swordfish_udf#

swordfish.function.swordfish_udf(func=None, *, is_aggregation=False, is_state=False, mode='default')#

Registers a user-defined function (UDF) in Swordfish.

This enables Python functions to be used within Swordfish SQL queries and higher-order functions. The function can be registered in different modes.

Parameters:

  • func (Optional[Callable], optional) – The UDF to be registered. Defaults to None.

  • is_aggregation (bool, optional) – Specifies whether the UDF performs aggregation operations. Defaults to False.

  • is_state (bool, optional) – Specifies whether the UDF maintains state across calls. Defaults to False.

  • mode (Literal["default", "translate"], optional) –

    Determines how the UDF is registered. Defaults to “default”.

    • ”defalut”: The function is saved as a Python UDF object which depends on the Python environment to run.

    • ”translate”: The function is translated into Swordfish’s internal representation. This eliminates Python environment dependency but supports a limited subset of Python syntax.

Returns:

The FunctionDef object representing the registered UDF.

Return type:

FunctionDef

Examples

>>> import swordfish.function as F
>>> @F.swordfish_udf(is_aggregation=True)
>>> def avg_func(a: int, b: int) -> float:
...     return (a + b) / 2

>>> @F.swordfish_udf(is_state=True)
>>> def counter(state: int, increment: int) -> int:
...     return state + increment

>>> @F.swordfish_udf(mode="translate")
>>> def translate_to_uppercase(text: str) -> str:
...     return text.upper()

>>> @F.swordfish_udf()
>>> def add(a: int, b: int) -> int:
...     return a + b
On this page