anchor link checking is broken

[OmarTawfik]

Thanks for the awesome linter. Long time user, first time contributor.
I have a bunch of files that define HTML code blocks, with id attributes:

<code id="foo"> .... </code>
<code id="bar"> .... </code>

Then I have some links that reference them:

<a href="#foo"> ... </a>
<a href="../other-file/#bar"> ... </a>

After upgrading to the latest version 3.10.3, it broke CI because of new errors everywhere. I traced it down to the following issues:

1. The linter doesn't collect links from id attributes. So any links pointing to it (like href="#foo") produce errors.
2. The linter doesn't lint href attributes that have a file path AND a header.
   1. href=".#id" for the same markdown file.
   2. href="../relative/path#id" for a markdown file at ../relative/path.md or ../relative/path/index.md.

Any feedback/pointers are greatly appreciated. I imagine it is not too hard to fix, given that the linter has partial support for it already.

Although my plate is currently full, I welcome any code pointers/green light on how to fix it, and I can take a stab at it at some point in the future.

[OmarTawfik]

cc @tcort any thoughts about how involving the fix is? I imagine it is a simple change, but probably involves a few different packages.

[MikeMcC399]

• This is also discussed in issue Parse HTML in order to find anchors (like #tomato) in the same document #195

[github-actions]

This issue has been marked as stale because it has been open 60 days with no activity. It will be closed in 30 days unless the stale label is removed or someone adds a comment.

[per1234]

This has been fixed (I assume by 0348390 / #331, but can't precisely bisect due to #368).

$ printf '[link to HTML anchor](#html-anchor)\n\n<code id="html-anchor"> .... </code>\n' > /tmp/foo.md

$ npx github:tcort/markdown-link-check#29694c8 /tmp/foo.md  # Last usable bad commit.

FILE: C:/Users/per/AppData/Local/Temp/foo.md
  [✖] #html-anchor

  1 links checked.

  ERROR: 1 dead links found!
  [✖] #html-anchor → Status: 404

$ npx github:tcort/markdown-link-check#834a4e2 /tmp/foo.md  # First usable good commit.

FILE: C:/Users/per/AppData/Local/Temp/foo.md
  [✓] #html-anchor

  1 links checked.