mirror of
https://github.com/xtekky/gpt4free.git
synced 2024-11-10 19:11:01 +00:00
6ef282de3a
Fix many providers Add selenium-wire to requierments
54 lines
1.9 KiB
Python
54 lines
1.9 KiB
Python
from __future__ import annotations
|
|
|
|
from aiohttp import ClientSession, ClientTimeout
|
|
|
|
from ...typing import AsyncResult, Messages
|
|
from ..base_provider import AsyncGeneratorProvider
|
|
|
|
|
|
class ChatAnywhere(AsyncGeneratorProvider):
|
|
url = "https://chatanywhere.cn"
|
|
supports_gpt_35_turbo = True
|
|
supports_message_history = True
|
|
working = False
|
|
|
|
@classmethod
|
|
async def create_async_generator(
|
|
cls,
|
|
model: str,
|
|
messages: Messages,
|
|
proxy: str = None,
|
|
timeout: int = 120,
|
|
temperature: float = 0.5,
|
|
**kwargs
|
|
) -> AsyncResult:
|
|
headers = {
|
|
"User-Agent": "Mozilla/5.0 (X11; Ubuntu; Linux x86_64; rv:109.0) Gecko/20100101 Firefox/119.0",
|
|
"Accept": "application/json, text/plain, */*",
|
|
"Accept-Language": "de,en-US;q=0.7,en;q=0.3",
|
|
"Accept-Encoding": "gzip, deflate, br",
|
|
"Content-Type": "application/json",
|
|
"Referer": f"{cls.url}/",
|
|
"Origin": cls.url,
|
|
"Sec-Fetch-Dest": "empty",
|
|
"Sec-Fetch-Mode": "cors",
|
|
"Sec-Fetch-Site": "same-origin",
|
|
"Authorization": "",
|
|
"Connection": "keep-alive",
|
|
"TE": "trailers"
|
|
}
|
|
async with ClientSession(headers=headers, timeout=ClientTimeout(timeout)) as session:
|
|
data = {
|
|
"list": messages,
|
|
"id": "s1_qYuOLXjI3rEpc7WHfQ",
|
|
"title": messages[-1]["content"],
|
|
"prompt": "",
|
|
"temperature": temperature,
|
|
"models": "61490748",
|
|
"continuous": True
|
|
}
|
|
async with session.post(f"{cls.url}/v1/chat/gpt/", json=data, proxy=proxy) as response:
|
|
response.raise_for_status()
|
|
async for chunk in response.content.iter_any():
|
|
if chunk:
|
|
yield chunk.decode() |