phind.com | how to set cf_clearance and user_agent

[zjrwtx]

[Z3R0F0]

same

[coding-bug-git]

same, found cf_clearance token still not work

[joshyorko]

same

[xtekky]

did u guys set the correct user_agent from your browser ?

[ozkanuysal]

Same error for me.

[ozkanuysal]

did u guys set the correct user_agent from your browser ?

Yeah, actually i set my agent from whatismybrowser.com

[AndroidTest10]

@zjrwtx we need to set user_agent only in the file streamlit_app.py ?

and is there a need to set any value in cf_clearance in the same file?

[giottobright]

did u guys set the correct user_agent from your browser ?

Can you please tell me where exactly i can set the user_agent from my browser?

and where if im using phind should i insert this code? `# if using CompletionModel.load set these ora.user_id = '...' ora.session_token = '...'

model = ora.CompletionModel.load(chatbot_id, 'gpt-4')

[giottobright]

and which cf_clearance token to use?

[shubhamwagh]

I too get the same error: Following are my settings in streamlit_app.py:

phind.user_agent = '<set using https://whatmyuseragent.com>'
phind.cf_clearance = ''

[maxolie00]

same issue, running on localhost

[maxolie00]

Main things are: make sure your using the CORRECT useragent, 2 make sure your getting the clearance token FROM THE API SITE aka if its from phind go to the phind API and get the clearance and lastly IF ITS ON A REMOTE SERVER like a vps the clearance WILL NOT WORK because the IP has to match, make sure to use a proxy server hosted on that VPS then get the clerance with that proxy

Bare with me please, i got my useragent and put it in phind.user_agent '" and im on phind.com, how am i supposed to get my clearence token to fill in the phind.cf_clearence = i suppose ?

[Z3R0F0]

Main things are: make sure your using the CORRECT useragent, 2 make sure your getting the clearance token FROM THE API SITE aka if its from phind go to the phind API and get the clearance and lastly IF ITS ON A REMOTE SERVER like a vps the clearance WILL NOT WORK because the IP has to match, make sure to use a proxy server hosted on that VPS then get the clerance with that proxy

so, i got my useragent from whatismybrowser, and got cf_clearance from phind.com (cookies). Still same error. Am I wrong somewhere ?

[maxolie00]

Bare with me please, i got my useragent and put it in phind.user_agent '" and im on phind.com, how am i supposed to get my clearence token to fill in the phind.cf_clearence = i suppose ?

You get it using the chrome dev tools and going to cookies

Thank you for showing me this, I got my cf_clearance and i got the user id from whatmyuseragent.com and filled them in the right places in the streamlit_app.py running it again i get this new error : An error occured, please make sure you are using a cf_clearance token and correct useragent | 'rawBingResults'

[shubhamwagh]

@maxolie00 , I got it working jut after couple of refreshes Thanks @localuser-isback for your help :)

[maxolie00]

@maxolie00 , I got it working jut after couple of refreshes Thanks @localuser-isback for your help :)

Same, now working after refreshes too, thanks all

[xtekky]

Hi, phind in their panick are sometimes disabling and re-enabling cloudflare, which may require you to not set it and to set it, please refer to the comment of @localuser-isback

"make sure your using the CORRECT useragent, 2 make sure your getting the clearance token FROM THE API SITE aka if its from phind go to the phind API and get the clearance and lastly IF ITS ON A REMOTE SERVER like a vps the clearance WILL NOT WORK because the IP has to match, make sure to use a proxy server hosted on that VPS then get the clerance with that proxy"

you can also just use chatgpt and ask: "how do I get the user agent of my browser" and "how do I see the cookies a site is using"

[pheonis2]

Bare with me please, i got my useragent and put it in phind.user_agent '" and im on phind.com, how am i supposed to get my clearence token to fill in the phind.cf_clearence = i suppose ?

You get it using the chrome dev tools and going to cookies

Thanks, I managed to get this running. My question is, do i need to get the cf_clearance token everytime i start the streamlit app?

[Ilan1274]

Why is there only “https://www.phind.com/” in my cookies ，and I don't see the cf_clearance

[Ilan1274]

can i use the chai.openai.com cf_clearance? I can see two cf_clearance from the cookies

[Ilan1274]

maybe the project need a clearer tutorial /(ㄒoㄒ)/~~

[arianyambao]

EDIT 2: https://github.com/xtekky/gpt4free/issues/153 - The use of phind.com is now forbidden EDIT 1: This is just a band-aid solution to make it work. Also, please keep in mind that the cf_clearance expires and you need to refresh the page to get a new one. Here: Step 1: Go to phind.com and get your cf_clearance via inspect element -> application -> storage -> cookies (see the images of others above) -- They expire. Step 2: Get your user-agent here Step 3: Edit your phind folder's __init__.py (basically copy this and follow through. I've removed some of the lines, you may compare them later with the original code.

• Manually paste your cf_clearance on lines: 79, 139, and 203
• Manually paste your user-agent on lines: 88, 146, and 212

  
  from urllib.parse import quote
  from time         import time
  from datetime     import datetime
  from queue        import Queue, Empty
  from threading    import Thread
  from re           import findall

from curl_cffi.requests import post

cf_clearance = '' # No need to put anything here (gets wiped) user_agent = '' # No need to put anything here (gets wiped)

class PhindResponse:

class Completion:

    class Choices:
        def __init__(self, choice: dict) -> None:
            self.text           = choice['text']
            self.content        = self.text.encode()
            self.index          = choice['index']
            self.logprobs       = choice['logprobs']
            self.finish_reason  = choice['finish_reason']

        def __repr__(self) -> str:
            return f'''<__main__.APIResponse.Completion.Choices(\n    text           = {self.text.encode()},\n    index          = {self.index},\n    logprobs       = {self.logprobs},\n    finish_reason  = {self.finish_reason})object at 0x1337>'''

    def __init__(self, choices: dict) -> None:
        self.choices = [self.Choices(choice) for choice in choices]

class Usage:
    def __init__(self, usage_dict: dict) -> None:
        self.prompt_tokens      = usage_dict['prompt_tokens']
        self.completion_tokens  = usage_dict['completion_tokens']
        self.total_tokens       = usage_dict['total_tokens']

    def __repr__(self):
        return f'''<__main__.APIResponse.Usage(\n    prompt_tokens      = {self.prompt_tokens},\n    completion_tokens  = {self.completion_tokens},\n    total_tokens       = {self.total_tokens})object at 0x1337>'''

def __init__(self, response_dict: dict) -> None:

    self.response_dict  = response_dict
    self.id             = response_dict['id']
    self.object         = response_dict['object']
    self.created        = response_dict['created']
    self.model          = response_dict['model']
    self.completion     = self.Completion(response_dict['choices'])
    self.usage          = self.Usage(response_dict['usage'])

def json(self) -> dict:
    return self.response_dict

class Search: def create(prompt: str, actualSearch: bool = True, language: str = 'en') -> dict: # None = no search

    if not actualSearch:
        return {
            '_type': 'SearchResponse',
            'queryContext': {
                'originalQuery': prompt
            },
            'webPages': {
                'webSearchUrl': f'https://www.bing.com/search?q={quote(prompt)}',
                'totalEstimatedMatches': 0,
                'value': []
            },
            'rankingResponse': {
                'mainline': {
                    'items': []
                }
            }
        }

    headers = {
        'authority': 'www.phind.com',
        'accept': '*/*',
        'accept-language': 'en,fr-FR;q=0.9,fr;q=0.8,es-ES;q=0.7,es;q=0.6,en-US;q=0.5,am;q=0.4,de;q=0.3',
        'cookie': f'cf_clearance=PASTEHERE',
        'origin': 'https://www.phind.com',
        'referer': 'https://www.phind.com/search?q=hi&c=&source=searchbox&init=true',
        'sec-ch-ua': '"Chromium";v="112", "Google Chrome";v="112", "Not:A-Brand";v="99"',
        'sec-ch-ua-mobile': '?0',
        'sec-ch-ua-platform': '"macOS"',
        'sec-fetch-dest': 'empty',
        'sec-fetch-mode': 'cors',
        'sec-fetch-site': 'same-origin',
        'user-agent': '' # PASTE HERE
    }

    return post('https://www.phind.com/api/bing/search', headers = headers, json = { 
        'q': prompt,
        'userRankList': {},
        'browserLanguage': language}).json()['rawBingResults']

class Completion: def create( model = 'gpt-4', prompt: str = '', results: dict = None, creative: bool = False, detailed: bool = False, codeContext: str = '', language: str = 'en') -> PhindResponse:

    if results is None:
        results = Search.create(prompt, actualSearch = True)

    if len(codeContext) > 2999:
        raise ValueError('codeContext must be less than 3000 characters')

    models = {
        'gpt-4' : 'expert',
        'gpt-3.5-turbo' : 'intermediate',
        'gpt-3.5': 'intermediate',
    }

    json_data = {
        'question'    : prompt,
        'bingResults' : results, #response.json()['rawBingResults'],
        'codeContext' : codeContext,
        'options': {
            'skill'   : models[model],
            'date'    : datetime.now().strftime("%d/%m/%Y"),
            'language': language,
            'detailed': detailed,
            'creative': creative
        }
    }

    headers = {
        'authority': 'www.phind.com',
        'accept': '*/*',
        'accept-language': 'en,fr-FR;q=0.9,fr;q=0.8,es-ES;q=0.7,es;q=0.6,en-US;q=0.5,am;q=0.4,de;q=0.3',
        'content-type': 'application/json',
        'cookie': f'cf_clearance=PASTEHERE',
        'origin': 'https://www.phind.com',
        'referer': 'https://www.phind.com/search?q=hi&c=&source=searchbox&init=true',
        'sec-ch-ua': '"Chromium";v="112", "Google Chrome";v="112", "Not:A-Brand";v="99"',
        'sec-ch-ua-mobile': '?0',
        'sec-ch-ua-platform': '"macOS"',
        'sec-fetch-dest': 'empty',
        'sec-fetch-mode': 'cors',
        'sec-fetch-site': 'same-origin',
        'user-agent': '' # PASTE HERE
    }

    completion = ''
    response   = post('https://www.phind.com/api/infer/answer', headers = headers, json = json_data, timeout=99999, impersonate='chrome110')
    for line in response.text.split('\r\n\r\n'):
        completion += (line.replace('data: ', ''))

    return PhindResponse({
        'id'     : f'cmpl-1337-{int(time())}', 
        'object' : 'text_completion', 
        'created': int(time()), 
        'model'  : models[model], 
        'choices': [{
                'text'          : completion, 
                'index'         : 0, 
                'logprobs'      : None, 
                'finish_reason' : 'stop'
        }], 
        'usage': {
            'prompt_tokens'     : len(prompt), 
            'completion_tokens' : len(completion), 
            'total_tokens'      : len(prompt) + len(completion)
        }
    })

class StreamingCompletion: message_queue = Queue() stream_completed = False

def request(model, prompt, results, creative, detailed, codeContext, language) -> None:

    models = {
        'gpt-4' : 'expert',
        'gpt-3.5-turbo' : 'intermediate',
        'gpt-3.5': 'intermediate',
    }

    json_data = {
        'question'    : prompt,
        'bingResults' : results,
        'codeContext' : codeContext,
        'options': {
            'skill'   : models[model],
            'date'    : datetime.now().strftime("%d/%m/%Y"),
            'language': language,
            'detailed': detailed,
            'creative': creative
        }
    }

    headers = {
        'authority': 'www.phind.com',
        'accept': '*/*',
        'accept-language': 'en,fr-FR;q=0.9,fr;q=0.8,es-ES;q=0.7,es;q=0.6,en-US;q=0.5,am;q=0.4,de;q=0.3',
        'content-type': 'application/json',
        'cookie': f'cf_clearance=PASTEHERE',
        'origin': 'https://www.phind.com',
        'referer': 'https://www.phind.com/search?q=hi&c=&source=searchbox&init=true',
        'sec-ch-ua': '"Chromium";v="112", "Google Chrome";v="112", "Not:A-Brand";v="99"',
        'sec-ch-ua-mobile': '?0',
        'sec-ch-ua-platform': '"macOS"',
        'sec-fetch-dest': 'empty',
        'sec-fetch-mode': 'cors',
        'sec-fetch-site': 'same-origin',
        'user-agent': '' # PASTE HERE
    }

    response   = post('https://www.phind.com/api/infer/answer', 
        headers = headers, json = json_data, timeout=99999, impersonate='chrome110', content_callback=StreamingCompletion.handle_stream_response)

    StreamingCompletion.stream_completed = True

@staticmethod
def create(
    model       : str = 'gpt-4', 
    prompt      : str = '', 
    results     : dict = None, 
    creative    : bool = False, 
    detailed    : bool = False, 
    codeContext : str = '',
    language    : str = 'en'):

    if results is None:
        results = Search.create(prompt, actualSearch = True)

    if len(codeContext) > 2999:
        raise ValueError('codeContext must be less than 3000 characters')

    Thread(target = StreamingCompletion.request, args = [
        model, prompt, results, creative, detailed, codeContext, language]).start()

    while StreamingCompletion.stream_completed != True or not StreamingCompletion.message_queue.empty():
        try:
            chunk = StreamingCompletion.message_queue.get(timeout=0)

            if chunk == b'data:  \r\ndata: \r\ndata: \r\n\r\n':
                chunk = b'data:  \n\n\r\n\r\n'

            chunk = chunk.decode()

            chunk = chunk.replace('data: \r\n\r\ndata: ', 'data: \n')
            chunk = chunk.replace('\r\ndata: \r\ndata: \r\n\r\n', '\n\n\r\n\r\n')
            chunk = chunk.replace('data: ', '').replace('\r\n\r\n', '')

            yield PhindResponse({
                'id'     : f'cmpl-1337-{int(time())}', 
                'object' : 'text_completion', 
                'created': int(time()), 
                'model'  : model, 
                'choices': [{
                        'text'          : chunk, 
                        'index'         : 0, 
                        'logprobs'      : None, 
                        'finish_reason' : 'stop'
                }], 
                'usage': {
                    'prompt_tokens'     : len(prompt), 
                    'completion_tokens' : len(chunk), 
                    'total_tokens'      : len(prompt) + len(chunk)
                }
            })

        except Empty:
            pass

@staticmethod
def handle_stream_response(response):
    StreamingCompletion.message_queue.put(response)

![image](https://user-images.githubusercontent.com/29201095/234626169-0ff36e98-5e2b-4144-a48a-4094e53d866e.png)

P.S. Improvements here is probably setting the keys via streamlit's implementation of `session_state` kinda looks like the keys get wiped every call for some reason

[AbdelrhmanNile]

@arianyambao is it working normally for you as of this moment? im still getting the error

[AbdelrhmanNile]

@arianyambao im getting this error

[AbdelrhmanNile]

i examined the response, it returns an html page, looks like its not passing the cloudflare captcha

[MDanyalT]

Im also getting the same error, "An error occured, please make sure you are using a cf_clearance token and correct useragent | Expecting value: line 1 column 1 (char 0)"

I've made sure my useragent and cf_clearance are both filled in correctly as well

[BastienYTB]

;-;

[jarvisar]

My solution was to set phind.cf_clearance and phind.user_agent on lines 4 and 5 in streamlit_app.py

[richen666]

@jarvisar Thank you, and you're right. But after trying several times, the similar errors still occur and you must gain cf_clearance and copy it again. It doesn't seem very convenient.

[arianyambao]

@jarvisar Thank you, and you're right. But after trying several times, the similar errors still occur and you must gain cf_clearance and copy it. It doesn't seem very convenient.

Yes, the clearance, similar to cookies expires and gets renewed...

[arianyambao]

@AbdelrhmanNile have you tried pasting a new cf_clearance? It might be because the one you use expired... Expires after some time cc: @suckurmom @BastienYTB

Since I also got the same error after my token expired. (It's really not efficient)

[tikserziku]

The local version works, but slower than the real chatgpt

[EricDylan47]

编辑：这只是使其工作的创可贴解决方案。另外，请记住，过期，您需要刷新页面才能获得新页面。在这里：第 1 步：转到 phind.com 并通过检查元素获取您的cf_clearance ->应用程序 ->存储 -> cookie（请参阅上面其他人的图像）-- 它们过期。第 2 步：在这里获取您的 第 3 步：编辑您的 phind 文件夹（基本上复制此文件夹并继续执行。我已经删除了一些行，您可以稍后将它们与原始代码进行比较。cf_clearance``user-agent``__init__.py

• 手动粘贴您的行：、 和cf_clearance``79``139``203
• 手动粘贴您的行：、 和user-agent``88``146``212

from urllib.parse import quote
from time         import time
from datetime     import datetime
from queue        import Queue, Empty
from threading    import Thread
from re           import findall

from curl_cffi.requests import post

cf_clearance = '' # No need to put anything here (gets wiped)
user_agent   = '' # No need to put anything here (gets wiped)

class PhindResponse:

    class Completion:

        class Choices:
            def __init__(self, choice: dict) -> None:
                self.text           = choice['text']
                self.content        = self.text.encode()
                self.index          = choice['index']
                self.logprobs       = choice['logprobs']
                self.finish_reason  = choice['finish_reason']

            def __repr__(self) -> str:
                return f'''<__main__.APIResponse.Completion.Choices(\n    text           = {self.text.encode()},\n    index          = {self.index},\n    logprobs       = {self.logprobs},\n    finish_reason  = {self.finish_reason})object at 0x1337>'''

        def __init__(self, choices: dict) -> None:
            self.choices = [self.Choices(choice) for choice in choices]

    class Usage:
        def __init__(self, usage_dict: dict) -> None:
            self.prompt_tokens      = usage_dict['prompt_tokens']
            self.completion_tokens  = usage_dict['completion_tokens']
            self.total_tokens       = usage_dict['total_tokens']

        def __repr__(self):
            return f'''<__main__.APIResponse.Usage(\n    prompt_tokens      = {self.prompt_tokens},\n    completion_tokens  = {self.completion_tokens},\n    total_tokens       = {self.total_tokens})object at 0x1337>'''

    def __init__(self, response_dict: dict) -> None:

        self.response_dict  = response_dict
        self.id             = response_dict['id']
        self.object         = response_dict['object']
        self.created        = response_dict['created']
        self.model          = response_dict['model']
        self.completion     = self.Completion(response_dict['choices'])
        self.usage          = self.Usage(response_dict['usage'])

    def json(self) -> dict:
        return self.response_dict

class Search:
    def create(prompt: str, actualSearch: bool = True, language: str = 'en') -> dict: # None = no search

        if not actualSearch:
            return {
                '_type': 'SearchResponse',
                'queryContext': {
                    'originalQuery': prompt
                },
                'webPages': {
                    'webSearchUrl': f'https://www.bing.com/search?q={quote(prompt)}',
                    'totalEstimatedMatches': 0,
                    'value': []
                },
                'rankingResponse': {
                    'mainline': {
                        'items': []
                    }
                }
            }

        headers = {
            'authority': 'www.phind.com',
            'accept': '*/*',
            'accept-language': 'en,fr-FR;q=0.9,fr;q=0.8,es-ES;q=0.7,es;q=0.6,en-US;q=0.5,am;q=0.4,de;q=0.3',
            'cookie': f'cf_clearance=PASTEHERE',
            'origin': 'https://www.phind.com',
            'referer': 'https://www.phind.com/search?q=hi&c=&source=searchbox&init=true',
            'sec-ch-ua': '"Chromium";v="112", "Google Chrome";v="112", "Not:A-Brand";v="99"',
            'sec-ch-ua-mobile': '?0',
            'sec-ch-ua-platform': '"macOS"',
            'sec-fetch-dest': 'empty',
            'sec-fetch-mode': 'cors',
            'sec-fetch-site': 'same-origin',
            'user-agent': '' # PASTE HERE
        }

        return post('https://www.phind.com/api/bing/search', headers = headers, json = { 
            'q': prompt,
            'userRankList': {},
            'browserLanguage': language}).json()['rawBingResults']

class Completion:
    def create(
        model = 'gpt-4', 
        prompt: str = '', 
        results: dict = None, 
        creative: bool = False, 
        detailed: bool = False, 
        codeContext: str = '',
        language: str = 'en') -> PhindResponse:

        if results is None:
            results = Search.create(prompt, actualSearch = True)

        if len(codeContext) > 2999:
            raise ValueError('codeContext must be less than 3000 characters')

        models = {
            'gpt-4' : 'expert',
            'gpt-3.5-turbo' : 'intermediate',
            'gpt-3.5': 'intermediate',
        }

        json_data = {
            'question'    : prompt,
            'bingResults' : results, #response.json()['rawBingResults'],
            'codeContext' : codeContext,
            'options': {
                'skill'   : models[model],
                'date'    : datetime.now().strftime("%d/%m/%Y"),
                'language': language,
                'detailed': detailed,
                'creative': creative
            }
        }

        headers = {
            'authority': 'www.phind.com',
            'accept': '*/*',
            'accept-language': 'en,fr-FR;q=0.9,fr;q=0.8,es-ES;q=0.7,es;q=0.6,en-US;q=0.5,am;q=0.4,de;q=0.3',
            'content-type': 'application/json',
            'cookie': f'cf_clearance=PASTEHERE',
            'origin': 'https://www.phind.com',
            'referer': 'https://www.phind.com/search?q=hi&c=&source=searchbox&init=true',
            'sec-ch-ua': '"Chromium";v="112", "Google Chrome";v="112", "Not:A-Brand";v="99"',
            'sec-ch-ua-mobile': '?0',
            'sec-ch-ua-platform': '"macOS"',
            'sec-fetch-dest': 'empty',
            'sec-fetch-mode': 'cors',
            'sec-fetch-site': 'same-origin',
            'user-agent': '' # PASTE HERE
        }

        completion = ''
        response   = post('https://www.phind.com/api/infer/answer', headers = headers, json = json_data, timeout=99999, impersonate='chrome110')
        for line in response.text.split('\r\n\r\n'):
            completion += (line.replace('data: ', ''))

        return PhindResponse({
            'id'     : f'cmpl-1337-{int(time())}', 
            'object' : 'text_completion', 
            'created': int(time()), 
            'model'  : models[model], 
            'choices': [{
                    'text'          : completion, 
                    'index'         : 0, 
                    'logprobs'      : None, 
                    'finish_reason' : 'stop'
            }], 
            'usage': {
                'prompt_tokens'     : len(prompt), 
                'completion_tokens' : len(completion), 
                'total_tokens'      : len(prompt) + len(completion)
            }
        })

class StreamingCompletion:
    message_queue    = Queue()
    stream_completed = False

    def request(model, prompt, results, creative, detailed, codeContext, language) -> None:

        models = {
            'gpt-4' : 'expert',
            'gpt-3.5-turbo' : 'intermediate',
            'gpt-3.5': 'intermediate',
        }

        json_data = {
            'question'    : prompt,
            'bingResults' : results,
            'codeContext' : codeContext,
            'options': {
                'skill'   : models[model],
                'date'    : datetime.now().strftime("%d/%m/%Y"),
                'language': language,
                'detailed': detailed,
                'creative': creative
            }
        }

        headers = {
            'authority': 'www.phind.com',
            'accept': '*/*',
            'accept-language': 'en,fr-FR;q=0.9,fr;q=0.8,es-ES;q=0.7,es;q=0.6,en-US;q=0.5,am;q=0.4,de;q=0.3',
            'content-type': 'application/json',
            'cookie': f'cf_clearance=PASTEHERE',
            'origin': 'https://www.phind.com',
            'referer': 'https://www.phind.com/search?q=hi&c=&source=searchbox&init=true',
            'sec-ch-ua': '"Chromium";v="112", "Google Chrome";v="112", "Not:A-Brand";v="99"',
            'sec-ch-ua-mobile': '?0',
            'sec-ch-ua-platform': '"macOS"',
            'sec-fetch-dest': 'empty',
            'sec-fetch-mode': 'cors',
            'sec-fetch-site': 'same-origin',
            'user-agent': '' # PASTE HERE
        }

        response   = post('https://www.phind.com/api/infer/answer', 
            headers = headers, json = json_data, timeout=99999, impersonate='chrome110', content_callback=StreamingCompletion.handle_stream_response)

        StreamingCompletion.stream_completed = True

    @staticmethod
    def create(
        model       : str = 'gpt-4', 
        prompt      : str = '', 
        results     : dict = None, 
        creative    : bool = False, 
        detailed    : bool = False, 
        codeContext : str = '',
        language    : str = 'en'):

        if results is None:
            results = Search.create(prompt, actualSearch = True)

        if len(codeContext) > 2999:
            raise ValueError('codeContext must be less than 3000 characters')

        Thread(target = StreamingCompletion.request, args = [
            model, prompt, results, creative, detailed, codeContext, language]).start()

        while StreamingCompletion.stream_completed != True or not StreamingCompletion.message_queue.empty():
            try:
                chunk = StreamingCompletion.message_queue.get(timeout=0)

                if chunk == b'data:  \r\ndata: \r\ndata: \r\n\r\n':
                    chunk = b'data:  \n\n\r\n\r\n'

                chunk = chunk.decode()

                chunk = chunk.replace('data: \r\n\r\ndata: ', 'data: \n')
                chunk = chunk.replace('\r\ndata: \r\ndata: \r\n\r\n', '\n\n\r\n\r\n')
                chunk = chunk.replace('data: ', '').replace('\r\n\r\n', '')

                yield PhindResponse({
                    'id'     : f'cmpl-1337-{int(time())}', 
                    'object' : 'text_completion', 
                    'created': int(time()), 
                    'model'  : model, 
                    'choices': [{
                            'text'          : chunk, 
                            'index'         : 0, 
                            'logprobs'      : None, 
                            'finish_reason' : 'stop'
                    }], 
                    'usage': {
                        'prompt_tokens'     : len(prompt), 
                        'completion_tokens' : len(chunk), 
                        'total_tokens'      : len(prompt) + len(chunk)
                    }
                })

            except Empty:
                pass

    @staticmethod
    def handle_stream_response(response):
        StreamingCompletion.message_queue.put(response)

附言这里的改进可能是通过 streamlit 的实现设置键，看起来由于某种原因每次调用都会擦除键session_state

sstillll not work

[shubhamwagh]

Check this issue #153 , phind.com is forbidden now

[MDanyalT]

After updating my cookies it seems to work now, thanks

[xtekky]

194 and #153

[liuSir2022]

This problem is so serious that the project cannot be deployed. Most of us have encountered this problem. Please come out and give a correct answer