app.py

import gradio as gr
from transformers import pipeline
from nltk.tokenize import sent_tokenize

# Initialize the summarizer pipelines (abstractive and extractive)
summarizer_abstractive = pipeline("summarization", model="facebook/bart-large-cnn")
summarizer_extractive = pipeline("summarization", model="bert-extractive-summarizer")

# Function to summarize text using the selected model
def summarize_text(text, model_type="Abstractive", max_length=150, min_length=50):
    # Tokenize the text into sentences
    sentences = sent_tokenize(text)
    if model_type == "Abstractive":
        # Process using the abstractive model (BART)
        summary = summarizer_abstractive(text, max_length=max_length, min_length=min_length, do_sample=False)
        return summary[0]['summary_text']
    
    elif model_type == "Extractive":
        # Process using the extractive model (BERT)
        summary = summarizer_extractive(text)
        return ' '.join([sentence['sentence'] for sentence in summary])

# Create Gradio Interface
demo = gr.Interface(
    fn=summarize_text,
    inputs=[
        gr.Textbox(placeholder="Enter your long text here", label="Input Text", lines=10),
        gr.Radio(choices=["Abstractive", "Extractive"], label="Summarization Method", value="Abstractive"),
        gr.Slider(minimum=50, maximum=500, step=10, label="Max Length", value=150),
        gr.Slider(minimum=10, maximum=150, step=5, label="Min Length", value=50),
    ],
    outputs="text",
    title="Advanced Text Summarizer",
    description="This tool provides both abstractive and extractive summarization options, allowing you to select the best method and adjust summary length.",
    live=True,
)

# Launch the interface
demo.launch()