Do you put a space between the "fence" and the language name, as it seems from your code?

I do not> 

I do, but the regex can be made more flexible :)

I think I'll publish this as a library if more people are interested.

> After that we escape some markdown-specific characters, so they will be preserved after markdown compilation. I guess that is necessary b/c your md parser is interpreting text inside the code block? I think that more modern / better (?) parser don't do that. I think I did not have that problem with flexmark, before I migrated to asciidoctor.

Well, the code block has been converted to HTML before it goes through the markdown transformer. I don't know what the behavior of is modern markdown parsers when you do:

<pre><code>[foo](bar)</code></pre>

but markdown-clj does the wrong thing here ;)

perhaps replacing it with an HTML entity also works

yeah that works

The correct behavior according to the "standard" is NOT to process markdown inside html block elements. But some parsers still do 😞 Very interesting post. Good to learn about some of the powers of kondo and rewrite! 🙂 Thanks for sharing!

yeah it might be better to just escape all markdown-related things with html entities, just to be on the safe side

Updated the blog post now

with the new code