import gradio as gr
import requests
from bs4 import BeautifulSoup
import pandas as pd
from transformers import pipeline
import plotly.express as px
from datetime import datetime, timedelta

# Sentiment Analysis Model
sentiment_model = pipeline(model="finiteautomata/bertweet-base-sentiment-analysis")

# Function to encode special characters in the search query
def encode_special_characters(text):
    encoded_text = ''
    special_characters = {'&': '%26', '=': '%3D', '+': '%2B', ' ': '%20'}
    for char in text.lower():
        encoded_text += special_characters.get(char, char)
    return encoded_text

# Function to fetch news articles
def fetch_news(query, num_articles=10):
    encoded_query = encode_special_characters(query)
    url = f"https://news.google.com/search?q={encoded_query}&hl=en-US&gl=in&ceid=US%3Aen&num={num_articles}"
    
    try:
        response = requests.get(url)
        response.raise_for_status()
    except requests.RequestException as e:
        print(f"Error fetching news: {e}")
        return pd.DataFrame()
    
    soup = BeautifulSoup(response.text, 'html.parser')
    articles = soup.find_all('article')
    
    news_data = []
    for article in articles[:num_articles]:
        link = article.find('a')['href'].replace("./articles/", "https://news.google.com/articles/")
        text_parts = article.get_text(separator='\n').split('\n')
        
        news_data.append({
            'Title': text_parts[2] if len(text_parts) > 2 else 'Missing',
            'Source': text_parts[0] if len(text_parts) > 0 else 'Missing',
            'Time': text_parts[3] if len(text_parts) > 3 else 'Missing',
            'Author': text_parts[4].split('By ')[-1] if len(text_parts) > 4 else 'Missing',
            'Link': link
        })
    
    return pd.DataFrame(news_data)

# Function to perform sentiment analysis
def analyze_sentiment(text):
    result = sentiment_model(text)[0]
    return result['label'], result['score']

# Main function to process news and perform analysis
def news_and_analysis(query):
    # Fetch news
    news_df = fetch_news(query)
    
    if news_df.empty:
        return "No news articles found.", None
    
    # Perform sentiment analysis
    news_df['Sentiment'], news_df['Sentiment_Score'] = zip(*news_df['Title'].apply(analyze_sentiment))
    
    # Create sentiment plot
    sentiment_fig = px.bar(
        news_df,
        x='Time',
        y='Sentiment_Score',
        color='Sentiment',
        color_discrete_map={'positive': 'green', 'neutral': 'gray', 'negative': 'red'},
        title='News Sentiment Over Time',
        labels={'Time': 'Publication Time', 'Sentiment_Score': 'Sentiment Score'}
    )
    
    return news_df, sentiment_fig

# Gradio interface
with gr.Blocks() as demo:
    gr.Markdown(
        """
        # Financial News Sentiment Analysis
        
        Analyze the sentiment of news articles related to financial topics or companies. 
        Enter a topic or company name to get started.
        """
    )
    
    with gr.Row():
        with gr.Column():
            topic = gr.Textbox(label="Enter a financial topic or company name", placeholder="e.g., Apple Inc.")
            analyze_btn = gr.Button(value="Analyze")
        
        with gr.Column():
            gr.Empty()
            gr.Empty()  # Using Empty objects to maintain spacing
            gr.Empty()
            news_output = gr.DataFrame(label="News and Sentiment Analysis")
            sentiment_plot = gr.Plot(label="Sentiment Analysis")
    
    analyze_btn.click(
        news_and_analysis,
        inputs=[topic],
        outputs=[news_output, sentiment_plot]
    )

if __name__ == "__main__":
    demo.launch()