logging proxy server

eroproxy.py

from twisted.web import proxy, http
from twisted.internet import reactor
import re
import psyco
psyco.full()

store_type = ['text/html', 'json']
ignore_host = []
tube8_re = re.compile("^http://www.tube8.com/[a-z]+/[\w\d.:_-]+/\d{3,6}/", re.M)

def _parse_type(val):
  res = val.split(';')
  if len(res) == 2:
    type, char = res
    char = char.split('=')[1]
  else:
    res.append('utf8')
    type, char = res
  
  return (type.strip(), char.strip())

class LogProxyClient(proxy.ProxyClient):
    
    def handleHeader(self, key, value):
        proxy.ProxyClient.handleHeader(self, key, value)
        if key == 'Content-Type':
            type, char = _parse_type(value)
            if type in store_type:
                self.content_type = type
                self.charset = char
                self._save = ""
    
    def handleResponsePart(self, data):
        proxy.ProxyClient.handleResponsePart(self, data)
        #if hasattr(self, "_save"):
        #    self._save += data
    
    def handleResponseEnd(self):
        proxy.ProxyClient.handleResponseEnd(self)
        uri = self.father.uri
        if uri.startswith("http://video.xnxx.com/video"):
            print(uri)
            f = open("xnxx.txt", "a+")
            f.write(uri+"\n")
            f.close()
        elif tube8_re.search(uri):
            print(uri)
            f = open("tube8.txt", "a+")
            f.write(uri+"\n")
            f.close()
        elif uri.startswith("http://www.jizzhut.com/videos/"):
            print(uri)
            f = open("jizzhut.txt", "a+")
            f.write(uri+"\n")
            f.close()
        elif uri.startswith("http://xhamster.com/movies/"):
            print(uri)
            f = open("xhamster.txt", "a+")
            f.write(uri+"\n")
            f.close()
        elif uri.startswith("http://www.empflix.com/view.php"):
            print(uri)
            f = open("empflix.txt", "a+")
            f.write(uri+"\n")
            f.close()

            
        #if hasattr(self, "_save"):
        #    req_host = self.father.getRequestHostname()
        #    if not req_host in ignore_host:
                #self.father.proxylog.write(self.father.uri, self.charset, self.content_type, self._save)
                #print "write %s %s" % (self.father.uri, self.content_type)
        #        del(self._save)

class LogProxyClientFactory(proxy.ProxyClientFactory):

    def buildProtocol(self, addr):
        client = proxy.ProxyClientFactory.buildProtocol(self, addr)
        client.__class__ = LogProxyClient
        return client

class LogProxyRequest(proxy.ProxyRequest):
   
    protocols = {'http': LogProxyClientFactory}

    def __init__(self, *args):
        proxy.ProxyRequest.__init__(self, *args)

class LogProxy(proxy.Proxy):
    
    def __init__(self):
        proxy.Proxy.__init__(self)

    def requestFactory(self, *args):
        return LogProxyRequest(*args)


class ProxyFactory(http.HTTPFactory):

    def __init__(self):
        http.HTTPFactory.__init__(self)
    
    def buildProtocol(self, addr):
        protocol = LogProxy()
        return protocol

def start_server(port=8000):
    reactor.listenTCP(port, ProxyFactory())
    reactor.run()

if __name__ == '__main__':
    start_server()