package org.daisy.pipeline.nlp.impl.matchrules;
   
   import java.util.regex.Matcher;
   import java.util.regex.Pattern;
   
   import org.daisy.pipeline.nlp.TextCategorizer.Category;
   import org.daisy.pipeline.nlp.TextCategorizer.MatchMode;
   import org.daisy.pipeline.nlp.impl.MatchRule;
   
  /**
   * MatchRule based on the Java core's Regexp engine. See dk.brics.automaton for
   * a future faster implementation.
   */
  public class RegexMatchRule extends MatchRule {
  
  	// warning: the Java Regexp engine always break the search if X matches in
  	// "(X|Y)"
  	// even if Y is longer than X.
  
  	private Pattern mPattern;
  
  	public RegexMatchRule(Category category, int priority, boolean caseSensitive,
  	        MatchMode matchMode) {
  		super(category, priority, caseSensitive, matchMode);
  	}
  
  	public void init(String regexp) {
  		switch (mMatchMode) {
  		case FULL_MATCH:
  			regexp += "$";
  			break;
  		default:
  			break;
  		}
  
  		int flags = Pattern.UNICODE_CASE | Pattern.MULTILINE | Pattern.DOTALL;
  		// java7: flags |= Pattern.UNICODE_CHARACTER_CLASS;
  
  		if (!mCaseSensitive) {
  			flags |= Pattern.CASE_INSENSITIVE;
  		}
  
  		mPattern = Pattern.compile(regexp, flags);
  	}
  
  	@Override
  	protected String match(String input) {
  		Matcher m = mPattern.matcher(input);
  
  		if (m.lookingAt()) {
  			return m.group(0);
  		}
  		return null;
  	}
  
  	@Override
  	public boolean threadsafe() {
  		return true;
  	}
  }