// BSD License (http://www.galagosearch.org/license)

 package org.galagosearch.core.scoring;

 import java.io.IOException;

 import java.util.HashMap;

 import org.galagosearch.core.types.DocumentLengthWordCount;

 import org.galagosearch.core.types.DocumentWordProbability;

 import org.galagosearch.tupleflow.Parameters;

 import org.galagosearch.tupleflow.StandardStep;

 import org.galagosearch.tupleflow.Utility;

 /**

  *

  * @author trevor

  */

 public class LinearSmoother extends StandardStep<DocumentLengthWordCount, DocumentWordProbability>

         implements DistributionSmoother {

     double lambda;

     HashMap<String, Double> backgrounds;

     public LinearSmoother(Parameters.Value value, HashMap<String, Double> backgrounds) {

         double lm = 0.4;

         if (value.containsKey("lambda")) {

             lm = Double.parseDouble(value.get("lambda"));

         }

         this.lambda = lm;

         this.backgrounds = backgrounds;

     }

     public LinearSmoother(double lambda, HashMap<String, Double> backgrounds) {

         this.lambda = lambda;

         this.backgrounds = backgrounds;

     }

     public void process(DocumentLengthWordCount object) throws IOException {

         double background = backgrounds.get(object.word);

         double foreground = 0;

         if (object.length > 0) {

             foreground = (double) object.count / (double) object.length;

         }

         double probability = lambda * foreground + (1 - lambda) * background;

         processor.process(new DocumentWordProbability(object.document,

                                                       Utility.makeBytes(object.word), probability));

     }

     public double smooth(double background, int count, int length) {

         return (1 - lambda) * (double) count / (double) length + lambda * background;

     }

     public double smooth(String word, int count, int length) {

         return smooth(backgrounds.get(word), count, length);

     }

     public Class<DocumentLengthWordCount> getInputClass() {

         return DocumentLengthWordCount.class;

     }

     public Class<DocumentWordProbability> getOutputClass() {

         return DocumentWordProbability.class;

     }

 }