Backport PR #39442 BUG: DataFrame constructor reordering elements with ndarray from datetime dtype not datetime64[ns (#39475)

(cherry picked from commit 624621b)

Author: phofl

doc/source/whatsnew/v1.2.2.rst

@@ -14,6 +14,7 @@ including other versions of pandas.
 
 Fixed regressions
 ~~~~~~~~~~~~~~~~~
+- Fixed regression in :class:`DataFrame` constructor reordering element when construction from datetime ndarray with dtype not ``"datetime64[ns]"`` (:issue:`39422`)
 - Fixed regression in :meth:`~DataFrame.to_pickle` failing to create bz2/xz compressed pickle files with ``protocol=5`` (:issue:`39002`)
 - Fixed regression in :func:`pandas.testing.assert_series_equal` and :func:`pandas.testing.assert_frame_equal` always raising ``AssertionError`` when comparing extension dtypes (:issue:`39410`)
 - Fixed regression in :meth:`~DataFrame.to_csv` opening ``codecs.StreamWriter`` in binary mode instead of in text mode and ignoring user-provided ``mode`` (:issue:`39247`)

pandas/_libs/tslibs/conversion.pyx

@@ -224,7 +224,7 @@ def ensure_datetime64ns(arr: ndarray, copy: bool=True):
 
     ivalues = arr.view(np.int64).ravel("K")
 
-    result = np.empty(shape, dtype=DT64NS_DTYPE)
+    result = np.empty_like(arr, dtype=DT64NS_DTYPE)
     iresult = result.ravel("K").view(np.int64)
 
     if len(iresult) == 0:

pandas/tests/frame/test_constructors.py

@@ -1936,6 +1936,70 @@ def test_constructor_datetimes_with_nulls(self, arr):
         expected = Series([np.dtype("datetime64[ns]")])
         tm.assert_series_equal(result, expected)
 
+    @pytest.mark.parametrize("order", ["K", "A", "C", "F"])
+    @pytest.mark.parametrize(
+        "dtype",
+        [
+            "datetime64[M]",
+            "datetime64[D]",
+            "datetime64[h]",
+            "datetime64[m]",
+            "datetime64[s]",
+            "datetime64[ms]",
+            "datetime64[us]",
+            "datetime64[ns]",
+        ],
+    )
+    def test_constructor_datetimes_non_ns(self, order, dtype):
+        na = np.array(
+            [
+                ["2015-01-01", "2015-01-02", "2015-01-03"],
+                ["2017-01-01", "2017-01-02", "2017-02-03"],
+            ],
+            dtype=dtype,
+            order=order,
+        )
+        df = DataFrame(na)
+        expected = DataFrame(
+            [
+                ["2015-01-01", "2015-01-02", "2015-01-03"],
+                ["2017-01-01", "2017-01-02", "2017-02-03"],
+            ]
+        )
+        expected = expected.astype(dtype=dtype)
+        tm.assert_frame_equal(df, expected)
+
+    @pytest.mark.parametrize("order", ["K", "A", "C", "F"])
+    @pytest.mark.parametrize(
+        "dtype",
+        [
+            "timedelta64[D]",
+            "timedelta64[h]",
+            "timedelta64[m]",
+            "timedelta64[s]",
+            "timedelta64[ms]",
+            "timedelta64[us]",
+            "timedelta64[ns]",
+        ],
+    )
+    def test_constructor_timedelta_non_ns(self, order, dtype):
+        na = np.array(
+            [
+                [np.timedelta64(1, "D"), np.timedelta64(2, "D")],
+                [np.timedelta64(4, "D"), np.timedelta64(5, "D")],
+            ],
+            dtype=dtype,
+            order=order,
+        )
+        df = DataFrame(na).astype("timedelta64[ns]")
+        expected = DataFrame(
+            [
+                [Timedelta(1, "D"), Timedelta(2, "D")],
+                [Timedelta(4, "D"), Timedelta(5, "D")],
+            ],
+        )
+        tm.assert_frame_equal(df, expected)
+
     def test_constructor_for_list_with_dtypes(self):
         # test list of lists/ndarrays
         df = DataFrame([np.arange(5) for x in range(5)])