extra NAs with `droplevels = FALSE`

[d-morrison]

When I use droplevels = FALSE with table1.formula(), I'm getting extra NAs on the variable name lines. For example:

 iris |> 
   dplyr::filter(Species != "setosa") |>
   table1::table1(data = _, x =  ~ . | Species, droplevels = FALSE)

produces:

Probably an unusual case, but a colleague asked me for a table with an all-NA column like this. Is there a way to remove the extra NAs (on lines 1, 4, ,7, and 10)?

[benjaminrich]

I agree that this is not the expected behavior, so I'll look into fixing it. For now, a quick workaround would be:

rndr <- function(x, ...) {
    if (is.numeric(x) & length(x)==0) {
        return((function(y) ifelse(y=="", "", "NA"))(table1::render.default(0, ...)))
    }
    table1::render.default(x, ...)
}

iris |> 
  dplyr::filter(Species != "setosa") |>
  table1::table1(data = _, x =  ~ . | Species, droplevels = FALSE, render=rndr)