Exclude paths from search index

config/sync/search_api.index.server_dev.yml

@@ -9,6 +9,7 @@ dependencies:
   module:
     - node
     - search_api
+    - server_search
 id: server_dev
 name: server_dev
 description: ''
@@ -72,7 +73,7 @@ field_settings:
     datasource_id: 'entity:node'
     property_path: title
     type: text
-    boost: !!float 3
+    boost: 3.0
     dependencies:
       module:
         - node
@@ -114,6 +115,9 @@ processor_settings:
       preprocess_query: -30
   entity_status: {  }
   entity_type: {  }
+  exclude_nodes_by_path_alias:
+    excluded_nodes:
+      - /search
   language_with_fallback: {  }
   rendered_item: {  }
 tracker_settings:

web/modules/custom/server_general/tests/src/ExistingSite/ServerGeneralSearchTest.php

@@ -2,6 +2,8 @@
 
 namespace Drupal\Tests\server_general\ExistingSite;
 
+use Drupal\paragraphs\Entity\Paragraph;
+
 /**
  * A test case to test search integration.
  */
@@ -127,4 +129,94 @@ public function testFacetConfigSanity() {
     $this->expectNotToPerformAssertions();
   }
 
+  /**
+   * Checks if search page (LP with Search PT) isn't indexed.
+   *
+   * See "ExcludeNodeByPathAliasProcessor.php".
+   */
+  public function testSearchPageNotIndexed(): void {
+    $paragraph = Paragraph::create([
+      'type' => 'search',
+      'field_title' => [
+        'value' => 'Search Test',
+      ],
+    ]);
+    $paragraph->save();
+    $this->markEntityForCleanup($paragraph);
+
+    $path_alias = '/search_6KmX9x99aG5o13xvqjkO868iR';
+    $title = 'Search_6KmX9x99aG5o13xvqjkO868iR';
+
+    // Create a search LP with a specific title and path.
+    $node = $this->createNode([
+      'title' => $title,
+      'langcode' => 'en',
+      'type' => 'news',
+      'status' => 1,
+      'field_paragraphs' => [
+        $paragraph,
+      ],
+      'path' => [
+        'pathauto' => FALSE,
+        'alias' => $path_alias,
+      ],
+    ]);
+    $node->setPublished()->save();
+
+    // Trigger indexing.
+    $this->triggerPostRequestIndexing();
+
+    $this->waitForElasticSearchIndex(function () use ($node): void {
+      $this->drupalGet('/search', [
+        'query' => [
+          'key' => $node->label(),
+        ],
+      ]);
+      $session = $this->assertSession();
+      // Without search index processor this page should be in the results.
+      $session->elementTextContains('css', '.view-search', $node->label());
+    });
+
+    // Get the configuration factory service.
+    $config_factory = \Drupal::configFactory();
+
+    // Load the configuration of our "exclude_nodes_by_path_alias"
+    // search processor.
+    $config = $config_factory->getEditable('search_api.index.server_dev');
+
+    // Get the existing "excluded nodes" config array.
+    $excluded_nodes_original = $excluded_nodes_temporary = $config->get('processor_settings.exclude_nodes_by_path_alias.excluded_nodes');
+
+    if (!in_array($path_alias, $excluded_nodes_original)) {
+      // Add a new entry to the excluded nodes array.
+      $excluded_nodes_temporary[] = $path_alias;
+      // Set the updated excluded nodes array back to the configuration.
+      $config->set('processor_settings.exclude_nodes_by_path_alias.excluded_nodes', $excluded_nodes_temporary);
+
+      // Save the configuration.
+      $config->save();
+    }
+
+    // Save LP again so that cache gets cleared.
+    $node->save();
+
+    // Trigger indexing.
+    $this->triggerPostRequestIndexing();
+
+    // Wait for indexing to complete.
+    $this->waitForElasticSearchIndex(function () use ($node): void {
+      // First search using the exact long phrase.
+      $this->drupalGet('/search', [
+        'query' => [
+          'key' => $node->label(),
+        ],
+      ]);
+      $session = $this->assertSession();
+      $session->elementTextContains('css', '.view-empty', 'No results found');
+    });
+
+    // Restore original settings.
+    $config->set('processor_settings.exclude_nodes_by_path_alias.excluded_nodes', $excluded_nodes_original);
+  }
+
 }

web/modules/custom/server_search/src/Plugin/search_api/processor/ExcludeNodeByPathAliasProcessor.php

@@ -0,0 +1,128 @@
+<?php
+
+namespace Drupal\server_search\Plugin\search_api\processor;
+
+use Drupal\Core\Form\FormStateInterface;
+use Drupal\Core\Language\LanguageManagerInterface;
+use Drupal\Core\Plugin\PluginFormInterface;
+use Drupal\node\NodeInterface;
+use Drupal\path_alias\AliasManagerInterface;
+use Drupal\search_api\Plugin\PluginFormTrait;
+use Drupal\search_api\Processor\ProcessorPluginBase;
+use Symfony\Component\DependencyInjection\ContainerInterface;
+
+/**
+ * Provides a processor to exclude specific nodes from being indexed.
+ *
+ * @SearchApiProcessor(
+ *   id = "exclude_nodes_by_path_alias",
+ *   label = @Translation("Exclude nodes by path alias"),
+ *   description = @Translation("Exclude specific nodes from being indexed given path alias."),
+ *   stages = {
+ *     "alter_items" = 0,
+ *   },
+ *   locked = true,
+ *   hidden = false,
+ * )
+ */
+class ExcludeNodeByPathAliasProcessor extends ProcessorPluginBase implements PluginFormInterface {
+
+  use PluginFormTrait;
+
+  /**
+   * The alias manager.
+   *
+   * @var \Drupal\path_alias\AliasManagerInterface
+   */
+  protected AliasManagerInterface $aliasManager;
+
+  /**
+   * The core language manager service.
+   *
+   * @var \Drupal\Core\Language\LanguageManagerInterface
+   */
+  protected LanguageManagerInterface $languageManager;
+
+  /**
+   * {@inheritdoc}
+   */
+  public static function create(ContainerInterface $container, array $configuration, $plugin_id, $plugin_definition) {
+    /** @var static $processor */
+    $processor = parent::create($container, $configuration, $plugin_id, $plugin_definition);
+    $processor->aliasManager = $container->get('path_alias.manager');
+    $processor->languageManager = $container->get('language_manager');
+    return $processor;
+  }
+
+  /**
+   * {@inheritdoc}
+   */
+  public function defaultConfiguration() {
+    return [
+      'excluded_nodes' => [],
+    ];
+  }
+
+  /**
+   * {@inheritdoc}
+   */
+  public function alterIndexedItems(array &$items) {
+    $excluded_nodes_array = $this->getConfiguration()['excluded_nodes'];
+    if (empty($excluded_nodes_array)) {
+      return;
+    }
+
+    foreach ($items as $id => $item) {
+      /** @var \Drupal\Core\Entity\ContentEntityInterface $object */
+      $object = $item->getOriginalObject()->getValue();
+
+      if (!$object instanceof NodeInterface || $object->bundle() !== 'landing_page') {
+        continue;
+      }
+
+      $languages = array_keys($this->languageManager->getLanguages());
+      foreach ($languages as $lang_id) {
+        // Get current node item path aliases for each language.
+        $path_alias = $this->aliasManager->getAliasByPath('/node/' . $object->id(), $lang_id);
+        // Check if the nodes' path alias is in the list of excluded aliases.
+        if (in_array($path_alias, $excluded_nodes_array)) {
+          unset($items[$id]);
+        }
+      }
+    }
+  }
+
+  /**
+   * {@inheritdoc}
+   */
+  public function buildConfigurationForm(array $form, FormStateInterface $form_state) {
+    $excluded_nodes = $this->getConfiguration()['excluded_nodes'];
+    $form['#description'] = $this->t('Excludes nodes from indexing given their path aliases.');
+
+    $form['excluded_nodes'] = [
+      '#type' => 'textarea',
+      '#title' => $this->t('Node path aliases to exclude'),
+      '#description' => $this->t('Enter the node path aliases to exclude from the search index one per line with leading slash (e.g. "/search" and not "search").'),
+      '#default_value' => $excluded_nodes ? implode("\n", $excluded_nodes) : '',
+    ];
+
+    return $form;
+  }
+
+  /**
+   * {@inheritdoc}
+   */
+  public function submitConfigurationForm(array &$form, FormStateInterface $form_state) {
+    $excluded_nodes = $form_state->getValue('excluded_nodes');
+
+    // Split the textarea value by newline.
+    $excluded_nodes_array = preg_split('/[\r\n]+/', $excluded_nodes);
+
+    // Remove empty values.
+    $excluded_nodes_array = array_filter($excluded_nodes_array);
+
+    // Save each line as a separate item in the configuration array.
+    $this->setConfiguration(['excluded_nodes' => $excluded_nodes_array]);
+  }
+
+}