json-five

JSON5 for Python

Installation

pip install json-five

This project requires Python 3.8+

Key features

Supports the JSON5 spec
Supports similar interfaces to stdlib json module
Provides an API for working with abstract model representations of JSON5 documents.
Supports round-trip loading, editing, and dumping, preserving non-data elements such as comments (in model-based load/dump)

Usage

NOTE: the import name is json5 which differs from the install name.

For basic loading/dumping, the interface is nearly identical to that of the json module.

import json5
json_text = """{ // This is a JSON5 comment
"foo": "bar", /* this is a JSON5 block
comment that can span lines */
bacon: "eggs"  // unquoted Identifiers also work
}
"""
print(json5.loads(json_text))
# {"foo": "bar", "bacon": "eggs"}

with open('myfile.json5') as f:
    data = json5.load(f)

For loading JSON5, the same parameters object_hook, object_pairs_hook and parse_* keyword arguments are available here for load/loads.

Additionally, a new hook, parse_json5_identifiers, is available to help users control the output of parsing identifiers. By default, JSON5 Identifiers in object keys are returned as a JsonIdentifier object, which is a subclass of str (meaning it's compatible anywhere str is accepted). This helps keep keys the same round-trip, rather than converting unquoted identifiers into strings:

>>> text = '{bacon: "eggs"}'
>>> json5.dumps(json5.loads(text)) == text
True

You can change this behavior with the parse_json5_identifiers argument with a callable that receives the JsonIdentifier object and its return value is used instead. For example, you can specify parse_json5_identifiers=str to convert identifiers to strings.

>>> json5.dumps(json5.loads(text, parse_json5_identifiers=str))
'{"bacon": "eggs"}'

Custom loaders; Abstract JSON5 Models

Note: the underlying model API and tokens are not stable and are subject to breaking changes, even in minor releases.

json-five also features an API for representing JSON5 as an abstract model. This enables a wide degree of capabilities for various use-cases, such as linters, formatters, custom serialization/deserialization, and more.

Example: a simple model

from json5.loader import loads, ModelLoader
json_string = """{foo: "bar"}"""
model = loads(json_string, loader=ModelLoader())

The resulting model object looks something like this:

JSONText(
    value=JSONObject(
        keys=[Identifier(name="foo", raw_value="foo")],
        values=[DoubleQuotedString(characters="bar", raw_value='"bar"')],
        trailing_comma=None,
    )
)

It is possible to make edits to the model, which will affect the output when dumped using the model dumper. However, there is (currently) no validation to ensure your model edits won't result in invalid JSON5 when dumped.

You may also implement custom loaders and dumpers to control serialization and deserialization. See the full documentation for more information.

Tokenization

You can also leverage tokenization of JSON5:

from json5.tokenizer import tokenize

json_string = """{foo: "bar"}"""
for tok in tokenize(json_string):
    print(tok.type)

Output would be:

LBRACE
NAME
COLON
WHITESPACE
DOUBLE_QUOTE_STRING
RBRACE

Status

This project currently fully supports the JSON5 spec and its interfaces for loading and dumping JSON5 is stable as of v1.0.0. There is still active development underway, particularly for the underlying abstract JSON5 model representations and ability to perform edits using the abstract model.