retrieve markdown table class from attributes

[priiduonu]

I would like to be able to write

| header1 | header2 |
|---------|---------|
| r1c1    | r1c2    |
| r2c1    | r2c2    |

: mycaption {.myclass}

and get the table class assigned to the table element, while removing it from the caption.

I know I can achieve it with fenced div blocks, but the shorthand version would be nice.

In the real world I am using pandoc-crossref, so my table usually looks like:

| header1 | header2 |
|---------|---------|
| r1c1    | r1c2    |
| r2c1    | r2c2    |

: mycaption {#tbl:id}

therefore I prefer the short version instead of wrapping the whole thing in extra div.

[jgm]

Pandoc's Markdown doesn't allow attributes that aren't attached to a heading, code, link, image, bracketed span, or fenced div.

I'm not sure how pandoc-crossref lets you do that.

But with commonmark_x (which includes the attributes extension to commonmark), you can do

{.myclass}
| header1 | header2 |
|---------|---------|
| r1c1    | r1c2    |
| r2c1    | r2c2    |

and it will attach the class to the table.

[rnwst]

@priiduonu -- I wrote a filter for cross-references in Lua which supports the same syntax that pandoc-crossref is using to specify Table Attrs (the syntax you are using above). You can find it here. If you are only interested in the code that enables use of this syntax to specify Table Attrs, you could copy the function parse_table_attr into a separate Lua filter. The way this function works is quite simple -- it parses the table caption as a Heading, since Headings use the same syntax to specify Attrs.

@jgm -- would you be open to having pandoc's Markdown extended to support this syntax for specifying Table Attrs? The syntax seems natural to me, since it is already used in Headings to specify Attrs.

[jgm]

@jgm -- would you be open to having pandoc's Markdown extended to support this syntax for specifying Table Attrs? The syntax seems natural to me, since it is already used in Headings to specify Attrs.

Worth considering. Comments welcome.

[tarleb]

It's a fairly common and natural extension to pandoc's syntax, Quarto has it, too.

I'd be in support of building this into pandoc. The only mildly awkward thing, IMHO, arises when using captions with multiple paragraphs. Attaching the attribute to the last paragraph is slightly weird. But, as noted above, people are already doing that, so it seems fine.

[rnwst]

@tarleb -- I did not realise it was possible to create table captions with multiple paragraphs in pandoc's Markdown. Could you give an example of how this can be done? I tried a few things and couldn't get it to work (I also don't think this is documented in the manual?).

[tarleb]

Whoops, you're right. I must have imagined that.