if I have a dictionary like this
>>> d = {10: 3, 100: 2, 1000: 1}
I can type something like:
>>> d.get(10), d
You can derive from dict
to change the behaviour of the get()
method:
class ClosestDict(dict):
def get(self, key):
key = min(self.iterkeys(), key=lambda x: abs(x - key))
return dict.get(self, key)
d = ClosestDict({10: 3, 100: 2, 1000: 1})
print (d.get(20), d.get(60), d.get(200))
prints
(3, 2, 2)
Note that the complexity of get()
no longer is O(1), but O(n).
bisect module allows fast lookup of insertion position in a sorted list.
from bisect import bisect_right
def closest_matches(data, query):
keys = sorted(data)
return [data[i] for i in (min(map(abs, (keys[p-1], keys[p]))) for p in (bisect_right(keys, k) for k in query))]
>>> d = {10: 3, 100: 2, 1000: 1}
>>> closest_matches(d, [20, 60, 200])
[3, 3, 2]
Checkout this recipe Fuzzy matching dictionary.