Docx to Markdown: write tables as HTML

[denismaier]

When converting docx to Markdown, us it possible to use Html for the tables? I'm having tables where the Markdown table syntax is hard to read, and I'd prefer to work with html tables.

[massifrg]

Maybe a lua filter like this could work (I did not test it):

return {
  {
    Table: function(t)
      if FORMAT == 'markdown' then
        local doc = pandoc.Pandoc( { t } )
        local html = pandoc.write(doc, 'html')
        return pandoc.RawBlock('markdown', html)
      end
    end,
  }
}

[denismaier]

Cool thanks. I'll also need to make sure the html tables get interpreted as tables when converting to other formats than HTML, but @tarleb seems to have already written a filter for that.

[denismaier]

And I still need to test your filter....

[denismaier]

Corrected filter:

return {
  {
    Table = function(t)
      if FORMAT == 'markdown' then
        local doc = pandoc.Pandoc( { t } )
        local html = pandoc.write(doc, 'html')
        return pandoc.RawBlock('markdown', html)
      end
    end,
  }
}

[HeirOfNorton]

The filter shouldn't be necessary. Tables in Markdown are extensions, so you just have to disable the table extensions in the pandoc command and it should automatically convert tables to HTML. Like so: pandoc -t markdown-simple_tables-multiline_tables-grid_tables-pipe_tables -o output.md input.docx. It's kind of a long bit to add to the command line since you have to disable all four types of markdown tables, but it works.

[denismaier]

The filter and the -t markdown-simple_tables-multiline_tables-grid_tables-pipe_tables produce essentially the same result.

However, I've noticed one thing: both solutions write the tables like so:

Bla bla bla.

<table>
<colgroup>
<col style="width: 33%" />
<col style="width: 33%" />
<col style="width: 33%" />
</colgroup>
<thead>
<tr class="header">
<th>1</th>
<th>2</th>
<th>3</th>
</tr>
</thead>
<tbody>
<tr class="odd">
<td>4</td>
<td>5</td>
<td>6</td>
</tr>
</tbody>
</table>

Bla bla bla

I.e., there are no rawBlock markers, and without those the table will not be picked up by this filter:

function RawBlock (raw)
  return raw.format:match 'html'
    and pandoc.read(raw.text, 'html').blocks
    or raw
end

Further processing only works with this:

Bla bla bla.

~~~{=html}
<table>
<colgroup>
<col style="width: 33%" />
<col style="width: 33%" />
<col style="width: 33%" />
</colgroup>
<thead>
<tr class="header">
<th>1</th>
<th>2</th>
<th>3</th>
</tr>
</thead>
<tbody>
<tr class="odd">
<td>4</td>
<td>5</td>
<td>6</td>
</tr>
</tbody>
</table>
~~~

Bla bla bla

Is there a way to get these rawBlock attributes automatically?

[massifrg]

Try this:

return {
  {
    Table = function(t)
      if FORMAT == 'markdown' then
        local doc = pandoc.Pandoc( { t } )
        local html = pandoc.write(doc, 'html')
        return pandoc.List({
          pandoc.RawBlock('markdown', '~~~{=html}'),
          pandoc.RawBlock('markdown', html),
          pandoc.RawBlock('markdown', '~~~'),
        })
      end
    end,
  }
}

or this:

return {
  {
    Table = function(t)
      if FORMAT == 'markdown' then
        local doc = pandoc.Pandoc( { t } )
        local html = pandoc.write(doc, 'html')
        return pandoc.RawBlock('markdown', '~~~{=html}\n' .. html .. '\n~~~'),
      end
    end,
  }
}

[denismaier]

Great. That works.