mirror of
https://github.com/ruvnet/ruflo.git
synced 2026-09-14 14:01:28 +08:00
dc8463872d
* docs: add ADR-057 — RVF native storage backend to replace sql.js Proposes replacing the 18MB sql.js WASM SQLite dependency with RuVector Format (RVF) across @claude-flow/shared, memory, and embeddings packages. Includes automatic and manual migration paths for existing .db and .json files with full backward compatibility. Key benefits: -18MB install size, 150x-12,500x faster vector search via native HNSW, quantization support, crash-safe append-only format. Co-Authored-By: claude-flow <ruv@ruv.net> * docs: extend ADR-057 with RVF embedding provider, ruvLLM persistence, and progressive download Add three major sections to ADR-057: - Section 3A: RVF as 5th EmbeddingProvider (52KB WASM replaces 540MB agentic-flow for basic use cases). Updated auto-select hierarchy: rvf > agentic-flow > transformers > mock. - Section 3B: ruvLLM learning persistence via RVF segments. SONA patterns (ReasoningBank), LoRA adapters, and EWC++ weights now survive process restarts. RuVectorProvider.searchMemory works offline via RVF without HTTP server. - Section 10: Progressive download architecture. Replaces npm optionalDependencies (all-or-nothing 1.3GB) with on-demand capability downloads stored as .rvf files. 5 phases from 22MB core CLI to full LLM/CNN/GNN/kernel capabilities. Implementation plan expanded from 5 to 8 phases. Testing strategy updated with 11 new test cases covering the additions. Co-Authored-By: claude-flow <ruv@ruv.net> * feat: add optimized Docker images for Ruflo lite/full/workers - Dockerfile: Multi-stage Alpine build with aggressive dependency pruning. Result: 324MB image, 0.6s startup (vs 1.3GB/35s baseline). Uses --omit=optional + explicit rm -rf of heavy transitive deps. - Dockerfile.full: Full image variant with all optional dependencies including native compilation (onnxruntime-node, better-sqlite3). - Dockerfile.headless: Updated from node:20 to node:22, claude-flow branding to ruflo, same prune strategy as lite. - docker-compose.yml: Three profiles (lite, full, workers) with appropriate resource limits and health checks. - .dockerignore: Excludes node_modules, tests, source maps, IDE configs, and secrets from build context. Co-Authored-By: claude-flow <ruv@ruv.net> * feat(ADR-057): implement RVF native storage backend across all packages Replace sql.js (18MB WASM) with zero-dependency RVF format across memory, events, embeddings, and learning subsystems. Pure TypeScript fallback with optional native @ruvector/rvf acceleration. New files: - memory/rvf-backend.ts: IMemoryBackend impl with HNSW + binary persistence - memory/hnsw-lite.ts: Lightweight HNSW graph for vector search - memory/rvf-learning-store.ts: SONA pattern/LoRA/EWC/trajectory persistence - memory/persistent-sona.ts: PersistentSonaCoordinator wrapping learning store - shared/events/rvf-event-log.ts: Append-only event log (replaces EventStore) - embeddings/rvf-embedding-cache.ts: Binary file embedding cache (FNV-1a keys) - embeddings/rvf-embedding-service.ts: Hash-based embedding provider (<0.1ms) Integrations: - DatabaseProvider: 'rvf' added, auto-selected first (always available) - EmbeddingProvider: 'rvf' added, auto-selected first in async factory - All packages export new RVF types and classes Tests: 163 passing (5 suites), validated in Docker (node:22-alpine) - rvf-backend: 50 tests (CRUD, query, search, persistence, HNSW) - rvf-event-log: 33 tests (append, replay, snapshots, crash recovery) - rvf-embeddings: 30 tests (cache, service, determinism, normalization) - rvf-learning-store: 31 tests (patterns, LoRA, EWC, trajectories) - rvf-integration: 19 tests (provider selection, migration, concurrency) Co-Authored-By: claude-flow <ruv@ruv.net> * fix(ADR-057): harden RVF storage with security, performance, and migration Security: atomic writes (tmp+rename) across all backends, path validation, input validation at boundaries. Performance: persist lock prevents concurrent flushes, stack overflow fix for large datasets, timer.unref() prevents hangs, O(1) eviction lookup in embedding cache. New: bidirectional RVF migration utility (JSON/SQLite/RVF), format v2 with backward compat, auto-persist patterns and trajectories before eviction. Added HnswLite and RvfMigrator exports. 193 tests passing, 0 TS errors. Refs #1242 Co-Authored-By: claude-flow <ruv@ruv.net> * fix: doctor version detection and init settings warnings (#1235, #1230) doctor: Replace unreliable path-walking heuristic with import.meta.url traversal to find the correct package.json. Removes stale npx cache scanning that could report wrong versions. Fixes #1235. init: Remove invalid TeammateIdle and TaskCompleted keys from generated .claude/settings.json hooks section. These are not recognized by Claude Code and caused warnings. Config remains available under the claudeFlow namespace. Fixes #1230. Co-Authored-By: claude-flow <ruv@ruv.net> * fix(doctor): remove Windows-incompatible shell redirects, simplify root detection - Remove fragile Windows root calculation (split on backslash); use dirname(dir) === dir instead which works on all platforms - Remove 2>/dev/null from npm view command (fails on Windows cmd.exe); the catch block handles errors gracefully already Co-Authored-By: claude-flow <ruv@ruv.net> * security: harden RVF against crafted files and race conditions - rvf-backend: validate dimensions (1-10000), validate RVF header structure and size before JSON.parse, guard persist interval against concurrent flushes - rvf-embedding-cache: use random suffix for tmp files to prevent predictable-path attacks on multi-user systems - rvf-event-log: add 100MB max payload size limit to prevent DoS from crafted event log files - rvf-migration: add null byte and type validation on all paths Co-Authored-By: claude-flow <ruv@ruv.net>
444 lines
15 KiB
TypeScript
444 lines
15 KiB
TypeScript
/**
|
|
* Tests for RvfLearningStore
|
|
*
|
|
* Covers initialization, pattern/LoRA/EWC/trajectory CRUD, persistence
|
|
* with RVLS magic header, stats, error handling, and edge cases.
|
|
*
|
|
* Run: npx tsx --test tests/rvf-learning-store.test.ts
|
|
*/
|
|
|
|
import { describe, it, before, after, beforeEach, afterEach } from 'node:test';
|
|
import assert from 'node:assert/strict';
|
|
import fs from 'node:fs';
|
|
import path from 'node:path';
|
|
import os from 'node:os';
|
|
import {
|
|
RvfLearningStore,
|
|
type PatternRecord,
|
|
type LoraRecord,
|
|
type EwcRecord,
|
|
type TrajectoryRecord,
|
|
} from '../v3/@claude-flow/memory/src/rvf-learning-store.js';
|
|
|
|
// ===== Fixtures =====
|
|
|
|
function tmpDir(): string {
|
|
return fs.mkdtempSync(path.join(os.tmpdir(), 'rvf-test-'));
|
|
}
|
|
|
|
function makePattern(overrides: Partial<PatternRecord> = {}): PatternRecord {
|
|
return {
|
|
id: 'p1',
|
|
type: 'query_response',
|
|
embedding: [0.1, 0.2, 0.3],
|
|
successRate: 0.95,
|
|
useCount: 10,
|
|
lastUsed: new Date().toISOString(),
|
|
...overrides,
|
|
};
|
|
}
|
|
|
|
function makeLora(overrides: Partial<LoraRecord> = {}): LoraRecord {
|
|
return {
|
|
id: 'lora-1',
|
|
config: { rank: 8 },
|
|
weights: 'base64weights==',
|
|
frozen: false,
|
|
numParameters: 1024,
|
|
...overrides,
|
|
};
|
|
}
|
|
|
|
function makeEwc(overrides: Partial<EwcRecord> = {}): EwcRecord {
|
|
return {
|
|
tasksLearned: 5,
|
|
protectionStrength: 0.8,
|
|
forgettingRate: 0.02,
|
|
taskWeights: { task1: [0.1, 0.2], task2: [0.3, 0.4] },
|
|
...overrides,
|
|
};
|
|
}
|
|
|
|
function makeTrajectory(overrides: Partial<TrajectoryRecord> = {}): TrajectoryRecord {
|
|
return {
|
|
id: 'traj-1',
|
|
steps: [
|
|
{ type: 'query', input: 'hello', output: 'world', durationMs: 50, confidence: 0.9 },
|
|
],
|
|
outcome: 'success',
|
|
durationMs: 100,
|
|
timestamp: new Date().toISOString(),
|
|
...overrides,
|
|
};
|
|
}
|
|
|
|
// ===== Tests =====
|
|
|
|
describe('RvfLearningStore', () => {
|
|
let dir: string;
|
|
let storePath: string;
|
|
let store: RvfLearningStore;
|
|
|
|
beforeEach(() => {
|
|
dir = tmpDir();
|
|
storePath = path.join(dir, 'test.rvls');
|
|
store = new RvfLearningStore({ storePath, autoPersistInterval: 0 });
|
|
});
|
|
|
|
afterEach(async () => {
|
|
try { await store.close(); } catch { /* already closed or not initialized */ }
|
|
fs.rmSync(dir, { recursive: true, force: true });
|
|
});
|
|
|
|
// ===== 1. Initialize =====
|
|
|
|
describe('initialize', () => {
|
|
it('creates parent directory if it does not exist', async () => {
|
|
const nested = path.join(dir, 'deep', 'nested', 'store.rvls');
|
|
const s = new RvfLearningStore({ storePath: nested, autoPersistInterval: 0 });
|
|
await s.initialize();
|
|
assert.ok(fs.existsSync(path.dirname(nested)));
|
|
await s.close();
|
|
});
|
|
|
|
it('loads existing data on init', async () => {
|
|
await store.initialize();
|
|
await store.savePatterns([makePattern()]);
|
|
await store.persist();
|
|
await store.close();
|
|
|
|
const store2 = new RvfLearningStore({ storePath, autoPersistInterval: 0 });
|
|
await store2.initialize();
|
|
const patterns = await store2.loadPatterns();
|
|
assert.equal(patterns.length, 1);
|
|
assert.equal(patterns[0].id, 'p1');
|
|
await store2.close();
|
|
});
|
|
|
|
it('is idempotent when called twice', async () => {
|
|
await store.initialize();
|
|
await store.initialize(); // should not throw
|
|
const count = await store.getPatternCount();
|
|
assert.equal(count, 0);
|
|
});
|
|
});
|
|
|
|
// ===== 2. Patterns =====
|
|
|
|
describe('patterns', () => {
|
|
before(async () => {});
|
|
|
|
it('savePatterns stores and loadPatterns retrieves', async () => {
|
|
await store.initialize();
|
|
const saved = await store.savePatterns([makePattern(), makePattern({ id: 'p2' })]);
|
|
assert.equal(saved, 2);
|
|
const loaded = await store.loadPatterns();
|
|
assert.equal(loaded.length, 2);
|
|
const ids = loaded.map((p) => p.id).sort();
|
|
assert.deepEqual(ids, ['p1', 'p2']);
|
|
});
|
|
|
|
it('getPatternCount returns correct count', async () => {
|
|
await store.initialize();
|
|
assert.equal(await store.getPatternCount(), 0);
|
|
await store.savePatterns([makePattern()]);
|
|
assert.equal(await store.getPatternCount(), 1);
|
|
await store.savePatterns([makePattern({ id: 'p2' }), makePattern({ id: 'p3' })]);
|
|
assert.equal(await store.getPatternCount(), 3);
|
|
});
|
|
|
|
it('updates existing pattern when ID matches', async () => {
|
|
await store.initialize();
|
|
await store.savePatterns([makePattern({ successRate: 0.5 })]);
|
|
await store.savePatterns([makePattern({ successRate: 0.99 })]);
|
|
assert.equal(await store.getPatternCount(), 1);
|
|
const loaded = await store.loadPatterns();
|
|
assert.equal(loaded[0].successRate, 0.99);
|
|
});
|
|
});
|
|
|
|
// ===== 3. LoRA =====
|
|
|
|
describe('lora', () => {
|
|
it('saveLoraAdapter stores and loadLoraAdapters retrieves', async () => {
|
|
await store.initialize();
|
|
await store.saveLoraAdapter(makeLora());
|
|
const adapters = await store.loadLoraAdapters();
|
|
assert.equal(adapters.length, 1);
|
|
assert.equal(adapters[0].id, 'lora-1');
|
|
assert.equal(adapters[0].numParameters, 1024);
|
|
});
|
|
|
|
it('deleteLoraAdapter removes adapter and returns true', async () => {
|
|
await store.initialize();
|
|
await store.saveLoraAdapter(makeLora());
|
|
const deleted = await store.deleteLoraAdapter('lora-1');
|
|
assert.equal(deleted, true);
|
|
const adapters = await store.loadLoraAdapters();
|
|
assert.equal(adapters.length, 0);
|
|
});
|
|
|
|
it('deleteLoraAdapter returns false for nonexistent ID', async () => {
|
|
await store.initialize();
|
|
const deleted = await store.deleteLoraAdapter('nope');
|
|
assert.equal(deleted, false);
|
|
});
|
|
});
|
|
|
|
// ===== 4. EWC =====
|
|
|
|
describe('ewc', () => {
|
|
it('saveEwcState stores and loadEwcState retrieves', async () => {
|
|
await store.initialize();
|
|
await store.saveEwcState(makeEwc());
|
|
const state = await store.loadEwcState();
|
|
assert.ok(state);
|
|
assert.equal(state.tasksLearned, 5);
|
|
assert.equal(state.protectionStrength, 0.8);
|
|
});
|
|
|
|
it('loadEwcState returns null when no state stored', async () => {
|
|
await store.initialize();
|
|
const state = await store.loadEwcState();
|
|
assert.equal(state, null);
|
|
});
|
|
|
|
it('saveEwcState replaces previous state', async () => {
|
|
await store.initialize();
|
|
await store.saveEwcState(makeEwc({ tasksLearned: 3 }));
|
|
await store.saveEwcState(makeEwc({ tasksLearned: 10 }));
|
|
const state = await store.loadEwcState();
|
|
assert.ok(state);
|
|
assert.equal(state.tasksLearned, 10);
|
|
});
|
|
});
|
|
|
|
// ===== 5. Trajectories =====
|
|
|
|
describe('trajectories', () => {
|
|
it('appendTrajectory adds a record', async () => {
|
|
await store.initialize();
|
|
await store.appendTrajectory(makeTrajectory());
|
|
assert.equal(await store.getTrajectoryCount(), 1);
|
|
});
|
|
|
|
it('getTrajectories returns newest first', async () => {
|
|
await store.initialize();
|
|
await store.appendTrajectory(makeTrajectory({ id: 't1', timestamp: '2026-01-01' }));
|
|
await store.appendTrajectory(makeTrajectory({ id: 't2', timestamp: '2026-01-02' }));
|
|
await store.appendTrajectory(makeTrajectory({ id: 't3', timestamp: '2026-01-03' }));
|
|
const trajs = await store.getTrajectories();
|
|
assert.equal(trajs.length, 3);
|
|
assert.equal(trajs[0].id, 't3'); // newest (last appended) first
|
|
assert.equal(trajs[2].id, 't1');
|
|
});
|
|
|
|
it('getTrajectoryCount returns correct count', async () => {
|
|
await store.initialize();
|
|
assert.equal(await store.getTrajectoryCount(), 0);
|
|
await store.appendTrajectory(makeTrajectory({ id: 't1' }));
|
|
await store.appendTrajectory(makeTrajectory({ id: 't2' }));
|
|
assert.equal(await store.getTrajectoryCount(), 2);
|
|
});
|
|
|
|
it('limit parameter restricts results', async () => {
|
|
await store.initialize();
|
|
for (let i = 0; i < 5; i++) {
|
|
await store.appendTrajectory(makeTrajectory({ id: `t${i}` }));
|
|
}
|
|
const limited = await store.getTrajectories(2);
|
|
assert.equal(limited.length, 2);
|
|
assert.equal(limited[0].id, 't4'); // newest first
|
|
assert.equal(limited[1].id, 't3');
|
|
});
|
|
});
|
|
|
|
// ===== 6. Persistence =====
|
|
|
|
describe('persistence', () => {
|
|
it('persist() writes file to disk', async () => {
|
|
await store.initialize();
|
|
await store.savePatterns([makePattern()]);
|
|
assert.ok(!fs.existsSync(storePath)); // not yet persisted
|
|
await store.persist();
|
|
assert.ok(fs.existsSync(storePath));
|
|
});
|
|
|
|
it('data survives close and reinitialize', async () => {
|
|
await store.initialize();
|
|
await store.savePatterns([makePattern(), makePattern({ id: 'p2' })]);
|
|
await store.saveLoraAdapter(makeLora());
|
|
await store.saveEwcState(makeEwc());
|
|
await store.appendTrajectory(makeTrajectory());
|
|
await store.persist();
|
|
await store.close();
|
|
|
|
const store2 = new RvfLearningStore({ storePath, autoPersistInterval: 0 });
|
|
await store2.initialize();
|
|
assert.equal(await store2.getPatternCount(), 2);
|
|
assert.equal((await store2.loadLoraAdapters()).length, 1);
|
|
assert.ok(await store2.loadEwcState());
|
|
assert.equal(await store2.getTrajectoryCount(), 1);
|
|
await store2.close();
|
|
});
|
|
|
|
it('file starts with RVLS magic header', async () => {
|
|
await store.initialize();
|
|
await store.savePatterns([makePattern()]);
|
|
await store.persist();
|
|
const content = fs.readFileSync(storePath, 'utf-8');
|
|
assert.ok(content.startsWith('RVLS\n'));
|
|
});
|
|
|
|
it('skips persist when store is not dirty', async () => {
|
|
await store.initialize();
|
|
await store.persist(); // nothing changed, should be no-op
|
|
assert.ok(!fs.existsSync(storePath));
|
|
});
|
|
|
|
it('close() auto-persists dirty state', async () => {
|
|
await store.initialize();
|
|
await store.savePatterns([makePattern()]);
|
|
await store.close();
|
|
assert.ok(fs.existsSync(storePath));
|
|
|
|
const store2 = new RvfLearningStore({ storePath, autoPersistInterval: 0 });
|
|
await store2.initialize();
|
|
assert.equal(await store2.getPatternCount(), 1);
|
|
await store2.close();
|
|
});
|
|
});
|
|
|
|
// ===== 7. Stats =====
|
|
|
|
describe('stats', () => {
|
|
it('returns correct counts and file size', async () => {
|
|
await store.initialize();
|
|
await store.savePatterns([makePattern(), makePattern({ id: 'p2' })]);
|
|
await store.saveLoraAdapter(makeLora());
|
|
await store.saveEwcState(makeEwc());
|
|
await store.appendTrajectory(makeTrajectory());
|
|
await store.appendTrajectory(makeTrajectory({ id: 't2' }));
|
|
await store.persist();
|
|
|
|
const stats = await store.getStats();
|
|
assert.equal(stats.patterns, 2);
|
|
assert.equal(stats.loraAdapters, 1);
|
|
assert.equal(stats.trajectories, 2);
|
|
assert.equal(stats.hasEwcState, true);
|
|
assert.ok(stats.fileSizeBytes > 0);
|
|
});
|
|
|
|
it('fileSizeBytes is 0 when nothing persisted', async () => {
|
|
await store.initialize();
|
|
const stats = await store.getStats();
|
|
assert.equal(stats.fileSizeBytes, 0);
|
|
assert.equal(stats.hasEwcState, false);
|
|
});
|
|
});
|
|
|
|
// ===== 8. Error Handling =====
|
|
|
|
describe('error handling', () => {
|
|
it('operations before initialize throw', async () => {
|
|
const uninit = new RvfLearningStore({ storePath, autoPersistInterval: 0 });
|
|
await assert.rejects(() => uninit.savePatterns([makePattern()]), {
|
|
message: /not been initialized/,
|
|
});
|
|
await assert.rejects(() => uninit.loadPatterns(), { message: /not been initialized/ });
|
|
await assert.rejects(() => uninit.getPatternCount(), { message: /not been initialized/ });
|
|
await assert.rejects(() => uninit.saveLoraAdapter(makeLora()), {
|
|
message: /not been initialized/,
|
|
});
|
|
await assert.rejects(() => uninit.loadLoraAdapters(), { message: /not been initialized/ });
|
|
await assert.rejects(() => uninit.deleteLoraAdapter('x'), {
|
|
message: /not been initialized/,
|
|
});
|
|
await assert.rejects(() => uninit.saveEwcState(makeEwc()), {
|
|
message: /not been initialized/,
|
|
});
|
|
await assert.rejects(() => uninit.loadEwcState(), { message: /not been initialized/ });
|
|
await assert.rejects(() => uninit.appendTrajectory(makeTrajectory()), {
|
|
message: /not been initialized/,
|
|
});
|
|
await assert.rejects(() => uninit.getTrajectories(), { message: /not been initialized/ });
|
|
await assert.rejects(() => uninit.getTrajectoryCount(), {
|
|
message: /not been initialized/,
|
|
});
|
|
await assert.rejects(() => uninit.getStats(), { message: /not been initialized/ });
|
|
});
|
|
|
|
it('corrupt file is handled gracefully (bad header)', async () => {
|
|
fs.writeFileSync(storePath, 'NOT_RVLS\n{"type":"pattern","data":{}}\n');
|
|
await store.initialize(); // should not throw
|
|
assert.equal(await store.getPatternCount(), 0);
|
|
});
|
|
|
|
it('corrupt file is handled gracefully (bad JSON lines)', async () => {
|
|
fs.writeFileSync(storePath, 'RVLS\n{invalid json}\n{"type":"pattern","data":' +
|
|
JSON.stringify(makePattern()) + '}\n');
|
|
await store.initialize(); // should not throw, skips bad line
|
|
assert.equal(await store.getPatternCount(), 1);
|
|
});
|
|
});
|
|
|
|
// ===== 9. Edge Cases =====
|
|
|
|
describe('edge cases', () => {
|
|
it('empty store returns empty arrays and zero counts', async () => {
|
|
await store.initialize();
|
|
assert.deepEqual(await store.loadPatterns(), []);
|
|
assert.deepEqual(await store.loadLoraAdapters(), []);
|
|
assert.equal(await store.loadEwcState(), null);
|
|
assert.deepEqual(await store.getTrajectories(), []);
|
|
assert.equal(await store.getPatternCount(), 0);
|
|
assert.equal(await store.getTrajectoryCount(), 0);
|
|
});
|
|
|
|
it('large number of entries persists and reloads correctly', async () => {
|
|
await store.initialize();
|
|
const patterns: PatternRecord[] = [];
|
|
for (let i = 0; i < 200; i++) {
|
|
patterns.push(makePattern({ id: `p${i}`, useCount: i }));
|
|
}
|
|
await store.savePatterns(patterns);
|
|
for (let i = 0; i < 50; i++) {
|
|
await store.appendTrajectory(makeTrajectory({ id: `t${i}` }));
|
|
}
|
|
await store.persist();
|
|
await store.close();
|
|
|
|
const store2 = new RvfLearningStore({ storePath, autoPersistInterval: 0 });
|
|
await store2.initialize();
|
|
assert.equal(await store2.getPatternCount(), 200);
|
|
assert.equal(await store2.getTrajectoryCount(), 50);
|
|
const last = (await store2.loadPatterns()).find((p) => p.id === 'p199');
|
|
assert.ok(last);
|
|
assert.equal(last.useCount, 199);
|
|
await store2.close();
|
|
});
|
|
|
|
it('savePatterns returns count of items passed', async () => {
|
|
await store.initialize();
|
|
assert.equal(await store.savePatterns([]), 0);
|
|
assert.equal(await store.savePatterns([makePattern()]), 1);
|
|
});
|
|
|
|
it('stored data is a copy, not a reference', async () => {
|
|
await store.initialize();
|
|
const p = makePattern();
|
|
await store.savePatterns([p]);
|
|
p.successRate = 0.0; // mutate original
|
|
const loaded = await store.loadPatterns();
|
|
assert.equal(loaded[0].successRate, 0.95); // should still be original value
|
|
});
|
|
|
|
it('getTrajectories with limit larger than count returns all', async () => {
|
|
await store.initialize();
|
|
await store.appendTrajectory(makeTrajectory({ id: 't1' }));
|
|
const result = await store.getTrajectories(100);
|
|
assert.equal(result.length, 1);
|
|
});
|
|
});
|
|
});
|