requirement.txt

lxml

test.py

#!/usr/bin/env python
# coding: utf-8

import re
import lxml.html
import time
import urllib
import threading
import webbrowser
import SocketServer
import SimpleHTTPServer
from optparse import OptionParser
from StringIO import StringIO


def get_options():
    parser = OptionParser()
    parser.add_option("-p", "--port", dest="port", help="Port. Default is 12345", metavar="PORT",
                      default=12345, type='int')
    parser.add_option("-s", "--site", dest="site", help="Site. Default is 'http://gist.github.com'",
                      metavar="SITE", default='http://gist.github.com')
    return parser.parse_args()[0]


def replacer(text):
    def repl(m):
        word = m.groups()[0]
        if len(word) == 6:
            return word + u'™'
        return word
    return re.sub(u'([\w\-]*)', repl, text, flags=re.U | re.M)


def serve(port, site):
    class Proxy(SimpleHTTPServer.SimpleHTTPRequestHandler):
        def do_GET(self):
            page_source = urllib.urlopen(site + self.path).read()
            if '</html>' in page_source.lower():
                try:
                    doc = lxml.html.document_fromstring(page_source.decode('utf-8'))
                    for tag in doc.xpath('//*'):
                        tag.text = replacer(tag.text) if tag.text else tag.text
                        tag.tail = replacer(tag.tail) if tag.tail else tag.tail
                    new_page_source = lxml.html.tostring(doc, encoding='unicode')
                    ifile = StringIO(new_page_source.encode('utf-8'))
                except UnicodeError:
                    ifile = StringIO(page_source)
            else:
                ifile = StringIO(page_source)
            self.copyfile(ifile, self.wfile)

    httpd = SocketServer.ForkingTCPServer(('', port), Proxy)
    httpd.serve_forever()


def open_browser(url):
    time.sleep(3)
    webbrowser.open_new(url)

if __name__ == "__main__":
    options = get_options()
    start_url = 'http://127.0.0.1:{}/anonymous/06e0bd519490c8f03404'.format(options.port)
    thread = threading.Thread(target=open_browser, args=(start_url,))
    thread.start()
    serve(options.port, options.site)