001 package org.maltparser.core.feature.spec.reader;
002
003 import java.io.IOException;
004 import java.net.URL;
005
006 import javax.xml.parsers.DocumentBuilder;
007 import javax.xml.parsers.DocumentBuilderFactory;
008 import javax.xml.parsers.ParserConfigurationException;
009
010 import org.maltparser.core.exception.MaltChainedException;
011 import org.maltparser.core.feature.FeatureException;
012 import org.maltparser.core.feature.spec.SpecificationModels;
013 import org.w3c.dom.Element;
014 import org.w3c.dom.NodeList;
015 import org.xml.sax.SAXException;
016 /**
017 *
018 *
019 * @author Johan Hall
020 */
021 public class XmlReader implements FeatureSpecReader{
022
023 public XmlReader() { }
024
025 public void load(URL specModelURL, SpecificationModels featureSpecModels) throws MaltChainedException {
026 try {
027 DocumentBuilderFactory dbf = DocumentBuilderFactory.newInstance();
028 DocumentBuilder db = dbf.newDocumentBuilder();
029 Element root = null;
030
031 root = db.parse(specModelURL.openStream()).getDocumentElement();
032
033 if (root == null) {
034 throw new FeatureException("The feature specification file '"+specModelURL.getFile()+"' cannot be found. ");
035 }
036
037 readFeatureModels(root, featureSpecModels);
038 } catch (IOException e) {
039 throw new FeatureException("The feature specification file '"+specModelURL.getFile()+"' cannot be found. ", e);
040 } catch (ParserConfigurationException e) {
041 throw new FeatureException("Problem parsing the file "+specModelURL.getFile()+". ", e);
042 } catch (SAXException e) {
043 throw new FeatureException("Problem parsing the file "+specModelURL.getFile()+". ", e);
044 }
045 }
046
047 private void readFeatureModels(Element featuremodels, SpecificationModels featureSpecModels) throws MaltChainedException {
048 NodeList featureModelList = featuremodels.getElementsByTagName("featuremodel");
049 for (int i = 0; i < featureModelList.getLength(); i++) {
050 readFeatureModel((Element)featureModelList.item(i), featureSpecModels);
051 }
052 }
053
054 private void readFeatureModel(Element featuremodel, SpecificationModels featureSpecModels) throws MaltChainedException {
055 int specModelIndex = featureSpecModels.getNextIndex();
056 NodeList submodelList = featuremodel.getElementsByTagName("submodel");
057 if (submodelList.getLength() == 0) {
058 NodeList featureList = featuremodel.getElementsByTagName("feature");
059 for (int i = 0; i < featureList.getLength(); i++) {
060 String featureText = ((Element)featureList.item(i)).getTextContent().trim();
061 if (featureText.length() > 1) {
062 featureSpecModels.add(specModelIndex, featureText);
063 }
064 }
065 } else {
066 for (int i = 0; i < submodelList.getLength(); i++) {
067 String name = ((Element)submodelList.item(i)).getAttribute("name");
068 NodeList featureList = ((Element)submodelList.item(i)).getElementsByTagName("feature");
069 for (int j = 0; j < featureList.getLength(); j++) {
070 String featureText = ((Element)featureList.item(j)).getTextContent().trim();
071 if (featureText.length() > 1) {
072 featureSpecModels.add(specModelIndex, name, featureText);
073 }
074 }
075 }
076 }
077 }
078 }