直接上代码:
代码如下:
#!/usr/bin/python
# -*- coding: utf-8 -*-
import urllib
import os,datetime,string
import sys
from bs4 import BeautifulSoup
reload(sys)
sys.setdefaultencoding('utf-8')
__BASEURL__ = 'http://bj.58.com/'
__INITURL__ = "http://bj.58.com/shoujiweixiu/"
soup = BeautifulSoup(urllib.urlopen(__INITURL__))
lvlELements = soup.html.body.find('div','selectbarTable').find('tr').find_next_sibling('tr')('a',href=True)
f = open('data1.txt','a')
for element in lvlELements[1:]:
f.write((element.get_text()+'\n\r' ))
url = __BASEURL__ + element.get('href')
print url
soup = BeautifulSoup(urllib.urlopen(url))
lv2ELements = soup.html.body.fi