[kupfer] relevance: _findBestMatch must take first best match

[Ulrik Sverdrup <usverdrup src gnome org>]

commit 9744979115280d5fd2d3ee41305253612151b469
Author: Ulrik Sverdrup <ulrik sverdrup gmail com>
Date:   Sat Sep 12 22:58:10 2009 +0200

    relevance: _findBestMatch must take first best match
    
    If a match is perfect, it can not get better. we take the first
    perfect match; we only update best match if it's better.

 kupfer/relevance.py |    8 +++++++-
 1 files changed, 7 insertions(+), 1 deletions(-)
---
diff --git a/kupfer/relevance.py b/kupfer/relevance.py
index 6088f01..6fc17f1 100644
--- a/kupfer/relevance.py
+++ b/kupfer/relevance.py
@@ -216,7 +216,13 @@ def _findBestMatch(s, query):
                 return bestMatch
             cur += 1
             qcur += 1
-        bestMatch = (index, cur)
+
+        # take match if it is shorter
+        # if perfect match, we are done
+        if bestMatch[0] == -1 or (cur - index) < (bestMatch[1] - bestMatch[0]):
+            bestMatch = (index, cur)
+            if cur - index == queryLength:
+                break
 
         index = s.find(query[0], index + 1)