API

jsonutils

JSON related utilities.

This module provides a few things:

1) A handy function for getting an object down to something that can be JSON serialized. See to_primitive().

2) Wrappers around loads() and dumps(). The dumps() wrapper will automatically use to_primitive() for you if needed.

3) This sets up anyjson to use the loads() and dumps() wrappers if anyjson is available.

oslo_serialization.jsonutils.dump(obj, fp, *args, **kwargs)

Serialize obj as a JSON formatted stream to fp

Parameters:

• obj – object to be serialized
• fp – a .write()-supporting file-like object
• default – function that returns a serializable version of an object
• args – extra arguments, please see documentation of json.dump
• kwargs –

  extra named parameters, please see documentation of json.dump

oslo_serialization.jsonutils.dumps(obj, default=<function to_primitive at 0x28c6758>, **kwargs)

Serialize obj to a JSON formatted str.

Parameters:

• obj – object to be serialized
• default – function that returns a serializable version of an object
• kwargs – extra named parameters, please see documentation of json.dumps

Returns:

json formatted string

oslo_serialization.jsonutils.load(fp, encoding='utf-8', **kwargs)

Deserialize fp to a Python object.

Parameters:

• fp – a .read() -supporting file-like object
• encoding – encoding used to interpret the string
• kwargs – extra named parameters, please see documentation of json.loads

Returns:

python object

oslo_serialization.jsonutils.loads(s, encoding='utf-8', **kwargs)

Deserialize s (a str or unicode instance containing a JSON

Parameters:

• s – string to deserialize
• encoding – encoding used to interpret the string
• kwargs –

  extra named parameters, please see documentation of json.loads

Returns:

python object

oslo_serialization.jsonutils.to_primitive(value, convert_instances=False, convert_datetime=True, level=0, max_depth=3)

Convert a complex object into primitives.

Handy for JSON serialization. We can optionally handle instances, but since this is a recursive function, we could have cyclical data structures.

To handle cyclical data structures we could track the actual objects visited in a set, but not all objects are hashable. Instead we just track the depth of the object inspections and don’t go too deep.

Therefore, convert_instances=True is lossy ... be aware.