CODEWITHSUNDEEP
pythondictionary
Roee Adler
Roee Adler

Reputation: 33980

Python: finding keys with unique values in a dictionary?

I receive a dictionary as input, and want to return a list of keys for which the dictionary values are unique in the scope of that dictionary.

I will clarify with an example. Say my input is dictionary a, constructed as follows:

a = dict()
a['cat'] =      1
a['fish'] =     1
a['dog'] =      2  # <-- unique
a['bat'] =      3
a['aardvark'] = 3
a['snake'] =    4  # <-- unique
a['wallaby'] =  5
a['badger'] =   5

The result I expect is ['dog', 'snake'].

There are obvious brute force ways to achieve this, however I wondered if there's a neat Pythonian way to get the job done.

Upvotes: 18

Views: 18622

Answers (9)

Anurag Uniyal
Anurag Uniyal

Reputation: 88737

I think efficient way if dict is too large would be

countMap = {}
for v in a.itervalues():
    countMap[v] = countMap.get(v,0) + 1
uni = [ k for k, v in a.iteritems() if countMap[v] == 1]

Upvotes: 15

Bartosz Radaczyński
Bartosz Radaczyński

Reputation: 18564

Note that this actually is a bruteforce:

l = a.values()
b = [x for x in a if l.count(a[x]) == 1]

Upvotes: 6

S.Lott
S.Lott

Reputation: 391818

Here's another variation.

>>> import collections
>>> inverse= collections.defaultdict(list)
>>> for k,v in a.items():
...     inverse[v].append(k)
... 
>>> [ v[0] for v in inverse.values() if len(v) == 1 ]
['dog', 'snake']

I'm partial to this because the inverted dictionary is such a common design pattern.

Upvotes: 0

fabmilo
fabmilo

Reputation: 48310

What about subclassing? 

class UniqueValuesDict(dict):

    def __init__(self, *args):
        dict.__init__(self, *args)
        self._inverse = {}

    def __setitem__(self, key, value):
        if value in self.values():
            if value in self._inverse:
                del self._inverse[value]
        else:
            self._inverse[value] = key
        dict.__setitem__(self, key, value)

    def unique_values(self):
        return self._inverse.values()

a = UniqueValuesDict()

a['cat'] =      1
a['fish'] =     1
a[None] =       1
a['duck'] =     1
a['dog'] =      2  # <-- unique
a['bat'] =      3
a['aardvark'] = 3
a['snake'] =    4  # <-- unique
a['wallaby'] =  5
a['badger'] =   5

assert a.unique_values() == ['dog', 'snake']

Upvotes: 2

Rick Copeland
Rick Copeland

Reputation: 11902

Here is a solution that only requires traversing the dict once:

def unique_values(d):
    seen = {} # dict (value, key)
    result = set() # keys with unique values
    for k,v in d.iteritems():
        if v in seen:
            result.discard(seen[v])
        else:
            seen[v] = k
            result.add(k)
    return list(result)

Upvotes: 5

Greg Bacon
Greg Bacon

Reputation: 139441

Use nested list comprehensions!

print [v[0] for v in 
           dict([(v, [k for k in a.keys() if a[k] == v])
                     for v in set(a.values())]).values()
       if len(v) == 1]

Upvotes: -2

Juergen
Juergen

Reputation: 12728

A little more verbose, but does need only one pass over a:

revDict = {}
for k, v in a.iteritems():
  if v in revDict:
     revDict[v] = None
  else:
     revDict[v] = k

[ x for x in revDict.itervalues() if x != None ]

( I hope it works, since I can't test it here )

Upvotes: 2

John Machin
John Machin

Reputation: 82924

>>> b = []
>>> import collections
>>> bag = collections.defaultdict(lambda: 0)
>>> for v in a.itervalues():
...     bag[v] += 1
...
>>> b = [k for (k, v) in a.iteritems() if bag[v] == 1]
>>> b.sort() # optional
>>> print b
['dog', 'snake']
>>>

Upvotes: 4

Alex Morega
Alex Morega

Reputation: 4228

You could do something like this (just count the number of occurrences for each value):

def unique(a):
    from collections import defaultdict
    count = defaultdict(lambda: 0)
    for k, v in a.iteritems():
        count[v] += 1
    for v, c in count.iteritems():
        if c <= 1:
            yield v

Upvotes: -1

Related Questions