CopyWriter/docker-compose.yml at main · JaggerH/CopyWriter · GitHub

1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
66
67
68
69
70
71
72
73
74
75
76
77
78
79
80
81
82
83
84
85
86
87
88
89
90
91
92
93
94
95
96
97
98
99
100
101
102
103
104
105
106
107
108
109
110
111
112
113
114
115
116
117
118
119
120
121
122
123
124
125
126
127
128
volumes:
  # 共享媒体文件存储
  media-pipeline:
    driver: local
    driver_opts:
      type: none
      o: bind
      device: ./shared/media

  # Redis数据持久化
  redis-data:
    driver: local

  # ASR模型缓存
  asr-models:
    driver: local

  # ASR向量数据库（RAG热词系统）
  asr-vector-db:
    driver: local

  # ModelScope 缓存（避免重复下载模型）
  modelscope-cache:
    driver: local

services:
  # 1. 视频解析下载服务 (基于Douyin_TikTok_Download_API)
  video-service:
    build:
      context: ./services/Douyin_TikTok_Download_API
      dockerfile: Dockerfile
    container_name: copywriter-video-service
    # 内部服务，仅通过 orchestrator 访问，无需暴露端口
    volumes:
      - media-pipeline:/app/downloads
      - ./config.yml:/app/config.yaml  # 自定义配置文件
    environment:
      TZ: Asia/Shanghai  # 设置时区为亚洲/上海
      PUID: 1026  # 设置容器内部的用户 ID
      PGID: 100  # 设置容器内部的用户组 ID
    networks:
      - app-network
    restart: always

  # 2. ASR语音识别服务 (NVIDIA NGC - CUDA 12.8 最高性能)
  asr-service:
    build: ./services/asr-service
    container_name: copywriter-asr-service
    ports:
      - "8082:8000"
    volumes:
      - media-pipeline:/app/media
      - ./services/asr-service/models:/app/models  # 挂载本地模型目录（开发用）
      - ./services/asr-service/user_models:/app/user_models  # 热词和用户数据持久化
      - asr-vector-db:/app/vector_db  # RAG 向量数据库持久化
      - modelscope-cache:/root/.cache/modelscope  # ModelScope 缓存持久化（避免重复下载）
    environment:
      - MODEL_CACHE_DIR=/app/models
      - PYTHONPATH=/app
      - CUDA_VISIBLE_DEVICES=0  # 使用第一个GPU
      - ASR_MODEL=funasr-nano-2512  # Fun-ASR-Nano-2512 模型
      - RAG_ENABLED=true  # 启用 RAG 热词检索
      - QDRANT_PATH=/app/vector_db  # RAG 向量库路径
      - DEFAULT_SCENE=${DEFAULT_SCENE:-tech}  # 默认场景配置
      - USER_SCENE_MERGE=${USER_SCENE_MERGE:-}  # user场景要合并的场景列表
    networks:
      - app-network
    restart: always
    runtime: nvidia  # 启用NVIDIA GPU支持

  # 3. 任务编排服务 (API Gateway + 集成FFmpeg)
  orchestrator-service:
    build: ./services/orchestrator
    container_name: copywriter-orchestrator
    ports:
      - "8081:8000"
    volumes:
      - media-pipeline:/app/media
    environment:
      - VIDEO_SERVICE_URL=http://video-service:80
      - ASR_SERVICE_URL=http://asr-service:8000
      - REDIS_URL=redis://redis:6379/0
      - MEDIA_PATH=/app/media
      - TELEGRAM_BOT_SERVICE_URL=http://telegram-bot:8000  # Telegram Bot 服务回调 URL
      - TASK_AUTO_CLEANUP=${TASK_AUTO_CLEANUP:-true}
      - TASK_TTL_SECONDS=${TASK_TTL_SECONDS:-3600}
    depends_on:
      - redis
      - video-service
      - asr-service
    networks:
      - app-network
    restart: always

  # Redis任务队列和缓存
  redis:
    image: redis:alpine
    container_name: copywriter-redis
    # 内部服务，仅通过 Docker 网络访问，无需暴露端口
    volumes:
      - redis-data:/data
    networks:
      - app-network
    restart: always
    command: redis-server --appendonly yes --notify-keyspace-events Ex

  # 4. Telegram Bot 服务 (混合: Bot + 回调 API)
  telegram-bot:
    build: ./services/telegram-bot
    container_name: copywriter-telegram-bot
    # 回调 API 仅供 orchestrator 内部调用 (http://telegram-bot:8000)
    volumes:
      - media-pipeline:/app/media  # 访问共享媒体文件
    environment:
      - TELEGRAM_BOT_TOKEN=${TELEGRAM_BOT_TOKEN}
      - ORCHESTRATOR_URL=http://orchestrator-service:8000
      - ORCHESTRATOR_TIMEOUT=30
    depends_on:
      - orchestrator-service
      - redis
    networks:
      - app-network
    restart: always

networks:
  app-network:
    driver: bridge
    name: copywriter-network