*: fix upper-cased charset and collation name

ast/ddl_test.go

@@ -234,12 +234,12 @@ func (ts *testDDLSuite) TestDDLColumnDefRestore(c *C) {
 		{"id set('''a''','''b''')", "`id` SET('''a''','''b''')"},
 		{"id set('a\\nb','a''	\\r\\nb','a\\rb')", "`id` SET('a\nb','a''	\r\nb','a\rb')"},
 		{`id set("a'\nb","a'b\tc")`, "`id` SET('a''\nb','a''b\tc')"},
-		{"id TEXT CHARACTER SET UTF8 COLLATE UTF8_UNICODE_G", "`id` TEXT CHARACTER SET UTF8 COLLATE UTF8_UNICODE_G"},
+		{"id TEXT CHARACTER SET UTF8 COLLATE UTF8_UNICODE_CI", "`id` TEXT CHARACTER SET UTF8 COLLATE utf8_unicode_ci"},
 		{"id text character set UTF8", "`id` TEXT CHARACTER SET UTF8"},
 		{"id text charset UTF8", "`id` TEXT CHARACTER SET UTF8"},
-		{"id varchar(50) collate UTF8MB4_CZECH_CI", "`id` VARCHAR(50) COLLATE UTF8MB4_CZECH_CI"},
-		{"id varchar(50) collate utf8", "`id` VARCHAR(50) COLLATE utf8"},
-		{"id varchar(50) collate utf8 collate utf8mb4_bin", "`id` VARCHAR(50) COLLATE utf8 COLLATE utf8mb4_bin"},
+		{"id varchar(50) collate UTF8MB4_CZECH_CI", "`id` VARCHAR(50) COLLATE utf8mb4_czech_ci"},
+		{"id varchar(50) collate utf8_bin", "`id` VARCHAR(50) COLLATE utf8_bin"},
+		{"id varchar(50) collate utf8_unicode_ci collate utf8mb4_bin", "`id` VARCHAR(50) COLLATE utf8_unicode_ci COLLATE utf8mb4_bin"},
 		{"c1 char(10) character set LATIN1 collate latin1_german1_ci", "`c1` CHAR(10) CHARACTER SET LATIN1 COLLATE latin1_german1_ci"},
 
 		{"id int(11) PRIMARY KEY", "`id` INT(11) PRIMARY KEY"},

charset/charset.go

@@ -4,7 +4,7 @@
 // you may not use this file except in compliance with the License.
 // You may obtain a copy of the License at
 //
-//     http://www.apache.org/licenses/LICENSE-2.0
+//	http://www.apache.org/licenses/LICENSE-2.0
 //
 // Unless required by applicable law or agreed to in writing, software
 // distributed under the License is distributed on an "AS IS" BASIS,
@@ -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.
@@ -40,7 +51,8 @@ type Collation struct {
 }
 
 var charsets = make(map[string]*Charset)
-var collationsMap = make(map[int]*Collation)
+var collationsIDMap = make(map[int]*Collation)
+var collationsNameMap = make(map[string]*Collation)
 var descs = make([]*Desc, 0, len(charsetInfos))
 var supportedCollations = make([]*Collation, 0, len(supportedCollationNames))
 
@@ -155,7 +167,7 @@ func GetCharsetInfoByID(coID int) (string, string, error) {
 	if coID == mysql.DefaultCollationID {
 		return mysql.DefaultCharset, mysql.DefaultCollationName, nil
 	}
-	if collation, ok := collationsMap[coID]; ok {
+	if collation, ok := collationsIDMap[coID]; ok {
 		return collation.CharsetName, collation.Name, nil
 	}
 	return "", "", errors.Errorf("Unknown charset id %d", coID)
@@ -166,6 +178,14 @@ 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"
@@ -426,7 +446,7 @@ func init() {
 	}
 
 	for _, c := range collations {
-		collationsMap[c.ID] = c
+		collationsIDMap[c.ID] = c
 
 		if _, ok := supportedCollationNames[c.Name]; ok {
 			supportedCollations = append(supportedCollations, c)
@@ -436,4 +456,8 @@ func init() {
 			charset.Collations[c.Name] = c
 		}
 	}
+
+	for id, name := range mysql.Collations {
+		collationsNameMap[name] = collationsIDMap[int(id)]
+	}
 }

charset/charset_test.go

@@ -156,6 +156,19 @@ func (s *testCharsetSuite) TestGetCharsetDesc(c *C) {
 	}
 }
 
+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'")
+}
+
 func BenchmarkGetCharsetDesc(b *testing.B) {
 	b.ResetTimer()
 	charsets := []string{CharsetUTF8, CharsetUTF8MB4, CharsetASCII, CharsetLatin1, CharsetBin}
@@ -165,4 +178,5 @@ func BenchmarkGetCharsetDesc(b *testing.B) {
 	for i := 0; i < b.N; i++ {
 		GetCharsetDesc(cs)
 	}
+
 }

model/model.go

@@ -175,9 +175,15 @@ 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

@@ -343,6 +343,7 @@ 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 collation name to its ID
@@ -566,6 +567,7 @@ 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.