def test_all(self):
tedgen = TEDGenerator(costs=[FanoutWeightedTreeEditDistanceCost(),
TreeEditDistanceCost(),
SubtreeWeightedTreeEditDistanceCost(),
SubtreeHeightWeightedTreeEditDistanceCost()],
operation_generator=RandomOperation(delete_probability=0.25,
insert_probability=0.25,
edit_probability=0.25,
move_probability=0),
probability=.5)
prototype = Prototype()
root = prototype.add_node("root", pid=1, ppid=0)
one = root.add_node("test1", pid=2, ppid=1)
root.add_node("test2", pid=3, ppid=1)
root.add_node("test3", pid=4, ppid=1)
one.add_node("test1.1", pid=5, ppid=2)
one.add_node("test1.2", pid=6, ppid=2)
one.add_node("test1.3", pid=7, ppid=2)
two = one.add_node("test1.4", pid=8, ppid=2)
two.add_node("test2.1", pid=9, ppid=8)
two.add_node("test2.2", pid=10, ppid=8)
two.add_node("test2.3", pid=11, ppid=8)
two.add_node("test2.4", pid=12, ppid=8)
result = tedgen.generate(tree=prototype)
result2 = tedgen.generate(tree=prototype)
print("received %s" % result.distance)
print("received %s" % result2.distance)
def test_ensemble_result(self):
"""
I recognised that apparently sometimes (or ever?) values of ensemble
methods don't match results from single calculations. Therefore this
test should uniquely test for this.
"""
tree = real_tree()
tree_generator = TEDGenerator(
operation_generator=RandomOperation(
insert_probability=.5, delete_probability=.5),
costs=[TreeEditDistanceCost()],
seed=1234)
disturbed_tree = tree_generator.generate(tree)
signatures = [ParentChildByNameTopologySignature(),
ParentCountedChildrenByNameTopologySignature(count=2)]
# First test results from ensemble
ensemble_signature = EnsembleSignature(signatures=signatures)
decorator = DistanceMatrixDecorator(normalized=False)
algorithm = IncrementalDistanceAlgorithm(
signature=ensemble_signature,
distance=StartDistance
)
decorator.wrap_algorithm(algorithm)
algorithm.prototypes = [tree]
algorithm.start_tree()
for event in disturbed_tree.event_iter(supported=algorithm.supported):
try:
algorithm.add_event(event)
except EventNotSupportedException:
pass
algorithm.finish_tree()
ensemble_data = decorator.data()
# Second, validate this result with single measurements
single_data = {}
for index, signature in enumerate(signatures):
decorator = DistanceMatrixDecorator(normalized=False)
algorithm = IncrementalDistanceAlgorithm(
signature=signature,
distance=StartDistance
)
decorator.wrap_algorithm(algorithm)
algorithm.prototypes = [real_tree()]
algorithm.start_tree()
for event in disturbed_tree.event_iter(supported=algorithm.supported):
try:
algorithm.add_event(event)
except EventNotSupportedException:
pass
algorithm.finish_tree()
single_data[index] = decorator.data()
for index, _ in enumerate(signatures):
self.assertEqual(ensemble_data[0][index][0], single_data[index][0][0][0])
def _generate_perturbated_tree(kwargs):
"""
:param kwargs:
:param filepath: Path to consider
:param probabilities: List of probabilites
:param repeat: How often to repeat a single probability
:param insert_probability: Probability to insert item
:param delete_probability: Probability to delete item
:param change_probability: Probability to change item
:param move_probability: Probability to move item
:param leaf_nodes_only: Only include leaf nodes?
:param internal_nodes_only: Only include internal nodes?
:param attribute_nodes_only: Only include attribute nodes?
:param cost: True or False
:return:
"""
result = MulticoreResult()
filepath = kwargs.get("filepath", None)
probabilities = kwargs.get("probabilities", [])
repeat = kwargs.get("repeat", 1)
insert_probability = kwargs.get("insert_probability", 0)
delete_probability = kwargs.get("delete_probability", 0)
change_probability = kwargs.get("change_probability", 0)
move_probability = kwargs.get("move_probability", 0)
leaf_nodes_only = kwargs.get("leaf_nodes_only", False)
internal_nodes_only = kwargs.get("internal_nodes_only", False)
attribute_nodes_only = kwargs.get("attribute_nodes_only", False)
cost = kwargs.get("cost", True)
tree_builder = CSVTreeBuilder()
tree = tree_builder.build(filepath)
if tree is not None:
result.setdefault(filepath, {})
result[filepath]["tree"] = tree
result[filepath].setdefault("perturbated_tree", {})
for probability in probabilities:
if attribute_nodes_only:
ted_generator = TEDGenerator(
costs=[],
operation_generator=RandomOperation(
delete_probability=1,
delete_operation=DeleteAttributeTreeEditOperation(
probability=probability)),
probability=1,
skip_node=skip_all_but_attribute_nodes)
else:
ted_generator = TEDGenerator(
costs=[
TreeEditDistanceCost(),
FanoutWeightedTreeEditDistanceCost(),
SubtreeWeightedTreeEditDistanceCost(),
SubtreeHeightWeightedTreeEditDistanceCost(),
SubtreeWeightedTreeEditDistanceCostWithMove()
] if cost else [],
operation_generator=RandomOperation(
insert_probability=insert_probability,
delete_probability=delete_probability,
edit_probability=change_probability,
move_probability=move_probability),
probability=probability,
skip_node=skip_leaf if internal_nodes_only else
(skip_inner_node if leaf_nodes_only else skip_no_node))
for _ in range(repeat):
perturbated_tree = ted_generator.generate(tree)
result[filepath]["perturbated_tree"].setdefault(
probability, []).append(perturbated_tree)
# reload tree
tree = tree_builder.build(filepath)
return result
def test_edit(self):
tedgen = TEDGenerator(costs=[FanoutWeightedTreeEditDistanceCost(), TreeEditDistanceCost()],
operation_generator=RandomOperation(edit_probability=1))
result = tedgen.generate(tree=self.prototype)
print("received %s" % result.distance)
def test_move(self):
tedgen = TEDGenerator(costs=[],
operation_generator=RandomOperation(move_probability=1))
result = tedgen.generate(tree=self.prototype)
print(result)