Which functions need to be used to be still undetectable with uc=True

[MacMarde]

@mdmintz
As we know it is not possible to enter the URL automatically into the browser. If you do so, you will no longer pass CF captcha here: nowsecure.nl
Therefore you need to use the driver.get("https://www.nowsecure.nl") function.

See here

So the question is, are there more functions which should not be performed manually, for example sendkeys, click an element etc. ?

I want to do some tests to see if the whole SeleniumBase uc-driver is working for me before automating everything.
The idea is to start a browser with uc=True in SB and then just perform as much actions manually as possible without risking being detected.

[mdmintz]

Hello, In UC Mode (--uc / uc=True) SeleniumBase's driver.get(url) function has been modified to disconnect chromedriver from Chrome before navigating to the URL. This disconnect is necessary to prevent detection. There are other special methods, such as driver.uc_click(selector) and driver.uc_open_with_reconnect(url, reconnect_time), etc.

See https://www.youtube.com/watch?v=5dMFI3e85ig for details.

[MacMarde]

Hello, In UC Mode (--uc / uc=True) SeleniumBase's driver.get(url) function has been modified to disconnect chromedriver from Chrome before navigating to the URL. This disconnect is necessary to prevent detection. There are other special methods, such as driver.uc_click(selector) and driver.uc_open_with_reconnect(url, reconnect_time), etc.

See https://www.youtube.com/watch?v=5dMFI3e85ig for details.

Thanks so much. I am on it.

[MacMarde]

@mdmintz
Many clicks in my project on website elements opens a new website in a new browser tab.
How to make sure the new loading website in the new browser tab is also undetected?
As far as I understood you in your youtube video, driver.uc_click(selector) only disconnects for a short time. See here.
So how to handle it and make sure I am undetectable in this situations? Can I change the disconnect time for driver.uc_click(selector)?

[MacMarde]

@mdmintz
Ok thanks for the clarification. I think I understand it.
Sometimes after clicking a button, Selenium is waiting and pausing my python script until the new content has been loaded (probably not good for using this function).
But in cases where Selenium just clicks a button and immediately continues to process the python script it might help (when I am fast enough to call driver.reconnect() instantly).

[mdmintz]

You can set the page-load-strategy to "none" (pls="none") to prevent some extra waiting. That could also lead to unintended consequences (flaky tests, etc.)

[MacMarde]

Ah yes you are right. Thanks for reminding me. I did not consider this.

[MacMarde]

@mdmintz
I am thinking about a more dynamically way to solve this issue.
Because it is always just a matter of time when the reconnect time you set is not long enough when loading a website with driver.uc_open_with_reconnect("https://www.ebay.com", 6)
There could be many reasons for this. For example internet connection or proxy is slow, windows updates need high pc ressources, etc.

So I thought about something like this:

driver.uc_open_with_reconnect("https://www.ebay.com", 6)
while browser.execute_script("return document.readyState") != "complete":
        browser.reconnect(5)

Problem is that browser.reconnect(5) takes about 8 seconds in my test not 5 seconds. This means the rest of the function to disconnet and reconnect takes about 3 seconds plus 5 seconds timeout. The question is how long is the webdriver really connected in this example?
If it would be possible to make the actual time the webdriver is connected much shorter, you could use it to determine when the website has been completely loaded and the probability of getting detected this way would be less if the time being connected is for example within 100 ms.

What do you think? Can we measure the time actually connected? Can we make it faster?

[MacMarde]

It seems like it is not working as expected.
browser.execute_script("return document.readyState") sometimes returns "complete" when website is still loading and sometimes returns not instantly, instead it waits until website has been loaded completely and then also returns "complete".

I am just not happy with the UC solution. It is so easy for anti-bot measures to get around this solution. It would be great to have a reliable solution again like it was with the original UC from @ultrafunkamsterdam

[MacMarde]

@mdmintz
Maybe it would be a great feature to have a function for connecting/disconnecting chromedriver manually whenever needed.
What do you think?

[mdmintz]

driver.reconnect(seconds) in UC Mode. Upgrade to the latest version.

[MacMarde]

@mdmintz
What is with webelement.click() ?
Sometimes it is hard or not possible to get the element with a simple selector.

[mdmintz]

That feature already exists and is already more than 5 days old.

SeleniumBase/seleniumbase/core/browser_launcher.py

Line 438 in f9a31a3

reconnect_time=None,

Please make sure you are using the latest version of SeleniumBase before opening tickets.

[MacMarde]

@mdmintz
What selectors are supported with SB?
I can not click an element with its class name.

browser.uc_click('center', by="class_name", reconnect_time = random.uniform(4, 6))

"class" and "class name" also not working.

[MacMarde]

@mdmintz
Nevermind.
Something was weird.
It is working with "class name" instead of "class_name"

[MacMarde]

@mdmintz

I have tried to select an element like in your example.
I need to select a span element that matches 4 class names. In this example I want to select the third element:

driver.uc_click('span.class1.class2.class3.class4:nth-of-type(3)', by="css selector", reconnect_time = random.uniform(1, 2))

no matter what I am trying it is not working. Also not with xpath. It always gives me an error. element not found.

When using Selenium methods I find every element available on the website. In this case over 100 elements:

browser.find_elements(By.XPATH, ('//span[@class = "class1 class2 class3 class4"]'))

same true when using css selector.

I can select any element I want with Selenium function. Here for example element 25.

browser.find_elements(By.XPATH, ('//span[@class = "class1 class2 class3 class4"]'))[25]

But this does not help, because this way I think I can not parse it to uc_click()

Any idea how to do it? I tried for hours. No way to tell uc_click() which element to use.

[mdmintz]

Use a better CSS Selector. Or use XPath. Also your random.uniform(1, 2) will be less than 2, which might not be enough.

[MacMarde]

Could you please implement a way to use uc_click with elements like this:

webelement.uc_click()

It would be so much better and easier.

[mdmintz]

@MacMarde Have you tried that in UC Mode?

[iamumairayub]

@MacMarde AFAIK it already works

driver.find_element(by=By.CSS_SELECTOR, value='#some-id').uc_click(reconnect_time=5)

[mdmintz]

Yes, element.uc_click() already exists:

SeleniumBase/seleniumbase/undetected/webelement.py

Line 5 in 7d9d5c5

def uc_click(

[MacMarde]

@iamumairayub
Thanks so much. I will try it.