Merge branch 'main' into polars-1.10.0

Author: etiennebacher

NEWS.md

@@ -27,6 +27,8 @@
 
 * Fix `sample()` to make it work correctly (@Yousa-Mirage, #338).
 
+* Fix `unite()` behavior when `na.rm = TRUE` (#344).
+
 # tidypolars 0.17.0
 
 `tidypolars` requires `polars` >= 1.9.0 and `dplyr` >= 1.2.0.

R/unite.R

@@ -50,13 +50,17 @@ unite.polars_data_frame <- function(
   col <- rlang::as_string(rlang::ensym(col))
 
   if (isTRUE(na.rm)) {
-    fill <- ""
+    vars_to_concat <- vars
   } else {
-    fill <- "NA"
+    vars_to_concat <- list(pl$col(!!!vars)$fill_null("NA"))
   }
 
   out <- data$with_columns(
-    pl$concat_str(!!!vars, separator = sep, ignore_nulls = TRUE)$alias(col)
+    pl$concat_str(
+      !!!vars_to_concat,
+      separator = sep,
+      ignore_nulls = TRUE
+    )$alias(col)
   )
 
   out <- if (isTRUE(remove)) {

tests/testthat/test-unite-lazy.R

@@ -21,7 +21,7 @@ test_that("basic behavior works", {
   )
 })
 
-test_that("argument remove works", {
+test_that("argument 'remove' works", {
   test_df <- tibble(
     year = 2009:2011,
     month = 10:12,
@@ -52,6 +52,28 @@ test_that("argument remove works", {
   )
 })
 
+test_that("argument 'na.rm' works", {
+  test_df <- tibble(
+    name = c(NA, "Jack", "Thomas"),
+    middle = c("T.", NA, "F."),
+    surname = c(NA, "Thompson", "Jones")
+  )
+  test_pl <- as_polars_lf(test_df)
+
+  expect_equal_lazy(
+    unite(test_pl, col = "out", name, middle, surname, sep = "-"),
+    unite(test_df, col = "out", name, middle, surname, sep = "-")
+  )
+  expect_equal_lazy(
+    unite(test_pl, col = "out", name, middle, surname, na.rm = FALSE),
+    unite(test_df, col = "out", name, middle, surname, na.rm = FALSE)
+  )
+  expect_equal_lazy(
+    unite(test_pl, col = "out", name, middle, surname, na.rm = TRUE),
+    unite(test_df, col = "out", name, middle, surname, na.rm = TRUE)
+  )
+})
+
 test_that("tidy selection works", {
   test_df <- tibble(
     name = c("John", "Jack", "Thomas"),
@@ -61,20 +83,8 @@ test_that("tidy selection works", {
   test_pl <- as_polars_lf(test_df)
 
   expect_equal_lazy(
-    unite(
-      test_pl,
-      col = "full_name",
-      everything(),
-      sep = " ",
-      na.rm = TRUE
-    ),
-    unite(
-      test_df,
-      col = "full_name",
-      everything(),
-      sep = " ",
-      na.rm = TRUE
-    )
+    unite(test_pl, col = "full_name", everything(), sep = " "),
+    unite(test_df, col = "full_name", everything(), sep = " ")
   )
 })
 

tests/testthat/test-unite.R

@@ -17,7 +17,7 @@ test_that("basic behavior works", {
   )
 })
 
-test_that("argument remove works", {
+test_that("argument 'remove' works", {
   test_df <- tibble(
     year = 2009:2011,
     month = 10:12,
@@ -48,6 +48,28 @@ test_that("argument remove works", {
   )
 })
 
+test_that("argument 'na.rm' works", {
+  test_df <- tibble(
+    name = c(NA, "Jack", "Thomas"),
+    middle = c("T.", NA, "F."),
+    surname = c(NA, "Thompson", "Jones")
+  )
+  test_pl <- as_polars_df(test_df)
+
+  expect_equal(
+    unite(test_pl, col = "out", name, middle, surname, sep = "-"),
+    unite(test_df, col = "out", name, middle, surname, sep = "-")
+  )
+  expect_equal(
+    unite(test_pl, col = "out", name, middle, surname, na.rm = FALSE),
+    unite(test_df, col = "out", name, middle, surname, na.rm = FALSE)
+  )
+  expect_equal(
+    unite(test_pl, col = "out", name, middle, surname, na.rm = TRUE),
+    unite(test_df, col = "out", name, middle, surname, na.rm = TRUE)
+  )
+})
+
 test_that("tidy selection works", {
   test_df <- tibble(
     name = c("John", "Jack", "Thomas"),
@@ -57,20 +79,8 @@ test_that("tidy selection works", {
   test_pl <- as_polars_df(test_df)
 
   expect_equal(
-    unite(
-      test_pl,
-      col = "full_name",
-      everything(),
-      sep = " ",
-      na.rm = TRUE
-    ),
-    unite(
-      test_df,
-      col = "full_name",
-      everything(),
-      sep = " ",
-      na.rm = TRUE
-    )
+    unite(test_pl, col = "full_name", everything(), sep = " "),
+    unite(test_df, col = "full_name", everything(), sep = " ")
   )
 })