Make top score collector about 7 times faster

search/collectors/bench_test.go

@@ -0,0 +1,32 @@
+package collectors
+
+import (
+	"math/rand"
+	"strconv"
+	"testing"
+
+	"github.com/blevesearch/bleve/search"
+	"golang.org/x/net/context"
+)
+
+func benchHelper(numOfMatches int, collector search.Collector, b *testing.B) {
+	matches := make(search.DocumentMatchCollection, 0, numOfMatches)
+	for i := 0; i < numOfMatches; i++ {
+		matches = append(matches, &search.DocumentMatch{
+			ID:    strconv.Itoa(i),
+			Score: rand.Float64(),
+		})
+	}
+
+	b.ResetTimer()
+
+	for run := 0; run < b.N; run++ {
+		searcher := &stubSearcher{
+			matches: matches,
+		}
+		err := collector.Collect(context.Background(), searcher)
+		if err != nil {
+			b.Fatal(err)
+		}
+	}
+}

search/collectors/collector_top_score.go

@@ -24,6 +24,7 @@ type TopScoreCollector struct {
 	results       *list.List
 	took          time.Duration
 	maxScore      float64
+	minScore      float64
 	total         uint64
 	facetsBuilder *search.FacetsBuilder
 }
@@ -98,6 +99,10 @@ func (tksc *TopScoreCollector) collectSingle(dm *search.DocumentMatch) {
 		tksc.maxScore = dm.Score
 	}
 
+	if dm.Score <= tksc.minScore {
+		return
+	}
+
 	for e := tksc.results.Front(); e != nil; e = e.Next() {
 		curr := e.Value.(*search.DocumentMatch)
 		if dm.Score < curr.Score {
@@ -106,7 +111,7 @@ func (tksc *TopScoreCollector) collectSingle(dm *search.DocumentMatch) {
 			// if we just made the list too long
 			if tksc.results.Len() > (tksc.k + tksc.skip) {
 				// remove the head
-				tksc.results.Remove(tksc.results.Front())
+				tksc.minScore = tksc.results.Remove(tksc.results.Front()).(*search.DocumentMatch).Score
 			}
 			return
 		}
@@ -115,7 +120,7 @@ func (tksc *TopScoreCollector) collectSingle(dm *search.DocumentMatch) {
 	tksc.results.PushBack(dm)
 	if tksc.results.Len() > (tksc.k + tksc.skip) {
 		// remove the head
-		tksc.results.Remove(tksc.results.Front())
+		tksc.minScore = tksc.results.Remove(tksc.results.Front()).(*search.DocumentMatch).Score
 	}
 }
 

search/collectors/collector_top_score_test.go

@@ -10,8 +10,6 @@
 package collectors
 
 import (
-	"math/rand"
-	"strconv"
 	"testing"
 
 	"golang.org/x/net/context"
@@ -225,27 +223,17 @@ func TestTop10ScoresSkip10(t *testing.T) {
 }
 
 func BenchmarkTop10of100000Scores(b *testing.B) {
+	benchHelper(10000, NewTopScorerCollector(10), b)
+}
 
-	matches := make(search.DocumentMatchCollection, 0, 100000)
-	for i := 0; i < 100000; i++ {
-		matches = append(matches, &search.DocumentMatch{
-			ID:    strconv.Itoa(i),
-			Score: rand.Float64(),
-		})
-	}
-	searcher := &stubSearcher{
-		matches: matches,
-	}
+func BenchmarkTop100of100000Scores(b *testing.B) {
+	benchHelper(10000, NewTopScorerCollector(100), b)
+}
 
-	collector := NewTopScorerCollector(10)
-	b.ResetTimer()
+func BenchmarkTop10of1000000Scores(b *testing.B) {
+	benchHelper(100000, NewTopScorerCollector(10), b)
+}
 
-	err := collector.Collect(context.Background(), searcher)
-	if err != nil {
-		b.Fatal(err)
-	}
-	res := collector.Results()
-	for _, dm := range res {
-		b.Logf("%s - %f\n", dm.ID, dm.Score)
-	}
+func BenchmarkTop100of1000000Scores(b *testing.B) {
+	benchHelper(100000, NewTopScorerCollector(100), b)
 }