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, afterEach } from 'vitest';
import { fireEvent, render, screen, waitFor } from '@testing-library/react';
import MediaLightbox from './MediaLightbox';
// The footer's AddToCollectionMenu and prompt modals pull the whole API surface
// (and useProviderModels) into the import graph. Keep the mocks inert except
// for PromptRefineModal's open state, which lets the Escape regression assert
// that it closes without pulling its provider hooks into the test.
vi.mock('./AddToCollectionMenu', () => ({ default: () => null }));
vi.mock('./PromptRefineModal', () => ({
default: ({ open }) => open ? <div data-testid="refine-modal" /> : null,
}));
vi.mock('./FilmLookEditor.jsx', () => ({
default: ({ open }) => open ? <div data-testid="film-look-editor" /> : null,
}));
vi.mock('./PromptFromMedia', () => ({
default: () => null,
PromptFromMediaModal: ({ open }) => open ? <div data-testid="prompt-from-modal" /> : null,
}));
const videoItem = {
kind: 'video',
key: 'video:abc',
id: 'abc',
filename: 'abc.mp4',
previewUrl: '/data/video-thumbnails/abc.jpg',
downloadUrl: '/data/videos/abc.mp4',
prompt: 'a cat',
createdAt: Date.now(),
};
const imageItem = {
kind: 'image',
key: 'image:frame.png',
filename: 'frame.png',
previewUrl: '/data/images/frame.png',
downloadUrl: '/data/images/frame.png',
prompt: 'a cat portrait',
createdAt: Date.now(),
};
// The overlay portals to <body>, so it is outside render()'s container —
// query the whole document for the media element.
const videoEl = () => document.body.querySelector('video');
describe('MediaLightbox video element (mobile playback)', () => {
// jsdom doesn't implement HTMLMediaElement.play; stub it per-test so we can
// drive the unmute-on-open effect down both the granted and blocked paths.
let playMock;
beforeEach(() => {
playMock = vi.fn(() => Promise.resolve());
HTMLMediaElement.prototype.play = playMock;
});
afterEach(() => {
delete HTMLMediaElement.prototype.play;
});
it('renders the <video> with a poster + playsInline + muted autoplay baseline so it loads on mobile', () => {
render(<MediaLightbox item={videoItem} onClose={() => {}} />);
const video = videoEl();
expect(video).toBeTruthy();
// src points at the full asset
expect(video.getAttribute('src')).toBe('/data/videos/abc.mp4');
// poster = thumbnail so a blank box never shows while the clip buffers,
// and the frame is visible even if mobile autoplay is deferred.
expect(video.getAttribute('poster')).toBe('/data/video-thumbnails/abc.jpg');
// muted autoplay is the baseline that lets the clip start under the mobile
// media-engagement policy; the effect then unmutes for sound.
expect(video.hasAttribute('autoplay')).toBe(true);
// playsInline keeps iOS from promoting to a native fullscreen player.
expect(video.hasAttribute('playsinline')).toBe(true);
expect(video.hasAttribute('loop')).toBe(true);
expect(video.hasAttribute('controls')).toBe(true);
});
it('unmutes and plays for sound when the opening gesture allows audible playback', async () => {
render(<MediaLightbox item={videoItem} onClose={() => {}} />);
const video = videoEl();
await waitFor(() => expect(playMock).toHaveBeenCalled());
// play() resolved (gesture activation present) → stays unmuted for sound.
expect(video.muted).toBe(false);
});
it('falls back to muted playback when audible autoplay is blocked', async () => {
playMock.mockImplementation(() => Promise.reject(new Error('NotAllowedError')));
render(<MediaLightbox item={videoItem} onClose={() => {}} />);
const video = videoEl();
// First (unmuted) play rejects → effect re-mutes and re-plays so the clip
// still runs; the user can unmute via the controls.
await waitFor(() => expect(video.muted).toBe(true));
expect(playMock.mock.calls.length).toBeGreaterThanOrEqual(2);
});
it('omits poster when the video has no thumbnail rather than rendering an empty poster', () => {
render(
<MediaLightbox item={{ ...videoItem, previewUrl: null }} onClose={() => {}} />
);
const video = videoEl();
expect(video.hasAttribute('poster')).toBe(false);
});
});
describe('MediaLightbox overlay portal', () => {
it('portals the overlay to <body>, escaping a backdrop-filter containing-block ancestor', () => {
// Mirror a themed gallery: the lightbox is opened from inside a
// `.bg-port-card` tile, which gains `backdrop-filter` on "glass" themes. A
// backdrop-filter ancestor becomes the containing block for position:fixed
// descendants, so an inline overlay would be sized to the card instead of
// the viewport. The portal has to move it to <body> to escape that trap.
const { container } = render(
<div className="bg-port-card border rounded-xl" style={{ backdropFilter: 'blur(22px)' }} data-testid="glass-card">
<MediaLightbox item={imageItem} onClose={() => {}} />
</div>
);
const dialog = screen.getByRole('dialog');
expect(dialog.className).toContain('fixed inset-0');
expect(screen.getByTestId('glass-card').contains(dialog)).toBe(false);
expect(dialog.parentElement).toBe(document.body);
// The component's own rendered subtree holds nothing at all.
expect(container.querySelector('[role="dialog"]')).toBeNull();
});
});
describe('MediaLightbox touch gestures', () => {
it('keeps native pinch input from navigating the gallery', () => {
const onPrevious = vi.fn();
const onNext = vi.fn();
render(
<MediaLightbox
item={imageItem}
onClose={() => {}}
onPrevious={onPrevious}
onNext={onNext}
hasPrevious
hasNext
/>
);
const surface = screen.getByRole('img').parentElement;
expect(surface.style.touchAction).toBe('manipulation');
fireEvent.touchStart(surface, { touches: [{ clientX: 300, clientY: 400 }] });
fireEvent.touchStart(surface, {
touches: [
{ clientX: 300, clientY: 400 },
{ clientX: 500, clientY: 400 },
],
});
fireEvent.touchEnd(surface, {
touches: [{ clientX: 300, clientY: 400 }],
changedTouches: [{ clientX: 80, clientY: 400 }],
});
fireEvent.touchEnd(surface, {
touches: [],
changedTouches: [{ clientX: 80, clientY: 400 }],
});
expect(onPrevious).not.toHaveBeenCalled();
expect(onNext).not.toHaveBeenCalled();
});
it('keeps one-finger horizontal swipes for gallery navigation', () => {
const onNext = vi.fn();
render(<MediaLightbox item={imageItem} onClose={() => {}} onNext={onNext} hasNext />);
const surface = screen.getByRole('img').parentElement;
fireEvent.touchStart(surface, { touches: [{ clientX: 300, clientY: 400 }] });
fireEvent.touchEnd(surface, {
touches: [],
changedTouches: [{ clientX: 80, clientY: 400 }],
});
expect(onNext).toHaveBeenCalledTimes(1);
});
});
describe('MediaLightbox Escape cascade', () => {
it('exits full screen without closing the lightbox', () => {
const onClose = vi.fn();
render(<MediaLightbox item={imageItem} onClose={onClose} />);
fireEvent.keyDown(screen.getByRole('button', { name: 'Full screen' }), { key: 'f' });
const exitFullScreen = screen.getByRole('button', { name: 'Exit full screen' });
fireEvent.keyDown(exitFullScreen, { key: 'Escape' });
expect(onClose).not.toHaveBeenCalled();
expect(screen.getByRole('button', { name: 'Full screen' })).toBeTruthy();
});
it('closes the refine modal without closing the lightbox', () => {
const onClose = vi.fn();
render(<MediaLightbox item={imageItem} onClose={onClose} />);
fireEvent.click(screen.getByRole('button', { name: 'Refine Prompt' }));
const refineModal = screen.getByTestId('refine-modal');
fireEvent.keyDown(refineModal, { key: 'Escape' });
expect(onClose).not.toHaveBeenCalled();
expect(screen.queryByTestId('refine-modal')).toBeNull();
expect(screen.getByRole('dialog')).toBeTruthy();
});
it('closes the prompt-from-media modal without closing the lightbox', () => {
const onClose = vi.fn();
render(<MediaLightbox item={imageItem} onClose={onClose} />);
fireEvent.click(screen.getByRole('button', { name: 'Prompt from this' }));
const promptFromModal = screen.getByTestId('prompt-from-modal');
fireEvent.keyDown(promptFromModal, { key: 'Escape' });
expect(onClose).not.toHaveBeenCalled();
expect(screen.queryByTestId('prompt-from-modal')).toBeNull();
expect(screen.getByRole('dialog')).toBeTruthy();
});
it('opens the film look editor from the image actions and closes it without closing the lightbox', () => {
const onClose = vi.fn();
render(<MediaLightbox item={imageItem} onClose={onClose} />);
fireEvent.click(screen.getByRole('button', { name: 'Film look' }));
const editor = screen.getByTestId('film-look-editor');
fireEvent.keyDown(editor, { key: 'Escape' });
expect(onClose).not.toHaveBeenCalled();
expect(screen.queryByTestId('film-look-editor')).toBeNull();
expect(screen.getByRole('dialog')).toBeTruthy();
});
it('closes the lightbox when Escape is pressed with nothing else open', () => {
const onClose = vi.fn();
render(<MediaLightbox item={imageItem} onClose={onClose} />);
fireEvent.keyDown(screen.getByRole('dialog'), { key: 'Escape' });
expect(onClose).toHaveBeenCalledTimes(1);
});
});
describe('MediaLightbox route-changing actions', () => {
it('closes the preview before Send to Video runs so query cleanup cannot override navigation', () => {
const calls = [];
const onClose = vi.fn(() => calls.push('close'));
const onSendToVideo = vi.fn(() => calls.push('send'));
render(
<MediaLightbox
item={imageItem}
onClose={onClose}
onSendToVideo={onSendToVideo}
/>
);
fireEvent.click(screen.getByRole('button', { name: /send to video/i }));
expect(onClose).toHaveBeenCalledTimes(1);
expect(onSendToVideo).toHaveBeenCalledWith(imageItem);
expect(calls).toEqual(['close', 'send']);
});
});
// The conditioning promise (#4874) is provenance: a clip whose opening frame was
// GENERATED rather than reproduced has to say so where the render is inspected.
// The Image Strength number alone never carried that distinction.
describe('MediaLightbox — i2v reference provenance (#4874)', () => {
// jsdom has no HTMLMediaElement.play; the lightbox calls it on mount for a
// video item, so stub it here the way the playback suite above does.
beforeEach(() => { HTMLMediaElement.prototype.play = vi.fn(() => Promise.resolve()); });
afterEach(() => { delete HTMLMediaElement.prototype.play; });
const videoItem = (raw) => ({
kind: 'video',
key: 'video:ref-1',
id: 'ref-1',
filename: 'ref-1.mp4',
downloadUrl: '/data/videos/ref-1.mp4',
prompt: 'a fox in the rain',
modelId: 'ltx25_mlx_q8',
numFrames: 121,
fps: 24,
createdAt: '2026-08-23T10:00:00.000Z',
raw,
});
it('names the promise and the strength that delivered it', () => {
render(<MediaLightbox item={videoItem({ i2vReferenceMode: 'inspire', imageStrength: 0.35 })} onClose={() => {}} />);
expect(screen.getByText('Reference')).toBeTruthy();
expect(screen.getByText('Inspire')).toBeTruthy();
expect(screen.getByText('Image strength')).toBeTruthy();
});
it('shows no Reference row for an anchored render, so the row means something', () => {
render(<MediaLightbox item={videoItem({})} onClose={() => {}} />);
expect(screen.queryByText('Reference')).toBeNull();
});
});
describe('MediaLightbox — local image execution provenance', () => {
it('distinguishes confirmed, degraded, and unknown legacy image execution', () => {
const { rerender } = render(<MediaLightbox item={{ ...imageItem, executionProvenance: { state: 'confirmed', effectiveDevice: 'cuda', placement: 'cuda+offload' } }} onClose={() => {}} />);
expect(screen.getByText('Execution')).toBeTruthy();
expect(screen.getByText('Confirmed · cuda (cuda+offload)')).toBeTruthy();
rerender(<MediaLightbox item={{ ...imageItem, executionProvenance: { state: 'degraded', requestedDevice: 'mps' } }} onClose={() => {}} />);
expect(screen.getByText('Degraded · CPU fallback (requested mps)')).toBeTruthy();
rerender(<MediaLightbox item={imageItem} onClose={() => {}} />);
expect(screen.getByText('Unknown (legacy runner)')).toBeTruthy();
});
it('does not mistake a malformed marker for confirmed execution', () => {
render(<MediaLightbox item={{ ...imageItem, executionProvenance: { state: 'malformed' } }} onClose={() => {}} />);
expect(screen.getByText('Unknown (invalid runner marker)')).toBeTruthy();
});
});
// Render duration (#5878) — the detail panel is where the number is read
// deliberately ("why did this one take four minutes?"), so it needs both the
// humanized value and the row label, and must not invent a row for a record
// that carries no timing.
describe('MediaLightbox render-time row', () => {
it('shows a humanized render time beside its label', () => {
render(<MediaLightbox item={{ ...imageItem, renderMs: 254_000 }} onClose={() => {}} />);
const label = screen.getByText('Render time');
expect(label.nextElementSibling).toHaveTextContent('4m 14s');
});
it('omits the row entirely when the record was never timed', () => {
render(<MediaLightbox item={{ ...imageItem, renderMs: null }} onClose={() => {}} />);
expect(screen.queryByText('Render time')).toBeNull();
});
});
// A compact gallery row (#8292) holds a prompt preview until the host hydrates
// its record: editing or refining it would treat the truncation as the prompt.
describe('MediaLightbox compact preview item', () => {
it('withholds prompt editing and refine until the full record arrives, and reports a failed load', () => {
const onPromptChange = vi.fn();
const compact = { ...imageItem, compact: true, prompt: 'a cat port…' };
const { rerender } = render(<MediaLightbox item={compact} onClose={() => {}} onPromptChange={onPromptChange} />);
expect(screen.queryByLabelText('Prompt')).toBeNull();
expect(screen.queryByRole('button', { name: 'Refine Prompt' })).toBeNull();
expect(screen.getByRole('status')).toHaveTextContent('Loading full details');
rerender(<MediaLightbox item={{ ...compact, detailError: true }} onClose={() => {}} onPromptChange={onPromptChange} />);
expect(screen.getByRole('alert')).toHaveTextContent('could not be loaded');
rerender(<MediaLightbox item={imageItem} onClose={() => {}} onPromptChange={onPromptChange} />);
expect(screen.getByLabelText('Prompt')).toHaveValue('a cat portrait');
expect(screen.getByRole('button', { name: 'Refine Prompt' })).toBeInTheDocument();
});
});
it('saves the current playhead, resets the poster, and offers a sharing copy', async () => {
HTMLMediaElement.prototype.play = vi.fn().mockResolvedValue();
const onPosterChange = vi.fn().mockResolvedValue({});
render(<MediaLightbox item={videoItem} onClose={() => {}} onPosterChange={onPosterChange} />);
videoEl().currentTime = 12.4;
fireEvent.click(screen.getByRole('button', { name: 'Use this frame as poster' }));
await waitFor(() => expect(onPosterChange).toHaveBeenCalledWith(videoItem, 12.4));
await waitFor(() => expect(screen.getByRole('button', { name: 'Reset poster' })).not.toBeDisabled());
fireEvent.click(screen.getByRole('button', { name: 'Reset poster' }));
await waitFor(() => expect(onPosterChange).toHaveBeenCalledWith(videoItem, null));
expect(screen.getByRole('button', { name: 'Download for sharing' })).toBeInTheDocument();
});