blob: 6965fbdb8a94d16088312a0bdc20f04cf05ac873 [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
3import string
4
5# 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
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)
Guido van Rossum93a7c0f2000-03-28 21:45:46 +000069 s.connect((host, port))
Guido van Rossum54f22ed2000-02-04 15:10:34 +000070 s.send(selector + CRLF)
71 s.shutdown(1)
72 return s.makefile('rb')
Guido van Rossum23acc951994-02-21 16:36:04 +000073
Guido van Rossum2922c6d1994-05-06 14:28:19 +000074def send_query(selector, query, host, port = 0):
Guido van Rossum54f22ed2000-02-04 15:10:34 +000075 """Send a selector and a query string."""
76 return send_selector(selector + '\t' + query, host, port)
Guido van Rossum23acc951994-02-21 16:36:04 +000077
Guido van Rossumd2dd9a81998-01-19 21:59:48 +000078def path_to_selector(path):
Guido van Rossum54f22ed2000-02-04 15:10:34 +000079 """Takes a path as returned by urlparse and returns the appropriate selector."""
80 if path=="/":
81 return "/"
82 else:
83 return path[2:] # Cuts initial slash and data type identifier
Guido van Rossumd2dd9a81998-01-19 21:59:48 +000084
Guido van Rossumd2dd9a81998-01-19 21:59:48 +000085def path_to_datatype_name(path):
Guido van Rossum54f22ed2000-02-04 15:10:34 +000086 """Takes a path as returned by urlparse and maps it to a string.
87 See section 3.4 of RFC 1738 for details."""
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])
Guido van Rossum8ca84201998-03-26 20:56:10 +000093
Guido van Rossum23acc951994-02-21 16:36:04 +000094# The following functions interpret the data returned by the gopher
95# server according to the expected type, e.g. textfile or directory
96
Guido van Rossum23acc951994-02-21 16:36:04 +000097def get_directory(f):
Guido van Rossum54f22ed2000-02-04 15:10:34 +000098 """Get a directory in the form of a list of entries."""
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
Guido van Rossum23acc951994-02-21 16:36:04 +0000129
Guido van Rossum23acc951994-02-21 16:36:04 +0000130def get_textfile(f):
Guido van Rossum54f22ed2000-02-04 15:10:34 +0000131 """Get a text file as a list of lines, with trailing CRLF stripped."""
132 list = []
133 get_alt_textfile(f, list.append)
134 return list
Guido van Rossum23acc951994-02-21 16:36:04 +0000135
Guido van Rossum23acc951994-02-21 16:36:04 +0000136def get_alt_textfile(f, func):
Guido van Rossum54f22ed2000-02-04 15:10:34 +0000137 """Get a text file and pass each line to a function, with trailing CRLF stripped."""
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)
Guido van Rossum23acc951994-02-21 16:36:04 +0000152
Guido van Rossum23acc951994-02-21 16:36:04 +0000153def get_binary(f):
Guido van Rossum54f22ed2000-02-04 15:10:34 +0000154 """Get a binary file as one solid data block."""
155 data = f.read()
156 return data
Guido van Rossum23acc951994-02-21 16:36:04 +0000157
Guido van Rossum23acc951994-02-21 16:36:04 +0000158def get_alt_binary(f, func, blocksize):
Guido van Rossum54f22ed2000-02-04 15:10:34 +0000159 """Get a binary file and pass each block to a function."""
160 while 1:
161 data = f.read(blocksize)
162 if not data:
163 break
164 func(data)
Guido van Rossum23acc951994-02-21 16:36:04 +0000165
Guido van Rossum23acc951994-02-21 16:36:04 +0000166def test():
Guido van Rossum54f22ed2000-02-04 15:10:34 +0000167 """Trivial test program."""
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]
Guido van Rossum23acc951994-02-21 16:36:04 +0000205
206# Run the test when run as script
207if __name__ == '__main__':
Guido van Rossum54f22ed2000-02-04 15:10:34 +0000208 test()