Merge pull request #122 from Rogach/master

Change default encoding in XML.save to UTF-8

Author: ashawley

jvm/src/test/scala/scala/xml/XMLTest.scala

@@ -300,6 +300,29 @@ class XMLTestJVM {
     </wsdl:definitions>""", wsdlTemplate4("service4", () => "target4") toString)
   }
 
+  // Issue found with ISO-8859-1 in #121 that was fixed with UTF-8 default
+  @UnitTest
+  def writeReadNoDeclarationDefaultEncoding: Unit = {
+    val chars = ((32 to 126) ++ (160 to 255)).map(_.toChar)
+    val xml = <x>{ chars.mkString }</x>
+
+    // com.sun.org.apache.xerces.internal.impl.io.MalformedByteSequenceException:
+    // Invalid byte 1 of 1-byte UTF-8 sequence.
+    // scala.xml.XML.save("foo.xml", xml)
+    // scala.xml.XML.loadFile("foo.xml").toString)
+
+    val outputStream = new java.io.ByteArrayOutputStream
+    val streamWriter = new java.io.OutputStreamWriter(outputStream, XML.encoding)
+
+    XML.write(streamWriter, xml, XML.encoding, false, null)
+    streamWriter.flush
+
+    val inputStream = new java.io.ByteArrayInputStream(outputStream.toByteArray)
+    val streamReader = new java.io.InputStreamReader(inputStream)
+
+    assertEquals(xml.toString, XML.load(streamReader).toString)
+  }
+
   @UnitTest
   def t0663 = {
     val src = scala.io.Source.fromString("<?xml version='1.0' encoding='UTF-8'?><feed/>")

shared/src/main/scala/scala/xml/XML.scala

@@ -63,7 +63,7 @@ object XML extends XMLLoader[Elem] {
   val preserve = "preserve"
   val space = "space"
   val lang = "lang"
-  val encoding = "ISO-8859-1"
+  val encoding = "UTF-8"
 
   /** Returns an XMLLoader whose load* methods will use the supplied SAXParser. */
   def withSAXParser(p: SAXParser): XMLLoader[Elem] =
@@ -73,6 +73,10 @@ object XML extends XMLLoader[Elem] {
    * Saves a node to a file with given filename using given encoding
    *  optionally with xmldecl and doctype declaration.
    *
+   *  Note: default encoding was ISO-8859-1 (latin1) in pre-1.0.7 scala-xml versions.
+   *  If your code depends on characters in non-ASCII latin1 range, specify
+   *  ISO-8859-1 encoding explicitly.
+   *
    *  @param filename the filename
    *  @param node     the xml node we want to write
    *  @param enc      encoding to use
@@ -82,7 +86,7 @@ object XML extends XMLLoader[Elem] {
   final def save(
     filename: String,
     node: Node,
-    enc: String = encoding,
+    enc: String = "UTF-8",
     xmlDecl: Boolean = false,
     doctype: dtd.DocType = null): Unit =
     {