Bugfix - MAG Video Indexer API & Tweak to Speech Service Defaults (#344)

Xeelee33 · paullizer · web-flow · commit cf24854da732 · 2025-09-03T14:28:14.000-04:00
* Delete logfile.log

* Removed external

these are pending release and will show up in next version.

* Update route_external_health.py

* Adding health check

* Added health check feature

* updating app.py, removed external

* Updated to v0.215.36

* Update app.py

* v0.215.37

* v0.215.37

* updated so video indexer api default changes based on public/government cloud and is not read-only in the admin settings, removed hardcoded commercial speech_service_endpoint and speech_service_location

* added placeholder for speech_service_endpoint to show default endpoint format based on location and cloud type

* added reminders to save pending changes before fetching GPT models in admin settings

* Added process to delete videos from VI service when deleted from front end/blob/cosmos. Includes adding the file's video indexer ID to the file metadata in cosmos.

* Added unmerged files from Development branch

---------

Co-authored-by: Paul Lizer &lt;paullizer@microsoft.com&gt;
Co-authored-by: Paul Lizer &lt;paul.lizer@live.com&gt;
diff --git a/application/single_app/config.py b/application/single_app/config.py
@@ -133,6 +133,9 @@
 else:
     AUTHORITY = f"https://login.microsoftonline.us/{TENANT_ID}"
 
+# Commercial Azure Video Indexer Endpoint
+video_indexer_endpoint = "https://api.videoindexer.ai"
+
 WORD_CHUNK_SIZE = 400
 
 if AZURE_ENVIRONMENT == "usgovernment":
@@ -141,7 +144,9 @@
     authority = AzureAuthorityHosts.AZURE_GOVERNMENT
     credential_scopes=[resource_manager + "/.default"]
     cognitive_services_scope = "https://cognitiveservices.azure.us/.default"
+    video_indexer_endpoint = "https://api.videoindexer.ai.azure.us"
     search_resource_manager = "https://search.azure.us"
+
 elif AZURE_ENVIRONMENT == "custom":
     resource_manager = CUSTOM_RESOURCE_MANAGER_URL_VALUE
     authority = CUSTOM_IDENTITY_URL_VALUE
diff --git a/application/single_app/functions_documents.py b/application/single_app/functions_documents.py
@@ -359,6 +359,17 @@ def to_seconds(ts: str) -> float:
             raise ValueError("no video ID returned")
         print(f"[VIDEO] UPLOAD OK, videoId={vid}", flush=True)
         update_callback(status=f"VIDEO: uploaded id={vid}")
+        try:
+            # Update the document's metadata with the video indexer ID
+            update_document(
+                document_id=document_id,
+                user_id=user_id,
+                group_id=group_id,
+                video_indexer_id=vid
+            )
+        except Exception as e:
+            print(f"[VIDEO] Failed to update document metadata with video_indexer_id: {e}", flush=True)
+
     except Exception as e:
         print(f"[VIDEO] UPLOAD ERROR: {e}", flush=True)
         update_callback(status=f"VIDEO: upload failed → {e}")
@@ -1444,8 +1455,29 @@ def delete_document(user_id, document_id, group_id=None, public_workspace_id=Non
             
         # Get the file name from the document to use for blob deletion
         file_name = document_item.get('file_name')
-        
-        # First try to delete from blob storage
+        file_ext = os.path.splitext(file_name)[1].lower() if file_name else None
+
+        # First try to delete video from Video Indexer if applicable
+        if file_ext in ('.mp4', '.mov', '.avi', '.mkv', '.flv'):
+            try:
+                settings = get_settings()
+                vi_ep = settings.get("video_indexer_endpoint")
+                vi_loc = settings.get("video_indexer_location")
+                vi_acc = settings.get("video_indexer_account_id")
+                token = get_video_indexer_account_token(settings)
+                # You need to store the video ID in the document metadata when uploading
+                video_id = document_item.get("video_indexer_id")
+                if video_id:
+                    delete_url = f"{vi_ep}/{vi_loc}/Accounts/{vi_acc}/Videos/{video_id}?accessToken={token}"
+                    resp = requests.delete(delete_url, timeout=60)
+                    resp.raise_for_status()
+                    print(f"Deleted video from Video Indexer: {video_id}")
+                else:
+                    print("No video_indexer_id found in document metadata; skipping Video Indexer deletion.")
+            except Exception as e:
+                print(f"Error deleting video from Video Indexer: {e}")
+
+        # Second try to delete from blob storage
         try:
             if file_name:
                 delete_from_blob_storage(document_id, user_id, file_name, group_id, public_workspace_id)
diff --git a/application/single_app/functions_settings.py b/application/single_app/functions_settings.py
@@ -294,7 +294,7 @@ def get_settings():
         'enable_external_healthcheck': False,
 
         # Video file settings with Azure Video Indexer Settings
-        'video_indexer_endpoint': 'https://api.videoindexer.ai',
+        'video_indexer_endpoint': video_indexer_endpoint,
         'video_indexer_location': '',
         'video_indexer_account_id': '',
         'video_indexer_api_key': '',
@@ -305,8 +305,8 @@ def get_settings():
         'video_index_timeout': 600,
 
         # Audio file settings with Azure speech service
-        "speech_service_endpoint": "https://eastus.api.cognitive.microsoft.com",
-        "speech_service_location": "eastus",
+        "speech_service_endpoint": '',
+        "speech_service_location": '',
         "speech_service_locale": "en-US",
         "speech_service_key": ""
     }
diff --git a/application/single_app/route_frontend_admin_settings.py b/application/single_app/route_frontend_admin_settings.py
@@ -570,7 +570,7 @@ def is_valid_url(url):
                 'default_system_prompt': form_data.get('default_system_prompt', '').strip(),
 
                 # Video file settings with Azure Video Indexer Settings
-                'video_indexer_endpoint': form_data.get('video_indexer_endpoint', 'https://api.videoindexer.ai').strip(),
+                'video_indexer_endpoint': form_data.get('video_indexer_endpoint', video_indexer_endpoint).strip(),
                 'video_indexer_location': form_data.get('video_indexer_location', '').strip(),
                 'video_indexer_account_id': form_data.get('video_indexer_account_id', '').strip(),
                 'video_indexer_api_key': form_data.get('video_indexer_api_key', '').strip(),
diff --git a/application/single_app/templates/admin_settings.html b/application/single_app/templates/admin_settings.html
@@ -1012,6 +1012,7 @@ <h6>External Links</h6>
                         <div class="mb-3">
                             <label for="gpt_model" class="form-label">Azure OpenAI GPT Model Selection</label>
                             <br><small>Each selected model will be available in the Chat UI as an option for the User. You can select multiple models.</small>
+                            <br><small>Save Pending Changes to settings before clicking Fetch Models</small>
                             <div id="gpt_models_list" class="mt-2 mb-3">
                                 
                             </div>
@@ -1153,6 +1154,7 @@ <h6>External Links</h6>
                             <div id="embedding_models_list" class="mt-2 mb-3">
                                 
                             </div>
+                            <small>Save pending changes to settings before clicking Fetch Embedding Models</small>
                             
                             <input type="hidden" name="embedding_model_json" id="embedding_model_json" />
 
@@ -1599,14 +1601,14 @@ <h5>Video Indexer Settings</h5>
 
                         <div class="mb-3">
                             <label for="video_indexer_endpoint" class="form-label">Endpoint</label>
-                            <input type="text" readonly class="form-control"
+                            <input type="text" class="form-control"
                                 id="video_indexer_endpoint" name="video_indexer_endpoint"
                                 value="{{ settings.video_indexer_endpoint or 'https://api.videoindexer.ai' }}">
                         </div>
 
                         <div class="mb-3">
                             <label for="video_indexer_arm_api_version" class="form-label">ARM API Version</label>
-                            <input type="text" readonly class="form-control"
+                            <input type="text" class="form-control"
                                 id="video_indexer_arm_api_version" name="video_indexer_arm_api_version"
                                 value="{{ settings.video_indexer_arm_api_version or '2021-11-10-preview' }}">
                         </div>
@@ -1687,7 +1689,8 @@ <h5>Speech Service Settings</h5>
                             <label for="speech_service_endpoint" class="form-label">Endpoint</label>
                             <input type="text" class="form-control"
                                 id="speech_service_endpoint" name="speech_service_endpoint"
-                                value="{{ settings.speech_service_endpoint or '' }}">
+                                value="{{ settings.speech_service_endpoint or '' }}"
+                                placeholder="https://<location>.cognitiveservices.azure.<com or us>/">
                         </div>
 
                         <div class="mb-3">
