guix-commits
[Top][All Lists]
Advanced

[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]

01/01: ui: Increase relevance score for exact matches.


From: Ludovic Courtès
Subject: 01/01: ui: Increase relevance score for exact matches.
Date: Fri, 29 Jun 2018 17:08:48 -0400 (EDT)

civodul pushed a commit to branch master
in repository guix.

commit fd1395c4982dd8cf7fe9843317d1ebb1e5d78bee
Author: Ludovic Courtès <address@hidden>
Date:   Fri Jun 29 12:17:41 2018 +0200

    ui: Increase relevance score for exact matches.
    
    Previously "guix package -s python" would have 'python2-zope-interface' as
    its first result (relevance: 10), followed by many other python-*
    packages with the same score, while 'python' itself would come
    later (relevance: 7).
    
    This change makes 'python' the first result (relevance: 27).
    
    Reported by Gábor Boskovits.
    
    * guix/ui.scm (relevance)[score]: Use 'fold-matches' instead of
    'match:count' to counter the number of maches.  Give more weight to
    exact matches.
---
 guix/ui.scm | 13 ++++++++-----
 1 file changed, 8 insertions(+), 5 deletions(-)

diff --git a/guix/ui.scm b/guix/ui.scm
index ec70945..6996b7f 100644
--- a/guix/ui.scm
+++ b/guix/ui.scm
@@ -1222,11 +1222,14 @@ field in the final score.
 A score of zero means that OBJ does not match any of REGEXPS.  The higher the
 score, the more relevant OBJ is to REGEXPS."
   (define (score str)
-    (let ((counts (filter-map (lambda (regexp)
-                                (match (regexp-exec regexp str)
-                                  (#f #f)
-                                  (m  (match:count m))))
-                              regexps)))
+    (let ((counts (map (lambda (regexp)
+                         (match (fold-matches regexp str '() cons)
+                           (()  0)
+                           ((m) (if (string=? (match:substring m) str)
+                                    5              ;exact match
+                                    1))
+                           (lst (length lst))))
+                       regexps)))
       ;; Compute a score that's proportional to the number of regexps matched
       ;; and to the number of matches for each regexp.
       (* (length counts) (reduce + 0 counts))))



reply via email to

[Prev in Thread] Current Thread [Next in Thread]