import tensorflow as tf
from tensorflow.examples.tutorials.mnist import input_data
#number 1 to 10 data
mnist=input_data.read_data_sets('MNIST_data',one_hot=True)
def compute_accuracy(v_xs,v_ys):
    global prediction
    y_pre=sess.run(prediction,feed_dict={xs:v_xs,keep_prob:1})
    correct_prediction=tf.equal(tf.argmax(y_pre,1),tf.argmax(v_ys,1))
    accuracy=tf.reduce_mean(tf.cast(correct_prediction,tf.float32))
    result=sess.run(accuracy,feed_dict={xs:v_xs,ys:v_ys})
    return result
def weight_variable(shape):
    inital=tf.truncated_normal(shape,stddev=0.1)
    
    return tf.Variable(inital)
def bias_variable(shape):
    initial=tf.constant(0.1,shape=shape)
    return tf.Variable(initial)
def conv2d(x,W):
#    stride[1,x_movement,y_movement,1]
#    Must have strides[0]=strides[4]=1
    return tf.nn.conv2d(x,W,strides=[1,1,1,1],padding='SAME')#MUST have strides
def max_pool_2x2(x):
    return tf.nn.max_pool(x,ksize=[1,2,2,1],strides=[1,2,2,1],padding='SAME')
#define placeholder for inputs to network
xs=tf.placeholder(tf.float32,[None,784])#28x28
ys=tf.placeholder(tf.float32,[None,10])
keep_prob=tf.placeholder(tf.float32)

x_image=tf.reshape(xs,[-1,28,28,1])
print(x_image.shape)#[n_samples,28,28,1]
##conv1 layer##
W_conv1=weight_variable([5,5,1,32])#patch5x5,in size 1,out size 32
b_conv1=bias_variable([32])
h_conv1=tf.nn.relu(conv2d(x_image,W_conv1)+b_conv1)#output size 28x28x32
h_pool1=max_pool_2x2(h_conv1)    #output size 14x14x32

##conv2 layer##
W_conv2=weight_variable([5,5,32,64])#patch5x5,in size 32,out size 64
b_conv2=bias_variable([64])
h_conv2=tf.nn.relu(conv2d(h_pool1,W_conv2)+b_conv2)#output size 14x14x64
h_pool2=max_pool_2x2(h_conv2)    #output size 

##func1 layer
W_fc1=weight_variable([7*7*64,1024])
b_fc1=bias_variable([1024])
#[n_samples,7,7,64]---->[n_samples,7*7*64]
h_pool2_flat=tf.reshape(h_pool2,[-1,7*7*64])
h_fc1=tf.nn.relu(tf.matmul(h_pool2_flat,W_fc1)+b_fc1)
h_fc1_drop=tf.nn.dropout(h_fc1,keep_prob)

##func2 layer    
W_fc2=weight_variable([1024,10])
b_fc2=bias_variable([10])
prediction=tf.nn.softmax(tf.matmul(h_fc1_drop,W_fc2)+b_fc2)

#the error between prediction and read data
cross_entropy=tf.reduce_mean(-tf.reduce_sum(ys*tf.log(prediction),reduction_indices=[1]))
train_step=tf.train.AdamOptimizer(1e-4).minimize(cross_entropy)
sess=tf.Session()

#important step
sess.run(tf.global_variables_initializer())
for i in range(1000):
    batch_xs,batch_ys=mnist.train.next_batch(100)
    sess.run(train_step,feed_dict={xs:batch_xs,ys:batch_ys,keep_prob:0.5})
    if i%50==0:
        print(compute_accuracy(mnist.test.images,mnist.test.labels))