import tensorflow as tf 
import pathlib
import gradio as gr
import matplotlib.pyplot as plt 
from huggingface_hub import from_pretrained_keras
import numpy as np

# Normalizing the images to [-1, 1]
def normalize_test(input_image):
  input_image = tf.cast(input_image, tf.float32)
  input_image = (input_image / 127.5) - 1
  return input_image
  
def resize(input_image, height, width):
  input_image = tf.image.resize(input_image, [height, width],
                                method=tf.image.ResizeMethod.NEAREST_NEIGHBOR)
  return input_image
  
def load_image_infer(image_file):
  input_image = resize(image_file, 256, 256)
  input_image = normalize_test(input_image)

  return input_image
  
def generate_images(test_input):
  test_input = load_image_infer(test_input)
  prediction = generator(np.expand_dims(test_input, axis=0), training=True)
  fig = plt.figure(figsize=(128, 128))
  title = ['Predicted Image']

  plt.title('Predicted Image')
  # Getting the pixel values in the [0, 1] range to plot.
  plt.imshow(prediction[0,:,:,:] * 0.5 + 0.5)
  plt.axis('off')
  return fig
  
  
generator = from_pretrained_keras("keras-io/pix2pix-generator")


img = gr.inputs.Image(shape=(256,256))
plot = gr.outputs.Image(type="plot")

description = "Conditional GAN model that translates image-to-image."
gr.Interface(generate_images, inputs = img, outputs = plot, 
title = "Pix2Pix Shoes Reconstructor", description = description, examples = [["./img.png"]]).launch()