def test_serialize_custom_class_with_custom_name(self):
@serialization.register_softlearning_serializable(
'TestPackage', 'CustomName')
class OtherTestClass(object):
def __init__(self, val):
self._val = val
def get_config(self):
return {'val': self._val}
serialized_name = 'TestPackage>CustomName'
inst = OtherTestClass(val=5)
class_name = serialization._GLOBAL_CUSTOM_NAMES[OtherTestClass]
self.assertEqual(serialized_name, class_name)
fn_class_name = serialization.get_registered_name(OtherTestClass)
self.assertEqual(fn_class_name, class_name)
cls = serialization.get_registered_object(fn_class_name)
self.assertEqual(OtherTestClass, cls)
config = serialization.serialize_softlearning_object(inst)
self.assertEqual(class_name, config['class_name'])
new_inst = serialization.deserialize_softlearning_object(config)
self.assertIsNot(inst, new_inst)
self.assertIsInstance(new_inst, OtherTestClass)
self.assertEqual(5, new_inst._val)
def deserialize(name, custom_objects=None):
"""Returns a value function or class denoted by input string.
Arguments:
name : String
Returns:
Value function function or class denoted by input string.
For example:
>>> softlearning.value_functions.get('double_feedforward_Q_function')
<function double_feedforward_Q_function at 0x7f86e3691e60>
>>> softlearning.value_functions.get('abcd')
Traceback (most recent call last):
...
ValueError: Unknown value function: abcd
Args:
name: The name of the value function.
Raises:
ValueError: `Unknown value function` if the input string does not
denote any defined value function.
"""
return deserialize_softlearning_object(
name,
module_objects=globals(),
custom_objects=custom_objects,
printable_module_name='value function')
def test_serialize_custom_class_with_default_name(self):
@serialization.register_softlearning_serializable()
class TestClass(object):
def __init__(self, value):
self._value = value
def get_config(self):
return {'value': self._value}
serialized_name = 'Custom>TestClass'
inst = TestClass(value=10)
class_name = serialization._GLOBAL_CUSTOM_NAMES[TestClass]
self.assertEqual(serialized_name, class_name)
config = serialization.serialize_softlearning_object(inst)
self.assertEqual(class_name, config['class_name'])
new_inst = serialization.deserialize_softlearning_object(config)
self.assertIsNot(inst, new_inst)
self.assertIsInstance(new_inst, TestClass)
self.assertEqual(10, new_inst._value)
# Make sure registering a new class with same name will fail.
with self.assertRaisesRegex(ValueError,
".*has already been registered.*"):
@serialization.register_softlearning_serializable() # pylint: disable=function-redefined
class TestClass(object):
def __init__(self, value):
self._value = value
def get_config(self):
return {'value': self._value}
def deserialize(name, custom_objects=None):
"""Returns a replay pool function or class denoted by input string.
Arguments:
name : String
Returns:
Replay Pool function or class denoted by input string.
For example:
>>> softlearning.replay_pools.get({'class_name': 'SimpleReplayPool', ...})
<softlearning.replay_pools.simple_replay_pool.SimpleReplayPool object at 0x7fea93d6cdd0>
>>> softlearning.replay_pools.get('abcd')
Traceback (most recent call last):
...
ValueError: Unknown replay pool: abcd
Args:
name: The name of the replay pool.
Raises:
ValueError: `Unknown replay pool` if the input string does not
denote any defined replay pool.
"""
return deserialize_softlearning_object(name,
module_objects=globals(),
custom_objects=custom_objects,
printable_module_name='replay pool')
def deserialize(name, custom_objects=None):
"""Returns a algorithm function or class denoted by input string.
Arguments:
name : String
Returns:
Algorithm function or class denoted by input string.
For example:
>>> softlearning.algorithms.get({'class_name': 'SAC', ...})
<softlearning.algorithms.sac.SAC object at 0x7fea93d6cdd0>
>>> softlearning.algorithms.get('abcd')
Traceback (most recent call last):
...
ValueError: Unknown algorithm: abcd
Args:
name: The name of the algorithm.
Raises:
ValueError: `Unknown algorithm` if the input string does not
denote any defined algorithm.
"""
return deserialize_softlearning_object(name,
module_objects=globals(),
custom_objects=custom_objects,
printable_module_name='algorithm')
def deserialize(name, custom_objects=None):
"""Returns a preprocessor function or class denoted by input string.
Arguments:
name : String
Returns:
Preprocessor function or class denoted by input string.
For example:
>>> softlearning.preprocessors.get('convnet_preprocessor')
<function convnet_preprocessor at 0x7fd170125950>
>>> softlearning.preprocessors.get('abcd')
Traceback (most recent call last):
...
ValueError: Unknown preprocessor: abcd
Args:
name: The name of the preprocessor.
Raises:
ValueError: `Unknown preprocessor` if the input string does not
denote any defined preprocessor.
"""
return deserialize_softlearning_object(
name,
module_objects=globals(),
custom_objects=custom_objects,
printable_module_name='preprocessor')
def deserialize(name, custom_objects=None):
"""Returns a sampler function or class denoted by input string.
Arguments:
name : String
Returns:
Sampler function or class denoted by input string.
For example:
>>> softlearning.samplers.get({'class_name': 'SimpleSampler', ...})
<softlearning.samplers.simple_sampler.SimpleSampler object at 0x7fea93d6cdd0>
>>> softlearning.samplers.get('abcd')
Traceback (most recent call last):
...
ValueError: Unknown sampler: abcd
Args:
name: The name of the sampler.
Raises:
ValueError: `Unknown sampler` if the input string does not
denote any defined sampler.
"""
return deserialize_softlearning_object(name,
module_objects=globals(),
custom_objects=custom_objects,
printable_module_name='sampler')
def deserialize(name, custom_objects=None):
"""Returns a policy function or class denoted by input string.
Arguments:
name : String
Returns:
Policy function or class denoted by input string.
For example:
>>> softlearning.policies.get({
... 'class_name': 'ContinuousUniformPolicy',
... 'config': {
... 'action_range': [[-1], [1]],
... 'input_shapes': tf.TensorShape((3, )),
... 'output_shape': 2
... }
... })
<softlearning.policies.uniform_policy.ContinuousUniformPolicy object at 0x7fea93d6cdd0>
>>> softlearning.policies.get('abcd')
Traceback (most recent call last):
...
ValueError: Unknown policy: abcd
Args:
name: The name of the policy.
Raises:
ValueError: `Unknown policy` if the input string does not
denote any defined policy.
"""
return deserialize_softlearning_object(name,
module_objects=globals(),
custom_objects=custom_objects,
printable_module_name='policy')
def test_serialize_custom_function(self):
@serialization.register_softlearning_serializable()
def my_fn():
return 42
serialized_name = 'Custom>my_fn'
class_name = serialization._GLOBAL_CUSTOM_NAMES[my_fn]
self.assertEqual(serialized_name, class_name)
fn_class_name = serialization.get_registered_name(my_fn)
self.assertEqual(fn_class_name, class_name)
config = serialization.serialize_softlearning_object(my_fn)
self.assertEqual(class_name, config)
fn = serialization.deserialize_softlearning_object(config)
self.assertEqual(42, fn())
fn_2 = serialization.get_registered_object(fn_class_name)
self.assertEqual(42, fn_2())
def test_serialize_none(self):
serialized = serialization.serialize_softlearning_object(None)
self.assertEqual(serialized, None)
deserialized = serialization.deserialize_softlearning_object(
serialized)
self.assertEqual(deserialized, None)
