# Standalone Bumparr — runs next to your *arr stack on its own port.
# Point the volumes at your storage; edit config_files/ for your cams & fetch queue.
services:
  bumparr:
    build:
      context: .
      dockerfile: Dockerfile
    container_name: bumparr
    ports:
      - "${PORT:-8780}:8780"
    environment:
      - ASSET_ROOT=/assets
      - DB_PATH=/data/bumparr.db
      - BRAND=${BRAND:-Bumparr}
      # Set behind a reverse proxy so emitted M3U / bumper URLs are absolute.
      - PUBLIC_URL=${PUBLIC_URL:-}
      # Optional local model for the model-generated card kinds; blank = skip.
      # Grounded, procedural, and starter cards work with no model.
      - LLM_BASE=${LLM_BASE:-}
      - LLM_MODEL=${LLM_MODEL:-}
      - LLM_DISABLE_THINKING=${LLM_DISABLE_THINKING:-}
      # Stock-footage keys for the shipped starter seeds (free self-signup).
      - PEXELS_API_KEY=${PEXELS_API_KEY:-}
      - PIXABAY_API_KEY=${PIXABAY_API_KEY:-}
      # Timezone for the clock card; containers are UTC without it.
      - TZ=${TZ:-UTC}
    volumes:
      # Your bumper library (rw so live-cam snippets + fetched PD content can land).
      - ${ASSETS:-./assets}:/assets:rw
      - ${DATA:-./data}:/data
      # Live edits without rebuild (optional; safe to remove for a pinned deploy).
      - ./bumparr:/app/bumparr:ro
    restart: unless-stopped