daft.functions.split_part#

def split_part(
    expr: Expression,
    delim: str | Expression,
    part: int | Expression,
) -> Expression:
    """Splits the string on occurrences of the delimiter and returns the requested part (1-based).

    If part is negative, the parts are counted backward from the end of the string.
    If part is out of range, an empty string is returned.
    If part is 0, an error is raised.
    If the delimiter is an empty string, the string is not split.
    This is compatible with Spark's split_part function.

    Args:
        expr: The string expression to split
        delim: The delimiter string to split on (not a regular expression)
        part: The 1-based index of the part to return

    Returns:
        Expression: a String expression with the requested part

    Examples:
        >>> import daft
        >>> from daft.functions import split_part
        >>> df = daft.from_pydict({"x": ["a,b,c", "x,y,z"]})
        >>> df.select(split_part(df["x"], ",", 2)).show()
        ╭────────╮
        │ x      │
        │ ---    │
        │ String │
        ╞════════╡
        │ b      │
        ├╌╌╌╌╌╌╌╌┤
        │ y      │
        ╰────────╯
        <BLANKLINE>
        (Showing first 2 of 2 rows)

        Negative parts are counted from the end of the string:

        >>> df.select(split_part(df["x"], ",", -1)).show()
        ╭────────╮
        │ x      │
        │ ---    │
        │ String │
        ╞════════╡
        │ c      │
        ├╌╌╌╌╌╌╌╌┤
        │ z      │
        ╰────────╯
        <BLANKLINE>
        (Showing first 2 of 2 rows)

    """
    return Expression._call_builtin_scalar_fn("split_part", expr, delim, part)