Table captions

[coryasilva]

How can I get table captions to be included in markdown output? As you can see in the example the table caption is quite important and should not be dropped. (I am not sure if this is intentional or a bug.)

Actual

Table captions are ignored

Expected

Table caption to be included

Steps to reproduce

import rehypeParse from "rehype-parse";
import rehypeRemark from "rehype-remark";
import remarkGfm from "remark-gfm";
import remarkStringify from "remark-stringify";
import { unified } from "unified";

const processor = unified()
  .use(rehypeParse)
  .use(rehypeRemark)
  .use(remarkGfm, { tablePipeAlign: false })
  .use(remarkStringify);
      
const vfile = processor.processSync(`
  <table>
    <caption>
      Table 1: Orencia IV Dosing for Adult Rheumatoid Arthritis and Adult Psoriatic Arthritis
    </caption>
    <tbody>
      <tr>
        <th>Body Weight</th>
        <th>Dose</th>
        <th>Number of Vials</th>
      </tr>
      <tr>
        <td>Less than 60 kg</td>
        <td>500 mg</td>
        <td>2</td>
      </tr>
      <tr>
        <td>60 to 100 kg</td>
        <td>750 mg</td>
        <td>3</td>
      </tr>
      <tr>
        <td>Greater than 100 kg</td>
        <td>1,000 mg</td>
        <td>4</td>
      </tr>
    </tbody>
  </table>
`);
 

[ChristianMurphy]

Table captions don't exist in gfm markdown tables.
https://github.github.com/gfm/#tables-extension-

So where would you propose they go?

[coryasilva]

Above or below the table as a paragraph, just like in HTML.

It was unexpected that the hast-util-to-mdast does not just write a paragraph before or after the table. I see that GitHub markdown does not have a semantic expression for table captions, but not sure that means we should omit them.

Speculation: Maybe they thought it superfluous since it is just a paragraph above or below the table...

Table caption
|a|b|
|-|-|
|1|2|

[wooorm]

There are lots of ways to go about this, I am not sure everyone would want it as a paragraph above, or below. Markdown exists to not have everything that HTML has.
This may make sense for your use case tho, in which case, you should be able to write a handler!

[coryasilva]

Thanks, I will give the handler a try.

Ignoring the caption content still feels wrong. I understand ignoring structural things like colgroup which require bespoke expression but ignoring caption seems a bit draconian when the expression is simply a paragraph -- a feature that Markdown does have.

Either way I love this ecosystem of parsers/trees/utils!

[wooorm]

I can see it, not 100% convinced, but it may be possible.

The HTML spec talks about it as being a “title” for the table though. That sounds like a header/heading.

And examples there show that it contains several paragraphs (its content model is “flow”).

So it may be a problem to some people, but then what is the solution for everyone?

Feel free to ponder this and all the trade-offs in an issue/PR there! :)

[coryasilva]

Good point regarding the flow content model. Gets murky there.