Wrapper for stringr functions that keeps factor levels.

[orgadish]

This is related to stringr issue #394. I'm not sure if this is best handled in forcats or stringr.

I often have factor data that I want to rework slightly for a plot. When I use something like str_to_upper or str_replace to do so, I lose the factor levels. However, this should be pretty straightforward to do because you just want to apply the function to the levels and the values.

I created a wrapper function for myself fct_stringr to do this and I think it could be useful for others.

library(tidyverse)

iris_with_reordered_levels <- iris %>% 
  mutate(across(Species, fct_relevel, "virginica"))

iris_with_reordered_levels %>% 
  pull(Species) %>% 
  levels    
#> [1] "virginica"  "setosa"     "versicolor"

iris_with_reordered_levels %>% 
  mutate(across(Species, str_to_upper)) %>% 
  pull(Species) %>% 
  levels    
#> NULL

fct_stringr <- function(.f, .fn, ...) {
  new_levels <- .fn(levels(.f), ...)
  new_str <- .fn(.f, ...)
  factor(new_str, levels=new_levels)
}

iris_with_reordered_levels %>% 
  mutate(across(Species, fct_stringr, str_to_upper)) %>% 
  pull(Species) %>% 
  levels
#> [1] "VIRGINICA"  "SETOSA"     "VERSICOLOR"

^{Created on 2021-09-30 by the reprex package (v2.0.1)}

[orgadish]

I had created a similar issue in stringr to suggest something like this and @hadley made the excellent point that forcats::fct_relabel already does this:

forcats::fct_relabel(head(iris$Species), stringr::str_to_upper)
#> [1] SETOSA SETOSA SETOSA SETOSA SETOSA SETOSA
#> Levels: SETOSA VERSICOLOR VIRGINICA

^{Created on 2021-11-29 by the reprex package (v2.0.1)}

As such this feature is already developed! Though it might be useful at some point to add this use case more explicitly to the help or vignette, as it's something I come across often.