Use mutate + across with case_when with a list of conditions

Question

I want to modify a list of columns. Each column has a list of conditions that need to be applied with a case_when.

I tried the following:

df <- tibble(value=c("a","b","c"))

# minimal example modifying only one column
case_when_exprs <- list("value" =  rlang::exprs(.data  == "a" ~ "A", TRUE ~ .data))

# if I were to apply the case_when conditions manually it works
df  %>% mutate(across(names(case_when_exprs), ~ case_when(.x  == "a" ~ "A", TRUE ~ .x)))

# but when trying to do it dynamically it fails
df  %>% mutate(across(names(case_when_exprs), ~ case_when(!!!case_when_exprs[cur_column()])))

but I'm getting this error:

Error in local_error_context(dots = dots, .index = i, mask = mask): promise already under evaluation: recursive default argument reference or earlier problems?

what am I doing wrong?

Answer 1

While I already encountered the error you mention myself I still have no clue what's the reason.

But one option to make your code work would be to use a wrapper around case_when and minor fix, i.e. we need to use [[ instead of [:

library(dplyr, warn=FALSE)

df <- tibble(value=c("a","b","c"))

case_when_exprs <- list("value" =  rlang::exprs(.data  == "a" ~ "A", TRUE ~ .data))

case_expr <- function(.data, expr) {
  case_when(!!!expr)  
}

df  %>% 
  mutate(
    across(
      names(case_when_exprs), 
      ~ case_expr(.x, case_when_exprs[[cur_column()]]))
  )
#> # A tibble: 3 × 1
#>   value
#>   <chr>
#> 1 A    
#> 2 b    
#> 3 c

Or as a slight variation:

case_expr2 <- function(.data, expr) {
  case_when(!!!case_when_exprs[[expr]])  
}

df  %>% 
  mutate(
    across(
      names(case_when_exprs), 
      ~ case_expr2(.x, cur_column()))
  )
#> # A tibble: 3 × 1
#>   value
#>   <chr>
#> 1 A    
#> 2 b    
#> 3 c