
+            <Button onClick={() => setEditPasswordModalVisible(true)}>{userProfile.is_password_set ? t('common.account.resetPassword') : t('common.account.setPassword')}</Button>
+          </div>
+        )
+      }
+      <div className='mb-6 border-[0.5px] border-gray-100' />
+      <div className='mb-8'>
+        <div className={titleClassName}>{t('common.account.langGeniusAccount')}</div>
+        <div className={descriptionClassName}>{t('common.account.langGeniusAccountTip')}</div>
+        {!!apps.length && (
+          <Collapse
+            title={`${t('common.account.showAppLength', { length: apps.length })}`}
+            items={apps.map(app => ({ key: app.id, name: app.name }))}
+            renderItem={renderAppItem}
+            wrapperClassName='mt-2'
+          />
+        )}
+        {!IS_CE_EDITION && <Button className='mt-2 text-[#D92D20]' onClick={() => setShowDeleteAccountModal(true)}>{t('common.account.delete')}</Button>}
+      </div>
+      {
+        editNameModalVisible && (
+          <Modal
+            isShow
+            onClose={() => setEditNameModalVisible(false)}
+            className={s.modal}
+          >
+            <div className='mb-6 text-lg font-medium text-gray-900'>{t('common.account.editName')}</div>
+            <div className={titleClassName}>{t('common.account.name')}</div>
+            <input
+              className={inputClassName}
+              value={editName}
+              onChange={e => setEditName(e.target.value)}
+            />
+            <div className='flex justify-end mt-10'>
+              <Button className='mr-2' onClick={() => setEditNameModalVisible(false)}>{t('common.operation.cancel')}</Button>
+              <Button
+                disabled={editing || !editName}
+                variant='primary'
+                onClick={handleSaveName}
+              >
+                {t('common.operation.save')}
+              </Button>
+            </div>
+          </Modal>
+        )
+      }
+      {
+        editPasswordModalVisible && (
+          <Modal
+            isShow
+            onClose={() => {
+              setEditPasswordModalVisible(false)
+              resetPasswordForm()
+            }}
+            className={s.modal}
+          >
+            <div className='mb-6 text-lg font-medium text-gray-900'>{userProfile.is_password_set ? t('common.account.resetPassword') : t('common.account.setPassword')}</div>
+            {userProfile.is_password_set && (
+              <>
+                <div className={titleClassName}>{t('common.account.currentPassword')}</div>
+                <input
+                  type="password"
+                  className={inputClassName}
+                  value={currentPassword}
+                  onChange={e => setCurrentPassword(e.target.value)}
+                />
+              </>
+            )}
+            <div className='mt-8 text-sm font-medium text-gray-900'>
+              {userProfile.is_password_set ? t('common.account.newPassword') : t('common.account.password')}
+            </div>
+            <input
+              type="password"
+              className={inputClassName}
+              value={password}
+              onChange={e => setPassword(e.target.value)}
+            />
+            <div className='mt-8 text-sm font-medium text-gray-900'>{t('common.account.confirmPassword')}</div>
+            <input
+              type="password"
+              className={inputClassName}
+              value={confirmPassword}
+              onChange={e => setConfirmPassword(e.target.value)}
+            />
+            <div className='flex justify-end mt-10'>
+              <Button className='mr-2' onClick={() => {
+                setEditPasswordModalVisible(false)
+                resetPasswordForm()
+              }}>{t('common.operation.cancel')}</Button>
+              <Button
+                disabled={editing}
+                variant='primary'
+                onClick={handleSavePassword}
+              >
+                {userProfile.is_password_set ? t('common.operation.reset') : t('common.operation.save')}
+              </Button>
+            </div>
+          </Modal>
+        )
+      }
+      {
+        showDeleteAccountModal && (
+          <Confirm
+            isShow
+            onCancel={() => setShowDeleteAccountModal(false)}
+            onConfirm={() => setShowDeleteAccountModal(false)}
+            showCancel={false}
+            type='warning'
+            title={t('common.account.delete')}
+            content={
+              <>
+                <div className='my-1 text-[#D92D20] text-sm leading-5'>
+                  {t('common.account.deleteTip')}
+                </div>
+                <div className='mt-3 text-sm leading-5'>
+                  <span>{t('common.account.deleteConfirmTip')}</span>
+                  <a
+                    className='text-primary-600 cursor'
+                    href={`mailto:support@dify.ai?subject=Delete Account Request&body=Delete Account: ${userProfile.email}`}
+                    target='_blank'
+                    rel='noreferrer noopener'
+                    onClick={(e) => {
+                      e.preventDefault()
+                      window.location.href = e.currentTarget.href
+                    }}
+                  >
+                    support@dify.ai
+                  </a>
+                </div>
+                <div className='my-2 px-3 py-2 rounded-lg bg-gray-100 text-sm font-medium leading-5 text-gray-800'>{`${t('common.account.delete')}: ${userProfile.email}`}</div>
+              </>
+            }
+            confirmText={t('common.operation.ok') as string}
+          />
+        )
+      }
+    </>
+  )
+}
diff --git a/web/app/account/avatar.tsx b/web/app/account/avatar.tsx
new file mode 100644
index 0000000000000..29bd0cb5a581c
--- /dev/null
+++ b/web/app/account/avatar.tsx
@@ -0,0 +1,94 @@
+'use client'
+import { useTranslation } from 'react-i18next'
+import { Fragment } from 'react'
+import { useRouter } from 'next/navigation'
+import { Menu, Transition } from '@headlessui/react'
+import Avatar from '@/app/components/base/avatar'
+import { logout } from '@/service/common'
+import { useAppContext } from '@/context/app-context'
+import { LogOut01 } from '@/app/components/base/icons/src/vender/line/general'
+
+export type IAppSelector = {
+  isMobile: boolean
+}
+
+export default function AppSelector() {
+  const router = useRouter()
+  const { t } = useTranslation()
+  const { userProfile } = useAppContext()
+
+  const handleLogout = async () => {
+    await logout({
+      url: '/logout',
+      params: {},
+    })
+
+    if (localStorage?.getItem('console_token'))
+      localStorage.removeItem('console_token')
+
+    router.push('/signin')
+  }
+
+  return (
+    <Menu as="div" className="relative inline-block text-left">
+      {
+        ({ open }) => (
+          <>
+            <div>
+              <Menu.Button
+                className={`
+                    inline-flex items-center
+                    rounded-[20px] p-1x text-sm
+                  text-gray-700 hover:bg-gray-200
+                    mobile:px-1
+                    ${open && 'bg-gray-200'}
+                  `}
+              >
+                <Avatar name={userProfile.name} size={32} />
+              </Menu.Button>
+            </div>
+            <Transition
+              as={Fragment}
+              enter="transition ease-out duration-100"
+              enterFrom="transform opacity-0 scale-95"
+              enterTo="transform opacity-100 scale-100"
+              leave="transition ease-in duration-75"
+              leaveFrom="transform opacity-100 scale-100"
+              leaveTo="transform opacity-0 scale-95"
+            >
+              <Menu.Items
+                className="
+                    absolute -right-3 -top-3 w-60 max-w-80
+                    divide-y divide-gray-100 origin-top-right rounded-lg bg-white
+                    shadow-lg
+                  "
+              >
+                <Menu.Item>
+                  <div className='p-1'>
+                    <div className='flex flex-nowrap items-center px-3 py-2'>
+                      <div className='grow'>
+                        <div className='system-md-medium text-text-primary break-all'>{userProfile.name}</div>
+                        <div className='system-xs-regular text-text-tertiary break-all'>{userProfile.email}</div>
+                      </div>
+                      <Avatar name={userProfile.name} size={32} />
+                    </div>
+                  </div>
+                </Menu.Item>
+                <Menu.Item>
+                  <div className='p-1' onClick={() => handleLogout()}>
+                    <div
+                      className='flex items-center justify-start h-9 px-3 rounded-lg cursor-pointer group hover:bg-gray-50'
+                    >
+                      <LogOut01 className='w-4 h-4 text-gray-500 flex mr-1' />
+                      <div className='font-normal text-[14px] text-gray-700'>{t('common.userProfile.logout')}</div>
+                    </div>
+                  </div>
+                </Menu.Item>
+              </Menu.Items>
+            </Transition>
+          </>
+        )
+      }
+    </Menu>
+  )
+}
diff --git a/web/app/account/header.tsx b/web/app/account/header.tsx
new file mode 100644
index 0000000000000..694533e5ab7cb
--- /dev/null
+++ b/web/app/account/header.tsx
@@ -0,0 +1,37 @@
+'use client'
+import { useTranslation } from 'react-i18next'
+import { RiArrowRightUpLine, RiRobot2Line } from '@remixicon/react'
+import { useRouter } from 'next/navigation'
+import Button from '../components/base/button'
+import Avatar from './avatar'
+import LogoSite from '@/app/components/base/logo/logo-site'
+
+const Header = () => {
+  const { t } = useTranslation()
+  const router = useRouter()
+
+  const back = () => {
+    router.back()
+  }
+  return (
+    <div className='flex flex-1 items-center justify-between px-4'>
+      <div className='flex items-center gap-3'>
+        <div className='flex items-center cursor-pointer' onClick={back}>
+          <LogoSite className='object-contain' />
+        </div>
+        <div className='w-[1px] h-4 bg-divider-regular' />
+        <p className='text-text-primary title-3xl-semi-bold'>{t('common.account.account')}</p>
+      </div>
+      <div className='flex items-center flex-shrink-0 gap-3'>
+        <Button className='gap-2 py-2 px-3 system-sm-medium' onClick={back}>
+          <RiRobot2Line className='w-4 h-4' />
+          <p>{t('common.account.studio')}</p>
+          <RiArrowRightUpLine className='w-4 h-4' />
+        </Button>
+        <div className='w-[1px] h-4 bg-divider-regular' />
+        <Avatar />
+      </div>
+    </div>
+  )
+}
+export default Header
diff --git a/web/app/account/layout.tsx b/web/app/account/layout.tsx
new file mode 100644
index 0000000000000..5aa8b05cbfd07
--- /dev/null
+++ b/web/app/account/layout.tsx
@@ -0,0 +1,40 @@
+import React from 'react'
+import type { ReactNode } from 'react'
+import Header from './header'
+import SwrInitor from '@/app/components/swr-initor'
+import { AppContextProvider } from '@/context/app-context'
+import GA, { GaType } from '@/app/components/base/ga'
+import HeaderWrapper from '@/app/components/header/header-wrapper'
+import { EventEmitterContextProvider } from '@/context/event-emitter'
+import { ProviderContextProvider } from '@/context/provider-context'
+import { ModalContextProvider } from '@/context/modal-context'
+
+const Layout = ({ children }: { children: ReactNode }) => {
+  return (
+    <>
+      <GA gaType={GaType.admin} />
+      <SwrInitor>
+        <AppContextProvider>
+          <EventEmitterContextProvider>
+            <ProviderContextProvider>
+              <ModalContextProvider>
+                <HeaderWrapper>
+                  <Header />
+                </HeaderWrapper>
+                <div className='relative flex flex-col overflow-y-auto bg-white shrink-0 h-0 grow'>
+                  {children}
+                </div>
+              </ModalContextProvider>
+            </ProviderContextProvider>
+          </EventEmitterContextProvider>
+        </AppContextProvider>
+      </SwrInitor>
+    </>
+  )
+}
+
+export const metadata = {
+  title: 'Dify',
+}
+
+export default Layout
diff --git a/web/app/account/page.tsx b/web/app/account/page.tsx
new file mode 100644
index 0000000000000..bb7e7f7feb184
--- /dev/null
+++ b/web/app/account/page.tsx
@@ -0,0 +1,7 @@
+import AccountPage from './account-page'
+
+export default function Account() {
+  return <div className='max-w-[640px] w-full mx-auto pt-12 px-6'>
+    <AccountPage />
+  </div>
+}
diff --git a/web/app/components/header/account-dropdown/index.tsx b/web/app/components/header/account-dropdown/index.tsx
index 03157ed7cb4e9..712906ebae381 100644
--- a/web/app/components/header/account-dropdown/index.tsx
+++ b/web/app/components/header/account-dropdown/index.tsx
@@ -107,7 +107,16 @@ export default function AppSelector({ isMobile }: IAppSelector) {
                   </div>
                   <div className="px-1 py-1">
                     <Menu.Item>
-                      <div className={itemClassName} onClick={() => setShowAccountSettingModal({ payload: 'account' })}>
+                      <Link
+                        className={classNames(itemClassName, 'group justify-between')}
+                        href='/account'
+                        target='_self' rel='noopener noreferrer'>
+                        <div>{t('common.account.account')}</div>
+                        <ArrowUpRight className='hidden w-[14px] h-[14px] text-gray-500 group-hover:flex' />
+                      </Link>
+                    </Menu.Item>
+                    <Menu.Item>
+                      <div className={itemClassName} onClick={() => setShowAccountSettingModal({ payload: 'members' })}>
                         <div>{t('common.userProfile.settings')}</div>
                       </div>
                     </Menu.Item>
diff --git a/web/app/components/header/account-setting/index.tsx b/web/app/components/header/account-setting/index.tsx
index 253b9f1b4c2cd..d829f6b77b0cc 100644
--- a/web/app/components/header/account-setting/index.tsx
+++ b/web/app/components/header/account-setting/index.tsx
@@ -2,10 +2,6 @@
 import { useTranslation } from 'react-i18next'
 import { useEffect, useRef, useState } from 'react'
 import {
-  RiAccountCircleFill,
-  RiAccountCircleLine,
-  RiApps2AddFill,
-  RiApps2AddLine,
   RiBox3Fill,
   RiBox3Line,
   RiCloseLine,
@@ -21,9 +17,7 @@ import {
   RiPuzzle2Line,
   RiTranslate2,
 } from '@remixicon/react'
-import AccountPage from './account-page'
 import MembersPage from './members-page'
-import IntegrationsPage from './Integrations-page'
 import LanguagePage from './language-page'
 import ApiBasedExtensionPage from './api-based-extension-page'
 import DataSourcePage from './data-source-page'
@@ -60,7 +54,7 @@ type GroupItem = {
 
 export default function AccountSetting({
   onCancel,
-  activeTab = 'account',
+  activeTab = 'members',
 }: IAccountSettingProps) {
   const [activeMenu, setActiveMenu] = useState(activeTab)
   const { t } = useTranslation()
@@ -125,18 +119,6 @@ export default function AccountSetting({
       key: 'account-group',
       name: t('common.settings.accountGroup'),
       items: [
-        {
-          key: 'account',
-          name: t('common.settings.account'),
-          icon: <RiAccountCircleLine className={iconClassName} />,
-          activeIcon: <RiAccountCircleFill className={iconClassName} />,
-        },
-        {
-          key: 'integrations',
-          name: t('common.settings.integrations'),
-          icon: <RiApps2AddLine className={iconClassName} />,
-          activeIcon: <RiApps2AddFill className={iconClassName} />,
-        },
         {
           key: 'language',
           name: t('common.settings.language'),
@@ -217,10 +199,8 @@ export default function AccountSetting({
             </div>
           </div>
           <div className='px-4 sm:px-8 pt-2'>
-            {activeMenu === 'account' && <AccountPage />}
             {activeMenu === 'members' && <MembersPage />}
             {activeMenu === 'billing' && <BillingPage />}
-            {activeMenu === 'integrations' && <IntegrationsPage />}
             {activeMenu === 'language' && <LanguagePage />}
             {activeMenu === 'provider' && <ModelProviderPage />}
             {activeMenu === 'data-source' && <DataSourcePage />}
diff --git a/web/app/components/header/header-wrapper.tsx b/web/app/components/header/header-wrapper.tsx
index 205a379a90301..360cf8e5607ba 100644
--- a/web/app/components/header/header-wrapper.tsx
+++ b/web/app/components/header/header-wrapper.tsx
@@ -11,7 +11,7 @@ const HeaderWrapper = ({
   children,
 }: HeaderWrapperProps) => {
   const pathname = usePathname()
-  const isBordered = ['/apps', '/datasets', '/datasets/create', '/tools'].includes(pathname)
+  const isBordered = ['/apps', '/datasets', '/datasets/create', '/tools', '/account'].includes(pathname)
 
   return (
     <div className={classNames(
diff --git a/web/i18n/en-US/common.ts b/web/i18n/en-US/common.ts
index 23e301485adfb..c348a580ebcdf 100644
--- a/web/i18n/en-US/common.ts
+++ b/web/i18n/en-US/common.ts
@@ -140,7 +140,7 @@ const translation = {
     logout: 'Log out',
   },
   settings: {
-    accountGroup: 'ACCOUNT',
+    accountGroup: 'GENERAL',
     workplaceGroup: 'WORKSPACE',
     account: 'My account',
     members: 'Members',
@@ -153,6 +153,9 @@ const translation = {
     apiBasedExtension: 'API Extension',
   },
   account: {
+    account: 'Account',
+    myAccount: 'My Account',
+    studio: 'Dify Studio',
     avatar: 'Avatar',
     name: 'Name',
     email: 'Email',
diff --git a/web/i18n/zh-Hans/common.ts b/web/i18n/zh-Hans/common.ts
index 7947d32f251f1..fe2b3bf92d0b2 100644
--- a/web/i18n/zh-Hans/common.ts
+++ b/web/i18n/zh-Hans/common.ts
@@ -140,9 +140,8 @@ const translation = {
     logout: '登出',
   },
   settings: {
-    accountGroup: '账户',
+    accountGroup: '通用',
     workplaceGroup: '工作空间',
-    account: '我的账户',
     members: '成员',
     billing: '账单',
     integrations: '集成',
@@ -153,6 +152,9 @@ const translation = {
     apiBasedExtension: 'API 扩展',
   },
   account: {
+    account: '账户',
+    myAccount: '我的账户',
+    studio: 'Dify 工作室',
     avatar: '头像',
     name: '用户名',
     email: '邮箱',

From d97d3ff5fc43a066c605a04c4ac5d522155527dd Mon Sep 17 00:00:00 2001
From: zhuhao <37029601+hwzhuhao@users.noreply.github.com>
Date: Sat, 12 Oct 2024 23:58:41 +0800
Subject: [PATCH 224/235] chore: add abstract decorator and output log when
 query embedding fails (#9264)

---
 api/core/embedding/cached_embedding.py                     | 7 ++++++-
 api/core/rag/datasource/keyword/keyword_base.py            | 2 ++
 .../datasource/vdb/elasticsearch/elasticsearch_vector.py   | 2 +-
 3 files changed, 9 insertions(+), 2 deletions(-)

diff --git a/api/core/embedding/cached_embedding.py b/api/core/embedding/cached_embedding.py
index 75219051cd30c..31d2171e7267d 100644
--- a/api/core/embedding/cached_embedding.py
+++ b/api/core/embedding/cached_embedding.py
@@ -5,6 +5,7 @@
 import numpy as np
 from sqlalchemy.exc import IntegrityError
 
+from configs import dify_config
 from core.embedding.embedding_constant import EmbeddingInputType
 from core.model_manager import ModelInstance
 from core.model_runtime.entities.model_entities import ModelPropertyKey
@@ -110,6 +111,8 @@ def embed_query(self, text: str) -> list[float]:
             embedding_results = embedding_result.embeddings[0]
             embedding_results = (embedding_results / np.linalg.norm(embedding_results)).tolist()
         except Exception as ex:
+            if dify_config.DEBUG:
+                logging.exception(f"Failed to embed query text: {ex}")
             raise ex
 
         try:
@@ -122,6 +125,8 @@ def embed_query(self, text: str) -> list[float]:
             encoded_str = encoded_vector.decode("utf-8")
             redis_client.setex(embedding_cache_key, 600, encoded_str)
         except Exception as ex:
-            logging.exception("Failed to add embedding to redis %s", ex)
+            if dify_config.DEBUG:
+                logging.exception("Failed to add embedding to redis %s", ex)
+            raise ex
 
         return embedding_results
diff --git a/api/core/rag/datasource/keyword/keyword_base.py b/api/core/rag/datasource/keyword/keyword_base.py
index 4b9ec460e61fe..be00687abd502 100644
--- a/api/core/rag/datasource/keyword/keyword_base.py
+++ b/api/core/rag/datasource/keyword/keyword_base.py
@@ -27,9 +27,11 @@ def text_exists(self, id: str) -> bool:
     def delete_by_ids(self, ids: list[str]) -> None:
         raise NotImplementedError
 
+    @abstractmethod
     def delete(self) -> None:
         raise NotImplementedError
 
+    @abstractmethod
     def search(self, query: str, **kwargs: Any) -> list[Document]:
         raise NotImplementedError
 
diff --git a/api/core/rag/datasource/vdb/elasticsearch/elasticsearch_vector.py b/api/core/rag/datasource/vdb/elasticsearch/elasticsearch_vector.py
index f585e12b2e99c..66bc31a4bfaff 100644
--- a/api/core/rag/datasource/vdb/elasticsearch/elasticsearch_vector.py
+++ b/api/core/rag/datasource/vdb/elasticsearch/elasticsearch_vector.py
@@ -77,7 +77,7 @@ def _check_version(self):
             raise ValueError("Elasticsearch vector database version must be greater than 8.0.0")
 
     def get_type(self) -> str:
-        return "elasticsearch"
+        return VectorType.ELASTICSEARCH
 
     def add_texts(self, documents: list[Document], embeddings: list[list[float]], **kwargs):
         uuids = self._get_uuids(documents)

From 9275760599abf6df762b6f8babf7eba54554119d Mon Sep 17 00:00:00 2001
From: zhuhao <37029601+hwzhuhao@users.noreply.github.com>
Date: Sun, 13 Oct 2024 09:44:53 +0800
Subject: [PATCH 225/235] chore: add baidu-obs and supabase for .env.example
 (#9289)

---
 docker/.env.example | 20 +++++++++++++++++++-
 1 file changed, 19 insertions(+), 1 deletion(-)

diff --git a/docker/.env.example b/docker/.env.example
index c7b4f38d2ec11..969deadf67921 100644
--- a/docker/.env.example
+++ b/docker/.env.example
@@ -264,7 +264,7 @@ CONSOLE_CORS_ALLOW_ORIGINS=*
 # ------------------------------
 
 # The type of storage to use for storing user files.
-# Supported values are `local` and `s3` and `azure-blob` and `google-storage` and `tencent-cos` and `huawei-obs`
+# Supported values are `local` , `s3` , `azure-blob` , `google-storage`, `tencent-cos`, `huawei-obs`, `volcengine-tos`, `baidu-obs`, `supabase`
 # Default: `local`
 STORAGE_TYPE=local
 
@@ -344,6 +344,24 @@ VOLCENGINE_TOS_ENDPOINT=your-server-url
 # The region of the Volcengine TOS service.
 VOLCENGINE_TOS_REGION=your-region
 
+# Baidu OBS Storage Configuration
+# The name of the Baidu OBS bucket to use for storing files.
+BAIDU_OBS_BUCKET_NAME=your-bucket-name
+# The secret key to use for authenticating with the Baidu OBS service.
+BAIDU_OBS_SECRET_KEY=your-secret-key
+# The access key to use for authenticating with the Baidu OBS service.
+BAIDU_OBS_ACCESS_KEY=your-access-key
+# The endpoint of the Baidu OBS service.
+BAIDU_OBS_ENDPOINT=your-server-url
+
+# Supabase Storage Configuration
+# The name of the Supabase bucket to use for storing files.
+SUPABASE_BUCKET_NAME=your-bucket-name
+# The api key to use for authenticating with the Supabase service.
+SUPABASE_API_KEY=your-access-key
+# The project endpoint url of the Supabase service.
+SUPABASE_URL=your-server-url
+
 # ------------------------------
 # Vector Database Configuration
 # ------------------------------

From 1ec83e4969a2a78cd5a2e563a3cc8815328562ef Mon Sep 17 00:00:00 2001
From: "github-actions[bot]"
 <41898282+github-actions[bot]@users.noreply.github.com>
Date: Sun, 13 Oct 2024 14:56:26 +0800
Subject: [PATCH 226/235] chore: translate i18n files (#9288)

Co-authored-by: douxc <7553076+douxc@users.noreply.github.com>
Co-authored-by: crazywoola <427733928@qq.com>
---
 web/i18n/de-DE/common.ts   | 3 +++
 web/i18n/es-ES/common.ts   | 3 +++
 web/i18n/fa-IR/common.ts   | 3 +++
 web/i18n/fr-FR/common.ts   | 3 +++
 web/i18n/hi-IN/common.ts   | 3 +++
 web/i18n/it-IT/common.ts   | 3 +++
 web/i18n/ja-JP/common.ts   | 3 +++
 web/i18n/ko-KR/common.ts   | 3 +++
 web/i18n/pl-PL/common.ts   | 3 +++
 web/i18n/pt-BR/common.ts   | 3 +++
 web/i18n/ro-RO/common.ts   | 3 +++
 web/i18n/ru-RU/common.ts   | 3 +++
 web/i18n/tr-TR/common.ts   | 3 +++
 web/i18n/uk-UA/common.ts   | 3 +++
 web/i18n/vi-VN/common.ts   | 3 +++
 web/i18n/zh-Hans/common.ts | 1 +
 web/i18n/zh-Hant/common.ts | 3 +++
 17 files changed, 49 insertions(+)

diff --git a/web/i18n/de-DE/common.ts b/web/i18n/de-DE/common.ts
index 6ea06bc8b1c72..8b221ca3bbbf4 100644
--- a/web/i18n/de-DE/common.ts
+++ b/web/i18n/de-DE/common.ts
@@ -167,6 +167,9 @@ const translation = {
     delete: 'Konto löschen',
     deleteTip: 'Wenn Sie Ihr Konto löschen, werden alle Ihre Daten dauerhaft gelöscht und können nicht wiederhergestellt werden.',
     deleteConfirmTip: 'Zur Bestätigung senden Sie bitte Folgendes von Ihrer registrierten E-Mail-Adresse an ',
+    myAccount: 'Mein Konto',
+    studio: 'Dify Studio',
+    account: 'Konto',
   },
   members: {
     team: 'Team',
diff --git a/web/i18n/es-ES/common.ts b/web/i18n/es-ES/common.ts
index 59a05f63d8a3e..748c9d152d4c1 100644
--- a/web/i18n/es-ES/common.ts
+++ b/web/i18n/es-ES/common.ts
@@ -171,6 +171,9 @@ const translation = {
     delete: 'Eliminar cuenta',
     deleteTip: 'Eliminar tu cuenta borrará permanentemente todos tus datos y no se podrán recuperar.',
     deleteConfirmTip: 'Para confirmar, por favor envía lo siguiente desde tu correo electrónico registrado a ',
+    account: 'Cuenta',
+    myAccount: 'Mi Cuenta',
+    studio: 'Estudio Dify',
   },
   members: {
     team: 'Equipo',
diff --git a/web/i18n/fa-IR/common.ts b/web/i18n/fa-IR/common.ts
index c75ab11a63cd8..a369a0ba5ebb1 100644
--- a/web/i18n/fa-IR/common.ts
+++ b/web/i18n/fa-IR/common.ts
@@ -171,6 +171,9 @@ const translation = {
     delete: 'حذف حساب کاربری',
     deleteTip: 'حذف حساب کاربری شما تمام داده‌های شما را به طور دائمی پاک می‌کند و قابل بازیابی نیست.',
     deleteConfirmTip: 'برای تأیید، لطفاً موارد زیر را از ایمیل ثبت‌نام شده خود به این آدرس ارسال کنید ',
+    account: 'حساب',
+    myAccount: 'حساب من',
+    studio: 'استودیو Dify',
   },
   members: {
     team: 'تیم',
diff --git a/web/i18n/fr-FR/common.ts b/web/i18n/fr-FR/common.ts
index c4fed4405dc4f..0cd301aed2c5e 100644
--- a/web/i18n/fr-FR/common.ts
+++ b/web/i18n/fr-FR/common.ts
@@ -167,6 +167,9 @@ const translation = {
     delete: 'Supprimer le compte',
     deleteTip: 'La suppression de votre compte effacera définitivement toutes vos données et elles ne pourront pas être récupérées.',
     deleteConfirmTip: 'Pour confirmer, veuillez envoyer ce qui suit depuis votre adresse e-mail enregistrée à ',
+    myAccount: 'Mon compte',
+    account: 'Compte',
+    studio: 'Dify Studio',
   },
   members: {
     team: 'Équipe',
diff --git a/web/i18n/hi-IN/common.ts b/web/i18n/hi-IN/common.ts
index 256cb9d426f23..224090437e449 100644
--- a/web/i18n/hi-IN/common.ts
+++ b/web/i18n/hi-IN/common.ts
@@ -177,6 +177,9 @@ const translation = {
     deleteConfirmTip: 'पुष्टि करने के लिए, कृपया अपने पंजीकृत ईमेल से निम्नलिखित भेजें',
     delete: 'खाता हटाएं',
     deleteTip: 'अपना खाता हटाने से आपका सारा डेटा स्थायी रूप से मिट जाएगा और इसे पुनर्प्राप्त नहीं किया जा सकता है।',
+    account: 'खाता',
+    studio: 'डिफाई स्टूडियो',
+    myAccount: 'मेरा खाता',
   },
   members: {
     team: 'टीम',
diff --git a/web/i18n/it-IT/common.ts b/web/i18n/it-IT/common.ts
index aa675bb471829..5c180a8b69710 100644
--- a/web/i18n/it-IT/common.ts
+++ b/web/i18n/it-IT/common.ts
@@ -179,6 +179,9 @@ const translation = {
       'Eliminando il tuo account cancellerai permanentemente tutti i tuoi dati e non sarà possibile recuperarli.',
     deleteConfirmTip:
       'Per confermare, invia il seguente messaggio dalla tua email registrata a ',
+    myAccount: 'Il mio account',
+    account: 'Conto',
+    studio: 'Dify Studio',
   },
   members: {
     team: 'Team',
diff --git a/web/i18n/ja-JP/common.ts b/web/i18n/ja-JP/common.ts
index e2517a619dc97..bd50e68b0949e 100644
--- a/web/i18n/ja-JP/common.ts
+++ b/web/i18n/ja-JP/common.ts
@@ -171,6 +171,9 @@ const translation = {
     delete: 'アカウントを削除',
     deleteTip: 'アカウントを削除すると、すべてのデータが完全に消去され、復元できなくなります。',
     deleteConfirmTip: '確認のため、登録したメールから次の内容をに送信してください ',
+    account: 'アカウント',
+    myAccount: 'マイアカウント',
+    studio: 'Difyスタジオ',
   },
   members: {
     team: 'チーム',
diff --git a/web/i18n/ko-KR/common.ts b/web/i18n/ko-KR/common.ts
index 8ef55da3f7153..d70b7ebb108d4 100644
--- a/web/i18n/ko-KR/common.ts
+++ b/web/i18n/ko-KR/common.ts
@@ -163,6 +163,9 @@ const translation = {
     delete: '계정 삭제',
     deleteTip: '계정을 삭제하면 모든 데이터가 영구적으로 지워지며 복구할 수 없습니다.',
     deleteConfirmTip: '확인하려면 등록된 이메일에서 다음 내용을 로 보내주세요 ',
+    myAccount: '내 계정',
+    studio: '디파이 스튜디오',
+    account: '계좌',
   },
   members: {
     team: '팀',
diff --git a/web/i18n/pl-PL/common.ts b/web/i18n/pl-PL/common.ts
index 91f5fb2899296..b0706787855d3 100644
--- a/web/i18n/pl-PL/common.ts
+++ b/web/i18n/pl-PL/common.ts
@@ -173,6 +173,9 @@ const translation = {
     delete: 'Usuń konto',
     deleteTip: 'Usunięcie konta spowoduje trwałe usunięcie wszystkich danych i nie będzie można ich odzyskać.',
     deleteConfirmTip: 'Aby potwierdzić, wyślij następujące informacje z zarejestrowanego adresu e-mail na adres ',
+    myAccount: 'Moje konto',
+    studio: 'Dify Studio',
+    account: 'Rachunek',
   },
   members: {
     team: 'Zespół',
diff --git a/web/i18n/pt-BR/common.ts b/web/i18n/pt-BR/common.ts
index f9e9eb78889e1..9343fdf5605e7 100644
--- a/web/i18n/pt-BR/common.ts
+++ b/web/i18n/pt-BR/common.ts
@@ -167,6 +167,9 @@ const translation = {
     delete: 'Excluir conta',
     deleteTip: 'Excluir sua conta apagará permanentemente todos os seus dados e eles não poderão ser recuperados.',
     deleteConfirmTip: 'Para confirmar, envie o seguinte do seu e-mail registrado para ',
+    myAccount: 'Minha Conta',
+    account: 'Conta',
+    studio: 'Estúdio Dify',
   },
   members: {
     team: 'Equipe',
diff --git a/web/i18n/ro-RO/common.ts b/web/i18n/ro-RO/common.ts
index 1fd8778106213..dc3bfcc45a845 100644
--- a/web/i18n/ro-RO/common.ts
+++ b/web/i18n/ro-RO/common.ts
@@ -167,6 +167,9 @@ const translation = {
     delete: 'Șterge contul',
     deleteTip: 'Ștergerea contului vă va șterge definitiv toate datele și nu pot fi recuperate.',
     deleteConfirmTip: 'Pentru a confirma, trimiteți următoarele din e-mailul înregistrat la ',
+    account: 'Cont',
+    studio: 'Dify Studio',
+    myAccount: 'Contul meu',
   },
   members: {
     team: 'Echipă',
diff --git a/web/i18n/ru-RU/common.ts b/web/i18n/ru-RU/common.ts
index 82e3471e607dd..a829fb27b1151 100644
--- a/web/i18n/ru-RU/common.ts
+++ b/web/i18n/ru-RU/common.ts
@@ -171,6 +171,9 @@ const translation = {
     delete: 'Удалить учетную запись',
     deleteTip: 'Удаление вашей учетной записи приведет к безвозвратному удалению всех ваших данных, и их невозможно будет восстановить.',
     deleteConfirmTip: 'Для подтверждения, пожалуйста, отправьте следующее с вашего зарегистрированного адреса электронной почты на ',
+    account: 'Счет',
+    studio: 'Студия Dify',
+    myAccount: 'Моя учетная запись',
   },
   members: {
     team: 'Команда',
diff --git a/web/i18n/tr-TR/common.ts b/web/i18n/tr-TR/common.ts
index a41925cd2002a..dc4b1cccbab5a 100644
--- a/web/i18n/tr-TR/common.ts
+++ b/web/i18n/tr-TR/common.ts
@@ -171,6 +171,9 @@ const translation = {
     delete: 'Hesabı Sil',
     deleteTip: 'Hesabınızı silmek tüm verilerinizi kalıcı olarak siler ve geri alınamaz.',
     deleteConfirmTip: 'Onaylamak için, kayıtlı e-postanızdan şu adrese e-posta gönderin: ',
+    account: 'Hesap',
+    myAccount: 'Hesabım',
+    studio: 'Dify Stüdyo',
   },
   members: {
     team: 'Takım',
diff --git a/web/i18n/uk-UA/common.ts b/web/i18n/uk-UA/common.ts
index cc70772be33fc..ef0bc55203787 100644
--- a/web/i18n/uk-UA/common.ts
+++ b/web/i18n/uk-UA/common.ts
@@ -167,6 +167,9 @@ const translation = {
     delete: 'Видалити обліковий запис',
     deleteTip: 'Видалення вашого облікового запису призведе до остаточного видалення всіх ваших даних, і їх неможливо буде відновити.',
     deleteConfirmTip: 'Щоб підтвердити, будь ласка, надішліть наступне з вашої зареєстрованої електронної пошти на ',
+    account: 'Рахунок',
+    studio: 'Студія Dify',
+    myAccount: 'Особистий кабінет',
   },
   members: {
     team: 'Команда',
diff --git a/web/i18n/vi-VN/common.ts b/web/i18n/vi-VN/common.ts
index 252fa7e1dfa95..5336ec4f667f4 100644
--- a/web/i18n/vi-VN/common.ts
+++ b/web/i18n/vi-VN/common.ts
@@ -167,6 +167,9 @@ const translation = {
     delete: 'Xóa tài khoản',
     deleteTip: 'Xóa tài khoản của bạn sẽ xóa vĩnh viễn tất cả dữ liệu của bạn và không thể khôi phục được.',
     deleteConfirmTip: 'Để xác nhận, vui lòng gửi thông tin sau từ email đã đăng ký của bạn tới ',
+    studio: 'Dify Studio',
+    myAccount: 'Tài khoản của tôi',
+    account: 'Tài khoản',
   },
   members: {
     team: 'Nhóm',
diff --git a/web/i18n/zh-Hans/common.ts b/web/i18n/zh-Hans/common.ts
index fe2b3bf92d0b2..3c9b61f56b0e1 100644
--- a/web/i18n/zh-Hans/common.ts
+++ b/web/i18n/zh-Hans/common.ts
@@ -142,6 +142,7 @@ const translation = {
   settings: {
     accountGroup: '通用',
     workplaceGroup: '工作空间',
+    account: '我的账户',
     members: '成员',
     billing: '账单',
     integrations: '集成',
diff --git a/web/i18n/zh-Hant/common.ts b/web/i18n/zh-Hant/common.ts
index 8cd51b1991f1a..e43b49bd3c618 100644
--- a/web/i18n/zh-Hant/common.ts
+++ b/web/i18n/zh-Hant/common.ts
@@ -167,6 +167,9 @@ const translation = {
     delete: '刪除帳戶',
     deleteTip: '刪除您的帳戶將永久刪除您的所有資料並且無法恢復。',
     deleteConfirmTip: '請將以下內容從您的註冊電子郵件發送至 ',
+    account: '帳戶',
+    myAccount: '我的帳戶',
+    studio: 'Dify 工作室',
   },
   members: {
     team: '團隊',

From d15ba3939d39c14fd52cd2566bc3bf04cdf5f1f2 Mon Sep 17 00:00:00 2001
From: ice yao <yao3690093@gmail.com>
Date: Sun, 13 Oct 2024 21:26:05 +0800
Subject: [PATCH 227/235] Add Volcengine VikingDB as new vector provider
 (#9287)

---
 api/.env.example                              |  11 +-
 api/configs/middleware/__init__.py            |   2 +
 api/configs/middleware/vdb/vikingdb_config.py |  37 +++
 api/controllers/console/datasets/datasets.py  |   2 +
 api/core/rag/datasource/vdb/vector_factory.py |   4 +
 api/core/rag/datasource/vdb/vector_type.py    |   1 +
 .../rag/datasource/vdb/vikingdb/__init__.py   |   0
 .../vdb/vikingdb/vikingdb_vector.py           | 239 ++++++++++++++++++
 api/poetry.lock                               |  73 +++++-
 api/pyproject.toml                            |   1 +
 .../integration_tests/vdb/__mock/vikingdb.py  | 215 ++++++++++++++++
 .../vdb/vikingdb/__init__.py                  |   0
 .../vdb/vikingdb/test_vikingdb.py             |  37 +++
 dev/pytest/pytest_vdb.sh                      |   3 +-
 docker/docker-compose.yaml                    |   5 +
 15 files changed, 627 insertions(+), 3 deletions(-)
 create mode 100644 api/configs/middleware/vdb/vikingdb_config.py
 create mode 100644 api/core/rag/datasource/vdb/vikingdb/__init__.py
 create mode 100644 api/core/rag/datasource/vdb/vikingdb/vikingdb_vector.py
 create mode 100644 api/tests/integration_tests/vdb/__mock/vikingdb.py
 create mode 100644 api/tests/integration_tests/vdb/vikingdb/__init__.py
 create mode 100644 api/tests/integration_tests/vdb/vikingdb/test_vikingdb.py

diff --git a/api/.env.example b/api/.env.example
index 468130b1628e9..aa155003abd3f 100644
--- a/api/.env.example
+++ b/api/.env.example
@@ -111,7 +111,7 @@ SUPABASE_URL=your-server-url
 WEB_API_CORS_ALLOW_ORIGINS=http://127.0.0.1:3000,*
 CONSOLE_CORS_ALLOW_ORIGINS=http://127.0.0.1:3000,*
 
-# Vector database configuration, support: weaviate, qdrant, milvus, myscale, relyt, pgvecto_rs, pgvector, pgvector, chroma, opensearch, tidb_vector
+# Vector database configuration, support: weaviate, qdrant, milvus, myscale, relyt, pgvecto_rs, pgvector, pgvector, chroma, opensearch, tidb_vector, vikingdb
 VECTOR_STORE=weaviate
 
 # Weaviate configuration
@@ -220,6 +220,15 @@ BAIDU_VECTOR_DB_DATABASE=dify
 BAIDU_VECTOR_DB_SHARD=1
 BAIDU_VECTOR_DB_REPLICAS=3
 
+# ViKingDB configuration
+VIKINGDB_ACCESS_KEY=your-ak
+VIKINGDB_SECRET_KEY=your-sk
+VIKINGDB_REGION=cn-shanghai
+VIKINGDB_HOST=api-vikingdb.xxx.volces.com
+VIKINGDB_SCHEMA=http
+VIKINGDB_CONNECTION_TIMEOUT=30
+VIKINGDB_SOCKET_TIMEOUT=30
+
 # Upload configuration
 UPLOAD_FILE_SIZE_LIMIT=15
 UPLOAD_FILE_BATCH_LIMIT=5
diff --git a/api/configs/middleware/__init__.py b/api/configs/middleware/__init__.py
index 25f3df6dde41d..fa7f41d630965 100644
--- a/api/configs/middleware/__init__.py
+++ b/api/configs/middleware/__init__.py
@@ -28,6 +28,7 @@
 from configs.middleware.vdb.relyt_config import RelytConfig
 from configs.middleware.vdb.tencent_vector_config import TencentVectorDBConfig
 from configs.middleware.vdb.tidb_vector_config import TiDBVectorConfig
+from configs.middleware.vdb.vikingdb_config import VikingDBConfig
 from configs.middleware.vdb.weaviate_config import WeaviateConfig
 
 
@@ -243,5 +244,6 @@ class MiddlewareConfig(
     WeaviateConfig,
     ElasticsearchConfig,
     InternalTestConfig,
+    VikingDBConfig,
 ):
     pass
diff --git a/api/configs/middleware/vdb/vikingdb_config.py b/api/configs/middleware/vdb/vikingdb_config.py
new file mode 100644
index 0000000000000..5ad98d898a16e
--- /dev/null
+++ b/api/configs/middleware/vdb/vikingdb_config.py
@@ -0,0 +1,37 @@
+from typing import Optional
+
+from pydantic import BaseModel, Field
+
+
+class VikingDBConfig(BaseModel):
+    """
+    Configuration for connecting to Volcengine VikingDB.
+    Refer to the following documentation for details on obtaining credentials:
+    https://www.volcengine.com/docs/6291/65568
+    """
+
+    VIKINGDB_ACCESS_KEY: Optional[str] = Field(
+        default=None, description="The Access Key provided by Volcengine VikingDB for API authentication."
+    )
+    VIKINGDB_SECRET_KEY: Optional[str] = Field(
+        default=None, description="The Secret Key provided by Volcengine VikingDB for API authentication."
+    )
+    VIKINGDB_REGION: Optional[str] = Field(
+        default="cn-shanghai",
+        description="The region of the Volcengine VikingDB service.(e.g., 'cn-shanghai', 'cn-beijing').",
+    )
+    VIKINGDB_HOST: Optional[str] = Field(
+        default="api-vikingdb.mlp.cn-shanghai.volces.com",
+        description="The host of the Volcengine VikingDB service.(e.g., 'api-vikingdb.volces.com', \
+            'api-vikingdb.mlp.cn-shanghai.volces.com')",
+    )
+    VIKINGDB_SCHEME: Optional[str] = Field(
+        default="http",
+        description="The scheme of the Volcengine VikingDB service.(e.g., 'http', 'https').",
+    )
+    VIKINGDB_CONNECTION_TIMEOUT: Optional[int] = Field(
+        default=30, description="The connection timeout of the Volcengine VikingDB service."
+    )
+    VIKINGDB_SOCKET_TIMEOUT: Optional[int] = Field(
+        default=30, description="The socket timeout of the Volcengine VikingDB service."
+    )
diff --git a/api/controllers/console/datasets/datasets.py b/api/controllers/console/datasets/datasets.py
index 102089bf071ac..6583356d23c9e 100644
--- a/api/controllers/console/datasets/datasets.py
+++ b/api/controllers/console/datasets/datasets.py
@@ -618,6 +618,7 @@ def get(self):
                 | VectorType.TENCENT
                 | VectorType.PGVECTO_RS
                 | VectorType.BAIDU
+                | VectorType.VIKINGDB
             ):
                 return {"retrieval_method": [RetrievalMethod.SEMANTIC_SEARCH.value]}
             case (
@@ -655,6 +656,7 @@ def get(self, vector_type):
                 | VectorType.TENCENT
                 | VectorType.PGVECTO_RS
                 | VectorType.BAIDU
+                | VectorType.VIKINGDB
             ):
                 return {"retrieval_method": [RetrievalMethod.SEMANTIC_SEARCH.value]}
             case (
diff --git a/api/core/rag/datasource/vdb/vector_factory.py b/api/core/rag/datasource/vdb/vector_factory.py
index 1f4a4d44a23ee..873b2890277d6 100644
--- a/api/core/rag/datasource/vdb/vector_factory.py
+++ b/api/core/rag/datasource/vdb/vector_factory.py
@@ -107,6 +107,10 @@ def get_vector_factory(vector_type: str) -> type[AbstractVectorFactory]:
                 from core.rag.datasource.vdb.baidu.baidu_vector import BaiduVectorFactory
 
                 return BaiduVectorFactory
+            case VectorType.VIKINGDB:
+                from core.rag.datasource.vdb.vikingdb.vikingdb_vector import VikingDBVectorFactory
+
+                return VikingDBVectorFactory
             case _:
                 raise ValueError(f"Vector store {vector_type} is not supported.")
 
diff --git a/api/core/rag/datasource/vdb/vector_type.py b/api/core/rag/datasource/vdb/vector_type.py
index 996ff48615c90..b4d604a080899 100644
--- a/api/core/rag/datasource/vdb/vector_type.py
+++ b/api/core/rag/datasource/vdb/vector_type.py
@@ -17,3 +17,4 @@ class VectorType(str, Enum):
     ORACLE = "oracle"
     ELASTICSEARCH = "elasticsearch"
     BAIDU = "baidu"
+    VIKINGDB = "vikingdb"
diff --git a/api/core/rag/datasource/vdb/vikingdb/__init__.py b/api/core/rag/datasource/vdb/vikingdb/__init__.py
new file mode 100644
index 0000000000000..e69de29bb2d1d
diff --git a/api/core/rag/datasource/vdb/vikingdb/vikingdb_vector.py b/api/core/rag/datasource/vdb/vikingdb/vikingdb_vector.py
new file mode 100644
index 0000000000000..22d0e92586fcb
--- /dev/null
+++ b/api/core/rag/datasource/vdb/vikingdb/vikingdb_vector.py
@@ -0,0 +1,239 @@
+import json
+from typing import Any
+
+from pydantic import BaseModel
+from volcengine.viking_db import (
+    Data,
+    DistanceType,
+    Field,
+    FieldType,
+    IndexType,
+    QuantType,
+    VectorIndexParams,
+    VikingDBService,
+)
+
+from configs import dify_config
+from core.rag.datasource.entity.embedding import Embeddings
+from core.rag.datasource.vdb.field import Field as vdb_Field
+from core.rag.datasource.vdb.vector_base import BaseVector
+from core.rag.datasource.vdb.vector_factory import AbstractVectorFactory
+from core.rag.datasource.vdb.vector_type import VectorType
+from core.rag.models.document import Document
+from extensions.ext_redis import redis_client
+from models.dataset import Dataset
+
+
+class VikingDBConfig(BaseModel):
+    access_key: str
+    secret_key: str
+    host: str
+    region: str
+    scheme: str
+    connection_timeout: int
+    socket_timeout: int
+    index_type: str = IndexType.HNSW
+    distance: str = DistanceType.L2
+    quant: str = QuantType.Float
+
+
+class VikingDBVector(BaseVector):
+    def __init__(self, collection_name: str, group_id: str, config: VikingDBConfig):
+        super().__init__(collection_name)
+        self._group_id = group_id
+        self._client_config = config
+        self._index_name = f"{self._collection_name}_idx"
+        self._client = VikingDBService(
+            host=config.host,
+            region=config.region,
+            scheme=config.scheme,
+            connection_timeout=config.connection_timeout,
+            socket_timeout=config.socket_timeout,
+            ak=config.access_key,
+            sk=config.secret_key,
+        )
+
+    def _has_collection(self) -> bool:
+        try:
+            self._client.get_collection(self._collection_name)
+        except Exception:
+            return False
+        return True
+
+    def _has_index(self) -> bool:
+        try:
+            self._client.get_index(self._collection_name, self._index_name)
+        except Exception:
+            return False
+        return True
+
+    def _create_collection(self, dimension: int):
+        lock_name = f"vector_indexing_lock_{self._collection_name}"
+        with redis_client.lock(lock_name, timeout=20):
+            collection_exist_cache_key = f"vector_indexing_{self._collection_name}"
+            if redis_client.get(collection_exist_cache_key):
+                return
+
+            if not self._has_collection():
+                fields = [
+                    Field(field_name=vdb_Field.PRIMARY_KEY.value, field_type=FieldType.String, is_primary_key=True),
+                    Field(field_name=vdb_Field.METADATA_KEY.value, field_type=FieldType.String),
+                    Field(field_name=vdb_Field.GROUP_KEY.value, field_type=FieldType.String),
+                    Field(field_name=vdb_Field.CONTENT_KEY.value, field_type=FieldType.Text),
+                    Field(field_name=vdb_Field.VECTOR.value, field_type=FieldType.Vector, dim=dimension),
+                ]
+
+                self._client.create_collection(
+                    collection_name=self._collection_name,
+                    fields=fields,
+                    description="Collection For Dify",
+                )
+
+            if not self._has_index():
+                vector_index = VectorIndexParams(
+                    distance=self._client_config.distance,
+                    index_type=self._client_config.index_type,
+                    quant=self._client_config.quant,
+                )
+
+                self._client.create_index(
+                    collection_name=self._collection_name,
+                    index_name=self._index_name,
+                    vector_index=vector_index,
+                    partition_by=vdb_Field.GROUP_KEY.value,
+                    description="Index For Dify",
+                )
+            redis_client.set(collection_exist_cache_key, 1, ex=3600)
+
+    def get_type(self) -> str:
+        return VectorType.VIKINGDB
+
+    def create(self, texts: list[Document], embeddings: list[list[float]], **kwargs):
+        dimension = len(embeddings[0])
+        self._create_collection(dimension)
+        self.add_texts(texts, embeddings, **kwargs)
+
+    def add_texts(self, documents: list[Document], embeddings: list[list[float]], **kwargs):
+        page_contents = [doc.page_content for doc in documents]
+        metadatas = [doc.metadata for doc in documents]
+        docs = []
+
+        for i, page_content in enumerate(page_contents):
+            metadata = {}
+            if metadatas is not None:
+                for key, val in metadatas[i].items():
+                    metadata[key] = val
+            doc = Data(
+                {
+                    vdb_Field.PRIMARY_KEY.value: metadatas[i]["doc_id"],
+                    vdb_Field.VECTOR.value: embeddings[i] if embeddings else None,
+                    vdb_Field.CONTENT_KEY.value: page_content,
+                    vdb_Field.METADATA_KEY.value: json.dumps(metadata),
+                    vdb_Field.GROUP_KEY.value: self._group_id,
+                }
+            )
+            docs.append(doc)
+
+        self._client.get_collection(self._collection_name).upsert_data(docs)
+
+    def text_exists(self, id: str) -> bool:
+        docs = self._client.get_collection(self._collection_name).fetch_data(id)
+        not_exists_str = "data does not exist"
+        if docs is not None and not_exists_str not in docs.fields.get("message", ""):
+            return True
+        return False
+
+    def delete_by_ids(self, ids: list[str]) -> None:
+        self._client.get_collection(self._collection_name).delete_data(ids)
+
+    def get_ids_by_metadata_field(self, key: str, value: str):
+        # Note: Metadata field value is an dict, but vikingdb field
+        # not support json type
+        results = self._client.get_index(self._collection_name, self._index_name).search(
+            filter={"op": "must", "field": vdb_Field.GROUP_KEY.value, "conds": [self._group_id]},
+            # max value is 5000
+            limit=5000,
+        )
+
+        if not results:
+            return []
+
+        ids = []
+        for result in results:
+            metadata = result.fields.get(vdb_Field.METADATA_KEY.value)
+            if metadata is not None:
+                metadata = json.loads(metadata)
+                if metadata.get(key) == value:
+                    ids.append(result.id)
+        return ids
+
+    def delete_by_metadata_field(self, key: str, value: str) -> None:
+        ids = self.get_ids_by_metadata_field(key, value)
+        self.delete_by_ids(ids)
+
+    def search_by_vector(self, query_vector: list[float], **kwargs: Any) -> list[Document]:
+        results = self._client.get_index(self._collection_name, self._index_name).search_by_vector(
+            query_vector, limit=kwargs.get("top_k", 50)
+        )
+        score_threshold = float(kwargs.get("score_threshold") or 0.0)
+        return self._get_search_res(results, score_threshold)
+
+    def _get_search_res(self, results, score_threshold):
+        if len(results) == 0:
+            return []
+
+        docs = []
+        for result in results:
+            metadata = result.fields.get(vdb_Field.METADATA_KEY.value)
+            if metadata is not None:
+                metadata = json.loads(metadata)
+            if result.score > score_threshold:
+                metadata["score"] = result.score
+                doc = Document(page_content=result.fields.get(vdb_Field.CONTENT_KEY.value), metadata=metadata)
+                docs.append(doc)
+        docs = sorted(docs, key=lambda x: x.metadata["score"], reverse=True)
+        return docs
+
+    def search_by_full_text(self, query: str, **kwargs: Any) -> list[Document]:
+        return []
+
+    def delete(self) -> None:
+        if self._has_index():
+            self._client.drop_index(self._collection_name, self._index_name)
+        if self._has_collection():
+            self._client.drop_collection(self._collection_name)
+
+
+class VikingDBVectorFactory(AbstractVectorFactory):
+    def init_vector(self, dataset: Dataset, attributes: list, embeddings: Embeddings) -> VikingDBVector:
+        if dataset.index_struct_dict:
+            class_prefix: str = dataset.index_struct_dict["vector_store"]["class_prefix"]
+            collection_name = class_prefix.lower()
+        else:
+            dataset_id = dataset.id
+            collection_name = Dataset.gen_collection_name_by_id(dataset_id).lower()
+            dataset.index_struct = json.dumps(self.gen_index_struct_dict(VectorType.VIKINGDB, collection_name))
+
+        if dify_config.VIKINGDB_ACCESS_KEY is None:
+            raise ValueError("VIKINGDB_ACCESS_KEY should not be None")
+        if dify_config.VIKINGDB_SECRET_KEY is None:
+            raise ValueError("VIKINGDB_SECRET_KEY should not be None")
+        if dify_config.VIKINGDB_HOST is None:
+            raise ValueError("VIKINGDB_HOST should not be None")
+        if dify_config.VIKINGDB_REGION is None:
+            raise ValueError("VIKINGDB_REGION should not be None")
+        if dify_config.VIKINGDB_SCHEME is None:
+            raise ValueError("VIKINGDB_SCHEME should not be None")
+        return VikingDBVector(
+            collection_name=collection_name,
+            group_id=dataset.id,
+            config=VikingDBConfig(
+                access_key=dify_config.VIKINGDB_ACCESS_KEY,
+                secret_key=dify_config.VIKINGDB_SECRET_KEY,
+                host=dify_config.VIKINGDB_HOST,
+                region=dify_config.VIKINGDB_REGION,
+                scheme=dify_config.VIKINGDB_SCHEME,
+                connection_timeout=dify_config.VIKINGDB_CONNECTION_TIMEOUT,
+                socket_timeout=dify_config.VIKINGDB_SOCKET_TIMEOUT,
+            ),
+        )
diff --git a/api/poetry.lock b/api/poetry.lock
index 6565db27ad572..efefedfb21cd2 100644
--- a/api/poetry.lock
+++ b/api/poetry.lock
@@ -2038,6 +2038,17 @@ packaging = ">=17.0"
 pandas = ">=0.24.2"
 pyarrow = ">=3.0.0"
 
+[[package]]
+name = "decorator"
+version = "5.1.1"
+description = "Decorators for Humans"
+optional = false
+python-versions = ">=3.5"
+files = [
+    {file = "decorator-5.1.1-py3-none-any.whl", hash = "sha256:b8c3f85900b9dc423225913c5aace94729fe1fa9763b38939a95226f02d37186"},
+    {file = "decorator-5.1.1.tar.gz", hash = "sha256:637996211036b6385ef91435e4fae22989472f9d571faba8927ba8253acbc330"},
+]
+
 [[package]]
 name = "defusedxml"
 version = "0.7.1"
@@ -3027,6 +3038,20 @@ files = [
 docs = ["sphinx (>=4)", "sphinx-rtd-theme (>=1)"]
 tests = ["cython", "hypothesis", "mpmath", "pytest", "setuptools"]
 
+[[package]]
+name = "google"
+version = "3.0.0"
+description = "Python bindings to the Google search engine."
+optional = false
+python-versions = "*"
+files = [
+    {file = "google-3.0.0-py2.py3-none-any.whl", hash = "sha256:889cf695f84e4ae2c55fbc0cfdaf4c1e729417fa52ab1db0485202ba173e4935"},
+    {file = "google-3.0.0.tar.gz", hash = "sha256:143530122ee5130509ad5e989f0512f7cb218b2d4eddbafbad40fd10e8d8ccbe"},
+]
+
+[package.dependencies]
+beautifulsoup4 = "*"
+
 [[package]]
 name = "google-ai-generativelanguage"
 version = "0.6.9"
@@ -6670,6 +6695,17 @@ files = [
     {file = "psycopg2_binary-2.9.9-cp39-cp39-win_amd64.whl", hash = "sha256:f7ae5d65ccfbebdfa761585228eb4d0df3a8b15cfb53bd953e713e09fbb12957"},
 ]
 
+[[package]]
+name = "py"
+version = "1.11.0"
+description = "library with cross-python path, ini-parsing, io, code, log facilities"
+optional = false
+python-versions = ">=2.7, !=3.0.*, !=3.1.*, !=3.2.*, !=3.3.*, !=3.4.*"
+files = [
+    {file = "py-1.11.0-py2.py3-none-any.whl", hash = "sha256:607c53218732647dff4acdfcd50cb62615cedf612e72d1724fb1a0cc6405b378"},
+    {file = "py-1.11.0.tar.gz", hash = "sha256:51c75c4126074b472f746a24399ad32f6053d1b34b68d2fa41e558e6f4a98719"},
+]
+
 [[package]]
 name = "py-cpuinfo"
 version = "9.0.0"
@@ -8012,6 +8048,21 @@ files = [
 [package.dependencies]
 requests = "2.31.0"
 
+[[package]]
+name = "retry"
+version = "0.9.2"
+description = "Easy to use retry decorator."
+optional = false
+python-versions = "*"
+files = [
+    {file = "retry-0.9.2-py2.py3-none-any.whl", hash = "sha256:ccddf89761fa2c726ab29391837d4327f819ea14d244c232a1d24c67a2f98606"},
+    {file = "retry-0.9.2.tar.gz", hash = "sha256:f8bfa8b99b69c4506d6f5bd3b0aabf77f98cdb17f3c9fc3f5ca820033336fba4"},
+]
+
+[package.dependencies]
+decorator = ">=3.4.2"
+py = ">=1.4.26,<2.0.0"
+
 [[package]]
 name = "rich"
 version = "13.9.2"
@@ -9829,6 +9880,26 @@ files = [
     {file = "vine-5.1.0.tar.gz", hash = "sha256:8b62e981d35c41049211cf62a0a1242d8c1ee9bd15bb196ce38aefd6799e61e0"},
 ]
 
+[[package]]
+name = "volcengine-compat"
+version = "1.0.156"
+description = "Be Compatible with the Volcengine SDK for Python, The version of package dependencies has been modified. like pycryptodome, pytz."
+optional = false
+python-versions = "*"
+files = [
+    {file = "volcengine_compat-1.0.156-py3-none-any.whl", hash = "sha256:4abc149a7601ebad8fa2d28fab50c7945145cf74daecb71bca797b0bdc82c5a5"},
+    {file = "volcengine_compat-1.0.156.tar.gz", hash = "sha256:e357d096828e31a202dc6047bbc5bf6fff3f54a98cd35a99ab5f965ea741a267"},
+]
+
+[package.dependencies]
+google = ">=3.0.0"
+protobuf = ">=3.18.3"
+pycryptodome = ">=3.9.9"
+pytz = ">=2020.5"
+requests = ">=2.25.1"
+retry = ">=0.9.2"
+six = ">=1.0"
+
 [[package]]
 name = "volcengine-python-sdk"
 version = "1.0.103"
@@ -10636,4 +10707,4 @@ cffi = ["cffi (>=1.11)"]
 [metadata]
 lock-version = "2.0"
 python-versions = ">=3.10,<3.13"
-content-hash = "375ac3a91760513924647e67376cb6018505ec61d967651b254c68af9808d774"
+content-hash = "edb5e3b0d50e84a239224cc77f3f615fdbdd6b504bce5b1075b29363f3054957"
diff --git a/api/pyproject.toml b/api/pyproject.toml
index 594517771b34f..dff74750f0f55 100644
--- a/api/pyproject.toml
+++ b/api/pyproject.toml
@@ -246,6 +246,7 @@ pymochow = "1.3.1"
 qdrant-client = "1.7.3"
 tcvectordb = "1.3.2"
 tidb-vector = "0.0.9"
+volcengine-compat = "~1.0.156"
 weaviate-client = "~3.21.0"
 
 ############################################################
diff --git a/api/tests/integration_tests/vdb/__mock/vikingdb.py b/api/tests/integration_tests/vdb/__mock/vikingdb.py
new file mode 100644
index 0000000000000..0f40337feba6e
--- /dev/null
+++ b/api/tests/integration_tests/vdb/__mock/vikingdb.py
@@ -0,0 +1,215 @@
+import os
+from typing import Union
+from unittest.mock import MagicMock
+
+import pytest
+from _pytest.monkeypatch import MonkeyPatch
+from volcengine.viking_db import (
+    Collection,
+    Data,
+    DistanceType,
+    Field,
+    FieldType,
+    Index,
+    IndexType,
+    QuantType,
+    VectorIndexParams,
+    VikingDBService,
+)
+
+from core.rag.datasource.vdb.field import Field as vdb_Field
+
+
+class MockVikingDBClass:
+    def __init__(
+        self,
+        host="api-vikingdb.volces.com",
+        region="cn-north-1",
+        ak="",
+        sk="",
+        scheme="http",
+        connection_timeout=30,
+        socket_timeout=30,
+        proxy=None,
+    ):
+        self._viking_db_service = MagicMock()
+        self._viking_db_service.get_exception = MagicMock(return_value='{"data": {"primary_key": "test_id"}}')
+
+    def get_collection(self, collection_name) -> Collection:
+        return Collection(
+            collection_name=collection_name,
+            description="Collection For Dify",
+            viking_db_service=self._viking_db_service,
+            primary_key=vdb_Field.PRIMARY_KEY.value,
+            fields=[
+                Field(field_name=vdb_Field.PRIMARY_KEY.value, field_type=FieldType.String, is_primary_key=True),
+                Field(field_name=vdb_Field.METADATA_KEY.value, field_type=FieldType.String),
+                Field(field_name=vdb_Field.GROUP_KEY.value, field_type=FieldType.String),
+                Field(field_name=vdb_Field.CONTENT_KEY.value, field_type=FieldType.Text),
+                Field(field_name=vdb_Field.VECTOR.value, field_type=FieldType.Vector, dim=768),
+            ],
+            indexes=[
+                Index(
+                    collection_name=collection_name,
+                    index_name=f"{collection_name}_idx",
+                    vector_index=VectorIndexParams(
+                        distance=DistanceType.L2,
+                        index_type=IndexType.HNSW,
+                        quant=QuantType.Float,
+                    ),
+                    scalar_index=None,
+                    stat=None,
+                    viking_db_service=self._viking_db_service,
+                )
+            ],
+        )
+
+    def drop_collection(self, collection_name):
+        assert collection_name != ""
+
+    def create_collection(self, collection_name, fields, description="") -> Collection:
+        return Collection(
+            collection_name=collection_name,
+            description=description,
+            primary_key=vdb_Field.PRIMARY_KEY.value,
+            viking_db_service=self._viking_db_service,
+            fields=fields,
+        )
+
+    def get_index(self, collection_name, index_name) -> Index:
+        return Index(
+            collection_name=collection_name,
+            index_name=index_name,
+            viking_db_service=self._viking_db_service,
+            stat=None,
+            scalar_index=None,
+            vector_index=VectorIndexParams(
+                distance=DistanceType.L2,
+                index_type=IndexType.HNSW,
+                quant=QuantType.Float,
+            ),
+        )
+
+    def create_index(
+        self,
+        collection_name,
+        index_name,
+        vector_index=None,
+        cpu_quota=2,
+        description="",
+        partition_by="",
+        scalar_index=None,
+        shard_count=None,
+        shard_policy=None,
+    ):
+        return Index(
+            collection_name=collection_name,
+            index_name=index_name,
+            vector_index=vector_index,
+            cpu_quota=cpu_quota,
+            description=description,
+            partition_by=partition_by,
+            scalar_index=scalar_index,
+            shard_count=shard_count,
+            shard_policy=shard_policy,
+            viking_db_service=self._viking_db_service,
+            stat=None,
+        )
+
+    def drop_index(self, collection_name, index_name):
+        assert collection_name != ""
+        assert index_name != ""
+
+    def upsert_data(self, data: Union[Data, list[Data]]):
+        assert data is not None
+
+    def fetch_data(self, id: Union[str, list[str], int, list[int]]):
+        return Data(
+            fields={
+                vdb_Field.GROUP_KEY.value: "test_group",
+                vdb_Field.METADATA_KEY.value: "{}",
+                vdb_Field.CONTENT_KEY.value: "content",
+                vdb_Field.PRIMARY_KEY.value: id,
+                vdb_Field.VECTOR.value: [-0.00762577411336441, -0.01949881482151406, 0.008832383941428398],
+            },
+            id=id,
+        )
+
+    def delete_data(self, id: Union[str, list[str], int, list[int]]):
+        assert id is not None
+
+    def search_by_vector(
+        self,
+        vector,
+        sparse_vectors=None,
+        filter=None,
+        limit=10,
+        output_fields=None,
+        partition="default",
+        dense_weight=None,
+    ) -> list[Data]:
+        return [
+            Data(
+                fields={
+                    vdb_Field.GROUP_KEY.value: "test_group",
+                    vdb_Field.METADATA_KEY.value: '\
+                    {"source": "/var/folders/ml/xxx/xxx.txt", \
+                    "document_id": "test_document_id", \
+                    "dataset_id": "test_dataset_id", \
+                    "doc_id": "test_id", \
+                    "doc_hash": "test_hash"}',
+                    vdb_Field.CONTENT_KEY.value: "content",
+                    vdb_Field.PRIMARY_KEY.value: "test_id",
+                    vdb_Field.VECTOR.value: vector,
+                },
+                id="test_id",
+                score=0.10,
+            )
+        ]
+
+    def search(
+        self, order=None, filter=None, limit=10, output_fields=None, partition="default", dense_weight=None
+    ) -> list[Data]:
+        return [
+            Data(
+                fields={
+                    vdb_Field.GROUP_KEY.value: "test_group",
+                    vdb_Field.METADATA_KEY.value: '\
+                    {"source": "/var/folders/ml/xxx/xxx.txt", \
+                    "document_id": "test_document_id", \
+                    "dataset_id": "test_dataset_id", \
+                    "doc_id": "test_id", \
+                    "doc_hash": "test_hash"}',
+                    vdb_Field.CONTENT_KEY.value: "content",
+                    vdb_Field.PRIMARY_KEY.value: "test_id",
+                    vdb_Field.VECTOR.value: [-0.00762577411336441, -0.01949881482151406, 0.008832383941428398],
+                },
+                id="test_id",
+                score=0.10,
+            )
+        ]
+
+
+MOCK = os.getenv("MOCK_SWITCH", "false").lower() == "true"
+
+
+@pytest.fixture
+def setup_vikingdb_mock(monkeypatch: MonkeyPatch):
+    if MOCK:
+        monkeypatch.setattr(VikingDBService, "__init__", MockVikingDBClass.__init__)
+        monkeypatch.setattr(VikingDBService, "get_collection", MockVikingDBClass.get_collection)
+        monkeypatch.setattr(VikingDBService, "create_collection", MockVikingDBClass.create_collection)
+        monkeypatch.setattr(VikingDBService, "drop_collection", MockVikingDBClass.drop_collection)
+        monkeypatch.setattr(VikingDBService, "get_index", MockVikingDBClass.get_index)
+        monkeypatch.setattr(VikingDBService, "create_index", MockVikingDBClass.create_index)
+        monkeypatch.setattr(VikingDBService, "drop_index", MockVikingDBClass.drop_index)
+        monkeypatch.setattr(Collection, "upsert_data", MockVikingDBClass.upsert_data)
+        monkeypatch.setattr(Collection, "fetch_data", MockVikingDBClass.fetch_data)
+        monkeypatch.setattr(Collection, "delete_data", MockVikingDBClass.delete_data)
+        monkeypatch.setattr(Index, "search_by_vector", MockVikingDBClass.search_by_vector)
+        monkeypatch.setattr(Index, "search", MockVikingDBClass.search)
+
+    yield
+
+    if MOCK:
+        monkeypatch.undo()
diff --git a/api/tests/integration_tests/vdb/vikingdb/__init__.py b/api/tests/integration_tests/vdb/vikingdb/__init__.py
new file mode 100644
index 0000000000000..e69de29bb2d1d
diff --git a/api/tests/integration_tests/vdb/vikingdb/test_vikingdb.py b/api/tests/integration_tests/vdb/vikingdb/test_vikingdb.py
new file mode 100644
index 0000000000000..2572012ea03aa
--- /dev/null
+++ b/api/tests/integration_tests/vdb/vikingdb/test_vikingdb.py
@@ -0,0 +1,37 @@
+from core.rag.datasource.vdb.vikingdb.vikingdb_vector import VikingDBConfig, VikingDBVector
+from tests.integration_tests.vdb.__mock.vikingdb import setup_vikingdb_mock
+from tests.integration_tests.vdb.test_vector_store import AbstractVectorTest, get_example_text, setup_mock_redis
+
+
+class VikingDBVectorTest(AbstractVectorTest):
+    def __init__(self):
+        super().__init__()
+        self.vector = VikingDBVector(
+            "test_collection",
+            "test_group",
+            config=VikingDBConfig(
+                access_key="test_access_key",
+                host="test_host",
+                region="test_region",
+                scheme="test_scheme",
+                secret_key="test_secret_key",
+                connection_timeout=30,
+                socket_timeout=30,
+            ),
+        )
+
+    def search_by_vector(self):
+        hits_by_vector = self.vector.search_by_vector(query_vector=self.example_embedding)
+        assert len(hits_by_vector) == 1
+
+    def search_by_full_text(self):
+        hits_by_full_text = self.vector.search_by_full_text(query=get_example_text())
+        assert len(hits_by_full_text) == 0
+
+    def get_ids_by_metadata_field(self):
+        ids = self.vector.get_ids_by_metadata_field(key="document_id", value="test_document_id")
+        assert len(ids) > 0
+
+
+def test_vikingdb_vector(setup_mock_redis, setup_vikingdb_mock):
+    VikingDBVectorTest().run_all_tests()
diff --git a/dev/pytest/pytest_vdb.sh b/dev/pytest/pytest_vdb.sh
index bad809cbfdb90..6809ef7c6f74c 100755
--- a/dev/pytest/pytest_vdb.sh
+++ b/dev/pytest/pytest_vdb.sh
@@ -7,4 +7,5 @@ pytest api/tests/integration_tests/vdb/chroma \
   api/tests/integration_tests/vdb/pgvector \
   api/tests/integration_tests/vdb/qdrant \
   api/tests/integration_tests/vdb/weaviate \
-  api/tests/integration_tests/vdb/elasticsearch
\ No newline at end of file
+  api/tests/integration_tests/vdb/elasticsearch \
+  api/tests/integration_tests/vdb/vikingdb
diff --git a/docker/docker-compose.yaml b/docker/docker-compose.yaml
index cb6ecba2791db..5db11d1961ce0 100644
--- a/docker/docker-compose.yaml
+++ b/docker/docker-compose.yaml
@@ -173,6 +173,11 @@ x-shared-env: &shared-api-worker-env
   BAIDU_VECTOR_DB_DATABASE: ${BAIDU_VECTOR_DB_DATABASE:-dify}
   BAIDU_VECTOR_DB_SHARD: ${BAIDU_VECTOR_DB_SHARD:-1}
   BAIDU_VECTOR_DB_REPLICAS: ${BAIDU_VECTOR_DB_REPLICAS:-3}
+  VIKINGDB_ACCESS_KEY: ${VIKINGDB_ACCESS_KEY:-dify}
+  VIKINGDB_SECRET_KEY: ${VIKINGDB_SECRET_KEY:-dify}
+  VIKINGDB_REGION: ${VIKINGDB_REGION:-cn-shanghai}
+  VIKINGDB_HOST: ${VIKINGDB_HOST:-api-vikingdb.xxx.volces.com}
+  VIKINGDB_SCHEMA: ${VIKINGDB_SCHEMA:-http}
   UPLOAD_FILE_SIZE_LIMIT: ${UPLOAD_FILE_SIZE_LIMIT:-15}
   UPLOAD_FILE_BATCH_LIMIT: ${UPLOAD_FILE_BATCH_LIMIT:-5}
   ETL_TYPE: ${ETL_TYPE:-dify}

From 857055b797b88fd0fc01a4f8423acd3c399454d5 Mon Sep 17 00:00:00 2001
From: kurokobo <kuro664@gmail.com>
Date: Mon, 14 Oct 2024 00:25:50 +0900
Subject: [PATCH 228/235] fix: remove the latest message from the user that
 does not have any answer yet (#9297)

---
 api/core/memory/token_buffer_memory.py | 4 ++--
 1 file changed, 2 insertions(+), 2 deletions(-)

diff --git a/api/core/memory/token_buffer_memory.py b/api/core/memory/token_buffer_memory.py
index 72da3b0c6f251..bc94912c1eb84 100644
--- a/api/core/memory/token_buffer_memory.py
+++ b/api/core/memory/token_buffer_memory.py
@@ -60,8 +60,8 @@ def get_history_prompt_messages(
         thread_messages = extract_thread_messages(messages)
 
         # for newly created message, its answer is temporarily empty, we don't need to add it to memory
-        if thread_messages and not thread_messages[-1].answer:
-            thread_messages.pop()
+        if thread_messages and not thread_messages[0].answer:
+            thread_messages.pop(0)
 
         messages = list(reversed(thread_messages))
 

From ffc3f3367058b48dab4a801d3031251339811f1f Mon Sep 17 00:00:00 2001
From: =?UTF-8?q?=E9=9D=9E=E6=B3=95=E6=93=8D=E4=BD=9C?= <hjlarry@163.com>
Date: Mon, 14 Oct 2024 10:53:45 +0800
Subject: [PATCH 229/235] chore: remove the copied zhipu_ai sdk (#9270)

---
 .../model_providers/zhipuai/llm/llm.py        |   7 +-
 .../zhipuai/text_embedding/text_embedding.py  |   3 +-
 .../zhipuai/zhipuai_sdk/__init__.py           |  15 -
 .../zhipuai/zhipuai_sdk/__version__.py        |   1 -
 .../zhipuai/zhipuai_sdk/_client.py            |  82 --
 .../zhipuai_sdk/api_resource/__init__.py      |  34 -
 .../api_resource/assistant/__init__.py        |   3 -
 .../api_resource/assistant/assistant.py       | 122 ---
 .../zhipuai_sdk/api_resource/batches.py       | 146 ---
 .../zhipuai_sdk/api_resource/chat/__init__.py |   5 -
 .../api_resource/chat/async_completions.py    | 115 ---
 .../zhipuai_sdk/api_resource/chat/chat.py     |  18 -
 .../api_resource/chat/completions.py          | 108 ---
 .../zhipuai_sdk/api_resource/embeddings.py    |  50 -
 .../zhipuai/zhipuai_sdk/api_resource/files.py | 194 ----
 .../api_resource/fine_tuning/__init__.py      |   5 -
 .../api_resource/fine_tuning/fine_tuning.py   |  18 -
 .../api_resource/fine_tuning/jobs/__init__.py |   3 -
 .../api_resource/fine_tuning/jobs/jobs.py     | 152 ---
 .../fine_tuning/models/__init__.py            |   3 -
 .../fine_tuning/models/fine_tuned_models.py   |  41 -
 .../zhipuai_sdk/api_resource/images.py        |  59 --
 .../api_resource/knowledge/__init__.py        |   3 -
 .../knowledge/document/__init__.py            |   3 -
 .../knowledge/document/document.py            | 217 -----
 .../api_resource/knowledge/knowledge.py       | 173 ----
 .../api_resource/tools/__init__.py            |   3 -
 .../zhipuai_sdk/api_resource/tools/tools.py   |  65 --
 .../api_resource/videos/__init__.py           |   7 -
 .../zhipuai_sdk/api_resource/videos/videos.py |  77 --
 .../zhipuai/zhipuai_sdk/core/__init__.py      | 108 ---
 .../zhipuai/zhipuai_sdk/core/_base_api.py     |  19 -
 .../zhipuai/zhipuai_sdk/core/_base_compat.py  | 209 ----
 .../zhipuai/zhipuai_sdk/core/_base_models.py  | 670 -------------
 .../zhipuai/zhipuai_sdk/core/_base_type.py    | 170 ----
 .../zhipuai/zhipuai_sdk/core/_constants.py    |  12 -
 .../zhipuai/zhipuai_sdk/core/_errors.py       |  86 --
 .../zhipuai/zhipuai_sdk/core/_files.py        |  75 --
 .../zhipuai/zhipuai_sdk/core/_http_client.py  | 910 ------------------
 .../zhipuai/zhipuai_sdk/core/_jwt_token.py    |  31 -
 .../core/_legacy_binary_response.py           | 207 ----
 .../zhipuai_sdk/core/_legacy_response.py      | 341 -------
 .../zhipuai/zhipuai_sdk/core/_request_opt.py  |  97 --
 .../zhipuai/zhipuai_sdk/core/_response.py     | 398 --------
 .../zhipuai/zhipuai_sdk/core/_sse_client.py   | 206 ----
 .../zhipuai_sdk/core/_utils/__init__.py       |  52 -
 .../zhipuai_sdk/core/_utils/_transform.py     | 383 --------
 .../zhipuai_sdk/core/_utils/_typing.py        | 122 ---
 .../zhipuai/zhipuai_sdk/core/_utils/_utils.py | 409 --------
 .../zhipuai/zhipuai_sdk/core/logs.py          |  78 --
 .../zhipuai/zhipuai_sdk/core/pagination.py    |  62 --
 .../zhipuai/zhipuai_sdk/types/__init__.py     |   0
 .../zhipuai_sdk/types/assistant/__init__.py   |   5 -
 .../types/assistant/assistant_completion.py   |  40 -
 .../assistant_conversation_params.py          |   7 -
 .../assistant/assistant_conversation_resp.py  |  29 -
 .../assistant/assistant_create_params.py      |  32 -
 .../types/assistant/assistant_support_resp.py |  21 -
 .../types/assistant/message/__init__.py       |   3 -
 .../assistant/message/message_content.py      |  13 -
 .../assistant/message/text_content_block.py   |  14 -
 .../tools/code_interpreter_delta_block.py     |  27 -
 .../message/tools/drawing_tool_delta_block.py |  21 -
 .../message/tools/function_delta_block.py     |  22 -
 .../message/tools/retrieval_delta_black.py    |  41 -
 .../assistant/message/tools/tools_type.py     |  16 -
 .../message/tools/web_browser_delta_block.py  |  48 -
 .../assistant/message/tools_delta_block.py    |  16 -
 .../zhipuai/zhipuai_sdk/types/batch.py        |  82 --
 .../zhipuai_sdk/types/batch_create_params.py  |  37 -
 .../zhipuai/zhipuai_sdk/types/batch_error.py  |  21 -
 .../zhipuai_sdk/types/batch_list_params.py    |  20 -
 .../zhipuai_sdk/types/batch_request_counts.py |  14 -
 .../zhipuai_sdk/types/chat/__init__.py        |   0
 .../types/chat/async_chat_completion.py       |  22 -
 .../zhipuai_sdk/types/chat/chat_completion.py |  43 -
 .../types/chat/chat_completion_chunk.py       |  57 --
 .../chat/chat_completions_create_param.py     |   8 -
 .../types/chat/code_geex/code_geex_params.py  | 146 ---
 .../zhipuai/zhipuai_sdk/types/embeddings.py   |  21 -
 .../zhipuai_sdk/types/files/__init__.py       |   5 -
 .../types/files/file_create_params.py         |  38 -
 .../zhipuai_sdk/types/files/file_deleted.py   |  13 -
 .../zhipuai_sdk/types/files/file_object.py    |  22 -
 .../zhipuai_sdk/types/files/upload_detail.py  |  13 -
 .../zhipuai_sdk/types/fine_tuning/__init__.py |   4 -
 .../types/fine_tuning/fine_tuning_job.py      |  51 -
 .../fine_tuning/fine_tuning_job_event.py      |  35 -
 .../types/fine_tuning/job_create_params.py    |  15 -
 .../types/fine_tuning/models/__init__.py      |   1 -
 .../fine_tuning/models/fine_tuned_models.py   |  13 -
 .../zhipuai/zhipuai_sdk/types/image.py        |  18 -
 .../zhipuai_sdk/types/knowledge/__init__.py   |   8 -
 .../types/knowledge/document/__init__.py      |   8 -
 .../types/knowledge/document/document.py      |  51 -
 .../document/document_edit_params.py          |  29 -
 .../document/document_list_params.py          |  26 -
 .../knowledge/document/document_list_resp.py  |  11 -
 .../zhipuai_sdk/types/knowledge/knowledge.py  |  21 -
 .../knowledge/knowledge_create_params.py      |  30 -
 .../types/knowledge/knowledge_list_params.py  |  15 -
 .../types/knowledge/knowledge_list_resp.py    |  11 -
 .../types/knowledge/knowledge_used.py         |  21 -
 .../types/sensitive_word_check/__init__.py    |   3 -
 .../sensitive_word_check.py                   |  14 -
 .../zhipuai_sdk/types/tools/__init__.py       |   9 -
 .../types/tools/tools_web_search_params.py    |  35 -
 .../zhipuai_sdk/types/tools/web_search.py     |  71 --
 .../types/tools/web_search_chunk.py           |  33 -
 .../zhipuai_sdk/types/video/__init__.py       |   3 -
 .../types/video/video_create_params.py        |  27 -
 .../zhipuai_sdk/types/video/video_object.py   |  30 -
 .../builtin/cogview/tools/cogview3.py         |   3 +-
 api/poetry.lock                               |  36 +-
 api/pyproject.toml                            |   2 +-
 115 files changed, 19 insertions(+), 7909 deletions(-)
 delete mode 100644 api/core/model_runtime/model_providers/zhipuai/zhipuai_sdk/__init__.py
 delete mode 100644 api/core/model_runtime/model_providers/zhipuai/zhipuai_sdk/__version__.py
 delete mode 100644 api/core/model_runtime/model_providers/zhipuai/zhipuai_sdk/_client.py
 delete mode 100644 api/core/model_runtime/model_providers/zhipuai/zhipuai_sdk/api_resource/__init__.py
 delete mode 100644 api/core/model_runtime/model_providers/zhipuai/zhipuai_sdk/api_resource/assistant/__init__.py
 delete mode 100644 api/core/model_runtime/model_providers/zhipuai/zhipuai_sdk/api_resource/assistant/assistant.py
 delete mode 100644 api/core/model_runtime/model_providers/zhipuai/zhipuai_sdk/api_resource/batches.py
 delete mode 100644 api/core/model_runtime/model_providers/zhipuai/zhipuai_sdk/api_resource/chat/__init__.py
 delete mode 100644 api/core/model_runtime/model_providers/zhipuai/zhipuai_sdk/api_resource/chat/async_completions.py
 delete mode 100644 api/core/model_runtime/model_providers/zhipuai/zhipuai_sdk/api_resource/chat/chat.py
 delete mode 100644 api/core/model_runtime/model_providers/zhipuai/zhipuai_sdk/api_resource/chat/completions.py
 delete mode 100644 api/core/model_runtime/model_providers/zhipuai/zhipuai_sdk/api_resource/embeddings.py
 delete mode 100644 api/core/model_runtime/model_providers/zhipuai/zhipuai_sdk/api_resource/files.py
 delete mode 100644 api/core/model_runtime/model_providers/zhipuai/zhipuai_sdk/api_resource/fine_tuning/__init__.py
 delete mode 100644 api/core/model_runtime/model_providers/zhipuai/zhipuai_sdk/api_resource/fine_tuning/fine_tuning.py
 delete mode 100644 api/core/model_runtime/model_providers/zhipuai/zhipuai_sdk/api_resource/fine_tuning/jobs/__init__.py
 delete mode 100644 api/core/model_runtime/model_providers/zhipuai/zhipuai_sdk/api_resource/fine_tuning/jobs/jobs.py
 delete mode 100644 api/core/model_runtime/model_providers/zhipuai/zhipuai_sdk/api_resource/fine_tuning/models/__init__.py
 delete mode 100644 api/core/model_runtime/model_providers/zhipuai/zhipuai_sdk/api_resource/fine_tuning/models/fine_tuned_models.py
 delete mode 100644 api/core/model_runtime/model_providers/zhipuai/zhipuai_sdk/api_resource/images.py
 delete mode 100644 api/core/model_runtime/model_providers/zhipuai/zhipuai_sdk/api_resource/knowledge/__init__.py
 delete mode 100644 api/core/model_runtime/model_providers/zhipuai/zhipuai_sdk/api_resource/knowledge/document/__init__.py
 delete mode 100644 api/core/model_runtime/model_providers/zhipuai/zhipuai_sdk/api_resource/knowledge/document/document.py
 delete mode 100644 api/core/model_runtime/model_providers/zhipuai/zhipuai_sdk/api_resource/knowledge/knowledge.py
 delete mode 100644 api/core/model_runtime/model_providers/zhipuai/zhipuai_sdk/api_resource/tools/__init__.py
 delete mode 100644 api/core/model_runtime/model_providers/zhipuai/zhipuai_sdk/api_resource/tools/tools.py
 delete mode 100644 api/core/model_runtime/model_providers/zhipuai/zhipuai_sdk/api_resource/videos/__init__.py
 delete mode 100644 api/core/model_runtime/model_providers/zhipuai/zhipuai_sdk/api_resource/videos/videos.py
 delete mode 100644 api/core/model_runtime/model_providers/zhipuai/zhipuai_sdk/core/__init__.py
 delete mode 100644 api/core/model_runtime/model_providers/zhipuai/zhipuai_sdk/core/_base_api.py
 delete mode 100644 api/core/model_runtime/model_providers/zhipuai/zhipuai_sdk/core/_base_compat.py
 delete mode 100644 api/core/model_runtime/model_providers/zhipuai/zhipuai_sdk/core/_base_models.py
 delete mode 100644 api/core/model_runtime/model_providers/zhipuai/zhipuai_sdk/core/_base_type.py
 delete mode 100644 api/core/model_runtime/model_providers/zhipuai/zhipuai_sdk/core/_constants.py
 delete mode 100644 api/core/model_runtime/model_providers/zhipuai/zhipuai_sdk/core/_errors.py
 delete mode 100644 api/core/model_runtime/model_providers/zhipuai/zhipuai_sdk/core/_files.py
 delete mode 100644 api/core/model_runtime/model_providers/zhipuai/zhipuai_sdk/core/_http_client.py
 delete mode 100644 api/core/model_runtime/model_providers/zhipuai/zhipuai_sdk/core/_jwt_token.py
 delete mode 100644 api/core/model_runtime/model_providers/zhipuai/zhipuai_sdk/core/_legacy_binary_response.py
 delete mode 100644 api/core/model_runtime/model_providers/zhipuai/zhipuai_sdk/core/_legacy_response.py
 delete mode 100644 api/core/model_runtime/model_providers/zhipuai/zhipuai_sdk/core/_request_opt.py
 delete mode 100644 api/core/model_runtime/model_providers/zhipuai/zhipuai_sdk/core/_response.py
 delete mode 100644 api/core/model_runtime/model_providers/zhipuai/zhipuai_sdk/core/_sse_client.py
 delete mode 100644 api/core/model_runtime/model_providers/zhipuai/zhipuai_sdk/core/_utils/__init__.py
 delete mode 100644 api/core/model_runtime/model_providers/zhipuai/zhipuai_sdk/core/_utils/_transform.py
 delete mode 100644 api/core/model_runtime/model_providers/zhipuai/zhipuai_sdk/core/_utils/_typing.py
 delete mode 100644 api/core/model_runtime/model_providers/zhipuai/zhipuai_sdk/core/_utils/_utils.py
 delete mode 100644 api/core/model_runtime/model_providers/zhipuai/zhipuai_sdk/core/logs.py
 delete mode 100644 api/core/model_runtime/model_providers/zhipuai/zhipuai_sdk/core/pagination.py
 delete mode 100644 api/core/model_runtime/model_providers/zhipuai/zhipuai_sdk/types/__init__.py
 delete mode 100644 api/core/model_runtime/model_providers/zhipuai/zhipuai_sdk/types/assistant/__init__.py
 delete mode 100644 api/core/model_runtime/model_providers/zhipuai/zhipuai_sdk/types/assistant/assistant_completion.py
 delete mode 100644 api/core/model_runtime/model_providers/zhipuai/zhipuai_sdk/types/assistant/assistant_conversation_params.py
 delete mode 100644 api/core/model_runtime/model_providers/zhipuai/zhipuai_sdk/types/assistant/assistant_conversation_resp.py
 delete mode 100644 api/core/model_runtime/model_providers/zhipuai/zhipuai_sdk/types/assistant/assistant_create_params.py
 delete mode 100644 api/core/model_runtime/model_providers/zhipuai/zhipuai_sdk/types/assistant/assistant_support_resp.py
 delete mode 100644 api/core/model_runtime/model_providers/zhipuai/zhipuai_sdk/types/assistant/message/__init__.py
 delete mode 100644 api/core/model_runtime/model_providers/zhipuai/zhipuai_sdk/types/assistant/message/message_content.py
 delete mode 100644 api/core/model_runtime/model_providers/zhipuai/zhipuai_sdk/types/assistant/message/text_content_block.py
 delete mode 100644 api/core/model_runtime/model_providers/zhipuai/zhipuai_sdk/types/assistant/message/tools/code_interpreter_delta_block.py
 delete mode 100644 api/core/model_runtime/model_providers/zhipuai/zhipuai_sdk/types/assistant/message/tools/drawing_tool_delta_block.py
 delete mode 100644 api/core/model_runtime/model_providers/zhipuai/zhipuai_sdk/types/assistant/message/tools/function_delta_block.py
 delete mode 100644 api/core/model_runtime/model_providers/zhipuai/zhipuai_sdk/types/assistant/message/tools/retrieval_delta_black.py
 delete mode 100644 api/core/model_runtime/model_providers/zhipuai/zhipuai_sdk/types/assistant/message/tools/tools_type.py
 delete mode 100644 api/core/model_runtime/model_providers/zhipuai/zhipuai_sdk/types/assistant/message/tools/web_browser_delta_block.py
 delete mode 100644 api/core/model_runtime/model_providers/zhipuai/zhipuai_sdk/types/assistant/message/tools_delta_block.py
 delete mode 100644 api/core/model_runtime/model_providers/zhipuai/zhipuai_sdk/types/batch.py
 delete mode 100644 api/core/model_runtime/model_providers/zhipuai/zhipuai_sdk/types/batch_create_params.py
 delete mode 100644 api/core/model_runtime/model_providers/zhipuai/zhipuai_sdk/types/batch_error.py
 delete mode 100644 api/core/model_runtime/model_providers/zhipuai/zhipuai_sdk/types/batch_list_params.py
 delete mode 100644 api/core/model_runtime/model_providers/zhipuai/zhipuai_sdk/types/batch_request_counts.py
 delete mode 100644 api/core/model_runtime/model_providers/zhipuai/zhipuai_sdk/types/chat/__init__.py
 delete mode 100644 api/core/model_runtime/model_providers/zhipuai/zhipuai_sdk/types/chat/async_chat_completion.py
 delete mode 100644 api/core/model_runtime/model_providers/zhipuai/zhipuai_sdk/types/chat/chat_completion.py
 delete mode 100644 api/core/model_runtime/model_providers/zhipuai/zhipuai_sdk/types/chat/chat_completion_chunk.py
 delete mode 100644 api/core/model_runtime/model_providers/zhipuai/zhipuai_sdk/types/chat/chat_completions_create_param.py
 delete mode 100644 api/core/model_runtime/model_providers/zhipuai/zhipuai_sdk/types/chat/code_geex/code_geex_params.py
 delete mode 100644 api/core/model_runtime/model_providers/zhipuai/zhipuai_sdk/types/embeddings.py
 delete mode 100644 api/core/model_runtime/model_providers/zhipuai/zhipuai_sdk/types/files/__init__.py
 delete mode 100644 api/core/model_runtime/model_providers/zhipuai/zhipuai_sdk/types/files/file_create_params.py
 delete mode 100644 api/core/model_runtime/model_providers/zhipuai/zhipuai_sdk/types/files/file_deleted.py
 delete mode 100644 api/core/model_runtime/model_providers/zhipuai/zhipuai_sdk/types/files/file_object.py
 delete mode 100644 api/core/model_runtime/model_providers/zhipuai/zhipuai_sdk/types/files/upload_detail.py
 delete mode 100644 api/core/model_runtime/model_providers/zhipuai/zhipuai_sdk/types/fine_tuning/__init__.py
 delete mode 100644 api/core/model_runtime/model_providers/zhipuai/zhipuai_sdk/types/fine_tuning/fine_tuning_job.py
 delete mode 100644 api/core/model_runtime/model_providers/zhipuai/zhipuai_sdk/types/fine_tuning/fine_tuning_job_event.py
 delete mode 100644 api/core/model_runtime/model_providers/zhipuai/zhipuai_sdk/types/fine_tuning/job_create_params.py
 delete mode 100644 api/core/model_runtime/model_providers/zhipuai/zhipuai_sdk/types/fine_tuning/models/__init__.py
 delete mode 100644 api/core/model_runtime/model_providers/zhipuai/zhipuai_sdk/types/fine_tuning/models/fine_tuned_models.py
 delete mode 100644 api/core/model_runtime/model_providers/zhipuai/zhipuai_sdk/types/image.py
 delete mode 100644 api/core/model_runtime/model_providers/zhipuai/zhipuai_sdk/types/knowledge/__init__.py
 delete mode 100644 api/core/model_runtime/model_providers/zhipuai/zhipuai_sdk/types/knowledge/document/__init__.py
 delete mode 100644 api/core/model_runtime/model_providers/zhipuai/zhipuai_sdk/types/knowledge/document/document.py
 delete mode 100644 api/core/model_runtime/model_providers/zhipuai/zhipuai_sdk/types/knowledge/document/document_edit_params.py
 delete mode 100644 api/core/model_runtime/model_providers/zhipuai/zhipuai_sdk/types/knowledge/document/document_list_params.py
 delete mode 100644 api/core/model_runtime/model_providers/zhipuai/zhipuai_sdk/types/knowledge/document/document_list_resp.py
 delete mode 100644 api/core/model_runtime/model_providers/zhipuai/zhipuai_sdk/types/knowledge/knowledge.py
 delete mode 100644 api/core/model_runtime/model_providers/zhipuai/zhipuai_sdk/types/knowledge/knowledge_create_params.py
 delete mode 100644 api/core/model_runtime/model_providers/zhipuai/zhipuai_sdk/types/knowledge/knowledge_list_params.py
 delete mode 100644 api/core/model_runtime/model_providers/zhipuai/zhipuai_sdk/types/knowledge/knowledge_list_resp.py
 delete mode 100644 api/core/model_runtime/model_providers/zhipuai/zhipuai_sdk/types/knowledge/knowledge_used.py
 delete mode 100644 api/core/model_runtime/model_providers/zhipuai/zhipuai_sdk/types/sensitive_word_check/__init__.py
 delete mode 100644 api/core/model_runtime/model_providers/zhipuai/zhipuai_sdk/types/sensitive_word_check/sensitive_word_check.py
 delete mode 100644 api/core/model_runtime/model_providers/zhipuai/zhipuai_sdk/types/tools/__init__.py
 delete mode 100644 api/core/model_runtime/model_providers/zhipuai/zhipuai_sdk/types/tools/tools_web_search_params.py
 delete mode 100644 api/core/model_runtime/model_providers/zhipuai/zhipuai_sdk/types/tools/web_search.py
 delete mode 100644 api/core/model_runtime/model_providers/zhipuai/zhipuai_sdk/types/tools/web_search_chunk.py
 delete mode 100644 api/core/model_runtime/model_providers/zhipuai/zhipuai_sdk/types/video/__init__.py
 delete mode 100644 api/core/model_runtime/model_providers/zhipuai/zhipuai_sdk/types/video/video_create_params.py
 delete mode 100644 api/core/model_runtime/model_providers/zhipuai/zhipuai_sdk/types/video/video_object.py

diff --git a/api/core/model_runtime/model_providers/zhipuai/llm/llm.py b/api/core/model_runtime/model_providers/zhipuai/llm/llm.py
index e0c4980523075..43bffad2a0bc3 100644
--- a/api/core/model_runtime/model_providers/zhipuai/llm/llm.py
+++ b/api/core/model_runtime/model_providers/zhipuai/llm/llm.py
@@ -1,6 +1,10 @@
 from collections.abc import Generator
 from typing import Optional, Union
 
+from zhipuai import ZhipuAI
+from zhipuai.types.chat.chat_completion import Completion
+from zhipuai.types.chat.chat_completion_chunk import ChatCompletionChunk
+
 from core.model_runtime.entities.llm_entities import LLMResult, LLMResultChunk, LLMResultChunkDelta
 from core.model_runtime.entities.message_entities import (
     AssistantPromptMessage,
@@ -16,9 +20,6 @@
 from core.model_runtime.errors.validate import CredentialsValidateFailedError
 from core.model_runtime.model_providers.__base.large_language_model import LargeLanguageModel
 from core.model_runtime.model_providers.zhipuai._common import _CommonZhipuaiAI
-from core.model_runtime.model_providers.zhipuai.zhipuai_sdk._client import ZhipuAI
-from core.model_runtime.model_providers.zhipuai.zhipuai_sdk.types.chat.chat_completion import Completion
-from core.model_runtime.model_providers.zhipuai.zhipuai_sdk.types.chat.chat_completion_chunk import ChatCompletionChunk
 from core.model_runtime.utils import helper
 
 GLM_JSON_MODE_PROMPT = """You should always follow the instructions and output a valid JSON object.
diff --git a/api/core/model_runtime/model_providers/zhipuai/text_embedding/text_embedding.py b/api/core/model_runtime/model_providers/zhipuai/text_embedding/text_embedding.py
index 14a529dddf82d..5a34a3d593997 100644
--- a/api/core/model_runtime/model_providers/zhipuai/text_embedding/text_embedding.py
+++ b/api/core/model_runtime/model_providers/zhipuai/text_embedding/text_embedding.py
@@ -1,13 +1,14 @@
 import time
 from typing import Optional
 
+from zhipuai import ZhipuAI
+
 from core.embedding.embedding_constant import EmbeddingInputType
 from core.model_runtime.entities.model_entities import PriceType
 from core.model_runtime.entities.text_embedding_entities import EmbeddingUsage, TextEmbeddingResult
 from core.model_runtime.errors.validate import CredentialsValidateFailedError
 from core.model_runtime.model_providers.__base.text_embedding_model import TextEmbeddingModel
 from core.model_runtime.model_providers.zhipuai._common import _CommonZhipuaiAI
-from core.model_runtime.model_providers.zhipuai.zhipuai_sdk._client import ZhipuAI
 
 
 class ZhipuAITextEmbeddingModel(_CommonZhipuaiAI, TextEmbeddingModel):
diff --git a/api/core/model_runtime/model_providers/zhipuai/zhipuai_sdk/__init__.py b/api/core/model_runtime/model_providers/zhipuai/zhipuai_sdk/__init__.py
deleted file mode 100644
index fc71d64714bd9..0000000000000
--- a/api/core/model_runtime/model_providers/zhipuai/zhipuai_sdk/__init__.py
+++ /dev/null
@@ -1,15 +0,0 @@
-from .__version__ import __version__
-from ._client import ZhipuAI
-from .core import (
-    APIAuthenticationError,
-    APIConnectionError,
-    APIInternalError,
-    APIReachLimitError,
-    APIRequestFailedError,
-    APIResponseError,
-    APIResponseValidationError,
-    APIServerFlowExceedError,
-    APIStatusError,
-    APITimeoutError,
-    ZhipuAIError,
-)
diff --git a/api/core/model_runtime/model_providers/zhipuai/zhipuai_sdk/__version__.py b/api/core/model_runtime/model_providers/zhipuai/zhipuai_sdk/__version__.py
deleted file mode 100644
index 51f8c49ecb827..0000000000000
--- a/api/core/model_runtime/model_providers/zhipuai/zhipuai_sdk/__version__.py
+++ /dev/null
@@ -1 +0,0 @@
-__version__ = "v2.1.0"
diff --git a/api/core/model_runtime/model_providers/zhipuai/zhipuai_sdk/_client.py b/api/core/model_runtime/model_providers/zhipuai/zhipuai_sdk/_client.py
deleted file mode 100644
index 705d371e628f0..0000000000000
--- a/api/core/model_runtime/model_providers/zhipuai/zhipuai_sdk/_client.py
+++ /dev/null
@@ -1,82 +0,0 @@
-from __future__ import annotations
-
-import os
-from collections.abc import Mapping
-from typing import Union
-
-import httpx
-from httpx import Timeout
-from typing_extensions import override
-
-from . import api_resource
-from .core import NOT_GIVEN, ZHIPUAI_DEFAULT_MAX_RETRIES, HttpClient, NotGiven, ZhipuAIError, _jwt_token
-
-
-class ZhipuAI(HttpClient):
-    chat: api_resource.chat.Chat
-    api_key: str
-    _disable_token_cache: bool = True
-
-    def __init__(
-        self,
-        *,
-        api_key: str | None = None,
-        base_url: str | httpx.URL | None = None,
-        timeout: Union[float, Timeout, None, NotGiven] = NOT_GIVEN,
-        max_retries: int = ZHIPUAI_DEFAULT_MAX_RETRIES,
-        http_client: httpx.Client | None = None,
-        custom_headers: Mapping[str, str] | None = None,
-        disable_token_cache: bool = True,
-        _strict_response_validation: bool = False,
-    ) -> None:
-        if api_key is None:
-            api_key = os.environ.get("ZHIPUAI_API_KEY")
-        if api_key is None:
-            raise ZhipuAIError("未提供api_key，请通过参数或环境变量提供")
-        self.api_key = api_key
-        self._disable_token_cache = disable_token_cache
-
-        if base_url is None:
-            base_url = os.environ.get("ZHIPUAI_BASE_URL")
-        if base_url is None:
-            base_url = "https://open.bigmodel.cn/api/paas/v4"
-        from .__version__ import __version__
-
-        super().__init__(
-            version=__version__,
-            base_url=base_url,
-            max_retries=max_retries,
-            timeout=timeout,
-            custom_httpx_client=http_client,
-            custom_headers=custom_headers,
-            _strict_response_validation=_strict_response_validation,
-        )
-        self.chat = api_resource.chat.Chat(self)
-        self.images = api_resource.images.Images(self)
-        self.embeddings = api_resource.embeddings.Embeddings(self)
-        self.files = api_resource.files.Files(self)
-        self.fine_tuning = api_resource.fine_tuning.FineTuning(self)
-        self.batches = api_resource.Batches(self)
-        self.knowledge = api_resource.Knowledge(self)
-        self.tools = api_resource.Tools(self)
-        self.videos = api_resource.Videos(self)
-        self.assistant = api_resource.Assistant(self)
-
-    @property
-    @override
-    def auth_headers(self) -> dict[str, str]:
-        api_key = self.api_key
-        if self._disable_token_cache:
-            return {"Authorization": f"Bearer {api_key}"}
-        else:
-            return {"Authorization": f"Bearer {_jwt_token.generate_token(api_key)}"}
-
-    def __del__(self) -> None:
-        if not hasattr(self, "_has_custom_http_client") or not hasattr(self, "close") or not hasattr(self, "_client"):
-            # if the '__init__' method raised an error, self would not have client attr
-            return
-
-        if self._has_custom_http_client:
-            return
-
-        self.close()
diff --git a/api/core/model_runtime/model_providers/zhipuai/zhipuai_sdk/api_resource/__init__.py b/api/core/model_runtime/model_providers/zhipuai/zhipuai_sdk/api_resource/__init__.py
deleted file mode 100644
index 4fe0719dde3e0..0000000000000
--- a/api/core/model_runtime/model_providers/zhipuai/zhipuai_sdk/api_resource/__init__.py
+++ /dev/null
@@ -1,34 +0,0 @@
-from .assistant import (
-    Assistant,
-)
-from .batches import Batches
-from .chat import (
-    AsyncCompletions,
-    Chat,
-    Completions,
-)
-from .embeddings import Embeddings
-from .files import Files, FilesWithRawResponse
-from .fine_tuning import FineTuning
-from .images import Images
-from .knowledge import Knowledge
-from .tools import Tools
-from .videos import (
-    Videos,
-)
-
-__all__ = [
-    "Videos",
-    "AsyncCompletions",
-    "Chat",
-    "Completions",
-    "Images",
-    "Embeddings",
-    "Files",
-    "FilesWithRawResponse",
-    "FineTuning",
-    "Batches",
-    "Knowledge",
-    "Tools",
-    "Assistant",
-]
diff --git a/api/core/model_runtime/model_providers/zhipuai/zhipuai_sdk/api_resource/assistant/__init__.py b/api/core/model_runtime/model_providers/zhipuai/zhipuai_sdk/api_resource/assistant/__init__.py
deleted file mode 100644
index ce619aa7f0922..0000000000000
--- a/api/core/model_runtime/model_providers/zhipuai/zhipuai_sdk/api_resource/assistant/__init__.py
+++ /dev/null
@@ -1,3 +0,0 @@
-from .assistant import Assistant
-
-__all__ = ["Assistant"]
diff --git a/api/core/model_runtime/model_providers/zhipuai/zhipuai_sdk/api_resource/assistant/assistant.py b/api/core/model_runtime/model_providers/zhipuai/zhipuai_sdk/api_resource/assistant/assistant.py
deleted file mode 100644
index c29b05749847c..0000000000000
--- a/api/core/model_runtime/model_providers/zhipuai/zhipuai_sdk/api_resource/assistant/assistant.py
+++ /dev/null
@@ -1,122 +0,0 @@
-from __future__ import annotations
-
-from typing import TYPE_CHECKING, Optional
-
-import httpx
-
-from ...core import (
-    NOT_GIVEN,
-    BaseAPI,
-    Body,
-    Headers,
-    NotGiven,
-    StreamResponse,
-    deepcopy_minimal,
-    make_request_options,
-    maybe_transform,
-)
-from ...types.assistant import AssistantCompletion
-from ...types.assistant.assistant_conversation_resp import ConversationUsageListResp
-from ...types.assistant.assistant_support_resp import AssistantSupportResp
-
-if TYPE_CHECKING:
-    from ..._client import ZhipuAI
-
-from ...types.assistant import assistant_conversation_params, assistant_create_params
-
-__all__ = ["Assistant"]
-
-
-class Assistant(BaseAPI):
-    def __init__(self, client: ZhipuAI) -> None:
-        super().__init__(client)
-
-    def conversation(
-        self,
-        assistant_id: str,
-        model: str,
-        messages: list[assistant_create_params.ConversationMessage],
-        *,
-        stream: bool = True,
-        conversation_id: Optional[str] = None,
-        attachments: Optional[list[assistant_create_params.AssistantAttachments]] = None,
-        metadata: dict | None = None,
-        request_id: Optional[str] = None,
-        user_id: Optional[str] = None,
-        extra_headers: Headers | None = None,
-        extra_body: Body | None = None,
-        timeout: float | httpx.Timeout | None | NotGiven = NOT_GIVEN,
-    ) -> StreamResponse[AssistantCompletion]:
-        body = deepcopy_minimal(
-            {
-                "assistant_id": assistant_id,
-                "model": model,
-                "messages": messages,
-                "stream": stream,
-                "conversation_id": conversation_id,
-                "attachments": attachments,
-                "metadata": metadata,
-                "request_id": request_id,
-                "user_id": user_id,
-            }
-        )
-        return self._post(
-            "/assistant",
-            body=maybe_transform(body, assistant_create_params.AssistantParameters),
-            options=make_request_options(extra_headers=extra_headers, extra_body=extra_body, timeout=timeout),
-            cast_type=AssistantCompletion,
-            stream=stream or True,
-            stream_cls=StreamResponse[AssistantCompletion],
-        )
-
-    def query_support(
-        self,
-        *,
-        assistant_id_list: Optional[list[str]] = None,
-        request_id: Optional[str] = None,
-        user_id: Optional[str] = None,
-        extra_headers: Headers | None = None,
-        extra_body: Body | None = None,
-        timeout: float | httpx.Timeout | None | NotGiven = NOT_GIVEN,
-    ) -> AssistantSupportResp:
-        body = deepcopy_minimal(
-            {
-                "assistant_id_list": assistant_id_list,
-                "request_id": request_id,
-                "user_id": user_id,
-            }
-        )
-        return self._post(
-            "/assistant/list",
-            body=body,
-            options=make_request_options(extra_headers=extra_headers, extra_body=extra_body, timeout=timeout),
-            cast_type=AssistantSupportResp,
-        )
-
-    def query_conversation_usage(
-        self,
-        assistant_id: str,
-        page: int = 1,
-        page_size: int = 10,
-        *,
-        request_id: Optional[str] = None,
-        user_id: Optional[str] = None,
-        extra_headers: Headers | None = None,
-        extra_body: Body | None = None,
-        timeout: float | httpx.Timeout | None | NotGiven = NOT_GIVEN,
-    ) -> ConversationUsageListResp:
-        body = deepcopy_minimal(
-            {
-                "assistant_id": assistant_id,
-                "page": page,
-                "page_size": page_size,
-                "request_id": request_id,
-                "user_id": user_id,
-            }
-        )
-        return self._post(
-            "/assistant/conversation/list",
-            body=maybe_transform(body, assistant_conversation_params.ConversationParameters),
-            options=make_request_options(extra_headers=extra_headers, extra_body=extra_body, timeout=timeout),
-            cast_type=ConversationUsageListResp,
-        )
diff --git a/api/core/model_runtime/model_providers/zhipuai/zhipuai_sdk/api_resource/batches.py b/api/core/model_runtime/model_providers/zhipuai/zhipuai_sdk/api_resource/batches.py
deleted file mode 100644
index ae2f2be85eb9b..0000000000000
--- a/api/core/model_runtime/model_providers/zhipuai/zhipuai_sdk/api_resource/batches.py
+++ /dev/null
@@ -1,146 +0,0 @@
-from __future__ import annotations
-
-from typing import TYPE_CHECKING, Literal, Optional
-
-import httpx
-
-from ..core import NOT_GIVEN, BaseAPI, Body, Headers, NotGiven, make_request_options, maybe_transform
-from ..core.pagination import SyncCursorPage
-from ..types import batch_create_params, batch_list_params
-from ..types.batch import Batch
-
-if TYPE_CHECKING:
-    from .._client import ZhipuAI
-
-
-class Batches(BaseAPI):
-    def __init__(self, client: ZhipuAI) -> None:
-        super().__init__(client)
-
-    def create(
-        self,
-        *,
-        completion_window: str | None = None,
-        endpoint: Literal["/v1/chat/completions", "/v1/embeddings"],
-        input_file_id: str,
-        metadata: Optional[dict[str, str]] | NotGiven = NOT_GIVEN,
-        auto_delete_input_file: bool = True,
-        extra_headers: Headers | None = None,
-        extra_body: Body | None = None,
-        timeout: float | httpx.Timeout | None | NotGiven = NOT_GIVEN,
-    ) -> Batch:
-        return self._post(
-            "/batches",
-            body=maybe_transform(
-                {
-                    "completion_window": completion_window,
-                    "endpoint": endpoint,
-                    "input_file_id": input_file_id,
-                    "metadata": metadata,
-                    "auto_delete_input_file": auto_delete_input_file,
-                },
-                batch_create_params.BatchCreateParams,
-            ),
-            options=make_request_options(extra_headers=extra_headers, extra_body=extra_body, timeout=timeout),
-            cast_type=Batch,
-        )
-
-    def retrieve(
-        self,
-        batch_id: str,
-        *,
-        extra_headers: Headers | None = None,
-        extra_body: Body | None = None,
-        timeout: float | httpx.Timeout | None | NotGiven = NOT_GIVEN,
-    ) -> Batch:
-        """
-        Retrieves a batch.
-
-        Args:
-          extra_headers: Send extra headers
-
-          extra_body: Add additional JSON properties to the request
-
-          timeout: Override the client-level default timeout for this request, in seconds
-        """
-        if not batch_id:
-            raise ValueError(f"Expected a non-empty value for `batch_id` but received {batch_id!r}")
-        return self._get(
-            f"/batches/{batch_id}",
-            options=make_request_options(extra_headers=extra_headers, extra_body=extra_body, timeout=timeout),
-            cast_type=Batch,
-        )
-
-    def list(
-        self,
-        *,
-        after: str | NotGiven = NOT_GIVEN,
-        limit: int | NotGiven = NOT_GIVEN,
-        extra_headers: Headers | None = None,
-        extra_body: Body | None = None,
-        timeout: float | httpx.Timeout | None | NotGiven = NOT_GIVEN,
-    ) -> SyncCursorPage[Batch]:
-        """List your organization's batches.
-
-        Args:
-          after: A cursor for use in pagination.
-
-            `after` is an object ID that defines your place
-              in the list. For instance, if you make a list request and receive 100 objects,
-              ending with obj_foo, your subsequent call can include after=obj_foo in order to
-              fetch the next page of the list.
-
-          limit: A limit on the number of objects to be returned. Limit can range between 1 and
-              100, and the default is 20.
-
-          extra_headers: Send extra headers
-
-          extra_body: Add additional JSON properties to the request
-
-          timeout: Override the client-level default timeout for this request, in seconds
-        """
-        return self._get_api_list(
-            "/batches",
-            page=SyncCursorPage[Batch],
-            options=make_request_options(
-                extra_headers=extra_headers,
-                extra_body=extra_body,
-                timeout=timeout,
-                query=maybe_transform(
-                    {
-                        "after": after,
-                        "limit": limit,
-                    },
-                    batch_list_params.BatchListParams,
-                ),
-            ),
-            model=Batch,
-        )
-
-    def cancel(
-        self,
-        batch_id: str,
-        *,
-        extra_headers: Headers | None = None,
-        extra_body: Body | None = None,
-        timeout: float | httpx.Timeout | None | NotGiven = NOT_GIVEN,
-    ) -> Batch:
-        """
-        Cancels an in-progress batch.
-
-        Args:
-          batch_id: The ID of the batch to cancel.
-          extra_headers: Send extra headers
-
-          extra_body: Add additional JSON properties to the request
-
-          timeout: Override the client-level default timeout for this request, in seconds
-
-        """
-        if not batch_id:
-            raise ValueError(f"Expected a non-empty value for `batch_id` but received {batch_id!r}")
-        return self._post(
-            f"/batches/{batch_id}/cancel",
-            options=make_request_options(extra_headers=extra_headers, extra_body=extra_body, timeout=timeout),
-            cast_type=Batch,
-        )
diff --git a/api/core/model_runtime/model_providers/zhipuai/zhipuai_sdk/api_resource/chat/__init__.py b/api/core/model_runtime/model_providers/zhipuai/zhipuai_sdk/api_resource/chat/__init__.py
deleted file mode 100644
index 5cd8dc6f339a6..0000000000000
--- a/api/core/model_runtime/model_providers/zhipuai/zhipuai_sdk/api_resource/chat/__init__.py
+++ /dev/null
@@ -1,5 +0,0 @@
-from .async_completions import AsyncCompletions
-from .chat import Chat
-from .completions import Completions
-
-__all__ = ["AsyncCompletions", "Chat", "Completions"]
diff --git a/api/core/model_runtime/model_providers/zhipuai/zhipuai_sdk/api_resource/chat/async_completions.py b/api/core/model_runtime/model_providers/zhipuai/zhipuai_sdk/api_resource/chat/async_completions.py
deleted file mode 100644
index 05510a3ec421d..0000000000000
--- a/api/core/model_runtime/model_providers/zhipuai/zhipuai_sdk/api_resource/chat/async_completions.py
+++ /dev/null
@@ -1,115 +0,0 @@
-from __future__ import annotations
-
-import logging
-from typing import TYPE_CHECKING, Literal, Optional, Union
-
-import httpx
-
-from ...core import (
-    NOT_GIVEN,
-    BaseAPI,
-    Body,
-    Headers,
-    NotGiven,
-    drop_prefix_image_data,
-    make_request_options,
-    maybe_transform,
-)
-from ...types.chat.async_chat_completion import AsyncCompletion, AsyncTaskStatus
-from ...types.chat.code_geex import code_geex_params
-from ...types.sensitive_word_check import SensitiveWordCheckRequest
-
-logger = logging.getLogger(__name__)
-
-if TYPE_CHECKING:
-    from ..._client import ZhipuAI
-
-
-class AsyncCompletions(BaseAPI):
-    def __init__(self, client: ZhipuAI) -> None:
-        super().__init__(client)
-
-    def create(
-        self,
-        *,
-        model: str,
-        request_id: Optional[str] | NotGiven = NOT_GIVEN,
-        user_id: Optional[str] | NotGiven = NOT_GIVEN,
-        do_sample: Optional[Literal[False]] | Literal[True] | NotGiven = NOT_GIVEN,
-        temperature: Optional[float] | NotGiven = NOT_GIVEN,
-        top_p: Optional[float] | NotGiven = NOT_GIVEN,
-        max_tokens: int | NotGiven = NOT_GIVEN,
-        seed: int | NotGiven = NOT_GIVEN,
-        messages: Union[str, list[str], list[int], list[list[int]], None],
-        stop: Optional[Union[str, list[str], None]] | NotGiven = NOT_GIVEN,
-        sensitive_word_check: Optional[SensitiveWordCheckRequest] | NotGiven = NOT_GIVEN,
-        tools: Optional[object] | NotGiven = NOT_GIVEN,
-        tool_choice: str | NotGiven = NOT_GIVEN,
-        meta: Optional[dict[str, str]] | NotGiven = NOT_GIVEN,
-        extra: Optional[code_geex_params.CodeGeexExtra] | NotGiven = NOT_GIVEN,
-        extra_headers: Headers | None = None,
-        extra_body: Body | None = None,
-        timeout: float | httpx.Timeout | None | NotGiven = NOT_GIVEN,
-    ) -> AsyncTaskStatus:
-        _cast_type = AsyncTaskStatus
-        logger.debug(f"temperature:{temperature}, top_p:{top_p}")
-        if temperature is not None and temperature != NOT_GIVEN:
-            if temperature <= 0:
-                do_sample = False
-                temperature = 0.01
-                # logger.warning("temperature:取值范围是：(0.0, 1.0) 开区间，do_sample重写为:false（参数top_p temperature不生效）")  # noqa: E501
-            if temperature >= 1:
-                temperature = 0.99
-                # logger.warning("temperature:取值范围是：(0.0, 1.0) 开区间")
-        if top_p is not None and top_p != NOT_GIVEN:
-            if top_p >= 1:
-                top_p = 0.99
-                # logger.warning("top_p:取值范围是：(0.0, 1.0) 开区间，不能等于 0 或 1")
-            if top_p <= 0:
-                top_p = 0.01
-                # logger.warning("top_p:取值范围是：(0.0, 1.0) 开区间，不能等于 0 或 1")
-
-        logger.debug(f"temperature:{temperature}, top_p:{top_p}")
-        if isinstance(messages, list):
-            for item in messages:
-                if item.get("content"):
-                    item["content"] = drop_prefix_image_data(item["content"])
-
-        body = {
-            "model": model,
-            "request_id": request_id,
-            "user_id": user_id,
-            "temperature": temperature,
-            "top_p": top_p,
-            "do_sample": do_sample,
-            "max_tokens": max_tokens,
-            "seed": seed,
-            "messages": messages,
-            "stop": stop,
-            "sensitive_word_check": sensitive_word_check,
-            "tools": tools,
-            "tool_choice": tool_choice,
-            "meta": meta,
-            "extra": maybe_transform(extra, code_geex_params.CodeGeexExtra),
-        }
-        return self._post(
-            "/async/chat/completions",
-            body=body,
-            options=make_request_options(extra_headers=extra_headers, extra_body=extra_body, timeout=timeout),
-            cast_type=_cast_type,
-            stream=False,
-        )
-
-    def retrieve_completion_result(
-        self,
-        id: str,
-        extra_headers: Headers | None = None,
-        extra_body: Body | None = None,
-        timeout: float | httpx.Timeout | None | NotGiven = NOT_GIVEN,
-    ) -> Union[AsyncCompletion, AsyncTaskStatus]:
-        _cast_type = Union[AsyncCompletion, AsyncTaskStatus]
-        return self._get(
-            path=f"/async-result/{id}",
-            cast_type=_cast_type,
-            options=make_request_options(extra_headers=extra_headers, extra_body=extra_body, timeout=timeout),
-        )
diff --git a/api/core/model_runtime/model_providers/zhipuai/zhipuai_sdk/api_resource/chat/chat.py b/api/core/model_runtime/model_providers/zhipuai/zhipuai_sdk/api_resource/chat/chat.py
deleted file mode 100644
index b3cc46566c7bf..0000000000000
--- a/api/core/model_runtime/model_providers/zhipuai/zhipuai_sdk/api_resource/chat/chat.py
+++ /dev/null
@@ -1,18 +0,0 @@
-from typing import TYPE_CHECKING
-
-from ...core import BaseAPI, cached_property
-from .async_completions import AsyncCompletions
-from .completions import Completions
-
-if TYPE_CHECKING:
-    pass
-
-
-class Chat(BaseAPI):
-    @cached_property
-    def completions(self) -> Completions:
-        return Completions(self._client)
-
-    @cached_property
-    def asyncCompletions(self) -> AsyncCompletions:  # noqa: N802
-        return AsyncCompletions(self._client)
diff --git a/api/core/model_runtime/model_providers/zhipuai/zhipuai_sdk/api_resource/chat/completions.py b/api/core/model_runtime/model_providers/zhipuai/zhipuai_sdk/api_resource/chat/completions.py
deleted file mode 100644
index 8e5bb454e6ce7..0000000000000
--- a/api/core/model_runtime/model_providers/zhipuai/zhipuai_sdk/api_resource/chat/completions.py
+++ /dev/null
@@ -1,108 +0,0 @@
-from __future__ import annotations
-
-import logging
-from typing import TYPE_CHECKING, Literal, Optional, Union
-
-import httpx
-
-from ...core import (
-    NOT_GIVEN,
-    BaseAPI,
-    Body,
-    Headers,
-    NotGiven,
-    StreamResponse,
-    deepcopy_minimal,
-    drop_prefix_image_data,
-    make_request_options,
-    maybe_transform,
-)
-from ...types.chat.chat_completion import Completion
-from ...types.chat.chat_completion_chunk import ChatCompletionChunk
-from ...types.chat.code_geex import code_geex_params
-from ...types.sensitive_word_check import SensitiveWordCheckRequest
-
-logger = logging.getLogger(__name__)
-
-if TYPE_CHECKING:
-    from ..._client import ZhipuAI
-
-
-class Completions(BaseAPI):
-    def __init__(self, client: ZhipuAI) -> None:
-        super().__init__(client)
-
-    def create(
-        self,
-        *,
-        model: str,
-        request_id: Optional[str] | NotGiven = NOT_GIVEN,
-        user_id: Optional[str] | NotGiven = NOT_GIVEN,
-        do_sample: Optional[Literal[False]] | Literal[True] | NotGiven = NOT_GIVEN,
-        stream: Optional[Literal[False]] | Literal[True] | NotGiven = NOT_GIVEN,
-        temperature: Optional[float] | NotGiven = NOT_GIVEN,
-        top_p: Optional[float] | NotGiven = NOT_GIVEN,
-        max_tokens: int | NotGiven = NOT_GIVEN,
-        seed: int | NotGiven = NOT_GIVEN,
-        messages: Union[str, list[str], list[int], object, None],
-        stop: Optional[Union[str, list[str], None]] | NotGiven = NOT_GIVEN,
-        sensitive_word_check: Optional[SensitiveWordCheckRequest] | NotGiven = NOT_GIVEN,
-        tools: Optional[object] | NotGiven = NOT_GIVEN,
-        tool_choice: str | NotGiven = NOT_GIVEN,
-        meta: Optional[dict[str, str]] | NotGiven = NOT_GIVEN,
-        extra: Optional[code_geex_params.CodeGeexExtra] | NotGiven = NOT_GIVEN,
-        extra_headers: Headers | None = None,
-        extra_body: Body | None = None,
-        timeout: float | httpx.Timeout | None | NotGiven = NOT_GIVEN,
-    ) -> Completion | StreamResponse[ChatCompletionChunk]:
-        logger.debug(f"temperature:{temperature}, top_p:{top_p}")
-        if temperature is not None and temperature != NOT_GIVEN:
-            if temperature <= 0:
-                do_sample = False
-                temperature = 0.01
-                # logger.warning("temperature:取值范围是：(0.0, 1.0) 开区间，do_sample重写为:false（参数top_p temperature不生效）")  # noqa: E501
-            if temperature >= 1:
-                temperature = 0.99
-                # logger.warning("temperature:取值范围是：(0.0, 1.0) 开区间")
-        if top_p is not None and top_p != NOT_GIVEN:
-            if top_p >= 1:
-                top_p = 0.99
-                # logger.warning("top_p:取值范围是：(0.0, 1.0) 开区间，不能等于 0 或 1")
-            if top_p <= 0:
-                top_p = 0.01
-                # logger.warning("top_p:取值范围是：(0.0, 1.0) 开区间，不能等于 0 或 1")
-
-        logger.debug(f"temperature:{temperature}, top_p:{top_p}")
-        if isinstance(messages, list):
-            for item in messages:
-                if item.get("content"):
-                    item["content"] = drop_prefix_image_data(item["content"])
-
-        body = deepcopy_minimal(
-            {
-                "model": model,
-                "request_id": request_id,
-                "user_id": user_id,
-                "temperature": temperature,
-                "top_p": top_p,
-                "do_sample": do_sample,
-                "max_tokens": max_tokens,
-                "seed": seed,
-                "messages": messages,
-                "stop": stop,
-                "sensitive_word_check": sensitive_word_check,
-                "stream": stream,
-                "tools": tools,
-                "tool_choice": tool_choice,
-                "meta": meta,
-                "extra": maybe_transform(extra, code_geex_params.CodeGeexExtra),
-            }
-        )
-        return self._post(
-            "/chat/completions",
-            body=body,
-            options=make_request_options(extra_headers=extra_headers, extra_body=extra_body, timeout=timeout),
-            cast_type=Completion,
-            stream=stream or False,
-            stream_cls=StreamResponse[ChatCompletionChunk],
-        )
diff --git a/api/core/model_runtime/model_providers/zhipuai/zhipuai_sdk/api_resource/embeddings.py b/api/core/model_runtime/model_providers/zhipuai/zhipuai_sdk/api_resource/embeddings.py
deleted file mode 100644
index 4b4baef9421ba..0000000000000
--- a/api/core/model_runtime/model_providers/zhipuai/zhipuai_sdk/api_resource/embeddings.py
+++ /dev/null
@@ -1,50 +0,0 @@
-from __future__ import annotations
-
-from typing import TYPE_CHECKING, Optional, Union
-
-import httpx
-
-from ..core import NOT_GIVEN, BaseAPI, Body, Headers, NotGiven, make_request_options
-from ..types.embeddings import EmbeddingsResponded
-
-if TYPE_CHECKING:
-    from .._client import ZhipuAI
-
-
-class Embeddings(BaseAPI):
-    def __init__(self, client: ZhipuAI) -> None:
-        super().__init__(client)
-
-    def create(
-        self,
-        *,
-        input: Union[str, list[str], list[int], list[list[int]]],
-        model: Union[str],
-        dimensions: Union[int] | NotGiven = NOT_GIVEN,
-        encoding_format: str | NotGiven = NOT_GIVEN,
-        user: str | NotGiven = NOT_GIVEN,
-        request_id: Optional[str] | NotGiven = NOT_GIVEN,
-        sensitive_word_check: Optional[object] | NotGiven = NOT_GIVEN,
-        extra_headers: Headers | None = None,
-        extra_body: Body | None = None,
-        disable_strict_validation: Optional[bool] | None = None,
-        timeout: float | httpx.Timeout | None | NotGiven = NOT_GIVEN,
-    ) -> EmbeddingsResponded:
-        _cast_type = EmbeddingsResponded
-        if disable_strict_validation:
-            _cast_type = object
-        return self._post(
-            "/embeddings",
-            body={
-                "input": input,
-                "model": model,
-                "dimensions": dimensions,
-                "encoding_format": encoding_format,
-                "user": user,
-                "request_id": request_id,
-                "sensitive_word_check": sensitive_word_check,
-            },
-            options=make_request_options(extra_headers=extra_headers, extra_body=extra_body, timeout=timeout),
-            cast_type=_cast_type,
-            stream=False,
-        )
diff --git a/api/core/model_runtime/model_providers/zhipuai/zhipuai_sdk/api_resource/files.py b/api/core/model_runtime/model_providers/zhipuai/zhipuai_sdk/api_resource/files.py
deleted file mode 100644
index c723f6f66e41c..0000000000000
--- a/api/core/model_runtime/model_providers/zhipuai/zhipuai_sdk/api_resource/files.py
+++ /dev/null
@@ -1,194 +0,0 @@
-from __future__ import annotations
-
-from collections.abc import Mapping
-from typing import TYPE_CHECKING, Literal, Optional, cast
-
-import httpx
-
-from ..core import (
-    NOT_GIVEN,
-    BaseAPI,
-    Body,
-    FileTypes,
-    Headers,
-    NotGiven,
-    _legacy_binary_response,
-    _legacy_response,
-    deepcopy_minimal,
-    extract_files,
-    make_request_options,
-    maybe_transform,
-)
-from ..types.files import FileDeleted, FileObject, ListOfFileObject, UploadDetail, file_create_params
-
-if TYPE_CHECKING:
-    from .._client import ZhipuAI
-
-__all__ = ["Files", "FilesWithRawResponse"]
-
-
-class Files(BaseAPI):
-    def __init__(self, client: ZhipuAI) -> None:
-        super().__init__(client)
-
-    def create(
-        self,
-        *,
-        file: Optional[FileTypes] = None,
-        upload_detail: Optional[list[UploadDetail]] = None,
-        purpose: Literal["fine-tune", "retrieval", "batch"],
-        knowledge_id: Optional[str] = None,
-        sentence_size: Optional[int] = None,
-        extra_headers: Headers | None = None,
-        extra_body: Body | None = None,
-        timeout: float | httpx.Timeout | None | NotGiven = NOT_GIVEN,
-    ) -> FileObject:
-        if not file and not upload_detail:
-            raise ValueError("At least one of `file` and `upload_detail` must be provided.")
-        body = deepcopy_minimal(
-            {
-                "file": file,
-                "upload_detail": upload_detail,
-                "purpose": purpose,
-                "knowledge_id": knowledge_id,
-                "sentence_size": sentence_size,
-            }
-        )
-        files = extract_files(cast(Mapping[str, object], body), paths=[["file"]])
-        if files:
-            # It should be noted that the actual Content-Type header that will be
-            # sent to the server will contain a `boundary` parameter, e.g.
-            # multipart/form-data; boundary=---abc--
-            extra_headers = {"Content-Type": "multipart/form-data", **(extra_headers or {})}
-        return self._post(
-            "/files",
-            body=maybe_transform(body, file_create_params.FileCreateParams),
-            files=files,
-            options=make_request_options(extra_headers=extra_headers, extra_body=extra_body, timeout=timeout),
-            cast_type=FileObject,
-        )
-
-    # def retrieve(
-    #         self,
-    #         file_id: str,
-    #         *,
-    #         extra_headers: Headers | None = None,
-    #         extra_body: Body | None = None,
-    #         timeout: float | httpx.Timeout | None | NotGiven = NOT_GIVEN,
-    # ) -> FileObject:
-    #     """
-    #     Returns information about a specific file.
-    #
-    #     Args:
-    #       file_id: The ID of the file to retrieve information about
-    #       extra_headers: Send extra headers
-    #
-    #       extra_body: Add additional JSON properties to the request
-    #
-    #       timeout: Override the client-level default timeout for this request, in seconds
-    #     """
-    #     if not file_id:
-    #         raise ValueError(f"Expected a non-empty value for `file_id` but received {file_id!r}")
-    #     return self._get(
-    #         f"/files/{file_id}",
-    #         options=make_request_options(
-    #             extra_headers=extra_headers, extra_body=extra_body, timeout=timeout
-    #         ),
-    #         cast_type=FileObject,
-    #     )
-
-    def list(
-        self,
-        *,
-        purpose: str | NotGiven = NOT_GIVEN,
-        limit: int | NotGiven = NOT_GIVEN,
-        after: str | NotGiven = NOT_GIVEN,
-        order: str | NotGiven = NOT_GIVEN,
-        extra_headers: Headers | None = None,
-        extra_body: Body | None = None,
-        timeout: float | httpx.Timeout | None | NotGiven = NOT_GIVEN,
-    ) -> ListOfFileObject:
-        return self._get(
-            "/files",
-            cast_type=ListOfFileObject,
-            options=make_request_options(
-                extra_headers=extra_headers,
-                extra_body=extra_body,
-                timeout=timeout,
-                query={
-                    "purpose": purpose,
-                    "limit": limit,
-                    "after": after,
-                    "order": order,
-                },
-            ),
-        )
-
-    def delete(
-        self,
-        file_id: str,
-        *,
-        extra_headers: Headers | None = None,
-        extra_body: Body | None = None,
-        timeout: float | httpx.Timeout | None | NotGiven = NOT_GIVEN,
-    ) -> FileDeleted:
-        """
-        Delete a file.
-
-        Args:
-          file_id: The ID of the file to delete
-          extra_headers: Send extra headers
-
-          extra_body: Add additional JSON properties to the request
-
-          timeout: Override the client-level default timeout for this request, in seconds
-        """
-        if not file_id:
-            raise ValueError(f"Expected a non-empty value for `file_id` but received {file_id!r}")
-        return self._delete(
-            f"/files/{file_id}",
-            options=make_request_options(extra_headers=extra_headers, extra_body=extra_body, timeout=timeout),
-            cast_type=FileDeleted,
-        )
-
-    def content(
-        self,
-        file_id: str,
-        *,
-        extra_headers: Headers | None = None,
-        extra_body: Body | None = None,
-        timeout: float | httpx.Timeout | None | NotGiven = NOT_GIVEN,
-    ) -> _legacy_response.HttpxBinaryResponseContent:
-        """
-        Returns the contents of the specified file.
-
-        Args:
-          extra_headers: Send extra headers
-
-          extra_body: Add additional JSON properties to the request
-
-          timeout: Override the client-level default timeout for this request, in seconds
-        """
-        if not file_id:
-            raise ValueError(f"Expected a non-empty value for `file_id` but received {file_id!r}")
-        extra_headers = {"Accept": "application/binary", **(extra_headers or {})}
-        return self._get(
-            f"/files/{file_id}/content",
-            options=make_request_options(extra_headers=extra_headers, extra_body=extra_body, timeout=timeout),
-            cast_type=_legacy_binary_response.HttpxBinaryResponseContent,
-        )
-
-
-class FilesWithRawResponse:
-    def __init__(self, files: Files) -> None:
-        self._files = files
-
-        self.create = _legacy_response.to_raw_response_wrapper(
-            files.create,
-        )
-        self.list = _legacy_response.to_raw_response_wrapper(
-            files.list,
-        )
-        self.content = _legacy_response.to_raw_response_wrapper(
-            files.content,
-        )
diff --git a/api/core/model_runtime/model_providers/zhipuai/zhipuai_sdk/api_resource/fine_tuning/__init__.py b/api/core/model_runtime/model_providers/zhipuai/zhipuai_sdk/api_resource/fine_tuning/__init__.py
deleted file mode 100644
index 7c309b8341680..0000000000000
--- a/api/core/model_runtime/model_providers/zhipuai/zhipuai_sdk/api_resource/fine_tuning/__init__.py
+++ /dev/null
@@ -1,5 +0,0 @@
-from .fine_tuning import FineTuning
-from .jobs import Jobs
-from .models import FineTunedModels
-
-__all__ = ["Jobs", "FineTunedModels", "FineTuning"]
diff --git a/api/core/model_runtime/model_providers/zhipuai/zhipuai_sdk/api_resource/fine_tuning/fine_tuning.py b/api/core/model_runtime/model_providers/zhipuai/zhipuai_sdk/api_resource/fine_tuning/fine_tuning.py
deleted file mode 100644
index 8670f7de00df8..0000000000000
--- a/api/core/model_runtime/model_providers/zhipuai/zhipuai_sdk/api_resource/fine_tuning/fine_tuning.py
+++ /dev/null
@@ -1,18 +0,0 @@
-from typing import TYPE_CHECKING
-
-from ...core import BaseAPI, cached_property
-from .jobs import Jobs
-from .models import FineTunedModels
-
-if TYPE_CHECKING:
-    pass
-
-
-class FineTuning(BaseAPI):
-    @cached_property
-    def jobs(self) -> Jobs:
-        return Jobs(self._client)
-
-    @cached_property
-    def models(self) -> FineTunedModels:
-        return FineTunedModels(self._client)
diff --git a/api/core/model_runtime/model_providers/zhipuai/zhipuai_sdk/api_resource/fine_tuning/jobs/__init__.py b/api/core/model_runtime/model_providers/zhipuai/zhipuai_sdk/api_resource/fine_tuning/jobs/__init__.py
deleted file mode 100644
index 40777a153f272..0000000000000
--- a/api/core/model_runtime/model_providers/zhipuai/zhipuai_sdk/api_resource/fine_tuning/jobs/__init__.py
+++ /dev/null
@@ -1,3 +0,0 @@
-from .jobs import Jobs
-
-__all__ = ["Jobs"]
diff --git a/api/core/model_runtime/model_providers/zhipuai/zhipuai_sdk/api_resource/fine_tuning/jobs/jobs.py b/api/core/model_runtime/model_providers/zhipuai/zhipuai_sdk/api_resource/fine_tuning/jobs/jobs.py
deleted file mode 100644
index 8b038cadc0640..0000000000000
--- a/api/core/model_runtime/model_providers/zhipuai/zhipuai_sdk/api_resource/fine_tuning/jobs/jobs.py
+++ /dev/null
@@ -1,152 +0,0 @@
-from __future__ import annotations
-
-from typing import TYPE_CHECKING, Optional
-
-import httpx
-
-from ....core import (
-    NOT_GIVEN,
-    BaseAPI,
-    Body,
-    Headers,
-    NotGiven,
-    make_request_options,
-)
-from ....types.fine_tuning import (
-    FineTuningJob,
-    FineTuningJobEvent,
-    ListOfFineTuningJob,
-    job_create_params,
-)
-
-if TYPE_CHECKING:
-    from ...._client import ZhipuAI
-
-__all__ = ["Jobs"]
-
-
-class Jobs(BaseAPI):
-    def __init__(self, client: ZhipuAI) -> None:
-        super().__init__(client)
-
-    def create(
-        self,
-        *,
-        model: str,
-        training_file: str,
-        hyperparameters: job_create_params.Hyperparameters | NotGiven = NOT_GIVEN,
-        suffix: Optional[str] | NotGiven = NOT_GIVEN,
-        request_id: Optional[str] | NotGiven = NOT_GIVEN,
-        validation_file: Optional[str] | NotGiven = NOT_GIVEN,
-        extra_headers: Headers | None = None,
-        extra_body: Body | None = None,
-        timeout: float | httpx.Timeout | None | NotGiven = NOT_GIVEN,
-    ) -> FineTuningJob:
-        return self._post(
-            "/fine_tuning/jobs",
-            body={
-                "model": model,
-                "training_file": training_file,
-                "hyperparameters": hyperparameters,
-                "suffix": suffix,
-                "validation_file": validation_file,
-                "request_id": request_id,
-            },
-            options=make_request_options(extra_headers=extra_headers, extra_body=extra_body, timeout=timeout),
-            cast_type=FineTuningJob,
-        )
-
-    def retrieve(
-        self,
-        fine_tuning_job_id: str,
-        *,
-        extra_headers: Headers | None = None,
-        extra_body: Body | None = None,
-        timeout: float | httpx.Timeout | None | NotGiven = NOT_GIVEN,
-    ) -> FineTuningJob:
-        return self._get(
-            f"/fine_tuning/jobs/{fine_tuning_job_id}",
-            options=make_request_options(extra_headers=extra_headers, extra_body=extra_body, timeout=timeout),
-            cast_type=FineTuningJob,
-        )
-
-    def list(
-        self,
-        *,
-        after: str | NotGiven = NOT_GIVEN,
-        limit: int | NotGiven = NOT_GIVEN,
-        extra_headers: Headers | None = None,
-        extra_body: Body | None = None,
-        timeout: float | httpx.Timeout | None | NotGiven = NOT_GIVEN,
-    ) -> ListOfFineTuningJob:
-        return self._get(
-            "/fine_tuning/jobs",
-            cast_type=ListOfFineTuningJob,
-            options=make_request_options(
-                extra_headers=extra_headers,
-                extra_body=extra_body,
-                timeout=timeout,
-                query={
-                    "after": after,
-                    "limit": limit,
-                },
-            ),
-        )
-
-    def cancel(
-        self,
-        fine_tuning_job_id: str,
-        *,
-        # Use the following arguments if you need to pass additional parameters to the API that aren't available via kwargs.  # noqa: E501
-        # The extra values given here take precedence over values defined on the client or passed to this method.
-        extra_headers: Headers | None = None,
-        extra_body: Body | None = None,
-        timeout: float | httpx.Timeout | None | NotGiven = NOT_GIVEN,
-    ) -> FineTuningJob:
-        if not fine_tuning_job_id:
-            raise ValueError(f"Expected a non-empty value for `fine_tuning_job_id` but received {fine_tuning_job_id!r}")
-        return self._post(
-            f"/fine_tuning/jobs/{fine_tuning_job_id}/cancel",
-            options=make_request_options(extra_headers=extra_headers, extra_body=extra_body, timeout=timeout),
-            cast_type=FineTuningJob,
-        )
-
-    def list_events(
-        self,
-        fine_tuning_job_id: str,
-        *,
-        after: str | NotGiven = NOT_GIVEN,
-        limit: int | NotGiven = NOT_GIVEN,
-        extra_headers: Headers | None = None,
-        extra_body: Body | None = None,
-        timeout: float | httpx.Timeout | None | NotGiven = NOT_GIVEN,
-    ) -> FineTuningJobEvent:
-        return self._get(
-            f"/fine_tuning/jobs/{fine_tuning_job_id}/events",
-            cast_type=FineTuningJobEvent,
-            options=make_request_options(
-                extra_headers=extra_headers,
-                extra_body=extra_body,
-                timeout=timeout,
-                query={
-                    "after": after,
-                    "limit": limit,
-                },
-            ),
-        )
-
-    def delete(
-        self,
-        fine_tuning_job_id: str,
-        *,
-        extra_headers: Headers | None = None,
-        extra_body: Body | None = None,
-        timeout: float | httpx.Timeout | None | NotGiven = NOT_GIVEN,
-    ) -> FineTuningJob:
-        if not fine_tuning_job_id:
-            raise ValueError(f"Expected a non-empty value for `fine_tuning_job_id` but received {fine_tuning_job_id!r}")
-        return self._delete(
-            f"/fine_tuning/jobs/{fine_tuning_job_id}",
-            options=make_request_options(extra_headers=extra_headers, extra_body=extra_body, timeout=timeout),
-            cast_type=FineTuningJob,
-        )
diff --git a/api/core/model_runtime/model_providers/zhipuai/zhipuai_sdk/api_resource/fine_tuning/models/__init__.py b/api/core/model_runtime/model_providers/zhipuai/zhipuai_sdk/api_resource/fine_tuning/models/__init__.py
deleted file mode 100644
index d832635bafbc6..0000000000000
--- a/api/core/model_runtime/model_providers/zhipuai/zhipuai_sdk/api_resource/fine_tuning/models/__init__.py
+++ /dev/null
@@ -1,3 +0,0 @@
-from .fine_tuned_models import FineTunedModels
-
-__all__ = ["FineTunedModels"]
diff --git a/api/core/model_runtime/model_providers/zhipuai/zhipuai_sdk/api_resource/fine_tuning/models/fine_tuned_models.py b/api/core/model_runtime/model_providers/zhipuai/zhipuai_sdk/api_resource/fine_tuning/models/fine_tuned_models.py
deleted file mode 100644
index 29c023e3b1cd5..0000000000000
--- a/api/core/model_runtime/model_providers/zhipuai/zhipuai_sdk/api_resource/fine_tuning/models/fine_tuned_models.py
+++ /dev/null
@@ -1,41 +0,0 @@
-from __future__ import annotations
-
-from typing import TYPE_CHECKING
-
-import httpx
-
-from ....core import (
-    NOT_GIVEN,
-    BaseAPI,
-    Body,
-    Headers,
-    NotGiven,
-    make_request_options,
-)
-from ....types.fine_tuning.models import FineTunedModelsStatus
-
-if TYPE_CHECKING:
-    from ...._client import ZhipuAI
-
-__all__ = ["FineTunedModels"]
-
-
-class FineTunedModels(BaseAPI):
-    def __init__(self, client: ZhipuAI) -> None:
-        super().__init__(client)
-
-    def delete(
-        self,
-        fine_tuned_model: str,
-        *,
-        extra_headers: Headers | None = None,
-        extra_body: Body | None = None,
-        timeout: float | httpx.Timeout | None | NotGiven = NOT_GIVEN,
-    ) -> FineTunedModelsStatus:
-        if not fine_tuned_model:
-            raise ValueError(f"Expected a non-empty value for `fine_tuned_model` but received {fine_tuned_model!r}")
-        return self._delete(
-            f"fine_tuning/fine_tuned_models/{fine_tuned_model}",
-            options=make_request_options(extra_headers=extra_headers, extra_body=extra_body, timeout=timeout),
-            cast_type=FineTunedModelsStatus,
-        )
diff --git a/api/core/model_runtime/model_providers/zhipuai/zhipuai_sdk/api_resource/images.py b/api/core/model_runtime/model_providers/zhipuai/zhipuai_sdk/api_resource/images.py
deleted file mode 100644
index 8ad411913fa11..0000000000000
--- a/api/core/model_runtime/model_providers/zhipuai/zhipuai_sdk/api_resource/images.py
+++ /dev/null
@@ -1,59 +0,0 @@
-from __future__ import annotations
-
-from typing import TYPE_CHECKING, Optional
-
-import httpx
-
-from ..core import NOT_GIVEN, BaseAPI, Body, Headers, NotGiven, make_request_options
-from ..types.image import ImagesResponded
-from ..types.sensitive_word_check import SensitiveWordCheckRequest
-
-if TYPE_CHECKING:
-    from .._client import ZhipuAI
-
-
-class Images(BaseAPI):
-    def __init__(self, client: ZhipuAI) -> None:
-        super().__init__(client)
-
-    def generations(
-        self,
-        *,
-        prompt: str,
-        model: str | NotGiven = NOT_GIVEN,
-        n: Optional[int] | NotGiven = NOT_GIVEN,
-        quality: Optional[str] | NotGiven = NOT_GIVEN,
-        response_format: Optional[str] | NotGiven = NOT_GIVEN,
-        size: Optional[str] | NotGiven = NOT_GIVEN,
-        style: Optional[str] | NotGiven = NOT_GIVEN,
-        sensitive_word_check: Optional[SensitiveWordCheckRequest] | NotGiven = NOT_GIVEN,
-        user: str | NotGiven = NOT_GIVEN,
-        request_id: Optional[str] | NotGiven = NOT_GIVEN,
-        user_id: Optional[str] | NotGiven = NOT_GIVEN,
-        extra_headers: Headers | None = None,
-        extra_body: Body | None = None,
-        disable_strict_validation: Optional[bool] | None = None,
-        timeout: float | httpx.Timeout | None | NotGiven = NOT_GIVEN,
-    ) -> ImagesResponded:
-        _cast_type = ImagesResponded
-        if disable_strict_validation:
-            _cast_type = object
-        return self._post(
-            "/images/generations",
-            body={
-                "prompt": prompt,
-                "model": model,
-                "n": n,
-                "quality": quality,
-                "response_format": response_format,
-                "sensitive_word_check": sensitive_word_check,
-                "size": size,
-                "style": style,
-                "user": user,
-                "user_id": user_id,
-                "request_id": request_id,
-            },
-            options=make_request_options(extra_headers=extra_headers, extra_body=extra_body, timeout=timeout),
-            cast_type=_cast_type,
-            stream=False,
-        )
diff --git a/api/core/model_runtime/model_providers/zhipuai/zhipuai_sdk/api_resource/knowledge/__init__.py b/api/core/model_runtime/model_providers/zhipuai/zhipuai_sdk/api_resource/knowledge/__init__.py
deleted file mode 100644
index 5a67d743c35b9..0000000000000
--- a/api/core/model_runtime/model_providers/zhipuai/zhipuai_sdk/api_resource/knowledge/__init__.py
+++ /dev/null
@@ -1,3 +0,0 @@
-from .knowledge import Knowledge
-
-__all__ = ["Knowledge"]
diff --git a/api/core/model_runtime/model_providers/zhipuai/zhipuai_sdk/api_resource/knowledge/document/__init__.py b/api/core/model_runtime/model_providers/zhipuai/zhipuai_sdk/api_resource/knowledge/document/__init__.py
deleted file mode 100644
index fd289e2232b95..0000000000000
--- a/api/core/model_runtime/model_providers/zhipuai/zhipuai_sdk/api_resource/knowledge/document/__init__.py
+++ /dev/null
@@ -1,3 +0,0 @@
-from .document import Document
-
-__all__ = ["Document"]
diff --git a/api/core/model_runtime/model_providers/zhipuai/zhipuai_sdk/api_resource/knowledge/document/document.py b/api/core/model_runtime/model_providers/zhipuai/zhipuai_sdk/api_resource/knowledge/document/document.py
deleted file mode 100644
index 492c49da6636c..0000000000000
--- a/api/core/model_runtime/model_providers/zhipuai/zhipuai_sdk/api_resource/knowledge/document/document.py
+++ /dev/null
@@ -1,217 +0,0 @@
-from __future__ import annotations
-
-from collections.abc import Mapping
-from typing import TYPE_CHECKING, Literal, Optional, cast
-
-import httpx
-
-from ....core import (
-    NOT_GIVEN,
-    BaseAPI,
-    Body,
-    FileTypes,
-    Headers,
-    NotGiven,
-    deepcopy_minimal,
-    extract_files,
-    make_request_options,
-    maybe_transform,
-)
-from ....types.files import UploadDetail, file_create_params
-from ....types.knowledge.document import DocumentData, DocumentObject, document_edit_params, document_list_params
-from ....types.knowledge.document.document_list_resp import DocumentPage
-
-if TYPE_CHECKING:
-    from ...._client import ZhipuAI
-
-__all__ = ["Document"]
-
-
-class Document(BaseAPI):
-    def __init__(self, client: ZhipuAI) -> None:
-        super().__init__(client)
-
-    def create(
-        self,
-        *,
-        file: Optional[FileTypes] = None,
-        custom_separator: Optional[list[str]] = None,
-        upload_detail: Optional[list[UploadDetail]] = None,
-        purpose: Literal["retrieval"],
-        knowledge_id: Optional[str] = None,
-        sentence_size: Optional[int] = None,
-        extra_headers: Headers | None = None,
-        extra_body: Body | None = None,
-        timeout: float | httpx.Timeout | None | NotGiven = NOT_GIVEN,
-    ) -> DocumentObject:
-        if not file and not upload_detail:
-            raise ValueError("At least one of `file` and `upload_detail` must be provided.")
-        body = deepcopy_minimal(
-            {
-                "file": file,
-                "upload_detail": upload_detail,
-                "purpose": purpose,
-                "custom_separator": custom_separator,
-                "knowledge_id": knowledge_id,
-                "sentence_size": sentence_size,
-            }
-        )
-        files = extract_files(cast(Mapping[str, object], body), paths=[["file"]])
-        if files:
-            # It should be noted that the actual Content-Type header that will be
-            # sent to the server will contain a `boundary` parameter, e.g.
-            # multipart/form-data; boundary=---abc--
-            extra_headers = {"Content-Type": "multipart/form-data", **(extra_headers or {})}
-        return self._post(
-            "/files",
-            body=maybe_transform(body, file_create_params.FileCreateParams),
-            files=files,
-            options=make_request_options(extra_headers=extra_headers, extra_body=extra_body, timeout=timeout),
-            cast_type=DocumentObject,
-        )
-
-    def edit(
-        self,
-        document_id: str,
-        knowledge_type: str,
-        *,
-        custom_separator: Optional[list[str]] = None,
-        sentence_size: Optional[int] = None,
-        callback_url: Optional[str] = None,
-        callback_header: Optional[dict[str, str]] = None,
-        extra_headers: Headers | None = None,
-        extra_body: Body | None = None,
-        timeout: float | httpx.Timeout | None | NotGiven = NOT_GIVEN,
-    ) -> httpx.Response:
-        """
-
-        Args:
-          document_id: 知识id
-          knowledge_type: 知识类型:
-                        1:文章知识: 支持pdf,url,docx
-                        2.问答知识-文档:  支持pdf,url,docx
-                        3.问答知识-表格:  支持xlsx
-                        4.商品库-表格:  支持xlsx
-                        5.自定义:  支持pdf,url,docx
-          extra_headers: Send extra headers
-
-          extra_body: Add additional JSON properties to the request
-
-          timeout: Override the client-level default timeout for this request, in seconds
-          :param knowledge_type:
-          :param document_id:
-          :param timeout:
-          :param extra_body:
-          :param callback_header:
-          :param sentence_size:
-          :param extra_headers:
-          :param callback_url:
-          :param custom_separator:
-        """
-        if not document_id:
-            raise ValueError(f"Expected a non-empty value for `document_id` but received {document_id!r}")
-
-        body = deepcopy_minimal(
-            {
-                "id": document_id,
-                "knowledge_type": knowledge_type,
-                "custom_separator": custom_separator,
-                "sentence_size": sentence_size,
-                "callback_url": callback_url,
-                "callback_header": callback_header,
-            }
-        )
-
-        return self._put(
-            f"/document/{document_id}",
-            body=maybe_transform(body, document_edit_params.DocumentEditParams),
-            options=make_request_options(extra_headers=extra_headers, extra_body=extra_body, timeout=timeout),
-            cast_type=httpx.Response,
-        )
-
-    def list(
-        self,
-        knowledge_id: str,
-        *,
-        purpose: str | NotGiven = NOT_GIVEN,
-        page: str | NotGiven = NOT_GIVEN,
-        limit: str | NotGiven = NOT_GIVEN,
-        order: Literal["desc", "asc"] | NotGiven = NOT_GIVEN,
-        extra_headers: Headers | None = None,
-        extra_body: Body | None = None,
-        timeout: float | httpx.Timeout | None | NotGiven = NOT_GIVEN,
-    ) -> DocumentPage:
-        return self._get(
-            "/files",
-            options=make_request_options(
-                extra_headers=extra_headers,
-                extra_body=extra_body,
-                timeout=timeout,
-                query=maybe_transform(
-                    {
-                        "knowledge_id": knowledge_id,
-                        "purpose": purpose,
-                        "page": page,
-                        "limit": limit,
-                        "order": order,
-                    },
-                    document_list_params.DocumentListParams,
-                ),
-            ),
-            cast_type=DocumentPage,
-        )
-
-    def delete(
-        self,
-        document_id: str,
-        *,
-        extra_headers: Headers | None = None,
-        extra_body: Body | None = None,
-        timeout: float | httpx.Timeout | None | NotGiven = NOT_GIVEN,
-    ) -> httpx.Response:
-        """
-        Delete a file.
-
-        Args:
-
-          document_id: 知识id
-          extra_headers: Send extra headers
-
-          extra_body: Add additional JSON properties to the request
-
-          timeout: Override the client-level default timeout for this request, in seconds
-        """
-        if not document_id:
-            raise ValueError(f"Expected a non-empty value for `document_id` but received {document_id!r}")
-
-        return self._delete(
-            f"/document/{document_id}",
-            options=make_request_options(extra_headers=extra_headers, extra_body=extra_body, timeout=timeout),
-            cast_type=httpx.Response,
-        )
-
-    def retrieve(
-        self,
-        document_id: str,
-        *,
-        extra_headers: Headers | None = None,
-        extra_body: Body | None = None,
-        timeout: float | httpx.Timeout | None | NotGiven = NOT_GIVEN,
-    ) -> DocumentData:
-        """
-
-        Args:
-          extra_headers: Send extra headers
-
-          extra_body: Add additional JSON properties to the request
-
-          timeout: Override the client-level default timeout for this request, in seconds
-        """
-        if not document_id:
-            raise ValueError(f"Expected a non-empty value for `document_id` but received {document_id!r}")
-
-        return self._get(
-            f"/document/{document_id}",
-            options=make_request_options(extra_headers=extra_headers, extra_body=extra_body, timeout=timeout),
-            cast_type=DocumentData,
-        )
diff --git a/api/core/model_runtime/model_providers/zhipuai/zhipuai_sdk/api_resource/knowledge/knowledge.py b/api/core/model_runtime/model_providers/zhipuai/zhipuai_sdk/api_resource/knowledge/knowledge.py
deleted file mode 100644
index fea4c73ac997c..0000000000000
--- a/api/core/model_runtime/model_providers/zhipuai/zhipuai_sdk/api_resource/knowledge/knowledge.py
+++ /dev/null
@@ -1,173 +0,0 @@
-from __future__ import annotations
-
-from typing import TYPE_CHECKING, Literal, Optional
-
-import httpx
-
-from ...core import (
-    NOT_GIVEN,
-    BaseAPI,
-    Body,
-    Headers,
-    NotGiven,
-    cached_property,
-    deepcopy_minimal,
-    make_request_options,
-    maybe_transform,
-)
-from ...types.knowledge import KnowledgeInfo, KnowledgeUsed, knowledge_create_params, knowledge_list_params
-from ...types.knowledge.knowledge_list_resp import KnowledgePage
-from .document import Document
-
-if TYPE_CHECKING:
-    from ..._client import ZhipuAI
-
-__all__ = ["Knowledge"]
-
-
-class Knowledge(BaseAPI):
-    def __init__(self, client: ZhipuAI) -> None:
-        super().__init__(client)
-
-    @cached_property
-    def document(self) -> Document:
-        return Document(self._client)
-
-    def create(
-        self,
-        embedding_id: int,
-        name: str,
-        *,
-        customer_identifier: Optional[str] = None,
-        description: Optional[str] = None,
-        background: Optional[Literal["blue", "red", "orange", "purple", "sky"]] = None,
-        icon: Optional[Literal["question", "book", "seal", "wrench", "tag", "horn", "house"]] = None,
-        bucket_id: Optional[str] = None,
-        extra_headers: Headers | None = None,
-        extra_body: Body | None = None,
-        timeout: float | httpx.Timeout | None | NotGiven = NOT_GIVEN,
-    ) -> KnowledgeInfo:
-        body = deepcopy_minimal(
-            {
-                "embedding_id": embedding_id,
-                "name": name,
-                "customer_identifier": customer_identifier,
-                "description": description,
-                "background": background,
-                "icon": icon,
-                "bucket_id": bucket_id,
-            }
-        )
-        return self._post(
-            "/knowledge",
-            body=maybe_transform(body, knowledge_create_params.KnowledgeBaseParams),
-            options=make_request_options(extra_headers=extra_headers, extra_body=extra_body, timeout=timeout),
-            cast_type=KnowledgeInfo,
-        )
-
-    def modify(
-        self,
-        knowledge_id: str,
-        embedding_id: int,
-        *,
-        name: str,
-        description: Optional[str] = None,
-        background: Optional[Literal["blue", "red", "orange", "purple", "sky"]] = None,
-        icon: Optional[Literal["question", "book", "seal", "wrench", "tag", "horn", "house"]] = None,
-        extra_headers: Headers | None = None,
-        extra_body: Body | None = None,
-        timeout: float | httpx.Timeout | None | NotGiven = NOT_GIVEN,
-    ) -> httpx.Response:
-        body = deepcopy_minimal(
-            {
-                "id": knowledge_id,
-                "embedding_id": embedding_id,
-                "name": name,
-                "description": description,
-                "background": background,
-                "icon": icon,
-            }
-        )
-        return self._put(
-            f"/knowledge/{knowledge_id}",
-            body=maybe_transform(body, knowledge_create_params.KnowledgeBaseParams),
-            options=make_request_options(extra_headers=extra_headers, extra_body=extra_body, timeout=timeout),
-            cast_type=httpx.Response,
-        )
-
-    def query(
-        self,
-        *,
-        page: int | NotGiven = 1,
-        size: int | NotGiven = 10,
-        extra_headers: Headers | None = None,
-        extra_body: Body | None = None,
-        timeout: float | httpx.Timeout | None | NotGiven = NOT_GIVEN,
-    ) -> KnowledgePage:
-        return self._get(
-            "/knowledge",
-            options=make_request_options(
-                extra_headers=extra_headers,
-                extra_body=extra_body,
-                timeout=timeout,
-                query=maybe_transform(
-                    {
-                        "page": page,
-                        "size": size,
-                    },
-                    knowledge_list_params.KnowledgeListParams,
-                ),
-            ),
-            cast_type=KnowledgePage,
-        )
-
-    def delete(
-        self,
-        knowledge_id: str,
-        *,
-        extra_headers: Headers | None = None,
-        extra_body: Body | None = None,
-        timeout: float | httpx.Timeout | None | NotGiven = NOT_GIVEN,
-    ) -> httpx.Response:
-        """
-        Delete a file.
-
-        Args:
-          knowledge_id: 知识库ID
-          extra_headers: Send extra headers
-
-          extra_body: Add additional JSON properties to the request
-
-          timeout: Override the client-level default timeout for this request, in seconds
-        """
-        if not knowledge_id:
-            raise ValueError("Expected a non-empty value for `knowledge_id`")
-
-        return self._delete(
-            f"/knowledge/{knowledge_id}",
-            options=make_request_options(extra_headers=extra_headers, extra_body=extra_body, timeout=timeout),
-            cast_type=httpx.Response,
-        )
-
-    def used(
-        self,
-        *,
-        extra_headers: Headers | None = None,
-        extra_body: Body | None = None,
-        timeout: float | httpx.Timeout | None | NotGiven = NOT_GIVEN,
-    ) -> KnowledgeUsed:
-        """
-        Returns the contents of the specified file.
-
-        Args:
-          extra_headers: Send extra headers
-
-          extra_body: Add additional JSON properties to the request
-
-          timeout: Override the client-level default timeout for this request, in seconds
-        """
-        return self._get(
-            "/knowledge/capacity",
-            options=make_request_options(extra_headers=extra_headers, extra_body=extra_body, timeout=timeout),
-            cast_type=KnowledgeUsed,
-        )
diff --git a/api/core/model_runtime/model_providers/zhipuai/zhipuai_sdk/api_resource/tools/__init__.py b/api/core/model_runtime/model_providers/zhipuai/zhipuai_sdk/api_resource/tools/__init__.py
deleted file mode 100644
index 43e4e37da1779..0000000000000
--- a/api/core/model_runtime/model_providers/zhipuai/zhipuai_sdk/api_resource/tools/__init__.py
+++ /dev/null
@@ -1,3 +0,0 @@
-from .tools import Tools
-
-__all__ = ["Tools"]
diff --git a/api/core/model_runtime/model_providers/zhipuai/zhipuai_sdk/api_resource/tools/tools.py b/api/core/model_runtime/model_providers/zhipuai/zhipuai_sdk/api_resource/tools/tools.py
deleted file mode 100644
index 3c3a630aff47d..0000000000000
--- a/api/core/model_runtime/model_providers/zhipuai/zhipuai_sdk/api_resource/tools/tools.py
+++ /dev/null
@@ -1,65 +0,0 @@
-from __future__ import annotations
-
-import logging
-from typing import TYPE_CHECKING, Literal, Optional, Union
-
-import httpx
-
-from ...core import (
-    NOT_GIVEN,
-    BaseAPI,
-    Body,
-    Headers,
-    NotGiven,
-    StreamResponse,
-    deepcopy_minimal,
-    make_request_options,
-    maybe_transform,
-)
-from ...types.tools import WebSearch, WebSearchChunk, tools_web_search_params
-
-logger = logging.getLogger(__name__)
-
-if TYPE_CHECKING:
-    from ..._client import ZhipuAI
-
-__all__ = ["Tools"]
-
-
-class Tools(BaseAPI):
-    def __init__(self, client: ZhipuAI) -> None:
-        super().__init__(client)
-
-    def web_search(
-        self,
-        *,
-        model: str,
-        request_id: Optional[str] | NotGiven = NOT_GIVEN,
-        stream: Optional[Literal[False]] | Literal[True] | NotGiven = NOT_GIVEN,
-        messages: Union[str, list[str], list[int], object, None],
-        scope: Optional[str] | NotGiven = NOT_GIVEN,
-        location: Optional[str] | NotGiven = NOT_GIVEN,
-        recent_days: Optional[int] | NotGiven = NOT_GIVEN,
-        extra_headers: Headers | None = None,
-        extra_body: Body | None = None,
-        timeout: float | httpx.Timeout | None | NotGiven = NOT_GIVEN,
-    ) -> WebSearch | StreamResponse[WebSearchChunk]:
-        body = deepcopy_minimal(
-            {
-                "model": model,
-                "request_id": request_id,
-                "messages": messages,
-                "stream": stream,
-                "scope": scope,
-                "location": location,
-                "recent_days": recent_days,
-            }
-        )
-        return self._post(
-            "/tools",
-            body=maybe_transform(body, tools_web_search_params.WebSearchParams),
-            options=make_request_options(extra_headers=extra_headers, extra_body=extra_body, timeout=timeout),
-            cast_type=WebSearch,
-            stream=stream or False,
-            stream_cls=StreamResponse[WebSearchChunk],
-        )
diff --git a/api/core/model_runtime/model_providers/zhipuai/zhipuai_sdk/api_resource/videos/__init__.py b/api/core/model_runtime/model_providers/zhipuai/zhipuai_sdk/api_resource/videos/__init__.py
deleted file mode 100644
index 6b0f99ed09efe..0000000000000
--- a/api/core/model_runtime/model_providers/zhipuai/zhipuai_sdk/api_resource/videos/__init__.py
+++ /dev/null
@@ -1,7 +0,0 @@
-from .videos import (
-    Videos,
-)
-
-__all__ = [
-    "Videos",
-]
diff --git a/api/core/model_runtime/model_providers/zhipuai/zhipuai_sdk/api_resource/videos/videos.py b/api/core/model_runtime/model_providers/zhipuai/zhipuai_sdk/api_resource/videos/videos.py
deleted file mode 100644
index 71c8316602a08..0000000000000
--- a/api/core/model_runtime/model_providers/zhipuai/zhipuai_sdk/api_resource/videos/videos.py
+++ /dev/null
@@ -1,77 +0,0 @@
-from __future__ import annotations
-
-from typing import TYPE_CHECKING, Optional
-
-import httpx
-
-from ...core import (
-    NOT_GIVEN,
-    BaseAPI,
-    Body,
-    Headers,
-    NotGiven,
-    deepcopy_minimal,
-    make_request_options,
-    maybe_transform,
-)
-from ...types.sensitive_word_check import SensitiveWordCheckRequest
-from ...types.video import VideoObject, video_create_params
-
-if TYPE_CHECKING:
-    from ..._client import ZhipuAI
-
-__all__ = ["Videos"]
-
-
-class Videos(BaseAPI):
-    def __init__(self, client: ZhipuAI) -> None:
-        super().__init__(client)
-
-    def generations(
-        self,
-        model: str,
-        *,
-        prompt: Optional[str] = None,
-        image_url: Optional[str] = None,
-        sensitive_word_check: Optional[SensitiveWordCheckRequest] | NotGiven = NOT_GIVEN,
-        request_id: Optional[str] = None,
-        user_id: Optional[str] = None,
-        extra_headers: Headers | None = None,
-        extra_body: Body | None = None,
-        timeout: float | httpx.Timeout | None | NotGiven = NOT_GIVEN,
-    ) -> VideoObject:
-        if not model and not model:
-            raise ValueError("At least one of `model` and `prompt` must be provided.")
-        body = deepcopy_minimal(
-            {
-                "model": model,
-                "prompt": prompt,
-                "image_url": image_url,
-                "sensitive_word_check": sensitive_word_check,
-                "request_id": request_id,
-                "user_id": user_id,
-            }
-        )
-        return self._post(
-            "/videos/generations",
-            body=maybe_transform(body, video_create_params.VideoCreateParams),
-            options=make_request_options(extra_headers=extra_headers, extra_body=extra_body, timeout=timeout),
-            cast_type=VideoObject,
-        )
-
-    def retrieve_videos_result(
-        self,
-        id: str,
-        *,
-        extra_headers: Headers | None = None,
-        extra_body: Body | None = None,
-        timeout: float | httpx.Timeout | None | NotGiven = NOT_GIVEN,
-    ) -> VideoObject:
-        if not id:
-            raise ValueError("At least one of `id` must be provided.")
-
-        return self._get(
-            f"/async-result/{id}",
-            options=make_request_options(extra_headers=extra_headers, extra_body=extra_body, timeout=timeout),
-            cast_type=VideoObject,
-        )
diff --git a/api/core/model_runtime/model_providers/zhipuai/zhipuai_sdk/core/__init__.py b/api/core/model_runtime/model_providers/zhipuai/zhipuai_sdk/core/__init__.py
deleted file mode 100644
index 3d6466d279861..0000000000000
--- a/api/core/model_runtime/model_providers/zhipuai/zhipuai_sdk/core/__init__.py
+++ /dev/null
@@ -1,108 +0,0 @@
-from ._base_api import BaseAPI
-from ._base_compat import (
-    PYDANTIC_V2,
-    ConfigDict,
-    GenericModel,
-    cached_property,
-    field_get_default,
-    get_args,
-    get_model_config,
-    get_model_fields,
-    get_origin,
-    is_literal_type,
-    is_union,
-    parse_obj,
-)
-from ._base_models import BaseModel, construct_type
-from ._base_type import (
-    NOT_GIVEN,
-    Body,
-    FileTypes,
-    Headers,
-    IncEx,
-    ModelT,
-    NotGiven,
-    Query,
-)
-from ._constants import (
-    ZHIPUAI_DEFAULT_LIMITS,
-    ZHIPUAI_DEFAULT_MAX_RETRIES,
-    ZHIPUAI_DEFAULT_TIMEOUT,
-)
-from ._errors import (
-    APIAuthenticationError,
-    APIConnectionError,
-    APIInternalError,
-    APIReachLimitError,
-    APIRequestFailedError,
-    APIResponseError,
-    APIResponseValidationError,
-    APIServerFlowExceedError,
-    APIStatusError,
-    APITimeoutError,
-    ZhipuAIError,
-)
-from ._files import is_file_content
-from ._http_client import HttpClient, make_request_options
-from ._sse_client import StreamResponse
-from ._utils import (
-    deepcopy_minimal,
-    drop_prefix_image_data,
-    extract_files,
-    is_given,
-    is_list,
-    is_mapping,
-    maybe_transform,
-    parse_date,
-    parse_datetime,
-)
-
-__all__ = [
-    "BaseModel",
-    "construct_type",
-    "BaseAPI",
-    "NOT_GIVEN",
-    "Headers",
-    "NotGiven",
-    "Body",
-    "IncEx",
-    "ModelT",
-    "Query",
-    "FileTypes",
-    "PYDANTIC_V2",
-    "ConfigDict",
-    "GenericModel",
-    "get_args",
-    "is_union",
-    "parse_obj",
-    "get_origin",
-    "is_literal_type",
-    "get_model_config",
-    "get_model_fields",
-    "field_get_default",
-    "is_file_content",
-    "ZhipuAIError",
-    "APIStatusError",
-    "APIRequestFailedError",
-    "APIAuthenticationError",
-    "APIReachLimitError",
-    "APIInternalError",
-    "APIServerFlowExceedError",
-    "APIResponseError",
-    "APIResponseValidationError",
-    "APITimeoutError",
-    "make_request_options",
-    "HttpClient",
-    "ZHIPUAI_DEFAULT_TIMEOUT",
-    "ZHIPUAI_DEFAULT_MAX_RETRIES",
-    "ZHIPUAI_DEFAULT_LIMITS",
-    "is_list",
-    "is_mapping",
-    "parse_date",
-    "parse_datetime",
-    "is_given",
-    "maybe_transform",
-    "deepcopy_minimal",
-    "extract_files",
-    "StreamResponse",
-]
diff --git a/api/core/model_runtime/model_providers/zhipuai/zhipuai_sdk/core/_base_api.py b/api/core/model_runtime/model_providers/zhipuai/zhipuai_sdk/core/_base_api.py
deleted file mode 100644
index 3592ea6bacd17..0000000000000
--- a/api/core/model_runtime/model_providers/zhipuai/zhipuai_sdk/core/_base_api.py
+++ /dev/null
@@ -1,19 +0,0 @@
-from __future__ import annotations
-
-from typing import TYPE_CHECKING
-
-if TYPE_CHECKING:
-    from .._client import ZhipuAI
-
-
-class BaseAPI:
-    _client: ZhipuAI
-
-    def __init__(self, client: ZhipuAI) -> None:
-        self._client = client
-        self._delete = client.delete
-        self._get = client.get
-        self._post = client.post
-        self._put = client.put
-        self._patch = client.patch
-        self._get_api_list = client.get_api_list
diff --git a/api/core/model_runtime/model_providers/zhipuai/zhipuai_sdk/core/_base_compat.py b/api/core/model_runtime/model_providers/zhipuai/zhipuai_sdk/core/_base_compat.py
deleted file mode 100644
index 92a5d683be673..0000000000000
--- a/api/core/model_runtime/model_providers/zhipuai/zhipuai_sdk/core/_base_compat.py
+++ /dev/null
@@ -1,209 +0,0 @@
-from __future__ import annotations
-
-from collections.abc import Callable
-from datetime import date, datetime
-from typing import TYPE_CHECKING, Any, Generic, TypeVar, Union, cast, overload
-
-import pydantic
-from pydantic.fields import FieldInfo
-from typing_extensions import Self
-
-from ._base_type import StrBytesIntFloat
-
-_T = TypeVar("_T")
-_ModelT = TypeVar("_ModelT", bound=pydantic.BaseModel)
-
-# --------------- Pydantic v2 compatibility ---------------
-
-# Pyright incorrectly reports some of our functions as overriding a method when they don't
-# pyright: reportIncompatibleMethodOverride=false
-
-PYDANTIC_V2 = pydantic.VERSION.startswith("2.")
-
-# v1 re-exports
-if TYPE_CHECKING:
-
-    def parse_date(value: date | StrBytesIntFloat) -> date: ...
-
-    def parse_datetime(value: Union[datetime, StrBytesIntFloat]) -> datetime: ...
-
-    def get_args(t: type[Any]) -> tuple[Any, ...]: ...
-
-    def is_union(tp: type[Any] | None) -> bool: ...
-
-    def get_origin(t: type[Any]) -> type[Any] | None: ...
-
-    def is_literal_type(type_: type[Any]) -> bool: ...
-
-    def is_typeddict(type_: type[Any]) -> bool: ...
-
-else:
-    if PYDANTIC_V2:
-        from pydantic.v1.typing import (  # noqa: I001
-            get_args as get_args,  # noqa: PLC0414
-            is_union as is_union,  # noqa: PLC0414
-            get_origin as get_origin,  # noqa: PLC0414
-            is_typeddict as is_typeddict,  # noqa: PLC0414
-            is_literal_type as is_literal_type,  # noqa: PLC0414
-        )
-        from pydantic.v1.datetime_parse import parse_date as parse_date, parse_datetime as parse_datetime  # noqa: PLC0414
-    else:
-        from pydantic.typing import (  # noqa: I001
-            get_args as get_args,  # noqa: PLC0414
-            is_union as is_union,  # noqa: PLC0414
-            get_origin as get_origin,  # noqa: PLC0414
-            is_typeddict as is_typeddict,  # noqa: PLC0414
-            is_literal_type as is_literal_type,  # noqa: PLC0414
-        )
-        from pydantic.datetime_parse import parse_date as parse_date, parse_datetime as parse_datetime  # noqa: PLC0414
-
-
-# refactored config
-if TYPE_CHECKING:
-    from pydantic import ConfigDict
-else:
-    if PYDANTIC_V2:
-        from pydantic import ConfigDict
-    else:
-        # TODO: provide an error message here?
-        ConfigDict = None
-
-
-# renamed methods / properties
-def parse_obj(model: type[_ModelT], value: object) -> _ModelT:
-    if PYDANTIC_V2:
-        return model.model_validate(value)
-    else:
-        # pyright: ignore[reportDeprecated, reportUnnecessaryCast]
-        return cast(_ModelT, model.parse_obj(value))
-
-
-def field_is_required(field: FieldInfo) -> bool:
-    if PYDANTIC_V2:
-        return field.is_required()
-    return field.required  # type: ignore
-
-
-def field_get_default(field: FieldInfo) -> Any:
-    value = field.get_default()
-    if PYDANTIC_V2:
-        from pydantic_core import PydanticUndefined
-
-        if value == PydanticUndefined:
-            return None
-        return value
-    return value
-
-
-def field_outer_type(field: FieldInfo) -> Any:
-    if PYDANTIC_V2:
-        return field.annotation
-    return field.outer_type_  # type: ignore
-
-
-def get_model_config(model: type[pydantic.BaseModel]) -> Any:
-    if PYDANTIC_V2:
-        return model.model_config
-    return model.__config__  # type: ignore
-
-
-def get_model_fields(model: type[pydantic.BaseModel]) -> dict[str, FieldInfo]:
-    if PYDANTIC_V2:
-        return model.model_fields
-    return model.__fields__  # type: ignore
-
-
-def model_copy(model: _ModelT) -> _ModelT:
-    if PYDANTIC_V2:
-        return model.model_copy()
-    return model.copy()  # type: ignore
-
-
-def model_json(model: pydantic.BaseModel, *, indent: int | None = None) -> str:
-    if PYDANTIC_V2:
-        return model.model_dump_json(indent=indent)
-    return model.json(indent=indent)  # type: ignore
-
-
-def model_dump(
-    model: pydantic.BaseModel,
-    *,
-    exclude_unset: bool = False,
-    exclude_defaults: bool = False,
-) -> dict[str, Any]:
-    if PYDANTIC_V2:
-        return model.model_dump(
-            exclude_unset=exclude_unset,
-            exclude_defaults=exclude_defaults,
-        )
-    return cast(
-        "dict[str, Any]",
-        model.dict(  # pyright: ignore[reportDeprecated, reportUnnecessaryCast]
-            exclude_unset=exclude_unset,
-            exclude_defaults=exclude_defaults,
-        ),
-    )
-
-
-def model_parse(model: type[_ModelT], data: Any) -> _ModelT:
-    if PYDANTIC_V2:
-        return model.model_validate(data)
-    return model.parse_obj(data)  # pyright: ignore[reportDeprecated]
-
-
-# generic models
-if TYPE_CHECKING:
-
-    class GenericModel(pydantic.BaseModel): ...
-
-else:
-    if PYDANTIC_V2:
-        # there no longer needs to be a distinction in v2 but
-        # we still have to create our own subclass to avoid
-        # inconsistent MRO ordering errors
-        class GenericModel(pydantic.BaseModel): ...
-
-    else:
-        import pydantic.generics
-
-        class GenericModel(pydantic.generics.GenericModel, pydantic.BaseModel): ...
-
-
-# cached properties
-if TYPE_CHECKING:
-    cached_property = property
-
-    # we define a separate type (copied from typeshed)
-    # that represents that `cached_property` is `set`able
-    # at runtime, which differs from `@property`.
-    #
-    # this is a separate type as editors likely special case
-    # `@property` and we don't want to cause issues just to have
-    # more helpful internal types.
-
-    class typed_cached_property(Generic[_T]):  # noqa: N801
-        func: Callable[[Any], _T]
-        attrname: str | None
-
-        def __init__(self, func: Callable[[Any], _T]) -> None: ...
-
-        @overload
-        def __get__(self, instance: None, owner: type[Any] | None = None) -> Self: ...
-
-        @overload
-        def __get__(self, instance: object, owner: type[Any] | None = None) -> _T: ...
-
-        def __get__(self, instance: object, owner: type[Any] | None = None) -> _T | Self:
-            raise NotImplementedError()
-
-        def __set_name__(self, owner: type[Any], name: str) -> None: ...
-
-        # __set__ is not defined at runtime, but @cached_property is designed to be settable
-        def __set__(self, instance: object, value: _T) -> None: ...
-else:
-    try:
-        from functools import cached_property
-    except ImportError:
-        from cached_property import cached_property
-
-    typed_cached_property = cached_property
diff --git a/api/core/model_runtime/model_providers/zhipuai/zhipuai_sdk/core/_base_models.py b/api/core/model_runtime/model_providers/zhipuai/zhipuai_sdk/core/_base_models.py
deleted file mode 100644
index 69b1d3a83dfef..0000000000000
--- a/api/core/model_runtime/model_providers/zhipuai/zhipuai_sdk/core/_base_models.py
+++ /dev/null
@@ -1,670 +0,0 @@
-from __future__ import annotations
-
-import inspect
-import os
-from collections.abc import Callable
-from datetime import date, datetime
-from typing import TYPE_CHECKING, Any, ClassVar, Generic, Literal, TypeGuard, TypeVar, cast
-
-import pydantic
-import pydantic.generics
-from pydantic.fields import FieldInfo
-from typing_extensions import (
-    ParamSpec,
-    Protocol,
-    override,
-    runtime_checkable,
-)
-
-from ._base_compat import (
-    PYDANTIC_V2,
-    ConfigDict,
-    field_get_default,
-    get_args,
-    get_model_config,
-    get_model_fields,
-    get_origin,
-    is_literal_type,
-    is_union,
-    parse_obj,
-)
-from ._base_compat import (
-    GenericModel as BaseGenericModel,
-)
-from ._base_type import (
-    IncEx,
-    ModelT,
-)
-from ._utils import (
-    PropertyInfo,
-    coerce_boolean,
-    extract_type_arg,
-    is_annotated_type,
-    is_list,
-    is_mapping,
-    parse_date,
-    parse_datetime,
-    strip_annotated_type,
-)
-
-if TYPE_CHECKING:
-    from pydantic_core.core_schema import ModelField
-
-__all__ = ["BaseModel", "GenericModel"]
-_BaseModelT = TypeVar("_BaseModelT", bound="BaseModel")
-
-_T = TypeVar("_T")
-P = ParamSpec("P")
-
-
-@runtime_checkable
-class _ConfigProtocol(Protocol):
-    allow_population_by_field_name: bool
-
-
-class BaseModel(pydantic.BaseModel):
-    if PYDANTIC_V2:
-        model_config: ClassVar[ConfigDict] = ConfigDict(
-            extra="allow", defer_build=coerce_boolean(os.environ.get("DEFER_PYDANTIC_BUILD", "true"))
-        )
-    else:
-
-        @property
-        @override
-        def model_fields_set(self) -> set[str]:
-            # a forwards-compat shim for pydantic v2
-            return self.__fields_set__  # type: ignore
-
-        class Config(pydantic.BaseConfig):  # pyright: ignore[reportDeprecated]
-            extra: Any = pydantic.Extra.allow  # type: ignore
-
-    def to_dict(
-        self,
-        *,
-        mode: Literal["json", "python"] = "python",
-        use_api_names: bool = True,
-        exclude_unset: bool = True,
-        exclude_defaults: bool = False,
-        exclude_none: bool = False,
-        warnings: bool = True,
-    ) -> dict[str, object]:
-        """Recursively generate a dictionary representation of the model, optionally specifying which fields to include or exclude.
-
-        By default, fields that were not set by the API will not be included,
-        and keys will match the API response, *not* the property names from the model.
-
-        For example, if the API responds with `"fooBar": true` but we've defined a `foo_bar: bool` property,
-        the output will use the `"fooBar"` key (unless `use_api_names=False` is passed).
-
-        Args:
-            mode:
-                If mode is 'json', the dictionary will only contain JSON serializable types. e.g. `datetime` will be turned into a string, `"2024-3-22T18:11:19.117000Z"`.
-                If mode is 'python', the dictionary may contain any Python objects. e.g. `datetime(2024, 3, 22)`
-
-            use_api_names: Whether to use the key that the API responded with or the property name. Defaults to `True`.
-            exclude_unset: Whether to exclude fields that have not been explicitly set.
-            exclude_defaults: Whether to exclude fields that are set to their default value from the output.
-            exclude_none: Whether to exclude fields that have a value of `None` from the output.
-            warnings: Whether to log warnings when invalid fields are encountered. This is only supported in Pydantic v2.
-        """  # noqa: E501
-        return self.model_dump(
-            mode=mode,
-            by_alias=use_api_names,
-            exclude_unset=exclude_unset,
-            exclude_defaults=exclude_defaults,
-            exclude_none=exclude_none,
-            warnings=warnings,
-        )
-
-    def to_json(
-        self,
-        *,
-        indent: int | None = 2,
-        use_api_names: bool = True,
-        exclude_unset: bool = True,
-        exclude_defaults: bool = False,
-        exclude_none: bool = False,
-        warnings: bool = True,
-    ) -> str:
-        """Generates a JSON string representing this model as it would be received from or sent to the API (but with indentation).
-
-        By default, fields that were not set by the API will not be included,
-        and keys will match the API response, *not* the property names from the model.
-
-        For example, if the API responds with `"fooBar": true` but we've defined a `foo_bar: bool` property,
-        the output will use the `"fooBar"` key (unless `use_api_names=False` is passed).
-
-        Args:
-            indent: Indentation to use in the JSON output. If `None` is passed, the output will be compact. Defaults to `2`
-            use_api_names: Whether to use the key that the API responded with or the property name. Defaults to `True`.
-            exclude_unset: Whether to exclude fields that have not been explicitly set.
-            exclude_defaults: Whether to exclude fields that have the default value.
-            exclude_none: Whether to exclude fields that have a value of `None`.
-            warnings: Whether to show any warnings that occurred during serialization. This is only supported in Pydantic v2.
-        """  # noqa: E501
-        return self.model_dump_json(
-            indent=indent,
-            by_alias=use_api_names,
-            exclude_unset=exclude_unset,
-            exclude_defaults=exclude_defaults,
-            exclude_none=exclude_none,
-            warnings=warnings,
-        )
-
-    @override
-    def __str__(self) -> str:
-        # mypy complains about an invalid self arg
-        return f'{self.__repr_name__()}({self.__repr_str__(", ")})'  # type: ignore[misc]
-
-    # Override the 'construct' method in a way that supports recursive parsing without validation.
-    # Based on https://github.com/samuelcolvin/pydantic/issues/1168#issuecomment-817742836.
-    @classmethod
-    @override
-    def construct(
-        cls: type[ModelT],
-        _fields_set: set[str] | None = None,
-        **values: object,
-    ) -> ModelT:
-        m = cls.__new__(cls)
-        fields_values: dict[str, object] = {}
-
-        config = get_model_config(cls)
-        populate_by_name = (
-            config.allow_population_by_field_name
-            if isinstance(config, _ConfigProtocol)
-            else config.get("populate_by_name")
-        )
-
-        if _fields_set is None:
-            _fields_set = set()
-
-        model_fields = get_model_fields(cls)
-        for name, field in model_fields.items():
-            key = field.alias
-            if key is None or (key not in values and populate_by_name):
-                key = name
-
-            if key in values:
-                fields_values[name] = _construct_field(value=values[key], field=field, key=key)
-                _fields_set.add(name)
-            else:
-                fields_values[name] = field_get_default(field)
-
-        _extra = {}
-        for key, value in values.items():
-            if key not in model_fields:
-                if PYDANTIC_V2:
-                    _extra[key] = value
-                else:
-                    _fields_set.add(key)
-                    fields_values[key] = value
-
-        object.__setattr__(m, "__dict__", fields_values)  # noqa: PLC2801
-
-        if PYDANTIC_V2:
-            # these properties are copied from Pydantic's `model_construct()` method
-            object.__setattr__(m, "__pydantic_private__", None)  # noqa: PLC2801
-            object.__setattr__(m, "__pydantic_extra__", _extra)  # noqa: PLC2801
-            object.__setattr__(m, "__pydantic_fields_set__", _fields_set)  # noqa: PLC2801
-        else:
-            # init_private_attributes() does not exist in v2
-            m._init_private_attributes()  # type: ignore
-
-            # copied from Pydantic v1's `construct()` method
-            object.__setattr__(m, "__fields_set__", _fields_set)  # noqa: PLC2801
-
-        return m
-
-    if not TYPE_CHECKING:
-        # type checkers incorrectly complain about this assignment
-        # because the type signatures are technically different
-        # although not in practice
-        model_construct = construct
-
-    if not PYDANTIC_V2:
-        # we define aliases for some of the new pydantic v2 methods so
-        # that we can just document these methods without having to specify
-        # a specific pydantic version as some users may not know which
-        # pydantic version they are currently using
-
-        @override
-        def model_dump(
-            self,
-            *,
-            mode: Literal["json", "python"] | str = "python",
-            include: IncEx = None,
-            exclude: IncEx = None,
-            by_alias: bool = False,
-            exclude_unset: bool = False,
-            exclude_defaults: bool = False,
-            exclude_none: bool = False,
-            round_trip: bool = False,
-            warnings: bool | Literal["none", "warn", "error"] = True,
-            context: dict[str, Any] | None = None,
-            serialize_as_any: bool = False,
-        ) -> dict[str, Any]:
-            """Usage docs: https://docs.pydantic.dev/2.4/concepts/serialization/#modelmodel_dump
-
-            Generate a dictionary representation of the model, optionally specifying which fields to include or exclude.
-
-            Args:
-                mode: The mode in which `to_python` should run.
-                    If mode is 'json', the dictionary will only contain JSON serializable types.
-                    If mode is 'python', the dictionary may contain any Python objects.
-                include: A list of fields to include in the output.
-                exclude: A list of fields to exclude from the output.
-                by_alias: Whether to use the field's alias in the dictionary key if defined.
-                exclude_unset: Whether to exclude fields that are unset or None from the output.
-                exclude_defaults: Whether to exclude fields that are set to their default value from the output.
-                exclude_none: Whether to exclude fields that have a value of `None` from the output.
-                round_trip: Whether to enable serialization and deserialization round-trip support.
-                warnings: Whether to log warnings when invalid fields are encountered.
-
-            Returns:
-                A dictionary representation of the model.
-            """
-            if mode != "python":
-                raise ValueError("mode is only supported in Pydantic v2")
-            if round_trip != False:
-                raise ValueError("round_trip is only supported in Pydantic v2")
-            if warnings != True:
-                raise ValueError("warnings is only supported in Pydantic v2")
-            if context is not None:
-                raise ValueError("context is only supported in Pydantic v2")
-            if serialize_as_any != False:
-                raise ValueError("serialize_as_any is only supported in Pydantic v2")
-            return super().dict(  # pyright: ignore[reportDeprecated]
-                include=include,
-                exclude=exclude,
-                by_alias=by_alias,
-                exclude_unset=exclude_unset,
-                exclude_defaults=exclude_defaults,
-                exclude_none=exclude_none,
-            )
-
-        @override
-        def model_dump_json(
-            self,
-            *,
-            indent: int | None = None,
-            include: IncEx = None,
-            exclude: IncEx = None,
-            by_alias: bool = False,
-            exclude_unset: bool = False,
-            exclude_defaults: bool = False,
-            exclude_none: bool = False,
-            round_trip: bool = False,
-            warnings: bool | Literal["none", "warn", "error"] = True,
-            context: dict[str, Any] | None = None,
-            serialize_as_any: bool = False,
-        ) -> str:
-            """Usage docs: https://docs.pydantic.dev/2.4/concepts/serialization/#modelmodel_dump_json
-
-            Generates a JSON representation of the model using Pydantic's `to_json` method.
-
-            Args:
-                indent: Indentation to use in the JSON output. If None is passed, the output will be compact.
-                include: Field(s) to include in the JSON output. Can take either a string or set of strings.
-                exclude: Field(s) to exclude from the JSON output. Can take either a string or set of strings.
-                by_alias: Whether to serialize using field aliases.
-                exclude_unset: Whether to exclude fields that have not been explicitly set.
-                exclude_defaults: Whether to exclude fields that have the default value.
-                exclude_none: Whether to exclude fields that have a value of `None`.
-                round_trip: Whether to use serialization/deserialization between JSON and class instance.
-                warnings: Whether to show any warnings that occurred during serialization.
-
-            Returns:
-                A JSON string representation of the model.
-            """
-            if round_trip != False:
-                raise ValueError("round_trip is only supported in Pydantic v2")
-            if warnings != True:
-                raise ValueError("warnings is only supported in Pydantic v2")
-            if context is not None:
-                raise ValueError("context is only supported in Pydantic v2")
-            if serialize_as_any != False:
-                raise ValueError("serialize_as_any is only supported in Pydantic v2")
-            return super().json(  # type: ignore[reportDeprecated]
-                indent=indent,
-                include=include,
-                exclude=exclude,
-                by_alias=by_alias,
-                exclude_unset=exclude_unset,
-                exclude_defaults=exclude_defaults,
-                exclude_none=exclude_none,
-            )
-
-
-def _construct_field(value: object, field: FieldInfo, key: str) -> object:
-    if value is None:
-        return field_get_default(field)
-
-    if PYDANTIC_V2:
-        type_ = field.annotation
-    else:
-        type_ = cast(type, field.outer_type_)  # type: ignore
-
-    if type_ is None:
-        raise RuntimeError(f"Unexpected field type is None for {key}")
-
-    return construct_type(value=value, type_=type_)
-
-
-def is_basemodel(type_: type) -> bool:
-    """Returns whether or not the given type is either a `BaseModel` or a union of `BaseModel`"""
-    if is_union(type_):
-        return any(is_basemodel(variant) for variant in get_args(type_))
-
-    return is_basemodel_type(type_)
-
-
-def is_basemodel_type(type_: type) -> TypeGuard[type[BaseModel] | type[GenericModel]]:
-    origin = get_origin(type_) or type_
-    return issubclass(origin, BaseModel) or issubclass(origin, GenericModel)
-
-
-def build(
-    base_model_cls: Callable[P, _BaseModelT],
-    *args: P.args,
-    **kwargs: P.kwargs,
-) -> _BaseModelT:
-    """Construct a BaseModel class without validation.
-
-    This is useful for cases where you need to instantiate a `BaseModel`
-    from an API response as this provides type-safe params which isn't supported
-    by helpers like `construct_type()`.
-
-    ```py
-    build(MyModel, my_field_a="foo", my_field_b=123)
-    ```
-    """
-    if args:
-        raise TypeError(
-            "Received positional arguments which are not supported; Keyword arguments must be used instead",
-        )
-
-    return cast(_BaseModelT, construct_type(type_=base_model_cls, value=kwargs))
-
-
-def construct_type_unchecked(*, value: object, type_: type[_T]) -> _T:
-    """Loose coercion to the expected type with construction of nested values.
-
-    Note: the returned value from this function is not guaranteed to match the
-    given type.
-    """
-    return cast(_T, construct_type(value=value, type_=type_))
-
-
-def construct_type(*, value: object, type_: type) -> object:
-    """Loose coercion to the expected type with construction of nested values.
-
-    If the given value does not match the expected type then it is returned as-is.
-    """
-    # we allow `object` as the input type because otherwise, passing things like
-    # `Literal['value']` will be reported as a type error by type checkers
-    type_ = cast("type[object]", type_)
-
-    # unwrap `Annotated[T, ...]` -> `T`
-    if is_annotated_type(type_):
-        meta: tuple[Any, ...] = get_args(type_)[1:]
-        type_ = extract_type_arg(type_, 0)
-    else:
-        meta = ()
-    # we need to use the origin class for any types that are subscripted generics
-    # e.g. Dict[str, object]
-    origin = get_origin(type_) or type_
-    args = get_args(type_)
-
-    if is_union(origin):
-        try:
-            return validate_type(type_=cast("type[object]", type_), value=value)
-        except Exception:
-            pass
-
-        # if the type is a discriminated union then we want to construct the right variant
-        # in the union, even if the data doesn't match exactly, otherwise we'd break code
-        # that relies on the constructed class types, e.g.
-        #
-        # class FooType:
-        #   kind: Literal['foo']
-        #   value: str
-        #
-        # class BarType:
-        #   kind: Literal['bar']
-        #   value: int
-        #
-        # without this block, if the data we get is something like `{'kind': 'bar', 'value': 'foo'}` then
-        # we'd end up constructing `FooType` when it should be `BarType`.
-        discriminator = _build_discriminated_union_meta(union=type_, meta_annotations=meta)
-        if discriminator and is_mapping(value):
-            variant_value = value.get(discriminator.field_alias_from or discriminator.field_name)
-            if variant_value and isinstance(variant_value, str):
-                variant_type = discriminator.mapping.get(variant_value)
-                if variant_type:
-                    return construct_type(type_=variant_type, value=value)
-
-        # if the data is not valid, use the first variant that doesn't fail while deserializing
-        for variant in args:
-            try:
-                return construct_type(value=value, type_=variant)
-            except Exception:
-                continue
-
-        raise RuntimeError(f"Could not convert data into a valid instance of {type_}")
-    if origin == dict:
-        if not is_mapping(value):
-            return value
-
-        _, items_type = get_args(type_)  # Dict[_, items_type]
-        return {key: construct_type(value=item, type_=items_type) for key, item in value.items()}
-
-    if not is_literal_type(type_) and (issubclass(origin, BaseModel) or issubclass(origin, GenericModel)):
-        if is_list(value):
-            return [cast(Any, type_).construct(**entry) if is_mapping(entry) else entry for entry in value]
-
-        if is_mapping(value):
-            if issubclass(type_, BaseModel):
-                return type_.construct(**value)  # type: ignore[arg-type]
-
-            return cast(Any, type_).construct(**value)
-
-    if origin == list:
-        if not is_list(value):
-            return value
-
-        inner_type = args[0]  # List[inner_type]
-        return [construct_type(value=entry, type_=inner_type) for entry in value]
-
-    if origin == float:
-        if isinstance(value, int):
-            coerced = float(value)
-            if coerced != value:
-                return value
-            return coerced
-
-        return value
-
-    if type_ == datetime:
-        try:
-            return parse_datetime(value)  # type: ignore
-        except Exception:
-            return value
-
-    if type_ == date:
-        try:
-            return parse_date(value)  # type: ignore
-        except Exception:
-            return value
-
-    return value
-
-
-@runtime_checkable
-class CachedDiscriminatorType(Protocol):
-    __discriminator__: DiscriminatorDetails
-
-
-class DiscriminatorDetails:
-    field_name: str
-    """The name of the discriminator field in the variant class, e.g.
-
-    ```py
-    class Foo(BaseModel):
-        type: Literal['foo']
-    ```
-
-    Will result in field_name='type'
-    """
-
-    field_alias_from: str | None
-    """The name of the discriminator field in the API response, e.g.
-
-    ```py
-    class Foo(BaseModel):
-        type: Literal['foo'] = Field(alias='type_from_api')
-    ```
-
-    Will result in field_alias_from='type_from_api'
-    """
-
-    mapping: dict[str, type]
-    """Mapping of discriminator value to variant type, e.g.
-
-    {'foo': FooVariant, 'bar': BarVariant}
-    """
-
-    def __init__(
-        self,
-        *,
-        mapping: dict[str, type],
-        discriminator_field: str,
-        discriminator_alias: str | None,
-    ) -> None:
-        self.mapping = mapping
-        self.field_name = discriminator_field
-        self.field_alias_from = discriminator_alias
-
-
-def _build_discriminated_union_meta(*, union: type, meta_annotations: tuple[Any, ...]) -> DiscriminatorDetails | None:
-    if isinstance(union, CachedDiscriminatorType):
-        return union.__discriminator__
-
-    discriminator_field_name: str | None = None
-
-    for annotation in meta_annotations:
-        if isinstance(annotation, PropertyInfo) and annotation.discriminator is not None:
-            discriminator_field_name = annotation.discriminator
-            break
-
-    if not discriminator_field_name:
-        return None
-
-    mapping: dict[str, type] = {}
-    discriminator_alias: str | None = None
-
-    for variant in get_args(union):
-        variant = strip_annotated_type(variant)
-        if is_basemodel_type(variant):
-            if PYDANTIC_V2:
-                field = _extract_field_schema_pv2(variant, discriminator_field_name)
-                if not field:
-                    continue
-
-                # Note: if one variant defines an alias then they all should
-                discriminator_alias = field.get("serialization_alias")
-
-                field_schema = field["schema"]
-
-                if field_schema["type"] == "literal":
-                    for entry in cast("LiteralSchema", field_schema)["expected"]:
-                        if isinstance(entry, str):
-                            mapping[entry] = variant
-            else:
-                field_info = cast("dict[str, FieldInfo]", variant.__fields__).get(discriminator_field_name)  # pyright: ignore[reportDeprecated, reportUnnecessaryCast]
-                if not field_info:
-                    continue
-
-                # Note: if one variant defines an alias then they all should
-                discriminator_alias = field_info.alias
-
-                if field_info.annotation and is_literal_type(field_info.annotation):
-                    for entry in get_args(field_info.annotation):
-                        if isinstance(entry, str):
-                            mapping[entry] = variant
-
-    if not mapping:
-        return None
-
-    details = DiscriminatorDetails(
-        mapping=mapping,
-        discriminator_field=discriminator_field_name,
-        discriminator_alias=discriminator_alias,
-    )
-    cast(CachedDiscriminatorType, union).__discriminator__ = details
-    return details
-
-
-def _extract_field_schema_pv2(model: type[BaseModel], field_name: str) -> ModelField | None:
-    schema = model.__pydantic_core_schema__
-    if schema["type"] != "model":
-        return None
-
-    fields_schema = schema["schema"]
-    if fields_schema["type"] != "model-fields":
-        return None
-
-    fields_schema = cast("ModelFieldsSchema", fields_schema)
-
-    field = fields_schema["fields"].get(field_name)
-    if not field:
-        return None
-
-    return cast("ModelField", field)  # pyright: ignore[reportUnnecessaryCast]
-
-
-def validate_type(*, type_: type[_T], value: object) -> _T:
-    """Strict validation that the given value matches the expected type"""
-    if inspect.isclass(type_) and issubclass(type_, pydantic.BaseModel):
-        return cast(_T, parse_obj(type_, value))
-
-    return cast(_T, _validate_non_model_type(type_=type_, value=value))
-
-
-# Subclassing here confuses type checkers, so we treat this class as non-inheriting.
-if TYPE_CHECKING:
-    GenericModel = BaseModel
-else:
-
-    class GenericModel(BaseGenericModel, BaseModel):
-        pass
-
-
-if PYDANTIC_V2:
-    from pydantic import TypeAdapter
-
-    def _validate_non_model_type(*, type_: type[_T], value: object) -> _T:
-        return TypeAdapter(type_).validate_python(value)
-
-elif not TYPE_CHECKING:
-
-    class TypeAdapter(Generic[_T]):
-        """Used as a placeholder to easily convert runtime types to a Pydantic format
-        to provide validation.
-
-        For example:
-        ```py
-        validated = RootModel[int](__root__="5").__root__
-        # validated: 5
-        ```
-        """
-
-        def __init__(self, type_: type[_T]):
-            self.type_ = type_
-
-        def validate_python(self, value: Any) -> _T:
-            if not isinstance(value, self.type_):
-                raise ValueError(f"Invalid type: {value} is not of type {self.type_}")
-            return value
-
-    def _validate_non_model_type(*, type_: type[_T], value: object) -> _T:
-        return TypeAdapter(type_).validate_python(value)
diff --git a/api/core/model_runtime/model_providers/zhipuai/zhipuai_sdk/core/_base_type.py b/api/core/model_runtime/model_providers/zhipuai/zhipuai_sdk/core/_base_type.py
deleted file mode 100644
index ea1d3f09dc42e..0000000000000
--- a/api/core/model_runtime/model_providers/zhipuai/zhipuai_sdk/core/_base_type.py
+++ /dev/null
@@ -1,170 +0,0 @@
-from __future__ import annotations
-
-from collections.abc import Callable, Mapping, Sequence
-from os import PathLike
-from typing import (
-    IO,
-    TYPE_CHECKING,
-    Any,
-    Literal,
-    Optional,
-    TypeAlias,
-    TypeVar,
-    Union,
-)
-
-import pydantic
-from httpx import Response
-from typing_extensions import Protocol, TypedDict, override, runtime_checkable
-
-Query = Mapping[str, object]
-Body = object
-AnyMapping = Mapping[str, object]
-PrimitiveData = Union[str, int, float, bool, None]
-Data = Union[PrimitiveData, list[Any], tuple[Any], "Mapping[str, Any]"]
-ModelT = TypeVar("ModelT", bound=pydantic.BaseModel)
-_T = TypeVar("_T")
-
-if TYPE_CHECKING:
-    NoneType: type[None]
-else:
-    NoneType = type(None)
-
-
-# Sentinel class used until PEP 0661 is accepted
-class NotGiven:
-    """
-    A sentinel singleton class used to distinguish omitted keyword arguments
-    from those passed in with the value None (which may have different behavior).
-
-    For example:
-
-    ```py
-    def get(timeout: Union[int, NotGiven, None] = NotGiven()) -> Response: ...
-
-    get(timeout=1) # 1s timeout
-    get(timeout=None) # No timeout
-    get() # Default timeout behavior, which may not be statically known at the method definition.
-    ```
-    """
-
-    def __bool__(self) -> Literal[False]:
-        return False
-
-    @override
-    def __repr__(self) -> str:
-        return "NOT_GIVEN"
-
-
-NotGivenOr = Union[_T, NotGiven]
-NOT_GIVEN = NotGiven()
-
-
-class Omit:
-    """In certain situations you need to be able to represent a case where a default value has
-    to be explicitly removed and `None` is not an appropriate substitute, for example:
-
-    ```py
-    # as the default `Content-Type` header is `application/json` that will be sent
-    client.post('/upload/files', files={'file': b'my raw file content'})
-
-    # you can't explicitly override the header as it has to be dynamically generated
-    # to look something like: 'multipart/form-data; boundary=0d8382fcf5f8c3be01ca2e11002d2983'
-    client.post(..., headers={'Content-Type': 'multipart/form-data'})
-
-    # instead you can remove the default `application/json` header by passing Omit
-    client.post(..., headers={'Content-Type': Omit()})
-    ```
-    """
-
-    def __bool__(self) -> Literal[False]:
-        return False
-
-
-@runtime_checkable
-class ModelBuilderProtocol(Protocol):
-    @classmethod
-    def build(
-        cls: type[_T],
-        *,
-        response: Response,
-        data: object,
-    ) -> _T: ...
-
-
-Headers = Mapping[str, Union[str, Omit]]
-
-
-class HeadersLikeProtocol(Protocol):
-    def get(self, __key: str) -> str | None: ...
-
-
-HeadersLike = Union[Headers, HeadersLikeProtocol]
-
-ResponseT = TypeVar(
-    "ResponseT",
-    bound="Union[str, None, BaseModel, list[Any], dict[str, Any], Response, UnknownResponse, ModelBuilderProtocol, BinaryResponseContent]",  # noqa: E501
-)
-
-StrBytesIntFloat = Union[str, bytes, int, float]
-
-# Note: copied from Pydantic
-# https://github.com/pydantic/pydantic/blob/32ea570bf96e84234d2992e1ddf40ab8a565925a/pydantic/main.py#L49
-IncEx: TypeAlias = "set[int] | set[str] | dict[int, Any] | dict[str, Any] | None"
-
-PostParser = Callable[[Any], Any]
-
-
-@runtime_checkable
-class InheritsGeneric(Protocol):
-    """Represents a type that has inherited from `Generic`
-
-    The `__orig_bases__` property can be used to determine the resolved
-    type variable for a given base class.
-    """
-
-    __orig_bases__: tuple[_GenericAlias]
-
-
-class _GenericAlias(Protocol):
-    __origin__: type[object]
-
-
-class HttpxSendArgs(TypedDict, total=False):
-    auth: httpx.Auth
-
-
-# for user input files
-if TYPE_CHECKING:
-    Base64FileInput = Union[IO[bytes], PathLike[str]]
-    FileContent = Union[IO[bytes], bytes, PathLike[str]]
-else:
-    Base64FileInput = Union[IO[bytes], PathLike]
-    FileContent = Union[IO[bytes], bytes, PathLike]
-
-FileTypes = Union[
-    # file (or bytes)
-    FileContent,
-    # (filename, file (or bytes))
-    tuple[Optional[str], FileContent],
-    # (filename, file (or bytes), content_type)
-    tuple[Optional[str], FileContent, Optional[str]],
-    # (filename, file (or bytes), content_type, headers)
-    tuple[Optional[str], FileContent, Optional[str], Mapping[str, str]],
-]
-RequestFiles = Union[Mapping[str, FileTypes], Sequence[tuple[str, FileTypes]]]
-
-# duplicate of the above but without our custom file support
-HttpxFileContent = Union[bytes, IO[bytes]]
-HttpxFileTypes = Union[
-    # file (or bytes)
-    HttpxFileContent,
-    # (filename, file (or bytes))
-    tuple[Optional[str], HttpxFileContent],
-    # (filename, file (or bytes), content_type)
-    tuple[Optional[str], HttpxFileContent, Optional[str]],
-    # (filename, file (or bytes), content_type, headers)
-    tuple[Optional[str], HttpxFileContent, Optional[str], Mapping[str, str]],
-]
-
-HttpxRequestFiles = Union[Mapping[str, HttpxFileTypes], Sequence[tuple[str, HttpxFileTypes]]]
diff --git a/api/core/model_runtime/model_providers/zhipuai/zhipuai_sdk/core/_constants.py b/api/core/model_runtime/model_providers/zhipuai/zhipuai_sdk/core/_constants.py
deleted file mode 100644
index 8e43bdebecb61..0000000000000
--- a/api/core/model_runtime/model_providers/zhipuai/zhipuai_sdk/core/_constants.py
+++ /dev/null
@@ -1,12 +0,0 @@
-import httpx
-
-RAW_RESPONSE_HEADER = "X-Stainless-Raw-Response"
-# 通过 `Timeout` 控制接口`connect` 和 `read` 超时时间，默认为`timeout=300.0, connect=8.0`
-ZHIPUAI_DEFAULT_TIMEOUT = httpx.Timeout(timeout=300.0, connect=8.0)
-# 通过 `retry` 参数控制重试次数，默认为3次
-ZHIPUAI_DEFAULT_MAX_RETRIES = 3
-# 通过 `Limits` 控制最大连接数和保持连接数，默认为`max_connections=50, max_keepalive_connections=10`
-ZHIPUAI_DEFAULT_LIMITS = httpx.Limits(max_connections=50, max_keepalive_connections=10)
-
-INITIAL_RETRY_DELAY = 0.5
-MAX_RETRY_DELAY = 8.0
diff --git a/api/core/model_runtime/model_providers/zhipuai/zhipuai_sdk/core/_errors.py b/api/core/model_runtime/model_providers/zhipuai/zhipuai_sdk/core/_errors.py
deleted file mode 100644
index e2c9d24c6c0d2..0000000000000
--- a/api/core/model_runtime/model_providers/zhipuai/zhipuai_sdk/core/_errors.py
+++ /dev/null
@@ -1,86 +0,0 @@
-from __future__ import annotations
-
-import httpx
-
-__all__ = [
-    "ZhipuAIError",
-    "APIStatusError",
-    "APIRequestFailedError",
-    "APIAuthenticationError",
-    "APIReachLimitError",
-    "APIInternalError",
-    "APIServerFlowExceedError",
-    "APIResponseError",
-    "APIResponseValidationError",
-    "APITimeoutError",
-    "APIConnectionError",
-]
-
-
-class ZhipuAIError(Exception):
-    def __init__(
-        self,
-        message: str,
-    ) -> None:
-        super().__init__(message)
-
-
-class APIStatusError(ZhipuAIError):
-    response: httpx.Response
-    status_code: int
-
-    def __init__(self, message: str, *, response: httpx.Response) -> None:
-        super().__init__(message)
-        self.response = response
-        self.status_code = response.status_code
-
-
-class APIRequestFailedError(APIStatusError): ...
-
-
-class APIAuthenticationError(APIStatusError): ...
-
-
-class APIReachLimitError(APIStatusError): ...
-
-
-class APIInternalError(APIStatusError): ...
-
-
-class APIServerFlowExceedError(APIStatusError): ...
-
-
-class APIResponseError(ZhipuAIError):
-    message: str
-    request: httpx.Request
-    json_data: object
-
-    def __init__(self, message: str, request: httpx.Request, json_data: object):
-        self.message = message
-        self.request = request
-        self.json_data = json_data
-        super().__init__(message)
-
-
-class APIResponseValidationError(APIResponseError):
-    status_code: int
-    response: httpx.Response
-
-    def __init__(self, response: httpx.Response, json_data: object | None, *, message: str | None = None) -> None:
-        super().__init__(
-            message=message or "Data returned by API invalid for expected schema.",
-            request=response.request,
-            json_data=json_data,
-        )
-        self.response = response
-        self.status_code = response.status_code
-
-
-class APIConnectionError(APIResponseError):
-    def __init__(self, *, message: str = "Connection error.", request: httpx.Request) -> None:
-        super().__init__(message, request, json_data=None)
-
-
-class APITimeoutError(APIConnectionError):
-    def __init__(self, request: httpx.Request) -> None:
-        super().__init__(message="Request timed out.", request=request)
diff --git a/api/core/model_runtime/model_providers/zhipuai/zhipuai_sdk/core/_files.py b/api/core/model_runtime/model_providers/zhipuai/zhipuai_sdk/core/_files.py
deleted file mode 100644
index f9d2e14d9ecb9..0000000000000
--- a/api/core/model_runtime/model_providers/zhipuai/zhipuai_sdk/core/_files.py
+++ /dev/null
@@ -1,75 +0,0 @@
-from __future__ import annotations
-
-import io
-import os
-import pathlib
-from typing import TypeGuard, overload
-
-from ._base_type import (
-    Base64FileInput,
-    FileContent,
-    FileTypes,
-    HttpxFileContent,
-    HttpxFileTypes,
-    HttpxRequestFiles,
-    RequestFiles,
-)
-from ._utils import is_mapping_t, is_sequence_t, is_tuple_t
-
-
-def is_base64_file_input(obj: object) -> TypeGuard[Base64FileInput]:
-    return isinstance(obj, io.IOBase | os.PathLike)
-
-
-def is_file_content(obj: object) -> TypeGuard[FileContent]:
-    return isinstance(obj, bytes | tuple | io.IOBase | os.PathLike)
-
-
-def assert_is_file_content(obj: object, *, key: str | None = None) -> None:
-    if not is_file_content(obj):
-        prefix = f"Expected entry at `{key}`" if key is not None else f"Expected file input `{obj!r}`"
-        raise RuntimeError(
-            f"{prefix} to be bytes, an io.IOBase instance, PathLike or a tuple but received {type(obj)} instead. See https://github.com/openai/openai-python/tree/main#file-uploads"
-        ) from None
-
-
-@overload
-def to_httpx_files(files: None) -> None: ...
-
-
-@overload
-def to_httpx_files(files: RequestFiles) -> HttpxRequestFiles: ...
-
-
-def to_httpx_files(files: RequestFiles | None) -> HttpxRequestFiles | None:
-    if files is None:
-        return None
-
-    if is_mapping_t(files):
-        files = {key: _transform_file(file) for key, file in files.items()}
-    elif is_sequence_t(files):
-        files = [(key, _transform_file(file)) for key, file in files]
-    else:
-        raise TypeError(f"Unexpected file type input {type(files)}, expected mapping or sequence")
-
-    return files
-
-
-def _transform_file(file: FileTypes) -> HttpxFileTypes:
-    if is_file_content(file):
-        if isinstance(file, os.PathLike):
-            path = pathlib.Path(file)
-            return (path.name, path.read_bytes())
-
-        return file
-
-    if is_tuple_t(file):
-        return (file[0], _read_file_content(file[1]), *file[2:])
-
-    raise TypeError("Expected file types input to be a FileContent type or to be a tuple")
-
-
-def _read_file_content(file: FileContent) -> HttpxFileContent:
-    if isinstance(file, os.PathLike):
-        return pathlib.Path(file).read_bytes()
-    return file
diff --git a/api/core/model_runtime/model_providers/zhipuai/zhipuai_sdk/core/_http_client.py b/api/core/model_runtime/model_providers/zhipuai/zhipuai_sdk/core/_http_client.py
deleted file mode 100644
index ffdafb85d581f..0000000000000
--- a/api/core/model_runtime/model_providers/zhipuai/zhipuai_sdk/core/_http_client.py
+++ /dev/null
@@ -1,910 +0,0 @@
-from __future__ import annotations
-
-import inspect
-import logging
-import time
-import warnings
-from collections.abc import Iterator, Mapping
-from itertools import starmap
-from random import random
-from typing import TYPE_CHECKING, Any, Generic, Literal, Optional, TypeVar, Union, cast, overload
-
-import httpx
-import pydantic
-from httpx import URL, Timeout
-
-from . import _errors, get_origin
-from ._base_compat import model_copy
-from ._base_models import GenericModel, construct_type, validate_type
-from ._base_type import (
-    NOT_GIVEN,
-    AnyMapping,
-    Body,
-    Data,
-    Headers,
-    HttpxSendArgs,
-    ModelBuilderProtocol,
-    NotGiven,
-    Omit,
-    PostParser,
-    Query,
-    RequestFiles,
-    ResponseT,
-)
-from ._constants import (
-    INITIAL_RETRY_DELAY,
-    MAX_RETRY_DELAY,
-    RAW_RESPONSE_HEADER,
-    ZHIPUAI_DEFAULT_LIMITS,
-    ZHIPUAI_DEFAULT_MAX_RETRIES,
-    ZHIPUAI_DEFAULT_TIMEOUT,
-)
-from ._errors import APIConnectionError, APIResponseValidationError, APIStatusError, APITimeoutError
-from ._files import to_httpx_files
-from ._legacy_response import LegacyAPIResponse
-from ._request_opt import FinalRequestOptions, UserRequestInput
-from ._response import APIResponse, BaseAPIResponse, extract_response_type
-from ._sse_client import StreamResponse
-from ._utils import flatten, is_given, is_mapping
-
-log: logging.Logger = logging.getLogger(__name__)
-
-# TODO: make base page type vars covariant
-SyncPageT = TypeVar("SyncPageT", bound="BaseSyncPage[Any]")
-# AsyncPageT = TypeVar("AsyncPageT", bound="BaseAsyncPage[Any]")
-
-_T = TypeVar("_T")
-_T_co = TypeVar("_T_co", covariant=True)
-
-if TYPE_CHECKING:
-    from httpx._config import DEFAULT_TIMEOUT_CONFIG as HTTPX_DEFAULT_TIMEOUT
-else:
-    try:
-        from httpx._config import DEFAULT_TIMEOUT_CONFIG as HTTPX_DEFAULT_TIMEOUT
-    except ImportError:
-        # taken from https://github.com/encode/httpx/blob/3ba5fe0d7ac70222590e759c31442b1cab263791/httpx/_config.py#L366
-        HTTPX_DEFAULT_TIMEOUT = Timeout(5.0)
-
-
-headers = {
-    "Accept": "application/json",
-    "Content-Type": "application/json; charset=UTF-8",
-}
-
-
-class PageInfo:
-    """Stores the necessary information to build the request to retrieve the next page.
-
-    Either `url` or `params` must be set.
-    """
-
-    url: URL | NotGiven
-    params: Query | NotGiven
-
-    @overload
-    def __init__(
-        self,
-        *,
-        url: URL,
-    ) -> None: ...
-
-    @overload
-    def __init__(
-        self,
-        *,
-        params: Query,
-    ) -> None: ...
-
-    def __init__(
-        self,
-        *,
-        url: URL | NotGiven = NOT_GIVEN,
-        params: Query | NotGiven = NOT_GIVEN,
-    ) -> None:
-        self.url = url
-        self.params = params
-
-
-class BasePage(GenericModel, Generic[_T]):
-    """
-    Defines the core interface for pagination.
-
-    Type Args:
-        ModelT: The pydantic model that represents an item in the response.
-
-    Methods:
-        has_next_page(): Check if there is another page available
-        next_page_info(): Get the necessary information to make a request for the next page
-    """
-
-    _options: FinalRequestOptions = pydantic.PrivateAttr()
-    _model: type[_T] = pydantic.PrivateAttr()
-
-    def has_next_page(self) -> bool:
-        items = self._get_page_items()
-        if not items:
-            return False
-        return self.next_page_info() is not None
-
-    def next_page_info(self) -> Optional[PageInfo]: ...
-
-    def _get_page_items(self) -> Iterable[_T]:  # type: ignore[empty-body]
-        ...
-
-    def _params_from_url(self, url: URL) -> httpx.QueryParams:
-        # TODO: do we have to preprocess params here?
-        return httpx.QueryParams(cast(Any, self._options.params)).merge(url.params)
-
-    def _info_to_options(self, info: PageInfo) -> FinalRequestOptions:
-        options = model_copy(self._options)
-        options._strip_raw_response_header()
-
-        if not isinstance(info.params, NotGiven):
-            options.params = {**options.params, **info.params}
-            return options
-
-        if not isinstance(info.url, NotGiven):
-            params = self._params_from_url(info.url)
-            url = info.url.copy_with(params=params)
-            options.params = dict(url.params)
-            options.url = str(url)
-            return options
-
-        raise ValueError("Unexpected PageInfo state")
-
-
-class BaseSyncPage(BasePage[_T], Generic[_T]):
-    _client: HttpClient = pydantic.PrivateAttr()
-
-    def _set_private_attributes(
-        self,
-        client: HttpClient,
-        model: type[_T],
-        options: FinalRequestOptions,
-    ) -> None:
-        self._model = model
-        self._client = client
-        self._options = options
-
-    # Pydantic uses a custom `__iter__` method to support casting BaseModels
-    # to dictionaries. e.g. dict(model).
-    # As we want to support `for item in page`, this is inherently incompatible
-    # with the default pydantic behavior. It is not possible to support both
-    # use cases at once. Fortunately, this is not a big deal as all other pydantic
-    # methods should continue to work as expected as there is an alternative method
-    # to cast a model to a dictionary, model.dict(), which is used internally
-    # by pydantic.
-    def __iter__(self) -> Iterator[_T]:  # type: ignore
-        for page in self.iter_pages():
-            yield from page._get_page_items()
-
-    def iter_pages(self: SyncPageT) -> Iterator[SyncPageT]:
-        page = self
-        while True:
-            yield page
-            if page.has_next_page():
-                page = page.get_next_page()
-            else:
-                return
-
-    def get_next_page(self: SyncPageT) -> SyncPageT:
-        info = self.next_page_info()
-        if not info:
-            raise RuntimeError(
-                "No next page expected; please check `.has_next_page()` before calling `.get_next_page()`."
-            )
-
-        options = self._info_to_options(info)
-        return self._client._request_api_list(self._model, page=self.__class__, options=options)
-
-
-class HttpClient:
-    _client: httpx.Client
-    _version: str
-    _base_url: URL
-    max_retries: int
-    timeout: Union[float, Timeout, None]
-    _limits: httpx.Limits
-    _has_custom_http_client: bool
-    _default_stream_cls: type[StreamResponse[Any]] | None = None
-
-    _strict_response_validation: bool
-
-    def __init__(
-        self,
-        *,
-        version: str,
-        base_url: URL,
-        _strict_response_validation: bool,
-        max_retries: int = ZHIPUAI_DEFAULT_MAX_RETRIES,
-        timeout: Union[float, Timeout, None],
-        limits: httpx.Limits | None = None,
-        custom_httpx_client: httpx.Client | None = None,
-        custom_headers: Mapping[str, str] | None = None,
-    ) -> None:
-        if limits is not None:
-            warnings.warn(
-                "The `connection_pool_limits` argument is deprecated. The `http_client` argument should be passed instead",  # noqa: E501
-                category=DeprecationWarning,
-                stacklevel=3,
-            )
-            if custom_httpx_client is not None:
-                raise ValueError("The `http_client` argument is mutually exclusive with `connection_pool_limits`")
-        else:
-            limits = ZHIPUAI_DEFAULT_LIMITS
-
-        if not is_given(timeout):
-            if custom_httpx_client and custom_httpx_client.timeout != HTTPX_DEFAULT_TIMEOUT:
-                timeout = custom_httpx_client.timeout
-            else:
-                timeout = ZHIPUAI_DEFAULT_TIMEOUT
-        self.max_retries = max_retries
-        self.timeout = timeout
-        self._limits = limits
-        self._has_custom_http_client = bool(custom_httpx_client)
-        self._client = custom_httpx_client or httpx.Client(
-            base_url=base_url,
-            timeout=self.timeout,
-            limits=limits,
-        )
-        self._version = version
-        url = URL(url=base_url)
-        if not url.raw_path.endswith(b"/"):
-            url = url.copy_with(raw_path=url.raw_path + b"/")
-        self._base_url = url
-        self._custom_headers = custom_headers or {}
-        self._strict_response_validation = _strict_response_validation
-
-    def _prepare_url(self, url: str) -> URL:
-        sub_url = URL(url)
-        if sub_url.is_relative_url:
-            request_raw_url = self._base_url.raw_path + sub_url.raw_path.lstrip(b"/")
-            return self._base_url.copy_with(raw_path=request_raw_url)
-
-        return sub_url
-
-    @property
-    def _default_headers(self):
-        return {
-            "Accept": "application/json",
-            "Content-Type": "application/json; charset=UTF-8",
-            "ZhipuAI-SDK-Ver": self._version,
-            "source_type": "zhipu-sdk-python",
-            "x-request-sdk": "zhipu-sdk-python",
-            **self.auth_headers,
-            **self._custom_headers,
-        }
-
-    @property
-    def custom_auth(self) -> httpx.Auth | None:
-        return None
-
-    @property
-    def auth_headers(self):
-        return {}
-
-    def _prepare_headers(self, options: FinalRequestOptions) -> httpx.Headers:
-        custom_headers = options.headers or {}
-        headers_dict = _merge_mappings(self._default_headers, custom_headers)
-
-        httpx_headers = httpx.Headers(headers_dict)
-
-        return httpx_headers
-
-    def _remaining_retries(
-        self,
-        remaining_retries: Optional[int],
-        options: FinalRequestOptions,
-    ) -> int:
-        return remaining_retries if remaining_retries is not None else options.get_max_retries(self.max_retries)
-
-    def _calculate_retry_timeout(
-        self,
-        remaining_retries: int,
-        options: FinalRequestOptions,
-        response_headers: Optional[httpx.Headers] = None,
-    ) -> float:
-        max_retries = options.get_max_retries(self.max_retries)
-
-        # If the API asks us to wait a certain amount of time (and it's a reasonable amount), just do what it says.
-        # retry_after = self._parse_retry_after_header(response_headers)
-        # if retry_after is not None and 0 < retry_after <= 60:
-        #     return retry_after
-
-        nb_retries = max_retries - remaining_retries
-
-        # Apply exponential backoff, but not more than the max.
-        sleep_seconds = min(INITIAL_RETRY_DELAY * pow(2.0, nb_retries), MAX_RETRY_DELAY)
-
-        # Apply some jitter, plus-or-minus half a second.
-        jitter = 1 - 0.25 * random()
-        timeout = sleep_seconds * jitter
-        return max(timeout, 0)
-
-    def _build_request(self, options: FinalRequestOptions) -> httpx.Request:
-        kwargs: dict[str, Any] = {}
-        headers = self._prepare_headers(options)
-        url = self._prepare_url(options.url)
-        json_data = options.json_data
-        if options.extra_json is not None:
-            if json_data is None:
-                json_data = cast(Body, options.extra_json)
-            elif is_mapping(json_data):
-                json_data = _merge_mappings(json_data, options.extra_json)
-            else:
-                raise RuntimeError(f"Unexpected JSON data type, {type(json_data)}, cannot merge with `extra_body`")
-
-        content_type = headers.get("Content-Type")
-        # multipart/form-data; boundary=---abc--
-        if headers.get("Content-Type") == "multipart/form-data":
-            if "boundary" not in content_type:
-                # only remove the header if the boundary hasn't been explicitly set
-                # as the caller doesn't want httpx to come up with their own boundary
-                headers.pop("Content-Type")
-
-            if json_data:
-                kwargs["data"] = self._make_multipartform(json_data)
-
-        return self._client.build_request(
-            headers=headers,
-            timeout=self.timeout if isinstance(options.timeout, NotGiven) else options.timeout,
-            method=options.method,
-            url=url,
-            json=json_data,
-            files=options.files,
-            params=options.params,
-            **kwargs,
-        )
-
-    def _object_to_formdata(self, key: str, value: Data | Mapping[object, object]) -> list[tuple[str, str]]:
-        items = []
-
-        if isinstance(value, Mapping):
-            for k, v in value.items():
-                items.extend(self._object_to_formdata(f"{key}[{k}]", v))
-            return items
-        if isinstance(value, list | tuple):
-            for v in value:
-                items.extend(self._object_to_formdata(key + "[]", v))
-            return items
-
-        def _primitive_value_to_str(val) -> str:
-            # copied from httpx
-            if val is True:
-                return "true"
-            elif val is False:
-                return "false"
-            elif val is None:
-                return ""
-            return str(val)
-
-        str_data = _primitive_value_to_str(value)
-
-        if not str_data:
-            return []
-        return [(key, str_data)]
-
-    def _make_multipartform(self, data: Mapping[object, object]) -> dict[str, object]:
-        items = flatten(list(starmap(self._object_to_formdata, data.items())))
-
-        serialized: dict[str, object] = {}
-        for key, value in items:
-            if key in serialized:
-                raise ValueError(f"存在重复的键: {key};")
-            serialized[key] = value
-        return serialized
-
-    def _process_response_data(
-        self,
-        *,
-        data: object,
-        cast_type: type[ResponseT],
-        response: httpx.Response,
-    ) -> ResponseT:
-        if data is None:
-            return cast(ResponseT, None)
-
-        if cast_type is object:
-            return cast(ResponseT, data)
-
-        try:
-            if inspect.isclass(cast_type) and issubclass(cast_type, ModelBuilderProtocol):
-                return cast(ResponseT, cast_type.build(response=response, data=data))
-
-            if self._strict_response_validation:
-                return cast(ResponseT, validate_type(type_=cast_type, value=data))
-
-            return cast(ResponseT, construct_type(type_=cast_type, value=data))
-        except pydantic.ValidationError as err:
-            raise APIResponseValidationError(response=response, json_data=data) from err
-
-    def _should_stream_response_body(self, request: httpx.Request) -> bool:
-        return request.headers.get(RAW_RESPONSE_HEADER) == "stream"  # type: ignore[no-any-return]
-
-    def _should_retry(self, response: httpx.Response) -> bool:
-        # Note: this is not a standard header
-        should_retry_header = response.headers.get("x-should-retry")
-
-        # If the server explicitly says whether or not to retry, obey.
-        if should_retry_header == "true":
-            log.debug("Retrying as header `x-should-retry` is set to `true`")
-            return True
-        if should_retry_header == "false":
-            log.debug("Not retrying as header `x-should-retry` is set to `false`")
-            return False
-
-        # Retry on request timeouts.
-        if response.status_code == 408:
-            log.debug("Retrying due to status code %i", response.status_code)
-            return True
-
-        # Retry on lock timeouts.
-        if response.status_code == 409:
-            log.debug("Retrying due to status code %i", response.status_code)
-            return True
-
-        # Retry on rate limits.
-        if response.status_code == 429:
-            log.debug("Retrying due to status code %i", response.status_code)
-            return True
-
-        # Retry internal errors.
-        if response.status_code >= 500:
-            log.debug("Retrying due to status code %i", response.status_code)
-            return True
-
-        log.debug("Not retrying")
-        return False
-
-    def is_closed(self) -> bool:
-        return self._client.is_closed
-
-    def close(self):
-        self._client.close()
-
-    def __enter__(self):
-        return self
-
-    def __exit__(self, exc_type, exc_val, exc_tb):
-        self.close()
-
-    def request(
-        self,
-        cast_type: type[ResponseT],
-        options: FinalRequestOptions,
-        remaining_retries: Optional[int] = None,
-        *,
-        stream: bool = False,
-        stream_cls: type[StreamResponse] | None = None,
-    ) -> ResponseT | StreamResponse:
-        return self._request(
-            cast_type=cast_type,
-            options=options,
-            stream=stream,
-            stream_cls=stream_cls,
-            remaining_retries=remaining_retries,
-        )
-
-    def _request(
-        self,
-        *,
-        cast_type: type[ResponseT],
-        options: FinalRequestOptions,
-        remaining_retries: int | None,
-        stream: bool,
-        stream_cls: type[StreamResponse] | None,
-    ) -> ResponseT | StreamResponse:
-        retries = self._remaining_retries(remaining_retries, options)
-        request = self._build_request(options)
-
-        kwargs: HttpxSendArgs = {}
-        if self.custom_auth is not None:
-            kwargs["auth"] = self.custom_auth
-        try:
-            response = self._client.send(
-                request,
-                stream=stream or self._should_stream_response_body(request=request),
-                **kwargs,
-            )
-        except httpx.TimeoutException as err:
-            log.debug("Encountered httpx.TimeoutException", exc_info=True)
-
-            if retries > 0:
-                return self._retry_request(
-                    options,
-                    cast_type,
-                    retries,
-                    stream=stream,
-                    stream_cls=stream_cls,
-                    response_headers=None,
-                )
-
-            log.debug("Raising timeout error")
-            raise APITimeoutError(request=request) from err
-        except Exception as err:
-            log.debug("Encountered Exception", exc_info=True)
-
-            if retries > 0:
-                return self._retry_request(
-                    options,
-                    cast_type,
-                    retries,
-                    stream=stream,
-                    stream_cls=stream_cls,
-                    response_headers=None,
-                )
-
-            log.debug("Raising connection error")
-            raise APIConnectionError(request=request) from err
-
-        log.debug(
-            'HTTP Request: %s %s "%i %s"', request.method, request.url, response.status_code, response.reason_phrase
-        )
-
-        try:
-            response.raise_for_status()
-        except httpx.HTTPStatusError as err:  # thrown on 4xx and 5xx status code
-            log.debug("Encountered httpx.HTTPStatusError", exc_info=True)
-
-            if retries > 0 and self._should_retry(err.response):
-                err.response.close()
-                return self._retry_request(
-                    options,
-                    cast_type,
-                    retries,
-                    err.response.headers,
-                    stream=stream,
-                    stream_cls=stream_cls,
-                )
-
-            # If the response is streamed then we need to explicitly read the response
-            # to completion before attempting to access the response text.
-            if not err.response.is_closed:
-                err.response.read()
-
-            log.debug("Re-raising status error")
-            raise self._make_status_error(err.response) from None
-
-        # return self._parse_response(
-        #     cast_type=cast_type,
-        #     options=options,
-        #     response=response,
-        #     stream=stream,
-        #     stream_cls=stream_cls,
-        # )
-        return self._process_response(
-            cast_type=cast_type,
-            options=options,
-            response=response,
-            stream=stream,
-            stream_cls=stream_cls,
-        )
-
-    def _retry_request(
-        self,
-        options: FinalRequestOptions,
-        cast_type: type[ResponseT],
-        remaining_retries: int,
-        response_headers: httpx.Headers | None,
-        *,
-        stream: bool,
-        stream_cls: type[StreamResponse] | None,
-    ) -> ResponseT | StreamResponse:
-        remaining = remaining_retries - 1
-        if remaining == 1:
-            log.debug("1 retry left")
-        else:
-            log.debug("%i retries left", remaining)
-
-        timeout = self._calculate_retry_timeout(remaining, options, response_headers)
-        log.info("Retrying request to %s in %f seconds", options.url, timeout)
-
-        # In a synchronous context we are blocking the entire thread. Up to the library user to run the client in a
-        # different thread if necessary.
-        time.sleep(timeout)
-
-        return self._request(
-            options=options,
-            cast_type=cast_type,
-            remaining_retries=remaining,
-            stream=stream,
-            stream_cls=stream_cls,
-        )
-
-    def _process_response(
-        self,
-        *,
-        cast_type: type[ResponseT],
-        options: FinalRequestOptions,
-        response: httpx.Response,
-        stream: bool,
-        stream_cls: type[StreamResponse] | None,
-    ) -> ResponseT:
-        # _legacy_response with raw_response_header to parser method
-        if response.request.headers.get(RAW_RESPONSE_HEADER) == "true":
-            return cast(
-                ResponseT,
-                LegacyAPIResponse(
-                    raw=response,
-                    client=self,
-                    cast_type=cast_type,
-                    stream=stream,
-                    stream_cls=stream_cls,
-                    options=options,
-                ),
-            )
-
-        origin = get_origin(cast_type) or cast_type
-
-        if inspect.isclass(origin) and issubclass(origin, BaseAPIResponse):
-            if not issubclass(origin, APIResponse):
-                raise TypeError(f"API Response types must subclass {APIResponse}; Received {origin}")
-
-            response_cls = cast("type[BaseAPIResponse[Any]]", cast_type)
-            return cast(
-                ResponseT,
-                response_cls(
-                    raw=response,
-                    client=self,
-                    cast_type=extract_response_type(response_cls),
-                    stream=stream,
-                    stream_cls=stream_cls,
-                    options=options,
-                ),
-            )
-
-        if cast_type == httpx.Response:
-            return cast(ResponseT, response)
-
-        api_response = APIResponse(
-            raw=response,
-            client=self,
-            cast_type=cast("type[ResponseT]", cast_type),  # pyright: ignore[reportUnnecessaryCast]
-            stream=stream,
-            stream_cls=stream_cls,
-            options=options,
-        )
-        if bool(response.request.headers.get(RAW_RESPONSE_HEADER)):
-            return cast(ResponseT, api_response)
-
-        return api_response.parse()
-
-    def _request_api_list(
-        self,
-        model: type[object],
-        page: type[SyncPageT],
-        options: FinalRequestOptions,
-    ) -> SyncPageT:
-        def _parser(resp: SyncPageT) -> SyncPageT:
-            resp._set_private_attributes(
-                client=self,
-                model=model,
-                options=options,
-            )
-            return resp
-
-        options.post_parser = _parser
-
-        return self.request(page, options, stream=False)
-
-    @overload
-    def get(
-        self,
-        path: str,
-        *,
-        cast_type: type[ResponseT],
-        options: UserRequestInput = {},
-        stream: Literal[False] = False,
-    ) -> ResponseT: ...
-
-    @overload
-    def get(
-        self,
-        path: str,
-        *,
-        cast_type: type[ResponseT],
-        options: UserRequestInput = {},
-        stream: Literal[True],
-        stream_cls: type[StreamResponse],
-    ) -> StreamResponse: ...
-
-    @overload
-    def get(
-        self,
-        path: str,
-        *,
-        cast_type: type[ResponseT],
-        options: UserRequestInput = {},
-        stream: bool,
-        stream_cls: type[StreamResponse] | None = None,
-    ) -> ResponseT | StreamResponse: ...
-
-    def get(
-        self,
-        path: str,
-        *,
-        cast_type: type[ResponseT],
-        options: UserRequestInput = {},
-        stream: bool = False,
-        stream_cls: type[StreamResponse] | None = None,
-    ) -> ResponseT:
-        opts = FinalRequestOptions.construct(method="get", url=path, **options)
-        return cast(ResponseT, self.request(cast_type, opts, stream=stream, stream_cls=stream_cls))
-
-    @overload
-    def post(
-        self,
-        path: str,
-        *,
-        cast_type: type[ResponseT],
-        body: Body | None = None,
-        options: UserRequestInput = {},
-        files: RequestFiles | None = None,
-        stream: Literal[False] = False,
-    ) -> ResponseT: ...
-
-    @overload
-    def post(
-        self,
-        path: str,
-        *,
-        cast_type: type[ResponseT],
-        body: Body | None = None,
-        options: UserRequestInput = {},
-        files: RequestFiles | None = None,
-        stream: Literal[True],
-        stream_cls: type[StreamResponse],
-    ) -> StreamResponse: ...
-
-    @overload
-    def post(
-        self,
-        path: str,
-        *,
-        cast_type: type[ResponseT],
-        body: Body | None = None,
-        options: UserRequestInput = {},
-        files: RequestFiles | None = None,
-        stream: bool,
-        stream_cls: type[StreamResponse] | None = None,
-    ) -> ResponseT | StreamResponse: ...
-
-    def post(
-        self,
-        path: str,
-        *,
-        cast_type: type[ResponseT],
-        body: Body | None = None,
-        options: UserRequestInput = {},
-        files: RequestFiles | None = None,
-        stream: bool = False,
-        stream_cls: type[StreamResponse[Any]] | None = None,
-    ) -> ResponseT | StreamResponse:
-        opts = FinalRequestOptions.construct(
-            method="post", url=path, json_data=body, files=to_httpx_files(files), **options
-        )
-
-        return cast(ResponseT, self.request(cast_type, opts, stream=stream, stream_cls=stream_cls))
-
-    def patch(
-        self,
-        path: str,
-        *,
-        cast_type: type[ResponseT],
-        body: Body | None = None,
-        options: UserRequestInput = {},
-    ) -> ResponseT:
-        opts = FinalRequestOptions.construct(method="patch", url=path, json_data=body, **options)
-
-        return self.request(
-            cast_type=cast_type,
-            options=opts,
-        )
-
-    def put(
-        self,
-        path: str,
-        *,
-        cast_type: type[ResponseT],
-        body: Body | None = None,
-        options: UserRequestInput = {},
-        files: RequestFiles | None = None,
-    ) -> ResponseT | StreamResponse:
-        opts = FinalRequestOptions.construct(
-            method="put", url=path, json_data=body, files=to_httpx_files(files), **options
-        )
-
-        return self.request(
-            cast_type=cast_type,
-            options=opts,
-        )
-
-    def delete(
-        self,
-        path: str,
-        *,
-        cast_type: type[ResponseT],
-        body: Body | None = None,
-        options: UserRequestInput = {},
-    ) -> ResponseT | StreamResponse:
-        opts = FinalRequestOptions.construct(method="delete", url=path, json_data=body, **options)
-
-        return self.request(
-            cast_type=cast_type,
-            options=opts,
-        )
-
-    def get_api_list(
-        self,
-        path: str,
-        *,
-        model: type[object],
-        page: type[SyncPageT],
-        body: Body | None = None,
-        options: UserRequestInput = {},
-        method: str = "get",
-    ) -> SyncPageT:
-        opts = FinalRequestOptions.construct(method=method, url=path, json_data=body, **options)
-        return self._request_api_list(model, page, opts)
-
-    def _make_status_error(self, response) -> APIStatusError:
-        response_text = response.text.strip()
-        status_code = response.status_code
-        error_msg = f"Error code: {status_code}, with error text {response_text}"
-
-        if status_code == 400:
-            return _errors.APIRequestFailedError(message=error_msg, response=response)
-        elif status_code == 401:
-            return _errors.APIAuthenticationError(message=error_msg, response=response)
-        elif status_code == 429:
-            return _errors.APIReachLimitError(message=error_msg, response=response)
-        elif status_code == 500:
-            return _errors.APIInternalError(message=error_msg, response=response)
-        elif status_code == 503:
-            return _errors.APIServerFlowExceedError(message=error_msg, response=response)
-        return APIStatusError(message=error_msg, response=response)
-
-
-def make_request_options(
-    *,
-    query: Query | None = None,
-    extra_headers: Headers | None = None,
-    extra_query: Query | None = None,
-    extra_body: Body | None = None,
-    timeout: float | httpx.Timeout | None | NotGiven = NOT_GIVEN,
-    post_parser: PostParser | NotGiven = NOT_GIVEN,
-) -> UserRequestInput:
-    """Create a dict of type RequestOptions without keys of NotGiven values."""
-    options: UserRequestInput = {}
-    if extra_headers is not None:
-        options["headers"] = extra_headers
-
-    if extra_body is not None:
-        options["extra_json"] = cast(AnyMapping, extra_body)
-
-    if query is not None:
-        options["params"] = query
-
-    if extra_query is not None:
-        options["params"] = {**options.get("params", {}), **extra_query}
-
-    if not isinstance(timeout, NotGiven):
-        options["timeout"] = timeout
-
-    if is_given(post_parser):
-        # internal
-        options["post_parser"] = post_parser  # type: ignore
-
-    return options
-
-
-def _merge_mappings(
-    obj1: Mapping[_T_co, Union[_T, Omit]],
-    obj2: Mapping[_T_co, Union[_T, Omit]],
-) -> dict[_T_co, _T]:
-    """Merge two mappings of the same type, removing any values that are instances of `Omit`.
-
-    In cases with duplicate keys the second mapping takes precedence.
-    """
-    merged = {**obj1, **obj2}
-    return {key: value for key, value in merged.items() if not isinstance(value, Omit)}
diff --git a/api/core/model_runtime/model_providers/zhipuai/zhipuai_sdk/core/_jwt_token.py b/api/core/model_runtime/model_providers/zhipuai/zhipuai_sdk/core/_jwt_token.py
deleted file mode 100644
index 21f158a5f4525..0000000000000
--- a/api/core/model_runtime/model_providers/zhipuai/zhipuai_sdk/core/_jwt_token.py
+++ /dev/null
@@ -1,31 +0,0 @@
-import time
-
-import cachetools.func
-import jwt
-
-# 缓存时间 3分钟
-CACHE_TTL_SECONDS = 3 * 60
-
-# token 有效期比缓存时间 多30秒
-API_TOKEN_TTL_SECONDS = CACHE_TTL_SECONDS + 30
-
-
-@cachetools.func.ttl_cache(maxsize=10, ttl=CACHE_TTL_SECONDS)
-def generate_token(apikey: str):
-    try:
-        api_key, secret = apikey.split(".")
-    except Exception as e:
-        raise Exception("invalid api_key", e)
-
-    payload = {
-        "api_key": api_key,
-        "exp": int(round(time.time() * 1000)) + API_TOKEN_TTL_SECONDS * 1000,
-        "timestamp": int(round(time.time() * 1000)),
-    }
-    ret = jwt.encode(
-        payload,
-        secret,
-        algorithm="HS256",
-        headers={"alg": "HS256", "sign_type": "SIGN"},
-    )
-    return ret
diff --git a/api/core/model_runtime/model_providers/zhipuai/zhipuai_sdk/core/_legacy_binary_response.py b/api/core/model_runtime/model_providers/zhipuai/zhipuai_sdk/core/_legacy_binary_response.py
deleted file mode 100644
index 51623bd860951..0000000000000
--- a/api/core/model_runtime/model_providers/zhipuai/zhipuai_sdk/core/_legacy_binary_response.py
+++ /dev/null
@@ -1,207 +0,0 @@
-from __future__ import annotations
-
-import os
-from collections.abc import AsyncIterator, Iterator
-from typing import Any
-
-import httpx
-
-
-class HttpxResponseContent:
-    @property
-    def content(self) -> bytes:
-        raise NotImplementedError("This method is not implemented for this class.")
-
-    @property
-    def text(self) -> str:
-        raise NotImplementedError("This method is not implemented for this class.")
-
-    @property
-    def encoding(self) -> str | None:
-        raise NotImplementedError("This method is not implemented for this class.")
-
-    @property
-    def charset_encoding(self) -> str | None:
-        raise NotImplementedError("This method is not implemented for this class.")
-
-    def json(self, **kwargs: Any) -> Any:
-        raise NotImplementedError("This method is not implemented for this class.")
-
-    def read(self) -> bytes:
-        raise NotImplementedError("This method is not implemented for this class.")
-
-    def iter_bytes(self, chunk_size: int | None = None) -> Iterator[bytes]:
-        raise NotImplementedError("This method is not implemented for this class.")
-
-    def iter_text(self, chunk_size: int | None = None) -> Iterator[str]:
-        raise NotImplementedError("This method is not implemented for this class.")
-
-    def iter_lines(self) -> Iterator[str]:
-        raise NotImplementedError("This method is not implemented for this class.")
-
-    def iter_raw(self, chunk_size: int | None = None) -> Iterator[bytes]:
-        raise NotImplementedError("This method is not implemented for this class.")
-
-    def write_to_file(
-        self,
-        file: str | os.PathLike[str],
-    ) -> None:
-        raise NotImplementedError("This method is not implemented for this class.")
-
-    def stream_to_file(
-        self,
-        file: str | os.PathLike[str],
-        *,
-        chunk_size: int | None = None,
-    ) -> None:
-        raise NotImplementedError("This method is not implemented for this class.")
-
-    def close(self) -> None:
-        raise NotImplementedError("This method is not implemented for this class.")
-
-    async def aread(self) -> bytes:
-        raise NotImplementedError("This method is not implemented for this class.")
-
-    async def aiter_bytes(self, chunk_size: int | None = None) -> AsyncIterator[bytes]:
-        raise NotImplementedError("This method is not implemented for this class.")
-
-    async def aiter_text(self, chunk_size: int | None = None) -> AsyncIterator[str]:
-        raise NotImplementedError("This method is not implemented for this class.")
-
-    async def aiter_lines(self) -> AsyncIterator[str]:
-        raise NotImplementedError("This method is not implemented for this class.")
-
-    async def aiter_raw(self, chunk_size: int | None = None) -> AsyncIterator[bytes]:
-        raise NotImplementedError("This method is not implemented for this class.")
-
-    async def astream_to_file(
-        self,
-        file: str | os.PathLike[str],
-        *,
-        chunk_size: int | None = None,
-    ) -> None:
-        raise NotImplementedError("This method is not implemented for this class.")
-
-    async def aclose(self) -> None:
-        raise NotImplementedError("This method is not implemented for this class.")
-
-
-class HttpxBinaryResponseContent(HttpxResponseContent):
-    response: httpx.Response
-
-    def __init__(self, response: httpx.Response) -> None:
-        self.response = response
-
-    @property
-    def content(self) -> bytes:
-        return self.response.content
-
-    @property
-    def encoding(self) -> str | None:
-        return self.response.encoding
-
-    @property
-    def charset_encoding(self) -> str | None:
-        return self.response.charset_encoding
-
-    def read(self) -> bytes:
-        return self.response.read()
-
-    def text(self) -> str:
-        raise NotImplementedError("Not implemented for binary response content")
-
-    def json(self, **kwargs: Any) -> Any:
-        raise NotImplementedError("Not implemented for binary response content")
-
-    def iter_text(self, chunk_size: int | None = None) -> Iterator[str]:
-        raise NotImplementedError("Not implemented for binary response content")
-
-    def iter_lines(self) -> Iterator[str]:
-        raise NotImplementedError("Not implemented for binary response content")
-
-    async def aiter_text(self, chunk_size: int | None = None) -> AsyncIterator[str]:
-        raise NotImplementedError("Not implemented for binary response content")
-
-    async def aiter_lines(self) -> AsyncIterator[str]:
-        raise NotImplementedError("Not implemented for binary response content")
-
-    def iter_bytes(self, chunk_size: int | None = None) -> Iterator[bytes]:
-        return self.response.iter_bytes(chunk_size)
-
-    def iter_raw(self, chunk_size: int | None = None) -> Iterator[bytes]:
-        return self.response.iter_raw(chunk_size)
-
-    def write_to_file(
-        self,
-        file: str | os.PathLike[str],
-    ) -> None:
-        """Write the output to the given file.
-
-        Accepts a filename or any path-like object, e.g. pathlib.Path
-
-        Note: if you want to stream the data to the file instead of writing
-        all at once then you should use `.with_streaming_response` when making
-        the API request, e.g. `client.with_streaming_response.foo().stream_to_file('my_filename.txt')`
-        """
-        with open(file, mode="wb") as f:
-            for data in self.response.iter_bytes():
-                f.write(data)
-
-    def stream_to_file(
-        self,
-        file: str | os.PathLike[str],
-        *,
-        chunk_size: int | None = None,
-    ) -> None:
-        with open(file, mode="wb") as f:
-            for data in self.response.iter_bytes(chunk_size):
-                f.write(data)
-
-    def close(self) -> None:
-        return self.response.close()
-
-    async def aread(self) -> bytes:
-        return await self.response.aread()
-
-    async def aiter_bytes(self, chunk_size: int | None = None) -> AsyncIterator[bytes]:
-        return self.response.aiter_bytes(chunk_size)
-
-    async def aiter_raw(self, chunk_size: int | None = None) -> AsyncIterator[bytes]:
-        return self.response.aiter_raw(chunk_size)
-
-    async def astream_to_file(
-        self,
-        file: str | os.PathLike[str],
-        *,
-        chunk_size: int | None = None,
-    ) -> None:
-        path = anyio.Path(file)
-        async with await path.open(mode="wb") as f:
-            async for data in self.response.aiter_bytes(chunk_size):
-                await f.write(data)
-
-    async def aclose(self) -> None:
-        return await self.response.aclose()
-
-
-class HttpxTextBinaryResponseContent(HttpxBinaryResponseContent):
-    response: httpx.Response
-
-    @property
-    def text(self) -> str:
-        return self.response.text
-
-    def json(self, **kwargs: Any) -> Any:
-        return self.response.json(**kwargs)
-
-    def iter_text(self, chunk_size: int | None = None) -> Iterator[str]:
-        return self.response.iter_text(chunk_size)
-
-    def iter_lines(self) -> Iterator[str]:
-        return self.response.iter_lines()
-
-    async def aiter_text(self, chunk_size: int | None = None) -> AsyncIterator[str]:
-        return self.response.aiter_text(chunk_size)
-
-    async def aiter_lines(self) -> AsyncIterator[str]:
-        return self.response.aiter_lines()
diff --git a/api/core/model_runtime/model_providers/zhipuai/zhipuai_sdk/core/_legacy_response.py b/api/core/model_runtime/model_providers/zhipuai/zhipuai_sdk/core/_legacy_response.py
deleted file mode 100644
index 51bf21bcdc17a..0000000000000
--- a/api/core/model_runtime/model_providers/zhipuai/zhipuai_sdk/core/_legacy_response.py
+++ /dev/null
@@ -1,341 +0,0 @@
-from __future__ import annotations
-
-import datetime
-import functools
-import inspect
-import logging
-from collections.abc import Callable
-from typing import TYPE_CHECKING, Any, Generic, TypeVar, Union, cast, get_origin, overload
-
-import httpx
-import pydantic
-from typing_extensions import ParamSpec, override
-
-from ._base_models import BaseModel, is_basemodel
-from ._base_type import NoneType
-from ._constants import RAW_RESPONSE_HEADER
-from ._errors import APIResponseValidationError
-from ._legacy_binary_response import HttpxResponseContent, HttpxTextBinaryResponseContent
-from ._sse_client import StreamResponse, extract_stream_chunk_type, is_stream_class_type
-from ._utils import extract_type_arg, is_annotated_type, is_given
-
-if TYPE_CHECKING:
-    from ._http_client import HttpClient
-    from ._request_opt import FinalRequestOptions
-
-P = ParamSpec("P")
-R = TypeVar("R")
-_T = TypeVar("_T")
-
-log: logging.Logger = logging.getLogger(__name__)
-
-
-class LegacyAPIResponse(Generic[R]):
-    """This is a legacy class as it will be replaced by `APIResponse`
-    and `AsyncAPIResponse` in the `_response.py` file in the next major
-    release.
-
-    For the sync client this will mostly be the same with the exception
-    of `content` & `text` will be methods instead of properties. In the
-    async client, all methods will be async.
-
-    A migration script will be provided & the migration in general should
-    be smooth.
-    """
-
-    _cast_type: type[R]
-    _client: HttpClient
-    _parsed_by_type: dict[type[Any], Any]
-    _stream: bool
-    _stream_cls: type[StreamResponse[Any]] | None
-    _options: FinalRequestOptions
-
-    http_response: httpx.Response
-
-    def __init__(
-        self,
-        *,
-        raw: httpx.Response,
-        cast_type: type[R],
-        client: HttpClient,
-        stream: bool,
-        stream_cls: type[StreamResponse[Any]] | None,
-        options: FinalRequestOptions,
-    ) -> None:
-        self._cast_type = cast_type
-        self._client = client
-        self._parsed_by_type = {}
-        self._stream = stream
-        self._stream_cls = stream_cls
-        self._options = options
-        self.http_response = raw
-
-    @property
-    def request_id(self) -> str | None:
-        return self.http_response.headers.get("x-request-id")  # type: ignore[no-any-return]
-
-    @overload
-    def parse(self, *, to: type[_T]) -> _T: ...
-
-    @overload
-    def parse(self) -> R: ...
-
-    def parse(self, *, to: type[_T] | None = None) -> R | _T:
-        """Returns the rich python representation of this response's data.
-
-        NOTE: For the async client: this will become a coroutine in the next major version.
-
-        For lower-level control, see `.read()`, `.json()`, `.iter_bytes()`.
-
-        You can customize the type that the response is parsed into through
-        the `to` argument, e.g.
-
-        ```py
-        from zhipuai import BaseModel
-
-
-        class MyModel(BaseModel):
-            foo: str
-
-
-        obj = response.parse(to=MyModel)
-        print(obj.foo)
-        ```
-
-        We support parsing:
-          - `BaseModel`
-          - `dict`
-          - `list`
-          - `Union`
-          - `str`
-          - `int`
-          - `float`
-          - `httpx.Response`
-        """
-        cache_key = to if to is not None else self._cast_type
-        cached = self._parsed_by_type.get(cache_key)
-        if cached is not None:
-            return cached  # type: ignore[no-any-return]
-
-        parsed = self._parse(to=to)
-        if is_given(self._options.post_parser):
-            parsed = self._options.post_parser(parsed)
-
-        self._parsed_by_type[cache_key] = parsed
-        return parsed
-
-    @property
-    def headers(self) -> httpx.Headers:
-        return self.http_response.headers
-
-    @property
-    def http_request(self) -> httpx.Request:
-        return self.http_response.request
-
-    @property
-    def status_code(self) -> int:
-        return self.http_response.status_code
-
-    @property
-    def url(self) -> httpx.URL:
-        return self.http_response.url
-
-    @property
-    def method(self) -> str:
-        return self.http_request.method
-
-    @property
-    def content(self) -> bytes:
-        """Return the binary response content.
-
-        NOTE: this will be removed in favour of `.read()` in the
-        next major version.
-        """
-        return self.http_response.content
-
-    @property
-    def text(self) -> str:
-        """Return the decoded response content.
-
-        NOTE: this will be turned into a method in the next major version.
-        """
-        return self.http_response.text
-
-    @property
-    def http_version(self) -> str:
-        return self.http_response.http_version
-
-    @property
-    def is_closed(self) -> bool:
-        return self.http_response.is_closed
-
-    @property
-    def elapsed(self) -> datetime.timedelta:
-        """The time taken for the complete request/response cycle to complete."""
-        return self.http_response.elapsed
-
-    def _parse(self, *, to: type[_T] | None = None) -> R | _T:
-        # unwrap `Annotated[T, ...]` -> `T`
-        if to and is_annotated_type(to):
-            to = extract_type_arg(to, 0)
-
-        if self._stream:
-            if to:
-                if not is_stream_class_type(to):
-                    raise TypeError(f"Expected custom parse type to be a subclass of {StreamResponse}")
-
-                return cast(
-                    _T,
-                    to(
-                        cast_type=extract_stream_chunk_type(
-                            to,
-                            failure_message="Expected custom stream type to be passed with a type argument, e.g. StreamResponse[ChunkType]",  # noqa: E501
-                        ),
-                        response=self.http_response,
-                        client=cast(Any, self._client),
-                    ),
-                )
-
-            if self._stream_cls:
-                return cast(
-                    R,
-                    self._stream_cls(
-                        cast_type=extract_stream_chunk_type(self._stream_cls),
-                        response=self.http_response,
-                        client=cast(Any, self._client),
-                    ),
-                )
-
-            stream_cls = cast("type[StreamResponse[Any]] | None", self._client._default_stream_cls)
-            if stream_cls is None:
-                raise MissingStreamClassError()
-
-            return cast(
-                R,
-                stream_cls(
-                    cast_type=self._cast_type,
-                    response=self.http_response,
-                    client=cast(Any, self._client),
-                ),
-            )
-
-        cast_type = to if to is not None else self._cast_type
-
-        # unwrap `Annotated[T, ...]` -> `T`
-        if is_annotated_type(cast_type):
-            cast_type = extract_type_arg(cast_type, 0)
-
-        if cast_type is NoneType:
-            return cast(R, None)
-
-        response = self.http_response
-        if cast_type == str:
-            return cast(R, response.text)
-
-        if cast_type == int:
-            return cast(R, int(response.text))
-
-        if cast_type == float:
-            return cast(R, float(response.text))
-
-        origin = get_origin(cast_type) or cast_type
-
-        if inspect.isclass(origin) and issubclass(origin, HttpxResponseContent):
-            # in the response, e.g. mime file
-            *_, filename = response.headers.get("content-disposition", "").split("filename=")
-            # 判断文件类型是jsonl类型的使用HttpxTextBinaryResponseContent
-            if filename and filename.endswith(".jsonl") or filename and filename.endswith(".xlsx"):
-                return cast(R, HttpxTextBinaryResponseContent(response))
-            else:
-                return cast(R, cast_type(response))  # type: ignore
-
-        if origin == LegacyAPIResponse:
-            raise RuntimeError("Unexpected state - cast_type is `APIResponse`")
-
-        if inspect.isclass(origin) and issubclass(origin, httpx.Response):
-            # Because of the invariance of our ResponseT TypeVar, users can subclass httpx.Response
-            # and pass that class to our request functions. We cannot change the variance to be either
-            # covariant or contravariant as that makes our usage of ResponseT illegal. We could construct
-            # the response class ourselves but that is something that should be supported directly in httpx
-            # as it would be easy to incorrectly construct the Response object due to the multitude of arguments.
-            if cast_type != httpx.Response:
-                raise ValueError("Subclasses of httpx.Response cannot be passed to `cast_type`")
-            return cast(R, response)
-
-        if inspect.isclass(origin) and not issubclass(origin, BaseModel) and issubclass(origin, pydantic.BaseModel):
-            raise TypeError("Pydantic models must subclass our base model type, e.g. `from openai import BaseModel`")
-
-        if (
-            cast_type is not object
-            and origin is not list
-            and origin is not dict
-            and origin is not Union
-            and not issubclass(origin, BaseModel)
-        ):
-            raise RuntimeError(
-                f"Unsupported type, expected {cast_type} to be a subclass of {BaseModel}, {dict}, {list}, {Union}, {NoneType}, {str} or {httpx.Response}."  # noqa: E501
-            )
-
-        # split is required to handle cases where additional information is included
-        # in the response, e.g. application/json; charset=utf-8
-        content_type, *_ = response.headers.get("content-type", "*").split(";")
-        if content_type != "application/json":
-            if is_basemodel(cast_type):
-                try:
-                    data = response.json()
-                except Exception as exc:
-                    log.debug("Could not read JSON from response data due to %s - %s", type(exc), exc)
-                else:
-                    return self._client._process_response_data(
-                        data=data,
-                        cast_type=cast_type,  # type: ignore
-                        response=response,
-                    )
-
-            if self._client._strict_response_validation:
-                raise APIResponseValidationError(
-                    response=response,
-                    message=f"Expected Content-Type response header to be `application/json` but received `{content_type}` instead.",  # noqa: E501
-                    json_data=response.text,
-                )
-
-            # If the API responds with content that isn't JSON then we just return
-            # the (decoded) text without performing any parsing so that you can still
-            # handle the response however you need to.
-            return response.text  # type: ignore
-
-        data = response.json()
-
-        return self._client._process_response_data(
-            data=data,
-            cast_type=cast_type,  # type: ignore
-            response=response,
-        )
-
-    @override
-    def __repr__(self) -> str:
-        return f"<APIResponse [{self.status_code} {self.http_response.reason_phrase}] type={self._cast_type}>"
-
-
-class MissingStreamClassError(TypeError):
-    def __init__(self) -> None:
-        super().__init__(
-            "The `stream` argument was set to `True` but the `stream_cls` argument was not given. See `openai._streaming` for reference",  # noqa: E501
-        )
-
-
-def to_raw_response_wrapper(func: Callable[P, R]) -> Callable[P, LegacyAPIResponse[R]]:
-    """Higher order function that takes one of our bound API methods and wraps it
-    to support returning the raw `APIResponse` object directly.
-    """
-
-    @functools.wraps(func)
-    def wrapped(*args: P.args, **kwargs: P.kwargs) -> LegacyAPIResponse[R]:
-        extra_headers: dict[str, str] = {**(cast(Any, kwargs.get("extra_headers")) or {})}
-        extra_headers[RAW_RESPONSE_HEADER] = "true"
-
-        kwargs["extra_headers"] = extra_headers
-
-        return cast(LegacyAPIResponse[R], func(*args, **kwargs))
-
-    return wrapped
diff --git a/api/core/model_runtime/model_providers/zhipuai/zhipuai_sdk/core/_request_opt.py b/api/core/model_runtime/model_providers/zhipuai/zhipuai_sdk/core/_request_opt.py
deleted file mode 100644
index c3b894b3a3d88..0000000000000
--- a/api/core/model_runtime/model_providers/zhipuai/zhipuai_sdk/core/_request_opt.py
+++ /dev/null
@@ -1,97 +0,0 @@
-from __future__ import annotations
-
-from collections.abc import Callable
-from typing import TYPE_CHECKING, Any, ClassVar, Union, cast
-
-import pydantic.generics
-from httpx import Timeout
-from typing_extensions import Required, TypedDict, Unpack, final
-
-from ._base_compat import PYDANTIC_V2, ConfigDict
-from ._base_type import AnyMapping, Body, Headers, HttpxRequestFiles, NotGiven, Query
-from ._constants import RAW_RESPONSE_HEADER
-from ._utils import is_given, strip_not_given
-
-
-class UserRequestInput(TypedDict, total=False):
-    headers: Headers
-    max_retries: int
-    timeout: float | Timeout | None
-    params: Query
-    extra_json: AnyMapping
-
-
-class FinalRequestOptionsInput(TypedDict, total=False):
-    method: Required[str]
-    url: Required[str]
-    params: Query
-    headers: Headers
-    max_retries: int
-    timeout: float | Timeout | None
-    files: HttpxRequestFiles | None
-    json_data: Body
-    extra_json: AnyMapping
-
-
-@final
-class FinalRequestOptions(pydantic.BaseModel):
-    method: str
-    url: str
-    params: Query = {}
-    headers: Union[Headers, NotGiven] = NotGiven()
-    max_retries: Union[int, NotGiven] = NotGiven()
-    timeout: Union[float, Timeout, None, NotGiven] = NotGiven()
-    files: Union[HttpxRequestFiles, None] = None
-    idempotency_key: Union[str, None] = None
-    post_parser: Union[Callable[[Any], Any], NotGiven] = NotGiven()
-
-    # It should be noted that we cannot use `json` here as that would override
-    # a BaseModel method in an incompatible fashion.
-    json_data: Union[Body, None] = None
-    extra_json: Union[AnyMapping, None] = None
-
-    if PYDANTIC_V2:
-        model_config: ClassVar[ConfigDict] = ConfigDict(arbitrary_types_allowed=True)
-    else:
-
-        class Config(pydantic.BaseConfig):  # pyright: ignore[reportDeprecated]
-            arbitrary_types_allowed: bool = True
-
-    def get_max_retries(self, max_retries: int) -> int:
-        if isinstance(self.max_retries, NotGiven):
-            return max_retries
-        return self.max_retries
-
-    def _strip_raw_response_header(self) -> None:
-        if not is_given(self.headers):
-            return
-
-        if self.headers.get(RAW_RESPONSE_HEADER):
-            self.headers = {**self.headers}
-            self.headers.pop(RAW_RESPONSE_HEADER)
-
-    # override the `construct` method so that we can run custom transformations.
-    # this is necessary as we don't want to do any actual runtime type checking
-    # (which means we can't use validators) but we do want to ensure that `NotGiven`
-    # values are not present
-    #
-    # type ignore required because we're adding explicit types to `**values`
-    @classmethod
-    def construct(  # type: ignore
-        cls,
-        _fields_set: set[str] | None = None,
-        **values: Unpack[UserRequestInput],
-    ) -> FinalRequestOptions:
-        kwargs: dict[str, Any] = {
-            # we unconditionally call `strip_not_given` on any value
-            # as it will just ignore any non-mapping types
-            key: strip_not_given(value)
-            for key, value in values.items()
-        }
-        if PYDANTIC_V2:
-            return super().model_construct(_fields_set, **kwargs)
-        return cast(FinalRequestOptions, super().construct(_fields_set, **kwargs))  # pyright: ignore[reportDeprecated]
-
-    if not TYPE_CHECKING:
-        # type checkers incorrectly complain about this assignment
-        model_construct = construct
diff --git a/api/core/model_runtime/model_providers/zhipuai/zhipuai_sdk/core/_response.py b/api/core/model_runtime/model_providers/zhipuai/zhipuai_sdk/core/_response.py
deleted file mode 100644
index 92e601805569f..0000000000000
--- a/api/core/model_runtime/model_providers/zhipuai/zhipuai_sdk/core/_response.py
+++ /dev/null
@@ -1,398 +0,0 @@
-from __future__ import annotations
-
-import datetime
-import inspect
-import logging
-from collections.abc import Iterator
-from typing import TYPE_CHECKING, Any, Generic, TypeVar, Union, cast, get_origin, overload
-
-import httpx
-import pydantic
-from typing_extensions import ParamSpec, override
-
-from ._base_models import BaseModel, is_basemodel
-from ._base_type import NoneType
-from ._errors import APIResponseValidationError, ZhipuAIError
-from ._sse_client import StreamResponse, extract_stream_chunk_type, is_stream_class_type
-from ._utils import extract_type_arg, extract_type_var_from_base, is_annotated_type, is_given
-
-if TYPE_CHECKING:
-    from ._http_client import HttpClient
-    from ._request_opt import FinalRequestOptions
-
-P = ParamSpec("P")
-R = TypeVar("R")
-_T = TypeVar("_T")
-_APIResponseT = TypeVar("_APIResponseT", bound="APIResponse[Any]")
-log: logging.Logger = logging.getLogger(__name__)
-
-
-class BaseAPIResponse(Generic[R]):
-    _cast_type: type[R]
-    _client: HttpClient
-    _parsed_by_type: dict[type[Any], Any]
-    _is_sse_stream: bool
-    _stream_cls: type[StreamResponse[Any]]
-    _options: FinalRequestOptions
-    http_response: httpx.Response
-
-    def __init__(
-        self,
-        *,
-        raw: httpx.Response,
-        cast_type: type[R],
-        client: HttpClient,
-        stream: bool,
-        stream_cls: type[StreamResponse[Any]] | None = None,
-        options: FinalRequestOptions,
-    ) -> None:
-        self._cast_type = cast_type
-        self._client = client
-        self._parsed_by_type = {}
-        self._is_sse_stream = stream
-        self._stream_cls = stream_cls
-        self._options = options
-        self.http_response = raw
-
-    def _parse(self, *, to: type[_T] | None = None) -> R | _T:
-        # unwrap `Annotated[T, ...]` -> `T`
-        if to and is_annotated_type(to):
-            to = extract_type_arg(to, 0)
-
-        if self._is_sse_stream:
-            if to:
-                if not is_stream_class_type(to):
-                    raise TypeError(f"Expected custom parse type to be a subclass of {StreamResponse}")
-
-                return cast(
-                    _T,
-                    to(
-                        cast_type=extract_stream_chunk_type(
-                            to,
-                            failure_message="Expected custom stream type to be passed with a type argument, e.g. StreamResponse[ChunkType]",  # noqa: E501
-                        ),
-                        response=self.http_response,
-                        client=cast(Any, self._client),
-                    ),
-                )
-
-            if self._stream_cls:
-                return cast(
-                    R,
-                    self._stream_cls(
-                        cast_type=extract_stream_chunk_type(self._stream_cls),
-                        response=self.http_response,
-                        client=cast(Any, self._client),
-                    ),
-                )
-
-            stream_cls = cast("type[Stream[Any]] | None", self._client._default_stream_cls)
-            if stream_cls is None:
-                raise MissingStreamClassError()
-
-            return cast(
-                R,
-                stream_cls(
-                    cast_type=self._cast_type,
-                    response=self.http_response,
-                    client=cast(Any, self._client),
-                ),
-            )
-
-        cast_type = to if to is not None else self._cast_type
-
-        # unwrap `Annotated[T, ...]` -> `T`
-        if is_annotated_type(cast_type):
-            cast_type = extract_type_arg(cast_type, 0)
-
-        if cast_type is NoneType:
-            return cast(R, None)
-
-        response = self.http_response
-        if cast_type == str:
-            return cast(R, response.text)
-
-        if cast_type == bytes:
-            return cast(R, response.content)
-
-        if cast_type == int:
-            return cast(R, int(response.text))
-
-        if cast_type == float:
-            return cast(R, float(response.text))
-
-        origin = get_origin(cast_type) or cast_type
-
-        # handle the legacy binary response case
-        if inspect.isclass(cast_type) and cast_type.__name__ == "HttpxBinaryResponseContent":
-            return cast(R, cast_type(response))  # type: ignore
-
-        if origin == APIResponse:
-            raise RuntimeError("Unexpected state - cast_type is `APIResponse`")
-
-        if inspect.isclass(origin) and issubclass(origin, httpx.Response):
-            # Because of the invariance of our ResponseT TypeVar, users can subclass httpx.Response
-            # and pass that class to our request functions. We cannot change the variance to be either
-            # covariant or contravariant as that makes our usage of ResponseT illegal. We could construct
-            # the response class ourselves but that is something that should be supported directly in httpx
-            # as it would be easy to incorrectly construct the Response object due to the multitude of arguments.
-            if cast_type != httpx.Response:
-                raise ValueError("Subclasses of httpx.Response cannot be passed to `cast_type`")
-            return cast(R, response)
-
-        if inspect.isclass(origin) and not issubclass(origin, BaseModel) and issubclass(origin, pydantic.BaseModel):
-            raise TypeError("Pydantic models must subclass our base model type, e.g. `from openai import BaseModel`")
-
-        if (
-            cast_type is not object
-            and origin is not list
-            and origin is not dict
-            and origin is not Union
-            and not issubclass(origin, BaseModel)
-        ):
-            raise RuntimeError(
-                f"Unsupported type, expected {cast_type} to be a subclass of {BaseModel}, {dict}, {list}, {Union}, {NoneType}, {str} or {httpx.Response}."  # noqa: E501
-            )
-
-        # split is required to handle cases where additional information is included
-        # in the response, e.g. application/json; charset=utf-8
-        content_type, *_ = response.headers.get("content-type", "*").split(";")
-        if content_type != "application/json":
-            if is_basemodel(cast_type):
-                try:
-                    data = response.json()
-                except Exception as exc:
-                    log.debug("Could not read JSON from response data due to %s - %s", type(exc), exc)
-                else:
-                    return self._client._process_response_data(
-                        data=data,
-                        cast_type=cast_type,  # type: ignore
-                        response=response,
-                    )
-
-            if self._client._strict_response_validation:
-                raise APIResponseValidationError(
-                    response=response,
-                    message=f"Expected Content-Type response header to be `application/json` but received `{content_type}` instead.",  # noqa: E501
-                    json_data=response.text,
-                )
-
-            # If the API responds with content that isn't JSON then we just return
-            # the (decoded) text without performing any parsing so that you can still
-            # handle the response however you need to.
-            return response.text  # type: ignore
-
-        data = response.json()
-
-        return self._client._process_response_data(
-            data=data,
-            cast_type=cast_type,  # type: ignore
-            response=response,
-        )
-
-    @property
-    def headers(self) -> httpx.Headers:
-        return self.http_response.headers
-
-    @property
-    def http_request(self) -> httpx.Request:
-        """Returns the httpx Request instance associated with the current response."""
-        return self.http_response.request
-
-    @property
-    def status_code(self) -> int:
-        return self.http_response.status_code
-
-    @property
-    def url(self) -> httpx.URL:
-        """Returns the URL for which the request was made."""
-        return self.http_response.url
-
-    @property
-    def method(self) -> str:
-        return self.http_request.method
-
-    @property
-    def http_version(self) -> str:
-        return self.http_response.http_version
-
-    @property
-    def elapsed(self) -> datetime.timedelta:
-        """The time taken for the complete request/response cycle to complete."""
-        return self.http_response.elapsed
-
-    @property
-    def is_closed(self) -> bool:
-        """Whether or not the response body has been closed.
-
-        If this is False then there is response data that has not been read yet.
-        You must either fully consume the response body or call `.close()`
-        before discarding the response to prevent resource leaks.
-        """
-        return self.http_response.is_closed
-
-    @override
-    def __repr__(self) -> str:
-        return f"<{self.__class__.__name__} [{self.status_code} {self.http_response.reason_phrase}] type={self._cast_type}>"  # noqa: E501
-
-
-class APIResponse(BaseAPIResponse[R]):
-    @property
-    def request_id(self) -> str | None:
-        return self.http_response.headers.get("x-request-id")  # type: ignore[no-any-return]
-
-    @overload
-    def parse(self, *, to: type[_T]) -> _T: ...
-
-    @overload
-    def parse(self) -> R: ...
-
-    def parse(self, *, to: type[_T] | None = None) -> R | _T:
-        """Returns the rich python representation of this response's data.
-
-        For lower-level control, see `.read()`, `.json()`, `.iter_bytes()`.
-
-        You can customize the type that the response is parsed into through
-        the `to` argument, e.g.
-
-        ```py
-        from openai import BaseModel
-
-
-        class MyModel(BaseModel):
-            foo: str
-
-
-        obj = response.parse(to=MyModel)
-        print(obj.foo)
-        ```
-
-        We support parsing:
-          - `BaseModel`
-          - `dict`
-          - `list`
-          - `Union`
-          - `str`
-          - `int`
-          - `float`
-          - `httpx.Response`
-        """
-        cache_key = to if to is not None else self._cast_type
-        cached = self._parsed_by_type.get(cache_key)
-        if cached is not None:
-            return cached  # type: ignore[no-any-return]
-
-        if not self._is_sse_stream:
-            self.read()
-
-        parsed = self._parse(to=to)
-        if is_given(self._options.post_parser):
-            parsed = self._options.post_parser(parsed)
-
-        self._parsed_by_type[cache_key] = parsed
-        return parsed
-
-    def read(self) -> bytes:
-        """Read and return the binary response content."""
-        try:
-            return self.http_response.read()
-        except httpx.StreamConsumed as exc:
-            # The default error raised by httpx isn't very
-            # helpful in our case so we re-raise it with
-            # a different error message.
-            raise StreamAlreadyConsumed() from exc
-
-    def text(self) -> str:
-        """Read and decode the response content into a string."""
-        self.read()
-        return self.http_response.text
-
-    def json(self) -> object:
-        """Read and decode the JSON response content."""
-        self.read()
-        return self.http_response.json()
-
-    def close(self) -> None:
-        """Close the response and release the connection.
-
-        Automatically called if the response body is read to completion.
-        """
-        self.http_response.close()
-
-    def iter_bytes(self, chunk_size: int | None = None) -> Iterator[bytes]:
-        """
-        A byte-iterator over the decoded response content.
-
-        This automatically handles gzip, deflate and brotli encoded responses.
-        """
-        yield from self.http_response.iter_bytes(chunk_size)
-
-    def iter_text(self, chunk_size: int | None = None) -> Iterator[str]:
-        """A str-iterator over the decoded response content
-        that handles both gzip, deflate, etc but also detects the content's
-        string encoding.
-        """
-        yield from self.http_response.iter_text(chunk_size)
-
-    def iter_lines(self) -> Iterator[str]:
-        """Like `iter_text()` but will only yield chunks for each line"""
-        yield from self.http_response.iter_lines()
-
-
-class MissingStreamClassError(TypeError):
-    def __init__(self) -> None:
-        super().__init__(
-            "The `stream` argument was set to `True` but the `stream_cls` argument was not given. See `openai._streaming` for reference",  # noqa: E501
-        )
-
-
-class StreamAlreadyConsumed(ZhipuAIError):  # noqa: N818
-    """
-    Attempted to read or stream content, but the content has already
-    been streamed.
-
-    This can happen if you use a method like `.iter_lines()` and then attempt
-    to read th entire response body afterwards, e.g.
-
-    ```py
-    response = await client.post(...)
-    async for line in response.iter_lines():
-        ...  # do something with `line`
-
-    content = await response.read()
-    # ^ error
-    ```
-
-    If you want this behavior you'll need to either manually accumulate the response
-    content or call `await response.read()` before iterating over the stream.
-    """
-
-    def __init__(self) -> None:
-        message = (
-            "Attempted to read or stream some content, but the content has "
-            "already been streamed. "
-            "This could be due to attempting to stream the response "
-            "content more than once."
-            "\n\n"
-            "You can fix this by manually accumulating the response content while streaming "
-            "or by calling `.read()` before starting to stream."
-        )
-        super().__init__(message)
-
-
-def extract_response_type(typ: type[BaseAPIResponse[Any]]) -> type:
-    """Given a type like `APIResponse[T]`, returns the generic type variable `T`.
-
-    This also handles the case where a concrete subclass is given, e.g.
-    ```py
-    class MyResponse(APIResponse[bytes]):
-        ...
-
-    extract_response_type(MyResponse) -> bytes
-    ```
-    """
-    return extract_type_var_from_base(
-        typ,
-        generic_bases=cast("tuple[type, ...]", (BaseAPIResponse, APIResponse)),
-        index=0,
-    )
diff --git a/api/core/model_runtime/model_providers/zhipuai/zhipuai_sdk/core/_sse_client.py b/api/core/model_runtime/model_providers/zhipuai/zhipuai_sdk/core/_sse_client.py
deleted file mode 100644
index cbc449d24421d..0000000000000
--- a/api/core/model_runtime/model_providers/zhipuai/zhipuai_sdk/core/_sse_client.py
+++ /dev/null
@@ -1,206 +0,0 @@
-from __future__ import annotations
-
-import inspect
-import json
-from collections.abc import Iterator, Mapping
-from typing import TYPE_CHECKING, Generic, TypeGuard, cast
-
-import httpx
-
-from . import get_origin
-from ._base_type import ResponseT
-from ._errors import APIResponseError
-from ._utils import extract_type_var_from_base, is_mapping
-
-_FIELD_SEPARATOR = ":"
-
-if TYPE_CHECKING:
-    from ._http_client import HttpClient
-
-
-class StreamResponse(Generic[ResponseT]):
-    response: httpx.Response
-    _cast_type: type[ResponseT]
-
-    def __init__(
-        self,
-        *,
-        cast_type: type[ResponseT],
-        response: httpx.Response,
-        client: HttpClient,
-    ) -> None:
-        self.response = response
-        self._cast_type = cast_type
-        self._data_process_func = client._process_response_data
-        self._stream_chunks = self.__stream__()
-
-    def __next__(self) -> ResponseT:
-        return self._stream_chunks.__next__()
-
-    def __iter__(self) -> Iterator[ResponseT]:
-        yield from self._stream_chunks
-
-    def __stream__(self) -> Iterator[ResponseT]:
-        sse_line_parser = SSELineParser()
-        iterator = sse_line_parser.iter_lines(self.response.iter_lines())
-
-        for sse in iterator:
-            if sse.data.startswith("[DONE]"):
-                break
-
-            if sse.event is None:
-                data = sse.json_data()
-                if isinstance(data, Mapping) and data.get("error"):
-                    raise APIResponseError(
-                        message="An error occurred during streaming",
-                        request=self.response.request,
-                        json_data=data["error"],
-                    )
-            if sse.event is None:
-                data = sse.json_data()
-                if is_mapping(data) and data.get("error"):
-                    message = None
-                    error = data.get("error")
-                    if is_mapping(error):
-                        message = error.get("message")
-                    if not message or not isinstance(message, str):
-                        message = "An error occurred during streaming"
-
-                    raise APIResponseError(
-                        message=message,
-                        request=self.response.request,
-                        json_data=data["error"],
-                    )
-                yield self._data_process_func(data=data, cast_type=self._cast_type, response=self.response)
-
-            else:
-                data = sse.json_data()
-
-                if sse.event == "error" and is_mapping(data) and data.get("error"):
-                    message = None
-                    error = data.get("error")
-                    if is_mapping(error):
-                        message = error.get("message")
-                    if not message or not isinstance(message, str):
-                        message = "An error occurred during streaming"
-
-                    raise APIResponseError(
-                        message=message,
-                        request=self.response.request,
-                        json_data=data["error"],
-                    )
-                yield self._data_process_func(data=data, cast_type=self._cast_type, response=self.response)
-
-        for sse in iterator:
-            pass
-
-
-class Event:
-    def __init__(
-        self, event: str | None = None, data: str | None = None, id: str | None = None, retry: int | None = None
-    ):
-        self._event = event
-        self._data = data
-        self._id = id
-        self._retry = retry
-
-    def __repr__(self):
-        data_len = len(self._data) if self._data else 0
-        return (
-            f"Event(event={self._event}, data={self._data} ,data_length={data_len}, id={self._id}, retry={self._retry}"
-        )
-
-    @property
-    def event(self):
-        return self._event
-
-    @property
-    def data(self):
-        return self._data
-
-    def json_data(self):
-        return json.loads(self._data)
-
-    @property
-    def id(self):
-        return self._id
-
-    @property
-    def retry(self):
-        return self._retry
-
-
-class SSELineParser:
-    _data: list[str]
-    _event: str | None
-    _retry: int | None
-    _id: str | None
-
-    def __init__(self):
-        self._event = None
-        self._data = []
-        self._id = None
-        self._retry = None
-
-    def iter_lines(self, lines: Iterator[str]) -> Iterator[Event]:
-        for line in lines:
-            line = line.rstrip("\n")
-            if not line:
-                if self._event is None and not self._data and self._id is None and self._retry is None:
-                    continue
-                sse_event = Event(event=self._event, data="\n".join(self._data), id=self._id, retry=self._retry)
-                self._event = None
-                self._data = []
-                self._id = None
-                self._retry = None
-
-                yield sse_event
-            self.decode_line(line)
-
-    def decode_line(self, line: str):
-        if line.startswith(":") or not line:
-            return
-
-        field, _p, value = line.partition(":")
-
-        value = value.removeprefix(" ")
-        if field == "data":
-            self._data.append(value)
-        elif field == "event":
-            self._event = value
-        elif field == "retry":
-            try:
-                self._retry = int(value)
-            except (TypeError, ValueError):
-                pass
-        return
-
-
-def is_stream_class_type(typ: type) -> TypeGuard[type[StreamResponse[object]]]:
-    """TypeGuard for determining whether or not the given type is a subclass of `Stream` / `AsyncStream`"""
-    origin = get_origin(typ) or typ
-    return inspect.isclass(origin) and issubclass(origin, StreamResponse)
-
-
-def extract_stream_chunk_type(
-    stream_cls: type,
-    *,
-    failure_message: str | None = None,
-) -> type:
-    """Given a type like `StreamResponse[T]`, returns the generic type variable `T`.
-
-    This also handles the case where a concrete subclass is given, e.g.
-    ```py
-    class MyStream(StreamResponse[bytes]):
-        ...
-
-    extract_stream_chunk_type(MyStream) -> bytes
-    ```
-    """
-
-    return extract_type_var_from_base(
-        stream_cls,
-        index=0,
-        generic_bases=cast("tuple[type, ...]", (StreamResponse,)),
-        failure_message=failure_message,
-    )
diff --git a/api/core/model_runtime/model_providers/zhipuai/zhipuai_sdk/core/_utils/__init__.py b/api/core/model_runtime/model_providers/zhipuai/zhipuai_sdk/core/_utils/__init__.py
deleted file mode 100644
index a66b095816b8b..0000000000000
--- a/api/core/model_runtime/model_providers/zhipuai/zhipuai_sdk/core/_utils/__init__.py
+++ /dev/null
@@ -1,52 +0,0 @@
-from ._utils import (  # noqa: I001
-    remove_notgiven_indict as remove_notgiven_indict,  # noqa: PLC0414
-    flatten as flatten,  # noqa: PLC0414
-    is_dict as is_dict,  # noqa: PLC0414
-    is_list as is_list,  # noqa: PLC0414
-    is_given as is_given,  # noqa: PLC0414
-    is_tuple as is_tuple,  # noqa: PLC0414
-    is_mapping as is_mapping,  # noqa: PLC0414
-    is_tuple_t as is_tuple_t,  # noqa: PLC0414
-    parse_date as parse_date,  # noqa: PLC0414
-    is_iterable as is_iterable,  # noqa: PLC0414
-    is_sequence as is_sequence,  # noqa: PLC0414
-    coerce_float as coerce_float,  # noqa: PLC0414
-    is_mapping_t as is_mapping_t,  # noqa: PLC0414
-    removeprefix as removeprefix,  # noqa: PLC0414
-    removesuffix as removesuffix,  # noqa: PLC0414
-    extract_files as extract_files,  # noqa: PLC0414
-    is_sequence_t as is_sequence_t,  # noqa: PLC0414
-    required_args as required_args,  # noqa: PLC0414
-    coerce_boolean as coerce_boolean,  # noqa: PLC0414
-    coerce_integer as coerce_integer,  # noqa: PLC0414
-    file_from_path as file_from_path,  # noqa: PLC0414
-    parse_datetime as parse_datetime,  # noqa: PLC0414
-    strip_not_given as strip_not_given,  # noqa: PLC0414
-    deepcopy_minimal as deepcopy_minimal,  # noqa: PLC0414
-    get_async_library as get_async_library,  # noqa: PLC0414
-    maybe_coerce_float as maybe_coerce_float,  # noqa: PLC0414
-    get_required_header as get_required_header,  # noqa: PLC0414
-    maybe_coerce_boolean as maybe_coerce_boolean,  # noqa: PLC0414
-    maybe_coerce_integer as maybe_coerce_integer,  # noqa: PLC0414
-    drop_prefix_image_data as drop_prefix_image_data,  # noqa: PLC0414
-)
-
-
-from ._typing import (
-    is_list_type as is_list_type,  # noqa: PLC0414
-    is_union_type as is_union_type,  # noqa: PLC0414
-    extract_type_arg as extract_type_arg,  # noqa: PLC0414
-    is_iterable_type as is_iterable_type,  # noqa: PLC0414
-    is_required_type as is_required_type,  # noqa: PLC0414
-    is_annotated_type as is_annotated_type,  # noqa: PLC0414
-    strip_annotated_type as strip_annotated_type,  # noqa: PLC0414
-    extract_type_var_from_base as extract_type_var_from_base,  # noqa: PLC0414
-)
-
-from ._transform import (
-    PropertyInfo as PropertyInfo,  # noqa: PLC0414
-    transform as transform,  # noqa: PLC0414
-    async_transform as async_transform,  # noqa: PLC0414
-    maybe_transform as maybe_transform,  # noqa: PLC0414
-    async_maybe_transform as async_maybe_transform,  # noqa: PLC0414
-)
diff --git a/api/core/model_runtime/model_providers/zhipuai/zhipuai_sdk/core/_utils/_transform.py b/api/core/model_runtime/model_providers/zhipuai/zhipuai_sdk/core/_utils/_transform.py
deleted file mode 100644
index e8ef1f79358a9..0000000000000
--- a/api/core/model_runtime/model_providers/zhipuai/zhipuai_sdk/core/_utils/_transform.py
+++ /dev/null
@@ -1,383 +0,0 @@
-from __future__ import annotations
-
-import base64
-import io
-import pathlib
-from collections.abc import Mapping
-from datetime import date, datetime
-from typing import Any, Literal, TypeVar, cast, get_args, get_type_hints
-
-import anyio
-import pydantic
-from typing_extensions import override
-
-from .._base_compat import is_typeddict, model_dump
-from .._files import is_base64_file_input
-from ._typing import (
-    extract_type_arg,
-    is_annotated_type,
-    is_iterable_type,
-    is_list_type,
-    is_required_type,
-    is_union_type,
-    strip_annotated_type,
-)
-from ._utils import (
-    is_iterable,
-    is_list,
-    is_mapping,
-)
-
-_T = TypeVar("_T")
-
-
-# TODO: support for drilling globals() and locals()
-# TODO: ensure works correctly with forward references in all cases
-
-
-PropertyFormat = Literal["iso8601", "base64", "custom"]
-
-
-class PropertyInfo:
-    """Metadata class to be used in Annotated types to provide information about a given type.
-
-    For example:
-
-    class MyParams(TypedDict):
-        account_holder_name: Annotated[str, PropertyInfo(alias='accountHolderName')]
-
-    This means that {'account_holder_name': 'Robert'} will be transformed to {'accountHolderName': 'Robert'} before being sent to the API.
-    """  # noqa: E501
-
-    alias: str | None
-    format: PropertyFormat | None
-    format_template: str | None
-    discriminator: str | None
-
-    def __init__(
-        self,
-        *,
-        alias: str | None = None,
-        format: PropertyFormat | None = None,
-        format_template: str | None = None,
-        discriminator: str | None = None,
-    ) -> None:
-        self.alias = alias
-        self.format = format
-        self.format_template = format_template
-        self.discriminator = discriminator
-
-    @override
-    def __repr__(self) -> str:
-        return f"{self.__class__.__name__}(alias='{self.alias}', format={self.format}, format_template='{self.format_template}', discriminator='{self.discriminator}')"  # noqa: E501
-
-
-def maybe_transform(
-    data: object,
-    expected_type: object,
-) -> Any | None:
-    """Wrapper over `transform()` that allows `None` to be passed.
-
-    See `transform()` for more details.
-    """
-    if data is None:
-        return None
-    return transform(data, expected_type)
-
-
-# Wrapper over _transform_recursive providing fake types
-def transform(
-    data: _T,
-    expected_type: object,
-) -> _T:
-    """Transform dictionaries based off of type information from the given type, for example:
-
-    ```py
-    class Params(TypedDict, total=False):
-        card_id: Required[Annotated[str, PropertyInfo(alias="cardID")]]
-
-
-    transformed = transform({"card_id": "<my card ID>"}, Params)
-    # {'cardID': '<my card ID>'}
-    ```
-
-    Any keys / data that does not have type information given will be included as is.
-
-    It should be noted that the transformations that this function does are not represented in the type system.
-    """
-    transformed = _transform_recursive(data, annotation=cast(type, expected_type))
-    return cast(_T, transformed)
-
-
-def _get_annotated_type(type_: type) -> type | None:
-    """If the given type is an `Annotated` type then it is returned, if not `None` is returned.
-
-    This also unwraps the type when applicable, e.g. `Required[Annotated[T, ...]]`
-    """
-    if is_required_type(type_):
-        # Unwrap `Required[Annotated[T, ...]]` to `Annotated[T, ...]`
-        type_ = get_args(type_)[0]
-
-    if is_annotated_type(type_):
-        return type_
-
-    return None
-
-
-def _maybe_transform_key(key: str, type_: type) -> str:
-    """Transform the given `data` based on the annotations provided in `type_`.
-
-    Note: this function only looks at `Annotated` types that contain `PropertInfo` metadata.
-    """
-    annotated_type = _get_annotated_type(type_)
-    if annotated_type is None:
-        # no `Annotated` definition for this type, no transformation needed
-        return key
-
-    # ignore the first argument as it is the actual type
-    annotations = get_args(annotated_type)[1:]
-    for annotation in annotations:
-        if isinstance(annotation, PropertyInfo) and annotation.alias is not None:
-            return annotation.alias
-
-    return key
-
-
-def _transform_recursive(
-    data: object,
-    *,
-    annotation: type,
-    inner_type: type | None = None,
-) -> object:
-    """Transform the given data against the expected type.
-
-    Args:
-        annotation: The direct type annotation given to the particular piece of data.
-            This may or may not be wrapped in metadata types, e.g. `Required[T]`, `Annotated[T, ...]` etc
-
-        inner_type: If applicable, this is the "inside" type. This is useful in certain cases where the outside type
-            is a container type such as `List[T]`. In that case `inner_type` should be set to `T` so that each entry in
-            the list can be transformed using the metadata from the container type.
-
-            Defaults to the same value as the `annotation` argument.
-    """
-    if inner_type is None:
-        inner_type = annotation
-
-    stripped_type = strip_annotated_type(inner_type)
-    if is_typeddict(stripped_type) and is_mapping(data):
-        return _transform_typeddict(data, stripped_type)
-
-    if (
-        # List[T]
-        (is_list_type(stripped_type) and is_list(data))
-        # Iterable[T]
-        or (is_iterable_type(stripped_type) and is_iterable(data) and not isinstance(data, str))
-    ):
-        inner_type = extract_type_arg(stripped_type, 0)
-        return [_transform_recursive(d, annotation=annotation, inner_type=inner_type) for d in data]
-
-    if is_union_type(stripped_type):
-        # For union types we run the transformation against all subtypes to ensure that everything is transformed.
-        #
-        # TODO: there may be edge cases where the same normalized field name will transform to two different names
-        # in different subtypes.
-        for subtype in get_args(stripped_type):
-            data = _transform_recursive(data, annotation=annotation, inner_type=subtype)
-        return data
-
-    if isinstance(data, pydantic.BaseModel):
-        return model_dump(data, exclude_unset=True)
-
-    annotated_type = _get_annotated_type(annotation)
-    if annotated_type is None:
-        return data
-
-    # ignore the first argument as it is the actual type
-    annotations = get_args(annotated_type)[1:]
-    for annotation in annotations:
-        if isinstance(annotation, PropertyInfo) and annotation.format is not None:
-            return _format_data(data, annotation.format, annotation.format_template)
-
-    return data
-
-
-def _format_data(data: object, format_: PropertyFormat, format_template: str | None) -> object:
-    if isinstance(data, date | datetime):
-        if format_ == "iso8601":
-            return data.isoformat()
-
-        if format_ == "custom" and format_template is not None:
-            return data.strftime(format_template)
-
-    if format_ == "base64" and is_base64_file_input(data):
-        binary: str | bytes | None = None
-
-        if isinstance(data, pathlib.Path):
-            binary = data.read_bytes()
-        elif isinstance(data, io.IOBase):
-            binary = data.read()
-
-            if isinstance(binary, str):  # type: ignore[unreachable]
-                binary = binary.encode()
-
-        if not isinstance(binary, bytes):
-            raise RuntimeError(f"Could not read bytes from {data}; Received {type(binary)}")
-
-        return base64.b64encode(binary).decode("ascii")
-
-    return data
-
-
-def _transform_typeddict(
-    data: Mapping[str, object],
-    expected_type: type,
-) -> Mapping[str, object]:
-    result: dict[str, object] = {}
-    annotations = get_type_hints(expected_type, include_extras=True)
-    for key, value in data.items():
-        type_ = annotations.get(key)
-        if type_ is None:
-            # we do not have a type annotation for this field, leave it as is
-            result[key] = value
-        else:
-            result[_maybe_transform_key(key, type_)] = _transform_recursive(value, annotation=type_)
-    return result
-
-
-async def async_maybe_transform(
-    data: object,
-    expected_type: object,
-) -> Any | None:
-    """Wrapper over `async_transform()` that allows `None` to be passed.
-
-    See `async_transform()` for more details.
-    """
-    if data is None:
-        return None
-    return await async_transform(data, expected_type)
-
-
-async def async_transform(
-    data: _T,
-    expected_type: object,
-) -> _T:
-    """Transform dictionaries based off of type information from the given type, for example:
-
-    ```py
-    class Params(TypedDict, total=False):
-        card_id: Required[Annotated[str, PropertyInfo(alias="cardID")]]
-
-
-    transformed = transform({"card_id": "<my card ID>"}, Params)
-    # {'cardID': '<my card ID>'}
-    ```
-
-    Any keys / data that does not have type information given will be included as is.
-
-    It should be noted that the transformations that this function does are not represented in the type system.
-    """
-    transformed = await _async_transform_recursive(data, annotation=cast(type, expected_type))
-    return cast(_T, transformed)
-
-
-async def _async_transform_recursive(
-    data: object,
-    *,
-    annotation: type,
-    inner_type: type | None = None,
-) -> object:
-    """Transform the given data against the expected type.
-
-    Args:
-        annotation: The direct type annotation given to the particular piece of data.
-            This may or may not be wrapped in metadata types, e.g. `Required[T]`, `Annotated[T, ...]` etc
-
-        inner_type: If applicable, this is the "inside" type. This is useful in certain cases where the outside type
-            is a container type such as `List[T]`. In that case `inner_type` should be set to `T` so that each entry in
-            the list can be transformed using the metadata from the container type.
-
-            Defaults to the same value as the `annotation` argument.
-    """
-    if inner_type is None:
-        inner_type = annotation
-
-    stripped_type = strip_annotated_type(inner_type)
-    if is_typeddict(stripped_type) and is_mapping(data):
-        return await _async_transform_typeddict(data, stripped_type)
-
-    if (
-        # List[T]
-        (is_list_type(stripped_type) and is_list(data))
-        # Iterable[T]
-        or (is_iterable_type(stripped_type) and is_iterable(data) and not isinstance(data, str))
-    ):
-        inner_type = extract_type_arg(stripped_type, 0)
-        return [await _async_transform_recursive(d, annotation=annotation, inner_type=inner_type) for d in data]
-
-    if is_union_type(stripped_type):
-        # For union types we run the transformation against all subtypes to ensure that everything is transformed.
-        #
-        # TODO: there may be edge cases where the same normalized field name will transform to two different names
-        # in different subtypes.
-        for subtype in get_args(stripped_type):
-            data = await _async_transform_recursive(data, annotation=annotation, inner_type=subtype)
-        return data
-
-    if isinstance(data, pydantic.BaseModel):
-        return model_dump(data, exclude_unset=True)
-
-    annotated_type = _get_annotated_type(annotation)
-    if annotated_type is None:
-        return data
-
-    # ignore the first argument as it is the actual type
-    annotations = get_args(annotated_type)[1:]
-    for annotation in annotations:
-        if isinstance(annotation, PropertyInfo) and annotation.format is not None:
-            return await _async_format_data(data, annotation.format, annotation.format_template)
-
-    return data
-
-
-async def _async_format_data(data: object, format_: PropertyFormat, format_template: str | None) -> object:
-    if isinstance(data, date | datetime):
-        if format_ == "iso8601":
-            return data.isoformat()
-
-        if format_ == "custom" and format_template is not None:
-            return data.strftime(format_template)
-
-    if format_ == "base64" and is_base64_file_input(data):
-        binary: str | bytes | None = None
-
-        if isinstance(data, pathlib.Path):
-            binary = await anyio.Path(data).read_bytes()
-        elif isinstance(data, io.IOBase):
-            binary = data.read()
-
-            if isinstance(binary, str):  # type: ignore[unreachable]
-                binary = binary.encode()
-
-        if not isinstance(binary, bytes):
-            raise RuntimeError(f"Could not read bytes from {data}; Received {type(binary)}")
-
-        return base64.b64encode(binary).decode("ascii")
-
-    return data
-
-
-async def _async_transform_typeddict(
-    data: Mapping[str, object],
-    expected_type: type,
-) -> Mapping[str, object]:
-    result: dict[str, object] = {}
-    annotations = get_type_hints(expected_type, include_extras=True)
-    for key, value in data.items():
-        type_ = annotations.get(key)
-        if type_ is None:
-            # we do not have a type annotation for this field, leave it as is
-            result[key] = value
-        else:
-            result[_maybe_transform_key(key, type_)] = await _async_transform_recursive(value, annotation=type_)
-    return result
diff --git a/api/core/model_runtime/model_providers/zhipuai/zhipuai_sdk/core/_utils/_typing.py b/api/core/model_runtime/model_providers/zhipuai/zhipuai_sdk/core/_utils/_typing.py
deleted file mode 100644
index c7c54dcc37458..0000000000000
--- a/api/core/model_runtime/model_providers/zhipuai/zhipuai_sdk/core/_utils/_typing.py
+++ /dev/null
@@ -1,122 +0,0 @@
-from __future__ import annotations
-
-from collections import abc as _c_abc
-from collections.abc import Iterable
-from typing import Annotated, Any, TypeVar, cast, get_args, get_origin
-
-from typing_extensions import Required
-
-from .._base_compat import is_union as _is_union
-from .._base_type import InheritsGeneric
-
-
-def is_annotated_type(typ: type) -> bool:
-    return get_origin(typ) == Annotated
-
-
-def is_list_type(typ: type) -> bool:
-    return (get_origin(typ) or typ) == list
-
-
-def is_iterable_type(typ: type) -> bool:
-    """If the given type is `typing.Iterable[T]`"""
-    origin = get_origin(typ) or typ
-    return origin in {Iterable, _c_abc.Iterable}
-
-
-def is_union_type(typ: type) -> bool:
-    return _is_union(get_origin(typ))
-
-
-def is_required_type(typ: type) -> bool:
-    return get_origin(typ) == Required
-
-
-def is_typevar(typ: type) -> bool:
-    # type ignore is required because type checkers
-    # think this expression will always return False
-    return type(typ) == TypeVar  # type: ignore
-
-
-# Extracts T from Annotated[T, ...] or from Required[Annotated[T, ...]]
-def strip_annotated_type(typ: type) -> type:
-    if is_required_type(typ) or is_annotated_type(typ):
-        return strip_annotated_type(cast(type, get_args(typ)[0]))
-
-    return typ
-
-
-def extract_type_arg(typ: type, index: int) -> type:
-    args = get_args(typ)
-    try:
-        return cast(type, args[index])
-    except IndexError as err:
-        raise RuntimeError(f"Expected type {typ} to have a type argument at index {index} but it did not") from err
-
-
-def extract_type_var_from_base(
-    typ: type,
-    *,
-    generic_bases: tuple[type, ...],
-    index: int,
-    failure_message: str | None = None,
-) -> type:
-    """Given a type like `Foo[T]`, returns the generic type variable `T`.
-
-    This also handles the case where a concrete subclass is given, e.g.
-    ```py
-    class MyResponse(Foo[bytes]):
-        ...
-
-    extract_type_var(MyResponse, bases=(Foo,), index=0) -> bytes
-    ```
-
-    And where a generic subclass is given:
-    ```py
-    _T = TypeVar('_T')
-    class MyResponse(Foo[_T]):
-        ...
-
-    extract_type_var(MyResponse[bytes], bases=(Foo,), index=0) -> bytes
-    ```
-    """
-    cls = cast(object, get_origin(typ) or typ)
-    if cls in generic_bases:
-        # we're given the class directly
-        return extract_type_arg(typ, index)
-
-    # if a subclass is given
-    # ---
-    # this is needed as __orig_bases__ is not present in the typeshed stubs
-    # because it is intended to be for internal use only, however there does
-    # not seem to be a way to resolve generic TypeVars for inherited subclasses
-    # without using it.
-    if isinstance(cls, InheritsGeneric):
-        target_base_class: Any | None = None
-        for base in cls.__orig_bases__:
-            if base.__origin__ in generic_bases:
-                target_base_class = base
-                break
-
-        if target_base_class is None:
-            raise RuntimeError(
-                "Could not find the generic base class;\n"
-                "This should never happen;\n"
-                f"Does {cls} inherit from one of {generic_bases} ?"
-            )
-
-        extracted = extract_type_arg(target_base_class, index)
-        if is_typevar(extracted):
-            # If the extracted type argument is itself a type variable
-            # then that means the subclass itself is generic, so we have
-            # to resolve the type argument from the class itself, not
-            # the base class.
-            #
-            # Note: if there is more than 1 type argument, the subclass could
-            # change the ordering of the type arguments, this is not currently
-            # supported.
-            return extract_type_arg(typ, index)
-
-        return extracted
-
-    raise RuntimeError(failure_message or f"Could not resolve inner type variable at index {index} for {typ}")
diff --git a/api/core/model_runtime/model_providers/zhipuai/zhipuai_sdk/core/_utils/_utils.py b/api/core/model_runtime/model_providers/zhipuai/zhipuai_sdk/core/_utils/_utils.py
deleted file mode 100644
index 3a7b234ab0c06..0000000000000
--- a/api/core/model_runtime/model_providers/zhipuai/zhipuai_sdk/core/_utils/_utils.py
+++ /dev/null
@@ -1,409 +0,0 @@
-from __future__ import annotations
-
-import functools
-import inspect
-import os
-import re
-from collections.abc import Callable, Iterable, Mapping, Sequence
-from pathlib import Path
-from typing import (
-    Any,
-    TypeGuard,
-    TypeVar,
-    Union,
-    cast,
-    overload,
-)
-
-import sniffio
-
-from .._base_compat import parse_date as parse_date  # noqa: PLC0414
-from .._base_compat import parse_datetime as parse_datetime  # noqa: PLC0414
-from .._base_type import FileTypes, Headers, HeadersLike, NotGiven, NotGivenOr
-
-
-def remove_notgiven_indict(obj):
-    if obj is None or (not isinstance(obj, Mapping)):
-        return obj
-    return {key: value for key, value in obj.items() if not isinstance(value, NotGiven)}
-
-
-_T = TypeVar("_T")
-_TupleT = TypeVar("_TupleT", bound=tuple[object, ...])
-_MappingT = TypeVar("_MappingT", bound=Mapping[str, object])
-_SequenceT = TypeVar("_SequenceT", bound=Sequence[object])
-CallableT = TypeVar("CallableT", bound=Callable[..., Any])
-
-
-def flatten(t: Iterable[Iterable[_T]]) -> list[_T]:
-    return [item for sublist in t for item in sublist]
-
-
-def extract_files(
-    # TODO: this needs to take Dict but variance issues.....
-    # create protocol type ?
-    query: Mapping[str, object],
-    *,
-    paths: Sequence[Sequence[str]],
-) -> list[tuple[str, FileTypes]]:
-    """Recursively extract files from the given dictionary based on specified paths.
-
-    A path may look like this ['foo', 'files', '<array>', 'data'].
-
-    Note: this mutates the given dictionary.
-    """
-    files: list[tuple[str, FileTypes]] = []
-    for path in paths:
-        files.extend(_extract_items(query, path, index=0, flattened_key=None))
-    return files
-
-
-def _extract_items(
-    obj: object,
-    path: Sequence[str],
-    *,
-    index: int,
-    flattened_key: str | None,
-) -> list[tuple[str, FileTypes]]:
-    try:
-        key = path[index]
-    except IndexError:
-        if isinstance(obj, NotGiven):
-            # no value was provided - we can safely ignore
-            return []
-
-        # cyclical import
-        from .._files import assert_is_file_content
-
-        # We have exhausted the path, return the entry we found.
-        assert_is_file_content(obj, key=flattened_key)
-        assert flattened_key is not None
-        return [(flattened_key, cast(FileTypes, obj))]
-
-    index += 1
-    if is_dict(obj):
-        try:
-            # We are at the last entry in the path so we must remove the field
-            if (len(path)) == index:
-                item = obj.pop(key)
-            else:
-                item = obj[key]
-        except KeyError:
-            # Key was not present in the dictionary, this is not indicative of an error
-            # as the given path may not point to a required field. We also do not want
-            # to enforce required fields as the API may differ from the spec in some cases.
-            return []
-        if flattened_key is None:
-            flattened_key = key
-        else:
-            flattened_key += f"[{key}]"
-        return _extract_items(
-            item,
-            path,
-            index=index,
-            flattened_key=flattened_key,
-        )
-    elif is_list(obj):
-        if key != "<array>":
-            return []
-
-        return flatten(
-            [
-                _extract_items(
-                    item,
-                    path,
-                    index=index,
-                    flattened_key=flattened_key + "[]" if flattened_key is not None else "[]",
-                )
-                for item in obj
-            ]
-        )
-
-    # Something unexpected was passed, just ignore it.
-    return []
-
-
-def is_given(obj: NotGivenOr[_T]) -> TypeGuard[_T]:
-    return not isinstance(obj, NotGiven)
-
-
-# Type safe methods for narrowing types with TypeVars.
-# The default narrowing for isinstance(obj, dict) is dict[unknown, unknown],
-# however this cause Pyright to rightfully report errors. As we know we don't
-# care about the contained types we can safely use `object` in it's place.
-#
-# There are two separate functions defined, `is_*` and `is_*_t` for different use cases.
-# `is_*` is for when you're dealing with an unknown input
-# `is_*_t` is for when you're narrowing a known union type to a specific subset
-
-
-def is_tuple(obj: object) -> TypeGuard[tuple[object, ...]]:
-    return isinstance(obj, tuple)
-
-
-def is_tuple_t(obj: _TupleT | object) -> TypeGuard[_TupleT]:
-    return isinstance(obj, tuple)
-
-
-def is_sequence(obj: object) -> TypeGuard[Sequence[object]]:
-    return isinstance(obj, Sequence)
-
-
-def is_sequence_t(obj: _SequenceT | object) -> TypeGuard[_SequenceT]:
-    return isinstance(obj, Sequence)
-
-
-def is_mapping(obj: object) -> TypeGuard[Mapping[str, object]]:
-    return isinstance(obj, Mapping)
-
-
-def is_mapping_t(obj: _MappingT | object) -> TypeGuard[_MappingT]:
-    return isinstance(obj, Mapping)
-
-
-def is_dict(obj: object) -> TypeGuard[dict[object, object]]:
-    return isinstance(obj, dict)
-
-
-def is_list(obj: object) -> TypeGuard[list[object]]:
-    return isinstance(obj, list)
-
-
-def is_iterable(obj: object) -> TypeGuard[Iterable[object]]:
-    return isinstance(obj, Iterable)
-
-
-def deepcopy_minimal(item: _T) -> _T:
-    """Minimal reimplementation of copy.deepcopy() that will only copy certain object types:
-
-    - mappings, e.g. `dict`
-    - list
-
-    This is done for performance reasons.
-    """
-    if is_mapping(item):
-        return cast(_T, {k: deepcopy_minimal(v) for k, v in item.items()})
-    if is_list(item):
-        return cast(_T, [deepcopy_minimal(entry) for entry in item])
-    return item
-
-
-# copied from https://github.com/Rapptz/RoboDanny
-def human_join(seq: Sequence[str], *, delim: str = ", ", final: str = "or") -> str:
-    size = len(seq)
-    if size == 0:
-        return ""
-
-    if size == 1:
-        return seq[0]
-
-    if size == 2:
-        return f"{seq[0]} {final} {seq[1]}"
-
-    return delim.join(seq[:-1]) + f" {final} {seq[-1]}"
-
-
-def quote(string: str) -> str:
-    """Add single quotation marks around the given string. Does *not* do any escaping."""
-    return f"'{string}'"
-
-
-def required_args(*variants: Sequence[str]) -> Callable[[CallableT], CallableT]:
-    """Decorator to enforce a given set of arguments or variants of arguments are passed to the decorated function.
-
-    Useful for enforcing runtime validation of overloaded functions.
-
-    Example usage:
-    ```py
-    @overload
-    def foo(*, a: str) -> str:
-        ...
-
-
-    @overload
-    def foo(*, b: bool) -> str:
-        ...
-
-
-    # This enforces the same constraints that a static type checker would
-    # i.e. that either a or b must be passed to the function
-    @required_args(["a"], ["b"])
-    def foo(*, a: str | None = None, b: bool | None = None) -> str:
-        ...
-    ```
-    """
-
-    def inner(func: CallableT) -> CallableT:
-        params = inspect.signature(func).parameters
-        positional = [
-            name
-            for name, param in params.items()
-            if param.kind
-            in {
-                param.POSITIONAL_ONLY,
-                param.POSITIONAL_OR_KEYWORD,
-            }
-        ]
-
-        @functools.wraps(func)
-        def wrapper(*args: object, **kwargs: object) -> object:
-            given_params: set[str] = set()
-            for i in range(len(args)):
-                try:
-                    given_params.add(positional[i])
-                except IndexError:
-                    raise TypeError(
-                        f"{func.__name__}() takes {len(positional)} argument(s) but {len(args)} were given"
-                    ) from None
-
-            given_params.update(kwargs.keys())
-
-            for variant in variants:
-                matches = all(param in given_params for param in variant)
-                if matches:
-                    break
-            else:  # no break
-                if len(variants) > 1:
-                    variations = human_join(
-                        ["(" + human_join([quote(arg) for arg in variant], final="and") + ")" for variant in variants]
-                    )
-                    msg = f"Missing required arguments; Expected either {variations} arguments to be given"
-                else:
-                    # TODO: this error message is not deterministic
-                    missing = list(set(variants[0]) - given_params)
-                    if len(missing) > 1:
-                        msg = f"Missing required arguments: {human_join([quote(arg) for arg in missing])}"
-                    else:
-                        msg = f"Missing required argument: {quote(missing[0])}"
-                raise TypeError(msg)
-            return func(*args, **kwargs)
-
-        return wrapper  # type: ignore
-
-    return inner
-
-
-_K = TypeVar("_K")
-_V = TypeVar("_V")
-
-
-@overload
-def strip_not_given(obj: None) -> None: ...
-
-
-@overload
-def strip_not_given(obj: Mapping[_K, _V | NotGiven]) -> dict[_K, _V]: ...
-
-
-@overload
-def strip_not_given(obj: object) -> object: ...
-
-
-def strip_not_given(obj: object | None) -> object:
-    """Remove all top-level keys where their values are instances of `NotGiven`"""
-    if obj is None:
-        return None
-
-    if not is_mapping(obj):
-        return obj
-
-    return {key: value for key, value in obj.items() if not isinstance(value, NotGiven)}
-
-
-def coerce_integer(val: str) -> int:
-    return int(val, base=10)
-
-
-def coerce_float(val: str) -> float:
-    return float(val)
-
-
-def coerce_boolean(val: str) -> bool:
-    return val in {"true", "1", "on"}
-
-
-def maybe_coerce_integer(val: str | None) -> int | None:
-    if val is None:
-        return None
-    return coerce_integer(val)
-
-
-def maybe_coerce_float(val: str | None) -> float | None:
-    if val is None:
-        return None
-    return coerce_float(val)
-
-
-def maybe_coerce_boolean(val: str | None) -> bool | None:
-    if val is None:
-        return None
-    return coerce_boolean(val)
-
-
-def removeprefix(string: str, prefix: str) -> str:
-    """Remove a prefix from a string.
-
-    Backport of `str.removeprefix` for Python < 3.9
-    """
-    if string.startswith(prefix):
-        return string[len(prefix) :]
-    return string
-
-
-def removesuffix(string: str, suffix: str) -> str:
-    """Remove a suffix from a string.
-
-    Backport of `str.removesuffix` for Python < 3.9
-    """
-    if string.endswith(suffix):
-        return string[: -len(suffix)]
-    return string
-
-
-def file_from_path(path: str) -> FileTypes:
-    contents = Path(path).read_bytes()
-    file_name = os.path.basename(path)
-    return (file_name, contents)
-
-
-def get_required_header(headers: HeadersLike, header: str) -> str:
-    lower_header = header.lower()
-    if isinstance(headers, Mapping):
-        headers = cast(Headers, headers)
-        for k, v in headers.items():
-            if k.lower() == lower_header and isinstance(v, str):
-                return v
-
-    """ to deal with the case where the header looks like Stainless-Event-Id """
-    intercaps_header = re.sub(r"([^\w])(\w)", lambda pat: pat.group(1) + pat.group(2).upper(), header.capitalize())
-
-    for normalized_header in [header, lower_header, header.upper(), intercaps_header]:
-        value = headers.get(normalized_header)
-        if value:
-            return value
-
-    raise ValueError(f"Could not find {header} header")
-
-
-def get_async_library() -> str:
-    try:
-        return sniffio.current_async_library()
-    except Exception:
-        return "false"
-
-
-def drop_prefix_image_data(content: Union[str, list[dict]]) -> Union[str, list[dict]]:
-    """
-    删除 ;base64, 前缀
-    :param image_data:
-    :return:
-    """
-    if isinstance(content, list):
-        for data in content:
-            if data.get("type") == "image_url":
-                image_data = data.get("image_url").get("url")
-                if image_data.startswith("data:image/"):
-                    image_data = image_data.split("base64,")[-1]
-                    data["image_url"]["url"] = image_data
-
-    return content
diff --git a/api/core/model_runtime/model_providers/zhipuai/zhipuai_sdk/core/logs.py b/api/core/model_runtime/model_providers/zhipuai/zhipuai_sdk/core/logs.py
deleted file mode 100644
index e5fce94c00e9e..0000000000000
--- a/api/core/model_runtime/model_providers/zhipuai/zhipuai_sdk/core/logs.py
+++ /dev/null
@@ -1,78 +0,0 @@
-import logging
-import os
-import time
-
-logger = logging.getLogger(__name__)
-
-
-class LoggerNameFilter(logging.Filter):
-    def filter(self, record):
-        # return record.name.startswith("loom_core") or record.name in "ERROR" or (
-        #         record.name.startswith("uvicorn.error")
-        #         and record.getMessage().startswith("Uvicorn running on")
-        # )
-        return True
-
-
-def get_log_file(log_path: str, sub_dir: str):
-    """
-    sub_dir should contain a timestamp.
-    """
-    log_dir = os.path.join(log_path, sub_dir)
-    # Here should be creating a new directory each time, so `exist_ok=False`
-    os.makedirs(log_dir, exist_ok=False)
-    return os.path.join(log_dir, "zhipuai.log")
-
-
-def get_config_dict(log_level: str, log_file_path: str, log_backup_count: int, log_max_bytes: int) -> dict:
-    # for windows, the path should be a raw string.
-    log_file_path = log_file_path.encode("unicode-escape").decode() if os.name == "nt" else log_file_path
-    log_level = log_level.upper()
-    config_dict = {
-        "version": 1,
-        "disable_existing_loggers": False,
-        "formatters": {
-            "formatter": {"format": ("%(asctime)s %(name)-12s %(process)d %(levelname)-8s %(message)s")},
-        },
-        "filters": {
-            "logger_name_filter": {
-                "()": __name__ + ".LoggerNameFilter",
-            },
-        },
-        "handlers": {
-            "stream_handler": {
-                "class": "logging.StreamHandler",
-                "formatter": "formatter",
-                "level": log_level,
-                # "stream": "ext://sys.stdout",
-                # "filters": ["logger_name_filter"],
-            },
-            "file_handler": {
-                "class": "logging.handlers.RotatingFileHandler",
-                "formatter": "formatter",
-                "level": log_level,
-                "filename": log_file_path,
-                "mode": "a",
-                "maxBytes": log_max_bytes,
-                "backupCount": log_backup_count,
-                "encoding": "utf8",
-            },
-        },
-        "loggers": {
-            "loom_core": {
-                "handlers": ["stream_handler", "file_handler"],
-                "level": log_level,
-                "propagate": False,
-            }
-        },
-        "root": {
-            "level": log_level,
-            "handlers": ["stream_handler", "file_handler"],
-        },
-    }
-    return config_dict
-
-
-def get_timestamp_ms():
-    t = time.time()
-    return int(round(t * 1000))
diff --git a/api/core/model_runtime/model_providers/zhipuai/zhipuai_sdk/core/pagination.py b/api/core/model_runtime/model_providers/zhipuai/zhipuai_sdk/core/pagination.py
deleted file mode 100644
index 7f0b1b91d9855..0000000000000
--- a/api/core/model_runtime/model_providers/zhipuai/zhipuai_sdk/core/pagination.py
+++ /dev/null
@@ -1,62 +0,0 @@
-# File generated from our OpenAPI spec by Stainless. See CONTRIBUTING.md for details.
-
-from typing import Any, Generic, Optional, TypeVar, cast
-
-from typing_extensions import Protocol, override, runtime_checkable
-
-from ._http_client import BasePage, BaseSyncPage, PageInfo
-
-__all__ = ["SyncPage", "SyncCursorPage"]
-
-_T = TypeVar("_T")
-
-
-@runtime_checkable
-class CursorPageItem(Protocol):
-    id: Optional[str]
-
-
-class SyncPage(BaseSyncPage[_T], BasePage[_T], Generic[_T]):
-    """Note: no pagination actually occurs yet, this is for forwards-compatibility."""
-
-    data: list[_T]
-    object: str
-
-    @override
-    def _get_page_items(self) -> list[_T]:
-        data = self.data
-        if not data:
-            return []
-        return data
-
-    @override
-    def next_page_info(self) -> None:
-        """
-        This page represents a response that isn't actually paginated at the API level
-        so there will never be a next page.
-        """
-        return None
-
-
-class SyncCursorPage(BaseSyncPage[_T], BasePage[_T], Generic[_T]):
-    data: list[_T]
-
-    @override
-    def _get_page_items(self) -> list[_T]:
-        data = self.data
-        if not data:
-            return []
-        return data
-
-    @override
-    def next_page_info(self) -> Optional[PageInfo]:
-        data = self.data
-        if not data:
-            return None
-
-        item = cast(Any, data[-1])
-        if not isinstance(item, CursorPageItem) or item.id is None:
-            # TODO emit warning log
-            return None
-
-        return PageInfo(params={"after": item.id})
diff --git a/api/core/model_runtime/model_providers/zhipuai/zhipuai_sdk/types/__init__.py b/api/core/model_runtime/model_providers/zhipuai/zhipuai_sdk/types/__init__.py
deleted file mode 100644
index e69de29bb2d1d..0000000000000
diff --git a/api/core/model_runtime/model_providers/zhipuai/zhipuai_sdk/types/assistant/__init__.py b/api/core/model_runtime/model_providers/zhipuai/zhipuai_sdk/types/assistant/__init__.py
deleted file mode 100644
index 9f941fb91c877..0000000000000
--- a/api/core/model_runtime/model_providers/zhipuai/zhipuai_sdk/types/assistant/__init__.py
+++ /dev/null
@@ -1,5 +0,0 @@
-from .assistant_completion import AssistantCompletion
-
-__all__ = [
-    "AssistantCompletion",
-]
diff --git a/api/core/model_runtime/model_providers/zhipuai/zhipuai_sdk/types/assistant/assistant_completion.py b/api/core/model_runtime/model_providers/zhipuai/zhipuai_sdk/types/assistant/assistant_completion.py
deleted file mode 100644
index cbfb6edaeb1f1..0000000000000
--- a/api/core/model_runtime/model_providers/zhipuai/zhipuai_sdk/types/assistant/assistant_completion.py
+++ /dev/null
@@ -1,40 +0,0 @@
-from typing import Any, Optional
-
-from ...core import BaseModel
-from .message import MessageContent
-
-__all__ = ["AssistantCompletion", "CompletionUsage"]
-
-
-class ErrorInfo(BaseModel):
-    code: str  # 错误码
-    message: str  # 错误信息
-
-
-class AssistantChoice(BaseModel):
-    index: int  # 结果下标
-    delta: MessageContent  # 当前会话输出消息体
-    finish_reason: str
-    """
-    # 推理结束原因 stop代表推理自然结束或触发停止词。  sensitive 代表模型推理内容被安全审核接口拦截。请注意，针对此类内容，请用户自行判断并决定是否撤回已公开的内容。 
-    # network_error 代表模型推理服务异常。
-    """  # noqa: E501
-    metadata: dict  # 元信息，拓展字段
-
-
-class CompletionUsage(BaseModel):
-    prompt_tokens: int  # 输入的 tokens 数量
-    completion_tokens: int  # 输出的 tokens 数量
-    total_tokens: int  # 总 tokens 数量
-
-
-class AssistantCompletion(BaseModel):
-    id: str  # 请求 ID
-    conversation_id: str  # 会话 ID
-    assistant_id: str  # 智能体 ID
-    created: int  # 请求创建时间，Unix 时间戳
-    status: str  # 返回状态，包括：`completed` 表示生成结束`in_progress`表示生成中 `failed` 表示生成异常
-    last_error: Optional[ErrorInfo]  # 异常信息
-    choices: list[AssistantChoice]  # 增量返回的信息
-    metadata: Optional[dict[str, Any]]  # 元信息，拓展字段
-    usage: Optional[CompletionUsage]  # tokens 数量统计
diff --git a/api/core/model_runtime/model_providers/zhipuai/zhipuai_sdk/types/assistant/assistant_conversation_params.py b/api/core/model_runtime/model_providers/zhipuai/zhipuai_sdk/types/assistant/assistant_conversation_params.py
deleted file mode 100644
index 03f14f4238f37..0000000000000
--- a/api/core/model_runtime/model_providers/zhipuai/zhipuai_sdk/types/assistant/assistant_conversation_params.py
+++ /dev/null
@@ -1,7 +0,0 @@
-from typing import TypedDict
-
-
-class ConversationParameters(TypedDict, total=False):
-    assistant_id: str  # 智能体 ID
-    page: int  # 当前分页
-    page_size: int  # 分页数量
diff --git a/api/core/model_runtime/model_providers/zhipuai/zhipuai_sdk/types/assistant/assistant_conversation_resp.py b/api/core/model_runtime/model_providers/zhipuai/zhipuai_sdk/types/assistant/assistant_conversation_resp.py
deleted file mode 100644
index d1833d220a2e3..0000000000000
--- a/api/core/model_runtime/model_providers/zhipuai/zhipuai_sdk/types/assistant/assistant_conversation_resp.py
+++ /dev/null
@@ -1,29 +0,0 @@
-from ...core import BaseModel
-
-__all__ = ["ConversationUsageListResp"]
-
-
-class Usage(BaseModel):
-    prompt_tokens: int  # 用户输入的 tokens 数量
-    completion_tokens: int  # 模型输入的 tokens 数量
-    total_tokens: int  # 总 tokens 数量
-
-
-class ConversationUsage(BaseModel):
-    id: str  # 会话 id
-    assistant_id: str  # 智能体Assistant id
-    create_time: int  # 创建时间
-    update_time: int  # 更新时间
-    usage: Usage  # 会话中 tokens 数量统计
-
-
-class ConversationUsageList(BaseModel):
-    assistant_id: str  # 智能体id
-    has_more: bool  # 是否还有更多页
-    conversation_list: list[ConversationUsage]  # 返回的
-
-
-class ConversationUsageListResp(BaseModel):
-    code: int
-    msg: str
-    data: ConversationUsageList
diff --git a/api/core/model_runtime/model_providers/zhipuai/zhipuai_sdk/types/assistant/assistant_create_params.py b/api/core/model_runtime/model_providers/zhipuai/zhipuai_sdk/types/assistant/assistant_create_params.py
deleted file mode 100644
index 2def1025cd2b3..0000000000000
--- a/api/core/model_runtime/model_providers/zhipuai/zhipuai_sdk/types/assistant/assistant_create_params.py
+++ /dev/null
@@ -1,32 +0,0 @@
-from typing import Optional, TypedDict, Union
-
-
-class AssistantAttachments:
-    file_id: str
-
-
-class MessageTextContent:
-    type: str  # 目前支持 type = text
-    text: str
-
-
-MessageContent = Union[MessageTextContent]
-
-
-class ConversationMessage(TypedDict):
-    """会话消息体"""
-
-    role: str  # 用户的输入角色，例如 'user'
-    content: list[MessageContent]  # 会话消息体的内容
-
-
-class AssistantParameters(TypedDict, total=False):
-    """智能体参数类"""
-
-    assistant_id: str  # 智能体 ID
-    conversation_id: Optional[str]  # 会话 ID，不传则创建新会话
-    model: str  # 模型名称，默认为 'GLM-4-Assistant'
-    stream: bool  # 是否支持流式 SSE，需要传入 True
-    messages: list[ConversationMessage]  # 会话消息体
-    attachments: Optional[list[AssistantAttachments]]  # 会话指定的文件，非必填
-    metadata: Optional[dict]  # 元信息，拓展字段，非必填
diff --git a/api/core/model_runtime/model_providers/zhipuai/zhipuai_sdk/types/assistant/assistant_support_resp.py b/api/core/model_runtime/model_providers/zhipuai/zhipuai_sdk/types/assistant/assistant_support_resp.py
deleted file mode 100644
index 0709cdbcad25e..0000000000000
--- a/api/core/model_runtime/model_providers/zhipuai/zhipuai_sdk/types/assistant/assistant_support_resp.py
+++ /dev/null
@@ -1,21 +0,0 @@
-from ...core import BaseModel
-
-__all__ = ["AssistantSupportResp"]
-
-
-class AssistantSupport(BaseModel):
-    assistant_id: str  # 智能体的 Assistant id，用于智能体会话
-    created_at: int  # 创建时间
-    updated_at: int  # 更新时间
-    name: str  # 智能体名称
-    avatar: str  # 智能体头像
-    description: str  # 智能体描述
-    status: str  # 智能体状态，目前只有 publish
-    tools: list[str]  # 智能体支持的工具名
-    starter_prompts: list[str]  # 智能体启动推荐的 prompt
-
-
-class AssistantSupportResp(BaseModel):
-    code: int
-    msg: str
-    data: list[AssistantSupport]  # 智能体列表
diff --git a/api/core/model_runtime/model_providers/zhipuai/zhipuai_sdk/types/assistant/message/__init__.py b/api/core/model_runtime/model_providers/zhipuai/zhipuai_sdk/types/assistant/message/__init__.py
deleted file mode 100644
index 562e0151e53b4..0000000000000
--- a/api/core/model_runtime/model_providers/zhipuai/zhipuai_sdk/types/assistant/message/__init__.py
+++ /dev/null
@@ -1,3 +0,0 @@
-from .message_content import MessageContent
-
-__all__ = ["MessageContent"]
diff --git a/api/core/model_runtime/model_providers/zhipuai/zhipuai_sdk/types/assistant/message/message_content.py b/api/core/model_runtime/model_providers/zhipuai/zhipuai_sdk/types/assistant/message/message_content.py
deleted file mode 100644
index 6a1a438a6fe03..0000000000000
--- a/api/core/model_runtime/model_providers/zhipuai/zhipuai_sdk/types/assistant/message/message_content.py
+++ /dev/null
@@ -1,13 +0,0 @@
-from typing import Annotated, TypeAlias, Union
-
-from ....core._utils import PropertyInfo
-from .text_content_block import TextContentBlock
-from .tools_delta_block import ToolsDeltaBlock
-
-__all__ = ["MessageContent"]
-
-
-MessageContent: TypeAlias = Annotated[
-    Union[ToolsDeltaBlock, TextContentBlock],
-    PropertyInfo(discriminator="type"),
-]
diff --git a/api/core/model_runtime/model_providers/zhipuai/zhipuai_sdk/types/assistant/message/text_content_block.py b/api/core/model_runtime/model_providers/zhipuai/zhipuai_sdk/types/assistant/message/text_content_block.py
deleted file mode 100644
index 865fb1139e2f7..0000000000000
--- a/api/core/model_runtime/model_providers/zhipuai/zhipuai_sdk/types/assistant/message/text_content_block.py
+++ /dev/null
@@ -1,14 +0,0 @@
-from typing import Literal
-
-from ....core import BaseModel
-
-__all__ = ["TextContentBlock"]
-
-
-class TextContentBlock(BaseModel):
-    content: str
-
-    role: str = "assistant"
-
-    type: Literal["content"] = "content"
-    """Always `content`."""
diff --git a/api/core/model_runtime/model_providers/zhipuai/zhipuai_sdk/types/assistant/message/tools/code_interpreter_delta_block.py b/api/core/model_runtime/model_providers/zhipuai/zhipuai_sdk/types/assistant/message/tools/code_interpreter_delta_block.py
deleted file mode 100644
index 9d569b282ef9f..0000000000000
--- a/api/core/model_runtime/model_providers/zhipuai/zhipuai_sdk/types/assistant/message/tools/code_interpreter_delta_block.py
+++ /dev/null
@@ -1,27 +0,0 @@
-from typing import Literal
-
-__all__ = ["CodeInterpreterToolBlock"]
-
-from .....core import BaseModel
-
-
-class CodeInterpreterToolOutput(BaseModel):
-    """代码工具输出结果"""
-
-    type: str  # 代码执行日志，目前只有 logs
-    logs: str  # 代码执行的日志结果
-    error_msg: str  # 错误信息
-
-
-class CodeInterpreter(BaseModel):
-    """代码解释器"""
-
-    input: str  # 生成的代码片段，输入给代码沙盒
-    outputs: list[CodeInterpreterToolOutput]  # 代码执行后的输出结果
-
-
-class CodeInterpreterToolBlock(BaseModel):
-    """代码工具块"""
-
-    code_interpreter: CodeInterpreter  # 代码解释器对象
-    type: Literal["code_interpreter"]  # 调用工具的类型，始终为 `code_interpreter`
diff --git a/api/core/model_runtime/model_providers/zhipuai/zhipuai_sdk/types/assistant/message/tools/drawing_tool_delta_block.py b/api/core/model_runtime/model_providers/zhipuai/zhipuai_sdk/types/assistant/message/tools/drawing_tool_delta_block.py
deleted file mode 100644
index 0b6895556b616..0000000000000
--- a/api/core/model_runtime/model_providers/zhipuai/zhipuai_sdk/types/assistant/message/tools/drawing_tool_delta_block.py
+++ /dev/null
@@ -1,21 +0,0 @@
-from typing import Literal
-
-from .....core import BaseModel
-
-__all__ = ["DrawingToolBlock"]
-
-
-class DrawingToolOutput(BaseModel):
-    image: str
-
-
-class DrawingTool(BaseModel):
-    input: str
-    outputs: list[DrawingToolOutput]
-
-
-class DrawingToolBlock(BaseModel):
-    drawing_tool: DrawingTool
-
-    type: Literal["drawing_tool"]
-    """Always `drawing_tool`."""
diff --git a/api/core/model_runtime/model_providers/zhipuai/zhipuai_sdk/types/assistant/message/tools/function_delta_block.py b/api/core/model_runtime/model_providers/zhipuai/zhipuai_sdk/types/assistant/message/tools/function_delta_block.py
deleted file mode 100644
index c439bc4b3fbbb..0000000000000
--- a/api/core/model_runtime/model_providers/zhipuai/zhipuai_sdk/types/assistant/message/tools/function_delta_block.py
+++ /dev/null
@@ -1,22 +0,0 @@
-from typing import Literal, Union
-
-__all__ = ["FunctionToolBlock"]
-
-from .....core import BaseModel
-
-
-class FunctionToolOutput(BaseModel):
-    content: str
-
-
-class FunctionTool(BaseModel):
-    name: str
-    arguments: Union[str, dict]
-    outputs: list[FunctionToolOutput]
-
-
-class FunctionToolBlock(BaseModel):
-    function: FunctionTool
-
-    type: Literal["function"]
-    """Always `drawing_tool`."""
diff --git a/api/core/model_runtime/model_providers/zhipuai/zhipuai_sdk/types/assistant/message/tools/retrieval_delta_black.py b/api/core/model_runtime/model_providers/zhipuai/zhipuai_sdk/types/assistant/message/tools/retrieval_delta_black.py
deleted file mode 100644
index 4789e9378a8a3..0000000000000
--- a/api/core/model_runtime/model_providers/zhipuai/zhipuai_sdk/types/assistant/message/tools/retrieval_delta_black.py
+++ /dev/null
@@ -1,41 +0,0 @@
-from typing import Literal
-
-from .....core import BaseModel
-
-
-class RetrievalToolOutput(BaseModel):
-    """
-    This class represents the output of a retrieval tool.
-
-    Attributes:
-    - text (str): The text snippet retrieved from the knowledge base.
-    - document (str): The name of the document from which the text snippet was retrieved, returned only in intelligent configuration.
-    """  # noqa: E501
-
-    text: str
-    document: str
-
-
-class RetrievalTool(BaseModel):
-    """
-    This class represents the outputs of a retrieval tool.
-
-    Attributes:
-    - outputs (List[RetrievalToolOutput]): A list of text snippets and their respective document names retrieved from the knowledge base.
-    """  # noqa: E501
-
-    outputs: list[RetrievalToolOutput]
-
-
-class RetrievalToolBlock(BaseModel):
-    """
-    This class represents a block for invoking the retrieval tool.
-
-    Attributes:
-    - retrieval (RetrievalTool): An instance of the RetrievalTool class containing the retrieval outputs.
-    - type (Literal["retrieval"]): The type of tool being used, always set to "retrieval".
-    """
-
-    retrieval: RetrievalTool
-    type: Literal["retrieval"]
-    """Always `retrieval`."""
diff --git a/api/core/model_runtime/model_providers/zhipuai/zhipuai_sdk/types/assistant/message/tools/tools_type.py b/api/core/model_runtime/model_providers/zhipuai/zhipuai_sdk/types/assistant/message/tools/tools_type.py
deleted file mode 100644
index 98544053d4c83..0000000000000
--- a/api/core/model_runtime/model_providers/zhipuai/zhipuai_sdk/types/assistant/message/tools/tools_type.py
+++ /dev/null
@@ -1,16 +0,0 @@
-from typing import Annotated, TypeAlias, Union
-
-from .....core._utils import PropertyInfo
-from .code_interpreter_delta_block import CodeInterpreterToolBlock
-from .drawing_tool_delta_block import DrawingToolBlock
-from .function_delta_block import FunctionToolBlock
-from .retrieval_delta_black import RetrievalToolBlock
-from .web_browser_delta_block import WebBrowserToolBlock
-
-__all__ = ["ToolsType"]
-
-
-ToolsType: TypeAlias = Annotated[
-    Union[DrawingToolBlock, CodeInterpreterToolBlock, WebBrowserToolBlock, RetrievalToolBlock, FunctionToolBlock],
-    PropertyInfo(discriminator="type"),
-]
diff --git a/api/core/model_runtime/model_providers/zhipuai/zhipuai_sdk/types/assistant/message/tools/web_browser_delta_block.py b/api/core/model_runtime/model_providers/zhipuai/zhipuai_sdk/types/assistant/message/tools/web_browser_delta_block.py
deleted file mode 100644
index 966e6fe0c84fe..0000000000000
--- a/api/core/model_runtime/model_providers/zhipuai/zhipuai_sdk/types/assistant/message/tools/web_browser_delta_block.py
+++ /dev/null
@@ -1,48 +0,0 @@
-from typing import Literal
-
-from .....core import BaseModel
-
-__all__ = ["WebBrowserToolBlock"]
-
-
-class WebBrowserOutput(BaseModel):
-    """
-    This class represents the output of a web browser search result.
-
-    Attributes:
-    - title (str): The title of the search result.
-    - link (str): The URL link to the search result's webpage.
-    - content (str): The textual content extracted from the search result.
-    - error_msg (str): Any error message encountered during the search or retrieval process.
-    """
-
-    title: str
-    link: str
-    content: str
-    error_msg: str
-
-
-class WebBrowser(BaseModel):
-    """
-    This class represents the input and outputs of a web browser search.
-
-    Attributes:
-    - input (str): The input query for the web browser search.
-    - outputs (List[WebBrowserOutput]): A list of search results returned by the web browser.
-    """
-
-    input: str
-    outputs: list[WebBrowserOutput]
-
-
-class WebBrowserToolBlock(BaseModel):
-    """
-    This class represents a block for invoking the web browser tool.
-
-    Attributes:
-    - web_browser (WebBrowser): An instance of the WebBrowser class containing the search input and outputs.
-    - type (Literal["web_browser"]): The type of tool being used, always set to "web_browser".
-    """
-
-    web_browser: WebBrowser
-    type: Literal["web_browser"]
diff --git a/api/core/model_runtime/model_providers/zhipuai/zhipuai_sdk/types/assistant/message/tools_delta_block.py b/api/core/model_runtime/model_providers/zhipuai/zhipuai_sdk/types/assistant/message/tools_delta_block.py
deleted file mode 100644
index 781a1ab819c28..0000000000000
--- a/api/core/model_runtime/model_providers/zhipuai/zhipuai_sdk/types/assistant/message/tools_delta_block.py
+++ /dev/null
@@ -1,16 +0,0 @@
-from typing import Literal
-
-from ....core import BaseModel
-from .tools.tools_type import ToolsType
-
-__all__ = ["ToolsDeltaBlock"]
-
-
-class ToolsDeltaBlock(BaseModel):
-    tool_calls: list[ToolsType]
-    """The index of the content part in the message."""
-
-    role: str = "tool"
-
-    type: Literal["tool_calls"] = "tool_calls"
-    """Always `tool_calls`."""
diff --git a/api/core/model_runtime/model_providers/zhipuai/zhipuai_sdk/types/batch.py b/api/core/model_runtime/model_providers/zhipuai/zhipuai_sdk/types/batch.py
deleted file mode 100644
index 560562915c9d3..0000000000000
--- a/api/core/model_runtime/model_providers/zhipuai/zhipuai_sdk/types/batch.py
+++ /dev/null
@@ -1,82 +0,0 @@
-# File generated from our OpenAPI spec by Stainless. See CONTRIBUTING.md for details.
-
-import builtins
-from typing import Literal, Optional
-
-from ..core import BaseModel
-from .batch_error import BatchError
-from .batch_request_counts import BatchRequestCounts
-
-__all__ = ["Batch", "Errors"]
-
-
-class Errors(BaseModel):
-    data: Optional[list[BatchError]] = None
-
-    object: Optional[str] = None
-    """这个类型，一直是`list`。"""
-
-
-class Batch(BaseModel):
-    id: str
-
-    completion_window: str
-    """用于执行请求的地址信息。"""
-
-    created_at: int
-    """这是 Unix timestamp (in seconds) 表示的创建时间。"""
-
-    endpoint: str
-    """这是ZhipuAI endpoint的地址。"""
-
-    input_file_id: str
-    """标记为batch的输入文件的ID。"""
-
-    object: Literal["batch"]
-    """这个类型，一直是`batch`."""
-
-    status: Literal[
-        "validating", "failed", "in_progress", "finalizing", "completed", "expired", "cancelling", "cancelled"
-    ]
-    """batch 的状态。"""
-
-    cancelled_at: Optional[int] = None
-    """Unix timestamp (in seconds) 表示的取消时间。"""
-
-    cancelling_at: Optional[int] = None
-    """Unix timestamp (in seconds) 表示发起取消的请求时间 """
-
-    completed_at: Optional[int] = None
-    """Unix timestamp (in seconds) 表示的完成时间。"""
-
-    error_file_id: Optional[str] = None
-    """这个文件id包含了执行请求失败的请求的输出。"""
-
-    errors: Optional[Errors] = None
-
-    expired_at: Optional[int] = None
-    """Unix timestamp (in seconds) 表示的将在过期时间。"""
-
-    expires_at: Optional[int] = None
-    """Unix timestamp (in seconds) 触发过期"""
-
-    failed_at: Optional[int] = None
-    """Unix timestamp (in seconds) 表示的失败时间。"""
-
-    finalizing_at: Optional[int] = None
-    """Unix timestamp (in seconds) 表示的最终时间。"""
-
-    in_progress_at: Optional[int] = None
-    """Unix timestamp (in seconds) 表示的开始处理时间。"""
-
-    metadata: Optional[builtins.object] = None
-    """ 
-    key:value形式的元数据，以便将信息存储
-        结构化格式。键的长度是64个字符，值最长512个字符
-    """
-
-    output_file_id: Optional[str] = None
-    """完成请求的输出文件的ID。"""
-
-    request_counts: Optional[BatchRequestCounts] = None
-    """批次中不同状态的请求计数"""
diff --git a/api/core/model_runtime/model_providers/zhipuai/zhipuai_sdk/types/batch_create_params.py b/api/core/model_runtime/model_providers/zhipuai/zhipuai_sdk/types/batch_create_params.py
deleted file mode 100644
index 3dae65ea46fcb..0000000000000
--- a/api/core/model_runtime/model_providers/zhipuai/zhipuai_sdk/types/batch_create_params.py
+++ /dev/null
@@ -1,37 +0,0 @@
-from __future__ import annotations
-
-from typing import Literal, Optional
-
-from typing_extensions import Required, TypedDict
-
-__all__ = ["BatchCreateParams"]
-
-
-class BatchCreateParams(TypedDict, total=False):
-    completion_window: Required[str]
-    """The time frame within which the batch should be processed.
-
-    Currently only `24h` is supported.
-    """
-
-    endpoint: Required[Literal["/v1/chat/completions", "/v1/embeddings"]]
-    """The endpoint to be used for all requests in the batch.
-
-    Currently `/v1/chat/completions` and `/v1/embeddings` are supported.
-    """
-
-    input_file_id: Required[str]
-    """The ID of an uploaded file that contains requests for the new batch.
-
-    See [upload file](https://platform.openai.com/docs/api-reference/files/create)
-    for how to upload a file.
-
-    Your input file must be formatted as a
-    [JSONL file](https://platform.openai.com/docs/api-reference/batch/requestInput),
-    and must be uploaded with the purpose `batch`.
-    """
-
-    metadata: Optional[dict[str, str]]
-    """Optional custom metadata for the batch."""
-
-    auto_delete_input_file: Optional[bool]
diff --git a/api/core/model_runtime/model_providers/zhipuai/zhipuai_sdk/types/batch_error.py b/api/core/model_runtime/model_providers/zhipuai/zhipuai_sdk/types/batch_error.py
deleted file mode 100644
index f934db19781e4..0000000000000
--- a/api/core/model_runtime/model_providers/zhipuai/zhipuai_sdk/types/batch_error.py
+++ /dev/null
@@ -1,21 +0,0 @@
-# File generated from our OpenAPI spec by Stainless. See CONTRIBUTING.md for details.
-
-from typing import Optional
-
-from ..core import BaseModel
-
-__all__ = ["BatchError"]
-
-
-class BatchError(BaseModel):
-    code: Optional[str] = None
-    """定义的业务错误码"""
-
-    line: Optional[int] = None
-    """文件中的行号"""
-
-    message: Optional[str] = None
-    """关于对话文件中的错误的描述"""
-
-    param: Optional[str] = None
-    """参数名称，如果有的话"""
diff --git a/api/core/model_runtime/model_providers/zhipuai/zhipuai_sdk/types/batch_list_params.py b/api/core/model_runtime/model_providers/zhipuai/zhipuai_sdk/types/batch_list_params.py
deleted file mode 100644
index 1a681671320ec..0000000000000
--- a/api/core/model_runtime/model_providers/zhipuai/zhipuai_sdk/types/batch_list_params.py
+++ /dev/null
@@ -1,20 +0,0 @@
-from __future__ import annotations
-
-from typing_extensions import TypedDict
-
-__all__ = ["BatchListParams"]
-
-
-class BatchListParams(TypedDict, total=False):
-    after: str
-    """分页的游标，用于获取下一页的数据。
-
-    `after` 是一个指向当前页面的游标，用于获取下一页的数据。如果没有提供 `after`，则返回第一页的数据。
-    list.
-    """
-
-    limit: int
-    """这个参数用于限制返回的结果数量。
-
-    Limit 用于限制返回的结果数量。默认值为 10
-    """
diff --git a/api/core/model_runtime/model_providers/zhipuai/zhipuai_sdk/types/batch_request_counts.py b/api/core/model_runtime/model_providers/zhipuai/zhipuai_sdk/types/batch_request_counts.py
deleted file mode 100644
index ca3ccae625052..0000000000000
--- a/api/core/model_runtime/model_providers/zhipuai/zhipuai_sdk/types/batch_request_counts.py
+++ /dev/null
@@ -1,14 +0,0 @@
-from ..core import BaseModel
-
-__all__ = ["BatchRequestCounts"]
-
-
-class BatchRequestCounts(BaseModel):
-    completed: int
-    """这个数字表示已经完成的请求。"""
-
-    failed: int
-    """这个数字表示失败的请求。"""
-
-    total: int
-    """这个数字表示总的请求。"""
diff --git a/api/core/model_runtime/model_providers/zhipuai/zhipuai_sdk/types/chat/__init__.py b/api/core/model_runtime/model_providers/zhipuai/zhipuai_sdk/types/chat/__init__.py
deleted file mode 100644
index e69de29bb2d1d..0000000000000
diff --git a/api/core/model_runtime/model_providers/zhipuai/zhipuai_sdk/types/chat/async_chat_completion.py b/api/core/model_runtime/model_providers/zhipuai/zhipuai_sdk/types/chat/async_chat_completion.py
deleted file mode 100644
index c1eed070f32d9..0000000000000
--- a/api/core/model_runtime/model_providers/zhipuai/zhipuai_sdk/types/chat/async_chat_completion.py
+++ /dev/null
@@ -1,22 +0,0 @@
-from typing import Optional
-
-from ...core import BaseModel
-from .chat_completion import CompletionChoice, CompletionUsage
-
-__all__ = ["AsyncTaskStatus", "AsyncCompletion"]
-
-
-class AsyncTaskStatus(BaseModel):
-    id: Optional[str] = None
-    request_id: Optional[str] = None
-    model: Optional[str] = None
-    task_status: Optional[str] = None
-
-
-class AsyncCompletion(BaseModel):
-    id: Optional[str] = None
-    request_id: Optional[str] = None
-    model: Optional[str] = None
-    task_status: str
-    choices: list[CompletionChoice]
-    usage: CompletionUsage
diff --git a/api/core/model_runtime/model_providers/zhipuai/zhipuai_sdk/types/chat/chat_completion.py b/api/core/model_runtime/model_providers/zhipuai/zhipuai_sdk/types/chat/chat_completion.py
deleted file mode 100644
index 1945a826cda2d..0000000000000
--- a/api/core/model_runtime/model_providers/zhipuai/zhipuai_sdk/types/chat/chat_completion.py
+++ /dev/null
@@ -1,43 +0,0 @@
-from typing import Optional
-
-from ...core import BaseModel
-
-__all__ = ["Completion", "CompletionUsage"]
-
-
-class Function(BaseModel):
-    arguments: str
-    name: str
-
-
-class CompletionMessageToolCall(BaseModel):
-    id: str
-    function: Function
-    type: str
-
-
-class CompletionMessage(BaseModel):
-    content: Optional[str] = None
-    role: str
-    tool_calls: Optional[list[CompletionMessageToolCall]] = None
-
-
-class CompletionUsage(BaseModel):
-    prompt_tokens: int
-    completion_tokens: int
-    total_tokens: int
-
-
-class CompletionChoice(BaseModel):
-    index: int
-    finish_reason: str
-    message: CompletionMessage
-
-
-class Completion(BaseModel):
-    model: Optional[str] = None
-    created: Optional[int] = None
-    choices: list[CompletionChoice]
-    request_id: Optional[str] = None
-    id: Optional[str] = None
-    usage: CompletionUsage
diff --git a/api/core/model_runtime/model_providers/zhipuai/zhipuai_sdk/types/chat/chat_completion_chunk.py b/api/core/model_runtime/model_providers/zhipuai/zhipuai_sdk/types/chat/chat_completion_chunk.py
deleted file mode 100644
index 27fad0008a1dd..0000000000000
--- a/api/core/model_runtime/model_providers/zhipuai/zhipuai_sdk/types/chat/chat_completion_chunk.py
+++ /dev/null
@@ -1,57 +0,0 @@
-from typing import Any, Optional
-
-from ...core import BaseModel
-
-__all__ = [
-    "CompletionUsage",
-    "ChatCompletionChunk",
-    "Choice",
-    "ChoiceDelta",
-    "ChoiceDeltaFunctionCall",
-    "ChoiceDeltaToolCall",
-    "ChoiceDeltaToolCallFunction",
-]
-
-
-class ChoiceDeltaFunctionCall(BaseModel):
-    arguments: Optional[str] = None
-    name: Optional[str] = None
-
-
-class ChoiceDeltaToolCallFunction(BaseModel):
-    arguments: Optional[str] = None
-    name: Optional[str] = None
-
-
-class ChoiceDeltaToolCall(BaseModel):
-    index: int
-    id: Optional[str] = None
-    function: Optional[ChoiceDeltaToolCallFunction] = None
-    type: Optional[str] = None
-
-
-class ChoiceDelta(BaseModel):
-    content: Optional[str] = None
-    role: Optional[str] = None
-    tool_calls: Optional[list[ChoiceDeltaToolCall]] = None
-
-
-class Choice(BaseModel):
-    delta: ChoiceDelta
-    finish_reason: Optional[str] = None
-    index: int
-
-
-class CompletionUsage(BaseModel):
-    prompt_tokens: int
-    completion_tokens: int
-    total_tokens: int
-
-
-class ChatCompletionChunk(BaseModel):
-    id: Optional[str] = None
-    choices: list[Choice]
-    created: Optional[int] = None
-    model: Optional[str] = None
-    usage: Optional[CompletionUsage] = None
-    extra_json: dict[str, Any]
diff --git a/api/core/model_runtime/model_providers/zhipuai/zhipuai_sdk/types/chat/chat_completions_create_param.py b/api/core/model_runtime/model_providers/zhipuai/zhipuai_sdk/types/chat/chat_completions_create_param.py
deleted file mode 100644
index 6ee4dc4794b20..0000000000000
--- a/api/core/model_runtime/model_providers/zhipuai/zhipuai_sdk/types/chat/chat_completions_create_param.py
+++ /dev/null
@@ -1,8 +0,0 @@
-from typing import Optional
-
-from typing_extensions import TypedDict
-
-
-class Reference(TypedDict, total=False):
-    enable: Optional[bool]
-    search_query: Optional[str]
diff --git a/api/core/model_runtime/model_providers/zhipuai/zhipuai_sdk/types/chat/code_geex/code_geex_params.py b/api/core/model_runtime/model_providers/zhipuai/zhipuai_sdk/types/chat/code_geex/code_geex_params.py
deleted file mode 100644
index 666b38855cd63..0000000000000
--- a/api/core/model_runtime/model_providers/zhipuai/zhipuai_sdk/types/chat/code_geex/code_geex_params.py
+++ /dev/null
@@ -1,146 +0,0 @@
-from typing import Literal, Optional
-
-from typing_extensions import Required, TypedDict
-
-__all__ = [
-    "CodeGeexTarget",
-    "CodeGeexContext",
-    "CodeGeexExtra",
-]
-
-
-class CodeGeexTarget(TypedDict, total=False):
-    """补全的内容参数"""
-
-    path: Optional[str]
-    """文件路径"""
-    language: Required[
-        Literal[
-            "c",
-            "c++",
-            "cpp",
-            "c#",
-            "csharp",
-            "c-sharp",
-            "css",
-            "cuda",
-            "dart",
-            "lua",
-            "objectivec",
-            "objective-c",
-            "objective-c++",
-            "python",
-            "perl",
-            "prolog",
-            "swift",
-            "lisp",
-            "java",
-            "scala",
-            "tex",
-            "jsx",
-            "tsx",
-            "vue",
-            "markdown",
-            "html",
-            "php",
-            "js",
-            "javascript",
-            "typescript",
-            "go",
-            "shell",
-            "rust",
-            "sql",
-            "kotlin",
-            "vb",
-            "ruby",
-            "pascal",
-            "r",
-            "fortran",
-            "lean",
-            "matlab",
-            "delphi",
-            "scheme",
-            "basic",
-            "assembly",
-            "groovy",
-            "abap",
-            "gdscript",
-            "haskell",
-            "julia",
-            "elixir",
-            "excel",
-            "clojure",
-            "actionscript",
-            "solidity",
-            "powershell",
-            "erlang",
-            "cobol",
-            "alloy",
-            "awk",
-            "thrift",
-            "sparql",
-            "augeas",
-            "cmake",
-            "f-sharp",
-            "stan",
-            "isabelle",
-            "dockerfile",
-            "rmarkdown",
-            "literate-agda",
-            "tcl",
-            "glsl",
-            "antlr",
-            "verilog",
-            "racket",
-            "standard-ml",
-            "elm",
-            "yaml",
-            "smalltalk",
-            "ocaml",
-            "idris",
-            "visual-basic",
-            "protocol-buffer",
-            "bluespec",
-            "applescript",
-            "makefile",
-            "tcsh",
-            "maple",
-            "systemverilog",
-            "literate-coffeescript",
-            "vhdl",
-            "restructuredtext",
-            "sas",
-            "literate-haskell",
-            "java-server-pages",
-            "coffeescript",
-            "emacs-lisp",
-            "mathematica",
-            "xslt",
-            "zig",
-            "common-lisp",
-            "stata",
-            "agda",
-            "ada",
-        ]
-    ]
-    """代码语言类型，如python"""
-    code_prefix: Required[str]
-    """补全位置的前文"""
-    code_suffix: Required[str]
-    """补全位置的后文"""
-
-
-class CodeGeexContext(TypedDict, total=False):
-    """附加代码"""
-
-    path: Required[str]
-    """附加代码文件的路径"""
-    code: Required[str]
-    """附加的代码内容"""
-
-
-class CodeGeexExtra(TypedDict, total=False):
-    target: Required[CodeGeexTarget]
-    """补全的内容参数"""
-    contexts: Optional[list[CodeGeexContext]]
-    """附加代码"""
diff --git a/api/core/model_runtime/model_providers/zhipuai/zhipuai_sdk/types/embeddings.py b/api/core/model_runtime/model_providers/zhipuai/zhipuai_sdk/types/embeddings.py
deleted file mode 100644
index 8425b5c86688d..0000000000000
--- a/api/core/model_runtime/model_providers/zhipuai/zhipuai_sdk/types/embeddings.py
+++ /dev/null
@@ -1,21 +0,0 @@
-from __future__ import annotations
-
-from typing import Optional
-
-from ..core import BaseModel
-from .chat.chat_completion import CompletionUsage
-
-__all__ = ["Embedding", "EmbeddingsResponded"]
-
-
-class Embedding(BaseModel):
-    object: str
-    index: Optional[int] = None
-    embedding: list[float]
-
-
-class EmbeddingsResponded(BaseModel):
-    object: str
-    data: list[Embedding]
-    model: str
-    usage: CompletionUsage
diff --git a/api/core/model_runtime/model_providers/zhipuai/zhipuai_sdk/types/files/__init__.py b/api/core/model_runtime/model_providers/zhipuai/zhipuai_sdk/types/files/__init__.py
deleted file mode 100644
index bbaf59e4d7d17..0000000000000
--- a/api/core/model_runtime/model_providers/zhipuai/zhipuai_sdk/types/files/__init__.py
+++ /dev/null
@@ -1,5 +0,0 @@
-from .file_deleted import FileDeleted
-from .file_object import FileObject, ListOfFileObject
-from .upload_detail import UploadDetail
-
-__all__ = ["FileObject", "ListOfFileObject", "UploadDetail", "FileDeleted"]
diff --git a/api/core/model_runtime/model_providers/zhipuai/zhipuai_sdk/types/files/file_create_params.py b/api/core/model_runtime/model_providers/zhipuai/zhipuai_sdk/types/files/file_create_params.py
deleted file mode 100644
index 4ef93b1c05aca..0000000000000
--- a/api/core/model_runtime/model_providers/zhipuai/zhipuai_sdk/types/files/file_create_params.py
+++ /dev/null
@@ -1,38 +0,0 @@
-from __future__ import annotations
-
-from typing import Literal, Optional
-
-from typing_extensions import Required, TypedDict
-
-__all__ = ["FileCreateParams"]
-
-from ...core import FileTypes
-from . import UploadDetail
-
-
-class FileCreateParams(TypedDict, total=False):
-    file: FileTypes
-    """file和 upload_detail二选一必填"""
-
-    upload_detail: list[UploadDetail]
-    """file和 upload_detail二选一必填"""
-
-    purpose: Required[Literal["fine-tune", "retrieval", "batch"]]
-    """ 
-    上传文件的用途，支持 "fine-tune和 "retrieval"
-    retrieval支持上传Doc、Docx、PDF、Xlsx、URL类型文件，且单个文件的大小不超过 5MB。
-    fine-tune支持上传.jsonl文件且当前单个文件的大小最大可为 100 MB ，文件中语料格式需满足微调指南中所描述的格式。
-    """
-    custom_separator: Optional[list[str]]
-    """ 
-    当 purpose 为 retrieval 且文件类型为 pdf, url, docx 时上传，切片规则默认为 `\n`。
-    """
-    knowledge_id: str
-    """ 
-        当文件上传目的为  retrieval 时，需要指定知识库ID进行上传。
-    """
-
-    sentence_size: int
-    """ 
-        当文件上传目的为  retrieval 时，需要指定知识库ID进行上传。
-    """
diff --git a/api/core/model_runtime/model_providers/zhipuai/zhipuai_sdk/types/files/file_deleted.py b/api/core/model_runtime/model_providers/zhipuai/zhipuai_sdk/types/files/file_deleted.py
deleted file mode 100644
index a384b1a69a573..0000000000000
--- a/api/core/model_runtime/model_providers/zhipuai/zhipuai_sdk/types/files/file_deleted.py
+++ /dev/null
@@ -1,13 +0,0 @@
-from typing import Literal
-
-from ...core import BaseModel
-
-__all__ = ["FileDeleted"]
-
-
-class FileDeleted(BaseModel):
-    id: str
-
-    deleted: bool
-
-    object: Literal["file"]
diff --git a/api/core/model_runtime/model_providers/zhipuai/zhipuai_sdk/types/files/file_object.py b/api/core/model_runtime/model_providers/zhipuai/zhipuai_sdk/types/files/file_object.py
deleted file mode 100644
index 8f9d0fbb8e6ce..0000000000000
--- a/api/core/model_runtime/model_providers/zhipuai/zhipuai_sdk/types/files/file_object.py
+++ /dev/null
@@ -1,22 +0,0 @@
-from typing import Optional
-
-from ...core import BaseModel
-
-__all__ = ["FileObject", "ListOfFileObject"]
-
-
-class FileObject(BaseModel):
-    id: Optional[str] = None
-    bytes: Optional[int] = None
-    created_at: Optional[int] = None
-    filename: Optional[str] = None
-    object: Optional[str] = None
-    purpose: Optional[str] = None
-    status: Optional[str] = None
-    status_details: Optional[str] = None
-
-
-class ListOfFileObject(BaseModel):
-    object: Optional[str] = None
-    data: list[FileObject]
-    has_more: Optional[bool] = None
diff --git a/api/core/model_runtime/model_providers/zhipuai/zhipuai_sdk/types/files/upload_detail.py b/api/core/model_runtime/model_providers/zhipuai/zhipuai_sdk/types/files/upload_detail.py
deleted file mode 100644
index 8f1ca5ce5756a..0000000000000
--- a/api/core/model_runtime/model_providers/zhipuai/zhipuai_sdk/types/files/upload_detail.py
+++ /dev/null
@@ -1,13 +0,0 @@
-from typing import Optional
-
-from ...core import BaseModel
-
-
-class UploadDetail(BaseModel):
-    url: str
-    knowledge_type: int
-    file_name: Optional[str] = None
-    sentence_size: Optional[int] = None
-    custom_separator: Optional[list[str]] = None
-    callback_url: Optional[str] = None
-    callback_header: Optional[dict[str, str]] = None
diff --git a/api/core/model_runtime/model_providers/zhipuai/zhipuai_sdk/types/fine_tuning/__init__.py b/api/core/model_runtime/model_providers/zhipuai/zhipuai_sdk/types/fine_tuning/__init__.py
deleted file mode 100644
index 416f516ef7bf1..0000000000000
--- a/api/core/model_runtime/model_providers/zhipuai/zhipuai_sdk/types/fine_tuning/__init__.py
+++ /dev/null
@@ -1,4 +0,0 @@
-from __future__ import annotations
-
-from .fine_tuning_job import FineTuningJob, ListOfFineTuningJob
-from .fine_tuning_job_event import FineTuningJobEvent
diff --git a/api/core/model_runtime/model_providers/zhipuai/zhipuai_sdk/types/fine_tuning/fine_tuning_job.py b/api/core/model_runtime/model_providers/zhipuai/zhipuai_sdk/types/fine_tuning/fine_tuning_job.py
deleted file mode 100644
index 75c7553dbe35c..0000000000000
--- a/api/core/model_runtime/model_providers/zhipuai/zhipuai_sdk/types/fine_tuning/fine_tuning_job.py
+++ /dev/null
@@ -1,51 +0,0 @@
-from typing import Optional, Union
-
-from ...core import BaseModel
-
-__all__ = ["FineTuningJob", "Error", "Hyperparameters", "ListOfFineTuningJob"]
-
-
-class Error(BaseModel):
-    code: str
-    message: str
-    param: Optional[str] = None
-
-
-class Hyperparameters(BaseModel):
-    n_epochs: Union[str, int, None] = None
-
-
-class FineTuningJob(BaseModel):
-    id: Optional[str] = None
-
-    request_id: Optional[str] = None
-
-    created_at: Optional[int] = None
-
-    error: Optional[Error] = None
-
-    fine_tuned_model: Optional[str] = None
-
-    finished_at: Optional[int] = None
-
-    hyperparameters: Optional[Hyperparameters] = None
-
-    model: Optional[str] = None
-
-    object: Optional[str] = None
-
-    result_files: list[str]
-
-    status: str
-
-    trained_tokens: Optional[int] = None
-
-    training_file: str
-
-    validation_file: Optional[str] = None
-
-
-class ListOfFineTuningJob(BaseModel):
-    object: Optional[str] = None
-    data: list[FineTuningJob]
-    has_more: Optional[bool] = None
diff --git a/api/core/model_runtime/model_providers/zhipuai/zhipuai_sdk/types/fine_tuning/fine_tuning_job_event.py b/api/core/model_runtime/model_providers/zhipuai/zhipuai_sdk/types/fine_tuning/fine_tuning_job_event.py
deleted file mode 100644
index f996cff11430b..0000000000000
--- a/api/core/model_runtime/model_providers/zhipuai/zhipuai_sdk/types/fine_tuning/fine_tuning_job_event.py
+++ /dev/null
@@ -1,35 +0,0 @@
-from typing import Optional, Union
-
-from ...core import BaseModel
-
-__all__ = ["FineTuningJobEvent", "Metric", "JobEvent"]
-
-
-class Metric(BaseModel):
-    epoch: Optional[Union[str, int, float]] = None
-    current_steps: Optional[int] = None
-    total_steps: Optional[int] = None
-    elapsed_time: Optional[str] = None
-    remaining_time: Optional[str] = None
-    trained_tokens: Optional[int] = None
-    loss: Optional[Union[str, int, float]] = None
-    eval_loss: Optional[Union[str, int, float]] = None
-    acc: Optional[Union[str, int, float]] = None
-    eval_acc: Optional[Union[str, int, float]] = None
-    learning_rate: Optional[Union[str, int, float]] = None
-
-
-class JobEvent(BaseModel):
-    object: Optional[str] = None
-    id: Optional[str] = None
-    type: Optional[str] = None
-    created_at: Optional[int] = None
-    level: Optional[str] = None
-    message: Optional[str] = None
-    data: Optional[Metric] = None
-
-
-class FineTuningJobEvent(BaseModel):
-    object: Optional[str] = None
-    data: list[JobEvent]
-    has_more: Optional[bool] = None
diff --git a/api/core/model_runtime/model_providers/zhipuai/zhipuai_sdk/types/fine_tuning/job_create_params.py b/api/core/model_runtime/model_providers/zhipuai/zhipuai_sdk/types/fine_tuning/job_create_params.py
deleted file mode 100644
index e1ebc352bc97f..0000000000000
--- a/api/core/model_runtime/model_providers/zhipuai/zhipuai_sdk/types/fine_tuning/job_create_params.py
+++ /dev/null
@@ -1,15 +0,0 @@
-from __future__ import annotations
-
-from typing import Literal, Union
-
-from typing_extensions import TypedDict
-
-__all__ = ["Hyperparameters"]
-
-
-class Hyperparameters(TypedDict, total=False):
-    batch_size: Union[Literal["auto"], int]
-
-    learning_rate_multiplier: Union[Literal["auto"], float]
-
-    n_epochs: Union[Literal["auto"], int]
diff --git a/api/core/model_runtime/model_providers/zhipuai/zhipuai_sdk/types/fine_tuning/models/__init__.py b/api/core/model_runtime/model_providers/zhipuai/zhipuai_sdk/types/fine_tuning/models/__init__.py
deleted file mode 100644
index 57d0d2511dbc1..0000000000000
--- a/api/core/model_runtime/model_providers/zhipuai/zhipuai_sdk/types/fine_tuning/models/__init__.py
+++ /dev/null
@@ -1 +0,0 @@
-from .fine_tuned_models import FineTunedModelsStatus
diff --git a/api/core/model_runtime/model_providers/zhipuai/zhipuai_sdk/types/fine_tuning/models/fine_tuned_models.py b/api/core/model_runtime/model_providers/zhipuai/zhipuai_sdk/types/fine_tuning/models/fine_tuned_models.py
deleted file mode 100644
index b286a5b5774d3..0000000000000
--- a/api/core/model_runtime/model_providers/zhipuai/zhipuai_sdk/types/fine_tuning/models/fine_tuned_models.py
+++ /dev/null
@@ -1,13 +0,0 @@
-from typing import ClassVar
-
-from ....core import PYDANTIC_V2, BaseModel, ConfigDict
-
-__all__ = ["FineTunedModelsStatus"]
-
-
-class FineTunedModelsStatus(BaseModel):
-    if PYDANTIC_V2:
-        model_config: ClassVar[ConfigDict] = ConfigDict(extra="allow", protected_namespaces=())
-    request_id: str  # 请求id
-    model_name: str  # 模型名称
-    delete_status: str  # 删除状态 deleting（删除中）, deleted （已删除）
diff --git a/api/core/model_runtime/model_providers/zhipuai/zhipuai_sdk/types/image.py b/api/core/model_runtime/model_providers/zhipuai/zhipuai_sdk/types/image.py
deleted file mode 100644
index 3bcad0acabd21..0000000000000
--- a/api/core/model_runtime/model_providers/zhipuai/zhipuai_sdk/types/image.py
+++ /dev/null
@@ -1,18 +0,0 @@
-from __future__ import annotations
-
-from typing import Optional
-
-from ..core import BaseModel
-
-__all__ = ["GeneratedImage", "ImagesResponded"]
-
-
-class GeneratedImage(BaseModel):
-    b64_json: Optional[str] = None
-    url: Optional[str] = None
-    revised_prompt: Optional[str] = None
-
-
-class ImagesResponded(BaseModel):
-    created: int
-    data: list[GeneratedImage]
diff --git a/api/core/model_runtime/model_providers/zhipuai/zhipuai_sdk/types/knowledge/__init__.py b/api/core/model_runtime/model_providers/zhipuai/zhipuai_sdk/types/knowledge/__init__.py
deleted file mode 100644
index 8c81d703e214a..0000000000000
--- a/api/core/model_runtime/model_providers/zhipuai/zhipuai_sdk/types/knowledge/__init__.py
+++ /dev/null
@@ -1,8 +0,0 @@
-from .knowledge import KnowledgeInfo
-from .knowledge_used import KnowledgeStatistics, KnowledgeUsed
-
-__all__ = [
-    "KnowledgeInfo",
-    "KnowledgeStatistics",
-    "KnowledgeUsed",
-]
diff --git a/api/core/model_runtime/model_providers/zhipuai/zhipuai_sdk/types/knowledge/document/__init__.py b/api/core/model_runtime/model_providers/zhipuai/zhipuai_sdk/types/knowledge/document/__init__.py
deleted file mode 100644
index 59cb41d7124a7..0000000000000
--- a/api/core/model_runtime/model_providers/zhipuai/zhipuai_sdk/types/knowledge/document/__init__.py
+++ /dev/null
@@ -1,8 +0,0 @@
-from .document import DocumentData, DocumentFailedInfo, DocumentObject, DocumentSuccessInfo
-
-__all__ = [
-    "DocumentData",
-    "DocumentObject",
-    "DocumentSuccessInfo",
-    "DocumentFailedInfo",
-]
diff --git a/api/core/model_runtime/model_providers/zhipuai/zhipuai_sdk/types/knowledge/document/document.py b/api/core/model_runtime/model_providers/zhipuai/zhipuai_sdk/types/knowledge/document/document.py
deleted file mode 100644
index 980bc6f4a7c40..0000000000000
--- a/api/core/model_runtime/model_providers/zhipuai/zhipuai_sdk/types/knowledge/document/document.py
+++ /dev/null
@@ -1,51 +0,0 @@
-from typing import Optional
-
-from ....core import BaseModel
-
-__all__ = ["DocumentData", "DocumentObject", "DocumentSuccessInfo", "DocumentFailedInfo"]
-
-
-class DocumentSuccessInfo(BaseModel):
-    documentId: Optional[str] = None
-    """文件id"""
-    filename: Optional[str] = None
-    """文件名称"""
-
-
-class DocumentFailedInfo(BaseModel):
-    failReason: Optional[str] = None
-    """上传失败的原因，包括：文件格式不支持、文件大小超出限制、知识库容量已满、容量上限为 50 万字。"""
-    filename: Optional[str] = None
-    """文件名称"""
-    documentId: Optional[str] = None
-    """知识库id"""
-
-
-class DocumentObject(BaseModel):
-    """文档信息"""
-
-    successInfos: Optional[list[DocumentSuccessInfo]] = None
-    """上传成功的文件信息"""
-    failedInfos: Optional[list[DocumentFailedInfo]] = None
-    """上传失败的文件信息"""
-
-
-class DocumentDataFailInfo(BaseModel):
-    """失败原因"""
-
-    embedding_code: Optional[int] = (
-        None  # 失败码 10001：知识不可用，知识库空间已达上限 10002：知识不可用，知识库空间已达上限(字数超出限制)
-    )
-    embedding_msg: Optional[str] = None  # 失败原因
-
-
-class DocumentData(BaseModel):
-    id: str = None  # 知识唯一id
-    custom_separator: list[str] = None  # 切片规则
-    sentence_size: str = None  # 切片大小
-    length: int = None  # 文件大小（字节）
-    word_num: int = None  # 文件字数
-    name: str = None  # 文件名
-    url: str = None  # 文件下载链接
-    embedding_stat: int = None  # 0:向量化中 1:向量化完成 2:向量化失败
-    failInfo: Optional[DocumentDataFailInfo] = None  # 失败原因 向量化失败embedding_stat=2的时候 会有此值
diff --git a/api/core/model_runtime/model_providers/zhipuai/zhipuai_sdk/types/knowledge/document/document_edit_params.py b/api/core/model_runtime/model_providers/zhipuai/zhipuai_sdk/types/knowledge/document/document_edit_params.py
deleted file mode 100644
index 509cb3a451af5..0000000000000
--- a/api/core/model_runtime/model_providers/zhipuai/zhipuai_sdk/types/knowledge/document/document_edit_params.py
+++ /dev/null
@@ -1,29 +0,0 @@
-from typing import Optional, TypedDict
-
-__all__ = ["DocumentEditParams"]
-
-
-class DocumentEditParams(TypedDict):
-    """
-    知识参数类型定义
-
-    Attributes:
-        id (str): 知识ID
-        knowledge_type (int): 知识类型:
-                        1:文章知识: 支持pdf,url,docx
-                        2.问答知识-文档:  支持pdf,url,docx
-                        3.问答知识-表格:  支持xlsx
-                        4.商品库-表格:  支持xlsx
-                        5.自定义:  支持pdf,url,docx
-        custom_separator (Optional[List[str]]): 当前知识类型为自定义(knowledge_type=5)时的切片规则，默认\n
-        sentence_size (Optional[int]): 当前知识类型为自定义(knowledge_type=5)时的切片字数，取值范围: 20-2000，默认300
-        callback_url (Optional[str]): 回调地址
-        callback_header (Optional[dict]): 回调时携带的header
-    """
-
-    id: str
-    knowledge_type: int
-    custom_separator: Optional[list[str]]
-    sentence_size: Optional[int]
-    callback_url: Optional[str]
-    callback_header: Optional[dict[str, str]]
diff --git a/api/core/model_runtime/model_providers/zhipuai/zhipuai_sdk/types/knowledge/document/document_list_params.py b/api/core/model_runtime/model_providers/zhipuai/zhipuai_sdk/types/knowledge/document/document_list_params.py
deleted file mode 100644
index 910c8c045e1b9..0000000000000
--- a/api/core/model_runtime/model_providers/zhipuai/zhipuai_sdk/types/knowledge/document/document_list_params.py
+++ /dev/null
@@ -1,26 +0,0 @@
-from __future__ import annotations
-
-from typing import Optional
-
-from typing_extensions import TypedDict
-
-
-class DocumentListParams(TypedDict, total=False):
-    """
-    文件查询参数类型定义
-
-    Attributes:
-        purpose (Optional[str]): 文件用途
-        knowledge_id (Optional[str]): 当文件用途为 retrieval 时，需要提供查询的知识库ID
-        page (Optional[int]): 页，默认1
-        limit (Optional[int]): 查询文件列表数，默认10
-        after (Optional[str]): 查询指定fileID之后的文件列表（当文件用途为 fine-tune 时需要）
-        order (Optional[str]): 排序规则，可选值['desc', 'asc']，默认desc（当文件用途为 fine-tune 时需要）
-    """
-
-    purpose: Optional[str]
-    knowledge_id: Optional[str]
-    page: Optional[int]
-    limit: Optional[int]
-    after: Optional[str]
-    order: Optional[str]
diff --git a/api/core/model_runtime/model_providers/zhipuai/zhipuai_sdk/types/knowledge/document/document_list_resp.py b/api/core/model_runtime/model_providers/zhipuai/zhipuai_sdk/types/knowledge/document/document_list_resp.py
deleted file mode 100644
index acae4fad9ff36..0000000000000
--- a/api/core/model_runtime/model_providers/zhipuai/zhipuai_sdk/types/knowledge/document/document_list_resp.py
+++ /dev/null
@@ -1,11 +0,0 @@
-from __future__ import annotations
-
-from ....core import BaseModel
-from . import DocumentData
-
-__all__ = ["DocumentPage"]
-
-
-class DocumentPage(BaseModel):
-    list: list[DocumentData]
-    object: str
diff --git a/api/core/model_runtime/model_providers/zhipuai/zhipuai_sdk/types/knowledge/knowledge.py b/api/core/model_runtime/model_providers/zhipuai/zhipuai_sdk/types/knowledge/knowledge.py
deleted file mode 100644
index bc6f159eb211e..0000000000000
--- a/api/core/model_runtime/model_providers/zhipuai/zhipuai_sdk/types/knowledge/knowledge.py
+++ /dev/null
@@ -1,21 +0,0 @@
-from typing import Optional
-
-from ...core import BaseModel
-
-__all__ = ["KnowledgeInfo"]
-
-
-class KnowledgeInfo(BaseModel):
-    id: Optional[str] = None
-    """知识库唯一 id"""
-    embedding_id: Optional[str] = (
-        None  # 知识库绑定的向量化模型 见模型列表 [内部服务开放接口文档](https://lslfd0slxc.feishu.cn/docx/YauWdbBiMopV0FxB7KncPWCEn8f#H15NduiQZo3ugmxnWQFcfAHpnQ4)
-    )
-    name: Optional[str] = None  # 知识库名称 100字限制
-    customer_identifier: Optional[str] = None  # 用户标识 长度32位以内
-    description: Optional[str] = None  # 知识库描述 500字限制
-    background: Optional[str] = None  # 背景颜色（给枚举）'blue', 'red', 'orange', 'purple', 'sky'
-    icon: Optional[str] = (
-        None  # 知识库图标（给枚举） question: 问号、book: 书籍、seal: 印章、wrench: 扳手、tag: 标签、horn: 喇叭、house: 房子  # noqa: E501
-    )
-    bucket_id: Optional[str] = None  # 桶id 限制32位
diff --git a/api/core/model_runtime/model_providers/zhipuai/zhipuai_sdk/types/knowledge/knowledge_create_params.py b/api/core/model_runtime/model_providers/zhipuai/zhipuai_sdk/types/knowledge/knowledge_create_params.py
deleted file mode 100644
index c3da201727c34..0000000000000
--- a/api/core/model_runtime/model_providers/zhipuai/zhipuai_sdk/types/knowledge/knowledge_create_params.py
+++ /dev/null
@@ -1,30 +0,0 @@
-from __future__ import annotations
-
-from typing import Literal, Optional
-
-from typing_extensions import TypedDict
-
-__all__ = ["KnowledgeBaseParams"]
-
-
-class KnowledgeBaseParams(TypedDict):
-    """
-    知识库参数类型定义
-
-    Attributes:
-        embedding_id (int): 知识库绑定的向量化模型ID
-        name (str): 知识库名称，限制100字
-        customer_identifier (Optional[str]): 用户标识，长度32位以内
-        description (Optional[str]): 知识库描述，限制500字
-        background (Optional[Literal['blue', 'red', 'orange', 'purple', 'sky']]): 背景颜色
-        icon (Optional[Literal['question', 'book', 'seal', 'wrench', 'tag', 'horn', 'house']]): 知识库图标
-        bucket_id (Optional[str]): 桶ID，限制32位
-    """
-
-    embedding_id: int
-    name: str
-    customer_identifier: Optional[str]
-    description: Optional[str]
-    background: Optional[Literal["blue", "red", "orange", "purple", "sky"]] = None
-    icon: Optional[Literal["question", "book", "seal", "wrench", "tag", "horn", "house"]] = None
-    bucket_id: Optional[str]
diff --git a/api/core/model_runtime/model_providers/zhipuai/zhipuai_sdk/types/knowledge/knowledge_list_params.py b/api/core/model_runtime/model_providers/zhipuai/zhipuai_sdk/types/knowledge/knowledge_list_params.py
deleted file mode 100644
index a221b28e4603b..0000000000000
--- a/api/core/model_runtime/model_providers/zhipuai/zhipuai_sdk/types/knowledge/knowledge_list_params.py
+++ /dev/null
@@ -1,15 +0,0 @@
-from __future__ import annotations
-
-from typing_extensions import TypedDict
-
-__all__ = ["KnowledgeListParams"]
-
-
-class KnowledgeListParams(TypedDict, total=False):
-    page: int = 1
-    """ 页码，默认 1，第一页
-    """
-
-    size: int = 10
-    """每页数量 默认10
-    """
diff --git a/api/core/model_runtime/model_providers/zhipuai/zhipuai_sdk/types/knowledge/knowledge_list_resp.py b/api/core/model_runtime/model_providers/zhipuai/zhipuai_sdk/types/knowledge/knowledge_list_resp.py
deleted file mode 100644
index e462eddc550d6..0000000000000
--- a/api/core/model_runtime/model_providers/zhipuai/zhipuai_sdk/types/knowledge/knowledge_list_resp.py
+++ /dev/null
@@ -1,11 +0,0 @@
-from __future__ import annotations
-
-from ...core import BaseModel
-from . import KnowledgeInfo
-
-__all__ = ["KnowledgePage"]
-
-
-class KnowledgePage(BaseModel):
-    list: list[KnowledgeInfo]
-    object: str
diff --git a/api/core/model_runtime/model_providers/zhipuai/zhipuai_sdk/types/knowledge/knowledge_used.py b/api/core/model_runtime/model_providers/zhipuai/zhipuai_sdk/types/knowledge/knowledge_used.py
deleted file mode 100644
index cfda7097026c5..0000000000000
--- a/api/core/model_runtime/model_providers/zhipuai/zhipuai_sdk/types/knowledge/knowledge_used.py
+++ /dev/null
@@ -1,21 +0,0 @@
-from typing import Optional
-
-from ...core import BaseModel
-
-__all__ = ["KnowledgeStatistics", "KnowledgeUsed"]
-
-
-class KnowledgeStatistics(BaseModel):
-    """
-    使用量统计
-    """
-
-    word_num: Optional[int] = None
-    length: Optional[int] = None
-
-
-class KnowledgeUsed(BaseModel):
-    used: Optional[KnowledgeStatistics] = None
-    """已使用量"""
-    total: Optional[KnowledgeStatistics] = None
-    """知识库总量"""
diff --git a/api/core/model_runtime/model_providers/zhipuai/zhipuai_sdk/types/sensitive_word_check/__init__.py b/api/core/model_runtime/model_providers/zhipuai/zhipuai_sdk/types/sensitive_word_check/__init__.py
deleted file mode 100644
index c9bd60419ce60..0000000000000
--- a/api/core/model_runtime/model_providers/zhipuai/zhipuai_sdk/types/sensitive_word_check/__init__.py
+++ /dev/null
@@ -1,3 +0,0 @@
-from .sensitive_word_check import SensitiveWordCheckRequest
-
-__all__ = ["SensitiveWordCheckRequest"]
diff --git a/api/core/model_runtime/model_providers/zhipuai/zhipuai_sdk/types/sensitive_word_check/sensitive_word_check.py b/api/core/model_runtime/model_providers/zhipuai/zhipuai_sdk/types/sensitive_word_check/sensitive_word_check.py
deleted file mode 100644
index 0c37d99e65329..0000000000000
--- a/api/core/model_runtime/model_providers/zhipuai/zhipuai_sdk/types/sensitive_word_check/sensitive_word_check.py
+++ /dev/null
@@ -1,14 +0,0 @@
-from typing import Optional
-
-from typing_extensions import TypedDict
-
-
-class SensitiveWordCheckRequest(TypedDict, total=False):
-    type: Optional[str]
-    """敏感词类型，当前仅支持ALL"""
-    status: Optional[str]
-    """敏感词启用禁用状态
-        启用：ENABLE
-        禁用：DISABLE
-       备注：默认开启敏感词校验，如果要关闭敏感词校验，需联系商务获取对应权限，否则敏感词禁用不生效。
-    """
diff --git a/api/core/model_runtime/model_providers/zhipuai/zhipuai_sdk/types/tools/__init__.py b/api/core/model_runtime/model_providers/zhipuai/zhipuai_sdk/types/tools/__init__.py
deleted file mode 100644
index 62f77344eee56..0000000000000
--- a/api/core/model_runtime/model_providers/zhipuai/zhipuai_sdk/types/tools/__init__.py
+++ /dev/null
@@ -1,9 +0,0 @@
-from .web_search import (
-    SearchIntent,
-    SearchRecommend,
-    SearchResult,
-    WebSearch,
-)
-from .web_search_chunk import WebSearchChunk
-
-__all__ = ["WebSearch", "SearchIntent", "SearchResult", "SearchRecommend", "WebSearchChunk"]
diff --git a/api/core/model_runtime/model_providers/zhipuai/zhipuai_sdk/types/tools/tools_web_search_params.py b/api/core/model_runtime/model_providers/zhipuai/zhipuai_sdk/types/tools/tools_web_search_params.py
deleted file mode 100644
index b3a3b26f07ee5..0000000000000
--- a/api/core/model_runtime/model_providers/zhipuai/zhipuai_sdk/types/tools/tools_web_search_params.py
+++ /dev/null
@@ -1,35 +0,0 @@
-from __future__ import annotations
-
-from typing import Optional, Union
-
-from typing_extensions import TypedDict
-
-__all__ = ["WebSearchParams"]
-
-
-class WebSearchParams(TypedDict):
-    """
-    工具名：web-search-pro参数类型定义
-
-    Attributes:
-        :param model: str, 模型名称
-        :param request_id: Optional[str], 请求ID
-        :param stream: Optional[bool], 是否流式
-        :param messages: Union[str, List[str], List[int], object, None],
-                        包含历史对话上下文的内容，按照 {"role": "user", "content": "你好"} 的json 数组形式进行传参
-                        当前版本仅支持 User Message 单轮对话，工具会理解User Message并进行搜索，
-                        请尽可能传入不带指令格式的用户原始提问，以提高搜索准确率。
-        :param scope: Optional[str], 指定搜索范围，全网、学术等，默认全网
-        :param location: Optional[str], 指定搜索用户地区 location 提高相关性
-        :param recent_days: Optional[int],支持指定返回 N 天（1-30）更新的搜索结果
-
-
-    """
-
-    model: str
-    request_id: Optional[str]
-    stream: Optional[bool]
-    messages: Union[str, list[str], list[int], object, None]
-    scope: Optional[str] = None
-    location: Optional[str] = None
-    recent_days: Optional[int] = None
diff --git a/api/core/model_runtime/model_providers/zhipuai/zhipuai_sdk/types/tools/web_search.py b/api/core/model_runtime/model_providers/zhipuai/zhipuai_sdk/types/tools/web_search.py
deleted file mode 100644
index ac9fa3821e979..0000000000000
--- a/api/core/model_runtime/model_providers/zhipuai/zhipuai_sdk/types/tools/web_search.py
+++ /dev/null
@@ -1,71 +0,0 @@
-from typing import Optional
-
-from ...core import BaseModel
-
-__all__ = [
-    "WebSearch",
-    "SearchIntent",
-    "SearchResult",
-    "SearchRecommend",
-]
-
-
-class SearchIntent(BaseModel):
-    index: int
-    # 搜索轮次，默认为 0
-    query: str
-    # 搜索优化 query
-    intent: str
-    # 判断的意图类型
-    keywords: str
-    # 搜索关键词
-
-
-class SearchResult(BaseModel):
-    index: int
-    # 搜索轮次，默认为 0
-    title: str
-    # 标题
-    link: str
-    # 链接
-    content: str
-    # 内容
-    icon: str
-    # 图标
-    media: str
-    # 来源媒体
-    refer: str
-    # 角标序号 [ref_1]
-
-
-class SearchRecommend(BaseModel):
-    index: int
-    # 搜索轮次，默认为 0
-    query: str
-    # 推荐query
-
-
-class WebSearchMessageToolCall(BaseModel):
-    id: str
-    search_intent: Optional[SearchIntent]
-    search_result: Optional[SearchResult]
-    search_recommend: Optional[SearchRecommend]
-    type: str
-
-
-class WebSearchMessage(BaseModel):
-    role: str
-    tool_calls: Optional[list[WebSearchMessageToolCall]] = None
-
-
-class WebSearchChoice(BaseModel):
-    index: int
-    finish_reason: str
-    message: WebSearchMessage
-
-
-class WebSearch(BaseModel):
-    created: Optional[int] = None
-    choices: list[WebSearchChoice]
-    request_id: Optional[str] = None
-    id: Optional[str] = None
diff --git a/api/core/model_runtime/model_providers/zhipuai/zhipuai_sdk/types/tools/web_search_chunk.py b/api/core/model_runtime/model_providers/zhipuai/zhipuai_sdk/types/tools/web_search_chunk.py
deleted file mode 100644
index 7fb0e02bb5871..0000000000000
--- a/api/core/model_runtime/model_providers/zhipuai/zhipuai_sdk/types/tools/web_search_chunk.py
+++ /dev/null
@@ -1,33 +0,0 @@
-from typing import Optional
-
-from ...core import BaseModel
-from .web_search import SearchIntent, SearchRecommend, SearchResult
-
-__all__ = ["WebSearchChunk"]
-
-
-class ChoiceDeltaToolCall(BaseModel):
-    index: int
-    id: Optional[str] = None
-
-    search_intent: Optional[SearchIntent] = None
-    search_result: Optional[SearchResult] = None
-    search_recommend: Optional[SearchRecommend] = None
-    type: Optional[str] = None
-
-
-class ChoiceDelta(BaseModel):
-    role: Optional[str] = None
-    tool_calls: Optional[list[ChoiceDeltaToolCall]] = None
-
-
-class Choice(BaseModel):
-    delta: ChoiceDelta
-    finish_reason: Optional[str] = None
-    index: int
-
-
-class WebSearchChunk(BaseModel):
-    id: Optional[str] = None
-    choices: list[Choice]
-    created: Optional[int] = None
diff --git a/api/core/model_runtime/model_providers/zhipuai/zhipuai_sdk/types/video/__init__.py b/api/core/model_runtime/model_providers/zhipuai/zhipuai_sdk/types/video/__init__.py
deleted file mode 100644
index b14072b1a771a..0000000000000
--- a/api/core/model_runtime/model_providers/zhipuai/zhipuai_sdk/types/video/__init__.py
+++ /dev/null
@@ -1,3 +0,0 @@
-from .video_object import VideoObject, VideoResult
-
-__all__ = ["VideoObject", "VideoResult"]
diff --git a/api/core/model_runtime/model_providers/zhipuai/zhipuai_sdk/types/video/video_create_params.py b/api/core/model_runtime/model_providers/zhipuai/zhipuai_sdk/types/video/video_create_params.py
deleted file mode 100644
index f5489d708e722..0000000000000
--- a/api/core/model_runtime/model_providers/zhipuai/zhipuai_sdk/types/video/video_create_params.py
+++ /dev/null
@@ -1,27 +0,0 @@
-from __future__ import annotations
-
-from typing import Optional
-
-from typing_extensions import TypedDict
-
-__all__ = ["VideoCreateParams"]
-
-from ..sensitive_word_check import SensitiveWordCheckRequest
-
-
-class VideoCreateParams(TypedDict, total=False):
-    model: str
-    """模型编码"""
-    prompt: str
-    """所需视频的文本描述"""
-    image_url: str
-    """所需视频的文本描述"""
-    sensitive_word_check: Optional[SensitiveWordCheckRequest]
-    """支持 URL 或者 Base64、传入 image 奖进行图生视频
-     * 图片格式：
-     *   图片大小："""
-    request_id: str
-    """由用户端传参，需保证唯一性；用于区分每次请求的唯一标识，用户端不传时平台会默认生成。"""
-
-    user_id: str
-    """用户端。"""
diff --git a/api/core/model_runtime/model_providers/zhipuai/zhipuai_sdk/types/video/video_object.py b/api/core/model_runtime/model_providers/zhipuai/zhipuai_sdk/types/video/video_object.py
deleted file mode 100644
index 85c3844d8a791..0000000000000
--- a/api/core/model_runtime/model_providers/zhipuai/zhipuai_sdk/types/video/video_object.py
+++ /dev/null
@@ -1,30 +0,0 @@
-from typing import Optional
-
-from ...core import BaseModel
-
-__all__ = ["VideoObject", "VideoResult"]
-
-
-class VideoResult(BaseModel):
-    url: str
-    """视频url"""
-    cover_image_url: str
-    """预览图"""
-
-
-class VideoObject(BaseModel):
-    id: Optional[str] = None
-    """智谱 AI 开放平台生成的任务订单号，调用请求结果接口时请使用此订单号"""
-
-    model: str
-    """模型名称"""
-
-    video_result: list[VideoResult]
-    """视频生成结果"""
-
-    task_status: str
-    """处理状态，PROCESSING（处理中），SUCCESS（成功），FAIL（失败）
-    注：处理中状态需通过查询获取结果"""
-
-    request_id: str
-    """用户在客户端请求时提交的任务编号或者平台生成的任务编号"""
diff --git a/api/core/tools/provider/builtin/cogview/tools/cogview3.py b/api/core/tools/provider/builtin/cogview/tools/cogview3.py
index 085084ca38355..12b4173fa4027 100644
--- a/api/core/tools/provider/builtin/cogview/tools/cogview3.py
+++ b/api/core/tools/provider/builtin/cogview/tools/cogview3.py
@@ -1,7 +1,8 @@
 import random
 from typing import Any, Union
 
-from core.model_runtime.model_providers.zhipuai.zhipuai_sdk._client import ZhipuAI
+from zhipuai import ZhipuAI
+
 from core.tools.entities.tool_entities import ToolInvokeMessage
 from core.tools.tool.builtin_tool import BuiltinTool
 
diff --git a/api/poetry.lock b/api/poetry.lock
index efefedfb21cd2..f1c5d949fee73 100644
--- a/api/poetry.lock
+++ b/api/poetry.lock
@@ -1995,17 +1995,6 @@ dev = ["Sphinx (==5.3.0)", "bump2version (==1.0.1)", "coverage (>=6.2)", "datacl
 timedelta = ["pytimeparse (>=1.1.7)"]
 yaml = ["PyYAML (>=5.3)"]
 
-[[package]]
-name = "dataclasses"
-version = "0.6"
-description = "A backport of the dataclasses module for Python 3.6"
-optional = false
-python-versions = "*"
-files = [
-    {file = "dataclasses-0.6-py3-none-any.whl", hash = "sha256:454a69d788c7fda44efd71e259be79577822f5e3f53f029a22d08004e951dc9f"},
-    {file = "dataclasses-0.6.tar.gz", hash = "sha256:6988bd2b895eef432d562370bb707d540f32f7360ab13da45340101bc2307d84"},
-]
-
 [[package]]
 name = "dataclasses-json"
 version = "0.6.7"
@@ -4965,7 +4954,6 @@ python-versions = ">=3.7"
 files = [
     {file = "milvus_lite-2.4.10-py3-none-macosx_10_9_x86_64.whl", hash = "sha256:fc4246d3ed7d1910847afce0c9ba18212e93a6e9b8406048436940578dfad5cb"},
     {file = "milvus_lite-2.4.10-py3-none-macosx_11_0_arm64.whl", hash = "sha256:74a8e07c5e3b057df17fbb46913388e84df1dc403a200f4e423799a58184c800"},
-    {file = "milvus_lite-2.4.10-py3-none-manylinux2014_aarch64.whl", hash = "sha256:240c7386b747bad696ecb5bd1f58d491e86b9d4b92dccee3315ed7256256eddc"},
     {file = "milvus_lite-2.4.10-py3-none-manylinux2014_x86_64.whl", hash = "sha256:211d2e334a043f9282bdd9755f76b9b2d93b23bffa7af240919ffce6a8dfe325"},
 ]
 
@@ -8484,11 +8472,6 @@ files = [
     {file = "scikit_learn-1.5.2-cp312-cp312-manylinux_2_17_aarch64.manylinux2014_aarch64.whl", hash = "sha256:f60021ec1574e56632be2a36b946f8143bf4e5e6af4a06d85281adc22938e0dd"},
     {file = "scikit_learn-1.5.2-cp312-cp312-manylinux_2_17_x86_64.manylinux2014_x86_64.whl", hash = "sha256:394397841449853c2290a32050382edaec3da89e35b3e03d6cc966aebc6a8ae6"},
     {file = "scikit_learn-1.5.2-cp312-cp312-win_amd64.whl", hash = "sha256:57cc1786cfd6bd118220a92ede80270132aa353647684efa385a74244a41e3b1"},
-    {file = "scikit_learn-1.5.2-cp313-cp313-macosx_10_13_x86_64.whl", hash = "sha256:e9a702e2de732bbb20d3bad29ebd77fc05a6b427dc49964300340e4c9328b3f5"},
-    {file = "scikit_learn-1.5.2-cp313-cp313-macosx_12_0_arm64.whl", hash = "sha256:b0768ad641981f5d3a198430a1d31c3e044ed2e8a6f22166b4d546a5116d7908"},
-    {file = "scikit_learn-1.5.2-cp313-cp313-manylinux_2_17_aarch64.manylinux2014_aarch64.whl", hash = "sha256:178ddd0a5cb0044464fc1bfc4cca5b1833bfc7bb022d70b05db8530da4bb3dd3"},
-    {file = "scikit_learn-1.5.2-cp313-cp313-manylinux_2_17_x86_64.manylinux2014_x86_64.whl", hash = "sha256:f7284ade780084d94505632241bf78c44ab3b6f1e8ccab3d2af58e0e950f9c12"},
-    {file = "scikit_learn-1.5.2-cp313-cp313-win_amd64.whl", hash = "sha256:b7b0f9a0b1040830d38c39b91b3a44e1b643f4b36e36567b80b7c6bd2202a27f"},
     {file = "scikit_learn-1.5.2-cp39-cp39-macosx_10_9_x86_64.whl", hash = "sha256:757c7d514ddb00ae249832fe87100d9c73c6ea91423802872d9e74970a0e40b9"},
     {file = "scikit_learn-1.5.2-cp39-cp39-macosx_12_0_arm64.whl", hash = "sha256:52788f48b5d8bca5c0736c175fa6bdaab2ef00a8f536cda698db61bd89c551c1"},
     {file = "scikit_learn-1.5.2-cp39-cp39-manylinux_2_17_aarch64.manylinux2014_aarch64.whl", hash = "sha256:643964678f4b5fbdc95cbf8aec638acc7aa70f5f79ee2cdad1eec3df4ba6ead8"},
@@ -10486,20 +10469,21 @@ repair = ["scipy (>=1.6.3)"]
 
 [[package]]
 name = "zhipuai"
-version = "1.0.7"
+version = "2.1.5.20230904"
 description = "A SDK library for accessing big model apis from ZhipuAI"
 optional = false
-python-versions = ">=3.6"
+python-versions = "!=2.7.*,!=3.0.*,!=3.1.*,!=3.2.*,!=3.3.*,!=3.4.*,!=3.5.*,!=3.6.*,!=3.7.*,>=3.8"
 files = [
-    {file = "zhipuai-1.0.7-py3-none-any.whl", hash = "sha256:360c01b8c2698f366061452e86d5a36a5ff68a576ea33940da98e4806f232530"},
-    {file = "zhipuai-1.0.7.tar.gz", hash = "sha256:b80f699543d83cce8648acf1ce32bc2725d1c1c443baffa5882abc2cc704d581"},
+    {file = "zhipuai-2.1.5.20230904-py3-none-any.whl", hash = "sha256:8485ca452c2f07fea476fb0666abc8fbbdf1b2e4feeee46a3bb3c1a2b51efccd"},
+    {file = "zhipuai-2.1.5.20230904.tar.gz", hash = "sha256:2c19dd796b12e2f19b93d8f9be6fd01e85d3320737a187ebf3c75a9806a7c2b5"},
 ]
 
 [package.dependencies]
-cachetools = "*"
-dataclasses = "*"
-PyJWT = "*"
-requests = "*"
+cachetools = ">=4.2.2"
+httpx = ">=0.23.0"
+pydantic = ">=1.9.0,<3.0"
+pydantic-core = ">=2.14.6"
+pyjwt = ">=2.8.0,<2.9.0"
 
 [[package]]
 name = "zipp"
@@ -10707,4 +10691,4 @@ cffi = ["cffi (>=1.11)"]
 [metadata]
 lock-version = "2.0"
 python-versions = ">=3.10,<3.13"
-content-hash = "edb5e3b0d50e84a239224cc77f3f615fdbdd6b504bce5b1075b29363f3054957"
+content-hash = "75a7e7eab36b9386c11a3e9808da28102ad20a43a0e8ae08c37594ecf50da02b"
diff --git a/api/pyproject.toml b/api/pyproject.toml
index dff74750f0f55..cc85ec3af6e9a 100644
--- a/api/pyproject.toml
+++ b/api/pyproject.toml
@@ -182,7 +182,7 @@ websocket-client = "~1.7.0"
 werkzeug = "~3.0.1"
 xinference-client = "0.15.2"
 yarl = "~1.9.4"
-zhipuai = "1.0.7"
+zhipuai = "~2.1.5"
 # Before adding new dependency, consider place it in alphabet order (a-z) and suitable group.
 
 ############################################################

From 7a405b86c92fdb70d7210234b42a61c7e3db6e0f Mon Sep 17 00:00:00 2001
From: zhuhao <37029601+hwzhuhao@users.noreply.github.com>
Date: Mon, 14 Oct 2024 13:26:21 +0800
Subject: [PATCH 230/235] refactor: Refactor the service of retrieval the
 recommend app (#9302)

---
 api/services/recommend_app/__init__.py        |   0
 .../recommend_app/buildin/__init__.py         |   0
 .../buildin/buildin_retrieval.py              |  64 +++++
 .../recommend_app/database/__init__.py        |   0
 .../database/database_retrieval.py            | 111 +++++++++
 .../recommend_app/recommend_app_base.py       |  17 ++
 .../recommend_app/recommend_app_factory.py    |  23 ++
 .../recommend_app/recommend_app_type.py       |   7 +
 api/services/recommend_app/remote/__init__.py |   0
 .../recommend_app/remote/remote_retrieval.py  |  71 ++++++
 api/services/recommended_app_service.py       | 235 +-----------------
 11 files changed, 302 insertions(+), 226 deletions(-)
 create mode 100644 api/services/recommend_app/__init__.py
 create mode 100644 api/services/recommend_app/buildin/__init__.py
 create mode 100644 api/services/recommend_app/buildin/buildin_retrieval.py
 create mode 100644 api/services/recommend_app/database/__init__.py
 create mode 100644 api/services/recommend_app/database/database_retrieval.py
 create mode 100644 api/services/recommend_app/recommend_app_base.py
 create mode 100644 api/services/recommend_app/recommend_app_factory.py
 create mode 100644 api/services/recommend_app/recommend_app_type.py
 create mode 100644 api/services/recommend_app/remote/__init__.py
 create mode 100644 api/services/recommend_app/remote/remote_retrieval.py

diff --git a/api/services/recommend_app/__init__.py b/api/services/recommend_app/__init__.py
new file mode 100644
index 0000000000000..e69de29bb2d1d
diff --git a/api/services/recommend_app/buildin/__init__.py b/api/services/recommend_app/buildin/__init__.py
new file mode 100644
index 0000000000000..e69de29bb2d1d
diff --git a/api/services/recommend_app/buildin/buildin_retrieval.py b/api/services/recommend_app/buildin/buildin_retrieval.py
new file mode 100644
index 0000000000000..4704d533a950e
--- /dev/null
+++ b/api/services/recommend_app/buildin/buildin_retrieval.py
@@ -0,0 +1,64 @@
+import json
+from os import path
+from pathlib import Path
+from typing import Optional
+
+from flask import current_app
+
+from services.recommend_app.recommend_app_base import RecommendAppRetrievalBase
+from services.recommend_app.recommend_app_type import RecommendAppType
+
+
+class BuildInRecommendAppRetrieval(RecommendAppRetrievalBase):
+    """
+    Retrieval recommended app from buildin, the location  is constants/recommended_apps.json
+    """
+
+    builtin_data: Optional[dict] = None
+
+    def get_type(self) -> str:
+        return RecommendAppType.BUILDIN
+
+    def get_recommended_apps_and_categories(self, language: str) -> dict:
+        result = self.fetch_recommended_apps_from_builtin(language)
+        return result
+
+    def get_recommend_app_detail(self, app_id: str):
+        result = self.fetch_recommended_app_detail_from_builtin(app_id)
+        return result
+
+    @classmethod
+    def _get_builtin_data(cls) -> dict:
+        """
+        Get builtin data.
+        :return:
+        """
+        if cls.builtin_data:
+            return cls.builtin_data
+
+        root_path = current_app.root_path
+        cls.builtin_data = json.loads(
+            Path(path.join(root_path, "constants", "recommended_apps.json")).read_text(encoding="utf-8")
+        )
+
+        return cls.builtin_data
+
+    @classmethod
+    def fetch_recommended_apps_from_builtin(cls, language: str) -> dict:
+        """
+        Fetch recommended apps from builtin.
+        :param language: language
+        :return:
+        """
+        builtin_data = cls._get_builtin_data()
+        return builtin_data.get("recommended_apps", {}).get(language)
+
+    @classmethod
+    def fetch_recommended_app_detail_from_builtin(cls, app_id: str) -> Optional[dict]:
+        """
+        Fetch recommended app detail from builtin.
+        :param app_id: App ID
+        :return:
+        """
+        builtin_data = cls._get_builtin_data()
+        return builtin_data.get("app_details", {}).get(app_id)
diff --git a/api/services/recommend_app/database/__init__.py b/api/services/recommend_app/database/__init__.py
new file mode 100644
index 0000000000000..e69de29bb2d1d
diff --git a/api/services/recommend_app/database/database_retrieval.py b/api/services/recommend_app/database/database_retrieval.py
new file mode 100644
index 0000000000000..995d3755bb5b1
--- /dev/null
+++ b/api/services/recommend_app/database/database_retrieval.py
@@ -0,0 +1,111 @@
+from typing import Optional
+
+from constants.languages import languages
+from extensions.ext_database import db
+from models.model import App, RecommendedApp
+from services.app_dsl_service import AppDslService
+from services.recommend_app.recommend_app_base import RecommendAppRetrievalBase
+from services.recommend_app.recommend_app_type import RecommendAppType
+
+
+class DatabaseRecommendAppRetrieval(RecommendAppRetrievalBase):
+    """
+    Retrieval recommended app from database
+    """
+
+    def get_recommended_apps_and_categories(self, language: str) -> dict:
+        result = self.fetch_recommended_apps_from_db(language)
+        return result
+
+    def get_recommend_app_detail(self, app_id: str):
+        result = self.fetch_recommended_app_detail_from_db(app_id)
+        return result
+
+    def get_type(self) -> str:
+        return RecommendAppType.DATABASE
+
+    @classmethod
+    def fetch_recommended_apps_from_db(cls, language: str) -> dict:
+        """
+        Fetch recommended apps from db.
+        :param language: language
+        :return:
+        """
+        recommended_apps = (
+            db.session.query(RecommendedApp)
+            .filter(RecommendedApp.is_listed == True, RecommendedApp.language == language)
+            .all()
+        )
+
+        if len(recommended_apps) == 0:
+            recommended_apps = (
+                db.session.query(RecommendedApp)
+                .filter(RecommendedApp.is_listed == True, RecommendedApp.language == languages[0])
+                .all()
+            )
+
+        categories = set()
+        recommended_apps_result = []
+        for recommended_app in recommended_apps:
+            app = recommended_app.app
+            if not app or not app.is_public:
+                continue
+
+            site = app.site
+            if not site:
+                continue
+
+            recommended_app_result = {
+                "id": recommended_app.id,
+                "app": {
+                    "id": app.id,
+                    "name": app.name,
+                    "mode": app.mode,
+                    "icon": app.icon,
+                    "icon_background": app.icon_background,
+                },
+                "app_id": recommended_app.app_id,
+                "description": site.description,
+                "copyright": site.copyright,
+                "privacy_policy": site.privacy_policy,
+                "custom_disclaimer": site.custom_disclaimer,
+                "category": recommended_app.category,
+                "position": recommended_app.position,
+                "is_listed": recommended_app.is_listed,
+            }
+            recommended_apps_result.append(recommended_app_result)
+
+            categories.add(recommended_app.category)
+
+        return {"recommended_apps": recommended_apps_result, "categories": sorted(categories)}
+
+    @classmethod
+    def fetch_recommended_app_detail_from_db(cls, app_id: str) -> Optional[dict]:
+        """
+        Fetch recommended app detail from db.
+        :param app_id: App ID
+        :return:
+        """
+        # is in public recommended list
+        recommended_app = (
+            db.session.query(RecommendedApp)
+            .filter(RecommendedApp.is_listed == True, RecommendedApp.app_id == app_id)
+            .first()
+        )
+
+        if not recommended_app:
+            return None
+
+        # get app detail
+        app_model = db.session.query(App).filter(App.id == app_id).first()
+        if not app_model or not app_model.is_public:
+            return None
+
+        return {
+            "id": app_model.id,
+            "name": app_model.name,
+            "icon": app_model.icon,
+            "icon_background": app_model.icon_background,
+            "mode": app_model.mode,
+            "export_data": AppDslService.export_dsl(app_model=app_model),
+        }
diff --git a/api/services/recommend_app/recommend_app_base.py b/api/services/recommend_app/recommend_app_base.py
new file mode 100644
index 0000000000000..00c037710e869
--- /dev/null
+++ b/api/services/recommend_app/recommend_app_base.py
@@ -0,0 +1,17 @@
+from abc import ABC, abstractmethod
+
+
+class RecommendAppRetrievalBase(ABC):
+    """Interface for recommend app retrieval."""
+
+    @abstractmethod
+    def get_recommended_apps_and_categories(self, language: str) -> dict:
+        raise NotImplementedError
+
+    @abstractmethod
+    def get_recommend_app_detail(self, app_id: str):
+        raise NotImplementedError
+
+    @abstractmethod
+    def get_type(self) -> str:
+        raise NotImplementedError
diff --git a/api/services/recommend_app/recommend_app_factory.py b/api/services/recommend_app/recommend_app_factory.py
new file mode 100644
index 0000000000000..e53667c0b06dd
--- /dev/null
+++ b/api/services/recommend_app/recommend_app_factory.py
@@ -0,0 +1,23 @@
+from services.recommend_app.buildin.buildin_retrieval import BuildInRecommendAppRetrieval
+from services.recommend_app.database.database_retrieval import DatabaseRecommendAppRetrieval
+from services.recommend_app.recommend_app_base import RecommendAppRetrievalBase
+from services.recommend_app.recommend_app_type import RecommendAppType
+from services.recommend_app.remote.remote_retrieval import RemoteRecommendAppRetrieval
+
+
+class RecommendAppRetrievalFactory:
+    @staticmethod
+    def get_recommend_app_factory(mode: str) -> type[RecommendAppRetrievalBase]:
+        match mode:
+            case RecommendAppType.REMOTE:
+                return RemoteRecommendAppRetrieval
+            case RecommendAppType.DATABASE:
+                return DatabaseRecommendAppRetrieval
+            case RecommendAppType.BUILDIN:
+                return BuildInRecommendAppRetrieval
+            case _:
+                raise ValueError(f"invalid fetch recommended apps mode: {mode}")
+
+    @staticmethod
+    def get_buildin_recommend_app_retrieval():
+        return BuildInRecommendAppRetrieval
diff --git a/api/services/recommend_app/recommend_app_type.py b/api/services/recommend_app/recommend_app_type.py
new file mode 100644
index 0000000000000..7ea93b3f64b1d
--- /dev/null
+++ b/api/services/recommend_app/recommend_app_type.py
@@ -0,0 +1,7 @@
+from enum import Enum
+
+
+class RecommendAppType(str, Enum):
+    REMOTE = "remote"
+    BUILDIN = "builtin"
+    DATABASE = "db"
diff --git a/api/services/recommend_app/remote/__init__.py b/api/services/recommend_app/remote/__init__.py
new file mode 100644
index 0000000000000..e69de29bb2d1d
diff --git a/api/services/recommend_app/remote/remote_retrieval.py b/api/services/recommend_app/remote/remote_retrieval.py
new file mode 100644
index 0000000000000..b0607a21323ac
--- /dev/null
+++ b/api/services/recommend_app/remote/remote_retrieval.py
@@ -0,0 +1,71 @@
+import logging
+from typing import Optional
+
+import requests
+
+from configs import dify_config
+from services.recommend_app.buildin.buildin_retrieval import BuildInRecommendAppRetrieval
+from services.recommend_app.recommend_app_base import RecommendAppRetrievalBase
+from services.recommend_app.recommend_app_type import RecommendAppType
+
+logger = logging.getLogger(__name__)
+
+
+class RemoteRecommendAppRetrieval(RecommendAppRetrievalBase):
+    """
+    Retrieval recommended app from dify official
+    """
+
+    def get_recommend_app_detail(self, app_id: str):
+        try:
+            result = self.fetch_recommended_app_detail_from_dify_official(app_id)
+        except Exception as e:
+            logger.warning(f"fetch recommended app detail from dify official failed: {e}, switch to built-in.")
+            result = BuildInRecommendAppRetrieval.fetch_recommended_app_detail_from_builtin(app_id)
+        return result
+
+    def get_recommended_apps_and_categories(self, language: str) -> dict:
+        try:
+            result = self.fetch_recommended_apps_from_dify_official(language)
+        except Exception as e:
+            logger.warning(f"fetch recommended apps from dify official failed: {e}, switch to built-in.")
+            result = BuildInRecommendAppRetrieval.fetch_recommended_apps_from_builtin(language)
+        return result
+
+    def get_type(self) -> str:
+        return RecommendAppType.REMOTE
+
+    @classmethod
+    def fetch_recommended_app_detail_from_dify_official(cls, app_id: str) -> Optional[dict]:
+        """
+        Fetch recommended app detail from dify official.
+        :param app_id: App ID
+        :return:
+        """
+        domain = dify_config.HOSTED_FETCH_APP_TEMPLATES_REMOTE_DOMAIN
+        url = f"{domain}/apps/{app_id}"
+        response = requests.get(url, timeout=(3, 10))
+        if response.status_code != 200:
+            return None
+
+        return response.json()
+
+    @classmethod
+    def fetch_recommended_apps_from_dify_official(cls, language: str) -> dict:
+        """
+        Fetch recommended apps from dify official.
+        :param language: language
+        :return:
+        """
+        domain = dify_config.HOSTED_FETCH_APP_TEMPLATES_REMOTE_DOMAIN
+        url = f"{domain}/apps?language={language}"
+        response = requests.get(url, timeout=(3, 10))
+        if response.status_code != 200:
+            raise ValueError(f"fetch recommended apps failed, status code: {response.status_code}")
+
+        result = response.json()
+
+        if "categories" in result:
+            result["categories"] = sorted(result["categories"])
+
+        return result
diff --git a/api/services/recommended_app_service.py b/api/services/recommended_app_service.py
index daec8393d092e..4660316fcfcf7 100644
--- a/api/services/recommended_app_service.py
+++ b/api/services/recommended_app_service.py
@@ -1,24 +1,10 @@
-import json
-import logging
-from os import path
-from pathlib import Path
 from typing import Optional
 
-import requests
-from flask import current_app
-
 from configs import dify_config
-from constants.languages import languages
-from extensions.ext_database import db
-from models.model import App, RecommendedApp
-from services.app_dsl_service import AppDslService
-
-logger = logging.getLogger(__name__)
+from services.recommend_app.recommend_app_factory import RecommendAppRetrievalFactory
 
 
 class RecommendedAppService:
-    builtin_data: Optional[dict] = None
-
     @classmethod
     def get_recommended_apps_and_categories(cls, language: str) -> dict:
         """
@@ -27,109 +13,17 @@ def get_recommended_apps_and_categories(cls, language: str) -> dict:
         :return:
         """
         mode = dify_config.HOSTED_FETCH_APP_TEMPLATES_MODE
-        if mode == "remote":
-            try:
-                result = cls._fetch_recommended_apps_from_dify_official(language)
-            except Exception as e:
-                logger.warning(f"fetch recommended apps from dify official failed: {e}, switch to built-in.")
-                result = cls._fetch_recommended_apps_from_builtin(language)
-        elif mode == "db":
-            result = cls._fetch_recommended_apps_from_db(language)
-        elif mode == "builtin":
-            result = cls._fetch_recommended_apps_from_builtin(language)
-        else:
-            raise ValueError(f"invalid fetch recommended apps mode: {mode}")
-
+        retrieval_instance = RecommendAppRetrievalFactory.get_recommend_app_factory(mode)()
+        result = retrieval_instance.get_recommended_apps_and_categories(language)
         if not result.get("recommended_apps") and language != "en-US":
-            result = cls._fetch_recommended_apps_from_builtin("en-US")
-
-        return result
-
-    @classmethod
-    def _fetch_recommended_apps_from_db(cls, language: str) -> dict:
-        """
-        Fetch recommended apps from db.
-        :param language: language
-        :return:
-        """
-        recommended_apps = (
-            db.session.query(RecommendedApp)
-            .filter(RecommendedApp.is_listed == True, RecommendedApp.language == language)
-            .all()
-        )
-
-        if len(recommended_apps) == 0:
-            recommended_apps = (
-                db.session.query(RecommendedApp)
-                .filter(RecommendedApp.is_listed == True, RecommendedApp.language == languages[0])
-                .all()
+            result = (
+                RecommendAppRetrievalFactory.get_buildin_recommend_app_retrieval().fetch_recommended_apps_from_builtin(
+                    "en-US"
+                )
             )
 
-        categories = set()
-        recommended_apps_result = []
-        for recommended_app in recommended_apps:
-            app = recommended_app.app
-            if not app or not app.is_public:
-                continue
-
-            site = app.site
-            if not site:
-                continue
-
-            recommended_app_result = {
-                "id": recommended_app.id,
-                "app": {
-                    "id": app.id,
-                    "name": app.name,
-                    "mode": app.mode,
-                    "icon": app.icon,
-                    "icon_background": app.icon_background,
-                },
-                "app_id": recommended_app.app_id,
-                "description": site.description,
-                "copyright": site.copyright,
-                "privacy_policy": site.privacy_policy,
-                "custom_disclaimer": site.custom_disclaimer,
-                "category": recommended_app.category,
-                "position": recommended_app.position,
-                "is_listed": recommended_app.is_listed,
-            }
-            recommended_apps_result.append(recommended_app_result)
-
-            categories.add(recommended_app.category)  # add category to categories
-
-        return {"recommended_apps": recommended_apps_result, "categories": sorted(categories)}
-
-    @classmethod
-    def _fetch_recommended_apps_from_dify_official(cls, language: str) -> dict:
-        """
-        Fetch recommended apps from dify official.
-        :param language: language
-        :return:
-        """
-        domain = dify_config.HOSTED_FETCH_APP_TEMPLATES_REMOTE_DOMAIN
-        url = f"{domain}/apps?language={language}"
-        response = requests.get(url, timeout=(3, 10))
-        if response.status_code != 200:
-            raise ValueError(f"fetch recommended apps failed, status code: {response.status_code}")
-
-        result = response.json()
-
-        if "categories" in result:
-            result["categories"] = sorted(result["categories"])
-
         return result
 
-    @classmethod
-    def _fetch_recommended_apps_from_builtin(cls, language: str) -> dict:
-        """
-        Fetch recommended apps from builtin.
-        :param language: language
-        :return:
-        """
-        builtin_data = cls._get_builtin_data()
-        return builtin_data.get("recommended_apps", {}).get(language)
-
     @classmethod
     def get_recommend_app_detail(cls, app_id: str) -> Optional[dict]:
         """
@@ -138,117 +32,6 @@ def get_recommend_app_detail(cls, app_id: str) -> Optional[dict]:
         :return:
         """
         mode = dify_config.HOSTED_FETCH_APP_TEMPLATES_MODE
-        if mode == "remote":
-            try:
-                result = cls._fetch_recommended_app_detail_from_dify_official(app_id)
-            except Exception as e:
-                logger.warning(f"fetch recommended app detail from dify official failed: {e}, switch to built-in.")
-                result = cls._fetch_recommended_app_detail_from_builtin(app_id)
-        elif mode == "db":
-            result = cls._fetch_recommended_app_detail_from_db(app_id)
-        elif mode == "builtin":
-            result = cls._fetch_recommended_app_detail_from_builtin(app_id)
-        else:
-            raise ValueError(f"invalid fetch recommended app detail mode: {mode}")
-
+        retrieval_instance = RecommendAppRetrievalFactory.get_recommend_app_factory(mode)()
+        result = retrieval_instance.get_recommend_app_detail(app_id)
         return result
-
-    @classmethod
-    def _fetch_recommended_app_detail_from_dify_official(cls, app_id: str) -> Optional[dict]:
-        """
-        Fetch recommended app detail from dify official.
-        :param app_id: App ID
-        :return:
-        """
-        domain = dify_config.HOSTED_FETCH_APP_TEMPLATES_REMOTE_DOMAIN
-        url = f"{domain}/apps/{app_id}"
-        response = requests.get(url, timeout=(3, 10))
-        if response.status_code != 200:
-            return None
-
-        return response.json()
-
-    @classmethod
-    def _fetch_recommended_app_detail_from_db(cls, app_id: str) -> Optional[dict]:
-        """
-        Fetch recommended app detail from db.
-        :param app_id: App ID
-        :return:
-        """
-        # is in public recommended list
-        recommended_app = (
-            db.session.query(RecommendedApp)
-            .filter(RecommendedApp.is_listed == True, RecommendedApp.app_id == app_id)
-            .first()
-        )
-
-        if not recommended_app:
-            return None
-
-        # get app detail
-        app_model = db.session.query(App).filter(App.id == app_id).first()
-        if not app_model or not app_model.is_public:
-            return None
-
-        return {
-            "id": app_model.id,
-            "name": app_model.name,
-            "icon": app_model.icon,
-            "icon_background": app_model.icon_background,
-            "mode": app_model.mode,
-            "export_data": AppDslService.export_dsl(app_model=app_model),
-        }
-
-    @classmethod
-    def _fetch_recommended_app_detail_from_builtin(cls, app_id: str) -> Optional[dict]:
-        """
-        Fetch recommended app detail from builtin.
-        :param app_id: App ID
-        :return:
-        """
-        builtin_data = cls._get_builtin_data()
-        return builtin_data.get("app_details", {}).get(app_id)
-
-    @classmethod
-    def _get_builtin_data(cls) -> dict:
-        """
-        Get builtin data.
-        :return:
-        """
-        if cls.builtin_data:
-            return cls.builtin_data
-
-        root_path = current_app.root_path
-        cls.builtin_data = json.loads(
-            Path(path.join(root_path, "constants", "recommended_apps.json")).read_text(encoding="utf-8")
-        )
-
-        return cls.builtin_data
-
-    @classmethod
-    def fetch_all_recommended_apps_and_export_datas(cls):
-        """
-        Fetch all recommended apps and export datas
-        :return:
-        """
-        templates = {"recommended_apps": {}, "app_details": {}}
-        for language in languages:
-            try:
-                result = cls._fetch_recommended_apps_from_dify_official(language)
-            except Exception as e:
-                logger.warning(f"fetch recommended apps from dify official failed: {e}, skip.")
-                continue
-
-            templates["recommended_apps"][language] = result
-
-            for recommended_app in result.get("recommended_apps"):
-                app_id = recommended_app.get("app_id")
-
-                # get app detail
-                app_detail = cls._fetch_recommended_app_detail_from_dify_official(app_id)
-                if not app_detail:
-                    continue
-
-                templates["app_details"][app_id] = app_detail
-
-        return templates

From 5ee7e03c1b163b8c4bbdcc051fdc2c426810e421 Mon Sep 17 00:00:00 2001
From: =?UTF-8?q?=E6=88=91=E6=9C=89=E4=B8=80=E6=8A=8A=E5=A6=96=E5=88=80?=
 <bling_faker@163.com>
Date: Mon, 14 Oct 2024 13:32:13 +0800
Subject: [PATCH 231/235] chore: Optimize operations in Q&A mode (#9274)

Co-authored-by: billsyli <billsyli@tencent.com>
---
 web/app/components/base/popover/index.tsx     |  3 ++
 .../datasets/create/step-two/index.tsx        | 44 ++++++++++++-------
 .../create/step-two/language-select/index.tsx |  3 ++
 3 files changed, 33 insertions(+), 17 deletions(-)

diff --git a/web/app/components/base/popover/index.tsx b/web/app/components/base/popover/index.tsx
index 141ac8ff70d62..1e7ba76269d75 100644
--- a/web/app/components/base/popover/index.tsx
+++ b/web/app/components/base/popover/index.tsx
@@ -17,6 +17,7 @@ type IPopover = {
   btnElement?: string | React.ReactNode
   btnClassName?: string | ((open: boolean) => string)
   manualClose?: boolean
+  disabled?: boolean
 }
 
 const timeoutDuration = 100
@@ -30,6 +31,7 @@ export default function CustomPopover({
   className,
   btnClassName,
   manualClose,
+  disabled = false,
 }: IPopover) {
   const buttonRef = useRef<HTMLButtonElement>(null)
   const timeOutRef = useRef<NodeJS.Timeout | null>(null)
@@ -60,6 +62,7 @@ export default function CustomPopover({
             >
               <Popover.Button
                 ref={buttonRef}
+                disabled={disabled}
                 className={`group ${s.popupBtn} ${open ? '' : 'bg-gray-100'} ${!btnClassName
                   ? ''
                   : typeof btnClassName === 'string'
diff --git a/web/app/components/datasets/create/step-two/index.tsx b/web/app/components/datasets/create/step-two/index.tsx
index ee01652de8dd2..c4c80053b06e5 100644
--- a/web/app/components/datasets/create/step-two/index.tsx
+++ b/web/app/components/datasets/create/step-two/index.tsx
@@ -132,6 +132,7 @@ const StepTwo = ({
       ? IndexingType.QUALIFIED
       : IndexingType.ECONOMICAL,
   )
+  const [isLanguageSelectDisabled, setIsLanguageSelectDisabled] = useState(false)
   const [docForm, setDocForm] = useState<DocForm | string>(
     (datasetId && documentDetail) ? documentDetail.doc_form : DocForm.TEXT,
   )
@@ -200,9 +201,9 @@ const StepTwo = ({
     }
   }
 
-  const fetchFileIndexingEstimate = async (docForm = DocForm.TEXT) => {
+  const fetchFileIndexingEstimate = async (docForm = DocForm.TEXT, language?: string) => {
     // eslint-disable-next-line @typescript-eslint/no-use-before-define
-    const res = await didFetchFileIndexingEstimate(getFileIndexingEstimateParams(docForm)!)
+    const res = await didFetchFileIndexingEstimate(getFileIndexingEstimateParams(docForm, language)!)
     if (segmentationType === SegmentType.CUSTOM)
       setCustomFileIndexingEstimate(res)
     else
@@ -270,7 +271,7 @@ const StepTwo = ({
     }
   }
 
-  const getFileIndexingEstimateParams = (docForm: DocForm): IndexingEstimateParams | undefined => {
+  const getFileIndexingEstimateParams = (docForm: DocForm, language?: string): IndexingEstimateParams | undefined => {
     if (dataSourceType === DataSourceType.FILE) {
       return {
         info_list: {
@@ -282,7 +283,7 @@ const StepTwo = ({
         indexing_technique: getIndexing_technique() as string,
         process_rule: getProcessRule(),
         doc_form: docForm,
-        doc_language: docLanguage,
+        doc_language: language || docLanguage,
         dataset_id: datasetId as string,
       }
     }
@@ -295,7 +296,7 @@ const StepTwo = ({
         indexing_technique: getIndexing_technique() as string,
         process_rule: getProcessRule(),
         doc_form: docForm,
-        doc_language: docLanguage,
+        doc_language: language || docLanguage,
         dataset_id: datasetId as string,
       }
     }
@@ -308,7 +309,7 @@ const StepTwo = ({
         indexing_technique: getIndexing_technique() as string,
         process_rule: getProcessRule(),
         doc_form: docForm,
-        doc_language: docLanguage,
+        doc_language: language || docLanguage,
         dataset_id: datasetId as string,
       }
     }
@@ -483,8 +484,26 @@ const StepTwo = ({
       setDocForm(DocForm.TEXT)
   }
 
+  const previewSwitch = async (language?: string) => {
+    setPreviewSwitched(true)
+    setIsLanguageSelectDisabled(true)
+    if (segmentationType === SegmentType.AUTO)
+      setAutomaticFileIndexingEstimate(null)
+    else
+      setCustomFileIndexingEstimate(null)
+    try {
+      await fetchFileIndexingEstimate(DocForm.QA, language)
+    }
+    finally {
+      setIsLanguageSelectDisabled(false)
+    }
+  }
+
   const handleSelect = (language: string) => {
     setDocLanguage(language)
+    // Switch language, re-cutter
+    if (docForm === DocForm.QA && previewSwitched)
+      previewSwitch(language)
   }
 
   const changeToEconomicalType = () => {
@@ -494,15 +513,6 @@ const StepTwo = ({
     }
   }
 
-  const previewSwitch = async () => {
-    setPreviewSwitched(true)
-    if (segmentationType === SegmentType.AUTO)
-      setAutomaticFileIndexingEstimate(null)
-    else
-      setCustomFileIndexingEstimate(null)
-    await fetchFileIndexingEstimate(DocForm.QA)
-  }
-
   useEffect(() => {
     // fetch rules
     if (!isSetting) {
@@ -777,7 +787,7 @@ const StepTwo = ({
                     <div className='mb-[2px] text-md font-medium text-gray-900'>{t('datasetCreation.stepTwo.QATitle')}</div>
                     <div className='inline-flex items-center text-[13px] leading-[18px] text-gray-500'>
                       <span className='pr-1'>{t('datasetCreation.stepTwo.QALanguage')}</span>
-                      <LanguageSelect currentLanguage={docLanguage} onSelect={handleSelect} />
+                      <LanguageSelect currentLanguage={docLanguage} onSelect={handleSelect} disabled={isLanguageSelectDisabled} />
                     </div>
                   </div>
                   <div className='shrink-0'>
@@ -948,7 +958,7 @@ const StepTwo = ({
               <div className='grow flex items-center'>
                 <div>{t('datasetCreation.stepTwo.previewTitle')}</div>
                 {docForm === DocForm.QA && !previewSwitched && (
-                  <Button className='ml-2' variant='secondary-accent' onClick={previewSwitch}>{t('datasetCreation.stepTwo.previewButton')}</Button>
+                  <Button className='ml-2' variant='secondary-accent' onClick={() => previewSwitch()}>{t('datasetCreation.stepTwo.previewButton')}</Button>
                 )}
               </div>
               <div className='flex items-center justify-center w-6 h-6 cursor-pointer' onClick={hidePreview}>
diff --git a/web/app/components/datasets/create/step-two/language-select/index.tsx b/web/app/components/datasets/create/step-two/language-select/index.tsx
index f8709c89f3a6b..fab2bb1c71389 100644
--- a/web/app/components/datasets/create/step-two/language-select/index.tsx
+++ b/web/app/components/datasets/create/step-two/language-select/index.tsx
@@ -9,16 +9,19 @@ import { languages } from '@/i18n/language'
 export type ILanguageSelectProps = {
   currentLanguage: string
   onSelect: (language: string) => void
+  disabled?: boolean
 }
 
 const LanguageSelect: FC<ILanguageSelectProps> = ({
   currentLanguage,
   onSelect,
+  disabled,
 }) => {
   return (
     <Popover
       manualClose
       trigger='click'
+      disabled={disabled}
       htmlContent={
         <div className='w-full py-1'>
           {languages.filter(language => language.supported).map(({ prompt_name, name }) => (

From de3c5751db8bca0c0795abbc7f6bddfb31d7779d Mon Sep 17 00:00:00 2001
From: =?UTF-8?q?=E6=88=91=E6=9C=89=E4=B8=80=E6=8A=8A=E5=A6=96=E5=88=80?=
 <bling_faker@163.com>
Date: Mon, 14 Oct 2024 13:32:52 +0800
Subject: [PATCH 232/235] chore: add reopen preview btn (#9279)

Co-authored-by: billsyli <billsyli@tencent.com>
---
 web/app/components/datasets/create/step-two/index.tsx | 2 +-
 1 file changed, 1 insertion(+), 1 deletion(-)

diff --git a/web/app/components/datasets/create/step-two/index.tsx b/web/app/components/datasets/create/step-two/index.tsx
index c4c80053b06e5..5d92e30deb8cc 100644
--- a/web/app/components/datasets/create/step-two/index.tsx
+++ b/web/app/components/datasets/create/step-two/index.tsx
@@ -585,7 +585,7 @@ const StepTwo = ({
       <div ref={scrollRef} className='relative h-full w-full overflow-y-scroll'>
         <div className={cn(s.pageHeader, scrolled && s.fixed, isMobile && '!px-6')}>
           <span>{t('datasetCreation.steps.two')}</span>
-          {isMobile && (
+          {(isMobile || !showPreview) && (
             <Button
               className='border-[0.5px] !h-8 hover:outline hover:outline-[0.5px] hover:outline-gray-300 text-gray-700 font-medium bg-white shadow-[0px_1px_2px_0px_rgba(16,24,40,0.05)]'
               onClick={setShowPreview}

From 7838f9f3a3f4566ec6240dc175938bfc9044edd0 Mon Sep 17 00:00:00 2001
From: -LAN- <laipz8200@outlook.com>
Date: Mon, 14 Oct 2024 16:27:26 +0800
Subject: [PATCH 233/235] fix: Add new Milvus Lite wheel for
 manylinux2014_aarch64 (#9316)

---
 api/poetry.lock | 6 ++++++
 1 file changed, 6 insertions(+)

diff --git a/api/poetry.lock b/api/poetry.lock
index f1c5d949fee73..289b905356098 100644
--- a/api/poetry.lock
+++ b/api/poetry.lock
@@ -4954,6 +4954,7 @@ python-versions = ">=3.7"
 files = [
     {file = "milvus_lite-2.4.10-py3-none-macosx_10_9_x86_64.whl", hash = "sha256:fc4246d3ed7d1910847afce0c9ba18212e93a6e9b8406048436940578dfad5cb"},
     {file = "milvus_lite-2.4.10-py3-none-macosx_11_0_arm64.whl", hash = "sha256:74a8e07c5e3b057df17fbb46913388e84df1dc403a200f4e423799a58184c800"},
+    {file = "milvus_lite-2.4.10-py3-none-manylinux2014_aarch64.whl", hash = "sha256:240c7386b747bad696ecb5bd1f58d491e86b9d4b92dccee3315ed7256256eddc"},
     {file = "milvus_lite-2.4.10-py3-none-manylinux2014_x86_64.whl", hash = "sha256:211d2e334a043f9282bdd9755f76b9b2d93b23bffa7af240919ffce6a8dfe325"},
 ]
 
@@ -8472,6 +8473,11 @@ files = [
     {file = "scikit_learn-1.5.2-cp312-cp312-manylinux_2_17_aarch64.manylinux2014_aarch64.whl", hash = "sha256:f60021ec1574e56632be2a36b946f8143bf4e5e6af4a06d85281adc22938e0dd"},
     {file = "scikit_learn-1.5.2-cp312-cp312-manylinux_2_17_x86_64.manylinux2014_x86_64.whl", hash = "sha256:394397841449853c2290a32050382edaec3da89e35b3e03d6cc966aebc6a8ae6"},
     {file = "scikit_learn-1.5.2-cp312-cp312-win_amd64.whl", hash = "sha256:57cc1786cfd6bd118220a92ede80270132aa353647684efa385a74244a41e3b1"},
+    {file = "scikit_learn-1.5.2-cp313-cp313-macosx_10_13_x86_64.whl", hash = "sha256:e9a702e2de732bbb20d3bad29ebd77fc05a6b427dc49964300340e4c9328b3f5"},
+    {file = "scikit_learn-1.5.2-cp313-cp313-macosx_12_0_arm64.whl", hash = "sha256:b0768ad641981f5d3a198430a1d31c3e044ed2e8a6f22166b4d546a5116d7908"},
+    {file = "scikit_learn-1.5.2-cp313-cp313-manylinux_2_17_aarch64.manylinux2014_aarch64.whl", hash = "sha256:178ddd0a5cb0044464fc1bfc4cca5b1833bfc7bb022d70b05db8530da4bb3dd3"},
+    {file = "scikit_learn-1.5.2-cp313-cp313-manylinux_2_17_x86_64.manylinux2014_x86_64.whl", hash = "sha256:f7284ade780084d94505632241bf78c44ab3b6f1e8ccab3d2af58e0e950f9c12"},
+    {file = "scikit_learn-1.5.2-cp313-cp313-win_amd64.whl", hash = "sha256:b7b0f9a0b1040830d38c39b91b3a44e1b643f4b36e36567b80b7c6bd2202a27f"},
     {file = "scikit_learn-1.5.2-cp39-cp39-macosx_10_9_x86_64.whl", hash = "sha256:757c7d514ddb00ae249832fe87100d9c73c6ea91423802872d9e74970a0e40b9"},
     {file = "scikit_learn-1.5.2-cp39-cp39-macosx_12_0_arm64.whl", hash = "sha256:52788f48b5d8bca5c0736c175fa6bdaab2ef00a8f536cda698db61bd89c551c1"},
     {file = "scikit_learn-1.5.2-cp39-cp39-manylinux_2_17_aarch64.manylinux2014_aarch64.whl", hash = "sha256:643964678f4b5fbdc95cbf8aec638acc7aa70f5f79ee2cdad1eec3df4ba6ead8"},

From 3938d8863ebb59ee0aaf2da72661adab1747a7b1 Mon Sep 17 00:00:00 2001
From: Wu Tianwei <30284043+WTW0313@users.noreply.github.com>
Date: Mon, 14 Oct 2024 16:54:23 +0800
Subject: [PATCH 234/235] fix: token refreshing logic issue (#9308)

---
 web/app/components/swr-initor.tsx |  6 +++---
 web/app/signin/normalForm.tsx     |  2 +-
 web/app/signin/userSSOForm.tsx    |  2 +-
 web/hooks/use-refresh-token.ts    | 33 ++++++++++++++-----------------
 4 files changed, 20 insertions(+), 23 deletions(-)

diff --git a/web/app/components/swr-initor.tsx b/web/app/components/swr-initor.tsx
index 85e05499e6d1e..ce126512fa977 100644
--- a/web/app/components/swr-initor.tsx
+++ b/web/app/components/swr-initor.tsx
@@ -14,12 +14,12 @@ const SwrInitor = ({
 }: SwrInitorProps) => {
   const router = useRouter()
   const searchParams = useSearchParams()
+  const { getNewAccessToken } = useRefreshToken()
   const consoleToken = searchParams.get('access_token')
   const refreshToken = searchParams.get('refresh_token')
   const consoleTokenFromLocalStorage = localStorage?.getItem('console_token')
   const refreshTokenFromLocalStorage = localStorage?.getItem('refresh_token')
   const [init, setInit] = useState(false)
-  const { getNewAccessToken } = useRefreshToken()
 
   useEffect(() => {
     if (!(consoleToken || refreshToken || consoleTokenFromLocalStorage || refreshTokenFromLocalStorage)) {
@@ -27,12 +27,12 @@ const SwrInitor = ({
       return
     }
     if (consoleTokenFromLocalStorage && refreshTokenFromLocalStorage)
-      getNewAccessToken(consoleTokenFromLocalStorage, refreshTokenFromLocalStorage)
+      getNewAccessToken()
 
     if (consoleToken && refreshToken) {
       localStorage.setItem('console_token', consoleToken)
       localStorage.setItem('refresh_token', refreshToken)
-      getNewAccessToken(consoleToken, refreshToken).then(() => {
+      getNewAccessToken().then(() => {
         router.replace('/apps', { forceOptimisticNavigation: false } as any)
       }).catch(() => {
         router.replace('/signin')
diff --git a/web/app/signin/normalForm.tsx b/web/app/signin/normalForm.tsx
index 0ae4eb1f43210..113ed64b57b9e 100644
--- a/web/app/signin/normalForm.tsx
+++ b/web/app/signin/normalForm.tsx
@@ -99,7 +99,7 @@ const NormalForm = () => {
       if (res.result === 'success') {
         localStorage.setItem('console_token', res.data.access_token)
         localStorage.setItem('refresh_token', res.data.refresh_token)
-        getNewAccessToken(res.data.access_token, res.data.refresh_token)
+        getNewAccessToken()
         router.replace('/apps')
       }
       else {
diff --git a/web/app/signin/userSSOForm.tsx b/web/app/signin/userSSOForm.tsx
index e4b61413bc0ef..f01afa9eaf274 100644
--- a/web/app/signin/userSSOForm.tsx
+++ b/web/app/signin/userSSOForm.tsx
@@ -31,7 +31,7 @@ const UserSSOForm: FC<UserSSOFormProps> = ({
     if (refreshToken && consoleToken) {
       localStorage.setItem('console_token', consoleToken)
       localStorage.setItem('refresh_token', refreshToken)
-      getNewAccessToken(consoleToken, refreshToken)
+      getNewAccessToken()
       router.replace('/apps')
     }
 
diff --git a/web/hooks/use-refresh-token.ts b/web/hooks/use-refresh-token.ts
index 3d8779636f4b4..293f3159de9ad 100644
--- a/web/hooks/use-refresh-token.ts
+++ b/web/hooks/use-refresh-token.ts
@@ -14,7 +14,6 @@ const useRefreshToken = () => {
   const router = useRouter()
   const timer = useRef<NodeJS.Timeout>()
   const advanceTime = useRef<number>(5 * 60 * 1000)
-  const interval = useRef<number>(55 * 60 * 1000)
 
   const getExpireTime = useCallback((token: string) => {
     if (!token)
@@ -31,18 +30,24 @@ const useRefreshToken = () => {
     localStorage?.removeItem('is_refreshing')
     localStorage?.removeItem('console_token')
     localStorage?.removeItem('refresh_token')
-    localStorage?.removeItem('last_refresh_time')
     router.replace('/signin')
   }, [])
 
-  const getNewAccessToken = useCallback(async (currentAccessToken: string, currentRefreshToken: string) => {
-    if (localStorage?.getItem('is_refreshing') === '1')
+  const getNewAccessToken = useCallback(async () => {
+    const currentAccessToken = localStorage?.getItem('console_token')
+    const currentRefreshToken = localStorage?.getItem('refresh_token')
+    if (!currentAccessToken || !currentRefreshToken) {
+      handleError()
+      return new Error('No access token or refresh token found')
+    }
+    if (localStorage?.getItem('is_refreshing') === '1') {
+      timer.current = setTimeout(() => {
+        getNewAccessToken()
+      }, 1000)
       return null
+    }
     const currentTokenExpireTime = getExpireTime(currentAccessToken)
-    let lastRefreshTime = parseInt(localStorage?.getItem('last_refresh_time') || '0')
-    lastRefreshTime = isNaN(lastRefreshTime) ? 0 : lastRefreshTime
-    if (getCurrentTimeStamp() + advanceTime.current > currentTokenExpireTime
-      && lastRefreshTime + interval.current < getCurrentTimeStamp()) {
+    if (getCurrentTimeStamp() + advanceTime.current > currentTokenExpireTime) {
       localStorage?.setItem('is_refreshing', '1')
       const [e, res] = await fetchWithRetry(fetchNewToken({
         body: { refresh_token: currentRefreshToken },
@@ -53,24 +58,17 @@ const useRefreshToken = () => {
       }
       const { access_token, refresh_token } = res.data
       localStorage?.setItem('is_refreshing', '0')
-      localStorage?.setItem('last_refresh_time', getCurrentTimeStamp().toString())
       localStorage?.setItem('console_token', access_token)
       localStorage?.setItem('refresh_token', refresh_token)
       const newTokenExpireTime = getExpireTime(access_token)
       timer.current = setTimeout(() => {
-        const consoleTokenFromLocalStorage = localStorage?.getItem('console_token')
-        const refreshTokenFromLocalStorage = localStorage?.getItem('refresh_token')
-        if (consoleTokenFromLocalStorage && refreshTokenFromLocalStorage)
-          getNewAccessToken(consoleTokenFromLocalStorage, refreshTokenFromLocalStorage)
+        getNewAccessToken()
       }, newTokenExpireTime - advanceTime.current - getCurrentTimeStamp())
     }
     else {
       const newTokenExpireTime = getExpireTime(currentAccessToken)
       timer.current = setTimeout(() => {
-        const consoleTokenFromLocalStorage = localStorage?.getItem('console_token')
-        const refreshTokenFromLocalStorage = localStorage?.getItem('refresh_token')
-        if (consoleTokenFromLocalStorage && refreshTokenFromLocalStorage)
-          getNewAccessToken(consoleTokenFromLocalStorage, refreshTokenFromLocalStorage)
+        getNewAccessToken()
       }, newTokenExpireTime - advanceTime.current - getCurrentTimeStamp())
     }
     return null
@@ -80,7 +78,6 @@ const useRefreshToken = () => {
     return () => {
       clearTimeout(timer.current)
       localStorage?.removeItem('is_refreshing')
-      localStorage?.removeItem('last_refresh_time')
     }
   }, [])
 

From 5c7b1358d49c8df54bc39eea7b9c4d9b41c09fcb Mon Sep 17 00:00:00 2001
From: -LAN- <laipz8200@outlook.com>
Date: Mon, 14 Oct 2024 17:21:58 +0800
Subject: [PATCH 235/235] chore(release): bump version to 0.9.2 (#9314)

---
 api/configs/packaging/__init__.py | 2 +-
 docker-legacy/docker-compose.yaml | 6 +++---
 docker/docker-compose.yaml        | 6 +++---
 web/package.json                  | 2 +-
 4 files changed, 8 insertions(+), 8 deletions(-)

diff --git a/api/configs/packaging/__init__.py b/api/configs/packaging/__init__.py
index c311a989b418c..c832fb671e5af 100644
--- a/api/configs/packaging/__init__.py
+++ b/api/configs/packaging/__init__.py
@@ -9,7 +9,7 @@ class PackagingInfo(BaseSettings):
 
     CURRENT_VERSION: str = Field(
         description="Dify version",
-        default="0.9.1",
+        default="0.9.2",
     )
 
     COMMIT_SHA: str = Field(
diff --git a/docker-legacy/docker-compose.yaml b/docker-legacy/docker-compose.yaml
index 3f230b47ab9f1..273f95b4284dd 100644
--- a/docker-legacy/docker-compose.yaml
+++ b/docker-legacy/docker-compose.yaml
@@ -2,7 +2,7 @@ version: '3'
 services:
   # API service
   api:
-    image: langgenius/dify-api:0.9.1
+    image: langgenius/dify-api:0.9.2
     restart: always
     environment:
       # Startup mode, 'api' starts the API server.
@@ -227,7 +227,7 @@ services:
   # worker service
   # The Celery worker for processing the queue.
   worker:
-    image: langgenius/dify-api:0.9.1
+    image: langgenius/dify-api:0.9.2
     restart: always
     environment:
       CONSOLE_WEB_URL: ''
@@ -396,7 +396,7 @@ services:
 
   # Frontend web application.
   web:
-    image: langgenius/dify-web:0.9.1
+    image: langgenius/dify-web:0.9.2
     restart: always
     environment:
       # The base URL of console application api server, refers to the Console base URL of WEB service if console domain is
diff --git a/docker/docker-compose.yaml b/docker/docker-compose.yaml
index 5db11d1961ce0..6c3331aadc5f0 100644
--- a/docker/docker-compose.yaml
+++ b/docker/docker-compose.yaml
@@ -227,7 +227,7 @@ x-shared-env: &shared-api-worker-env
 services:
   # API service
   api:
-    image: langgenius/dify-api:0.9.1
+    image: langgenius/dify-api:0.9.2
     restart: always
     environment:
       # Use the shared environment variables.
@@ -247,7 +247,7 @@ services:
   # worker service
   # The Celery worker for processing the queue.
   worker:
-    image: langgenius/dify-api:0.9.1
+    image: langgenius/dify-api:0.9.2
     restart: always
     environment:
       # Use the shared environment variables.
@@ -266,7 +266,7 @@ services:
 
   # Frontend web application.
   web:
-    image: langgenius/dify-web:0.9.1
+    image: langgenius/dify-web:0.9.2
     restart: always
     environment:
       CONSOLE_API_URL: ${CONSOLE_API_URL:-}
diff --git a/web/package.json b/web/package.json
index 8a17997fe963c..9707b9753feaf 100644
--- a/web/package.json
+++ b/web/package.json
@@ -1,6 +1,6 @@
 {
   "name": "dify-web",
-  "version": "0.9.1",
+  "version": "0.9.2",
   "private": true,
   "engines": {
     "node": ">=18.17.0"