def is_controllable(expr, flag=None): # 获取表达式中的变量，看是否在用户可控变量列表中
"""
判断赋值表达式是否是用户可控的
:param expr:
:return:
"""
controlled_params = [
'$_GET',
'$_POST',
'$_REQUEST',
'$_COOKIE',
'$_FILES',
'$_SERVER',
'$HTTP_POST_FILES',
'$HTTP_COOKIE_VARS',
'$HTTP_REQUEST_VARS',
'$HTTP_POST_VARS',
'$HTTP_RAW_POST_DATA',
'$HTTP_GET_VARS'
]
if expr in controlled_params: # 当为可控变量时 返回1
logger.debug('[AST] is_controllable --> {expr}'.format(expr=expr))
if flag:
return 1, expr
return 1, php.Variable(expr)
if expr.startswith("$"):
if flag:
return 3, expr
return 3, php.Variable(expr)
return -1, php.Variable(expr)
def array_back(param, nodes): # 回溯数组定义赋值
"""
递归回溯数组赋值定义
:param param:
:param nodes:
:return:
"""
param_name = param.node.name
param_expr = param.expr
is_co = 3
cp = param
expr_lineno = 0
# print nodes
for node in nodes[::-1]:
if isinstance(node, php.Assignment):
param_node_name = get_node_name(node.node)
param_node = node.node
param_node_expr = node.expr
if param_node_name == param_name: # 处理数组中值被改变的问题
if isinstance(node.expr, php.Array):
for p_node in node.expr.nodes:
if p_node.key == param_expr:
if isinstance(p_node.value, php.ArrayOffset): # 如果赋值值仍然是数组，先经过判断在进入递归
is_co, cp = is_controllable(p_node.value.node.name)
if is_co != 1:
is_co, cp, expr_lineno = array_back(param, nodes)
else:
n_node = php.Variable(p_node.value)
is_co, cp, expr_lineno = parameters_back(n_node, nodes)
if param == param_node: # 处理数组一次性赋值，左值为数组
if isinstance(param_node_expr, php.ArrayOffset): # 如果赋值值仍然是数组，先经过判断在进入递归
is_co, cp = is_controllable(param_node_expr.node.name)
if is_co != 1:
is_co, cp, expr_lineno = array_back(param, nodes)
else:
is_co, cp = is_controllable(param_node_expr)
print is_co
if is_co != 1 and is_co != -1:
n_node = php.Variable(param_node_expr.node.value)
is_co, cp, expr_lineno = parameters_back(n_node, nodes)
return is_co, cp, expr_lineno
def deep_parameters_back(node, back_node, function_params, count, file_path):
"""
深度递归遍历
:param node:
:param back_node:
:param function_params:
:param file_path:
:return:
"""
count += 1
params = get_node_name(node)
is_co, cp, expr_lineno = parameters_back(params, back_node,
function_params)
if count > 20:
logger.warning("[Deep AST] depth too big to auto exit...")
return is_co, cp, expr_lineno
if is_co == 3:
logger.debug("[Deep AST] try to find include, start deep AST")
for node in back_node[::-1]:
if isinstance(node, php.Include):
filename = node.expr
file_path = re.split(r"[\/\\]", file_path)
file_path.pop()
file_path.append(filename)
file_path = "/".join(file_path)
try:
logger.debug("[Deep AST] open new file {file_path}".format(
file_path=file_path))
f = open(file_path, 'r')
file_content = f.read()
except:
logger.warning(
"[Deep AST] error to open new file...continue")
continue
parser = make_parser()
all_nodes = parser.parse(file_content,
debug=False,
lexer=lexer.clone(),
tracking=with_line)
node = php.Variable(cp)
is_co, cp, expr_lineno = deep_parameters_back(
node, all_nodes, function_params, count, file_path)
if is_co == -1:
break
return is_co, cp, expr_lineno
def analysis_functioncall_node(node, back_node, vul_function, vul_lineno, function_params=None, file_path=None):
"""
处理FunctionCall类型节点-->取出参数-->回溯判断参数是否可控-->输出结果
:param node:
:param back_node:
:param vul_function:
:param vul_lineno:
:param function_params:
:return:
"""
logger.debug('[AST] vul_function:{v}'.format(v=vul_function))
params = get_all_params(node.params)
for param in params:
param = php.Variable(param)
is_co, cp, expr_lineno = parameters_back(param, back_node, function_params)
set_scan_results(is_co, cp, expr_lineno, vul_function, param, vul_lineno)
def anlysis_params(param, code_content, file_path, lineno):
"""
在cast调用时做中转数据预处理
:param lineno:
:param param:
:param code_content:
:param file_path:
:return:
"""
count = 0
function_params = None
param = php.Variable(param)
parser = make_parser()
all_nodes = parser.parse(code_content, debug=False, lexer=lexer.clone(), tracking=with_line)
vul_nodes = []
for node in all_nodes:
if node.lineno < int(lineno):
vul_nodes.append(node)
is_co, cp, expr_lineno = deep_parameters_back(param, vul_nodes, function_params, count, file_path, lineno)
return is_co, cp, expr_lineno
import os
# for django
os.environ.setdefault('DJANGO_SETTINGS_MODULE', 'Kunlun_M.settings')
import django
django.setup()
from core.engine import scan
from core.engine import init_match_rule
from Kunlun_M.settings import EXAMPLES_PATH
from utils.log import logger
from phply import phpast as php
def test_scan():
logger.info('Examples Path: {path}'.format(path=EXAMPLES_PATH))
assert scan(EXAMPLES_PATH)
data = (php.Method(u'eval_function', [], [php.FormalParameter(u'$a', None, False, None)], [php.Eval(php.Variable(u'$a'))], False), php.Variable(u'$a'), "eval_function")
def test_init_match_rule():
assert isinstance(init_match_rule(data), tuple)
assert "eval_function" in init_match_rule(data)[1]
from cobra.engine import scan
from cobra.engine import init_match_rule
from cobra.config import examples_path
from cobra.log import logger
from phply import phpast as php
def test_scan():
logger.info('Examples Path: {path}'.format(path=examples_path))
assert scan(examples_path)
data = (php.Method(u'eval_function', [],
[php.FormalParameter(u'$a', None, False, None)],
[php.Eval(php.Variable(u'$a'))],
False), php.Variable(u'$a'))
def test_init_match_rule():
assert isinstance(init_match_rule(data), tuple)
assert "eval_function" in init_match_rule(data)[1]
def parameters_back(param, nodes, function_params=None, lineno=0, function_flag=0): # 用来得到回溯过程中的被赋值的变量是否与敏感函数变量相等,param是当前需要跟踪的污点
"""
递归回溯敏感函数的赋值流程，param为跟踪的污点，当找到param来源时-->分析复制表达式-->获取新污点；否则递归下一个节点
:param param:
:param nodes:
:param function_params:
:param lineno
:param flineno: 已经分析过的行数，用于解决已经分析完成函数内容问题
:return:
"""
if isinstance(param, php.FunctionCall): # 当污点为寻找函数时，递归进入寻找函数
is_co, cp, expr_lineno = function_back(param, nodes, function_params)
return is_co, cp, expr_lineno
if isinstance(param, php.ArrayOffset): # 当污点为数组时，递归进入寻找数组声明或赋值
is_co, cp, expr_lineno = array_back(param, nodes)
return is_co, cp, expr_lineno
expr_lineno = 0 # source所在行号
param_name = param.name
is_co, cp = is_controllable(param_name)
if len(nodes) != 0 and is_co != 1:
node = nodes[len(nodes) - 1]
if isinstance(node, php.Assignment): # 回溯的过程中，对出现赋值情况的节点进行跟踪
param_node = get_node_name(node.node) # param_node为被赋值的变量
param_expr, expr_lineno, is_re = get_expr_name(node.expr) # param_expr为赋值表达式,param_expr为变量或者列表
if param_name == param_node and not isinstance(param_expr, list): # 找到变量的来源，开始继续分析变量的赋值表达式是否可控
is_co, cp = is_controllable(param_expr) # 开始判断变量是否可控
if is_co != 1:
is_co, cp = is_sink_function(param_expr, function_params)
if isinstance(node.expr, php.ArrayOffset):
param = node.expr
else:
param = php.Variable(param_expr) # 每次找到一个污点的来源时，开始跟踪新污点，覆盖旧污点
if param_name == param_node and isinstance(node.expr, php.FunctionCall): # 当变量来源是函数时，处理函数内容
function_name = node.expr.name
param = node.expr # 如果没找到函数定义，则将函数作为变量回溯
for node in nodes[::-1]:
if isinstance(node, php.Function):
if node.name == function_name:
function_nodes = node.nodes
# 进入递归函数内语句
for function_node in function_nodes:
if isinstance(function_node, php.Return):
return_node = function_node.node
return_param = return_node.node
is_co, cp, expr_lineno = parameters_back(return_param, function_nodes, function_params, lineno, function_flag=1)
if param_name == param_node and isinstance(param_expr, list):
for expr in param_expr:
param = expr
is_co, cp = is_controllable(expr)
if is_co == 1:
return is_co, cp, expr_lineno
param = php.Variable(param)
_is_co, _cp, expr_lineno = parameters_back(param, nodes[:-1], function_params, lineno, function_flag=1)
if _is_co != -1: # 当参数可控时，值赋给is_co 和 cp，有一个参数可控，则认定这个函数可能可控
is_co = _is_co
cp = _cp
elif isinstance(node, php.Function) and function_flag == 0:
function_nodes = node.nodes
function_lineno = node.lineno
function_params = node.params
vul_nodes = []
for function_node in function_nodes:
if int(function_lineno) <= function_node.lineno < int(lineno):
vul_nodes.append(function_node)
if len(vul_nodes) > 0:
is_co, cp, expr_lineno = parameters_back(param, function_nodes, function_params, function_lineno, function_flag=1)
if is_co == 3: # 出现新的敏感函数，重新生成新的漏洞结构，进入新的遍历结构
logger.info("[Deep AST] Now vulnerability param become to function {}() param {}".format(node.name, cp.name))
is_co = 4
cp = tuple([node, param])
return is_co, cp, 0
if is_co != 1 and is_co != -1: # 当is_co为True时找到可控，停止递归
is_co, cp, expr_lineno = parameters_back(param, nodes[:-1], function_params, lineno, function_flag=1) # 找到可控的输入时，停止递归
elif len(nodes) == 0 and function_params is not None: # 考虑函数参数情况
for function_param in function_params:
if function_param == param:
is_co = 2
cp = function_param
return is_co, cp, expr_lineno