services:
  cleverdatasets:
    build:
      context: .
      dockerfile: Dockerfile
    image: cleverdatasets:latest
    container_name: cleverdatasets

    # Mount volumes for data persistence
    volumes:
      # Data directory for downloads and conversions
      - ./dataset_processing:/data/dataset_processing
      # HuggingFace cache for faster subsequent runs
      - huggingface_cache:/data/.cache/huggingface
      # Optional: mount config for custom dataset configurations
      - ./scripts/dataset_config.json:/app/scripts/dataset_config.json:ro

    # Environment variables
    environment:
      # HuggingFace authentication token (required for uploads)
      - HF_TOKEN=${HF_TOKEN:-}
      # Optional: customize base directory
      - BASE_DIR=/data/dataset_processing
      # Python configuration
      - PYTHONUNBUFFERED=1

    # Resource limits (adjust based on workload)
    deploy:
      resources:
        limits:
          memory: 16G
          cpus: '4'
        reservations:
          memory: 4G
          cpus: '2'

    # Restart policy
    restart: on-failure

    # Network mode
    network_mode: bridge

volumes:
  # Named volume for HuggingFace cache persistence
  huggingface_cache:
    driver: local