Merge pull request #105 from duckdblabs/f-94-group-rest

NAMESPACE

@@ -16,17 +16,22 @@ S3method(explain,duckplyr_df)
 S3method(full_join,duckplyr_df)
 S3method(group_by,duckplyr_df)
 S3method(group_data,duckplyr_df)
+S3method(group_indices,duckplyr_df)
 S3method(group_keys,duckplyr_df)
 S3method(group_map,duckplyr_df)
 S3method(group_modify,duckplyr_df)
+S3method(group_nest,duckplyr_df)
 S3method(group_size,duckplyr_df)
+S3method(group_split,duckplyr_df)
+S3method(group_trim,duckplyr_df)
 S3method(group_vars,duckplyr_df)
 S3method(groups,duckplyr_df)
 S3method(head,duckplyr_df)
 S3method(inner_join,duckplyr_df)
 S3method(intersect,duckplyr_df)
 S3method(left_join,duckplyr_df)
 S3method(mutate,duckplyr_df)
+S3method(n_groups,duckplyr_df)
 S3method(nest_by,duckplyr_df)
 S3method(nest_join,duckplyr_df)
 S3method(print,relational_relexpr)
@@ -252,12 +257,15 @@ importFrom(dplyr,funs_)
 importFrom(dplyr,group_by)
 importFrom(dplyr,group_by_prepare)
 importFrom(dplyr,group_data)
+importFrom(dplyr,group_indices)
 importFrom(dplyr,group_keys)
 importFrom(dplyr,group_map)
 importFrom(dplyr,group_modify)
+importFrom(dplyr,group_nest)
 importFrom(dplyr,group_rows)
 importFrom(dplyr,group_size)
 importFrom(dplyr,group_split)
+importFrom(dplyr,group_trim)
 importFrom(dplyr,group_vars)
 importFrom(dplyr,group_walk)
 importFrom(dplyr,grouped_df)
@@ -276,6 +284,7 @@ importFrom(dplyr,mutate)
 importFrom(dplyr,mutate_all)
 importFrom(dplyr,n)
 importFrom(dplyr,n_distinct)
+importFrom(dplyr,n_groups)
 importFrom(dplyr,nest_by)
 importFrom(dplyr,nest_join)
 importFrom(dplyr,nth)

R/dplyr.R

@@ -58,6 +58,8 @@ ensure_group_vars <- dplyr$ensure_group_vars
 env_bind_do_pronouns <- dplyr$env_bind_do_pronouns
 filter_rows <- dplyr$filter_rows
 groups <- dplyr$groups
+group_nest_impl <- dplyr$group_nest_impl
+group_split_impl <- dplyr$group_split_impl
 join_cast_common <- dplyr$join_cast_common
 join_filter <- dplyr$join_filter
 join_mutate <- dplyr$join_mutate

R/duckplyr-package.R

@@ -9,9 +9,13 @@
 #' @importFrom collections dict
 #' @importFrom collections queue
 #' @importFrom dplyr group_by_prepare
+#' @importFrom dplyr group_indices
+#' @importFrom dplyr group_nest
 #' @importFrom dplyr group_split
+#' @importFrom dplyr group_trim
 #' @importFrom dplyr grouped_df
 #' @importFrom dplyr if_else
+#' @importFrom dplyr n_groups
 #' @importFrom glue glue
 #' @importFrom lifecycle deprecated
 #' @importFrom tibble as_tibble

R/group_indices.R

@@ -0,0 +1,41 @@
+# Generated by 02-duckplyr_df-methods.R
+#' @export
+group_indices.duckplyr_df <- function(.data, ...) {
+  # Our implementation
+  rel_try(
+    # Always fall back to dplyr
+    "No relational implementation for group_indices()" = TRUE,
+    {
+      return(out)
+    }
+  )
+
+  # dplyr forward
+  group_indices <- dplyr$group_indices.data.frame
+  out <- group_indices(.data, ...)
+  return(out)
+
+  # dplyr implementation
+  if (dots_n(...) > 0) {
+    lifecycle::deprecate_warn(
+      "1.0.0", "group_indices(... = )",
+      details = "Please `group_by()` first",
+      always = TRUE
+    )
+    .data <- group_by(.data, ...)
+  }
+
+  # .Call(`dplyr_group_indices`, .data, group_rows(.data))
+}
+
+duckplyr_group_indices <- function(.data, ...) {
+  try_fetch(
+    .data <- as_duckplyr_df(.data),
+    error = function(e) {
+      testthat::skip(conditionMessage(e))
+    }
+  )
+  out <- group_indices(.data, ...)
+  class(out) <- setdiff(class(out), "duckplyr_df")
+  out
+}

R/group_nest.R

@@ -0,0 +1,36 @@
+# Generated by 02-duckplyr_df-methods.R
+#' @export
+group_nest.duckplyr_df <- function(.tbl, ..., .key = "data", keep = FALSE) {
+  # Our implementation
+  rel_try(
+    # Always fall back to dplyr
+    "No relational implementation for group_nest()" = TRUE,
+    {
+      return(out)
+    }
+  )
+
+  # dplyr forward
+  group_nest <- dplyr$group_nest.data.frame
+  out <- group_nest(.tbl, ..., .key = .key, keep = keep)
+  return(out)
+
+  # dplyr implementation
+  if (dots_n(...)) {
+    group_nest_impl(group_by(.tbl, ...), .key = .key, keep = keep)
+  } else {
+    tibble(!!.key := list(.tbl))
+  }
+}
+
+duckplyr_group_nest <- function(.tbl, ...) {
+  try_fetch(
+    .tbl <- as_duckplyr_df(.tbl),
+    error = function(e) {
+      testthat::skip(conditionMessage(e))
+    }
+  )
+  out <- group_nest(.tbl, ...)
+  class(out) <- setdiff(class(out), "duckplyr_df")
+  out
+}

R/group_split.R

@@ -0,0 +1,37 @@
+# Generated by 02-duckplyr_df-methods.R
+#' @export
+group_split.duckplyr_df <- function(.tbl, ..., .keep = TRUE, keep = deprecated()) {
+  # Our implementation
+  rel_try(
+    # Always fall back to dplyr
+    "No relational implementation for group_split()" = TRUE,
+    {
+      return(out)
+    }
+  )
+
+  # dplyr forward
+  group_split <- dplyr$group_split.data.frame
+  out <- group_split(.tbl, ..., .keep = .keep, keep = keep)
+  return(out)
+
+  # dplyr implementation
+  if (!missing(keep)) {
+    lifecycle::deprecate_warn("1.0.0", "group_split(keep = )", "group_split(.keep = )", always = TRUE)
+    .keep <- keep
+  }
+  data <- group_by(.tbl, ...)
+  group_split_impl(data, .keep = .keep)
+}
+
+duckplyr_group_split <- function(.tbl, ...) {
+  try_fetch(
+    .tbl <- as_duckplyr_df(.tbl),
+    error = function(e) {
+      testthat::skip(conditionMessage(e))
+    }
+  )
+  out <- group_split(.tbl, ...)
+  class(out) <- setdiff(class(out), "duckplyr_df")
+  out
+}

R/group_trim.R

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

R/n_groups.R

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

R/overwrite.R

@@ -16,16 +16,21 @@ methods_overwrite <- function() {
   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_indices", "data.frame", group_indices.duckplyr_df)
   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_nest", "data.frame", group_nest.duckplyr_df)
   vctrs::s3_register("dplyr::group_size", "data.frame", group_size.duckplyr_df)
+  vctrs::s3_register("dplyr::group_split", "data.frame", group_split.duckplyr_df)
+  vctrs::s3_register("dplyr::group_trim", "data.frame", group_trim.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)
   vctrs::s3_register("dplyr::mutate", "data.frame", mutate.duckplyr_df)
+  vctrs::s3_register("dplyr::n_groups", "data.frame", n_groups.duckplyr_df)
   vctrs::s3_register("dplyr::nest_by", "data.frame", nest_by.duckplyr_df)
   vctrs::s3_register("dplyr::nest_join", "data.frame", nest_join.duckplyr_df)
   vctrs::s3_register("dplyr::pull", "data.frame", pull.duckplyr_df)

R/restore.R

@@ -16,16 +16,21 @@ methods_restore <- function() {
   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$group_data.data.frame)
+  vctrs::s3_register("dplyr::group_indices", "data.frame", dplyr$group_indices.data.frame)
   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$group_modify.data.frame)
+  vctrs::s3_register("dplyr::group_nest", "data.frame", dplyr$group_nest.data.frame)
   vctrs::s3_register("dplyr::group_size", "data.frame", dplyr$group_size.data.frame)
+  vctrs::s3_register("dplyr::group_split", "data.frame", dplyr$group_split.data.frame)
+  vctrs::s3_register("dplyr::group_trim", "data.frame", dplyr$group_trim.data.frame)
   vctrs::s3_register("dplyr::group_vars", "data.frame", dplyr$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)
   vctrs::s3_register("dplyr::mutate", "data.frame", dplyr$mutate.data.frame)
+  vctrs::s3_register("dplyr::n_groups", "data.frame", dplyr$n_groups.data.frame)
   vctrs::s3_register("dplyr::nest_by", "data.frame", dplyr$nest_by.data.frame)
   vctrs::s3_register("dplyr::nest_join", "data.frame", dplyr$nest_join.data.frame)
   vctrs::s3_register("dplyr::pull", "data.frame", dplyr$pull.data.frame)

dplyr-methods/group_indices.txt

@@ -0,0 +1,12 @@
+group_indices.data.frame <- function(.data, ...) {
+  if (dots_n(...) > 0) {
+    lifecycle::deprecate_warn(
+      "1.0.0", "group_indices(... = )",
+      details = "Please `group_by()` first",
+      always = TRUE
+    )
+    .data <- group_by(.data, ...)
+  }
+
+  .Call(`dplyr_group_indices`, .data, group_rows(.data))
+}

dplyr-methods/group_nest.txt

@@ -0,0 +1,7 @@
+group_nest.data.frame <- function(.tbl, ..., .key = "data", keep = FALSE) {
+  if (dots_n(...)) {
+    group_nest_impl(group_by(.tbl, ...), .key = .key, keep = keep)
+  } else {
+    tibble(!!.key := list(.tbl))
+  }
+}

dplyr-methods/group_split.txt

@@ -0,0 +1,8 @@
+group_split.data.frame <- function(.tbl, ..., .keep = TRUE, keep = deprecated()) {
+  if (!missing(keep)) {
+    lifecycle::deprecate_warn("1.0.0", "group_split(keep = )", "group_split(.keep = )", always = TRUE)
+    .keep <- keep
+  }
+  data <- group_by(.tbl, ...)
+  group_split_impl(data, .keep = .keep)
+}

dplyr-methods/group_trim.txt

@@ -0,0 +1,3 @@
+group_trim.data.frame <- function(.tbl, .drop = group_by_drop_default(.tbl)) {
+  .tbl
+}

dplyr-methods/n_groups.txt

@@ -0,0 +1,3 @@
+n_groups.data.frame <- function(x) {
+  nrow(group_data(x))
+}

patch/group_indices.patch

@@ -0,0 +1,13 @@
+diff --git b/R/group_indices.R a/R/group_indices.R
+index 48bcf3e..6c1bf15 100644
+--- b/R/group_indices.R
++++ a/R/group_indices.R
+@@ -25,7 +25,7 @@ group_indices.duckplyr_df <- function(.data, ...) {
+     .data <- group_by(.data, ...)
+   }
+
+-  .Call(`dplyr_group_indices`, .data, group_rows(.data))
++  # .Call(`dplyr_group_indices`, .data, group_rows(.data))
+ }
+
+ duckplyr_group_indices <- function(.data, ...) {