An introduction to web scraping with playwright, nodejs and firebase.
Prerequisites
If you want to follow along this tutorial, you'll need the following:
- Basic Knowledge of Firebase and a Firebase account https://firebase.google.com/
- Basic knowledge of javascript
- A coding Editor Vscode Preferred
- API Development/Debugging tool.
What is web scrapping?
Web scrapping refers to the extraction of data from a website. This information
is collected and exported into a format (i.e csv) that is more useful to the
user.
What is a Headless Browser?
You might have heard of the term Headless Browser but still don't know what it
means. You don't have to worry because the Internet's got our back 🙂
Headless browsers provide automated control of a web page in an environment
similar to popular web browsers, but are executed via a command-line interface
or using network communication.
wikipedia.
Here are few most popular Headless Browsers 👇
Puppeteer: Puppeteer is a
Node library which provides a high-level API to control headless Chrome or
Chromium over the DevTools Protocol. It can also be configured to use full
(non-headless) Chrome or Chromium.
Playwright: Playwright is a Node library developed by
microsoft to automate Chromium, Firefox and WebKit with a single API. Playwright
is built to enable cross-browser web automation that is ever-green, capable,
reliable and fast.
Initial Setup 🚀
Lets start off by initializing firebase cloud functions for javascript:
firebase init functions
cd functions
npm install playwright
This installs Playwright and browser binaries for Chromium, Firefox and WebKit.
Once installed, you can require Playwright in a Node.js script and automate web
browser interactions.
Now lets create our Instagram Scraper
Instagram on the web uses React, which means we won’t see any dynamic content
util the page is fully loaded. Playwright is available in the Clould Functions
runtime, allowing you to spin up a Chrome/Firefox/Webkit browser on your server.
It will render JavaScript and handle events just like the browser you’re using
right now.
First, the function logs into a real instagram account. The page.type method
will find the cooresponding DOM element and type characters into it. Once logged
in, we navigate to a specific username and wait for the img tags to render on
the screen, then scrape the src attribute from them.
const functions = require('firebase-functions');
const playwright = require('playwright');
exports.scrapeImages = functions.https.onRequest(async (req, res) => {
// Randomly select a browser
// You can also specify a single browser that you prefer
for (const browserType of ['firefox', 'chromium', 'webkit']) {
console.log(browserType); // To know the chosen one 😁
const browser = await playwright[browserType].launch();
const context = await browser.newContext();
const page = await context.newPage();
await page.goto('https://www.instagram.com/accounts/login/');
await page.waitForSelector('[type=submit]', {
state: 'visible',
});
// You can also take screenshots of pages
await page.screenshot({
path: `ig-sign-in.png`,
});
await page.type('[name=username]', '<your-username>'); // ->
await page.type('[type="password"]', '<your-password>'); // ->
await page.click('[type=submit]');
await page.waitForSelector('[placeholder=Search]', { state: 'visible' });
await page.goto(`https://www.instagram.com/<your-username>`); // ->
await page.waitForSelector('img', {
state: 'visible',
});
await page.screenshot({ path: `profile.png` });
// Execute code in the DOM
const data = await page.evaluate(() => {
const images = document.querySelectorAll('img');
const urls = Array.from(images).map((v) => v.src);
return urls;
});
await browser.close();
console.log(data);
// Return the data in form of json
return res.status(200).json(data);
}
});
Replace highlighted fields with valid credentials.
Now we need to test out our API and to do that we're going to need an API
debugging tool and I'd recommend Insomnia because it's
the best API tool i've ever used and it has tons of features. You can also use
tools like Postman.
start the dev server by running:
npm run serve
Open Insomnia and send a post request to the generated url from your firebase
dev server.
Example http://localhost:5001/playwright-faecb/us-central1/scrapeImages
In the above image, you can see the response JSON data colored with yellow on
the right side. It is an array of urls that points to individual images on your
Instagram page.
Let's reflect back step by step to better understand what we're doing.
Playwright provides us with three different browser types so why not try three
of them at the same time.
We loop through the three browser types
['chromium', 'firefox', 'webkit'].launch()
We launch a new browser with the launch method.newContext()
Creates a new browser context. A
Browser context
provides us with most operations like creating a new tab in the browser.goto
method navigates to a specified URL path.waitForSelector()
waits for an element to either be or not be present in the browser.screeshot()
takes a screenshot of the current page.text()
provides us with the ability to fill out form fields.click()
allows us to click on an element or page in the dom.evaluate().
You can do a lot in the callback of this function; However, we're just
returning thesrcvalue of each image in the callback.browser.close()
destroys our current browser as you already know.
You can do whatever you want with the data (Download the images) but in our case
we are just logging it out on the console and returning it as a JSON string with
a 200 response to see what it looks like.
Conclusion
Most times Bad dudes applies this technique to Illegally extract data from a
website and I'm pretty sure that the person reading this is not one of them.
Always remember to Use your code for good 🙂
Complete code: https://github.com/dnature/playwright-example
I hope you find this helpful.
Happy codding 💻 🙂.
You might as well check my blog for other tech related articles. https://divinehycenth.com/blog
Top comments (10)
Worth noting that the documentation has alternatives to hard-coding the wait times (page.waitForTimeout). Some commands like
fillandclickhave auto waits built-in. Or you can explicitly wait for an object to appear in the DOM.https://playwright.dev/path=docs%2Fcore-concepts.md&q=auto-waiting#version=master
Hi Aleccc,
I've updated the article to use this approach as recommended in the docs. Thank you for pointing that out :)
Hi Divine,
Just note that currently this approach only works on localhost with firebase serve, it would fail when you deploy it to the Cloud. In my observation, Firebase can't figure out where the binary browsers used for scraping are stored, therefore can't initialize the browsers. I am still finding a way to modify this behaviour. Do you have any ideas?
Hi Johnny,
I apologize for my late response.
What you said is true and I haven't figured out a way to make it run on firebase cloud. I will be glad to know if you've figured that out :).
Thank you for your patience.
Hi, any o you find a solution for this bug?
Nice writetup! If someone decides to launch this script on datacenter, I would definitely recommend using some clean (preferably residential) proxies to avoid your accounts being flagged and save your cookies to re-use them later (this was actually mentioned here in comments).
I've recently published a simple tutorial on Instagram scraping and discovering micro-influencers via Node.js and MySQL.
How to scrape Instagram followers with Node.js, put results to MySQL, and discover micro-influencers
restyler ・ Oct 3 ・ 9 min read
Good luck!
Do u know, how to save login session after browser close and want to scraping again and again ?
go to your project directory
using the terminal,run the command below,
npx playwright open --save-storage websitename.json
a browser will open,now navigate to the website and sign-in/solve captcha,
then close browser. You will notice that a file "websitename.json" has been created.
now in playwright,set you browser context using this code below
const context = await browser.newContext({
storageState: "websitename.json"
});
you are now automatically logged in. :)
I haven't tried that and maybe it's possible but i'm sure it's not going to work if you randomly spin up browsers using Playwright.
Let me know if you are able to do that :)
Why not Functions Framework? Is the use of Firebase tools really necessary?