cherry pick pingcap#37242 to release-6.2

Signed-off-by: ti-srebot <ti-srebot@pingcap.com>
ti-srebot · Aug 24, 2022 · 79a40c4 · 79a40c4
1 parent daf2b17
commit 79a40c4
Show file tree

Hide file tree

Showing 4 changed files with 86 additions and 2 deletions.
diff --git a/executor/executor_issue_test.go b/executor/executor_issue_test.go
@@ -26,6 +26,7 @@ import (
 	"github.com/pingcap/tidb/config"
 	"github.com/pingcap/tidb/kv"
 	"github.com/pingcap/tidb/parser/auth"
+	"github.com/pingcap/tidb/parser/charset"
 	"github.com/pingcap/tidb/parser/mysql"
 	"github.com/pingcap/tidb/sessionctx/variable"
 	"github.com/pingcap/tidb/statistics"
@@ -409,6 +410,70 @@ func TestIssue30971(t *testing.T) {
 	}
 }
 
+func TestIssue31678(t *testing.T) {
+	// The issue31678 is mainly about type conversion in UNION
+	store := testkit.CreateMockStore(t)
+	tk := testkit.NewTestKit(t, store)
+	tk.MustExec("USE test")
+	tk.MustExec("DROP TABLE IF EXISTS t1, t2;")
+
+	// https://github.com/pingcap/tidb/issues/31678
+	tk.MustExec("CREATE TABLE t1 (c VARCHAR(11)) CHARACTER SET utf8mb4")
+	tk.MustExec("CREATE TABLE t2 (b CHAR(1) CHARACTER SET binary, i INT)")
+	tk.MustExec("INSERT INTO t1 (c) VALUES ('н1234567890')")
+	tk.MustExec("INSERT INTO t2 (b, i) VALUES ('1', 1)")
+	var tests = []struct {
+		query           string
+		expectedFlen    int
+		expectedCharset string
+		result          []string
+	}{
+		{"SELECT c FROM t1 UNION SELECT b FROM t2", 44, "binary", []string{"1", "н1234567890"}},
+		{"SELECT c FROM t1 UNION SELECT i FROM t2", 20, "utf8mb4", []string{"1", "н1234567890"}},
+		{"SELECT i FROM t2 UNION SELECT c FROM t1", 20, "utf8mb4", []string{"1", "н1234567890"}},
+		{"SELECT b FROM t2 UNION SELECT c FROM t1", 44, "binary", []string{"1", "н1234567890"}},
+	}
+	for _, test := range tests {
+		tk.MustQuery(test.query).Sort().Check(testkit.Rows(test.result...))
+		rs, err := tk.Exec(test.query)
+		require.NoError(t, err)
+		resultFields := rs.Fields()
+		require.Equal(t, 1, len(resultFields), test.query)
+		require.Equal(t, test.expectedFlen, resultFields[0].Column.FieldType.GetFlen(), test.query)
+		require.Equal(t, test.expectedCharset, resultFields[0].Column.FieldType.GetCharset(), test.query)
+	}
+	tk.MustExec("DROP TABLE t1, t2;")
+
+	// test some other charset
+	tk.MustExec("CREATE TABLE t1 (c1 VARCHAR(5) CHARACTER SET utf8mb4, c2 VARCHAR(1) CHARACTER SET binary)")
+	tk.MustExec("CREATE TABLE t2 (c1 CHAR(10) CHARACTER SET GBK, c2 VARCHAR(50) CHARACTER SET binary)")
+	tk.MustExec("INSERT INTO t1 VALUES ('一二三四五', '1')")
+	tk.MustExec("INSERT INTO t2 VALUES ('一二三四五六七八九十', '1234567890')")
+	gbkResult, err := charset.NewCustomGBKEncoder().String("一二三四五六七八九十")
+	require.NoError(t, err)
+	tests = []struct {
+		query           string
+		expectedFlen    int
+		expectedCharset string
+		result          []string
+	}{
+		{"SELECT c1 FROM t1 UNION SELECT c1 FROM t2", 10, "utf8mb4", []string{"一二三四五", "一二三四五六七八九十"}},
+		{"SELECT c1 FROM t1 UNION SELECT c2 FROM t2", 50, "binary", []string{"1234567890", "一二三四五"}},
+		{"SELECT c2 FROM t1 UNION SELECT c1 FROM t2", 20, "binary", []string{"1", gbkResult}},
+		{"SELECT c2 FROM t1 UNION SELECT c2 FROM t2", 50, "binary", []string{"1", "1234567890"}},
+	}
+	for _, test := range tests {
+		tk.MustQuery(test.query).Sort().Check(testkit.Rows(test.result...))
+		rs, err := tk.Exec(test.query)
+		require.NoError(t, err)
+		resultFields := rs.Fields()
+		require.Equal(t, 1, len(resultFields), test.query)
+		require.Equal(t, test.expectedFlen, resultFields[0].Column.FieldType.GetFlen(), test.query)
+		require.Equal(t, test.expectedCharset, resultFields[0].Column.FieldType.GetCharset(), test.query)
+	}
+	tk.MustExec("DROP TABLE t1, t2;")
+}
+
 func TestIndexJoin31494(t *testing.T) {
 	store, dom, clean := testkit.CreateMockStoreAndDomain(t)
 	defer clean()

diff --git a/planner/core/logical_plan_builder.go b/planner/core/logical_plan_builder.go
@@ -34,6 +34,7 @@ import (
 	"github.com/pingcap/tidb/metrics"
 	"github.com/pingcap/tidb/parser"
 	"github.com/pingcap/tidb/parser/ast"
+	"github.com/pingcap/tidb/parser/charset"
 	"github.com/pingcap/tidb/parser/format"
 	"github.com/pingcap/tidb/parser/model"
 	"github.com/pingcap/tidb/parser/mysql"
@@ -1507,7 +1508,24 @@ func unionJoinFieldType(a, b *types.FieldType) *types.FieldType {
 	return resultTp
 }
 
+<<<<<<< HEAD
 func (b *PlanBuilder) buildProjection4Union(ctx context.Context, u *LogicalUnionAll) error {
+=======
+// Set the flen of the union column using the max flen in children.
+func (b *PlanBuilder) setUnionFlen(resultTp *types.FieldType, cols []expression.Expression) {
+	isBinary := resultTp.GetCharset() == charset.CharsetBin
+	for i := 0; i < len(cols); i++ {
+		childTp := cols[i].GetType()
+		childTpCharLen := 1
+		if isBinary {
+			childTpCharLen = charset.CharacterSetInfos[childTp.GetCharset()].Maxlen
+		}
+		resultTp.SetFlen(mathutil.Max(resultTp.GetFlen(), childTpCharLen*childTp.GetFlen()))
+	}
+}
+
+func (b *PlanBuilder) buildProjection4Union(_ context.Context, u *LogicalUnionAll) error {
+>>>>>>> 25dda978f... types: fix a bug in casting str2str when `union` (#37242)
 	unionCols := make([]*expression.Column, 0, u.children[0].Schema().Len())
 	names := make([]*types.FieldName, 0, u.children[0].Schema().Len())
 
@@ -1527,6 +1545,7 @@ func (b *PlanBuilder) buildProjection4Union(ctx context.Context, u *LogicalUnion
 		}
 		resultTp.SetCharset(collation.Charset)
 		resultTp.SetCollate(collation.Collation)
+		b.setUnionFlen(resultTp, tmpExprs)
 		names = append(names, &types.FieldName{ColName: u.children[0].OutputNames()[i].ColName})
 		unionCols = append(unionCols, &expression.Column{
 			RetType:  resultTp,

diff --git a/types/convert_test.go b/types/convert_test.go
@@ -389,7 +389,7 @@ func TestConvertToString(t *testing.T) {
 		sc := new(stmtctx.StatementContext)
 		outputDatum, err := inputDatum.ConvertTo(sc, ft)
 		if tt.input != tt.output {
-			require.True(t, ErrDataTooLong.Equal(err))
+			require.True(t, ErrDataTooLong.Equal(err), "flen: %d, charset: %s, input: %s, output: %s", tt.flen, tt.charset, tt.input, tt.output)
 		} else {
 			require.NoError(t, err)
 		}

diff --git a/types/datum.go b/types/datum.go
@@ -1069,7 +1069,7 @@ func (d *Datum) convertToString(sc *stmtctx.StatementContext, target *FieldType)
 func ProduceStrWithSpecifiedTp(s string, tp *FieldType, sc *stmtctx.StatementContext, padZero bool) (_ string, err error) {
 	flen, chs := tp.GetFlen(), tp.GetCharset()
 	if flen >= 0 {
-		// overflowed stores the part of the string that is out of the length contraint, it is later checked to see if the
+		// overflowed stores the part of the string that is out of the length constraint, it is later checked to see if the
 		// overflowed part is all whitespaces
 		var overflowed string
 		var characterLen int