From ad7b90bebdad6ad6f0392b86f0390b98ebe9808c Mon Sep 17 00:00:00 2001 From: "shumeiko.it" Date: Sat, 24 Oct 2026 17:07:10 +0200 Subject: [PATCH] test(server): drop tests of the legacy memory engine (P1.5) - harvest-provenance: tested the removed harvest pipeline (RawArchive and frame metadata), which no longer exists. - benchmarks: drop the FTS5 frame search and frame/session write benchmarks; they measured the legacy engine only. --- .../tests/local/harvest-provenance.test.ts | 130 ------------ .../tests/performance/benchmarks.test.ts | 197 +----------------- 2 files changed, 9 insertions(+), 318 deletions(-) delete mode 100644 packages/server/tests/local/harvest-provenance.test.ts diff --git a/packages/server/tests/local/harvest-provenance.test.ts b/packages/server/tests/local/harvest-provenance.test.ts deleted file mode 100644 index 4180a4b..0000000 --- a/packages/server/tests/local/harvest-provenance.test.ts +++ /dev/null @@ -1,130 +0,0 @@ -import { describe, it, expect, beforeEach, afterEach } from 'vitest'; -import { MindDB, FrameStore, SessionStore, RawArchive, readArchiveUids, withArchiveUid } from '@waggle/core'; - -interface HItem { source: string; id: string; title: string; content: string } - -// Faithfully mirrors the harvest route's per-item persistence: archive the full -// verbatim (best-effort try/catch), create the (truncated) summary frame, then -// stamp metadata.archiveUids — with the dedup/backfill else-if from the route. -function persistHarvestItem(db: MindDB, item: HItem, archive: RawArchive = new RawArchive(db)) { - const frames = new FrameStore(db); - let archiveUid: string | undefined; - try { - archiveUid = archive.append({ - source: item.source, sourceRef: item.id, title: item.title, content: item.content, - }).archiveUid; - } catch { /* best-effort — frame still persists without the link */ } - const frame = frames.createIFrame('harvest', `${item.title}\n\n${item.content.slice(0, 10_000)}`, 'normal', 'import'); - if (!frame.metadata || frame.metadata === '{}') { - frames.setMetadata(frame.id, JSON.stringify({ status: 'unreviewed', sourceId: item.id, ...(archiveUid ? { archiveUids: [archiveUid] } : {}) })); - } else if (archiveUid) { - const meta = JSON.parse(frame.metadata) as Record; - if (!readArchiveUids(meta).includes(archiveUid)) { - frames.setMetadata(frame.id, JSON.stringify(withArchiveUid(meta, archiveUid))); - } - } - return { archive, frame: frames.getById(frame.id)! }; -} - -describe('harvest provenance archive', () => { - let db: MindDB; - beforeEach(() => { - db = new MindDB(':memory:'); - new SessionStore(db).ensure('harvest', 'harvest', 'test'); - }); - afterEach(() => { db.close(); }); - - it('a harvested frame links to its full immutable raw_archive row', () => { - const big = 'A'.repeat(25_000); - const { archive, frame } = persistHarvestItem(db, { source: 'claude', id: 'c1', title: 'T', content: big }); - const rows = archive.reconstructSource(frame.id); - expect(rows).toHaveLength(1); - expect(rows[0].content.length).toBe(25_000); // full source survived (frame is capped at 10K) - expect(frame.content.length).toBeLessThanOrEqual(10_000 + 4); - // canonical link shape is the archiveUids array (no legacy scalar) - const meta = JSON.parse(frame.metadata ?? '{}') as { archiveUids?: string[]; archiveUid?: string }; - expect(meta.archiveUids).toHaveLength(1); - expect(meta.archiveUid).toBeUndefined(); - }); - - it('re-importing the same item does not duplicate the archive row', () => { - const item = { source: 'claude', id: 'c2', title: 'T', content: 'same content' }; - const { archive } = persistHarvestItem(db, item); - persistHarvestItem(db, item); - expect(archive.count()).toBe(1); - }); - - it('a failed archive append still persists the frame, without an archiveUid link', () => { - const archive = new RawArchive(db); - archive.append = () => { throw new Error('boom'); }; - const { frame } = persistHarvestItem(db, { source: 'claude', id: 'c3', title: 'T', content: 'body' }, archive); - expect(frame.id).toBeGreaterThan(0); - const meta = JSON.parse(frame.metadata ?? '{}'); - expect(meta.archiveUids).toBeUndefined(); - expect(meta.archiveUid).toBeUndefined(); - expect(meta.sourceId).toBe('c3'); - }); - - it('backfills archiveUid on a re-import after a prior append failure', () => { - // First import: archive append throws → frame persists without a link. - const failing = new RawArchive(db); - failing.append = () => { throw new Error('boom'); }; - persistHarvestItem(db, { source: 'claude', id: 'c4', title: 'T', content: 'recoverable' }, failing); - - // Second import (working archive): createIFrame dedups → same frame; the - // else-if backfill adds archiveUid without clobbering the existing sourceId. - const working = new RawArchive(db); - const { frame } = persistHarvestItem(db, { source: 'claude', id: 'c4', title: 'T', content: 'recoverable' }, working); - const meta = JSON.parse(frame.metadata ?? '{}'); - expect(meta.archiveUids).toHaveLength(1); - expect(typeof meta.archiveUids[0]).toBe('string'); - expect(meta.sourceId).toBe('c4'); - const rows = working.reconstructSource(frame.id); - expect(rows).toHaveLength(1); - expect(rows[0].content).toBe('recoverable'); - }); - - it('accumulates both uids on one frame when identical content arrives from two sourceRefs', () => { - // Same source + same content (→ content-dedup to ONE frame) but different - // sourceRef → two distinct per-source archive rows. Both uids must end up on - // the single frame's archiveUids array (backfill grows the set). - const archive = new RawArchive(db); - const { frame: f1 } = persistHarvestItem(db, { source: 'claude', id: 'ref-a', title: 'T', content: 'shared body' }, archive); - const { frame: f2 } = persistHarvestItem(db, { source: 'claude', id: 'ref-b', title: 'T', content: 'shared body' }, archive); - - expect(f2.id).toBe(f1.id); // content-dedup → one frame - expect(archive.count()).toBe(2); // two per-source archive rows - - const meta = JSON.parse(f2.metadata ?? '{}') as { archiveUids?: string[]; archiveUid?: string }; - expect(meta.archiveUids).toHaveLength(2); - expect(new Set(meta.archiveUids)).toHaveProperty('size', 2); - expect(meta.archiveUid).toBeUndefined(); // no legacy scalar lingering - - const rows = archive.reconstructSource(f2.id); - expect(rows).toHaveLength(2); - expect(new Set(rows.map(r => r.source_ref))).toEqual(new Set(['ref-a', 'ref-b'])); - }); - - // (e) pins the `.includes(archiveUid)` grow-check skip-path: re-importing an - // IDENTICAL item must not duplicate the archive row NOR grow archiveUids beyond - // length 1. The existing "does not duplicate the archive row" test only asserts - // the row count; this test additionally pins uid-array cardinality and - // reconstructSource resolution. - it('idempotent backfill: re-importing the same item does not grow archiveUids or the archive row count', () => { - const archive = new RawArchive(db); - const item = { source: 'claude', id: 'idem-1', title: 'T', content: 'idem content' }; - - // First import: creates archive row + frame, stamps archiveUids: [uid]. - const { frame: f1 } = persistHarvestItem(db, item, archive); - // Second import of the IDENTICAL item: archive.append is a no-op (INSERT OR IGNORE); - // createIFrame deduplicates; the grow-check skips setMetadata because the uid is - // already in archiveUids. - const { frame: f2 } = persistHarvestItem(db, item, archive); - - expect(f2.id).toBe(f1.id); // same frame (content dedup) - expect(archive.count()).toBe(1); // archive row NOT duplicated - const meta = JSON.parse(f2.metadata ?? '{}') as { archiveUids?: string[] }; - expect(meta.archiveUids).toHaveLength(1); // uid-array NOT grown by the skip-path - expect(archive.reconstructSource(f2.id)).toHaveLength(1); // resolves exactly one row - }); -}); diff --git a/packages/server/tests/performance/benchmarks.test.ts b/packages/server/tests/performance/benchmarks.test.ts index 442ebde..9da7334 100644 --- a/packages/server/tests/performance/benchmarks.test.ts +++ b/packages/server/tests/performance/benchmarks.test.ts @@ -6,19 +6,17 @@ * * Coverage: * 1. Cold start: server boot → first response - * 2. FTS5 search with 1000+ frames - * 3. Workspace list with 50 workspaces - * 4. Session load with 500 messages - * 5. Marketplace FTS5 search (bundled DB) - * 6. Vault set+get cycle - * 7. Memory frame batch write (100 frames) + * 2. Workspace list with 50 workspaces + * 3. Session load with 500 messages + * 4. Marketplace FTS5 search (bundled DB) + * 5. Vault set+get cycle */ import { describe, it, expect, beforeAll, afterAll, beforeEach, afterEach } from 'vitest'; import fs from 'node:fs'; import path from 'node:path'; import os from 'node:os'; -import { MindDB, FrameStore, SessionStore, VaultStore, WorkspaceManager } from '@waggle/core'; +import { VaultStore, WorkspaceManager } from '@waggle/core'; import { buildLocalServer } from '../../src/local/index.js'; import type { FastifyInstance } from 'fastify'; import { injectWithAuth } from '../test-utils.js'; @@ -55,11 +53,6 @@ describe('Performance Benchmarks', () => { it('server boots and responds within 5 seconds', async () => { const tmpDir = createTmpDir('coldstart'); try { - // Create a minimal personal.mind so the server can start - const personalPath = path.join(tmpDir, 'personal.mind'); - const mind = new MindDB(personalPath); - mind.close(); - const start = Date.now(); const server = await buildLocalServer({ dataDir: tmpDir }); const res = await injectWithAuth(server, { method: 'GET', url: '/health' }); @@ -76,97 +69,7 @@ describe('Performance Benchmarks', () => { }, 15_000); }); - // ── 2. FTS5 Search Performance ────────────────────────────────────── - - describe('FTS5 Search with 1000 frames', () => { - let tmpDir: string; - let mind: MindDB; - let frames: FrameStore; - let sessions: SessionStore; - let gopId: string; - - beforeAll(() => { - tmpDir = createTmpDir('fts5'); - const dbPath = path.join(tmpDir, 'personal.mind'); - mind = new MindDB(dbPath); - frames = new FrameStore(mind); - sessions = new SessionStore(mind); - - // Create a session for FK constraints - const session = sessions.create('perf-test'); - gopId = session.gop_id; - - // Insert 1000 frames with varied content - const topics = [ - 'machine learning model training optimization', - 'database indexing strategies for SQLite', - 'TypeScript generics and type inference', - 'React component lifecycle management', - 'Node.js event loop performance tuning', - 'memory management garbage collection patterns', - 'API design RESTful best practices', - 'testing strategies unit integration e2e', - 'CI/CD pipeline automation deployment', - 'microservices architecture communication patterns', - ]; - - for (let i = 0; i < 1000; i++) { - const topic = topics[i % topics.length]; - frames.createIFrame(gopId, `Frame ${i}: ${topic} — detailed notes about ${topic} iteration ${i}`, 'normal'); - } - }); - - afterAll(() => { - mind.close(); - cleanupDir(tmpDir); - }); - - it('FTS5 keyword search completes within 200ms for 1000 frames', () => { - const raw = mind.getDatabase(); - - const start = Date.now(); - const results = raw.prepare(` - SELECT mf.* FROM memory_frames mf - INNER JOIN memory_frames_fts fts ON mf.id = fts.rowid - WHERE memory_frames_fts MATCH ? - LIMIT 20 - `).all('machine learning'); - const elapsed = Date.now() - start; - - expect(results.length).toBeGreaterThan(0); - // 200ms is generous — SQLite FTS5 on 1000 rows should be <10ms - expect(elapsed).toBeLessThan(perfBudget(200)); - }); - - it('FTS5 wildcard search completes within 200ms', () => { - const raw = mind.getDatabase(); - - const start = Date.now(); - const results = raw.prepare(` - SELECT mf.* FROM memory_frames mf - INNER JOIN memory_frames_fts fts ON mf.id = fts.rowid - WHERE memory_frames_fts MATCH ? - LIMIT 20 - `).all('database*'); - const elapsed = Date.now() - start; - - expect(results.length).toBeGreaterThan(0); - expect(elapsed).toBeLessThan(perfBudget(200)); - }); - - it('frame count query completes within 50ms', () => { - const raw = mind.getDatabase(); - - const start = Date.now(); - const row = raw.prepare('SELECT COUNT(*) as cnt FROM memory_frames').get() as { cnt: number }; - const elapsed = Date.now() - start; - - expect(row.cnt).toBe(1000); - expect(elapsed).toBeLessThan(perfBudget(50)); - }); - }); - - // ── 3. Workspace List Performance ─────────────────────────────────── + // ── 2. Workspace List Performance ─────────────────────────────────── describe('Workspace Listing with 50 workspaces', () => { let tmpDir: string; @@ -175,11 +78,6 @@ describe('Performance Benchmarks', () => { beforeAll(async () => { tmpDir = createTmpDir('ws-list'); - // Create personal.mind - const personalPath = path.join(tmpDir, 'personal.mind'); - const mind = new MindDB(personalPath); - mind.close(); - // Create 50 workspaces via WorkspaceManager const wsManager = new WorkspaceManager(tmpDir); for (let i = 0; i < 50; i++) { @@ -211,7 +109,7 @@ describe('Performance Benchmarks', () => { }); }); - // ── 4. Session Load Performance ───────────────────────────────────── + // ── 3. Session Load Performance ───────────────────────────────────── describe('Session File Loading', () => { let tmpDir: string; @@ -258,7 +156,7 @@ describe('Performance Benchmarks', () => { }); }); - // ── 5. Marketplace FTS5 Search ────────────────────────────────────── + // ── 4. Marketplace FTS5 Search ────────────────────────────────────── describe('Marketplace Search', () => { it('marketplace FTS5 search completes within 200ms (bundled DB)', () => { @@ -317,7 +215,7 @@ describe('Performance Benchmarks', () => { }); }); - // ── 6. Vault Operations ───────────────────────────────────────────── + // ── 5. Vault Operations ───────────────────────────────────────────── describe('Vault Encrypt/Decrypt Cycle', () => { let tmpDir: string; @@ -372,81 +270,4 @@ describe('Performance Benchmarks', () => { expect(elapsed).toBeLessThan(perfBudget(50)); }); }); - - // ── 7. Memory Frame Write Performance ─────────────────────────────── - - describe('Memory Frame Batch Write', () => { - let tmpDir: string; - let mind: MindDB; - let frames: FrameStore; - let sessions: SessionStore; - - beforeEach(() => { - tmpDir = createTmpDir('frame-write'); - const dbPath = path.join(tmpDir, 'write-bench.mind'); - mind = new MindDB(dbPath); - frames = new FrameStore(mind); - sessions = new SessionStore(mind); - }); - - afterEach(() => { - mind.close(); - cleanupDir(tmpDir); - }); - - it('writing 100 I-frames completes within 2 seconds', () => { - const session = sessions.create('write-bench'); - const gopId = session.gop_id; - - const start = Date.now(); - for (let i = 0; i < 100; i++) { - frames.createIFrame( - gopId, - `Batch write test frame ${i}: This is a moderately sized content block that simulates real-world memory frame sizes with enough text to be meaningful.`, - 'normal' - ); - } - const elapsed = Date.now() - start; - - // Verify all frames were written - const allFrames = frames.getGopFrames(gopId); - expect(allFrames.length).toBe(100); - // 2 seconds is generous — SQLite WAL mode batch writes should be <500ms - expect(elapsed).toBeLessThan(perfBudget(2000)); - }); - - it('writing 100 P-frames completes within 2 seconds', () => { - const session = sessions.create('pframe-bench'); - const gopId = session.gop_id; - - // Create an I-frame as the base - const baseFrame = frames.createIFrame(gopId, 'Base frame for P-frame benchmark', 'normal'); - - const start = Date.now(); - for (let i = 0; i < 100; i++) { - frames.createPFrame( - gopId, - `P-frame delta ${i}: Updated information about the topic with new details and corrections.`, - baseFrame.id, - 'normal' - ); - } - const elapsed = Date.now() - start; - - // 1 base + 100 P-frames = 101 - const allFrames = frames.getGopFrames(gopId); - expect(allFrames.length).toBe(101); - expect(elapsed).toBeLessThan(perfBudget(2000)); - }); - - it('creating 50 sessions completes within 1 second', () => { - const start = Date.now(); - for (let i = 0; i < 50; i++) { - sessions.create(`session-bench-${i}`); - } - const elapsed = Date.now() - start; - - expect(elapsed).toBeLessThan(perfBudget(1000)); - }); - }); });