`@scope (a) to (b) {}` applies its contained style rules to elements
that have `a` as a parent, and do not have `a b` as a parent. Both the
`a` and `b` selector lists are optional.
Because it's situational whether a `@scope` will apply to a given
element, we store the ancestor scope on the `MatchingRule`, similar to
`@container`, and then determine during matching whether all the parent
`@scope`s match or not.
The rules for how selectors inside `@scope` are adjusted and interpreted
are a bit confusing. Unlike for other at-rules, nested style rules
inside `@scope` do not get a leading `&` added during parsing. To
support this, `adapt_nested_relative_selector_list()` now takes a flag
for whether its parent is a `@scope` or not.
`@scope` can also contain nested declarations without itself being
nested inside a style rule.
When determining their selectors, nested declarations rules adopt the
`@scope`'s scoping root if it has one, or otherwise fall back to the
parent element of the `<style>` element (not implemented here,) or the
`:root`. These are required to have zero specificity, so we wrap the
selector in `:where()`.
Matches if we support the at-rule in some form.
Keeping this list up to date is a bit awkward, but we don't add at-rules
too often, and having all at-rules defined in JSON would just move the
awkward-to-maintain list somewhere else.
Similar to GC::Root<T>, make GC::RootVector<T> constructible without
explicitly passing a Heap.
This is implemented by having RootVectorBase use GC::Heap::the() for
heap-free construction.
e.g., `@container (width >= 300px) {}` and similar.
During style computation, flag any elements whose style depends on a
size container. Then re-evaluate their style after the initial layout
has been computed and size containers have a size. This may take
multiple passes, as these may have further descendants that depend on
their size, etc. We limit this to 8 passes currently.
SizeFeature itself is very similar to MediaFeature, but queries the
container element instead. There are only 6 size features specified, so
they're hard-coded instead of generated from JSON.
Also add a counter test for the narrower restyle path.
Different users of BooleanExpression have different requirements for
evaluation:
- `@media` needs a Document
- `@supports` doesn't need anything
- `@container` needs a container Element
To support these without expanding the API, replace the Document*
parameter with a BooleanExpressionEvaluationContext type which contains
these different values.
No behaviour changes.
DecodedImageFrame only wraps a ref-counted Bitmap and color-space
metadata. The frame object itself does not provide shared mutable
state or lifetime ownership beyond those members, so ref-counting it
adds an unnecessary layer of indirection.
DecodedImageFrame now owns decoded bitmap pixels directly, so the
separate ImmutableBitmap wrapper no longer carries useful semantics.
Remove the class and pass decoded image frames or bitmaps at the
boundaries where pixels are actually required.
The Skia image cache now keys off DecodedImageFrame, matching the
display-list commands that paint decoded images. Video frames stay
owned by LibMedia, with the explicit YUV-to-bitmap conversion living
at HTMLVideoElement's decoded-frame entry point for canvas and WebGL
callers.
This brings a couple of advantages:
- Previously we relied on the caller validating the parsed value was in
bounds after the fact - this was usually fine but there are a couple
of places that it was forgotten (see the tests added in this commit),
requiring the bounds to be passed as arguments makes us consider the
desired range more explicitly.
- In a future commit we will use the passed bounds as the clamping
bounds for computed values, removing the need for the existing
`ValueParsingContext` based method we have at the moment.
- Generating code is easier with this approach
This introduces two new top-level `ValueParsingContext`s,
`OnScreenCanvasContextFontValue` and `CanvasContextGenericValue`, while
these are handled the same for now, there is a distinction is whether or
not they allow tree counting functions (which will come in a later
commit)
The HTML sizes algorithm does not use full media queries. It evaluates
a restricted media-condition grammar and then parses the selected
source size value as a length.
Teach the parser to follow that split more closely: treat sizes
conditions as two-valued booleans, validate MQ5 <general-enclosed>
contents more strictly, accept calc(0) for media feature values, and
reject only source-size math results that are negative or non-finite.
The imported sizes parsing tests then progress from 140/171 to
171/171 in all four cases.
We now ignore all animation properties from `css-animations-1` declared
within keyframes, except `animation-timing-function`, which is treated
specially.
The main limitation here are that none of the container-query features
are parsed in a meaningful way; they all become `<general-enclosed>`.
Parsing for them will be added as they are implemented.
When parsing declarations within a nested grouping rule, we don't store
these directly, but within a "nested declarations rule", in most cases
this is `CSSNestedDeclarations`, but this isn't always the case e.g.
`@function` rules and others (e.g. @media) within them should instead
use `CSSFunctionDeclarations`
Some at-rules (i.e. `@function`) require us to support custom
descriptors (e.g. `--foo`).
We do this by adding `DescriptorID::Custom` and using a new
`DescriptorNameAndID` class in a bunch of places where we previously
just used `DescriptorID`
When parsing block contents, the CSS parser speculatively tries to parse
each item as a declaration first. If that fails, it restores the token
position and tries again as a qualified rule. This means every qualified
rule inside an at-rule block (e.g. @layer, @media) gets parsed twice:
once as a failed declaration (which consumes all tokens via
consume_the_remnants_of_a_bad_declaration), and then again successfully
as a rule.
Add a lookahead that checks for the `ident whitespace* ':'` pattern
before attempting declaration parsing. Since declarations must start
with this pattern per spec, we can skip the attempt entirely when it
doesn't match and go straight to qualified rule parsing.
This is a massive win on large Tailwind CSS stylesheets (like the one
used by chatgpt.com) where thousands of rules inside @layer blocks were
being double-parsed. On a 1.2MB Tailwind v4 stylesheet, parse time goes
from ~2000ms to ~95ms (21x speedup).
Previously, we registered `@property` rules during parsing, and treated
them the same as `CSS.registerProperty()` calls. This is not correct
for a couple of reasons: One, the spec wants us to distinguish between
those two sources of registered custom properties, with
`CSS.registerProperty()` calls taking precedence. Two, we never removed
the registered property when its `@property` was removed from the
document.
This commit deals with this by iterating active CSSPropertyRules to find
which ones currently apply, and storing those in a cache. This cache is
invalidated whenever the Document's style is invalidated, which happens
whenever a CSSRule is added or removed from the Document.
The attached test demonstrates this now working as it should.
...instead of returning the one from its associated style sheet.
This reverts 848a250b29 where I made
`CSSImportRule.media` nullable.
CSSImportRule may not have an associated style sheet, because of not
matching a supports condition, or just failing to load the URL.
Regardless of whether we do or not, the expected (non-spec) behaviour
is that we should return a MediaList always, which matches the media
queries specified on the `@import` rule.
Fixes some WPT tests that expected `supports(foo:bar)` to serialize as
`supports(foo:bar)`, instead of `supports(foo: bar)` with a space
between.
Reading the original_full_text directly also lets us delete
Declaration::to_string(), which was only used here.
The previous implementation assumed that the contents of `supports()`
was either a raw declaration, or a block containing some number of them.
This meant we wouldn't parse things like `supports(not (a:b))` or
`supports(selector(*))`.
`parse_a_supports()` actually does what we want in every case except for
raw declarations (`supports(a: b)`), so let's always call it first, and
then fall back to parsing a single declaration.
This works by generating random values using XorShift128PlusRNG at
compute time and then caching them on the document using the relevant
random-caching-key
In a few places, user code wants to parse a `<color>` or `<length>` etc,
but we didn't have a way to do so, so they would do something
similar-ish instead, like parse the value of the `color` property.
Let's make that available instead.
Some contexts (e.g. descriptors, media conditions) don't allow tree
counting functions, this commit adds an easy way to check if the current
value context is one of those.
Previously we would always use the window's viewport which was incorrect
if we were within an iframe.
This is likely applicable to all uses of
`Length::ResolutionContext::for_window`.
