Lib/gopherlib.py

   1 # Gopher protocol client interface
   2
   3 import string
   4
   5 # Default selector, host and port
   6 DEF_SELECTOR = '1/'
   7 DEF_HOST     = 'gopher.micro.umn.edu'
   8 DEF_PORT     = 70
   9
  10 # Recognized file types
  11 A_TEXT       = '0'
  12 A_MENU       = '1'
  13 A_CSO        = '2'
  14 A_ERROR      = '3'
  15 A_MACBINHEX  = '4'
  16 A_PCBINHEX   = '5'
  17 A_UUENCODED  = '6'
  18 A_INDEX      = '7'
  19 A_TELNET     = '8'
  20 A_BINARY     = '9'
  21 A_DUPLICATE  = '+'
  22 A_SOUND      = 's'
  23 A_EVENT      = 'e'
  24 A_CALENDAR   = 'c'
  25 A_HTML       = 'h'
  26 A_TN3270     = 'T'
  27 A_MIME       = 'M'
  28 A_IMAGE      = 'I'
  29 A_WHOIS      = 'w'
  30 A_QUERY      = 'q'
  31 A_GIF        = 'g'
  32 A_HTML       = 'h'                      # HTML file
  33 A_WWW        = 'w'                      # WWW address
  34 A_PLUS_IMAGE = ':'
  35 A_PLUS_MOVIE = ';'
  36 A_PLUS_SOUND = '<'
  37
  38
  39 # Function mapping all file types to strings; unknown types become TYPE='x'
  40 _names = dir()
  41 _type_to_name_map = None
  42 def type_to_name(gtype):
  43         global _type_to_name_map
  44         if not _type_to_name_map:
  45                 for name in _names:
  46                         if name[:2] == 'A_':
  47                                 _type_to_name_map[eval(name)] = name[2:]
  48         if _type_to_name_map.has_key(gtype):
  49                 return _type_to_name_map[gtype]
  50         return 'TYPE=' + `gtype`
  51
  52 # Names for characters and strings
  53 CRLF = '\r\n'
  54 TAB = '\t'
  55
  56 # Send a selector to a given host and port, return a file with the reply
  57 def send_selector(selector, host, port = 0):
  58         import socket
  59         import string
  60         if not port:
  61                 i = string.find(host, ':')
  62                 if i >= 0:
  63                         host, port = host[:i], string.atoi(host[i+1:])
  64         if not port:
  65                 port = DEF_PORT
  66         elif type(port) == type(''):
  67                 port = string.atoi(port)
  68         s = socket.socket(socket.AF_INET, socket.SOCK_STREAM)
  69         s.connect(host, port)
  70         s.send(selector + CRLF)
  71         s.shutdown(1)
  72         return s.makefile('r')
  73
  74 # Send a selector and a query string
  75 def send_query(selector, query, host, port = 0):
  76         return send_selector(selector + '\t' + query, host, port)
  77
  78 # The following functions interpret the data returned by the gopher
  79 # server according to the expected type, e.g. textfile or directory
  80
  81 # Get a directory in the form of a list of entries
  82 def get_directory(f):
  83         import string
  84         list = []
  85         while 1:
  86                 line = f.readline()
  87                 if not line:
  88                         print '(Unexpected EOF from server)'
  89                         break
  90                 if line[-2:] == CRLF:
  91                         line = line[:-2]
  92                 elif line[-1:] in CRLF:
  93                         line = line[:-1]
  94                 if line == '.':
  95                         break
  96                 if not line:
  97                         print '(Empty line from server)'
  98                         continue
  99                 gtype = line[0]
 100                 parts = string.splitfields(line[1:], TAB)
 101                 if len(parts) < 4:
 102                         print '(Bad line from server:', `line`, ')'
 103                         continue
 104                 if len(parts) > 4:
 105                         if parts[4:] != ['+']:
 106                             print '(Extra info from server:', parts[4:], ')'
 107                 else:
 108                         parts.append('')
 109                 parts.insert(0, gtype)
 110                 list.append(parts)
 111         return list
 112
 113 # Get a text file as a list of lines, with trailing CRLF stripped
 114 def get_textfile(f):
 115         list = []
 116         get_alt_textfile(f, list.append)
 117         return list
 118
 119 # Get a text file and pass each line to a function, with trailing CRLF stripped
 120 def get_alt_textfile(f, func):
 121         while 1:
 122                 line = f.readline()
 123                 if not line:
 124                         print '(Unexpected EOF from server)'
 125                         break
 126                 if line[-2:] == CRLF:
 127                         line = line[:-2]
 128                 elif line[-1:] in CRLF:
 129                         line = line[:-1]
 130                 if line == '.':
 131                         break
 132                 if line[:2] == '..':
 133                         line = line[1:]
 134                 func(line)
 135
 136 # Get a binary file as one solid data block
 137 def get_binary(f):
 138         data = f.read()
 139         return data
 140
 141 # Get a binary file and pass each block to a function
 142 def get_alt_binary(f, func, blocksize):
 143         while 1:
 144                 data = f.read(blocksize)
 145                 if not data:
 146                         break
 147                 func(data)
 148
 149 # Trivial test program
 150 def test():
 151         import sys
 152         import getopt
 153         opts, args = getopt.getopt(sys.argv[1:], '')
 154         selector = DEF_SELECTOR
 155         type = selector[0]
 156         host = DEF_HOST
 157         port = DEF_PORT
 158         if args:
 159                 host = args[0]
 160                 args = args[1:]
 161         if args:
 162                 type = args[0]
 163                 args = args[1:]
 164                 if len(type) > 1:
 165                         type, selector = type[0], type
 166                 else:
 167                         selector = ''
 168                         if args:
 169                                 selector = args[0]
 170                                 args = args[1:]
 171                 query = ''
 172                 if args:
 173                         query = args[0]
 174                         args = args[1:]
 175         if type == A_INDEX:
 176                 f = send_query(selector, query, host)
 177         else:
 178                 f = send_selector(selector, host)
 179         if type == A_TEXT:
 180                 list = get_textfile(f)
 181                 for item in list: print item
 182         elif type in (A_MENU, A_INDEX):
 183                 list = get_directory(f)
 184                 for item in list: print item
 185         else:
 186                 data = get_binary(f)
 187                 print 'binary data:', len(data), 'bytes:', `data[:100]`[:40]
 188
 189 # Run the test when run as script
 190 if __name__ == '__main__':
 191         test()