Fix and correct some aspects of language handling.

In particular, take better care to ensure that when an element is
removed which might have an xml:lang attribute, its children keep
their original language.

6 files changed, 26 insertions, 6 deletions

diff --git a/org/madore/damlengine/LangHelper.java b/org/madore/damlengine/LangHelper.java
index a94f5be..aea1240 100644
--- a/org/madore/damlengine/LangHelper.java
+++ b/org/madore/damlengine/LangHelper.java
@@ -32,6 +32,14 @@ public class LangHelper {
 	node.removeAttributeNS(DamlEngine.XML_NS, LANG);
     }
 
+    public static void setWeakLangNorec(Element node, String lang) {
+	if ( lang == null )
+	    throw new NullPointerException("lang is null in setWeakLangNorec");
+	String currentLang = getLangNorec(node);
+	if ( currentLang == null )
+	    setLangNorec(node,lang);
+    }
+
     public static String getLangRec(Node node) {
 	// Never returns null (returns the empty string for no language).
 	// node may be null (in which case, return "").
diff --git a/org/madore/damlengine/TodoBodyElement.java b/org/madore/damlengine/TodoBodyElement.java
index 1db657b..2e48ee8 100644
--- a/org/madore/damlengine/TodoBodyElement.java
+++ b/org/madore/damlengine/TodoBodyElement.java
@@ -27,6 +27,7 @@ public final class TodoBodyElement extends TodoDefaultElement {
 	    throw new IllegalArgumentException("body node can only be child of daml node");
 
 	Element parent = (Element)(this.node.getParentNode());
+	String explicitLang = LangHelper.getLangNorec(this.node);
 	ArrayList<Node> childList = getChildList(this.node);
 	ArrayList<TodoElement> toProcess = new ArrayList<TodoElement>(childList.size());
 	for ( Node child : childList ) {
@@ -37,6 +38,8 @@ public final class TodoBodyElement extends TodoDefaultElement {
 	    }
 	    parent.insertBefore(child, node);
 	    if ( child.getNodeType() == Node.ELEMENT_NODE ) {
+		if ( explicitLang != null )
+		    LangHelper.setWeakLangNorec((Element)child, explicitLang);
 		TodoElement it
 		    = TodoElement.getTodoElement((Element)child, this.ctx, this);
 		toProcess.add(it);
diff --git a/org/madore/damlengine/TodoCommentsScript.java b/org/madore/damlengine/TodoCommentsScript.java
index 224d547..eb250e8 100644
--- a/org/madore/damlengine/TodoCommentsScript.java
+++ b/org/madore/damlengine/TodoCommentsScript.java
@@ -1,5 +1,6 @@
 package org.madore.damlengine;
 
+import java.util.regex.Pattern;
 import org.w3c.dom.*;
 
 public final class TodoCommentsScript extends TodoElement {
@@ -13,6 +14,11 @@ public final class TodoCommentsScript extends TodoElement {
     @Override
     public void handle() {
 	if ( ctx.gc.commentsJSParam == null ) {
+	    Node ws = node.getNextSibling();
+	    if ( ( ws.getNodeType() == Node.TEXT_NODE
+		   || ws.getNodeType() == Node.CDATA_SECTION_NODE )
+		 && Pattern.matches("^\\s*$",((CharacterData)ws).getData()) )
+		node.getParentNode().removeChild(ws);
 	    node.getParentNode().removeChild(node);
 	    return;
 	}
diff --git a/org/madore/damlengine/TodoKillAElement.java b/org/madore/damlengine/TodoKillAElement.java
index 4478fae..f230ee0 100644
--- a/org/madore/damlengine/TodoKillAElement.java
+++ b/org/madore/damlengine/TodoKillAElement.java
@@ -22,11 +22,14 @@ public final class TodoKillAElement extends TodoDefaultElement {
 
     @Override
     public void handleNodeOnly() {
+	String explicitLang = LangHelper.getLangNorec(this.node);
 	ArrayList<Node> childList = getChildList(node);
 	ArrayList<TodoElement> toProcess = new ArrayList<TodoElement>(childList.size());
 	for ( Node child : childList ) {
 	    node.getParentNode().insertBefore(child, node);
 	    if ( child.getNodeType() == Node.ELEMENT_NODE ) {
+		if ( explicitLang != null )
+		    LangHelper.setWeakLangNorec((Element)child, explicitLang);
 		TodoElement it
 		    = TodoElement.getTodoElement((Element)child, this.ctx, this);
 		toProcess.add(it);
diff --git a/org/madore/damlengine/TodoKillAcronymElement.java b/org/madore/damlengine/TodoKillAcronymElement.java
index 65664e9..f7e6190 100644
--- a/org/madore/damlengine/TodoKillAcronymElement.java
+++ b/org/madore/damlengine/TodoKillAcronymElement.java
@@ -28,9 +28,9 @@ public final class TodoKillAcronymElement extends TodoDefaultElement {
 				   "acronym "+node.getAttributeNS(null, "class"));
 	else
 	    newNode.setAttributeNS(null, "class", "acronym");
-	String lang = LangHelper.getLangNorec(node);
-	if ( lang != null )
-	    LangHelper.setLangNorec(newNode, lang);
+	String explicitLang = LangHelper.getLangNorec(node);
+	if ( explicitLang != null )
+	    LangHelper.setLangNorec(newNode, explicitLang);
 	ArrayList<Attr> attrList = getAttrList(this.node);
 	for ( Attr attr : attrList ) {
 	    if ( attr.getNamespaceURI() == null ) {
diff --git a/org/madore/damlengine/TodoWeblogElement.java b/org/madore/damlengine/TodoWeblogElement.java
index 3d457a8..1cbc53c 100644
--- a/org/madore/damlengine/TodoWeblogElement.java
+++ b/org/madore/damlengine/TodoWeblogElement.java
@@ -25,7 +25,7 @@ public final class TodoWeblogElement extends TodoDefaultElement {
     public void handleNodeOnly() {
 	System.err.println("warning: weblog element should not be processed this way!");
 	Node parent = node.getParentNode();
-	String lang = LangHelper.getLangNorec(node);
+	String explicitLang = LangHelper.getLangNorec(node);
 	ArrayList<Node> childList = getChildList(this.node);
 	ArrayList<TodoElement> toProcess = new ArrayList<TodoElement>(childList.size()+8);
 	for ( Node child : childList ) {
@@ -36,8 +36,8 @@ public final class TodoWeblogElement extends TodoDefaultElement {
 	    }
 	    parent.insertBefore(child, node);
 	    if ( child.getNodeType() == Node.ELEMENT_NODE ) {
-		if ( lang != null )
-		    LangHelper.setLangRec((Element)child, lang);  // Dubious
+		if ( explicitLang != null )
+		    LangHelper.setWeakLangNorec((Element)child, explicitLang);
 		TodoElement it
 		    = TodoElement.getTodoElement((Element)child, this.ctx, this);
 		toProcess.add(it);