import os
os.system('pip install streamlit transformers torch')

import streamlit as st
from transformers import BartTokenizer, BartForConditionalGeneration

# Load the model and tokenizer
model_name = 'ibrahimgiki/facebook_bart_base_new'

tokenizer = BartTokenizer.from_pretrained(model_name)
model = BartForConditionalGeneration.from_pretrained(model_name)

def summarize_text(text):
    inputs = tokenizer(text, return_tensors="pt", truncation=True, padding="longest")
    summary_ids = model.generate(
        inputs["input_ids"],
        max_length=150,
        min_length=30,
        length_penalty=2.0,
        num_beams=4,
        early_stopping=True
    )
    summary = tokenizer.decode(summary_ids[0], skip_special_tokens=True)
    return summary

st.title("Text Summarization with Fine-Tuned Model")
st.write("Enter text to generate a summary using the fine-tuned summarization model.")

text = st.text_area("Input Text", height=200)
if st.button("Summarize"):
    if text:
        with st.spinner("Summarizing..."):
            summary = summarize_text(text)
            st.success("Summary Generated")
            st.write(summary)
    else:
        st.warning("Please enter some text to summarize.")

if __name__ == "__main__":
    st.set_option('deprecation.showfileUploaderEncoding', False)
    st.markdown(
        """
        <style>
        .reportview-container {
            flex-direction: row;
            justify-content: center.
        }
        </style>
        """,
        unsafe_allow_html=True
    )