summaryrefslogtreecommitdiffstats
path: root/g4f/Provider/TwitterBio.py
diff options
context:
space:
mode:
Diffstat (limited to 'g4f/Provider/TwitterBio.py')
-rw-r--r--g4f/Provider/TwitterBio.py103
1 files changed, 0 insertions, 103 deletions
diff --git a/g4f/Provider/TwitterBio.py b/g4f/Provider/TwitterBio.py
deleted file mode 100644
index c143e4ff..00000000
--- a/g4f/Provider/TwitterBio.py
+++ /dev/null
@@ -1,103 +0,0 @@
-from __future__ import annotations
-
-import json
-import re
-from aiohttp import ClientSession
-
-from ..typing import AsyncResult, Messages
-from .base_provider import AsyncGeneratorProvider, ProviderModelMixin
-from .helper import format_prompt
-
-class TwitterBio(AsyncGeneratorProvider, ProviderModelMixin):
- url = "https://www.twitterbio.io"
- api_endpoint_mistral = "https://www.twitterbio.io/api/mistral"
- api_endpoint_openai = "https://www.twitterbio.io/api/openai"
- working = True
- supports_gpt_35_turbo = True
-
- default_model = 'gpt-3.5-turbo'
- models = [
- 'mistralai/Mixtral-8x7B-Instruct-v0.1',
- 'gpt-3.5-turbo',
- ]
-
- model_aliases = {
- "mixtral-8x7b": "mistralai/Mixtral-8x7B-Instruct-v0.1",
- }
-
- @classmethod
- def get_model(cls, model: str) -> str:
- if model in cls.models:
- return model
- return cls.default_model
-
- @staticmethod
- def format_text(text: str) -> str:
- text = re.sub(r'\s+', ' ', text.strip())
- text = re.sub(r'\s+([,.!?])', r'\1', text)
- return text
-
- @classmethod
- async def create_async_generator(
- cls,
- model: str,
- messages: Messages,
- proxy: str = None,
- **kwargs
- ) -> AsyncResult:
- model = cls.get_model(model)
-
- headers = {
- "accept": "*/*",
- "accept-language": "en-US,en;q=0.9",
- "cache-control": "no-cache",
- "content-type": "application/json",
- "origin": cls.url,
- "pragma": "no-cache",
- "priority": "u=1, i",
- "referer": f"{cls.url}/",
- "sec-ch-ua": '"Chromium";v="127", "Not)A;Brand";v="99"',
- "sec-ch-ua-mobile": "?0",
- "sec-ch-ua-platform": '"Linux"',
- "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/127.0.0.0 Safari/537.36"
- }
- async with ClientSession(headers=headers) as session:
- prompt = format_prompt(messages)
- data = {
- "prompt": f'{prompt}.'
- }
-
- if model == 'mistralai/Mixtral-8x7B-Instruct-v0.1':
- api_endpoint = cls.api_endpoint_mistral
- elif model == 'gpt-3.5-turbo':
- api_endpoint = cls.api_endpoint_openai
- else:
- raise ValueError(f"Unsupported model: {model}")
-
- async with session.post(api_endpoint, json=data, proxy=proxy) as response:
- response.raise_for_status()
- buffer = ""
- async for line in response.content:
- line = line.decode('utf-8').strip()
- if line.startswith('data: '):
- try:
- json_data = json.loads(line[6:])
- if model == 'mistralai/Mixtral-8x7B-Instruct-v0.1':
- if 'choices' in json_data and len(json_data['choices']) > 0:
- text = json_data['choices'][0].get('text', '')
- if text:
- buffer += text
- elif model == 'gpt-3.5-turbo':
- text = json_data.get('text', '')
- if text:
- buffer += text
- except json.JSONDecodeError:
- continue
- elif line == 'data: [DONE]':
- break
-
- if buffer:
- yield cls.format_text(buffer)