首次提交

tools/check_srt.py

@@ -0,0 +1,117 @@
+"""
+使用 cos-python-sdk-v5 全量扫描 0004/data/ 下的文件，
+找出已有 SRT 文件的目录，将对应的 MP4 COS 路径写入 txt 文件。
+
+安装依赖：pip install cos-python-sdk-v5
+"""
+import os
+import re
+from collections import defaultdict
+from qcloud_cos import CosConfig, CosS3Client
+from config.config_utils import config as app_config
+
+SCAN_PREFIX = "0004/data/"
+OUTPUT_TXT  = "./text/0004_has_srt.txt"
+DIR_RE      = re.compile(r'.+/data/[^/]+/[^/]+/$')
+
+
+def build_client() -> tuple[CosS3Client, str]:
+    cos_cfg = app_config.get_cos_config()
+    secret_id  = cos_cfg.get("secret_id")
+    secret_key = cos_cfg.get("secret_key")
+    region     = cos_cfg.get("region", "ap-chongqing")
+    bucket     = cos_cfg.get("bucket", "asr-dig-data-1253181912")
+
+    if not secret_id or not secret_key or "填入" in secret_id:
+        raise ValueError("请先在 config/config.json 的 tencentcloud.cos 中填入 secret_id 和 secret_key")
+
+    cfg    = CosConfig(Region=region, SecretId=secret_id, SecretKey=secret_key)
+    client = CosS3Client(cfg)
+    return client, bucket
+
+
+def list_all_objects(client: CosS3Client, bucket: str, prefix: str) -> list[str]:
+    """分页获取 prefix 下全部 Key"""
+    keys   = []
+    marker = ""
+    page   = 0
+
+    while True:
+        page += 1
+        kwargs = dict(Bucket=bucket, Prefix=prefix, MaxKeys=1000)
+        if marker:
+            kwargs["Marker"] = marker
+
+        resp     = client.list_objects(**kwargs)
+        contents = resp.get("Contents", [])
+        keys.extend(item["Key"] for item in contents)
+
+        print(f"\r  第 {page} 页，已获取 {len(keys)} 个文件...", end="", flush=True)
+
+        if resp.get("IsTruncated") == "true":
+            marker = resp.get("NextMarker", "")
+        else:
+            break
+
+    print()
+    return keys
+
+
+def main():
+    client, bucket = build_client()
+    bucket_prefix  = f"cos://{bucket}"
+
+    print(f"正在全量扫描 {bucket_prefix}/{SCAN_PREFIX} ...")
+    keys = list_all_objects(client, bucket, SCAN_PREFIX)
+    print(f"共找到 {len(keys)} 个文件")
+
+    if not keys:
+        print("未找到任何文件，请检查密钥和路径是否正确")
+        return
+
+    # 按父目录分组
+    dir_files: dict[str, list[str]] = defaultdict(list)
+    for key in keys:
+        idx      = key.rfind("/")
+        dir_path = key[:idx + 1]   # e.g. "0004/data/BV1xx/BV1xx/"
+        filename = key[idx + 1:]
+        if filename:
+            dir_files[dir_path].append(filename)
+
+    # 找有 SRT 的目录，收集 MP4 路径
+    mp4_paths   = []
+    has_srt_cnt = 0
+    no_srt_cnt  = 0
+
+    for dir_path, filenames in sorted(dir_files.items()):
+        # 只处理 data/{id}/{id}/ 结构
+        if not DIR_RE.match(dir_path):
+            continue
+
+        has_srt   = any(f.lower().endswith(".srt") for f in filenames)
+        mp4_files = [f for f in filenames if f.lower().endswith(".mp4")]
+
+        if has_srt:
+            has_srt_cnt += 1
+            for mp4 in mp4_files:
+                mp4_paths.append(f"{bucket_prefix}/{dir_path}{mp4}")
+        else:
+            no_srt_cnt += 1
+
+    print(f"有 SRT 的目录: {has_srt_cnt} 个  →  {len(mp4_paths)} 个 MP4")
+    print(f"无 SRT 的目录: {no_srt_cnt} 个")
+
+    if not mp4_paths:
+        print("没有找到含 SRT 的目录")
+        return
+
+    os.makedirs("./text", exist_ok=True)
+    with open(OUTPUT_TXT, "w", encoding="utf-8") as f:
+        for path in mp4_paths:
+            f.write(path + "\n")
+
+    print(f"已写入: {OUTPUT_TXT}  ({len(mp4_paths)} 条)")
+
+
+if __name__ == "__main__":
+    main()