expression: fix incorrect collation when cast non-string type arg to …

…string type (#19186) (#22599)
pingcap · Jan 28, 2021 · 33fdb15 · 33fdb15
1 parent f74dd5d
commit 33fdb15
Show file tree

Hide file tree

Showing 5 changed files with 17 additions and 3 deletions.
diff --git a/expression/builtin_cast.go b/expression/builtin_cast.go
@@ -28,7 +28,6 @@ import (
 
 	"github.com/pingcap/errors"
 	"github.com/pingcap/parser/ast"
-	"github.com/pingcap/parser/charset"
 	"github.com/pingcap/parser/model"
 	"github.com/pingcap/parser/mysql"
 	"github.com/pingcap/parser/terror"
@@ -1900,7 +1899,7 @@ func WrapWithCastAsString(ctx sessionctx.Context, expr Expression) Expression {
 		argLen = -1
 	}
 	tp := types.NewFieldType(mysql.TypeVarString)
-	tp.Charset, tp.Collate = charset.GetDefaultCharsetAndCollate()
+	tp.Charset, tp.Collate = expr.CharsetAndCollation(ctx)
 	tp.Flen, tp.Decimal = argLen, types.UnspecifiedLength
 	return BuildCastFunction(ctx, expr, tp)
 }

diff --git a/expression/collation.go b/expression/collation.go
@@ -151,12 +151,18 @@ func deriveCoercibilityForScarlarFunc(sf *ScalarFunction) Coercibility {
 	if sf.RetType.EvalType() != types.ETString {
 		return CoercibilityNumeric
 	}
-	coer := CoercibilityCoercible
+	coer := CoercibilityIgnorable
 	for _, arg := range sf.GetArgs() {
 		if arg.Coercibility() < coer {
 			coer = arg.Coercibility()
 		}
 	}
+
+	// it is weird if a ScalarFunction is CoercibilityNumeric but return string type
+	if coer == CoercibilityNumeric {
+		return CoercibilityCoercible
+	}
+
 	return coer
 }
 

diff --git a/expression/integration_test.go b/expression/integration_test.go
@@ -7674,7 +7674,14 @@ func (s *testIntegrationSerialSuite) TestIssue19116(c *C) {
 	defer collate.SetNewCollationEnabledForTest(false)
 
 	tk := testkit.NewTestKit(c, s.store)
+	tk.MustExec("set names utf8mb4 collate utf8mb4_general_ci;")
+	tk.MustQuery("select collation(concat(1 collate `binary`));").Check(testkit.Rows("binary"))
+	tk.MustQuery("select coercibility(concat(1 collate `binary`));").Check(testkit.Rows("0"))
 	tk.MustQuery("select collation(concat(NULL,NULL));").Check(testkit.Rows("binary"))
+	tk.MustQuery("select coercibility(concat(NULL,NULL));").Check(testkit.Rows("6"))
+	tk.MustQuery("select collation(concat(1,1));").Check(testkit.Rows("utf8mb4_general_ci"))
 	tk.MustQuery("select coercibility(concat(1,1));").Check(testkit.Rows("4"))
+	tk.MustQuery("select collation(1);").Check(testkit.Rows("binary"))
 	tk.MustQuery("select coercibility(1);").Check(testkit.Rows("5"))
+	tk.MustQuery("select coercibility(1=1);").Check(testkit.Rows("5"))
 }
diff --git a/expression/simple_rewriter.go b/expression/simple_rewriter.go
@@ -297,6 +297,7 @@ func (sr *simpleRewriter) Leave(originInNode ast.Node) (retNode ast.Node, ok boo
 			arg.GetType().Collate = v.Collate
 		}
 		sr.stack[len(sr.stack)-1].SetCoercibility(CoercibilityExplicit)
+		sr.stack[len(sr.stack)-1].SetCharsetAndCollation(arg.GetType().Charset, arg.GetType().Collate)
 	default:
 		sr.err = errors.Errorf("UnknownType: %T", v)
 		return retNode, false

diff --git a/planner/core/expression_rewriter.go b/planner/core/expression_rewriter.go
@@ -1120,6 +1120,7 @@ func (er *expressionRewriter) Leave(originInNode ast.Node) (retNode ast.Node, ok
 			arg.GetType().Collate = v.Collate
 		}
 		er.ctxStack[len(er.ctxStack)-1].SetCoercibility(expression.CoercibilityExplicit)
+		er.ctxStack[len(er.ctxStack)-1].SetCharsetAndCollation(arg.GetType().Charset, arg.GetType().Collate)
 	default:
 		er.err = errors.Errorf("UnknownType: %T", v)
 		return retNode, false