import mdp
import random

class Qlearner :

    def __init__(self, states, actions, alpha, gamma, world) :
        self.qtable = {}
        self.states = states
        self.actions = actions
        for s in states :
            for a in actions :
                self.qtable[(s,a)] = 0.0
        self.world = world
        self.temp = 2.0
        self.alpha = alpha
        self.gamma = gamma

        
    ### use boltzmann exploration to select an action
    ### you do this one 
    def selectAction(self, state) :


    ### you do this one
    def update(self, oldstate, action, newstate, reward) :
        

class World :
    def __init__(self, map) :
        self.map = map

    ### move the agent from one state to another according to the map's 
    ### transition function.
    def newState(self, statenum, action) :
        ttable = self.map.states[statenum].transitions[action]
        val = random.random()
        total = 0.0
        i = -1
        while total < val :
            i += 1
            total += ttable[i][0]
        return ttable[i][1].coords
        

    def atGoal(self, state) :
        return self.map.states[state].isGoal

    def reward(self, state) :
        if self.map.states[state].isGoal :
            return self.map.states[state].utility
        else :
            return -0.04


### you do this one.
def learn() :