gistfile1.txt

<?php

namespace Drupal\custom_migration\Plugin\migrate\process;

use Drupal\Component\Utility\Html;
use Drupal\migrate\ProcessPluginBase;
use Drupal\migrate\MigrateExecutableInterface;
use Drupal\migrate\Row;

/**
 * If any of the conditions are met, we skip the current row.
 *
 * @MigrateProcessPlugin(
 *   id = "image_insert",
 *   handle_multiples = TRUE
 * )
 */
class ImageInsert extends ProcessPluginBase {

  /**
   * @var \Drupal\Core\Entity\EntityStorageInterface
   */
  protected $fileStorage;

  /**
   * {@inheritdoc}
   */
  public function transform($value, MigrateExecutableInterface $migrate_executable, Row $row, $destination_property) {
    // Prevent useless processing if there are no image-insert tags at all.
    if (stripos($value, 'image-insert') !== FALSE) {
      $dom = Html::load($value);
      $xpath = new \DOMXPath($dom);
      $items = $xpath->query('//div[@class and contains(concat(\' \', normalize-space(@class), \' \'), \' image-insert \')]');
      /** @var \DomNode $item */
      foreach ($items as $item) {
        // Get the first image.
        $image = $xpath->query(".//img", $item)->item(0);
        if ($image === null) {
          // There's no image. Nothing to do here.
          continue;
        }

        // Get the caption.
        $caption = $xpath->query('.//div[@class and contains(concat(\' \', normalize-space(@class), \' \'), \' image-insert-caption \')]', $item);
        $captions = [];
        foreach ($caption as $caption_item) {
          $captions[] = $caption_item->textContent;
        }
        // Convert html entities into UTF-8 characters so &nbsp; is trimmed.
        $caption = trim(html_entity_decode(implode(' ', $captions)));
        if (!preg_match('/^\W*$/', $caption)) {
          $image->setAttribute('data-caption', $caption);
        }

        // Set the aligment.
        $class = $item->getAttribute('class');
        if (strpos($class, 'image-insert-left') !== FALSE) {
          $image->setAttribute('data-align', 'left');
        }
        elseif (strpos($class, 'image-insert-right') !== FALSE) {
          $image->setAttribute('data-align', 'right');
        }

        $image->setAttribute('data-entity-type', "file");


        $src = $image->getAttribute('src');
        // Make the URL relative just in case.
        $http_host = 'HTTP_HOST_SHOULD_BE_CONFIGURABLE';
        $src = str_replace('/sites/all/files/', '/sites/default/files/', $src);
        $src = preg_replace('|^https?://' . $http_host . '|', '', $src);
        $image->setAttribute('src', $src);

        // Need to look up the file to add the data-entity-uuid attribute.
        $src = str_replace('/sites/default/files/', 'public://', $src);
        $this->fileStorage = \Drupal::entityTypeManager()->getStorage('file');
        $file = $this->fileStorage->loadByProperties(['uri' => urldecode($src)]);
        $file = reset($file);
        if ($file) {
          $image->setAttribute('data-entity-uuid', $file->uuid());
        }

        // If the image is wrapped in an A tag, preserve that.
        $image_parent = $image->parentNode;
        if ($image_parent->tagName === 'a') {
          $clone = $image_parent->cloneNode(TRUE);
        }
        else {
          $clone = $image->cloneNode(TRUE);
        }

        // Return the new HTML.
        $item->parentNode->replaceChild($clone, $item);
      }
      $value = Html::serialize($dom);
    }
    return $value;
  }

}