import gradio as gr
import random
from transformers import AutoTokenizer, AutoModelForSeq2SeqLM
from datasets import load_dataset
import requests
from bs4 import BeautifulSoup

tokenizer = AutoTokenizer.from_pretrained("liamvbetts/bart-large-cnn-v4")
model = AutoModelForSeq2SeqLM.from_pretrained("liamvbetts/bart-large-cnn-v4")

dataset = load_dataset("cnn_dailymail", "3.0.0")

NEWS_API_KEY = "da2cc601304341e7a39cb5604d0b076b"  

def summarize(article):
    inputs = tokenizer(article, return_tensors="pt").input_ids
    outputs = model.generate(inputs, max_new_tokens=128, do_sample=False)
    summary = tokenizer.decode(outputs[0], skip_special_tokens=True)
    return summary

def get_random_article():
    random.seed()
    val_example = dataset["validation"].shuffle().select(range(1))
    val_article = val_example['article'][0][:1024]
    return val_article

def load_article():
    return get_random_article()

def get_news_article():
    url = 'https://newsapi.org/v2/top-headlines'
    news_url = ''
    params = {
        'apiKey': NEWS_API_KEY,
        'country': 'us',  # You can change this as needed
        'pageSize': 100
    }
    response = requests.get(url, params=params)
    articles = response.json().get('articles', [])
    if articles:
        random_article = random.choice(articles)
        news_url = random_article.get('url')
    else:
        return None
    
    if news_url:
        full_article, title = scrape_article(news_url)
        return full_article, title
    else:
        return "No news article found.", ""

def scrape_article(url):
    try:
        response = requests.get(url)
        soup = BeautifulSoup(response.content, 'html.parser')
        
        # Extracting the title - this is a general approach
        title = soup.title.string if soup.title else "No Title Available"

        article_content = soup.find_all('p')  # This is a simplification
        
        text = ' '.join([p.get_text() for p in article_content])
        words = text.split()
        truncated_text = ' '.join(words[:1024])  # Truncate to first 1024 words

        return truncated_text, title
    except Exception as e:
        return "Error scraping article: " + str(e), ""

# Using Gradio Blocks
with gr.Blocks() as demo:
    gr.Markdown("## News Summary App")
    gr.Markdown("Enter a news text and get its summary, or load a random article.")
    with gr.Row():
        article_title = gr.Label()  # Component to display the article title
        input_text = gr.Textbox(lines=10, label="Input Text")
        output_text = gr.Textbox(label="Summary")

    load_dataset_article_button = gr.Button("Load Random Article from Dataset")
    load_news_article_button = gr.Button("Load News Article")

    load_dataset_article_button.click(fn=load_article, inputs=[], outputs=input_text)
    load_news_article_button.click(fn=get_news_article, inputs=[], outputs=[input_text, article_title])

    summarize_button = gr.Button("Summarize")
    summarize_button.click(fn=summarize, inputs=input_text, outputs=output_text)

demo.launch()