def player_selection_to_bot(selection):
if selection == 1:
return gofish.HumanController()
elif selection == 2:
return bots.RandomBot()
elif selection == 3:
return bots.BadBot()
elif selection == 4:
return bots.GoodBot(default_ask='random_high')
elif selection == 5:
return bots.GoodBot(default_ask='same_high')
elif selection == 6:
return bots.GoodBot(default_ask='random_all')
raise ValueError('Incorrect Selection')
#!/usr/bin/env python3
# -*- coding: utf-8 -*-
"""
Created on Fri Apr 3 07:23:14 2020
@author: terrylines
"""
import OX_short
import DP_bot
import bots
import MonteCarloEpsilonGreedy_bot
player1 = bots.RandomBot("Terry")
player2 = bots.RandomBot("Mikey")
player3 = bots.OXMasterBot("Peter")
player4 = DP_bot.OXDynamicProgrammingBot("Nick")
player5 = DP_bot.OXDynamicProgrammingBot("Rob")
# show that two random players get a zero avg scroe
scores2 = player2.play(Game=OX_short.OX,
reward=bots.marginReward,
opponents=[player1],
repeats=1)
sum(scores2) / 10000
# a simple tactic of taking centre gets an 0.2 avg score
scores3 = player3.play(Game=OX_short.OX,
reward=bots.marginReward,
opponents=[player1],
repeats=10000)
import gofish_env
import bots
replay_buffer_size = 20000
fc_layer_params = (256,64)
learning_rate = 5e-5
init_collection_steps = 10000
num_iterations = 100000
collect_steps_per_iteration = 1
log_interval = 100
eval_interval = 10000
num_eval_episodes = 100
train_reward_run_amount = 0.99
bot = bots.RandomBot()
memory_features = True
lose_on_illegal_move = False
drawless = False
conv_features = True
train_py_environment = gofish_env.GoFishEnv(bot, max_visible_opponent_hand_size=10, max_visible_deck_size=10, drawless=drawless, lose_on_illegal_move=lose_on_illegal_move, memory_features=memory_features, conv_features=conv_features)
print('Validating env.')
utils.validate_py_environment(train_py_environment, episodes=5)
print('Validation complete.')
eval_py_environment = gofish_env.GoFishEnv(bot, max_visible_opponent_hand_size=10, max_visible_deck_size=10, drawless=drawless, lose_on_illegal_move=lose_on_illegal_move, memory_features=memory_features, conv_features=conv_features)
train_env = tf_py_environment.TFPyEnvironment(train_py_environment)
eval_env = tf_py_environment.TFPyEnvironment(eval_py_environment)
def main(argv):
if len(argv) == 1:
print(
'To play against bot or local, run binary with appropriate flags:')
print('\tpython3 run_game.py bot')
print('\tpython3 run_game.py remote SERVER_ADDRESS')
window = pygame.display.set_mode((WIDTH, WIDTH))
pygame.display.set_caption("Split")
opponent = 'local' if len(argv) == 1 else argv[1] # bot/local/remote
if opponent == 'remote':
host = argv[2]
ui_state = UIState(
window,
players=[remote_play.RemotePlayer(host),
game.GamePlayer()])
if ui_state.players[0].client_id == 0:
temp_p = ui_state.players.pop(0)
ui_state.players.append(temp_p)
else:
opponent_player = bots.RandomBot(
) if opponent == 'bot' else game.GamePlayer()
ui_state = UIState(window,
players=[opponent_player,
game.GamePlayer()])
random.shuffle(ui_state.players)
run = True
while run:
current_player = ui_state.players[ui_state.state.next_player - 1]
other_player = ui_state.players[2 - ui_state.state.next_player]
if not current_player.local_human:
if ui_state.intro_state == 1:
choice = current_player.choose_color(ui_state.state)
ui_state.choose_color(choice)
other_player.update_color_choice(choice)
else:
move = current_player.get_move(ui_state.state)
ui_state.make_move(move)
other_player.update_move(move)
for event in pygame.event.get():
if event.type == pygame.QUIT:
pygame.quit()
# Reverse transform mouse
mouse_position = pygame.mouse.get_pos()
mouse_position = game.GamePoint(
(mouse_position[0] - ui_state.pan[0]) / ui_state.zoom,
(mouse_position[1] - ui_state.pan[1]) / ui_state.zoom)
if ui_state.can_draw_fill():
if event.type == pygame.MOUSEBUTTONDOWN:
if event.button == 4:
ui_state.zoom *= 1.5
elif event.button == 5:
ui_state.zoom /= 1.5
elif event.button == 3:
ui_state.cancel_move()
else:
ui_state.mouse_click(mouse_position)
if event.type == pygame.MOUSEMOTION:
ui_state.mouse_move(mouse_position)
keys = pygame.key.get_pressed() #checking pressed keys
if keys[pygame.K_w]:
ui_state.pan[1] += 1
if keys[pygame.K_a]:
ui_state.pan[0] += 1
if keys[pygame.K_s]:
ui_state.pan[1] -= 1
if keys[pygame.K_d]:
ui_state.pan[0] -= 1
if ui_state.can_choose_color():
if keys[pygame.K_1]:
ui_state.choose_color(True)
if keys[pygame.K_2]:
ui_state.choose_color(False)
ui_state.render()
def test():
p1 = bots.RandomBot(consts.X)
p2 = bots.RandomBot(consts.O)
game = TicTacToe(p1, p2)
game.run(100000)