From 40f16cf7b1319c212f95c4638c8be8efb8f15588 Mon Sep 17 00:00:00 2001
From: Christoph Reiter <reiter.christoph@gmail.com>
Date: Wed, 21 Aug 2024 16:10:55 +0200
Subject: [PATCH] typense: process document conversion in chunks to reduce
 memory

Also delete all no longer active collections, to reduce memory usage
of typensense too, before and after a full sync.
---
 src/TypesenseClient/SearchIndex.php | 38 ++++++++---------------------
 src/TypesenseSync/TypesenseSync.php | 36 +++++++++++++++------------
 2 files changed, 31 insertions(+), 43 deletions(-)

diff --git a/src/TypesenseClient/SearchIndex.php b/src/TypesenseClient/SearchIndex.php
index a8c1a7d..b379879 100644
--- a/src/TypesenseClient/SearchIndex.php
+++ b/src/TypesenseClient/SearchIndex.php
@@ -106,7 +106,7 @@ public function purgeAll(): void
     {
         $newName = $this->createNewCollection();
         $this->updateAlias($newName);
-        $this->expireOldCollections(0);
+        $this->deleteOldCollections();
     }
 
     protected function isAliasExists(string $aliasName): bool
@@ -130,28 +130,20 @@ public function ensureSetup(): void
         }
     }
 
-    public function expireOldCollections(int $keepLast = 3): bool
+    /**
+     * Delete all collections that are no longer actively used.
+     */
+    public function deleteOldCollections(): void
     {
         $client = $this->getClient();
+        $collectionNameSkipList = [];
+
         // Don't delete the currently linked collection in all cases
         $alias = $client->aliases[$this->getAliasName()]->retrieve();
-        $collectionNameSkipList = [$alias['collection_name']];
-
-        // TODO: remove this once we are done testing
-        // We use these for testing as well, so skip for now
-        $collectionNameSkipList = array_merge(
-            $collectionNameSkipList,
-            ['cabinet-students', 'cabinet-files']
-        );
-
-        // Fetch all collections
-        try {
-            $collections = $client->collections->retrieve();
-        } catch (Exception|TypesenseClientErrorAlias) {
-            return false;
-        }
+        $collectionNameSkipList[] = $alias['collection_name'];
 
         // Collect all collections with the given prefix that are not in the skip list
+        $collections = $client->collections->retrieve();
         $collectionNameList = [];
         foreach ($collections as $collection) {
             if (str_starts_with($collection['name'], $this->getCollectionPrefix())
@@ -160,21 +152,11 @@ public function expireOldCollections(int $keepLast = 3): bool
             }
         }
 
-        rsort($collectionNameList);
-        // Slice off $keepLast collections
-        $collectionNameList = array_slice($collectionNameList, $keepLast);
-
         // Delete the remaining collections
         foreach ($collectionNameList as $collectionName) {
             $this->logger->info("Deleting old collection '$collectionName'");
-            try {
-                $client->collections[$collectionName]->delete();
-            } catch (Exception|TypesenseClientErrorAlias $e) {
-                $this->logger->error('Deleting collection failed', ['exception' => $e]);
-            }
+            $client->collections[$collectionName]->delete();
         }
-
-        return true;
     }
 
     public function setLogger(LoggerInterface $logger): void
diff --git a/src/TypesenseSync/TypesenseSync.php b/src/TypesenseSync/TypesenseSync.php
index d46fb5c..66c3e60 100644
--- a/src/TypesenseSync/TypesenseSync.php
+++ b/src/TypesenseSync/TypesenseSync.php
@@ -62,40 +62,46 @@ public function sync(bool $full = false)
         }
         $cursor = $this->getCursor();
 
+        // Process in chunks to reduce memory consumption
+        $chunkSize = 10000;
+
         if ($cursor === null) {
             $this->logger->info('Starting a full sync');
             $schema = $this->translator->getSchema();
 
             $this->searchIndex->setSchema($schema);
             $this->searchIndex->ensureSetup();
+            $this->searchIndex->deleteOldCollections();
             $collectionName = $this->searchIndex->createNewCollection();
 
             $res = $this->personSync->getAllPersons();
-            $documents = [];
-            foreach ($res->getPersons() as $person) {
-                $documents[] = $this->personToDocument($person);
-            }
-
-            $this->searchIndex->addDocumentsToCollection($collectionName, $documents);
-            if ($documents !== []) {
-                $this->addDummyDocuments($collectionName, $documents);
+            foreach (array_chunk($res->getPersons(), $chunkSize) as $persons) {
+                $documents = [];
+                foreach ($persons as $person) {
+                    $documents[] = $this->personToDocument($person);
+                }
+                $this->searchIndex->addDocumentsToCollection($collectionName, $documents);
+                if ($documents !== []) {
+                    $this->addDummyDocuments($collectionName, $documents);
+                }
             }
-            $this->searchIndex->ensureSetup();
 
             $this->searchIndex->updateAlias($collectionName);
-            $this->searchIndex->expireOldCollections();
+            $this->searchIndex->deleteOldCollections();
 
             $this->saveCursor($res->getCursor());
         } else {
             $this->logger->info('Starting a partial sync');
             $res = $this->personSync->getAllPersons($cursor);
+            $collectionName = $this->searchIndex->getCollectionName();
 
-            $documents = [];
-            foreach ($res->getPersons() as $person) {
-                $documents[] = $this->personToDocument($person);
+            foreach (array_chunk($res->getPersons(), $chunkSize) as $persons) {
+                $documents = [];
+                foreach ($persons as $person) {
+                    $documents[] = $this->personToDocument($person);
+                }
+                $this->searchIndex->addDocumentsToCollection($collectionName, $documents);
             }
-            $collectionName = $this->searchIndex->getCollectionName();
-            $this->searchIndex->addDocumentsToCollection($collectionName, $documents);
 
             $this->saveCursor($res->getCursor());
         }