dd

import random

class Layer:
    def __init__(self, nodesIn, nodesOut):
        self.nodesIn = nodesIn
        self.nodesOut = nodesOut
        self.weights = [[0.0] * self.nodesOut] * self.nodesIn

        for x in range(self.nodesIn):
            for y in range(self.nodesOut):
                self.weights[x][y] = random.random() - 0.5

    def forward(self, inputs):
        self.inputs = inputs
        if(len(self.inputs)) != self.nodesIn:
            print("Incorrect number of inputs to the layer")
        else:
            self.output = [0.0] * self.nodesOut
            for x in range(self.nodesIn):
                for y in range(self.nodesOut):
                    self.output[y] += self.inputs[x] * self.weights[x][y]

    def getCost(self, expectedOutput):
        self.expectedOutput = expectedOutput
        self.error = [0.0] * len(self.output)
        for x in range(len(self.nodesOut)):
            self.error[x] = self.output[x] - self.expectedOutput[x]

        return self.error

    def updateWeightsGradients(self, gradientsPerNeuron):
        self.dWeights = [[0.0] * self.nodesOut] * self.nodesIn
        for x in range(self.nodesIn):
            for y in range(self.nodesOut):
                self.dWeights[x][y] += gradientsPerNeuron[y] * self.inputs[x]

        for x in range(self.nodesIn):
            for y in range(self.nodesOut):
                self.weights[x][y] -= self.dWeights[x][y]



class Network:
    def __init__(self, layers):
        self.layers = layers

    def forward(self, inputs):
        self.layers[0].forward(inputs)
        for x in range(1, len(self.layers)):
            self.layers[x].forward(self.layers[x - 1].output)
        self.output = self.layers[len(self.layers) - 1].output

    def updateWeights(self, error):
        #Get the cost of the final output
        self.error = error
        self.dError = 2 * self.error
        self.gradientsPerNeuron = [[] * 20] * len(self.layers)
        print(self.gradientsPerNeuron)
        #self.layers[2].updateWeightsGradients(self.gradientsPerNeuron)


neuralNetwork = Network([
    Layer(1, 3),
    Layer(3, 2),
    Layer(2, 1)])

neuralNetwork.forward([3])

print(neuralNetwork.output)

#neuralNetwork.layers[2].updateWeightsGradients([3])
neuralNetwork.updateWeights([5])
neuralNetwork.forward([3])

print(neuralNetwork.output)