blob: 8dea5660e696a282468f46954bb5eb3d247ccd3b [file] [log] [blame]
Guido van Rossum54f22ed2000-02-04 15:10:34 +00001"""Gopher protocol client interface."""
Guido van Rossum23acc951994-02-21 16:36:04 +00002
Skip Montanaro2dd42762001-01-23 15:35:05 +00003__all__ = ["send_selector","send_query"]
4
Guido van Rossum23acc951994-02-21 16:36:04 +00005# Default selector, host and port
6DEF_SELECTOR = '1/'
7DEF_HOST = 'gopher.micro.umn.edu'
8DEF_PORT = 70
9
10# Recognized file types
11A_TEXT = '0'
12A_MENU = '1'
13A_CSO = '2'
14A_ERROR = '3'
15A_MACBINHEX = '4'
16A_PCBINHEX = '5'
17A_UUENCODED = '6'
18A_INDEX = '7'
19A_TELNET = '8'
20A_BINARY = '9'
21A_DUPLICATE = '+'
22A_SOUND = 's'
23A_EVENT = 'e'
24A_CALENDAR = 'c'
25A_HTML = 'h'
26A_TN3270 = 'T'
27A_MIME = 'M'
28A_IMAGE = 'I'
29A_WHOIS = 'w'
30A_QUERY = 'q'
31A_GIF = 'g'
Guido van Rossum54f22ed2000-02-04 15:10:34 +000032A_HTML = 'h' # HTML file
33A_WWW = 'w' # WWW address
Guido van Rossum23acc951994-02-21 16:36:04 +000034A_PLUS_IMAGE = ':'
35A_PLUS_MOVIE = ';'
36A_PLUS_SOUND = '<'
37
38
Guido van Rossum23acc951994-02-21 16:36:04 +000039_names = dir()
Guido van Rossumd2dd9a81998-01-19 21:59:48 +000040_type_to_name_map = {}
Guido van Rossum23acc951994-02-21 16:36:04 +000041def type_to_name(gtype):
Guido van Rossum54f22ed2000-02-04 15:10:34 +000042 """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 _type_to_name_map.has_key(gtype):
49 return _type_to_name_map[gtype]
50 return 'TYPE=' + `gtype`
Guido van Rossum23acc951994-02-21 16:36:04 +000051
52# Names for characters and strings
53CRLF = '\r\n'
54TAB = '\t'
55
Guido van Rossum2922c6d1994-05-06 14:28:19 +000056def send_selector(selector, host, port = 0):
Guido van Rossum54f22ed2000-02-04 15:10:34 +000057 """Send a selector to a given host and port, return a file with the reply."""
58 import socket
Guido van Rossum54f22ed2000-02-04 15:10:34 +000059 if not port:
Eric S. Raymond19e6d622001-02-09 10:10:02 +000060 i = host.find(':')
Guido van Rossum54f22ed2000-02-04 15:10:34 +000061 if i >= 0:
Eric S. Raymond19e6d622001-02-09 10:10:02 +000062 host, port = host[:i], int(host[i+1:])
Guido van Rossum54f22ed2000-02-04 15:10:34 +000063 if not port:
64 port = DEF_PORT
65 elif type(port) == type(''):
Eric S. Raymond19e6d622001-02-09 10:10:02 +000066 port = int(port)
Guido van Rossum54f22ed2000-02-04 15:10:34 +000067 s = socket.socket(socket.AF_INET, socket.SOCK_STREAM)
Guido van Rossum93a7c0f2000-03-28 21:45:46 +000068 s.connect((host, port))
Guido van Rossum54f22ed2000-02-04 15:10:34 +000069 s.send(selector + CRLF)
70 s.shutdown(1)
71 return s.makefile('rb')
Guido van Rossum23acc951994-02-21 16:36:04 +000072
Guido van Rossum2922c6d1994-05-06 14:28:19 +000073def send_query(selector, query, host, port = 0):
Guido van Rossum54f22ed2000-02-04 15:10:34 +000074 """Send a selector and a query string."""
75 return send_selector(selector + '\t' + query, host, port)
Guido van Rossum23acc951994-02-21 16:36:04 +000076
Guido van Rossumd2dd9a81998-01-19 21:59:48 +000077def path_to_selector(path):
Guido van Rossum54f22ed2000-02-04 15:10:34 +000078 """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
Guido van Rossumd2dd9a81998-01-19 21:59:48 +000083
Guido van Rossumd2dd9a81998-01-19 21:59:48 +000084def path_to_datatype_name(path):
Guido van Rossum54f22ed2000-02-04 15:10:34 +000085 """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])
Guido van Rossum8ca84201998-03-26 20:56:10 +000092
Guido van Rossum23acc951994-02-21 16:36:04 +000093# The following functions interpret the data returned by the gopher
94# server according to the expected type, e.g. textfile or directory
95
Guido van Rossum23acc951994-02-21 16:36:04 +000096def get_directory(f):
Guido van Rossum54f22ed2000-02-04 15:10:34 +000097 """Get a directory in the form of a list of entries."""
Guido van Rossum54f22ed2000-02-04 15:10:34 +000098 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]
Eric S. Raymond19e6d622001-02-09 10:10:02 +0000114 parts = line[1:].split(TAB)
Guido van Rossum54f22ed2000-02-04 15:10:34 +0000115 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
Guido van Rossum23acc951994-02-21 16:36:04 +0000127
Guido van Rossum23acc951994-02-21 16:36:04 +0000128def get_textfile(f):
Guido van Rossum54f22ed2000-02-04 15:10:34 +0000129 """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
Guido van Rossum23acc951994-02-21 16:36:04 +0000133
Guido van Rossum23acc951994-02-21 16:36:04 +0000134def get_alt_textfile(f, func):
Guido van Rossum54f22ed2000-02-04 15:10:34 +0000135 """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)
Guido van Rossum23acc951994-02-21 16:36:04 +0000150
Guido van Rossum23acc951994-02-21 16:36:04 +0000151def get_binary(f):
Guido van Rossum54f22ed2000-02-04 15:10:34 +0000152 """Get a binary file as one solid data block."""
153 data = f.read()
154 return data
Guido van Rossum23acc951994-02-21 16:36:04 +0000155
Guido van Rossum23acc951994-02-21 16:36:04 +0000156def get_alt_binary(f, func, blocksize):
Guido van Rossum54f22ed2000-02-04 15:10:34 +0000157 """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)
Guido van Rossum23acc951994-02-21 16:36:04 +0000163
Guido van Rossum23acc951994-02-21 16:36:04 +0000164def test():
Guido van Rossum54f22ed2000-02-04 15:10:34 +0000165 """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
Guido van Rossum54f22ed2000-02-04 15:10:34 +0000172 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]
Guido van Rossum23acc951994-02-21 16:36:04 +0000202
203# Run the test when run as script
204if __name__ == '__main__':
Guido van Rossum54f22ed2000-02-04 15:10:34 +0000205 test()