def create_function(self, func, name=None, database=None):
"""
Creates a function within Impala
Parameters
----------
func : ImpalaUDF or ImpalaUDA
Created with wrap_udf or wrap_uda
name : string (optional)
database : string (optional)
"""
if name is None:
name = func.name
database = database or self.current_database
if isinstance(func, udf.ImpalaUDF):
stmt = ddl.CreateFunction(func.lib_path, func.so_symbol,
func.input_type, func.output, name,
database)
elif isinstance(func, udf.ImpalaUDA):
stmt = ddl.CreateAggregateFunction(func.lib_path, func.input_type,
func.output, func.update_fn,
func.init_fn, func.merge_fn,
func.serialize_fn,
func.finalize_fn, name,
database)
else:
raise TypeError(func)
self._execute(stmt)
def test_create_udf(self):
stmt = ddl.CreateFunction('/foo/bar.so', 'testFunc', self.inputs,
self.output, self.name)
result = stmt.compile()
expected = ("CREATE FUNCTION test_name(string, string) returns bigint "
"location '/foo/bar.so' symbol='testFunc'")
assert result == expected
def test_create_udf_type_conversions(self):
stmt = ddl.CreateFunction('/foo/bar.so', 'testFunc',
['string', 'int8', 'int16', 'int32'],
self.output, self.name)
result = stmt.compile()
expected = ("CREATE FUNCTION `test_name`(string, tinyint, "
"smallint, int) returns bigint "
"location '/foo/bar.so' symbol='testFunc'")
assert result == expected
def create_udf(self, udf_info, name=None, database=None):
"""
Creates a function within Impala
Parameters
----------
udf_info : UDFCreator object
name : string (optional)
database : string (optional)
"""
if name is None:
name = udf_info.get_name()
statement = ddl.CreateFunction(udf_info.hdfs_file, udf_info.so_symbol,
udf_info.inputs, udf_info.output, name,
database)
self._execute(statement)
