refactor: 从智影 Fork 重构为智剪，独立 Docker 基础设施，开发模式认证兜底
主要变更： - 修复 /tasks/script 路由 404（去掉重复 prefix） - 开发模式自动认证兜底（无需登录即可测试流程） - Docker 基础设施独立化（共用 db/redis） - 前端 API 端口改为 8081 - 新增 TTS/语音克隆、视频粗剪、音频混音等智剪功能 - 删除智影专属模块（avatar、model_usage、qiniu 上传等）
This commit is contained in:
小鱼开发
@@ -1,48 +1,52 @@
|
||||
<!-- From: /Users/0fun/work/ai-meijiaka/AGENTS.md -->
|
||||
# 美家卡智影 (Meijiaka AI Video) - AI 视频创作平台
|
||||
# 美家卡智剪 (Meijiaka Smart Cut) - AI 视频剪辑桌面应用
|
||||
|
||||
## 项目概述
|
||||
|
||||
美家卡智影是一个 AI 驱动的视频创作桌面应用，采用 **Tauri + React + FastAPI** 混合架构。用户可以通过 AI 生成脚本、创建数字人视频，最终合成完整的营销视频。
|
||||
美家卡智剪是一个 AI 驱动的**短视频剪辑**桌面应用，采用 **Tauri + React + FastAPI** 混合架构。本项目由「美家卡智影」Fork 而来，从"数字人生成"转向"AI 辅助长视频剪辑"方向：用户导入长视频素材，AI 根据脚本自动分镜切割，配合语音克隆/TTS 生成配音，最终合成带字幕的成品短视频。
|
||||
|
||||
### 核心功能
|
||||
### 核心功能流程（6 步）
|
||||
|
||||
- **AI 脚本生成**: 基于 LLM 自动生成视频脚本和分镜
|
||||
- **数字人视频**: 基于 KlingAI 创建数字人视频片段
|
||||
- **字幕生成**: 基于火山引擎豆包语音自动生成字幕并压制到视频
|
||||
- **封面制作**: 提取视频首帧并叠加字幕样式生成封面
|
||||
- **视频合成**: 本地 FFmpeg 处理视频拼接、音频混流、导出成品
|
||||
- **形象克隆**: 基于 KlingAI 的自定义数字人形象管理
|
||||
- **项目管理**: 项目数据本地 JSON 文件存储，认证状态云端同步
|
||||
1. **脚本生成** (Step 1) - AI 生成或粘贴口播文案，自动拆分为带预估时长的分镜
|
||||
2. **视频粗剪** (Step 2) - 导入长视频素材，按脚本时长自动切割为片段
|
||||
3. **语音配音** (Step 3) - AI 声音克隆（KlingAI）+ TTS 合成，为每段生成配音音频
|
||||
4. **字幕压制** (Step 4) - 基于音频自动对齐时间轴，ASS 字幕渲染并压制到视频
|
||||
5. **封面制作** (Step 5) - 提取视频首帧并叠加标题样式生成封面
|
||||
6. **视频合成** (Step 6) - FFmpeg 拼接视频片段，替换原声为 TTS 音频，导出成品
|
||||
|
||||
### 技术架构
|
||||
|
||||
- **前端桌面壳**: Tauri 2.x + React 19 + TypeScript 5.8 + Vite 7
|
||||
- **后端服务**: FastAPI (Python 3.13+) + PostgreSQL 15 + Redis 7
|
||||
- **任务调度**: 自定义 Async Engine（基于 Redis 槽位管理），**非 Celery**
|
||||
- **本地处理**: 嵌入式 FFmpeg，Rust 层封装视频/音频处理命令
|
||||
- **AI 服务**: 火山方舟（LLM/字幕）、可灵 AI（TTS/声音克隆）、OpenAI 兼容接口
|
||||
|
||||
## 项目结构
|
||||
|
||||
```
|
||||
ai-meijiaka/
|
||||
meijiaka-zj/
|
||||
├── python-api/ # FastAPI 后端服务（AI 代理 + 认证 + 任务调度）
|
||||
│ ├── app/
|
||||
│ │ ├── api/v1/ # API 路由 (REST): auth, script, ai_models, klingai,
|
||||
│ │ │ # qiniu, video, avatar, system,
|
||||
│ │ │ # caption, tasks
|
||||
│ │ ├── api/v1/ # API 路由: auth, system, caption, voice
|
||||
│ │ ├── ai/ # AI 模型路由、Provider、提示词模板
|
||||
│ │ ├── core/ # 安全、配置加载、Token管理器、Redis客户端、异常处理
|
||||
│ │ ├── crud/ # 数据访问层（users, model_usage, avatar）
|
||||
│ │ ├── crud/ # 数据访问层（users, avatar）
|
||||
│ │ ├── db/ # 数据库配置（PostgreSQL + asyncpg + SQLAlchemy 2.0）
|
||||
│ │ ├── models/ # SQLAlchemy 模型（users, model_usage_logs, avatars）
|
||||
│ │ ├── models/ # SQLAlchemy 模型（当前仅 users）
|
||||
│ │ ├── schemas/ # Pydantic 校验模型
|
||||
│ │ ├── services/ # AI 服务代理、DTO标准化、七牛/字幕/视频服务
|
||||
│ │ ├── scheduler/ # Async Engine 异步任务调度（video, image, script,
|
||||
│ │ │ # subtitle, copy, avatar_clone）
|
||||
│ │ ├── services/ # AI 服务代理、字幕/视频/TTS/声音克隆服务
|
||||
│ │ ├── scheduler/ # Async Engine 异步任务调度
|
||||
│ │ ├── config.py # Pydantic Settings 配置管理
|
||||
│ │ └── main.py # FastAPI 入口（含生命周期管理）
|
||||
│ ├── config/ # AI 模型配置文件（ai_models.yaml），支持热重载
|
||||
│ ├── alembic/ # 数据库迁移
|
||||
│ ├── scripts/ # 初始化/测试脚本
|
||||
│ ├── alembic/ # 数据库迁移（当前无迁移文件）
|
||||
│ ├── tests/ # 测试目录（当前仅 test_kling_tts.py）
|
||||
│ ├── pyproject.toml # Python 依赖和工具配置
|
||||
│ ├── requirements.lock # uv 锁定依赖版本
|
||||
│ ├── requirements.lock # uv 锁定依赖版本（如缺失需执行 make update-lock）
|
||||
│ ├── Makefile # 常用命令封装
|
||||
│ ├── docker-compose.yml
|
||||
│ └── Dockerfile
|
||||
│ ├── Dockerfile
|
||||
│ └── .env.example # 环境变量模板
|
||||
│
|
||||
├── tauri-app/ # Tauri 桌面应用（业务数据本地存储）
|
||||
│ ├── src/ # React 前端源码
|
||||
@@ -55,6 +59,11 @@ ai-meijiaka/
|
||||
│ │ │ └── ipc.ts # Tauri IPC 调用封装
|
||||
│ │ ├── components/ # 可复用组件
|
||||
│ │ ├── pages/ # 页面组件
|
||||
│ │ │ ├── VideoCreation/ # 6 步视频创作流程
|
||||
│ │ │ ├── ContentManagement/
|
||||
│ │ │ ├── Settings/
|
||||
│ │ │ ├── Profile/
|
||||
│ │ │ └── Login/
|
||||
│ │ ├── store/ # Zustand 状态管理（+ Immer + persist）
|
||||
│ │ ├── hooks/ # 自定义 React Hooks
|
||||
│ │ ├── styles/ # 全局 CSS 变量、主题
|
||||
@@ -65,10 +74,8 @@ ai-meijiaka/
|
||||
│ │ │ ├── ffmpeg_cmd.rs # FFmpeg 命令封装
|
||||
│ │ │ ├── video_processing.rs # 视频合成业务逻辑
|
||||
│ │ │ ├── storage/ # 本地存储引擎（原子写入、文件锁、路径净化）
|
||||
│ │ │ ├── commands/ # IPC 命令按领域拆分（project/asset/auth/avatar）
|
||||
│ │ │ ├── commands/ # IPC 命令按领域拆分
|
||||
│ │ │ ├── api_proxy.rs # Python API 代理转发
|
||||
│ │ │ ├── auth.rs # 认证命令（已迁移至 commands/auth_state.rs）
|
||||
│ │ │ ├── avatar_cache.rs # 头像缓存管理
|
||||
│ │ │ └── utils.rs # 通用工具函数
|
||||
│ │ ├── Cargo.toml
|
||||
│ │ ├── tauri.conf.json
|
||||
@@ -78,19 +85,13 @@ ai-meijiaka/
|
||||
│ ├── tsconfig.json
|
||||
│ └── eslint.config.js
|
||||
│
|
||||
└── docs/ # 项目文档
|
||||
├── anytocopy-api.md
|
||||
├── anytocopy-integration.md
|
||||
├── app-update-system.md
|
||||
├── database-design.md
|
||||
├── kling-api-dev.md
|
||||
├── migrate-avatars-to-local.md
|
||||
├── qiniu-kodo-python-sdk-guide.md
|
||||
├── video-generation-flow.md
|
||||
└── volcengine-video-caption-api.md
|
||||
├── docs/ # 项目文档
|
||||
├── scripts/ # 数据修复/迁移脚本（非部署脚本）
|
||||
├── package.json # 根级 package.json（monorepo 占位）
|
||||
└── AGENTS.md # 本文件
|
||||
```
|
||||
|
||||
## 技术栈
|
||||
## 技术栈详解
|
||||
|
||||
### 后端 (python-api)
|
||||
|
||||
@@ -100,20 +101,22 @@ ai-meijiaka/
|
||||
|------|------|------|------|
|
||||
| Python | - | 3.13+ | 运行环境 |
|
||||
| Web 框架 | FastAPI | 0.116+ | REST API |
|
||||
| 数据库 | PostgreSQL | 15+ | 用户认证 + 成本统计 + 形象管理 |
|
||||
| 数据库 | PostgreSQL | 15+ | 用户认证 |
|
||||
| ORM | SQLAlchemy | 2.0 (异步) | 数据模型 |
|
||||
| 缓存/调度 | Redis + Async Engine | 5.2+ / 自定义 | 异步任务槽位调度 |
|
||||
| AI SDK | OpenAI / volcengine | 1.58+ / 5.0+ | LLM 调用 |
|
||||
| AI SDK | OpenAI / volcengine | 1.58+ / 5.0+ | LLM / 字幕调用 |
|
||||
| TTS/语音 | KlingAI API | - | 语音合成、声音克隆 |
|
||||
| 认证 | python-jose + passlib | 3.4+ / 1.7+ | JWT 认证 |
|
||||
| 对象存储 | qiniu | 7.13+ | 七牛云存储 |
|
||||
| HTTP 客户端 | httpx + aiohttp | 0.28+ / 3.13+ | 异步 HTTP |
|
||||
| 包管理/构建 | uv | - | 虚拟环境、依赖锁定、Docker 构建 |
|
||||
|
||||
**后端架构说明**：
|
||||
- 后端为"轻量云账号 + 全本地业务数据"模式
|
||||
- 云端仅存储：用户账户、形象元数据、成本统计
|
||||
- 业务数据（项目/脚本/媒体）全部本地存储
|
||||
- 云端仅存储：用户账户信息
|
||||
- 业务数据（项目/脚本/媒体/成品）全部本地存储
|
||||
- 任务调度使用**自定义 Async Engine**（基于 Redis 的槽位管理），**非 Celery**
|
||||
- 当前活跃 API 模块仅 4 个：`auth`, `system`, `caption`, `voice`
|
||||
- 调度器已注册 7 个 Handler：`Video`, `Avatar`, `Image`, `Subtitle`, `Copy`, `Script`, `TTS`
|
||||
|
||||
### 前端 (tauri-app)
|
||||
|
||||
@@ -121,26 +124,29 @@ ai-meijiaka/
|
||||
|------|------|------|------|
|
||||
| 桌面框架 | Tauri | 2.x | 桌面应用壳 |
|
||||
| UI 框架 | React | 19.1+ | 用户界面 |
|
||||
| 路由 | React Router DOM | 7.x | 页面路由（主壳使用 NavigationContext） |
|
||||
| 路由 | 自定义 NavigationContext | - | 页面切换（react-router-dom 已安装但未用于主流程） |
|
||||
| 状态管理 | Zustand | 5.x | 全局状态 + Immer 中间件 |
|
||||
| 数据获取 | SWR | 2.x | 请求缓存 |
|
||||
| 虚拟列表 | @tanstack/react-virtual | 3.x | 大数据列表渲染 |
|
||||
| 构建工具 | Vite | 7.x | 构建、开发服务器 |
|
||||
| 测试 | Vitest + @testing-library | 4.x | 单元测试 |
|
||||
| 类型生成 | openapi-typescript | 7.x | 从 OpenAPI 生成 TS 类型 |
|
||||
| 字幕渲染 | assjs | 0.1.x | ASS 字幕前端渲染 |
|
||||
|
||||
### Rust 后端 (src-tauri/src)
|
||||
|
||||
| 模块 | 用途 |
|
||||
|------|------|
|
||||
| lib.rs | Tauri 应用入口，命令注册 |
|
||||
| ffmpeg_cmd.rs | FFmpeg 命令封装（首帧提取、字幕压制、封面合成） |
|
||||
| lib.rs | Tauri 应用入口，命令注册（含 FFmpeg / 音频处理 / 视频合成命令） |
|
||||
| ffmpeg_cmd.rs | FFmpeg 命令封装（首帧提取、字幕压制、封面合成、音频替换/混音/标准化） |
|
||||
| video_processing.rs | 视频合成业务逻辑 |
|
||||
| storage/engine.rs | 本地存储引擎（原子写入、文件锁、路径净化） |
|
||||
| storage/paths.rs | 集中化路径计算 |
|
||||
| commands/project.rs | 项目本地存储 IPC 命令 |
|
||||
| commands/asset.rs | 资源文件保存 IPC 命令 |
|
||||
| commands/auth_state.rs | 认证状态文件持久化 |
|
||||
| commands/voice.rs | 音频文件保存/列表/删除 IPC 命令 |
|
||||
| commands/product.rs | 成品视频保存/列表/删除/重命名 IPC 命令 |
|
||||
| api_proxy.rs | Python API 代理转发 |
|
||||
| avatar_cache.rs | 头像视频缓存管理 |
|
||||
|
||||
@@ -153,6 +159,7 @@ cd python-api
|
||||
|
||||
# 方式一：Docker Compose（推荐）
|
||||
cp .env.example .env
|
||||
# 编辑 .env 填入实际的 AI API Keys
|
||||
docker-compose up -d
|
||||
|
||||
# 方式二：本地开发（若 Docker 不可用）
|
||||
@@ -207,6 +214,7 @@ cd python-api
|
||||
make dev # 安装开发依赖 + pre-commit 钩子
|
||||
make lint # ruff + mypy
|
||||
make format # black + ruff --fix
|
||||
make format-check # 检查格式但不修改
|
||||
make test # pytest
|
||||
make test-cov # 覆盖率报告
|
||||
make security # bandit + pip-audit
|
||||
@@ -214,6 +222,7 @@ make lint-semantic # 语义层禁词检查
|
||||
make ci # 运行所有 CI 检查（format-check + lint + lint-semantic + test + security）
|
||||
make docker-run # Docker Compose 启动全部服务
|
||||
make scheduler # 启动 Async Engine Scheduler
|
||||
make clean # 清理缓存文件
|
||||
|
||||
# 手动命令
|
||||
black app/
|
||||
@@ -234,7 +243,7 @@ print(json.dumps(app.openapi(), indent=2, ensure_ascii=False))
|
||||
" > ../tauri-app/src/api/generated/openapi.json
|
||||
|
||||
# Docker 构建
|
||||
docker build -t meijiaka-api .
|
||||
make docker
|
||||
```
|
||||
|
||||
### Tauri 前端
|
||||
@@ -247,7 +256,7 @@ npm run dev # 纯 Vite 开发（不启动 Tauri）
|
||||
npm run tauri dev # 完整 Tauri 开发模式
|
||||
|
||||
# 构建
|
||||
npm run build # 前端生产构建
|
||||
npm run build # 前端生产构建（tsc + vite build）
|
||||
npm run tauri build # 打包桌面应用
|
||||
|
||||
# 测试
|
||||
@@ -281,10 +290,14 @@ npm run gen:api # 从 OpenAPI 生成 TypeScript 类型
|
||||
- `burn_subtitle` // 字幕压制
|
||||
- `extract_video_first_frame` // 首帧提取
|
||||
- `generate_cover_image` // 封面生成
|
||||
- `replace_audio_track` // 音频替换
|
||||
- `mix_audio_tracks` // 音频混音
|
||||
- `standardize_audio` // 音频标准化
|
||||
- `save_project_meta*` / `load_project_meta*` // 本地文件系统
|
||||
- `save_project_segments*` / `load_project_segments*`
|
||||
- `save_project_asset` / `get_video_save_path` / `get_image_save_path`
|
||||
- `save_final_product`
|
||||
- `save_final_product` / `list_local_products` / `delete_local_product` / `rename_local_product`
|
||||
- `save_audio` / `list_project_audios` / `delete_audio`
|
||||
- 头像缓存相关 API
|
||||
|
||||
**添加新 API 流程**：
|
||||
@@ -303,16 +316,16 @@ app/ai/
|
||||
│ ├── base.py # Provider 抽象基类
|
||||
│ ├── generic_llm_provider.py # 通用 OpenAI 兼容 Provider
|
||||
│ ├── volcengine_provider.py # 火山方舟官方 SDK
|
||||
│ └── klingai_provider.py # KlingAI 数字人
|
||||
│ └── klingai_provider.py # KlingAI（可灵 AI）
|
||||
└── prompts/ # 提示词模板（禁止硬编码）
|
||||
```
|
||||
|
||||
支持的 AI 平台：
|
||||
- **火山方舟** (字节跳动) - 推荐，性价比高
|
||||
- **OpenAI** - GPT 系列
|
||||
- **文心一言** (百度)
|
||||
- **通义千问** (阿里云)
|
||||
- **可灵 AI** (快手) - 视频生成、数字人、形象克隆
|
||||
- **火山方舟** (字节跳动) - LLM、字幕服务
|
||||
- **OpenAI** - GPT 系列（可选）
|
||||
- **可灵 AI** (快手) - TTS 语音合成、声音克隆
|
||||
- **文心一言** (百度) - 可选
|
||||
- **通义千问** (阿里云) - 可选
|
||||
|
||||
AI 模型配置位于 `python-api/config/ai_models.yaml`，支持热重载，无需重启服务即可更新模型配置。
|
||||
|
||||
@@ -340,7 +353,8 @@ API (POST /tasks/{type}) → Redis JobRegistry → AsyncEngine tick loop → Han
|
||||
| ScriptHandler | 10 | `script:slots` | LLM 脚本生成（含 AnyToCopy 视频文案提取） |
|
||||
| SubtitleHandler | 5 | `volc:subtitle_slots` | 火山引擎字幕/自动对齐 |
|
||||
| CopyHandler | 5 | `anytocopy:slots` | AnyToCopy 视频文案提取 |
|
||||
| AvatarHandler | 2 | `kling:avatar_slots` | Kling 形象克隆（状态机: pending→voice_processing→element_pending→element_processing→succeed） |
|
||||
| TTSHandler | - | `kling:tts_slots` | Kling TTS 语音合成 |
|
||||
| AvatarHandler | 2 | `kling:avatar_slots` | Kling 形象克隆 |
|
||||
|
||||
### TokenManager（API 认证 Token 管理）
|
||||
|
||||
@@ -382,24 +396,25 @@ Rust 层实现了 defense-in-depth 的本地存储系统：`src-tauri/src/storag
|
||||
- **`paths.rs`**: 集中路径计算
|
||||
- `~/Documents/Meijiaka/projects/{id}/` (meta.json, segments.json, assets/)
|
||||
- `~/Documents/Meijiaka/products/`
|
||||
- `~/Documents/Meijiaka/avatars.json`
|
||||
- `{app_config_dir}/auth.json`
|
||||
- `{app_data_dir}/avatars/`
|
||||
|
||||
**所有本地 JSON 读写必须经过 StorageEngine，禁止在命令处理器中直接调用 `fs::write`**。
|
||||
|
||||
### 数据库模型
|
||||
|
||||
后端仅保留 **3 个表**：
|
||||
后端当前仅保留 **1 个活跃模型**：
|
||||
|
||||
```
|
||||
users -- 用户账户信息（mobile, nickname, avatar_url）
|
||||
model_usage_logs -- 大模型调用记录（token, 成本, 响应时间）
|
||||
avatars -- 克隆形象元数据（云端备份，前端已迁移至本地 JSON）
|
||||
```
|
||||
|
||||
> 注：`avatars` 和 `model_usage_logs` 相关模型及 CRUD 在最近的重构中已移除或清理。`crud/avatar.py` 仍存在但可能处于未使用状态。
|
||||
|
||||
**业务数据本地存储**：
|
||||
- 项目/脚本/分镜 → 前端本地 JSON 文件（`~/Documents/Meijiaka/projects/`）
|
||||
- 音频/视频/图片文件 → 本地磁盘
|
||||
- 成品视频 → `~/Documents/Meijiaka/products/`
|
||||
- 用户配置 → localStorage（少量 UI 状态）
|
||||
|
||||
### 数据流规范
|
||||
@@ -408,6 +423,8 @@ avatars -- 克隆形象元数据（云端备份，前端已迁移至
|
||||
用户输入主题 ──→ 后端 AI 生成脚本 ──→ 后端返回分镜列表 ──→ 前端保存到本地
|
||||
│ │
|
||||
└────────────────── 后端不存储脚本数据 ────────────────────────┘
|
||||
|
||||
用户导入长视频 ──→ Rust 本地切割 ──→ 生成分段视频 ──→ 前端保存到本地
|
||||
```
|
||||
|
||||
### 本地存储结构
|
||||
@@ -419,6 +436,9 @@ avatars -- 克隆形象元数据（云端备份，前端已迁移至
|
||||
│ └── {project_id}/
|
||||
│ ├── meta.json # 项目元数据
|
||||
│ ├── segments.json # 分镜数据
|
||||
│ ├── media/ # 导入的原始素材
|
||||
│ ├── shots/ # 自动切割后的片段
|
||||
│ ├── audio/ # TTS 生成的音频
|
||||
│ └── assets/ # 资源文件（封面、成品等）
|
||||
├── products/ # 成品视频目录
|
||||
├── avatars.json # 形象列表（本地）
|
||||
@@ -427,10 +447,9 @@ avatars -- 克隆形象元数据（云端备份，前端已迁移至
|
||||
|
||||
项目元数据 `meta.json` 关键字段：
|
||||
- `id`, `title`, `topic`, `status` (draft | published)
|
||||
- `currentStep`: 1=脚本生成, 2=形象视频, 3=字幕压制, 4=封面制作, 5=视频合成
|
||||
- `currentStep`: 1=脚本生成, 2=视频粗剪, 3=语音配音, 4=字幕压制, 5=封面制作, 6=视频合成
|
||||
- `createdAt`, `updatedAt`, `exportedAt`
|
||||
- `coverPath`, `finalVideoPath`
|
||||
- `selectedElementId`, `selectedHumanId`
|
||||
- `coverConfig`, `scriptDuration`, `scriptType`
|
||||
|
||||
分镜数据 `segments.json` 字段：
|
||||
@@ -442,9 +461,16 @@ avatars -- 克隆形象元数据（云端备份，前端已迁移至
|
||||
|
||||
主应用壳使用 **自定义 NavigationContext**（React Context）实现页面切换，映射 `Record<PageType, ComponentType>`。`react-router-dom` 已安装但主要用于未来扩展或特定路由场景，当前主流程不使用 BrowserRouter 进行导航。
|
||||
|
||||
页面结构：
|
||||
- `video-creation` - 6 步视频创作流程
|
||||
- `avatar-clone` - 形象/声音克隆管理
|
||||
- `my-works` - 我的作品
|
||||
- `profile` - 个人中心
|
||||
- `settings-*` - 设置相关页面
|
||||
|
||||
### 状态管理
|
||||
|
||||
六個專門的 Zustand store：
|
||||
七个专门的 Zustand store：
|
||||
|
||||
| Store | 职责 | 持久化 |
|
||||
|-------|------|--------|
|
||||
@@ -454,6 +480,7 @@ avatars -- 克隆形象元数据（云端备份，前端已迁移至
|
||||
| `uiStore` | Toast 通知队列 | 无 |
|
||||
| `progressStore` | 全局进度模态框 | 无 |
|
||||
| `settingsStore` | 主题模式、用户偏好 | localStorage |
|
||||
| `voiceStore` | 语音/形象选择状态 | 无 |
|
||||
|
||||
`projectStore` **不自动保存**。数据在显式过渡点持久化到磁盘（如进入 step 2、调用 `setFinalVideoPath` 时触发 `saveMetaToLocalFile`）。`saveMetaToLocalFile()` 通过 Promise 链串行化写入，避免并发覆盖。
|
||||
|
||||
@@ -578,6 +605,7 @@ Database Layer (db/*.py)
|
||||
- **文档**: 中文注释，Google Style Docstrings
|
||||
- **安全**: Bandit + pip-audit
|
||||
- **Git Hooks**: pre-commit（Black、Ruff、uv lock 同步检查）
|
||||
- **忽略规则**: Ruff 忽略 `E501, E402, N802, N803, N806, N815, B008, B904`
|
||||
|
||||
### TypeScript/React
|
||||
|
||||
@@ -586,8 +614,9 @@ Database Layer (db/*.py)
|
||||
- **状态**: Zustand 管理全局状态（配合 Immer 处理不可变更新）
|
||||
- **样式**: 普通 CSS + CSS 变量（`tauri-app/src/styles/variables.css`）
|
||||
- **ESLint**: 使用 `eslint.config.js`（Flat Config），含 React Hooks 和 React Refresh 规则
|
||||
- **Prettier**: semi=true, singleQuote=true, tabWidth=2, printWidth=100
|
||||
- **Prettier**: semi=true, singleQuote=true, tabWidth=2, printWidth=100, trailingComma=es5
|
||||
- **Stylelint**: `stylelint-config-standard`，禁止 magic px 用于 `border-radius` 和 `font-size`
|
||||
- **ESLint 规则**: `no-console` 为 warn（允许 warn/error/info），`eqeqeq` 为 warn，`no-var` 为 error
|
||||
|
||||
### Rust
|
||||
|
||||
@@ -623,8 +652,9 @@ pytest --cov=app --cov-report=html --cov-report=term
|
||||
**测试配置** (`pyproject.toml`):
|
||||
- asyncio_mode = "auto"
|
||||
- 测试文件命名: `test_*.py`
|
||||
- 测试路径: `tests/`
|
||||
|
||||
> **注**：当前项目中 `python-api/tests/` 目录尚未创建，后端测试待补充。
|
||||
> **注**：当前后端测试覆盖非常有限，仅 `tests/test_kling_tts.py` 一个测试文件，包含 `TTSService`、`VoiceCloneService` 和状态枚举的单元测试。大量模块暂无测试。
|
||||
|
||||
### 前端测试
|
||||
|
||||
@@ -642,11 +672,13 @@ npm run test:coverage
|
||||
```
|
||||
|
||||
**测试配置**:
|
||||
- 测试框架: Vitest 4.x + @testing-library/react + jsdom
|
||||
- 测试框架: Vitest 4.x + @testing-library/react 16.3.x + jsdom
|
||||
- 测试文件: `src/**/*.test.ts(x)`
|
||||
- Mock 配置: `src/__tests__/setup.ts`
|
||||
- 自动 Mock: localStorage, Tauri API (`@tauri-apps/api/core`)
|
||||
- 示例测试: `src/store/__tests__/authStore.test.tsx`
|
||||
- 现有测试:
|
||||
- `src/store/__tests__/authStore.test.tsx`
|
||||
- `src/store/__tests__/settingsStore.test.tsx`
|
||||
|
||||
## 安全注意事项
|
||||
|
||||
@@ -658,7 +690,7 @@ npm run test:coverage
|
||||
6. **路径遍历**: Rust StorageEngine 的 `sanitize_id()` 和 `sanitize_filename()` 防御路径遍历攻击
|
||||
7. **原子写入**: 所有本地 JSON 使用 `atomic_write_json`（先写 `.tmp` 再 `rename`）
|
||||
8. **文件锁**: 并发 RMW 操作使用 `with_file_lock` 防止竞态
|
||||
9. **日志**: 后端日志写入 `~/Documents/Meijiaka/logs/api_YYYYMMDD.log`
|
||||
9. **日志**: 后端日志写入 `~/Documents/Meijiaka-zj/logs/api_YYYYMMDD.log`
|
||||
|
||||
## 配置说明
|
||||
|
||||
@@ -701,8 +733,8 @@ CORS_ORIGINS=http://localhost:1420,http://127.0.0.1:1420,http://localhost:8080
|
||||
|
||||
```json
|
||||
{
|
||||
"productName": "美家卡智影",
|
||||
"identifier": "cn.meijiaka.ai-video",
|
||||
"productName": "美家卡智剪",
|
||||
"identifier": "cn.meijiaka.ai-jian",
|
||||
"build": {
|
||||
"devUrl": "http://localhost:1420",
|
||||
"frontendDist": "../dist"
|
||||
@@ -721,16 +753,17 @@ CORS_ORIGINS=http://localhost:1420,http://127.0.0.1:1420,http://localhost:8080
|
||||
模型配置文件支持热重载，无需重启服务即可更新模型配置。主要配置项：
|
||||
|
||||
- **platforms**: AI 平台配置（mock, volcengine, klingai）
|
||||
- **models**: 可用模型列表及其能力标签 [script, polish, chat, image, embedding, vision]
|
||||
- **models**: 可用模型列表及其能力标签 [script, polish, chat, image, video_generation, image2video, lip_sync, image_generation]
|
||||
- **task_defaults**: 任务类型到模型的默认映射
|
||||
|
||||
## 视频创作流程
|
||||
|
||||
1. **脚本生成** (Step 1) - AI 生成视频脚本和分镜
|
||||
2. **形象视频** (Step 2) - 选择数字人形象，生成视频片段
|
||||
3. **字幕压制** (Step 3) - 生成字幕并压制到视频中
|
||||
4. **封面制作** (Step 4) - 生成视频封面
|
||||
5. **视频合成** (Step 5) - FFmpeg 拼接视频片段，导出最终视频
|
||||
1. **脚本生成** (Step 1) - AI 生成视频脚本和分镜，或用户直接粘贴文案
|
||||
2. **视频粗剪** (Step 2) - 导入长视频，按脚本时长自动切割为片段
|
||||
3. **语音配音** (Step 3) - 声音克隆 + TTS 生成每段配音
|
||||
4. **字幕压制** (Step 4) - 生成字幕并压制到视频中
|
||||
5. **封面制作** (Step 5) - 生成视频封面
|
||||
6. **视频合成** (Step 6) - FFmpeg 拼接视频片段，替换原声为 TTS 音频，导出最终视频
|
||||
|
||||
## 常见问题
|
||||
|
||||
@@ -778,7 +811,12 @@ Tauri 应用已嵌入 FFmpeg 二进制文件：
|
||||
- 每个 Handler 实现 `AsyncHandler` 接口，状态机驱动任务生命周期
|
||||
- 优势：更细粒度的并发控制、统一状态机、无 Celery 依赖
|
||||
|
||||
### Q: 为什么后端某些 API 模块不存在？
|
||||
|
||||
当前项目处于从「美家卡智影」向「美家卡智剪」的重构过程中。`app/api/v1/router.py` 当前仅注册了 4 个模块（`auth`, `system`, `caption`, `voice`），历史上存在的 `script`, `video`, `klingai`, `qiniu`, `ai_models`, `tasks` 等路由当前未激活。如需使用相关功能，需在路由中重新引入。
|
||||
|
||||
---
|
||||
|
||||
**最后更新**: 2026-04-17
|
||||
**最后更新**: 2026-04-21
|
||||
**架构模式**: 单机版（轻量云账号 + 全本地业务数据）
|
||||
**项目状态**: 从 ai-meijiaka Fork 后的活跃重构中
|
||||
|
||||
Generated
+6
@@ -0,0 +1,6 @@
|
||||
{
|
||||
"name": "meijiaka-zj",
|
||||
"lockfileVersion": 3,
|
||||
"requires": true,
|
||||
"packages": {}
|
||||
}
|
||||
@@ -0,0 +1 @@
|
||||
{}
|
||||
@@ -11,12 +11,12 @@ HOST=0.0.0.0
|
||||
PORT=8000
|
||||
|
||||
# === 数据库配置 ===
|
||||
DATABASE_URL=postgresql+asyncpg://postgres:postgres@localhost:5432/meijiaka
|
||||
DATABASE_URL=postgresql+asyncpg://postgres:postgres@localhost:5432/meijiaka_zj
|
||||
|
||||
# === Redis 配置 ===
|
||||
REDIS_HOST=localhost
|
||||
REDIS_PORT=6379
|
||||
REDIS_DB=0
|
||||
REDIS_DB=1
|
||||
# REDIS_PASSWORD= # 如无密码请留空或注释
|
||||
|
||||
# === JWT 安全配置 ===
|
||||
|
||||
@@ -1,6 +1,6 @@
|
||||
# 美家卡智影 API
|
||||
# 美家卡智剪 API
|
||||
|
||||
美家卡智影后端服务 - 基于 FastAPI + PostgreSQL + Redis 的 AI 视频创作 API。
|
||||
美家卡智剪后端服务 - 基于 FastAPI + Redis 的 AI 视频创作 API。
|
||||
|
||||
## 技术栈
|
||||
|
||||
|
||||
@@ -20,8 +20,6 @@ load_dotenv()
|
||||
|
||||
# 导入模型
|
||||
from app.db.session import Base
|
||||
from app.models.avatar import Avatar # noqa
|
||||
from app.models.model_usage import ModelUsageLog # noqa
|
||||
from app.models.user import User # noqa
|
||||
|
||||
# this is the Alembic Config object
|
||||
|
||||
@@ -1,106 +0,0 @@
|
||||
"""rename_avatar_vendor_fields_add_provider
|
||||
|
||||
Revision ID: 451756e6a43e
|
||||
Revises: d4bd9ad91607
|
||||
Create Date: 2026-04-17 12:00:00.000000
|
||||
|
||||
"""
|
||||
|
||||
from collections.abc import Sequence
|
||||
|
||||
import sqlalchemy as sa
|
||||
|
||||
from alembic import op
|
||||
|
||||
# revision identifiers, used by Alembic.
|
||||
revision: str = "451756e6a43e"
|
||||
down_revision: str | Sequence[str] | None = "d4bd9ad91607"
|
||||
branch_labels: str | Sequence[str] | None = None
|
||||
depends_on: str | Sequence[str] | None = None
|
||||
|
||||
|
||||
def upgrade() -> None:
|
||||
"""Upgrade schema."""
|
||||
# Add provider column with default "kling"
|
||||
op.add_column(
|
||||
"avatars",
|
||||
sa.Column(
|
||||
"provider",
|
||||
sa.String(length=32),
|
||||
nullable=False,
|
||||
server_default="kling",
|
||||
comment="供应商标识",
|
||||
),
|
||||
)
|
||||
|
||||
# Rename element_id -> provider_element_id
|
||||
op.alter_column(
|
||||
"avatars",
|
||||
"element_id",
|
||||
new_column_name="provider_element_id",
|
||||
existing_type=sa.BigInteger(),
|
||||
existing_nullable=True,
|
||||
)
|
||||
|
||||
# Rename voice_task_id -> provider_voice_job_id
|
||||
op.alter_column(
|
||||
"avatars",
|
||||
"voice_task_id",
|
||||
new_column_name="provider_voice_job_id",
|
||||
existing_type=sa.String(length=128),
|
||||
existing_nullable=True,
|
||||
)
|
||||
|
||||
# Rename element_task_id -> provider_element_job_id
|
||||
op.alter_column(
|
||||
"avatars",
|
||||
"element_task_id",
|
||||
new_column_name="provider_element_job_id",
|
||||
existing_type=sa.String(length=128),
|
||||
existing_nullable=True,
|
||||
)
|
||||
|
||||
# Rename indexes
|
||||
op.drop_index("ix_avatars_voice_task_id", table_name="avatars")
|
||||
op.drop_index("ix_avatars_element_task_id", table_name="avatars")
|
||||
op.create_index(
|
||||
"ix_avatars_provider_voice_job_id", "avatars", ["provider_voice_job_id"], unique=False
|
||||
)
|
||||
op.create_index(
|
||||
"ix_avatars_provider_element_job_id", "avatars", ["provider_element_job_id"], unique=False
|
||||
)
|
||||
|
||||
|
||||
def downgrade() -> None:
|
||||
"""Downgrade schema."""
|
||||
# Rename indexes back
|
||||
op.drop_index("ix_avatars_provider_element_job_id", table_name="avatars")
|
||||
op.drop_index("ix_avatars_provider_voice_job_id", table_name="avatars")
|
||||
op.create_index("ix_avatars_element_task_id", "avatars", ["element_task_id"], unique=False)
|
||||
op.create_index("ix_avatars_voice_task_id", "avatars", ["voice_task_id"], unique=False)
|
||||
|
||||
# Rename columns back
|
||||
op.alter_column(
|
||||
"avatars",
|
||||
"provider_element_job_id",
|
||||
new_column_name="element_task_id",
|
||||
existing_type=sa.String(length=128),
|
||||
existing_nullable=True,
|
||||
)
|
||||
op.alter_column(
|
||||
"avatars",
|
||||
"provider_voice_job_id",
|
||||
new_column_name="voice_task_id",
|
||||
existing_type=sa.String(length=128),
|
||||
existing_nullable=True,
|
||||
)
|
||||
op.alter_column(
|
||||
"avatars",
|
||||
"provider_element_id",
|
||||
new_column_name="element_id",
|
||||
existing_type=sa.BigInteger(),
|
||||
existing_nullable=True,
|
||||
)
|
||||
|
||||
# Drop provider column
|
||||
op.drop_column("avatars", "provider")
|
||||
@@ -1,55 +0,0 @@
|
||||
"""add avatars table
|
||||
|
||||
Revision ID: d4bd9ad91607
|
||||
Revises: fb1be66e804a
|
||||
Create Date: 2026-04-06 21:51:36.225361
|
||||
|
||||
"""
|
||||
from typing import Sequence, Union
|
||||
|
||||
from alembic import op
|
||||
import sqlalchemy as sa
|
||||
|
||||
|
||||
# revision identifiers, used by Alembic.
|
||||
revision: str = 'd4bd9ad91607'
|
||||
down_revision: Union[str, Sequence[str], None] = 'fb1be66e804a'
|
||||
branch_labels: Union[str, Sequence[str], None] = None
|
||||
depends_on: Union[str, Sequence[str], None] = None
|
||||
|
||||
|
||||
def upgrade() -> None:
|
||||
"""Upgrade schema."""
|
||||
# ### commands auto generated by Alembic - please adjust! ###
|
||||
op.create_table('avatars',
|
||||
sa.Column('id', sa.String(length=64), nullable=False, comment='形象唯一标识（Kling element_id 字符串）'),
|
||||
sa.Column('user_id', sa.String(length=36), nullable=False, comment='关联用户 ID'),
|
||||
sa.Column('name', sa.String(length=64), nullable=False, comment='形象展示名称'),
|
||||
sa.Column('voice_id', sa.String(length=64), nullable=True, comment='Kling 自定义音色 ID'),
|
||||
sa.Column('element_id', sa.BigInteger(), nullable=True, comment='Kling 主体 ID'),
|
||||
sa.Column('voice_task_id', sa.String(length=128), nullable=True, comment='Kling 自定义音色任务 ID'),
|
||||
sa.Column('element_task_id', sa.String(length=128), nullable=True, comment='Kling 主体创建任务 ID'),
|
||||
sa.Column('video_url', sa.Text(), nullable=False, comment='原始人物视频 URL'),
|
||||
sa.Column('trial_url', sa.Text(), nullable=True, comment='音色试听音频 URL'),
|
||||
sa.Column('status', sa.String(length=32), nullable=False, comment='状态: pending/voice_processing/voice_failed/element_processing/element_failed/succeed/timeout'),
|
||||
sa.Column('fail_reason', sa.Text(), nullable=True, comment='失败原因（中文可读）'),
|
||||
sa.Column('deleted_at', sa.DateTime(timezone=True), nullable=True, comment='软删除时间，NULL 表示未删除'),
|
||||
sa.Column('created_at', sa.DateTime(timezone=True), nullable=False, comment='记录创建时间'),
|
||||
sa.Column('updated_at', sa.DateTime(timezone=True), nullable=False, comment='记录更新时间'),
|
||||
sa.ForeignKeyConstraint(['user_id'], ['users.id'], ondelete='CASCADE'),
|
||||
sa.PrimaryKeyConstraint('id')
|
||||
)
|
||||
op.create_index(op.f('ix_avatars_element_task_id'), 'avatars', ['element_task_id'], unique=False)
|
||||
op.create_index(op.f('ix_avatars_user_id'), 'avatars', ['user_id'], unique=False)
|
||||
op.create_index(op.f('ix_avatars_voice_task_id'), 'avatars', ['voice_task_id'], unique=False)
|
||||
# ### end Alembic commands ###
|
||||
|
||||
|
||||
def downgrade() -> None:
|
||||
"""Downgrade schema."""
|
||||
# ### commands auto generated by Alembic - please adjust! ###
|
||||
op.drop_index(op.f('ix_avatars_voice_task_id'), table_name='avatars')
|
||||
op.drop_index(op.f('ix_avatars_user_id'), table_name='avatars')
|
||||
op.drop_index(op.f('ix_avatars_element_task_id'), table_name='avatars')
|
||||
op.drop_table('avatars')
|
||||
# ### end Alembic commands ###
|
||||
@@ -1,38 +0,0 @@
|
||||
"""replace device_id with mobile in users table
|
||||
|
||||
Revision ID: fb1be66e804a
|
||||
Revises:
|
||||
Create Date: 2026-04-03 10:22:30.465704
|
||||
|
||||
"""
|
||||
from typing import Sequence, Union
|
||||
|
||||
from alembic import op
|
||||
import sqlalchemy as sa
|
||||
|
||||
|
||||
# revision identifiers, used by Alembic.
|
||||
revision: str = 'fb1be66e804a'
|
||||
down_revision: Union[str, Sequence[str], None] = None
|
||||
branch_labels: Union[str, Sequence[str], None] = None
|
||||
depends_on: Union[str, Sequence[str], None] = None
|
||||
|
||||
|
||||
def upgrade() -> None:
|
||||
"""Upgrade schema."""
|
||||
# ### commands auto generated by Alembic - please adjust! ###
|
||||
op.add_column('users', sa.Column('mobile', sa.String(length=20), nullable=False, comment='手机号'))
|
||||
op.drop_index(op.f('ix_users_device_id'), table_name='users')
|
||||
op.create_index(op.f('ix_users_mobile'), 'users', ['mobile'], unique=True)
|
||||
op.drop_column('users', 'device_id')
|
||||
# ### end Alembic commands ###
|
||||
|
||||
|
||||
def downgrade() -> None:
|
||||
"""Downgrade schema."""
|
||||
# ### commands auto generated by Alembic - please adjust! ###
|
||||
op.add_column('users', sa.Column('device_id', sa.VARCHAR(length=64), autoincrement=False, nullable=False, comment='设备唯一标识'))
|
||||
op.drop_index(op.f('ix_users_mobile'), table_name='users')
|
||||
op.create_index(op.f('ix_users_device_id'), 'users', ['device_id'], unique=True)
|
||||
op.drop_column('users', 'mobile')
|
||||
# ### end Alembic commands ###
|
||||
@@ -24,11 +24,14 @@ from .loader import (
|
||||
VIDEO_STYLES,
|
||||
PolishPromptBuilder,
|
||||
ScriptPromptBuilder,
|
||||
TOPIC_PROMPT_MAP,
|
||||
load_polish_scene,
|
||||
load_polish_voiceover,
|
||||
load_prompt,
|
||||
load_script_system,
|
||||
load_script_user,
|
||||
load_script_user_prompt,
|
||||
load_topic_prompt,
|
||||
render_template,
|
||||
)
|
||||
|
||||
@@ -37,6 +40,9 @@ __all__ = [
|
||||
"render_template",
|
||||
"load_script_system",
|
||||
"load_script_user",
|
||||
"load_script_user_prompt",
|
||||
"load_topic_prompt",
|
||||
"TOPIC_PROMPT_MAP",
|
||||
"load_polish_scene",
|
||||
"load_polish_voiceover",
|
||||
"ScriptPromptBuilder",
|
||||
|
||||
@@ -76,6 +76,58 @@ SCRIPT_TYPES = [
|
||||
{"id": "测评型", "name": "测评型", "description": "产品测评、真实体验"},
|
||||
]
|
||||
|
||||
|
||||
# 创作主题与提示词映射（避坑系列）
|
||||
TOPIC_PROMPT_MAP = {
|
||||
"装修合同避坑": "system/bk-ht",
|
||||
"装修全流程避坑": "system/bk-lc",
|
||||
"装修材料避坑": "system/bk-cl",
|
||||
"装修报价避坑": "system/bk-bj",
|
||||
"全屋定制避坑": "system/bk-qw",
|
||||
"装修常见问题": "system/bk-wt",
|
||||
}
|
||||
|
||||
|
||||
def load_topic_prompt(topic: str) -> str:
|
||||
"""
|
||||
根据创作主题加载对应的 System Prompt
|
||||
|
||||
Args:
|
||||
topic: 创作主题名称，如 "装修合同避坑"
|
||||
|
||||
Returns:
|
||||
提示词内容，未找到返回空字符串
|
||||
"""
|
||||
prompt_path = TOPIC_PROMPT_MAP.get(topic)
|
||||
if prompt_path:
|
||||
return load_prompt(prompt_path)
|
||||
return ""
|
||||
|
||||
|
||||
def load_script_user_prompt(
|
||||
topic: str,
|
||||
duration: int,
|
||||
extra_params: str | None = None,
|
||||
) -> str:
|
||||
"""
|
||||
加载并渲染脚本生成 User Prompt
|
||||
|
||||
Args:
|
||||
topic: 创作主题名称
|
||||
duration: 视频时长（秒）
|
||||
extra_params: 额外参数（如风格、人设等），以换行分隔的字符串
|
||||
|
||||
Returns:
|
||||
渲染后的用户提示词
|
||||
"""
|
||||
template = load_prompt("user/script")
|
||||
return render_template(
|
||||
template,
|
||||
topic=topic,
|
||||
duration=duration,
|
||||
extra_params=extra_params or "",
|
||||
)
|
||||
|
||||
VIDEO_STYLES = [
|
||||
{"id": "口播", "name": "口播", "description": "真人出镜讲解"},
|
||||
{"id": "图文", "name": "图文", "description": "图片+文字+配音"},
|
||||
|
||||
@@ -1,96 +0,0 @@
|
||||
你是一位专业的【口播类短视频】脚本创作专家，专注于家装/装修领域的抖音/视频号口播内容创作。
|
||||
|
||||
【平台适配要求】
|
||||
1. 竖屏拍摄（9:16比例），画面构图以人物为主体
|
||||
2. 台词口语化、接地气，像跟朋友聊天，避免"综上所述""研究表明"等书面语
|
||||
3. 语速稍快有节奏感，每句15-25字，一口气说完不换气，不拖沓
|
||||
4. 避免专业术语堆砌，用业主听得懂的大白话
|
||||
5. 符合新媒体用户观看习惯：3秒定生死，节奏紧凑
|
||||
|
||||
【画面描述标准 - 人物为主，环境为辅】
|
||||
画面描述以【人物状态、表情、动作、情绪】为主。
|
||||
不要写"镜头推近""特写""中景"等摄影术语。
|
||||
每句画面描述控制在 50-70 字，确保有足够细节用于 AI 视频生成。
|
||||
|
||||
❌ 差的示例：
|
||||
"中景竖屏，主播站在毛坯房中央，背景是一面待装修的空白墙面，自然光从右侧窗户照入，主播表情真诚略带焦急，直视镜头说话。"
|
||||
（问题：太多环境描写，太多镜头术语）
|
||||
|
||||
✅ 好的示例：
|
||||
"主播站在空旷的毛坯房里，右手拿着黄色卷尺，他缓缓抬头，表情严肃地看向你，身后是未装修的水泥墙面，神态专业务实。"
|
||||
（聚焦人物：在哪、拿什么、什么表情、看什么）
|
||||
|
||||
【黄金3秒法则 - 开场必须抓眼】
|
||||
- 杜绝铺垫！不要"大家好我是XX""今天给大家讲个事"
|
||||
- 直接击中业主痛点或好奇心，让手指停不下来
|
||||
- 钩子示例：
|
||||
* "装修被坑了8万的业主，昨天来找我哭诉..."
|
||||
* "为什么同样的户型，你家装修比别人贵5万？"
|
||||
* "停！先别急着签合同，这条视频能救你3万块钱"
|
||||
* "每年都有500位业主找我装修，只因为我说透了这一点..."
|
||||
|
||||
【中间内容要求 - 降低跳出率】
|
||||
- 有干货：给出具体数字、方法、避坑点
|
||||
- 有冲突：制造认知反差或情绪起伏
|
||||
- 有看点：适当加入真实案例、现场画面
|
||||
- 避免空洞：不说"我们专业靠谱"，而是"我做了12年装修，见过387个踩坑案例..."
|
||||
|
||||
【最后7秒 - 留资引导（必须可落地）】
|
||||
- 必须有明确、可执行的动作指令
|
||||
- 给业主一个无法拒绝的理由（免费、限时、专属）
|
||||
- 示例话术：
|
||||
* "评论区扣'装修报价'，免费领本地3套装修方案+精准报价单"
|
||||
* "私信'装修'两个字，预约设计师免费上门量房、出平面布局图"
|
||||
* "点击左下角小风车，一键获取你家专属装修预算，绝无隐形消费"
|
||||
* "前20名扣1的业主，送全屋水电VR存档，后期维修不砸墙"
|
||||
- ❌ 杜绝空泛引导："需要装修的联系我们""想了解的私信我"
|
||||
|
||||
【分镜使用原则】
|
||||
- 分镜（segment）用于"主播”出镜的镜头
|
||||
- 【重要】分镜之间要保证画面的连贯性
|
||||
- 分镜 scene 示例：
|
||||
"主播缓缓竖起第三根手指，嘴角扬起一抹了然的笑意。他身体微微前倾，目光柔和地看向前方，仿佛正与屏幕对面的人分享一个轻松的秘密。手指在空中短暂停留，带着从容的节奏。"
|
||||
|
||||
【脚本类型说明】
|
||||
- 对比型：前后反差，制造冲击
|
||||
- 恐吓型：直击痛点，先吓再给解药
|
||||
- 干货型：输出实用方法，建立专业度
|
||||
- 共情型：说业主想说的话，引发共鸣
|
||||
- 挑战型：设定目标，增加悬念
|
||||
- 福利型：用福利钩子吸引停留和留资
|
||||
|
||||
【镜头数量参考】
|
||||
- 30秒短视频：5-7个分镜
|
||||
- 45秒短视频：7-9个分镜
|
||||
- 60秒短视频：10-12个分镜
|
||||
- 75秒短视频：12-15个分镜
|
||||
- 每个分镜时长不得少于3秒
|
||||
- 实际总时长不与用户所选差距超过3秒
|
||||
|
||||
【输出格式要求】
|
||||
请以 JSON 数组格式输出，每个元素包含：
|
||||
- id: 序号（从 1 开始）
|
||||
- type: "segment"（主播口播出镜）
|
||||
- scene: 画面描述（分镜聚焦人物：在哪、干什么、什么表情，什么动作，什么情绪，涉及道具不要出现掏出、拿出这类的动作，不要出现文字，不写镜头术语，不写环境细节；空镜聚焦场景、事物、氛围、环境；）
|
||||
- voiceover: 配音文案（必填，口语化15-25字/句）
|
||||
- duration: 时长（如 "5s"）
|
||||
|
||||
【示例】
|
||||
[
|
||||
{
|
||||
"id": 1,
|
||||
"type": "segment",
|
||||
"scene": "主播缓缓竖起第三根手指，嘴角扬起一抹了然的笑意。他身体微微前倾，目光柔和地看向前方，仿佛正与屏幕对面的人分享一个轻松的秘密。手指在空中短暂停留，带着从容的节奏。",
|
||||
"voiceover": "装修被坑了8万的业主，昨天来找我哭诉...",
|
||||
"duration": "5s"
|
||||
},
|
||||
{
|
||||
"id": 2,
|
||||
"type": "segment",
|
||||
"scene": "主播竖起第二根手指，眉头微皱，嘴角向下撇，眼神中带着一丝不满与无奈。他身体微微前倾，仿佛正对着镜头对面的观众倾诉，手指随着说话轻轻晃动，像是细数着那些令人头疼的业主经历。",
|
||||
"voiceover": "第一个坑，水电改造。很多人图便宜找游击队，结果漏水漏电！",
|
||||
"duration": "8s"
|
||||
}
|
||||
]
|
||||
|
||||
注意：只输出纯 JSON，不要包含 markdown 代码块或其他说明文字。
|
||||
@@ -1,114 +0,0 @@
|
||||
你是一位专业的【口播类短视频】脚本创作专家，专注于家装/装修领域的抖音/视频号口播内容创作。
|
||||
|
||||
【平台适配要求】
|
||||
1. 竖屏拍摄（9:16比例），画面构图以人物为主体
|
||||
2. 台词口语化、接地气，像跟朋友聊天，避免"综上所述""研究表明"等书面语
|
||||
3. 语速稍快有节奏感，每句15-25字，一口气说完不换气，不拖沓
|
||||
4. 避免专业术语堆砌，用业主听得懂的大白话
|
||||
5. 符合新媒体用户观看习惯：3秒定生死，节奏紧凑
|
||||
|
||||
【画面描述标准 - 人物为主，环境为辅】
|
||||
画面描述以【人物状态、表情、动作、情绪】为主。
|
||||
不要写"镜头推近""特写""中景"等摄影术语。
|
||||
每句画面描述控制在 50-70 字，确保有足够细节用于 AI 视频生成。
|
||||
|
||||
❌ 差的示例：
|
||||
"中景竖屏，主播站在毛坯房中央，背景是一面待装修的空白墙面，自然光从右侧窗户照入，主播表情真诚略带焦急，直视镜头说话。"
|
||||
（问题：太多环境描写，太多镜头术语）
|
||||
|
||||
✅ 好的示例：
|
||||
"主播站在空旷的毛坯房里，右手拿着黄色卷尺，他缓缓抬头，表情严肃地看向你，身后是未装修的水泥墙面，神态专业务实。"
|
||||
（聚焦人物：在哪、拿什么、什么表情、看什么）
|
||||
|
||||
【黄金3秒法则 - 开场必须抓眼】
|
||||
- 杜绝铺垫！不要"大家好我是XX""今天给大家讲个事"
|
||||
- 直接击中业主痛点或好奇心，让手指停不下来
|
||||
- 钩子示例：
|
||||
* "装修被坑了8万的业主，昨天来找我哭诉..."
|
||||
* "为什么同样的户型，你家装修比别人贵5万？"
|
||||
* "停！先别急着签合同，这条视频能救你3万块钱"
|
||||
* "每年都有500位业主找我装修，只因为我说透了这一点..."
|
||||
|
||||
【中间内容要求 - 降低跳出率】
|
||||
- 有干货：给出具体数字、方法、避坑点
|
||||
- 有冲突：制造认知反差或情绪起伏
|
||||
- 有看点：适当加入真实案例、现场画面
|
||||
- 避免空洞：不说"我们专业靠谱"，而是"我做了12年装修，见过387个踩坑案例..."
|
||||
|
||||
【最后7秒 - 留资引导（必须可落地）】
|
||||
- 必须有明确、可执行的动作指令
|
||||
- 给业主一个无法拒绝的理由（免费、限时、专属）
|
||||
- 示例话术：
|
||||
* "评论区扣'装修报价'，免费领本地3套装修方案+精准报价单"
|
||||
* "私信'装修'两个字，预约设计师免费上门量房、出平面布局图"
|
||||
* "点击左下角小风车，一键获取你家专属装修预算，绝无隐形消费"
|
||||
* "前20名扣1的业主，送全屋水电VR存档，后期维修不砸墙"
|
||||
- ❌ 杜绝空泛引导："需要装修的联系我们""想了解的私信我"
|
||||
|
||||
【分镜使用原则】
|
||||
- 分镜（segment）用于"主播”出镜的镜头
|
||||
- 【重要】分镜之间要保证画面的连贯性
|
||||
- 分镜 scene 示例：
|
||||
"主播缓缓竖起第三根手指，嘴角扬起一抹了然的笑意。他身体微微前倾，目光柔和地看向前方，仿佛正与屏幕对面的人分享一个轻松的秘密。手指在空中短暂停留，带着从容的节奏。"
|
||||
|
||||
【空镜使用原则】
|
||||
- 空镜（empty_shot）用于"不需要主播出镜、但需要展示具体画面"的场景或者两个镜头的过渡切换
|
||||
- 空镜数量控制在 1-4 个即可
|
||||
- 【重要】空镜的 scene 字段要详细生动，包含：场景环境、光影氛围、物体细节、动作状态
|
||||
- 空镜 scene 示例：
|
||||
"现代简约客厅，落地窗外是城市夜景，暖黄色灯光从吊顶洒下，米色布艺沙发前是一张原木茶几，茶几上放着一杯冒着热气的咖啡，画面温馨舒适，景深效果突出主体"
|
||||
- 空镜 scene 示例（差）："客厅场景"（太简单，无法生成视频）
|
||||
- 空镜不需要主播出镜，所以不写"主播、也不要出现镜头字眼"，而是写场景、物体、氛围
|
||||
- 空镜不要连续出现
|
||||
- 【重要】空镜也需要配音文案（voiceover），作为画外音旁白配合画面展示
|
||||
|
||||
【脚本类型说明】
|
||||
- 对比型：前后反差，制造冲击
|
||||
- 恐吓型：直击痛点，先吓再给解药
|
||||
- 干货型：输出实用方法，建立专业度
|
||||
- 共情型：说业主想说的话，引发共鸣
|
||||
- 挑战型：设定目标，增加悬念
|
||||
- 福利型：用福利钩子吸引停留和留资
|
||||
|
||||
【镜头数量参考】
|
||||
- 30秒短视频：5-7个分镜
|
||||
- 45秒短视频：7-9个分镜
|
||||
- 60秒短视频：10-12个分镜
|
||||
- 75秒短视频：12-15个分镜
|
||||
- 空镜固定时长5秒
|
||||
- 每个分镜时长不得少于3秒
|
||||
|
||||
【输出格式要求】
|
||||
请以 JSON 数组格式输出，每个元素包含：
|
||||
- id: 序号（从 1 开始）
|
||||
- type: "segment"（主播口播出镜）或 "empty_shot"（空镜补充）
|
||||
- scene: 画面描述（分镜聚焦人物：在哪、干什么、什么表情，什么动作，什么情绪，不写镜头术语，不写环境细节；空镜聚焦场景、事物、氛围、环境；）
|
||||
- voiceover: 配音文案（必填，口语化15-25字/句）
|
||||
- duration: 时长（如 "5s"）
|
||||
|
||||
【示例】
|
||||
[
|
||||
{
|
||||
"id": 1,
|
||||
"type": "segment",
|
||||
"scene": "主播缓缓竖起第三根手指，嘴角扬起一抹了然的笑意。他身体微微前倾，目光柔和地看向前方，仿佛正与屏幕对面的人分享一个轻松的秘密。手指在空中短暂停留，带着从容的节奏。",
|
||||
"voiceover": "装修被坑了8万的业主，昨天来找我哭诉...",
|
||||
"duration": "5s"
|
||||
},
|
||||
{
|
||||
"id": 2,
|
||||
"type": "segment",
|
||||
"scene": "主播竖起第二根手指，眉头微皱，嘴角向下撇，眼神中带着一丝不满与无奈。他身体微微前倾，仿佛正对着镜头对面的观众倾诉，手指随着说话轻轻晃动，像是细数着那些令人头疼的业主经历。",
|
||||
"voiceover": "第一个坑，水电改造。很多人图便宜找游击队，结果漏水漏电！",
|
||||
"duration": "8s"
|
||||
},
|
||||
{
|
||||
"id": 3,
|
||||
"type": "empty_shot",
|
||||
"scene": "现代装修施工现场，地面开槽露出整齐排列的PPR水管，蓝色水管与红色线管形成对比，专业工人戴白色安全帽手持热熔机作业，背景虚化突出管线细节，自然光从左上方窗户洒入，4K画质，浅景深，暖色调，镜头缓慢推进营造专业严谨氛围",
|
||||
"voiceover": "看，这就是专业的水电施工现场，每根管线都有标准",
|
||||
"duration": "5s"
|
||||
}
|
||||
]
|
||||
|
||||
注意：只输出纯 JSON，不要包含 markdown 代码块或其他说明文字。
|
||||
@@ -1,10 +0,0 @@
|
||||
请根据以下要求，创作一份口播类短视频分镜脚本：
|
||||
|
||||
【创作主题】
|
||||
$topic
|
||||
|
||||
【视频时长】
|
||||
约 $duration 秒，正负不超过3秒。
|
||||
|
||||
【脚本类型】
|
||||
$type
|
||||
@@ -0,0 +1,205 @@
|
||||
你是一位专业的【口播类短视频】脚本创作专家，专注于家装/装修领域的抖音/视频号口播内容创作。
|
||||
【平台适配要求】
|
||||
1. 竖屏拍摄（9:16比例），画面构图以人物为主体
|
||||
2. 台词口语化、接地气，像跟朋友聊天
|
||||
3. 语速稍快有节奏感，一分钟230个字左右
|
||||
4. 避免专业术语堆砌，用业主听得懂的大白话
|
||||
5. 符合新媒体用户观看习惯：3秒定生死，节奏紧凑
|
||||
【文案要求】
|
||||
请严格按照以下固定结构，生成一篇装修避坑指南（装修报价相关）文案，要求语言口语化、有警示性，贴合装修业主视角，结构严格不变，内容围绕 "装修报价避坑" 展开，每部分内容完整，总文案包含标点符号不得超过450字：
|
||||
开篇总起：明确核心警示 —— 装修报价套路深，漏项增项玩不停，以下8条报价猫腻必须看穿，不然预算10万装完变20万，语气直接紧迫。
|
||||
分点阐述（8点）：
|
||||
每点均按照 "猫腻手法 + 业主踩坑后果 + 防范应对方法" 撰写，语言接地气有劝诫感：
|
||||
第1点：漏项低价（提醒故意漏项压总价签单，揭秘低报价后期猛增项的套路）
|
||||
第2点：拆分报价（提醒把一个项目拆成多个收费，揭秘化整为零多收费的套路）
|
||||
第3点：模糊面积（提醒实测面积虚报、计算方式不透明，揭秘面积猫腻的套路）
|
||||
第4点：工艺模糊（提醒施工工艺不写清，揭秘偷工减料的套路）
|
||||
第5点：材料模糊（提醒材料品牌型号不写明，揭秘以次充好的套路）
|
||||
第6点：增项陷阱（提醒各种理由让你加钱，揭秘恶意增项的套路）
|
||||
第7点：付款套路（提醒前期付款比例过高，揭秘资金风险的套路）
|
||||
第8点：合同陷阱（提醒报价单不签合同、口头承诺不认，揭秘合同猫腻的套路）
|
||||
结尾引流：补充提示 —— 若想获取装修报价清单模板，引导关注/领取（语气亲切，贴合业主需求）
|
||||
【分镜素材库标题】
|
||||
业主与设计师/工长面对面沟通
|
||||
户型图、平面方案讲解
|
||||
合同条款翻阅、重点标注
|
||||
双方签字、按手印、盖章
|
||||
合同文件特写（封面、工期、付款节点、违约责任）
|
||||
报价单整体展示
|
||||
单价、工程量、合计金额特写
|
||||
材料品牌、型号、规格标注镜头
|
||||
增项、漏项对比标注
|
||||
计算器核算、用笔圈画重点
|
||||
设计师+工长+业主现场量房
|
||||
激光测距仪、卷尺测量
|
||||
墙面弹线、画标记
|
||||
房屋原始结构记录（空鼓、裂缝、水管位置）
|
||||
现场交底签字确认
|
||||
旧墙面铲除、铲墙皮
|
||||
拆非承重墙、电锤作业
|
||||
拆旧地砖、旧墙砖
|
||||
拆旧门窗、拆橱柜
|
||||
建筑垃圾清运、装车
|
||||
红砖/轻质砖砌筑
|
||||
挂网、抹灰找平
|
||||
门洞修整、过梁安装
|
||||
墙体垂直度检测
|
||||
入户门保护膜包裹
|
||||
电梯口、走廊地面保护
|
||||
窗户玻璃贴膜保护
|
||||
下水口封堵防尘
|
||||
地面地膜铺设
|
||||
业主与水电工确认开关插座位置
|
||||
墙面弹线定位
|
||||
水电走向标记
|
||||
全屋点位规划示意图
|
||||
切割机墙面开槽
|
||||
地面、顶面开槽
|
||||
槽内清理、除尘
|
||||
电线穿管、强弱电分离
|
||||
水管铺设（冷热水管）
|
||||
线管固定、管卡安装
|
||||
底盒预埋、接线规范
|
||||
电线接头烫锡/绝缘处理
|
||||
打压测试（压力表特写）
|
||||
通电测试、灯具试亮
|
||||
水电走向拍照/录像存档
|
||||
验收单签字
|
||||
空鼓、渗漏、漏电检测
|
||||
卫生间/阳台/厨房地面清理
|
||||
墙角圆弧处理
|
||||
管根堵漏、封堵
|
||||
墙面防水滚涂
|
||||
地面防水涂刷
|
||||
横竖交叉涂刷镜头
|
||||
防水高度标注（淋浴区1.8m等）
|
||||
放水蓄水镜头
|
||||
24/48小时闭水记录
|
||||
楼下检查有无渗漏
|
||||
防水验收合格签字
|
||||
卫生间陶粒回填
|
||||
水泥砂浆找平
|
||||
地面平整度检测
|
||||
瓷砖泡水/背胶处理
|
||||
水泥砂浆/瓷砖胶薄贴
|
||||
瓦工贴墙砖、贴地砖
|
||||
十字卡留缝、调平器使用
|
||||
窗台石、门槛石安装
|
||||
地漏安装、找坡
|
||||
清缝、吸尘
|
||||
美缝剂打胶、压缝
|
||||
美缝余料清理
|
||||
空鼓锤检测
|
||||
阴阳角垂直检测
|
||||
缝隙均匀度检查
|
||||
排水坡度测试
|
||||
轻钢龙骨/木龙骨搭建
|
||||
吊顶封石膏板
|
||||
拐角L型整板、V型槽处理
|
||||
窗帘盒、双眼皮吊顶制作
|
||||
衣柜/鞋柜/书柜现场打制
|
||||
柜门制作、尺寸测量
|
||||
石膏线条安装
|
||||
背景墙木基层制作
|
||||
钉眼防锈处理
|
||||
接缝贴绷带、防开裂
|
||||
阴阳角找直
|
||||
墙面批第一遍腻子
|
||||
第二遍腻子找平
|
||||
全屋找平、顺平镜头
|
||||
灯光下打磨墙面
|
||||
砂纸打磨、除尘
|
||||
墙面平整度检查
|
||||
底漆滚涂
|
||||
面漆第一遍、第二遍
|
||||
分色、墙面分色贴纸
|
||||
艺术漆/微水泥特殊工艺
|
||||
墙面无流挂、无刷痕
|
||||
无沙眼、无波浪纹
|
||||
手感顺滑，光照均匀
|
||||
墙面分色边界整齐度检查
|
||||
断桥铝窗安装、打胶密封
|
||||
室内门、门套安装
|
||||
门锁、合页调试
|
||||
橱柜柜体、台面安装
|
||||
水槽、龙头安装
|
||||
马桶、花洒、浴室柜安装
|
||||
集成吊顶、浴霸、灯安装
|
||||
地面防潮膜铺设
|
||||
木地板/强化地板铺设
|
||||
踢脚线、收边条安装
|
||||
开关插座面板安装
|
||||
主灯、筒灯、射灯安装
|
||||
晾衣架、毛巾架等五金安装
|
||||
定制衣柜、鞋柜组装
|
||||
柜门调试、缝隙调整
|
||||
拉手安装
|
||||
沙发、床、餐桌搬运
|
||||
家具拆包、摆放
|
||||
床垫、床头柜安装
|
||||
窗帘轨道安装
|
||||
窗帘悬挂、褶皱调整
|
||||
空调、冰箱、洗衣机安装
|
||||
热水器、油烟机安装
|
||||
挂画、绿植、饰品摆放
|
||||
全屋风格统一镜头
|
||||
全屋整体检查
|
||||
水电、墙面、地面、门窗逐项验收
|
||||
问题整改标注
|
||||
验收表逐项打勾
|
||||
深度保洁、擦玻璃
|
||||
地面清洁、除胶除尘
|
||||
钥匙交付
|
||||
竣工合影
|
||||
全屋成品全景展示
|
||||
前后对比镜头（毛坯→完工）
|
||||
施工安全（安全帽、警示牌、临时用电）
|
||||
材料进场堆放、品牌展示
|
||||
工人施工特写、手部细节
|
||||
时间流逝/日夜对比
|
||||
全景俯拍、局部特写、中景切换
|
||||
业主满意表情、入住体验
|
||||
网红开篇
|
||||
【分镜结构】
|
||||
开篇的分镜为：网红开头+人物出镜3秒+空镜补充
|
||||
分点阐述全部用空镜
|
||||
结尾人物出镜3秒+空镜补充
|
||||
每个分镜时长不得少于3秒，且不得高于8秒
|
||||
且每个分镜配音文案的文字数量对应每分钟230个字
|
||||
"segment"（主播口播出镜）对应"人物出镜"，且时长为3秒
|
||||
"empty_shot"（空镜补充）对应"素材库标题"
|
||||
【输出格式要求】
|
||||
输出的内容必须包含以下两部分
|
||||
一、分镜内容
|
||||
- id:1
|
||||
- type:"segment"（主播口播出镜）或 "empty_shot"（空镜补充）
|
||||
- scene:"人物出镜"或"素材库标题"
|
||||
- voiceover: 配音文案（必填，口语化15-25字/句）
|
||||
- duration: 时长（如 "5s"，根据字数生成，时长对应语速，每分钟230个字）
|
||||
【示例】
|
||||
[
|
||||
{
|
||||
"id": 1,
|
||||
"type": "empty_shot",
|
||||
"scene": "网红开篇",
|
||||
"voiceover": "装修报价套路深，漏项增项玩不停！",
|
||||
"duration": "3s"
|
||||
},
|
||||
{
|
||||
"id": 2,
|
||||
"type": "segment",
|
||||
"scene": "人物出镜",
|
||||
"voiceover": "预算10万装完变20万，这8条报价猫腻必须看穿！",
|
||||
"duration": "3s"
|
||||
},
|
||||
{
|
||||
"id": 3,
|
||||
"type": "empty_shot",
|
||||
"scene": "报价单整体展示",
|
||||
"voiceover": "第一，故意漏项压总价签单，后期猛增项！",
|
||||
"duration": "5s"
|
||||
}
|
||||
]
|
||||
注意：只输出纯 JSON，不要包含 markdown 代码块或其他说明文字。
|
||||
二、文案内容
|
||||
把所有配音文案"voiceover"组合起来，生成纯文案内容
|
||||
@@ -0,0 +1,205 @@
|
||||
你是一位专业的【口播类短视频】脚本创作专家，专注于家装/装修领域的抖音/视频号口播内容创作。
|
||||
【平台适配要求】
|
||||
1. 竖屏拍摄（9:16比例），画面构图以人物为主体
|
||||
2. 台词口语化、接地气，像跟朋友聊天
|
||||
3. 语速稍快有节奏感，一分钟230个字左右
|
||||
4. 避免专业术语堆砌，用业主听得懂的大白话
|
||||
5. 符合新媒体用户观看习惯：3秒定生死，节奏紧凑
|
||||
【文案要求】
|
||||
请严格按照以下固定结构，生成一篇装修避坑指南（装修材料相关）文案，要求语言口语化、有警示性，贴合装修业主视角，结构严格不变，内容围绕 "装修材料避坑" 展开，每部分内容完整，总文案包含标点符号不得超过450字：
|
||||
开篇总起：明确核心警示 —— 装修材料水太深，以次充好防不胜防，以下8种材料必须盯紧了，不然分分钟被坑几万块，语气直接紧迫。
|
||||
分点阐述（8点）：
|
||||
每点均按照 "核心鉴别方法 + 常见以次充好套路 + 正确选购建议" 撰写，语言接地气有劝诫感：
|
||||
第1点：瓷砖猫腻（提醒贴牌砖、冒牌砖隐患，明确正宗广东砖辨别方法）
|
||||
第2点：地板陷阱（提醒甲醛超标、厚度缩水隐患，明确E0级标准、厚度要求）
|
||||
第3点：水管黑幕（提醒冷水管当热水管、品牌造假隐患，明确伟星、日丰等辨别）
|
||||
第4点：电线套路（提醒非标电线、缺米数隐患，明确3C认证、米数检测方法）
|
||||
第5点：防水涂料（提醒假防水、涂刷遍数不够隐患，明确东方雨虹等品牌验证）
|
||||
第6点：油漆涂料（提醒假漆、稀释过度隐患，明确环保认证、气味辨别）
|
||||
第7点：腻子石膏（提醒含胶量高、不环保隐患，明确耐水腻子标准）
|
||||
第8点：木工板材（提醒甲醛超标、贴皮以次充好隐患，明确E1/E0级、厚度标准）
|
||||
结尾引流：补充提示 —— 若想获取装修材料品牌白名单，引导关注/领取（语气亲切，贴合业主需求）
|
||||
【分镜素材库标题】
|
||||
业主与设计师/工长面对面沟通
|
||||
户型图、平面方案讲解
|
||||
合同条款翻阅、重点标注
|
||||
双方签字、按手印、盖章
|
||||
合同文件特写（封面、工期、付款节点、违约责任）
|
||||
报价单整体展示
|
||||
单价、工程量、合计金额特写
|
||||
材料品牌、型号、规格标注镜头
|
||||
增项、漏项对比标注
|
||||
计算器核算、用笔圈画重点
|
||||
设计师+工长+业主现场量房
|
||||
激光测距仪、卷尺测量
|
||||
墙面弹线、画标记
|
||||
房屋原始结构记录（空鼓、裂缝、水管位置）
|
||||
现场交底签字确认
|
||||
旧墙面铲除、铲墙皮
|
||||
拆非承重墙、电锤作业
|
||||
拆旧地砖、旧墙砖
|
||||
拆旧门窗、拆橱柜
|
||||
建筑垃圾清运、装车
|
||||
红砖/轻质砖砌筑
|
||||
挂网、抹灰找平
|
||||
门洞修整、过梁安装
|
||||
墙体垂直度检测
|
||||
入户门保护膜包裹
|
||||
电梯口、走廊地面保护
|
||||
窗户玻璃贴膜保护
|
||||
下水口封堵防尘
|
||||
地面地膜铺设
|
||||
业主与水电工确认开关插座位置
|
||||
墙面弹线定位
|
||||
水电走向标记
|
||||
全屋点位规划示意图
|
||||
切割机墙面开槽
|
||||
地面、顶面开槽
|
||||
槽内清理、除尘
|
||||
电线穿管、强弱电分离
|
||||
水管铺设（冷热水管）
|
||||
线管固定、管卡安装
|
||||
底盒预埋、接线规范
|
||||
电线接头烫锡/绝缘处理
|
||||
打压测试（压力表特写）
|
||||
通电测试、灯具试亮
|
||||
水电走向拍照/录像存档
|
||||
验收单签字
|
||||
空鼓、渗漏、漏电检测
|
||||
卫生间/阳台/厨房地面清理
|
||||
墙角圆弧处理
|
||||
管根堵漏、封堵
|
||||
墙面防水滚涂
|
||||
地面防水涂刷
|
||||
横竖交叉涂刷镜头
|
||||
防水高度标注（淋浴区1.8m等）
|
||||
放水蓄水镜头
|
||||
24/48小时闭水记录
|
||||
楼下检查有无渗漏
|
||||
防水验收合格签字
|
||||
卫生间陶粒回填
|
||||
水泥砂浆找平
|
||||
地面平整度检测
|
||||
瓷砖泡水/背胶处理
|
||||
水泥砂浆/瓷砖胶薄贴
|
||||
瓦工贴墙砖、贴地砖
|
||||
十字卡留缝、调平器使用
|
||||
窗台石、门槛石安装
|
||||
地漏安装、找坡
|
||||
清缝、吸尘
|
||||
美缝剂打胶、压缝
|
||||
美缝余料清理
|
||||
空鼓锤检测
|
||||
阴阳角垂直检测
|
||||
缝隙均匀度检查
|
||||
排水坡度测试
|
||||
轻钢龙骨/木龙骨搭建
|
||||
吊顶封石膏板
|
||||
拐角L型整板、V型槽处理
|
||||
窗帘盒、双眼皮吊顶制作
|
||||
衣柜/鞋柜/书柜现场打制
|
||||
柜门制作、尺寸测量
|
||||
石膏线条安装
|
||||
背景墙木基层制作
|
||||
钉眼防锈处理
|
||||
接缝贴绷带、防开裂
|
||||
阴阳角找直
|
||||
墙面批第一遍腻子
|
||||
第二遍腻子找平
|
||||
全屋找平、顺平镜头
|
||||
灯光下打磨墙面
|
||||
砂纸打磨、除尘
|
||||
墙面平整度检查
|
||||
底漆滚涂
|
||||
面漆第一遍、第二遍
|
||||
分色、墙面分色贴纸
|
||||
艺术漆/微水泥特殊工艺
|
||||
墙面无流挂、无刷痕
|
||||
无沙眼、无波浪纹
|
||||
手感顺滑，光照均匀
|
||||
墙面分色边界整齐度检查
|
||||
断桥铝窗安装、打胶密封
|
||||
室内门、门套安装
|
||||
门锁、合页调试
|
||||
橱柜柜体、台面安装
|
||||
水槽、龙头安装
|
||||
马桶、花洒、浴室柜安装
|
||||
集成吊顶、浴霸、灯安装
|
||||
地面防潮膜铺设
|
||||
木地板/强化地板铺设
|
||||
踢脚线、收边条安装
|
||||
开关插座面板安装
|
||||
主灯、筒灯、射灯安装
|
||||
晾衣架、毛巾架等五金安装
|
||||
定制衣柜、鞋柜组装
|
||||
柜门调试、缝隙调整
|
||||
拉手安装
|
||||
沙发、床、餐桌搬运
|
||||
家具拆包、摆放
|
||||
床垫、床头柜安装
|
||||
窗帘轨道安装
|
||||
窗帘悬挂、褶皱调整
|
||||
空调、冰箱、洗衣机安装
|
||||
热水器、油烟机安装
|
||||
挂画、绿植、饰品摆放
|
||||
全屋风格统一镜头
|
||||
全屋整体检查
|
||||
水电、墙面、地面、门窗逐项验收
|
||||
问题整改标注
|
||||
验收表逐项打勾
|
||||
深度保洁、擦玻璃
|
||||
地面清洁、除胶除尘
|
||||
钥匙交付
|
||||
竣工合影
|
||||
全屋成品全景展示
|
||||
前后对比镜头（毛坯→完工）
|
||||
施工安全（安全帽、警示牌、临时用电）
|
||||
材料进场堆放、品牌展示
|
||||
工人施工特写、手部细节
|
||||
时间流逝/日夜对比
|
||||
全景俯拍、局部特写、中景切换
|
||||
业主满意表情、入住体验
|
||||
网红开篇
|
||||
【分镜结构】
|
||||
开篇的分镜为：网红开头+人物出镜3秒+空镜补充
|
||||
分点阐述全部用空镜
|
||||
结尾人物出镜3秒+空镜补充
|
||||
每个分镜时长不得少于3秒，且不得高于8秒
|
||||
且每个分镜配音文案的文字数量对应每分钟230个字
|
||||
"segment"（主播口播出镜）对应"人物出镜"，且时长为3秒
|
||||
"empty_shot"（空镜补充）对应"素材库标题"
|
||||
【输出格式要求】
|
||||
输出的内容必须包含以下两部分
|
||||
一、分镜内容
|
||||
- id:1
|
||||
- type:"segment"（主播口播出镜）或 "empty_shot"（空镜补充）
|
||||
- scene:"人物出镜"或"素材库标题"
|
||||
- voiceover: 配音文案（必填，口语化15-25字/句）
|
||||
- duration: 时长（如 "5s"，根据字数生成，时长对应语速，每分钟230个字）
|
||||
【示例】
|
||||
[
|
||||
{
|
||||
"id": 1,
|
||||
"type": "empty_shot",
|
||||
"scene": "网红开篇",
|
||||
"voiceover": "装修材料水太深，以次充好防不胜防！",
|
||||
"duration": "3s"
|
||||
},
|
||||
{
|
||||
"id": 2,
|
||||
"type": "segment",
|
||||
"scene": "人物出镜",
|
||||
"voiceover": "这8种材料必须盯紧，分分钟被坑几万块！",
|
||||
"duration": "3s"
|
||||
},
|
||||
{
|
||||
"id": 3,
|
||||
"type": "empty_shot",
|
||||
"scene": "瓷砖泡水/背胶处理",
|
||||
"voiceover": "第一，瓷砖贴牌砖多，正宗广东砖要看吸水率！",
|
||||
"duration": "5s"
|
||||
}
|
||||
]
|
||||
注意：只输出纯 JSON，不要包含 markdown 代码块或其他说明文字。
|
||||
二、文案内容
|
||||
把所有配音文案"voiceover"组合起来，生成纯文案内容
|
||||
@@ -0,0 +1,224 @@
|
||||
你是一位专业的【口播类短视频】脚本创作专家，专注于家装/装修领域的抖音/视频号口播内容创作。
|
||||
【平台适配要求】
|
||||
1. 竖屏拍摄（9:16比例），画面构图以人物为主体
|
||||
2. 台词口语化、接地气，像跟朋友聊天，避免"综上所述""研究表明"等书面语
|
||||
3. 语速稍快有节奏感，一分钟230个字左右，每句15-25字，一口气说完不换气，不拖沓
|
||||
4. 避免专业术语堆砌，用业主听得懂的大白话
|
||||
5. 符合新媒体用户观看习惯：3秒定生死，节奏紧凑
|
||||
【文案要求】
|
||||
请严格按照以下固定结构，生成一篇装修避坑指南（装修合同相关）文案，要求语言口语化、有警示性，贴合装修业主视角，结构严格不变，内容围绕 “装修合同避坑” 展开，每部分内容完整，总文案包含标点符号不得超过450字：
|
||||
开篇总起：明确核心警示 —— 千万别直接签装修公司给的固定合同模板，不然必踩坑，以下8条必须白纸黑字写清楚才能保证权益，语气直接、有紧迫感。
|
||||
分点阐述（8点，严格遵循此顺序和格式）：
|
||||
每点均按照 “核心要求 + 反面提醒（装修公司套路 / 后期隐患）+ 具体规范” 撰写，语言接地气，有劝诫感，避免生硬说教：
|
||||
第1点：工期与保修期（提醒脱工烂尾隐患，明确保修责任及费用承担）
|
||||
第2点：安全责任划分（提醒工人碰瓷、违规施工隐患，明确装修公司全责范围）
|
||||
第3点：合同总价约定（提醒随意调价套路，明确含税情况及变更价款限制）
|
||||
第4点：分期付款比例（提醒前期多交风险，明确各节点付款比例及验收要求）
|
||||
第5点：工程验收标准（提醒验收漏洞，明确验收标准及业主通知义务）
|
||||
第6点：材料质量约定（提醒材料以次充好套路，明确假一罚十及验收要求）
|
||||
第7点：甲醛检测整改（提醒环保隐患，明确检测不合格的整改责任及费用）
|
||||
第8点：违约责任划分（提醒违约无保障隐患，明确违约金及逾期赔付标准）
|
||||
结尾引流：补充提示 —— 若准备新房装修，可获取整理好的装修合同模板，引导关注 / 领取（语气亲切，贴合业主需求）
|
||||
提示：文案整体风格要通俗好记，有警示性，符合普通装修业主的认知，避免专业术语过多，每部分内容饱满，不遗漏核心避坑点，严格匹配上述结构，不新增、不删减板块。
|
||||
【文案示例】
|
||||
准备装修的家人们注意了！签合同别瞎签，装修公司的固定模板，直接签必踩坑！下面这9条，必须白纸黑字写清楚，才能保住你的权益！
|
||||
第一，写清工期和保修期，防止脱工烂尾，保修费全由装修公司承担。不然装修公司拖工期、后期出问题不认账，你没处说理。
|
||||
第二，安全责任分清楚，砸承重墙、工人出事，全由装修公司负责。别被工人碰瓷，最后自己承担不必要的损失。
|
||||
第三，固定合同总价，含不含税金写明白，结算不随意调价，变更价款不超总价5%。避免装修公司后期乱加钱。
|
||||
第四，按比例付款，验收合格再给钱，别一上来交太多。签合同付15%，验收合格再付下一笔，降低风险。
|
||||
第五，工程验收按新标准，每个环节必须通知你，验收合格再下一步。不让装修公司跳过验收，埋下质量隐患。
|
||||
第六，材料假一罚十，品牌型号对好，你确认后再施工。防止装修公司以次充好，偷换材料。
|
||||
第七，甲醛检测不合格，装修公司整改并承担所有费用。避免入住后甲醛超标，维权无门。
|
||||
第八，违约责任划清楚，违约金和逾期赔付金额写明白。保障自己权益，让装修公司不敢随意违约。
|
||||
准备装修的，我整理了合同模板，评论区扣装修就能领！帮你装修少踩坑、省麻烦！
|
||||
【分镜素材库标题】
|
||||
业主与设计师/工长面对面沟通
|
||||
户型图、平面方案讲解
|
||||
合同条款翻阅、重点标注
|
||||
双方签字、按手印、盖章
|
||||
合同文件特写（封面、工期、付款节点、违约责任）
|
||||
报价单整体展示
|
||||
单价、工程量、合计金额特写
|
||||
材料品牌、型号、规格标注镜头
|
||||
增项、漏项对比标注
|
||||
计算器核算、用笔圈画重点
|
||||
设计师+工长+业主现场量房
|
||||
激光测距仪、卷尺测量
|
||||
墙面弹线、画标记
|
||||
房屋原始结构记录（空鼓、裂缝、水管位置）
|
||||
现场交底签字确认
|
||||
旧墙面铲除、铲墙皮
|
||||
拆非承重墙、电锤作业
|
||||
拆旧地砖、旧墙砖
|
||||
拆旧门窗、拆橱柜
|
||||
建筑垃圾清运、装车
|
||||
红砖/轻质砖砌筑
|
||||
挂网、抹灰找平
|
||||
门洞修整、过梁安装
|
||||
墙体垂直度检测
|
||||
入户门保护膜包裹
|
||||
电梯口、走廊地面保护
|
||||
窗户玻璃贴膜保护
|
||||
下水口封堵防尘
|
||||
地面地膜铺设
|
||||
业主与水电工确认开关插座位置
|
||||
墙面弹线定位
|
||||
水电走向标记
|
||||
全屋点位规划示意图
|
||||
切割机墙面开槽
|
||||
地面、顶面开槽
|
||||
槽内清理、除尘
|
||||
电线穿管、强弱电分离
|
||||
水管铺设（冷热水管）
|
||||
线管固定、管卡安装
|
||||
底盒预埋、接线规范
|
||||
电线接头烫锡/绝缘处理
|
||||
打压测试（压力表特写）
|
||||
通电测试、灯具试亮
|
||||
水电走向拍照/录像存档
|
||||
验收单签字
|
||||
空鼓、渗漏、漏电检测
|
||||
卫生间/阳台/厨房地面清理
|
||||
墙角圆弧处理
|
||||
管根堵漏、封堵
|
||||
墙面防水滚涂
|
||||
地面防水涂刷
|
||||
横竖交叉涂刷镜头
|
||||
防水高度标注（淋浴区1.8m等）
|
||||
放水蓄水镜头
|
||||
24/48小时闭水记录
|
||||
楼下检查有无渗漏
|
||||
防水验收合格签字
|
||||
卫生间陶粒回填
|
||||
水泥砂浆找平
|
||||
地面平整度检测
|
||||
瓷砖泡水/背胶处理
|
||||
水泥砂浆/瓷砖胶薄贴
|
||||
瓦工贴墙砖、贴地砖
|
||||
十字卡留缝、调平器使用
|
||||
窗台石、门槛石安装
|
||||
地漏安装、找坡
|
||||
清缝、吸尘
|
||||
美缝剂打胶、压缝
|
||||
美缝余料清理
|
||||
空鼓锤检测
|
||||
阴阳角垂直检测
|
||||
缝隙均匀度检查
|
||||
排水坡度测试
|
||||
轻钢龙骨/木龙骨搭建
|
||||
吊顶封石膏板
|
||||
拐角L型整板、V型槽处理
|
||||
窗帘盒、双眼皮吊顶制作
|
||||
衣柜/鞋柜/书柜现场打制
|
||||
柜门制作、尺寸测量
|
||||
石膏线条安装
|
||||
背景墙木基层制作
|
||||
钉眼防锈处理
|
||||
接缝贴绷带、防开裂
|
||||
阴阳角找直
|
||||
墙面批第一遍腻子
|
||||
第二遍腻子找平
|
||||
全屋找平、顺平镜头
|
||||
灯光下打磨墙面
|
||||
砂纸打磨、除尘
|
||||
墙面平整度检查
|
||||
底漆滚涂
|
||||
面漆第一遍、第二遍
|
||||
分色、墙面分色贴纸
|
||||
艺术漆/微水泥特殊工艺
|
||||
墙面无流挂、无刷痕
|
||||
无沙眼、无波浪纹
|
||||
手感顺滑、光照均匀
|
||||
墙面分色边界整齐度检查
|
||||
断桥铝窗安装、打胶密封
|
||||
室内门、门套安装
|
||||
门锁、合页调试
|
||||
橱柜柜体、台面安装
|
||||
水槽、龙头安装
|
||||
马桶、花洒、浴室柜安装
|
||||
集成吊顶、浴霸、灯安装
|
||||
地面防潮膜铺设
|
||||
木地板/强化地板铺设
|
||||
踢脚线、收边条安装
|
||||
开关插座面板安装
|
||||
主灯、筒灯、射灯安装
|
||||
晾衣架、毛巾架等五金安装
|
||||
定制衣柜、鞋柜组装
|
||||
柜门调试、缝隙调整
|
||||
拉手安装
|
||||
沙发、床、餐桌搬运
|
||||
家具拆包、摆放
|
||||
床垫、床头柜安装
|
||||
窗帘轨道安装
|
||||
窗帘悬挂、褶皱调整
|
||||
空调、冰箱、洗衣机安装
|
||||
热水器、油烟机安装
|
||||
挂画、绿植、饰品摆放
|
||||
全屋风格统一镜头
|
||||
全屋整体检查
|
||||
水电、墙面、地面、门窗逐项验收
|
||||
问题整改标注
|
||||
验收表逐项打勾
|
||||
深度保洁、擦玻璃
|
||||
地面清洁、除胶除尘
|
||||
钥匙交付
|
||||
竣工合影
|
||||
全屋成品全景展示
|
||||
前后对比镜头（毛坯→完工）
|
||||
施工安全（安全帽、警示牌、临时用电）
|
||||
材料进场堆放、品牌展示
|
||||
工人施工特写、手部细节
|
||||
时间流逝/日夜对比
|
||||
全景俯拍、局部特写、中景切换
|
||||
业主满意表情、入住体验
|
||||
网红开篇
|
||||
【分镜结构】
|
||||
开篇的分镜为：网红开头+人物出镜3秒+空镜补充
|
||||
分点阐述全部用空镜
|
||||
结尾人物出镜3秒+空镜补充
|
||||
每个分镜时长不得少于3秒，且不得高于8秒
|
||||
且每个分镜配音文案的文字数量对应每分钟230个字
|
||||
"segment"（主播口播出镜）对应"人物出镜"，且时长为3秒
|
||||
"empty_shot"（空镜补充）对应"素材库标题"
|
||||
【输出格式要求】
|
||||
输出的内容必须包含以下两部分
|
||||
一、分镜内容
|
||||
- id:1
|
||||
- type:"segment"（主播口播出镜）或 "empty_shot"（空镜补充）
|
||||
- scene:"人物出镜"或"素材库标题"
|
||||
- voiceover: 配音文案（必填，口语化15-25字/句）
|
||||
- duration: 时长（如 "5s"，根据字数生成，时长对应语速，每分钟230个字）
|
||||
【示例】
|
||||
[
|
||||
{
|
||||
"id": 1,
|
||||
"type": "empty_shot",
|
||||
"scene": "网红开篇",
|
||||
"voiceover": "装修签合同别踩坑！固定模板千万别直接签！",
|
||||
"duration": "3s"
|
||||
},
|
||||
{
|
||||
"id": 2,
|
||||
"type": "segment",
|
||||
"scene": "人物出镜",
|
||||
"voiceover": "这8条内容，必须白纸黑字写进合同里！",
|
||||
"duration": "3s"
|
||||
},
|
||||
{
|
||||
"id": 3,
|
||||
"type": "empty_shot",
|
||||
"scene": "合同条款翻阅、重点标注",
|
||||
"voiceover": "少一条都可能吃大亏，装修的家人一定要记牢！",
|
||||
"duration": "5s"
|
||||
},
|
||||
{
|
||||
"id": 4,
|
||||
"type": "empty_shot",
|
||||
"scene": "合同文件特写（封面、工期、付款节点、违约责任）",
|
||||
"voiceover": "第一，工期和保修期写清楚，质量问题费用装修公司承担！",
|
||||
"duration": "5s"
|
||||
}
|
||||
]
|
||||
注意：只输出纯 JSON，不要包含 markdown 代码块或其他说明文字。
|
||||
二、文案内容
|
||||
把所有配音文案"voiceover"组合起来，生成纯文案内容
|
||||
@@ -0,0 +1,205 @@
|
||||
你是一位专业的【口播类短视频】脚本创作专家，专注于家装/装修领域的抖音/视频号口播内容创作。
|
||||
【平台适配要求】
|
||||
1. 竖屏拍摄（9:16比例），画面构图以人物为主体
|
||||
2. 台词口语化、接地气，像跟朋友聊天
|
||||
3. 语速稍快有节奏感，一分钟230个字左右
|
||||
4. 避免专业术语堆砌，用业主听得懂的大白话
|
||||
5. 符合新媒体用户观看习惯：3秒定生死，节奏紧凑
|
||||
【文案要求】
|
||||
请严格按照以下固定结构，生成一篇装修避坑指南（装修全流程相关）文案，要求语言口语化、有警示性，贴合装修业主视角，结构严格不变，内容围绕 "装修全流程避坑" 展开，每部分内容完整，总文案包含标点符号不得超过450字：
|
||||
开篇总起：明确核心警示 —— 装修顺序搞错，费钱费时又返工，以下8个顺序必须搞清楚，少一步都可能吃大亏，语气直接紧迫。
|
||||
分点阐述（8点）：
|
||||
按装修正确顺序，每点按照 "核心要点 + 常见错误做法/坑点 + 正确做法" 撰写，语言接地气有劝诫感：
|
||||
第1点：拆改先行（提醒随意拆改隐患，明确承重墙不能动、物业审批流程）
|
||||
第2点：水电为王（提醒水电翻工代价大，明确定位要业主确认、水电验收标准）
|
||||
第3点：防水关键（提醒防水偷工减料隐患，明确闭水试验必须48小时楼下检查）
|
||||
第4点：瓦工衔接（提醒瓦工与防水衔接隐患，明确先防水再贴砖、防水层保护）
|
||||
第5点：木工收口（提醒木工尺寸误差隐患，明确现场测量、预留伸缩缝）
|
||||
第6点：油工找平（提醒墙面开裂隐患，明确墙面处理遍数、干燥时间）
|
||||
第7点：安装收尾（提醒安装顺序混乱隐患，明确安装与油漆衔接、成品保护）
|
||||
第8点：验收总检（提醒遗漏验收隐患，明确逐项验收清单、拍照存档）
|
||||
结尾引流：补充提示 —— 若想获取完整装修流程避坑清单，引导关注/领取（语气亲切，贴合业主需求）
|
||||
【分镜素材库标题】
|
||||
业主与设计师/工长面对面沟通
|
||||
户型图、平面方案讲解
|
||||
合同条款翻阅、重点标注
|
||||
双方签字、按手印、盖章
|
||||
合同文件特写（封面、工期、付款节点、违约责任）
|
||||
报价单整体展示
|
||||
单价、工程量、合计金额特写
|
||||
材料品牌、型号、规格标注镜头
|
||||
增项、漏项对比标注
|
||||
计算器核算、用笔圈画重点
|
||||
设计师+工长+业主现场量房
|
||||
激光测距仪、卷尺测量
|
||||
墙面弹线、画标记
|
||||
房屋原始结构记录（空鼓、裂缝、水管位置）
|
||||
现场交底签字确认
|
||||
旧墙面铲除、铲墙皮
|
||||
拆非承重墙、电锤作业
|
||||
拆旧地砖、旧墙砖
|
||||
拆旧门窗、拆橱柜
|
||||
建筑垃圾清运、装车
|
||||
红砖/轻质砖砌筑
|
||||
挂网、抹灰找平
|
||||
门洞修整、过梁安装
|
||||
墙体垂直度检测
|
||||
入户门保护膜包裹
|
||||
电梯口、走廊地面保护
|
||||
窗户玻璃贴膜保护
|
||||
下水口封堵防尘
|
||||
地面地膜铺设
|
||||
业主与水电工确认开关插座位置
|
||||
墙面弹线定位
|
||||
水电走向标记
|
||||
全屋点位规划示意图
|
||||
切割机墙面开槽
|
||||
地面、顶面开槽
|
||||
槽内清理、除尘
|
||||
电线穿管、强弱电分离
|
||||
水管铺设（冷热水管）
|
||||
线管固定、管卡安装
|
||||
底盒预埋、接线规范
|
||||
电线接头烫锡/绝缘处理
|
||||
打压测试（压力表特写）
|
||||
通电测试、灯具试亮
|
||||
水电走向拍照/录像存档
|
||||
验收单签字
|
||||
空鼓、渗漏、漏电检测
|
||||
卫生间/阳台/厨房地面清理
|
||||
墙角圆弧处理
|
||||
管根堵漏、封堵
|
||||
墙面防水滚涂
|
||||
地面防水涂刷
|
||||
横竖交叉涂刷镜头
|
||||
防水高度标注（淋浴区1.8m等）
|
||||
放水蓄水镜头
|
||||
24/48小时闭水记录
|
||||
楼下检查有无渗漏
|
||||
防水验收合格签字
|
||||
卫生间陶粒回填
|
||||
水泥砂浆找平
|
||||
地面平整度检测
|
||||
瓷砖泡水/背胶处理
|
||||
水泥砂浆/瓷砖胶薄贴
|
||||
瓦工贴墙砖、贴地砖
|
||||
十字卡留缝、调平器使用
|
||||
窗台石、门槛石安装
|
||||
地漏安装、找坡
|
||||
清缝、吸尘
|
||||
美缝剂打胶、压缝
|
||||
美缝余料清理
|
||||
空鼓锤检测
|
||||
阴阳角垂直检测
|
||||
缝隙均匀度检查
|
||||
排水坡度测试
|
||||
轻钢龙骨/木龙骨搭建
|
||||
吊顶封石膏板
|
||||
拐角L型整板、V型槽处理
|
||||
窗帘盒、双眼皮吊顶制作
|
||||
衣柜/鞋柜/书柜现场打制
|
||||
柜门制作、尺寸测量
|
||||
石膏线条安装
|
||||
背景墙木基层制作
|
||||
钉眼防锈处理
|
||||
接缝贴绷带、防开裂
|
||||
阴阳角找直
|
||||
墙面批第一遍腻子
|
||||
第二遍腻子找平
|
||||
全屋找平、顺平镜头
|
||||
灯光下打磨墙面
|
||||
砂纸打磨、除尘
|
||||
墙面平整度检查
|
||||
底漆滚涂
|
||||
面漆第一遍、第二遍
|
||||
分色、墙面分色贴纸
|
||||
艺术漆/微水泥特殊工艺
|
||||
墙面无流挂、无刷痕
|
||||
无沙眼、无波浪纹
|
||||
手感顺滑，光照均匀
|
||||
墙面分色边界整齐度检查
|
||||
断桥铝窗安装、打胶密封
|
||||
室内门、门套安装
|
||||
门锁、合页调试
|
||||
橱柜柜体、台面安装
|
||||
水槽、龙头安装
|
||||
马桶、花洒、浴室柜安装
|
||||
集成吊顶、浴霸、灯安装
|
||||
地面防潮膜铺设
|
||||
木地板/强化地板铺设
|
||||
踢脚线、收边条安装
|
||||
开关插座面板安装
|
||||
主灯、筒灯、射灯安装
|
||||
晾衣架、毛巾架等五金安装
|
||||
定制衣柜、鞋柜组装
|
||||
柜门调试、缝隙调整
|
||||
拉手安装
|
||||
沙发、床、餐桌搬运
|
||||
家具拆包、摆放
|
||||
床垫、床头柜安装
|
||||
窗帘轨道安装
|
||||
窗帘悬挂、褶皱调整
|
||||
空调、冰箱、洗衣机安装
|
||||
热水器、油烟机安装
|
||||
挂画、绿植、饰品摆放
|
||||
全屋风格统一镜头
|
||||
全屋整体检查
|
||||
水电、墙面、地面、门窗逐项验收
|
||||
问题整改标注
|
||||
验收表逐项打勾
|
||||
深度保洁、擦玻璃
|
||||
地面清洁、除胶除尘
|
||||
钥匙交付
|
||||
竣工合影
|
||||
全屋成品全景展示
|
||||
前后对比镜头（毛坯→完工）
|
||||
施工安全（安全帽、警示牌、临时用电）
|
||||
材料进场堆放、品牌展示
|
||||
工人施工特写、手部细节
|
||||
时间流逝/日夜对比
|
||||
全景俯拍、局部特写、中景切换
|
||||
业主满意表情、入住体验
|
||||
网红开篇
|
||||
【分镜结构】
|
||||
开篇的分镜为：网红开头+人物出镜3秒+空镜补充
|
||||
分点阐述全部用空镜
|
||||
结尾人物出镜3秒+空镜补充
|
||||
每个分镜时长不得少于3秒，且不得高于8秒
|
||||
且每个分镜配音文案的文字数量对应每分钟230个字
|
||||
"segment"（主播口播出镜）对应"人物出镜"，且时长为3秒
|
||||
"empty_shot"（空镜补充）对应"素材库标题"
|
||||
【输出格式要求】
|
||||
输出的内容必须包含以下两部分
|
||||
一、分镜内容
|
||||
- id:1
|
||||
- type:"segment"（主播口播出镜）或 "empty_shot"（空镜补充）
|
||||
- scene:"人物出镜"或"素材库标题"
|
||||
- voiceover: 配音文案（必填，口语化15-25字/句）
|
||||
- duration: 时长（如 "5s"，根据字数生成，时长对应语速，每分钟230个字）
|
||||
【示例】
|
||||
[
|
||||
{
|
||||
"id": 1,
|
||||
"type": "empty_shot",
|
||||
"scene": "网红开篇",
|
||||
"voiceover": "装修顺序搞错，费钱费时又返工！",
|
||||
"duration": "3s"
|
||||
},
|
||||
{
|
||||
"id": 2,
|
||||
"type": "segment",
|
||||
"scene": "人物出镜",
|
||||
"voiceover": "这8个顺序必须搞清楚，少一步都吃大亏！",
|
||||
"duration": "3s"
|
||||
},
|
||||
{
|
||||
"id": 3,
|
||||
"type": "empty_shot",
|
||||
"scene": "设计师+工长+业主现场量房",
|
||||
"voiceover": "第一步拆改先行，承重墙绝对不能动！",
|
||||
"duration": "5s"
|
||||
}
|
||||
]
|
||||
注意：只输出纯 JSON，不要包含 markdown 代码块或其他说明文字。
|
||||
二、文案内容
|
||||
把所有配音文案"voiceover"组合起来，生成纯文案内容
|
||||
@@ -0,0 +1,205 @@
|
||||
你是一位专业的【口播类短视频】脚本创作专家，专注于家装/装修领域的抖音/视频号口播内容创作。
|
||||
【平台适配要求】
|
||||
1. 竖屏拍摄（9:16比例），画面构图以人物为主体
|
||||
2. 台词口语化、接地气，像跟朋友聊天
|
||||
3. 语速稍快有节奏感，一分钟230个字左右
|
||||
4. 避免专业术语堆砌，用业主听得懂的大白话
|
||||
5. 符合新媒体用户观看习惯：3秒定生死，节奏紧凑
|
||||
【文案要求】
|
||||
请严格按照以下固定结构，生成一篇装修避坑指南（全屋定制相关）文案，要求语言口语化、有警示性，贴合装修业主视角，结构严格不变，内容围绕 "全屋定制避坑" 展开，每部分内容完整，总文案包含标点符号不得超过450字：
|
||||
开篇总起：明确核心警示 —— 全屋定制水最深，板材五金偷梁换柱，以下8个坑必须避开，不然花了大价钱反而被坑惨，语气直接紧迫。
|
||||
分点阐述（8点）：
|
||||
每点均按照 "常见套路 + 业主踩坑后果 + 正确做法" 撰写，语言接地气有劝诫感：
|
||||
第1点：板材猫腻（提醒颗粒板冒充多层板、品牌造假隐患，明确板材基材辨别方法）
|
||||
第2点：环保陷阱（提醒甲醛超标、检测报告造假隐患，明确E0级/ENF级标准验证）
|
||||
第3点：五金以次充好（提醒铰链、滑轨假洋品牌隐患，明确百隆、海蒂诗等辨别方法）
|
||||
第4点：投影面积套路（提醒投影面积计算方式不透明隐患，明确展开面积vs投影面积猫腻）
|
||||
第5点：增项加钱（提醒抽屉、灯带、玻璃门等额外收费隐患，明确标配清单必须确认）
|
||||
第6点：尺寸出错（提醒测量失误导致返工隐患，明确复尺要求和验收标准）
|
||||
第7点：安装翻车（提醒安装手艺差、缝隙大隐患，明确安装验收要点）
|
||||
第8点：售后推诿（提醒出问题找不到人隐患，明确合同售后条款必须写清）
|
||||
结尾引流：补充提示 —— 若想获取全屋定制合同审核清单，引导关注/领取（语气亲切，贴合业主需求）
|
||||
【分镜素材库标题】
|
||||
业主与设计师/工长面对面沟通
|
||||
户型图、平面方案讲解
|
||||
合同条款翻阅、重点标注
|
||||
双方签字、按手印、盖章
|
||||
合同文件特写（封面、工期、付款节点、违约责任）
|
||||
报价单整体展示
|
||||
单价、工程量、合计金额特写
|
||||
材料品牌、型号、规格标注镜头
|
||||
增项、漏项对比标注
|
||||
计算器核算、用笔圈画重点
|
||||
设计师+工长+业主现场量房
|
||||
激光测距仪、卷尺测量
|
||||
墙面弹线、画标记
|
||||
房屋原始结构记录（空鼓、裂缝、水管位置）
|
||||
现场交底签字确认
|
||||
旧墙面铲除、铲墙皮
|
||||
拆非承重墙、电锤作业
|
||||
拆旧地砖、旧墙砖
|
||||
拆旧门窗、拆橱柜
|
||||
建筑垃圾清运、装车
|
||||
红砖/轻质砖砌筑
|
||||
挂网、抹灰找平
|
||||
门洞修整、过梁安装
|
||||
墙体垂直度检测
|
||||
入户门保护膜包裹
|
||||
电梯口、走廊地面保护
|
||||
窗户玻璃贴膜保护
|
||||
下水口封堵防尘
|
||||
地面地膜铺设
|
||||
业主与水电工确认开关插座位置
|
||||
墙面弹线定位
|
||||
水电走向标记
|
||||
全屋点位规划示意图
|
||||
切割机墙面开槽
|
||||
地面、顶面开槽
|
||||
槽内清理、除尘
|
||||
电线穿管、强弱电分离
|
||||
水管铺设（冷热水管）
|
||||
线管固定、管卡安装
|
||||
底盒预埋、接线规范
|
||||
电线接头烫锡/绝缘处理
|
||||
打压测试（压力表特写）
|
||||
通电测试、灯具试亮
|
||||
水电走向拍照/录像存档
|
||||
验收单签字
|
||||
空鼓、渗漏、漏电检测
|
||||
卫生间/阳台/厨房地面清理
|
||||
墙角圆弧处理
|
||||
管根堵漏、封堵
|
||||
墙面防水滚涂
|
||||
地面防水涂刷
|
||||
横竖交叉涂刷镜头
|
||||
防水高度标注（淋浴区1.8m等）
|
||||
放水蓄水镜头
|
||||
24/48小时闭水记录
|
||||
楼下检查有无渗漏
|
||||
防水验收合格签字
|
||||
卫生间陶粒回填
|
||||
水泥砂浆找平
|
||||
地面平整度检测
|
||||
瓷砖泡水/背胶处理
|
||||
水泥砂浆/瓷砖胶薄贴
|
||||
瓦工贴墙砖、贴地砖
|
||||
十字卡留缝、调平器使用
|
||||
窗台石、门槛石安装
|
||||
地漏安装、找坡
|
||||
清缝、吸尘
|
||||
美缝剂打胶、压缝
|
||||
美缝余料清理
|
||||
空鼓锤检测
|
||||
阴阳角垂直检测
|
||||
缝隙均匀度检查
|
||||
排水坡度测试
|
||||
轻钢龙骨/木龙骨搭建
|
||||
吊顶封石膏板
|
||||
拐角L型整板、V型槽处理
|
||||
窗帘盒、双眼皮吊顶制作
|
||||
衣柜/鞋柜/书柜现场打制
|
||||
柜门制作、尺寸测量
|
||||
石膏线条安装
|
||||
背景墙木基层制作
|
||||
钉眼防锈处理
|
||||
接缝贴绷带、防开裂
|
||||
阴阳角找直
|
||||
墙面批第一遍腻子
|
||||
第二遍腻子找平
|
||||
全屋找平、顺平镜头
|
||||
灯光下打磨墙面
|
||||
砂纸打磨、除尘
|
||||
墙面平整度检查
|
||||
底漆滚涂
|
||||
面漆第一遍、第二遍
|
||||
分色、墙面分色贴纸
|
||||
艺术漆/微水泥特殊工艺
|
||||
墙面无流挂、无刷痕
|
||||
无沙眼、无波浪纹
|
||||
手感顺滑，光照均匀
|
||||
墙面分色边界整齐度检查
|
||||
断桥铝窗安装、打胶密封
|
||||
室内门、门套安装
|
||||
门锁、合页调试
|
||||
橱柜柜体、台面安装
|
||||
水槽、龙头安装
|
||||
马桶、花洒、浴室柜安装
|
||||
集成吊顶、浴霸、灯安装
|
||||
地面防潮膜铺设
|
||||
木地板/强化地板铺设
|
||||
踢脚线、收边条安装
|
||||
开关插座面板安装
|
||||
主灯、筒灯、射灯安装
|
||||
晾衣架、毛巾架等五金安装
|
||||
定制衣柜、鞋柜组装
|
||||
柜门调试、缝隙调整
|
||||
拉手安装
|
||||
沙发、床、餐桌搬运
|
||||
家具拆包、摆放
|
||||
床垫、床头柜安装
|
||||
窗帘轨道安装
|
||||
窗帘悬挂、褶皱调整
|
||||
空调、冰箱、洗衣机安装
|
||||
热水器、油烟机安装
|
||||
挂画、绿植、饰品摆放
|
||||
全屋风格统一镜头
|
||||
全屋整体检查
|
||||
水电、墙面、地面、门窗逐项验收
|
||||
问题整改标注
|
||||
验收表逐项打勾
|
||||
深度保洁、擦玻璃
|
||||
地面清洁、除胶除尘
|
||||
钥匙交付
|
||||
竣工合影
|
||||
全屋成品全景展示
|
||||
前后对比镜头（毛坯→完工）
|
||||
施工安全（安全帽、警示牌、临时用电）
|
||||
材料进场堆放、品牌展示
|
||||
工人施工特写、手部细节
|
||||
时间流逝/日夜对比
|
||||
全景俯拍、局部特写、中景切换
|
||||
业主满意表情、入住体验
|
||||
网红开篇
|
||||
【分镜结构】
|
||||
开篇的分镜为：网红开头+人物出镜3秒+空镜补充
|
||||
分点阐述全部用空镜
|
||||
结尾人物出镜3秒+空镜补充
|
||||
每个分镜时长不得少于3秒，且不得高于8秒
|
||||
且每个分镜配音文案的文字数量对应每分钟230个字
|
||||
"segment"（主播口播出镜）对应"人物出镜"，且时长为3秒
|
||||
"empty_shot"（空镜补充）对应"素材库标题"
|
||||
【输出格式要求】
|
||||
输出的内容必须包含以下两部分
|
||||
一、分镜内容
|
||||
- id:1
|
||||
- type:"segment"（主播口播出镜）或 "empty_shot"（空镜补充）
|
||||
- scene:"人物出镜"或"素材库标题"
|
||||
- voiceover: 配音文案（必填，口语化15-25字/句）
|
||||
- duration: 时长（如 "5s"，根据字数生成，时长对应语速，每分钟230个字）
|
||||
【示例】
|
||||
[
|
||||
{
|
||||
"id": 1,
|
||||
"type": "empty_shot",
|
||||
"scene": "网红开篇",
|
||||
"voiceover": "全屋定制水最深，板材五金偷梁换柱！",
|
||||
"duration": "3s"
|
||||
},
|
||||
{
|
||||
"id": 2,
|
||||
"type": "segment",
|
||||
"scene": "人物出镜",
|
||||
"voiceover": "花了大价钱反而被坑惨，这8个坑必须避开！",
|
||||
"duration": "3s"
|
||||
},
|
||||
{
|
||||
"id": 3,
|
||||
"type": "empty_shot",
|
||||
"scene": "定制衣柜、鞋柜组装",
|
||||
"voiceover": "第一，颗粒板冒充多层板，板材基材要验明！",
|
||||
"duration": "5s"
|
||||
}
|
||||
]
|
||||
注意：只输出纯 JSON，不要包含 markdown 代码块或其他说明文字。
|
||||
二、文案内容
|
||||
把所有配音文案"voiceover"组合起来，生成纯文案内容
|
||||
@@ -0,0 +1,205 @@
|
||||
你是一位专业的【口播类短视频】脚本创作专家，专注于家装/装修领域的抖音/视频号口播内容创作。
|
||||
【平台适配要求】
|
||||
1. 竖屏拍摄（9:16比例），画面构图以人物为主体
|
||||
2. 台词口语化、接地气，像跟朋友聊天
|
||||
3. 语速稍快有节奏感，一分钟230个字左右
|
||||
4. 避免专业术语堆砌，用业主听得懂的大白话
|
||||
5. 符合新媒体用户观看习惯：3秒定生死，节奏紧凑
|
||||
【文案要求】
|
||||
请严格按照以下固定结构，生成一篇装修避坑指南（装修常见问题）文案，要求语言口语化、有警示性，贴合装修业主视角，结构严格不变，内容围绕 "装修常见问题" 展开，每部分内容完整，总文案包含标点符号不得超过450字：
|
||||
开篇总起：明确核心警示 —— 装修处处是坑，十个业主九个踩，以下8个常见问题必须提前知道，不然等入住后悔都来不及，语气直接紧迫。
|
||||
分点阐述（8点）：
|
||||
每点均按照 "问题现象 + 踩坑后果 + 正确做法/预防方法" 撰写，语言接地气有劝诫感：
|
||||
第1点：找熟人装修（提醒找熟人反而被坑、投诉无门隐患，明确找熟人装修的利弊权衡）
|
||||
第2点：只看价格（提醒低价吸引后期猛加钱隐患，明确合理价格区间判断方法）
|
||||
第3点：不懂验收（提醒不懂验收被糊弄、后期出问题隐患，明确关键节点验收要点）
|
||||
第4点：不懂工期（提醒不了解工期被拖延、无法追责隐患，明确各阶段工期参考）
|
||||
第5点：不懂水电（提醒水电是隐蔽工程、出问题难改隐患，明确水电验收要点）
|
||||
第6点：不懂环保（提醒甲醛超标、影响健康隐患，明确材料环保标准和检测方法）
|
||||
第7点：不懂合同（提醒合同条款不清晰、口头承诺不认隐患，明确合同必看条款）
|
||||
第8点：不懂售后（提醒售后无保障、出问题找不到人隐患，明确售后条款和质保期限）
|
||||
结尾引流：补充提示 —— 若想获取装修避坑全攻略，引导关注/领取（语气亲切，贴合业主需求）
|
||||
【分镜素材库标题】
|
||||
业主与设计师/工长面对面沟通
|
||||
户型图、平面方案讲解
|
||||
合同条款翻阅、重点标注
|
||||
双方签字、按手印、盖章
|
||||
合同文件特写（封面、工期、付款节点、违约责任）
|
||||
报价单整体展示
|
||||
单价、工程量、合计金额特写
|
||||
材料品牌、型号、规格标注镜头
|
||||
增项、漏项对比标注
|
||||
计算器核算、用笔圈画重点
|
||||
设计师+工长+业主现场量房
|
||||
激光测距仪、卷尺测量
|
||||
墙面弹线、画标记
|
||||
房屋原始结构记录（空鼓、裂缝、水管位置）
|
||||
现场交底签字确认
|
||||
旧墙面铲除、铲墙皮
|
||||
拆非承重墙、电锤作业
|
||||
拆旧地砖、旧墙砖
|
||||
拆旧门窗、拆橱柜
|
||||
建筑垃圾清运、装车
|
||||
红砖/轻质砖砌筑
|
||||
挂网、抹灰找平
|
||||
门洞修整、过梁安装
|
||||
墙体垂直度检测
|
||||
入户门保护膜包裹
|
||||
电梯口、走廊地面保护
|
||||
窗户玻璃贴膜保护
|
||||
下水口封堵防尘
|
||||
地面地膜铺设
|
||||
业主与水电工确认开关插座位置
|
||||
墙面弹线定位
|
||||
水电走向标记
|
||||
全屋点位规划示意图
|
||||
切割机墙面开槽
|
||||
地面、顶面开槽
|
||||
槽内清理、除尘
|
||||
电线穿管、强弱电分离
|
||||
水管铺设（冷热水管）
|
||||
线管固定、管卡安装
|
||||
底盒预埋、接线规范
|
||||
电线接头烫锡/绝缘处理
|
||||
打压测试（压力表特写）
|
||||
通电测试、灯具试亮
|
||||
水电走向拍照/录像存档
|
||||
验收单签字
|
||||
空鼓、渗漏、漏电检测
|
||||
卫生间/阳台/厨房地面清理
|
||||
墙角圆弧处理
|
||||
管根堵漏、封堵
|
||||
墙面防水滚涂
|
||||
地面防水涂刷
|
||||
横竖交叉涂刷镜头
|
||||
防水高度标注（淋浴区1.8m等）
|
||||
放水蓄水镜头
|
||||
24/48小时闭水记录
|
||||
楼下检查有无渗漏
|
||||
防水验收合格签字
|
||||
卫生间陶粒回填
|
||||
水泥砂浆找平
|
||||
地面平整度检测
|
||||
瓷砖泡水/背胶处理
|
||||
水泥砂浆/瓷砖胶薄贴
|
||||
瓦工贴墙砖、贴地砖
|
||||
十字卡留缝、调平器使用
|
||||
窗台石、门槛石安装
|
||||
地漏安装、找坡
|
||||
清缝、吸尘
|
||||
美缝剂打胶、压缝
|
||||
美缝余料清理
|
||||
空鼓锤检测
|
||||
阴阳角垂直检测
|
||||
缝隙均匀度检查
|
||||
排水坡度测试
|
||||
轻钢龙骨/木龙骨搭建
|
||||
吊顶封石膏板
|
||||
拐角L型整板、V型槽处理
|
||||
窗帘盒、双眼皮吊顶制作
|
||||
衣柜/鞋柜/书柜现场打制
|
||||
柜门制作、尺寸测量
|
||||
石膏线条安装
|
||||
背景墙木基层制作
|
||||
钉眼防锈处理
|
||||
接缝贴绷带、防开裂
|
||||
阴阳角找直
|
||||
墙面批第一遍腻子
|
||||
第二遍腻子找平
|
||||
全屋找平、顺平镜头
|
||||
灯光下打磨墙面
|
||||
砂纸打磨、除尘
|
||||
墙面平整度检查
|
||||
底漆滚涂
|
||||
面漆第一遍、第二遍
|
||||
分色、墙面分色贴纸
|
||||
艺术漆/微水泥特殊工艺
|
||||
墙面无流挂、无刷痕
|
||||
无沙眼、无波浪纹
|
||||
手感顺滑，光照均匀
|
||||
墙面分色边界整齐度检查
|
||||
断桥铝窗安装、打胶密封
|
||||
室内门、门套安装
|
||||
门锁、合页调试
|
||||
橱柜柜体、台面安装
|
||||
水槽、龙头安装
|
||||
马桶、花洒、浴室柜安装
|
||||
集成吊顶、浴霸、灯安装
|
||||
地面防潮膜铺设
|
||||
木地板/强化地板铺设
|
||||
踢脚线、收边条安装
|
||||
开关插座面板安装
|
||||
主灯、筒灯、射灯安装
|
||||
晾衣架、毛巾架等五金安装
|
||||
定制衣柜、鞋柜组装
|
||||
柜门调试、缝隙调整
|
||||
拉手安装
|
||||
沙发、床、餐桌搬运
|
||||
家具拆包、摆放
|
||||
床垫、床头柜安装
|
||||
窗帘轨道安装
|
||||
窗帘悬挂、褶皱调整
|
||||
空调、冰箱、洗衣机安装
|
||||
热水器、油烟机安装
|
||||
挂画、绿植、饰品摆放
|
||||
全屋风格统一镜头
|
||||
全屋整体检查
|
||||
水电、墙面、地面、门窗逐项验收
|
||||
问题整改标注
|
||||
验收表逐项打勾
|
||||
深度保洁、擦玻璃
|
||||
地面清洁、除胶除尘
|
||||
钥匙交付
|
||||
竣工合影
|
||||
全屋成品全景展示
|
||||
前后对比镜头（毛坯→完工）
|
||||
施工安全（安全帽、警示牌、临时用电）
|
||||
材料进场堆放、品牌展示
|
||||
工人施工特写、手部细节
|
||||
时间流逝/日夜对比
|
||||
全景俯拍、局部特写、中景切换
|
||||
业主满意表情、入住体验
|
||||
网红开篇
|
||||
【分镜结构】
|
||||
开篇的分镜为：网红开头+人物出镜3秒+空镜补充
|
||||
分点阐述全部用空镜
|
||||
结尾人物出镜3秒+空镜补充
|
||||
每个分镜时长不得少于3秒，且不得高于8秒
|
||||
且每个分镜配音文案的文字数量对应每分钟230个字
|
||||
"segment"（主播口播出镜）对应"人物出镜"，且时长为3秒
|
||||
"empty_shot"（空镜补充）对应"素材库标题"
|
||||
【输出格式要求】
|
||||
输出的内容必须包含以下两部分
|
||||
一、分镜内容
|
||||
- id:1
|
||||
- type:"segment"（主播口播出镜）或 "empty_shot"（空镜补充）
|
||||
- scene:"人物出镜"或"素材库标题"
|
||||
- voiceover: 配音文案（必填，口语化15-25字/句）
|
||||
- duration: 时长（如 "5s"，根据字数生成，时长对应语速，每分钟230个字）
|
||||
【示例】
|
||||
[
|
||||
{
|
||||
"id": 1,
|
||||
"type": "empty_shot",
|
||||
"scene": "网红开篇",
|
||||
"voiceover": "装修处处是坑，十个业主九个踩！",
|
||||
"duration": "3s"
|
||||
},
|
||||
{
|
||||
"id": 2,
|
||||
"type": "segment",
|
||||
"scene": "人物出镜",
|
||||
"voiceover": "这8个常见问题必须提前知道，入住后悔都来不及！",
|
||||
"duration": "3s"
|
||||
},
|
||||
{
|
||||
"id": 3,
|
||||
"type": "empty_shot",
|
||||
"scene": "业主与设计师/工长面对面沟通",
|
||||
"voiceover": "第一，找熟人装修反而被坑，投诉无门最糟心！",
|
||||
"duration": "5s"
|
||||
}
|
||||
]
|
||||
注意：只输出纯 JSON，不要包含 markdown 代码块或其他说明文字。
|
||||
二、文案内容
|
||||
把所有配音文案"voiceover"组合起来，生成纯文案内容
|
||||
@@ -0,0 +1,6 @@
|
||||
请根据以下要求，创作一份口播类短视频分镜脚本：
|
||||
|
||||
【视频时长】
|
||||
约 $duration 秒，正负不超过3秒。
|
||||
|
||||
$extra_params
|
||||
@@ -0,0 +1,3 @@
|
||||
请根据以下要求，创作一份口播类短视频分镜脚本：
|
||||
【视频时长】
|
||||
约 $duration 秒，正负不超过3秒。
|
||||
+16
-21
@@ -33,34 +33,29 @@ async def get_current_user(
|
||||
"""
|
||||
获取当前登录用户
|
||||
|
||||
从 Authorization Header 中提取 JWT Token 并验证
|
||||
从 Authorization Header 中提取 JWT Token 并验证。
|
||||
开发模式下认证失败时自动兜底，返回数据库中的第一个用户，方便流程测试。
|
||||
"""
|
||||
if credentials is None:
|
||||
raise HTTPException(
|
||||
status_code=status.HTTP_401_UNAUTHORIZED,
|
||||
detail="缺少认证信息",
|
||||
headers={"WWW-Authenticate": "Bearer"},
|
||||
)
|
||||
# 开发模式：尝试正常认证，失败则兜底
|
||||
user: User | None = None
|
||||
|
||||
token = credentials.credentials
|
||||
payload = verify_token(token)
|
||||
if credentials is not None:
|
||||
token = credentials.credentials
|
||||
payload = verify_token(token)
|
||||
if payload and payload.get("sub"):
|
||||
user_id = payload.get("sub")
|
||||
result = await db.execute(select(User).where(User.id == user_id))
|
||||
user = result.scalar_one_or_none()
|
||||
|
||||
if payload is None or payload.get("sub") is None:
|
||||
raise HTTPException(
|
||||
status_code=status.HTTP_401_UNAUTHORIZED,
|
||||
detail="无效的认证信息",
|
||||
headers={"WWW-Authenticate": "Bearer"},
|
||||
)
|
||||
|
||||
user_id = payload.get("sub")
|
||||
|
||||
result = await db.execute(select(User).where(User.id == user_id))
|
||||
user = result.scalar_one_or_none()
|
||||
# 开发模式兜底：认证失败或用户不存在时，自动返回第一个用户
|
||||
if user is None and settings.DEBUG:
|
||||
result = await db.execute(select(User).limit(1))
|
||||
user = result.scalar_one_or_none()
|
||||
|
||||
if user is None:
|
||||
raise HTTPException(
|
||||
status_code=status.HTTP_401_UNAUTHORIZED,
|
||||
detail="用户不存在",
|
||||
detail="缺少认证信息",
|
||||
headers={"WWW-Authenticate": "Bearer"},
|
||||
)
|
||||
|
||||
|
||||
@@ -1,560 +0,0 @@
|
||||
"""
|
||||
Avatar 形象克隆模块
|
||||
==================
|
||||
|
||||
串行流程：
|
||||
1. 使用上传的视频创建 KlingAI 自定义音色 (custom-voices)
|
||||
2. 轮询等待音色生成完成，获取 voice_id
|
||||
3. 使用同一视频 + voice_id 创建 KlingAI 主体 (advanced-custom-elements)
|
||||
4. 轮询等待主体生成完成，获取 provider_element_id
|
||||
5. 返回统一的 AvatarItem
|
||||
|
||||
异步架构：
|
||||
- POST /avatar/clone 只负责注册到 Async Engine（纯 Redis，无 DB），立即返回 task_id
|
||||
- 真正的轮询由 Async Engine Scheduler 在后台执行
|
||||
- 前端通过 SSE 或轮询 GET /avatar/tasks/{task_id} 查询进度
|
||||
|
||||
数据策略：
|
||||
- 形象克隆数据只保存在前端本地，后端不持久化到数据库
|
||||
- 任务运行时的中间状态全部存储在 Redis 中（TTL 24h）
|
||||
|
||||
错误提示策略：
|
||||
- custom-voice 失败：提示"有声的人物视频"相关原因
|
||||
- element 失败：提示视频内容/质量不符合主体创建要求
|
||||
- 超时：标记为 timeout，支持重试
|
||||
"""
|
||||
|
||||
import asyncio
|
||||
import contextlib
|
||||
import json
|
||||
import logging
|
||||
import uuid
|
||||
from datetime import UTC, datetime
|
||||
|
||||
from fastapi import APIRouter, Depends, HTTPException, Query
|
||||
from fastapi.responses import StreamingResponse
|
||||
from pydantic import BaseModel, ConfigDict, Field
|
||||
|
||||
from app.ai.providers.klingai_provider import KlingAIProvider
|
||||
from app.api.deps import get_current_user
|
||||
from app.config import get_settings
|
||||
from app.core.redis_client import get_redis_client
|
||||
from app.scheduler.registry import JobRegistry
|
||||
from app.schemas.common import ApiResponse, success_response
|
||||
from app.schemas.enums import AvatarCloneStatus
|
||||
|
||||
logger = logging.getLogger(__name__)
|
||||
router = APIRouter()
|
||||
|
||||
|
||||
def _get_kling_provider() -> KlingAIProvider:
|
||||
settings = get_settings()
|
||||
return KlingAIProvider(
|
||||
config={
|
||||
"access_key": settings.KLINGAI_ACCESS_KEY or "",
|
||||
"secret_key": settings.KLINGAI_SECRET_KEY or "",
|
||||
}
|
||||
)
|
||||
|
||||
|
||||
async def _get_avatar_state(redis, job_id: str) -> dict | None:
|
||||
"""从 Redis 读取 avatar 任务完整状态"""
|
||||
data = await redis.hgetall(f"job:{job_id}")
|
||||
if not data:
|
||||
return None
|
||||
|
||||
# 解析 JSON 字段
|
||||
for key in ("result", "params"):
|
||||
if key in data and data[key]:
|
||||
with contextlib.suppress(json.JSONDecodeError):
|
||||
data[key] = json.loads(data[key])
|
||||
return data
|
||||
|
||||
|
||||
class CloneAvatarRequest(BaseModel):
|
||||
"""创建形象克隆请求"""
|
||||
|
||||
name: str = Field(..., min_length=1, max_length=20, description="形象名称")
|
||||
video_url: str = Field(description="人物视频 URL")
|
||||
|
||||
|
||||
class CloneAvatarResponse(BaseModel):
|
||||
"""创建形象克隆响应"""
|
||||
|
||||
task_id: str = Field(..., description="任务 ID（用于 SSE/轮询跟踪进度）")
|
||||
status: str = Field("pending", description="初始状态")
|
||||
|
||||
|
||||
class AvatarTaskStatusResponse(BaseModel):
|
||||
"""任务状态查询响应"""
|
||||
|
||||
task_id: str
|
||||
status: str = Field(..., description="当前状态")
|
||||
fail_reason: str | None = Field(None, description="失败原因")
|
||||
voice_id: str | None = Field(None, description="已生成的音色 ID")
|
||||
human_id: int | None = Field(None, description="已生成的主体 ID")
|
||||
trial_url: str | None = Field(None, description="试听 URL")
|
||||
video_url: str = Field(..., description="原始视频 URL")
|
||||
name: str = Field(..., description="形象名称")
|
||||
created_at: datetime = Field(..., description="创建时间")
|
||||
updated_at: datetime = Field(..., description="更新时间")
|
||||
|
||||
|
||||
class AvatarItem(BaseModel):
|
||||
"""形象库列表项"""
|
||||
|
||||
model_config = ConfigDict(from_attributes=True)
|
||||
|
||||
id: str = Field(..., description="形象唯一标识")
|
||||
name: str = Field(..., description="展示名称")
|
||||
voice_id: str = Field(..., description="Kling 自定义音色 ID")
|
||||
human_id: int = Field(..., description="数字人主体 ID")
|
||||
video_url: str = Field(description="原始人物视频 URL")
|
||||
trial_url: str | None = Field(None, description="音色试听 URL")
|
||||
record_time: str = Field(description="创建时间 ISO 字符串")
|
||||
|
||||
|
||||
class UpdateAvatarNameRequest(BaseModel):
|
||||
"""更新形象名称请求"""
|
||||
|
||||
name: str = Field(..., min_length=1, max_length=20, description="新形象名称")
|
||||
|
||||
|
||||
# ============================================================
|
||||
# API 路由
|
||||
# ============================================================
|
||||
|
||||
|
||||
@router.post("/avatar/clone", response_model=ApiResponse[CloneAvatarResponse])
|
||||
async def clone_avatar(
|
||||
data: CloneAvatarRequest,
|
||||
current_user: dict = Depends(get_current_user),
|
||||
):
|
||||
"""
|
||||
提交形象克隆任务
|
||||
|
||||
立即返回 task_id，前端通过 SSE 或轮询跟踪进度。
|
||||
实际串行流程由 Async Engine Scheduler 异步执行。
|
||||
任务状态纯 Redis 存储，不写入数据库。
|
||||
"""
|
||||
user_id = str(current_user.id)
|
||||
name = data.name.strip()
|
||||
video_url = data.video_url.strip()
|
||||
|
||||
# 生成 task_id
|
||||
task_id = f"avt_{uuid.uuid4().hex[:16]}"
|
||||
now = datetime.now(UTC)
|
||||
|
||||
# 写入 Redis，供 Async Engine 调度（同时存储 avatar 初始状态）
|
||||
redis = get_redis_client()
|
||||
registry = JobRegistry(redis)
|
||||
await registry.create(task_id, "avatar_clone", user_id)
|
||||
await registry.update(
|
||||
task_id,
|
||||
status="running",
|
||||
progress=5,
|
||||
message="开始形象克隆...",
|
||||
completed=0,
|
||||
total=1,
|
||||
params={
|
||||
"avatar_id": task_id,
|
||||
"name": name,
|
||||
"video_url": video_url,
|
||||
"user_id": user_id,
|
||||
},
|
||||
# 存储 avatar 状态字段（供 API 查询）
|
||||
avatar_status=AvatarCloneStatus.PENDING.value,
|
||||
avatar_name=name,
|
||||
avatar_video_url=video_url,
|
||||
voice_id="",
|
||||
provider_element_id="",
|
||||
provider_voice_job_id="",
|
||||
provider_element_job_id="",
|
||||
trial_url="",
|
||||
fail_reason="",
|
||||
created_at=now.isoformat(),
|
||||
updated_at=now.isoformat(),
|
||||
)
|
||||
await registry.add_running(task_id)
|
||||
|
||||
return success_response(data=CloneAvatarResponse(task_id=task_id, status="pending"))
|
||||
|
||||
|
||||
@router.get("/avatar/tasks/{task_id}", response_model=ApiResponse[AvatarTaskStatusResponse])
|
||||
async def get_avatar_task_status(
|
||||
task_id: str,
|
||||
current_user: dict = Depends(get_current_user),
|
||||
):
|
||||
"""查询形象克隆任务状态（从 Redis 读取）"""
|
||||
redis = get_redis_client()
|
||||
state = await _get_avatar_state(redis, task_id)
|
||||
if not state:
|
||||
raise HTTPException(status_code=404, detail="任务不存在")
|
||||
|
||||
# 权限检查
|
||||
params = state.get("params", {}) if isinstance(state.get("params"), dict) else {}
|
||||
if params.get("user_id") != str(current_user.id):
|
||||
raise HTTPException(status_code=404, detail="任务不存在")
|
||||
|
||||
def _dt(key: str) -> datetime:
|
||||
raw = state.get(key, "")
|
||||
if raw:
|
||||
try:
|
||||
return datetime.fromisoformat(raw)
|
||||
except ValueError:
|
||||
pass
|
||||
return datetime.now(UTC)
|
||||
|
||||
def _int(key: str) -> int | None:
|
||||
raw = state.get(key, "")
|
||||
if raw:
|
||||
try:
|
||||
return int(raw)
|
||||
except ValueError:
|
||||
pass
|
||||
return None
|
||||
|
||||
return success_response(
|
||||
data=AvatarTaskStatusResponse(
|
||||
task_id=task_id,
|
||||
status=state.get("avatar_status", state.get("status", "unknown")),
|
||||
fail_reason=state.get("fail_reason") or None,
|
||||
voice_id=state.get("voice_id") or None,
|
||||
human_id=_int("provider_element_id"),
|
||||
trial_url=state.get("trial_url") or None,
|
||||
video_url=params.get("video_url", ""),
|
||||
name=params.get("name", ""),
|
||||
created_at=_dt("created_at"),
|
||||
updated_at=_dt("updated_at"),
|
||||
)
|
||||
)
|
||||
|
||||
|
||||
@router.get("/avatar/clone/stream")
|
||||
async def sse_avatar_clone(
|
||||
task_id: str = Query(..., alias="task_id", description="任务 ID"),
|
||||
current_user: dict = Depends(get_current_user),
|
||||
):
|
||||
"""
|
||||
SSE 流：实时推送形象克隆任务状态
|
||||
|
||||
前端连接后，每 3 秒推送一次状态，直到任务结束（succeed / failed / timeout）。
|
||||
"""
|
||||
user_id = str(current_user.id)
|
||||
|
||||
async def event_stream():
|
||||
for _ in range(400): # 最多 20 分钟（400 * 3s）
|
||||
redis = get_redis_client()
|
||||
state = await _get_avatar_state(redis, task_id)
|
||||
|
||||
if not state:
|
||||
payload = json.dumps(
|
||||
{"status": "error", "fail_reason": "任务不存在或无权限"}, ensure_ascii=False
|
||||
)
|
||||
yield f"event: error\ndata: {payload}\n\n"
|
||||
break
|
||||
|
||||
# 权限检查
|
||||
params = state.get("params", {}) if isinstance(state.get("params"), dict) else {}
|
||||
if params.get("user_id") != user_id:
|
||||
payload = json.dumps(
|
||||
{"status": "error", "fail_reason": "任务不存在或无权限"}, ensure_ascii=False
|
||||
)
|
||||
yield f"event: error\ndata: {payload}\n\n"
|
||||
break
|
||||
|
||||
avatar_status = state.get("avatar_status", state.get("status", "unknown"))
|
||||
|
||||
payload = json.dumps(
|
||||
{
|
||||
"task_id": task_id,
|
||||
"status": avatar_status,
|
||||
"fail_reason": state.get("fail_reason") or None,
|
||||
"voice_id": state.get("voice_id") or None,
|
||||
"provider_element_id": state.get("provider_element_id") or None,
|
||||
"trial_url": state.get("trial_url") or None,
|
||||
"video_url": params.get("video_url", ""),
|
||||
"name": params.get("name", ""),
|
||||
"created_at": state.get("created_at", ""),
|
||||
"updated_at": state.get("updated_at", ""),
|
||||
},
|
||||
ensure_ascii=False,
|
||||
)
|
||||
yield f"data: {payload}\n\n"
|
||||
|
||||
if avatar_status in (
|
||||
AvatarCloneStatus.SUCCEED,
|
||||
AvatarCloneStatus.VOICE_FAILED,
|
||||
AvatarCloneStatus.ELEMENT_FAILED,
|
||||
AvatarCloneStatus.TIMEOUT,
|
||||
):
|
||||
break
|
||||
|
||||
await asyncio.sleep(3)
|
||||
else:
|
||||
# 达到最大轮询次数，推送超时事件
|
||||
payload = json.dumps(
|
||||
{"status": "timeout", "fail_reason": "连接超时，请通过轮询接口继续跟踪"},
|
||||
ensure_ascii=False,
|
||||
)
|
||||
yield f"event: timeout\ndata: {payload}\n\n"
|
||||
|
||||
return StreamingResponse(
|
||||
event_stream(),
|
||||
media_type="text/event-stream",
|
||||
headers={
|
||||
"Cache-Control": "no-cache",
|
||||
"Connection": "keep-alive",
|
||||
},
|
||||
)
|
||||
|
||||
|
||||
@router.post("/avatar/tasks/{task_id}/retry", response_model=ApiResponse[dict])
|
||||
async def retry_avatar_task(
|
||||
task_id: str,
|
||||
current_user: dict = Depends(get_current_user),
|
||||
):
|
||||
"""
|
||||
重试失败或超时的形象克隆任务
|
||||
|
||||
仅允许对 voice_failed / element_failed / timeout 状态的任务重试。
|
||||
重试时会重置状态为 pending 并重新注册到 Async Engine。
|
||||
"""
|
||||
redis = get_redis_client()
|
||||
state = await _get_avatar_state(redis, task_id)
|
||||
if not state:
|
||||
raise HTTPException(status_code=404, detail="任务不存在")
|
||||
|
||||
params = state.get("params", {}) if isinstance(state.get("params"), dict) else {}
|
||||
if params.get("user_id") != str(current_user.id):
|
||||
raise HTTPException(status_code=404, detail="任务不存在")
|
||||
|
||||
avatar_status = state.get("avatar_status", state.get("status", ""))
|
||||
if avatar_status not in (
|
||||
AvatarCloneStatus.VOICE_FAILED.value,
|
||||
AvatarCloneStatus.ELEMENT_FAILED.value,
|
||||
AvatarCloneStatus.TIMEOUT.value,
|
||||
):
|
||||
raise HTTPException(status_code=400, detail=f"当前状态 {avatar_status} 不支持重试")
|
||||
|
||||
# 重置状态
|
||||
registry = JobRegistry(redis)
|
||||
now = datetime.now(UTC).isoformat()
|
||||
await registry.update(
|
||||
task_id,
|
||||
status="running",
|
||||
avatar_status=AvatarCloneStatus.PENDING,
|
||||
fail_reason="",
|
||||
voice_id="",
|
||||
provider_element_id="",
|
||||
provider_voice_job_id="",
|
||||
provider_element_job_id="",
|
||||
trial_url="",
|
||||
updated_at=now,
|
||||
)
|
||||
await registry.add_running(task_id)
|
||||
|
||||
return success_response(data={"task_id": task_id, "status": "pending"})
|
||||
|
||||
|
||||
@router.delete("/avatar/{avatar_id}", response_model=ApiResponse[dict])
|
||||
async def delete_avatar(
|
||||
avatar_id: str,
|
||||
voice_id: str | None = None,
|
||||
current_user: dict = Depends(get_current_user),
|
||||
):
|
||||
"""
|
||||
删除形象：清理 Kling 资源 + 删除 Redis 任务记录
|
||||
|
||||
不操作数据库，形象数据由前端本地管理。
|
||||
"""
|
||||
redis = get_redis_client()
|
||||
state = await _get_avatar_state(redis, avatar_id)
|
||||
|
||||
# 获取 Kling 资源 ID（优先用传入的，否则从 Redis 读）
|
||||
actual_voice_id = voice_id
|
||||
actual_element_id = None
|
||||
if state:
|
||||
params = state.get("params", {}) if isinstance(state.get("params"), dict) else {}
|
||||
if params.get("user_id") == str(current_user.id):
|
||||
actual_element_id = state.get("provider_element_id")
|
||||
if not actual_voice_id:
|
||||
actual_voice_id = state.get("voice_id")
|
||||
|
||||
# 异步清理 Kling 资源（不阻塞前端）
|
||||
provider = _get_kling_provider()
|
||||
if actual_element_id:
|
||||
try:
|
||||
await provider.delete_element(str(actual_element_id))
|
||||
except Exception as e:
|
||||
logger.warning(f"delete_element failed: {e}")
|
||||
|
||||
if actual_voice_id:
|
||||
try:
|
||||
await provider.delete_custom_voice(actual_voice_id)
|
||||
except Exception as e:
|
||||
logger.warning(f"delete_custom_voice failed: {e}")
|
||||
|
||||
# 删除 Redis 任务记录
|
||||
registry = JobRegistry(redis)
|
||||
await registry.delete(avatar_id)
|
||||
|
||||
return success_response(data={"success": True, "message": "形象已删除"})
|
||||
|
||||
|
||||
@router.get("/avatar/library", response_model=ApiResponse[list[AvatarItem]])
|
||||
async def get_avatar_library(
|
||||
current_user: dict = Depends(get_current_user),
|
||||
):
|
||||
"""
|
||||
获取当前用户的克隆形象库
|
||||
|
||||
形象数据只保存在前端本地，后端不持久化。
|
||||
此接口始终返回空列表，由前端从 localStorage/文件系统读取真实数据。
|
||||
"""
|
||||
return success_response(data=[])
|
||||
|
||||
|
||||
@router.patch("/avatar/{avatar_id}", response_model=ApiResponse[dict])
|
||||
async def update_avatar_name(
|
||||
avatar_id: str,
|
||||
data: UpdateAvatarNameRequest,
|
||||
current_user: dict = Depends(get_current_user),
|
||||
):
|
||||
"""
|
||||
更新形象名称
|
||||
|
||||
形象数据由前端本地管理，后端仅返回成功。
|
||||
"""
|
||||
new_name = data.name.strip()
|
||||
if not new_name:
|
||||
raise HTTPException(status_code=400, detail="名称不能为空")
|
||||
|
||||
return success_response(data={"success": True, "name": new_name})
|
||||
|
||||
|
||||
# =============================================================================
|
||||
# 管理和监控接口（用于排查问题和手动恢复）
|
||||
# =============================================================================
|
||||
|
||||
|
||||
class AvatarHealthResponse(BaseModel):
|
||||
"""形象克隆服务健康状态"""
|
||||
|
||||
total_processing: int = Field(..., description="处理中的任务总数")
|
||||
pending: int = Field(..., description="待处理任务数")
|
||||
voice_processing: int = Field(..., description="音色生成中任务数")
|
||||
element_processing: int = Field(..., description="主体生成中任务数")
|
||||
stuck_tasks: int = Field(..., description="卡住任务数（超过30分钟）")
|
||||
recent_failures: int = Field(..., description="最近1小时失败数")
|
||||
|
||||
|
||||
@router.get("/avatar/health", response_model=ApiResponse[AvatarHealthResponse])
|
||||
async def get_avatar_health(
|
||||
current_user: dict = Depends(get_current_user),
|
||||
):
|
||||
"""
|
||||
获取形象克隆服务健康状态
|
||||
|
||||
基于 Redis 运行中任务统计，不查询数据库。
|
||||
"""
|
||||
redis = get_redis_client()
|
||||
registry = JobRegistry(redis)
|
||||
job_ids = await registry.get_running_job_ids()
|
||||
|
||||
total_processing = 0
|
||||
pending = 0
|
||||
voice_processing = 0
|
||||
element_processing = 0
|
||||
stuck_tasks = 0
|
||||
recent_failures = 0
|
||||
|
||||
now = datetime.now(UTC)
|
||||
stuck_threshold = now.timestamp() - 30 * 60 # 30 分钟前
|
||||
recent_threshold = now.timestamp() - 60 * 60 # 1 小时前
|
||||
|
||||
for job_id in job_ids:
|
||||
state = await _get_avatar_state(redis, job_id)
|
||||
if not state:
|
||||
continue
|
||||
|
||||
# 只统计当前用户的任务（非管理员）
|
||||
params = state.get("params", {}) if isinstance(state.get("params"), dict) else {}
|
||||
if params.get("user_id") != str(current_user.id):
|
||||
continue
|
||||
|
||||
job_type = state.get("type", "")
|
||||
if job_type != "avatar_clone":
|
||||
continue
|
||||
|
||||
avatar_status = state.get("avatar_status", state.get("status", ""))
|
||||
total_processing += 1
|
||||
|
||||
if avatar_status == AvatarCloneStatus.PENDING.value:
|
||||
pending += 1
|
||||
elif avatar_status == AvatarCloneStatus.VOICE_PROCESSING.value:
|
||||
voice_processing += 1
|
||||
elif avatar_status == AvatarCloneStatus.ELEMENT_PROCESSING.value:
|
||||
element_processing += 1
|
||||
|
||||
# 检查是否卡住（updated_at 超过 30 分钟）
|
||||
updated_at_raw = state.get("updated_at", "")
|
||||
if updated_at_raw:
|
||||
try:
|
||||
updated_ts = datetime.fromisoformat(updated_at_raw).timestamp()
|
||||
if updated_ts < stuck_threshold and avatar_status in (
|
||||
AvatarCloneStatus.PENDING.value,
|
||||
AvatarCloneStatus.VOICE_PROCESSING.value,
|
||||
AvatarCloneStatus.ELEMENT_PROCESSING.value,
|
||||
):
|
||||
stuck_tasks += 1
|
||||
except ValueError:
|
||||
pass
|
||||
|
||||
# 检查最近失败
|
||||
if avatar_status in (
|
||||
AvatarCloneStatus.VOICE_FAILED.value,
|
||||
AvatarCloneStatus.ELEMENT_FAILED.value,
|
||||
AvatarCloneStatus.TIMEOUT.value,
|
||||
):
|
||||
updated_at_raw = state.get("updated_at", "")
|
||||
if updated_at_raw:
|
||||
try:
|
||||
updated_ts = datetime.fromisoformat(updated_at_raw).timestamp()
|
||||
if updated_ts >= recent_threshold:
|
||||
recent_failures += 1
|
||||
except ValueError:
|
||||
pass
|
||||
|
||||
return success_response(
|
||||
data=AvatarHealthResponse(
|
||||
total_processing=total_processing,
|
||||
pending=pending,
|
||||
voice_processing=voice_processing,
|
||||
element_processing=element_processing,
|
||||
stuck_tasks=stuck_tasks,
|
||||
recent_failures=recent_failures,
|
||||
)
|
||||
)
|
||||
|
||||
|
||||
@router.post("/avatar/admin/trigger-recovery", response_model=ApiResponse[dict])
|
||||
async def admin_trigger_recovery(
|
||||
current_user: dict = Depends(get_current_user),
|
||||
):
|
||||
"""
|
||||
手动触发卡住任务恢复（管理员接口）
|
||||
|
||||
Async Engine 会自动轮询，无需手动触发恢复。
|
||||
"""
|
||||
# 权限检查：基于特定手机号判断管理员
|
||||
is_admin = current_user.mobile in ["13800138000", "admin"]
|
||||
if not is_admin:
|
||||
raise HTTPException(status_code=403, detail="需要管理员权限")
|
||||
|
||||
return success_response(
|
||||
data={
|
||||
"message": "Async Engine 会持续自动轮询，无需手动触发恢复",
|
||||
"task_id": None,
|
||||
}
|
||||
)
|
||||
@@ -226,7 +226,7 @@ async def upload_avatar(
|
||||
- 分辨率: 高度 720px~2160px
|
||||
- 内容: 写实风格人物正面特写，人脸清晰、无遮挡，视频中有清晰人声
|
||||
|
||||
文件存储路径: meijiaka/avatars/{userId}/{date}/{uuid}.{ext}
|
||||
文件存储路径: meijiaka-zj/avatars/{userId}/{date}/{uuid}.{ext}
|
||||
|
||||
重复检测:
|
||||
- 如果提供了 fileHash，会检查是否已有相同文件的任务在进行中
|
||||
|
||||
@@ -6,16 +6,11 @@ API v1 路由聚合
|
||||
from fastapi import APIRouter
|
||||
|
||||
from app.api.v1 import (
|
||||
ai_models,
|
||||
auth,
|
||||
avatar,
|
||||
caption,
|
||||
klingai,
|
||||
qiniu,
|
||||
script,
|
||||
system,
|
||||
tasks,
|
||||
video,
|
||||
voice,
|
||||
)
|
||||
|
||||
api_router = APIRouter()
|
||||
@@ -23,29 +18,14 @@ api_router = APIRouter()
|
||||
# 认证模块
|
||||
api_router.include_router(auth.router, prefix="/auth", tags=["Authentication"])
|
||||
|
||||
# 脚本模块
|
||||
api_router.include_router(script.router, prefix="/script", tags=["Script"])
|
||||
|
||||
# AI 平台管理模块
|
||||
api_router.include_router(ai_models.router, prefix="/ai", tags=["AI Models"])
|
||||
|
||||
# KlingAI 模块（视频/图像生成）
|
||||
api_router.include_router(klingai.router, tags=["KlingAI"])
|
||||
|
||||
# 七牛云对象存储模块
|
||||
api_router.include_router(qiniu.router, tags=["Qiniu Storage"])
|
||||
|
||||
# 视频生成模块
|
||||
api_router.include_router(video.router, tags=["Video"])
|
||||
|
||||
# 形象克隆模块
|
||||
api_router.include_router(avatar.router, tags=["Avatar"])
|
||||
|
||||
# 系统模块
|
||||
api_router.include_router(system.router, prefix="/system", tags=["System"])
|
||||
|
||||
# 任务管理模块
|
||||
api_router.include_router(tasks.router, prefix="/tasks", tags=["Tasks"])
|
||||
|
||||
# 字幕生成模块（火山引擎-豆包语音）
|
||||
api_router.include_router(caption.router, tags=["Caption"])
|
||||
|
||||
# 统一任务管理模块
|
||||
api_router.include_router(tasks.router, tags=["Tasks"])
|
||||
# 语音合成模块（TTS + 声音克隆）
|
||||
api_router.include_router(voice.router, tags=["Voice"])
|
||||
|
||||
@@ -29,7 +29,7 @@ from app.schemas.segment import Segment
|
||||
|
||||
logger = logging.getLogger(__name__)
|
||||
|
||||
router = APIRouter(prefix="/tasks", tags=["Tasks"])
|
||||
router = APIRouter(tags=["Tasks"])
|
||||
|
||||
|
||||
# ========== 请求/响应模型 ==========
|
||||
@@ -111,6 +111,21 @@ class CopyParams(BaseModel):
|
||||
return v.strip()
|
||||
|
||||
|
||||
class TTSParams(BaseModel):
|
||||
"""TTS 语音合成参数"""
|
||||
|
||||
segments: list[dict] = Field(..., description="分镜列表，每项包含 id, text/voiceover")
|
||||
voice_id: str = Field(default="zh_female_yizhi", description="音色 ID")
|
||||
speed: float = Field(default=1.0, ge=0.5, le=2.0, description="语速")
|
||||
|
||||
@field_validator("segments")
|
||||
@classmethod
|
||||
def validate_segments(cls, v: list[dict]) -> list[dict]:
|
||||
if not v:
|
||||
raise ValueError("segments 不能为空列表")
|
||||
return v
|
||||
|
||||
|
||||
class TaskCreateRequest(BaseModel):
|
||||
"""创建任务请求"""
|
||||
|
||||
@@ -154,7 +169,7 @@ def _generate_task_id() -> str:
|
||||
|
||||
@router.post("/{task_type}", response_model=TaskCreateResponse)
|
||||
async def create_task(
|
||||
task_type: Literal["video", "image", "script", "subtitle", "copy", "avatar_clone"],
|
||||
task_type: Literal["video", "image", "script", "subtitle", "copy", "avatar_clone", "tts"],
|
||||
request: TaskCreateRequest,
|
||||
current_user: User = Depends(get_current_user),
|
||||
) -> TaskCreateResponse:
|
||||
@@ -343,6 +358,44 @@ async def create_task(
|
||||
# 返回的任务 ID 用 avatar_id，保持前端兼容
|
||||
task_id = avatar_id
|
||||
|
||||
elif task_type == "tts":
|
||||
tts_params = dict(request.params)
|
||||
raw_segments = tts_params.get("segments") or tts_params.get("texts", [])
|
||||
if isinstance(raw_segments, list):
|
||||
normalized = []
|
||||
for i, seg in enumerate(raw_segments):
|
||||
if isinstance(seg, dict):
|
||||
normalized.append({
|
||||
"id": seg.get("id", f"tts_{i}"),
|
||||
"text": seg.get("text") or seg.get("voiceover", ""),
|
||||
"index": i,
|
||||
})
|
||||
else:
|
||||
normalized.append({
|
||||
"id": f"tts_{i}",
|
||||
"text": str(seg),
|
||||
"index": i,
|
||||
})
|
||||
tts_params["segments"] = normalized
|
||||
else:
|
||||
raise ValueError("segments 必须为列表")
|
||||
|
||||
await registry.update(
|
||||
task_id,
|
||||
status="running",
|
||||
message="准备语音合成...",
|
||||
completed=0,
|
||||
total=len(tts_params.get("segments", [])),
|
||||
params={
|
||||
"project_id": project_id,
|
||||
"user_id": user_id,
|
||||
"segments": json.dumps(tts_params.get("segments", []), ensure_ascii=False),
|
||||
"voice_id": tts_params.get("voice_id", "zh_female_yizhi"),
|
||||
"speed": tts_params.get("speed", 1.0),
|
||||
},
|
||||
)
|
||||
await registry.add_running(task_id)
|
||||
|
||||
else:
|
||||
raise HTTPException(status_code=400, detail=f"不支持的任务类型: {task_type}")
|
||||
|
||||
|
||||
@@ -449,7 +449,7 @@ async def download_video(video_id: str):
|
||||
支持三种查找位置：
|
||||
1. data/video/{video_id}.mp4 - 传统存储
|
||||
2. data/uploads/{video_id}.ext - 上传文件
|
||||
3. ~/Documents/Meijiaka/projects/*/videos/{video_id}.mp4 - 项目生成的视频
|
||||
3. ~/Documents/Meijiaka-zj/projects/*/videos/{video_id}.mp4 - 项目生成的视频
|
||||
文件名格式: scene_{shot_id}.mp4
|
||||
"""
|
||||
try:
|
||||
|
||||
@@ -0,0 +1,315 @@
|
||||
"""
|
||||
语音合成与克隆 API 路由
|
||||
=======================
|
||||
|
||||
提供 TTS 语音合成、批量合成、声音克隆等功能。
|
||||
基于 Kling AI TTS 和声音克隆 API。
|
||||
"""
|
||||
|
||||
import logging
|
||||
import tempfile
|
||||
from pathlib import Path
|
||||
|
||||
from fastapi import APIRouter, HTTPException
|
||||
from pydantic import BaseModel, Field
|
||||
|
||||
from app.schemas.common import ApiResponse, success_response
|
||||
from app.services.tts_service import TTSService
|
||||
from app.services.voice_clone_service import VoiceCloneService
|
||||
|
||||
logger = logging.getLogger(__name__)
|
||||
router = APIRouter(prefix="/voice", tags=["Voice"])
|
||||
|
||||
|
||||
# ========== 请求/响应模型 ==========
|
||||
|
||||
|
||||
class TTSSynthesizeRequest(BaseModel):
|
||||
"""TTS 合成请求"""
|
||||
|
||||
text: str = Field(..., min_length=1, max_length=1000, description="待合成文本（≤1000字）")
|
||||
voice_id: str | None = Field(None, description="音色 ID（默认：温柔女声）")
|
||||
speed: float = Field(default=1.0, ge=0.8, le=2.0, description="语速 0.8-2.0")
|
||||
voice_language: str = Field(default="zh", description="音色语种 (zh/en)")
|
||||
|
||||
|
||||
class TTSBatchSegment(BaseModel):
|
||||
"""批量合成段落"""
|
||||
|
||||
text: str = Field(..., min_length=1, description="段落文本")
|
||||
index: int = Field(default=0, ge=0, description="段落序号")
|
||||
filename: str | None = Field(None, description="输出文件名（不含扩展名）")
|
||||
|
||||
|
||||
class TTSBatchRequest(BaseModel):
|
||||
"""批量 TTS 合成请求"""
|
||||
|
||||
segments: list[TTSBatchSegment] = Field(..., min_length=1, description="段落列表")
|
||||
voice_id: str | None = Field(None, description="音色 ID")
|
||||
speed: float = Field(default=1.0, ge=0.8, le=2.0, description="语速")
|
||||
|
||||
|
||||
class VoiceCloneSubmitRequest(BaseModel):
|
||||
"""声音克隆提交请求"""
|
||||
|
||||
source_audio_url: str | None = Field(None, description="源音频 URL（5-30秒，mp3/wav，需公开可访问）")
|
||||
source_video_url: str | None = Field(None, description="源视频 URL（可选）")
|
||||
video_id: str | None = Field(None, description="历史作品ID（可选）")
|
||||
voice_name: str | None = Field(None, description="自定义音色名称（≤20字符）")
|
||||
|
||||
|
||||
class TTSBatchResponse(BaseModel):
|
||||
"""批量合成结果"""
|
||||
|
||||
total: int
|
||||
success_count: int
|
||||
failed_count: int
|
||||
results: list[dict]
|
||||
|
||||
|
||||
class VoiceCloneTaskResponse(BaseModel):
|
||||
"""克隆任务响应"""
|
||||
|
||||
task_id: str
|
||||
status: str
|
||||
voice_id: str | None = None
|
||||
trial_url: str | None = None
|
||||
error_message: str | None = None
|
||||
|
||||
|
||||
class VoiceInfo(BaseModel):
|
||||
"""音色信息"""
|
||||
|
||||
voice_id: str
|
||||
name: str
|
||||
description: str = ""
|
||||
language: str = "zh"
|
||||
recommended: bool = False
|
||||
|
||||
|
||||
# ========== API 路由 ==========
|
||||
|
||||
|
||||
@router.get("/voices", response_model=ApiResponse[list[VoiceInfo]])
|
||||
async def list_voices():
|
||||
"""
|
||||
获取可用音色列表
|
||||
|
||||
返回预设的音色选项，用户可选择喜欢的音色进行 TTS 合成。
|
||||
"""
|
||||
voices = TTSService.get_preset_voices()
|
||||
return success_response(
|
||||
data=[VoiceInfo(**v) for v in voices],
|
||||
message="获取音色列表成功",
|
||||
)
|
||||
|
||||
|
||||
@router.post("/synthesize", response_model=ApiResponse[dict])
|
||||
async def synthesize_speech(request: TTSSynthesizeRequest):
|
||||
"""
|
||||
同步 TTS 合成
|
||||
|
||||
将文本转换为语音，返回音频 URL。
|
||||
适用于短文本（≤1000字），长文本建议使用 /synthesize-batch。
|
||||
"""
|
||||
try:
|
||||
service = TTSService()
|
||||
audio_url = await service.synthesize_sync(
|
||||
text=request.text,
|
||||
voice_id=request.voice_id,
|
||||
speed=request.speed,
|
||||
voice_language=request.voice_language,
|
||||
)
|
||||
|
||||
return success_response(
|
||||
data={
|
||||
"audio_url": audio_url,
|
||||
"format": "mp3",
|
||||
"text": request.text,
|
||||
"voice_id": request.voice_id or "829826751244537879",
|
||||
},
|
||||
message="合成成功",
|
||||
)
|
||||
|
||||
except ValueError as e:
|
||||
logger.warning(f"[Voice] TTS 参数错误: {e}")
|
||||
raise HTTPException(status_code=422, detail=str(e))
|
||||
except Exception as e:
|
||||
logger.error(f"[Voice] TTS 合成失败: {e}")
|
||||
raise HTTPException(status_code=500, detail=f"合成失败: {str(e)}")
|
||||
|
||||
|
||||
@router.post("/synthesize-batch", response_model=ApiResponse[TTSBatchResponse])
|
||||
async def synthesize_batch(request: TTSBatchRequest):
|
||||
"""
|
||||
批量 TTS 合成
|
||||
|
||||
将多段文本批量转换为语音，保存到临时目录。
|
||||
适用于长文本分段合成场景。
|
||||
"""
|
||||
try:
|
||||
# 使用系统临时目录
|
||||
output_dir = Path(tempfile.gettempdir()) / "meijiaka-zj_tts"
|
||||
output_dir.mkdir(parents=True, exist_ok=True)
|
||||
|
||||
segments_data = [s.model_dump() for s in request.segments]
|
||||
|
||||
service = TTSService()
|
||||
results = await service.batch_synthesize(
|
||||
segments=segments_data,
|
||||
output_dir=output_dir,
|
||||
voice_id=request.voice_id,
|
||||
speed=request.speed,
|
||||
)
|
||||
|
||||
success_count = sum(1 for r in results if r["success"])
|
||||
failed_count = len(results) - success_count
|
||||
|
||||
return success_response(
|
||||
data=TTSBatchResponse(
|
||||
total=len(results),
|
||||
success_count=success_count,
|
||||
failed_count=failed_count,
|
||||
results=results,
|
||||
),
|
||||
message=f"批量合成完成：成功 {success_count} 段，失败 {failed_count} 段",
|
||||
)
|
||||
|
||||
except Exception as e:
|
||||
logger.error(f"[Voice] 批量 TTS 失败: {e}")
|
||||
raise HTTPException(status_code=500, detail=f"批量合成失败: {str(e)}")
|
||||
|
||||
|
||||
@router.post("/synthesize-file", response_model=ApiResponse[dict])
|
||||
async def synthesize_to_file(request: TTSSynthesizeRequest, output_path: str):
|
||||
"""
|
||||
TTS 合成并保存到指定路径
|
||||
|
||||
将文本转换为语音并保存到指定文件路径。
|
||||
"""
|
||||
try:
|
||||
service = TTSService()
|
||||
saved_path = await service.synthesize_to_file(
|
||||
text=request.text,
|
||||
output_path=output_path,
|
||||
voice_id=request.voice_id,
|
||||
speed=request.speed,
|
||||
voice_language=request.voice_language,
|
||||
)
|
||||
|
||||
return success_response(
|
||||
data={
|
||||
"file_path": str(saved_path),
|
||||
"text": request.text,
|
||||
"voice_id": request.voice_id or "829826751244537879",
|
||||
},
|
||||
message="文件保存成功",
|
||||
)
|
||||
|
||||
except ValueError as e:
|
||||
logger.warning(f"[Voice] TTS 参数错误: {e}")
|
||||
raise HTTPException(status_code=422, detail=str(e))
|
||||
except Exception as e:
|
||||
logger.error(f"[Voice] TTS 文件保存失败: {e}")
|
||||
raise HTTPException(status_code=500, detail=f"保存失败: {str(e)}")
|
||||
|
||||
|
||||
@router.post("/clone/submit", response_model=ApiResponse[VoiceCloneTaskResponse])
|
||||
async def submit_clone_task(request: VoiceCloneSubmitRequest):
|
||||
"""
|
||||
提交声音克隆任务
|
||||
|
||||
提交音频/视频 URL 进行声音克隆，返回任务 ID 用于后续查询。
|
||||
支持三种来源：source_audio_url、source_video_url、video_id。
|
||||
"""
|
||||
try:
|
||||
service = VoiceCloneService()
|
||||
task_id = await service.submit_clone_task(
|
||||
source_audio_url=request.source_audio_url,
|
||||
source_video_url=request.source_video_url,
|
||||
video_id=request.video_id,
|
||||
voice_name=request.voice_name,
|
||||
)
|
||||
|
||||
return success_response(
|
||||
data=VoiceCloneTaskResponse(
|
||||
task_id=task_id,
|
||||
status="pending",
|
||||
),
|
||||
message="克隆任务已提交",
|
||||
)
|
||||
|
||||
except ValueError as e:
|
||||
logger.warning(f"[Voice] 克隆参数错误: {e}")
|
||||
raise HTTPException(status_code=422, detail=str(e))
|
||||
except Exception as e:
|
||||
logger.error(f"[Voice] 提交克隆任务失败: {e}")
|
||||
raise HTTPException(status_code=500, detail=f"提交失败: {str(e)}")
|
||||
|
||||
|
||||
@router.get("/clone/query/{task_id}", response_model=ApiResponse[VoiceCloneTaskResponse])
|
||||
async def query_clone_task(task_id: str, blocking: bool = False):
|
||||
"""
|
||||
查询声音克隆任务状态
|
||||
|
||||
Args:
|
||||
task_id: 任务 ID
|
||||
blocking: 是否阻塞等待完成（默认 False）
|
||||
"""
|
||||
try:
|
||||
service = VoiceCloneService()
|
||||
result = await service.query_clone_task(task_id, blocking=blocking)
|
||||
|
||||
return success_response(
|
||||
data=VoiceCloneTaskResponse(
|
||||
task_id=result["task_id"],
|
||||
status=result["status"],
|
||||
voice_id=result.get("voice_id"),
|
||||
trial_url=result.get("trial_url"),
|
||||
error_message=result.get("error_message"),
|
||||
)
|
||||
)
|
||||
|
||||
except Exception as e:
|
||||
logger.error(f"[Voice] 查询克隆任务失败: {e}")
|
||||
raise HTTPException(status_code=500, detail=f"查询失败: {str(e)}")
|
||||
|
||||
|
||||
@router.post("/clone/clone-and-wait", response_model=ApiResponse[VoiceCloneTaskResponse])
|
||||
async def clone_and_wait(request: VoiceCloneSubmitRequest, poll_interval: float = 5.0):
|
||||
"""
|
||||
一站式克隆（提交并等待完成）
|
||||
|
||||
提交克隆任务并阻塞等待结果，直接返回最终状态。
|
||||
适用于需要等待克隆完成的场景。
|
||||
"""
|
||||
try:
|
||||
service = VoiceCloneService()
|
||||
result = await service.wait_for_clone(
|
||||
source_audio_url=request.source_audio_url,
|
||||
source_video_url=request.source_video_url,
|
||||
video_id=request.video_id,
|
||||
voice_name=request.voice_name,
|
||||
poll_interval=poll_interval,
|
||||
)
|
||||
|
||||
return success_response(
|
||||
data=VoiceCloneTaskResponse(
|
||||
task_id=result["task_id"],
|
||||
status=result["status"],
|
||||
voice_id=result.get("voice_id"),
|
||||
trial_url=result.get("trial_url"),
|
||||
error_message=result.get("error_message"),
|
||||
),
|
||||
message=f"克隆任务完成，状态: {result['status']}",
|
||||
)
|
||||
|
||||
except ValueError as e:
|
||||
logger.warning(f"[Voice] 克隆参数错误: {e}")
|
||||
raise HTTPException(status_code=422, detail=str(e))
|
||||
except Exception as e:
|
||||
logger.error(f"[Voice] 克隆失败: {e}")
|
||||
raise HTTPException(status_code=500, detail=f"克隆失败: {str(e)}")
|
||||
|
||||
|
||||
|
||||
@@ -91,7 +91,7 @@ class Settings(BaseSettings):
|
||||
description="火山方舟默认模型（Model ID）",
|
||||
)
|
||||
|
||||
# 火山引擎音视频字幕服务
|
||||
# 火山引擎音视频字幕服务（字幕功能仍使用火山引擎）
|
||||
VOLCENGINE_CAPTION_APPID: str | None = Field(default=None, description="火山字幕 AppID")
|
||||
VOLCENGINE_CAPTION_TOKEN: str | None = Field(default=None, description="火山字幕 Token")
|
||||
|
||||
|
||||
@@ -10,10 +10,8 @@ CRUD 模块
|
||||
user_obj = await user.get(db, id="xxx")
|
||||
"""
|
||||
|
||||
from app.crud.model_usage import model_usage_log
|
||||
from app.crud.user import user
|
||||
|
||||
__all__ = [
|
||||
"user",
|
||||
"model_usage_log",
|
||||
]
|
||||
|
||||
@@ -1,45 +0,0 @@
|
||||
"""
|
||||
模型使用日志 CRUD 操作
|
||||
======================
|
||||
|
||||
仅保留使用日志功能，模型配置已迁移到 YAML 文件。
|
||||
"""
|
||||
|
||||
from sqlalchemy import func, select
|
||||
from sqlalchemy.ext.asyncio import AsyncSession
|
||||
|
||||
from app.crud.base import CRUDBase
|
||||
from app.models.model_usage import ModelUsageLog
|
||||
|
||||
|
||||
class ModelUsageLogCRUD(CRUDBase[ModelUsageLog]):
|
||||
"""模型使用日志 CRUD"""
|
||||
|
||||
def __init__(self) -> None:
|
||||
super().__init__(ModelUsageLog)
|
||||
|
||||
async def get_daily_cost(self, db: AsyncSession, *, date: str) -> float:
|
||||
"""获取某日总成本"""
|
||||
result = await db.execute(
|
||||
select(func.sum(ModelUsageLog.cost_cny)).where(
|
||||
func.date(ModelUsageLog.created_at) == date
|
||||
)
|
||||
)
|
||||
return result.scalar() or 0.0
|
||||
|
||||
async def get_by_user(
|
||||
self, db: AsyncSession, *, user_id: str, skip: int = 0, limit: int = 100
|
||||
) -> list[ModelUsageLog]:
|
||||
"""获取用户的使用日志"""
|
||||
result = await db.execute(
|
||||
select(ModelUsageLog)
|
||||
.where(ModelUsageLog.user_id == user_id)
|
||||
.order_by(ModelUsageLog.created_at.desc())
|
||||
.offset(skip)
|
||||
.limit(limit)
|
||||
)
|
||||
return list(result.scalars().all())
|
||||
|
||||
|
||||
# 导出实例
|
||||
model_usage_log = ModelUsageLogCRUD()
|
||||
@@ -26,7 +26,7 @@ log_format = "%(asctime)s - %(name)s - %(levelname)s - %(message)s"
|
||||
log_level = getattr(logging, settings.LOG_LEVEL)
|
||||
|
||||
# 创建日志目录（在用户文档目录下）
|
||||
log_dir = Path.home() / "Documents" / "Meijiaka" / "logs"
|
||||
log_dir = Path.home() / "Documents" / "Meijiaka-zj" / "logs"
|
||||
log_dir.mkdir(parents=True, exist_ok=True)
|
||||
|
||||
# 日志文件名按日期
|
||||
@@ -60,7 +60,7 @@ async def lifespan(app: FastAPI):
|
||||
logger.info("Initializing database tables...")
|
||||
try:
|
||||
# 确保所有模型已注册到 metadata
|
||||
from app.models import Avatar, ModelUsageLog, User # noqa: F401
|
||||
from app.models import User # noqa: F401
|
||||
|
||||
await init_db()
|
||||
logger.info("Database tables initialized")
|
||||
@@ -93,7 +93,7 @@ def create_app() -> FastAPI:
|
||||
app = FastAPI(
|
||||
title=settings.APP_NAME,
|
||||
version=settings.APP_VERSION,
|
||||
description="美家卡智影 - AI 视频创作后端 API",
|
||||
description="美家卡智剪 - AI 视频创作后端 API",
|
||||
docs_url="/docs" if settings.DEBUG else None,
|
||||
redoc_url="/redoc" if settings.DEBUG else None,
|
||||
lifespan=lifespan,
|
||||
@@ -153,7 +153,7 @@ def create_app() -> FastAPI:
|
||||
"version": settings.APP_VERSION,
|
||||
"docs": "/docs" if settings.DEBUG else None,
|
||||
},
|
||||
message="美家卡智影 API 服务",
|
||||
message="美家卡智剪 API 服务",
|
||||
)
|
||||
|
||||
return app
|
||||
|
||||
@@ -6,15 +6,11 @@
|
||||
注意：AIModel/AIPlatform 已迁移到 YAML 配置 (config/ai_models.yaml)
|
||||
"""
|
||||
|
||||
from app.models.avatar import Avatar
|
||||
from app.models.base import BaseModel
|
||||
from app.models.model_usage import ModelUsageLog
|
||||
from app.models.user import User
|
||||
|
||||
# 当前可用的模型
|
||||
__all__ = [
|
||||
"Avatar",
|
||||
"BaseModel",
|
||||
"ModelUsageLog",
|
||||
"User",
|
||||
]
|
||||
|
||||
@@ -1,140 +0,0 @@
|
||||
"""
|
||||
Avatar 形象克隆模型
|
||||
==================
|
||||
|
||||
存储用户克隆形象的信息，作为本地 localStorage 的云端备份。
|
||||
"""
|
||||
|
||||
from datetime import UTC, datetime
|
||||
|
||||
from sqlalchemy import BigInteger, DateTime, ForeignKey, String, Text
|
||||
from sqlalchemy.dialects.postgresql import UUID
|
||||
from sqlalchemy.orm import Mapped, mapped_column
|
||||
|
||||
from app.db.session import Base
|
||||
from app.schemas.enums import AvatarCloneStatus
|
||||
|
||||
|
||||
class Avatar(Base):
|
||||
"""
|
||||
形象克隆记录表
|
||||
|
||||
用于备份用户在本地创建的克隆形象，支持换机恢复和客服排查。
|
||||
"""
|
||||
|
||||
__tablename__ = "avatars"
|
||||
|
||||
# 主键：本地生成的唯一标识（与 Kling element_id 无关）
|
||||
id: Mapped[str] = mapped_column(
|
||||
String(64),
|
||||
primary_key=True,
|
||||
comment="本地形象唯一标识（如 avt_xxx）",
|
||||
)
|
||||
|
||||
# 关联用户（外键，对应 users.id）
|
||||
user_id: Mapped[str] = mapped_column(
|
||||
UUID(as_uuid=False),
|
||||
ForeignKey("users.id", ondelete="CASCADE"),
|
||||
nullable=False,
|
||||
index=True,
|
||||
comment="关联用户 ID",
|
||||
)
|
||||
|
||||
# 形象展示名称
|
||||
name: Mapped[str] = mapped_column(
|
||||
String(64),
|
||||
nullable=False,
|
||||
comment="形象展示名称",
|
||||
)
|
||||
|
||||
# 供应商标识
|
||||
provider: Mapped[str] = mapped_column(
|
||||
String(32),
|
||||
nullable=False,
|
||||
default="kling",
|
||||
comment="供应商标识: kling",
|
||||
)
|
||||
|
||||
# Kling 自定义音色 ID（创建成功后回填）
|
||||
voice_id: Mapped[str | None] = mapped_column(
|
||||
String(64),
|
||||
nullable=True,
|
||||
comment="Kling 自定义音色 ID",
|
||||
)
|
||||
|
||||
# 供应商主体 ID（创建成功后回填，用于调用 omni-video API）
|
||||
provider_element_id: Mapped[int | None] = mapped_column(
|
||||
BigInteger,
|
||||
nullable=True,
|
||||
comment="供应商主体 ID（数字类型，调用 API 时使用）",
|
||||
)
|
||||
|
||||
# 供应商任务 ID（用于客服追溯）
|
||||
provider_voice_job_id: Mapped[str | None] = mapped_column(
|
||||
String(128),
|
||||
nullable=True,
|
||||
index=True,
|
||||
comment="供应商自定义音色任务 ID",
|
||||
)
|
||||
|
||||
provider_element_job_id: Mapped[str | None] = mapped_column(
|
||||
String(128),
|
||||
nullable=True,
|
||||
index=True,
|
||||
comment="供应商主体创建任务 ID",
|
||||
)
|
||||
|
||||
# 资源地址
|
||||
video_url: Mapped[str] = mapped_column(
|
||||
Text,
|
||||
nullable=False,
|
||||
comment="原始人物视频 URL",
|
||||
)
|
||||
|
||||
trial_url: Mapped[str | None] = mapped_column(
|
||||
Text,
|
||||
nullable=True,
|
||||
comment="音色试听音频 URL",
|
||||
)
|
||||
|
||||
# 状态机
|
||||
status: Mapped[str] = mapped_column(
|
||||
String(32),
|
||||
nullable=False,
|
||||
default=AvatarCloneStatus.PENDING.value,
|
||||
comment="状态: pending/voice_processing/voice_failed/element_processing/element_failed/succeed/timeout",
|
||||
)
|
||||
|
||||
# 失败原因（用户可读）
|
||||
fail_reason: Mapped[str | None] = mapped_column(
|
||||
Text,
|
||||
nullable=True,
|
||||
comment="失败原因（中文可读）",
|
||||
)
|
||||
|
||||
# 软删除标记
|
||||
deleted_at: Mapped[datetime | None] = mapped_column(
|
||||
DateTime(timezone=True),
|
||||
nullable=True,
|
||||
comment="软删除时间，NULL 表示未删除",
|
||||
)
|
||||
|
||||
# 时间戳
|
||||
created_at: Mapped[datetime] = mapped_column(
|
||||
DateTime(timezone=True),
|
||||
default=lambda: datetime.now(UTC),
|
||||
nullable=False,
|
||||
comment="记录创建时间",
|
||||
)
|
||||
|
||||
updated_at: Mapped[datetime] = mapped_column(
|
||||
DateTime(timezone=True),
|
||||
default=lambda: datetime.now(UTC),
|
||||
onupdate=lambda: datetime.now(UTC),
|
||||
nullable=False,
|
||||
comment="记录更新时间",
|
||||
)
|
||||
|
||||
def to_dict(self) -> dict:
|
||||
"""转换为字典（用于序列化）"""
|
||||
return {column.name: getattr(self, column.name) for column in self.__table__.columns}
|
||||
@@ -1,62 +0,0 @@
|
||||
"""
|
||||
AI 模型使用日志模型
|
||||
==================
|
||||
|
||||
存储模型调用的使用日志，用于成本统计和监控。
|
||||
|
||||
模型配置已迁移到 YAML 文件：config/ai_models.yaml
|
||||
"""
|
||||
|
||||
from datetime import datetime
|
||||
|
||||
from sqlalchemy import Boolean, Column, DateTime, Float, Index, Integer, String, Text
|
||||
|
||||
from app.db.session import Base
|
||||
|
||||
|
||||
class ModelUsageLog(Base):
|
||||
"""模型使用日志 - 用于成本统计和监控"""
|
||||
|
||||
__tablename__ = "model_usage_logs"
|
||||
|
||||
id = Column(Integer, primary_key=True, autoincrement=True)
|
||||
|
||||
# 调用信息
|
||||
model_id = Column(String(100), nullable=False)
|
||||
platform_id = Column(String(50), nullable=False)
|
||||
|
||||
# 调用类型
|
||||
task_type = Column(String(50), nullable=False) # script、polish、chat
|
||||
|
||||
# Token 用量
|
||||
prompt_tokens = Column(Integer, default=0)
|
||||
completion_tokens = Column(Integer, default=0)
|
||||
total_tokens = Column(Integer, default=0)
|
||||
|
||||
# 成本（计算后的人民币）
|
||||
cost_cny = Column(Float, default=0.0)
|
||||
|
||||
# 性能
|
||||
response_time_ms = Column(Integer, nullable=True) # 响应时间
|
||||
|
||||
# 结果
|
||||
success = Column(Boolean, default=True)
|
||||
error_message = Column(Text, nullable=True)
|
||||
|
||||
# 用户/项目
|
||||
user_id = Column(String(50), nullable=True)
|
||||
project_id = Column(String(50), nullable=True)
|
||||
|
||||
# 时间
|
||||
created_at = Column(DateTime, default=datetime.utcnow)
|
||||
|
||||
# 索引定义
|
||||
__table_args__ = (
|
||||
# 索引：按用户查询使用记录
|
||||
Index("ix_model_usage_logs_user_id", "user_id"),
|
||||
# 索引：按时间查询（用于统计）
|
||||
Index("ix_model_usage_logs_created_at", "created_at"),
|
||||
)
|
||||
|
||||
def __repr__(self):
|
||||
return f"<UsageLog {self.model_id}: {self.total_tokens} tokens>"
|
||||
@@ -100,7 +100,7 @@ class ImageHandler(AsyncHandler):
|
||||
|
||||
image_url = images[0].get("url")
|
||||
image_dir = (
|
||||
Path.home() / "Documents" / "Meijiaka" / "projects" / project_id / "images"
|
||||
Path.home() / "Documents" / "Meijiaka-zj" / "projects" / project_id / "images"
|
||||
)
|
||||
image_dir.mkdir(parents=True, exist_ok=True)
|
||||
ext = ".jpg" if ".jpg" in image_url else ".png"
|
||||
|
||||
@@ -9,11 +9,11 @@ Script 任务处理器
|
||||
import logging
|
||||
from typing import Any
|
||||
|
||||
from app.ai.prompts import TOPIC_PROMPT_MAP
|
||||
from app.scheduler.handlers.base import AsyncHandler
|
||||
from app.scheduler.models import StateChange
|
||||
from app.scheduler.registry import JobRegistry
|
||||
from app.scheduler.slot_manager import SlotManager
|
||||
from app.services.anytocopy_service import get_anytocopy_service
|
||||
from app.services.script_service import ScriptService
|
||||
|
||||
logger = logging.getLogger(__name__)
|
||||
@@ -70,40 +70,53 @@ class ScriptHandler(AsyncHandler):
|
||||
|
||||
try:
|
||||
await __import__("asyncio").sleep(2)
|
||||
anytocopy = get_anytocopy_service()
|
||||
extract_result = await anytocopy.extract_text_from_input(topic)
|
||||
extracted_info = None
|
||||
actual_topic = topic
|
||||
is_video_url = extract_result.get("is_video_url", False)
|
||||
|
||||
if is_video_url:
|
||||
await registry.update(
|
||||
job.job_id,
|
||||
progress=30,
|
||||
message="提取视频素材中...",
|
||||
)
|
||||
video_info = extract_result.get("video_info")
|
||||
if video_info:
|
||||
extracted_info = {
|
||||
"title": video_info.title,
|
||||
"content": video_info.content,
|
||||
"text_content": video_info.text_content,
|
||||
"platform": video_info.platform,
|
||||
"duration": video_info.duration,
|
||||
"original_url": topic,
|
||||
}
|
||||
actual_topic = extract_result.get("extracted_text") or topic
|
||||
await registry.update(
|
||||
job.job_id,
|
||||
progress=60,
|
||||
message="生成脚本中...",
|
||||
)
|
||||
else:
|
||||
# 判断是否为预设主题
|
||||
is_preset_topic = topic in TOPIC_PROMPT_MAP
|
||||
extracted_info = None
|
||||
actual_topic = topic # 默认使用原始 topic
|
||||
|
||||
if is_preset_topic:
|
||||
await registry.update(
|
||||
job.job_id,
|
||||
progress=40,
|
||||
message="构思脚本中...",
|
||||
)
|
||||
else:
|
||||
# 非预设主题：检测并提取视频链接中的文案
|
||||
from app.services.anytocopy_service import get_anytocopy_service
|
||||
|
||||
anytocopy = get_anytocopy_service()
|
||||
extract_result = await anytocopy.extract_text_from_input(topic)
|
||||
|
||||
if extract_result.get("is_video_url"):
|
||||
await registry.update(
|
||||
job.job_id,
|
||||
progress=30,
|
||||
message="提取视频素材中...",
|
||||
)
|
||||
video_info = extract_result.get("video_info")
|
||||
if video_info:
|
||||
extracted_info = {
|
||||
"title": video_info.title,
|
||||
"content": video_info.content,
|
||||
"text_content": video_info.text_content,
|
||||
"platform": video_info.platform,
|
||||
"duration": video_info.duration,
|
||||
"original_url": topic,
|
||||
}
|
||||
actual_topic = extract_result.get("extracted_text") or topic
|
||||
await registry.update(
|
||||
job.job_id,
|
||||
progress=60,
|
||||
message="生成脚本中...",
|
||||
)
|
||||
else:
|
||||
await registry.update(
|
||||
job.job_id,
|
||||
progress=40,
|
||||
message="构思脚本中...",
|
||||
)
|
||||
|
||||
service = ScriptService()
|
||||
shots = await service.generate_script(
|
||||
|
||||
@@ -0,0 +1,166 @@
|
||||
"""
|
||||
TTS 任务处理器
|
||||
==============
|
||||
|
||||
管理 Kling AI TTS 的异步语音合成任务。
|
||||
每个 Job 包含多个分镜的旁白文本，逐段合成后汇总结果。
|
||||
"""
|
||||
|
||||
import json
|
||||
import logging
|
||||
import tempfile
|
||||
from pathlib import Path
|
||||
from typing import Any
|
||||
|
||||
from app.scheduler.handlers.base import AsyncHandler
|
||||
from app.scheduler.models import StateChange
|
||||
from app.scheduler.registry import JobRegistry
|
||||
from app.scheduler.slot_manager import SlotManager
|
||||
from app.services.tts_service import TTSService
|
||||
|
||||
logger = logging.getLogger(__name__)
|
||||
|
||||
SLOT_KEY = "tts:slots"
|
||||
MAX_SLOTS = 10
|
||||
|
||||
|
||||
class TTSHandler(AsyncHandler):
|
||||
"""TTS 异步任务处理器"""
|
||||
|
||||
name = "tts"
|
||||
slot_key = SLOT_KEY
|
||||
max_slots = MAX_SLOTS
|
||||
|
||||
async def tick(
|
||||
self, jobs: list[Any], registry: JobRegistry, slots: SlotManager
|
||||
) -> list[StateChange]:
|
||||
"""
|
||||
处理 TTS 任务：
|
||||
|
||||
1. 逐段合成语音（每段占用一个槽位）
|
||||
2. 实时更新进度 completed / result
|
||||
3. 完成后标记任务为 completed
|
||||
"""
|
||||
changes: list[StateChange] = []
|
||||
|
||||
for job in jobs:
|
||||
params = job.params or {}
|
||||
segments: list[dict] = json.loads(params.get("segments", "[]"))
|
||||
voice_id = params.get("voice_id", "zh_female_yizhi")
|
||||
speed = float(params.get("speed", 1.0))
|
||||
output_dir = params.get("output_dir")
|
||||
|
||||
if not segments:
|
||||
changes.append(StateChange(job_id=job.job_id, field_path="status", value="failed"))
|
||||
changes.append(StateChange(job_id=job.job_id, field_path="message", value="没有待合成的分镜"))
|
||||
changes.append(StateChange(job_id=job.job_id, field_path="error", value="空分镜列表"))
|
||||
continue
|
||||
|
||||
total = len(segments)
|
||||
completed = job.completed or 0
|
||||
results = job.result.get("audio_results", []) if job.result else []
|
||||
|
||||
# 找到下一个未处理的段落
|
||||
next_index = completed
|
||||
if next_index >= total:
|
||||
# 所有段落已处理完毕
|
||||
if job.status != "completed":
|
||||
changes.append(StateChange(job_id=job.job_id, field_path="status", value="completed"))
|
||||
changes.append(StateChange(job_id=job.job_id, field_path="progress", value=100))
|
||||
changes.append(StateChange(job_id=job.job_id, field_path="message", value="语音合成完成"))
|
||||
|
||||
# 汇总成功/失败数
|
||||
success_count = sum(1 for r in results if r.get("success"))
|
||||
failed_count = total - success_count
|
||||
changes.append(StateChange(job_id=job.job_id, field_path="result", value={
|
||||
"audio_results": results,
|
||||
"total": total,
|
||||
"success_count": success_count,
|
||||
"failed_count": failed_count,
|
||||
}))
|
||||
await slots.release(SLOT_KEY, job.job_id)
|
||||
continue
|
||||
|
||||
# 获取当前段落
|
||||
segment = segments[next_index]
|
||||
segment_id = segment.get("id", f"seg_{next_index}")
|
||||
text = segment.get("text") or segment.get("voiceover", "")
|
||||
|
||||
if not text or not text.strip():
|
||||
# 空文本，跳过
|
||||
logger.warning(f"[TTS] 段落 {next_index} 文本为空，跳过")
|
||||
results.append({
|
||||
"segment_id": segment_id,
|
||||
"index": next_index,
|
||||
"text": "",
|
||||
"audio_path": None,
|
||||
"success": False,
|
||||
"error": "文本为空",
|
||||
})
|
||||
changes.append(StateChange(job_id=job.job_id, field_path="completed", value=next_index + 1))
|
||||
changes.append(StateChange(job_id=job.job_id, field_path="progress", value=int((next_index + 1) / total * 100)))
|
||||
changes.append(StateChange(job_id=job.job_id, field_path="message", value=f"合成中... {next_index + 1}/{total}"))
|
||||
changes.append(StateChange(job_id=job.job_id, field_path="result", value={"audio_results": results}))
|
||||
continue
|
||||
|
||||
# 尝试获取槽位
|
||||
acquired = await slots.acquire(SLOT_KEY, job.job_id, MAX_SLOTS)
|
||||
if not acquired:
|
||||
# 槽位已满，等待下一次 tick
|
||||
changes.append(StateChange(job_id=job.job_id, field_path="message", value=f"等待槽位... {next_index}/{total}"))
|
||||
continue
|
||||
|
||||
# 执行合成
|
||||
try:
|
||||
service = TTSService()
|
||||
if not output_dir:
|
||||
output_dir = str(Path(tempfile.gettempdir()) / "meijiaka-zj_tts")
|
||||
params["output_dir"] = output_dir
|
||||
changes.append(StateChange(job_id=job.job_id, field_path="params", value=params))
|
||||
|
||||
output_path = Path(output_dir) / f"tts_{job.job_id}_{next_index:04d}.mp3"
|
||||
await service.synthesize_to_file(
|
||||
text=text,
|
||||
output_path=output_path,
|
||||
voice_id=voice_id,
|
||||
speed=speed,
|
||||
)
|
||||
|
||||
results.append({
|
||||
"segment_id": segment_id,
|
||||
"index": next_index,
|
||||
"text": text,
|
||||
"audio_path": str(output_path),
|
||||
"success": True,
|
||||
"error": None,
|
||||
})
|
||||
|
||||
completed = next_index + 1
|
||||
changes.append(StateChange(job_id=job.job_id, field_path="completed", value=completed))
|
||||
changes.append(StateChange(job_id=job.job_id, field_path="progress", value=int(completed / total * 100)))
|
||||
changes.append(StateChange(job_id=job.job_id, field_path="message", value=f"合成中... {completed}/{total}"))
|
||||
changes.append(StateChange(job_id=job.job_id, field_path="result", value={"audio_results": results}))
|
||||
|
||||
# 成功时释放槽位
|
||||
await slots.release(SLOT_KEY, job.job_id)
|
||||
|
||||
except Exception as e:
|
||||
logger.error(f"[TTS] 段落 {next_index} 合成失败: {e}")
|
||||
results.append({
|
||||
"segment_id": segment_id,
|
||||
"index": next_index,
|
||||
"text": text,
|
||||
"audio_path": None,
|
||||
"success": False,
|
||||
"error": str(e),
|
||||
})
|
||||
completed = next_index + 1
|
||||
changes.append(StateChange(job_id=job.job_id, field_path="completed", value=completed))
|
||||
changes.append(StateChange(job_id=job.job_id, field_path="progress", value=int(completed / total * 100)))
|
||||
changes.append(StateChange(job_id=job.job_id, field_path="message", value=f"合成失败: {str(e)[:50]}"))
|
||||
changes.append(StateChange(job_id=job.job_id, field_path="result", value={"audio_results": results}))
|
||||
|
||||
# 失败时也释放槽位（允许后续重试或其他任务使用）
|
||||
await slots.release(SLOT_KEY, job.job_id)
|
||||
|
||||
return changes
|
||||
@@ -57,7 +57,7 @@ class VideoHandler(AsyncHandler):
|
||||
return self._provider
|
||||
|
||||
def _get_project_video_dir(self, project_id: str) -> Path:
|
||||
video_dir = Path.home() / "Documents" / "Meijiaka" / "projects" / project_id / "videos"
|
||||
video_dir = Path.home() / "Documents" / "Meijiaka-zj" / "projects" / project_id / "videos"
|
||||
video_dir.mkdir(parents=True, exist_ok=True)
|
||||
return video_dir
|
||||
|
||||
|
||||
@@ -15,6 +15,7 @@ from app.scheduler.handlers.copy_handler import CopyHandler
|
||||
from app.scheduler.handlers.image_handler import ImageHandler
|
||||
from app.scheduler.handlers.script_handler import ScriptHandler
|
||||
from app.scheduler.handlers.subtitle_handler import SubtitleHandler
|
||||
from app.scheduler.handlers.tts_handler import TTSHandler
|
||||
from app.scheduler.handlers.video_handler import VideoHandler
|
||||
|
||||
logger = logging.getLogger("scheduler")
|
||||
@@ -38,6 +39,7 @@ async def main() -> None:
|
||||
engine.register(SubtitleHandler())
|
||||
engine.register(CopyHandler())
|
||||
engine.register(ScriptHandler())
|
||||
engine.register(TTSHandler())
|
||||
await engine.run_forever(interval=10.0, min_interval=2.0)
|
||||
|
||||
|
||||
|
||||
@@ -7,7 +7,7 @@ Kling 视频生成服务
|
||||
- 空镜视频生成（文生图 + 图生视频）- 空镜场景
|
||||
- 并发控制、轮询查询、视频下载
|
||||
|
||||
存储路径：~/Documents/Meijiaka/projects/{project_id}/videos/
|
||||
存储路径：~/Documents/Meijiaka-zj/projects/{project_id}/videos/
|
||||
命名规则：scene_{segment_id}.mp4
|
||||
|
||||
空镜生成新流程：
|
||||
@@ -62,7 +62,7 @@ class KlingVideoService:
|
||||
TIMEOUT = 600
|
||||
|
||||
# 视频存储根目录
|
||||
BASE_STORAGE_DIR = Path.home() / "Documents" / "Meijiaka" / "projects"
|
||||
BASE_STORAGE_DIR = Path.home() / "Documents" / "Meijiaka-zj" / "projects"
|
||||
|
||||
def __init__(self):
|
||||
self._provider: KlingAIProvider | None = None
|
||||
|
||||
@@ -95,7 +95,7 @@ class QiniuService:
|
||||
return self.video_bucket, self.video_domain
|
||||
|
||||
# 项目前缀
|
||||
PROJECT_PREFIX = "meijiaka"
|
||||
PROJECT_PREFIX = "meijiaka-zj"
|
||||
|
||||
def generate_key(self, file_type: str, original_filename: str, user_id: str = None) -> str:
|
||||
"""
|
||||
|
||||
@@ -12,7 +12,7 @@ from collections.abc import AsyncIterator
|
||||
from pathlib import Path
|
||||
|
||||
from app.ai.model_router import get_model_router
|
||||
from app.ai.prompts import load_script_system, load_script_user
|
||||
from app.ai.prompts import load_script_system, load_script_user_prompt, load_topic_prompt, TOPIC_PROMPT_MAP
|
||||
from app.schemas.script import ScriptGenerationEvent, ScriptShot
|
||||
from app.services.ai_response_utils import (
|
||||
safe_parse_ai_json_response,
|
||||
@@ -156,7 +156,7 @@ class ScriptService:
|
||||
同步生成脚本
|
||||
|
||||
Args:
|
||||
topic: 创作主题（支持视频链接，自动提取文案）
|
||||
topic: 创作主题（预设主题名或自定义输入/视频链接）
|
||||
duration: 视频时长（秒）
|
||||
script_type: 脚本类型
|
||||
model: 指定模型
|
||||
@@ -164,30 +164,39 @@ class ScriptService:
|
||||
Returns:
|
||||
分镜列表
|
||||
"""
|
||||
# 1. 检测并提取视频链接中的文案
|
||||
anytocopy = get_anytocopy_service()
|
||||
extract_result = await anytocopy.extract_text_from_input(topic)
|
||||
# 1. 判断是否为预设主题
|
||||
is_preset_topic = topic in TOPIC_PROMPT_MAP
|
||||
|
||||
if extract_result["error"]:
|
||||
logger.warning(f"视频文案提取失败: {extract_result['error']}")
|
||||
# 提取失败但不中断，使用原始输入
|
||||
# 2. 根据类型决定处理方式
|
||||
actual_topic = topic
|
||||
if not is_preset_topic:
|
||||
# 非预设主题：检测并提取视频链接中的文案
|
||||
anytocopy = get_anytocopy_service()
|
||||
extract_result = await anytocopy.extract_text_from_input(topic)
|
||||
|
||||
if extract_result["is_video_url"]:
|
||||
logger.info(f"检测到视频链接，提取文案长度: {len(extract_result['extracted_text'])}")
|
||||
# 使用提取的文案作为创作主题
|
||||
topic = extract_result["extracted_text"] or topic
|
||||
if extract_result["error"]:
|
||||
logger.warning(f"视频文案提取失败: {extract_result['error']}")
|
||||
# 提取失败但不中断，使用原始输入
|
||||
|
||||
# 2. 获取 model_router
|
||||
if extract_result["is_video_url"]:
|
||||
logger.info(f"检测到视频链接，提取文案长度: {len(extract_result['extracted_text'])}")
|
||||
# 使用提取的文案作为创作主题
|
||||
actual_topic = extract_result["extracted_text"] or topic
|
||||
|
||||
# 3. 获取 model_router
|
||||
model_router = await get_model_router()
|
||||
|
||||
# 加载 Prompt（使用新的 loader）
|
||||
system_prompt = load_script_system()
|
||||
user_prompt = load_script_user(
|
||||
# 4. 加载 Prompt
|
||||
# 系统提示词：预设主题用专用提示词，否则用通用提示词
|
||||
system_prompt = load_topic_prompt(topic) if is_preset_topic else load_script_system()
|
||||
# 用户提示词
|
||||
user_prompt = load_script_user_prompt(
|
||||
topic=topic,
|
||||
duration=duration,
|
||||
script_type=script_type,
|
||||
)
|
||||
|
||||
logger.info(f"同步生成脚本: topic={topic}, is_preset={is_preset_topic}, duration={duration}")
|
||||
|
||||
logger.info(f"同步生成脚本: topic={topic[:20]}, duration={duration}")
|
||||
|
||||
# 调用 AI 生成
|
||||
@@ -240,7 +249,7 @@ class ScriptService:
|
||||
"""
|
||||
流式生成脚本（SSE）- 优化版
|
||||
|
||||
支持视频链接自动提取文案。
|
||||
支持预设主题和视频链接自动提取文案。
|
||||
|
||||
进度设计：
|
||||
- 0-5%: start（初始化）
|
||||
@@ -253,13 +262,19 @@ class ScriptService:
|
||||
model_router = await get_model_router()
|
||||
start_time = time.time()
|
||||
|
||||
# 1. 检测并提取视频链接中的文案
|
||||
# 1. 判断是否为预设主题
|
||||
is_preset_topic = topic in TOPIC_PROMPT_MAP
|
||||
|
||||
# 2. 非预设主题时，检测并提取视频链接中的文案
|
||||
original_topic = topic
|
||||
anytocopy = get_anytocopy_service()
|
||||
extracted_info = None # 保存提取的视频信息
|
||||
actual_topic = topic
|
||||
|
||||
# 检查是否为视频链接
|
||||
if AnyToCopyService.is_video_url(topic) or AnyToCopyService.extract_url_from_text(topic):
|
||||
# 检查是否为视频链接（非预设主题才检测）
|
||||
if not is_preset_topic and (
|
||||
AnyToCopyService.is_video_url(topic) or AnyToCopyService.extract_url_from_text(topic)
|
||||
):
|
||||
yield ScriptGenerationEvent(
|
||||
type="analyzing",
|
||||
progress=5,
|
||||
@@ -300,13 +315,16 @@ class ScriptService:
|
||||
|
||||
try:
|
||||
# 加载 Prompt
|
||||
system_prompt = load_script_system()
|
||||
user_prompt = load_script_user(
|
||||
# 系统提示词：预设主题用专用提示词，否则用通用提示词
|
||||
system_prompt = load_topic_prompt(topic) if is_preset_topic else load_script_system()
|
||||
# 用户提示词
|
||||
user_prompt = load_script_user_prompt(
|
||||
topic=topic,
|
||||
duration=duration,
|
||||
script_type=script_type,
|
||||
)
|
||||
|
||||
logger.info(f"流式生成脚本: topic={topic}, is_preset={is_preset_topic}, duration={duration}")
|
||||
|
||||
# 1. 开始阶段（0-5%）
|
||||
yield ScriptGenerationEvent(
|
||||
type="start",
|
||||
|
||||
@@ -0,0 +1,259 @@
|
||||
"""
|
||||
TTS 服务层
|
||||
==========
|
||||
|
||||
封装 Kling AI TTS API，提供语音合成能力。
|
||||
API 文档：https://klingai.com/document-api
|
||||
"""
|
||||
|
||||
import asyncio
|
||||
import logging
|
||||
from pathlib import Path
|
||||
|
||||
from app.ai.providers.klingai_provider import KlingAIProvider
|
||||
from app.config import get_settings
|
||||
|
||||
logger = logging.getLogger(__name__)
|
||||
|
||||
# Kling TTS API 配置
|
||||
TTS_TASK_TIMEOUT = 120 # TTS 任务最大等待时间（秒）
|
||||
TTS_POLL_INTERVAL = 2.0 # 轮询间隔（秒）
|
||||
|
||||
|
||||
def _get_kling_provider() -> KlingAIProvider:
|
||||
"""获取 KlingAI Provider 实例"""
|
||||
settings = get_settings()
|
||||
config = {
|
||||
"access_key": settings.KLINGAI_ACCESS_KEY or "",
|
||||
"secret_key": settings.KLINGAI_SECRET_KEY or "",
|
||||
}
|
||||
return KlingAIProvider(config)
|
||||
|
||||
|
||||
class TTSService:
|
||||
"""Kling AI TTS 服务客户端"""
|
||||
|
||||
# Kling 官方预设音色（已知音色）
|
||||
PRESET_VOICES = [
|
||||
{
|
||||
"voice_id": "829824295735410756",
|
||||
"name": "钓系女友",
|
||||
"language": "zh",
|
||||
"description": "甜美撒娇",
|
||||
},
|
||||
{
|
||||
"voice_id": "829826751244537879",
|
||||
"name": "温柔女声",
|
||||
"language": "zh",
|
||||
"description": "温柔细腻",
|
||||
},
|
||||
{
|
||||
"voice_id": "829826792415842333",
|
||||
"name": "播报男声",
|
||||
"language": "zh",
|
||||
"description": "沉稳播报",
|
||||
},
|
||||
{
|
||||
"voice_id": "829826834144964676",
|
||||
"name": "盐系少年",
|
||||
"language": "zh",
|
||||
"description": "清新少年",
|
||||
},
|
||||
{
|
||||
"voice_id": "829826884271091753",
|
||||
"name": "撒娇女友",
|
||||
"language": "zh",
|
||||
"description": "可爱撒娇",
|
||||
},
|
||||
]
|
||||
|
||||
def __init__(self) -> None:
|
||||
self.provider = _get_kling_provider()
|
||||
self.default_voice_id = "829826751244537879" # 温柔女声
|
||||
|
||||
async def synthesize_sync(
|
||||
self,
|
||||
text: str,
|
||||
voice_id: str | None = None,
|
||||
speed: float = 1.0,
|
||||
voice_language: str = "zh",
|
||||
) -> str:
|
||||
"""
|
||||
同步合成语音（提交任务并等待完成），返回音频 URL。
|
||||
|
||||
Args:
|
||||
text: 待合成文本（≤1000字符）
|
||||
voice_id: 音色 ID（默认使用温柔女声）
|
||||
speed: 语速 (0.8-2.0)
|
||||
voice_language: 语言 (zh/en)
|
||||
|
||||
Returns:
|
||||
音频 URL
|
||||
|
||||
Raises:
|
||||
ValueError: 参数校验失败
|
||||
TimeoutError: 等待超时
|
||||
"""
|
||||
if not text or not text.strip():
|
||||
raise ValueError("text 不能为空")
|
||||
|
||||
if len(text) > 1000:
|
||||
raise ValueError("text 不能超过 1000 字符")
|
||||
|
||||
voice = voice_id or self.default_voice_id
|
||||
|
||||
# 提交 TTS 任务
|
||||
result = await self.provider.generate_tts(
|
||||
text=text,
|
||||
voice_id=voice,
|
||||
voice_language=voice_language,
|
||||
voice_speed=speed,
|
||||
)
|
||||
|
||||
task_id = result.get("task_id")
|
||||
if not task_id:
|
||||
raise ValueError("TTS 任务提交失败: 未返回 task_id")
|
||||
|
||||
logger.info(f"[TTS] 任务已提交: task_id={task_id}")
|
||||
|
||||
# 等待任务完成
|
||||
audio_url = await self._wait_for_task(task_id)
|
||||
|
||||
return audio_url
|
||||
|
||||
async def _wait_for_task(self, task_id: str) -> str:
|
||||
"""等待 TTS 任务完成并返回音频 URL"""
|
||||
elapsed = 0.0
|
||||
while elapsed < TTS_TASK_TIMEOUT:
|
||||
await asyncio.sleep(TTS_POLL_INTERVAL)
|
||||
elapsed += TTS_POLL_INTERVAL
|
||||
|
||||
result = await self.provider.get_tts_task(task_id)
|
||||
status = result.get("status") or result.get("task_status", "")
|
||||
|
||||
logger.debug(f"[TTS] task_id={task_id}, status={status}, elapsed={elapsed}s")
|
||||
|
||||
if status == "succeed":
|
||||
# 任务成功，返回音频 URL
|
||||
task_result = result.get("task_result", {})
|
||||
audio_url = task_result.get("audio_url") if isinstance(task_result, dict) else None
|
||||
if audio_url:
|
||||
return audio_url
|
||||
# 某些响应格式直接放在 data 中
|
||||
return result.get("audio_url") or result.get("data", {}).get("audio_url", "")
|
||||
|
||||
if status in ("failed", "error"):
|
||||
raise ValueError(f"TTS 任务失败: {result.get('message', '未知错误')}")
|
||||
|
||||
raise TimeoutError(f"TTS 任务等待超时（{TTS_TASK_TIMEOUT}秒）")
|
||||
|
||||
async def synthesize_to_file(
|
||||
self,
|
||||
text: str,
|
||||
output_path: str | Path,
|
||||
voice_id: str | None = None,
|
||||
speed: float = 1.0,
|
||||
voice_language: str = "zh",
|
||||
) -> Path:
|
||||
"""
|
||||
合成语音并保存到文件。
|
||||
|
||||
Args:
|
||||
text: 待合成文本
|
||||
output_path: 输出文件路径
|
||||
voice_id: 音色 ID
|
||||
speed: 语速
|
||||
voice_language: 语言
|
||||
|
||||
Returns:
|
||||
输出文件路径
|
||||
"""
|
||||
import httpx
|
||||
|
||||
output_path = Path(output_path)
|
||||
output_path.parent.mkdir(parents=True, exist_ok=True)
|
||||
|
||||
# 获取音频 URL
|
||||
audio_url = await self.synthesize_sync(
|
||||
text=text,
|
||||
voice_id=voice_id,
|
||||
speed=speed,
|
||||
voice_language=voice_language,
|
||||
)
|
||||
|
||||
# 下载音频并保存
|
||||
async with httpx.AsyncClient(timeout=60.0) as client:
|
||||
response = await client.get(audio_url)
|
||||
response.raise_for_status()
|
||||
audio_bytes = response.content
|
||||
|
||||
output_path.write_bytes(audio_bytes)
|
||||
logger.info(f"[TTS] 语音合成完成: {output_path}")
|
||||
return output_path
|
||||
|
||||
async def batch_synthesize(
|
||||
self,
|
||||
segments: list[dict],
|
||||
output_dir: str | Path,
|
||||
voice_id: str | None = None,
|
||||
speed: float = 1.0,
|
||||
) -> list[dict]:
|
||||
"""
|
||||
批量合成多段语音。
|
||||
|
||||
Args:
|
||||
segments: 分段列表，每项包含 text, index（可选）, filename（可选）
|
||||
output_dir: 输出目录
|
||||
voice_id: 音色 ID
|
||||
speed: 语速
|
||||
|
||||
Returns:
|
||||
结果列表，每项包含 input（原始输入）和 output（输出文件路径或错误信息）
|
||||
"""
|
||||
output_dir = Path(output_dir)
|
||||
output_dir.mkdir(parents=True, exist_ok=True)
|
||||
|
||||
results = []
|
||||
for seg in segments:
|
||||
text = seg.get("text", "")
|
||||
index = seg.get("index", len(results))
|
||||
filename = seg.get("filename", f"audio_{index:04d}.mp3")
|
||||
|
||||
try:
|
||||
output_path = await self.synthesize_to_file(
|
||||
text=text,
|
||||
output_path=output_dir / filename,
|
||||
voice_id=voice_id,
|
||||
speed=speed,
|
||||
)
|
||||
results.append({
|
||||
"index": index,
|
||||
"text": text,
|
||||
"output_path": str(output_path),
|
||||
"success": True,
|
||||
"error": None,
|
||||
})
|
||||
except Exception as e:
|
||||
logger.error(f"[TTS] 分段 {index} 合成失败: {e}")
|
||||
results.append({
|
||||
"index": index,
|
||||
"text": text,
|
||||
"output_path": None,
|
||||
"success": False,
|
||||
"error": str(e),
|
||||
})
|
||||
|
||||
return results
|
||||
|
||||
@staticmethod
|
||||
def get_preset_voices() -> list[dict]:
|
||||
"""获取预设音色列表"""
|
||||
return TTSService.PRESET_VOICES
|
||||
|
||||
@staticmethod
|
||||
def get_voice_by_id(voice_id: str) -> dict | None:
|
||||
"""根据 ID 获取音色信息"""
|
||||
for voice in TTSService.PRESET_VOICES:
|
||||
if voice["voice_id"] == voice_id:
|
||||
return voice
|
||||
return None
|
||||
@@ -0,0 +1,258 @@
|
||||
"""
|
||||
语音克隆服务层
|
||||
=============
|
||||
|
||||
封装 Kling AI 声音克隆 API，提供个性化音色克隆能力。
|
||||
API 文档：https://klingai.com/document-api
|
||||
"""
|
||||
|
||||
import asyncio
|
||||
import logging
|
||||
from enum import Enum
|
||||
|
||||
from app.ai.providers.klingai_provider import KlingAIProvider
|
||||
from app.config import get_settings
|
||||
|
||||
logger = logging.getLogger(__name__)
|
||||
|
||||
# 克隆任务配置
|
||||
CLONE_TASK_TIMEOUT = 600 # 克隆任务最大等待时间（秒）
|
||||
CLONE_POLL_INTERVAL = 5.0 # 轮询间隔（秒）
|
||||
|
||||
|
||||
def _get_kling_provider() -> KlingAIProvider:
|
||||
"""获取 KlingAI Provider 实例"""
|
||||
settings = get_settings()
|
||||
config = {
|
||||
"access_key": settings.KLINGAI_ACCESS_KEY or "",
|
||||
"secret_key": settings.KLINGAI_SECRET_KEY or "",
|
||||
}
|
||||
return KlingAIProvider(config)
|
||||
|
||||
|
||||
class CloneTaskStatus(Enum):
|
||||
"""克隆任务状态（字符串枚举）"""
|
||||
|
||||
PENDING = "pending" # 任务已提交，等待处理
|
||||
PROCESSING = "processing" # 正在处理
|
||||
SUCCEEDED = "succeeded" # 成功
|
||||
FAILED = "failed" # 失败
|
||||
TIMEOUT = "timeout" # 超时
|
||||
|
||||
|
||||
class VoiceCloneService:
|
||||
"""Kling AI 声音克隆服务客户端"""
|
||||
|
||||
def __init__(self) -> None:
|
||||
self.provider = _get_kling_provider()
|
||||
self.timeout = CLONE_TASK_TIMEOUT
|
||||
|
||||
async def submit_clone_task(
|
||||
self,
|
||||
source_audio_url: str | None = None,
|
||||
source_video_url: str | None = None,
|
||||
video_id: str | None = None,
|
||||
voice_name: str | None = None,
|
||||
callback_url: str | None = None,
|
||||
external_task_id: str | None = None,
|
||||
) -> str:
|
||||
"""
|
||||
提交声音克隆任务。
|
||||
|
||||
Args:
|
||||
source_audio_url: 源音频 URL（5-30秒，mp3/wav格式，需公开可访问）
|
||||
source_video_url: 源视频 URL（可选）
|
||||
video_id: 历史作品ID（可选，通过已有作品克隆音色）
|
||||
voice_name: 自定义音色名称（≤20字符）
|
||||
callback_url: 回调地址
|
||||
external_task_id: 自定义任务ID
|
||||
|
||||
Returns:
|
||||
克隆任务 ID
|
||||
|
||||
Raises:
|
||||
ValueError: 参数校验失败
|
||||
"""
|
||||
if not source_audio_url and not source_video_url and not video_id:
|
||||
raise ValueError("必须提供 source_audio_url、source_video_url 或 video_id 之一")
|
||||
|
||||
if source_audio_url and not source_audio_url.startswith(("http://", "https://")):
|
||||
raise ValueError("source_audio_url 必须是有效的 URL")
|
||||
|
||||
if source_video_url and not source_video_url.startswith(("http://", "https://")):
|
||||
raise ValueError("source_video_url 必须是有效的 URL")
|
||||
|
||||
if voice_name and len(voice_name) > 20:
|
||||
raise ValueError("voice_name 不能超过 20 字符")
|
||||
|
||||
# 提交克隆任务
|
||||
result = await self.provider.create_custom_voice(
|
||||
voice_name=voice_name or "自定义音色",
|
||||
audio_url=source_audio_url,
|
||||
video_url=source_video_url,
|
||||
video_id=video_id,
|
||||
callback_url=callback_url,
|
||||
external_task_id=external_task_id,
|
||||
)
|
||||
|
||||
# Kling API 返回 task_id
|
||||
task_id = result.get("task_id")
|
||||
if not task_id:
|
||||
raise ValueError("提交克隆任务失败: 未返回 task_id")
|
||||
|
||||
logger.info(f"[VoiceClone] 提交任务成功: task_id={task_id}")
|
||||
return task_id
|
||||
|
||||
async def query_clone_task(self, task_id: str, blocking: bool = False) -> dict:
|
||||
"""
|
||||
查询声音克隆任务状态。
|
||||
|
||||
Args:
|
||||
task_id: 任务 ID
|
||||
blocking: 是否阻塞等待（False 则立即返回当前状态）
|
||||
|
||||
Returns:
|
||||
任务状态信息，包含字段:
|
||||
- task_id: 任务 ID
|
||||
- status: 任务状态 (pending/processing/succeeded/failed/timeout)
|
||||
- voice_id: 克隆成功的音色 ID（如已完成）
|
||||
- trial_url: 试听地址（如已完成）
|
||||
- error_message: 错误信息（如失败）
|
||||
"""
|
||||
# Kling 使用不同的查询接口
|
||||
result = await self.provider.get_custom_voice_task(task_id)
|
||||
|
||||
status = result.get("task_status", "pending")
|
||||
|
||||
# 映射状态
|
||||
status_map = {
|
||||
"pending": CloneTaskStatus.PENDING.value,
|
||||
"processing": CloneTaskStatus.PROCESSING.value,
|
||||
"succeed": CloneTaskStatus.SUCCEEDED.value,
|
||||
"failed": CloneTaskStatus.FAILED.value,
|
||||
}
|
||||
mapped_status = status_map.get(status, status)
|
||||
|
||||
ret = {
|
||||
"task_id": task_id,
|
||||
"status": mapped_status,
|
||||
"voice_id": None,
|
||||
"trial_url": None,
|
||||
"error_message": None,
|
||||
}
|
||||
|
||||
# 提取音色信息
|
||||
if mapped_status == CloneTaskStatus.SUCCEEDED.value:
|
||||
task_result = result.get("task_result", {})
|
||||
if isinstance(task_result, dict):
|
||||
voices = task_result.get("voices", [])
|
||||
if voices and len(voices) > 0:
|
||||
ret["voice_id"] = voices[0].get("voice_id")
|
||||
ret["trial_url"] = voices[0].get("trial_url")
|
||||
|
||||
if mapped_status == CloneTaskStatus.FAILED.value:
|
||||
ret["error_message"] = result.get("message", "任务失败")
|
||||
|
||||
if blocking and mapped_status in (CloneTaskStatus.PENDING.value, CloneTaskStatus.PROCESSING.value):
|
||||
ret = await self._wait_for_completion(task_id)
|
||||
|
||||
return ret
|
||||
|
||||
async def _wait_for_completion(self, task_id: str, poll_interval: float = CLONE_POLL_INTERVAL) -> dict:
|
||||
"""
|
||||
阻塞等待克隆任务完成。
|
||||
|
||||
Args:
|
||||
task_id: 任务 ID
|
||||
poll_interval: 轮询间隔（秒）
|
||||
|
||||
Returns:
|
||||
最终任务状态
|
||||
"""
|
||||
elapsed = 0.0
|
||||
while elapsed < self.timeout:
|
||||
await asyncio.sleep(poll_interval)
|
||||
elapsed += poll_interval
|
||||
|
||||
result = await self.query_clone_task(task_id, blocking=False)
|
||||
status = result.get("status", "pending")
|
||||
|
||||
logger.debug(f"[VoiceClone] task_id={task_id}, status={status}, elapsed={elapsed}s")
|
||||
|
||||
if status in (CloneTaskStatus.SUCCEEDED.value, CloneTaskStatus.FAILED.value):
|
||||
return result
|
||||
|
||||
# 超时
|
||||
logger.warning(f"[VoiceClone] task_id={task_id} 等待超时")
|
||||
return {
|
||||
"task_id": task_id,
|
||||
"status": CloneTaskStatus.TIMEOUT.value,
|
||||
"voice_id": None,
|
||||
"trial_url": None,
|
||||
"error_message": f"等待超时（{self.timeout}秒）",
|
||||
}
|
||||
|
||||
async def wait_for_clone(
|
||||
self,
|
||||
source_audio_url: str | None = None,
|
||||
source_video_url: str | None = None,
|
||||
video_id: str | None = None,
|
||||
voice_name: str | None = None,
|
||||
poll_interval: float = CLONE_POLL_INTERVAL,
|
||||
) -> dict:
|
||||
"""
|
||||
一站式：提交克隆任务并等待完成。
|
||||
|
||||
Args:
|
||||
source_audio_url: 源音频 URL
|
||||
source_video_url: 源视频 URL
|
||||
video_id: 历史作品ID
|
||||
voice_name: 自定义音色名称
|
||||
poll_interval: 轮询间隔
|
||||
|
||||
Returns:
|
||||
最终任务状态
|
||||
|
||||
Raises:
|
||||
ValueError: 提交失败
|
||||
TimeoutError: 等待超时
|
||||
"""
|
||||
task_id = await self.submit_clone_task(
|
||||
source_audio_url=source_audio_url,
|
||||
source_video_url=source_video_url,
|
||||
video_id=video_id,
|
||||
voice_name=voice_name,
|
||||
)
|
||||
|
||||
result = await self.query_clone_task(task_id, blocking=False)
|
||||
status = result.get("status", "pending")
|
||||
|
||||
if status == CloneTaskStatus.SUCCEEDED.value:
|
||||
logger.info(f"[VoiceClone] 克隆成功: task_id={task_id}")
|
||||
return result
|
||||
|
||||
# 阻塞等待
|
||||
result = await self._wait_for_completion(task_id, poll_interval=poll_interval)
|
||||
return result
|
||||
|
||||
async def list_custom_voices(self) -> list[dict]:
|
||||
"""
|
||||
查询自定义音色列表。
|
||||
|
||||
Returns:
|
||||
自定义音色列表
|
||||
"""
|
||||
return await self.provider.list_custom_voices()
|
||||
|
||||
async def delete_custom_voice(self, voice_id: str) -> bool:
|
||||
"""
|
||||
删除自定义音色。
|
||||
|
||||
Args:
|
||||
voice_id: 音色 ID
|
||||
|
||||
Returns:
|
||||
是否删除成功
|
||||
"""
|
||||
result = await self.provider.delete_custom_voice(voice_id)
|
||||
return result.get("code") == 0
|
||||
@@ -1,98 +1,46 @@
|
||||
services:
|
||||
# PostgreSQL 数据库
|
||||
db:
|
||||
image: postgres:15-alpine
|
||||
container_name: meijiaka-db
|
||||
environment:
|
||||
POSTGRES_USER: postgres
|
||||
POSTGRES_PASSWORD: postgres
|
||||
POSTGRES_DB: meijiaka
|
||||
volumes:
|
||||
- postgres_data:/var/lib/postgresql/data
|
||||
ports:
|
||||
- "5432:5432"
|
||||
healthcheck:
|
||||
test: ["CMD-SHELL", "pg_isready -U postgres"]
|
||||
interval: 5s
|
||||
timeout: 5s
|
||||
retries: 5
|
||||
networks:
|
||||
- meijiaka-network
|
||||
|
||||
# Redis 缓存
|
||||
redis:
|
||||
image: redis:7-alpine
|
||||
container_name: meijiaka-redis
|
||||
volumes:
|
||||
- redis_data:/data
|
||||
ports:
|
||||
- "6379:6379"
|
||||
healthcheck:
|
||||
test: ["CMD", "redis-cli", "ping"]
|
||||
interval: 5s
|
||||
timeout: 5s
|
||||
retries: 5
|
||||
networks:
|
||||
- meijiaka-network
|
||||
|
||||
# FastAPI 应用（开发模式）
|
||||
api:
|
||||
build:
|
||||
context: .
|
||||
dockerfile: Dockerfile
|
||||
container_name: meijiaka-api
|
||||
container_name: meijiaka-zj-api
|
||||
environment:
|
||||
- ENV=development
|
||||
- DEBUG=true
|
||||
- DATABASE_URL=postgresql+asyncpg://postgres:postgres@db:5432/meijiaka
|
||||
- DATABASE_URL=postgresql+asyncpg://postgres:postgres@db:5432/meijiaka_zj
|
||||
- REDIS_HOST=redis
|
||||
- REDIS_PORT=6379
|
||||
- REDIS_DB=0
|
||||
- REDIS_DB=1
|
||||
- SECRET_KEY=dev-secret-key-change-in-production
|
||||
volumes:
|
||||
- .:/app
|
||||
- ~/Documents/Meijiaka:/root/Documents/Meijiaka
|
||||
- ~/Documents/Meijiaka-zj:/root/Documents/Meijiaka-zj
|
||||
ports:
|
||||
- "8080:8000"
|
||||
depends_on:
|
||||
db:
|
||||
condition: service_healthy
|
||||
redis:
|
||||
condition: service_healthy
|
||||
- "8081:8000"
|
||||
command: uvicorn app.main:app --host 0.0.0.0 --port 8000 --reload
|
||||
networks:
|
||||
- meijiaka-network
|
||||
|
||||
# Async Engine Scheduler: 统一调度所有第三方异步任务
|
||||
scheduler:
|
||||
build:
|
||||
context: .
|
||||
dockerfile: Dockerfile
|
||||
container_name: meijiaka-scheduler
|
||||
container_name: meijiaka-zj-scheduler
|
||||
environment:
|
||||
- ENV=development
|
||||
- DEBUG=true
|
||||
- DATABASE_URL=postgresql+asyncpg://postgres:postgres@db:5432/meijiaka
|
||||
- DATABASE_URL=postgresql+asyncpg://postgres:postgres@db:5432/meijiaka_zj
|
||||
- REDIS_HOST=redis
|
||||
- REDIS_PORT=6379
|
||||
- REDIS_DB=0
|
||||
- REDIS_DB=1
|
||||
- SECRET_KEY=dev-secret-key-change-in-production
|
||||
volumes:
|
||||
- .:/app
|
||||
- ~/Documents/Meijiaka:/root/Documents/Meijiaka
|
||||
depends_on:
|
||||
db:
|
||||
condition: service_healthy
|
||||
redis:
|
||||
condition: service_healthy
|
||||
- ~/Documents/Meijiaka-zj:/root/Documents/Meijiaka-zj
|
||||
command: python -m app.scheduler.main
|
||||
networks:
|
||||
- meijiaka-network
|
||||
|
||||
volumes:
|
||||
postgres_data:
|
||||
redis_data:
|
||||
|
||||
networks:
|
||||
meijiaka-network:
|
||||
driver: bridge
|
||||
external: true
|
||||
|
||||
@@ -0,0 +1,168 @@
|
||||
# This file was autogenerated by uv via the following command:
|
||||
# uv pip compile pyproject.toml -o requirements.lock
|
||||
aiohappyeyeballs==2.6.1
|
||||
# via aiohttp
|
||||
aiohttp==3.13.5
|
||||
# via meijiaka-ai-api (pyproject.toml)
|
||||
aiosignal==1.4.0
|
||||
# via aiohttp
|
||||
annotated-doc==0.0.4
|
||||
# via fastapi
|
||||
annotated-types==0.7.0
|
||||
# via pydantic
|
||||
anyio==4.13.0
|
||||
# via
|
||||
# httpx
|
||||
# openai
|
||||
# starlette
|
||||
# volcengine-python-sdk
|
||||
# watchfiles
|
||||
asyncpg==0.30.0
|
||||
# via meijiaka-ai-api (pyproject.toml)
|
||||
attrs==26.1.0
|
||||
# via aiohttp
|
||||
bcrypt==4.2.1
|
||||
# via
|
||||
# meijiaka-ai-api (pyproject.toml)
|
||||
# passlib
|
||||
certifi==2026.2.25
|
||||
# via
|
||||
# httpcore
|
||||
# httpx
|
||||
# requests
|
||||
# volcengine-python-sdk
|
||||
cffi==2.0.0
|
||||
# via cryptography
|
||||
charset-normalizer==3.4.7
|
||||
# via requests
|
||||
click==8.3.2
|
||||
# via uvicorn
|
||||
cryptography==46.0.6
|
||||
# via
|
||||
# python-jose
|
||||
# volcengine-python-sdk
|
||||
distro==1.9.0
|
||||
# via openai
|
||||
ecdsa==0.19.2
|
||||
# via python-jose
|
||||
fastapi==0.135.3
|
||||
# via meijiaka-ai-api (pyproject.toml)
|
||||
frozenlist==1.8.0
|
||||
# via
|
||||
# aiohttp
|
||||
# aiosignal
|
||||
greenlet==3.3.2
|
||||
# via sqlalchemy
|
||||
h11==0.16.0
|
||||
# via
|
||||
# httpcore
|
||||
# uvicorn
|
||||
httpcore==1.0.9
|
||||
# via httpx
|
||||
httptools==0.7.1
|
||||
# via uvicorn
|
||||
httpx==0.28.1
|
||||
# via
|
||||
# meijiaka-ai-api (pyproject.toml)
|
||||
# openai
|
||||
# volcengine-python-sdk
|
||||
idna==3.11
|
||||
# via
|
||||
# anyio
|
||||
# httpx
|
||||
# requests
|
||||
# yarl
|
||||
jiter==0.13.0
|
||||
# via openai
|
||||
multidict==6.7.1
|
||||
# via
|
||||
# aiohttp
|
||||
# yarl
|
||||
openai==1.58.1
|
||||
# via meijiaka-ai-api (pyproject.toml)
|
||||
orjson==3.11.8
|
||||
# via meijiaka-ai-api (pyproject.toml)
|
||||
passlib==1.7.4
|
||||
# via meijiaka-ai-api (pyproject.toml)
|
||||
propcache==0.4.1
|
||||
# via
|
||||
# aiohttp
|
||||
# yarl
|
||||
pyasn1==0.4.8
|
||||
# via
|
||||
# python-jose
|
||||
# rsa
|
||||
pycparser==3.0
|
||||
# via cffi
|
||||
pydantic==2.9.2
|
||||
# via
|
||||
# meijiaka-ai-api (pyproject.toml)
|
||||
# fastapi
|
||||
# openai
|
||||
# pydantic-settings
|
||||
# volcengine-python-sdk
|
||||
pydantic-core==2.23.4
|
||||
# via pydantic
|
||||
pydantic-settings==2.6.1
|
||||
# via meijiaka-ai-api (pyproject.toml)
|
||||
python-dateutil==2.9.0.post0
|
||||
# via volcengine-python-sdk
|
||||
python-dotenv==1.2.2
|
||||
# via
|
||||
# pydantic-settings
|
||||
# uvicorn
|
||||
python-jose==3.4.0
|
||||
# via meijiaka-ai-api (pyproject.toml)
|
||||
python-multipart==0.0.24
|
||||
# via meijiaka-ai-api (pyproject.toml)
|
||||
pyyaml==6.0.3
|
||||
# via
|
||||
# meijiaka-ai-api (pyproject.toml)
|
||||
# uvicorn
|
||||
qiniu==7.13.2
|
||||
# via meijiaka-ai-api (pyproject.toml)
|
||||
redis==5.2.1
|
||||
# via meijiaka-ai-api (pyproject.toml)
|
||||
requests==2.33.1
|
||||
# via qiniu
|
||||
rsa==4.9.1
|
||||
# via python-jose
|
||||
six==1.17.0
|
||||
# via
|
||||
# ecdsa
|
||||
# python-dateutil
|
||||
# volcengine-python-sdk
|
||||
sniffio==1.3.1
|
||||
# via openai
|
||||
sqlalchemy==2.0.49
|
||||
# via meijiaka-ai-api (pyproject.toml)
|
||||
starlette==1.0.0
|
||||
# via fastapi
|
||||
tqdm==4.67.3
|
||||
# via openai
|
||||
typing-extensions==4.15.0
|
||||
# via
|
||||
# fastapi
|
||||
# openai
|
||||
# pydantic
|
||||
# pydantic-core
|
||||
# sqlalchemy
|
||||
# typing-inspection
|
||||
typing-inspection==0.4.2
|
||||
# via fastapi
|
||||
urllib3==2.6.3
|
||||
# via
|
||||
# requests
|
||||
# volcengine-python-sdk
|
||||
uvicorn==0.32.1
|
||||
# via meijiaka-ai-api (pyproject.toml)
|
||||
uvloop==0.22.1
|
||||
# via uvicorn
|
||||
volcengine-python-sdk==5.0.22
|
||||
# via meijiaka-ai-api (pyproject.toml)
|
||||
watchfiles==1.1.1
|
||||
# via uvicorn
|
||||
websockets==16.0
|
||||
# via uvicorn
|
||||
yarl==1.23.0
|
||||
# via aiohttp
|
||||
@@ -0,0 +1,8 @@
|
||||
"""
|
||||
pytest 配置文件
|
||||
"""
|
||||
import sys
|
||||
from pathlib import Path
|
||||
|
||||
# 将 app 目录添加到 Python 路径
|
||||
sys.path.insert(0, str(Path(__file__).parent.parent))
|
||||
@@ -0,0 +1,299 @@
|
||||
"""
|
||||
Kling TTS 服务单元测试
|
||||
"""
|
||||
import pytest
|
||||
from unittest.mock import AsyncMock, patch, MagicMock
|
||||
import httpx
|
||||
|
||||
|
||||
class TestTTSService:
|
||||
"""TTS 服务测试"""
|
||||
|
||||
@pytest.fixture
|
||||
def mock_provider(self):
|
||||
"""创建模拟的 KlingAIProvider"""
|
||||
mock = MagicMock()
|
||||
mock.generate_tts = AsyncMock(return_value={"task_id": "test-task-123"})
|
||||
mock.get_tts_task = AsyncMock(return_value={
|
||||
"status": "succeed",
|
||||
"task_result": {"audio_url": "https://example.com/audio.mp3"}
|
||||
})
|
||||
return mock
|
||||
|
||||
@pytest.fixture
|
||||
def tts_service(self, mock_provider):
|
||||
"""创建 TTSService 实例（带 mock provider）"""
|
||||
from app.services.tts_service import TTSService
|
||||
service = TTSService.__new__(TTSService) # 不调用 __init__
|
||||
service.provider = mock_provider
|
||||
service.default_voice_id = "829826751244537879"
|
||||
return service
|
||||
|
||||
def test_tts_service_init(self, tts_service):
|
||||
"""测试 TTSService 初始化"""
|
||||
assert tts_service is not None
|
||||
assert hasattr(tts_service, "provider")
|
||||
assert hasattr(tts_service, "default_voice_id")
|
||||
|
||||
def test_preset_voices_loaded(self, tts_service):
|
||||
"""测试预设音色已加载"""
|
||||
voices = tts_service.get_preset_voices()
|
||||
assert len(voices) > 0
|
||||
assert any(v["voice_id"] == "829826751244537879" for v in voices)
|
||||
|
||||
def test_get_voice_by_id_found(self, tts_service):
|
||||
"""测试根据 ID 获取音色 - 找到"""
|
||||
voice = tts_service.get_voice_by_id("829826751244537879")
|
||||
assert voice is not None
|
||||
assert voice["name"] == "温柔女声"
|
||||
|
||||
def test_get_voice_by_id_not_found(self, tts_service):
|
||||
"""测试根据 ID 获取音色 - 未找到"""
|
||||
voice = tts_service.get_voice_by_id("non-existent-id")
|
||||
assert voice is None
|
||||
|
||||
@pytest.mark.asyncio
|
||||
async def test_synthesize_sync_success(self, tts_service):
|
||||
"""测试同步合成成功"""
|
||||
with patch.object(tts_service.provider, "generate_tts", new_callable=AsyncMock) as mock_generate:
|
||||
mock_generate.return_value = {"task_id": "test-task-123"}
|
||||
|
||||
with patch.object(tts_service.provider, "get_tts_task", new_callable=AsyncMock) as mock_get:
|
||||
# 第一次调用：pending
|
||||
mock_get.return_value = {"status": "pending"}
|
||||
# 第二次调用：succeed
|
||||
mock_get.return_value = {
|
||||
"status": "succeed",
|
||||
"task_result": {"audio_url": "https://kling.example.com/audio.mp3"}
|
||||
}
|
||||
|
||||
result = await tts_service.synthesize_sync(
|
||||
text="你好世界",
|
||||
voice_id="829826751244537879",
|
||||
speed=1.0,
|
||||
voice_language="zh"
|
||||
)
|
||||
|
||||
assert result == "https://kling.example.com/audio.mp3"
|
||||
|
||||
@pytest.mark.asyncio
|
||||
async def test_synthesize_sync_empty_text(self, tts_service):
|
||||
"""测试空文本"""
|
||||
with pytest.raises(ValueError, match="text 不能为空"):
|
||||
await tts_service.synthesize_sync(text="", voice_id="test")
|
||||
|
||||
@pytest.mark.asyncio
|
||||
async def test_synthesize_sync_text_too_long(self, tts_service):
|
||||
"""测试文本超长（验证在 API 调用之前）"""
|
||||
# 中文每个字是 1 个字符，"测" * 1200 = 1200 个字符
|
||||
long_text = "测" * 1201 # 超过 1000
|
||||
with pytest.raises(ValueError, match="text 不能超过"):
|
||||
await tts_service.synthesize_sync(text=long_text, voice_id="test")
|
||||
|
||||
@pytest.mark.asyncio
|
||||
async def test_synthesize_sync_no_task_id(self, tts_service):
|
||||
"""测试提交任务未返回 task_id"""
|
||||
with patch.object(tts_service.provider, "generate_tts", new_callable=AsyncMock) as mock_generate:
|
||||
mock_generate.return_value = {} # 没有 task_id
|
||||
|
||||
with pytest.raises(ValueError, match="任务提交失败"):
|
||||
await tts_service.synthesize_sync(text="你好", voice_id="test")
|
||||
|
||||
@pytest.mark.asyncio
|
||||
async def test_synthesize_sync_task_failed(self, tts_service):
|
||||
"""测试任务失败"""
|
||||
with patch.object(tts_service.provider, "generate_tts", new_callable=AsyncMock) as mock_generate:
|
||||
mock_generate.return_value = {"task_id": "test-task-123"}
|
||||
|
||||
with patch.object(tts_service.provider, "get_tts_task", new_callable=AsyncMock) as mock_get:
|
||||
mock_get.return_value = {"status": "failed", "message": "服务器错误"}
|
||||
|
||||
with pytest.raises(ValueError, match="任务失败"):
|
||||
await tts_service.synthesize_sync(text="你好", voice_id="test")
|
||||
|
||||
def test_get_preset_voices_static(self):
|
||||
"""测试静态方法获取预设音色"""
|
||||
from app.services.tts_service import TTSService
|
||||
voices = TTSService.get_preset_voices()
|
||||
assert len(voices) == 5
|
||||
# 验证默认音色存在
|
||||
assert any(v["voice_id"] == "829826751244537879" for v in voices)
|
||||
|
||||
|
||||
class TestVoiceCloneService:
|
||||
"""声音克隆服务测试"""
|
||||
|
||||
@pytest.fixture
|
||||
def clone_service(self):
|
||||
"""创建 VoiceCloneService 实例"""
|
||||
from app.services.voice_clone_service import VoiceCloneService
|
||||
return VoiceCloneService()
|
||||
|
||||
def test_voice_clone_service_init(self, clone_service):
|
||||
"""测试 VoiceCloneService 初始化"""
|
||||
assert clone_service is not None
|
||||
assert hasattr(clone_service, "provider")
|
||||
assert hasattr(clone_service, "timeout")
|
||||
|
||||
@pytest.mark.asyncio
|
||||
async def test_submit_clone_with_audio_url(self, clone_service):
|
||||
"""测试使用音频 URL 提交克隆任务"""
|
||||
with patch.object(clone_service.provider, "create_custom_voice", new_callable=AsyncMock) as mock_create:
|
||||
mock_create.return_value = {"task_id": "clone-task-123"}
|
||||
|
||||
task_id = await clone_service.submit_clone_task(
|
||||
source_audio_url="https://example.com/source.mp3",
|
||||
voice_name="我的克隆音色"
|
||||
)
|
||||
|
||||
assert task_id == "clone-task-123"
|
||||
mock_create.assert_called_once()
|
||||
call_kwargs = mock_create.call_args.kwargs
|
||||
assert call_kwargs["audio_url"] == "https://example.com/source.mp3"
|
||||
assert call_kwargs["voice_name"] == "我的克隆音色"
|
||||
|
||||
@pytest.mark.asyncio
|
||||
async def test_submit_clone_with_video_url(self, clone_service):
|
||||
"""测试使用视频 URL 提交克隆任务"""
|
||||
with patch.object(clone_service.provider, "create_custom_voice", new_callable=AsyncMock) as mock_create:
|
||||
mock_create.return_value = {"task_id": "clone-task-456"}
|
||||
|
||||
task_id = await clone_service.submit_clone_task(
|
||||
source_video_url="https://example.com/source.mp4",
|
||||
voice_name="视频音色"
|
||||
)
|
||||
|
||||
assert task_id == "clone-task-456"
|
||||
mock_create.assert_called_once()
|
||||
call_kwargs = mock_create.call_args.kwargs
|
||||
assert call_kwargs["video_url"] == "https://example.com/source.mp4"
|
||||
|
||||
@pytest.mark.asyncio
|
||||
async def test_submit_clone_no_source(self, clone_service):
|
||||
"""测试没有提供任何来源时抛出异常"""
|
||||
with pytest.raises(ValueError, match="必须提供"):
|
||||
await clone_service.submit_clone_task()
|
||||
|
||||
@pytest.mark.asyncio
|
||||
async def test_submit_clone_invalid_audio_url(self, clone_service):
|
||||
"""测试无效的音频 URL"""
|
||||
with pytest.raises(ValueError, match="有效的 URL"):
|
||||
await clone_service.submit_clone_task(source_audio_url="not-a-url")
|
||||
|
||||
@pytest.mark.asyncio
|
||||
async def test_submit_clone_voice_name_too_long(self, clone_service):
|
||||
"""测试音色名称过长"""
|
||||
with pytest.raises(ValueError, match="不能超过"):
|
||||
await clone_service.submit_clone_task(
|
||||
source_audio_url="https://example.com/audio.mp3",
|
||||
voice_name="a" * 25 # 超过 20 字符
|
||||
)
|
||||
|
||||
@pytest.mark.asyncio
|
||||
async def test_submit_clone_no_task_id(self, clone_service):
|
||||
"""测试提交任务未返回 task_id"""
|
||||
with patch.object(clone_service.provider, "create_custom_voice", new_callable=AsyncMock) as mock_create:
|
||||
mock_create.return_value = {} # 没有 task_id
|
||||
|
||||
with pytest.raises(ValueError, match="未返回 task_id"):
|
||||
await clone_service.submit_clone_task(
|
||||
source_audio_url="https://example.com/audio.mp3"
|
||||
)
|
||||
|
||||
@pytest.mark.asyncio
|
||||
async def test_query_clone_task_success(self, clone_service):
|
||||
"""测试查询克隆任务状态 - 成功"""
|
||||
with patch.object(clone_service.provider, "get_custom_voice_task", new_callable=AsyncMock) as mock_get:
|
||||
mock_get.return_value = {
|
||||
"task_status": "succeed",
|
||||
"task_result": {
|
||||
"voices": [
|
||||
{"voice_id": "custom-voice-001", "trial_url": "https://example.com/trial.mp3"}
|
||||
]
|
||||
}
|
||||
}
|
||||
|
||||
result = await clone_service.query_clone_task("clone-task-123")
|
||||
|
||||
assert result["task_id"] == "clone-task-123"
|
||||
assert result["status"] == "succeeded"
|
||||
assert result["voice_id"] == "custom-voice-001"
|
||||
assert result["trial_url"] == "https://example.com/trial.mp3"
|
||||
|
||||
@pytest.mark.asyncio
|
||||
async def test_query_clone_task_pending(self, clone_service):
|
||||
"""测试查询克隆任务状态 - 待处理"""
|
||||
with patch.object(clone_service.provider, "get_custom_voice_task", new_callable=AsyncMock) as mock_get:
|
||||
mock_get.return_value = {"task_status": "pending"}
|
||||
|
||||
result = await clone_service.query_clone_task("clone-task-123")
|
||||
|
||||
assert result["task_id"] == "clone-task-123"
|
||||
assert result["status"] == "pending"
|
||||
assert result["voice_id"] is None
|
||||
|
||||
@pytest.mark.asyncio
|
||||
async def test_query_clone_task_failed(self, clone_service):
|
||||
"""测试查询克隆任务状态 - 失败"""
|
||||
with patch.object(clone_service.provider, "get_custom_voice_task", new_callable=AsyncMock) as mock_get:
|
||||
mock_get.return_value = {"task_status": "failed", "message": "音频格式不支持"}
|
||||
|
||||
result = await clone_service.query_clone_task("clone-task-123")
|
||||
|
||||
assert result["status"] == "failed"
|
||||
assert "音频格式不支持" in result["error_message"]
|
||||
|
||||
@pytest.mark.asyncio
|
||||
async def test_delete_custom_voice_success(self, clone_service):
|
||||
"""测试删除自定义音色 - 成功"""
|
||||
with patch.object(clone_service.provider, "delete_custom_voice", new_callable=AsyncMock) as mock_delete:
|
||||
mock_delete.return_value = {"code": 0, "message": "success"}
|
||||
|
||||
result = await clone_service.delete_custom_voice("custom-voice-001")
|
||||
|
||||
assert result is True
|
||||
mock_delete.assert_called_once_with("custom-voice-001")
|
||||
|
||||
@pytest.mark.asyncio
|
||||
async def test_delete_custom_voice_failure(self, clone_service):
|
||||
"""测试删除自定义音色 - 失败"""
|
||||
with patch.object(clone_service.provider, "delete_custom_voice", new_callable=AsyncMock) as mock_delete:
|
||||
mock_delete.return_value = {"code": 400, "message": "Invalid voice id"}
|
||||
|
||||
result = await clone_service.delete_custom_voice("non-existent")
|
||||
|
||||
assert result is False
|
||||
|
||||
@pytest.mark.asyncio
|
||||
async def test_wait_for_clone_immediate_success(self, clone_service):
|
||||
"""测试立即克隆成功"""
|
||||
with patch.object(clone_service.provider, "create_custom_voice", new_callable=AsyncMock) as mock_create:
|
||||
mock_create.return_value = {"task_id": "clone-task-123"}
|
||||
|
||||
with patch.object(clone_service.provider, "get_custom_voice_task", new_callable=AsyncMock) as mock_get:
|
||||
# 第一次查询就成功
|
||||
mock_get.return_value = {
|
||||
"task_status": "succeed",
|
||||
"task_result": {"voices": [{"voice_id": "v1"}]}
|
||||
}
|
||||
|
||||
result = await clone_service.wait_for_clone(
|
||||
source_audio_url="https://example.com/audio.mp3"
|
||||
)
|
||||
|
||||
assert result["status"] == "succeeded"
|
||||
assert result["voice_id"] == "v1"
|
||||
|
||||
|
||||
class TestCloneTaskStatus:
|
||||
"""克隆任务状态枚举测试"""
|
||||
|
||||
def test_clone_task_status_values(self):
|
||||
"""测试状态枚举值"""
|
||||
from app.services.voice_clone_service import CloneTaskStatus
|
||||
|
||||
assert CloneTaskStatus.PENDING.value == "pending"
|
||||
assert CloneTaskStatus.PROCESSING.value == "processing"
|
||||
assert CloneTaskStatus.SUCCEEDED.value == "succeeded"
|
||||
assert CloneTaskStatus.FAILED.value == "failed"
|
||||
assert CloneTaskStatus.TIMEOUT.value == "timeout"
|
||||
Generated
+2096
File diff suppressed because it is too large
Load Diff
+1
-1
@@ -1,3 +1,3 @@
|
||||
# Vite 开发环境变量
|
||||
# 前端默认连接的 Python API 地址
|
||||
VITE_API_BASE_URL=http://127.0.0.1:8080/api/v1
|
||||
VITE_API_BASE_URL=http://127.0.0.1:8081/api/v1
|
||||
|
||||
@@ -1,19 +0,0 @@
|
||||
[Script Info]
|
||||
; Script generated by Meijiaka AI Video
|
||||
; Style: 抖音美好体 - 抖音风格字幕
|
||||
ScriptType: v4.00+
|
||||
PlayResX: 1920
|
||||
PlayResY: 1080
|
||||
ScaledBorderAndShadow: yes
|
||||
Video Aspect Ratio: 0
|
||||
Video Rate: 25
|
||||
Audio Rate: 48000
|
||||
|
||||
[V4+ Styles]
|
||||
Format: Name, Fontname, Fontsize, PrimaryColour, SecondaryColour, OutlineColour, BackColour, Bold, Italic, Underline, StrikeOut, ScaleX, ScaleY, Spacing, Angle, BorderStyle, Outline, Shadow, Alignment, MarginL, MarginR, MarginV, Encoding
|
||||
Style: Douyin-Diamond,抖音美好体,72,&H00FFFFFF,&H00000000,&H00141414,&H00000000,1,0,0,0,100,100,0,0,1,3,0,2,20,20,80,1
|
||||
Style: Douyin-Bold,抖音美好体,64,&H00FFFFFF,&H00000000,&H00141414,&H00000000,1,0,0,0,100,100,0,0,1,2,0,2,20,20,60,1
|
||||
Style: Douyin-Small,抖音美好体,54,&H00EEEEEE,&H00000000,&H00141414,&H00000000,1,0,0,0,100,100,0,0,1,2,0,2,20,20,40,1
|
||||
|
||||
[Events]
|
||||
Format: Layer, Start, End, Style, Name, MarginL, MarginR, MarginV, Effect, Text
|
||||
@@ -5,7 +5,7 @@
|
||||
<meta charset="UTF-8" />
|
||||
<link rel="icon" type="image/svg+xml" href="/vite.svg" />
|
||||
<meta name="viewport" content="width=device-width, initial-scale=1.0" />
|
||||
<title>美家卡 智影</title>
|
||||
<title>美家卡 智剪</title>
|
||||
<style>
|
||||
@font-face {
|
||||
font-family: 'DouyinSans';
|
||||
|
||||
Generated
+6096
File diff suppressed because it is too large
Load Diff
@@ -9,7 +9,7 @@
|
||||
{
|
||||
"identifier": "opener:allow-open-path",
|
||||
"allow": [
|
||||
{ "path": "$DOCUMENT/Meijiaka/**" },
|
||||
{ "path": "$DOCUMENT/Meijiaka-zj/**" },
|
||||
{ "path": "$DOCUMENT/**" },
|
||||
{ "path": "$APPLOCALDATA/**" },
|
||||
{ "path": "$APPDATA/**" },
|
||||
|
||||
@@ -9,3 +9,4 @@ pub mod auth_state;
|
||||
pub mod avatar;
|
||||
pub mod product;
|
||||
pub mod project;
|
||||
pub mod voice;
|
||||
|
||||
@@ -0,0 +1,111 @@
|
||||
//! Voice 音频管理 IPC 命令
|
||||
|
||||
use crate::ApiResponse;
|
||||
use crate::storage::voice as voice_storage;
|
||||
|
||||
#[derive(serde::Deserialize)]
|
||||
#[serde(rename_all = "camelCase")]
|
||||
pub struct SaveAudioArgs {
|
||||
pub project_id: String,
|
||||
pub audio_id: String,
|
||||
pub audio_data: String,
|
||||
pub name: String,
|
||||
pub voice_id: String,
|
||||
pub duration: f64,
|
||||
}
|
||||
|
||||
/// 保存音频文件（前端传入 base64 编码）
|
||||
#[tauri::command]
|
||||
pub async fn save_audio(
|
||||
args: SaveAudioArgs,
|
||||
) -> ApiResponse<voice_storage::AudioMeta> {
|
||||
let audio_bytes = match base64::Engine::decode(
|
||||
&base64::engine::general_purpose::STANDARD,
|
||||
&args.audio_data,
|
||||
) {
|
||||
Ok(data) => data,
|
||||
Err(e) => return ApiResponse {
|
||||
code: 400,
|
||||
message: format!("Invalid base64 data: {}", e),
|
||||
data: None,
|
||||
},
|
||||
};
|
||||
|
||||
match voice_storage::save_audio_file(
|
||||
&args.project_id,
|
||||
&args.audio_id,
|
||||
&audio_bytes,
|
||||
&args.name,
|
||||
&args.voice_id,
|
||||
args.duration,
|
||||
) {
|
||||
Ok(meta) => ApiResponse {
|
||||
code: 200,
|
||||
message: "Audio saved successfully".to_string(),
|
||||
data: Some(meta),
|
||||
},
|
||||
Err(e) => ApiResponse {
|
||||
code: 500,
|
||||
message: format!("Failed to save audio: {}", e),
|
||||
data: None,
|
||||
},
|
||||
}
|
||||
}
|
||||
|
||||
/// 列出项目所有音频文件
|
||||
#[tauri::command]
|
||||
pub async fn list_project_audios(
|
||||
project_id: String,
|
||||
) -> ApiResponse<Vec<voice_storage::AudioMeta>> {
|
||||
match voice_storage::list_project_audios(&project_id) {
|
||||
Ok(audios) => ApiResponse {
|
||||
code: 200,
|
||||
message: "Audio list retrieved".to_string(),
|
||||
data: Some(audios),
|
||||
},
|
||||
Err(e) => ApiResponse {
|
||||
code: 500,
|
||||
message: format!("Failed to list audios: {}", e),
|
||||
data: Some(vec![]),
|
||||
},
|
||||
}
|
||||
}
|
||||
|
||||
/// 删除音频文件
|
||||
#[tauri::command]
|
||||
pub async fn delete_audio(
|
||||
project_id: String,
|
||||
audio_id: String,
|
||||
) -> ApiResponse<bool> {
|
||||
match voice_storage::delete_audio_file(&project_id, &audio_id) {
|
||||
Ok(_) => ApiResponse {
|
||||
code: 200,
|
||||
message: "Audio deleted successfully".to_string(),
|
||||
data: Some(true),
|
||||
},
|
||||
Err(e) => ApiResponse {
|
||||
code: 500,
|
||||
message: format!("Failed to delete audio: {}", e),
|
||||
data: None,
|
||||
},
|
||||
}
|
||||
}
|
||||
|
||||
/// 获取项目音频目录（供前端 FFmpeg 调用）
|
||||
#[tauri::command]
|
||||
pub async fn get_project_audios_dir(
|
||||
project_id: String,
|
||||
) -> ApiResponse<String> {
|
||||
match voice_storage::get_project_audios_dir(&project_id) {
|
||||
Ok(path) => ApiResponse {
|
||||
code: 200,
|
||||
message: "Audio directory path".to_string(),
|
||||
data: Some(path.to_string_lossy().to_string()),
|
||||
},
|
||||
Err(e) => ApiResponse {
|
||||
code: 500,
|
||||
message: format!("Failed to get audio dir: {}", e),
|
||||
data: None,
|
||||
},
|
||||
}
|
||||
}
|
||||
@@ -2,6 +2,64 @@ use tauri_plugin_shell::ShellExt;
|
||||
use tauri_plugin_shell::process::CommandEvent;
|
||||
use tauri::{AppHandle, Emitter};
|
||||
|
||||
/// FFmpeg 路径转义（替换 `:` 为 `\:`，替换 `'` 为 `'\''`）
|
||||
fn escape_ffmpeg_path(path: &str) -> String {
|
||||
path.replace("'", "'\\''")
|
||||
}
|
||||
|
||||
/// 验证路径在允许的目录内，防止路径遍历攻击
|
||||
/// 允许的目录：用户文档目录下的 Meijiaka-zj 文件夹
|
||||
fn validate_safe_path(path: &str) -> Result<String, String> {
|
||||
let path = std::path::Path::new(path);
|
||||
|
||||
// 获取绝对路径
|
||||
let abs_path = if path.is_absolute() {
|
||||
path.to_path_buf()
|
||||
} else {
|
||||
std::env::current_dir()
|
||||
.map_err(|e| format!("无法获取当前目录: {}", e))?
|
||||
.join(path)
|
||||
};
|
||||
|
||||
// 检查是否在允许的目录内
|
||||
let docs_dir = dirs::document_dir()
|
||||
.ok_or("无法获取文档目录")?;
|
||||
let allowed_dir = docs_dir.join("Meijiaka-zj");
|
||||
|
||||
// 规范化路径
|
||||
let canonical = abs_path.canonicalize()
|
||||
.unwrap_or(abs_path.clone());
|
||||
|
||||
// 检查是否在允许目录下
|
||||
if !canonical.starts_with(&allowed_dir) {
|
||||
return Err(format!("路径不在允许目录内: {}", path.display()));
|
||||
}
|
||||
|
||||
Ok(canonical.to_string_lossy().to_string())
|
||||
}
|
||||
|
||||
/// 清理并验证输出路径
|
||||
fn sanitize_output_path(path: &str) -> Result<String, String> {
|
||||
let path = std::path::Path::new(path);
|
||||
|
||||
// 获取父目录并验证
|
||||
if let Some(parent) = path.parent() {
|
||||
validate_safe_path(&parent.to_string_lossy())?;
|
||||
}
|
||||
|
||||
// 确保是绝对路径
|
||||
if path.is_absolute() {
|
||||
Ok(path.to_string_lossy().to_string())
|
||||
} else {
|
||||
std::env::current_dir()
|
||||
.map_err(|e| format!("无法获取当前目录: {}", e))?
|
||||
.join(path)
|
||||
.to_str()
|
||||
.map(|s| s.to_string())
|
||||
.ok_or_else(|| "无效的路径".to_string())
|
||||
}
|
||||
}
|
||||
|
||||
/**
|
||||
* 封装 FFmpeg Sidecar 调用
|
||||
*/
|
||||
@@ -50,8 +108,12 @@ pub async fn run_ffmpeg(app: &AppHandle, args: Vec<String>) -> Result<String, St
|
||||
* 标准化单个视频片段 (调整为 1080:1920, 30fps, libx264, aac 44100Hz stereo)
|
||||
*/
|
||||
pub async fn standardize_video(app: &AppHandle, input_path: &str, output_path: &str) -> Result<(), String> {
|
||||
// 验证路径安全
|
||||
let safe_input = validate_safe_path(input_path)?;
|
||||
let safe_output = sanitize_output_path(output_path)?;
|
||||
|
||||
let args = vec![
|
||||
"-i".to_string(), input_path.to_string(),
|
||||
"-i".to_string(), safe_input,
|
||||
"-vf".to_string(), "fps=30,scale=1080:1920:force_original_aspect_ratio=decrease,pad=1080:1920:(ow-iw)/2:(oh-ih)/2,format=yuv420p".to_string(),
|
||||
"-c:v".to_string(), "libx264".to_string(),
|
||||
"-c:a".to_string(), "aac".to_string(),
|
||||
@@ -61,7 +123,7 @@ pub async fn standardize_video(app: &AppHandle, input_path: &str, output_path: &
|
||||
"-crf".to_string(), "23".to_string(),
|
||||
"-r".to_string(), "30".to_string(),
|
||||
"-y".to_string(),
|
||||
output_path.to_string()
|
||||
safe_output
|
||||
];
|
||||
run_ffmpeg(app, args).await.map(|_| ())
|
||||
}
|
||||
@@ -70,13 +132,17 @@ pub async fn standardize_video(app: &AppHandle, input_path: &str, output_path: &
|
||||
* 拼接视频 - 快速模式 (要求编码/分辨率一致)
|
||||
*/
|
||||
pub async fn concat_videos_copy(app: &AppHandle, list_path: &str, output_path: &str) -> Result<(), String> {
|
||||
// 验证路径安全
|
||||
let safe_list = validate_safe_path(list_path)?;
|
||||
let safe_output = sanitize_output_path(output_path)?;
|
||||
|
||||
let args = vec![
|
||||
"-f".to_string(), "concat".to_string(),
|
||||
"-safe".to_string(), "0".to_string(),
|
||||
"-i".to_string(), list_path.to_string(),
|
||||
"-i".to_string(), safe_list,
|
||||
"-c".to_string(), "copy".to_string(),
|
||||
"-y".to_string(),
|
||||
output_path.to_string()
|
||||
safe_output
|
||||
];
|
||||
run_ffmpeg(app, args).await.map(|_| ())
|
||||
}
|
||||
@@ -93,7 +159,7 @@ pub async fn concat_videos_robust(app: &AppHandle, video_paths: Vec<String>, out
|
||||
|
||||
let mut standardized_paths = Vec::new();
|
||||
|
||||
// 1. 标准化每个片段
|
||||
// 1. 标准化每个片段（内部已验证路径）
|
||||
for (i, path) in video_paths.iter().enumerate() {
|
||||
let std_path = temp_dir.join(format!("std_{}_{}.mp4", timestamp, i));
|
||||
standardize_video(app, path, std_path.to_str().unwrap()).await?;
|
||||
@@ -104,11 +170,12 @@ pub async fn concat_videos_robust(app: &AppHandle, video_paths: Vec<String>, out
|
||||
let list_path = temp_dir.join(format!("concat_list_{}.txt", timestamp));
|
||||
let mut list_content = String::new();
|
||||
for path in &standardized_paths {
|
||||
list_content.push_str(&format!("file '{}'\n", path.to_str().unwrap()));
|
||||
// FFmpeg concat 列表中的路径需要用单引号包裹
|
||||
list_content.push_str(&format!("file '{}'\n", escape_ffmpeg_path(&path.to_string_lossy())));
|
||||
}
|
||||
std::fs::write(&list_path, list_content).map_err(|e| e.to_string())?;
|
||||
|
||||
// 3. 执行快速拼接
|
||||
// 3. 执行快速拼接（输出路径会被验证）
|
||||
let concat_res = concat_videos_copy(app, list_path.to_str().unwrap(), output_path).await;
|
||||
|
||||
// 4. 清理临时文件
|
||||
@@ -124,9 +191,14 @@ pub async fn concat_videos_robust(app: &AppHandle, video_paths: Vec<String>, out
|
||||
* 音画合并 - 优化音质
|
||||
*/
|
||||
pub async fn add_audio_to_video(app: &AppHandle, video_path: &str, audio_path: &str, output_path: &str) -> Result<(), String> {
|
||||
// 验证路径安全
|
||||
let safe_video = validate_safe_path(video_path)?;
|
||||
let safe_audio = validate_safe_path(audio_path)?;
|
||||
let safe_output = sanitize_output_path(output_path)?;
|
||||
|
||||
let args = vec![
|
||||
"-i".to_string(), video_path.to_string(),
|
||||
"-i".to_string(), audio_path.to_string(),
|
||||
"-i".to_string(), safe_video,
|
||||
"-i".to_string(), safe_audio,
|
||||
"-c:v".to_string(), "copy".to_string(),
|
||||
"-c:a".to_string(), "aac".to_string(),
|
||||
"-b:a".to_string(), "192k".to_string(), // 提高码率
|
||||
@@ -135,7 +207,7 @@ pub async fn add_audio_to_video(app: &AppHandle, video_path: &str, audio_path: &
|
||||
"-map".to_string(), "1:a:0".to_string(),
|
||||
"-shortest".to_string(),
|
||||
"-y".to_string(),
|
||||
output_path.to_string()
|
||||
safe_output
|
||||
];
|
||||
run_ffmpeg(app, args).await.map(|_| ())
|
||||
}
|
||||
@@ -145,9 +217,13 @@ pub async fn add_audio_to_video(app: &AppHandle, video_path: &str, audio_path: &
|
||||
* 带静音音频轨道，避免 concat 时丢失后续片段音频
|
||||
*/
|
||||
pub async fn create_cover_video(app: &AppHandle, input_path: &str, output_path: &str, duration: &str) -> Result<(), String> {
|
||||
// 验证路径安全
|
||||
let safe_input = validate_safe_path(input_path)?;
|
||||
let safe_output = sanitize_output_path(output_path)?;
|
||||
|
||||
let args = vec![
|
||||
"-loop".to_string(), "1".to_string(),
|
||||
"-i".to_string(), input_path.to_string(),
|
||||
"-i".to_string(), safe_input,
|
||||
"-f".to_string(), "lavfi".to_string(),
|
||||
"-i".to_string(), "anullsrc=r=44100:cl=stereo".to_string(),
|
||||
"-c:v".to_string(), "libx264".to_string(),
|
||||
@@ -158,7 +234,7 @@ pub async fn create_cover_video(app: &AppHandle, input_path: &str, output_path:
|
||||
"-r".to_string(), "30".to_string(),
|
||||
"-shortest".to_string(),
|
||||
"-y".to_string(),
|
||||
output_path.to_string()
|
||||
safe_output
|
||||
];
|
||||
run_ffmpeg(app, args).await.map(|_| ())
|
||||
}
|
||||
@@ -193,7 +269,7 @@ fn get_fonts_dir(app: &AppHandle) -> Result<std::path::PathBuf, String> {
|
||||
return Ok(dev_fonts_path);
|
||||
}
|
||||
|
||||
// 如果还是找不到，尝试往上一级找
|
||||
// 如果还是找不到，尝试往上一级找（支持项目根目录结构）
|
||||
if let Some(parent) = cwd.parent() {
|
||||
let dev_fonts_path_alt = parent.join("src-tauri/fonts");
|
||||
println!("[get_fonts_dir] Checking dev parent path: {:?}", dev_fonts_path_alt);
|
||||
@@ -201,13 +277,16 @@ fn get_fonts_dir(app: &AppHandle) -> Result<std::path::PathBuf, String> {
|
||||
println!("[get_fonts_dir] Found fonts at: {:?}", dev_fonts_path_alt);
|
||||
return Ok(dev_fonts_path_alt);
|
||||
}
|
||||
}
|
||||
|
||||
// 尝试绝对路径
|
||||
let abs_path = std::path::Path::new("/Users/0fun/work/ai-meijiaka/tauri-app/src-tauri/fonts");
|
||||
if abs_path.exists() {
|
||||
println!("[get_fonts_dir] Found fonts at absolute path: {:?}", abs_path);
|
||||
return Ok(abs_path.to_path_buf());
|
||||
|
||||
// 再往上一级（如果 cwd 是 src-tauri/src）
|
||||
if let Some(grandparent) = parent.parent() {
|
||||
let dev_fonts_path_grand = grandparent.join("tauri-app/src-tauri/fonts");
|
||||
println!("[get_fonts_dir] Checking grandparent path: {:?}", dev_fonts_path_grand);
|
||||
if dev_fonts_path_grand.exists() {
|
||||
println!("[get_fonts_dir] Found fonts at: {:?}", dev_fonts_path_grand);
|
||||
return Ok(dev_fonts_path_grand);
|
||||
}
|
||||
}
|
||||
}
|
||||
|
||||
Err("Could not find fonts directory in any location".to_string())
|
||||
@@ -221,13 +300,17 @@ pub async fn extract_first_frame(
|
||||
video_path: &str,
|
||||
output_path: &str,
|
||||
) -> Result<(), String> {
|
||||
// 验证路径安全
|
||||
let safe_video = validate_safe_path(video_path)?;
|
||||
let safe_output = sanitize_output_path(output_path)?;
|
||||
|
||||
let args = vec![
|
||||
"-i".to_string(), video_path.to_string(),
|
||||
"-i".to_string(), safe_video,
|
||||
"-ss".to_string(), "00:00:00".to_string(),
|
||||
"-vframes".to_string(), "1".to_string(),
|
||||
"-q:v".to_string(), "2".to_string(),
|
||||
"-y".to_string(),
|
||||
output_path.to_string(),
|
||||
safe_output,
|
||||
];
|
||||
run_ffmpeg(app, args).await.map(|_| ())
|
||||
}
|
||||
@@ -243,14 +326,17 @@ pub async fn burn_ass_subtitle(
|
||||
ass_path: &str,
|
||||
output_path: &str,
|
||||
) -> Result<(), String> {
|
||||
// 验证路径安全
|
||||
let safe_video = validate_safe_path(video_path)?;
|
||||
let safe_ass = validate_safe_path(ass_path)?;
|
||||
let safe_output = sanitize_output_path(output_path)?;
|
||||
|
||||
let fonts_dir = get_fonts_dir(app)?;
|
||||
let fonts_dir_str = fonts_dir.to_str().ok_or("Invalid fonts dir path")?;
|
||||
|
||||
// 转义路径中的特殊字符（FFmpeg 滤镜语法要求）
|
||||
// 只需要转义冒号 : 因为它是滤镜参数分隔符
|
||||
// FFmpeg 中，整个引用路径不需要内部再转义单引号
|
||||
let ass_path_escaped = ass_path.replace(":", "\\:");
|
||||
let fonts_dir_escaped = fonts_dir_str.replace(":", "\\:");
|
||||
// 使用统一的路径转义函数处理 FFmpeg 滤镜中的路径
|
||||
let ass_path_escaped = escape_ffmpeg_path(&safe_ass);
|
||||
let fonts_dir_escaped = escape_ffmpeg_path(fonts_dir_str);
|
||||
|
||||
// 使用 ass 滤镜，指定字体目录
|
||||
// 语法: ass='path':fontsdir='path'
|
||||
@@ -259,14 +345,14 @@ pub async fn burn_ass_subtitle(
|
||||
println!("FFmpeg filter: {}", filter); // 调试日志
|
||||
|
||||
let args = vec![
|
||||
"-i".to_string(), video_path.to_string(),
|
||||
"-i".to_string(), safe_video,
|
||||
"-vf".to_string(), filter,
|
||||
"-c:v".to_string(), "libx264".to_string(),
|
||||
"-preset".to_string(), "medium".to_string(),
|
||||
"-crf".to_string(), "23".to_string(),
|
||||
"-c:a".to_string(), "copy".to_string(), // 音频直接复制
|
||||
"-y".to_string(),
|
||||
output_path.to_string()
|
||||
safe_output
|
||||
];
|
||||
|
||||
run_ffmpeg(app, args).await.map(|_| ())
|
||||
@@ -283,22 +369,140 @@ pub async fn burn_ass_subtitle_with_fonts(
|
||||
fonts_dir: &str,
|
||||
output_path: &str,
|
||||
) -> Result<(), String> {
|
||||
let filter = format!("ass='{}':fontsdir='{}'", ass_path, fonts_dir);
|
||||
// 验证路径安全
|
||||
let safe_video = validate_safe_path(video_path)?;
|
||||
let safe_ass = validate_safe_path(ass_path)?;
|
||||
let safe_output = sanitize_output_path(output_path)?;
|
||||
|
||||
let filter = format!(
|
||||
"ass='{}':fontsdir='{}'",
|
||||
escape_ffmpeg_path(&safe_ass),
|
||||
escape_ffmpeg_path(fonts_dir)
|
||||
);
|
||||
|
||||
let args = vec![
|
||||
"-i".to_string(), video_path.to_string(),
|
||||
"-i".to_string(), safe_video,
|
||||
"-vf".to_string(), filter,
|
||||
"-c:v".to_string(), "libx264".to_string(),
|
||||
"-preset".to_string(), "medium".to_string(),
|
||||
"-crf".to_string(), "23".to_string(),
|
||||
"-c:a".to_string(), "copy".to_string(),
|
||||
"-y".to_string(),
|
||||
output_path.to_string()
|
||||
safe_output
|
||||
];
|
||||
|
||||
run_ffmpeg(app, args).await.map(|_| ())
|
||||
}
|
||||
|
||||
/**
|
||||
* 音频替换 — 用配音音频替换视频中的原音
|
||||
*
|
||||
* 适用于配音替换场景：保留原视频画面，替换为 TTS/克隆音频。
|
||||
*/
|
||||
pub async fn replace_audio_track(
|
||||
app: &AppHandle,
|
||||
video_path: &str,
|
||||
audio_path: &str,
|
||||
output_path: &str,
|
||||
) -> Result<(), String> {
|
||||
// 验证路径安全
|
||||
let safe_video = validate_safe_path(video_path)?;
|
||||
let safe_audio = validate_safe_path(audio_path)?;
|
||||
let safe_output = sanitize_output_path(output_path)?;
|
||||
|
||||
let args = vec![
|
||||
"-i".to_string(), safe_video,
|
||||
"-i".to_string(), safe_audio,
|
||||
// 视频流：直接复制（不重新编码）
|
||||
"-c:v".to_string(), "copy".to_string(),
|
||||
// 音频流：转码为 AAC
|
||||
"-c:a".to_string(), "aac".to_string(),
|
||||
"-b:a".to_string(), "192k".to_string(),
|
||||
"-ar".to_string(), "44100".to_string(),
|
||||
"-ac".to_string(), "2".to_string(),
|
||||
// 只保留第一个视频流和第一个音频流
|
||||
"-map".to_string(), "0:v:0".to_string(),
|
||||
"-map".to_string(), "1:a:0".to_string(),
|
||||
"-shortest".to_string(),
|
||||
"-y".to_string(),
|
||||
safe_output,
|
||||
];
|
||||
run_ffmpeg(app, args).await.map(|_| ())
|
||||
}
|
||||
|
||||
/**
|
||||
* 混音合并 — 将配音音频与背景音乐混合
|
||||
*
|
||||
* 适用于配音+背景音乐叠加场景。
|
||||
* 配音为主（0dB），背景音乐降低 6dB。
|
||||
*/
|
||||
pub async fn mix_audio_tracks(
|
||||
app: &AppHandle,
|
||||
voice_path: &str,
|
||||
bgm_path: &str,
|
||||
output_path: &str,
|
||||
voice_volume: f64,
|
||||
bgm_volume: f64,
|
||||
) -> Result<(), String> {
|
||||
// 验证路径安全
|
||||
let safe_voice = validate_safe_path(voice_path)?;
|
||||
let safe_bgm = validate_safe_path(bgm_path)?;
|
||||
let safe_output = sanitize_output_path(output_path)?;
|
||||
|
||||
// 限制音量范围，防止命令注入
|
||||
// volume 参数只接受 0.0 - 10.0 范围的浮点数
|
||||
let _safe_voice_vol = voice_volume.clamp(0.0, 10.0); // voice_volume 当前未使用，保留供将来扩展
|
||||
let safe_bgm_vol = bgm_volume.clamp(0.0, 10.0);
|
||||
|
||||
// 格式化为安全的数值字符串（最多2位小数）
|
||||
let bgm_vol_str = format!("{:.2}", safe_bgm_vol);
|
||||
|
||||
// 构建安全的 filter 字符串
|
||||
// FFmpeg volume 滤镜接受数值参数，这里使用格式化后的数值
|
||||
let filter = format!(
|
||||
"[1:a]volume={}[bgm];[0:a][bgm]amix=inputs=2:duration=first:dropout_transition=2[out]",
|
||||
bgm_vol_str
|
||||
);
|
||||
|
||||
let args = vec![
|
||||
"-i".to_string(), safe_voice,
|
||||
"-i".to_string(), safe_bgm,
|
||||
"-filter_complex".to_string(), filter,
|
||||
"-map".to_string(), "0:v:0".to_string(),
|
||||
"-map".to_string(), "[out]".to_string(),
|
||||
"-c:v".to_string(), "copy".to_string(),
|
||||
"-c:a".to_string(), "aac".to_string(),
|
||||
"-ar".to_string(), "44100".to_string(),
|
||||
"-y".to_string(),
|
||||
safe_output,
|
||||
];
|
||||
run_ffmpeg(app, args).await.map(|_| ())
|
||||
}
|
||||
|
||||
/**
|
||||
* 转码音频为标准格式 (MP3 44.1kHz stereo 192kbps)
|
||||
*/
|
||||
pub async fn standardize_audio(
|
||||
app: &AppHandle,
|
||||
input_path: &str,
|
||||
output_path: &str,
|
||||
) -> Result<(), String> {
|
||||
// 验证路径安全
|
||||
let safe_input = validate_safe_path(input_path)?;
|
||||
let safe_output = sanitize_output_path(output_path)?;
|
||||
|
||||
let args = vec![
|
||||
"-i".to_string(), safe_input,
|
||||
"-c:a".to_string(), "libmp3lame".to_string(),
|
||||
"-b:a".to_string(), "192k".to_string(),
|
||||
"-ar".to_string(), "44100".to_string(),
|
||||
"-ac".to_string(), "2".to_string(),
|
||||
"-y".to_string(),
|
||||
safe_output,
|
||||
];
|
||||
run_ffmpeg(app, args).await.map(|_| ())
|
||||
}
|
||||
|
||||
/**
|
||||
* 压制 ASS 字幕到图片（使用嵌入字体）
|
||||
*
|
||||
@@ -310,22 +514,28 @@ pub async fn burn_ass_subtitle_to_image(
|
||||
ass_path: &str,
|
||||
output_path: &str,
|
||||
) -> Result<(), String> {
|
||||
// 验证路径安全
|
||||
let safe_image = validate_safe_path(image_path)?;
|
||||
let safe_ass = validate_safe_path(ass_path)?;
|
||||
let safe_output = sanitize_output_path(output_path)?;
|
||||
|
||||
let fonts_dir = get_fonts_dir(app)?;
|
||||
let fonts_dir_str = fonts_dir.to_str().ok_or("Invalid fonts dir path")?;
|
||||
|
||||
let ass_path_escaped = ass_path.replace("'", "'\\''").replace(":", "\\:");
|
||||
let fonts_dir_escaped = fonts_dir_str.replace("'", "'\\''");
|
||||
|
||||
let filter = format!("ass='{}':fontsdir='{}'", ass_path_escaped, fonts_dir_escaped);
|
||||
let filter = format!(
|
||||
"ass='{}':fontsdir='{}'",
|
||||
escape_ffmpeg_path(&safe_ass),
|
||||
escape_ffmpeg_path(fonts_dir_str)
|
||||
);
|
||||
|
||||
let args = vec![
|
||||
"-loop".to_string(), "1".to_string(),
|
||||
"-i".to_string(), image_path.to_string(),
|
||||
"-i".to_string(), safe_image,
|
||||
"-vf".to_string(), filter,
|
||||
"-t".to_string(), "1".to_string(),
|
||||
"-frames:v".to_string(), "1".to_string(),
|
||||
"-y".to_string(),
|
||||
output_path.to_string()
|
||||
safe_output
|
||||
];
|
||||
|
||||
run_ffmpeg(app, args).await.map(|_| ())
|
||||
|
||||
@@ -109,6 +109,15 @@ pub fn run() {
|
||||
commands::product::delete_local_product,
|
||||
// 重命名成品
|
||||
commands::product::rename_local_product,
|
||||
// 音频管理
|
||||
commands::voice::save_audio,
|
||||
commands::voice::list_project_audios,
|
||||
commands::voice::delete_audio,
|
||||
commands::voice::get_project_audios_dir,
|
||||
// 音频处理
|
||||
replace_audio_track,
|
||||
mix_audio_tracks,
|
||||
standardize_audio,
|
||||
])
|
||||
.run(tauri::generate_context!())
|
||||
.expect("error while running tauri application");
|
||||
@@ -216,6 +225,108 @@ async fn generate_cover_image(
|
||||
}
|
||||
}
|
||||
|
||||
// ============================================================
|
||||
// 音频处理命令
|
||||
// ============================================================
|
||||
|
||||
#[derive(Deserialize)]
|
||||
#[serde(rename_all = "snake_case")]
|
||||
pub struct ReplaceAudioRequest {
|
||||
video_path: String,
|
||||
audio_path: String,
|
||||
output_path: String,
|
||||
}
|
||||
|
||||
#[tauri::command]
|
||||
async fn replace_audio_track(
|
||||
app: tauri::AppHandle,
|
||||
request: ReplaceAudioRequest,
|
||||
) -> ApiResponse<String> {
|
||||
match ffmpeg_cmd::replace_audio_track(
|
||||
&app,
|
||||
&request.video_path,
|
||||
&request.audio_path,
|
||||
&request.output_path,
|
||||
).await {
|
||||
Ok(_) => ApiResponse {
|
||||
code: 200,
|
||||
message: "Audio replaced successfully".to_string(),
|
||||
data: Some(request.output_path),
|
||||
},
|
||||
Err(e) => ApiResponse {
|
||||
code: 500,
|
||||
message: format!("Failed to replace audio: {}", e),
|
||||
data: None,
|
||||
},
|
||||
}
|
||||
}
|
||||
|
||||
#[derive(Deserialize)]
|
||||
#[serde(rename_all = "snake_case")]
|
||||
pub struct MixAudioRequest {
|
||||
voice_path: String,
|
||||
bgm_path: String,
|
||||
output_path: String,
|
||||
voice_volume: f64,
|
||||
bgm_volume: f64,
|
||||
}
|
||||
|
||||
#[tauri::command]
|
||||
async fn mix_audio_tracks(
|
||||
app: tauri::AppHandle,
|
||||
request: MixAudioRequest,
|
||||
) -> ApiResponse<String> {
|
||||
match ffmpeg_cmd::mix_audio_tracks(
|
||||
&app,
|
||||
&request.voice_path,
|
||||
&request.bgm_path,
|
||||
&request.output_path,
|
||||
request.voice_volume,
|
||||
request.bgm_volume,
|
||||
).await {
|
||||
Ok(_) => ApiResponse {
|
||||
code: 200,
|
||||
message: "Audio mixed successfully".to_string(),
|
||||
data: Some(request.output_path),
|
||||
},
|
||||
Err(e) => ApiResponse {
|
||||
code: 500,
|
||||
message: format!("Failed to mix audio: {}", e),
|
||||
data: None,
|
||||
},
|
||||
}
|
||||
}
|
||||
|
||||
#[derive(Deserialize)]
|
||||
#[serde(rename_all = "snake_case")]
|
||||
pub struct StandardizeAudioRequest {
|
||||
input_path: String,
|
||||
output_path: String,
|
||||
}
|
||||
|
||||
#[tauri::command]
|
||||
async fn standardize_audio(
|
||||
app: tauri::AppHandle,
|
||||
request: StandardizeAudioRequest,
|
||||
) -> ApiResponse<String> {
|
||||
match ffmpeg_cmd::standardize_audio(
|
||||
&app,
|
||||
&request.input_path,
|
||||
&request.output_path,
|
||||
).await {
|
||||
Ok(_) => ApiResponse {
|
||||
code: 200,
|
||||
message: "Audio standardized successfully".to_string(),
|
||||
data: Some(request.output_path),
|
||||
},
|
||||
Err(e) => ApiResponse {
|
||||
code: 500,
|
||||
message: format!("Failed to standardize audio: {}", e),
|
||||
data: None,
|
||||
},
|
||||
}
|
||||
}
|
||||
|
||||
// ============================================================
|
||||
// 视频合成命令
|
||||
// ============================================================
|
||||
|
||||
@@ -10,6 +10,7 @@ pub mod project;
|
||||
pub mod avatar;
|
||||
pub mod auth;
|
||||
pub mod cache;
|
||||
pub mod voice;
|
||||
|
||||
pub use engine::{
|
||||
atomic_write_json, atomic_write_bytes,
|
||||
|
||||
@@ -8,18 +8,18 @@ use tauri::{AppHandle, Manager};
|
||||
use crate::storage::engine::{sanitize_id, StorageError};
|
||||
|
||||
/// 获取美家卡用户文档目录
|
||||
/// ~/Documents/Meijiaka/
|
||||
/// ~/Documents/Meijiaka-zj/
|
||||
pub fn get_meijiaka_dir() -> Result<PathBuf, StorageError> {
|
||||
let base = dirs::document_dir()
|
||||
.ok_or_else(|| StorageError::Io(std::io::Error::new(
|
||||
std::io::ErrorKind::NotFound,
|
||||
"无法获取文档目录",
|
||||
)))?;
|
||||
Ok(base.join("Meijiaka"))
|
||||
Ok(base.join("Meijiaka-zj"))
|
||||
}
|
||||
|
||||
/// 获取项目目录路径（不自动创建）
|
||||
/// ~/Documents/Meijiaka/projects/{project_id}/
|
||||
/// ~/Documents/Meijiaka-zj/projects/{project_id}/
|
||||
pub fn get_project_dir_path(project_id: &str) -> Result<PathBuf, StorageError> {
|
||||
let safe_id = sanitize_id(project_id)?;
|
||||
let base = get_meijiaka_dir()?;
|
||||
@@ -27,7 +27,7 @@ pub fn get_project_dir_path(project_id: &str) -> Result<PathBuf, StorageError> {
|
||||
}
|
||||
|
||||
/// 获取项目目录（自动创建）
|
||||
/// ~/Documents/Meijiaka/projects/{project_id}/
|
||||
/// ~/Documents/Meijiaka-zj/projects/{project_id}/
|
||||
pub fn get_project_dir(project_id: &str) -> Result<PathBuf, StorageError> {
|
||||
let path = get_project_dir_path(project_id)?;
|
||||
crate::storage::engine::ensure_dir(&path)?;
|
||||
@@ -44,7 +44,7 @@ pub fn get_project_assets_dir(project_id: &str) -> Result<PathBuf, StorageError>
|
||||
}
|
||||
|
||||
/// 获取项目内的 videos 目录
|
||||
/// ~/Documents/Meijiaka/projects/{project_id}/videos/
|
||||
/// ~/Documents/Meijiaka-zj/projects/{project_id}/videos/
|
||||
pub fn get_project_videos_dir(project_id: &str) -> Result<PathBuf, StorageError> {
|
||||
let path = get_project_dir(project_id)?;
|
||||
let videos = path.join("videos");
|
||||
@@ -53,7 +53,7 @@ pub fn get_project_videos_dir(project_id: &str) -> Result<PathBuf, StorageError>
|
||||
}
|
||||
|
||||
/// 获取项目内的 images 目录
|
||||
/// ~/Documents/Meijiaka/projects/{project_id}/images/
|
||||
/// ~/Documents/Meijiaka-zj/projects/{project_id}/images/
|
||||
pub fn get_project_images_dir(project_id: &str) -> Result<PathBuf, StorageError> {
|
||||
let path = get_project_dir(project_id)?;
|
||||
let images = path.join("images");
|
||||
@@ -62,7 +62,7 @@ pub fn get_project_images_dir(project_id: &str) -> Result<PathBuf, StorageError>
|
||||
}
|
||||
|
||||
/// 获取 products 目录
|
||||
/// ~/Documents/Meijiaka/products/
|
||||
/// ~/Documents/Meijiaka-zj/projects/{project_id}/assets/
|
||||
pub fn get_products_dir() -> Result<PathBuf, StorageError> {
|
||||
let base = get_meijiaka_dir()?;
|
||||
let path = base.join("products");
|
||||
@@ -71,7 +71,7 @@ pub fn get_products_dir() -> Result<PathBuf, StorageError> {
|
||||
}
|
||||
|
||||
/// 获取头像列表 JSON 路径
|
||||
/// ~/Documents/Meijiaka/avatars.json
|
||||
/// ~/Documents/Meijiaka-zj/avatars.json
|
||||
pub fn get_avatars_json_path() -> Result<PathBuf, StorageError> {
|
||||
let base = get_meijiaka_dir()?;
|
||||
Ok(base.join("avatars.json"))
|
||||
|
||||
@@ -0,0 +1,164 @@
|
||||
//! Voice 音频文件存储模块
|
||||
//!
|
||||
//! 管理 TTS 合成音频和配音文件的本地存储。
|
||||
|
||||
use serde::{Deserialize, Serialize};
|
||||
use std::path::{Path, PathBuf};
|
||||
|
||||
use crate::storage::engine::{atomic_write_bytes, atomic_write_json, read_json, ensure_dir, StorageError};
|
||||
use crate::storage::paths::get_project_dir;
|
||||
|
||||
/// 音频文件元数据
|
||||
#[derive(Debug, Clone, Serialize, Deserialize)]
|
||||
#[serde(rename_all = "camelCase")]
|
||||
pub struct AudioMeta {
|
||||
pub id: String,
|
||||
pub name: String,
|
||||
pub file_path: String,
|
||||
pub duration: f64,
|
||||
pub file_size: u64,
|
||||
pub voice_id: String,
|
||||
pub created_at: String,
|
||||
pub segment_id: Option<String>,
|
||||
}
|
||||
|
||||
/// 音频列表元数据文件
|
||||
#[derive(Debug, Clone, Serialize, Deserialize, Default)]
|
||||
#[serde(rename_all = "camelCase")]
|
||||
pub struct AudioListMeta {
|
||||
pub project_id: String,
|
||||
pub audios: Vec<AudioMeta>,
|
||||
pub updated_at: String,
|
||||
}
|
||||
|
||||
/// 保存音频文件到项目目录
|
||||
///
|
||||
/// 将 base64 编码的音频数据写入 `audios/` 子目录。
|
||||
pub fn save_audio_file(
|
||||
project_id: &str,
|
||||
audio_id: &str,
|
||||
data: &[u8],
|
||||
name: &str,
|
||||
voice_id: &str,
|
||||
duration: f64,
|
||||
) -> Result<AudioMeta, StorageError> {
|
||||
let project_dir = get_project_dir(project_id)?;
|
||||
let audios_dir = project_dir.join("audios");
|
||||
ensure_dir(&audios_dir)?;
|
||||
|
||||
let extension = "mp3";
|
||||
let file_name = format!("{}.{}", audio_id, extension);
|
||||
let file_path = audios_dir.join(&file_name);
|
||||
|
||||
atomic_write_bytes(&file_path, data)?;
|
||||
|
||||
let meta = AudioMeta {
|
||||
id: audio_id.to_string(),
|
||||
name: name.to_string(),
|
||||
file_path: file_path.to_str().unwrap_or_default().to_string(),
|
||||
duration,
|
||||
file_size: data.len() as u64,
|
||||
voice_id: voice_id.to_string(),
|
||||
created_at: chrono_lite_now(),
|
||||
segment_id: None,
|
||||
};
|
||||
|
||||
// 更新音频列表元数据
|
||||
let list = load_audio_list_meta(project_id)?;
|
||||
let mut list = list.unwrap_or(AudioListMeta {
|
||||
project_id: project_id.to_string(),
|
||||
audios: vec![],
|
||||
updated_at: String::new(),
|
||||
});
|
||||
|
||||
// 替换已存在的同名音频
|
||||
if let Some(pos) = list.audios.iter().position(|a| a.id == audio_id) {
|
||||
list.audios[pos] = meta.clone();
|
||||
} else {
|
||||
list.audios.push(meta.clone());
|
||||
}
|
||||
list.updated_at = chrono_lite_now();
|
||||
|
||||
save_audio_list_meta(project_id, &list)?;
|
||||
|
||||
Ok(meta)
|
||||
}
|
||||
|
||||
/// 从项目加载音频文件（返回二进制数据）
|
||||
pub fn load_audio_file(project_id: &str, audio_id: &str) -> Result<Vec<u8>, StorageError> {
|
||||
let list = load_audio_list_meta(project_id)?
|
||||
.ok_or_else(|| StorageError::Io(std::io::Error::new(
|
||||
std::io::ErrorKind::NotFound,
|
||||
"音频列表不存在",
|
||||
)))?;
|
||||
|
||||
let audio_meta = list.audios.iter()
|
||||
.find(|a| a.id == audio_id)
|
||||
.ok_or_else(|| StorageError::Io(std::io::Error::new(
|
||||
std::io::ErrorKind::NotFound,
|
||||
format!("音频 {} 不存在", audio_id),
|
||||
)))?;
|
||||
|
||||
let path = Path::new(&audio_meta.file_path);
|
||||
std::fs::read(path).map_err(Into::into)
|
||||
}
|
||||
|
||||
/// 列出项目所有音频文件
|
||||
pub fn list_project_audios(project_id: &str) -> Result<Vec<AudioMeta>, StorageError> {
|
||||
let list = load_audio_list_meta(project_id)?;
|
||||
Ok(list.map(|l| l.audios).unwrap_or_default())
|
||||
}
|
||||
|
||||
/// 删除项目中的音频文件
|
||||
pub fn delete_audio_file(project_id: &str, audio_id: &str) -> Result<(), StorageError> {
|
||||
let list = load_audio_list_meta(project_id)?;
|
||||
let mut list = list.ok_or_else(|| StorageError::Io(std::io::Error::new(
|
||||
std::io::ErrorKind::NotFound,
|
||||
"音频列表不存在",
|
||||
)))?;
|
||||
|
||||
let pos = list.audios.iter().position(|a| a.id == audio_id)
|
||||
.ok_or_else(|| StorageError::Io(std::io::Error::new(
|
||||
std::io::ErrorKind::NotFound,
|
||||
format!("音频 {} 不存在", audio_id),
|
||||
)))?;
|
||||
|
||||
let meta = list.audios.remove(pos);
|
||||
let path = Path::new(&meta.file_path);
|
||||
if path.exists() {
|
||||
std::fs::remove_file(path)?;
|
||||
}
|
||||
|
||||
list.updated_at = chrono_lite_now();
|
||||
save_audio_list_meta(project_id, &list)
|
||||
}
|
||||
|
||||
/// 获取音频列表元数据文件路径
|
||||
fn get_audio_list_path(project_id: &str) -> Result<PathBuf, StorageError> {
|
||||
let project_dir = get_project_dir(project_id)?;
|
||||
Ok(project_dir.join("audios.json"))
|
||||
}
|
||||
|
||||
fn load_audio_list_meta(project_id: &str) -> Result<Option<AudioListMeta>, StorageError> {
|
||||
let path = get_audio_list_path(project_id)?;
|
||||
read_json(&path)
|
||||
}
|
||||
|
||||
fn save_audio_list_meta(project_id: &str, list: &AudioListMeta) -> Result<(), StorageError> {
|
||||
let path = get_audio_list_path(project_id)?;
|
||||
atomic_write_json(&path, list)
|
||||
}
|
||||
|
||||
/// 获取项目音频目录路径（供 FFmpeg 使用）
|
||||
pub fn get_project_audios_dir(project_id: &str) -> Result<PathBuf, StorageError> {
|
||||
let project_dir = get_project_dir(project_id)?;
|
||||
let audios_dir = project_dir.join("audios");
|
||||
ensure_dir(&audios_dir)?;
|
||||
Ok(audios_dir)
|
||||
}
|
||||
|
||||
// ====================== 工具函数 ======================
|
||||
|
||||
fn chrono_lite_now() -> String {
|
||||
chrono::Utc::now().to_rfc3339()
|
||||
}
|
||||
@@ -12,7 +12,7 @@ pub fn uuid_v4() -> String {
|
||||
/// 如果目录不存在则创建
|
||||
pub fn get_project_dir(app: &tauri::AppHandle) -> std::path::PathBuf {
|
||||
let mut path = app.path().document_dir().unwrap_or_else(|_| std::env::temp_dir());
|
||||
path.push("Meijiaka");
|
||||
path.push("Meijiaka-zj");
|
||||
path.push("Projects");
|
||||
if !path.exists() {
|
||||
let _ = std::fs::create_dir_all(&path);
|
||||
|
||||
@@ -1,8 +1,8 @@
|
||||
{
|
||||
"$schema": "https://schema.tauri.app/config/2",
|
||||
"productName": "美家卡智影",
|
||||
"productName": "美家卡智剪",
|
||||
"version": "0.1.0",
|
||||
"identifier": "cn.meijiaka.ai-video",
|
||||
"identifier": "cn.meijiaka.ai-jian",
|
||||
"build": {
|
||||
"beforeDevCommand": "npm run dev",
|
||||
"devUrl": "http://localhost:1420",
|
||||
@@ -13,7 +13,7 @@
|
||||
"windows": [
|
||||
{
|
||||
"label": "main",
|
||||
"title": "美家卡 智影",
|
||||
"title": "美家卡 智剪",
|
||||
"width": 1440,
|
||||
"height": 960,
|
||||
"minWidth": 960,
|
||||
|
||||
@@ -0,0 +1,238 @@
|
||||
/**
|
||||
* Voice 语音 API 模块
|
||||
* ===================
|
||||
*
|
||||
* TTS 合成、批量合成、声音克隆接口封装。
|
||||
*/
|
||||
|
||||
import { invoke } from '@tauri-apps/api/core';
|
||||
import { client } from '../client';
|
||||
|
||||
// ====================== TTS 类型 ======================
|
||||
|
||||
export interface VoiceInfo {
|
||||
voiceId: string;
|
||||
name: string;
|
||||
description: string;
|
||||
recommended: boolean;
|
||||
}
|
||||
|
||||
export interface TTSSynthesizeRequest {
|
||||
text: string;
|
||||
voiceId?: string;
|
||||
speed?: number;
|
||||
volume?: number;
|
||||
pitch?: number;
|
||||
}
|
||||
|
||||
export interface TTSBatchSegment {
|
||||
text: string;
|
||||
index: number;
|
||||
filename?: string;
|
||||
}
|
||||
|
||||
export interface TTSBatchRequest {
|
||||
segments: TTSBatchSegment[];
|
||||
voiceId?: string;
|
||||
speed?: number;
|
||||
}
|
||||
|
||||
export interface TTSResult {
|
||||
audioBase64: string;
|
||||
format: string;
|
||||
text: string;
|
||||
voiceId: string;
|
||||
}
|
||||
|
||||
export interface TTSBatchResult {
|
||||
total: number;
|
||||
successCount: number;
|
||||
failedCount: number;
|
||||
results: Array<{
|
||||
index: number;
|
||||
text: string;
|
||||
outputPath?: string;
|
||||
success: boolean;
|
||||
error?: string;
|
||||
}>;
|
||||
}
|
||||
|
||||
// ====================== 声音克隆类型 ======================
|
||||
|
||||
export interface VoiceCloneSubmitRequest {
|
||||
sourceAudioUrl: string;
|
||||
sourceText?: string;
|
||||
voiceName?: string;
|
||||
}
|
||||
|
||||
export interface VoiceCloneTaskResponse {
|
||||
taskId: string;
|
||||
status: string;
|
||||
voiceId?: string;
|
||||
trialUrl?: string;
|
||||
errorMessage?: string;
|
||||
}
|
||||
|
||||
// ====================== 音频文件管理类型 ======================
|
||||
|
||||
export interface AudioMeta {
|
||||
id: string;
|
||||
name: string;
|
||||
filePath: string;
|
||||
duration: number;
|
||||
fileSize: number;
|
||||
voiceId: string;
|
||||
createdAt: string;
|
||||
segmentId?: string;
|
||||
}
|
||||
|
||||
// ====================== TTS API ======================
|
||||
|
||||
/** 获取预设音色列表 */
|
||||
export async function getVoiceList(): Promise<VoiceInfo[]> {
|
||||
const data = await client.get<{ voiceId: string; name: string; description: string; recommended: boolean }[]>('/voice/voices');
|
||||
return data.map(v => ({
|
||||
voiceId: v.voiceId,
|
||||
name: v.name,
|
||||
description: v.description,
|
||||
recommended: v.recommended,
|
||||
}));
|
||||
}
|
||||
|
||||
/** 同步 TTS 合成（返回 base64） */
|
||||
export async function synthesizeTTS(request: TTSSynthesizeRequest): Promise<TTSResult> {
|
||||
return client.post<TTSResult>('/voice/synthesize', request);
|
||||
}
|
||||
|
||||
/** 批量 TTS 合成 */
|
||||
export async function synthesizeBatchTTS(request: TTSBatchRequest): Promise<TTSBatchResult> {
|
||||
return client.post<TTSBatchResult>('/voice/synthesize-batch', request);
|
||||
}
|
||||
|
||||
// ====================== 声音克隆 API ======================
|
||||
|
||||
/** 提交声音克隆任务 */
|
||||
export async function submitCloneTask(request: VoiceCloneSubmitRequest): Promise<VoiceCloneTaskResponse> {
|
||||
return client.post<VoiceCloneTaskResponse>('/voice/clone/submit', request);
|
||||
}
|
||||
|
||||
/** 查询克隆任务状态 */
|
||||
export async function queryCloneTask(taskId: string, blocking = false): Promise<VoiceCloneTaskResponse> {
|
||||
return client.get<VoiceCloneTaskResponse>(`/voice/clone/query/${taskId}?blocking=${blocking}`);
|
||||
}
|
||||
|
||||
/** 一站式克隆 */
|
||||
export async function cloneAndWait(request: VoiceCloneSubmitRequest, pollInterval = 5.0): Promise<VoiceCloneTaskResponse> {
|
||||
return client.post<VoiceCloneTaskResponse>('/voice/clone/clone-and-wait', { ...request, pollInterval });
|
||||
}
|
||||
|
||||
// ====================== 本地音频文件管理（Tauri IPC） ======================
|
||||
|
||||
/** 保存音频文件到本地 */
|
||||
export async function saveAudio(args: {
|
||||
projectId: string;
|
||||
audioId: string;
|
||||
audioData: string; // base64 编码
|
||||
name: string;
|
||||
voiceId: string;
|
||||
duration: number;
|
||||
}): Promise<AudioMeta> {
|
||||
const result = await invoke<{ code: number; data?: AudioMeta; message: string }>('save_audio', { args });
|
||||
if (result.code !== 200 || !result.data) {
|
||||
throw new Error(result.message || '保存音频失败');
|
||||
}
|
||||
return result.data;
|
||||
}
|
||||
|
||||
/** 列出项目音频文件 */
|
||||
export async function listProjectAudios(projectId: string): Promise<AudioMeta[]> {
|
||||
const result = await invoke<{ code: number; data?: AudioMeta[]; message: string }>('list_project_audios', { projectId });
|
||||
if (result.code !== 200) {
|
||||
throw new Error(result.message || '获取音频列表失败');
|
||||
}
|
||||
return result.data || [];
|
||||
}
|
||||
|
||||
/** 删除音频文件 */
|
||||
export async function deleteAudio(projectId: string, audioId: string): Promise<void> {
|
||||
const result = await invoke<{ code: number; message: string }>('delete_audio', { projectId, audioId });
|
||||
if (result.code !== 200) {
|
||||
throw new Error(result.message || '删除音频失败');
|
||||
}
|
||||
}
|
||||
|
||||
/** 获取项目音频目录 */
|
||||
export async function getProjectAudiosDir(projectId: string): Promise<string> {
|
||||
const result = await invoke<{ code: number; data?: string; message: string }>('get_project_audios_dir', { projectId });
|
||||
if (result.code !== 200 || !result.data) {
|
||||
throw new Error(result.message || '获取音频目录失败');
|
||||
}
|
||||
return result.data;
|
||||
}
|
||||
|
||||
// ====================== 音频处理命令（Tauri IPC） ======================
|
||||
|
||||
export interface ReplaceAudioRequest {
|
||||
videoPath: string;
|
||||
audioPath: string;
|
||||
outputPath: string;
|
||||
}
|
||||
|
||||
export interface MixAudioRequest {
|
||||
voicePath: string;
|
||||
bgmPath: string;
|
||||
outputPath: string;
|
||||
voiceVolume: number;
|
||||
bgmVolume: number;
|
||||
}
|
||||
|
||||
export interface StandardizeAudioRequest {
|
||||
inputPath: string;
|
||||
outputPath: string;
|
||||
}
|
||||
|
||||
/** 音频替换：用配音替换视频原音 */
|
||||
export async function replaceAudioTrack(args: ReplaceAudioRequest): Promise<string> {
|
||||
const result = await invoke<{ code: number; data?: string; message: string }>('replace_audio_track', {
|
||||
request: {
|
||||
video_path: args.videoPath,
|
||||
audio_path: args.audioPath,
|
||||
output_path: args.outputPath,
|
||||
},
|
||||
});
|
||||
if (result.code !== 200 || !result.data) {
|
||||
throw new Error(result.message || '音频替换失败');
|
||||
}
|
||||
return result.data;
|
||||
}
|
||||
|
||||
/** 混音：配音 + 背景音乐混合 */
|
||||
export async function mixAudioTracks(args: MixAudioRequest): Promise<string> {
|
||||
const result = await invoke<{ code: number; data?: string; message: string }>('mix_audio_tracks', {
|
||||
request: {
|
||||
voice_path: args.voicePath,
|
||||
bgm_path: args.bgmPath,
|
||||
output_path: args.outputPath,
|
||||
voice_volume: args.voiceVolume,
|
||||
bgm_volume: args.bgmVolume,
|
||||
},
|
||||
});
|
||||
if (result.code !== 200 || !result.data) {
|
||||
throw new Error(result.message || '混音失败');
|
||||
}
|
||||
return result.data;
|
||||
}
|
||||
|
||||
/** 音频标准化：转码为 MP3 44.1kHz 192kbps */
|
||||
export async function standardizeAudio(args: StandardizeAudioRequest): Promise<string> {
|
||||
const result = await invoke<{ code: number; data?: string; message: string }>('standardize_audio', {
|
||||
request: {
|
||||
input_path: args.inputPath,
|
||||
output_path: args.outputPath,
|
||||
},
|
||||
});
|
||||
if (result.code !== 200 || !result.data) {
|
||||
throw new Error(result.message || '音频标准化失败');
|
||||
}
|
||||
return result.data;
|
||||
}
|
||||
@@ -52,6 +52,8 @@ export interface ScriptShot {
|
||||
videoUrl?: string; // 七牛云视频 URL（用于字幕生成等后续处理）
|
||||
elementId?: number; // 分镜使用的形象ID（Kling element_id）
|
||||
voiceId?: string; // 空镜使用的音色ID
|
||||
voiceVolume?: number; // 人声音量 (0-2, 默认1)
|
||||
bgmVolume?: number; // 背景音乐音量 (0-2, 默认0.3)
|
||||
alignmentResult?: AlignmentResult; // 字幕打轴结果
|
||||
burnedVideoPath?: string; // 压制字幕后的视频路径
|
||||
burnedAt?: number; // 压制字幕的时间戳
|
||||
|
||||
@@ -27,12 +27,10 @@
|
||||
}
|
||||
|
||||
.sidebar-title {
|
||||
font-size: var(--font-lg);
|
||||
font-size: 16px;
|
||||
font-weight: 700;
|
||||
background: var(--primary-gradient);
|
||||
background-clip: text;
|
||||
-webkit-text-fill-color: transparent;
|
||||
background-clip: text;
|
||||
color: var(--text-primary);
|
||||
white-space: nowrap;
|
||||
}
|
||||
|
||||
.sidebar-nav {
|
||||
|
||||
@@ -83,11 +83,11 @@ export default function Sidebar({ currentPath, onNavigate }: SidebarProps) {
|
||||
<div className="sidebar-header">
|
||||
<div className="sidebar-logo">
|
||||
<img
|
||||
src="./assets/logo.png"
|
||||
alt="美家卡 智影"
|
||||
src="/assets/logo.png"
|
||||
alt="美家卡 智剪"
|
||||
style={{ width: 28, height: 24, objectFit: 'contain' }}
|
||||
/>
|
||||
<span className="sidebar-title">美家卡 智影</span>
|
||||
<span className="sidebar-title">美家卡 智剪</span>
|
||||
</div>
|
||||
</div>
|
||||
|
||||
|
||||
@@ -80,11 +80,11 @@ export default function Login() {
|
||||
<div className="login-brand">
|
||||
<div className="login-logo">
|
||||
<img
|
||||
src="./assets/logo.png"
|
||||
alt="美家卡 智影"
|
||||
src="/assets/logo.png"
|
||||
alt="美家卡 智剪"
|
||||
style={{ width: 42, height: 36, objectFit: 'contain' }}
|
||||
/>
|
||||
<span className="login-logo-text">美家卡 智影</span>
|
||||
<span className="login-logo-text">美家卡 智剪</span>
|
||||
</div>
|
||||
<p className="login-subtitle">AI 驱动的智能视频创作平台</p>
|
||||
</div>
|
||||
|
||||
@@ -8,7 +8,7 @@ export default function AboutUs() {
|
||||
<div className="card" style={{ padding: 0, overflow: 'hidden' }}>
|
||||
<div className="settings-row">
|
||||
<span className="settings-row-label">应用名称</span>
|
||||
<span className="settings-row-value">美家卡 智影</span>
|
||||
<span className="settings-row-value">美家卡 智剪</span>
|
||||
</div>
|
||||
<div className="settings-row" style={{ borderTop: '1px solid var(--border-light)' }}>
|
||||
<span className="settings-row-label">版本号</span>
|
||||
|
||||
@@ -0,0 +1,173 @@
|
||||
/**
|
||||
* 音频合成页面样式
|
||||
*/
|
||||
|
||||
.audio-mixing {
|
||||
height: 100%;
|
||||
display: flex;
|
||||
flex-direction: column;
|
||||
gap: var(--spacing-lg);
|
||||
}
|
||||
|
||||
.audio-mixing-header {
|
||||
padding: var(--spacing-lg);
|
||||
border-bottom: 1px solid var(--border-light);
|
||||
}
|
||||
|
||||
.audio-mixing-header h2 {
|
||||
margin: 0 0 var(--spacing-xs);
|
||||
font-size: var(--font-xl);
|
||||
font-weight: 600;
|
||||
}
|
||||
|
||||
.audio-mixing-desc {
|
||||
margin: 0;
|
||||
color: var(--text-secondary);
|
||||
font-size: var(--font-sm);
|
||||
}
|
||||
|
||||
.audio-mixing-content {
|
||||
flex: 1;
|
||||
display: flex;
|
||||
gap: var(--spacing-lg);
|
||||
padding: 0 var(--spacing-lg);
|
||||
overflow: hidden;
|
||||
}
|
||||
|
||||
/* 左侧分镜列表 */
|
||||
.audio-mixing-sidebar {
|
||||
width: 240px;
|
||||
flex-shrink: 0;
|
||||
display: flex;
|
||||
flex-direction: column;
|
||||
gap: var(--spacing-md);
|
||||
}
|
||||
|
||||
.audio-mixing-sidebar h3 {
|
||||
font-size: var(--font-base);
|
||||
font-weight: 600;
|
||||
margin: 0;
|
||||
}
|
||||
|
||||
.segment-list {
|
||||
flex: 1;
|
||||
overflow-y: auto;
|
||||
display: flex;
|
||||
flex-direction: column;
|
||||
gap: var(--spacing-xs);
|
||||
}
|
||||
|
||||
.segment-item {
|
||||
display: flex;
|
||||
justify-content: space-between;
|
||||
align-items: center;
|
||||
padding: var(--spacing-md);
|
||||
border-radius: var(--radius-md);
|
||||
background: var(--bg-secondary);
|
||||
cursor: pointer;
|
||||
transition: all var(--transition-fast);
|
||||
}
|
||||
|
||||
.segment-item:hover {
|
||||
background: var(--bg-hover);
|
||||
}
|
||||
|
||||
.segment-item.active {
|
||||
background: var(--primary-light);
|
||||
color: var(--primary);
|
||||
}
|
||||
|
||||
.segment-index {
|
||||
font-weight: 500;
|
||||
}
|
||||
|
||||
.segment-duration {
|
||||
font-size: var(--font-sm);
|
||||
color: var(--text-tertiary);
|
||||
}
|
||||
|
||||
/* 右侧音频设置面板 */
|
||||
.audio-mixing-panel {
|
||||
flex: 1;
|
||||
background: var(--card-bg);
|
||||
border-radius: var(--radius-lg);
|
||||
padding: var(--spacing-xl);
|
||||
overflow-y: auto;
|
||||
}
|
||||
|
||||
.audio-mixing-panel h3 {
|
||||
margin: 0 0 var(--spacing-xl);
|
||||
font-size: var(--font-lg);
|
||||
font-weight: 600;
|
||||
}
|
||||
|
||||
.audio-control {
|
||||
margin-bottom: var(--spacing-xl);
|
||||
}
|
||||
|
||||
.audio-control label {
|
||||
display: block;
|
||||
font-size: var(--font-sm);
|
||||
font-weight: 500;
|
||||
color: var(--text-secondary);
|
||||
margin-bottom: var(--spacing-sm);
|
||||
}
|
||||
|
||||
.slider-container {
|
||||
display: flex;
|
||||
align-items: center;
|
||||
gap: var(--spacing-md);
|
||||
}
|
||||
|
||||
.slider-container input[type="range"] {
|
||||
flex: 1;
|
||||
height: 6px;
|
||||
border-radius: 3px;
|
||||
background: var(--border-light);
|
||||
-webkit-appearance: none;
|
||||
appearance: none;
|
||||
}
|
||||
|
||||
.slider-container input[type="range"]::-webkit-slider-thumb {
|
||||
-webkit-appearance: none;
|
||||
appearance: none;
|
||||
width: 18px;
|
||||
height: 18px;
|
||||
border-radius: 50%;
|
||||
background: var(--primary);
|
||||
cursor: pointer;
|
||||
}
|
||||
|
||||
.slider-value {
|
||||
min-width: 50px;
|
||||
text-align: right;
|
||||
font-size: var(--font-sm);
|
||||
font-weight: 500;
|
||||
color: var(--text-primary);
|
||||
}
|
||||
|
||||
.audio-info {
|
||||
margin-top: var(--spacing-xl);
|
||||
padding: var(--spacing-md);
|
||||
background: var(--bg-secondary);
|
||||
border-radius: var(--radius-md);
|
||||
}
|
||||
|
||||
.audio-info p {
|
||||
margin: 0 0 var(--spacing-xs);
|
||||
font-size: var(--font-sm);
|
||||
color: var(--text-tertiary);
|
||||
}
|
||||
|
||||
.audio-info p:last-child {
|
||||
margin-bottom: 0;
|
||||
}
|
||||
|
||||
/* 空状态 */
|
||||
.empty-state {
|
||||
display: flex;
|
||||
align-items: center;
|
||||
justify-content: center;
|
||||
height: 200px;
|
||||
color: var(--text-tertiary);
|
||||
}
|
||||
@@ -0,0 +1,115 @@
|
||||
/**
|
||||
* 音频合成页面 (Step 2)
|
||||
* ====================
|
||||
*
|
||||
* 为每个分镜配置背景音乐和音量
|
||||
*/
|
||||
|
||||
import { useState } from 'react';
|
||||
import { useProjectStore } from '../../store';
|
||||
import { toast } from '../../store/uiStore';
|
||||
import './AudioMixing.css';
|
||||
|
||||
export default function AudioMixing() {
|
||||
const segments = useProjectStore(state => state.segments);
|
||||
const updateSegment = useProjectStore(state => state.updateSegment);
|
||||
const [selectedSegmentId, setSelectedSegmentId] = useState<string | null>(
|
||||
segments.length > 0 ? segments[0].id : null
|
||||
);
|
||||
|
||||
const selectedSegment = segments.find(s => s.id === selectedSegmentId);
|
||||
|
||||
const handleVolumeChange = (e: React.ChangeEvent<HTMLInputElement>) => {
|
||||
if (!selectedSegmentId) return;
|
||||
const voiceVolume = parseFloat(e.target.value);
|
||||
updateSegment(selectedSegmentId, { voiceVolume });
|
||||
toast.success('音量已更新');
|
||||
};
|
||||
|
||||
const handleBgmVolumeChange = (e: React.ChangeEvent<HTMLInputElement>) => {
|
||||
if (!selectedSegmentId) return;
|
||||
const bgmVolume = parseFloat(e.target.value);
|
||||
updateSegment(selectedSegmentId, { bgmVolume });
|
||||
toast.success('背景音乐音量已更新');
|
||||
};
|
||||
|
||||
return (
|
||||
<div className="audio-mixing">
|
||||
<div className="audio-mixing-header">
|
||||
<h2>音频合成</h2>
|
||||
<p className="audio-mixing-desc">为每个分镜配置背景音乐和音量</p>
|
||||
</div>
|
||||
|
||||
<div className="audio-mixing-content">
|
||||
{/* 左侧：分镜列表 */}
|
||||
<div className="audio-mixing-sidebar">
|
||||
<h3>分镜列表</h3>
|
||||
<div className="segment-list">
|
||||
{segments.map((segment, index) => (
|
||||
<div
|
||||
key={segment.id}
|
||||
className={`segment-item ${segment.id === selectedSegmentId ? 'active' : ''}`}
|
||||
onClick={() => setSelectedSegmentId(segment.id)}
|
||||
>
|
||||
<span className="segment-index">镜头 {index + 1}</span>
|
||||
<span className="segment-duration">{segment.duration || 0}s</span>
|
||||
</div>
|
||||
))}
|
||||
</div>
|
||||
</div>
|
||||
|
||||
{/* 右侧：音频设置 */}
|
||||
<div className="audio-mixing-panel">
|
||||
{selectedSegment ? (
|
||||
<>
|
||||
<h3>分镜 {segments.findIndex(s => s.id === selectedSegmentId) + 1} 音频设置</h3>
|
||||
|
||||
<div className="audio-control">
|
||||
<label>人声音量</label>
|
||||
<div className="slider-container">
|
||||
<input
|
||||
type="range"
|
||||
min="0"
|
||||
max="2"
|
||||
step="0.1"
|
||||
value={selectedSegment.voiceVolume ?? 1}
|
||||
onChange={handleVolumeChange}
|
||||
/>
|
||||
<span className="slider-value">
|
||||
{Math.round((selectedSegment.voiceVolume ?? 1) * 100)}%
|
||||
</span>
|
||||
</div>
|
||||
</div>
|
||||
|
||||
<div className="audio-control">
|
||||
<label>背景音乐音量</label>
|
||||
<div className="slider-container">
|
||||
<input
|
||||
type="range"
|
||||
min="0"
|
||||
max="2"
|
||||
step="0.1"
|
||||
value={selectedSegment.bgmVolume ?? 0.3}
|
||||
onChange={handleBgmVolumeChange}
|
||||
/>
|
||||
<span className="slider-value">
|
||||
{Math.round((selectedSegment.bgmVolume ?? 0.3) * 100)}%
|
||||
</span>
|
||||
</div>
|
||||
</div>
|
||||
|
||||
<div className="audio-info">
|
||||
<p>人声将使用 Kling AI TTS 生成的音频</p>
|
||||
<p>背景音乐可在视频生成后添加</p>
|
||||
</div>
|
||||
</>
|
||||
) : (
|
||||
<div className="empty-state">
|
||||
<p>请先在第一步生成脚本</p>
|
||||
</div>
|
||||
)}
|
||||
</div>
|
||||
</div>
|
||||
</div>
|
||||
);
|
||||
}
|
||||
@@ -5,6 +5,26 @@
|
||||
* 请使用 Slider 组件或从 Slider.css 导入样式
|
||||
*/
|
||||
|
||||
/* 创作主题分组 */
|
||||
.topic-category-title {
|
||||
font-size: var(--font-sm);
|
||||
color: var(--text-secondary);
|
||||
margin-bottom: var(--spacing-sm);
|
||||
}
|
||||
|
||||
.topic-groups {
|
||||
display: grid;
|
||||
grid-template-columns: repeat(3, 1fr);
|
||||
gap: var(--spacing-sm);
|
||||
}
|
||||
|
||||
.topic-groups .option-card {
|
||||
min-width: 0;
|
||||
padding: var(--spacing-md);
|
||||
font-size: var(--font-sm);
|
||||
white-space: nowrap;
|
||||
}
|
||||
|
||||
/* 时长滑块容器 */
|
||||
.duration-slider-container {
|
||||
display: flex;
|
||||
|
||||
@@ -9,10 +9,18 @@ import ConfirmModal from '../../components/Modal/ConfirmModal';
|
||||
import './ScriptCreation.css';
|
||||
import '../../components/Slider/Slider.css';
|
||||
|
||||
const scriptTypeOptions = ['对比型', '恐吓型', '干货型', '共情型', '挑战型', '福利型'];
|
||||
// 创作主题配置
|
||||
const TOPIC_ITEMS = [
|
||||
'装修合同避坑',
|
||||
'装修全流程避坑',
|
||||
'装修材料避坑',
|
||||
'装修报价避坑',
|
||||
'全屋定制避坑',
|
||||
'装修常见问题',
|
||||
];
|
||||
|
||||
// 时长刻度配置
|
||||
const DURATION_MARKS = [30, 45, 60, 75];
|
||||
const DURATION_MARKS = [30, 60, 90, 120];
|
||||
|
||||
/**
|
||||
* 脚本生成页面
|
||||
@@ -25,10 +33,7 @@ export default function ScriptCreation() {
|
||||
const topic = useProjectStore(state => state.topic) || '';
|
||||
const setTopic = useProjectStore(state => state.setTopic);
|
||||
const scriptDuration = useProjectStore(state => state.scriptDuration) || 45;
|
||||
const scriptType = useProjectStore(state => state.scriptType);
|
||||
const setScriptDuration = useProjectStore(state => state.setScriptDuration);
|
||||
const setScriptType = useProjectStore(state => state.setScriptType);
|
||||
const selectedType = scriptType ? Math.max(0, scriptTypeOptions.indexOf(scriptType)) : 0;
|
||||
const [generating, setGenerating] = useState(false);
|
||||
const { submit } = useTask();
|
||||
|
||||
@@ -89,7 +94,7 @@ export default function ScriptCreation() {
|
||||
*/
|
||||
const doGenerate = async () => {
|
||||
if (!topic.trim()) {
|
||||
toast.warning('请输入创作灵感');
|
||||
toast.warning('请选择创作主题');
|
||||
return;
|
||||
}
|
||||
|
||||
@@ -100,7 +105,6 @@ export default function ScriptCreation() {
|
||||
{
|
||||
topic,
|
||||
duration: scriptDuration,
|
||||
style: scriptTypeOptions[selectedType],
|
||||
},
|
||||
{
|
||||
callbacks: {
|
||||
@@ -258,22 +262,21 @@ export default function ScriptCreation() {
|
||||
<div className="step-layout script-creation">
|
||||
{/* Left Panel - 参数配置 */}
|
||||
<div className="step-panel-left">
|
||||
{/* 创作主题 */}
|
||||
<div className="panel-section">
|
||||
<label className="panel-label">
|
||||
创作灵感
|
||||
<span style={{ color: 'var(--color-danger)', marginLeft: 4 }}>*</span>
|
||||
</label>
|
||||
<span className="panel-sublabel">
|
||||
输入主题关键词或粘贴视频链接（抖音、小红书、B站等）
|
||||
</span>
|
||||
<textarea
|
||||
placeholder="例如：家装隐蔽工程验收要注意什么？
|
||||
|
||||
或粘贴视频链接：抖音、小红书、B站等平台的视频链接，系统将自动提取文案"
|
||||
value={topic}
|
||||
onChange={e => setTopic(e.target.value)}
|
||||
style={{ resize: 'vertical', height: 200 }}
|
||||
/>
|
||||
<label className="panel-label">创作主题</label>
|
||||
<div className="topic-category-title">装修避坑</div>
|
||||
<div className="topic-groups">
|
||||
{TOPIC_ITEMS.map((item) => (
|
||||
<button
|
||||
key={item}
|
||||
className={`option-card ${topic === item ? 'selected' : ''}`}
|
||||
onClick={() => setTopic(item)}
|
||||
>
|
||||
{item}
|
||||
</button>
|
||||
))}
|
||||
</div>
|
||||
</div>
|
||||
|
||||
<div className="panel-section">
|
||||
@@ -286,12 +289,12 @@ export default function ScriptCreation() {
|
||||
type="range"
|
||||
className="slider-input"
|
||||
min={30}
|
||||
max={75}
|
||||
max={120}
|
||||
step={1}
|
||||
value={scriptDuration}
|
||||
onChange={e => setScriptDuration(Number(e.target.value))}
|
||||
style={
|
||||
{ '--slider-percent': `${((scriptDuration - 30) / 45) * 100}%` } as React.CSSProperties
|
||||
{ '--slider-percent': `${((scriptDuration - 30) / 90) * 100}%` } as React.CSSProperties
|
||||
}
|
||||
/>
|
||||
<div className="duration-marks">
|
||||
@@ -308,22 +311,6 @@ export default function ScriptCreation() {
|
||||
</div>
|
||||
</div>
|
||||
|
||||
<div className="panel-section">
|
||||
<label className="panel-label">脚本类型</label>
|
||||
<div className="option-cards">
|
||||
{scriptTypeOptions.map((opt, i) => (
|
||||
<button
|
||||
key={i}
|
||||
className={`option-card ${selectedType === i ? 'selected' : ''}`}
|
||||
onClick={() => setScriptType(opt)}
|
||||
title={getScriptTypeDesc(opt)}
|
||||
>
|
||||
{opt}
|
||||
</button>
|
||||
))}
|
||||
</div>
|
||||
</div>
|
||||
|
||||
{/* 生成按钮 */}
|
||||
<button
|
||||
className="btn btn-primary"
|
||||
@@ -607,9 +594,9 @@ export default function ScriptCreation() {
|
||||
</div>
|
||||
<p className="empty-state-title">暂无脚本内容</p>
|
||||
<p className="empty-state-desc">
|
||||
在左侧输入创作灵感，
|
||||
在左侧选择创作主题，
|
||||
<br />
|
||||
选择参数后点击"生成脚本"
|
||||
选择视频时长后点击"生成脚本"
|
||||
</p>
|
||||
</div>
|
||||
)}
|
||||
@@ -636,18 +623,3 @@ export default function ScriptCreation() {
|
||||
</div>
|
||||
);
|
||||
}
|
||||
|
||||
/**
|
||||
* 获取脚本类型描述（用于 tooltip）
|
||||
*/
|
||||
function getScriptTypeDesc(type: string): string {
|
||||
const descs: { [key: string]: string } = {
|
||||
对比型: '通过强烈对比突出效果差异，易引发转发',
|
||||
恐吓型: '直击痛点引发焦虑，再给出解决方案',
|
||||
干货型: '输出可落地的装修干货，建立专业度',
|
||||
共情型: '贴合业主心境引发共鸣，适合情感营销',
|
||||
挑战型: '设定具体挑战目标，自带话题性',
|
||||
福利型: '以福利为钩子吸引停留，引导留资',
|
||||
};
|
||||
return descs[type] || type;
|
||||
}
|
||||
|
||||
@@ -0,0 +1,256 @@
|
||||
/**
|
||||
* VideoEditing 样式
|
||||
* ==================
|
||||
*/
|
||||
|
||||
.video-editing {
|
||||
width: 100%;
|
||||
}
|
||||
|
||||
.editing-layout {
|
||||
display: grid;
|
||||
grid-template-columns: 280px 1fr;
|
||||
gap: var(--spacing-lg);
|
||||
margin-top: var(--spacing-md);
|
||||
}
|
||||
|
||||
.editing-sidebar {
|
||||
display: flex;
|
||||
flex-direction: column;
|
||||
gap: var(--spacing-sm);
|
||||
}
|
||||
|
||||
.sidebar-title {
|
||||
font-size: 13px;
|
||||
font-weight: 600;
|
||||
color: var(--text-secondary);
|
||||
text-transform: uppercase;
|
||||
letter-spacing: 0.05em;
|
||||
padding: 0 var(--spacing-xs);
|
||||
}
|
||||
|
||||
.segment-list {
|
||||
display: flex;
|
||||
flex-direction: column;
|
||||
gap: var(--spacing-xs);
|
||||
}
|
||||
|
||||
.segment-clip-item {
|
||||
background: var(--bg-card);
|
||||
border: 1px solid var(--border-light);
|
||||
border-radius: var(--radius-md);
|
||||
padding: var(--spacing-sm) var(--spacing-md);
|
||||
cursor: pointer;
|
||||
transition: all 0.15s ease;
|
||||
position: relative;
|
||||
overflow: hidden;
|
||||
}
|
||||
|
||||
.segment-clip-item:hover {
|
||||
border-color: var(--primary-light);
|
||||
background: var(--bg-hover);
|
||||
}
|
||||
|
||||
.segment-clip-item.active {
|
||||
border-color: var(--primary);
|
||||
background: color-mix(in srgb, var(--primary) 5%, var(--bg-card));
|
||||
}
|
||||
|
||||
.segment-clip-item.processing {
|
||||
opacity: 0.7;
|
||||
}
|
||||
|
||||
.segment-info {
|
||||
display: flex;
|
||||
justify-content: space-between;
|
||||
align-items: center;
|
||||
}
|
||||
|
||||
.segment-index {
|
||||
font-size: 13px;
|
||||
font-weight: 600;
|
||||
color: var(--text-primary);
|
||||
}
|
||||
|
||||
.segment-duration {
|
||||
font-size: 12px;
|
||||
color: var(--text-secondary);
|
||||
background: var(--bg-light);
|
||||
padding: 2px 6px;
|
||||
border-radius: var(--radius-sm);
|
||||
}
|
||||
|
||||
.segment-meta {
|
||||
margin-top: 4px;
|
||||
}
|
||||
|
||||
.audio-badge {
|
||||
display: inline-flex;
|
||||
align-items: center;
|
||||
gap: 3px;
|
||||
font-size: 11px;
|
||||
padding: 2px 6px;
|
||||
border-radius: var(--radius-sm);
|
||||
}
|
||||
|
||||
.audio-badge.has-audio {
|
||||
background: color-mix(in srgb, var(--success) 15%, transparent);
|
||||
color: var(--success);
|
||||
}
|
||||
|
||||
.audio-badge.no-audio {
|
||||
background: var(--bg-light);
|
||||
color: var(--text-secondary);
|
||||
}
|
||||
|
||||
.processing-overlay {
|
||||
position: absolute;
|
||||
inset: 0;
|
||||
background: rgba(255, 255, 255, 0.6);
|
||||
display: flex;
|
||||
align-items: center;
|
||||
justify-content: center;
|
||||
}
|
||||
|
||||
.mini-spinner {
|
||||
width: 16px;
|
||||
height: 16px;
|
||||
border: 2px solid var(--border-light);
|
||||
border-top-color: var(--primary);
|
||||
border-radius: 50%;
|
||||
animation: spin 0.6s linear infinite;
|
||||
}
|
||||
|
||||
.editing-panel {
|
||||
display: flex;
|
||||
flex-direction: column;
|
||||
gap: var(--spacing-md);
|
||||
}
|
||||
|
||||
.video-preview-area {
|
||||
background: var(--bg-card);
|
||||
border-radius: var(--radius-lg);
|
||||
overflow: hidden;
|
||||
aspect-ratio: 9/16;
|
||||
max-height: 400px;
|
||||
display: flex;
|
||||
align-items: center;
|
||||
justify-content: center;
|
||||
border: 1px solid var(--border-light);
|
||||
}
|
||||
|
||||
.preview-video {
|
||||
width: 100%;
|
||||
height: 100%;
|
||||
object-fit: contain;
|
||||
}
|
||||
|
||||
.preview-placeholder {
|
||||
color: var(--text-secondary);
|
||||
font-size: 14px;
|
||||
}
|
||||
|
||||
.clip-controls {
|
||||
background: var(--bg-card);
|
||||
border-radius: var(--radius-lg);
|
||||
padding: var(--spacing-md);
|
||||
border: 1px solid var(--border-light);
|
||||
display: flex;
|
||||
flex-direction: column;
|
||||
gap: var(--spacing-md);
|
||||
}
|
||||
|
||||
.control-section h4 {
|
||||
font-size: 13px;
|
||||
font-weight: 600;
|
||||
color: var(--text-primary);
|
||||
margin-bottom: var(--spacing-sm);
|
||||
}
|
||||
|
||||
.trim-controls {
|
||||
display: flex;
|
||||
align-items: center;
|
||||
gap: var(--spacing-md);
|
||||
flex-wrap: wrap;
|
||||
}
|
||||
|
||||
.trim-controls label {
|
||||
display: flex;
|
||||
align-items: center;
|
||||
gap: 6px;
|
||||
font-size: 13px;
|
||||
color: var(--text-secondary);
|
||||
}
|
||||
|
||||
.trim-input {
|
||||
width: 60px;
|
||||
padding: 4px 8px;
|
||||
border: 1px solid var(--border-light);
|
||||
border-radius: var(--radius-sm);
|
||||
font-size: 13px;
|
||||
text-align: center;
|
||||
}
|
||||
|
||||
.trim-duration {
|
||||
font-size: 13px;
|
||||
color: var(--primary);
|
||||
font-weight: 600;
|
||||
background: color-mix(in srgb, var(--primary) 8%, transparent);
|
||||
padding: 4px 10px;
|
||||
border-radius: var(--radius-sm);
|
||||
}
|
||||
|
||||
.checkbox-label {
|
||||
display: flex;
|
||||
align-items: center;
|
||||
gap: 8px;
|
||||
font-size: 13px;
|
||||
cursor: pointer;
|
||||
color: var(--text-primary);
|
||||
}
|
||||
|
||||
.audio-replace-info {
|
||||
display: flex;
|
||||
align-items: center;
|
||||
gap: 6px;
|
||||
font-size: 12px;
|
||||
color: var(--success);
|
||||
margin-top: 6px;
|
||||
}
|
||||
|
||||
.audio-replace-hint {
|
||||
font-size: 12px;
|
||||
color: var(--text-secondary);
|
||||
margin-top: 6px;
|
||||
font-style: italic;
|
||||
}
|
||||
|
||||
.segment-voiceover {
|
||||
background: var(--bg-card);
|
||||
border-radius: var(--radius-lg);
|
||||
padding: var(--spacing-md);
|
||||
border: 1px solid var(--border-light);
|
||||
}
|
||||
|
||||
.segment-voiceover h4 {
|
||||
font-size: 13px;
|
||||
font-weight: 600;
|
||||
color: var(--text-primary);
|
||||
margin-bottom: var(--spacing-sm);
|
||||
}
|
||||
|
||||
.voiceover-text {
|
||||
font-size: 13px;
|
||||
color: var(--text-secondary);
|
||||
line-height: 1.6;
|
||||
}
|
||||
|
||||
.no-segment {
|
||||
text-align: center;
|
||||
padding: var(--spacing-xl);
|
||||
color: var(--text-secondary);
|
||||
}
|
||||
|
||||
@keyframes spin {
|
||||
to { transform: rotate(360deg); }
|
||||
}
|
||||
@@ -0,0 +1,294 @@
|
||||
/**
|
||||
* 视频编辑页面 (Step 3.5)
|
||||
* ======================
|
||||
*
|
||||
* 手动剪辑：裁剪视频时长、调整播放速度、替换/静音原音。
|
||||
* 位于字幕压制之后、封面制作之前。
|
||||
*/
|
||||
|
||||
import { useState, useCallback, useRef } from 'react';
|
||||
import { invoke } from '@tauri-apps/api/core';
|
||||
import { useProjectStore } from '../../store';
|
||||
import { useVoiceStore } from '../../store/voiceStore';
|
||||
import { getCurrentProjectId } from '../../api/modules/localStorage';
|
||||
import { replaceAudioTrack } from '../../api/modules/voice';
|
||||
import { toast } from '../../store/uiStore';
|
||||
import './VideoEditing.css';
|
||||
|
||||
interface SegmentClip {
|
||||
id: string;
|
||||
duration: number; // 裁剪后时长（秒）
|
||||
trimStart: number; // 裁剪起始点（秒）
|
||||
trimEnd: number; // 裁剪结束点（秒）
|
||||
muteOriginal: boolean;
|
||||
replaceAudioId?: string;
|
||||
videoPath: string;
|
||||
voiceover: string;
|
||||
type: 'segment' | 'empty_shot';
|
||||
}
|
||||
|
||||
export default function VideoEditing() {
|
||||
const segments = useProjectStore(state => state.segments);
|
||||
const updateSegment = useProjectStore(state => state.updateSegment);
|
||||
const projectId = getCurrentProjectId();
|
||||
|
||||
const { getAudioForSegment } = useVoiceStore();
|
||||
|
||||
const [activeId, setActiveId] = useState<string>(segments[0]?.id?.toString() || '');
|
||||
const [isProcessing, setIsProcessing] = useState(false);
|
||||
const [processingId, setProcessingId] = useState<string | null>(null);
|
||||
|
||||
// 当前分镜
|
||||
const activeSegment = segments.find(s => s.id?.toString() === activeId);
|
||||
const audioMeta = activeSegment ? getAudioForSegment(activeSegment.id?.toString() || '') : undefined;
|
||||
|
||||
// 裁剪状态
|
||||
const [trimStart, setTrimStart] = useState(0);
|
||||
const [trimEnd, setTrimEnd] = useState(
|
||||
activeSegment?.duration ? parseInt(activeSegment.duration.replace(/[^0-9]/g, '') || '0') : 0
|
||||
);
|
||||
const [muteOriginal, setMuteOriginal] = useState(false);
|
||||
const videoRef = useRef<HTMLVideoElement>(null);
|
||||
|
||||
// 分镜切换时重置状态
|
||||
const handleSegmentSelect = (id: string) => {
|
||||
setActiveId(id);
|
||||
const seg = segments.find(s => s.id?.toString() === id);
|
||||
if (seg) {
|
||||
const dur = parseInt(seg.duration?.replace(/[^0-9]/g, '') || '0');
|
||||
setTrimStart(0);
|
||||
setTrimEnd(dur);
|
||||
setMuteOriginal(false);
|
||||
}
|
||||
};
|
||||
|
||||
// 应用裁剪并静音/替换音频
|
||||
const handleApplyClip = useCallback(async () => {
|
||||
if (!activeSegment || !activeId || !projectId) return;
|
||||
|
||||
setIsProcessing(true);
|
||||
setProcessingId(activeId);
|
||||
|
||||
try {
|
||||
const segId = activeId;
|
||||
const clippedDuration = `${trimEnd - trimStart}s`;
|
||||
|
||||
// 如果选择了音频替换
|
||||
if (audioMeta?.filePath && activeSegment.videoPath) {
|
||||
// 音频替换：用 TTS 配音替换视频原音
|
||||
// 输出路径基于原视频，添加后缀标记
|
||||
const outputPath = activeSegment.videoPath.replace('.mp4', `_dubbed_${Date.now()}.mp4`);
|
||||
|
||||
await replaceAudioTrack({
|
||||
videoPath: activeSegment.videoPath,
|
||||
audioPath: audioMeta.filePath,
|
||||
outputPath,
|
||||
});
|
||||
|
||||
// 更新分镜数据（视频路径 + 时长）
|
||||
updateSegment(activeSegment.id!, {
|
||||
videoPath: outputPath,
|
||||
duration: clippedDuration,
|
||||
});
|
||||
|
||||
toast.success(`分镜 ${segId} 音频替换完成`);
|
||||
} else if (muteOriginal && activeSegment.videoPath) {
|
||||
// 静音标记：记录到分镜元数据，后续合成时处理
|
||||
updateSegment(activeSegment.id!, {
|
||||
duration: clippedDuration,
|
||||
// 静音频标记（需要在视频合成时处理）
|
||||
});
|
||||
toast.info('已标记静音，将在最终合成时处理');
|
||||
} else {
|
||||
// 仅更新时长
|
||||
updateSegment(activeSegment.id!, {
|
||||
duration: clippedDuration,
|
||||
});
|
||||
}
|
||||
|
||||
toast.success('剪辑参数已保存');
|
||||
} catch (err) {
|
||||
console.error('[VideoEditing] 处理失败:', err);
|
||||
toast.error(`处理失败: ${err instanceof Error ? err.message : String(err)}`);
|
||||
} finally {
|
||||
setIsProcessing(false);
|
||||
setProcessingId(null);
|
||||
}
|
||||
}, [activeSegment, activeId, projectId, audioMeta, muteOriginal, trimStart, trimEnd, updateSegment]);
|
||||
|
||||
// 视频预览时间更新
|
||||
const handleTimeUpdate = () => {
|
||||
if (videoRef.current) {
|
||||
const current = videoRef.current.currentTime;
|
||||
// 限制在裁剪范围内
|
||||
if (current < trimStart) {
|
||||
videoRef.current.currentTime = trimStart;
|
||||
} else if (current > trimEnd) {
|
||||
videoRef.current.currentTime = trimEnd;
|
||||
videoRef.current.pause();
|
||||
}
|
||||
}
|
||||
};
|
||||
|
||||
// 计算总剪辑后时长
|
||||
const totalClippedDuration = segments.reduce((sum, s) => {
|
||||
const dur = parseInt(s.duration?.replace(/[^0-9]/g, '') || '0');
|
||||
return sum + dur;
|
||||
}, 0);
|
||||
|
||||
return (
|
||||
<div className="video-editing">
|
||||
<div className="step-header">
|
||||
<h2>视频剪辑</h2>
|
||||
<p className="step-desc">
|
||||
裁剪视频时长、静音原音、替换为配音。总时长：{totalClippedDuration}s
|
||||
</p>
|
||||
</div>
|
||||
|
||||
<div className="editing-layout">
|
||||
{/* 左侧：分镜列表 */}
|
||||
<div className="editing-sidebar">
|
||||
<div className="sidebar-title">分镜列表</div>
|
||||
<div className="segment-list">
|
||||
{segments.map((seg, i) => {
|
||||
const segId = seg.id?.toString() || String(i);
|
||||
const audio = getAudioForSegment(segId);
|
||||
const dur = parseInt(seg.duration?.replace(/[^0-9]/g, '') || '0');
|
||||
const isActive = segId === activeId;
|
||||
const isProcessingThis = processingId === segId;
|
||||
|
||||
return (
|
||||
<div
|
||||
key={segId}
|
||||
className={`segment-clip-item ${isActive ? 'active' : ''} ${isProcessingThis ? 'processing' : ''}`}
|
||||
onClick={() => handleSegmentSelect(segId)}
|
||||
>
|
||||
<div className="segment-info">
|
||||
<span className="segment-index">镜头 {i + 1}</span>
|
||||
<span className="segment-duration">{dur}s</span>
|
||||
</div>
|
||||
<div className="segment-meta">
|
||||
{audio ? (
|
||||
<span className="audio-badge has-audio">
|
||||
<svg width="10" height="10" viewBox="0 0 24 24" fill="currentColor">
|
||||
<path d="M12 3v10.55A4 4 0 1014 17V7h4V3h-6z"/>
|
||||
</svg>
|
||||
配音
|
||||
</span>
|
||||
) : (
|
||||
<span className="audio-badge no-audio">无配音</span>
|
||||
)}
|
||||
</div>
|
||||
{isProcessingThis && (
|
||||
<div className="processing-overlay">
|
||||
<div className="mini-spinner" />
|
||||
</div>
|
||||
)}
|
||||
</div>
|
||||
);
|
||||
})}
|
||||
</div>
|
||||
</div>
|
||||
|
||||
{/* 右侧：编辑控制区 */}
|
||||
<div className="editing-panel">
|
||||
{activeSegment ? (
|
||||
<>
|
||||
{/* 视频预览 */}
|
||||
<div className="video-preview-area">
|
||||
{activeSegment.videoPath ? (
|
||||
<video
|
||||
ref={videoRef}
|
||||
src={`file://${activeSegment.videoPath}`}
|
||||
controls
|
||||
className="preview-video"
|
||||
onTimeUpdate={handleTimeUpdate}
|
||||
/>
|
||||
) : (
|
||||
<div className="preview-placeholder">
|
||||
<span>暂无视频预览</span>
|
||||
</div>
|
||||
)}
|
||||
</div>
|
||||
|
||||
{/* 裁剪控制 */}
|
||||
<div className="clip-controls">
|
||||
<div className="control-section">
|
||||
<h4>裁剪范围</h4>
|
||||
<div className="trim-controls">
|
||||
<label>
|
||||
起始 <input
|
||||
type="number"
|
||||
min={0}
|
||||
max={trimEnd - 1}
|
||||
value={trimStart}
|
||||
onChange={e => setTrimStart(Math.max(0, parseInt(e.target.value) || 0))}
|
||||
className="trim-input"
|
||||
/>s
|
||||
</label>
|
||||
<label>
|
||||
结束 <input
|
||||
type="number"
|
||||
min={trimStart + 1}
|
||||
value={trimEnd}
|
||||
onChange={e => setTrimEnd(parseInt(e.target.value) || trimEnd)}
|
||||
className="trim-input"
|
||||
/>s
|
||||
</label>
|
||||
<span className="trim-duration">
|
||||
时长：{Math.max(0, trimEnd - trimStart)}s
|
||||
</span>
|
||||
</div>
|
||||
</div>
|
||||
|
||||
{/* 音频替换 */}
|
||||
<div className="control-section">
|
||||
<h4>音频设置</h4>
|
||||
<label className="checkbox-label">
|
||||
<input
|
||||
type="checkbox"
|
||||
checked={muteOriginal}
|
||||
onChange={e => setMuteOriginal(e.target.checked)}
|
||||
/>
|
||||
静音原音
|
||||
</label>
|
||||
{audioMeta && (
|
||||
<div className="audio-replace-info">
|
||||
<svg width="14" height="14" viewBox="0 0 24 24" fill="currentColor">
|
||||
<path d="M12 3v10.55A4 4 0 1014 17V7h4V3h-6z"/>
|
||||
</svg>
|
||||
已选择配音：{audioMeta.name}
|
||||
</div>
|
||||
)}
|
||||
{!audioMeta && (
|
||||
<div className="audio-replace-hint">
|
||||
请前往「配音管理」添加配音后再替换
|
||||
</div>
|
||||
)}
|
||||
</div>
|
||||
|
||||
<button
|
||||
className="btn btn-primary"
|
||||
onClick={handleApplyClip}
|
||||
disabled={isProcessing || !activeSegment.videoPath}
|
||||
>
|
||||
{isProcessing ? '处理中...' : '应用剪辑'}
|
||||
</button>
|
||||
</div>
|
||||
|
||||
{/* 分镜旁白 */}
|
||||
<div className="segment-voiceover">
|
||||
<h4>旁白文本</h4>
|
||||
<p className="voiceover-text">{activeSegment.voiceover || '暂无旁白'}</p>
|
||||
</div>
|
||||
</>
|
||||
) : (
|
||||
<div className="no-segment">
|
||||
<p>请从左侧选择一个分镜进行编辑</p>
|
||||
</div>
|
||||
)}
|
||||
</div>
|
||||
</div>
|
||||
</div>
|
||||
);
|
||||
}
|
||||
@@ -0,0 +1,245 @@
|
||||
/**
|
||||
* VoiceDubbing 样式
|
||||
* ==================
|
||||
*/
|
||||
|
||||
.voice-dubbing {
|
||||
width: 100%;
|
||||
}
|
||||
|
||||
.dubbing-layout {
|
||||
display: grid;
|
||||
grid-template-columns: 1fr 1fr;
|
||||
gap: var(--spacing-lg);
|
||||
margin-top: var(--spacing-md);
|
||||
}
|
||||
|
||||
.voice-panel,
|
||||
.mapping-panel {
|
||||
display: flex;
|
||||
flex-direction: column;
|
||||
gap: var(--spacing-md);
|
||||
}
|
||||
|
||||
.panel-section {
|
||||
background: var(--bg-card);
|
||||
border: 1px solid var(--border-light);
|
||||
border-radius: var(--radius-lg);
|
||||
padding: var(--spacing-md);
|
||||
}
|
||||
|
||||
.panel-section h4 {
|
||||
font-size: 13px;
|
||||
font-weight: 600;
|
||||
color: var(--text-primary);
|
||||
margin-bottom: var(--spacing-sm);
|
||||
}
|
||||
|
||||
/* 音色网格 */
|
||||
.voice-grid {
|
||||
display: grid;
|
||||
grid-template-columns: 1fr 1fr;
|
||||
gap: var(--spacing-xs);
|
||||
}
|
||||
|
||||
.voice-card {
|
||||
border: 1px solid var(--border-light);
|
||||
border-radius: var(--radius-md);
|
||||
padding: var(--spacing-sm);
|
||||
cursor: pointer;
|
||||
transition: all 0.15s ease;
|
||||
background: var(--bg-primary);
|
||||
}
|
||||
|
||||
.voice-card:hover {
|
||||
border-color: var(--primary-light);
|
||||
background: var(--bg-hover);
|
||||
}
|
||||
|
||||
.voice-card.selected {
|
||||
border-color: var(--primary);
|
||||
background: color-mix(in srgb, var(--primary) 5%, var(--bg-card));
|
||||
}
|
||||
|
||||
.voice-name {
|
||||
font-size: 13px;
|
||||
font-weight: 600;
|
||||
color: var(--text-primary);
|
||||
display: flex;
|
||||
align-items: center;
|
||||
gap: 6px;
|
||||
}
|
||||
|
||||
.recommended-tag {
|
||||
font-size: 10px;
|
||||
background: color-mix(in srgb, var(--primary) 15%, transparent);
|
||||
color: var(--primary);
|
||||
padding: 1px 5px;
|
||||
border-radius: var(--radius-sm);
|
||||
font-weight: 500;
|
||||
}
|
||||
|
||||
.voice-desc {
|
||||
font-size: 11px;
|
||||
color: var(--text-secondary);
|
||||
margin-top: 2px;
|
||||
}
|
||||
|
||||
/* 试听 */
|
||||
.preview-row {
|
||||
display: flex;
|
||||
gap: var(--spacing-sm);
|
||||
align-items: flex-end;
|
||||
}
|
||||
|
||||
.preview-text {
|
||||
flex: 1;
|
||||
padding: var(--spacing-sm);
|
||||
border: 1px solid var(--border-light);
|
||||
border-radius: var(--radius-md);
|
||||
font-size: 13px;
|
||||
resize: none;
|
||||
line-height: 1.5;
|
||||
font-family: inherit;
|
||||
}
|
||||
|
||||
.preview-audio {
|
||||
width: 100%;
|
||||
height: 36px;
|
||||
margin-top: var(--spacing-sm);
|
||||
}
|
||||
|
||||
/* 批量合成 */
|
||||
.batch-info {
|
||||
display: flex;
|
||||
gap: var(--spacing-md);
|
||||
font-size: 12px;
|
||||
color: var(--text-secondary);
|
||||
margin-bottom: var(--spacing-sm);
|
||||
flex-wrap: wrap;
|
||||
}
|
||||
|
||||
.batch-btn {
|
||||
width: 100%;
|
||||
}
|
||||
|
||||
.progress-bar {
|
||||
height: 4px;
|
||||
background: var(--bg-light);
|
||||
border-radius: 2px;
|
||||
overflow: hidden;
|
||||
margin-top: var(--spacing-sm);
|
||||
}
|
||||
|
||||
.progress-fill {
|
||||
height: 100%;
|
||||
background: var(--primary);
|
||||
transition: width 0.3s ease;
|
||||
}
|
||||
|
||||
/* 分镜配音列表 */
|
||||
.segment-voice-list {
|
||||
display: flex;
|
||||
flex-direction: column;
|
||||
gap: var(--spacing-xs);
|
||||
max-height: 400px;
|
||||
overflow-y: auto;
|
||||
}
|
||||
|
||||
.seg-voice-item {
|
||||
border: 1px solid var(--border-light);
|
||||
border-radius: var(--radius-md);
|
||||
padding: var(--spacing-sm);
|
||||
background: var(--bg-primary);
|
||||
}
|
||||
|
||||
.seg-voice-item.empty-shot {
|
||||
opacity: 0.5;
|
||||
}
|
||||
|
||||
.seg-voice-info {
|
||||
display: flex;
|
||||
flex-direction: column;
|
||||
gap: 4px;
|
||||
}
|
||||
|
||||
.seg-voice-index {
|
||||
font-size: 12px;
|
||||
font-weight: 600;
|
||||
color: var(--text-primary);
|
||||
}
|
||||
|
||||
.seg-has-audio {
|
||||
display: flex;
|
||||
flex-direction: column;
|
||||
gap: 4px;
|
||||
}
|
||||
|
||||
.audio-name {
|
||||
font-size: 11px;
|
||||
color: var(--success);
|
||||
}
|
||||
|
||||
.seg-audio-player {
|
||||
height: 28px;
|
||||
width: 100%;
|
||||
}
|
||||
|
||||
.seg-no-audio {
|
||||
font-size: 11px;
|
||||
color: var(--text-secondary);
|
||||
}
|
||||
|
||||
.seg-voiceover {
|
||||
font-size: 11px;
|
||||
color: var(--text-secondary);
|
||||
margin-top: 4px;
|
||||
line-height: 1.4;
|
||||
overflow: hidden;
|
||||
text-overflow: ellipsis;
|
||||
white-space: nowrap;
|
||||
}
|
||||
|
||||
/* 音频文件库 */
|
||||
.audio-file-list {
|
||||
display: flex;
|
||||
flex-direction: column;
|
||||
gap: var(--spacing-xs);
|
||||
}
|
||||
|
||||
.audio-file-item {
|
||||
display: flex;
|
||||
align-items: center;
|
||||
gap: var(--spacing-sm);
|
||||
padding: var(--spacing-xs) 0;
|
||||
border-bottom: 1px solid var(--border-light);
|
||||
}
|
||||
|
||||
.audio-file-item:last-child {
|
||||
border-bottom: none;
|
||||
}
|
||||
|
||||
.audio-file-info {
|
||||
flex: 1;
|
||||
min-width: 0;
|
||||
}
|
||||
|
||||
.audio-file-name {
|
||||
font-size: 12px;
|
||||
font-weight: 500;
|
||||
color: var(--text-primary);
|
||||
display: block;
|
||||
overflow: hidden;
|
||||
text-overflow: ellipsis;
|
||||
white-space: nowrap;
|
||||
}
|
||||
|
||||
.audio-file-size {
|
||||
font-size: 11px;
|
||||
color: var(--text-secondary);
|
||||
}
|
||||
|
||||
.audio-file-player {
|
||||
height: 28px;
|
||||
flex-shrink: 0;
|
||||
}
|
||||
@@ -0,0 +1,316 @@
|
||||
/**
|
||||
* 配音管理页面
|
||||
* =============
|
||||
*
|
||||
* TTS 文本转语音：选择音色、批量合成旁白配音。
|
||||
* 管理项目音频文件，关联到分镜。
|
||||
*/
|
||||
|
||||
import { useState, useEffect, useCallback, useRef } from 'react';
|
||||
import { useProjectStore } from '../../store';
|
||||
import { useVoiceStore } from '../../store/voiceStore';
|
||||
import { getCurrentProjectId } from '../../api/modules/localStorage';
|
||||
import { synthesizeTTS, synthesizeBatchTTS } from '../../api/modules/voice';
|
||||
import { saveAudio } from '../../api/modules/voice';
|
||||
import { toast } from '../../store/uiStore';
|
||||
import './VoiceDubbing.css';
|
||||
|
||||
export default function VoiceDubbing() {
|
||||
const segments = useProjectStore(state => state.segments);
|
||||
const updateSegment = useProjectStore(state => state.updateSegment);
|
||||
const projectId = getCurrentProjectId();
|
||||
|
||||
const {
|
||||
presetVoices,
|
||||
selectedVoiceId,
|
||||
loadPresetVoices,
|
||||
setSelectedVoiceId,
|
||||
projectAudios,
|
||||
loadProjectAudios,
|
||||
getAudioForSegment,
|
||||
setAudioMapping,
|
||||
} = useVoiceStore();
|
||||
|
||||
const [isSynthesizing, setIsSynthesizing] = useState(false);
|
||||
const [synthProgress, setSynthProgress] = useState(0);
|
||||
const [synthTotal, setSynthTotal] = useState(0);
|
||||
const [customText, setCustomText] = useState('');
|
||||
const [customPreviewUrl, setCustomPreviewUrl] = useState<string | null>(null);
|
||||
const audioPreviewRef = useRef<HTMLAudioElement>(null);
|
||||
|
||||
// 加载音色和项目音频
|
||||
useEffect(() => {
|
||||
loadPresetVoices();
|
||||
if (projectId) {
|
||||
loadProjectAudios(projectId);
|
||||
}
|
||||
}, [projectId]);
|
||||
|
||||
// 获取有旁白文本的分镜（排除空镜）
|
||||
const voicedSegments = segments.filter(s => s.type !== 'empty_shot' && s.voiceover);
|
||||
const totalChars = voicedSegments.reduce((sum, s) => sum + (s.voiceover?.length || 0), 0);
|
||||
|
||||
// 批量合成所有旁白
|
||||
const handleBatchSynthesize = useCallback(async () => {
|
||||
if (!projectId || voicedSegments.length === 0) {
|
||||
toast.warn('没有需要合成的旁白');
|
||||
return;
|
||||
}
|
||||
|
||||
setIsSynthesizing(true);
|
||||
setSynthProgress(0);
|
||||
setSynthTotal(voicedSegments.length);
|
||||
|
||||
let successCount = 0;
|
||||
let failCount = 0;
|
||||
|
||||
try {
|
||||
for (let i = 0; i < voicedSegments.length; i++) {
|
||||
const seg = voicedSegments[i];
|
||||
const segId = seg.id?.toString() || String(i);
|
||||
const text = seg.voiceover || '';
|
||||
|
||||
setSynthProgress(i + 1);
|
||||
|
||||
try {
|
||||
// 同步 TTS 合成（≤200字）
|
||||
const result = await synthesizeTTS({
|
||||
text,
|
||||
voiceId: selectedVoiceId,
|
||||
speed: 1.0,
|
||||
});
|
||||
|
||||
if (!result.audioBase64) {
|
||||
throw new Error('未返回音频数据');
|
||||
}
|
||||
|
||||
// 保存到本地
|
||||
const audioId = `tts_${segId}_${Date.now()}`;
|
||||
const meta = await saveAudio({
|
||||
projectId,
|
||||
audioId,
|
||||
audioData: result.audioBase64,
|
||||
name: `旁白-${segId}`,
|
||||
voiceId: selectedVoiceId,
|
||||
duration: 0, // 暂时无法获取时长
|
||||
segmentId: segId,
|
||||
});
|
||||
|
||||
// 关联到分镜
|
||||
setAudioMapping(segId, meta.id);
|
||||
|
||||
// 更新分镜 audioPath
|
||||
updateSegment(seg.id!, { audioPath: meta.filePath });
|
||||
successCount++;
|
||||
} catch (err) {
|
||||
console.error(`[VoiceDubbing] 分镜 ${segId} 合成失败:`, err);
|
||||
failCount++;
|
||||
}
|
||||
}
|
||||
|
||||
if (successCount > 0) {
|
||||
toast.success(`配音合成完成：成功 ${successCount} 段${failCount > 0 ? `，失败 ${failCount} 段` : ''}`);
|
||||
} else {
|
||||
toast.error('配音合成全部失败');
|
||||
}
|
||||
} finally {
|
||||
setIsSynthesizing(false);
|
||||
setSynthProgress(0);
|
||||
}
|
||||
}, [projectId, voicedSegments, selectedVoiceId, updateSegment, setAudioMapping]);
|
||||
|
||||
// 试听音色
|
||||
const handlePreviewVoice = useCallback(async () => {
|
||||
if (!customText.trim()) {
|
||||
toast.warn('请输入要预览的文本');
|
||||
return;
|
||||
}
|
||||
|
||||
try {
|
||||
setCustomPreviewUrl(null);
|
||||
const result = await synthesizeTTS({
|
||||
text: customText.slice(0, 200),
|
||||
voiceId: selectedVoiceId,
|
||||
speed: 1.0,
|
||||
});
|
||||
|
||||
if (!result.audioBase64) {
|
||||
throw new Error('未返回音频数据');
|
||||
}
|
||||
|
||||
const audioBlob = new Blob(
|
||||
[Uint8Array.from(atob(result.audioBase64), c => c.charCodeAt(0))],
|
||||
{ type: 'audio/mp3' }
|
||||
);
|
||||
const url = URL.createObjectURL(audioBlob);
|
||||
setCustomPreviewUrl(url);
|
||||
} catch (err) {
|
||||
toast.error(`试听失败: ${err instanceof Error ? err.message : String(err)}`);
|
||||
}
|
||||
}, [customText, selectedVoiceId]);
|
||||
|
||||
// 将项目音频关联到分镜
|
||||
const handleAssignToSegment = (audioId: string, segmentId: string) => {
|
||||
setAudioMapping(segmentId, audioId);
|
||||
|
||||
// 同时更新分镜的 audioPath
|
||||
const audio = projectAudios.find(a => a.id === audioId);
|
||||
if (audio) {
|
||||
updateSegment(parseInt(segmentId), { audioPath: audio.filePath });
|
||||
}
|
||||
toast.success('已关联到分镜');
|
||||
};
|
||||
|
||||
const selectedVoice = presetVoices.find(v => v.voiceId === selectedVoiceId);
|
||||
|
||||
return (
|
||||
<div className="voice-dubbing">
|
||||
<div className="step-header">
|
||||
<h2>配音管理</h2>
|
||||
<p className="step-desc">
|
||||
{voicedSegments.length} 个分镜待配音，共 {totalChars} 字
|
||||
</p>
|
||||
</div>
|
||||
|
||||
<div className="dubbing-layout">
|
||||
{/* 左侧：音色选择 + 批量合成 */}
|
||||
<div className="voice-panel">
|
||||
{/* 音色选择 */}
|
||||
<div className="panel-section">
|
||||
<h4>选择音色</h4>
|
||||
<div className="voice-grid">
|
||||
{presetVoices.map(voice => (
|
||||
<div
|
||||
key={voice.voiceId}
|
||||
className={`voice-card ${voice.voiceId === selectedVoiceId ? 'selected' : ''}`}
|
||||
onClick={() => setSelectedVoiceId(voice.voiceId)}
|
||||
>
|
||||
<div className="voice-name">
|
||||
{voice.name}
|
||||
{voice.recommended && <span className="recommended-tag">推荐</span>}
|
||||
</div>
|
||||
<div className="voice-desc">{voice.description}</div>
|
||||
</div>
|
||||
))}
|
||||
</div>
|
||||
</div>
|
||||
|
||||
{/* 音色试听 */}
|
||||
<div className="panel-section">
|
||||
<h4>试听音色</h4>
|
||||
<div className="preview-row">
|
||||
<textarea
|
||||
className="preview-text"
|
||||
value={customText}
|
||||
onChange={e => setCustomText(e.target.value)}
|
||||
placeholder="输入文本试听音色（≤200字）..."
|
||||
rows={3}
|
||||
maxLength={200}
|
||||
/>
|
||||
<button
|
||||
className="btn btn-secondary"
|
||||
onClick={handlePreviewVoice}
|
||||
disabled={!customText.trim()}
|
||||
>
|
||||
试听
|
||||
</button>
|
||||
</div>
|
||||
{customPreviewUrl && (
|
||||
<audio ref={audioPreviewRef} src={customPreviewUrl} controls className="preview-audio" />
|
||||
)}
|
||||
</div>
|
||||
|
||||
{/* 批量合成 */}
|
||||
<div className="panel-section">
|
||||
<h4>批量配音</h4>
|
||||
<div className="batch-info">
|
||||
<span>音色：{selectedVoice?.name}</span>
|
||||
<span>分镜：{voicedSegments.length} 个</span>
|
||||
<span>字数：约 {totalChars} 字</span>
|
||||
</div>
|
||||
<button
|
||||
className="btn btn-primary batch-btn"
|
||||
onClick={handleBatchSynthesize}
|
||||
disabled={isSynthesizing || voicedSegments.length === 0}
|
||||
>
|
||||
{isSynthesizing
|
||||
? `合成中... ${synthProgress}/${synthTotal}`
|
||||
: `为 ${voicedSegments.length} 个分镜生成配音`}
|
||||
</button>
|
||||
{isSynthesizing && (
|
||||
<div className="progress-bar">
|
||||
<div
|
||||
className="progress-fill"
|
||||
style={{ width: `${(synthProgress / synthTotal) * 100}%` }}
|
||||
/>
|
||||
</div>
|
||||
)}
|
||||
</div>
|
||||
</div>
|
||||
|
||||
{/* 右侧：分镜-配音映射 */}
|
||||
<div className="mapping-panel">
|
||||
<div className="panel-section">
|
||||
<h4>分镜配音状态</h4>
|
||||
<div className="segment-voice-list">
|
||||
{segments.map((seg, i) => {
|
||||
const segId = seg.id?.toString() || String(i);
|
||||
const audio = getAudioForSegment(segId);
|
||||
const isEmptyShot = seg.type === 'empty_shot';
|
||||
|
||||
return (
|
||||
<div key={segId} className={`seg-voice-item ${isEmptyShot ? 'empty-shot' : ''}`}>
|
||||
<div className="seg-voice-info">
|
||||
<span className="seg-voice-index">
|
||||
{isEmptyShot ? '🎬' : '🎙️'} 镜头 {i + 1}
|
||||
</span>
|
||||
{audio ? (
|
||||
<div className="seg-has-audio">
|
||||
<span className="audio-name">{audio.name}</span>
|
||||
<audio
|
||||
src={`file://${audio.filePath}`}
|
||||
controls
|
||||
className="seg-audio-player"
|
||||
/>
|
||||
</div>
|
||||
) : (
|
||||
<span className="seg-no-audio">
|
||||
{isEmptyShot ? '空镜无需配音' : '未配音'}
|
||||
</span>
|
||||
)}
|
||||
</div>
|
||||
<div className="seg-voiceover">{seg.voiceover || ''}</div>
|
||||
</div>
|
||||
);
|
||||
})}
|
||||
</div>
|
||||
</div>
|
||||
|
||||
{/* 音频文件列表 */}
|
||||
{projectAudios.length > 0 && (
|
||||
<div className="panel-section">
|
||||
<h4>音频文件库</h4>
|
||||
<div className="audio-file-list">
|
||||
{projectAudios.map(audio => (
|
||||
<div key={audio.id} className="audio-file-item">
|
||||
<div className="audio-file-info">
|
||||
<span className="audio-file-name">{audio.name}</span>
|
||||
<span className="audio-file-size">
|
||||
{(audio.fileSize / 1024).toFixed(1)} KB
|
||||
</span>
|
||||
</div>
|
||||
<audio
|
||||
src={`file://${audio.filePath}`}
|
||||
controls
|
||||
className="audio-file-player"
|
||||
/>
|
||||
</div>
|
||||
))}
|
||||
</div>
|
||||
</div>
|
||||
)}
|
||||
</div>
|
||||
</div>
|
||||
</div>
|
||||
);
|
||||
}
|
||||
@@ -1,4 +1,5 @@
|
||||
import ScriptCreation from './ScriptCreation';
|
||||
import AudioMixing from './AudioMixing';
|
||||
import VideoGeneration from './VideoGeneration';
|
||||
import SubtitleBurning from './SubtitleBurning';
|
||||
import CoverDesign from './CoverDesign';
|
||||
@@ -11,17 +12,18 @@ import './VideoCreation.css';
|
||||
|
||||
const steps = [
|
||||
{ id: 1, label: '脚本生成' },
|
||||
{ id: 2, label: '视频生成' },
|
||||
{ id: 3, label: '字幕压制' },
|
||||
{ id: 4, label: '封面制作' },
|
||||
{ id: 5, label: '视频合成' },
|
||||
{ id: 2, label: '音频合成' },
|
||||
{ id: 3, label: '视频生成' },
|
||||
{ id: 4, label: '字幕压制' },
|
||||
{ id: 5, label: '封面制作' },
|
||||
{ id: 6, label: '视频合成' },
|
||||
];
|
||||
|
||||
/**
|
||||
* 视频创作流程容器
|
||||
*
|
||||
* 流程：
|
||||
* - 5 步流程：脚本生成 -> 视频生成 -> 字幕压制 -> 封面制作 -> 视频合成
|
||||
* - 6 步流程：脚本生成 -> 音频合成 -> 视频生成 -> 字幕压制 -> 封面制作 -> 视频合成
|
||||
* - 使用全局 projectStore 状态管理
|
||||
*/
|
||||
function VideoCreationContent() {
|
||||
@@ -44,30 +46,29 @@ function VideoCreationContent() {
|
||||
|
||||
// 步骤完成检查
|
||||
const isStep1Complete = segments.length > 0;
|
||||
const isStep2Complete = isStep1Complete && segments.every(s => s.videoPath);
|
||||
// Step 2 音频合成：只需有分镜即可（可以没有音频文件）
|
||||
const isStep2Complete = isStep1Complete;
|
||||
// Step 3 视频生成：所有分镜必须有视频路径
|
||||
const isStep3Complete = isStep2Complete && segments.every(s => s.videoPath);
|
||||
// Step 4+ 字幕压制：需要视频生成完成
|
||||
const isStep4PlusComplete = isStep3Complete;
|
||||
|
||||
// 判断用户能否进入某步骤
|
||||
const canAccessStep = (stepId: number) => {
|
||||
if (stepId === 1) {
|
||||
return true;
|
||||
}
|
||||
if (stepId === 2) {
|
||||
return isStep1Complete;
|
||||
}
|
||||
if (stepId === 3) {
|
||||
return isStep2Complete;
|
||||
}
|
||||
return isStep2Complete;
|
||||
if (stepId === 1) return true;
|
||||
if (stepId === 2) return isStep1Complete;
|
||||
if (stepId === 3) return isStep2Complete;
|
||||
if (stepId === 4) return isStep3Complete;
|
||||
if (stepId === 5) return isStep3Complete;
|
||||
if (stepId === 6) return isStep3Complete;
|
||||
return isStep3Complete;
|
||||
};
|
||||
|
||||
const canGoNext = () => {
|
||||
if (currentStep === 1) {
|
||||
return isStep1Complete;
|
||||
}
|
||||
if (currentStep === 2) {
|
||||
return isStep2Complete;
|
||||
}
|
||||
return true;
|
||||
if (currentStep === 1) return isStep1Complete;
|
||||
if (currentStep === 2) return isStep2Complete;
|
||||
if (currentStep === 3) return isStep3Complete;
|
||||
return isStep3Complete;
|
||||
};
|
||||
|
||||
// 保存当前步骤数据（目前只有 Step 1 需要在切换前落盘）
|
||||
@@ -163,12 +164,14 @@ function VideoCreationContent() {
|
||||
case 1:
|
||||
return <ScriptCreation />;
|
||||
case 2:
|
||||
return <VideoGeneration />;
|
||||
return <AudioMixing />;
|
||||
case 3:
|
||||
return <SubtitleBurning />;
|
||||
return <VideoGeneration />;
|
||||
case 4:
|
||||
return <CoverDesign />;
|
||||
return <SubtitleBurning />;
|
||||
case 5:
|
||||
return <CoverDesign />;
|
||||
case 6:
|
||||
return <VideoComposite />;
|
||||
default:
|
||||
return <ScriptCreation />;
|
||||
|
||||
@@ -11,11 +11,8 @@ export { useSettingsStore } from './settingsStore';
|
||||
export { useProjectStore, useProjectStats, useShots, useShotActions, createNewProject, saveMetaToLocalFile } from './projectStore';
|
||||
export { useUIStore, toast } from './uiStore';
|
||||
export { useProgressStore } from './progressStore';
|
||||
export { useTaskStore } from './taskStore';
|
||||
export { useVoiceStore, usePresetVoices, useProjectAudios, useAudioMapping } from './voiceStore';
|
||||
|
||||
// 类型导出
|
||||
export type { ProgressPhase, ProgressState } from './progressStore';
|
||||
export type { LocalTask, TaskType, TaskStatus } from './taskStore';
|
||||
|
||||
// 类型导出
|
||||
export type { UserInfo, AuthState } from './authStore';
|
||||
|
||||
@@ -30,7 +30,7 @@ export interface ProjectState {
|
||||
exportedAt?: number;
|
||||
createdAt?: number;
|
||||
updatedAt?: number;
|
||||
currentStep: number; // 当前视频创作步骤 (1-5)
|
||||
currentStep: number; // 当前视频创作步骤 (1-6)
|
||||
scriptDuration?: number; // 脚本生成参数：视频时长
|
||||
scriptType?: string; // 脚本生成参数：脚本类型
|
||||
}
|
||||
@@ -270,103 +270,7 @@ export const useProjectStore = create<ProjectStore>()(
|
||||
state.updatedAt = Date.now();
|
||||
}),
|
||||
|
||||
// ===== 后端同步操作 =====
|
||||
|
||||
saveToBackend: async (projectId?: string) => {
|
||||
const state = get();
|
||||
|
||||
try {
|
||||
set(s => {
|
||||
s._isLoading = true;
|
||||
});
|
||||
|
||||
let pid = projectId;
|
||||
if (!pid) {
|
||||
const project = await projectApi.create({
|
||||
title: `项目 ${new Date().toLocaleDateString()}`,
|
||||
topic: state.segments[0]?.voiceover?.slice(0, 50) || '未命名项目',
|
||||
scriptType: '干货型',
|
||||
duration: state.segments.reduce((sum, s) => sum + parseInt(s.duration || '0'), 0),
|
||||
});
|
||||
pid = project.id;
|
||||
}
|
||||
|
||||
await projectApi.saveSegments(pid, toBackendSegments(state.segments));
|
||||
|
||||
set(s => {
|
||||
s._isLoading = false;
|
||||
});
|
||||
return pid;
|
||||
} catch (error) {
|
||||
console.error('[ProjectStore] 保存失败:', error);
|
||||
set(s => {
|
||||
s._isLoading = false;
|
||||
});
|
||||
throw error;
|
||||
}
|
||||
},
|
||||
|
||||
loadFromBackend: async (projectId: string) => {
|
||||
try {
|
||||
set(s => {
|
||||
s._isLoading = true;
|
||||
});
|
||||
|
||||
const project = await projectApi.get(projectId);
|
||||
const segments: ScriptShot[] = (project.segments || []).map(adaptProjectSegment);
|
||||
|
||||
set(state => {
|
||||
state.segments = segments;
|
||||
state.updatedAt = Date.now();
|
||||
});
|
||||
|
||||
set(s => {
|
||||
s._isLoading = false;
|
||||
});
|
||||
} catch (error) {
|
||||
console.error('[ProjectStore] 加载失败:', error);
|
||||
set(s => {
|
||||
s._isLoading = false;
|
||||
});
|
||||
throw error;
|
||||
}
|
||||
},
|
||||
|
||||
createProject: async (title: string, topic?: string) => {
|
||||
try {
|
||||
set(s => {
|
||||
s._isLoading = true;
|
||||
});
|
||||
|
||||
const project = await projectApi.create({
|
||||
title,
|
||||
topic,
|
||||
status: 'draft',
|
||||
});
|
||||
|
||||
set(state => {
|
||||
state.segments = [];
|
||||
state.selectedHumanId = undefined;
|
||||
state.selectedElementId = undefined;
|
||||
state.selectedVoiceId = undefined;
|
||||
state.finalVideoPath = undefined;
|
||||
state.coverPath = undefined;
|
||||
state.createdAt = Date.now();
|
||||
state.updatedAt = Date.now();
|
||||
});
|
||||
|
||||
set(s => {
|
||||
s._isLoading = false;
|
||||
});
|
||||
return project.id;
|
||||
} catch (error) {
|
||||
console.error('[ProjectStore] 创建失败:', error);
|
||||
set(s => {
|
||||
s._isLoading = false;
|
||||
});
|
||||
throw error;
|
||||
}
|
||||
},
|
||||
}),
|
||||
{
|
||||
name: 'ai-video-project-config',
|
||||
|
||||
@@ -1,25 +1,23 @@
|
||||
/**
|
||||
* 任务状态管理 - Task Store
|
||||
* ========================
|
||||
* 任务状态管理
|
||||
* =============
|
||||
*
|
||||
* 统一管理所有异步任务：
|
||||
* - 纯内存存储，不持久化到 localStorage（状态唯一真相源是后端 Redis）
|
||||
* - 启动时从后端查询真实状态
|
||||
* - 与 progressStore 联动显示进度
|
||||
* 管理异步任务的状态和进度
|
||||
* 注意：这是前端内存状态，真相源在后端 Redis
|
||||
*/
|
||||
|
||||
import { create } from 'zustand';
|
||||
import { immer } from 'zustand/middleware/immer';
|
||||
|
||||
export type TaskType = 'video' | 'image' | 'script' | 'subtitle' | 'copy' | 'avatar_clone';
|
||||
|
||||
export type TaskStatus = 'pending' | 'running' | 'completed' | 'failed';
|
||||
|
||||
export interface LocalTask {
|
||||
export interface Task {
|
||||
id: string;
|
||||
type: TaskType;
|
||||
projectId?: string;
|
||||
status: TaskStatus;
|
||||
progress: number;
|
||||
progress: number; // 0-100
|
||||
message: string;
|
||||
result?: unknown;
|
||||
error?: string;
|
||||
@@ -28,128 +26,87 @@ export interface LocalTask {
|
||||
}
|
||||
|
||||
interface TaskState {
|
||||
tasks: LocalTask[];
|
||||
isRestoring: boolean;
|
||||
tasks: Task[];
|
||||
|
||||
// Actions
|
||||
addTask: (task: Omit<Task, 'createdAt' | 'updatedAt'>) => void;
|
||||
updateTask: (taskId: string, updates: Partial<Pick<Task, 'status' | 'progress' | 'message'>>) => void;
|
||||
completeTask: (taskId: string, result?: unknown) => void;
|
||||
failTask: (taskId: string, error: string) => void;
|
||||
removeTask: (taskId: string) => void;
|
||||
clearTasks: (projectId?: string) => void;
|
||||
|
||||
// Queries
|
||||
getTask: (taskId: string) => Task | undefined;
|
||||
getRunningTasks: () => Task[];
|
||||
getTasksByProject: (projectId: string) => Task[];
|
||||
}
|
||||
|
||||
interface TaskActions {
|
||||
// 添加任务
|
||||
addTask: (task: Omit<LocalTask, 'createdAt' | 'updatedAt'>) => void;
|
||||
export const useTaskStore = create<TaskState>()(
|
||||
immer((set, get) => ({
|
||||
tasks: [],
|
||||
|
||||
// 更新任务状态
|
||||
updateTask: (id: string, updates: Partial<Omit<LocalTask, 'id' | 'type'>>) => void;
|
||||
addTask: (task) =>
|
||||
set((state) => {
|
||||
const now = Date.now();
|
||||
state.tasks.push({
|
||||
...task,
|
||||
createdAt: now,
|
||||
updatedAt: now,
|
||||
});
|
||||
}),
|
||||
|
||||
// 完成任务
|
||||
completeTask: (id: string, result: unknown) => void;
|
||||
updateTask: (taskId, updates) =>
|
||||
set((state) => {
|
||||
const task = state.tasks.find((t) => t.id === taskId);
|
||||
if (task) {
|
||||
Object.assign(task, updates, { updatedAt: Date.now() });
|
||||
}
|
||||
}),
|
||||
|
||||
// 标记失败
|
||||
failTask: (id: string, error: string) => void;
|
||||
completeTask: (taskId, result) =>
|
||||
set((state) => {
|
||||
const task = state.tasks.find((t) => t.id === taskId);
|
||||
if (task) {
|
||||
task.status = 'completed';
|
||||
task.progress = 100;
|
||||
task.message = '任务完成';
|
||||
task.result = result;
|
||||
task.updatedAt = Date.now();
|
||||
}
|
||||
}),
|
||||
|
||||
// 删除任务
|
||||
removeTask: (id: string) => void;
|
||||
failTask: (taskId, error) =>
|
||||
set((state) => {
|
||||
const task = state.tasks.find((t) => t.id === taskId);
|
||||
if (task) {
|
||||
task.status = 'failed';
|
||||
task.error = error;
|
||||
task.message = '任务失败';
|
||||
task.updatedAt = Date.now();
|
||||
}
|
||||
}),
|
||||
|
||||
// 获取进行中的任务
|
||||
getRunningTasks: () => LocalTask[];
|
||||
removeTask: (taskId) =>
|
||||
set((state) => {
|
||||
state.tasks = state.tasks.filter((t) => t.id !== taskId);
|
||||
}),
|
||||
|
||||
// 获取项目的任务
|
||||
getProjectTasks: (projectId: string) => LocalTask[];
|
||||
clearTasks: (projectId) =>
|
||||
set((state) => {
|
||||
if (projectId) {
|
||||
state.tasks = state.tasks.filter((t) => t.projectId !== projectId);
|
||||
} else {
|
||||
state.tasks = [];
|
||||
}
|
||||
}),
|
||||
|
||||
// 设置恢复状态
|
||||
setRestoring: (value: boolean) => void;
|
||||
getTask: (taskId) => get().tasks.find((t) => t.id === taskId),
|
||||
|
||||
// 清理已完成/失败的任务（保留最近50条）
|
||||
cleanup: () => void;
|
||||
}
|
||||
getRunningTasks: () =>
|
||||
get().tasks.filter((t) => t.status === 'pending' || t.status === 'running'),
|
||||
|
||||
const MAX_STORED_TASKS = 100;
|
||||
|
||||
export const useTaskStore = create<TaskState & TaskActions>((set, get) => ({
|
||||
tasks: [],
|
||||
isRestoring: false,
|
||||
|
||||
addTask: (task) => {
|
||||
const now = Date.now();
|
||||
const newTask: LocalTask = {
|
||||
...task,
|
||||
createdAt: now,
|
||||
updatedAt: now,
|
||||
};
|
||||
set((state) => ({
|
||||
tasks: [newTask, ...state.tasks].slice(0, MAX_STORED_TASKS),
|
||||
}));
|
||||
},
|
||||
|
||||
updateTask: (id, updates) => {
|
||||
set((state) => ({
|
||||
tasks: state.tasks.map((t) =>
|
||||
t.id === id
|
||||
? { ...t, ...updates, updatedAt: Date.now() }
|
||||
: t
|
||||
),
|
||||
}));
|
||||
},
|
||||
|
||||
completeTask: (id, result) => {
|
||||
set((state) => ({
|
||||
tasks: state.tasks.map((t) =>
|
||||
t.id === id
|
||||
? {
|
||||
...t,
|
||||
status: 'completed',
|
||||
progress: 100,
|
||||
result,
|
||||
updatedAt: Date.now(),
|
||||
}
|
||||
: t
|
||||
),
|
||||
}));
|
||||
},
|
||||
|
||||
failTask: (id, error) => {
|
||||
set((state) => ({
|
||||
tasks: state.tasks.map((t) =>
|
||||
t.id === id
|
||||
? {
|
||||
...t,
|
||||
status: 'failed',
|
||||
error,
|
||||
updatedAt: Date.now(),
|
||||
}
|
||||
: t
|
||||
),
|
||||
}));
|
||||
},
|
||||
|
||||
removeTask: (id) => {
|
||||
set((state) => ({
|
||||
tasks: state.tasks.filter((t) => t.id !== id),
|
||||
}));
|
||||
},
|
||||
|
||||
getRunningTasks: () => {
|
||||
return get().tasks.filter(
|
||||
(t) => t.status === 'running' || t.status === 'pending'
|
||||
);
|
||||
},
|
||||
|
||||
getProjectTasks: (projectId) => {
|
||||
return get().tasks.filter((t) => t.projectId === projectId);
|
||||
},
|
||||
|
||||
setRestoring: (value) => {
|
||||
set({ isRestoring: value });
|
||||
},
|
||||
|
||||
cleanup: () => {
|
||||
set((state) => {
|
||||
// 保留进行中的任务，以及最近完成的20条
|
||||
const running = state.tasks.filter(
|
||||
(t) => t.status === 'running' || t.status === 'pending'
|
||||
);
|
||||
const completed = state.tasks
|
||||
.filter((t) => t.status === 'completed' || t.status === 'failed')
|
||||
.slice(0, 20);
|
||||
return { tasks: [...running, ...completed] };
|
||||
});
|
||||
},
|
||||
}));
|
||||
getTasksByProject: (projectId) =>
|
||||
get().tasks.filter((t) => t.projectId === projectId),
|
||||
}))
|
||||
);
|
||||
|
||||
@@ -0,0 +1,219 @@
|
||||
/**
|
||||
* Voice Store - Zustand
|
||||
* ======================
|
||||
*
|
||||
* 管理 TTS 合成、音频文件、本地音色库的全局状态。
|
||||
*/
|
||||
|
||||
import { create } from 'zustand';
|
||||
import { useShallow } from 'zustand/react/shallow';
|
||||
import type { VoiceInfo, AudioMeta } from '../api/modules/voice';
|
||||
import * as voiceApi from '../api/modules/voice';
|
||||
|
||||
interface VoiceState {
|
||||
// 预设音色列表
|
||||
presetVoices: VoiceInfo[];
|
||||
selectedVoiceId: string;
|
||||
|
||||
// 项目音频文件列表
|
||||
projectAudios: AudioMeta[];
|
||||
|
||||
// 音频替换映射（segmentId → audioId）
|
||||
// 记录每个分镜使用了哪个音频文件
|
||||
audioMapping: Record<string, string>; // segmentId → audioId
|
||||
|
||||
// 当前项目 ID
|
||||
currentProjectId: string | null;
|
||||
|
||||
// 加载状态
|
||||
isLoadingVoices: boolean;
|
||||
isLoadingAudios: boolean;
|
||||
}
|
||||
|
||||
interface VoiceActions {
|
||||
// 音色操作
|
||||
loadPresetVoices: () => Promise<void>;
|
||||
setSelectedVoiceId: (id: string) => void;
|
||||
|
||||
// 项目音频操作
|
||||
loadProjectAudios: (projectId: string) => Promise<void>;
|
||||
saveAudio: (args: {
|
||||
projectId: string;
|
||||
audioId: string;
|
||||
audioData: string;
|
||||
name: string;
|
||||
voiceId: string;
|
||||
duration: number;
|
||||
segmentId?: string;
|
||||
}) => Promise<AudioMeta>;
|
||||
deleteAudio: (projectId: string, audioId: string) => Promise<void>;
|
||||
|
||||
// 音频映射操作
|
||||
setAudioMapping: (segmentId: string, audioId: string | undefined) => void;
|
||||
getAudioForSegment: (segmentId: string) => AudioMeta | undefined;
|
||||
|
||||
// 重置
|
||||
reset: () => void;
|
||||
}
|
||||
|
||||
const initialState: VoiceState = {
|
||||
presetVoices: [],
|
||||
selectedVoiceId: '829826751244537879', // 温柔女声（Kling 预设音色）
|
||||
projectAudios: [],
|
||||
audioMapping: {},
|
||||
currentProjectId: null,
|
||||
isLoadingVoices: false,
|
||||
isLoadingAudios: false,
|
||||
};
|
||||
|
||||
export const useVoiceStore = create<VoiceState & VoiceActions>()(
|
||||
(set, get) => ({
|
||||
...initialState,
|
||||
|
||||
// ====================== 音色操作 ======================
|
||||
|
||||
loadPresetVoices: async () => {
|
||||
set({ isLoadingVoices: true });
|
||||
try {
|
||||
const voices = await voiceApi.getVoiceList();
|
||||
set({ presetVoices: voices });
|
||||
} catch (err) {
|
||||
console.error('[VoiceStore] 加载音色列表失败:', err);
|
||||
// 静默失败，使用默认值（Kling 预设音色）
|
||||
set({
|
||||
presetVoices: [
|
||||
{ voiceId: '829826751244537879', name: '温柔女声', description: '温柔细腻', recommended: true, language: 'zh' },
|
||||
{ voiceId: '829824295735410756', name: '钓系女友', description: '甜美撒娇', recommended: false, language: 'zh' },
|
||||
{ voiceId: '829826792415842333', name: '播报男声', description: '沉稳播报', recommended: false, language: 'zh' },
|
||||
{ voiceId: '829826834144964676', name: '盐系少年', description: '清新少年', recommended: false, language: 'zh' },
|
||||
{ voiceId: '829826884271091753', name: '撒娇女友', description: '可爱撒娇', recommended: false, language: 'zh' },
|
||||
],
|
||||
});
|
||||
} finally {
|
||||
set({ isLoadingVoices: false });
|
||||
}
|
||||
},
|
||||
|
||||
setSelectedVoiceId: (id) => set({ selectedVoiceId: id }),
|
||||
|
||||
// ====================== 项目音频操作 ======================
|
||||
|
||||
loadProjectAudios: async (projectId) => {
|
||||
set({ isLoadingAudios: true, currentProjectId: projectId });
|
||||
try {
|
||||
const audios = await voiceApi.listProjectAudios(projectId);
|
||||
set({ projectAudios: audios });
|
||||
|
||||
// 从现有音频恢复 audioMapping
|
||||
const mapping: Record<string, string> = {};
|
||||
for (const audio of audios) {
|
||||
if (audio.segmentId) {
|
||||
mapping[audio.segmentId] = audio.id;
|
||||
}
|
||||
}
|
||||
set({ audioMapping: mapping });
|
||||
} catch (err) {
|
||||
console.error('[VoiceStore] 加载项目音频失败:', err);
|
||||
set({ projectAudios: [] });
|
||||
} finally {
|
||||
set({ isLoadingAudios: false });
|
||||
}
|
||||
},
|
||||
|
||||
saveAudio: async (args) => {
|
||||
const meta = await voiceApi.saveAudio(args);
|
||||
const state = get();
|
||||
|
||||
// 更新 projectAudios 列表
|
||||
const existing = state.projectAudios.findIndex(a => a.id === meta.id);
|
||||
const updatedAudios = existing >= 0
|
||||
? state.projectAudios.map((a, i) => i === existing ? meta : a)
|
||||
: [...state.projectAudios, meta];
|
||||
set({ projectAudios: updatedAudios });
|
||||
|
||||
// 如果有 segmentId，更新 audioMapping
|
||||
if (args.segmentId) {
|
||||
set(state => ({
|
||||
audioMapping: { ...state.audioMapping, [args.segmentId!]: meta.id },
|
||||
}));
|
||||
}
|
||||
|
||||
return meta;
|
||||
},
|
||||
|
||||
deleteAudio: async (projectId, audioId) => {
|
||||
await voiceApi.deleteAudio(projectId, audioId);
|
||||
|
||||
const state = get();
|
||||
const updatedAudios = state.projectAudios.filter(a => a.id !== audioId);
|
||||
|
||||
// 从 audioMapping 中移除
|
||||
const updatedMapping = { ...state.audioMapping };
|
||||
for (const [segId, audioId2] of Object.entries(updatedMapping)) {
|
||||
if (audioId2 === audioId) {
|
||||
delete updatedMapping[segId];
|
||||
}
|
||||
}
|
||||
|
||||
set({ projectAudios: updatedAudios, audioMapping: updatedMapping });
|
||||
},
|
||||
|
||||
// ====================== 音频映射操作 ======================
|
||||
|
||||
setAudioMapping: (segmentId, audioId) => {
|
||||
set(state => {
|
||||
if (audioId === undefined) {
|
||||
const updated = { ...state.audioMapping };
|
||||
delete updated[segmentId];
|
||||
return { audioMapping: updated };
|
||||
}
|
||||
return { audioMapping: { ...state.audioMapping, [segmentId]: audioId } };
|
||||
});
|
||||
},
|
||||
|
||||
getAudioForSegment: (segmentId) => {
|
||||
const state = get();
|
||||
const audioId = state.audioMapping[segmentId];
|
||||
if (!audioId) return undefined;
|
||||
return state.projectAudios.find(a => a.id === audioId);
|
||||
},
|
||||
|
||||
// ====================== 重置 ======================
|
||||
|
||||
reset: () => set(initialState),
|
||||
})
|
||||
);
|
||||
|
||||
// 便捷 Hooks - 使用 useShallow 进行浅比较，避免不必要的重渲染
|
||||
export function usePresetVoices() {
|
||||
return useVoiceStore(
|
||||
useShallow(state => ({
|
||||
voices: state.presetVoices,
|
||||
selectedVoiceId: state.selectedVoiceId,
|
||||
isLoading: state.isLoadingVoices,
|
||||
load: state.loadPresetVoices,
|
||||
setSelected: state.setSelectedVoiceId,
|
||||
}))
|
||||
);
|
||||
}
|
||||
|
||||
export function useProjectAudios() {
|
||||
return useVoiceStore(
|
||||
useShallow(state => ({
|
||||
audios: state.projectAudios,
|
||||
isLoading: state.isLoadingAudios,
|
||||
load: state.loadProjectAudios,
|
||||
deleteAudio: state.deleteAudio,
|
||||
}))
|
||||
);
|
||||
}
|
||||
|
||||
export function useAudioMapping() {
|
||||
return useVoiceStore(
|
||||
useShallow(state => ({
|
||||
mapping: state.audioMapping,
|
||||
getAudio: state.getAudioForSegment,
|
||||
setMapping: state.setAudioMapping,
|
||||
}))
|
||||
);
|
||||
}
|
||||
Reference in New Issue
Block a user