Merge pull request #37 from gowrizrh/feature/index-large-count

Index large amount of event records

Model/Indexer/AsyncEventSubscriber.php

@@ -4,21 +4,20 @@
 
 namespace MageOS\AsyncEvents\Model\Indexer;
 
-use Magento\Elasticsearch\Model\Adapter\ElasticsearchFactory;
-use MageOS\AsyncEvents\Helper\Config;
-use MageOS\AsyncEvents\Model\Adapter\BatchDataMapper\AsyncEventLogMapper;
-use MageOS\AsyncEvents\Model\Indexer\DataProvider\AsyncEventSubscriberLogs;
-use MageOS\AsyncEvents\Model\Resolver\AsyncEvent;
 use ArrayIterator;
 use Magento\CatalogSearch\Model\Indexer\IndexerHandlerFactory;
+use Magento\Elasticsearch\Model\Adapter\ElasticsearchFactory;
 use Magento\Framework\App\DeploymentConfig;
 use Magento\Framework\App\ObjectManager;
 use Magento\Framework\Indexer\ActionInterface as IndexerActionInterface;
 use Magento\Framework\Indexer\DimensionalIndexerInterface;
 use Magento\Framework\Indexer\DimensionProviderInterface;
-use Magento\Framework\Indexer\IndexStructureInterface;
 use Magento\Framework\Indexer\SaveHandler\IndexerInterface;
 use Magento\Framework\Mview\ActionInterface as MviewActionInterface;
+use MageOS\AsyncEvents\Helper\Config;
+use MageOS\AsyncEvents\Model\Adapter\BatchDataMapper\AsyncEventLogMapper;
+use MageOS\AsyncEvents\Model\Indexer\DataProvider\AsyncEventSubscriberLogs;
+use MageOS\AsyncEvents\Model\Resolver\AsyncEvent;
 use Traversable;
 
 class AsyncEventSubscriber implements
@@ -29,7 +28,7 @@ class AsyncEventSubscriber implements
     /**
      * Indexer ID in configuration
      */
-    private const INDEXER_ID = 'asynchronous_event_subscriber_log';
+    public const INDEXER_ID = 'asynchronous_event_subscriber_log';
 
     /**
      * Default batch size
@@ -71,10 +70,10 @@ public function __construct(
         private readonly IndexerHandlerFactory $indexerHandlerFactory,
         private readonly AsyncEventSubscriberLogs $asyncEventSubscriberLogsDataProvider,
         private readonly AsyncEvent $asyncEventScopeResolver,
-        private readonly IndexStructureInterface $indexStructure,
         private readonly Config $config,
         private readonly ElasticsearchFactory $adapterFactory,
         private readonly AsyncEventLogMapper $loggerMapper,
+        private readonly IndexStructureFactory $indexStructureFactory,
         private readonly array $data,
         int $batchSize = null,
         DeploymentConfig $deploymentConfig = null
@@ -106,21 +105,27 @@ public function executeByDimensions(array $dimensions, ?Traversable $entityIds)
             'batchDocumentDataMapper' => $this->loggerMapper
         ]);
 
+        $indexStructure = $this->indexStructureFactory->create([
+            'adapter' => $adapter
+        ]);
+
         $saveHandler = $this->indexerHandlerFactory->create(
             [
                 'data' => $this->data,
                 'adapter' => $adapter,
                 'scopeResolver' => $this->asyncEventScopeResolver,
-                'indexStructure' => $this->indexStructure
+                'indexStructure' => $indexStructure
             ]
         );
 
         if ($entityIds === null) {
             $asyncEventDimension = $dimensions[AsyncEventDimensionProvider::DIMENSION_NAME]->getValue();
+
+
             $saveHandler->cleanIndex($dimensions);
             $saveHandler->saveIndex(
                 $dimensions,
-                $this->asyncEventSubscriberLogsDataProvider->getAsyncEventLogs($asyncEventDimension, null)
+                $this->asyncEventSubscriberLogsDataProvider->rebuildIndex($asyncEventDimension)
             );
         } else {
             $asyncEventIds = iterator_to_array($entityIds);

Model/Indexer/DataProvider/AsyncEventSubscriberLogs.php

@@ -4,17 +4,32 @@
 
 namespace MageOS\AsyncEvents\Model\Indexer\DataProvider;
 
+use Generator;
+use Magento\Framework\App\DeploymentConfig;
+use Magento\Framework\App\ResourceConnection;
+use Magento\Framework\DB\Adapter\AdapterInterface;
+use MageOS\AsyncEvents\Model\Indexer\AsyncEventSubscriber;
 use MageOS\AsyncEvents\Model\ResourceModel\AsyncEventLog\Collection;
 use MageOS\AsyncEvents\Model\ResourceModel\AsyncEventLog\CollectionFactory as AsyncEventLogCollectionFactory;
+use Zend_Db_Expr;
 
 class AsyncEventSubscriberLogs
 {
+    private AdapterInterface $connection;
+
+    private const DEPLOYMENT_CONFIG_INDEXER_BATCHES_PREFIX = 'indexer/batch_size/';
+
     /**
      * @param AsyncEventLogCollectionFactory $collectionFactory
+     * @param ResourceConnection $resource
+     * @param DeploymentConfig $deploymentConfig
      */
     public function __construct(
-        private readonly AsyncEventLogCollectionFactory $collectionFactory
+        private readonly AsyncEventLogCollectionFactory $collectionFactory,
+        private readonly ResourceConnection $resource,
+        private readonly DeploymentConfig $deploymentConfig
     ) {
+        $this->connection = $resource->getConnection();
     }
 
     /**
@@ -42,4 +57,35 @@ public function getAsyncEventLogs(string $asyncEvent, ?array $logIds): Collectio
 
         return $logCollection;
     }
+
+    public function rebuildIndex(string $asyncEvent): Generator
+    {
+        $tableName = $this->resource->getTableName('async_event_subscriber_log');
+
+        $batchSize = $this->deploymentConfig->get(
+            self::DEPLOYMENT_CONFIG_INDEXER_BATCHES_PREFIX . AsyncEventSubscriber::INDEXER_ID . '/mysql_get'
+        ) ?? 10_000;
+
+        $lastId = 0;
+
+        while (true) {
+            $select = $this->connection->select()
+                ->from($tableName, ['*', new Zend_Db_Expr("'$asyncEvent' AS event_name")])
+                ->where('log_id > ?', $lastId)
+                ->order('log_id ASC')
+                ->limit($batchSize);
+
+            $result = $this->connection->fetchAll($select);
+
+            if (empty($result)) {
+                break;
+            }
+
+            foreach ($result as $row) {
+                yield $row['log_id'] => $row;
+            }
+
+            $lastId = end($result)['log_id'];
+        }
+    }
 }