YSP-672: AI: Add media payloads and pipelines

ai_engine.permissions.yml

@@ -1,7 +1,7 @@
 # Create permission for managin AI Engine settings.
 administer ai engine:
   title: 'Administer AI Engine'
-  description: 'Enable services and change sensative settings.'
+  description: 'Enable services and change sensitive settings.'
 manage ai engine settings:
   title: 'Manage AI Engine Settings'
   description: 'Set and update AI Engine content and settings.'

modules/ai_engine_embedding/src/Form/AiEngineEmbeddingSettings.php

@@ -2,9 +2,11 @@
 
 namespace Drupal\ai_engine_embedding\Form;
 
-use Drupal\ai_engine_embedding\Service\EntityUpdate;
+use Drupal\Core\Entity\EntityTypeManagerInterface;
 use Drupal\Core\Form\ConfigFormBase;
 use Drupal\Core\Form\FormStateInterface;
+use Drupal\ai_engine_embedding\Service\EntityUpdate;
+use Symfony\Component\DependencyInjection\ContainerInterface;
 
 /**
  * Setting form for the AI Engine Embedding module.
@@ -18,6 +20,13 @@ class AiEngineEmbeddingSettings extends ConfigFormBase {
    */
   const CONFIG_NAME = 'ai_engine_embedding.settings';
 
+  /**
+   * Entity type manager.
+   *
+   * @var \Drupal\Core\Entity\EntityTypeManager
+   */
+  protected $entityTypeManager;
+
   /**
    * {@inheritdoc}
    */
@@ -32,6 +41,21 @@ protected function getEditableConfigNames() {
     return [self::CONFIG_NAME];
   }
 
+  public function __construct(
+    EntityTypeManagerInterface $entityTypeManager,
+  ) {
+    $this->entityTypeManager = $entityTypeManager;
+  }
+
+  /**
+   * {@inheritdoc}
+   */
+  public static function create(ContainerInterface $container) {
+    return new static(
+      $container->get('entity_type.manager'),
+    );
+  }
+
   /**
    * {@inheritdoc}
    */
@@ -67,6 +91,13 @@ public function buildForm(array $form, FormStateInterface $form_state) {
       '#description' => $this->t('The chunk size to split each document into'),
       '#default_value' => $config->get('azure_chunk_size') ?? 3000,
     ];
+    $form['included_media_types'] = [
+      '#type' => 'checkboxes',
+      '#title' => $this->t('Included Media Types'),
+      '#options' => $this->getMediaTypes(),
+      '#default_value' => array_keys($config->get('included_media_types') ?? []),
+      '#default_value' => $config->get('included_media_types') ?? [],
+    ];
     $form['actions'] = [
       '#type' => 'details',
       '#title' => $this->t('Embedding Operations'),
@@ -112,6 +143,7 @@ public function submitForm(array &$form, FormStateInterface $form_state) {
       ->set('azure_search_service_index', $form_state->getValue('azure_search_service_index'))
       ->set('azure_embedding_service_url', $form_state->getValue('azure_embedding_service_url'))
       ->set('azure_chunk_size', $azure_chunk_size)
+      ->set('included_media_types', $form_state->getValue('included_media_types'))
       ->save();
     parent::submitForm($form, $form_state);
   }
@@ -124,4 +156,19 @@ public function actionUpsertAllDocuments(array &$form, FormStateInterface $form_
     $service->addAllDocuments();
   }
 
+  /**
+   * Retrieves the list of media types.
+   *
+   * @return array
+   *   An array of media type labels.
+   */
+  protected function getMediaTypes() {
+    $media_types = [];
+    foreach ($this->entityTypeManager->getStorage('media_type')->loadMultiple() as $media_type) {
+      $media_types[$media_type->id()] = $media_type->label();
+    }
+
+    return $media_types;
+  }
+
 }

modules/ai_engine_embedding/src/Service/EntityUpdate.php

@@ -2,20 +2,32 @@
 
 namespace Drupal\ai_engine_embedding\Service;
 
-use Drupal\ai_engine_feed\Service\Sources;
 use Drupal\Core\Config\ConfigFactoryInterface;
 use Drupal\Core\Entity\EntityInterface;
 use Drupal\Core\Entity\EntityPublishedInterface;
 use Drupal\Core\Http\ClientFactory;
 use Drupal\Core\Logger\LoggerChannelInterface;
+use Drupal\ai_engine_feed\Service\Sources;
 use Drupal\metatag\MetatagManager;
 
 /**
  * Service for updating the vector database as content is updated.
  */
 class EntityUpdate {
+  /**
+   * The default chunk size for sending data to the AI Embedding service.
+   *
+   * @var int
+   */
   const CHUNK_SIZE_DEFAULT = 3000;
 
+  /**
+   * The allowed entity types for indexing.
+   *
+   * @var array
+   */
+  const ALLOWED_ENTITIES = ['node', 'media'];
+
   /**
    * The configuration factory.
    *
@@ -150,36 +162,56 @@ public function delete(EntityInterface $entity) {
    * a cleanup routine to find and delete out of date chunks.
    */
   public function addAllDocuments() {
+    $docTypes = ['node' => 'text', 'media' => 'media'];
     $config = $this->configFactory->get('ai_engine_embedding.settings');
-    $data = $this->getData("upsert", $config, [], "");
-    $httpClient = $this->httpClientFactory->fromOptions([
-      'headers' => [
-        'Content-Type' => 'application/json',
-      ],
-    ]);
-    $endpoint = $config->get('azure_embedding_service_url') . '/api/upsert';
 
-    try {
-      $response = $httpClient->post($endpoint, ['json' => $data]);
+    // Loop through entityTypesToSend and send.
+    foreach (self::ALLOWED_ENTITIES as $entityType) {
+      $data = $this->getData("upsert", $config, ['entityType' => $entityType], "", $docTypes[$entityType]);
+      $endpoint = $config->get('azure_embedding_service_url') . '/api/upsert';
+      $response = $this->sendJsonPost($endpoint, $data);
 
       if ($response->getStatusCode() === 200) {
         $responseData = json_decode($response->getBody()->getContents(), TRUE);
         $this->logger->notice(
-          'Removed node @id from vector database. Service response: @response',
+          'Upserted node @id from vector database. Service response: @response',
           ['@response' => print_r($responseData, TRUE)]
         );
       }
       else {
         $this->logger->notice(
-          'Unable to remove node @id from vector database. POST failed with status code: @code',
+          'Unable to upsert node @id to vector database. POST failed with status code: @code',
           ['@code' => $response->getStatusCode()]
         );
         return NULL;
       }
     }
+  }
+
+  /**
+   * Sends a post request to an endpoint with data.
+   *
+   * @param string $endpoint
+   *   The endpoint to send the data to.
+   * @param array $data
+   *   The data to send.
+   *
+   * @return \Psr\Http\Message\ResponseInterface
+   *   The response from the post request.
+   */
+  protected function sendJsonPost($endpoint, $data) {
+    $httpClient = $this->httpClientFactory->fromOptions([
+      'headers' => [
+        'Content-Type' => 'application/json',
+      ],
+    ]);
+
+    try {
+      return $httpClient->post($endpoint, ['json' => $data]);
+    }
     catch (\Exception $e) {
       $this->logger->error(
-        'An error occurred while upserting document: @error',
+        'An error occurred while posting document: @error',
         ['@error' => $e->getMessage()]
       );
       return NULL;
@@ -198,41 +230,26 @@ public function addAllDocuments() {
    */
   public function upsertDocument(EntityInterface $entity) {
     $config = $this->configFactory->get('ai_engine_embedding.settings');
-    $chunk_size = $config->get('azure_chunk_size') || CHUNK_SIZE_DEFAULT;
+    $entityTypeId = $entity->getEntityTypeId();
     $route_params = [
-      'entityType' => $entity->getEntityTypeId(),
+      'entityType' => $entityTypeId,
       'id' => $entity->id(),
     ];
     $data = $this->getData("upsert", $config, $route_params, "");
-    $httpClient = $this->httpClientFactory->fromOptions([
-      'headers' => [
-        'Content-Type' => 'application/json',
-      ],
-    ]);
     $endpoint = $config->get('azure_embedding_service_url') . '/api/upsert';
+    $response = $this->sendJsonPost($endpoint, $data);
 
-    try {
-      $response = $httpClient->post($endpoint, ['json' => $data]);
-
-      if ($response->getStatusCode() === 200) {
-        $responseData = json_decode($response->getBody()->getContents(), TRUE);
-        $this->logger->notice(
-          'Removed node @id from vector database. Service response: @response',
-          ['@id' => $entity->id(), '@response' => print_r($responseData, TRUE)]
-        );
-      }
-      else {
-        $this->logger->notice(
-          'Unable to remove node @id from vector database. POST failed with status code: @code',
-          ['@id' => $entity->id(), '@code' => $response->getStatusCode()]
-        );
-        return NULL;
-      }
+    if ($response->getStatusCode() === 200) {
+      $responseData = json_decode($response->getBody()->getContents(), TRUE);
+      $this->logger->notice(
+        'Upserted node @id to vector database. Service response: @response',
+        ['@id' => $entity->id(), '@response' => print_r($responseData, TRUE)]
+      );
     }
-    catch (\Exception $e) {
-      $this->logger->error(
-        'An error occurred while upserting document: @error',
-        ['@error' => $e->getMessage()]
+    else {
+      $this->logger->notice(
+        'Unable to upsert node @id to vector database. POST failed with status code: @code',
+        ['@id' => $entity->id(), '@code' => $response->getStatusCode()]
       );
       return NULL;
     }
@@ -253,35 +270,20 @@ protected function removeDocument(EntityInterface $entity) {
       "id_list" => [],
       "id_filter_list" => [$this->sources->getSearchIndexId($entity)],
     ];
-    $httpClient = $this->httpClientFactory->fromOptions([
-      'headers' => [
-        'Content-Type' => 'application/json',
-      ],
-    ]);
     $endpoint = $config->get('azure_embedding_service_url') . '/api/deletebyid';
+    $response = $this->sendJsonPost($endpoint, $data);
 
-    try {
-      $response = $httpClient->post($endpoint, ['json' => $data]);
-
-      if ($response->getStatusCode() === 200) {
-        $responseData = json_decode($response->getBody()->getContents(), TRUE);
-        $this->logger->notice(
-          'Removed node @id from vector database. Service response: @response',
-          ['@id' => $entity->id(), '@response' => print_r($responseData, TRUE)]
-        );
-      }
-      else {
-        $this->logger->notice(
-          'Unable to remove node @id from vector database. POST failed with status code: @code',
-          ['@id' => $entity->id(), '@code' => $response->getStatusCode()]
-        );
-        return NULL;
-      }
+    if ($response->getStatusCode() === 200) {
+      $responseData = json_decode($response->getBody()->getContents(), TRUE);
+      $this->logger->notice(
+        'Removed node @id from vector database. Service response: @response',
+        ['@id' => $entity->id(), '@response' => print_r($responseData, TRUE)]
+      );
     }
-    catch (\Exception $e) {
-      $this->logger->error(
-        'An error occurred while deleting document: @error',
-        ['@error' => $e->getMessage()]
+    else {
+      $this->logger->notice(
+        'Unable to remove node @id from vector database. POST failed with status code: @code',
+        ['@id' => $entity->id(), '@code' => $response->getStatusCode()]
       );
       return NULL;
     }
@@ -330,7 +332,29 @@ protected function isIndexable(EntityInterface $entity) {
    *   TRUE if the entity should be embedded, FALSE otherwise.
    */
   protected function isSupportedEntityType(EntityInterface $entity) {
-    return $entity->getEntityTypeId() === 'node';
+    $entity_type_id = $entity->getEntityTypeId();
+
+    if ($entity_type_id == 'media') {
+      return $this->isSupportedMediaType($entity);
+    }
+    else {
+      return in_array($entity->getEntityTypeId(), self::ALLOWED_ENTITIES);
+    }
+  }
+
+  /**
+   * Checks if an entity is supported by the embedding system.
+   *
+   * @param \Drupal\Core\Entity\EntityInterface $entity
+   *   The entity to check.
+   *
+   * @return bool
+   *   TRUE if the entity should be embedded, FALSE otherwise.
+   */
+  protected function isSupportedMediaType(EntityInterface $entity) {
+    $config = $this->configFactory->get('ai_engine_embedding.settings');
+    $allowed_media_types = $config->get('included_media_types');
+    return in_array($entity->bundle(), $allowed_media_types);
   }
 
   /**
@@ -386,7 +410,7 @@ public function isIndexingEnabled(EntityInterface $entity) {
    * @return array
    *   An array of data to send to the AI Embedding service.
    */
-  protected function getData($action = 'upsert', $config, $route_params = [], $data = ""): array {
+  protected function getData($action = 'upsert', $config = NULL, $route_params = [], $data = "", $doctype = 'text'): array {
     $allowed_actions = ['upsert'];
     if (!$config) {
       throw new \Exception('Missing configuration object.');
@@ -396,7 +420,13 @@ protected function getData($action = 'upsert', $config, $route_params = [], $dat
       throw new \Exception('Invalid action provided.');
     }
 
-    $chunk_size = $config->get('azure_chunk_size') ?? CHUNK_SIZE_DEFAULT;
+    $allowed_doctypes = ['text', 'media'];
+
+    if (!in_array($doctype, $allowed_doctypes)) {
+      throw new \Exception('Invalid doctype provided.');
+    }
+
+    $chunk_size = $config->get('azure_chunk_size') ?? self::CHUNK_SIZE_DEFAULT;
 
     $data_endpoint = "";
     if ($data == "") {
@@ -405,7 +435,7 @@ protected function getData($action = 'upsert', $config, $route_params = [], $dat
 
     return [
       "action" => $action,
-      "doctype" => "text",
+      "doctype" => $doctype,
       "service_name" => $config->get('azure_search_service_name'),
       "index_name" => $config->get('azure_search_service_index'),
       "data" => $data,

modules/ai_engine_feed/ai_engine_feed.services.yml

@@ -2,4 +2,7 @@ services:
   # Service to query and prepare content for the feed.
   ai_engine_feed.sources:
     class: Drupal\ai_engine_feed\Service\Sources
-    arguments: ['@entity_type.manager', '@logger.channel.default', '@renderer', '@request_stack', '@ai_engine_metadata.manager', '@entity_field.manager', '@config.factory']
+    arguments: ['@entity_type.manager', '@logger.channel.default', '@renderer', '@request_stack', '@ai_engine_metadata.manager', '@entity_field.manager', '@config.factory', '@plugin.manager.ai_engine_feed.content_feed_manager']
+  plugin.manager.ai_engine_feed.content_feed_manager:
+    class: Drupal\ai_engine_feed\ContentFeedManager
+    parent: default_plugin_manager

modules/ai_engine_feed/src/Annotation/ContentFeedPlugin.php

@@ -0,0 +1,20 @@
+<?php
+
+namespace Drupal\ai_engine_feed\Annotation;
+
+use Drupal\Component\Annotation\Plugin;
+
+/**
+ * Defines a content feed plugin annotation object.
+ *
+ * @Annotation
+ */
+class ContentFeedPlugin extends Plugin {
+  /**
+   * The plugin ID.
+   *
+   * @var string
+   */
+  public $id;
+
+}