# Εκπαιδευμένα βάρη και διαδικασία prediction
import random
import math
import csv
import torch
import torch.nn as nn


class SimpleMLP(nn.Module):
    def __init__(self):
        super(SimpleMLP, self).__init__()
        self.layer1 = nn.Linear(6, 32)
        self.layer2 = nn.Linear(32, 16)
        self.output = nn.Linear(16, 1)


def sigmoid(x):
    return 1 / (1 + math.exp(-x))

def relu(x):
  if x>0 :
    return x
  else :
    return 0.001*x



def calculate_hidden1(inputs, hidden1):
    for j in range(hidden1):
        a = 0
        for i in range(inputs):
            a = a + inp[i] * i_h1_w[j][i]  #i, j
        a = relu(a + h1_b[j])
        h1.append(a)
       
        
def calculate_hidden2(hidden1, hidden2):    
    for j in range(hidden2):
        a = 0
        for i in range(hidden1):
            a = a + h1[i] * h1_h2_w[j][i]  #i,j
        a = relu(a + h2_b[j])
        h2.append(a)
        
def calculate_output(hidden2):
        a = 0
        for i in range(hidden2):
            a = a + h2[i] * h2_o_w[i]
        return sigmoid(a + o_b)   


# ΚΥΡΙΩΣ ΠΡΟΓΡΑΜΜΑ
i_h1_w = []
h1_h2_w = []
h2_o_w = []
h1_b = []
h2_b = []
outp = 0
o_b = 0


model = SimpleMLP()
model.load_state_dict(torch.load("road_danger_mlp_weights.pth"))
model_dict = model.state_dict()
i_h1_w = model_dict['layer1.weight'].tolist()
h1_b = model_dict['layer1.bias'].tolist()
h1_h2_w = model_dict['layer2.weight'].tolist()
h2_b = model_dict['layer2.bias'].tolist()
h2_o_w = model_dict['output.weight'].tolist()[0]
o_b = float(model_dict['output.bias'].tolist()[0])

mse = 0
count = 0
# Open file
with open("datafile2.csv") as file_obj:
    # Skips the heading
    heading = next(file_obj)
    # Create reader object
    reader_obj = csv.reader(file_obj) 
    # Iterate over each row in the csv file
    for row in reader_obj:
        #print(row)
        inp = []
        for i in range(7):
            if i<6:
                inp.append(float(row[i]))
            else:
                target = float(row[i])
        #print(inp)
        h1 = []
        calculate_hidden1(6, 32)
        h2 = []
        calculate_hidden2(32, 16)
        outp = calculate_output(16) 
        error = target - outp
        print("output: ", outp, "  target: ", target, "  error: ", error)
        mse = mse + error**2
        count = count + 1
 
mse = mse / count
print("MSE: ", mse)