From 265f3afc2a20805dd16a7adee7eed6fc30a2b548 Mon Sep 17 00:00:00 2001
From: Andy Balholm <andy@balholm.com>
Date: Tue, 9 Jan 2024 06:03:56 -0800
Subject: [PATCH] matchfinder: penalize score for overlapping matches

---
 matchfinder/m4.go        | 16 +++++++++++++++-
 matchfinder/multihash.go | 16 +++++++++++++++-
 2 files changed, 30 insertions(+), 2 deletions(-)

diff --git a/matchfinder/m4.go b/matchfinder/m4.go
index a1fc824..5b2acba 100644
--- a/matchfinder/m4.go
+++ b/matchfinder/m4.go
@@ -167,7 +167,21 @@ func (q *M4) FindMatches(dst []Match, src []byte) []Match {
 			}
 		}
 
-		if q.score(currentMatch) <= q.score(matches[0]) {
+		if currentMatch.End-currentMatch.Start < q.MinLength {
+			continue
+		}
+
+		overlapPenalty := 0
+		if matches[0] != (absoluteMatch{}) {
+			overlapPenalty = 275
+			if currentMatch.Start <= matches[1].End {
+				// This match would completely replace the previous match,
+				// so there is no penalty for overlap.
+				overlapPenalty = 0
+			}
+		}
+
+		if q.score(currentMatch) <= q.score(matches[0])+overlapPenalty {
 			continue
 		}
 
diff --git a/matchfinder/multihash.go b/matchfinder/multihash.go
index adc0b40..369b99d 100644
--- a/matchfinder/multihash.go
+++ b/matchfinder/multihash.go
@@ -172,7 +172,21 @@ func (q *MultiHash) FindMatches(dst []Match, src []byte) []Match {
 			}
 		}
 
-		if currentMatch == (absoluteMatch{}) || q.score(currentMatch) <= q.score(matches[0]) {
+		if currentMatch.End-currentMatch.Start < q.MinLength {
+			continue
+		}
+
+		overlapPenalty := 0
+		if matches[0] != (absoluteMatch{}) {
+			overlapPenalty = 275
+			if currentMatch.Start <= matches[1].End {
+				// This match would completely replace the previous match,
+				// so there is no penalty for overlap.
+				overlapPenalty = 0
+			}
+		}
+
+		if q.score(currentMatch) <= q.score(matches[0])+overlapPenalty {
 			continue
 		}