def test_get_vertex():
"""Check if vertex can be retrieved from directed graph."""
g = DirectedGraph()
g.add_vertex(vertex_props=None)
v = g.get_vertex(vertex_id=0)
assert v is not None
assert v.id == 0
def test_get_vertex_ids_graph_with_one_vertex():
"""Test whether method DirectedGraph.get_vertex_ids() works correctly for graph with vertex."""
g = DirectedGraph()
v0 = g.add_vertex(vertex_props=None)
list_ids = g.get_vertex_ids()
assert len(list_ids) == 1
assert v0.id in list_ids
def test_get_vertices_graph_with_one_vertex():
"""Test whether method DirectedGraph.get_vertices() works correctly."""
g = DirectedGraph()
v0 = g.add_vertex(vertex_props=None)
list_vertices = g.get_vertices()
assert len(list_vertices) == 1
assert v0 in list_vertices
def test_get_vertex_ids():
g = DirectedGraph()
v0 = g.add_vertex(vertex_props=None)
v1 = g.add_vertex(vertex_props=None)
list_ids = g.get_vertex_ids()
assert len(list_ids) == 2
assert v0.id in list_ids
assert v1.id in list_ids
def test_get_vertices():
g = DirectedGraph()
v0 = g.add_vertex(vertex_props=None)
v1 = g.add_vertex(vertex_props=None)
list_vertices = g.get_vertices()
assert len(list_vertices) == 2
assert v0 in list_vertices
assert v1 in list_vertices
def test_get_vertex_ids():
"""Test whether method DirectedGraph.get_vertex_ids() works correctly."""
g = DirectedGraph()
v0 = g.add_vertex(vertex_props=None)
v1 = g.add_vertex(vertex_props=None)
list_ids = g.get_vertex_ids()
assert len(list_ids) == 2
assert v0.id in list_ids
assert v1.id in list_ids
def test_add_edge_between_nonexistent_vertices():
"""Test whether method DirectedGraph.add_edge() works correctly."""
g = DirectedGraph()
v0 = Vertex(0, {})
v1 = Vertex(1, {})
# neither vertex are put into the directed graph, so it is impossible
# to add an edge to connect them. ATM the exception thrown by the
# add_edge method is not very specific, it might need to be improved.
with pytest.raises(Exception):
g.add_edge(from_id=v0.id, to_id=v1.id)
def test_add_cycle_to_the_same_vertex():
"""Test whether method DirectedGraph.add_edge() works correctly."""
g = DirectedGraph()
v0 = g.add_vertex(vertex_props=None)
# create edge from and to the same vertex
g.add_edge(from_id=v0.id, to_id=v0.id)
v0_neighbours = v0.get_neighbours()
# only one neighbour
assert len(v0_neighbours) == 1
# neighbour has id = 1
assert v0_neighbours[0].id == 0
def test_find_common_reachable_vertex_empty_graph():
"""Test if method DirectedGraph.find_common_reachable_vertices() works correctly."""
list_vertices = DirectedGraph.find_common_reachable_vertices(input_vertices=None)
assert list_vertices is None
list_vertices = DirectedGraph.find_common_reachable_vertices(input_vertices=[])
assert list_vertices is None
v0 = Vertex(0, {})
list_vertices = DirectedGraph.find_common_reachable_vertices(input_vertices=[v0])
assert list_vertices is not None
assert list_vertices[0] == v0
def test_add_edge():
g = DirectedGraph()
v0 = g.add_vertex(vertex_props=None)
v1 = g.add_vertex(vertex_props=None)
g.add_edge(from_id=v0.id, to_id=v1.id)
v0_neighbours = v0.get_neighbours()
# only one neighbour
assert len(v0_neighbours) == 1
# neighbour has id = 1
assert v0_neighbours[0].id == 1
v1_neighbours = v1.get_neighbours()
# no neighbour of v1 as it is directed graph
assert len(v1_neighbours) == 0
def __init__(self, graph_store, synonyms_store):
"""Initialize the analyzer and read known synonyms."""
# load graph from given data store
self.g = DirectedGraph.read_from_json(graph_store)
self.known_licenses = [
'public domain', 'mit', 'bsd-new', 'bsd-simplified', 'apache 2.0',
'lgplv2.1', 'lgplv2.1+', 'lgplv3+', 'mpl 1.1', 'gplv2', 'gplv2+',
'gplv3+', 'affero gplv3', 'epl 1.0', 'epl 2.0', 'cddlv1.1+',
'mpl 2.0', 'w3c', 'bouncycastle', 'cc0v1.0', 'cc-by-3.0', 'edl',
'json', 'postgresql', 'apache 1.1', 'cpl 1.0', 'cpal 1.0', 'ISC',
'gwt', 'psfl'
]
# IMPORTANT: Order matters in the following tuple
self.license_type_tuple = ('P', 'WP', 'SP', 'NP')
# read the json that contains known synonyms
list_synonym_jsons = synonyms_store.list_files()
for synonym_json in list_synonym_jsons:
self.syn = synonyms_store.read_json_file(synonym_json)
break # currently only one synonym json is supported
# identify compatibility classes among the licenses in graph
self.dict_compatibility_classes = {}
self.dict_type_compatibility_classes = {}
self._find_compatibility_classes()
self._find_type_compatibility_classes()
def test_add_cycle():
"""Test whether method DirectedGraph.add_edge() works correctly."""
g = DirectedGraph()
v0 = g.add_vertex(vertex_props=None)
v1 = g.add_vertex(vertex_props=None)
g.add_edge(from_id=v0.id, to_id=v1.id)
g.add_edge(from_id=v1.id, to_id=v0.id)
v0_neighbours = v0.get_neighbours()
# only one neighbour
assert len(v0_neighbours) == 1
# neighbour has id = 1
assert v0_neighbours[0].id == 1
v1_neighbours = v1.get_neighbours()
# no neighbour of v1 as it is directed graph
assert len(v1_neighbours) == 1
# neighbour has id = 0
assert v1_neighbours[0].id == 0
def test_num_vertices_counter():
"""Check the num_vertices counter."""
g = DirectedGraph()
assert g.num_vertices == 0
g.add_vertex(vertex_props=None)
assert g.num_vertices == 1
g.add_vertex(vertex_props=None)
assert g.num_vertices == 2
def test_iterator():
"""Test that the iterator over all vertexes works correctly."""
g = DirectedGraph()
items = [item for item in g]
assert not items
v0 = g.add_vertex(vertex_props={'license': 'L0', 'type': 'P'})
items = [item for item in g]
assert len(items) == 1
assert items[0] == v0
v1 = g.add_vertex(vertex_props={'license': 'L1', 'type': 'WP'})
items = [item for item in g]
assert len(items) == 2
assert v0 in items
assert v1 in items
v2 = g.add_vertex(vertex_props={})
items = [item for item in g]
assert len(items) == 3
assert v0 in items
assert v1 in items
assert v2 in items
def test_find_vertex():
g = DirectedGraph()
v0 = g.add_vertex(vertex_props={'license': 'L0', 'type': 'P'})
v1 = g.add_vertex(vertex_props={'license': 'L1', 'type': 'WP'})
v = g.find_vertex(prop_name='license', prop_value='L1')
assert v != v0
assert v == v1
v = g.find_vertex(prop_name='license', prop_value='L2')
assert v is None
def test_find_vertex():
"""Test if method DirectedGraph.find_vertex() works correctly."""
g = DirectedGraph()
v0 = g.add_vertex(vertex_props={'license': 'L0', 'type': 'P'})
v1 = g.add_vertex(vertex_props={'license': 'L1', 'type': 'WP'})
v = g.find_vertex(prop_name='license', prop_value='L1')
assert v != v0
assert v == v1
v = g.find_vertex(prop_name='license', prop_value='L2')
assert v is None
def test_find_vertex_graph_with_one_vertex():
"""Test if method DirectedGraph.find_vertex() works correctly."""
g = DirectedGraph()
v0 = g.add_vertex(vertex_props={'license': 'L0', 'type': 'P'})
v = g.find_vertex(prop_name='license', prop_value='L0')
assert v is not None
assert v == v0
v = g.find_vertex(prop_name='license', prop_value='something_else')
assert v is None
with pytest.raises(Exception):
v = g.find_vertex(prop_name='unknown_property', prop_value='L0')
def __init__(self, args):
super(GraphSearchPolicy, self).__init__()
self.args = args
self.model = args.model
self.relation_only = args.relation_only
self.history_dim = args.history_dim
self.history_num_layers = args.history_num_layers
self.entity_dim = args.entity_dim
self.relation_dim = args.relation_dim
if self.relation_only:
self.action_dim = args.relation_dim
else:
self.action_dim = args.entity_dim + args.relation_dim
self.ff_dropout_rate = args.ff_dropout_rate
self.rnn_dropout_rate = args.rnn_dropout_rate
self.action_dropout_rate = args.action_dropout_rate
self.num_rollouts = args.num_rollouts
self.bandwidth = args.bandwidth
self.num_rollout_steps = args.num_rollout_steps
self.emb_dropout_rate = args.emb_dropout_rate
self.beam_size = args.beam_size
self.xavier_initialization = args.xavier_initialization
self.relation_only_in_path = args.relation_only_in_path
self.path = None
# Directed Graph
self.dg = DirectedGraph(args.data_dir)
# Graph Transformer hyper params
self.num_heads = 4
self.head_dim = args.entity_dim // self.num_heads
self.hidden_dim = self.history_dim
# Set policy network modules
self.define_modules()
self.initialize_modules()
# Fact network modules
self.fn = None
self.fn_kg = None
def __init__(self, graph_store, synonyms_store):
# load graph from given data store
self.g = DirectedGraph.read_from_json(graph_store)
self.known_licenses = [
'public domain', 'mit', 'bsd-new', 'apache 2.0', 'lgplv2.1',
'lgplv2.1+', 'lgplv3+', 'mpl 1.1', 'gplv2', 'gplv2+', 'gplv3+',
'affero gplv3'
]
# IMPORTANT: Order matters in the following tuple
self.license_type_tuple = ('P', 'WP', 'SP', 'NP')
# read the json that contains known synonyms
list_synonym_jsons = synonyms_store.list_files()
for synonym_json in list_synonym_jsons:
self.syn = synonyms_store.read_json_file(synonym_json)
break # currently only one synonym json is supported
# identify compatibility classes among the licenses in graph
self.dict_compatibility_classes = {}
self.dict_type_compatibility_classes = {}
self._find_compatibility_classes()
self._find_type_compatibility_classes()
def test_get_vertices_empty_graph():
"""Test whether method DirectedGraph.get_vertices() works correctly for empty graph."""
g = DirectedGraph()
list_vertices = g.get_vertices()
assert not list_vertices
def _create_graph():
g = DirectedGraph()
pd = g.add_vertex(vertex_props={'license': 'public domain', 'type': 'P'})
mit = g.add_vertex(vertex_props={'license': 'mit', 'type': 'P'})
bsd = g.add_vertex(vertex_props={'license': 'bsd-new', 'type': 'P'})
apache = g.add_vertex(vertex_props={'license': 'apache 2.0', 'type': 'P'})
lgpl2 = g.add_vertex(vertex_props={'license': 'lgplv2.1', 'type': 'WP'})
lgpl22 = g.add_vertex(vertex_props={'license': 'lgplv2.1+', 'type': 'WP'})
lgpl3 = g.add_vertex(vertex_props={'license': 'lgplv3+', 'type': 'WP'})
mpl = g.add_vertex(vertex_props={'license': 'mpl 1.1', 'type': 'WP'})
gpl2 = g.add_vertex(vertex_props={'license': 'gplv2', 'type': 'SP'})
gpl22 = g.add_vertex(vertex_props={'license': 'gplv2+', 'type': 'SP'})
gpl3 = g.add_vertex(vertex_props={'license': 'gplv3+', 'type': 'SP'})
agpl3 = g.add_vertex(vertex_props={'license': 'agplv3', 'type': 'NP'})
g.add_edge(pd.id, mit.id)
g.add_edge(mit.id, bsd.id)
g.add_edge(bsd.id, apache.id)
g.add_edge(bsd.id, mpl.id)
g.add_edge(bsd.id, lgpl2.id)
g.add_edge(bsd.id, lgpl22.id)
g.add_edge(bsd.id, lgpl3.id)
g.add_edge(apache.id, lgpl3.id)
g.add_edge(lgpl22.id, lgpl2.id)
g.add_edge(lgpl22.id, lgpl3.id)
g.add_edge(lgpl2.id, gpl2.id)
g.add_edge(lgpl2.id, gpl22.id)
g.add_edge(lgpl22.id, gpl22.id)
g.add_edge(lgpl3.id, gpl3.id)
g.add_edge(gpl22.id, gpl2.id)
g.add_edge(gpl22.id, gpl3.id)
g.add_edge(gpl3.id, agpl3.id)
return g
def compute_representative_license(self, input_licenses):
"""Compute representative license for given list of licenses.
First, it tries to identify the input licenses by using known synonyms.
If there exists at least one unknown license, this method gives up.
It makes use of a very popular license graph available in [1]. After
identifying license vertices in the graph, it tries to find the
common reachable vertex from the input license vertices.
If a common reachable vertex is not possible then there is a conflict
and all pairs of conflicting licenses are identified by using the
concept of compatibility classes.
If a common reachable vertex is available, then its license becomes
representative license. Note that we also try to find outlier
licenses when representative license is available.
[1] https://www.dwheeler.com/essays/floss-license-slide.html
:param input_licenses: list of input licenses
:return: representative license with supporting information
"""
output = {
'status': 'Failure',
'reason': 'Input is invalid',
'representative_license': None,
'unknown_licenses': [],
'conflict_licenses': [],
'outlier_licenses': [],
'synonyms': {}
}
if input_licenses is None:
return output
if len(input_licenses) == 0:
return output
# Find synonyms
input_lic_synonyms = [self.find_synonym(y) for y in input_licenses]
output['synonyms'] = dict(list(zip(input_licenses, input_lic_synonyms)))
# Check if all input licenses are known
if len(set(input_lic_synonyms) - set(self.known_licenses)) > 0:
output['status'] = 'Unknown'
output['reason'] = 'Some unknown licenses found'
output['unknown_licenses'] = list(
set(input_lic_synonyms) - set(self.known_licenses))
output['representative_license'] = None
return output
# Let's try to find a representative license
# First, we need to find vertices for input licenses
license_vertices = []
for lic in input_lic_synonyms:
v = self.g.find_vertex(prop_name='license', prop_value=lic)
assert v is not None
license_vertices.append(v)
assert len(license_vertices) == len(input_lic_synonyms)
# Find common reachable vertices from input license vertices
reachable_vertices = DirectedGraph.find_common_reachable_vertices(license_vertices)
if len(reachable_vertices) == 0: # i.e. conflict
output['status'] = 'Conflict'
output['reason'] = 'Some licenses are in conflict'
output['conflict_licenses'] = self._find_conflict_licenses(license_vertices)
output['representative_license'] = None
return output
# Some representative license is possible :)
# Check if one of the input licenses is the representative one
common_destination = None
license_vertex_ids = [x.id for x in license_vertices]
for v in reachable_vertices:
if v.id in license_vertex_ids:
common_destination = v # TODO: should we break after this ?
if common_destination is not None:
output['status'] = 'Successful'
output['reason'] = 'Representative license found'
output['representative_license'] = \
common_destination.get_prop_value(prop_name='license')
rep_lic_vertex = self.g.find_vertex('license', output['representative_license'])
rep_lic_type = rep_lic_vertex.get_prop_value('type')
output['outlier_licenses'] = self._find_outlier_licenses(license_vertices, rep_lic_type)
return output
# If one of the input licenses is NOT the representative one, then
# let us pick up the least restrictive one
for license_type in self.license_type_tuple:
list_common_destinations = [
x
for x in reachable_vertices
if x.get_prop_value(prop_name='type') == license_type
]
if len(list_common_destinations) > 0:
output['status'] = 'Successful'
output['reason'] = 'Representative license found'
output['representative_license'] = \
list_common_destinations[0].get_prop_value(prop_name='license')
rep_lic_vertex = self.g.find_vertex('license', output['representative_license'])
rep_lic_type = rep_lic_vertex.get_prop_value('type')
output['outlier_licenses'] = self._find_outlier_licenses(license_vertices,
rep_lic_type)
return output
# We should have returned by now ! Returning from here is unexpected !
output['status'] = 'Failure'
output['reason'] = 'Something unexpected happened!'
output['representative_license'] = None
return output
def test_get_vertex():
g = DirectedGraph()
g.add_vertex(vertex_props=None)
v = g.get_vertex(vertex_id=0)
assert v is not None
def test_find_vertex_empty_graph():
"""Test if method DirectedGraph.find_vertex() works correctly for empty graph."""
g = DirectedGraph()
v = g.find_vertex(prop_name='license', prop_value='L1')
assert v is None
def test_find_common_reachable_vertex():
g = DirectedGraph()
v0 = g.add_vertex(vertex_props={'license': 'L0', 'type': 'P'})
v1 = g.add_vertex(vertex_props={'license': 'L1', 'type': 'WP'})
g.add_edge(from_id=v0.id, to_id=v1.id)
list_vertices = DirectedGraph.find_common_reachable_vertices(
input_vertices=None)
assert list_vertices is None
list_vertices = DirectedGraph.find_common_reachable_vertices(
input_vertices=[])
assert list_vertices is None
list_vertices = DirectedGraph.find_common_reachable_vertices(
input_vertices=[v0])
assert list_vertices is not None
assert len(list_vertices) == 2
assert list_vertices[0] == v0
assert list_vertices[1] == v1
list_vertices = DirectedGraph.find_common_reachable_vertices(
input_vertices=[v0, v1])
assert list_vertices is not None
assert len(list_vertices) == 1
assert list_vertices[0] == v1
v2 = g.add_vertex(vertex_props={'license': 'L2', 'type': 'SP'})
g.add_edge(from_id=v2.id, to_id=v1.id)
list_vertices = DirectedGraph.find_common_reachable_vertices(
input_vertices=[v0, v2])
assert list_vertices is not None
assert len(list_vertices) == 1
assert list_vertices[0] == v1
def test_get_nonexistent_vertex():
"""Check if non existent vertex can not be retrieved from directed graph."""
g = DirectedGraph()
g.add_vertex(vertex_props=None)
v = g.get_vertex(vertex_id=42)
assert v is None
def test_initial_state():
"""Check the initial state of directed graph."""
g = DirectedGraph()
assert g.num_vertices == 0
def test_add_vertex():
g = DirectedGraph()
v0 = g.add_vertex(vertex_props=None)
assert v0.id >= 0
def test_add_vertex():
"""Check that new vertex can be added to directed graph."""
g = DirectedGraph()
v0 = g.add_vertex(vertex_props=None)
assert v0.id >= 0