What's Python JSON serializable?

If you are asking about vanilla Python, serialization could be done this way:

import json


class Person:
   def __init__(self, name, age, city):
      self.name = name
      self.age = age
      self.city = city

   def to_json(self):
       return json.dumps(self, default=lambda o: o.__dict__)


person = Person("John", 30, "New York")
print(person.to_json())

So we're just converting an object to a dict using __dict__ attribute.

But if you need something more sophisticated, you might need DRF (Django REST Framework) or pydantic. An example how it could be done with DRF:

from rest_framework import serializers
from rest_framework.serializers import Serializer


class Person:
   def __init__(self, name, age, city):
      self.name = name
      self.age = age
      self.city = city


class PersonSerializer(Serializer):
    name = serializers.CharField()
    age = serializers.IntegerField()
    city = serializers.CharField()

    def create(self, validated_data):
        return Person(**validated_data)


person = Person("John", 30, "New York")
print(PersonSerializer(person).data)

This way you have a much better control over it. See docs.

Yes, the json module only knows how to serialize certain built-in types. An easy way to work with classes that have "fields" is the dataclasses module.

Example:

from dataclasses import dataclass, asdict
import json

@dataclass
class Person:
    name: str
    age: int
    city: str

person = Person("John", 30, "New York")
print(json.dumps(asdict(person)))

The asdict() function converts the dataclass instance into a dict which json can serialize.

With Python 3.6+ you can use dataclasses in Python along with the asdict helper function to convert a dataclass instance to a Python dict object.

Note: For 3.6, you'll need a backport for the dataclasses module, as it only became a builtin to Python starting in 3.7.

import json
from dataclasses import dataclass, asdict


@dataclass
class Person:
    name: str
    age: int
    city: str


person = Person("John", 30, "New York")
person_dict = asdict(person)

json_string = json.dumps(person_dict, indent=2)

print(json_string)

Out:

{
  "name": "John",
  "age": 30,
  "city": "New York"
}

If you have a more complex use case, or end up with needing to (de)serialize a nested dataclass model, I'd check out an external library like the dataclass-wizard that supports such a use case in particular.