Lib/dos-8x3/gopherli.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 = {}
  42 def type_to_name(gtype):
  43         global _type_to_name_map
  44         if _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('rb')
  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 # Takes a path as returned by urlparse and returns the appropriate selector
  79 def path_to_selector(path):
  80         if path=="/":
  81                 return "/"
  82         else:
  83                 return path[2:] # Cuts initial slash and data type identifier
  84
  85 # Takes a path as returned by urlparse and maps it to a string
  86 # See section 3.4 of RFC 1738 for details
  87 def path_to_datatype_name(path):
  88         if path=="/":
  89                 # No way to tell, although "INDEX" is likely
  90                 return "TYPE='unknown'"
  91         else:
  92                 return type_to_name(path[1])
  93
  94 # The following functions interpret the data returned by the gopher
  95 # server according to the expected type, e.g. textfile or directory
  96
  97 # Get a directory in the form of a list of entries
  98 def get_directory(f):
  99         import string
 100         list = []
 101         while 1:
 102                 line = f.readline()
 103                 if not line:
 104                         print '(Unexpected EOF from server)'
 105                         break
 106                 if line[-2:] == CRLF:
 107                         line = line[:-2]
 108                 elif line[-1:] in CRLF:
 109                         line = line[:-1]
 110                 if line == '.':
 111                         break
 112                 if not line:
 113                         print '(Empty line from server)'
 114                         continue
 115                 gtype = line[0]
 116                 parts = string.splitfields(line[1:], TAB)
 117                 if len(parts) < 4:
 118                         print '(Bad line from server:', `line`, ')'
 119                         continue
 120                 if len(parts) > 4:
 121                         if parts[4:] != ['+']:
 122                                 print '(Extra info from server:',
 123                                 print parts[4:], ')'
 124                 else:
 125                         parts.append('')
 126                 parts.insert(0, gtype)
 127                 list.append(parts)
 128         return list
 129
 130 # Get a text file as a list of lines, with trailing CRLF stripped
 131 def get_textfile(f):
 132         list = []
 133         get_alt_textfile(f, list.append)
 134         return list
 135
 136 # Get a text file and pass each line to a function, with trailing CRLF stripped
 137 def get_alt_textfile(f, func):
 138         while 1:
 139                 line = f.readline()
 140                 if not line:
 141                         print '(Unexpected EOF from server)'
 142                         break
 143                 if line[-2:] == CRLF:
 144                         line = line[:-2]
 145                 elif line[-1:] in CRLF:
 146                         line = line[:-1]
 147                 if line == '.':
 148                         break
 149                 if line[:2] == '..':
 150                         line = line[1:]
 151                 func(line)
 152
 153 # Get a binary file as one solid data block
 154 def get_binary(f):
 155         data = f.read()
 156         return data
 157
 158 # Get a binary file and pass each block to a function
 159 def get_alt_binary(f, func, blocksize):
 160         while 1:
 161                 data = f.read(blocksize)
 162                 if not data:
 163                         break
 164                 func(data)
 165
 166 # Trivial test program
 167 def test():
 168         import sys
 169         import getopt
 170         opts, args = getopt.getopt(sys.argv[1:], '')
 171         selector = DEF_SELECTOR
 172         type = selector[0]
 173         host = DEF_HOST
 174         port = DEF_PORT
 175         if args:
 176                 host = args[0]
 177                 args = args[1:]
 178         if args:
 179                 type = args[0]
 180                 args = args[1:]
 181                 if len(type) > 1:
 182                         type, selector = type[0], type
 183                 else:
 184                         selector = ''
 185                         if args:
 186                                 selector = args[0]
 187                                 args = args[1:]
 188                 query = ''
 189                 if args:
 190                         query = args[0]
 191                         args = args[1:]
 192         if type == A_INDEX:
 193                 f = send_query(selector, query, host)
 194         else:
 195                 f = send_selector(selector, host)
 196         if type == A_TEXT:
 197                 list = get_textfile(f)
 198                 for item in list: print item
 199         elif type in (A_MENU, A_INDEX):
 200                 list = get_directory(f)
 201                 for item in list: print item
 202         else:
 203                 data = get_binary(f)
 204                 print 'binary data:', len(data), 'bytes:', `data[:100]`[:40]
 205
 206 # Run the test when run as script
 207 if __name__ == '__main__':
 208         test()