encoding/xml: rewrite invalid code points to U+FFFD in Marshal, Escape

Fixes #4235.

R=rsc, dave, r, dr.volker.dobler
CC=golang-dev
https://golang.org/cl/7438051

src/pkg/encoding/xml/xml.go

@@ -1729,6 +1729,7 @@ var (
 	esc_tab  = []byte("	")
 	esc_nl   = []byte("
")
 	esc_cr   = []byte("
")
+	esc_fffd = []byte("\uFFFD") // Unicode replacement character
 )
 
 // EscapeText writes to w the properly escaped XML equivalent
@@ -1736,8 +1737,10 @@ var (
 func EscapeText(w io.Writer, s []byte) error {
 	var esc []byte
 	last := 0
-	for i, c := range s {
+	for i := 0; i < len(s); {
-		switch c {
+		r, width := utf8.DecodeRune(s[i:])
+		i += width
+		switch r {
 		case '"':
 			esc = esc_quot
 		case '\'':
@@ -1755,15 +1758,19 @@ func EscapeText(w io.Writer, s []byte) error {
 		case '\r':
 			esc = esc_cr
 		default:
+			if !isInCharacterRange(r) {
+				esc = esc_fffd
+				break
+			}
 			continue
 		}
-		if _, err := w.Write(s[last:i]); err != nil {
+		if _, err := w.Write(s[last : i-width]); err != nil {
 			return err
 		}
 		if _, err := w.Write(esc); err != nil {
 			return err
 		}
-		last = i + 1
+		last = i
 	}
 	if _, err := w.Write(s[last:]); err != nil {
 		return err

src/pkg/encoding/xml/xml_test.go

@@ -5,6 +5,7 @@
 package xml
 
 import (
+	"bytes"
 	"fmt"
 	"io"
 	"reflect"
@@ -695,6 +696,21 @@ func TestEscapeTextIOErrors(t *testing.T) {
 	err := EscapeText(errWriter{}, []byte{'A'})
 
 	if err == nil || err.Error() != expectErr {
-		t.Errorf("EscapeTest = [error] %v, want %v", err, expectErr)
+		t.Errorf("have %v, want %v", err, expectErr)
+	}
+}
+
+func TestEscapeTextInvalidChar(t *testing.T) {
+	input := []byte("A \x00 terminated string.")
+	expected := "A \uFFFD terminated string."
+
+	buff := new(bytes.Buffer)
+	if err := EscapeText(buff, input); err != nil {
+		t.Fatalf("have %v, want nil", err)
+	}
+	text := buff.String()
+
+	if text != expected {
+		t.Errorf("have %v, want %v", text, expected)
 	}
 }