group_size

NAMESPACE

@@ -19,6 +19,7 @@ S3method(group_data,duckplyr_df)
 S3method(group_keys,duckplyr_df)
 S3method(group_map,duckplyr_df)
 S3method(group_modify,duckplyr_df)
+S3method(group_size,duckplyr_df)
 S3method(group_vars,duckplyr_df)
 S3method(head,duckplyr_df)
 S3method(inner_join,duckplyr_df)

R/group_size.R

@@ -0,0 +1,32 @@
+# Generated by 02-duckplyr_df-methods.R
+#' @export
+group_size.duckplyr_df <- function(x) {
+  # Our implementation
+  rel_try(
+    # Always fall back to dplyr
+    "No relational implementation for group_size()" = TRUE,
+    {
+      return(out)
+    }
+  )
+
+  # dplyr forward
+  group_size <- dplyr$group_size.data.frame
+  out <- group_size(x)
+  return(out)
+
+  # dplyr implementation
+  lengths(group_rows(x))
+}
+
+duckplyr_group_size <- function(x, ...) {
+  try_fetch(
+    x <- as_duckplyr_df(x),
+    error = function(e) {
+      testthat::skip(conditionMessage(e))
+    }
+  )
+  out <- group_size(x, ...)
+  class(out) <- setdiff(class(out), "duckplyr_df")
+  out
+}

R/overwrite.R

@@ -19,6 +19,7 @@ methods_overwrite <- function() {
   vctrs::s3_register("dplyr::group_keys", "data.frame", group_keys.duckplyr_df)
   vctrs::s3_register("dplyr::group_map", "data.frame", group_map.duckplyr_df)
   vctrs::s3_register("dplyr::group_modify", "data.frame", group_modify.duckplyr_df)
+  vctrs::s3_register("dplyr::group_size", "data.frame", group_size.duckplyr_df)
   vctrs::s3_register("dplyr::group_vars", "data.frame", group_vars.duckplyr_df)
   vctrs::s3_register("dplyr::inner_join", "data.frame", inner_join.duckplyr_df)
   vctrs::s3_register("dplyr::intersect", "data.frame", intersect.duckplyr_df)

R/restore.R

@@ -19,6 +19,7 @@ methods_restore <- function() {
   vctrs::s3_register("dplyr::group_keys", "data.frame", dplyr$group_keys.data.frame)
   vctrs::s3_register("dplyr::group_map", "data.frame", dplyr$group_map.data.frame)
   vctrs::s3_register("dplyr::group_modify", "data.frame", dplyr$group_modify.data.frame)
+  vctrs::s3_register("dplyr::group_size", "data.frame", dplyr$group_size.data.frame)
   vctrs::s3_register("dplyr::group_vars", "data.frame", dplyr$group_vars.data.frame)
   vctrs::s3_register("dplyr::inner_join", "data.frame", dplyr$inner_join.data.frame)
   vctrs::s3_register("dplyr::intersect", "data.frame", dplyr$intersect.data.frame)

dplyr-methods/group_size.txt

@@ -0,0 +1,3 @@
+group_size.data.frame <- function(x) {
+  lengths(group_rows(x))
+}

tests/testthat/test-arrange.R

@@ -334,8 +334,8 @@ test_that("arrange keeps zero length groups",{
   )
   df <- duckplyr_group_by(df, e, f, g, .drop = FALSE)
 
-  expect_equal( group_size(duckplyr_arrange(df)), c(2, 2, 0) )
-  expect_equal( group_size(duckplyr_arrange(df, x)), c(2, 2, 0) )
+  expect_equal( duckplyr_group_size(duckplyr_arrange(df)), c(2, 2, 0) )
+  expect_equal( duckplyr_group_size(duckplyr_arrange(df, x)), c(2, 2, 0) )
 })
 
 # legacy --------------------------------------------------------------

tests/testthat/test-as_duckplyr_df.R

@@ -818,6 +818,22 @@ test_that("as_duckplyr_df() and group_modify(~ .x)", {
   expect_equal(pre, post)
 })
 
+test_that("as_duckplyr_df() and group_size()", {
+  withr::local_envvar(DUCKPLYR_FORCE = "FALSE")
+
+  skip("Special")
+
+  # Data
+  test_df <- data.frame(a = 1:6 + 0, b = 2, g = rep(1:3, 1:3))
+
+  # Run
+  pre <- test_df %>% as_duckplyr_df() %>% group_size()
+  post <- test_df %>% group_size() %>% as_duckplyr_df()
+
+  # Compare
+  expect_equal(pre, post)
+})
+
 test_that("as_duckplyr_df() and group_vars()", {
   withr::local_envvar(DUCKPLYR_FALLBACK_FORCE = "TRUE")
 

tests/testthat/test-filter.R

@@ -618,7 +618,7 @@ test_that("filter keeps zero length groups", {
   )
   df <- duckplyr_group_by(df, e, f, g, .drop = FALSE)
 
-  expect_equal(group_size(duckplyr_filter(df, f == 1)), c(2, 0, 0) )
+  expect_equal(duckplyr_group_size(duckplyr_filter(df, f == 1)), c(2, 0, 0) )
 })
 
 test_that("filtering retains labels for zero length groups", {

tests/testthat/test-join.R

@@ -647,23 +647,23 @@ test_that("joins respect zero length groups", {
   df2 <- tibble(f = factor( c(2,2,3,3), levels = 1:3), y = c(1,2,3,4)) %>%
     duckplyr_group_by(f)
 
-  expect_equal(group_size(duckplyr_left_join( df1, df2, by = "f", relationship = "many-to-many")),  c(2,4))
-  expect_equal(group_size(duckplyr_right_join( df1, df2, by = "f", relationship = "many-to-many")),  c(4,2))
-  expect_equal(group_size(duckplyr_full_join( df1, df2, by = "f", relationship = "many-to-many")),  c(2,4,2))
-  expect_equal(group_size(duckplyr_anti_join( df1, df2, by = "f")),  c(2))
-  expect_equal(group_size(duckplyr_inner_join( df1, df2, by = "f", relationship = "many-to-many")),  c(4))
+  expect_equal(duckplyr_group_size(duckplyr_left_join( df1, df2, by = "f", relationship = "many-to-many")),  c(2,4))
+  expect_equal(duckplyr_group_size(duckplyr_right_join( df1, df2, by = "f", relationship = "many-to-many")),  c(4,2))
+  expect_equal(duckplyr_group_size(duckplyr_full_join( df1, df2, by = "f", relationship = "many-to-many")),  c(2,4,2))
+  expect_equal(duckplyr_group_size(duckplyr_anti_join( df1, df2, by = "f")),  c(2))
+  expect_equal(duckplyr_group_size(duckplyr_inner_join( df1, df2, by = "f", relationship = "many-to-many")),  c(4))
 
 
   df1 <- tibble(f = factor( c(1,1,2,2), levels = 1:3), x = c(1,2,1,4)) %>%
     duckplyr_group_by(f, .drop = FALSE)
   df2 <- tibble(f = factor( c(2,2,3,3), levels = 1:3), y = c(1,2,3,4)) %>%
     duckplyr_group_by(f, .drop = FALSE)
 
-  expect_equal(group_size(duckplyr_left_join( df1, df2, by = "f", relationship = "many-to-many")),  c(2,4,0))
-  expect_equal(group_size(duckplyr_right_join( df1, df2, by = "f", relationship = "many-to-many")),  c(0,4,2))
-  expect_equal(group_size(duckplyr_full_join( df1, df2, by = "f", relationship = "many-to-many")),  c(2,4,2))
-  expect_equal(group_size(duckplyr_anti_join( df1, df2, by = "f")),  c(2,0,0))
-  expect_equal(group_size(duckplyr_inner_join( df1, df2, by = "f", relationship = "many-to-many")),  c(0,4,0))
+  expect_equal(duckplyr_group_size(duckplyr_left_join( df1, df2, by = "f", relationship = "many-to-many")),  c(2,4,0))
+  expect_equal(duckplyr_group_size(duckplyr_right_join( df1, df2, by = "f", relationship = "many-to-many")),  c(0,4,2))
+  expect_equal(duckplyr_group_size(duckplyr_full_join( df1, df2, by = "f", relationship = "many-to-many")),  c(2,4,2))
+  expect_equal(duckplyr_group_size(duckplyr_anti_join( df1, df2, by = "f")),  c(2,0,0))
+  expect_equal(duckplyr_group_size(duckplyr_inner_join( df1, df2, by = "f", relationship = "many-to-many")),  c(0,4,0))
 })
 
 test_that("group column names reflect renamed duplicate columns (#2330)", {

tests/testthat/test-mutate.R

@@ -423,7 +423,7 @@ test_that("mutate keeps zero length groups", {
   )
   df <- duckplyr_group_by(df, e, f, g, .drop = FALSE)
 
-  expect_equal( group_size(duckplyr_mutate(df, z = 2)), c(2, 2, 0) )
+  expect_equal( duckplyr_group_size(duckplyr_mutate(df, z = 2)), c(2, 2, 0) )
 })
 
 # other -------------------------------------------------------------------

tests/testthat/test-slice.R

@@ -120,7 +120,7 @@ test_that("slice keeps zero length groups", {
   )
   df <- duckplyr_group_by(df, e, f, g, .drop = FALSE)
 
-  expect_equal(group_size(duckplyr_slice(df, 1)), c(1, 1, 0) )
+  expect_equal(duckplyr_group_size(duckplyr_slice(df, 1)), c(1, 1, 0) )
 })
 
 test_that("slicing retains labels for zero length groups", {

tools/00-funs.R

@@ -16,7 +16,7 @@ df_methods <-
   filter(!grepl("_$|^as[.]tbl$", name)) %>%
   # special dplyr methods, won't implement
   filter(!(name %in% c(
-    "group_indices", "group_nest", "group_size", "group_split", "group_trim", "groups", "n_groups",
+    "group_indices", "group_nest", "group_split", "group_trim", "groups", "n_groups",
     "same_src", # data frames can be copied into duck-frames with zero cost
     NULL
   ))) %>%
@@ -29,6 +29,7 @@ df_methods <-
     "group_keys",
     "group_map",
     "group_modify",
+    "group_size",
     "rowwise",
     NULL
   ))) %>%
@@ -700,6 +701,7 @@ test_skip_map <- c(
   group_map = "WAT",
   group_modify = "Grouped",
   group_nest = "Always returns tibble",
+  group_size = "Special",
   group_split = "WAT",
   group_trim = "Grouped",
   nest_by = "WAT",