Get face descriptors only when needed in batches.

In my tests with 37128 faces and a batch size of 5000 it improves
memory consumption by 38,81%.

I guess we can celebrate a little. 🎉 😉

lib/BackgroundJob/Tasks/CreateClustersTask.php

@@ -167,14 +167,20 @@ private function createClusterIfNeeded(string $userId) {
 			$sliceSize = ceil($facesCount / $noSlices);
 		}
 
-		$this->logDebug('We will cluster with ' . $noSlices . ' batch(es) of ' . $sliceSize . ' faces');
+		$this->logDebug('We will cluster these with ' . $noSlices . ' batch(es) of ' . $sliceSize . ' faces.');
 
 		$newClusters = [];
-
 		// Obtain the clusters in batches and append them.
 		for ($i = 0; $i < $noSlices ; $i++) {
+			// Get the batches.
 			$facesSliced = array_slice($faces, $i * $sliceSize, $sliceSize);
-			$newClusters = array_merge($newClusters, $this->getNewClusters($facesSliced));
+			// Get the indices, obtain the partial clusters and incorporate them.
+			$faceIds = array_map(function ($face) { return $face->getId(); }, $facesSliced);
+			$facesDescripted = $this->faceMapper->findDescriptorsBathed($faceIds);
+			$newClusters = array_merge($newClusters, $this->getNewClusters($facesDescripted));
+			// Discard variables aggressively to improve memory consumption.
+			unset($facesDescripted);
+			unset($facesSliced);
 		}
 
 		// Append non groupable faces on a single step.
@@ -184,7 +190,7 @@ private function createClusterIfNeeded(string $userId) {
 		// Value is array of face IDs. For old clusters, person IDs are some existing person IDs,
 		// and for new clusters is whatever chinese whispers decides to identify them.
 		//
-		$currentClusters = $this->getCurrentClusters($faces);
+		$currentClusters = $this->getCurrentClusters(array_merge($faces, $nonGroupables));
 
 		$this->logInfo(count($newClusters) . ' clusters found after clustering');
 

lib/Db/FaceMapper.php

@@ -41,14 +41,23 @@ public function find (int $faceId): ?Face {
 		$qb = $this->db->getQueryBuilder();
 		$qb->select('id', 'image', 'person', 'x', 'y', 'width', 'height', 'landmarks', 'descriptor', 'confidence')
 			->from($this->getTableName(), 'f')
-			->andWhere($qb->expr()->eq('id', $qb->createNamedParameter($faceId)));
+			->where($qb->expr()->eq('id', $qb->createNamedParameter($faceId)));
 		try {
 			return $this->findEntity($qb);
 		} catch (DoesNotExistException $e) {
 			return null;
 		}
 	}
 
+	public function findDescriptorsBathed (array $faceIds): array {
+		$qb = $this->db->getQueryBuilder();
+		$qb->select('id', 'descriptor')
+			->from($this->getTableName(), 'f')
+			->where($qb->expr()->in('id', $qb->createParameter('face_ids')));
+		$qb->setParameter('face_ids', $faceIds, IQueryBuilder::PARAM_INT_ARRAY);
+		return $this->findEntities($qb);
+	}
+
 	/**
 	 * Based on a given fileId, takes all faces that belong to that file
 	 * and return an array with that.
@@ -147,7 +156,7 @@ public function getFaces(string $userId, int $model): array {
 
 	public function getGroupableFaces(string $userId, int $model, int $minSize, float $minConfidence): array {
 		$qb = $this->db->getQueryBuilder();
-		$qb->select('f.id', 'f.person', 'f.descriptor')
+		$qb->select('f.id', 'f.person')
 			->from($this->getTableName(), 'f')
 			->innerJoin('f', 'facerecog_images' ,'i', $qb->expr()->eq('f.image', 'i.id'))
 			->where($qb->expr()->eq('user', $qb->createParameter('user')))