unrealon 1.0.9__py3-none-any.whl → 1.1.0__py3-none-any.whl

unrealon_llm/src/utils/models_cache.py

@@ -1,592 +0,0 @@
-"""
-Models Cache Utilities
-
-Automatically fetch and cache LLM model information with pricing from providers.
-Eliminates hardcoding and provides up-to-date model data and pricing.
-"""
-
-import asyncio
-import json
-import logging
-import os
-from datetime import datetime, timedelta
-from pathlib import Path
-from typing import Any, Dict, List, Optional
-
-import aiohttp
-from cachetools import TTLCache
-
-from unrealon_llm.src.dto import LLMProvider
-from unrealon_llm.src.exceptions import APIError, CacheError, ConfigurationError
-
-logger = logging.getLogger(__name__)
-
-
-class ModelInfo:
-    """Model information with pricing and capabilities"""
-
-    def __init__(
-        self,
-        id: str,
-        name: str,
-        provider: str,
-        context_length: int = 0,
-        prompt_price: float = 0.0,
-        completion_price: float = 0.0,
-        description: Optional[str] = None,
-        tags: Optional[List[str]] = None,
-        is_available: bool = True,
-        currency: str = "USD",
-    ):
-        self.id = id
-        self.name = name
-        self.provider = provider
-        self.context_length = context_length
-        self.prompt_price = prompt_price  # Price per 1M tokens
-        self.completion_price = completion_price  # Price per 1M tokens
-        self.description = description or ""
-        self.tags = tags or []
-        self.is_available = is_available
-        self.currency = currency
-
-    def estimate_cost(self, input_tokens: int, output_tokens: int) -> float:
-        """Estimate cost for given token usage"""
-        input_cost = (input_tokens / 1_000_000) * self.prompt_price
-        output_cost = (output_tokens / 1_000_000) * self.completion_price
-        return input_cost + output_cost
-
-    def to_dict(self) -> Dict[str, Any]:
-        """Convert to dictionary representation"""
-        return {
-            "id": self.id,
-            "name": self.name,
-            "provider": self.provider,
-            "context_length": self.context_length,
-            "prompt_price": self.prompt_price,
-            "completion_price": self.completion_price,
-            "description": self.description,
-            "tags": self.tags,
-            "is_available": self.is_available,
-            "currency": self.currency,
-        }
-
-    @classmethod
-    def from_dict(cls, data: Dict[str, Any]) -> "ModelInfo":
-        """Create ModelInfo from dictionary"""
-        return cls(
-            id=data.get("id", ""),
-            name=data.get("name", ""),
-            provider=data.get("provider", ""),
-            context_length=data.get("context_length", 0),
-            prompt_price=data.get("prompt_price", 0.0),
-            completion_price=data.get("completion_price", 0.0),
-            description=data.get("description"),
-            tags=data.get("tags", []),
-            is_available=data.get("is_available", True),
-            currency=data.get("currency", "USD"),
-        )
-
-
-class ModelsCache:
-    """
-    Automatic models cache with provider integration
-    Fetches current model data and pricing from APIs
-    """
-
-    def __init__(
-        self,
-        openrouter_api_key: Optional[str] = None,
-        openai_api_key: Optional[str] = None,
-        anthropic_api_key: Optional[str] = None,
-        cache_ttl_hours: int = 24,  # Cache for 24 hours
-        max_cache_size: int = 1000,
-    ):
-        """
-        Initialize models cache with provider API keys
-
-        Args:
-            openrouter_api_key: OpenRouter API key
-            openai_api_key: OpenAI API key
-            anthropic_api_key: Anthropic API key
-            cache_ttl_hours: Cache TTL in hours
-            max_cache_size: Maximum cache entries
-        """
-        self.openrouter_api_key = openrouter_api_key
-        self.openai_api_key = openai_api_key
-        self.anthropic_api_key = anthropic_api_key
-
-        # Initialize cache
-        cache_ttl = cache_ttl_hours * 3600  # Convert to seconds
-        self.cache = TTLCache(maxsize=max_cache_size, ttl=cache_ttl)
-
-        # Models storage
-        self.models: Dict[str, ModelInfo] = {}
-        self.last_fetch_time: Optional[datetime] = None
-
-        # Cache keys
-        self.MODELS_CACHE_KEY = "all_models"
-        self.OPENROUTER_CACHE_KEY = "openrouter_models"
-        self.OPENAI_CACHE_KEY = "openai_models"
-        self.ANTHROPIC_CACHE_KEY = "anthropic_models"
-
-        # Cache settings
-        self.cache_ttl_hours = cache_ttl_hours
-
-        # Provider endpoints
-        self.provider_endpoints = {
-            "openrouter": "https://openrouter.ai/api/v1/models",
-            "openai": "https://api.openai.com/v1/models",
-            "anthropic": "https://api.anthropic.com/v1/models",
-        }
-
-    async def fetch_all_models(
-        self, force_refresh: bool = False
-    ) -> Dict[str, ModelInfo]:
-        """
-        Fetch models from all available providers
-
-        Args:
-            force_refresh: Force refresh even if cache is valid
-
-        Returns:
-            Dictionary of model_id -> ModelInfo
-        """
-        # Quick check - if models already loaded, return them
-        if not force_refresh and self.models:
-            logger.info(f"🔥 Using already loaded {len(self.models)} models from memory cache")
-            return self.models
-
-        # Check cache first
-        if not force_refresh and self.MODELS_CACHE_KEY in self.cache:
-            logger.info("💾 Using models from TTL cache")
-            cached_data = self.cache[self.MODELS_CACHE_KEY]
-            self.models = {
-                k: ModelInfo.from_dict(v) for k, v in cached_data["models"].items()
-            }
-            self.last_fetch_time = datetime.fromisoformat(cached_data["fetch_time"])
-            return self.models
-
-        logger.info("🌐 Fetching fresh models from all providers (no cache)")
-
-        # Fetch from all available providers
-        all_models = {}
-
-        # Fetch OpenRouter models (priority provider)
-        if self.openrouter_api_key:
-            try:
-                openrouter_models = await self._fetch_openrouter_models()
-                all_models.update(openrouter_models)
-                logger.info(f"Fetched {len(openrouter_models)} models from OpenRouter")
-            except Exception as e:
-                logger.warning(f"Failed to fetch OpenRouter models: {e}")
-
-        # Fetch OpenAI models
-        if self.openai_api_key:
-            try:
-                openai_models = await self._fetch_openai_models()
-                all_models.update(openai_models)
-                logger.info(f"Fetched {len(openai_models)} models from OpenAI")
-            except Exception as e:
-                logger.warning(f"Failed to fetch OpenAI models: {e}")
-
-        # Fetch Anthropic models
-        if self.anthropic_api_key:
-            try:
-                anthropic_models = await self._fetch_anthropic_models()
-                all_models.update(anthropic_models)
-                logger.info(f"Fetched {len(anthropic_models)} models from Anthropic")
-            except Exception as e:
-                logger.warning(f"Failed to fetch Anthropic models: {e}")
-
-        # Add fallback models if no providers available
-        if not all_models:
-            logger.warning("No models fetched from providers, using fallback models")
-            all_models = self._get_fallback_models()
-
-        # Update cache
-        self.models = all_models
-        self.last_fetch_time = datetime.now()
-
-        cache_data = {
-            "models": {k: v.to_dict() for k, v in all_models.items()},
-            "fetch_time": self.last_fetch_time.isoformat(),
-        }
-        self.cache[self.MODELS_CACHE_KEY] = cache_data
-
-        logger.info(f"Total models cached: {len(all_models)}")
-        return self.models
-
-    async def _fetch_openrouter_models(self) -> Dict[str, ModelInfo]:
-        """Fetch models from OpenRouter API"""
-        if not self.openrouter_api_key:
-            return {}
-
-        async with aiohttp.ClientSession() as session:
-            try:
-                async with session.get(
-                    self.provider_endpoints["openrouter"],
-                    headers={
-                        "Authorization": f"Bearer {self.openrouter_api_key}",
-                        "Content-Type": "application/json",
-                    },
-                    timeout=aiohttp.ClientTimeout(total=30),
-                ) as response:
-                    response.raise_for_status()
-                    data = await response.json()
-
-                    models = {}
-                    for model_data in data.get("data", []):
-                        model_info = self._parse_openrouter_model(model_data)
-                        if model_info:
-                            models[model_info.id] = model_info
-
-                    return models
-
-            except aiohttp.ClientError as e:
-                raise APIError(f"OpenRouter API error: {e}")
-            except Exception as e:
-                raise APIError(f"Failed to fetch OpenRouter models: {e}")
-
-    async def _fetch_openai_models(self) -> Dict[str, ModelInfo]:
-        """Fetch models from OpenAI API"""
-        if not self.openai_api_key:
-            return {}
-
-        async with aiohttp.ClientSession() as session:
-            try:
-                async with session.get(
-                    self.provider_endpoints["openai"],
-                    headers={
-                        "Authorization": f"Bearer {self.openai_api_key}",
-                        "Content-Type": "application/json",
-                    },
-                    timeout=aiohttp.ClientTimeout(total=30),
-                ) as response:
-                    response.raise_for_status()
-                    data = await response.json()
-
-                    models = {}
-                    for model_data in data.get("data", []):
-                        model_info = self._parse_openai_model(model_data)
-                        if model_info:
-                            models[model_info.id] = model_info
-
-                    return models
-
-            except aiohttp.ClientError as e:
-                raise APIError(f"OpenAI API error: {e}")
-            except Exception as e:
-                raise APIError(f"Failed to fetch OpenAI models: {e}")
-
-    async def _fetch_anthropic_models(self) -> Dict[str, ModelInfo]:
-        """Fetch models from Anthropic API"""
-        # Anthropic doesn't provide a public models endpoint yet
-        # Return known Anthropic models with approximate pricing
-        return {
-            "claude-3-opus-20240229": ModelInfo(
-                id="claude-3-opus-20240229",
-                name="Claude 3 Opus",
-                provider="anthropic",
-                context_length=200000,
-                prompt_price=15.0,
-                completion_price=75.0,
-                description="Most powerful Claude 3 model",
-                tags=["reasoning", "analysis", "coding"],
-            ),
-            "claude-3-sonnet-20240229": ModelInfo(
-                id="claude-3-sonnet-20240229",
-                name="Claude 3 Sonnet",
-                provider="anthropic",
-                context_length=200000,
-                prompt_price=3.0,
-                completion_price=15.0,
-                description="Balanced Claude 3 model",
-                tags=["general", "reasoning", "coding"],
-            ),
-            "claude-sonnet-4-20240307": ModelInfo(
-                id="claude-sonnet-4-20240307",
-                name="Claude 3 Haiku",
-                provider="anthropic",
-                context_length=200000,
-                prompt_price=0.25,
-                completion_price=1.25,
-                description="Fastest Claude 3 model",
-                tags=["speed", "general", "cost-effective"],
-            ),
-        }
-
-    def _parse_openrouter_model(
-        self, model_data: Dict[str, Any]
-    ) -> Optional[ModelInfo]:
-        """Parse OpenRouter model data"""
-        try:
-            model_id = model_data.get("id")
-            if not model_id:
-                return None
-
-            pricing = model_data.get("pricing", {})
-
-            return ModelInfo(
-                id=model_id,
-                name=model_data.get("name", model_id),
-                provider="openrouter",
-                context_length=model_data.get("context_length", 0),
-                prompt_price=float(pricing.get("prompt", 0))
-                * 1_000_000,  # Convert to per 1M
-                completion_price=float(pricing.get("completion", 0))
-                * 1_000_000,  # Convert to per 1M
-                description=model_data.get("description"),
-                tags=model_data.get("tags", []),
-                is_available=model_data.get("available", True),
-            )
-        except (ValueError, TypeError):
-            return None
-
-    def _parse_openai_model(self, model_data: Dict[str, Any]) -> Optional[ModelInfo]:
-        """Parse OpenAI model data"""
-        try:
-            model_id = model_data.get("id")
-            if not model_id:
-                return None
-
-            # OpenAI pricing (approximate, should be updated regularly)
-            pricing_map = {
-                "gpt-4": {"prompt": 30.0, "completion": 60.0},
-                "gpt-4-32k": {"prompt": 60.0, "completion": 120.0},
-                "gpt-3.5-turbo": {"prompt": 0.5, "completion": 1.5},
-                "gpt-3.5-turbo-16k": {"prompt": 3.0, "completion": 4.0},
-            }
-
-            pricing = pricing_map.get(model_id, {"prompt": 2.0, "completion": 4.0})
-
-            return ModelInfo(
-                id=model_id,
-                name=model_data.get("id", model_id),
-                provider="openai",
-                context_length=self._get_openai_context_length(model_id),
-                prompt_price=pricing["prompt"],
-                completion_price=pricing["completion"],
-                description=f"OpenAI {model_id} model",
-                tags=["openai", "chat"] if "gpt" in model_id else ["openai"],
-            )
-        except (ValueError, TypeError):
-            return None
-
-    def _get_openai_context_length(self, model_id: str) -> int:
-        """Get context length for OpenAI models"""
-        context_map = {
-            "gpt-4": 8192,
-            "gpt-4-32k": 32768,
-            "gpt-3.5-turbo": 4096,
-            "gpt-3.5-turbo-16k": 16384,
-        }
-        return context_map.get(model_id, 4096)
-
-    def _get_fallback_models(self) -> Dict[str, ModelInfo]:
-        """Fallback models when API fetching fails"""
-        return {
-            "gpt-3.5-turbo": ModelInfo(
-                id="gpt-3.5-turbo",
-                name="GPT-3.5 Turbo",
-                provider="openai",
-                context_length=4096,
-                prompt_price=0.5,
-                completion_price=1.5,
-                description="OpenAI GPT-3.5 Turbo",
-                tags=["openai", "chat", "cost-effective"],
-            ),
-            "claude-sonnet-4": ModelInfo(
-                id="claude-sonnet-4",
-                name="Claude 3 Haiku",
-                provider="anthropic",
-                context_length=200000,
-                prompt_price=0.25,
-                completion_price=1.25,
-                description="Anthropic Claude 3 Haiku",
-                tags=["anthropic", "fast", "cost-effective"],
-            ),
-            "llama-2-70b": ModelInfo(
-                id="llama-2-70b",
-                name="Llama 2 70B",
-                provider="meta",
-                context_length=4096,
-                prompt_price=0.7,
-                completion_price=0.8,
-                description="Meta Llama 2 70B",
-                tags=["meta", "open-source", "reasoning"],
-            ),
-        }
-
-    # Query methods
-    def get_model(self, model_id: str) -> Optional[ModelInfo]:
-        """Get model by ID"""
-        return self.models.get(model_id)
-
-    def get_models_by_provider(self, provider: str) -> List[ModelInfo]:
-        """Get models by provider"""
-        return [model for model in self.models.values() if model.provider == provider]
-
-    def get_free_models(self) -> List[ModelInfo]:
-        """Get free models (no cost)"""
-        return [
-            model
-            for model in self.models.values()
-            if model.prompt_price == 0.0
-            and model.completion_price == 0.0
-            and model.is_available
-        ]
-
-    def get_budget_models(self, max_price_per_1m: float = 1.0) -> List[ModelInfo]:
-        """Get budget-friendly models"""
-        return [
-            model
-            for model in self.models.values()
-            if model.prompt_price <= max_price_per_1m and model.is_available
-        ]
-
-    def get_premium_models(self, min_price_per_1m: float = 10.0) -> List[ModelInfo]:
-        """Get premium models"""
-        return [
-            model
-            for model in self.models.values()
-            if model.prompt_price >= min_price_per_1m and model.is_available
-        ]
-
-    def search_models(self, query: str) -> List[ModelInfo]:
-        """Search models by name, description, or tags"""
-        query_lower = query.lower()
-        results = []
-
-        for model in self.models.values():
-            if not model.is_available:
-                continue
-
-            # Search in name
-            if query_lower in model.name.lower():
-                results.append(model)
-                continue
-
-            # Search in description
-            if query_lower in model.description.lower():
-                results.append(model)
-                continue
-
-            # Search in tags
-            if any(query_lower in tag.lower() for tag in model.tags):
-                results.append(model)
-                continue
-
-        return results
-
-    def get_cheapest_model(self, provider: Optional[str] = None) -> Optional[ModelInfo]:
-        """Get cheapest available model"""
-        available_models = [m for m in self.models.values() if m.is_available]
-
-        if provider:
-            available_models = [m for m in available_models if m.provider == provider]
-
-        if not available_models:
-            return None
-
-        return min(available_models, key=lambda m: m.prompt_price)
-
-    def estimate_cost(
-        self, model_id: str, input_tokens: int, output_tokens: int
-    ) -> Optional[float]:
-        """Estimate cost for model usage"""
-        model = self.get_model(model_id)
-        if not model:
-            return None
-
-        return model.estimate_cost(input_tokens, output_tokens)
-
-    def get_models_summary(self) -> Dict[str, Any]:
-        """Get summary of available models"""
-        if not self.models:
-            return {"error": "No models loaded"}
-
-        available_models = [m for m in self.models.values() if m.is_available]
-
-        # Provider breakdown
-        provider_counts = {}
-        for model in available_models:
-            provider_counts[model.provider] = provider_counts.get(model.provider, 0) + 1
-
-        # Price statistics
-        prices = [m.prompt_price for m in available_models if m.prompt_price > 0]
-
-        return {
-            "total_models": len(self.models),
-            "available_models": len(available_models),
-            "providers": provider_counts,
-            "free_models": len(self.get_free_models()),
-            "budget_models": len(self.get_budget_models()),
-            "premium_models": len(self.get_premium_models()),
-            "price_range": {
-                "min": min(prices) if prices else 0,
-                "max": max(prices) if prices else 0,
-                "avg": sum(prices) / len(prices) if prices else 0,
-            },
-            "last_updated": (
-                self.last_fetch_time.isoformat() if self.last_fetch_time else None
-            ),
-        }
-
-    def clear_cache(self) -> None:
-        """Clear all cached data"""
-        self.cache.clear()
-        self.models.clear()
-        self.last_fetch_time = None
-        logger.info("Models cache cleared")
-
-
-# Global cache instance
-_models_cache: Optional[ModelsCache] = None
-
-
-def get_models_cache(
-    openrouter_api_key: Optional[str] = None,
-    openai_api_key: Optional[str] = None,
-    anthropic_api_key: Optional[str] = None,
-) -> ModelsCache:
-    """Get global models cache instance"""
-    global _models_cache
-
-    if _models_cache is None:
-        _models_cache = ModelsCache(
-            openrouter_api_key=openrouter_api_key,
-            openai_api_key=openai_api_key,
-            anthropic_api_key=anthropic_api_key,
-        )
-
-    return _models_cache
-
-
-# Convenience functions
-async def get_available_models(force_refresh: bool = False) -> Dict[str, ModelInfo]:
-    """Get all available models"""
-    cache = get_models_cache()
-    return await cache.fetch_all_models(force_refresh=force_refresh)
-
-
-async def get_model_info(model_id: str) -> Optional[ModelInfo]:
-    """Get information for specific model"""
-    cache = get_models_cache()
-    await cache.fetch_all_models()
-    return cache.get_model(model_id)
-
-
-async def estimate_model_cost(
-    model_id: str, input_tokens: int, output_tokens: int
-) -> Optional[float]:
-    """Estimate cost for model usage"""
-    cache = get_models_cache()
-    await cache.fetch_all_models()
-    return cache.estimate_cost(model_id, input_tokens, output_tokens)
-
-
-async def find_cheapest_model(provider: Optional[str] = None) -> Optional[ModelInfo]:
-    """Find the cheapest available model"""
-    cache = get_models_cache()
-    await cache.fetch_all_models()
-    return cache.get_cheapest_model(provider=provider)