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 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'] ))
def get_type_mapping_table(neo4j_db, func_name):
ast_root = get_function_ast_root(neo4j_db, func_name)
if ast_root is None:
print u"节点不存在"
return {'other':'v'}
else:
ser = serializedAST(neo4j_db)
ser.genSerilizedAST(ast_root)
return ser.variable_maps
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, db_table):
# @db1 待比对数据库
# @db2 代码段数据库
# @func_name 代码段构成的函数名
neo4j_db1 = Graph(db1)
neo4j_db2 = Graph(db2)
suffix_tree_obj = suffixtree()
#sqlite
db_conn = sqlite3.connect("/home/bert/Documents/data/soft_test.db")
db_conn.execute("""create table if not exists %s(
func_id INT PRIMARY KEY,
func_name CHAR(100) NOT NULL,
file CHAR(200) NOT NULL,
vuln_segement CHAR(100) NOT NULL,
distinct_type_and_const BOOLEAN,
distinct_const_no_type BOOLEAN,
distinct_type_no_const BOOLEAN,
no_type_no_const BOOLEAN)""" % db_table)
db_conn.commit()
target_func = get_function_ast_root(neo4j_db2, func_name)
if target_func is None:
print "%s is not found" % func_name
return
ret = serializedAST(neo4j_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 funcs:
print "[%s] processing %s VS %s" % (
datetime.datetime.now().strftime("%y-%m-%d %H:%M:%S"),
func[1], func_name)
ast_root = get_function_ast_root(neo4j_db1, func[0])
if ast_root is None:
print "function not found:", func[0], func[1]
tmp = serializedAST(neo4j_db1).genSerilizedAST(ast_root)
s1 = ";".join(tmp[0])
s2 = ";".join(tmp[1])
s3 = ";".join(tmp[2])
s4 = ";".join(tmp[3])
report = {}
try:
if suffix_tree_obj.search(s1, pattern1):
report['distinct_type_and_const'] = True
else:
report['distinct_type_and_const'] = False
if suffix_tree_obj.search(s2, pattern2):
report['distinct_const_no_type'] = True
else:
report['distinct_const_no_type'] = False
if suffix_tree_obj.search(s3, pattern3):
report['distinct_type_no_const'] = True
else:
report['distinct_type_no_const'] = False
if suffix_tree_obj.search(s4, pattern4):
report['no_type_no_const'] = True
else:
report['no_type_no_const'] = False
query = "insert into %s values(?,?,?,?,?,?,?,?)" % db_table
db_conn.execute(query, (func[0], func[1], func[2], func_name, report['distinct_type_and_const'],
report['distinct_const_no_type'],
report['distinct_type_no_const'],
report['no_type_no_const'])
)
db_conn.commit()
except Exception,e:
log_file = open("suffix_tree_error.log","a")
log_file.writelines(
[datetime.datetime.now().strftime("%y-%m-%d %H:%M:%S") + " " + e.__str__(),
s1, pattern1])
log_file.flush()
log_file.close()
def vuln_patch_compare(conn, neo4jdb, vuln_info, worksheet, suffix_tree_obj):
cve_info = vuln_info.get_cve_info(conn)
print "[%s] processing %s" % (datetime.datetime.now().strftime("%y-%m-%d %H:%M:%S"), cve_info.cveid)
vuln_name = cve_info.cveid.replace("-", "_").upper() + "_VULN_" + vuln_info.vuln_func
patch_name = cve_info.cveid.replace("-", "_").upper() + "_PATCHED_" + vuln_info.vuln_func
start_time = time.time()
status = ""
vuln_func = get_function_ast_root(neo4jdb, vuln_name)
if vuln_func is None:
status = "vuln_func_not_found"
line = process_line(conn, vuln_info, status, None, 0)
worksheet.append(line)
return
patched_func = get_function_ast_root(neo4jdb, patch_name)
if patched_func is None:
status = "patched_func_not_found"
line = process_line(conn, vuln_info, status, None, 0)
worksheet.append(line)
return
#序列化AST返回值是一个数组，0元素是序列化的AST字符串，1元素是节点个数，AST字符串以;结尾，需要去掉结尾的;
pattern1 = serializedAST(neo4jdb, True, True).genSerilizedAST(vuln_func)[0][:-1]
pattern2 = serializedAST(neo4jdb, False, True).genSerilizedAST(vuln_func)[0][:-1]
pattern3 = serializedAST(neo4jdb, True, False).genSerilizedAST(vuln_func)[0][:-1]
pattern4 = serializedAST(neo4jdb, False, False).genSerilizedAST(vuln_func)[0][:-1]
#delete FunctionDef and CompoundStatement node
prefix_str = r"^FunctionDef\([0-9]+\);CompoundStatement\([0-9]+\);"
pattern1 = re.sub(prefix_str, "", pattern1)
pattern2 = re.sub(prefix_str, "", pattern2)
pattern3 = re.sub(prefix_str, "", pattern3)
pattern4 = re.sub(prefix_str, "", pattern4)
s1 = serializedAST(neo4jdb, True, True).genSerilizedAST(patched_func)[0][:-1]
s2 = serializedAST(neo4jdb, False, True).genSerilizedAST(patched_func)[0][:-1]
s3 = serializedAST(neo4jdb, True, False).genSerilizedAST(patched_func)[0][:-1]
s4 = serializedAST(neo4jdb, False, False).genSerilizedAST(patched_func)[0][:-1]
report = {}
if suffix_tree_obj.search(s1, pattern1):
report['distinct_type_and_const'] = True
else:
report['distinct_type_and_const'] = False
if suffix_tree_obj.search(s2, pattern2):
report['distinct_const_no_type'] = True
else:
report['distinct_const_no_type'] = False
if suffix_tree_obj.search(s3, pattern3):
report['distinct_type_no_const'] = True
else:
report['distinct_type_no_const'] = False
if suffix_tree_obj.search(s4, pattern4):
report['no_type_no_const'] = True
else:
report['no_type_no_const'] = False
status = "success"
end_time = time.time()
cost = round(end_time - start_time, 2)
line = process_line(conn, vuln_info, status, report, cost)
worksheet.append(line)
return
def segement_ast_similarity_process(vuln_name, patch_name, neo4jdb,
org_func_name, type_mapping, worksheet,
suffix_tree_obj):
start_time = time.time()
print "[%s] processing %s" % (datetime.datetime.now().strftime(
"%y-%m-%d %H:%M:%S"), vuln_name + " vs " + patch_name)
#检查数据库里面是否可以找到该函数
vuln_func = get_function_ast_root(neo4jdb, vuln_name)
if vuln_func is None:
line = (vuln_name, patch_name, "vuln_func_not_found", "-", "-", "-",
"-", 0, org_func_name, type_mapping.__str__())
worksheet.append(line)
return
#检查数据库里面是否可以找到该函数
patched_func = get_function_ast_root(neo4jdb, patch_name)
if patched_func is None:
line = (vuln_name, patch_name, "patch_func_not_found", "-", "-", "-",
"-", 0, org_func_name, type_mapping.__str__())
worksheet.append(line)
return
#序列化AST返回值是一个数组，0元素是序列化的AST字符串，1元素是节点个数，AST字符串以;结尾，需要去掉结尾的;
o1 = serializedAST(neo4jdb)
o1.data_type_mapping = type_mapping
ret = o1.genSerilizedAST(vuln_func)
#delete FunctionDef and CompoundStatement node [2:]
pattern1 = ";".join(ret[0][2:])
pattern2 = ";".join(ret[1][2:])
pattern3 = ";".join(ret[2][2:])
pattern4 = ";".join(ret[3][2:])
tmp = o1.genSerilizedAST(patched_func)
s1 = ";".join(tmp[0])
s2 = ";".join(tmp[0])
s3 = ";".join(tmp[0])
s4 = ";".join(tmp[0])
report = {}
if suffix_tree_obj.search(s1, pattern1):
report['distinct_type_and_const'] = True
else:
report['distinct_type_and_const'] = False
if suffix_tree_obj.search(s2, pattern2):
report['distinct_const_no_type'] = True
else:
report['distinct_const_no_type'] = False
if suffix_tree_obj.search(s3, pattern3):
report['distinct_type_no_const'] = True
else:
report['distinct_type_no_const'] = False
if suffix_tree_obj.search(s4, pattern4):
report['distinct_type_no_const'] = True
else:
report['no_type_no_const'] = False
end_time = time.time()
cost = round(end_time - start_time, 2)
line = (vuln_name, patch_name, "success",
report['distinct_type_and_const'],
report['distinct_const_no_type'], report['distinct_type_no_const'],
report['distinct_type_no_const'], cost, org_func_name,
type_mapping.__str__())
worksheet.append(line)
def search_vuln_seg_in_func(db1, vuln_seg, vuln_func, var_map, db2, func_name, suffix_obj):
print "[%s] processing %s VS %s" % (
datetime.datetime.now().strftime("%y-%m-%d %H:%M:%S"),
vuln_seg, func_name)
vuln_seg_func = get_function_ast_root(db1, vuln_seg)
if vuln_seg_func is None:
vuln_seg_func = get_function_ast_root(db1, vuln_func)
if vuln_seg_func is None:
print "%s %s not found" % (vuln_seg, vuln_func)
return (vuln_seg+"-"+vuln_func, func_name, "vuln_not_found")
patched_func = get_function_ast_root(db2, func_name)
if patched_func is None:
print "%s is not found" % func_name
return (vuln_seg, func_name, "patch_not_found")
o1 = serializedAST(db1)
o1.variable_maps = var_map
ret = o1.genSerilizedAST(vuln_seg_func)
#delete FunctionDef and CompoundStatement node
pattern1 = ";".join(ret[0][2:])
pattern2 = ";".join(ret[1][2:])
pattern3 = ";".join(ret[2][2:])
pattern4 = ";".join(ret[3][2:])
pattern5 = ";".join(ret[4][2:])
tmp = serializedAST(db2).genSerilizedAST(patched_func)
s1 = ";".join(tmp[0][2:])
s2 = ";".join(tmp[1][2:])
s3 = ";".join(tmp[2][2:])
s4 = ";".join(tmp[3][2:])
s5 = ";".join(tmp[4][2:])
report = {}
if suffix_obj.search(s1, pattern1):
report['distinct_type_and_const'] = True
else:
report['distinct_type_and_const'] = False
if suffix_obj.search(s2, pattern2):
report['distinct_const_no_type'] = True
else:
report['distinct_const_no_type'] = False
if suffix_obj.search(s3, pattern3):
report['distinct_type_no_const'] = True
else:
report['distinct_type_no_const'] = False
if suffix_obj.search(s4, pattern4):
report['no_type_no_const'] = True
else:
report['no_type_no_const'] = False
if suffix_obj.search(s5, pattern5):
report['no_mapping'] = True
else:
report['no_mapping'] = False
#begin cfg
# patch_root = get_function_node_by_ast_root(db2, patched_func)
# vuln_seg_root = get_function_node_by_ast_root(db1, vuln_seg_func)
# match, simi = func_cfg_similarity(patch_root, db2, vuln_seg_root, db1)
return (vuln_seg, func_name, "success", report["distinct_type_and_const"],
report["distinct_const_no_type"], report["distinct_type_no_const"],
report["no_type_no_const"], report['no_mapping'])
def segement_ast_similarity_process(
vuln_name, patch_name, neo4jdb, org_func_name, type_mapping, worksheet, suffix_tree_obj
):
start_time = time.time()
print "[%s] processing %s" % (
datetime.datetime.now().strftime("%y-%m-%d %H:%M:%S"),
vuln_name + " vs " + patch_name,
)
# 检查数据库里面是否可以找到该函数
vuln_func = get_function_ast_root(neo4jdb, vuln_name)
if vuln_func is None:
line = (
vuln_name,
patch_name,
"vuln_func_not_found",
"-",
"-",
"-",
"-",
0,
org_func_name,
type_mapping.__str__(),
)
worksheet.append(line)
return
# 检查数据库里面是否可以找到该函数
patched_func = get_function_ast_root(neo4jdb, patch_name)
if patched_func is None:
line = (
vuln_name,
patch_name,
"patch_func_not_found",
"-",
"-",
"-",
"-",
0,
org_func_name,
type_mapping.__str__(),
)
worksheet.append(line)
return
# 序列化AST返回值是一个数组，0元素是序列化的AST字符串，1元素是节点个数，AST字符串以;结尾，需要去掉结尾的;
o1 = serializedAST(neo4jdb)
o1.data_type_mapping = type_mapping
ret = o1.genSerilizedAST(vuln_func)
# delete FunctionDef and CompoundStatement node [2:]
pattern1 = ";".join(ret[0][2:])
pattern2 = ";".join(ret[1][2:])
pattern3 = ";".join(ret[2][2:])
pattern4 = ";".join(ret[3][2:])
tmp = o1.genSerilizedAST(patched_func)
s1 = ";".join(tmp[0])
s2 = ";".join(tmp[0])
s3 = ";".join(tmp[0])
s4 = ";".join(tmp[0])
report = {}
if suffix_tree_obj.search(s1, pattern1):
report["distinct_type_and_const"] = True
else:
report["distinct_type_and_const"] = False
if suffix_tree_obj.search(s2, pattern2):
report["distinct_const_no_type"] = True
else:
report["distinct_const_no_type"] = False
if suffix_tree_obj.search(s3, pattern3):
report["distinct_type_no_const"] = True
else:
report["distinct_type_no_const"] = False
if suffix_tree_obj.search(s4, pattern4):
report["distinct_type_no_const"] = True
else:
report["no_type_no_const"] = False
end_time = time.time()
cost = round(end_time - start_time, 2)
line = (
vuln_name,
patch_name,
"success",
report["distinct_type_and_const"],
report["distinct_const_no_type"],
report["distinct_type_no_const"],
report["distinct_type_no_const"],
cost,
org_func_name,
type_mapping.__str__(),
)
worksheet.append(line)
def search_vuln_seg_in_func(db1, vuln_seg, vuln_func, var_map, db2, func_name,
suffix_obj):
print "[%s] processing %s VS %s" % (datetime.datetime.now().strftime(
"%y-%m-%d %H:%M:%S"), vuln_seg, func_name)
vuln_seg_func = get_function_ast_root(db1, vuln_seg)
if vuln_seg_func is None:
vuln_seg_func = get_function_ast_root(db1, vuln_func)
if vuln_seg_func is None:
print "%s %s not found" % (vuln_seg, vuln_func)
return (vuln_seg + "-" + vuln_func, func_name, "vuln_not_found")
patched_func = get_function_ast_root(db2, func_name)
if patched_func is None:
print "%s is not found" % func_name
return (vuln_seg, func_name, "patch_not_found")
o1 = serializedAST(db1)
o1.variable_maps = var_map
ret = o1.genSerilizedAST(vuln_seg_func)
#delete FunctionDef and CompoundStatement node
pattern1 = ";".join(ret[0][2:])
pattern2 = ";".join(ret[1][2:])
pattern3 = ";".join(ret[2][2:])
pattern4 = ";".join(ret[3][2:])
pattern5 = ";".join(ret[4][2:])
tmp = serializedAST(db2).genSerilizedAST(patched_func)
s1 = ";".join(tmp[0][2:])
s2 = ";".join(tmp[1][2:])
s3 = ";".join(tmp[2][2:])
s4 = ";".join(tmp[3][2:])
s5 = ";".join(tmp[4][2:])
report = {}
if suffix_obj.search(s1, pattern1):
report['distinct_type_and_const'] = True
else:
report['distinct_type_and_const'] = False
if suffix_obj.search(s2, pattern2):
report['distinct_const_no_type'] = True
else:
report['distinct_const_no_type'] = False
if suffix_obj.search(s3, pattern3):
report['distinct_type_no_const'] = True
else:
report['distinct_type_no_const'] = False
if suffix_obj.search(s4, pattern4):
report['no_type_no_const'] = True
else:
report['no_type_no_const'] = False
if suffix_obj.search(s5, pattern5):
report['no_mapping'] = True
else:
report['no_mapping'] = False
#begin cfg
# patch_root = get_function_node_by_ast_root(db2, patched_func)
# vuln_seg_root = get_function_node_by_ast_root(db1, vuln_seg_func)
# match, simi = func_cfg_similarity(patch_root, db2, vuln_seg_root, db1)
return (vuln_seg, func_name, "success", report["distinct_type_and_const"],
report["distinct_const_no_type"], report["distinct_type_no_const"],
report["no_type_no_const"], report['no_mapping'])