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