add #39 to 7.x-2.x umhs-28: use canonical URL for results (#94)

add #39 to 7.x-2.x umhs-28: use canonical URL for results

README.md

@@ -48,7 +48,7 @@ When changes to [federated-search-react](https://github.com/palantirnet/federate
 
 ## More information
 
-Full documentation for this module is in the [handbook on Drupal.org](https://www.drupal.org/docs/7/modules/search-api-federated-solr/search-api-federated-solr-module)
+Full documentation for this module is available in the [handbook on Drupal.org](https://www.drupal.org/docs/7/modules/search-api-federated-solr/search-api-federated-solr-module)
 
 * [How to use this module](https://www.drupal.org/docs/7/modules/search-api-federated-solr/search-api-federated-solr-module/intro-install-configure)
 * [How to configure a Search API Index for federated search](https://www.drupal.org/docs/8/modules/search-api-federated-solr/federated-search-schema)

search_api_federated_solr.info

@@ -9,6 +9,7 @@ dependencies[] = search_api_solr
 dependencies[] = token
 
 files[] = search-api_federated_solr.module
+files[] = src/SearchApiFederatedSolrCanonicalUrl.php
 files[] = src/SearchApiFederatedSolrField.php
 files[] = src/SearchApiFederatedSolrTerms.php
 files[] = src/SearchApiFederatedSolrRemap.php

search_api_federated_solr.module

@@ -69,6 +69,11 @@ function search_api_federated_solr_search_api_alter_callback_info() {
     'description' => t('The links to the node on all available sites. This can be useful if indexing multiple sites with a single search index.'),
     'class' => 'SearchApiFederatedSolrUrls',
   );
+  $callbacks['canonical_url'] = array(
+    'name' => t('Canonical URL'),
+    'description' => t('Preferred URL for this content.'),
+    'class' => 'SearchApiFederatedSolrCanonicalUrl',
+  );
   $callbacks['federated_field'] = array(
     'name' => t('Federated Field'),
     'description' => t('A token or free text field that can be customized per-bundle.'),

src/SearchApiFederatedSolrCanonicalUrl.php

@@ -0,0 +1,94 @@
+<?php
+
+/**
+ * Class SearchApiFederatedSolrCanonicalUrl
+ * Provides a Search API index data alteration that indicates the preferred
+ * URL content is available on to each indexed item.
+ */
+class SearchApiFederatedSolrCanonicalUrl extends SearchApiAbstractAlterCallback {
+
+  /**
+   * @var SearchApiIndex
+   */
+  protected $index;
+
+  /**
+   * @var array
+   */
+  protected $options;
+
+  /**
+   * {@inheritdoc}
+   */
+  public function propertyInfo() {
+    return [
+      'canonical_url' => [
+        'label' => t('Canonical URL'),
+        'description' => t('Preferred URL for this content'),
+        'type' => 'uri',
+        'cardinality' => -1,
+      ],
+    ];
+  }
+
+  /**
+   * {@inheritdoc}
+   */
+  public function alterItems(array &$items) {
+
+    if ($this->useDomainAccess()) {
+      $this->addDomainUrl($items);
+    }
+    else {
+      $this->addUrl($items);
+    }
+
+  }
+
+  protected function addUrl(array &$items) {
+    foreach ($items as &$item) {
+      $url = $this->index->datasource()->getItemUrl($item);
+      if (!$url) {
+        $item->canonical_url = NULL;
+        continue;
+      }
+      $item->canonical_url = url($url['path'], array('absolute' => TRUE) + $url['options']);
+    }
+  }
+
+  protected function addDomainUrl(array &$items) {
+    $entity_type = $this->index->getEntityType();
+    $entity_info = entity_get_info($entity_type);
+
+    foreach ($items as $item) {
+      $id = entity_id($entity_type, $item);
+
+      // Get the entity object for the item being indexed, exit if there's somehow not one.
+      $entity = current(entity_load($entity_type, [$id]));
+      if (!$entity) {
+        return;
+      }
+
+      // Determine if there is a canonical URL for the content.
+      // This only comes into play if domain source is used.
+      if (isset($entity->domain_source) && $entity->domain_source == DOMAIN_SOURCE_USE_ACTIVE) {
+        $this->canonical_url = '';
+      }
+      else {
+        $list = [$item];
+        $this->addUrl($list);
+      }
+    }
+
+  }
+
+  /**
+   * Whether to use the canonical value from Domain Source.
+   *
+   * @return bool
+   */
+  protected function useDomainAccess() {
+    return defined('DOMAIN_SOURCE_USE_ACTIVE');
+  }
+
+}

src/SearchApiFederatedSolrUrls.php

@@ -70,8 +70,13 @@ protected function addDomainUrls(array &$items) {
       }
 
       $urls = domain_get_content_urls($entity);
-
-      $item->urls = $urls;
+      if (!empty($urls)) {
+        $item->urls = $urls;
+      }
+      else {
+        $list = [$item];
+        $this->addUrl($list);
+      }
     }
 
   }