import os
import subprocess
import json

def find_webm_without_txt(root_dir):
    """
    Finds all .webm files in the specified directory and its subdirectories
    that do not have a corresponding .txt file with the same base name.

    Args:
        root_dir: The root directory to search within.

    Returns:
        A list of tuples, where each tuple contains the path to a .webm file,
        its size in bytes, and its duration in seconds.
    """
    found_files = []
    for subdir, _, files in os.walk(root_dir):
        for file in files:
            if file.endswith('.webm'):
                webm_path = os.path.join(subdir, file)
                base_name = os.path.splitext(file)[0]
                txt_file = base_name + '.txt'
                txt_path = os.path.join(subdir, txt_file)

                if not os.path.exists(txt_path):
                    try:
                        # 获取文件大小
                        file_size = os.path.getsize(webm_path)
                        # 获取视频时长
                        duration = get_video_duration(webm_path)
                        # 添加路径、大小和时长到列表
                        found_files.append((webm_path, file_size, duration))
                    except OSError as e:
                        # 处理获取文件大小时的错误
                        print(f"无法获取文件信息 {webm_path}: {e}")

    return found_files

def get_video_duration(video_path):
    """
    使用ffprobe获取视频时长，如果出错则返回-1
    """
    try:
        # ffprobe命令来获取视频持续时间
        cmd = [
            'ffprobe', 
            '-v', 'error', 
            '-show_entries', 'format=duration', 
            '-of', 'json', 
            video_path
        ]
        result = subprocess.run(cmd, stdout=subprocess.PIPE, stderr=subprocess.PIPE, universal_newlines=True)
        
        if result.returncode != 0:
            print(f"ffprobe错误 {video_path}: {result.stderr.strip()}")
            return -1
            
        # 解析JSON输出
        data = json.loads(result.stdout)
        duration = float(data['format']['duration'])
        return duration
    except (subprocess.SubprocessError, json.JSONDecodeError, KeyError, ValueError) as e:
        print(f"处理视频时长时出错 {video_path}: {e}")
        return -1
    except FileNotFoundError:
        print("ffprobe命令没有找到。请确保FFmpeg已安装并添加到PATH中。")
        return -1

if __name__ == "__main__":
    video_dir = os.path.abspath(os.path.join(os.path.dirname(__file__), '..', 'video'))
    if not os.path.isdir(video_dir):
        print(f"错误：目录 '{video_dir}' 不存在。")
    else:
        missing_txt_files = find_webm_without_txt(video_dir)
        if missing_txt_files:
            print("找到以下没有对应 .txt 文件的 .webm 文件及其大小和时长：")
            # 修改打印部分以包含文件大小和时长
            for file_path, file_size, duration in missing_txt_files:
                duration_str = f"{duration:.2f} 秒" if duration is not None and duration >= 0 else "未知"
                print(f"{file_path} ({file_size} bytes, {duration_str})")
        else:
            print("所有 .webm 文件都有对应的 .txt 文件。")