1. rcrnstn/fzy
  2. Commit ae67b76802fa3d7d301b37b4b13e2ff2a1ece490
Browse code

Adjust scoring system

-0.05 for a skipped character in the candidate.
+1 for a match following a previous match
+1.5 for a match at the beginning of a word
No change for any other match.

John Hawthorn authored on 27/07/2014 05:38:37
Showing 2 changed files
fzytest.c
History View file @ ae67b76
... ... 
@@ -62,6 +62,15 @@ int test_positions_2(){
62 62 
 	return 0;
63 63 
 }
64 64
65 
+int test_positions_3(){
66 
+	size_t positions[2];
67 
+	match_positions("as", "tags", positions);
68 
+	assert(positions[0] == 1);
69 
+	assert(positions[1] == 3);
70 
+
71 
+	return 0;
72 
+}
73 
+
65 74 
 int test_positions_exact(){
66 75 
 	size_t positions[3];
67 76 
 	match_positions("foo", "foo", positions);
... ... 
@@ -84,6 +93,7 @@ int main(int argc, char *argv[]){
84 93 
 	runtest(test_scoring);
85 94 
 	runtest(test_positions_1);
86 95 
 	runtest(test_positions_2);
96 
+	runtest(test_positions_3);
87 97 
 	runtest(test_positions_exact);
88 98
89 99 
 	summary();
match.c
History View file @ ae67b76
... ... 
@@ -29,7 +29,7 @@ void mat_print(int *mat, int n, int m){
29 29 
 }
30 30
31 31 
 #define max(a, b) (((a) > (b)) ? (a) : (b))
32 
-typedef int score_t;
32 
+typedef double score_t;
33 33 
 #define SCORE_MAX DBL_MAX
34 34 
 #define SCORE_MIN -DBL_MAX
35 35
... ... 
@@ -50,7 +50,7 @@ double calculate_score(const char *needle, const char *haystack, size_t *positio
50 50 
 	}
51 51
52 52 
 	int bow[m];
53 
-	int D[n][m], M[n][m];
53 
+	score_t D[n][m], M[n][m];
54 54 
 	bzero(D, sizeof(D));
55 55 
 	bzero(M, sizeof(M));
56 56
... ... 
@@ -72,39 +72,38 @@ double calculate_score(const char *needle, const char *haystack, size_t *positio
72 72
73 73 
 	for(int i = 0; i < n; i++){
74 74 
 		for(int j = 0; j < m; j++){
75 
+			D[i][j] = SCORE_MIN;
75 76 
 			int match = tolower(needle[i]) == tolower(haystack[j]);
76 77 
 			if(match){
77 78 
 				score_t score = 0;
78 79 
 				if(i && j)
79 80 
 					score = M[i-1][j-1];
80 81 
 				if(bow[j])
81 
-					score += 2;
82 
+					score += 1.5;
82 83 
 				else if(i && j && D[i-1][j-1])
83 84 
 					score = max(score, 1 + D[i-1][j-1]);
84 85 
 				M[i][j] = D[i][j] = score;
85 86 
 			}
86 87 
 			if(j)
87 
-				M[i][j] = max(M[i][j], M[i][j-1]);
88 
+				M[i][j] = max(M[i][j], M[i][j-1] - 0.05);
88 89 
 		}
89 90 
 	}
90 91
91 92 
 	/* backtrace to find the positions of optimal matching */
92 93 
 	if(positions){
93 94 
 		for(int i = n-1, j = m-1; i >= 0; i--){
94 
-			int last = M[i][j];
95 
-			for(; j >= 0 && M[i][j] == last; j--){
95 
+			for(; j >= 0; j--){
96 96 
 				/*
97 97 
 				 * There may be multiple paths which result in
98 98 
 				 * the optimal weight.
99 99 
 				 *
100 
-				 * Since we don't exit the loop on the first
101 
-				 * match, positions[i] may be assigned to
102 
-				 * multiple times. Since we are decrementing i
103 
-				 * and j, this favours the optimal path
104 
-				 * occurring earlier in the string.
100 
+				 * For simplicity, we will pick the first one
101 
+				 * we encounter, the latest in the candidate
102 
+				 * string.
105 103 
 				 */
106 
-				if(tolower(needle[i]) == tolower(haystack[j])){
104 
+				if(D[i][j] == M[i][j]){
107 105 
 					positions[i] = j;
106 
+					break;
108 107 
 				}
109 108 
 			}
110 109 
 		}