import numpy as np
from GEO_parameters import GEO_Parameters
import torch


class GEOEnv:
    def __init__(self, num_satellite = 1, num_carrier_wave = 100, num_beam = 1, num_user_dim = 32, delt_t =5, dataset = "month_2.txt", max_intervals = 8):
        self.action_space = num_carrier_wave
        self.user_space_dim = num_user_dim
        self.user_space = num_user_dim * num_user_dim
        self.delt_t =  delt_t
        self.beams = num_beam
        self.satellite = num_satellite
        self.steps_before_termination = 0
        self.datas = []
        self.dataset = dataset
        self.num_data = len(self.datas)
        self.state = np.zeros((self.action_space+1, self.user_space_dim, self.user_space_dim), dtype=np.uint8)
        self.max_intervals = 8

    def load_data(self):
        with open(self.dataset, 'r') as f:
            self.datas = f.readlines()
            self.num_data = len(self.datas)

    def compute_state(self, index):
        # state = self.state.copy()
        self.state[-1] = np.zeros_like(self.state[-1])
        [date_t, beam_type, user, ap] = self.datas[index].split('\t')
        user = int(user)
        ap = int(ap)
        beam_type = int(beam_type)
        i = user//10
        j = user%10
        t = 0
        if ap == 0:
            t = beam_type
        else:
            t = -beam_type
        self.state[-1][i][j] = t
        return self.state

    def compute_reward(self, action):
        assert type(action) == int
        assert action >= 0 and action < self.action_space
        # 功率暂时不用

        # 频谱效率
        axis2_sum = np.sum(self.state, axis= 2)
        matrix_sum = np.sum(axis2_sum, axis=1)
        if (matrix_sum)
        t = 0.0
        for i in range(self.action_space):
            if matrix_sum[i] == 0:
                t += 1/self.action_space
            else:
                t += 1/self.action_space * matrix_sum[i]/self.max_intervals
        # 用户满意率，即action所在的信道的最大可用带宽的满足程度
        b = 0.0
        acquire = matrix_sum[-1]
        supply = self.max_intervals - matrix_sum[action]
        b = min(1.0, acquire/supply)
        # 如果用户被拒绝了会有一个非常大的惩罚
        if abs(b) < 1e-7 or supply <= 0:
            b = -1.0

        # 载波共用的惩罚
        c:float = 0
        # 暂时不动
        return t + b


def renew_environment():
    pass
def step(self, action):
    self.state = self.compute_state(self.steps_before_termination)
    reward = self.compute_reward(action)

    # self.pre




if __name__ == "__main__":
    e = GEOEnv()
    e.load_data()
    print(len(e.datas))
    state = e.compute_state(0)
    print(state)
    state[1][1][2] = 4
    t = e.compute_reward(state, 1)
    print(t)