def match2Graph(G1,G2):
global a
# obtain min_edit_path and min_edit_distance
para2, para3 = nx.optimal_edit_paths(G1, G2)
maxOrder = 0
for i in range(0,len(G1.nodes)):
if G1.nodes[i]['order'] > maxOrder:
maxOrder = G1.nodes[i]['order']
# match 2 graph
for i in range(0,len(para2[0][0])):
if (para2[0][0][i][0] != None) and (para2[0][0][i][1] != None):
G2.nodes[para2[0][0][i][1]]['order'] = G1.nodes[para2[0][0][i][0]]['order']
elif para2[0][0][i][0] == None:
maxOrder += 1
G2.nodes[para2[0][0][i][1]]['order'] = maxOrder
else:
continue
for paraTemp in para2:
matchNumber = 0
for matchNodes in paraTemp[0]:
if matchNodes[0] != matchNodes[1]:
matchNumber += 1
print(para3)
def compute_path(tree1, tree2):
g1 = convert_nx(tree1)
g2 = convert_nx(tree2)
g, cost = nx.optimal_edit_paths(g1, g2, node_subst_cost=node_subst_cost, node_del_cost=node_del_cost, node_ins_cost=node_ins_cost, edge_subst_cost=edge_subst_cost, edge_del_cost=edge_del_cost, edge_ins_cost=edge_ins_cost)
path = g[0]
#print("edit distance between %s and %s" % (tree1.smiles, tree2.smiles))
#print_path(g1, g2, path)
return path, g, cost
def ged_paths(g1,g2,node_weight_mode='proportional'):
#need to incorporate edges attributes in order for ged edge costs to work correctly
'''for e, attr in g1.edges.items():
pdb.set_trace()
attr['nodes'] = '{}-{}'.format(e[0],e[1])
for e, attr in g2.edges.items():
attr['nodes'] = '{}-{}'.format(e[0],e[1])'''
def edge_subst_cost(gattr, hattr):
if (gattr['relation'] == hattr['relation']):# and (gattr['nodes'] == hattr['nodes']):
return 0
else:
return -1
def node_subst_cost_proportional(uattr, vattr):
cost = 0
attributes = list(uattr.keys())
unit_cost = 1/len(attributes)
for attr in attributes:
if(uattr[attr] != vattr[attr]): cost=cost+unit_cost
return cost
def node_subst_cost_atleastone(uattr, vattr):
cost = 0
attributes = list(uattr.keys())
for attr in attributes:
if(uattr[attr] != vattr[attr]):
cost=1
break
return cost
if node_weight_mode == 'proportional':
node_subst_cost = node_subst_cost_proportional
elif node_weight_mode == 'atleastone':
node_subst_cost = node_subst_cost_atleastone
else:
raise ValueError('Node weight mode {} not known'.format(node_weight_mode))
return nx.optimal_edit_paths(g1, g2,
edge_subst_cost = edge_subst_cost,
node_subst_cost = node_subst_cost)
g2 = loadGXL(args.g2)
# if you want to only output the distance, you can use the graph_edit_distance as follows:
distance = nx.graph_edit_distance(
g1,
g2,
node_subst_cost=GREC.node_substitution_cost,
node_del_cost=GREC.node_deletion_cost,
node_ins_cost=GREC.node_insertion_cost,
edge_subst_cost=GREC.edge_substitution_cost,
edge_del_cost=GREC.edge_deletion_cost,
edge_ins_cost=GREC.node_insertion_cost)
print("optimal distance :::", distance)
# if you want to output the distance and all the possible optimal pathz, you can use the optimal_edit_paths function
paths, cost = nx.optimal_edit_paths(
g1,
g2,
node_subst_cost=GREC.node_substitution_cost,
node_del_cost=GREC.node_deletion_cost,
node_ins_cost=GREC.node_insertion_cost,
edge_subst_cost=GREC.edge_substitution_cost,
edge_del_cost=GREC.edge_deletion_cost,
edge_ins_cost=GREC.node_insertion_cost)
print("the possible paths are: ", paths)
print("distance: ", distance)
def func2(codebox1: Text, codebox2: Text):
# codebox1 和 codebox2是窗口中的两个文本输入框
# 函数将会从文本框中读入代码
codeA = codebox1.get(0.0, END) # 从文本框读入代码
codeB = codebox2.get(0.0, END)
codeResourceA = cp.ProcessCode(codeA) # 创建一个文本处理的对象
codeResourceB = cp.ProcessCode(codeB)
codeResourceA.processComment() # 对文本处理对象去除注释信息
codeResourceB.processComment()
codeResourceA.processMacro() # 对文本处理对象替换宏定义
codeResourceB.processMacro()
codebox1.delete(0.0, END) # 清空文本框中的信息
codebox2.delete(0.0, END)
codebox1.insert(END, codeResourceA.code) # 用处理后的代码信息替代原来的代码
codebox2.insert(END, codeResourceB.code)
codeResourceA.collectFunctionInfo() # 收集代码中函数定义的信息
codeResourceB.collectFunctionInfo()
codeResourceA.functionCallInfo() # 获取代码中函数调用的信息
codeResourceB.functionCallInfo()
func_name_listA = codeResourceA.function_name_list # 获取代码中定义的所有函数名
func_name_listB = codeResourceB.function_name_list
call_matrixA = codeResourceA.call_matrix # 获取函数调用的矩阵
call_matrixB = codeResourceB.call_matrix
print(call_matrixA)
print(call_matrixB)
GraphA = nx.MultiDiGraph() # 建立一个空的有向多重图
GraphB = nx.MultiDiGraph()
GraphA.add_nodes_from(func_name_listA) # 从列表里添加节点
GraphB.add_nodes_from(func_name_listB)
for function_from in range(len(call_matrixA)):
for function_to in range(len(call_matrixA[function_from])):
edge_num = call_matrixA[function_from][function_to]
if edge_num != 0:
for count in range(edge_num):
GraphA.add_edge(
func_name_listA[function_from],
func_name_listA[function_to]) # 根据矩阵中的信息添加边
GraphA_info = len(GraphA.nodes) + len(GraphA.edges)
for function_from in range(len(call_matrixB)):
for function_to in range(len(call_matrixB[function_from])):
edge_num = call_matrixB[function_from][function_to]
if edge_num != 0:
for count in range(edge_num):
GraphB.add_edge(
func_name_listB[function_from],
func_name_listB[function_to]) # 根据矩阵中的信息添加边
nx.draw(GraphA, with_labels=True, font_weight='bold') # 绘制有向图
plt.show()
nx.draw(GraphB, with_labels=True, font_weight='bold')
plt.show()
nx.draw(GraphA, with_labels=True, font_weight='bold') # 绘制有向图
path, cost = nx.optimal_edit_paths(GraphA, GraphB)
sim_rate = (GraphA_info - cost) / GraphA_info
return sim_rate