empty strings -> nat

Author: MarcoGorelli

doc/source/whatsnew/v2.0.0.rst

@@ -776,6 +776,7 @@ Datetimelike
 - Bug in rendering :class:`DatetimeIndex` and :class:`Series` and :class:`DataFrame` with timezone-aware dtypes with ``dateutil`` or ``zoneinfo`` timezones near daylight-savings transitions (:issue:`49684`)
 - Bug in :func:`to_datetime` was raising ``ValueError`` when parsing :class:`Timestamp`, ``datetime.datetime``, ``datetime.date``, or ``np.datetime64`` objects when non-ISO8601 ``format`` was passed (:issue:`49298`, :issue:`50036`)
 - Bug in :class:`Timestamp` was showing ``UserWarning`` which was not actionable by users (:issue:`50232`)
+- Bug in :func:`to_datetime` was raising ``ValueError`` when parsing empty string and non-ISO8601 format was passed. Now, empty strings will be parsed as :class:`NaT`, for compatibility with how is done for ISO8601 formats (:issue:`50251`)
 -
 
 Timedelta

pandas/_libs/tslibs/strptime.pyx

@@ -153,7 +153,7 @@ def array_strptime(
     for i in range(n):
         val = values[i]
         if isinstance(val, str):
-            if val in nat_strings:
+            if len(val) == 0 or val in nat_strings:
                 iresult[i] = NPY_NAT
                 continue
         elif checknull_with_nat_and_na(val):

pandas/tests/tools/test_to_datetime.py

@@ -2976,24 +2976,17 @@ def test_na_to_datetime(nulls_fixture, klass):
         assert result[0] is NaT
 
 
-def test_empty_string_datetime_coerce_format():
-    # GH13044
+@pytest.mark.parametrize("errors", ["raise", "coerce", "ignore"])
+def test_empty_string_datetime_coerce_format(errors):
+    # GH13044, GH50251
     td = Series(["03/24/2016", "03/25/2016", ""])
     format = "%m/%d/%Y"
 
     # coerce empty string to pd.NaT
-    result = to_datetime(td, format=format, errors="coerce")
+    result = to_datetime(td, format=format, errors=errors)
     expected = Series(["2016-03-24", "2016-03-25", NaT], dtype="datetime64[ns]")
     tm.assert_series_equal(expected, result)
 
-    # raise an exception in case a format is given
-    with pytest.raises(ValueError, match="does not match format"):
-        to_datetime(td, format=format, errors="raise")
-
-    # still raise an exception in case no format is given
-    with pytest.raises(ValueError, match="does not match format"):
-        to_datetime(td, errors="raise")
-
 
 def test_empty_string_datetime_coerce__unit():
     # GH13044