agents.py

import numpy as np
import collections


class Agent:
    '''Agent Base.'''

    def __init__(self, game, display=None):
        self.game = game
        self.display = display

    def play(self, max_iter=np.inf, verbose=False):
        n_iter = 0
        while (n_iter < max_iter) and (not self.game.end):
            direction = self.step()
            self.game.move(direction)
            n_iter += 1
            if verbose:
                print("Iter: {}".format(n_iter))
                print("======Direction: {}======".format(
                    ["left", "down", "right", "up"][direction]))
            
                if self.display is not None:
                    self.display.display(self.game)

    # added to keep track of exploration for the learning agent
    def play_learn(self, max_iter=np.inf, verbose=False):
        n_iter = 0
        transitions = []

        while (n_iter < max_iter) and (not self.game.end):
            direction = self.step()
            state = convert_state(self.game.board)
            self.game.move(direction)
            new_state = convert_state(self.game.board)
            n_iter += 1
            transition = (state, new_state, direction)

            transitions.append(transition)

            if verbose:
                print("Iter: {}".format(n_iter))
                print("======Direction: {}======".format(
                    ["left", "down", "right", "up"][direction]))
            
                if self.display is not None:
                    self.display.display(self.game)

        counter = collections.Counter(transitions)
        return counter

    def step(self):
        direction = int(input("0: left, 1: down, 2: right, 3: up = ")) % 4
        return direction
    
# in order to represent states more simply, convert from matrix to string
# required for use of Counter()
def convert_state(state):
    numbers = ""
    
    for row in range(len(state)):
        for col in range(len(state[row])):
            numbers += str(state[row][col])
                
    return numbers


class RandomAgent(Agent):

    def step(self):
        direction = np.random.randint(0, 4)
        return direction


class ExpectiMaxAgent(Agent):

    def __init__(self, game, display=None):
        if game.size != 4:
            raise ValueError(
                "`%s` can only work with game of `size` 4." % self.__class__.__name__)
        super().__init__(game, display)
        from .expectimax import board_to_move
        self.search_func = board_to_move

    def step(self):
        direction = self.search_func(self.game.board)
        return direction