Self-hosted, Self-Improving, Federated Everything App — dev management, agent orchestration, persistent mind, digital twin, second brain, and creative pipeline from a single dashboard over Tailscale to port the user into the OS
JavaScript ai-agentsdashboarddeveloper-toolsdigital-twinquantified-selfsecond-brainself-hostedtailscale
import { describe, it, expect, vi, beforeEach } from 'vitest';
import { render, screen, fireEvent, act } from '@testing-library/react';
// Drive the component off a mocked training hook so we don't need a real mic /
// AudioContext (neither exists in jsdom). The hook + the pure progress math are
// tested on their own (useColorMatch + songProgress.test.js).
const start = vi.fn();
const stop = vi.fn();
let hookState = { running: false, countingIn: false, noteColors: {}, activeIndex: null, lastSummary: null };
vi.mock('../../hooks/useSongTraining.js', () => ({
__esModule: true,
default: () => ({ ...hookState, start, stop }),
}));
import SongTraining from './SongTraining.jsx';
import { recordAttempt, WHOLE_SONG_SCOPE } from '../../lib/songProgress.js';
const SCORE = [
'clef: treble', 'key: G', 'time: 4/4', 'tempo: 68', '',
'| [G] D4q(If) D4q(you) | [G] B4h(miss) |',
'',
'| [Em] B4h(on) | [C] C5q(know) G4q(am) |',
].join('\n');
describe('SongTraining', () => {
beforeEach(() => {
start.mockClear();
stop.mockClear();
hookState = { running: false, countingIn: false, noteColors: {}, activeIndex: null, lastSummary: null };
});
it('prompts to add a melody when the song has no notated music', () => {
render(<SongTraining score="" />);
expect(screen.getByText(/add a notated melody/i)).toBeTruthy();
});
it('renders the scope picker with the whole song and each section', () => {
render(<SongTraining score={SCORE} lyricSections={[{ label: 'Verse 1' }, { label: 'Chorus' }]} />);
const select = screen.getByLabelText(/practice/i);
const options = Array.from(select.querySelectorAll('option')).map((o) => o.textContent);
expect(options[0]).toMatch(/whole song/i);
expect(options.some((o) => /Verse 1/.test(o))).toBe(true);
expect(options.some((o) => /Chorus/.test(o))).toBe(true);
});
it('shows a Start button that triggers a run after the mic opens', async () => {
// getUserMedia isn't in jsdom — provide a stub the component awaits.
const getUserMedia = vi.fn().mockResolvedValue({ getTracks: () => [] });
Object.defineProperty(global.navigator, 'mediaDevices', {
value: { getUserMedia }, configurable: true,
});
render(<SongTraining score={SCORE} />);
fireEvent.click(screen.getByRole('button', { name: /start/i }));
expect(getUserMedia).toHaveBeenCalled();
// The Start handler awaits getUserMedia then arms the run via setState —
// flush that resolution inside act() so it doesn't fire after the test returns.
await act(async () => {});
});
it('renders progress for trained scopes', () => {
// Whole song trained to 90% once.
const progress = recordAttempt(null, WHOLE_SONG_SCOPE, { percentInTune: 90, graded: 6 });
render(<SongTraining score={SCORE} progress={progress} />);
// The whole-song best/avg (both 90% after one take) are surfaced.
expect(screen.getAllByText(/90%/).length).toBeGreaterThan(0);
// The memorization bar label is present.
expect(screen.getByText(/memorization/i)).toBeTruthy();
});
it('surfaces the last-take readout from the hook', () => {
hookState = { ...hookState, lastSummary: { percentInTune: 75, graded: 4 } };
render(<SongTraining score={SCORE} />);
expect(screen.getByText(/last take/i)).toBeTruthy();
expect(screen.getByText(/75%/)).toBeTruthy();
});
it('closes the mic when a non-loop run finishes naturally', async () => {
// Track stop() on the mic's track so we can assert teardown without a real
// AudioContext. The hook is mocked, so we drive `running` true→false by hand
// and re-render — that edge is the natural-completion signal.
const trackStop = vi.fn();
const getUserMedia = vi.fn().mockResolvedValue({ getTracks: () => [{ stop: trackStop }] });
Object.defineProperty(global.navigator, 'mediaDevices', {
value: { getUserMedia }, configurable: true,
});
const { rerender } = render(<SongTraining score={SCORE} />);
fireEvent.click(screen.getByRole('button', { name: /start/i }));
await screen.findByRole('button', { name: /stop/i }); // mic landed → armed run
// Hook reports the run started, then finished (running flips true → false).
hookState = { ...hookState, running: true };
rerender(<SongTraining score={SCORE} />);
hookState = { ...hookState, running: false };
rerender(<SongTraining score={SCORE} />);
// Natural completion (loop off) must stop the mic track and drop back to Start.
expect(trackStop).toHaveBeenCalled();
expect(screen.getByRole('button', { name: /start/i })).toBeTruthy();
});
it('warns when the browser kept processing on for the training mic', async () => {
const getUserMedia = vi.fn().mockResolvedValue({
getTracks: () => [{ stop: vi.fn(), getSettings: () => ({ noiseSuppression: true }) }],
});
Object.defineProperty(global.navigator, "mediaDevices", {
value: { getUserMedia }, configurable: true,
});
render(<SongTraining score={SCORE} />);
fireEvent.click(screen.getByRole("button", { name: /start/i }));
await act(async () => {});
expect(screen.getByText(/Browser audio processing is on \(noise suppression\)/)).toBeTruthy();
});
});