Handle multiple classes on code blocks

[metasoarous]

First off, thanks for this library!

I see that it's possible to specify a single class for a code block by append it to the tick marks in the markdown, as is possible with GitHub markdown in order to specify language type for formatting. However, while GitHub is fine with you specifying multiple such classes, markdown-to-hiccup treats the second class name as though it's part of the content of the code block. Example:

```edn vega-lite
{:data {:values [{:a 1 :b 3} {:a 4 :b 8} {:a 7 :b 5}]}
 :mark "point"
 :encoding {:x {:field "a"} :y {:field "b"}}}
```

in markdown-to-hiccup =>

vega
{:data {:values [{:a 1 :b 3} {:a 4 :b 8} {:a 7 :b 5}]}
 :mark "point"
 :encoding {:x {:field "a"} :y {:field "b"}}}

While GitHub gives us this:

{:data {:values [{:a 1 :b 3} {:a 4 :b 8} {:a 7 :b 5}]}
 :mark "point"
 :encoding {:x {:field "a"} :y {:field "b"}}}

It would be extremely valuable for me as I'm adding some functionality to https://github.com/metasoarous/oz which would make it possible to embed vega/vega-lite visualizations in markdown documents using this syntax. The reason two classes is important is that visualizations might either be json or edn, but I obviously don't want to assume every such chunk of code should be rendered as vega-lite, so I need another class to accomplish that. I could have separate classes like json-vega-lite & edn-vega-lite, but that is somewhat annoying, and won't lead to good rendering as regular markdown (say on GH).

Are you open to markdown-to-hiccup matching the GitHub behavior with respect to this issue?

Thanks in advance!

[mpcarolin]

Hi there!

This library does a few small things: it links together markdown-clj, which takes markdown and outputs a standard HTML string, and hickory, which then takes that string and converts it into hiccup. I do some decoding/encoding work and other small manipulations of the resulting hiccup, but I don't have that much control, at least until I find the time to write a direct markdown -> hiccup converter. :)

I think I've narrowed down the culprit to markdown-clj. When I tried entering your code block into that library's live demo site here, the result was exactly what you encountered.

I know the author @yogthos is open to both user suggestions and pull requests (see yogthos/markdown-clj#73). I suggest heading there first.

But if you still think markdown-to-hiccup is more responsible for this type of change, let me know!

[yogthos]

Hi,

As a note, commonmark-hiccup might be a better fit here because it parses markdown into Hiccup directly. This avoids the step of generating the HTML string and parsing it using hickory.

[mpcarolin]

Thanks for the link! You may have deprecated my library in one comment haha. I will check it out and if it covers all the bases this library does then I will post a note in the readme directing people to commonmark-hiccup.

[yogthos]

It doesn't look like commonmark-hiccup is actively maintained though, so might make sense to take it over. :)

[mpcarolin]

I may try just that 👍

[metasoarous]

Great! Thanks for the feedback!

[mpcarolin]

@metasoarous, I saw your PR to clj-markdown. I will update this library’s version of clj-markdown later today to include those changes.

[metasoarous]

@mpcarolin Wonderful! Thank you! I don't see a release on Clojars yet, but I noticed you're using deps.edn, which should allow you to specify a git dependency pointing to the merge commit in the interim, if you're so inclined.

Thanks again!

[mpcarolin]

@metasoarous

Okay just deployed it as version 6.1. Thank you for your patience!

[metasoarous]

Amazing! Thanks so much!