源码如下 :保存成utf-8 bd.py 文件
复制代码 代码如下:
#!/usr/bin/env python
# -*- coding: utf-8 -*-
import sys
import urllib ,urllib2
import re
def baidu(w):
url= \”http://www.baidu.com/s?\”
values = {
\”w\”:w.encode(\’gbk\’,\’ignore\’)
}
data = urllib.urlencode(values)
newurl = url + data
response = urllib2.urlopen(newurl)
the_page = response.read().decode(\’gbk\’,\’ignore\’)
return the_page
def ana(data,mysite):
o = re.compile(\”href=\\\”(.+?)\\\”\”)
f = o.findall(data)
line = 1
for ff in f:
if not re.search(\”baidu\”,ff) and not re.search(\”^s\\?\”,ff) and re.search(\”^http:\\/\\/\”,ff):
if re.search(mysite,ff):
print \”* \” ,line ,ff
else:
print line ,ff
line += 1
if __name__ == \”__main__\”:
mysite = sys.argv[2]
data = baidu(sys.argv[1].decode(\’utf-8\’))
ana(data,mysite)
用法 python bd.py \”关键字\” 我的域名部分 :
例:
复制代码 代码如下:
python bd.py \”vbs\” \”jb51.net\”