You cannot tell the default class pickler to ignore something, no.
jsonpickle
does support the pickle
module __getstate__
and __setstate__
methods. If your classes implement those two methods, whatever is returned is then used by jsonpickle
to represent the state instead. Both methods do need to be implemented.
If __getstate__
is not implemented, jsonpickle
uses the __dict__
attribute instead, so your own version merely needs to use that same dictionary, remove the _sa_instance_state
key and you are done:
def __getstate__(self):
state = self.__dict__.copy()
del state['_sa_instance_state']
return state
def __setstate__(self, state):
self.__dict__.update(state)
Whatever __getstate__
returns will be processed further, recursively, there is no need to worry about handling subobjects there.
If adding __getstate__
and __setstate__
is not an option, you can also register a custom serialization handler for your class; the disadvantage is that while __getstate__
can get away with just returning a dictionary, a custom handler will need to return a fully flattened value.
__getstate__
and__setstate__
works like a charm for now, thank you. Instead of writing same lines of code for each model, implementing a decorator to remove ignored fields from__dict__
might be a good idea for the future. – Sharpen