# Load the mystery data here and cluster using k-means (now you can use libraries e.g. sklearn)
import numpy as np
import pandas as pd
mystery = np.load('mystery.npy')
mystery.shape
meta_data=pd.read_csv('mnist_784.csv', low_memory=False)


data = meta_data.to_numpy()
print(type(data))
# mystery[0]


what = mystery[500]
rgba = np.reshape(what, (196,4))
# print(rgba)

import matplotlib.pyplot as plt
mystery.shape
x = mystery[500]
# x = x[:-1]
# 196 RGBA tuples
x2 = np.reshape(x, (196,4))
# print(x2)
# plt.imshow(x2, cmap="gray")
x_vals = []
y_vals = []
c_vals = []
# print('#%02x%02x%02x' % tuple(x2[0]))
for i in range(14):
    for j in range(14):

            x_vals.append(i)
            y_vals.append(j)
            c_vals.append(tuple(x2[i*i+ j]))
#             opacity =
#             c_vals.append('#%02x%02x%02x' % tuple(x2[j*j+ i]))
# print(c_vals)
c_vals2 = []
for c in c_vals:
    if c[3] != 0:
        lst = list(c)
        # lst[3] = lst[3] / 255.0
        del lst[3]
        c_vals2.append(tuple(lst))
    else:
        c_vals2.append(tuple(lst))
#         c[3] = c[3] / 255.0
# c_vals[0] = (0,0,0,0.2)
# print(c_vals2)


plt.scatter(x_vals, y_vals, color=c_vals2)
# plt.scatter(x_vals, y_vals, c=c_vals)
        # x_vals
# plt.imshow(arr, cmap='gray', vmin=0, vmax=255)
# x = np.reshape(x, (28, 28))
# print(x)
# f = open("out.txt", "w+")
# f.write(np.array_str(x))
# f.close()