Fixed error 1 (openAI title sanitization) and added unit tests

server/api/views/uploadFile/test_title.py

@@ -53,11 +53,11 @@ def test_falls_back_to_first_page_text_if_metadata_title_does_not_match_regex(se
         expected_title = "Advances in Mood Disorder Pharmacotherapy: Evaluating New Antipsychotics and Mood Stabilizers for Bipolar Disorder and Schizophrenia"
         self.assertEqual(expected_title, title.generate_title(doc))
 
-    @patch("api.services.openai_services.openAIServices.openAI")
+    @patch("api.views.uploadFile.title.openAIServices.openAI")
     def test_falls_back_to_chatgpt_if_no_title_found(self, mock_openAI):
         doc = MagicMock()
         doc.metadata = {"title": None}
-        doc.get_text.return_value = []
+        doc[0].get_text.return_value = []
 
         mock_response = MagicMock()
         mock_response.choices = [MagicMock()]
@@ -67,3 +67,34 @@ def test_falls_back_to_chatgpt_if_no_title_found(self, mock_openAI):
         title.generate_title(doc)
 
         self.assertTrue(mock_openAI.called)
+
+    @patch("api.views.uploadFile.title.openAIServices.openAI")
+    def test_strips_quotes_from_openai_title(self, mock_openAI):
+        doc = MagicMock()
+        doc.metadata = {"title": None}
+        doc[0].get_text.return_value = []
+
+        mock_response = MagicMock()
+        mock_response.choices = [MagicMock()]
+        mock_response.choices[0].message.content = '"Updated CANMAT/ISBD Guidelines for Treating Mixed Features in Bipolar Disorder"'
+        mock_openAI.return_value = mock_response
+
+        result = title.generate_title(doc)
+
+        self.assertEqual(result, "Updated CANMAT/ISBD Guidelines for Treating Mixed Features in Bipolar Disorder")
+
+    @patch("api.views.uploadFile.title.openAIServices.openAI")
+    def test_truncates_long_openai_title(self, mock_openAI):
+        doc = MagicMock()
+        doc.metadata = {"title": None}
+        doc[0].get_text.return_value = []
+
+        mock_response = MagicMock()
+        mock_response.choices = [MagicMock()]
+        mock_response.choices[0].message.content = "A" * 300
+        mock_openAI.return_value = mock_response
+
+        result = title.generate_title(doc)
+
+        # Ensure the title is truncated to fit the UploadFile model's title field (max_length=255), since OpenAI responses may exceed this limit
+        self.assertLessEqual(len(result), 255)

server/api/views/uploadFile/title.py

@@ -58,4 +58,6 @@ def summarize_pdf(pdf: fitz.Document) -> str:
     prompt = "Please provide a title for this document. The title should be less than 256 characters and will be displayed on a webpage."
     response = openAIServices.openAI(
         first_page_content, prompt, model='gpt-4o', temp=0.0)
-    return response.choices[0].message.content
+    title = response.choices[0].message.content.strip().strip('"').strip("'")
+    # Truncate to fit UploadFile model's max_length=255 title field as a final safeguard
+    return title[:255]

server/api/views/uploadFile/views.py

@@ -12,6 +12,9 @@
 import fitz
 from django.db import transaction
 from .title import generate_title
+import logging
+
+logger = logging.getLogger(__name__)
 
 
 class UploadFileView(APIView):
@@ -124,6 +127,7 @@ def post(self, request, format=None):
             )
         except Exception as e:
             # Handle potential errors
+            logger.exception("File upload failed for '%s': %s", uploaded_file.name, e)
             return Response({"message": f"Error processing file and embeddings: {str(e)}"},
                             status=status.HTTP_400_BAD_REQUEST)