Add group_other option to fct_collapse, to allow for optional recodin…

…g of all non-targeted levels into an "Other" category (#160)

Fixes #100

NEWS.md

@@ -1,6 +1,8 @@
 # forcats 0.3.0.9000
 
 * `fct_relabel()` now accepts character input.
+* `fct_collapse()` gains a `group_other` argument to allow you to group all 
+  un-named levels into `"Other"`. (#100, @AmeliaMN)
 
 * `fct_cross()` creates a new factor containing the combined levels from two or more input factors, similar to `base::interaction` (@tslumley, #136)
 

R/collapse.R

@@ -3,6 +3,7 @@
 #' @param .f A factor (or character vector).
 #' @param ... A series of named character vectors. The levels in
 #'   each vector will be replaced with the name.
+#' @param group_other Replace all levels not named in `...` with "Other"?
 #' @export
 #' @examples
 #' fct_count(gss_cat$partyid)
@@ -15,10 +16,16 @@
 #'   dem = c("Not str democrat", "Strong democrat")
 #' )
 #' fct_count(partyid2)
-fct_collapse <- function(.f, ...) {
+fct_collapse <- function(.f, ..., group_other = FALSE) {
   new <- rlang::dots_list(...)
-
   levs <- as.list(unlist(new, use.names = FALSE))
+  if (group_other){
+    f <- check_factor(.f)
+    levels <- levels(f)
+    new[["Other"]] <- levels[!levels %in% levs]
+    levs <- levels
+  }
+
   names(levs) <- names(new)[rep(seq_along(new), vapply(new, length, integer(1)))]
 
   fct_recode(.f, !!!levs)

tests/testthat/test-collapse.R

@@ -21,3 +21,17 @@ test_that("can collapse missing levels", {
 
   expect_equal(f2, factor(c("x", "y")))
 })
+
+test_that("can collapse un-named levels to Other", {
+  f1 <- factor(letters[1:3])
+  f2 <- fct_collapse(f1, xy = c("a", "b"), group_other = TRUE)
+
+  expect_equal(f2, factor(c("xy", "xy", "Other"), levels = c("xy", "Other")))
+})
+
+test_that("does not automatically collapse unnamed levels to Other", {
+  f1 <- factor(letters[1:3])
+  f2 <- fct_collapse(f1, xy = c("a", "b"))
+
+  expect_equal(f2, factor(c("xy", "xy", "c"), levels = c("xy", "c")))
+})