I have a pickled structure consisting of nested builtin primitives (list, dictionaries) and instances of classes that are not in the project anymore, that therefore cause errors during unpickling. I do not really care about those objects, I wish I could extract numerical values stored in this nested structure. Is there any way to unpickle from a file and replace everything that was broken due to import issues with, let's say, None
?
I was trying to inherit from Unpickler
and override find_class(self, module, name)
to return Dummy
if class can not be found, but for some reason I keep getting TypeError: 'NoneType' object is not callable
in load reduce
after that.
class Dummy(object):
def __init__(self, *argv, **kwargs):
pass
I tried something like
class RobustJoblibUnpickle(Unpickler):
def find_class(self, _module, name):
try:
super(RobustJoblibUnpickle, self).find_class(_module, name)
except ImportError:
return Dummy
Maybe you can catch the exception in a try
block, and do what you want (set some object to use a Dummy class)?None
edit:
Take a look at this, I don't know if it is the right way to do it, but it seems to work fine:
import sys
import pickle
class Dummy:
pass
class MyUnpickler(pickle._Unpickler):
def find_class(self, module, name): # from the pickle module code but with a try
# Subclasses may override this. # we are doing it right now...
try:
if self.proto < 3 and self.fix_imports:
if (module, name) in _compat_pickle.NAME_MAPPING:
module, name = _compat_pickle.NAME_MAPPING[(module, name)]
elif module in _compat_pickle.IMPORT_MAPPING:
module = _compat_pickle.IMPORT_MAPPING[module]
__import__(module, level=0)
if self.proto >= 4:
return _getattribute(sys.modules[module], name)[0]
else:
return getattr(sys.modules[module], name)
except AttributeError:
return Dummy
# edit: as per Ben suggestion an even simpler subclass can be used
# instead of the above
class MyUnpickler2(pickle._Unpickler):
def find_class(self, module, name):
try:
return super().find_class(module, name)
except AttributeError:
return Dummy
class C:
pass
c1 = C()
with open('data1.dat', 'wb') as f:
pickle.dump(c1,f)
del C # simulate the missing class
with open('data1.dat', 'rb') as f:
unpickler = MyUnpickler(f) # or MyUnpickler2(f)
c1 = unpickler.load()
print(c1) # got a Dummy object because of missing class