From ad0070fa445216690b8027624d8da3a8a7e060a1 Mon Sep 17 00:00:00 2001
From: Jim Posen <jimpo@coinbase.com>
Date: Mon, 22 Jan 2018 22:05:04 -0800
Subject: [PATCH] gcs/builder: Deduplicate items before creating block filters.

---
 gcs/builder/builder.go      | 22 +++++++++++++++-------
 gcs/builder/builder_test.go | 12 ++++++++++++
 2 files changed, 27 insertions(+), 7 deletions(-)

diff --git a/gcs/builder/builder.go b/gcs/builder/builder.go
index 264085c..081e8bd 100644
--- a/gcs/builder/builder.go
+++ b/gcs/builder/builder.go
@@ -20,9 +20,12 @@ const DefaultP = 20
 
 // GCSBuilder is a utility class that makes building GCS filters convenient.
 type GCSBuilder struct {
-	p    uint8
-	key  [gcs.KeySize]byte
-	data [][]byte
+	p   uint8
+	key [gcs.KeySize]byte
+
+	// data is a set of entries represented as strings. This is done to
+	// deduplicate items as they are added.
+	data map[string]struct{}
 	err  error
 }
 
@@ -125,8 +128,8 @@ func (b *GCSBuilder) Preallocate(n uint32) *GCSBuilder {
 		return b
 	}
 
-	if len(b.data) == 0 {
-		b.data = make([][]byte, 0, n)
+	if b.data == nil {
+		b.data = make(map[string]struct{}, n)
 	}
 
 	return b
@@ -140,7 +143,7 @@ func (b *GCSBuilder) AddEntry(data []byte) *GCSBuilder {
 		return b
 	}
 
-	b.data = append(b.data, data)
+	b.data[string(data)] = struct{}{}
 	return b
 }
 
@@ -217,7 +220,12 @@ func (b *GCSBuilder) Build() (*gcs.Filter, error) {
 		return nil, b.err
 	}
 
-	return gcs.BuildGCSFilter(b.p, b.key, b.data)
+	dataSlice := make([][]byte, 0, len(b.data))
+	for item := range b.data {
+		dataSlice = append(dataSlice, []byte(item))
+	}
+
+	return gcs.BuildGCSFilter(b.p, b.key, dataSlice)
 }
 
 // WithKeyPN creates a GCSBuilder with specified key and the passed probability
diff --git a/gcs/builder/builder_test.go b/gcs/builder/builder_test.go
index 29f847b..5cfd17e 100644
--- a/gcs/builder/builder_test.go
+++ b/gcs/builder/builder_test.go
@@ -286,4 +286,16 @@ func BuilderTest(b *builder.GCSBuilder, hash *chainhash.Hash, p uint8,
 	if !match {
 		t.Fatal("Filter didn't match when it should have!")
 	}
+
+	// Check that adding duplicate items does not increase filter size.
+	originalSize := f.N()
+	b.AddScript(addrBytes)
+	b.AddWitness(witness)
+	f, err = b.Build()
+	if err != nil {
+		t.Fatalf("Filter build failed: %s", err.Error())
+	}
+	if f.N() != originalSize {
+		t.Fatal("Filter size increased with duplicate items")
+	}
 }