This is an automated cherry-pick of pingcap#61080

xhebox · xhebox · commit 79710e2bb3d3 · 2025-05-21T15:24:10.000+08:00
Signed-off-by: ti-chi-bot &lt;ti-community-prow-bot@tidb.io&gt;
Signed-off-by: xhe &lt;xw897002528@gmail.com&gt;
diff --git a/cmd/explaintest/r/new_character_set.result b/cmd/explaintest/r/new_character_set.result
@@ -2,129 +2,103 @@ set tidb_cost_model_version=1;
 drop table if exists t;
 set names utf8mb4;
 create table t (a varchar(255) charset utf8mb4);
-insert into t values ('一二三四一');
+insert into t values ('ä¸äºä¸åä¸');
 select hex(a) from t;
 hex(a)
 E4B880E4BA8CE4B889E59B9BE4B880
 select a from t;
 a
-一二三四一
+ä¸äºä¸åä¸
 set @@character_set_results = 'gbk';
 select a from t;
 a
-һ������һ
-drop table if exists 一;
-create table 一 (二 char(20));
-show create table 一;
+Ò»¶þÈýËÄÒ»
+drop table if exists ä¸;
+create table ä¸ (äº char(20));
+show create table ä¸;
 Table	Create Table
-һ	CREATE TABLE `һ` (
-  `��` char(20) DEFAULT NULL
+Ò»	CREATE TABLE `Ò»` (
+  `¶þ` char(20) DEFAULT NULL
 ) ENGINE=InnoDB DEFAULT CHARSET=utf8mb4 COLLATE=utf8mb4_bin
 drop table if exists t;
 set names utf8mb4;
 create table t (a varchar(255) charset gbk, b varchar(255) charset utf8mb4, c varchar(255) charset binary);
-insert into t values ('一', '一', '一');
+insert into t values ('ä¸', 'ä¸', 'ä¸');
 set @@character_set_results = null;
 select * from t;
 a	b	c
-һ	一	一
+Ò»	ä¸	ä¸
 set @@character_set_results = BINARY;
 select * from t;
 a	b	c
-һ	一	一
+Ò»	ä¸	ä¸
 set @@character_set_results = "BINARY";
 select * from t;
 a	b	c
-һ	一	一
+Ò»	ä¸	ä¸
 set names utf8mb4;
 select * from t;
 a	b	c
-一	一	一
+ä¸	ä¸	ä¸
 set @@character_set_results = 'utf8mb4';
 drop table if exists t;
 create table t (a varchar(255) charset utf8mb4);
 set @@character_set_client = 'gbk';
-insert into t values ('中文');
+insert into t values ('ä¸­æ');
 set @@character_set_client = 'utf8mb4';
-insert into t values ('中文');
+insert into t values ('ä¸­æ');
 set @@character_set_client = 'gbk';
-prepare p1 from "insert into t values ('中文');";
+prepare p1 from "insert into t values ('ä¸­æ');";
 execute p1;
 
 select a, hex(a) from t;
 a	hex(a)
-涓?枃	E6B6933FE69E83
-中文	E4B8ADE69687
-涓?枃	E6B6933FE69E83
+æ¶?æ	E6B6933FE69E83
+ä¸­æ	E4B8ADE69687
+æ¶?æ	E6B6933FE69E83
 set @@character_set_client = 'utf8mb4';
 set names gbk;
 drop table if exists t;
 create table t (b blob, d json);
-insert into t values ('你好', '{"测试": "你好"}');
+insert into t values ('ä½ å¥½', '{"æµè¯": "ä½ å¥½"}');
 select b, d from t;
 b	d
-你好	{"测试": "你好"}
+ä½ å¥½	{"æµè¯": "ä½ å¥½"}
 select hex(b), hex(d) from t;
 hex(b)	hex(d)
 E4BDA0E5A5BD	7B22E6B58BE8AF95223A2022E4BDA0E5A5BD227D
 set names utf8mb4;
 drop table if exists t;
 create table t(a blob, b char(10));
-insert into t values (0x61, '啊');
-insert into t values (0x61, '一');
+insert into t values (0x61, 'å');
+insert into t values (0x61, 'ä¸');
 set names gbk;
 select * from t;
 a	b
-a	��
-a	һ
+a	°¡
+a	Ò»
 drop table t;
 set names gbk;
 set character_set_connection = utf8mb4;
-create table t(j json, b blob, s1 varchar(255) collate binary, s2 varchar(255), st set('{"点赞": "你好"}'), en enum('{"点赞": "你好"}'));
-insert into t values('{"点赞": "你好"}', '{"点赞": "你好"}', '{"点赞": "你好"}', '{"点赞": "你好"}', '{"点赞": "你好"}', '{"点赞": "你好"}');
+create table t(j json, b blob, s1 varchar(255) collate binary, s2 varchar(255), st set('{"ç¹èµ": "ä½ å¥½"}'), en enum('{"ç¹èµ": "ä½ å¥½"}'));
+insert into t values('{"ç¹èµ": "ä½ å¥½"}', '{"ç¹èµ": "ä½ å¥½"}', '{"ç¹èµ": "ä½ å¥½"}', '{"ç¹èµ": "ä½ å¥½"}', '{"ç¹èµ": "ä½ å¥½"}', '{"ç¹èµ": "ä½ å¥½"}');
 select * from t;
 j	b	s1	s2	st	en
-{"点赞": "你好"}	{"鐐硅禐": "浣犲ソ"}	{"鐐硅禐": "浣犲ソ"}	{"点赞": "你好"}	{"点赞": "你好"}	{"点赞": "你好"}
+{"ç¹èµ": "ä½ å¥½"}	{"éç¡ç¦": "æµ£ç²ã½"}	{"éç¡ç¦": "æµ£ç²ã½"}	{"ç¹èµ": "ä½ å¥½"}	{"ç¹èµ": "ä½ å¥½"}	{"ç¹èµ": "ä½ å¥½"}
 set names utf8mb4;
 set @@character_set_client=gbk;
 set @@character_set_connection=gbk;
-select hex('一a'), '一a';
-hex('涓?')	涓?
-E4B83F	涓?
-<<<<<<< HEAD:cmd/explaintest/r/new_character_set.result
-=======
-set @@sql_mode=default;
-set @@character_set_client=default;
-set @@character_set_connection=default;
-show variables like 'collation_connection';
-Variable_name	Value
-collation_connection	utf8mb4_bin
-set default_collation_for_utf8mb4 = 'utf8mb4_general_ci';
-set names utf8mb4;
-show variables like 'collation_connection';
-Variable_name	Value
-collation_connection	utf8mb4_general_ci
-set default_collation_for_utf8mb4 = 'utf8mb4_0900_ai_ci';
-set names utf8mb4;
-show variables like 'collation_connection';
-Variable_name	Value
-collation_connection	utf8mb4_0900_ai_ci
-set default_collation_for_utf8mb4 = default;
-set names utf8mb4;
-show variables like 'collation_connection';
-Variable_name	Value
-collation_connection	utf8mb4_bin
+select hex('ä¸a'), 'ä¸a';
+hex('æ¶?')	æ¶?
+E4B83F	æ¶?
 set character_set_results = "gbk";
 select cast(0x414141E280A9424242 as char charset utf8mb4);
 cast(0x414141E280A9424242 as char charset utf8mb4)
 AAA?BBB
 SET character_set_results = @undefined_var;
-DROP TABLE if exists t61085;
 create table t61085 (a char(255) charset gbk);
 insert into t61085 values ('AAA');
 set SESSION sql_mode = '';
 select * from t61085 where a = cast(0x41414180424242 as char charset gbk);
-a
-AAA
+Error 1105: Cannot convert string 'AAA\x80BB...' from binary to gbk
 DROP TABLE t61085;
->>>>>>> d5dcbdf3792 (server: replace instead of truncate encoded result (#61080)):tests/integrationtest/r/new_character_set.result
diff --git a/cmd/explaintest/t/new_character_set.test b/cmd/explaintest/t/new_character_set.test
@@ -67,32 +67,14 @@ set names utf8mb4;
 set @@character_set_client=gbk;
 set @@character_set_connection=gbk;
 select hex('一a'), '一a';
-<<<<<<< HEAD:cmd/explaintest/t/new_character_set.test
-=======
 
-set @@sql_mode=default;
-set @@character_set_client=default;
-set @@character_set_connection=default;
-
-show variables like 'collation_connection';
-set default_collation_for_utf8mb4 = 'utf8mb4_general_ci';
-set names utf8mb4;
-show variables like 'collation_connection';
-set default_collation_for_utf8mb4 = 'utf8mb4_0900_ai_ci';
-set names utf8mb4;
-show variables like 'collation_connection';
-set default_collation_for_utf8mb4 = default;
-set names utf8mb4;
-show variables like 'collation_connection';
-
-# Bug#61085: https://github.com/pingcap/tidb/issues/61085 should replace instead of truncation for result charset
+-- Bug#61085: https://github.com/pingcap/tidb/issues/61085 should replace instead of truncation for result charset
 set character_set_results = "gbk";
 select cast(0x414141E280A9424242 as char charset utf8mb4);
 SET character_set_results = @undefined_var;
-DROP TABLE if exists t61085;
 create table t61085 (a char(255) charset gbk);
 insert into t61085 values ('AAA');
 set SESSION sql_mode = '';
+--error 1105
 select * from t61085 where a = cast(0x41414180424242 as char charset gbk);
 DROP TABLE t61085;
->>>>>>> d5dcbdf3792 (server: replace instead of truncate encoded result (#61080)):tests/integrationtest/t/new_character_set.test
diff --git a/server/util.go b/server/util.go
@@ -385,7 +385,7 @@ func (d *resultEncoder) encodeData(src []byte) []byte {
 }
 
 func (d *resultEncoder) encodeWith(src []byte, enc charset.Encoding) []byte {
-	data, err := enc.Transform(d.buffer, src, charset.OpEncode)
+	data, err := enc.Transform(d.buffer, src, charset.OpEncodeReplace)
 	if err != nil {
 		logutil.BgLogger().Debug("encode error", zap.Error(err))
 	}

Original file line number	Diff line number	Diff line change
`@@ -385,7 +385,7 @@ func (d *resultEncoder) encodeData(src []byte) []byte {`
`385`	`385`	`}`
`386`	`386`
`387`	`387`	`func (d *resultEncoder) encodeWith(src []byte, enc charset.Encoding) []byte {`
`388`		`- data, err := enc.Transform(d.buffer, src, charset.OpEncode)`
	`388`	`+ data, err := enc.Transform(d.buffer, src, charset.OpEncodeReplace)`
`389`	`389`	`if err != nil {`
`390`	`390`	`logutil.BgLogger().Debug("encode error", zap.Error(err))`
`391`	`391`	`}`