Determine if a group contains a value in Polars

I would like return rows of a dataset where groups satisfy two conditions. That is,

1. I want to know if the number of unique values in a group is greater than 1, and
2. I want to know if the group contains a particular value.

I am getting hung up on the latter!

Suppose I have this dataframe:

import polars as pl

df = pl.from_repr("""
┌──────┬──────────┬─────────┐
│ item ┆ location ┆ store   │
│ ---  ┆ ---      ┆ ---     │
│ i64  ┆ str      ┆ str     │
╞══════╪══════════╪═════════╡
│ 0    ┆ new york ┆ store 1 │
│ 0    ┆ boston   ┆ store 1 │
│ 1    ┆ boston   ┆ store 1 │
│ 1    ┆ boston   ┆ store 2 │
│ 0    ┆ ohio     ┆ store 1 │
│ 0    ┆ ohio     ┆ store 3 │
└──────┴──────────┴─────────┘
""")

I can check for #1 easily using filter, n_unique(), and over as follows:

df.filter(
    (pl.col("store").n_unique() > 1).over("item", "location")
)

┌──────┬──────────┬─────────┐
│ item ┆ location ┆ store   │
│ ---  ┆ ---      ┆ ---     │
│ i64  ┆ str      ┆ str     │
╞══════╪══════════╪═════════╡
│ 1    ┆ boston   ┆ store 1 │
│ 1    ┆ boston   ┆ store 2 │
│ 0    ┆ ohio     ┆ store 1 │
│ 0    ┆ ohio     ┆ store 3 │
└──────┴──────────┴─────────┘

However, I would like to add to the filter to return rows where the corresponding group contains "store 3".

I tried this (and variations of it):

df.filter(
    (pl.col("store").n_unique() > 1).over("item", "location")
    & (pl.col("store").is_in(["store 3"]))
)

┌──────┬──────────┬─────────┐
│ item ┆ location ┆ store   │
│ ---  ┆ ---      ┆ ---     │
│ i64  ┆ str      ┆ str     │
╞══════╪══════════╪═════════╡
│ 0    ┆ ohio     ┆ store 3 │
└──────┴──────────┴─────────┘

But this is only returning a single row. And actually the result makes some sense to me but I was hoping to return the whole group. For example,

┌──────┬──────────┬─────────┐
│ item ┆ location ┆ store   │
│ ---  ┆ ---      ┆ ---     │
│ i64  ┆ str      ┆ str     │
╞══════╪══════════╪═════════╡
│ 0    ┆ ohio     ┆ store 1 │
│ 0    ┆ ohio     ┆ store 3 │
└──────┴──────────┴─────────┘

I'm quite sure I'm missing something obvious. Any help is appreciated. Loving Polars so far without a doubt!

(pl.col(...) == value).any().over(group) can be used to test if a value exists within a group.

You can "and" the multiple conditions and run them through .over()

(one & two).over(group)

df.filter(
   (
      (pl.col("store").n_unique() > 1) &
      (pl.col('store') == 'store 3').any()
   )
   .over('item', 'location')
)

Using pl.all_horizontal() is also another option.

df.filter(
   pl.all_horizontal(
      pl.col("store").n_unique() > 1,
      (pl.col('store') == 'store 3').any()
   )
   .over('item', 'location')
)

shape: (2, 3)
┌──────┬──────────┬─────────┐
│ item ┆ location ┆ store   │
│ ---  ┆ ---      ┆ ---     │
│ i64  ┆ str      ┆ str     │
╞══════╪══════════╪═════════╡
│ 0    ┆ ohio     ┆ store 1 │
│ 0    ┆ ohio     ┆ store 3 │
└──────┴──────────┴─────────┘