docker-compose.yml

services:
  elasticsearch:
    image: docker.elastic.co/elasticsearch/elasticsearch:8.18.2
    container_name: es8
    ports:
      - "9200:9200"
      - "9300:9300"
    environment:
      # Run as a single-node cluster (suitable for local development)
      - discovery.type=single-node

      # Disable security for local development (no authentication required)
      - xpack.security.enabled=false
      - xpack.security.http.ssl.enabled=false
      # Limit JVM memory usage to avoid excessive RAM consumption
      - ES_JAVA_OPTS=-Xms512m -Xmx512m
      # Disable disk watermark checks (useful on dev machines with limited disk space)
      - cluster.routing.allocation.disk.threshold_enabled=false
    # Persist Elasticsearch data across container restarts
    volumes:
      - esdata:/usr/share/elasticsearch/data
    networks:
      - backend
  
  redis:
    image: redis:7-alpine
    container_name: redis

    # Only available inside the Docker network
    networks:
      - backend
  ##########################################################################
  # Helper container responsible for managing the XML corpus repository.
  #
  # Responsibilities:
  # - Clone the formulae-corpora repository into a persistent Docker volume
  # - Update the repository when it already exists
  # - Optionally rebuild Elasticsearch indices from the XML data
  #
  # This container is normally executed manually:
  #
  #   docker compose run --rm formulae_corpora
  #
  # It is therefore placed in the "init" profile.
  formulae_corpora:
    build:
      context: .
      dockerfile: docker/Dockerfile.corpora
     # Only started when explicitly requested
    profiles: ["init"]
    environment:
      - FORMULAE_CORPORA_REPO_URL=${FORMULAE_CORPORA_REPO_URL}
      - GITHUB_TOKEN=${GITHUB_TOKEN}
      # Optional branch/tag/commit to checkout
      - FORMULAE_CORPORA_REF=${FORMULAE_CORPORA_REF:-}
      - ELASTICSEARCH_URL=${ELASTICSEARCH_URL}
      - ELASTICSEARCH_VERSION=8.18.2
      # If true, rebuild Elasticsearch indices after updating the corpus
      - REBUILD_ELASTICSEARCH=${REBUILD_ELASTICSEARCH:-false}
    volumes:
      - formulae_corpora_data:/data
      - ./docker/clone_corpora.sh:/clone_corpora.sh:ro
    entrypoint: ["/bin/sh", "/clone_corpora.sh"]

    depends_on:
      - elasticsearch
      - redis
    networks:
      - backend

  nemo:
    build: .
    env_file: ".env"
    environment:
      - ELASTICSEARCH_URL=http://elasticsearch:9200
      - CORPUS_FOLDERS=/data
      - PYTHONUNBUFFERED=1
      - REDIS_URL=redis://redis:6379/0
    volumes:
      - formulae_corpora_data:/data:ro
    ports:
      - "5000:5000"
    networks:
      - backend

volumes:
  esdata:
  formulae_corpora_data:

networks:
  backend:
    driver: bridge