Source

wd_pys60 / html_jie.py

Full commit

#!/usr/bin/envpython
 
import sys
import urllib
import HTMLParser
 
class CustomParser(HTMLParser.HTMLParser):
    selected=('input')
    #selected=('table','h1','font','ul','li','tr','td','a')    
    def reset(self):
        HTMLParser.HTMLParser.reset(self)
        self._level_stack=[]
    def handle_starttag(self,tag,attrs):
        if tag in CustomParser.selected:
            self._level_stack.append(tag)
    def handle_endtag(self,tag):
        if self._level_stack\
        and tag in CustomParser.selected\
        and tag==self._level_stack[-1]:
            self._level_stack.pop()
    def handle_data(self,data):
        if "/".join(self._level_stack) in (
            'table/tr/td',
            'table/tr/td/h1/font',
            'table/tr/td/ul/li'):
            print self._level_stack,data
        
if len(sys.argv)>1:
    params=urllib.urlencode({'ip':sys.argv[1],'action':2})
else:
    params=None

content=urllib.urlopen('http://202.114.90.117/',params).read().decode("gbk")
print content
import re
print re.findall('name="([^<>()"]*)"[^<>()]*value="([^<>()"]*)"',content)
print re.findall('<input [^<>()]*name="([^<>()"]*)"[^<>()]*"',content)