import requests
import torch
import gradio as gr
from transformers import AutoModelForCausalLM, AutoTokenizer
from datetime import datetime

# GPT-2 setup
device = torch.device("cuda" if torch.cuda.is_available() else "cpu")
model_name = "gpt2"
tokenizer = AutoTokenizer.from_pretrained(model_name)
model = AutoModelForCausalLM.from_pretrained(model_name).to(device)

# NewsAPI Setup (Replace with your own API key)
news_api_key = "35cbd14c45184a109fc2bbb5fff7fb1b"  # Replace with your NewsAPI key

def fetch_trending_topics(search_term="artificial intelligence OR machine learning", page=1, page_size=9):
    try:
        # Fetch AI and Machine Learning related news from NewsAPI with search term
        url = f"https://newsapi.org/v2/everything?q={search_term}&sortBy=publishedAt&pageSize={page_size + 5}&page={page}&language=en&apiKey={news_api_key}"  # Fetch extra to avoid duplicates
        response = requests.get(url)
        data = response.json()
        
        # Check for valid response
        if response.status_code == 200 and "articles" in data:
            # Collect articles without duplicates
            trending_topics = []
            seen_titles = set()
            for article in data["articles"]:
                title = article["title"]
                if title not in seen_titles:  # Avoid duplicate titles
                    seen_titles.add(title)
                    trending_topics.append({
                        "title": title,
                        "description": article["description"] if article["description"] else "No description available.",
                        "url": article["url"],
                        "publishedAt": article["publishedAt"],
                    })

            if not trending_topics:
                return [{"title": "No news available", "description": "", "url": "", "publishedAt": ""}]
                
            return trending_topics
        else:
            print(f"Error: {data.get('message', 'No articles found')}")
            return [{"title": "No news available", "description": "", "url": "", "publishedAt": ""}]
    except Exception as e:
        print(f"Error fetching news: {e}")
        return [{"title": "Error fetching news", "description": "", "url": "", "publishedAt": ""}]

# Analyze the trending topic using GPT-2
def generate_analysis(trending_topic):
    input_text = f"Provide a concise analysis about the following topic: '{trending_topic['title']}'. Please summarize its significance in the AI and Machine Learning field."

    # Tokenize and generate text with a max limit on tokens
    inputs = tokenizer(input_text, return_tensors="pt").to(device)
    outputs = model.generate(**inputs, max_length=80, num_return_sequences=1, do_sample=True, top_k=50, top_p=0.95)

    analysis = tokenizer.decode(outputs[0], skip_special_tokens=True)

    return analysis

# Combine both functions for Gradio
def analyze_trends(page=1, page_size=9):
    search_term = "artificial intelligence OR machine learning"  # Fixed search term
    trending_topics = fetch_trending_topics(search_term=search_term, page=page, page_size=page_size)
    topic_analysis = []
    
    for topic in trending_topics:
        if topic["title"] not in ["Error fetching news", "No news available"]:
            analysis = generate_analysis(topic)
            topic_analysis.append({
                "title": topic["title"],
                "description": topic["description"],
                "analysis": analysis,
                "url": topic["url"],
                "publishedAt": topic["publishedAt"],
            })
        else:
            topic_analysis.append({
                "title": topic["title"],
                "description": topic["description"],
                "analysis": "Unable to retrieve or analyze data.",
                "url": topic["url"],
                "publishedAt": topic["publishedAt"],
            })
    
    # Limit the results to the specified page size
    return topic_analysis[:page_size]  # Ensure only the specified number of articles are returned

# Gradio UI with 3 Columns Layout for Displaying News
def display_news_cards(page=1, page_size=9):
    analysis_results = analyze_trends(page=page, page_size=page_size)
    current_date = datetime.now().strftime("%d-%m-%Y")  # Format: DD-MM-YYYY

    display = f"### **AI & Machine Learning News for {current_date}**\n\n"
    
    # Create a 3-column layout
    display += "<div style='display:flex; flex-wrap:wrap; justify-content:space-between;'>"
    for news_item in analysis_results:
        # Each news box in a flex box with equal width
        display += f"""
        <div style='flex: 1 1 30%; border:1px solid black; margin:10px; padding:10px; box-sizing:border-box;'>
        <b>{news_item['title']}</b><br/>
        <i>{news_item['publishedAt']}</i><br/><br/>
        {news_item['description']}<br/><br/>
        <a href='{news_item['url']}' target='_blank'>Read more</a><br/><br/>
        <b>Analysis:</b> {news_item['analysis']}<br/><br/>
        </div>
        """
    display += "</div>"

    return display

# Gradio UI with Header, Search Option, and Submit Button
def gradio_interface():
    with gr.Blocks() as demo:
        # Header with background colour
        gr.Markdown("""<h1 style='text-align:center; color:white; background-color:#007BFF; padding:20px; border-radius:10px;'>AI & Machine Learning News Analyzer</h1>""", elem_id="header")

        # Fixed search term displayed to the user
        gr.Markdown("<p style='text-align:center;'>Search term: <b>artificial intelligence OR machine learning</b></p>")

        # Sliders for page number and news per page
        page = gr.Slider(minimum=1, maximum=5, step=1, label="Page Number", value=1)
        page_size = gr.Slider(minimum=6, maximum=15, step=3, label="News per Page", value=9)

        # Button to fetch and analyze news
        analyze_button = gr.Button("Submit")

        # Output area for displaying the news
        news_output = gr.HTML()

        # Link the button click to the display function
        analyze_button.click(display_news_cards, inputs=[page, page_size], outputs=news_output)

    return demo

# Launch the Gradio UI
if __name__ == "__main__":
    gradio_interface().launch()