bin/HEAD.py

   1 #! /usr/bin/env python3
   2
   3 try:
   4     PY2 = False
   5     from http.client import HTTPConnection
   6     from urllib.parse import parse_qsl, urlencode, \
   7         quote, quote_plus, unquote, unquote_plus, \
   8         splittype, splithost, splituser, splitpasswd, \
   9         splitport, splittag, splitquery
  10     import urllib.request
  11 except ImportError:
  12     PY2 = True
  13     from cgi import parse_qsl
  14     from urllib import urlencode, quote, quote_plus, unquote, unquote_plus, \
  15         splittype, splithost, splituser, splitpasswd, \
  16         splitport, splittag, splitquery
  17     from httplib import HTTPConnection
  18
  19 import sys
  20 import urllib
  21 from m_lib.defenc import default_encoding
  22
  23 url = sys.argv[1]
  24
  25 protocol, request = splittype(url)
  26 user, password, port = None, None, None
  27 host, path = splithost(request)
  28 if host:
  29     user, host = splituser(host)
  30     if user:
  31         user, password = splitpasswd(user)
  32     host, port = splitport(host)
  33     if port: port = int(port)
  34 path, tag = splittag(path)
  35 path, query = splitquery(path)
  36 path = unquote(path)
  37 if tag: tag = unquote_plus(tag)
  38
  39 if query:
  40     qlist = []
  41     for name, value in parse_qsl(query):
  42         qlist.append((name, value))
  43
  44 url = protocol + "://"
  45 if user:
  46     url += quote(user)
  47     if password:
  48         url += ':' + quote(password)
  49     url += '@'
  50 if host:
  51     if PY2:
  52         host = host.decode(default_encoding)
  53     host = host.encode('idna')
  54     if not PY2:
  55         host = host.decode('ascii')
  56     url += host
  57     if port:
  58         url += ':%d' % port
  59 if path:
  60     if protocol == "file":
  61         url += quote(path)
  62     else:
  63         url += quote(path)
  64 if query:
  65     url += '?' + urlencode(qlist)
  66 if tag:
  67     url += '#' + quote_plus(tag)
  68
  69 server = HTTPConnection(host, port)
  70 server.set_debuglevel(1)
  71
  72 server.putrequest("HEAD", path)
  73 if port:
  74     server.putheader("Host", '%s:%d' % (host, port))
  75 else:
  76     server.putheader("Host", host)
  77
  78 # I remember seeing some sites that return broken HTML or even HTTP response
  79 # without "compatible" user agent; I don't know if such sites are still around,
  80 # but this header doesn't cause any harm so I'd better continue to use it.
  81 # UPDATE: I saw a number of sites that forbid "Mozilla compatible"
  82 if PY2:
  83     urllib_version = urllib.__version__
  84 else:
  85     urllib_version = urllib.request.__version__
  86 client_version = "Python-urllib/%s" % urllib_version
  87 server.putheader('User-agent', client_version)
  88
  89 server.putheader('Accept-Charset', "koi8-r;q=1.0")
  90 server.endheaders()
  91
  92 server.getresponse()