Strikethrough serialization issue with mdast-util-to-markdown (GFM)

[martin-lysk]

First of all thank you for the awesome project!

I'm using mdast-util-to-markdown with the GFM extensions, and I'm running into issues with roundtripping content that mixes bold (**) and strikethrough (~) markers.

Problem

I expect the following input Markdown:

**Hello ~world~**~ bold~

to serialize back to the same string. Instead, it gets transformed into:

**Hello ~~world~~**~~ bold~~

That breaks the roundtrip, because the internal space is no longer escaped, and the strikethrough range has changed.

Interestingly, a similar pattern works fine when using emphasis (_) instead of strikethrough:

Input:

**Hello _world_**_ bold_

Output (correct):

**Hello _world_**_ bold_

Reproduction

Here's a minimal script to reproduce:

import {fromMarkdown} from 'mdast-util-from-markdown'
import {toMarkdown} from 'mdast-util-to-markdown'
import {gfmFromMarkdown, gfmToMarkdown} from 'mdast-util-gfm'
import {gfm} from 'micromark-extension-gfm'

const boldStrikethrough = '**Hello ~world~**~ bold~'
const boldEmphasis = '**Hello _world_**_ bold_'

const tree1 = fromMarkdown(boldStrikethrough, {
  extensions: [gfm()],
  mdastExtensions: [gfmFromMarkdown()],
})
console.log(toMarkdown(tree1, {extensions: [gfmToMarkdown()]}))

const tree2 = fromMarkdown(boldEmphasis, {
  extensions: [gfm()],
  mdastExtensions: [gfmFromMarkdown()],
})
console.log(toMarkdown(tree2, {extensions: [gfmToMarkdown()]}))

Question

Is this a configuration issue, or a bug in mdast-util-to-markdown? It looks like [encodeInfo](https://github.com/syntax-tree/mdast-util-to-markdown/blob/main/lib/util/encode-info.js#L30) only accounts for _ and *, but not ~.

Would love any guidance or confirmation here

[wooorm]

Hi! Yes, probably a bug. I fixed it for emphasis/strong (standard, common mark), and that should probably also be applied to extensions (gfm strike through being the only one maintained by us!)