app.py

import os
import yaml
import gdown 
import gradio as gr
from predict import PredictTri

output_path = "tashkeela-d2.pt"
if not os.path.exists(output_path):
    model_gdrive_id = "1FGelqImFkESbTyRsx_elkKIOZ9VbhRuo"
    gdown.download(id=model_gdrive_id, output=output_path, quiet=False)

output_path = "vocab.vec"
if not os.path.exists(output_path):
    vocab_gdrive_id = "1-0muGvcSYEf8RAVRcwXay4MRex6kmCii"
    gdown.download(id=vocab_gdrive_id, output=output_path, quiet=False)

with open("config.yaml", 'r', encoding="utf-8") as file:
    config = yaml.load(file, Loader=yaml.FullLoader)

config["train"]["max-sent-len"] = config["predictor"]["window"]
config["train"]["max-token-count"] = config["predictor"]["window"] * 3

def diacritze(text):
    print(text)
    predictor = PredictTri(config, text)
    diacritized_lines = predictor.predict_majority_vote()
    return '\n'.join(diacritized_lines)

with gr.Blocks() as demo:
    gr.Markdown(
    """
    # Partial Diacritization
    TODO: put paper links here
    """)
    input_txt = gr.Textbox(
        placeholder="اكتب هنا", 
        lines=5, 
        label="Input",
        type='text',
        rtl=True,
        text_align='right',
    )

    output_txt = gr.Textbox(
        lines=5, 
        label="Output",
        type='text',
        rtl=True,  
        text_align='right',
    )

    btn = gr.Button(value="Shakkel")
    btn.click(diacritze, inputs=input_txt, outputs=output_txt)

if __name__ == "__main__":
    demo.launch()