yangdx commited on
Commit
989a2b6
·
1 Parent(s): 22483ea

fix: Improve file path handling and logging for document scanning

Browse files

• Convert relative paths to absolute paths
• Add logging for file scanning progress
• Log total number of new files found
• Enhance file scanning feedback
• Improve path resolution safety

Files changed (1) hide show
  1. lightrag/api/lightrag_server.py +6 -0
lightrag/api/lightrag_server.py CHANGED
@@ -564,6 +564,10 @@ def parse_args() -> argparse.Namespace:
564
 
565
  args = parser.parse_args()
566
 
 
 
 
 
567
  ollama_server_infos.LIGHTRAG_MODEL = args.simulated_model_name
568
 
569
  return args
@@ -595,6 +599,7 @@ class DocumentManager:
595
  """Scan input directory for new files"""
596
  new_files = []
597
  for ext in self.supported_extensions:
 
598
  for file_path in self.input_dir.rglob(f"*{ext}"):
599
  if file_path not in self.indexed_files:
600
  new_files.append(file_path)
@@ -1198,6 +1203,7 @@ def create_app(args):
1198
  new_files = doc_manager.scan_directory_for_new_files()
1199
  scan_progress["total_files"] = len(new_files)
1200
 
 
1201
  for file_path in new_files:
1202
  try:
1203
  with progress_lock:
 
564
 
565
  args = parser.parse_args()
566
 
567
+ # conver relative path to absolute path
568
+ args.working_dir = os.path.abspath(args.working_dir)
569
+ args.input_dir = os.path.abspath(args.input_dir)
570
+
571
  ollama_server_infos.LIGHTRAG_MODEL = args.simulated_model_name
572
 
573
  return args
 
599
  """Scan input directory for new files"""
600
  new_files = []
601
  for ext in self.supported_extensions:
602
+ logger.info(f"Scanning for {ext} files in {self.input_dir}")
603
  for file_path in self.input_dir.rglob(f"*{ext}"):
604
  if file_path not in self.indexed_files:
605
  new_files.append(file_path)
 
1203
  new_files = doc_manager.scan_directory_for_new_files()
1204
  scan_progress["total_files"] = len(new_files)
1205
 
1206
+ logger.info(f"Found {len(new_files)} new files to index.")
1207
  for file_path in new_files:
1208
  try:
1209
  with progress_lock: