docker-compose.yml

services:
  db:
    container_name: scraper-db
    image: chromadb/chroma:0.4.24
    volumes:
      - ./chromadb:/chroma/chroma
    environment:
      IS_PERSISTENT: True
      ANONYMIZED_TELEMETRY: False
      PERSIST_DIRECTORY: ./chroma/chroma
      CHROMA_SERVER_AUTH_CREDENTIALS: ${CHROMA_SERVER_AUTH_CREDENTIALS:-test-token}
      CHROMA_SERVER_AUTH_CREDENTIALS_PROVIDER: ${CHROMA_SERVER_AUTH_CREDENTIALS_PROVIDER:-chromadb.auth.token.TokenConfigServerAuthCredentialsProvider}
      CHROMA_SERVER_AUTH_PROVIDER: ${CHROMA_SERVER_AUTH_PROVIDER:-chromadb.auth.token.TokenAuthServerProvider}
      CHROMA_SERVER_AUTH_TOKEN_TRANSPORT_HEADER: ${CHROMA_SERVER_AUTH_TOKEN_TRANSPORT_HEADER:-X_CHROMA_TOKEN}
    ports:
      - "8000:8000"
    healthcheck:
      test: [ "CMD", "curl" ,"--fail", "-s", "http://localhost:8000/api/v1/heartbeat", "||", "exit 1" ]
      timeout: 10s
      retries: 10

  scraper:
    container_name: scraper
    build:
      context: ./
    #        target: local
    volumes:
      - ./:/usr/src/app
    environment:
      ENVIRONMENT: ${ENVIRONMENT:-development}
      DEPLOY_STAGE: ${DEPLOY_STAGE:-local}
      AWS_DEFAULT_REGION: ${AWS_DEFAULT_REGION:-us-east-1}
      AWS_BEDROCK_REGION: ${AWS_BEDROCK_REGION:-us-east-1}
      AWS_EMBEDDING_MODEL: ${AWS_EMBEDDING_MODEL:-amazon.titan-embed-text-v1}
      AWS_ACCESS_KEY_ID: ${AWS_ACCESS_KEY_ID:-access-key}
      AWS_SECRET_ACCESS_KEY: ${AWS_SECRET_ACCESS_KEY:-secret-access-key}
      AWS_SESSION_TOKEN: ${AWS_SESSION_TOKEN:-session-token}
      CHROMA_AUTH_CREDENTIALS: ${CHROMA_AUTH_CREDENTIALS:-test-token}
      CHROMADB_HOST: ${CHROMADB_HOST:-db}
      CHROMADB_PORT: ${CHROMADB_PORT:-8000}
      CHROMADB_COLLECTION_NAME: ${CHROMADB_COLLECTION_NAME:-cheesecake-blog-aws}
    depends_on:
      db:
        condition: service_healthy