gpt4free/g4f/client/async_client.py
2024-05-21 20:39:31 +02:00

276 lines
9.7 KiB
Python

from __future__ import annotations
import time
import random
import string
import asyncio
import base64
from aiohttp import ClientSession, BaseConnector
from .types import Client as BaseClient
from .types import ProviderType, FinishReason
from .stubs import ChatCompletion, ChatCompletionChunk, ImagesResponse, Image
from .types import AsyncIterResponse, ImageProvider
from .image_models import ImageModels
from .helper import filter_json, find_stop, filter_none, cast_iter_async
from .service import get_last_provider, get_model_and_provider
from ..Provider import ProviderUtils
from ..typing import Union, Messages, AsyncIterator, ImageType
from ..errors import NoImageResponseError, ProviderNotFoundError
from ..requests.aiohttp import get_connector
from ..providers.conversation import BaseConversation
from ..image import ImageResponse as ImageProviderResponse, ImageDataResponse
try:
anext
except NameError:
async def anext(iter):
async for chunk in iter:
return chunk
async def iter_response(
response: AsyncIterator[str],
stream: bool,
response_format: dict = None,
max_tokens: int = None,
stop: list = None
) -> AsyncIterResponse:
content = ""
finish_reason = None
completion_id = ''.join(random.choices(string.ascii_letters + string.digits, k=28))
count: int = 0
async for chunk in response:
if isinstance(chunk, FinishReason):
finish_reason = chunk.reason
break
elif isinstance(chunk, BaseConversation):
yield chunk
continue
content += str(chunk)
count += 1
if max_tokens is not None and count >= max_tokens:
finish_reason = "length"
first, content, chunk = find_stop(stop, content, chunk)
if first != -1:
finish_reason = "stop"
if stream:
yield ChatCompletionChunk(chunk, None, completion_id, int(time.time()))
if finish_reason is not None:
break
finish_reason = "stop" if finish_reason is None else finish_reason
if stream:
yield ChatCompletionChunk(None, finish_reason, completion_id, int(time.time()))
else:
if response_format is not None and "type" in response_format:
if response_format["type"] == "json_object":
content = filter_json(content)
yield ChatCompletion(content, finish_reason, completion_id, int(time.time()))
async def iter_append_model_and_provider(response: AsyncIterResponse) -> AsyncIterResponse:
last_provider = None
async for chunk in response:
last_provider = get_last_provider(True) if last_provider is None else last_provider
chunk.model = last_provider.get("model")
chunk.provider = last_provider.get("name")
yield chunk
class AsyncClient(BaseClient):
def __init__(
self,
provider: ProviderType = None,
image_provider: ImageProvider = None,
**kwargs
):
super().__init__(**kwargs)
self.chat: Chat = Chat(self, provider)
self.images: Images = Images(self, image_provider)
def create_response(
messages: Messages,
model: str,
provider: ProviderType = None,
stream: bool = False,
proxy: str = None,
max_tokens: int = None,
stop: list[str] = None,
api_key: str = None,
**kwargs
):
has_asnyc = hasattr(provider, "create_async_generator")
if has_asnyc:
create = provider.create_async_generator
else:
create = provider.create_completion
response = create(
model, messages,
stream=stream,
**filter_none(
proxy=proxy,
max_tokens=max_tokens,
stop=stop,
api_key=api_key
),
**kwargs
)
if not has_asnyc:
response = cast_iter_async(response)
return response
class Completions():
def __init__(self, client: AsyncClient, provider: ProviderType = None):
self.client: AsyncClient = client
self.provider: ProviderType = provider
def create(
self,
messages: Messages,
model: str,
provider: ProviderType = None,
stream: bool = False,
proxy: str = None,
max_tokens: int = None,
stop: Union[list[str], str] = None,
api_key: str = None,
response_format: dict = None,
ignored : list[str] = None,
ignore_working: bool = False,
ignore_stream: bool = False,
**kwargs
) -> Union[ChatCompletion, AsyncIterator[ChatCompletionChunk]]:
model, provider = get_model_and_provider(
model,
self.provider if provider is None else provider,
stream,
ignored,
ignore_working,
ignore_stream
)
stop = [stop] if isinstance(stop, str) else stop
response = create_response(
messages, model,
provider, stream,
proxy=self.client.get_proxy() if proxy is None else proxy,
max_tokens=max_tokens,
stop=stop,
api_key=self.client.api_key if api_key is None else api_key,
**kwargs
)
response = iter_response(response, stream, response_format, max_tokens, stop)
response = iter_append_model_and_provider(response)
return response if stream else anext(response)
class Chat():
completions: Completions
def __init__(self, client: AsyncClient, provider: ProviderType = None):
self.completions = Completions(client, provider)
async def iter_image_response(
response: AsyncIterator,
response_format: str = None,
connector: BaseConnector = None,
proxy: str = None
) -> Union[ImagesResponse, None]:
async for chunk in response:
if isinstance(chunk, ImageProviderResponse):
if response_format == "b64_json":
async with ClientSession(
connector=get_connector(connector, proxy),
cookies=chunk.options.get("cookies")
) as session:
async def fetch_image(image):
async with session.get(image) as response:
return base64.b64encode(await response.content.read()).decode()
images = await asyncio.gather(*[fetch_image(image) for image in chunk.get_list()])
return ImagesResponse([Image(None, image, chunk.alt) for image in images], int(time.time()))
return ImagesResponse([Image(image, None, chunk.alt) for image in chunk.get_list()], int(time.time()))
elif isinstance(chunk, ImageDataResponse):
return ImagesResponse([Image(None, image, chunk.alt) for image in chunk.get_list()], int(time.time()))
def create_image(provider: ProviderType, prompt: str, model: str = "", **kwargs) -> AsyncIterator:
if isinstance(provider, type) and provider.__name__ == "You":
kwargs["chat_mode"] = "create"
else:
prompt = f"create a image with: {prompt}"
return provider.create_async_generator(
model,
[{"role": "user", "content": prompt}],
stream=True,
**kwargs
)
class Images():
def __init__(self, client: AsyncClient, provider: ImageProvider = None):
self.client: AsyncClient = client
self.provider: ImageProvider = provider
self.models: ImageModels = ImageModels(client)
def get_provider(self, model: str, provider: ProviderType = None):
if isinstance(provider, str):
if provider in ProviderUtils.convert:
provider = ProviderUtils.convert[provider]
else:
raise ProviderNotFoundError(f'Provider not found: {provider}')
else:
provider = self.models.get(model, self.provider)
return provider
async def generate(
self,
prompt,
model: str = "",
provider: ProviderType = None,
response_format: str = None,
connector: BaseConnector = None,
proxy: str = None,
**kwargs
) -> ImagesResponse:
provider = self.get_provider(model, provider)
if hasattr(provider, "create_async_generator"):
response = create_image(
provider,
prompt,
**filter_none(
response_format=response_format,
connector=connector,
proxy=self.client.get_proxy() if proxy is None else proxy,
),
**kwargs
)
else:
response = await provider.create_async(prompt)
return ImagesResponse([Image(image) for image in response.get_list()])
image = await iter_image_response(response, response_format, connector, proxy)
if image is None:
raise NoImageResponseError()
return image
async def create_variation(
self,
image: ImageType,
model: str = None,
response_format: str = None,
connector: BaseConnector = None,
proxy: str = None,
**kwargs
):
provider = self.get_provider(model, provider)
result = None
if hasattr(provider, "create_async_generator"):
response = provider.create_async_generator(
"",
[{"role": "user", "content": "create a image like this"}],
stream=True,
image=image,
**filter_none(
response_format=response_format,
connector=connector,
proxy=self.client.get_proxy() if proxy is None else proxy,
),
**kwargs
)
result = iter_image_response(response, response_format, connector, proxy)
if result is None:
raise NoImageResponseError()
return result