From f2927a56dbdb7d169081a979c8310c7afa085800 Mon Sep 17 00:00:00 2001
From: Ti Chi Robot <ti-community-prow-bot@tidb.io>
Date: Tue, 27 Feb 2024 19:41:14 +0800
Subject: [PATCH] planner: a better way to round scale factor when collecting
 TopN stats (#49808) (#49820)

close pingcap/tidb#49801
---
 statistics/builder.go       |  4 +---
 statistics/cmsketch.go      |  7 +++++++
 statistics/cmsketch_test.go | 20 ++++++++++++++++++++
 3 files changed, 28 insertions(+), 3 deletions(-)

diff --git a/statistics/builder.go b/statistics/builder.go
index ec116803e952d..22946dd922db6 100644
--- a/statistics/builder.go
+++ b/statistics/builder.go
@@ -419,10 +419,8 @@ func BuildHistAndTopN(
 		}
 	}
 
-	for i := 0; i < len(topNList); i++ {
-		topNList[i].Count *= uint64(sampleFactor)
-	}
 	topn := &TopN{TopN: topNList}
+	topn.Scale(sampleFactor)
 
 	if uint64(count) <= topn.TotalCount() || int(hg.NDV) <= len(topn.TopN) {
 		// TopN includes all sample data
diff --git a/statistics/cmsketch.go b/statistics/cmsketch.go
index 43b29e7226a0e..829a824981f8f 100644
--- a/statistics/cmsketch.go
+++ b/statistics/cmsketch.go
@@ -529,6 +529,13 @@ type TopN struct {
 	TopN []TopNMeta
 }
 
+// Scale scales the TopN by the given factor.
+func (c *TopN) Scale(scaleFactor float64) {
+	for i := range c.TopN {
+		c.TopN[i].Count = uint64(float64(c.TopN[i].Count) * scaleFactor)
+	}
+}
+
 // AppendTopN appends a topn into the TopN struct.
 func (c *TopN) AppendTopN(data []byte, count uint64) {
 	if c == nil {
diff --git a/statistics/cmsketch_test.go b/statistics/cmsketch_test.go
index bed7181d3fd70..044dc0fcf163d 100644
--- a/statistics/cmsketch_test.go
+++ b/statistics/cmsketch_test.go
@@ -389,3 +389,23 @@ func TestMergePartTopN2GlobalTopNWithHists(t *testing.T) {
 	require.Equal(t, uint64(55), globalTopN.TotalCount(), "should have 55")
 	require.Len(t, leftTopN, 1, "should have 1 left topN")
 }
+
+func TestTopNScale(t *testing.T) {
+	for _, scaleFactor := range []float64{0.9999, 1.00001, 1.9999, 4.9999, 5.001, 9.99} {
+		var data []TopNMeta
+		sumCount := uint64(0)
+		for i := 0; i < 20; i++ {
+			cnt := uint64(rand.Intn(100000))
+			data = append(data, TopNMeta{
+				Count: cnt,
+			})
+			sumCount += cnt
+		}
+		topN := TopN{TopN: data}
+		topN.Scale(scaleFactor)
+		scaleCount := float64(sumCount) * scaleFactor
+		delta := math.Abs(float64(topN.TotalCount()) - scaleCount)
+		roundErrorRatio := delta / scaleCount
+		require.Less(t, roundErrorRatio, 0.0001)
+	}
+}