ddl: make TiDB treat charset and collate case insensitive pingcap#8577 (

pingcap#9338)
eurekaka · Feb 18, 2019 · f9df5fa · f9df5fa
1 parent 1e962c9
commit f9df5fa
Show file tree

Hide file tree

Showing 3 changed files with 21 additions and 0 deletions.
diff --git a/ddl/integration_test.go b/ddl/integration_test.go
@@ -118,6 +118,19 @@ func (s *testIntegrationSuite) TestEndIncluded(c *C) {
 	tk.MustExec("admin check table t")
 }
 
+func (s *testIntegrationSuite) TestCaseInsensitiveCharsetAndCollate(c *C) {
+	tk := testkit.NewTestKit(c, s.store)
+
+	tk.MustExec("create database if not exists test_charset_collate")
+	defer tk.MustExec("drop database test_charset_collate")
+	tk.MustExec("use test_charset_collate")
+	tk.MustExec("create table t(id int) ENGINE=InnoDB DEFAULT CHARSET=UTF8 COLLATE=UTF8_BIN;")
+	tk.MustExec("create table t1(id int) ENGINE=InnoDB DEFAULT CHARSET=UTF8 COLLATE=uTF8_BIN;")
+	tk.MustExec("create table t2(id int) ENGINE=InnoDB DEFAULT CHARSET=Utf8 COLLATE=utf8_BIN;")
+	tk.MustExec("create table t3(id int) ENGINE=InnoDB DEFAULT CHARSET=Utf8mb4 COLLATE=utf8MB4_BIN;")
+	tk.MustExec("create table t4(id int) ENGINE=InnoDB DEFAULT CHARSET=Utf8mb4 COLLATE=utf8MB4_general_ci;")
+}
+
 func newStoreWithBootstrap() (kv.Storage, *domain.Domain, error) {
 	store, err := mockstore.NewMockTikvStore()
 	if err != nil {

diff --git a/util/charset/charset.go b/util/charset/charset.go
@@ -86,6 +86,7 @@ func ValidCharsetAndCollation(cs string, co string) bool {
 		cs = "utf8"
 	}
 
+	cs = strings.ToLower(cs)
 	c, ok := charsets[cs]
 	if !ok {
 		return false
@@ -94,6 +95,7 @@ func ValidCharsetAndCollation(cs string, co string) bool {
 	if co == "" {
 		return true
 	}
+	co = strings.ToLower(co)
 	_, ok = c.Collations[co]
 	if !ok {
 		return false

diff --git a/util/charset/charset_test.go b/util/charset/charset_test.go
@@ -49,6 +49,12 @@ func (s *testCharsetSuite) TestValidCharset(c *C) {
 		{"utf8", "utf8_invalid_ci", false},
 		{"utf16", "utf16_bin", false},
 		{"gb2312", "gb2312_chinese_ci", false},
+		{"UTF8", "UTF8_BIN", true},
+		{"UTF8", "utf8_bin", true},
+		{"UTF8MB4", "utf8mb4_bin", true},
+		{"UTF8MB4", "UTF8MB4_bin", true},
+		{"UTF8MB4", "UTF8MB4_general_ci", true},
+		{"Utf8", "uTf8_bIN", true},
 	}
 	for _, tt := range tests {
 		testValidCharset(c, tt.cs, tt.co, tt.succ)