使用requests和BeautifulSoup爬取教务系统个人信息

Python

import requests
from bs4 import BeautifulSoup
if __name__ == "__main__":
    def getData(myInfo):
        # 首页
        homeUrl = "http://uia.whxy.edu.cn/cas/login?service=http%3A%2F%2Fportal.whxy.edu.cn%2Fc%2Fportal%2Flogin"
        # 登录
        loginUrl = homeUrl + "%3Fredirect%3D%252Fweb%252Fguest%252Findex%26p_l_id%3D10213"
        # 教务登录
        jwLogin = "http://59.172.226.5/eams/sso/login.action"
        # 教务主页
        jwHome = "http://59.172.226.5/eams/home.action"
        # 学籍信息
        xj = "http://59.172.226.5/eams/stdDetail.action"
        # 课表
        kb = "http://59.172.226.5/eams/courseTableForStd.action"
        kb2 = "http://59.172.226.5/eams/courseTableForStd!courseTable.action"
        s = requests.Session()
        r = s.get(homeUrl)
        r.encoding = r.apparent_encoding
        soup = BeautifulSoup(r.text, "html.parser")
        lt = soup.find(attrs={"name": "lt"})["value"]
        # 获取 lt
        myInfo["lt"] = lt
        myInfo["_eventId"] = "submit"
        r = s.post(loginUrl, data=myInfo)
        soup = BeautifulSoup(r.text, "html.parser")
        # 一卡通余额
        balance = soup.select_one("#balance").string
        r = s.get(jwLogin)
        r = s.get(jwHome)
        r = s.get(xj)
        soup = BeautifulSoup(r.text, "html.parser")
        data_list = {"一卡通余额": balance}
        for idx, tr in enumerate(soup.find_all('tr')):
            for tdx, td in enumerate(tr.find_all("td")):
                if len(td.contents) != 0:
                    if idx == 1 and tdx == 1:
                        data_list["学号"] = td.contents[0]
                    elif idx == 1 and tdx == 3:
                        data_list["姓名"] = td.contents[0]
                    elif idx == 2 and tdx == 3:
                        data_list["性别"] = td.contents[0]
                    elif idx == 5 and tdx == 3:
                        data_list["学院"] = td.contents[0]
                    elif idx == 6 and tdx == 1:
                        data_list["专业"] = td.contents[0]
                    elif idx == 11 and tdx == 3:
                        data_list["班级"] = td.contents[0]
        print(data_list)
    info = {"username": "2015****", "password": "Whxy27****"}
    getData(info)