Added method to override the bulk request size in ProviderInterface

Author: pmishev

Document/Provider/AbstractProvider.php

@@ -9,7 +9,6 @@
  */
 abstract class AbstractProvider implements ProviderInterface
 {
-
     /**
      * @var string The type the provider is for
      */
@@ -48,4 +47,15 @@ abstract public function getDocuments();
      * @return DocumentInterface|array
      */
     abstract public function getDocument($id);
+
+    /**
+     * Returns the number of Elasticsearch documents to persist in a single bulk request
+     * If null is returned, the 'bulk_batch_size' of the Connection will be used
+     *
+     * @return int|null
+     */
+    public function getPersistRequestBatchSize() : ?int
+    {
+        return null;
+    }
 }

Document/Provider/ProviderInterface.php

@@ -34,4 +34,12 @@ public function getDocuments();
      * @return DocumentInterface|array
      */
     public function getDocument($id);
+
+    /**
+     * Returns the number of Elasticsearch documents to persist in a single bulk request
+     * If null is returned, the 'bulk_batch_size' of the Connection will be used
+     *
+     * @return int|null
+     */
+    public function getPersistRequestBatchSize() : ?int;
 }

Manager/IndexManager.php

@@ -687,15 +687,15 @@ protected function createNewIndexWithUniqueName()
      */
     protected function copyDataToNewIndex(string $newIndex, string $oldIndex)
     {
-        $batchSize = $this->connection->getConnectionSettings()['bulk_batch_size'];
-
         // Make sure we don't autocommit on every item in the bulk request
         $autocommit = $this->getConnection()->isAutocommit();
         $this->getConnection()->setAutocommit(false);
 
-        $typeDataProvider = $this->getDataProvider();
+        $indexDataProvider = $this->getDataProvider();
+        $batchSize = $indexDataProvider->getPersistRequestBatchSize() ?? $this->connection->getConnectionSettings()['bulk_batch_size'];
+
         $i = 1;
-        foreach ($typeDataProvider->getDocuments() as $document) {
+        foreach ($indexDataProvider->getDocuments() as $document) {
             // Temporarily override the write alias with the new physical index name, so rebuilding only happens in the new index
             $originalWriteAlias = $this->writeAlias;
             $this->setWriteAlias($newIndex);