check_progress.py

#!/usr/bin/env python3

import os
import glob

def check_processing_progress():
    """Check the progress of FT_work dataset processing"""
    
    # Paths
    data_path = "/apdcephfs_gy4/share_303628665/joywu/dataset/FT_work/videos/train"
    save_path = "/apdcephfs_gy5/share_303628665/joyewu/research/AVH-Align/ft_work_preprocessed/train"
    
    print("=== FT_Work 数据集处理进度检查 ===")
    print(f"数据路径: {data_path}")
    print(f"输出路径: {save_path}")
    print()
    
    # Find all MP4 files in the dataset
    print("扫描数据集中的MP4文件...")
    mp4_files = []
    for mp4_path in glob.glob(os.path.join(data_path, "**", "*.mp4"), recursive=True):
        rel_path = os.path.relpath(mp4_path, data_path)
        mp4_files.append(rel_path)
    
    total_files = len(mp4_files)
    print(f"总MP4文件数: {total_files}")
    print()
    
    # Check processing status
    processed_files = []
    unprocessed_files = []
    failed_files = []
    
    for mp4_file in mp4_files:
        video_dir = os.path.dirname(mp4_file)
        video_basename = os.path.basename(mp4_file)
        
        output_dir = os.path.join(save_path, video_dir)
        roi_path = os.path.join(output_dir, video_basename[:-4] + '_roi.mp4')
        wav_path = os.path.join(output_dir, video_basename[:-4] + '.wav')
        
        if os.path.exists(roi_path) and os.path.exists(wav_path):
            processed_files.append(mp4_file)
        else:
            unprocessed_files.append(mp4_file)
            # Check if it's a known problematic file
            if "Real_CelebV-HQ" in mp4_file or "Real_HDTF" in mp4_file:
                failed_files.append(mp4_file)
    
    # Display results
    processed_count = len(processed_files)
    unprocessed_count = len(unprocessed_files)
    failed_count = len(failed_files)
    
    print("=== 处理状态汇总 ===")
    print(f"✅ 已处理完成: {processed_count} 个文件 ({processed_count/total_files*100:.1f}%)")
    print(f"⏳ 待处理: {unprocessed_count} 个文件 ({unprocessed_count/total_files*100:.1f}%)")
    print(f"❌ 已知问题文件: {failed_count} 个文件")
    print()
    
    if processed_files:
        print("✅ 已处理的文件示例:")
        for i, file in enumerate(processed_files[:5]):
            print(f"  {i+1}. {file}")
        if len(processed_files) > 5:
            print(f"  ... 还有 {len(processed_files) - 5} 个文件")
        print()
    
    if unprocessed_files:
        print("⏳ 待处理的文件示例:")
        for i, file in enumerate(unprocessed_files[:10]):
            print(f"  {i+1}. {file}")
        if len(unprocessed_files) > 10:
            print(f"  ... 还有 {len(unprocessed_files) - 10} 个文件")
        print()
    
    if failed_files:
        print("❌ 已知问题文件 (可能需要特殊处理):")
        for i, file in enumerate(failed_files[:10]):
            print(f"  {i+1}. {file}")
        if len(failed_files) > 10:
            print(f"  ... 还有 {len(failed_files) - 10} 个文件")
        print()
    
    # Recommendations
    print("=== 建议 ===")
    if unprocessed_count > 0:
        print("1. 可以继续运行预处理命令，程序会自动跳过已处理的文件")
        print("2. 建议使用较少的worker数量进行测试:")
        print("   python deepfake_preprocess.py --dataset FT_work --split train --data_path /apdcephfs_gy4/share_303628665/joywu/dataset/FT_work/videos --save_path /apdcephfs_gy5/share_303628665/joyewu/research/AVH-Align/ft_work_preprocessed --max_workers 8")
    
    if failed_count > 0:
        print("3. 对于Real_CelebV-HQ和Real_HDTF文件，可能需要检查视频格式或质量")
    
    print()
    print("=== 重新运行命令 ===")
    print("python deepfake_preprocess.py \\")
    print("    --dataset FT_work \\")
    print("    --split train \\")
    print("    --data_path /apdcephfs_gy4/share_303628665/joywu/dataset/FT_work/videos \\")
    print("    --save_path /apdcephfs_gy5/share_303628665/joyewu/research/AVH-Align/ft_work_preprocessed")

if __name__ == "__main__":
    check_processing_progress()