Allow Z in datetime string parsing in non pandas compat mode (#14701)

xref #14039 (comment)

Authors:
  - Matthew Roeschke (https://github.com/mroeschke)

Approvers:
  - Christopher Harris (https://github.com/cwharris)
  - Bradley Dice (https://github.com/bdice)

URL: #14701

python/cudf/cudf/core/column/datetime.py

@@ -1,4 +1,4 @@
-# Copyright (c) 2019-2023, NVIDIA CORPORATION.
+# Copyright (c) 2019-2024, NVIDIA CORPORATION.
 
 from __future__ import annotations
 
@@ -102,6 +102,9 @@ def infer_format(element: str, **kwargs) -> str:
     """
     Infers datetime format from a string, also takes cares for `ms` and `ns`
     """
+    if not cudf.get_option("mode.pandas_compatible"):
+        # We allow "Z" but don't localize it to datetime64[ns, UTC] type (yet)
+        element = element.replace("Z", "")
     fmt = _guess_datetime_format(element, **kwargs)
 
     if fmt is not None:

python/cudf/cudf/tests/test_datetime.py

@@ -1,4 +1,4 @@
-# Copyright (c) 2019-2023, NVIDIA CORPORATION.
+# Copyright (c) 2019-2024, NVIDIA CORPORATION.
 
 import datetime
 import operator
@@ -1306,8 +1306,10 @@ def test_datetime_infer_format(data, timezone, dtype):
 
         assert_eq(expected, actual)
     else:
-        with pytest.raises(NotImplementedError):
-            sr.astype(dtype)
+        with cudf.option_context("mode.pandas_compatible", True):
+            with pytest.raises(NotImplementedError):
+                # pandas doesn't allow parsing "Z" to naive type
+                sr.astype(dtype)
 
 
 def test_dateoffset_instance_subclass_check():
@@ -2308,12 +2310,22 @@ def test_format_timezone_not_implemented(code):
         )
 
 
-@pytest.mark.parametrize("tz", ["Z", "UTC-3", "+01:00"])
-def test_no_format_timezone_not_implemented(tz):
+@pytest.mark.parametrize("tz", ["UTC-3", "+01:00"])
+def test_utc_offset_not_implemented(tz):
     with pytest.raises(NotImplementedError):
         cudf.to_datetime([f"2020-01-01 00:00:00{tz}"])
 
 
+def test_Z_utc_offset():
+    with cudf.option_context("mode.pandas_compatible", True):
+        with pytest.raises(NotImplementedError):
+            cudf.to_datetime(["2020-01-01 00:00:00Z"])
+
+    result = cudf.to_datetime(["2020-01-01 00:00:00Z"])
+    expected = cudf.to_datetime(["2020-01-01 00:00:00"])
+    assert_eq(result, expected)
+
+
 @pytest.mark.parametrize("arg", [True, False])
 def test_args_not_datetime_typerror(arg):
     with pytest.raises(TypeError):