FIX: DatetimeIndex.array should return DatetimeArray instead of Exten…

…sionArray

- Override array property in DatetimeIndex to return DatetimeArray
- Add test case to verify the correct type is returned
- Fixes issue where type checkers incorrectly identified return type

pandas-stubs/core/indexes/datetimes.pyi

@@ -20,6 +20,7 @@ from pandas import (
     TimedeltaIndex,
     Timestamp,
 )
+from pandas.core.arrays import DatetimeArray
 from pandas.core.indexes.accessors import DatetimeIndexProperties
 from pandas.core.indexes.datetimelike import DatetimeTimedeltaMixin
 from pandas.core.series import (
@@ -61,6 +62,11 @@ class DatetimeIndex(
         name: Hashable = ...,
     ) -> Self: ...
     def __reduce__(self): ...
+
+    # Override the array property to return DatetimeArray instead of ExtensionArray
+    @property
+    def array(self) -> DatetimeArray: ...
+
     # various ignores needed for mypy, as we do want to restrict what can be used in
     # arithmetic for these types
     @overload  # type: ignore[override]

tests/indexes/test_indexes.py

@@ -14,6 +14,8 @@
 import pandas as pd
 from pandas.core.arrays.categorical import Categorical
 from pandas.core.indexes.base import Index
+from pandas.core.arrays import DatetimeArray
+from typing import assert_type
 from typing_extensions import (
     Never,
     assert_type,
@@ -1516,3 +1518,15 @@ def test_period_index_asof_locs() -> None:
         assert_type(idx.asof_locs(where, mask), np_1darray[np.intp]),
         np_1darray[np.intp],
     )
+
+
+def test_datetime_index_array_property():
+    """Test that DatetimeIndex.array returns DatetimeArray instead of ExtensionArray."""
+    # Test with to_datetime
+    arr = pd.to_datetime(["2020-01-01", "2020-01-02"]).array
+    assert_type(arr, DatetimeArray)
+
+    # Test with DatetimeIndex directly
+    dt_index = pd.DatetimeIndex(["2020-01-01", "2020-01-02"])
+    assert_type(dt_index.array, DatetimeArray)
+