Clustering: Cap clustering batch size at 10k

Signed-off-by: Marcel Klehr <[email protected]>
nextcloud · Feb 9, 2023 · 75e1641 · 75e1641
1 parent c4ba824
commit 75e1641
Show file tree

Hide file tree

Showing 2 changed files with 7 additions and 2 deletions.
diff --git a/lib/BackgroundJobs/ClusterFacesJob.php b/lib/BackgroundJobs/ClusterFacesJob.php
@@ -19,6 +19,7 @@ class ClusterFacesJob extends QueuedJob {
 	private IJobList $jobList;
 	private LoggerInterface $logger;
 
+	public const BATCH_SIZE = 10000;
 	public function __construct(ITimeFactory $time, Logger $logger, IJobList $jobList, FaceClusterAnalyzer $clusterAnalyzer) {
 		parent::__construct($time);
 		$this->logger = $logger;
@@ -35,7 +36,7 @@ protected function run($argument) {
 		/** @var string $userId */
 		$userId = $argument['userId'];
 		try {
-			$this->clusterAnalyzer->calculateClusters($userId);
+			$this->clusterAnalyzer->calculateClusters($userId, self::BATCH_SIZE);
 		} catch (\JsonException|Exception $e) {
 			$this->logger->error('Failed to calculate face clusters', ['exception' => $e]);
 		}

diff --git a/lib/Service/FaceClusterAnalyzer.php b/lib/Service/FaceClusterAnalyzer.php
@@ -36,7 +36,7 @@ public function __construct(FaceDetectionMapper $faceDetections, FaceClusterMapp
 	 * @throws \OCP\DB\Exception
 	 * @throws \JsonException
 	 */
-	public function calculateClusters(string $userId): void {
+	public function calculateClusters(string $userId, int $batchSize = -1): void {
 		$this->logger->debug('ClusterDebug: Retrieving face detections for user ' . $userId);
 
 		$unclusteredDetections = $this->faceDetections->findUnclusteredByUserId($userId);
@@ -50,6 +50,10 @@ public function calculateClusters(string $userId): void {
 			return;
 		}
 
+		if ($batchSize > 0 && count($unclusteredDetections) > $batchSize) {
+			$unclusteredDetections = array_slice($unclusteredDetections, 0, $batchSize);
+		}
+
 		$this->logger->debug('ClusterDebug: Found ' . count($unclusteredDetections) . " unclustered detections. Calculating clusters.");
 
 		$sampledDetections = [];