search

acheong08 / ChatGPT

Reverse engineered ChatGPT API
GNU General Public License v2.0
28.04k stars 4.48k forks source link

[BUG] Can't login | Cloudflare protection #261

Closed lambor590 closed 1 year ago

lambor590 commented 1 year ago

Describe the bug I'm not able to login.

To Reproduce Steps to reproduce the behavior:

  1. Use the library as normal, with email/password or session token.
  2. If you are using email/password you'll get an API Error at start
  3. If you are using a session token you'll get an Exception: Wrong response code at start

Expected behavior Be able to access the website as normal.

Output Using session token:

Traceback (most recent call last):
  File "/home/container/main.py", line 1, in <module>
    from src import bot
  File "/home/container/src/bot.py", line 3, in <module>
    from src import responses
  File "/home/container/src/responses.py", line 36, in <module>
    chatbot = Chatbot(config, conversation_id=None)
  File "/home/container/.local/lib/python3.10/site-packages/revChatGPT/revChatGPT.py", line 95, in __init__
    self.refresh_session()
  File "/home/container/.local/lib/python3.10/site-packages/revChatGPT/revChatGPT.py", line 296, in refresh_session
    raise Exception("Wrong response code")
Exception: Wrong response code

Using email/password:

Traceback (most recent call last):
  File "/home/container/main.py", line 1, in <module>
    from src import bot
  File "/home/container/src/bot.py", line 3, in <module>
    from src import responses
  File "/home/container/src/responses.py", line 36, in <module>
    chatbot = Chatbot(config, conversation_id=None)
  File "/home/container/.local/lib/python3.10/site-packages/revChatGPT/revChatGPT.py", line 95, in __init__
    self.refresh_session()
  File "/home/container/.local/lib/python3.10/site-packages/revChatGPT/revChatGPT.py", line 340, in refresh_session
    raise exc
  File "/home/container/.local/lib/python3.10/site-packages/revChatGPT/revChatGPT.py", line 337, in refresh_session
    self.login(self.config["email"], self.config["password"])
  File "/home/container/.local/lib/python3.10/site-packages/revChatGPT/revChatGPT.py", line 373, in login
    raise exc
  File "/home/container/.local/lib/python3.10/site-packages/revChatGPT/revChatGPT.py", line 366, in login
    auth.begin()
  File "/home/container/.local/lib/python3.10/site-packages/OpenAIAuth/OpenAIAuth.py", line 95, in begin
    raise Exception("API error")
Exception: API error

Environment (please complete the following information):

Please update your packages before reporting! pip3 install --upgrade OpenAIAuth revChatGPT

Additional context I saw that they applied Cloudflare on the website, and there are more cookies than before: image If I delete the cookie cf_clearance, the next time I reload the website I will get the Cloudflare is verifying your connection page, otherwise, it will not show.

michelleyin96 commented 1 year ago

I'm also seeing the same error - getting a 403 back

TheMihirNaik commented 1 year ago

I am seeing the same error as well.

coldhotstart commented 1 year ago

It will be difficult to bypass this, TBH.

MohamedAliRashad commented 1 year ago

same here

Soul-AI-LLM commented 1 year ago

I'm also seeing the same error - getting a 403 back

Same issue here.

birdup000 commented 1 year ago

same here

Lukium commented 1 year ago

Probably not super helpful but same here

CarlZielinski commented 1 year ago

Seeing the same issue. Is this something that could be bypassed using FlareSolverr?

acheong08 commented 1 year ago

I am working on this

acheong08 commented 1 year ago

This is the response:

<html lang="en-US">
<head>
    <title>Just a moment...</title>
    <meta http-equiv="Content-Type" content="text/html; charset=UTF-8">
    <meta http-equiv="X-UA-Compatible" content="IE=Edge">
    <meta name="robots" content="noindex,nofollow">
    <meta name="viewport" content="width=device-width,initial-scale=1">
    <link href="/cdn-cgi/styles/challenges.css" rel="stylesheet">

</head>
<body class="no-js">
    <div class="main-wrapper" role="main">
    <div class="main-content">
        <h1 class="zone-name-title h1">
            <img class="heading-favicon" src="/favicon.ico"
                 onerror="this.onerror=null;this.parentNode.removeChild(this)">
            chat.openai.com
        </h1>
        <h2 class="h2" id="challenge-running">
            Checking if the site connection is secure
        </h2>
        <noscript>
            <div id="challenge-error-title">
                <div class="h2">
                    <span class="icon-wrapper">
                        <div class="heading-icon warning-icon"></div>
                    </span>
                    <span id="challenge-error-text">
                        Enable JavaScript and cookies to continue
                    </span>
                </div>
            </div>
        </noscript>
        <div id="trk_jschal_js" style="display:none;background-image:url('/cdn-cgi/images/trace/managed/nojs/transparent.gif?ray=77823f748f39c5a0')"></div>
        <div id="challenge-body-text" class="core-msg spacer">
            chat.openai.com needs to review the security of your connection before proceeding.
        </div>
        <form id="challenge-form" action="/auth/login?__cf_chl_f_tk=YcrSC6GxLdFJV3nDQnB94iOY874kZRHn7xo.qxkz0ik-1670803940-0-gaNycGzNB2U" method="POST" enctype="application/x-www-form-urlencoded">
            <input type="hidden" name="md" value="neq8DhZscgOk9ZqTwAD01JFh9zb97x2wAxRY.AXXQZA-1670803940-0-ARsRYvxWGameB3Ago3KgmFO6sj1qc8uTru2TF6WetlBz22SNZXPxN1ww_9-tpkuzLjKUrYnZYELy7IK3lVMZVfyOoXUJdGnpR6bSjqPQXhpDdDzASo-ZuLx13VmjwGmi0DnoM2OOHEO3Mysl-ARX3TMqPF2T4PldmUCtKnFSmH-UeG2x7SKDp7dvaj9Bpuc-8s-HgVdVb7ARu1yiRtszSgxrPXPpbJNK34cqlNt1YMzvBeEjwN9kZxELK3mFkxbkvs_n6JTBvfF_iNrzVxdZUh4o5OCTkPdfwnaVZNvnU2LEIOd6QCZW-9ceUXUItt5OiKdzWhiiLiCPcLBqcQFiMgdEmCKypO7iulABlDraCARgkMz0p9zEDA99R3Ayh-aGESQUfjaarRHh6iFDbumoPxiguHWP0P21GYRFdYpzktFoasucmtjHfDBk7yW1LedHmYqyPlT4oC5c4BXcBxYZMDB8yxcNtM8QKWvPbg1BIJEUx_osPPyWTobkrJ7iFHEGsWL45MQpL7eMzx_wAqqp_2FdZwPKHkGjmCOpx6A3bngYRcHimCQjZNMqeje1iMitQjgI1-wTUM9RyvSOh53C5kD02s6qZ9PBmdicNCRTE5keLHLUd7pieJFQDIF4SjMmzZCiZ0d0Zv9JXYBa4XxzV7o6O5RBW76hTB8v06wIHKt8">
            <input type="hidden" name="r" value="laVUElSO5KeLxtGkMbmU9a0FovPXUSlj31DGVyIuyaQ-1670803940-0-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[370](https://github.com/acheong08/ChatGPT/actions/runs/3671477743/jobs/6206722710#step:6:371)57Jsqw4JfxOQSCrKMUdIRvMrUpuU9RL7/iIL9Ueq5biuqN6scDRyfpkHiiYKo5jBLO7qMNtZ0LraijOdccc6StyzlT5njjAxIKeDeuN9p8ni1MvTvcCe8kVGk/oB71IThCeaqNaKLCV8rMuEMqldEEujykA3KREeZVA+AOp2A5o5nkuTTfMluaFzu34TspGeIKeCKi5DxNo2Gu2xxlAgtlyECZLR7iV8c9uij64tSui6iFL6Mu">
        </form>
    </div>
</div>
<script>
    (function(){
        window._cf_chl_opt={
            cvId: '2',
            cType: 'managed',
            cNounce: '44898',
            cRay: '77823f748f39c5a0',
            cHash: '8065dd636c78e94',
            cUPMDTk: "\/auth\/login?__cf_chl_tk=YcrSC6GxLdFJV3nDQnB94iOY874kZRHn7xo.qxkz0ik-1670803940-0-gaNycGzNB2U",
            cFPWv: 'b',
            cTTimeMs: '1000',
            cTplV: 4,
            cTplB: 'cf',
            cRq: {
                ru: 'aHR0cHM6Ly9jaGF0Lm9wZW5haS5jb20vY2hhdC9mcm9udGVuZC8vYXV0aC9sb2dpbg==',
                ra: 'TW96aWxsYS81LjAgKE1hY2ludG9zaDsgSW50ZWwgTWFjIE9TIFggMTBfMTVfNykgQXBwbGVXZWJLaXQvNjA1LjEuMTUgKEtIVE1MLCBsaWtlIEdlY2tvKSBWZXJzaW9uLzE2LjEgU2FmYXJpLzYwNS4xLjE1',
                rm: 'R0VU',
                d: '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',
                t: 'MTY3MDgwMzk0MC41NzQwMDA=',
                m: '7Lx9YJ/uJ/1Rg0ZT5arYdwR4kn/s/tT0mBDT+sn5T4Y=',
                i1: 'MUxuc2VFZXTkaQ6kBzJ0Lw==',
                i2: 'pU5AjHJ1/ZnqMxuNQUy35g==',
                zh: 'iuyN59AGNgeBFFEOPj72EN+xtBfXXvoCyd1dJnIVFAc=',
                uh: 'cm2wjUr6M+ZSWJGGpQRwl8TWzcHP0274MxfQYzAQXGE=',
                hh: 'foaWmAUHGGlVCZaNUJIHhxzxFNzSPNnVe6rJjiQw728=',
            }
        };
        var trkjs = document.createElement('img');
        trkjs.setAttribute('src', '/cdn-cgi/images/trace/managed/js/transparent.gif?ray=77823f748f39c5a0');
        trkjs.setAttribute('style', 'display: none');
        document.body.appendChild(trkjs);
        var cpo = document.createElement('script');
        cpo.src = '/cdn-cgi/challenge-platform/h/b/orchestrate/managed/v1?ray=77823f748f39c5a0';
        window._cf_chl_opt.cOgUHash = location.hash === '' && location.href.indexOf('#') !== -1 ? '#' : location.hash;
        window._cf_chl_opt.cOgUQuery = location.search === '' && location.href.slice(0, -window._cf_chl_opt.cOgUHash.length).indexOf('?') !== -1 ? '?' : location.search;
        if (window.history && window.history.replaceState) {
            var ogU = location.pathname + window._cf_chl_opt.cOgUQuery + window._cf_chl_opt.cOgUHash;
            history.replaceState(null, null, "\/auth\/login?__cf_chl_rt_tk=YcrSC6GxLdFJV3nDQnB94iOY874kZRHn7xo.qxkz0ik-1670803940-0-gaNycGzNB2U" + window._cf_chl_opt.cOgUHash);
            cpo.onload = function() {
                history.replaceState(null, null, ogU);
            };
        }
        document.getElementsByTagName('head')[0].appendChild(cpo);
    }());
</script>

    <div class="footer" role="contentinfo">
        <div class="footer-inner">
            <div class="clearfix diagnostic-wrapper">
                <div class="ray-id">Ray ID: <code>77823f748f39c5a0</code></div>
            </div>
            <div class="text-center">Performance &amp; security by <a rel="noopener noreferrer" href="https://www.cloudflare.com?utm_source=challenge&utm_campaign=m" target="_blank">Cloudflare</a></div>
        </div>
    </div>
</body>
</html>

I will check if there is a way to solve it

nick-haoran commented 1 year ago

same error

jabontz09 commented 1 year ago

does this have to do with chatGPT using cloudflare now?

acheong08 commented 1 year ago

does this have to do with chatGPT using cloudflare now?

see title. yes

coldhotstart commented 1 year ago

Looks like the underlying openAI auth is not working 

with sync_playwright() as p:
            browser = p.chromium.launch(headless=False)
            page = browser.new_page()
            sync_stealth(page, pure=True)
            page.goto('https://chat.openai.com/auth/login')
            res = sync_cf_retry(page)
            if res:
                cookies = page.context.cookies()
                for cookie in cookies:
                    if cookie.get('name') == 'cf_clearance':
                        cf_clearance_value = cookie.get('value')
                        print(cf_clearance_value)
                ua = page.evaluate('() => {return navigator.userAgent}')
                print(ua)
            else:
                print("cf challenge fail")
            browser.close()

try to bypass it, but have some other issues too.

nick-haoran commented 1 year ago

Looks like the underlying openAI auth is not working

with sync_playwright() as p:
            browser = p.chromium.launch(headless=False)
            page = browser.new_page()
            sync_stealth(page, pure=True)
            page.goto('https://chat.openai.com/auth/login')
            res = sync_cf_retry(page)
            if res:
                cookies = page.context.cookies()
                for cookie in cookies:
                    if cookie.get('name') == 'cf_clearance':
                        cf_clearance_value = cookie.get('value')
                        print(cf_clearance_value)
                ua = page.evaluate('() => {return navigator.userAgent}')
                print(ua)
            else:
                print("cf challenge fail")
            browser.close()

try to bypass it, but have some other issues too.

it was broken by cloudflare.......

acheong08 commented 1 year ago

I am fixing. please wait

coldhotstart commented 1 year ago

Looks like the underlying openAI auth is not working

with sync_playwright() as p:
            browser = p.chromium.launch(headless=False)
            page = browser.new_page()
            sync_stealth(page, pure=True)
            page.goto('https://chat.openai.com/auth/login')
            res = sync_cf_retry(page)
            if res:
                cookies = page.context.cookies()
                for cookie in cookies:
                    if cookie.get('name') == 'cf_clearance':
                        cf_clearance_value = cookie.get('value')
                        print(cf_clearance_value)
                ua = page.evaluate('() => {return navigator.userAgent}')
                print(ua)
            else:
                print("cf challenge fail")
            browser.close()

try to bypass it, but have some other issues too.

it was broken by cloudflare.......

It doesn't matter, you can always launch a browser and then simulate the traffic to the openAI, there is no way to block it.

acheong08 commented 1 year ago

The point here is to use their API. A browser is too heavy

nick-haoran commented 1 year ago

The point here is to use their API. A browser is too heavy

As you think, selenium is disgusting to use

Cubicpath commented 1 year ago

For reference, the cf_clearance cookie is tied to a combination of your IP address and User Agent, and is currently valid for 2 hours.

msnyder5 commented 1 year ago

You could maybe try using cloudscraper. I'm not sure if it works atm https://pypi.org/project/cloudscraper/

BooBooKittyFck commented 1 year ago

@msnyder5

I tried already - and it wants you to use the anticaptcha services - actually i don't have a account to test it. Screenshot_20221212_014651_com sonelli juicessh

transitive-bullshit commented 1 year ago

cf_clearance will be required going forwards. The open questions are:

  1. How to reliably renew your accessToken if you have a valid clearance token
  2. How to reliably renew a clearanceToken

Sharing some of my WIP attempts at resolving the issue:

https://github.com/transitive-bullshit/chatgpt-api/pull/99 - attempting to more closely match the official HTTP requests https://github.com/transitive-bullshit/chatgpt-twitter-bot/pull/5 - added attempts at using puppeteer and playwright

alexandreteles commented 1 year ago

You could maybe try using cloudscraper. I'm not sure if it works atm https://pypi.org/project/cloudscraper/

I tried already - and it wants you to use the anticaptcha services - actually i don't have account to test it.

I do have an account to test it if the maintainer wants to try,

coldhotstart commented 1 year ago

The point here is to use their API. A browser is too heavy

the browser is used to get the cf_clearance cookie and UA, then use that to call the API. I am able to get to the

def part_three(self, token: str) -> None: """ We reuse the token from part to make a request to /api/auth/signin/auth0?prompt=login """ self.debugger.log("Beginning part three") url = "https://chat.openai.com/api/auth/signin/auth0?prompt=login" headers = { "Host": "ask.openai.com", "Origin": "https://chat.openai.com", "Connection": "keep-alive", "Accept": "/", "User-Agent": self.ua, "Referer": "https://chat.openai.com/auth/login", "Content-Length": "100", "Accept-Language": "en-GB,en-US;q=0.9,en;q=0.8", "Content-Type": "application/x-www-form-urlencoded", } payload = f"callbackUrl=%2F&csrfToken={token}&json=true" response = self.session.post(url=url, headers=headers, cookies=self.cookies, data=payload) print("part_three response=" + str(response))

Part1 and Part2 passed .

acheong08 commented 1 year ago

I can't get pass part1 even with the clearance cookie. I might be doing something wrong.

Cubicpath commented 1 year ago

@transitive-bullshit Once we get a clearanceToken, getting the accessToken should be the same as before, by calling https://chat.openai.com/api/auth/session with a valid session token. We should only need the clearanceToken for accessing the website.

alexandreteles commented 1 year ago

I can't get pass part1 even with the clearance cookie. I might be doing something wrong.

Are you matching the IP and user_agent of the browser used to obtain the clearance cookie?

coldhotstart commented 1 year ago

def begin(self) -> None: """ Begin the auth process """ self.debugger.log("Beginning auth process") if not self.email_address or not self.password: return

    if self.use_proxy:
        if not self.proxy:
            return

        proxies = {
            "http": self.proxy,
            "https": self.proxy,
        }
        self.session.proxies = proxies

    # First, make a request to https://chat.openai.com/auth/login
    # get cf_clearance
    self.ua = {}
    self.cookies = {}
    with sync_playwright() as p:
        browser = p.chromium.launch(headless=False)
        page = browser.new_page()
        sync_stealth(page, pure=True)
        page.goto('https://chat.openai.com/auth/login')
        res = sync_cf_retry(page)
        if res:
            cookies = page.context.cookies()
            for cookie in cookies:
                if cookie.get('name') == 'cf_clearance':
                    cf_clearance_value = cookie.get('value')
                    print(cf_clearance_value)
            ua = page.evaluate('() => {return navigator.userAgent}')
            print(ua)
        else:
            print("cf challenge fail")
        browser.close()
    url = "https://chat.openai.com/auth/login"
    self.ua = ua
    headers = {
                "Host": "ask.openai.com",
                "Accept": "text/html,application/xhtml+xml,application/xml;q=0.9,*/*;q=0.8",
                "User-Agent": self.ua,
                "Accept-Language": "en-GB,en-US;q=0.9,en;q=0.8",
                "Accept-Encoding": "gzip, deflate, br",
                "Connection": "keep-alive",
            }
    self.cookies = {"cf_clearance": cf_clearance_value}
    response = self.session.get(url=url, headers=headers, cookies=self.cookies)
    if response.status_code == 200:
        self.part_two()
    else:
        self.debugger.log("Error in part one")
        self.debugger.log("Response: ", end="")
        print(response.text)
        self.debugger.log("Status code: ", end="")
        self.debugger.log(response.status_code)
        raise Exception("API error")

def part_two(self) -> None:
    """
    In part two, We make a request to https://chat.openai.com/api/auth/csrf and grab a fresh csrf token
    """
    self.debugger.log("Beginning part two")
    headers = {
                "Host": "ask.openai.com",
                "Accept": "*/*",
                "Connection": "keep-alive",
                "User-Agent": self.ua,
                "Accept-Language": "en-GB,en-US;q=0.9,en;q=0.8",
                "Referer": "https://chat.openai.com/auth/login",
                "Accept-Encoding": "gzip, deflate, br",
            }
    url = "https://chat.openai.com/api/auth/csrf"
    response = self.session.get(url=url, headers=headers, cookies=self.cookies)
    if response.status_code == 200 and "json" in response.headers["Content-Type"]:
        csrf_token = response.json()["csrfToken"]
        print("csrf_token=" + csrf_token)
        self.part_three(token=csrf_token)
    else:
        self.debugger.log("Error in part two")
        self.debugger.log("Response: ", end="")
        self.debugger.log(response.text)
        self.debugger.log("Status code: ", end="")
        self.debugger.log(response.status_code)
        raise Exception("Error logging in")
acheong08 commented 1 year ago

Since cloudflare expires, might as well use session tokens along with the clearance. Logging in with the token is too much work to maintain without adding functionality due to the clearance limit.

nick-haoran commented 1 year ago

由于 cloudflare 过期，还不如使用会话令牌和许可。由于许可限制，在不添加功能的情况下，使用令牌登录的维护工作量太大。

so do we have other solutions?

acheong08 commented 1 year ago

Yes. Manually getting the cookies from the browser and putting them in the config (only allows 2 hours before manual refresh needed)

acheong08 commented 1 year ago

This is a temporary solution which I will try to find a better one later

coldhotstart commented 1 year ago

the better way is to separate the token getting process from the token usage process. creating a token pool and just use chrome robot to get it, you are able to use that for 2Hrs , that's already enough. I am also working on one solution.

Cubicpath commented 1 year ago

I don't see an issue with keeping the current implementation, we only have to change step 1 so we get the clearanceToken, preferable with browser automation. Unless I'm missing something.

acheong08 commented 1 year ago

I don't see an issue with keeping the current implementation, we only have to change step 1 so we get the clearanceToken, preferable with browser automation. Unless I'm missing something.

Browser automation is the issue. Not supported on headless instances

acheong08 commented 1 year ago

I will build a way to use the clearance token but getting that token will be left to the users

MohamedAliRashad commented 1 year ago

This may not help with the current issue but i got the chance to try svg-captcha-solver and succeed in fixing the captcha request in the start.

alexandreteles commented 1 year ago

I will build a way to use the clearance token but getting that token will be left to the users

Sounds good to me.

Browser automation is the issue. Not supported on headless instances

It most definitely is, but I agree with you that how the token is obtained is a question users should answer. If I may, would it be possible to have it as an optional parameter instead of a required one? One could be using a proxy to bypass Cloudflare's shenanigans and not necessarily be forced to provide the clearenceToken sounds like a good idea.

acheong08 commented 1 year ago

Current bypass: https://github.com/acheong08/ChatGPT/releases/tag/0.0.38.7

In config.json:

{
    "session_token": "<YOUR TOKEN>",
    "cf_clearance": "<cf_clearance cookie value>",
    "user_agent": "<USER_AGENT>"
}
acheong08 commented 1 year ago

One could be using a proxy to bypass Cloudflare's shenanigans and not necessarily be forced to provide the clearenceToken sounds like a good idea.

Proxies don't help in this case

acheong08 commented 1 year ago

If I may, would it be possible to have it as an optional parameter instead of a required one?

The whole thing doesn't work anymore without the cloudflare token

Cubicpath commented 1 year ago

Proxies don't help in this case

Do proxies work at all anymore? As long as you use the same proxy for all requests cf_clearance should be valid. Unless cloudflare blocks proxies :/

htser commented 1 year ago

I'm trying to use automated scripts to periodically fill in cookie

Harry-Jing commented 1 year ago

I'm trying to use automated scripts to periodically fill in cookie

by playwrite?

alexandreteles commented 1 year ago

One could be using a proxy to bypass Cloudflare's shenanigans and not necessarily be forced to provide the clearenceToken sounds like a good idea.

Proxies don't help in this case

There are solutions like cloudscraper and some paid/proprietary local proxy servers that are 100% able to bypass and/or make it easier to work around Cloudflare (the kind of tool scalpers genarally use), reason why I asked if passing as an optional argument would be possible

acheong08 commented 1 year ago

Proxies don't help in this case

Do proxies work at all anymore? As long as you use the same proxy for all requests cf_clearance should be valid. Unless cloudflare blocks proxies :/

What I meant was that proxies can't bypass the need for a cf_clearance

acheong08 commented 1 year ago

One could be using a proxy to bypass Cloudflare's shenanigans and not necessarily be forced to provide the clearenceToken sounds like a good idea.

Proxies don't help in this case

There are solutions like cloudscraper and some paid/proprietary local proxy servers that are 100% able to bypass and/or make it easier to work around Cloudflare (the kind of tool scalpers genarally use), reason why I asked if passing as an optional argument would be possible

I'm currently having class (school) right now. I'll try to look into it during lunch

abacaj commented 1 year ago

Proxies don't help in this case

Do proxies work at all anymore? As long as you use the same proxy for all requests cf_clearance should be valid. Unless cloudflare blocks proxies :/

What I meant was that proxies can't bypass the need for a cf_clearance

Just for reference any solution will be limited to IP/machine. Can't pass the cookie around.

On remote VM using same cookie: image

coldhotstart commented 1 year ago

The hardest part is that the cloudflare will ask for Hcaptcha , so I don't think totally getting ride of web browser is possible at this point .