Update main.py

main.py

@@ -184,42 +184,64 @@ import logging
 from typing import Optional
 
 
+# def get_collection_files(collection_id: str, user_id: str) -> str:
+#     """
+#     Synchronously get list of files in the specified collection using the external API
+#     with proper timeout and error handling.
+#     """
+#     try:
+#         url = "https://pvanand-documind-api-v2.hf.space/rag/get_collection_files"
+#         params = {
+#             "collection_id": collection_id,
+#             "user_id": user_id
+#         }
+#         headers = {
+#             'accept': 'application/json'
+#         }
+        
+#         logger.debug(f"Requesting collection files for user {user_id}, collection {collection_id}")
+        
+#         # Set timeout to 5 seconds
+#         response = requests.post(url, params=params, headers=headers, data='', timeout=5)
+        
+#         if response.status_code == 200:
+#             logger.info(f"Successfully retrieved collection files: {response.text[:100]}...")
+#             return response.text
+#         else:
+#             logger.error(f"API error (status {response.status_code}): {response.text}")
+#             return f"Error fetching files (status {response.status_code})"
+            
+#     except Timeout:
+#         logger.error("Timeout while fetching collection files")
+#         return "Error: Request timed out"
+#     except RequestException as e:
+#         logger.error(f"Network error fetching collection files: {str(e)}")
+#         return f"Error: Network issue - {str(e)}"
+#     except Exception as e:
+#         logger.error(f"Error fetching collection files: {str(e)}", exc_info=True)
+#         return f"Error fetching files: {str(e)}"
+
+
 def get_collection_files(collection_id: str, user_id: str) -> str:
-    """
-    Synchronously get list of files in the specified collection using the external API
-    with proper timeout and error handling.
-    """
+    """Get list of files in the specified collection"""
     try:
-        url = "https://pvanand-documind-api-v2.hf.space/rag/get_collection_files"
-        params = {
-            "collection_id": collection_id,
-            "user_id": user_id
-        }
-        headers = {
-            'accept': 'application/json'
-        }
+        # Get the full collection name
+        collection_name = f"{user_id}_{collection_id}"
         
-        logger.debug(f"Requesting collection files for user {user_id}, collection {collection_id}")
+        # Open the table and convert to pandas
+        table = db.open_table(collection_name)
+        df = table.to_pandas()
+        print(df.head())
         
-        # Set timeout to 5 seconds
-        response = requests.post(url, params=params, headers=headers, data='', timeout=5)
         
-        if response.status_code == 200:
-            logger.info(f"Successfully retrieved collection files: {response.text[:100]}...")
-            return response.text
-        else:
-            logger.error(f"API error (status {response.status_code}): {response.text}")
-            return f"Error fetching files (status {response.status_code})"
-            
-    except Timeout:
-        logger.error("Timeout while fetching collection files")
-        return "Error: Request timed out"
-    except RequestException as e:
-        logger.error(f"Network error fetching collection files: {str(e)}")
-        return f"Error: Network issue - {str(e)}"
+        # Get unique file names
+        unique_files = df['file_name'].unique()
+        
+        # Join the file names into a string
+        return ", ".join(unique_files)
     except Exception as e:
-        logger.error(f"Error fetching collection files: {str(e)}", exc_info=True)
-        return f"Error fetching files: {str(e)}"
+        logging.error(f"Error getting collection files: {str(e)}")
+        return f"Error getting files: {str(e)}"
 
 def format_for_model(state: dict, config: Optional[RunnableConfig] = None) -> list[BaseMessage]:
     """