From 228db07a97ae9c58d1af57244efbf1f9a715a7f4 Mon Sep 17 00:00:00 2001
From: Robert Lemke <robert@flownative.com>
Date: Tue, 19 Feb 2019 16:32:50 +0100
Subject: [PATCH] Optimize memory usage for updateResourceMetadata

---
 Classes/Command/GcsCommandController.php | 61 ++++++++++++++++++------
 Classes/GcsTarget.php                    | 21 --------
 Classes/StorageFactory.php               |  2 +-
 3 files changed, 48 insertions(+), 36 deletions(-)

diff --git a/Classes/Command/GcsCommandController.php b/Classes/Command/GcsCommandController.php
index 28fc0c4..86b4224 100644
--- a/Classes/Command/GcsCommandController.php
+++ b/Classes/Command/GcsCommandController.php
@@ -11,11 +11,16 @@
  * source code.
  */
 
+use Doctrine\DBAL\FetchMode;
+use Doctrine\ORM\EntityManagerInterface;
 use Flownative\Google\CloudStorage\Exception;
 use Flownative\Google\CloudStorage\GcsTarget;
 use Flownative\Google\CloudStorage\StorageFactory;
+use Google\Cloud\Core\Exception\NotFoundException;
+use Google\Cloud\Core\Exception\ServiceException;
 use Neos\Flow\Annotations as Flow;
 use Neos\Flow\Cli\CommandController;
+use Neos\Flow\Persistence\PersistenceManagerInterface;
 use Neos\Flow\ResourceManagement\ResourceManager;
 use Neos\Flow\ResourceManagement\Storage\StorageObject;
 
@@ -38,6 +43,12 @@ final class GcsCommandController extends CommandController
      */
     protected $resourceManager;
 
+    /**
+     * @Flow\Inject(lazy=false)
+     * @var EntityManagerInterface
+     */
+    protected $entityManager;
+
     /**
      * Checks the connection
      *
@@ -46,7 +57,7 @@ final class GcsCommandController extends CommandController
      * @param string $bucket The bucket which is used for trying to upload and retrieve some test data
      * @return void
      */
-    public function connectCommand(string $bucket)
+    public function connectCommand(string $bucket): void
     {
         try {
             $storageClient = $this->storageFactory->create();
@@ -86,10 +97,10 @@ public function connectCommand(string $bucket)
      *
      * @param string $collection Name of the collection to publish
      */
-    public function republishCommand(string $collection = 'persistent')
+    public function republishCommand(string $collection = 'persistent'): void
     {
         $collectionName = $collection;
-        $collection = $this->resourceManager->getCollection($collection);
+        $collection = $this->resourceManager->getCollection($collectionName);
         if (!$collection) {
             $this->outputLine('<error>The collection %s does not exist.</error>', [$collectionName]);
             exit(1);
@@ -132,8 +143,9 @@ public function republishCommand(string $collection = 'persistent')
      * the same bucket.
      *
      * @param string $collection Name of the collection to publish
+     * @throws \Doctrine\DBAL\DBALException
      */
-    public function updateResourceMetadataCommand(string $collection = 'persistent')
+    public function updateResourceMetadataCommand(string $collection = 'persistent'): void
     {
         $collectionName = $collection;
         $collection = $this->resourceManager->getCollection($collection);
@@ -151,17 +163,38 @@ public function updateResourceMetadataCommand(string $collection = 'persistent')
         $this->outputLine();
         $this->outputLine('Updating metadata for resources in bucket %s ...', [$target->getBucketName()]);
         $this->outputLine();
+
+
         try {
-            foreach ($collection->getObjects() as $object) {
-                /** @var StorageObject $object */
-                $resource = $this->resourceManager->getResourceBySha1($object->getSha1());
-                if ($resource) {
-                    try {
-                        $target->updateResourceMetadata($resource);
-                        $this->outputLine('   ✅  %s %s ', [$resource->getSha1(), $resource->getFilename()]);
-                    } catch (Exception $exception) {
-                        $this->outputLine('   ❌  %s <error>%s</error>', [$resource->getSha1(), $exception->getMessage()]);
-                    }
+            $storageClient = $this->storageFactory->create();
+        } catch (\Exception $e) {
+            $this->outputLine('<error>%s</error>', [$e->getMessage()]);
+            exit(1);
+        }
+
+        $targetBucket = $storageClient->bucket($target->getBucketName());
+        $targetKeyPrefix = $target->getKeyPrefix();
+        $queryResult = $this->entityManager->getConnection()->executeQuery(
+            'SELECT sha1, filename, mediatype FROM neos_flow_resourcemanagement_persistentresource AS r WHERE collectionname = :collectionName ORDER BY sha1',
+            [
+                'collectionName' => $collectionName
+            ]
+        );
+
+        try {
+            $previousSha1 = null;
+            while ($resourceRecord = $queryResult->fetch(FetchMode::ASSOCIATIVE)) {
+                if ($resourceRecord['sha1'] === $previousSha1) {
+                    continue;
+                }
+                $previousSha1 = $resourceRecord['sha1'];
+
+                try {
+                    $object = $targetBucket->object($targetKeyPrefix . $resourceRecord['sha1']);
+                    $object->update(['contentType' => $resourceRecord['mediatype']]);
+                    $this->outputLine('   ✅  %s %s ', [$resourceRecord['sha1'], $resourceRecord['filename']]);
+                } catch (ServiceException | NotFoundException $exception) {
+                    $this->outputLine('   ❌  <error>%s %s</error>', [$resourceRecord['sha1'], $resourceRecord['filename']]);
                 }
             }
         } catch (\Exception $e) {
diff --git a/Classes/GcsTarget.php b/Classes/GcsTarget.php
index 4afba34..f929f85 100644
--- a/Classes/GcsTarget.php
+++ b/Classes/GcsTarget.php
@@ -424,27 +424,6 @@ public function publishResource(PersistentResource $resource, CollectionInterfac
         }
     }
 
-    /**
-     * Updates the metadata (currently content type) of a resource object already stored in this target
-     *
-     * @param PersistentResource $resource
-     * @throws \Flownative\Google\CloudStorage\Exception
-     */
-    public function updateResourceMetadata(PersistentResource $resource)
-    {
-        try {
-            $targetBucket = $this->storageClient->bucket($this->bucketName);
-        } catch (NotFoundException $exception) {
-            throw new \Flownative\Google\CloudStorage\Exception(sprintf('Failed retrieving bucket information for "%s".', $this->bucketName), 1538462744);
-        }
-        try {
-            $object = $targetBucket->object($this->getKeyPrefix() . $resource->getSha1());
-            $object->update(['contentType' => $resource->getMediaType()]);
-        } catch (ServiceException | NotFoundException $exception) {
-            throw new \Flownative\Google\CloudStorage\Exception(sprintf('Resource "%s" (%s) not found in bucket %s.', $resource->getSha1(), $resource->getFilename(), $this->bucketName), 1538462744);
-        }
-    }
-
     /**
      * Unpublishes the given persistent resource
      *
diff --git a/Classes/StorageFactory.php b/Classes/StorageFactory.php
index 9a0d32d..bb83991 100644
--- a/Classes/StorageFactory.php
+++ b/Classes/StorageFactory.php
@@ -34,7 +34,7 @@ class StorageFactory
     protected $environment;
 
     /**
-     * Creates a new Storage instance and authenticates agains the Google API
+     * Creates a new Storage instance and authenticates against the Google API
      *
      * @param string $credentialsProfileName
      * @return \Google\Cloud\Storage\StorageClient