How to include header in table for docx format?

[okenk]

Description

Hi. Apologies if the answer to this exists, but I could not find it. I am trying to create a table in a document that compiles to a .docx format. I would like two of the columns to have a little header above them.

First, I tried using gt::tab_spanner(). Even a gt table without a spanner was causing headaches, which I figured out is a known issue that does not seem to have been resolved (#12022).

There are an overwhelming number of packages that will format tables, so then I moved along to {kableExtra}. A table produced using knitr::kable() works fine, but once I started trying to use kableExtra::add_header_above() I got the following error:

Error: Functions that produce HTML output found in document targeting docx output.
Please change the output type of this document to HTML.
If you're aiming to have some HTML widgets shown in non-HTML format as a screenshot,
please install webshot or webshot2 R package for knitr to do the screenshot, and configure it by looking at its documentation.

Is there a trick to getting {kableExtra} to work, or is there a different package that will produce this desired result?

Example:

---
format: 
  html: default
  docx: default
---

 ```{r}
data.frame(apples = 1, bananas = 2, carrots = 3) |>
  knitr::kable() |>
  kableExtra::add_header_above(c("fruits" = 2, " " = 1))

[Steinthal]

Try flextable:

---
format: 
  html: default
  docx: default
---

```{r}
#| echo: false
#| label: output
data.frame(apples = 1, bananas = 2, carrots = 3) |>
  flextable::flextable() |>
  flextable::add_header_row( values=c("fruits", " "), colwidths = c(2,1)) |>
  flextable::align(align = "center", part = "all")
```

[cderv]

This is currently a limitation from knitr and how kableExtra works, from back then in the R Markdown context.

Basically what is happening ?

kableExtra is producing an HTML table. However, you target format: docx. So, an HTML table for a Docx document is not supposed to work.
The above answer about using flextable #13099 (comment) will work, because flextable will output raw openxml when you target a docx document, not an HTML table.

So why am I saying this is a limitation ?

Because Quarto has a specific feature of support HTML table, by parsing them to convert to a Table AST element for Pandoc, which makes it possibly to output a Table in the targeted output.

knitr or kableExtra currently does not take this into account, and kableExtra does try to include HTML dependencies in Quarto docx output. So seeing a HTML output with HTML deps from a cell inside a non-HTML target format leads to this issue (unless prefer-html is set).

To make it work, something like this could be used

---
format: 
  docx: default
keep-md: true
---

```{r}
tab <- data.frame(apples = 1, bananas = 2, carrots = 3) |>
  knitr::kable() |>
  kableExtra::add_header_above(c("fruits" = 2, " " = 1))
```

```{r}
#| output: asis
#| echo: false
if (knitr::pandoc_to("docx")) {
  as.character(tab) |> knitr::raw_html()
}
```

• In docx output, the tab is not output as such, but some HTML code is produced as character, and inserted as raw inside the document (using output: asis)

First, I tried using gt::tab_spanner(). Even a gt table without a spanner was causing headaches, which I figured out is a known issue that does not seem to have been resolved (#12022).

I am surprised by this not working. This is working for me

---
format: 
  docx: default
keep-md: true
---

```{r}
library(gt)
towny |>
  dplyr::slice_max(population_2021, n = 5) |>
  dplyr::select(
    name, latitude, longitude,
    ends_with("2016"), ends_with("2021")
  ) |>
  gt() |>
  tab_spanner(
    label = "Population",
    columns = starts_with("pop")
  ) |>
  tab_spanner(
    label = "Density",
    columns = starts_with("den")
  ) |>
  tab_spanner(
    label = md("*Location*"),
    columns = ends_with("itude"),
    id = "loc"
  )
```

So maybe another issue with gt and Quarto due to other stuff than tab_spanner

Hope it helps understand.

[cderv]

I have opened

• Consider different behavior in Quarto context for HTML tables rstudio/rmarkdown#2598
  to reflect on that

[okenk]

Thank you. flextable almost worked, but I was hoping to include some latex (sorry, my minimal example was too minimal), which it looks like is not possible?

[Steinthal]

Could you please elaborate why or what LaTeX you'd like or need to include?

[okenk]

Just some simple math mode text. For example:

---
format: 
  html: default
  docx: default
---

 ```{r}
data.frame(apples = $L_{A_\\min}$, bananas = $\\sigma_R$, carrots = 3) |>
  flextable::flextable |>
  flextable::add_header_row(values = c("fruits", " "), colwidths = c(2, 1))

I also tried with $$ instead of $ but no dice.

[Steinthal]

Is this what you have in mind?

---
format: 
  html: default
  docx: default
---

```{r}
#| echo: false
#| label: r-libraries
library(flextable)
library(equatags)
```

```{r}
#| echo: false
#| label: output
data.frame(apples = "L_{A\\min_{2}}", bananas = "\\sigma_R", carrots = 3) |>
  flextable() |>
  compose(j = "apples", value = as_paragraph(as_equation(apples))) |>
  compose(j = "bananas", value = as_paragraph(as_equation(bananas))) |>
  add_header_row( values=c("fruits", " "), colwidths = c(2,1)) |>
  align(align = "center", part = "all")
```

[okenk]

Ah, I actually needed to combine some math and non-math text (e.g., "the standard deviation is $\sigma_R$"), but I figured out that it just requires a few arguments to as_paragraph().

Thank you!

[cderv]

@Steinthal Thanks a lot for sharing the flextable way !