This commit was manufactured by cvs2svn to create tag 'cnrisync'.
[python/dscho.git] / Lib / pickle.py
blob9ebe03aa9c2a921467660c6fd72eb7653a24f20e
1 """\
2 Pickling Algorithm
3 ------------------
5 This module implements a basic but powerful algorithm for "pickling" (a.k.a.
6 serializing, marshalling or flattening) nearly arbitrary Python objects.
7 This is a more primitive notion than persistency -- although pickle
8 reads and writes file objects, it does not handle the issue of naming
9 persistent objects, nor the (even more complicated) area of concurrent
10 access to persistent objects. The pickle module can transform a complex
11 object into a byte stream and it can transform the byte stream into
12 an object with the same internal structure. The most obvious thing to
13 do with these byte streams is to write them onto a file, but it is also
14 conceivable to send them across a network or store them in a database.
16 Unlike the built-in marshal module, pickle handles the following correctly:
18 - recursive objects
19 - pointer sharing
20 - classes and class instances
22 Pickle is Python-specific. This has the advantage that there are no
23 restrictions imposed by external standards such as CORBA (which probably
24 can't represent pointer sharing or recursive objects); however it means
25 that non-Python programs may not be able to reconstruct pickled Python
26 objects.
28 Pickle uses a printable ASCII representation. This is slightly more
29 voluminous than a binary representation. However, small integers actually
30 take *less* space when represented as minimal-size decimal strings than
31 when represented as 32-bit binary numbers, and strings are only much longer
32 if they contain control characters or 8-bit characters. The big advantage
33 of using printable ASCII (and of some other characteristics of pickle's
34 representation) is that for debugging or recovery purposes it is possible
35 for a human to read the pickled file with a standard text editor. (I could
36 have gone a step further and used a notation like S-expressions, but the
37 parser would have been considerably more complicated and slower, and the
38 files would probably have become much larger.)
40 Pickle doesn't handle code objects, which marshal does.
41 I suppose pickle could, and maybe it should, but there's probably no
42 great need for it right now (as long as marshal continues to be used
43 for reading and writing code objects), and at least this avoids
44 the possibility of smuggling Trojan horses into a program.
46 For the benefit of persistency modules written using pickle, it supports
47 the notion of a reference to an object outside the pickled data stream.
48 Such objects are referenced by a name, which is an arbitrary string of
49 printable ASCII characters. The resolution of such names is not defined
50 by the pickle module -- the persistent object module will have to implement
51 a method "persistent_load". To write references to persistent objects,
52 the persistent module must define a method "persistent_id" which returns
53 either None or the persistent ID of the object.
55 There are some restrictions on the pickling of class instances.
57 First of all, the class must be defined at the top level in a module.
59 Next, it must normally be possible to create class instances by calling
60 the class without arguments. If this is undesirable, the class can
61 define a method __getinitargs__ (XXX not a pretty name!), which should
62 return a *tuple* containing the arguments to be passed to the class
63 constructor.
65 Classes can influence how their instances are pickled -- if the class defines
66 the method __getstate__, it is called and the return state is pickled
67 as the contents for the instance, and if the class defines the
68 method __setstate__, it is called with the unpickled state. (Note
69 that these methods can also be used to implement copying class instances.)
70 If there is no __getstate__ method, the instance's __dict__
71 is pickled. If there is no __setstate__ method, the pickled object
72 must be a dictionary and its items are assigned to the new instance's
73 dictionary. (If a class defines both __getstate__ and __setstate__,
74 the state object needn't be a dictionary -- these methods can do what they
75 want.)
77 Note that when class instances are pickled, their class's code and data
78 is not pickled along with them. Only the instance data is pickled.
79 This is done on purpose, so you can fix bugs in a class or add methods and
80 still load objects that were created with an earlier version of the
81 class. If you plan to have long-lived objects that will see many versions
82 of a class, it may be worth to put a version number in the objects so
83 that suitable conversions can be made by the class's __setstate__ method.
85 The interface is as follows:
87 To pickle an object x onto a file f, open for writing:
89 p = pickle.Pickler(f)
90 p.dump(x)
92 To unpickle an object x from a file f, open for reading:
94 u = pickle.Unpickler(f)
95 x = u.load()
97 The Pickler class only calls the method f.write with a string argument
98 (XXX possibly the interface should pass f.write instead of f).
99 The Unpickler calls the methods f.read(with an integer argument)
100 and f.readline(without argument), both returning a string.
101 It is explicitly allowed to pass non-file objects here, as long as they
102 have the right methods.
104 The following types can be pickled:
106 - None
107 - integers, long integers, floating point numbers
108 - strings
109 - tuples, lists and dictionaries containing only picklable objects
110 - class instances whose __dict__ or __setstate__() is picklable
111 - classes
113 Attempts to pickle unpicklable objects will raise an exception
114 after having written an unspecified number of bytes to the file argument.
116 It is possible to make multiple calls to Pickler.dump() or to
117 Unpickler.load(), as long as there is a one-to-one correspondence
118 between pickler and Unpickler objects and between dump and load calls
119 for any pair of corresponding Pickler and Unpicklers. WARNING: this
120 is intended for pickleing multiple objects without intervening modifications
121 to the objects or their parts. If you modify an object and then pickle
122 it again using the same Pickler instance, the object is not pickled
123 again -- a reference to it is pickled and the Unpickler will return
124 the old value, not the modified one. (XXX There are two problems here:
125 (a) detecting changes, and (b) marshalling a minimal set of changes.
126 I have no answers. Garbage Collection may also become a problem here.)
129 __version__ = "1.5" # Code version
131 from types import *
132 import string
134 format_version = "1.1" # File format version we write
135 compatible_formats = ["1.0"] # Old format versions we can read
137 PicklingError = "pickle.PicklingError"
139 AtomicTypes = [NoneType, IntType, FloatType, StringType]
141 def safe(object):
142 t = type(object)
143 if t in AtomicTypes:
144 return 1
145 if t is TupleType:
146 for item in object:
147 if not safe(item): return 0
148 return 1
149 return 0
151 MARK = '('
152 POP = '0'
153 DUP = '2'
154 STOP = '.'
155 TUPLE = 't'
156 LIST = 'l'
157 DICT = 'd'
158 INST = 'i'
159 CLASS = 'c'
160 GET = 'g'
161 PUT = 'p'
162 APPEND = 'a'
163 SETITEM = 's'
164 BUILD = 'b'
165 NONE = 'N'
166 INT = 'I'
167 LONG = 'L'
168 FLOAT = 'F'
169 STRING = 'S'
170 PERSID = 'P'
171 AtomicKeys = [NONE, INT, LONG, FLOAT, STRING]
172 AtomicMap = {
173 NoneType: NONE,
174 IntType: INT,
175 LongType: LONG,
176 FloatType: FLOAT,
177 StringType: STRING,
180 class Pickler:
182 def __init__(self, file):
183 self.write = file.write
184 self.memo = {}
186 def dump(self, object):
187 self.save(object)
188 self.write(STOP)
190 def save(self, object):
191 pid = self.persistent_id(object)
192 if pid:
193 self.write(PERSID + str(pid) + '\n')
194 return
195 d = id(object)
196 if self.memo.has_key(d):
197 self.write(GET + `d` + '\n')
198 return
199 t = type(object)
200 try:
201 f = self.dispatch[t]
202 except KeyError:
203 raise PicklingError, \
204 "can't pickle %s objects" % `t.__name__`
205 f(self, object)
207 def persistent_id(self, object):
208 return None
210 dispatch = {}
212 def save_none(self, object):
213 self.write(NONE)
214 dispatch[NoneType] = save_none
216 def save_int(self, object):
217 self.write(INT + `object` + '\n')
218 dispatch[IntType] = save_int
220 def save_long(self, object):
221 self.write(LONG + `object` + '\n')
222 dispatch[LongType] = save_long
224 def save_float(self, object):
225 self.write(FLOAT + `object` + '\n')
226 dispatch[FloatType] = save_float
228 def save_string(self, object):
229 d = id(object)
230 self.write(STRING + `object` + '\n')
231 self.write(PUT + `d` + '\n')
232 self.memo[d] = object
233 dispatch[StringType] = save_string
235 def save_tuple(self, object):
236 d = id(object)
237 self.write(MARK)
238 n = len(object)
239 for k in range(n):
240 self.save(object[k])
241 if self.memo.has_key(d):
242 # Saving object[k] has saved us!
243 while k >= 0:
244 self.write(POP)
245 k = k-1
246 self.write(GET + `d` + '\n')
247 break
248 else:
249 self.write(TUPLE + PUT + `d` + '\n')
250 self.memo[d] = object
251 dispatch[TupleType] = save_tuple
253 def save_list(self, object):
254 d = id(object)
255 self.write(MARK)
256 n = len(object)
257 for k in range(n):
258 item = object[k]
259 if not safe(item):
260 break
261 self.save(item)
262 else:
263 k = n
264 self.write(LIST + PUT + `d` + '\n')
265 self.memo[d] = object
266 for k in range(k, n):
267 item = object[k]
268 self.save(item)
269 self.write(APPEND)
270 dispatch[ListType] = save_list
272 def save_dict(self, object):
273 d = id(object)
274 self.write(MARK)
275 items = object.items()
276 n = len(items)
277 for k in range(n):
278 key, value = items[k]
279 if not safe(key) or not safe(value):
280 break
281 self.save(key)
282 self.save(value)
283 else:
284 k = n
285 self.write(DICT + PUT + `d` + '\n')
286 self.memo[d] = object
287 for k in range(k, n):
288 key, value = items[k]
289 self.save(key)
290 self.save(value)
291 self.write(SETITEM)
292 dispatch[DictionaryType] = save_dict
294 def save_inst(self, object):
295 d = id(object)
296 cls = object.__class__
297 module = whichmodule(cls)
298 name = cls.__name__
299 if hasattr(object, '__getinitargs__'):
300 args = object.__getinitargs__()
301 len(args) # XXX Assert it's a sequence
302 else:
303 args = ()
304 self.write(MARK)
305 for arg in args:
306 self.save(arg)
307 self.write(INST + module + '\n' + name + '\n' +
308 PUT + `d` + '\n')
309 self.memo[d] = object
310 try:
311 getstate = object.__getstate__
312 except AttributeError:
313 stuff = object.__dict__
314 else:
315 stuff = getstate()
316 self.save(stuff)
317 self.write(BUILD)
318 dispatch[InstanceType] = save_inst
320 def save_class(self, object):
321 d = id(object)
322 module = whichmodule(object)
323 name = object.__name__
324 self.write(CLASS + module + '\n' + name + '\n' +
325 PUT + `d` + '\n')
326 dispatch[ClassType] = save_class
329 classmap = {}
331 def whichmodule(cls):
332 """Figure out the module in which a class occurs.
334 Search sys.modules for the module.
335 Cache in classmap.
336 Return a module name.
337 If the class cannot be found, return __main__.
339 if classmap.has_key(cls):
340 return classmap[cls]
341 import sys
342 clsname = cls.__name__
343 for name, module in sys.modules.items():
344 if name != '__main__' and \
345 hasattr(module, clsname) and \
346 getattr(module, clsname) is cls:
347 break
348 else:
349 name = '__main__'
350 classmap[cls] = name
351 return name
354 class Unpickler:
356 def __init__(self, file):
357 self.readline = file.readline
358 self.read = file.read
359 self.memo = {}
361 def load(self):
362 self.mark = ['spam'] # Any new unique object
363 self.stack = []
364 try:
365 while 1:
366 key = self.read(1)
367 self.dispatch[key](self)
368 except STOP, value:
369 return value
371 def marker(self):
372 k = len(self.stack)-1
373 while self.stack[k] != self.mark: k = k-1
374 return k
376 dispatch = {}
378 def load_eof(self):
379 raise EOFError
380 dispatch[''] = load_eof
382 def load_persid(self):
383 pid = self.readline()[:-1]
384 self.stack.append(self.persistent_load(pid))
385 dispatch[PERSID] = load_persid
387 def load_none(self):
388 self.stack.append(None)
389 dispatch[NONE] = load_none
391 def load_atomic(self):
392 self.stack.append(eval(self.readline()[:-1]))
393 dispatch[INT] = load_atomic
394 dispatch[LONG] = load_atomic
395 dispatch[FLOAT] = load_atomic
396 dispatch[STRING] = load_atomic
398 def load_tuple(self):
399 k = self.marker()
400 self.stack[k:] = [tuple(self.stack[k+1:])]
401 dispatch[TUPLE] = load_tuple
403 def load_list(self):
404 k = self.marker()
405 self.stack[k:] = [self.stack[k+1:]]
406 dispatch[LIST] = load_list
408 def load_dict(self):
409 k = self.marker()
410 d = {}
411 items = self.stack[k+1:]
412 for i in range(0, len(items), 2):
413 key = items[i]
414 value = items[i+1]
415 d[key] = value
416 self.stack[k:] = [d]
417 dispatch[DICT] = load_dict
419 def load_inst(self):
420 k = self.marker()
421 args = tuple(self.stack[k+1:])
422 del self.stack[k:]
423 module = self.readline()[:-1]
424 name = self.readline()[:-1]
425 klass = self.find_class(module, name)
426 value = apply(klass, args)
427 self.stack.append(value)
428 dispatch[INST] = load_inst
430 def load_class(self):
431 module = self.readline()[:-1]
432 name = self.readline()[:-1]
433 klass = self.find_class(module, name)
434 self.stack.append(klass)
435 return klass
436 dispatch[CLASS] = load_class
438 def find_class(self, module, name):
439 env = {}
440 try:
441 exec 'from %s import %s' % (module, name) in env
442 except ImportError:
443 raise SystemError, \
444 "Failed to import class %s from module %s" % \
445 (name, module)
446 klass = env[name]
447 if type(klass) != ClassType:
448 raise SystemError, \
449 "Imported object %s from module %s is not a class" % \
450 (name, module)
451 return klass
453 def load_pop(self):
454 del self.stack[-1]
455 dispatch[POP] = load_pop
457 def load_dup(self):
458 stack.append(stack[-1])
459 dispatch[DUP] = load_dup
461 def load_get(self):
462 self.stack.append(self.memo[string.atoi(self.readline()[:-1])])
463 dispatch[GET] = load_get
465 def load_put(self):
466 self.memo[string.atoi(self.readline()[:-1])] = self.stack[-1]
467 dispatch[PUT] = load_put
469 def load_append(self):
470 value = self.stack[-1]
471 del self.stack[-1]
472 list = self.stack[-1]
473 list.append(value)
474 dispatch[APPEND] = load_append
476 def load_setitem(self):
477 value = self.stack[-1]
478 key = self.stack[-2]
479 del self.stack[-2:]
480 dict = self.stack[-1]
481 dict[key] = value
482 dispatch[SETITEM] = load_setitem
484 def load_build(self):
485 value = self.stack[-1]
486 del self.stack[-1]
487 inst = self.stack[-1]
488 try:
489 setstate = inst.__setstate__
490 except AttributeError:
491 for key in value.keys():
492 inst.__dict__[key] = value[key]
493 else:
494 setstate(value)
495 dispatch[BUILD] = load_build
497 def load_mark(self):
498 self.stack.append(self.mark)
499 dispatch[MARK] = load_mark
501 def load_stop(self):
502 value = self.stack[-1]
503 del self.stack[-1]
504 raise STOP, value
505 dispatch[STOP] = load_stop
508 # Shorthands
510 def dump(object, file):
511 Pickler(file).dump(object)
513 def dumps(object):
514 import StringIO
515 file = StringIO.StringIO()
516 Pickler(file).dump(object)
517 return file.getvalue()
519 def load(file):
520 return Unpickler(file).load()
522 def loads(str):
523 import StringIO
524 file = StringIO.StringIO(str)
525 return Unpickler(file).load()
528 # The rest is used for testing only
530 class C:
531 def __cmp__(self, other):
532 return cmp(self.__dict__, other.__dict__)
534 def test():
535 fn = 'pickle_tmp'
536 c = C()
537 c.foo = 1
538 c.bar = 2
539 x = [0,1,2,3]
540 y = ('abc', 'abc', c, c)
541 x.append(y)
542 x.append(y)
543 x.append(5)
544 f = open(fn, 'w')
545 F = Pickler(f)
546 F.dump(x)
547 f.close()
548 f = open(fn, 'r')
549 U = Unpickler(f)
550 x2 = U.load()
551 print x
552 print x2
553 print x == x2
554 print map(id, x)
555 print map(id, x2)
556 print F.memo
557 print U.memo
559 if __name__ == '__main__':
560 test()