arHSM / NamedDict

Its like typing.NamedTuple but dictionaries!!!

NamedDicts aka NamedTuples but dictionaries

How does it work?

- No magic involved here, nameddict.NamedDict is a just subclass of dict, only the __getattribute__ and __setattr__ have been replaced, and thats where the magic happens.

How to create a NamedDict?

Well its as simple as this -

>>> import nameddict.NamedDict as NamedDict
>>>
>>>
>>> class Data(NamedDict):
...     __slots__ = ()  # you need this line to avoid the creation of 
...                     # `__dict__`
...     value_a: int
...     value_b: str
...
>>> # initialise it just like you would do with `dict()`
>>> data = Data(value_a=1, value_b="hi!")
>>> print(data)
Data {"value_a": 1, "value_b": "hi!"}
>>> print(data.value_a)
1
>>> print(data.value_b)
"hi!"

Points to note:

1. Attributes that conflict with the methods in dict object such as items, keys, values, etc... cannot be accessed by instance.items instead you need to do instance["items"].

2. You need to add __slots__ = () to prevent the creation of __dict__ (Which you don't need and want, trust me).

NOTE: These type hints mean nothing!

Extending NamedDicts

With the new NamedDictMeta you can extend your NamedDicts.

>>> form nameddict import NamedDict, NamedDictMeta
>>>
>>>
>>> class Data(NamedDict, metaclass=NamedDictMeta):
...     val_1: str
...     val_2 = 5
...     val_3: bool = False
...
>>> data = Data()
>>> data
Data {"val_1": None, "val_3": False}
>>> data.val_2 = 5
# Simplified error
AttributeError: Data has no attribute val_2.
>>> data.val_3
True
>>> data.val_2
# Simplified error
KeyError: 'val_2'

Points to note:

1. Type hints still don't mean anything, you can always pass in a different type of object.

2. Attribute without an annotation will be ignored if there are attributes with annotations present (look into the implementation for better understanding).

3. Default values is None unless explicitly set.

4. You do not need to add the __slots__ line, because the metaclass handles that.

Why use (NamedDict, metaclass=NamedDictMeta), instead of handling it in the meta class?

• Long story short, it's mainly for type-checkers to stop being annoying.

How to install?

Well simply download the nameddict.py file and put it in your project folder, then import it like just any other file!