initial commit

hocr_sanitizer.info.yml

@@ -0,0 +1,7 @@
+name: 'HOCR Media Sanitizer'
+type: module
+description: Updates HOCR media'
+package: Custom
+core_version_requirement: ^10 || ^11
+dependencies:
+  - islandora:islandora

hocr_sanitizer.module

@@ -0,0 +1,15 @@
+<?php
+
+/**
+ * @file
+ * hocr_sanitizer.module
+ */
+
+/**
+ * Implements hook_help().
+ */
+function hocr_sanitizer_help($route_name, $route_match) {
+  if ($route_name == 'help.page.hocr_sanitizer') {
+    return t('The HOCR Sanitizer module batch-renames managed files ending in _HOCR.txt to .html and updates their MIME type to text/html.');
+  }
+}

hocr_sanitizer.routing.yml

@@ -0,0 +1,7 @@
+hocr_sanitizer.admin:
+  path: '/admin/config/media/hocr-sanitizer'
+  defaults:
+    _form: '\Drupal\hocr_sanitizer\Form\HocrSanitizerForm'
+    _title: 'HOCR Sanitizer'
+  requirements:
+    _permission: 'administer site configuration'

src/Form/HocrSanitizerForm.php

@@ -0,0 +1,59 @@
+<?php
+
+namespace Drupal\hocr_sanitizer\Form;
+
+use Drupal\Core\Form\FormBase;
+use Drupal\Core\Form\FormStateInterface;
+use Drupal\Core\Batch\BatchBuilder;
+use Drupal\hocr_sanitizer\Service\HocrSanitizerBatch;
+
+class HocrSanitizerForm extends FormBase {
+
+  /**
+   * {@inheritdoc}
+   */
+  public function getFormId() {
+    return 'hocr_sanitizer_admin_form';
+  }
+
+  /**
+   * {@inheritdoc}
+   */
+  public function buildForm(array $form, FormStateInterface $form_state) {
+    $form['description'] = [
+      '#markup' => $this->t('This will rename all files ending in <code>_HOCR.txt</code> to <code>.html</code> and update their MIME type to <code>text/html</code>.'),
+    ];
+
+    $form['submit'] = [
+      '#type' => 'submit',
+      '#value' => $this->t('Run HOCR Sanitizer Batch'),
+    ];
+
+    return $form;
+  }
+
+  /**
+   * {@inheritdoc}
+   */
+  public function submitForm(array &$form, FormStateInterface $form_state) {
+    $query = \Drupal::entityQuery('file')
+      ->condition('uri', '%_HOCR.txt', 'LIKE')
+      ->accessCheck(FALSE)
+      ->condition('filemime', 'text/plain');
+    $fids = $query->execute();
+
+    $batch_builder = (new BatchBuilder())
+      ->setTitle($this->t('Renaming HOCR text files to HTML...'))
+      ->setInitMessage($this->t('Starting HOCR rename...'))
+      ->setProgressMessage($this->t('Renaming file @current of @total...'))
+      ->setErrorMessage($this->t('An error occurred during file renaming.'));
+
+    foreach ($fids as $fid) {
+      $batch_builder->addOperation([HocrSanitizerBatch::class, 'renameFile'], [$fid]);
+    }
+
+    $batch_builder->setFinishCallback([HocrSanitizerBatch::class, 'batchFinished']);
+
+    batch_set($batch_builder->toArray());
+  }
+}

src/Service/HocrSanitizerBatch.php

@@ -0,0 +1,57 @@
+<?php
+
+namespace Drupal\hocr_sanitizer\Service;
+
+use Drupal\file\Entity\File;
+use Drupal\Core\File\FileExists;
+
+class HocrSanitizerBatch {
+
+  /**
+   * Batch operation: rename one HOCR file.
+   */
+  public static function renameFile($fid, array &$context) {
+    $file = File::load($fid);
+    if (!$file) {
+      $context['results']['errors'][] = "File with fid $fid not found.";
+      return;
+    }
+
+    $old_uri = $file->getFileUri();
+    $dirname = dirname($old_uri);
+    $old_filename = basename($old_uri);
+
+    if (!preg_match('/_HOCR\.txt$/i', $old_filename)) {
+      return; // Skip if no match.
+    }
+
+    $new_filename = preg_replace('/_HOCR\.txt$/i', '_HOCR.html', $old_filename);
+    $new_uri = $dirname . '/' . $new_filename;
+    $file_system = \Drupal::service('file_system');
+
+    try {
+      $file_system->move($old_uri, $new_uri, FileExists::Rename);
+      $file->setFileUri($new_uri);
+      $file->setMimeType('text/vnd.hocr+html');
+      $file->setFilename($new_filename);
+      $file->save();
+
+      $context['results']['renamed'][] = $new_filename;
+    }
+    catch (\Exception $e) {
+      $context['results']['errors'][] = "Failed to rename $old_filename: " . $e->getMessage();
+    }
+  }
+
+  /**
+   * Batch finished callback.
+   */
+  public static function batchFinished($success, $results, $operations) {
+    if (!empty($results['renamed'])) {
+      \Drupal::messenger()->addStatus(t('Renamed @count files.', ['@count' => count($results['renamed'])]));
+    }
+    if (!empty($results['errors'])) {
+      \Drupal::messenger()->addWarning(t('Some errors occurred: @count issues.', ['@count' => count($results['errors'])]));
+    }
+  }
+}