Skip to content

fix: Preserve pandas column name attribute #2363

New issue

Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.

By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.

Already on GitHub? Sign in to your account

Open
wants to merge 17 commits into
base: main
Choose a base branch
from
Open
Show file tree
Hide file tree
Changes from all commits
Commits
File filter

Filter by extension

Filter by extension

Conversations
Failed to load comments.
Loading
Jump to
Jump to file
Failed to load files.
Loading
Diff view
Diff view
15 changes: 12 additions & 3 deletions narwhals/_dask/dataframe.py
Copy link
Member Author

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

Dask select and with_columns have no issues since we use .assign method for both!

Original file line number Diff line number Diff line change
Expand Up @@ -12,6 +12,7 @@
from narwhals._dask.utils import add_row_index
from narwhals._dask.utils import evaluate_exprs
from narwhals._pandas_like.utils import native_to_narwhals_dtype
from narwhals._pandas_like.utils import rename_axis
from narwhals._pandas_like.utils import select_columns_by_name
from narwhals.typing import CompliantDataFrame
from narwhals.typing import CompliantLazyFrame
Expand Down Expand Up @@ -112,10 +113,18 @@ def collect(
if backend is None or backend is Implementation.PANDAS:
from narwhals._pandas_like.dataframe import PandasLikeDataFrame

implementation = Implementation.PANDAS
backend_version = parse_version(pd)

return PandasLikeDataFrame(
result,
implementation=Implementation.PANDAS,
backend_version=parse_version(pd),
rename_axis(
result,
implementation=implementation,
backend_version=backend_version,
columns=self.native.columns.name,
),
implementation=implementation,
backend_version=backend_version,
version=self._version,
validate_column_names=True,
)
Expand Down
16 changes: 14 additions & 2 deletions narwhals/_pandas_like/dataframe.py
Original file line number Diff line number Diff line change
Expand Up @@ -25,6 +25,7 @@
from narwhals._pandas_like.utils import object_native_to_narwhals_dtype
from narwhals._pandas_like.utils import pivot_table
from narwhals._pandas_like.utils import rename
from narwhals._pandas_like.utils import rename_axis
from narwhals._pandas_like.utils import select_columns_by_name
from narwhals._pandas_like.utils import set_index
from narwhals.dependencies import is_numpy_array_1d
Expand Down Expand Up @@ -120,6 +121,7 @@ def __init__(
validate_backend_version(self._implementation, self._backend_version)
if validate_column_names:
check_column_names_are_unique(native_dataframe.columns)
self._native_columns_name = native_dataframe.columns.name

@classmethod
def from_arrow(cls, data: IntoArrowTable, /, *, context: _FullContext) -> Self:
Expand Down Expand Up @@ -251,7 +253,12 @@ def _with_version(self, version: Version) -> Self:

def _with_native(self, df: Any, *, validate_column_names: bool = True) -> Self:
return self.__class__(
df,
rename_axis(
df,
implementation=self._implementation,
backend_version=self._backend_version,
columns=self._native_columns_name,
),
implementation=self._implementation,
backend_version=self._backend_version,
version=self._version,
Expand Down Expand Up @@ -621,7 +628,12 @@ def collect(
import pandas as pd # ignore-banned-import

return PandasLikeDataFrame(
self.to_pandas(),
rename_axis(
self.to_pandas(),
implementation=self._implementation,
backend_version=self._backend_version,
columns=self._native_columns_name,
),
implementation=Implementation.PANDAS,
backend_version=parse_version(pd),
version=self._version,
Expand Down
15 changes: 15 additions & 0 deletions narwhals/_pandas_like/utils.py
Original file line number Diff line number Diff line change
Expand Up @@ -736,6 +736,21 @@ def check_column_names_are_unique(columns: pd.Index[str]) -> None:
raise DuplicateError(msg)


def rename_axis(
obj: T,
*args: Any,
implementation: Implementation,
backend_version: tuple[int, ...],
**kwargs: Any,
) -> T:
"""Wrapper around pandas' rename_axis so that we can set `copy` based on implementation/version."""
if implementation is Implementation.PANDAS and (
backend_version >= (3,)
): # pragma: no cover
return obj.rename_axis(*args, **kwargs) # type: ignore[attr-defined]
return obj.rename_axis(*args, **kwargs, copy=False) # type: ignore[attr-defined]


class PandasLikeSeriesNamespace(EagerSeriesNamespace["PandasLikeSeries", Any]):
@property
def implementation(self) -> Implementation:
Expand Down
2 changes: 1 addition & 1 deletion narwhals/translate.py
Original file line number Diff line number Diff line change
Expand Up @@ -475,8 +475,8 @@ def _from_native_impl( # noqa: PLR0915
return DataFrame(
PandasLikeDataFrame(
native_object,
backend_version=parse_version(pd),
implementation=Implementation.PANDAS,
backend_version=parse_version(pd),
Comment on lines -478 to +479
Copy link
Member Author

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

Mental sanity for symmetry with other pandas-like and order of the spec πŸ˜…

version=version,
validate_column_names=True,
),
Expand Down
12 changes: 7 additions & 5 deletions tests/frame/join_test.py
Original file line number Diff line number Diff line change
Expand Up @@ -9,7 +9,6 @@
import pytest

import narwhals as nw
from narwhals.utils import Implementation
from tests.utils import DUCKDB_VERSION
from tests.utils import PANDAS_VERSION
from tests.utils import POLARS_VERSION
Expand Down Expand Up @@ -237,11 +236,14 @@ def test_cross_join_suffix(constructor: Constructor, suffix: str) -> None:


def test_cross_join_non_pandas() -> None:
_ = pytest.importorskip("modin")

import modin.pandas as mpd

data = {"antananarivo": [1, 3, 2]}
df = nw.from_native(pd.DataFrame(data))
# HACK to force testing for a non-pandas codepath
df._compliant_frame._implementation = Implementation.MODIN
result = df.join(df, how="cross")
df1 = nw.from_native(mpd.DataFrame(pd.DataFrame(data)), eager_only=True)
df2 = nw.from_native(mpd.DataFrame(pd.DataFrame(data)), eager_only=True)
result = df1.join(df2, how="cross")
expected = {
"antananarivo": [1, 1, 1, 3, 3, 3, 2, 2, 2],
"antananarivo_right": [1, 3, 2, 1, 3, 2, 1, 3, 2],
Expand Down
28 changes: 28 additions & 0 deletions tests/preserve_pandas_like_columns_name_attr_test.py
Original file line number Diff line number Diff line change
@@ -0,0 +1,28 @@
from __future__ import annotations

from typing import TYPE_CHECKING

import pytest

import narwhals.stable.v1 as nw

if TYPE_CHECKING:
from tests.utils import Constructor


def test_ops_preserve_column_index_name(constructor: Constructor) -> None:
if not any(x in str(constructor) for x in ("pandas", "modin", "cudf", "dask")):
pytest.skip(
reason="Dataframe columns is a list and do not have a `name` like a pandas Index does"
)

data = {"a": [1, 3, 2], "b": [4, 4, 6], "z": [7.0, 8.0, 9.0]}
df_native = constructor(data)
df_native.columns.name = "foo" # type: ignore[union-attr]

df = nw.from_native(df_native)

result = df.with_columns(b=nw.col("a") + 1, c=nw.col("a") * 2).select("c", "b")
Copy link
Member Author

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

I wanted to concatenate two methods here, mostly for the sake of it.
Might be worth reparametrizing it


assert result.to_native().columns.name == "foo" # type: ignore[union-attr]
assert result.lazy().collect(backend="pandas").to_native().columns.name == "foo"
Loading