blob: c6a16f1d298938d98f46b5ffe1ab804585da99cd [file] [log] [blame]
#include <ctype.h>
#include <string.h>
#include <stdio.h>
#include <stdlib.h>
#include "jellyfish.h"
#define NOTNUM(c) ((c>57) || (c<48))
#define INRANGE(c) ((c>0) && (c<91))
#ifndef NaN
#define NaN (0.0 / 0.0)
#endif
/* borrowed heavily from strcmp95.c
* http://www.census.gov/geo/msb/stand/strcmp.c
*/
double _jaro_winkler(const char *ying, const char *yang,
bool long_tolerance, bool winklerize)
{
/* Arguments:
ying
yang
pointers to the 2 strings to be compared.
long_tolerance
Increase the probability of a match when the number of matched
characters is large. This option allows for a little more
tolerance when the strings are large. It is not an appropriate
test when comparing fixed length fields such as phone and
social security numbers.
*/
char *ying_flag=0, *yang_flag=0;
double weight;
long ying_length, yang_length, min_len;
long search_range;
long lowlim, hilim;
long trans_count, common_chars;
int i, j, k;
// ensure that neither string is blank
ying_length = strlen(ying);
yang_length = strlen(yang);
if (!ying_length || !yang_length) return 0;
search_range = min_len = (ying_length > yang_length) ? ying_length : yang_length;
// Blank out the flags
ying_flag = alloca(ying_length + 1);
if (!ying_flag) return NaN;
yang_flag = alloca(yang_length + 1);
if (!yang_flag) return NaN;
memset(ying_flag, 0, ying_length + 1);
memset(yang_flag, 0, yang_length + 1);
search_range = (search_range/2) - 1;
if (search_range < 0) search_range = 0;
// Looking only within the search range, count and flag the matched pairs.
common_chars = 0;
for (i = 0; i < ying_length; i++) {
lowlim = (i >= search_range) ? i - search_range : 0;
hilim = (i + search_range <= yang_length-1) ? (i + search_range) : yang_length-1;
for (j = lowlim; j <= hilim; j++) {
if (!yang_flag[j] && yang[j] == ying[i]) {
yang_flag[j] = 1;
ying_flag[i] = 1;
common_chars++;
break;
}
}
}
// If no characters in common - return
if (!common_chars) return 0;
// Count the number of transpositions
k = trans_count = 0;
for (i = 0; i < ying_length; i++) {
if (ying_flag[i]) {
for (j = k; j < yang_length; j++) {
if (yang_flag[j]) {
k = j + 1;
break;
}
}
if (ying[i] != yang[j]) {
trans_count++;
}
}
}
trans_count /= 2;
// adjust for similarities in nonmatched characters
// Main weight computation.
weight= common_chars / ((double) ying_length) + common_chars / ((double) yang_length)
+ ((double) (common_chars - trans_count)) / ((double) common_chars);
weight /= 3.0;
// Continue to boost the weight if the strings are similar
if (winklerize && weight > 0.7) {
// Adjust for having up to the first 4 characters in common
j = (min_len >= 4) ? 4 : min_len;
for (i=0; ((i<j) && (ying[i] == yang[i]) && (NOTNUM(ying[i]))); i++);
if (i) {
weight += i * 0.1 * (1.0 - weight);
}
/* Optionally adjust for long strings. */
/* After agreeing beginning chars, at least two more must agree and
the agreeing characters must be > .5 of remaining characters.
*/
if ((long_tolerance) && (min_len>4) && (common_chars>i+1) && (2*common_chars>=min_len+i)) {
if (NOTNUM(ying[0])) {
weight += (double) (1.0-weight) *
((double) (common_chars-i-1) / ((double) (ying_length+yang_length-i*2+2)));
}
}
}
return weight;
}
double jaro_winkler(const char *ying, const char *yang, bool long_tolerance)
{
return _jaro_winkler(ying, yang, long_tolerance, true);
}
double jaro_distance(const char *ying, const char *yang)
{
return _jaro_winkler(ying, yang, false, false);
}