:nth-child() (and its siblings) support complex expressions, e.g.
`:nth-child(2n of :is(.foo, .bar))`. Previously we'd choke because of
the `:is()`. Now, we reuse the `_parenSuffix` subexpression to match
nested parentheses the same way we do for :host() and :host-context().
Note that we only support 3 levels of nesting, so a selector like
`:nth-child(n of :is(:has(:not(.foo))))` will still break.
I'll say yet again that we really should add a proper parser so we stop
getting bug reports like this :)
Fixes#64913
I took a quick look at my recent changes to see if I had inadvertently
fixed this bug, but I couldn't seem to reproduce it even before my
changes. Seems like it's working, though.
Closes#58436
This change adds support for commas in :host() arguments (e.g.
`:host(:not(.foo, .bar))` as well as in nested parens when the argument
is applied without parens (e.g. `:host:not(:has(.foo, .bar))`).
Previously these selectors would receive an extra `[nghost]` attr, e.g.
`[nghost]:not(.foo, [nghost].bar)`.
I didn't file a bug for this one, but it's also blocking on an internal
LSC. Like the other CSS changes, I'll run a TGP to confirm this isn't
breaking.
Previously we supported one level of nested parentheses inside of a
`:host()` selector, e.g. `:host(:not(p))`. This caused a breakage in g3
when I migrated a selector from `:host:not(:has(p))` to
`:host(:not(:has(p)))`. This change adds support for just one more level
of nesting.
It'd be nice to move everything to a real CSS parser (or even update it
to count parentheses like I did with :host-context()), but I wasn't able
to get that to work in ~20 minutes and I'm focusing on other things at
the moment.
This change punts the problem until somebody tries to use just one more
level of nesting in a selector.
Fixes#64830
Previously we supported one level of nested pseudo-element selectors
inside :host-context(), e.g. :host-context(:is(.foo, .bar)). This was
based on a regex-based approach. We could support deeper levels of
nesting by updating the regex, but using a regex approach prohibits us
from supporting arbitrary nesting.
Rather than just adding one more level to the existing expression, I've
added a new generator function which splits selectors on commas in a
parenthesis-aware way. This allows us to support arbitrary nesting.
It's likely we'll want to reuse this in other places where we're not as
careful today. We'll probably do this on a request-based basis, though.
Fixes#59176
I've updated the test to assert what I believe it was trying to assert
before. Without this change, the CSS is invalid so it's unclear what
behavior we're demonstrating.
PR Close#64036
Fixes two issues that were preventing template literals from being recovered properly if one of the interpolated expressions is broken:
1. We weren't updating the expected brace counter when an interpolation starts which in turn was throwing off the recovery logic in `skip`.
2. When producing tokens for template literals, we were treating the closing brace as an operator whereas other places treat it as a character. Even after fixing the first issue, this was preventing the recovery logic from working correctly.
Fixes#63940.
PR Close#64150
The new animations was not correctly looking for the `.` when parsing bindings. This resulted in arbitrary event bindings creating animate.leave instruction calls.
fixes: #63466
PR Close#63470
Similar fix as #63082, but for template attributes. The root cause is the same where we should be using `fullStart` instead of `start` in order to account for whitespaces being skipped.
Fixes#63157.
PR Close#63175
When parsing expressions inside a bound attribute, we offset all of its spans by an `absoluteOffset` in order to get the right spans in the source file. The offset was incorrect when parsing an attribute with leading spaces during the construction of the Ivy AST, because of the combination of:
1. We were setting the offset by looking at `valueSpan.start`.
2. The Ivy parser sets `leadingTriviaChars: [' ', '\n']` which means that spaces and new lines will be ignored in the `sourceSpan.start`.
These changes resolve the issue by using `valueSpan.fullStart` which includes the leading spaces.
Fixes#63069.
PR Close#63082
Currently the HTML parser will stop parsing as soon as it hits an end character in the name of an attribute (e.g. `/` or `>`). This ends up being problematic with some third-party packages like Tailwind which uses a wider range of characters for its class names. While the characters are fine when inside the `class` attribute, our current parser behavior prevents users from setting those classes conditionally through `[class.]` bindings.
These changes adjust the parser to handle such cases.
Fixes#61671.
PR Close#62742
When we introduced blocks, we made a deliberate decision to treat the `@` character as a reserved character in case we need to use it for other syntax in the future. This meant that some common cases, like writing out an email address in the template, can be broken.
After some recent discussions we decided to relax the requirement and only treat `@` as a reserve character if it's followed by a character sequence that matches a known block.
PR Close#62644
There were 26 duplicated block tests in `lexer_spec.ts`, likely due to merge conflicts. These changes remove the duplicates while keeping the 6 tests that were different.
PR Close#62644
Currently when there's a parser error in interpolated text, the compiler reports an error on the entire text node. This can be really noisy in long strings.
These changes switch to reporting the errors on the specific expressions that caused them.
PR Close#62258
Currently we have a `ParserError` that is used for the expression parser and a `ParseError` that is used everywhere else. These changes consolidate them into the `ParseError` to avoid confusion and make it easier to add more context in the future.
PR Close#62160
Currently our expression parser produces two different expressions for writes: `PropertyWrite` (e.g. `foo.bar = 123`) or `KeyedWrite` (e.g. `foo[0] = 123`). This is inconsistent with other ASTs, like TypeScript's, where writes are represented as binary expressions with a `=` operator and it makes it difficult to implement more write operators like `??=`, because we'd essentially have to duplicate them.
These changes switch the expression parser over to produce binary expressions instead.
PR Close#61682
When the expression parser consumes tokens inside a parenthesized expression, it looks for valid tokens until it hits and invalid one or a closing paren. If it finds an invalid token, it reports and error and tries to recover until it finds a closing paren. The problem is that in such cases, it would produce the `ParenthesizedExpression` and continue parsing **from** from the closing paren which would then produce more errors that add noise to the output and result in an incorrect representation of the user's code. E.g. `foo((event.target as HTMLElement).value)` would be recovered to `foo((event.target)).value` instead of `foo((event.target).value)`.
These changes resolve the issue by skipping over the closing paren at the recovery point.
Fixes#61792.
PR Close#61815
Currently we reuse the same binding parser for all expressions in the template. Under the hood, the parser has a single `errors` array that it passes into all ASTs which means that if there's one binding with an error, those errors will be propagated to all other ASTs in the template.
These changes switch to having a unique `errors` array for each AST so we only report errors once.
Relates to #61792.
PR Close#61793
