This is an automated cherry-pick of pingcap#51319

Signed-off-by: ti-chi-bot <[email protected]>

bindinfo/bind_test.go

@@ -1388,3 +1388,47 @@ func TestDropBindBySQLDigest(t *testing.T) {
 	tk.MustGetErrMsg(fmt.Sprintf("drop binding for sql digest '%s'", "1"), "can't find any binding for '1'")
 	tk.MustGetErrMsg(fmt.Sprintf("drop binding for sql digest '%s'", ""), "sql digest is empty")
 }
+<<<<<<< HEAD:bindinfo/bind_test.go
+=======
+
+func TestJoinOrderHintWithBinding(t *testing.T) {
+	store := testkit.CreateMockStore(t)
+
+	tk := testkit.NewTestKit(t, store)
+	tk.MustExec("use test")
+	tk.MustExec("drop table if exists t, t1, t2, t3;")
+	tk.MustExec("create table t(a int, b int, key(a));")
+	tk.MustExec("create table t1(a int, b int, key(a));")
+	tk.MustExec("create table t2(a int, b int, key(a));")
+	tk.MustExec("create table t3(a int, b int, key(a));")
+
+	tk.MustExec("create global binding for select * from t1 join t2 on t1.a=t2.a left join t3 on t2.b=t3.b using select /*+ leading(t2) */ * from t1 join t2 on t1.a=t2.a left join t3 on t2.b=t3.b")
+	tk.MustExec("select * from t1 join t2 on t1.a=t2.a left join t3 on t2.b=t3.b")
+	tk.MustQuery("select @@last_plan_from_binding").Check(testkit.Rows("1"))
+	res := tk.MustQuery("show global bindings").Rows()
+	require.Equal(t, res[0][0], "select * from ( `test` . `t1` join `test` . `t2` on `t1` . `a` = `t2` . `a` ) left join `test` . `t3` on `t2` . `b` = `t3` . `b`")
+
+	tk.MustExec("drop global binding for select * from t1 join t2 on t1.a=t2.a join t3 on t2.b=t3.b")
+}
+
+func TestNormalizeStmtForBinding(t *testing.T) {
+	tests := []struct {
+		sql        string
+		normalized string
+		digest     string
+	}{
+		{"select 1 from b where (x,y) in ((1, 3), ('3', 1))", "select ? from `b` where row ( `x` , `y` ) in ( ... )", "ab6c607d118c24030807f8d1c7c846ec23e3b752fd88ed763bb8e26fbfa56a83"},
+		{"select 1 from b where (x,y) in ((1, 3), ('3', 1), (2, 3))", "select ? from `b` where row ( `x` , `y` ) in ( ... )", "ab6c607d118c24030807f8d1c7c846ec23e3b752fd88ed763bb8e26fbfa56a83"},
+		{"select 1 from b where (x,y) in ((1, 3), ('3', 1), (2, 3),('x', 'y'))", "select ? from `b` where row ( `x` , `y` ) in ( ... )", "ab6c607d118c24030807f8d1c7c846ec23e3b752fd88ed763bb8e26fbfa56a83"},
+		{"select 1 from b where (x,y) in ((1, 3), ('3', 1), (2, 3),('x', 'y'),('x', 'y'))", "select ? from `b` where row ( `x` , `y` ) in ( ... )", "ab6c607d118c24030807f8d1c7c846ec23e3b752fd88ed763bb8e26fbfa56a83"},
+		{"select 1 from b where (x) in ((1), ('3'), (2),('x'),('x'))", "select ? from `b` where ( `x` ) in ( ( ... ) )", "03e6e1eb3d76b69363922ff269284b359ca73351001ba0e82d3221c740a6a14c"},
+		{"select 1 from b where (x) in ((1), ('3'), (2),('x'))", "select ? from `b` where ( `x` ) in ( ( ... ) )", "03e6e1eb3d76b69363922ff269284b359ca73351001ba0e82d3221c740a6a14c"},
+	}
+	for _, test := range tests {
+		stmt, _, _ := internal.UtilNormalizeWithDefaultDB(t, test.sql)
+		n, digest := norm.NormalizeStmtForBinding(stmt, norm.WithFuzz(true))
+		require.Equal(t, test.normalized, n)
+		require.Equal(t, test.digest, digest)
+	}
+}
+>>>>>>> 707b0a4e381 (parser: support (Row(..),Row(..))=>(..) in the binding mode (#51319)):pkg/bindinfo/tests/bind_test.go

parser/digester.go

@@ -192,7 +192,20 @@ func (d *sqlDigester) normalize(sql string, keepHint bool) {
 			continue
 		}
 
+<<<<<<< HEAD:parser/digester.go
 		d.reduceLit(&currTok)
+=======
+		d.reduceLit(&currTok, forBinding)
+		if forPlanReplayerReload {
+			// Apply for plan replayer to match specific rules, changing IN (...) to IN (?). This can avoid plan replayer load failures caused by parse errors.
+			d.replaceSingleLiteralWithInList(&currTok)
+		} else if forBinding {
+			// Apply binding matching specific rules, IN (?) => IN ( ... ) #44298
+			d.reduceInListWithSingleLiteral(&currTok)
+			// In (Row(...)) => In (...) #51222
+			d.reduceInRowListWithSingleLiteral(&currTok)
+		}
+>>>>>>> 707b0a4e381 (parser: support (Row(..),Row(..))=>(..) in the binding mode (#51319)):pkg/parser/digester.go
 
 		if currTok.tok == identifier {
 			if strings.HasPrefix(currTok.lit, "_") {
@@ -267,7 +280,7 @@ func (d *sqlDigester) reduceOptimizerHint(tok *token) (reduced bool) {
 	return
 }
 
-func (d *sqlDigester) reduceLit(currTok *token) {
+func (d *sqlDigester) reduceLit(currTok *token, forBinding bool) {
 	if !d.isLit(*currTok) {
 		return
 	}
@@ -294,6 +307,29 @@ func (d *sqlDigester) reduceLit(currTok *token) {
 		return
 	}
 
+<<<<<<< HEAD:parser/digester.go
+=======
+	// Aggressive reduce lists.
+	last4 := d.tokens.back(4)
+	if d.isGenericLists(last4) {
+		d.tokens.popBack(4)
+		currTok.tok = genericSymbolList
+		currTok.lit = "..."
+		return
+	}
+	// reduce "In (row(...), row(...))" to "In (row(...))"
+	// final, it will be reduced to "In (...)". Issue: #51222
+	if forBinding {
+		last9 := d.tokens.back(9)
+		if d.isGenericRowListsWithIn(last9) {
+			d.tokens.popBack(5)
+			currTok.tok = genericSymbolList
+			currTok.lit = "..."
+			return
+		}
+	}
+
+>>>>>>> 707b0a4e381 (parser: support (Row(..),Row(..))=>(..) in the binding mode (#51319)):pkg/parser/digester.go
 	// order by n => order by n
 	if currTok.tok == intLit {
 		if d.isOrderOrGroupBy() {
@@ -306,6 +342,110 @@ func (d *sqlDigester) reduceLit(currTok *token) {
 	currTok.lit = "?"
 }
 
+<<<<<<< HEAD:parser/digester.go
+=======
+func (d *sqlDigester) isGenericLists(last4 []token) bool {
+	if len(last4) < 4 {
+		return false
+	}
+	if !(last4[0].tok == genericSymbol || last4[0].tok == genericSymbolList) {
+		return false
+	}
+	if last4[1].lit != ")" {
+		return false
+	}
+	if !d.isComma(last4[2]) {
+		return false
+	}
+	if last4[3].lit != "(" {
+		return false
+	}
+	return true
+}
+
+// In (Row(...), Row(...)) => In (Row(...))
+func (d *sqlDigester) isGenericRowListsWithIn(last9 []token) bool {
+	if len(last9) < 7 {
+		return false
+	}
+	if !d.isInKeyword(last9[0]) {
+		return false
+	}
+	if last9[1].lit != "(" {
+		return false
+	}
+	if !d.isRowKeyword(last9[2]) {
+		return false
+	}
+	if last9[3].lit != "(" {
+		return false
+	}
+	if !(last9[4].tok == genericSymbol || last9[4].tok == genericSymbolList) {
+		return false
+	}
+	if last9[5].lit != ")" {
+		return false
+	}
+	if !d.isComma(last9[6]) {
+		return false
+	}
+	if !d.isRowKeyword(last9[7]) {
+		return false
+	}
+	if last9[8].lit != "(" {
+		return false
+	}
+	return true
+}
+
+// IN (...) => IN (?) Issue: #43192
+func (d *sqlDigester) replaceSingleLiteralWithInList(currTok *token) {
+	last5 := d.tokens.back(5)
+	if len(last5) == 5 &&
+		d.isInKeyword(last5[0]) &&
+		d.isLeftParen(last5[1]) &&
+		last5[2].lit == "." &&
+		last5[3].lit == "." &&
+		last5[4].lit == "." &&
+		d.isRightParen(*currTok) {
+		d.tokens.popBack(3)
+		d.tokens.pushBack(token{genericSymbol, "?"})
+		return
+	}
+}
+
+// IN (?) => IN (...) Issue: #44298
+func (d *sqlDigester) reduceInListWithSingleLiteral(currTok *token) {
+	last3 := d.tokens.back(3)
+	if len(last3) == 3 &&
+		d.isInKeyword(last3[0]) &&
+		d.isLeftParen(last3[1]) &&
+		last3[2].tok == genericSymbol &&
+		d.isRightParen(*currTok) {
+		d.tokens.popBack(1)
+		d.tokens.pushBack(token{genericSymbolList, "..."})
+		return
+	}
+}
+
+// In (Row(...)) => In (...) #51222
+func (d *sqlDigester) reduceInRowListWithSingleLiteral(currTok *token) {
+	last5 := d.tokens.back(6)
+	if len(last5) == 6 &&
+		d.isInKeyword(last5[0]) &&
+		d.isLeftParen(last5[1]) &&
+		d.isRowKeyword(last5[2]) &&
+		d.isLeftParen(last5[3]) &&
+		(last5[4].tok == genericSymbolList || last5[4].tok == genericSymbol) &&
+		d.isRightParen(last5[5]) &&
+		d.isRightParen(*currTok) {
+		d.tokens.popBack(4)
+		d.tokens.pushBack(token{genericSymbolList, "..."})
+		return
+	}
+}
+
+>>>>>>> 707b0a4e381 (parser: support (Row(..),Row(..))=>(..) in the binding mode (#51319)):pkg/parser/digester.go
 func (d *sqlDigester) isPrefixByUnary(currTok int) (isUnary bool) {
 	if !d.isNumLit(currTok) {
 		return
@@ -414,6 +554,29 @@ func (*sqlDigester) isComma(tok token) (isComma bool) {
 	return
 }
 
+<<<<<<< HEAD:parser/digester.go
+=======
+func (*sqlDigester) isLeftParen(tok token) (isLeftParen bool) {
+	isLeftParen = tok.lit == "("
+	return
+}
+
+func (*sqlDigester) isRightParen(tok token) (isLeftParen bool) {
+	isLeftParen = tok.lit == ")"
+	return
+}
+
+func (*sqlDigester) isInKeyword(tok token) (isInKeyword bool) {
+	isInKeyword = tok.lit == "in"
+	return
+}
+
+func (*sqlDigester) isRowKeyword(tok token) (isRowKeyword bool) {
+	isRowKeyword = tok.lit == "row"
+	return
+}
+
+>>>>>>> 707b0a4e381 (parser: support (Row(..),Row(..))=>(..) in the binding mode (#51319)):pkg/parser/digester.go
 type token struct {
 	tok int
 	lit string

pkg/bindinfo/tests/BUILD.bazel

@@ -0,0 +1,29 @@
+load("@io_bazel_rules_go//go:def.bzl", "go_test")
+
+go_test(
+    name = "tests_test",
+    timeout = "short",
+    srcs = [
+        "bind_test.go",
+        "main_test.go",
+    ],
+    flaky = True,
+    race = "on",
+    shard_count = 16,
+    deps = [
+        "//pkg/bindinfo",
+        "//pkg/bindinfo/internal",
+        "//pkg/bindinfo/norm",
+        "//pkg/domain",
+        "//pkg/parser",
+        "//pkg/parser/model",
+        "//pkg/parser/terror",
+        "//pkg/testkit",
+        "//pkg/testkit/testsetup",
+        "//pkg/util",
+        "//pkg/util/parser",
+        "//pkg/util/stmtsummary",
+        "@com_github_stretchr_testify//require",
+        "@org_uber_go_goleak//:goleak",
+    ],
+)