Output the endpoint from Bicep

app/backend/prepdocslib/mediadescriber.py

@@ -105,4 +105,4 @@ async def describe_image(self, image_bytes) -> str:
                     results = await self.poll_api(session, poll_url, headers)
 
                 fields = results["result"]["contents"][0]["fields"]
-                return fields["DescriptionHTML"]["valueString"]
+                return fields["Description"]["valueString"]

app/backend/prepdocslib/pdfparser.py

@@ -138,17 +138,19 @@ class ObjectType(Enum):
                 added_objects = set()  # set of object types todo mypy
                 for idx, mask_char in enumerate(mask_chars):
                     object_type, object_idx = mask_char
-                    if object_idx is None:
-                        raise ValueError("object_idx should not be None")
                     if object_type == ObjectType.NONE:
                         page_text += form_recognizer_results.content[page_offset + idx]
                     elif object_type == ObjectType.TABLE:
+                        if object_idx is None:
+                            raise ValueError("Expected object_idx to be set")
                         if mask_char not in added_objects:
                             page_text += DocumentAnalysisParser.table_to_html(tables_on_page[object_idx])
                             added_objects.add(mask_char)
                     elif object_type == ObjectType.FIGURE:
                         if cu_describer is None:
                             raise ValueError("cu_describer should not be None, unable to describe figure")
+                        if object_idx is None:
+                            raise ValueError("Expected object_idx to be set")
                         if mask_char not in added_objects:
                             figure_html = await DocumentAnalysisParser.figure_to_html(
                                 doc_for_pymupdf, cu_describer, figures_on_page[object_idx]
@@ -176,7 +178,7 @@ async def figure_to_html(
         doc: pymupdf.Document, cu_describer: ContentUnderstandingDescriber, figure: DocumentFigure
     ) -> str:
         figure_title = (figure.caption and figure.caption.content) or ""
-        logger.info("Describing figure '%s' with title", figure.id, figure_title)
+        logger.info("Describing figure %s with title '%s'", figure.id, figure_title)
         if not figure.bounding_regions:
             return f"<figure><figcaption>{figure_title}</figcaption></figure>"
         for region in figure.bounding_regions:

infra/main.bicep

@@ -412,6 +412,7 @@ var appEnvVariables = {
   USE_LOCAL_PDF_PARSER: useLocalPdfParser
   USE_LOCAL_HTML_PARSER: useLocalHtmlParser
   USE_MEDIA_DESCRIBER_AZURE_CU: useMediaDescriberAzureCU
+  AZURE_CONTENTUNDERSTANDING_ENDPOINT: useMediaDescriberAzureCU ? contentUnderstanding.outputs.endpoint : ''
   RUNNING_IN_PRODUCTION: 'true'
 }
 
@@ -1193,6 +1194,7 @@ output AZURE_SPEECH_SERVICE_ID string = useSpeechOutputAzure ? speech.outputs.re
 output AZURE_SPEECH_SERVICE_LOCATION string = useSpeechOutputAzure ? speech.outputs.location : ''
 
 output AZURE_VISION_ENDPOINT string = useGPT4V ? computerVision.outputs.endpoint : ''
+output AZURE_CONTENTUNDERSTANDING_ENDPOINT string = useMediaDescriberAzureCU ? contentUnderstanding.outputs.endpoint : ''
 
 output AZURE_DOCUMENTINTELLIGENCE_SERVICE string = documentIntelligence.outputs.name
 output AZURE_DOCUMENTINTELLIGENCE_RESOURCE_GROUP string = documentIntelligenceResourceGroup.name

scripts/prepdocs.sh

@@ -9,4 +9,4 @@ if [ $# -gt 0 ]; then
   additionalArgs="$@"
 fi
 
-./.venv/bin/python ./app/backend/prepdocs.py './data/GPT4V_Examples/Financial Market Analysis Report 2023.pdf' --verbose $additionalArgs
+./.venv/bin/python ./app/backend/prepdocs.py './data/*' --verbose $additionalArgs

tests/conftest.py

@@ -34,6 +34,7 @@
     MockBlobClient,
     MockResponse,
     mock_computervision_response,
+    mock_contentunderstanding_response,
     mock_speak_text_cancelled,
     mock_speak_text_failed,
     mock_speak_text_success,
@@ -54,10 +55,12 @@ async def mock_search(self, *args, **kwargs):
 
 
 @pytest.fixture
-def mock_compute_embeddings_call(monkeypatch):
+def mock_azurehttp_calls(monkeypatch):
     def mock_post(*args, **kwargs):
         if kwargs.get("url").endswith("computervision/retrieval:vectorizeText"):
             return mock_computervision_response()
+        elif kwargs.get("url").endswith("/contentunderstanding/analyzers/image_analyzer:analyze"):
+            return mock_contentunderstanding_response()
         else:
             raise Exception("Unexpected URL for mock call to ClientSession.post()")
 
@@ -327,7 +330,7 @@ async def client(
     mock_openai_embedding,
     mock_acs_search,
     mock_blob_container_client,
-    mock_compute_embeddings_call,
+    mock_azurehttp_calls,
 ):
     quart_app = app.create_app()
 
@@ -346,7 +349,7 @@ async def client_with_expiring_token(
     mock_openai_embedding,
     mock_acs_search,
     mock_blob_container_client,
-    mock_compute_embeddings_call,
+    mock_azurehttp_calls,
 ):
     quart_app = app.create_app()
 

tests/mocks.py

@@ -203,6 +203,15 @@ def mock_computervision_response():
     )
 
 
+def mock_contentunderstanding_response():
+    return MockResponse(
+        status=200,
+        headers={
+            "Operation-Location": "https://cu-ztmfrxlgtk3nq.cognitiveservices.azure.com/contentunderstanding/analyzers/image_analyzer/results/53e4c016-d2c0-48a9-a9f4-38891f7d45f0?api-version=2024-12-01-preview"
+        },
+    )
+
+
 class MockAudio:
     def __init__(self, audio_data):
         self.audio_data = audio_data