version: "3.8"

services:
  htknow:
    build:
      context: .
      dockerfile: Dockerfile
    container_name: htknow
    ports:
      - "8080:3000"
    volumes:
      # Mount data directory for persistent storage
      - ./data:/app/data
    environment:
      - RUST_LOG=info
      # 服务器配置
      - HTKNOW_SERVER_HOST=0.0.0.0
      - HTKNOW_SERVER_PORT=3000
      - HTKNOW_SERVER_UPLOAD_LIMIT_MB=500
      - HTKNOW_SERVER_PROCESS_INTERVAL_SECS=10
      - HTKNOW_BUILD_KNOWLEDGE_GRAPH=false
      - "HTKNOW_LANCEDB_COMPACT_CRON=0 0 3 * * *"
      # 外部服务地址
      - HTKNOW_MINERU_URL=http://192.168.0.46:10001/file_parse
      - HTKNOW_REQUEST_TIMEOUT_SECS=600
      - HTKNOW_MINERU_MAX_PAGES=0
      - HTKNOW_OFFICE_CONVERT_URL=http://192.168.0.46:8003/convert
      - HTKNOW_EMBEDDING_URL=http://192.168.0.46:9700/v1/embeddings
      - HTKNOW_RERANK_URL=http://192.168.0.46:9600/v1/rerank
      # AI 模型配置
      - HTKNOW_EMBEDDING_MODEL=bge-m3
      - HTKNOW_EMBEDDING_DIM=1024
      - HTKNOW_EMBEDDING_BATCH_SIZE=8
      - HTKNOW_RERANK_MODEL=bge-rerank
      - HTKNOW_RERANK_THRESHOLD=0.1
      # 数据库配置
      - DATABASE_URL=sqlite://data/app.sqlite
      - HTKNOW_DB_MAX_CONNECTIONS=10
      - HTKNOW_DB_BUSY_TIMEOUT_MS=5000
      # 存储路径配置
      - HTKNOW_DATA_DIR=data
      - HTKNOW_LANCEDB_PATH=data/lancedb_data
      - HTKNOW_TEMP_PATH=data/temp
      - HTKNOW_IMAGES_PATH=data/images
      # 搜索配置
      - HTKNOW_SEARCH_LIMIT=10
      - HTKNOW_TANTIVY_INDEX_PATH=data/tantivy_index
      - HTKNOW_TANTIVY_MEMORY_MB=50
      # LLM 配置（可选，用于知识图谱构建）
      # - LLM_API_URL=https://api.openai.com/v1/chat/completions
      # - LLM_API_KEY=your-api-key
      # - LLM_MODEL=gpt-3.5-turbo
    restart: unless-stopped
    # Uncomment if you need to connect to other services
    # networks:
    #   - htknow-network

# Uncomment if you need a network
# networks:
#   htknow-network:
#     driver: bridge