feat: optimize image cleanup process with improved S3 key scanning and repository query

src/main/java/com/magamochi/image/model/repository/ImageRepository.java

@@ -1,10 +1,16 @@
 package com.magamochi.image.model.repository;
 
 import com.magamochi.image.model.entity.Image;
+import java.util.List;
 import java.util.Optional;
 import java.util.UUID;
 import org.springframework.data.jpa.repository.JpaRepository;
+import org.springframework.data.jpa.repository.Query;
+import org.springframework.data.repository.query.Param;
 
 public interface ImageRepository extends JpaRepository<Image, UUID> {
   Optional<Image> findByFileHash(String fileHash);
+
+  @Query("SELECT i.objectKey FROM Image i WHERE i.objectKey IN :objectKeys")
+  List<String> findExistingObjectKeys(@Param("objectKeys") List<String> objectKeys);
 }

src/main/java/com/magamochi/image/service/ImageService.java

@@ -4,7 +4,9 @@ import com.magamochi.common.exception.NotFoundException;
 import com.magamochi.image.model.entity.Image;
 import com.magamochi.image.model.repository.ImageRepository;
 import java.io.InputStream;
+import java.util.HashSet;
 import java.util.List;
+import java.util.Set;
 import java.util.UUID;
 import lombok.RequiredArgsConstructor;
 import lombok.extern.log4j.Log4j2;
@@ -49,8 +51,8 @@ public class ImageService {
         .orElseThrow(() -> new NotFoundException("Image not found with ID " + id));
   }
 
-  public List<Image> findAll() {
-    return imageRepository.findAll();
+  public Set<String> findExistingObjectKeys(List<String> objectKeys) {
+    return new HashSet<>(imageRepository.findExistingObjectKeys(objectKeys));
   }
 
   public InputStream getStream(Image image) {

src/main/java/com/magamochi/image/service/S3Service.java

@@ -4,7 +4,6 @@ import static java.util.Objects.nonNull;
 
 import java.io.InputStream;
 import java.time.Duration;
-import java.util.ArrayList;
 import java.util.List;
 import java.util.Set;
 import lombok.Getter;
@@ -42,8 +41,7 @@ public class S3Service {
     return filename;
   }
 
-  public List<String> listAllObjectKeys() {
-    var keys = new ArrayList<String>();
+  public void processObjectKeyPages(java.util.function.Consumer<List<String>> pageConsumer) {
     String continuationToken = null;
 
     do {
@@ -55,13 +53,12 @@ public class S3Service {
 
       var response = s3Client.listObjectsV2(requestBuilder.build());
 
-      response.contents().forEach(s3Object -> keys.add(s3Object.key()));
+      var page = response.contents().stream().map(S3Object::key).toList();
+      pageConsumer.accept(page);
 
       continuationToken = response.isTruncated() ? response.nextContinuationToken() : null;
 
     } while (nonNull(continuationToken));
-
-    return keys;
   }
 
   public void deleteObjects(Set<String> objectKeys) {

src/main/java/com/magamochi/image/task/ImageCleanupTask.java

@@ -1,9 +1,7 @@
 package com.magamochi.image.task;
 
-import com.magamochi.image.model.entity.Image;
 import com.magamochi.image.service.ImageService;
 import com.magamochi.image.service.S3Service;
-import java.util.stream.Collectors;
 import lombok.RequiredArgsConstructor;
 import lombok.extern.log4j.Log4j2;
 import org.springframework.beans.factory.annotation.Value;
@@ -31,21 +29,21 @@ public class ImageCleanupTask {
   }
 
   public void cleanupImages() {
-    log.info("Getting unused S3 object keys to remove.");
+    log.info("Scanning S3 pages for orphaned object keys.");
+    var keysToRemove = new java.util.HashSet<String>();
 
-    var imageKeys = s3Service.listAllObjectKeys();
+    s3Service.processObjectKeyPages(
+        page -> {
+          var existing = imageService.findExistingObjectKeys(page);
+          page.stream().filter(key -> !existing.contains(key)).forEach(keysToRemove::add);
+        });
 
-    var existingImages =
-        imageService.findAll().parallelStream()
-            .map(Image::getObjectKey)
-            .collect(Collectors.toSet());
+    if (keysToRemove.isEmpty()) {
+      log.info("No orphaned objects found.");
+      return;
+    }
 
-    var keysToRemove =
-        imageKeys.parallelStream()
-            .filter(imageKey -> !existingImages.contains(imageKey))
-            .collect(Collectors.toSet());
-
-    log.info("Removing {} objects from S3 storage", keysToRemove.size());
+    log.info("Removing {} orphaned objects from S3 storage", keysToRemove.size());
 
     s3Service.deleteObjects(keysToRemove);
 

src/main/resources/db/migration/V0012__ADD_INDEXES.sql

@@ -0,0 +1,98 @@
+-- ============================================================
+-- V0012 — Add performance indexes
+-- ============================================================
+
+-- mangas
+-- follow: findByFollowTrue() hot path for the scheduled follow-update task
+CREATE INDEX IF NOT EXISTS idx_mangas_follow ON mangas (follow);
+
+-- state: natural filter (PENDING, ACTIVE, etc.)
+CREATE INDEX IF NOT EXISTS idx_mangas_state ON mangas (state);
+
+-- adult: MangaSpecification always filters adult = false by default
+CREATE INDEX IF NOT EXISTS idx_mangas_adult ON mangas (adult);
+
+-- score: MangaSpecification filters score >= ?
+CREATE INDEX IF NOT EXISTS idx_mangas_score ON mangas (score);
+
+-- status: plain index for exact-match; functional index for LOWER(status) IN (...)
+CREATE INDEX IF NOT EXISTS idx_mangas_status ON mangas (status);
+CREATE INDEX IF NOT EXISTS idx_mangas_status_lower ON mangas (LOWER(status));
+
+-- title: findByTitleIgnoreCase + LIKE search via LOWER(title)
+CREATE INDEX IF NOT EXISTS idx_mangas_title_lower ON mangas (LOWER(title));
+
+-- ============================================================
+
+-- images
+-- object_key: needed for the page-by-page IN-query in ImageCleanupTask
+CREATE INDEX IF NOT EXISTS idx_images_object_key ON images (object_key);
+
+-- ============================================================
+
+-- manga_contents
+-- manga_content_provider_id: FK join for every chapter lookup
+CREATE INDEX IF NOT EXISTS idx_manga_contents_provider_id ON manga_contents (manga_content_provider_id);
+
+-- (manga_content_provider_id, LOWER(url)): covers existsByMangaContentProvider_IdAndUrlIgnoreCase
+CREATE INDEX IF NOT EXISTS idx_manga_contents_provider_url ON manga_contents (manga_content_provider_id, LOWER(url));
+
+-- downloaded: ingestion pipelines filter on downloaded = false
+CREATE INDEX IF NOT EXISTS idx_manga_contents_downloaded ON manga_contents (downloaded);
+
+-- ============================================================
+
+-- manga_content_images
+-- manga_content_id: findAllByMangaContent + existsByMangaContent_IdAndPosition
+CREATE INDEX IF NOT EXISTS idx_manga_content_images_content_id ON manga_content_images (manga_content_id);
+
+-- (manga_content_id, position): covers the position-existence check and ordered reads
+CREATE INDEX IF NOT EXISTS idx_manga_content_images_content_position ON manga_content_images (manga_content_id, position);
+
+-- image_id: FK into the 500k-row images table — cascade deletes and joins
+CREATE INDEX IF NOT EXISTS idx_manga_content_images_image_id ON manga_content_images (image_id);
+
+-- ============================================================
+
+-- manga_content_provider
+-- manga_id: findByManga_IdAndContentProvider_Id + @OneToMany collection load
+CREATE INDEX IF NOT EXISTS idx_manga_content_provider_manga_id ON manga_content_provider (manga_id);
+
+-- LOWER(manga_title): existsByMangaTitleIgnoreCaseAndContentProvider_Id
+CREATE INDEX IF NOT EXISTS idx_manga_content_provider_title_lower ON manga_content_provider (LOWER(manga_title));
+
+-- ============================================================
+
+-- manga_ingest_reviews
+-- content_provider_id: FK lookups and cascade deletes
+CREATE INDEX IF NOT EXISTS idx_manga_ingest_reviews_provider_id ON manga_ingest_reviews (content_provider_id);
+
+-- ============================================================
+
+-- user_manga_content_read
+-- user_id alone: countByUser, countDistinctMangaByUser
+-- (The UNIQUE constraint on (user_id, manga_content_id) exists but a single-column index
+--  on user_id allows efficient count/scan queries without reading the full composite key.)
+CREATE INDEX IF NOT EXISTS idx_user_manga_content_read_user_id ON user_manga_content_read (user_id);
+
+-- (user_id, created_at DESC): eliminates the sort step in findTop10ByUserOrderByCreatedAtDesc
+CREATE INDEX IF NOT EXISTS idx_user_manga_content_read_user_created ON user_manga_content_read (user_id, created_at DESC);
+
+-- ============================================================
+
+-- user_manga_follow
+-- user_id: findByUser, existsByUserAndManga
+CREATE INDEX IF NOT EXISTS idx_user_manga_follow_user_id ON user_manga_follow (user_id);
+
+-- manga_id: findByManga, existsByManga
+CREATE INDEX IF NOT EXISTS idx_user_manga_follow_manga_id ON user_manga_follow (manga_id);
+
+-- Composite unique: existsByUserAndManga / findByUserAndManga point lookups
+-- (no constraint existed on this table before)
+CREATE UNIQUE INDEX IF NOT EXISTS idx_user_manga_follow_user_manga ON user_manga_follow (user_id, manga_id);
+
+-- ============================================================
+
+-- manga_import_job
+-- status: MangaImportJobSpecification filters on status
+CREATE INDEX IF NOT EXISTS idx_manga_import_job_status ON manga_import_job (status);