docker-compose.yml

version: '3.8'

services:
  mongodb:
    image: mongo:latest
    container_name: ditector_mongo
    profiles: [db]
    ports:
      - "27017:27017"
    volumes:
      - ./mongo_data_secure:/data/db

  neo4j:
    image: neo4j:latest
    container_name: ditector_neo4j
    profiles: [db]
    ports:
      - "7474:7474"
      - "7687:7687"
    environment:
      - NEO4J_AUTH=none
    volumes:
      - ./neo4j_data:/data

  crawler:
    image: golang:1.22
    container_name: ditector_crawler
    restart: unless-stopped
    volumes:
      - .:/app
    working_dir: /app
    network_mode: host
    environment:
      - MONGO_URI=${MONGO_URI}
      - NEO4J_URI=${NEO4J_URI}
      - PAGE_CONCURRENCY=${PAGE_CONCURRENCY:-0}
      - ROLE=${ROLE:-primary}
    # Removed manual log redirection to prevent IO deadlocks
    command: sh -c "go build -buildvcs=false -o /tmp/ditector . && /tmp/ditector crawl --workers ${WORKERS:-3} --shard ${SHARD:-0} --shards ${SHARDS:-1} --accounts ${ACCOUNTS:-accounts.json} --config config.yaml"

# No named Docker volumes — all data mounted to host paths so docker system
# prune -a (with or without --volumes) cannot destroy the dataset.