Is there a way to lint the url in a link?

[p0fi]

Given the explanation in #320 it currently seems that url in the following example is never linted.

<a href="url">text</a>

I know that you could lint url using the raw scope and build a regex and I have done that. The problem is that the raw scope is so wide that the regex starts to fail me because it catches so many parts of the source text which in raw looks almost like a link in asciidoc but is not. The pattern for a link is [[some reference text]] where some reference text becomes the url in the converted HTML. If the source includes for example code snippets a similar pattern (double square brackets) often occur but that are no links and they will not be converted in

<a href="#some reference text">some reference text</a>

by asciidoctor.

Is there a scope which lints url instead of text in the example on the top?

[jdkato]

No, there is not a more specific scope than raw for this case.

I think this is beyond Vale's purpose as a generic (many markup inputs) prose linter and ventures into the space of an AsciiDoc-specific linter.

[jdkato]

The problem is that no such thing exists.

This is true, but I am hesitant to expand the scope of Vale into this territory. There are many related concerns/features that I'm not interested in exploring at this time.

[p0fi]

I completely understand your desire to maintain a focused approach in addressing the initial problem with Vale. However, it would be incredibly beneficial to include an option for linting the url component of a link. This addition would greatly assist in resolving a challenging problem that currently requires a seemingly impossibly complex regex – if it is at all solvable. Moreover, implementing this feature would possibly only require a minor adjustment on Vale's side since it's already aware of the syntax and actually knows what a link is and what other text is, which looks syntactically similar.

[eskwayrd]

Vale is a prose linter, not a link checker. While part of Vale's processing shares some similarities with a link checker, the differences are where all the complexity lies (and it's not minor). As you noted, a "seemingly impossibly complex regex" means a regex engine is not the right tool. You need a link checker that lets you apply custom logic.

FYI: AsciiDoc's [[some reference text]] is syntax for specifying a custom id to the following block. It does not, by itself, create a link. Also, id attribute values cannot contain spaces, so that specific syntax is parsed as content and appears in the output.

[p0fi]

Thanks for your comment @eskwayrd! Your remarks made me think more. Especially this made me think

a regex engine is not the right tool

I did solve my problem with modifying my rule. Instead of checking for existence and only use regex, I changed to script and wrote a tengo script in which I could narrow the scope down on my own! Thanks for that!

[PeterBurner]

Could you please share your solution here?

[PeterBurner]

This proposed feature would be extremely helpful for me as well.

I'm currently facing a similar problem, where I am trying to restrict the content and structure of anchors. Implementing this feature would help solve it. I fully support this proposal and look forward to seeing it implemented.

Great suggestion!