orderedattrdict 1.2

Usage

AttrDict behaves exactly like collections.OrderedDict, but also allows keys to be accessed as attributes:

>>> from orderedattrdict import AttrDict
>>> conf = AttrDict()
>>> conf['z'] = 1
>>> assert conf.z == 1
>>> conf.y = 2
>>> assert conf['y'] == 2
>>> conf.x = 3
>>> assert conf.keys() == ['z', 'y', 'x']

NOTE: If the key clashes with an OrderedDict attribute or starts with __ (two underscores), you can’t access it as an attribute. For example:

>>> a = AttrDict(keys=1)
>>> a.keys
<bound method AttrDict.keys of AttrDict([('keys', 1)])>
>>> a['keys']
1

Load JSON preserving the order of keys:

>>> import json
>>> data = json.load(open('test.json'), object_pairs_hook=AttrDict)

Load YAML preserving the order of keys:

>>> import yaml
>>> from orderedattrdict.yamlutils import AttrDictYAMLLoader
>>> data = yaml.load(open('test.yaml'), Loader=AttrDictYAMLLoader)

Make PyYAML always load all dictionaries as AttrDict:

>>> from orderedattrdict.yamlutils import from_yaml
>>> yaml.add_constructor(u'tag:yaml.org,2002:map', from_yaml)
>>> yaml.add_constructor(u'tag:yaml.org,2002:omap', from_yaml)

json.dump, yaml.dump and yaml.safe_dump convert AttrDict into dictionaries, retaining the order:

>>> json.dumps(data)
>>> yaml.dump(data)

You can subclass AttrDict with mixins for more functionality. Set the __exclude_keys__ to exclude certain keys from attribute access. For example, to implement a Counter that uses AttrDict, create this class:

>>> from collections import Counter
>>> class CounterAttrDict(AttrDict, Counter):
>>>     def __init__(self, *args, **kwargs):
>>>         AttrDict.__init__(self, *args, **kwargs)
>>>         Counter.__init__(self)
>>>         self.__exclude_keys__ |= {'most_common', 'elements', 'subtract'}

This CounterAttrDict acts like a counter but with ordered keys via attribute access. However, .most_common, .elements and .subtract will not be used as keys:

>>> c = CounterAttrDict()
>>> c.x
0
>>> c.elements
<bound method CounterAttrDict.elements of CounterAttrDict()>
>>> c.x += 1
>>> c.y += 2
>>> c.most_common()
[('y', 2), ('x', 1)]
>>> list(c.elements())
['x', 'y', 'y']
>>> c.subtract(y=1)
>>> c
CounterAttrDict([('x', 1), ('y', 1)])

To create a defaultdict that is ordered and has attribute access, subclass from AttrDict and defaultdict:

>>> from collections import defaultdict
>>> class DefaultAttrDict(AttrDict, defaultdict):
>>>     def __init__(self, default_factory, *args, **kwargs):
>>>         AttrDict.__init__(self, *args, **kwargs)
>>>         defaultdict.__init__(self, default_factory)
>>>         self.__exclude_keys__ |= {'default_factory', '_ipython_display_'}

This can be used with a list factory:

>>> d.x
[]
>>> d.y.append(10)
>>> d
DefaultAttrDict([('x', []), ('y', [10])])

You can create a tree structure where you can set attributes in any level of the hierarchy:

>>> tree = lambda: DefaultAttrDict(tree)
>>> node = tree()
>>> node.x.y.z = 1
>>> node
DefaultAttrDict([('x', DefaultAttrDict([('y', DefaultAttrDict([('z', 1)]))]))])

Installation

This is a pure-Python package is build for Python 2.7+ and Python 3.0+. Set it up using:

pip install orderedattdict

Changelog

• 1.0: Basic implementation

• 1.1: Add utilities to load and save as YAML

• 1.2: Allow specific keys to be excluded from attribute access