Any chance to have the dependencies needed for Chromium installed on the server (cloud version)?

Discussions
1

Hi!

I’m trying Activepieces out to see if it could help me with automatic testing of an SPA.

I’m thinking of using Chromium together with Playwright, Puppeteer or Nightmare in a Code piece.

Now, after installing Chromium using NPM I get the following error when trying to open a website with Chromium:

error while loading shared libraries: libnss3.so: cannot open shared object file: No such file or directory

Any change of having the dependencies needed for chromium installed on the Activepieces server?

I think this would open up a bunch of interesting uses of Activepieces: Automated UI testing, Browser automation and SPA web scraping.

All the best,
Frazze

2

Hi @Frazze I tried adding a code step and added chromium as an npm package and ran a simple code using it and it worked, maybe this problem with the many updates we have rolled out since this post, apologies for the delayed response.

import * as c from 'chromium'
import {execFile} from 'child_process';
export const code = async (inputs) => {
    execFile(c.path, ['https://google.com'], err => {
	console.log('Hello Google!');
});
  return c;
};
3

Could you please tell us about what code you were trying to run ?

4

Hi @Abdul !

Thanks for reaching out.

I’m trying to run the following code:

import * as c from 'chromium';
import puppeteer from 'puppeteer';
export const code = async (inputs) => {
    const browserTest3 = await puppeteer.launch({ executablePath: c.path });
}

This results in the following response:

"Failed to launch the browser process!
(…) Error while loading shared libraries: libnss3.so: cannot open shared object file: No such file or directory.

I started trying with the more basic:

const browserTest1 = await puppeteer.launch();

and

const browserTest2 = await puppeteer.launch({headless: "new"});

But those two calls fails, saying it cannot find Chrome.

Thanks for helping me out.

All the best,
Johan

5

@Frazze

I was playing around this but got it somehow get past the previous issue you’re having but it’s complaining about error while loading shared libraries: libnss3.so: cannot open shared object file: No such file or directory;

Here’s the relevant code when I launch using Playwright.

import { chromium } from 'playwright';
import chromiumPath from 'chromium';

export const code = async (inputs) => {
    console.log(chromiumPath.path);
  const browser = await chromium.launch({ executablePath: chromiumPath.path });

...

Same error when I launch using Puppeteer:

import puppeteer from 'puppeteer';
import chromiumPath from 'chromium';

export const code = async (inputs) => {
  const browser = await puppeteer.launch({
    executablePath: chromiumPath.path
  });

...

@Abdul is it possible to install this library on the server? libnss3

6

Hi @menacestudio !

Thanks for posting your findings - for a while I thought I was the only one having issues with this.

Hope that @Abdul and crew can find a solution soon - would be really cool to get Playwright / Puppeteer to run!

7

Chiming in to say that I am having the same issue when using puppeteer – Error: Could not find Chrome…

Any update on this? thanks!

1 Like
8

Hi,

I haven’t updated the ticket here. We spent a great time figuring this out, but the issue is that Chrome/Firefox requires certain permissions that are dangerous to give. also we have two sand-boxing modes to make it easy to be self hostable, we need a specific configuration for each environment.

TL;DR: We are still looking for a way to introduce it without breaking security best practices. So, we postponed it a little, as this is not straightforward.

3 Likes
9

I’m stuck on this also with Puppeteer, would love to use it on activepieces, it will help me a lot to automate things :slight_smile:

1 Like
10

Ok I found a workaround : connect to a remote chrome that you host on a docker server.
GitHub - browserless/browserless: Deploy headless browsers in Docker. Run on our cloud or bring your own. Free for non-commercial uses.

I hosted it on a 6$ digitalocean droplet and it works perfectly
Then to instanciate your browser with puppetter, just use this line :

const browser = await puppeteer.connect({ browserWSEndpoint: 'ws://YOUR_SERVER_IP:3000/chrome' });
2 Likes
11

Hi @yann120,

That’s a really clever solution. Thanks for sharing!

12

Hi @yann120

Your solution is elegant!

Do you mind post it on Tutorials, It will be great addition there!

Thank you,
Mo.

13

For anyone landing here now: you no longer need Chromium inside a Code step. Activepieces has a Browserless piece that does what yann120 described without writing the connect code yourself.

Run Puppeteer Function takes an ES module whose default export receives { page, context } and returns { data, type }, so most existing Puppeteer code moves over with small changes. For the common jobs there are ready actions: Capture Screenshot (PNG, JPEG or WebP, full page, viewport size, wait for a selector, delay) and Generate PDF (from a URL or from a raw HTML string, with paper format, margins and print background).

The connection has a region dropdown for the hosted service plus a Custom Endpoint option, so a self hosted Browserless container like the one on a small droplet works too. Put its base URL there and keep your token.

Each call is a fresh browser session, so pass any values your test needs (login, target URL) through Context rather than expecting state to carry over between steps.

(Disclosure: I’m connected to thirds.ai.)