#3020911

README.md

@@ -1,12 +1,12 @@
 # Search API Elasticsearch Attachments
 [![CircleCI](https://circleci.com/gh/dakkusingh/search_api_elasticsearch_attachments.svg?style=svg)](https://circleci.com/gh/dakkusingh/search_api_elasticsearch_attachments)
 
-Elasticsearch is generally used to index data of types like string, 
-number, date, etc. 
-However, what if you wanted to index a file like a .pdf or a .doc 
+Elasticsearch is generally used to index data of types like string,
+number, date, etc.
+However, what if you wanted to index a file like a .pdf or a .doc
 directly and make it searchable?
 
-This module allows Drupal to index files (attachments) to Elasticsearch by 
+This module allows Drupal to index files (attachments) to Elasticsearch by
 making use of Elasticsearch data type "attachment".
 
 ![Search_API_Elasticsearch_Attachments](https://www.drupal.org/files/search_api_elasticsearch_attachments.jpg)
@@ -20,9 +20,9 @@ This module requires:
 * Elasticsearch `ingest-attachment` plugin
 
 ## Elasticsearch Plugin Installation
-The first step is to install the Elasticsearch plugin: `ingest-attachment`, 
-which enables ES to recognise the "attachment" data type. In turn, it uses 
-Apache Tika for content extraction and supports several file types such as 
+The first step is to install the Elasticsearch plugin: `ingest-attachment`,
+which enables ES to recognise the "attachment" data type. In turn, it uses
+Apache Tika for content extraction and supports several file types such as
 .pdf, .doc, .xls, .rtf, .html, .odt, etc.
 
 ```
@@ -41,7 +41,7 @@ You have to choose the correct versions of the module depending on your
 Elastic Search Server setup. Please see the table below for
 compatibility.
 
-If you are using Elasticsearch Connector 8.x-5.x, 
+If you are using Elasticsearch Connector 8.x-5.x,
 please use 8.x-5.x of
 *search_api_elasticsearch_attachments* module.
 

composer.json

@@ -1,9 +1,16 @@
 {
-  "name": "drupal/search_api_elasticsearch_attachments",
-  "description": "Search API Elasticsearch Attachments",
-  "type": "drupal-module",
-  "minimum-stability": "dev",
-  "require": {
-    "drupal/elasticsearch_connector": "5.0-alpha3"
-  }
+    "name": "drupal/search_api_elasticsearch_attachments",
+    "description": "Search API Elasticsearch Attachments",
+    "type": "drupal-module",
+    "minimum-stability": "dev",
+    "require": {
+        "drupal/elasticsearch_connector": "5.0-alpha3"
+    },
+    "extra": {
+        "patches": {
+            "drupal/elasticsearch_connector": {
+                "Issue #2918138": "https://www.drupal.org/files/issues/2018-12-14/elasticsearch_connector-alter_params-2918138-5.patch"
+            }
+        }
+    }
 }

search_api_elasticsearch_attachments.services.yml

@@ -1,16 +1,4 @@
 services:
-  search_api_elasticsearch_attachments.prepare_index:
-    class: Drupal\search_api_elasticsearch_attachments\EventSubscriber\PrepareIndex
-    tags:
-      - { name: event_subscriber }
-  search_api_elasticsearch_attachments.prepare_query:
-    class: Drupal\search_api_elasticsearch_attachments\EventSubscriber\PrepareQuery
-    tags:
-      - { name: event_subscriber }
-  search_api_elasticsearch_attachments.prepare_mapping:
-    class: Drupal\search_api_elasticsearch_attachments\EventSubscriber\PrepareMapping
-    tags:
-      - { name: event_subscriber }
   search_api_elasticsearch_attachments.build_query:
     class: Drupal\search_api_elasticsearch_attachments\EventSubscriber\BuildSearchParams
     tags:
@@ -19,3 +7,8 @@ services:
     class: Drupal\search_api_elasticsearch_attachments\EventSubscriber\PrepareIndexMapping
     tags:
       - { name: event_subscriber }
+  search_api_elasticsearch_attachments.build_index_params:
+    class: Drupal\search_api_elasticsearch_attachments\EventSubscriber\BuildIndexParams
+    arguments: ['@elasticsearch_connector.client_manager', '@entity_type.manager']
+    tags:
+      - { name: event_subscriber }

src/EventSubscriber/BuildIndexParams.php

@@ -0,0 +1,188 @@
+<?php
+
+namespace Drupal\search_api_elasticsearch_attachments\EventSubscriber;
+
+use Drupal\elasticsearch_connector\Event\BuildIndexParamsEvent;
+use Symfony\Component\EventDispatcher\EventSubscriberInterface;
+use Drupal\search_api\Entity\Index;
+use Drupal\search_api_elasticsearch_attachments\Helpers;
+use Drupal\elasticsearch_connector\ElasticSearch\ClientManagerInterface;
+use Drupal\Core\Entity\EntityTypeManagerInterface;
+use Drupal\search_api\IndexInterface;
+use Elasticsearch\Common\Exceptions\Missing404Exception;
+
+/**
+ * {@inheritdoc}
+ */
+class BuildIndexParams implements EventSubscriberInterface {
+
+  protected $pipelineName = 'es_attachment';
+  protected $targetFieldId = 'es_attachment';
+
+  /**
+   * {@inheritdoc}
+   */
+  public function __construct(ClientManagerInterface $client_manager, EntityTypeManagerInterface $entity_type_manager) {
+    $this->clientManager = $client_manager;
+    $this->entityTypeManager = $entity_type_manager;
+  }
+
+  /**
+   * {@inheritdoc}
+   */
+  public static function getSubscribedEvents() {
+    $events[BuildIndexParamsEvent::BUILD_PARAMS][] = ['indexParams', 100];
+    $events[BuildIndexParamsEvent::BUILD_PARAMS][] = ['pipelineProcessing', 101];
+    return $events;
+  }
+
+  /**
+   * Method to build Params.
+   *
+   * @param \Drupal\elasticsearch_connector\Event\BuildIndexParamsEvent $event
+   *   The BuildIndexParamsEvent event.
+   */
+  public function indexParams(BuildIndexParamsEvent $event) {
+    // We need to react only on our processor.
+    $indexName = $this->getIndexName($event);
+    $processors = $this->getIndexProcessors($indexName);
+    // Add pipeline param.
+    if (!empty($processors['elasticsearch_attachments'])) {
+      $params = $event->getElasticIndexParams();
+      // Add pipeline param for attachment processing.
+      $params['pipeline'] = $this->pipelineName;
+      // Set updated params array.
+      $event->setElasticIndexParams($params);
+    }
+  }
+
+  /**
+   * Valdiate pipeline. Create new one or delete existing.
+   *
+   * @param \Drupal\elasticsearch_connector\Event\BuildIndexParamsEvent $event
+   *   The BuildIndexParamsEvent event.
+   */
+  public function pipelineProcessing(BuildIndexParamsEvent $event) {
+    // Get incex name and list of available processors.
+    $indexName = $this->getIndexName($event);
+    $processors = $this->getIndexProcessors($indexName);
+    /** @var \Drupal\search_api\IndexInterface $index */
+    $index = $this->getIndex($indexName);
+    // Initialize client to work with.
+    $this->initializeClient($index);
+    // Pipeline registration.
+    if (!empty($processors['elasticsearch_attachments'])) {
+      // If there is no pipeline yet, Elastic will return Missing404Exception.
+      // There is no other way to check if pipeline exist.
+      try {
+        $this->getPipeline();
+      }
+      catch (Missing404Exception $e) {
+        $this->putPipeline();
+      }
+    }
+    else {
+      // If there is no pipeline yet, Elastic will return Missing404Exception.
+      // There is no other way to check if pipeline exist.
+      try {
+        $this->getPipeline();
+        $this->deletePipeline();
+      }
+      catch (Missing404Exception $e) {
+        // Nothing to do here.
+      }
+    }
+  }
+
+  /**
+   * Get index name.
+   *
+   * @param \Drupal\elasticsearch_connector\Event\BuildIndexParamsEvent $event
+   *   The BuildIndexParamsEvent event.
+   *
+   * @return string
+   *   Index name
+   */
+  public function getIndexName(BuildIndexParamsEvent $event) {
+    return Helpers::getIndexName($event->getIndexName());
+  }
+
+  /**
+   * Get list of all available index processors.
+   *
+   * @param string $indexName
+   *   Name of index.
+   *
+   * @return array
+   *   List of all available processors.
+   */
+  public function getIndexProcessors($indexName) {
+    return Index::load($indexName)->getProcessors();
+  }
+
+  /**
+   * Get list of all available index processors.
+   *
+   * @param string $indexName
+   *   Name of index.
+   *
+   * @return \Drupal\search_api\IndexInterface
+   *   Index object.
+   */
+  public function getIndex($indexName) {
+    return Index::load($indexName);
+  }
+
+  /**
+   * ElasticSearch client initialization.
+   *
+   * @param \Drupal\search_api\IndexInterface $index
+   *   The index scheduled for indexing.
+   */
+  public function initializeClient(IndexInterface $index) {
+    $cluster_name = $index->getServerInstance()->getBackend()->getCluster();
+    $cluster = $this->entityTypeManager->getStorage('elasticsearch_cluster')->load($cluster_name);
+    $this->client = $this->clientManager->getClientForCluster($cluster);
+  }
+
+  /**
+   * Helper to register new pipeline.
+   */
+  public function putPipeline() {
+    $params = [];
+    $params['id'] = $this->pipelineName;
+    $params['body'] = [
+      'description' => 'Extract attachment information from arrays',
+      'processors' => [
+        [
+          'foreach' => [
+            'field' => $this->targetFieldId,
+            'ignore_failure' => TRUE,
+            'processor' => [
+              'attachment' => [
+                'target_field' => '_ingest._value.attachment',
+                'field' => '_ingest._value.data',
+              ],
+            ],
+          ],
+        ],
+      ],
+    ];
+    $this->client->ingest()->putPipeline($params);
+  }
+
+  /**
+   * Helper to delete exiting pipeline.
+   */
+  public function deletePipeline() {
+    $this->client->ingest()->deletePipeline(['id' => $this->pipelineName]);
+  }
+
+  /**
+   * Helper to get exiting pipeline.
+   */
+  public function getPipeline() {
+    return $this->client->ingest()->getPipeline(['id' => $this->pipelineName]);
+  }
+
+}

src/EventSubscriber/BuildSearchParams.php

@@ -28,26 +28,59 @@ public static function getSubscribedEvents() {
    */
   public function searchParams(BuildSearchParamsEvent $event) {
     $params = $event->getElasticSearchParams();
-
-    // Default Prefix and Suffix.
-    $prefix = '<strong>';
-    $suffix = '</strong>';
-
-    // We need to get the Prefix and Suffix from processor.
+    // Set default boost.
+    $boost = 1.0;
+    // We need to get the processor.
     $indexName = Helpers::getIndexName($event->getIndexName());
     $processors = Index::load($indexName)->getProcessors();
-
+    // Try to load boost value from config form.
+    if (!empty($processors['elasticsearch_attachments'])) {
+      $boost = $processors['elasticsearch_attachments']->getConfiguration()['boost'];
+      // Get original query.
+      $originalBoolQuery = $params['body']['query']['bool']['must'];
+      // Get query string.
+      if (isset($originalBoolQuery['query_string'])) {
+        $queryString = $originalBoolQuery['query_string']['query'];
+        // Build nestedQuery.
+        // @see https://www.elastic.co/guide/en/elasticsearch/guide/current/nested-query.html.
+        $nestedQuery = [
+          'nested' => [
+            'path' => 'es_attachment',
+            'query' => [
+              'bool' => [
+                'must' => [
+                  'query_string' => [
+                    'query' => $queryString,
+                    'fields' => [
+                      'es_attachment.attachment.content^' . $boost,
+                    ],
+                  ],
+                ],
+              ],
+            ],
+          ],
+        ];
+        // We need to change the bool query from must to should.
+        // This is requried to add support for nested and string queries.
+        $params['body']['query']['bool']['should'] = [];
+        unset($params['body']['query']['bool']['must']);
+        $params['body']['query']['bool']['should'][] = $originalBoolQuery;
+        $params['body']['query']['bool']['should'][] = $nestedQuery;
+        // Add min match param.
+        $params['body']['query']['bool']['minimum_should_match'] = 1;
+      }
+    }
+    // Add highlight if enabled.
     if (!empty($processors['elasticsearch_attachments_highlight'])) {
       $processorConf = $processors['elasticsearch_attachments_highlight']->getConfiguration();
       $prefix = $processorConf['prefix'];
       $suffix = $processorConf['suffix'];
+      // See: https://github.com/elastic/elasticsearch-php/issues/394
+      $params['body']['highlight']['fields']['es_attachment.attachment.content'] = (object) [];
+      $params['body']['highlight']['pre_tags'] = [$prefix];
+      $params['body']['highlight']['post_tags'] = [$suffix];
     }
-
-    // See: https://github.com/elastic/elasticsearch-php/issues/394
-    $params['body']['highlight']['fields']['es_attachment.content'] = (object) [];
-    $params['body']['highlight']['pre_tags'] = [$prefix];
-    $params['body']['highlight']['post_tags'] = [$suffix];
-
+    // Set updated params array.
     $event->setElasticSearchParams($params);
   }