from __future__ import annotations import json import uuid from aiohttp import ClientSession, ClientTimeout, ClientResponseError from ...typing import AsyncResult, Messages from ..base_provider import AsyncGeneratorProvider, ProviderModelMixin from ..helper import format_prompt from ...image import ImageResponse class AmigoChat(AsyncGeneratorProvider, ProviderModelMixin): url = "https://amigochat.io/chat/" chat_api_endpoint = "https://api.amigochat.io/v1/chat/completions" image_api_endpoint = "https://api.amigochat.io/v1/images/generations" working = False supports_stream = True supports_system_message = True supports_message_history = True default_model = 'gpt-4o-mini' chat_models = [ 'gpt-4o', default_model, 'o1-preview', 'o1-mini', 'meta-llama/Meta-Llama-3.1-405B-Instruct-Turbo', 'meta-llama/Llama-3.2-90B-Vision-Instruct-Turbo', 'claude-3-sonnet-20240229', 'gemini-1.5-pro', ] image_models = [ 'flux-pro/v1.1', 'flux-realism', 'flux-pro', 'dalle-e-3', ] models = [*chat_models, *image_models] model_aliases = { "o1": "o1-preview", "llama-3.1-405b": "meta-llama/Meta-Llama-3.1-405B-Instruct-Turbo", "llama-3.2-90b": "meta-llama/Llama-3.2-90B-Vision-Instruct-Turbo", "claude-3.5-sonnet": "claude-3-sonnet-20240229", "gemini-pro": "gemini-1.5-pro", "flux-pro": "flux-pro/v1.1", "dalle-3": "dalle-e-3", } persona_ids = { 'gpt-4o': "gpt", 'gpt-4o-mini': "amigo", 'o1-preview': "openai-o-one", 'o1-mini': "openai-o-one-mini", 'meta-llama/Meta-Llama-3.1-405B-Instruct-Turbo': "llama-three-point-one", 'meta-llama/Llama-3.2-90B-Vision-Instruct-Turbo': "llama-3-2", 'claude-3-sonnet-20240229': "claude", 'gemini-1.5-pro': "gemini-1-5-pro", 'flux-pro/v1.1': "flux-1-1-pro", 'flux-realism': "flux-realism", 'flux-pro': "flux-pro", 'dalle-e-3': "dalle-three", } @classmethod def get_model(cls, model: str) -> str: if model in cls.models: return model elif model in cls.model_aliases: return cls.model_aliases[model] else: return cls.default_model @classmethod def get_personaId(cls, model: str) -> str: return cls.persona_ids[model] @classmethod async def create_async_generator( cls, model: str, messages: Messages, proxy: str = None, stream: bool = False, **kwargs ) -> AsyncResult: model = cls.get_model(model) device_uuid = str(uuid.uuid4()) max_retries = 3 retry_count = 0 while retry_count < max_retries: try: headers = { "accept": "*/*", "accept-language": "en-US,en;q=0.9", "authorization": "Bearer", "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="129", "Not=A?Brand";v="8"', "sec-ch-ua-mobile": "?0", "sec-ch-ua-platform": '"Linux"', "user-agent": "Mozilla/5.0 (X11; Linux x86_64) AppleWebKit/537.36 (KHTML, like Gecko) Chrome/129.0.0.0 Safari/537.36", "x-device-language": "en-US", "x-device-platform": "web", "x-device-uuid": device_uuid, "x-device-version": "1.0.32" } async with ClientSession(headers=headers) as session: if model in cls.chat_models: # Chat completion data = { "messages": [{"role": m["role"], "content": m["content"]} for m in messages], "model": model, "personaId": cls.get_personaId(model), "frequency_penalty": 0, "max_tokens": 4000, "presence_penalty": 0, "stream": stream, "temperature": 0.5, "top_p": 0.95 } timeout = ClientTimeout(total=300) # 5 minutes timeout async with session.post(cls.chat_api_endpoint, json=data, proxy=proxy, timeout=timeout) as response: if response.status not in (200, 201): error_text = await response.text() raise Exception(f"Error {response.status}: {error_text}") async for line in response.content: line = line.decode('utf-8').strip() if line.startswith('data: '): if line == 'data: [DONE]': break try: chunk = json.loads(line[6:]) # Remove 'data: ' prefix if 'choices' in chunk and len(chunk['choices']) > 0: choice = chunk['choices'][0] if 'delta' in choice: content = choice['delta'].get('content') elif 'text' in choice: content = choice['text'] else: content = None if content: yield content except json.JSONDecodeError: pass else: # Image generation prompt = messages[-1]['content'] data = { "prompt": prompt, "model": model, "personaId": cls.get_personaId(model) } async with session.post(cls.image_api_endpoint, json=data, proxy=proxy) as response: response.raise_for_status() response_data = await response.json() if "data" in response_data: image_urls = [] for item in response_data["data"]: if "url" in item: image_url = item["url"] image_urls.append(image_url) if image_urls: yield ImageResponse(image_urls, prompt) else: yield None break except (ClientResponseError, Exception) as e: retry_count += 1 if retry_count >= max_retries: raise e device_uuid = str(uuid.uuid4())