| 1 |
|
|---|
| 2 |
|
|---|
| 3 |
|
|---|
| 4 |
|
|---|
| 5 |
""" |
|---|
| 6 |
Simplistic HTTP proxy support. |
|---|
| 7 |
|
|---|
| 8 |
This comes in two main variants - the Proxy and the ReverseProxy. |
|---|
| 9 |
|
|---|
| 10 |
When a Proxy is in use, a browser trying to connect to a server (say, |
|---|
| 11 |
www.yahoo.com) will be intercepted by the Proxy, and the proxy will covertly |
|---|
| 12 |
connect to the server, and return the result. |
|---|
| 13 |
|
|---|
| 14 |
When a ReverseProxy is in use, the client connects directly to the ReverseProxy |
|---|
| 15 |
(say, www.yahoo.com) which farms off the request to one of a pool of servers, |
|---|
| 16 |
and returns the result. |
|---|
| 17 |
|
|---|
| 18 |
Normally, a Proxy is used on the client end of an Internet connection, while a |
|---|
| 19 |
ReverseProxy is used on the server end. |
|---|
| 20 |
""" |
|---|
| 21 |
|
|---|
| 22 |
import urlparse |
|---|
| 23 |
from urllib import quote as urlquote |
|---|
| 24 |
|
|---|
| 25 |
from twisted.internet import reactor |
|---|
| 26 |
from twisted.internet.protocol import ClientFactory |
|---|
| 27 |
from twisted.web.resource import Resource |
|---|
| 28 |
from twisted.web.server import NOT_DONE_YET |
|---|
| 29 |
from twisted.web.http import HTTPClient, Request, HTTPChannel |
|---|
| 30 |
|
|---|
| 31 |
|
|---|
| 32 |
|
|---|
| 33 |
class ProxyClient(HTTPClient): |
|---|
| 34 |
""" |
|---|
| 35 |
Used by ProxyClientFactory to implement a simple web proxy. |
|---|
| 36 |
|
|---|
| 37 |
@ivar _finished: A flag which indicates whether or not the original request |
|---|
| 38 |
has been finished yet. |
|---|
| 39 |
""" |
|---|
| 40 |
_finished = False |
|---|
| 41 |
|
|---|
| 42 |
def __init__(self, command, rest, version, headers, data, father): |
|---|
| 43 |
self.father = father |
|---|
| 44 |
self.command = command |
|---|
| 45 |
self.rest = rest |
|---|
| 46 |
if "proxy-connection" in headers: |
|---|
| 47 |
del headers["proxy-connection"] |
|---|
| 48 |
headers["connection"] = "close" |
|---|
| 49 |
self.headers = headers |
|---|
| 50 |
self.data = data |
|---|
| 51 |
|
|---|
| 52 |
|
|---|
| 53 |
def connectionMade(self): |
|---|
| 54 |
self.sendCommand(self.command, self.rest) |
|---|
| 55 |
for header, value in self.headers.items(): |
|---|
| 56 |
self.sendHeader(header, value) |
|---|
| 57 |
self.endHeaders() |
|---|
| 58 |
self.transport.write(self.data) |
|---|
| 59 |
|
|---|
| 60 |
|
|---|
| 61 |
def handleStatus(self, version, code, message): |
|---|
| 62 |
self.father.setResponseCode(int(code), message) |
|---|
| 63 |
|
|---|
| 64 |
|
|---|
| 65 |
def handleHeader(self, key, value): |
|---|
| 66 |
self.father.responseHeaders.addRawHeader(key, value) |
|---|
| 67 |
|
|---|
| 68 |
|
|---|
| 69 |
def handleResponsePart(self, buffer): |
|---|
| 70 |
self.father.write(buffer) |
|---|
| 71 |
|
|---|
| 72 |
|
|---|
| 73 |
def handleResponseEnd(self): |
|---|
| 74 |
""" |
|---|
| 75 |
Finish the original request, indicating that the response has been |
|---|
| 76 |
completely written to it, and disconnect the outgoing transport. |
|---|
| 77 |
""" |
|---|
| 78 |
if not self._finished: |
|---|
| 79 |
self._finished = True |
|---|
| 80 |
self.father.finish() |
|---|
| 81 |
self.transport.loseConnection() |
|---|
| 82 |
|
|---|
| 83 |
|
|---|
| 84 |
|
|---|
| 85 |
class ProxyClientFactory(ClientFactory): |
|---|
| 86 |
""" |
|---|
| 87 |
Used by ProxyRequest to implement a simple web proxy. |
|---|
| 88 |
""" |
|---|
| 89 |
|
|---|
| 90 |
protocol = ProxyClient |
|---|
| 91 |
|
|---|
| 92 |
|
|---|
| 93 |
def __init__(self, command, rest, version, headers, data, father): |
|---|
| 94 |
self.father = father |
|---|
| 95 |
self.command = command |
|---|
| 96 |
self.rest = rest |
|---|
| 97 |
self.headers = headers |
|---|
| 98 |
self.data = data |
|---|
| 99 |
self.version = version |
|---|
| 100 |
|
|---|
| 101 |
|
|---|
| 102 |
def buildProtocol(self, addr): |
|---|
| 103 |
return self.protocol(self.command, self.rest, self.version, |
|---|
| 104 |
self.headers, self.data, self.father) |
|---|
| 105 |
|
|---|
| 106 |
|
|---|
| 107 |
def clientConnectionFailed(self, connector, reason): |
|---|
| 108 |
""" |
|---|
| 109 |
Report a connection failure in a response to the incoming request as |
|---|
| 110 |
an error. |
|---|
| 111 |
""" |
|---|
| 112 |
self.father.setResponseCode(501, "Gateway error") |
|---|
| 113 |
self.father.responseHeaders.addRawHeader("Content-Type", "text/html") |
|---|
| 114 |
self.father.write("<H1>Could not connect</H1>") |
|---|
| 115 |
self.father.finish() |
|---|
| 116 |
|
|---|
| 117 |
|
|---|
| 118 |
|
|---|
| 119 |
class ProxyRequest(Request): |
|---|
| 120 |
""" |
|---|
| 121 |
Used by Proxy to implement a simple web proxy. |
|---|
| 122 |
|
|---|
| 123 |
@ivar reactor: the reactor used to create connections. |
|---|
| 124 |
@type reactor: object providing L{twisted.internet.interfaces.IReactorTCP} |
|---|
| 125 |
""" |
|---|
| 126 |
|
|---|
| 127 |
protocols = {'http': ProxyClientFactory} |
|---|
| 128 |
ports = {'http': 80} |
|---|
| 129 |
|
|---|
| 130 |
def __init__(self, channel, queued, reactor=reactor): |
|---|
| 131 |
Request.__init__(self, channel, queued) |
|---|
| 132 |
self.reactor = reactor |
|---|
| 133 |
|
|---|
| 134 |
|
|---|
| 135 |
def process(self): |
|---|
| 136 |
parsed = urlparse.urlparse(self.uri) |
|---|
| 137 |
protocol = parsed[0] |
|---|
| 138 |
host = parsed[1] |
|---|
| 139 |
port = self.ports[protocol] |
|---|
| 140 |
if ':' in host: |
|---|
| 141 |
host, port = host.split(':') |
|---|
| 142 |
port = int(port) |
|---|
| 143 |
rest = urlparse.urlunparse(('', '') + parsed[2:]) |
|---|
| 144 |
if not rest: |
|---|
| 145 |
rest = rest + '/' |
|---|
| 146 |
class_ = self.protocols[protocol] |
|---|
| 147 |
headers = self.getAllHeaders().copy() |
|---|
| 148 |
if 'host' not in headers: |
|---|
| 149 |
headers['host'] = host |
|---|
| 150 |
self.content.seek(0, 0) |
|---|
| 151 |
s = self.content.read() |
|---|
| 152 |
clientFactory = class_(self.method, rest, self.clientproto, headers, |
|---|
| 153 |
s, self) |
|---|
| 154 |
self.reactor.connectTCP(host, port, clientFactory) |
|---|
| 155 |
|
|---|
| 156 |
|
|---|
| 157 |
|
|---|
| 158 |
class Proxy(HTTPChannel): |
|---|
| 159 |
""" |
|---|
| 160 |
This class implements a simple web proxy. |
|---|
| 161 |
|
|---|
| 162 |
Since it inherits from L{twisted.protocols.http.HTTPChannel}, to use it you |
|---|
| 163 |
should do something like this:: |
|---|
| 164 |
|
|---|
| 165 |
from twisted.web import http |
|---|
| 166 |
f = http.HTTPFactory() |
|---|
| 167 |
f.protocol = Proxy |
|---|
| 168 |
|
|---|
| 169 |
Make the HTTPFactory a listener on a port as per usual, and you have |
|---|
| 170 |
a fully-functioning web proxy! |
|---|
| 171 |
""" |
|---|
| 172 |
|
|---|
| 173 |
requestFactory = ProxyRequest |
|---|
| 174 |
|
|---|
| 175 |
|
|---|
| 176 |
|
|---|
| 177 |
class ReverseProxyRequest(Request): |
|---|
| 178 |
""" |
|---|
| 179 |
Used by ReverseProxy to implement a simple reverse proxy. |
|---|
| 180 |
|
|---|
| 181 |
@ivar proxyClientFactoryClass: a proxy client factory class, used to create |
|---|
| 182 |
new connections. |
|---|
| 183 |
@type proxyClientFactoryClass: L{ClientFactory} |
|---|
| 184 |
|
|---|
| 185 |
@ivar reactor: the reactor used to create connections. |
|---|
| 186 |
@type reactor: object providing L{twisted.internet.interfaces.IReactorTCP} |
|---|
| 187 |
""" |
|---|
| 188 |
|
|---|
| 189 |
proxyClientFactoryClass = ProxyClientFactory |
|---|
| 190 |
|
|---|
| 191 |
def __init__(self, channel, queued, reactor=reactor): |
|---|
| 192 |
Request.__init__(self, channel, queued) |
|---|
| 193 |
self.reactor = reactor |
|---|
| 194 |
|
|---|
| 195 |
|
|---|
| 196 |
def process(self): |
|---|
| 197 |
""" |
|---|
| 198 |
Handle this request by connecting to the proxied server and forwarding |
|---|
| 199 |
it there, then forwarding the response back as the response to this |
|---|
| 200 |
request. |
|---|
| 201 |
""" |
|---|
| 202 |
self.received_headers['host'] = self.factory.host |
|---|
| 203 |
clientFactory = self.proxyClientFactoryClass( |
|---|
| 204 |
self.method, self.uri, self.clientproto, self.getAllHeaders(), |
|---|
| 205 |
self.content.read(), self) |
|---|
| 206 |
self.reactor.connectTCP(self.factory.host, self.factory.port, |
|---|
| 207 |
clientFactory) |
|---|
| 208 |
|
|---|
| 209 |
|
|---|
| 210 |
|
|---|
| 211 |
class ReverseProxy(HTTPChannel): |
|---|
| 212 |
""" |
|---|
| 213 |
Implements a simple reverse proxy. |
|---|
| 214 |
|
|---|
| 215 |
For details of usage, see the file examples/proxy.py. |
|---|
| 216 |
""" |
|---|
| 217 |
|
|---|
| 218 |
requestFactory = ReverseProxyRequest |
|---|
| 219 |
|
|---|
| 220 |
|
|---|
| 221 |
|
|---|
| 222 |
class ReverseProxyResource(Resource): |
|---|
| 223 |
""" |
|---|
| 224 |
Resource that renders the results gotten from another server |
|---|
| 225 |
|
|---|
| 226 |
Put this resource in the tree to cause everything below it to be relayed |
|---|
| 227 |
to a different server. |
|---|
| 228 |
|
|---|
| 229 |
@ivar proxyClientFactoryClass: a proxy client factory class, used to create |
|---|
| 230 |
new connections. |
|---|
| 231 |
@type proxyClientFactoryClass: L{ClientFactory} |
|---|
| 232 |
|
|---|
| 233 |
@ivar reactor: the reactor used to create connections. |
|---|
| 234 |
@type reactor: object providing L{twisted.internet.interfaces.IReactorTCP} |
|---|
| 235 |
""" |
|---|
| 236 |
|
|---|
| 237 |
proxyClientFactoryClass = ProxyClientFactory |
|---|
| 238 |
|
|---|
| 239 |
|
|---|
| 240 |
def __init__(self, host, port, path, reactor=reactor): |
|---|
| 241 |
""" |
|---|
| 242 |
@param host: the host of the web server to proxy. |
|---|
| 243 |
@type host: C{str} |
|---|
| 244 |
|
|---|
| 245 |
@param port: the port of the web server to proxy. |
|---|
| 246 |
@type port: C{port} |
|---|
| 247 |
|
|---|
| 248 |
@param path: the base path to fetch data from. Note that you shouldn't |
|---|
| 249 |
put any trailing slashes in it, it will be added automatically in |
|---|
| 250 |
request. For example, if you put B{/foo}, a request on B{/bar} will |
|---|
| 251 |
be proxied to B{/foo/bar}. Any required encoding of special |
|---|
| 252 |
characters (such as " " or "/") should have been done already. |
|---|
| 253 |
|
|---|
| 254 |
@type path: C{str} |
|---|
| 255 |
""" |
|---|
| 256 |
Resource.__init__(self) |
|---|
| 257 |
self.host = host |
|---|
| 258 |
self.port = port |
|---|
| 259 |
self.path = path |
|---|
| 260 |
self.reactor = reactor |
|---|
| 261 |
|
|---|
| 262 |
|
|---|
| 263 |
def getChild(self, path, request): |
|---|
| 264 |
""" |
|---|
| 265 |
Create and return a proxy resource with the same proxy configuration |
|---|
| 266 |
as this one, except that its path also contains the segment given by |
|---|
| 267 |
C{path} at the end. |
|---|
| 268 |
""" |
|---|
| 269 |
return ReverseProxyResource( |
|---|
| 270 |
self.host, self.port, self.path + '/' + urlquote(path, safe="")) |
|---|
| 271 |
|
|---|
| 272 |
|
|---|
| 273 |
def render(self, request): |
|---|
| 274 |
""" |
|---|
| 275 |
Render a request by forwarding it to the proxied server. |
|---|
| 276 |
""" |
|---|
| 277 |
|
|---|
| 278 |
|
|---|
| 279 |
if self.port == 80: |
|---|
| 280 |
host = self.host |
|---|
| 281 |
else: |
|---|
| 282 |
host = "%s:%d" % (self.host, self.port) |
|---|
| 283 |
request.received_headers['host'] = host |
|---|
| 284 |
request.content.seek(0, 0) |
|---|
| 285 |
qs = urlparse.urlparse(request.uri)[4] |
|---|
| 286 |
if qs: |
|---|
| 287 |
rest = self.path + '?' + qs |
|---|
| 288 |
else: |
|---|
| 289 |
rest = self.path |
|---|
| 290 |
clientFactory = self.proxyClientFactoryClass( |
|---|
| 291 |
request.method, rest, request.clientproto, |
|---|
| 292 |
request.getAllHeaders(), request.content.read(), request) |
|---|
| 293 |
self.reactor.connectTCP(self.host, self.port, clientFactory) |
|---|
| 294 |
return NOT_DONE_YET |
|---|