source4/dsdb/tests/python/sort.py

   1 #!/usr/bin/env python
   2 # -*- coding: utf-8 -*-
   3 # Originally based on ./sam.py
   4 from __future__ import print_function
   5 from unicodedata import normalize
   6 import locale
   7 locale.setlocale(locale.LC_ALL, ('en_US', 'UTF-8'))
   8
   9 import optparse
  10 import sys
  11 import os
  12 import re
  13
  14 sys.path.insert(0, "bin/python")
  15 import samba
  16 from samba.tests.subunitrun import SubunitOptions, TestProgram
  17 from samba.compat import cmp_fn
  18 from samba.compat import cmp_to_key_fn
  19 from samba.compat import text_type
  20 import samba.getopt as options
  21
  22 from samba.auth import system_session
  23 import ldb
  24 from samba.samdb import SamDB
  25
  26 parser = optparse.OptionParser("sort.py [options] <host>")
  27 sambaopts = options.SambaOptions(parser)
  28 parser.add_option_group(sambaopts)
  29 parser.add_option_group(options.VersionOptions(parser))
  30 # use command line creds if available
  31 credopts = options.CredentialsOptions(parser)
  32 parser.add_option_group(credopts)
  33 subunitopts = SubunitOptions(parser)
  34 parser.add_option_group(subunitopts)
  35
  36 parser.add_option('--elements', type='int', default=33,
  37                   help="use this many elements in the tests")
  38
  39 opts, args = parser.parse_args()
  40
  41 if len(args) < 1:
  42     parser.print_usage()
  43     sys.exit(1)
  44
  45 host = args[0]
  46
  47 lp = sambaopts.get_loadparm()
  48 creds = credopts.get_credentials(lp)
  49
  50
  51 def norm(x):
  52     if not isinstance(x, text_type):
  53         x = x.decode('utf8')
  54     return normalize('NFKC', x).upper()
  55
  56
  57 # Python, Windows, and Samba all sort the following sequence in
  58 # drastically different ways. The order here is what you get from
  59 # Windows2012R2.
  60 FIENDISH_TESTS = [' ', ' e', '\t-\t', '\n\t\t', '!@#!@#!', '¼', '¹', '1',
  61                   '1/4', '1⁄4', '1\xe2\x81\x845', '3', 'abc', 'fo\x00od',
  62
  63                   # Here we also had '\x00food', but that seems to sort
  64                   # non-deterministically on Windows vis-a-vis 'fo\x00od'.
  65
  66                   'kōkako', 'ŋđ¼³ŧ “«đð', 'ŋđ¼³ŧ“«đð',
  67                   'ｓorttest', 'sorttēst11,', 'śorttest2', 'śoRttest2',
  68                   'ś-o-r-t-t-e-s-t-2', 'soRTTēst2,', 'ṡorttest4', 'ṡorttesT4',
  69                   'sörttest-5', 'sÖrttest-5', 'so-rttest7,', '桑巴']
  70
  71
  72 class BaseSortTests(samba.tests.TestCase):
  73     avoid_tricky_sort = False
  74     maxDiff = 2000
  75
  76     def create_user(self, i, n, prefix='sorttest', suffix='', attrs=None,
  77                     tricky=False):
  78         name = "%s%d%s" % (prefix, i, suffix)
  79         user = {
  80             'cn': name,
  81             "objectclass": "user",
  82             'givenName': "abcdefghijklmnopqrstuvwxyz"[i % 26],
  83             "roomNumber": "%sb\x00c" % (n - i),
  84             "carLicense": "后来经",
  85             "employeeNumber": "%s%sx" % (abs(i * (99 - i)), '\n' * (i & 255)),
  86             "accountExpires": "%s" % (10 ** 9 + 1000000 * i),
  87             "msTSExpireDate4": "19%02d0101010000.0Z" % (i % 100),
  88             "flags": str(i * (n - i)),
  89             "serialNumber": "abc %s%s%s" % ('AaBb |-/'[i & 7],
  90                                             ' 3z}'[i & 3],
  91                                             '"@'[i & 1],),
  92             "comment": "Favourite colour is %d" % (n % (i + 1)),
  93         }
  94
  95         if self.avoid_tricky_sort:
  96             # We are not even going to try passing tests that assume
  97             # some kind of Unicode awareness.
  98             for k, v in user.items():
  99                 user[k] = re.sub(r'[^\w,.]', 'X', v)
 100         else:
 101             # Add some even trickier ones!
 102             fiendish_index = i % len(FIENDISH_TESTS)
 103             user.update({
 104                 # Sort doesn't look past a NUL byte.
 105                 "photo": "\x00%d" % (n - i),
 106                 "audio": "%sn octet string %s%s ♫♬\x00lalala" % ('Aa'[i & 1],
 107                                                                  chr(i & 255),
 108                                                                  i),
 109                 "displayNamePrintable": "%d\x00%c" % (i, i & 255),
 110                 "adminDisplayName": "%d\x00b" % (n - i),
 111                 "title": "%d%sb" % (n - i, '\x00' * i),
 112
 113                 # Names that vary only in case. Windows returns
 114                 # equivalent addresses in the order they were put
 115                 # in ('a st', 'A st',...). We don't check that.
 116                 "street": "%s st" % (chr(65 | (i & 14) | ((i & 1) * 32))),
 117
 118                 "streetAddress": FIENDISH_TESTS[fiendish_index],
 119                 "postalAddress": FIENDISH_TESTS[-fiendish_index],
 120             })
 121
 122         if attrs is not None:
 123             user.update(attrs)
 124
 125         user['dn'] = "cn=%s,%s" % (user['cn'], self.ou)
 126
 127         self.users.append(user)
 128         self.ldb.add(user)
 129         return user
 130
 131     def setUp(self):
 132         super(BaseSortTests, self).setUp()
 133         self.ldb = SamDB(host, credentials=creds,
 134                          session_info=system_session(lp), lp=lp)
 135
 136         self.base_dn = self.ldb.domain_dn()
 137         self.ou = "ou=sort,%s" % self.base_dn
 138         if False:
 139             try:
 140                 self.ldb.delete(self.ou, ['tree_delete:1'])
 141             except ldb.LdbError as e:
 142                 print("tried deleting %s, got error %s" % (self.ou, e))
 143
 144         self.ldb.add({
 145             "dn": self.ou,
 146             "objectclass": "organizationalUnit"})
 147         self.users = []
 148         n = opts.elements
 149         for i in range(n):
 150             self.create_user(i, n)
 151
 152         attrs = set(self.users[0].keys()) - set([
 153             'objectclass', 'dn'])
 154         self.binary_sorted_keys = attrs.intersection(['audio',
 155                                                       'photo',
 156                                                       "msTSExpireDate4",
 157                                                       'serialNumber',
 158                                                       "displayNamePrintable"])
 159
 160         self.numeric_sorted_keys = attrs.intersection(['flags',
 161                                                        'accountExpires'])
 162
 163         self.timestamp_keys = attrs.intersection(['msTSExpireDate4'])
 164
 165         self.int64_keys = set(['accountExpires'])
 166
 167         self.locale_sorted_keys = [x for x in attrs if
 168                                    x not in (self.binary_sorted_keys |
 169                                              self.numeric_sorted_keys)]
 170
 171         self.expected_results = {}
 172         self.expected_results_binary = {}
 173
 174         for k in self.locale_sorted_keys:
 175             # Using key=locale.strxfrm fails on \x00
 176             forward = sorted((norm(x[k]) for x in self.users),
 177                              key=cmp_to_key_fn(locale.strcoll))
 178             reverse = list(reversed(forward))
 179             self.expected_results[k] = (forward, reverse)
 180
 181         for k in self.binary_sorted_keys:
 182             forward = sorted((x[k] for x in self.users))
 183             reverse = list(reversed(forward))
 184             self.expected_results_binary[k] = (forward, reverse)
 185             self.expected_results[k] = (forward, reverse)
 186
 187         # Fix up some because Python gets it wrong, using Schwartzian tramsform
 188         for k in ('adminDisplayName', 'title', 'streetAddress',
 189                   'employeeNumber'):
 190             if k in self.expected_results:
 191                 broken = self.expected_results[k][0]
 192                 tmp = [(x.replace('\x00', ''), x) for x in broken]
 193                 tmp.sort()
 194                 fixed = [x[1] for x in tmp]
 195                 self.expected_results[k] = (fixed, list(reversed(fixed)))
 196         for k in ('streetAddress', 'postalAddress'):
 197             if k in self.expected_results:
 198                 c = {}
 199                 for u in self.users:
 200                     x = u[k]
 201                     if x in c:
 202                         c[x] += 1
 203                         continue
 204                     c[x] = 1
 205                 fixed = []
 206                 for x in FIENDISH_TESTS:
 207                     fixed += [norm(x)] * c.get(x, 0)
 208
 209                 rev = list(reversed(fixed))
 210                 self.expected_results[k] = (fixed, rev)
 211
 212     def tearDown(self):
 213         super(BaseSortTests, self).tearDown()
 214         self.ldb.delete(self.ou, ['tree_delete:1'])
 215
 216     def _test_server_sort_default(self):
 217         attrs = self.locale_sorted_keys
 218
 219         for attr in attrs:
 220             for rev in (0, 1):
 221                 res = self.ldb.search(self.ou,
 222                                       scope=ldb.SCOPE_ONELEVEL, attrs=[attr],
 223                                       controls=["server_sort:1:%d:%s" %
 224                                                 (rev, attr)])
 225                 self.assertEqual(len(res), len(self.users))
 226
 227                 expected_order = self.expected_results[attr][rev]
 228                 received_order = [norm(x[attr][0]) for x in res]
 229                 if expected_order != received_order:
 230                     print(attr, ['forward', 'reverse'][rev])
 231                     print("expected", expected_order)
 232                     print("recieved", received_order)
 233                     print("unnormalised:", [x[attr][0] for x in res])
 234                     print("unnormalised: «%s»" % '»  «'.join(str(x[attr][0])
 235                                                              for x in res))
 236                 self.assertEquals(expected_order, received_order)
 237
 238     def _test_server_sort_binary(self):
 239         for attr in self.binary_sorted_keys:
 240             for rev in (0, 1):
 241                 res = self.ldb.search(self.ou,
 242                                       scope=ldb.SCOPE_ONELEVEL, attrs=[attr],
 243                                       controls=["server_sort:1:%d:%s" %
 244                                                 (rev, attr)])
 245
 246                 self.assertEqual(len(res), len(self.users))
 247                 expected_order = self.expected_results_binary[attr][rev]
 248                 received_order = [str(x[attr][0]) for x in res]
 249                 if expected_order != received_order:
 250                     print(attr)
 251                     print(expected_order)
 252                     print(received_order)
 253                 self.assertEquals(expected_order, received_order)
 254
 255     def _test_server_sort_us_english(self):
 256         # Windows doesn't support many matching rules, but does allow
 257         # the locale specific sorts -- if it has the locale installed.
 258         # The most reliable locale is the default US English, which
 259         # won't change the sort order.
 260
 261         for lang, oid in [('en_US', '1.2.840.113556.1.4.1499'),
 262                           ]:
 263
 264             for attr in self.locale_sorted_keys:
 265                 for rev in (0, 1):
 266                     res = self.ldb.search(self.ou,
 267                                           scope=ldb.SCOPE_ONELEVEL,
 268                                           attrs=[attr],
 269                                           controls=["server_sort:1:%d:%s:%s" %
 270                                                     (rev, attr, oid)])
 271
 272                     self.assertTrue(len(res) == len(self.users))
 273                     expected_order = self.expected_results[attr][rev]
 274                     received_order = [norm(x[attr][0]) for x in res]
 275                     if expected_order != received_order:
 276                         print(attr, lang)
 277                         print(['forward', 'reverse'][rev])
 278                         print("expected: ", expected_order)
 279                         print("recieved: ", received_order)
 280                         print("unnormalised:", [x[attr][0] for x in res])
 281                         print("unnormalised: «%s»" % '»  «'.join(str(x[attr][0])
 282                                                                  for x in res))
 283
 284                     self.assertEquals(expected_order, received_order)
 285
 286     def _test_server_sort_different_attr(self):
 287
 288         def cmp_locale(a, b):
 289             return locale.strcoll(a[0], b[0])
 290
 291         def cmp_binary(a, b):
 292             return cmp_fn(a[0], b[0])
 293
 294         def cmp_numeric(a, b):
 295             return cmp_fn(int(a[0]), int(b[0]))
 296
 297         # For testing simplicity, the attributes in here need to be
 298         # unique for each user. Otherwise there are multiple possible
 299         # valid answers.
 300         sort_functions = {'cn': cmp_binary,
 301                           "employeeNumber": cmp_locale,
 302                           "accountExpires": cmp_numeric,
 303                           "msTSExpireDate4": cmp_binary}
 304         attrs = list(sort_functions.keys())
 305         attr_pairs = zip(attrs, attrs[1:] + attrs[:1])
 306
 307         for sort_attr, result_attr in attr_pairs:
 308             forward = sorted(((norm(x[sort_attr]), norm(x[result_attr]))
 309                              for x in self.users),
 310                              key=cmp_to_key_fn(sort_functions[sort_attr]))
 311             reverse = list(reversed(forward))
 312
 313             for rev in (0, 1):
 314                 res = self.ldb.search(self.ou,
 315                                       scope=ldb.SCOPE_ONELEVEL,
 316                                       attrs=[result_attr],
 317                                       controls=["server_sort:1:%d:%s" %
 318                                                 (rev, sort_attr)])
 319                 self.assertEqual(len(res), len(self.users))
 320                 pairs = (forward, reverse)[rev]
 321
 322                 expected_order = [x[1] for x in pairs]
 323                 received_order = [norm(x[result_attr][0]) for x in res]
 324
 325                 if expected_order != received_order:
 326                     print(sort_attr, result_attr, ['forward', 'reverse'][rev])
 327                     print("expected", expected_order)
 328                     print("recieved", received_order)
 329                     print("unnormalised:", [x[result_attr][0] for x in res])
 330                     print("unnormalised: «%s»" % '»  «'.join(str(x[result_attr][0])
 331                                                              for x in res))
 332                     print("pairs:", pairs)
 333                     # There are bugs in Windows that we don't want (or
 334                     # know how) to replicate regarding timestamp sorting.
 335                     # Let's remind ourselves.
 336                     if result_attr == "msTSExpireDate4":
 337                         print('-' * 72)
 338                         print("This test fails against Windows with the "
 339                               "default number of elements (33).")
 340                         print("Try with --elements=27 (or similar).")
 341                         print('-' * 72)
 342
 343                 self.assertEquals(expected_order, received_order)
 344                 for x in res:
 345                     if sort_attr in x:
 346                         self.fail('the search for %s should not return %s' %
 347                                   (result_attr, sort_attr))
 348
 349
 350 class SimpleSortTests(BaseSortTests):
 351     avoid_tricky_sort = True
 352
 353     def test_server_sort_different_attr(self):
 354         self._test_server_sort_different_attr()
 355
 356     def test_server_sort_default(self):
 357         self._test_server_sort_default()
 358
 359     def test_server_sort_binary(self):
 360         self._test_server_sort_binary()
 361
 362     def test_server_sort_us_english(self):
 363         self._test_server_sort_us_english()
 364
 365
 366 class UnicodeSortTests(BaseSortTests):
 367     avoid_tricky_sort = False
 368
 369     def test_server_sort_default(self):
 370         self._test_server_sort_default()
 371
 372     def test_server_sort_us_english(self):
 373         self._test_server_sort_us_english()
 374
 375     def test_server_sort_different_attr(self):
 376         self._test_server_sort_different_attr()
 377
 378
 379 if "://" not in host:
 380     if os.path.isfile(host):
 381         host = "tdb://%s" % host
 382     else:
 383         host = "ldap://%s" % host
 384
 385
 386 TestProgram(module=__name__, opts=subunitopts)