Boolean "true" and "false" values for type inference or automatic conversion (a la pyarrow)

[thedeadliestcatch]

Boolean "true" and "false" values for type inference or automatic conversion (a la pyarrow)

[dangotbanned]

Feel free to adapt this:

def cast_str_bool(
    *cols: str, replacer: Literal["lower", "upper", "capwords"] = "lower"
) -> Expr:
    """
    Casts `cols` through `Int64` to `Boolean`, via Expr.`replace`.

    Selection is via cs.`matches`, and can be applied transparently.

    Parameters
    ----------
    *cols
        One or more column names, which are of type String
    replacer
        remapping strategy from _ to 1,0.
            `lower`: 'true', 'false'
            `upper`: 'TRUE', 'FALSE'
            `capwords`: 'True', 'False'
    """
    match replacer:
        case "lower":
            mapping = {
                "true": 1,
                "false": 0,
            }
        case "upper":
            mapping = {
                "TRUE": 1,
                "FALSE": 0,
            }
        case "capwords":
            mapping = {
                "True": 1,
                "False": 0,
            }
        case _:
            msg = (
                f"`replacer` must be one of {["lower", "upper", "capwords"]!r}\n"
                f"but got {replacer!r}"
            )
            raise ValueError(msg)
    return (
        cs.matches(f"^{"|".join(cols)}$")
        .replace(mapping)
        .str.to_integer()
        .cast(pl.Boolean)
    )