WIP Modify GenerateOCRDerivativeFile to support hOCR

modules/islandora_text_extraction/src/Plugin/Action/GenerateOCRDerivativeFile.php

@@ -8,7 +8,7 @@ use Drupal\Core\Url;
 use Drupal\islandora\Plugin\Action\AbstractGenerateDerivativeMediaFile;
 
 /**
- * Emits a Node for generating fits derivatives event.
+ * Generates OCR derivatives event.
  *
  * @Action(
  *   id = "generate_extracted_text_file",
@@ -29,6 +29,7 @@ class GenerateOCRDerivativeFile extends AbstractGenerateDerivativeMediaFile {
     $config['destination_media_type'] = 'file';
     $config['scheme'] = $this->config->get('default_scheme');
     $config['destination_text_field_name'] = '';
+    $config['text_format'] = 'plain_text';
     return $config;
   }
 
@@ -38,7 +39,7 @@ class GenerateOCRDerivativeFile extends AbstractGenerateDerivativeMediaFile {
   public function buildConfigurationForm(array $form, FormStateInterface $form_state) {
     $map = $this->entityFieldManager->getFieldMapByFieldType('text_long');
     $file_fields = $map['media'];
-    $field_options = array_combine(array_keys($file_fields), array_keys($file_fields));
+    $field_options = ['none' => $this->t('None')] + array_combine(array_keys($file_fields), array_keys($file_fields));
     $form = parent::buildConfigurationForm($form, $form_state);
     $form['mimetype']['#description'] = $this->t('Mimetype to convert to (e.g. application/xml, etc...)');
     $form['mimetype']['#value'] = 'text/plain';
@@ -48,13 +49,23 @@ class GenerateOCRDerivativeFile extends AbstractGenerateDerivativeMediaFile {
     $last = array_slice($form, count($form) - $position + 1);
 
     $middle['destination_text_field_name'] = [
-      '#required' => TRUE,
+      '#required' => FALSE,
       '#type' => 'select',
       '#options' => $field_options,
       '#title' => $this->t('Destination Text field Name'),
       '#default_value' => $this->configuration['destination_text_field_name'],
       '#description' => $this->t('Text field on Media Type to hold extracted text.'),
     ];
+    $middle['text_format'] = [
+      '#type' => 'select',
+      '#title' => $this->t('Format'),
+      '#options' => [
+        'plain_text' => $this->t('Plain text'),
+        'hocr' => $this->t('hOCR text with positional data'),
+      ],
+      '#default_value' => $this->configuration['text_format'],
+      '#description' => $this->t("The type of text to be returned."),
+    ];
     $form = array_merge($first, $middle, $last);
 
     unset($form['args']);
@@ -81,17 +92,28 @@ class GenerateOCRDerivativeFile extends AbstractGenerateDerivativeMediaFile {
   public function submitConfigurationForm(array &$form, FormStateInterface $form_state) {
     parent::submitConfigurationForm($form, $form_state);
     $this->configuration['destination_text_field_name'] = $form_state->getValue('destination_text_field_name');
+    $this->configuration['text_format'] = $form_state->getValue('text_format');
+    switch ($form_state->getValue('text_format')) {
+      case 'hocr':
+        $this->configuration['args'] = '-c tessedit_create_hocr=1 -c hocr_font_info=0';
+        break;
+      case 'plain_text':
+        $his->configuration['args'] = '';
+        break;
+    }
   }
 
   /**
    * Override this to return arbitrary data as an array to be json encoded.
    */
   protected function generateData(EntityInterface $entity) {
+
     $data = parent::generateData($entity);
     $route_params = [
       'media' => $entity->id(),
       'destination_field' => $this->configuration['destination_field_name'],
       'destination_text_field' => $this->configuration['destination_text_field_name'],
+      'text_format' => $this->configuration['text_format'],
     ];
     $data['destination_uri'] = Url::fromRoute('islandora_text_extraction.attach_file_to_media', $route_params)
       ->setAbsolute()