@@ -496,3 +496,176 @@ async def mock_delete_documents(self, documents):
496496 assert len (searched_filters ) == 1 , "It should have searched once"
497497 assert searched_filters [0 ] == "sourcefile eq 'foo.pdf'"
498498 assert len (deleted_documents ) == 0 , "It should have deleted no documents"
499+
500+
501+ @pytest .mark .asyncio
502+ async def test_create_index_with_search_images (monkeypatch , search_info ):
503+ """Test that SearchManager correctly creates an index with image search capabilities."""
504+ indexes = []
505+
506+ async def mock_create_index (self , index ):
507+ indexes .append (index )
508+
509+ async def mock_list_index_names (self ):
510+ for index in []:
511+ yield index
512+
513+ monkeypatch .setattr (SearchIndexClient , "create_index" , mock_create_index )
514+ monkeypatch .setattr (SearchIndexClient , "list_index_names" , mock_list_index_names )
515+
516+ # Create a SearchInfo with an Azure Vision endpoint
517+ search_info_with_vision = SearchInfo (
518+ endpoint = search_info .endpoint ,
519+ credential = search_info .credential ,
520+ index_name = search_info .index_name ,
521+ azure_vision_endpoint = "https://testvision.cognitiveservices.azure.com/" ,
522+ )
523+
524+ # Create a SearchManager with search_images=True
525+ manager = SearchManager (search_info_with_vision , search_images = True , field_name_embedding = "embedding" )
526+ await manager .create_index ()
527+
528+ # Verify the index was created correctly
529+ assert len (indexes ) == 1 , "It should have created one index"
530+ assert indexes [0 ].name == "test"
531+
532+ # Find the "images" field in the index
533+ images_field = next ((field for field in indexes [0 ].fields if field .name == "images" ), None )
534+ assert images_field is not None , "The index should include an 'images' field"
535+
536+ # Verify the "images" field structure
537+ assert images_field .type .startswith (
538+ "Collection(Edm.ComplexType)"
539+ ), "The 'images' field should be a collection of complex type"
540+
541+ # Check subfields of the images field
542+ image_subfields = images_field .fields
543+ assert len (image_subfields ) == 4 , "The 'images' field should have 4 subfields"
544+
545+ # Verify specific subfields
546+ assert any (field .name == "embedding" for field in image_subfields ), "Should have an 'embedding' subfield"
547+ assert any (field .name == "url" for field in image_subfields ), "Should have a 'url' subfield"
548+ assert any (field .name == "description" for field in image_subfields ), "Should have a 'description' subfield"
549+ assert any (field .name == "boundingbox" for field in image_subfields ), "Should have a 'boundingbox' subfield"
550+
551+ # Verify vector search configuration
552+ vectorizers = indexes [0 ].vector_search .vectorizers
553+ assert any (
554+ v .vectorizer_name == "images-vision-vectorizer" for v in vectorizers
555+ ), "Should have an AI Vision vectorizer"
556+
557+ # Verify vector search profile
558+ profiles = indexes [0 ].vector_search .profiles
559+ assert any (p .name == "images_embedding_profile" for p in profiles ), "Should have an image embedding profile"
560+
561+
562+ @pytest .mark .asyncio
563+ async def test_create_index_with_search_images_no_endpoint (monkeypatch , search_info ):
564+ """Test that SearchManager raises an error when search_images=True but no Azure Vision endpoint is provided."""
565+ indexes = []
566+
567+ async def mock_create_index (self , index ):
568+ indexes .append (index )
569+
570+ async def mock_list_index_names (self ):
571+ for index in []:
572+ yield index
573+
574+ monkeypatch .setattr (SearchIndexClient , "create_index" , mock_create_index )
575+ monkeypatch .setattr (SearchIndexClient , "list_index_names" , mock_list_index_names )
576+
577+ # Create a SearchManager with search_images=True but no Azure Vision endpoint
578+ manager = SearchManager (
579+ search_info , # search_info doesn't have azure_vision_endpoint
580+ search_images = True ,
581+ field_name_embedding = "embedding" ,
582+ )
583+
584+ # Verify that create_index raises a ValueError
585+ with pytest .raises (ValueError ) as excinfo :
586+ await manager .create_index ()
587+
588+ # Check the error message
589+ assert "Azure AI Vision endpoint must be provided to use image embeddings" in str (excinfo .value )
590+
591+
592+ @pytest .mark .asyncio
593+ async def test_create_index_with_search_images_and_embeddings (monkeypatch , search_info ):
594+ """Test that SearchManager correctly creates an index with both image search and embeddings."""
595+ indexes = []
596+
597+ async def mock_create_index (self , index ):
598+ indexes .append (index )
599+
600+ async def mock_list_index_names (self ):
601+ for index in []:
602+ yield index
603+
604+ async def mock_create_client (* args , ** kwargs ):
605+ return MockClient (
606+ embeddings_client = MockEmbeddingsClient (
607+ create_embedding_response = openai .types .CreateEmbeddingResponse (
608+ object = "list" ,
609+ data = [
610+ openai .types .Embedding (
611+ embedding = [0.1 , 0.2 , 0.3 ],
612+ index = 0 ,
613+ object = "embedding" ,
614+ )
615+ ],
616+ model = MOCK_EMBEDDING_MODEL_NAME ,
617+ usage = Usage (prompt_tokens = 8 , total_tokens = 8 ),
618+ )
619+ )
620+ )
621+
622+ monkeypatch .setattr (SearchIndexClient , "create_index" , mock_create_index )
623+ monkeypatch .setattr (SearchIndexClient , "list_index_names" , mock_list_index_names )
624+
625+ # Create a SearchInfo with an Azure Vision endpoint
626+ search_info_with_vision = SearchInfo (
627+ endpoint = search_info .endpoint ,
628+ credential = search_info .credential ,
629+ index_name = search_info .index_name ,
630+ azure_vision_endpoint = "https://testvision.cognitiveservices.azure.com/" ,
631+ )
632+
633+ # Create embeddings service
634+ embeddings = AzureOpenAIEmbeddingService (
635+ open_ai_service = "x" ,
636+ open_ai_deployment = "x" ,
637+ open_ai_model_name = MOCK_EMBEDDING_MODEL_NAME ,
638+ open_ai_dimensions = MOCK_EMBEDDING_DIMENSIONS ,
639+ open_ai_api_version = "test-api-version" ,
640+ credential = AzureKeyCredential ("test" ),
641+ disable_batch = True ,
642+ )
643+ monkeypatch .setattr (embeddings , "create_client" , mock_create_client )
644+
645+ # Create a SearchManager with both search_images and embeddings
646+ manager = SearchManager (
647+ search_info_with_vision , search_images = True , embeddings = embeddings , field_name_embedding = "embedding3"
648+ )
649+ await manager .create_index ()
650+
651+ # Verify the index was created correctly
652+ assert len (indexes ) == 1 , "It should have created one index"
653+
654+ # Find both the embeddings field and images field
655+ embedding_field = next ((field for field in indexes [0 ].fields if field .name == "embedding3" ), None )
656+ images_field = next ((field for field in indexes [0 ].fields if field .name == "images" ), None )
657+
658+ assert embedding_field is not None , "The index should include an 'embedding3' field"
659+ assert images_field is not None , "The index should include an 'images' field"
660+
661+ # Verify vector search configuration includes both text and image vectorizers
662+ vectorizers = indexes [0 ].vector_search .vectorizers
663+ assert any (
664+ v .vectorizer_name == "images-vision-vectorizer" for v in vectorizers
665+ ), "Should have an AI Vision vectorizer"
666+ assert any (hasattr (v , "ai_services_vision_parameters" ) for v in vectorizers ), "Should have AI vision parameters"
667+
668+ # Verify vector search profiles for both text and images
669+ profiles = indexes [0 ].vector_search .profiles
670+ assert any (p .name == "images_embedding_profile" for p in profiles ), "Should have an image embedding profile"
671+ assert any (p .name == "embedding3-profile" for p in profiles ), "Should have a text embedding profile"
0 commit comments