Behavior of str_remove when called with NA_character_ pattern

[davidciani]

When called with NA as a pattern, str_remove fails loudly about a type mismatch. When called with NA_character_ as a pattern, it silently returns NA. Is this the expected behavior? If so it doesn't seem to be documented anywhere. Personally, I was expecting that str_remove would have returned the input string unmodified in this scenario (as there is nothing specified to remove), or alternatively failed with an error message.

library(stringr)

str_remove("test", NA)
#> Error in `str_replace()`:
#> ! `pattern` must be a string, not `NA`.

str_remove("test", NA_character_)
#> [1] NA

# What lead to discovery of this, though str_c is behaving as expected returning NA_character_

string_to_remove = NA # a blank cell in a data frame

str_remove_all("test", str_c("\\b",string_to_remove,"\\b"))
#> [1] NA

^{Created on 2024-04-11 with reprex v2.1.0}

[hadley]

Hmmm, looks like type.character() should probably error on this. But we'd need to double check what the docs for stringr says just to make sure there's no reason for anyone to be relying on this behavkour.