celery 容灾升级

video-gen-api/app/services/shot_video_split_service.py

@@ -4,6 +4,7 @@ import asyncio
 import os
 import subprocess
 from dataclasses import dataclass
+from datetime import datetime
 from pathlib import Path
 
 from app.config import settings
@@ -19,12 +20,10 @@ class ShotSplitResult:
     url: str
     path: str
     file_size_bytes: int
+    temporary_path: str | None = None
 
 
 def _date_dir_from_segment_id(segment_id: str) -> str:
-    # 由调用方更适合按 created_at 传入；这里兜底按当前日期。
-    from datetime import datetime
-
     return datetime.now().strftime("%Y/%m/%d")
 
 
@@ -43,17 +42,21 @@ def split_video_segment(
     start_second: float,
     end_second: float,
     date_dir: str | None = None,
+    attempt_key: str | None = None,
+    finalize: bool = False,
 ) -> ShotSplitResult:
-    """使用 ffmpeg 拆出单个视频片段，输出到 storage/uploads/shot_segments。"""
-    source_path = Path(source_path)
+    """执行 ffmpeg 切片。
 
+    默认只产出 attempt 专属临时文件；调用方完成 Redis/DB fencing 后再原子移动到正式路径，
+    避免旧 Worker 在恢复任务接管后覆盖新结果。
+    """
+    source_path = Path(source_path)
     if not source_path.exists():
         raise RuntimeError(f"ffmpeg 拆镜失败：源视频不存在 {source_path}")
 
     start = max(float(start_second), 0.0)
     end = max(float(end_second), 0.0)
     duration = end - start
-
     if duration <= 0:
         raise RuntimeError(
             f"ffmpeg 拆镜失败：非法时间范围 start_second={start_second}, end_second={end_second}"
@@ -62,17 +65,11 @@ def split_video_segment(
     date_dir = date_dir or _date_dir_from_segment_id(segment_id)
     output_dir = ensure_shot_segment_dir(date_dir)
     output_path = output_dir / f"{segment_id}.mp4"
-
-    # 注意：
-    # 不能用 xxx.mp4.part，因为 ffmpeg 会按最后一个扩展名 .part 判断输出格式，导致：
-    # Unable to choose an output format
-    # 这里改为 xxx.part.mp4，让 ffmpeg 能识别 mp4 容器。
-    part_path = output_dir / f"{segment_id}.part.mp4"
-
+    suffix = str(attempt_key or "default").replace("/", "_").replace(":", "_")[:80]
+    part_path = output_dir / f"{segment_id}.{suffix}.part.mp4"
     _safe_unlink(part_path)
 
     timeout = int(getattr(settings, "SHOT_FFMPEG_TIMEOUT_SECONDS", 120) or 120)
-
     cmd = [
         get_ffmpeg_bin(),
         "-y",
@@ -100,19 +97,16 @@ def split_video_segment(
         "23",
         "-pix_fmt",
         "yuv420p",
-
         "-c:a",
         "aac",
         "-b:a",
         "128k",
-
         "-movflags",
         "+faststart",
 
         # 即使临时文件扩展名未来被改坏，也强制指定 mp4 muxer。
         "-f",
         "mp4",
-
         str(part_path),
     ]
 
@@ -136,20 +130,49 @@ def split_video_segment(
         raise RuntimeError(
             f"ffmpeg 拆镜失败: {completed.stderr.strip() or completed.stdout.strip()}"
         )
-
     if not part_path.exists() or part_path.stat().st_size <= 0:
         _safe_unlink(part_path)
         raise RuntimeError("ffmpeg 拆镜失败：输出文件为空")
 
-    os.replace(part_path, output_path)
+    if finalize:
+        os.replace(part_path, output_path)
+        return ShotSplitResult(
+            url=build_upload_url_from_path(output_path),
+            path=str(output_path),
+            file_size_bytes=output_path.stat().st_size,
+            temporary_path=None,
+        )
 
     return ShotSplitResult(
         url=build_upload_url_from_path(output_path),
         path=str(output_path),
-        file_size_bytes=output_path.stat().st_size,
+        file_size_bytes=part_path.stat().st_size,
+        temporary_path=str(part_path),
     )
 
 
+def finalize_split_result(result: ShotSplitResult) -> ShotSplitResult:
+    if not result.temporary_path:
+        return result
+    temp_path = Path(result.temporary_path)
+    final_path = Path(result.path)
+    if not temp_path.exists() or temp_path.stat().st_size <= 0:
+        raise RuntimeError("ffmpeg 拆镜临时结果不存在或为空")
+    final_path.parent.mkdir(parents=True, exist_ok=True)
+    os.replace(temp_path, final_path)
+    return ShotSplitResult(
+        url=result.url,
+        path=str(final_path),
+        file_size_bytes=final_path.stat().st_size,
+        temporary_path=None,
+    )
+
+
+def cleanup_split_result(result: ShotSplitResult | None) -> None:
+    if result and result.temporary_path:
+        _safe_unlink(Path(result.temporary_path))
+
+
 async def split_video_segment_async(
     *,
     source_path: str | Path,
@@ -157,12 +180,9 @@ async def split_video_segment_async(
     start_second: float,
     end_second: float,
     date_dir: str | None = None,
+    attempt_key: str | None = None,
+    finalize: bool = False,
 ) -> ShotSplitResult:
-    """异步拆镜入口。
-
-    ffmpeg 本身是同步阻塞命令，不能直接在 Celery 进程内唯一 event loop 中执行。
-    这里通过 asyncio.to_thread 跑同步拆镜函数，避免阻塞 asyncpg / Redis / HTTP 等异步任务。
-    """
     return await asyncio.to_thread(
         split_video_segment,
         source_path=source_path,
@@ -170,4 +190,6 @@ async def split_video_segment_async(
         start_second=start_second,
         end_second=end_second,
         date_dir=date_dir,
-    )
+        attempt_key=attempt_key,
+        finalize=finalize,
+    )