groups

NAMESPACE

@@ -21,6 +21,7 @@ S3method(group_map,duckplyr_df)
 S3method(group_modify,duckplyr_df)
 S3method(group_size,duckplyr_df)
 S3method(group_vars,duckplyr_df)
+S3method(groups,duckplyr_df)
 S3method(head,duckplyr_df)
 S3method(inner_join,duckplyr_df)
 S3method(intersect,duckplyr_df)

R/groups.R

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

R/overwrite.R

@@ -21,6 +21,7 @@ methods_overwrite <- function() {
   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::groups", "data.frame", groups.duckplyr_df)
   vctrs::s3_register("dplyr::inner_join", "data.frame", inner_join.duckplyr_df)
   vctrs::s3_register("dplyr::intersect", "data.frame", intersect.duckplyr_df)
   vctrs::s3_register("dplyr::left_join", "data.frame", left_join.duckplyr_df)

R/restore.R

@@ -21,6 +21,7 @@ methods_restore <- function() {
   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$group_vars.data.frame)
+  vctrs::s3_register("dplyr::groups", "data.frame", dplyr$groups.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)
   vctrs::s3_register("dplyr::left_join", "data.frame", dplyr$left_join.data.frame)

dplyr-methods/groups.txt

@@ -0,0 +1,3 @@
+groups.data.frame <- function(x) {
+  syms(group_vars(x))
+}

tests/testthat/test-as_duckplyr_df.R

@@ -863,6 +863,22 @@ test_that("as_duckplyr_df() and group_vars()", {
   expect_equal(pre, post)
 })
 
+test_that("as_duckplyr_df() and groups()", {
+  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() %>% groups()
+  post <- test_df %>% groups() %>% as_duckplyr_df()
+
+  # Compare
+  expect_equal(pre, post)
+})
+
 test_that("as_duckplyr_df() and inner_join(join_by(a))", {
   withr::local_envvar(DUCKPLYR_FALLBACK_FORCE = "TRUE")
 

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_split", "group_trim", "groups", "n_groups",
+    "group_indices", "group_nest", "group_split", "group_trim", "n_groups",
     "same_src", # data frames can be copied into duck-frames with zero cost
     NULL
   ))) %>%
@@ -30,6 +30,7 @@ df_methods <-
     "group_map",
     "group_modify",
     "group_size",
+    "groups",
     "rowwise",
     NULL
   ))) %>%
@@ -704,6 +705,7 @@ test_skip_map <- c(
   group_size = "Special",
   group_split = "WAT",
   group_trim = "Grouped",
+  groups = "Special",
   nest_by = "WAT",
   # FIXME: Fail with rowwise()
   rowwise = "Stack overflow",