Inserting a cudf.NA into a DataFrame (#8923)

Addressing #8922 - just have to check if it's a `cudf.NA` and set to `None` if necessary.

Authors:
  - Sarah Yurick (https://github.com/sarahyurick)

Approvers:
  - https://github.com/brandon-b-miller
  - Ram (Ramakrishna Prabhu) (https://github.com/rgsl888prabhu)

URL: #8923

python/cudf/cudf/tests/test_dataframe.py

@@ -5054,6 +5054,18 @@ def test_insert(data):
     assert_eq(pdf, gdf)
 
 
+@pytest.mark.parametrize(
+    "data", [{"A": [1, 2, 3], "B": ["a", "b", "c"]}],
+)
+def test_insert_NA(data):
+    pdf = pd.DataFrame.from_dict(data)
+    gdf = cudf.DataFrame.from_pandas(pdf)
+
+    pdf["C"] = pd.NA
+    gdf["C"] = cudf.NA
+    assert_eq(pdf, gdf)
+
+
 def test_cov():
     gdf = cudf.datasets.randomdata(10)
     pdf = gdf.to_pandas()

python/cudf/cudf/utils/utils.py

@@ -42,10 +42,7 @@ def scalar_broadcast_to(scalar, size, dtype=None):
     if isinstance(size, (tuple, list)):
         size = size[0]
 
-    if scalar is None or (
-        isinstance(scalar, (np.datetime64, np.timedelta64))
-        and np.isnat(scalar)
-    ):
+    if cudf._lib.scalar._is_null_host_scalar(scalar):
         if dtype is None:
             dtype = "object"
         return column.column_empty(size, dtype=dtype, masked=True)