【Python】爬虫

参考：http://www.cnblogs.com/fnng/p/3576154.html

 

import urllib,re
page = urllib.urlopen('http://trend.baidu.lecai.com/ssq/')
html = page.read()
reg = r"<td class='chart_table_td omission_entry omission_hit .+_ball background_color_.+'>\d{2}</td>"
htmlre = re.compile(reg)
balllist = re.findall(htmlre,html)
len(balllist)
# 30天，6红1蓝，210个结果

last30=[]
redlist=[]
for item in imglist:
    if 'red_ball' in item:
        redlist.append(item[-7:-5])
    elif 'blue_ball' in item:
        last30.append({'blue':item[-7:-5],'red':redlist})
        redlist=[]

print last30