fix: Keeps Enum order when converting to R

DESCRIPTION

@@ -35,6 +35,7 @@ Suggests:
     nanoarrow (>= 0.6.0),
     nycflights13,
     patrick,
+    quickcheck,
     pillar,
     rlang,
     rmarkdown,
@@ -118,5 +119,5 @@ Collate:
     'zzz.R'
 Config/rextendr/version: 0.3.1
 VignetteBuilder: knitr
-Config/polars/LibVersion: 0.43.0
+Config/polars/LibVersion: 0.43.1
 Config/polars/RustToolchainVersion: nightly-2024-09-19

NEWS.md

@@ -2,6 +2,10 @@
 
 ## Polars R Package (development version)
 
+### Bug fixes
+
+- Maintain level order when converting Enums to factors (#1252, @andyquinterom).
+
 ## Polars R Package 0.20.0
 
 - Updated rust-polars to 0.43.1 (#1230).

src/rust/Cargo.toml

@@ -1,6 +1,6 @@
 [package]
 name = "r-polars"
-version = "0.43.0"
+version = "0.43.1"
 edition = "2021"
 rust-version = "1.80.0"
 publish = false

src/rust/src/conversion_s_to_r.rs

@@ -110,9 +110,11 @@ pub fn pl_series_to_list(
                     .set_class(["rpolars_raw_list", "list"])
                     .expect("this class label is always valid")
             }),
-            Enum(_, _) => s
-                .categorical()
-                .map(|ca| extendr_api::call!("factor", ca.iter_str().collect_robj()).unwrap()),
+            Enum(_, _) => s.categorical().map(|ca| {
+                let levels = ca.unique().unwrap().iter_str().collect_robj();
+                let values = ca.iter_str().collect_robj();
+                extendr_api::call!("factor", values, levels).unwrap()
+            }),
             Categorical(_, _) => s
                 .categorical()
                 .map(|ca| extendr_api::call!("factor", ca.iter_str().collect_robj()).unwrap()),

tests/testthat/test-datatype.R

@@ -224,11 +224,16 @@ test_that("contains_* functions for datatype work", {
 })
 
 test_that("Enum", {
-  expect_identical(
-    as_polars_series(c("z", "z", "k", "a"))$
-      cast(pl$Enum(c("z", "k", "a")))$
-      to_r(),
-    factor(c("z", "z", "k", "a"))
+  quickcheck::for_all(
+    factors = quickcheck::factor_(),
+    property = function(factors) {
+      expect_identical(
+        as_polars_series(as.character(factors))$
+          cast(pl$Enum(levels(factors)))$
+          to_r(),
+        factors
+      )
+    }
   )
 
   expect_grepl_error(pl$Enum(), "missing")