vectra 0.12.2 → 0.14.0

package/lib/LocalIndex.spec.js

@@ -13,33 +13,269 @@ var __importDefault = (this && this.__importDefault) || function (mod) {
 };
 Object.defineProperty(exports, "__esModule", { value: true });
 const node_assert_1 = __importDefault(require("node:assert"));
+const sinon_1 = __importDefault(require("sinon"));
 const LocalIndex_1 = require("./LocalIndex");
 const promises_1 = __importDefault(require("fs/promises"));
 const path_1 = __importDefault(require("path"));
+const storage_1 = require("./storage");
+const LocalDocument_1 = require("./LocalDocument");
 describe('LocalIndex', () => {
     const testIndexDir = path_1.default.join(__dirname, 'test_index');
+    const basicIndexItems = [
+        { id: '1', vector: [1, 2, 3] },
+        { id: '2', vector: [2, 3, 4] },
+        { id: '3', vector: [3, 4, 5] }
+    ];
     beforeEach(() => __awaiter(void 0, void 0, void 0, function* () {
         yield promises_1.default.rm(testIndexDir, { recursive: true, force: true });
     }));
     afterEach(() => __awaiter(void 0, void 0, void 0, function* () {
         yield promises_1.default.rm(testIndexDir, { recursive: true, force: true });
+        sinon_1.default.restore();
     }));
     it('should create a new index', () => __awaiter(void 0, void 0, void 0, function* () {
         const index = new LocalIndex_1.LocalIndex(testIndexDir);
         yield index.createIndex();
         const created = yield index.isIndexCreated();
         node_assert_1.default.equal(created, true);
+        node_assert_1.default.equal(index.folderPath, testIndexDir);
     }));
+    it('exposes getters indexName and storage', () => __awaiter(void 0, void 0, void 0, function* () {
+        const index = new LocalIndex_1.LocalIndex(testIndexDir);
+        node_assert_1.default.equal(index.indexName, 'index.json');
+        node_assert_1.default.ok(index.storage);
+    }));
+    it('blocks concurrent operations when lock is held', () => __awaiter(void 0, void 0, void 0, function* () {
+        const index = new LocalIndex_1.LocalIndex(testIndexDir);
+        yield index.createIndex();
+        yield index.beginUpdate();
+        yield node_assert_1.default.rejects(() => __awaiter(void 0, void 0, void 0, function* () {
+            yield index.beginUpdate();
+        }), new Error('Update already in progress'));
+    }));
+    describe('createIndex', () => {
+        it('checks for existing index on creation', () => __awaiter(void 0, void 0, void 0, function* () {
+            const index = new LocalIndex_1.LocalIndex(testIndexDir);
+            yield index.createIndex();
+            yield node_assert_1.default.rejects(() => __awaiter(void 0, void 0, void 0, function* () {
+                yield index.createIndex();
+            }), new Error('Index already exists'));
+            yield index.insertItem({ id: '1', vector: [1, 2, 3] });
+            const lengthBefore = (yield index.listItems()).length;
+            node_assert_1.default.equal(lengthBefore, 1);
+            yield index.createIndex({ deleteIfExists: true, version: 2, metadata_config: {} });
+            const lengthAfter = (yield index.listItems()).length;
+            node_assert_1.default.equal(lengthAfter, 0);
+        }));
+        it('delete index if file creation fails', () => __awaiter(void 0, void 0, void 0, function* () {
+            const index = new LocalIndex_1.LocalIndex(testIndexDir);
+            sinon_1.default.stub(promises_1.default, 'writeFile').rejects(new Error('fs error'));
+            yield node_assert_1.default.rejects(() => __awaiter(void 0, void 0, void 0, function* () {
+                yield index.createIndex();
+            }), new Error('Error creating index'));
+            yield node_assert_1.default.rejects(() => __awaiter(void 0, void 0, void 0, function* () {
+                yield index.listItems();
+            }));
+        }));
+        it('persists version and metadata_config', () => __awaiter(void 0, void 0, void 0, function* () {
+            const index = new LocalIndex_1.LocalIndex(testIndexDir);
+            yield index.createIndex({ deleteIfExists: true, version: 2, metadata_config: { indexed: [] } });
+            const stats = yield index.getIndexStats();
+            node_assert_1.default.deepStrictEqual(stats, { version: 2, metadata_config: { indexed: [] }, items: 0 });
+        }));
+    });
+    describe('deleteIndex and load guard', () => {
+        it('listItems before createIndex rejects with not-exist error', () => __awaiter(void 0, void 0, void 0, function* () {
+            const index = new LocalIndex_1.LocalIndex(testIndexDir);
+            yield node_assert_1.default.rejects(() => __awaiter(void 0, void 0, void 0, function* () { return index.listItems(); }), new Error('Index does not exist'));
+        }));
+        it('deleteIndex removes the index and prevents further reads', () => __awaiter(void 0, void 0, void 0, function* () {
+            const index = new LocalIndex_1.LocalIndex(testIndexDir);
+            yield index.createIndex();
+            yield index.insertItem({ id: 'a', vector: [1] });
+            node_assert_1.default.equal(yield index.isIndexCreated(), true);
+            yield index.deleteIndex();
+            node_assert_1.default.equal(yield index.isIndexCreated(), false);
+            yield node_assert_1.default.rejects(() => __awaiter(void 0, void 0, void 0, function* () { return index.listItems(); }), new Error('Index does not exist'));
+        }));
+    });
+    describe('insert/upsert validation and behavior', () => {
+        it('insertItem validation: requires vector', () => __awaiter(void 0, void 0, void 0, function* () {
+            const index = new LocalIndex_1.LocalIndex(testIndexDir);
+            yield index.createIndex();
+            yield index.beginUpdate();
+            yield node_assert_1.default.rejects(() => __awaiter(void 0, void 0, void 0, function* () { return index.insertItem({ id: 'x' }); }), new Error('Vector is required'));
+            yield index.cancelUpdate();
+        }));
+        it('upsertItem inserts when not existing', () => __awaiter(void 0, void 0, void 0, function* () {
+            const index = new LocalIndex_1.LocalIndex(testIndexDir);
+            yield index.createIndex();
+            const inserted = yield index.upsertItem({ id: 'up', vector: [9, 9] });
+            node_assert_1.default.equal(inserted.id, 'up');
+            node_assert_1.default.deepStrictEqual(inserted.vector, [9, 9]);
+            const all = yield index.listItems();
+            node_assert_1.default.equal(all.length, 1);
+            node_assert_1.default.equal(all[0].id, 'up');
+        }));
+        it('upsertItem replaces existing item contents', () => __awaiter(void 0, void 0, void 0, function* () {
+            const index = new LocalIndex_1.LocalIndex(testIndexDir);
+            yield index.createIndex();
+            yield index.insertItem({ id: 'same', vector: [1], metadata: { a: 1 } });
+            const updated = yield index.upsertItem({ id: 'same', vector: [2], metadata: { a: 2 } });
+            node_assert_1.default.equal(updated.id, 'same');
+            node_assert_1.default.deepStrictEqual(updated.vector, [2]);
+            node_assert_1.default.deepStrictEqual(updated.metadata, { a: 2 });
+            const reread = yield index.getItem('same');
+            node_assert_1.default.deepStrictEqual(reread === null || reread === void 0 ? void 0 : reread.vector, [2]);
+            node_assert_1.default.deepStrictEqual(reread === null || reread === void 0 ? void 0 : reread.metadata, { a: 2 });
+        }));
+        it('auto-generates id when not provided and does not create metadataFile with no metadata even if indexed config exists', () => __awaiter(void 0, void 0, void 0, function* () {
+            const index = new LocalIndex_1.LocalIndex(testIndexDir);
+            yield index.createIndex({ version: 1, metadata_config: { indexed: ['keep'] } });
+            const inserted = yield index.insertItem({ vector: [1, 2, 3] });
+            node_assert_1.default.ok(inserted.id && typeof inserted.id === 'string');
+            const reread = yield index.getItem(inserted.id);
+            node_assert_1.default.deepStrictEqual(reread === null || reread === void 0 ? void 0 : reread.metadata, {});
+            node_assert_1.default.equal(reread.metadataFile, undefined);
+        }));
+    });
+    describe('cancelUpdate', () => {
+        it('discards staged changes', () => __awaiter(void 0, void 0, void 0, function* () {
+            const index = new LocalIndex_1.LocalIndex(testIndexDir);
+            yield index.createIndex();
+            yield index.beginUpdate();
+            yield index.insertItem({ id: 'temp', vector: [1, 1, 1] });
+            index.cancelUpdate();
+            const items = yield index.listItems();
+            node_assert_1.default.deepStrictEqual(items, []);
+        }));
+    });
+    describe('isIndexCreated transitions', () => {
+        it('reports false -> true -> false across lifecycle', () => __awaiter(void 0, void 0, void 0, function* () {
+            const index = new LocalIndex_1.LocalIndex(testIndexDir);
+            node_assert_1.default.equal(yield index.isIndexCreated(), false);
+            yield index.createIndex();
+            node_assert_1.default.equal(yield index.isIndexCreated(), true);
+            yield index.deleteIndex();
+            node_assert_1.default.equal(yield index.isIndexCreated(), false);
+        }));
+    });
+    describe('deleteItem', () => {
+        it('does nothing when id not found', () => __awaiter(void 0, void 0, void 0, function* () {
+            const index = new LocalIndex_1.LocalIndex(testIndexDir);
+            yield index.createIndex();
+            yield index.beginUpdate();
+            yield index.insertItem(basicIndexItems[0]);
+            yield index.insertItem(basicIndexItems[1]);
+            yield index.insertItem(basicIndexItems[2]);
+            yield index.endUpdate();
+            yield node_assert_1.default.doesNotReject(() => __awaiter(void 0, void 0, void 0, function* () {
+                yield index.deleteItem('dne');
+            }));
+            node_assert_1.default.equal((yield index.listItems()).length, 3);
+        }));
+        it('leaves existing empty index when last el deleted', () => __awaiter(void 0, void 0, void 0, function* () {
+            var _a;
+            const index = new LocalIndex_1.LocalIndex(testIndexDir);
+            yield index.createIndex();
+            yield index.insertItem(basicIndexItems[0]);
+            yield index.deleteItem((_a = basicIndexItems[0].id) !== null && _a !== void 0 ? _a : '');
+            node_assert_1.default.equal(yield index.isIndexCreated(), true);
+            node_assert_1.default.equal((yield index.listItems()).length, 0);
+        }));
+        it('removes elements from any position', () => __awaiter(void 0, void 0, void 0, function* () {
+            const index = new LocalIndex_1.LocalIndex(testIndexDir);
+            yield index.createIndex();
+            yield index.batchInsertItems([
+                { id: '1', vector: [] },
+                { id: '2', vector: [] },
+                { id: '3', vector: [] },
+                { id: '4', vector: [] },
+                { id: '5', vector: [] }
+            ]);
+            yield index.beginUpdate();
+            yield index.deleteItem('1');
+            yield index.deleteItem('3');
+            yield index.deleteItem('5');
+            yield index.endUpdate();
+            node_assert_1.default.deepStrictEqual(yield index.listItems(), [
+                { id: '2', vector: [], metadata: {}, norm: 0 },
+                { id: '4', vector: [], metadata: {}, norm: 0 }
+            ]);
+        }));
+        it('no-op delete while update active when id missing', () => __awaiter(void 0, void 0, void 0, function* () {
+            const index = new LocalIndex_1.LocalIndex(testIndexDir);
+            yield index.createIndex();
+            yield index.batchInsertItems([
+                { id: '1', vector: [1] },
+                { id: '2', vector: [2] }
+            ]);
+            yield index.beginUpdate();
+            yield node_assert_1.default.doesNotReject(() => __awaiter(void 0, void 0, void 0, function* () { return index.deleteItem('missing'); }));
+            yield index.endUpdate();
+            const items = yield index.listItems();
+            node_assert_1.default.deepStrictEqual(items.map(i => i.id), ['1', '2']);
+        }));
+    });
+    describe('endUpdate', () => {
+        it('throws an error if no update has begun', () => __awaiter(void 0, void 0, void 0, function* () {
+            const index = new LocalIndex_1.LocalIndex(testIndexDir);
+            yield node_assert_1.default.rejects(() => __awaiter(void 0, void 0, void 0, function* () {
+                yield index.endUpdate();
+            }), new Error('No update in progress'));
+        }));
+        it('throws an error if the index could not be saved', () => __awaiter(void 0, void 0, void 0, function* () {
+            const index = new LocalIndex_1.LocalIndex(testIndexDir);
+            yield index.createIndex();
+            yield index.beginUpdate();
+            sinon_1.default.stub(promises_1.default, 'writeFile').rejects(new Error('fs error'));
+            yield node_assert_1.default.rejects(() => __awaiter(void 0, void 0, void 0, function* () {
+                yield index.endUpdate();
+            }), new Error('Error saving index: Error: fs error'));
+        }));
+    });
+    describe('getIndexStats', () => {
+        it('reports empty index correctly', () => __awaiter(void 0, void 0, void 0, function* () {
+            const index = new LocalIndex_1.LocalIndex(testIndexDir);
+            yield index.createIndex();
+            node_assert_1.default.deepStrictEqual(yield index.getIndexStats(), {
+                version: 1,
+                metadata_config: {},
+                items: 0
+            });
+        }));
+        it('correctly reports non-empty index stats', () => __awaiter(void 0, void 0, void 0, function* () {
+            const index = new LocalIndex_1.LocalIndex(testIndexDir);
+            yield index.createIndex({ version: 1, metadata_config: { indexed: [] } });
+            yield index.batchInsertItems(basicIndexItems);
+            node_assert_1.default.deepStrictEqual(yield index.getIndexStats(), {
+                version: 1,
+                metadata_config: { indexed: [] },
+                items: 3
+            });
+        }));
+    });
+    describe('getItem', () => {
+        it('returns undefined when item not found', () => __awaiter(void 0, void 0, void 0, function* () {
+            const index = new LocalIndex_1.LocalIndex(testIndexDir);
+            yield index.createIndex();
+            node_assert_1.default.equal(yield index.getItem('1'), undefined);
+        }));
+        it('returns requested item', () => __awaiter(void 0, void 0, void 0, function* () {
+            const index = new LocalIndex_1.LocalIndex(testIndexDir);
+            yield index.createIndex();
+            yield index.batchInsertItems(basicIndexItems);
+            const item2 = yield index.getItem('2');
+            node_assert_1.default.equal(item2 === null || item2 === void 0 ? void 0 : item2.id, basicIndexItems[1].id);
+            node_assert_1.default.deepStrictEqual(item2 === null || item2 === void 0 ? void 0 : item2.vector, basicIndexItems[1].vector);
+            node_assert_1.default.equal((yield index.listItems()).length, 3);
+        }));
+    });
     describe('batchInsertItems', () => {
-        const indexItems = [
-            { id: '1', vector: [1, 2, 3] },
-            { id: '2', vector: [2, 3, 4] },
-            { id: '3', vector: [3, 4, 5] }
-        ];
         it('should insert provided items', () => __awaiter(void 0, void 0, void 0, function* () {
             const index = new LocalIndex_1.LocalIndex(testIndexDir);
             yield index.createIndex();
-            const newItems = yield index.batchInsertItems(indexItems);
+            const newItems = yield index.batchInsertItems(basicIndexItems);
             node_assert_1.default.equal(newItems.length, 3);
             const retrievedItems = yield index.listItems();
             node_assert_1.default.equal(retrievedItems.length, 3);
@@ -48,17 +284,383 @@ describe('LocalIndex', () => {
             const index = new LocalIndex_1.LocalIndex(testIndexDir);
             yield index.createIndex();
             yield index.insertItem({ id: '2', vector: [9, 9, 9] });
-            // ensures insert error is bubbled up to batchIndexItems caller
             yield node_assert_1.default.rejects(() => __awaiter(void 0, void 0, void 0, function* () {
-                yield index.batchInsertItems(indexItems);
+                yield index.batchInsertItems(basicIndexItems);
             }), {
                 name: 'Error',
                 message: 'Item with id 2 already exists'
             });
-            // ensures no partial update is applied
             const storedItems = yield index.listItems();
             node_assert_1.default.equal(storedItems.length, 1);
         }));
     });
+    describe('listItemsByMetadata', () => {
+        it('returns items matching metadata filter', () => __awaiter(void 0, void 0, void 0, function* () {
+            const index = new LocalIndex_1.LocalIndex(testIndexDir);
+            yield index.createIndex();
+            yield index.batchInsertItems([
+                { id: '1', vector: [], metadata: { category: 'food' } },
+                { id: '2', vector: [], metadata: { category: 'food' } },
+                { id: '3', vector: [], metadata: { category: 'electronics' } },
+                { id: '4', vector: [], metadata: { category: 'drink' } },
+                { id: '5', vector: [], metadata: { category: 'food' } }
+            ]);
+            const foodItems = yield index.listItemsByMetadata({ category: { $eq: 'food' } });
+            node_assert_1.default.deepStrictEqual(foodItems.map(item => item.id), ['1', '2', '5']);
+            const drinkItems = yield index.listItemsByMetadata({ category: { $eq: 'drink' } });
+            node_assert_1.default.deepStrictEqual(drinkItems.map(item => item.id), ['4']);
+            const clothingItems = yield index.listItemsByMetadata({ category: { $eq: 'clothes' } });
+            node_assert_1.default.deepStrictEqual(clothingItems, []);
+        }));
+        it('returns nothing when no items in index', () => __awaiter(void 0, void 0, void 0, function* () {
+            const index = new LocalIndex_1.LocalIndex(testIndexDir);
+            yield index.createIndex();
+            const items = yield index.listItemsByMetadata({});
+            node_assert_1.default.deepStrictEqual(items, []);
+        }));
+        it('empty filter {} with items returns all', () => __awaiter(void 0, void 0, void 0, function* () {
+            const index = new LocalIndex_1.LocalIndex(testIndexDir);
+            yield index.createIndex();
+            yield index.batchInsertItems([
+                { id: 'a', vector: [1] },
+                { id: 'b', vector: [2] }
+            ]);
+            const items = yield index.listItemsByMetadata({});
+            node_assert_1.default.deepStrictEqual(items.map(i => i.id), ['a', 'b']);
+        }));
+    });
+    describe('metadata indexing behavior', () => {
+        it('stores only indexed metadata in index and writes full metadata to external file', () => __awaiter(void 0, void 0, void 0, function* () {
+            const index = new LocalIndex_1.LocalIndex(testIndexDir);
+            yield index.createIndex({ version: 1, metadata_config: { indexed: ['keep'] } });
+            const realWriteFile = promises_1.default.writeFile.bind(promises_1.default);
+            const writes = [];
+            sinon_1.default.stub(promises_1.default, 'writeFile').callsFake((file, data, options) => __awaiter(void 0, void 0, void 0, function* () {
+                writes.push({ file, data });
+                return realWriteFile(file, data, options);
+            }));
+            yield index.insertItem({ id: 'm1', vector: [1], metadata: { keep: 'x', drop: 'y' } });
+            const items = yield index.listItems();
+            const stored = items.find(i => i.id === 'm1');
+            node_assert_1.default.deepStrictEqual(stored.metadata, { keep: 'x' });
+            node_assert_1.default.ok(stored.metadataFile);
+            const textPayloads = writes
+                .map(w => { var _a, _b; return (typeof w.data === 'string' ? w.data : (_b = (_a = w.data) === null || _a === void 0 ? void 0 : _a.toString) === null || _b === void 0 ? void 0 : _b.call(_a)); })
+                .filter(Boolean);
+            const hasFullMetadata = textPayloads.some(t => {
+                try {
+                    const obj = JSON.parse(t);
+                    return obj && obj.keep === 'x' && obj.drop === 'y';
+                }
+                catch (_a) {
+                    return false;
+                }
+            });
+            node_assert_1.default.equal(hasFullMetadata, true);
+        }));
+        it('does not write external file when metadata only contains indexed keys', () => __awaiter(void 0, void 0, void 0, function* () {
+            const index = new LocalIndex_1.LocalIndex(testIndexDir);
+            yield index.createIndex({ version: 1, metadata_config: { indexed: ['only'] } });
+            yield index.insertItem({ id: 'm2', vector: [1], metadata: { only: 'ok' } });
+            const item = yield index.getItem('m2');
+            node_assert_1.default.deepStrictEqual(item === null || item === void 0 ? void 0 : item.metadata, { only: 'ok' });
+            node_assert_1.default.equal(item.metadataFile, undefined);
+        }));
+    });
+    describe('queryItems', () => {
+        it('upsertItem uses the in-progress update path when _update is present', () => __awaiter(void 0, void 0, void 0, function* () {
+            const index = new LocalIndex_1.LocalIndex(testIndexDir);
+            yield index.createIndex();
+            yield index.beginUpdate();
+            // goes through the if (this._update) early return branch in upsertItem
+            const up = yield index.upsertItem({ id: 'inplace', vector: [7, 7], metadata: { k: 'v' } });
+            // not persisted until endUpdate
+            let items = yield index.listItems();
+            node_assert_1.default.equal(items.length, 0);
+            yield index.endUpdate();
+            items = yield index.listItems();
+            node_assert_1.default.equal(items.length, 1);
+            node_assert_1.default.equal(items[0].id, 'inplace');
+            node_assert_1.default.deepStrictEqual(items[0].metadata, { k: 'v' });
+        }));
+        it('with no indexed config, full metadata is stored inline and no metadataFile is written', () => __awaiter(void 0, void 0, void 0, function* () {
+            const index = new LocalIndex_1.LocalIndex(testIndexDir);
+            yield index.createIndex({ version: 1 /* metadata_config undefined => inline */ });
+            yield index.insertItem({ id: 'fullMd', vector: [1], metadata: { a: 1, b: 2 } });
+            const item = yield index.getItem('fullMd');
+            node_assert_1.default.deepStrictEqual(item === null || item === void 0 ? void 0 : item.metadata, { a: 1, b: 2 });
+            node_assert_1.default.equal(item.metadataFile, undefined);
+        }));
+        it('BM25 uses the default docReader (LocalDocument.loadText) when not injected', () => __awaiter(void 0, void 0, void 0, function* () {
+            // stub LocalDocument.loadText so default docReader path runs without touching disk
+            const stub = sinon_1.default.stub(LocalDocument_1.LocalDocument.prototype, 'loadText').resolves('Hello default docReader text');
+            const addDocSpy = sinon_1.default.spy();
+            const fakeEngine = {
+                defineConfig: sinon_1.default.stub(),
+                definePrepTasks: sinon_1.default.stub(),
+                addDoc: (...a) => addDocSpy(...a),
+                consolidate: sinon_1.default.stub(),
+                search: (_q) => []
+            };
+            const index = new LocalIndex_1.LocalIndex(testIndexDir, undefined, undefined, undefined, { bm25Factory: () => fakeEngine /* no docReader injected */ });
+            yield index.createIndex();
+            yield index.batchInsertItems([
+                { id: 'a', vector: [1, 0, 0], metadata: { documentId: 'docA', startPos: 0, endPos: 4 } },
+                { id: 'b', vector: [0, 1, 0], metadata: { documentId: 'docB', startPos: 0, endPos: 4 } }
+            ]);
+            yield index.queryItems([1, 0, 0], 'kw', 1, undefined, true);
+            node_assert_1.default.ok(addDocSpy.called); // default docReader loaded and chunks added
+            stub.restore();
+        }));
+        it('executes the prepTask defined in setupBm25 to cover tokenization/negation pipeline', () => __awaiter(void 0, void 0, void 0, function* () {
+            let capturedTasks = [];
+            const fakeEngine = {
+                defineConfig: sinon_1.default.stub(),
+                definePrepTasks: (arr) => { capturedTasks = arr; },
+                addDoc: sinon_1.default.stub(),
+                consolidate: sinon_1.default.stub(),
+                search: (_q) => []
+            };
+            const index = new LocalIndex_1.LocalIndex(testIndexDir, undefined, undefined, undefined, { bm25Factory: () => fakeEngine, docReader: () => __awaiter(void 0, void 0, void 0, function* () { return 'text'; }) });
+            yield index.createIndex();
+            // trigger setupBm25 via a BM25 query
+            yield index.batchInsertItems([{ id: 'x', vector: [1], metadata: { documentId: 'd', startPos: 0, endPos: 0 } }]);
+            yield index.queryItems([1], 'q', 1, undefined, true);
+            node_assert_1.default.ok(Array.isArray(capturedTasks) && capturedTasks.length === 1);
+            const tokens = capturedTasks[0]("I don't like complicated tokens!");
+            // ensure pipeline ran: non-empty tokens, punctuation removed, possible negation handling
+            node_assert_1.default.ok(Array.isArray(tokens) && tokens.length > 0);
+        }));
+        it('upsertItem uses the in-progress update path when _update is present', () => __awaiter(void 0, void 0, void 0, function* () {
+            const index = new LocalIndex_1.LocalIndex(testIndexDir);
+            yield index.createIndex();
+            yield index.beginUpdate();
+            yield index.upsertItem({ id: 'inplace', vector: [7, 7], metadata: { k: 'v' } });
+            // Not persisted yet because endUpdate not called
+            let items = yield index.listItems();
+            node_assert_1.default.equal(items.length, 0);
+            yield index.endUpdate();
+            items = yield index.listItems();
+            node_assert_1.default.equal(items.length, 1);
+            node_assert_1.default.equal(items[0].id, 'inplace');
+            node_assert_1.default.deepStrictEqual(items[0].metadata, { k: 'v' });
+        }));
+        it('with no indexed config, full metadata is stored inline and no metadataFile is written', () => __awaiter(void 0, void 0, void 0, function* () {
+            const index = new LocalIndex_1.LocalIndex(testIndexDir);
+            yield index.createIndex({ version: 1 }); // metadata_config undefined => inline
+            yield index.insertItem({ id: 'fullMd', vector: [1], metadata: { a: 1, b: 2 } });
+            const item = yield index.getItem('fullMd');
+            node_assert_1.default.deepStrictEqual(item === null || item === void 0 ? void 0 : item.metadata, { a: 1, b: 2 });
+            node_assert_1.default.equal(item.metadataFile, undefined);
+        }));
+        it('BM25 uses the default docReader (LocalDocument.loadText) when not injected', () => __awaiter(void 0, void 0, void 0, function* () {
+            const stub = sinon_1.default.stub(LocalDocument_1.LocalDocument.prototype, 'loadText').resolves('Hello default docReader text');
+            const addDocSpy = sinon_1.default.spy();
+            const fakeEngine = {
+                defineConfig: sinon_1.default.stub(),
+                definePrepTasks: sinon_1.default.stub(),
+                addDoc: (...a) => addDocSpy(...a),
+                consolidate: sinon_1.default.stub(),
+                search: (_q) => []
+            };
+            const index = new LocalIndex_1.LocalIndex(testIndexDir, undefined, undefined, undefined, { bm25Factory: () => fakeEngine });
+            yield index.createIndex();
+            yield index.batchInsertItems([
+                { id: 'a', vector: [1, 0, 0], metadata: { documentId: 'docA', startPos: 0, endPos: 4 } },
+                { id: 'b', vector: [0, 1, 0], metadata: { documentId: 'docB', startPos: 0, endPos: 4 } }
+            ]);
+            yield index.queryItems([1, 0, 0], 'kw', 1, undefined, true);
+            node_assert_1.default.ok(addDocSpy.called);
+            stub.restore();
+        }));
+        it('executes the prepTask defined in setupBm25 to cover tokenization/negation pipeline', () => __awaiter(void 0, void 0, void 0, function* () {
+            let capturedTasks = [];
+            const fakeEngine = {
+                defineConfig: sinon_1.default.stub(),
+                definePrepTasks: (arr) => { capturedTasks = arr; },
+                addDoc: sinon_1.default.stub(),
+                consolidate: sinon_1.default.stub(),
+                search: (_q) => []
+            };
+            const index = new LocalIndex_1.LocalIndex(testIndexDir, undefined, undefined, undefined, { bm25Factory: () => fakeEngine, docReader: () => __awaiter(void 0, void 0, void 0, function* () { return 'text'; }) });
+            yield index.createIndex();
+            yield index.batchInsertItems([{ id: 'x', vector: [1], metadata: { documentId: 'd', startPos: 0, endPos: 0 } }]);
+            yield index.queryItems([1], 'q', 1, undefined, true);
+            node_assert_1.default.ok(Array.isArray(capturedTasks) && capturedTasks.length === 1);
+            const tokens = capturedTasks[0]("I don't like complicated tokens!");
+            node_assert_1.default.ok(Array.isArray(tokens) && tokens.length > 0);
+        }));
+        it('returns empty array on empty index search', () => __awaiter(void 0, void 0, void 0, function* () {
+            const index = new LocalIndex_1.LocalIndex(testIndexDir);
+            yield index.createIndex();
+            const result = yield index.queryItems([1, 2, 3], '', 10);
+            node_assert_1.default.deepStrictEqual(result, []);
+        }));
+        it('returns bad match when no better match exists', () => __awaiter(void 0, void 0, void 0, function* () {
+            var _a, _b;
+            const index = new LocalIndex_1.LocalIndex(testIndexDir);
+            yield index.createIndex();
+            yield index.insertItem({ id: '1', vector: [0.9, 0, 0, 0, 0] });
+            const result = yield index.queryItems([0, 0, 0, 0, 0.1], '', 1);
+            node_assert_1.default.equal((_a = result[0]) === null || _a === void 0 ? void 0 : _a.score, 0);
+            node_assert_1.default.equal((_b = result[0]) === null || _b === void 0 ? void 0 : _b.item.id, '1');
+        }));
+        it('returns all vectors when fewer than topK exist', () => __awaiter(void 0, void 0, void 0, function* () {
+            const index = new LocalIndex_1.LocalIndex(testIndexDir);
+            yield index.createIndex();
+            yield index.batchInsertItems(basicIndexItems);
+            const result = yield index.queryItems([0, 0, 1], '', 10);
+            node_assert_1.default.equal(result.length, 3);
+            node_assert_1.default.deepStrictEqual(result.map(({ item }) => item.id), basicIndexItems.map(item => item.id));
+        }));
+        it('limits results to topK when more items exist', () => __awaiter(void 0, void 0, void 0, function* () {
+            const index = new LocalIndex_1.LocalIndex(testIndexDir);
+            yield index.createIndex();
+            yield index.batchInsertItems([
+                { id: 'a', vector: [1, 0, 0] },
+                { id: 'b', vector: [0, 1, 0] },
+                { id: 'c', vector: [0, 0, 1] },
+                { id: 'd', vector: [1, 1, 0] },
+                { id: 'e', vector: [0, 1, 1] }
+            ]);
+            const result = yield index.queryItems([1, 0, 0], '', 2);
+            node_assert_1.default.equal(result.length, 2);
+        }));
+        it('filters by metadata when filter provided', () => __awaiter(void 0, void 0, void 0, function* () {
+            const index = new LocalIndex_1.LocalIndex(testIndexDir);
+            yield index.createIndex();
+            yield index.batchInsertItems([
+                { id: '1', vector: [1, 0, 0], metadata: { category: 'food' } },
+                { id: '2', vector: [0, 0, 1], metadata: { category: 'drink' } }
+            ]);
+            const bestGeneralMatch = yield index.queryItems([1, 0, 0], '', 1);
+            const bestDrinkMatch = yield index.queryItems([1, 0, 0], '', 1, {
+                category: { $eq: 'drink' }
+            });
+            node_assert_1.default.equal(bestGeneralMatch[0].item.id, '1');
+            node_assert_1.default.equal(bestDrinkMatch[0].item.id, '2');
+        }));
+        it('reads item metadata file when provided', () => __awaiter(void 0, void 0, void 0, function* () {
+            const index = new LocalIndex_1.LocalIndex(testIndexDir);
+            yield index.createIndex({ version: 1, metadata_config: { indexed: ['category'] } });
+            yield index.batchInsertItems([
+                { id: '1', vector: [1, 0, 0] },
+                { id: '2', vector: [0, 0, 1], metadata: { category: 'drink', extra: 'x' } }
+            ]);
+            sinon_1.default.stub(promises_1.default, 'readFile').resolves(JSON.stringify({ category: 'drink', extra: 'x' }));
+            const bestDrinkMatch = yield index.queryItems([1, 0, 0], '', 2, { category: { $eq: 'drink' } });
+            node_assert_1.default.notEqual(bestDrinkMatch[0].item.metadataFile, undefined);
+            node_assert_1.default.equal(bestDrinkMatch[0].item.id, '2');
+        }));
+        it('appends BM25 results when isBm25=true (tuple results)', () => __awaiter(void 0, void 0, void 0, function* () {
+            // Inject a fake BM25 engine and docReader
+            const addDocSpy = sinon_1.default.spy();
+            const consolidateSpy = sinon_1.default.spy();
+            const fakeEngine = {
+                defineConfig: sinon_1.default.stub(),
+                definePrepTasks: sinon_1.default.stub(),
+                addDoc: (...a) => addDocSpy(...a),
+                consolidate: (...a) => consolidateSpy(...a),
+                search: (q) => {
+                    void q;
+                    // results as tuples [index, score]
+                    return [[0, 0.123], [1, 0.122]];
+                }
+            };
+            const index = new LocalIndex_1.LocalIndex(testIndexDir, undefined, undefined, undefined, {
+                bm25Factory: () => fakeEngine,
+                docReader: (_docId) => __awaiter(void 0, void 0, void 0, function* () { return 'SAMPLETEXT'; })
+            });
+            yield index.createIndex();
+            yield index.batchInsertItems([
+                { id: 's1', vector: [1, 0, 0], metadata: { category: 'alpha', documentId: 'd1', startPos: 0, endPos: 3 } },
+                { id: 's2', vector: [0, 1, 0], metadata: { category: 'beta', documentId: 'd2', startPos: 0, endPos: 3 } },
+                { id: 's3', vector: [0, 0, 1], metadata: { category: 'gamma', documentId: 'd3', startPos: 0, endPos: 3 } }
+            ]);
+            const topK = 1;
+            const results = yield index.queryItems([1, 0, 0], 'keyword', topK, undefined, true);
+            node_assert_1.default.ok(results.length >= 1);
+            node_assert_1.default.ok(results.some(r => { var _a; return ((_a = r.item.metadata) === null || _a === void 0 ? void 0 : _a.isBm25) === true; }));
+            node_assert_1.default.ok(consolidateSpy.calledOnce);
+            // only non-top items should have been offered to addDoc (top item excluded)
+            node_assert_1.default.ok(addDocSpy.callCount >= 1);
+        }));
+        it('BM25 indexes only non-top items with document metadata, calls consolidate, and swallows load errors', () => __awaiter(void 0, void 0, void 0, function* () {
+            // docReader returns text for docB, throws for docC
+            const addDocSpy = sinon_1.default.spy();
+            const consolidateSpy = sinon_1.default.spy();
+            const fakeEngine = {
+                defineConfig: sinon_1.default.stub(),
+                definePrepTasks: sinon_1.default.stub(),
+                addDoc: (...a) => addDocSpy(...a),
+                consolidate: (...a) => consolidateSpy(...a),
+                search: (_q) => {
+                    // invalid tuple index is ignored
+                    return [[999, 0.77]];
+                }
+            };
+            const docReader = (docId) => __awaiter(void 0, void 0, void 0, function* () {
+                if (docId === 'docB')
+                    return 'ABCDEFGHIJ';
+                if (docId === 'docC')
+                    throw new Error('load failed');
+                return '';
+            });
+            const index = new LocalIndex_1.LocalIndex(testIndexDir, undefined, undefined, undefined, { bm25Factory: () => fakeEngine, docReader });
+            yield index.createIndex();
+            yield index.batchInsertItems([
+                { id: 'sTop', vector: [1, 0, 0], metadata: { documentId: 'docA', startPos: 0, endPos: 4 } },
+                { id: 'd1', vector: [0, 1, 0], metadata: { documentId: 'docB', startPos: 0, endPos: 4 } },
+                { id: 'd2', vector: [0, 0, 1], metadata: { documentId: 'docC', startPos: 0, endPos: 4 } }
+            ]);
+            const res = yield index.queryItems([1, 0, 0], 'kw', 1, undefined, true);
+            node_assert_1.default.equal(addDocSpy.callCount, 1); // only d1 added; d2 threw; sTop excluded
+            node_assert_1.default.equal(consolidateSpy.callCount, 1);
+            node_assert_1.default.ok(res.length >= 1);
+        }));
+        it('BM25 object results are appended and marked isBm25 even if item is undefined', () => __awaiter(void 0, void 0, void 0, function* () {
+            const fakeEngine = {
+                defineConfig: sinon_1.default.stub(),
+                definePrepTasks: sinon_1.default.stub(),
+                addDoc: sinon_1.default.stub(),
+                consolidate: sinon_1.default.stub(),
+                search: (_q) => [{ item: undefined, score: 0.55 }]
+            };
+            const index = new LocalIndex_1.LocalIndex(testIndexDir, undefined, undefined, undefined, { bm25Factory: () => fakeEngine, docReader: () => __awaiter(void 0, void 0, void 0, function* () { return 'YY'; }) });
+            yield index.createIndex();
+            yield index.batchInsertItems([
+                { id: 'x1', vector: [1, 0], metadata: { documentId: 'doc1', startPos: 0, endPos: 1 } },
+                { id: 'x2', vector: [0, 1], metadata: { documentId: 'doc2', startPos: 0, endPos: 1 } }
+            ]);
+            const res = yield index.queryItems([1, 0], 'any', 1, undefined, true);
+            node_assert_1.default.ok(res.some(r => { var _a; return ((_a = r.item.metadata) === null || _a === void 0 ? void 0 : _a.isBm25) === true; }));
+        }));
+        it('BM25 engine missing addDoc/consolidate does not throw and returns semantic results', () => __awaiter(void 0, void 0, void 0, function* () {
+            const fakeEngine = {
+                defineConfig: sinon_1.default.stub(),
+                definePrepTasks: sinon_1.default.stub(),
+                search: (_q) => []
+            };
+            const index = new LocalIndex_1.LocalIndex(testIndexDir, undefined, undefined, undefined, { bm25Factory: () => fakeEngine, docReader: () => __awaiter(void 0, void 0, void 0, function* () { return 'YY'; }) });
+            yield index.createIndex();
+            yield index.batchInsertItems([
+                { id: 's1', vector: [1, 0], metadata: { documentId: 'doc1', startPos: 0, endPos: 1 } },
+                { id: 's2', vector: [0, 1], metadata: { documentId: 'doc2', startPos: 0, endPos: 1 } }
+            ]);
+            const res = yield index.queryItems([1, 0], 'kw', 1, undefined, true);
+            node_assert_1.default.ok(res.length >= 1);
+            node_assert_1.default.ok(res.every(r => { var _a; return !((_a = r.item.metadata) === null || _a === void 0 ? void 0 : _a.isBm25); }));
+        }));
+    });
+    describe('VirtualFileStorage (in-memory)', () => {
+        it('works end-to-end in-memory without touching disk', () => __awaiter(void 0, void 0, void 0, function* () {
+            const storage = new storage_1.VirtualFileStorage();
+            const index = new LocalIndex_1.LocalIndex('mem://idx', undefined, storage);
+            yield index.createIndex({ version: 3, metadata_config: { indexed: ['t'] } });
+            yield index.insertItem({ vector: [1], metadata: { t: 'x', other: 'y' } });
+            const stats = yield index.getIndexStats();
+            node_assert_1.default.deepStrictEqual(stats, { version: 3, metadata_config: { indexed: ['t'] }, items: 1 });
+        }));
+    });
 });
 //# sourceMappingURL=LocalIndex.spec.js.map