uuid: 73a5bbd6-ca98-4600-82f7-429111327a17
langcode: en
status: true
dependencies:
  module:
    - islandora_text_extraction
id: get_hocr_from_image
label: 'Get hOCR from image'
type: node
plugin: generate_ocr_derivative
configuration:
  queue: islandora-connector-ocr
  event: 'Generate Derivative'
  source_term_uri: 'http://pcdm.org/use#OriginalFile'
  derivative_term_uri: 'https://discoverygarden.ca/use#hocr'
  mimetype: text/html
  args: '-c tessedit_create_hocr=1 -c hocr_font_info=0'
  destination_media_type: extracted_text
  scheme: public
  path: '[date:custom:Y]-[date:custom:m]/[node:nid]-[term:name].txt'