gtsummary table with difference doesn't work in typst, but is fine in html

[aghaynes]

Description

I think I may have found a bug when using gtsummary tables with differences, which seems to be a lua filter issue.

Consider the following file including two tables, the only difference being the use of add_difference.

---
title: "Untitled"
format: typst
editor: visual
---

```{r}
data(mtcars)

library(gtsummary)

mtcars |> 
  select(mpg, am) |> 
  tbl_summary(by = am)
```

```{r}
mtcars |> 
  select(mpg, am) |> 
  tbl_summary(by = am) |> 
  add_difference()
```

When rendering to html, both tables are included (considering that typst is still relatively new, i often try html as a reference when something doesn't work). But when rendering to typst, only the first table is included, with the following message from quarto:

WARNING (C:/Users/haynes/AppData/Local/Programs/Quarto/share/filters/main.lua:9521) Unable to parse table from raw html block: skipping.

Actually, the warning is produced when rendering to both HTML and typst, but the table is included anyway in the case of HTML... (the fact that the warning is produced under both formats suggests that there might also be an issue on the gtsummary or gt side I suppose, but I'm posting here as the table clearly works in HTML...)

quarto check
Quarto 1.7.31
[>] Checking environment information...
      Quarto cache location: C:\Users\haynes\AppData\Local\quarto
[>] Checking versions of quarto binary dependencies...
      Pandoc version 3.6.3: OK
      Dart Sass version 1.85.1: OK
      Deno version 1.46.3: OK
      Typst version 0.13.0: OK
[>] Checking versions of quarto dependencies......OK
[>] Checking Quarto installation......OK
      Version: 1.7.31
      Path: C:\Users\haynes\AppData\Local\Programs\Quarto\bin
      CodePage: 1252

[>] Checking tools....................OK
      TinyTeX: (external install)
      Chromium: (not installed)

[>] Checking LaTeX....................OK
      Using: TinyTex
      Path: C:\Users\haynes\AppData\Roaming\TinyTeX\bin\windows\
      Version: 2025

[>] Checking Chrome Headless....................OK
      Using: Chrome found on system
      Path: C:\Program Files\Google\Chrome\Application\chrome.exe
      Source: Windows Registry

[>] Checking basic markdown render....OK

[>] Checking Python 3 installation....(None)

      Unable to locate an installed version of Python 3.
      Install Python 3 from https://www.python.org/downloads/

[>] Checking R installation...........OK
      Version: 4.5.0
      Path: C:/Users/haynes/AppData/Local/Programs/R/R-4.5.0
      LibPaths:
        - C:/Users/haynes/AppData/Local/Programs/R/R-4.5.0/library
      knitr: 1.50
      rmarkdown: 2.29

[>] Checking Knitr engine render......OK

[cderv]

This thread in gt will explain a little the support:

• typst can't see/use/output gt objects if quarto.disable_processing = TRUE rstudio/gt#2006

gt does not yet support Typst output natively

• Add as_typst() function to support styled tables in typst Quarto output format rstudio/gt#1500

So it will produce HTML table, and then it is Quarto that does handle the transformation from HTML table to a Table AST node that can be written to Typst (https://quarto.org/docs/authoring/tables.html#html-tables)

However, this assumes the HTML table produced can be correctly parsed using pandoc.read()

It seems that the second HTML table is not parseable by Quarto. So, it is skipped as raw HTML does not show in Typst output, and the table can't be converted.

The next steps are:

• Which is the HTML that causes trouble ?
• Is this HTML valid ? So is it a limitation of Pandoc HTML reader, or Quarto processing ?
• Who is emitting this HTML table ? Can it be adapted.

Thanks for the report!

[cderv]

So issue is that there is two tfoot produced

  <tfoot class="gt_sourcenotes">
    <tr>
      <td class="gt_sourcenote" colspan="6"><span data-qmd-base64="QWJicmV2aWF0aW9uOiBDSSA9IENvbmZpZGVuY2UgSW50ZXJ2YWw="><span class='gt_from_md'>Abbreviation: CI = Confidence Interval</span></span></td>
    </tr>
  </tfoot>
  <tfoot class="gt_footnotes">
    <tr>
      <td class="gt_footnote" colspan="6"><span class="gt_footnote_marks" style="white-space:nowrap;font-style:italic;font-weight:normal;line-height:0;"><sup>1</sup></span> <span data-qmd-base64="TWVkaWFuIChRMSwgUTMp"><span class='gt_from_md'>Median (Q1, Q3)</span></span></td>
    </tr>
  </tfoot>

It seems from specification that a single one needs to be present
From https://html.spec.whatwg.org/multipage/tables.html#the-tfoot-element

In this order: optionally a caption element, followed by zero or more colgroup elements, followed optionally by a thead element, followed by either zero or more tbody elements or one or more tr elements, followed optionally by a tfoot element, optionally intermixed with one or more script-supporting elements.

And pandoc respects that. Here is a minimal HTML table that won't be parsed as table by Pandoc because two tfoot

<table>
  <thead>
    <tr>
      <th>Characteristic</th>
    </tr>
  </thead>
  <tbody>
    <tr>
      <td>mpg</td>
    </tr>
  </tbody>
  <tfoot>
    <tr>
      <td colspan="3">Footnote: Median (Q1, Q3)</td>
    </tr>
  </tfoot>
  <tfoot>
    <tr>
      <td colspan="3">Source note: CI = Confidence Interval</td>
    </tr>
  </tfoot>
</table>

So they need to be merged in one tfoot element.

Probably a gtsummary issue.

[cderv]

Reported upstream:

• Bug Report: HTML table produced failed to be parsed by Quarto because of multiple tfoot ddsjoberg/gtsummary#2276

[aghaynes]

Thanks for investigating @cderv!

[cderv]

This is in fact known issue in gt:

• Using tab_source_note and tab_footnote together in quarto distorts table rstudio/gt#1296