You might not know, but Puppeteer’s team is making good progress on a Puppeteer version for Firefox. That’s a big deal. If we have trouble sometimes trying to get a site look the same in all browsers, imagine trying to use the same debugging protocol. They even had to fork Firefox’s debugging protocol to make it “Puppeteer friendly”.
Implementation on Puppeteer
Architecturally wise, Firefox implementation in Puppeteer is quite simple (as it’s supposed to be):
- They created an
- Replicated there all the classes they have in the main
- Load the puppeteer library dynamically using
const puppeteer = require(puppeteerPath);
And voilá! Duck typing does the rest.
Implementation on Puppeteer-Sharp
Things won’t be so easy in Puppeteer-Sharp. I’m not saying that it will be hard. But the challenge is how to include a layer of abstraction without breaking our users.
Introducing Puppeteer-Sharp Abstractions
Puppeteer-Sharp Abstractions will be a set of interfaces, such as IBrowser, IPage, etc., as a layer of abstraction. So you can use the same code to run against Chromium and Firefox.
We might have some common classes here, but the final goal is having a set of interfaces.
Abstractions in Puppeteer-Sharp
The idea is implementing in Puppeteer-Sharp all the interfaces we created in the Abstractions project. To avoid breaking the users’ code, we’ll do two things:
1. Implement explicit interfaces
We are going to implement explicit interfaces. This means that, for instance, the Browser class will have methods expecting and returning concrete classes, and methods expecting and returning interfaces.
So if you have an instance of a Browser class you will see a:
public async Task<Page> PagesAsync()
And if you have an instance of an IBrowser interface you will see a:
public async Task<IPage> PagesAsync()
2. Mark concrete methods as obsolete
Task<Page> PagesAsync() will be marked as obsolete. We will be keeping those methods for two versions, giving users some time to replace their concrete declarations with interfaces.
I’m not saying that we are going to start with Firefox right now. I want to wait for Puppeteer’s code to get more stable so we can begin moving their code to C#. But implementing these interfaces now will lay the groundwork to start working on this great new package.
Don’t stop coding!