fix: Handled the file format if its in UPPER CASE (#1200)

Azure-Samples · Aug 6, 2024 · 1786e21 · 1786e21
1 parent 20a6843
commit 1786e21
Show file tree

Hide file tree

Showing 3 changed files with 23 additions and 1 deletion.
diff --git a/code/backend/batch/utilities/helpers/embedders/push_embedder.py b/code/backend/batch/utilities/helpers/embedders/push_embedder.py
@@ -38,7 +38,7 @@ def __init__(self, blob_client: AzureBlobStorageClient, env_helper: EnvHelper):
             self.embedding_configs[ext] = processor
 
     def embed_file(self, source_url: str, file_name: str):
-        file_extension = file_name.split(".")[-1]
+        file_extension = file_name.split(".")[-1].lower()
         embedding_config = self.embedding_configs.get(file_extension)
         self.__embed(
             source_url=source_url,

diff --git a/code/tests/functional/conftest.py b/code/tests/functional/conftest.py
@@ -334,6 +334,12 @@ def setup_config_mocking(httpserver: HTTPServer):
                     "loading": {"strategy": "web"},
                     "use_advanced_image_processing": False,
                 },
+                {
+                    "document_type": "htm",
+                    "chunking": {"strategy": "layout", "size": 500, "overlap": 100},
+                    "loading": {"strategy": "web"},
+                    "use_advanced_image_processing": False,
+                },
                 {
                     "document_type": "docx",
                     "chunking": {"strategy": "layout", "size": 500, "overlap": 100},

diff --git a/code/tests/utilities/helpers/test_push_embedder.py b/code/tests/utilities/helpers/test_push_embedder.py
@@ -347,6 +347,22 @@ def test_embed_file_chunks_documents(document_loading_mock, document_chunking_mo
     )
 
 
+def test_embed_file_chunks_documents_upper_case(document_loading_mock, document_chunking_mock, env_helper_mock):
+    # given
+    push_embedder = PushEmbedder(MagicMock(), env_helper_mock)
+
+    # when
+    push_embedder.embed_file(
+        "some-url",
+        "some-file-name.PDF",
+    )
+
+    # then
+    document_chunking_mock.return_value.chunk.assert_called_once_with(
+        document_loading_mock.return_value.load.return_value, CHUNKING_SETTINGS
+    )
+
+
 def test_embed_file_generates_embeddings_for_documents(llm_helper_mock, env_helper_mock):
     # given
     push_embedder = PushEmbedder(MagicMock(), env_helper_mock)