feat: streamlit page transcription

CHANGELOG.md

@@ -15,6 +15,7 @@ Tous les changements notables de l'application sont documentés dans ce fichier.
 ## [Alpha] - 2024-11-05
 
 - 🎉 Ajout de l'endpoint POST `/audio/transcriptions` pour la transcription d'audio
+- 🎉 Ajout d'une page de transcription d'audio dans l'UI
 
 ## [Alpha] - 2024-10-23
 

CONTRIBUTING.md

@@ -2,9 +2,11 @@
 
 Pour contribuer au projet, merci de suivre les instructions suivantes.
 
+> ⚠️ **Attention** : Vous devez disposer d'une API de modèle de language et d'embeddings pour lancer l'API en local.
+
 # Commit 
 
-Merci de respecter la convention suivante pour vos commits : 
+Merci de respecter la convention suivante pour vos commits :
 
 ```
 [doc|feat|fix](*) commit object (in english)
@@ -17,25 +19,41 @@ feat(collections): collection name retriever
 
 # Packages
 
-1. Installez [libmagic](https://man7.org/linux/man-pages/man3/libmagic.3.html)
-
-2. Dans un environnement virtuel Python, installez les packages Python présents dans le fichier *[pyproject.toml](./pyproject.toml)*
+1. Dans un environnement virtuel Python, installez les packages Python présents dans le fichier *[pyproject.toml](./pyproject.toml)*
 
   ```bash 
   pip install ".[ui,app,dev,test]"
   pre-commit install
   ```
 
-# Tests
+# Lancement des services
 
-Merci, avant chaque pull request, de vérifier le bon déploiement de votre API en exécutant des tests unitaires.
+Pour plus d'information sur le déploiement des services, veuillez consulter la [documentation dédiée](./docs/deployment.md).
+
+## API (FastAPI)
 
 1. Après avoir créé un fichier *config.yml*, lancez l'API en local
 
     ```bash
     uvicorn app.main:app --port 8080 --log-level debug --reload
     ```
 
+## User interface (Streamlit)
+
+1. Lancez l'API en local (voir la section[Lancement de l'API](#lancement-de-l-api))
+
+2. Lancez l'UI en local
+
+    ```bash
+    python -m streamlit run ui/chat.py --server.port 8501 --browser.gatherUsageStats false --theme.base light
+    ```
+
+# Tests
+
+Merci, avant chaque pull request, de vérifier le bon déploiement de votre API en exécutant des tests unitaires.
+
+1. Lancez l'API en local (voir la section[Lancement de l'API](#lancement-de-l-api))
+
 2. Exécutez les tests unitaires
     
     ```bash

README.md

@@ -2,8 +2,7 @@
 <summary><h1>Albert API</h1></summary>
 
 ![](https://img.shields.io/badge/version-alpha-yellow) ![](https://img.shields.io/badge/Python-3.12-green) ![](https://img.shields.io/badge/vLLM-v0.6.3.post1-blue) ![](https://img.shields.io/badge/HuggingFace%20Text%20Embeddings%20Inference-1.5-red)<br>
-<a href="https://albert.api.etalab.gouv.fr/documentation"><b>Documentation</b></a> | <a href="https://github.com/etalab-ia/albert-api/blob/main/CHANGELOG.md"><b>Changelog</b></a> | <a href="https://huggingface.co/AgentPublic"><b>HuggingFace</b></a>
-| <a href="https://albert.api.etalab.gouv.fr/swagger"><b>Swagger</b></a> <br><br>
+<a href="https://github.com/etalab-ia/albert-api/blob/main/CHANGELOG.md"><b>Changelog</b></a> | <a href="https://albert.api.etalab.gouv.fr/documentation"><b>Documentation</b></a> | <a href="https://albert.api.etalab.gouv.fr/status"><b>Status</b></a> | <a href="https://albert.api.etalab.gouv.fr/swagger"><b>Swagger</b></a> <br><br>
 </ul></div>
 
 Albert API est une initiative d'[Etalab](https://www.etalab.gouv.fr/). Il s'agit d'une API open source d'IA générative développée par Etalab. Elle permet d'être un proxy entre des modèles de langage et vos données. Elle agrège les services suivants :

app/endpoints/audio.py

@@ -1,4 +1,4 @@
-from typing import List
+from typing import List, Literal
 
 from fastapi import APIRouter, Form, Security, Request, UploadFile, File
 
@@ -8,9 +8,11 @@
 from app.utils.security import check_api_key, check_rate_limit, User
 from app.utils.lifespan import clients, limiter
 from app.utils.exceptions import ModelNotFoundException
+from app.utils.variables import SUPPORTED_LANGUAGES
 
 
 router = APIRouter()
+SUPPORTED_LANGUAGES_VALUES = sorted(set(SUPPORTED_LANGUAGES.values())) + sorted(set(SUPPORTED_LANGUAGES.keys()))
 
 
 @router.post("/audio/transcriptions")
@@ -19,7 +21,7 @@ async def audio_transcriptions(
     request: Request,
     file: UploadFile = File(...),
     model: str = Form(...),
-    language: str = Form(None),
+    language: Literal[*SUPPORTED_LANGUAGES_VALUES] = Form("fr"),
     prompt: str = Form(None),
     response_format: str = Form("json"),
     temperature: float = Form(0),

app/utils/variables.py

@@ -10,7 +10,121 @@
 JSON_TYPE = "application/json"
 TXT_TYPE = "text/plain"
 HTML_TYPE = "text/html"
+# @TODO : add DOCX_TYPE (application/vnd.openxmlformats-officedocument.wordprocessingml.document)
 ROLE_LEVEL_0 = 0
 ROLE_LEVEL_1 = 1
 ROLE_LEVEL_2 = 2
-# @TODO : add DOCX_TYPE (application/vnd.openxmlformats-officedocument.wordprocessingml.document)
+SUPPORTED_LANGUAGES = {
+    "afrikaans": "af",
+    "albanian": "sq",
+    "amharic": "am",
+    "arabic": "ar",
+    "armenian": "hy",
+    "assamese": "as",
+    "azerbaijani": "az",
+    "bashkir": "ba",
+    "basque": "eu",
+    "belarusian": "be",
+    "bengali": "bn",
+    "bosnian": "bs",
+    "breton": "br",
+    "bulgarian": "bg",
+    "burmese": "my",
+    "cantonese": "yue",
+    "castilian": "es",
+    "catalan": "ca",
+    "chinese": "zh",
+    "croatian": "hr",
+    "czech": "cs",
+    "danish": "da",
+    "dutch": "nl",
+    "english": "en",
+    "estonian": "et",
+    "faroese": "fo",
+    "finnish": "fi",
+    "flemish": "nl",
+    "french": "fr",
+    "galician": "gl",
+    "georgian": "ka",
+    "german": "de",
+    "greek": "el",
+    "gujarati": "gu",
+    "haitian": "ht",
+    "haitian creole": "ht",
+    "hausa": "ha",
+    "hawaiian": "haw",
+    "hebrew": "he",
+    "hindi": "hi",
+    "hungarian": "hu",
+    "icelandic": "is",
+    "indonesian": "id",
+    "italian": "it",
+    "japanese": "ja",
+    "javanese": "jw",
+    "kannada": "kn",
+    "kazakh": "kk",
+    "khmer": "km",
+    "korean": "ko",
+    "lao": "lo",
+    "latin": "la",
+    "latvian": "lv",
+    "letzeburgesch": "lb",
+    "lingala": "ln",
+    "lithuanian": "lt",
+    "luxembourgish": "lb",
+    "macedonian": "mk",
+    "malagasy": "mg",
+    "malay": "ms",
+    "malayalam": "ml",
+    "maltese": "mt",
+    "mandarin": "zh",
+    "maori": "mi",
+    "marathi": "mr",
+    "moldavian": "ro",
+    "moldovan": "ro",
+    "mongolian": "mn",
+    "myanmar": "my",
+    "nepali": "ne",
+    "norwegian": "no",
+    "nynorsk": "nn",
+    "occitan": "oc",
+    "panjabi": "pa",
+    "pashto": "ps",
+    "persian": "fa",
+    "polish": "pl",
+    "portuguese": "pt",
+    "punjabi": "pa",
+    "pushto": "ps",
+    "romanian": "ro",
+    "russian": "ru",
+    "sanskrit": "sa",
+    "serbian": "sr",
+    "shona": "sn",
+    "sindhi": "sd",
+    "sinhala": "si",
+    "sinhalese": "si",
+    "slovak": "sk",
+    "slovenian": "sl",
+    "somali": "so",
+    "spanish": "es",
+    "sundanese": "su",
+    "swahili": "sw",
+    "swedish": "sv",
+    "tagalog": "tl",
+    "tajik": "tg",
+    "tamil": "ta",
+    "tatar": "tt",
+    "telugu": "te",
+    "thai": "th",
+    "tibetan": "bo",
+    "turkish": "tr",
+    "turkmen": "tk",
+    "ukrainian": "uk",
+    "urdu": "ur",
+    "uzbek": "uz",
+    "valencian": "ca",
+    "vietnamese": "vi",
+    "welsh": "cy",
+    "yiddish": "yi",
+    "yoruba": "yo",
+}

pyproject.toml

@@ -6,12 +6,13 @@ requires-python = ">=3.12"
 license = { text = "MIT" }
 dependencies = [
     "openai==1.43.0",
+    "requests==2.32.3",
 ]
 
 [project.optional-dependencies]
 ui = [
-    "requests==2.32.3",
-    "streamlit==1.38.0",
+    "streamlit==1.39.0",
+    "streamlit-extras==0.5.0", 
 ]
 app = [
     "langchain==0.2.15",

ui/chat.py

@@ -4,6 +4,7 @@
 from openai import OpenAI
 import requests
 import streamlit as st
+from streamlit_extras.stylable_container import stylable_container
 
 from config import BASE_URL
 from utils import get_collections, get_models, header, set_config
@@ -14,7 +15,7 @@
 
 # Data
 try:
-    language_models, embeddings_models = get_models(api_key=API_KEY)
+    language_models, embeddings_models, _ = get_models(api_key=API_KEY)
     collections = get_collections(api_key=API_KEY)
 except Exception:
     st.error("Error to fetch user data.")
@@ -25,11 +26,10 @@
 
 # Sidebar
 with st.sidebar:
-    st.title("Model parameters")
     params = {"sampling_params": dict(), "rag": dict()}
-
+    st.title("Chat parameters")
     params["sampling_params"]["model"] = st.selectbox("Language model", language_models)
-    params["sampling_params"]["temperature"] = st.number_input("Temperature", value=0.1)
+    params["sampling_params"]["temperature"] = st.slider("Temperature", value=0.2, min_value=0.0, max_value=1.0, step=0.1)
     params["sampling_params"]["max_tokens"] = st.number_input("Max tokens (optional)", value=400)
 
     st.title("RAG parameters")
@@ -43,18 +43,27 @@
         params["rag"]["k"] = st.number_input("Top K", value=3)
 
 # Main
-col1, col2 = st.columns([0.85, 0.15])
-with col1:
-    new_chat = st.button("New chat")
-with col2:
-    if model_collections:
-        rag = st.toggle("Activated RAG", value=False, disabled=not bool(params["rag"]["collections"]))
-    else:
-        rag = st.toggle("Activated RAG", value=False, disabled=True)
-if new_chat:
-    st.session_state.pop("messages", None)
-    st.session_state.pop("sources", None)
-    st.rerun()
+with stylable_container(
+    key="Chat",
+    css_styles="""
+    button{
+        float: right;
+    }
+    """,
+):
+    col1, col2 = st.columns(2)
+    with col2:
+        new_chat = st.button("New chat")
+    with col1:
+        if model_collections:
+            rag = st.toggle("Activated RAG", value=True, disabled=not bool(params["rag"]["collections"]))
+        else:
+            rag = st.toggle("Activated RAG", value=False, disabled=True)
+
+    if new_chat:
+        st.session_state.pop("messages", None)
+        st.session_state.pop("sources", None)
+        st.rerun()
 
 if "messages" not in st.session_state:
     st.session_state.messages = []