D-Net

package eu.dnetlib.pace.clustering;

import java.util.Collection;

import java.util.List;

import java.util.Map;

import com.google.common.collect.Sets;

import eu.dnetlib.pace.common.AbstractPaceFunctions;

import eu.dnetlib.pace.model.Field;

/**

 * The Class AbstractClusteringFunction.

 */

public abstract class AbstractClusteringFunction extends AbstractPaceFunctions implements ClusteringFunction {

	/** The params. */

	protected Map<String, Integer> params;

	/**

	 * Instantiates a new abstract clustering function.

	 * 

	 * @param params

	 *            the params

	 */

	public AbstractClusteringFunction(final Map<String, Integer> params) {

		this.params = params;

	}

	/**

	 * Do apply.

	 * 

	 * @param s

	 *            the s

	 * @return the collection

	 */

	protected abstract Collection<String> doApply(String s);

	/*

	 * (non-Javadoc)

	 * 

	 * @see eu.dnetlib.pace.clustering.ClusteringFunction#apply(java.util.List)

	 */

	@Override

	public Collection<String> apply(final List<Field> fields) {

		Collection<String> c = Sets.newLinkedHashSet();

		for (Field f : fields) {

			c.addAll(filterBlacklisted(doApply(filterStopWords(normalize(f.stringValue()), stopwords)), ngramBlacklist));

		}

		return c;

	}

	/*

	 * (non-Javadoc)

	 * 

	 * @see eu.dnetlib.pace.clustering.ClusteringFunction#getParams()

	 */

	@Override

	public Map<String, Integer> getParams() {

		return params;

	}

	/**

	 * Param.

	 * 

	 * @param name

	 *            the name

	 * @return the integer

	 */

	protected Integer param(final String name) {

		return params.get(name);

	}

}