MatasanoCrypto/utils/hamming.c

#include <stdio.h>
#include <string.h>
#include <stdlib.h>
#include "utils/base64.h"

// calculate distance between s1 and s2
// -1 : failure (different sizes of the strings)
int block_distance(const unsigned char* const s1, const unsigned char* const s2, const unsigned i_size)
{
	int dist = 0;
	int counter = 0;

	for(counter=0; counter<i_size; ++counter)
	{
		// calc nb bits set
		unsigned char ch = s1[counter]^s2[counter];
		while(ch!=0)
		{
			dist++;
			ch &= ch - 1;   // this is neet. Wegner technique for counting ones in binary computer
							// http://gurmeet.net/puzzles/fast-bit-counting-routines/
		}
	}

	return dist;
}

int choose_min_block_size(const unsigned char* const i_string,
                          const unsigned long i_size,
                          const unsigned i_min_block_size,
                          const unsigned i_max_block_size)
{
	unsigned block_size = 0;
	double min_dist=i_max_block_size*8;
	int min_block_size = i_max_block_size;

	// temporary results
	double tmp_dist1, tmp_dist2, tmp_dist3;
	tmp_dist1 = tmp_dist2 = tmp_dist3 = 0;

	for(block_size=i_min_block_size; block_size<=i_max_block_size; block_size++)
	{
		// take 4 blocks, calc distance, normalize and get min

		tmp_dist1 = block_distance(i_string, i_string+block_size, block_size);
		tmp_dist1 = tmp_dist1/(block_size*8);
		tmp_dist2 = block_distance(i_string+(block_size*2), i_string+(block_size*3), block_size);
		tmp_dist2 = tmp_dist2/(block_size*8);

		if(tmp_dist2<tmp_dist1)
			tmp_dist1 = tmp_dist2;
		if(tmp_dist1<min_dist) {
			min_dist = tmp_dist1;
			min_block_size = block_size;
		}
		// printf("DIST: %f BS: %d\n", tmp_dist1, block_size);

	}
	return min_block_size;
}