Skip to content

Commit

Permalink
fix: use object dtype for ARRAY columns in to_pandas() with pandas …
Browse files Browse the repository at this point in the history
…1.x (#329)

* fix: use object dtype for ARRAY columns in `to_pandas()` with pandas 1.x

* fix tests

* simplify dtype logic
  • Loading branch information
tswast authored Jan 22, 2024
1 parent c91f70c commit 374ddb5
Show file tree
Hide file tree
Showing 8 changed files with 1,451 additions and 550 deletions.
42 changes: 42 additions & 0 deletions bigframes/features.py
Original file line number Diff line number Diff line change
@@ -0,0 +1,42 @@
# Copyright 2024 Google LLC
#
# Licensed under the Apache License, Version 2.0 (the "License");
# you may not use this file except in compliance with the License.
# You may obtain a copy of the License at
#
# http://www.apache.org/licenses/LICENSE-2.0
#
# Unless required by applicable law or agreed to in writing, software
# distributed under the License is distributed on an "AS IS" BASIS,
# WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
# See the License for the specific language governing permissions and
# limitations under the License.

from __future__ import annotations

from typing import Tuple


class PandasVersions:
"""Version comparisons for pandas package"""

def __init__(self):
self._installed_version = None

@property
def installed_version(self) -> Tuple[str, ...]:
"""pandas version"""
if self._installed_version is None:
import pandas

self._installed_version = tuple(pandas.__version__.split("."))
return self._installed_version

@property
def is_arrow_list_dtype_usable(self):
"""True if pandas.ArrowDtype is usable."""
version = self.installed_version
return version[0] != "1"


PANDAS_VERSIONS = PandasVersions()
32 changes: 26 additions & 6 deletions bigframes/session/_io/pandas.py
Original file line number Diff line number Diff line change
Expand Up @@ -19,8 +19,33 @@
import pandas.arrays
import pyarrow # type: ignore
import pyarrow.compute # type: ignore
import pyarrow.types # type: ignore

import bigframes.constants
import bigframes.features


def _arrow_to_pandas_arrowdtype(
column: pyarrow.Array, dtype: pandas.ArrowDtype
) -> pandas.Series:
if (
pyarrow.types.is_list(dtype.pyarrow_dtype)
and not bigframes.features.PANDAS_VERSIONS.is_arrow_list_dtype_usable
):
# This version of pandas doesn't really support ArrowDtype
# well. See internal issue 321013333 where array type has
# several problems converting a string.
return pandas.Series(
column.to_pylist(), # type: ignore
dtype="object",
)

# Avoid conversion logic if we are backing the pandas Series by the
# arrow array.
return pandas.Series(
pandas.arrays.ArrowExtensionArray(column), # type: ignore
dtype=dtype,
)


def arrow_to_pandas(
Expand Down Expand Up @@ -75,12 +100,7 @@ def arrow_to_pandas(
)
series = pandas.Series(pd_array, dtype=dtype)
elif isinstance(dtype, pandas.ArrowDtype):
# Avoid conversion logic if we are backing the pandas Series by the
# arrow array.
series = pandas.Series(
pandas.arrays.ArrowExtensionArray(column), # type: ignore
dtype=dtype,
)
series = _arrow_to_pandas_arrowdtype(column, dtype)
else:
series = column.to_pandas(types_mapper=lambda _: dtype)

Expand Down
Loading

0 comments on commit 374ddb5

Please sign in to comment.