import numpy as np
import tensorflow as tf
import random
import linecache
import cv2


path = "D:\\Audio_Dataset\\Animals\\annotations\\annotations\\trainval.txt"
imagepath = "D:\\Audio_Dataset\\Animals\\annotations\\images"
def get_batch(path) :
    file_object  = open(path, 'r')
    num_lines = sum(1 for line in open(path))
    list = [random.randint(0,num_lines) for x in range(20)]
    x_batch = []
    y_labels = []
    for x in list:
        photoname = linecache.getline(path, x)
        photoname = photoname.split(" ")
        imagename = photoname[0]
        imagearray = read_image(imagepath+"\\"+imagename+".jpg", 28)
        y_labels.append(photoname[1])
        x_batch.append(imagearray.flatten())


    file_object.close()

    return x_batch, np.array(y_labels)


def read_image(image_path, size):
    image = cv2.imread(image_path, cv2.IMREAD_GRAYSCALE)
    image = cv2.resize(image,(size,size))
    #data = tf.convert_to_tensor(image)
    data = np.array(image)
    return data

def init_weights(shape) :
    init_random_dist = tf.truncated_normal(shape, stddev=0.1)
    return tf.Variable(init_random_dist)

def init_bias(shape) :
    init_bias_val = tf.constant(0.1,shape=shape)
    return tf.Variable(init_bias_val)

def conv2d(x,W) :
    return tf.nn.conv2d(x,W,strides=[1,1,1,1], padding='SAME')

def max_pool_2by2(x) :
    return tf.nn.max_pool(x,ksize=[1,2,2,1],strides=[1,2,2,1],padding='SAME')

def conv_layer(input_x,shape) :
    W = init_weights(shape)
    b = init_bias([shape[3]])
    return tf.nn.relu(conv2d(input_x, W)+b)

def full_layer(input_layer, size) :
    input_size = int(input_layer.get_shape()[1])
    W = init_weights([input_size, size])
    b = init_bias([size])
    return tf.matmul(input_layer,W) + b



## Placeholders

x = tf.placeholder(tf.float32, shape=[None, 784])
y_true = tf.placeholder(tf.float32,shape=[None, 37])

x_image = tf.reshape(x, [-1,28,28,1])

convo_1 = conv_layer(x_image, shape=[5, 5, 1, 32])
convo_1_pooling =  max_pool_2by2(convo_1)

convo_2 = conv_layer(convo_1_pooling,shape=[5, 5, 32, 64])
convo_2_pooling = max_pool_2by2(convo_2)

convo_2_flat = tf.reshape(convo_2_pooling, [-1, 7*7*64])
full_layer_one = tf.nn.relu(full_layer(convo_2_flat, 1024))

hold_prob = tf.placeholder(tf.float32)
full_one_dropout = tf.nn.dropout(full_layer_one,keep_prob=hold_prob)

y_pred = full_layer(full_one_dropout, 2)

cross_entropy = tf.reduce_mean(tf.nn.softmax_cross_entropy_with_logits_v2(labels=y_true, logits=y_pred))
optimizer = tf.train.AdamOptimizer(learning_rate=0.001)
train = optimizer.minimize(cross_entropy)

init = tf.global_variables_initializer()

steps = 500
with tf.Session() as sess:
    sess.run(init)
    batch_size = 20
    for i in range(steps) :
        batch_x , batch_y = get_batch(path)
        y2 = ['1','2']
        batch_size = 20
        sess.run(train, feed_dict={x: batch_x, y_true: batch_y, hold_prob: 0.5})

        if i%100==0 :
            print("ON STEP: {}".format(i))
            print("Accuracy:")
            matchs = tf.euqal(tf.argmax(y_pred,1), tf.argmax(y_true,1))
            acc = tf.reduce_mean(tf.cast(matchs,tf.float32))