LingYing Agent 32c3d2f263
CI/CD Pipeline / Dedup Check - skip PR tests when covered by push pipeline (pull_request) Successful in 1s
CI/CD Pipeline / Check if frontend-only change (pull_request) Successful in 1s
CI/CD Pipeline / PR Build Worker Image (pull_request) Successful in 27s
CI/CD Pipeline / PR Build API Image (pull_request) Successful in 1m43s
Preview Deploy / Deploy Preview Environment (pull_request) Successful in 1m55s
PR Automation / Auto Approve on CI Green (pull_request) Successful in 2m44s
AI Code Review / AI Code Review (pull_request) Successful in 6m34s
CI/CD Pipeline / Integration Tests (pull_request) Successful in 8m17s
CI/CD Pipeline / Unit Tests (pull_request) Successful in 8m45s
CI/CD Pipeline / Validate - Style (pull_request) Has been cancelled
CI/CD Pipeline / Validate - Security (pull_request) Has been cancelled
CI/CD Pipeline / Validate - Python (mypy + alembic) (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 Merge on CI Green + Approved (pull_request) Has been cancelled
CI/CD Pipeline / Staging E2E Tests (pull_request) Failing after 112h13m34s
CI/CD Pipeline / Frontend Unit Tests (pull_request) Failing after 112h13m49s
CI/CD Pipeline / Deploy Staging (Watchtower auto-deploy) (pull_request) Failing after 112h13m21s
CI/CD Pipeline / Retag skipped Staging Worker Image (pull_request) Failing after 112h13m26s
CI/CD Pipeline / Retag skipped Staging Web Image (pull_request) Failing after 112h13m27s
CI/CD Pipeline / Retag skipped Staging API Image (pull_request) Failing after 112h13m27s
CI/CD Pipeline / Build Staging Worker Image (pull_request) Failing after 112h13m32s
CI/CD Pipeline / Build Staging API Image (pull_request) Failing after 112h13m32s
CI/CD Pipeline / Staging API Integration Tests (pull_request) Failing after 112h13m16s
CI/CD Pipeline / ACR Image Cleanup (pull_request) Failing after 112h13m15s
CI/CD Pipeline / PR Build Web Image (pull_request) Failing after 112h13m32s
CI/CD Pipeline / Frontend Lint (pull_request) Failing after 112h13m33s
CI/CD Pipeline / Check push changed paths (pull_request) Failing after 112h13m36s
CI/CD Pipeline / Build Staging Web Image (pull_request) Failing after 112h48m43s
fix(ai-avatar): 全盘修复 FFmpeg 渲染滤镜链路(exit 234 P0)
根因分析(退出码 234 = Invalid argument):
1) drawtext 使用了无效参数 font=bold,导致整个 filter_complex 解析失败
   —— FFmpeg drawtext 没有 bold 参数;改为通过 borderw 模拟粗体视觉效果
2) DRAWTEXT_FONT_SEARCH_PATHS 未包含 Dockerfile 中 COPY 的
   NotoSansSC-VF.ttf 路径,且把不支持中文的 DejaVuSans 放在 fallback 首位,
   导致字体 fallback 到拉丁字体,中文渲染乱码/方框
3) build_broll_overlay_filter 硬编码 1280:720 横屏尺寸,AI 数字人是 9:16 竖屏
4) B-roll PIP 输入索引错误(用 len(sorted_segments) 而非原始下标映射),
   PIP 标签链断裂([pip0]→[vout0] 而非 [vout])
5) 渲染命令缺 -map 0:a?,合成后音频丢失
6) 标题滤镜与 B-roll 输出标签拼接用分号有缺陷,末尾分号处理偶发问题

修复内容(packages/domain/video_filter_builder.py):
- DRAWTEXT_FONT_SEARCH_PATHS: VF 字体置顶、移除 DejaVuSans、加 Bold.ttc 路径
- DRAWTEXT_FONT_MAP: 思源黑体等关键字改为 NotoSansSC(匹配 VF 文件名)
- 删除 font=bold 无效参数;bold 无显式描边时自动用 borderw=3+同色描边模拟粗体
- build_broll_overlay_filter 返回 (filter_str, final_label) 元组,解决标签问题
- 重写 fullscreen/PIP 拆分:原始列表下标决定 -i 输入序号,sorted 只用于时序处理
- PIP overlay 正确基于 fullscreen 输出([vout_fs])或主视频([0:v])链接
- output_width/output_height 贯穿所有 scale/pad/overlay,默认 1280x720 兼容旧调用

修复内容(apps/api/app/services/ai_avatar_render_service.py):
- 新增 _probe_video_resolution() 用 ffprobe 探测输入视频实际分辨率
- AI 数字人默认竖屏 720x1280,探测失败兜底不阻断渲染
- build_broll_overlay_filter 调用传实际 output_width/output_height
- 标题滤镜传实际尺寸,保证位置/坐标计算正确
- filter_complex 拼接重写:broll+title/broll-only/title-only/无滤镜四分支清晰
- _build_ffmpeg_command 补 -map 0:a? + -c:a aac,音频不再丢失
- 清理冗余内联 import

测试:
- 更新 test_ai_avatar_render_routes.py / test_video_filter_builder.py 适配新元组签名
- 新增 test_bold_true_does_not_use_font_bold_param 防回归
- 本地 ffmpeg 实测中文标题+B-roll+竖屏720x1280合成成功
- 相关 3524 个单测全通过
2026-09-11 23:09:22 +08:00
2026-06-27 19:02:31 +08:00

小虾 SaaS - 自动化视频剪辑平台

Python 3.12+ FastAPI PostgreSQL

自动化视频剪辑 SaaS 平台,支持素材上传、AI 分类、智能剪辑计划生成、自动化视频合成与成片管理。


✨ 核心功能

🎬 视频剪辑主链路

  • 素材上传(直传 OSS + 分片上传大文件,最大 2GB)
  • AI 智能分类与质量评分
  • 4 种剪辑模式:one_take / pip(画中画)/ voice_over(口播+B-roll)/ voice_pip
  • 剪辑计划模板 + 智能生成
  • 自动化视频合成任务(Celery 异步)
  • 成片下载与审核管理
  • 资产诊断(素材就绪度评估、缺口分析)

🔐 认证系统

  • JWT Bearer Token 认证
  • 邮箱注册 + 邮箱验证
  • 密码重置(邮箱找回)
  • bcrypt 密码加密

📋 项目管理

  • 项目 CRUD + 共享
  • 任务管理(创建/更新/状态流转/进度追踪)
  • 里程碑管理
  • 任务问题追踪

📊 素材库管理

  • 素材库创建与管理
  • 素材上传、审核状态流转(pending_review → approved/rejected)
  • 素材诊断(就绪度评分、缺口分析、智能视图)

🚀 快速开始

方式 1: Docker Compose(推荐)

# 1. 克隆仓库
git clone https://git.xiaoxiajianji.com/xiaoxia/xiaoxia-saas.git
cd xiaoxia-saas

# 2. 配置环境变量
cp .env.example .env
# 编辑 .env 填写数据库、Redis、OSS 等配置

# 3. 启动所有服务
docker-compose up -d

# 4. 访问 API 文档
open http://localhost:8000/docs

方式 2: 本地开发

# 1. 克隆仓库
git clone https://git.xiaoxiajianji.com/xiaoxia/xiaoxia-saas.git
cd xiaoxia-saas

# 2. 创建虚拟环境
python -m venv venv
source venv/bin/activate  # Windows: venv\Scripts\activate

# 3. 安装依赖
pip install -r requirements.txt

# 4. 配置环境变量
cp .env.example .env

# 5. 启动 API 服务
uvicorn apps.api.main:app --reload

# 6. 访问 API 文档
open http://localhost:8000/docs

📚 API 文档

交互式文档

核心 API 路径

认证 (/api/v1/auth)

方法 路径 说明
POST /register 用户注册
POST /login 用户登录
GET /me 获取当前用户信息
POST /password/forgot 忘记密码
POST /password/reset 重置密码

视频剪辑主链路

上传素材 → POST /api/v1/upload(直传)或 /api/v1/upload/chunk/init(分片)
   ↓
创建素材 → POST /api/v1/assets
   ↓
AI 分类 → POST /api/v1/classification-jobs
   ↓
生成剪辑计划 → POST /api/v1/projects/{id}/edit-plans/auto-generate
   ↓
创建生成任务 → POST /api/v1/generation/tasks/
   ↓
查询结果 → GET /api/v1/generation/tasks/{task_id}/results/
   ↓
获取成片 → GET /api/v1/generated-videos/{video_id}/download-url

项目管理 (/api/v1/project-management)

方法 路径 说明
GET/POST /tasks 任务列表/创建
PATCH /tasks/{id} 更新任务信息
PATCH /tasks/{id}/status 更新任务状态
PATCH /tasks/{id}/progress 更新任务进度
GET/POST /milestones 里程碑列表/创建
GET/POST /issues 问题列表/创建
PATCH /issues/{id}/resolve 解决问题

素材与上传

方法 路径 说明
POST /api/v1/upload 直传素材(multipart/form-data)
POST /api/v1/upload/direct/prepare 准备 OSS 直传签名
POST /api/v1/upload/direct/complete 确认直传完成
POST /api/v1/upload/chunk/init 初始化分片上传
POST /api/v1/upload/chunk/{id}/{index} 上传分片
POST /api/v1/upload/chunk/{id}/complete 完成分片上传
GET /api/v1/assets 素材列表
PATCH /api/v1/assets/{id}/review 更新素材审核状态
GET /api/v1/projects/{id}/asset-diagnosis 资产诊断

成片管理 (/api/v1/generated-videos)

方法 路径 说明
GET / 成片列表
GET /{video_id} 成片详情
GET /{video_id}/download-url 下载链接
PATCH /{video_id}/review 审核状态

完整 API 列表请查看 API 主线清单


🏗️ 架构

小虾 SaaS
├── packages/              # 核心业务逻辑(Clean Architecture)
│   ├── domain/           # 领域模型(dataclass)
│   ├── application/      # 用例(Use Cases)
│   ├── ports/            # 接口定义(抽象端口)
│   └── adapters/         # 适配器实现(SQLAlchemy、Redis、SMTP 等)
├── apps/                 # 应用层
│   ├── api/             # FastAPI 应用 + 路由 + Pydantic schemas
│   ├── web/             # React + Vite 前端
│   └── worker/          # Celery 异步任务(视频处理、分类等)
├── migrations/           # Alembic 数据库迁移
├── tests/               # 测试
│   ├── unit/           # 单元测试
│   ├── integration/    # 集成测试
│   └── e2e/            # 端到端测试
└── docs/                # 文档

设计模式:

  • Clean Architecture(依赖方向:外层 → 内层)
  • 依赖注入(FastAPI Depends)
  • Repository 模式(通过 ports 抽象)
  • Domain-Driven Design

🛠️ 技术栈

后端:

  • Python 3.12 + FastAPI 0.115.0
  • PostgreSQL 16(生产)
  • Redis 7(缓存 + Celery Broker)
  • Celery(异步任务:视频处理、素材导入、分类)
  • 阿里云 OSS(文件存储)

前端:

  • React 18 + TypeScript
  • Vite(构建工具)
  • Ant Design(UI 组件)

部署:

  • Docker + Docker Compose
  • Gitea + Gitea Actions(CI/CD)
  • Nginx(反向代理)

🧪 测试

# 运行所有测试
pytest tests/ -v

# 运行单元测试
pytest tests/unit -v

# 运行集成测试
pytest tests/integration -v

# 生成覆盖率报告
pytest --cov=packages --cov-report=html

📊 当前状态

模块 状态
视频剪辑主链路(Phase 7) ✅ 已完成
分片上传(最大 2GB) ✅ 已完成
4 种剪辑模式 ✅ 已完成
项目管理 + 任务追踪 ✅ 已完成
资产诊断 ✅ 已完成
认证系统(JWT) ✅ 已完成
CI/CD 流水线 ✅ 运行中
前端界面(Vite) ✅ 已完成

📄 更多文档


🤝 贡献

欢迎贡献!请查看 贡献指南

仓库地址: https://git.xiaoxiajianji.com/xiaoxia/xiaoxia-saas


License: MIT

S
Description
小虾 SaaS 自动化剪辑系统
Readme MIT 187 MiB
v2.1.0-test.9 Latest
2026-07-09 08:32:46 +08:00
Languages
Python 78.5%
TypeScript 15.8%
CSS 2.8%
Shell 2.7%
Dockerfile 0.1%
Other 0.1%