even better dtd parsing! handled URLs now

4 files changed, 128 insertions, 50 deletions

diff --git a/config/list/library.lst b/config/list/library.lst
index f20bba95ac..d0c694e253 100644
--- a/config/list/library.lst
+++ b/config/list/library.lst
@@ -256,6 +256,7 @@ xml/factory/LoggedNodeFactory.scala
 
 xml/parsing/ConstructingHandler.scala
 xml/parsing/ConstructingParser.scala
+xml/parsing/ExternalSources.scala
 xml/parsing/FatalError.scala
 xml/parsing/MarkupHandler.scala
 xml/parsing/MarkupParser.scala
diff --git a/sources/scala/xml/parsing/ConstructingParser.scala b/sources/scala/xml/parsing/ConstructingParser.scala
index 51299ec916..791d3b2f2c 100644
--- a/sources/scala/xml/parsing/ConstructingParser.scala
+++ b/sources/scala/xml/parsing/ConstructingParser.scala
@@ -14,24 +14,7 @@ import scala.io.Source;
 object ConstructingParser {
 
   def fromFile(inp: java.io.File, preserveWS: Boolean) = {
-    /* DEBUG
-    val src = Source.fromFile(inp);
-    while(src.hasNext) {
-      Console.print(src.next);
-      if(!src.hasNext) {
-        Console.print("last character!");
-        Console.print(src.ch);
-      }
-    }
-    */
     val p = new ConstructingParser(Source.fromFile(inp), preserveWS);
-    /*
-    {
-      override def externalSource(systemLiteral: String): Source = {
-        Source.fromFile(new java.io.File(inp.getParent(), systemLiteral));
-      }
-    }
-    */
     p.nextch;
     p
   }
@@ -47,19 +30,12 @@ object ConstructingParser {
  */
 class ConstructingParser(inp: Source, presWS:Boolean)
 extends  ConstructingHandler
+with     ExternalSources
 with     MarkupParser  {
 
   override val isValidating = true;
   val preserveWS = presWS;
   val input = inp;
   val handle = this;
-
-  override def externalSource(systemLiteral: String): Source = {
-    var fileStr = inp.descr;
-    if(inp.descr.startsWith("file:")) {
-      fileStr = inp.descr.substring(5, inp.descr.length());
-    }
-    fileStr = fileStr.substring(0,fileStr.lastIndexOf(java.io.File.separator)+1);
-    Source.fromFile(fileStr + systemLiteral);
-  }
 }
+
diff --git a/sources/scala/xml/parsing/ExternalSources.scala b/sources/scala/xml/parsing/ExternalSources.scala
new file mode 100644
index 0000000000..7d42d6ec6a
--- /dev/null
+++ b/sources/scala/xml/parsing/ExternalSources.scala
@@ -0,0 +1,76 @@
+/*                     __                                               *\
+**     ________ ___   / /  ___     Scala API                            **
+**    / __/ __// _ | / /  / _ |    (c) 2003-2005, LAMP/EPFL             **
+**  __\ \/ /__/ __ |/ /__/ __ |                                         **
+** /____/\___/_/ |_/____/_/ | |                                         **
+**                          |/                                          **
+** $Id$
+\*                                                                      */
+
+package scala.xml.parsing ;
+
+import scala.io.Source;
+import java.net._;
+import java.io._;
+
+trait ExternalSources : (MarkupParser with MarkupHandler)  {
+
+
+  private def externalSourceFromURL(url:URL): Source = {
+    val in =
+      new BufferedReader(
+        new InputStreamReader(
+	  url.openStream()));
+
+    //@todo: replace this hack with proper Source implementation
+
+    val str = new StringBuffer();
+    var inputLine:String = _;
+
+    //while (inputLine = in.readLine()) != null) {
+    while ({inputLine = in.readLine(); inputLine} != null) {
+      // Console.println(inputLine); // DEBUG
+      str.append(inputLine);
+      str.append('\n'); // readable output
+    }
+    in.close();
+
+    class MyClass extends Source {
+
+      def newIter = new Iterator[Char] {
+        var i = -1;
+        private val len = str.length()-1;
+        def hasNext = i < len;
+        def next = {
+          i = i + 1;
+          str.charAt(i);
+        }
+      }
+
+      val iter = newIter;
+
+      def reset: Source = new MyClass;
+
+      override var descr = url.toExternalForm();
+    }
+
+    return new MyClass;
+  }
+
+  def externalSource(systemId: String): Source = {
+    //Console.println("in external source("+systemId+")");
+    if(systemId.startsWith("http:")) {
+      return externalSourceFromURL(new URL(systemId));
+    }
+
+    var fileStr = input.descr;
+
+    if(input.descr.startsWith("file:")) {
+      fileStr = input.descr.substring(5, input.descr.length());
+    } else
+    fileStr = fileStr.substring(0,
+                                fileStr.lastIndexOf(java.io.File.separator)+1);
+    Source.fromFile(fileStr + systemId);
+  }
+
+}
diff --git a/sources/scala/xml/parsing/MarkupParser.scala b/sources/scala/xml/parsing/MarkupParser.scala
index 8b52bd8db4..95c7b3d440 100644
--- a/sources/scala/xml/parsing/MarkupParser.scala
+++ b/sources/scala/xml/parsing/MarkupParser.scala
@@ -45,8 +45,8 @@ abstract class MarkupParser: (MarkupParser with MarkupHandler) extends AnyRef wi
   var pos: Int = _;
 
 
-  /* true if reading external subset */
-  var inExtSubSet = false;
+  /* used when reading external subset */
+  var extIndex = -1;
 
   /** holds temporary values of pos */
   var tmppos: Int = _;
@@ -226,9 +226,9 @@ abstract class MarkupParser: (MarkupParser with MarkupHandler) extends AnyRef wi
       ch = curInput.next;
       pos = curInput.pos;
     } else {
-      //Console.println("nextch, curInput.hasNext == false ")      ;
-      //Console.println("nextch, Nil != inpStack == "+(Nil!=inpStack));
-      if (Nil != inpStack) {
+      val ilen = inpStack.length;
+      //Console.println("  ilen = "+ilen+ " extIndex = "+extIndex);
+      if ((ilen != extIndex) && (ilen > 0)) {
         /** for external source, inpStack == Nil ! need notify of eof! */
         pop();
       } else {
@@ -580,9 +580,11 @@ abstract class MarkupParser: (MarkupParser with MarkupHandler) extends AnyRef wi
      */
 
     if(null != extID) {
-      val extSubsetSrc = externalSource( extID.systemId );
 
-      inExtSubSet = true;
+      pushExternal(extID.systemId);
+      //val extSubsetSrc = externalSource( extID.systemId );
+
+      extIndex = inpStack.length;
       /*
        .indexOf(':') != -1) { // assume URI
          Source.fromFile(new java.net.URI(extID.systemLiteral));
@@ -591,20 +593,23 @@ abstract class MarkupParser: (MarkupParser with MarkupHandler) extends AnyRef wi
        }
       */
       //Console.println("I'll print it now");
-      val old = curInput;
-      tmppos = curInput.pos;
-      val oldch = ch;
-      curInput = extSubsetSrc;
-      pos = 0;
-      nextch;
+      //val old = curInput;
+      //tmppos = curInput.pos;
+      //val oldch = ch;
+      //curInput = extSubsetSrc;
+      //pos = 0;
+      //nextch;
+
       extSubset();
 
-      inExtSubSet = false;
+      pop();
 
-      curInput = old;
-      pos = curInput.pos;
-      ch = curInput.ch;
-      eof = false;
+      extIndex = -1;
+
+      //curInput = old;
+      //pos = curInput.pos;
+      //ch = curInput.ch;
+      //eof = false;
       //while(extSubsetSrc.hasNext)
       //Console.print(extSubsetSrc.next);
 
@@ -806,7 +811,7 @@ abstract class MarkupParser: (MarkupParser with MarkupHandler) extends AnyRef wi
       } else
         markupDecl1();
     }
-    while(curInput.hasNext) {
+    while(!eof) {
       markupDecl();
     }
   }
@@ -843,7 +848,7 @@ abstract class MarkupParser: (MarkupParser with MarkupHandler) extends AnyRef wi
           nextch;
           notationDecl();
 
-        case '[' if inExtSubSet =>
+        case '[' if inpStack.length >= extIndex =>
           nextch;
           xSpaceOpt;
           ch match {
@@ -949,9 +954,11 @@ abstract class MarkupParser: (MarkupParser with MarkupHandler) extends AnyRef wi
     val n = xName;
     xSpace;
     while ('>' != ch) {
+      //Console.println("["+ch+"]");
       putChar(ch);
       nextch;
     }
+    //Console.println("END["+ch+"]");
     nextch;
     val cmstr = cbuf.toString();
     cbuf.setLength(0);
@@ -1102,10 +1109,10 @@ abstract class MarkupParser: (MarkupParser with MarkupHandler) extends AnyRef wi
   def push(entityName:String) = {
     //Console.println("BEFORE PUSHING  "+ch);
     //Console.println("BEFORE PUSHING  "+pos);
-    //Console.println("PUSHING "+entityName);
-    if(!eof) {
+    //Console.print("[PUSHING "+entityName+"]");
+    if(!eof)
       inpStack = curInput :: inpStack;
-    }
+
     curInput = replacementText(entityName);
     nextch;
   }
@@ -1116,13 +1123,31 @@ abstract class MarkupParser: (MarkupParser with MarkupHandler) extends AnyRef wi
     nextch;
   }
   */
+
+  def pushExternal(systemId:String) = {
+    //Console.print("BEFORE PUSH, curInput = $"+curInput.descr);
+    //Console.println(" stack = "+inpStack.map { x => "$"+x.descr });
+
+    //Console.print("[PUSHING EXTERNAL "+systemId+"]");
+    if(!eof)
+      inpStack = curInput :: inpStack;
+
+    curInput = externalSource(systemId);
+
+    //Console.print("AFTER PUSH, curInput = $"+curInput.descr);
+    //Console.println(" stack = "+inpStack.map { x => "$"+x.descr });
+
+    nextch;
+  }
   def pop() = {
-    //Console.println("POPPING");
+
     curInput = inpStack.head;
     inpStack = inpStack.tail;
     ch = curInput.ch;
     pos = curInput.pos;
     eof = false; // must be false, because of places where entity refs occur
+    //Console.println("\n AFTER POP, curInput = $"+curInput.descr);
+    //Console.println(inpStack.map { x => x.descr });
   }
 
 }