Use orjson instead of json for JSON serialization

[advance512]

Description

The authors of orjson describe it like this:

orjson is a fast, correct JSON library for Python. It benchmarks as the fastest Python library for JSON and is more correct than the standard json library or other third-party libraries.

Based on the benchmarks the authors have shared, orjson is 5 to 50 times faster than the standard json library.

Expected behaviour

Add an option to use orjson instead of json.

Actual behaviour

json is the only option for JSON serialization.

[advance512]

Comparing the standard Jsonifier used by Connexion and orjson in an example API call in my application, I see the difference in performance can be extreme, 570 times faster:

FlaskApi._serialize_data():

# 2434078 bytes, total time 8ms
body = orjson.dumps(data, option=orjson.OPT_NAIVE_UTC | orjson.OPT_NON_STR_KEYS)

# 3899087 bytes, total time 4565ms
body = cls.jsonifier.dumps(data)

To validate the output, I used json.loads() and verified that it creates identical objects. It does.

[backbord]

Please also include the option to use orjson for deserialisation.

[lucas03]

Isn't this possible by custom json encoder? In connexion 2.x

connexion_app.app.json_encoder = CustomJSONEncoder

and by jsonifier in connexion=3.x

jsonifier=Jsonifier(kw.json)

[houbie]

Jsonifier unfortunately always adds JSONEncoder to dumps calls: self.dumps_args.setdefault("cls", JSONEncoder) which is not allowed for orjson.

You can work around this by subclassing Jsonifier:

import orjson

class OrJsonifier(Jsonifier):
    """A Jsonifier that uses orjson for serialization and deserialization."""

    def dumps(self, data, **_kwargs):
        return orjson.dumps(data).decode()

connexion_app = AsyncApp(name, jsonifier=OrJsonifier())

Note that the original Jsonifier adds an (unnecessary) trailing newline.