模拟登录丁香园并抓取所有信息

丁香园论坛：http://www.dxy.cn/bbs/thread/626626#626626

import time
from selenium import webdriver
from selenium.common.exceptions import TimeoutException, NoSuchElementException

def login(browser, username, password):
    try:
        browser.find_element_by_link_text("登录").click()
        browser.find_element_by_class_name("ico_pc").click()
        element = browser.find_element_by_name("username")
        element.clear()
        element.send_keys(username)

        element = browser.find_element_by_name("password")
        element.clear()
        element.send_keys(password)

        browser.find_element_by_class_name("button").click()

    except TimeoutException:
        print("time out")
    except NoSuchElementException:
        print("No Element")


def get_content(browser):
    print("success login")
    time.sleep(5)
    auth = browser.find_elements_by_class_name('auth')  # 昵称
    level = browser.find_elements_by_class_name('info') # 等级
    user_atten = browser.find_elements_by_class_name('user_atten')  # 用户属性
    content = browser.find_elements_by_class_name('postbody')   # 内容
    sfile = open('data.txt','a',encoding='utf-8')
    for i in range(len(content)):
        num = user_atten[i].find_elements_by_tag_name('a')
        data = str({'num':i+1,'name':auth[i].text,'level':level[i].text,'score':num[0].text,'vote':num[2].text,'dingdang':num[4].text,'content':content[i].text.replace(" ", "").replace("\n", "")}) +"\n"
        sfile.writelines(data)
    print("success write")
    sfile.close()

if __name__ == '__main__':
    browser = webdriver.Chrome(executable_path = 'C:\Program Files (x86)\Google\Chrome\Application\chromedriver.exe')
    browser.get('http://www.dxy.cn/bbs/thread/626626#626626')
    login(browser, 'xxxxxxname', 'pwd')
    get_content(browser)
    browser.quit()

···