• Home
  • Line#
  • Scopes#
  • Navigate#
  • Raw
  • Download
1from test import test_support
2from test.test_support import bigmemtest, _1G, _2G, _4G, precisionbigmemtest
3
4import unittest
5import operator
6import string
7import 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 assertEqual 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
36if test_support.have_unicode:
37    character_size = 4 if sys.maxunicode > 0xFFFF else 2
38else:
39    character_size = 1
40
41class StrTest(unittest.TestCase):
42    @bigmemtest(minsize=_2G, memuse=2)
43    def test_capitalize(self, size):
44        SUBSTR = ' abc def ghi'
45        s = '-' * size + SUBSTR
46        caps = s.capitalize()
47        self.assertEqual(caps[-len(SUBSTR):],
48                         SUBSTR.capitalize())
49        self.assertEqual(caps.lstrip('-'), SUBSTR)
50
51    @bigmemtest(minsize=_2G + 10, memuse=1)
52    def test_center(self, size):
53        SUBSTR = ' abc def ghi'
54        s = SUBSTR.center(size)
55        self.assertEqual(len(s), size)
56        lpadsize = rpadsize = (len(s) - len(SUBSTR)) // 2
57        if len(s) % 2:
58            lpadsize += 1
59        self.assertEqual(s[lpadsize:-rpadsize], SUBSTR)
60        self.assertEqual(s.strip(), SUBSTR.strip())
61
62    @test_support.requires_unicode
63    @precisionbigmemtest(size=_2G - 1, memuse=character_size)
64    def test_center_unicode(self, size):
65        SUBSTR = u' abc def ghi'
66        try:
67            s = SUBSTR.center(size)
68        except OverflowError:
69            pass # acceptable on 32-bit
70        else:
71            self.assertEqual(len(s), size)
72            lpadsize = rpadsize = (len(s) - len(SUBSTR)) // 2
73            if len(s) % 2:
74                lpadsize += 1
75            self.assertEqual(s[lpadsize:-rpadsize], SUBSTR)
76            self.assertEqual(s.strip(), SUBSTR.strip())
77            del s
78
79    @bigmemtest(minsize=_2G, memuse=2)
80    def test_count(self, size):
81        SUBSTR = ' abc def ghi'
82        s = '.' * size + SUBSTR
83        self.assertEqual(s.count('.'), size)
84        s += '.'
85        self.assertEqual(s.count('.'), size + 1)
86        self.assertEqual(s.count(' '), 3)
87        self.assertEqual(s.count('i'), 1)
88        self.assertEqual(s.count('j'), 0)
89
90    @test_support.requires_unicode
91    @bigmemtest(minsize=_2G + 2, memuse=1 + character_size)
92    def test_decode(self, size):
93        s = '.' * size
94        self.assertEqual(len(s.decode('utf-8')), size)
95
96    def basic_encode_test(self, size, enc, c=u'.', expectedsize=None):
97        if expectedsize is None:
98            expectedsize = size
99
100        s = c * size
101        self.assertEqual(len(s.encode(enc)), expectedsize)
102
103    @test_support.requires_unicode
104    @bigmemtest(minsize=_2G + 2, memuse=character_size + 4)
105    def test_encode(self, size):
106        self.basic_encode_test(size, 'utf-8')
107
108    @test_support.requires_unicode
109    @precisionbigmemtest(size=_4G // 6 + 2, memuse=character_size + 6)
110    def test_encode_raw_unicode_escape(self, size):
111        self.basic_encode_test(size, 'raw_unicode_escape')
112
113    @test_support.requires_unicode
114    @precisionbigmemtest(size=_4G // 5 + 70, memuse=character_size + 8)
115    def test_encode_utf7(self, size):
116        self.basic_encode_test(size, 'utf7')
117
118    @test_support.requires_unicode
119    @precisionbigmemtest(size=_4G // 4 + 5, memuse=character_size + 4)
120    def test_encode_utf32(self, size):
121        self.basic_encode_test(size, 'utf32', expectedsize=4*size+4)
122
123    @test_support.requires_unicode
124    @precisionbigmemtest(size=_2G-1, memuse=4)
125    def test_decodeascii(self, size):
126        self.basic_encode_test(size, 'ascii', c='A')
127
128    @bigmemtest(minsize=_2G, memuse=2)
129    def test_endswith(self, size):
130        SUBSTR = ' abc def ghi'
131        s = '-' * size + SUBSTR
132        self.assertTrue(s.endswith(SUBSTR))
133        self.assertTrue(s.endswith(s))
134        s2 = '...' + s
135        self.assertTrue(s2.endswith(s))
136        self.assertFalse(s.endswith('a' + SUBSTR))
137        self.assertFalse(SUBSTR.endswith(s))
138
139    @bigmemtest(minsize=_2G + 10, memuse=2)
140    def test_expandtabs(self, size):
141        s = '-' * size
142        tabsize = 8
143        self.assertEqual(s.expandtabs(), s)
144        del s
145        slen, remainder = divmod(size, tabsize)
146        s = '       \t' * slen
147        s = s.expandtabs(tabsize)
148        self.assertEqual(len(s), size - remainder)
149        self.assertEqual(len(s.strip(' ')), 0)
150
151    @bigmemtest(minsize=_2G, memuse=2)
152    def test_find(self, size):
153        SUBSTR = ' abc def ghi'
154        sublen = len(SUBSTR)
155        s = ''.join([SUBSTR, '-' * size, SUBSTR])
156        self.assertEqual(s.find(' '), 0)
157        self.assertEqual(s.find(SUBSTR), 0)
158        self.assertEqual(s.find(' ', sublen), sublen + size)
159        self.assertEqual(s.find(SUBSTR, len(SUBSTR)), sublen + size)
160        self.assertEqual(s.find('i'), SUBSTR.find('i'))
161        self.assertEqual(s.find('i', sublen),
162                         sublen + size + SUBSTR.find('i'))
163        self.assertEqual(s.find('i', size),
164                         sublen + size + SUBSTR.find('i'))
165        self.assertEqual(s.find('j'), -1)
166
167    @bigmemtest(minsize=_2G, memuse=2)
168    def test_index(self, size):
169        SUBSTR = ' abc def ghi'
170        sublen = len(SUBSTR)
171        s = ''.join([SUBSTR, '-' * size, SUBSTR])
172        self.assertEqual(s.index(' '), 0)
173        self.assertEqual(s.index(SUBSTR), 0)
174        self.assertEqual(s.index(' ', sublen), sublen + size)
175        self.assertEqual(s.index(SUBSTR, sublen), sublen + size)
176        self.assertEqual(s.index('i'), SUBSTR.index('i'))
177        self.assertEqual(s.index('i', sublen),
178                         sublen + size + SUBSTR.index('i'))
179        self.assertEqual(s.index('i', size),
180                         sublen + size + SUBSTR.index('i'))
181        self.assertRaises(ValueError, s.index, 'j')
182
183    @bigmemtest(minsize=_2G, memuse=2)
184    def test_isalnum(self, size):
185        SUBSTR = '123456'
186        s = 'a' * size + SUBSTR
187        self.assertTrue(s.isalnum())
188        s += '.'
189        self.assertFalse(s.isalnum())
190
191    @bigmemtest(minsize=_2G, memuse=2)
192    def test_isalpha(self, size):
193        SUBSTR = 'zzzzzzz'
194        s = 'a' * size + SUBSTR
195        self.assertTrue(s.isalpha())
196        s += '.'
197        self.assertFalse(s.isalpha())
198
199    @bigmemtest(minsize=_2G, memuse=2)
200    def test_isdigit(self, size):
201        SUBSTR = '123456'
202        s = '9' * size + SUBSTR
203        self.assertTrue(s.isdigit())
204        s += 'z'
205        self.assertFalse(s.isdigit())
206
207    @bigmemtest(minsize=_2G, memuse=2)
208    def test_islower(self, size):
209        chars = ''.join([ chr(c) for c in range(255) if not chr(c).isupper() ])
210        repeats = size // len(chars) + 2
211        s = chars * repeats
212        self.assertTrue(s.islower())
213        s += 'A'
214        self.assertFalse(s.islower())
215
216    @bigmemtest(minsize=_2G, memuse=2)
217    def test_isspace(self, size):
218        whitespace = ' \f\n\r\t\v'
219        repeats = size // len(whitespace) + 2
220        s = whitespace * repeats
221        self.assertTrue(s.isspace())
222        s += 'j'
223        self.assertFalse(s.isspace())
224
225    @bigmemtest(minsize=_2G, memuse=2)
226    def test_istitle(self, size):
227        SUBSTR = '123456'
228        s = ''.join(['A', 'a' * size, SUBSTR])
229        self.assertTrue(s.istitle())
230        s += 'A'
231        self.assertTrue(s.istitle())
232        s += 'aA'
233        self.assertFalse(s.istitle())
234
235    @bigmemtest(minsize=_2G, memuse=2)
236    def test_isupper(self, size):
237        chars = ''.join([ chr(c) for c in range(255) if not chr(c).islower() ])
238        repeats = size // len(chars) + 2
239        s = chars * repeats
240        self.assertTrue(s.isupper())
241        s += 'a'
242        self.assertFalse(s.isupper())
243
244    @bigmemtest(minsize=_2G, memuse=2)
245    def test_join(self, size):
246        s = 'A' * size
247        x = s.join(['aaaaa', 'bbbbb'])
248        self.assertEqual(x.count('a'), 5)
249        self.assertEqual(x.count('b'), 5)
250        self.assertTrue(x.startswith('aaaaaA'))
251        self.assertTrue(x.endswith('Abbbbb'))
252
253    @bigmemtest(minsize=_2G + 10, memuse=1)
254    def test_ljust(self, size):
255        SUBSTR = ' abc def ghi'
256        s = SUBSTR.ljust(size)
257        self.assertTrue(s.startswith(SUBSTR + '  '))
258        self.assertEqual(len(s), size)
259        self.assertEqual(s.strip(), SUBSTR.strip())
260
261    @bigmemtest(minsize=_2G + 10, memuse=2)
262    def test_lower(self, size):
263        s = 'A' * size
264        s = s.lower()
265        self.assertEqual(len(s), size)
266        self.assertEqual(s.count('a'), size)
267
268    @bigmemtest(minsize=_2G + 10, memuse=1)
269    def test_lstrip(self, size):
270        SUBSTR = 'abc def ghi'
271        s = SUBSTR.rjust(size)
272        self.assertEqual(len(s), size)
273        self.assertEqual(s.lstrip(), SUBSTR.lstrip())
274        del s
275        s = SUBSTR.ljust(size)
276        self.assertEqual(len(s), size)
277        stripped = s.lstrip()
278        self.assertTrue(stripped is s)
279
280    @bigmemtest(minsize=_2G + 10, memuse=2)
281    def test_replace(self, size):
282        replacement = 'a'
283        s = ' ' * size
284        s = s.replace(' ', replacement)
285        self.assertEqual(len(s), size)
286        self.assertEqual(s.count(replacement), size)
287        s = s.replace(replacement, ' ', size - 4)
288        self.assertEqual(len(s), size)
289        self.assertEqual(s.count(replacement), 4)
290        self.assertEqual(s[-10:], '      aaaa')
291
292    @bigmemtest(minsize=_2G, memuse=2)
293    def test_rfind(self, size):
294        SUBSTR = ' abc def ghi'
295        sublen = len(SUBSTR)
296        s = ''.join([SUBSTR, '-' * size, SUBSTR])
297        self.assertEqual(s.rfind(' '), sublen + size + SUBSTR.rfind(' '))
298        self.assertEqual(s.rfind(SUBSTR), sublen + size)
299        self.assertEqual(s.rfind(' ', 0, size), SUBSTR.rfind(' '))
300        self.assertEqual(s.rfind(SUBSTR, 0, sublen + size), 0)
301        self.assertEqual(s.rfind('i'), sublen + size + SUBSTR.rfind('i'))
302        self.assertEqual(s.rfind('i', 0, sublen), SUBSTR.rfind('i'))
303        self.assertEqual(s.rfind('i', 0, sublen + size),
304                         SUBSTR.rfind('i'))
305        self.assertEqual(s.rfind('j'), -1)
306
307    @bigmemtest(minsize=_2G, memuse=2)
308    def test_rindex(self, size):
309        SUBSTR = ' abc def ghi'
310        sublen = len(SUBSTR)
311        s = ''.join([SUBSTR, '-' * size, SUBSTR])
312        self.assertEqual(s.rindex(' '),
313                          sublen + size + SUBSTR.rindex(' '))
314        self.assertEqual(s.rindex(SUBSTR), sublen + size)
315        self.assertEqual(s.rindex(' ', 0, sublen + size - 1),
316                         SUBSTR.rindex(' '))
317        self.assertEqual(s.rindex(SUBSTR, 0, sublen + size), 0)
318        self.assertEqual(s.rindex('i'),
319                         sublen + size + SUBSTR.rindex('i'))
320        self.assertEqual(s.rindex('i', 0, sublen), SUBSTR.rindex('i'))
321        self.assertEqual(s.rindex('i', 0, sublen + size),
322                         SUBSTR.rindex('i'))
323        self.assertRaises(ValueError, s.rindex, 'j')
324
325    @bigmemtest(minsize=_2G + 10, memuse=1)
326    def test_rjust(self, size):
327        SUBSTR = ' abc def ghi'
328        s = SUBSTR.ljust(size)
329        self.assertTrue(s.startswith(SUBSTR + '  '))
330        self.assertEqual(len(s), size)
331        self.assertEqual(s.strip(), SUBSTR.strip())
332
333    @bigmemtest(minsize=_2G + 10, memuse=1)
334    def test_rstrip(self, size):
335        SUBSTR = ' abc def ghi'
336        s = SUBSTR.ljust(size)
337        self.assertEqual(len(s), size)
338        self.assertEqual(s.rstrip(), SUBSTR.rstrip())
339        del s
340        s = SUBSTR.rjust(size)
341        self.assertEqual(len(s), size)
342        stripped = s.rstrip()
343        self.assertTrue(stripped is s)
344
345    # The test takes about size bytes to build a string, and then about
346    # sqrt(size) substrings of sqrt(size) in size and a list to
347    # hold sqrt(size) items. It's close but just over 2x size.
348    @bigmemtest(minsize=_2G, memuse=2.1)
349    def test_split_small(self, size):
350        # Crudely calculate an estimate so that the result of s.split won't
351        # take up an inordinate amount of memory
352        chunksize = int(size ** 0.5 + 2)
353        SUBSTR = 'a' + ' ' * chunksize
354        s = SUBSTR * chunksize
355        l = s.split()
356        self.assertEqual(len(l), chunksize)
357        self.assertEqual(set(l), set(['a']))
358        del l
359        l = s.split('a')
360        self.assertEqual(len(l), chunksize + 1)
361        self.assertEqual(set(l), set(['', ' ' * chunksize]))
362
363    # Allocates a string of twice size (and briefly two) and a list of
364    # size.  Because of internal affairs, the s.split() call produces a
365    # list of size times the same one-character string, so we only
366    # suffer for the list size. (Otherwise, it'd cost another 48 times
367    # size in bytes!) Nevertheless, a list of size takes
368    # 8*size bytes.
369    @bigmemtest(minsize=_2G + 5, memuse=10)
370    def test_split_large(self, size):
371        s = ' a' * size + ' '
372        l = s.split()
373        self.assertEqual(len(l), size)
374        self.assertEqual(set(l), set(['a']))
375        del l
376        l = s.split('a')
377        self.assertEqual(len(l), size + 1)
378        self.assertEqual(set(l), set([' ']))
379
380    @bigmemtest(minsize=_2G, memuse=2.1)
381    def test_splitlines(self, size):
382        # Crudely calculate an estimate so that the result of s.split won't
383        # take up an inordinate amount of memory
384        chunksize = int(size ** 0.5 + 2) // 2
385        SUBSTR = ' ' * chunksize + '\n' + ' ' * chunksize + '\r\n'
386        s = SUBSTR * chunksize
387        l = s.splitlines()
388        self.assertEqual(len(l), chunksize * 2)
389        self.assertEqual(set(l), set([' ' * chunksize]))
390
391    @bigmemtest(minsize=_2G, memuse=2)
392    def test_startswith(self, size):
393        SUBSTR = ' abc def ghi'
394        s = '-' * size + SUBSTR
395        self.assertTrue(s.startswith(s))
396        self.assertTrue(s.startswith('-' * size))
397        self.assertFalse(s.startswith(SUBSTR))
398
399    @bigmemtest(minsize=_2G, memuse=1)
400    def test_strip(self, size):
401        SUBSTR = '   abc def ghi   '
402        s = SUBSTR.rjust(size)
403        self.assertEqual(len(s), size)
404        self.assertEqual(s.strip(), SUBSTR.strip())
405        del s
406        s = SUBSTR.ljust(size)
407        self.assertEqual(len(s), size)
408        self.assertEqual(s.strip(), SUBSTR.strip())
409
410    @bigmemtest(minsize=_2G, memuse=2)
411    def test_swapcase(self, size):
412        SUBSTR = "aBcDeFG12.'\xa9\x00"
413        sublen = len(SUBSTR)
414        repeats = size // sublen + 2
415        s = SUBSTR * repeats
416        s = s.swapcase()
417        self.assertEqual(len(s), sublen * repeats)
418        self.assertEqual(s[:sublen * 3], SUBSTR.swapcase() * 3)
419        self.assertEqual(s[-sublen * 3:], SUBSTR.swapcase() * 3)
420
421    @bigmemtest(minsize=_2G, memuse=2)
422    def test_title(self, size):
423        SUBSTR = 'SpaaHAaaAaham'
424        s = SUBSTR * (size // len(SUBSTR) + 2)
425        s = s.title()
426        self.assertTrue(s.startswith((SUBSTR * 3).title()))
427        self.assertTrue(s.endswith(SUBSTR.lower() * 3))
428
429    @bigmemtest(minsize=_2G, memuse=2)
430    def test_translate(self, size):
431        trans = string.maketrans('.aZ', '-!$')
432        SUBSTR = 'aZz.z.Aaz.'
433        sublen = len(SUBSTR)
434        repeats = size // sublen + 2
435        s = SUBSTR * repeats
436        s = s.translate(trans)
437        self.assertEqual(len(s), repeats * sublen)
438        self.assertEqual(s[:sublen], SUBSTR.translate(trans))
439        self.assertEqual(s[-sublen:], SUBSTR.translate(trans))
440        self.assertEqual(s.count('.'), 0)
441        self.assertEqual(s.count('!'), repeats * 2)
442        self.assertEqual(s.count('z'), repeats * 3)
443
444    @bigmemtest(minsize=_2G + 5, memuse=2)
445    def test_upper(self, size):
446        s = 'a' * size
447        s = s.upper()
448        self.assertEqual(len(s), size)
449        self.assertEqual(s.count('A'), size)
450
451    @bigmemtest(minsize=_2G + 20, memuse=1)
452    def test_zfill(self, size):
453        SUBSTR = '-568324723598234'
454        s = SUBSTR.zfill(size)
455        self.assertTrue(s.endswith('0' + SUBSTR[1:]))
456        self.assertTrue(s.startswith('-0'))
457        self.assertEqual(len(s), size)
458        self.assertEqual(s.count('0'), size - len(SUBSTR))
459
460    @bigmemtest(minsize=_2G + 10, memuse=2)
461    def test_format(self, size):
462        s = '-' * size
463        sf = '%s' % (s,)
464        self.assertTrue(s == sf)
465        del sf
466        sf = '..%s..' % (s,)
467        self.assertEqual(len(sf), len(s) + 4)
468        self.assertTrue(sf.startswith('..-'))
469        self.assertTrue(sf.endswith('-..'))
470        del s, sf
471
472        size //= 2
473        edge = '-' * size
474        s = ''.join([edge, '%s', edge])
475        del edge
476        s = s % '...'
477        self.assertEqual(len(s), size * 2 + 3)
478        self.assertEqual(s.count('.'), 3)
479        self.assertEqual(s.count('-'), size * 2)
480
481    @bigmemtest(minsize=_2G + 10, memuse=5)
482    def test_repr_small(self, size):
483        s = '-' * size
484        s = repr(s)
485        self.assertEqual(len(s), size + 2)
486        self.assertEqual(s[0], "'")
487        self.assertEqual(s[-1], "'")
488        self.assertEqual(s.count('-'), size)
489        del s
490        # repr() will create a string four times as large as this 'binary
491        # string', but we don't want to allocate much more than twice
492        # size in total.  (We do extra testing in test_repr_large())
493        s = '\x00' * size
494        s = repr(s)
495        self.assertEqual(len(s), size * 4 + 2)
496        self.assertEqual(s[0], "'")
497        self.assertEqual(s[-1], "'")
498        self.assertEqual(s.count('\\'), size)
499        self.assertEqual(s.count('0'), size * 2)
500
501    @bigmemtest(minsize=_2G + 10, memuse=5)
502    def test_repr_large(self, size):
503        s = '\x00' * size
504        s = repr(s)
505        self.assertEqual(len(s), size * 4 + 2)
506        self.assertEqual(s[0], "'")
507        self.assertEqual(s[-1], "'")
508        self.assertEqual(s.count('\\'), size)
509        self.assertEqual(s.count('0'), size * 2)
510
511    @test_support.requires_unicode
512    @bigmemtest(minsize=2**32 // 6, memuse=character_size + 6)
513    def test_unicode_repr(self, size):
514        s = unichr(0xABCD) * size
515        try:
516            r = repr(s)
517            self.assertEqual(len(r), 3 + 6 * size)
518            self.assertTrue(r.endswith(r"\uabcd'"), r[-10:])
519        finally:
520            s = r = None
521
522    @test_support.requires_unicode
523    @precisionbigmemtest(size=_4G // 6 + 1, memuse=character_size + 6)
524    def test_unicode_repr_oflw(self, size):
525        s = unichr(0xABCD) * size
526        try:
527            r = repr(s)
528            self.assertEqual(len(r), 3 + 6 * size)
529            self.assertTrue(r.endswith(r"\uabcd'"), r[-10:])
530        finally:
531            r = s = None
532
533    # This test is meaningful even with size < 2G, as long as the
534    # doubled string is > 2G (but it tests more if both are > 2G :)
535    @bigmemtest(minsize=_1G + 2, memuse=3)
536    def test_concat(self, size):
537        s = '.' * size
538        self.assertEqual(len(s), size)
539        s = s + s
540        self.assertEqual(len(s), size * 2)
541        self.assertEqual(s.count('.'), size * 2)
542
543    # This test is meaningful even with size < 2G, as long as the
544    # repeated string is > 2G (but it tests more if both are > 2G :)
545    @bigmemtest(minsize=_1G + 2, memuse=3)
546    def test_repeat(self, size):
547        s = '.' * size
548        self.assertEqual(len(s), size)
549        s = s * 2
550        self.assertEqual(len(s), size * 2)
551        self.assertEqual(s.count('.'), size * 2)
552
553    @bigmemtest(minsize=_2G + 20, memuse=2)
554    def test_slice_and_getitem(self, size):
555        SUBSTR = '0123456789'
556        sublen = len(SUBSTR)
557        s = SUBSTR * (size // sublen)
558        stepsize = len(s) // 100
559        stepsize = stepsize - (stepsize % sublen)
560        for i in range(0, len(s) - stepsize, stepsize):
561            self.assertEqual(s[i], SUBSTR[0])
562            self.assertEqual(s[i:i + sublen], SUBSTR)
563            self.assertEqual(s[i:i + sublen:2], SUBSTR[::2])
564            if i > 0:
565                self.assertEqual(s[i + sublen - 1:i - 1:-3],
566                                 SUBSTR[sublen::-3])
567        # Make sure we do some slicing and indexing near the end of the
568        # string, too.
569        self.assertEqual(s[len(s) - 1], SUBSTR[-1])
570        self.assertEqual(s[-1], SUBSTR[-1])
571        self.assertEqual(s[len(s) - 10], SUBSTR[0])
572        self.assertEqual(s[-sublen], SUBSTR[0])
573        self.assertEqual(s[len(s):], '')
574        self.assertEqual(s[len(s) - 1:], SUBSTR[-1])
575        self.assertEqual(s[-1:], SUBSTR[-1])
576        self.assertEqual(s[len(s) - sublen:], SUBSTR)
577        self.assertEqual(s[-sublen:], SUBSTR)
578        self.assertEqual(len(s[:]), len(s))
579        self.assertEqual(len(s[:len(s) - 5]), len(s) - 5)
580        self.assertEqual(len(s[5:-5]), len(s) - 10)
581
582        self.assertRaises(IndexError, operator.getitem, s, len(s))
583        self.assertRaises(IndexError, operator.getitem, s, len(s) + 1)
584        self.assertRaises(IndexError, operator.getitem, s, len(s) + 1<<31)
585
586    @bigmemtest(minsize=_2G, memuse=2)
587    def test_contains(self, size):
588        SUBSTR = '0123456789'
589        edge = '-' * (size // 2)
590        s = ''.join([edge, SUBSTR, edge])
591        del edge
592        self.assertIn(SUBSTR, s)
593        self.assertNotIn(SUBSTR * 2, s)
594        self.assertIn('-', s)
595        self.assertNotIn('a', s)
596        s += 'a'
597        self.assertIn('a', s)
598
599    @bigmemtest(minsize=_2G + 10, memuse=2)
600    def test_compare(self, size):
601        s1 = '-' * size
602        s2 = '-' * size
603        self.assertTrue(s1 == s2)
604        del s2
605        s2 = s1 + 'a'
606        self.assertFalse(s1 == s2)
607        del s2
608        s2 = '.' * size
609        self.assertFalse(s1 == s2)
610
611    @bigmemtest(minsize=_2G + 10, memuse=1)
612    def test_hash(self, size):
613        # Not sure if we can do any meaningful tests here...  Even if we
614        # start relying on the exact algorithm used, the result will be
615        # different depending on the size of the C 'long int'.  Even this
616        # test is dodgy (there's no *guarantee* that the two things should
617        # have a different hash, even if they, in the current
618        # implementation, almost always do.)
619        s = '\x00' * size
620        h1 = hash(s)
621        del s
622        s = '\x00' * (size + 1)
623        self.assertFalse(h1 == hash(s))
624
625class TupleTest(unittest.TestCase):
626
627    # Tuples have a small, fixed-sized head and an array of pointers to
628    # data.  Since we're testing 64-bit addressing, we can assume that the
629    # pointers are 8 bytes, and that thus that the tuples take up 8 bytes
630    # per size.
631
632    # As a side-effect of testing long tuples, these tests happen to test
633    # having more than 2<<31 references to any given object. Hence the
634    # use of different types of objects as contents in different tests.
635
636    @bigmemtest(minsize=_2G + 2, memuse=16)
637    def test_compare(self, size):
638        t1 = (u'',) * size
639        t2 = (u'',) * size
640        self.assertTrue(t1 == t2)
641        del t2
642        t2 = (u'',) * (size + 1)
643        self.assertFalse(t1 == t2)
644        del t2
645        t2 = (1,) * size
646        self.assertFalse(t1 == t2)
647
648    # Test concatenating into a single tuple of more than 2G in length,
649    # and concatenating a tuple of more than 2G in length separately, so
650    # the smaller test still gets run even if there isn't memory for the
651    # larger test (but we still let the tester know the larger test is
652    # skipped, in verbose mode.)
653    def basic_concat_test(self, size):
654        t = ((),) * size
655        self.assertEqual(len(t), size)
656        t = t + t
657        self.assertEqual(len(t), size * 2)
658
659    @bigmemtest(minsize=_2G // 2 + 2, memuse=24)
660    def test_concat_small(self, size):
661        return self.basic_concat_test(size)
662
663    @bigmemtest(minsize=_2G + 2, memuse=24)
664    def test_concat_large(self, size):
665        return self.basic_concat_test(size)
666
667    @bigmemtest(minsize=_2G // 5 + 10, memuse=8 * 5)
668    def test_contains(self, size):
669        t = (1, 2, 3, 4, 5) * size
670        self.assertEqual(len(t), size * 5)
671        self.assertIn(5, t)
672        self.assertNotIn((1, 2, 3, 4, 5), t)
673        self.assertNotIn(0, t)
674
675    @bigmemtest(minsize=_2G + 10, memuse=8)
676    def test_hash(self, size):
677        t1 = (0,) * size
678        h1 = hash(t1)
679        del t1
680        t2 = (0,) * (size + 1)
681        self.assertFalse(h1 == hash(t2))
682
683    @bigmemtest(minsize=_2G + 10, memuse=8)
684    def test_index_and_slice(self, size):
685        t = (None,) * size
686        self.assertEqual(len(t), size)
687        self.assertEqual(t[-1], None)
688        self.assertEqual(t[5], None)
689        self.assertEqual(t[size - 1], None)
690        self.assertRaises(IndexError, operator.getitem, t, size)
691        self.assertEqual(t[:5], (None,) * 5)
692        self.assertEqual(t[-5:], (None,) * 5)
693        self.assertEqual(t[20:25], (None,) * 5)
694        self.assertEqual(t[-25:-20], (None,) * 5)
695        self.assertEqual(t[size - 5:], (None,) * 5)
696        self.assertEqual(t[size - 5:size], (None,) * 5)
697        self.assertEqual(t[size - 6:size - 2], (None,) * 4)
698        self.assertEqual(t[size:size], ())
699        self.assertEqual(t[size:size+5], ())
700
701    # Like test_concat, split in two.
702    def basic_test_repeat(self, size):
703        t = ('',) * size
704        self.assertEqual(len(t), size)
705        t = t * 2
706        self.assertEqual(len(t), size * 2)
707
708    @bigmemtest(minsize=_2G // 2 + 2, memuse=24)
709    def test_repeat_small(self, size):
710        return self.basic_test_repeat(size)
711
712    @bigmemtest(minsize=_2G + 2, memuse=24)
713    def test_repeat_large(self, size):
714        return self.basic_test_repeat(size)
715
716    @bigmemtest(minsize=_1G - 1, memuse=12)
717    def test_repeat_large_2(self, size):
718        return self.basic_test_repeat(size)
719
720    @precisionbigmemtest(size=_1G - 1, memuse=9)
721    def test_from_2G_generator(self, size):
722        try:
723            t = tuple(xrange(size))
724        except MemoryError:
725            pass # acceptable on 32-bit
726        else:
727            count = 0
728            for item in t:
729                self.assertEqual(item, count)
730                count += 1
731            self.assertEqual(count, size)
732
733    @precisionbigmemtest(size=_1G - 25, memuse=9)
734    def test_from_almost_2G_generator(self, size):
735        try:
736            t = tuple(xrange(size))
737            count = 0
738            for item in t:
739                self.assertEqual(item, count)
740                count += 1
741            self.assertEqual(count, size)
742        except MemoryError:
743            pass # acceptable, expected on 32-bit
744
745    # Like test_concat, split in two.
746    def basic_test_repr(self, size):
747        t = (0,) * size
748        s = repr(t)
749        # The repr of a tuple of 0's is exactly three times the tuple length.
750        self.assertEqual(len(s), size * 3)
751        self.assertEqual(s[:5], '(0, 0')
752        self.assertEqual(s[-5:], '0, 0)')
753        self.assertEqual(s.count('0'), size)
754
755    @bigmemtest(minsize=_2G // 3 + 2, memuse=8 + 3)
756    def test_repr_small(self, size):
757        return self.basic_test_repr(size)
758
759    @bigmemtest(minsize=_2G + 2, memuse=8 + 3)
760    def test_repr_large(self, size):
761        return self.basic_test_repr(size)
762
763class ListTest(unittest.TestCase):
764
765    # Like tuples, lists have a small, fixed-sized head and an array of
766    # pointers to data, so 8 bytes per size. Also like tuples, we make the
767    # lists hold references to various objects to test their refcount
768    # limits.
769
770    @bigmemtest(minsize=_2G + 2, memuse=16)
771    def test_compare(self, size):
772        l1 = [u''] * size
773        l2 = [u''] * size
774        self.assertTrue(l1 == l2)
775        del l2
776        l2 = [u''] * (size + 1)
777        self.assertFalse(l1 == l2)
778        del l2
779        l2 = [2] * size
780        self.assertFalse(l1 == l2)
781
782    # Test concatenating into a single list of more than 2G in length,
783    # and concatenating a list of more than 2G in length separately, so
784    # the smaller test still gets run even if there isn't memory for the
785    # larger test (but we still let the tester know the larger test is
786    # skipped, in verbose mode.)
787    def basic_test_concat(self, size):
788        l = [[]] * size
789        self.assertEqual(len(l), size)
790        l = l + l
791        self.assertEqual(len(l), size * 2)
792
793    @bigmemtest(minsize=_2G // 2 + 2, memuse=24)
794    def test_concat_small(self, size):
795        return self.basic_test_concat(size)
796
797    @bigmemtest(minsize=_2G + 2, memuse=24)
798    def test_concat_large(self, size):
799        return self.basic_test_concat(size)
800
801    def basic_test_inplace_concat(self, size):
802        l = [sys.stdout] * size
803        l += l
804        self.assertEqual(len(l), size * 2)
805        self.assertTrue(l[0] is l[-1])
806        self.assertTrue(l[size - 1] is l[size + 1])
807
808    @bigmemtest(minsize=_2G // 2 + 2, memuse=24)
809    def test_inplace_concat_small(self, size):
810        return self.basic_test_inplace_concat(size)
811
812    @bigmemtest(minsize=_2G + 2, memuse=24)
813    def test_inplace_concat_large(self, size):
814        return self.basic_test_inplace_concat(size)
815
816    @bigmemtest(minsize=_2G // 5 + 10, memuse=8 * 5)
817    def test_contains(self, size):
818        l = [1, 2, 3, 4, 5] * size
819        self.assertEqual(len(l), size * 5)
820        self.assertIn(5, l)
821        self.assertNotIn([1, 2, 3, 4, 5], l)
822        self.assertNotIn(0, l)
823
824    @bigmemtest(minsize=_2G + 10, memuse=8)
825    def test_hash(self, size):
826        l = [0] * size
827        self.assertRaises(TypeError, hash, l)
828
829    @bigmemtest(minsize=_2G + 10, memuse=8)
830    def test_index_and_slice(self, size):
831        l = [None] * size
832        self.assertEqual(len(l), size)
833        self.assertEqual(l[-1], None)
834        self.assertEqual(l[5], None)
835        self.assertEqual(l[size - 1], None)
836        self.assertRaises(IndexError, operator.getitem, l, size)
837        self.assertEqual(l[:5], [None] * 5)
838        self.assertEqual(l[-5:], [None] * 5)
839        self.assertEqual(l[20:25], [None] * 5)
840        self.assertEqual(l[-25:-20], [None] * 5)
841        self.assertEqual(l[size - 5:], [None] * 5)
842        self.assertEqual(l[size - 5:size], [None] * 5)
843        self.assertEqual(l[size - 6:size - 2], [None] * 4)
844        self.assertEqual(l[size:size], [])
845        self.assertEqual(l[size:size+5], [])
846
847        l[size - 2] = 5
848        self.assertEqual(len(l), size)
849        self.assertEqual(l[-3:], [None, 5, None])
850        self.assertEqual(l.count(5), 1)
851        self.assertRaises(IndexError, operator.setitem, l, size, 6)
852        self.assertEqual(len(l), size)
853
854        l[size - 7:] = [1, 2, 3, 4, 5]
855        size -= 2
856        self.assertEqual(len(l), size)
857        self.assertEqual(l[-7:], [None, None, 1, 2, 3, 4, 5])
858
859        l[:7] = [1, 2, 3, 4, 5]
860        size -= 2
861        self.assertEqual(len(l), size)
862        self.assertEqual(l[:7], [1, 2, 3, 4, 5, None, None])
863
864        del l[size - 1]
865        size -= 1
866        self.assertEqual(len(l), size)
867        self.assertEqual(l[-1], 4)
868
869        del l[-2:]
870        size -= 2
871        self.assertEqual(len(l), size)
872        self.assertEqual(l[-1], 2)
873
874        del l[0]
875        size -= 1
876        self.assertEqual(len(l), size)
877        self.assertEqual(l[0], 2)
878
879        del l[:2]
880        size -= 2
881        self.assertEqual(len(l), size)
882        self.assertEqual(l[0], 4)
883
884    # Like test_concat, split in two.
885    def basic_test_repeat(self, size):
886        l = [] * size
887        self.assertFalse(l)
888        l = [''] * size
889        self.assertEqual(len(l), size)
890        l = l * 2
891        self.assertEqual(len(l), size * 2)
892
893    @bigmemtest(minsize=_2G // 2 + 2, memuse=24)
894    def test_repeat_small(self, size):
895        return self.basic_test_repeat(size)
896
897    @bigmemtest(minsize=_2G + 2, memuse=24)
898    def test_repeat_large(self, size):
899        return self.basic_test_repeat(size)
900
901    def basic_test_inplace_repeat(self, size):
902        l = ['']
903        l *= size
904        self.assertEqual(len(l), size)
905        self.assertTrue(l[0] is l[-1])
906        del l
907
908        l = [''] * size
909        l *= 2
910        self.assertEqual(len(l), size * 2)
911        self.assertTrue(l[size - 1] is l[-1])
912
913    @bigmemtest(minsize=_2G // 2 + 2, memuse=16)
914    def test_inplace_repeat_small(self, size):
915        return self.basic_test_inplace_repeat(size)
916
917    @bigmemtest(minsize=_2G + 2, memuse=16)
918    def test_inplace_repeat_large(self, size):
919        return self.basic_test_inplace_repeat(size)
920
921    def basic_test_repr(self, size):
922        l = [0] * size
923        s = repr(l)
924        # The repr of a list of 0's is exactly three times the list length.
925        self.assertEqual(len(s), size * 3)
926        self.assertEqual(s[:5], '[0, 0')
927        self.assertEqual(s[-5:], '0, 0]')
928        self.assertEqual(s.count('0'), size)
929
930    @bigmemtest(minsize=_2G // 3 + 2, memuse=8 + 3)
931    def test_repr_small(self, size):
932        return self.basic_test_repr(size)
933
934    @bigmemtest(minsize=_2G + 2, memuse=8 + 3)
935    def test_repr_large(self, size):
936        return self.basic_test_repr(size)
937
938    # list overallocates ~1/8th of the total size (on first expansion) so
939    # the single list.append call puts memuse at 9 bytes per size.
940    @bigmemtest(minsize=_2G, memuse=9)
941    def test_append(self, size):
942        l = [object()] * size
943        l.append(object())
944        self.assertEqual(len(l), size+1)
945        self.assertTrue(l[-3] is l[-2])
946        self.assertFalse(l[-2] is l[-1])
947
948    @bigmemtest(minsize=_2G // 5 + 2, memuse=8 * 5)
949    def test_count(self, size):
950        l = [1, 2, 3, 4, 5] * size
951        self.assertEqual(l.count(1), size)
952        self.assertEqual(l.count("1"), 0)
953
954    def basic_test_extend(self, size):
955        l = [file] * size
956        l.extend(l)
957        self.assertEqual(len(l), size * 2)
958        self.assertTrue(l[0] is l[-1])
959        self.assertTrue(l[size - 1] is l[size + 1])
960
961    @bigmemtest(minsize=_2G // 2 + 2, memuse=16)
962    def test_extend_small(self, size):
963        return self.basic_test_extend(size)
964
965    @bigmemtest(minsize=_2G + 2, memuse=16)
966    def test_extend_large(self, size):
967        return self.basic_test_extend(size)
968
969    @bigmemtest(minsize=_2G // 5 + 2, memuse=8 * 5)
970    def test_index(self, size):
971        l = [1L, 2L, 3L, 4L, 5L] * size
972        size *= 5
973        self.assertEqual(l.index(1), 0)
974        self.assertEqual(l.index(5, size - 5), size - 1)
975        self.assertEqual(l.index(5, size - 5, size), size - 1)
976        self.assertRaises(ValueError, l.index, 1, size - 4, size)
977        self.assertRaises(ValueError, l.index, 6L)
978
979    # This tests suffers from overallocation, just like test_append.
980    @bigmemtest(minsize=_2G + 10, memuse=9)
981    def test_insert(self, size):
982        l = [1.0] * size
983        l.insert(size - 1, "A")
984        size += 1
985        self.assertEqual(len(l), size)
986        self.assertEqual(l[-3:], [1.0, "A", 1.0])
987
988        l.insert(size + 1, "B")
989        size += 1
990        self.assertEqual(len(l), size)
991        self.assertEqual(l[-3:], ["A", 1.0, "B"])
992
993        l.insert(1, "C")
994        size += 1
995        self.assertEqual(len(l), size)
996        self.assertEqual(l[:3], [1.0, "C", 1.0])
997        self.assertEqual(l[size - 3:], ["A", 1.0, "B"])
998
999    @bigmemtest(minsize=_2G // 5 + 4, memuse=8 * 5)
1000    def test_pop(self, size):
1001        l = [u"a", u"b", u"c", u"d", u"e"] * size
1002        size *= 5
1003        self.assertEqual(len(l), size)
1004
1005        item = l.pop()
1006        size -= 1
1007        self.assertEqual(len(l), size)
1008        self.assertEqual(item, u"e")
1009        self.assertEqual(l[-2:], [u"c", u"d"])
1010
1011        item = l.pop(0)
1012        size -= 1
1013        self.assertEqual(len(l), size)
1014        self.assertEqual(item, u"a")
1015        self.assertEqual(l[:2], [u"b", u"c"])
1016
1017        item = l.pop(size - 2)
1018        size -= 1
1019        self.assertEqual(len(l), size)
1020        self.assertEqual(item, u"c")
1021        self.assertEqual(l[-2:], [u"b", u"d"])
1022
1023    @bigmemtest(minsize=_2G + 10, memuse=8)
1024    def test_remove(self, size):
1025        l = [10] * size
1026        self.assertEqual(len(l), size)
1027
1028        l.remove(10)
1029        size -= 1
1030        self.assertEqual(len(l), size)
1031
1032        # Because of the earlier l.remove(), this append doesn't trigger
1033        # a resize.
1034        l.append(5)
1035        size += 1
1036        self.assertEqual(len(l), size)
1037        self.assertEqual(l[-2:], [10, 5])
1038        l.remove(5)
1039        size -= 1
1040        self.assertEqual(len(l), size)
1041        self.assertEqual(l[-2:], [10, 10])
1042
1043    @bigmemtest(minsize=_2G // 5 + 2, memuse=8 * 5)
1044    def test_reverse(self, size):
1045        l = [1, 2, 3, 4, 5] * size
1046        l.reverse()
1047        self.assertEqual(len(l), size * 5)
1048        self.assertEqual(l[-5:], [5, 4, 3, 2, 1])
1049        self.assertEqual(l[:5], [5, 4, 3, 2, 1])
1050
1051    @bigmemtest(minsize=_2G // 5 + 2, memuse=8 * 5)
1052    def test_sort(self, size):
1053        l = [1, 2, 3, 4, 5] * size
1054        l.sort()
1055        self.assertEqual(len(l), size * 5)
1056        self.assertEqual(l.count(1), size)
1057        self.assertEqual(l[:10], [1] * 10)
1058        self.assertEqual(l[-10:], [5] * 10)
1059
1060class BufferTest(unittest.TestCase):
1061
1062    @precisionbigmemtest(size=_1G, memuse=4)
1063    def test_repeat(self, size):
1064        try:
1065            with test_support.check_py3k_warnings():
1066                b = buffer("AAAA")*size
1067        except MemoryError:
1068            pass # acceptable on 32-bit
1069        else:
1070            count = 0
1071            for c in b:
1072                self.assertEqual(c, 'A')
1073                count += 1
1074            self.assertEqual(count, size*4)
1075
1076def test_main():
1077    test_support.run_unittest(StrTest, TupleTest, ListTest, BufferTest)
1078
1079if __name__ == '__main__':
1080    if len(sys.argv) > 1:
1081        test_support.set_memlimit(sys.argv[1])
1082    test_main()
1083