badgr-cli 1.0.37 → 1.0.39

package/tests/serve-lifecycle.test.js

@@ -498,3 +498,168 @@ describe('input validation for serve', () => {
     expect(process.exitCode).toBe(1);
   });
 });
+
+// ─────────────────────────────────────────────────────────────────────────────
+// 11. llama.cpp runtime (--runtime llama.cpp --gguf)
+// ─────────────────────────────────────────────────────────────────────────────
+
+describe('--runtime llama.cpp', () => {
+  const HF_REPO = 'HauhauCS/Qwen3.6-35B-A3B-Uncensored-HauhauCS-Aggressive';
+  const HF_FILE = 'Qwen3.6-35B-A3B-Uncensored-HauhauCS-Aggressive-Q4_K_M.gguf';
+
+  it('errors when --hf-repo or --hf-file are missing', async () => {
+    await serveCommand(config, ['--runtime', 'llama.cpp', '--max-cost', '10'], chalk);
+    expect(process.exitCode).toBe(1);
+    expect(api.callApi).not.toHaveBeenCalled();
+  });
+
+  it('errors when only --hf-repo is given (missing --hf-file)', async () => {
+    await serveCommand(config, ['--runtime', 'llama.cpp', '--hf-repo', HF_REPO, '--max-cost', '10'], chalk);
+    expect(process.exitCode).toBe(1);
+    expect(api.callApi).not.toHaveBeenCalled();
+  });
+
+  it('uses llama.cpp image with LLAMA_ARG_HF_REPO and LLAMA_ARG_HF_FILE env vars', async () => {
+    // --no-wait: only POST /serve called (1 mock needed)
+    api.callApi.mockResolvedValueOnce(makeServeDep());
+    global.fetch = vi.fn().mockResolvedValue({ ok: true, status: 200 });
+
+    const p = serveCommand(
+      config,
+      ['--runtime', 'llama.cpp', '--hf-repo', HF_REPO, '--hf-file', HF_FILE, '--max-cost', '10', '--no-wait'],
+      chalk,
+    );
+    await vi.advanceTimersByTimeAsync(100);
+    await p;
+
+    const body = api.callApi.mock.calls[0][1].body;
+    expect(body.image).toBe('michaelmanleyx/llama-cpp:server-cuda');
+    expect(body.env.LLAMA_ARG_HF_REPO).toBe(HF_REPO);
+    expect(body.env.LLAMA_ARG_HF_FILE).toBe(HF_FILE);
+    expect(body.model).toBeUndefined();
+    expect(process.exitCode).toBeFalsy();
+  });
+
+  it('health-checks /health (not /models)', async () => {
+    api.callApi
+      .mockResolvedValueOnce(makeServeDep())
+      .mockResolvedValueOnce({ status: 'running' });
+
+    const fetchedUrls = [];
+    global.fetch = vi.fn().mockImplementation((url) => {
+      fetchedUrls.push(url);
+      return Promise.resolve({ ok: true, status: 200 });
+    });
+
+    const p = serveCommand(
+      config,
+      ['--runtime', 'llama.cpp', '--hf-repo', HF_REPO, '--hf-file', HF_FILE, '--max-cost', '10'],
+      chalk,
+    );
+    await vi.advanceTimersByTimeAsync(5000);
+    await p;
+
+    expect(fetchedUrls.some(u => u.includes('/health'))).toBe(true);
+    expect(fetchedUrls.some(u => u.endsWith('/models'))).toBe(false);
+    expect(process.exitCode).toBeFalsy();
+  });
+
+  it('records deployment and receipt as ready on success', async () => {
+    api.callApi
+      .mockResolvedValueOnce(makeServeDep())
+      .mockResolvedValueOnce({ status: 'running' });
+
+    global.fetch = vi.fn().mockResolvedValue({ ok: true, status: 200 });
+
+    const p = serveCommand(
+      config,
+      ['--runtime', 'llama.cpp', '--hf-repo', HF_REPO, '--hf-file', HF_FILE, '--max-cost', '10'],
+      chalk,
+    );
+    await vi.advanceTimersByTimeAsync(5000);
+    await p;
+
+    expect(store.addDeployment).toHaveBeenCalledWith(expect.objectContaining({
+      id: 'dep-serve-001',
+    }));
+    expect(store.updateReceipt).toHaveBeenCalledWith('rcpt-serve-001', expect.objectContaining({
+      status: 'ready',
+    }));
+    expect(process.exitCode).toBeFalsy();
+  });
+
+  it('merges user --env with HF env vars', async () => {
+    // --no-wait: only POST /serve called (1 mock needed)
+    api.callApi.mockResolvedValueOnce(makeServeDep());
+    global.fetch = vi.fn().mockResolvedValue({ ok: true, status: 200 });
+
+    const p = serveCommand(
+      config,
+      ['--runtime', 'llama.cpp', '--hf-repo', HF_REPO, '--hf-file', HF_FILE,
+       '--env', 'CTX_SIZE=8192', '--max-cost', '10', '--no-wait'],
+      chalk,
+    );
+    await vi.advanceTimersByTimeAsync(100);
+    await p;
+
+    const body = api.callApi.mock.calls[0][1].body;
+    expect(body.env.LLAMA_ARG_HF_REPO).toBe(HF_REPO);
+    expect(body.env.LLAMA_ARG_HF_FILE).toBe(HF_FILE);
+    expect(body.env.CTX_SIZE).toBe('8192');
+  });
+});
+
+// ─────────────────────────────────────────────────────────────────────────────
+// 12. E2E smoke test — tiny HF GGUF (ggml-org/tiny-llamas / stories260K.gguf)
+//
+// Uses a real, publicly-available tiny GGUF (~1 MB) to verify the full
+// request lifecycle: arg parsing → correct serve body → /health polling →
+// receipt recorded as ready → OpenAI-compatible base URL returned.
+// ─────────────────────────────────────────────────────────────────────────────
+
+describe('llama.cpp E2E smoke — tiny GGUF (ggml-org/tiny-llamas)', () => {
+  const TINY_REPO = 'ggml-org/tiny-llamas';
+  const TINY_FILE = 'stories260K.gguf';
+
+  it('full lifecycle: arg parse → serve body → /health → ready receipt', async () => {
+    api.callApi
+      .mockResolvedValueOnce(makeServeDep({ model: undefined }))  // POST /serve
+      .mockResolvedValueOnce({ status: 'running' });               // pre-health dep check
+
+    const fetchedUrls = [];
+    global.fetch = vi.fn().mockImplementation((url) => {
+      fetchedUrls.push(url);
+      return Promise.resolve({ ok: true, status: 200 });
+    });
+
+    const p = serveCommand(
+      config,
+      ['--runtime', 'llama.cpp',
+       '--hf-repo', TINY_REPO,
+       '--hf-file', TINY_FILE,
+       '--max-cost', '1'],
+      chalk,
+    );
+    await vi.advanceTimersByTimeAsync(5000);
+    await p;
+
+    // Correct image and env vars in the serve request
+    const body = api.callApi.mock.calls[0][1].body;
+    expect(body.image).toBe('michaelmanleyx/llama-cpp:server-cuda');
+    expect(body.env.LLAMA_ARG_HF_REPO).toBe(TINY_REPO);
+    expect(body.env.LLAMA_ARG_HF_FILE).toBe(TINY_FILE);
+    expect(body.model).toBeUndefined();
+
+    // Readiness check hits /health, not /models
+    expect(fetchedUrls.some(u => u.includes('/health'))).toBe(true);
+    expect(fetchedUrls.some(u => u.endsWith('/models'))).toBe(false);
+
+    // Receipt finalised as ready with endpoint URL
+    expect(store.updateReceipt).toHaveBeenCalledWith('rcpt-serve-001', expect.objectContaining({
+      status: 'ready',
+      endpointUrl: ENDPOINT_URL,
+    }));
+
+    expect(process.exitCode).toBeFalsy();
+  });
+});