Precedence between two "text" constructs

[StaloneLab]

Hello 👋

I maintain the ZMarkdown repository, which has a lot of extensions to remark. I have so far almost managed to get remark-kbd switched to micromark, but I struggle to find how to give precedence to my custom kbd element over emphasis. The problem specified here only concern micromark by itself, remark is not involved.

The code for the extension can be found here: https://github.com/zestedesavoir/zmarkdown/tree/next/packages/micromark-extension-kbd , and here is a test, that will not give what I expect it to:

*foo||*||

gives

<p><em>foo||</em>||</p>

but I would want it to give:

<p>*foo<kbd>*</kbd></p>

As you can see, currently, attention has a priority over kbd. I have not found how it would be possible to invert this behavior, excepted by switching kbd from text to flow, which doesn't seem like a good solution. Could you indicate me how I could manage to get it working as expected?

Thanks a lot 😃

[wooorm]

hey hey!

That conflicts a bit with markdown:

*a~b*c~

~a*b~c*

Yields:

a~bc~

a*bc*

So the earlier one “wins”.
That’s because emphasis/strong/delete/link/image are parsed separately: as potential starts/ends, and finally when done resolved based on whether they match and which comes first, and the starts/ends that don’t match reverted back to text.

But, kbd could perhaps be parsed similarly to code / math? So that it parses from start to end in one go.

And then its contents chunks could be linked to form more text? That would be similar to titles (but then using text instead of string as the content type)

[wooorm]

depends on what you want:

• Do character references and character entities work in kbd?
• Do emphasis and links and such work in kbd?

[StaloneLab]

Inline constructs do not work inside kbd, so no emphasis, links, or anything. I'm not sure about character references, but it seems to me like we do not want them to work in kbd (as in code? I'm unsure about it).

I'm curious though, how would you make another construct work inside kbd? Should I call the tokenizer from inside?

[wooorm]

I was assuming you did want such things, hence these questions and the “And then its contents chunks could be linked to form more text? That would be similar to titles (but then using text instead of string as the content type)”

• If you don‘t want anything inside, take code as inspiration
• If you want character references, take title
• If you want all phrasing constructs, take title but use text instead of string as a content type.

Adding a contentType on a token, and linking them together, will be seen by micromark as something to parse in the future. When you’re done with what you parsed, micromark will recursively find that embedded content, and parse it

[wooorm]

did you see https://github.com/micromark/micromark#creating-a-micromark-extension, that might help clarify such things.
And, you can also look at the types in the code. It’s quite extensive: https://github.com/micromark/micromark/blob/main/packages/micromark-util-types/index.js

[StaloneLab]

I knew the document existed, but hadn't taken time to read it until now; it answers my questions, thanks.

Also, hats off for the work on micromark, seems very powerful 👍