Python: categorical to list

py-polars/polars/eager/series.py

@@ -1681,10 +1681,15 @@ def cast(self, dtype: Type[DataType], strict: bool = True) -> "Series":
             dtype = Float64
         return wrap_s(self._s.cast(str(dtype), strict))
 
-    def to_list(self) -> tp.List[Optional[Any]]:
+    def to_list(self, use_pyarrow: bool = False) -> tp.List[Optional[Any]]:
         """
         Convert this Series to a Python List. This operation clones data.
 
+        Parameters
+        ----------
+        use_pyarrow
+            Use pyarrow for the conversion.
+
         Examples
         --------
         >>> s = pl.Series("a", [1, 2, 3])
@@ -1694,8 +1699,7 @@ def to_list(self) -> tp.List[Optional[Any]]:
         <class 'list'>
 
         """
-        # maybe we should not use pyarrow at all.
-        if (self.dtype != Object) and _PYARROW_AVAILABLE:
+        if use_pyarrow:
             return self.to_arrow().to_pylist()
         return self._s.to_list()
 

py-polars/src/series.rs

@@ -649,7 +649,7 @@ impl PySeries {
         };
 
         let pylist = match series.dtype() {
-            DataType::Categorical => PyList::new(python, series.categorical().unwrap()),
+            DataType::Categorical => PyList::new(python, series.categorical().unwrap().iter_str()),
             DataType::Object(_) => {
                 let v = PyList::empty(python);
                 for i in 0..series.len() {

py-polars/tests/test_interop.py

@@ -74,3 +74,10 @@ def test_from_pandas_nested_list():
     pldf = pl.from_pandas(pddf)
     print(pldf)
     assert pldf.shape == (4, 2)
+
+
+def test_from_pandas_categorical_none():
+    s = pd.Series(["a", "b", "c", pd.NA], dtype="category")
+    out = pl.from_pandas(s)
+    assert out.dtype == pl.Categorical
+    assert out.to_list() == ["a", "b", "c", None]

py-polars/tests/test_sort.py

@@ -13,7 +13,7 @@ def test_sort_dates_multiples():
                     "2021-01-02 00:00:00",
                     "2021-01-03 00:00:00",
                 ],
-            ).str.strptime(pl.datatypes.Datetime, "%Y-%m-%d %T"),
+            ).str.strptime(pl.Datetime, "%Y-%m-%d %T"),
             pl.Series("values", [5, 4, 3, 2, 1]),
         ]
     )