onbuzz 4.9.0 → 4.9.5

package/src/services/embeddings/__tests__/embeddingService.test.js

@@ -0,0 +1,417 @@
+/**
+ * EmbeddingService orchestrator tests.
+ *
+ * Pins:
+ *   - 'disabled' is a first-class state (consumers can detect via isEnabled).
+ *   - Provider is lazy-instantiated based on config; hot-swap via updateConfig.
+ *   - VectorStores are per-scope, get-or-created, file-backed under stateDir.
+ *   - Telemetry tracks calls + tokens, per-day buckets, and budget enforcement.
+ *   - Budget exhaustion throws RATE_LIMITED; local provider is exempt
+ *     from the cloud budget cap.
+ *   - Health surface reflects provider readiness + budget state.
+ *   - Per-scope store path layout matches the documented contract
+ *     (agents/<id>/<surface>.vec.json vs. global/<surface>.vec.json).
+ */
+
+import { describe, test, expect, jest, beforeEach, afterEach } from '@jest/globals';
+import os from 'os';
+import fsp from 'fs/promises';
+import path from 'path';
+import { EmbeddingService } from '../embeddingService.js';
+import {
+  EmbeddingError,
+  EMBEDDING_ERROR_CODES,
+  EmbeddingProvider,
+} from '../providerInterface.js';
+
+class StubProvider extends EmbeddingProvider {
+  constructor({ dimensions = 4, name = 'azure', fingerprint = 'azure:stub-v1' } = {}) {
+    super();
+    this._info = {
+      name,
+      modelId: 'stub-model',
+      dimensions,
+      maxBatchSize: 16,
+      maxTokensPerText: 100,
+      distinguishesQueryAndDocument: false,
+      isLocal: name === 'local',
+    };
+    this._fingerprint = fingerprint;
+    this.calls = 0;
+  }
+  getInfo() { return this._info; }
+  getModelFingerprint() { return this._fingerprint; }
+  async _embed(texts) {
+    this.calls += texts.length;
+    return texts.map((_, i) =>
+      Array.from({ length: this._info.dimensions }, (_, j) => (i + 1) * (j + 1) * 0.1));
+  }
+}
+
+/** Build an EmbeddingService with stub providers + fake apiKeyManager. */
+function makeService({ tmpDir, providerName = 'azure', dimensions = 4, configExtra = {} }) {
+  const apiKeyManager = {
+    getEmbeddingApiKey: jest.fn(() => 'stub-key'),
+  };
+  const stubProvider = new StubProvider({ name: providerName, dimensions });
+  const svc = new EmbeddingService({
+    config: { provider: providerName, ...configExtra },
+    apiKeyManager,
+    baseUrl: 'https://stub-backend.example',
+    stateDir: tmpDir,
+    providerOverrides: { [providerName]: () => stubProvider },
+  });
+  return { svc, stubProvider, apiKeyManager };
+}
+
+describe('EmbeddingService — disabled by default', () => {
+  let tmpDir;
+  beforeEach(async () => {
+    tmpDir = await fsp.mkdtemp(path.join(os.tmpdir(), 'loxia-emb-svc-'));
+  });
+  afterEach(async () => {
+    await fsp.rm(tmpDir, { recursive: true, force: true });
+  });
+
+  test('no config → disabled', () => {
+    const svc = new EmbeddingService({
+      apiKeyManager: { getEmbeddingApiKey: () => null },
+      baseUrl: 'https://x',
+      stateDir: tmpDir,
+    });
+    expect(svc.isEnabled).toBe(false);
+  });
+
+  test('disabled state: embedQuery throws NOT_INITIALIZED', async () => {
+    const svc = new EmbeddingService({
+      apiKeyManager: { getEmbeddingApiKey: () => null },
+      baseUrl: 'https://x',
+      stateDir: tmpDir,
+    });
+    await expect(svc.embedQuery('hi')).rejects.toMatchObject({
+      code: EMBEDDING_ERROR_CODES.NOT_INITIALIZED,
+    });
+  });
+
+  test('disabled state: health() returns ok:false with reason', async () => {
+    const svc = new EmbeddingService({
+      apiKeyManager: { getEmbeddingApiKey: () => null },
+      baseUrl: 'https://x',
+      stateDir: tmpDir,
+    });
+    const h = await svc.health();
+    expect(h.ok).toBe(false);
+    expect(h.reason).toMatch(/disabled/i);
+  });
+
+  test('disabled state: getStore throws', async () => {
+    const svc = new EmbeddingService({
+      apiKeyManager: { getEmbeddingApiKey: () => null },
+      baseUrl: 'https://x',
+      stateDir: tmpDir,
+    });
+    await expect(svc.getStore({ agentId: 'a1', surface: 'memory' })).rejects.toThrow(/disabled/i);
+  });
+});
+
+describe('EmbeddingService — provider routing', () => {
+  let tmpDir;
+  beforeEach(async () => { tmpDir = await fsp.mkdtemp(path.join(os.tmpdir(), 'loxia-emb-svc-')); });
+  afterEach(async () => { await fsp.rm(tmpDir, { recursive: true, force: true }); });
+
+  test('provider=azure: uses azure factory', () => {
+    const { svc, stubProvider } = makeService({ tmpDir, providerName: 'azure' });
+    expect(svc.getProvider()).toBe(stubProvider);
+    expect(svc.getProvider().getInfo().name).toBe('azure');
+  });
+
+  test('provider lookup is lazy (no work until first request)', () => {
+    const apiKeyManager = { getEmbeddingApiKey: jest.fn(() => 'k') };
+    const factory = jest.fn(() => new StubProvider());
+    const svc = new EmbeddingService({
+      config: { provider: 'azure' },
+      apiKeyManager,
+      baseUrl: 'https://x',
+      stateDir: tmpDir,
+      providerOverrides: { azure: factory },
+    });
+    expect(factory).not.toHaveBeenCalled();
+    svc.getProvider();
+    expect(factory).toHaveBeenCalledTimes(1);
+    svc.getProvider();
+    expect(factory).toHaveBeenCalledTimes(1);  // cached
+  });
+
+  test('unknown provider name surfaces NOT_INITIALIZED', () => {
+    const svc = new EmbeddingService({
+      config: { provider: 'cohere' },
+      apiKeyManager: { getEmbeddingApiKey: () => 'k' },
+      baseUrl: 'https://x',
+      stateDir: tmpDir,
+    });
+    expect(() => svc.getProvider()).toThrow(/cohere/);
+  });
+
+  test('updateConfig hot-swaps the provider', async () => {
+    const { svc, stubProvider } = makeService({ tmpDir, providerName: 'azure' });
+    expect(svc.getProvider()).toBe(stubProvider);
+    const local = new StubProvider({ name: 'local', dimensions: 8, fingerprint: 'local:foo' });
+    await svc.updateConfig({ provider: 'local' });
+    // After updateConfig we need a new override for the new provider name.
+    svc._providerOverrides.local = () => local;
+    expect(svc.getProvider()).toBe(local);
+    expect(svc.getProvider()).not.toBe(stubProvider);
+  });
+});
+
+describe('EmbeddingService — embed surfaces + telemetry', () => {
+  let tmpDir;
+  beforeEach(async () => { tmpDir = await fsp.mkdtemp(path.join(os.tmpdir(), 'loxia-emb-svc-')); });
+  afterEach(async () => { await fsp.rm(tmpDir, { recursive: true, force: true }); });
+
+  test('embedQuery increments telemetry', async () => {
+    const { svc } = makeService({ tmpDir });
+    expect(svc.getTelemetry().callsTotal).toBe(0);
+    await svc.embedQuery('hello world');
+    const t = svc.getTelemetry();
+    expect(t.callsTotal).toBe(1);
+    expect(t.tokensTotal).toBeGreaterThan(0);
+    expect(t.today.calls).toBe(1);
+    expect(t.lastCallAt).not.toBeNull();
+  });
+
+  test('embedDocuments counts batch size in call total', async () => {
+    const { svc } = makeService({ tmpDir });
+    await svc.embedDocuments(['a', 'b', 'c']);
+    expect(svc.getTelemetry().callsTotal).toBe(3);
+  });
+
+  test('lastError is captured on failure', async () => {
+    const apiKeyManager = { getEmbeddingApiKey: () => 'k' };
+    class FailingProvider extends StubProvider {
+      async _embed() {
+        throw new EmbeddingError(EMBEDDING_ERROR_CODES.RATE_LIMITED, 'too fast', { provider: 'azure', transient: true });
+      }
+    }
+    const svc = new EmbeddingService({
+      config: { provider: 'azure' },
+      apiKeyManager,
+      baseUrl: 'https://x',
+      stateDir: tmpDir,
+      providerOverrides: { azure: () => new FailingProvider() },
+    });
+    await expect(svc.embedQuery('hi')).rejects.toMatchObject({ code: EMBEDDING_ERROR_CODES.RATE_LIMITED });
+    const t = svc.getTelemetry();
+    expect(t.lastError).toMatchObject({ code: 'RATE_LIMITED' });
+  });
+});
+
+describe('EmbeddingService — budget enforcement', () => {
+  let tmpDir;
+  beforeEach(async () => { tmpDir = await fsp.mkdtemp(path.join(os.tmpdir(), 'loxia-emb-svc-')); });
+  afterEach(async () => { await fsp.rm(tmpDir, { recursive: true, force: true }); });
+
+  test('exceeding daily token limit (cloud) throws RATE_LIMITED', async () => {
+    const { svc } = makeService({
+      tmpDir,
+      providerName: 'azure',
+      configExtra: { budget: { dailyTokenLimit: 4 } }, // ~16 chars total
+    });
+    // First call uses ~estimateTokens('hello world')=3 tokens. Allowed.
+    await svc.embedQuery('hello world');
+    // Now should still allow the next small call but block once we cross 4.
+    await expect(svc.embedQuery('hello world')).rejects.toMatchObject({
+      code: EMBEDDING_ERROR_CODES.RATE_LIMITED,
+    });
+  });
+
+  test('local provider is exempt from the cloud token cap', async () => {
+    const { svc } = makeService({
+      tmpDir,
+      providerName: 'local',
+      configExtra: { budget: { dailyTokenLimit: 1 } }, // trivially exceeded
+    });
+    // Many calls; none should hit the budget.
+    for (let i = 0; i < 5; i++) {
+      // eslint-disable-next-line no-await-in-loop
+      await svc.embedQuery('a longer string that would exceed 1 token easily');
+    }
+    expect(svc.getTelemetry().callsTotal).toBe(5);
+  });
+
+  test('dailyTokenLimit=0 disables the cap entirely', async () => {
+    const { svc } = makeService({
+      tmpDir,
+      providerName: 'azure',
+      configExtra: { budget: { dailyTokenLimit: 0 } },
+    });
+    for (let i = 0; i < 3; i++) {
+      // eslint-disable-next-line no-await-in-loop
+      await svc.embedQuery('hello world');
+    }
+    expect(svc.getTelemetry().callsTotal).toBe(3);
+  });
+});
+
+describe('EmbeddingService — VectorStore per-scope', () => {
+  let tmpDir;
+  beforeEach(async () => { tmpDir = await fsp.mkdtemp(path.join(os.tmpdir(), 'loxia-emb-svc-')); });
+  afterEach(async () => { await fsp.rm(tmpDir, { recursive: true, force: true }); });
+
+  test('agent-scoped store path = embeddings/agents/<id>/<surface>.vec.json', async () => {
+    const { svc } = makeService({ tmpDir });
+    const store = await svc.getStore({ agentId: 'a1', surface: 'memory' });
+    expect(store._filePath).toBe(path.join(tmpDir, 'embeddings', 'agents', 'a1', 'memory.vec.json'));
+  });
+
+  test('global-scoped store path = embeddings/global/<surface>.vec.json', async () => {
+    const { svc } = makeService({ tmpDir });
+    const store = await svc.getStore({ surface: 'composio' });
+    expect(store._filePath).toBe(path.join(tmpDir, 'embeddings', 'global', 'composio.vec.json'));
+  });
+
+  test('getStore is get-or-create: second call returns same instance', async () => {
+    const { svc } = makeService({ tmpDir });
+    const a = await svc.getStore({ agentId: 'a1', surface: 'memory' });
+    const b = await svc.getStore({ agentId: 'a1', surface: 'memory' });
+    expect(a).toBe(b);
+  });
+
+  test('different scopes return different stores', async () => {
+    const { svc } = makeService({ tmpDir });
+    const a = await svc.getStore({ agentId: 'a1', surface: 'memory' });
+    const b = await svc.getStore({ agentId: 'a2', surface: 'memory' });
+    const c = await svc.getStore({ surface: 'composio' });
+    expect(a).not.toBe(b);
+    expect(a).not.toBe(c);
+    expect(b).not.toBe(c);
+  });
+
+  test('store carries provider fingerprint + dimensions', async () => {
+    const { svc, stubProvider } = makeService({ tmpDir, dimensions: 8 });
+    const store = await svc.getStore({ agentId: 'a1', surface: 'memory' });
+    const s = await store.stats();
+    expect(s.dimensions).toBe(8);
+    expect(s.modelFingerprint).toBe(stubProvider.getModelFingerprint());
+  });
+
+  test('updateConfig flushes the store map (new provider = re-load)', async () => {
+    const { svc } = makeService({ tmpDir, providerName: 'azure' });
+    const before = await svc.getStore({ agentId: 'a1', surface: 'memory' });
+    await svc.updateConfig({ provider: 'local' });
+    svc._providerOverrides.local = () => new StubProvider({ name: 'local', dimensions: 8, fingerprint: 'local:swap' });
+    const after = await svc.getStore({ agentId: 'a1', surface: 'memory' });
+    expect(after).not.toBe(before);
+    expect((await after.stats()).modelFingerprint).toBe('local:swap');
+  });
+
+  test('scope missing surface throws', async () => {
+    const { svc } = makeService({ tmpDir });
+    await expect(svc.getStore({ agentId: 'a1' })).rejects.toThrow(/surface/);
+    await expect(svc.getStore({})).rejects.toThrow(/surface/);
+  });
+});
+
+describe('EmbeddingService — warmup', () => {
+  let tmpDir;
+  beforeEach(async () => { tmpDir = await fsp.mkdtemp(path.join(os.tmpdir(), 'loxia-emb-svc-')); });
+  afterEach(async () => { await fsp.rm(tmpDir, { recursive: true, force: true }); });
+
+  test('warmup calls provider.warmup() when available', async () => {
+    class LocalLike extends StubProvider {
+      constructor() { super({ name: 'local' }); this.warmupCalled = 0; }
+      async warmup() { this.warmupCalled++; }
+    }
+    const provider = new LocalLike();
+    const svc = new EmbeddingService({
+      config: { provider: 'local' },
+      apiKeyManager: { getEmbeddingApiKey: () => null },
+      baseUrl: 'https://x',
+      stateDir: tmpDir,
+      providerOverrides: { local: () => provider },
+    });
+    await svc.warmup();
+    expect(provider.warmupCalled).toBe(1);
+  });
+
+  test('warmup is a no-op when disabled', async () => {
+    const svc = new EmbeddingService({
+      apiKeyManager: { getEmbeddingApiKey: () => null },
+      baseUrl: 'https://x',
+      stateDir: tmpDir,
+    });
+    await expect(svc.warmup()).resolves.toBeUndefined();
+  });
+});
+
+describe('EmbeddingService — config persistence (Settings UI path)', () => {
+  let tmpDir;
+  beforeEach(async () => { tmpDir = await fsp.mkdtemp(path.join(os.tmpdir(), 'loxia-emb-cfg-')); });
+  afterEach(async () => { await fsp.rm(tmpDir, { recursive: true, force: true }); });
+
+  function makeSvc(extra = {}) {
+    return new EmbeddingService({
+      apiKeyManager: { getEmbeddingApiKey: () => null },
+      baseUrl: 'https://x',
+      stateDir: tmpDir,
+      ...extra,
+    });
+  }
+
+  test('updateConfig({persist:true}) writes to disk; loadPersistedConfig reads it back', async () => {
+    const first = makeSvc();
+    await first.updateConfig({ provider: 'azure', budget: { dailyTokenLimit: 500 } }, { persist: true });
+    expect(first.config.provider).toBe('azure');
+
+    const second = makeSvc();
+    expect(second.config.provider).toBe('disabled');     // default before load
+    const loaded = await second.loadPersistedConfig();
+    expect(loaded).toBe(true);
+    expect(second.config.provider).toBe('azure');
+    expect(second.config.budget.dailyTokenLimit).toBe(500);
+  });
+
+  test('updateConfig({persist:false}) does NOT touch the file', async () => {
+    const svc = makeSvc();
+    await svc.updateConfig({ provider: 'local' });           // default persist=false
+
+    const second = makeSvc();
+    const loaded = await second.loadPersistedConfig();
+    expect(loaded).toBe(false);                              // nothing on disk
+    expect(second.config.provider).toBe('disabled');
+  });
+
+  test('loadPersistedConfig is graceful when the file is missing', async () => {
+    const svc = makeSvc();
+    const loaded = await svc.loadPersistedConfig();
+    expect(loaded).toBe(false);
+    expect(svc.config.provider).toBe('disabled');
+  });
+
+  test('loadPersistedConfig is graceful with corrupt JSON', async () => {
+    const file = path.join(tmpDir, 'embeddings', 'embeddings-config.json');
+    await fsp.mkdir(path.dirname(file), { recursive: true });
+    await fsp.writeFile(file, 'not json{{');
+    const svc = makeSvc();
+    const loaded = await svc.loadPersistedConfig();
+    expect(loaded).toBe(false);                              // ignored
+    expect(svc.config.provider).toBe('disabled');            // safe default
+  });
+
+  test('persist uses atomic .tmp + rename (no partial file on disk after success)', async () => {
+    const svc = makeSvc();
+    await svc.updateConfig({ provider: 'azure' }, { persist: true });
+    const dir = path.join(tmpDir, 'embeddings');
+    const entries = await fsp.readdir(dir);
+    expect(entries).toContain('embeddings-config.json');
+    expect(entries.filter(e => e.endsWith('.tmp'))).toEqual([]);
+  });
+
+  test('invalid config does NOT persist (validation fails first)', async () => {
+    const svc = makeSvc();
+    const result = await svc.updateConfig({ provider: 'cohere' }, { persist: true });
+    expect(result.ok).toBe(false);
+    const file = path.join(tmpDir, 'embeddings', 'embeddings-config.json');
+    await expect(fsp.access(file)).rejects.toMatchObject({ code: 'ENOENT' });
+  });
+});

package/src/services/embeddings/__tests__/embeddingsConfig.test.js

@@ -0,0 +1,182 @@
+/**
+ * embeddingsConfig — defaults, deep-merge, validation, model catalog,
+ * token estimator.
+ */
+
+import { describe, test, expect } from '@jest/globals';
+import {
+  DEFAULT_CONFIG,
+  PROVIDERS,
+  mergeConfig,
+  validateConfig,
+  getProviderModelOptions,
+  estimateTokens,
+} from '../embeddingsConfig.js';
+
+describe('DEFAULT_CONFIG', () => {
+  test('starts disabled', () => {
+    expect(DEFAULT_CONFIG.provider).toBe('disabled');
+  });
+
+  test('default local model is nomic-embed-text-v1.5', () => {
+    expect(DEFAULT_CONFIG.local.modelId).toBe('Xenova/nomic-embed-text-v1.5');
+  });
+
+  test('default cloud model is text-embedding-3-small (both providers)', () => {
+    expect(DEFAULT_CONFIG.azure.modelId).toBe('text-embedding-3-small');
+    expect(DEFAULT_CONFIG.openai.modelId).toBe('text-embedding-3-small');
+  });
+
+  test('default budget has a sane daily cap', () => {
+    expect(DEFAULT_CONFIG.budget.dailyTokenLimit).toBeGreaterThan(0);
+  });
+
+  test('PROVIDERS lists all known options', () => {
+    expect(PROVIDERS).toEqual(['local', 'azure', 'openai', 'azure-custom', 'disabled']);
+  });
+
+  test('azure-custom defaults to empty endpoint + deployment', () => {
+    expect(DEFAULT_CONFIG.azureCustom).toEqual(expect.objectContaining({
+      endpoint: '', deployment: '', dimensions: 1536, maxBatchSize: 16,
+    }));
+  });
+});
+
+describe('mergeConfig / validateConfig — azure-custom', () => {
+  test('mergeConfig honours azureCustom overrides', () => {
+    const m = mergeConfig({
+      provider: 'azure-custom',
+      azureCustom: {
+        endpoint: 'https://x.azure.com/openai/v1/',
+        deployment: 'embed-v-4-0',
+        dimensions: 1024,
+      },
+    });
+    expect(m.azureCustom.endpoint).toBe('https://x.azure.com/openai/v1/');
+    expect(m.azureCustom.deployment).toBe('embed-v-4-0');
+    expect(m.azureCustom.dimensions).toBe(1024);
+  });
+
+  test('validateConfig requires endpoint + deployment when provider=azure-custom', () => {
+    const v1 = validateConfig(mergeConfig({ provider: 'azure-custom' }));
+    expect(v1.ok).toBe(false);
+    expect(v1.errors.join(' ')).toMatch(/endpoint/);
+    expect(v1.errors.join(' ')).toMatch(/deployment/);
+
+    const v2 = validateConfig(mergeConfig({
+      provider: 'azure-custom',
+      azureCustom: { endpoint: 'x', deployment: 'm' },
+    }));
+    expect(v2.ok).toBe(true);
+  });
+
+  test('validateConfig rejects non-positive dimensions for azure-custom', () => {
+    const v = validateConfig(mergeConfig({
+      provider: 'azure-custom',
+      azureCustom: { endpoint: 'x', deployment: 'm', dimensions: -5 },
+    }));
+    expect(v.ok).toBe(false);
+    expect(v.errors.join(' ')).toMatch(/dimensions/);
+  });
+});
+
+describe('mergeConfig', () => {
+  test('null user config → defaults', () => {
+    expect(mergeConfig(null).provider).toBe('disabled');
+    expect(mergeConfig(undefined).local.modelId).toBe('Xenova/nomic-embed-text-v1.5');
+  });
+
+  test('top-level provider override wins', () => {
+    expect(mergeConfig({ provider: 'azure' }).provider).toBe('azure');
+  });
+
+  test('per-provider overrides merge (do NOT replace) defaults', () => {
+    const m = mergeConfig({ local: { modelId: 'Custom/x' } });
+    expect(m.local.modelId).toBe('Custom/x');
+    expect(m.local.maxBatchSize).toBe(32);   // default preserved
+  });
+
+  test('budget merges field-by-field', () => {
+    const m = mergeConfig({ budget: { dailyTokenLimit: 100 } });
+    expect(m.budget.dailyTokenLimit).toBe(100);
+  });
+
+  test('returned object is mutable (defaults stay frozen)', () => {
+    const m = mergeConfig({});
+    expect(() => { m.provider = 'azure'; }).not.toThrow();
+    // The frozen DEFAULT_CONFIG must NOT be mutated through the merge.
+    expect(DEFAULT_CONFIG.provider).toBe('disabled');
+  });
+});
+
+describe('validateConfig', () => {
+  test('rejects unknown provider', () => {
+    const v = validateConfig({ provider: 'cohere' });
+    expect(v.ok).toBe(false);
+    expect(v.errors.some(e => /provider/.test(e))).toBe(true);
+  });
+
+  test('local requires modelId', () => {
+    const v = validateConfig(mergeConfig({ provider: 'local', local: { modelId: '' } }));
+    expect(v.ok).toBe(false);
+    expect(v.errors.some(e => /local.modelId/.test(e))).toBe(true);
+  });
+
+  test('local.maxBatchSize must be positive', () => {
+    const v = validateConfig(mergeConfig({ provider: 'local', local: { maxBatchSize: -1 } }));
+    expect(v.ok).toBe(false);
+  });
+
+  test('valid config passes', () => {
+    expect(validateConfig(mergeConfig({ provider: 'azure' })).ok).toBe(true);
+    expect(validateConfig(mergeConfig({ provider: 'local' })).ok).toBe(true);
+    expect(validateConfig(mergeConfig({ provider: 'disabled' })).ok).toBe(true);
+  });
+
+  test('budget.dailyTokenLimit must be 0 or positive int', () => {
+    expect(validateConfig(mergeConfig({ provider: 'azure', budget: { dailyTokenLimit: 0 } })).ok).toBe(true);
+    expect(validateConfig(mergeConfig({ provider: 'azure', budget: { dailyTokenLimit: 100 } })).ok).toBe(true);
+    expect(validateConfig(mergeConfig({ provider: 'azure', budget: { dailyTokenLimit: -1 } })).ok).toBe(false);
+    expect(validateConfig(mergeConfig({ provider: 'azure', budget: { dailyTokenLimit: 0.5 } })).ok).toBe(false);
+  });
+});
+
+describe('getProviderModelOptions', () => {
+  test('local options come from the LOCAL_MODELS catalog', () => {
+    const opts = getProviderModelOptions('local');
+    expect(opts.length).toBeGreaterThanOrEqual(2);
+    const ids = opts.map(o => o.id);
+    expect(ids).toContain('Xenova/nomic-embed-text-v1.5');
+    expect(ids).toContain('Xenova/bge-small-en-v1.5');
+  });
+
+  test('azure + openai expose 3-small and 3-large', () => {
+    for (const provider of ['azure', 'openai']) {
+      const opts = getProviderModelOptions(provider);
+      expect(opts.map(o => o.id)).toEqual([
+        'text-embedding-3-small',
+        'text-embedding-3-large',
+      ]);
+    }
+  });
+
+  test('unknown provider → empty list (no throw)', () => {
+    expect(getProviderModelOptions('does-not-exist')).toEqual([]);
+  });
+});
+
+describe('estimateTokens', () => {
+  test('empty / non-string → 0', () => {
+    expect(estimateTokens('')).toBe(0);
+    expect(estimateTokens(null)).toBe(0);
+    expect(estimateTokens(undefined)).toBe(0);
+    expect(estimateTokens(42)).toBe(0);
+  });
+
+  test('~4 chars/token rule of thumb', () => {
+    expect(estimateTokens('a'.repeat(4))).toBe(1);
+    expect(estimateTokens('a'.repeat(40))).toBe(10);
+    // Ceil for partial tokens (5 chars → 2)
+    expect(estimateTokens('a'.repeat(5))).toBe(2);
+  });
+});