Update app.py

app.py

@@ -22,13 +22,6 @@ def download_dataset_file(dataset_id, local_dir):
     dataset = load_dataset(dataset_id)
     cache_file_info = dataset.cache_files
     print(f"Cache File Info: {cache_file_info}\n")
-    # filename = cache_file_info['test'][0]['filename']
-    # snapshot_id = filename.split('/')[-2]
-    # dataset_name = filename.split('/')[-5]
-    # dataset_name = dataset_name.replace('___', '--')
-    # base_path = os.path.join('/home/user/.cache/huggingface/hub', 'datasets--' + dataset_name)
-    # snapshot_path = os.path.join(base_path, "snapshots", snapshot_id)
-    # print(f"snapshot_path: {snapshot_path}")
     
     snapshot_path = "/home/user/.cache/huggingface/hub/datasets--350016z--flores_plus_Taiwanese/snapshots/9a8fe738903c9ab08620db4553c582376bcdd64f"
     contents = os.listdir(snapshot_path)
@@ -42,18 +35,25 @@ def download_dataset_file(dataset_id, local_dir):
             source_file_path = os.path.join(snapshot_path, file_name)
             local_file_path = os.path.join(local_dir, file_name)
             
-            shutil.move(source_file_path, local_file_path)
-            print(f"Moved {file_name} to {local_file_path}")
+            shutil.copy(source_file_path, local_file_path)  # 改為複製操作
+            print(f"Copied {file_name} to {local_file_path}")
+            
+            # 確認檔案存在
             if not os.path.exists(local_file_path):
-                print(f"Error: {local_file_path} does not exist after moving.")
+                print(f"Error: {local_file_path} does not exist after copying.")
                 continue
+            
+            # 檢查檔案權限
+            print(f"Permissions for {local_file_path}: {oct(os.stat(local_file_path).st_mode)}")
+            
+            # 延遲以確保檔案系統同步
+            time.sleep(1)
           
     return local_dir
 
 DATASET_ID = "350016z/flores_plus_Taiwanese"
 current_dir = os.getcwd()
 print(f"Before -------- Contents of current_dir: {os.listdir(current_dir)}--------")
-
 download_dataset_file(DATASET_ID, current_dir)
 print(f"After -------- Contents of current_dir: {os.listdir(current_dir)}--------")