Adding a char shingling FX for sgdtk struct

ShingledCharFeatureExtractor.java

package org.sgdtk.struct;

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

/**
 * Extracts a an array of character shingle features from a sequence
 *
 * @author dpressel
 */
public class ShingledCharFeatureExtractor implements FeatureExtractorInterface
{
    private final String ns;
    private final int featureIndex;
    private final int ngram;

    public ShingledCharFeatureExtractor(String ns, int ngram)
    {
        this(ns, ngram, 0);
    }
    /**
     * Constructor, you would probably not need to do this under normal circumstances
     *
     * @param ns namespace
     * @param featureIndex The offset of the feature in a CONLL file
     */
    public ShingledCharFeatureExtractor(String ns, int ngram, int featureIndex)
    {
        this.ns = ns;
        this.featureIndex = featureIndex;
        this.ngram = ngram;
    }

    /**
     * Given a sequence of {@link State} objects which are multi-dimensional,
     * made up of components themselves, and a position in that sequence, extract a single feature
     *
     * @param states A sequence
     * @param current The current absolute position in the sequence
     * @return
     */
    @Override
    public String[] run(List<State> states, int current)
    {
        String term = states.get(current).atIndex(this.featureIndex);

        List<String> buffer = new ArrayList<>();
        for (int i = ngram, sz = term.length() - ngram + 1; i < sz; ++i)
        {
            String aPart = term.substring(i - ngram, i);
            buffer.add(ns + aPart);

        }
        return buffer.toArray(new String[buffer.size()]);
    }

    /**
     * The order of this feature is the max of all of its parts
     * @return The order (1 or 2)
     */
    @Override
    public int getOrder()
    {
        return 1;
    }

    /**
     * Get the number of parts in the extractor
     * @return
     */
    @Override
    public int size()
    {
        return 1;
    }


}