feat: support Series.dt.strftime (#453)

* feat: support Series.dt.strftime

* 🦉 Updates from OwlBot post-processor

See https://github.com/googleapis/repo-automation-bots/blob/main/packages/owl-bot/README.md

* address comments

* fix imports

* 🦉 Updates from OwlBot post-processor

See https://github.com/googleapis/repo-automation-bots/blob/main/packages/owl-bot/README.md

---------

Co-authored-by: Owl Bot <gcf-owl-bot[bot]@users.noreply.github.com>

bigframes/core/compile/scalar_op_compiler.py

@@ -613,6 +613,15 @@ def second_op_impl(x: ibis_types.Value):
     return typing.cast(ibis_types.TimestampValue, x).second().cast(ibis_dtypes.int64)
 
 
+@scalar_op_compiler.register_unary_op(ops.StrftimeOp, pass_op=True)
+def strftime_op_impl(x: ibis_types.Value, op: ops.StrftimeOp):
+    return (
+        typing.cast(ibis_types.TimestampValue, x)
+        .strftime(op.date_format)
+        .cast(ibis_dtypes.str)
+    )
+
+
 @scalar_op_compiler.register_unary_op(ops.time_op)
 def time_op_impl(x: ibis_types.Value):
     return typing.cast(ibis_types.TimestampValue, x).time()

bigframes/operations/__init__.py

@@ -415,6 +415,15 @@ def output_type(self, *input_types):
         return input_types[0]
 
 
+@dataclasses.dataclass(frozen=True)
+class StrftimeOp(UnaryOp):
+    name: typing.ClassVar[str] = "strftime"
+    date_format: str
+
+    def output_type(self, *input_types):
+        return dtypes.STRING_DTYPE
+
+
 # Binary Ops
 fillna_op = create_binary_op(name="fillna")
 cliplower_op = create_binary_op(name="clip_lower")

bigframes/operations/datetimes.py

@@ -17,6 +17,7 @@
 import datetime as dt
 from typing import Optional
 
+import bigframes_vendored.pandas.core.arrays.datetimelike as vendored_pandas_datetimelike
 import bigframes_vendored.pandas.core.indexes.accessor as vendordt
 
 from bigframes.core import log_adapter
@@ -27,7 +28,9 @@
 
 @log_adapter.class_logger
 class DatetimeMethods(
-    bigframes.operations.base.SeriesMethods, vendordt.DatetimeProperties
+    bigframes.operations.base.SeriesMethods,
+    vendordt.DatetimeProperties,
+    vendored_pandas_datetimelike.DatelikeOps,
 ):
     __doc__ = vendordt.DatetimeProperties.__doc__
 
@@ -88,3 +91,6 @@ def tz(self) -> Optional[dt.timezone]:
     def unit(self) -> str:
         # Assumption: pyarrow dtype
         return self._dtype.pyarrow_dtype.unit
+
+    def strftime(self, date_format: str) -> series.Series:
+        return self._apply_unary_op(ops.StrftimeOp(date_format=date_format))

tests/system/small/operations/test_datetimes.py

@@ -219,3 +219,50 @@ def test_dt_unit(scalars_dfs, col_name):
     pd_result = scalars_pandas_df[col_name].dt.unit
 
     assert bf_result == pd_result
+
+
+@pytest.mark.parametrize(
+    ("column", "date_format"),
+    [
+        ("timestamp_col", "%B %d, %Y, %r"),
+        ("timestamp_col", "%m-%d-%Y %H:%M"),
+        ("datetime_col", "%m-%d-%Y %H:%M"),
+        ("datetime_col", "%H:%M"),
+    ],
+)
+@skip_legacy_pandas
+def test_dt_strftime(scalars_df_index, scalars_pandas_df_index, column, date_format):
+    bf_result = scalars_df_index[column].dt.strftime(date_format).to_pandas()
+    pd_result = scalars_pandas_df_index[column].dt.strftime(date_format)
+    pd.testing.assert_series_equal(bf_result, pd_result, check_dtype=False)
+    assert bf_result.dtype == "string[pyarrow]"
+
+
+def test_dt_strftime_date():
+    bf_series = bigframes.series.Series(
+        ["2014-08-15", "2215-08-15", "2016-02-29"]
+    ).astype("date32[day][pyarrow]")
+
+    expected_result = pd.Series(["08/15/2014", "08/15/2215", "02/29/2016"])
+    bf_result = bf_series.dt.strftime("%m/%d/%Y").to_pandas()
+
+    pd.testing.assert_series_equal(
+        bf_result, expected_result, check_index_type=False, check_dtype=False
+    )
+    assert bf_result.dtype == "string[pyarrow]"
+
+
+def test_dt_strftime_time():
+    bf_series = bigframes.series.Series(
+        [143542314, 345234512341, 75543252344, 626546437654754, 8543523452345234]
+    ).astype("time64[us][pyarrow]")
+
+    expected_result = pd.Series(
+        ["00:02:23", "23:53:54", "20:59:03", "16:40:37", "08:57:32"]
+    )
+    bf_result = bf_series.dt.strftime("%X").to_pandas()
+
+    pd.testing.assert_series_equal(
+        bf_result, expected_result, check_index_type=False, check_dtype=False
+    )
+    assert bf_result.dtype == "string[pyarrow]"

third_party/bigframes_vendored/pandas/core/arrays/datetimelike.py

@@ -0,0 +1,38 @@
+# Contains code from https://github.com/pandas-dev/pandas/blob/main/pandas/core/arrays/datetimelike.py
+
+from bigframes import constants
+
+
+class DatelikeOps:
+    def strftime(self, date_format: str):
+        """
+        Convert to string Series using specified date_format.
+
+        Return a Series of formatted strings specified by date_format. Details
+        of the string format can be found in `BigQuery format elements doc
+        <%(https://cloud.google.com/bigquery/docs/reference/standard-sql/format-elements)s>`__.
+
+        **Examples:**
+
+            >>> import bigframes.pandas as bpd
+            >>> bpd.options.display.progress_bar = None
+
+            >>> s = bpd.to_datetime(
+            ...     ['2014-08-15 08:15:12', '2012-02-29 08:15:12+06:00', '2015-08-15 08:15:12+05:00'],
+            ...     utc=True
+            ... ).astype("timestamp[us, tz=UTC][pyarrow]")
+
+            >>> s.dt.strftime("%B %d, %Y, %r")
+            0      August 15, 2014, 08:15:12 AM
+            1    February 29, 2012, 02:15:12 AM
+            2      August 15, 2015, 03:15:12 AM
+            Name: 0, dtype: string
+
+        Args:
+            date_format (str):
+                Date format string (e.g. "%Y-%m-%d").
+
+        Returns:
+            bigframes.series.Series of formatted strings.
+        """
+        raise NotImplementedError(constants.ABSTRACT_METHOD_ERROR_MESSAGE)