""""
Patrick E. Lanigan
CS 1573 - Homework 02

Problems 2.7, 2.8, 2.9

Based on agents2.py, which itself was adapted from the AIMA website.

Implements a performance measuring environment simulator for the 2D vacuum
cleaner world pictured in figure 2.2 on page 34 of AIMA.

Also implements a random agent, simple reflex agent and a model based
reflex agent which tracks the state of the environment. The agents correctly
perceive their location and their location's status.

The performance measure awards points for each clean square at each time
step, and penalizes for movements. The amount of reward and penalty
can be specified in the environment constructor. The values may be
set to zero (ie to remove the penalty). The compare_agents function
can be used to compare the performance measures of the agents.

The TableDrivenVacuumAgent was left unimplemented. Since the precept sequence
in this case would be 1000 time-steps long, the table would require an
unreasonable number of entries to complete.

"""

from utils import *
import random, copy, operator

loc_A, loc_B = (0, 0), (1, 0)       # The two locations for the Vacuum world

#______________________________________________________________________________

class Object:
    """This represents any physical object that can appear in an Environment.
    You subclass Object to get the objects you want.  Each object can have a
    .__name__  slot (used for output only)."""

    def __repr__(self):
        return '<%s>' % getattr(self, '__name__', self.__class__.__name__)

    def is_alive(self):
        """MOD: Objects should never die"""
        return True

class Agent(Object):
    """An Agent is a subclass of Object with one required slot,
    .program, which should hold a function that takes one argument, the
    percept, and returns an action. (What counts as a percept or action
    will depend on the specific environment in which the agent exists.)
    Note that 'program' is a slot, not a method.  If it were a method,
    then the program could 'cheat' and look at aspects of the agent.
    It's not supposed to do that: the program can only look at the
    percepts.  An agent program that needs a model of the world (and of
    the agent itself) will have to build and maintain its own model.
    There is an optional slot, .performance, which is a number giving
    the performance measure of the agent in its environment."""

    def __init__(self):
        def program(percept):
            return raw_input('Percept=%s; action? ' % percept)

        self.program = program
        self.alive = True

def TraceAgent(agent):
    """Modify the agent's program to print its input and output. This will let
    you see what the agent is doing in the environment."""
    old_program = agent.program

    def new_program(percept):
        action = old_program(percept)
        print '%s perceives %s and does %s' % (agent, percept, action)
        return action

    agent.program = new_program
    return agent

#______________________________________________________________________________

#class TableDrivenVacuumAgent(Agent):
#    """This agent selects an action based current precept using a table
#       lookup based on the percept sequence. Since the precept sequence
#       in this case would be 1000 time steps long, it would be unreasonable
#       to implement the entire table."""

#table = {((loc_A, 'Clean'),): 'Right',
#         ((loc_A, 'Dirty'),): 'Suck',
#         ((loc_B, 'Clean'),): 'Left',
#         ((loc_B, 'Dirty'),): 'Suck',
#         ((loc_A, 'Clean'), (loc_A, 'Clean')): 'Right',
#         ((loc_A, 'Clean'), (loc_A, 'Dirty')): 'Suck',
#         # ...
#         ((loc_A, 'Clean'), (loc_A, 'Clean'), (loc_A, 'Clean')): 'Right',
#         ((loc_A, 'Clean'), (loc_A, 'Clean'), (loc_A, 'Dirty')): 'Suck',
#         # ...
#         }

#    def __init__(self):
#        "Supply as table a dictionary of all {percept_sequence:action} pairs."
#        ## The agent program could in principle be a function, but because
#        ## it needs to store state, we make it a callable instance of a class.
#        percepts = []

#        def program(percept):
#            percepts.append(percept)
#            action = self.table.get(tuple(percepts))
#            return action

#        self.program = program

class RandomVacuumAgent(Agent):
    "An agent that chooses an action at random, ignoring all percepts."
    def __init__(self, actions = ['Right', 'Left', 'Suck', 'NoOp']):
        self.program = lambda percept: random.choice(actions)

class ReflexVacuumAgent(Agent):
    "A reflex agent for the two-state vacuum environment. [Fig. 2.8]"
    def __init__(self):
        def program((location, status)):
            if status == 'Dirty': return 'Suck'
            elif location == loc_A: return 'Right'
            elif location == loc_B: return 'Left'

        self.program = program

class ModelBasedVacuumAgent(Agent):
    "An reflex agent that keeps track of what locations are clean or dirty."
    def __init__(self):
        model = {loc_A: None, loc_B: None}

        def program((location, status)):
            model[location] = status # Update the model here

            if model[loc_A] == model[loc_B] == 'Clean': return 'NoOp'
            elif status == 'Dirty': return 'Suck'
            elif location == loc_A: return 'Right'
            elif location == loc_B: return 'Left'

        self.program = program

#______________________________________________________________________________

class Environment:
    """Abstract class representing an Environment.  'Real' Environment classes
    inherit from this. Your Environment will typically need to implement:
        percept:           Define the percept that an agent sees.
        execute_action:    Define the effects of executing an action.
                           Also update the agent.performance slot.
    The environment keeps a list of .objects and .agents (which is a subset
    of .objects). Each agent has a .performance slot, initialized to 0.
    Each object has a .location slot, even though some environments may not
    need this."""

    object_classes = [] ## List of classes that can go into environment

    def __init__(self,):
        self.objects = []
        self.agents = []

    def add_object(self, object, location = random.choice([loc_A, loc_B])):
        """Add an object to the environment, setting its location. Also keep
        track of objects that are agents.  Shouldn't need to override this."""
        object.location = location
        self.objects.append(object)

        if isinstance(object, Agent):
            object.performance = 0
            self.agents.append(object)

        return self

    def execute_action(self, agent, action):
        "Change the world to reflect this action. Override this."
        abstract()

    def is_done(self):
        "By default, we're done when we can't find a live agent."
        return not find_if(method('is_alive'), self.agents)

    def percept(self, agent):
        "Return the percept that the agent sees at this point. Override this."
        abstract()

    def run(self, steps = 1000):
        """Run the Environment for given number of time steps."""
        for step in range(steps):
            if self.is_done(): return
            self.step()

    def step(self):
        """Run the environment for one time step. If the
        actions and exogenous changes are independent, this method will
        do. If there are interactions between them, you'll need to
        override this method."""
        if not self.is_done():
            actions = [agent.program(self.percept(agent)) for agent in self.agents]

            for (agent, action) in zip(self.agents, actions):
                self.execute_action(agent, action)

#______________________________________________________________________________

class VacuumEnvironment(Environment):
    """This environment has two locations, A and B. Each can be Dirty or Clean.
    The agent perceives its location and the location's status. This serves as
    an example of how to implement a simple Environment."""

    def __init__(self):
        "Randomly initializes the clean/dirty status of the two squares"
        Environment.__init__(self)
        self.status = {loc_A:random.choice(['Clean', 'Dirty']),
                       loc_B:random.choice(['Clean', 'Dirty'])}
        self.reward = 1
        self.penalty = 0

    def __init__(self, status_A, status_B, reward, penalty):
        "MOD: Constructor to specify the clean/dirty status of the two squares"
        Environment.__init__(self)
        self.status = {loc_A:status_A, loc_B:status_B }
        self.reward = reward
        self.penalty = penalty

    def execute_action(self, agent, action):
        """Change agent's location and/or location's status; track performance.
        Score 1 for each clean square; no movement penalty"""
        if action == 'Right':
            agent.location = loc_B
            agent.performance -= self.penalty
        elif action == 'Left':
            agent.location = loc_A
            agent.performance -= self.penalty
        elif action == 'Suck':
            self.status[agent.location] = 'Clean'

        # MOD: update performace measure for each clean square at each time step
        if self.status[loc_A] == 'Clean':
            agent.performance += self.reward;
        if self.status[loc_B] == 'Clean':
            agent.performance += self.reward;

    def percept(self, agent):
        "Returns the agent's location, and the location status (Dirty/Clean)."
        return (agent.location, self.status[agent.location])

#______________________________________________________________________________

def compare_agents(envs, AgentFactories, init_loc, steps):
    """See how well each of several agents do in n instances of an environment.
    Pass in a factory (constructor) for environments, and several for agents.
    Create n instances of the environment, and run each agent in copies of each one.
    Return a list of (agent-name, scores, avg) tuples."""
    results = []

    for A in AgentFactories:
        scores = []
        total = 0

        for env in copy.deepcopy(envs):
            agent = A()
            env.add_object(agent, init_loc)
            env.run(steps)
            total += agent.performance
            scores.append(agent.performance);   # MOD: save each score for later

        results.append((agent, scores, total / float(len(envs))))

    return results

#______________________________________________________________________________

### Output results header ###

print "\n" + ("-" * 72)
print """Output takes the following form:
\t<Agent Name>\t[Raw Scores*]\tAverage Score

\t*Raw scores are listed by configuration...
\t(Clean, Clean), (Clean, Dirty), (Dirty, Clean), (Dirty, Dirty)"""

### Experiments for 2.7 & 2.8 ###

agents = [RandomVacuumAgent, ReflexVacuumAgent, ModelBasedVacuumAgent]

envs = [VacuumEnvironment('Clean', 'Clean', 1, 0),
        VacuumEnvironment('Clean', 'Dirty', 1, 0),
        VacuumEnvironment('Dirty', 'Clean', 1, 0),
        VacuumEnvironment('Dirty', 'Dirty', 1, 0)]

results_A = compare_agents(envs, agents, loc_A, 1000)
results_B = compare_agents(envs, agents, loc_B, 1000)

### Output results of experiments ###

print "\n:::Problems 2.7 & 2.8:::"

print "\nFrom location (%s, %s)" % loc_A
print "\t%-23s\t%-16s\t%s" % results_A[0]
print "\t%-23s\t%-16s\t%s" % results_A[1]
print "\t%-23s\t%-16s\t%s" % results_A[2]

print "\nFrom location (%s, %s)" % loc_B
print "\t%-23s\t%-16s\t%s" % results_B[0]
print "\t%-23s\t%-16s\t%s" % results_B[1]
print "\t%-23s\t%-16s\t%s" % results_B[2]

### Experiments for 2.9b ###

agents = [RandomVacuumAgent, ReflexVacuumAgent, ModelBasedVacuumAgent]

envs = [VacuumEnvironment('Clean', 'Clean', 1, 1),
        VacuumEnvironment('Clean', 'Dirty', 1, 1),
        VacuumEnvironment('Dirty', 'Clean', 1, 1),
        VacuumEnvironment('Dirty', 'Dirty', 1, 1)]

results_A = compare_agents(envs, agents, loc_A, 1000)
results_B = compare_agents(envs, agents, loc_B, 1000)

### Output results of experiments ###

print "\n:::Problem 2.9b:::"

print "\nFrom location (%s, %s)" % loc_A
print "\t%-23s\t%-16s\t%s" % results_A[0]
print "\t%-23s\t%-16s\t%s" % results_A[1]
print "\t%-23s\t%-16s\t%s" % results_A[2]

print "\nFrom location (%s, %s)" % loc_B
print "\t%-23s\t%-16s\t%s" % results_B[0]
print "\t%-23s\t%-16s\t%s" % results_B[1]
print "\t%-23s\t%-16s\t%s" % results_B[2]