fix(cdc): skip foreign key tables in scanner

Author: jiangxinmeng1

pkg/cdc/sql_builder.go

@@ -248,19 +248,20 @@ const (
 		"table_name = '%s'"
 
 	CDCCollectTableInfoSqlTemplate = "SELECT " +
-		" rel_id, " +
-		" relname, " +
-		" reldatabase_id, " +
-		" reldatabase, " +
-		" rel_createsql, " +
-		" account_id " +
-		"FROM `mo_catalog`.`mo_tables` " +
+		" tbl.rel_id, " +
+		" tbl.relname, " +
+		" tbl.reldatabase_id, " +
+		" tbl.reldatabase, " +
+		" tbl.rel_createsql, " +
+		" tbl.account_id, " +
+		" tbl.constraint " +
+		"FROM `mo_catalog`.`mo_tables` tbl " +
 		"WHERE " +
-		" account_id IN (%s) " +
+		" tbl.account_id IN (%s) " +
 		"%s" +
 		"%s" +
-		" AND relkind = '%s' " +
-		" AND reldatabase NOT IN (%s)"
+		" AND tbl.relkind = '%s' " +
+		" AND tbl.reldatabase NOT IN (%s)"
 	CDCInsertMOISCPLogSqlTemplate = `REPLACE INTO mo_catalog.mo_iscp_log (` +
 		`account_id,` +
 		`table_id,` +
@@ -475,6 +476,7 @@ var CDCSQLTemplates = [CDCSqlTemplateCount]struct {
 			"reldatabase",
 			"rel_createsql",
 			"account_id",
+			"constraint",
 		},
 	},
 	CDCGetWatermarkWhereSqlTemplate_Idx: {
@@ -1071,13 +1073,13 @@ func (b cdcSQLBuilder) CollectTableInfoSQL(accountIDs string, dbNames string, ta
 			if dbNames == "*" {
 				return ""
 			}
-			return " AND reldatabase IN (" + dbNames + ") "
+			return " AND tbl.reldatabase IN (" + dbNames + ") "
 		}(),
 		func() string {
 			if tableNames == "*" {
 				return ""
 			}
-			return " AND relname IN (" + tableNames + ") "
+			return " AND tbl.relname IN (" + tableNames + ") "
 		}(),
 		catalog.SystemOrdinaryRel,
 		AddSingleQuotesJoin(catalog.SystemDatabases),

pkg/cdc/table_scanner.go

@@ -19,7 +19,6 @@ import (
 	"fmt"
 	"runtime/debug"
 	"slices"
-	"strings"
 	"sync"
 	"sync/atomic"
 	"time"
@@ -37,6 +36,7 @@ import (
 	"github.com/matrixorigin/matrixone/pkg/defines"
 	"github.com/matrixorigin/matrixone/pkg/logutil"
 	"github.com/matrixorigin/matrixone/pkg/util/executor"
+	"github.com/matrixorigin/matrixone/pkg/vm/engine"
 )
 
 const (
@@ -752,9 +752,20 @@ func (s *TableDetector) scanTable() error {
 			dbName := cols[3].GetStringAt(i)
 			createSql := cols[4].GetStringAt(i)
 			accountId := vector.MustFixedColWithTypeCheck[uint32](cols[5])[i]
+			hasForeignKey, err := tableHasForeignKeyConstraint(cols[6].GetBytesAt(i))
+			if err != nil {
+				logutil.Warn(
+					"cdc.table_detector.scan_constraint_failed",
+					zap.Uint32("account-id", accountId),
+					zap.String("db", dbName),
+					zap.String("table", tblName),
+					zap.Error(err),
+				)
+				return false
+			}
 
 			// skip table with foreign key
-			if strings.Contains(strings.ToLower("createSql"), "foreign key") {
+			if hasForeignKey {
 				continue
 			}
 
@@ -787,10 +798,36 @@ func (s *TableDetector) scanTable() error {
 		}
 		return true
 	})
+	if err != nil {
+		return err
+	}
 
 	// replace the old table map
 	s.mu.Lock()
 	s.Mp = mp
 	s.mu.Unlock()
 	return nil
 }
+
+func tableHasForeignKeyConstraint(data []byte) (hasForeignKey bool, err error) {
+	if len(data) == 0 {
+		return false, nil
+	}
+
+	defer func() {
+		if r := recover(); r != nil {
+			err = moerr.NewInternalErrorNoCtxf("unmarshal table constraint failed: %v", r)
+		}
+	}()
+
+	constraintDef := &engine.ConstraintDef{}
+	if err := constraintDef.UnmarshalBinary(data); err != nil {
+		return false, err
+	}
+	for _, constraint := range constraintDef.Cts {
+		if foreignKeyDef, ok := constraint.(*engine.ForeignKeyDef); ok && len(foreignKeyDef.Fkeys) > 0 {
+			return true, nil
+		}
+	}
+	return false, nil
+}