A transparent caching proxy server built using WSGI.

caching-proxy.py

#!/usr/bin/env python

"""A transparent caching proxy.

To use, simply create a directory and run this script a la:

    ./caching-proxy.py cache

Then point your browser's HTTP proxy at localhost:8888.

Everything requested will be cached (regardless of no-cache headers).

To clear the cache, simply nuke the contents of the folder.
"""


import os
import sys
import hashlib
import httplib


from wsgiref.simple_server import make_server



def filename(method, path):
    return os.path.join(sys.argv[1], hashlib.sha256(method + ' ' + path).hexdigest())


def proxy(environ, start_response):
    fn = filename(environ['REQUEST_METHOD'], environ['PATH_INFO'])
    
    status = '200 OK'
    headers = []
    body = None
    
    if os.path.exists(fn):
        print "Cache hit:", fn
        data = open(fn, 'r').readlines()
        
        status = data[0].strip()
        
        j = 0
        for i, header in enumerate(data[1:]):
            if header.strip() == "":
                j = i + 1
                break
            
            headers.append(tuple(header.strip().split(':', 1)))
        
        body = data[j+1:]
        
    else:
        print "Cache miss:", fn
        fh = open(fn, 'w')
        
        rheaders = {}
        
        for i, j in environ.iteritems():
            if i.startswith("HTTP_"):
                rheaders[i[5:].lower().replace('_', '-')] = j
        
        conn = httplib.HTTPConnection(environ['PATH_INFO'][7:].split('/', 1)[0] + ':80')
        conn.request(environ['REQUEST_METHOD'], '/' + environ['PATH_INFO'][7:].split('/', 1)[1], None, rheaders)
        response = conn.getresponse()
        status = "%s %s" % (response.status, response.reason)
        headers = [(i, j) for i, j in response.getheaders() if i not in ('connection', 'keep-alive', 'transfer-encoding')]
        body = [response.read()]
        
        fh.write("%s\n" % (status, ))
        fh.write("%s\n\n" % ("\n".join([":".join((i, j)) for i, j in headers])))
        fh.write(body[0])
        
        fh.close()
    
    start_response(status, headers)
    return body


httpd = make_server('', 8888, proxy)
print "Serving on port 127.0.0.1:8888..."

# Serve until process is killed
httpd.serve_forever()