| 
							 | 
						
 | 
					
					
						
						| 
							 | 
						
 | 
					
					
						
						| 
							 | 
						
 | 
					
					
						
						| 
							 | 
						
 | 
					
					
						
						| 
							 | 
						
 | 
					
					
						
						| 
							 | 
						
 | 
					
					
						
						| 
							 | 
						
 | 
					
					
						
						| 
							 | 
						
 | 
					
					
						
						| 
							 | 
						
 | 
					
					
						
						| 
							 | 
						
 | 
					
					
						
						| 
							 | 
						
 | 
					
					
						
						| 
							 | 
						
 | 
					
					
						
						| 
							 | 
						
 | 
					
					
						
						| 
							 | 
						
 | 
					
					
						
						| 
							 | 
						import os
 | 
					
					
						
						| 
							 | 
						import shutil
 | 
					
					
						
						| 
							 | 
						from tqdm import tqdm
 | 
					
					
						
						| 
							 | 
						
 | 
					
					
						
						| 
							 | 
						paths = []
 | 
					
					
						
						| 
							 | 
						
 | 
					
					
						
						| 
							 | 
						
 | 
					
					
						
						| 
							 | 
						def gather_paths(input_dir, output_dir):
 | 
					
					
						
						| 
							 | 
						    os.makedirs(output_dir, exist_ok=True)
 | 
					
					
						
						| 
							 | 
						
 | 
					
					
						
						| 
							 | 
						    for video in sorted(os.listdir(input_dir)):
 | 
					
					
						
						| 
							 | 
						        if video.endswith(".mp4"):
 | 
					
					
						
						| 
							 | 
						            video_input = os.path.join(input_dir, video)
 | 
					
					
						
						| 
							 | 
						            video_output = os.path.join(output_dir, video)
 | 
					
					
						
						| 
							 | 
						            if os.path.isfile(video_output):
 | 
					
					
						
						| 
							 | 
						                continue
 | 
					
					
						
						| 
							 | 
						            paths.append([video_input, output_dir])
 | 
					
					
						
						| 
							 | 
						        elif os.path.isdir(os.path.join(input_dir, video)):
 | 
					
					
						
						| 
							 | 
						            gather_paths(os.path.join(input_dir, video), os.path.join(output_dir, video))
 | 
					
					
						
						| 
							 | 
						
 | 
					
					
						
						| 
							 | 
						
 | 
					
					
						
						| 
							 | 
						def main(input_dir, output_dir):
 | 
					
					
						
						| 
							 | 
						    print(f"Recursively gathering video paths of {input_dir} ...")
 | 
					
					
						
						| 
							 | 
						    gather_paths(input_dir, output_dir)
 | 
					
					
						
						| 
							 | 
						
 | 
					
					
						
						| 
							 | 
						    for video_input, output_dir in tqdm(paths):
 | 
					
					
						
						| 
							 | 
						        shutil.move(video_input, output_dir)
 | 
					
					
						
						| 
							 | 
						
 | 
					
					
						
						| 
							 | 
						
 | 
					
					
						
						| 
							 | 
						if __name__ == "__main__":
 | 
					
					
						
						| 
							 | 
						    input_dir = "/mnt/bn/maliva-gen-ai-v2/chunyu.li/multilingual_dcc"
 | 
					
					
						
						| 
							 | 
						    output_dir = "/mnt/bn/maliva-gen-ai-v2/chunyu.li/multilingual"
 | 
					
					
						
						| 
							 | 
						
 | 
					
					
						
						| 
							 | 
						    main(input_dir, output_dir)
 | 
					
					
						
						| 
							 | 
						
 |