import gradio as gr
import os
import torch

from model import create_efficientnet
from timeit import default_timer as timer
from typing import Tuple, Dict

# Setup class names
with open("class_names.txt", "r") as f:
  class_names = [food_name.strip() for food_name in f.readlines()]

### Model and transforms preparation ###
# Create model and transforms
effnet, effnet_transforms = create_efficientnet(output_shape=101)

# Load saved weights
effnet.load_state_dict(
    torch.load(f="effnetv2L_100_percent.pth",
               map_location=torch.device("cpu")) # load to CPU
)

### Predict function ###
def predict(img) -> Tuple[Dict, float]:
  # Start a timer
  start_time = timer()

  # Transform the input image for use with EffNetB2
  img = effnet_transforms(img).unsqueeze(0) # unsqueeze = add batch dimension on 0th index

  # Put model into eval mode, make prediction
  effnet.eval()
  with torch.inference_mode():
    # Pass transformed image through the model and turn the prediction logits into probaiblities
    pred_probs = torch.softmax(effnet(img), dim=1)

  # Create a prediction label and prediction probability dictionary
  pred_labels_and_probs = {class_names[i]: float(pred_probs[0][i]) for i in range(len(class_names))}

  # Calculate pred time
  end_time = timer()
  pred_time = round(end_time - start_time, 4)

  # Return pred dict and pred time
  return pred_labels_and_probs, pred_time

### 4. Gradio app ###
# Create title, description and article
title = "Food101 classifier"
description = "An [EfficientNetV2 feature extractor](https://pytorch.org/vision/main/models/efficientnetv2.html) computer vision model to classify images [101 classes of food from the Food101 dataset](https://github.com/mrdbourke/pytorch-deep-learning/blob/main/extras/food101_class_names.txt)."
article = "Built with [Gradio](https://github.com/gradio-app/gradio) and [PyTorch](https://pytorch.org/)."

# Create example list
example_list = [["examples/" + example] for example in os.listdir("examples")]

# Create the Gradio demo
demo = gr.Interface(fn=predict, # maps inputs to outputs
                    inputs=gr.Image(type="pil"),
                    outputs=[gr.Label(num_top_classes=5, label="Predictions"),
                             gr.Number(label="Prediction time (s)")],
                    examples=example_list,
                    title=title,
                    description=description,
                    article=article)

# Launch the demo
demo.launch()