Escape Next Line and Line Start in XML

aws · Feb 8, 2021 · 360664a · 360664a
1 parent 61f63d8
commit 360664a
Show file tree

Hide file tree

Showing 3 changed files with 105 additions and 2 deletions.
diff --git a/encoding/xml/encoder_test.go b/encoding/xml/encoder_test.go
@@ -3,6 +3,7 @@ package xml_test
 import (
 	"bytes"
 	"log"
+	"sort"
 	"testing"
 
 	"github.com/aws/smithy-go/encoding/xml"
@@ -478,6 +479,53 @@ func TestEncodeListNamedShape(t *testing.T) {
 	verify(t, encoder, ex)
 }
 
+func TestEncodeEscaping(t *testing.T) {
+	b := bytes.NewBuffer(nil)
+	encoder := xml.NewEncoder(b)
+
+	func() {
+		r := encoder.RootElement(root)
+		defer r.Close()
+
+		cases := map[string]rune{
+			"quote":          '"',
+			"apos":           '\'',
+			"amp":            '&',
+			"lt":             '<',
+			"gt":             '>',
+			"tab":            '\t',
+			"newLine":        '\n',
+			"carriageReturn": '\r',
+			"nextLine":       '\u0085',
+			"lineSeparator":  '\u2028',
+		}
+
+		var sortedKeys []string
+		for name := range cases {
+			sortedKeys = append(sortedKeys, name)
+		}
+
+		sort.Strings(sortedKeys)
+
+		for _, name := range sortedKeys {
+			rr := cases[name]
+
+			st := xml.StartElement{Name: xml.Name{Local: name}}
+			st.Attr = append(st.Attr, xml.Attr{
+				Name: xml.Name{
+					Local: "key",
+				},
+				Value: name + string(rr) + name,
+			})
+			value := r.MemberElement(st)
+			value.String(name + string(rr) + name)
+		}
+	}()
+
+	ex := []byte(`<root><amp key="amp&amp;amp">amp&amp;amp</amp><apos key="apos&#39;apos">apos&#39;apos</apos><carriageReturn key="carriageReturn&#xD;carriageReturn">carriageReturn&#xD;carriageReturn</carriageReturn><gt key="gt&gt;gt">gt&gt;gt</gt><lineSeparator key="lineSeparator&#x2028;lineSeparator">lineSeparator&#x2028;lineSeparator</lineSeparator><lt key="lt&lt;lt">lt&lt;lt</lt><newLine key="newLine&#xA;newLine">newLine&#xA;newLine</newLine><nextLine key="nextLine&#x85;nextLine">nextLine&#x85;nextLine</nextLine><quote key="quote&#34;quote">quote&#34;quote</quote><tab key="tab&#x9;tab">tab&#x9;tab</tab></root>`)
+	verify(t, encoder, ex)
+}
+
 // ExampleEncoder is the example function on how to use an encoder
 func ExampleEncoder() {
 	b := bytes.NewBuffer(nil)

diff --git a/encoding/xml/escape.go b/encoding/xml/escape.go
@@ -21,6 +21,10 @@ var (
 	escNL   = []byte("&#xA;")
 	escCR   = []byte("&#xD;")
 	escFFFD = []byte("\uFFFD") // Unicode replacement character
+
+	// Additional Escapes
+	escNextLine = []byte("&#x85;")
+	escLS       = []byte("&#x2028;")
 )
 
 // Decide whether the given rune is in the XML Character Range, per
@@ -61,6 +65,12 @@ func escapeString(e writer, s string) {
 			esc = escNL
 		case '\r':
 			esc = escCR
+		case '\u0085':
+			// Not escaped by stdlib
+			esc = escNextLine
+		case '\u2028':
+			// Not escaped by stdlib
+			esc = escLS
 		default:
 			if !isInCharacterRange(r) || (r == 0xFFFD && width == 1) {
 				esc = escFFFD
@@ -101,11 +111,17 @@ func escapeText(e writer, s []byte) {
 		case '\t':
 			esc = escTab
 		case '\n':
-			// TODO: This always escapes newline, which is different than stdlib's optional
-			// escape of new line
+			// This always escapes newline, which is different than stdlib's optional
+			// escape of new line.
 			esc = escNL
 		case '\r':
 			esc = escCR
+		case '\u0085':
+			// Not escaped by stdlib
+			esc = escNextLine
+		case '\u2028':
+			// Not escaped by stdlib
+			esc = escLS
 		default:
 			if !isInCharacterRange(r) || (r == 0xFFFD && width == 1) {
 				esc = escFFFD

diff --git a/local-mod-replace.sh b/local-mod-replace.sh
@@ -0,0 +1,39 @@
+#1/usr/bin/env bash
+
+PROJECT_DIR=""
+SMITHY_SOURCE_DIR=$(cd `dirname $0` && pwd)
+
+usage() {
+  echo "Usage: $0 [-s SMITHY_SOURCE_DIR] [-d PROJECT_DIR]" 1>&2
+  exit 1
+}
+
+while getopts "hs:d:" options; do
+  case "${options}" in
+  s)
+    SMITHY_SOURCE_DIR=${OPTARG}
+    if [ "$SMITHY_SOURCE_DIR" == "" ]; then
+      echo "path to smithy-go source directory is required" || exit
+      usage
+    fi
+    ;;
+  d)
+    PROJECT_DIR=${OPTARG}
+    ;;
+  h)
+    usage
+    ;;
+  *)
+    usage
+    ;;
+  esac
+done
+
+if [ "$PROJECT_DIR" != "" ]; then
+  cd $PROJECT_DIR || exit
+fi
+
+go mod graph | awk '{print $1}' | cut -d '@' -f 1 | sort | uniq | grep "github.com/aws/smithy-go" | while read x; do
+  repPath=${x/github.com\/aws\/smithy-go/${SMITHY_SOURCE_DIR}}
+  echo -replace $x=$repPath
+done | xargs go mod edit