Agents

import requests
from bs4 import BeautifulSoup
import pandas as pd
from litai import LLM

llm = LLM('openai/gpt-4o')

def scrape_yahoo_finance_gainers(url):
    # these headers make our request look like a regular browser visit,
    # which helps prevent websites from blocking our scraper.
    headers = {
        'User-Agent': 'Mozilla/5.0 (Windows NT 10.0; Win64; x64) AppleWebKit/537.36 (KHTML, like Gecko) Chrome/125.0.6422.112 Safari/537.36',
        'Accept': 'text/html,application/xhtml+xml,application/xml;q=0.9,image/webp,image/apng,*/*;q=0.8',
        'Accept-Language': 'en-US,en;q=0.9',
        'Accept-Encoding': 'gzip, deflate, br',
        'Connection': 'keep-alive',
        'Referer': 'https://www.google.com'
    }

    # we send a request to the yahoo finance page to get its content.
    response = requests.get(url, headers=headers)
    # this line checks if the website responded successfully; if not, it'll tell us there was an issue.
    response.raise_for_status()

    # beautifulsoup helps us parse the html content so we can easily find specific elements like tables.
    soup = BeautifulSoup(response.text, 'html.parser')

    # we're looking for the main table that contains the stock data.
    table = soup.find('table')

    if not table:
        print("error: could not find table on the page.")
        return pd.DataFrame()

    # we extract the column headers from the table.
    headers = [header.text.strip() for header in table.find_all('th')]
    data = []

    # then we go through each row (skipping the header) and pull out the data for each stock.
    for row in table.find_all('tr')[1:]:
        cols = row.find_all('td')
        cols = [ele.text.strip() for ele in cols]
        data.append(cols)

    # finally, we put all this data into a pandas dataframe, making it structured and easy to work with.
    df = pd.DataFrame(data, columns=headers)

    # now for the fun part: using the llm!
    # we convert our dataframe to a csv string so the llm can easily understand the tabular data.
    csv_string = df.to_csv(index=False)

    # this is the prompt we send to the llm. we're asking it to act like a finance expert
    # and explain its choices simply, avoiding confusing jargon.
    question_prompt = """
    pick the top 2 stocks that had the most real movement today — based on strong price gain and healthy volume trends.
    ignore stocks where the percent gain is big just because the price is tiny, or where today's volume is way higher than usual (a one-day spike).
    explain clearly why each one looks like it has real investor momentum to a non investment expert.
    """

    # here's where litai sends your data and question to the llm and gets back an answer.
    # it's like having a very smart, on-demand analyst!
    answer = llm.chat(f'{question_prompt} here is the data: <data>{csv_string}</data>')
    print(answer)
    return df

if __name__ == "__main__":
    url = "https://finance.yahoo.com/markets/stocks/gainers"
    print(f"scraping data from: {url}")
    stock_data_df = scrape_yahoo_finance_gainers(url)

    if not stock_data_df.empty:
        output_filename = "yahoo_finance_gainers.csv"
        # we also save the raw scraped data to a csv file just in case you want to look at it yourself later.
        stock_data_df.to_csv(output_filename, index=False)
        print(f"data successfully saved to {output_filename}")
    else:
        print("no data to save.")