improve relevancy calculation

lib/completion.js

@@ -21,16 +21,23 @@ var completion = (function() {
     }
 
     /** Calculates a very simple similarity value between a :query and a :string. The current
-      * implementation simply returns the length of the longest prefix of :query that is found within :str.
+      * implementation simply returns the cumulated length of query parts that are also in the string.
       */
     self.calculateRelevancy = function(query, str) {
       query = self.normalize(query);
       str   = self.normalize(str);
-      for (var i = query.length; i >= 0; --i) {
-        if (str.indexOf(query.slice(0, i)) >= 0)
-          return i;
+      var sum = 0;
+      // only iterate over slices of the query starting at an offset up to 10 to save resources
+      for (var start = 0; start < 10; ++start) {
+        for (var i = query.length; i >= start; --i) {
+          if (str.indexOf(query.slice(start, i)) >= 0) {
+            sum += i - start;
+            break;
+          }
+
+        }
       }
-      return 0;
+      return sum;
     }
 
     /** Trims the size of the regex cache to the configured size using a FIFO algorithm. */