# 1. Import the required packages
import torch
import gradio as gr

from typing import Dict
from transformers import pipeline

# 2. Define function to use our model on given text 
def food_not_food_classifier(text: str) -> Dict[str, float]:
    # Set up text classification pipeline
    food_not_food_classifier = pipeline(task="text-classification", 
                                        # Because our model is on Hugging Face already, we can pass in the model name directly
                                        model="devagonal/bert-f1-durga-muhammad-c", # link to model on HF Hub
                                        device="cuda" if torch.cuda.is_available() else "cpu",
                                        top_k=None) # return all possible scores (not just top-1)
    
    # Get outputs from pipeline (as a list of dicts)
    outputs = food_not_food_classifier(text)[0]

    # Format output for Gradio (e.g. {"label_1": probability_1, "label_2": probability_2})
    output_dict = {}
    for item in outputs:
        output_dict[item["label"]] = item["score"]

    return output_dict

# 3. Create a Gradio interface with details about our app
description = """
a text classifier to determine question class.
label0 = durga
label1 = muhammad
"""

demo = gr.Interface(fn=food_not_food_classifier, 
             inputs="text", 
             outputs=gr.Label(num_top_classes=2), # show top 2 classes (that's all we have)
             title="Bert F1 Durga Muhammad c",
             description=description,
             examples=[["siapakah durga"],
                       ["siapakah muhammad"]])

# 4. Launch the interface
if __name__ == "__main__":
    demo.launch()