Fixes null propagation in boolean casts

Eventual-Inc · Feb 4, 2025 · 13e8638 · 13e8638
1 parent 2e682a3
commit 13e8638
Show file tree

Hide file tree

Showing 3 changed files with 5 additions and 5 deletions.
diff --git a/src/arrow2/src/compute/cast/boolean_to.rs b/src/arrow2/src/compute/cast/boolean_to.rs
@@ -30,7 +30,7 @@ where
 /// Casts the [`BooleanArray`] to a [`Utf8Array`], casting trues to `"true"` and falses to `"false"`
 pub fn boolean_to_utf8<O: Offset>(from: &BooleanArray) -> Utf8Array<O> {
     let iter = from.values().iter().map(|x| if x { "true" } else { "false" });
-    Utf8Array::from_trusted_len_values_iter(iter)
+    Utf8Array::from_trusted_len_values_iter(iter).with_validity(from.validity().cloned())
 }
 
 pub(super) fn boolean_to_utf8_dyn<O: Offset>(array: &dyn Array) -> Result<Box<dyn Array>> {
@@ -41,7 +41,7 @@ pub(super) fn boolean_to_utf8_dyn<O: Offset>(array: &dyn Array) -> Result<Box<dy
 /// Casts the [`BooleanArray`] to a [`BinaryArray`], casting trues to `"1"` and falses to `"0"`
 pub fn boolean_to_binary<O: Offset>(from: &BooleanArray) -> BinaryArray<O> {
     let iter = from.values().iter().map(|x| if x { b"1" } else { b"0" });
-    BinaryArray::from_trusted_len_values_iter(iter)
+    BinaryArray::from_trusted_len_values_iter(iter).with_validity(from.validity().cloned())
 }
 
 pub(super) fn boolean_to_binary_dyn<O: Offset>(array: &dyn Array) -> Result<Box<dyn Array>> {

diff --git a/tests/expressions/test_list_.py b/tests/expressions/test_list_.py
@@ -68,7 +68,7 @@ def test_list_constructor_heterogeneous():
 def test_list_constructor_heterogeneous_with_cast():
     df = daft.from_pydict({"x": [1, 2, 3], "y": [True, True, False]})
     df = df.select(list_(col("x").cast(dt.string()), col("y").cast(dt.string())).alias("strs"))
-    assert df.to_pydict() == {"strs": [["1", "1"], ["2", "1"], ["3", "0"]]}
+    assert df.to_pydict() == {"strs": [["1", "true"], ["2", "true"], ["3", "false"]]}
 
 
 def test_list_constructor_mixed_null_first():

diff --git a/tests/series/test_cast.py b/tests/series/test_cast.py
@@ -70,8 +70,8 @@ def test_series_cast_null(target_dtype) -> None:
     [
         (DataType.null(), [None, None, None]),
         (DataType.bool(), [True, False, None]),
-        (DataType.string(), ["true", "false", "false"]),
-        (DataType.binary(), [b"1", b"0", b"0"]),
+        (DataType.string(), ["true", "false", None]),
+        (DataType.binary(), [b"1", b"0", None]),
         (DataType.python(), [True, False, None]),
     ]
     + [(dtype, [1, 0, None]) for dtype in daft_int_types]