feat: Add support for %notin% (#349)

Author: etiennebacher

NEWS.md

@@ -1,3 +1,9 @@
+# tidypolars (development)
+
+## New features
+
+* Added support for `%notin%` (#349).
+
 # tidypolars 0.18.0
 
 `tidypolars` requires `polars` >= 1.10.0.

R/utils-expr.R

@@ -492,6 +492,37 @@ translate <- function(
           )
           return(out)
         },
+        # Same thing as "%in%" but with the "$not()" at the end.
+        "%notin%" = {
+          out <- tryCatch(
+            {
+              lhs <- translate(
+                expr[[2]],
+                .data = .data,
+                new_vars = new_vars,
+                env = env,
+                caller = caller,
+                expr_uses_col = expr_uses_col
+              ) |>
+                as_polars_expr(as_lit = TRUE)
+              rhs <- translate(
+                expr[[3]],
+                .data = .data,
+                new_vars = new_vars,
+                env = env,
+                caller = caller,
+                expr_uses_col = expr_uses_col
+              ) |>
+                as_polars_expr(as_lit = TRUE)
+              if (is.list(rhs)) {
+                rhs <- unlist(rhs)
+              }
+              lhs$is_in(rhs$implode(), nulls_equal = TRUE)$not()
+            },
+            error = identity
+          )
+          return(out)
+        },
         "base::ifelse" = ,
         "ifelse" = ,
         "dplyr::if_else" = ,

tests/testthat/test-filter-lazy.R

@@ -190,11 +190,6 @@ test_that("%in% works", {
     test_df |> filter(Species %in% c("setosa", "virginica"))
   )
 
-  expect_equal_lazy(
-    test_pl |> filter(Species %in% c("setosa", "virginica")),
-    test_df |> filter(Species %in% c("setosa", "virginica"))
-  )
-
   test_df <- tibble(x = c(1, 2, 3), y = c(1, 3, 2))
   test_pl <- pl$LazyFrame(x = c(1, 2, 3), y = c(1, 3, 2))
 
@@ -214,6 +209,69 @@ test_that("%in% works with NA", {
   )
 })
 
+test_that("%notin% works", {
+  test_df <- as_tibble(mtcars)
+  test_pl <- as_polars_lf(test_df)
+
+  expect_equal_lazy(
+    filter(test_pl, cyl %notin% 4:5),
+    filter(test_df, cyl %notin% 4:5)
+  )
+
+  expect_equal_lazy(
+    filter(test_pl, cyl %notin% 4:5 & am %notin% 1),
+    filter(test_df, cyl %notin% 4:5 & am %notin% 1)
+  )
+
+  expect_equal_lazy(
+    filter(test_pl, cyl %notin% 4:5, am %notin% 1),
+    filter(test_df, cyl %notin% 4:5, am %notin% 1)
+  )
+
+  expect_equal_lazy(
+    filter(test_pl, cyl %notin% 4:5 | am %notin% 1),
+    filter(test_df, cyl %notin% 4:5 | am %notin% 1)
+  )
+
+  expect_equal_lazy(
+    filter(test_pl, cyl %notin% 4:5, vs == 1),
+    filter(test_df, cyl %notin% 4:5, vs == 1)
+  )
+
+  expect_equal_lazy(
+    filter(test_pl, cyl %notin% 4:5 | carb == 4),
+    filter(test_df, cyl %notin% 4:5 | carb == 4)
+  )
+
+  test_df <- as_tibble(iris)
+  # TODO: this shouldn't be necessary
+  test_df$Species <- as.character(test_df$Species)
+  test_pl <- as_polars_lf(test_df)
+
+  expect_equal_lazy(
+    test_pl |> filter(Species %notin% c("setosa", "virginica")),
+    test_df |> filter(Species %notin% c("setosa", "virginica"))
+  )
+
+  test_df <- tibble(x = c(1, 2, 3), y = c(1, 3, 2))
+  test_pl <- pl$LazyFrame(x = c(1, 2, 3), y = c(1, 3, 2))
+
+  expect_equal_lazy(
+    test_pl |> filter(x %notin% y),
+    test_df |> filter(x %notin% y)
+  )
+})
+
+test_that("%notin% works with NA", {
+  test_df <- tibble(x = c(1, 2, NA))
+  test_pl <- as_polars_lf(test_df)
+
+  expect_equal_lazy(
+    test_pl |> filter(x %notin% c(1, NA)),
+    test_df |> filter(x %notin% c(1, NA))
+  )
+})
+
 test_that("between() works", {
   test_df <- as_tibble(iris)
   # TODO: this shouldn't be necessary

tests/testthat/test-filter.R

@@ -186,11 +186,6 @@ test_that("%in% works", {
     test_df |> filter(Species %in% c("setosa", "virginica"))
   )
 
-  expect_equal(
-    test_pl |> filter(Species %in% c("setosa", "virginica")),
-    test_df |> filter(Species %in% c("setosa", "virginica"))
-  )
-
   test_df <- tibble(x = c(1, 2, 3), y = c(1, 3, 2))
   test_pl <- pl$DataFrame(x = c(1, 2, 3), y = c(1, 3, 2))
 
@@ -210,6 +205,69 @@ test_that("%in% works with NA", {
   )
 })
 
+test_that("%notin% works", {
+  test_df <- as_tibble(mtcars)
+  test_pl <- as_polars_df(test_df)
+
+  expect_equal(
+    filter(test_pl, cyl %notin% 4:5),
+    filter(test_df, cyl %notin% 4:5)
+  )
+
+  expect_equal(
+    filter(test_pl, cyl %notin% 4:5 & am %notin% 1),
+    filter(test_df, cyl %notin% 4:5 & am %notin% 1)
+  )
+
+  expect_equal(
+    filter(test_pl, cyl %notin% 4:5, am %notin% 1),
+    filter(test_df, cyl %notin% 4:5, am %notin% 1)
+  )
+
+  expect_equal(
+    filter(test_pl, cyl %notin% 4:5 | am %notin% 1),
+    filter(test_df, cyl %notin% 4:5 | am %notin% 1)
+  )
+
+  expect_equal(
+    filter(test_pl, cyl %notin% 4:5, vs == 1),
+    filter(test_df, cyl %notin% 4:5, vs == 1)
+  )
+
+  expect_equal(
+    filter(test_pl, cyl %notin% 4:5 | carb == 4),
+    filter(test_df, cyl %notin% 4:5 | carb == 4)
+  )
+
+  test_df <- as_tibble(iris)
+  # TODO: this shouldn't be necessary
+  test_df$Species <- as.character(test_df$Species)
+  test_pl <- as_polars_df(test_df)
+
+  expect_equal(
+    test_pl |> filter(Species %notin% c("setosa", "virginica")),
+    test_df |> filter(Species %notin% c("setosa", "virginica"))
+  )
+
+  test_df <- tibble(x = c(1, 2, 3), y = c(1, 3, 2))
+  test_pl <- pl$DataFrame(x = c(1, 2, 3), y = c(1, 3, 2))
+
+  expect_equal(
+    test_pl |> filter(x %notin% y),
+    test_df |> filter(x %notin% y)
+  )
+})
+
+test_that("%notin% works with NA", {
+  test_df <- tibble(x = c(1, 2, NA))
+  test_pl <- as_polars_df(test_df)
+
+  expect_equal(
+    test_pl |> filter(x %notin% c(1, NA)),
+    test_df |> filter(x %notin% c(1, NA))
+  )
+})
+
 test_that("between() works", {
   test_df <- as_tibble(iris)
   # TODO: this shouldn't be necessary

vignettes/supported-functions.Rmd

@@ -85,6 +85,8 @@ out <- tribble(
   "`base`", "`unique`",
   "`base`", "`which.min`",
   "`base`", "`which.max`",
+  "`base`", "`%in%`",
+  "`base`", "`%notin%`",
   "`dplyr`", "`between`",
   "`dplyr`", "`case_match`",
   "`dplyr`", "`case_when`",