source: trunk/essentials/dev-lang/python/Lib/test/pickletester.py@ 3951

Last change on this file since 3951 was 3225, checked in by bird, 19 years ago

Python 2.5

File size: 29.3 KB
Line 
1import unittest
2import pickle
3import cPickle
4import pickletools
5import copy_reg
6
7from test.test_support import TestFailed, have_unicode, TESTFN, \
8 run_with_locale
9
10# Tests that try a number of pickle protocols should have a
11# for proto in protocols:
12# kind of outer loop.
13assert pickle.HIGHEST_PROTOCOL == cPickle.HIGHEST_PROTOCOL == 2
14protocols = range(pickle.HIGHEST_PROTOCOL + 1)
15
16
17# Return True if opcode code appears in the pickle, else False.
18def opcode_in_pickle(code, pickle):
19 for op, dummy, dummy in pickletools.genops(pickle):
20 if op.code == code:
21 return True
22 return False
23
24# Return the number of times opcode code appears in pickle.
25def count_opcode(code, pickle):
26 n = 0
27 for op, dummy, dummy in pickletools.genops(pickle):
28 if op.code == code:
29 n += 1
30 return n
31
32# We can't very well test the extension registry without putting known stuff
33# in it, but we have to be careful to restore its original state. Code
34# should do this:
35#
36# e = ExtensionSaver(extension_code)
37# try:
38# fiddle w/ the extension registry's stuff for extension_code
39# finally:
40# e.restore()
41
42class ExtensionSaver:
43 # Remember current registration for code (if any), and remove it (if
44 # there is one).
45 def __init__(self, code):
46 self.code = code
47 if code in copy_reg._inverted_registry:
48 self.pair = copy_reg._inverted_registry[code]
49 copy_reg.remove_extension(self.pair[0], self.pair[1], code)
50 else:
51 self.pair = None
52
53 # Restore previous registration for code.
54 def restore(self):
55 code = self.code
56 curpair = copy_reg._inverted_registry.get(code)
57 if curpair is not None:
58 copy_reg.remove_extension(curpair[0], curpair[1], code)
59 pair = self.pair
60 if pair is not None:
61 copy_reg.add_extension(pair[0], pair[1], code)
62
63class C:
64 def __cmp__(self, other):
65 return cmp(self.__dict__, other.__dict__)
66
67import __main__
68__main__.C = C
69C.__module__ = "__main__"
70
71class myint(int):
72 def __init__(self, x):
73 self.str = str(x)
74
75class initarg(C):
76
77 def __init__(self, a, b):
78 self.a = a
79 self.b = b
80
81 def __getinitargs__(self):
82 return self.a, self.b
83
84class metaclass(type):
85 pass
86
87class use_metaclass(object):
88 __metaclass__ = metaclass
89
90# DATA0 .. DATA2 are the pickles we expect under the various protocols, for
91# the object returned by create_data().
92
93# break into multiple strings to avoid confusing font-lock-mode
94DATA0 = """(lp1
95I0
96aL1L
97aF2
98ac__builtin__
99complex
100p2
101""" + \
102"""(F3
103F0
104tRp3
105aI1
106aI-1
107aI255
108aI-255
109aI-256
110aI65535
111aI-65535
112aI-65536
113aI2147483647
114aI-2147483647
115aI-2147483648
116a""" + \
117"""(S'abc'
118p4
119g4
120""" + \
121"""(i__main__
122C
123p5
124""" + \
125"""(dp6
126S'foo'
127p7
128I1
129sS'bar'
130p8
131I2
132sbg5
133tp9
134ag9
135aI5
136a.
137"""
138
139# Disassembly of DATA0.
140DATA0_DIS = """\
141 0: ( MARK
142 1: l LIST (MARK at 0)
143 2: p PUT 1
144 5: I INT 0
145 8: a APPEND
146 9: L LONG 1L
147 13: a APPEND
148 14: F FLOAT 2.0
149 17: a APPEND
150 18: c GLOBAL '__builtin__ complex'
151 39: p PUT 2
152 42: ( MARK
153 43: F FLOAT 3.0
154 46: F FLOAT 0.0
155 49: t TUPLE (MARK at 42)
156 50: R REDUCE
157 51: p PUT 3
158 54: a APPEND
159 55: I INT 1
160 58: a APPEND
161 59: I INT -1
162 63: a APPEND
163 64: I INT 255
164 69: a APPEND
165 70: I INT -255
166 76: a APPEND
167 77: I INT -256
168 83: a APPEND
169 84: I INT 65535
170 91: a APPEND
171 92: I INT -65535
172 100: a APPEND
173 101: I INT -65536
174 109: a APPEND
175 110: I INT 2147483647
176 122: a APPEND
177 123: I INT -2147483647
178 136: a APPEND
179 137: I INT -2147483648
180 150: a APPEND
181 151: ( MARK
182 152: S STRING 'abc'
183 159: p PUT 4
184 162: g GET 4
185 165: ( MARK
186 166: i INST '__main__ C' (MARK at 165)
187 178: p PUT 5
188 181: ( MARK
189 182: d DICT (MARK at 181)
190 183: p PUT 6
191 186: S STRING 'foo'
192 193: p PUT 7
193 196: I INT 1
194 199: s SETITEM
195 200: S STRING 'bar'
196 207: p PUT 8
197 210: I INT 2
198 213: s SETITEM
199 214: b BUILD
200 215: g GET 5
201 218: t TUPLE (MARK at 151)
202 219: p PUT 9
203 222: a APPEND
204 223: g GET 9
205 226: a APPEND
206 227: I INT 5
207 230: a APPEND
208 231: . STOP
209highest protocol among opcodes = 0
210"""
211
212DATA1 = (']q\x01(K\x00L1L\nG@\x00\x00\x00\x00\x00\x00\x00'
213 'c__builtin__\ncomplex\nq\x02(G@\x08\x00\x00\x00\x00\x00'
214 '\x00G\x00\x00\x00\x00\x00\x00\x00\x00tRq\x03K\x01J\xff\xff'
215 '\xff\xffK\xffJ\x01\xff\xff\xffJ\x00\xff\xff\xffM\xff\xff'
216 'J\x01\x00\xff\xffJ\x00\x00\xff\xffJ\xff\xff\xff\x7fJ\x01\x00'
217 '\x00\x80J\x00\x00\x00\x80(U\x03abcq\x04h\x04(c__main__\n'
218 'C\nq\x05oq\x06}q\x07(U\x03fooq\x08K\x01U\x03barq\tK\x02ubh'
219 '\x06tq\nh\nK\x05e.'
220 )
221
222# Disassembly of DATA1.
223DATA1_DIS = """\
224 0: ] EMPTY_LIST
225 1: q BINPUT 1
226 3: ( MARK
227 4: K BININT1 0
228 6: L LONG 1L
229 10: G BINFLOAT 2.0
230 19: c GLOBAL '__builtin__ complex'
231 40: q BINPUT 2
232 42: ( MARK
233 43: G BINFLOAT 3.0
234 52: G BINFLOAT 0.0
235 61: t TUPLE (MARK at 42)
236 62: R REDUCE
237 63: q BINPUT 3
238 65: K BININT1 1
239 67: J BININT -1
240 72: K BININT1 255
241 74: J BININT -255
242 79: J BININT -256
243 84: M BININT2 65535
244 87: J BININT -65535
245 92: J BININT -65536
246 97: J BININT 2147483647
247 102: J BININT -2147483647
248 107: J BININT -2147483648
249 112: ( MARK
250 113: U SHORT_BINSTRING 'abc'
251 118: q BINPUT 4
252 120: h BINGET 4
253 122: ( MARK
254 123: c GLOBAL '__main__ C'
255 135: q BINPUT 5
256 137: o OBJ (MARK at 122)
257 138: q BINPUT 6
258 140: } EMPTY_DICT
259 141: q BINPUT 7
260 143: ( MARK
261 144: U SHORT_BINSTRING 'foo'
262 149: q BINPUT 8
263 151: K BININT1 1
264 153: U SHORT_BINSTRING 'bar'
265 158: q BINPUT 9
266 160: K BININT1 2
267 162: u SETITEMS (MARK at 143)
268 163: b BUILD
269 164: h BINGET 6
270 166: t TUPLE (MARK at 112)
271 167: q BINPUT 10
272 169: h BINGET 10
273 171: K BININT1 5
274 173: e APPENDS (MARK at 3)
275 174: . STOP
276highest protocol among opcodes = 1
277"""
278
279DATA2 = ('\x80\x02]q\x01(K\x00\x8a\x01\x01G@\x00\x00\x00\x00\x00\x00\x00'
280 'c__builtin__\ncomplex\nq\x02G@\x08\x00\x00\x00\x00\x00\x00G\x00'
281 '\x00\x00\x00\x00\x00\x00\x00\x86Rq\x03K\x01J\xff\xff\xff\xffK'
282 '\xffJ\x01\xff\xff\xffJ\x00\xff\xff\xffM\xff\xffJ\x01\x00\xff\xff'
283 'J\x00\x00\xff\xffJ\xff\xff\xff\x7fJ\x01\x00\x00\x80J\x00\x00\x00'
284 '\x80(U\x03abcq\x04h\x04(c__main__\nC\nq\x05oq\x06}q\x07(U\x03foo'
285 'q\x08K\x01U\x03barq\tK\x02ubh\x06tq\nh\nK\x05e.')
286
287# Disassembly of DATA2.
288DATA2_DIS = """\
289 0: \x80 PROTO 2
290 2: ] EMPTY_LIST
291 3: q BINPUT 1
292 5: ( MARK
293 6: K BININT1 0
294 8: \x8a LONG1 1L
295 11: G BINFLOAT 2.0
296 20: c GLOBAL '__builtin__ complex'
297 41: q BINPUT 2
298 43: G BINFLOAT 3.0
299 52: G BINFLOAT 0.0
300 61: \x86 TUPLE2
301 62: R REDUCE
302 63: q BINPUT 3
303 65: K BININT1 1
304 67: J BININT -1
305 72: K BININT1 255
306 74: J BININT -255
307 79: J BININT -256
308 84: M BININT2 65535
309 87: J BININT -65535
310 92: J BININT -65536
311 97: J BININT 2147483647
312 102: J BININT -2147483647
313 107: J BININT -2147483648
314 112: ( MARK
315 113: U SHORT_BINSTRING 'abc'
316 118: q BINPUT 4
317 120: h BINGET 4
318 122: ( MARK
319 123: c GLOBAL '__main__ C'
320 135: q BINPUT 5
321 137: o OBJ (MARK at 122)
322 138: q BINPUT 6
323 140: } EMPTY_DICT
324 141: q BINPUT 7
325 143: ( MARK
326 144: U SHORT_BINSTRING 'foo'
327 149: q BINPUT 8
328 151: K BININT1 1
329 153: U SHORT_BINSTRING 'bar'
330 158: q BINPUT 9
331 160: K BININT1 2
332 162: u SETITEMS (MARK at 143)
333 163: b BUILD
334 164: h BINGET 6
335 166: t TUPLE (MARK at 112)
336 167: q BINPUT 10
337 169: h BINGET 10
338 171: K BININT1 5
339 173: e APPENDS (MARK at 5)
340 174: . STOP
341highest protocol among opcodes = 2
342"""
343
344def create_data():
345 c = C()
346 c.foo = 1
347 c.bar = 2
348 x = [0, 1L, 2.0, 3.0+0j]
349 # Append some integer test cases at cPickle.c's internal size
350 # cutoffs.
351 uint1max = 0xff
352 uint2max = 0xffff
353 int4max = 0x7fffffff
354 x.extend([1, -1,
355 uint1max, -uint1max, -uint1max-1,
356 uint2max, -uint2max, -uint2max-1,
357 int4max, -int4max, -int4max-1])
358 y = ('abc', 'abc', c, c)
359 x.append(y)
360 x.append(y)
361 x.append(5)
362 return x
363
364class AbstractPickleTests(unittest.TestCase):
365 # Subclass must define self.dumps, self.loads, self.error.
366
367 _testdata = create_data()
368
369 def setUp(self):
370 pass
371
372 def test_misc(self):
373 # test various datatypes not tested by testdata
374 for proto in protocols:
375 x = myint(4)
376 s = self.dumps(x, proto)
377 y = self.loads(s)
378 self.assertEqual(x, y)
379
380 x = (1, ())
381 s = self.dumps(x, proto)
382 y = self.loads(s)
383 self.assertEqual(x, y)
384
385 x = initarg(1, x)
386 s = self.dumps(x, proto)
387 y = self.loads(s)
388 self.assertEqual(x, y)
389
390 # XXX test __reduce__ protocol?
391
392 def test_roundtrip_equality(self):
393 expected = self._testdata
394 for proto in protocols:
395 s = self.dumps(expected, proto)
396 got = self.loads(s)
397 self.assertEqual(expected, got)
398
399 def test_load_from_canned_string(self):
400 expected = self._testdata
401 for canned in DATA0, DATA1, DATA2:
402 got = self.loads(canned)
403 self.assertEqual(expected, got)
404
405 # There are gratuitous differences between pickles produced by
406 # pickle and cPickle, largely because cPickle starts PUT indices at
407 # 1 and pickle starts them at 0. See XXX comment in cPickle's put2() --
408 # there's a comment with an exclamation point there whose meaning
409 # is a mystery. cPickle also suppresses PUT for objects with a refcount
410 # of 1.
411 def dont_test_disassembly(self):
412 from cStringIO import StringIO
413 from pickletools import dis
414
415 for proto, expected in (0, DATA0_DIS), (1, DATA1_DIS):
416 s = self.dumps(self._testdata, proto)
417 filelike = StringIO()
418 dis(s, out=filelike)
419 got = filelike.getvalue()
420 self.assertEqual(expected, got)
421
422 def test_recursive_list(self):
423 l = []
424 l.append(l)
425 for proto in protocols:
426 s = self.dumps(l, proto)
427 x = self.loads(s)
428 self.assertEqual(len(x), 1)
429 self.assert_(x is x[0])
430
431 def test_recursive_dict(self):
432 d = {}
433 d[1] = d
434 for proto in protocols:
435 s = self.dumps(d, proto)
436 x = self.loads(s)
437 self.assertEqual(x.keys(), [1])
438 self.assert_(x[1] is x)
439
440 def test_recursive_inst(self):
441 i = C()
442 i.attr = i
443 for proto in protocols:
444 s = self.dumps(i, 2)
445 x = self.loads(s)
446 self.assertEqual(dir(x), dir(i))
447 self.assert_(x.attr is x)
448
449 def test_recursive_multi(self):
450 l = []
451 d = {1:l}
452 i = C()
453 i.attr = d
454 l.append(i)
455 for proto in protocols:
456 s = self.dumps(l, proto)
457 x = self.loads(s)
458 self.assertEqual(len(x), 1)
459 self.assertEqual(dir(x[0]), dir(i))
460 self.assertEqual(x[0].attr.keys(), [1])
461 self.assert_(x[0].attr[1] is x)
462
463 def test_garyp(self):
464 self.assertRaises(self.error, self.loads, 'garyp')
465
466 def test_insecure_strings(self):
467 insecure = ["abc", "2 + 2", # not quoted
468 #"'abc' + 'def'", # not a single quoted string
469 "'abc", # quote is not closed
470 "'abc\"", # open quote and close quote don't match
471 "'abc' ?", # junk after close quote
472 "'\\'", # trailing backslash
473 # some tests of the quoting rules
474 #"'abc\"\''",
475 #"'\\\\a\'\'\'\\\'\\\\\''",
476 ]
477 for s in insecure:
478 buf = "S" + s + "\012p0\012."
479 self.assertRaises(ValueError, self.loads, buf)
480
481 if have_unicode:
482 def test_unicode(self):
483 endcases = [unicode(''), unicode('<\\u>'), unicode('<\\\u1234>'),
484 unicode('<\n>'), unicode('<\\>')]
485 for proto in protocols:
486 for u in endcases:
487 p = self.dumps(u, proto)
488 u2 = self.loads(p)
489 self.assertEqual(u2, u)
490
491 def test_ints(self):
492 import sys
493 for proto in protocols:
494 n = sys.maxint
495 while n:
496 for expected in (-n, n):
497 s = self.dumps(expected, proto)
498 n2 = self.loads(s)
499 self.assertEqual(expected, n2)
500 n = n >> 1
501
502 def test_maxint64(self):
503 maxint64 = (1L << 63) - 1
504 data = 'I' + str(maxint64) + '\n.'
505 got = self.loads(data)
506 self.assertEqual(got, maxint64)
507
508 # Try too with a bogus literal.
509 data = 'I' + str(maxint64) + 'JUNK\n.'
510 self.assertRaises(ValueError, self.loads, data)
511
512 def test_long(self):
513 for proto in protocols:
514 # 256 bytes is where LONG4 begins.
515 for nbits in 1, 8, 8*254, 8*255, 8*256, 8*257:
516 nbase = 1L << nbits
517 for npos in nbase-1, nbase, nbase+1:
518 for n in npos, -npos:
519 pickle = self.dumps(n, proto)
520 got = self.loads(pickle)
521 self.assertEqual(n, got)
522 # Try a monster. This is quadratic-time in protos 0 & 1, so don't
523 # bother with those.
524 nbase = long("deadbeeffeedface", 16)
525 nbase += nbase << 1000000