forked from uqfoundation/dill
-
Notifications
You must be signed in to change notification settings - Fork 1
/
_dill.py
2035 lines (1837 loc) · 80.4 KB
/
_dill.py
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
66
67
68
69
70
71
72
73
74
75
76
77
78
79
80
81
82
83
84
85
86
87
88
89
90
91
92
93
94
95
96
97
98
99
100
101
102
103
104
105
106
107
108
109
110
111
112
113
114
115
116
117
118
119
120
121
122
123
124
125
126
127
128
129
130
131
132
133
134
135
136
137
138
139
140
141
142
143
144
145
146
147
148
149
150
151
152
153
154
155
156
157
158
159
160
161
162
163
164
165
166
167
168
169
170
171
172
173
174
175
176
177
178
179
180
181
182
183
184
185
186
187
188
189
190
191
192
193
194
195
196
197
198
199
200
201
202
203
204
205
206
207
208
209
210
211
212
213
214
215
216
217
218
219
220
221
222
223
224
225
226
227
228
229
230
231
232
233
234
235
236
237
238
239
240
241
242
243
244
245
246
247
248
249
250
251
252
253
254
255
256
257
258
259
260
261
262
263
264
265
266
267
268
269
270
271
272
273
274
275
276
277
278
279
280
281
282
283
284
285
286
287
288
289
290
291
292
293
294
295
296
297
298
299
300
301
302
303
304
305
306
307
308
309
310
311
312
313
314
315
316
317
318
319
320
321
322
323
324
325
326
327
328
329
330
331
332
333
334
335
336
337
338
339
340
341
342
343
344
345
346
347
348
349
350
351
352
353
354
355
356
357
358
359
360
361
362
363
364
365
366
367
368
369
370
371
372
373
374
375
376
377
378
379
380
381
382
383
384
385
386
387
388
389
390
391
392
393
394
395
396
397
398
399
400
401
402
403
404
405
406
407
408
409
410
411
412
413
414
415
416
417
418
419
420
421
422
423
424
425
426
427
428
429
430
431
432
433
434
435
436
437
438
439
440
441
442
443
444
445
446
447
448
449
450
451
452
453
454
455
456
457
458
459
460
461
462
463
464
465
466
467
468
469
470
471
472
473
474
475
476
477
478
479
480
481
482
483
484
485
486
487
488
489
490
491
492
493
494
495
496
497
498
499
500
501
502
503
504
505
506
507
508
509
510
511
512
513
514
515
516
517
518
519
520
521
522
523
524
525
526
527
528
529
530
531
532
533
534
535
536
537
538
539
540
541
542
543
544
545
546
547
548
549
550
551
552
553
554
555
556
557
558
559
560
561
562
563
564
565
566
567
568
569
570
571
572
573
574
575
576
577
578
579
580
581
582
583
584
585
586
587
588
589
590
591
592
593
594
595
596
597
598
599
600
601
602
603
604
605
606
607
608
609
610
611
612
613
614
615
616
617
618
619
620
621
622
623
624
625
626
627
628
629
630
631
632
633
634
635
636
637
638
639
640
641
642
643
644
645
646
647
648
649
650
651
652
653
654
655
656
657
658
659
660
661
662
663
664
665
666
667
668
669
670
671
672
673
674
675
676
677
678
679
680
681
682
683
684
685
686
687
688
689
690
691
692
693
694
695
696
697
698
699
700
701
702
703
704
705
706
707
708
709
710
711
712
713
714
715
716
717
718
719
720
721
722
723
724
725
726
727
728
729
730
731
732
733
734
735
736
737
738
739
740
741
742
743
744
745
746
747
748
749
750
751
752
753
754
755
756
757
758
759
760
761
762
763
764
765
766
767
768
769
770
771
772
773
774
775
776
777
778
779
780
781
782
783
784
785
786
787
788
789
790
791
792
793
794
795
796
797
798
799
800
801
802
803
804
805
806
807
808
809
810
811
812
813
814
815
816
817
818
819
820
821
822
823
824
825
826
827
828
829
830
831
832
833
834
835
836
837
838
839
840
841
842
843
844
845
846
847
848
849
850
851
852
853
854
855
856
857
858
859
860
861
862
863
864
865
866
867
868
869
870
871
872
873
874
875
876
877
878
879
880
881
882
883
884
885
886
887
888
889
890
891
892
893
894
895
896
897
898
899
900
901
902
903
904
905
906
907
908
909
910
911
912
913
914
915
916
917
918
919
920
921
922
923
924
925
926
927
928
929
930
931
932
933
934
935
936
937
938
939
940
941
942
943
944
945
946
947
948
949
950
951
952
953
954
955
956
957
958
959
960
961
962
963
964
965
966
967
968
969
970
971
972
973
974
975
976
977
978
979
980
981
982
983
984
985
986
987
988
989
990
991
992
993
994
995
996
997
998
999
1000
# -*- coding: utf-8 -*-
#
# Author: Mike McKerns (mmckerns @caltech and @uqfoundation)
# Copyright (c) 2008-2015 California Institute of Technology.
# Copyright (c) 2016-2022 The Uncertainty Quantification Foundation.
# License: 3-clause BSD. The full license text is available at:
# - https://github.com/uqfoundation/dill/blob/master/LICENSE
"""
dill: a utility for serialization of python objects
Based on code written by Oren Tirosh and Armin Ronacher.
Extended to a (near) full set of the builtin types (in types module),
and coded to the pickle interface, by <mmckerns@caltech.edu>.
Initial port to python3 by Jonathan Dobson, continued by mmckerns.
Test against "all" python types (Std. Lib. CH 1-15 @ 2.7) by mmckerns.
Test against CH16+ Std. Lib. ... TBD.
"""
__all__ = [
'Pickler','Unpickler',
'check','copy','dump','dumps','load','loads','pickle','pickles','register',
'DEFAULT_PROTOCOL','HIGHEST_PROTOCOL','CONTENTS_FMODE','FILE_FMODE','HANDLE_FMODE',
'PickleError','PickleWarning','PicklingError','PicklingWarning','UnpicklingError',
'UnpicklingWarning',
]
__module__ = 'dill'
import warnings
from .logger import adapter as logger
from .logger import trace as _trace
import os
import sys
diff = None
_use_diff = False
OLD38 = (sys.hexversion < 0x3080000)
OLD39 = (sys.hexversion < 0x3090000)
OLD310 = (sys.hexversion < 0x30a0000)
#XXX: get types from .objtypes ?
import builtins as __builtin__
from pickle import _Pickler as StockPickler, Unpickler as StockUnpickler
from _thread import LockType
from _thread import RLock as RLockType
#from io import IOBase
from types import CodeType, FunctionType, MethodType, GeneratorType, \
TracebackType, FrameType, ModuleType, BuiltinMethodType
BufferType = memoryview #XXX: unregistered
ClassType = type # no 'old-style' classes
EllipsisType = type(Ellipsis)
#FileType = IOBase
NotImplementedType = type(NotImplemented)
SliceType = slice
TypeType = type # 'new-style' classes #XXX: unregistered
XRangeType = range
from types import MappingProxyType as DictProxyType
from pickle import DEFAULT_PROTOCOL, HIGHEST_PROTOCOL, PickleError, PicklingError, UnpicklingError
import __main__ as _main_module
import marshal
import gc
# import zlib
from weakref import ReferenceType, ProxyType, CallableProxyType
from collections import OrderedDict
from functools import partial
from operator import itemgetter, attrgetter
GENERATOR_FAIL = False
import importlib.machinery
EXTENSION_SUFFIXES = tuple(importlib.machinery.EXTENSION_SUFFIXES)
try:
import ctypes
HAS_CTYPES = True
# if using `pypy`, pythonapi is not found
IS_PYPY = not hasattr(ctypes, 'pythonapi')
except ImportError:
HAS_CTYPES = False
IS_PYPY = False
NumpyUfuncType = None
NumpyDType = None
NumpyArrayType = None
try:
if not importlib.machinery.PathFinder().find_spec('numpy'):
raise ImportError("No module named 'numpy'")
NumpyUfuncType = True
NumpyDType = True
NumpyArrayType = True
except ImportError:
pass
def __hook__():
global NumpyArrayType, NumpyDType, NumpyUfuncType
from numpy import ufunc as NumpyUfuncType
from numpy import ndarray as NumpyArrayType
from numpy import dtype as NumpyDType
return True
if NumpyArrayType: # then has numpy
def ndarraysubclassinstance(obj_type):
if all((c.__module__, c.__name__) != ('numpy', 'ndarray') for c in obj_type.__mro__):
return False
# anything below here is a numpy array (or subclass) instance
__hook__() # import numpy (so the following works!!!)
# verify that __reduce__ has not been overridden
if obj_type.__reduce_ex__ is not NumpyArrayType.__reduce_ex__ \
or obj_type.__reduce__ is not NumpyArrayType.__reduce__:
return False
return True
def numpyufunc(obj_type):
return any((c.__module__, c.__name__) == ('numpy', 'ufunc') for c in obj_type.__mro__)
def numpydtype(obj_type):
if all((c.__module__, c.__name__) != ('numpy', 'dtype') for c in obj_type.__mro__):
return False
# anything below here is a numpy dtype
__hook__() # import numpy (so the following works!!!)
return obj_type is type(NumpyDType) # handles subclasses
else:
def ndarraysubclassinstance(obj): return False
def numpyufunc(obj): return False
def numpydtype(obj): return False
from types import GetSetDescriptorType, ClassMethodDescriptorType, \
WrapperDescriptorType, MethodDescriptorType, MemberDescriptorType, \
MethodWrapperType #XXX: unused
# make sure to add these 'hand-built' types to _typemap
CellType = type((lambda x: lambda y: x)(0).__closure__[0])
PartialType = type(partial(int, base=2))
SuperType = type(super(Exception, TypeError()))
ItemGetterType = type(itemgetter(0))
AttrGetterType = type(attrgetter('__repr__'))
try:
from functools import _lru_cache_wrapper as LRUCacheType
except ImportError:
LRUCacheType = None
if not isinstance(LRUCacheType, type):
LRUCacheType = None
def get_file_type(*args, **kwargs):
open = kwargs.pop("open", __builtin__.open)
f = open(os.devnull, *args, **kwargs)
t = type(f)
f.close()
return t
FileType = get_file_type('rb', buffering=0)
TextWrapperType = get_file_type('r', buffering=-1)
BufferedRandomType = get_file_type('r+b', buffering=-1)
BufferedReaderType = get_file_type('rb', buffering=-1)
BufferedWriterType = get_file_type('wb', buffering=-1)
try:
from _pyio import open as _open
PyTextWrapperType = get_file_type('r', buffering=-1, open=_open)
PyBufferedRandomType = get_file_type('r+b', buffering=-1, open=_open)
PyBufferedReaderType = get_file_type('rb', buffering=-1, open=_open)
PyBufferedWriterType = get_file_type('wb', buffering=-1, open=_open)
except ImportError:
PyTextWrapperType = PyBufferedRandomType = PyBufferedReaderType = PyBufferedWriterType = None
from io import BytesIO as StringIO
InputType = OutputType = None
from socket import socket as SocketType
#FIXME: additionally calls ForkingPickler.register several times
from multiprocessing.reduction import _reduce_socket as reduce_socket
try:
IS_IPYTHON = __IPYTHON__ # is True
ExitType = None # IPython.core.autocall.ExitAutocall
singletontypes = ['exit', 'quit', 'get_ipython']
except NameError:
IS_IPYTHON = False
try: ExitType = type(exit) # apparently 'exit' can be removed
except NameError: ExitType = None
singletontypes = []
import inspect
import dataclasses
import typing
from pickle import GLOBAL
### Shims for different versions of Python and dill
class Sentinel(object):
"""
Create a unique sentinel object that is pickled as a constant.
"""
def __init__(self, name, module_name=None):
self.name = name
if module_name is None:
# Use the calling frame's module
self.__module__ = inspect.currentframe().f_back.f_globals['__name__']
else:
self.__module__ = module_name # pragma: no cover
def __repr__(self):
return self.__module__ + '.' + self.name # pragma: no cover
def __copy__(self):
return self # pragma: no cover
def __deepcopy__(self, memo):
return self # pragma: no cover
def __reduce__(self):
return self.name
def __reduce_ex__(self, protocol):
return self.name
from . import _shims
from ._shims import Reduce, Getattr
### File modes
#: Pickles the file handle, preserving mode. The position of the unpickled
#: object is as for a new file handle.
HANDLE_FMODE = 0
#: Pickles the file contents, creating a new file if on load the file does
#: not exist. The position = min(pickled position, EOF) and mode is chosen
#: as such that "best" preserves behavior of the original file.
CONTENTS_FMODE = 1
#: Pickles the entire file (handle and contents), preserving mode and position.
FILE_FMODE = 2
### Shorthands (modified from python2.5/lib/pickle.py)
def copy(obj, *args, **kwds):
"""
Use pickling to 'copy' an object (i.e. `loads(dumps(obj))`).
See :func:`dumps` and :func:`loads` for keyword arguments.
"""
ignore = kwds.pop('ignore', Unpickler.settings['ignore'])
return loads(dumps(obj, *args, **kwds), ignore=ignore)
def dump(obj, file, protocol=None, byref=None, fmode=None, recurse=None, **kwds):#, strictio=None):
"""
Pickle an object to a file.
See :func:`dumps` for keyword arguments.
"""
from .settings import settings
protocol = settings['protocol'] if protocol is None else int(protocol)
_kwds = kwds.copy()
_kwds.update(dict(byref=byref, fmode=fmode, recurse=recurse))
Pickler(file, protocol, **_kwds).dump(obj)
return
def dumps(obj, protocol=None, byref=None, fmode=None, recurse=None, **kwds):#, strictio=None):
"""
Pickle an object to a string.
*protocol* is the pickler protocol, as defined for Python *pickle*.
If *byref=True*, then dill behaves a lot more like pickle as certain
objects (like modules) are pickled by reference as opposed to attempting
to pickle the object itself.
If *recurse=True*, then objects referred to in the global dictionary
are recursively traced and pickled, instead of the default behavior
of attempting to store the entire global dictionary. This is needed for
functions defined via *exec()*.
*fmode* (:const:`HANDLE_FMODE`, :const:`CONTENTS_FMODE`,
or :const:`FILE_FMODE`) indicates how file handles will be pickled.
For example, when pickling a data file handle for transfer to a remote
compute service, *FILE_FMODE* will include the file contents in the
pickle and cursor position so that a remote method can operate
transparently on an object with an open file handle.
Default values for keyword arguments can be set in :mod:`dill.settings`.
"""
file = StringIO()
dump(obj, file, protocol, byref, fmode, recurse, **kwds)#, strictio)
return file.getvalue()
def load(file, ignore=None, **kwds):
"""
Unpickle an object from a file.
See :func:`loads` for keyword arguments.
"""
return Unpickler(file, ignore=ignore, **kwds).load()
def loads(str, ignore=None, **kwds):
"""
Unpickle an object from a string.
If *ignore=False* then objects whose class is defined in the module
*__main__* are updated to reference the existing class in *__main__*,
otherwise they are left to refer to the reconstructed type, which may
be different.
Default values for keyword arguments can be set in :mod:`dill.settings`.
"""
file = StringIO(str)
return load(file, ignore, **kwds)
# def dumpzs(obj, protocol=None):
# """pickle an object to a compressed string"""
# return zlib.compress(dumps(obj, protocol))
# def loadzs(str):
# """unpickle an object from a compressed string"""
# return loads(zlib.decompress(str))
### End: Shorthands ###
class MetaCatchingDict(dict):
def get(self, key, default=None):
try:
return self[key]
except KeyError:
return default
def __missing__(self, key):
if issubclass(key, type):
return save_type
else:
raise KeyError()
class PickleWarning(Warning, PickleError):
pass
class PicklingWarning(PickleWarning, PicklingError):
pass
class UnpicklingWarning(PickleWarning, UnpicklingError):
pass
### Extend the Picklers
class Pickler(StockPickler):
"""python's Pickler extended to interpreter sessions"""
dispatch = MetaCatchingDict(StockPickler.dispatch.copy())
_session = False
from .settings import settings
def __init__(self, file, *args, **kwds):
settings = Pickler.settings
_byref = kwds.pop('byref', None)
#_strictio = kwds.pop('strictio', None)
_fmode = kwds.pop('fmode', None)
_recurse = kwds.pop('recurse', None)
StockPickler.__init__(self, file, *args, **kwds)
self._main = _main_module
self._diff_cache = {}
self._byref = settings['byref'] if _byref is None else _byref
self._strictio = False #_strictio
self._fmode = settings['fmode'] if _fmode is None else _fmode
self._recurse = settings['recurse'] if _recurse is None else _recurse
self._postproc = OrderedDict()
self._file = file
def save(self, obj, save_persistent_id=True):
# register if the object is a numpy ufunc
# thanks to Paul Kienzle for pointing out ufuncs didn't pickle
obj_type = type(obj)
if NumpyArrayType and not (obj_type is type or obj_type in Pickler.dispatch):
if NumpyUfuncType and numpyufunc(obj_type):
@register(obj_type)
def save_numpy_ufunc(pickler, obj):
logger.trace(pickler, "Nu: %s", obj)
name = getattr(obj, '__qualname__', getattr(obj, '__name__', None))
StockPickler.save_global(pickler, obj, name=name)
logger.trace(pickler, "# Nu")
return
# NOTE: the above 'save' performs like:
# import copy_reg
# def udump(f): return f.__name__
# def uload(name): return getattr(numpy, name)
# copy_reg.pickle(NumpyUfuncType, udump, uload)
# register if the object is a numpy dtype
if NumpyDType and numpydtype(obj_type):
@register(obj_type)
def save_numpy_dtype(pickler, obj):
logger.trace(pickler, "Dt: %s", obj)
pickler.save_reduce(_create_dtypemeta, (obj.type,), obj=obj)
logger.trace(pickler, "# Dt")
return
# NOTE: the above 'save' performs like:
# import copy_reg
# def uload(name): return type(NumpyDType(name))
# def udump(f): return uload, (f.type,)
# copy_reg.pickle(NumpyDTypeType, udump, uload)
# register if the object is a subclassed numpy array instance
if NumpyArrayType and ndarraysubclassinstance(obj_type):
@register(obj_type)
def save_numpy_array(pickler, obj):
logger.trace(pickler, "Nu: (%s, %s)", obj.shape, obj.dtype)
npdict = getattr(obj, '__dict__', None)
f, args, state = obj.__reduce__()
pickler.save_reduce(_create_array, (f,args,state,npdict), obj=obj)
logger.trace(pickler, "# Nu")
return
# end hack
if GENERATOR_FAIL and type(obj) == GeneratorType:
msg = "Can't pickle %s: attribute lookup builtins.generator failed" % GeneratorType
raise PicklingError(msg)
StockPickler.save(self, obj, save_persistent_id)
save.__doc__ = StockPickler.save.__doc__
def dump(self, obj): #NOTE: if settings change, need to update attributes
logger.trace_setup(self)
StockPickler.dump(self, obj)
dump.__doc__ = StockPickler.dump.__doc__
class Unpickler(StockUnpickler):
"""python's Unpickler extended to interpreter sessions and more types"""
from .settings import settings
_session = False
def find_class(self, module, name):
if (module, name) == ('__builtin__', '__main__'):
return self._main.__dict__ #XXX: above set w/save_module_dict
elif (module, name) == ('__builtin__', 'NoneType'):
return type(None) #XXX: special case: NoneType missing
if module == 'dill.dill': module = 'dill._dill'
return StockUnpickler.find_class(self, module, name)
def __init__(self, *args, **kwds):
settings = Pickler.settings
_ignore = kwds.pop('ignore', None)
StockUnpickler.__init__(self, *args, **kwds)
self._main = _main_module
self._ignore = settings['ignore'] if _ignore is None else _ignore
def load(self): #NOTE: if settings change, need to update attributes
obj = StockUnpickler.load(self)
if type(obj).__module__ == getattr(_main_module, '__name__', '__main__'):
if not self._ignore:
# point obj class to main
try: obj.__class__ = getattr(self._main, type(obj).__name__)
except (AttributeError,TypeError): pass # defined in a file
#_main_module.__dict__.update(obj.__dict__) #XXX: should update globals ?
return obj
load.__doc__ = StockUnpickler.load.__doc__
pass
'''
def dispatch_table():
"""get the dispatch table of registered types"""
return Pickler.dispatch
'''
pickle_dispatch_copy = StockPickler.dispatch.copy()
def pickle(t, func):
"""expose dispatch table for user-created extensions"""
Pickler.dispatch[t] = func
return
def register(t):
"""register type to Pickler's dispatch table """
def proxy(func):
Pickler.dispatch[t] = func
return func
return proxy
def _revert_extension():
"""drop dill-registered types from pickle's dispatch table"""
for type, func in list(StockPickler.dispatch.items()):
if func.__module__ == __name__:
del StockPickler.dispatch[type]
if type in pickle_dispatch_copy:
StockPickler.dispatch[type] = pickle_dispatch_copy[type]
def use_diff(on=True):
"""
Reduces size of pickles by only including object which have changed.
Decreases pickle size but increases CPU time needed.
Also helps avoid some unpicklable objects.
MUST be called at start of script, otherwise changes will not be recorded.
"""
global _use_diff, diff
_use_diff = on
if _use_diff and diff is None:
try:
from . import diff as d
except ImportError:
import diff as d
diff = d
def _create_typemap():
import types
d = dict(list(__builtin__.__dict__.items()) + \
list(types.__dict__.items())).items()
for key, value in d:
if getattr(value, '__module__', None) == 'builtins' \
and type(value) is type:
yield key, value
return
_reverse_typemap = dict(_create_typemap())
_reverse_typemap.update({
'PartialType': PartialType,
'SuperType': SuperType,
'ItemGetterType': ItemGetterType,
'AttrGetterType': AttrGetterType,
})
if sys.hexversion < 0x30800a2:
_reverse_typemap.update({
'CellType': CellType,
})
# "Incidental" implementation specific types. Unpickling these types in another
# implementation of Python (PyPy -> CPython) is not guaranteed to work
# This dictionary should contain all types that appear in Python implementations
# but are not defined in https://docs.python.org/3/library/types.html#standard-interpreter-types
x=OrderedDict()
_incedental_reverse_typemap = {
'FileType': FileType,
'BufferedRandomType': BufferedRandomType,
'BufferedReaderType': BufferedReaderType,
'BufferedWriterType': BufferedWriterType,
'TextWrapperType': TextWrapperType,
'PyBufferedRandomType': PyBufferedRandomType,
'PyBufferedReaderType': PyBufferedReaderType,
'PyBufferedWriterType': PyBufferedWriterType,
'PyTextWrapperType': PyTextWrapperType,
}
_incedental_reverse_typemap.update({
"DictKeysType": type({}.keys()),
"DictValuesType": type({}.values()),
"DictItemsType": type({}.items()),
"OdictKeysType": type(x.keys()),
"OdictValuesType": type(x.values()),
"OdictItemsType": type(x.items()),
})
if ExitType:
_incedental_reverse_typemap['ExitType'] = ExitType
if InputType:
_incedental_reverse_typemap['InputType'] = InputType
_incedental_reverse_typemap['OutputType'] = OutputType
'''
try:
import symtable
_incedental_reverse_typemap["SymtableEntryType"] = type(symtable.symtable("", "string", "exec")._table)
except: #FIXME: fails to pickle
pass
if sys.hexversion >= 0x30a00a0:
_incedental_reverse_typemap['LineIteratorType'] = type(compile('3', '', 'eval').co_lines())
'''
if sys.hexversion >= 0x30b00b0:
from types import GenericAlias
_incedental_reverse_typemap["GenericAliasIteratorType"] = type(iter(GenericAlias(list, (int,))))
'''
_incedental_reverse_typemap['PositionsIteratorType'] = type(compile('3', '', 'eval').co_positions())
'''
try:
import winreg
_incedental_reverse_typemap["HKEYType"] = winreg.HKEYType
except ImportError:
pass
_reverse_typemap.update(_incedental_reverse_typemap)
_incedental_types = set(_incedental_reverse_typemap.values())
del x
_typemap = dict((v, k) for k, v in _reverse_typemap.items())
def _unmarshal(string):
return marshal.loads(string)
def _load_type(name):
return _reverse_typemap[name]
def _create_type(typeobj, *args):
return typeobj(*args)
def _create_function(fcode, fglobals, fname=None, fdefaults=None,
fclosure=None, fdict=None, fkwdefaults=None):
# same as FunctionType, but enable passing __dict__ to new function,
# __dict__ is the storehouse for attributes added after function creation
func = FunctionType(fcode, fglobals or dict(), fname, fdefaults, fclosure)
if fdict is not None:
func.__dict__.update(fdict) #XXX: better copy? option to copy?
if fkwdefaults is not None:
func.__kwdefaults__ = fkwdefaults
# 'recurse' only stores referenced modules/objects in fglobals,
# thus we need to make sure that we have __builtins__ as well
if "__builtins__" not in func.__globals__:
func.__globals__["__builtins__"] = globals()["__builtins__"]
# assert id(fglobals) == id(func.__globals__)
return func
class match:
"""
Make avaialable a limited structural pattern matching-like syntax for Python < 3.10
Patterns can be only tuples (without types) currently.
Inspired by the package pattern-matching-PEP634.
Usage:
>>> with match(args) as m:
>>> if m.case(('x', 'y')):
>>> # use m.x and m.y
>>> elif m.case(('x', 'y', 'z')):
>>> # use m.x, m.y and m.z
Equivalent native code for Python >= 3.10:
>>> match args:
>>> case (x, y):
>>> # use x and y
>>> case (x, y, z):
>>> # use x, y and z
"""
def __init__(self, value):
self.value = value
self._fields = None
def __enter__(self):
return self
def __exit__(self, *exc_info):
return False
def case(self, args): # *args, **kwargs):
"""just handles tuple patterns"""
if len(self.value) != len(args): # + len(kwargs):
return False
#if not all(isinstance(arg, pat) for arg, pat in zip(self.value[len(args):], kwargs.values())):
# return False
self.args = args # (*args, *kwargs)
return True
@property
def fields(self):
# Only bind names to values if necessary.
if self._fields is None:
self._fields = dict(zip(self.args, self.value))
return self._fields
def __getattr__(self, item):
return self.fields[item]
ALL_CODE_PARAMS = [
# Version New attribute CodeType parameters
((3,11,'a'), 'co_endlinetable', 'argcount posonlyargcount kwonlyargcount nlocals stacksize flags code consts names varnames filename name qualname firstlineno linetable endlinetable columntable exceptiontable freevars cellvars'),
((3,11), 'co_exceptiontable', 'argcount posonlyargcount kwonlyargcount nlocals stacksize flags code consts names varnames filename name qualname firstlineno linetable exceptiontable freevars cellvars'),
((3,10), 'co_linetable', 'argcount posonlyargcount kwonlyargcount nlocals stacksize flags code consts names varnames filename name firstlineno linetable freevars cellvars'),
((3,8), 'co_posonlyargcount', 'argcount posonlyargcount kwonlyargcount nlocals stacksize flags code consts names varnames filename name firstlineno lnotab freevars cellvars'),
((3,7), 'co_kwonlyargcount', 'argcount kwonlyargcount nlocals stacksize flags code consts names varnames filename name firstlineno lnotab freevars cellvars'),
]
for version, new_attr, params in ALL_CODE_PARAMS:
if hasattr(CodeType, new_attr):
CODE_VERSION = version
CODE_PARAMS = params.split()
break
ENCODE_PARAMS = set(CODE_PARAMS).intersection(
['code', 'lnotab', 'linetable', 'endlinetable', 'columntable', 'exceptiontable'])
def _create_code(*args):
if not isinstance(args[0], int): # co_lnotab stored from >= 3.10
LNOTAB, *args = args
else: # from < 3.10 (or pre-LNOTAB storage)
LNOTAB = b''
with match(args) as m:
# Python 3.11/3.12a (18 members)
if m.case((
'argcount', 'posonlyargcount', 'kwonlyargcount', 'nlocals', 'stacksize', 'flags', # args[0:6]
'code', 'consts', 'names', 'varnames', 'filename', 'name', 'qualname', 'firstlineno', # args[6:14]
'linetable', 'exceptiontable', 'freevars', 'cellvars' # args[14:]
)):
if CODE_VERSION == (3,11):
return CodeType(
*args[:6],
args[6].encode() if hasattr(args[6], 'encode') else args[6], # code
*args[7:14],
args[14].encode() if hasattr(args[14], 'encode') else args[14], # linetable
args[15].encode() if hasattr(args[15], 'encode') else args[15], # exceptiontable
args[16],
args[17],
)
fields = m.fields
# Python 3.10 or 3.8/3.9 (16 members)
elif m.case((
'argcount', 'posonlyargcount', 'kwonlyargcount', 'nlocals', 'stacksize', 'flags', # args[0:6]
'code', 'consts', 'names', 'varnames', 'filename', 'name', 'firstlineno', # args[6:13]
'LNOTAB_OR_LINETABLE', 'freevars', 'cellvars' # args[13:]
)):
if CODE_VERSION == (3,10) or CODE_VERSION == (3,8):
return CodeType(
*args[:6],
args[6].encode() if hasattr(args[6], 'encode') else args[6], # code
*args[7:13],
args[13].encode() if hasattr(args[13], 'encode') else args[13], # lnotab/linetable
args[14],
args[15],
)
fields = m.fields
if CODE_VERSION >= (3,10):
fields['linetable'] = m.LNOTAB_OR_LINETABLE
else:
fields['lnotab'] = LNOTAB if LNOTAB else m.LNOTAB_OR_LINETABLE
# Python 3.7 (15 args)
elif m.case((
'argcount', 'kwonlyargcount', 'nlocals', 'stacksize', 'flags', # args[0:5]
'code', 'consts', 'names', 'varnames', 'filename', 'name', 'firstlineno', # args[5:12]
'lnotab', 'freevars', 'cellvars' # args[12:]
)):
if CODE_VERSION == (3,7):
return CodeType(
*args[:5],
args[5].encode() if hasattr(args[5], 'encode') else args[5], # code
*args[6:12],
args[12].encode() if hasattr(args[12], 'encode') else args[12], # lnotab
args[13],
args[14],
)
fields = m.fields
# Python 3.11a (20 members)
elif m.case((
'argcount', 'posonlyargcount', 'kwonlyargcount', 'nlocals', 'stacksize', 'flags', # args[0:6]
'code', 'consts', 'names', 'varnames', 'filename', 'name', 'qualname', 'firstlineno', # args[6:14]
'linetable', 'endlinetable', 'columntable', 'exceptiontable', 'freevars', 'cellvars' # args[14:]
)):
if CODE_VERSION == (3,11,'a'):
return CodeType(
*args[:6],
args[6].encode() if hasattr(args[6], 'encode') else args[6], # code
*args[7:14],
*(a.encode() if hasattr(a, 'encode') else a for a in args[14:18]), # linetable-exceptiontable
args[18],
args[19],
)
fields = m.fields
else:
raise UnpicklingError("pattern match for code object failed")
# The args format doesn't match this version.
fields.setdefault('posonlyargcount', 0) # from python <= 3.7
fields.setdefault('lnotab', LNOTAB) # from python >= 3.10
fields.setdefault('linetable', b'') # from python <= 3.9
fields.setdefault('qualname', fields['name']) # from python <= 3.10
fields.setdefault('exceptiontable', b'') # from python <= 3.10
fields.setdefault('endlinetable', None) # from python != 3.11a
fields.setdefault('columntable', None) # from python != 3.11a
args = (fields[k].encode() if k in ENCODE_PARAMS and hasattr(fields[k], 'encode') else fields[k]
for k in CODE_PARAMS)
return CodeType(*args)
def _create_ftype(ftypeobj, func, args, kwds):
if kwds is None:
kwds = {}
if args is None:
args = ()
return ftypeobj(func, *args, **kwds)
def _create_typing_tuple(argz, *args): #NOTE: workaround python/cpython#94245
if not argz:
return typing.Tuple[()].copy_with(())
if argz == ((),):
return typing.Tuple[()]
return typing.Tuple[argz]
def _create_lock(locked, *args): #XXX: ignores 'blocking'
from threading import Lock
lock = Lock()
if locked:
if not lock.acquire(False):
raise UnpicklingError("Cannot acquire lock")
return lock
def _create_rlock(count, owner, *args): #XXX: ignores 'blocking'
lock = RLockType()
if owner is not None:
lock._acquire_restore((count, owner))
if owner and not lock._is_owned():
raise UnpicklingError("Cannot acquire lock")
return lock
# thanks to matsjoyce for adding all the different file modes
def _create_filehandle(name, mode, position, closed, open, strictio, fmode, fdata): # buffering=0
# only pickles the handle, not the file contents... good? or StringIO(data)?
# (for file contents see: http://effbot.org/librarybook/copy-reg.htm)
# NOTE: handle special cases first (are there more special cases?)
names = {'<stdin>':sys.__stdin__, '<stdout>':sys.__stdout__,
'<stderr>':sys.__stderr__} #XXX: better fileno=(0,1,2) ?
if name in list(names.keys()):
f = names[name] #XXX: safer "f=sys.stdin"
elif name == '<tmpfile>':
f = os.tmpfile()
elif name == '<fdopen>':
import tempfile
f = tempfile.TemporaryFile(mode)
else:
try:
exists = os.path.exists(name)
except Exception:
exists = False
if not exists:
if strictio:
raise FileNotFoundError("[Errno 2] No such file or directory: '%s'" % name)
elif "r" in mode and fmode != FILE_FMODE:
name = '<fdopen>' # or os.devnull?
current_size = 0 # or maintain position?
else:
current_size = os.path.getsize(name)
if position > current_size:
if strictio:
raise ValueError("invalid buffer size")
elif fmode == CONTENTS_FMODE:
position = current_size
# try to open the file by name
# NOTE: has different fileno
try:
#FIXME: missing: *buffering*, encoding, softspace
if fmode == FILE_FMODE:
f = open(name, mode if "w" in mode else "w")
f.write(fdata)
if "w" not in mode:
f.close()
f = open(name, mode)
elif name == '<fdopen>': # file did not exist
import tempfile
f = tempfile.TemporaryFile(mode)
# treat x mode as w mode
elif fmode == CONTENTS_FMODE \
and ("w" in mode or "x" in mode):
# stop truncation when opening
flags = os.O_CREAT
if "+" in mode:
flags |= os.O_RDWR
else:
flags |= os.O_WRONLY
f = os.fdopen(os.open(name, flags), mode)
# set name to the correct value
r = getattr(f, "buffer", f)
r = getattr(r, "raw", r)
r.name = name
assert f.name == name
else:
f = open(name, mode)
except (IOError, FileNotFoundError):
err = sys.exc_info()[1]
raise UnpicklingError(err)
if closed:
f.close()
elif position >= 0 and fmode != HANDLE_FMODE:
f.seek(position)
return f
def _create_stringi(value, position, closed):
f = StringIO(value)
if closed: f.close()
else: f.seek(position)
return f
def _create_stringo(value, position, closed):
f = StringIO()
if closed: f.close()
else:
f.write(value)
f.seek(position)
return f
class _itemgetter_helper(object):
def __init__(self):
self.items = []
def __getitem__(self, item):
self.items.append(item)
return
class _attrgetter_helper(object):
def __init__(self, attrs, index=None):
self.attrs = attrs
self.index = index
def __getattribute__(self, attr):
attrs = object.__getattribute__(self, "attrs")
index = object.__getattribute__(self, "index")
if index is None:
index = len(attrs)
attrs.append(attr)
else:
attrs[index] = ".".join([attrs[index], attr])
return type(self)(attrs, index)
class _dictproxy_helper(dict):
def __ror__(self, a):
return a
_dictproxy_helper_instance = _dictproxy_helper()
__d = {}
try:
# In CPython 3.9 and later, this trick can be used to exploit the
# implementation of the __or__ function of MappingProxyType to get the true
# mapping referenced by the proxy. It may work for other implementations,
# but is not guaranteed.
MAPPING_PROXY_TRICK = __d is (DictProxyType(__d) | _dictproxy_helper_instance)
except Exception:
MAPPING_PROXY_TRICK = False
del __d
# _CELL_REF and _CELL_EMPTY are used to stay compatible with versions of dill
# whose _create_cell functions do not have a default value.
# _CELL_REF can be safely removed entirely (replaced by empty tuples for calls
# to _create_cell) once breaking changes are allowed.
_CELL_REF = None
_CELL_EMPTY = Sentinel('_CELL_EMPTY')
def _create_cell(contents=None):
if contents is not _CELL_EMPTY:
value = contents
return (lambda: value).__closure__[0]
def _create_weakref(obj, *args):
from weakref import ref
if obj is None: # it's dead
from collections import UserDict
return ref(UserDict(), *args)
return ref(obj, *args)
def _create_weakproxy(obj, callable=False, *args):
from weakref import proxy
if obj is None: # it's dead
if callable: return proxy(lambda x:x, *args)
from collections import UserDict
return proxy(UserDict(), *args)
return proxy(obj, *args)
def _eval_repr(repr_str):
return eval(repr_str)
def _create_array(f, args, state, npdict=None):
#array = numpy.core.multiarray._reconstruct(*args)
array = f(*args)
array.__setstate__(state)
if npdict is not None: # we also have saved state in __dict__
array.__dict__.update(npdict)
return array
def _create_dtypemeta(scalar_type):
if NumpyDType is True: __hook__() # a bit hacky I think
if scalar_type is None:
return NumpyDType
return type(NumpyDType(scalar_type))
def _create_namedtuple(name, fieldnames, modulename, defaults=None):
class_ = _import_module(modulename + '.' + name, safe=True)
if class_ is not None:
return class_
import collections
t = collections.namedtuple(name, fieldnames, defaults=defaults, module=modulename)
return t
def _create_capsule(pointer, name, context, destructor):
attr_found = False
try:
# based on https://github.com/python/cpython/blob/f4095e53ab708d95e019c909d5928502775ba68f/Objects/capsule.c#L209-L231
uname = name.decode('utf8')
for i in range(1, uname.count('.')+1):
names = uname.rsplit('.', i)
try:
module = __import__(names[0])
except ImportError:
pass
obj = module
for attr in names[1:]:
obj = getattr(obj, attr)
capsule = obj
attr_found = True
break
except Exception:
pass
if attr_found:
if _PyCapsule_IsValid(capsule, name):
return capsule
raise UnpicklingError("%s object exists at %s but a PyCapsule object was expected." % (type(capsule), name))
else:
warnings.warn('Creating a new PyCapsule %s for a C data structure that may not be present in memory. Segmentation faults or other memory errors are possible.' % (name,), UnpicklingWarning)
capsule = _PyCapsule_New(pointer, name, destructor)
_PyCapsule_SetContext(capsule, context)
return capsule
def _getattr(objclass, name, repr_str):
# hack to grab the reference directly
try: #XXX: works only for __builtin__ ?
attr = repr_str.split("'")[3]
return eval(attr+'.__dict__["'+name+'"]')
except Exception:
try:
attr = objclass.__dict__
if type(attr) is DictProxyType:
attr = attr[name]
else:
attr = getattr(objclass,name)
except (AttributeError, KeyError):
attr = getattr(objclass,name)
return attr
def _get_attr(self, name):
# stop recursive pickling
return getattr(self, name, None) or getattr(__builtin__, name)
def _import_module(import_name, safe=False):
try:
if import_name.startswith('__runtime__.'):
return sys.modules[import_name]
elif '.' in import_name:
items = import_name.split('.')