orjson

orjson is a fast JSON library for Python. It benchmarks as the fastest Python library for JSON. Its serialization performance is 2x to 3x the nearest other library and 4.5x to 11.5x the standard library. Its deserialization performance is 1.05x to 1.2x the nearest other library and 1.2x to 4x the standard library.

It supports CPython 3.5, 3.6, and 3.7. Its API is a subset of the API of the standard library's json module.

Usage

Install

To install a manylinux wheel from PyPI:

pip install --upgrade orjson

To build a release wheel from source, assuming a Rust nightly toolchain and Python environment:

git clone --recurse-submodules https://github.com/ijl/orjson.git && cd orjson
pip install --upgrade pyo3-pack
pyo3-pack build --release --strip --interpreter python3.7

There is no runtime dependency other than a manylinux environment (i.e., deploying this does not require Rust or non-libc type libraries.)

Serialize

def dumps(obj: Any, default=Optional[Callable[Any]]) -> bytes: ...

dumps() serializes Python objects to JSON. It natively serializes str, dict, list, tuple, int, float, and None instances. It supports arbitrary types through default.

It raises JSONEncodeError on an unsupported type. This exception message describes the invalid object.

It raises JSONEncodeError on a str that contains invalid UTF-8.

It raises JSONEncodeError on an integer that exceeds 64 bits. This is the same as the standard library's json module.

It raises JSONEncodeError if a dict has a key of a type other than str.

It raises JSONEncodeError if the output of default recurses to handling by default more than five levels deep.

JSONEncodeError is a subclass of TypeError.

import orjson

try:
    val = orjson.dumps(...)
except orjson.JSONEncodeError:
    raise

To serialize arbitrary types, specify default as a callable that returns a supported type. default may be a function, lambda, or callable class instance.

>>> import orjson, numpy
>>> def default(obj):
        if isinstance(obj, numpy.ndarray):
            return obj.tolist()
>>> orjson.dumps(numpy.random.rand(2, 2), default=default)
b'[[0.08423896597867486,0.854121264944197],[0.8452845446981371,0.19227780743524303]]'

If the default callable does not return an object, and an exception was raised within the default function, an error describing this is returned. If no object is returned by the default callable but also no exception was raised, it falls through to raising JSONEncodeError on an unsupported type.

The default callable may return an object that itself must be handled by default up to five levels deep before an exception is raised.

Deserialize

def loads(obj: Union[bytes, str]) -> Union[dict, list, int, float, str, None]: ...

loads() deserializes JSON to Python objects.

It raises orjson.JSONDecodeError if given an invalid type or invalid JSON. This exception is a subclass of ValueError.

import orjson

try:
    val = orjson.loads(...)
except orjson.JSONDecodeError:
    raise

Comparison

There are slight differences in output between libraries. The differences are not an issue for interoperability. Note orjson returns bytes. Its output is slightly smaller as well.

>>> import orjson, ujson, rapidjson, json
>>> data = {'bool': True, '🐈':'哈哈', 'int': 9223372036854775807, 'float': 1.337e+40}
>>> orjson.dumps(data)
b'{"bool":true,"\xf0\x9f\x90\x88":"\xe5\x93\x88\xe5\x93\x88","int":9223372036854775807,"float":1.337e40}'
>>> ujson.dumps(data)
'{"bool":true,"\\ud83d\\udc08":"\\u54c8\\u54c8","int":9223372036854775807,"float":1.337000000000000e+40}'
>>> rapidjson.dumps(data)
'{"bool":true,"\\uD83D\\uDC08":"\\u54C8\\u54C8","int":9223372036854775807,"float":1.337e+40}'
>>> json.dumps(data)
'{"bool": true, "\\ud83d\\udc08": "\\u54c8\\u54c8", "int": 9223372036854775807, "float": 1.337e+40}'

UTF-8

orjson raises an exception on invalid UTF-8. This is necessary because Python 3 str objects may contain UTF-16 surrogates. The standard library's json module accepts invalid UTF-8.

>>> import orjson, ujson, rapidjson, json
>>> orjson.dumps('\ud800')
JSONEncodeError: str is not valid UTF-8: surrogates not allowed
>>> ujson.dumps('\ud800')
UnicodeEncodeError: 'utf-8' codec ...
>>> rapidjson.dumps('\ud800')
UnicodeEncodeError: 'utf-8' codec ...
>>> json.dumps('\ud800')
'"\\ud800"'

>>> import orjson, ujson, rapidjson, json
>>> orjson.loads('"\\ud800"')
JSONDecodeError: unexpected end of hex escape at line 1 column 8: line 1 column 1 (char 0)
>>> ujson.loads('"\\ud800"')
''
>>> rapidjson.loads('"\\ud800"')
ValueError: Parse error at offset 1: The surrogate pair in string is invalid.
>>> json.loads('"\\ud800"')
'\ud800'

Testing

The library has comprehensive tests. There are unit tests against the roundtrip, jsonchecker, and fixtures files of the nativejson-benchmark repository. It is tested to not crash against the Big List of Naughty Strings. It is tested to not leak memory. It is tested to be correct against input from the PyJFuzz JSON fuzzer. It is tested to not crash against and not accept invalid UTF-8. There are integration tests exercising the library's use in web servers (uwsgi and gunicorn, using multiprocess/forked workers) and when multithreaded. It also uses some tests from the ultrajson library.

Performance

Serialization and deserialization performance of orjson is better than ultrajson, rapidjson, or json. The benchmarks are done on fixtures of real data:

• twitter.json, 631.5KiB, results of a search on Twitter for "一", containing CJK strings, dictionaries of strings and arrays of dictionaries, indented.

• github.json, 55.8KiB, a GitHub activity feed, containing dictionaries of strings and arrays of dictionaries, not indented.

• citm_catalog.json, 1.7MiB, concert data, containing nested dictionaries of strings and arrays of integers, indented.

• canada.json, 2.2MiB, coordinates of the Canadian border in GeoJSON format, containing floats and arrays, indented.

twitter.json serialization

Library	Median latency (milliseconds)	Operations per second	Relative (latency)
orjson	0.48	2077.6	1
ujson	1.48	664.6	3.09
rapidjson	1.59	626.5	3.32
json	2.24	443.9	4.68

twitter.json deserialization

Library	Median latency (milliseconds)	Operations per second	Relative (latency)
orjson	2.38	418.8	1
ujson	2.67	373	1.12
rapidjson	2.78	359.5	1.16
json	2.77	359.7	1.16

github.json serialization

Library	Median latency (milliseconds)	Operations per second	Relative (latency)
orjson	0.06	17745	1
ujson	0.14	7107.1	2.49
rapidjson	0.16	6253.9	2.86
json	0.25	3972.5	4.49

github.json deserialization

Library	Median latency (milliseconds)	Operations per second	Relative (latency)
orjson	0.2	4929.7	1
ujson	0.22	4605.2	1.08
rapidjson	0.24	4166.5	1.19
json	0.24	4150.8	1.19

citm_catalog.json serialization

Library	Median latency (milliseconds)	Operations per second	Relative (latency)
orjson	0.76	1302	1
ujson	2.58	387.2	3.38
rapidjson	2.37	421.1	3.11
json	5.41	184.4	7.09

citm_catalog.json deserialization

Library	Median latency (milliseconds)	Operations per second	Relative (latency)
orjson	4.28	233.1	1
ujson	5.06	197.2	1.18
rapidjson	5.82	171.7	1.36
json	5.81	171.8	1.36

canada.json serialization

Library	Median latency (milliseconds)	Operations per second	Relative (latency)
orjson	4.04	247.7	1
ujson	8.43	118.6	2.09
rapidjson	43.93	22.7	10.88
json	47.23	21.1	11.7

canada.json deserialization

Library	Median latency (milliseconds)	Operations per second	Relative (latency)
orjson	6.69	147.6	1
ujson	7.17	139.4	1.07
rapidjson	26.77	37.4	4
json	26.59	37.6	3.97

This was measured using orjson 1.3.0 on Python 3.7.2 and Linux.

The results can be reproduced using the pybench and graph scripts.

License

orjson is dual licensed under the Apache 2.0 and MIT licenses. It contains tests from the hyperjson and ultrajson libraries. It is implemented using the serde_json and pyo3 libraries.