Increase test coverage

Author: pamelafox

.github/workflows/python-test.yaml

@@ -62,10 +62,8 @@ jobs:
         - name: Check formatting with black
           run: black . --check --verbose
         - name: Run Python tests
-          if: runner.os != 'Windows'
-          run: pytest -s -vv --cov --cov-report=xml --cov-fail-under=89
+          run: pytest -s -vv --cov --cov-report=xml --cov-fail-under=90
         - name: Check diff coverage
-          if: runner.os != 'Windows' && github.event_name == 'pull_request'
           run: |
             git fetch origin main:refs/remotes/origin/main
             diff-cover coverage.xml --compare-branch=origin/main --fail-under=90

tests/conftest.py

@@ -42,6 +42,7 @@
     MockAzureCredential,
     MockAzureCredentialExpired,
     MockBlobClient,
+    MockDirectoryClient,
     MockResponse,
     MockTransport,
     mock_retrieval_response,
@@ -237,6 +238,8 @@ async def mock_acreate(*args, **kwargs):
             answer = "capital of France"
         elif last_question == "Generate search query for: Are interest rates high?":
             answer = "interest rates"
+        elif last_question == "Generate search query for: Flowers in westbrae nursery logo?":
+            answer = "westbrae nursery logo"
         elif isinstance(last_question, list) and any([part.get("image_url") for part in last_question]):
             answer = "From the provided sources, the impact of interest rates and GDP growth on financial markets can be observed through the line graph. [Financial Market Analysis Report 2023-7.png]"
         else:
@@ -362,6 +365,8 @@ async def mock_exists(*args, **kwargs):
         "AZURE_SERVER_APP_SECRET": "SECRET",
         "AZURE_CLIENT_APP_ID": "CLIENT_APP",
         "AZURE_TENANT_ID": "TENANT_ID",
+        "USE_MULTIMODAL": "true",
+        "AZURE_VISION_ENDPOINT": "https://testvision.cognitiveservices.azure.com/",
     },
 ]
 
@@ -666,6 +671,7 @@ async def auth_client(
     mock_validate_token_success,
     mock_list_groups_success,
     mock_acs_search_filter,
+    mock_azurehttp_calls,
     request,
 ):
     monkeypatch.setenv("AZURE_STORAGE_ACCOUNT", "test-storage-account")
@@ -757,7 +763,6 @@ async def vision_client(
         test_app.app.config.update({"TESTING": True})
         mock_openai_chatcompletion(test_app.app.config[app.CONFIG_OPENAI_CLIENT])
         mock_openai_embedding(test_app.app.config[app.CONFIG_OPENAI_CLIENT])
-        # need to mock app.CONFIG_GLOBAL_BLOB_MANAGER to
         mock_blob_service_client = BlobServiceClient(
             f"https://{os.environ['AZURE_STORAGE_ACCOUNT']}.blob.core.windows.net",
             credential=MockAzureCredential(),
@@ -1083,6 +1088,15 @@ def mock_readinto(self, stream: IO[bytes]):
     monkeypatch.setattr(azure.storage.filedatalake.aio.StorageStreamDownloader, "readinto", mock_readinto)
 
 
+@pytest.fixture
+def mock_user_directory_client(monkeypatch):
+    monkeypatch.setattr(
+        azure.storage.filedatalake.aio.FileSystemClient,
+        "get_directory_client",
+        lambda *args, **kwargs: MockDirectoryClient(),
+    )
+
+
 @pytest.fixture
 def chat_approach():
     return ChatReadRetrieveReadApproach(

tests/mocks.py

@@ -136,7 +136,78 @@ def __init__(self, text, highlights=None, additional_properties=None):
 
 class MockAsyncSearchResultsIterator:
     def __init__(self, search_text, vector_queries: Optional[list[VectorQuery]]):
-        if search_text == "interest rates" or (
+        if search_text == "westbrae nursery logo" and (
+            vector_queries and any([vector.fields == "images/embedding" for vector in vector_queries])
+        ):
+            self.data = [
+                [
+                    {
+                        "id": "file-westbrae_jun28_pdf-77657374627261655F6A756E32382E7064667B276F696473273A205B2766653437353262612D623565652D343531632D623065312D393332316664663365353962275D7D-page-0",
+                        "content": '<figure><figcaption>1.1 <br>The image displays the Gmail logo. It consists of a stylized letter "M" with four colors: red, blue, green, and yellow. To the right of the "M" is the word "Gmail" written in gray text. The design is modern and clean. The colors used are characteristic of Google\'s branding.</figcaption></figure>\n\n\nPamela Fox <[email protected]>\n\nReceipt / Tax invoice (#2-108442)\n\nWestbrae Nursery <[email protected]>\nReply-To: [email protected]\nTo: [email protected]\n\nSat, Jun 28, 2025 at 1:21 PM\n\n\n<figure><figcaption>1.2 <br>The image shows the logo of Westbrae Nursery. The logo features three daffodil flowers on the left side. The text "Westbrae" is positioned to the right of the flowers. Below "Westbrae" is the word "Nursery." The design is simple and rendered in black and white.</figcaption></figure>\n\n\nAn Employee-Owned Co-op\n1272 Gilman St, Berkeley, CA 94706\n510-526-5517\n\nMain Outlet\n\nReceipt / Tax Invoice #2-108442 28 Jun 2025 1:21pm\n\n\n<figure><table><tr><td>1 Gopher Baskets</td><td>@ $7.',
+                        "category": None,
+                        "sourcepage": "westbrae_jun28.pdf#page=1",
+                        "sourcefile": "westbrae_jun28.pdf",
+                        "oids": ["OID_X"],
+                        "groups": [],
+                        "captions": [],
+                        "score": 0.05000000447034836,
+                        "reranker_score": 3.2427687644958496,
+                        "search_agent_query": None,
+                        "images": [
+                            {
+                                "url": "https://userst5gj4l5eootrlo.dfs.core.windows.net/user-content/OID_X/images/westbrae_jun28.pdf/page_0/figure1_1.png",
+                                "description": '<figure><figcaption>1.1 <br>The image displays the Gmail logo. It consists of a stylized letter "M" with four colors: red, blue, green, and yellow. To the right of the "M" is the word "Gmail" written in gray text. The design is modern and clean. The colors used are characteristic of Google\'s branding.</figcaption></figure>',
+                                "boundingbox": [32.99, 43.65, 126.14, 67.72],
+                            },
+                            {
+                                "url": "https://userst5gj4l5eootrlo.dfs.core.windows.net/user-content/OID_X/images/westbrae_jun28.pdf/page_0/figure1_2.png",
+                                "description": '<figure><figcaption>1.2 <br>The image shows the logo of Westbrae Nursery. The logo features three daffodil flowers on the left side. The text "Westbrae" is positioned to the right of the flowers. Below "Westbrae" is the word "Nursery." The design is simple and rendered in black and white.</figcaption></figure>',
+                                "boundingbox": [40.76, 163.42, 347.1, 354.15],
+                            },
+                        ],
+                    },
+                    {
+                        "id": "file-westbrae_jun28_pdf-77657374627261655F6A756E32382E7064667B276F696473273A205B2766653437353262612D623565652D343531632D623065312D393332316664663365353962275D7D-page-1",
+                        "content": "</figcaption></figure>\n\n\nAn Employee-Owned Co-op\n1272 Gilman St, Berkeley, CA 94706\n510-526-5517\n\nMain Outlet\n\nReceipt / Tax Invoice #2-108442 28 Jun 2025 1:21pm\n\n\n<figure><table><tr><td>1 Gopher Baskets</td><td>@ $7.99</td><td>$7.99</td></tr><tr><td>1 qt</td><td></td><td></td></tr><tr><td rowSpan=2>1 Gopher Baskets 1 gal</td><td>@ $14.99</td><td>$14.99</td></tr><tr><td></td><td></td></tr><tr><td>1 Edible 4.99</td><td>@ $4.99</td><td>$4.99</td></tr><tr><td>4 Color 11.99</td><td>@ $11.99</td><td>$47.96</td></tr><tr><td>1 Edible $6.99</td><td>@ $6.99</td><td>$6.99</td></tr><tr><td>Subtotal</td><td></td><td>$82.",
+                        "category": None,
+                        "sourcepage": "westbrae_jun28.pdf#page=1",
+                        "sourcefile": "westbrae_jun28.pdf",
+                        "oids": ["OID_X"],
+                        "groups": [],
+                        "captions": [],
+                        "score": 0.04696394503116608,
+                        "reranker_score": 1.8582123517990112,
+                        "search_agent_query": None,
+                        "images": [
+                            {
+                                "url": "https://userst5gj4l5eootrlo.dfs.core.windows.net/user-content/OID_X/images/westbrae_jun28.pdf/page_0/figure1_1.png",
+                                "description": '<figure><figcaption>1.1 <br>The image displays the Gmail logo. It consists of a stylized letter "M" with four colors: red, blue, green, and yellow. To the right of the "M" is the word "Gmail" written in gray text. The design is modern and clean. The colors used are characteristic of Google\'s branding.</figcaption></figure>',
+                                "boundingbox": [32.99, 43.65, 126.14, 67.72],
+                            },
+                            {
+                                "url": "https://userst5gj4l5eootrlo.dfs.core.windows.net/user-content/OID_X/images/westbrae_jun28.pdf/page_0/figure1_2.png",
+                                "description": '<figure><figcaption>1.2 <br>The image shows the logo of Westbrae Nursery. The logo features three daffodil flowers on the left side. The text "Westbrae" is positioned to the right of the flowers. Below "Westbrae" is the word "Nursery." The design is simple and rendered in black and white.</figcaption></figure>',
+                                "boundingbox": [40.76, 163.42, 347.1, 354.15],
+                            },
+                        ],
+                    },
+                    {
+                        "id": "file-westbrae_jun28_pdf-77657374627261655F6A756E32382E7064667B276F696473273A205B2766653437353262612D623565652D343531632D623065312D393332316664663365353962275D7D-page-4",
+                        "content": "\n\nIf you have any questions about how to take care of the plants you purchase\nor if they start to show symptoms of ill health, please, give us a call (510-526-\n5517)\n\nreceipt.pdf\n50K",
+                        "category": None,
+                        "sourcepage": "westbrae_jun28.pdf#page=2",
+                        "sourcefile": "westbrae_jun28.pdf",
+                        "oids": ["OID_X"],
+                        "groups": [],
+                        "captions": [],
+                        "score": 0.016393441706895828,
+                        "reranker_score": 1.7518715858459473,
+                        "search_agent_query": None,
+                        "images": [],
+                    },
+                ]
+            ]
+        elif search_text == "interest rates" or (
             vector_queries and any([vector.fields == "images/embedding" for vector in vector_queries])
         ):
             self.data = [
@@ -363,6 +434,30 @@ def get(self):
         return self.__result
 
 
+# Mock DirectoryClient used in blobmanager.py:AdlsBlobManager
+class MockDirectoryClient:
+    async def get_directory_properties(self):
+        # Return dummy properties to indicate directory exists
+        return {"name": "test-directory"}
+
+    async def get_access_control(self):
+        # Return a dictionary with the owner matching the auth_client's user_oid
+        return {"owner": "OID_X"}  # This should match the user_oid in auth_client
+
+    def get_file_client(self, filename):
+        # Return a file client for the given filename
+        return MockFileClient(filename)
+
+
+# Mock FileClient used in blobmanager.py:AdlsBlobManager
+class MockFileClient:
+    def __init__(self, path_name):
+        self.path_name = path_name
+
+    async def download_file(self):
+        return MockBlob()
+
+
 def mock_speak_text_success(self, text):
     return MockSynthesisResult(MockAudio("mock_audio_data"))
 

tests/snapshots/test_app/test_ask_rtr_text_filter/auth_client0/result.json

@@ -14,7 +14,7 @@
                 "description": "What is the capital of France?",
                 "props": {
                     "filter": "category ne 'excluded' and (oids/any(g:search.in(g, 'OID_X')) or groups/any(g:search.in(g, 'GROUP_Y, GROUP_Z')))",
-                    "search_image_embeddings": false,
+                    "search_image_embeddings": true,
                     "search_text_embeddings": true,
                     "top": 3,
                     "use_query_rewriting": false,