Chore: Added test for OCR search

packages/lib/services/ocr/OcrService.test.ts

@@ -1,18 +1,10 @@
-import { createNoteAndResource, ocrSampleDir, resourceFetcher, setupDatabaseAndSynchronizer, supportDir, switchClient, synchronizerStart } from '../../testing/test-utils';
+import { createNoteAndResource, newOcrService, ocrSampleDir, resourceFetcher, setupDatabaseAndSynchronizer, supportDir, switchClient, synchronizerStart } from '../../testing/test-utils';
-import OcrDriverTesseract from './drivers/OcrDriverTesseract';
-import OcrService from './OcrService';
 import { supportedMimeTypes } from './OcrService';
-import { createWorker } from 'tesseract.js';
 import Resource from '../../models/Resource';
 import { ResourceEntity, ResourceOcrStatus } from '../database/types';
 import { msleep } from '@joplin/utils/time';
 import Logger from '@joplin/utils/Logger';
 
-const newService = () => {
-	const driver = new OcrDriverTesseract({ createWorker });
-	return new OcrService(driver);
-};
-
 describe('OcrService', () => {
 
 	beforeEach(async () => {
@@ -31,7 +23,7 @@ describe('OcrService', () => {
 
 		expect(await Resource.needOcrCount(supportedMimeTypes)).toBe(3);
 
-		const service = newService();
+		const service = newOcrService();
 		await service.processResources();
 
 		const expectedText = 'This is a lot of 12 point text to test the\n' +
@@ -88,7 +80,7 @@ describe('OcrService', () => {
 	it('should process PDF resources', async () => {
 		const { resource } = await createNoteAndResource({ path: `${ocrSampleDir}/dummy.pdf` });
 
-		const service = newService();
+		const service = newOcrService();
 
 		await service.processResources();
 
@@ -111,7 +103,7 @@ describe('OcrService', () => {
 
 		await msleep(1);
 
-		const service = newService();
+		const service = newOcrService();
 
 		await service.processResources();
 
@@ -157,7 +149,7 @@ describe('OcrService', () => {
 			fetch_error: 'cannot be downloaded',
 		});
 
-		const service = newService();
+		const service = newOcrService();
 
 		// The service will print a warnign so we disable it in tests
 		Logger.globalLogger.enabled = false;
@@ -196,7 +188,7 @@ describe('OcrService', () => {
 	it('should handle conflicts if two clients process the same resource then sync', async () => {
 		await createNoteAndResource({ path: `${ocrSampleDir}/dummy.pdf` });
 
-		const service1 = newService();
+		const service1 = newOcrService();
 		await synchronizerStart();
 		await service1.processResources();
 
@@ -205,7 +197,7 @@ describe('OcrService', () => {
 		await synchronizerStart();
 		await msleep(1);
 		await resourceFetcher().startAndWait();
-		const service2 = newService();
+		const service2 = newOcrService();
 		await service2.processResources();
 		await synchronizerStart();
 		const expectedResouceUpatedTime = (await Resource.all())[0].updated_time;
@@ -236,7 +228,7 @@ describe('OcrService', () => {
 	// it('should process resources 2', async () => {
 	// 	await createNoteAndResource({ path: `${require('os').homedir()}/Desktop/AllClients.png` });
 
-	// 	const service = newService();
+	// 	const service = newOcrService();
 	// 	await service.processResources();
 
 	// 	console.info(await Resource.all());

packages/lib/services/search/SearchEngine.resources.test.ts

@@ -1,5 +1,8 @@
+import { ModelType } from '../../BaseModel';
+import Note from '../../models/Note';
 import Resource from '../../models/Resource';
-import { db, msleep, setupDatabaseAndSynchronizer, switchClient } from '../../testing/test-utils';
+import shim from '../../shim';
+import { db, msleep, newOcrService, ocrSampleDir, resourceService, setupDatabaseAndSynchronizer, switchClient } from '../../testing/test-utils';
 import { ResourceOcrStatus } from '../database/types';
 import SearchEngine from './SearchEngine';
 
@@ -35,6 +38,27 @@ describe('SearchEngine.resources', () => {
 		expect(normalized[0].body).toBe('hello, how are you ?');
 	});
 
+	it('should return notes associated with indexed resources', (async () => {
+		const note1 = await Note.save({});
+		await Note.save({});
+		await shim.attachFileToNote(note1, `${ocrSampleDir}/testocr.png`);
+		const resource = (await Resource.all())[0];
+
+		await resourceService().indexNoteResources();
+
+		const ocrService = newOcrService();
+		await ocrService.processResources();
+
+		const searchEngine = newSearchEngine();
+		await searchEngine.syncTables();
+
+		const results = await searchEngine.search('lazy fox');
+		expect(results.length).toBe(1);
+		expect(results[0].id).toBe(note1.id);
+		expect(results[0].item_id).toBe(resource.id);
+		expect(results[0].item_type).toBe(ModelType.Resource);
+	}));
+
 	it('should delete normalized data when a resource is deleted', async () => {
 		const engine = newSearchEngine();
 

packages/lib/services/search/SearchEngine.test.ts

@@ -534,4 +534,5 @@ describe('services/SearchEngine', () => {
 		expect((await engine.search(n2.id))[0].id).toBe(n2.id);
 		expect(await engine.search(f1.id)).toEqual([]);
 	}));
+
 });

packages/lib/testing/test-utils.ts

@@ -63,6 +63,9 @@ const { Dirnames } = require('../services/synchronizer/utils/types');
 import RSA from '../services/e2ee/RSA.node';
 import { State as ShareState } from '../services/share/reducer';
 import initLib from '../initLib';
+import OcrDriverTesseract from '../services/ocr/drivers/OcrDriverTesseract';
+import OcrService from '../services/ocr/OcrService';
+import { createWorker } from 'tesseract.js';
 
 // Each suite has its own separate data and temp directory so that multiple
 // suites can be run at the same time. suiteName is what is used to
@@ -1027,4 +1030,9 @@ const simulateReadOnlyShareEnv = (shareId: string) => {
 	};
 };
 
+export const newOcrService = () => {
+	const driver = new OcrDriverTesseract({ createWorker });
+	return new OcrService(driver);
+};
+
 export { supportDir, createNoteAndResource, createTempFile, createTestShareData, simulateReadOnlyShareEnv, waitForFolderCount, afterAllCleanUp, exportDir, synchronizerStart, afterEachCleanUp, syncTargetName, setSyncTargetName, syncDir, createTempDir, isNetworkSyncTarget, kvStore, expectThrow, logger, expectNotThrow, resourceService, resourceFetcher, tempFilePath, allSyncTargetItemsEncrypted, msleep, setupDatabase, revisionService, setupDatabaseAndSynchronizer, db, synchronizer, fileApi, sleep, clearDatabase, switchClient, syncTargetId, objectsEqual, checkThrowAsync, checkThrow, encryptionService, loadEncryptionMasterKey, fileContentEqual, decryptionWorker, currentClientId, id, ids, sortedIds, at, createNTestNotes, createNTestFolders, createNTestTags, TestApp };