|
- import re
- import sys
- import io
- from bs4 import BeautifulSoup
- from selenium import webdriver
- from datetime import datetime
- def hot():
- driver = webdriver.PhantomJS()
- myUrl = 'http://qiqu.uc.cn/?uc_param_str=frpfvedncpssntnwbipreime#!/changwen?id=0902389688c6731bf26928a6ecfedbec!!tag=xjrb!!from=index__index'
- driver.get(myUrl)
- fp = open(datetime.now().date().isoformat()+'.html', 'a+', encoding='utf-8') # 用来将主页上的个人信息存储
- bsObj=BeautifulSoup(driver.page_source,features= "html.parser")
- myItems = bsObj.find_all('div',{'class':'cw-content'})
- for item in myItems:
- fp.write(item.prettify())
- print(item)
- print(u'------------------------------结束-----------------------------------------')
- if __name__=='__main__':
- sys.stdout = io.TextIOWrapper(sys.stdout.buffer, errors='replace', line_buffering=True) # print(打印中文)
- hot()
- sys.stdout.flush()
- input()
复制代码
|
|