Repeat gt column labels on each page-CodePudding

I am using

I can edit the TeX document, adding \endhead after the column header, and I get the result I want on the second (and subsequent) pages.

   158 \begin{longtable}{rr}
   159 \toprule
   160 id & random \\
   161 \midrule
   162 \endhead

Like this:

One option to add the \endhead would be to use a Pandoc filter to add the extra longtable command, but that sounds like a lot of effort for something so basic.

Is there a simple way with this workflow (Quarto options, pandoc, gt) to get column headers on each page in the PDF output?

CodePudding user response：

Your question is a topical issue.

So, I can suggest a next one solution.

As you know - we can represent our gt table as a LaTeX code:

Considering the place for the header - updating our code (adding \\toprule\nid & random \\\\ \n\\midrule between 42 and 43:

New code:

---
format:
  pdf:
    documentclass: article
    keep-tex: true
include-in-header:
- \usepackage{amsmath, booktabs, caption, longtable}
---


```{r, results='asis', echo=F}

k <- "\\captionsetup[table]{labelformat=empty,skip=1pt}\n\\begin{longtable}{rr}\n\\toprule\nid & random \\\\ \n\\midrule\n1 & 0.83452294 \\\\ \n2 & 0.43130524 \\\\ \n3 & 0.38702311 \\\\ \n4 & 0.75589167 \\\\ \n5 & 0.34432672 \\\\ \n6 & 0.75283676 \\\\ \n7 & 0.72770533 \\\\ \n8 & 0.20232485 \\\\ \n9 & 0.07900835 \\\\ \n10 & 0.80229610 \\\\ \n11 & 0.03800323 \\\\ \n12 & 0.03144921 \\\\ \n13 & 0.87364981 \\\\ \n14 & 0.47587592 \\\\ \n15 & 0.56736249 \\\\ \n16 & 0.83990532 \\\\ \n17 & 0.64138659 \\\\ \n18 & 0.76073451 \\\\ \n19 & 0.66013426 \\\\ \n20 & 0.54178832 \\\\ \n21 & 0.70443849 \\\\ \n22 & 0.81238956 \\\\ \n23 & 0.11305811 \\\\ \n24 & 0.12530444 \\\\ \n25 & 0.22497031 \\\\ \n26 & 0.35093560 \\\\ \n27 & 0.91268645 \\\\ \n28 & 0.19127469 \\\\ \n29 & 0.21638115 \\\\ \n30 & 0.24333619 \\\\ \n31 & 0.21178737 \\\\ \n32 & 0.50896038 \\\\ \n33 & 0.30918302 \\\\ \n34 & 0.44719810 \\\\ \n35 & 0.20445296 \\\\ \n36 & 0.67448123 \\\\ \n37 & 0.78824962 \\\\ \n38 & 0.62888684 \\\\ \n39 & 0.39583956 \\\\ \n40 & 0.99218439 \\\\ \n41 & 0.27330819 \\\\ \n42 & 0.20861772 \\\\ \\toprule\nid & random \\\\ \n\\midrule \n43 & 0.67226113 \\\\ \n44 & 0.11676375 \\\\ \n45 & 0.24479000 \\\\ \n46 & 0.28968861 \\\\ \n47 & 0.72271636 \\\\ \n48 & 0.58395987 \\\\ \n49 & 0.74914347 \\\\ \n50 & 0.32712809 \\\\ \n\\bottomrule\n\\end{longtable}\n"

knitr::asis_output(k)

```

The desired output:

So, the main idea of this method: first render of qmd-file -> detect the break -> manually add the header.

CodePudding user response：

Since we only need to add the \endhead after the \midrule, one option could be inserting \endhead after the \midrule (Using the answer from @manro),


---
title: "Random Table"
format:
  pdf:
    keep-tex: true
---

```{r}
#| echo: false
#| warning: false
#| results: asis

library(tidyverse)
library(gt)

random <- tibble(
  id = seq(1:50),
  random = runif(50)
)

t <- gt(random) %>%
  as_latex() %>%
  as.character()

knitr::asis_output(gsub("(.*\\\\begin\\{longtable\\}.*?\\midrule\\\n)", t, replacement = "\\1\\\\endhead\\\n"))
```

CodePudding user response：

I found a way to solve this problem for my documents using pandoc filters. I'm still hoping there is a more elegant solution or that this is a feature that could be added to gt, but here's what I did.

Add the filter to the pdf section of the YAML header:

format:
  pdf:
    filters:
      - endhead.lua

And here's the endhead.lua filter:

function RawBlock(elem)
    -- print(elem.text)
    return {
        pandoc.RawInline(
            "latex",
            string.gsub(
              elem.text, 
              "\\toprule\n(.-)\\midrule\n",
              "\\toprule\n%1\\midrule\n\\endhead\n"
            )
        ),
    }
end

gt within an R code block produces a RawBlock element in Pandoc's AST, so that's what needs to be changed using the filter. You can figure out the element you need to manipulate by adding keep-md: true to the YAML header of your Quarto doc, then after rendering, do:

$ pandoc -s -t native tables.md

That will show you the AST.

Once you know the Pandoc element, it's just a question of figuring out the regular expression that will add an \endhead element to the proper section of the table. I did this by finding the first \midrule after \toprule in each RawBlock (gt table). It took a bit for me to figure out how to use non-greedy capturing ((.-)) in order to only add \endhead once, right after the header of the table.