Gitiles
Code Review Sign In
gerrit-public.fairphone.software / platform / external / python / cpython2 / 67d1981c5172e840ac3e2fc3505ce9c3abe5cd63 / . / Lib / test / test_urlparse.py
blob: e7d9e5a9164448d936a354f86eeeb24bddd2a611 [file] [log] [blame]
Skip Montanaro6ec967d2002-03-23 05:32:10 +0000[diff] [blame]1#! /usr/bin/env python
2
Barry Warsaw04f357c2002-07-23 19:04:11 +0000[diff] [blame]3from test import test_support
Skip Montanaro6ec967d2002-03-23 05:32:10 +0000[diff] [blame]4import unittest
Fred Drakea4d18a02001-01-05 05:57:04 +0000[diff] [blame]5import urlparse
6
Fred Drakea4d18a02001-01-05 05:57:04 +0000[diff] [blame]7RFC1808_BASE = "http://a/b/c/d;p?q#f"
Skip Montanaro6ec967d2002-03-23 05:32:10 +0000[diff] [blame]8RFC2396_BASE = "http://a/b/c/d;p?q"
Facundo Batista67d19812008-08-14 16:51:00 +0000[diff] [blame^]9RFC3986_BASE = "http://a/b/c/d;p?q"
Fred Drakea4d18a02001-01-05 05:57:04 +0000[diff] [blame]10
Skip Montanaro6ec967d2002-03-23 05:32:10 +0000[diff] [blame]11class UrlParseTestCase(unittest.TestCase):
Johannes Gijsbers41e4faa2005-01-09 15:29:10 +0000[diff] [blame]12
13 def checkRoundtrips(self, url, parsed, split):
14 result = urlparse.urlparse(url)
15 self.assertEqual(result, parsed)
Fred Drakead5177c2006-04-01 22:14:43 +0000[diff] [blame]16 t = (result.scheme, result.netloc, result.path,
17 result.params, result.query, result.fragment)
18 self.assertEqual(t, parsed)
Johannes Gijsbers41e4faa2005-01-09 15:29:10 +0000[diff] [blame]19 # put it back together and it should be the same
20 result2 = urlparse.urlunparse(result)
21 self.assertEqual(result2, url)
Fred Drakead5177c2006-04-01 22:14:43 +0000[diff] [blame]22 self.assertEqual(result2, result.geturl())
23
24 # the result of geturl() is a fixpoint; we can always parse it
25 # again to get the same result:
26 result3 = urlparse.urlparse(result.geturl())
27 self.assertEqual(result3.geturl(), result.geturl())
28 self.assertEqual(result3, result)
29 self.assertEqual(result3.scheme, result.scheme)
30 self.assertEqual(result3.netloc, result.netloc)
31 self.assertEqual(result3.path, result.path)
32 self.assertEqual(result3.params, result.params)
33 self.assertEqual(result3.query, result.query)
34 self.assertEqual(result3.fragment, result.fragment)
35 self.assertEqual(result3.username, result.username)
36 self.assertEqual(result3.password, result.password)
37 self.assertEqual(result3.hostname, result.hostname)
38 self.assertEqual(result3.port, result.port)
Johannes Gijsbers41e4faa2005-01-09 15:29:10 +0000[diff] [blame]39
40 # check the roundtrip using urlsplit() as well
41 result = urlparse.urlsplit(url)
42 self.assertEqual(result, split)
Fred Drakead5177c2006-04-01 22:14:43 +0000[diff] [blame]43 t = (result.scheme, result.netloc, result.path,
44 result.query, result.fragment)
45 self.assertEqual(t, split)
Johannes Gijsbers41e4faa2005-01-09 15:29:10 +0000[diff] [blame]46 result2 = urlparse.urlunsplit(result)
47 self.assertEqual(result2, url)
Fred Drakead5177c2006-04-01 22:14:43 +0000[diff] [blame]48 self.assertEqual(result2, result.geturl())
49
50 # check the fixpoint property of re-parsing the result of geturl()
51 result3 = urlparse.urlsplit(result.geturl())
52 self.assertEqual(result3.geturl(), result.geturl())
53 self.assertEqual(result3, result)
54 self.assertEqual(result3.scheme, result.scheme)
55 self.assertEqual(result3.netloc, result.netloc)
56 self.assertEqual(result3.path, result.path)
57 self.assertEqual(result3.query, result.query)
58 self.assertEqual(result3.fragment, result.fragment)
59 self.assertEqual(result3.username, result.username)
60 self.assertEqual(result3.password, result.password)
61 self.assertEqual(result3.hostname, result.hostname)
62 self.assertEqual(result3.port, result.port)
Johannes Gijsbers41e4faa2005-01-09 15:29:10 +0000[diff] [blame]63
64 def test_roundtrips(self):
65 testcases = [
Fred Drake70705652002-10-16 21:02:36 +0000[diff] [blame]66 ('file:///tmp/junk.txt',
67 ('file', '', '/tmp/junk.txt', '', '', ''),
68 ('file', '', '/tmp/junk.txt', '', '')),
Neal Norwitz68b539e2003-01-06 06:58:31 +0000[diff] [blame]69 ('imap://mail.python.org/mbox1',
70 ('imap', 'mail.python.org', '/mbox1', '', '', ''),
71 ('imap', 'mail.python.org', '/mbox1', '', '')),
Skip Montanarof09b88e2003-01-06 20:27:03 +0000[diff] [blame]72 ('mms://wms.sys.hinet.net/cts/Drama/09006251100.asf',
Johannes Gijsbers41e4faa2005-01-09 15:29:10 +0000[diff] [blame]73 ('mms', 'wms.sys.hinet.net', '/cts/Drama/09006251100.asf',
74 '', '', ''),
75 ('mms', 'wms.sys.hinet.net', '/cts/Drama/09006251100.asf',
76 '', '')),
Fred Drake50747fc2005-07-29 15:56:32 +0000[diff] [blame]77 ('svn+ssh://svn.zope.org/repos/main/ZConfig/trunk/',
78 ('svn+ssh', 'svn.zope.org', '/repos/main/ZConfig/trunk/',
79 '', '', ''),
80 ('svn+ssh', 'svn.zope.org', '/repos/main/ZConfig/trunk/',
81 '', ''))
Johannes Gijsbers41e4faa2005-01-09 15:29:10 +0000[diff] [blame]82 ]
83 for url, parsed, split in testcases:
84 self.checkRoundtrips(url, parsed, split)
Michael W. Hudsonbd3e7712002-03-18 13:06:00 +0000[diff] [blame]85
Johannes Gijsbers41e4faa2005-01-09 15:29:10 +0000[diff] [blame]86 def test_http_roundtrips(self):
87 # urlparse.urlsplit treats 'http:' as an optimized special case,
88 # so we test both 'http:' and 'https:' in all the following.
89 # Three cheers for white box knowledge!
90 testcases = [
91 ('://www.python.org',
92 ('www.python.org', '', '', '', ''),
93 ('www.python.org', '', '', '')),
94 ('://www.python.org#abc',
95 ('www.python.org', '', '', '', 'abc'),
96 ('www.python.org', '', '', 'abc')),
97 ('://www.python.org?q=abc',
98 ('www.python.org', '', '', 'q=abc', ''),
99 ('www.python.org', '', 'q=abc', '')),
100 ('://www.python.org/#abc',
101 ('www.python.org', '/', '', '', 'abc'),
102 ('www.python.org', '/', '', 'abc')),
103 ('://a/b/c/d;p?q#f',
104 ('a', '/b/c/d', 'p', 'q', 'f'),
105 ('a', '/b/c/d;p', 'q', 'f')),
106 ]
107 for scheme in ('http', 'https'):
108 for url, parsed, split in testcases:
109 url = scheme + url
110 parsed = (scheme,) + parsed
111 split = (scheme,) + split
112 self.checkRoundtrips(url, parsed, split)
Fred Drake70705652002-10-16 21:02:36 +0000[diff] [blame]113
Skip Montanaro6ec967d2002-03-23 05:32:10 +0000[diff] [blame]114 def checkJoin(self, base, relurl, expected):
Guido van Rossumbbc05682002-10-14 19:59:54 +0000[diff] [blame]115 self.assertEqual(urlparse.urljoin(base, relurl), expected,
116 (base, relurl, expected))
117
118 def test_unparse_parse(self):
119 for u in ['Python', './Python']:
Fred Drake70705652002-10-16 21:02:36 +0000[diff] [blame]120 self.assertEqual(urlparse.urlunsplit(urlparse.urlsplit(u)), u)
Guido van Rossumbbc05682002-10-14 19:59:54 +0000[diff] [blame]121 self.assertEqual(urlparse.urlunparse(urlparse.urlparse(u)), u)
Fred Drakea4d18a02001-01-05 05:57:04 +0000[diff] [blame]122
Skip Montanaro6ec967d2002-03-23 05:32:10 +0000[diff] [blame]123 def test_RFC1808(self):
124 # "normal" cases from RFC 1808:
125 self.checkJoin(RFC1808_BASE, 'g:h', 'g:h')
126 self.checkJoin(RFC1808_BASE, 'g', 'http://a/b/c/g')
127 self.checkJoin(RFC1808_BASE, './g', 'http://a/b/c/g')
128 self.checkJoin(RFC1808_BASE, 'g/', 'http://a/b/c/g/')
129 self.checkJoin(RFC1808_BASE, '/g', 'http://a/g')
130 self.checkJoin(RFC1808_BASE, '//g', 'http://g')
Skip Montanaro6ec967d2002-03-23 05:32:10 +0000[diff] [blame]131 self.checkJoin(RFC1808_BASE, 'g?y', 'http://a/b/c/g?y')
132 self.checkJoin(RFC1808_BASE, 'g?y/./x', 'http://a/b/c/g?y/./x')
133 self.checkJoin(RFC1808_BASE, '#s', 'http://a/b/c/d;p?q#s')
134 self.checkJoin(RFC1808_BASE, 'g#s', 'http://a/b/c/g#s')
135 self.checkJoin(RFC1808_BASE, 'g#s/./x', 'http://a/b/c/g#s/./x')
136 self.checkJoin(RFC1808_BASE, 'g?y#s', 'http://a/b/c/g?y#s')
Skip Montanaro6ec967d2002-03-23 05:32:10 +0000[diff] [blame]137 self.checkJoin(RFC1808_BASE, 'g;x', 'http://a/b/c/g;x')
138 self.checkJoin(RFC1808_BASE, 'g;x?y#s', 'http://a/b/c/g;x?y#s')
139 self.checkJoin(RFC1808_BASE, '.', 'http://a/b/c/')
140 self.checkJoin(RFC1808_BASE, './', 'http://a/b/c/')
141 self.checkJoin(RFC1808_BASE, '..', 'http://a/b/')
142 self.checkJoin(RFC1808_BASE, '../', 'http://a/b/')
143 self.checkJoin(RFC1808_BASE, '../g', 'http://a/b/g')
144 self.checkJoin(RFC1808_BASE, '../..', 'http://a/')
145 self.checkJoin(RFC1808_BASE, '../../', 'http://a/')
146 self.checkJoin(RFC1808_BASE, '../../g', 'http://a/g')
Fred Drakea4d18a02001-01-05 05:57:04 +0000[diff] [blame]147
Skip Montanaro6ec967d2002-03-23 05:32:10 +0000[diff] [blame]148 # "abnormal" cases from RFC 1808:
149 self.checkJoin(RFC1808_BASE, '', 'http://a/b/c/d;p?q#f')
150 self.checkJoin(RFC1808_BASE, '../../../g', 'http://a/../g')
151 self.checkJoin(RFC1808_BASE, '../../../../g', 'http://a/../../g')
152 self.checkJoin(RFC1808_BASE, '/./g', 'http://a/./g')
153 self.checkJoin(RFC1808_BASE, '/../g', 'http://a/../g')
154 self.checkJoin(RFC1808_BASE, 'g.', 'http://a/b/c/g.')
155 self.checkJoin(RFC1808_BASE, '.g', 'http://a/b/c/.g')
156 self.checkJoin(RFC1808_BASE, 'g..', 'http://a/b/c/g..')
157 self.checkJoin(RFC1808_BASE, '..g', 'http://a/b/c/..g')
158 self.checkJoin(RFC1808_BASE, './../g', 'http://a/b/g')
159 self.checkJoin(RFC1808_BASE, './g/.', 'http://a/b/c/g/')
160 self.checkJoin(RFC1808_BASE, 'g/./h', 'http://a/b/c/g/h')
161 self.checkJoin(RFC1808_BASE, 'g/../h', 'http://a/b/c/h')
Fred Drakea4d18a02001-01-05 05:57:04 +0000[diff] [blame]162
Skip Montanaro6ec967d2002-03-23 05:32:10 +0000[diff] [blame]163 # RFC 1808 and RFC 1630 disagree on these (according to RFC 1808),
164 # so we'll not actually run these tests (which expect 1808 behavior).
165 #self.checkJoin(RFC1808_BASE, 'http:g', 'http:g')
166 #self.checkJoin(RFC1808_BASE, 'http:', 'http:')
Fred Drakea4d18a02001-01-05 05:57:04 +0000[diff] [blame]167
Skip Montanaro6ec967d2002-03-23 05:32:10 +0000[diff] [blame]168 def test_RFC2396(self):
169 # cases from RFC 2396
Fred Drakea4d18a02001-01-05 05:57:04 +0000[diff] [blame]170
Skip Montanaro6ec967d2002-03-23 05:32:10 +0000[diff] [blame]171
172 self.checkJoin(RFC2396_BASE, 'g:h', 'g:h')
173 self.checkJoin(RFC2396_BASE, 'g', 'http://a/b/c/g')
174 self.checkJoin(RFC2396_BASE, './g', 'http://a/b/c/g')
175 self.checkJoin(RFC2396_BASE, 'g/', 'http://a/b/c/g/')
176 self.checkJoin(RFC2396_BASE, '/g', 'http://a/g')
177 self.checkJoin(RFC2396_BASE, '//g', 'http://g')
178 self.checkJoin(RFC2396_BASE, 'g?y', 'http://a/b/c/g?y')
179 self.checkJoin(RFC2396_BASE, '#s', 'http://a/b/c/d;p?q#s')
180 self.checkJoin(RFC2396_BASE, 'g#s', 'http://a/b/c/g#s')
181 self.checkJoin(RFC2396_BASE, 'g?y#s', 'http://a/b/c/g?y#s')
182 self.checkJoin(RFC2396_BASE, 'g;x', 'http://a/b/c/g;x')
183 self.checkJoin(RFC2396_BASE, 'g;x?y#s', 'http://a/b/c/g;x?y#s')
184 self.checkJoin(RFC2396_BASE, '.', 'http://a/b/c/')
185 self.checkJoin(RFC2396_BASE, './', 'http://a/b/c/')
186 self.checkJoin(RFC2396_BASE, '..', 'http://a/b/')
187 self.checkJoin(RFC2396_BASE, '../', 'http://a/b/')
188 self.checkJoin(RFC2396_BASE, '../g', 'http://a/b/g')
189 self.checkJoin(RFC2396_BASE, '../..', 'http://a/')
190 self.checkJoin(RFC2396_BASE, '../../', 'http://a/')
191 self.checkJoin(RFC2396_BASE, '../../g', 'http://a/g')
192 self.checkJoin(RFC2396_BASE, '', RFC2396_BASE)
193 self.checkJoin(RFC2396_BASE, '../../../g', 'http://a/../g')
194 self.checkJoin(RFC2396_BASE, '../../../../g', 'http://a/../../g')
195 self.checkJoin(RFC2396_BASE, '/./g', 'http://a/./g')
196 self.checkJoin(RFC2396_BASE, '/../g', 'http://a/../g')
197 self.checkJoin(RFC2396_BASE, 'g.', 'http://a/b/c/g.')
198 self.checkJoin(RFC2396_BASE, '.g', 'http://a/b/c/.g')
199 self.checkJoin(RFC2396_BASE, 'g..', 'http://a/b/c/g..')
200 self.checkJoin(RFC2396_BASE, '..g', 'http://a/b/c/..g')
201 self.checkJoin(RFC2396_BASE, './../g', 'http://a/b/g')
202 self.checkJoin(RFC2396_BASE, './g/.', 'http://a/b/c/g/')
203 self.checkJoin(RFC2396_BASE, 'g/./h', 'http://a/b/c/g/h')
204 self.checkJoin(RFC2396_BASE, 'g/../h', 'http://a/b/c/h')
205 self.checkJoin(RFC2396_BASE, 'g;x=1/./y', 'http://a/b/c/g;x=1/y')
206 self.checkJoin(RFC2396_BASE, 'g;x=1/../y', 'http://a/b/c/y')
207 self.checkJoin(RFC2396_BASE, 'g?y/./x', 'http://a/b/c/g?y/./x')
208 self.checkJoin(RFC2396_BASE, 'g?y/../x', 'http://a/b/c/g?y/../x')
209 self.checkJoin(RFC2396_BASE, 'g#s/./x', 'http://a/b/c/g#s/./x')
210 self.checkJoin(RFC2396_BASE, 'g#s/../x', 'http://a/b/c/g#s/../x')
211
Facundo Batista67d19812008-08-14 16:51:00 +0000[diff] [blame^]212 #The following scenarios have been updated in RFC3986
213 #self.checkJoin(RFC2396_BASE, '?y', 'http://a/b/c/?y')
214 #self.checkJoin(RFC2396_BASE, ';x', 'http://a/b/c/;x')
215
216 def test_RFC3986(self):
217 self.checkJoin(RFC3986_BASE, '?y','http://a/b/c/d;p?y')
218 self.checkJoin(RFC2396_BASE, ';x', 'http://a/b/c/;x')
219
Fred Drake70705652002-10-16 21:02:36 +0000[diff] [blame]220 def test_urldefrag(self):
221 for url, defrag, frag in [
222 ('http://python.org#frag', 'http://python.org', 'frag'),
223 ('http://python.org', 'http://python.org', ''),
224 ('http://python.org/#frag', 'http://python.org/', 'frag'),
225 ('http://python.org/', 'http://python.org/', ''),
226 ('http://python.org/?q#frag', 'http://python.org/?q', 'frag'),
227 ('http://python.org/?q', 'http://python.org/?q', ''),
228 ('http://python.org/p#frag', 'http://python.org/p', 'frag'),
229 ('http://python.org/p?q', 'http://python.org/p?q', ''),
230 (RFC1808_BASE, 'http://a/b/c/d;p?q', 'f'),
231 (RFC2396_BASE, 'http://a/b/c/d;p?q', ''),
232 ]:
233 self.assertEqual(urlparse.urldefrag(url), (defrag, frag))
234
Fred Drakead5177c2006-04-01 22:14:43 +0000[diff] [blame]235 def test_urlsplit_attributes(self):
236 url = "HTTP://WWW.PYTHON.ORG/doc/#frag"
237 p = urlparse.urlsplit(url)
238 self.assertEqual(p.scheme, "http")
239 self.assertEqual(p.netloc, "WWW.PYTHON.ORG")
240 self.assertEqual(p.path, "/doc/")
241 self.assertEqual(p.query, "")
242 self.assertEqual(p.fragment, "frag")
243 self.assertEqual(p.username, None)
244 self.assertEqual(p.password, None)
245 self.assertEqual(p.hostname, "www.python.org")
246 self.assertEqual(p.port, None)
247 # geturl() won't return exactly the original URL in this case
248 # since the scheme is always case-normalized
249 #self.assertEqual(p.geturl(), url)
250
251 url = "http://User:Pass@www.python.org:080/doc/?query=yes#frag"
252 p = urlparse.urlsplit(url)
253 self.assertEqual(p.scheme, "http")
254 self.assertEqual(p.netloc, "User:Pass@www.python.org:080")
255 self.assertEqual(p.path, "/doc/")
256 self.assertEqual(p.query, "query=yes")
257 self.assertEqual(p.fragment, "frag")
258 self.assertEqual(p.username, "User")
259 self.assertEqual(p.password, "Pass")
260 self.assertEqual(p.hostname, "www.python.org")
261 self.assertEqual(p.port, 80)
262 self.assertEqual(p.geturl(), url)
263
Guido van Rossumced4eb02008-01-05 01:21:57 +0000[diff] [blame]264 # Addressing issue1698, which suggests Username can contain
Andrew M. Kuchling05899142008-01-05 15:13:49 +0000[diff] [blame]265 # "@" characters. Though not RFC compliant, many ftp sites allow
Fred Drakef7476c42008-01-05 04:38:38 +0000[diff] [blame]266 # and request email addresses as usernames.
Guido van Rossumced4eb02008-01-05 01:21:57 +0000[diff] [blame]267
268 url = "http://User@example.com:Pass@www.python.org:080/doc/?query=yes#frag"
269 p = urlparse.urlsplit(url)
270 self.assertEqual(p.scheme, "http")
271 self.assertEqual(p.netloc, "User@example.com:Pass@www.python.org:080")
272 self.assertEqual(p.path, "/doc/")
273 self.assertEqual(p.query, "query=yes")
274 self.assertEqual(p.fragment, "frag")
275 self.assertEqual(p.username, "User@example.com")
276 self.assertEqual(p.password, "Pass")
277 self.assertEqual(p.hostname, "www.python.org")
278 self.assertEqual(p.port, 80)
279 self.assertEqual(p.geturl(), url)
280
281
Fred Drakead5177c2006-04-01 22:14:43 +0000[diff] [blame]282 def test_attributes_bad_port(self):
283 """Check handling of non-integer ports."""
284 p = urlparse.urlsplit("http://www.example.net:foo")
285 self.assertEqual(p.netloc, "www.example.net:foo")
286 self.assertRaises(ValueError, lambda: p.port)
287
288 p = urlparse.urlparse("http://www.example.net:foo")
289 self.assertEqual(p.netloc, "www.example.net:foo")
290 self.assertRaises(ValueError, lambda: p.port)
291
292 def test_attributes_without_netloc(self):
293 # This example is straight from RFC 3261. It looks like it
294 # should allow the username, hostname, and port to be filled
295 # in, but doesn't. Since it's a URI and doesn't use the
296 # scheme://netloc syntax, the netloc and related attributes
297 # should be left empty.
298 uri = "sip:alice@atlanta.com;maddr=239.255.255.1;ttl=15"
299 p = urlparse.urlsplit(uri)
300 self.assertEqual(p.netloc, "")
301 self.assertEqual(p.username, None)
302 self.assertEqual(p.password, None)
303 self.assertEqual(p.hostname, None)
304 self.assertEqual(p.port, None)
305 self.assertEqual(p.geturl(), uri)
306
307 p = urlparse.urlparse(uri)
308 self.assertEqual(p.netloc, "")
309 self.assertEqual(p.username, None)
310 self.assertEqual(p.password, None)
311 self.assertEqual(p.hostname, None)
312 self.assertEqual(p.port, None)
313 self.assertEqual(p.geturl(), uri)
314
Alexandre Vassalotti2f9ca292007-12-13 17:58:23 +0000[diff] [blame]315 def test_caching(self):
316 # Test case for bug #1313119
317 uri = "http://example.com/doc/"
318 unicode_uri = unicode(uri)
319
320 urlparse.urlparse(unicode_uri)
321 p = urlparse.urlparse(uri)
322 self.assertEqual(type(p.scheme), type(uri))
323 self.assertEqual(type(p.hostname), type(uri))
324 self.assertEqual(type(p.path), type(uri))
Fred Drakead5177c2006-04-01 22:14:43 +0000[diff] [blame]325
Guido van Rossumc6a04c22008-01-05 22:19:06 +0000[diff] [blame]326 def test_noslash(self):
327 # Issue 1637: http://foo.com?query is legal
328 self.assertEqual(urlparse.urlparse("http://example.com?blahblah=/foo"),
329 ('http', 'example.com', '', '', 'blahblah=/foo', ''))
330
Skip Montanaro6ec967d2002-03-23 05:32:10 +0000[diff] [blame]331def test_main():
332 test_support.run_unittest(UrlParseTestCase)
333
334if __name__ == "__main__":
335 test_main()