Desktop, Mobile, Cli: Fixes #16792: Avoid loading every matching note body in search_notes

Load bodies only for the limited search results while keeping the full match count. Match bodies by ID while iterating the ranked results because byIds does not guarantee row order.

Validation could not run: dependency installation failed with a server-aborted request, leaving Yarn without a node_modules state file.

Assisted-by: GPT-6.1 Sol

Author: anshurajbisoyi98-ctrl

.gitignore

@@ -1672,6 +1672,7 @@ packages/lib/services/ai/tools/global/manageTags.js
 packages/lib/services/ai/tools/global/readImage.js
 packages/lib/services/ai/tools/global/readNote.js
 packages/lib/services/ai/tools/global/searchNotes.js
+packages/lib/services/ai/tools/global/searchNotes.test.js
 packages/lib/services/ai/tools/global/semanticSearchNotes.js
 packages/lib/services/ai/tools/global/updateNote.js
 packages/lib/services/ai/tools/types.js

.ignore.eslint

@@ -1698,6 +1698,7 @@ packages/lib/services/ai/tools/global/manageTags.js
 packages/lib/services/ai/tools/global/readImage.js
 packages/lib/services/ai/tools/global/readNote.js
 packages/lib/services/ai/tools/global/searchNotes.js
+packages/lib/services/ai/tools/global/searchNotes.test.js
 packages/lib/services/ai/tools/global/semanticSearchNotes.js
 packages/lib/services/ai/tools/global/updateNote.js
 packages/lib/services/ai/tools/types.js

packages/lib/services/ai/tools/global/searchNotes.test.ts

@@ -0,0 +1,68 @@
+import Note from '../../../../models/Note';
+import { db, setupDatabaseAndSynchronizer, switchClient } from '../../../../testing/test-utils';
+import SearchEngine from '../../../search/SearchEngine';
+import SearchEngineUtils from '../../../search/SearchEngineUtils';
+import searchNotes from './searchNotes';
+
+interface SearchOutput {
+	results: { id: string; snippet: string }[];
+	total: number;
+}
+
+const runSearch = async (limit = 20) => await searchNotes.handler({ query: 'keyword', limit }, {}) as SearchOutput;
+
+describe('search_notes', () => {
+	beforeEach(async () => {
+		await setupDatabaseAndSynchronizer(1);
+		await switchClient(1);
+		SearchEngine.instance().setDb(db());
+	});
+
+	afterEach(() => {
+		jest.restoreAllMocks();
+	});
+
+	const createNotes = async () => {
+		const notes = [];
+		for (const [title, body] of [['Other', 'keyword once'], ['keyword', 'keyword keyword repeated'], ['Last', 'keyword last']]) {
+			notes.push(await Note.save({ id: String(notes.length + 1).padStart(32, '0'), title, body }));
+		}
+		await SearchEngine.instance().syncTables();
+		return notes;
+	};
+
+	test('preserves ranking and matches snippets to bodies returned in a different order', async () => {
+		const notes = await createNotes();
+		const ranked = (await SearchEngineUtils.notesForQuery('keyword', false, { fields: ['id'] })).notes;
+		expect(ranked.map(n => n.id)).not.toEqual(notes.map(n => n.id));
+		expect(ranked.map(n => n.id)).not.toEqual(notes.map(n => n.id).sort());
+		const byIds = Note.byIds.bind(Note);
+		jest.spyOn(Note, 'byIds').mockImplementation(async (ids, options) => {
+			const rows = await byIds(ids, options);
+			return ids.slice().reverse().map(id => rows.find(n => n.id === id));
+		});
+		const output = await runSearch();
+		expect(output.results.map(n => n.id)).toEqual(ranked.map(n => n.id));
+		for (const result of output.results) {
+			expect(result.snippet).toBe(notes.find(n => n.id === result.id).body);
+		}
+	});
+
+	test('limits results while keeping the full match count', async () => {
+		await createNotes();
+		const output = await runSearch(1);
+		expect(output.results).toHaveLength(1);
+		expect(output.total).toBe(3);
+	});
+
+	test('fetches bodies only for the limited results and handles a missing body', async () => {
+		await createNotes();
+		const querySpy = jest.spyOn(SearchEngineUtils, 'notesForQuery');
+		const bodySpy = jest.spyOn(Note, 'byIds').mockResolvedValue([]);
+		const output = await runSearch(1);
+		expect(querySpy.mock.calls[0][2].fields).not.toContain('body');
+		expect(bodySpy).toHaveBeenCalledWith([output.results[0].id], { fields: ['id', 'body'] });
+		expect(output.results[0].snippet).toBe('');
+		expect(output.total).toBe(3);
+	});
+});

packages/lib/services/ai/tools/global/searchNotes.ts

@@ -3,13 +3,14 @@ import { NoteEntity } from '../../../database/types';
 import { _ } from '../../../../locale';
 import buildTool from '../utils/buildTool';
 import { ToolError } from '../types';
+import Note from '../../../../models/Note';
 
 interface Input {
 	query?: string;
 	limit?: number;
 }
 
-const fields = ['id', 'title', 'parent_id', 'updated_time', 'body'];
+const fields = ['id', 'title', 'parent_id', 'updated_time'];
 const defaultLimit = 20;
 const maxLimit = 100;
 const snippetChars = 240;
@@ -57,6 +58,10 @@ const tool = buildTool({
 		const limit = Math.min(Math.max(input.limit ?? defaultLimit, 1), maxLimit);
 		const { notes } = await SearchEngineUtils.notesForQuery(input.query, false, { fields });
 
+		const limitedNotes = notes.slice(0, limit);
+		const bodies = await Note.byIds(limitedNotes.map(n => n.id), { fields: ['id', 'body'] });
+		const bodyById = new Map(bodies.map(n => [n.id, n.body]));
+
 		// Pull keywords out of the query so we can anchor the snippet near a
 		// match. Filters like `notebook:"X"` aren't useful for that.
 		const keywords = input.query
@@ -65,12 +70,12 @@ const tool = buildTool({
 			.map(t => t.replace(/^["*]+|["*]+$/g, '').toLowerCase())
 			.filter(Boolean);
 
-		const results = notes.slice(0, limit).map((n: NoteEntity) => ({
+		const results = limitedNotes.map((n: NoteEntity) => ({
 			id: n.id,
 			title: n.title,
 			notebook_id: n.parent_id,
 			updated_time: n.updated_time,
-			snippet: makeSnippet(n.body ?? '', keywords),
+			snippet: makeSnippet(bodyById.get(n.id) ?? '', keywords),
 		}));
 
 		return { results, total: notes.length };