Fix class extraction followed by ( in Slim

[RobinMalfait]

This PR fixes an issue where using the class shorthand in Slim templates, followed by an ( results in the last class being ignored.

E.g.:

body.border-t-4.p-8(class="#{body_classes}" data-hotwire-native="#{hotwire_native_app?}" data-controller="update-time-zone")

This is because we will eventually extract p-8 but it's followed by an invalid boundary character (.

To solve this, we make sure to replace the ( with a space. We already do a similar thing when the classes are followed by an [.

One caveat, we can have ( in our classes, like bg-(--my-color). But in my testing this is not something that can be used in the shorthand version.

E.g.:

div.bg-(--my-color)

Compiles to:

<div --my-color="" class="bg-"></div>

So I didn't add any special handling for this. Even when trying to escape the (, - and ) characters, it still doesn't work. E.g.:

div.bg-\(--my-color\)

Compiles to:

<div class="bg-">\(--my-color\)</div>

Test plan

1. Added test for the issue
2. Existing tests pass
3. Verified via the extractor tool:

Before	After

---

Fixes: #17277

[thecrypticace]

idk if it matters but should we worry about bg-red-500/(--my-var) here too? If so we should do !matches!(cursor.prev, b'-' | b'/')

[thecrypticace]

Okay, answering my own question with a note: I think we can drop that matches check entirely based on the PR desc

This is because the check (after adding the /) makes these two tests I just wrote pass:

let input = r#"
    body.border-(--my-color).p-8(class="\#{body_classes}" data-hotwire-native="\#{hotwire_native_app?}" data-controller="update-time-zone")
"#;
Slim::test_extract_contains(input, vec!["border-(--my-color)", "p-8"]);

let input = r#"
    body.border-red-500/(--my-color).p-8(class="\#{body_classes}" data-hotwire-native="\#{hotwire_native_app?}" data-controller="update-time-zone")
"#;
Slim::test_extract_contains(input, vec!["border-red-500/(--my-color)", "p-8"]);

but as you mentioned this is invalid Slim syntax

[RobinMalfait]

Yep yep, added an additional test just so that this case is covered: 1a9c242

[RobinMalfait]

Alright we do need this for this situation: 645db11

div class="bg-(--my-color) bg-(--my-color)/(--my-opacity)"