Skip to content
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.

Sign up for GitHub

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

feat: add Series.__contains__ #1480

Merged
merged 10 commits into from
Dec 3, 2024
Merged

feat: add Series.__contains__ #1480

Show file tree
Hide file tree
Changes from 4 commits
Commits
Show all changes
10 commits
Select commit Hold shift + click to select a range
e36d4ed
feat: add Series.__contains__
FBruzzesi Dec 1, 2024
e6d9419
use pa.scalar with explicit type
FBruzzesi Dec 1, 2024
9a90632
use pa.scalar with explicit type
FBruzzesi Dec 1, 2024
3aa5b75
use pa.scalar with explicit type
FBruzzesi Dec 1, 2024
c653e48
Merge branch 'main' into feat/series-contains
FBruzzesi Dec 1, 2024
3f66045
I got fooled
FBruzzesi Dec 1, 2024
c9f0b36
Merge branch 'main' into feat/series-contains
FBruzzesi Dec 3, 2024
b2b005a
use pandas-like equals, raise custom error, deal with arrow scalar
FBruzzesi Dec 3, 2024
38c5833
arrow None
FBruzzesi Dec 3, 2024
634c262
add ArrowInvalid
FBruzzesi Dec 3, 2024
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
10 changes: 10 additions & 0 deletions narwhals/_arrow/series.py
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -1017,6 +1017,16 @@ def __iter__(self: Self) -> Iterator[Any]:
for x in self._native_series.__iter__()
)

def __contains__(self: Self, other: Any) -> bool:
import pyarrow as pa # ignore-banned-imports
import pyarrow.compute as pc # ignore-banned-imports

native_series = self._native_series
return maybe_extract_py_scalar( # type: ignore[no-any-return]
pc.is_in(pa.scalar(other, type=native_series.type), native_series),
Copy link
Member

Choose a reason for hiding this comment

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

if other is of a type not convertible to native_series.type, will this raise? I'm wondering if we should return False in such a case?

Copy link
Member

Choose a reason for hiding this comment

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

looks like polars would also raise here:

In [2]: 'foo' in pl.Series([1,2,3])
---------------------------------------------------------------------------
InvalidOperationError                     Traceback (most recent call last)
Cell In[2], line 1
----> 1 'foo' in pl.Series([1,2,3])

File ~/scratch/.venv/lib/python3.12/site-packages/polars/series/series.py:1233, in Series.__contains__(self, item)
   1231 if item is None:
   1232     return self.has_nulls()
-> 1233 return self.implode().list.contains(item).item()

File ~/scratch/.venv/lib/python3.12/site-packages/polars/series/utils.py:106, in call_expr.<locals>.wrapper(self, *args, **kwargs)
    104     expr = getattr(expr, namespace)
    105 f = getattr(expr, func.__name__)
--> 106 return s.to_frame().select_seq(f(*args, **kwargs)).to_series()

File ~/scratch/.venv/lib/python3.12/site-packages/polars/dataframe/frame.py:9138, in DataFrame.select_seq(self, *exprs, **named_exprs)
   9115 def select_seq(
   9116     self, *exprs: IntoExpr | Iterable[IntoExpr], **named_exprs: IntoExpr
   9117 ) -> DataFrame:
   9118     """
   9119     Select columns from this DataFrame.
   9120
   (...)
   9136     select
   9137     """
-> 9138     return self.lazy().select_seq(*exprs, **named_exprs).collect(_eager=True)

File ~/scratch/.venv/lib/python3.12/site-packages/polars/lazyframe/frame.py:2029, in LazyFrame.collect(self, type_coercion, predicate_pushdown, projection_pushdown, simplify_expression, slice_pushdown, comm_subplan_elim, comm_subexpr_elim, cluster_with_columns, collapse_joins, no_optimization, streaming, engine, background, _eager, **_kwargs)
   2027 # Only for testing purposes
   2028 callback = _kwargs.get("post_opt_callback", callback)
-> 2029 return wrap_df(ldf.collect(callback))

InvalidOperationError: is_in operation not supported for dtypes `str` and `list[i64]`

probably ok to just follow them here then

Copy link
Member

Choose a reason for hiding this comment

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

still, probably good to have a test to check it raises? we can leave the expression unification for a separate topic

Copy link
Member Author

@FBruzzesi FBruzzesi Dec 3, 2024 •
edited
Loading

Choose a reason for hiding this comment

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

Will add a test and make sure that same exception is raised 👌

Copy link
Member Author

@FBruzzesi FBruzzesi Dec 3, 2024 •
edited
Loading

Choose a reason for hiding this comment

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

Ok this is definitly wrong behaviour for arrow as the casting would convert from float to int, leading to the following:

import pyarrow as pa
import narwhals as nw

data = [100, 200, None]
s = nw.from_native(pa.table({"a": data}), eager_only=True)["a"]

100.3 in s
True

On the Exception side, I would not be sure how to catch the error for pandas as:

(pd.Series([1,2,3]) == "foo").any()
np.False_

In a way, I find polars behaviour more surprising

Copy link
Member

@MarcoGorelli MarcoGorelli Dec 3, 2024 •
edited
Loading

Choose a reason for hiding this comment

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

tbh we could just dispatch to (s._compliant_series == other).any() for all backends 😄 that would still be better than the status quo

return_py_scalar=True,
)

@property
def shape(self: Self) -> tuple[int]:
return (len(self._native_series),)
Expand Down
5 changes: 5 additions & 0 deletions narwhals/_pandas_like/series.py
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -851,6 +851,11 @@ def rolling_mean(
def __iter__(self: Self) -> Iterator[Any]:
yield from self._native_series.__iter__()

def __contains__(self: Self, other: Any) -> bool:
if other is None:
return self._native_series.isna().any() # type: ignore[no-any-return]
FBruzzesi marked this conversation as resolved.
Show resolved Hide resolved
return other in self._native_series
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 fell for the example contained in our why section!

There are two ways of doing this:

  • return other in self.to_numpy(): yet I don't think this is a good idea for cudf
  • return self._native_series.isin({other}).any()

Copy link
Contributor

@AlessandroMiola AlessandroMiola Dec 1, 2024 •
edited
Loading

Choose a reason for hiding this comment

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

May I ask a general, possibly silly, question, just for the sake of better understanding? 😇

I see the point in implementing __contains__ for pandas-like and arrow and also what you mention here. Instead, I haven't properly got the details behind #1443 (comment), as I don't see func(1, pa.chunked_array([data])) returning False.

Should we implement __contains__ at narwhals/series.py to ensure the above behaviour is explicitly triggered via __contains__ (provided that what I'm stating above is correct)?

Copy link
Member Author

Choose a reason for hiding this comment

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

as I don't see func(1, pa.chunked_array([data])) returning False.

Damn that's true... but now I don't know why, and I wonder what is being called. The following returns false:

1 in pa.chunked_array([[1,2,3]])
False

Should we implement contains at narwhals/series.py to ensure the above behaviour is explicitly triggered via contains?

I am addressing it in the PR

Copy link
Contributor

Choose a reason for hiding this comment

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

Damn that's true... but now I don't know why, and I wonder what is being called.

Me neither, had issues in trying to figure it out

I am addressing it in the PR

yeah, I see it; was just asking to confirm my understanding as I couldn't indeed see where this was coming from, without the explicit reference to __contains__ 😃

Copy link
Member Author

@FBruzzesi FBruzzesi Dec 1, 2024 •
edited
Loading

Choose a reason for hiding this comment

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

Ok, since we implement __iter__ the check works. Example:

class A:
    pass

class B:
    def __iter__(self):
        yield from [1,2,3]

0 in B(), 1 in B()
> (False, True)

1 in A()
> TypeError: argument of type 'A' is not iterable

When I wrote the comment, the result was false, and now it returns true due to the changes in #1471, lines. In fact before such change, iteration was returning arrow scalars

Copy link
Contributor

Choose a reason for hiding this comment

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

nice, thanks for the clarification!:)


def is_finite(self: Self) -> Self:
s = self._native_series
return self._from_native_series((s > float("-inf")) & (s < float("inf")))
Expand Down
3 changes: 3 additions & 0 deletions narwhals/series.py
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -3324,6 +3324,9 @@ def rolling_mean(
def __iter__(self: Self) -> Iterator[Any]:
yield from self._compliant_series.__iter__()

def __contains__(self: Self, other: Any) -> bool:
return other in self._compliant_series

@property
def str(self: Self) -> SeriesStringNamespace[Self]:
return SeriesStringNamespace(self)
Expand Down
23 changes: 23 additions & 0 deletions tests/series_only/__contains___test.py
View file
Open in desktop
Original file line number Diff line number Diff line change
@@ -0,0 +1,23 @@
from __future__ import annotations

from typing import TYPE_CHECKING

import pytest

import narwhals.stable.v1 as nw

if TYPE_CHECKING:
from tests.utils import ConstructorEager

data = [1, 2, None]


@pytest.mark.parametrize(("other", "expected"), [(1, True), (None, True), (3, False)])
def test_contains(
constructor_eager: ConstructorEager,
other: int | None,
expected: bool, # noqa: FBT001
) -> None:
s = nw.from_native(constructor_eager({"a": data}), eager_only=True)["a"]

assert (other in s) == expected
1 change: 1 addition & 0 deletions utils/check_api_reference.py
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -32,6 +32,7 @@
"value_counts",
"zip_with",
"__iter__",
"__contains__",
}
BASE_DTYPES = {
"NumericType",
Expand Down
Loading