old/check.py

import json
import os
import torchaudio
import subprocess

def check_audio_file(audio_path):
    # 检查文件是否存在
    if not os.path.exists(audio_path):
        print(f"[ERROR] 音频文件不存在: {audio_path}")
        return False
    # 尝试用torchaudio加载
    try:
        waveform, sr = torchaudio.load(audio_path)
        print(f"[OK] torchaudio加载成功: {audio_path}")
        print(f"    采样率: {sr}, 时长: {waveform.shape[1]/sr:.2f}s, 通道数: {waveform.shape[0]}")
    except Exception as e:
        print(f"[ERROR] torchaudio加载失败: {audio_path}, 错误: {e}")
        return False
    # 用sox/ffprobe获取详细信息
    try:
        sox_info = subprocess.check_output(['sox', '--i', audio_path], stderr=subprocess.STDOUT).decode()
        print(f"    sox信息:\n{sox_info}")
    except Exception as e:
        print(f"    [WARN] sox信息获取失败: {e}")
    return True

def check_json_fields(obj):
    # 检查messages字段
    messages = obj.get("messages", [])
    for i, msg in enumerate(messages):
        content = msg.get("content", "")
        if not isinstance(content, str):
            print(f"[ERROR] messages[{i}].content 不是字符串")
        if len(content) > 2000:
            print(f"[WARN] messages[{i}].content 超长: {len(content)} 字符")
        if any(ord(c) < 32 and c not in '\n\r\t' for c in content):
            print(f"[WARN] messages[{i}].content 含有不可见字符")
    # 检查solution字段
    if "solution" not in obj:
        print("[WARN] 缺少 solution 字段")
    return True

def main():
    jsonl_path = "dataset_10k_train.jsonl"  # 替换为你的文件
    with open(jsonl_path, "r", encoding="utf-8") as f:
        for idx, line in enumerate(f):
            print(f"\n==== 检查第 {idx+1} 条数据 ====")
            try:
                obj = json.loads(line)
            except Exception as e:
                print(f"[ERROR] JSON解析失败: {e}")
                continue
            check_json_fields(obj)
            audios = obj.get("audios", [])
            for audio_path in audios:
                check_audio_file(audio_path)
            print("==== 检查结束 ====")

if __name__ == "__main__":
    main()