`fct()` doesn't work on factors

[DanChaltiel]

Hi,

Like its old cousin factor(), it would be nice if fct() could be useful on factors too. For instance, I would expect this to work with fct() (although I know I could do it with fct_expand()):

factor("a", levels=c("a", "b"))
#> [1] a
#> Levels: a b
factor("a", levels=c("a", "b")) %>% factor(levels=c("a", "b", "c"))
#> [1] a
#> Levels: a b c

In fact, when dealing with tables containing character and factor columns, it can happen to be very useful to be able to use fct() on both indifferently.

However, using fct() on factors will throw an error (unlike factor()):

a = factor("a")
factor(a)
#> [1] a
#> Levels: a
forcats::fct(a)
#> Error in `forcats::fct()`:
#> ! `x` must be a character vector

^{Created on 2022-12-02 with reprex v2.0.2}

Even on factors, fct() would keep its loved advantages over factor(), such as preserving attributes.

[hadley]

I think this is probably a good idea, but I don't think it would preserve attributes; fct() is a constructor not a coercer.

[hadley]

I've changed my mind: I think fct() is basically a constructor function; and it should only work with character vectors.