1. Introduction
This section is non-normative.
The Custom Highlight API extends the concept of highlight pseudo-elements (see CSS Pseudo-Elements 4 § 3 Highlight Pseudo-elements) by providing a way for web developers to style the text of arbitrary Range objects, rather than being limited to the user agent defined ::selection, ::inactive-selection, ::spelling-error, and '::grammar-error'. This is useful in a variety of scenarios, including editing frameworks that wish to implement their own selection, find-on-page over virtualized documents, multiple selection to represent online collaboration, or spellchecking frameworks.
The Custom Highlight API provides a programmatic way of adding and removing highlights that do not affect the underlying DOM structure, but instead applies styles to text based on range objects, accessed via the ::highlight() pseudo element.
One two. It does so by adding a
Highlight
to the HighlightRegistry
(both of these are new concepts introduced by this specification).
The Highlight
will contain a Range
whose boundary points surround the text One two.
< style > : root :: highlight ( example-highlight ) { background-color : yellow ; color : blue ; } </ style > < body >< span > One</ span >< span > two</ span >< span > three…</ span > < script > let r= new Range(); r. setStart( document. body, 0 ); r. setEnd( document. body, 2 ); CSS. highlights. set( "example-highlight" , new Highlight( r)); </ script >
The result would look like:
2. Module Interactions
This module depends on the Infra Standard [INFRA] and on WebIDL [WebIDL].
It assumes general familiarity with CSS and with the DOM Standard [DOM], and specifically extends the mechanisms defined in CSS Pseudo-Elements Module Level 4 [css-pseudo-4] to handle highlight pseudo-elements. The Selectors Level 4 [selectors-4] specification defines how pseudo-elements work in general.
See References for a full list of dependencies.
Note: This draft is an early version. As it matures, the CSS-WG could decide to keep it as an independent module, or might prefer to fold it into [css-pseudo-4], or a later version of that module.
3. Setting up Custom Highlights
3.1. Creating Custom Highlights
A custom highlight is a collection of ranges representing portions of a document. They do not necessarily fit into the element tree, and can arbitrarily cross element boundaries without honoring its nesting structure. They can be used to affect the appearance of these portions of the document (see § 4 Styling Custom Highlights), or to handle to events associated with them (see § 6 Event Handling).
Custom highlights are represented by Highlight
objects, setlike objects whose set entries are AbstractRange
objects. Ranges can be added to a custom highlight either by passing them to its constructor,
or by using the usual API of setlike objects
to manipulate its set entries.
Note: As the ranges in a custom highlight are AbstractRange
objects,
authors can chose between using Range
objects and StaticRange
objects.
See § 5.2 Range Updating and Invalidation for more details about this choice and its implications.
enum {
HighlightType ,
"highlight" ,
"spelling-error" }; [
"grammar-error" Exposed =Window ]interface Highlight {constructor (AbstractRange ...);
initialRanges setlike <AbstractRange >;attribute long ;
priority attribute HighlightType ; };
type
See § 4.2.5 Priority of Overlapping Highlights for more information on the priority
attribute.
See § 4.2.6 Highlight types for more information on the type
attribute.
Highlight(AbstractRange... initialRanges)
constructor is invoked,
run the following steps:
- Let highlight be the new
Highlight
object. - Set highlight’s
priority
to0
. - Set highlight’s
type
tohighlight
. - For each range of
initialRanges
, let rangeArg be the result of converting range to an ECMAScript value, then run the steps for a built-in setlike add function, with highlight as thethis
value, and rangeArg as the argument. - Return highlight.
3.2. Registering Custom Highlights
In order to have any effect, custom highlights need to be registered into the highlight registry.
The highlight registry is accessed via the highlights
attribute of the CSS
namespace,
and represents all the custom highlights registered for the current global object’s associated Document.
It is a maplike, and can be updated using the usual methods.
It’s map entries is initially empty.
A custom highlight is said to be registered if it is in the highlight registry. It stops being registered if it is later removed.
partial namespace CSS {readonly attribute HighlightRegistry ; }; [
highlights Exposed =Window ]interface {
HighlightRegistry maplike <DOMString ,Highlight >; };
set
method of the highlight registry which will run the steps for a built-in maplike set function,
with the context object as the this
value,
the passed-in custom highlight name as keyArg,
and the passed-in highlight as valueArg.
The custom highlight name assigned to a custom highlight when it is registered is used to identify the highlight during styling (see § 4 Styling Custom Highlights).
Note: When registering a custom highlight, authors are advised to use a custom highlight name that is a valid CSS identifier. Using a name that is not a valid identifier can make the highlight hard, and in some cases impossible, to style via CSS.
Note: It is possible to register a custom highlight with more than one custom highlight name. However, using more than one name to style a highlight will assign the highlight multiple different sets of styles, without a way to control the stacking order of conflicting styles within these sets during painting. This could be limiting for authors and could cause confusing painting behavior (see the example below for more context). Therefore, authors are advised to only use one name per highlight during styling.
< style > div :: highlight ( bar ) { color : red ; } div :: highlight ( foo ) { color : green ; } </ style > < body >< div > abc</ div > < script > let div= document. body. firstChild; let r= new Range(); r. setStart( div, 0 ); r. setEnd( div, 1 ); let h= new Highlight( r); CSS. highlights. set( 'foo' , h); CSS. highlights. set( 'bar' , h); </ script >
In the example above,
the same custom highlight object is registered under the names foo
and bar
.
Since each of the style rules target the same highlight and have the same specificity,
authors might expect the last rule to win in cascading order
and the highlighted content to be green.
However, each highlight name gets an independent set of highlight styles,
and the highlight will be painted once per name.
In this case, because foo
was registered before bar
,
the highlight will be first painted with foo
’s color (green)
and then with bar
’s color (red).
As a result, the highlighted content will appear red.
4. Styling Custom Highlights
4.1. The Custom Highlight Pseudo-element: ::highlight()
The ::highlight(<custom-highlight-name>) pseudo-element (also known as the custom highlight pseudo-element) represents the portion of a document that is being contained or partially contained in all the ranges of the registered custom highlight with the custom highlight name <custom-highlight-name>, if any. <custom-highlight-name> must be a valid CSS <ident-token>.
4.2. Processing Model
4.2.1. Applicable Properties
Custom highlight pseudo-elements, like the built-in highlight pseudo-elements, can only be styled with a limited set of properties. See CSS Pseudo-Elements 4 § 3.2 Styling Highlights for the full list.
4.2.2. Default Styles
UAs must not define any styles for custom highlight pseudo-elements in the default UA stylesheet. A custom highlight pseudo-element inherits the styles of its originating element.
4.2.3. Cascading and Inheritance
The cascading and inheritance of custom highlight pseudo-elements is handled identically to that of the built-in highlight pseudo-elements, as defined in CSS Pseudo-Elements 4 § 3.5 Cascading and Per-Element Highlight Styles.
4.2.4. Painting
The painting of custom highlights is also handled identically to that of the built-in highlight pseudo-elements, as specified in CSS Pseudo-Elements 4 § 3.4 Area of a Highlight and CSS Pseudo-Elements 4 § 3.6 Painting the Highlight, with the following clarifications:
- Collapsed ranges are not rendered.
-
Overlapping ranges within a single custom highlight are rendered
as if a single range representing the union of the overlapping ones
had been specified.
The following example renders in a single highlight with semi-transparent blue background, not two overlapping ones which can be seen through each other.
< style > :: highlight ( sample ) { background-color : rgba( 0 , 0 , 255 , 0.3 ); } </ style > < body > Lorem Ipsum.< script > let textNode= document. body. firstChild; let r1= new Range(); r1. setStart( textNode, 1 ); r1. setEnd( textNode, 5 ); let r2= new Range(); r2. setStart( textNode, 3 ); r2. setEnd( textNode, 7 ); CSS. highlights. set( "sample" , new Highlight( r1, r2)); </ script > In other words, this rendering would be correct:
Lorem Ipsum.However, this one would be incorrect:
Lorem Ipsum. - The highlight overlays of the custom highlights are below those of the built-in highlight pseudo-elements in the stacking order described in CSS Pseudo-Elements 4 § 3.6 Painting the Highlight.
- The relative stacking order of the highlight overlays of multiple custom highlights is defined by their priority (see § 4.2.5 Priority of Overlapping Highlights).
4.2.5. Priority of Overlapping Highlights
A custom highlight's priority
attribute
defines its priority.
This is used to determine the stacking order of the corresponding highlight overlay during painting operations (see § 4.2.4 Painting).
A higher priority results in being above in the stacking order.
A custom highlight will have a default numerical priority of 0
if its priority
attribute has not been explicitly set.
When two or more custom highlights have the same numerical priority, the one most recently registered has the higher effective priority.
< style > : root :: highlight ( foo ) { color : blue ; background-color : yellow ; } : root :: highlight ( bar ) { background-color : orange ; } </ style > < body > Some text< script > let textNode= document. body. firstChild; let r1= new Range(); r1. setStart( textNode, 0 ); r1. setEnd( textNode, 6 ); let r2= new Range(); r2. setStart( textNode, 3 ); r2. setEnd( textNode, 9 ); let h1= new Highlight( r1); let h2= new Highlight( r2); CSS. highlights. set( "foo" , h1); CSS. highlights. set( "bar" , h2); </ script >
As there are no priorities set
(i.e. there is a tie between h1
and h2
),
the custom highlights' styles are stacked
in order of insertion into the highlight registry.
The rendered results will have "Som" with blue text on yellow background,
"e t" with blue text on orange background,
and "ext" with the default color on orange background.
Setting h1
would cause h1
to stack higher than h2
,
which would result in "Some t" being blue on yellow,
and "ext" being default color on orange.
4.2.6. Highlight types
A custom highlight's type
attribute is used by authors
to specify the semantic meaning of the highlight.
This allows assistive technologies to include this meaning
when exposing the highlight to users.
A custom highlight will have a default type of highlight
if its type
attribute has not been explicitly set.
Note: Authors are advised to set a custom highlight's type
to spelling-error
when that custom highlight is being used to emphasize misspelled content.
Authors are advised to set a custom highlight's type
to grammar-error
when that custom highlight is being used to emphasize content that is grammatically incorrect.
For all other use cases type
is best left as highlight
.
UAs should make custom highlights available to assistive technologies.
When exposing a highlight using a given platform accessibility API,
UAs should expose the semantic meaning of the highlight
as specified by its type
attribute
with as much specificity as possible for that accessibility API.
Note: For example,
if a platform accessibility API has the capability to express spelling errors and grammar errors specifically,
then UAs is expected to use these capabilities to convey the semantics for highlights
with spelling-error
and spelling-error
.
If an accessibility API only has the capability to express spelling errors,
then UAs would be expected to convey both highlights with spelling-error
and with grammar-error
using spelling error semantics.
If an accessibility API has support for expressing neither spelling errors nor grammar errors,
then UAs would expose all highlights as generic highlight
regardless of their actual type
.
Note: This initial set of types was chosen
because they are expected to be popular use cases for Highlight API
and there is some existing support for expressing their semantics in platform accessibility APIs today.
Accessibility APIs currently don’t have any way to express the specific semantics
of other expected Highlight API use cases.
More types could later be added to HighlightType
as accessibility APIs gain support for expressing additional popular use cases of Highlight API.
5. Responding to Changes
5.1. Repaints
The addition or removal of a custom highlight in the highlight registry, or of a range in a registered custom highlight, must cause the user agent to reevaluate the rendering, and to repaint if appropriate.
The user agent must also repaint highlights as needed
in response to changes by the author
to the priority
,
or to the boundary points of Range
s
of a registered custom highlight.
How should we specify the timing (and synchronicity) of this reevaluation? [Issue #4596]
5.2. Range Updating and Invalidation
Authors can build custom highlights using either Range
s or StaticRange
s.
The resulting custom highlight represents the same parts of the document, and can be styled identically. However, the behavior is different in case the underlying document is modified.
Range
s are live ranges.
The user agent will adjust the boundary points of Range
s
in response to DOM changes overlapping the range or at its boundary,
and repaint accordingly. Boundary points of live ranges can also be changed
by the author.
On the other hand,
the user agent must not adjust the boundary points of StaticRange
s
in response to DOM changes,
nor can they be modified by the author after creation.
The user agent is expected to store the actual StaticRange
s, rather than backing
them up with live Range
s.
Range
objects as the DOM is modified
has a significant performance cost.
Authors who intend to observe DOM changes and react to them
by adjusting or recreating the ranges in their custom highlights are strongly encouraged to user StaticRange
s
in order to avoid this costly but unnecessary step.
Conversely, authors who use StaticRange
s
should observe and react to DOM changes,
by discarding stale ranges or custom highlights and recreating new ones.
When computing how to render a document,
if start node or end node of any range in the highlight registry associated with that document’s window
refer to a Node
whose shadow-including root is not that document,
the user agent must ignore that range.
If any StaticRange
in the highlight registry associated with that document’s window
is not valid,
the user agent must ignore that range.
The interaction of StaticRange
s in a custom highlight and [css-contain-2] seems problematic:
on a fully contained element,
you should expect that DOM changes to descendants of that element
will not cause invalidation and restyling/repainting
of elements outside the contained one.
However, if a static range has a boundary point inside the contained subtree
and another boundary point outside of it,
and the DOM in the contained subtree is changed
so that the boundary point inside no longer points to a valid node,
the whole range should be ignored,
which would affect painting outside the contained subtree.
Is this a weakness of style containment,
or of the invalidation logic above,
or something else? [Issue #4598]
6. Event Handling
Section on Events TBD, based on https://github.com/MicrosoftEdge/MSEdgeExplainers/blob/master/highlight/events-explainer.md
should custom highlights have a dedicated event handling mechanism, or should that be added to pseudo-elements in general?
Appendix A. Privacy and Security Considerations
This section is non-normative.
This specification is not thought to introduce any new security or privacy concern. Anyone suspecting that this is not accurate is encouraged to get in touch with the CSS Working Group or the co-editors.
Appendix B. Acknowledgements
This section is non-normative.
Acknowledge people (other than editors) who deserve credit for this.
Appendix C. Changes
This section is non-normative.
Changes since the 8 December 2020 Working Draft
In addition to various editorial improvements and minor tweaks, the main changes are:
-
Renamed
HighlightsRegister
toHighlightRegistry
-
Removed the redundant
add()
method fromHighlightRegistry
. (See Issue 6092) -
Make custom highlight overlays stack below native highlight overlays. (See Issue 4595)
-
Handle highlight priority with integers rather than floats. (See Issue 4592)
-
Define the default value for highlight priority to be 0. (See Issue 6136)
-
Made HighlightRegistry maplike (rather than setlike) and remove
name
property from Highlight. (See Issue 5910) -
Clarified that ranges from the wrong window are not painted. (See Issue 6417)
-
Specify that custom highlights have no UA styles. (See Issue 6375)
-
Deferred to the [DOM] specification for range invalidation (See Issue 4597)
-
Added a
type
attribute toHighlight
to give clearer semantics to different highlights, in support of exposing highlights to accessibility tools. (See Issue 6498)
Changes since the 22 October 2020 Working Draft
There have been only editorial changes since the 22 October 2020 Working Draft; see diffs.