Implement count() method

Fixes #159

NAMESPACE

@@ -10,6 +10,7 @@ S3method(as_tibble,dtplyr_step)
 S3method(auto_copy,dtplyr_step)
 S3method(collect,dtplyr_step)
 S3method(compute,dtplyr_step)
+S3method(count,dtplyr_step)
 S3method(dim,dtplyr_step)
 S3method(dim,dtplyr_step_first)
 S3method(distinct,data.table)
@@ -87,6 +88,7 @@ importFrom(dplyr,arrange)
 importFrom(dplyr,auto_copy)
 importFrom(dplyr,collect)
 importFrom(dplyr,compute)
+importFrom(dplyr,count)
 importFrom(dplyr,distinct)
 importFrom(dplyr,do)
 importFrom(dplyr,full_join)

NEWS.md

@@ -1,5 +1,7 @@
 # dtplyr (development version)
 
+* Implement `count()` method for compatibility with dplyr 1.0.0 (#159).
+
 * Anonymous functions work in `mutate()` and `summarise()` (@smingerson, #155)
 
 * Environment variables used in the `i` argument of `[.data.table` are

R/dtplyr-package.R

@@ -6,4 +6,4 @@
 #' @export
 .datatable.aware <- TRUE
 
-globalVariables(".N")
+globalVariables(c(".N", "desc"))

R/step-subset.R

@@ -178,6 +178,37 @@ transmute.dtplyr_step <- function(.data, ...) {
   step_subset_j(.data, vars = names(dots), j = j)
 }
 
+#' @importFrom dplyr count
+#' @export
+count.dtplyr_step <- function(.data, ..., wt = NULL, sort = FALSE, name = NULL) {
+  if (!missing(...)) {
+    out <- group_by(.data, ..., .add = TRUE)
+  } else {
+    out <- .data
+  }
+
+  wt <- enexpr(wt)
+  if (is.null(wt)) {
+    n <- expr(n())
+  } else {
+    n <- expr(sum(!!wt, na.rm = TRUE))
+  }
+
+  if (is.null(name)) {
+    name <- "n"
+  } else if (!is_string(name)) {
+    abort("`name` must be a string")
+  }
+
+  out <- summarise(out, !!name := !!n)
+
+  if (sort) {
+    out <- arrange(out, desc(!!sym(name)))
+  }
+
+  out
+}
+
 # exported onLoad
 filter.dtplyr_step <- function(.data, ...) {
   dots <- capture_dots(.data, ..., .j = FALSE)

R/step.R

@@ -125,6 +125,7 @@ as.data.frame.dtplyr_step <- function(x, ...) {
 as_tibble.dtplyr_step <- function(x, ...) {
   out <- as_tibble(dt_eval(x))
   attr(out, ".internal.selfref") <- NULL
+  attr(out, "sorted") <- NULL
   out
 }
 

tests/testthat/_snaps/step-subset.md

@@ -0,0 +1,7 @@
+# can control name
+
+    Code
+      dt %>% count(name = 10) %>% collect()
+    Error <rlang_error>
+      `name` must be a string
+

tests/testthat/test-step-subset.R

@@ -169,6 +169,52 @@ test_that("if for unsupported resummarise", {
   expect_error(dt %>% summarise(x = mean(x), x2 = sd(x)), "mutate")
 })
 
+
+# count -------------------------------------------------------------------
+
+test_that("can be used grouped or ungrouped", {
+  dt <- lazy_dt(data.table(x = c(1, 1, 1, 2)), "DT")
+
+  expect_equal(
+    dt %>% count(x) %>% collect(),
+    tibble(x = c(1, 2), n = c(3, 1))
+  )
+  expect_equal(
+    dt %>% group_by(x) %>% count() %>% collect(),
+    tibble(x = c(1, 2), n = c(3, 1))
+  )
+})
+
+test_that("can control name", {
+  dt <- lazy_dt(data.table(x = c(1, 1, 1, 2)), "DT")
+
+  expect_equal(
+    dt %>% count(x, name = "y") %>% collect(),
+    tibble(x = c(1, 2), y = c(3, 1))
+  )
+  expect_snapshot(
+    dt %>% count(name = 10) %>% collect(),
+    error = TRUE
+  )
+})
+
+
+test_that("can weight", {
+  dt <- lazy_dt(data.table(x = c(1, 1, 2), y = c(1, 2, 10)), "DT")
+  expect_equal(
+    dt %>% count(x, wt = y) %>% collect(),
+    tibble(x = c(1, 2), n = c(3, 10))
+  )
+})
+
+test_that("can sort", {
+  dt <- lazy_dt(data.table(x = c(1, 1, 2), y = c(1, 2, 10)), "DT")
+  expect_equal(
+    dt %>% count(x, wt = y, sort = TRUE) %>% collect(),
+    tibble(x = c(2, 1), n = c(10, 3))
+  )
+})
+
 # select/rename ------------------------------------------------------------------
 
 test_that("renames grouping vars", {