merging adjacent table cells

[sbonaretti]

Hi! Is there a way to merge to (horizontally) adjacent cells in a table?
Thanks!!!

[slochower]

I don't think this is possible until the upstream Markdown processor that pandoc uses, implements it. There is a very long thread discussing the possibility and implementation details: jgm/pandoc#1024. I can't quite tell what the timescale is for this, but I'm confident that a lot of people care about this and the authors of pandoc seem to be taking it seriously.

[vincerubinetti]

Unfortunately the only way to accomplish this right now is to write out the table "manually" (in html) in your markdown, and use colspan, like this:

<table>
 <tr>
  <td colspan="2"> </td>
  <td> </td>
 </tr>
 <tr>
  <td> </td>
  <td> </td>
  <td> </td>
 </tr>
 <tr>
  <td> </td>
  <td> </td>
  <td> </td>
 </tr>
</table>

Reference:
https://www.computerhope.com/issues/ch001655.htm

[dhimmel]

Unfortunately the only way to accomplish this right now is to write out the table "manually" (in html) in your markdown

@vincerubinetti are you sure the HTML table method works when passing through Pandoc? My understanding is that since Pandoc's Astract Syntax Tree doesn't support colspans, if the table gets interpreted by Pandoc, this solution won't work. However, I'm not sure whether Pandoc would interpret the table or just pass-through the HTML verbatim.

[vincerubinetti]

Just tested it, it does work. My understanding is that anything you write in plain HTML within a markdown file will just get passed through/ignored by pandoc. Is that consistent with your understanding?

[dhimmel]

Just tested it, it does work.

Good to know

Is that consistent with your understanding?

I don't have a good understanding, but I think it may be dependent on which input and output formats are chosen. When going from md to html, it seems that it keeps the HTML untouched rather than interpreting it as part of the AST.

[slochower]

It will pass through to the HTML output, but did you check that it also passes through to PDF and DOCX? That part I'm not sure about.

[vincerubinetti]

PDF output comes right from the HTML output, so as long as WeasyPrint/Athena supports colspan, it should work. As for docx, I don't know but I doubt it.

[slochower]

I think that writing the table as HTML will also break the functionality used by by pandoc-tablenos.

@vincerubinetti thanks for checking!

[sbonaretti]

I have finally tried this out. It works, and it was not hard. I opened the built .html and copied the table to my .md file.
I just have a new issue now: In some cells I reference to the literature. The md command @doi:something becomes something like <span class="citation" data-cites="aNdiyMUf">[<a href="#ref-aNdiyMUf" role="doc-biblioref">3</a>]</span> in the html, which obviously does not reference to anything - and I get an error during the built. Is there an easy fix to this? Thanks!!!

[dhimmel]

I opened the built .html and copied the table to my .md file.

Nifty solution. May be kind of a pain if you have to change the table or get table numbering to work, but cool to see the workaround.

I get an error during the built.

Can you paste the error text?

does not reference to anything

If Pandoc is still able to render the markdown in the HTML table cells, you could just revert the HTML citations back to the original citations. If you don't have too many citations in the table, it may just be easiest to do this manually. The output/citations.tsv file will show you the mapping from manuscript_id (e.g. doi:something) to short_id (e.g. @aNdiyMUf).

[vincerubinetti]

I have one solution. The others might have different solutions.

Instead of copying the html of the table into your input markdown, I'd just leave the table as regular markdown, and add the colspan attribute into the output itself. The downside of this, though, is that every time you re-build the output, that colspan change will go away.

To solve this, we could create a plugin that looks for any cell that has some flag in commented-out text, say <!-- colspan:2 -->, and add the colspan html attribute to that cell dynamically. This should work for html output, and pdf output if Athena is being used (because it supports javascript).

So basically you'd use markdown tables like normal, but add a little extra text to specify the colspan (and also leave the appropriate preceding cells blank):

| Tables                      |     Are     |  Cool |
|-----------------------------|:-----------:|------:|
| col 1 <!-- colspan: 2 -->   |             | $1600 |
| col 2                       |     money   |   $12 |
| col 3                       |     money   |    $1 |

In fact, this could be useful for more than just tables. It could be a way to bypass the issue of markdown/pandoc not being able to attach parameters to any element. It could look for any comment in the format <!-- attribute: value -->, and add that property to whatever html element it's in.

[dhimmel]

we could create a plugin that looks for any cell that has some flag in commented-out text

This is a cool idea. Won't work for DOCX outputs (but will an HTML table in a markdown input get converted to DOCX?).

@vincerubinetti happy for you to prototype this solution if you have time. If it works, we could include it as in the plugins directory (perhaps not added to the build command by default). I'm interested to see whether it works, and how general of a workaround it could be for "unsupported by pandoc" things.

[slochower]

Would be nice to look for a <!-- colspan: 2 --> attribute. That would be super helpful.

I'm fairly certain an HTML table embedded in Markdown will not get converted to DOCX with pandoc. I know that other HTML, like <span class="red">this is red text</span> embedded in the Markdown get dropped going to DOCX.

[sbonaretti]

@dhimmel

Can you paste the error text?

Here it is: ERROR: No anchor #aNdiyMUf for internal URI reference

I have noticed that:

• If I cite the paper somewhere else in my text, I get the clickable link to the citation in the rendered HTML table - I simply add the markdown command into the HTML line (e.g. <td>Clark et al. [@doi:10.1101/657676]</td>)
• If I do NOT cite the paper somewhere else in my text, I get the error above

Hope this helps