Python爬虫工程师学习路径 update 2018/08/11

import requests
import json

url = "https://www.sojson.com/open/api/weather/json.shtml?city={}"


def get_weather(city):
    """
    通过参数city获取不同城市天气
    :param city: 城市
    :return: 天气详细信息的字典
    """
    res = requests.get(url=url.format(city))
    print(json.loads(res.text))


if __name__ == '__main__':
    get_weather("武汉")

import requests
from bs4 import BeautifulSoup

session = requests.session()
headers = {
    "User-Agent": "Mozilla/5.0 (Windows NT 10.0; Win64; x64) AppleWebKit/537.36 (KHTML, like Gecko) Chrome/68.0.3440.106 Safari/537.36",
}
login_url = "http://www.guanggoo.com/login"


def login(user, password, xsrf):
    """
    登陆光谷社区请求函数
    :param user: 用户名
    :param password: 加密后的密码
    :param xsrf: 前端验证码
    :return: True/False
    """

    data = {
        "email": user,
        "password": password,
        "_xsrf": xsrf,
    }
    res = session.post(login_url, data=data, headers=headers)
    if res.history and res.history[0].status_code == 302:
        print("登陆成功")
        return True
    else:
        print("登陆失败")
        return False


def parse_xsrf():
    """
    获取登陆过程中参数_xsrf的值
    :return:
    """
    res = session.get(login_url, headers=headers)
    if res.status_code != 200:
        print("获取xsrf失败")
        raise requests.RequestException
    soup = BeautifulSoup(res.text, "html.parser")
    result = soup.find("input", attrs={"name": "_xsrf"}).get("value")
    return result


def get_information():
    """
    通过访问用户中心判断是否登陆成功,如果没有登陆的用户,访问用户中心会直接跳转到登陆界面
    :return:
    """
    information_url = "http://www.guanggoo.com/setting"
    res = session.get(information_url, headers=headers)
    soup = BeautifulSoup(res.text, "html.parser")
    result = soup.find("input", id="username")
    if result:
        print("获取用户名成功,username:[{}]".format(result.get("value")))
    else:
        print("获取用户名失败,可能前面登陆过程没有成功")


if __name__ == '__main__':
    user = input("请输入用户名:")
    password = input("请输入加密后的密码:")
    code = parse_xsrf()
    login_result = login(user, password, code)
    get_information()