*: migrate some collation tests to explain test (pingcap#31993)

ref pingcap#31625

cmd/explaintest/r/collation_misc_disabled.result

@@ -0,0 +1,80 @@
+create database collation_misc;
+use collation_misc;
+create table t1(a varchar(20) charset utf8);
+insert into t1 values ("t1_value");
+alter table t1 collate uTf8mB4_uNiCoDe_Ci charset Utf8mB4 charset uTF8Mb4 collate UTF8MB4_BiN;
+alter table t1 modify column a varchar(20) charset utf8mb4;
+select * from t1;
+a
+t1_value
+create table t(a varchar(20) charset latin1);
+insert into t values ("t_value");
+alter table t modify column a varchar(20) charset latin1;
+select * from t;
+a
+t_value
+alter table t modify column a varchar(20) charset utf8;
+Error 8200: Unsupported modify charset from latin1 to utf8
+alter table t modify column a varchar(20) charset utf8mb4;
+Error 8200: Unsupported modify charset from latin1 to utf8mb4
+alter table t modify column a varchar(20) charset utf8 collate utf8_bin;
+Error 8200: Unsupported modify charset from latin1 to utf8
+alter table t modify column a varchar(20) charset utf8mb4 collate utf8mb4_general_ci;
+Error 8200: Unsupported modify charset from latin1 to utf8mb4
+alter table t modify column a varchar(20) charset utf8mb4 collate utf8bin;
+[ddl:1273]Unknown collation: 'utf8bin'
+alter table t collate LATIN1_GENERAL_CI charset utf8 collate utf8_bin;
+Error 1302: Conflicting declarations: 'CHARACTER SET latin1' and 'CHARACTER SET utf8'
+alter table t collate LATIN1_GENERAL_CI collate UTF8MB4_UNICODE_ci collate utf8_bin;
+Error 1253: COLLATION 'utf8mb4_unicode_ci' is not valid for CHARACTER SET 'latin1'
+create database if not exists cd_test_utf8 CHARACTER SET utf8 COLLATE utf8_bin;
+create database if not exists cd_test_latin1 CHARACTER SET latin1 COLLATE latin1_swedish_ci;
+use cd_test_utf8;
+select @@character_set_database;
+@@character_set_database
+utf8
+select @@collation_database;
+@@collation_database
+utf8_bin
+use cd_test_latin1;
+select @@character_set_database;
+@@character_set_database
+latin1
+select @@collation_database;
+@@collation_database
+latin1_swedish_ci
+create database if not exists test_db CHARACTER SET latin1 COLLATE latin1_swedish_ci;
+with cte as (select cast('2010-09-09' as date) a union select  '2010-09-09  ') select count(*) from cte;
+count(*)
+2
+set names utf8mb4 collate utf8mb4_general_ci;
+select position('a' in 'AA');
+position('a' in 'AA')
+0
+select locate('a', 'AA');
+locate('a', 'AA')
+0
+select locate('a', 'a');
+locate('a', 'a')
+1
+set names utf8mb4;
+SELECT default_collate_name, maxlen FROM information_schema.character_sets ORDER BY character_set_name;
+default_collate_name	maxlen
+ascii_bin	1
+binary	1
+gbk_bin	2
+latin1_bin	1
+utf8_bin	3
+utf8mb4_bin	4
+SELECT character_set_name, id, sortlen FROM information_schema.collations ORDER BY collation_name, id;
+character_set_name	id	sortlen
+ascii	65	1
+binary	63	1
+gbk	87	1
+latin1	47	1
+utf8	83	1
+utf8mb4	46	1
+select * from information_schema.COLLATION_CHARACTER_SET_APPLICABILITY where COLLATION_NAME='utf8mb4_bin';
+COLLATION_NAME	CHARACTER_SET_NAME
+utf8mb4_bin	utf8mb4
+use test;

cmd/explaintest/r/collation_misc_enabled.result

@@ -0,0 +1,88 @@
+create database collation_misc;
+use collation_misc;
+create table t1(a varchar(20) charset utf8);
+insert into t1 values ("t1_value");
+alter table t1 collate uTf8mB4_uNiCoDe_Ci charset Utf8mB4 charset uTF8Mb4 collate UTF8MB4_BiN;
+alter table t1 modify column a varchar(20) charset utf8mb4;
+select * from t1;
+a
+t1_value
+create table t(a varchar(20) charset latin1);
+insert into t values ("t_value");
+alter table t modify column a varchar(20) charset latin1;
+select * from t;
+a
+t_value
+alter table t modify column a varchar(20) charset utf8;
+Error 8200: Unsupported modify charset from latin1 to utf8
+alter table t modify column a varchar(20) charset utf8mb4;
+Error 8200: Unsupported modify charset from latin1 to utf8mb4
+alter table t modify column a varchar(20) charset utf8 collate utf8_bin;
+Error 8200: Unsupported modify charset from latin1 to utf8
+alter table t modify column a varchar(20) charset utf8mb4 collate utf8mb4_general_ci;
+Error 8200: Unsupported modify charset from latin1 to utf8mb4
+alter table t modify column a varchar(20) charset utf8mb4 collate utf8bin;
+[ddl:1273]Unknown collation: 'utf8bin'
+alter table t collate LATIN1_GENERAL_CI charset utf8 collate utf8_bin;
+Error 1273: Unsupported collation when new collation is enabled: 'latin1_general_ci'
+alter table t collate LATIN1_GENERAL_CI collate UTF8MB4_UNICODE_ci collate utf8_bin;
+Error 1273: Unsupported collation when new collation is enabled: 'latin1_general_ci'
+create database if not exists cd_test_utf8 CHARACTER SET utf8 COLLATE utf8_bin;
+create database if not exists cd_test_latin1 CHARACTER SET latin1 COLLATE latin1_swedish_ci;
+Error 1273: Unsupported collation when new collation is enabled: 'latin1_swedish_ci'
+use cd_test_utf8;
+select @@character_set_database;
+@@character_set_database
+utf8
+select @@collation_database;
+@@collation_database
+utf8_bin
+use cd_test_latin1;
+Error 1049: Unknown database 'cd_test_latin1'
+select @@character_set_database;
+@@character_set_database
+utf8
+select @@collation_database;
+@@collation_database
+utf8_bin
+create database if not exists test_db CHARACTER SET latin1 COLLATE latin1_swedish_ci;
+Error 1273: Unsupported collation when new collation is enabled: 'latin1_swedish_ci'
+with cte as (select cast('2010-09-09' as date) a union select  '2010-09-09  ') select count(*) from cte;
+count(*)
+1
+set names utf8mb4 collate utf8mb4_general_ci;
+select position('a' in 'AA');
+position('a' in 'AA')
+1
+select locate('a', 'AA');
+locate('a', 'AA')
+1
+select locate('a', 'a');
+locate('a', 'a')
+1
+set names utf8mb4;
+SELECT default_collate_name, maxlen FROM information_schema.character_sets ORDER BY character_set_name;
+default_collate_name	maxlen
+ascii_bin	1
+binary	1
+gbk_chinese_ci	2
+latin1_bin	1
+utf8_bin	3
+utf8mb4_bin	4
+SELECT character_set_name, id, sortlen FROM information_schema.collations ORDER BY collation_name, id;
+character_set_name	id	sortlen
+ascii	65	1
+binary	63	1
+gbk	87	1
+gbk	28	1
+latin1	47	1
+utf8	83	1
+utf8	33	1
+utf8	192	1
+utf8mb4	46	1
+utf8mb4	45	1
+utf8mb4	224	1
+select * from information_schema.COLLATION_CHARACTER_SET_APPLICABILITY where COLLATION_NAME='utf8mb4_bin';
+COLLATION_NAME	CHARACTER_SET_NAME
+utf8mb4_bin	utf8mb4
+use test;

cmd/explaintest/t/collation_misc.test

@@ -0,0 +1,61 @@
+# prepare database
+create database collation_misc;
+use collation_misc;
+
+# ChangingCharsetToUtf8
+create table t1(a varchar(20) charset utf8);
+insert into t1 values ("t1_value");
+alter table t1 collate uTf8mB4_uNiCoDe_Ci charset Utf8mB4 charset uTF8Mb4 collate UTF8MB4_BiN;
+alter table t1 modify column a varchar(20) charset utf8mb4;
+select * from t1;
+create table t(a varchar(20) charset latin1);
+insert into t values ("t_value");
+alter table t modify column a varchar(20) charset latin1;
+select * from t;
+--error 8200
+alter table t modify column a varchar(20) charset utf8;
+--error 8200
+alter table t modify column a varchar(20) charset utf8mb4;
+--error 8200
+alter table t modify column a varchar(20) charset utf8 collate utf8_bin;
+--error 8200
+alter table t modify column a varchar(20) charset utf8mb4 collate utf8mb4_general_ci;
+--error 1273
+alter table t modify column a varchar(20) charset utf8mb4 collate utf8bin;
+--error 1302, 1273
+alter table t collate LATIN1_GENERAL_CI charset utf8 collate utf8_bin;
+--error 1253, 1273
+alter table t collate LATIN1_GENERAL_CI collate UTF8MB4_UNICODE_ci collate utf8_bin;
+
+# TestCharsetDatabase
+create database if not exists cd_test_utf8 CHARACTER SET utf8 COLLATE utf8_bin;
+--error 1273
+create database if not exists cd_test_latin1 CHARACTER SET latin1 COLLATE latin1_swedish_ci;
+use cd_test_utf8;
+select @@character_set_database;
+select @@collation_database;
+--error 1049
+use cd_test_latin1;
+select @@character_set_database;
+select @@collation_database;
+
+# DefaultDBAfterDropCurDB
+--error 1273
+create database if not exists test_db CHARACTER SET latin1 COLLATE latin1_swedish_ci;
+
+# CollationUnion
+with cte as (select cast('2010-09-09' as date) a union select  '2010-09-09  ') select count(*) from cte;
+
+# Issue26989
+set names utf8mb4 collate utf8mb4_general_ci;
+select position('a' in 'AA');
+select locate('a', 'AA');
+select locate('a', 'a');
+set names utf8mb4;
+
+# CharacterSetCollations
+SELECT default_collate_name, maxlen FROM information_schema.character_sets ORDER BY character_set_name;
+SELECT character_set_name, id, sortlen FROM information_schema.collations ORDER BY collation_name, id;
+select * from information_schema.COLLATION_CHARACTER_SET_APPLICABILITY where COLLATION_NAME='utf8mb4_bin';
+
+use test;

ddl/db_integration_test.go

@@ -831,32 +831,6 @@ func (s *testIntegrationSuite2) TestDependedGeneratedColumnPrior2GeneratedColumn
 	tk.MustExec("alter table t add column(e int as (c+1))")
 }
 
-func (s *testIntegrationSuite3) TestChangingCharsetToUtf8(c *C) {
-	tk := testkit.NewTestKit(c, s.store)
-
-	tk.MustExec("use test")
-	tk.MustExec("create table t1(a varchar(20) charset utf8)")
-	tk.MustExec("insert into t1 values (?)", "t1_value")
-	tk.MustExec("alter table t1 collate uTf8mB4_uNiCoDe_Ci charset Utf8mB4 charset uTF8Mb4 collate UTF8MB4_BiN")
-	tk.MustExec("alter table t1 modify column a varchar(20) charset utf8mb4")
-	tk.MustQuery("select * from t1;").Check(testkit.Rows("t1_value"))
-
-	tk.MustExec("create table t(a varchar(20) charset latin1)")
-	tk.MustExec("insert into t values (?)", "t_value")
-
-	tk.MustExec("alter table t modify column a varchar(20) charset latin1")
-	tk.MustQuery("select * from t;").Check(testkit.Rows("t_value"))
-
-	tk.MustGetErrCode("alter table t modify column a varchar(20) charset utf8", errno.ErrUnsupportedDDLOperation)
-	tk.MustGetErrCode("alter table t modify column a varchar(20) charset utf8mb4", errno.ErrUnsupportedDDLOperation)
-	tk.MustGetErrCode("alter table t modify column a varchar(20) charset utf8 collate utf8_bin", errno.ErrUnsupportedDDLOperation)
-	tk.MustGetErrCode("alter table t modify column a varchar(20) charset utf8mb4 collate utf8mb4_general_ci", errno.ErrUnsupportedDDLOperation)
-
-	tk.MustGetErrCode("alter table t modify column a varchar(20) charset utf8mb4 collate utf8bin", errno.ErrUnknownCollation)
-	tk.MustGetErrCode("alter table t collate LATIN1_GENERAL_CI charset utf8 collate utf8_bin", errno.ErrConflictingDeclarations)
-	tk.MustGetErrCode("alter table t collate LATIN1_GENERAL_CI collate UTF8MB4_UNICODE_ci collate utf8_bin", errno.ErrCollationCharsetMismatch)
-}
-
 func (s *testIntegrationSuite4) TestChangingTableCharset(c *C) {
 	tk := testkit.NewTestKit(c, s.store)
 

executor/ddl_test.go

@@ -671,25 +671,6 @@ func (s *testSuite6) TestAlterTableModifyColumn(c *C) {
 
 }
 
-func (s *testSuite6) TestDefaultDBAfterDropCurDB(c *C) {
-	tk := testkit.NewTestKit(c, s.store)
-
-	testSQL := `create database if not exists test_db CHARACTER SET latin1 COLLATE latin1_swedish_ci;`
-	tk.MustExec(testSQL)
-
-	testSQL = `use test_db;`
-	tk.MustExec(testSQL)
-	tk.MustQuery(`select database();`).Check(testkit.Rows("test_db"))
-	tk.MustQuery(`select @@character_set_database;`).Check(testkit.Rows("latin1"))
-	tk.MustQuery(`select @@collation_database;`).Check(testkit.Rows("latin1_swedish_ci"))
-
-	testSQL = `drop database test_db;`
-	tk.MustExec(testSQL)
-	tk.MustQuery(`select database();`).Check(testkit.Rows("<nil>"))
-	tk.MustQuery(`select @@character_set_database;`).Check(testkit.Rows(mysql.DefaultCharset))
-	tk.MustQuery(`select @@collation_database;`).Check(testkit.Rows(mysql.DefaultCollationName))
-}
-
 func (s *testSuite6) TestColumnCharsetAndCollate(c *C) {
 	tk := testkit.NewTestKit(c, s.store)
 	dbName := "col_charset_collate"

executor/infoschema_reader_test.go

@@ -218,22 +218,6 @@ func (s *testInfoschemaTableSuite) TestEngines(c *C) {
 	tk.MustQuery("select * from information_schema.ENGINES;").Check(testkit.Rows("InnoDB DEFAULT Supports transactions, row-level locking, and foreign keys YES YES YES"))
 }
 
-func (s *testInfoschemaTableSuite) TestCharacterSetCollations(c *C) {
-	tk := testkit.NewTestKit(c, s.store)
-
-	// The description column is not important
-	tk.MustQuery("SELECT default_collate_name, maxlen FROM information_schema.character_sets ORDER BY character_set_name").Check(
-		testkit.Rows("ascii_bin 1", "binary 1", "gbk_bin 2", "latin1_bin 1", "utf8_bin 3", "utf8mb4_bin 4"))
-
-	// The is_default column is not important
-	// but the id's are used by client libraries and must be stable
-	tk.MustQuery("SELECT character_set_name, id, sortlen FROM information_schema.collations ORDER BY collation_name").Check(
-		testkit.Rows("ascii 65 1", "binary 63 1", "gbk 87 1", "latin1 47 1", "utf8 83 1", "utf8mb4 46 1"))
-
-	tk.MustQuery("select * from information_schema.COLLATION_CHARACTER_SET_APPLICABILITY where COLLATION_NAME='utf8mb4_bin';").Check(
-		testkit.Rows("utf8mb4_bin utf8mb4"))
-}
-
 // https://github.com/pingcap/tidb/issues/25467.
 func (s *testInfoschemaTableSuite) TestDataTypesMaxLengthAndOctLength(c *C) {
 	tk := testkit.NewTestKit(c, s.store)

executor/simple_test.go

@@ -38,25 +38,6 @@ import (
 	"github.com/pingcap/tidb/util/testutil"
 )
 
-func (s *testSuite3) TestCharsetDatabase(c *C) {
-	tk := testkit.NewTestKit(c, s.store)
-	testSQL := `create database if not exists cd_test_utf8 CHARACTER SET utf8 COLLATE utf8_bin;`
-	tk.MustExec(testSQL)
-
-	testSQL = `create database if not exists cd_test_latin1 CHARACTER SET latin1 COLLATE latin1_swedish_ci;`
-	tk.MustExec(testSQL)
-
-	testSQL = `use cd_test_utf8;`
-	tk.MustExec(testSQL)
-	tk.MustQuery(`select @@character_set_database;`).Check(testkit.Rows("utf8"))
-	tk.MustQuery(`select @@collation_database;`).Check(testkit.Rows("utf8_bin"))
-
-	testSQL = `use cd_test_latin1;`
-	tk.MustExec(testSQL)
-	tk.MustQuery(`select @@character_set_database;`).Check(testkit.Rows("latin1"))
-	tk.MustQuery(`select @@collation_database;`).Check(testkit.Rows("latin1_swedish_ci"))
-}
-
 func (s *testSuite3) TestDo(c *C) {
 	tk := testkit.NewTestKit(c, s.store)
 	tk.MustExec("do 1, @a:=1")

expression/integration_serial_test.go

@@ -1450,22 +1450,6 @@ func TestLikeWithCollation(t *testing.T) {
 	tk.MustQuery(`select '😛' collate utf8mb4_unicode_ci = '😋';`).Check(testkit.Rows("1"))
 }
 
-func TestCollationUnion(t *testing.T) {
-	// For issue 19694.
-	store, clean := testkit.CreateMockStore(t)
-	defer clean()
-
-	tk := testkit.NewTestKit(t, store)
-
-	tk.MustQuery("select cast('2010-09-09' as date) a union select  '2010-09-09  ' order by a;").Check(testkit.Rows("2010-09-09", "2010-09-09  "))
-	res := tk.MustQuery("select cast('2010-09-09' as date) a union select  '2010-09-09  ';")
-	require.Len(t, res.Rows(), 2)
-	collate.SetNewCollationEnabledForTest(true)
-	defer collate.SetNewCollationEnabledForTest(false)
-	res = tk.MustQuery("select cast('2010-09-09' as date) a union select  '2010-09-09  ';")
-	require.Len(t, res.Rows(), 1)
-}
-
 func TestCollationPrefixClusteredIndex(t *testing.T) {
 	store, clean := testkit.CreateMockStore(t)
 	defer clean()

expression/integration_test.go

@@ -4763,17 +4763,6 @@ func TestIssue17287(t *testing.T) {
 	tk.MustQuery("execute stmt7 using @val2;").Check(testkit.Rows("1589873946"))
 }
 
-func TestIssue26989(t *testing.T) {
-	store, clean := testkit.CreateMockStore(t)
-	defer clean()
-
-	tk := testkit.NewTestKit(t, store)
-	tk.MustExec("set names utf8mb4 collate utf8mb4_general_ci;")
-	tk.MustQuery("select position('a' in 'AA');").Check(testkit.Rows("0"))
-	tk.MustQuery("select locate('a', 'AA');").Check(testkit.Rows("0"))
-	tk.MustQuery("select locate('a', 'a');").Check(testkit.Rows("1"))
-}
-
 func TestIssue17898(t *testing.T) {
 	store, clean := testkit.CreateMockStore(t)
 	defer clean()