Pandoc now supports djot

[jgm]

both as an input and an output format. https://github.com/jgm/pandoc/releases/tag/3.1.12

[dai]

Finally, Pandoc supports Djot!

Big shout it out Professer.

[uvtc]

Thank you!

[Omikhleia]

Erm... What about round-tripping?
(Edit: walking on eggs...)

[jgm]

The djoths library does do round-trip tests for all of the test cases. I caught quite a few issues in the writer this way. Caveat: currently this is only tested without automatic line-wrapping, which may introduce issues we haven't caught yet.

[Omikhleia]

I was actually thinking to round-tripping conversions between formats with same base features (A --> B --> A), such as (running pandoc 3.1.13):

$ pandoc -f markdown+mark -t djot
==toto==
[toto]{.mark}
$ pandoc -f djot -t markdown+mark
{=toto=}
[toto]{.highlighted}

Or

$ pandoc -f djot -t markdown+strikeout
{-toto-}
[toto]{.deleted}
$ pandoc -f markdown+strikeout -t djot
~~toto~~
{-toto-}

This does not seem as consistent as e.g.

$ pandoc -f html -t markdown+strikeout
<p><del>toto</del></p>
~~toto~~
$pandoc -f html -t markdown+mark
<p><mark>mark</mark></p>
==mark==

But there might be more that meets the eye there, too:

$ pandoc -f djot -t html
{-toto-} {=toto=}
<p><span class="deleted">toto</span> <span class="highlighted">toto</span></p>
$ pandoc -f markdown+mark+strikeout -t html
~~toto~~ ==toto==
<p><del>toto</del> <mark>toto</mark></p>

[Omikhleia]

Or

$ pandoc -f html -t djot
<p><del>toto</del> <ins>toto</ins></p>
{-toto-} [toto]{.underline}
$ pandoc -t html -f djot
{-toto-} {+toto+}
<p><span class="deleted">toto</span> <span
class="inserted">toto</span></p>

(non-symmetric mapping, and underline vs. inserted)

[jgm]

To understand what is going on, you'll want to look at how pandoc is parsing things: instead of djot -> X -> djot, do:

• djot -> native
• native -> X
• X -> native
• native -> djot

This will tell you where things are changing. Perhaps a concrete suggestion for specific changes might come out of this, but it's delicate -- pandoc has its own concepts and they differ a bit from djot's.

[Omikhleia]

What is the rationale for parsing {=xxx=} into an AST Span with .highlighted:
https://github.com/jgm/pandoc/blob/main/src/Text/Pandoc/Readers/Djot.hs#L192

Rather than .mark, which is what the Markdown reader does for the (seemingly) semantically identical ==xxx==:
https://github.com/jgm/pandoc/blob/edf21e6dab8cddf15dac6f494291396d4f4df8db/src/Text/Pandoc/Readers/Markdown.hs#L1726

And which is supported for rendering as <mark> in the HTML writer, around those lines it seems...
https://github.com/jgm/pandoc/blob/edf21e6dab8cddf15dac6f494291396d4f4df8db/src/Text/Pandoc/Writers/HTML.hs#L1402

... Or by the LaTeX Write as \hl:
https://github.com/jgm/pandoc/blob/edf21e6dab8cddf15dac6f494291396d4f4df8db/src/Text/Pandoc/Writers/LaTeX.hs#L822-L824

Or is {=djot=} considered different from extended ==markdown== for any reason?

[bpj]

@jgm what is the status of this? Will it be in the next release of Pandoc?

[jgm]

It's already in the current release.