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
36class 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.assertEqual(caps[-len(SUBSTR):],
43                         SUBSTR.capitalize())
44        self.assertEqual(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.assertEqual(len(s), size)
51        lpadsize = rpadsize = (len(s) - len(SUBSTR)) // 2
52        if len(s) % 2:
53            lpadsize += 1
54        self.assertEqual(s[lpadsize:-rpadsize], SUBSTR)
55        self.assertEqual(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.assertEqual(len(s), size)
66            lpadsize = rpadsize = (len(s) - len(SUBSTR)) // 2
67            if len(s) % 2:
68                lpadsize += 1
69            self.assertEqual(s[lpadsize:-rpadsize], SUBSTR)
70            self.assertEqual(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.assertEqual(s.count('.'), size)
78        s += '.'
79        self.assertEqual(s.count('.'), size + 1)
80        self.assertEqual(s.count(' '), 3)
81        self.assertEqual(s.count('i'), 1)
82        self.assertEqual(s.count('j'), 0)
83
84    @bigmemtest(minsize=_2G + 2, memuse=3)
85    def test_decode(self, size):
86        s = '.' * size
87        self.assertEqual(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.assertEqual(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=4)
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        self.skipTest("test crashes - see issue #14904")
128        try:
129            s = u"\uAAAA"*size
130            r = repr(s)
131        except MemoryError:
132            pass # acceptable on 32-bit
133        else:
134            self.assertTrue(s == eval(r))
135
136    @bigmemtest(minsize=_2G, memuse=2)
137    def test_endswith(self, size):
138        SUBSTR = ' abc def ghi'
139        s = '-' * size + SUBSTR
140        self.assertTrue(s.endswith(SUBSTR))
141        self.assertTrue(s.endswith(s))
142        s2 = '...' + s
143        self.assertTrue(s2.endswith(s))
144        self.assertFalse(s.endswith('a' + SUBSTR))
145        self.assertFalse(SUBSTR.endswith(s))
146
147    @bigmemtest(minsize=_2G + 10, memuse=2)
148    def test_expandtabs(self, size):
149        s = '-' * size
150        tabsize = 8
151        self.assertEqual(s.expandtabs(), s)
152        del s
153        slen, remainder = divmod(size, tabsize)
154        s = '       \t' * slen
155        s = s.expandtabs(tabsize)
156        self.assertEqual(len(s), size - remainder)
157        self.assertEqual(len(s.strip(' ')), 0)
158
159    @bigmemtest(minsize=_2G, memuse=2)
160    def test_find(self, size):
161        SUBSTR = ' abc def ghi'
162        sublen = len(SUBSTR)
163        s = ''.join([SUBSTR, '-' * size, SUBSTR])
164        self.assertEqual(s.find(' '), 0)
165        self.assertEqual(s.find(SUBSTR), 0)
166        self.assertEqual(s.find(' ', sublen), sublen + size)
167        self.assertEqual(s.find(SUBSTR, len(SUBSTR)), sublen + size)
168        self.assertEqual(s.find('i'), SUBSTR.find('i'))
169        self.assertEqual(s.find('i', sublen),
170                         sublen + size + SUBSTR.find('i'))
171        self.assertEqual(s.find('i', size),
172                         sublen + size + SUBSTR.find('i'))
173        self.assertEqual(s.find('j'), -1)
174
175    @bigmemtest(minsize=_2G, memuse=2)
176    def test_index(self, size):
177        SUBSTR = ' abc def ghi'
178        sublen = len(SUBSTR)
179        s = ''.join([SUBSTR, '-' * size, SUBSTR])
180        self.assertEqual(s.index(' '), 0)
181        self.assertEqual(s.index(SUBSTR), 0)
182        self.assertEqual(s.index(' ', sublen), sublen + size)
183        self.assertEqual(s.index(SUBSTR, sublen), sublen + size)
184        self.assertEqual(s.index('i'), SUBSTR.index('i'))
185        self.assertEqual(s.index('i', sublen),
186                         sublen + size + SUBSTR.index('i'))
187        self.assertEqual(s.index('i', size),
188                         sublen + size + SUBSTR.index('i'))
189        self.assertRaises(ValueError, s.index, 'j')
190
191    @bigmemtest(minsize=_2G, memuse=2)
192    def test_isalnum(self, size):
193        SUBSTR = '123456'
194        s = 'a' * size + SUBSTR
195        self.assertTrue(s.isalnum())
196        s += '.'
197        self.assertFalse(s.isalnum())
198
199    @bigmemtest(minsize=_2G, memuse=2)
200    def test_isalpha(self, size):
201        SUBSTR = 'zzzzzzz'
202        s = 'a' * size + SUBSTR
203        self.assertTrue(s.isalpha())
204        s += '.'
205        self.assertFalse(s.isalpha())
206
207    @bigmemtest(minsize=_2G, memuse=2)
208    def test_isdigit(self, size):
209        SUBSTR = '123456'
210        s = '9' * size + SUBSTR
211        self.assertTrue(s.isdigit())
212        s += 'z'
213        self.assertFalse(s.isdigit())
214
215    @bigmemtest(minsize=_2G, memuse=2)
216    def test_islower(self, size):
217        chars = ''.join([ chr(c) for c in range(255) if not chr(c).isupper() ])
218        repeats = size // len(chars) + 2
219        s = chars * repeats
220        self.assertTrue(s.islower())
221        s += 'A'
222        self.assertFalse(s.islower())
223
224    @bigmemtest(minsize=_2G, memuse=2)
225    def test_isspace(self, size):
226        whitespace = ' \f\n\r\t\v'
227        repeats = size // len(whitespace) + 2
228        s = whitespace * repeats
229        self.assertTrue(s.isspace())
230        s += 'j'
231        self.assertFalse(s.isspace())
232
233    @bigmemtest(minsize=_2G, memuse=2)
234    def test_istitle(self, size):
235        SUBSTR = '123456'
236        s = ''.join(['A', 'a' * size, SUBSTR])
237        self.assertTrue(s.istitle())
238        s += 'A'
239        self.assertTrue(s.istitle())
240        s += 'aA'
241        self.assertFalse(s.istitle())
242
243    @bigmemtest(minsize=_2G, memuse=2)
244    def test_isupper(self, size):
245        chars = ''.join([ chr(c) for c in range(255) if not chr(c).islower() ])
246        repeats = size // len(chars) + 2
247        s = chars * repeats
248        self.assertTrue(s.isupper())
249        s += 'a'
250        self.assertFalse(s.isupper())
251
252    @bigmemtest(minsize=_2G, memuse=2)
253    def test_join(self, size):
254        s = 'A' * size
255        x = s.join(['aaaaa', 'bbbbb'])
256        self.assertEqual(x.count('a'), 5)
257        self.assertEqual(x.count('b'), 5)
258        self.assertTrue(x.startswith('aaaaaA'))
259        self.assertTrue(x.endswith('Abbbbb'))
260
261    @bigmemtest(minsize=_2G + 10, memuse=1)
262    def test_ljust(self, size):
263        SUBSTR = ' abc def ghi'
264        s = SUBSTR.ljust(size)
265        self.assertTrue(s.startswith(SUBSTR + '  '))
266        self.assertEqual(len(s), size)
267        self.assertEqual(s.strip(), SUBSTR.strip())
268
269    @bigmemtest(minsize=_2G + 10, memuse=2)
270    def test_lower(self, size):
271        s = 'A' * size
272        s = s.lower()
273        self.assertEqual(len(s), size)
274        self.assertEqual(s.count('a'), size)
275
276    @bigmemtest(minsize=_2G + 10, memuse=1)
277    def test_lstrip(self, size):
278        SUBSTR = 'abc def ghi'
279        s = SUBSTR.rjust(size)
280        self.assertEqual(len(s), size)
281        self.assertEqual(s.lstrip(), SUBSTR.lstrip())
282        del s
283        s = SUBSTR.ljust(size)
284        self.assertEqual(len(s), size)
285        stripped = s.lstrip()
286        self.assertTrue(stripped is s)
287
288    @bigmemtest(minsize=_2G + 10, memuse=2)
289    def test_replace(self, size):
290        replacement = 'a'
291        s = ' ' * size
292        s = s.replace(' ', replacement)
293        self.assertEqual(len(s), size)
294        self.assertEqual(s.count(replacement), size)
295        s = s.replace(replacement, ' ', size - 4)
296        self.assertEqual(len(s), size)
297        self.assertEqual(s.count(replacement), 4)
298        self.assertEqual(s[-10:], '      aaaa')
299
300    @bigmemtest(minsize=_2G, memuse=2)
301    def test_rfind(self, size):
302        SUBSTR = ' abc def ghi'
303        sublen = len(SUBSTR)
304        s = ''.join([SUBSTR, '-' * size, SUBSTR])
305        self.assertEqual(s.rfind(' '), sublen + size + SUBSTR.rfind(' '))
306        self.assertEqual(s.rfind(SUBSTR), sublen + size)
307        self.assertEqual(s.rfind(' ', 0, size), SUBSTR.rfind(' '))
308        self.assertEqual(s.rfind(SUBSTR, 0, sublen + size), 0)
309        self.assertEqual(s.rfind('i'), sublen + size + SUBSTR.rfind('i'))
310        self.assertEqual(s.rfind('i', 0, sublen), SUBSTR.rfind('i'))
311        self.assertEqual(s.rfind('i', 0, sublen + size),
312                         SUBSTR.rfind('i'))
313        self.assertEqual(s.rfind('j'), -1)
314
315    @bigmemtest(minsize=_2G, memuse=2)
316    def test_rindex(self, size):
317        SUBSTR = ' abc def ghi'
318        sublen = len(SUBSTR)
319        s = ''.join([SUBSTR, '-' * size, SUBSTR])
320        self.assertEqual(s.rindex(' '),
321                          sublen + size + SUBSTR.rindex(' '))
322        self.assertEqual(s.rindex(SUBSTR), sublen + size)
323        self.assertEqual(s.rindex(' ', 0, sublen + size - 1),
324                         SUBSTR.rindex(' '))
325        self.assertEqual(s.rindex(SUBSTR, 0, sublen + size), 0)
326        self.assertEqual(s.rindex('i'),
327                         sublen + size + SUBSTR.rindex('i'))
328        self.assertEqual(s.rindex('i', 0, sublen), SUBSTR.rindex('i'))
329        self.assertEqual(s.rindex('i', 0, sublen + size),
330                         SUBSTR.rindex('i'))
331        self.assertRaises(ValueError, s.rindex, 'j')
332
333    @bigmemtest(minsize=_2G + 10, memuse=1)
334    def test_rjust(self, size):
335        SUBSTR = ' abc def ghi'
336        s = SUBSTR.ljust(size)
337        self.assertTrue(s.startswith(SUBSTR + '  '))
338        self.assertEqual(len(s), size)
339        self.assertEqual(s.strip(), SUBSTR.strip())
340
341    @bigmemtest(minsize=_2G + 10, memuse=1)
342    def test_rstrip(self, size):
343        SUBSTR = ' abc def ghi'
344        s = SUBSTR.ljust(size)
345        self.assertEqual(len(s), size)
346        self.assertEqual(s.rstrip(), SUBSTR.rstrip())
347        del s
348        s = SUBSTR.rjust(size)
349        self.assertEqual(len(s), size)
350        stripped = s.rstrip()
351        self.assertTrue(stripped is s)
352
353    # The test takes about size bytes to build a string, and then about
354    # sqrt(size) substrings of sqrt(size) in size and a list to
355    # hold sqrt(size) items. It's close but just over 2x size.
356    @bigmemtest(minsize=_2G, memuse=2.1)
357    def test_split_small(self, size):
358        # Crudely calculate an estimate so that the result of s.split won't
359        # take up an inordinate amount of memory
360        chunksize = int(size ** 0.5 + 2)
361        SUBSTR = 'a' + ' ' * chunksize
362        s = SUBSTR * chunksize
363        l = s.split()
364        self.assertEqual(len(l), chunksize)
365        self.assertEqual(set(l), set(['a']))
366        del l
367        l = s.split('a')
368        self.assertEqual(len(l), chunksize + 1)
369        self.assertEqual(set(l), set(['', ' ' * chunksize]))
370
371    # Allocates a string of twice size (and briefly two) and a list of
372    # size.  Because of internal affairs, the s.split() call produces a
373    # list of size times the same one-character string, so we only
374    # suffer for the list size. (Otherwise, it'd cost another 48 times
375    # size in bytes!) Nevertheless, a list of size takes
376    # 8*size bytes.
377    @bigmemtest(minsize=_2G + 5, memuse=10)
378    def test_split_large(self, size):
379        s = ' a' * size + ' '
380        l = s.split()
381        self.assertEqual(len(l), size)
382        self.assertEqual(set(l), set(['a']))
383        del l
384        l = s.split('a')
385        self.assertEqual(len(l), size + 1)
386        self.assertEqual(set(l), set([' ']))
387
388    @bigmemtest(minsize=_2G, memuse=2.1)
389    def test_splitlines(self, size):
390        # Crudely calculate an estimate so that the result of s.split won't
391        # take up an inordinate amount of memory
392        chunksize = int(size ** 0.5 + 2) // 2
393        SUBSTR = ' ' * chunksize + '\n' + ' ' * chunksize + '\r\n'
394        s = SUBSTR * chunksize
395        l = s.splitlines()
396        self.assertEqual(len(l), chunksize * 2)
397        self.assertEqual(set(l), set([' ' * chunksize]))
398
399    @bigmemtest(minsize=_2G, memuse=2)
400    def test_startswith(self, size):
401        SUBSTR = ' abc def ghi'
402        s = '-' * size + SUBSTR
403        self.assertTrue(s.startswith(s))
404        self.assertTrue(s.startswith('-' * size))
405        self.assertFalse(s.startswith(SUBSTR))
406
407    @bigmemtest(minsize=_2G, memuse=1)
408    def test_strip(self, size):
409        SUBSTR = '   abc def ghi   '
410        s = SUBSTR.rjust(size)
411        self.assertEqual(len(s), size)
412        self.assertEqual(s.strip(), SUBSTR.strip())
413        del s
414        s = SUBSTR.ljust(size)
415        self.assertEqual(len(s), size)
416        self.assertEqual(s.strip(), SUBSTR.strip())
417
418    @bigmemtest(minsize=_2G, memuse=2)
419    def test_swapcase(self, size):
420        SUBSTR = "aBcDeFG12.'\xa9\x00"
421        sublen = len(SUBSTR)
422        repeats = size // sublen + 2
423        s = SUBSTR * repeats
424        s = s.swapcase()
425        self.assertEqual(len(s), sublen * repeats)
426        self.assertEqual(s[:sublen * 3], SUBSTR.swapcase() * 3)
427        self.assertEqual(s[-sublen * 3:], SUBSTR.swapcase() * 3)
428
429    @bigmemtest(minsize=_2G, memuse=2)
430    def test_title(self, size):
431        SUBSTR = 'SpaaHAaaAaham'
432        s = SUBSTR * (size // len(SUBSTR) + 2)
433        s = s.title()
434        self.assertTrue(s.startswith((SUBSTR * 3).title()))
435        self.assertTrue(s.endswith(SUBSTR.lower() * 3))
436
437    @bigmemtest(minsize=_2G, memuse=2)
438    def test_translate(self, size):
439        trans = string.maketrans('.aZ', '-!$')
440        SUBSTR = 'aZz.z.Aaz.'
441        sublen = len(SUBSTR)
442        repeats = size // sublen + 2
443        s = SUBSTR * repeats
444        s = s.translate(trans)
445        self.assertEqual(len(s), repeats * sublen)
446        self.assertEqual(s[:sublen], SUBSTR.translate(trans))
447        self.assertEqual(s[-sublen:], SUBSTR.translate(trans))
448        self.assertEqual(s.count('.'), 0)
449        self.assertEqual(s.count('!'), repeats * 2)
450        self.assertEqual(s.count('z'), repeats * 3)
451
452    @bigmemtest(minsize=_2G + 5, memuse=2)
453    def test_upper(self, size):
454        s = 'a' * size
455        s = s.upper()
456        self.assertEqual(len(s), size)
457        self.assertEqual(s.count('A'), size)
458
459    @bigmemtest(minsize=_2G + 20, memuse=1)
460    def test_zfill(self, size):
461        SUBSTR = '-568324723598234'
462        s = SUBSTR.zfill(size)
463        self.assertTrue(s.endswith('0' + SUBSTR[1:]))
464        self.assertTrue(s.startswith('-0'))
465        self.assertEqual(len(s), size)
466        self.assertEqual(s.count('0'), size - len(SUBSTR))
467
468    @bigmemtest(minsize=_2G + 10, memuse=2)
469    def test_format(self, size):
470        s = '-' * size
471        sf = '%s' % (s,)
472        self.assertTrue(s == sf)
473        del sf
474        sf = '..%s..' % (s,)
475        self.assertEqual(len(sf), len(s) + 4)
476        self.assertTrue(sf.startswith('..-'))
477        self.assertTrue(sf.endswith('-..'))
478        del s, sf
479
480        size //= 2
481        edge = '-' * size
482        s = ''.join([edge, '%s', edge])
483        del edge
484        s = s % '...'
485        self.assertEqual(len(s), size * 2 + 3)
486        self.assertEqual(s.count('.'), 3)
487        self.assertEqual(s.count('-'), size * 2)
488
489    @bigmemtest(minsize=_2G + 10, memuse=5)
490    def test_repr_small(self, size):
491        s = '-' * size
492        s = repr(s)
493        self.assertEqual(len(s), size + 2)
494        self.assertEqual(s[0], "'")
495        self.assertEqual(s[-1], "'")
496        self.assertEqual(s.count('-'), size)
497        del s
498        # repr() will create a string four times as large as this 'binary
499        # string', but we don't want to allocate much more than twice
500        # size in total.  (We do extra testing in test_repr_large())
501        s = '\x00' * size
502        s = repr(s)
503        self.assertEqual(len(s), size * 4 + 2)
504        self.assertEqual(s[0], "'")
505        self.assertEqual(s[-1], "'")
506        self.assertEqual(s.count('\\'), size)
507        self.assertEqual(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.assertEqual(len(s), size * 4 + 2)
514        self.assertEqual(s[0], "'")
515        self.assertEqual(s[-1], "'")
516        self.assertEqual(s.count('\\'), size)
517        self.assertEqual(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.assertTrue(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.assertEqual(len(s), size)
530        s = s + s
531        self.assertEqual(len(s), size * 2)
532        self.assertEqual(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.assertEqual(len(s), size)
540        s = s * 2
541        self.assertEqual(len(s), size * 2)
542        self.assertEqual(s.count('.'), size * 2)
543
544    @bigmemtest(minsize=_2G + 20, memuse=2)
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.assertEqual(s[i], SUBSTR[0])
553            self.assertEqual(s[i:i + sublen], SUBSTR)
554            self.assertEqual(s[i:i + sublen:2], SUBSTR[::2])
555            if i > 0:
556                self.assertEqual(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.assertEqual(s[len(s) - 1], SUBSTR[-1])
561        self.assertEqual(s[-1], SUBSTR[-1])
562        self.assertEqual(s[len(s) - 10], SUBSTR[0])
563        self.assertEqual(s[-sublen], SUBSTR[0])
564        self.assertEqual(s[len(s):], '')
565        self.assertEqual(s[len(s) - 1:], SUBSTR[-1])
566        self.assertEqual(s[-1:], SUBSTR[-1])
567        self.assertEqual(s[len(s) - sublen:], SUBSTR)
568        self.assertEqual(s[-sublen:], SUBSTR)
569        self.assertEqual(len(s[:]), len(s))
570        self.assertEqual(len(s[:len(s) - 5]), len(s) - 5)
571        self.assertEqual(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.assertIn(SUBSTR, s)
584        self.assertNotIn(SUBSTR * 2, s)
585        self.assertIn('-', s)
586        self.assertNotIn('a', s)
587        s += 'a'
588        self.assertIn('a', s)
589
590    @bigmemtest(minsize=_2G + 10, memuse=2)
591    def test_compare(self, size):
592        s1 = '-' * size
593        s2 = '-' * size
594        self.assertTrue(s1 == s2)
595        del s2
596        s2 = s1 + 'a'
597        self.assertFalse(s1 == s2)
598        del s2
599        s2 = '.' * size
600        self.assertFalse(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.assertFalse(h1 == hash(s))
615
616class 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.assertTrue(t1 == t2)
632        del t2
633        t2 = (u'',) * (size + 1)
634        self.assertFalse(t1 == t2)
635        del t2
636        t2 = (1,) * size
637        self.assertFalse(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.assertEqual(len(t), size)
647        t = t + t
648        self.assertEqual(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.assertEqual(len(t), size * 5)
662        self.assertIn(5, t)
663        self.assertNotIn((1, 2, 3, 4, 5), t)
664        self.assertNotIn(0, 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.assertFalse(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.assertEqual(len(t), size)
678        self.assertEqual(t[-1], None)
679        self.assertEqual(t[5], None)
680        self.assertEqual(t[size - 1], None)
681        self.assertRaises(IndexError, operator.getitem, t, size)
682        self.assertEqual(t[:5], (None,) * 5)
683        self.assertEqual(t[-5:], (None,) * 5)
684        self.assertEqual(t[20:25], (None,) * 5)
685        self.assertEqual(t[-25:-20], (None,) * 5)
686        self.assertEqual(t[size - 5:], (None,) * 5)
687        self.assertEqual(t[size - 5:size], (None,) * 5)
688        self.assertEqual(t[size - 6:size - 2], (None,) * 4)
689        self.assertEqual(t[size:size], ())
690        self.assertEqual(t[size:size+5], ())
691
692    # Like test_concat, split in two.
693    def basic_test_repeat(self, size):
694        t = ('',) * size
695        self.assertEqual(len(t), size)
696        t = t * 2
697        self.assertEqual(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.assertEqual(item, count)
721                count += 1
722            self.assertEqual(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.assertEqual(item, count)
731                count += 1
732            self.assertEqual(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.assertEqual(len(s), size * 3)
742        self.assertEqual(s[:5], '(0, 0')
743        self.assertEqual(s[-5:], '0, 0)')
744        self.assertEqual(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
754class 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.assertTrue(l1 == l2)
766        del l2
767        l2 = [u''] * (size + 1)
768        self.assertFalse(l1 == l2)
769        del l2
770        l2 = [2] * size
771        self.assertFalse(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.assertEqual(len(l), size)
781        l = l + l
782        self.assertEqual(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.assertEqual(len(l), size * 2)
796        self.assertTrue(l[0] is l[-1])
797        self.assertTrue(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.assertEqual(len(l), size * 5)
811        self.assertIn(5, l)
812        self.assertNotIn([1, 2, 3, 4, 5], l)
813        self.assertNotIn(0, l)
814
815    @bigmemtest(minsize=_2G + 10, memuse=8)
816    def test_hash(self, size):
817        l = [0] * size
818        self.assertRaises(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.assertEqual(len(l), size)
824        self.assertEqual(l[-1], None)
825        self.assertEqual(l[5], None)
826        self.assertEqual(l[size - 1], None)
827        self.assertRaises(IndexError, operator.getitem, l, size)
828        self.assertEqual(l[:5], [None] * 5)
829        self.assertEqual(l[-5:], [None] * 5)
830        self.assertEqual(l[20:25], [None] * 5)
831        self.assertEqual(l[-25:-20], [None] * 5)
832        self.assertEqual(l[size - 5:], [None] * 5)
833        self.assertEqual(l[size - 5:size], [None] * 5)
834        self.assertEqual(l[size - 6:size - 2], [None] * 4)
835        self.assertEqual(l[size:size], [])
836        self.assertEqual(l[size:size+5], [])
837
838        l[size - 2] = 5
839        self.assertEqual(len(l), size)
840        self.assertEqual(l[-3:], [None, 5, None])
841        self.assertEqual(l.count(5), 1)
842        self.assertRaises(IndexError, operator.setitem, l, size, 6)
843        self.assertEqual(len(l), size)
844
845        l[size - 7:] = [1, 2, 3, 4, 5]
846        size -= 2
847        self.assertEqual(len(l), size)
848        self.assertEqual(l[-7:], [None, None, 1, 2, 3, 4, 5])
849
850        l[:7] = [1, 2, 3, 4, 5]
851        size -= 2
852        self.assertEqual(len(l), size)
853        self.assertEqual(l[:7], [1, 2, 3, 4, 5, None, None])
854
855        del l[size - 1]
856        size -= 1
857        self.assertEqual(len(l), size)
858        self.assertEqual(l[-1], 4)
859
860        del l[-2:]
861        size -= 2
862        self.assertEqual(len(l), size)
863        self.assertEqual(l[-1], 2)
864
865        del l[0]
866        size -= 1
867        self.assertEqual(len(l), size)
868        self.assertEqual(l[0], 2)
869
870        del l[:2]
871        size -= 2
872        self.assertEqual(len(l), size)
873        self.assertEqual(l[0], 4)
874
875    # Like test_concat, split in two.
876    def basic_test_repeat(self, size):
877        l = [] * size
878        self.assertFalse(l)
879        l = [''] * size
880        self.assertEqual(len(l), size)
881        l = l * 2
882        self.assertEqual(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.assertEqual(len(l), size)
896        self.assertTrue(l[0] is l[-1])
897        del l
898
899        l = [''] * size
900        l *= 2
901        self.assertEqual(len(l), size * 2)
902        self.assertTrue(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.assertEqual(len(s), size * 3)
917        self.assertEqual(s[:5], '[0, 0')
918        self.assertEqual(s[-5:], '0, 0]')
919        self.assertEqual(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.assertEqual(len(l), size+1)
936        self.assertTrue(l[-3] is l[-2])
937        self.assertFalse(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.assertEqual(l.count(1), size)
943        self.assertEqual(l.count("1"), 0)
944
945    def basic_test_extend(self, size):
946        l = [file] * size
947        l.extend(l)
948        self.assertEqual(len(l), size * 2)
949        self.assertTrue(l[0] is l[-1])
950        self.assertTrue(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.assertEqual(l.index(1), 0)
965        self.assertEqual(l.index(5, size - 5), size - 1)
966        self.assertEqual(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.assertEqual(len(l), size)
977        self.assertEqual(l[-3:], [1.0, "A", 1.0])
978
979        l.insert(size + 1, "B")
980        size += 1
981        self.assertEqual(len(l), size)
982        self.assertEqual(l[-3:], ["A", 1.0, "B"])
983
984        l.insert(1, "C")
985        size += 1
986        self.assertEqual(len(l), size)
987        self.assertEqual(l[:3], [1.0, "C", 1.0])
988        self.assertEqual(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.assertEqual(len(l), size)
995
996        item = l.pop()
997        size -= 1
998        self.assertEqual(len(l), size)
999        self.assertEqual(item, u"e")
1000        self.assertEqual(l[-2:], [u"c", u"d"])
1001
1002        item = l.pop(0)
1003        size -= 1
1004        self.assertEqual(len(l), size)
1005        self.assertEqual(item, u"a")
1006        self.assertEqual(l[:2], [u"b", u"c"])
1007
1008        item = l.pop(size - 2)
1009        size -= 1
1010        self.assertEqual(len(l), size)
1011        self.assertEqual(item, u"c")
1012        self.assertEqual(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.assertEqual(len(l), size)
1018
1019        l.remove(10)
1020        size -= 1
1021        self.assertEqual(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.assertEqual(len(l), size)
1028        self.assertEqual(l[-2:], [10, 5])
1029        l.remove(5)
1030        size -= 1
1031        self.assertEqual(len(l), size)
1032        self.assertEqual(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.assertEqual(len(l), size * 5)
1039        self.assertEqual(l[-5:], [5, 4, 3, 2, 1])
1040        self.assertEqual(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.assertEqual(len(l), size * 5)
1047        self.assertEqual(l.count(1), size)
1048        self.assertEqual(l[:10], [1] * 10)
1049        self.assertEqual(l[-10:], [5] * 10)
1050
1051class BufferTest(unittest.TestCase):
1052
1053    @precisionbigmemtest(size=_1G, memuse=4)
1054    def test_repeat(self, size):
1055        try:
1056            with test_support.check_py3k_warnings():
1057                b = buffer("AAAA")*size
1058        except MemoryError:
1059            pass # acceptable on 32-bit
1060        else:
1061            count = 0
1062            for c in b:
1063                self.assertEqual(c, 'A')
1064                count += 1
1065            self.assertEqual(count, size*4)
1066
1067def test_main():
1068    test_support.run_unittest(StrTest, TupleTest, ListTest, BufferTest)
1069
1070if __name__ == '__main__':
1071    if len(sys.argv) > 1:
1072        test_support.set_memlimit(sys.argv[1])
1073    test_main()
1074