A Python implementation of Extended JSON (EJSON) as used in Meteor and DDP.
pip install meteor-ejson
Supports Python 2.7, Python 3.x, and PyPy
The meteor-ejson Python package is open source under the MIT license.
This implementation piggybacks on the native json package by providing a EJSONEncoder and a EJSONDecoder class. Use either the ejson.dumps(o) and ejson.loads(s) shortcuts or the built-ins.
>>> import ejson
>>> import datetime
>>> ejson.dumps({"createdAt": datetime.date(2010, 10, 10)})
'{"createdAt": {"$date": 1286668800000}}'
>>>
>>> ejson.loads('{"createdAt": {"$date": 1286668800000}}')
{'createdAt': datetime.datetime(2010, 10, 10)}
Use with built-in JSON functions:
import json
import ejson
print(json.loads('{"$binary": "SGVsbG8gd29ybGQh"}', cls=ejson.EJSONDecoder))
| EJSON type | Python type |
|---|---|
| $date | datetime.datetime (and datetime.date when encoding) |
| $binary | str (Python 2.x; decoding only), bytes (Python 3.x) |
| $type/$value | user-specified |
The EJSON format allows for user-specified types:
{"$type": TYPENAME, "$value": VALUE}
These can be handled with the custom_type_hooks keyword. The encoder expects these hooks as a list of 3-tuples in the format (cls, type_name, encoder_callback) and the decoder expects a dict or a list of 2-tuples in the format (type_name, decoder_callback). For example:
>>> v = ejson.dumps(set(['a', 'b', 'c']), custom_type_hooks=[(set, 'set', list)])
>>> v
'{"$value": ["a", "c", "b"], "$type": "set"}'
>>> ejson.loads(v, custom_type_hooks=[('set', set)])
{'a', 'c', 'b'}
If the decoder encounters a user-specified type it cannot handle, it raises an ejson.UnknownTypeError(ValueError).