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_uda(self):
stmt = ddl.CreateAggregateFunction('/foo/bar.so', self.inputs,
self.output, 'Init', 'Update',
'Merge', 'Finalize', self.name)
result = stmt.compile()
expected = ("CREATE AGGREGATE FUNCTION test_name(string, string)"
" returns bigint location '/foo/bar.so'"
" init_fn='Init' update_fn='Update'"
" merge_fn='Merge' finalize_fn='Finalize'")
assert result == expected
def test_create_uda(self):
def make_ex(serialize=False):
if serialize:
serialize = "\nserialize_fn='Serialize'"
else:
serialize = ""
return (("CREATE AGGREGATE FUNCTION "
"bar.`test_name`(string, string)"
" returns bigint location '/foo/bar.so'"
"\ninit_fn='Init'"
"\nupdate_fn='Update'"
"\nmerge_fn='Merge'") + serialize +
("\nfinalize_fn='Finalize'"))
for ser in [True, False]:
stmt = ddl.CreateAggregateFunction('/foo/bar.so', self.inputs,
self.output, 'Update', 'Init',
'Merge',
'Serialize' if ser else None,
'Finalize', self.name, 'bar')
result = stmt.compile()
expected = make_ex(ser)
assert result == expected
def create_uda(self, uda_info, name=None, database=None):
"""
Creates a user-defined aggregate function within Impala
Parameters
----------
uda_info : UDAInfo object
name : string (optional)
database : string (optional)
"""
if name is None:
name = uda_info.get_name()
statement = ddl.CreateAggregateFunction(uda_info.hdfs_file,
uda_info.inputs,
uda_info.output,
uda_info.init_fn,
uda_info.update_fn,
uda_info.merge_fn,
uda_info.finalize_fn,
name,
database=database)
self._execute(statement)