feat: Add partial support for stringr::str_equal()

DESCRIPTION

@@ -1,7 +1,7 @@
 Package: tidypolars
 Type: Package
 Title: More Efficient Tidyverse Code, Using Polars in the Background
-Version: 0.14.1
+Version: 0.14.1.9000
 Authors@R:
     c(person(given = "Etienne",
              family = "Bacher",

NEWS.md

@@ -1,3 +1,9 @@
+# tidypolars (development)
+
+## New features
+
+* Add partial support for `stringr::str_equal()` (#228).
+
 # tidypolars 0.14.1
 
 * `tidypolars` requires `polars` >= 1.1.0 (#222).

R/funs-string.R

@@ -57,11 +57,6 @@ pl_str_dup_stringr <- function(string, times) {
   )
 }
 
-# TODO: this requires https://github.com/pola-rs/polars/issues/11455
-# pl_str_equal_string <- function(x, y, ...) {
-#
-# }
-
 pl_str_ends_stringr <- function(string, pattern, negate = FALSE, ...) {
   check_empty_dots(...)
   pattern <- check_pattern(pattern)
@@ -76,6 +71,12 @@ pl_str_ends_stringr <- function(string, pattern, negate = FALSE, ...) {
   out
 }
 
+pl_str_equal_stringr <- function(x, y, ...) {
+  check_empty_dots(...)
+  x$str$normalize("NFC") == y$str$normalize("NFC")
+}
+
+
 # group = 0 means the whole match
 pl_str_extract_stringr <- function(string, pattern, group = 0, ...) {
   check_empty_dots(...)

tests/testthat/test-funs-string-lazy.R

@@ -198,4 +198,26 @@ test_that("substr() works", {
   )
 })
 
+test_that("str_equal() works", {
+  # need equal length of inputs
+  length <- sample(0:10, 1)
+
+  for_all(
+    tests = 40,
+    x = character_(any_na = TRUE, len = length),
+    y = character_(any_na = TRUE, len = length),
+    property = function(x, y) {
+      test_df <- data.frame(x = x, y = y)
+      test <- as_polars_lf(test_df)
+
+      expect_equal_or_both_error(
+        mutate(test, foo = str_equal(x, y)) |>
+          pull(foo),
+        mutate(test_df, foo = str_equal(x, y)) |>
+          pull(foo)
+      )
+    }
+  )
+})
+
 Sys.setenv('TIDYPOLARS_TEST' = FALSE)

tests/testthat/test-funs-string.R

@@ -193,3 +193,25 @@ test_that("substr() works", {
     }
   )
 })
+
+test_that("str_equal() works", {
+  # need equal length of inputs
+  length <- sample(0:10, 1)
+
+  for_all(
+    tests = 40,
+    x = character_(any_na = TRUE, len = length),
+    y = character_(any_na = TRUE, len = length),
+    property = function(x, y) {
+      test_df <- data.frame(x = x, y = y)
+      test <- as_polars_df(test_df)
+
+      expect_equal_or_both_error(
+        mutate(test, foo = str_equal(x, y)) |>
+          pull(foo),
+        mutate(test_df, foo = str_equal(x, y)) |>
+          pull(foo)
+      )
+    }
+  )
+})

tests/testthat/test-funs_string-lazy.R

@@ -1233,4 +1233,31 @@ test_that("stringr::str_replace_na works", {
   )
 })
 
+test_that("str_equal() works", {
+  test_df <- data.frame(x = c("\u00e1", "\u2126"), y = c("a\u0301", "\u03A9"))
+  test <- as_polars_lf(test_df)
+
+  expect_equal_lazy(
+    mutate(test, eq = str_equal(x, y)) |>
+      pull(eq),
+    mutate(test_df, eq = str_equal(x, y)) |>
+      pull(eq)
+  )
+
+  expect_warning(
+    mutate(test, eq = str_equal(x, y, ignore_case = TRUE)),
+    "doesn't know how to use some arguments"
+  )
+
+  test_df <- data.frame(x = character(), y = character())
+  test <- as_polars_lf(test_df)
+
+  expect_equal_lazy(
+    mutate(test, eq = str_equal(x, y)) |>
+      pull(eq),
+    mutate(test_df, eq = str_equal(x, y)) |>
+      pull(eq)
+  )
+})
+
 Sys.setenv('TIDYPOLARS_TEST' = FALSE)

tests/testthat/test-funs_string.R

@@ -1228,3 +1228,30 @@ test_that("stringr::str_replace_na works", {
     error = TRUE
   )
 })
+
+test_that("str_equal() works", {
+  test_df <- data.frame(x = c("\u00e1", "\u2126"), y = c("a\u0301", "\u03A9"))
+  test <- as_polars_df(test_df)
+
+  expect_equal(
+    mutate(test, eq = str_equal(x, y)) |>
+      pull(eq),
+    mutate(test_df, eq = str_equal(x, y)) |>
+      pull(eq)
+  )
+
+  expect_warning(
+    mutate(test, eq = str_equal(x, y, ignore_case = TRUE)),
+    "doesn't know how to use some arguments"
+  )
+
+  test_df <- data.frame(x = character(), y = character())
+  test <- as_polars_df(test_df)
+
+  expect_equal(
+    mutate(test, eq = str_equal(x, y)) |>
+      pull(eq),
+    mutate(test_df, eq = str_equal(x, y)) |>
+      pull(eq)
+  )
+})

vignettes/r-and-polars-expressions.Rmd

@@ -359,6 +359,7 @@ out <- tribble(
   "`stringr`", "`str_detect`",
   "`stringr`", "`str_dup`",
   "`stringr`", "`str_ends`",
+  "`stringr`", "`str_equal`",
   "`stringr`", "`str_extract`",
   "`stringr`", "`str_extract_all`",
   "`stringr`", "`str_length`",