Parses Jekyll-style metadata as a heading

[lamby]

Jekyll (and similar, including Python-markdown) uses the following top-of-file metadata headers:

---
key: value
---
body

However, cmark parses this as a heading:

<hr />
<h2>key: value</h2>
<p>body</p>

Presumably because the --- underneath is the RestructuredText-style underscored headings. Indeed, a more-minimal testcase is actually:

a
-
b

<h2>a</h2>
<p>b</p>

… which I was expecting in a human sense to be parsed as:

<p>a</p>
<hr />
<p>b</p>

I'm not sure what the solution is here. Special-case the triple --- due to Jekyll's widespread use of this?

(I discovered this via https://github.com/wereturtle/ghostwriter/issues/518)

[jgm]

This is out of scope. Use a separate tool to extract the heading and parse it.

[lamby]

This is out of scope.

I am 100% in agreement with you idea that commonmark should not parse and return the metadata as metadata in some kind of of side-channel. 👍

However, would you consider special-casing not parsing it "incorrectly" (kinda...) as a heading in the case of the this style? I am not sure there would be many false-positives given that (at least in my experience) the underlined header style is not as common as the # foo style.

[jgm]

That would amount to blessing this particular style of including front matter as official. Third-party tools make sense here. My lmark, for example (jgm/lmark), wraps cmark with support for yaml metadata and templates.

[lamby]

Oh that's excellent and a much cleaner solution for me. 👍 Thanks for the explanation.

[milahu]

for example (jgm/lmark)

https://github.com/jgm/lcmark

see also: pulldown-cmark (docs)