blob: aaa2dc20a92c79d98f1427648375f09fd92b462e [file] [log] [blame]
Walter Dörwald0fd583c2003-02-21 12:53:50 +00001"""
2Common tests shared by test_str, test_unicode, test_userstring and test_string.
3"""
Jeremy Hyltonf82b04e2000-07-10 17:08:42 +00004
Walter Dörwald0fd583c2003-02-21 12:53:50 +00005import unittest, string, sys
6from test import test_support
Jeremy Hylton20f41b62000-07-11 03:31:55 +00007from UserList import UserList
8
Jeremy Hyltonf82b04e2000-07-10 17:08:42 +00009class Sequence:
Walter Dörwald0fd583c2003-02-21 12:53:50 +000010 def __init__(self, seq='wxyz'): self.seq = seq
Jeremy Hyltonf82b04e2000-07-10 17:08:42 +000011 def __len__(self): return len(self.seq)
12 def __getitem__(self, i): return self.seq[i]
13
14class BadSeq1(Sequence):
15 def __init__(self): self.seq = [7, 'hello', 123L]
16
17class BadSeq2(Sequence):
18 def __init__(self): self.seq = ['a', 'b', 'c']
19 def __len__(self): return 8
20
Walter Dörwald0fd583c2003-02-21 12:53:50 +000021class CommonTest(unittest.TestCase):
22 # This testcase contains test that can be used in all
23 # stringlike classes. Currently this is str, unicode
24 # UserString and the string module.
Jeremy Hyltonf82b04e2000-07-10 17:08:42 +000025
Walter Dörwald0fd583c2003-02-21 12:53:50 +000026 # The type to be tested
27 # Change in subclasses to change the behaviour of fixtesttype()
28 type2test = None
Jeremy Hyltonf82b04e2000-07-10 17:08:42 +000029
Walter Dörwald0fd583c2003-02-21 12:53:50 +000030 # All tests pass their arguments to the testing methods
31 # as str objects. fixtesttype() can be used to propagate
32 # these arguments to the appropriate type
33 def fixtype(self, obj):
34 if isinstance(obj, str):
35 return self.__class__.type2test(obj)
36 elif isinstance(obj, list):
37 return [self.fixtype(x) for x in obj]
38 elif isinstance(obj, tuple):
39 return tuple([self.fixtype(x) for x in obj])
40 elif isinstance(obj, dict):
41 return dict([
42 (self.fixtype(key), self.fixtype(value))
43 for (key, value) in obj.iteritems()
44 ])
45 else:
46 return obj
Jeremy Hyltonf82b04e2000-07-10 17:08:42 +000047
Walter Dörwald0fd583c2003-02-21 12:53:50 +000048 # check that object.method(*args) returns result
49 def checkequal(self, result, object, methodname, *args):
50 result = self.fixtype(result)
51 object = self.fixtype(object)
52 args = self.fixtype(args)
53 realresult = getattr(object, methodname)(*args)
54 self.assertEqual(
55 result,
56 realresult
57 )
58 # if the original is returned make sure that
59 # this doesn't happen with subclasses
60 if object == realresult:
61 class subtype(self.__class__.type2test):
62 pass
63 object = subtype(object)
64 realresult = getattr(object, methodname)(*args)
65 self.assert_(object is not realresult)
Jeremy Hyltonf82b04e2000-07-10 17:08:42 +000066
Walter Dörwald0fd583c2003-02-21 12:53:50 +000067 # check that object.method(*args) raises exc
68 def checkraises(self, exc, object, methodname, *args):
69 object = self.fixtype(object)
70 args = self.fixtype(args)
71 self.assertRaises(
72 exc,
73 getattr(object, methodname),
74 *args
75 )
Jeremy Hyltonf82b04e2000-07-10 17:08:42 +000076
Walter Dörwald0fd583c2003-02-21 12:53:50 +000077 # call object.method(*args) without any checks
78 def checkcall(self, object, methodname, *args):
79 object = self.fixtype(object)
80 args = self.fixtype(args)
81 getattr(object, methodname)(*args)
82
Raymond Hettinger561fbf12004-10-26 01:52:37 +000083 def test_hash(self):
84 # SF bug 1054139: += optimization was not invalidating cached hash value
85 a = self.type2test('DNSSEC')
86 b = self.type2test('')
87 for c in a:
88 b += c
89 hash(b)
90 self.assertEqual(hash(a), hash(b))
91
Walter Dörwald0fd583c2003-02-21 12:53:50 +000092 def test_capitalize(self):
93 self.checkequal(' hello ', ' hello ', 'capitalize')
94 self.checkequal('Hello ', 'Hello ','capitalize')
95 self.checkequal('Hello ', 'hello ','capitalize')
96 self.checkequal('Aaaa', 'aaaa', 'capitalize')
97 self.checkequal('Aaaa', 'AaAa', 'capitalize')
98
99 self.checkraises(TypeError, 'hello', 'capitalize', 42)
100
101 def test_count(self):
102 self.checkequal(3, 'aaa', 'count', 'a')
103 self.checkequal(0, 'aaa', 'count', 'b')
104 self.checkequal(3, 'aaa', 'count', 'a')
105 self.checkequal(0, 'aaa', 'count', 'b')
106 self.checkequal(3, 'aaa', 'count', 'a')
107 self.checkequal(0, 'aaa', 'count', 'b')
108 self.checkequal(0, 'aaa', 'count', 'b')
Thomas Wouters4d70c3d2006-06-08 14:42:34 +0000109 self.checkequal(2, 'aaa', 'count', 'a', 1)
110 self.checkequal(0, 'aaa', 'count', 'a', 10)
Walter Dörwald0fd583c2003-02-21 12:53:50 +0000111 self.checkequal(1, 'aaa', 'count', 'a', -1)
112 self.checkequal(3, 'aaa', 'count', 'a', -10)
Thomas Wouters4d70c3d2006-06-08 14:42:34 +0000113 self.checkequal(1, 'aaa', 'count', 'a', 0, 1)
114 self.checkequal(3, 'aaa', 'count', 'a', 0, 10)
Walter Dörwald0fd583c2003-02-21 12:53:50 +0000115 self.checkequal(2, 'aaa', 'count', 'a', 0, -1)
116 self.checkequal(0, 'aaa', 'count', 'a', 0, -10)
Thomas Wouters4d70c3d2006-06-08 14:42:34 +0000117 self.checkequal(3, 'aaa', 'count', '', 1)
118 self.checkequal(1, 'aaa', 'count', '', 3)
119 self.checkequal(0, 'aaa', 'count', '', 10)
120 self.checkequal(2, 'aaa', 'count', '', -1)
121 self.checkequal(4, 'aaa', 'count', '', -10)
Walter Dörwald0fd583c2003-02-21 12:53:50 +0000122
123 self.checkraises(TypeError, 'hello', 'count')
124 self.checkraises(TypeError, 'hello', 'count', 42)
125
Raymond Hettinger57e74472005-02-20 09:54:53 +0000126 # For a variety of combinations,
127 # verify that str.count() matches an equivalent function
128 # replacing all occurrences and then differencing the string lengths
129 charset = ['', 'a', 'b']
130 digits = 7
131 base = len(charset)
132 teststrings = set()
133 for i in xrange(base ** digits):
134 entry = []
135 for j in xrange(digits):
136 i, m = divmod(i, base)
137 entry.append(charset[m])
138 teststrings.add(''.join(entry))
139 teststrings = list(teststrings)
140 for i in teststrings:
141 i = self.fixtype(i)
142 n = len(i)
143 for j in teststrings:
144 r1 = i.count(j)
145 if j:
146 r2, rem = divmod(n - len(i.replace(j, '')), len(j))
147 else:
148 r2, rem = len(i)+1, 0
149 if rem or r1 != r2:
150 self.assertEqual(rem, 0)
151 self.assertEqual(r1, r2)
152
Walter Dörwald0fd583c2003-02-21 12:53:50 +0000153 def test_find(self):
154 self.checkequal(0, 'abcdefghiabc', 'find', 'abc')
155 self.checkequal(9, 'abcdefghiabc', 'find', 'abc', 1)
156 self.checkequal(-1, 'abcdefghiabc', 'find', 'def', 4)
157
Thomas Wouters4d70c3d2006-06-08 14:42:34 +0000158 self.checkequal(0, 'abc', 'find', '', 0)
159 self.checkequal(3, 'abc', 'find', '', 3)
160 self.checkequal(-1, 'abc', 'find', '', 4)
161
Walter Dörwald0fd583c2003-02-21 12:53:50 +0000162 self.checkraises(TypeError, 'hello', 'find')
163 self.checkraises(TypeError, 'hello', 'find', 42)
164
Raymond Hettinger7cbf1bc2005-02-20 04:07:08 +0000165 # For a variety of combinations,
166 # verify that str.find() matches __contains__
167 # and that the found substring is really at that location
168 charset = ['', 'a', 'b', 'c']
169 digits = 5
170 base = len(charset)
171 teststrings = set()
172 for i in xrange(base ** digits):
173 entry = []
174 for j in xrange(digits):
175 i, m = divmod(i, base)
176 entry.append(charset[m])
177 teststrings.add(''.join(entry))
Raymond Hettinger57e74472005-02-20 09:54:53 +0000178 teststrings = list(teststrings)
Raymond Hettinger7cbf1bc2005-02-20 04:07:08 +0000179 for i in teststrings:
180 i = self.fixtype(i)
181 for j in teststrings:
182 loc = i.find(j)
183 r1 = (loc != -1)
184 r2 = j in i
185 if r1 != r2:
186 self.assertEqual(r1, r2)
187 if loc != -1:
188 self.assertEqual(i[loc:loc+len(j)], j)
189
Walter Dörwald0fd583c2003-02-21 12:53:50 +0000190 def test_rfind(self):
191 self.checkequal(9, 'abcdefghiabc', 'rfind', 'abc')
192 self.checkequal(12, 'abcdefghiabc', 'rfind', '')
193 self.checkequal(0, 'abcdefghiabc', 'rfind', 'abcd')
194 self.checkequal(-1, 'abcdefghiabc', 'rfind', 'abcz')
195
Thomas Wouters4d70c3d2006-06-08 14:42:34 +0000196 self.checkequal(3, 'abc', 'rfind', '', 0)
197 self.checkequal(3, 'abc', 'rfind', '', 3)
198 self.checkequal(-1, 'abc', 'rfind', '', 4)
199
Walter Dörwald0fd583c2003-02-21 12:53:50 +0000200 self.checkraises(TypeError, 'hello', 'rfind')
201 self.checkraises(TypeError, 'hello', 'rfind', 42)
202
203 def test_index(self):
204 self.checkequal(0, 'abcdefghiabc', 'index', '')
205 self.checkequal(3, 'abcdefghiabc', 'index', 'def')
206 self.checkequal(0, 'abcdefghiabc', 'index', 'abc')
207 self.checkequal(9, 'abcdefghiabc', 'index', 'abc', 1)
208
209 self.checkraises(ValueError, 'abcdefghiabc', 'index', 'hib')
210 self.checkraises(ValueError, 'abcdefghiab', 'index', 'abc', 1)
211 self.checkraises(ValueError, 'abcdefghi', 'index', 'ghi', 8)
212 self.checkraises(ValueError, 'abcdefghi', 'index', 'ghi', -1)
213
214 self.checkraises(TypeError, 'hello', 'index')
215 self.checkraises(TypeError, 'hello', 'index', 42)
216
217 def test_rindex(self):
218 self.checkequal(12, 'abcdefghiabc', 'rindex', '')
219 self.checkequal(3, 'abcdefghiabc', 'rindex', 'def')
220 self.checkequal(9, 'abcdefghiabc', 'rindex', 'abc')
221 self.checkequal(0, 'abcdefghiabc', 'rindex', 'abc', 0, -1)
222
223 self.checkraises(ValueError, 'abcdefghiabc', 'rindex', 'hib')
224 self.checkraises(ValueError, 'defghiabc', 'rindex', 'def', 1)
225 self.checkraises(ValueError, 'defghiabc', 'rindex', 'abc', 0, -1)
226 self.checkraises(ValueError, 'abcdefghi', 'rindex', 'ghi', 0, 8)
227 self.checkraises(ValueError, 'abcdefghi', 'rindex', 'ghi', 0, -1)
228
229 self.checkraises(TypeError, 'hello', 'rindex')
230 self.checkraises(TypeError, 'hello', 'rindex', 42)
231
232 def test_lower(self):
233 self.checkequal('hello', 'HeLLo', 'lower')
234 self.checkequal('hello', 'hello', 'lower')
235 self.checkraises(TypeError, 'hello', 'lower', 42)
236
237 def test_upper(self):
238 self.checkequal('HELLO', 'HeLLo', 'upper')
239 self.checkequal('HELLO', 'HELLO', 'upper')
240 self.checkraises(TypeError, 'hello', 'upper', 42)
241
242 def test_expandtabs(self):
243 self.checkequal('abc\rab def\ng hi', 'abc\rab\tdef\ng\thi', 'expandtabs')
244 self.checkequal('abc\rab def\ng hi', 'abc\rab\tdef\ng\thi', 'expandtabs', 8)
245 self.checkequal('abc\rab def\ng hi', 'abc\rab\tdef\ng\thi', 'expandtabs', 4)
246 self.checkequal('abc\r\nab def\ng hi', 'abc\r\nab\tdef\ng\thi', 'expandtabs', 4)
247 self.checkequal('abc\rab def\ng hi', 'abc\rab\tdef\ng\thi', 'expandtabs')
248 self.checkequal('abc\rab def\ng hi', 'abc\rab\tdef\ng\thi', 'expandtabs', 8)
249 self.checkequal('abc\r\nab\r\ndef\ng\r\nhi', 'abc\r\nab\r\ndef\ng\r\nhi', 'expandtabs', 4)
250
251 self.checkraises(TypeError, 'hello', 'expandtabs', 42, 42)
252
253 def test_split(self):
254 self.checkequal(['this', 'is', 'the', 'split', 'function'],
255 'this is the split function', 'split')
Hye-Shik Chang75c00ef2004-01-05 00:29:51 +0000256
257 # by whitespace
258 self.checkequal(['a', 'b', 'c', 'd'], 'a b c d ', 'split')
Walter Dörwald0fd583c2003-02-21 12:53:50 +0000259 self.checkequal(['a', 'b c d'], 'a b c d', 'split', None, 1)
260 self.checkequal(['a', 'b', 'c d'], 'a b c d', 'split', None, 2)
261 self.checkequal(['a', 'b', 'c', 'd'], 'a b c d', 'split', None, 3)
262 self.checkequal(['a', 'b', 'c', 'd'], 'a b c d', 'split', None, 4)
Thomas Wouters477c8d52006-05-27 19:21:47 +0000263 self.checkequal(['a', 'b', 'c', 'd'], 'a b c d', 'split', None,
264 sys.maxint-1)
Walter Dörwald0fd583c2003-02-21 12:53:50 +0000265 self.checkequal(['a b c d'], 'a b c d', 'split', None, 0)
Thomas Wouters477c8d52006-05-27 19:21:47 +0000266 self.checkequal(['a b c d'], ' a b c d', 'split', None, 0)
Walter Dörwald0fd583c2003-02-21 12:53:50 +0000267 self.checkequal(['a', 'b', 'c d'], 'a b c d', 'split', None, 2)
Hye-Shik Chang75c00ef2004-01-05 00:29:51 +0000268
Thomas Wouters477c8d52006-05-27 19:21:47 +0000269 self.checkequal([], ' ', 'split')
270 self.checkequal(['a'], ' a ', 'split')
271 self.checkequal(['a', 'b'], ' a b ', 'split')
272 self.checkequal(['a', 'b '], ' a b ', 'split', None, 1)
273 self.checkequal(['a', 'b c '], ' a b c ', 'split', None, 1)
274 self.checkequal(['a', 'b', 'c '], ' a b c ', 'split', None, 2)
275 self.checkequal(['a', 'b'], '\n\ta \t\r b \v ', 'split')
276 aaa = ' a '*20
277 self.checkequal(['a']*20, aaa, 'split')
278 self.checkequal(['a'] + [aaa[4:]], aaa, 'split', None, 1)
279 self.checkequal(['a']*19 + ['a '], aaa, 'split', None, 19)
280
Hye-Shik Chang75c00ef2004-01-05 00:29:51 +0000281 # by a char
282 self.checkequal(['a', 'b', 'c', 'd'], 'a|b|c|d', 'split', '|')
Thomas Wouters477c8d52006-05-27 19:21:47 +0000283 self.checkequal(['a|b|c|d'], 'a|b|c|d', 'split', '|', 0)
Hye-Shik Chang75c00ef2004-01-05 00:29:51 +0000284 self.checkequal(['a', 'b|c|d'], 'a|b|c|d', 'split', '|', 1)
285 self.checkequal(['a', 'b', 'c|d'], 'a|b|c|d', 'split', '|', 2)
286 self.checkequal(['a', 'b', 'c', 'd'], 'a|b|c|d', 'split', '|', 3)
287 self.checkequal(['a', 'b', 'c', 'd'], 'a|b|c|d', 'split', '|', 4)
Thomas Wouters477c8d52006-05-27 19:21:47 +0000288 self.checkequal(['a', 'b', 'c', 'd'], 'a|b|c|d', 'split', '|',
289 sys.maxint-2)
Hye-Shik Chang75c00ef2004-01-05 00:29:51 +0000290 self.checkequal(['a|b|c|d'], 'a|b|c|d', 'split', '|', 0)
291 self.checkequal(['a', '', 'b||c||d'], 'a||b||c||d', 'split', '|', 2)
292 self.checkequal(['endcase ', ''], 'endcase |', 'split', '|')
Thomas Wouters477c8d52006-05-27 19:21:47 +0000293 self.checkequal(['', ' startcase'], '| startcase', 'split', '|')
294 self.checkequal(['', 'bothcase', ''], '|bothcase|', 'split', '|')
Hye-Shik Chang75c00ef2004-01-05 00:29:51 +0000295 self.checkequal(['a', '', 'b\x00c\x00d'], 'a\x00\x00b\x00c\x00d', 'split', '\x00', 2)
296
Thomas Wouters477c8d52006-05-27 19:21:47 +0000297 self.checkequal(['a']*20, ('a|'*20)[:-1], 'split', '|')
298 self.checkequal(['a']*15 +['a|a|a|a|a'],
299 ('a|'*20)[:-1], 'split', '|', 15)
300
Hye-Shik Chang75c00ef2004-01-05 00:29:51 +0000301 # by string
Walter Dörwald0fd583c2003-02-21 12:53:50 +0000302 self.checkequal(['a', 'b', 'c', 'd'], 'a//b//c//d', 'split', '//')
Hye-Shik Chang75c00ef2004-01-05 00:29:51 +0000303 self.checkequal(['a', 'b//c//d'], 'a//b//c//d', 'split', '//', 1)
304 self.checkequal(['a', 'b', 'c//d'], 'a//b//c//d', 'split', '//', 2)
305 self.checkequal(['a', 'b', 'c', 'd'], 'a//b//c//d', 'split', '//', 3)
306 self.checkequal(['a', 'b', 'c', 'd'], 'a//b//c//d', 'split', '//', 4)
Thomas Wouters477c8d52006-05-27 19:21:47 +0000307 self.checkequal(['a', 'b', 'c', 'd'], 'a//b//c//d', 'split', '//',
308 sys.maxint-10)
Hye-Shik Chang75c00ef2004-01-05 00:29:51 +0000309 self.checkequal(['a//b//c//d'], 'a//b//c//d', 'split', '//', 0)
310 self.checkequal(['a', '', 'b////c////d'], 'a////b////c////d', 'split', '//', 2)
Walter Dörwald0fd583c2003-02-21 12:53:50 +0000311 self.checkequal(['endcase ', ''], 'endcase test', 'split', 'test')
Thomas Wouters477c8d52006-05-27 19:21:47 +0000312 self.checkequal(['', ' begincase'], 'test begincase', 'split', 'test')
313 self.checkequal(['', ' bothcase ', ''], 'test bothcase test',
314 'split', 'test')
315 self.checkequal(['a', 'bc'], 'abbbc', 'split', 'bb')
316 self.checkequal(['', ''], 'aaa', 'split', 'aaa')
317 self.checkequal(['aaa'], 'aaa', 'split', 'aaa', 0)
318 self.checkequal(['ab', 'ab'], 'abbaab', 'split', 'ba')
319 self.checkequal(['aaaa'], 'aaaa', 'split', 'aab')
320 self.checkequal([''], '', 'split', 'aaa')
321 self.checkequal(['aa'], 'aa', 'split', 'aaa')
322 self.checkequal(['A', 'bobb'], 'Abbobbbobb', 'split', 'bbobb')
323 self.checkequal(['A', 'B', ''], 'AbbobbBbbobb', 'split', 'bbobb')
324
325 self.checkequal(['a']*20, ('aBLAH'*20)[:-4], 'split', 'BLAH')
326 self.checkequal(['a']*20, ('aBLAH'*20)[:-4], 'split', 'BLAH', 19)
327 self.checkequal(['a']*18 + ['aBLAHa'], ('aBLAH'*20)[:-4],
328 'split', 'BLAH', 18)
Walter Dörwald0fd583c2003-02-21 12:53:50 +0000329
Hye-Shik Chang75c00ef2004-01-05 00:29:51 +0000330 # mixed use of str and unicode
331 self.checkequal([u'a', u'b', u'c d'], 'a b c d', 'split', u' ', 2)
332
333 # argument type
Walter Dörwald0fd583c2003-02-21 12:53:50 +0000334 self.checkraises(TypeError, 'hello', 'split', 42, 42, 42)
335
Thomas Wouters477c8d52006-05-27 19:21:47 +0000336 # null case
337 self.checkraises(ValueError, 'hello', 'split', '')
338 self.checkraises(ValueError, 'hello', 'split', '', 0)
339
Hye-Shik Chang3ae811b2003-12-15 18:49:53 +0000340 def test_rsplit(self):
341 self.checkequal(['this', 'is', 'the', 'rsplit', 'function'],
342 'this is the rsplit function', 'rsplit')
Hye-Shik Chang75c00ef2004-01-05 00:29:51 +0000343
344 # by whitespace
345 self.checkequal(['a', 'b', 'c', 'd'], 'a b c d ', 'rsplit')
Hye-Shik Chang3ae811b2003-12-15 18:49:53 +0000346 self.checkequal(['a b c', 'd'], 'a b c d', 'rsplit', None, 1)
347 self.checkequal(['a b', 'c', 'd'], 'a b c d', 'rsplit', None, 2)
348 self.checkequal(['a', 'b', 'c', 'd'], 'a b c d', 'rsplit', None, 3)
349 self.checkequal(['a', 'b', 'c', 'd'], 'a b c d', 'rsplit', None, 4)
Thomas Wouters477c8d52006-05-27 19:21:47 +0000350 self.checkequal(['a', 'b', 'c', 'd'], 'a b c d', 'rsplit', None,
351 sys.maxint-20)
Hye-Shik Chang3ae811b2003-12-15 18:49:53 +0000352 self.checkequal(['a b c d'], 'a b c d', 'rsplit', None, 0)
Thomas Wouters477c8d52006-05-27 19:21:47 +0000353 self.checkequal(['a b c d'], 'a b c d ', 'rsplit', None, 0)
Hye-Shik Chang3ae811b2003-12-15 18:49:53 +0000354 self.checkequal(['a b', 'c', 'd'], 'a b c d', 'rsplit', None, 2)
Hye-Shik Chang75c00ef2004-01-05 00:29:51 +0000355
Thomas Wouters477c8d52006-05-27 19:21:47 +0000356 self.checkequal([], ' ', 'rsplit')
357 self.checkequal(['a'], ' a ', 'rsplit')
358 self.checkequal(['a', 'b'], ' a b ', 'rsplit')
359 self.checkequal([' a', 'b'], ' a b ', 'rsplit', None, 1)
360 self.checkequal([' a b','c'], ' a b c ', 'rsplit',
361 None, 1)
362 self.checkequal([' a', 'b', 'c'], ' a b c ', 'rsplit',
363 None, 2)
364 self.checkequal(['a', 'b'], '\n\ta \t\r b \v ', 'rsplit', None, 88)
365 aaa = ' a '*20
366 self.checkequal(['a']*20, aaa, 'rsplit')
367 self.checkequal([aaa[:-4]] + ['a'], aaa, 'rsplit', None, 1)
368 self.checkequal([' a a'] + ['a']*18, aaa, 'rsplit', None, 18)
369
370
Hye-Shik Chang75c00ef2004-01-05 00:29:51 +0000371 # by a char
372 self.checkequal(['a', 'b', 'c', 'd'], 'a|b|c|d', 'rsplit', '|')
373 self.checkequal(['a|b|c', 'd'], 'a|b|c|d', 'rsplit', '|', 1)
374 self.checkequal(['a|b', 'c', 'd'], 'a|b|c|d', 'rsplit', '|', 2)
375 self.checkequal(['a', 'b', 'c', 'd'], 'a|b|c|d', 'rsplit', '|', 3)
376 self.checkequal(['a', 'b', 'c', 'd'], 'a|b|c|d', 'rsplit', '|', 4)
Thomas Wouters477c8d52006-05-27 19:21:47 +0000377 self.checkequal(['a', 'b', 'c', 'd'], 'a|b|c|d', 'rsplit', '|',
378 sys.maxint-100)
Hye-Shik Chang75c00ef2004-01-05 00:29:51 +0000379 self.checkequal(['a|b|c|d'], 'a|b|c|d', 'rsplit', '|', 0)
380 self.checkequal(['a||b||c', '', 'd'], 'a||b||c||d', 'rsplit', '|', 2)
381 self.checkequal(['', ' begincase'], '| begincase', 'rsplit', '|')
Thomas Wouters477c8d52006-05-27 19:21:47 +0000382 self.checkequal(['endcase ', ''], 'endcase |', 'rsplit', '|')
383 self.checkequal(['', 'bothcase', ''], '|bothcase|', 'rsplit', '|')
384
Hye-Shik Chang75c00ef2004-01-05 00:29:51 +0000385 self.checkequal(['a\x00\x00b', 'c', 'd'], 'a\x00\x00b\x00c\x00d', 'rsplit', '\x00', 2)
386
Thomas Wouters477c8d52006-05-27 19:21:47 +0000387 self.checkequal(['a']*20, ('a|'*20)[:-1], 'rsplit', '|')
388 self.checkequal(['a|a|a|a|a']+['a']*15,
389 ('a|'*20)[:-1], 'rsplit', '|', 15)
390
Hye-Shik Chang75c00ef2004-01-05 00:29:51 +0000391 # by string
392 self.checkequal(['a', 'b', 'c', 'd'], 'a//b//c//d', 'rsplit', '//')
393 self.checkequal(['a//b//c', 'd'], 'a//b//c//d', 'rsplit', '//', 1)
394 self.checkequal(['a//b', 'c', 'd'], 'a//b//c//d', 'rsplit', '//', 2)
395 self.checkequal(['a', 'b', 'c', 'd'], 'a//b//c//d', 'rsplit', '//', 3)
396 self.checkequal(['a', 'b', 'c', 'd'], 'a//b//c//d', 'rsplit', '//', 4)
Thomas Wouters477c8d52006-05-27 19:21:47 +0000397 self.checkequal(['a', 'b', 'c', 'd'], 'a//b//c//d', 'rsplit', '//',
398 sys.maxint-5)
Hye-Shik Chang75c00ef2004-01-05 00:29:51 +0000399 self.checkequal(['a//b//c//d'], 'a//b//c//d', 'rsplit', '//', 0)
400 self.checkequal(['a////b////c', '', 'd'], 'a////b////c////d', 'rsplit', '//', 2)
401 self.checkequal(['', ' begincase'], 'test begincase', 'rsplit', 'test')
Thomas Wouters477c8d52006-05-27 19:21:47 +0000402 self.checkequal(['endcase ', ''], 'endcase test', 'rsplit', 'test')
403 self.checkequal(['', ' bothcase ', ''], 'test bothcase test',
404 'rsplit', 'test')
405 self.checkequal(['ab', 'c'], 'abbbc', 'rsplit', 'bb')
406 self.checkequal(['', ''], 'aaa', 'rsplit', 'aaa')
407 self.checkequal(['aaa'], 'aaa', 'rsplit', 'aaa', 0)
408 self.checkequal(['ab', 'ab'], 'abbaab', 'rsplit', 'ba')
409 self.checkequal(['aaaa'], 'aaaa', 'rsplit', 'aab')
410 self.checkequal([''], '', 'rsplit', 'aaa')
411 self.checkequal(['aa'], 'aa', 'rsplit', 'aaa')
412 self.checkequal(['bbob', 'A'], 'bbobbbobbA', 'rsplit', 'bbobb')
413 self.checkequal(['', 'B', 'A'], 'bbobbBbbobbA', 'rsplit', 'bbobb')
414
415 self.checkequal(['a']*20, ('aBLAH'*20)[:-4], 'rsplit', 'BLAH')
416 self.checkequal(['a']*20, ('aBLAH'*20)[:-4], 'rsplit', 'BLAH', 19)
417 self.checkequal(['aBLAHa'] + ['a']*18, ('aBLAH'*20)[:-4],
418 'rsplit', 'BLAH', 18)
Hye-Shik Chang75c00ef2004-01-05 00:29:51 +0000419
420 # mixed use of str and unicode
Hye-Shik Chang3ae811b2003-12-15 18:49:53 +0000421 self.checkequal([u'a b', u'c', u'd'], 'a b c d', 'rsplit', u' ', 2)
Hye-Shik Chang75c00ef2004-01-05 00:29:51 +0000422
423 # argument type
424 self.checkraises(TypeError, 'hello', 'rsplit', 42, 42, 42)
Hye-Shik Chang3ae811b2003-12-15 18:49:53 +0000425
Thomas Wouters477c8d52006-05-27 19:21:47 +0000426 # null case
427 self.checkraises(ValueError, 'hello', 'rsplit', '')
428 self.checkraises(ValueError, 'hello', 'rsplit', '', 0)
429
Walter Dörwald0fd583c2003-02-21 12:53:50 +0000430 def test_strip(self):
431 self.checkequal('hello', ' hello ', 'strip')
432 self.checkequal('hello ', ' hello ', 'lstrip')
433 self.checkequal(' hello', ' hello ', 'rstrip')
434 self.checkequal('hello', 'hello', 'strip')
435
Neal Norwitzffe33b72003-04-10 22:35:32 +0000436 # strip/lstrip/rstrip with None arg
437 self.checkequal('hello', ' hello ', 'strip', None)
438 self.checkequal('hello ', ' hello ', 'lstrip', None)
439 self.checkequal(' hello', ' hello ', 'rstrip', None)
440 self.checkequal('hello', 'hello', 'strip', None)
441
442 # strip/lstrip/rstrip with str arg
443 self.checkequal('hello', 'xyzzyhelloxyzzy', 'strip', 'xyz')
444 self.checkequal('helloxyzzy', 'xyzzyhelloxyzzy', 'lstrip', 'xyz')
445 self.checkequal('xyzzyhello', 'xyzzyhelloxyzzy', 'rstrip', 'xyz')
446 self.checkequal('hello', 'hello', 'strip', 'xyz')
447
448 # strip/lstrip/rstrip with unicode arg
449 if test_support.have_unicode:
450 self.checkequal(unicode('hello', 'ascii'), 'xyzzyhelloxyzzy',
451 'strip', unicode('xyz', 'ascii'))
452 self.checkequal(unicode('helloxyzzy', 'ascii'), 'xyzzyhelloxyzzy',
453 'lstrip', unicode('xyz', 'ascii'))
454 self.checkequal(unicode('xyzzyhello', 'ascii'), 'xyzzyhelloxyzzy',
455 'rstrip', unicode('xyz', 'ascii'))
456 self.checkequal(unicode('hello', 'ascii'), 'hello',
457 'strip', unicode('xyz', 'ascii'))
458
459 self.checkraises(TypeError, 'hello', 'strip', 42, 42)
460 self.checkraises(TypeError, 'hello', 'lstrip', 42, 42)
461 self.checkraises(TypeError, 'hello', 'rstrip', 42, 42)
462
Walter Dörwald0fd583c2003-02-21 12:53:50 +0000463 def test_ljust(self):
464 self.checkequal('abc ', 'abc', 'ljust', 10)
465 self.checkequal('abc ', 'abc', 'ljust', 6)
466 self.checkequal('abc', 'abc', 'ljust', 3)
467 self.checkequal('abc', 'abc', 'ljust', 2)
Raymond Hettinger4f8f9762003-11-26 08:21:35 +0000468 self.checkequal('abc*******', 'abc', 'ljust', 10, '*')
Walter Dörwald0fd583c2003-02-21 12:53:50 +0000469 self.checkraises(TypeError, 'abc', 'ljust')
470
471 def test_rjust(self):
472 self.checkequal(' abc', 'abc', 'rjust', 10)
473 self.checkequal(' abc', 'abc', 'rjust', 6)
474 self.checkequal('abc', 'abc', 'rjust', 3)
475 self.checkequal('abc', 'abc', 'rjust', 2)
Raymond Hettinger4f8f9762003-11-26 08:21:35 +0000476 self.checkequal('*******abc', 'abc', 'rjust', 10, '*')
Walter Dörwald0fd583c2003-02-21 12:53:50 +0000477 self.checkraises(TypeError, 'abc', 'rjust')
478
479 def test_center(self):
480 self.checkequal(' abc ', 'abc', 'center', 10)
481 self.checkequal(' abc ', 'abc', 'center', 6)
482 self.checkequal('abc', 'abc', 'center', 3)
483 self.checkequal('abc', 'abc', 'center', 2)
Raymond Hettinger4f8f9762003-11-26 08:21:35 +0000484 self.checkequal('***abc****', 'abc', 'center', 10, '*')
Walter Dörwald0fd583c2003-02-21 12:53:50 +0000485 self.checkraises(TypeError, 'abc', 'center')
486
487 def test_swapcase(self):
488 self.checkequal('hEllO CoMPuTErS', 'HeLLo cOmpUteRs', 'swapcase')
489
490 self.checkraises(TypeError, 'hello', 'swapcase', 42)
491
492 def test_replace(self):
Thomas Wouters477c8d52006-05-27 19:21:47 +0000493 EQ = self.checkequal
494
495 # Operations on the empty string
496 EQ("", "", "replace", "", "")
Thomas Wouters4d70c3d2006-06-08 14:42:34 +0000497 EQ("A", "", "replace", "", "A")
Thomas Wouters477c8d52006-05-27 19:21:47 +0000498 EQ("", "", "replace", "A", "")
499 EQ("", "", "replace", "A", "A")
500 EQ("", "", "replace", "", "", 100)
501 EQ("", "", "replace", "", "", sys.maxint)
502
503 # interleave (from=="", 'to' gets inserted everywhere)
504 EQ("A", "A", "replace", "", "")
505 EQ("*A*", "A", "replace", "", "*")
506 EQ("*1A*1", "A", "replace", "", "*1")
507 EQ("*-#A*-#", "A", "replace", "", "*-#")
508 EQ("*-A*-A*-", "AA", "replace", "", "*-")
509 EQ("*-A*-A*-", "AA", "replace", "", "*-", -1)
510 EQ("*-A*-A*-", "AA", "replace", "", "*-", sys.maxint)
511 EQ("*-A*-A*-", "AA", "replace", "", "*-", 4)
512 EQ("*-A*-A*-", "AA", "replace", "", "*-", 3)
513 EQ("*-A*-A", "AA", "replace", "", "*-", 2)
514 EQ("*-AA", "AA", "replace", "", "*-", 1)
515 EQ("AA", "AA", "replace", "", "*-", 0)
516
517 # single character deletion (from=="A", to=="")
518 EQ("", "A", "replace", "A", "")
519 EQ("", "AAA", "replace", "A", "")
520 EQ("", "AAA", "replace", "A", "", -1)
521 EQ("", "AAA", "replace", "A", "", sys.maxint)
522 EQ("", "AAA", "replace", "A", "", 4)
523 EQ("", "AAA", "replace", "A", "", 3)
524 EQ("A", "AAA", "replace", "A", "", 2)
525 EQ("AA", "AAA", "replace", "A", "", 1)
526 EQ("AAA", "AAA", "replace", "A", "", 0)
527 EQ("", "AAAAAAAAAA", "replace", "A", "")
528 EQ("BCD", "ABACADA", "replace", "A", "")
529 EQ("BCD", "ABACADA", "replace", "A", "", -1)
530 EQ("BCD", "ABACADA", "replace", "A", "", sys.maxint)
531 EQ("BCD", "ABACADA", "replace", "A", "", 5)
532 EQ("BCD", "ABACADA", "replace", "A", "", 4)
533 EQ("BCDA", "ABACADA", "replace", "A", "", 3)
534 EQ("BCADA", "ABACADA", "replace", "A", "", 2)
535 EQ("BACADA", "ABACADA", "replace", "A", "", 1)
536 EQ("ABACADA", "ABACADA", "replace", "A", "", 0)
537 EQ("BCD", "ABCAD", "replace", "A", "")
538 EQ("BCD", "ABCADAA", "replace", "A", "")
539 EQ("BCD", "BCD", "replace", "A", "")
540 EQ("*************", "*************", "replace", "A", "")
541 EQ("^A^", "^"+"A"*1000+"^", "replace", "A", "", 999)
542
543 # substring deletion (from=="the", to=="")
544 EQ("", "the", "replace", "the", "")
545 EQ("ater", "theater", "replace", "the", "")
546 EQ("", "thethe", "replace", "the", "")
547 EQ("", "thethethethe", "replace", "the", "")
548 EQ("aaaa", "theatheatheathea", "replace", "the", "")
549 EQ("that", "that", "replace", "the", "")
550 EQ("thaet", "thaet", "replace", "the", "")
551 EQ("here and re", "here and there", "replace", "the", "")
552 EQ("here and re and re", "here and there and there",
553 "replace", "the", "", sys.maxint)
554 EQ("here and re and re", "here and there and there",
555 "replace", "the", "", -1)
556 EQ("here and re and re", "here and there and there",
557 "replace", "the", "", 3)
558 EQ("here and re and re", "here and there and there",
559 "replace", "the", "", 2)
560 EQ("here and re and there", "here and there and there",
561 "replace", "the", "", 1)
562 EQ("here and there and there", "here and there and there",
563 "replace", "the", "", 0)
564 EQ("here and re and re", "here and there and there", "replace", "the", "")
565
566 EQ("abc", "abc", "replace", "the", "")
567 EQ("abcdefg", "abcdefg", "replace", "the", "")
568
569 # substring deletion (from=="bob", to=="")
570 EQ("bob", "bbobob", "replace", "bob", "")
571 EQ("bobXbob", "bbobobXbbobob", "replace", "bob", "")
572 EQ("aaaaaaa", "aaaaaaabob", "replace", "bob", "")
573 EQ("aaaaaaa", "aaaaaaa", "replace", "bob", "")
574
575 # single character replace in place (len(from)==len(to)==1)
576 EQ("Who goes there?", "Who goes there?", "replace", "o", "o")
577 EQ("WhO gOes there?", "Who goes there?", "replace", "o", "O")
578 EQ("WhO gOes there?", "Who goes there?", "replace", "o", "O", sys.maxint)
579 EQ("WhO gOes there?", "Who goes there?", "replace", "o", "O", -1)
580 EQ("WhO gOes there?", "Who goes there?", "replace", "o", "O", 3)
581 EQ("WhO gOes there?", "Who goes there?", "replace", "o", "O", 2)
582 EQ("WhO goes there?", "Who goes there?", "replace", "o", "O", 1)
583 EQ("Who goes there?", "Who goes there?", "replace", "o", "O", 0)
584
585 EQ("Who goes there?", "Who goes there?", "replace", "a", "q")
586 EQ("who goes there?", "Who goes there?", "replace", "W", "w")
587 EQ("wwho goes there?ww", "WWho goes there?WW", "replace", "W", "w")
588 EQ("Who goes there!", "Who goes there?", "replace", "?", "!")
589 EQ("Who goes there!!", "Who goes there??", "replace", "?", "!")
590
591 EQ("Who goes there?", "Who goes there?", "replace", ".", "!")
592
593 # substring replace in place (len(from)==len(to) > 1)
594 EQ("Th** ** a t**sue", "This is a tissue", "replace", "is", "**")
595 EQ("Th** ** a t**sue", "This is a tissue", "replace", "is", "**", sys.maxint)
596 EQ("Th** ** a t**sue", "This is a tissue", "replace", "is", "**", -1)
597 EQ("Th** ** a t**sue", "This is a tissue", "replace", "is", "**", 4)
598 EQ("Th** ** a t**sue", "This is a tissue", "replace", "is", "**", 3)
599 EQ("Th** ** a tissue", "This is a tissue", "replace", "is", "**", 2)
600 EQ("Th** is a tissue", "This is a tissue", "replace", "is", "**", 1)
601 EQ("This is a tissue", "This is a tissue", "replace", "is", "**", 0)
602 EQ("cobob", "bobob", "replace", "bob", "cob")
603 EQ("cobobXcobocob", "bobobXbobobob", "replace", "bob", "cob")
604 EQ("bobob", "bobob", "replace", "bot", "bot")
605
606 # replace single character (len(from)==1, len(to)>1)
607 EQ("ReyKKjaviKK", "Reykjavik", "replace", "k", "KK")
608 EQ("ReyKKjaviKK", "Reykjavik", "replace", "k", "KK", -1)
609 EQ("ReyKKjaviKK", "Reykjavik", "replace", "k", "KK", sys.maxint)
610 EQ("ReyKKjaviKK", "Reykjavik", "replace", "k", "KK", 2)
611 EQ("ReyKKjavik", "Reykjavik", "replace", "k", "KK", 1)
612 EQ("Reykjavik", "Reykjavik", "replace", "k", "KK", 0)
613 EQ("A----B----C----", "A.B.C.", "replace", ".", "----")
614
615 EQ("Reykjavik", "Reykjavik", "replace", "q", "KK")
616
617 # replace substring (len(from)>1, len(to)!=len(from))
618 EQ("ham, ham, eggs and ham", "spam, spam, eggs and spam",
619 "replace", "spam", "ham")
620 EQ("ham, ham, eggs and ham", "spam, spam, eggs and spam",
621 "replace", "spam", "ham", sys.maxint)
622 EQ("ham, ham, eggs and ham", "spam, spam, eggs and spam",
623 "replace", "spam", "ham", -1)
624 EQ("ham, ham, eggs and ham", "spam, spam, eggs and spam",
625 "replace", "spam", "ham", 4)
626 EQ("ham, ham, eggs and ham", "spam, spam, eggs and spam",
627 "replace", "spam", "ham", 3)
628 EQ("ham, ham, eggs and spam", "spam, spam, eggs and spam",
629 "replace", "spam", "ham", 2)
630 EQ("ham, spam, eggs and spam", "spam, spam, eggs and spam",
631 "replace", "spam", "ham", 1)
632 EQ("spam, spam, eggs and spam", "spam, spam, eggs and spam",
633 "replace", "spam", "ham", 0)
634
635 EQ("bobob", "bobobob", "replace", "bobob", "bob")
636 EQ("bobobXbobob", "bobobobXbobobob", "replace", "bobob", "bob")
637 EQ("BOBOBOB", "BOBOBOB", "replace", "bob", "bobby")
638
639 #
Walter Dörwald0fd583c2003-02-21 12:53:50 +0000640 self.checkequal('one@two!three!', 'one!two!three!', 'replace', '!', '@', 1)
641 self.checkequal('onetwothree', 'one!two!three!', 'replace', '!', '')
642 self.checkequal('one@two@three!', 'one!two!three!', 'replace', '!', '@', 2)
643 self.checkequal('one@two@three@', 'one!two!three!', 'replace', '!', '@', 3)
644 self.checkequal('one@two@three@', 'one!two!three!', 'replace', '!', '@', 4)
645 self.checkequal('one!two!three!', 'one!two!three!', 'replace', '!', '@', 0)
646 self.checkequal('one@two@three@', 'one!two!three!', 'replace', '!', '@')
647 self.checkequal('one!two!three!', 'one!two!three!', 'replace', 'x', '@')
648 self.checkequal('one!two!three!', 'one!two!three!', 'replace', 'x', '@', 2)
649 self.checkequal('-a-b-c-', 'abc', 'replace', '', '-')
650 self.checkequal('-a-b-c', 'abc', 'replace', '', '-', 3)
651 self.checkequal('abc', 'abc', 'replace', '', '-', 0)
652 self.checkequal('', '', 'replace', '', '')
653 self.checkequal('abc', 'abc', 'replace', 'ab', '--', 0)
654 self.checkequal('abc', 'abc', 'replace', 'xy', '--')
655 # Next three for SF bug 422088: [OSF1 alpha] string.replace(); died with
656 # MemoryError due to empty result (platform malloc issue when requesting
657 # 0 bytes).
658 self.checkequal('', '123', 'replace', '123', '')
659 self.checkequal('', '123123', 'replace', '123', '')
660 self.checkequal('x', '123x123', 'replace', '123', '')
661
662 self.checkraises(TypeError, 'hello', 'replace')
663 self.checkraises(TypeError, 'hello', 'replace', 42)
664 self.checkraises(TypeError, 'hello', 'replace', 42, 'h')
665 self.checkraises(TypeError, 'hello', 'replace', 'h', 42)
666
Thomas Wouters477c8d52006-05-27 19:21:47 +0000667 def test_replace_overflow(self):
668 # Check for overflow checking on 32 bit machines
669 if sys.maxint != 2147483647:
670 return
671 A2_16 = "A" * (2**16)
672 self.checkraises(OverflowError, A2_16, "replace", "", A2_16)
673 self.checkraises(OverflowError, A2_16, "replace", "A", A2_16)
674 self.checkraises(OverflowError, A2_16, "replace", "AA", A2_16+A2_16)
675
Walter Dörwald0fd583c2003-02-21 12:53:50 +0000676 def test_zfill(self):
677 self.checkequal('123', '123', 'zfill', 2)
678 self.checkequal('123', '123', 'zfill', 3)
679 self.checkequal('0123', '123', 'zfill', 4)
680 self.checkequal('+123', '+123', 'zfill', 3)
681 self.checkequal('+123', '+123', 'zfill', 4)
682 self.checkequal('+0123', '+123', 'zfill', 5)
683 self.checkequal('-123', '-123', 'zfill', 3)
684 self.checkequal('-123', '-123', 'zfill', 4)
685 self.checkequal('-0123', '-123', 'zfill', 5)
686 self.checkequal('000', '', 'zfill', 3)
687 self.checkequal('34', '34', 'zfill', 1)
688 self.checkequal('0034', '34', 'zfill', 4)
689
690 self.checkraises(TypeError, '123', 'zfill')
691
692class MixinStrUnicodeUserStringTest:
693 # additional tests that only work for
694 # stringlike objects, i.e. str, unicode, UserString
695 # (but not the string module)
696
697 def test_islower(self):
698 self.checkequal(False, '', 'islower')
699 self.checkequal(True, 'a', 'islower')
700 self.checkequal(False, 'A', 'islower')
701 self.checkequal(False, '\n', 'islower')
702 self.checkequal(True, 'abc', 'islower')
703 self.checkequal(False, 'aBc', 'islower')
704 self.checkequal(True, 'abc\n', 'islower')
705 self.checkraises(TypeError, 'abc', 'islower', 42)
706
707 def test_isupper(self):
708 self.checkequal(False, '', 'isupper')
709 self.checkequal(False, 'a', 'isupper')
710 self.checkequal(True, 'A', 'isupper')
711 self.checkequal(False, '\n', 'isupper')
712 self.checkequal(True, 'ABC', 'isupper')
713 self.checkequal(False, 'AbC', 'isupper')
714 self.checkequal(True, 'ABC\n', 'isupper')
715 self.checkraises(TypeError, 'abc', 'isupper', 42)
716
717 def test_istitle(self):
718 self.checkequal(False, '', 'istitle')
719 self.checkequal(False, 'a', 'istitle')
720 self.checkequal(True, 'A', 'istitle')
721 self.checkequal(False, '\n', 'istitle')
722 self.checkequal(True, 'A Titlecased Line', 'istitle')
723 self.checkequal(True, 'A\nTitlecased Line', 'istitle')
724 self.checkequal(True, 'A Titlecased, Line', 'istitle')
725 self.checkequal(False, 'Not a capitalized String', 'istitle')
726 self.checkequal(False, 'Not\ta Titlecase String', 'istitle')
727 self.checkequal(False, 'Not--a Titlecase String', 'istitle')
728 self.checkequal(False, 'NOT', 'istitle')
729 self.checkraises(TypeError, 'abc', 'istitle', 42)
730
731 def test_isspace(self):
732 self.checkequal(False, '', 'isspace')
733 self.checkequal(False, 'a', 'isspace')
734 self.checkequal(True, ' ', 'isspace')
735 self.checkequal(True, '\t', 'isspace')
736 self.checkequal(True, '\r', 'isspace')
737 self.checkequal(True, '\n', 'isspace')
738 self.checkequal(True, ' \t\r\n', 'isspace')
739 self.checkequal(False, ' \t\r\na', 'isspace')
740 self.checkraises(TypeError, 'abc', 'isspace', 42)
741
742 def test_isalpha(self):
743 self.checkequal(False, '', 'isalpha')
744 self.checkequal(True, 'a', 'isalpha')
745 self.checkequal(True, 'A', 'isalpha')
746 self.checkequal(False, '\n', 'isalpha')
747 self.checkequal(True, 'abc', 'isalpha')
748 self.checkequal(False, 'aBc123', 'isalpha')
749 self.checkequal(False, 'abc\n', 'isalpha')
750 self.checkraises(TypeError, 'abc', 'isalpha', 42)
751
752 def test_isalnum(self):
753 self.checkequal(False, '', 'isalnum')
754 self.checkequal(True, 'a', 'isalnum')
755 self.checkequal(True, 'A', 'isalnum')
756 self.checkequal(False, '\n', 'isalnum')
757 self.checkequal(True, '123abc456', 'isalnum')
758 self.checkequal(True, 'a1b3c', 'isalnum')
759 self.checkequal(False, 'aBc000 ', 'isalnum')
760 self.checkequal(False, 'abc\n', 'isalnum')
761 self.checkraises(TypeError, 'abc', 'isalnum', 42)
762
763 def test_isdigit(self):
764 self.checkequal(False, '', 'isdigit')
765 self.checkequal(False, 'a', 'isdigit')
766 self.checkequal(True, '0', 'isdigit')
767 self.checkequal(True, '0123456789', 'isdigit')
768 self.checkequal(False, '0123456789a', 'isdigit')
769
770 self.checkraises(TypeError, 'abc', 'isdigit', 42)
771
772 def test_title(self):
773 self.checkequal(' Hello ', ' hello ', 'title')
774 self.checkequal('Hello ', 'hello ', 'title')
775 self.checkequal('Hello ', 'Hello ', 'title')
776 self.checkequal('Format This As Title String', "fOrMaT thIs aS titLe String", 'title')
777 self.checkequal('Format,This-As*Title;String', "fOrMaT,thIs-aS*titLe;String", 'title', )
778 self.checkequal('Getint', "getInt", 'title')
779 self.checkraises(TypeError, 'hello', 'title', 42)
780
781 def test_splitlines(self):
782 self.checkequal(['abc', 'def', '', 'ghi'], "abc\ndef\n\rghi", 'splitlines')
783 self.checkequal(['abc', 'def', '', 'ghi'], "abc\ndef\n\r\nghi", 'splitlines')
784 self.checkequal(['abc', 'def', 'ghi'], "abc\ndef\r\nghi", 'splitlines')
785 self.checkequal(['abc', 'def', 'ghi'], "abc\ndef\r\nghi\n", 'splitlines')
786 self.checkequal(['abc', 'def', 'ghi', ''], "abc\ndef\r\nghi\n\r", 'splitlines')
787 self.checkequal(['', 'abc', 'def', 'ghi', ''], "\nabc\ndef\r\nghi\n\r", 'splitlines')
788 self.checkequal(['\n', 'abc\n', 'def\r\n', 'ghi\n', '\r'], "\nabc\ndef\r\nghi\n\r", 'splitlines', 1)
789
790 self.checkraises(TypeError, 'abc', 'splitlines', 42, 42)
791
792 def test_startswith(self):
793 self.checkequal(True, 'hello', 'startswith', 'he')
794 self.checkequal(True, 'hello', 'startswith', 'hello')
795 self.checkequal(False, 'hello', 'startswith', 'hello world')
796 self.checkequal(True, 'hello', 'startswith', '')
797 self.checkequal(False, 'hello', 'startswith', 'ello')
798 self.checkequal(True, 'hello', 'startswith', 'ello', 1)
799 self.checkequal(True, 'hello', 'startswith', 'o', 4)
800 self.checkequal(False, 'hello', 'startswith', 'o', 5)
801 self.checkequal(True, 'hello', 'startswith', '', 5)
802 self.checkequal(False, 'hello', 'startswith', 'lo', 6)
803 self.checkequal(True, 'helloworld', 'startswith', 'lowo', 3)
804 self.checkequal(True, 'helloworld', 'startswith', 'lowo', 3, 7)
805 self.checkequal(False, 'helloworld', 'startswith', 'lowo', 3, 6)
806
807 # test negative indices
808 self.checkequal(True, 'hello', 'startswith', 'he', 0, -1)
809 self.checkequal(True, 'hello', 'startswith', 'he', -53, -1)
810 self.checkequal(False, 'hello', 'startswith', 'hello', 0, -1)
811 self.checkequal(False, 'hello', 'startswith', 'hello world', -1, -10)
812 self.checkequal(False, 'hello', 'startswith', 'ello', -5)
813 self.checkequal(True, 'hello', 'startswith', 'ello', -4)
814 self.checkequal(False, 'hello', 'startswith', 'o', -2)
815 self.checkequal(True, 'hello', 'startswith', 'o', -1)
816 self.checkequal(True, 'hello', 'startswith', '', -3, -3)
817 self.checkequal(False, 'hello', 'startswith', 'lo', -9)
818
819 self.checkraises(TypeError, 'hello', 'startswith')
820 self.checkraises(TypeError, 'hello', 'startswith', 42)
821
822 def test_endswith(self):
823 self.checkequal(True, 'hello', 'endswith', 'lo')
824 self.checkequal(False, 'hello', 'endswith', 'he')
825 self.checkequal(True, 'hello', 'endswith', '')
826 self.checkequal(False, 'hello', 'endswith', 'hello world')
827 self.checkequal(False, 'helloworld', 'endswith', 'worl')
828 self.checkequal(True, 'helloworld', 'endswith', 'worl', 3, 9)
829 self.checkequal(True, 'helloworld', 'endswith', 'world', 3, 12)
830 self.checkequal(True, 'helloworld', 'endswith', 'lowo', 1, 7)
831 self.checkequal(True, 'helloworld', 'endswith', 'lowo', 2, 7)
832 self.checkequal(True, 'helloworld', 'endswith', 'lowo', 3, 7)
833 self.checkequal(False, 'helloworld', 'endswith', 'lowo', 4, 7)
834 self.checkequal(False, 'helloworld', 'endswith', 'lowo', 3, 8)
835 self.checkequal(False, 'ab', 'endswith', 'ab', 0, 1)
836 self.checkequal(False, 'ab', 'endswith', 'ab', 0, 0)
837
838 # test negative indices
839 self.checkequal(True, 'hello', 'endswith', 'lo', -2)
840 self.checkequal(False, 'hello', 'endswith', 'he', -2)
841 self.checkequal(True, 'hello', 'endswith', '', -3, -3)
842 self.checkequal(False, 'hello', 'endswith', 'hello world', -10, -2)
843 self.checkequal(False, 'helloworld', 'endswith', 'worl', -6)
844 self.checkequal(True, 'helloworld', 'endswith', 'worl', -5, -1)
845 self.checkequal(True, 'helloworld', 'endswith', 'worl', -5, 9)
846 self.checkequal(True, 'helloworld', 'endswith', 'world', -7, 12)
847 self.checkequal(True, 'helloworld', 'endswith', 'lowo', -99, -3)
848 self.checkequal(True, 'helloworld', 'endswith', 'lowo', -8, -3)
849 self.checkequal(True, 'helloworld', 'endswith', 'lowo', -7, -3)
850 self.checkequal(False, 'helloworld', 'endswith', 'lowo', 3, -4)
851 self.checkequal(False, 'helloworld', 'endswith', 'lowo', -8, -2)
852
853 self.checkraises(TypeError, 'hello', 'endswith')
854 self.checkraises(TypeError, 'hello', 'endswith', 42)
855
Walter Dörwald0fd583c2003-02-21 12:53:50 +0000856 def test___contains__(self):
857 self.checkequal(True, '', '__contains__', '') # vereq('' in '', True)
858 self.checkequal(True, 'abc', '__contains__', '') # vereq('' in 'abc', True)
859 self.checkequal(False, 'abc', '__contains__', '\0') # vereq('\0' in 'abc', False)
860 self.checkequal(True, '\0abc', '__contains__', '\0') # vereq('\0' in '\0abc', True)
861 self.checkequal(True, 'abc\0', '__contains__', '\0') # vereq('\0' in 'abc\0', True)
862 self.checkequal(True, '\0abc', '__contains__', 'a') # vereq('a' in '\0abc', True)
863 self.checkequal(True, 'asdf', '__contains__', 'asdf') # vereq('asdf' in 'asdf', True)
864 self.checkequal(False, 'asd', '__contains__', 'asdf') # vereq('asdf' in 'asd', False)
865 self.checkequal(False, '', '__contains__', 'asdf') # vereq('asdf' in '', False)
866
867 def test_subscript(self):
868 self.checkequal(u'a', 'abc', '__getitem__', 0)
869 self.checkequal(u'c', 'abc', '__getitem__', -1)
870 self.checkequal(u'a', 'abc', '__getitem__', 0L)
871 self.checkequal(u'abc', 'abc', '__getitem__', slice(0, 3))
872 self.checkequal(u'abc', 'abc', '__getitem__', slice(0, 1000))
873 self.checkequal(u'a', 'abc', '__getitem__', slice(0, 1))
874 self.checkequal(u'', 'abc', '__getitem__', slice(0, 0))
875 # FIXME What about negative indizes? This is handled differently by [] and __getitem__(slice)
876
877 self.checkraises(TypeError, 'abc', '__getitem__', 'def')
878
879 def test_slice(self):
880 self.checkequal('abc', 'abc', '__getslice__', 0, 1000)
881 self.checkequal('abc', 'abc', '__getslice__', 0, 3)
882 self.checkequal('ab', 'abc', '__getslice__', 0, 2)
883 self.checkequal('bc', 'abc', '__getslice__', 1, 3)
884 self.checkequal('b', 'abc', '__getslice__', 1, 2)
885 self.checkequal('', 'abc', '__getslice__', 2, 2)
886 self.checkequal('', 'abc', '__getslice__', 1000, 1000)
887 self.checkequal('', 'abc', '__getslice__', 2000, 1000)
888 self.checkequal('', 'abc', '__getslice__', 2, 1)
889 # FIXME What about negative indizes? This is handled differently by [] and __getslice__
890
891 self.checkraises(TypeError, 'abc', '__getslice__', 'def')
892
893 def test_mul(self):
894 self.checkequal('', 'abc', '__mul__', -1)
895 self.checkequal('', 'abc', '__mul__', 0)
896 self.checkequal('abc', 'abc', '__mul__', 1)
897 self.checkequal('abcabcabc', 'abc', '__mul__', 3)
898 self.checkraises(TypeError, 'abc', '__mul__')
899 self.checkraises(TypeError, 'abc', '__mul__', '')
Martin v. Löwis18e16552006-02-15 17:27:45 +0000900 # XXX: on a 64-bit system, this doesn't raise an overflow error,
901 # but either raises a MemoryError, or succeeds (if you have 54TiB)
902 #self.checkraises(OverflowError, 10000*'abc', '__mul__', 2000000000)
Walter Dörwald0fd583c2003-02-21 12:53:50 +0000903
904 def test_join(self):
905 # join now works with any sequence type
906 # moved here, because the argument order is
907 # different in string.join (see the test in
908 # test.test_string.StringTest.test_join)
909 self.checkequal('a b c d', ' ', 'join', ['a', 'b', 'c', 'd'])
910 self.checkequal('abcd', '', 'join', ('a', 'b', 'c', 'd'))
911 self.checkequal('w x y z', ' ', 'join', Sequence())
912 self.checkequal('abc', 'a', 'join', ('abc',))
913 self.checkequal('z', 'a', 'join', UserList(['z']))
914 if test_support.have_unicode:
915 self.checkequal(unicode('a.b.c'), unicode('.'), 'join', ['a', 'b', 'c'])
916 self.checkequal(unicode('a.b.c'), '.', 'join', [unicode('a'), 'b', 'c'])
917 self.checkequal(unicode('a.b.c'), '.', 'join', ['a', unicode('b'), 'c'])
918 self.checkequal(unicode('a.b.c'), '.', 'join', ['a', 'b', unicode('c')])
919 self.checkraises(TypeError, '.', 'join', ['a', unicode('b'), 3])
920 for i in [5, 25, 125]:
921 self.checkequal(((('a' * i) + '-') * i)[:-1], '-', 'join',
922 ['a' * i] * i)
923 self.checkequal(((('a' * i) + '-') * i)[:-1], '-', 'join',
924 ('a' * i,) * i)
925
926 self.checkraises(TypeError, ' ', 'join', BadSeq1())
927 self.checkequal('a b c', ' ', 'join', BadSeq2())
928
929 self.checkraises(TypeError, ' ', 'join')
930 self.checkraises(TypeError, ' ', 'join', 7)
931 self.checkraises(TypeError, ' ', 'join', Sequence([7, 'hello', 123L]))
Michael W. Hudsonb2308bb2005-10-21 11:45:01 +0000932 try:
933 def f():
934 yield 4 + ""
935 self.fixtype(' ').join(f())
936 except TypeError, e:
937 if '+' not in str(e):
938 self.fail('join() ate exception message')
939 else:
940 self.fail('exception not raised')
Walter Dörwald0fd583c2003-02-21 12:53:50 +0000941
942 def test_formatting(self):
943 self.checkequal('+hello+', '+%s+', '__mod__', 'hello')
944 self.checkequal('+10+', '+%d+', '__mod__', 10)
945 self.checkequal('a', "%c", '__mod__', "a")
946 self.checkequal('a', "%c", '__mod__', "a")
947 self.checkequal('"', "%c", '__mod__', 34)
948 self.checkequal('$', "%c", '__mod__', 36)
949 self.checkequal('10', "%d", '__mod__', 10)
Walter Dörwald43440a62003-03-31 18:07:50 +0000950 self.checkequal('\x7f', "%c", '__mod__', 0x7f)
Walter Dörwald0fd583c2003-02-21 12:53:50 +0000951
952 for ordinal in (-100, 0x200000):
953 # unicode raises ValueError, str raises OverflowError
954 self.checkraises((ValueError, OverflowError), '%c', '__mod__', ordinal)
955
956 self.checkequal(' 42', '%3ld', '__mod__', 42)
957 self.checkequal('0042.00', '%07.2f', '__mod__', 42)
Raymond Hettinger9bfe5332003-08-27 04:55:52 +0000958 self.checkequal('0042.00', '%07.2F', '__mod__', 42)
Walter Dörwald0fd583c2003-02-21 12:53:50 +0000959
960 self.checkraises(TypeError, 'abc', '__mod__')
961 self.checkraises(TypeError, '%(foo)s', '__mod__', 42)
962 self.checkraises(TypeError, '%s%s', '__mod__', (42,))
963 self.checkraises(TypeError, '%c', '__mod__', (None,))
964 self.checkraises(ValueError, '%(foo', '__mod__', {})
965 self.checkraises(TypeError, '%(foo)s %(bar)s', '__mod__', ('foo', 42))
966
967 # argument names with properly nested brackets are supported
968 self.checkequal('bar', '%((foo))s', '__mod__', {'(foo)': 'bar'})
969
970 # 100 is a magic number in PyUnicode_Format, this forces a resize
971 self.checkequal(103*'a'+'x', '%sx', '__mod__', 103*'a')
972
973 self.checkraises(TypeError, '%*s', '__mod__', ('foo', 'bar'))
974 self.checkraises(TypeError, '%10.*f', '__mod__', ('foo', 42.))
975 self.checkraises(ValueError, '%10', '__mod__', (42,))
976
977 def test_floatformatting(self):
978 # float formatting
979 for prec in xrange(100):
980 format = '%%.%if' % prec
981 value = 0.01
982 for x in xrange(60):
983 value = value * 3.141592655 / 3.0 * 10.0
984 # The formatfloat() code in stringobject.c and
985 # unicodeobject.c uses a 120 byte buffer and switches from
986 # 'f' formatting to 'g' at precision 50, so we expect
987 # OverflowErrors for the ranges x < 50 and prec >= 67.
988 if x < 50 and prec >= 67:
989 self.checkraises(OverflowError, format, "__mod__", value)
990 else:
991 self.checkcall(format, "__mod__", value)
992
Thomas Wouters477c8d52006-05-27 19:21:47 +0000993 def test_inplace_rewrites(self):
994 # Check that strings don't copy and modify cached single-character strings
995 self.checkequal('a', 'A', 'lower')
996 self.checkequal(True, 'A', 'isupper')
997 self.checkequal('A', 'a', 'upper')
998 self.checkequal(True, 'a', 'islower')
999
1000 self.checkequal('a', 'A', 'replace', 'A', 'a')
1001 self.checkequal(True, 'A', 'isupper')
1002
1003 self.checkequal('A', 'a', 'capitalize')
1004 self.checkequal(True, 'a', 'islower')
1005
1006 self.checkequal('A', 'a', 'swapcase')
1007 self.checkequal(True, 'a', 'islower')
1008
1009 self.checkequal('A', 'a', 'title')
1010 self.checkequal(True, 'a', 'islower')
1011
1012 def test_partition(self):
1013
1014 self.checkequal(('this is the par', 'ti', 'tion method'),
1015 'this is the partition method', 'partition', 'ti')
1016
1017 # from raymond's original specification
1018 S = 'http://www.python.org'
1019 self.checkequal(('http', '://', 'www.python.org'), S, 'partition', '://')
1020 self.checkequal(('http://www.python.org', '', ''), S, 'partition', '?')
1021 self.checkequal(('', 'http://', 'www.python.org'), S, 'partition', 'http://')
1022 self.checkequal(('http://www.python.', 'org', ''), S, 'partition', 'org')
1023
1024 self.checkraises(ValueError, S, 'partition', '')
1025 self.checkraises(TypeError, S, 'partition', None)
1026
1027 def test_rpartition(self):
1028
1029 self.checkequal(('this is the rparti', 'ti', 'on method'),
1030 'this is the rpartition method', 'rpartition', 'ti')
1031
1032 # from raymond's original specification
1033 S = 'http://www.python.org'
1034 self.checkequal(('http', '://', 'www.python.org'), S, 'rpartition', '://')
1035 self.checkequal(('http://www.python.org', '', ''), S, 'rpartition', '?')
1036 self.checkequal(('', 'http://', 'www.python.org'), S, 'rpartition', 'http://')
1037 self.checkequal(('http://www.python.', 'org', ''), S, 'rpartition', 'org')
1038
1039 self.checkraises(ValueError, S, 'rpartition', '')
1040 self.checkraises(TypeError, S, 'rpartition', None)
1041
Walter Dörwald57d88e52004-08-26 16:53:04 +00001042
Walter Dörwald0fd583c2003-02-21 12:53:50 +00001043class MixinStrStringUserStringTest:
1044 # Additional tests for 8bit strings, i.e. str, UserString and
1045 # the string module
1046
1047 def test_maketrans(self):
1048 self.assertEqual(
1049 ''.join(map(chr, xrange(256))).replace('abc', 'xyz'),
1050 string.maketrans('abc', 'xyz')
1051 )
1052 self.assertRaises(ValueError, string.maketrans, 'abc', 'xyzw')
1053
1054 def test_translate(self):
1055 table = string.maketrans('abc', 'xyz')
1056 self.checkequal('xyzxyz', 'xyzabcdef', 'translate', table, 'def')
1057
1058 table = string.maketrans('a', 'A')
1059 self.checkequal('Abc', 'abc', 'translate', table)
1060 self.checkequal('xyz', 'xyz', 'translate', table)
1061 self.checkequal('yz', 'xyz', 'translate', table, 'x')
1062 self.checkraises(ValueError, 'xyz', 'translate', 'too short', 'strip')
1063 self.checkraises(ValueError, 'xyz', 'translate', 'too short')
Jeremy Hyltonf82b04e2000-07-10 17:08:42 +00001064
1065
Walter Dörwald0fd583c2003-02-21 12:53:50 +00001066class MixinStrUserStringTest:
1067 # Additional tests that only work with
1068 # 8bit compatible object, i.e. str and UserString
Jeremy Hyltonf82b04e2000-07-10 17:08:42 +00001069
Walter Dörwald6eea7892005-07-28 16:49:15 +00001070 if test_support.have_unicode:
1071 def test_encoding_decoding(self):
1072 codecs = [('rot13', 'uryyb jbeyq'),
1073 ('base64', 'aGVsbG8gd29ybGQ=\n'),
1074 ('hex', '68656c6c6f20776f726c64'),
1075 ('uu', 'begin 666 <data>\n+:&5L;&\\@=V]R;&0 \n \nend\n')]
1076 for encoding, data in codecs:
1077 self.checkequal(data, 'hello world', 'encode', encoding)
1078 self.checkequal('hello world', data, 'decode', encoding)
1079 # zlib is optional, so we make the test optional too...
1080 try:
1081 import zlib
1082 except ImportError:
1083 pass
1084 else:
1085 data = 'x\x9c\xcbH\xcd\xc9\xc9W(\xcf/\xcaI\x01\x00\x1a\x0b\x04]'
1086 self.checkequal(data, 'hello world', 'encode', 'zlib')
1087 self.checkequal('hello world', data, 'decode', 'zlib')
Walter Dörwald97951de2003-03-26 14:31:25 +00001088
Walter Dörwald6eea7892005-07-28 16:49:15 +00001089 self.checkraises(TypeError, 'xyz', 'decode', 42)
1090 self.checkraises(TypeError, 'xyz', 'encode', 42)
Walter Dörwald57d88e52004-08-26 16:53:04 +00001091
1092
1093class MixinStrUnicodeTest:
Tim Peters108f1372004-08-27 05:36:07 +00001094 # Additional tests that only work with str and unicode.
Walter Dörwald57d88e52004-08-26 16:53:04 +00001095
1096 def test_bug1001011(self):
1097 # Make sure join returns a NEW object for single item sequences
Tim Peters108f1372004-08-27 05:36:07 +00001098 # involving a subclass.
1099 # Make sure that it is of the appropriate type.
1100 # Check the optimisation still occurs for standard objects.
Walter Dörwald57d88e52004-08-26 16:53:04 +00001101 t = self.type2test
1102 class subclass(t):
1103 pass
1104 s1 = subclass("abcd")
1105 s2 = t().join([s1])
1106 self.assert_(s1 is not s2)
1107 self.assert_(type(s2) is t)
Tim Peters108f1372004-08-27 05:36:07 +00001108
1109 s1 = t("abcd")
1110 s2 = t().join([s1])
1111 self.assert_(s1 is s2)
1112
1113 # Should also test mixed-type join.
1114 if t is unicode:
1115 s1 = subclass("abcd")
1116 s2 = "".join([s1])
1117 self.assert_(s1 is not s2)
1118 self.assert_(type(s2) is t)
1119
1120 s1 = t("abcd")
1121 s2 = "".join([s1])
1122 self.assert_(s1 is s2)
1123
1124 elif t is str:
1125 s1 = subclass("abcd")
1126 s2 = u"".join([s1])
1127 self.assert_(s1 is not s2)
1128 self.assert_(type(s2) is unicode) # promotes!
1129
1130 s1 = t("abcd")
1131 s2 = u"".join([s1])
1132 self.assert_(s1 is not s2)
1133 self.assert_(type(s2) is unicode) # promotes!
1134
1135 else:
1136 self.fail("unexpected type for MixinStrUnicodeTest %r" % t)