Question

I’m trying to create a custom hashing function for strings. I want to hash strings by their character frequency by weight. So that hi and ih will yield the same hash. Can I override __hash__?

Or is creating a wrapper class that holds the string and overriding __hash__ and __eq__ the only way?

Answer 1

You want a derived type with different equality semantics. Usually the approach taken will be to define how equality works, then build the hash method from the structures derived there, since it’s neccesary that the hash agree with equality. That might be:

import collections

class FrequencyString(str):
    @property
    def normalized(self):
        try:
            return self._normalized
        except AttributeError:
            self._normalized = normalized = ''.join(sorted(collections.Counter(self).elements()))
            return normalized

    def __eq__(self, other):
        return self.normalized == other.normalized

    def __hash__(self):
        return hash(self.normalized)