Python code coverage for Lib/test/test_bigmem.py

#	count	content
1	n/a	"""Bigmem tests - tests for the 32-bit boundary in containers.
2	n/a
3	n/a	These tests try to exercise the 32-bit boundary that is sometimes, if
4	n/a	rarely, exceeded in practice, but almost never tested. They are really only
5	n/a	meaningful on 64-bit builds on machines with a lot of memory, but the
6	n/a	tests are always run, usually with very low memory limits to make sure the
7	n/a	tests themselves don't suffer from bitrot. To run them for real, pass a
8	n/a	high memory limit to regrtest, with the -M option.
9	n/a	"""
10	n/a
11	n/a	from test import support
12	n/a	from test.support import bigmemtest, _1G, _2G, _4G
13	n/a
14	n/a	import unittest
15	n/a	import operator
16	n/a	import sys
17	n/a
18	n/a	# These tests all use one of the bigmemtest decorators to indicate how much
19	n/a	# memory they use and how much memory they need to be even meaningful. The
20	n/a	# decorators take two arguments: a 'memuse' indicator declaring
21	n/a	# (approximate) bytes per size-unit the test will use (at peak usage), and a
22	n/a	# 'minsize' indicator declaring a minimum useful size. A test that
23	n/a	# allocates a bytestring to test various operations near the end will have a
24	n/a	# minsize of at least 2Gb (or it wouldn't reach the 32-bit limit, so the
25	n/a	# test wouldn't be very useful) and a memuse of 1 (one byte per size-unit,
26	n/a	# if it allocates only one big string at a time.)
27	n/a	#
28	n/a	# When run with a memory limit set, both decorators skip tests that need
29	n/a	# more memory than available to be meaningful. The precisionbigmemtest will
30	n/a	# always pass minsize as size, even if there is much more memory available.
31	n/a	# The bigmemtest decorator will scale size upward to fill available memory.
32	n/a	#
33	n/a	# Bigmem testing houserules:
34	n/a	#
35	n/a	# - Try not to allocate too many large objects. It's okay to rely on
36	n/a	# refcounting semantics, and don't forget that 's = create_largestring()'
37	n/a	# doesn't release the old 's' (if it exists) until well after its new
38	n/a	# value has been created. Use 'del s' before the create_largestring call.
39	n/a	#
40	n/a	# - Do not compare large objects using assertEqual, assertIn or similar.
41	n/a	# It's a lengthy operation and the errormessage will be utterly useless
42	n/a	# due to its size. To make sure whether a result has the right contents,
43	n/a	# better to use the strip or count methods, or compare meaningful slices.
44	n/a	#
45	n/a	# - Don't forget to test for large indices, offsets and results and such,
46	n/a	# in addition to large sizes. Anything that probes the 32-bit boundary.
47	n/a	#
48	n/a	# - When repeating an object (say, a substring, or a small list) to create
49	n/a	# a large object, make the subobject of a length that is not a power of
50	n/a	# 2. That way, int-wrapping problems are more easily detected.
51	n/a	#
52	n/a	# - Despite the bigmemtest decorator, all tests will actually be called
53	n/a	# with a much smaller number too, in the normal test run (5Kb currently.)
54	n/a	# This is so the tests themselves get frequent testing.
55	n/a	# Consequently, always make all large allocations based on the
56	n/a	# passed-in 'size', and don't rely on the size being very large. Also,
57	n/a	# memuse-per-size should remain sane (less than a few thousand); if your
58	n/a	# test uses more, adjust 'size' upward, instead.
59	n/a
60	n/a	# BEWARE: it seems that one failing test can yield other subsequent tests to
61	n/a	# fail as well. I do not know whether it is due to memory fragmentation
62	n/a	# issues, or other specifics of the platform malloc() routine.
63	n/a
64	n/a	ascii_char_size = 1
65	n/a	ucs2_char_size = 2
66	n/a	ucs4_char_size = 4
67	n/a
68	n/a
69	n/a	class BaseStrTest:
70	n/a
71	n/a	def _test_capitalize(self, size):
72	n/a	_ = self.from_latin1
73	n/a	SUBSTR = self.from_latin1(' abc def ghi')
74	n/a	s = _('-') * size + SUBSTR
75	n/a	caps = s.capitalize()
76	n/a	self.assertEqual(caps[-len(SUBSTR):],
77	n/a	SUBSTR.capitalize())
78	n/a	self.assertEqual(caps.lstrip(_('-')), SUBSTR)
79	n/a
80	n/a	@bigmemtest(size=_2G + 10, memuse=1)
81	n/a	def test_center(self, size):
82	n/a	SUBSTR = self.from_latin1(' abc def ghi')
83	n/a	s = SUBSTR.center(size)
84	n/a	self.assertEqual(len(s), size)
85	n/a	lpadsize = rpadsize = (len(s) - len(SUBSTR)) // 2
86	n/a	if len(s) % 2:
87	n/a	lpadsize += 1
88	n/a	self.assertEqual(s[lpadsize:-rpadsize], SUBSTR)
89	n/a	self.assertEqual(s.strip(), SUBSTR.strip())
90	n/a
91	n/a	@bigmemtest(size=_2G, memuse=2)
92	n/a	def test_count(self, size):
93	n/a	_ = self.from_latin1
94	n/a	SUBSTR = _(' abc def ghi')
95	n/a	s = _('.') * size + SUBSTR
96	n/a	self.assertEqual(s.count(_('.')), size)
97	n/a	s += _('.')
98	n/a	self.assertEqual(s.count(_('.')), size + 1)
99	n/a	self.assertEqual(s.count(_(' ')), 3)
100	n/a	self.assertEqual(s.count(_('i')), 1)
101	n/a	self.assertEqual(s.count(_('j')), 0)
102	n/a
103	n/a	@bigmemtest(size=_2G, memuse=2)
104	n/a	def test_endswith(self, size):
105	n/a	_ = self.from_latin1
106	n/a	SUBSTR = _(' abc def ghi')
107	n/a	s = _('-') * size + SUBSTR
108	n/a	self.assertTrue(s.endswith(SUBSTR))
109	n/a	self.assertTrue(s.endswith(s))
110	n/a	s2 = _('...') + s
111	n/a	self.assertTrue(s2.endswith(s))
112	n/a	self.assertFalse(s.endswith(_('a') + SUBSTR))
113	n/a	self.assertFalse(SUBSTR.endswith(s))
114	n/a
115	n/a	@bigmemtest(size=_2G + 10, memuse=2)
116	n/a	def test_expandtabs(self, size):
117	n/a	_ = self.from_latin1
118	n/a	s = _('-') * size
119	n/a	tabsize = 8
120	n/a	self.assertTrue(s.expandtabs() == s)
121	n/a	del s
122	n/a	slen, remainder = divmod(size, tabsize)
123	n/a	s = _(' \t') * slen
124	n/a	s = s.expandtabs(tabsize)
125	n/a	self.assertEqual(len(s), size - remainder)
126	n/a	self.assertEqual(len(s.strip(_(' '))), 0)
127	n/a
128	n/a	@bigmemtest(size=_2G, memuse=2)
129	n/a	def test_find(self, size):
130	n/a	_ = self.from_latin1
131	n/a	SUBSTR = _(' abc def ghi')
132	n/a	sublen = len(SUBSTR)
133	n/a	s = _('').join([SUBSTR, _('-') * size, SUBSTR])
134	n/a	self.assertEqual(s.find(_(' ')), 0)
135	n/a	self.assertEqual(s.find(SUBSTR), 0)
136	n/a	self.assertEqual(s.find(_(' '), sublen), sublen + size)
137	n/a	self.assertEqual(s.find(SUBSTR, len(SUBSTR)), sublen + size)
138	n/a	self.assertEqual(s.find(_('i')), SUBSTR.find(_('i')))
139	n/a	self.assertEqual(s.find(_('i'), sublen),
140	n/a	sublen + size + SUBSTR.find(_('i')))
141	n/a	self.assertEqual(s.find(_('i'), size),
142	n/a	sublen + size + SUBSTR.find(_('i')))
143	n/a	self.assertEqual(s.find(_('j')), -1)
144	n/a
145	n/a	@bigmemtest(size=_2G, memuse=2)
146	n/a	def test_index(self, size):
147	n/a	_ = self.from_latin1
148	n/a	SUBSTR = _(' abc def ghi')
149	n/a	sublen = len(SUBSTR)
150	n/a	s = _('').join([SUBSTR, _('-') * size, SUBSTR])
151	n/a	self.assertEqual(s.index(_(' ')), 0)
152	n/a	self.assertEqual(s.index(SUBSTR), 0)
153	n/a	self.assertEqual(s.index(_(' '), sublen), sublen + size)
154	n/a	self.assertEqual(s.index(SUBSTR, sublen), sublen + size)
155	n/a	self.assertEqual(s.index(_('i')), SUBSTR.index(_('i')))
156	n/a	self.assertEqual(s.index(_('i'), sublen),
157	n/a	sublen + size + SUBSTR.index(_('i')))
158	n/a	self.assertEqual(s.index(_('i'), size),
159	n/a	sublen + size + SUBSTR.index(_('i')))
160	n/a	self.assertRaises(ValueError, s.index, _('j'))
161	n/a
162	n/a	@bigmemtest(size=_2G, memuse=2)
163	n/a	def test_isalnum(self, size):
164	n/a	_ = self.from_latin1
165	n/a	SUBSTR = _('123456')
166	n/a	s = _('a') * size + SUBSTR
167	n/a	self.assertTrue(s.isalnum())
168	n/a	s += _('.')
169	n/a	self.assertFalse(s.isalnum())
170	n/a
171	n/a	@bigmemtest(size=_2G, memuse=2)
172	n/a	def test_isalpha(self, size):
173	n/a	_ = self.from_latin1
174	n/a	SUBSTR = _('zzzzzzz')
175	n/a	s = _('a') * size + SUBSTR
176	n/a	self.assertTrue(s.isalpha())
177	n/a	s += _('.')
178	n/a	self.assertFalse(s.isalpha())
179	n/a
180	n/a	@bigmemtest(size=_2G, memuse=2)
181	n/a	def test_isdigit(self, size):
182	n/a	_ = self.from_latin1
183	n/a	SUBSTR = _('123456')
184	n/a	s = _('9') * size + SUBSTR
185	n/a	self.assertTrue(s.isdigit())
186	n/a	s += _('z')
187	n/a	self.assertFalse(s.isdigit())
188	n/a
189	n/a	@bigmemtest(size=_2G, memuse=2)
190	n/a	def test_islower(self, size):
191	n/a	_ = self.from_latin1
192	n/a	chars = _(''.join(
193	n/a	chr(c) for c in range(255) if not chr(c).isupper()))
194	n/a	repeats = size // len(chars) + 2
195	n/a	s = chars * repeats
196	n/a	self.assertTrue(s.islower())
197	n/a	s += _('A')
198	n/a	self.assertFalse(s.islower())
199	n/a
200	n/a	@bigmemtest(size=_2G, memuse=2)
201	n/a	def test_isspace(self, size):
202	n/a	_ = self.from_latin1
203	n/a	whitespace = _(' \f\n\r\t\v')
204	n/a	repeats = size // len(whitespace) + 2
205	n/a	s = whitespace * repeats
206	n/a	self.assertTrue(s.isspace())
207	n/a	s += _('j')
208	n/a	self.assertFalse(s.isspace())
209	n/a
210	n/a	@bigmemtest(size=_2G, memuse=2)
211	n/a	def test_istitle(self, size):
212	n/a	_ = self.from_latin1
213	n/a	SUBSTR = _('123456')
214	n/a	s = _('').join([_('A'), _('a') * size, SUBSTR])
215	n/a	self.assertTrue(s.istitle())
216	n/a	s += _('A')
217	n/a	self.assertTrue(s.istitle())
218	n/a	s += _('aA')
219	n/a	self.assertFalse(s.istitle())
220	n/a
221	n/a	@bigmemtest(size=_2G, memuse=2)
222	n/a	def test_isupper(self, size):
223	n/a	_ = self.from_latin1
224	n/a	chars = _(''.join(
225	n/a	chr(c) for c in range(255) if not chr(c).islower()))
226	n/a	repeats = size // len(chars) + 2
227	n/a	s = chars * repeats
228	n/a	self.assertTrue(s.isupper())
229	n/a	s += _('a')
230	n/a	self.assertFalse(s.isupper())
231	n/a
232	n/a	@bigmemtest(size=_2G, memuse=2)
233	n/a	def test_join(self, size):
234	n/a	_ = self.from_latin1
235	n/a	s = _('A') * size
236	n/a	x = s.join([_('aaaaa'), _('bbbbb')])
237	n/a	self.assertEqual(x.count(_('a')), 5)
238	n/a	self.assertEqual(x.count(_('b')), 5)
239	n/a	self.assertTrue(x.startswith(_('aaaaaA')))
240	n/a	self.assertTrue(x.endswith(_('Abbbbb')))
241	n/a
242	n/a	@bigmemtest(size=_2G + 10, memuse=1)
243	n/a	def test_ljust(self, size):
244	n/a	_ = self.from_latin1
245	n/a	SUBSTR = _(' abc def ghi')
246	n/a	s = SUBSTR.ljust(size)
247	n/a	self.assertTrue(s.startswith(SUBSTR + _(' ')))
248	n/a	self.assertEqual(len(s), size)
249	n/a	self.assertEqual(s.strip(), SUBSTR.strip())
250	n/a
251	n/a	@bigmemtest(size=_2G + 10, memuse=2)
252	n/a	def test_lower(self, size):
253	n/a	_ = self.from_latin1
254	n/a	s = _('A') * size
255	n/a	s = s.lower()
256	n/a	self.assertEqual(len(s), size)
257	n/a	self.assertEqual(s.count(_('a')), size)
258	n/a
259	n/a	@bigmemtest(size=_2G + 10, memuse=1)
260	n/a	def test_lstrip(self, size):
261	n/a	_ = self.from_latin1
262	n/a	SUBSTR = _('abc def ghi')
263	n/a	s = SUBSTR.rjust(size)
264	n/a	self.assertEqual(len(s), size)
265	n/a	self.assertEqual(s.lstrip(), SUBSTR.lstrip())
266	n/a	del s
267	n/a	s = SUBSTR.ljust(size)
268	n/a	self.assertEqual(len(s), size)
269	n/a	# Type-specific optimization
270	n/a	if isinstance(s, (str, bytes)):
271	n/a	stripped = s.lstrip()
272	n/a	self.assertTrue(stripped is s)
273	n/a
274	n/a	@bigmemtest(size=_2G + 10, memuse=2)
275	n/a	def test_replace(self, size):
276	n/a	_ = self.from_latin1
277	n/a	replacement = _('a')
278	n/a	s = _(' ') * size
279	n/a	s = s.replace(_(' '), replacement)
280	n/a	self.assertEqual(len(s), size)
281	n/a	self.assertEqual(s.count(replacement), size)
282	n/a	s = s.replace(replacement, _(' '), size - 4)
283	n/a	self.assertEqual(len(s), size)
284	n/a	self.assertEqual(s.count(replacement), 4)
285	n/a	self.assertEqual(s[-10:], _(' aaaa'))
286	n/a
287	n/a	@bigmemtest(size=_2G, memuse=2)
288	n/a	def test_rfind(self, size):
289	n/a	_ = self.from_latin1
290	n/a	SUBSTR = _(' abc def ghi')
291	n/a	sublen = len(SUBSTR)
292	n/a	s = _('').join([SUBSTR, _('-') * size, SUBSTR])
293	n/a	self.assertEqual(s.rfind(_(' ')), sublen + size + SUBSTR.rfind(_(' ')))
294	n/a	self.assertEqual(s.rfind(SUBSTR), sublen + size)
295	n/a	self.assertEqual(s.rfind(_(' '), 0, size), SUBSTR.rfind(_(' ')))
296	n/a	self.assertEqual(s.rfind(SUBSTR, 0, sublen + size), 0)
297	n/a	self.assertEqual(s.rfind(_('i')), sublen + size + SUBSTR.rfind(_('i')))
298	n/a	self.assertEqual(s.rfind(_('i'), 0, sublen), SUBSTR.rfind(_('i')))
299	n/a	self.assertEqual(s.rfind(_('i'), 0, sublen + size),
300	n/a	SUBSTR.rfind(_('i')))
301	n/a	self.assertEqual(s.rfind(_('j')), -1)
302	n/a
303	n/a	@bigmemtest(size=_2G, memuse=2)
304	n/a	def test_rindex(self, size):
305	n/a	_ = self.from_latin1
306	n/a	SUBSTR = _(' abc def ghi')
307	n/a	sublen = len(SUBSTR)
308	n/a	s = _('').join([SUBSTR, _('-') * size, SUBSTR])
309	n/a	self.assertEqual(s.rindex(_(' ')),
310	n/a	sublen + size + SUBSTR.rindex(_(' ')))
311	n/a	self.assertEqual(s.rindex(SUBSTR), sublen + size)
312	n/a	self.assertEqual(s.rindex(_(' '), 0, sublen + size - 1),
313	n/a	SUBSTR.rindex(_(' ')))
314	n/a	self.assertEqual(s.rindex(SUBSTR, 0, sublen + size), 0)
315	n/a	self.assertEqual(s.rindex(_('i')),
316	n/a	sublen + size + SUBSTR.rindex(_('i')))
317	n/a	self.assertEqual(s.rindex(_('i'), 0, sublen), SUBSTR.rindex(_('i')))
318	n/a	self.assertEqual(s.rindex(_('i'), 0, sublen + size),
319	n/a	SUBSTR.rindex(_('i')))
320	n/a	self.assertRaises(ValueError, s.rindex, _('j'))
321	n/a
322	n/a	@bigmemtest(size=_2G + 10, memuse=1)
323	n/a	def test_rjust(self, size):
324	n/a	_ = self.from_latin1
325	n/a	SUBSTR = _(' abc def ghi')
326	n/a	s = SUBSTR.ljust(size)
327	n/a	self.assertTrue(s.startswith(SUBSTR + _(' ')))
328	n/a	self.assertEqual(len(s), size)
329	n/a	self.assertEqual(s.strip(), SUBSTR.strip())
330	n/a
331	n/a	@bigmemtest(size=_2G + 10, memuse=1)
332	n/a	def test_rstrip(self, size):
333	n/a	_ = self.from_latin1
334	n/a	SUBSTR = _(' abc def ghi')
335	n/a	s = SUBSTR.ljust(size)
336	n/a	self.assertEqual(len(s), size)
337	n/a	self.assertEqual(s.rstrip(), SUBSTR.rstrip())
338	n/a	del s
339	n/a	s = SUBSTR.rjust(size)
340	n/a	self.assertEqual(len(s), size)
341	n/a	# Type-specific optimization
342	n/a	if isinstance(s, (str, bytes)):
343	n/a	stripped = s.rstrip()
344	n/a	self.assertTrue(stripped is s)
345	n/a
346	n/a	# The test takes about size bytes to build a string, and then about
347	n/a	# sqrt(size) substrings of sqrt(size) in size and a list to
348	n/a	# hold sqrt(size) items. It's close but just over 2x size.
349	n/a	@bigmemtest(size=_2G, memuse=2.1)
350	n/a	def test_split_small(self, size):
351	n/a	_ = self.from_latin1
352	n/a	# Crudely calculate an estimate so that the result of s.split won't
353	n/a	# take up an inordinate amount of memory
354	n/a	chunksize = int(size ** 0.5 + 2)
355	n/a	SUBSTR = _('a') + _(' ') * chunksize
356	n/a	s = SUBSTR * chunksize
357	n/a	l = s.split()
358	n/a	self.assertEqual(len(l), chunksize)
359	n/a	expected = _('a')
360	n/a	for item in l:
361	n/a	self.assertEqual(item, expected)
362	n/a	del l
363	n/a	l = s.split(_('a'))
364	n/a	self.assertEqual(len(l), chunksize + 1)
365	n/a	expected = _(' ') * chunksize
366	n/a	for item in filter(None, l):
367	n/a	self.assertEqual(item, expected)
368	n/a
369	n/a	# Allocates a string of twice size (and briefly two) and a list of
370	n/a	# size. Because of internal affairs, the s.split() call produces a
371	n/a	# list of size times the same one-character string, so we only
372	n/a	# suffer for the list size. (Otherwise, it'd cost another 48 times
373	n/a	# size in bytes!) Nevertheless, a list of size takes
374	n/a	# 8*size bytes.
375	n/a	@bigmemtest(size=_2G + 5, memuse=2 * ascii_char_size + 8)
376	n/a	def test_split_large(self, size):
377	n/a	_ = self.from_latin1
378	n/a	s = _(' a') * size + _(' ')
379	n/a	l = s.split()
380	n/a	self.assertEqual(len(l), size)
381	n/a	self.assertEqual(set(l), set([_('a')]))
382	n/a	del l
383	n/a	l = s.split(_('a'))
384	n/a	self.assertEqual(len(l), size + 1)
385	n/a	self.assertEqual(set(l), set([_(' ')]))
386	n/a
387	n/a	@bigmemtest(size=_2G, memuse=2.1)
388	n/a	def test_splitlines(self, size):
389	n/a	_ = self.from_latin1
390	n/a	# Crudely calculate an estimate so that the result of s.split won't
391	n/a	# take up an inordinate amount of memory
392	n/a	chunksize = int(size ** 0.5 + 2) // 2
393	n/a	SUBSTR = _(' ') * chunksize + _('\n') + _(' ') * chunksize + _('\r\n')
394	n/a	s = SUBSTR * (chunksize * 2)
395	n/a	l = s.splitlines()
396	n/a	self.assertEqual(len(l), chunksize * 4)
397	n/a	expected = _(' ') * chunksize
398	n/a	for item in l:
399	n/a	self.assertEqual(item, expected)
400	n/a
401	n/a	@bigmemtest(size=_2G, memuse=2)
402	n/a	def test_startswith(self, size):
403	n/a	_ = self.from_latin1
404	n/a	SUBSTR = _(' abc def ghi')
405	n/a	s = _('-') * size + SUBSTR
406	n/a	self.assertTrue(s.startswith(s))
407	n/a	self.assertTrue(s.startswith(_('-') * size))
408	n/a	self.assertFalse(s.startswith(SUBSTR))
409	n/a
410	n/a	@bigmemtest(size=_2G, memuse=1)
411	n/a	def test_strip(self, size):
412	n/a	_ = self.from_latin1
413	n/a	SUBSTR = _(' abc def ghi ')
414	n/a	s = SUBSTR.rjust(size)
415	n/a	self.assertEqual(len(s), size)
416	n/a	self.assertEqual(s.strip(), SUBSTR.strip())
417	n/a	del s
418	n/a	s = SUBSTR.ljust(size)
419	n/a	self.assertEqual(len(s), size)
420	n/a	self.assertEqual(s.strip(), SUBSTR.strip())
421	n/a
422	n/a	def _test_swapcase(self, size):
423	n/a	_ = self.from_latin1
424	n/a	SUBSTR = _("aBcDeFG12.'\xa9\x00")
425	n/a	sublen = len(SUBSTR)
426	n/a	repeats = size // sublen + 2
427	n/a	s = SUBSTR * repeats
428	n/a	s = s.swapcase()
429	n/a	self.assertEqual(len(s), sublen * repeats)
430	n/a	self.assertEqual(s[:sublen * 3], SUBSTR.swapcase() * 3)
431	n/a	self.assertEqual(s[-sublen * 3:], SUBSTR.swapcase() * 3)
432	n/a
433	n/a	def _test_title(self, size):
434	n/a	_ = self.from_latin1
435	n/a	SUBSTR = _('SpaaHAaaAaham')
436	n/a	s = SUBSTR * (size // len(SUBSTR) + 2)
437	n/a	s = s.title()
438	n/a	self.assertTrue(s.startswith((SUBSTR * 3).title()))
439	n/a	self.assertTrue(s.endswith(SUBSTR.lower() * 3))
440	n/a
441	n/a	@bigmemtest(size=_2G, memuse=2)
442	n/a	def test_translate(self, size):
443	n/a	_ = self.from_latin1
444	n/a	SUBSTR = _('aZz.z.Aaz.')
445	n/a	trans = bytes.maketrans(b'.aZ', b'-!$')
446	n/a	sublen = len(SUBSTR)
447	n/a	repeats = size // sublen + 2
448	n/a	s = SUBSTR * repeats
449	n/a	s = s.translate(trans)
450	n/a	self.assertEqual(len(s), repeats * sublen)
451	n/a	self.assertEqual(s[:sublen], SUBSTR.translate(trans))
452	n/a	self.assertEqual(s[-sublen:], SUBSTR.translate(trans))
453	n/a	self.assertEqual(s.count(_('.')), 0)
454	n/a	self.assertEqual(s.count(_('!')), repeats * 2)
455	n/a	self.assertEqual(s.count(_('z')), repeats * 3)
456	n/a
457	n/a	@bigmemtest(size=_2G + 5, memuse=2)
458	n/a	def test_upper(self, size):
459	n/a	_ = self.from_latin1
460	n/a	s = _('a') * size
461	n/a	s = s.upper()
462	n/a	self.assertEqual(len(s), size)
463	n/a	self.assertEqual(s.count(_('A')), size)
464	n/a
465	n/a	@bigmemtest(size=_2G + 20, memuse=1)
466	n/a	def test_zfill(self, size):
467	n/a	_ = self.from_latin1
468	n/a	SUBSTR = _('-568324723598234')
469	n/a	s = SUBSTR.zfill(size)
470	n/a	self.assertTrue(s.endswith(_('0') + SUBSTR[1:]))
471	n/a	self.assertTrue(s.startswith(_('-0')))
472	n/a	self.assertEqual(len(s), size)
473	n/a	self.assertEqual(s.count(_('0')), size - len(SUBSTR))
474	n/a
475	n/a	# This test is meaningful even with size < 2G, as long as the
476	n/a	# doubled string is > 2G (but it tests more if both are > 2G :)
477	n/a	@bigmemtest(size=_1G + 2, memuse=3)
478	n/a	def test_concat(self, size):
479	n/a	_ = self.from_latin1
480	n/a	s = _('.') * size
481	n/a	self.assertEqual(len(s), size)
482	n/a	s = s + s
483	n/a	self.assertEqual(len(s), size * 2)
484	n/a	self.assertEqual(s.count(_('.')), size * 2)
485	n/a
486	n/a	# This test is meaningful even with size < 2G, as long as the
487	n/a	# repeated string is > 2G (but it tests more if both are > 2G :)
488	n/a	@bigmemtest(size=_1G + 2, memuse=3)
489	n/a	def test_repeat(self, size):
490	n/a	_ = self.from_latin1
491	n/a	s = _('.') * size
492	n/a	self.assertEqual(len(s), size)
493	n/a	s = s * 2
494	n/a	self.assertEqual(len(s), size * 2)
495	n/a	self.assertEqual(s.count(_('.')), size * 2)
496	n/a
497	n/a	@bigmemtest(size=_2G + 20, memuse=2)
498	n/a	def test_slice_and_getitem(self, size):
499	n/a	_ = self.from_latin1
500	n/a	SUBSTR = _('0123456789')
501	n/a	sublen = len(SUBSTR)
502	n/a	s = SUBSTR * (size // sublen)
503	n/a	stepsize = len(s) // 100
504	n/a	stepsize = stepsize - (stepsize % sublen)
505	n/a	for i in range(0, len(s) - stepsize, stepsize):
506	n/a	self.assertEqual(s[i], SUBSTR[0])
507	n/a	self.assertEqual(s[i:i + sublen], SUBSTR)
508	n/a	self.assertEqual(s[i:i + sublen:2], SUBSTR[::2])
509	n/a	if i > 0:
510	n/a	self.assertEqual(s[i + sublen - 1:i - 1:-3],
511	n/a	SUBSTR[sublen::-3])
512	n/a	# Make sure we do some slicing and indexing near the end of the
513	n/a	# string, too.
514	n/a	self.assertEqual(s[len(s) - 1], SUBSTR[-1])
515	n/a	self.assertEqual(s[-1], SUBSTR[-1])
516	n/a	self.assertEqual(s[len(s) - 10], SUBSTR[0])
517	n/a	self.assertEqual(s[-sublen], SUBSTR[0])
518	n/a	self.assertEqual(s[len(s):], _(''))
519	n/a	self.assertEqual(s[len(s) - 1:], SUBSTR[-1:])
520	n/a	self.assertEqual(s[-1:], SUBSTR[-1:])
521	n/a	self.assertEqual(s[len(s) - sublen:], SUBSTR)
522	n/a	self.assertEqual(s[-sublen:], SUBSTR)
523	n/a	self.assertEqual(len(s[:]), len(s))
524	n/a	self.assertEqual(len(s[:len(s) - 5]), len(s) - 5)
525	n/a	self.assertEqual(len(s[5:-5]), len(s) - 10)
526	n/a
527	n/a	self.assertRaises(IndexError, operator.getitem, s, len(s))
528	n/a	self.assertRaises(IndexError, operator.getitem, s, len(s) + 1)
529	n/a	self.assertRaises(IndexError, operator.getitem, s, len(s) + 1<<31)
530	n/a
531	n/a	@bigmemtest(size=_2G, memuse=2)
532	n/a	def test_contains(self, size):
533	n/a	_ = self.from_latin1
534	n/a	SUBSTR = _('0123456789')
535	n/a	edge = _('-') * (size // 2)
536	n/a	s = _('').join([edge, SUBSTR, edge])
537	n/a	del edge
538	n/a	self.assertTrue(SUBSTR in s)
539	n/a	self.assertFalse(SUBSTR * 2 in s)
540	n/a	self.assertTrue(_('-') in s)
541	n/a	self.assertFalse(_('a') in s)
542	n/a	s += _('a')
543	n/a	self.assertTrue(_('a') in s)
544	n/a
545	n/a	@bigmemtest(size=_2G + 10, memuse=2)
546	n/a	def test_compare(self, size):
547	n/a	_ = self.from_latin1
548	n/a	s1 = _('-') * size
549	n/a	s2 = _('-') * size
550	n/a	self.assertTrue(s1 == s2)
551	n/a	del s2
552	n/a	s2 = s1 + _('a')
553	n/a	self.assertFalse(s1 == s2)
554	n/a	del s2
555	n/a	s2 = _('.') * size
556	n/a	self.assertFalse(s1 == s2)
557	n/a
558	n/a	@bigmemtest(size=_2G + 10, memuse=1)
559	n/a	def test_hash(self, size):
560	n/a	# Not sure if we can do any meaningful tests here... Even if we
561	n/a	# start relying on the exact algorithm used, the result will be
562	n/a	# different depending on the size of the C 'long int'. Even this
563	n/a	# test is dodgy (there's no guarantee that the two things should
564	n/a	# have a different hash, even if they, in the current
565	n/a	# implementation, almost always do.)
566	n/a	_ = self.from_latin1
567	n/a	s = _('\x00') * size
568	n/a	h1 = hash(s)
569	n/a	del s
570	n/a	s = _('\x00') * (size + 1)
571	n/a	self.assertNotEqual(h1, hash(s))
572	n/a
573	n/a
574	n/a	class StrTest(unittest.TestCase, BaseStrTest):
575	n/a
576	n/a	def from_latin1(self, s):
577	n/a	return s
578	n/a
579	n/a	def basic_encode_test(self, size, enc, c='.', expectedsize=None):
580	n/a	if expectedsize is None:
581	n/a	expectedsize = size
582	n/a	try:
583	n/a	s = c * size
584	n/a	self.assertEqual(len(s.encode(enc)), expectedsize)
585	n/a	finally:
586	n/a	s = None
587	n/a
588	n/a	def setUp(self):
589	n/a	# HACK: adjust memory use of tests inherited from BaseStrTest
590	n/a	# according to character size.
591	n/a	self._adjusted = {}
592	n/a	for name in dir(BaseStrTest):
593	n/a	if not name.startswith('test_'):
594	n/a	continue
595	n/a	meth = getattr(type(self), name)
596	n/a	try:
597	n/a	memuse = meth.memuse
598	n/a	except AttributeError:
599	n/a	continue
600	n/a	meth.memuse = ascii_char_size * memuse
601	n/a	self._adjusted[name] = memuse
602	n/a
603	n/a	def tearDown(self):
604	n/a	for name, memuse in self._adjusted.items():
605	n/a	getattr(type(self), name).memuse = memuse
606	n/a
607	n/a	@bigmemtest(size=_2G, memuse=ucs4_char_size * 3)
608	n/a	def test_capitalize(self, size):
609	n/a	self._test_capitalize(size)
610	n/a
611	n/a	@bigmemtest(size=_2G, memuse=ucs4_char_size * 3)
612	n/a	def test_title(self, size):
613	n/a	self._test_title(size)
614	n/a
615	n/a	@bigmemtest(size=_2G, memuse=ucs4_char_size * 3)
616	n/a	def test_swapcase(self, size):
617	n/a	self._test_swapcase(size)
618	n/a
619	n/a	# Many codecs convert to the legacy representation first, explaining
620	n/a	# why we add 'ucs4_char_size' to the 'memuse' below.
621	n/a
622	n/a	@bigmemtest(size=_2G + 2, memuse=ascii_char_size + 1)
623	n/a	def test_encode(self, size):
624	n/a	return self.basic_encode_test(size, 'utf-8')
625	n/a
626	n/a	@bigmemtest(size=_4G // 6 + 2, memuse=ascii_char_size + ucs4_char_size + 1)
627	n/a	def test_encode_raw_unicode_escape(self, size):
628	n/a	try:
629	n/a	return self.basic_encode_test(size, 'raw_unicode_escape')
630	n/a	except MemoryError:
631	n/a	pass # acceptable on 32-bit
632	n/a
633	n/a	@bigmemtest(size=_4G // 5 + 70, memuse=ascii_char_size + ucs4_char_size + 1)
634	n/a	def test_encode_utf7(self, size):
635	n/a	try:
636	n/a	return self.basic_encode_test(size, 'utf7')
637	n/a	except MemoryError:
638	n/a	pass # acceptable on 32-bit
639	n/a
640	n/a	@bigmemtest(size=_4G // 4 + 5, memuse=ascii_char_size + ucs4_char_size + 4)
641	n/a	def test_encode_utf32(self, size):
642	n/a	try:
643	n/a	return self.basic_encode_test(size, 'utf32', expectedsize=4 * size + 4)
644	n/a	except MemoryError:
645	n/a	pass # acceptable on 32-bit
646	n/a
647	n/a	@bigmemtest(size=_2G - 1, memuse=ascii_char_size + 1)
648	n/a	def test_encode_ascii(self, size):
649	n/a	return self.basic_encode_test(size, 'ascii', c='A')
650	n/a
651	n/a	# str % (...) uses a Py_UCS4 intermediate representation
652	n/a
653	n/a	@bigmemtest(size=_2G + 10, memuse=ascii_char_size * 2 + ucs4_char_size)
654	n/a	def test_format(self, size):
655	n/a	s = '-' * size
656	n/a	sf = '%s' % (s,)
657	n/a	self.assertTrue(s == sf)
658	n/a	del sf
659	n/a	sf = '..%s..' % (s,)
660	n/a	self.assertEqual(len(sf), len(s) + 4)
661	n/a	self.assertTrue(sf.startswith('..-'))
662	n/a	self.assertTrue(sf.endswith('-..'))
663	n/a	del s, sf
664	n/a
665	n/a	size //= 2
666	n/a	edge = '-' * size
667	n/a	s = ''.join([edge, '%s', edge])
668	n/a	del edge
669	n/a	s = s % '...'
670	n/a	self.assertEqual(len(s), size * 2 + 3)
671	n/a	self.assertEqual(s.count('.'), 3)
672	n/a	self.assertEqual(s.count('-'), size * 2)
673	n/a
674	n/a	@bigmemtest(size=_2G + 10, memuse=ascii_char_size * 2)
675	n/a	def test_repr_small(self, size):
676	n/a	s = '-' * size
677	n/a	s = repr(s)
678	n/a	self.assertEqual(len(s), size + 2)
679	n/a	self.assertEqual(s[0], "'")
680	n/a	self.assertEqual(s[-1], "'")
681	n/a	self.assertEqual(s.count('-'), size)
682	n/a	del s
683	n/a	# repr() will create a string four times as large as this 'binary
684	n/a	# string', but we don't want to allocate much more than twice
685	n/a	# size in total. (We do extra testing in test_repr_large())
686	n/a	size = size // 5 * 2
687	n/a	s = '\x00' * size
688	n/a	s = repr(s)
689	n/a	self.assertEqual(len(s), size * 4 + 2)
690	n/a	self.assertEqual(s[0], "'")
691	n/a	self.assertEqual(s[-1], "'")
692	n/a	self.assertEqual(s.count('\\'), size)
693	n/a	self.assertEqual(s.count('0'), size * 2)
694	n/a
695	n/a	@bigmemtest(size=_2G + 10, memuse=ascii_char_size * 5)
696	n/a	def test_repr_large(self, size):
697	n/a	s = '\x00' * size
698	n/a	s = repr(s)
699	n/a	self.assertEqual(len(s), size * 4 + 2)
700	n/a	self.assertEqual(s[0], "'")
701	n/a	self.assertEqual(s[-1], "'")
702	n/a	self.assertEqual(s.count('\\'), size)
703	n/a	self.assertEqual(s.count('0'), size * 2)
704	n/a
705	n/a	# ascii() calls encode('ascii', 'backslashreplace'), which itself
706	n/a	# creates a temporary Py_UNICODE representation in addition to the
707	n/a	# original (Py_UCS2) one
708	n/a	# There's also some overallocation when resizing the ascii() result
709	n/a	# that isn't taken into account here.
710	n/a	@bigmemtest(size=_2G // 5 + 1, memuse=ucs2_char_size +
711	n/a	ucs4_char_size + ascii_char_size * 6)
712	n/a	def test_unicode_repr(self, size):
713	n/a	# Use an assigned, but not printable code point.
714	n/a	# It is in the range of the low surrogates \uDC00-\uDFFF.
715	n/a	char = "\uDCBA"
716	n/a	s = char * size
717	n/a	try:
718	n/a	for f in (repr, ascii):
719	n/a	r = f(s)
720	n/a	self.assertEqual(len(r), 2 + (len(f(char)) - 2) * size)
721	n/a	self.assertTrue(r.endswith(r"\udcba'"), r[-10:])
722	n/a	r = None
723	n/a	finally:
724	n/a	r = s = None
725	n/a
726	n/a	@bigmemtest(size=_2G // 5 + 1, memuse=ucs4_char_size * 2 + ascii_char_size * 10)
727	n/a	def test_unicode_repr_wide(self, size):
728	n/a	char = "\U0001DCBA"
729	n/a	s = char * size
730	n/a	try:
731	n/a	for f in (repr, ascii):
732	n/a	r = f(s)
733	n/a	self.assertEqual(len(r), 2 + (len(f(char)) - 2) * size)
734	n/a	self.assertTrue(r.endswith(r"\U0001dcba'"), r[-12:])
735	n/a	r = None
736	n/a	finally:
737	n/a	r = s = None
738	n/a
739	n/a	# The original test_translate is overridden here, so as to get the
740	n/a	# correct size estimate: str.translate() uses an intermediate Py_UCS4
741	n/a	# representation.
742	n/a
743	n/a	@bigmemtest(size=_2G, memuse=ascii_char_size * 2 + ucs4_char_size)
744	n/a	def test_translate(self, size):
745	n/a	_ = self.from_latin1
746	n/a	SUBSTR = _('aZz.z.Aaz.')
747	n/a	trans = {
748	n/a	ord(_('.')): _('-'),
749	n/a	ord(_('a')): _('!'),
750	n/a	ord(_('Z')): _('$'),
751	n/a	}
752	n/a	sublen = len(SUBSTR)
753	n/a	repeats = size // sublen + 2
754	n/a	s = SUBSTR * repeats
755	n/a	s = s.translate(trans)
756	n/a	self.assertEqual(len(s), repeats * sublen)
757	n/a	self.assertEqual(s[:sublen], SUBSTR.translate(trans))
758	n/a	self.assertEqual(s[-sublen:], SUBSTR.translate(trans))
759	n/a	self.assertEqual(s.count(_('.')), 0)
760	n/a	self.assertEqual(s.count(_('!')), repeats * 2)
761	n/a	self.assertEqual(s.count(_('z')), repeats * 3)
762	n/a
763	n/a
764	n/a	class BytesTest(unittest.TestCase, BaseStrTest):
765	n/a
766	n/a	def from_latin1(self, s):
767	n/a	return s.encode("latin-1")
768	n/a
769	n/a	@bigmemtest(size=_2G + 2, memuse=1 + ascii_char_size)
770	n/a	def test_decode(self, size):
771	n/a	s = self.from_latin1('.') * size
772	n/a	self.assertEqual(len(s.decode('utf-8')), size)
773	n/a
774	n/a	@bigmemtest(size=_2G, memuse=2)
775	n/a	def test_capitalize(self, size):
776	n/a	self._test_capitalize(size)
777	n/a
778	n/a	@bigmemtest(size=_2G, memuse=2)
779	n/a	def test_title(self, size):
780	n/a	self._test_title(size)
781	n/a
782	n/a	@bigmemtest(size=_2G, memuse=2)
783	n/a	def test_swapcase(self, size):
784	n/a	self._test_swapcase(size)
785	n/a
786	n/a
787	n/a	class BytearrayTest(unittest.TestCase, BaseStrTest):
788	n/a
789	n/a	def from_latin1(self, s):
790	n/a	return bytearray(s.encode("latin-1"))
791	n/a
792	n/a	@bigmemtest(size=_2G + 2, memuse=1 + ascii_char_size)
793	n/a	def test_decode(self, size):
794	n/a	s = self.from_latin1('.') * size
795	n/a	self.assertEqual(len(s.decode('utf-8')), size)
796	n/a
797	n/a	@bigmemtest(size=_2G, memuse=2)
798	n/a	def test_capitalize(self, size):
799	n/a	self._test_capitalize(size)
800	n/a
801	n/a	@bigmemtest(size=_2G, memuse=2)
802	n/a	def test_title(self, size):
803	n/a	self._test_title(size)
804	n/a
805	n/a	@bigmemtest(size=_2G, memuse=2)
806	n/a	def test_swapcase(self, size):
807	n/a	self._test_swapcase(size)
808	n/a
809	n/a	test_hash = None
810	n/a	test_split_large = None
811	n/a
812	n/a	class TupleTest(unittest.TestCase):
813	n/a
814	n/a	# Tuples have a small, fixed-sized head and an array of pointers to
815	n/a	# data. Since we're testing 64-bit addressing, we can assume that the
816	n/a	# pointers are 8 bytes, and that thus that the tuples take up 8 bytes
817	n/a	# per size.
818	n/a
819	n/a	# As a side-effect of testing long tuples, these tests happen to test
820	n/a	# having more than 2<<31 references to any given object. Hence the
821	n/a	# use of different types of objects as contents in different tests.
822	n/a
823	n/a	@bigmemtest(size=_2G + 2, memuse=16)
824	n/a	def test_compare(self, size):
825	n/a	t1 = ('',) * size
826	n/a	t2 = ('',) * size
827	n/a	self.assertTrue(t1 == t2)
828	n/a	del t2
829	n/a	t2 = ('',) * (size + 1)
830	n/a	self.assertFalse(t1 == t2)
831	n/a	del t2
832	n/a	t2 = (1,) * size
833	n/a	self.assertFalse(t1 == t2)
834	n/a
835	n/a	# Test concatenating into a single tuple of more than 2G in length,
836	n/a	# and concatenating a tuple of more than 2G in length separately, so
837	n/a	# the smaller test still gets run even if there isn't memory for the
838	n/a	# larger test (but we still let the tester know the larger test is
839	n/a	# skipped, in verbose mode.)
840	n/a	def basic_concat_test(self, size):
841	n/a	t = ((),) * size
842	n/a	self.assertEqual(len(t), size)
843	n/a	t = t + t
844	n/a	self.assertEqual(len(t), size * 2)
845	n/a
846	n/a	@bigmemtest(size=_2G // 2 + 2, memuse=24)
847	n/a	def test_concat_small(self, size):
848	n/a	return self.basic_concat_test(size)
849	n/a
850	n/a	@bigmemtest(size=_2G + 2, memuse=24)
851	n/a	def test_concat_large(self, size):
852	n/a	return self.basic_concat_test(size)
853	n/a
854	n/a	@bigmemtest(size=_2G // 5 + 10, memuse=8 * 5)
855	n/a	def test_contains(self, size):
856	n/a	t = (1, 2, 3, 4, 5) * size
857	n/a	self.assertEqual(len(t), size * 5)
858	n/a	self.assertTrue(5 in t)
859	n/a	self.assertFalse((1, 2, 3, 4, 5) in t)
860	n/a	self.assertFalse(0 in t)
861	n/a
862	n/a	@bigmemtest(size=_2G + 10, memuse=8)
863	n/a	def test_hash(self, size):
864	n/a	t1 = (0,) * size
865	n/a	h1 = hash(t1)
866	n/a	del t1
867	n/a	t2 = (0,) * (size + 1)
868	n/a	self.assertFalse(h1 == hash(t2))
869	n/a
870	n/a	@bigmemtest(size=_2G + 10, memuse=8)
871	n/a	def test_index_and_slice(self, size):
872	n/a	t = (None,) * size
873	n/a	self.assertEqual(len(t), size)
874	n/a	self.assertEqual(t[-1], None)
875	n/a	self.assertEqual(t[5], None)
876	n/a	self.assertEqual(t[size - 1], None)
877	n/a	self.assertRaises(IndexError, operator.getitem, t, size)
878	n/a	self.assertEqual(t[:5], (None,) * 5)
879	n/a	self.assertEqual(t[-5:], (None,) * 5)
880	n/a	self.assertEqual(t[20:25], (None,) * 5)
881	n/a	self.assertEqual(t[-25:-20], (None,) * 5)
882	n/a	self.assertEqual(t[size - 5:], (None,) * 5)
883	n/a	self.assertEqual(t[size - 5:size], (None,) * 5)
884	n/a	self.assertEqual(t[size - 6:size - 2], (None,) * 4)
885	n/a	self.assertEqual(t[size:size], ())
886	n/a	self.assertEqual(t[size:size+5], ())
887	n/a
888	n/a	# Like test_concat, split in two.
889	n/a	def basic_test_repeat(self, size):
890	n/a	t = ('',) * size
891	n/a	self.assertEqual(len(t), size)
892	n/a	t = t * 2
893	n/a	self.assertEqual(len(t), size * 2)
894	n/a
895	n/a	@bigmemtest(size=_2G // 2 + 2, memuse=24)
896	n/a	def test_repeat_small(self, size):
897	n/a	return self.basic_test_repeat(size)
898	n/a
899	n/a	@bigmemtest(size=_2G + 2, memuse=24)
900	n/a	def test_repeat_large(self, size):
901	n/a	return self.basic_test_repeat(size)
902	n/a
903	n/a	@bigmemtest(size=_1G - 1, memuse=12)
904	n/a	def test_repeat_large_2(self, size):
905	n/a	return self.basic_test_repeat(size)
906	n/a
907	n/a	@bigmemtest(size=_1G - 1, memuse=9)
908	n/a	def test_from_2G_generator(self, size):
909	n/a	self.skipTest("test needs much more memory than advertised, see issue5438")
910	n/a	try:
911	n/a	t = tuple(range(size))
912	n/a	except MemoryError:
913	n/a	pass # acceptable on 32-bit
914	n/a	else:
915	n/a	count = 0
916	n/a	for item in t:
917	n/a	self.assertEqual(item, count)
918	n/a	count += 1
919	n/a	self.assertEqual(count, size)
920	n/a
921	n/a	@bigmemtest(size=_1G - 25, memuse=9)
922	n/a	def test_from_almost_2G_generator(self, size):
923	n/a	self.skipTest("test needs much more memory than advertised, see issue5438")
924	n/a	try:
925	n/a	t = tuple(range(size))
926	n/a	count = 0
927	n/a	for item in t:
928	n/a	self.assertEqual(item, count)
929	n/a	count += 1
930	n/a	self.assertEqual(count, size)
931	n/a	except MemoryError:
932	n/a	pass # acceptable, expected on 32-bit
933	n/a
934	n/a	# Like test_concat, split in two.
935	n/a	def basic_test_repr(self, size):
936	n/a	t = (0,) * size
937	n/a	s = repr(t)
938	n/a	# The repr of a tuple of 0's is exactly three times the tuple length.
939	n/a	self.assertEqual(len(s), size * 3)
940	n/a	self.assertEqual(s[:5], '(0, 0')
941	n/a	self.assertEqual(s[-5:], '0, 0)')
942	n/a	self.assertEqual(s.count('0'), size)
943	n/a
944	n/a	@bigmemtest(size=_2G // 3 + 2, memuse=8 + 3 * ascii_char_size)
945	n/a	def test_repr_small(self, size):
946	n/a	return self.basic_test_repr(size)
947	n/a
948	n/a	@bigmemtest(size=_2G + 2, memuse=8 + 3 * ascii_char_size)
949	n/a	def test_repr_large(self, size):
950	n/a	return self.basic_test_repr(size)
951	n/a
952	n/a	class ListTest(unittest.TestCase):
953	n/a
954	n/a	# Like tuples, lists have a small, fixed-sized head and an array of
955	n/a	# pointers to data, so 8 bytes per size. Also like tuples, we make the
956	n/a	# lists hold references to various objects to test their refcount
957	n/a	# limits.
958	n/a
959	n/a	@bigmemtest(size=_2G + 2, memuse=16)
960	n/a	def test_compare(self, size):
961	n/a	l1 = [''] * size
962	n/a	l2 = [''] * size
963	n/a	self.assertTrue(l1 == l2)
964	n/a	del l2
965	n/a	l2 = [''] * (size + 1)
966	n/a	self.assertFalse(l1 == l2)
967	n/a	del l2
968	n/a	l2 = [2] * size
969	n/a	self.assertFalse(l1 == l2)
970	n/a
971	n/a	# Test concatenating into a single list of more than 2G in length,
972	n/a	# and concatenating a list of more than 2G in length separately, so
973	n/a	# the smaller test still gets run even if there isn't memory for the
974	n/a	# larger test (but we still let the tester know the larger test is
975	n/a	# skipped, in verbose mode.)
976	n/a	def basic_test_concat(self, size):
977	n/a	l = [[]] * size
978	n/a	self.assertEqual(len(l), size)
979	n/a	l = l + l
980	n/a	self.assertEqual(len(l), size * 2)
981	n/a
982	n/a	@bigmemtest(size=_2G // 2 + 2, memuse=24)
983	n/a	def test_concat_small(self, size):
984	n/a	return self.basic_test_concat(size)
985	n/a
986	n/a	@bigmemtest(size=_2G + 2, memuse=24)
987	n/a	def test_concat_large(self, size):
988	n/a	return self.basic_test_concat(size)
989	n/a
990	n/a	def basic_test_inplace_concat(self, size):
991	n/a	l = [sys.stdout] * size
992	n/a	l += l
993	n/a	self.assertEqual(len(l), size * 2)
994	n/a	self.assertTrue(l[0] is l[-1])
995	n/a	self.assertTrue(l[size - 1] is l[size + 1])
996	n/a
997	n/a	@bigmemtest(size=_2G // 2 + 2, memuse=24)
998	n/a	def test_inplace_concat_small(self, size):
999	n/a	return self.basic_test_inplace_concat(size)
1000	n/a
1001	n/a	@bigmemtest(size=_2G + 2, memuse=24)
1002	n/a	def test_inplace_concat_large(self, size):
1003	n/a	return self.basic_test_inplace_concat(size)
1004	n/a
1005	n/a	@bigmemtest(size=_2G // 5 + 10, memuse=8 * 5)
1006	n/a	def test_contains(self, size):
1007	n/a	l = [1, 2, 3, 4, 5] * size
1008	n/a	self.assertEqual(len(l), size * 5)
1009	n/a	self.assertTrue(5 in l)
1010	n/a	self.assertFalse([1, 2, 3, 4, 5] in l)
1011	n/a	self.assertFalse(0 in l)
1012	n/a
1013	n/a	@bigmemtest(size=_2G + 10, memuse=8)
1014	n/a	def test_hash(self, size):
1015	n/a	l = [0] * size
1016	n/a	self.assertRaises(TypeError, hash, l)
1017	n/a
1018	n/a	@bigmemtest(size=_2G + 10, memuse=8)
1019	n/a	def test_index_and_slice(self, size):
1020	n/a	l = [None] * size
1021	n/a	self.assertEqual(len(l), size)
1022	n/a	self.assertEqual(l[-1], None)
1023	n/a	self.assertEqual(l[5], None)
1024	n/a	self.assertEqual(l[size - 1], None)
1025	n/a	self.assertRaises(IndexError, operator.getitem, l, size)
1026	n/a	self.assertEqual(l[:5], [None] * 5)
1027	n/a	self.assertEqual(l[-5:], [None] * 5)
1028	n/a	self.assertEqual(l[20:25], [None] * 5)
1029	n/a	self.assertEqual(l[-25:-20], [None] * 5)
1030	n/a	self.assertEqual(l[size - 5:], [None] * 5)
1031	n/a	self.assertEqual(l[size - 5:size], [None] * 5)
1032	n/a	self.assertEqual(l[size - 6:size - 2], [None] * 4)
1033	n/a	self.assertEqual(l[size:size], [])
1034	n/a	self.assertEqual(l[size:size+5], [])
1035	n/a
1036	n/a	l[size - 2] = 5
1037	n/a	self.assertEqual(len(l), size)
1038	n/a	self.assertEqual(l[-3:], [None, 5, None])
1039	n/a	self.assertEqual(l.count(5), 1)
1040	n/a	self.assertRaises(IndexError, operator.setitem, l, size, 6)
1041	n/a	self.assertEqual(len(l), size)
1042	n/a
1043	n/a	l[size - 7:] = [1, 2, 3, 4, 5]
1044	n/a	size -= 2
1045	n/a	self.assertEqual(len(l), size)
1046	n/a	self.assertEqual(l[-7:], [None, None, 1, 2, 3, 4, 5])
1047	n/a
1048	n/a	l[:7] = [1, 2, 3, 4, 5]
1049	n/a	size -= 2
1050	n/a	self.assertEqual(len(l), size)
1051	n/a	self.assertEqual(l[:7], [1, 2, 3, 4, 5, None, None])
1052	n/a
1053	n/a	del l[size - 1]
1054	n/a	size -= 1
1055	n/a	self.assertEqual(len(l), size)
1056	n/a	self.assertEqual(l[-1], 4)
1057	n/a
1058	n/a	del l[-2:]
1059	n/a	size -= 2
1060	n/a	self.assertEqual(len(l), size)
1061	n/a	self.assertEqual(l[-1], 2)
1062	n/a
1063	n/a	del l[0]
1064	n/a	size -= 1
1065	n/a	self.assertEqual(len(l), size)
1066	n/a	self.assertEqual(l[0], 2)
1067	n/a
1068	n/a	del l[:2]
1069	n/a	size -= 2
1070	n/a	self.assertEqual(len(l), size)
1071	n/a	self.assertEqual(l[0], 4)
1072	n/a
1073	n/a	# Like test_concat, split in two.
1074	n/a	def basic_test_repeat(self, size):
1075	n/a	l = [] * size
1076	n/a	self.assertFalse(l)
1077	n/a	l = [''] * size
1078	n/a	self.assertEqual(len(l), size)
1079	n/a	l = l * 2
1080	n/a	self.assertEqual(len(l), size * 2)
1081	n/a
1082	n/a	@bigmemtest(size=_2G // 2 + 2, memuse=24)
1083	n/a	def test_repeat_small(self, size):
1084	n/a	return self.basic_test_repeat(size)
1085	n/a
1086	n/a	@bigmemtest(size=_2G + 2, memuse=24)
1087	n/a	def test_repeat_large(self, size):
1088	n/a	return self.basic_test_repeat(size)
1089	n/a
1090	n/a	def basic_test_inplace_repeat(self, size):
1091	n/a	l = ['']
1092	n/a	l *= size
1093	n/a	self.assertEqual(len(l), size)
1094	n/a	self.assertTrue(l[0] is l[-1])
1095	n/a	del l
1096	n/a
1097	n/a	l = [''] * size
1098	n/a	l *= 2
1099	n/a	self.assertEqual(len(l), size * 2)
1100	n/a	self.assertTrue(l[size - 1] is l[-1])
1101	n/a
1102	n/a	@bigmemtest(size=_2G // 2 + 2, memuse=16)
1103	n/a	def test_inplace_repeat_small(self, size):
1104	n/a	return self.basic_test_inplace_repeat(size)
1105	n/a
1106	n/a	@bigmemtest(size=_2G + 2, memuse=16)
1107	n/a	def test_inplace_repeat_large(self, size):
1108	n/a	return self.basic_test_inplace_repeat(size)
1109	n/a
1110	n/a	def basic_test_repr(self, size):
1111	n/a	l = [0] * size
1112	n/a	s = repr(l)
1113	n/a	# The repr of a list of 0's is exactly three times the list length.
1114	n/a	self.assertEqual(len(s), size * 3)
1115	n/a	self.assertEqual(s[:5], '[0, 0')
1116	n/a	self.assertEqual(s[-5:], '0, 0]')
1117	n/a	self.assertEqual(s.count('0'), size)
1118	n/a
1119	n/a	@bigmemtest(size=_2G // 3 + 2, memuse=8 + 3 * ascii_char_size)
1120	n/a	def test_repr_small(self, size):
1121	n/a	return self.basic_test_repr(size)
1122	n/a
1123	n/a	@bigmemtest(size=_2G + 2, memuse=8 + 3 * ascii_char_size)
1124	n/a	def test_repr_large(self, size):
1125	n/a	return self.basic_test_repr(size)
1126	n/a
1127	n/a	# list overallocates ~1/8th of the total size (on first expansion) so
1128	n/a	# the single list.append call puts memuse at 9 bytes per size.
1129	n/a	@bigmemtest(size=_2G, memuse=9)
1130	n/a	def test_append(self, size):
1131	n/a	l = [object()] * size
1132	n/a	l.append(object())
1133	n/a	self.assertEqual(len(l), size+1)
1134	n/a	self.assertTrue(l[-3] is l[-2])
1135	n/a	self.assertFalse(l[-2] is l[-1])
1136	n/a
1137	n/a	@bigmemtest(size=_2G // 5 + 2, memuse=8 * 5)
1138	n/a	def test_count(self, size):
1139	n/a	l = [1, 2, 3, 4, 5] * size
1140	n/a	self.assertEqual(l.count(1), size)
1141	n/a	self.assertEqual(l.count("1"), 0)
1142	n/a
1143	n/a	def basic_test_extend(self, size):
1144	n/a	l = [object] * size
1145	n/a	l.extend(l)
1146	n/a	self.assertEqual(len(l), size * 2)
1147	n/a	self.assertTrue(l[0] is l[-1])
1148	n/a	self.assertTrue(l[size - 1] is l[size + 1])
1149	n/a
1150	n/a	@bigmemtest(size=_2G // 2 + 2, memuse=16)
1151	n/a	def test_extend_small(self, size):
1152	n/a	return self.basic_test_extend(size)
1153	n/a
1154	n/a	@bigmemtest(size=_2G + 2, memuse=16)
1155	n/a	def test_extend_large(self, size):
1156	n/a	return self.basic_test_extend(size)
1157	n/a
1158	n/a	@bigmemtest(size=_2G // 5 + 2, memuse=8 * 5)
1159	n/a	def test_index(self, size):
1160	n/a	l = [1, 2, 3, 4, 5] * size
1161	n/a	size *= 5
1162	n/a	self.assertEqual(l.index(1), 0)
1163	n/a	self.assertEqual(l.index(5, size - 5), size - 1)
1164	n/a	self.assertEqual(l.index(5, size - 5, size), size - 1)
1165	n/a	self.assertRaises(ValueError, l.index, 1, size - 4, size)
1166	n/a	self.assertRaises(ValueError, l.index, 6)
1167	n/a
1168	n/a	# This tests suffers from overallocation, just like test_append.
1169	n/a	@bigmemtest(size=_2G + 10, memuse=9)
1170	n/a	def test_insert(self, size):
1171	n/a	l = [1.0] * size
1172	n/a	l.insert(size - 1, "A")
1173	n/a	size += 1
1174	n/a	self.assertEqual(len(l), size)
1175	n/a	self.assertEqual(l[-3:], [1.0, "A", 1.0])
1176	n/a
1177	n/a	l.insert(size + 1, "B")
1178	n/a	size += 1
1179	n/a	self.assertEqual(len(l), size)
1180	n/a	self.assertEqual(l[-3:], ["A", 1.0, "B"])
1181	n/a
1182	n/a	l.insert(1, "C")
1183	n/a	size += 1
1184	n/a	self.assertEqual(len(l), size)
1185	n/a	self.assertEqual(l[:3], [1.0, "C", 1.0])
1186	n/a	self.assertEqual(l[size - 3:], ["A", 1.0, "B"])
1187	n/a
1188	n/a	@bigmemtest(size=_2G // 5 + 4, memuse=8 * 5)
1189	n/a	def test_pop(self, size):
1190	n/a	l = ["a", "b", "c", "d", "e"] * size
1191	n/a	size *= 5
1192	n/a	self.assertEqual(len(l), size)
1193	n/a
1194	n/a	item = l.pop()
1195	n/a	size -= 1
1196	n/a	self.assertEqual(len(l), size)
1197	n/a	self.assertEqual(item, "e")
1198	n/a	self.assertEqual(l[-2:], ["c", "d"])
1199	n/a
1200	n/a	item = l.pop(0)
1201	n/a	size -= 1
1202	n/a	self.assertEqual(len(l), size)
1203	n/a	self.assertEqual(item, "a")
1204	n/a	self.assertEqual(l[:2], ["b", "c"])
1205	n/a
1206	n/a	item = l.pop(size - 2)
1207	n/a	size -= 1
1208	n/a	self.assertEqual(len(l), size)
1209	n/a	self.assertEqual(item, "c")
1210	n/a	self.assertEqual(l[-2:], ["b", "d"])
1211	n/a
1212	n/a	@bigmemtest(size=_2G + 10, memuse=8)
1213	n/a	def test_remove(self, size):
1214	n/a	l = [10] * size
1215	n/a	self.assertEqual(len(l), size)
1216	n/a
1217	n/a	l.remove(10)
1218	n/a	size -= 1
1219	n/a	self.assertEqual(len(l), size)
1220	n/a
1221	n/a	# Because of the earlier l.remove(), this append doesn't trigger
1222	n/a	# a resize.
1223	n/a	l.append(5)
1224	n/a	size += 1
1225	n/a	self.assertEqual(len(l), size)
1226	n/a	self.assertEqual(l[-2:], [10, 5])
1227	n/a	l.remove(5)
1228	n/a	size -= 1
1229	n/a	self.assertEqual(len(l), size)
1230	n/a	self.assertEqual(l[-2:], [10, 10])
1231	n/a
1232	n/a	@bigmemtest(size=_2G // 5 + 2, memuse=8 * 5)
1233	n/a	def test_reverse(self, size):
1234	n/a	l = [1, 2, 3, 4, 5] * size
1235	n/a	l.reverse()
1236	n/a	self.assertEqual(len(l), size * 5)
1237	n/a	self.assertEqual(l[-5:], [5, 4, 3, 2, 1])
1238	n/a	self.assertEqual(l[:5], [5, 4, 3, 2, 1])
1239	n/a
1240	n/a	@bigmemtest(size=_2G // 5 + 2, memuse=8 * 5)
1241	n/a	def test_sort(self, size):
1242	n/a	l = [1, 2, 3, 4, 5] * size
1243	n/a	l.sort()
1244	n/a	self.assertEqual(len(l), size * 5)
1245	n/a	self.assertEqual(l.count(1), size)
1246	n/a	self.assertEqual(l[:10], [1] * 10)
1247	n/a	self.assertEqual(l[-10:], [5] * 10)
1248	n/a
1249	n/a	def test_main():
1250	n/a	support.run_unittest(StrTest, BytesTest, BytearrayTest,
1251	n/a	TupleTest, ListTest)
1252	n/a
1253	n/a	if __name__ == '__main__':
1254	n/a	if len(sys.argv) > 1:
1255	n/a	support.set_memlimit(sys.argv[1])
1256	n/a	test_main()