DEV Community

Cover image for How to Scrape Login-Protected Websites with Selenium (Step by Step Guide)
Sacha Dumay
Sacha Dumay

Posted on

How to Scrape Login-Protected Websites with Selenium (Step by Step Guide)

My Steps to Scrape a Password-Protected Website:

  1. Capture the HTML form elements: username ID, password ID, and login button class
  2. - Use a tool like requests or Selenium to automate the login: fill username, wait, fill password, wait, click login
  3. - Store session cookies for authentication
  4. - Continue scraping the authenticated pages

Disclaimer: I’ve built an API for this specific use case at https://www.scrapewebapp.com/. So if you want to just get it done fast, use it, otherwise read on.

Let’s use this example: let’s say I want to scrape my own API key from my account at https://www.scrapewebapp.com/. It is on this page: https://app.scrapewebapp.com/account/api_key

1. The Login Page

First, you need to find the login page. Most websites will give you a redirect 303 if you try to access a page behind login, so if you try to scrape directly https://app.scrapewebapp.com/account/api_key, you will automatically get the login page https://app.scrapewebapp.com/login. So it is a good way to automate finding the login page if not provided already.

Ok, now that we have the login page, we need to find the place to add username or email as well as password and the actual sign-in button. The best way is to create a simple script that finds the ID of the inputs using their type “email”, “username”, “password” and finds the button with the type “submit”. I made a code for you below:

from bs4 import BeautifulSoup


def extract_login_form(html_content: str):
    """
    Extracts the login form elements from the given HTML content and returns their CSS selectors.
    """
    soup = BeautifulSoup(html_content, "html.parser")

    # Finding the username/email field
    username_email = (
        soup.find("input", {"type": "email"})
        or soup.find("input", {"name": "username"})
        or soup.find("input", {"type": "text"})
    )  # Fallback to input type text if no email type is found

    # Finding the password field
    password = soup.find("input", {"type": "password"})

    # Finding the login button
    # Searching for buttons/input of type submit closest to the password or username field
    login_button = None

    # First try to find a submit button within the same form
    if password:
        form = password.find_parent("form")
        if form:
            login_button = form.find("button", {"type": "submit"}) or form.find(
                "input", {"type": "submit"}
            )
    # If no button is found in the form, fall back to finding any submit button
    if not login_button:
        login_button = soup.find("button", {"type": "submit"}) or soup.find(
            "input", {"type": "submit"}
        )

    # Extracting CSS selectors
    def generate_css_selector(element, element_type):
        if "id" in element.attrs:
            return f"#{element['id']}"
        elif "type" in element.attrs:
            return f"{element_type}[type='{element['type']}']"
        else:
            return element_type

    # Generate CSS selectors with the updated logic
    username_email_css_selector = None
    if username_email:
        username_email_css_selector = generate_css_selector(username_email, "input")

    password_css_selector = None
    if password:
        password_css_selector = generate_css_selector(password, "input")

    login_button_css_selector = None
    if login_button:
        login_button_css_selector = generate_css_selector(
            login_button, "button" if login_button.name == "button" else "input"
        )

    return username_email_css_selector, password_css_selector, login_button_css_selector


def main(html_content: str):
    # Call the extract_login_form function and return its result
    return extract_login_form(html_content)
Enter fullscreen mode Exit fullscreen mode

2. Using Selenium to Actually Log In

Now you need to create a selenium webdriver. We will use chrome headless to run it with Python. This is how to install it:

# Install selenium and chromium

!pip install selenium
!apt-get update 
!apt install chromium-chromedriver

!cp /usr/lib/chromium-browser/chromedriver /usr/bin
import sys
sys.path.insert(0,'/usr/lib/chromium-browser/chromedriver')
Enter fullscreen mode Exit fullscreen mode

Then actually log into our website and save the cookies. We will save all cookies, but you could only save the auth cookies if you wanted.

# Imports
from selenium import webdriver
from selenium.webdriver.common.by import By
import requests
import time

# Set up Chrome options
chrome_options = webdriver.ChromeOptions()
chrome_options.add_argument('--headless')
chrome_options.add_argument('--no-sandbox')
chrome_options.add_argument('--disable-dev-shm-usage')

# Initialize the WebDriver
driver = webdriver.Chrome(options=chrome_options)

try:
    # Open the login page
    driver.get("https://app.scrapewebapp.com/login")

    # Find the email input field by ID and input your email
    email_input = driver.find_element(By.ID, "email")
    email_input.send_keys("******@gmail.com")

    # Find the password input field by ID and input your password
    password_input = driver.find_element(By.ID, "password")
    password_input.send_keys("*******")

    # Find the login button and submit the form
    login_button = driver.find_element(By.CSS_SELECTOR, "button[type='submit']")
    login_button.click()

    # Wait for the login process to complete
    time.sleep(5)  # Adjust this depending on your site's response time


finally:
    # Close the browser
    driver.quit()
Enter fullscreen mode Exit fullscreen mode

3. Store Cookies

It is as simple as saving them into a dictionary from the driver.getcookies() function.

def save_cookies(driver):
    """Save cookies from the Selenium WebDriver into a dictionary."""
    cookies = driver.get_cookies()
    cookie_dict = {}
    for cookie in cookies:
        cookie_dict[cookie['name']] = cookie['value']
    return cookie_dict
Enter fullscreen mode Exit fullscreen mode

Save the cookies from the WebDriver

cookies = save_cookies(driver)

4. Get Data from Our Logged-in Session

In this part, we will use the simple library requests, but you could keep using selenium too.

Now we want to get the actual API from this page: https://app.scrapewebapp.com/account/api_key.

So we create a session from the requests library and add each cookie into it. Then request the URL and print the response text.

def scrape_api_key(cookies):
    """Use cookies to scrape the /account/api_key page."""
    url = 'https://app.scrapewebapp.com/account/api_key'

    # Set up the session to persist cookies
    session = requests.Session()

    # Add cookies from Selenium to the requests session
    for name, value in cookies.items():
        session.cookies.set(name, value)

    # Make the request to the /account/api_key page
    response = session.get(url)

    # Check if the request is successful
    if response.status_code == 200:
        print("API Key page content:")
        print(response.text)  # Print the page content (could contain the API key)
    else:
        print(f"Failed to retrieve API key page, status code: {response.status_code}")
Enter fullscreen mode Exit fullscreen mode

5. Get the Actual Data You Want (BONUS)

We got the page text we wanted, but there is a lot of data that we do not care about. We just want the api_key.

The best and easiest way to do that is to use AI like ChatGPT (GPT4o model).

Prompt the model like this: “You are an expert scraper and you will extract only the information asked from the context. I need the value of my api-key from {context}”

from openai import OpenAI

# Initialize OpenAI client
client = OpenAI(api_key="sk-proj-***********")

def extract_api_key_using_ai(response_text):
    """Use OpenAI's GPT model to extract the API key."""
    prompt = f"""
    You are an expert scraper, and you will extract only the information asked from the context. 
    I need the value of my api-key from the following context:

    {response_text}
    """

    try:
        # Use OpenAI client to create a chat completion
        chat_completion = client.chat.completions.create(
            messages=[
                {
                    "role": "user",
                    "content": prompt,
                }
            ],
            model="gpt-4o",
        )

        # Extract the response from the AI
        extracted_api_key = chat_completion.choices[0].message.content
        return extracted_api_key

    except Exception as e:
        print(f"An error occurred with OpenAI API: {e}")
        return None
Enter fullscreen mode Exit fullscreen mode

If you want all that in a simple and reliable API, please give a try to my new product https://www.scrapewebapp.com/

If you like this post, please give me claps and follow me. It does help a lot!

Top comments (1)

Collapse
 
kate_park_8025b47632ec265 profile image
Kate Park

Thanks very useful for my own scraping! 👍