@ariakit/test@0.7.0

Renamed the includesHidden query variant to hidden

BREAKING if you use the includesHidden role-query variant from @ariakit/test.

The role-query variant that also matches elements hidden from the accessibility tree (such as inert or aria-hidden elements) has been renamed from includesHidden to hidden.

Before:

q.dialog.includesHidden("Settings");
q.menuitemcheckbox.all.includesHidden();

After:

q.dialog.hidden("Settings");
q.menuitemcheckbox.all.hidden();

Applied the browser shims for the whole test environment

BREAKING if you import helpers from individual subpaths such as @ariakit/test/click.

@ariakit/test now exposes only the @ariakit/test, @ariakit/test/react, and @ariakit/test/playwright entry points; the individual helper subpaths have been removed. Import the helpers from the root @ariakit/test instead.

Before:

import { click } from "@ariakit/test/click";

After:

import { click } from "@ariakit/test";

@ariakit/test installs browser shims (most importantly a getClientRects visibility shim that jsdom lacks). They were only active while a simulated interaction ran, but component code reads layout and focusability between interactions too — for example a dialog's auto-focus picks the first tabbable element with getFirstTabbableIn, which depends on getClientRects. When such a read ran outside an interaction (e.g. when asserting with expect.poll), it hit jsdom's empty layout and misbehaved. The shims now apply for the whole test environment, and are applied automatically when importing @ariakit/test or @ariakit/test/react.

Faster simulated interactions

The click, type, press, hover, and select helpers now settle between their internal sub-steps using microtasks and animation frames instead of a wall-clock delay, and the delay of the final settle after each interaction was shortened. The components under test schedule their per-step updates with microtasks and animation frames, so this speeds up test suites while preserving their behavior. Interactions that rely on a real timer (animations, tooltip/typeahead timeouts) are unaffected — pass an explicit delay to sleep(ms) when you need one.

Improved happy-dom support

When simulating interactions in a happy-dom environment, @ariakit/test now polyfills several gaps so behavior matches jsdom and real browsers:

• validationMessage returns a non-empty message for elements failing built-in constraint validation (happy-dom leaves it empty).
• Disabled <select> and <textarea> controls are excluded from FormData (happy-dom incorrectly includes them).
• The selectionchange event that happy-dom fires synchronously inside Selection.removeAllRanges() is deferred to a task, as the spec requires.
• window.alert is stubbed as a no-op (happy-dom doesn't implement it).

Batched requestAnimationFrame callbacks under happy-dom

@ariakit/test now runs happy-dom's requestAnimationFrame callbacks as a batched frame: all callbacks scheduled before a frame run together with a single shared timestamp, and a callback scheduled while the frame is running is deferred to the next frame. This matches the HTML specification, jsdom, and real browsers.

happy-dom otherwise runs each callback as its own task, so components that schedule work in the same frame could observe each other's mid-flight state — for example, an animated Dialog and its backdrop reading each other's computed styles while a leave animation is being set up. The batching keeps happy-dom's fast cadence, so test runs are not slowed down.

Ran click listeners and activation on disabled controls under happy-dom

@ariakit/test now runs both the event listeners and the activation behavior for a click dispatched on a disabled button or input under happy-dom. This matches jsdom and real browsers, which only bar clicks queued from user interaction on a disabled control — not a scripted dispatchEvent.

A disabled checkbox/radio now toggles before its click listener runs (which still sees it disabled) and fires input/change, reverting — and restoring the rest of the radio group — when a listener calls preventDefault(). Disabled submit/reset controls run their listeners without submitting or resetting the form.

happy-dom otherwise drops such a click entirely, so a test couldn't observe a click reaching a control that became disabled mid-interaction, and a disabled checkbox wouldn't toggle.

Added press.down and press.up

The press utility now provides press.down and press.up to fire only the keydown or keyup half of a key press, each with the same per-key shortcuts as press (press.down.Space(), press.up.Enter(), and so on). Both default to the currently focused element, so a key released after focus moved away — for example, an element that disables itself on keydown — lands where a real browser would deliver it.

// Press and release in two steps; the keyup lands wherever focus is at release
// time, not necessarily where the keydown happened.
await press.down.Space();
await press.up.Space();

Hardened interaction settling under load

The settle that runs after each simulated interaction now drains pending React work before resolving. React 18 renders concurrently while interactions run, so under CPU contention it can split a render or commit across several scheduler tasks; a fixed delay could return between two of them and leave the DOM momentarily unsettled. Waiting for those slices to finish makes assertions after click, press, type, and the other helpers reliable under load.

Populated window.event during simulated events under happy-dom

@ariakit/test now exposes the event currently being dispatched on the legacy window.event global for the synchronous duration of each simulated event under happy-dom, matching jsdom and real browsers.

React 18 reads window.event to give state updates triggered from native event listeners discrete-event priority. happy-dom doesn't implement the global, so without this those updates fell back to a lower priority and flushed later, reordering commits — for example, a controlled Dialog hidden by clicking outside could momentarily re-open and restore focus to the wrong element. Simulated interactions on React 18 now match the other environments; the divergence this fixes doesn't reproduce under React 19.

Other updates

• Slightly reduced the default delay between simulated interactions in non-browser test environments, speeding up test suites.
• Added JSDoc descriptions and usage examples to the @ariakit/test public API, surfaced in editor hover hints and the package's API reference.
• Added .lazy to @ariakit/test queries so a query can be stored and re-run later by calling the returned function.
• Fixed press.Enter to respect disabled default submit buttons and activate enabled ones during implicit form submission.
• Fixed @ariakit/test event helpers to use empty-string defaults instead of "undefined" for omitted keyboard and input event string fields.
• Fixed press.Enter() and type("\n") to emit an Enter keypress with charCode 13 when typing into text fields.
• Fixed mouseDown() preserving or clearing the document selection based on the pressed target, matching browser behavior for native non-text controls and text targets.
• Fixed click() on native options to preserve default selectedness and anchor shift-click selection from the current selected option.
• Fixed click, tap, mouseDown, and mouseUp to suppress compatibility mouse events after a canceled pointerdown.
• Fixed press extending text selections past the anchor when pressing Shift+Home or Shift+End.
• Fixed type so readonly fields, prevented keydowns, non-text-field targets, and no-op deletion keys no longer cause blur or focus to dispatch a change event when the value did not change.
• Fixed nested test helpers to reuse browser polyfills and act-environment overrides instead of reapplying them.
• Updated dependencies: @ariakit/utils@0.1.3