python - Case insensitive dictionary

Question

I'd like my dictionary to be case insensitive.

I have this example code:

text = "practice changing the color"

words = {'color': 'colour',
        'practice': 'practise'}

def replace(words,text):

    keys = words.keys()

    for i in keys:
        text= text.replace(i ,words[i])
    return  text

text = replace(words,text)

print text

Output = practise changing the colour

I'd like another string, "practice changing the Color", (where Color starts with a capital) to also give the same output.

I believe there is a general way to convert to lowercase using mydictionary[key.lower()] but I'm not sure how to best integrate this into my existing code. (If this would be a reasonable, simple approach anyway).

Answer 1

If I understand you correctly and you want a way to key dictionaries in a non case-sensitive fashion, one way would be to subclass dict and overload the setter / getter:

class CaseInsensitiveDict(dict):
    def __setitem__(self, key, value):
        super(CaseInsensitiveDict, self).__setitem__(key.lower(), value)

    def __getitem__(self, key):
        return super(CaseInsensitiveDict, self).__getitem__(key.lower())

Answer 2

The currently approved answer doesn't work for a lot of cases, so it cannot be used as a drop-in dict replacement. Some tricky points in getting a proper dict replacement:

• overloading all of the methods that involve keys
• properly handling non-string keys
• properly handling the constructor of the class

The following should work much better:

class CaseInsensitiveDict(dict):
    @classmethod
    def _k(cls, key):
        return key.lower() if isinstance(key, basestring) else key

    def __init__(self, *args, **kwargs):
        super(CaseInsensitiveDict, self).__init__(*args, **kwargs)
        self._convert_keys()
    def __getitem__(self, key):
        return super(CaseInsensitiveDict, self).__getitem__(self.__class__._k(key))
    def __setitem__(self, key, value):
        super(CaseInsensitiveDict, self).__setitem__(self.__class__._k(key), value)
    def __delitem__(self, key):
        return super(CaseInsensitiveDict, self).__delitem__(self.__class__._k(key))
    def __contains__(self, key):
        return super(CaseInsensitiveDict, self).__contains__(self.__class__._k(key))
    def has_key(self, key):
        return super(CaseInsensitiveDict, self).has_key(self.__class__._k(key))
    def pop(self, key, *args, **kwargs):
        return super(CaseInsensitiveDict, self).pop(self.__class__._k(key), *args, **kwargs)
    def get(self, key, *args, **kwargs):
        return super(CaseInsensitiveDict, self).get(self.__class__._k(key), *args, **kwargs)
    def setdefault(self, key, *args, **kwargs):
        return super(CaseInsensitiveDict, self).setdefault(self.__class__._k(key), *args, **kwargs)
    def update(self, E={}, **F):
        super(CaseInsensitiveDict, self).update(self.__class__(E))
        super(CaseInsensitiveDict, self).update(self.__class__(**F))
    def _convert_keys(self):
        for k in list(self.keys()):
            v = super(CaseInsensitiveDict, self).pop(k)
            self.__setitem__(k, v)

Answer 3

Just for the record. I found an awesome impementation on Requests:

https://github.com/kennethreitz/requests/blob/v1.2.3/requests/structures.py#L37

Answer 4

In my particular instance, I needed a case insensitive lookup, however, I did not want to modify the original case of the key. For example:

>>> d = {}
>>> d['MyConfig'] = 'value'
>>> d['myconfig'] = 'new_value'
>>> d
{'MyConfig': 'new_value'}

You can see that the dictionary still has the original key, however it is accessible case-insensitively. Here's a simple solution:

class CaseInsensitiveKey(object):
    def __init__(self, key):
        self.key = key
    def __hash__(self):
        return hash(self.key.lower())
    def __eq__(self, other):
        return self.key.lower() == other.key.lower()
    def __str__(self):
        return self.key

The __hash__ and __eq__ overrides are required for both getting and setting entries in the dictionary. This is creating keys that hash to the same position in the dictionary if they are case-insensitively equal.

Now either create a custom dictionary that initializes a CaseInsensitiveKey using the provided key:

class CaseInsensitiveDict(dict):
    def __setitem__(self, key, value):
        key = CaseInsensitiveKey(key)
        super(CaseInsensitiveDict, self).__setitem__(key, value)
    def __getitem__(self, key):
        key = CaseInsensitiveKey(key)
        return super(CaseInsensitiveDict, self).__getitem__(key)

or simply make sure to always pass an instance of CaseInsensitiveKey as the key when using the dictionary.

Answer 5

Would you consider using string.lower() on your inputs and using a fully lowercase dictionary? It's a bit of a hacky solution, but it works

Answer 6

I've modified the simple yet good solution by pleasemorebacon (thanks!) making it slightly more compact, self-contained and with minor updates to allow construction from {'a':1, 'B':2} and support __contains__ protocol. Finally, since the CaseInsensitiveDict.Key is expected to be string (what else can be case-sensitive or not), it is a good idea to derive Key class from the str, then it is possible, for instance, to dump CaseInsensitiveDict with json.dumps out of the box.

# caseinsensitivedict.py
class CaseInsensitiveDict(dict):

    class Key(str):
        def __init__(self, key):
            str.__init__(key)
        def __hash__(self):
            return hash(self.lower())
        def __eq__(self, other):
            return self.lower() == other.lower()

    def __init__(self, data=None):
        super(CaseInsensitiveDict, self).__init__()
        if data is None:
            data = {}
        for key, val in data.items():
            self[key] = val
    def __contains__(self, key):
        key = self.Key(key)
        return super(CaseInsensitiveDict, self).__contains__(key)
    def __setitem__(self, key, value):
        key = self.Key(key)
        super(CaseInsensitiveDict, self).__setitem__(key, value)
    def __getitem__(self, key):
        key = self.Key(key)
        return super(CaseInsensitiveDict, self).__getitem__(key)

Here is a basic test script for those who like to check things in action:

# test_CaseInsensitiveDict.py
import json
import unittest
from caseinsensitivedict import *

class Key(unittest.TestCase):
    def setUp(self):
        self.Key = CaseInsensitiveDict.Key
        self.lower = self.Key('a')
        self.upper = self.Key('A')

    def test_eq(self):
        self.assertEqual(self.lower, self.upper)

    def test_hash(self):
        self.assertEqual(hash(self.lower), hash(self.upper))

    def test_str(self):
        self.assertEqual(str(self.lower), 'a')
        self.assertEqual(str(self.upper), 'A')

class Dict(unittest.TestCase):
    def setUp(self):
        self.Dict = CaseInsensitiveDict
        self.d1 = self.Dict()
        self.d2 = self.Dict()
        self.d1['a'] = 1
        self.d1['B'] = 2
        self.d2['A'] = 1
        self.d2['b'] = 2

    def test_contains(self):
        self.assertIn('B', self.d1)
        d = self.Dict({'a':1, 'B':2})
        self.assertIn('b', d)

    def test_init(self):
        d = self.Dict()
        self.assertFalse(d)
        d = self.Dict({'a':1, 'B':2})
        self.assertTrue(d)

    def test_items(self):
        self.assertDictEqual(self.d1, self.d2)
        self.assertEqual(
            [v for v in self.d1.items()],
            [v for v in self.d2.items()])

    def test_json_dumps(self):
        s = json.dumps(self.d1)
        self.assertIn('a', s)
        self.assertIn('B', s)

    def test_keys(self):
        self.assertEqual(self.d1.keys(), self.d2.keys())

    def test_values(self):
        self.assertEqual(
            [v for v in self.d1.values()],
            [v for v in self.d2.values()])

Answer 7

While a case insensitive dictionary is a solution, and there are answers to how to achieve that, there is a possibly easier way in this case. A case insensitive search is sufficient:

import re

text = "Practice changing the Color"
words = {'color': 'colour', 'practice': 'practise'}

def replace(words,text):
        keys = words.keys()
        for i in keys:
                exp = re.compile(i, re.I)
                text = re.sub(exp, words[i], text)
        return text

text = replace(words,text)
print text

Answer 8

You can do a dict key case insensitive search with a one liner:

>>> input_dict = {'aBc':1, 'xyZ':2}
>>> search_string = 'ABC'
>>> next((value for key, value in input_dict.items() if key.lower()==search_string.lower()), None)
1
>>> search_string = 'EFG'
>>> next((value for key, value in input_dict.items() if key.lower()==search_string.lower()), None)
>>>

You can place that into a function:

def get_case_insensitive_key_value(input_dict, key):
    return next((value for dict_key, value in input_dict.items() if dict_key.lower() == key.lower()), None)

Note that only the first match is returned.

Answer 9

If you only need to do this once in your code (hence, no point to a function), the most straightforward way to deal with the problem is this:

lowercase_dict = {key.lower(): value for (key, value) in original_dict}

I'm assuming here that the dict in question isn't all that large--it might be inelegant to duplicate it, but if it's not large, it isn't going to hurt anything.

The advantage of this over @Fred's answer (though that also works) is that it produces the same result as a dict when the key isn't present: a KeyError.

Answer 10

I just set up a function to handle this:

def setLCdict(d, k, v):
    k = k.lower()
    d[k] = v
    return d

myDict = {}

So instead of

myDict['A'] = 1
myDict['B'] = 2

You can:

myDict = setLCdict(myDict, 'A', 1)
myDict = setLCdict(myDict, 'B', 2)

You can then either lower case the value before looking it up or write a function to do so.

    def lookupLCdict(d, k):
        k = k.lower()
        return d[k]

    myVal = lookupLCdict(myDict, 'a')

Probably not ideal if you want to do this globally but works well if its just a subset you wish to use it for.