SimHash for chromaprint

simhash4chromaprint.py

#!/usr/env python
# -*- coding: UTF-8 -*-
__author__ = 'Myautsai Pan'
import sys
import numpy as np

class SimHash(object):
    '''Implements of SimHash for chromaprint'''
    def __init__(self, chromaprint=[], bits=32):
        self.hash_bits = 32
        self.value = self.cal_hash(chromaprint)

    def cal_hash(self, chromaprint):
        ''' input a hashed array `chromaprint`, output `self.hash_bits` bit hash'''
        Q = [0]*self.hash_bits
        for fp_hash in chromaprint:
            #TODO assert type(fp_hash) == int, "assert fingerprint type being integer."
            for i in xrange(self.hash_bits):
                r =(((fp_hash >> i) & 1) << 1) -1 # right started i'th(start from 0) bit * 2 -1
                assert r == 1 or r == -1
                Q[i] += r
        sim_hash = 0
        for i in xrange(self.hash_bits):
            w = 1 if Q[i]>0 else 0
            sim_hash |= w << i
        return sim_hash
    
    def __int__(self):
        return int(self.value)
    
    def __long__(self):
        return long(self.value)

    def __hash__(self):
        return self.value

    def __str__(self):
        s = [str(self.value >> i & 1) for i in xrange(self.hash_bits)]
        s.reverse()
        return ''.join(s)

    def __repr__(self):
        return ' '.join(('<SimHash',self.__str__(),' (',str(self.__int__()),')>'))
    
    def __call__(self):
        return self.value


def main(argv):
    a = SimHash([1,2,8,8,8,8,8])
    a = SimHash([1,2,8,8,8,8,7])
    print 'SimHash a:',repr(a)
    return 0

if __name__ == '__main__':
    sys.exit(main(sys.argv))