fix: support unquoted aria-hidden value in require-alt-text

[KumJungMin]

Disclosure: I'm a participant of open source contribution program OSSCA

Prerequisites checklist

• I have read the contributing guidelines.

AI acknowledgment

• I did not use AI to generate this PR.
• (If the above is not checked) I have reviewed the AI-generated content before submitting.

What is the purpose of this pull request?

This PR updates require-alt-text to recognize an unquoted aria-hidden=true attribute.

HTML permits attribute values to be written without quotes, but the existing implementation handles the following forms differently:

<img src="quoted.png" aria-hidden="true" />
<img src="unquoted.png" aria-hidden=true />

Since both forms set the value of aria-hidden to true, this PR updates the rule to skip the alternative text check in both cases.

What changes did you make? (Give an overview)

• Added a dedicated regular expression that recognizes double-quoted, single-quoted, and unquoted aria-hidden=true attributes.
• Added support for optional whitespace around the equals sign.
• Made matching case-insensitive for both the attribute name and value.
• Added a boundary check to ensure that the alternative text check is skipped only when the value is exactly true.
• Added valid test cases for the following unquoted forms:
  • aria-hidden=true
  • aria-hidden = true
  • ARIA-HIDDEN=TRUE
• Added invalid test cases to verify that aria-hidden=false and aria-hidden=truefoo are still reported.

The regex test cases for this change are available on RegExr.

Related Issues

fixes #718

Is there anything you'd like reviewers to focus on?

To keep the scope of this change minimal, the implementation adds a dedicated regular expression for detecting aria-hidden=true instead of extending getHtmlAttributeRe().

There are two reasons for this approach:

1. It avoids affecting the existing alt attribute parsing behavior and avoids recreating the aria-hidden regular expression each time an image tag is processed.
2. aria-hidden requires checking whether its value is exactly true, while getHtmlAttributeRe("alt") captures the value of the alt attribute to determine whether it consists only of whitespace. I considered these attributes to have different validation purposes and criteria.

I would appreciate feedback on whether this dedicated regular expression is appropriate or whether it would be better to generalize getHtmlAttributeRe() to handle both quoted and unquoted attribute values. :)

[DMartens]

There does not have to be whitespace around an HTML attribute, e.g. <a before="true"aria-hidden="true"after="true"></a> is valid.
In other words the before "aria-hidden" can be a quote or whitespace and alpha characters if the attribute value is quoted.

[KumJungMin]

I’d like to update it as follows to account for the additional cases.

Case	Before	After	Reason
Boundary before aria-hidden	\s	(?:\s|["'])	Allows aria-hidden to be preceded not only by whitespace, but also directly by the closing quote (" or ') of the previous attribute.
Boundary after true	(?:"true"|'true'|true)(?=\s|/?>)	(?:"true"|'true'|true(?=\s|/?>))	Allows another attribute to immediately follow "true" or 'true' without whitespace, while keeping the existing boundary check for unquoted true so that cases like aria-hidden=truefoo are not matched incorrectly.

I’m thinking of updating the final pattern as follows. Is there any case I may have misunderstood? (https://regexr.com/8o87r)

/(?:\s|["'])aria-hidden\s*=\s*(?:"true"|'true'|true(?=\s|\/?>))/iu

[DMartens]

Yes the new pattern works for the mentioned cases.
Can you please add test cases for this and replace > with > (no need to escape it in this case) in the pattern?

[KumJungMin]

I updated the regular expression to address the regexp/prefer-character-class ESLint error.

The previous pattern used a disjunction of single-character alternatives:

(?:\s|["'])

Since this matches either whitespace, " or ', it can be expressed more directly as a character class without changing its meaning:

[\s"']

So I updated the pattern as follows:

/[\s"']aria-hidden\s*=\s*(?:"true"|'true'|true(?=\s|\/?>))/iu

Here, [\s"'] allows aria-hidden to be preceded by whitespace or by the closing quote of the previous attribute. For the unquoted true form, the lookahead (?=\s|\/?>) is kept to ensure that values such as aria-hidden=truefoo are not matched.

I also added test cases for attributes without whitespace between them and confirmed that the previously discussed cases still work as expected. (https://regexr.com/8o8jo)

[KumJungMin]

The updated commit is b403e09

[DMartens]

Changes LGTM, thanks.