From 45abd8c7ebc3ef824c9d2195c8c95b5c5e1f35ae Mon Sep 17 00:00:00 2001
From: Yorick Barbanneau <ephase@xieme-art.org>
Date: Thu, 14 Dec 2023 21:58:35 +0100
Subject: [PATCH] First commit: minmax and alphabeta implemented

---
 README.md                     |  48 +++++++
 src/.envrc                    |   1 +
 src/classes/CustomFormater.py |  24 ++++
 src/classes/Engines.py        | 187 +++++++++++++++++++++++++++
 src/classes/Reversi.py        | 232 ++++++++++++++++++++++++++++++++++
 src/game.py                   |  93 ++++++++++++++
 src/shell.nix                 |   8 ++
 7 files changed, 593 insertions(+)
 create mode 100644 README.md
 create mode 100644 src/.envrc
 create mode 100644 src/classes/CustomFormater.py
 create mode 100644 src/classes/Engines.py
 create mode 100644 src/classes/Reversi.py
 create mode 100755 src/game.py
 create mode 100644 src/shell.nix

diff --git a/README.md b/README.md
new file mode 100644
index 0000000..44b9f12
--- /dev/null
+++ b/README.md
@@ -0,0 +1,48 @@
+IA: jeu de Reversi
+------------------
+
+Le but de ce projet est d'implémenter plusieurs mécanisme de jeu (humain et 
+intelligence artificielle) pour le jeu de Reversi
+
+## Installation
+
+Le programme utilise des outils standard de Python installé de base : `random`,
+`math`, `argpase` et `logging`. Le project est fourni avec un shell *Nix* dans
+le répertoire `src`
+
+## Utilisation
+
+le programme propose un emsemble d'options en ligne de commande afin de définir
+les options du jeu comme le choix des implementations de jeu (aléatoine, MinMax
+etc.) ou encore les paramètres (profondeur de recherche). Une aide est intégrée
+au programme via la commande `./game.py -h`. Voici quelques exemple de
+lancement:
+
+```shell
+# Lancement de base: les deux joueurs jouent avec le moteur aléatoire:
+./game.py
+
+# joueur noir humain et joueur blanc MinMax avec une profondeur de 5
+./game.py -b human -w minmax --depth 5
+```
+
+## Choix d'implémentation
+
+### Classes PlayerEngine
+
+Définies dans le fichier `./src/classes/Engines.py`, les classes utilisées h
+eritent de la classe de base `PlayerEngines` :
+
+```python
+class PlayerEngine(Object):
+    def __init__(self, logger, options):
+    def get_move(self, board):
+
+class MinmaxPlayerEngine(PlayerEngine):
+    def get_move(board):
+
+class RandomPlayerEngnine(PlayerEngine):
+    def get_move(board):
+```
+
+Il est ainsi plus aisé de tester les moteur dans notre programme de base.
diff --git a/src/.envrc b/src/.envrc
new file mode 100644
index 0000000..1d953f4
--- /dev/null
+++ b/src/.envrc
@@ -0,0 +1 @@
+use nix
diff --git a/src/classes/CustomFormater.py b/src/classes/CustomFormater.py
new file mode 100644
index 0000000..02ad065
--- /dev/null
+++ b/src/classes/CustomFormater.py
@@ -0,0 +1,24 @@
+import logging
+
+class CustomFormatter(logging.Formatter):
+
+    grey = "\x1b[0;35m"
+    blue = "\x1b[34;20m"
+    yellow = "\x1b[33;20m"
+    red = "\x1b[31;20m"
+    bold_red = "\x1b[31;1m"
+    reset = "\x1b[0m"
+    format = "%(levelname)s: %(message)s (%(filename)s:%(lineno)d)"
+
+    FORMATS = {
+        logging.DEBUG: blue + format + reset,
+        logging.INFO: grey + format + reset,
+        logging.WARNING: yellow + format + reset,
+        logging.ERROR: red + format + reset,
+        logging.CRITICAL: bold_red + format + reset
+    }
+
+    def format(self, record):
+        log_fmt = self.FORMATS.get(record.levelno)
+        formatter = logging.Formatter(log_fmt)
+        return formatter.format(record)
diff --git a/src/classes/Engines.py b/src/classes/Engines.py
new file mode 100644
index 0000000..c3b4ea5
--- /dev/null
+++ b/src/classes/Engines.py
@@ -0,0 +1,187 @@
+import random, math
+
+class PlayerEngine:
+    def __init__(self, logger, options: dict = {}):
+        # init logger do display informations
+        self.logger = logger
+        self.options = options
+        self.logger.info("Init engine {}".format(self.__class__.__name__))
+
+    def get_move(self, board):
+        self.logger.info("engine: {} - player:{}".format(
+            self.__class__.__name__, 
+            self.get_player_name(board._nextPlayer)
+        ))
+
+    @staticmethod
+    def get_player_name(player):
+        return 'White (O)' if player is 2 else 'Black (X)'
+
+class RandomPlayerEngine(PlayerEngine):
+    def get_move(self, board):
+        super().get_move(board)
+        moves = board.legal_moves()
+        return random.choice(moves)
+
+class HumanPlayerEngine(PlayerEngine):
+    def get_move(self, board):
+        super()
+        move = None
+        while move is None:
+            user_input = input("Please enter player {} move: ".format(
+                self.get_player_name(board._nextPlayer)
+            ))
+            move = self.validate_input(user_input, board, player)
+            print("{}".format(move))
+        return move
+
+    @staticmethod
+    def validate_input(input, board):
+        if input == 'print':
+            print("\n{}".format(board.__str__))
+            return None
+
+        if input == 'help':
+            print('{}'.format(board.legal_moves()))
+            return None
+
+        if len(input) != 2:
+            return None
+
+        x = int(input[0])
+        y = int(input[1])
+        if not board.is_valid_move(int(player), x, y):
+            return None
+
+        return [board._nextPlayer, x, y]
+
+class MinmaxPlayerEngine(PlayerEngine):
+    def get_move(self, board):
+        super().get_move(board)
+        value = -math.inf
+        nodes = 1
+        leafs = 0
+        move = ''
+        for m in board.legal_moves():
+            board.push(m)
+            v, n, l = self.checkMinMax(board, False, self.options['depth'] - 1)
+            if v > value:
+                value = v
+                move = m
+                self.logger.debug("found a better move: {} (heuristic:{})".format(
+                    move,
+                    value
+                    ))
+            nodes += n
+            leafs += l
+            board.pop()
+
+        self.logger.debug("Tree statistics:\n\tnodes:{}\n\tleafs:{}".format(
+            nodes,
+            leafs
+            ))
+        return move
+
+    def checkMinMax(self, board, friend_move:bool, depth :int = 2):
+        nodes = 1
+        leafs = 0
+        move = ''
+        if depth == 0:
+            leafs +=1
+            return board.heuristique(), nodes, leafs
+        
+        if friend_move:
+            value = -math.inf
+            for m in board.legal_moves():
+                board.push(m)
+                v, n, l = self.checkMinMax(board, False, depth - 1)
+                if v > value:
+                    value = v
+                nodes += n
+                leafs += l
+                board.pop()
+
+        else:
+            value = math.inf
+            for m in board.legal_moves():
+                board.push(m)
+                v, n, l = self.checkMinMax(board, True, depth - 1)
+                if v < value:
+                    value = v
+                board.pop();
+                nodes += n
+                leafs += l
+        return value, nodes, leafs
+
+class AlphabetaPlayerEngine(PlayerEngine):
+
+    def get_move(self, board):
+        super().get_move(board)
+        alpha = -math.inf
+        beta = math.inf
+        nodes = 1
+        leafs = 0
+        move = []
+        value = -math.inf
+        for m in board.legal_moves():
+            board.push(m)
+            v, n, l = self.checkAlphaBeta(board, False, self.options['depth'] - 1, alpha, beta)
+            board.pop()
+            alpha = max(alpha,v)
+            nodes += n
+            leafs += l
+            if alpha >= value:
+                value = alpha
+                move = m
+                self.logger.debug("found a better move: {} (heuristic:{})".format(
+                    move,
+                    alpha
+                ))
+
+        self.logger.debug("Tree statistics:\n\tnodes:{}\n\tleafs:{}".format(
+            nodes,
+            leafs
+        ))
+        return move
+
+
+    def checkAlphaBeta(self, board, friend_move : bool, depth, alpha, beta):
+        nodes = 1
+        leafs = 0
+        if depth == 0 :
+            leafs +=1
+            return board.heuristique(), nodes, leafs
+
+        if friend_move:
+            value = -math.inf
+            for m in board.legal_moves():
+                board.push(m)
+                v, n, l = self.checkAlphaBeta(board, False, depth - 1, alpha, beta)
+                board.pop()
+                alpha = max(value,v)
+                nodes += n
+                leafs += l
+                if alpha >= beta:
+                    self.logger.debug("Alpha pruning - alpha:{} / beta:{}".format(
+                        alpha,
+                        beta
+                    ))
+                    return beta, nodes, leafs
+            return alpha, nodes, leafs
+
+        else:
+            value = math.inf
+            for m in board.legal_moves():
+                board.push(m)
+                v, n, l = self.checkAlphaBeta(board, True, depth - 1, alpha, beta)
+                board.pop();
+                beta = min(beta,v)
+                nodes += n
+                leafs += l
+                if alpha >= beta:
+                    self.logger.debug("Beta pruning - alpha:{} / beta:{}".format(
+                        alpha,
+                        beta
+                    ))
+                    return alpha, nodes, leafs
+            return beta, nodes, leafs
diff --git a/src/classes/Reversi.py b/src/classes/Reversi.py
new file mode 100644
index 0000000..0d4d72d
--- /dev/null
+++ b/src/classes/Reversi.py
@@ -0,0 +1,232 @@
+# -*- coding: utf-8 -*-
+
+''' Fichier de règles du Reversi 
+    Certaines parties de ce code sont fortement inspirée de 
+    https://inventwithpython.com/chapter15.html
+
+    '''
+
+class Board:
+    _BLACK = 1
+    _WHITE = 2
+    _EMPTY = 0
+
+    # Attention, la taille du plateau est donnée en paramètre
+    def __init__(self, boardsize = 8):
+      self._nbWHITE = 2
+      self._nbBLACK = 2
+      self._nextPlayer = self._BLACK
+      self._boardsize = boardsize
+      self._board = []
+      for x in range(self._boardsize):
+          self._board.append([self._EMPTY]* self._boardsize)
+      _middle = int(self._boardsize / 2)
+      self._board[_middle-1][_middle-1] = self._BLACK 
+      self._board[_middle-1][_middle] = self._WHITE
+      self._board[_middle][_middle-1] = self._WHITE
+      self._board[_middle][_middle] = self._BLACK 
+      
+      self._stack= []
+      self._successivePass = 0
+
+    def reset(self):
+        self.__init__()
+
+    # Donne la taille du plateau 
+    def get_board_size(self):
+        return self._boardsize
+
+    # Donne le nombre de pieces de blanc et noir sur le plateau
+    # sous forme de tuple (blancs, noirs) 
+    # Peut être utilisé si le jeu est terminé pour déterminer le vainqueur
+    def get_nb_pieces(self):
+      return (self._nbWHITE, self._nbBLACK)
+
+    # Vérifie si player a le droit de jouer en (x,y)
+    def is_valid_move(self, player, x, y):
+        if x == -1 and y == -1:
+            return not self.at_least_one_legal_move(player)
+        return self.lazyTest_ValidMove(player,x,y)
+
+    def _isOnBoard(self,x,y):
+        return x >= 0 and x < self._boardsize and y >= 0 and y < self._boardsize 
+
+    # Renvoie la liste des pieces a retourner si le coup est valide
+    # Sinon renvoie False
+    # Ce code est très fortement inspiré de https://inventwithpython.com/chapter15.html
+    # y faire référence dans tous les cas
+    def testAndBuild_ValidMove(self, player, xstart, ystart):
+        if self._board[xstart][ystart] != self._EMPTY or not self._isOnBoard(xstart, ystart):
+            return False
+    
+        self._board[xstart][ystart] = player # On pourra remettre _EMPTY ensuite 
+    
+        otherPlayer = self._flip(player)
+    
+        tilesToFlip = [] # Si au moins un coup est valide, on collecte ici toutes les pieces a retourner
+        for xdirection, ydirection in [[0, 1], [1, 1], [1, 0], [1, -1], [0, -1], [-1, -1], [-1, 0], [-1, 1]]:
+            x, y = xstart, ystart
+            x += xdirection 
+            y += ydirection
+            if self._isOnBoard(x, y) and self._board[x][y] == otherPlayer:
+                # There is a piece belonging to the other player next to our piece.
+                x += xdirection
+                y += ydirection
+                if not self._isOnBoard(x, y):
+                    continue
+                while self._board[x][y] == otherPlayer:
+                    x += xdirection
+                    y += ydirection
+                    if not self._isOnBoard(x, y): # break out of while loop, then continue in for loop
+                        break
+                if not self._isOnBoard(x, y):
+                    continue
+                if self._board[x][y] == player: # We are sure we can at least build this move. Let's collect
+                    while True:
+                        x -= xdirection
+                        y -= ydirection
+                        if x == xstart and y == ystart:
+                            break
+                        tilesToFlip.append([x, y])
+    
+        self._board[xstart][ystart] = self._EMPTY # restore the empty space
+        if len(tilesToFlip) == 0: # If no tiles were flipped, this is not a valid move.
+            return False
+        return tilesToFlip
+
+    # Pareil que ci-dessus mais ne revoie que vrai / faux (permet de tester plus rapidement)
+    def lazyTest_ValidMove(self, player, xstart, ystart):
+        if self._board[xstart][ystart] != self._EMPTY or not self._isOnBoard(xstart, ystart):
+            return False
+    
+        self._board[xstart][ystart] = player # On pourra remettre _EMPTY ensuite 
+    
+        otherPlayer = self._flip(player)
+    
+        for xdirection, ydirection in [[0, 1], [1, 1], [1, 0], [1, -1], [0, -1], [-1, -1], [-1, 0], [-1, 1]]:
+            x, y = xstart, ystart
+            x += xdirection 
+            y += ydirection
+            if self._isOnBoard(x, y) and self._board[x][y] == otherPlayer:
+                # There is a piece belonging to the other player next to our piece.
+                x += xdirection
+                y += ydirection
+                if not self._isOnBoard(x, y):
+                    continue
+                while self._board[x][y] == otherPlayer:
+                    x += xdirection
+                    y += ydirection
+                    if not self._isOnBoard(x, y): # break out of while loop, then continue in for loop
+                        break
+                if not self._isOnBoard(x, y): # On a au moins 
+                    continue
+                if self._board[x][y] == player: # We are sure we can at least build this move. 
+                    self._board[xstart][ystart] = self._EMPTY
+                    return True
+                 
+        self._board[xstart][ystart] = self._EMPTY # restore the empty space
+        return False
+
+    def _flip(self, player):
+        if player == self._BLACK:
+            return self._WHITE 
+        return self._BLACK
+
+    def is_game_over(self):
+        if self.at_least_one_legal_move(self._nextPlayer):
+            return False
+        if self.at_least_one_legal_move(self._flip(self._nextPlayer)):
+            return False
+        return True 
+
+    def push(self, move):
+        [player, x, y] = move
+        assert player == self._nextPlayer
+        if x==-1 and y==-1: # pass
+            self._nextPlayer = self._flip(player)
+            self._stack.append([move, self._successivePass, []])
+            self._successivePass += 1
+            return
+        toflip = self.testAndBuild_ValidMove(player,x,y)
+        self._stack.append([move, self._successivePass, toflip])
+        self._successivePass = 0
+        self._board[x][y] = player
+        for xf,yf in toflip:
+            self._board[xf][yf] = self._flip(self._board[xf][yf])
+        if player == self._BLACK:
+            self._nbBLACK += 1 + len(toflip)
+            self._nbWHITE -= len(toflip)
+            self._nextPlayer = self._WHITE
+        else:
+            self._nbWHITE += 1 + len(toflip)
+            self._nbBLACK -= len(toflip)
+            self._nextPlayer = self._BLACK
+
+    def pop(self):
+        [move, self._successivePass, toflip] = self._stack.pop()
+        [player,x,y] = move
+        self._nextPlayer = player 
+        if len(toflip) == 0: # pass
+            assert x == -1 and y == -1
+            return
+        self._board[x][y] = self._EMPTY
+        for xf,yf in toflip:
+            self._board[xf][yf] = self._flip(self._board[xf][yf])
+        if player == self._BLACK:
+            self._nbBLACK -= 1 + len(toflip)
+            self._nbWHITE += len(toflip)
+        else:
+            self._nbWHITE -= 1 + len(toflip)
+            self._nbBLACK += len(toflip)
+
+    # Est-ce que on peut au moins jouer un coup ?
+    # Note: cette info pourrait être codée plus efficacement
+    def at_least_one_legal_move(self, player):
+        for x in range(0,self._boardsize):
+            for y in range(0,self._boardsize):
+                if self.lazyTest_ValidMove(player, x, y):
+                   return True
+        return False
+
+    # Renvoi la liste des coups possibles
+    # Note: cette méthode pourrait être codée plus efficacement
+    def legal_moves(self):
+        moves = []
+        for x in range(0,self._boardsize):
+            for y in range(0,self._boardsize):
+                if self.lazyTest_ValidMove(self._nextPlayer, x, y):
+                    moves.append([self._nextPlayer,x,y])
+        if len(moves) is 0:
+            moves = [[self._nextPlayer, -1, -1]] # We shall pass
+        return moves
+
+    # Exemple d'heuristique tres simple : compte simplement les pieces
+    def heuristique(self, player=None):
+        if player is None:
+            player = self._nextPlayer
+        if player is self._WHITE:
+            return self._nbWHITE - self._nbBLACK
+        return self._nbBLACK - self._nbWHITE
+
+    def _piece2str(self, c):
+        if c==self._WHITE:
+            return 'O'
+        elif c==self._BLACK:
+            return 'X'
+        else:
+            return '.'
+
+    def __str__(self):
+        toreturn=""
+        for l in self._board:
+            for c in l:
+                toreturn += self._piece2str(c)
+            toreturn += "\n"
+        toreturn += "Next player: " + ("BLACK" if self._nextPlayer == self._BLACK else "WHITE") + "\n"
+        toreturn += str(self._nbBLACK) + " blacks and " + str(self._nbWHITE) + " whites on board\n"
+        toreturn += "(successive pass: " + str(self._successivePass) + " )"
+        return toreturn
+
+    __repr__ = __str__
+
+
diff --git a/src/game.py b/src/game.py
new file mode 100755
index 0000000..027ab8d
--- /dev/null
+++ b/src/game.py
@@ -0,0 +1,93 @@
+#!/usr/bin/env python3
+from classes.Reversi import Board
+from classes.Engines import RandomPlayerEngine, HumanPlayerEngine, MinmaxPlayerEngine, AlphabetaPlayerEngine
+import logging as log
+import argparse as arg
+from classes.CustomFormater import CustomFormatter
+
+
+"""
+Function to parse command line arguments
+"""
+def parse_aguments():
+    engines_choices=['random', 'human', 'minmax', 'alphabeta']
+    parser = arg.ArgumentParser('Playing Reversi with (virtual) friend')
+
+    parser.add_argument('-w', '--white-engine',
+        choices=engines_choices,
+        help='white player engine (random)',
+        default='random'
+    )
+
+    parser.add_argument('-b', '--black-engine',
+        choices=engines_choices,
+        help='black player engine (random)',
+        default='random'
+    )
+
+    parser.add_argument('-D', '--depth',
+        help='Minmax exploration depth',
+        type=int,
+        default=3,
+    )
+
+    debug_group = parser.add_mutually_exclusive_group()
+    debug_group.add_argument('-V', '--verbose',
+        help='Verbose output',
+        action='store_true')
+    debug_group.add_argument('-d', '--debug',
+        help='Activate debug mode',
+        action='store_true')
+    return parser.parse_args()
+
+
+"""
+Main Function
+"""
+if __name__ == '__main__':
+    engines = {
+        "random": RandomPlayerEngine,
+        "human":  HumanPlayerEngine,
+        "minmax": MinmaxPlayerEngine,
+        "alphabeta": AlphabetaPlayerEngine,
+    }
+    print("Stating PyReverso...")
+    args = parse_aguments()
+    logger = log.getLogger()
+    print(args)
+    # Create handler for streaming to tty (stderr / stdout)
+    tty_handler = log.StreamHandler()
+    tty_handler.setFormatter(CustomFormatter())
+    logger.addHandler(tty_handler)
+    
+    # Activate verbose or debug mode
+    if args.verbose is True:
+        logger.setLevel(log.INFO)
+        logger.info('VERBOSE mode activated')
+    
+    if args.debug is True:
+        logger.setLevel(log.DEBUG)
+        logger.debug('DEBUG mode activated')
+
+    game = Board(10)
+    logger.debug("Init players engines - black:{} / white:{}".format(
+        args.black_engine,
+        args.white_engine
+        ))
+    options = {
+        'depth': args.depth
+    }
+    wplayer = engines[args.white_engine](logger, options)
+    bplayer = engines[args.black_engine](logger, options)
+    while ( not game.is_game_over()):
+        if game._nextPlayer == 1:
+            move = bplayer.get_move(game)
+        else:
+            move = wplayer.get_move(game)
+        print("Player {} move: {}".format(
+            game._nextPlayer, 
+            move
+        ))
+        game.push(move)
+    print("Game end - score black:{}, white:{}".format(game.heuristique(1), game.heuristique(2)))
+    print(game.__str__)
diff --git a/src/shell.nix b/src/shell.nix
new file mode 100644
index 0000000..55375ea
--- /dev/null
+++ b/src/shell.nix
@@ -0,0 +1,8 @@
+with (import <nixpkgs> {});
+mkShell {
+  buildInputs = [
+
+    # Defines a python + set of packages.
+    python3
+  ];
+}