env_100.py

from interrupt import Macro as mc
import image_matching2 as im
class env():

    def __init__(self):
        self.actions = []
        self.actions.append(0)
        self.pre_score = 0
        '''
        Test codes~~~~~~
        this should be changed to variables version
        def __init__(self,observations which is cv image matched data,actions which is macros):
        '''
        ### observation_space
        self.observation_space = im.get_shape()
        self.time_step = 0;
        self.best_time_step = 0;
        self.is_alive = True
        ### actions

        macros = []

        macro1 = mc.Macro()
        macro2 = mc.Macro()
        # new_macro.setDelay(3)
        macro1.setKeyPress('left')
        macros.append(macro1)
        macro2.setKeyPress('right')
        macros.append(macro2)

        # new_macro.setDelay(3)
        # macros.append(macro2)



        '''
        test codes
        end here
        '''
        self.actions.append(macro1)
        self.actions.append(macro2)

    def set_actions(self, macros):
        '''
        set actions
        for test this is temporary func
        '''
        self.actions = macros

    def step(self, action_num):


        ### should really think about how game over is recongnized
        # self.is_alive, next_state = get_state()#Capture
        ### should really think about how game over is recongnized


        self.time_step+=1
        if(action_num>0):#do nothing if no op
            self.actions[action_num].runMacro()

        score = 0
        '''
        score = get_score_from_ocr
        '''
#        print("t = ",self.time_step)
#        print("best_t = ",self.best_time_step)
        # if(pre_score < score):
        #     reward = 1
        # elif(pre_score == score):
        #     reward = 0
        # else:
        #     reward = -1

        # pre_score = score;

        #reward = 0 # get from ocr

        done, next_state = im.get_state()#Capture

        if(self.time_step>self.best_time_step):
            reward = 1
        elif(self.time_step>int(self.best_time_step*0.75)):
            reward = abs(self.time_step)/self.best_time_step
        elif(done):
            reward = -1
        else:
            reward = 0
        if(done):
            self.best_time_step = max(self.best_time_step,self.time_step)

        return (next_state, reward, done, 0)

    def reset(self):
        reset_macro = mc.Macro()
#         self.pre_score = 0
        reset_macro.setMouseClick('left')
        reset_macro.runMacro();
        self.time_step = 0
        self.is_alive = True
        _, state = im.get_state()
        return state