Lib/gopherlib.py

   1 """Gopher protocol client interface."""
   2
   3 __all__ = ["send_selector","send_query"]
   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 _names = dir()
  40 _type_to_name_map = {}
  41 def type_to_name(gtype):
  42     """Map all file types to strings; unknown types become TYPE='x'."""
  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 gtype in _type_to_name_map:
  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 def send_selector(selector, host, port = 0):
  57     """Send a selector to a given host and port, return a file with the reply."""
  58     import socket
  59     if not port:
  60         i = host.find(':')
  61         if i >= 0:
  62             host, port = host[:i], int(host[i+1:])
  63     if not port:
  64         port = DEF_PORT
  65     elif type(port) == type(''):
  66         port = int(port)
  67     s = socket.socket(socket.AF_INET, socket.SOCK_STREAM)
  68     s.connect((host, port))
  69     s.sendall(selector + CRLF)
  70     s.shutdown(1)
  71     return s.makefile('rb')
  72
  73 def send_query(selector, query, host, port = 0):
  74     """Send a selector and a query string."""
  75     return send_selector(selector + '\t' + query, host, port)
  76
  77 def path_to_selector(path):
  78     """Takes a path as returned by urlparse and returns the appropriate selector."""
  79     if path=="/":
  80         return "/"
  81     else:
  82         return path[2:] # Cuts initial slash and data type identifier
  83
  84 def path_to_datatype_name(path):
  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     if path=="/":
  88         # No way to tell, although "INDEX" is likely
  89         return "TYPE='unknown'"
  90     else:
  91         return type_to_name(path[1])
  92
  93 # The following functions interpret the data returned by the gopher
  94 # server according to the expected type, e.g. textfile or directory
  95
  96 def get_directory(f):
  97     """Get a directory in the form of a list of entries."""
  98     list = []
  99     while 1:
 100         line = f.readline()
 101         if not line:
 102             print '(Unexpected EOF from server)'
 103             break
 104         if line[-2:] == CRLF:
 105             line = line[:-2]
 106         elif line[-1:] in CRLF:
 107             line = line[:-1]
 108         if line == '.':
 109             break
 110         if not line:
 111             print '(Empty line from server)'
 112             continue
 113         gtype = line[0]
 114         parts = line[1:].split(TAB)
 115         if len(parts) < 4:
 116             print '(Bad line from server:', `line`, ')'
 117             continue
 118         if len(parts) > 4:
 119             if parts[4:] != ['+']:
 120                 print '(Extra info from server:',
 121                 print parts[4:], ')'
 122         else:
 123             parts.append('')
 124         parts.insert(0, gtype)
 125         list.append(parts)
 126     return list
 127
 128 def get_textfile(f):
 129     """Get a text file as a list of lines, with trailing CRLF stripped."""
 130     list = []
 131     get_alt_textfile(f, list.append)
 132     return list
 133
 134 def get_alt_textfile(f, func):
 135     """Get a text file and pass each line to a function, with trailing CRLF stripped."""
 136     while 1:
 137         line = f.readline()
 138         if not line:
 139             print '(Unexpected EOF from server)'
 140             break
 141         if line[-2:] == CRLF:
 142             line = line[:-2]
 143         elif line[-1:] in CRLF:
 144             line = line[:-1]
 145         if line == '.':
 146             break
 147         if line[:2] == '..':
 148             line = line[1:]
 149         func(line)
 150
 151 def get_binary(f):
 152     """Get a binary file as one solid data block."""
 153     data = f.read()
 154     return data
 155
 156 def get_alt_binary(f, func, blocksize):
 157     """Get a binary file and pass each block to a function."""
 158     while 1:
 159         data = f.read(blocksize)
 160         if not data:
 161             break
 162         func(data)
 163
 164 def test():
 165     """Trivial test program."""
 166     import sys
 167     import getopt
 168     opts, args = getopt.getopt(sys.argv[1:], '')
 169     selector = DEF_SELECTOR
 170     type = selector[0]
 171     host = DEF_HOST
 172     if args:
 173         host = args[0]
 174         args = args[1:]
 175     if args:
 176         type = args[0]
 177         args = args[1:]
 178         if len(type) > 1:
 179             type, selector = type[0], type
 180         else:
 181             selector = ''
 182             if args:
 183                 selector = args[0]
 184                 args = args[1:]
 185         query = ''
 186         if args:
 187             query = args[0]
 188             args = args[1:]
 189     if type == A_INDEX:
 190         f = send_query(selector, query, host)
 191     else:
 192         f = send_selector(selector, host)
 193     if type == A_TEXT:
 194         list = get_textfile(f)
 195         for item in list: print item
 196     elif type in (A_MENU, A_INDEX):
 197         list = get_directory(f)
 198         for item in list: print item
 199     else:
 200         data = get_binary(f)
 201         print 'binary data:', len(data), 'bytes:', `data[:100]`[:40]
 202
 203 # Run the test when run as script
 204 if __name__ == '__main__':
 205     test()