from ple.games.flappybird import FlappyBird

from ple import PLE

import numpy as np

from flappybird import Agent,Model,DQN,preprocess

import cv2

import time

game = FlappyBird()

env = PLE(game, fps=30, display_screen=True)

env.init()

# env.reset_game()

actionset = env.getActionSet()

state = env.getGameState()

action_dim = len(actionset)

obs_shape = len(state)

print(action_dim,obs_shape)

print(actionset,state)

# 根据parl框架构建agent

model = Model(act_dim=action_dim)

algorithm = DQN(model, act_dim=action_dim, gamma=0.99, lr=0.001)

agent = Agent(

e_greed_decrement=0) # 随着训练逐步收敛，探索的程度慢慢降低

# 加载模型

# save_path = '.\model_dir\model_6700_2823.0.ckpt' #episode_reward: 1785.0

save_path = '.\model_dir\model2_2400_735.6.ckpt'

agent.restore(save_path)

obs = list(env.getGameState().values())

#处理obs

obs = preprocess(obs)

episode_reward = 0

while True:

# 预测动作，只选最优动作

action = agent.predict(obs)

# 图像太快休眠

# time.sleep(0.02)

# # 新建窗口显示分数

# observation = env.getScreenRGB()

# score = env.score()

# # 格式转换

# observation = cv2.cvtColor(observation,cv2.COLOR_RGB2BGR)

# # 选择90度

# observation = cv2.transpose(observation)

# font = cv2.FONT_HERSHEY_SIMPLEX

# observation = cv2.putText(observation, "score:"+str(int(score)), (0, 30), font, 0.6, (0, 0, 255), 2)

# cv2.imshow("flappybird", observation)

# cv2.waitKey(10)

reward= env.act(actionset[action])

obs = list(env.getGameState().values())

#处理obs

obs = preprocess(obs)

done = env.game_over()

episode_reward += reward

if done:

break

print("episode_reward:",episode_reward)

cv2.destroyAllWindows()