import os
import matplotlib.pyplot as plt
import numpy as np
import pandas as pd

nb_files = os.listdir(".." + os.sep + "export")

size = len(nb_files)


def mean_mkn(arr: list[tuple[int, np.ndarray]]) -> np.ndarray:
    averages_mkn = np.empty((size, 2))
    nb = 0
    for nb_users, data in arr:
        rb = data[:, 4]

        total = 0.0
        for x in rb:
            total = total + x
        average = total / len(rb)
        averages_mkn[nb, 0] = nb_users
        averages_mkn[nb, 1] = average
        nb += 1
    return averages_mkn


def rb_available(arr: list[tuple[int, np.ndarray]]) -> np.ndarray:
    available = np.zeros((size, 2))
    nb = 0
    for nb_users, data in arr:
        available[nb, 0] = nb_users
        available[nb, 1] = (data.shape[0] / (200 * 10000)) * 100
        nb += 1
    return available


def delay(arr: list[tuple[int, np.ndarray]]) -> np.ndarray:
    delays = np.zeros((size, 2))
    nb = 0
    for nb_users, data in arr:
        d = data[:, 5]
        for x in d:
            delays[nb, 0] = nb_users
            delays[nb, 1] = float(x)
        nb += 1
    return delays

def rb_available_distance(arr: list[tuple[int, np.ndarray]], distance) -> np.ndarray:
    available = np.zeros((size, 2))
    nb = 0
    for nb_users, data in arr:
        n = 0
        for x in data[:,6]:
            if int(x) == distance:
                n+=1

        available[nb, 0] = nb_users
        available[nb, 1] = n/ (200 * 10000) * 100
        nb += 1
    return available


np_arr: list[tuple[int, np.ndarray]] = list()
for i in nb_files:
    np_arr.append((int(i.split(".")[0]), pd.read_csv(".." + os.sep + "export" + os.sep + i, delimiter=';').to_numpy()))

averages = mean_mkn(np_arr)
available = rb_available(np_arr)

available_lp1 = rb_available_distance(np_arr, 200)
available_lp2 = rb_available_distance(np_arr, 400)

delays = delay(np_arr)
delays.sort(axis=0)
# Data for plotting
averages.sort(axis=0)

fig, ax = plt.subplots(2, 2)
ax[0, 0].scatter(averages[:, 0], averages[:, 1])
ax[0, 0].set(xlabel='number of users', ylabel='Efficacité spectrale', title='Efficacité spectrale')
ax[0, 0].grid()

ax[0, 1].scatter(available[:, 0], available[:, 1])
ax[0, 1].set(xlabel='number of users', ylabel='RB utilisés', title='Pourcentage de RB utilisés')
ax[0, 1].grid()

ax[1, 0].scatter(delays[:, 0], delays[:, 1])
ax[1, 0].set(xlabel='number of users', ylabel='delays(ms)', title='Delay')
ax[1, 0].grid()

#ax[1, 1].scatter(available[:, 0], (available_lp1[:, 1]/available_lp2[:, 1])/available[:, 1] * 100)
ax[1, 1].scatter(available[:, 0], available_lp1[:, 1])
ax[1, 1].scatter(available[:, 0], available_lp2[:, 1])
ax[1, 1].set(xlabel='number of users', ylabel='RB utilisés proche/loin/total', title='RB utilisés distance')
ax[1, 1].grid()

plt.show()