statistics: avoid fmsketch calculation for single-column index (#41931)

ref #41930

Author: xuyifangreeneyes

statistics/handle/handletest/handle_test.go

@@ -1070,7 +1070,7 @@ partition by range (a) (
 	tk.MustQuery("select distinct_count, null_count, tot_col_size, correlation=0 from mysql.stats_histograms where is_index=0 order by table_id asc").Check(
 		testkit.Rows("15 1 17 1", "6 1 7 0", "9 0 10 0"))
 	tk.MustQuery("select distinct_count, null_count, tot_col_size, correlation=0 from mysql.stats_histograms where is_index=1 order by table_id asc").Check(
-		testkit.Rows("15 1 0 1", "6 1 6 1", "9 0 10 1"))
+		testkit.Rows("15 1 0 1", "6 1 7 1", "9 0 10 1"))
 
 	tk.MustQuery("show stats_buckets where is_index=0").Check(
 		// db table partition col is_idx bucket_id count repeats lower upper ndv

statistics/integration_test.go

@@ -669,3 +669,27 @@ func TestShowHistogramsLoadStatus(t *testing.T) {
 		}
 	}
 }
+
+func TestSingleColumnIndexNDV(t *testing.T) {
+	store, dom := testkit.CreateMockStoreAndDomain(t)
+	tk := testkit.NewTestKit(t, store)
+	h := dom.StatsHandle()
+	tk.MustExec("use test")
+	tk.MustExec("create table t(a int, b int, c varchar(20), d varchar(20), index idx_a(a), index idx_b(b), index idx_c(c), index idx_d(d))")
+	require.NoError(t, h.HandleDDLEvent(<-h.DDLEventCh()))
+	tk.MustExec("insert into t values (1, 1, 'xxx', 'zzz'), (2, 2, 'yyy', 'zzz'), (1, 3, null, 'zzz')")
+	for i := 0; i < 5; i++ {
+		tk.MustExec("insert into t select * from t")
+	}
+	tk.MustExec("analyze table t")
+	rows := tk.MustQuery("show stats_histograms where db_name = 'test' and table_name = 't'").Sort().Rows()
+	expectedResults := [][]string{
+		{"a", "2", "0"}, {"b", "3", "0"}, {"c", "2", "32"}, {"d", "1", "0"},
+		{"idx_a", "2", "0"}, {"idx_b", "3", "0"}, {"idx_c", "2", "32"}, {"idx_d", "1", "0"},
+	}
+	for i, row := range rows {
+		require.Equal(t, expectedResults[i][0], row[3]) // column_name
+		require.Equal(t, expectedResults[i][1], row[6]) // distinct_count
+		require.Equal(t, expectedResults[i][2], row[7]) // null_count
+	}
+}

statistics/row_sampler.go

@@ -172,7 +172,7 @@ func (s *RowSampleBuilder) Collect() (RowSampleCollector, error) {
 			return nil, err
 		}
 		if chk.NumRows() == 0 {
-			return collector, nil
+			break
 		}
 		collector.Base().Count += int64(chk.NumRows())
 		for row := it.Begin(); row != it.End(); row = it.Next() {
@@ -212,6 +212,20 @@ func (s *RowSampleBuilder) Collect() (RowSampleCollector, error) {
 			collector.sampleRow(newCols, s.Rng)
 		}
 	}
+	for i, group := range s.ColGroups {
+		if len(group) != 1 {
+			continue
+		}
+		// For the single-column group, its FMSketch is the same as that of the corresponding column. Hence, we don't
+		// maintain its FMSketch in collectColumnGroups. We just copy the corresponding column's FMSketch after
+		// iterating all rows. Also, we can directly copy TotalSize and NullCount.
+		colIdx := group[0]
+		colGroupIdx := len(s.ColsFieldType) + i
+		collector.Base().FMSketches[colGroupIdx] = collector.Base().FMSketches[colIdx].Copy()
+		collector.Base().NullCount[colGroupIdx] = collector.Base().NullCount[colIdx]
+		collector.Base().TotalSizes[colGroupIdx] = collector.Base().TotalSizes[colIdx]
+	}
+	return collector, nil
 }
 
 func (s *baseCollector) collectColumns(sc *stmtctx.StatementContext, cols []types.Datum, sizes []int64) error {
@@ -234,17 +248,19 @@ func (s *baseCollector) collectColumnGroups(sc *stmtctx.StatementContext, cols [
 	colLen := len(cols)
 	datumBuffer := make([]types.Datum, 0, len(cols))
 	for i, group := range colGroups {
+		if len(group) == 1 {
+			// For the single-column group, its FMSketch is the same as that of the corresponding column. Hence, we
+			// don't need to maintain its FMSketch. We just copy the corresponding column's FMSketch after iterating
+			// all rows. Also, we can directly copy TotalSize and NullCount.
+			continue
+		}
+		// We don't maintain the null counts information for the multi-column group.
 		datumBuffer = datumBuffer[:0]
-		hasNull := true
 		for _, c := range group {
 			datumBuffer = append(datumBuffer, cols[c])
-			hasNull = hasNull && cols[c].IsNull()
-			s.TotalSizes[colLen+i] += sizes[c] - 1
-		}
-		// We don't maintain the null counts information for the multi-column group
-		if hasNull && len(group) == 1 {
-			s.NullCount[colLen+i]++
-			continue
+			if !cols[c].IsNull() {
+				s.TotalSizes[colLen+i] += sizes[c] - 1
+			}
 		}
 		err := s.FMSketches[colLen+i].InsertRowValue(sc, datumBuffer)
 		if err != nil {