Merge pull request #897 from Islandora/hocr

Add hOCR option to Text Extraction Media Attachment action and IIIF Manifest

Author: wgilling

modules/islandora_iiif/src/Plugin/views/style/IIIFManifest.php

@@ -189,21 +189,27 @@ public function render() {
    */
   protected function getTileSourceFromRow(ResultRow $row, $iiif_address, $iiif_base_id) {
     $canvases = [];
-    foreach ($this->options['iiif_tile_field'] as $iiif_tile_field) {
+    foreach (array_filter(array_values($this->options['iiif_tile_field'])) as $iiif_tile_field) {
       $viewsField = $this->view->field[$iiif_tile_field];
+      $iiif_ocr_file_field = !empty($this->options['iiif_ocr_file_field']) ? array_filter(array_values($this->options['iiif_ocr_file_field'])) : [];
+      $ocrField = count($iiif_ocr_file_field) > 0 ? $this->view->field[$iiif_ocr_file_field[0]] : NULL;
       $entity = $viewsField->getEntity($row);
 
       if (isset($entity->{$viewsField->definition['field_name']})) {
 
         /** @var \Drupal\Core\Field\FieldItemListInterface $images */
         $images = $entity->{$viewsField->definition['field_name']};
-        foreach ($images as $image) {
+        foreach ($images as $i => $image) {
           if (!$image->entity->access('view')) {
             // If the user does not have permission to view the file, skip it.
             continue;
           }
+
+          $ocrs = $entity->{$ocrField->definition['field_name']};
+
           // Create the IIIF URL for this file
           // Visiting $iiif_url will resolve to the info.json for the image.
+          $ocr = isset($ocrs[$i]) ? $ocrs[$i] : FALSE;
           $file_url = $image->entity->createFileUrl(FALSE);
           $mime_type = $image->entity->getMimeType();
           $iiif_url = rtrim($iiif_address, '/') . '/' . urlencode($file_url);
@@ -241,8 +247,7 @@ protected function getTileSourceFromRow(ResultRow $row, $iiif_address, $iiif_bas
               }
             }
           }
-
-          $canvases[] = [
+          $tmp_canvas = [
             // @see https://iiif.io/api/presentation/2.1/#canvas
             '@id' => $canvas_id,
             '@type' => 'sc:Canvas',
@@ -271,6 +276,17 @@ protected function getTileSourceFromRow(ResultRow $row, $iiif_address, $iiif_bas
               ],
             ],
           ];
+
+          if (isset($ocr) && $ocr != FALSE) {
+            $tmp_canvas['seeAlso'] = [
+              '@id' => $ocr->entity->createFileUrl(FALSE),
+              'format' => 'text/vnd.hocr+html',
+              'profile' => 'http://kba.cloud/hocr-spec',
+              'label' => 'hOCR embedded text',
+            ];
+          }
+
+          $canvases[] = $tmp_canvas;
         }
       }
     }
@@ -313,6 +329,7 @@ protected function defineOptions() {
     $options = parent::defineOptions();
 
     $options['iiif_tile_field'] = ['default' => ''];
+    $options['iiif_ocr_file_field'] = ['default' => ''];
 
     return $options;
   }
@@ -368,6 +385,15 @@ public function buildOptionsForm(&$form, FormStateInterface $form_state) {
       // otherwise could lock up the form when setting up a View.
       '#required' => count($field_options) > 0,
     ];
+
+    $form['iiif_ocr_file_field'] = [
+      '#title' => $this->t('Structured OCR data file field'),
+      '#type' => 'checkboxes',
+      '#default_value' => $this->options['iiif_ocr_file_field'],
+      '#description' => $this->t('The source of structured OCR text for each entity.'),
+      '#options' => $field_options,
+      '#required' => FALSE,
+    ];
   }
 
   /**

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

@@ -8,7 +8,7 @@
 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 @@ public function defaultConfiguration() {
     $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 @@ public function defaultConfiguration() {
   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 @@ public function buildConfigurationForm(array $form, FormStateInterface $form_sta
     $last = array_slice($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($form['args']);
@@ -81,17 +92,29 @@ public function validateConfigurationForm(array &$form, FormStateInterface $form
   public function submitConfigurationForm(array &$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()