Lib/test/test_bigmem.py

   1 from test import test_support
   2 from test.test_support import bigmemtest, _1G, _2G, _4G, precisionbigmemtest
   3
   4 import unittest
   5 import operator
   6 import string
   7 import sys
   8
   9 # Bigmem testing houserules:
  10 #
  11 #  - Try not to allocate too many large objects. It's okay to rely on
  12 #    refcounting semantics, but don't forget that 's = create_largestring()'
  13 #    doesn't release the old 's' (if it exists) until well after its new
  14 #    value has been created. Use 'del s' before the create_largestring call.
  15 #
  16 #  - Do *not* compare large objects using assertEquals or similar. It's a
  17 #    lengty operation and the errormessage will be utterly useless due to
  18 #    its size. To make sure whether a result has the right contents, better
  19 #    to use the strip or count methods, or compare meaningful slices.
  20 #
  21 #  - Don't forget to test for large indices, offsets and results and such,
  22 #    in addition to large sizes.
  23 #
  24 #  - When repeating an object (say, a substring, or a small list) to create
  25 #    a large object, make the subobject of a length that is not a power of
  26 #    2. That way, int-wrapping problems are more easily detected.
  27 #
  28 #  - While the bigmemtest decorator speaks of 'minsize', all tests will
  29 #    actually be called with a much smaller number too, in the normal
  30 #    test run (5Kb currently.) This is so the tests themselves get frequent
  31 #    testing. Consequently, always make all large allocations based on the
  32 #    passed-in 'size', and don't rely on the size being very large. Also,
  33 #    memuse-per-size should remain sane (less than a few thousand); if your
  34 #    test uses more, adjust 'size' upward, instead.
  35
  36 class StrTest(unittest.TestCase):
  37     @bigmemtest(minsize=_2G, memuse=2)
  38     def test_capitalize(self, size):
  39         SUBSTR = ' abc def ghi'
  40         s = '-' * size + SUBSTR
  41         caps = s.capitalize()
  42         self.assertEquals(caps[-len(SUBSTR):],
  43                          SUBSTR.capitalize())
  44         self.assertEquals(caps.lstrip('-'), SUBSTR)
  45
  46     @bigmemtest(minsize=_2G + 10, memuse=1)
  47     def test_center(self, size):
  48         SUBSTR = ' abc def ghi'
  49         s = SUBSTR.center(size)
  50         self.assertEquals(len(s), size)
  51         lpadsize = rpadsize = (len(s) - len(SUBSTR)) // 2
  52         if len(s) % 2:
  53             lpadsize += 1
  54         self.assertEquals(s[lpadsize:-rpadsize], SUBSTR)
  55         self.assertEquals(s.strip(), SUBSTR.strip())
  56
  57     @precisionbigmemtest(size=_2G - 1, memuse=1)
  58     def test_center_unicode(self, size):
  59         SUBSTR = u' abc def ghi'
  60         try:
  61             s = SUBSTR.center(size)
  62         except OverflowError:
  63             pass # acceptable on 32-bit
  64         else:
  65             self.assertEquals(len(s), size)
  66             lpadsize = rpadsize = (len(s) - len(SUBSTR)) // 2
  67             if len(s) % 2:
  68                 lpadsize += 1
  69             self.assertEquals(s[lpadsize:-rpadsize], SUBSTR)
  70             self.assertEquals(s.strip(), SUBSTR.strip())
  71             del s
  72
  73     @bigmemtest(minsize=_2G, memuse=2)
  74     def test_count(self, size):
  75         SUBSTR = ' abc def ghi'
  76         s = '.' * size + SUBSTR
  77         self.assertEquals(s.count('.'), size)
  78         s += '.'
  79         self.assertEquals(s.count('.'), size + 1)
  80         self.assertEquals(s.count(' '), 3)
  81         self.assertEquals(s.count('i'), 1)
  82         self.assertEquals(s.count('j'), 0)
  83
  84     @bigmemtest(minsize=_2G + 2, memuse=3)
  85     def test_decode(self, size):
  86         s = '.' * size
  87         self.assertEquals(len(s.decode('utf-8')), size)
  88
  89     def basic_encode_test(self, size, enc, c=u'.', expectedsize=None):
  90         if expectedsize is None:
  91             expectedsize = size
  92
  93         s = c * size
  94         self.assertEquals(len(s.encode(enc)), expectedsize)
  95
  96     @bigmemtest(minsize=_2G + 2, memuse=3)
  97     def test_encode(self, size):
  98         return self.basic_encode_test(size, 'utf-8')
  99
 100     @precisionbigmemtest(size=_4G / 6 + 2, memuse=2)
 101     def test_encode_raw_unicode_escape(self, size):
 102         try:
 103             return self.basic_encode_test(size, 'raw_unicode_escape')
 104         except MemoryError:
 105             pass # acceptable on 32-bit
 106
 107     @precisionbigmemtest(size=_4G / 5 + 70, memuse=3)
 108     def test_encode_utf7(self, size):
 109         try:
 110             return self.basic_encode_test(size, 'utf7')
 111         except MemoryError:
 112             pass # acceptable on 32-bit
 113
 114     @precisionbigmemtest(size=_4G / 4 + 5, memuse=6)
 115     def test_encode_utf32(self, size):
 116         try:
 117             return self.basic_encode_test(size, 'utf32', expectedsize=4*size+4)
 118         except MemoryError:
 119             pass # acceptable on 32-bit
 120
 121     @precisionbigmemtest(size=_2G-1, memuse=2)
 122     def test_decodeascii(self, size):
 123         return self.basic_encode_test(size, 'ascii', c='A')
 124
 125     @precisionbigmemtest(size=_4G / 5, memuse=6+2)
 126     def test_unicode_repr_oflw(self, size):
 127         try:
 128             s = u"\uAAAA"*size
 129             r = repr(s)
 130         except MemoryError:
 131             pass # acceptable on 32-bit
 132         else:
 133             self.failUnless(s == eval(r))
 134
 135     @bigmemtest(minsize=_2G, memuse=2)
 136     def test_endswith(self, size):
 137         SUBSTR = ' abc def ghi'
 138         s = '-' * size + SUBSTR
 139         self.failUnless(s.endswith(SUBSTR))
 140         self.failUnless(s.endswith(s))
 141         s2 = '...' + s
 142         self.failUnless(s2.endswith(s))
 143         self.failIf(s.endswith('a' + SUBSTR))
 144         self.failIf(SUBSTR.endswith(s))
 145
 146     @bigmemtest(minsize=_2G + 10, memuse=2)
 147     def test_expandtabs(self, size):
 148         s = '-' * size
 149         tabsize = 8
 150         self.assertEquals(s.expandtabs(), s)
 151         del s
 152         slen, remainder = divmod(size, tabsize)
 153         s = '       \t' * slen
 154         s = s.expandtabs(tabsize)
 155         self.assertEquals(len(s), size - remainder)
 156         self.assertEquals(len(s.strip(' ')), 0)
 157
 158     @bigmemtest(minsize=_2G, memuse=2)
 159     def test_find(self, size):
 160         SUBSTR = ' abc def ghi'
 161         sublen = len(SUBSTR)
 162         s = ''.join([SUBSTR, '-' * size, SUBSTR])
 163         self.assertEquals(s.find(' '), 0)
 164         self.assertEquals(s.find(SUBSTR), 0)
 165         self.assertEquals(s.find(' ', sublen), sublen + size)
 166         self.assertEquals(s.find(SUBSTR, len(SUBSTR)), sublen + size)
 167         self.assertEquals(s.find('i'), SUBSTR.find('i'))
 168         self.assertEquals(s.find('i', sublen),
 169                          sublen + size + SUBSTR.find('i'))
 170         self.assertEquals(s.find('i', size),
 171                          sublen + size + SUBSTR.find('i'))
 172         self.assertEquals(s.find('j'), -1)
 173
 174     @bigmemtest(minsize=_2G, memuse=2)
 175     def test_index(self, size):
 176         SUBSTR = ' abc def ghi'
 177         sublen = len(SUBSTR)
 178         s = ''.join([SUBSTR, '-' * size, SUBSTR])
 179         self.assertEquals(s.index(' '), 0)
 180         self.assertEquals(s.index(SUBSTR), 0)
 181         self.assertEquals(s.index(' ', sublen), sublen + size)
 182         self.assertEquals(s.index(SUBSTR, sublen), sublen + size)
 183         self.assertEquals(s.index('i'), SUBSTR.index('i'))
 184         self.assertEquals(s.index('i', sublen),
 185                          sublen + size + SUBSTR.index('i'))
 186         self.assertEquals(s.index('i', size),
 187                          sublen + size + SUBSTR.index('i'))
 188         self.assertRaises(ValueError, s.index, 'j')
 189
 190     @bigmemtest(minsize=_2G, memuse=2)
 191     def test_isalnum(self, size):
 192         SUBSTR = '123456'
 193         s = 'a' * size + SUBSTR
 194         self.failUnless(s.isalnum())
 195         s += '.'
 196         self.failIf(s.isalnum())
 197
 198     @bigmemtest(minsize=_2G, memuse=2)
 199     def test_isalpha(self, size):
 200         SUBSTR = 'zzzzzzz'
 201         s = 'a' * size + SUBSTR
 202         self.failUnless(s.isalpha())
 203         s += '.'
 204         self.failIf(s.isalpha())
 205
 206     @bigmemtest(minsize=_2G, memuse=2)
 207     def test_isdigit(self, size):
 208         SUBSTR = '123456'
 209         s = '9' * size + SUBSTR
 210         self.failUnless(s.isdigit())
 211         s += 'z'
 212         self.failIf(s.isdigit())
 213
 214     @bigmemtest(minsize=_2G, memuse=2)
 215     def test_islower(self, size):
 216         chars = ''.join([ chr(c) for c in range(255) if not chr(c).isupper() ])
 217         repeats = size // len(chars) + 2
 218         s = chars * repeats
 219         self.failUnless(s.islower())
 220         s += 'A'
 221         self.failIf(s.islower())
 222
 223     @bigmemtest(minsize=_2G, memuse=2)
 224     def test_isspace(self, size):
 225         whitespace = ' \f\n\r\t\v'
 226         repeats = size // len(whitespace) + 2
 227         s = whitespace * repeats
 228         self.failUnless(s.isspace())
 229         s += 'j'
 230         self.failIf(s.isspace())
 231
 232     @bigmemtest(minsize=_2G, memuse=2)
 233     def test_istitle(self, size):
 234         SUBSTR = '123456'
 235         s = ''.join(['A', 'a' * size, SUBSTR])
 236         self.failUnless(s.istitle())
 237         s += 'A'
 238         self.failUnless(s.istitle())
 239         s += 'aA'
 240         self.failIf(s.istitle())
 241
 242     @bigmemtest(minsize=_2G, memuse=2)
 243     def test_isupper(self, size):
 244         chars = ''.join([ chr(c) for c in range(255) if not chr(c).islower() ])
 245         repeats = size // len(chars) + 2
 246         s = chars * repeats
 247         self.failUnless(s.isupper())
 248         s += 'a'
 249         self.failIf(s.isupper())
 250
 251     @bigmemtest(minsize=_2G, memuse=2)
 252     def test_join(self, size):
 253         s = 'A' * size
 254         x = s.join(['aaaaa', 'bbbbb'])
 255         self.assertEquals(x.count('a'), 5)
 256         self.assertEquals(x.count('b'), 5)
 257         self.failUnless(x.startswith('aaaaaA'))
 258         self.failUnless(x.endswith('Abbbbb'))
 259
 260     @bigmemtest(minsize=_2G + 10, memuse=1)
 261     def test_ljust(self, size):
 262         SUBSTR = ' abc def ghi'
 263         s = SUBSTR.ljust(size)
 264         self.failUnless(s.startswith(SUBSTR + '  '))
 265         self.assertEquals(len(s), size)
 266         self.assertEquals(s.strip(), SUBSTR.strip())
 267
 268     @bigmemtest(minsize=_2G + 10, memuse=2)
 269     def test_lower(self, size):
 270         s = 'A' * size
 271         s = s.lower()
 272         self.assertEquals(len(s), size)
 273         self.assertEquals(s.count('a'), size)
 274
 275     @bigmemtest(minsize=_2G + 10, memuse=1)
 276     def test_lstrip(self, size):
 277         SUBSTR = 'abc def ghi'
 278         s = SUBSTR.rjust(size)
 279         self.assertEquals(len(s), size)
 280         self.assertEquals(s.lstrip(), SUBSTR.lstrip())
 281         del s
 282         s = SUBSTR.ljust(size)
 283         self.assertEquals(len(s), size)
 284         stripped = s.lstrip()
 285         self.failUnless(stripped is s)
 286
 287     @bigmemtest(minsize=_2G + 10, memuse=2)
 288     def test_replace(self, size):
 289         replacement = 'a'
 290         s = ' ' * size
 291         s = s.replace(' ', replacement)
 292         self.assertEquals(len(s), size)
 293         self.assertEquals(s.count(replacement), size)
 294         s = s.replace(replacement, ' ', size - 4)
 295         self.assertEquals(len(s), size)
 296         self.assertEquals(s.count(replacement), 4)
 297         self.assertEquals(s[-10:], '      aaaa')
 298
 299     @bigmemtest(minsize=_2G, memuse=2)
 300     def test_rfind(self, size):
 301         SUBSTR = ' abc def ghi'
 302         sublen = len(SUBSTR)
 303         s = ''.join([SUBSTR, '-' * size, SUBSTR])
 304         self.assertEquals(s.rfind(' '), sublen + size + SUBSTR.rfind(' '))
 305         self.assertEquals(s.rfind(SUBSTR), sublen + size)
 306         self.assertEquals(s.rfind(' ', 0, size), SUBSTR.rfind(' '))
 307         self.assertEquals(s.rfind(SUBSTR, 0, sublen + size), 0)
 308         self.assertEquals(s.rfind('i'), sublen + size + SUBSTR.rfind('i'))
 309         self.assertEquals(s.rfind('i', 0, sublen), SUBSTR.rfind('i'))
 310         self.assertEquals(s.rfind('i', 0, sublen + size),
 311                           SUBSTR.rfind('i'))
 312         self.assertEquals(s.rfind('j'), -1)
 313
 314     @bigmemtest(minsize=_2G, memuse=2)
 315     def test_rindex(self, size):
 316         SUBSTR = ' abc def ghi'
 317         sublen = len(SUBSTR)
 318         s = ''.join([SUBSTR, '-' * size, SUBSTR])
 319         self.assertEquals(s.rindex(' '),
 320                           sublen + size + SUBSTR.rindex(' '))
 321         self.assertEquals(s.rindex(SUBSTR), sublen + size)
 322         self.assertEquals(s.rindex(' ', 0, sublen + size - 1),
 323                           SUBSTR.rindex(' '))
 324         self.assertEquals(s.rindex(SUBSTR, 0, sublen + size), 0)
 325         self.assertEquals(s.rindex('i'),
 326                           sublen + size + SUBSTR.rindex('i'))
 327         self.assertEquals(s.rindex('i', 0, sublen), SUBSTR.rindex('i'))
 328         self.assertEquals(s.rindex('i', 0, sublen + size),
 329                           SUBSTR.rindex('i'))
 330         self.assertRaises(ValueError, s.rindex, 'j')
 331
 332     @bigmemtest(minsize=_2G + 10, memuse=1)
 333     def test_rjust(self, size):
 334         SUBSTR = ' abc def ghi'
 335         s = SUBSTR.ljust(size)
 336         self.failUnless(s.startswith(SUBSTR + '  '))
 337         self.assertEquals(len(s), size)
 338         self.assertEquals(s.strip(), SUBSTR.strip())
 339
 340     @bigmemtest(minsize=_2G + 10, memuse=1)
 341     def test_rstrip(self, size):
 342         SUBSTR = ' abc def ghi'
 343         s = SUBSTR.ljust(size)
 344         self.assertEquals(len(s), size)
 345         self.assertEquals(s.rstrip(), SUBSTR.rstrip())
 346         del s
 347         s = SUBSTR.rjust(size)
 348         self.assertEquals(len(s), size)
 349         stripped = s.rstrip()
 350         self.failUnless(stripped is s)
 351
 352     # The test takes about size bytes to build a string, and then about
 353     # sqrt(size) substrings of sqrt(size) in size and a list to
 354     # hold sqrt(size) items. It's close but just over 2x size.
 355     @bigmemtest(minsize=_2G, memuse=2.1)
 356     def test_split_small(self, size):
 357         # Crudely calculate an estimate so that the result of s.split won't
 358         # take up an inordinate amount of memory
 359         chunksize = int(size ** 0.5 + 2)
 360         SUBSTR = 'a' + ' ' * chunksize
 361         s = SUBSTR * chunksize
 362         l = s.split()
 363         self.assertEquals(len(l), chunksize)
 364         self.assertEquals(set(l), set(['a']))
 365         del l
 366         l = s.split('a')
 367         self.assertEquals(len(l), chunksize + 1)
 368         self.assertEquals(set(l), set(['', ' ' * chunksize]))
 369
 370     # Allocates a string of twice size (and briefly two) and a list of
 371     # size.  Because of internal affairs, the s.split() call produces a
 372     # list of size times the same one-character string, so we only
 373     # suffer for the list size. (Otherwise, it'd cost another 48 times
 374     # size in bytes!) Nevertheless, a list of size takes
 375     # 8*size bytes.
 376     @bigmemtest(minsize=_2G + 5, memuse=10)
 377     def test_split_large(self, size):
 378         s = ' a' * size + ' '
 379         l = s.split()
 380         self.assertEquals(len(l), size)
 381         self.assertEquals(set(l), set(['a']))
 382         del l
 383         l = s.split('a')
 384         self.assertEquals(len(l), size + 1)
 385         self.assertEquals(set(l), set([' ']))
 386
 387     @bigmemtest(minsize=_2G, memuse=2.1)
 388     def test_splitlines(self, size):
 389         # Crudely calculate an estimate so that the result of s.split won't
 390         # take up an inordinate amount of memory
 391         chunksize = int(size ** 0.5 + 2) // 2
 392         SUBSTR = ' ' * chunksize + '\n' + ' ' * chunksize + '\r\n'
 393         s = SUBSTR * chunksize
 394         l = s.splitlines()
 395         self.assertEquals(len(l), chunksize * 2)
 396         self.assertEquals(set(l), set([' ' * chunksize]))
 397
 398     @bigmemtest(minsize=_2G, memuse=2)
 399     def test_startswith(self, size):
 400         SUBSTR = ' abc def ghi'
 401         s = '-' * size + SUBSTR
 402         self.failUnless(s.startswith(s))
 403         self.failUnless(s.startswith('-' * size))
 404         self.failIf(s.startswith(SUBSTR))
 405
 406     @bigmemtest(minsize=_2G, memuse=1)
 407     def test_strip(self, size):
 408         SUBSTR = '   abc def ghi   '
 409         s = SUBSTR.rjust(size)
 410         self.assertEquals(len(s), size)
 411         self.assertEquals(s.strip(), SUBSTR.strip())
 412         del s
 413         s = SUBSTR.ljust(size)
 414         self.assertEquals(len(s), size)
 415         self.assertEquals(s.strip(), SUBSTR.strip())
 416
 417     @bigmemtest(minsize=_2G, memuse=2)
 418     def test_swapcase(self, size):
 419         SUBSTR = "aBcDeFG12.'\xa9\x00"
 420         sublen = len(SUBSTR)
 421         repeats = size // sublen + 2
 422         s = SUBSTR * repeats
 423         s = s.swapcase()
 424         self.assertEquals(len(s), sublen * repeats)
 425         self.assertEquals(s[:sublen * 3], SUBSTR.swapcase() * 3)
 426         self.assertEquals(s[-sublen * 3:], SUBSTR.swapcase() * 3)
 427
 428     @bigmemtest(minsize=_2G, memuse=2)
 429     def test_title(self, size):
 430         SUBSTR = 'SpaaHAaaAaham'
 431         s = SUBSTR * (size // len(SUBSTR) + 2)
 432         s = s.title()
 433         self.failUnless(s.startswith((SUBSTR * 3).title()))
 434         self.failUnless(s.endswith(SUBSTR.lower() * 3))
 435
 436     @bigmemtest(minsize=_2G, memuse=2)
 437     def test_translate(self, size):
 438         trans = string.maketrans('.aZ', '-!$')
 439         SUBSTR = 'aZz.z.Aaz.'
 440         sublen = len(SUBSTR)
 441         repeats = size // sublen + 2
 442         s = SUBSTR * repeats
 443         s = s.translate(trans)
 444         self.assertEquals(len(s), repeats * sublen)
 445         self.assertEquals(s[:sublen], SUBSTR.translate(trans))
 446         self.assertEquals(s[-sublen:], SUBSTR.translate(trans))
 447         self.assertEquals(s.count('.'), 0)
 448         self.assertEquals(s.count('!'), repeats * 2)
 449         self.assertEquals(s.count('z'), repeats * 3)
 450
 451     @bigmemtest(minsize=_2G + 5, memuse=2)
 452     def test_upper(self, size):
 453         s = 'a' * size
 454         s = s.upper()
 455         self.assertEquals(len(s), size)
 456         self.assertEquals(s.count('A'), size)
 457
 458     @bigmemtest(minsize=_2G + 20, memuse=1)
 459     def test_zfill(self, size):
 460         SUBSTR = '-568324723598234'
 461         s = SUBSTR.zfill(size)
 462         self.failUnless(s.endswith('0' + SUBSTR[1:]))
 463         self.failUnless(s.startswith('-0'))
 464         self.assertEquals(len(s), size)
 465         self.assertEquals(s.count('0'), size - len(SUBSTR))
 466
 467     @bigmemtest(minsize=_2G + 10, memuse=2)
 468     def test_format(self, size):
 469         s = '-' * size
 470         sf = '%s' % (s,)
 471         self.failUnless(s == sf)
 472         del sf
 473         sf = '..%s..' % (s,)
 474         self.assertEquals(len(sf), len(s) + 4)
 475         self.failUnless(sf.startswith('..-'))
 476         self.failUnless(sf.endswith('-..'))
 477         del s, sf
 478
 479         size //= 2
 480         edge = '-' * size
 481         s = ''.join([edge, '%s', edge])
 482         del edge
 483         s = s % '...'
 484         self.assertEquals(len(s), size * 2 + 3)
 485         self.assertEquals(s.count('.'), 3)
 486         self.assertEquals(s.count('-'), size * 2)
 487
 488     @bigmemtest(minsize=_2G + 10, memuse=2)
 489     def test_repr_small(self, size):
 490         s = '-' * size
 491         s = repr(s)
 492         self.assertEquals(len(s), size + 2)
 493         self.assertEquals(s[0], "'")
 494         self.assertEquals(s[-1], "'")
 495         self.assertEquals(s.count('-'), size)
 496         del s
 497         # repr() will create a string four times as large as this 'binary
 498         # string', but we don't want to allocate much more than twice
 499         # size in total.  (We do extra testing in test_repr_large())
 500         size = size // 5 * 2
 501         s = '\x00' * size
 502         s = repr(s)
 503         self.assertEquals(len(s), size * 4 + 2)
 504         self.assertEquals(s[0], "'")
 505         self.assertEquals(s[-1], "'")
 506         self.assertEquals(s.count('\\'), size)
 507         self.assertEquals(s.count('0'), size * 2)
 508
 509     @bigmemtest(minsize=_2G + 10, memuse=5)
 510     def test_repr_large(self, size):
 511         s = '\x00' * size
 512         s = repr(s)
 513         self.assertEquals(len(s), size * 4 + 2)
 514         self.assertEquals(s[0], "'")
 515         self.assertEquals(s[-1], "'")
 516         self.assertEquals(s.count('\\'), size)
 517         self.assertEquals(s.count('0'), size * 2)
 518
 519     @bigmemtest(minsize=2**32 / 5, memuse=6+2)
 520     def test_unicode_repr(self, size):
 521         s = u"\uAAAA" * size
 522         self.failUnless(len(repr(s)) > size)
 523
 524     # This test is meaningful even with size < 2G, as long as the
 525     # doubled string is > 2G (but it tests more if both are > 2G :)
 526     @bigmemtest(minsize=_1G + 2, memuse=3)
 527     def test_concat(self, size):
 528         s = '.' * size
 529         self.assertEquals(len(s), size)
 530         s = s + s
 531         self.assertEquals(len(s), size * 2)
 532         self.assertEquals(s.count('.'), size * 2)
 533
 534     # This test is meaningful even with size < 2G, as long as the
 535     # repeated string is > 2G (but it tests more if both are > 2G :)
 536     @bigmemtest(minsize=_1G + 2, memuse=3)
 537     def test_repeat(self, size):
 538         s = '.' * size
 539         self.assertEquals(len(s), size)
 540         s = s * 2
 541         self.assertEquals(len(s), size * 2)
 542         self.assertEquals(s.count('.'), size * 2)
 543
 544     @bigmemtest(minsize=_2G + 20, memuse=1)
 545     def test_slice_and_getitem(self, size):
 546         SUBSTR = '0123456789'
 547         sublen = len(SUBSTR)
 548         s = SUBSTR * (size // sublen)
 549         stepsize = len(s) // 100
 550         stepsize = stepsize - (stepsize % sublen)
 551         for i in range(0, len(s) - stepsize, stepsize):
 552             self.assertEquals(s[i], SUBSTR[0])
 553             self.assertEquals(s[i:i + sublen], SUBSTR)
 554             self.assertEquals(s[i:i + sublen:2], SUBSTR[::2])
 555             if i > 0:
 556                 self.assertEquals(s[i + sublen - 1:i - 1:-3],
 557                                   SUBSTR[sublen::-3])
 558         # Make sure we do some slicing and indexing near the end of the
 559         # string, too.
 560         self.assertEquals(s[len(s) - 1], SUBSTR[-1])
 561         self.assertEquals(s[-1], SUBSTR[-1])
 562         self.assertEquals(s[len(s) - 10], SUBSTR[0])
 563         self.assertEquals(s[-sublen], SUBSTR[0])
 564         self.assertEquals(s[len(s):], '')
 565         self.assertEquals(s[len(s) - 1:], SUBSTR[-1])
 566         self.assertEquals(s[-1:], SUBSTR[-1])
 567         self.assertEquals(s[len(s) - sublen:], SUBSTR)
 568         self.assertEquals(s[-sublen:], SUBSTR)
 569         self.assertEquals(len(s[:]), len(s))
 570         self.assertEquals(len(s[:len(s) - 5]), len(s) - 5)
 571         self.assertEquals(len(s[5:-5]), len(s) - 10)
 572
 573         self.assertRaises(IndexError, operator.getitem, s, len(s))
 574         self.assertRaises(IndexError, operator.getitem, s, len(s) + 1)
 575         self.assertRaises(IndexError, operator.getitem, s, len(s) + 1<<31)
 576
 577     @bigmemtest(minsize=_2G, memuse=2)
 578     def test_contains(self, size):
 579         SUBSTR = '0123456789'
 580         edge = '-' * (size // 2)
 581         s = ''.join([edge, SUBSTR, edge])
 582         del edge
 583         self.failUnless(SUBSTR in s)
 584         self.failIf(SUBSTR * 2 in s)
 585         self.failUnless('-' in s)
 586         self.failIf('a' in s)
 587         s += 'a'
 588         self.failUnless('a' in s)
 589
 590     @bigmemtest(minsize=_2G + 10, memuse=2)
 591     def test_compare(self, size):
 592         s1 = '-' * size
 593         s2 = '-' * size
 594         self.failUnless(s1 == s2)
 595         del s2
 596         s2 = s1 + 'a'
 597         self.failIf(s1 == s2)
 598         del s2
 599         s2 = '.' * size
 600         self.failIf(s1 == s2)
 601
 602     @bigmemtest(minsize=_2G + 10, memuse=1)
 603     def test_hash(self, size):
 604         # Not sure if we can do any meaningful tests here...  Even if we
 605         # start relying on the exact algorithm used, the result will be
 606         # different depending on the size of the C 'long int'.  Even this
 607         # test is dodgy (there's no *guarantee* that the two things should
 608         # have a different hash, even if they, in the current
 609         # implementation, almost always do.)
 610         s = '\x00' * size
 611         h1 = hash(s)
 612         del s
 613         s = '\x00' * (size + 1)
 614         self.failIf(h1 == hash(s))
 615
 616 class TupleTest(unittest.TestCase):
 617
 618     # Tuples have a small, fixed-sized head and an array of pointers to
 619     # data.  Since we're testing 64-bit addressing, we can assume that the
 620     # pointers are 8 bytes, and that thus that the tuples take up 8 bytes
 621     # per size.
 622
 623     # As a side-effect of testing long tuples, these tests happen to test
 624     # having more than 2<<31 references to any given object. Hence the
 625     # use of different types of objects as contents in different tests.
 626
 627     @bigmemtest(minsize=_2G + 2, memuse=16)
 628     def test_compare(self, size):
 629         t1 = (u'',) * size
 630         t2 = (u'',) * size
 631         self.failUnless(t1 == t2)
 632         del t2
 633         t2 = (u'',) * (size + 1)
 634         self.failIf(t1 == t2)
 635         del t2
 636         t2 = (1,) * size
 637         self.failIf(t1 == t2)
 638
 639     # Test concatenating into a single tuple of more than 2G in length,
 640     # and concatenating a tuple of more than 2G in length separately, so
 641     # the smaller test still gets run even if there isn't memory for the
 642     # larger test (but we still let the tester know the larger test is
 643     # skipped, in verbose mode.)
 644     def basic_concat_test(self, size):
 645         t = ((),) * size
 646         self.assertEquals(len(t), size)
 647         t = t + t
 648         self.assertEquals(len(t), size * 2)
 649
 650     @bigmemtest(minsize=_2G // 2 + 2, memuse=24)
 651     def test_concat_small(self, size):
 652         return self.basic_concat_test(size)
 653
 654     @bigmemtest(minsize=_2G + 2, memuse=24)
 655     def test_concat_large(self, size):
 656         return self.basic_concat_test(size)
 657
 658     @bigmemtest(minsize=_2G // 5 + 10, memuse=8 * 5)
 659     def test_contains(self, size):
 660         t = (1, 2, 3, 4, 5) * size
 661         self.assertEquals(len(t), size * 5)
 662         self.failUnless(5 in t)
 663         self.failIf((1, 2, 3, 4, 5) in t)
 664         self.failIf(0 in t)
 665
 666     @bigmemtest(minsize=_2G + 10, memuse=8)
 667     def test_hash(self, size):
 668         t1 = (0,) * size
 669         h1 = hash(t1)
 670         del t1
 671         t2 = (0,) * (size + 1)
 672         self.failIf(h1 == hash(t2))
 673
 674     @bigmemtest(minsize=_2G + 10, memuse=8)
 675     def test_index_and_slice(self, size):
 676         t = (None,) * size
 677         self.assertEquals(len(t), size)
 678         self.assertEquals(t[-1], None)
 679         self.assertEquals(t[5], None)
 680         self.assertEquals(t[size - 1], None)
 681         self.assertRaises(IndexError, operator.getitem, t, size)
 682         self.assertEquals(t[:5], (None,) * 5)
 683         self.assertEquals(t[-5:], (None,) * 5)
 684         self.assertEquals(t[20:25], (None,) * 5)
 685         self.assertEquals(t[-25:-20], (None,) * 5)
 686         self.assertEquals(t[size - 5:], (None,) * 5)
 687         self.assertEquals(t[size - 5:size], (None,) * 5)
 688         self.assertEquals(t[size - 6:size - 2], (None,) * 4)
 689         self.assertEquals(t[size:size], ())
 690         self.assertEquals(t[size:size+5], ())
 691
 692     # Like test_concat, split in two.
 693     def basic_test_repeat(self, size):
 694         t = ('',) * size
 695         self.assertEquals(len(t), size)
 696         t = t * 2
 697         self.assertEquals(len(t), size * 2)
 698
 699     @bigmemtest(minsize=_2G // 2 + 2, memuse=24)
 700     def test_repeat_small(self, size):
 701         return self.basic_test_repeat(size)
 702
 703     @bigmemtest(minsize=_2G + 2, memuse=24)
 704     def test_repeat_large(self, size):
 705         return self.basic_test_repeat(size)
 706
 707     @bigmemtest(minsize=_1G - 1, memuse=12)
 708     def test_repeat_large_2(self, size):
 709         return self.basic_test_repeat(size)
 710
 711     @precisionbigmemtest(size=_1G - 1, memuse=9)
 712     def test_from_2G_generator(self, size):
 713         try:
 714             t = tuple(xrange(size))
 715         except MemoryError:
 716             pass # acceptable on 32-bit
 717         else:
 718             count = 0
 719             for item in t:
 720                 self.assertEquals(item, count)
 721                 count += 1
 722             self.assertEquals(count, size)
 723
 724     @precisionbigmemtest(size=_1G - 25, memuse=9)
 725     def test_from_almost_2G_generator(self, size):
 726         try:
 727             t = tuple(xrange(size))
 728             count = 0
 729             for item in t:
 730                 self.assertEquals(item, count)
 731                 count += 1
 732             self.assertEquals(count, size)
 733         except MemoryError:
 734             pass # acceptable, expected on 32-bit
 735
 736     # Like test_concat, split in two.
 737     def basic_test_repr(self, size):
 738         t = (0,) * size
 739         s = repr(t)
 740         # The repr of a tuple of 0's is exactly three times the tuple length.
 741         self.assertEquals(len(s), size * 3)
 742         self.assertEquals(s[:5], '(0, 0')
 743         self.assertEquals(s[-5:], '0, 0)')
 744         self.assertEquals(s.count('0'), size)
 745
 746     @bigmemtest(minsize=_2G // 3 + 2, memuse=8 + 3)
 747     def test_repr_small(self, size):
 748         return self.basic_test_repr(size)
 749
 750     @bigmemtest(minsize=_2G + 2, memuse=8 + 3)
 751     def test_repr_large(self, size):
 752         return self.basic_test_repr(size)
 753
 754 class ListTest(unittest.TestCase):
 755
 756     # Like tuples, lists have a small, fixed-sized head and an array of
 757     # pointers to data, so 8 bytes per size. Also like tuples, we make the
 758     # lists hold references to various objects to test their refcount
 759     # limits.
 760
 761     @bigmemtest(minsize=_2G + 2, memuse=16)
 762     def test_compare(self, size):
 763         l1 = [u''] * size
 764         l2 = [u''] * size
 765         self.failUnless(l1 == l2)
 766         del l2
 767         l2 = [u''] * (size + 1)
 768         self.failIf(l1 == l2)
 769         del l2
 770         l2 = [2] * size
 771         self.failIf(l1 == l2)
 772
 773     # Test concatenating into a single list of more than 2G in length,
 774     # and concatenating a list of more than 2G in length separately, so
 775     # the smaller test still gets run even if there isn't memory for the
 776     # larger test (but we still let the tester know the larger test is
 777     # skipped, in verbose mode.)
 778     def basic_test_concat(self, size):
 779         l = [[]] * size
 780         self.assertEquals(len(l), size)
 781         l = l + l
 782         self.assertEquals(len(l), size * 2)
 783
 784     @bigmemtest(minsize=_2G // 2 + 2, memuse=24)
 785     def test_concat_small(self, size):
 786         return self.basic_test_concat(size)
 787
 788     @bigmemtest(minsize=_2G + 2, memuse=24)
 789     def test_concat_large(self, size):
 790         return self.basic_test_concat(size)
 791
 792     def basic_test_inplace_concat(self, size):
 793         l = [sys.stdout] * size
 794         l += l
 795         self.assertEquals(len(l), size * 2)
 796         self.failUnless(l[0] is l[-1])
 797         self.failUnless(l[size - 1] is l[size + 1])
 798
 799     @bigmemtest(minsize=_2G // 2 + 2, memuse=24)
 800     def test_inplace_concat_small(self, size):
 801         return self.basic_test_inplace_concat(size)
 802
 803     @bigmemtest(minsize=_2G + 2, memuse=24)
 804     def test_inplace_concat_large(self, size):
 805         return self.basic_test_inplace_concat(size)
 806
 807     @bigmemtest(minsize=_2G // 5 + 10, memuse=8 * 5)
 808     def test_contains(self, size):
 809         l = [1, 2, 3, 4, 5] * size
 810         self.assertEquals(len(l), size * 5)
 811         self.failUnless(5 in l)
 812         self.failIf([1, 2, 3, 4, 5] in l)
 813         self.failIf(0 in l)
 814
 815     @bigmemtest(minsize=_2G + 10, memuse=8)
 816     def test_hash(self, size):
 817         l = [0] * size
 818         self.failUnlessRaises(TypeError, hash, l)
 819
 820     @bigmemtest(minsize=_2G + 10, memuse=8)
 821     def test_index_and_slice(self, size):
 822         l = [None] * size
 823         self.assertEquals(len(l), size)
 824         self.assertEquals(l[-1], None)
 825         self.assertEquals(l[5], None)
 826         self.assertEquals(l[size - 1], None)
 827         self.assertRaises(IndexError, operator.getitem, l, size)
 828         self.assertEquals(l[:5], [None] * 5)
 829         self.assertEquals(l[-5:], [None] * 5)
 830         self.assertEquals(l[20:25], [None] * 5)
 831         self.assertEquals(l[-25:-20], [None] * 5)
 832         self.assertEquals(l[size - 5:], [None] * 5)
 833         self.assertEquals(l[size - 5:size], [None] * 5)
 834         self.assertEquals(l[size - 6:size - 2], [None] * 4)
 835         self.assertEquals(l[size:size], [])
 836         self.assertEquals(l[size:size+5], [])
 837
 838         l[size - 2] = 5
 839         self.assertEquals(len(l), size)
 840         self.assertEquals(l[-3:], [None, 5, None])
 841         self.assertEquals(l.count(5), 1)
 842         self.assertRaises(IndexError, operator.setitem, l, size, 6)
 843         self.assertEquals(len(l), size)
 844
 845         l[size - 7:] = [1, 2, 3, 4, 5]
 846         size -= 2
 847         self.assertEquals(len(l), size)
 848         self.assertEquals(l[-7:], [None, None, 1, 2, 3, 4, 5])
 849
 850         l[:7] = [1, 2, 3, 4, 5]
 851         size -= 2
 852         self.assertEquals(len(l), size)
 853         self.assertEquals(l[:7], [1, 2, 3, 4, 5, None, None])
 854
 855         del l[size - 1]
 856         size -= 1
 857         self.assertEquals(len(l), size)
 858         self.assertEquals(l[-1], 4)
 859
 860         del l[-2:]
 861         size -= 2
 862         self.assertEquals(len(l), size)
 863         self.assertEquals(l[-1], 2)
 864
 865         del l[0]
 866         size -= 1
 867         self.assertEquals(len(l), size)
 868         self.assertEquals(l[0], 2)
 869
 870         del l[:2]
 871         size -= 2
 872         self.assertEquals(len(l), size)
 873         self.assertEquals(l[0], 4)
 874
 875     # Like test_concat, split in two.
 876     def basic_test_repeat(self, size):
 877         l = [] * size
 878         self.failIf(l)
 879         l = [''] * size
 880         self.assertEquals(len(l), size)
 881         l = l * 2
 882         self.assertEquals(len(l), size * 2)
 883
 884     @bigmemtest(minsize=_2G // 2 + 2, memuse=24)
 885     def test_repeat_small(self, size):
 886         return self.basic_test_repeat(size)
 887
 888     @bigmemtest(minsize=_2G + 2, memuse=24)
 889     def test_repeat_large(self, size):
 890         return self.basic_test_repeat(size)
 891
 892     def basic_test_inplace_repeat(self, size):
 893         l = ['']
 894         l *= size
 895         self.assertEquals(len(l), size)
 896         self.failUnless(l[0] is l[-1])
 897         del l
 898
 899         l = [''] * size
 900         l *= 2
 901         self.assertEquals(len(l), size * 2)
 902         self.failUnless(l[size - 1] is l[-1])
 903
 904     @bigmemtest(minsize=_2G // 2 + 2, memuse=16)
 905     def test_inplace_repeat_small(self, size):
 906         return self.basic_test_inplace_repeat(size)
 907
 908     @bigmemtest(minsize=_2G + 2, memuse=16)
 909     def test_inplace_repeat_large(self, size):
 910         return self.basic_test_inplace_repeat(size)
 911
 912     def basic_test_repr(self, size):
 913         l = [0] * size
 914         s = repr(l)
 915         # The repr of a list of 0's is exactly three times the list length.
 916         self.assertEquals(len(s), size * 3)
 917         self.assertEquals(s[:5], '[0, 0')
 918         self.assertEquals(s[-5:], '0, 0]')
 919         self.assertEquals(s.count('0'), size)
 920
 921     @bigmemtest(minsize=_2G // 3 + 2, memuse=8 + 3)
 922     def test_repr_small(self, size):
 923         return self.basic_test_repr(size)
 924
 925     @bigmemtest(minsize=_2G + 2, memuse=8 + 3)
 926     def test_repr_large(self, size):
 927         return self.basic_test_repr(size)
 928
 929     # list overallocates ~1/8th of the total size (on first expansion) so
 930     # the single list.append call puts memuse at 9 bytes per size.
 931     @bigmemtest(minsize=_2G, memuse=9)
 932     def test_append(self, size):
 933         l = [object()] * size
 934         l.append(object())
 935         self.assertEquals(len(l), size+1)
 936         self.failUnless(l[-3] is l[-2])
 937         self.failIf(l[-2] is l[-1])
 938
 939     @bigmemtest(minsize=_2G // 5 + 2, memuse=8 * 5)
 940     def test_count(self, size):
 941         l = [1, 2, 3, 4, 5] * size
 942         self.assertEquals(l.count(1), size)
 943         self.assertEquals(l.count("1"), 0)
 944
 945     def basic_test_extend(self, size):
 946         l = [file] * size
 947         l.extend(l)
 948         self.assertEquals(len(l), size * 2)
 949         self.failUnless(l[0] is l[-1])
 950         self.failUnless(l[size - 1] is l[size + 1])
 951
 952     @bigmemtest(minsize=_2G // 2 + 2, memuse=16)
 953     def test_extend_small(self, size):
 954         return self.basic_test_extend(size)
 955
 956     @bigmemtest(minsize=_2G + 2, memuse=16)
 957     def test_extend_large(self, size):
 958         return self.basic_test_extend(size)
 959
 960     @bigmemtest(minsize=_2G // 5 + 2, memuse=8 * 5)
 961     def test_index(self, size):
 962         l = [1L, 2L, 3L, 4L, 5L] * size
 963         size *= 5
 964         self.assertEquals(l.index(1), 0)
 965         self.assertEquals(l.index(5, size - 5), size - 1)
 966         self.assertEquals(l.index(5, size - 5, size), size - 1)
 967         self.assertRaises(ValueError, l.index, 1, size - 4, size)
 968         self.assertRaises(ValueError, l.index, 6L)
 969
 970     # This tests suffers from overallocation, just like test_append.
 971     @bigmemtest(minsize=_2G + 10, memuse=9)
 972     def test_insert(self, size):
 973         l = [1.0] * size
 974         l.insert(size - 1, "A")
 975         size += 1
 976         self.assertEquals(len(l), size)
 977         self.assertEquals(l[-3:], [1.0, "A", 1.0])
 978
 979         l.insert(size + 1, "B")
 980         size += 1
 981         self.assertEquals(len(l), size)
 982         self.assertEquals(l[-3:], ["A", 1.0, "B"])
 983
 984         l.insert(1, "C")
 985         size += 1
 986         self.assertEquals(len(l), size)
 987         self.assertEquals(l[:3], [1.0, "C", 1.0])
 988         self.assertEquals(l[size - 3:], ["A", 1.0, "B"])
 989
 990     @bigmemtest(minsize=_2G // 5 + 4, memuse=8 * 5)
 991     def test_pop(self, size):
 992         l = [u"a", u"b", u"c", u"d", u"e"] * size
 993         size *= 5
 994         self.assertEquals(len(l), size)
 995
 996         item = l.pop()
 997         size -= 1
 998         self.assertEquals(len(l), size)
 999         self.assertEquals(item, u"e")
1000         self.assertEquals(l[-2:], [u"c", u"d"])
1001
1002         item = l.pop(0)
1003         size -= 1
1004         self.assertEquals(len(l), size)
1005         self.assertEquals(item, u"a")
1006         self.assertEquals(l[:2], [u"b", u"c"])
1007
1008         item = l.pop(size - 2)
1009         size -= 1
1010         self.assertEquals(len(l), size)
1011         self.assertEquals(item, u"c")
1012         self.assertEquals(l[-2:], [u"b", u"d"])
1013
1014     @bigmemtest(minsize=_2G + 10, memuse=8)
1015     def test_remove(self, size):
1016         l = [10] * size
1017         self.assertEquals(len(l), size)
1018
1019         l.remove(10)
1020         size -= 1
1021         self.assertEquals(len(l), size)
1022
1023         # Because of the earlier l.remove(), this append doesn't trigger
1024         # a resize.
1025         l.append(5)
1026         size += 1
1027         self.assertEquals(len(l), size)
1028         self.assertEquals(l[-2:], [10, 5])
1029         l.remove(5)
1030         size -= 1
1031         self.assertEquals(len(l), size)
1032         self.assertEquals(l[-2:], [10, 10])
1033
1034     @bigmemtest(minsize=_2G // 5 + 2, memuse=8 * 5)
1035     def test_reverse(self, size):
1036         l = [1, 2, 3, 4, 5] * size
1037         l.reverse()
1038         self.assertEquals(len(l), size * 5)
1039         self.assertEquals(l[-5:], [5, 4, 3, 2, 1])
1040         self.assertEquals(l[:5], [5, 4, 3, 2, 1])
1041
1042     @bigmemtest(minsize=_2G // 5 + 2, memuse=8 * 5)
1043     def test_sort(self, size):
1044         l = [1, 2, 3, 4, 5] * size
1045         l.sort()
1046         self.assertEquals(len(l), size * 5)
1047         self.assertEquals(l.count(1), size)
1048         self.assertEquals(l[:10], [1] * 10)
1049         self.assertEquals(l[-10:], [5] * 10)
1050
1051 class BufferTest(unittest.TestCase):
1052
1053     @precisionbigmemtest(size=_1G, memuse=4)
1054     def test_repeat(self, size):
1055         try:
1056             b = buffer("AAAA")*size
1057         except MemoryError:
1058             pass # acceptable on 32-bit
1059         else:
1060             count = 0
1061             for c in b:
1062                 self.assertEquals(c, 'A')
1063                 count += 1
1064             self.assertEquals(count, size*4)
1065
1066 def test_main():
1067     test_support.run_unittest(StrTest, TupleTest, ListTest, BufferTest)
1068
1069 if __name__ == '__main__':
1070     if len(sys.argv) > 1:
1071         test_support.set_memlimit(sys.argv[1])
1072     test_main()