refactor: remove smart search entity (#17447)

refactor: smart search entity

server/src/db.d.ts

@@ -4,7 +4,7 @@
  */
 
 import type { ColumnType } from 'kysely';
-import { AssetType, MemoryType, Permission, SyncEntityType } from 'src/enum';
+import { AssetFileType, AssetType, MemoryType, Permission, SyncEntityType } from 'src/enum';
 import { UserTable } from 'src/schema/tables/user.table';
 import { OnThisDayData } from 'src/types';
 
@@ -106,7 +106,7 @@ export interface AssetFiles {
   createdAt: Generated<Timestamp>;
   id: Generated<string>;
   path: string;
-  type: string;
+  type: AssetFileType;
   updatedAt: Generated<Timestamp>;
   updateId: Generated<string>;
 }

server/src/entities/asset.entity.ts

@@ -7,7 +7,6 @@ import { AssetFileEntity } from 'src/entities/asset-files.entity';
 import { AssetJobStatusEntity } from 'src/entities/asset-job-status.entity';
 import { ExifEntity } from 'src/entities/exif.entity';
 import { SharedLinkEntity } from 'src/entities/shared-link.entity';
-import { SmartSearchEntity } from 'src/entities/smart-search.entity';
 import { StackEntity } from 'src/entities/stack.entity';
 import { TagEntity } from 'src/entities/tag.entity';
 import { UserEntity } from 'src/entities/user.entity';
@@ -50,7 +49,6 @@ export class AssetEntity {
   originalFileName!: string;
   sidecarPath!: string | null;
   exifInfo?: ExifEntity;
-  smartSearch?: SmartSearchEntity;
   tags!: TagEntity[];
   sharedLinks!: SharedLinkEntity[];
   albums?: AlbumEntity[];
@@ -97,9 +95,9 @@ export function withFiles(eb: ExpressionBuilder<DB, 'assets'>, type?: AssetFileT
   return jsonArrayFrom(
     eb
       .selectFrom('asset_files')
-      .selectAll()
+      .selectAll('asset_files')
       .whereRef('asset_files.assetId', '=', 'assets.id')
-      .$if(!!type, (qb) => qb.where('type', '=', type!)),
+      .$if(!!type, (qb) => qb.where('asset_files.type', '=', type!)),
   ).as('files');
 }
 

server/src/entities/smart-search.entity.ts

@@ -1,7 +0,0 @@
-import { AssetEntity } from 'src/entities/asset.entity';
-
-export class SmartSearchEntity {
-  asset?: AssetEntity;
-  assetId!: string;
-  embedding!: string;
-}

server/src/queries/asset.repository.sql

@@ -179,6 +179,37 @@ from
 where
   "livePhotoVideoId" = $1::uuid
 
+-- AssetRepository.getAssetForSearchDuplicatesJob
+select
+  "id",
+  "type",
+  "ownerId",
+  "duplicateId",
+  "stackId",
+  "isVisible",
+  "smart_search"."embedding",
+  (
+    select
+      coalesce(json_agg(agg), '[]')
+    from
+      (
+        select
+          "asset_files".*
+        from
+          "asset_files"
+        where
+          "asset_files"."assetId" = "assets"."id"
+          and "asset_files"."type" = $1
+      ) as agg
+  ) as "files"
+from
+  "assets"
+  left join "smart_search" on "assets"."id" = "smart_search"."assetId"
+where
+  "assets"."id" = $2::uuid
+limit
+  $3
+
 -- AssetRepository.getById
 select
   "assets".*

server/src/repositories/asset.repository.ts

@@ -475,6 +475,26 @@ export class AssetRepository {
     return count as number;
   }
 
+  @GenerateSql({ params: [DummyValue.UUID] })
+  getAssetForSearchDuplicatesJob(id: string) {
+    return this.db
+      .selectFrom('assets')
+      .where('assets.id', '=', asUuid(id))
+      .leftJoin('smart_search', 'assets.id', 'smart_search.assetId')
+      .select((eb) => [
+        'id',
+        'type',
+        'ownerId',
+        'duplicateId',
+        'stackId',
+        'isVisible',
+        'smart_search.embedding',
+        withFiles(eb, AssetFileType.PREVIEW),
+      ])
+      .limit(1)
+      .executeTakeFirst();
+  }
+
   @GenerateSql({ params: [DummyValue.UUID] })
   getById(
     id: string,

server/src/services/duplicate.service.spec.ts

@@ -1,4 +1,4 @@
-import { JobName, JobStatus } from 'src/enum';
+import { AssetFileType, AssetType, JobName, JobStatus } from 'src/enum';
 import { WithoutProperty } from 'src/repositories/asset.repository';
 import { DuplicateService } from 'src/services/duplicate.service';
 import { SearchService } from 'src/services/search.service';
@@ -9,6 +9,33 @@ import { beforeEach, vitest } from 'vitest';
 
 vitest.useFakeTimers();
 
+const hasEmbedding = {
+  id: 'asset-1',
+  ownerId: 'user-id',
+  files: [
+    {
+      assetId: 'asset-1',
+      createdAt: new Date(),
+      id: 'file-1',
+      path: 'preview.jpg',
+      type: AssetFileType.PREVIEW,
+      updatedAt: new Date(),
+      updateId: 'update-1',
+    },
+  ],
+  isVisible: true,
+  stackId: null,
+  type: AssetType.IMAGE,
+  duplicateId: null,
+  embedding: '[1, 2, 3, 4]',
+};
+
+const hasDupe = {
+  ...hasEmbedding,
+  id: 'asset-2',
+  duplicateId: 'duplicate-id',
+};
+
 describe(SearchService.name, () => {
   let sut: DuplicateService;
   let mocks: ServiceMocks;
@@ -25,16 +52,16 @@ describe(SearchService.name, () => {
     it('should get duplicates', async () => {
       mocks.asset.getDuplicates.mockResolvedValue([
         {
-          duplicateId: assetStub.hasDupe.duplicateId!,
+          duplicateId: 'duplicate-id',
-          assets: [assetStub.hasDupe, assetStub.hasDupe],
+          assets: [assetStub.image, assetStub.image],
         },
       ]);
       await expect(sut.getDuplicates(authStub.admin)).resolves.toEqual([
         {
-          duplicateId: assetStub.hasDupe.duplicateId,
+          duplicateId: 'duplicate-id',
           assets: [
-            expect.objectContaining({ id: assetStub.hasDupe.id }),
+            expect.objectContaining({ id: assetStub.image.id }),
-            expect.objectContaining({ id: assetStub.hasDupe.id }),
+            expect.objectContaining({ id: assetStub.image.id }),
           ],
         },
       ]);
@@ -175,7 +202,7 @@ describe(SearchService.name, () => {
 
     it('should skip if asset is part of stack', async () => {
       const id = assetStub.primaryImage.id;
-      mocks.asset.getById.mockResolvedValue(assetStub.primaryImage);
+      mocks.asset.getAssetForSearchDuplicatesJob.mockResolvedValue({ ...hasEmbedding, stackId: 'stack-id' });
 
       const result = await sut.handleSearchDuplicates({ id });
 
@@ -185,7 +212,7 @@ describe(SearchService.name, () => {
 
     it('should skip if asset is not visible', async () => {
       const id = assetStub.livePhotoMotionAsset.id;
-      mocks.asset.getById.mockResolvedValue(assetStub.livePhotoMotionAsset);
+      mocks.asset.getAssetForSearchDuplicatesJob.mockResolvedValue({ ...hasEmbedding, isVisible: false });
 
       const result = await sut.handleSearchDuplicates({ id });
 
@@ -194,7 +221,7 @@ describe(SearchService.name, () => {
     });
 
     it('should fail if asset is missing preview image', async () => {
-      mocks.asset.getById.mockResolvedValue(assetStub.noResizePath);
+      mocks.asset.getAssetForSearchDuplicatesJob.mockResolvedValue({ ...hasEmbedding, files: [] });
 
       const result = await sut.handleSearchDuplicates({ id: assetStub.noResizePath.id });
 
@@ -203,7 +230,7 @@ describe(SearchService.name, () => {
     });
 
     it('should fail if asset is missing embedding', async () => {
-      mocks.asset.getById.mockResolvedValue(assetStub.image);
+      mocks.asset.getAssetForSearchDuplicatesJob.mockResolvedValue({ ...hasEmbedding, embedding: null });
 
       const result = await sut.handleSearchDuplicates({ id: assetStub.image.id });
 
@@ -212,21 +239,21 @@ describe(SearchService.name, () => {
     });
 
     it('should search for duplicates and update asset with duplicateId', async () => {
-      mocks.asset.getById.mockResolvedValue(assetStub.hasEmbedding);
+      mocks.asset.getAssetForSearchDuplicatesJob.mockResolvedValue(hasEmbedding);
       mocks.search.searchDuplicates.mockResolvedValue([
         { assetId: assetStub.image.id, distance: 0.01, duplicateId: null },
       ]);
-      const expectedAssetIds = [assetStub.image.id, assetStub.hasEmbedding.id];
+      const expectedAssetIds = [assetStub.image.id, hasEmbedding.id];
 
-      const result = await sut.handleSearchDuplicates({ id: assetStub.hasEmbedding.id });
+      const result = await sut.handleSearchDuplicates({ id: hasEmbedding.id });
 
       expect(result).toBe(JobStatus.SUCCESS);
       expect(mocks.search.searchDuplicates).toHaveBeenCalledWith({
-        assetId: assetStub.hasEmbedding.id,
+        assetId: hasEmbedding.id,
-        embedding: assetStub.hasEmbedding.smartSearch!.embedding,
+        embedding: hasEmbedding.embedding,
         maxDistance: 0.01,
-        type: assetStub.hasEmbedding.type,
+        type: hasEmbedding.type,
-        userIds: [assetStub.hasEmbedding.ownerId],
+        userIds: [hasEmbedding.ownerId],
       });
       expect(mocks.asset.updateDuplicates).toHaveBeenCalledWith({
         assetIds: expectedAssetIds,
@@ -239,24 +266,24 @@ describe(SearchService.name, () => {
     });
 
     it('should use existing duplicate ID among matched duplicates', async () => {
-      const duplicateId = assetStub.hasDupe.duplicateId;
+      const duplicateId = hasDupe.duplicateId;
-      mocks.asset.getById.mockResolvedValue(assetStub.hasEmbedding);
+      mocks.asset.getAssetForSearchDuplicatesJob.mockResolvedValue(hasEmbedding);
-      mocks.search.searchDuplicates.mockResolvedValue([{ assetId: assetStub.hasDupe.id, distance: 0.01, duplicateId }]);
+      mocks.search.searchDuplicates.mockResolvedValue([{ assetId: hasDupe.id, distance: 0.01, duplicateId }]);
-      const expectedAssetIds = [assetStub.hasEmbedding.id];
+      const expectedAssetIds = [hasEmbedding.id];
 
-      const result = await sut.handleSearchDuplicates({ id: assetStub.hasEmbedding.id });
+      const result = await sut.handleSearchDuplicates({ id: hasEmbedding.id });
 
       expect(result).toBe(JobStatus.SUCCESS);
       expect(mocks.search.searchDuplicates).toHaveBeenCalledWith({
-        assetId: assetStub.hasEmbedding.id,
+        assetId: hasEmbedding.id,
-        embedding: assetStub.hasEmbedding.smartSearch!.embedding,
+        embedding: hasEmbedding.embedding,
         maxDistance: 0.01,
-        type: assetStub.hasEmbedding.type,
+        type: hasEmbedding.type,
-        userIds: [assetStub.hasEmbedding.ownerId],
+        userIds: [hasEmbedding.ownerId],
       });
       expect(mocks.asset.updateDuplicates).toHaveBeenCalledWith({
         assetIds: expectedAssetIds,
-        targetDuplicateId: assetStub.hasDupe.duplicateId,
+        targetDuplicateId: duplicateId,
         duplicateIds: [],
       });
       expect(mocks.asset.upsertJobStatus).toHaveBeenCalledWith(
@@ -265,15 +292,15 @@ describe(SearchService.name, () => {
     });
 
     it('should remove duplicateId if no duplicates found and asset has duplicateId', async () => {
-      mocks.asset.getById.mockResolvedValue(assetStub.hasDupe);
+      mocks.asset.getAssetForSearchDuplicatesJob.mockResolvedValue(hasDupe);
       mocks.search.searchDuplicates.mockResolvedValue([]);
 
-      const result = await sut.handleSearchDuplicates({ id: assetStub.hasDupe.id });
+      const result = await sut.handleSearchDuplicates({ id: hasDupe.id });
 
       expect(result).toBe(JobStatus.SUCCESS);
-      expect(mocks.asset.update).toHaveBeenCalledWith({ id: assetStub.hasDupe.id, duplicateId: null });
+      expect(mocks.asset.update).toHaveBeenCalledWith({ id: hasDupe.id, duplicateId: null });
       expect(mocks.asset.upsertJobStatus).toHaveBeenCalledWith({
-        assetId: assetStub.hasDupe.id,
+        assetId: hasDupe.id,
         duplicatesDetectedAt: expect.any(Date),
       });
     });

server/src/services/duplicate.service.ts

@@ -4,7 +4,6 @@ import { OnJob } from 'src/decorators';
 import { mapAsset } from 'src/dtos/asset-response.dto';
 import { AuthDto } from 'src/dtos/auth.dto';
 import { DuplicateResponseDto } from 'src/dtos/duplicate.dto';
-import { AssetEntity } from 'src/entities/asset.entity';
 import { AssetFileType, JobName, JobStatus, QueueName } from 'src/enum';
 import { WithoutProperty } from 'src/repositories/asset.repository';
 import { AssetDuplicateResult } from 'src/repositories/search.repository';
@@ -53,7 +52,7 @@ export class DuplicateService extends BaseService {
       return JobStatus.SKIPPED;
     }
 
-    const asset = await this.assetRepository.getById(id, { files: true, smartSearch: true });
+    const asset = await this.assetRepository.getAssetForSearchDuplicatesJob(id);
     if (!asset) {
       this.logger.error(`Asset ${id} not found`);
       return JobStatus.FAILED;
@@ -75,14 +74,14 @@ export class DuplicateService extends BaseService {
       return JobStatus.FAILED;
     }
 
-    if (!asset.smartSearch?.embedding) {
+    if (!asset.embedding) {
       this.logger.debug(`Asset ${id} is missing embedding`);
       return JobStatus.FAILED;
     }
 
     const duplicateAssets = await this.searchRepository.searchDuplicates({
       assetId: asset.id,
-      embedding: asset.smartSearch.embedding,
+      embedding: asset.embedding,
       maxDistance: machineLearning.duplicateDetection.maxDistance,
       type: asset.type,
       userIds: [asset.ownerId],
@@ -105,7 +104,10 @@ export class DuplicateService extends BaseService {
     return JobStatus.SUCCESS;
   }
 
-  private async updateDuplicates(asset: AssetEntity, duplicateAssets: AssetDuplicateResult[]): Promise<string[]> {
+  private async updateDuplicates(
+    asset: { id: string; duplicateId: string | null },
+    duplicateAssets: AssetDuplicateResult[],
+  ): Promise<string[]> {
     const duplicateIds = [
       ...new Set(
         duplicateAssets

server/src/utils/asset.util.ts

@@ -13,7 +13,10 @@ import { PartnerRepository } from 'src/repositories/partner.repository';
 import { IBulkAsset, ImmichFile, UploadFile } from 'src/types';
 import { checkAccess } from 'src/utils/access';
 
-export const getAssetFile = (files: AssetFileEntity[], type: AssetFileType | GeneratedImageType) => {
+export const getAssetFile = <T extends { type: AssetFileType }>(
+  files: T[],
+  type: AssetFileType | GeneratedImageType,
+) => {
   return (files || []).find((file) => file.type === type);
 };
 

server/test/fixtures/asset.stub.ts

@@ -851,88 +851,4 @@ export const assetStub = {
     duplicateId: null,
     isOffline: false,
   }),
-
-  hasEmbedding: Object.freeze<AssetEntity>({
-    id: 'asset-id-embedding',
-    status: AssetStatus.ACTIVE,
-    deviceAssetId: 'device-asset-id',
-    fileModifiedAt: new Date('2023-02-23T05:06:29.716Z'),
-    fileCreatedAt: new Date('2023-02-23T05:06:29.716Z'),
-    owner: userStub.user1,
-    ownerId: 'user-id',
-    deviceId: 'device-id',
-    originalPath: '/original/path.jpg',
-    checksum: Buffer.from('file hash', 'utf8'),
-    type: AssetType.IMAGE,
-    files,
-    thumbhash: Buffer.from('blablabla', 'base64'),
-    encodedVideoPath: null,
-    createdAt: new Date('2023-02-23T05:06:29.716Z'),
-    updatedAt: new Date('2023-02-23T05:06:29.716Z'),
-    localDateTime: new Date('2023-02-23T05:06:29.716Z'),
-    isFavorite: true,
-    isArchived: false,
-    duration: null,
-    isVisible: true,
-    isExternal: false,
-    livePhotoVideo: null,
-    livePhotoVideoId: null,
-    tags: [],
-    sharedLinks: [],
-    originalFileName: 'asset-id.jpg',
-    faces: [],
-    deletedAt: null,
-    sidecarPath: null,
-    exifInfo: {
-      fileSizeInByte: 5000,
-    } as ExifEntity,
-    duplicateId: null,
-    smartSearch: {
-      assetId: 'asset-id',
-      embedding: '[1, 2, 3, 4]',
-    },
-    isOffline: false,
-  }),
-
-  hasDupe: Object.freeze<AssetEntity>({
-    id: 'asset-id-dupe',
-    status: AssetStatus.ACTIVE,
-    deviceAssetId: 'device-asset-id',
-    fileModifiedAt: new Date('2023-02-23T05:06:29.716Z'),
-    fileCreatedAt: new Date('2023-02-23T05:06:29.716Z'),
-    owner: userStub.user1,
-    ownerId: 'user-id',
-    deviceId: 'device-id',
-    originalPath: '/original/path.jpg',
-    checksum: Buffer.from('file hash', 'utf8'),
-    type: AssetType.IMAGE,
-    files,
-    thumbhash: Buffer.from('blablabla', 'base64'),
-    encodedVideoPath: null,
-    createdAt: new Date('2023-02-23T05:06:29.716Z'),
-    updatedAt: new Date('2023-02-23T05:06:29.716Z'),
-    localDateTime: new Date('2023-02-23T05:06:29.716Z'),
-    isFavorite: true,
-    isArchived: false,
-    duration: null,
-    isVisible: true,
-    isExternal: false,
-    livePhotoVideo: null,
-    livePhotoVideoId: null,
-    tags: [],
-    sharedLinks: [],
-    originalFileName: 'asset-id.jpg',
-    faces: [],
-    deletedAt: null,
-    sidecarPath: null,
-    exifInfo: {
-      fileSizeInByte: 5000,
-    } as ExifEntity,
-    duplicateId: 'duplicate-id',
-    smartSearch: {
-      assetId: 'asset-id',
-      embedding: '[1, 2, 3, 4]',
-    },
-    isOffline: false,
-  }),
 };

server/test/repositories/asset.repository.mock.ts

@@ -11,6 +11,7 @@ export const newAssetRepositoryMock = (): Mocked<RepositoryInterface<AssetReposi
     upsertJobStatus: vitest.fn(),
     getByDayOfYear: vitest.fn(),
     getByIds: vitest.fn().mockResolvedValue([]),
+    getAssetForSearchDuplicatesJob: vitest.fn(),
     getByIdsWithAllRelations: vitest.fn().mockResolvedValue([]),
     getByAlbumId: vitest.fn(),
     getByDeviceIds: vitest.fn(),