Anchor tag breaks markdown

[craftzdog]

I'm not sure where the actual problem resides but I report it here.

Subject of the issue

If a given HTML has an anchor tag, it yields the broken markdown.

Your environment

• OS: macOS
• Packages:

  "rehype-parse": "^7.0.1",
  "rehype-remark": "^8.0.0",
  "remark-stringify": "^8.1.1",
  "unified": "^9.0.0"

• Env: node 14.3, npm 6.14.5

Steps to reproduce

If you have an HTML like so:

<div>
  <a name="foo" />
  <span>
    <div>test1</div>
  </span>
</div>

Convert it into Markdown:

export default function HTML2Markdown(html: string): string {
  var unified = require('unified')
  var parse = require('rehype-parse')
  var rehype2remark = require('rehype-remark')
  var stringify = require('remark-stringify')

  const c = unified()
    .use(parse)
    .use(rehype2remark)
    .use(stringify, {
      listItemIndent: '1',
      commonmark: true,
      fences: true
    })
  return (
    c
      .processSync(html)
      .toString()
}

Expected behavior

You should get:

test1

Actual behavior

It outputs:

[test1](<>)

[](<>)

[wooorm]

rehype is for HTML, not for XML. Your link isn’t closed, you have divs in spans, it’s all messed up! If I open your “HTML” in a browser, I get something very similar to what you’re getting with remark/rehype:

---

Why do you have such weird HTML? Can you fix that?

[craftzdog]

Thanks for the reply. Yes, I thought the same. I use remark-rehype for supporting importing HTML files into my app as Markdown. I got this broken HTML from my app's user who exported a note from Evernote. So, it looks like an Evernote's bug where it generates such weird HTML..

[wooorm]

oh gosh, that’s some really weird stuff that Evernote is doing 😅

It’s indeed a bug there, you could report it with them maybe, but I’m not sure they’d fix it. An alternative would be to do some processing with a plugin on “evernote html”? Or maybe support their .enex XML files (I see some projects and github that do that)? 🤷‍♂️

Anyway, not something for this project!