Warn against dead links within the document

[ben-clayton]

It would be fantastic if bad anchor-style links were detected and warned.
I wouldn't expect bad external URL links to be caught.

[koppor]

Dead external links can be discovered with markdown-link-check:

One file: npx markdown-link-check README.md.

Multiple files outputting the result into a text file:

find . -name \*.md -exec npx markdown-link-check -qq {} \; > bad-links.txt

Note that I used npx to avoid explicit global installation of the markdown-link-check package.

[nichtich]

Thanks for the pointer to markdown-link-check, but I'd prefer to have basic support as part of markdownlint instead of having to use yet another tool with different usage and a lot of functionality I don't need anyway. The dead-links use case in this issue is anchor-style links, so implementation would be:

1. extract all heading titles and create anchor ids
2. extract all link targets starting with # and check whether a matching id exists.

Note this would not catch link targets within inline HTML but this is discouraged by MD033 anyway.

[DavidAnson]

@nichtich Is the anchor ID algorithm part of a specification anywhere, or are you suggesting making this specific to GitHub?

[nichtich]

Anchor links from header titles have been discussed in CommonMark but it's not part of CommonMark. Popular implementations that create anchor ids from header titles include Pandoc and GitHub. Pandoc also supports GitHub algorithm (see feature gfm_auto_identifiers), so GitHub algorithm seems the best choice:

• spaces are converted to dashes (-)
• uppercase characters are converted to lowercase characters
• punctuation characters other than - and _ are removed
• if nothing is left, use the identifier section

For a more extensive comparison of algorithms see https://babelmark.github.io/

[DavidAnson]

Great info, thank you!