BUG: Return numpy types from ArrowExtensionArray.to_numpy for temporal types when possible

doc/source/whatsnew/v2.2.0.rst

@@ -107,11 +107,11 @@ documentation.
 
 .. _whatsnew_220.enhancements.to_numpy_ea:
 
-ExtensionArray.to_numpy converts to suitable NumPy dtype
-^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
+``to_numpy`` for NumPy nullable and Arrow types converts to suitable NumPy dtype
+^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
 
-:meth:`ExtensionArray.to_numpy` will now convert to a suitable NumPy dtype instead
-of ``object`` dtype for nullable extension dtypes.
+``to_numpy`` for NumPy nullable and Arrow types will now convert to a
+suitable NumPy dtype instead of ``object`` dtype for nullable extension dtypes.
 
 *Old behavior:*
 
@@ -128,13 +128,17 @@ of ``object`` dtype for nullable extension dtypes.
     ser = pd.Series([1, 2, 3], dtype="Int64")
     ser.to_numpy()
 
+    ser = pd.Series([1, 2, 3], dtype="timestamp[ns][pyarrow]")
+    ser.to_numpy()
+
 The default NumPy dtype (without any arguments) is determined as follows:
 
 - float dtypes are cast to NumPy floats
 - integer dtypes without missing values are cast to NumPy integer dtypes
 - integer dtypes with missing values are cast to NumPy float dtypes and ``NaN`` is used as missing value indicator
 - boolean dtypes without missing values are cast to NumPy bool dtype
 - boolean dtypes with missing values keep object dtype
+- datetime and timedelta types are cast to Numpy datetime64 and timedelta64 types respectively and ``NaT`` is used as missing value indicator
 
 .. _whatsnew_220.enhancements.struct_accessor:
 

pandas/core/arrays/arrow/array.py

@@ -47,6 +47,7 @@
 from pandas.core import (
     algorithms as algos,
     missing,
+    ops,
     roperator,
 )
 from pandas.core.arraylike import OpsMixin
@@ -655,7 +656,11 @@ def _cmp_method(self, other, op):
                 mask = isna(self) | isna(other)
                 valid = ~mask
                 result = np.zeros(len(self), dtype="bool")
-                result[valid] = op(np.array(self)[valid], other)
+                np_array = np.array(self)
+                try:
+                    result[valid] = op(np_array[valid], other)
+                except TypeError:
+                    result = ops.invalid_comparison(np_array, other, op)
                 result = pa.array(result, type=pa.bool_())
                 result = pc.if_else(valid, result, None)
         else:
@@ -1130,7 +1135,16 @@ def searchsorted(
         if isinstance(value, ExtensionArray):
             value = value.astype(object)
         # Base class searchsorted would cast to object, which is *much* slower.
-        return self.to_numpy().searchsorted(value, side=side, sorter=sorter)
+        dtype = None
+        if isinstance(self.dtype, ArrowDtype):
+            pa_dtype = self.dtype.pyarrow_dtype
+            if (
+                pa.types.is_timestamp(pa_dtype) or pa.types.is_duration(pa_dtype)
+            ) and pa_dtype.unit == "ns":
+                # np.array[datetime/timedelta].searchsorted(datetime/timedelta)
+                # erroneously fails when numpy type resolution is nanoseconds
+                dtype = object
+        return self.to_numpy(dtype=dtype).searchsorted(value, side=side, sorter=sorter)
 
     def take(
         self,
@@ -1281,10 +1295,15 @@ def to_numpy(
 
         if pa.types.is_timestamp(pa_type) or pa.types.is_duration(pa_type):
             result = data._maybe_convert_datelike_array()
-            if dtype is None or dtype.kind == "O":
-                result = result.to_numpy(dtype=object, na_value=na_value)
+            if (pa.types.is_timestamp(pa_type) and pa_type.tz is not None) or (
+                dtype is not None and dtype.kind == "O"
+            ):
+                dtype = object
             else:
-                result = result.to_numpy(dtype=dtype)
+                # GH 55997
+                dtype = None
+                na_value = pa_type.to_pandas_dtype().type("nat", pa_type.unit)
+            result = result.to_numpy(dtype=dtype, na_value=na_value)
         elif pa.types.is_time(pa_type) or pa.types.is_date(pa_type):
             # convert to list of python datetime.time objects before
             # wrapping in ndarray

pandas/io/formats/format.py

@@ -61,7 +61,6 @@
 )
 
 from pandas.core.arrays import (
-    BaseMaskedArray,
     Categorical,
     DatetimeArray,
     ExtensionArray,
@@ -1528,10 +1527,8 @@ def _format_strings(self) -> list[str]:
         if isinstance(values, Categorical):
             # Categorical is special for now, so that we can preserve tzinfo
             array = values._internal_get_values()
-        elif isinstance(values, BaseMaskedArray):
-            array = values.to_numpy(dtype=object)
         else:
-            array = np.asarray(values)
+            array = np.asarray(values, dtype=object)
 
         fmt_values = format_array(
             array,

pandas/tests/arrays/string_/test_string.py

@@ -263,7 +263,7 @@ def test_comparison_methods_scalar_not_string(comparison_op, dtype):
     other = 42
 
     if op_name not in ["__eq__", "__ne__"]:
-        with pytest.raises(TypeError, match="not supported between"):
+        with pytest.raises(TypeError, match="Invalid comparison|not supported between"):
             getattr(a, op_name)(other)
 
         return

pandas/tests/extension/test_arrow.py

@@ -41,6 +41,7 @@
     pa_version_under13p0,
     pa_version_under14p0,
 )
+import pandas.util._test_decorators as td
 
 from pandas.core.dtypes.dtypes import (
     ArrowDtype,
@@ -266,6 +267,19 @@ def data_for_twos(data):
 
 
 class TestArrowArray(base.ExtensionTests):
+    def test_compare_scalar(self, data, comparison_op):
+        ser = pd.Series(data)
+        self._compare_other(ser, data, comparison_op, data[0])
+
+    @pytest.mark.parametrize("na_action", [None, "ignore"])
+    def test_map(self, data_missing, na_action):
+        if data_missing.dtype.kind in "mM":
+            result = data_missing.map(lambda x: x, na_action=na_action)
+            expected = data_missing.to_numpy(dtype=object)
+            tm.assert_numpy_array_equal(result, expected)
+        else:
+            super().test_map(data_missing, na_action)
+
     def test_astype_str(self, data, request):
         pa_dtype = data.dtype.pyarrow_dtype
         if pa.types.is_binary(pa_dtype):
@@ -274,8 +288,35 @@ def test_astype_str(self, data, request):
                     reason=f"For {pa_dtype} .astype(str) decodes.",
                 )
             )
+        elif (
+            pa.types.is_timestamp(pa_dtype) and pa_dtype.tz is None
+        ) or pa.types.is_duration(pa_dtype):
+            request.applymarker(
+                pytest.mark.xfail(
+                    reason="pd.Timestamp/pd.Timedelta repr different from numpy repr",
+                )
+            )
         super().test_astype_str(data)
 
+    @pytest.mark.parametrize(
+        "nullable_string_dtype",
+        [
+            "string[python]",
+            pytest.param("string[pyarrow]", marks=td.skip_if_no("pyarrow")),
+        ],
+    )
+    def test_astype_string(self, data, nullable_string_dtype, request):
+        pa_dtype = data.dtype.pyarrow_dtype
+        if (
+            pa.types.is_timestamp(pa_dtype) and pa_dtype.tz is None
+        ) or pa.types.is_duration(pa_dtype):
+            request.applymarker(
+                pytest.mark.xfail(
+                    reason="pd.Timestamp/pd.Timedelta repr different from numpy repr",
+                )
+            )
+        super().test_astype_string(data, nullable_string_dtype)
+
     def test_from_dtype(self, data, request):
         pa_dtype = data.dtype.pyarrow_dtype
         if pa.types.is_string(pa_dtype) or pa.types.is_decimal(pa_dtype):
@@ -1511,11 +1552,9 @@ def test_to_numpy_with_defaults(data):
     result = data.to_numpy()
 
     pa_type = data._pa_array.type
-    if (
-        pa.types.is_duration(pa_type)
-        or pa.types.is_timestamp(pa_type)
-        or pa.types.is_date(pa_type)
-    ):
+    if pa.types.is_duration(pa_type) or pa.types.is_timestamp(pa_type):
+        pytest.skip("Tested in test_to_numpy_temporal")
+    elif pa.types.is_date(pa_type):
         expected = np.array(list(data))
     else:
         expected = np.array(data._pa_array)
@@ -2937,26 +2976,28 @@ def test_groupby_series_size_returns_pa_int(data):
 
 
 @pytest.mark.parametrize(
-    "pa_type", tm.DATETIME_PYARROW_DTYPES + tm.TIMEDELTA_PYARROW_DTYPES
+    "pa_type", tm.DATETIME_PYARROW_DTYPES + tm.TIMEDELTA_PYARROW_DTYPES, ids=repr
 )
-def test_to_numpy_temporal(pa_type):
+@pytest.mark.parametrize("dtype", [None, object])
+def test_to_numpy_temporal(pa_type, dtype):
     # GH 53326
+    # GH 55997: Return datetime64/timedelta64 types with NaT if possible
     arr = ArrowExtensionArray(pa.array([1, None], type=pa_type))
-    result = arr.to_numpy()
+    result = arr.to_numpy(dtype=dtype)
     if pa.types.is_duration(pa_type):
-        expected = [
-            pd.Timedelta(1, unit=pa_type.unit).as_unit(pa_type.unit),
-            pd.NA,
-        ]
-        assert isinstance(result[0], pd.Timedelta)
+        value = pd.Timedelta(1, unit=pa_type.unit).as_unit(pa_type.unit)
     else:
-        expected = [
-            pd.Timestamp(1, unit=pa_type.unit, tz=pa_type.tz).as_unit(pa_type.unit),
-            pd.NA,
-        ]
-        assert isinstance(result[0], pd.Timestamp)
-    expected = np.array(expected, dtype=object)
-    assert result[0].unit == expected[0].unit
+        value = pd.Timestamp(1, unit=pa_type.unit, tz=pa_type.tz).as_unit(pa_type.unit)
+
+    if dtype == object or (pa.types.is_timestamp(pa_type) and pa_type.tz is not None):
+        na = pd.NA
+        expected = np.array([value, na], dtype=object)
+        assert result[0].unit == value.unit
+    else:
+        na = pa_type.to_pandas_dtype().type("nat", pa_type.unit)
+        value = value.to_numpy()
+        expected = np.array([value, na])
+        assert np.datetime_data(result[0])[0] == pa_type.unit
     tm.assert_numpy_array_equal(result, expected)
 
 

pandas/tests/io/formats/test_format.py

@@ -1921,7 +1921,7 @@ def dtype(self):
         series = Series(ExtTypeStub(), copy=False)
         res = repr(series)  # This line crashed before #33770 was fixed.
         expected = "\n".join(
-            ["0    [False  True]", "1    [ True False]", "dtype: DtypeStub"]
+            ["0    [False True]", "1    [True False]", "dtype: DtypeStub"]
         )
         assert res == expected