Is bot updated to beat captcha for recent discord?

[ghost]

I noticed whenever I try to DM user, I get captcha twice. In old discord, there used to be only one time clicking which brings up box to click image and submit. So I guess this client code is used to solve that whereas in recent times there's two checkboxes (first doesn't require any kind of solving just clicking checkbox while second brings up solve toolbox to click image).

My question is that is discord.py-self==2.1.0a1 smart enough to bypass both checkboxes. I am thinking of buying 2captcha or captchas.io subscription and I saw some discussion regarding #456 people and this (#320 not sure if he managed to hardcode url stuff or not) not being able to solve it. So buying 2captcha or captchas.io won't be worth it if this selfbot client isn't updated to beat those two checkboxes. Just want to know general thoughts if it can solve it or not so I wont be wasting money trying to bypass captcha. @dolfies

[ghost]

class CaptchaSolver(discord.CaptchaHandler):
    async def fetch_token(self, data: dict, proxy: str, proxy_auth: aiohttp.BasicAuth) -> str:
        api_key = captcha_api_key
        loop = asyncio.get_running_loop()
        print("Starting to solve captcha")
        with ThreadPoolExecutor() as pool:
            solver = TwoCaptcha(api_key)
            result = await loop.run_in_executor(pool, lambda: solver.hcaptcha(
                sitekey=data['captcha_sitekey'],
                data=data['captcha_rqdata'],
                url='https://discord.com/channels/@me',
            ))
            print("Returning the token")
            return json.dumps(result)

[ghost]

[ghost]

Ok finally an update using proxy. I figured out difference between browser and selfbot client and why my code didnt work. It returned it and discord.py-self seralized it, but it sends in data field of POST request as json. It should rather send X-Captcha-Key: key and X-captcha-Rqtoken as headers in order for it to work. I analysed browser captcha requests and its the same as i said before. Json of data field had content, nonce, tts and flag while captcha tokens etc were in headers
My home laptop had old version, vps had new. Reinstalled new version and now debugging why it doesnt work

[dolfies]

This change is implemented in the master branch. Please install the library from GitHub.

[ghost]

any idea why it fails? referrer? or flags:0 ? @dolfies

[ghost]

similar error from refactor/tls branch, can't analyse request in burpsuite due to this error

[ghost]

For the time being I will try putting {"flags":0} in json parameter or discord.com/channels/@me/channel_id in referer and see if it bypasses did nothing

[dolfies]

Looking at your code, you seem to be returning the raw response from the API. What you want to be returning is just the solution. Either way, 2Captcha is not known for quality captcha solutions, so I don't recommend using them.

[ghost]

Yeah, initially I used json.dumps(result), but it didnt match with browser headers so I changed it to result["code"] which matches with browser headers, but yet still doesn't work. And what captcha bypasser you recommend?

[dolfies]

None of the public ones. Solve them manually, find a decent private service, or roll your own :)

[ghost]

Tried another human worker based captcha and still failed. Looking more into browser requests there's calls to /api/v9/science just before sending captcha keys in headers. I think it might be worth looking into

[dolfies]

Nah, tracking requests don't matter. Mind sending your discord/tg so I can investigate a little further and determine if this is a library issue?

[WyattRichartz]

Was this problem ever solved or have any other advancements been made? I've been trying to bypass their captchas for joining guilds from a direct invite link using 2captcha and requests, but I've had no luck. I was about to come to the discussions to ask how to implement the discord.CaptchaHandler class.