def func_similarity_cfg_level(soft_db, funcs, character_db, func_name, worksheet):
# @db1 待比对数据库
# @db2 漏洞特征数据库
# @func_name 目标函数名
#过滤一下
ast_root = get_function_ast_root(character_db, func_name)
if ast_root is None:
print "no function found"
return
return_type = get_function_return_type(character_db, ast_root) # 获取目标函数返回值类型
param_list = get_function_param_list(character_db, ast_root) # 获取目标函数参数类型列表
filter_funcs = filter_functions(soft_db, funcs, return_type, param_list) # 过滤待比较函数
tar_func = get_function_node(character_db, func_name)
for ast_root in filter_funcs:
src_func = get_function_node_by_ast_root(soft_db, ast_root)
match, simi = func_cfg_similarity(src_func, soft_db, tar_func, character_db)
if match:
file = get_function_file(soft_db, src_func.properties[u'name'])[41:]
worksheet.append(
(func_name, file, src_func.properties[u'name'],match,
round(simi,4) ))
elif simi == -1:
print u"节点太多，未进行比较 "
def get_software_var_map(soft, port):
neo4j_db = Graph("http://127.0.0.1:%d/db/data/" % port)
sql_db = sqlite3.connect("/home/bert/Documents/data/" + soft + ".db")
sql_db.execute('''create table if not exists %s(
func_id INT PRIMARY KEY,
func_name CHAR(100) NOT NULL,
file CHAR(200) NOT NULL,
var_map TEXT NOT NULL,
ast_type_const TEXT NOT NULL,
ast_type_only TEXT NOT NULL,
ast_const_only TEXT NOT NULL,
ast_no_type_const TEXT NOT NULL,
no_mapping TEXT NOT NULL)''' % soft)
sql_db.commit()
funcs = get_all_functions(neo4j_db)
open("" + len(funcs).__str__(), "w")
print "get all functions OK:", len(funcs)
for func in funcs:
# 查重
ret = sql_db.execute("select * from %s where func_id=?" % soft,
(func._id, ))
if ret.fetchone():
continue
print "[%s] processing %s " % (datetime.datetime.now().strftime(
"%y-%m-%d %H:%M:%S"), func.properties[u'name'])
try:
ast_root = get_function_ast_root(neo4j_db, func)
func_file = get_function_file(neo4j_db, func)
ser = serializedAST(neo4j_db)
ret = ser.genSerilizedAST(ast_root)
var_map = ser.variable_maps
ast1 = ";".join(ret[0])
ast2 = ";".join(ret[1])
ast3 = ";".join(ret[2])
ast4 = ";".join(ret[3])
ast5 = ";".join(ret[4])
except Exception, e:
traceback.print_exc()
try:
sql_db.execute(
'insert into %s values(?, ?, ?, ?, ?, ?, ?, ?,?)' % soft,
(func._id, func.properties[u'name'], func_file,
var_map.__str__(), ast1, ast2, ast3, ast4, ast5))
sql_db.commit()
except Exception, e:
print e
def func_similarity_segement_level(db1, funcs, db2, func_name, worksheet):
# @db1 待比对数据库
# @db2 漏洞特征数据库
# @func_name 目标函数名
tar_func = get_function_node(db2, func_name)
for src_func in funcs:
match, simi = func_cfg_similarity(src_func, db1, tar_func, db2)
if match:
file = get_function_file(db1, src_func.properties[u'name'])[41:]
worksheet.append(
(func_name, file, src_func.properties[u'name'], match,
round(simi, 4)))
elif simi == -1:
print u"节点太多，未进行比较 "
def func_similarity_segement_level(db1, funcs, db2, func_name, worksheet):
# @db1 待比对数据库
# @db2 漏洞特征数据库
# @func_name 目标函数名
tar_func = get_function_node(db2, func_name)
for src_func in funcs:
match, simi = func_cfg_similarity(src_func, db1, tar_func, db2)
if match:
file = get_function_file(db1, src_func.properties[u'name'])[41:]
worksheet.append(
(func_name, file, src_func.properties[u'name'],match,
round(simi,4) ))
elif simi == -1:
print u"节点太多，未进行比较 "
def func_similarity_astLevel(db1, funcs, db2, func_name, suffix_tree_obj, worksheet):
# @db1 待比对数据库
# @db2 漏洞特征数据库
# @func_name 目标函数名
target_func = get_function_ast_root(db2, func_name)
return_type = get_function_return_type(db2, target_func) # 获取目标函数返回值类型
param_list = get_function_param_list(db2, target_func) # 获取目标函数参数类型列表
# funcs = getAllFuncs(db1) #获取所有函数
filter_funcs = filter_functions(db1, funcs, return_type, param_list) # 过滤待比较函数
ret = serializedAST(db2).genSerilizedAST(target_func)
pattern1 = ";".join(ret[0][2:])
pattern2 = ";".join(ret[1][2:])
pattern3 = ";".join(ret[2][2:])
pattern4 = ";".join(ret[3][2:])
for func in filter_funcs:
ast_root = get_function_ast_root(db1, func.properties[u'name'])
s1 = serializedAST(db1, True, True).genSerilizedAST(ast_root)[0][:-1]
s2 = serializedAST(db1, False, True).genSerilizedAST(ast_root)[0][:-1]
s3 = serializedAST(db1, True, False).genSerilizedAST(ast_root)[0][:-1]
s4 = serializedAST(db1, False, False).genSerilizedAST(ast_root)[0][:-1]
report = {}
if suffix_tree_obj.search(s1, pattern1):
report['distinct_type_and_const'] = True
if suffix_tree_obj.search(s2, pattern2):
report['distinct_const_no_type'] = True
if suffix_tree_obj.search(s3, pattern3):
report['distinct_type_no_const'] = True
if suffix_tree_obj.search(s4, pattern4):
report['distinct_type_no_const'] = True
if report['distinct_type_and_const'] or report['distinct_const_no_type']\
or report['distinct_type_no_const'] or report['no_type_no_const']:
file = get_function_file(db1, func.properties[u'name'])
worksheet.append(
(func_name, file, func.properties[u'name'],report['distinct_type_and_const'],
report['distinct_const_no_type'], report['distinct_type_no_const'],
report['distinct_type_no_const'] ))