Explicitly Registering a Scalar UDF

Explicitly register a Scalar UDF. That is, you need to manually add the registration logic code intrusively in the Python code. For detailed usage, refer to Scalar UDF Explicit Registration Syntax.

• For Scalar Python UDFs, registering a Scalar Python UDF involves registering an original Python function or class into the database. The example code for explicit registration is as follows. For parameter descriptions, refer to Scalar Python UDF Registration Parameters.
  • Example of explicitly registering a Python function for a scalar Python UDF

    import ibis
    import ibis_fabric as fabric
    from ibis_fabric.udf import RegisterType
    
    def calculate_product(price: float, quantity: int) -> float:    
        return price * quantity
    
    con = ibis.fabric.connect(...)
    
    # Explicitly register a UDF directly.
    udf = con.udf.python.register(calculate_product, database="your-database", register_type=RegisterType.OBS)
    # Explicitly register a UDF from a file.
    udf = con.udf.python.register_from_file("your-current-file-path", "calculate_product", database="your-database", register_type=RegisterType.OBS)
    
    # Use the UDF.
    t = con.table(name="your-table", schema="your-schema")
    expression = t.select(udf(t.price, t.quantity).name("sum column"))
    
    print(expression.execute())
    

  • Example of explicitly registering a Python class for a scalar Python UDF

    import ibis
    import ibis_fabric as fabric
    from ibis_fabric.udf import RegisterType
    
    class Hello:
        def __init__(self, firstname: str, lastname: str): 
            self.firstname = firstname
            self.lastname = lastname
    
        def process(self, location: str) -> str:
            return f"Hello{self.firstname} {self.lastname}! Welcome to {location}"
    
    con = ibis.fabric.connect(...)
    
    # Explicitly register a UDF directly.
    udf = con.udf.python.register(Hello, database="your-database", register_type=RegisterType.OBS)
    # Explicitly register a UDF from a file.
    udf = con.udf.python.register_from_file("your-current-file-path", "Hello", database="your-database", register_type=RegisterType.OBS)
    
    # Use the UDF.
    t = con.table(name="your-table", schema="your-schema")
    expression = t.select(udf(t.city).with_arguments(firstname="Ethan", lastname="Carter"))
    
    print(expression.execute())

• For Scalar Builtin UDFs, registering a Scalar Builtin UDF serves to obtain the handle of an existing function in the database. There is no actual registration operation. The example code for explicit registration is as follows. For parameter descriptions, refer to Scalar Builtin UDF Registration Parameters.
  • Example of explicitly registering a Python function for a scalar builtin UDF

    import ibis
    import ibis_fabric as fabric
    # The power function already exists in the database.
    def power(a: float, b: float) -> float: 
        ...
    
    con = ibis.fabric.connect(...)
    
    # Explicitly register a UDF directly.
    udf = con.udf.builtin.register(power, database="your-database")
    # Explicitly register a UDF from a file.
    udf = con.udf.builtin.register_from_file("your-current-file-path", "power", database="your-database")
    
    # Use the UDF.
    t = con.table(name="your-table", schema="your-schema")
    expression = t.select(udf(t.interest, t.interval).name("yield column"))
    
    print(expression.execute())

  • Example of explicitly registering a Python class for a scalar builtin UDF

    import ibis
    import ibis_fabric as fabric
    # The Power class already exists in the database.
    class Power:
        def __init__(self, base: float): ...
    
        def process(self, exp: float) -> float: ...
    
    con = ibis.fabric.connect(...)
    
    # Explicitly register a UDF directly.
    udf = con.udf.builtin.register(Power, database="your-database")
    # Explicitly register a UDF from a file.
    udf = con.udf.builtin.register_from_file("your-current-file-path", "Power", database="your-database")
    
    # Use the UDF.
    t = con.table(name="your-table", schema="your-schema")
    expression = t.select(udf(t.interval).with_arguments(base=2048.0).name("yield column"))
    
    print(expression.execute())