When an HTTP proxy forwards traffic to a webserver, the server sees the proxy IP address rather than the original client IP address. Since the server may need the IP address for logging or authentication purposes, many HTTP proxies add an X-Forwarded-For header indicating the original client IP address.
As WSGI middleware, WsgiUnproxy sits between the WSGI server and your WSGI application. Before your application sees a request, WsgiUnproxy removes the X-Forwarded-For header and reinstates the client IP address, yielding a request that looks like it was never proxied to begin with.
Since anyone can add an X-Forwarded-For header, WsgiUnproxy only uses the header if it comes from a trusted proxy IP addresses.
Example WSGI application
from wsgiunproxy import unproxy @unproxy(trusted_proxies=[ '220.127.116.11', '18.104.22.168' ]) def application(environ, start_response): start_response('200 OK', ) return ['Your IP address is %s.' % environ.get('REMOTE_ADDR')]
Use with Paste Deployment
WsgiUnproxy can be used in a Paste Deployment pipeline:
[pipeline:main] pipeline = WsgiUnproxy MyApp [filter:WsgiUnproxy] use = egg:WsgiUnproxy trusted_proxies = 22.214.171.124, 126.96.36.199
If you need to specify a lot of trusted proxies (such as a whole subnet), you don't have to use give trusted_proxies as a list. All that WsgiUnproxy asks is that trusted_proxies supports the in operator (e.g. by implementing __contains__).
To the extent possible under law, the author has waived all copyright and related or neighboring rights to WsgiUnproxy.
For more information see: http://creativecommons.org/publicdomain/zero/1.0/