import os

os.environ['TF_CPP_MIN_LOG_LEVEL'] = '2'
import tensorflow as tf
from tensorflow import keras
from tensorflow.keras import datasets, layers, optimizers, Sequential, metrics
import datetime
from matplotlib import pyplot as plt
import io


# datasets用於資料集的管理，layers用於dense層，optimizer優化器，Sequential容器，metrics測試用的度量器


# 對資料集作預處理(對每一個x，y樣本)
def preprocess(x, y):
    x = tf.cast(x, dtype=tf.float32) / 255.
    y = tf.cast(y, dtype=tf.int32)
    return x, y


def plot_to_image(figure):
    # 把圖片儲存為png形式
    buf = io.BytesIO()
    plt.savefig(buf, format='png')
    # 關掉圖片，防止它直接在notebook裡display
    plt.close(figure)
    buf.seek(0)
    # 將圖片從PNG轉化為TF image
    image = tf.image.decode_png(buf.getvalue(), channels=4)
    image = tf.expand_dims(image, 0)
    return image


def image_grid(images):
    figure = plt.figure(figsize=(10, 10))
    for i in range(25):
        plt.subplot(5, 5, i + 1, title='name')
        plt.xticks([])
        plt.yticks([])
        plt.grid(False)
        plt.imshow(images[i], cmap=plt.cm.binary)
    return figure


batchsz = 128
(x, y), (x_val, y_val) = datasets.mnist.load_data()
db = tf.data.Dataset.from_tensor_slices((x, y))
db = db.map(preprocess).shuffle(60000).batch(batchsz).repeat(10)

db_val = tf.data.Dataset.from_tensor_slices((x_val, y_val))
db_val = db_val.map(preprocess).batch(batchsz, drop_remainder=True)  # test部分不需要shuffle

# 構建一個5層的網路
model = Sequential([
    layers.Dense(256, activation=tf.nn.relu),  # [b,784]---->[b,256]
    layers.Dense(128, activation=tf.nn.relu),  # [b,256]---->[b,128]
    layers.Dense(64, activation=tf.nn.relu),  # [b,128]---->[b,64]
    layers.Dense(32, activation=tf.nn.relu),  # [b,64]---->[b,32]
    layers.Dense(10)  # [b,32]---->[b,10] 最後一層不用啟用函式 330=32*10+10
])
model.build(input_shape=(None, 28 * 28))  # 完成一個建立工作
model.summary()  # 起到除錯作用，列印網路結構

# 優化器更新過程 w=w-lr*dw 只需要傳入一個list即可對所有的w進行更新
optimizer = optimizers.Adam(lr=0.01)  # Adam只是一種引數最優化的方法，其他的還有SGD、Momentum

# 建立一個summary，可以往中間喂資料
current_time = datetime.datetime.now().strftime("%Y%m%d-%H%M%S")
log_dir = 'logs/' + current_time
summary_writer = tf.summary.create_file_writer(log_dir)

# [128,28,28]
sample_img = next(iter(db))[0]
# [28,28]
sample_img = sample_img[0]

sample_img = tf.reshape(sample_img, [1, 28, 28, 1])
#餵了一張圖片
with summary_writer.as_default():
    tf.summary.image("Training sample:", sample_img, step=0)

for step, (x, y) in enumerate(db):
    with tf.GradientTape() as tape:
        x = tf.reshape(x, (-1, 28 * 28))
        out = model(x)
        y_onehot = tf.one_hot(y, depth=10)
        loss = tf.reduce_mean(tf.losses.categorical_crossentropy(y_onehot, out, from_logits=True))
    grads = tape.gradient(loss, model.trainable_variables)  # 求所有導數
    optimizer.apply_gradients(zip(grads, model.trainable_variables))  # 引數原地更新

    if step % 100 == 0:
        print(step, 'loss:', float(loss))
        with summary_writer.as_default():
            tf.summary.scalar('train-loss', float(loss), step=step)

    if step % 500 == 0:
        total_correct = 0.
        total_num = 0.
        for _,(x, y) in enumerate(db_val):
            x = tf.reshape(x, (-1, 28 * 28))
            # test的時候不需要求梯度
            out = model(x)
            # logits=>prob
            pred = tf.argmax(out, axis=1)  # 得到最大值的索引 int64
            pred = tf.cast(pred,dtype=tf.int32)

            correct = tf.reduce_sum(tf.cast(tf.equal(pred, y), dtype=tf.int32)).numpy()
            total_correct += correct  # correct型別是tensor，total_correct型別是numpy
            total_num += x.shape[0]  # 加上batch的個數

        acc = total_correct / total_num
        print(step, 'test acc:', acc)

        val_images = x[:25]
        val_images = tf.reshape(val_images,[-1,28,28,1])
        #喂accuracy
        with summary_writer.as_default():
            tf.summary.scalar('test-acc',float(acc),step=step)
            #喂很多圖片，但是以離散的方式呈現
            tf.summary.image("val-onebyond-images:",val_images,max_outputs=25,step=step)
            #喂很多圖片，但是經過人為的拼接了
            val_images = tf.reshape(val_images,[-1,28,28])
            figure = image_grid(val_images)
            tf.summary.image('val-images:',plot_to_image(figure),step=step)