Lib/httplib.py

   1 # HTTP client class
   2 #
   3 # See the following URL for a description of the HTTP/1.0 protocol:
   4 # http://www.w3.org/hypertext/WWW/Protocols/
   5 # (I actually implemented it from a much earlier draft.)
   6 #
   7 # Example:
   8 #
   9 # >>> from httplib import HTTP
  10 # >>> h = HTTP('www.python.org')
  11 # >>> h.putrequest('GET', '/index.html')
  12 # >>> h.putheader('Accept', 'text/html')
  13 # >>> h.putheader('Accept', 'text/plain')
  14 # >>> h.endheaders()
  15 # >>> errcode, errmsg, headers = h.getreply()
  16 # >>> if errcode == 200:
  17 # ...     f = h.getfile()
  18 # ...     print f.read() # Print the raw HTML
  19 # ...
  20 # <HEAD>
  21 # <TITLE>Python Language Home Page</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 mimetools
  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                 self.file = None
  50                 if host: self.connect(host, port)
  51
  52         def set_debuglevel(self, debuglevel):
  53                 self.debuglevel = debuglevel
  54
  55         def connect(self, host, port = 0):
  56                 if not port:
  57                         i = string.find(host, ':')
  58                         if i >= 0:
  59                                 host, port = host[:i], host[i+1:]
  60                                 try: port = string.atoi(port)
  61                                 except string.atoi_error:
  62                                     raise socket.error, "nonnumeric port"
  63                 if not port: port = HTTP_PORT
  64                 self.sock = socket.socket(socket.AF_INET, socket.SOCK_STREAM)
  65                 if self.debuglevel > 0: print 'connect:', (host, port)
  66                 self.sock.connect(host, port)
  67
  68         def send(self, str):
  69                 if self.debuglevel > 0: print 'send:', `str`
  70                 self.sock.send(str)
  71
  72         def putrequest(self, request, selector):
  73                 if not selector: selector = '/'
  74                 str = '%s %s %s\r\n' % (request, selector, HTTP_VERSION)
  75                 self.send(str)
  76
  77         def putheader(self, header, *args):
  78                 str = '%s: %s\r\n' % (header, string.joinfields(args,'\r\n\t'))
  79                 self.send(str)
  80
  81         def endheaders(self):
  82                 self.send('\r\n')
  83
  84         def getreply(self):
  85                 self.file = self.sock.makefile('rb')
  86                 self.sock = None
  87                 line = self.file.readline()
  88                 if self.debuglevel > 0: print 'reply:', `line`
  89                 if replyprog.match(line) < 0:
  90                         self.headers = None
  91                         return -1, line, self.headers
  92                 errcode, errmsg = replyprog.group(1, 2)
  93                 errcode = string.atoi(errcode)
  94                 errmsg = string.strip(errmsg)
  95                 self.headers = mimetools.Message(self.file, 0)
  96                 return errcode, errmsg, self.headers
  97
  98         def getfile(self):
  99                 return self.file
 100
 101         def close(self):
 102                 if self.file:
 103                         self.file.close()
 104                 self.file = None
 105
 106
 107 def test():
 108         import sys
 109         import getopt
 110         opts, args = getopt.getopt(sys.argv[1:], 'd')
 111         dl = 0
 112         for o, a in opts:
 113                 if o == '-d': dl = dl + 1
 114         host = 'www.python.org'
 115         selector = '/'
 116         if args[0:]: host = args[0]
 117         if args[1:]: selector = args[1]
 118         h = HTTP()
 119         h.set_debuglevel(dl)
 120         h.connect(host)
 121         h.putrequest('GET', selector)
 122         h.endheaders()
 123         errcode, errmsg, headers = h.getreply()
 124         print 'errcode =', errcode
 125         print 'errmsg  =', errmsg
 126         print
 127         if headers:
 128                 for header in headers.headers: print string.strip(header)
 129         print
 130         print h.getfile().read()
 131
 132
 133 if __name__ == '__main__':
 134         test()