Our BrowsingContext currently supports 1 target. So we have a per-BC target_id.
Previously, our target had 1 "frame" - our page. So we often treated the
targetId as the frameId. But to work with frames, we need page-specific
frameIds and loaderIds.
This tries to clean up our ids (a little). frameIds are now ids derived from
a new incrementing page.id. This page.id has to be passed around (via http
Requests and through notifications) in order to properly generate messages with
a frameId.
Added a get-only `getStyle` which doesn't lazily create a new style if none
exists. This can be used in the (frequently used) `checkVisibility` to avoid
an allocation. Added a specialized getBoundingClientRectForVisible which
skips the checkVisibility check, since a few callers have already done their
own visibility check.
DOMRect is now off the heap. This avoids _a lot_ of allocation when a DOMRect
is only needed for internal calculation, e.g. in Document.elementFromPoint.
implement proper tree-walking in writeTextContent to handle all cases:
same-element containers, cross-container ranges, and comment exclusion.
uncomment ~800 lines of Range tests and add 5 new toString tests.
handleClick() was setting _active_element directly, bypassing
Element.focus() — so no blur/focus events fired on click.
Now calls focus() for input, button, select, textarea, anchor.
focus() and blur() now dispatch all four spec-required FocusEvents:
blur (no bubble) → focusout (bubbles) → focus (no bubble) → focusin (bubbles)
Each event carries the correct relatedTarget: the element gaining focus
for blur/focusout, and the element losing focus for focus/focusin.
All four events are composed per W3C spec.
Relates to #1161
interactive elements (input, button, a, select, textarea, iframe)
default to 0 when tabindex attribute is absent; others default to -1.
also add TODO for hidden "until-found" tristate.
Tested in Chrome (headless, --autoplay-policy=no-user-gesture-required):
playing fires on every pause-to-play transition, not just the first
time. The _playing flag was incorrectly suppressing this. Removed it
and updated tests to match verified Chrome behavior.
Per spec, valid type values are absent, empty string, or a
case-insensitive match for "text/css". Also clear cached sheet
when the element is disconnected or type becomes invalid.
Per spec, complete returns true when the image has no src, is fully
fetched, or is broken. Since this is a headless browser that does
not fetch images, complete always returns true.
Per spec, connected style elements with type text/css should
return a CSSStyleSheet object. Previously always returned null.
The sheet is lazily created and cached on first access.
html5ever generally makes guarantees about nodes being parentless when
appending, but we've already seen 1 case where appendCallback receives a
connected node.
We're now seeing something in appendBeforeSiblingCallback, but we have a clearer
picture of how this is happening. In this case, it's via custom element
upgrading and the custom element constructor has already placed the node in
the document.
It's worth pointing, html5ever just has an opaque reference to our node. While
it guarantees that it will give us parent-less nodes, it doesn't actually know
anything about our nodes, or our node._parent. The guarantee is only from its
own point of view. There's nothing stopping us from giving a node a default
parent as soon as html5ever asks us to create a new node, in which case, the
node _will_ have a parent.
