ultra-mastermind-implementation/lib/ultra_mastermind_obj.py

# Librairie du projet en version orientée objet

import random
import Levenshtein

def min_i(array: list[int]) -> int:
    min_val = array[0]
    min_i = 0
    for i in range(len(array)):
        if array[i] < min_val:
            min_val = array[i]
            min_i = i 
    return min_i

def max_i(array: list[int]) -> int:
    max_val = array[0]
    max_i = 0
    for i in range(len(array)):
        if array[i] > max_val:
            max_val = array[i]
            max_i = i 
    return max_i

class Population:
    """
    Classe qui représente notre population d'individuts
    """
    def __init__(self, pm, ng, n, ts, tm, alpha, fm):
        self.individuals = [Individual() for _ in range(n)]
        for individual in self.individuals: 
            individual.randomize(len(pm))
        self.pm = pm
        self.ng = ng
        self.l = len(pm)
        self.n = n
        self.ts = ts
        self.tm = tm
        self.alpha = alpha
        self.fitness_method = fm

    def select(self) -> None:
        """
        Methode qui sélectionne les meilleurs individus
        """
        fitness_list = [] 
        for individual in self.individuals:
            match self.fitness_method:
                case 1:
                    fitness_list.append(individual.fitness1(self.pm))
                case 2:
                    fitness_list.append(individual.fitness2(self.pm, self.alpha))
                case 3:
                    fitness_list.append(individual.fitness3(self.pm))
                case _:
                    fitness_list.append(individual.fitness1(self.pm))         
        for i in range(int((1 - self.ts) * self.n)):
            least = min_i(fitness_list)
            fitness_list.pop(least)
            self.individuals.pop(least)

    def reproduct(self) -> None:
        """
        Methode qui reproduit les individus entre eux jusqu'à obtenir une population de taille N
        """
        new = []
        while len(self.individuals) + len(new) != self.n:
            cut = random.randint(int(self.l / 3), int(2 * self.l / 3))
            indivi_1 = self.individuals[random.randint(0, len(self.individuals) - 1)]
            indivi_2 = self.individuals[random.randint(0, len(self.individuals) - 1)]
            while indivi_1 == indivi_2:
                indivi_2 = self.individuals[random.randint(0, len(self.individuals) - 1)]
            new_chromozome = indivi_1.getChromozome()[:cut] + indivi_2.getChromozome()[cut:]
            child = Individual()
            child.setChromozome(new_chromozome)
            new.append(child)
        self.individuals += new

    def mutate(self) -> None:
        """
        Methode qui mute une partie de la population selon le taut de mutation
        """
        mutated = []
        for i in range(int(self.tm * self.n)):
            to_mutate = random.randint(0, self.n - 1)
            while to_mutate in mutated:
                to_mutate = random.randint(0, self.n - 1)
            self.individuals[to_mutate].mutate()
            mutated.append(to_mutate)

    def print_best(self) -> None:
        """
        Methode qui affiche le meilleur individu de la population
        """
        fitness_list = []
        for individual in self.individuals:
            match self.fitness_method:
                case 1:
                    fitness_list.append(individual.fitness1(self.pm))
                case 2:
                    fitness_list.append(individual.fitness2(self.pm, self.alpha))
                case 3:
                    fitness_list.append(individual.fitness3(self.pm))
                case _:
                    fitness_list.append(individual.fitness1(self.pm))
        print(self.individuals[max_i(fitness_list)].getChromozome())

    def run(self) -> None:
        """
        Boucle principale
        """
        for i in range(self.ng):
            self.select()
            self.reproduct()
            self.mutate()
        self.print_best()

class Individual:
    """
    Classe qui représente les individuts de la population (les solutions potentielles)
    """
    def __init__(self):
        self.chromozome = ""

    def setChromozome(self, c: str) -> None:
        self.chromozome = c
    
    def getChromozome(self) -> str:
        return self.chromozome

    def randomize(self, l) -> None:
        """
        Methode qui change la valeur d'un chromozome pour une valeur aléatoire
        """
        new = ""
        for i in range(l):
            new += chr(random.randint(0, 255))
        self.chromozome = new

    def fitness1(self, pm) -> int:
        """
        Première methode de fitness, fait la somme des différences entre les codages des caractères des deux chaînes.
        """
        sum = 0
        for i in range(len(self.chromozome)):
            sum += abs(ord(self.chromozome[i]) - ord(pm[i]))
        return -sum

    def fitness2(self, pm, alpha) -> int:
        """
        Deuxième methode de fitness qui compte les caractères bien placés et mal placés et qui renvoie un int pondéré par alpha
        """
        match = 0
        missed_placed = 0
        for i in range(len(self.chromozome)):
            if self.chromozome[i] == pm[i]:
                match += 1
            else:
                missed_placed += 1
        return match + alpha * missed_placed

    def fitness3(self, pm) -> int:
        """
        Troisième methode de fitness qui utilise la distance de Levenshtein
        """
        return -Levenshtein.distance(self.chromozome, pm)

    def mutate(self) -> None:
        """
        Methode qui change un des caractères du chromozome
        """
        new = list(self.chromozome)
        new[random.randint(0, len(new) - 1)] = chr(random.randint(0, 255))
        self.chromozome = "".join(new)