feat(server): add OCR extraction workflow filter (#32464)

feat(server): add OCR workflow filter

(cherry picked from commit 2e0eb35e93dcf8468398d7015daa5ab231c7dd9c)

i18n/en.json

@@ -2263,6 +2263,8 @@
   "trigger": "Trigger",
   "trigger_asset_metadata_extraction": "Asset Metadata Extraction",
   "trigger_asset_metadata_extraction_description": "Triggered when the EXIF metadata of an asset is extracted",
+  "trigger_asset_ocr_extraction": "Asset OCR Extraction",
+  "trigger_asset_ocr_extraction_description": "Triggered when the OCR text of an asset is extracted",
   "trigger_asset_tagged": "Asset Tagged",
   "trigger_asset_tagged_description": "Triggered when a tag is added to an asset",
   "trigger_asset_uploaded": "Asset Upload",

open-api/immich-openapi-specs.json

@@ -34673,6 +34673,7 @@
         "enum": [
           "AssetCreate",
           "AssetMetadataExtraction",
+          "AssetOcrExtraction",
           "AssetTagged"
         ],
         "type": "string"

packages/plugin-core/manifest.json

@@ -75,6 +75,32 @@
         }
       ],
       "uiHints": ["SmartAlbum"]
+    },
+    {
+      "name": "receipts",
+      "title": "Receipts",
+      "description": "Automatically add receipt assets to an album",
+      "trigger": "AssetOcrExtraction",
+      "steps": [
+        {
+          "method": "immich-plugin-core#assetOcrFilter",
+          "config": {
+            "caseSensitive": false,
+            "pattern": "receipt.+total",
+            "matchType": "regex",
+            "matchScope": "full",
+            "minScore": 0
+          }
+        },
+        {
+          "method": "immich-plugin-core#assetAddToAlbums",
+          "config": {
+            "albumName": "Receipts",
+            "albumIds": []
+          }
+        }
+      ],
+      "uiHints": ["SmartAlbum"]
     }
   ],
   "methods": [
@@ -428,6 +454,50 @@
       },
       "uiHints": ["Filter"]
     },
+    {
+      "name": "assetOcrFilter",
+      "title": "Filter by OCR text",
+      "description": "Filter assets by OCR text pattern using text matching or regular expressions",
+      "types": ["AssetV1"],
+      "schema": {
+        "type": "object",
+        "properties": {
+          "pattern": {
+            "type": "string",
+            "title": "OCR text pattern",
+            "description": "Text or regex pattern to match against OCR text"
+          },
+          "matchType": {
+            "type": "string",
+            "enum": ["contains", "startsWith", "exact", "regex"],
+            "default": "contains",
+            "title": "Match type",
+            "description": "Type of pattern matching to perform"
+          },
+          "caseSensitive": {
+            "type": "boolean",
+            "default": false,
+            "title": "Case sensitive",
+            "description": "Whether matching should be case-sensitive"
+          },
+          "minScore": {
+            "type": "number",
+            "default": 0,
+            "title": "Minimum score",
+            "description": "Minimum OCR confidence score to match"
+          },
+          "matchScope": {
+            "type": "string",
+            "enum": ["full", "box"],
+            "default": "full",
+            "title": "Match scope",
+            "description": "Scope of the OCR text to match against"
+          }
+        },
+        "required": ["pattern"]
+      },
+      "uiHints": ["Filter"]
+    },
     {
       "name": "assetArchive",
       "title": "Archive asset",

packages/plugin-core/src/index.ts

@@ -194,6 +194,13 @@ const methods = wrapper<Manifest>({
     return { workflow: { continue: hasTimeZone === needsTimeZone } };
   },
 
+  assetOcrFilter: ({ config, data }) => {
+    const entries = (data.asset.ocr ?? []).filter((ocr) => ocr.textScore >= (config.minScore ?? 0));
+    const texts = config.matchScope === 'full' ? [entries.map((ocr) => ocr.text).join(' ')] : entries.map((ocr) => ocr.text);
+    const matched = texts.some((t) => matchValueResult(t, config).workflow?.continue);
+    return { workflow: { continue: matched } };
+  },
+
   assetTagFilter: ({ config, data }) => {
     const assetTags = data.asset.tags.map((tag) => tag.id);
 
@@ -254,6 +261,7 @@ const {
   assetDateFilter,
   assetLock,
   assetMissingTimeZoneFilter,
+  assetOcrFilter,
   assetTagFilter,
   assetTypeFilter,
   assetVisibility,
@@ -274,6 +282,7 @@ export {
   assetDateFilter,
   assetLock,
   assetMissingTimeZoneFilter,
+  assetOcrFilter,
   assetTagFilter,
   assetTypeFilter,
   assetVisibility,

packages/plugin-sdk/src/types.ts

@@ -23,6 +23,7 @@ export type WorkflowEventData<T extends WorkflowType> = WorkflowEventMap[T];
 export enum WorkflowTrigger {
   AssetCreate = 'AssetCreate',
   AssetMetadataExtraction = 'AssetMetadataExtraction',
+  AssetOcrExtraction = 'AssetOcrExtraction',
   AssetTagged = 'AssetTagged',
   // PersonRecognized = 'PersonRecognized',
 }
@@ -127,6 +128,11 @@ export type AssetV1 = {
       tags: string[] | null;
       updatedAt: string | null;
     } | null;
+    ocr?: {
+      text: string;
+      textScore: number;
+      boxScore: number;
+    }[]
   };
 };
 

packages/sdk/src/fetch-client.ts

@@ -8603,6 +8603,7 @@ export enum WorkflowType {
 export enum WorkflowTrigger {
     AssetCreate = "AssetCreate",
     AssetMetadataExtraction = "AssetMetadataExtraction",
+    AssetOcrExtraction = "AssetOcrExtraction",
     AssetTagged = "AssetTagged"
 }
 export enum QueueJobStatus {

server/src/queries/asset.job.repository.sql

@@ -615,6 +615,7 @@ where
 
 -- AssetJobRepository.getForOcr
 select
+  "asset"."ownerId",
   "asset"."visibility",
   (
     select

server/src/repositories/asset-job.repository.ts

@@ -255,7 +255,7 @@ export class AssetJobRepository {
   getForOcr(id: string) {
     return this.db
       .selectFrom('asset')
-      .select((eb) => ['asset.visibility', withFilePath(eb, AssetFileType.Preview).as('previewFile')])
+      .select((eb) => ['asset.ownerId', 'asset.visibility', withFilePath(eb, AssetFileType.Preview).as('previewFile')])
       .where('asset.id', '=', id)
       .executeTakeFirst();
   }

server/src/repositories/event.repository.ts

@@ -53,6 +53,7 @@ type EventMap = {
   AssetTrash: [{ assetId: string; userId: string }];
   AssetDelete: [{ assetId: string; userId: string }];
   AssetMetadataExtracted: [{ assetId: string; userId: string; source?: JobSource }];
+  AssetOcrExtracted: [{ assetId: string; userId: string }];
 
   // asset bulk events
   AssetTrashAll: [{ assetIds: string[]; userId: string }];

server/src/repositories/workflow.repository.ts

@@ -1,3 +1,4 @@
+import { WorkflowTrigger } from '@immich/plugin-sdk';
 import { Injectable } from '@nestjs/common';
 import { jsonArrayFrom, jsonObjectFrom } from 'kysely/helpers/postgres';
 import { InjectKysely } from 'nestjs-kysely';
@@ -177,7 +178,7 @@ export class WorkflowRepository {
     await this.db.deleteFrom('workflow').where('id', '=', id).execute();
   }
 
-  getForAssetV1(assetId: string) {
+  getForAssetV1(assetId: string, trigger: WorkflowTrigger) {
     return this.db
       .selectFrom('asset')
       .leftJoin('asset_exif', 'asset_exif.assetId', 'asset.id')
@@ -221,6 +222,17 @@ export class WorkflowRepository {
             ])
             .whereRef('asset_exif.assetId', '=', 'asset.id'),
         ).as('exifInfo'),
+        ...(trigger === WorkflowTrigger.AssetOcrExtraction
+          ? [
+              jsonArrayFrom(
+                eb
+                  .selectFrom('asset_ocr')
+                  .select(['asset_ocr.text', 'asset_ocr.textScore', 'asset_ocr.boxScore'])
+                  .whereRef('asset_ocr.assetId', '=', 'asset.id')
+                  .where('isVisible', '=', true),
+              ).as('ocr'),
+            ]
+          : []),
       ])
       .where('id', '=', assetId)
       .executeTakeFirstOrThrow();

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

@@ -13,6 +13,7 @@ describe(OcrService.name, () => {
 
     mocks.config.getWorker.mockReturnValue(ImmichWorker.Microservices);
     mocks.assetJob.getForOcr.mockResolvedValue({
+      ownerId: 'user-id',
       visibility: AssetVisibility.Timeline,
       previewFile: '/uploads/user-id/thumbs/path.jpg',
     });
@@ -69,16 +70,22 @@ describe(OcrService.name, () => {
 
       expect(mocks.asset.getByIds).not.toHaveBeenCalled();
       expect(mocks.machineLearning.encodeImage).not.toHaveBeenCalled();
+      expect(mocks.event.emit).not.toHaveBeenCalled();
     });
 
     it('should skip assets without a resize path', async () => {
       const asset = AssetFactory.create();
-      mocks.assetJob.getForOcr.mockResolvedValue({ visibility: AssetVisibility.Timeline, previewFile: null });
+      mocks.assetJob.getForOcr.mockResolvedValue({
+        ownerId: 'userId',
+        visibility: AssetVisibility.Timeline,
+        previewFile: null,
+      });
 
       expect(await sut.handleOcr({ id: asset.id })).toEqual(JobStatus.Failed);
 
       expect(mocks.ocr.upsert).not.toHaveBeenCalled();
       expect(mocks.machineLearning.ocr).not.toHaveBeenCalled();
+      expect(mocks.event.emit).not.toHaveBeenCalled();
     });
 
     it('should save the returned objects', async () => {
@@ -170,6 +177,7 @@ describe(OcrService.name, () => {
     it('should skip invisible assets', async () => {
       const asset = AssetFactory.from().file({ type: AssetFileType.Preview }).build();
       mocks.assetJob.getForOcr.mockResolvedValue({
+        ownerId: 'userId',
         visibility: AssetVisibility.Hidden,
         previewFile: asset.files[0].path,
       });
@@ -178,6 +186,7 @@ describe(OcrService.name, () => {
 
       expect(mocks.machineLearning.ocr).not.toHaveBeenCalled();
       expect(mocks.ocr.upsert).not.toHaveBeenCalled();
+      expect(mocks.event.emit).not.toHaveBeenCalled();
     });
 
     it('should fail if asset could not be found', async () => {
@@ -187,6 +196,19 @@ describe(OcrService.name, () => {
 
       expect(mocks.machineLearning.ocr).not.toHaveBeenCalled();
       expect(mocks.ocr.upsert).not.toHaveBeenCalled();
+      expect(mocks.event.emit).not.toHaveBeenCalled();
+    });
+
+    it('should emit an event', async () => {
+      const asset = AssetFactory.create();
+      mockOcrResult('Hello World');
+
+      await sut.handleOcr({ id: asset.id });
+
+      expect(mocks.event.emit).toHaveBeenCalledWith('AssetOcrExtracted', {
+        userId: 'user-id',
+        assetId: asset.id,
+      });
     });
 
     describe('search tokenization', () => {

server/src/services/ocr.service.ts

@@ -50,6 +50,11 @@ export class OcrService extends BaseService {
 
     await this.assetRepository.upsertJobStatus({ assetId: id, ocrAt: new Date() });
 
+    await this.eventRepository.emit('AssetOcrExtracted', {
+      userId: asset.ownerId,
+      assetId: id,
+    });
+
     this.logger.debug(`Processed ${ocrResults.text.length} OCR result(s) for ${id}`);
     return JobStatus.Success;
   }

server/src/services/workflow-execution.service.ts

@@ -316,6 +316,11 @@ export class WorkflowExecutionService extends BaseService {
     return this.onAssetTrigger({ userId, assetId, trigger: WorkflowTrigger.AssetMetadataExtraction });
   }
 
+  @OnEvent({ name: 'AssetOcrExtracted' })
+  onAssetOcrExtracted({ userId, assetId }: ArgOf<'AssetOcrExtracted'>) {
+    return this.onAssetTrigger({ userId, assetId, trigger: WorkflowTrigger.AssetOcrExtraction });
+  }
+
   @OnEvent({ name: 'AssetTag' })
   onAssetTagged({ assetId, userId }: ArgOf<'AssetTag'>) {
     return this.onAssetTrigger({ userId, assetId, trigger: WorkflowTrigger.AssetTagged });
@@ -332,7 +337,7 @@ export class WorkflowExecutionService extends BaseService {
   }
 
   @OnJob({ name: JobName.WorkflowAssetTrigger, queue: QueueName.Workflow })
-  handleAssetTrigger({ workflowId, assetId }: JobOf<JobName.WorkflowAssetTrigger>) {
+  handleAssetTrigger({ workflowId, assetId, trigger }: JobOf<JobName.WorkflowAssetTrigger>) {
     return this.execute(workflowId, (type) => {
       const assetService = BaseService.create(AssetService, this);
 
@@ -340,7 +345,7 @@ export class WorkflowExecutionService extends BaseService {
         case WorkflowType.AssetV1: {
           return {
             read: async () => {
-              const asset = await this.workflowRepository.getForAssetV1(assetId);
+              const asset = await this.workflowRepository.getForAssetV1(assetId, trigger);
               return {
                 data: { asset } as any,
                 authUserId: asset.ownerId,

server/src/types.ts

@@ -1,3 +1,4 @@
+import { WorkflowTrigger } from '@immich/plugin-sdk';
 import { ShallowDehydrateObject } from 'kysely';
 import { Mocked } from 'vitest';
 import { VECTOR_EXTENSIONS } from 'src/constants.js';
@@ -439,7 +440,7 @@ export type JobItem =
   | { name: JobName.Ocr; data: IEntityJob }
 
   // Workflow
-  | { name: JobName.WorkflowAssetTrigger; data: { workflowId: string; assetId: string } }
+  | { name: JobName.WorkflowAssetTrigger; data: { workflowId: string; assetId: string; trigger: WorkflowTrigger } }
 
   // Integrity
   | { name: JobName.IntegrityUntrackedFilesQueueAll; data?: IIntegrityJob }

server/src/utils/workflow.ts

@@ -6,6 +6,7 @@ export const triggerMap: Record<WorkflowTrigger, WorkflowType[]> = {
   [WorkflowTrigger.AssetCreate]: [WorkflowType.AssetV1],
   // [WorkflowTrigger.PersonRecognized]: [WorkflowType.AssetPersonV1],
   [WorkflowTrigger.AssetMetadataExtraction]: [WorkflowType.AssetV1],
+  [WorkflowTrigger.AssetOcrExtraction]: [WorkflowType.AssetV1],
   [WorkflowTrigger.AssetTagged]: [WorkflowType.AssetV1],
 };
 

server/test/medium.factory.ts

@@ -78,6 +78,7 @@ import { AssetExifTable } from 'src/schema/tables/asset-exif.table.js';
 import { AssetFileTable } from 'src/schema/tables/asset-file.table.js';
 import { AssetJobStatusTable } from 'src/schema/tables/asset-job-status.table.js';
 import { AssetMetadataTable } from 'src/schema/tables/asset-metadata.table.js';
+import { AssetOcrTable } from 'src/schema/tables/asset-ocr.table.js';
 import { AssetTable } from 'src/schema/tables/asset.table.js';
 import { FaceSearchTable } from 'src/schema/tables/face-search.table.js';
 import { MemoryTable } from 'src/schema/tables/memory.table.js';
@@ -230,6 +231,28 @@ export class MediumTestContext<S extends ClassConstructor<typeof BaseService> =
     return { assetFace, result };
   }
 
+  async newAssetOcr(
+    dto: (Partial<Insertable<AssetOcrTable>> & { assetId: string; text: string; textScore: number })[],
+  ) {
+    const result = await this.get(OcrRepository).upsert(
+      dto[0].assetId,
+      dto.map((item) => ({
+        ...item,
+        x1: 0,
+        y1: 0,
+        x2: 1,
+        y2: 1,
+        x3: 2,
+        y3: 2,
+        x4: 3,
+        y4: 3,
+        boxScore: item.textScore,
+      })),
+      '',
+    );
+    return { ocr: dto, result };
+  }
+
   async newMemory(dto: Partial<Insertable<MemoryTable>> = {}) {
     const memory = mediumFactory.memoryInsert(dto);
     const result = await this.get(MemoryRepository).create(memory, new Set<string>());

server/test/medium/specs/services/ocr.service.spec.ts

@@ -3,6 +3,7 @@ import { AssetFileType, JobStatus } from 'src/enum.js';
 import { AssetJobRepository } from 'src/repositories/asset-job.repository.js';
 import { AssetRepository } from 'src/repositories/asset.repository.js';
 import { ConfigRepository } from 'src/repositories/config.repository.js';
+import { EventRepository } from 'src/repositories/event.repository.js';
 import { JobRepository } from 'src/repositories/job.repository.js';
 import { LoggingRepository } from 'src/repositories/logging.repository.js';
 import { MachineLearningRepository } from 'src/repositories/machine-learning.repository.js';
@@ -20,7 +21,7 @@ const setup = (db?: Kysely<DB>) => {
   return newMediumService(OcrService, {
     database: db || defaultDatabase,
     real: [AssetRepository, AssetJobRepository, ConfigRepository, OcrRepository, SystemMetadataRepository],
-    mock: [JobRepository, LoggingRepository, MachineLearningRepository],
+    mock: [EventRepository, JobRepository, LoggingRepository, MachineLearningRepository],
   });
 };
 
@@ -39,6 +40,7 @@ describe(OcrService.name, () => {
     const { user } = await ctx.newUser();
     const { asset } = await ctx.newAsset({ ownerId: user.id });
     await ctx.newAssetFile({ assetId: asset.id, type: AssetFileType.Preview, path: 'preview.jpg' });
+    ctx.getMock(EventRepository).emit.mockResolvedValue();
 
     const machineLearningMock = ctx.getMock(MachineLearningRepository);
     machineLearningMock.ocr.mockResolvedValue({
@@ -91,6 +93,7 @@ describe(OcrService.name, () => {
     const { user } = await ctx.newUser();
     const { asset } = await ctx.newAsset({ ownerId: user.id });
     await ctx.newAssetFile({ assetId: asset.id, type: AssetFileType.Preview, path: 'preview.jpg' });
+    ctx.getMock(EventRepository).emit.mockResolvedValue();
 
     const machineLearningMock = ctx.getMock(MachineLearningRepository);
     machineLearningMock.ocr.mockResolvedValue({
@@ -215,6 +218,7 @@ describe(OcrService.name, () => {
     const { user } = await ctx.newUser();
     const { asset } = await ctx.newAsset({ ownerId: user.id });
     await ctx.newAssetFile({ assetId: asset.id, type: AssetFileType.Preview, path: 'preview.jpg' });
+    ctx.getMock(EventRepository).emit.mockResolvedValue();
 
     const machineLearningMock = ctx.getMock(MachineLearningRepository);
     machineLearningMock.ocr.mockResolvedValue({ box: [], boxScore: [], text: [], textScore: [] });
@@ -240,6 +244,7 @@ describe(OcrService.name, () => {
     const { user } = await ctx.newUser();
     const { asset } = await ctx.newAsset({ ownerId: user.id });
     await ctx.newAssetFile({ assetId: asset.id, type: AssetFileType.Preview, path: 'preview.jpg' });
+    ctx.getMock(EventRepository).emit.mockResolvedValue();
 
     const machineLearningMock = ctx.getMock(MachineLearningRepository);
     machineLearningMock.ocr.mockResolvedValue({

server/test/medium/specs/workflow/workflow-core-plugin.spec.ts

@@ -146,7 +146,13 @@ describe('core plugin', () => {
         steps: [{ method: 'immich-plugin-core#assetArchive' }],
       });
 
-      await expect(ctx.sut.handleAssetTrigger({ workflowId: workflow.id, assetId: asset.id })).resolves.toBeUndefined();
+      await expect(
+        ctx.sut.handleAssetTrigger({
+          workflowId: workflow.id,
+          assetId: asset.id,
+          trigger: WorkflowTrigger.AssetCreate,
+        }),
+      ).resolves.toBeUndefined();
 
       await expect(ctx.get(AssetRepository).getById(asset.id)).resolves.toMatchObject({
         visibility: AssetVisibility.Archive,
@@ -163,7 +169,13 @@ describe('core plugin', () => {
         steps: [{ method: 'immich-plugin-core#assetArchive', config: { inverse: true } }],
       });
 
-      await expect(ctx.sut.handleAssetTrigger({ workflowId: workflow.id, assetId: asset.id })).resolves.toBeUndefined();
+      await expect(
+        ctx.sut.handleAssetTrigger({
+          workflowId: workflow.id,
+          assetId: asset.id,
+          trigger: WorkflowTrigger.AssetCreate,
+        }),
+      ).resolves.toBeUndefined();
 
       await expect(ctx.get(AssetRepository).getById(asset.id)).resolves.toMatchObject({
         visibility: AssetVisibility.Timeline,
@@ -182,7 +194,13 @@ describe('core plugin', () => {
         steps: [{ method: 'immich-plugin-core#assetLock' }],
       });
 
-      await expect(ctx.sut.handleAssetTrigger({ workflowId: workflow.id, assetId: asset.id })).resolves.toBeUndefined();
+      await expect(
+        ctx.sut.handleAssetTrigger({
+          workflowId: workflow.id,
+          assetId: asset.id,
+          trigger: WorkflowTrigger.AssetCreate,
+        }),
+      ).resolves.toBeUndefined();
 
       await expect(ctx.get(AssetRepository).getById(asset.id)).resolves.toMatchObject({
         visibility: AssetVisibility.Locked,
@@ -199,7 +217,13 @@ describe('core plugin', () => {
         steps: [{ method: 'immich-plugin-core#assetLock', config: { inverse: true } }],
       });
 
-      await expect(ctx.sut.handleAssetTrigger({ workflowId: workflow.id, assetId: asset.id })).resolves.toBeUndefined();
+      await expect(
+        ctx.sut.handleAssetTrigger({
+          workflowId: workflow.id,
+          assetId: asset.id,
+          trigger: WorkflowTrigger.AssetCreate,
+        }),
+      ).resolves.toBeUndefined();
 
       await expect(ctx.get(AssetRepository).getById(asset.id)).resolves.toMatchObject({
         visibility: AssetVisibility.Timeline,
@@ -218,7 +242,13 @@ describe('core plugin', () => {
         steps: [{ method: 'immich-plugin-core#assetFavorite' }],
       });
 
-      await expect(ctx.sut.handleAssetTrigger({ workflowId: workflow.id, assetId: asset.id })).resolves.toBeUndefined();
+      await expect(
+        ctx.sut.handleAssetTrigger({
+          workflowId: workflow.id,
+          assetId: asset.id,
+          trigger: WorkflowTrigger.AssetCreate,
+        }),
+      ).resolves.toBeUndefined();
 
       await expect(ctx.get(AssetRepository).getById(asset.id)).resolves.toMatchObject({ isFavorite: true });
     });
@@ -233,7 +263,13 @@ describe('core plugin', () => {
         steps: [{ method: 'immich-plugin-core#assetFavorite', config: { inverse: true } }],
       });
 
-      await expect(ctx.sut.handleAssetTrigger({ workflowId: workflow.id, assetId: asset.id })).resolves.toBeUndefined();
+      await expect(
+        ctx.sut.handleAssetTrigger({
+          workflowId: workflow.id,
+          assetId: asset.id,
+          trigger: WorkflowTrigger.AssetCreate,
+        }),
+      ).resolves.toBeUndefined();
 
       await expect(ctx.get(AssetRepository).getById(asset.id)).resolves.toMatchObject({ isFavorite: false });
     });
@@ -250,7 +286,13 @@ describe('core plugin', () => {
         steps: [{ method: 'immich-plugin-core#assetAddToAlbums', config: { albumIds: [], albumName: 'Screenshots' } }],
       });
 
-      await expect(ctx.sut.handleAssetTrigger({ workflowId: workflow.id, assetId: asset.id })).resolves.toBeUndefined();
+      await expect(
+        ctx.sut.handleAssetTrigger({
+          workflowId: workflow.id,
+          assetId: asset.id,
+          trigger: WorkflowTrigger.AssetCreate,
+        }),
+      ).resolves.toBeUndefined();
 
       const albums = await ctx.get(AlbumRepository).getAll(user.id);
       expect(albums).toHaveLength(1);
@@ -281,7 +323,13 @@ describe('core plugin', () => {
       expect(albums).toHaveLength(1);
       expect(albums[0].albumName).toEqual(album.albumName);
 
-      await expect(ctx.sut.handleAssetTrigger({ workflowId: workflow.id, assetId: asset.id })).resolves.toBeUndefined();
+      await expect(
+        ctx.sut.handleAssetTrigger({
+          workflowId: workflow.id,
+          assetId: asset.id,
+          trigger: WorkflowTrigger.AssetCreate,
+        }),
+      ).resolves.toBeUndefined();
 
       await expect(ctx.get(AlbumRepository).getAssetIds(album.id, [asset.id])).resolves.toContain(asset.id);
     });
@@ -297,7 +345,13 @@ describe('core plugin', () => {
         steps: [{ method: 'immich-plugin-core#assetAddToAlbums', config: { albumIds: [album.id] } }],
       });
 
-      await expect(ctx.sut.handleAssetTrigger({ workflowId: workflow.id, assetId: asset.id })).resolves.toBeUndefined();
+      await expect(
+        ctx.sut.handleAssetTrigger({
+          workflowId: workflow.id,
+          assetId: asset.id,
+          trigger: WorkflowTrigger.AssetCreate,
+        }),
+      ).resolves.toBeUndefined();
 
       await expect(ctx.get(AlbumRepository).getAssetIds(album.id, [asset.id])).resolves.toContain(asset.id);
     });
@@ -316,7 +370,13 @@ describe('core plugin', () => {
         steps: [{ method: 'immich-plugin-core#assetAddToAlbums', config: { albumIds: [album1.id, album2.id] } }],
       });
 
-      await expect(ctx.sut.handleAssetTrigger({ workflowId: workflow.id, assetId: asset.id })).resolves.toBeUndefined();
+      await expect(
+        ctx.sut.handleAssetTrigger({
+          workflowId: workflow.id,
+          assetId: asset.id,
+          trigger: WorkflowTrigger.AssetCreate,
+        }),
+      ).resolves.toBeUndefined();
 
       await expect(ctx.get(AlbumRepository).getAssetIds(album1.id, [asset.id])).resolves.toContain(asset.id);
       await expect(ctx.get(AlbumRepository).getAssetIds(album2.id, [asset.id])).resolves.toContain(asset.id);
@@ -334,7 +394,13 @@ describe('core plugin', () => {
         steps: [{ method: 'immich-plugin-core#assetAddToAlbums', config: { albumIds: [album.id] } }],
       });
 
-      await expect(ctx.sut.handleAssetTrigger({ workflowId: workflow.id, assetId: asset.id })).resolves.toBeTruthy();
+      await expect(
+        ctx.sut.handleAssetTrigger({
+          workflowId: workflow.id,
+          assetId: asset.id,
+          trigger: WorkflowTrigger.AssetCreate,
+        }),
+      ).resolves.toBeTruthy();
 
       await expect(ctx.get(AlbumRepository).getAssetIds(album.id, [asset.id])).resolves.not.toContain(asset.id);
     });
@@ -360,7 +426,13 @@ describe('core plugin', () => {
         ],
       });
 
-      await expect(ctx.sut.handleAssetTrigger({ workflowId: workflow.id, assetId: asset.id })).resolves.toBeUndefined();
+      await expect(
+        ctx.sut.handleAssetTrigger({
+          workflowId: workflow.id,
+          assetId: asset.id,
+          trigger: WorkflowTrigger.AssetCreate,
+        }),
+      ).resolves.toBeUndefined();
       await expect(ctx.get(AssetRepository).getById(asset.id)).resolves.toMatchObject({ isFavorite: true });
     });
 
@@ -383,7 +455,13 @@ describe('core plugin', () => {
         ],
       });
 
-      await expect(ctx.sut.handleAssetTrigger({ workflowId: workflow.id, assetId: asset.id })).resolves.toBeUndefined();
+      await expect(
+        ctx.sut.handleAssetTrigger({
+          workflowId: workflow.id,
+          assetId: asset.id,
+          trigger: WorkflowTrigger.AssetMetadataExtraction,
+        }),
+      ).resolves.toBeUndefined();
       await expect(ctx.get(AssetRepository).getById(asset.id)).resolves.toMatchObject({ isFavorite: false });
     });
 
@@ -406,7 +484,13 @@ describe('core plugin', () => {
         ],
       });
 
-      await expect(ctx.sut.handleAssetTrigger({ workflowId: workflow.id, assetId: asset.id })).resolves.toBeUndefined();
+      await expect(
+        ctx.sut.handleAssetTrigger({
+          workflowId: workflow.id,
+          assetId: asset.id,
+          trigger: WorkflowTrigger.AssetMetadataExtraction,
+        }),
+      ).resolves.toBeUndefined();
       await expect(ctx.get(AssetRepository).getById(asset.id)).resolves.toMatchObject({ isFavorite: true });
     });
   });
@@ -434,10 +518,18 @@ describe('core plugin', () => {
       });
 
       await expect(
-        ctx.sut.handleAssetTrigger({ workflowId: workflow.id, assetId: asset1.id }),
+        ctx.sut.handleAssetTrigger({
+          workflowId: workflow.id,
+          assetId: asset1.id,
+          trigger: WorkflowTrigger.AssetCreate,
+        }),
       ).resolves.toBeUndefined();
       await expect(
-        ctx.sut.handleAssetTrigger({ workflowId: workflow.id, assetId: asset2.id }),
+        ctx.sut.handleAssetTrigger({
+          workflowId: workflow.id,
+          assetId: asset2.id,
+          trigger: WorkflowTrigger.AssetCreate,
+        }),
       ).resolves.toBeUndefined();
       await expect(ctx.get(AssetRepository).getById(asset1.id)).resolves.toMatchObject({ isFavorite: true });
       await expect(ctx.get(AssetRepository).getById(asset2.id)).resolves.toMatchObject({ isFavorite: true });
@@ -465,10 +557,18 @@ describe('core plugin', () => {
       });
 
       await expect(
-        ctx.sut.handleAssetTrigger({ workflowId: workflow.id, assetId: asset1.id }),
+        ctx.sut.handleAssetTrigger({
+          workflowId: workflow.id,
+          assetId: asset1.id,
+          trigger: WorkflowTrigger.AssetCreate,
+        }),
       ).resolves.toBeUndefined();
       await expect(
-        ctx.sut.handleAssetTrigger({ workflowId: workflow.id, assetId: asset2.id }),
+        ctx.sut.handleAssetTrigger({
+          workflowId: workflow.id,
+          assetId: asset2.id,
+          trigger: WorkflowTrigger.AssetCreate,
+        }),
       ).resolves.toBeUndefined();
       await expect(ctx.get(AssetRepository).getById(asset1.id)).resolves.toMatchObject({ isFavorite: true });
       await expect(ctx.get(AssetRepository).getById(asset2.id)).resolves.toMatchObject({ isFavorite: false });
@@ -496,10 +596,18 @@ describe('core plugin', () => {
       });
 
       await expect(
-        ctx.sut.handleAssetTrigger({ workflowId: workflow.id, assetId: asset1.id }),
+        ctx.sut.handleAssetTrigger({
+          workflowId: workflow.id,
+          assetId: asset1.id,
+          trigger: WorkflowTrigger.AssetCreate,
+        }),
       ).resolves.toBeUndefined();
       await expect(
-        ctx.sut.handleAssetTrigger({ workflowId: workflow.id, assetId: asset2.id }),
+        ctx.sut.handleAssetTrigger({
+          workflowId: workflow.id,
+          assetId: asset2.id,
+          trigger: WorkflowTrigger.AssetCreate,
+        }),
       ).resolves.toBeUndefined();
       await expect(ctx.get(AssetRepository).getById(asset1.id)).resolves.toMatchObject({ isFavorite: true });
       await expect(ctx.get(AssetRepository).getById(asset2.id)).resolves.toMatchObject({ isFavorite: false });
@@ -525,7 +633,13 @@ describe('core plugin', () => {
         ],
       });
 
-      await expect(ctx.sut.handleAssetTrigger({ workflowId: workflow.id, assetId: asset.id })).resolves.toBeUndefined();
+      await expect(
+        ctx.sut.handleAssetTrigger({
+          workflowId: workflow.id,
+          assetId: asset.id,
+          trigger: WorkflowTrigger.AssetCreate,
+        }),
+      ).resolves.toBeUndefined();
       await expect(ctx.get(AssetRepository).getById(asset.id)).resolves.toMatchObject({ isFavorite: true });
     });
   });
@@ -663,13 +777,25 @@ describe('core plugin', () => {
         ],
       });
 
-      await ctx.sut.handleAssetTrigger({ workflowId: workflow.id, assetId: asset1.id });
+      await ctx.sut.handleAssetTrigger({
+        workflowId: workflow.id,
+        assetId: asset1.id,
+        trigger: WorkflowTrigger.AssetCreate,
+      });
       await expect(ctx.get(AssetRepository).getById(asset1.id)).resolves.toMatchObject({ isFavorite: true });
 
-      await ctx.sut.handleAssetTrigger({ workflowId: workflow.id, assetId: asset2.id });
+      await ctx.sut.handleAssetTrigger({
+        workflowId: workflow.id,
+        assetId: asset2.id,
+        trigger: WorkflowTrigger.AssetCreate,
+      });
       await expect(ctx.get(AssetRepository).getById(asset2.id)).resolves.toMatchObject({ isFavorite: true });
 
-      await ctx.sut.handleAssetTrigger({ workflowId: workflow.id, assetId: asset3.id });
+      await ctx.sut.handleAssetTrigger({
+        workflowId: workflow.id,
+        assetId: asset3.id,
+        trigger: WorkflowTrigger.AssetCreate,
+      });
       await expect(ctx.get(AssetRepository).getById(asset3.id)).resolves.toMatchObject({ isFavorite: false });
     });
 
@@ -700,17 +826,157 @@ describe('core plugin', () => {
         ],
       });
 
-      await ctx.sut.handleAssetTrigger({ workflowId: workflow.id, assetId: asset1.id });
+      await ctx.sut.handleAssetTrigger({
+        workflowId: workflow.id,
+        assetId: asset1.id,
+        trigger: WorkflowTrigger.AssetCreate,
+      });
       await expect(ctx.get(AssetRepository).getById(asset1.id)).resolves.toMatchObject({ isFavorite: true });
 
-      await ctx.sut.handleAssetTrigger({ workflowId: workflow.id, assetId: asset2.id });
+      await ctx.sut.handleAssetTrigger({
+        workflowId: workflow.id,
+        assetId: asset2.id,
+        trigger: WorkflowTrigger.AssetCreate,
+      });
       await expect(ctx.get(AssetRepository).getById(asset2.id)).resolves.toMatchObject({ isFavorite: true });
 
-      await ctx.sut.handleAssetTrigger({ workflowId: workflow.id, assetId: asset3.id });
+      await ctx.sut.handleAssetTrigger({
+        workflowId: workflow.id,
+        assetId: asset3.id,
+        trigger: WorkflowTrigger.AssetCreate,
+      });
       await expect(ctx.get(AssetRepository).getById(asset3.id)).resolves.toMatchObject({ isFavorite: false });
     });
   });
 
+  describe('assetOcrFilter', () => {
+    it('should match assets case-insensitively', async () => {
+      const { user } = await ctx.newUser();
+      const [{ asset: asset1 }, { asset: asset2 }] = await Promise.all([
+        ctx.newAsset({ ownerId: user.id }),
+        ctx.newAsset({ ownerId: user.id }),
+      ]);
+      await ctx.newAssetOcr([{ assetId: asset1.id, text: 'test', textScore: 0.9 }]);
+      await ctx.newAssetOcr([{ assetId: asset2.id, text: 'bogus', textScore: 0.9 }]);
+
+      const workflow = await createWorkflow({
+        ownerId: user.id,
+        trigger: WorkflowTrigger.AssetOcrExtraction,
+        steps: [
+          {
+            method: 'immich-plugin-core#assetOcrFilter',
+            config: { matchType: 'contains', matchScope: 'box', pattern: 'Test', minScore: 0 },
+          },
+          {
+            method: 'immich-plugin-core#assetFavorite',
+          },
+        ],
+      });
+
+      await expect(
+        ctx.sut.handleAssetTrigger({
+          workflowId: workflow.id,
+          assetId: asset1.id,
+          trigger: WorkflowTrigger.AssetOcrExtraction,
+        }),
+      ).resolves.toBeUndefined();
+      await expect(
+        ctx.sut.handleAssetTrigger({
+          workflowId: workflow.id,
+          assetId: asset2.id,
+          trigger: WorkflowTrigger.AssetOcrExtraction,
+        }),
+      ).resolves.toBeUndefined();
+      await expect(ctx.get(AssetRepository).getById(asset1.id)).resolves.toMatchObject({ isFavorite: true });
+      await expect(ctx.get(AssetRepository).getById(asset2.id)).resolves.toMatchObject({ isFavorite: false });
+    });
+
+    it('should match assets by regex in full text', async () => {
+      const { user } = await ctx.newUser();
+      const [{ asset: asset1 }, { asset: asset2 }] = await Promise.all([
+        ctx.newAsset({ ownerId: user.id }),
+        ctx.newAsset({ ownerId: user.id }),
+      ]);
+      await ctx.newAssetOcr([
+        { assetId: asset1.id, text: 'Test', textScore: 0.9 },
+        { assetId: asset1.id, text: 'Match', textScore: 0.5 },
+      ]);
+      await ctx.newAssetOcr([{ assetId: asset2.id, text: 'Test', textScore: 0.9 }]);
+
+      const workflow = await createWorkflow({
+        ownerId: user.id,
+        trigger: WorkflowTrigger.AssetCreate,
+        steps: [
+          {
+            method: 'immich-plugin-core#assetOcrFilter',
+            config: { matchType: 'regex', matchScope: 'full', pattern: 'Test.+Match', minScore: 0 },
+          },
+          {
+            method: 'immich-plugin-core#assetFavorite',
+          },
+        ],
+      });
+
+      await expect(
+        ctx.sut.handleAssetTrigger({
+          workflowId: workflow.id,
+          assetId: asset1.id,
+          trigger: WorkflowTrigger.AssetOcrExtraction,
+        }),
+      ).resolves.toBeUndefined();
+      await expect(
+        ctx.sut.handleAssetTrigger({
+          workflowId: workflow.id,
+          assetId: asset2.id,
+          trigger: WorkflowTrigger.AssetOcrExtraction,
+        }),
+      ).resolves.toBeUndefined();
+      await expect(ctx.get(AssetRepository).getById(asset1.id)).resolves.toMatchObject({ isFavorite: true });
+      await expect(ctx.get(AssetRepository).getById(asset2.id)).resolves.toMatchObject({ isFavorite: false });
+    });
+
+    it('should filter assets with low score', async () => {
+      const { user } = await ctx.newUser();
+      const [{ asset: asset1 }, { asset: asset2 }] = await Promise.all([
+        ctx.newAsset({ ownerId: user.id }),
+        ctx.newAsset({ ownerId: user.id }),
+      ]);
+      await ctx.newAssetOcr([{ assetId: asset1.id, text: 'test', textScore: 0.9 }]);
+      await ctx.newAssetOcr([{ assetId: asset2.id, text: 'test', textScore: 0.5 }]);
+
+      const workflow = await createWorkflow({
+        ownerId: user.id,
+        trigger: WorkflowTrigger.AssetOcrExtraction,
+        steps: [
+          {
+            method: 'immich-plugin-core#assetOcrFilter',
+            config: { matchType: 'contains', matchScope: 'box', pattern: 'test', minScore: 0.8 },
+          },
+          {
+            method: 'immich-plugin-core#assetFavorite',
+          },
+        ],
+      });
+
+      await expect(
+        ctx.sut.handleAssetTrigger({
+          workflowId: workflow.id,
+          assetId: asset1.id,
+          trigger: WorkflowTrigger.AssetOcrExtraction,
+        }),
+      ).resolves.toBeUndefined();
+      await expect(
+        ctx.sut.handleAssetTrigger({
+          workflowId: workflow.id,
+          assetId: asset2.id,
+          trigger: WorkflowTrigger.AssetOcrExtraction,
+        }),
+      ).resolves.toBeUndefined();
+      await expect(ctx.get(AssetRepository).getById(asset1.id)).resolves.toMatchObject({ isFavorite: true });
+      await expect(ctx.get(AssetRepository).getById(asset2.id)).resolves.toMatchObject({ isFavorite: false });
+    });
+  });
+
   describe('assetTagFilter', () => {
     it('should favorite asset if it matches any tag', async () => {
       const { user } = await ctx.newUser();
@@ -732,7 +998,13 @@ describe('core plugin', () => {
         ],
       });
 
-      await expect(ctx.sut.handleAssetTrigger({ workflowId: workflow.id, assetId: asset.id })).resolves.toBeUndefined();
+      await expect(
+        ctx.sut.handleAssetTrigger({
+          workflowId: workflow.id,
+          assetId: asset.id,
+          trigger: WorkflowTrigger.AssetCreate,
+        }),
+      ).resolves.toBeUndefined();
       await expect(ctx.get(AssetRepository).getById(asset.id)).resolves.toMatchObject({ isFavorite: true });
     });
 
@@ -755,7 +1027,13 @@ describe('core plugin', () => {
         ],
       });
 
-      await expect(ctx.sut.handleAssetTrigger({ workflowId: workflow.id, assetId: asset.id })).resolves.toBeUndefined();
+      await expect(
+        ctx.sut.handleAssetTrigger({
+          workflowId: workflow.id,
+          assetId: asset.id,
+          trigger: WorkflowTrigger.AssetCreate,
+        }),
+      ).resolves.toBeUndefined();
       await expect(ctx.get(AssetRepository).getById(asset.id)).resolves.toMatchObject({ isFavorite: false });
     });
 
@@ -779,7 +1057,13 @@ describe('core plugin', () => {
         ],
       });
 
-      await expect(ctx.sut.handleAssetTrigger({ workflowId: workflow.id, assetId: asset.id })).resolves.toBeUndefined();
+      await expect(
+        ctx.sut.handleAssetTrigger({
+          workflowId: workflow.id,
+          assetId: asset.id,
+          trigger: WorkflowTrigger.AssetCreate,
+        }),
+      ).resolves.toBeUndefined();
       await expect(ctx.get(AssetRepository).getById(asset.id)).resolves.toMatchObject({ isFavorite: true });
     });
 
@@ -802,7 +1086,13 @@ describe('core plugin', () => {
         ],
       });
 
-      await expect(ctx.sut.handleAssetTrigger({ workflowId: workflow.id, assetId: asset.id })).resolves.toBeUndefined();
+      await expect(
+        ctx.sut.handleAssetTrigger({
+          workflowId: workflow.id,
+          assetId: asset.id,
+          trigger: WorkflowTrigger.AssetCreate,
+        }),
+      ).resolves.toBeUndefined();
       await expect(ctx.get(AssetRepository).getById(asset.id)).resolves.toMatchObject({ isFavorite: false });
     });
 
@@ -826,7 +1116,13 @@ describe('core plugin', () => {
         ],
       });
 
-      await expect(ctx.sut.handleAssetTrigger({ workflowId: workflow.id, assetId: asset.id })).resolves.toBeUndefined();
+      await expect(
+        ctx.sut.handleAssetTrigger({
+          workflowId: workflow.id,
+          assetId: asset.id,
+          trigger: WorkflowTrigger.AssetCreate,
+        }),
+      ).resolves.toBeUndefined();
       await expect(ctx.get(AssetRepository).getById(asset.id)).resolves.toMatchObject({ isFavorite: true });
     });
   });
@@ -850,7 +1146,13 @@ describe('core plugin', () => {
         ],
       });
 
-      await expect(ctx.sut.handleAssetTrigger({ workflowId: workflow.id, assetId: asset.id })).resolves.toBeUndefined();
+      await expect(
+        ctx.sut.handleAssetTrigger({
+          workflowId: workflow.id,
+          assetId: asset.id,
+          trigger: WorkflowTrigger.AssetCreate,
+        }),
+      ).resolves.toBeUndefined();
       expect(fetchMock).toHaveBeenCalled();
     });
 
@@ -887,7 +1189,13 @@ describe('core plugin', () => {
           .execute();
       }
 
-      await expect(ctx.sut.handleAssetTrigger({ workflowId: workflow.id, assetId: asset.id })).resolves.toBeUndefined();
+      await expect(
+        ctx.sut.handleAssetTrigger({
+          workflowId: workflow.id,
+          assetId: asset.id,
+          trigger: WorkflowTrigger.AssetCreate,
+        }),
+      ).resolves.toBeUndefined();
       await expect(ctx.get(AssetRepository).getById(asset.id)).resolves.toMatchObject({ isFavorite: false });
     });
   });

web/src/lib/utils/workflow.ts

@@ -13,6 +13,9 @@ export const getTriggerName = ($t: MessageFormatter, type: WorkflowTrigger) => {
     case WorkflowTrigger.AssetMetadataExtraction: {
       return $t('trigger_asset_metadata_extraction');
     }
+    case WorkflowTrigger.AssetOcrExtraction: {
+      return $t('trigger_asset_ocr_extraction');
+    }
     case WorkflowTrigger.AssetTagged: {
       return $t('trigger_asset_tagged');
     }
@@ -33,6 +36,9 @@ export const getTriggerDescription = ($t: MessageFormatter, type: WorkflowTrigge
     case WorkflowTrigger.AssetMetadataExtraction: {
       return $t('trigger_asset_metadata_extraction_description');
     }
+    case WorkflowTrigger.AssetOcrExtraction: {
+      return $t('trigger_asset_ocr_extraction_description');
+    }
     case WorkflowTrigger.AssetTagged: {
       return $t('trigger_asset_tagged_description');
     }