app.py

import streamlit as st
from transformers import pipeline

st.set_page_config(page_title="LLM Text Summarizer", layout="centered")
#st.title("LLM-Powered Text Summarizer")
#st.markdown("This app summarizes long texts using a Hugging Face transformer model (`facebook/bart-large-cnn`).")

@st.cache_resource
def load_model():
    return pipeline("summarization", model="facebook/bart-large-cnn")

summarizer = load_model()
st.markdown("<h1 style='text-align: center;'>🧠 Smart Document Summarizer using BART</h1>", unsafe_allow_html=True)
st.markdown("<p style='text-align: center; color: gray;'>This app splits long text into chunks and summarizes each part individually for better results.</p>", unsafe_allow_html=True)

# Two columns: Input and Output
col1, col2 = st.columns(2)

with col1:
    st.markdown("### 📝 Enter Document")
    text = st.text_area("Paste your long document here...", height=400, label_visibility="collapsed")

with col2:
    st.markdown("### 🧾 Combined Summary")
    summary_box = st.empty()
#text = st.text_area("Enter text to summarize", height=300)
text = " ".join(text.split()[:700])  # Truncate input to ~700 words


if st.button("Summarize"):
    if len(text.strip()) == 0:
        st.warning("Please enter some text first.")
    else:
        word_count = len(text.split())

        if word_count > 1300:
            max_tokens = 1300
        elif word_count > 300:
            max_tokens = 350
        else:
            max_tokens = 100

        with st.spinner("Generating summary..."):
            summary = summarizer(
                text,
                max_length=max_tokens,
                min_length=int(max_tokens * 0.5),
                do_sample=False
            )
            st.subheader("Summary:")
            st.success(summary[0]['summary_text'])