stats: fix estimation in between row count (#5682)

Author: alivxxx

statistics/histogram.go

@@ -16,6 +16,7 @@ package statistics
 import (
 	"bytes"
 	"fmt"
+	"math"
 	"strings"
 	"time"
 
@@ -344,8 +345,10 @@ func (hg *Histogram) lessAndEqRowCount(value types.Datum) float64 {
 func (hg *Histogram) betweenRowCount(a, b types.Datum) float64 {
 	lessCountA := hg.lessRowCount(a)
 	lessCountB := hg.lessRowCount(b)
+	// If lessCountA is not less than lessCountB, it may be that they fall to the same bucket and we cannot estimate
+	// the fraction, so we use `totalCount / NDV` to estimate the row count, but the result should not greater than lessCountB.
 	if lessCountA >= lessCountB {
-		return hg.totalRowCount() / float64(hg.NDV)
+		return math.Min(lessCountB, hg.totalRowCount()/float64(hg.NDV))
 	}
 	return lessCountB - lessCountA
 }

statistics/statistics_test.go

@@ -271,6 +271,8 @@ func (s *testStatisticsSuite) TestBuild(c *C) {
 	c.Check(int(count), Equals, 19999)
 	count = col.betweenRowCount(encodeKey(types.NewIntDatum(30000)), encodeKey(types.NewIntDatum(35000)))
 	c.Check(int(count), Equals, 4999)
+	count = col.betweenRowCount(encodeKey(types.MinNotNullDatum()), encodeKey(types.NewIntDatum(0)))
+	c.Check(int(count), Equals, 0)
 	count = col.lessRowCount(encodeKey(types.NewIntDatum(0)))
 	c.Check(int(count), Equals, 0)