#! /usr/bin/env python3

from http.client import HTTPConnection, HTTPSConnection
from urllib.parse import urlsplit, parse_qsl, urlencode, quote, quote_plus
import sys
import urllib.request

url = sys.argv[1]

split_results = urlsplit(url)
protocol, netloc, path, query, tag = split_results
user = split_results.username
password = split_results.password
host = split_results.hostname
port = split_results.port
qlist = parse_qsl(query)

url = protocol + "://"
if user:
    url += quote(user)
    if password:
        url += ':' + quote(password)
    url += '@'
if host:
    host = host.encode('idna').decode('ascii')
    url += host
    if port:
        url += ':%d' % port
if path:
    if protocol == "file":
        url += quote(path)
    else:
        url += quote(path)
if query:
    url += '?' + urlencode(qlist)
if tag:
    url += '#' + quote_plus(tag)

if protocol == "https":
    server = HTTPSConnection(host, port)
else:
    server = HTTPConnection(host, port)
server.set_debuglevel(1)
server.putrequest("HEAD", path)

# I remember seeing some sites that return broken HTML or even HTTP response
# without "compatible" user agent; I don't know if such sites are still around,
# but this header doesn't cause any harm so I'd better continue to use it.
# UPDATE: I saw a number of sites that forbid "Mozilla compatible"
urllib_version = urllib.request.__version__
client_version = "Python-urllib/%s" % urllib_version
server.putheader('User-agent', client_version)

server.putheader('Accept-Charset', "koi8-r;q=1.0")
server.endheaders()

server.getresponse()