Lib/test/test_hashlib.py

   1 # Test hashlib module
   2 #
   3 # $Id$
   4 #
   5 #  Copyright (C) 2005-2010   Gregory P. Smith (greg@krypto.org)
   6 #  Licensed to PSF under a Contributor Agreement.
   7 #
   8
   9 import array
  10 import hashlib
  11 import StringIO
  12 import itertools
  13 import sys
  14 try:
  15     import threading
  16 except ImportError:
  17     threading = None
  18 import unittest
  19 import warnings
  20 from test import test_support
  21 from test.test_support import _4G, precisionbigmemtest
  22
  23 # Were we compiled --with-pydebug or with #define Py_DEBUG?
  24 COMPILED_WITH_PYDEBUG = hasattr(sys, 'gettotalrefcount')
  25
  26
  27 def hexstr(s):
  28     import string
  29     h = string.hexdigits
  30     r = ''
  31     for c in s:
  32         i = ord(c)
  33         r = r + h[(i >> 4) & 0xF] + h[i & 0xF]
  34     return r
  35
  36
  37 class HashLibTestCase(unittest.TestCase):
  38     supported_hash_names = ( 'md5', 'MD5', 'sha1', 'SHA1',
  39                              'sha224', 'SHA224', 'sha256', 'SHA256',
  40                              'sha384', 'SHA384', 'sha512', 'SHA512' )
  41
  42     _warn_on_extension_import = COMPILED_WITH_PYDEBUG
  43
  44     def _conditional_import_module(self, module_name):
  45         """Import a module and return a reference to it or None on failure."""
  46         try:
  47             exec('import '+module_name)
  48         except ImportError, error:
  49             if self._warn_on_extension_import:
  50                 warnings.warn('Did a C extension fail to compile? %s' % error)
  51         return locals().get(module_name)
  52
  53     def __init__(self, *args, **kwargs):
  54         algorithms = set()
  55         for algorithm in self.supported_hash_names:
  56             algorithms.add(algorithm.lower())
  57         self.constructors_to_test = {}
  58         for algorithm in algorithms:
  59             self.constructors_to_test[algorithm] = set()
  60
  61         # For each algorithm, test the direct constructor and the use
  62         # of hashlib.new given the algorithm name.
  63         for algorithm, constructors in self.constructors_to_test.items():
  64             constructors.add(getattr(hashlib, algorithm))
  65             def _test_algorithm_via_hashlib_new(data=None, _alg=algorithm):
  66                 if data is None:
  67                     return hashlib.new(_alg)
  68                 return hashlib.new(_alg, data)
  69             constructors.add(_test_algorithm_via_hashlib_new)
  70
  71         _hashlib = self._conditional_import_module('_hashlib')
  72         if _hashlib:
  73             # These two algorithms should always be present when this module
  74             # is compiled.  If not, something was compiled wrong.
  75             assert hasattr(_hashlib, 'openssl_md5')
  76             assert hasattr(_hashlib, 'openssl_sha1')
  77             for algorithm, constructors in self.constructors_to_test.items():
  78                 constructor = getattr(_hashlib, 'openssl_'+algorithm, None)
  79                 if constructor:
  80                     constructors.add(constructor)
  81
  82         _md5 = self._conditional_import_module('_md5')
  83         if _md5:
  84             self.constructors_to_test['md5'].add(_md5.new)
  85         _sha = self._conditional_import_module('_sha')
  86         if _sha:
  87             self.constructors_to_test['sha1'].add(_sha.new)
  88         _sha256 = self._conditional_import_module('_sha256')
  89         if _sha256:
  90             self.constructors_to_test['sha224'].add(_sha256.sha224)
  91             self.constructors_to_test['sha256'].add(_sha256.sha256)
  92         _sha512 = self._conditional_import_module('_sha512')
  93         if _sha512:
  94             self.constructors_to_test['sha384'].add(_sha512.sha384)
  95             self.constructors_to_test['sha512'].add(_sha512.sha512)
  96
  97         super(HashLibTestCase, self).__init__(*args, **kwargs)
  98
  99     def test_hash_array(self):
 100         a = array.array("b", range(10))
 101         constructors = self.constructors_to_test.itervalues()
 102         for cons in itertools.chain.from_iterable(constructors):
 103             c = cons(a)
 104             c.hexdigest()
 105
 106     def test_unknown_hash(self):
 107         try:
 108             hashlib.new('spam spam spam spam spam')
 109         except ValueError:
 110             pass
 111         else:
 112             self.assertTrue(0 == "hashlib didn't reject bogus hash name")
 113
 114     def test_hexdigest(self):
 115         for name in self.supported_hash_names:
 116             h = hashlib.new(name)
 117             self.assertTrue(hexstr(h.digest()) == h.hexdigest())
 118
 119     def test_large_update(self):
 120         aas = 'a' * 128
 121         bees = 'b' * 127
 122         cees = 'c' * 126
 123         abcs = aas + bees + cees
 124
 125         for name in self.supported_hash_names:
 126             m1 = hashlib.new(name)
 127             m1.update(aas)
 128             m1.update(bees)
 129             m1.update(cees)
 130
 131             m2 = hashlib.new(name)
 132             m2.update(abcs)
 133             self.assertEqual(m1.digest(), m2.digest(), name+' update problem.')
 134
 135             m3 = hashlib.new(name, abcs)
 136             self.assertEqual(m1.digest(), m3.digest(), name+' new problem.')
 137
 138     def check(self, name, data, digest):
 139         constructors = self.constructors_to_test[name]
 140         # 2 is for hashlib.name(...) and hashlib.new(name, ...)
 141         self.assertGreaterEqual(len(constructors), 2)
 142         for hash_object_constructor in constructors:
 143             computed = hash_object_constructor(data).hexdigest()
 144             self.assertEqual(
 145                     computed, digest,
 146                     "Hash algorithm %s constructed using %s returned hexdigest"
 147                     " %r for %d byte input data that should have hashed to %r."
 148                     % (name, hash_object_constructor,
 149                        computed, len(data), digest))
 150
 151     def check_unicode(self, algorithm_name):
 152         # Unicode objects are not allowed as input.
 153         expected = hashlib.new(algorithm_name, str(u'spam')).hexdigest()
 154         self.check(algorithm_name, u'spam', expected)
 155
 156     def test_unicode(self):
 157         # In python 2.x unicode is auto-encoded to the system default encoding
 158         # when passed to hashlib functions.
 159         self.check_unicode('md5')
 160         self.check_unicode('sha1')
 161         self.check_unicode('sha224')
 162         self.check_unicode('sha256')
 163         self.check_unicode('sha384')
 164         self.check_unicode('sha512')
 165
 166     def test_case_md5_0(self):
 167         self.check('md5', '', 'd41d8cd98f00b204e9800998ecf8427e')
 168
 169     def test_case_md5_1(self):
 170         self.check('md5', 'abc', '900150983cd24fb0d6963f7d28e17f72')
 171
 172     def test_case_md5_2(self):
 173         self.check('md5', 'ABCDEFGHIJKLMNOPQRSTUVWXYZabcdefghijklmnopqrstuvwxyz0123456789',
 174                    'd174ab98d277d9f5a5611c2c9f419d9f')
 175
 176     @precisionbigmemtest(size=_4G + 5, memuse=1)
 177     def test_case_md5_huge(self, size):
 178         if size == _4G + 5:
 179             try:
 180                 self.check('md5', 'A'*size, 'c9af2dff37468ce5dfee8f2cfc0a9c6d')
 181             except OverflowError:
 182                 pass # 32-bit arch
 183
 184     @precisionbigmemtest(size=_4G - 1, memuse=1)
 185     def test_case_md5_uintmax(self, size):
 186         if size == _4G - 1:
 187             try:
 188                 self.check('md5', 'A'*size, '28138d306ff1b8281f1a9067e1a1a2b3')
 189             except OverflowError:
 190                 pass # 32-bit arch
 191
 192     # use the three examples from Federal Information Processing Standards
 193     # Publication 180-1, Secure Hash Standard,  1995 April 17
 194     # http://www.itl.nist.gov/div897/pubs/fip180-1.htm
 195
 196     def test_case_sha1_0(self):
 197         self.check('sha1', "",
 198                    "da39a3ee5e6b4b0d3255bfef95601890afd80709")
 199
 200     def test_case_sha1_1(self):
 201         self.check('sha1', "abc",
 202                    "a9993e364706816aba3e25717850c26c9cd0d89d")
 203
 204     def test_case_sha1_2(self):
 205         self.check('sha1', "abcdbcdecdefdefgefghfghighijhijkijkljklmklmnlmnomnopnopq",
 206                    "84983e441c3bd26ebaae4aa1f95129e5e54670f1")
 207
 208     def test_case_sha1_3(self):
 209         self.check('sha1', "a" * 1000000,
 210                    "34aa973cd4c4daa4f61eeb2bdbad27316534016f")
 211
 212
 213     # use the examples from Federal Information Processing Standards
 214     # Publication 180-2, Secure Hash Standard,  2002 August 1
 215     # http://csrc.nist.gov/publications/fips/fips180-2/fips180-2.pdf
 216
 217     def test_case_sha224_0(self):
 218         self.check('sha224', "",
 219           "d14a028c2a3a2bc9476102bb288234c415a2b01f828ea62ac5b3e42f")
 220
 221     def test_case_sha224_1(self):
 222         self.check('sha224', "abc",
 223           "23097d223405d8228642a477bda255b32aadbce4bda0b3f7e36c9da7")
 224
 225     def test_case_sha224_2(self):
 226         self.check('sha224',
 227           "abcdbcdecdefdefgefghfghighijhijkijkljklmklmnlmnomnopnopq",
 228           "75388b16512776cc5dba5da1fd890150b0c6455cb4f58b1952522525")
 229
 230     def test_case_sha224_3(self):
 231         self.check('sha224', "a" * 1000000,
 232           "20794655980c91d8bbb4c1ea97618a4bf03f42581948b2ee4ee7ad67")
 233
 234
 235     def test_case_sha256_0(self):
 236         self.check('sha256', "",
 237           "e3b0c44298fc1c149afbf4c8996fb92427ae41e4649b934ca495991b7852b855")
 238
 239     def test_case_sha256_1(self):
 240         self.check('sha256', "abc",
 241           "ba7816bf8f01cfea414140de5dae2223b00361a396177a9cb410ff61f20015ad")
 242
 243     def test_case_sha256_2(self):
 244         self.check('sha256',
 245           "abcdbcdecdefdefgefghfghighijhijkijkljklmklmnlmnomnopnopq",
 246           "248d6a61d20638b8e5c026930c3e6039a33ce45964ff2167f6ecedd419db06c1")
 247
 248     def test_case_sha256_3(self):
 249         self.check('sha256', "a" * 1000000,
 250           "cdc76e5c9914fb9281a1c7e284d73e67f1809a48a497200e046d39ccc7112cd0")
 251
 252
 253     def test_case_sha384_0(self):
 254         self.check('sha384', "",
 255           "38b060a751ac96384cd9327eb1b1e36a21fdb71114be07434c0cc7bf63f6e1da"+
 256           "274edebfe76f65fbd51ad2f14898b95b")
 257
 258     def test_case_sha384_1(self):
 259         self.check('sha384', "abc",
 260           "cb00753f45a35e8bb5a03d699ac65007272c32ab0eded1631a8b605a43ff5bed"+
 261           "8086072ba1e7cc2358baeca134c825a7")
 262
 263     def test_case_sha384_2(self):
 264         self.check('sha384',
 265                    "abcdefghbcdefghicdefghijdefghijkefghijklfghijklmghijklmn"+
 266                    "hijklmnoijklmnopjklmnopqklmnopqrlmnopqrsmnopqrstnopqrstu",
 267           "09330c33f71147e83d192fc782cd1b4753111b173b3b05d22fa08086e3b0f712"+
 268           "fcc7c71a557e2db966c3e9fa91746039")
 269
 270     def test_case_sha384_3(self):
 271         self.check('sha384', "a" * 1000000,
 272           "9d0e1809716474cb086e834e310a4a1ced149e9c00f248527972cec5704c2a5b"+
 273           "07b8b3dc38ecc4ebae97ddd87f3d8985")
 274
 275
 276     def test_case_sha512_0(self):
 277         self.check('sha512', "",
 278           "cf83e1357eefb8bdf1542850d66d8007d620e4050b5715dc83f4a921d36ce9ce"+
 279           "47d0d13c5d85f2b0ff8318d2877eec2f63b931bd47417a81a538327af927da3e")
 280
 281     def test_case_sha512_1(self):
 282         self.check('sha512', "abc",
 283           "ddaf35a193617abacc417349ae20413112e6fa4e89a97ea20a9eeee64b55d39a"+
 284           "2192992a274fc1a836ba3c23a3feebbd454d4423643ce80e2a9ac94fa54ca49f")
 285
 286     def test_case_sha512_2(self):
 287         self.check('sha512',
 288                    "abcdefghbcdefghicdefghijdefghijkefghijklfghijklmghijklmn"+
 289                    "hijklmnoijklmnopjklmnopqklmnopqrlmnopqrsmnopqrstnopqrstu",
 290           "8e959b75dae313da8cf4f72814fc143f8f7779c6eb9f7fa17299aeadb6889018"+
 291           "501d289e4900f7e4331b99dec4b5433ac7d329eeb6dd26545e96e55b874be909")
 292
 293     def test_case_sha512_3(self):
 294         self.check('sha512', "a" * 1000000,
 295           "e718483d0ce769644e2e42c7bc15b4638e1f98b13b2044285632a803afa973eb"+
 296           "de0ff244877ea60a4cb0432ce577c31beb009c5c2c49aa2e4eadb217ad8cc09b")
 297
 298     def test_threaded_hashing(self):
 299         if not threading:
 300             raise unittest.SkipTest('No threading module.')
 301
 302         # Updating the same hash object from several threads at once
 303         # using data chunk sizes containing the same byte sequences.
 304         #
 305         # If the internal locks are working to prevent multiple
 306         # updates on the same object from running at once, the resulting
 307         # hash will be the same as doing it single threaded upfront.
 308         hasher = hashlib.sha1()
 309         num_threads = 5
 310         smallest_data = 'swineflu'
 311         data = smallest_data*200000
 312         expected_hash = hashlib.sha1(data*num_threads).hexdigest()
 313
 314         def hash_in_chunks(chunk_size, event):
 315             index = 0
 316             while index < len(data):
 317                 hasher.update(data[index:index+chunk_size])
 318                 index += chunk_size
 319             event.set()
 320
 321         events = []
 322         for threadnum in xrange(num_threads):
 323             chunk_size = len(data) // (10**threadnum)
 324             assert chunk_size > 0
 325             assert chunk_size % len(smallest_data) == 0
 326             event = threading.Event()
 327             events.append(event)
 328             threading.Thread(target=hash_in_chunks,
 329                              args=(chunk_size, event)).start()
 330
 331         for event in events:
 332             event.wait()
 333
 334         self.assertEqual(expected_hash, hasher.hexdigest())
 335
 336 @test_support.reap_threads
 337 def test_main():
 338     test_support.run_unittest(HashLibTestCase)
 339
 340 if __name__ == "__main__":
 341     test_main()