Skip to main content

A data structure which allows both object attributes and dictionary keys and values to be used simultaneously and interchangeably.

Project description




CleverDict is a hybrid Python data class which allows both object.attribute and dictionary['key'] notation to be used simultaneously and interchangeably. It's particularly handy when your code is mainly object-orientated but you want a 'DRY' and extensible way to import data in json/dictionary format into your objects... or vice versa... without having to write extra code just to handle the translation.

CleverDict also calls a .save method whenever an attribute or dictionary value is created or changed. Initially the .savemethod does nothing, but you can overwrite it with your own function to do useful things automatically every time an attribute changes in future e.g. pickle, encode, save data to a file or database, upload to the cloud etc. No more slavishly writing a call your own "save" function every... single... time...


No dependencies. Very lightweight:

pip install cleverdict

or to cover all bases...

python -m pip install cleverdict --upgrade --user


CleverDict objects behave like normal Python dictionaries, but with the convenience of immediately offering read and write access to their data (keys and values) using the object.attribute syntax, which many people find easier to type and more intuitive to read and understand.

You can create a CleverDict object in exactly the same way as a regular Python dictionary:

>>> from cleverdict import CleverDict
>>> x = CleverDict({'total':6, 'usergroup': "Knights of Ni"})

>>> x['total']
>>> x.usergroup
'Knights of Ni'
>>> x['usergroup']
'Knights of Ni'

The values are then immediately available using either dictionary or .attribute syntax:

>>> x['life'] = 42
>>> += 1
>>> x['life']

>>> del x['life']
# KeyError: 'life'


You can also create a CleverDict instance using keyword arguments like this:

>>> x = CleverDict(created = "today", review = "tomorrow")

>>> x.created
>>> x['review']

Or using a list of tuple/list pairs:

>>> x = CleverDict().fromlist([(1, "one"), [2, "two"], (3, "three")])

>>> x._1
>>> x._2
>>> x._3_

This can be helpful for serialisation issues like ```json.dumps()``` converting numeric dictionary keys to strings, and for use with Client/Server apps where there are limits on what object can be passed between the Client and Server (*).

Or using the .fromkeys() method like this:

>>> x = CleverDict.fromkeys(["Abigail", "Tino", "Isaac"], "Year 9")

>>> x
CleverDict({'Abigail': 'Year 9', 'Tino': 'Year 9', 'Isaac': 'Year 9'}, _aliases={}, _vars={})

Or by using vars() to import another object's data (but not its methods):

>>> class X: pass
>>> a = X(); = "Percival"
>>> x = CleverDict(vars(a))

>>> x
CleverDict({'name': 'Percival'}, _aliases={}, _vars={})

(*) You can use the .tolist() method to generate a list of key/value pairs from a CleverDict object:

>>> x = CleverDict({1: "one", 2: "two"})

>>> x.tolist()
[(1, 'one'), (2, 'two')]


By default CleverDict tries to find valid attribute names for dictionary keys which would otherwise fail. This includes keywords, null strings, most punctuation marks, and keys starting with a numeral. So for example 7 (integer) becomes "_7" (string):

>>> x = CleverDict({7: "Seven"})

>>> x._7
>>> x
CleverDict({7: 'Seven'}, _aliases={'_7': 7}, _vars={})

CleverDict keeps the original dictionary keys and values unchanged and remembers any normalised attribute names as aliases in ._alias. You can add or delete further aliases with .add_alias and .delete_alias, but the original key will never be deleted, even if all aliases and .attributes are removed:

>>> x.add_alias(7, "NumberSeven")

>>> x
CleverDict({7: 'Seven'}, _aliases={'_7': 7, 'NumberSeven': 7}, _vars={})

>>> x.delete_alias(["_7","NumberSeven"])

>>> x
CleverDict({7: 'Seven'}, _aliases={}, _vars={})
>>> x._7
AttributeError: '_7'


Did you know that the dictionary keys 0, 0.0, and False are considered the same in Python? Likewise 1, 1.0, and True, and 1234 and 1234.0? If you create a regular dictionary using more than one of these different identities, they'll appear to 'overwrite' each other, keeping the first Key specified but the last Value specified, reading left to right:

>>> x = {1: "one", True: "the truth"}

>>> x
{1: 'the truth'}

You'll be relieved to know CleverDict handles these cases but we thought it was worth mentioning in case you came across them first and wondered what the heck was going on! "Explicit is better than implicit", right? You can inspect all the key names and .attribute aliases using the .info() method, as well as any aliases for the object itself:

>>> x = y = z = CleverDict({1: "one", True: "the truth"})

x is y is z
x[1] == x['_1'] == x['_True'] == x._1 == x._True == 'the truth'

# Use 'as_str=True' to return the results as a string

"CleverDict:\n    x is y is z\n    x[1] == x['_1'] == x['_True'] == x._1 == x._True == 'the truth'"

Did you also know that since PEP3131 many unicode letters are valid in attribute names? CleverDict handles this and replaces all remaining invalid characters such as punctuation marks with "_" on a first come, first served basis. This can result in a KeyError, which you can get round by renaming the offending dictionary keys. For example:

>>> x = CleverDict({"one-two": "hypen",
                    "one/two": "forward slash"})
KeyError: "'one_two' already an alias for 'one-two'"

>>> x = CleverDict({"one-two": "hypen",
                    "one_or_two": "forward slash"})


We've included the .setattr_direct() method in case you want to set an object attribute without creating the corresponding dictionary key/value. This could be useful for storing save data for example, and is used internally to store aliases in ._aliases. Variables which have been set directly in this way are stored in _vars.

>>> x = CleverDict()
>>> x.setattr_direct("direct", False)

>>> x
CleverDict({}, _aliases={}, _vars={'direct': False})

Here's one way you could create a .store attribute and customise the auto-save behaviour for example:

class SaveDict(CleverDict):
    def __init__(self, *args, **kwargs):
        self.setattr_direct('store', [])
        super().__init__(*args, **kwargs)

    def save(self, name, value):, value))


You can set pretty much any function to run automatically whenever a CleverDict value is created or changed. There's an example function in cleverict.test_cleverdict which demonstrates how you just need to overwrite the .save method with your own:

>>> from cleverdict.test_cleverdict import example_save_function
>>> = example_save_function

>>> x = CleverDict({'total':6, 'usergroup': "Knights of Ni"})
Notional save to database: .total = 6 <class 'int'>
Notional save to database: .usergroup = Knights of Ni <class 'str'>

>>> = 42
Notional save to database: .life = 42 <class 'int'>

The example function above also appends output to a file, which you might want for debugging, auditing, further analysis etc.:

>>> with open("example.log","r") as file:
...     log =

>>> log.splitlines()
["Notional save to database: .total = 6 <class 'int'>",
"Notional save to database: .usergroup = Knights of Ni <class 'str'>"]

NB: The .save method is a class method, so changing will apply the new .save method to all previously created CleverDict objects as well as future ones.


When writing your own .save function, you'll need to specify three arguments as follows:

>>> def your_function(self, key: str, value: any):
...     print("Ni!")
  • self: because we're dealing with objects and classes...
  • key: a valid Python .attribute key preferably, otherwise you'll only be able to access it using dictionary['key'] notation later on.
  • value: anything


If you want to specify different .save behaviours for different objects, consider creating subclasses that inherit from CleverDict and set a different .save function for each subclass e.g.:

>>> class Type1(CleverDict): pass
>>> = my_save_function1

>>> class Type2(CleverDict): pass
>>> = my_save_function2


We'd love to see Pull Requests (and relevant tests) from other contributors, particularly if you can help evolve CleverDict to make it play nicely with other classes simply using inheritance, without causing recursion or requiring a rewrite/overwrite of the original class.

For example it would be great if we could graft on the CleverDict methods to other Classes, something like this:

>>> class MyDatetime(datetime.datetime, CleverDict):
...     pass

>>> mdt =
>>> mdt.hour
>>> mdt['hour']

Unfortunately at the moment this raises an error: TypeError: multiple bases have instance lay-out conflict

... which is beyond the author's current Python level!


CleverDict was developed jointly by Ruud van der Ham, Peter Fison, Loic Domaigne, and Rik Huygen who met on the friendly and excellent Pythonista Cafe forum ( Peter got the ball rolling after noticing a super-convenient, but not fully-fledged feature in Pandas that allows you to (mostly) use object.attribute syntax or dictionary['key'] syntax interchangeably. Ruud, Loic and Rik then started swapping ideas for a hybrid dictionary/data class, originally based on UserDict and the magic of __getattr__ and __setattr__.

(*) CleverDict was originally called attr_dict but several confusing flavours of this and AttrDict exist on PyPi and Github already. Hopefully this new tongue-in-cheek name is more memorable and raises a smile as you think about the Clever Dicks who created it ;)

Project details

Download files

Download the file for your platform. If you're not sure which to choose, learn more about installing packages.

Files for cleverdict, version 1.7.2
Filename, size File type Python version Upload date Hashes
Filename, size cleverdict-1.7.2.tar.gz (19.3 kB) File type Source Python version None Upload date Hashes View

Supported by

Pingdom Pingdom Monitoring Google Google Object Storage and Download Analytics Sentry Sentry Error logging AWS AWS Cloud computing DataDog DataDog Monitoring Fastly Fastly CDN DigiCert DigiCert EV certificate StatusPage StatusPage Status page