API: to_datetime allow mixed numeric/datetime with errors=coerce (pandas-dev#50453)

* API: to_datetime allow mixed numeric/datetime with errors=coerce

* GH ref

* remove unused exp

Author: jbrockmendel

doc/source/whatsnew/v2.0.0.rst

@@ -476,7 +476,7 @@ Other API changes
 - Changed behavior of :meth:`Series.quantile` and :meth:`DataFrame.quantile` with :class:`SparseDtype` to retain sparse dtype (:issue:`49583`)
 - When creating a :class:`Series` with a object-dtype :class:`Index` of datetime objects, pandas no longer silently converts the index to a :class:`DatetimeIndex` (:issue:`39307`, :issue:`23598`)
 - :meth:`Series.unique` with dtype "timedelta64[ns]" or "datetime64[ns]" now returns :class:`TimedeltaArray` or :class:`DatetimeArray` instead of ``numpy.ndarray`` (:issue:`49176`)
-- :func:`to_datetime` and :class:`DatetimeIndex` now allow sequences containing both ``datetime`` objects and numeric entries, matching :class:`Series` behavior (:issue:`49037`)
+- :func:`to_datetime` and :class:`DatetimeIndex` now allow sequences containing both ``datetime`` objects and numeric entries, matching :class:`Series` behavior (:issue:`49037`, :issue:`50453`)
 - :func:`pandas.api.dtypes.is_string_dtype` now only returns ``True`` for array-likes with ``dtype=object`` when the elements are inferred to be strings (:issue:`15585`)
 - Passing a sequence containing ``datetime`` objects and ``date`` objects to :class:`Series` constructor will return with ``object`` dtype instead of ``datetime64[ns]`` dtype, consistent with :class:`Index` behavior (:issue:`49341`)
 - Passing strings that cannot be parsed as datetimes to :class:`Series` or :class:`DataFrame` with ``dtype="datetime64[ns]"`` will raise instead of silently ignoring the keyword and returning ``object`` dtype (:issue:`24435`)

pandas/_libs/tslib.pyx

@@ -488,8 +488,6 @@ cpdef array_to_datetime(
         npy_datetimestruct dts
         NPY_DATETIMEUNIT out_bestunit
         bint utc_convert = bool(utc)
-        bint seen_integer = False
-        bint seen_datetime = False
         bint seen_datetime_offset = False
         bint is_raise = errors=="raise"
         bint is_ignore = errors=="ignore"
@@ -520,7 +518,6 @@ cpdef array_to_datetime(
                     iresult[i] = NPY_NAT
 
                 elif PyDateTime_Check(val):
-                    seen_datetime = True
                     if val.tzinfo is not None:
                         found_tz = True
                     else:
@@ -535,12 +532,10 @@ cpdef array_to_datetime(
                     result[i] = parse_pydatetime(val, &dts, utc_convert)
 
                 elif PyDate_Check(val):
-                    seen_datetime = True
                     iresult[i] = pydate_to_dt64(val, &dts)
                     check_dts_bounds(&dts)
 
                 elif is_datetime64_object(val):
-                    seen_datetime = True
                     iresult[i] = get_datetime64_nanos(val, NPY_FR_ns)
 
                 elif is_integer_object(val) or is_float_object(val):
@@ -555,7 +550,6 @@ cpdef array_to_datetime(
                             )
                         return values, tz_out
                     # these must be ns unit by-definition
-                    seen_integer = True
 
                     if val != val or val == NPY_NAT:
                         iresult[i] = NPY_NAT
@@ -688,17 +682,6 @@ cpdef array_to_datetime(
     except TypeError:
         return _array_to_datetime_object(values, errors, dayfirst, yearfirst)
 
-    if seen_datetime and seen_integer:
-        # we have mixed datetimes & integers
-
-        if is_coerce:
-            # coerce all of the integers/floats to NaT, preserve
-            # the datetimes and other convertibles
-            for i in range(n):
-                val = values[i]
-                if is_integer_object(val) or is_float_object(val):
-                    result[i] = NPY_NAT
-
     if seen_datetime_offset and not utc_convert:
         # GH#17697
         # 1) If all the offsets are equal, return one offset for

pandas/tests/tools/test_to_datetime.py

@@ -1628,29 +1628,24 @@ def test_unit_with_numeric_coerce(self, cache, exp, arr, warning):
         tm.assert_index_equal(result, expected)
 
     @pytest.mark.parametrize(
-        "exp, arr",
+        "arr",
         [
-            [
-                ["2013-01-01", "NaT", "NaT"],
-                [Timestamp("20130101"), 1.434692e18, 1.432766e18],
-            ],
-            [
-                ["NaT", "NaT", "2013-01-01"],
-                [1.434692e18, 1.432766e18, Timestamp("20130101")],
-            ],
+            [Timestamp("20130101"), 1.434692e18, 1.432766e18],
+            [1.434692e18, 1.432766e18, Timestamp("20130101")],
         ],
     )
-    def test_unit_mixed(self, cache, exp, arr):
-
+    def test_unit_mixed(self, cache, arr):
+        # GH#50453 pre-2.0 with mixed numeric/datetimes and errors="coerce"
+        #  the numeric entries would be coerced to NaT, was never clear exactly
+        #  why.
         # mixed integers/datetimes
-        expected = DatetimeIndex(exp)
+        expected = Index([Timestamp(x) for x in arr], dtype="M8[ns]")
         result = to_datetime(arr, errors="coerce", cache=cache)
         tm.assert_index_equal(result, expected)
 
         # GH#49037 pre-2.0 this raised, but it always worked with Series,
         #  was never clear why it was disallowed
         result = to_datetime(arr, errors="raise", cache=cache)
-        expected = Index([Timestamp(x) for x in arr], dtype="M8[ns]")
         tm.assert_index_equal(result, expected)
 
         result = DatetimeIndex(arr)