How to find looped references during serialization and make it possible to deserialize objects back?

Question

Python

Hello!
I came across a seemingly simple, but still incomprehensible puzzle for me. You need to pack various objects into a json and unpack them back on the other side when deserializing. Usually simple objects or classes are sent and this is not a problem at all, but objects with looped references to each other began to appear, for example

class Foo:
    def __init__(self):
        self.arg = None
  

class Bar:
    def __init__(self):
        self.baz = None
        
    
foo = Foo()
bar = Bar()

foo.arg = bar
bar.baz = foo

And if it is quite possible to pack in json, for example

import json


class CircRefEncoder(json.JSONEncoder):
    def __init__(self, *args, **argv):
        super().__init__(*args, **argv)
        self.proc_ids_map = {}
        
    def default(self, obj):
        if isinstance(obj, (Foo, Bar)):
            object_id = id(obj)
            if object_id in self.proc_ids_map:
                return {"pyref": object_id}  # short circle the object dumping
            self.proc_ids_map[object_id] = obj
            return {"$object_id": object_id, **obj.__dict__}
        return obj

print(json.dumps(foo, cls=CircRefEncoder, check_circular=False))

then I’ll never know what steps should be on the other side and how to understand the order for deserialization in general?

(Let's imagine that the one who packs and the one who unpacks are different processes and the second one simply creates objects according to the simple principle `types.SimpleNamespace(**all keys from the json, except for the object id)`

Answer 1

when exporting, you replace links with identifiers,
you export everything as a flat list
, when you import, after creating objects, you restore links from identifiers

Answer 2

pickle + base64 Throw
the received value into any json field.
Well, you and inventors on problems, of course, write some kind of encoders-decoders, horror.