Feature/am 427 sidebar only basic html

Closes feature/am-427-sidebar-only-basic-html

src/Drush/Commands/MigrateCleanupCommand.php

+<?php
+
+namespace Drupal\cambridge_migrations\Drush\Commands;
+
+use Drush\Commands\DrushCommands;
+use Drupal\Core\Entity\EntityTypeManagerInterface;
+use Drupal\paragraphs\Entity\Paragraph;
+use Symfony\Component\DependencyInjection\ContainerInterface;
+use DOMDocument;
+use DOMElement;
+
+/**
+ * Drush command for cleaning up paragraph HTML content.
+ */
+class MigrateCleanupCommand extends DrushCommands {
+
+  /**
+   * The entity type manager.
+   *
+   * @var \Drupal\Core\Entity\EntityTypeManagerInterface
+   */
+  protected $entityTypeManager;
+
+  /**
+   * Constructs a new MigrateCleanupCommand object.
+   */
+  public function __construct(EntityTypeManagerInterface $entity_type_manager) {
+    parent::__construct();
+    $this->entityTypeManager = $entity_type_manager;
+  }
+
+  /**
+   * Clean HTML content by removing unwanted attributes and empty tags.
+   *
+   * @param string $html
+   *   The HTML content to clean.
+   *
+   * @return string
+   *   The cleaned HTML content.
+   */
+  protected function cleanHtml($html) {
+    if (empty($html)) {
+      return $html;
+    }
+
+    // Create a new DOM document
+    $doc = new DOMDocument();
+    
+    // Preserve whitespace to maintain formatting
+    $doc->preserveWhiteSpace = true;
+    $doc->formatOutput = true;
+
+    // Load HTML with UTF-8 encoding and suppress warnings
+    $doc->loadHTML('<?xml encoding="UTF-8">' . $html, LIBXML_HTML_NOIMPLIED | LIBXML_HTML_NODEFDTD);
+
+    // Remove unwanted attributes and empty tags recursively
+    $this->cleanNode($doc->documentElement);
+
+    // Get the cleaned HTML
+    $cleanedHtml = $doc->saveHTML($doc->documentElement);
+
+    // Remove the XML declaration if present
+    $cleanedHtml = preg_replace('/<\?xml[^>]+\?>/', '', $cleanedHtml);
+
+    return trim($cleanedHtml);
+  }
+
+  /**
+   * Recursively clean a DOM node by removing unwanted attributes and empty tags.
+   *
+   * @param DOMElement $node
+   *   The DOM node to clean.
+   *
+   * @return bool
+   *   TRUE if the node should be kept, FALSE if it should be removed.
+   */
+  protected function cleanNode(DOMElement $node) {
+    // Remove style and script elements entirely
+    $tagName = strtolower($node->tagName);
+    if ($tagName === 'style' || $tagName === 'script') {
+      return false;
+    }
+
+    // Remove class and style attributes
+    $node->removeAttribute('class');
+    $node->removeAttribute('style');
+
+    // Process child nodes
+    $children = [];
+    foreach ($node->childNodes as $child) {
+      $children[] = $child;
+    }
+
+    $hasNonEmptyContent = false;
+
+    foreach ($children as $child) {
+      if ($child->nodeType === XML_ELEMENT_NODE) {
+        // Recursively clean child element
+        $keepChild = $this->cleanNode($child);
+        if (!$keepChild) {
+          $node->removeChild($child);
+        } else {
+          $hasNonEmptyContent = true;
+        }
+      }
+      elseif ($child->nodeType === XML_TEXT_NODE) {
+        // Check if text node contains more than just whitespace and non-breaking spaces
+        $text = trim($child->nodeValue);
+        $textWithoutNbsp = str_replace("\xC2\xA0", '', $text); // Remove non-breaking spaces
+        if (!empty($textWithoutNbsp)) {
+          $hasNonEmptyContent = true;
+        }
+      }
+    }
+
+    // Return true if node has meaningful content
+    return $hasNonEmptyContent;
+  }
+
+  /**
+   * Clean up HTML content in text paragraphs.
+   *
+   * @command cambridge:migrate-cleanup
+   * @aliases cm-cleanup
+   */
+  public function migrateCleanup() {
+    try {
+      // Query all paragraphs of type "text"
+      $storage = $this->entityTypeManager->getStorage('paragraph');
+      $query = $storage->getQuery()
+        ->condition('type', 'text')
+        ->accessCheck(FALSE);
+      $pids = $query->execute();
+
+      if (empty($pids)) {
+        $this->logger()->warning(dt('No text paragraphs found to clean up.'));
+        return;
+      }
+
+      $success_count = 0;
+      $error_count = 0;
+
+      foreach ($pids as $pid) {
+        try {
+          /** @var \Drupal\paragraphs\Entity\Paragraph $paragraph */
+          $paragraph = $storage->load($pid);
+          
+          if (!$paragraph->hasField('field_text')) {
+            $this->logger()->warning(dt('Paragraph @pid does not have field_text field.', [
+              '@pid' => $pid,
+            ]));
+            continue;
+          }
+
+          $text_field = $paragraph->get('field_text');
+          $value = $text_field->value;
+          $format = $text_field->format;
+
+          // Clean the HTML content
+          $cleaned_value = $this->cleanHtml($value);
+
+          // Update the paragraph if content changed
+          if ($cleaned_value !== $value) {
+            $text_field->setValue([
+              'value' => $cleaned_value,
+              'format' => $format,
+            ]);
+            $paragraph->save();
+
+            $success_count++;
+            $this->logger()->notice(dt('Successfully cleaned HTML in paragraph @pid', [
+              '@pid' => $pid,
+            ]));
+          }
+        }
+        catch (\Exception $e) {
+          $error_count++;
+          $this->logger()->error(dt('Error cleaning paragraph @pid: @message', [
+            '@pid' => $pid,
+            '@message' => $e->getMessage(),
+          ]));
+        }
+      }
+
+      $this->logger()->notice(dt('HTML cleanup complete. Successes: @success, Errors: @errors', [
+        '@success' => $success_count,
+        '@errors' => $error_count,
+      ]));
+    }
+    catch (\Exception $e) {
+      $this->logger()->error(dt('HTML cleanup failed: @message', [
+        '@message' => $e->getMessage(),
+      ]));
+    }
+  }
+
+  /**
+   * {@inheritdoc}
+   */
+  public static function create(ContainerInterface $container) {
+    return new static(
+      $container->get('entity_type_manager')
+    );
+  }
+
+}