providers.py

"""
Jade Code IDE - LLM Providers
Suporte a múltiplos providers: Groq, Cerebras, OpenRouter
"""

from abc import ABC, abstractmethod
from typing import Optional, List, Dict, Any
import httpx


class LLMProvider(ABC):
    """Classe base para providers de LLM."""
    
    def __init__(self, api_key: str):
        self.api_key = api_key
    
    @abstractmethod
    def chat(self, messages: List[Dict], model: str, tools: Optional[List] = None) -> Dict:
        """Envia mensagem pro LLM e retorna resposta."""
        pass
    
    @abstractmethod
    def list_models(self) -> List[Dict]:
        """Lista modelos disponíveis."""
        pass


class GroqProvider(LLMProvider):
    """Provider para Groq API."""
    
    BASE_URL = "https://api.groq.com/openai/v1"
    
    MODELS = [
        {"id": "moonshotai/kimi-k2-instruct-0905", "name": "Kimi K2 Instruct", "context": 128000},
        {"id": "meta-llama/llama-4-maverick-17b-128e-instruct", "name": "Llama 4 Maverick 17B", "context": 128000},
        {"id": "llama-3.3-70b-versatile", "name": "Llama 3.3 70B", "context": 128000},
    ]
    
    def chat(self, messages: List[Dict], model: str, tools: Optional[List] = None) -> Dict:
        headers = {
            "Authorization": f"Bearer {self.api_key}",
            "Content-Type": "application/json"
        }
        
        payload = {
            "model": model,
            "messages": messages,
            "temperature": 0.3,
            "max_tokens": 4096
        }
        
        if tools:
            payload["tools"] = tools
            payload["tool_choice"] = "auto"
        
        with httpx.Client(timeout=60.0) as client:
            response = client.post(
                f"{self.BASE_URL}/chat/completions",
                headers=headers,
                json=payload
            )
            response.raise_for_status()
            return response.json()
    
    def list_models(self) -> List[Dict]:
        return self.MODELS


class CerebrasProvider(LLMProvider):
    """Provider para Cerebras API (extremamente rápido)."""
    
    BASE_URL = "https://api.cerebras.ai/v1"
    
    MODELS = [
        {"id": "gpt-oss-120b", "name": "GPT OSS 120B", "context": 128000},
        {"id": "qwen-3-235b-a22b-instruct-2507", "name": "Qwen 3 235B", "context": 128000},
        {"id": "zai-glm-4.6", "name": "GLM 4.6", "context": 128000},
        {"id": "llama3.1-70b", "name": "Llama 3.1 70B", "context": 128000},
    ]
    
    def chat(self, messages: List[Dict], model: str, tools: Optional[List] = None) -> Dict:
        headers = {
            "Authorization": f"Bearer {self.api_key}",
            "Content-Type": "application/json"
        }
        
        payload = {
            "model": model,
            "messages": messages,
            "temperature": 0.3,
            "max_tokens": 4096
        }
        
        # Cerebras suporta tools também
        if tools:
            payload["tools"] = tools
            payload["tool_choice"] = "auto"
        
        with httpx.Client(timeout=60.0) as client:
            response = client.post(
                f"{self.BASE_URL}/chat/completions",
                headers=headers,
                json=payload
            )
            response.raise_for_status()
            return response.json()
    
    def list_models(self) -> List[Dict]:
        return self.MODELS


class OpenRouterProvider(LLMProvider):
    """Provider para OpenRouter (acesso a múltiplos modelos)."""
    
    BASE_URL = "https://openrouter.ai/api/v1"
    
    MODELS = [
        {"id": "xiaomi/mimo-v2-flash:free", "name": "Xiaomi Mimo V2 Flash", "context": 128000},
        {"id": "nex-agi/deepseek-v3.1-nex-n1:free", "name": "DeepSeek V3.1 Nex", "context": 128000},
        {"id": "qwen/qwen3-coder:free", "name": "Qwen 3 Coder", "context": 128000},
        {"id": "anthropic/claude-3.5-sonnet", "name": "Claude 3.5 Sonnet", "context": 200000},
        {"id": "google/gemini-2.0-flash-exp:free", "name": "Gemini 2.0 Flash (Free)", "context": 1000000},
    ]
    
    def chat(self, messages: List[Dict], model: str, tools: Optional[List] = None) -> Dict:
        headers = {
            "Authorization": f"Bearer {self.api_key}",
            "Content-Type": "application/json",
            "HTTP-Referer": "https://jade-code-ide.github.io",
            "X-Title": "Jade Code IDE"
        }
        
        payload = {
            "model": model,
            "messages": messages,
            "temperature": 0.3,
            "max_tokens": 4096
        }
        
        if tools:
            payload["tools"] = tools
            payload["tool_choice"] = "auto"
        
        with httpx.Client(timeout=120.0) as client:
            response = client.post(
                f"{self.BASE_URL}/chat/completions",
                headers=headers,
                json=payload
            )
            response.raise_for_status()
            return response.json()
    
    def list_models(self) -> List[Dict]:
        return self.MODELS


# === Factory ===
def get_provider(provider_name: str, api_key: str) -> LLMProvider:
    """Retorna o provider correto baseado no nome."""
    providers = {
        "groq": GroqProvider,
        "cerebras": CerebrasProvider,
        "openrouter": OpenRouterProvider
    }
    
    provider_class = providers.get(provider_name.lower())
    if not provider_class:
        raise ValueError(f"Provider '{provider_name}' não suportado. Use: {list(providers.keys())}")
    
    return provider_class(api_key)


def list_all_providers() -> List[Dict]:
    """Lista todos os providers disponíveis."""
    return [
        {"id": "groq", "name": "Groq", "description": "Rápido e gratuito"},
        {"id": "cerebras", "name": "Cerebras", "description": "Ultra rápido"},
        {"id": "openrouter", "name": "OpenRouter", "description": "Acesso a múltiplos modelos"},
    ]