chore(typing): Constrained `TypeVar`(s) for `pandas` #2227

dangotbanned · 2025-03-16T21:44:52Z

What type of PR is this? (check all applicable)

Related issues

https://github.com/narwhals-dev/narwhals/pull/2224/files#diff-91099b439be9fe51e2aaf06346cb576c3980d2d20ef03ff530abcf8bdecf0f4e
- Saw this and thought I'd try it out

Checklist

Code follows style guide (ruff)
Tests added
Documented the changes

If you have comments or can explain your changes, please do so below

mypy is happy
pyright still had 7 errors
- mypy disagrees that those are errors
- I side with pyright in (3648814)

Still have 7 for `pyright`

The stubs don't return `Self` for all types, so `pyright` is right

dangotbanned · 2025-03-16T22:49:32Z

I tried for about 30 mins to get PandasLikeDataframe[DataFrameT] working.

Throwing in the towel on that for now 😞

mypy yelling

narwhals\_pandas_like\utils.py:819: error: Redundant cast to "DataFrame"  [redundant-cast]
        return cast("DataFrameT", result)
               ^
narwhals\_pandas_like\dataframe.py:233: error: Call to untyped function "get_indexer" in typed context  [no-untyped-call]
                        self._native_frame.columns.get_indexer(item[1]),
                        ^~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
narwhals\_pandas_like\dataframe.py:352: error: Incompatible return value type (got "list[dict[Hashable, Any]]", expected "list[tuple[Any, ...]] | list[dict[str, Any]]")  [return-value]
            return self._native_frame.to_dict(orient="records")  # pyright: ignore[reportReturnType]
                   ^~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
narwhals\_pandas_like\dataframe.py:389: error: Argument 1 to "object_native_to_narwhals_dtype" has incompatible type "Series[Any]"; expected "PandasLikeSeries"  [arg-type]
                    self._native_frame[col],  # pyright: ignore[reportArgumentType]
                    ^~~~~~~~~~~~~~~~~~~~~~~
narwhals\_pandas_like\dataframe.py:411: error: Return type "PandasLikeDataFrame[DataFrame]" of "select" incompatible with return type "PandasLikeDataFrame[DataFrameT]" in supertype "CompliantDataFrame"  [override]
        def select(self: PandasLikeDataFrame, *exprs: PandasLikeExpr) -> PandasLikeDataFrame:
        ^
narwhals\_pandas_like\dataframe.py:426: error: Return type "PandasLikeDataFrame[DataFrame]" of "drop_nulls" incompatible with return type "PandasLikeDataFrame[DataFrameT]" in supertype "CompliantDataFrame"  [override]
        def drop_nulls(
        ^
narwhals\_pandas_like\dataframe.py:457: error: Return type "PandasLikeDataFrame[DataFrame]" of "filter" incompatible with return type "PandasLikeDataFrame[DataFrameT]" in supertype "CompliantDataFrame"  [override]
        def filter(
        ^
narwhals\_pandas_like\dataframe.py:533: error: No overload variant of "sort_values" of "DataFrame" matches argument types "list[str]", "bool | list[bool]", "str"  [call-overload]
                df.sort_values(list(by), ascending=ascending, na_position=na_position),
                ^~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
narwhals\_pandas_like\dataframe.py:533: note: Possible overload variants:
narwhals\_pandas_like\dataframe.py:533: note:     def sort_values(self, by: str | Sequence[str], *, axis: Literal['index', 0] | Literal['columns', 1] = ..., ascending: bool | Sequence[bool] = ..., kind: Literal['quicksort', 'mergesort', 'heapsort', 'stable'] = ..., na_position: Literal['first', 'last'] = ..., ignore_index: bool = ..., inplace: Literal[True], key: Callable[..., Any] | None = ...) -> None
narwhals\_pandas_like\dataframe.py:533: note:     def sort_values(self, by: str | Sequence[str], *, axis: Literal['index', 0] | Literal['columns', 1] = ..., ascending: bool | Sequence[bool] = ..., kind: Literal['quicksort', 'mergesort', 'heapsort', 'stable'] = ..., na_position: Literal['first', 'last'] = ..., ignore_index: bool = ..., inplace: Literal[False] = ..., key: Callable[..., Any] | None = ...) -> DataFrame
narwhals\_pandas_like\dataframe.py:641: error: Argument "how" to "merge" of "DataFrame" has incompatible type "Literal['leftanti']"; expected "Literal['left', 'right', 'outer', 'inner'] | Literal['cross']"  [arg-type]
                            how="leftanti",  # pyright: ignore[reportArgumentType]
                                ^~~~~~~~~~
narwhals\_pandas_like\dataframe.py:781: error: No overload variant of "drop_duplicates" of "DataFrame" matches argument types "Sequence[str] | None", "object"  [call-overload]
                self._native_frame.drop_duplicates(subset=subset, keep=mapped_keep),
                ^~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
narwhals\_pandas_like\dataframe.py:781: note: Possible overload variants:
narwhals\_pandas_like\dataframe.py:781: note:     def drop_duplicates(self, subset: Hashable | Iterable[Hashable] | None = ..., *, keep: Literal['first', 'last'] | bool = ..., inplace: Literal[True], ignore_index: bool = ...) -> None       
narwhals\_pandas_like\dataframe.py:781: note:     def drop_duplicates(self, subset: Hashable | Iterable[Hashable] | None = ..., *, keep: Literal['first', 'last'] | bool = ..., inplace: Literal[False] = ..., ignore_index: bool = ...) -> DataFrame
narwhals\_pandas_like\dataframe.py:840: error: Incompatible return value type (got "dict[Hashable, Any]", expected "dict[str, Any]")  [return-value]
            return self._native_frame.to_dict(orient="list")  # pyright: ignore[reportReturnType]
                   ^~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
narwhals\_pandas_like\dataframe.py:896: error: "Series[Any]" not callable  [operator]
                return self._native_frame.to_pandas()  # pyright: ignore[reportCallIssue]
                       ^~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
narwhals\_pandas_like\dataframe.py:898: error: "Series[Any]" not callable  [operator]
                return self._native_frame._to_pandas()  # pyright: ignore[reportCallIssue]
                       ^~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
narwhals\_pandas_like\dataframe.py:908: error: "Series[Any]" not callable  [operator]
                return pl.from_pandas(self._native_frame.to_pandas())  # pyright: ignore[reportCallIssue]
                                      ^~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
narwhals\_pandas_like\dataframe.py:910: error: "Series[Any]" not callable  [operator]
                return pl.from_pandas(self._native_frame._to_pandas())  # pyright: ignore[reportCallIssue]
                                      ^~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
narwhals\_pandas_like\dataframe.py:989: error: Expected iterable as variadic argument  [misc]
                values = [c for c in self.columns if c not in {*on, *index}]
                                                              ^~~~~~~~~~~~~
narwhals\_pandas_like\dataframe.py:995: error: Expected iterable as variadic argument  [misc]
                    frame.groupby([*on, *index])
                                  ^~~~~~~~~~~~~
narwhals\_pandas_like\dataframe.py:1012: error: "ndarray[Any, Any]" has no attribute "to_arrow"  [attr-defined]
                    col: sorted(self._native_frame[col].unique().to_arrow().to_pylist())  # pyright: ignore[reportAttributeAccessIssue]
                                ^~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
narwhals\_pandas_like\dataframe.py:1021: error: "ndarray[Any, Any]" has no attribute "to_arrow"  [attr-defined]
                    col: self._native_frame[col].unique().to_arrow().to_pylist()  # pyright: ignore[reportAttributeAccessIssue]
                         ^~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
narwhals\_pandas_like\dataframe.py:1025: error: Value expression in dictionary comprehension has incompatible type "str"; expected type "list[Any]"  [misc]
                uniques = {col: self._native_frame[col].unique().tolist() for col in on}
                                ^~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
narwhals\_pandas_like\dataframe.py:1043: error: Incompatible types in assignment (expression has type "list[str]", variable has type "Index[str]")  [assignment]
            result.columns = new_columns
                             ^~~~~~~~~~~
narwhals\_pandas_like\dataframe.py:1049: error: "Series[Any]" not callable  [operator]
                return self._native_frame.to_arrow(preserve_index=False)  # pyright: ignore[reportCallIssue]
                       ^~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
narwhals\_pandas_like\dataframe.py:1120: error: Cannot determine type of "other_columns"  [has-type]
                exploded_frame = native_frame[[*other_columns, columns[0]]].explode(
                                                ^~~~~~~~~~~~~
narwhals\_pandas_like\dataframe.py:1129: error: Cannot determine type of "exploded_series"  [has-type]
                    plx.concat([exploded_frame, *exploded_series], axis=1)[original_columns],
                                                 ^~~~~~~~~~~~~~~
narwhals\_pandas_like\group_by.py:68: error: Need type annotation for "_grouped"  [var-annotated]
                self._grouped = native_frame.groupby(
                                ^
narwhals\_pandas_like\expr.py:267: error: No overload variant of "groupby" of "DataFrame" matches argument type "Sequence[str]"  [call-overload]
                        rolling = df._native_frame.groupby(partition_by)[  # pyright: ignore[reportArgumentType, reportCallIssue]
                                  ^~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
narwhals\_pandas_like\expr.py:267: note: Possible overload variants:
narwhals\_pandas_like\expr.py:267: note:     def groupby(self, by: str | bytes | date | datetime | timedelta | <7 more items> | complex, axis: Literal['index', 0] | Literal[_NoDefault.no_default] = ..., level: Hashable | Sequence[Hashable] 
| None = ..., as_index: Literal[True] = ..., sort: bool = ..., group_keys: bool = ..., observed: bool | Literal[_NoDefault.no_default] = ..., dropna: bool = ...) -> DataFrameGroupBy[str | bytes | date | datetime | timedelta | <7 more items> | complex, Literal[True]]
narwhals\_pandas_like\expr.py:267: note:     def groupby(self, by: str | bytes | date | datetime | timedelta | <7 more items> | complex, axis: Literal['index', 0] | Literal[_NoDefault.no_default] = ..., level: Hashable | Sequence[Hashable] 
| None = ..., as_index: Literal[False] = ..., sort: bool = ..., group_keys: bool = ..., observed: bool | Literal[_NoDefault.no_default] = ..., dropna: bool = ...) -> DataFrameGroupBy[str | bytes | date | datetime | timedelta | <7 more items> | complex, Literal[False]]
narwhals\_pandas_like\expr.py:267: note:     def groupby(self, by: DatetimeIndex, axis: Literal['index', 0] | Literal[_NoDefault.no_default] = ..., level: Hashable | Sequence[Hashable] | None = ..., as_index: Literal[True] = ..., sort: bool = ..., group_keys: bool = ..., observed: bool | Literal[_NoDefault.no_default] = ..., dropna: bool = ...) -> DataFrameGroupBy[Timestamp, Literal[True]]
narwhals\_pandas_like\expr.py:267: note:     def groupby(self, by: DatetimeIndex, axis: Literal['index', 0] | Literal[_NoDefault.no_default] = ..., level: Hashable | Sequence[Hashable] | None = ..., as_index: Literal[False] = ..., sort: bool = ..., group_keys: bool = ..., observed: bool | Literal[_NoDefault.no_default] = ..., dropna: bool = ...) -> DataFrameGroupBy[Timestamp, Literal[False]]
narwhals\_pandas_like\expr.py:267: note:     def groupby(self, by: TimedeltaIndex, axis: Literal['index', 0] | Literal[_NoDefault.no_default] = ..., level: Hashable | Sequence[Hashable] | None = ..., as_index: Literal[True] = ..., sort: bool = ..., group_keys: bool = ..., observed: bool | Literal[_NoDefault.no_default] = ..., dropna: bool = ...) -> DataFrameGroupBy[Timedelta, Literal[True]]
narwhals\_pandas_like\expr.py:267: note:     def groupby(self, by: TimedeltaIndex, axis: Literal['index', 0] | Literal[_NoDefault.no_default] = ..., level: Hashable | Sequence[Hashable] | None = ..., as_index: Literal[False] = ..., sort: bool = ..., group_keys: bool = ..., observed: bool | Literal[_NoDefault.no_default] = ..., dropna: bool = ...) -> DataFrameGroupBy[Timedelta, Literal[False]]
narwhals\_pandas_like\expr.py:267: note:     def groupby(self, by: PeriodIndex, axis: Literal['index', 0] | Literal[_NoDefault.no_default] = ..., level: Hashable | Sequence[Hashable] | None = ..., as_index: Literal[True] = ..., sort: bool = ..., group_keys: bool = ..., observed: bool | Literal[_NoDefault.no_default] = ..., dropna: bool = ...) -> DataFrameGroupBy[Period, Literal[True]]
narwhals\_pandas_like\expr.py:267: note:     def groupby(self, by: PeriodIndex, axis: Literal['index', 0] | Literal[_NoDefault.no_default] = ..., level: Hashable | Sequence[Hashable] | None = ..., as_index: Literal[False] = ..., sort: bool 
= ..., group_keys: bool = ..., observed: bool | Literal[_NoDefault.no_default] = ..., dropna: bool = ...) -> DataFrameGroupBy[Period, Literal[False]]
narwhals\_pandas_like\expr.py:267: note:     def [IntervalT: Interval[Any]] groupby(self, by: IntervalIndex[IntervalT], axis: Literal['index', 0, _NoDefault.no_default] = ..., level: Hashable | Sequence[Hashable] | None = ..., as_index: Literal[True] = ..., sort: bool = ..., group_keys: bool = ..., observed: bool | Literal[_NoDefault.no_default] = ..., dropna: bool = ...) -> DataFrameGroupBy[IntervalT, Literal[True]]
narwhals\_pandas_like\expr.py:267: note:     def [IntervalT: Interval[Any]] groupby(self, by: IntervalIndex[IntervalT], axis: Literal['index', 0, _NoDefault.no_default] = ..., level: Hashable | Sequence[Hashable] | None = ..., as_index: Literal[False] = ..., sort: bool = ..., group_keys: bool = ..., observed: bool | Literal[_NoDefault.no_default] = ..., dropna: bool = ...) -> DataFrameGroupBy[IntervalT, Literal[False]]
narwhals\_pandas_like\expr.py:267: note:     def groupby(self, by: MultiIndex | tuple[Any, ...] | list[Any] | ufunc | Callable[..., Any] | list[ufunc | Callable[..., Any]] | list[Series[Any]] | <7 more items> | None = ..., axis: Literal['index', 0] | Literal[_NoDefault.no_default] = ..., level: Hashable | Sequence[Hashable] | None = ..., as_index: Literal[True] = ..., sort: bool = ..., group_keys: bool = ..., observed: bool | Literal[_NoDefault.no_default] = ..., dropna: bool = ...) -> DataFrameGroupBy[tuple[Any, ...], Literal[True]]
narwhals\_pandas_like\expr.py:267: note:     def groupby(self, by: MultiIndex | tuple[Any, ...] | list[Any] | ufunc | Callable[..., Any] | list[ufunc | Callable[..., Any]] | list[Series[Any]] | <7 more items> | None = ..., axis: Literal['index', 0] | Literal[_NoDefault.no_default] = ..., level: Hashable | Sequence[Hashable] | None = ..., as_index: Literal[False] = ..., sort: bool = ..., group_keys: bool = ..., observed: bool | Literal[_NoDefault.no_default] = ..., dropna: bool = ...) -> DataFrameGroupBy[tuple[Any, ...], Literal[False]]
narwhals\_pandas_like\expr.py:267: note:     def [SeriesByT: str | bytes | date | bool | int | <6 more items>] groupby(self, by: Series[SeriesByT], axis: Literal['index', 0, _NoDefault.no_default] = ..., level: Hashable | Sequence[Hashable] | None = ..., as_index: Literal[True] = ..., sort: bool = ..., group_keys: bool = ..., observed: bool | Literal[_NoDefault.no_default] = ..., dropna: bool = ...) -> DataFrameGroupBy[SeriesByT, Literal[True]]
narwhals\_pandas_like\expr.py:267: note:     def [SeriesByT: str | bytes | date | bool | int | <6 more items>] groupby(self, by: Series[SeriesByT], axis: Literal['index', 0, _NoDefault.no_default] = ..., level: Hashable | Sequence[Hashable] | None = ..., as_index: Literal[False] = ..., sort: bool = ..., group_keys: bool = ..., observed: bool | Literal[_NoDefault.no_default] = ..., dropna: bool = ...) -> DataFrameGroupBy[SeriesByT, Literal[False]]
narwhals\_pandas_like\expr.py:267: note:     def groupby(self, by: CategoricalIndex[Any] | Index[Any] | Series[Any], axis: Literal['index', 0] | Literal[_NoDefault.no_default] = ..., level: Hashable | Sequence[Hashable] | None = ..., as_index: Literal[True] = ..., sort: bool = ..., group_keys: bool = ..., observed: bool | Literal[_NoDefault.no_default] = ..., dropna: bool = ...) -> DataFrameGroupBy[Any, Literal[True]]
narwhals\_pandas_like\expr.py:267: note:     def groupby(self, by: CategoricalIndex[Any] | Index[Any] | Series[Any], axis: Literal['index', 0] | Literal[_NoDefault.no_default] = ..., level: Hashable | Sequence[Hashable] | None = ..., as_index: Literal[False] = ..., sort: bool = ..., group_keys: bool = ..., observed: bool | Literal[_NoDefault.no_default] = ..., dropna: bool = ...) -> DataFrameGroupBy[Any, Literal[False]]
narwhals\_pandas_like\expr.py:273: error: No overload variant of "groupby" of "DataFrame" matches argument type "Sequence[str]"  [call-overload]
                        res_native = df._native_frame.groupby(partition_by)[  # pyright: ignore[reportArgumentType, reportCallIssue]
                                     ^~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
narwhals\_pandas_like\expr.py:273: note: Possible overload variants:
narwhals\_pandas_like\expr.py:273: note:     def groupby(self, by: str | bytes | date | datetime | timedelta | <7 more items> | complex, axis: Literal['index', 0] | Literal[_NoDefault.no_default] = ..., level: Hashable | Sequence[Hashable] 
| None = ..., as_index: Literal[True] = ..., sort: bool = ..., group_keys: bool = ..., observed: bool | Literal[_NoDefault.no_default] = ..., dropna: bool = ...) -> DataFrameGroupBy[str | bytes | date | datetime | timedelta | <7 more items> | complex, Literal[True]]
narwhals\_pandas_like\expr.py:273: note:     def groupby(self, by: str | bytes | date | datetime | timedelta | <7 more items> | complex, axis: Literal['index', 0] | Literal[_NoDefault.no_default] = ..., level: Hashable | Sequence[Hashable] 
| None = ..., as_index: Literal[False] = ..., sort: bool = ..., group_keys: bool = ..., observed: bool | Literal[_NoDefault.no_default] = ..., dropna: bool = ...) -> DataFrameGroupBy[str | bytes | date | datetime | timedelta | <7 more items> | complex, Literal[False]]
narwhals\_pandas_like\expr.py:273: note:     def groupby(self, by: DatetimeIndex, axis: Literal['index', 0] | Literal[_NoDefault.no_default] = ..., level: Hashable | Sequence[Hashable] | None = ..., as_index: Literal[True] = ..., sort: bool = ..., group_keys: bool = ..., observed: bool | Literal[_NoDefault.no_default] = ..., dropna: bool = ...) -> DataFrameGroupBy[Timestamp, Literal[True]]
narwhals\_pandas_like\expr.py:273: note:     def groupby(self, by: DatetimeIndex, axis: Literal['index', 0] | Literal[_NoDefault.no_default] = ..., level: Hashable | Sequence[Hashable] | None = ..., as_index: Literal[False] = ..., sort: bool = ..., group_keys: bool = ..., observed: bool | Literal[_NoDefault.no_default] = ..., dropna: bool = ...) -> DataFrameGroupBy[Timestamp, Literal[False]]
narwhals\_pandas_like\expr.py:273: note:     def groupby(self, by: TimedeltaIndex, axis: Literal['index', 0] | Literal[_NoDefault.no_default] = ..., level: Hashable | Sequence[Hashable] | None = ..., as_index: Literal[True] = ..., sort: bool = ..., group_keys: bool = ..., observed: bool | Literal[_NoDefault.no_default] = ..., dropna: bool = ...) -> DataFrameGroupBy[Timedelta, Literal[True]]
narwhals\_pandas_like\expr.py:273: note:     def groupby(self, by: TimedeltaIndex, axis: Literal['index', 0] | Literal[_NoDefault.no_default] = ..., level: Hashable | Sequence[Hashable] | None = ..., as_index: Literal[False] = ..., sort: bool = ..., group_keys: bool = ..., observed: bool | Literal[_NoDefault.no_default] = ..., dropna: bool = ...) -> DataFrameGroupBy[Timedelta, Literal[False]]
narwhals\_pandas_like\expr.py:273: note:     def groupby(self, by: PeriodIndex, axis: Literal['index', 0] | Literal[_NoDefault.no_default] = ..., level: Hashable | Sequence[Hashable] | None = ..., as_index: Literal[True] = ..., sort: bool = ..., group_keys: bool = ..., observed: bool | Literal[_NoDefault.no_default] = ..., dropna: bool = ...) -> DataFrameGroupBy[Period, Literal[True]]
narwhals\_pandas_like\expr.py:273: note:     def groupby(self, by: PeriodIndex, axis: Literal['index', 0] | Literal[_NoDefault.no_default] = ..., level: Hashable | Sequence[Hashable] | None = ..., as_index: Literal[False] = ..., sort: bool 
= ..., group_keys: bool = ..., observed: bool | Literal[_NoDefault.no_default] = ..., dropna: bool = ...) -> DataFrameGroupBy[Period, Literal[False]]
narwhals\_pandas_like\expr.py:273: note:     def [IntervalT: Interval[Any]] groupby(self, by: IntervalIndex[IntervalT], axis: Literal['index', 0, _NoDefault.no_default] = ..., level: Hashable | Sequence[Hashable] | None = ..., as_index: Literal[True] = ..., sort: bool = ..., group_keys: bool = ..., observed: bool | Literal[_NoDefault.no_default] = ..., dropna: bool = ...) -> DataFrameGroupBy[IntervalT, Literal[True]]
narwhals\_pandas_like\expr.py:273: note:     def [IntervalT: Interval[Any]] groupby(self, by: IntervalIndex[IntervalT], axis: Literal['index', 0, _NoDefault.no_default] = ..., level: Hashable | Sequence[Hashable] | None = ..., as_index: Literal[False] = ..., sort: bool = ..., group_keys: bool = ..., observed: bool | Literal[_NoDefault.no_default] = ..., dropna: bool = ...) -> DataFrameGroupBy[IntervalT, Literal[False]]
narwhals\_pandas_like\expr.py:273: note:     def groupby(self, by: MultiIndex | tuple[Any, ...] | list[Any] | ufunc | Callable[..., Any] | list[ufunc | Callable[..., Any]] | list[Series[Any]] | <7 more items> | None = ..., axis: Literal['index', 0] | Literal[_NoDefault.no_default] = ..., level: Hashable | Sequence[Hashable] | None = ..., as_index: Literal[True] = ..., sort: bool = ..., group_keys: bool = ..., observed: bool | Literal[_NoDefault.no_default] = ..., dropna: bool = ...) -> DataFrameGroupBy[tuple[Any, ...], Literal[True]]
narwhals\_pandas_like\expr.py:273: note:     def groupby(self, by: MultiIndex | tuple[Any, ...] | list[Any] | ufunc | Callable[..., Any] | list[ufunc | Callable[..., Any]] | list[Series[Any]] | <7 more items> | None = ..., axis: Literal['index', 0] | Literal[_NoDefault.no_default] = ..., level: Hashable | Sequence[Hashable] | None = ..., as_index: Literal[False] = ..., sort: bool = ..., group_keys: bool = ..., observed: bool | Literal[_NoDefault.no_default] = ..., dropna: bool = ...) -> DataFrameGroupBy[tuple[Any, ...], Literal[False]]
narwhals\_pandas_like\expr.py:273: note:     def [SeriesByT: str | bytes | date | bool | int | <6 more items>] groupby(self, by: Series[SeriesByT], axis: Literal['index', 0, _NoDefault.no_default] = ..., level: Hashable | Sequence[Hashable] | None = ..., as_index: Literal[True] = ..., sort: bool = ..., group_keys: bool = ..., observed: bool | Literal[_NoDefault.no_default] = ..., dropna: bool = ...) -> DataFrameGroupBy[SeriesByT, Literal[True]]
narwhals\_pandas_like\expr.py:273: note:     def [SeriesByT: str | bytes | date | bool | int | <6 more items>] groupby(self, by: Series[SeriesByT], axis: Literal['index', 0, _NoDefault.no_default] = ..., level: Hashable | Sequence[Hashable] | None = ..., as_index: Literal[False] = ..., sort: bool = ..., group_keys: bool = ..., observed: bool | Literal[_NoDefault.no_default] = ..., dropna: bool = ...) -> DataFrameGroupBy[SeriesByT, Literal[False]]
narwhals\_pandas_like\expr.py:273: note:     def groupby(self, by: CategoricalIndex[Any] | Index[Any] | Series[Any], axis: Literal['index', 0] | Literal[_NoDefault.no_default] = ..., level: Hashable | Sequence[Hashable] | None = ..., as_index: Literal[True] = ..., sort: bool = ..., group_keys: bool = ..., observed: bool | Literal[_NoDefault.no_default] = ..., dropna: bool = ...) -> DataFrameGroupBy[Any, Literal[True]]
narwhals\_pandas_like\expr.py:273: note:     def groupby(self, by: CategoricalIndex[Any] | Index[Any] | Series[Any], axis: Literal['index', 0] | Literal[_NoDefault.no_default] = ..., level: Hashable | Sequence[Hashable] | None = ..., as_index: Literal[False] = ..., sort: bool = ..., group_keys: bool = ..., observed: bool | Literal[_NoDefault.no_default] = ..., dropna: bool = ...) -> DataFrameGroupBy[Any, Literal[False]]
Found 27 errors in 4 files (checked 355 source files)

MarcoGorelli · 2025-03-19T16:50:30Z

does this make any user-facing difference? if not, then internally i think i'd prefer to just use pandas, as modin doesn't export types, neither does cudf

dangotbanned · 2025-03-19T17:09:28Z

does this make any user-facing difference? if not, then internally i think i'd prefer to just use pandas

Right now - it doesn't even make an internal difference, since we don't type PandasLikeDataFrame.native.

The general idea is this:

SeriesT = TypeVar(
    "SeriesT", pd.Series[Any], mpd.Series, cudf.Series[Any], default=pd.Series[Any]
)

Means exactly one of those types, defaulting to pd.Series[Any] instead of Unknown.

I think that theoretically describes what you'd want - and accounts for missing imports for non-pandas.

However ...

as modin doesn't export types, neither does cudf

@MarcoGorelli well that certainly throws a spanner into this whole idea 😄

Note

This is low-priority, mainly just wanting to be able to get some IDE feedback to understand the current code better

dangotbanned · 2025-04-10T19:29:29Z

Closing in favor of the simpler (pandas-only) version in #2368

- Somewhat of a resurrection of #2227 - But this time building on #2693

* chore(typing): Kinda type `pandas_like.utils.`select_columns_by_name` - Somewhat of a resurrection of #2227 - But this time building on #2693 * chore(typing): "Fix" `pandas_like.utils.(set_index|rename)` * chore(typing): `NativeSeriesT` in `calculate_timestamp_date` Confused why this (seemingly unrelated) change triggered the `[has-type]` in `concat_str` * chore: `NativeSeriesT` in `calculate_timestamp_datetime` * fix(typing): Flip operands, use `ndarray.all` Resolves #2714 (comment) Did this before in https://github.com/narwhals-dev/narwhals/blob/0f37267b3d05b8f5d7a37ef8f0b43647f4afec48/narwhals/_pandas_like/utils.py#L767

dangotbanned added 3 commits March 16, 2025 20:34

chore(typing): Add constrained TypeVar(s) for pandas

b114d47

fix(typing): Resolve mypy errors

3dc7c03

Still have 7 for `pyright`

Merge remote-tracking branch 'upstream/main' into pandas-like-type-var

55dcfb7

dangotbanned added fix typing pandas-like Issue is related to pandas-like backends labels Mar 16, 2025

dangotbanned added 2 commits March 16, 2025 22:13

fix(typing): Resolve pyright errors`

dc62f1d

ignore mypy getting this wrong

3648814

The stubs don't return `Self` for all types, so `pyright` is right

dangotbanned marked this pull request as ready for review March 16, 2025 22:44

Merge branch 'main' into pandas-like-type-var

7403e7d

dangotbanned added 4 commits March 17, 2025 10:32

Merge branch 'main' into pandas-like-type-var

c70a041

Merge branch 'main' into pandas-like-type-var

b3e7b86

Merge branch 'main' into pandas-like-type-var

7cca156

Merge branch 'main' into pandas-like-type-var

0f37267

dangotbanned added the low priority label Mar 19, 2025

dangotbanned closed this Apr 10, 2025

dangotbanned added a commit that referenced this pull request Jun 21, 2025

chore(typing): Kinda type pandas_like.utils.select_columns_by_name`

c743de3

- Somewhat of a resurrection of #2227 - But this time building on #2693

dangotbanned mentioned this pull request Jun 21, 2025

chore(typing): Improving _pandas_like.utils #2714

Merged

10 tasks

dangotbanned deleted the pandas-like-type-var branch September 2, 2025 20:27

Provide feedback

Saved searches

Use saved searches to filter your results more quickly

Uh oh!

chore(typing): Constrained `TypeVar`(s) for `pandas` #2227

chore(typing): Constrained `TypeVar`(s) for `pandas` #2227

Uh oh!

dangotbanned commented Mar 16, 2025 •

edited

Loading

Uh oh!

dangotbanned commented Mar 16, 2025

Uh oh!

MarcoGorelli commented Mar 19, 2025

Uh oh!

dangotbanned commented Mar 19, 2025

Uh oh!

dangotbanned commented Apr 10, 2025

Uh oh!

Reviewers

Assignees

Labels

Projects

Milestone

Development

Uh oh!

3 participants

chore(typing): Constrained TypeVar(s) for pandas #2227

chore(typing): Constrained TypeVar(s) for pandas #2227

Uh oh!

Conversation

dangotbanned commented Mar 16, 2025 • edited Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

What type of PR is this? (check all applicable)

Related issues

Checklist

If you have comments or can explain your changes, please do so below

Uh oh!

dangotbanned commented Mar 16, 2025

Uh oh!

MarcoGorelli commented Mar 19, 2025

Uh oh!

dangotbanned commented Mar 19, 2025

Uh oh!

dangotbanned commented Apr 10, 2025

Uh oh!

Reviewers

Assignees

Labels

Projects

Milestone

Development

Uh oh!

3 participants

chore(typing): Constrained `TypeVar`(s) for `pandas` #2227

chore(typing): Constrained `TypeVar`(s) for `pandas` #2227

dangotbanned commented Mar 16, 2025 •

edited

Loading