diff options
| -rwxr-xr-x | proxy.py | 26 |
1 files changed, 12 insertions, 14 deletions
@@ -16,7 +16,7 @@ ENDPOINTS = [ kB = 1024 #minimum entity size to start a paralel fetch -THRESHOLD = 0 * kB #512 * kB +THRESHOLD = 512 * kB #first fetch-range blocksize INIT_BLOCKSIZE = 512 * kB #lower bound of fetch-range blocksize optimization @@ -53,7 +53,7 @@ class Fetcher(asynchat.async_chat): def handle_connect (self): print self, "Start" self.send("GET http://%s:%s%s HTTP/1.0\r\n" % ( self.url.hostname, self.url.port or 80, self.url.path )) - for key in filter(lambda k: k not in ("range"), self.headers.keys()): #origin request headers + for key in filter(lambda k: k not in ("range"), self.headers.keys()): #send origin request headers self.send("%s: %s\r\n" % (key, self.headers[key])) self.send("Range: bytes=%s-%s\r\n" % (self.range[0], self.range[1])) self.send("\r\n") @@ -75,8 +75,8 @@ class Fetcher(asynchat.async_chat): if self.pos >= self.range[1]: self.stop_time = time.time() print self, "finished" - #XXX just to make sure the next fetcher will be started - self.reader.handle_incoming_data(self, "") + #make sure the next fetcher will be started + self.reader.handle_incoming_data(self) self.close() elif self.state ==1: #header self.http_header += data @@ -112,8 +112,8 @@ class MultipleProxyReader(object): for proxy in ENDPOINTS: self.fetchers.append( Fetcher(self, proxy, self.url, self.header, self.next_range(INIT_BLOCKSIZE)) ) - def handle_incoming_data(self, fetcher, data): - if len(data) == 0: + def handle_incoming_data(self, fetcher, data=None): + if not data: self.fetchers = filter(lambda f: f != fetcher, self.fetchers) else: heappush(self.blocks, (fetcher.pos, data)) @@ -223,11 +223,9 @@ class HTTPProxyServer(asyncore.dispatcher): def handle_request(self, channel, method, path, headers): url = urlparse.urlparse(path) - print method, path if method != "GET" or url.query != "": - print "non-GET or query, bypass" #do not handle non-GET or GET with Query (?foo=bla) requests - return self._bypass_request(channel, method, url) + return self._bypass_request(channel, method, url, headers) #check for content-length header with a HEAD request conn = httplib.HTTPConnection(url.hostname, url.port or 80) @@ -237,24 +235,24 @@ class HTTPProxyServer(asyncore.dispatcher): if len( content_length ) == 0: # no content length given, bypass this request print "missing content-length, bypass" - self._bypass_request(channel, "GET", url) + self._bypass_request(channel, "GET", url, headers) else: content_length = int(content_length[0][1]) if content_length < THRESHOLD: - self._bypass_request(channel, "GET", url) + self._bypass_request(channel, "GET", url, headers) else: MultipleProxyReader(channel, url, headers, content_length) - def _bypass_request(self, channel, method, url): - print "_bypass request hostname=%s port=%s path=%s" % (url.hostname, url.port or 80, url.path) + def _bypass_request(self, channel, method, url, headers): + print "_bypass request: %s %s" % (method, urlparse.urlunparse(url)) #XXX hier sollte nicht proxy gespielt werden sondern #die daten 1-zu-1 durchgereicht werden. #Weiterhin sollte sichergestellt werden, dass die requests #zu Host X1 immer ueber Proxy Y1 geroutet werden # etwa proxy=proxies[ stuff(hostname) % len(proxies) ] conn = httplib.HTTPConnection(url.hostname, url.port or 80) - conn.request(method, url.path) + conn.request(method, url.path, body="", headers=headers) resp = conn.getresponse() channel.push("HTTP/1.0 200 OK\r\nX-Proxy: Magicproxy (superpower disabled)\r\n") channel.push( "\r\n".join(map(lambda k: "%s: %s" % (k[0],k[1]), resp.getheaders())) ) |