Merge pull request tidyverse#4100 from tidyverse/distinct_prepare

export distinct_prepare() (previously known as distinct_vars

NAMESPACE

@@ -270,6 +270,7 @@ export(distinct_)
 export(distinct_all)
 export(distinct_at)
 export(distinct_if)
+export(distinct_prepare)
 export(do)
 export(do_)
 export(dr_dplyr)

NEWS.md

@@ -76,6 +76,9 @@
     group_map(~ head(.x, 2L))
   ```
 
+* `distinct_prepare()`, previously known as `distinct_vars()` is exported. This is mostly useful for
+  alternative backends (e.g. `dbplyr`). 
+
 ## Major changes
 
 * `group_by()` respects levels of factors and keeps empty groups (#341). 

R/dataframe.R

@@ -210,7 +210,7 @@ reconstruct_set <- function(out, x) {
 
 #' @export
 distinct.data.frame <- function(.data, ..., .keep_all = FALSE) {
-  dist <- distinct_vars(.data, quos(...), .keep_all = .keep_all)
+  dist <- distinct_prepare(.data, quos(...), .keep_all = .keep_all)
   vars <- match_vars(dist$vars, dist$data)
   keep <- match_vars(dist$keep, dist$data)
   distinct_impl(dist$data, vars, keep, environment())

R/distinct.R

@@ -74,10 +74,11 @@ distinct_ <- function(.data, ..., .dots, .keep_all = FALSE) {
   UseMethod("distinct_")
 }
 
-#' Same basic philosophy as group_by: lazy_dots comes in, list of data and
+#' Same basic philosophy as group_by_prepare(): lazy_dots comes in, list of data and
 #' vars (character vector) comes out.
-#' @noRd
-distinct_vars <- function(.data, vars, group_vars = character(), .keep_all = FALSE) {
+#' @rdname group_by_prepare
+#' @export
+distinct_prepare <- function(.data, vars, group_vars = character(), .keep_all = FALSE) {
   stopifnot(is_quosures(vars), is.character(group_vars))
 
   # If no input, keep all variables

R/group-by.r

@@ -119,9 +119,9 @@ ungroup <- function(x, ...) {
 
 #' Prepare for grouping.
 #'
-#' Performs standard operations that should happen before individual methods
-#' process the data. This includes mutating the tbl to add new grouping columns
-#' and updating the groups (based on add)
+#' `*_prepare()` performs standard manipulation that is needed prior
+#' to actual data processing. They are only be needed by packages
+#' that implement dplyr backends.
 #'
 #' @return A list
 #'   \item{data}{Modified tbl}

R/grouped-df.r

@@ -349,7 +349,7 @@ do_.grouped_df <- function(.data, ..., env = caller_env(), .dots = list()) {
 
 #' @export
 distinct.grouped_df <- function(.data, ..., .keep_all = FALSE) {
-  dist <- distinct_vars(
+  dist <- distinct_prepare(
     .data,
     vars = quos(...),
     group_vars = group_vars(.data),