update signature interface

Author: ekzhu

lsh.go

@@ -11,10 +11,10 @@ const (
 	integrationPrecision = 0.01
 )
 
-type hashKeyFunc func(Signature) string
+type hashKeyFunc func([]uint64) string
 
 func hashKeyFuncGen(hashValueSize int) hashKeyFunc {
-	return func(sig Signature) string {
+	return func(sig []uint64) string {
 		s := make([]byte, hashValueSize*len(sig))
 		buf := make([]byte, 8)
 		for i, v := range sig {
@@ -163,7 +163,7 @@ func (f *MinhashLSH) Params() (k, l int) {
 
 // Add a key with MinHash signature into the index.
 // The key won't be searchable until Index() is called.
-func (f *MinhashLSH) Add(key interface{}, sig Signature) {
+func (f *MinhashLSH) Add(key interface{}, sig []uint64) {
 	// Generate hash keys
 	Hs := make([]string, f.l)
 	for i := 0; i < f.l; i++ {
@@ -213,7 +213,7 @@ func (f *MinhashLSH) Index() {
 }
 
 // Query returns candidate keys given the query signature.
-func (f *MinhashLSH) Query(sig Signature) []interface{} {
+func (f *MinhashLSH) Query(sig []uint64) []interface{} {
 	set := f.query(sig, f.k)
 	results := make([]interface{}, 0, len(set))
 	for key := range set {
@@ -222,7 +222,7 @@ func (f *MinhashLSH) Query(sig Signature) []interface{} {
 	return results
 }
 
-func (f *MinhashLSH) query(sig Signature, minK int) map[interface{}]bool {
+func (f *MinhashLSH) query(sig []uint64, minK int) map[interface{}]bool {
 	results := make(map[interface{}]bool)
 	for K := f.k; K >= minK; K-- {
 		prefixSize := f.hashValueSize * K

lsh_benchmark_test.go

@@ -6,7 +6,7 @@ import (
 )
 
 func Benchmark_Insert10000(b *testing.B) {
-	sigs := make([]Signature, 10000)
+	sigs := make([][]uint64, 10000)
 	for i := range sigs {
 		sigs[i] = randomSignature(64, int64(i))
 	}

lsh_test.go

@@ -5,9 +5,9 @@ import (
 	"testing"
 )
 
-func randomSignature(size int, seed int64) Signature {
+func randomSignature(size int, seed int64) []uint64 {
 	r := rand.New(rand.NewSource(seed))
-	sig := make(Signature, size)
+	sig := make([]uint64, size)
 	for i := range sig {
 		sig[i] = uint64(r.Int63())
 	}

minhash.go

@@ -1,6 +1,7 @@
 package minhashlsh
 
 import (
+	"bytes"
 	"encoding/binary"
 	"hash/fnv"
 	"math/rand"
@@ -11,15 +12,13 @@ import (
 // The number of byte in a hash value for Minhash
 const hashValueSize = 8
 
-type Signature []uint64
-
-// Represents a MinHash object
+// Minhash represents a MinHash object
 type Minhash struct {
 	mw   *minwise.MinWise
 	seed int64
 }
 
-// Initialize a MinHash object with a seed and the number of
+// NewMinhash initialize a MinHash object with a seed and the number of
 // hash functions.
 func NewMinhash(seed int64, numHash int) *Minhash {
 	r := rand.New(rand.NewSource(seed))
@@ -54,7 +53,7 @@ func (m *Minhash) Push(b []byte) {
 	m.mw.Push(b)
 }
 
-// Export the MinHash signature.
+// Signature exports the MinHash as a list of hash values.
 func (m *Minhash) Signature() []uint64 {
 	return m.mw.Signature()
 }
@@ -68,3 +67,27 @@ func (m *Minhash) Merge(o *Minhash) {
 	}
 	m.mw.Merge(o.mw)
 }
+
+// SigToBytes serializes the signature into byte slice
+func SigToBytes(sig []uint64) []byte {
+	buf := new(bytes.Buffer)
+	for _, v := range sig {
+		binary.Write(buf, binary.BigEndian, v)
+	}
+	return buf.Bytes()
+}
+
+// BytesToSig converts a byte slice into a signature
+func BytesToSig(data []byte) ([]uint64, error) {
+	size := len(data) / hashValueSize
+	sig := make([]uint64, size)
+	buf := bytes.NewReader(data)
+	var v uint64
+	for i := range sig {
+		if err := binary.Read(buf, binary.BigEndian, &v); err != nil {
+			return nil, err
+		}
+		sig[i] = v
+	}
+	return sig, nil
+}