From ad38e12c15af10a73447196ac0b0f1a5df8bf865 Mon Sep 17 00:00:00 2001
From: n-peugnet <n.peugnet@free.fr>
Date: Mon, 30 Aug 2021 16:35:09 +0200
Subject: fix: chunkId added multiple times for same superFeature

---
 repo.go | 15 ++++++++++++++-
 1 file changed, 14 insertions(+), 1 deletion(-)

(limited to 'repo.go')

diff --git a/repo.go b/repo.go
index 8df6fdf..457dadd 100644
--- a/repo.go
+++ b/repo.go
@@ -223,12 +223,25 @@ func hashChunks(chunks <-chan StoredChunk) (FingerprintMap, SketchMap) {
 		fingerprints[h] = c.Id()
 		sketch, _ := SketchChunk(c, 32, 3, 4)
 		for _, s := range sketch {
-			sketches[s] = append(sketches[s], c.Id())
+			prev := sketches[s]
+			if contains(prev, c.Id()) {
+				continue
+			}
+			sketches[s] = append(prev, c.Id())
 		}
 	}
 	return fingerprints, sketches
 }
 
+func contains(s []*ChunkId, id *ChunkId) bool {
+	for _, v := range s {
+		if v == id {
+			return true
+		}
+	}
+	return false
+}
+
 func findSimilarChunk(chunk Chunk, sketches SketchMap) (*ChunkId, bool) {
 	var similarChunks = make(map[ChunkId]int)
 	var max int
-- 
cgit v1.2.3