table with long text, bullet points and specific table width - r

Question

I want a table to have bullet points in one column and to have a specific table width (in order to be placed on one page when rendered to PDF).
How can I achieve this in rmarkdown using one of the many packages out there?
What I have tried and have so far:
---
output: pdf_document
---
```{r, include = FALSE}
df <- data.frame(col1 = "Some really long text here. I mean some reeeeeaaly loooong text. So long, it should be wrapped. Really.",
col2 = "* bullet point 1\n * bullet point 2", col3 = "Yes, there is still another column.")
```
# Attempt 1: kableExtra
```{r, echo = FALSE, warning = FALSE}
library(kableExtra)
df1 <- df
df1$col2 <- linebreak(df1$col2)
knitr::kable(df1, escape = FALSE) %>% column_spec(1, width = "15em")
```
# Attempt 2: pander
```{r, echo = FALSE}
pander::pander(df, keep.line.breaks = TRUE, style = 'grid', justify = 'left')
```
This renders to:
As you can see both options have caveats. The kableExtra version does have a specific table width that fits on one page but does not show bullet points nicely. Whereas the pander solution renders the bullet points nicely but spans multiple pages, because I don't know how to specifiy the table width in pander.
Is there a solution that can do both?
Related questions are for example here and there.

Answer 1

An alternative solution based on kableExtra (optional: with footnotes)
This approach allows to add a caption, to manually add footnotes within the table, and to fix column widths.
To create the bullet point list:
"thicker" \cdots from LaTex are employed as bullets (alternatives see here)
linebreaks are forced with \n (indentation is therefore not as nice as with the pander approach from #daroczig).
MWE
library(stringi); library(kableExtra); library(dplyr)
string_short <- "Descriptor"
string_long <- substr(stri_rand_lipsum(1), 1, 50)
# add footnotes manually within table
string_bulletlist <- "$\\boldsymbol{\\cdot}$ bullet point 1: foo$^{a}$ \n $\\boldsymbol{\\cdot}$ bullet point 2: bar$^{b}$"
df <- data.frame(col1 = c(string_short, string_short),
col2 = c(string_bulletlist, string_bulletlist),
col3 = c(string_long, string_long)
)
col_names <- c("Descriptor", "Description with list", "Some comment")
# header: bold column names
colnames(df) <- paste0("\\textbf{", col_names,"}")
# add footnote with kableExtra commands
names(df)[1] <- paste0(names(df)[1], footnote_marker_symbol(1))
df %>%
mutate_all(linebreak) %>% # required for linebreaks to work
kable(
"latex",
escape = F,
booktabs=TRUE,
align = "l",
caption = 'kableTable with bullet list and footnote') %>%
# kable_styling(full_width = F) %>% # makes no difference here
footnote(general = "General comment of the table.",
alphabet = c("Footnote A;", "Footnote B;"),
symbol = c("Footnote Symbol 1")) %>%
column_spec(1, width = "5em") %>% # fix width column 1
column_spec(2, width = "10em") %>% # fix width column 2
column_spec(3, width = "15em") # fix width column 3
To [improve line spacing[(https://stackoverflow.com/questions/53794142/increase-line-row-spacing-with-kableextra), one can add the following before & after code chunk in the Rmd:
\renewcommand{\arraystretch}{1.5} <!-- increase line spacing for the table -->
RMD CHUNK HERE
\renewcommand{\arraystretch}{1} <!-- reset row hight/line spacing -->
Comment:
I also tried the pander approach from #daroczig and made the following experiences:
Pro: nice line spacing plus a "real bullet list" (compared to the $\boldsymbol{\cdot}$-workaround in the kableExtra approach).
Con: a large blank space is included after the bullet list
In addition, when using the pander approach in an Rmd file using the huskydown thesis template the footnotes greatly messed up the table alignment.

Answer 2

Use the split.table parameter of pandoc.table (that is being called by pander in the background) or disable the table splitting in general via panderOptions's table.split.table, eg
pander::pander(df, keep.line.breaks = TRUE, style = 'grid', justify = 'left', split.table = Inf)
or
library(pander)
panderOptions('table.style', 'grid')
panderOptions('table.alignment.default', 'left')
panderOptions('table.split.table', Inf)
panderOptions('keep.line.breaks', TRUE)
pander(df)