def test_simple_cache(self):
signature = EnsembleSignature(
signatures=[ParentChildByNameTopologySignature()])
cache = signature.signature_cache_class()
for node in simple_unique_node_tree().nodes():
tokens = signature.get_signature(node, parent=node.parent())
if not cache.node_count():
# nothing added yet
self.assertEqual([], cache.multiplicity(signature=tokens))
else:
self.assertEqual([0], cache.multiplicity(signature=tokens))
cache[tokens, ProcessStartEvent] = {"value": 0}
self.assertEqual([1], cache.multiplicity(signature=tokens))
self.assertEqual([4], cache.node_count())
self.assertEqual([4], cache.multiplicity())
signature = EnsembleSignature(
signatures=[ParentChildByNameTopologySignature()])
cache = signature.signature_cache_class()
for node in simple_monitoring_tree().nodes():
cache[signature.get_signature(node, parent=node.parent()),
ProcessStartEvent] = {
"value": 0
}
self.assertEqual([3], cache.node_count())
self.assertEqual([4], cache.multiplicity())
def test_to_index(self):
prototype = simple_prototype()
index = prototype.to_index(signature=ParentChildByNameTopologySignature())
self.assertEqual(2, index.multiplicity(signature="root_1"))
self.assertEqual(4, index.multiplicity(signature="test_149160533"))
self.assertEqual(4, index.multiplicity(signature="muh_149160533"))
self.assertEqual(3, index.node_count())
self.assertEqual(2, index.get_statistics(
signature="muh_149160533",
key="value",
event_type=ProcessExitEvent)._statistics[1].mean)
self.assertEqual(0, index.get_statistics(
signature="muh_149160533",
key="value",
event_type=ProcessExitEvent).distance(2))
index = prototype.to_index(
signature=ParentChildByNameTopologySignature(), supported={
ProcessStartEvent: True,
ProcessExitEvent: False
})
self.assertEqual(3, index.node_count())
self.assertEqual(1, index.multiplicity(signature="root_1"))
self.assertEqual(2, index.multiplicity(signature="test_149160533"))
self.assertEqual(2, index.multiplicity(signature="muh_149160533"))
def test_simple_prototype_cache(self):
signature = EnsembleSignature(
signatures=[ParentChildByNameTopologySignature()])
cache = signature.prototype_signature_cache_class(
supported={ProcessExitEvent: True})
tree = simple_unique_node_tree()
for node in tree.nodes():
tokens = signature.get_signature(node, parent=node.parent())
if not cache.node_count():
# nothing added yet
self.assertEqual([], cache.get(signature=tokens))
else:
self.assertTrue(
isinstance(cache.get(signature=tokens)[0], dict))
cache[tokens, tree, ProcessExitEvent] = {"value": 1}
stats = cache.get_statistics(signature=tokens,
prototype=tree,
key="value",
event_type=ProcessExitEvent)
self.assertTrue(stats[0].count() >= 1)
self.assertEqual([4], cache.node_count())
self.assertEqual([4], cache.multiplicity())
signature = EnsembleSignature(
signatures=[ParentChildByNameTopologySignature()])
cache = signature.signature_cache_class()
for node in simple_monitoring_tree().nodes():
cache[signature.get_signature(node, parent=node.parent()),
ProcessStartEvent] = {
"value": 0
}
self.assertEqual([3], cache.node_count())
self.assertEqual([4], cache.multiplicity())
def test_real_trees(self):
def distance_builder(**kwargs):
distance = StartExitDistance()
distance.supported[TrafficEvent] = True
return distance
tree_one = real_tree()
tree_two = real_tree(path="data/c01-007-102/2/1129-2-process.csv")
prototype_names = ["1", "2"]
prototype_caches = []
for index, tree in enumerate([tree_one, tree_two]):
prototype_caches.append(PrototypeSignatureCache.from_signature_caches(
[tree.to_index(signature=ParentChildByNameTopologySignature(),
supported={
ProcessStartEvent: True,
ProcessExitEvent: True,
TrafficEvent: True
},
statistics_cls=SetStatistics)],
prototype=prototype_names[index], threshold=0))
decorator = AnomalyDecorator()
algorithm = IncrementalDistanceAlgorithm(
signature=ParentChildByNameTopologySignature(),
distance=distance_builder,
cache_statistics=SetStatistics
)
prototype_cache = prototype_caches[0]
for cache in prototype_caches[1:]:
prototype_cache += cache
algorithm.cluster_representatives(
signature_prototypes=[prototype_cache],
prototypes=prototype_names
)
decorator.wrap_algorithm(algorithm)
for tree in [tree_one, tree_two]:
algorithm.start_tree()
for event in tree.event_iter(supported=algorithm.supported):
try:
algorithm.add_event(event)
except EventNotSupportedException:
pass
algorithm.finish_tree()
# First tree vs. first is False for start
self.assertFalse(decorator.data()[0][0][0][0])
# ... and end
self.assertFalse(decorator.data()[0][0][0][-1])
# First tree vs. second is False for start
self.assertFalse(decorator.data()[0][0][1][0])
# and True for end
self.assertTrue(decorator.data()[0][0][1][-1])
# Second tree vs. first is False for start
self.assertFalse(decorator.data()[1][0][0][0])
# ... and True for end
self.assertTrue(decorator.data()[1][0][0][-1])
# Second tree vs. second is False for start
self.assertFalse(decorator.data()[1][0][1][0])
# ... and False for end
self.assertFalse(decorator.data()[1][0][1][-1])
def test_simple_clustering(self):
clusterer = Clustering(distance=StartExitDistance())
# create an index to cluster
builder = CSVTreeBuilder()
tree = builder.build(self.file_path_one)
tree_two = builder.build(self.file_path_two)
tree_index = tree.to_index(
signature=ParentChildByNameTopologySignature())
tree_two_index = tree_two.to_index(
signature=ParentChildByNameTopologySignature())
clusterer[1] = tree_index
clusterer[1] = tree_two_index
self.assertEqual(0, len(clusterer.clusterer.clusters))
self.assertEqual(2, len(clusterer.clusterer.noise))
def test_two_prototype_ensembles(self):
signature = EnsembleSignature(signatures=[
ParentChildByNameTopologySignature(),
ParentChildOrderTopologySignature()
])
cache = signature.prototype_signature_cache_class()
prototype = simple_unique_node_tree()
for node in prototype.nodes():
tokens = signature.get_signature(node, parent=node.parent())
if not cache.node_count():
self.assertEqual([], cache.get(signature=tokens))
else:
received = cache.get(signature=tokens)
self.assertEqual(2, len(received))
self.assertTrue(isinstance(received[0], dict))
cache[tokens, prototype, ProcessStartEvent] = {"value": 1}
received = cache.get(signature=tokens)
self.assertEqual(2, len(received))
self.assertTrue(isinstance(received[0], dict))
self.assertTrue(
received[0][prototype][ProcessStartEvent]["value"].count() >= 1
)
self.assertEqual([4, 4], cache.node_count())
self.assertEqual([4, 4], cache.multiplicity())
def test_parameter_distance(self):
prototype = Prototype()
root = prototype.add_node("root", tme=0, exit_tme=0, pid=1, ppid=0, param=1)
for i in range(5):
root.add_node("child_%d" % i, tme=0, exit_tme=0, pid=i + 2, ppid=1, param=1)
next(root.children()).add_node("child", tme=0, exit_tme=0, pid=8, ppid=2, param=1)
tree = Prototype()
root = tree.add_node("root", tme=0, exit_tme=0, pid=1, ppid=0, param=1)
for i in range(5):
root.add_node("child_%d" % i, tme=0, exit_tme=0, pid=i + 2, ppid=1, param=4)
next(root.children()).add_node("child", tme=0, exit_tme=0, pid=8, ppid=2, param=4)
for weight, result in [(1, 0), (.5, 6), (0, 12)]:
def distance(**kwargs):
distance = StartExitDistance(weight=weight, **kwargs)
distance.supported = {
ProcessStartEvent: True,
ProcessExitEvent: True,
ParameterEvent: True
}
return distance
signature = EnsembleSignature(signatures=[ParentChildByNameTopologySignature()])
algorithm = IncrementalDistanceAlgorithm(
signature=signature,
distance=distance,
cache_statistics=SetStatistics
)
decorator = DistanceMatrixDecorator(normalized=False)
decorator.wrap_algorithm(algorithm)
algorithm.prototypes = [prototype]
algorithm.start_tree()
algorithm.add_events(tree.event_iter(supported=algorithm.supported))
algorithm.finish_tree()
self.assertEqual(result, decorator.data()[0][0][0])
def test_signatures(self):
decorator = SignatureDecorator()
algorithm = IncrementalDistanceAlgorithm(
signature=ParentChildByNameTopologySignature())
algorithm.prototypes = [simple_prototype()]
decorator.wrap_algorithm(algorithm)
decorator.start_tree()
self.assertEqual(decorator.descriptive_data(), {"signature": [[[]]]})
decorator.finish_tree()
self.assertEqual(decorator.descriptive_data(), {"signature": [[[]]]})
decorator.wrap_algorithm(algorithm)
decorator.start_tree()
for event in Event.from_tree(simple_monitoring_tree(),
supported={ProcessStartEvent: True}):
decorator.add_event(event)
decorator.finish_tree()
description = decorator.descriptive_data()
self.assertEqual(len(description["signature"]), 1)
self.assertEqual(len(set(description["signature"][0][0])), 3)
self.assertEqual(len(description["signature"][0][0]), 4)
self.assertEqual(description["signature"][0][0][0], "root_1")
self.assertEqual(description["signature"][0][0][1], "test_149160533")
self.assertEqual(description["signature"][0][0][3], "muh_149160533")
decorator.start_tree()
for event in Event.from_tree(simple_monitoring_tree(),
supported={ProcessStartEvent: True}):
decorator.add_event(event)
decorator.finish_tree()
description = decorator.descriptive_data()
self.assertEqual(len(description["signature"]), 2)
def test_more_cluster_representatives(self):
tree_1 = real_tree()
tree_2 = real_tree()
signature = ParentChildByNameTopologySignature()
signature_caches = []
for tree in [tree_1, tree_2]:
signature_caches.append(
tree.to_index(signature=signature,
supported={
ProcessStartEvent: True,
ProcessExitEvent: True,
TrafficEvent: True
},
statistics_cls=SetStatistics))
cr = PrototypeSignatureCache.from_signature_caches(signature_caches,
prototype=tree_1)
for token in cr:
self.assertEqual(cr.multiplicity(token, tree_1),
signature_caches[0].multiplicity(token))
self.assertEqual(
cr.get_statistics(signature=token,
key="value",
prototype=tree_1,
event_type=ProcessExitEvent).count(),
signature_caches[0].get_statistics(
signature=token, key="value",
event_type=ProcessExitEvent).count())
def test_negative_values(self):
def distance_builder(**kwargs):
distance = SimpleDistance(**kwargs)
return distance
tree_one = real_tree(path="data/c01-007-102/2/1078-2-process.csv",
absolute=True)
tree_two = real_tree(path="data/c01-007-102/2/1165-2-process.csv",
absolute=True)
signature = ParentChildByNameTopologySignature()
algorithm = IncrementalDistanceAlgorithm(
signature=signature,
distance=distance_builder,
cache_statistics=SetStatistics)
algorithm.prototypes = [tree_one, tree_two]
decorator = DistanceMatrixDecorator(normalized=True)
decorator.wrap_algorithm(algorithm)
algorithm.start_tree()
for event in tree_one.event_iter(supported=algorithm.supported):
try:
algorithm.add_event(event)
except EventNotSupportedException:
pass
algorithm.finish_tree()
algorithm.start_tree()
for event in tree_two.event_iter(supported=algorithm.supported):
try:
algorithm.add_event(event)
except EventNotSupportedException:
pass
algorithm.finish_tree()
print(decorator.data())
self.assertTrue(False)
def test_external(self):
def distance(**kwargs):
distance = StartExitDistance(weight=0, **kwargs)
distance.supported[TrafficEvent] = True
return distance
algorithm = IncrementalDistanceAlgorithm(
signature=ParentChildByNameTopologySignature(),
distance=distance,
cache_statistics=SplittedStatistics)
decorator = DistanceMatrixDecorator(normalized=False)
decorator.wrap_algorithm(algorithm)
the_tree = real_tree(
path="data/c01-007-102/2/1146-2-process.csv",
absolute=True
)
algorithm.prototypes = [the_tree]
algorithm.start_tree()
for event in the_tree.event_iter(supported=algorithm.supported):
try:
algorithm.add_event(event)
except EventNotSupportedException:
pass
algorithm.finish_tree()
print(decorator.data())
self.assertTrue(False)
def test_from_signature_frequency(self):
def cache_builder():
return SignatureCache(supported={
ProcessStartEvent: True,
ProcessExitEvent: True,
TrafficEvent: True
},
statistics_cls=SetStatistics)
prototype = simple_prototype()
signature = ParentChildByNameTopologySignature()
validation_index = prototype.to_index(signature=signature,
cache=cache_builder())
prototype_index = PrototypeSignatureCache.from_signature_caches(
[
prototype.to_index(signature, cache=cache_builder())
for _ in range(10)
],
prototype=prototype,
threshold=0)
self.assertEqual(validation_index.multiplicity(),
prototype_index.multiplicity())
for token in validation_index:
self.assertEqual(
validation_index.multiplicity(signature=token),
prototype_index.multiplicity(signature=token,
prototype=prototype))
def test_ensemble(self):
def distance(**kwargs):
distance = StartExitDistance(weight=.5, **kwargs)
distance.supported[TrafficEvent] = False
return distance
signature = EnsembleSignature(
signatures=[
ParentChildByNameTopologySignature(), ParentSiblingSignature(width=2)])
algorithm = IncrementalDistanceAlgorithm(
signature=signature,
distance=distance,
cache_statistics=SetStatistics
)
decorator = DistanceMatrixDecorator(normalized=True)
decorator.wrap_algorithm(algorithm)
algorithm.prototypes = [simple_prototype()]
algorithm.start_tree()
for event in simple_monitoring_tree().event_iter(supported=algorithm.supported):
try:
algorithm.add_event(event)
except EventNotSupportedException:
pass
algorithm.finish_tree()
print(decorator.data())
self.assertTrue(False)
def test_ensemble_with_prototypes(self):
decorator = AnomalyDecorator()
algorithm = IncrementalDistanceAlgorithm(
signature=EnsembleSignature(
signatures=[ParentChildByNameTopologySignature(),
ParentChildOrderByNameTopologySignature()]))
algorithm.prototypes = [simple_prototype(), simple_monitoring_tree()]
decorator.wrap_algorithm(algorithm)
algorithm.start_tree()
for event in Event.from_tree(simple_monitoring_tree(), supported={ProcessStartEvent: True}):
algorithm.add_event(event)
algorithm.finish_tree()
self.assertEqual(decorator.data(), [
[
[
[False, False, False, False, False],
[False, False, False, False, False]
],
[
[False, False, False, False, True],
[False, False, False, False, False]
]
]
])
def test_normalized_results(self):
decorator = DistanceDecorator(normalized=True)
algorithm = IncrementalDistanceAlgorithm(
signature=EnsembleSignature(
signatures=[ParentChildByNameTopologySignature(),
ParentChildOrderByNameTopologySignature()])
)
algorithm.prototypes = [simple_prototype(), simple_monitoring_tree()]
decorator.wrap_algorithm(algorithm)
algorithm.start_tree()
self.assertEqual([[[[], []], [[], []]]], decorator.data())
for event in Event.from_tree(simple_monitoring_tree(), supported={ProcessStartEvent: True}):
algorithm.add_event(event)
algorithm.finish_tree()
self.assertEqual([[
[ # ParentChildByNameTopologySignature
[2 / 3, 1 / 3, 1 / 3, 0.0], # simple_prototype
[2 / 3, 1 / 3, 1 / 3, 0.0] # simple_monitoring_tree
],
[ # ParentChildOrderByNameTopologySignature
[4 / 5, 3 / 5, 3 / 5, 2 / 5],
[2 / 3, 1 / 3, 1 / 3, 0.0]
]]], decorator.data())
def test_same_attributes_different_count(self):
tree_1 = Prototype()
root = tree_1.add_node("root", pid=1, ppid=0, tme=0, exit_tme=0)
for _ in range(5):
root.add_node("node", pid=2, ppid=1, tme=0, exit_tme=0)
tree_2 = Prototype()
root = tree_2.add_node("root", pid=1, ppid=0, tme=0, exit_tme=0)
for _ in range(35):
root.add_node("node", pid=2, ppid=1, tme=0, exit_tme=0)
signature = ParentChildByNameTopologySignature()
algorithm = IncrementalDistanceAlgorithm(
signature=signature,
distance=lambda **kwargs: StartExitDistance(weight=0, **kwargs),
cache_statistics=SplittedStatistics)
algorithm.prototypes = [tree_1, tree_2]
decorator = DistanceMatrixDecorator(normalized=False)
decorator.wrap_algorithm(algorithm)
algorithm.start_tree()
for event in tree_1.event_iter(supported=algorithm.supported):
try:
algorithm.add_event(event)
except EventNotSupportedException:
pass
algorithm.finish_tree()
algorithm.start_tree()
for event in tree_2.event_iter(supported=algorithm.supported):
try:
algorithm.add_event(event)
except EventNotSupportedException:
pass
algorithm.finish_tree()
data = decorator.data()
self.assertEqual(data[0][0][1], data[1][0][0])
def test_merging(self):
signature = ParentChildByNameTopologySignature()
one = simple_prototype()
two = simple_additional_monitoring_tree()
prototype_one = one.to_prototype(signature=signature,
supported={
ProcessStartEvent: True,
ProcessExitEvent: False,
TrafficEvent: False
},
statistics_cls=SetStatistics)
prototype_two = two.to_prototype(signature=signature,
supported={
ProcessStartEvent: True,
ProcessExitEvent: False,
TrafficEvent: False
},
statistics_cls=SetStatistics)
self.assertEqual(one.node_count(),
prototype_one.multiplicity(prototype=one))
self.assertEqual(two.node_count(),
prototype_two.multiplicity(prototype=two))
self.assertEqual(0, prototype_one.multiplicity(two))
prototype_one += prototype_two
self.assertEqual(one.node_count(),
prototype_one.multiplicity(prototype=one))
self.assertEqual(two.node_count(),
prototype_one.multiplicity(prototype=two))
self.assertEqual(one.node_count() + two.node_count(),
prototype_one.multiplicity())
prototype_three = one.to_prototype(signature=signature,
supported={
ProcessStartEvent: True,
ProcessExitEvent: True,
TrafficEvent: False
},
statistics_cls=SetStatistics)
prototype_four = two.to_prototype(signature=signature,
supported={
ProcessStartEvent: True,
ProcessExitEvent: True,
TrafficEvent: False
},
statistics_cls=SetStatistics)
self.assertEqual(one.node_count() * 2,
prototype_three.multiplicity(prototype=one))
self.assertEqual(two.node_count() * 2,
prototype_four.multiplicity(prototype=two))
self.assertEqual(0, prototype_three.multiplicity(two))
prototype_three += prototype_four
self.assertEqual(one.node_count() * 2,
prototype_three.multiplicity(prototype=one))
self.assertEqual(two.node_count() * 2,
prototype_three.multiplicity(prototype=two))
self.assertEqual(one.node_count() * 2 + two.node_count() * 2,
prototype_three.multiplicity())
def test_simple_signature(self):
signature = EnsembleSignature(signatures=[ParentChildByNameTopologySignature()])
self.assertEqual(1, signature.count)
for node in simple_monitoring_tree().nodes():
signature.prepare_signature(node, parent=node.parent())
for node in simple_monitoring_tree().nodes():
token = signature.get_signature(node, parent=node.parent())
self.assertIsNotNone(token)
self.assertEqual(1, len(token))
def test_mean(self):
distance = ClusterDistance(distance=StartExitDistance())
prototype_one = simple_prototype()
prototype_two = simple_monitoring_tree()
signature = ParentChildByNameTopologySignature()
mean_prototype = distance.mean([
prototype_one.to_index(signature),
prototype_two.to_index(signature)
])
self.assertEqual(3, mean_prototype.node_count())
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 test_ensemble_signature(self):
decorator = SignatureDecorator()
algorithm = IncrementalDistanceAlgorithm(signature=EnsembleSignature(
signatures=[
ParentChildByNameTopologySignature(),
ParentChildOrderByNameTopologySignature()
]))
algorithm.prototypes = [simple_prototype(), simple_monitoring_tree()]
decorator.wrap_algorithm(algorithm)
algorithm.start_tree()
for event in Event.from_tree(simple_monitoring_tree(),
supported={ProcessStartEvent: True}):
algorithm.add_event(event)
algorithm.finish_tree()
self.assertEqual(
decorator.descriptive_data(), {
'signature': [[[
'root_1', 'test_149160533', 'test_149160533',
'muh_149160533'
],
[
'.0_root_1', '.0.0_test_245236498',
'.0.0_test_245236498', '.0.1_muh_245236498'
]]]
})
algorithm.start_tree()
for event in Event.from_tree(simple_prototype(),
supported={ProcessStartEvent: True}):
algorithm.add_event(event)
algorithm.finish_tree()
self.assertEqual(
decorator.descriptive_data(), {
'signature':
[[[
'root_1', 'test_149160533', 'test_149160533',
'muh_149160533'
],
[
'.0_root_1', '.0.0_test_245236498',
'.0.0_test_245236498', '.0.1_muh_245236498'
]],
[[
'root_1', 'test_149160533', 'muh_149160533',
'test_149160533', 'muh_149160533'
],
[
'.0_root_1', '.0.0_test_245236498', '.0.1_muh_245236498',
'.0.2_test_245236498', '.0.3_muh_245236498'
]]]
})
def test_two_ensembles_different_frequency(self):
signature = EnsembleSignature(signatures=[
ParentChildByNameTopologySignature(),
ParentChildOrderTopologySignature()
])
cache = signature.signature_cache_class()
for node in simple_monitoring_tree().nodes():
cache[signature.get_signature(node, parent=node.parent()),
ProcessStartEvent] = {
"value": 0
}
self.assertEqual([3, 4], cache.node_count())
self.assertEqual([4, 4], cache.multiplicity())
def test_maximum_generator_newerincremental_with_signature_distance(self):
rg = RandomGenerator(relative_matching=1)
signature = ParentChildByNameTopologySignature()
alg = IncrementalDistanceAlgorithm(signature=signature)
alg.prototypes = [rg.prototype]
decorator = DistanceDecorator(normalized=True)
decorator.algorithm = alg
decorator.start_tree()
for event in rg:
decorator.add_event(event)
decorator.finish_tree()
self.assertEqual(decorator.data()[-1][-1][-1][-1], 0)
def test_from_signatures(self):
cache = SignatureCache(supported={
ProcessStartEvent: True,
ProcessExitEvent: True,
TrafficEvent: True
},
statistics_cls=SetStatistics)
cache_two = SignatureCache(supported={
ProcessStartEvent: True,
ProcessExitEvent: True,
TrafficEvent: True
},
statistics_cls=SetStatistics)
prototype = simple_additional_monitoring_tree()
other_prototype = simple_prototype()
tree_index = prototype.to_index(
signature=ParentChildByNameTopologySignature(), cache=cache)
prototype_index = other_prototype.to_index(
signature=ParentChildByNameTopologySignature(), cache=cache_two)
prototype_cache = PrototypeSignatureCache.from_signature_caches(
[tree_index, prototype_index], prototype=1, threshold=.9)
self.assertEqual(9, prototype_cache.multiplicity())
self.assertEqual(3, prototype_cache.node_count())
def test_initialisation_of_cluster_representatitives(self):
def distance_builder(**kwargs):
distance = StartExitDistance(weight=.5, **kwargs)
distance.supported[TrafficEvent] = True
return distance
tree_one = real_tree()
tree_two = real_tree(path="data/c01-007-102/2/1129-2-process.csv")
tree_three = real_tree(path="data/c01-007-102/2/1136-3-process.csv")
distance = StartExitDistance(weight=.5)
distance.supported[TrafficEvent] = True
signature = ParentChildByNameTopologySignature()
tree_profiles = [
tree_one.to_index(signature=signature,
supported=distance.supported,
statistics_cls=SetStatistics),
tree_two.to_index(signature=signature,
supported=distance.supported,
statistics_cls=SetStatistics),
tree_three.to_index(signature=signature,
supported=distance.supported,
statistics_cls=SetStatistics)
]
algorithm = IncrementalDistanceAlgorithm(
signature=ParentChildByNameTopologySignature(),
distance=distance_builder,
cache_statistics=SetStatistics)
cluster_distance = ClusterDistance(distance=distance)
prototype_names = ["test"]
prototype_signatures = []
for prototype in prototype_names:
prototype_signatures.append(
cluster_distance.mean(tree_profiles, prototype=prototype))
algorithm.cluster_representatives(
signature_prototypes=prototype_signatures,
prototypes=prototype_names)
algorithm.start_tree()
def test_creation(self):
test_algorithm = algorithm(ParentChildByNameTopologySignature())
test_algorithm.prototypes = ["1", "2", "3"]
distance = Distance(signature_count=test_algorithm.signature.count)
distance.init_distance(
prototypes=test_algorithm.prototypes,
signature_prototypes=test_algorithm.signature_prototypes)
last_index = 0
for index, dist in enumerate(
distance.iter_on_prototypes(test_algorithm.prototypes)):
self.assertEqual(dist, [0])
last_index = index
self.assertEqual(last_index, 2)
self.assertEqual(distance.node_count(), [0])
self.assertFalse(distance.is_prototype_based_on_original())
def test_empty_nodes(self):
signature = EnsembleSignature(
signatures=[ParentChildByNameTopologySignature(),
ParentCountedChildrenByNameTopologySignature(count=3)])
algorithm = IncrementalDistanceAlgorithm(signature=signature)
algorithm.prototypes = [simple_prototype()]
algorithm.start_tree()
for event in simple_monitoring_tree().event_iter(supported=algorithm.supported):
try:
distance = algorithm.add_event(event)
except EventNotSupportedException:
pass
algorithm.finish_tree()
self.assertEqual([[0], [7]], distance[-1])
def test_two_prototype_ensembles_different_frequency(self):
signature = EnsembleSignature(signatures=[
ParentChildByNameTopologySignature(),
ParentChildOrderTopologySignature()
])
cache = signature.prototype_signature_cache_class(
supported={ProcessExitEvent: True})
prototype = simple_monitoring_tree()
for node in prototype.nodes():
cache[signature.get_signature(node, parent=node.parent()),
prototype, ProcessExitEvent] = {
"value": 1
}
self.assertEqual([3, 4], cache.node_count())
self.assertEqual([4, 4], cache.multiplicity())
def test_ensemble(self):
decorator = DistanceDecorator(normalized=False)
algorithm = IncrementalDistanceAlgorithm(
signature=EnsembleSignature(
signatures=[ParentChildByNameTopologySignature(),
ParentChildOrderByNameTopologySignature()]))
algorithm.prototypes = [simple_prototype()]
decorator.wrap_algorithm(algorithm)
algorithm.start_tree()
self.assertEqual([[[[]], [[]]]], decorator.data())
for event in Event.from_tree(simple_monitoring_tree(), supported={ProcessStartEvent: True}):
algorithm.add_event(event)
algorithm.finish_tree()
self.assertEqual([[[[2, 1, 1, 0]], [[4, 3, 3, 2]]]], decorator.data())
def prepare_signature(self, node, parent):
ordered_nodes = [node.name] + list(
self.sibling_generator(node, self._width + 1))
ordered_nodes.sort() # ascending order
algorithm_id = "%s_%s_%s" % (
"_".join(ordered_nodes[1:-1]), # up to width ordered siblings
ordered_nodes[-1], # last element as anchor node name
zlib.adler32(
self.get_signature(parent, None, dimension="p").
encode('utf-8', errors='surrogateescape'
) if parent is not None else b''))
p_signature = ParentChildByNameTopologySignature.signature_string(
node.name,
self.get_signature(parent, None, dimension="p")
if parent is not None else '')
self._prepare_signature(node, algorithm_id, p=p_signature)
