email.yaml

# Email Invoice Processor Configuration
# Uses URL-style configuration with .env variable support

routes:
  - name: "email_invoice_processor"
    description: "Process email attachments for invoices"

    # Source configuration (IMAP server)
    from: "imap://{{EMAIL_USER}}:{{EMAIL_PASSWORD}}@{{EMAIL_SERVER}}:{{EMAIL_PORT}}/INBOX"

    # Processors configuration
    processors:
      # Step 1: Filter emails from specific month
      - type: "filter"
        condition: "{{date|date('%Y-%m')}} == '{{YEAR|default(now|date('%Y'))}}-{{'%02d'|format(MONTH|default(now.month))}}'"

      # Step 2: Extract attachments
      - type: "external"
        command: "python -m email_invoice_processor.processor"
        input_format: "email"
        output_format: "json"
        config:
          output_dir: "{{OUTPUT_DIR|default('./output')}}"
          supported_extensions: [".pdf", ".jpg", ".jpeg", ".png"]

      # Step 3: Process PDFs with local OCR
      - type: "local_ocr"
        engine: "tesseract"
        config:
          tesseract_cmd: "{{TESSERACT_CMD|default('tesseract')}}"
          languages: ["eng", "pol"]
          dpi: 300
          oem: 1 # LSTM + Legacy OCR Engine
          psm: 6 # Assume a single uniform block of text
          # Local processing flags
          use_gpu: false
          max_threads: 4
          temp_dir: "/tmp/ocr_processing"

    # Error handling
    error_handlers:
      - type: "retry"
        max_attempts: 3
        delay: 5 # seconds

      - type: "notify"
        to: "smtp://{{ADMIN_EMAIL}}"
        template: |
          Subject: Invoice Processing Failed

          Error processing email: {{error}}

          Details:
          {{error_details}}

    # Output configuration
    to:
      - "file://{{OUTPUT_DIR|default('./output')}}/processed/{{date|date('%Y-%m')}}/{{message_id}}.json"
      - "smtp://{{REPORT_EMAIL}}?subject=Processed Invoice {{date|date('%Y-%m')}}"

# Global configuration
config:
  log_level: "{{LOG_LEVEL|default('INFO')}}"
  log_file: "{{LOG_FILE|default('email_processor.log')}}"

  # Email polling interval in seconds
  poll_interval: 300 # 5 minutes

  # Maximum number of emails to process in one batch
  batch_size: 50

  # Archive processed emails (move to another folder)
  archive_processed: true
  archive_folder: "Processed_Invoices"