"""A very simple MNIST classifier.

See extensive documentation at
https://www.tensorflow.org/get_started/mnist/beginners
"""
import argparse
import sys
from tensorflow.examples.tutorials.mnist import input_data
import tensorflow as tf

FLAGS = None
num_epochs = 1000

def main(_):
    # lecture des données
    mnist = input_data.read_data_sets(FLAGS.data_dir, one_hot=True)

    # Modèle
    x = tf.placeholder(tf.float32 , [None, 784]) # batchs d’images
    W = tf.Variable(tf.zeros([784, 10])) # poids
    b = tf.Variable(tf.zeros([10])) # biais
    y = tf.matmul(x, W) + b # sortie du réseau

    y_ = tf.placeholder(tf.float32 , [None, 10]) # labels théoriques

    #Fonction de perte
    cross_entropy = tf.reduce_mean(tf.nn.softmax_cross_entropy_with_logits(labels=y_, logits=y))
    #méthode d’optimisation
    train_step = tf.train.GradientDescentOptimizer(0.5).minimize( cross_entropy )

    #démarrage d’une session TF
    sess = tf.InteractiveSession()
    tf.global_variables_initializer().run()
    # Train
    for _ in range(num_epochs):
        batch_xs, batch_ys = mnist.train.next_batch(100)
        sess.run(train_step, feed_dict={x: batch_xs, y_: batch_ys})

    # test du modèle
    correct_prediction = tf.equal(tf.argmax(y, 1), tf.argmax(y_, 1))
    accuracy = tf.reduce_mean(tf.cast(correct_prediction , tf.float32))
    print(sess.run(accuracy, feed_dict={x: mnist.test.images, y_: mnist.test.labels }))

if __name__ == '__main__':
    parser = argparse.ArgumentParser()
    parser.add_argument('--data_dir', type=str , default='../data/mnistdata', help='Directory for storing input data')                                                                          ,
    FLAGS, unparsed = parser.parse_known_args()
    tf.app.run(main=main, argv=[sys.argv[0]] + unparsed)