tproxy is a simple TCP routing proxy (layer 7) built on Gevent that lets you configure the routine logic in Python. It's heavily inspired from proxy machine but have some unique features like the pre-fork worker model borrowed to Gunicorn.

tproxy requires Python 2.x >= 2.5. Python 3.x support is planned.

$ pip install gevent
$ pip install tproxy

To install from source:

$ git clone git://github.com/benoitc/tproxy.git
$ cd tproxy
$ pip install -r requirements.txt
$ python setup.py install

Test your installation by running the command line:

$ tproxy examples/transparent.py

And go on http://127.0.0.1:5000 , you should see the google homepage.

$ tproxy -h

Usage: tproxy [OPTIONS] script_path

Options:
  --version                     show program's version number and exit
  -h, --help                    show this help message and exit
  --log-file=FILE               The log file to write to. [-]
  --log-level=LEVEL             The granularity of log outputs. [info]
  --log-config=FILE             The log config file to use. [None]
  -n STRING, --name=STRING      A base to use with setproctitle for process naming.
                                [None]
  -D, --daemon                  Daemonize the tproxy process. [False]
  -p FILE, --pid=FILE           A filename to use for the PID file. [None]
  -u USER, --user=USER          Switch worker processes to run as this user. [501]
  -g GROUP, --group=GROUP
                                Switch worker process to run as this group. [20]
  -m INT, --umask=INT           A bit mask for the file mode on files written by
                                tproxy. [0]
  -b ADDRESS, --bind=ADDRESS    The socket to bind. [127.0.0.1:8000]
  --backlog=INT                 The maximum number of pending connections.     [2048]
  --ssl-keyfile=STRING          Ssl key file [None]
  --ssl-certfile=STRING         Ssl ca certs file. contains concatenated
                                "certification [None]
  --ssl-ca-certs=STRING         Ssl ca certs file. contains concatenated
                                "certification [None]
  --ssl-cert-reqs=INT           Specifies whether a certificate is required from the
                                other [0]
  -w INT, --workers=INT         The number of worker process for handling requests. [1]
  --worker-connections=INT      The maximum number of simultaneous clients per worker.
                                [1000]
  -t INT, --timeout=INT         Workers silent for more than this many seconds are
                                killed and restarted. [30]

QUIT    -   Graceful shutdown. Stop accepting connections immediatly
            and wait until all connections close

TERM    -   Fast shutdown. Stop accepting and close all conections
            after 10s.
INT     -   Same as TERM

HUP     -   Graceful reloading. Reload all workers with the new code
            in your routing script.

USR2    -   Upgrade tproxy on the fly

TTIN    -   Increase the number of worker from 1

TTOU    -   Decrease the number of worker from 1

import re
re_host = re.compile("Host:\s*(.*)\r\n")

class CouchDBRouter(object):
    # look at the routing table and return a couchdb node to use
    def lookup(self, name):
        """ do something """

router = CouchDBRouter()

# Perform content-aware routing based on the stream data. Here, the
# Host header information from the HTTP protocol is parsed to find the
# username and a lookup routine is run on the name to find the correct
# couchdb node. If no match can be made yet, do nothing with the
# connection. (make your own couchone server...)

def proxy(data):
    matches = re_host.findall(data)
    if matches:
        host = router.lookup(matches.pop())
        return {"remote": host}
    return None

import socket
import struct

def proxy(data):
    if len(data) < 9:
        return

    command = ord(data[1])
    ip, port = socket.inet_ntoa(data[4:8]), struct.unpack(">H", data[2:4])[0]
    idx = data.index("\0")
    userid = data[8:idx]

    if command == 1: #connect
        return dict(remote="%s:%s" % (ip, port),
                reply="\0\x5a\0\0\0\0\0\0",
                data=data[idx:])
    else:
        return {"close": "\0\x5b\0\0\0\0\0\0"}

import os

WELCOME_FILE = os.path.join(os.path.dirname(__file__), "welcome.txt")

def proxy(data):
    fno = os.open(WELCOME_FILE, os.O_RDONLY)
    return {
            "file": fno,
            "reply": "HTTP/1.1 200 OK\r\n\r\n"
           }

• { "remote:": string or tuple } - String is the host:port of the server that will be proxied.
• { "remote": String, "data": String} - Same as above, but send the given data instead.
• { "remote": String, "data": String, "reply": String} - Same as above, but reply with given data back to the client
• None - Do nothing.
• { "close": True } - Close the connection.
• { "close": String } - Close the connection after sending the String.
• { "file": String } - Return a file specify by the file path and close the connection.
• { "file": String, "reply": String } - Return a file specify by the file path and close the connection.
• { "file": Int, "reply": String} - Same as above but reply with given data back to the client
• { "file": Int } - Return a file specify by its file descriptor
• { "file": Int, "reply": String} - Same as above but reply with given data back to the client

If sendfile API available it will be used to send a file with "file" command.

The file command can have 2 optionnnal parameters:

• offset: argument specifies where to begin in the file.
• nbytes: specifies how many bytes of the file should be sent

To handle ssl for remote connection you can add these optionals arguments:

• ssl: True or False, if you want to connect with ssl
• ssl_args: dict, optionals ssl arguments. Read the ssl documentation for more informations about them.

You can easily handling error by adding a proxy_error function in your script:

def proxy_error(client, e):
    pass

This function get the ClientConnection instance (current connection) as first arguments and the error exception in second argument.

Main goal of tproxy is to allows you to route transparently tcp to your applications. But some case you want to do more. For example you need in HTTP 1.1 to change the Host header to make sure remote HTTP server will know what to do if uses virtual hosting.

To do that, add a rewrite_request function in your function to simply rewrite clienrt request and rewrite_response to rewrite the remote response. Both functions take a tproxy.rewrite.RewriteIO instance which is based on io.RawIOBase class.

See the httprewrite.py example for an example of HTTP rewrite.

2011 (c) Benoît Chesneau <[email protected]>