net.loomchild.segment.srx.io.Srx2Parser Maven / Gradle / Ivy

Go to download
Show more of this group Show more artifacts with this name
Show all versions of segment Show documentation
Library used to split text into segments.
There is a newer version: 2.0.4
package net.loomchild.segment.srx.io;

import static net.loomchild.segment.util.Util.getContext;
import static net.loomchild.segment.util.Util.getReader;
import static net.loomchild.segment.util.Util.getResourceStream;
import static net.loomchild.segment.util.Util.getSchema;

import java.io.Reader;
import java.util.HashMap;
import java.util.Map;

import net.loomchild.segment.srx.LanguageRule;
import net.loomchild.segment.srx.SrxDocument;
import net.loomchild.segment.srx.io.bind.*;
import net.loomchild.segment.srx.SrxParser;
import net.loomchild.segment.util.Bind;

import org.apache.commons.logging.Log;
import org.apache.commons.logging.LogFactory;


/**
 * Represents SRX 2.0 document parser. Responsible for creating and initializing
 * Document according to given SRX. Uses JAXB. Can validate.
 * 
 * @author loomchild
 */
public class Srx2Parser implements SrxParser {

	private static final Log log = LogFactory.getLog(Srx2Parser.class);

	private static final String CONTEXT = "net.loomchild.segment.srx.io.bind";
	
	private static final String SCHEMA = "net/loomchild/segment/res/xml/srx20.xsd";

	private static Bind bind = createBind();

	private static Bind createBind() {
        // Macintosh Java 1.5 work-around borrowed from okapi library
        // When you use -XstartOnFirstThread as a java -Xarg on Leopard, 
		// your ContextClassloader gets set to null.
		// On other Macs setting this value breaks everything.
		if (Thread.currentThread().getContextClassLoader() == null) {
			Thread.currentThread().setContextClassLoader(Srx2Parser.class.getClassLoader());
		}
		
		// Must pass the ClassLoader directly due to Java 1.5 bugs when using 
		// custom ClassLoader.
		Bind bind = new Bind(
				getContext(CONTEXT, Srx2Parser.class.getClassLoader()),
				getSchema(getReader(getResourceStream(SCHEMA))));
		return bind;
	}
	
	/**
	 * Parses SRX document from reader.
	 * 
	 * @param reader
	 * @return initialized document
	 */
	public SrxDocument parse(Reader reader) {
		Srx srx = (Srx) bind.unmarshal(reader);

		SrxDocument document = new SrxDocument();
		document.setCascade("yes".equals(srx.getHeader().getCascade()));

		Body body = srx.getBody();

		Map languageRuleMap = new HashMap();
		for (Languagerule lr : body.getLanguagerules().getLanguagerule()) {
			LanguageRule languageRule = new LanguageRule(lr
					.getLanguagerulename());
			for (Rule r : lr.getRule()) {
				boolean breakRule = !"no".equals(r.getBreak());

				String before;
				if (r.getBeforebreak() != null) {
					before = r.getBeforebreak().getContent();
				} else {
					before = "";
				}

				String after;
				if (r.getAfterbreak() != null) {
					after = r.getAfterbreak().getContent();
				} else {
					after = "";
				}

				net.loomchild.segment.srx.Rule rule = new net.loomchild.segment.srx.Rule(breakRule, before, after);
				languageRule.addRule(rule);
			}
			languageRuleMap.put(languageRule.getName(), languageRule);
		}

		for (Languagemap lm : body.getMaprules().getLanguagemap()) {
			LanguageRule languageRule = languageRuleMap.get(lm
					.getLanguagerulename());
			if (languageRule == null) {
				log.warn("Language map \"" + lm.getLanguagepattern()
						+ "\": language rule \"" + lm.getLanguagerulename()
						+ "\" not found.");
			} else {
				document.addLanguageMap(lm.getLanguagepattern(), languageRule);
			}
		}

		return document;
	}

}