complexalgorithm.java

基于词典和最大匹配算法的的中文分词组件

/**
 * 
 */
package org.solol.mmseg.internal;

import java.util.ArrayList;
import java.util.List;

import org.solol.mmseg.core.AlgorithmException;
import org.solol.mmseg.core.DictionaryFactory;
import org.solol.mmseg.core.IChunk;
import org.solol.mmseg.core.IWord;

/**
 * @author solo L
 * 
 */
public final class ComplexAlgorithm extends AbstractAlgorithm {

	public ComplexAlgorithm() throws AlgorithmException {
		super();
	}

	public ComplexAlgorithm(DictionaryFactory dictionaryFactory) {
		super(dictionaryFactory);
	}

	protected IChunk[] createChunks(char[] chars, int index) {

		List chunkList = new ArrayList(5);

		IWord[] words0 = findMatchWords(chars, index);

		for (int i = 0; i < words0.length; i++) {
			int index0 = index + words0[i].getLength();
			if (index0 < chars.length) {
				IWord[] words1 = findMatchWords(chars, index0);
				for (int j = 0; j < words1.length; j++) {
					int index1 = index0 + words1[j].getLength();
					if (index1 < chars.length) {
						IWord[] words2 = findMatchWords(chars, index1);
						for (int k = 0; k < words2.length; k++) {
							chunkList.add(new Chunk(new IWord[] { words0[i],
									words1[j], words2[k] }));
						}
					} else if (index1 == chars.length) {
						chunkList.add(new Chunk(new IWord[] { words0[i],
								words1[j] }));
					}
				}
			} else if (index0 == chars.length) {
				chunkList.add(new Chunk(new IWord[] { words0[i] }));
			}
		}

		IChunk[] chunks = new IChunk[chunkList.size()];
		chunkList.toArray(chunks);
		chunkList.clear();

		return chunks;
	}

}