[BUG] examples/tutorial/01-preprocess.ipynb: Convert timestamp from datetime - NotImplementedError: cuDF does not yet support timezone-aware datetimes

Bug description

raw_df['event_time_dt'] = raw_df['event_time'].astype('datetime64[s]')
raw_df['event_time_ts']= raw_df['event_time_dt'].astype('int')
raw_df.head()

---

NotImplementedError Traceback (most recent call last)
Cell In[4], line 4
1 #import datetime
2 #raw_df['event_time'] = cudf.to_datetime(raw_df['event_time'], format='%Y-%m-%d %H:%M:%S')
----> 4 raw_df['event_time_dt'] = raw_df['event_time'].astype('datetime64[s]')
5 raw_df['event_time_ts']= raw_df['event_time_dt'].astype('int')
6 raw_df.head()

File ~/miniconda3/lib/python3.10/site-packages/nvtx/nvtx.py:116, in annotate.call..inner(*args, **kwargs)
113 @wraps(func)
114 def inner(*args, **kwargs):
115 libnvtx_push_range(self.attributes, self.domain.handle)
--> 116 result = func(*args, **kwargs)
117 libnvtx_pop_range(self.domain.handle)
118 return result

File ~/miniconda3/lib/python3.10/site-packages/cudf/core/series.py:2102, in Series.astype(self, dtype, copy, errors)
2100 else:
2101 dtype = {self.name: dtype}
-> 2102 return super().astype(dtype, copy, errors)

File ~/miniconda3/lib/python3.10/site-packages/cudf/core/indexed_frame.py:5009, in IndexedFrame.astype(self, dtype, copy, errors)
5007 except Exception as e:
5008 if errors == "raise":
-> 5009 raise e
5010 return self
5012 return self._from_data(data, index=self._index)

File ~/miniconda3/lib/python3.10/site-packages/cudf/core/indexed_frame.py:5006, in IndexedFrame.astype(self, dtype, copy, errors)
5003 raise ValueError("invalid error value specified")
5005 try:
-> 5006 data = super().astype(dtype, copy)
5007 except Exception as e:
5008 if errors == "raise":

File ~/miniconda3/lib/python3.10/site-packages/nvtx/nvtx.py:116, in annotate.call..inner(*args, **kwargs)
113 @wraps(func)
114 def inner(*args, **kwargs):
115 libnvtx_push_range(self.attributes, self.domain.handle)
--> 116 result = func(*args, **kwargs)
117 libnvtx_pop_range(self.domain.handle)
118 return result

File ~/miniconda3/lib/python3.10/site-packages/cudf/core/frame.py:272, in Frame.astype(self, dtype, copy)
270 @_cudf_nvtx_annotate
271 def astype(self, dtype, copy: bool = False):
--> 272 result_data = {
273 col_name: col.astype(dtype.get(col_name, col.dtype), copy=copy)
274 for col_name, col in self._data.items()
275 }
277 return ColumnAccessor(
278 data=result_data,
279 multiindex=self._data.multiindex,
(...)
283 verify=False,
284 )

File ~/miniconda3/lib/python3.10/site-packages/cudf/core/frame.py:273, in (.0)
270 @_cudf_nvtx_annotate
271 def astype(self, dtype, copy: bool = False):
272 result_data = {
--> 273 col_name: col.astype(dtype.get(col_name, col.dtype), copy=copy)
274 for col_name, col in self._data.items()
275 }
277 return ColumnAccessor(
278 data=result_data,
279 multiindex=self._data.multiindex,
(...)
283 verify=False,
284 )

File ~/miniconda3/lib/python3.10/site-packages/cudf/core/column/column.py:1002, in ColumnBase.astype(self, dtype, copy)
1000 return col.as_decimal_column(dtype)
1001 elif np.issubdtype(cast(Any, dtype), np.datetime64):
-> 1002 return col.as_datetime_column(dtype)
1003 elif np.issubdtype(cast(Any, dtype), np.timedelta64):
1004 return col.as_timedelta_column(dtype)

File ~/miniconda3/lib/python3.10/site-packages/cudf/core/column/string.py:5749, in StringColumn.as_datetime_column(self, dtype, format)
5742 return cast(
5743 "cudf.core.column.DatetimeColumn",
5744 column.column_empty(
5745 len(self), dtype=out_dtype, masked=True
5746 ),
5747 )
5748 else:
-> 5749 format = datetime.infer_format(
5750 self.apply_boolean_mask(self.notnull()).element_indexing(0)
5751 )
5753 if format.endswith("%z"):
5754 raise NotImplementedError(
5755 "cuDF does not yet support timezone-aware datetimes"
5756 )

File ~/miniconda3/lib/python3.10/site-packages/cudf/core/column/datetime.py:108, in infer_format(element, **kwargs)
106 if fmt is not None:
107 if "%z" in fmt or "%Z" in fmt:
--> 108 raise NotImplementedError(
109 "cuDF does not yet support timezone-aware datetimes"
110 )
111 if ".%f" not in fmt:
112 # For context read:
113 # https://github.com/pandas-dev/pandas/issues/52418
114 # We cannot rely on format containing only %f
115 # c++/libcudf expects .%3f, .%6f, .%9f
116 # Logic below handles those cases well.
117 return fmt

NotImplementedError: cuDF does not yet support timezone-aware datetimes

Steps/Code to reproduce bug

1. just run the code raw_df['event_time'].astype('datetime64[s]')

Expected behavior

No error

Environment details

• Platform: WSL2 + win 11
• Python version: 3.10.14
• cuDF version: '24.04.00'
• nvtabular version: '23.08.00'
• PyTorch version (GPU): '2.2.2+cu121'

Additional context

Problem solved by adding one line

## added one line 
raw_df['event_time'] = cudf.to_datetime(raw_df['event_time'], format='%Y-%m-%d %H:%M:%S')
##
raw_df['event_time_dt'] = raw_df['event_time'].astype('datetime64[s]')
raw_df['event_time_ts']= raw_df['event_time_dt'].astype('int')
raw_df.head()

Open examples/tutorial/01-preprocess.ipynb and inspect the cell containing the event_time conversion. Start by running the preprocessing steps with the reported environment and compare them with the explicit cudf.to_datetime workaround shown in the issue. Done means the tutorial cell runs without the timezone-aware datetime error and produces event_time_dt and event_time_ts.