1. rcrnstn/fzy
  2. Commit 1eec4e7c2b5405a638c92a0142741d8fca11e1df
Browse code

Improve scoring

John Hawthorn authored on 30/07/2014 04:28:32
Showing 2 changed files
fzytest.c
History View file @ 1eec4e7
... ... 
@@ -35,6 +35,15 @@ int test_scoring(){
35 35 
 	/* App/MOdels/foo is better than App/M/fOo  */
36 36 
 	assert(match("amo", "app/m/foo") < match("amo", "app/models/foo"));
37 37
38 
+	/* GEMFIle.Lock < GEMFILe  */
39 
+	assert(match("gemfil", "Gemfile.lock") < match("gemfil", "Gemfile"));
40 
+
41 
+	/* GEMFIle.Lock < GEMFILe  */
42 
+	assert(match("gemfil", "Gemfile.lock") < match("gemfil", "Gemfile"));
43 
+
44 
+	/* Prefer shorter matches */
45 
+	assert(match("test", "tests") > match("test", "testing"));
46 
+
38 47 
 	return 0;
39 48 
 }
40 49
match.c
History View file @ 1eec4e7
... ... 
@@ -3,6 +3,7 @@
3 3 
 #include <strings.h>
4 4 
 #include <stdio.h>
5 5 
 #include <float.h>
6 
+#include <math.h>
6 7
7 8 
 #include "fzy.h"
8 9
... ... 
@@ -18,8 +19,8 @@ int has_match(const char *needle, const char *haystack){
18 19
19 20 
 #define max(a, b) (((a) > (b)) ? (a) : (b))
20 21 
 typedef double score_t;
21 
-#define SCORE_MAX DBL_MAX
22 
-#define SCORE_MIN -DBL_MAX
22 
+#define SCORE_MAX INFINITY
23 
+#define SCORE_MIN -INFINITY
23 24
24 25 
 /* print one of the internal matrices */
25 26 
 void mat_print(score_t *mat, int n, int m){
... ... 
@@ -65,34 +66,52 @@ double calculate_score(const char *needle, const char *haystack, size_t *positio
65 66 
 	 */
66 67
67 68 
 	/* Which positions are beginning of words */
68 
-	int at_bow = 1;
69 69 
 	char last_ch = '\0';
70 70 
 	for(int i = 0; i < m; i++){
71 71 
 		char ch = haystack[i];
72 
-		/* TODO: What about allcaps (ex. README) */
73 
-		int bow = (at_bow && isalnum(ch)) || (isupper(ch) && !isupper(last_ch));
74 
-		at_bow = !isalnum(ch);
75 
-		last_ch = ch;
76 72
77 
-		match_bonus[i] = bow ? 1.5 : 0;
73 
+		score_t score = 0;
74 
+		if(isalnum(ch)){
75 
+			if(last_ch == '/'){
76 
+				score = 1.5;
77 
+			}else if(last_ch == '-' ||
78 
+					last_ch == '_' ||
79 
+					last_ch == ' ' ||
80 
+					(last_ch >= '0' && last_ch <= '9')){
81 
+				score = 1.2;
82 
+			}else if(last_ch >= 'a' && last_ch <= 'z' &&
83 
+					ch >= 'A' && ch <= 'Z'){
84 
+				/* CamelCase */
85 
+				score = 1.1;
86 
+			}else if(last_ch == '.'){
87 
+				score = 0.8;
88 
+			}
89 
+		}
90 
+
91 
+		match_bonus[i] = score;
92 
+		last_ch = ch;
78 93 
 	}
79 94
80 95 
 	for(int i = 0; i < n; i++){
81 96 
 		for(int j = 0; j < m; j++){
82 
-			D[i][j] = SCORE_MIN;
97 
+			score_t score = j ? SCORE_MIN : 0;
83 98 
 			int match = tolower(needle[i]) == tolower(haystack[j]);
99 
+			D[i][j] = SCORE_MIN;
84 100 
 			if(match){
85 
-				score_t score = 0;
86 101 
 				if(i && j){
87 
-					score = M[i-1][j-1] + match_bonus[j];
102 
+					score = max(score, M[i-1][j-1] + match_bonus[j]);
88 103
89 104 
 					/* consecutive match, doesn't stack with match_bonus */
90 
-					score = max(score, 1 + D[i-1][j-1]);
105 
+					score = max(score, D[i-1][j-1] + 1.0);
106 
+				}else if(!i){
107 
+					score = (j * -0.01) + match_bonus[j];
91 108 
 				}
92 
-				M[i][j] = D[i][j] = score;
109 
+				D[i][j] = score;
110 
+			}
111 
+			if(j){
112 
+				score = max(score, M[i][j-1] - 0.01);
93 113 
 			}
94 
-			if(j)
95 
-				M[i][j] = max(M[i][j], M[i][j-1] - 0.05);
114 
+			M[i][j] = score;
96 115 
 		}
97 116 
 	}
98 117
... ... 
@@ -113,8 +132,8 @@ double calculate_score(const char *needle, const char *haystack, size_t *positio
113 132 
 				 * we encounter, the latest in the candidate
114 133 
 				 * string.
115 134 
 				 */
116 
-				if(D[i][j] == M[i][j]){
117 
-					positions[i] = j;
135 
+				if(tolower(needle[i]) == tolower(haystack[j]) && D[i][j] == M[i][j]){
136 
+					positions[i] = j--;
118 137 
 					break;
119 138 
 				}
120 139 
 			}