org.continuity.api.entities.deserialization.BehaviorModelSerializer Maven / Gradle / Ivy

Go to download
Show more of this group Show more artifacts with this name
Show all versions of continuity.api Show documentation
Library of continuITy defining exchange entities
The newest version!
package org.continuity.api.entities.deserialization;

import java.util.ArrayList;
import java.util.Collections;
import java.util.List;
import java.util.regex.Matcher;
import java.util.regex.Pattern;
import java.util.stream.Collectors;

import org.continuity.api.entities.artifact.BehaviorModel.Behavior;
import org.continuity.api.entities.artifact.BehaviorModel.MarkovState;
import org.continuity.api.entities.artifact.BehaviorModel.Transition;

/**
 * Provides method for (de-) serialization of multidimensional arrays (csv) and {@link Behavior}
 * objects.
 * 
 * @author Tobias Angerstein
 *
 */
public class BehaviorModelSerializer {

	/**
	 * Default zero edge.
	 */
	private static final String DEFAULT_ZERO_EDGE = "0.0; n(0 0)";

	/**
	 * Deserializes the behaviorModel csv representation into a {@link Behavior}
	 * 
	 * @param csvRepresentation
	 *            the csv representation of a {@link Behavior}
	 * @return {@link Behavior}
	 */
	public static Behavior deserializeBehaviorModel(String[][] csvRepresentation) {
		if (null == csvRepresentation) {
			return null;
		}
		List markovStates = new ArrayList();
		for (int i = 1; i < csvRepresentation.length; i++) {
			MarkovState markovState = new MarkovState();
			markovState.setId(csvRepresentation[i][0]);

			List transitions = new ArrayList();

			for (int j = 1; j < csvRepresentation[i].length; j++) {
				if (!csvRepresentation[i][j].equals(DEFAULT_ZERO_EDGE)) {
					Transition transition = new Transition();

					Pattern pattern = Pattern.compile("[0-9.]+");
					Matcher matcher = pattern.matcher(csvRepresentation[i][j]);
					matcher.find();
					transition.setProbability(Double.parseDouble(matcher.group(0)));

					matcher.find();
					transition.setMean(Double.parseDouble(matcher.group(0)));

					matcher.find();
					transition.setDeviation(Double.parseDouble(matcher.group(0)));

					transition.setTargetState(csvRepresentation[0][j]);

					transitions.add(transition);
				}
			}
			markovState.setTransitions(transitions);
			markovStates.add(markovState);
		}

		// Add ExitState
		MarkovState exitState = new MarkovState();
		exitState.setId("$");
		exitState.setTransitions(Collections.EMPTY_LIST);
		markovStates.add(exitState);

		Behavior behavior = new Behavior();
		behavior.setInitialState(markovStates.get(0).getId());
		behavior.setMarkovStates(markovStates);
		return behavior;
	}

	/**
	 * Provides serialization of {@link Behavior} into a multidimensional array representation.
	 * 
	 * @param rootBehaviorModel
	 *            the behavior model, which has to be serialized.
	 * @return {@link String[][]}
	 */
	public static String[][] serializeBehaviorModel(Behavior rootBehaviorModel) {
		String[][] csvRepresentation = new String[rootBehaviorModel.getMarkovStates().size()][rootBehaviorModel.getMarkovStates().size() + 1];
		List markovStateNames = rootBehaviorModel.getMarkovStates().stream().map(ms -> ms.getId()).collect(Collectors.toList());

		// Remove ExitState
		MarkovState exitState = new MarkovState();
		for(MarkovState state : rootBehaviorModel.getMarkovStates()) {
			if(state.getId().equals("$")) {
				exitState = state;
				break;
			}
		}
		rootBehaviorModel.getMarkovStates().remove(exitState);

		for (int i = 0; i < rootBehaviorModel.getMarkovStates().size(); i++) {

			// Write MarkovState name
			csvRepresentation[i + 1][0] = rootBehaviorModel.getMarkovStates().get(i).getId();

			// Write transitions
			for (Transition transition : rootBehaviorModel.getMarkovStates().get(i).getTransitions()) {
				csvRepresentation[i + 1][markovStateNames.indexOf(transition.getTargetState()) + 1] = getTransitionString(transition);
			}
		}
		// Write target state names (y-axis)
		for (int j = 0; j < markovStateNames.size(); j++) {
			if(markovStateNames.get(j).equals("INITIAL*")) {
				csvRepresentation[0][j + 1] = "INITIAL";
				continue;
			}
			csvRepresentation[0][j + 1] = markovStateNames.get(j);
		}

		fillEmptyFields(csvRepresentation);

		return csvRepresentation;
	}

	/**
	 * Fills empty fields with predefined zero edge, except csvRepresentation[0][0]
	 * 
	 * @param csvRepresentation
	 *            {@link String[][]}
	 */
	private static void fillEmptyFields(String[][] csvRepresentation) {
		for (int i = 0; i < csvRepresentation.length; i++) {
			for (int j = 0; j < csvRepresentation[i].length; j++) {
				if (i == 0 && j == 0) {
					csvRepresentation[i][j] = "";
				} else if (csvRepresentation[i][j] == null) {
					csvRepresentation[i][j] = DEFAULT_ZERO_EDGE;
				}
			}
		}
	}

	/**
	 * Provides transition properties String
	 * 
	 * @param transition
	 * @return
	 */
	private static String getTransitionString(Transition transition) {
		return String.format("%f; n(%f %f)", transition.getProbability(), transition.getMean(), transition.getDeviation());
	}

}