`<p><br /></p>` results `\` due to `wrapText`

[JounQin]

Initial checklist

• I read the support docs
• I read the contributing guide
• I agree to follow the code of conduct
• I searched issues and couldn’t find anything (or linked relevant results below)

Affected packages and versions

hast-util-to-mdast@8.2.1

Link to runnable example

No response

Steps to reproduce

import fs from 'fs'

import type { Element } from 'hast'
import { all } from 'hast-util-to-mdast'
import { code } from 'hast-util-to-mdast/lib/handlers/code.js'
import rehypeParse from 'rehype-parse'
import rehypeRemark from 'rehype-remark'
import remarkStringify from 'remark-stringify'
import { unified } from 'unified'

const processor = unified()
  .use(rehypeParse)
  .use(rehypeRemark, {
    handlers: {
      div(h, node: Element) {
        if (
          (node.properties?.className as string[] | undefined)?.includes(
            'toc-macro',
          )
        ) {
          return
        }
        return all(h, node)
      },
      pre(h, node: Element) {
        // `h.handlers.code` is actually `inlineCode`
        return code(h, {
          ...node,
          children: node.children.map(child =>
            child.type === 'text'
              ? {
                  type: 'element',
                  tagName: 'code',
                  children: [child],
                }
              : child,
          ),
        })
      },
    },
  })
  .use(remarkStringify, {
    fences: true,
  })

const main = async () => {
  const text = await fs.promises.readFile('index.html', 'utf8')
  const vfile = await processor.process(text)
  await fs.promises.writeFile('index.md', vfile.value)
}

main().catch(console.error)

<!-- index.html -->
<p><br /></p>

Expected behavior

empty line

Actual behavior

\

Runtime

Node v12

Package manager

yarn v1

OS

macOS

Build and bundle tools

esbuild

[JounQin]

It occurs when upgrading remark-stringify from 8.0 to 10.0, so maybe mdast-util-to-markdown is the root cause, I'll continue to debug.

[JounQin]

My workaround for now:

{
  handlers: {
    br(h, node) {
      return h(node, 'html', '<br>')
    },
  },
}

[wooorm]

What is the expected output you want?

[wooorm]

More specifically, why should the break be removed?

[JounQin]

What is the expected output you want?

I'm not sure actually, maybe a simple \n? Or should it check whether it is safe to yield \\\n with siblings? Although <br> will be better for table contents.

[wooorm]

Is your question specifically (a) about a break on its own in a paragraph, which is now broken and can’t roundtrip?

Or, due to the other issue, are you saying (b) all breaks should be different?

I think (a), right?

[JounQin]

I'm trying to do html -> markdown -> html

<p><br></p> -> \\\n -> <p>\</p>, is this expected?

[JounQin]

My original issue is about (a, but for <br> in table, it's another problem.

[wooorm]

• break in table isn’t a “problem”, because it is turned into a space right now, due to wrapText.
• <br>s and other HTML tags are not used by this project by default, see HTML in markdown if you want that
• This issue isn’t really about wrapText, because in this case (in a paragraph), it’s fine to use line endings. In other cases, such as tables, it is not, and that’s what the feature does (and does properly)

For a solution, does '<p><br></p>' -> '' work for you? So, both the break and the paragraph are not output at all?

[JounQin]

break in table isn’t a “problem”, because it is turned into a space right now, due to wrapText.

A space is not a new line, some markdown editors like typora support <br> in table, but nevermind, it is not related to this issue.

<br>s and other HTML tags are not used by this project by default, see HTML in markdown if you want that

This is what my workaround doing.

For a solution, does '<p><br></p>' -> '' work for you? So, both the break and the paragraph are not output at all?

I think it's fine to me.