From cd70092b8e6b7d9064e9893b10bfa7506269eb76 Mon Sep 17 00:00:00 2001
From: Santeri Hurnanen <santeri.hurnanen@fame.fi>
Date: Mon, 4 Sep 2023 10:55:25 +0300
Subject: [PATCH 1/3] UHF-8837: remove pages with X-Robots-Tag: noindex from
 sitemap

---
 helfi_proxy.module | 36 ++++++++++++++++++++++++++++++++++++
 1 file changed, 36 insertions(+)

diff --git a/helfi_proxy.module b/helfi_proxy.module
index fa91826..1c0364c 100644
--- a/helfi_proxy.module
+++ b/helfi_proxy.module
@@ -7,6 +7,7 @@
 
 use Drupal\Core\Asset\AttachedAssetsInterface;
 use Drupal\Core\Entity\EntityInterface;
+use Drupal\helfi_proxy\ProxyManagerInterface;
 
 /**
  * Implements hook_module_implements_alter().
@@ -82,3 +83,38 @@ function helfi_proxy_page_attachments_alter(array &$attachments) {
     $attachments['#attached']['html_head'][] = [$helfi_content_type, $tag_name];
   }
 }
+
+/**
+ * Implements hook_simple_sitemap_links_alter().
+ */
+function helfi_proxy_simple_sitemap_links_alter(array &$links, $sitemap_variant) {
+  /** @var \Drupal\Core\Config\ImmutableConfig $config */
+  $config = \Drupal::service('config.factory')->get('helfi_proxy.settings');
+
+  if (!$paths = implode("\n", $config->get(ProxyManagerInterface::ROBOTS_PATHS) ?? [])) {
+    return;
+  }
+
+  /** @var \Drupal\helfi_api_base\Environment\Environment $environment */
+  $environment = \Drupal::service('helfi_api_base.environment_resolver')->getActiveEnvironment();
+
+  /** @var \Drupal\Core\Path\PathMatcherInterface $pathMatcher */
+  $pathMatcher = \Drupal::service('path.matcher');
+
+  // helfi_proxy module sets "X-Robots-Tag: noindex" header for configured
+  // paths. These url should not be included in the sitemap.xml file.
+  foreach ($links as $key => $link) {
+    $baseUrl = $environment->getUrl($link['langcode']);
+    $url = $link['url'];
+
+    if (str_starts_with($url, $baseUrl)) {
+      $path = substr($url, strlen($baseUrl));
+
+      // Remove matched paths from sitemap.xml file.
+      if ($pathMatcher->matchPath($path, $paths)) {
+        unset($links[$key]);
+      }
+    }
+  }
+
+}

From 9ba62298bcfb025f2b84dd9fb337cbddd719795c Mon Sep 17 00:00:00 2001
From: Santeri Hurnanen <santeri.hurnanen@fame.fi>
Date: Mon, 4 Sep 2023 12:59:18 +0300
Subject: [PATCH 2/3] UHF-8837: handle environment_resolver exceptions

---
 helfi_proxy.module | 16 +++++++++++++---
 1 file changed, 13 insertions(+), 3 deletions(-)

diff --git a/helfi_proxy.module b/helfi_proxy.module
index 1c0364c..d9c6bcf 100644
--- a/helfi_proxy.module
+++ b/helfi_proxy.module
@@ -95,8 +95,12 @@ function helfi_proxy_simple_sitemap_links_alter(array &$links, $sitemap_variant)
     return;
   }
 
-  /** @var \Drupal\helfi_api_base\Environment\Environment $environment */
-  $environment = \Drupal::service('helfi_api_base.environment_resolver')->getActiveEnvironment();
+  try {
+    /** @var \Drupal\helfi_api_base\Environment\Environment $environment */
+    $environment = \Drupal::service('helfi_api_base.environment_resolver')->getActiveEnvironment();
+  } catch (\InvalidArgumentException) {
+    return;
+  }
 
   /** @var \Drupal\Core\Path\PathMatcherInterface $pathMatcher */
   $pathMatcher = \Drupal::service('path.matcher');
@@ -104,7 +108,13 @@ function helfi_proxy_simple_sitemap_links_alter(array &$links, $sitemap_variant)
   // helfi_proxy module sets "X-Robots-Tag: noindex" header for configured
   // paths. These url should not be included in the sitemap.xml file.
   foreach ($links as $key => $link) {
-    $baseUrl = $environment->getUrl($link['langcode']);
+    try {
+      $baseUrl = $environment->getUrl($link['langcode']);
+    } catch (\InvalidArgumentException) {
+      // Base url not found for given langcode.
+      continue;
+    }
+
     $url = $link['url'];
 
     if (str_starts_with($url, $baseUrl)) {

From dc19d76cad07575b4b89b58cdd976a02c5595fe2 Mon Sep 17 00:00:00 2001
From: Santeri Hurnanen <santeri.hurnanen@fame.fi>
Date: Mon, 4 Sep 2023 13:47:16 +0300
Subject: [PATCH 3/3] UHF-8837: phpcs fixes

---
 helfi_proxy.module | 6 ++++--
 1 file changed, 4 insertions(+), 2 deletions(-)

diff --git a/helfi_proxy.module b/helfi_proxy.module
index d9c6bcf..3233c01 100644
--- a/helfi_proxy.module
+++ b/helfi_proxy.module
@@ -98,7 +98,8 @@ function helfi_proxy_simple_sitemap_links_alter(array &$links, $sitemap_variant)
   try {
     /** @var \Drupal\helfi_api_base\Environment\Environment $environment */
     $environment = \Drupal::service('helfi_api_base.environment_resolver')->getActiveEnvironment();
-  } catch (\InvalidArgumentException) {
+  }
+  catch (\InvalidArgumentException) {
     return;
   }
 
@@ -110,7 +111,8 @@ function helfi_proxy_simple_sitemap_links_alter(array &$links, $sitemap_variant)
   foreach ($links as $key => $link) {
     try {
       $baseUrl = $environment->getUrl($link['langcode']);
-    } catch (\InvalidArgumentException) {
+    }
+    catch (\InvalidArgumentException) {
       // Base url not found for given langcode.
       continue;
     }