stats: specially handle unqiue key when estimate (#13354) (#13385)

statistics/feedback.go

@@ -930,7 +930,7 @@ func (q *QueryFeedback) recalculateExpectCount(h *Handle) error {
 		expected *= idx.getIncreaseFactor(t.Count)
 	} else {
 		c := t.Columns[id]
-		expected, err = c.getColumnRowCount(sc, ranges, t.ModifyCount)
+		expected, err = c.getColumnRowCount(sc, ranges, t.ModifyCount, true)
 		expected *= c.getIncreaseFactor(t.Count)
 	}
 	if err != nil {

statistics/histogram.go

@@ -806,7 +806,7 @@ func (c *Column) equalRowCount(sc *stmtctx.StatementContext, val types.Datum, mo
 }
 
 // getColumnRowCount estimates the row count by a slice of Range.
-func (c *Column) getColumnRowCount(sc *stmtctx.StatementContext, ranges []*ranger.Range, modifyCount int64) (float64, error) {
+func (c *Column) getColumnRowCount(sc *stmtctx.StatementContext, ranges []*ranger.Range, modifyCount int64, pkIsHandle bool) (float64, error) {
 	var rowCount float64
 	for _, rg := range ranges {
 		cmp, err := rg.LowVal[0].CompareDatum(sc, &rg.HighVal[0])
@@ -816,6 +816,11 @@ func (c *Column) getColumnRowCount(sc *stmtctx.StatementContext, ranges []*range
 		if cmp == 0 {
 			// the point case.
 			if !rg.LowExclude && !rg.HighExclude {
+				// In this case, the row count is at most 1.
+				if pkIsHandle {
+					rowCount += 1
+					continue
+				}
 				var cnt float64
 				cnt, err = c.equalRowCount(sc, rg.LowVal[0], modifyCount)
 				if err != nil {
@@ -909,6 +914,11 @@ func (idx *Index) getRowCount(sc *stmtctx.StatementContext, indexRanges []*range
 				continue
 			}
 			if fullLen {
+				// At most 1 in this case.
+				if idx.Info.Unique {
+					totalCount += 1
+					continue
+				}
 				count, err := idx.equalRowCount(sc, lb, modifyCount)
 				if err != nil {
 					return 0, err

statistics/selectivity_test.go

@@ -324,7 +324,40 @@ func (s *testSelectivitySuite) TestEstimationForUnknownValues(c *C) {
 	c.Assert(count, Equals, 0.0)
 }
 
+func (s *testSelectivitySuite) TestEstimationUniqueKeyEqualConds(c *C) {
+	defer cleanEnv(c, s.store, s.dom)
+	testKit := testkit.NewTestKit(c, s.store)
+	testKit.MustExec("use test")
+	testKit.MustExec("drop table if exists t")
+	testKit.MustExec("create table t(a int, b int, c int, unique key(b))")
+	testKit.MustExec("insert into t values (1,1,1),(2,2,2),(3,3,3),(4,4,4),(5,5,5),(6,6,6),(7,7,7)")
+	testKit.MustExec("analyze table t")
+	table, err := s.dom.InfoSchema().TableByName(model.NewCIStr("test"), model.NewCIStr("t"))
+	c.Assert(err, IsNil)
+	statsTbl := s.dom.StatsHandle().GetTableStats(table.Meta())
+
+	sc := &stmtctx.StatementContext{}
+	idxID := table.Meta().Indices[0].ID
+	count, err := statsTbl.GetRowCountByIndexRanges(sc, idxID, getRange(7, 7))
+	c.Assert(err, IsNil)
+	c.Assert(count, Equals, 1.0)
+
+	count, err = statsTbl.GetRowCountByIndexRanges(sc, idxID, getRange(6, 6))
+	c.Assert(err, IsNil)
+	c.Assert(count, Equals, 1.0)
+
+	colID := table.Meta().Columns[0].ID
+	count, err = statsTbl.GetRowCountByIntColumnRanges(sc, colID, getRange(7, 7))
+	c.Assert(err, IsNil)
+	c.Assert(count, Equals, 1.0)
+
+	count, err = statsTbl.GetRowCountByIntColumnRanges(sc, colID, getRange(6, 6))
+	c.Assert(err, IsNil)
+	c.Assert(count, Equals, 1.0)
+}
+
 func (s *testSelectivitySuite) TestPrimaryKeySelectivity(c *C) {
+	defer cleanEnv(c, s.store, s.dom)
 	testKit := testkit.NewTestKit(c, s.store)
 	testKit.MustExec("use test")
 	testKit.MustExec("drop table if exists t")

statistics/table.go

@@ -415,7 +415,7 @@ func (coll *HistColl) GetRowCountByIntColumnRanges(sc *stmtctx.StatementContext,
 		return getPseudoRowCountByUnsignedIntRanges(intRanges, float64(coll.Count)), nil
 	}
 	c := coll.Columns[colID]
-	result, err := c.getColumnRowCount(sc, intRanges, coll.ModifyCount)
+	result, err := c.getColumnRowCount(sc, intRanges, coll.ModifyCount, true)
 	result *= c.getIncreaseFactor(coll.Count)
 	return result, errors.Trace(err)
 }
@@ -427,7 +427,7 @@ func (coll *HistColl) GetRowCountByColumnRanges(sc *stmtctx.StatementContext, co
 		return getPseudoRowCountByColumnRanges(sc, float64(coll.Count), colRanges, 0)
 	}
 	c := coll.Columns[colID]
-	result, err := c.getColumnRowCount(sc, colRanges, coll.ModifyCount)
+	result, err := c.getColumnRowCount(sc, colRanges, coll.ModifyCount, false)
 	result *= c.getIncreaseFactor(coll.Count)
 	return result, errors.Trace(err)
 }
@@ -546,6 +546,12 @@ func (coll *HistColl) getIndexRowCount(sc *stmtctx.StatementContext, idxID int64
 	totalCount := float64(0)
 	for _, ran := range indexRanges {
 		rangePosition := getOrdinalOfRangeCond(sc, ran)
+		coverAll := len(ran.LowVal) == len(idx.Info.Columns) && rangePosition == len(ran.LowVal)
+		// // In this case, the row count is at most 1.
+		if coverAll && idx.Info.Unique {
+			totalCount += 1.0
+			continue
+		}
 		// If first one is range, just use the previous way to estimate; if it is [NULL, NULL] range
 		// on single-column index, use previous way as well, because CMSketch does not contain null
 		// values in this case.