findSimilarChunk returns an Id if found

3 files changed, 22 insertions, 10 deletions

diff --git a/TODO.md b/TODO.md
index a55e618..6c62b4c 100644
--- a/TODO.md
+++ b/TODO.md
@@ -2,6 +2,7 @@ priority 1
 ----------
 - join non-deduplicated chunks
   - choose when and how to
+- read from repo
 
 priority 2
 ----------
diff --git a/repo.go b/repo.go
index 599b8da..bf8b167 100644
--- a/repo.go
+++ b/repo.go
@@ -229,17 +229,26 @@ func hashChunks(chunks <-chan StoredChunk) (FingerprintMap, SketchMap) {
 	return fingerprints, sketches
 }
 
-func findSimilarChunks(chunks []Chunk, sketches SketchMap) {
-	for i, c := range chunks {
-		log.Println("New chunk:", i)
-		sketch, _ := SketchChunk(c, 32, 3, 4)
-		for _, s := range sketch {
-			chunkId, exists := sketches[s]
-			if exists {
-				log.Println("Found similar chunks: ", chunkId)
+func findSimilarChunk(chunk Chunk, sketches SketchMap) (*ChunkId, bool) {
+	var similarChunks = make(map[ChunkId]int)
+	var max int
+	var similarChunk *ChunkId
+	sketch, _ := SketchChunk(chunk, 32, 3, 4)
+	for _, s := range sketch {
+		chunkIds, exists := sketches[s]
+		if !exists {
+			continue
+		}
+		for _, id := range chunkIds {
+			count := similarChunks[*id]
+			count += 1
+			if count > max {
+				similarChunk = id
 			}
+			similarChunks[*id] = count
 		}
 	}
+	return similarChunk, similarChunk != nil
 }
 
 func readChunk(stream io.Reader) ([]byte, error) {
diff --git a/repo_test.go b/repo_test.go
index 40d4374..3a4af68 100644
--- a/repo_test.go
+++ b/repo_test.go
@@ -200,7 +200,9 @@ func TestBsdiff(t *testing.T) {
 	}
 	newChunks := extractNewChunks(recipe)
 	log.Println("Checking new chunks:", len(newChunks[0]))
-	for _, c := range newChunks {
-		findSimilarChunks(c, sketches)
+	for _, chunks := range newChunks {
+		for _, c := range chunks {
+			log.Println(findSimilarChunk(c, sketches))
+		}
 	}
 }