feat: 渲染失败检测+任务超时机制 (#1219)
CI/CD Pipeline / Check if frontend-only change (pull_request) Successful in 39s
CI/CD Pipeline / Validate - Type Check (mypy) (pull_request) Successful in 46s
CI/CD Pipeline / Validate - Migration (alembic) (pull_request) Successful in 50s
Preview Deploy / Deploy Preview Environment (pull_request) Successful in 1m25s
PR Automation / Auto Merge on CI Green + Approved (pull_request) Successful in 1m40s
AI Code Review / AI Code Review (pull_request) Failing after 2m13s
CI/CD Pipeline / PR Build API Image (pull_request) Successful in 1m52s
CI/CD Pipeline / PR Build Worker Image (pull_request) Successful in 1m53s
CI/CD Pipeline / Unit Tests (pull_request) Failing after 2m2s
CI/CD Pipeline / Validate - Code Quality (pull_request) Has been cancelled
CI/CD Pipeline / Integration Tests (pull_request) Has been cancelled
CI/CD Pipeline / Build Production API Image (pull_request) Has been cancelled
CI/CD Pipeline / Build Production Web Image (pull_request) Has been cancelled
CI/CD Pipeline / Build Production Worker Image (pull_request) Has been cancelled
CI/CD Pipeline / Deploy Production (pull_request) Has been cancelled
CI/CD Pipeline / Production Browser E2E (pull_request) Has been cancelled
CI/CD Pipeline / Canary Release to Production (pull_request) Has been cancelled
CI/CD Pipeline / CI Gate (pull_request) Has been cancelled
PR Automation / Auto Approve on CI Green (pull_request) Has been cancelled
CI/CD Pipeline / PR Build Web Image (pull_request) Failing after 1081h8m47s
CI/CD Pipeline / Build Staging Web Image (pull_request) Failing after 1081h9m28s
CI/CD Pipeline / Build Staging API Image (pull_request) Failing after 1081h9m29s
CI/CD Pipeline / Frontend Unit Tests (pull_request) Failing after 1081h8m48s
CI/CD Pipeline / Frontend Lint (pull_request) Failing after 1081h8m48s
CI/CD Pipeline / ACR Image Cleanup (pull_request) Failing after 1081h9m24s
CI/CD Pipeline / Staging API Integration Tests (pull_request) Failing after 1081h9m25s
CI/CD Pipeline / Staging E2E Tests (pull_request) Failing after 1081h9m25s
CI/CD Pipeline / Deploy Staging (Watchtower auto-deploy) (pull_request) Failing after 1081h9m27s
CI/CD Pipeline / Build Staging Worker Image (pull_request) Failing after 1081h42m42s

- 新增 video_validation 模块:渲染后校验输出完整性
  - 文件存在性/大小检查
  - moov atom 检测(MP4容器完整性)
  - ffprobe 视频流验证
  - FFmpeg 退出码中文映射(OOM/段错误/滤镜错误等)
- render_adapter: 渲染后自动校验输出,不通过则不上传OSS
- compose_video: 添加 soft_time_limit=600s (10分钟)
  - 区分超时/FFmpeg错误/通用异常的失败处理
- render_edit_plan: 添加 soft_time_limit=600s
  - 超时不重试,直接标记失败
- generate_video: 添加 soft_time_limit=600s
- _startup: 孤儿任务清理扩展到 Job 表
- 21 个单元测试全绿
This commit is contained in:
xiaoxia
2026-08-02 14:17:11 +08:00
parent 380274e635
commit 4fca2e48fb
7 changed files with 770 additions and 14 deletions
+267
View File
@@ -0,0 +1,267 @@
"""视频输出校验 — 渲染后验证输出文件完整性.
在 FFmpeg 渲染完成后,验证输出文件是否有效(非损坏/截断):
1. 文件存在且大小 > 0
2. moov atom 存在(MP4 容器完整性标志)
3. ffprobe 可正常读取视频流信息
用途:
- RenderAdapter 渲染后调用,避免将损坏文件上传到 OSS
- 提前发现 FFmpeg 异常退出但未抛异常的情况(如 exit=0 但输出截断)
"""
from __future__ import annotations
import logging
import subprocess
from dataclasses import dataclass
from pathlib import Path
from shared.ffmpeg_utils import FFPROBE_BIN
logger = logging.getLogger(__name__)
# ── 数据结构 ──────────────────────────────────────────────────────────────────
@dataclass
class VideoValidationResult:
"""视频校验结果。"""
valid: bool
file_exists: bool = False
file_size: int = 0
moov_atom_found: bool = False
has_video_stream: bool = False
duration: float = 0.0
width: int = 0
height: int = 0
error_message: str = ""
@property
def is_valid(self) -> bool:
"""是否通过所有校验。"""
return self.valid
# ── FFmpeg 退出码映射 ──────────────────────────────────────────────────────────
# 常见 FFmpeg 退出码及其含义
FFMPEG_EXIT_CODES: dict[int, tuple[str, str]] = {
0: ("成功", "渲染正常完成"),
1: ("通用错误", "FFmpeg 执行出错,请检查输入参数和素材"),
69: ("权限错误", "无权限写入输出文件或访问输入文件"),
126: ("权限不足", "命令不可执行"),
127: ("命令不存在", "FFmpeg 二进制文件未找到"),
134: ("Abnormal termination", "FFmpeg 异常终止(可能内存不足)"),
137: ("OOM Killed", "FFmpeg 被系统 OOM Killer 终止(内存不足)"),
139: ("段错误", "FFmpeg Segmentation Fault(可能是编解码器 Bug)"),
141: ("管道断裂", "FFmpeg 输出管道断裂"),
143: ("SIGTERM", "FFmpeg 收到终止信号"),
183: ("滤镜错误", "FFmpeg 滤镜链配置错误"),
234: ("素材异常", "输入素材格式不兼容或已损坏"),
255: ("严重错误", "FFmpeg 执行严重错误"),
}
def get_exit_code_message(exit_code: int) -> str:
"""获取 FFmpeg 退出码的中文描述。
Args:
exit_code: FFmpeg 进程退出码
Returns:
人类可读的错误描述
"""
if exit_code in FFMPEG_EXIT_CODES:
name, desc = FFMPEG_EXIT_CODES[exit_code]
return f"exit={exit_code} ({name}): {desc}"
if exit_code > 128:
signal_num = exit_code - 128
return f"exit={exit_code}: 被信号 {signal_num} 终止"
return f"exit={exit_code}: 未知错误"
# ── 校验函数 ──────────────────────────────────────────────────────────────────
def validate_video_output(video_path: str | Path, *, min_duration: float = 0.1) -> VideoValidationResult:
"""校验渲染输出视频文件的完整性。
校验项:
1. 文件存在且大小 > 0
2. 包含 moov atom(MP4 容器完整性)
3. ffprobe 可读取至少一个视频流
Args:
video_path: 输出视频文件路径
min_duration: 最小有效时长(秒),低于此值视为无效,默认 0.1s
Returns:
VideoValidationResult
"""
path = Path(video_path)
result = VideoValidationResult(valid=False)
# 1. 文件存在性检查
if not path.exists():
result.error_message = f"输出文件不存在: {path}"
logger.error("[video-validation] %s", result.error_message)
return result
result.file_exists = True
# 2. 文件大小检查
try:
result.file_size = path.stat().st_size
except OSError as e:
result.error_message = f"无法读取文件大小: {e}"
logger.error("[video-validation] %s", result.error_message)
return result
if result.file_size == 0:
result.error_message = "输出文件大小为 0(FFmpeg 未写入任何数据)"
logger.error("[video-validation] %s", result.error_message)
return result
# 极小文件(< 1KB)几乎不可能是有效视频
if result.file_size < 1024:
result.error_message = f"输出文件过小 ({result.file_size} bytes),可能渲染未完成"
logger.error("[video-validation] %s", result.error_message)
return result
# 3. moov atom 检查(MP4 容器完整性标志)
result.moov_atom_found = _check_moov_atom(path)
if not result.moov_atom_found:
result.error_message = (
"输出文件缺少 moov atom(MP4 容器不完整)。" "可能原因:FFmpeg 被强制终止、磁盘空间不足、或渲染过程中断。"
)
logger.error("[video-validation] %s — file_size=%d", result.error_message, result.file_size)
return result
# 4. ffprobe 验证视频流
try:
probe_result = subprocess.run( # nosec B603
[
FFPROBE_BIN,
"-v",
"error",
"-select_streams",
"v:0",
"-show_entries",
"stream=width,height,duration,codec_name",
"-show_entries",
"format=duration",
"-of",
"json",
str(path),
],
check=True,
stdout=subprocess.PIPE,
stderr=subprocess.PIPE,
text=True,
timeout=15,
)
import json
probe_data = json.loads(probe_result.stdout)
streams = probe_data.get("streams", [])
fmt = probe_data.get("format", {})
if not streams:
result.error_message = "输出文件无视频流(ffprobe 未检测到 video stream)"
logger.error("[video-validation] %s", result.error_message)
return result
result.has_video_stream = True
video_stream = streams[0]
result.width = int(video_stream.get("width", 0) or 0)
result.height = int(video_stream.get("height", 0) or 0)
result.duration = float(fmt.get("duration", 0) or video_stream.get("duration", 0) or 0)
# 时长校验
if result.duration > 0 and result.duration < min_duration:
result.error_message = (
f"输出视频时长过短 ({result.duration:.2f}s < {min_duration}s)," "可能渲染只处理了极少帧"
)
logger.warning("[video-validation] %s", result.error_message)
# 不标记为失败,只是警告(某些预览场景确实很短)
# 但如果时长为 0 且文件大小较大,说明 moov 有问题
elif result.duration == 0 and result.file_size > 0:
logger.warning(
"[video-validation] ffprobe 无法读取时长,但文件存在且大小=%d,标记为可疑",
result.file_size,
)
except subprocess.TimeoutExpired:
result.error_message = "ffprobe 超时(15s),输出文件可能已损坏"
logger.error("[video-validation] %s", result.error_message)
return result
except subprocess.CalledProcessError as e:
stderr_text = (e.stderr or "").strip()
result.error_message = f"ffprobe 校验失败: exit={e.returncode}, stderr={stderr_text[:200]}"
logger.error("[video-validation] %s", result.error_message)
return result
except (json.JSONDecodeError, KeyError, ValueError) as e:
result.error_message = f"ffprobe 输出解析失败: {e}"
logger.error("[video-validation] %s", result.error_message)
return result
# 全部通过
result.valid = True
logger.info(
"[video-validation] 校验通过: path=%s size=%d duration=%.2fs resolution=%dx%d",
path,
result.file_size,
result.duration,
result.width,
result.height,
)
return result
def _check_moov_atom(video_path: Path) -> bool:
"""检查 MP4 文件是否包含 moov atom。
moov atom 是 MP4 容器的元数据容器,包含视频时长、编解码器信息等。
FFmpeg 使用 -movflags +faststart 时 moov 在文件头部;否则在尾部。
如果 FFmpeg 被强制终止,moov 可能完全不存在。
方法:读取文件前 64KB + 尾部 64KB,搜索 "moov" 字节标记。
对于 -movflags +faststart 的快启文件,moov 在头部。
Args:
video_path: 视频文件路径
Returns:
True 表示找到 moov atom
"""
try:
file_size = video_path.stat().st_size
if file_size < 8:
return False
# 搜索范围:头部 64KB + 尾部 64KB(覆盖 faststart 和普通 MP4)
search_size = min(64 * 1024, file_size)
with open(video_path, "rb") as f:
# 读取头部
head_data = f.read(search_size)
if b"moov" in head_data:
return True
# 读取尾部
if file_size > search_size:
f.seek(file_size - search_size)
tail_data = f.read(search_size)
if b"moov" in tail_data:
return True
return False
except OSError as e:
logger.warning("[video-validation] 检查 moov atom 失败: %s", e)
return False