@so1ve/rehype-shiki

@so1ve/rehype-shiki

Rehype plugin to highlight code blocks with Shiki

Installation

$ npm install @so1ve/rehype-shiki shiki

Format

Code blocks must have the following format:

<pre>
<code class="language-javascript">
return unified();
</code>
</pre>

This is the format produced by remark-parse & remark-rehype from the following Markdown:

```javascript
return unified();
```

Usage

See source/index.test.ts for examples.

Options

• highlighter (required): An instance of the Shiki highlighter, or an object whose keys are identifiers and values are Shiki highlighters, in which case @so1ve/rehype-shiki combines the outputs of all the highlighters.
• throwOnUnsupportedLanguage (default: false): A boolean indicating whether to throw an exception if a code block refers to an unsupported language.

Security

@so1ve/rehype-shiki doesn’t open you up to cross-site scripting (XSS) attacks as long as Shiki doesn’t (which it doesn’t).

How Is This Different from rehype-shiki?

rehype-shiki is great! That’s how I learned about Shiki and I fell in love with it. The following are the ways in which @so1ve/rehype-shiki is different:

1. TypeScript support.
2. Shiki is declared as a peerDependency, so @so1ve/rehype-shiki doesn’t have to be updated when new versions of Shiki are released (as long as Shiki’s API remain compatible). See https://github.com/rsclarke/rehype-shiki/pull/48 https://github.com/rsclarke/rehype-shiki/pull/46 https://github.com/rsclarke/rehype-shiki/issues/47 https://github.com/rsclarke/rehype-shiki/issues/2.
3. You must pass in an instance of the Shiki highlighter, @so1ve/rehype-shiki won’t create one for you. This means that:
   1. The Shiki highlighter instance is reused on every invocation of the processor, instead of being recreated every time you call the processor.
   2. The transformer is synchronous, so you may use it with .processSync().
4. Instead of looking at the tokens produced by Shiki and generating hast, @so1ve/rehype-shiki lets Shiki produce HTML and parses the result. The advantage is that when Shiki improves the output with things like italics @so1ve/rehype-shiki will pick the changes up with no extra work. The disadvantage is that we’re producing HTML as a string and then parsing it right back; this is slower, but in most cases it won’t matter and I think the previous advantages outweighs this disadvantage. (Also, the language-* class will be removed from the produced HTML, so you may need to adapt your CSS.)
5. Support for multiple highlighters.

That said, I contacted the maintainers of rehype-shiki and try to merge the code bases. We’ll see…

Changelog

2.2.0

• Updated the peer dependency to shiki@0.11.1.

2.1.0

• Added a feature that preserves the position of the top element node. Useful for products that need to map the HTML back to the Markdown that generated it (see tests).

2.0.0

• ESM only.
• Compatible with unified 10.