Lib/httplib.py

   1 # HTTP client class
   2 #
   3 # See the following document for a tentative protocol description:
   4 #     Hypertext Transfer Protocol (HTTP)        Tim Berners-Lee, CERN
   5 #     Internet Draft                                       5 Nov 1993
   6 #     draft-ietf-iiir-http-00.txt                  Expires 5 May 1994
   7 #
   8 # Example:
   9 #
  10 # >>> from httplib import HTTP
  11 # >>> h = HTTP('www.cwi.nl')
  12 # >>> h.putreqest('GET', '/index.html')
  13 # >>> h.putheader('Accept', 'text/html')
  14 # >>> h.putheader('Accept', 'text/plain')
  15 # >>> h.endheaders()
  16 # >>> errcode, errmsg, headers = h.getreply()
  17 # >>> if errcode == 200:
  18 # ...     f = h.getfile()
  19 # ...     print f.read() # Print the raw HTML
  20 # ...
  21 # <TITLE>Home Page of CWI, Amsterdam</TITLE>
  22 # [...many more lines...]
  23 # >>>
  24 #
  25 # Note that an HTTP object is used for a single request -- to issue a
  26 # second request to the same server, you create a new HTTP object.
  27 # (This is in accordance with the protocol, which uses a new TCP
  28 # connection for each request.)
  29
  30
  31 import os
  32 import socket
  33 import string
  34 import regex
  35 import regsub
  36 import rfc822
  37
  38 HTTP_VERSION = 'HTTP/1.0'
  39 HTTP_PORT = 80
  40
  41 replypat = regsub.gsub('\\.', '\\\\.', HTTP_VERSION) + \
  42           '[ \t]+\([0-9][0-9][0-9]\)\(.*\)'
  43 replyprog = regex.compile(replypat)
  44
  45 class HTTP:
  46
  47         def __init__(self, host = '', port = 0):
  48                 self.debuglevel = 0
  49                 if host: self.connect(host, port)
  50
  51         def set_debuglevel(self, debuglevel):
  52                 self.debuglevel = debuglevel
  53
  54         def connect(self, host, port = 0):
  55                 if not port:
  56                         i = string.find(host, ':')
  57                         if i >= 0:
  58                                 host, port = host[:i], host[i+1:]
  59                                 try: port = string.atoi(port)
  60                                 except string.atoi_error: pass
  61                 if not port: port = HTTP_PORT
  62                 self.sock = socket.socket(socket.AF_INET, socket.SOCK_STREAM)
  63                 if self.debuglevel > 0: print 'connect:', (host, port)
  64                 self.sock.connect(host, port)
  65
  66         def send(self, str):
  67                 if self.debuglevel > 0: print 'send:', `str`
  68                 self.sock.send(str)
  69
  70         def putrequest(self, request, selector):
  71                 if not selector: selector = '/'
  72                 str = '%s %s %s\r\n' % (request, selector, HTTP_VERSION)
  73                 self.send(str)
  74
  75         def putheader(self, header, *args):
  76                 str = '%s: %s\r\n' % (header, string.joinfields(args,'\r\n\t'))
  77                 self.send(str)
  78
  79         def endheaders(self):
  80                 self.send('\r\n')
  81
  82         def getreply(self):
  83                 self.file = self.sock.makefile('r')
  84                 self.sock = None
  85                 line = self.file.readline()
  86                 if self.debuglevel > 0: print 'reply:', `line`
  87                 if replyprog.match(line) < 0:
  88                         self.headers = None
  89                         return -1, line, self.headers
  90                 errcode, errmsg = replyprog.group(1, 2)
  91                 errcode = string.atoi(errcode)
  92                 errmsg = string.strip(errmsg)
  93                 self.headers = rfc822.Message(self.file, 0)
  94                 return errcode, errmsg, self.headers
  95
  96         def getfile(self):
  97                 return self.file
  98
  99
 100 def test():
 101         import sys
 102         import getopt
 103         opts, args = getopt.getopt(sys.argv[1:], 'd')
 104         dl = 0
 105         for o, a in opts:
 106                 if o == '-d': dl = dl + 1
 107         host = 'www.cwi.nl:80'
 108         selector = '/index.html'
 109         if args[0:]: host = args[0]
 110         if args[1:]: selector = args[1]
 111         h = HTTP()
 112         h.set_debuglevel(dl)
 113         h.connect(host)
 114         h.putrequest('GET', selector)
 115         errcode, errmsg, headers = h.getreply()
 116         print 'errcode =', errcode
 117         print 'headers =', headers
 118         print 'errmsg  =', errmsg
 119         if headers:
 120                 for header in headers.headers: print string.strip(header)
 121         print h.getfile().read()
 122
 123 if __name__ == '__main__':
 124         test()