1 from myproject.dajie import Jobs, IpAgentPool
 2 from myproject import agentPool
 3 import http.cookiejar
 4 import urllib.request
 5 import urllib.parse
 6 import random
 7 import re
 8 #import pymssql
 9
10 class dj():
11
12     def __init__(self):
13         # -----BASEURL  为目标网站的URL
14         # -----ToSearchJob  为需要搜寻的工作
15         # -----Agent  为Agent池，用于伪装浏览器
16         # -----opener 为自己建造的一个opener，配合cookiejar可用于存储cookies
17         def Myopener(self):
18                 cookie = http.cookiejar.CookieJar()
19                 return urllib.request.build_opener(urllib.request.HTTPCookieProcessor(cookie))
20
21         self.BASEURL = "https://www.dajie.com/"
22         # self.ToSearchJob=Jobs.TheJobNeedToSearch
23         self.AgentPool = agentPool.userAgent
24         self.IpPool = IpAgentPool.ipPool
25
26         self.opener = Myopener(self=self)
27
28         pass
29
30     def getContext(self):
31         url="https://so.dajie.com/job/search?keyword=%E7%A8%8B%E5%BA%8F%E5%91%98&from=job"
32
33         header=\
34             {
35                 "User-Agent":agentPool.userAgent[int(random.random()*4)],
36                 "Referer":"https://so.dajie.com/job/search?keyword=%E7%A8%8B%E5%BA%8F%E5%91%98&from=job"
37             }
38
39         head=[]
40
41         for key,value in header.items():
42             head.append((key,value))
43
44         self.opener.addheaders=head
45
46         #session_cookie用于保存服务器返回的cookie，并将其保存
47         #
48         #其实只需要保存  session_cookie["SO_COOKIE_V2"]  即可，其余的是多余的。服务器在进行验证的时候，只会验证 SO_COOKIE_V2
49         #
50         session_cookie={}
51         session=self.opener.open("https://so.dajie.com/job/search?keyword=%E7%A8%8B%E5%BA%8F%E5%91%98&from=job")
52         print((session.info()))
53
54         session_cookie["DJ_RF"]= re.findall(r"DJ_RF=.+?;",str(session.info()))[0].split("=")[1]
55         session_cookie["DJ_EU"]=re.findall(r"DJ_EU=.+?;",str(session.info()))[0].split("=")[1]
56         session_cookie["DJ_UVID"] = re.findall(r"DJ_UVID=.+?;", str(session.info()))[0].split("=")[1]
57         session_cookie["SO_COOKIE_V2"] = re.findall(r"SO_COOKIE_V2=.+?;", str(session.info()))[0].split("=")[1]
58
59         #data 包含的是所需要传入的 cookie
60         data=\
61             {
62                 "DJ_RF":session_cookie["DJ_RF"].strip(";"),
63                 "DJ_EU":session_cookie["DJ_EU"].strip(";"),
64                 "DJ_UVID":session_cookie["DJ_UVID"].strip(";"),
65                 "SO_COOKIE_V2":session_cookie["SO_COOKIE_V2"].strip(";"),
66                 "__login_tips":1,
67             }
68         #将 数据解析为传入数据的格式
69         _data=urllib.parse.urlencode(data,"utf-8")
70         print("______________")
71         print(_data)
72         #
73         #_url 指的是ajax的链接地址
74         _url="https://so.dajie.com/job/ajax/search/filter?keyword=%E7%A8%8B%E5%BA%8F%E5%91%98&order=1&city=&recruitType=&salary=&experience=&page=1&positionFunction=&_CSRFToken=&ajax=1"
75         req=self.opener.open(_url,data=_data.encode("utf-8"))
76
77         print("-----------------")
78         print(req.read().decode("utf-8"))
79
80         #print(req.read().decode("utf-8"))
81
82
83
84 if __name__=="__main__":
85     _dj=dj()
86     _dj.getContext()