encoding/xml: normalize \t, \r and \n to space in attributes

DemiMarie · DemiMarie · commit 043668eba735 · 2024-09-01T10:58:04.000-04:00
This is required by the spec. Fixes: #20614
diff --git a/src/encoding/xml/marshal_test.go b/src/encoding/xml/marshal_test.go
@@ -602,6 +602,12 @@ var marshalTests = []struct {
 		ExpectXML:     `<Data Attr=""><Bytes></Bytes><Custom></Custom></Data>`,
 		UnmarshalOnly: true,
 	},
+	// Check attribute value normalization (issue 20614).
+	{
+		Value:         &Data{Bytes: []byte{}, Custom: MyBytes{}, Attr: []byte(" a b c d")},
+		ExpectXML:     "<Data Attr='\na\r\nb\tc\rd'><Bytes></Bytes><Custom></Custom></Data>",
+		UnmarshalOnly: true,
+	},
 
 	// Check that []byte works, including named []byte types.
 	{
diff --git a/src/encoding/xml/xml.go b/src/encoding/xml/xml.go
@@ -1118,11 +1118,17 @@ Input:
 			return nil
 		}
 
-		// We must rewrite unescaped \r and \r\n into \n.
-		if b == '\r' {
-			d.buf.WriteByte('\n')
+		// We must rewrite unescaped \r and \r\n into \n outside quotes,
+		// and unescaped \r, \r\n, \n, and \t into space inside quotes.
+		if b >= ' ' {
+			d.buf.WriteByte(b)
 		} else if b1 == '\r' && b == '\n' {
 			// Skip \r\n--we already wrote \n.
+		} else if quote >= 0 && (b == '\r' || b == '\n' || b == '\t') {
+			// Normalize newline, CR, and tab into space
+			d.buf.WriteByte(' ')
+		} else if b == '\r' {
+			d.buf.WriteByte('\n')
 		} else {
 			d.buf.WriteByte(b)
 		}
diff --git a/src/encoding/xml/xml_test.go b/src/encoding/xml/xml_test.go
@@ -626,6 +626,30 @@ type item struct {
 	FieldA string
 }
 
+func TestIssue20614(t *testing.T) {
+	data := "<item b='\n\r\t'/>"
+	dec := NewDecoder(strings.NewReader(data))
+	var tok1, tok2, tok3 Token
+	var err error
+	if tok1, err = dec.RawToken(); err != nil {
+		t.Fatalf("RawToken() failed: %#v", err)
+	}
+	if tok2, err = dec.RawToken(); err != nil {
+		t.Fatalf("RawToken() failed: %#v", err)
+	}
+	if tok3, err = dec.RawToken(); err != io.EOF || tok3 != nil {
+		t.Fatalf("Missed EOF")
+	}
+	s := StartElement{Name{"", "item"}, []Attr{Attr{Name{"","b"}, "   "}}}
+	if !reflect.DeepEqual(tok1.(StartElement), s) {
+		t.Error("Wrong start element")
+	}
+	e := EndElement{Name{"","item"}}
+	if tok2.(EndElement) != e {
+		t.Error("Wrong end element")
+	}
+}
+
 func TestIssue569(t *testing.T) {
 	data := `<item><FieldA>abcd</FieldA></item>`
 	var i item
