[cssom] There should be a way to test CSSStyleRules against Elements #10470
Description
To determine whether a given CSSStyleRule applies to a DOM element you can take its selectorText and use the element.matches() DOM API to check whether that selector applies to it. Doing so hasn't been sufficient for a long time:
- It might be in a media query, or in a
<style>or<link>element with a media attribute - It might be in a container query.
- It might be inside a supports at-rule.
- It might be nested, so its selectorText is incomplete on its own and potentially including
&. - It might be inside a scope, either explicitly or implicitly, the selectortext potentially including
:scope.
For media queries there is window.matchMedia(), for container queries there is going to be element.matchContainer(), for @supports there is CSS.supports(). This means you can at least get the information, but you will still need to walk the CSSOM to find conditionTexts to match against. (It seems that media and supports at-rules are getting a .matches property as discussed in #4240. This would be easier.)
For nesting and scoping however, you will need to walk the CSSOM and build up a complete, potentially very complex, selector to test against and, in the case of implicit scoping, also check the DOM and determine a selector to use for the common ancestor. This is far from ideal. For nesting this could potentially be solved with a resolved selectorText like discussed in #10246, but that wouldn't account for scoping.
All situations require you to loop over at least all ancestors in the CSSOM to get a full picture for a single CSSStyleRule, and potentially switching over to the DOM for attributes or common ancestors.
Proposal
The addition of a new API that explicitly tests if a given CSSStyleRule applies to a given Element with an optional pseudo-state (or class).
Applies here means that the content of the CSSStyleRule is used when determining the style of an element, regardless of whether the declarations inside the rule actually end up being used or are overwritten by style with higher specificity.
CSSStyleRule currently does not expose any functions so it makes sense to add a new method to the Element instead, possibly called matchRule(), that takes a CSSStyleRule and an optional second string argument indicating the pseudo state or element to match against, and returns a boolean value.
/* Match against the element itself */
Element.matchRule(CSSStyleRule);
/* Match against the element in hover state */
Element.matchRule(CSSStyleRule, ':hover');
/* Match against the element's first-letter */
Element.matchRule(CSSStyleRule, '::first-letter');Alternatively, when matching for example a p against a CSSStyleRule with the selector div:hover p the matcher could not take that :hover pseudo-state into account and instead return an object that has a .matches boolean property (like window.matchMedia) as well as a conditional property that has an object property that could be shaped { element: Element, state: string }:
/* Match a CSSStyleRule with selectorText 'div:hover p' */
Element.matchRule(CSSStyleRule);
/* {
matches: true,
conditional: {
element: <div>,
state: ':hover',
}
*/In addition it would be very useful if that function returned more information such as the resolved specificity or resolved/matched selectorText (for when the CSSStyleRule contains a selectorlist) and information on the conditionals (media, container, supports layer,) applied to the CSSStyleRule, though this is less important than the matching test itself.
I'm happy to do the work here in terms of spec writing (after consensus), if someone can nudge me in the right direction.
edit 20 june 24: expanded the proposal to include checking for pseudo classes and elements.