Adding buckets

Author: Dynom

finder/find.go

@@ -11,9 +11,11 @@ import (
 type Finder struct {
 	referenceMap    map[string]struct{}
 	reference       []string
+	referenceBucket map[rune][]string
 	Alg             Algorithm
 	LengthTolerance float64 // A number between 0.0-1.0 (percentage) to allow for length miss-match, anything outside this is considered not similar. Set to 0 to disable.
 	lock            sync.RWMutex
+	enableBuckets   bool
 }
 
 // Errors
@@ -33,11 +35,12 @@ const (
 func New(list []string, options ...Option) (*Finder, error) {
 	i := &Finder{}
 
-	i.Refresh(list)
 	for _, o := range options {
 		o(i)
 	}
 
+	i.Refresh(list)
+
 	if i.Alg == nil {
 		return i, ErrNoAlgorithmDefined
 	}
@@ -48,13 +51,23 @@ func New(list []string, options ...Option) (*Finder, error) {
 // Refresh replaces the internal reference list.
 func (t *Finder) Refresh(list []string) {
 	rm := make(map[string]struct{}, len(list))
+	rb := make(map[rune][]string, 26)
 	for _, r := range list {
 		rm[r] = struct{}{}
+
+		if t.enableBuckets {
+			l := rune(r[0])
+			if _, ok := rb[l]; !ok {
+				rb[l] = make([]string, 0, 128)
+			}
+			rb[l] = append(rb[l], r)
+		}
 	}
 
 	t.lock.Lock()
 	t.reference = list
 	t.referenceMap = rm
+	t.referenceBucket = rb
 	t.lock.Unlock()
 }
 
@@ -79,12 +92,20 @@ func (t *Finder) FindTopRankingCtx(ctx context.Context, input string) ([]string,
 	defer t.lock.RUnlock()
 
 	// Exact matches
-	if _, exists := t.referenceMap[input]; exists {
+	if _, exists := t.referenceMap[input]; exists || len(input) == 0 {
 		return []string{input}, BestScoreValue, true
 	}
 
+	var list []string
+	r := rune(input[0])
+	if l, ok := t.referenceBucket[r]; ok {
+		list = l
+	} else {
+		list = t.reference
+	}
+
 	var sameScore = []string{input}
-	for _, ref := range t.reference {
+	for _, ref := range list {
 		select {
 		case <-ctx.Done():
 			return []string{input}, WorstScoreValue, false

finder/find_test.go

@@ -2,6 +2,8 @@ package finder
 
 import (
 	"context"
+	"math"
+	"math/rand"
 	"testing"
 	"time"
 )
@@ -148,7 +150,29 @@ func TestMeetsLengthTolerance(t *testing.T) {
 			t.Errorf("Expected the tolerance to be %t\n%+v", td.Expect, td)
 		}
 	}
+}
+
+// Preventing the compiler to inline
+var ceilA, ceilB int
+
+func BenchmarkCeilOrNoCeil(b *testing.B) {
+	inputLen := 64
+	threshold := 0.195
+	b.Run("No Ceil", func(b *testing.B) {
+		for i := 0; i < b.N; i++ {
+			ceilA = int((float64(inputLen) * threshold) + 0.555)
+		}
+	})
+
+	b.Run("Ceil", func(b *testing.B) {
+		for i := 0; i < b.N; i++ {
+			ceilB = int(math.Ceil(float64(inputLen) * threshold))
+		}
+	})
 
+	if ceilA != ceilB {
+		b.Errorf("Implementation failure, a:%d != b:%d", ceilA, ceilB)
+	}
 }
 
 func BenchmarkSliceOrMap(b *testing.B) {
@@ -175,3 +199,55 @@ func BenchmarkSliceOrMap(b *testing.B) {
 		}
 	})
 }
+
+func BenchmarkFindWithBucket(b *testing.B) {
+	refs := generateRefs(1000, 20)
+	alg := NewJaroWinkler(.7, 4)
+
+	testRef := generateRef(20)
+	b.ReportAllocs()
+	b.Run("find with bucket", func(b *testing.B) {
+		f, _ := New(refs,
+			WithAlgorithm(alg),
+			WithLengthTolerance(0),
+			WithBuckets(false),
+		)
+
+		b.ResetTimer()
+		for i := 0; i < b.N; i++ {
+			f.Find(testRef)
+		}
+	})
+
+	b.Run("find without bucket", func(b *testing.B) {
+		f, _ := New(refs,
+			WithAlgorithm(alg),
+			WithLengthTolerance(0),
+			WithBuckets(true),
+		)
+
+		b.ResetTimer()
+		for i := 0; i < b.N; i++ {
+			f.Find(testRef)
+		}
+	})
+}
+
+func generateRefs(refNum, length uint64) []string {
+	refs := make([]string, refNum)
+	for i := uint64(0); i < refNum; i++ {
+		refs[i] = generateRef(length)
+	}
+
+	return refs
+}
+
+func generateRef(length uint64) string {
+	const alnum = "abcdefghijklmnopqrstuvwxyzABCDEFGHIJKLMNOPQRSTUVWXYZ0123456789"
+
+	var b = make([]byte, length)
+	for i := uint64(0); i < length; i++ {
+		b[i] = alnum[rand.Intn(len(alnum))]
+	}
+	return string(b)
+}

finder/option.go

@@ -18,3 +18,9 @@ func WithLengthTolerance(t float64) Option {
 		s.LengthTolerance = t
 	}
 }
+
+func WithBuckets(enable bool) Option {
+	return func(s *Finder) {
+		s.enableBuckets = enable
+	}
+}