import org.codehaus.stax2.XMLOutputFactory2; //导入依赖的package包/类
public void extract(String inputFile, String outputFile) throws IOException, XMLStreamException {
    System.out.println("Parsing pages and extracting links...");
    
    long startTime = System.currentTimeMillis();
    XMLOutputFactory outputFactory = XMLOutputFactory2.newInstance();
    
    XMLStreamWriter writer = outputFactory.createXMLStreamWriter(new FileOutputStream(outputFile), "UTF-8");
    writer.writeStartDocument();
    writer.writeStartElement("d");
    
    LinkExtractor linkExtractor = new LinkExtractor(writer);
    linkExtractor.parse(inputFile);

    writer.writeEndElement();
    writer.writeEndDocument();
    writer.close();
    
    long elapsedSeconds = (System.currentTimeMillis() - startTime) / 1000;
    System.out.printf("\n%d pages parsed in %d seconds.\n", linkExtractor.getPageCount(), elapsedSeconds);
}
 

import org.codehaus.stax2.XMLOutputFactory2; //导入依赖的package包/类
public Pair<ArticlesIdsRelations, Map<String, String>> extract(String inputFile) throws IOException, XMLStreamException
{
	System.out.println("Parsing pages and extracting Articles...");

	long startTime = System.currentTimeMillis();
	XMLOutputFactory outputFactory = XMLOutputFactory2.newInstance();

	

	ArticleExtractor articleExtractor = new ArticleExtractor(parameters);
	articleExtractor.parse(inputFile);
	ArticlesIdsRelations articles = articleExtractor.getArticles();
	Map<String, String> articlesText = articleExtractor.getArticlesText();
	/*for (Map.Entry<String, String> entry : articles.getTitlesMap().entrySet())
	{
	StringEscapeUtils.escapeCsv(entry.getKey());
	System.out.println(	StringEscapeUtils.escapeCsv(entry.getKey())+ " --> " + entry.getValue());
	}*/


	long elapsedSeconds = (System.currentTimeMillis() - startTime) / 1000;
	System.out.printf("\n%d pages parsed in %d seconds.\n", articleExtractor.getPageCount(), elapsedSeconds);
	return new Pair< > (articles,articlesText);
}
 

import org.codehaus.stax2.XMLOutputFactory2; //导入依赖的package包/类
private static XMLOutputFactory getOrCreateOutputFactory() throws FactoryConfigurationError {
	if (ourOutputFactory == null) {
		XMLOutputFactory outputFactory = XMLOutputFactory.newInstance();

		if (!ourHaveLoggedStaxImplementation) {
			logStaxImplementation(outputFactory.getClass());
		}

		/*
		 * Note that these properties are Woodstox specific and they cause a crash in environments where SJSXP is
		 * being used (e.g. glassfish) so we don't set them there.
		 */
		try {
			Class.forName("com.ctc.wstx.stax.WstxOutputFactory");
			if (outputFactory instanceof WstxOutputFactory) {
				outputFactory.setProperty(XMLOutputFactory2.P_TEXT_ESCAPER, new MyEscaper());
			}
		} catch (ClassNotFoundException e) {
			ourLog.debug("WstxOutputFactory (Woodstox) not found on classpath");
		}
		ourOutputFactory = outputFactory;
	}
	return ourOutputFactory;
}
 

import org.codehaus.stax2.XMLOutputFactory2; //导入依赖的package包/类
@Override
public XMLOutputFactory2 createOutputFactory() {
    return new WstxOutputFactory();
}