Lib/test/test_urlparse.py

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