org.evosuite.lm.LanguageModelSearch Maven / Gradle / Ivy
The newest version!
/**
* Copyright (C) 2010-2018 Gordon Fraser, Andrea Arcuri and EvoSuite
* contributors
*
* This file is part of EvoSuite.
*
* EvoSuite is free software: you can redistribute it and/or modify it
* under the terms of the GNU Lesser General Public License as published
* by the Free Software Foundation, either version 3.0 of the License, or
* (at your option) any later version.
*
* EvoSuite is distributed in the hope that it will be useful, but
* WITHOUT ANY WARRANTY; without even the implied warranty of
* MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the GNU
* Lesser Public License for more details.
*
* You should have received a copy of the GNU Lesser General Public
* License along with EvoSuite. If not, see {
protected static final int GENERATIONS = 1000000;
private static Logger logger = LoggerFactory.getLogger(LanguageModelSearch.class);
protected final LangModel languageModel;
protected final String startPoint;
protected final ValueMinimizer.Minimization objective;
protected final ConstantValue constantValue;
protected static final int MAX_EVALUATIONS;
static{
MAX_EVALUATIONS = Properties.LM_ITERATIONS;
}
public int getEvaluations() {
return evaluations;
}
private int evaluations = 0;
public LanguageModelSearch(ValueMinimizer.Minimization objective, ConstantValue constantValue) {
try {
this.languageModel = new LangModel(Properties.LM_SRC);
} catch (Exception e) {
//FIXME: remove this garbage
throw new RuntimeException("Couldn't create language model");
}
this.startPoint = (String)constantValue.getValue();
this.objective = objective;
// this.statement = statement;
this.constantValue = constantValue;
}
public static String mutateRandom(final String input){
if (input == null || input.length() == 0){
return input;
}
int position = (int)(Math.abs(Randomness.nextDouble() * input.length()));
String output = input.substring(0,position);
int tailSize = input.length() - position;
assert tailSize >= 0;
int replacementSize = 1;
if(tailSize > 1)
replacementSize = Randomness.nextInt(1,tailSize);
String replacementString = "";
for(int i =0; i < replacementSize; i++)
replacementString += (char) Randomness.nextInt(' ', '~' + 1);
output += replacementString;
output += input.substring(position+ replacementSize);
return output;
}
public static String mutateEvoSuite(final String input){
char[] array = input.toCharArray();
for (int index = 0; index < input.length(); index++) {
if (Randomness.nextDouble() < 1.0 / (0.5 * input.length() + 1)) {
char oldValue = array[index];
while (array[index] == oldValue) {
array[index] = (char) Randomness.nextInt(' ', '~' + 1);
}
}
}
assert String.valueOf(array).length() == input.length();
return String.valueOf(array);
}
public String mutate(final String input){
if(Properties.LM_MUTATION_TYPE == MutationType.LANGMODEL) {
return mutateLangModel(input);
}else if(Properties.LM_MUTATION_TYPE == MutationType.RANDOM){
return mutateRandom(input);
}else if(Properties.LM_MUTATION_TYPE == MutationType.EVOSUITE){
return mutateEvoSuite(input);
}else{
throw new RuntimeException("Mutation type " + Properties.LM_MUTATION_TYPE + " not supported");
}
}
/**
* Build a random string that is at most maxStringLength characters long.
*/
public String generateRandomStringFromModel(int maxStringLength, String previousChar){
//Start the string with a random choice from the 10 most likely characters to start a string:
String newString = "";
do{
Set choices = new HashSet();
choices.addAll(Arrays.asList(0, 1, 2, 3, 4, 5, 6, 7, 8, 9));
boolean foundValidChar = false;
String nextChar;
do{
int currentChoice = Randomness.choice(choices);
choices.remove(currentChoice);
if(previousChar == null){
nextChar = languageModel.predict_char(currentChoice);
}else{
nextChar = languageModel.predict_char(previousChar, currentChoice);
}
foundValidChar = nextChar != null && !languageModel.isMagicChar(nextChar);
}while(!choices.isEmpty() && !foundValidChar);
//We keep trying until we get an actual char, or we run out of slots. After that, we give up and
//reuse the input.
if(!foundValidChar){
logger.debug("Couldn't find any bigram or unigram for " + previousChar);
nextChar = "a"; //TODO: this is a kludge
}
if(!languageModel.isEndOfSentence(nextChar))
newString += nextChar;
previousChar = nextChar;
}while(newString.length() < maxStringLength && !languageModel.isEndOfSentence(previousChar));
return newString;
}
public String generateRandomStringFromModelWithExactLength(int targetStringLength, String previousChar){
String newString = generateRandomStringFromModel(targetStringLength,previousChar);
while(newString.length() < targetStringLength){
newString += generateRandomStringFromModel(targetStringLength - newString.length(), newString.charAt(newString.length()-1) + "");
}
assert newString.length() == targetStringLength;
return newString;
}
public String generateRandomStringFromModelWithExactLength(int targetStringLength){
return generateRandomStringFromModelWithExactLength(targetStringLength, null);
}
public String generateRandomStringFromModel(int maxStringLength){
return generateRandomStringFromModel(maxStringLength, null);
}
/**
* Produce a new string similar to input.
* Selects a random substring of the string and for each character in that substring:
* gets the character preceeding it
* gets a character the language model predicts to follow that prefix
* replaces the character with the predicted character
*/
public String mutateLangModel(final String input){
if(input.length() == 0){
//Mutation preserves length, so we can't do anything for empty string.
return input;
}
//e.g. string is "abcde"
// point is 0..4
//Since nextDouble never returns 1.0, taking the floor gives us values in the range 0..strlen-1:
int startPoint = (int)(Randomness.nextDouble() * (input.length()));
//we preserve string lengths; how many chars do we have to mutate?
int remainingLength = (int)Math.round(
Randomness.nextDouble() * (input.length() - startPoint)
);
if(remainingLength == 0){
//Mutation size is zero, so there's nothing to do.
return input;
}
//substring is start (inclusive) to end (exclusive).
//e.g. if startPoint is 3
// output = "abc" at this point
String output = input.substring(0,startPoint);
String replacementChunk = generateRandomStringFromModelWithExactLength(remainingLength, ""+ input.charAt(startPoint));
output += replacementChunk;
output += input.substring(startPoint + remainingLength, input.length());
return output;
}
/**
* @param individual
* @return the language model score of the string, plus 1 if it does not affect the coverage goal.
*/
protected double evaluate(Chromosome individual) throws EvaluationBudgetExpendedException {
if(isBudgetExpended()){
throw new EvaluationBudgetExpendedException();
}
evaluations++;
String oldValue = (String)constantValue.getValue();
constantValue.setValue(individual.getValue());
boolean isNotWorse = objective.isNotWorse();
constantValue.setValue(oldValue);
return (isNotWorse ? 1 : 0) + languageModel.score(individual.getValue());
}
protected boolean isBudgetExpended() {
return evaluations >= MAX_EVALUATIONS;
}
protected Chromosome mutate(Chromosome individual){
assert individual != null;
return new Chromosome(mutate(individual.getValue()));
}
public abstract String optimise();
@Override
public int compare(Chromosome o1, Chromosome o2) {
if (!o1.isEvaluated()) {
o1.setFitness(evaluate(o1));
}
if (!o2.isEvaluated()) {
o2.setFitness(evaluate(o2));
}
return o1.compareTo(o2);
}
protected void resetEvaluationCounter() {
evaluations = 0;
}
}
class Chromosome implements Comparable, Cloneable {
private String value;
private double fitness = -1;
private boolean isEvaluated = false;
public Chromosome(String value) {
this.value = value;
}
public String getValue() {
return value;
}
public void setValue(String value) {
if (this.value.equals(value)) {
return;
}
this.value = value;
this.isEvaluated = false;
}
public double getFitness() {
return fitness;
}
public void setFitness(double fitness) {
this.fitness = fitness;
this.isEvaluated = true;
}
public boolean isEvaluated() {
return isEvaluated;
}
public Chromosome clone() {
Chromosome other = new Chromosome(this.getValue());
other.setFitness(this.getFitness());
other.isEvaluated = this.isEvaluated();
return other;
}
@Override
public boolean equals(Object o) {
if (this == o) return true;
if (o == null || getClass() != o.getClass()) return false;
Chromosome that = (Chromosome) o;
//Don't care about fitness or evaluation status:
return value.equals(that.value);
}
@Override
public int hashCode() {
int result;
long temp;
result = value.hashCode();
return result;
/*temp = Double.doubleToLongBits(fitness);
result = 31 * result + (int) (temp ^ (temp >>> 32));
result = 31 * result + (isEvaluated ? 1 : 0);
return result;*/
}
@Override
public int compareTo(Chromosome o) {
if (o == null) {
return 1;
}
return Double.compare(this.getFitness(), o.getFitness());
}
} © 2015 - 2024 Weber Informatics LLC | Privacy Policy