Parse nodes before and after the root element

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

@@ -586,7 +586,7 @@ class XMLTestJVM {
   def issue508commentParsing: Unit = {
     // confirm that comments are processed correctly now
     roundtrip("<a><!-- comment --> suffix</a>")
-    roundtrip("<a>prefix <!-- comment --> suffix</a>")
+    roundtrip("<a>prefix <!-- comment -->   <!-- comment2 --> suffix</a>")
     roundtrip("<a>prefix <b><!-- comment --></b> suffix</a>")
     roundtrip("<a>prefix <b><!-- multi-\nline\n comment --></b> suffix</a>")
     roundtrip("""<a>prefix <b><!-- multi-
@@ -596,13 +596,7 @@ class XMLTestJVM {
     // confirm that processing instructions were always processed correctly
     roundtrip("<a><?target content ?> suffix</a>")
     roundtrip("<a>prefix <?target content ?> suffix</a>")
-    roundtrip("<a>prefix <b><?target content?></b> suffix</a>")
-
-    // TODO since XMLLoader retrieves FactoryAdapter.rootNode,
-    // capturing comments before and after the root element is not currently possible
-    // (by the way, the same applies to processing instructions).
-    //check("<!-- prologue --><a>text</a>")
-    //check("<a>text</a><!-- epilogue -->")
+    roundtrip("<a>prefix <b><?target content?> </b> suffix</a>")
   }
 
   @UnitTest
@@ -613,7 +607,26 @@ class XMLTestJVM {
     roundtrip("""<a>prefix <b><![CDATA[
                 | multi-
                 | line    cdata
-                |    section]]></b> suffix</a>""".stripMargin)
+                |    section]]>   </b> suffix</a>""".stripMargin)
+  }
+
+  def roundtripNodes(xml: String): Unit = assertEquals(xml, XML.loadStringNodes(xml).map(_.toString).mkString(""))
+
+  @UnitTest
+  def xmlLoaderLoadNodes: Unit = {
+    roundtripNodes("<!-- prolog --><a>text</a>")
+    roundtripNodes("<!-- prolog --><?target content ?><!-- comment2 --><a>text</a>")
+    roundtripNodes("""<!-- prolog
+                     |    --><?target content ?><!--
+                     |  comment2 --><a>text</a>""".stripMargin)
+
+    roundtripNodes("<a>text</a><!-- epilogue -->")
+    roundtripNodes("<a>text</a><!-- epilogue --><?target content ?><!-- comment2 -->")
+
+    // Note: at least with the JDK's Xerces, whitespace in the prolog and epilogue gets lost in parsing:
+    // the parser does not fire any white-space related events, so:
+    // does not work: roundtripNodes("<!-- c -->  <a/>")
+    // does not work: roundtripNodes("<a/> <!-- epilogue -->")
   }
 
   @UnitTest

shared/src/main/scala/scala/xml/factory/XMLLoader.scala

@@ -51,19 +51,29 @@ trait XMLLoader[T <: Node] {
    *  The methods available in scala.xml.XML use the XML parser in the JDK.
    */
   def loadXML(source: InputSource, parser: SAXParser): T = {
-    val newAdapter = adapter
+    val result: FactoryAdapter = parse(source, parser)
+    result.rootElem.asInstanceOf[T]
+  }
+
+  def loadXMLNodes(source: InputSource, parser: SAXParser): Seq[Node] = {
+    val result: FactoryAdapter = parse(source, parser)
+    result.prolog ++ (result.rootElem :: result.epilogue)
+  }
+
+  private def parse(source: InputSource, parser: SAXParser): FactoryAdapter = {
+    val result: FactoryAdapter = adapter
 
     try {
-      parser.setProperty("http://xml.org/sax/properties/lexical-handler", newAdapter)
+      parser.setProperty("http://xml.org/sax/properties/lexical-handler", result)
     } catch {
       case _: SAXNotRecognizedException =>
     }
 
-    newAdapter.scopeStack = TopScope :: newAdapter.scopeStack
-    parser.parse(source, newAdapter)
-    newAdapter.scopeStack = newAdapter.scopeStack.tail
+    result.scopeStack = TopScope :: result.scopeStack
+    parser.parse(source, result)
+    result.scopeStack = result.scopeStack.tail
 
-    newAdapter.rootElem.asInstanceOf[T]
+    result
   }
 
   /** Loads XML from the given file, file descriptor, or filename. */
@@ -80,4 +90,15 @@ trait XMLLoader[T <: Node] {
 
   /** Loads XML from the given String. */
   def loadString(string: String): T = loadXML(fromString(string), parser)
-}
+
+  /** Load XML nodes, including comments and processing instructions that precede and follow the root element. */
+  def loadFileNodes(file: File): Seq[Node] = loadXMLNodes(fromFile(file), parser)
+  def loadFileNodes(fd: FileDescriptor): Seq[Node] = loadXMLNodes(fromFile(fd), parser)
+  def loadFileNodes(name: String): Seq[Node] = loadXMLNodes(fromFile(name), parser)
+  def loadNodes(is: InputStream): Seq[Node] = loadXMLNodes(fromInputStream(is), parser)
+  def loadNodes(reader: Reader): Seq[Node] = loadXMLNodes(fromReader(reader), parser)
+  def loadNodes(sysID: String): Seq[Node] = loadXMLNodes(fromSysId(sysID), parser)
+  def loadNodes(source: InputSource): Seq[Node] = loadXMLNodes(source, parser)
+  def loadNodes(url: URL): Seq[Node] = loadXMLNodes(fromInputStream(url.openStream()), parser)
+  def loadStringNodes(string: String): Seq[Node] = loadXMLNodes(fromString(string), parser)
+}

shared/src/main/scala/scala/xml/parsing/FactoryAdapter.scala

@@ -40,9 +40,11 @@ trait ConsoleErrorHandler extends DefaultHandler2 {
  *  underlying SAX parser.
  */
 abstract class FactoryAdapter extends DefaultHandler2 with factory.XMLLoader[Node] {
+  var prolog: List[Node] = List.empty
   var rootElem: Node = _
+  var epilogue: List[Node] = List.empty
 
-  val buffer = new StringBuilder()
+  val buffer: StringBuilder = new StringBuilder()
   private var inCDATA: Boolean = false
 
   /** List of attributes
@@ -51,28 +53,28 @@ abstract class FactoryAdapter extends DefaultHandler2 with factory.XMLLoader[Nod
     * 
     * @since 2.0.0 
     */
-  var attribStack = List.empty[MetaData]
+  var attribStack: List[MetaData] = List.empty
   /** List of elements
     * 
     * Previously was a mutable [[scala.collection.mutable.Stack Stack]], but is now a mutable reference to an immutable [[scala.collection.immutable.List List]].
     * 
     * @since 2.0.0 
     */
-  var hStack = List.empty[Node] // [ element ] contains siblings
+  var hStack: List[Node] = List.empty // [ element ] contains siblings
   /** List of element names
     * 
     * Previously was a mutable [[scala.collection.mutable.Stack Stack]], but is now a mutable reference to an immutable [[scala.collection.immutable.List List]].
     * 
     * @since 2.0.0 
     */
-  var tagStack = List.empty[String]
+  var tagStack: List[String] = List.empty
   /** List of namespaces
     * 
     * Previously was a mutable [[scala.collection.mutable.Stack Stack]], but is now a mutable reference to an immutable [[scala.collection.immutable.List List]].
     * 
     * @since 2.0.0 
     */
-  var scopeStack = List.empty[NamespaceBinding]
+  var scopeStack: List[NamespaceBinding] = List.empty
 
   var curTag: String = _
   var capture: Boolean = false
@@ -123,7 +125,7 @@ abstract class FactoryAdapter extends DefaultHandler2 with factory.XMLLoader[Nod
   // ContentHandler methods
   //
 
-  val normalizeWhitespace = false
+  val normalizeWhitespace: Boolean = false
 
   /**
    * Capture characters, possibly normalizing whitespace.
@@ -177,13 +179,20 @@ abstract class FactoryAdapter extends DefaultHandler2 with factory.XMLLoader[Nod
     attributes: Attributes): Unit =
     {
       captureText()
+
+      // capture the prolog at the start of the root element
+      if (tagStack.isEmpty) {
+        prolog = hStack.reverse
+        hStack = List.empty
+      }
+
       tagStack = curTag :: tagStack
       curTag = qname
 
       val localName = splitName(qname)._2
       capture = nodeContainsText(localName)
 
-      hStack =  null :: hStack
+      hStack = null :: hStack
       var m: MetaData = Null
       var scpe: NamespaceBinding =
         if (scopeStack.isEmpty) TopScope
@@ -193,7 +202,7 @@ abstract class FactoryAdapter extends DefaultHandler2 with factory.XMLLoader[Nod
         val qname = attributes getQName i
         val value = attributes getValue i
         val (pre, key) = splitName(qname)
-        def nullIfEmpty(s: String) = if (s == "") null else s
+        def nullIfEmpty(s: String): String = if (s == "") null else s
 
         if (pre == "xmlns" || (pre == null && qname == "xmlns")) {
           val arg = if (pre == null) null else key
@@ -250,6 +259,12 @@ abstract class FactoryAdapter extends DefaultHandler2 with factory.XMLLoader[Nod
     capture = curTag != null && nodeContainsText(curTag) // root level
   }
 
+  override def endDocument(): Unit =  {
+    // capture the epilogue at the end of the document
+    epilogue = hStack.init.reverse
+    hStack = hStack.last :: Nil
+  }
+
   /**
    * Processing instruction.
    */

shared/src/main/scala/scala/xml/parsing/MarkupParser.scala

@@ -98,6 +98,8 @@ trait MarkupParser extends MarkupParserCommon with TokenTests {
   var extIndex = -1
 
   /** holds temporary values of pos */
+  // Note: this is clearly an override, but if marked as such it causes a "...cannot override a mutable variable"
+  // error with Scala 3; does it work with Scala 3 if not explicitly marked as an override remains to be seen...
   var tmppos: Int = _
 
   /** holds the next character */