statistics: fix "data too long" error when dumping stats from table with new collation data (#27033)

statistics/handle/dump.go

@@ -97,7 +97,7 @@ func (h *Handle) DumpStatsToJSONBySnapshot(dbName string, tableInfo *model.Table
 }
 
 func (h *Handle) tableStatsToJSON(dbName string, tableInfo *model.TableInfo, physicalID int64, snapshot uint64) (*JSONTable, error) {
-	tbl, err := h.tableStatsFromStorage(tableInfo, physicalID, true, snapshot)
+	tbl, err := h.TableStatsFromStorage(tableInfo, physicalID, true, snapshot)
 	if err != nil || tbl == nil {
 		return nil, err
 	}

statistics/handle/handle.go

@@ -217,7 +217,7 @@ func (h *Handle) Update(is infoschema.InfoSchema) error {
 		if oldTbl, ok := oldCache.tables[physicalID]; ok && oldTbl.Version >= version && tableInfo.UpdateTS == oldTbl.TblInfoUpdateTS {
 			continue
 		}
-		tbl, err := h.tableStatsFromStorage(tableInfo, physicalID, false, 0)
+		tbl, err := h.TableStatsFromStorage(tableInfo, physicalID, false, 0)
 		// Error is not nil may mean that there are some ddl changes on this table, we will not update it.
 		if err != nil {
 			logutil.BgLogger().Error("[stats] error occurred when read table stats", zap.String("table", tableInfo.Name.O), zap.Error(err))
@@ -531,8 +531,8 @@ func (h *Handle) columnStatsFromStorage(reader *statsReader, row chunk.Row, tabl
 	return nil
 }
 
-// tableStatsFromStorage loads table stats info from storage.
-func (h *Handle) tableStatsFromStorage(tableInfo *model.TableInfo, physicalID int64, loadAll bool, snapshot uint64) (_ *statistics.Table, err error) {
+// TableStatsFromStorage loads table stats info from storage.
+func (h *Handle) TableStatsFromStorage(tableInfo *model.TableInfo, physicalID int64, loadAll bool, snapshot uint64) (_ *statistics.Table, err error) {
 	reader, err := h.getStatsReader(snapshot)
 	if err != nil {
 		return nil, err
@@ -705,6 +705,12 @@ func (h *Handle) histogramFromStorage(reader *statsReader, tableID int64, colID
 		} else {
 			sc := &stmtctx.StatementContext{TimeZone: time.UTC}
 			d := rows[i].GetDatum(2, &fields[2].Column.FieldType)
+			// When there's new collation data, the length of bounds of histogram(the collate key) might be
+			// longer than the FieldType.Flen of this column.
+			// We change it to TypeBlob to bypass the length check here.
+			if tp.EvalType() == types.ETString && tp.Tp != mysql.TypeEnum && tp.Tp != mysql.TypeSet {
+				tp = types.NewFieldType(mysql.TypeBlob)
+			}
 			lowerBound, err = d.ConvertTo(sc, tp)
 			if err != nil {
 				return nil, errors.Trace(err)

statistics/handle/handle_test.go

@@ -33,6 +33,7 @@ import (
 	"github.com/pingcap/tidb/store/mockstore"
 	"github.com/pingcap/tidb/store/tikv/oracle"
 	"github.com/pingcap/tidb/types"
+	"github.com/pingcap/tidb/util/collate"
 	"github.com/pingcap/tidb/util/testkit"
 )
 
@@ -576,3 +577,31 @@ func (s *testStatsSuite) TestStatsCacheUpdateSkip(c *C) {
 	statsTbl2 := h.GetTableStats(tableInfo)
 	c.Assert(statsTbl1, Equals, statsTbl2)
 }
+
+var _ = SerialSuites(&statsSerialSuite{})
+
+type statsSerialSuite struct {
+	store kv.Storage
+	do    *domain.Domain
+}
+
+func (s *testSerialStatsSuite) TestLoadHistogramWithCollate(c *C) {
+	defer cleanEnv(c, s.store, s.do)
+	testKit := testkit.NewTestKit(c, s.store)
+	collate.SetNewCollationEnabledForTest(true)
+	defer collate.SetNewCollationEnabledForTest(false)
+	testKit.MustExec("use test")
+	testKit.MustExec("drop table if exists t")
+	testKit.MustExec("create table t(a varchar(10) collate utf8mb4_unicode_ci);")
+	testKit.MustExec("insert into t values('abcdefghij');")
+	testKit.MustExec("insert into t values('abcdufghij');")
+	testKit.MustExec("analyze table t with 0 topn;")
+	do := s.do
+	h := do.StatsHandle()
+	is := do.InfoSchema()
+	tbl, err := is.TableByName(model.NewCIStr("test"), model.NewCIStr("t"))
+	c.Assert(err, IsNil)
+	tblInfo := tbl.Meta()
+	_, err = h.TableStatsFromStorage(tblInfo, tblInfo.ID, true, 0)
+	c.Assert(err, IsNil)
+}