# cAIc — Docker Compose stack
# Coordinator: cAIc + SearXNG + Qdrant + RabbitMQ + llama-server + Ollama
#
# Usage:
#   cp .env.example .env    # edit with your values
#   mkdir -p models secrets searxng
#   # place .gguf model in ./models/
#   docker compose up -d

services:
  # ── cAIc (FastAPI) ──────────────────────────────────────
  caic:
    build: .
    ports:
      - "${CAIC_EXPOSE_PORT:-8080}:8080"
    volumes:
      - caic_data:/app/data
    secrets:
      - rabbitmq_password
    environment:
      - CAIC_AMQP_SECRET_PATH=/run/secrets/rabbitmq_password
    env_file: .env
    depends_on:
      qdrant:        { condition: service_started }
      rabbitmq:      { condition: service_healthy }
      llama-server:  { condition: service_healthy }
    restart: unless-stopped

  # ── SearXNG (web search) ────────────────────────────────
  searxng:
    image: searxng/searxng:latest
    ports:
      - "${SEARXNG_EXPOSE_PORT:-8888}:8080"
    volumes:
      - ./searxng/settings.yml:/etc/searxng/settings.yml:ro
      - searxng_config:/etc/searxng
    environment:
      - SEARXNG_BASE_URL=http://localhost:8888
    restart: unless-stopped

  # ── Qdrant (vector DB) ──────────────────────────────────
  qdrant:
    image: qdrant/qdrant:latest
    ports:
      - "${QDRANT_EXPOSE_PORT:-6333}:6333"
    volumes:
      - qdrant_storage:/qdrant/storage
    restart: unless-stopped

  # ── RabbitMQ (AMQP broker) ──────────────────────────────
  rabbitmq:
    image: rabbitmq:4-management
    ports:
      - "${RABBITMQ_EXPOSE_PORT:-5672}:5672"
    volumes:
      - rabbitmq_data:/var/lib/rabbitmq
    environment:
      RABBITMQ_DEFAULT_USER: caic
      RABBITMQ_DEFAULT_PASS_FILE: /run/secrets/rabbitmq_password
      RABBITMQ_DEFAULT_VHOST: /
    secrets:
      - rabbitmq_password
    healthcheck:
      test: ["CMD", "rabbitmq-diagnostics", "check_port_connectivity"]
      interval: 15s
      timeout: 5s
      retries: 3
    restart: unless-stopped

  # ── llama-server (LLM inference) ────────────────────────
  llama-server:
    image: ghcr.io/ggml-org/llama.cpp:server
    ports:
      - "${LLAMA_EXPOSE_PORT:-8081}:8081"
    volumes:
      - ./models:/models:ro
    command: >
      --model /models/${LLAMA_MODEL:?Set LLAMA_MODEL in .env}
      --host 0.0.0.0 --port 8081
      --ctx-size ${LLAMA_CTX_SIZE:-4096}
      --n-gpu-layers ${LLAMA_N_GPU_LAYERS:-0}
      --embeddings
      --logprobs
      ${LLAMA_RPC_ENDPOINTS:+--rpc ${LLAMA_RPC_ENDPOINTS}}
    healthcheck:
      test: ["CMD", "curl", "-fs", "http://localhost:8081/health"]
      interval: 30s
      timeout: 10s
      retries: 5
      start_period: 60s
    restart: unless-stopped

  # ── Ollama (embeddings) ─────────────────────────────────
  ollama:
    image: ollama/ollama:latest
    ports:
      - "${OLLAMA_EXPOSE_PORT:-11434}:11434"
    volumes:
      - ollama_models:/root/.ollama
    healthcheck:
      test: ["CMD", "ollama", "list"]
      interval: 30s
      timeout: 10s
      retries: 3
    restart: unless-stopped

volumes:
  caic_data:
  searxng_config:
  qdrant_storage:
  rabbitmq_data:
  ollama_models:

secrets:
  rabbitmq_password:
    file: ./secrets/rabbitmq_password.txt