StringWordIndexer.java

package edu.berkeley.nlp.lm;

import edu.berkeley.nlp.lm.collections.Indexer;

/**
 * Implementation of a WordIndexer in which words are represented as strings.
 * 
 * @author adampauls
 * 
 */
public class StringWordIndexer implements WordIndexer<String>
{

	/**
	 * 
	 */
	private static final long serialVersionUID = 1L;

	private final Indexer<String> sparseIndexer;

	private String startSymbol;

	private String endSymbol;

	private String unkSymbol;

	private int unkIndex = -1;

	public StringWordIndexer() {
		sparseIndexer = new Indexer<String>();
	}

	@Override
	public int getOrAddIndex(final String word) {
		return sparseIndexer.getIndex(word);
	}

	@Override
	public String getWord(final int index) {
		return sparseIndexer.getObject(index);
	}

	@Override
	public int numWords() {
		return sparseIndexer.size();
	}

	@Override
	public String getStartSymbol() {
		return startSymbol;
	}

	@Override
	public String getEndSymbol() {
		return endSymbol;
	}

	@Override
	public String getUnkSymbol() {
		return unkSymbol;
	}

	@Override
	public int getOrAddIndexFromString(final String word) {
		return getOrAddIndex(word);
	}

	@Override
	public void setStartSymbol(final String sym) {
		startSymbol = sym;
		sparseIndexer.add(sym);

	}

	@Override
	public void setEndSymbol(final String sym) {
		endSymbol = sym;
		sparseIndexer.add(sym);
	}

	@Override
	public void setUnkSymbol(final String sym) {
		unkSymbol = sym;
		unkIndex = sparseIndexer.getIndex(sym);
	}

	@Override
	public void trimAndLock() {
		sparseIndexer.trim();
		sparseIndexer.lock();
	}
	
	@Override
	public int getIndexPossiblyUnk(final String word) {
		final int id = sparseIndexer.indexOf(word);
		return id < 0 ? unkIndex : id;

	}
}