python3.7 爬取QQ空间好友

  1 from urllib import parse
  2 from selenium import webdriver
  3 import requests
  4 import json
  5 from json import loads
  6 import time
  7 import pymssql
  8 import datetime
  9 
 10 def get_key_values(body,key,end =‘;‘):
 11     """提取body中不包括的key，分片操作
 12 
 13     :param body: 父字符串
 14     :param key: 子字符串
 15     :param end: 结束字符串，默认为;
 16     :return: 不包括子字符串的字符串
 17     """
 18     return body[body.find(key) + len(key): body.find(‘;‘, body.find(key))]
 19 
 20 def get_key(cookies):
 21     """获取cookie中的相关键的值
 22     解密
 23 
 24     :param cookies: 缓存
 25     :return: 相关键的值
 26     """
 27     key = get_key_values(cookies,‘p_skey=‘)
 28     h = 5381
 29     for i in key:
 30         h += (h << 5) + ord(i)
 31     return h & 2147483647
 32 
 33 def web_login_cookie():
 34     """url = ‘https://user.qzone.qq.com/QQ号相关的缓存
 35     实现自动化登录
 36 
 37     :return: 浏览器的缓存
 38     """
 39     driver = webdriver.Chrome()
 40     qq_account = ‘1612893772‘
 41     qq_password = ‘13974162858x‘
 42     login(driver,qq_account,qq_password)
 43     time.sleep(10)
 44     driver.get(‘https://user.qzone.qq.com/{}‘.format(qq_account))
 45     cookie = ‘‘
 46     for elem in driver.get_cookies(): # 记录登录的Cookie
 47         # elem 为 dict类型
 48         cookie += elem["name"] + "=" + elem["value"] + ";"
 49     # cookies = cookie
 50     return cookie
 51 
 52 def login(driver,qq_account,qq_password):
 53     """登录
 54 
 55     :param driver: 浏览器对象
 56     :param qq_account: QQ账号
 57     :param qq_password: QQ密码
 58     :return:
 59     """
 60     driver.maximize_window()
 61     driver.get(‘http://user.qzone.qq.com‘)
 62     driver.switch_to.frame(‘login_frame‘)
 63     time.sleep(1)
 64     driver.find_element_by_id("switcher_plogin").click()
 65     driver.find_element_by_id("u").send_keys(qq_account)
 66     time.sleep(2)
 67     driver.find_element_by_id("p").send_keys(qq_password)
 68     time.sleep(2)
 69     driver.find_element_by_id("login_button").click()
 70 
 71 def send_requests(req,headers,url,params=None):
 72     """url_friend = ‘https://user.qzone.qq.com/proxy/domain/r.qzone.qq.com/cgi-bin/tfriend/friend_ship_manager.cgi?‘
 73     url_friend携带以下参数：uin（QQ号）、do（没有它，返回空，默认值为：1）
 74     rd，g_t，qzonetoken（每次登录都发生变化，从Cookiezz中获取）
 75     fupdate，clean（默认值为：1）
 76 
 77     :param req: 请求（Request）,该请求为会话
 78     :param headers: 请求头
 79     :param params: 请求参数
 80     :return: JSONP数据
 81     """
 82     if None != params:
 83         url = url + parse.urlencode(params)
 84     # url = url+‘&offset=‘+str(0)
 85     page = req.get(url=url, headers=headers)
 86     return page.text
 87 
 88 def get_each_str(req,uin,headers):
 89     each_url = ‘https://user.qzone.qq.com/{}‘.format(uin)
 90     page = req.get(url=each_url, headers=headers)
 91 
 92 def friend_db(dicts,name=‘‘):
 93     """操作DB
 94 
 95     :param dicts: 数据字典信息
 96     :param name: 备注名
 97     :return: void
 98     """
 99     if len(str(dicts[‘birthyear‘])) < 4:
100         dicts[‘birthyear‘] = ‘1900‘
101     if dicts[‘birthday‘][1:2] == ‘0‘:
102         dicts[‘birthday‘] = ‘01-01‘
103     if len(dicts[‘signature‘]) > 70:
104         dicts[‘signature‘] = ‘‘
105     friend_db_dict = {
106         ‘friendInfo‘: [
107         dicts[‘uin‘], name, dicts[‘age‘], ‘男‘ if dicts[‘sex‘] == 1 else ‘女‘
108         , datetime.datetime.strptime(str(dicts[‘birthyear‘]) + ‘-‘ + str(dicts[‘birthday‘]), ‘%Y-%m-%d‘)],
109         ‘friendPlace‘: [
110         dicts[‘uin‘], dicts[‘company‘],dicts[‘career‘], dicts[‘hco‘] + dicts[‘hp‘] + dicts[‘hc‘],
111         dicts[‘country‘] + dicts[‘province‘] + dicts[‘city‘],dicts[‘cco‘] + dicts[‘cp‘] + dicts[‘cc‘], dicts[‘cb‘]],
112         ‘friendNet‘: [
113         dicts[‘uin‘], dicts[‘nickname‘], dicts[‘spacename‘], dicts[‘desc‘], dicts[‘signature‘]]
114     }
115     conn = pymssql.connect(host=‘localhost‘, user=‘sa‘, password=‘123456‘, database=‘friendDB‘,
116                             charset=‘utf8‘)
117     cur = conn.cursor()
118     sql = "begin tran insertData insert into friendInfo values({},‘{}‘,{},‘{}‘,‘{}‘);" 119           "insert into friendPlace values({},‘{}‘,‘{}‘,‘{}‘,‘{}‘,‘{}‘,‘{}‘);" 120           "insert into friendNet values({},‘{}‘,‘{}‘,‘{}‘,‘{}‘);" 121           "commit tran insertData".122         format(friend_db_dict[‘friendInfo‘][0],friend_db_dict[‘friendInfo‘][1],friend_db_dict[‘friendInfo‘][2]
123                 ,friend_db_dict[‘friendInfo‘][3],friend_db_dict[‘friendInfo‘][4],friend_db_dict[‘friendPlace‘][0],
124                 friend_db_dict[‘friendPlace‘][1],friend_db_dict[‘friendPlace‘][2],friend_db_dict[‘friendPlace‘][3],
125                 friend_db_dict[‘friendPlace‘][4],friend_db_dict[‘friendPlace‘][5],friend_db_dict[‘friendPlace‘][6],
126                 friend_db_dict[‘friendNet‘][0],friend_db_dict[‘friendNet‘][1],friend_db_dict[‘friendNet‘][2],
127                 friend_db_dict[‘friendNet‘][3],friend_db_dict[‘friendNet‘][4])
128     print(‘sql: ‘,sql)
129     cur.execute(sql)
130     conn.commit()
131     cur.close()
132     conn.close()
133 
134 def main():
135     """主要操作
136 
137     :return: void
138     """
139     req = requests.session()
140     headers={‘host‘: ‘h5.qzone.qq.com‘,
141              ‘accept-encoding‘:‘gzip, deflate, br‘,
142              ‘accept-language‘:‘zh-CN,zh;q=0.8‘,
143              ‘accept‘:‘text/html,application/xhtml+xml,application/xml;q=0.9,image/webp,image/apng,*/*;q=0.8‘,
144              ‘user-agent‘:‘Mozilla/5.0 (X11; Linux x86_64) AppleWebKit/537.36 (KHTML, like Gecko) Chrome/‘
145                           ‘59.0.3071.115 Safari/537.36‘,
146              ‘connection‘: ‘keep-alive‘}
147     cookie = web_login_cookie()
148     print(‘cookie‘,cookie)
149     g_tk = get_key(cookie)
150     qzonetoken_friend = get_key_values(cookie,‘ptcz=‘)
151     uin_friend = get_key_values(cookie,‘ptui_loginuin=‘)
152     rd_friend = get_key_values(cookie,‘_qpsvr_localtk=‘)
153     print(‘friend_data‘,‘qzontoken:%s;uin:%s;rd:%s‘ %(qzonetoken_friend,uin_friend,rd_friend))
154     headers[‘Cookie‘]=cookie
155     params_friend = {"uin": uin_friend,"fupdate": 1,"action": 1,"do":1,"g_tk":g_tk,"rd":rd_friend,
156                      ‘qzonetoken‘:qzonetoken_friend}
157     url_friend = ‘https://user.qzone.qq.com/proxy/domain/r.qzone.qq.com/cgi-bin/tfriend/friend_ship_manager.cgi?‘
158     data_friend_str = send_requests(req,headers,url_friend,params=params_friend)
159     data_friend_dict = loads(data_friend_str[0+len(‘_Callback(‘):data_friend_str.find(‘);‘)])
160     print(‘data_friend_dict: ‘,data_friend_dict)
161     if data_friend_dict[‘code‘] != 0: # code = -3000 message = ‘请先登录‘
162         time.sleep(10)
163         main()
164     else:
165         data_friend_list = list(data_friend_dict[‘data‘][‘items_list‘])
166         for i in range(len(data_friend_list)):
167             each_uin = data_friend_list[i][‘uin‘]
168             each_url = ‘https://h5.qzone.qq.com/proxy/domain/base.qzone.qq.com/cgi-bin/user/cgi_userinfo_get_all?‘
169             params_each = {"uin": each_uin, "fupdate": 1, "vuin": uin_friend, "g_tk": g_tk, "rd": rd_friend,
170                            ‘qzonetoken‘: qzonetoken_friend}
171             time.sleep(1)
172             data_each_str = send_requests(req,headers,each_url,params_each)
173             try:
174                 data_each_dict = loads(data_each_str[0+len("_Callback("):data_each_str.find(");")])
175             except json.decoder.JSONDecodeError as e:
176                 with open(‘leak.txt‘,‘a‘,encoding=‘utf8‘) as file: # 数据持久化，统计错误信息
177                     file.write(‘except: ‘ + str(each_uin) + " " + data_friend_list[i][‘name‘] + " " + e.msg + "\n")
178                     continue
179             print(‘data_each_dict: ‘,data_each_dict)
180             if data_each_dict[‘code‘] == 0: # code = -4009 message = ‘没有访问权限‘
181                 friend_db(data_each_dict[‘data‘],name=data_friend_list[i][‘name‘])
182             else:
183                 with open(‘leak.txt‘,‘a‘,encoding=‘utf8‘) as file: # 数据持久化，统计错误信息
184                     file.write((‘没有访问权限: ‘ + str(each_uin) + " " + data_friend_list[i][‘name‘] + "\n"))
185 main()