How can I use tidy selection on variables created with `{{ }}`?

January 12, 2024

I create a new variable in a custom function using the {{ }} operator, but I can’t figure out how to use this newly created variable in a tidy select context, as in the ... of separate_rows():

library(dplyr, warn.conflicts = FALSE)
library(tidyr)

foo <- function(col) {
  head(iris) |>
    select({{ col }}) |> 
    mutate("{{ col }}_suffixed" := paste({{ col }}, "SUF")) |> 
    separate_rows(all_of("{{ col }}_suffixed"), sep = " ")
}

foo(Species)
#> Error in `all_of()`:
#> ! Can't subset columns that don't exist.
#> ✖ Column `{{ col }}_suffixed` doesn't exist.

What am I missing?

>Solution :

In this particular case, one could use contains('_suffixed'), but that lacks generality, so doesn’t really answer the question.

I always wondered why anyone would need rlang::englue, but this seems a perfect use case:

foo <- function(col) {
  head(iris) |>
    select({{ col }}) |> 
    mutate("{{ col }}_suffixed" := paste({{ col }}, "SUF")) |> 
    separate_rows(all_of(rlang::englue('{{col}}_suffixed')), sep = " ")
}

foo(Species)
#> # A tibble: 12 x 2
#>    Species Species_suffixed
#>    <fct>   <chr>           
#>  1 setosa  setosa          
#>  2 setosa  SUF             
#>  3 setosa  setosa          
#>  4 setosa  SUF             
#>  5 setosa  setosa          
#>  6 setosa  SUF             
#>  7 setosa  setosa          
#>  8 setosa  SUF             
#>  9 setosa  setosa          
#> 10 setosa  SUF             
#> 11 setosa  setosa          
#> 12 setosa  SUF