Port plyr::mapvalues() equivalent into dplyr

[aloboa]

ref. #4310
(cannot write into that issue because it was blocked).

The currently suggested alternatives for plyr::mapvalues() in dplyr are inconvenient.
In the case of

dplyr::recode(z, `1`  = 10, `5` = 50, `9` = 90)

the solution is unbearable as soon as the "from" and "to" vectors are longer than 3 elements.

In the case of

recode(z, !!! setNames(to, from))

the code is just inextricable.

It would be so much friendlier just having a function dplyr::mapvalues()
as a wrapper for the last solution, i.e. as suggested in
https://stackoverflow.com/questions/62216940/dplyr-alternative-for-plyrmapvalues-recode-using-dictionary

While implementing that solution as a personal function is easy, using personal functions for
such a general tool is not a good idea, as R scripts would easily become unintelligible to other people.

[romainfrancois]

We are in the process of designing a better recode() in the funs package. We're not sure at this point what it will be.

[aloboa]

I'm confused, do we have an equivalent to plyr::mapvalues() in dplyr or in any other tidyverse package?
If you have closed this, I understand it should exist, but cannot find it.

[hadley]

The new case_switch() is a good equivalent.

[thomasjwood]

Is case_switch() documented somewhere?

[olivroy]

Probably referring to case_match().

[aloboa]

Could you include as an example the equivalent to:

z <- c(1, 4, 5, 9)
dict <- list(from=c(1,5,9), to=c("a","x","r"))
z.new <- mapvalues(z, from = dict$from to = dict$to)

which is what is really needed by users?
Please note that being able to include an object such as dict within case_match() is much needed.
Also, I would prefer a devoted, standard "recode()" function for this specific, very usual, application of case_match()