ingestor context handling

utils/sources.py

@@ -122,15 +122,32 @@ def _process_context(context: Union[str, List[Dict[str, Any]]]) -> tuple[str, Li
             if isinstance(result, str):
                 result = ast.literal_eval(result)
             
+            # Handle both ingested files (metadata at top level) and retrieved documents (metadata in answer_metadata)
+            # Check if metadata is nested in 'answer_metadata' (retrieved documents)
             metadata = result.get('answer_metadata', {})
-            doc_info = {
-                'answer': result.get('answer', ''),
-                'filename': metadata.get('filename', 'Unknown'),
-                'page': metadata.get('page', 'Unknown'),
-                'year': metadata.get('year', 'Unknown'),
-                'source': metadata.get('source', 'Unknown'),
-                'document_id': metadata.get('_id', 'Unknown')
-            }
+            
+            # If answer_metadata is empty or missing, check top level (ingested files)
+            if not metadata or all(v is None or v == 'Unknown' for v in metadata.values()):
+                # For ingested files, metadata is at the top level
+                doc_info = {
+                    'answer': result.get('answer', result.get('content', '')),
+                    'filename': result.get('filename', 'Unknown'),
+                    'page': result.get('page', 'Unknown'),
+                    'year': result.get('year', 'Unknown'),
+                    'source': result.get('source', 'Unknown'),
+                    'document_id': result.get('_id', result.get('document_id', 'Unknown'))
+                }
+            else:
+                # For retrieved documents, use nested metadata
+                doc_info = {
+                    'answer': result.get('answer', ''),
+                    'filename': metadata.get('filename', 'Unknown'),
+                    'page': metadata.get('page', 'Unknown'),
+                    'year': metadata.get('year', 'Unknown'),
+                    'source': metadata.get('source', 'Unknown'),
+                    'document_id': metadata.get('_id', 'Unknown')
+                }
+            
             processed_results.append(doc_info)
         
         # Format context string - SIMPLIFIED TO ONLY USE [1], [2], [3]