toolshed-8.x-1.x-dev/modules/toolshed_search/src/Plugin/search_api/processor/ExcludeSystemUrlsProcessor.php

<?php

namespace Drupal\toolshed_search\Plugin\search_api\processor;

use Drupal\Core\Config\ConfigFactoryInterface;
use Drupal\Core\Entity\EntityInterface;
use Drupal\Core\Entity\EntityMalformedException;
use Drupal\Core\Entity\Exception\UndefinedLinkTemplateException;
use Drupal\Core\Path\PathValidatorInterface;
use Drupal\Core\Plugin\ContainerFactoryPluginInterface;
use Drupal\search_api\Processor\ProcessorPluginBase;
use Symfony\Component\DependencyInjection\ContainerInterface;

/**
 * Prevents the search indexing of the site 403 or 404 content.
 *
 * A content page (or other indexable entity) can be set as the response for
 * these statuses, but they often get indexed and appear in search results
 * unintentionally. This processor prevents them from getting added.
 *
 * @todo Extend this to have options for the home page or other special purpose
 * pages. This could be things like a welcome page.
 *
 * @SearchApiProcessor(
 *   id = "toolshed_exclude_system_urls",
 *   label = @Translation("Exclude Site URLs"),
 *   description = @Translation("Prevent the indexing of 404 and 403 pages."),
 *   stages = {
 *     "alter_items" = 0,
 *   },
 *   locked = true,
 *   hidden = true,
 * )
 */
class ExcludeSystemUrlsProcessor extends ProcessorPluginBase implements ContainerFactoryPluginInterface {

  /**
   * Build, fetch and update configurations.
   *
   * @var \Drupal\Core\Config\ConfigFactoryInterface
   */
  protected ConfigFactoryInterface $configFactory;

  /**
   * Service for validating and resolving URIs to Drupal routes and Url objects.
   *
   * @var \Drupal\Core\Path\PathValidatorInterface
   */
  protected PathValidatorInterface $pathValidator;

  /**
   * Create a new instance of the ExcludeSystemUrlsProcessor.
   *
   * @param array $configuration
   *   Configurations to use for this processor's options.
   * @param string $plugin_id
   *   The ID of the plugin.
   * @param mixed $plugin_definition
   *   The definition for this plugin instances.
   * @param \Drupal\Core\Config\ConfigFactoryInterface $config_factory
   *   Service for managing and fetching configurations.
   * @param \Drupal\Core\Path\PathValidatorInterface $path_validator
   *   Service which is used to resolve a URI to Drupal\Core\Url objects.
   */
  public function __construct(array $configuration, $plugin_id, $plugin_definition, ConfigFactoryInterface $config_factory, PathValidatorInterface $path_validator) {
    parent::__construct($configuration, $plugin_id, $plugin_definition);

    $this->configFactory = $config_factory;
    $this->pathValidator = $path_validator;
  }

  /**
   * {@inheritdoc}
   */
  public static function create(ContainerInterface $container, array $configuration, $plugin_id, $plugin_definition): static {
    return new static(
      $configuration,
      $plugin_id,
      $plugin_definition,
      $container->get('config.factory'),
      $container->get('path.validator')
    );
  }

  /**
   * {@inheritdoc}
   */
  public function alterIndexedItems(array &$items): void {
    $config = $this->configFactory
      ->get('system.site')
      ->get('page');

    foreach ($config as $uri) {
      $url = $this->pathValidator->getUrlIfValid($uri);

      if ($url && $url->isRouted()) {
        $routes[$url->getRouteName()] = $url->getRouteParameters();
      }
    }

    if (empty($routes)) {
      return;
    }

    foreach ($items as $item_id => $item) {
      $object = $item->getOriginalObject()->getValue();

      if ($object instanceof EntityInterface) {
        try {
          $entityUrl = $object->toUrl();
          $routeName = $entityUrl->getRouteName();

          // Only if the route and the parameters match do we consider these
          // the same target. This avoids assuming things about the entity ID
          // and route parameters for all entity types.
          if (!empty($routes[$routeName]) && $routes[$routeName] == $entityUrl->getRouteParameters()) {
            unset($items[$item_id]);
          }
        }
        catch (EntityMalformedException | UndefinedLinkTemplateException) {
          // Skip as this entity doesn't support regular canonical link route.
        }
      }
    }
  }

}