# Mirrors the upstream OpenRAG compose
# (https://github.com/langflow-ai/openrag/blob/main/docker-compose.yml)
# with one local override: the backend's :8000 port is published so our
# Next.js app can call it from the host.

services:
  opensearch:
    image: langflowai/openrag-opensearch:${OPENRAG_VERSION:-latest}
    container_name: os
    # Bypass the upstream wrapper that runs the security demo installer.
    # We disable the security plugin entirely (option A in setup notes), so
    # plain http on 9200 is fine for local dev and boot drops to ~30s.
    entrypoint: ["/usr/share/opensearch/opensearch-docker-entrypoint.sh"]
    command: ["opensearch"]
    environment:
      - discovery.type=single-node
      # Heap sizing: the openrag image bundles ~25 plugins (ML, AD, neural
      # search, geospatial, SQL...). 512m caused GC thrash + cluster-event
      # timeouts ("Failed to create index .plugins-ml-config within 30s"),
      # 1g caused OOM-kill (exit 137) on tight Docker Desktop allocations.
      # 1.5g + 256m direct is the sweet spot — needs Docker Desktop set to
      # at least 4 GB (Settings → Resources).
      - OPENSEARCH_JAVA_OPTS=-Xms1536m -Xmx1536m -XX:MaxDirectMemorySize=268435456
      - DISABLE_SECURITY_PLUGIN=true
      - DISABLE_INSTALL_DEMO_CONFIG=true
    ports:
      - "9200:9200"
      - "9600:9600"
    volumes:
      - opensearch-data:/usr/share/opensearch/data
    stop_grace_period: 2m
    healthcheck:
      test:
        [
          "CMD-SHELL",
          "curl -s 'http://localhost:9200/_cluster/health' | grep -qE '\"status\":\"(green|yellow)\"'",
        ]
      interval: 5s
      timeout: 5s
      retries: 30
      start_period: 30s
    restart: unless-stopped

  docling:
    # Backend's /router/upload_ingest path POSTs uploaded files to docling-serve
    # to extract text/tables/images. Without it, ingest fails with
    # "Cannot connect to docling-serve at http://host.docker.internal:5001".
    image: quay.io/docling-project/docling-serve-cpu:latest
    container_name: docling
    ports:
      - "5001:5001"
    environment:
      - DOCLING_SERVE_HOST=0.0.0.0
      - DOCLING_SERVE_PORT=5001
      - DOCLING_SERVE_MAX_SYNC_WAIT=300
    restart: unless-stopped

  langflow:
    image: langflowai/openrag-langflow:${OPENRAG_VERSION:-latest}
    container_name: langflow
    depends_on:
      opensearch:
        condition: service_healthy
      docling:
        condition: service_started
    ports:
      - "${LANGFLOW_PORT:-7860}:7860"
    volumes:
      - ./flows:/app/flows
      - ./langflow-data:/app/langflow-data
    extra_hosts:
      - "host.docker.internal:host-gateway"
    environment:
      - OPENAI_API_KEY=${OPENAI_API_KEY}
      - ANTHROPIC_API_KEY=${ANTHROPIC_API_KEY}
      - WATSONX_APIKEY=${WATSONX_API_KEY}
      - WATSONX_URL=${WATSONX_URL:-${WATSONX_ENDPOINT}}
      - WATSONX_PROJECT_ID=${WATSONX_PROJECT_ID}
      - OLLAMA_BASE_URL=${OLLAMA_ENDPOINT}
      - LANGFLOW_CONFIG_DIR=/app/langflow-data
      - LANGFLOW_DATABASE_URL=${LANGFLOW_DATABASE_URL:-sqlite:////app/langflow-data/langflow.db}
      - LANGFLOW_SECRET_KEY=${LANGFLOW_SECRET_KEY}
      - LANGFLOW_AUTO_LOGIN=${LANGFLOW_AUTO_LOGIN:-false}
      - LANGFLOW_SUPERUSER=${LANGFLOW_SUPERUSER:-admin}
      - LANGFLOW_SUPERUSER_PASSWORD=${LANGFLOW_SUPERUSER_PASSWORD}
      - LANGFLOW_NEW_USER_IS_ACTIVE=${LANGFLOW_NEW_USER_IS_ACTIVE:-true}
      - LANGFLOW_ENABLE_SUPERUSER_CLI=${LANGFLOW_ENABLE_SUPERUSER_CLI:-true}
      - LANGFLOW_WORKERS=${LANGFLOW_WORKERS:-1}
      - LANGFLOW_LOG_LEVEL=INFO
      - JWT=None
      - OWNER=None
      - OWNER_NAME=None
      - OWNER_EMAIL=None
      - CONNECTOR_TYPE=system
      - CONNECTOR_TYPE_URL=url
      - DOCUMENT_ID=None
      - SOURCE_URL=None
      - ALLOWED_USERS=[]
      - ALLOWED_GROUPS=[]
      - OPENRAG-QUERY-FILTER="{}"
      - OPENSEARCH_PASSWORD=${OPENSEARCH_PASSWORD}
      - OPENSEARCH_HOST=opensearch
      - OPENSEARCH_PORT=9200
      - OPENSEARCH_URL=http://opensearch:9200
      - OPENSEARCH_INDEX_NAME=${OPENSEARCH_INDEX_NAME:-documents}
      - DOCLING_SERVE_URL=${DOCLING_SERVE_URL:-http://host.docker.internal:5001}
      - FILENAME=None
      - MIMETYPE=None
      - FILESIZE=0
      - SELECTED_EMBEDDING_MODEL=${SELECTED_EMBEDDING_MODEL:-text-embedding-3-small}
      - LANGFLOW_VARIABLES_TO_GET_FROM_ENVIRONMENT=JWT,OPENRAG-QUERY-FILTER,OPENSEARCH_PASSWORD,OPENSEARCH_URL,DOCLING_SERVE_URL,OWNER,OWNER_NAME,OWNER_EMAIL,CONNECTOR_TYPE,DOCUMENT_ID,SOURCE_URL,ALLOWED_USERS,ALLOWED_GROUPS,FILENAME,MIMETYPE,FILESIZE,SELECTED_EMBEDDING_MODEL,OPENAI_API_KEY,ANTHROPIC_API_KEY,WATSONX_APIKEY,WATSONX_URL,WATSONX_PROJECT_ID,OLLAMA_BASE_URL,OPENSEARCH_INDEX_NAME
      - HIDE_GETTING_STARTED_PROGRESS=true

  openrag-backend:
    # Custom build patches the base image to talk to OpenSearch over plain
    # http with no auth (the upstream image hardcodes use_ssl=True +
    # http_auth=admin). See docker/openrag-backend.Dockerfile.
    build:
      context: .
      dockerfile: docker/openrag-backend.Dockerfile
      args:
        OPENRAG_VERSION: ${OPENRAG_VERSION:-latest}
    image: notebooklm/openrag-backend:local
    container_name: openrag-backend
    depends_on:
      opensearch:
        condition: service_healthy
      langflow:
        condition: service_started
      docling:
        condition: service_started
    # NOTE: upstream compose does NOT publish 8000 to the host (frontend
    # reaches it on the docker network). We publish it so our Next.js app,
    # running on the host, can call /v1/chat/completions and the ingest API.
    ports:
      - "8000:8000"
    environment:
      - OPENSEARCH_HOST=opensearch
      - OPENSEARCH_PORT=9200
      - OPENSEARCH_USERNAME=${OPENSEARCH_USERNAME:-admin}
      - OPENSEARCH_PASSWORD=${OPENSEARCH_PASSWORD}
      - OPENSEARCH_INDEX_NAME=${OPENSEARCH_INDEX_NAME:-documents}
      - LANGFLOW_URL=http://langflow:7860
      - LANGFLOW_PUBLIC_URL=${LANGFLOW_PUBLIC_URL:-http://localhost:7860}
      - LANGFLOW_AUTO_LOGIN=${LANGFLOW_AUTO_LOGIN:-false}
      - LANGFLOW_SUPERUSER=${LANGFLOW_SUPERUSER:-admin}
      - LANGFLOW_SUPERUSER_PASSWORD=${LANGFLOW_SUPERUSER_PASSWORD}
      - LANGFLOW_KEY=${LANGFLOW_KEY}
      - LANGFLOW_CHAT_FLOW_ID=${LANGFLOW_CHAT_FLOW_ID}
      - LANGFLOW_INGEST_FLOW_ID=${LANGFLOW_INGEST_FLOW_ID}
      - LANGFLOW_URL_INGEST_FLOW_ID=${LANGFLOW_URL_INGEST_FLOW_ID}
      - DISABLE_INGEST_WITH_LANGFLOW=${DISABLE_INGEST_WITH_LANGFLOW:-true}
      - INGEST_SAMPLE_DATA=${INGEST_SAMPLE_DATA:-false}
      - LANGFLOW_OPENSEARCH_HOST=opensearch
      - LANGFLOW_OPENSEARCH_PORT=9200
      - OPENAI_API_KEY=${OPENAI_API_KEY}
      - ANTHROPIC_API_KEY=${ANTHROPIC_API_KEY}
      - WATSONX_API_KEY=${WATSONX_API_KEY}
      - WATSONX_ENDPOINT=${WATSONX_ENDPOINT}
      - WATSONX_PROJECT_ID=${WATSONX_PROJECT_ID}
      - OLLAMA_ENDPOINT=${OLLAMA_ENDPOINT}
      - SESSION_SECRET=${SESSION_SECRET}
      - JWT_SIGNING_KEY=${JWT_SIGNING_KEY}
      - OPENRAG_ENCRYPTION_KEY=${OPENRAG_ENCRYPTION_KEY}
      - LOG_LEVEL=${LOG_LEVEL:-INFO}
      - ENVIRONMENT=${ENVIRONMENT:-production}
      - LANGFLOW_TIMEOUT=${LANGFLOW_TIMEOUT:-2400}
      - INGESTION_TIMEOUT=${INGESTION_TIMEOUT:-3600}
      - UPLOAD_BATCH_SIZE=${UPLOAD_BATCH_SIZE:-25}
      - MAX_WORKERS=${MAX_WORKERS:-4}
      - DOCLING_SERVE_URL=${DOCLING_SERVE_URL:-http://docling:5001}
    volumes:
      - ./openrag-documents:/app/openrag-documents
      - ./keys:/app/keys
      - ./flows:/app/flows
      - ./flows/backup:/app/flows/backup
      - ./config:/app/config
      - ./openrag-data:/app/data
    extra_hosts:
      - "host.docker.internal:host-gateway"

  openrag-frontend:
    image: langflowai/openrag-frontend:${OPENRAG_VERSION:-latest}
    container_name: openrag-frontend
    depends_on:
      - openrag-backend
    environment:
      - OPENRAG_BACKEND_HOST=openrag-backend
    ports:
      - "${FRONTEND_PORT:-3000}:3000"

volumes:
  opensearch-data: