[Bug] Darkreader breaks browser detection routines on Firefox

Bug Report

Bug Description

Dark Reader creates the objet window.chrome on Firefox, which breaks browser detection using “duck-typing” methods.

I’ve recently included Dark Reader in some webpack projects (using npm install darkreader) but now Firefox is seen as a Chrome/Chromium browser when checking for window.chrome

I know I could adjust my browser detection routines to use feature detection (or user agent detection), but I still think it’s a bit overkill to unconditionally create the window.chrome object on non-Chrome/Chromium browsers just for internal use. Other JavaScript libraries don’t do that, it pollutes the window object.

The culprit is following code: https://github.com/darkreader/darkreader/blob/0edbc3f2e3f84ed37a0368f1c5445d3157a9c8f0/src/api/chrome.ts#L7

PS: Dark Reader is a great library, thank you for your great work!

Website Address

https://stackoverflow.com/a/9851769/2332350

Steps To Reproduce

Use the code from above stackoverflow link in Firefox in a webpack project with Dark Reader installed:

var isChrome = !!window.chrome && (!!window.chrome.webstore || !!window.chrome.runtime);

Expected Behavior

Previous test should return false on Firefox, but Dark Reader breaks the test by creating a “fake” window.chrome object.

Actual Behavior

Previous test returns true on Firefox.

Screenshots

System Information

• OS: Windows 10
• Browser: Firefox 94.0.1
• Dark Reader Version: 4.9.40

Additional Context

[Gusted]

At least, would it be possible to put some comment/warning in the “Using Dark Reader for a website” section? Users should be aware that an object window.chrome could be created after the inclusion of the Dark Reader library and that window.chrome is then not reliable anymore for browser detection routines.

Good suggestion!

[tbl0605]

The name of the file chrome.ts is because we’re using the chrome. namespace - which is btw supported on Firefox.

@Gusted, thank you for the explanation!

As a lot of people(giving experience with current API) will not understand it properly even if it’s clear with warnings and errors.

At least, would it be possible to put some comment/warning in the “Using Dark Reader for a website” section? Users should be aware that an object window.chrome could be created after the inclusion of the Dark Reader library and that window.chrome is then not reliable anymore for browser detection routines.

After that, feel free to close my bug report 😉