Fixes the 32 failing tests in WPT /dom/events/passive-by-default.html
(68/100 -> 100/100): per the DOM spec's "default passive value",
touchstart, touchmove, wheel and mousewheel listeners added to the
Window, the Document, or the html or body elements default to
{passive: true} (so their preventDefault is ignored), unless the
options explicitly say {passive: false}.
addEventListener's options dictionary now keeps `passive` optional so
an absent or undefined member can fall back to the default passive
value computed from the event type and target; an explicit boolean
keeps its meaning.
Coverage: /dom/events/passive-by-default.html 68/100 -> 100/100. No
regressions across /dom/events.
Co-Authored-By: Claude Fable 5 <noreply@anthropic.com>
Fixes the "Real clicks on disabled elements must not dispatch events"
subtest of WPT /dom/events/Event-dispatch-on-disabled-elements.html
(verified in isolation): a testdriver click on a disabled
button/input/select/textarea must not fire any pointer or mouse events,
while a click on the re-enabled element must. WebDriver.click now
checks the disabled state like HTMLElement.click does.
The file's overall score is unchanged (4/9) because its four CSS
transition/animation subtests hang awaiting animation events (no CSS
animation engine in the renderless browser) and, being sequential
promise_tests, also mask this subtest's result in the full run.
Co-Authored-By: Claude Fable 5 <noreply@anthropic.com>
Fixes 3 failing tests in WPT /dom/events/Event-dispatch-click.html
(29/33 -> 32/33) and both tests in /dom/events/no-focus-events-at-
clicking-editable-content-in-link.html (1/2 -> 2/2).
- Only a MouseEvent click is an activation event.
- The activation target is the nearest inclusive ancestor with
activation behavior (ancestors only for bubbling events), shared by
ActivationState and handleClick via findClickActivationTarget.
- Clicks on editable content don't activate enclosing links; mousedown
focuses the editing host (CDP and WebDriver paths).
- Interactive submission of a disconnected form does nothing; the
programmatic form.submit() stays exempt per HTML 'cannot navigate'.
The remaining Event-dispatch-click.html failure needs javascript: URL
navigation.
Coverage: Event-dispatch-click.html 29/33 -> 32/33,
no-focus-events-at-clicking-editable-content-in-link.html 1/2 -> 2/2.
No regressions across /dom/events.
Co-Authored-By: Claude Fable 5 <noreply@anthropic.com>
Fixes 5 failing tests across WPT /dom/events/relatedTarget.window.html
(0/6 -> 3/6) and /dom/events/shadow-relatedTarget.html (0/2 -> 2/2).
The remaining 3 relatedTarget.window.html subtests also fail in Chrome
stable (verified on wpt.fyi), so they are out of scope.
Per the DOM dispatch algorithm:
- The event's relatedTarget is retargeted against the dispatch target
when dispatch starts (Event gains relatedTargetPtr() to reach the
MouseEvent/FocusEvent storage generically), so listeners and
post-dispatch reads observe e.g. the shadow host instead of a node
inside a shadow tree.
- The "clear targets" decision (reset target and relatedTarget to null
when they would expose nodes whose root is a shadow root) is now
computed when the event path is built — on the pre-dispatch tree —
instead of after dispatch, so listeners moving nodes out of the
shadow tree during dispatch can't defeat it, and it now also applies
to relatedTarget, not just target.
Coverage: /dom/events/relatedTarget.window.html 0/6 -> 3/6 (Chrome
parity), /dom/events/shadow-relatedTarget.html 0/2 -> 2/2. No
regressions across /dom/events.
Co-Authored-By: Claude Fable 5 <noreply@anthropic.com>
Fixes all 4 failing tests in WPT
/dom/events/Event-timestamp-high-resolution.html: event.timeStamp
returned a raw monotonic-clock value (milliseconds since boot) instead
of a DOMHighResTimeStamp sharing performance.now()'s time origin.
- Events now record their creation time with the same (already
coarsened) clock Performance uses, and the timeStamp getter reports
it in milliseconds relative to the time origin.
- Per spec the origin is the *event's relevant global*'s: the JS Event
constructor captures the creating realm's origin so cross-realm reads
(Event-timestamp-cross-realm-getter.html) stay correct; events
without a captured origin fall back to the accessing realm's, which
is the creating realm in all other paths.
Coverage: /dom/events/Event-timestamp-high-resolution.html 0/4 -> 4/4,
Event-timestamp-cross-realm-getter.html stays 1/1. No regressions
across /dom/events; Event-dispatch-single-activation-behavior.html now
runs (0/0 crash -> 3/132).
Co-Authored-By: Claude Fable 5 <noreply@anthropic.com>
Fixes the 20 failing tests in WPT /dom/events/EventTarget-
dispatchEvent.html (5/25 -> 25/25), all of the form "If the event's
initialized flag is not set, an InvalidStateError must be thrown".
Per the DOM spec, an event created via document.createEvent has its
initialized flag unset until one of the legacy init*Event methods runs,
and dispatchEvent must throw an InvalidStateError for it. Lightpanda
dispatched such events happily. Event now carries an _initialized flag:
true by default (constructor-created and internal events), cleared by
document.createEvent, and set by initEvent, initCustomEvent,
initUIEvent, initMouseEvent, initKeyboardEvent, initCompositionEvent
and initTextEvent. EventTarget.dispatchEvent rejects uninitialized
events.
document.createEvent also gains the spec-required aliases it was
missing: DragEvent, HashChangeEvent and SVGEvents, plus
BeforeUnloadEvent, DeviceMotionEvent, DeviceOrientationEvent and
StorageEvent mapped to a plain Event (logged as not_implemented) until
those interfaces exist.
Coverage: /dom/events/EventTarget-dispatchEvent.html 5/25 -> 25/25. No
regressions across /dom/events.
Co-Authored-By: Claude Fable 5 <noreply@anthropic.com>
Fixes the failing test in WPT /dom/events/handler-count.html?window and
?document (1/2 -> 2/2 each): setting `window.onclick` or
`document.onclick` did nothing, because neither interface exposed the
accessor — the assignment created an inert expando and the handler
never fired for clicks bubbling up from the page.
Both now store the handler in the frame's attribute-listener map (the
same mechanism as element and ShadowRoot property handlers), which the
dispatch propagation path already consults for any event target.
Non-callable values clear the handler per [LegacyTreatNonObjectAsNull].
Coverage: /dom/events/handler-count.html?window 1/2 -> 2/2,
/dom/events/handler-count.html?document 1/2 -> 2/2. No regressions
across /dom/events.
Co-Authored-By: Claude Fable 5 <noreply@anthropic.com>
Fixes the 3 failing tests in WPT /dom/events/event-global.html
(5/8 -> 8/8):
- "window.event is undefined if the target is in a shadow tree" (and
the window.onerror variant): per the DOM invoke algorithm, the
window's current event is only set while invoking listeners whose
invocation target is not in a shadow tree. window.event was set once
for the whole dispatch; it is now set per invocation target (in
dispatchPhase and around inline handler calls), left undefined for
targets whose root is a shadow root.
- "window.event is set to the current event, which is the event passed
to dispatch (2)": script-dispatched events on non-node targets only
ran addEventListener listeners. Property event handlers now fire too:
XMLHttpRequestEventTarget resolves its on* fields by event type, and
the Window's handler fields are consulted both when an event
propagates to the window (getInlineHandler) and when one is
dispatched directly on it.
Coverage: /dom/events/event-global.html 5/8 -> 8/8. No regressions
across /dom/events.
Co-Authored-By: Claude Fable 5 <noreply@anthropic.com>
Collect types with unforgeable upfront, so that detection in the child is only
O(N) over the very small set of types with an unforgeable property.
Remove duplicate comments.
Fixes the 2 remaining failures in WPT
/dom/events/EventListener-handleEvent.html (4/6 -> 6/6): "throws if
`handleEvent` is falsy/truthy and not callable".
Per the Web IDL callback interface invocation steps, when an object
event listener's handleEvent member is not callable, invoking the
listener reports a TypeError to the global (window.onerror / an "error"
event). Lightpanda silently skipped such listeners.
Listener.run's object branch now performs the Get(handleEvent)
explicitly: a throwing getter is reported like a throwing listener, a
non-callable value reports a TypeError, and only a callable one is
invoked (with exceptions reported as before).
Coverage: /dom/events/EventListener-handleEvent.html 4/6 -> 6/6. No
regressions across /dom/events.
Co-Authored-By: Claude Fable 5 <noreply@anthropic.com>
Fixes the 3 remaining failures in WPT
/dom/events/Event-subclasses-constructors.html (46/49 -> 49/49): the
"argument with non-default values" cases for MouseEvent, WheelEvent and
KeyboardEvent.
- MouseEvent.button is a WebIDL short: any value (e.g. 40) is allowed.
It was stored as the MouseButton enum and construction threw a
TypeError for values outside the named buttons. The field is now a
raw i16 (the enum remains for the named constants); same fix in
WheelEvent, DragEvent and PointerEvent, and in initMouseEvent.
- KeyboardEvent.location is an unsigned long with the same problem
(Location enum); now stored as a raw u32.
- KeyboardEvent supports the legacy KeyboardEventInit members charCode
and keyCode: synthetic events report the initialized values (still 0
by default), while trusted events keep deriving them from the key.
UIEvent.which now returns keyCode for keyboard events, and clamps
negative mouse buttons instead of overflowing.
Coverage: /dom/events/Event-subclasses-constructors.html 46/49 ->
49/49. No regressions across /dom/events.
Co-Authored-By: Claude Fable 5 <noreply@anthropic.com>
Fixes 4 failing tests in WPT /dom/events/Event-init-while-dispatching.html
(1/5 -> 5/5), and as a side effect most of
Event-subclasses-constructors.html (16/49 -> 46/49). Three distinct
bugs, all about synthetic (constructor-created) events not following
the spec's dictionary defaults:
- UIEvent.view defaulted to the window: the constructor stored
`opts.view orelse frame.window` and the getter fell back to the
window too, so `new UIEvent("x").view` was never null. Per
UIEventInit, view defaults to null. The window fallback is kept for
*trusted* (browser-generated) UI events only, which do target the
window, so internal dispatch call sites stay unchanged.
- initCustomEvent was missing the "short-circuit while dispatching"
guard that initEvent and the other legacy initializers have.
- KeyboardEvent's constructor unconditionally forced bubbles,
cancelable and composed to true (correct only for trusted key events
per UI Events); synthetic KeyboardEvents now follow the EventInit
defaults.
The unit tests asserting the old view-defaults-to-window behavior are
updated to expect null.
Coverage: /dom/events/Event-init-while-dispatching.html 1/5 -> 5/5,
/dom/events/Event-subclasses-constructors.html 16/49 -> 46/49. No
regressions across /dom/events.
Co-Authored-By: Claude Fable 5 <noreply@anthropic.com>
Fixes the failing test in WPT /dom/events/Event-isTrusted.any.html (and
its .worker variant): Object.getOwnPropertyDescriptor(event,
"isTrusted") returned undefined because the accessor lived on
Event.prototype. Per Web IDL, isTrusted is [LegacyUnforgeable]: an own,
non-configurable accessor property of every Event instance, with the
same getter function shared between instances.
- bridge.accessor gains an `unforgeable` option. Such accessors are
installed on the class's v8 instance template instead of the
prototype template, so every instance gets an own accessor backed by
one getter function per context.
- Since v8's Inherit only chains prototype templates, attachClass now
walks the ancestor chain (via the _proto fields) and re-installs
ancestors' unforgeable accessors on the subclass's instance template,
so MouseEvent instances & co. get their own isTrusted as well. The
accessor installation code is extracted into attachAccessorProperty
for reuse.
- Event.isTrusted is declared unforgeable and non-deletable.
Coverage: /dom/events/Event-isTrusted.any.html 0/1 -> 1/1,
/dom/events/Event-isTrusted.any.worker.html 0/1 -> 1/1. No regressions
across /dom/events.
Co-Authored-By: Claude Fable 5 <noreply@anthropic.com>
Fixes both failing tests in WPT /dom/events/Event-dispatch-throwing.html
("Throwing in event listener with a single/multiple listeners"): an
exception thrown by an event listener must be reported to the global —
firing window.onerror and an "error" event — while the dispatch
continues with the remaining listeners. Lightpanda only logged the
exception internally, so window.onerror never fired.
- js.Function gains callWithThisRethrow, built on the existing
callRethrow machinery (_tryCallWithThis with .rethrow = true): a
thrown JS exception is surfaced as error.TryCatchRethrow past the
internal TryCatch so an enclosing TryCatch of the caller can observe
the exception value. js.TryCatch gains exceptionValue() returning
the raw caught value.
- Listener.run wraps function and handleEvent callbacks in a TryCatch
and routes the caught exception value to Window.reportError (the
existing "report the exception" implementation, which invokes
onerror with the 5-argument form and dispatches the error event).
This merges with the pre-existing _listeners_did_throw tracking on
Event (used by IndexedDB): the reporting paths also set the flag.
Worker globals are left as-is for now.
- Window.reportError now implements the spec's "in error reporting
mode" guard: an exception thrown while reporting one (e.g. by a
throwing "error" listener) is not reported again, which would recurse
without bound.
Coverage: /dom/events/Event-dispatch-throwing.html 0/2 -> 2/2. No
regressions across /dom/events (verified against a baseline build).
Co-Authored-By: Claude Fable 5 <noreply@anthropic.com>
Fixes 1 failing test in WPT /dom/events/Event-dispatch-redispatch.html:
"Redispatching mouseup event whose default action dispatches a click
event".
The WPT testdriver-vendor shim implemented test_driver.click() with
HTMLElement.click(), which dispatches a single untrusted click event
and no press/release, so the test saw no trusted mouseup at all and an
untrusted click. window.webdriver (the -Dwpt_extensions WebAPI) now
exposes click(element): a full trusted primary-button sequence
(pointerdown, mousedown, pointerup, mouseup, click), dispatched
synchronously so the events are observable when the testdriver promise
resolves. The shim routes test_driver.click through it, like
action_sequence already routes through webdriver.actionSequence.
Note this test (like everything testdriver-based) needs the browser
built with -Dwpt_extensions=true, otherwise window.webdriver is
undefined.
Coverage: /dom/events/Event-dispatch-redispatch.html 3/4 -> 4/4 (with a
wpt_extensions build).
Co-Authored-By: Claude Fable 5 <noreply@anthropic.com>
Fixes the failing test in WPT /dom/events/Event-dispatch-handlers-
changed.html ("Dispatch additional events inside an event listener").
At the target, listeners were run in a single pass mixing capture and
bubble listeners, with one snapshot of the listener list. Per the DOM
dispatch algorithm, the target participates in both the capturing and
the bubbling iterations of the event path, and each invoke clones the
listener list separately: a bubble listener added by one of the
target's capture listeners must run during the bubbling pass. The
single pass used one snapshot, so such a listener never ran.
The at-target step now dispatches twice — capture listeners first, then
non-capture listeners — re-fetching the listener list (and taking a new
sentinel snapshot) for the second pass, and honoring stopPropagation /
stopImmediatePropagation between the two.
Coverage: /dom/events/Event-dispatch-handlers-changed.html 0/1 -> 1/1,
/dom/events/EventTarget-dispatchEvent.html 4/25 -> 5/25. No regressions
across /dom/events.
Co-Authored-By: Claude Fable 5 <noreply@anthropic.com>
Fixes 1 failing test each in WPT /dom/events/Event-dispatch-bubbles-
false.html and Event-dispatch-bubbles-true.html ("In
window.document.cloneNode(true)"): document.cloneNode() threw
NotSupportedError because the .document case of Node.cloneNode was
unimplemented.
Per the DOM spec cloning steps for Document, a new document of the same
type (HTML vs XML) is created with the same URL, and a deep clone
copies the document's children (doctype and document element subtree)
into it, reusing the existing per-child cloning path.
Coverage: /dom/events/Event-dispatch-bubbles-false.html 4/5 -> 5/5,
/dom/events/Event-dispatch-bubbles-true.html 4/5 -> 5/5.
Co-Authored-By: Claude Fable 5 <noreply@anthropic.com>
Introduces an incremental markdown renderer (`md_term.Stream`) to
format streamed assistant responses in real time. Buffers lines,
maintains code fence state across chunks, and temporarily withholds
tables to render them with proper column alignment once complete.
Introduces a stack-threaded `Style` chain to track and re-apply
parent styles upon nested resets. This fixes styling loss in
deeply nested markdown spans.
- Use a 256-ramp gray for `ansi.dim` to improve readability.
- Track and re-apply active styles in nested markdown spans.
- Collapse extra spaces after list markers.
CDP clients (Playwright/Puppeteer) never saw same-document navigations —
history.pushState / replaceState changed frame.url but fired no CDP event,
so frame.url() went stale and waitForNavigation missed SPA route changes.
Add a frame_navigated_within_document notification, dispatched from
History.pushState/replaceState, and emit Page.navigatedWithinDocument from
the Page domain with frameId, url, and navigationType. Unlike frameNavigated,
this deliberately does NOT send Runtime.executionContextsCleared or
DOM.documentUpdated — the document and its execution context are unchanged,
and clearing them would invalidate the client's live context ids and break
frame.evaluate(). Adds a cdp.frame test.
Scoped to the History-API path; fragment and Navigation-API paths are left
for follow-ups.
Same review pattern as the getNames enumerators: the bridge can inject
*Frame directly, so the manual switch on exec.js.global is unnecessary.
getIndexes was the one enumerator the earlier suggestions missed.
Co-Authored-By: Karl Seguin <karlseguin@users.noreply.github.com>
Review follow-up on #2944: the injected `frame` parameter of a bridged
function is the frame of the executing JS context - the caller's realm -
not the frame owning the element. The window-reflecting body/frameset
event handlers (onblur, onerror, onfocus, onload, onresize, onscroll)
used it directly, so a same-origin script reaching into another frame,
e.g. the parent doing
iframe.contentDocument.body.onblur = cb;
stored the handler on the parent's window instead of the iframe's. Per
the HTML spec these handlers alias the Window of the element's node
document, whichever realm performs the access.
All Body/FrameSet getters and setters now resolve that Window through
the existing Node.ownerFrame (element -> ownerDocument -> frame, falling
back to the caller's frame for detached/frameless documents), via a
shared reflectedWindow helper. The Build hooks (parse-complete,
attributeChange, attributeRemove) do the same, and additionally pass the
owner frame to setWindowReflectingHandlerFromAttribute so attribute
handler bodies are compiled in the element's document's realm.
Unit test added: tests/frames/window_reflecting_handlers.html covers the
IDL property path and the content-attribute path (parse-time and
runtime) across realms, asserting the handler lands on the iframe's
window while the parent's window stays untouched. It fails without this
fix.
Co-Authored-By: Karl Seguin <karlseguin@users.noreply.github.com>
Fixes 1 failing test in WPT /dom/events/Event-constructors.any.html
(and its .worker variant): "Event constructors 2", which does
`new Event({ toString: function() { throw test_error; } })` and expects
the exact test_error object to propagate.
When a JS value's conversion runs user code that throws (here, ToString
invoking a throwing toString method), the conversion fails with
error.JsException while the user's exception is pending in the isolate.
The argument-conversion catch folded that into error.InvalidArgument,
and handleError then threw a brand new "TypeError: invalid argument"
over the pending exception, so the script observed the wrong error.
error.JsException is now passed through the argument-conversion catch,
and handleError leaves the isolate untouched for it (like the existing
TryCatchRethrow case), letting the original exception surface.
Coverage: /dom/events/Event-constructors.any.html 13/14 -> 14/14,
/dom/events/Event-constructors.any.worker.html 13/14 -> 14/14.
Co-Authored-By: Claude Fable 5 <noreply@anthropic.com>
Fixes 24 failing tests in WPT /dom/events/Body-FrameSet-Event-
Handlers.html (24/48 -> 48/48).
Per the HTML spec, the onblur, onerror, onfocus, onload, onresize and
onscroll event handlers of body and frameset elements are aliases for
the Window's handlers (the "window-reflecting body element event
handler set"), and setting the corresponding content attribute on those
elements must activate the Window's handler too. Two groups of
failures:
- "Set ...": assigning a non-callable (e.g. "") to the IDL attribute
threw "invalid argument" instead of storing null, because the setters
took ?js.Function.Global. Per [LegacyTreatNonObjectAsNull], the
Body/FrameSet setters now use Window.FunctionSetter (function or
anything-else-means-null), like the existing Window handler setters.
- "Forward ... to Window": only body.onload was forwarded. Window now
stores onblur/onfocus/onresize/onscroll handlers too (and exposes the
matching accessors, which it lacked entirely), and Body/FrameSet
implement getters/setters forwarding all six handlers to the Window
of the current frame. Setting or removing one of the six content
attributes (at parse time via Build.complete or at runtime via the
attributeChange/attributeRemove build hooks) compiles the handler
body and stores it on the Window, so window.onX === element.onX.
Coverage: /dom/events/Body-FrameSet-Event-Handlers.html 24/48 -> 48/48.
Co-Authored-By: Claude Fable 5 <noreply@anthropic.com>
Fixes 2 failing tests in WPT /dom/events/AddEventListenerOptions-
signal.any.html (and its .worker variant): "Passing null as the signal
should throw" (with a valid and with a null listener).
The AddEventListenerOptions dictionary declares `signal` as a
non-nullable AbortSignal, so per Web IDL an explicit null (or any
non-AbortSignal value) must be rejected with a TypeError during the
dictionary conversion — even when the listener argument is null. The
bridge collapsed both a missing member and an explicit null into a Zig
null, silently ignoring the signal.
EventTarget.addEventListener now receives the signal member as a raw
js.Value: an absent or undefined member means "no signal", anything
else is converted to an AbortSignal (throwing a TypeError when it
isn't one), and the conversion runs before the null-callback early
return to match the spec's argument-conversion ordering.
Coverage: /dom/events/AddEventListenerOptions-signal.any.html
9/11 -> 11/11, /dom/events/AddEventListenerOptions-signal.any.worker.html
9/11 -> 11/11.
Co-Authored-By: Claude Fable 5 <noreply@anthropic.com>
The applied review suggestions switched the getNames/getIndexes
enumerator callbacks of HTMLCollection, NamedNodeMap and DOMStringMap
from *const js.Execution to the injected *Frame, but only partially:
signatures and bodies disagreed and the build broke. Complete the
conversion — the bridge resolves *Frame from the context, so the manual
exec.js.global switch goes away, and the arenas/JS locals are reached
through the frame.
Co-Authored-By: Karl Seguin <karlseguin@users.noreply.github.com>
The Caller.deleteIndex and Caller.deleteNamedIndex helpers are invoked
by the bridge for definer interceptors too, since definers share the
deleter's (self, index/name) -> bool calling shape. Rename them (and
their private counterparts) to deleteOrDefineIndex and
deleteOrDefineNamedIndex so the name reflects both uses.
No behavior change.
Co-Authored-By: Karl Seguin <karlseguin@users.noreply.github.com>
