tests: Add test for Expr.is_in() with Series input

[CangyuanLi]

Describe the bug

Hi, thanks for the awesome package! I noticed a bug with .is_in() when the Narwhals DataFrame is Polars-backed. When .si_in() is passed a Narwhals Polars-backed Series, it errors with

ValueError: can only call '.item()' if the Series is of length 1, or an explicit index is provided

You can see the behavior in the code below, which shows that it works when passed a list or a Polars Series. The code also shows the same bug does not happen with a Narwhals Pandas-backed Series.

Steps or code to reproduce the bug

import narwhals as nw
import narwhals.typing as nwt
import pandas as pd
import polars as pl

# Pandas

df = pd.DataFrame({"a": [1, 2, 3]})
df2 = pd.DataFrame({"b": [1, 0, 2]})
nw_df = nw.from_native(df)
nw_df2 = nw.from_native(df2)

print(nw_df.filter(nw.col("a").is_in(df2["b"])).to_native())
print(nw_df.filter(nw.col("a").is_in(nw_df2["b"])).to_native())

# Polars

df = pl.DataFrame({"a": [1, 2, 3]})
df2 = pl.DataFrame({"b": [1, 0, 2]})
nw_df = nw.from_native(df)
nw_df2 = nw.from_native(df2)

print(nw_df.filter(nw.col("a").is_in(df2["b"])).to_native())
print(nw_df.filter(nw.col("a").is_in(nw_df2["b"].to_list())).to_native())
print(nw_df.filter(nw.col("a").is_in(nw_df2["b"])).to_native())  # doesn't work!

Expected results

I expect the Pandas and Polars-backed frames to be the same.

Actual results

ValueError: can only call '.item()' if the Series is of length 1, or an explicit index is provided (Series is of length 3)

Please run narwhals.show_version() and enter the output below.

System:
    python: 3.10.6 (main, Oct 24 2022, 16:07:47) [GCC 11.2.0]
executable: /opt/conda/envs/Python-3.10-Premium/bin/python
   machine: Linux-4.18.0-372.98.1.el8_6.x86_64-x86_64-with-glibc2.28

Python dependencies:
     narwhals: 1.15.1
       pandas: 2.2.3
       polars: 1.16.0
         cudf: 
        modin: 
      pyarrow: 18.1.0
        numpy: 1.26.4

Relevant log output

No response

[MarcoGorelli]

thanks @CangyuanLi for your kind works, and for your report

I just tried this on the main branch, and it looks like it works - just gonna see when it might have got fixed

[MarcoGorelli]

In fact, it also works with 1.15.2 - could you try upgrading your Narwhals version please? pip install -U narwhals

[CangyuanLi]

Oh it does, thanks so much!

[MarcoGorelli]

We don't have an explicit test case for this though, it looks like it got fixed as a positive side effect of #1480, good one @FBruzzesi !

Looks like it's because Polars explicitly check for Collection:

https://github.com/pola-rs/polars/blob/3985573cd1b2aedce18e243140e6b0b57696e9ba/py-polars/polars/expr/expr.py#L5778-L5784

Reopening then as a "needs test" issue

The task is, in is_in_test.py, to add a test which tests df.filter(nw.col("a").is_in(df_other["b"])). It should use the constructor_eager fixture. Please see the contributing guide for how to get started https://github.com/narwhals-dev/narwhals/blob/main/CONTRIBUTING.md

[FBruzzesi]

Cool to see a nice side effect 😇
Long term plan would be to add support for expr's as well, similar to what's mentioned in #730 💪🏼

[jddelaune]

I'll give this a shot!

[skritsotalakis]

Hi!
Is this still available?

[MarcoGorelli]

yup!