爬蟲示例

pass

import requests
import re

session = requests.session()

# 第一步：訪問登錄頁,拿到X_Anti_Forge_Token，X_Anti_Forge_Code
# 一、請求url:https://passport.lagou.com/login/login.html
# 二、請求方法:GET
# 三、請求頭:
#    User-agent
r1 = session.get('https://passport.lagou.com/login/login.html',
                 headers={
                     'User-Agent': 'Mozilla/5.0 (Windows NT 6.1; WOW64) AppleWebKit/537.36 (KHTML, like Gecko) Chrome/63.0.3239.132 Safari/537.36',
                 },
                 )

X_Anti_Forge_Token = re.findall("X_Anti_Forge_Token = '(.*?)'", r1.text, re.S)[0]
X_Anti_Forge_Code = re.findall("X_Anti_Forge_Code = '(.*?)'", r1.text, re.S)[0]
# print(X_Anti_Forge_Token,X_Anti_Forge_Code)

# 第二步：登錄
# 一、請求url:https://passport.lagou.com/login/login.json
# 二、請求方法:POST
# 三、請求頭:
#    cookie
#    User-agent
#    Referer:https://passport.lagou.com/login/login.html
#    X-Anit-Forge-Code:53165984
#    X-Anit-Forge-Token:3b6a2f62-80f0-428b-8efb-ef72fc100d78
#    X-Requested-With:XMLHttpRequest
# 四、請求體：
# isValidate:true
# username:18611453110
# password:70621c64832c4d4d66a47be6150b4a8e
# request_form_verifyCode:''
# submit:''
r2 = session.post('https://passport.lagou.com/login/login.json',
                  headers={
                      'User-Agent': 'Mozilla/5.0 (Windows NT 6.1; WOW64) AppleWebKit/537.36 (KHTML, like Gecko) Chrome/63.0.3239.132 Safari/537.36',
                      'Referer': 'https://passport.lagou.com/login/login.html',
                      'X-Anit-Forge-Code': X_Anti_Forge_Code,
                      'X-Anit-Forge-Token': X_Anti_Forge_Token,
                      'X-Requested-With': 'XMLHttpRequest'
                  },
                  data={
                      "isValidate": True,
                      'username': '18611453110',
                      'password': '70621c64832c4d4d66a47be6150b4a8e',
                      'request_form_verifyCode': '',
                      'submit': ''
                  }
                  )
# 第三步：受權
# 一、請求url:https://passport.lagou.com/grantServiceTicket/grant.html
# 二、請求方法:GET
# 三、請求頭:
#    User-agent
#    Referer:https://passport.lagou.com/login/login.html

r3 = session.get('https://passport.lagou.com/grantServiceTicket/grant.html',
                 headers={
                     'User-Agent': 'Mozilla/5.0 (Windows NT 6.1; WOW64) AppleWebKit/537.36 (KHTML, like Gecko) Chrome/63.0.3239.132 Safari/537.36',
                     'Referer': 'https://passport.lagou.com/login/login.html',
                 }
                 )

# 第四步：驗證
r4 = session.get('https://www.lagou.com/resume/myresume.html',
                 headers={
                     'User-Agent': 'Mozilla/5.0 (Windows NT 6.1; WOW64) AppleWebKit/537.36 (KHTML, like Gecko) Chrome/63.0.3239.132 Safari/537.36',
                 }
                 )

# print('18611453110' in r4.text)


# 第五步：篩選職位信息
# 請求url：https://www.lagou.com/jobs/list_java%E9%AB%98%E7%BA%A7%E5%BC%80%E5%8F%91
# 請求方法：GET
# 請求頭：
# User-Agent
# 請求參數：
# gj:3年及如下
# px:default
# yx:25k-50k
# city:北京
from urllib.parse import urlencode

res = urlencode({'k': 'java高級開發'}, encoding='utf-8').split('=')[-1]
url = 'https://www.lagou.com/jobs/list_' + res
#
# r5 = session.get(url,
#                  headers={
#                      'User-Agent': 'Mozilla/5.0 (Windows NT 6.1; WOW64) AppleWebKit/537.36 (KHTML, like Gecko) Chrome/63.0.3239.132 Safari/537.36',
#                  },
#                  params={
#                      'gj': '3年及如下',
#                      'px': 'default',
#                      'yx': '25k-50k',
#                      'city': '北京'
#                  }
#                  )
#
# print(r5.text)

#請求url：https://www.lagou.com/jobs/positionAjax.json
#請求方法：POST
#請求頭
#    Referer
#    User-Agent
#請求體：
    # first:true
    # pn:1
    # kd:java高級開發
#請求參數
# params={
#      'gj': '3年及如下',
#      'px': 'default',
#      'yx': '25k-50k',
#      'city': '北京',
#     'needAddtionalResult':False,
#     'isSchoolJob':0
# }
r6=session.post('https://www.lagou.com/jobs/positionAjax.json',
             headers={
                    'Referer':url,
                     'User-Agent': 'Mozilla/5.0 (Windows NT 6.1; WOW64) AppleWebKit/537.36 (KHTML, like Gecko) Chrome/63.0.3239.132 Safari/537.36',

             },
             data={
                 'first':True,
                 'pn':2,
                 'kd':'java高級開發'
             },
             params={
                 'gj': '3年及如下',
                 'px': 'default',
                 'yx': '25k-50k',
                 'city': '北京',
                 'needAddtionalResult': False,
                 'isSchoolJob': 0
             }
             )


from pprint import pprint
# print(r6.json())
comapines_list=r6.json()['content']['positionResult']['result']
for comapiny in comapines_list:
    positionId=comapiny['positionId']
    company_link='https://www.lagou.com/jobs/{pos_id}.html'.format(pos_id=positionId)
    companyShortName = comapiny['companyShortName']
    positionName = comapiny['positionName']
    salary = comapiny['salary']
    print('''
    詳情鏈接:%s
    公司名：%s
    職位名：%s
    薪資：%s
    ''' %(company_link,companyShortName,positionName,salary))


    #第七步：訪問詳情頁，拿到X_Anti_Forge_Token，X_Anti_Forge_Code
    # 請求url：詳情頁地址
    # 請求方式：GET
    # 請求頭：User-Agent
    r7=session.get(company_link,
                headers={
                    'User-Agent': 'Mozilla/5.0 (Windows NT 6.1; WOW64) AppleWebKit/537.36 (KHTML, like Gecko) Chrome/63.0.3239.132 Safari/537.36',
                }
                )
    X_Anti_Forge_Token = re.findall("X_Anti_Forge_Token = '(.*?)'", r7.text, re.S)[0]
    X_Anti_Forge_Code = re.findall("X_Anti_Forge_Code = '(.*?)'", r7.text, re.S)[0]
    # print(X_Anti_Forge_Token,X_Anti_Forge_Code)


    #第八步：投遞簡歷
    #請求url：https://www.lagou.com/mycenterDelay/deliverResumeBeforce.json
    #請求方式：POST
    #請求頭：
        #Referer:詳情頁地址
        #User-agent
        #X-Anit-Forge-Code:53165984
        #X-Anit-Forge-Token:3b6a2f62-80f0-428b-8efb-ef72fc100d78
        #X-Requested-With:XMLHttpRequest
    #請求體：
    # positionId:職位ID
    # type:1
    # force:true

    session.post('https://www.lagou.com/mycenterDelay/deliverResumeBeforce.json',
                 headers={
                     'User-Agent': 'Mozilla/5.0 (Windows NT 6.1; WOW64) AppleWebKit/537.36 (KHTML, like Gecko) Chrome/63.0.3239.132 Safari/537.36',
                     'Referer': company_link,
                     'X-Anit-Forge-Code': X_Anti_Forge_Code,
                     'X-Anit-Forge-Token': X_Anti_Forge_Token,
                     'X-Requested-With': 'XMLHttpRequest'
                 },
                 data={
    'positionId':positionId,
    'type':1,
    'force':True
                 }
                 )
    print('%s 投遞成功' %(companyShortName))

爬取拉勾網