docker-compose.yml

# App for postprocessing scanned documents and photos.
# Crop photos and doing OCR on documents. See and manage files in http-filebrowser or in samba share.
# Reverse proxy for a single point of contact to network scanners and users.
# Should only be used in private LANs because of old/insecure SambaV1, used to be compatible with old network scanners.
services:
  # ScanCropper itself
  srv-scancropper:
    build:
      context: .
      dockerfile: Dockerfile
    container_name: scancropper
    environment:
      SC_DIR: "/mnt/scans/input"
      SC_ODIR: "/mnt/scans/results"
      SC_ODIR_PROCESSED: "/mnt/scans/processed"
      # SC_NO_DIRSCAN: 1
      SC_WATCH: 1
      SC_POLLING_TIMEOUT: 3
      SC_OUTPUT_JPEG_QUALITY: 86
      # SC_OUTPUT_FILE_NAME_PREFIX: "MyPhoto "
      # SC_OUTPUT_FILE_NAME_PREFIX_STRFTIME: "%Y-%m-%d_%H%M%S "
      TZ: "Europe/Berlin"
    restart: unless-stopped
    volumes:
      - v-photos-input:/mnt/scans/input
      - v-photos-processed:/mnt/scans/processed
      - v-photos-results:/mnt/scans/results

  # Proxy for central serving all services - by https://doc.traefik.io/traefik/
  srv-proxy:
    image: traefik:latest
    container_name: scanproxy
    command: >
      --log.level=DEBUG
      --api.insecure
      --providers.docker
      --providers.docker.exposedByDefault=false
      --entryPoints.filebrowser.address=:80
      --entryPoints.smb445.address=:445
    networks:
      - backend
      - frontend
    ports:
      - 80
      - 445
    volumes:
      - /var/run/docker.sock:/var/run/docker.sock:ro

  # Web file browser to upload / download / delete files - by https://github.com/filebrowser/filebrowser/
  srv-filebrowser:
    image: filebrowser/filebrowser:v2
    container_name: filebrowser
    environment:
      FB_USERNAME: scanner
      # Password is sc4nn3rPazzw()rrrd
      # todo: change password (bcrypt hash), https://bcrypt-generator.com or https://gchq.github.io/CyberChef/#recipe=Bcrypt(12)
      FB_PASSWORD: '$2a$12$7IIEB9y.pt7Q0S.03.g0qeHcRXBMI5IG7tkUbT7Dr6DOPwAe/LHnC'
      # or disable auth:
      # FB_NOAUTH: 1
    labels:
      - traefik.enable=true
      - traefik.http.routers.filebrowser.entrypoints=filebrowser
      - traefik.http.routers.filebrowser.rule=PathPrefix("/")
    networks:
      - backend
    ports:
      - 80
    volumes:
      - v-photos-input:/srv/photos-input
      - v-photos-processed:/srv/photos-processed
      - v-photos-results:/srv/photos-results
      - v-documents-input:/srv/documents-input
      - v-documents-ocr:/srv/documents-ocr
      - v-documents-processed:/srv/documents-processed

  # Samba by https://hub.docker.com/r/dperson/samba
  # todo: change the password sc4nn3rPazzw()rrrd
  srv-samba:
    image: dperson/samba
    container_name: samba
    command: >
      -p
      -S
      -r
      -u "scanner;sc4nn3rPazzw()rrrd"
      -s "Photoscan-Input;/share/scanner/photos-input;yes;no;no;scanner;;;Photos to crop"
      -s "OCR-Incomming-Documents;/share/scanner/documents-input;yes;no;no;scanner;;;Documents to OCR"
      -s "Scanned;/share/scanner;yes;no;no;scanner;;;Scanned and processed files"
    labels:
      - traefik.enable=true
      - traefik.tcp.routers.samba.entrypoints=smb445
      - traefik.tcp.routers.samba.rule=HostSNI("*")
      - traefik.tcp.routers.samba.service=samba-srv
      - traefik.tcp.services.samba-srv.loadbalancer.server.port=445
    networks:
      - backend
    ports:
      - 445
    volumes:
      - v-photos-input:/share/scanner/photos-input
      - v-photos-processed:/share/scanner/photos-processed
      - v-photos-results:/share/scanner/photos-results
      - v-documents-input:/share/scanner/documents-input
      - v-documents-ocr:/share/scanner/documents-ocr
      - v-documents-processed:/share/scanner/documents-processed

  # OCR PDFs by https://github.com/ocrmypdf/OCRmyPDF
  # https://ocrmypdf.readthedocs.io/en/latest/batch.html#hot-watched-folders
  srv-pdfocr:
    image: jbarlow83/ocrmypdf
    container_name: pdfocr
    entrypoint: python3
    command: watcher.py
    environment:
      # https://github.com/ocrmypdf/OCRmyPDF/blob/main/misc/watcher.py#L162C13-L162C19
      OCR_USE_POLLING: 1
      OCR_POLL_NEW_FILE_SECONDS: 3
      OCR_RETRIES_LOADING_FILE: 5
      OCR_ON_SUCCESS_ARCHIVE: 1
      # set ocrmypdf arguments as JSON, https://ocrmypdf.readthedocs.io/en/latest/cookbook.html
      OCR_JSON_SETTINGS: '{"language": "eng+deu", "deskew": true, "rotate_pages": true, "clean": true, "redo_ocr": true}'
      PYTHONBUFFERED: 1
    restart: unless-stopped
    volumes:
      - v-documents-input:/input
      - v-documents-ocr:/output
      - v-documents-processed:/processed


# Setup volumes where to store the input scanned files, results and the processed original files.
volumes:
  # Volumes for scanned photos
  v-photos-input:
    driver: local
    driver_opts:
      device: ./scans/photos/input/
      o: bind
      type: none
  # Cropped results output
  v-photos-results:
    driver: local
    driver_opts:
      device: ./scans/photos/results/
      o: bind
      type: none
  # Archive of original files after processing
  v-photos-processed:
    driver: local
    driver_opts:
      device: ./scans/photos/processed/
      o: bind
      type: none
  #
  # Volumes for pdf-documents to process OCR:
  v-documents-input:
    driver: local
    driver_opts:
      device: ./scans/documents/input
      o: bind
      type: none
  # OCRed results output
  v-documents-ocr:
    driver: local
    driver_opts:
      device: ./scans/documents/ocr/
      o: bind
      type: none
  # archive of original files after processing
  v-documents-processed:
    driver: local
    driver_opts:
      device: ./scans/documents/processed/
      o: bind
      type: none

# Define 2 networks: frontend to the scanner and user; backend is internal only
networks:
  frontend:
#    external: true # must be created before
  backend: