Enhanced ranking algorithm

Based on the patch by Matt Westcott (@mjwestcott).
But with a more conservative approach:
- Does not use linearly increasing penalties; It is agreed upon that we
  should prefer matching characters at the beginnings of the words, but
  it's not always clear that the relevance is inversely proportional to
  the distance from the beginning.
- The approach here is more conservative in that the bonus is never
  large enough to override the matchlen, so it can be thought of as the
  first implicit tiebreak criterion.
- One may argue the change breaks the contract of --tiebreak, but the
  judgement depends on the definition of "tie".

1 files changed, 7 insertions, 4 deletions

diff --git a/src/item.go b/src/item.go
index 43ce1a21..66b3e4d1 100644
--- a/src/item.go
+++ b/src/item.go
@@ -23,6 +23,7 @@ type Item struct {
 	offsets     []Offset
 	colors      []ansiOffset
 	rank        [5]int32
+	bonus       int32
 }
 
 // Sort criteria to use. Never changes once fzf is started.
@@ -73,15 +74,17 @@ func (item *Item) Rank(cache bool) [5]int32 {
 			matchlen += end - begin
 		}
 	}
-	if matchlen == 0 {
-		matchlen = math.MaxInt32
-	}
 	rank := buildEmptyRank(item.Index())
 	for idx, criterion := range sortCriteria {
 		var val int32
 		switch criterion {
 		case byMatchLen:
-			val = int32(matchlen)
+			if matchlen == 0 {
+				val = math.MaxInt32
+			} else {
+				// It is extremely unlikely that bonus exceeds 128
+				val = 128*int32(matchlen) - item.bonus
+			}
 		case byLength:
 			// It is guaranteed that .transformed in not null in normal execution
 			if item.transformed != nil {