Scalar Python UDF Registration Parameters

Registering a Scalar Python UDF involves registering an original Python function into the database.

Whether it is explicit or implicit registration, for registering a Python-type Scalar UDF, you can currently pass the following parameters:

**Table 1** Scalar Python UDF registration parameters
Registration Parameter	Description	Type	Default Value
name	Specifies the actual storage name of the UDF in the database.	str \| None	None
database	Specifies the LakeFormation database where the UDF resides.	str \| None	None
fn	Specifies the original Python function of the UDF.	Callable	None
signature (currently unavailable)	Specifies the UDF function signature and return value type.	ibis.common.annotations.Signature \| None	None
replace (currently unavailable)	Specifies whether the UDF supports in-place modification.	bool	False
temporary (currently unavailable)	Specifies whether the UDF has a session-level lifecycle.	bool	False
if_not_exist (currently unavailable)	Specifies whether to skip errors for existing UDFs.	bool	False
strict	Specifies whether the UDF automatically filters NULL values.	bool	True
volatility	Specifies the stability of the UDF.	VolatilityType.VOLATILE \| VolatilityType.STABLE \| VolatilityType.IMMUTABLE	VolatilityType.VOLATILE
runtime_version (currently unavailable)	Specifies the Python version for executing the UDF.	str	sys.version_info
imports	Specifies the external code files on which the UDF depends.	List[str]	None
packages	Specifies the Python modules on which the UDF depends.	List[Union[str, module]]	None
register_type (currently, only RegisterType.OBS is supported)	Specifies the registration form of the UDF.	RegisterType.TEXT \| RegisterType.OBS	RegisterType.OBS
comment	Specifies user comments for the UDF.	str \| None	None

Notes:

For the imports parameter, only file paths in the same directory or subdirectories as the .py file containing the current Python function are allowed.

For the fn parameter, if fn is not in the .py file where the Scalar Python UDF is being registered, then the file path defining fn must also be added to the imports parameter, for example:

from process import outer

con = ibis.fabric.connect(...)

# Register a UDF.
udf = con.udf.python.register(
    outer(), # fn introduced externally
    imports=["process.py"] # Add the file path for fn
)

For the signature parameter, user input is currently not allowed. Only automatic inference of parameter/return value types is supported. See Type Inference of the signature Parameter for details.
For the volatility parameter, the meanings of the three enumeration types are:
- VolatilityType.VOLATILE: Function results may change at any time.
- VolatilityType.STABLE: For fixed inputs, the function's result does not change during a single scan.
- VolatilityType.IMMUTABLE: The function always produces the same result for identical inputs.
This parameter, along with SHIPPABLE, affects the push-down execution of functions. Specifically: Functions of the IMMUTABLE type can always be pushed down to DNs for execution. For STABLE or VOLATILE type functions, they can only be pushed down to DNs for execution if their property is SHIPPABLE.