feat(python): Enable Null datatype and null values by default in pa…

…rametric testing

py-polars/tests/unit/interchange/test_roundtrip.py

@@ -36,7 +36,12 @@
 ]
 
 
-@given(dataframes(allowed_dtypes=protocol_dtypes))
+@given(
+    dataframes(
+        allowed_dtypes=protocol_dtypes,
+        allow_null=False,  # Bug: https://github.com/pola-rs/polars/issues/16190
+    )
+)
 def test_to_dataframe_pyarrow_parametric(df: pl.DataFrame) -> None:
     dfi = df.__dataframe__()
     df_pa = pa.interchange.from_dataframe(dfi)
@@ -71,7 +76,12 @@ def test_to_dataframe_pyarrow_zero_copy_parametric(df: pl.DataFrame) -> None:
 @pytest.mark.filterwarnings(
     "ignore:.*PEP3118 format string that does not match its itemsize:RuntimeWarning"
 )
-@given(dataframes(allowed_dtypes=protocol_dtypes))
+@given(
+    dataframes(
+        allowed_dtypes=protocol_dtypes,
+        allow_null=False,  # Bug: https://github.com/pola-rs/polars/issues/16190
+    )
+)
 def test_to_dataframe_pandas_parametric(df: pl.DataFrame) -> None:
     dfi = df.__dataframe__()
     df_pd = pd.api.interchange.from_dataframe(dfi)
@@ -94,6 +104,7 @@ def test_to_dataframe_pandas_parametric(df: pl.DataFrame) -> None:
             pl.Categorical,
         ],
         chunked=False,
+        allow_null=False,  # Bug: https://github.com/pola-rs/polars/issues/16190
     )
 )
 def test_to_dataframe_pandas_zero_copy_parametric(df: pl.DataFrame) -> None:
@@ -193,6 +204,7 @@ def test_from_dataframe_pandas_zero_copy_parametric(df: pl.DataFrame) -> None:
         # Empty string columns cause an error due to a bug in pandas.
         # https://github.com/pandas-dev/pandas/issues/56703
         min_size=1,
+        allow_null=False,  # Bug: https://github.com/pola-rs/polars/issues/16190
     )
 )
 def test_from_dataframe_pandas_native_parametric(df: pl.DataFrame) -> None:
@@ -217,6 +229,7 @@ def test_from_dataframe_pandas_native_parametric(df: pl.DataFrame) -> None:
         # https://github.com/pandas-dev/pandas/issues/56700
         min_size=1,
         chunked=False,
+        allow_null=False,  # Bug: https://github.com/pola-rs/polars/issues/16190
     )
 )
 def test_from_dataframe_pandas_native_zero_copy_parametric(df: pl.DataFrame) -> None:

py-polars/tests/unit/interop/numpy/test_to_numpy_series.py

@@ -326,11 +326,19 @@ def test_series_to_numpy_temporal() -> None:
 
 @given(
     s=series(
-        min_size=1, max_size=10, excluded_dtypes=[pl.Categorical, pl.List, pl.Struct]
+        min_size=1,
+        max_size=10,
+        excluded_dtypes=[
+            pl.Categorical,
+            pl.List,
+            pl.Struct,
+            pl.Datetime("ms"),
+            pl.Duration("ms"),
+        ],
+        allow_null=False,
     ).filter(
         lambda s: (
-            getattr(s.dtype, "time_unit", None) != "ms"
-            and not (s.dtype == pl.String and s.str.contains("\x00").any())
+            not (s.dtype == pl.String and s.str.contains("\x00").any())
             and not (s.dtype == pl.Binary and s.bin.contains(b"\x00").any())
         )
     ),
@@ -345,6 +353,7 @@ def test_series_to_numpy(s: pl.Series) -> None:
         pl.Datetime("us"): "datetime64[us]",
         pl.Duration("ns"): "timedelta64[ns]",
         pl.Duration("us"): "timedelta64[us]",
+        pl.Null(): np.float32,
     }
     np_dtype = dtype_map.get(s.dtype)  # type: ignore[call-overload]
     expected = np.array(values, dtype=np_dtype)

py-polars/tests/unit/operations/namespaces/temporal/test_datetime.py

@@ -1350,7 +1350,7 @@ def test_series_duration_timeunits(
 
 
 @given(
-    s=series(min_size=1, max_size=10, dtype=pl.Datetime),
+    s=series(min_size=1, max_size=10, dtype=pl.Datetime, allow_null=False),
 )
 def test_series_datetime_timeunits(
     s: pl.Series,

py-polars/tests/unit/series/buffers/test_from_buffer.py

@@ -14,6 +14,7 @@
     s=series(
         allowed_dtypes=(pl.INTEGER_DTYPES | pl.FLOAT_DTYPES | {pl.Boolean}),
         chunked=False,
+        allow_null=False,
     )
 )
 def test_series_from_buffer(s: pl.Series) -> None:

py-polars/tests/unit/testing/parametric/strategies/test_core.py

@@ -23,7 +23,6 @@
 def test_series_defaults(s: pl.Series) -> None:
     assert isinstance(s, pl.Series)
     assert s.name == ""
-    assert s.null_count() == 0
 
 
 @given(s=series(name="hello"))
@@ -39,7 +38,7 @@ def test_series_dtype(data: st.DataObject) -> None:
     assert s.dtype == dtype
 
 
-@given(s=series(dtype=pl.Enum))
+@given(s=series(dtype=pl.Enum, allow_null=False))
 @settings(max_examples=5)
 def test_series_dtype_enum(s: pl.Series) -> None:
     assert isinstance(s.dtype, pl.Enum)
@@ -58,9 +57,15 @@ def test_series_size_range(s: pl.Series) -> None:
     assert 3 <= s.len() <= 8
 
 
-@given(s=series(allow_null=True))
-def test_series_allow_null(s: pl.Series) -> None:
-    assert 0 <= s.null_count() <= s.len()
+@given(s=series(allow_null=False))
+def test_series_allow_null_false(s: pl.Series) -> None:
+    assert s.null_count() == 0
+    assert s.dtype != pl.Null
+
+
+@given(s=series(allowed_dtypes=[pl.Null], allow_null=False))
+def test_series_allow_null_allowed_dtypes(s: pl.Series) -> None:
+    assert s.dtype == pl.Null
 
 
 @given(df=dataframes())
@@ -121,6 +126,7 @@ def test_dataframes_allow_null_override(df: pl.DataFrame) -> None:
         # generate lazyframes with at least one row
         lazy=True,
         min_size=1,
+        allow_null=False,
         # test mix & match of bulk-assigned cols with custom cols
         cols=[column(n, dtype=pl.UInt8, unique=True) for n in ["a", "b"]],
         include_cols=[
@@ -190,7 +196,8 @@ def test_allow_infinities_deprecated(data: st.DataObject) -> None:
                     min_len=1,
                 ),
             ),
-        ]
+        ],
+        allow_null=False,
     ),
 )
 def test_dataframes_nested_strategies(df: pl.DataFrame) -> None:
@@ -255,9 +262,12 @@ def test_chunking(
 
 @given(
     df=dataframes(
-        allowed_dtypes=[pl.Float32, pl.Float64], max_cols=4, allow_infinity=False
+        allowed_dtypes=[pl.Float32, pl.Float64],
+        max_cols=4,
+        allow_null=False,
+        allow_infinity=False,
     ),
-    s=series(dtype=pl.Float64, allow_infinity=False),
+    s=series(dtype=pl.Float64, allow_null=False, allow_infinity=False),
 )
 def test_infinities(
     df: pl.DataFrame,