Validation / Coersion

class MyObject(Object):
    amethyst_verifyclass = False  # don't check json for class name
    foo = Attr(int)               # coerce to int
    bar = Attr(isa=str).strip()   # ensure str then strip whitespace

Serialization

JSON text can be produced and loaded, even for nested objects.

json_string = myobj.toJSON()
myobj2 = MyObject.newFromJSON(json_string)

Other serialization libraries can easily be used as well.

yaml_string = yaml.dump(myobj.deflate_data())
myobj2 = MyObject.inflate_new(yaml.safe_load(yaml_string))

By default the JSON serializer injects type hints to ensure that objects are de-serialized into the correct class:

# print(MyObject(foo=23, bar="plugh").toJSON())
{"__class__": "__mymodule.MyObject__", "foo": 23, "bar": "plugh"}

When building an object from JSON, the constructor will look for these hints and raise a ValueError if the type hint is missing or imported into the wrong class.

# These raise ValueError
MyObject.newFromJSON('{"foo":23, "bar":"plugh"}')
MyObject.newFromJSON('{"__class__": "__mymodule.MyOtherObject__", "foo":23}')

Class verification can be skipped by passing verifyclass=False to the loader.

myobj = MyObject.newFromJSON('{"foo":23, "bar":"plugh"}', verifyclass=False)

If you want no munging or class verification at all, set the class parameters:

class MyObject(Object):
    amethyst_includeclass  = False
    amethyst_verifyclass   = False

    foo = Attr(int)
    bar = Attr(isa=str).strip()

# No extra class info due to modified defaults:
myobj = MyObject.newFromJSON('{"foo":"23", "bar":"plugh"}')
print(myobj.toJSON())
# => { "foo": 23, "bar": "plugh" }