app.py

import gradio as gr
import pandas as pd
import requests
import datetime
import tempfile
import os
import matplotlib.pyplot as plt
from transformers import pipeline

# Initialize Models
summarizer = pipeline("summarization", model="facebook/bart-large-cnn")
chat_model = pipeline("text-generation", model="google/flan-t5-large", max_length=256)

# API Key
POLYGON_API_KEY = os.getenv("POLYGON_API_KEY")

# Sector Averages
sector_averages = {
    "Technology": {"P/E Ratio": 25, "P/S Ratio": 5, "P/B Ratio": 6},
    "Healthcare": {"P/E Ratio": 20, "P/S Ratio": 4, "P/B Ratio": 3},
    "Financials": {"P/E Ratio": 15, "P/S Ratio": 2, "P/B Ratio": 1.5},
    "Energy": {"P/E Ratio": 12, "P/S Ratio": 1.2, "P/B Ratio": 1.3},
}

# Tooltip dictionary
tooltips = {
    "P/E Ratio": "Price/Earnings: Lower can indicate better value.",
    "P/S Ratio": "Price/Sales: Lower can indicate better value relative to sales.",
    "P/B Ratio": "Price/Book: Lower can indicate undervaluation.",
    "PEG Ratio": "Price/Earnings to Growth: Closer to 1 is ideal.",
    "Dividend Yield": "Annual dividend income relative to price."
}

# Helper Functions
def safe_request(url):
    try:
        response = requests.get(url)
        response.raise_for_status()
        return response
    except:
        return None

def get_company_info(symbol):
    url = f"https://api.polygon.io/v3/reference/tickers/{symbol}?apiKey={POLYGON_API_KEY}"
    response = safe_request(url)
    if response:
        data = response.json().get('results', {})
        sector = data.get('market', 'Technology')
        # Dynamic Guess
        if sector.lower() == 'stocks':
            sector = "Technology"
        return {
            'Name': data.get('name', 'N/A'),
            'Industry': data.get('sic_description', 'N/A'),
            'Sector': sector,
            'Market Cap': data.get('market_cap', 0),
            'Total Revenue': data.get('total_employees', 0) * 100000
        }
    return None

def get_current_price(symbol):
    url = f"https://api.polygon.io/v2/aggs/ticker/{symbol}/prev?adjusted=true&apiKey={POLYGON_API_KEY}"
    response = safe_request(url)
    if response:
        return response.json()['results'][0]['c']
    return None

def get_dividends(symbol):
    url = f"https://api.polygon.io/v3/reference/dividends?ticker={symbol}&apiKey={POLYGON_API_KEY}"
    response = safe_request(url)
    if response:
        return response.json()['results'][0].get('cash_amount', 0)
    return 0

def get_historical_prices(symbol):
    end = datetime.date.today()
    start = end - datetime.timedelta(days=365)
    url = f"https://api.polygon.io/v2/aggs/ticker/{symbol}/range/1/day/{start}/{end}?adjusted=true&sort=asc&apiKey={POLYGON_API_KEY}"
    response = safe_request(url)
    if response:
        results = response.json()['results']
        dates = [datetime.datetime.fromtimestamp(r['t']/1000) for r in results]
        prices = [r['c'] for r in results]
        return dates, prices
    return [], []

def calculate_ratios(market_cap, total_revenue, price, dividend_amount, eps=5.0, growth=0.1, book_value=500000000):
    pe = price / eps if eps else 0
    ps = market_cap / total_revenue if total_revenue else 0
    pb = market_cap / book_value if book_value else 0
    peg = pe / (growth * 100) if growth else 0
    div_yield = (dividend_amount / price) * 100 if price else 0
    return {
        'P/E Ratio': pe,
        'P/S Ratio': ps,
        'P/B Ratio': pb,
        'PEG Ratio': peg,
        'Dividend Yield': div_yield
    }

def compare_to_sector(sector, ratios):
    if sector.lower() == 'stocks':
        sector = 'Technology'
    averages = sector_averages.get(sector, {})
    if not averages:
        return pd.DataFrame({"Metric": ["Sector data not available"], "Value": ["N/A"]})

    data = {
        "Ratio": [],
        "Stock Value": [],
        "Sector Average": [],
        "Difference": []
    }

    for key in averages:
        stock_value = ratios.get(key, 0)
        sector_value = averages.get(key, 0)
        diff = stock_value - sector_value

        # Add emoji based on difference
        if diff < 0:
            diff_display = f"{diff:.2f} 🟢"
        elif diff > 0:
            diff_display = f"{diff:.2f} 🔴"
        else:
            diff_display = f"{diff:.2f} ⚪"

        data["Ratio"].append(key)
        data["Stock Value"].append(round(stock_value, 2))
        data["Sector Average"].append(round(sector_value, 2))
        data["Difference"].append(diff_display)

    return pd.DataFrame(data)


def generate_summary(info, ratios):
    recommendation = "Hold"
    if ratios['P/E Ratio'] < 15 and ratios['P/B Ratio'] < 2 and ratios['PEG Ratio'] < 1.0 and ratios['Dividend Yield'] > 2:
        recommendation = "Buy"
    elif ratios['P/E Ratio'] > 30 and ratios['P/B Ratio'] > 5 and ratios['PEG Ratio'] > 2.0:
        recommendation = "Sell"
    report = (
        f"Company Overview:\n"
        f"Name: {info['Name']}\n"
        f"Industry: {info['Industry']}\n"
        f"Sector: {info['Sector']}\n"
        f"Market Cap: ${info['Market Cap']:,.2f}\n\n"
        f"Financial Metrics:\n"
        f"P/E Ratio: {ratios['P/E Ratio']:.2f}\n"
        f"P/S Ratio: {ratios['P/S Ratio']:.2f}\n"
        f"P/B Ratio: {ratios['P/B Ratio']:.2f}\n"
        f"PEG Ratio: {ratios['PEG Ratio']:.2f}\n"
        f"Dividend Yield: {ratios['Dividend Yield']:.2f}%\n\n"
        f"Recommended Investment Action: {recommendation}.\n"
    )
    return summarizer(report, max_length=250, min_length=100, do_sample=False)[0]['summary_text']

def answer_investing_question(question):
    response = chat_model(['generated_text'])
    return response


def stock_research(symbol, eps=5.0, growth=0.1, book=500000000):
    info = get_company_info(symbol)
    price = get_current_price(symbol)
    dividends = get_dividends(symbol)
    dates, prices = get_historical_prices(symbol)

    if not info or not price:
        return "⚠️ Error fetching stock info", None, None, None, None

    ratios = calculate_ratios(info['Market Cap'], info['Total Revenue'], price, dividends, eps, growth, book)
    summary = generate_summary(info, ratios)
    sector_comp = compare_to_sector(info['Sector'], ratios)

    fig, ax = plt.subplots()
    ax.plot(dates, prices)
    ax.set_title(f"{symbol} Historical Price (1Y)")
    ax.set_xlabel("Date")
    ax.set_ylabel("Price ($)")
    ax.grid(True)

    info_table = pd.DataFrame(info.items(), columns=["Metric", "Value"])
    ratios_table = pd.DataFrame(ratios.items(), columns=["Ratio", "Value"])

    return summary, info_table, ratios_table, sector_comp, fig

# --- Gradio UI ---
with gr.Blocks(theme="soft") as iface:
    with gr.Row():
        symbol = gr.Textbox(label="Stock Symbol (e.g., AAPL)")
        eps = gr.Number(label="Assumed EPS", value=5.0)
        growth = gr.Number(label="Assumed Growth Rate", value=0.1)
        book = gr.Number(label="Assumed Book Value", value=500000000)

    with gr.Tabs():
        with gr.Tab("AI Research Summary"):
            output_summary = gr.Textbox()
        with gr.Tab("Company Snapshot"):
            output_info = gr.Dataframe()
        with gr.Tab("Valuation Ratios"):
            output_ratios = gr.Dataframe()
        with gr.Tab("Sector Comparison"):
            output_sector = gr.Dataframe()
        with gr.Tab("Historical Price Chart"):
            output_chart = gr.Plot()
        with gr.Tab("Ask About Investing"):
            user_question = gr.Textbox(label="Ask about investing...")
            answer_box = gr.Textbox()
            ask_button = gr.Button("Get Answer")
            ask_button.click(fn=answer_investing_question, inputs=[user_question], outputs=[answer_box])

    submit_btn = gr.Button("Run Analysis")
    download_btn = gr.Button("Download Report")
    file_output = gr.File()

    submit_btn.click(fn=stock_research, inputs=[symbol, eps, growth, book],
                     outputs=[output_summary, output_info, output_ratios, output_sector, output_chart])

    # Sector Comparison Color Highlight
    def style_sector(df):
        def highlight(val):
            if isinstance(val, (int, float)):
                if val < 0:
                    return 'color: green'
                elif val > 0:
                    return 'color: red'
            return ''
        return df.style.applymap(highlight, subset=['Difference'])
    output_sector.style_fn = style_sector

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