group_data

NAMESPACE

@@ -15,6 +15,7 @@ S3method(dplyr_row_slice,duckplyr_df)
 S3method(explain,duckplyr_df)
 S3method(full_join,duckplyr_df)
 S3method(group_by,duckplyr_df)
+S3method(group_data,duckplyr_df)
 S3method(group_vars,duckplyr_df)
 S3method(head,duckplyr_df)
 S3method(inner_join,duckplyr_df)
@@ -330,6 +331,7 @@ importFrom(tidyselect,everything)
 importFrom(utils,globalVariables)
 importFrom(utils,head)
 importFrom(vctrs,new_data_frame)
+importFrom(vctrs,new_list_of)
 importFrom(vctrs,new_rcrd)
 importFrom(vctrs,unspecified)
 importFrom(vctrs,vec_as_names)

R/duckplyr-package.R

@@ -22,6 +22,7 @@
 #' @importFrom utils globalVariables
 #' @importFrom utils head
 #' @importFrom vctrs new_data_frame
+#' @importFrom vctrs new_list_of
 #' @importFrom vctrs new_rcrd
 #' @importFrom vctrs unspecified
 #' @importFrom vctrs vec_as_names

R/group_data.R

@@ -0,0 +1,37 @@
+# Generated by 02-duckplyr_df-methods.R
+#' @export
+group_data.duckplyr_df <- function(.data) {
+  # Our implementation
+  rel_try(
+    # Always fall back to dplyr
+    "No relational implementation for group_data()" = TRUE,
+    {
+      return(out)
+    }
+  )
+
+  # dplyr forward
+  group_data <- dplyr$group_data.data.frame
+  out <- group_data(.data)
+  return(out)
+
+  # dplyr implementation
+  size <- nrow(.data)
+  out <- seq_len(size)
+  out <- new_list_of(list(out), ptype = integer())
+  out <- list(.rows = out)
+  out <- new_data_frame(out, n = 1L)
+  out
+}
+
+duckplyr_group_data <- function(.data, ...) {
+  try_fetch(
+    .data <- as_duckplyr_df(.data),
+    error = function(e) {
+      testthat::skip(conditionMessage(e))
+    }
+  )
+  out <- group_data(.data, ...)
+  class(out) <- setdiff(class(out), "duckplyr_df")
+  out
+}

R/overwrite.R

@@ -15,6 +15,7 @@ methods_overwrite <- function() {
   vctrs::s3_register("dplyr::filter", "data.frame", filter.duckplyr_df)
   vctrs::s3_register("dplyr::full_join", "data.frame", full_join.duckplyr_df)
   vctrs::s3_register("dplyr::group_by", "data.frame", group_by.duckplyr_df)
+  vctrs::s3_register("dplyr::group_data", "data.frame", group_data.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

@@ -15,6 +15,7 @@ methods_restore <- function() {
   vctrs::s3_register("dplyr::filter", "data.frame", dplyr$filter.data.frame)
   vctrs::s3_register("dplyr::full_join", "data.frame", dplyr$full_join.data.frame)
   vctrs::s3_register("dplyr::group_by", "data.frame", dplyr$group_by.data.frame)
+  vctrs::s3_register("dplyr::group_data", "data.frame", dplyr$group_data.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_data.txt

@@ -0,0 +1,8 @@
+group_data.data.frame <- function(.data) {
+  size <- nrow(.data)
+  out <- seq_len(size)
+  out <- new_list_of(list(out), ptype = integer())
+  out <- list(.rows = out)
+  out <- new_data_frame(out, n = 1L)
+  out
+}

tests/testthat/test-as_duckplyr_df.R

@@ -754,6 +754,22 @@ test_that("as_duckplyr_df() and group_by()", {
   expect_equal(pre, post)
 })
 
+test_that("as_duckplyr_df() and group_data()", {
+  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_data()
+  post <- test_df %>% group_data() %>% 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-generics.R

@@ -1,16 +1,16 @@
 test_that("row_slice recomputes groups", {
   gf <- duckplyr_group_by(data.frame(g = c(1, 1, 2, 2, 3, 3)), g)
   out <- duckplyr_dplyr_row_slice(gf, c(1L, 3L, 5L))
-  expect_equal(group_data(out)$.rows, list_of(1L, 2L, 3L))
+  expect_equal(duckplyr_group_data(out)$.rows, list_of(1L, 2L, 3L))
 
   out <- duckplyr_dplyr_row_slice(gf, c(4L, 3L))
-  expect_equal(group_data(out)$.rows, list_of(c(1L, 2L)))
+  expect_equal(duckplyr_group_data(out)$.rows, list_of(c(1L, 2L)))
 })
 
 test_that("row_slice preserves empty groups if requested", {
   gf <- duckplyr_group_by(data.frame(g = c(1, 1, 2, 2, 3, 3)), g, .drop = FALSE)
   out <- duckplyr_dplyr_row_slice(gf, c(3L, 4L))
-  expect_equal(group_data(out)$.rows, list_of(integer(), c(1L, 2L), integer()))
+  expect_equal(duckplyr_group_data(out)$.rows, list_of(integer(), c(1L, 2L), integer()))
 })
 
 

tests/testthat/test-mutate.R

@@ -303,11 +303,11 @@ test_that("mutate preserves grouping", {
   i <- count_regroups(out <- duckplyr_mutate(gf, x = 1))
   expect_equal(i, 1L)
   expect_equal(duckplyr_group_vars(out), "x")
-  expect_equal(nrow(group_data(out)), 1)
+  expect_equal(nrow(duckplyr_group_data(out)), 1)
 
   i <- count_regroups(out <- duckplyr_mutate(gf, z = 1))
   expect_equal(i, 0)
-  expect_equal(group_data(out), group_data(gf))
+  expect_equal(duckplyr_group_data(out), duckplyr_group_data(gf))
 })
 
 test_that("mutate works on zero-row grouped data frame (#596)", {
@@ -319,7 +319,7 @@ test_that("mutate works on zero-row grouped data frame (#596)", {
 
   expect_type(group_rows(res), "list")
   expect_equal(attr(group_rows(res), "ptype"), integer())
-  expect_equal(group_data(res)$b, factor(character(0)))
+  expect_equal(duckplyr_group_data(res)$b, factor(character(0)))
 })
 
 test_that("mutate preserves class of zero-row rowwise (#4224, #6303)", {

tests/testthat/test-transmute.R

@@ -9,11 +9,11 @@ test_that("transmute preserves grouping", {
   i <- count_regroups(out <- duckplyr_transmute(gf, x = 1))
   expect_equal(i, 1L)
   expect_equal(duckplyr_group_vars(out), "x")
-  expect_equal(nrow(group_data(out)), 1)
+  expect_equal(nrow(duckplyr_group_data(out)), 1)
 
   i <- count_regroups(out <- duckplyr_transmute(gf, z = 1))
   expect_equal(i, 0)
-  expect_equal(group_data(out), group_data(gf))
+  expect_equal(duckplyr_group_data(out), duckplyr_group_data(gf))
 })
 
 # Empty transmutes -------------------------------------------------

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_data", "group_indices", "group_keys", "group_map", "group_modify", "group_nest", "group_size", "group_split", "group_trim", "groups", "n_groups",
+    "group_indices", "group_keys", "group_map", "group_modify", "group_nest", "group_size", "group_split", "group_trim", "groups", "n_groups",
     "same_src", # data frames can be copied into duck-frames with zero cost
     NULL
   ))) %>%
@@ -25,6 +25,7 @@ df_methods <-
     "dplyr_col_modify",
     "dplyr_row_slice",
     "group_by",
+    "group_data",
     "rowwise",
     NULL
   ))) %>%
@@ -691,6 +692,7 @@ test_skip_map <- c(
   # FIXME: Fail with group_by()
   dplyr_reconstruct = "Hack",
   group_by = "Grouped",
+  group_data = "Special",
   group_map = "WAT",
   group_modify = "Grouped",
   group_nest = "Always returns tibble",