This is an automated email from the ASF dual-hosted git repository.
npr pushed a commit to branch main
in repository https://gitbox.apache.org/repos/asf/arrow.git
The following commit(s) were added to refs/heads/main by this push:
new 81b94dcda6 GH-43440: [R] Unable to filter a factor column with %in%
(#43446)
81b94dcda6 is described below
commit 81b94dcda6180781effc331650d9cd413a4ac115
Author: Neal Richardson <[email protected]>
AuthorDate: Sun Sep 22 11:15:40 2024 -0400
GH-43440: [R] Unable to filter a factor column with %in% (#43446)
### Rationale for this change
Fixes #43440
### What changes are included in this PR?
The binding for `%in%` sends the DictionaryType's `value_type` to
`cast_or_parse()`. It's possible that it would be better to handle this
in `cast_or_parse()`, but it is used in lots of places and I wasn't sure
that was correct everywhere. We could certainly find out, but that's a
bigger testing exercise than I wanted to take on this afternoon.
### Are these changes tested?
Yes.
### Are there any user-facing changes?
The bug is fixed.
* GitHub Issue: #43440
---
r/R/dplyr-funcs-conditional.R | 11 +++++++++--
r/tests/testthat/test-dplyr-funcs-conditional.R | 10 ++++++++++
2 files changed, 19 insertions(+), 2 deletions(-)
diff --git a/r/R/dplyr-funcs-conditional.R b/r/R/dplyr-funcs-conditional.R
index 3ab955aa8a..52f47a128f 100644
--- a/r/R/dplyr-funcs-conditional.R
+++ b/r/R/dplyr-funcs-conditional.R
@@ -21,9 +21,16 @@ register_bindings_conditional <- function() {
value_set <- Array$create(table)
# If possible, `table` should be the same type as `x`
# Try downcasting here; otherwise Acero may upcast x to table's type
+ x_type <- x$type()
+ # GH-43440: `is_in` doesn't want a DictionaryType in the value_set,
+ # so we'll cast to its value_type
+ # TODO: should this be pushed into cast_or_parse? Is this a bigger issue?
+ if (inherits(x_type, "DictionaryType")) {
+ x_type <- x_type$value_type
+ }
try(
- value_set <- cast_or_parse(value_set, x$type()),
- silent = TRUE
+ value_set <- cast_or_parse(value_set, x_type),
+ silent = !getOption("arrow.debug", FALSE)
)
expr <- Expression$create("is_in", x,
diff --git a/r/tests/testthat/test-dplyr-funcs-conditional.R
b/r/tests/testthat/test-dplyr-funcs-conditional.R
index d90dc827b4..24ddd342a8 100644
--- a/r/tests/testthat/test-dplyr-funcs-conditional.R
+++ b/r/tests/testthat/test-dplyr-funcs-conditional.R
@@ -26,6 +26,16 @@ tbl <- example_data
tbl$verses <- verses[[1]]
tbl$another_chr <- tail(letters, 10)
+test_that("%in% handles dictionary type", {
+ df <- tibble::tibble(x = factor(c("a", "b", "c")))
+ compare_dplyr_binding(
+ .input %>%
+ filter(x %in% "a") %>%
+ collect(),
+ df
+ )
+})
+
test_that("if_else and ifelse", {
compare_dplyr_binding(
.input %>%
