Update app.py

app.py

@@ -637,7 +637,8 @@ def extract_sources_from_search_context(search_context: str, user_language: str
     found_files = set()
     
     # Pattern to match "From filename.pdf (category):" or "From filename.docx (category):"
-    from_pattern = r'From\s+([^:\s]+\.(?:pdf|docx))\s*\('
+    # Updated to handle assets/pdfs/ prefix and empty parentheses, and stop at the colon
+    from_pattern = r'From\s+assets/pdfs/([^:]+\.(?:pdf|docx))\s*\([^)]*\)'
     matches = re.findall(from_pattern, search_context)
     
     for match in matches:
@@ -650,38 +651,47 @@ def extract_sources_from_search_context(search_context: str, user_language: str
             if pdf in search_context:
                 found_files.add(pdf)
     
-    # Filter by user language preference
-    language_filtered_files = set()
+    # Filter by user language preference and prioritize by relevance
+    language_filtered_files = []
     
+    # First, collect all language-appropriate documents
     for pdf in found_files:
         # Determine document language from filename
         doc_language = get_document_language(pdf)
         
         # Language matching logic
+        should_include = False
         if user_language == 'en' or user_language == 'singlish':
             # English users can see all documents, but prefer English versions
             if doc_language in ['en', 'english'] or user_language == 'singlish':
-                language_filtered_files.add(pdf)
+                should_include = True
             elif not any(f for f in found_files if get_document_language(f) in ['en', 'english']):
                 # If no English version available, show other languages
-                language_filtered_files.add(pdf)
+                should_include = True
         elif user_language == 'si' or user_language == 'sinhala':
             # Sinhala users prefer Sinhala documents
             if doc_language in ['si', 'sinhala']:
-                language_filtered_files.add(pdf)
+                should_include = True
             elif not any(f for f in found_files if get_document_language(f) in ['si', 'sinhala']):
                 # If no Sinhala version available, show other languages
-                language_filtered_files.add(pdf)
+                should_include = True
         elif user_language == 'ta' or user_language == 'tamil':
             # Tamil users prefer Tamil documents
             if doc_language in ['ta', 'tamil']:
-                language_filtered_files.add(pdf)
+                should_include = True
             elif not any(f for f in found_files if get_document_language(f) in ['ta', 'tamil']):
                 # If no Tamil version available, show other languages
-                language_filtered_files.add(pdf)
+                should_include = True
         else:
             # Default: show all documents
-            language_filtered_files.add(pdf)
+            should_include = True
+            
+        if should_include:
+            language_filtered_files.append(pdf)
+    
+    # Prioritize by relevance - take only the first (most relevant) document
+    if language_filtered_files:
+        language_filtered_files = [language_filtered_files[0]]
     
     # Convert to list with short names
     for pdf in language_filtered_files: