normalize class property

[blackmann]

Changes

Attempting to use some of the common transformers raises an exception:

MDXError: classValue.replace is not a function
    at TransformContext.transform (node_modules/@astrojs/mdx/dist/index.js:93:27)
    at async Object.transform (node_modules/vite/dist/node/chunks/dep-V3BH7oO1.js:64060:30)
    at async loadAndTransform (node_modules/vite/dist/node/chunks/dep-V3BH7oO1.js:49741:29)
    at async instantiateModule (node_modules/vite/dist/node/chunks/dep-V3BH7oO1.js:50759:10)

This is because addClassToHast returns an array of strings. But Arrays don't have a .replace. See: https://github.com/antfu/shikiji/blob/31241ba4dc47c2d4d7809ee08c6573b26b07b812/packages/shikiji-core/src/utils.ts#L44

Testing

I couldn't figure out how to write a unit test for this. But I edited a build version with this logic and it works.

Docs

[changeset-bot]

🦋 Changeset detected

Latest commit: 528b57a

The changes in this PR will be included in the next version bump.

Not sure what this means? Click here to learn what changesets are.

Click here if you're a maintainer who wants to add another changeset to this PR

[ematipico]

If value is cast to string, when does the function return null?

[blackmann]

Updated @ematipico

[matthewp]

What does this fix? Is there an issue or code example?

[blackmann]

I gave a description with the PR; to elaborate, if you added a codeblock like the following to an .md content file:

  // &mut means mutable reference
  fn next_question(&mut self) -> &Question {
    let count = self.questions.len() - 1; // [!code --]
    if self.current_question().is_answered() && self.current_index < count {
      self.current_index += 1;
    }

    self.current_question()
  }

Notice on line 3, we do // [!code --], this triggers the transformerNotationDiff transformer. In that process, addClassToHast is called which converts the class of that line from string to an array of string.

This PR converts it back to a string by just concatenating them.

Here's a screenshot of the error

@matthewp

[blackmann]

This is a patch follow-up to #9643 which exposes the transformers API from shikiji which was published earlier today.

[bluwy]

Thanks for the fix! This makes sense to me. Could we also handle this for style? I assume transformers could also change that to a non-string now.

Maybe we can rename and generalize the function as normalizePropAsString

[blackmann]

@bluwy @ematipico

ready

[blackmann]

@florian-lefebvre @bluwy @ematipico

Can we get this merged soon?