Fix false positives for parens in selectors in indentation

[fregante]

Initially reported in #3461 (comment)

Clearly describe the bug

indentation does not cover the indentation of the content of parens in selectors, like :is()

What code is needed to reproduce the bug?

:is(
a,
b
) {}

What stylelint configuration is needed to reproduce the bug?

{
  "rules": {
    "indentation": 2
  }
}

Which version of stylelint are you using?

13.11.0

What did you expect to happen?

a and b should be indented:

:is(
  a,
  b
) {}

What actually happened (e.g. what warnings or errors did you get)?

They're accepted as is, without indentation.

Context

Originally posted by @jeddy3 in #3461 (comment)

We apply indentation to parens in at-rules and declaration-values, but not those in selectors. It should, though.

[jeddy3]

@fregante Thanks for the report and for using the template.

I can reproduce using the demo with:

:is(
 a,
 b
) {
  width:
    calc(
      100% - 10px
    );
}

I've labelled the issue as ready to implement. Please consider contributing if you have time.

There are steps on how to fix a bug in a rule in the Developer guide.

[fregante]

Interestingly enough, :is(:is()) is partially indented:

:is(
:is(
  a,
  b
),
b
) {
  width:
    calc(
      100% - 10px
    );
}

Whereas :is(:is(:is())) just fails between selector-pseudo-class-parentheses-space-inside and indentation

:is(
:is(
  :is(
    a,
    b
  ) /* Unexpected whitespace before ")" */
)
) {
  width:
    calc(
      100% - 10px
    );
}

[jeddy3]

Thanks for digging deeper.

Whereas :is(:is(:is())) just fails between selector-pseudo-class-parentheses-space-inside and indentation

It looks like we'll need to bring the same flexibility we have for functions to pseudo-classes, i.e. add a:

• never-single-line option to selector-pseudo-class-parentheses-space-inside, like function
• selector-pseudo-class-parentheses-newline-inside rule, like function

If anyone fancies tackling these, please create new issues for each so we can track them.

---

Incidentally, it's amazing to see how much :is() cleaned things up in that referenced pull request.

It makes me think there's scope for a new use-matches-any-pseudo-class rule (spec ref) that, that would work like the stylelint-use-nesting plugin but would enforce the use of :is() where appropriate.

[jeddy3]

Closing as stylistic rules are frozen. The community is welcome to migrate the rule to a plugin and fix this bug.