Python crawler: Analog 12306 Login

Tried for a long time to log on always show: System busy, please refresh,,, too tortured people, engaged for a half a day to think is the question of the request head ....

Verification code or to manually identify the diagram.

#!/bin/env python# -*- coding=utf-8 -*-import sslimport sysimport  urllib2import randomimport httplibimport jsonfrom cookielib import  Lwpcookiejarimport urllibimport reimport getpassreload (SYS) sys.setdefaultencoding (' UTF8 ') Cookiejar = lwpcookiejar () cookiesuppor = urllib2. Httpcookieprocessor (Cookiejar) Opener = urllib2.build_opener (cookiesuppor, urllib2. HttpHandler) Urllib2.install_opener (opener) ssl._create_default_https_context = ssl._create_unverified _contextcodeimg =  ' Https://kyfw.12306.cn/otn/passcodeNew/getPassCodeNew?module=login&rand=sjrand &%s '  % random.random () baner =  "" "##################################     12306 Login Script, author Mr rjl    python version: 2.7, for linux     verification Code Input Method:      input problem corresponds to the picture ordinal,1-8;     multiple with ', ' delimited. For example: 1,2,3################################## "" "Def get (URL):    try:         request = urllib2. Request (Url=url)         # req.add_header (' User-Agent ',  ' mozilla/5.0  (x11; fedora; linux x86_64; rv:52.0)  gecko/20100101 firefox/ 52.0 ')         request.add_header ("Content-type",  "application/ X-www-form-urlencoded; charset=utf-8 ")         request.add_header ( ' X-requested-with ',  ' XMLHttpRequest ')         request.add_header (' User-agent ',                             ' mozilla/5.0  (windows nt 6.1)  AppleWebKit/537.36  (Khtml, like gecko)  chrome/33.0.1750.154 safari/537.36 ')         request.add_header (' Referer ',  ' https:// Kyfw.12306.cn/otn/login/init ')         request.add_header (' Accept ',   ' */* ')         result = urllib2.urlopen (Request). Read ()         assert isinstance (Result, object)          return result    except httplib.error as  e:        print e         pass    except urllib2. urlerror as e:        print e         pass    except urllib2. Httpbasicauthhandler, urllib2. httperror:        print  ' ERROR '          passdef post (Url, data):    try:         request = urllib2. Request (Url=url, data=urllib.urlencode (data))         #  Req.add_header (' user-agent ',  ' mozilla/5.0  (x11; fedora; linux x86_64; rv:52.0)  gecko/20100101 firefox/52.0 ')         # request =  urllib2. Request (Ajax_url, urllib.urlencode (DC))         request.add_header ("Content-type",  "Application/x-www-form-urlencoded; charset=utf-8")          request.add_header (' X-requested-with ',  ' XMLHttpRequest ')          request.add_header (' user-agent ',                             ' mozilla/5.0  (windows nt 6.1)  AppleWebKit/ 537.36  (Khtml, like gecko)  chrome/33.0.1750.154 safari/537.36 ')          request.add_header (' Referer ',  ' https://kyfw.12306.cn/otn/login/init ')          request.add_header (' Accept ',  ' */* ')          # request.add_header (' accept-encoding ',  ' gzip, deflate ')          result = urllib2.urlopen (Request). Read ()          return result    except httplib.error as e:         return e    except urllib2. urlerror as e:        return e     Except urllib2. HttpbasicauthhAndler, urllib2. httperror:        return  ' ERROR ' DEF&NBSP;COOKIETP ():     stoidinput ("Get Cookie")     Url =  "https://kyfw.12306.cn/otn/login/ Init "    get (URL)     for index, c in enumerate ( Cookiejar):         stoidinput (c) def getimg ():     stoidinput ("Download Verification Code ...")     result = get (codeimg)     try :         if open ('/tmp/tkcode ',  ' WB '). Write (Result)  :             import os             os.system ("oeg /tmp/tkcode &")          else:            return false    except oserror as e:         print e        passdef stoidinput (text):      ""      normal information output     :p aram text:      :return:      ""     print  "\033[34m[*]\033[0m"  %s  " % textdef errorinput (text):    " "" "     Error message output     :p aram text:     :return:       ""     print  "\033[32m[!" \033[0m %s  " % text    return falsedef codexy ():      "" "     Get Verification Code     :return: str    " ""         ofset = raw_Input ("[*]  Please enter Verification code: ")     select = ofset.split (', ')      global randcode    post = []    offsetsx =  0  #  Select the left value of the answer by clicking on the midpoint of the 8 small graphs in the browser to get it, so basically no problem     offsetsy = 0   #  top value of the selected answer     for ofset in select:         if ofset ==  ' 1 ':             offsetsY = 46             offsetsx = 42        elif ofset ==   ' 2 ':             offsetsy = 46             offsetsX = 105         elif ofset ==  ' 3 ':             offsetsy = 45            offsetsx =  184        elif ofset ==  ' 4 ':             offsetsY = 48             offsetsX = 256         elif ofset ==  ' 5 ':             offsetsy = 36            offsetsx  = 117        elif ofset ==  ' 6 ':             offsetsY = 112              offsetsx = 115        elif ofset  ==  ' 7 ':             offsetsy =  114            offsetsx = 181         elif ofset ==  ' 8 ':             offsetsY = 111             offsetsX = 252         else:            pass         post.append (OFFSETSX)         post.append ( Offsetsy)     randcode = str (POST). Replace ('] ',  '). Replace (' [',  '). Replace ("'" ", "). Replace ('   ', ") Def login (user, passwd):    randurl =  ' https://kyfw.12306.cn/otn/ Passcodenew/checkrandcodeansyn '     logurl =  ' https://kyfw.12306.cn/otn/login/ Loginaysnsuggest '     surl =  ' https://kyfw.12306.cn/otn/login/userLogin '      geturl =  ' https://kyfw.12306.cn/otn/index/initMy12306 '      randdata = {         "Randcode": randCode,          "Rand":  "Sjrand"     }     logdata = {         "Loginuserdto.user_name":  user,          "Userdto.password": passwd,          "Randcode": randcode    }    ldata =  {&NBSP;&NBSP;&NBSP;&NBSP;&NBSP;&Nbsp;   "_json_att": none    }    fresult =  Json.loads (Post (randurl, randdata),  encoding= ' UTF8 ')     checkcode =  fresult[' data ' [' msg ']    if checkcode ==  ' FALSE ':         errorinput ("Verification code is wrong, please try again")     else:         stoidinput ("Verification code passed, start login ...")         try:             tresult = json.loads (Post (logurl, logdata),  encoding= ' UTF8 ')              if tresult[' data '].__len__ ()  == 0:                 errorinput (" Login failed:  %s " % tresult[' Messages '][0])             else:                 stoidinput ("Login Successful")                  sult = post (Surl, ldata)                  getuserinfo ()          except ValueError as e:             errorinput (e) def getuserinfo ():     "" "     login successful, display user name     :return:     ""     url =  ' https:// Kyfw.12306.cn/otn/modifyuser/initqueryuserinfo '     data = dict (_json_att=None)     result = post (Url, data)     userinfo =  Result   &nbsP;name = r ' <input name= "UserDTO.loginUserDTO.user_name"  style= "Display:none;"  type= "Text"  value= "(\s+)"  /> "    try:         stoidinput ("Welcome  %s  Login"  % re.search (name, result). Group (1))      except attributeerror:        passdef main ():     user = raw_input ("[+]  User name (username/email/mobile): ")      passwd = getpass.getpass ("[+]  password: ")     if user ==   '  or passwd ==  ':         errorinput (" User name or password cannot be empty! ") &NBSP;&NBSP;&NBSP;&NBSP;ELSE:&NBSP;&NBSP;&NBSP;&NBSP;&NBSP;&NBSP;&NBSP;&NBSP;COOKIETP ()          getimg ()         codexy ()          loGin (USER,&NBSP;PASSWD) def logout ():    url =  ' https://kyfw.12306.cn/otn/ Login/loginout '     result = get (URL)     if result:         stoidinput ("exited")     else:         errorinput ("Exit failed") if __name__ ==  "__main__":     print baner    main ()     logout ()

This article from "Lonely No Worry" blog, please be sure to keep this source http://jlnetc.blog.51cto.com/10920149/1907446

Python crawler: Analog 12306 Login