diff options
Diffstat (limited to 'g4f/Provider')
-rw-r--r-- | g4f/Provider/FreeChatgpt.py | 77 |
1 files changed, 35 insertions, 42 deletions
diff --git a/g4f/Provider/FreeChatgpt.py b/g4f/Provider/FreeChatgpt.py index 05a0016e..82fc3ef4 100644 --- a/g4f/Provider/FreeChatgpt.py +++ b/g4f/Provider/FreeChatgpt.py @@ -7,6 +7,15 @@ from ..typing import AsyncResult, Messages from .base_provider import AsyncGeneratorProvider + +models = { + "claude-v1":"claude-2.1", + "claude-v2":"claude-2.0", + "gpt_35_turbo":"gpt-3.5-turbo-1106", + "gpt-4":"gpt-4", + "gemini-pro":"google-gemini-pro" +} + class FreeChatgpt(AsyncGeneratorProvider): url = "https://free.chatgpt.org.uk" working = True @@ -22,47 +31,31 @@ class FreeChatgpt(AsyncGeneratorProvider): proxy: str = None, **kwargs ) -> AsyncResult: - if not model: - model = "gpt-3.5-turbo" + model = models[model] if model in models else "gpt-3.5-turbo-1106" headers = { - "User-Agent": "Mozilla/5.0 (X11; Ubuntu; Linux x86_64; rv:109.0) Gecko/20100101 Firefox/119.0", - "Accept": "application/json, text/event-stream", - "Accept-Language": "de,en-US;q=0.7,en;q=0.3", - "Accept-Encoding": "gzip, deflate, br", - "Content-Type": "application/json", - "Referer": "https://free.chatgpt.org.uk/", - "x-requested-with": "XMLHttpRequest", - "Origin": "https://free.chatgpt.org.uk", - "Sec-Fetch-Dest": "empty", - "Sec-Fetch-Mode": "cors", - "Sec-Fetch-Site": "same-origin", - "Connection": "keep-alive", - "Alt-Used": "free.chatgpt.org.uk", - "Pragma": "no-cache", - "Cache-Control": "no-cache", - "TE": "trailers", - } + "Accept": "application/json, text/event-stream", + "Content-Type":"application/json", + "Accept-Encoding": "gzip, deflate, br", + "Accept-Language": "en-US,en;q=0.5", + "Host":"free.chatgpt.org.uk", + "Referer":f"{cls.url}/", + "Origin":f"{cls.url}", + "Sec-Fetch-Dest": "empty", + "Sec-Fetch-Mode": "cors", + "Sec-Fetch-Site": "same-origin", + "User-Agent": "Mozilla/5.0 (X11; Linux x86_64) AppleWebKit/537.36 (KHTML, like Gecko) Chrome/120.0.0.0 Safari/537.36", +} async with ClientSession(headers=headers) as session: - data = { - "messages": messages, - "isAzure": False, - "azureApiVersion": "2023-08-01-preview", - "stream": True, - "model": model, - "temperature": 0.5, - "presence_penalty": 0, - "frequency_penalty": 0, - "top_p": 1, - "baseUrl": "/api/openai", - "maxIterations": 10, - "returnIntermediateSteps": True, - "useTools": ["web-search", "calculator", "web-browser"], - **kwargs - } - async with session.post(f"{cls.url}/api/langchain/tool/agent/nodejs", json=data, proxy=proxy) as response: - response.raise_for_status() - async for line in response.content: - if line.startswith(b"data: "): - data = json.loads(line[6:]) - if data["isSuccess"] and not data["isToolMessage"]: - yield data["message"]
\ No newline at end of file + data = {"messages":messages,"stream":True,"model":model,"temperature":0.5,"presence_penalty":0,"frequency_penalty":0,"top_p":1} + async with session.post(f'{cls.url}/api/openai/v1/chat/completions',json=data) as result: + async for chunk in result.content: + + line = chunk.decode() + if line.startswith("data: [DONE]"): + break + elif line.startswith("data: "): + line = json.loads(line[6:]) + if(line["choices"]==[]): + continue + if(line["choices"][0]["delta"].get("content") and line["choices"][0]["delta"]["content"]!=None): + yield line["choices"][0]["delta"]["content"]
\ No newline at end of file |