From 86078768231625d798de4d48615ea926369952a3 Mon Sep 17 00:00:00 2001
From: eutwt <11261404+eutwt@users.noreply.github.com>
Date: Mon, 30 Aug 2021 10:09:14 -0400
Subject: [PATCH 1/3] first draft

---
 R/step-subset-expand.R | 29 +++++++++++++++++++++++------
 1 file changed, 23 insertions(+), 6 deletions(-)

diff --git a/R/step-subset-expand.R b/R/step-subset-expand.R
index fefa753fa..1795279e2 100644
--- a/R/step-subset-expand.R
+++ b/R/step-subset-expand.R
@@ -19,12 +19,29 @@
 #'   weights = rnorm(6, as.numeric(size) + 2)
 #' ))
 #'
-#' # All possible combinations ---------------------------------------
-#' # Note that all defined, but not necessarily present, levels of the
-#' # factor variable `size` are retained.
-#' fruits %>% expand(type)
-#' fruits %>% expand(type, size)
-#' fruits %>% expand(type, size, year)
+#' # Factors ----------------------------------------------------------
+#' # The output of `expand(<dtplyr_step>)` only contains levels of factor 
+#' # variables present in the data. This is in contrast to the output of 
+#' # `expand()` when the input is a data.frame, which retains all factor levels.
+#' # Using the `fruits` data, the level "L" is not be present in `expand()` 
+#' # output, unlike the output of `expand()` after converting `fruits()` to a
+#' # tibble.
+#' 
+#' fruits %>% expand(size) %>% as_tibble()
+#' #> # A tibble: 3 × 1
+#' #>   size
+#' #>   <fct>
+#' #> 1 XS
+#' #> 2 S
+#' #> 3 M
+#' fruits %>% as_tibble() %>% expand(size)
+#' #> # A tibble: 4 × 1
+#' #>   size
+#' #>   <fct>
+#' #> 1 XS
+#' #> 2 S
+#' #> 3 M
+#' #> 4 L
 #'
 #' # Other uses -------------------------------------------------------
 #' fruits %>% expand(type, size, 2010:2012)

From ca6f19ac400c02235ab9fa6818d3085fa69c3f16 Mon Sep 17 00:00:00 2001
From: eutwt <11261404+eutwt@users.noreply.github.com>
Date: Mon, 30 Aug 2021 11:40:58 -0400
Subject: [PATCH 2/3] document factor behavior

---
 R/step-subset-expand.R    | 15 +++++++--------
 man/expand.dtplyr_step.Rd | 29 ++++++++++++++++++++++-------
 2 files changed, 29 insertions(+), 15 deletions(-)

diff --git a/R/step-subset-expand.R b/R/step-subset-expand.R
index 1795279e2..32ee2d7e3 100644
--- a/R/step-subset-expand.R
+++ b/R/step-subset-expand.R
@@ -2,7 +2,8 @@
 #'
 #' @description
 #' This is a method for the tidyr `expand()` generic. It is translated to
-#' [data.table::CJ()].
+#' [data.table::CJ()]. Unlike the data.frame method, this method only retains
+#' factor levels present in the data.
 #'
 #' @param data A [lazy_dt()].
 #' @inheritParams tidyr::expand
@@ -20,13 +21,10 @@
 #' ))
 #'
 #' # Factors ----------------------------------------------------------
-#' # The output of `expand(<dtplyr_step>)` only contains levels of factor 
-#' # variables present in the data. This is in contrast to the output of 
-#' # `expand()` when the input is a data.frame, which retains all factor levels.
-#' # Using the `fruits` data, the level "L" is not be present in `expand()` 
-#' # output, unlike the output of `expand()` after converting `fruits()` to a
-#' # tibble.
-#' 
+#' # When called on `fruits` defined above, the level "L" is not present in
+#' # the output of `expand(fruits, size)`, unlike the output of `expand()` when
+#' # `fruits` is a data.frame.
+#'
 #' fruits %>% expand(size) %>% as_tibble()
 #' #> # A tibble: 3 × 1
 #' #>   size
@@ -54,6 +52,7 @@
 #' # Use with `right_join()` to fill in missing rows
 #' fruits %>% dplyr::right_join(all)
 # exported onLoad
+
 expand.dtplyr_step <- function(data, ..., .name_repair = "check_unique") {
   dots <- capture_dots(data, ..., .j = FALSE)
   dots <- dots[!vapply(dots, is_null, logical(1))]
diff --git a/man/expand.dtplyr_step.Rd b/man/expand.dtplyr_step.Rd
index 31f4698c7..22e678405 100644
--- a/man/expand.dtplyr_step.Rd
+++ b/man/expand.dtplyr_step.Rd
@@ -49,7 +49,8 @@ to enforce them.}
 }
 \description{
 This is a method for the tidyr \code{expand()} generic. It is translated to
-\code{\link[data.table:J]{data.table::CJ()}}.
+\code{\link[data.table:J]{data.table::CJ()}}. Unlike the data.frame method, this method only retains
+factor levels present in the data.
 }
 \examples{
 library(tidyr)
@@ -64,12 +65,26 @@ fruits <- lazy_dt(tibble(
   weights = rnorm(6, as.numeric(size) + 2)
 ))
 
-# All possible combinations ---------------------------------------
-# Note that all defined, but not necessarily present, levels of the
-# factor variable `size` are retained.
-fruits \%>\% expand(type)
-fruits \%>\% expand(type, size)
-fruits \%>\% expand(type, size, year)
+# Factors ----------------------------------------------------------
+# When called on `fruits` defined above, the level "L" is not present in
+# the output of `expand(fruits, size)`, unlike the output of `expand()` when
+# `fruits` is a data.frame.
+
+fruits \%>\% expand(size) \%>\% as_tibble()
+#> # A tibble: 3 × 1
+#>   size
+#>   <fct>
+#> 1 XS
+#> 2 S
+#> 3 M
+fruits \%>\% as_tibble() \%>\% expand(size)
+#> # A tibble: 4 × 1
+#>   size
+#>   <fct>
+#> 1 XS
+#> 2 S
+#> 3 M
+#> 4 L
 
 # Other uses -------------------------------------------------------
 fruits \%>\% expand(type, size, 2010:2012)

From e91110c2f99b2b69d712ca34df9f7d91935f2a52 Mon Sep 17 00:00:00 2001
From: eutwt <11261404+eutwt@users.noreply.github.com>
Date: Mon, 30 Aug 2021 11:51:58 -0400
Subject: [PATCH 3/3] formatting

---
 R/step-subset-expand.R | 1 -
 1 file changed, 1 deletion(-)

diff --git a/R/step-subset-expand.R b/R/step-subset-expand.R
index 32ee2d7e3..cce3fd7be 100644
--- a/R/step-subset-expand.R
+++ b/R/step-subset-expand.R
@@ -52,7 +52,6 @@
 #' # Use with `right_join()` to fill in missing rows
 #' fruits %>% dplyr::right_join(all)
 # exported onLoad
-
 expand.dtplyr_step <- function(data, ..., .name_repair = "check_unique") {
   dots <- capture_dots(data, ..., .j = FALSE)
   dots <- dots[!vapply(dots, is_null, logical(1))]