From c49cd5438655043b487f4d986961746ae58edb3a Mon Sep 17 00:00:00 2001
From: Pim Jansen <pimjansen@gmail.com>
Date: Tue, 27 Aug 2019 16:43:06 +0200
Subject: [PATCH] Added batch inserts for doctrine orm populate

---
 readme.md                            |  2 ++
 src/Faker/ORM/Doctrine/Populator.php | 34 +++++++++++++++++++++++++---
 2 files changed, 33 insertions(+), 3 deletions(-)

diff --git a/readme.md b/readme.md
index 7c8148dfd6..6790c5ea21 100644
--- a/readme.md
+++ b/readme.md
@@ -452,6 +452,8 @@ print_r($insertedPKs);
 // )
 ```
 
+**Note:** Due to the fact that `Faker` returns all the primary keys inserted, the memory consumption will go up drastically when you do batch inserts due to the big list of data.
+
 In the previous example, the `Book` and `Author` models share a relationship. Since `Author` entities are populated first, Faker is smart enough to relate the populated `Book` entities to one of the populated `Author` entities.
 
 Lastly, if you want to execute an arbitrary function on an entity before insertion, use the fourth argument of the `addEntity()` method:
diff --git a/src/Faker/ORM/Doctrine/Populator.php b/src/Faker/ORM/Doctrine/Populator.php
index d4fe897c61..d4c5dfb879 100644
--- a/src/Faker/ORM/Doctrine/Populator.php
+++ b/src/Faker/ORM/Doctrine/Populator.php
@@ -3,6 +3,7 @@
 namespace Faker\ORM\Doctrine;
 
 use Doctrine\Common\Persistence\ObjectManager;
+use Faker\Generator;
 
 /**
  * Service class for populating a database using the Doctrine ORM or ODM.
@@ -10,20 +11,35 @@
  */
 class Populator
 {
+    /** @var int  */
+    protected $batchSize;
+
+    /** @var Generator  */
     protected $generator;
+
+    /** @var ObjectManager|null  */
     protected $manager;
+
+    /** @var array  */
     protected $entities = array();
+
+    /** @var array  */
     protected $quantities = array();
+
+    /** @var array  */
     protected $generateId = array();
 
     /**
-     * @param \Faker\Generator $generator
+     * Populator constructor.
+     * @param Generator $generator
      * @param ObjectManager|null $manager
+     * @param int $batchSize
      */
-    public function __construct(\Faker\Generator $generator, ObjectManager $manager = null)
+    public function __construct(Generator $generator, ObjectManager $manager = null, $batchSize = 1000)
     {
         $this->generator = $generator;
         $this->manager = $manager;
+        $this->batchSize = $batchSize;
     }
 
     /**
@@ -55,6 +71,9 @@ public function addEntity($entity, $number, $customColumnFormatters = array(), $
     /**
      * Populate the database using all the Entity classes previously added.
      *
+     * Please note that large amounts of data will result in more memory usage since the the Populator will return
+     * all newly created primary keys after executing.
+     *
      * @param null|EntityManager $entityManager A Doctrine connection object
      *
      * @return array A list of the inserted PKs
@@ -72,9 +91,18 @@ public function execute($entityManager = null)
         foreach ($this->quantities as $class => $number) {
             $generateId = $this->generateId[$class];
             for ($i=0; $i < $number; $i++) {
-                $insertedEntities[$class][]= $this->entities[$class]->execute($entityManager, $insertedEntities, $generateId);
+                $insertedEntities[$class][]= $this->entities[$class]->execute(
+                    $entityManager,
+                    $insertedEntities,
+                    $generateId
+                );
+                if (count($insertedEntities) % $this->batchSize === 0) {
+                    $entityManager->flush();
+                    $entityManager->clear($class);
+                }
             }
             $entityManager->flush();
+            $entityManager->clear($class);
         }
 
         return $insertedEntities;