Per the CSS Namespaces spec, an empty string declared in an
@namespace rule represents no namespace.
Fixes WPT:
- css/css-namespaces/prefix-002.xml
- css/css-namespaces/prefix-003.xml
This updates the `parse_text_decoration_line_value` to reject values
which non-exclusively include `none` e.g. `underline none`.
It also simplifies handling by always producing a Vector (except for
`none`) and adding VERIFY_NOT_REACHED in more places which shouldn't be
reachable.
If a CSS rule has a URL like `url("")`, it would resolve to the document
URL itself. There isn't a context in which this would result in a valid
resource, so the spec tells us to drop it.
No test here because there isn't really a way to know when a CSS URL is
fetched. We could use PerformanceResourceTiming, but that is only for
HTTP(S) URLs. There is already an existing test for serialization of
empty URLs, which still passes.
The exception returned here is never used or logged. Let's just return
a null request, to make it clearer that these are not exceptions shown
to the user (thus not observable).
We always want to return the mapped scaling mode here, since the bitmap
might end up transformed, scaled, etc. and we do not want to fall back
to nearest neighbour in those cases.
This changes Gfx::ScalingMode to reflect the three modes of scaling we
support using Skia, which makes it a bit easier to reason about the mode
to select. New is ::BilinearMipmap, which uses linear interpolation
between mipmap levels to produce higher quality downscaled images.
The cubic resampling options Mitchell and its sibling CatmullRom both
produced weird artifacts or resulted in a worse quality than
BilinearMipmap when downscaling. We might not have been using these
correctly, but the new ::BilinearMipmap method seems to mirror what
Chrome uses for downscaled images.
`play_or_cancel_animations_after_display_property_change` is called
whenever an element is inserted or removed, or it's display property
changes, but it is only required to run if we actually have animations
to play or cancel.
Reduces time spent in the aforementioned function from ~2% to ~0.03%
when loading https://en.wikipedia.org/wiki/2023_in_American_television
We don't need to iterate every property in start_needed_transitions,
only those that appear in transition-property or have an existing
transition
Reduces the time spent in start_needed_transitions from ~5% to ~0.03%
when loading https://en.wikipedia.org/wiki/2023_in_American_television
By doing this in computed properties rather than InlineLevelIterator we
only do it once per element rather than once per text fragment.
Reduces runtime of this process from ~15% to ~0.2% when loading
https://en.wikipedia.org/wiki/2023_in_American_television
We simplify these at style computation time so there is no need to
maintain them as {Number|Length}OrCalculated
Reduces the time spent in `Length::ResolutionContext::for_layout_node`
from 3.3% to 0.4% when loading
https://en.wikipedia.org/wiki/2023_in_American_television
Prevents observably calling Trusted Types, which can run arbitrary JS,
cause crashes due to use of MUST and allow arbitrary JS to modify
internal elements.
The majority of time in `compute_font()` was spent in
`font_matching_algorithm()` repeatedly computing the same values. We
now cache these values to avoid unnecessary work.
Makes the `compute_font_variation_settings` function match other
`compute_font_*` by absolutizing values passed into them. It resolves a
crash related to using `siblings-count` inside `font-variation-settings`
Co-authored-by: Sam Atkins <sam@ladybird.org>
Integrates the new `FontVariationSettings` from LibGfx into LibWeb to
enable initial variable font functionality. Currently, only the `wght`
(weight) axis is fully supported and tested. This update also introduces
support for the CSS `font-variation-settings` property.
This algorithm is used by Typed-OM for producing a list of internal
style values from a single one, for properties that take a list. We
will probably need to implement this for more StyleValues later.
Now that we don't remove whitespace when parsing, we don't need to
artificially insert it back in again when serializing. We do now need
to trim leading and trailing whitespace from UnresolvedStyleValues, as
this previously was done as part of the whitespace insertion.
This makes our serialization of UnresolvedStyleValues more correct and
gets us a few WPT passes for each property in the Typed OM tests.
Now that we don't strip out whitespace, this method was copying the
input TokenStream into a Vector, and then creating new TokenStreams
from that. So, stop doing that and use the input TokenStream instead.
