import tensorflow as tf
from tensorflow.keras.optimizers import RMSprop
from tensorflow.keras.preprocessing.image import ImageDataGenerator
from sklearn.model_selection import train_test_split
import os

!wget --no-check-certificate \
 https://dicodingacademy.blob.core.windows.net/picodiploma/ml_pemula_academy/rockpaperscissors.zip \
 -O /tmp/rockpaperscissors.zip
 

import zipfile
local_zip = '/tmp/rockpaperscissors.zip'
zip_ref = zipfile.ZipFile(local_zip,'r')
zip_ref.extractall('/tmp')
zip_ref.close()

base_dir ='/tmp/rockpaperscissors/rps-cv-images'

os.listdir(base_dir)

base_dir='/tmp/rockpaperscissors'
train_dir=os.path.join(base_dir,'train')
validation_dir=os.path.join(base_dir,'validation')
rock_dir=os.path.join(base_dir,'rock')
paper_dir=os.path.join(base_dir,'paper')
scissor_dir=os.path.join(base_dir,'scissors')
 

os.mkdir(train_dir)
os.mkdir(validation_dir)

train_rock=os.path.join(train_dir,'rock')
train_paper=os.path.join(train_dir,'paper')
train_scissor=os.path.join(train_dir,'scissor')
val_rock=os.path.join(validation_dir,'rock')
val_paper=os.path.join(validation_dir,'paper')
val_scissor=os.path.join(validation_dir,'scissor')

os.mkdir(train_rock)
os.mkdir(train_paper)
os.mkdir(train_scissor)
os.mkdir(val_rock)
os.mkdir(val_paper)
os.mkdir(val_scissor)

#split data rock dengan proporsi 60 40  antara train dan validation
train_rock_dir,val_rock_dir=train_test_split(os.listdir(rock_dir), test_size=0.4)

#split data paper dengan proporsi 60 40 antara train dan validation
train_paper_dir,val_paper_dir=train_test_split(os.listdir(paper_dir), test_size=0.4)

#split data scissor dengan proporsi 60 40 antara train dan validation
train_scissor_dir,val_scissors_dir=train_test_split(os.listdir(scissor_dir),test_size=0.4)

import tensorflow as tf
print(tf.__version__)

base_dir='/tmp/rockpaperscissors'
train_dir=os.path.join(base_dir,'train')
val_dir=os.path.join(base_dir,'validation')

os.listdir(train_dir)

os.listdir(val_dir)

#membuat directory rock pada direktori data training
train_rock_dir=os.path.join(train_dir,'rock')

#membuat directory paper pada directory data training
train_paper_dir=os.path.join(train_dir,'paper')

#membuat directory scissor pada directory data training
train_scissor_dir=os.path.join(train_dir,'scissors')

#membuat directory rock pada directory data validasi
val_rock_dir=os.path.join(val_dir,'rock')

#membuat directory paper pada directory data validasi
val_paper_dir=os.path.join(val_dir,'paper')

#membuat directory scissor pada directory data validasi
val_scissors_dir=os.path.join(val_dir,'scissor')

from tensorflow.keras.preprocessing.image import ImageDataGenerator

train_datagen=ImageDataGenerator(
                  rescale=1./255,
                  rotation_range=20,
                  horizontal_flip=True,
                  shear_range=0.2,
                  fill_mode='wrap',
                  validation_split=0.4)

test_datagen=ImageDataGenerator(
                  rescale=1./255,
                  rotation_range=20,
                  horizontal_flip=True,
                  shear_range=0.2,
                  fill_mode='wrap',
                  validation_split=0.4)

train_generator=train_datagen.flow_from_directory(
        base_dir,
        target_size=(150,150), 
        class_mode='categorical',
        shuffle=True,
        subset='training'
        )

validation_generator=test_datagen.flow_from_directory(
        base_dir,
        target_size=(150,150), 
        class_mode='categorical',
        shuffle=True,
        subset='validation',
        )

model = tf.keras.models.Sequential([
    tf.keras.layers.Conv2D(32, (3,3), activation='relu', input_shape=(150, 150, 3)),
    tf.keras.layers.MaxPooling2D(2, 2),
    tf.keras.layers.Conv2D(64, (3,3), activation='relu'),
    tf.keras.layers.MaxPooling2D(2,2),
    tf.keras.layers.Conv2D(128, (3,3), activation='relu'),
    tf.keras.layers.MaxPooling2D(2,2),
    tf.keras.layers.Conv2D(128, (3,3), activation='relu'),
    tf.keras.layers.MaxPooling2D(2,2),
    tf.keras.layers.Flatten(),
    tf.keras.layers.Dense(512, activation='relu'),
    tf.keras.layers.Dense(3, activation='softmax')
])

# compile model dengan 'adam' optimizer loss function 'categorical_crossentropy' 
model.compile(loss='categorical_crossentropy',
              optimizer=tf.optimizers.Adam(),
              metrics=['accuracy'])

# latih model dengan model.fit 
model.fit(
      train_generator,
      steps_per_epoch=25,  # berapa batch yang akan dieksekusi pada setiap epoch
      epochs=20, # tambahkan eposchs jika akurasi model belum optimal
      validation_data=validation_generator, # menampilkan akurasi pengujian data validasi
      validation_steps=5,  # berapa batch yang akan dieksekusi pada setiap epoch
      verbose=2)