一、一些用到的技术
1.1解析验证码（联众-收费，两分钱一个）
验证码识别设计机器学习，没有耗费时间去实现，所以采用了第三方（联众）的在线打码，注册充值便可使用，识别率挺高的，验证码识别种类不少，充了五十块玩了好久还剩不少。具体地址http://v1-http-api.jsdama.com/api.php?mod=php
1.2 具体实现代码,此部分负责登录的部分逻辑（__init__.py）
# -*- coding: utf-8 -*- import sys sys.path.append("..") import requests, json from common.base import Base from requests.packages.urllib3.exceptions import InsecureRequestWarning # 禁用安全请求警告 requests.packages.urllib3.disable_warnings(InsecureRequestWarning) __author__ = 'Rachel feng' PATH = 'https://kyfw.12306.cn' __all__ = ['LoginTic'] class LoginTic(Base): def __init__(self): super(LoginTic, self).__init__() # 验证结果 def checkYanZheng(self,solution): # # 分割用户输入的验证码位置 # 225,83|181,31|35,67 => 225,83,181,31,35,67 yanStr = solution.replace('|',',') print('校验验证码......') checkUrl = "https://kyfw.12306.cn/passport/captcha/captcha-check" data = { 'login_site':'E', #固定的 'rand':'sjrand', #固定的 'answer':yanStr #验证码对应的坐标，两个为一组，跟选择顺序有关,有几个正确的，输入几个 } print(data) # 发送验证 cont = self.requests.post(checkUrl, data=data, verify=False) # 返回json格式的字符串，用json模块解析 dic = cont.json() code = dic['result_code'] # 取出验证结果，4：成功 5：验证失败 7：过期 if str(code) == '4': return True else: print(dic) return False # 发送登录请求的方法 def loginTo(self, userName, pwd): yan = self.decode_captcha(url) check = False # #只有验证成功后才能执行登录操作 # while not check: check = self.checkYanZheng(yan) if not check: return 600, '验证失败，请重新验证' loginUrl = "https://kyfw.12306.cn/passport/web/login" data = { 'username':userName, 'password':pwd, 'appid':'otn' } print(data) headers = { 'Origin': 'https://kyfw.12306.cn', 'User-Agent': 'Mozilla/5.0 (Windows NT 6.1; Win64; x64) AppleWebKit/537.36 (KHTML, like Gecko) Chrome/59.0.3071.86 Safari/537.36', 'Referer': 'https://kyfw.12306.cn/otn/login/init', 'Content-Type': 'application/x-www-form-urlencoded; charset=UTF-8', 'Accept': 'application/json, text/javascript, */*; q=0.01' } result = self.requests.post(loginUrl,data=data,headers=headers) dic = result.json() print(dic) mes = dic['result_message'] # 结果的编码方式是Unicode编码，所以对比的时候字符串前面加u,或者mes.encode('utf-8') == '登录成功'进行判断，否则报错 if mes == u'登录成功': # 获取cookie r = self.requests.post('https://kyfw.12306.cn/otn/login/userLogin') data = {'appid': 'otn'} r = self.requests.post('https://kyfw.12306.cn/passport/web/auth/uamtk', data=data) if r.status_code != 200: return 600, '登录验证不通过' d = r.json() if d.get('result_code') == 0: data = { 'tk': d.get('newapptk') } r = self.requests.post('https://kyfw.12306.cn/otn/uamauthclient', data=data) d = r.json() print(d) if d.get('result_code') == 0: print('恭喜你，登录成功，可以购票!') return True else: return 600, d.get('result') else: print(dic.get('result_message')) return 600, dic.get('result_message') def get_info(self): data = {'_json_att': ''} headers = { 'Referer': 'https://kyfw.12306.cn/otn/index/initMy12306', 'Origin': 'https://kyfw.12306.cn', 'User-Agent': 'Mozilla/5.0 (Windows NT 6.1; Win64; x64) AppleWebKit/537.36 (KHTML, like Gecko) Chrome/59.0.3071.86 Safari/537.36', 'Content-Type': 'application/x-www-form-urlencoded' } r = self.requests.post('https://kyfw.12306.cn/otn/modifyUser/initQueryUserInfo', data=data, headers=headers) if r.status_code != 200: print('获取购票人信息异常') with open('__tmp/userInfo.html', 'wb') as f: f.write(r.content) f.close() print('获取信息成功') if __name__ == '__main__': userName = '’ # 12306登录账号 pwd = '' #12306登录密码 url = "https://kyfw.12306.cn/passport/captcha/captcha-image?login_site=E&module=login&rand=sjrand"; login = LoginTic() d = login.loginTo(userName, pwd) if type(d) != tuple: login.get_info() else: print(d)
1.2base.py(在comment目录底下共用)
# -*- coding: utf-8 -*- import importlib, logging, codecs, json, datetime, time import web, requests from pyquery import PyQuery as pq from jinja2 import Template __all__ = ["cache_session"] # API_PATH = 'http://bbb4.hyslt.com/api.php?mod=php' API_PATH = 'http://v1-http-api.jsdama.com/api.php?mod=php' USERNAME = '' #联众账号 PASSWORD = '' #联众密码 TOKEN = '' class Base(object): """docstring for base""" def __init__(self): self.session = requests.session() net = requests.Session() net.timeout = 30 net.headers.update({ "User-Agent":"Mozilla/5.0 (Macintosh; Intel Mac OS X 10_12_3) AppleWebKit/537.36 (KHTML, like Gecko) Chrome/59.0.3071.115 Safari/537.36" }) retries = requests.packages.urllib3.util.retry.Retry(total=3, backoff_factor=0.1, status_forcelist=[500, 502, 503, 504]) adapter = requests.adapters.HTTPAdapter(max_retries=retries) net.mount('http://', adapter) net.mount('https://', adapter) self.requests = net # 识别 def decode(self, img, codetype=1303, minlength=1, maxlength=8): """ img: 验证码图片 codetype: 验证码类型（https://www.jsdati.com/price） 1001: 四个字母,数字 1201: 计算题 1008: 纯字母验证码 1009: 纯数字验证码 1013: 5位字母加数字(5位纯字母） 1014: 6位字母加数字（6位纯字母） 1015: 7位字母加数字（7位纯字母） 1313: 坐标点击3次 1314: 坐标点击4次 minlength: 验证码最小长度 maxlength: 验证码最大长度 """ data = { 'user_name': USERNAME, 'user_pw': PASSWORD, 'zztool_token': TOKEN, 'yzmtype_mark': codetype, 'yzm_minlen': minlength, 'yzm_maxlen': maxlength } files = { 'upload': img } try: r = self.requests.post(API_PATH, params={'act':'upload'}, data=data, files=files, timeout=20) except: return 504, '连接超时' d = r.json() return d # 获取验证码图片 def decode_captcha(self, url): print('get code....') response = self.requests.get(url, verify=False) # 把验证码图片保存到本地 with open('vcode.png','wb') as f: f.write(response.content) # 用pillow模块打开并解析验证码,这里是假的，自动解析以后学会了再实现 try: content = response.content except: content = None print('获取验证码失败') # 识别 print('decode...') d = self.decode(content) # 识别成功 if type(d) is dict and d['result']: code = d['data']['val'] print(code) # 保存验证码图片 return code elif type(d) is str: return d
1.3一些乱七八糟补充的：
项目目录
comment
---__init__.py
---base.py
test
---__init__.py
验证码识别有时候识别不准确，导致校验有时候会不通过，得检查一下请求头什么的。
2.运行结果
