def test_get_graph_embeddings():
g1 = Graph('ls', convert_dot_to_networkx('./out/ls.dot'))
assert g1.graph is not None
g2 = Graph('cat', convert_dot_to_networkx('./out/cat.dot'))
assert g2.graph is not None
graphs = [g1, g2]
model = get_graph_embeddings(graphs)
assert type(model) is Doc2Vec
if __name__ == "__main__":
datasets = get_datasets(dir='/bin/')
graphs = []
for dataset in datasets:
dataset.set_dot_file_path(
generate_dot_by_radare2(dataset.bin_file_path,
dataset.sample_name))
if not os.path.exists(dataset.dot_file_path):
print(f'{dataset.dot_file_path} is not created.')
print('Please check the reason manually...')
continue
graphs.append(
Graph(dataset.sample_name,
convert_dot_to_networkx(dataset.dot_file_path)))
model = get_graph_embeddings(graphs)
results = pd.DataFrame({},
columns=[
'1st Similar file', '1st Similarity Score',
'2nd Similar file', '2nd Similarity Score',
'3rd Similar file', '3rd Similarity Score'
])
for graph in graphs:
print(f'get files which are similar to {graph.name}')
sim_files = model.dv.most_similar(f'g_{graph.name}')
print(sim_files)
print('-----------------------------------------------')
def test_get_graph_embeddings_with_single_graph():
G = convert_dot_to_networkx(test_file_path)
assert G is not None
with pytest.raises(AssertionError):
get_graph_embeddings(G)
def test_convert_to_networkx():
dot_file_path = './out/ls.dot'
G = convert_dot_to_networkx(dot_file_path)
assert G is not None
def test_get_edge_embeddings():
G = convert_dot_to_networkx(test_file_path)
assert G is not None
wv = get_edge_embeddings(G)
assert type(wv) is HadamardEmbedder
def test_convert_to_networkx_with_empty_saved_path():
dot_file_path = './out/ls.dot'
saved_file_path = './out/ls_cfg.png'
G = convert_dot_to_networkx(dot_file_path, saved_file_path)
assert G is not None
def test_convert_to_networkx_with_empty_dotfile_path():
with pytest.raises(FileNotFoundError):
convert_dot_to_networkx('')
def get_edge_embeddings(G: Union[MultiDiGraph, MultiGraph],
debug: bool = False) -> HadamardEmbedder:
if G is None:
raise TypeError(' A type of G must be MultiDiGraph or MultiGraph')
wv = None
try:
wv = HadamardEmbedder(keyed_vectors=get_node_embedings(G))
except Exception as e:
raise e
if debug:
for idxI in range(len(G.nodes())):
for idxJ in range(idxI):
print('similar vector: ', (idxI, idxJ))
print('similar_edge', wv.most_similar((idxI, idxJ)))
print('')
return wv
if __name__ == '__main__':
try:
G = convert_dot_to_networkx('./out/ls.dot')
embeds = get_edge_embeddings(G)
except Exception:
print(traceback.format_exc())
min_count=min_count)
return model
def save_embeddings(output_path: str, model: Doc2Vec,
graphs: List[TaggedDocument], dimensions: int):
if not os.path.exists(output_path):
with open(output_path, 'wb'):
pass
out = []
for graph in graphs:
out.append([graph.name] + list(model.dv["g_" + graph.name]))
column_names = ["types"] + ["x_" + str(dim) for dim in range(dimensions)]
out = pd.DataFrame(out, columns=column_names)
out.to_csv(output_path, index=None)
if __name__ == "__main__":
dimensions = 128
try:
g1 = Graph('ls', convert_dot_to_networkx('./out/ls.dot'))
g2 = Graph('cat', convert_dot_to_networkx('./out/cat.dot'))
graphs = [g1, g2]
model = get_graph_embeddings(graphs, dimensions=dimensions)
save_embeddings('./out/ls.model', model, graphs, dimensions)
except Exception:
print(traceback.format_exc())
def test_get_node_embeddings():
G = convert_dot_to_networkx(test_file_path)
assert G is not None
wv = get_node_embedings(G)
assert type(wv) is KeyedVectors