I have a polars dataframe with two columns, one contains lists of string and the other one string. I want to apply the following expression to both columns. However, for some reason isinstance(x, list) doesn’t work properly.

def process_column(column_name: str, alias_name: str) -> pl.Expr:

    return (
        pl.col(column_name).map_elements(lambda x: " ".join(x) if isinstance(x, list) else x)
        .str.to_lowercase()
        .str.split(by="-")
        .list.join(" ")
        #.str.contains("hello")
        .alias(alias_name)
        )

here is a sample dataframe

df = pl.DataFrame({
    "lists": [["hello", "World"], ["polars", "IS", "fast"]],
    "strings": ["foo-hello", "bOO"]
})

This line works just fine:

df2 = df.with_columns(process_column("strings", "processed_string"))

But this one throws the following error.

df2 = df.with_columns(process_column("lists", "processed_lists"))

The error:

polars.exceptions.SchemaError: invalid series dtype: expected `String`, got `list[str]`

I tried the map_elements with return_dtype=pl.String. It doesn’t return an error but the output is wrong.