Backport PR pandas-dev#52555 on branch 2.0.x (BUG: DatetimeArray.unit…

… when constructed from a non-nano ndarray) (pandas-dev#52608)

Backport PR pandas-dev#52555: BUG: DatetimeArray.unit when constructed from a non-nano ndarray

Co-authored-by: Luke Manley <lukemanley@gmail.com>

doc/source/whatsnew/v2.0.1.rst

@@ -30,6 +30,7 @@ Bug fixes
 - Bug in :func:`pandas.testing.assert_series_equal` where ``check_dtype=False`` would still raise for datetime or timedelta types with different resolutions (:issue:`52449`)
 - Bug in :meth:`ArrowDtype.__from_arrow__` not respecting if dtype is explicitly given (:issue:`52533`)
 - Bug in :func:`read_csv` casting PyArrow datetimes to NumPy when ``dtype_backend="pyarrow"`` and ``parse_dates`` is set causing a performance bottleneck in the process (:issue:`52546`)
+- Bug in :class:`arrays.DatetimeArray` constructor returning an incorrect unit when passed a non-nanosecond numpy datetime array (:issue:`52555`)
 
 .. ---------------------------------------------------------------------------
 .. _whatsnew_201.other:

pandas/core/arrays/datetimelike.py

@@ -1826,7 +1826,10 @@ def __init__(
             values = values._ndarray
 
         elif dtype is None:
-            dtype = self._default_dtype
+            if isinstance(values, np.ndarray) and values.dtype.kind in "Mm":
+                dtype = values.dtype
+            else:
+                dtype = self._default_dtype
 
         if not isinstance(values, np.ndarray):
             raise ValueError(

pandas/tests/arrays/datetimes/test_constructors.py

@@ -127,6 +127,13 @@ def test_copy(self):
         arr = DatetimeArray(data, copy=True)
         assert arr._ndarray is not data
 
+    @pytest.mark.parametrize("unit", ["s", "ms", "us", "ns"])
+    def test_numpy_datetime_unit(self, unit):
+        data = np.array([1, 2, 3], dtype=f"M8[{unit}]")
+        arr = DatetimeArray(data)
+        assert arr.unit == unit
+        assert arr[0].unit == unit
+
 
 class TestSequenceToDT64NS:
     def test_tz_dtype_mismatch_raises(self):