syncer(dm): fix wrong MySQL latin1 decoder (#8705) (#8710)

ref #7028

dm/syncer/dml.go

@@ -79,7 +79,9 @@ type genDMLParam struct {
 	extendData      [][]interface{}     // all data include extend data
 }
 
-var latin1Decoder = charmap.ISO8859_1.NewDecoder()
+// latin1Decider is not usually ISO8859_1 in MySQL.
+// ref https://dev.mysql.com/doc/refman/8.0/en/charset-we-sets.html
+var latin1Decoder = charmap.Windows1252.NewDecoder()
 
 // extractValueFromData adjust the values obtained from go-mysql so that
 // - the values can be correctly converted to TiDB datum

dm/tests/expression_filter/conf/dm-task2.yaml

@@ -49,7 +49,7 @@ expression-filter:
   only_muller:
     schema: "expr_filter"
     table: "t6"
-    insert-value-expr: "name != 'Müller'"
+    insert-value-expr: "name != 'Müller…'"
 
 black-white-list: # compatible with deprecated config
   instance:

dm/tests/expression_filter/data/db1.increment2.sql

@@ -41,8 +41,8 @@ update t5 set should_skip = 1, c = 2 where c = 1;
 insert into t5 values (4, 1, 1); -- check this `should_skip = 1` row must be updated to `should_skip = 0` in TiDB
 update t5 set should_skip = 0, c = 3 where c = 1;
 
-insert into t6 (id, name, msg) values (1, 'Müller', 'Müller'), (2, 'X Æ A-12', 'X Æ A-12');
-alter table t6 add column name2 varchar(20) character set latin1 default 'Müller';
+insert into t6 (id, name, msg) values (1, 'Müller…', 'Müller…'), (2, 'X Æ A-12', 'X Æ A-12');
+alter table t6 add column name2 varchar(20) character set latin1 default 'Müller…';
 
 -- trigger a flush
 alter table t5 add column dummy int;

dm/tests/expression_filter/data/db1.prepare2.sql

@@ -9,4 +9,4 @@ create table t2 (id int primary key,
 );
 
 create table t6 (id int, name varchar(20), msg text, primary key(`id`)) character set latin1;
-insert into t6 (id, name, msg) values (0, 'Müller', 'Müller');
+insert into t6 (id, name, msg) values (0, 'Müller…', 'Müller…');

dm/tests/expression_filter/run.sh

@@ -60,9 +60,9 @@ function complex_behaviour() {
 	run_sql_tidb "select count(8) from expr_filter.t5 where should_skip = 1"
 	check_contains "count(8): 0"
 
-	run_sql_tidb "select count(9) from expr_filter.t6 where name = 'Müller' and msg = 'Müller' and name2 = 'Müller'"
+	run_sql_tidb "select count(9) from expr_filter.t6 where name = 'Müller…' and msg = 'Müller…' and name2 = 'Müller…'"
 	check_contains "count(9): 2"
-	run_sql_tidb "select count(10) from expr_filter.t6 where name != 'Müller'"
+	run_sql_tidb "select count(10) from expr_filter.t6 where name != 'Müller…'"
 	check_contains "count(10): 0"
 
 	insert_num=$(grep -o '"number of filtered insert"=[0-9]\+' $WORK_DIR/worker1/log/dm-worker.log | grep -o '[0-9]\+' | awk '{n += $1}; END{print n}')