package predictor;

import java.util.Random;

import org.apache.log4j.*;

import weka.core.*;
import weka.classifiers.*;

public class ClassifierAdapter implements PredictorInterface {
	protected Logger logger = Logger.getLogger(ClassifierAdapter.class);

	protected String name;
	protected Classifier classifier;
	protected Evaluation eval;

	public ClassifierAdapter(Classifier classifier, String name) {
		this.classifier = classifier;
		this.name = name;
	}

	@Override
	public String getName() {
		return this.name;
	}

	@Override
	public void crossValidate(Instances instances, int numFold, Random rand) throws Exception {
		eval = new Evaluation(instances);
		eval.crossValidateModel(this.classifier, instances, numFold, rand);
	}

	public String getEvaluationSummaryString() {
		return this.eval.toSummaryString();
	}

	public String getEvaluationMatrixString() throws Exception {
		return this.eval.toMatrixString();
	}
	
	public FastVector getEvaluationPredictions() {
		return eval.predictions();
	}

	@Override
	public void train(Instances instances) throws Exception {
		this.classifier.buildClassifier(instances);
	}
	
	@Override
	public int predict(Instance instance) throws Exception {
		this.classifier.classifyInstance(instance);
		return 0;
	}

	@Override
	public int predict(Instances instances) throws Exception {
		for (int i=0;i<instances.numInstances();i++) {
			this.predict(instances.instance(i));
		}
		return 0;
	}
	
	@Override
	public String getEvaluationResults() throws Exception {
		String results = "\n-- ";
		results += this.getName();
		results += " --\n";
		results += this.eval.toSummaryString();
		results += this.eval.toMatrixString();
		results += this.eval.toClassDetailsString();
		results += "\n";
		return results;
	}
}