tree 09dcb9af52c43071ae229884441088a3067cfc31
parent f44e82893ffac9ff549fb44e09d1a9f0ccf9341d
author Anders Hartvoll Ruud <andruud@chromium.org> 1604333971 -0800
committer Blink WPT Bot <blink-w3c-test-autoroller@chromium.org> 1604337471 -0800

Support complex selectors in :not()

Until now we've only supported a single compound selector as an
argument to :not(). This CL adds support for complex selectors,
based on the recent work on :is() and :where().

Parsing
-------

Note that unlike :is() and :where(), :not() does not use
the forgiving selector list behavior, but instead fails parsing the
normal way. CSSSelectorParser now has both ConsumeNestedSelectorList
and ConsumeForgivingNestedSelectorList for the two behaviors.

Our ShadowDOMv0-disabling logic should apply to complex :not() as well,
but for backwards compatibility, we need to allow it whenever :not()
_doesn't_ have a complex argument. The TreatAsNonComplexArgumentToNot()
was added to detect this, which replaces the previous function
CSSParserSelector::IsSimple. (When v0 is removed, we can remove this
mess as well).

Invalidation
------------

There was a bug in ExtractInvalidationSetFeaturesFromSelectorList:
we're supposed to ignore any extracted features inside a :not().
For example, for ".a :not(.b)", we should not add ".b" to the
invalidation set for ".a". In ExtractInvalidationSetFeaturesFrom-
SelectorList there was code that attempted to block the inner features
from "crossing" the :not()-boundary, but it incorrectly used the
*outer* pseudo-type. In other words, when considering ":not(.b)",
the pseudo-type would be kUnknown, since we are not (yet) inside the
:not(). Instead what we need to do is check the pseudo-type of
of the currently held simple selector.

Now that we "block" inner features correctly at the :not() boundary,
the check in ExtractInvalidationSetFeaturesFromCompound is not needed
anymore. This in turns means we can remove several PseudoType params,
which is nice.

Matching
--------

Now traverses a CSSSelectorList and uses MatchSelector. Beyond that
it just works.

Bug: 1139865
Change-Id: Iea5513445e54303ba0685cffc3b52dabc115b7f1
I2S: https://groups.google.com/a/chromium.org/g/blink-dev/c/0alTkXvDCL8/m/-ClOGvOJBAAJ
Reviewed-on: https://chromium-review.googlesource.com/c/chromium/src/+/2507549
Reviewed-by: Rune Lillesveen <futhark@chromium.org>
Commit-Queue: Anders Hartvoll Ruud <andruud@chromium.org>
Cr-Commit-Position: refs/heads/master@{#823161}