xiaoxia-saas/scripts/backfill_generated_videos.py

"""补写 generated_videos 记录 — 修复一键生成场景下 project_id 为空导致未写表的问题.

背景：#535 之前的一键生成场景，由于 edit_plan.project_id 为空，
_finalize_render_success 中 `if generation_task_id and project_id:` 判断不通过，
导致渲染成功的视频没有写入 generated_videos 表，成片库看不到。

本脚本扫描所有 status=completed 的 generation_task，
如果没有对应的 generated_videos 记录，则从 plan.config.rendered_url 补写。

用法：
    PYTHONPATH="apps/worker:apps/api:packages" python scripts/backfill_generated_videos.py

安全说明：
- 只读事务扫描，补写前检查是否已存在，不会重复写入
- 失败不中断，继续处理下一个
"""

from __future__ import annotations

import logging
import sys
from datetime import datetime, UTC
from uuid import uuid4

# 配置日志
logging.basicConfig(
    level=logging.INFO,
    format="%(asctime)s [%(levelname)s] %(message)s",
)
logger = logging.getLogger(__name__)


def main() -> int:
    from sqlalchemy import exists
    from sqlalchemy.orm import Session

    from packages.adapters.sqlalchemy_impl.models import (
        EditPlanModel,
        GeneratedVideoModel,
        GenerationTaskModel,
    )
    from packages.adapters.sqlalchemy_impl.session import get_db_session

    db: Session = next(get_db_session())

    try:
        # 1. 查询所有 status=completed 的 generation_task
        completed_tasks = (
            db.query(GenerationTaskModel)
            .filter(
                GenerationTaskModel.status == "completed",
            )
            .all()
        )

        logger.info("找到 %d 个已完成的生成任务", len(completed_tasks))

        fixed_count = 0
        skipped_count = 0
        failed_count = 0

        for task in completed_tasks:
            task_id = task.id

            # 2. 检查是否已有 generated_videos 记录
            exists_video = db.query(exists().where(GeneratedVideoModel.generation_task_id == task_id)).scalar()

            if exists_video:
                skipped_count += 1
                continue

            # 3. 从 source_edit_plan_id 找到对应的 edit_plan，取 rendered_url
            if not task.source_edit_plan_id:
                logger.warning("任务 %s 无 source_edit_plan_id，跳过", task_id)
                skipped_count += 1
                continue

            plan = (
                db.query(EditPlanModel)
                .filter(
                    EditPlanModel.id == task.source_edit_plan_id,
                )
                .first()
            )

            if not plan:
                logger.warning("任务 %s 对应的剪辑计划不存在，跳过", task_id)
                skipped_count += 1
                continue

            # 从 config 中取 rendered_url
            config = plan.config or {}
            rendered_url = config.get("rendered_url", "")
            rendered_storage_key = config.get("rendered_storage_key", "")

            if not rendered_url:
                logger.warning("任务 %s 的剪辑计划无 rendered_url，跳过", task_id)
                skipped_count += 1
                continue

            # 4. 补写 generated_videos 记录
            try:
                video_id = uuid4().hex
                video = GeneratedVideoModel(
                    id=video_id,
                    project_id=task.project_id or "",
                    generation_task_id=task_id,
                    name=f"generated-{task_id[:8]}.mp4",
                    file_url=rendered_url,
                    storage_key=rendered_storage_key,
                    file_size=0,
                    duration=0.0,
                    width=0,
                    height=0,
                    fps=0.0,
                    status="completed",
                    generation_params={"mode": "backfill", "plan_id": plan.id},
                    review_status="pending",
                    created_at=task.completed_at or datetime.now(UTC),
                    updated_at=datetime.now(UTC),
                )
                db.add(video)
                db.commit()
                fixed_count += 1
                logger.info("已补写视频记录: task_id=%s video_id=%s url=%s", task_id, video_id, rendered_url[:80])
            except Exception as e:
                db.rollback()
                failed_count += 1
                logger.error("补写视频记录失败: task_id=%s error=%s", task_id, e)

        logger.info(
            "补写完成：共 %d 个已完成任务，补写 %d 条，跳过 %d 条，失败 %d 条",
            len(completed_tasks),
            fixed_count,
            skipped_count,
            failed_count,
        )
        return 0 if failed_count == 0 else 1

    finally:
        db.close()


if __name__ == "__main__":
    sys.exit(main())