sqlmodel(dm): generate key in lower case (#9498)

close #9489
pingcap · Aug 8, 2023 · b35406e · b35406e
1 parent e616696
commit b35406e
Show file tree

Hide file tree

Showing 3 changed files with 35 additions and 2 deletions.
diff --git a/pkg/sqlmodel/causality.go b/pkg/sqlmodel/causality.go
@@ -19,6 +19,7 @@ import (
 	"strings"
 
 	timodel "github.com/pingcap/tidb/parser/model"
+	"github.com/pingcap/tidb/parser/mysql"
 	"github.com/pingcap/tidb/sessionctx"
 	"github.com/pingcap/tidb/tablecodec"
 	"github.com/pingcap/tiflow/dm/pkg/log"
@@ -41,6 +42,19 @@ func (r *RowChange) CausalityKeys() []string {
 	return ret
 }
 
+func columnNeeds2LowerCase(col *timodel.ColumnInfo) bool {
+	switch col.GetType() {
+	case mysql.TypeVarchar, mysql.TypeString, mysql.TypeVarString, mysql.TypeTinyBlob,
+		mysql.TypeMediumBlob, mysql.TypeBlob, mysql.TypeLongBlob:
+		return collationNeeds2LowerCase(col.GetCollate())
+	}
+	return false
+}
+
+func collationNeeds2LowerCase(collation string) bool {
+	return strings.HasSuffix(collation, "_ci")
+}
+
 func columnValue2String(value interface{}) string {
 	var data string
 	switch v := value.(type) {
@@ -99,7 +113,12 @@ func genKeyString(
 			continue // ignore `null` value.
 		}
 		// one column key looks like:`column_val.column_name.`
-		buf.WriteString(columnValue2String(data))
+
+		val := columnValue2String(data)
+		if columnNeeds2LowerCase(columns[i]) {
+			val = strings.ToLower(val)
+		}
+		buf.WriteString(val)
 		buf.WriteString(".")
 		buf.WriteString(columns[i].Name.L)
 		buf.WriteString(".")

diff --git a/pkg/sqlmodel/causality_test.go b/pkg/sqlmodel/causality_test.go
@@ -155,6 +155,18 @@ func TestGetCausalityString(t *testing.T) {
 			values: []interface{}{16, "xyz"},
 			keys:   []string{"16.a.db.tbl", "xyz.b.db.tbl"},
 		},
+		{
+			// case insensitive
+			schema: `create table t_ci(a int unique, b varchar(16) primary key)default charset=utf8 collate=utf8_unicode_ci`,
+			values: []interface{}{16, "XyZ"},
+			keys:   []string{"16.a.db.tbl", "xyz.b.db.tbl"},
+		},
+		{
+			// case sensitive
+			schema: `create table t_bin(a int unique, b varchar(16) primary key)default charset=utf8 collate=utf8_bin`,
+			values: []interface{}{16, "XyZ"},
+			keys:   []string{"16.a.db.tbl", "XyZ.b.db.tbl"},
+		},
 		{
 			// primary key of multiple columns
 			schema: `create table t7(a int, b int, primary key(a, b))`,

diff --git a/pkg/sqlmodel/row_change_test.go b/pkg/sqlmodel/row_change_test.go
@@ -19,6 +19,7 @@ import (
 	"github.com/pingcap/tidb/ddl"
 	"github.com/pingcap/tidb/parser"
 	"github.com/pingcap/tidb/parser/ast"
+	"github.com/pingcap/tidb/parser/charset"
 	timodel "github.com/pingcap/tidb/parser/model"
 	timock "github.com/pingcap/tidb/util/mock"
 	cdcmodel "github.com/pingcap/tiflow/cdc/model"
@@ -33,7 +34,8 @@ func mockTableInfo(t *testing.T, sql string) *timodel.TableInfo {
 	se := timock.NewContext()
 	node, err := p.ParseOneStmt(sql, "", "")
 	require.NoError(t, err)
-	ti, err := ddl.MockTableInfo(se, node.(*ast.CreateTableStmt), 1)
+	dbChs, dbColl := charset.GetDefaultCharsetAndCollate()
+	ti, err := ddl.BuildTableInfoWithStmt(se, node.(*ast.CreateTableStmt), dbChs, dbColl, nil)
 	require.NoError(t, err)
 	return ti
 }