kdeps-components-ocr

version: 1.0.2
requires:
    schemas/v1: '>=1.0.1'
files:
    component.yaml: |
        apiVersion: kdeps.io/v1
        kind: Component
        metadata:
          name: ocr
          description: Extract text from an image with the tesseract CLI
          version: "1.0.2"
          targetActionId: ocrText
        setup:
          osPackages:
            - tesseract-ocr          # kdeps build installs it in the image
          provides:
            - tesseract              # already on PATH: no package manager runs
        interface:
          inputs:
            - name: file
              type: string
              required: true
              description: Path to the image (png, jpg, tiff, bmp, gif). PDFs with a text layer go through the loader resource instead.
            - name: language
              type: string
              default: eng
              description: tesseract -l value, e.g. eng or eng+fra
            - name: psm
              type: integer
              default: 3
              description: tesseract --psm page segmentation mode (3 = fully automatic)
            - name: oem
              type: integer
              default: 3
              description: tesseract --oem engine mode (3 = whatever is available)
        resources:
          - actionId: ocrText
            name: OCR
            exec:
              command: tesseract
              args:
                - "{{ get('ocr.file') }}"
                - stdout              # write the text to stdout: the result's stdout
                - "-l"
                - "{{ get('ocr.language') }}"
                - "--psm"
                - "{{ get('ocr.psm') }}"
                - "--oem"
                - "{{ get('ocr.oem') }}"
              timeout: 60s