Differences to undetected-chromedriver?

[alexreg]

Hey. This looks like a great project and one you've put a lot of work into... As a current user of undetected-chromedriver, I was just wondering if you might be able to summarise what pluses/minuses this library has vs undetected-chromedriver when it comes to detection by anti-bot mechanisms. Obviously, your library provides other features when it comes to device profiles and a few other things, though I'm mainly interested in the anti-detection capabilities for now.

[kaliiiiiiiiii]

Undetectability (headfull)

The most of it is similar like in undetected-chromediver (stuff in /scripts/undetected.py)

Profiling//Fingerprinting

A minus of undetected-chromedriver is, that it's fingerprint depends on the Operating-System, which leads to:

• inconsistency on different platforms
• websites might analyze and detect, when a fingerprint is way to often used, since they usually are unique (see amiunique.org)

That's where Selenium-Profiles comes into play. It makes specifying the fingerprint possible (==> emulation)

Proxies

To make it as easy as possible, Selenium-Profiles allready includes support for authentificated proxies. This is quite important, since residential proxy-providers like asocks usually require a password.

Headless

Headless would be usefull for reducing computing-resources, but is quite tricky:

• --headless argument leaves out a lot of javascript objects ==> detection
• --headless=chrome seems to be better, but still isn't traceless.

solution

• overwrite the missing javascript objects. That would also improve the fingerprinting-profiles. bowserless and creep-js would be a starting point.
• unofficial workaround. I can't tell a lot about it currently, since I didn't discover it, and the goal currently is, to keep it out of radar from sites like cloudfare.

Target

The target is for this project, to make it as complete as possible, as computing-efficient as possible and pass creep-js . It's the most advanced detection I've seen so far.

There are currently a lot of todos and things to improve, but I currently don't have the capacity to carry on. But I'd apprecheate contribution or working together with undetected-chromedriver. @ultrafunkamsterdam @sebdelsol

[alexreg]

@kaliiiiiiiiii What a great detailed reply... thank you very much! I can definitely see the reasoning behind this approach. Even if right now not many sites are (effectively) fingerprinting the browser/device, I'm sure more and more sites will start doing this in the future. And yes, good proxy support is very important too.

Anyway, I'm definitely up for trying out your library in the project I'm working on right now, so I can let you know how it goes and report bugs. Fingers crossed.

[alexreg]

P.S. I think you meant --headless=new is better but not traceless right? --headless=chrome should be the same as just --headless, i.e. the old method.

[kaliiiiiiiiii]

@kaliiiiiiiiii What a great detailed reply... thank you very much! I can definitely see the reasoning behind this approach.

Thanks!

Even if right now not many sites are (effectively) fingerprinting the browser/device, I'm sure more and more sites will start doing this in the future. And yes, good proxy support is very important too.

Yeah, but I assume that websites like cloudfare all ready have approaches. For example detecting when there's a big peak in the statistics for a specific device.

Anyway, I'm definitely up for trying out your library in the project I'm working on right now, so I can let you know how it goes and report bugs. Fingers crossed.

Great:) the newest version has some bugs (since proxy-support) regarding extensions. But I think the rest should be quite stable (except of the features I didn't ever use:) )

P.S. I think you meant --headless=new is better but not traceless right? --headless=chrome should be the same as just --headless, i.e. the old method.

Mhmm no, I actually meant --headless=chrome. Why do you mean it's the same?

[alexreg]

Sounds good.

Re headless mode, see this Selenium blog post for example, which makes clear --headless=chrome (= --headless) is the old one and --headless=new is the new one.

[ultrafunkamsterdam]

None of this a is true . Dig way deeper....

[kaliiiiiiiiii]

Regarding what do you mean? Headless? Or fingerprinting?
And if you say "dig way deeper", can you tell me what's wrong about?

[kaliiiiiiiiii]

Closing as it seems to be resolved