`polars.Expr.take` returns `null` if `ComputeError: index out of bounds`

[stevenlis]

Problem description

Sometimes you may have groups with different numbers of rows. It would be nice if polars.Expr.take could just return a null when index out of bounds.

df = pl.DataFrame(
{
        "group": [
            "one",
            "one",
            "one",
            "two",
            "two",
        ],
        "value": [1, 98, 2, 3, 99],
    }
)
df.groupby("group", maintain_order=True).agg(pl.col("value").take(2))

shape: (2, 2)
┌───────┬───────┐
│ group ┆ value │
│ ---   ┆ ---   │
│ str   ┆ i64   │
╞═══════╪═══════╡
│ one   ┆ 2     │
│ two   ┆ null  │
└───────┴───────┘

For now:

df.groupby("group", maintain_order=True).agg(pl.col("value").shift(-2).first())

[cmdlineluser]

.arr.get returns null instead of raising, which may be another option:

>>> df.groupby("group", maintain_order=True).agg(pl.col("value").list().arr.get(2))
shape: (2, 2)
┌───────┬───────┐
│ group ┆ value │
│ ---   ┆ ---   │
│ str   ┆ i64   │
╞═══════╪═══════╡
│ one   ┆ 2     │
│ two   ┆ null  │
└───────┴───────┘

[uditrana]

I would love for take to have this as a flag though... I think its a pretty reasonable and common use case! take syntactically is more canonical for operating on columns instead of treating it as an array imo.

Also seems that list.take has this flag but not take? https://pola-rs.github.io/polars/py-polars/html/reference/expressions/api/polars.Expr.list.take.html