Feature request `case_replace()`

[rossholmberg]

I don't know about others, but my main use for case_when and case_match is to selectively replace values in a column. For example:

Outline of the current situation (as I understand it)

df <- tibble(cola = letters[1:5])
df |>
    mutate(cola = case_match(cola, c("b", "e") ~ "changed", .default = cola))

This works fine, but I find I regularly wish the .default value didn't need to be specified (its default is NULL). For example when using as part of a piped series of functions:

df |>
    mutate(cola = cola |>
        toupper() |>
        case_match(c("B", "E") ~ "changed")
    )

This will replace all unspecified values with NA, which is not what I want, but it's difficult to pass the piped vector as the .default value. It can be done by specifying a function, but it's a bit cumbersome:

df |>
    mutate(cola = cola |>
        toupper() |>
        (\(x) case_match(x, c("B", "E") ~ "changed", .default = x))()
    )

Suggested solution

My suggestion is a new function case_replace. The same as case_match, except without the .default parameter, always specifying that parameter as the input vector (the use case is similar to base::replace, but that function requires specifying by index instead of value):

case_replace <- \(x, ...) case_match(.x = x, ..., .default = x)

With that, we can pipe a vector nicely:

df |>
    mutate(cola = cola |>
        toupper() |>
        case_replace(c("B", "E") ~ "changed")
    )

I realise I'm nitpicking here, and case_match is really close to what I need. I've just found this to be a useful thing for me, and if others have similar needs, maybe it would be worth adding as a separate function in dplyr::. I'd be happy to have a go at contributing a PR if the interest is there.

[rossholmberg]

I did search the issues for a similar thing, but just searching again now have come across an almost exact replica of what I'm asking for in this comment on another issue:
#7694 (comment)