/*
* Copyright (C) 2004, 2005, 2006 Joe Walnes.
* Copyright (C) 2006, 2007 XStream Committers.
* All rights reserved.
*
* The software in this package is published under the terms of the BSD
* style license a copy of which has been included with this distribution in
* the LICENSE.txt file.
*
* Created on 07. March 2004 by Joe Walnes
*/
package modmanager.utility.xml;
import com.thoughtworks.xstream.io.xml.AbstractDocumentReader;
import com.thoughtworks.xstream.io.xml.XmlFriendlyReplacer;
import org.w3c.dom.Attr;
import org.w3c.dom.Document;
import org.w3c.dom.Element;
import org.w3c.dom.Node;
import org.w3c.dom.NodeList;
import org.w3c.dom.Text;
import java.util.ArrayList;
import java.util.List;
/**
* This reader is used by ShirkitDriver to read the mod.xml files. The only difference is this: if (!text.getData().trim().isEmpty()). With it, I avoid a real bug/problem that XStream was retrieving blank spaces of code that the mod developer didn't thought of. For example:
* <find>
* CDATA[find string}</find>
* Without this reader, that line feed after <find> and before the CData would be passed as a complement of the mod's find action, wich the developer didn't meant to do.
* @author Shirkit
*/
public class ShirkitReader extends AbstractDocumentReader {
private Element currentElement;
private StringBuffer textBuffer;
private List childElements;
public ShirkitReader(Element rootElement) {
this(rootElement, new XmlFriendlyReplacer());
}
public ShirkitReader(Document document) {
this(document.getDocumentElement());
}
/**
* @since 1.2
*/
public ShirkitReader(Element rootElement, XmlFriendlyReplacer replacer) {
super(rootElement, replacer);
textBuffer = new StringBuffer();
}
/**
* @since 1.2
*/
public ShirkitReader(Document document, XmlFriendlyReplacer replacer) {
this(document.getDocumentElement(), replacer);
}
public String getNodeName() {
return unescapeXmlName(currentElement.getTagName());
}
public String getValue() {
NodeList childNodes = currentElement.getChildNodes();
textBuffer.setLength(0);
int length = childNodes.getLength();
for (int i = 0; i < length; i++) {
Node childNode = childNodes.item(i);
if (childNode instanceof Text) {
Text text = (Text) childNode;
// Avoid empty blocks of text. This is the main difference on this class!
if (!text.getData().trim().isEmpty()) {
textBuffer.append(text.getData());
}
}
}
return textBuffer.toString();
}
public String getAttribute(String name) {
Attr attribute = currentElement.getAttributeNode(name);
return attribute == null ? null : attribute.getValue();
}
public String getAttribute(int index) {
return ((Attr) currentElement.getAttributes().item(index)).getValue();
}
public int getAttributeCount() {
return currentElement.getAttributes().getLength();
}
public String getAttributeName(int index) {
return unescapeXmlName(((Attr) currentElement.getAttributes().item(index)).getName());
}
protected Object getParent() {
return currentElement.getParentNode();
}
protected Object getChild(int index) {
return childElements.get(index);
}
protected int getChildCount() {
return childElements.size();
}
protected void reassignCurrentElement(Object current) {
currentElement = (Element) current;
NodeList childNodes = currentElement.getChildNodes();
childElements = new ArrayList();
for (int i = 0; i < childNodes.getLength(); i++) {
Node node = childNodes.item(i);
if (node instanceof Element) {
childElements.add(node);
}
}
}
}