*: fix upper-cased charset and collation name (#301) (#314)

charset/charset.go

@@ -18,6 +18,17 @@ import (
 
 	"github.com/pingcap/errors"
 	"github.com/pingcap/parser/mysql"
+	"github.com/pingcap/parser/terror"
+)
+
+const (
+	codeCollationCharsetMismatch = terror.ErrCode(mysql.ErrCollationCharsetMismatch)
+	codeUnknownCollation         = terror.ErrCode(mysql.ErrUnknownCollation)
+)
+
+var (
+	ErrUnknownCollation         = terror.ClassDDL.New(codeUnknownCollation, mysql.MySQLErrName[mysql.ErrUnknownCollation])
+	ErrCollationCharsetMismatch = terror.ClassDDL.New(codeCollationCharsetMismatch, mysql.MySQLErrName[mysql.ErrCollationCharsetMismatch])
 )
 
 // Charset is a charset.
@@ -40,6 +51,8 @@ type Collation struct {
 }
 
 var charsets = make(map[string]*Charset)
+var collationsIDMap = make(map[int]*Collation)
+var collationsNameMap = make(map[string]*Collation)
 
 // All the supported charsets should be in the following table.
 var charsetInfos = []*Charset{
@@ -150,10 +163,8 @@ func GetCharsetInfoByID(coID int) (string, string, error) {
 	if coID == mysql.DefaultCollationID {
 		return mysql.DefaultCharset, mysql.DefaultCollationName, nil
 	}
-	for _, collation := range collations {
-		if coID == collation.ID {
-			return collation.CharsetName, collation.Name, nil
-		}
+	if collation, ok := collationsIDMap[coID]; ok {
+		return collation.CharsetName, collation.Name, nil
 	}
 	return "", "", errors.Errorf("Unknown charset id %d", coID)
 }
@@ -163,6 +174,14 @@ func GetCollations() []*Collation {
 	return collations
 }
 
+func GetCollationByName(name string) (*Collation, error) {
+	collation, ok := collationsNameMap[strings.ToLower(name)]
+	if !ok {
+		return nil, ErrUnknownCollation.GenWithStackByArgs(name)
+	}
+	return collation, nil
+}
+
 const (
 	// CharsetBin is used for marking binary charset.
 	CharsetBin = "binary"
@@ -415,10 +434,14 @@ func init() {
 		charsets[c.Name] = c
 	}
 	for _, c := range collations {
-		charset, ok := charsets[c.CharsetName]
-		if !ok {
-			continue
+		collationsIDMap[c.ID] = c
+
+		if charset, ok := charsets[c.CharsetName]; ok {
+			charset.Collations[c.Name] = c
 		}
-		charset.Collations[c.Name] = c
+	}
+
+	for id, name := range mysql.Collations {
+		collationsNameMap[name] = collationsIDMap[int(id)]
 	}
 }

charset/charset_test.go

@@ -98,3 +98,16 @@ func (s *testCharsetSuite) TestGetDefaultCollation(c *C) {
 		testGetDefaultCollation(c, tt.cs, tt.co, tt.succ)
 	}
 }
+
+func (s *testCharsetSuite) TestGetCollationByName(c *C) {
+	defer testleak.AfterTest(c)()
+
+	for _, collation := range collations {
+		coll, err := GetCollationByName(collation.Name)
+		c.Assert(err, IsNil)
+		c.Assert(coll, Equals, collation)
+	}
+
+	_, err := GetCollationByName("non_exist")
+	c.Assert(err, ErrorMatches, "\\[ddl:1273\\]Unknown collation: 'non_exist'")
+}

model/model.go

@@ -174,9 +174,15 @@ const (
 	// After version V2.1.2 (PR#8738) , TiDB add UTF8 check, then the user upgrade from v2.0.8 insert some UTF8MB4 characters will got error.
 	// This is not compatibility for user. Then we try to fix this in PR #9820, and increase the version number.
 	TableInfoVersion2 = uint16(2)
+	// TableInfoVersion3 means the table info version is 3.
+	// This version aims to deal with upper-cased charset name in TableInfo stored by versions prior to TiDB v2.1.9:
+	// TiDB always suppose all charsets / collations as lower-cased and try to convert them if they're not.
+	// However, the convert is missed in some scenarios before v2.1.9, so for all those tables prior to TableInfoVersion3, their
+	// charsets / collations will be converted to lower-case while loading from the storage.
+	TableInfoVersion3 = uint16(3)
 
 	// CurrLatestTableInfoVersion means the latest table info in the current TiDB.
-	CurrLatestTableInfoVersion = TableInfoVersion2
+	CurrLatestTableInfoVersion = TableInfoVersion3
 )
 
 // ExtraHandleName is the name of ExtraHandle Column.

mysql/charset.go

@@ -324,6 +324,7 @@ var Collations = map[uint8]string{
 	245: "utf8mb4_croatian_ci",
 	246: "utf8mb4_unicode_520_ci",
 	247: "utf8mb4_vietnamese_ci",
+	255: "utf8mb4_0900_ai_ci",
 }
 
 // CollationNames maps MySQL default collation name to its ID
@@ -547,6 +548,7 @@ var CollationNames = map[string]uint8{
 	"utf8mb4_croatian_ci":      245,
 	"utf8mb4_unicode_520_ci":   246,
 	"utf8mb4_vietnamese_ci":    247,
+	"utf8mb4_0900_ai_ci":       255,
 }
 
 // MySQL collation information.