Creating dictionaries with pre-defined keys

Question

In python, is there a way to create a class that is treated like a dictionary but have the keys pre-defined when a new instance is created?

Answer 1

You can easily extend any built in type. This is how you'd do it with a dict:

>>> class MyClass(dict):
...     def __init__(self, *args, **kwargs):
...             self['mykey'] = 'myvalue'
...             self['mykey2'] = 'myvalue2'
...
>>> x = MyClass()
>>> x['mykey']
'myvalue'
>>> x
{'mykey2': 'myvalue2', 'mykey': 'myvalue'}

I wasn't able to find the Python documentation that talks about this, but the very popular book Dive Into Python (available for free online) has a few examples on doing this.

Answer 2

You can also have the dict subclass restrict the keys to a predefined list, by overriding __setitem__()

>>> class LimitedDict(dict):
    _keys = "a b c".split()
    def __init__(self, valtype=int):
        for key in LimitedDict._keys:
            self[key] = valtype()
    def __setitem__(self, key, val):
        if key not in LimitedDict._keys:
            raise KeyError
        dict.__setitem__(self, key, val)


>>> limited = LimitedDict()
>>> limited['a']
0
>>> limited['a'] = 3
>>> limited['a']
3
>>> limited['z'] = 0

Traceback (most recent call last):
  File "<pyshell#61>", line 1, in <module>
    limited['z'] = 0
  File "<pyshell#56>", line 8, in __setitem__
    raise KeyError
KeyError
>>> len(limited)
3

Answer 3

Yes, in Python dict is a class , so you can subclass it:

    class SubDict(dict):
        def __init__(self):
            dict.__init__(self)
            self.update({
                'foo': 'bar',
                'baz': 'spam',})

Here you override dict's __init__() method (a method which is called when an instance of the class is created). Inside __init__ you first call supercalss's __init__() method, which is a common practice when you whant to expand on the functionality of the base class. Then you update the new instance of SubDictionary with your initial data.

    subDict = SubDict()
    print subDict    # prints {'foo': 'bar', 'baz': 'spam'}

Answer 4

I'm not sure this is what you're looking for, but when I read your post I immediately thought you were looking to dynamically generate keys for counting exercises.

Unlike perl, which will do this for you by default,

grep{$_{$_}++} qw/ a a b c c c /;
print map{$_."\t".$_{$_}."\n"} sort {$_{$b}$_{$a}} keys %_;

c   3
a   2
b   1

Python won't give you this for free:

l = ["a","a","b","c","c","c"]
d = {}
for item in l:
    d[item] += 1

Traceback (most recent call last):
  File "./y.py", line 6, in 
    d[item] += 1
KeyError: 'a'

however, defaultdict will do this for you,

from collections import defaultdict
from operator import itemgetter

l = ["a","a","b","c","c","c"]
d = defaultdict(int)
for item in l:
    d[item] += 1

dl = sorted(d.items(),key=itemgetter(1), reverse=True)
for item in dl:
    print item

('c', 3)
('a', 2)
('b', 1)

Answer 5

Just create a subclass of dict and add the keys in the init method.

class MyClass(dict)

def __init__(self):
    """Creates a new dict with default values""""

    self['key1'] = 'value1'

Remember though, that in python any class that 'acts like a dict' is usually treated like one, so you don't have to worry too much about it being a subclass, you could instead implement the dict methods, although the above approach is probably more useful to you :).