isa-model 0.4.0__py3-none-any.whl → 0.4.4__py3-none-any.whl

isa_model/inference/repositories/inference_repository.py

@@ -0,0 +1,828 @@
+"""
+Inference Repository - Data persistence layer for inference operations
+
+Provides standardized data access for inference requests, usage statistics, and history
+following the ISA Model architecture pattern.
+"""
+
+import logging
+import json
+import uuid
+from datetime import datetime, timezone, timedelta
+from typing import Dict, List, Optional, Any, Union
+from pathlib import Path
+from dataclasses import dataclass, asdict
+from enum import Enum
+
+try:
+    # Try to import Supabase for centralized data storage
+    from ...core.database.supabase_client import get_supabase_client
+    SUPABASE_AVAILABLE = True
+except ImportError:
+    SUPABASE_AVAILABLE = False
+
+logger = logging.getLogger(__name__)
+
+class InferenceStatus(str, Enum):
+    """Inference status enumeration"""
+    PENDING = "pending"
+    PROCESSING = "processing"
+    COMPLETED = "completed"
+    FAILED = "failed"
+    TIMEOUT = "timeout"
+    CANCELLED = "cancelled"
+
+class ServiceType(str, Enum):
+    """Service type enumeration"""
+    LLM = "llm"
+    VISION = "vision"
+    EMBEDDING = "embedding"
+    TTS = "tts"
+    STT = "stt"
+    IMAGE_GEN = "image_gen"
+    AUDIO = "audio"
+    RERANK = "rerank"
+    OCR = "ocr"
+
+@dataclass
+class InferenceRequest:
+    """Inference request record"""
+    request_id: str
+    service_type: str
+    model_id: str
+    provider: str
+    endpoint: str
+    request_data: Dict[str, Any]
+    status: str = InferenceStatus.PENDING
+    created_at: datetime = None
+    started_at: Optional[datetime] = None
+    completed_at: Optional[datetime] = None
+    user_id: Optional[str] = None
+    session_id: Optional[str] = None
+    ip_address: Optional[str] = None
+    user_agent: Optional[str] = None
+    response_data: Optional[Dict[str, Any]] = None
+    error_message: Optional[str] = None
+    execution_time_ms: Optional[int] = None
+    tokens_used: Optional[int] = None
+    cost_usd: Optional[float] = None
+    metadata: Optional[Dict[str, Any]] = None
+    
+    def __post_init__(self):
+        if self.created_at is None:
+            self.created_at = datetime.now(timezone.utc)
+
+@dataclass
+class UsageStatistics:
+    """Usage statistics record"""
+    stat_id: str
+    period_start: datetime
+    period_end: datetime
+    service_type: str
+    model_id: Optional[str] = None
+    provider: Optional[str] = None
+    user_id: Optional[str] = None
+    total_requests: int = 0
+    successful_requests: int = 0
+    failed_requests: int = 0
+    total_tokens: int = 0
+    total_cost_usd: float = 0.0
+    avg_response_time_ms: float = 0.0
+    p95_response_time_ms: float = 0.0
+    requests_per_hour: float = 0.0
+    error_rate: float = 0.0
+    created_at: datetime = None
+    
+    def __post_init__(self):
+        if self.created_at is None:
+            self.created_at = datetime.now(timezone.utc)
+
+@dataclass
+class ModelUsageSnapshot:
+    """Model usage snapshot for quick analytics"""
+    snapshot_id: str
+    model_id: str
+    provider: str
+    snapshot_time: datetime
+    hourly_requests: int = 0
+    daily_requests: int = 0
+    weekly_requests: int = 0
+    monthly_requests: int = 0
+    total_tokens_today: int = 0
+    total_cost_today: float = 0.0
+    avg_response_time_today: float = 0.0
+    success_rate_today: float = 100.0
+    last_used: Optional[datetime] = None
+    
+    def __post_init__(self):
+        if self.snapshot_time is None:
+            self.snapshot_time = datetime.now(timezone.utc)
+
+class InferenceRepository:
+    """
+    Repository for inference data persistence
+    
+    Supports multiple backend storage options:
+    1. Supabase (preferred for centralized storage)
+    2. Local file system (fallback for development)
+    3. In-memory storage (for testing)
+    """
+    
+    def __init__(self, storage_backend: str = "auto", **kwargs):
+        """
+        Initialize inference repository
+        
+        Args:
+            storage_backend: "supabase", "file", "memory", or "auto"
+            **kwargs: Backend-specific configuration
+        """
+        self.storage_backend = self._determine_backend(storage_backend)
+        self.config = kwargs
+        
+        # Initialize storage backend
+        if self.storage_backend == "supabase":
+            self._init_supabase()
+        elif self.storage_backend == "memory":
+            self._init_memory()
+        else:  # file system fallback
+            self._init_file_system()
+        
+        logger.info(f"Inference repository initialized with {self.storage_backend} backend")
+    
+    def _determine_backend(self, preference: str) -> str:
+        """Determine the best available storage backend"""
+        if preference == "supabase" and SUPABASE_AVAILABLE:
+            return "supabase"
+        elif preference in ["supabase", "file", "memory"]:
+            return preference
+        
+        # Auto-select best available backend
+        if SUPABASE_AVAILABLE:
+            return "supabase"
+        else:
+            return "file"
+    
+    def _init_supabase(self):
+        """Initialize Supabase backend"""
+        try:
+            self.supabase_client = get_supabase_client()
+            self._ensure_supabase_tables()
+            logger.info("Supabase backend initialized for inference")
+        except Exception as e:
+            logger.error(f"Failed to initialize Supabase backend: {e}")
+            self.storage_backend = "file"
+            self._init_file_system()
+    
+    def _init_file_system(self):
+        """Initialize file system backend"""
+        self.data_dir = Path(self.config.get("data_dir", "./inference_data"))
+        self.data_dir.mkdir(parents=True, exist_ok=True)
+        
+        # Create subdirectories
+        (self.data_dir / "requests").mkdir(exist_ok=True)
+        (self.data_dir / "statistics").mkdir(exist_ok=True)
+        (self.data_dir / "snapshots").mkdir(exist_ok=True)
+        
+        logger.info(f"File system backend initialized: {self.data_dir}")
+    
+    def _init_memory(self):
+        """Initialize in-memory backend for testing"""
+        self.requests = {}
+        self.statistics = {}
+        self.snapshots = {}
+        logger.info("In-memory backend initialized for inference")
+    
+    def _ensure_supabase_tables(self):
+        """Ensure required Supabase tables exist"""
+        try:
+            self.supabase_client.table("inference_requests").select("request_id").limit(1).execute()
+            self.supabase_client.table("usage_statistics").select("stat_id").limit(1).execute()
+            self.supabase_client.table("model_usage_snapshots").select("snapshot_id").limit(1).execute()
+        except Exception as e:
+            logger.warning(f"Some inference tables may not exist in Supabase: {e}")
+    
+    # Request Management Methods
+    
+    def create_inference_request(
+        self,
+        service_type: str,
+        model_id: str,
+        provider: str,
+        endpoint: str,
+        request_data: Dict[str, Any],
+        user_id: Optional[str] = None,
+        session_id: Optional[str] = None,
+        ip_address: Optional[str] = None,
+        user_agent: Optional[str] = None,
+        metadata: Optional[Dict[str, Any]] = None
+    ) -> str:
+        """Create a new inference request record"""
+        request_id = f"inf_{datetime.now().strftime('%Y%m%d_%H%M%S')}_{uuid.uuid4().hex[:8]}"
+        
+        request = InferenceRequest(
+            request_id=request_id,
+            service_type=service_type,
+            model_id=model_id,
+            provider=provider,
+            endpoint=endpoint,
+            request_data=request_data,
+            user_id=user_id,
+            session_id=session_id,
+            ip_address=ip_address,
+            user_agent=user_agent,
+            metadata=metadata
+        )
+        
+        if self.storage_backend == "supabase":
+            return self._create_request_supabase(request)
+        elif self.storage_backend == "memory":
+            return self._create_request_memory(request)
+        else:
+            return self._create_request_file(request)
+    
+    def update_inference_status(
+        self,
+        request_id: str,
+        status: str,
+        response_data: Optional[Dict[str, Any]] = None,
+        error_message: Optional[str] = None,
+        execution_time_ms: Optional[int] = None,
+        tokens_used: Optional[int] = None,
+        cost_usd: Optional[float] = None,
+        additional_updates: Optional[Dict[str, Any]] = None
+    ) -> bool:
+        """Update inference request status and results"""
+        updates = {"status": status}
+        
+        if status == InferenceStatus.PROCESSING:
+            updates["started_at"] = datetime.now(timezone.utc).isoformat()
+        elif status in [InferenceStatus.COMPLETED, InferenceStatus.FAILED, InferenceStatus.TIMEOUT]:
+            updates["completed_at"] = datetime.now(timezone.utc).isoformat()
+        
+        if response_data:
+            updates["response_data"] = response_data
+        if error_message:
+            updates["error_message"] = error_message
+        if execution_time_ms:
+            updates["execution_time_ms"] = execution_time_ms
+        if tokens_used:
+            updates["tokens_used"] = tokens_used
+        if cost_usd:
+            updates["cost_usd"] = cost_usd
+        
+        if additional_updates:
+            updates.update(additional_updates)
+        
+        if self.storage_backend == "supabase":
+            return self._update_request_supabase(request_id, updates)
+        elif self.storage_backend == "memory":
+            return self._update_request_memory(request_id, updates)
+        else:
+            return self._update_request_file(request_id, updates)
+    
+    def get_inference_request(self, request_id: str) -> Optional[InferenceRequest]:
+        """Get inference request by ID"""
+        if self.storage_backend == "supabase":
+            return self._get_request_supabase(request_id)
+        elif self.storage_backend == "memory":
+            return self._get_request_memory(request_id)
+        else:
+            return self._get_request_file(request_id)
+    
+    def list_recent_requests(
+        self,
+        service_type: Optional[str] = None,
+        model_id: Optional[str] = None,
+        user_id: Optional[str] = None,
+        status: Optional[str] = None,
+        hours: int = 24,
+        limit: int = 100
+    ) -> List[InferenceRequest]:
+        """List recent inference requests with optional filtering"""
+        if self.storage_backend == "supabase":
+            return self._list_requests_supabase(service_type, model_id, user_id, status, hours, limit)
+        elif self.storage_backend == "memory":
+            return self._list_requests_memory(service_type, model_id, user_id, status, hours, limit)
+        else:
+            return self._list_requests_file(service_type, model_id, user_id, status, hours, limit)
+    
+    # Usage Statistics Methods
+    
+    def record_usage_statistics(
+        self,
+        period_start: datetime,
+        period_end: datetime,
+        service_type: str,
+        model_id: Optional[str] = None,
+        provider: Optional[str] = None,
+        user_id: Optional[str] = None,
+        total_requests: int = 0,
+        successful_requests: int = 0,
+        failed_requests: int = 0,
+        total_tokens: int = 0,
+        total_cost_usd: float = 0.0,
+        avg_response_time_ms: float = 0.0,
+        p95_response_time_ms: float = 0.0,
+        requests_per_hour: float = 0.0,
+        error_rate: float = 0.0
+    ) -> str:
+        """Record usage statistics for a time period"""
+        stat_id = f"stat_{period_start.strftime('%Y%m%d_%H')}_{uuid.uuid4().hex[:6]}"
+        
+        stats = UsageStatistics(
+            stat_id=stat_id,
+            period_start=period_start,
+            period_end=period_end,
+            service_type=service_type,
+            model_id=model_id,
+            provider=provider,
+            user_id=user_id,
+            total_requests=total_requests,
+            successful_requests=successful_requests,
+            failed_requests=failed_requests,
+            total_tokens=total_tokens,
+            total_cost_usd=total_cost_usd,
+            avg_response_time_ms=avg_response_time_ms,
+            p95_response_time_ms=p95_response_time_ms,
+            requests_per_hour=requests_per_hour,
+            error_rate=error_rate
+        )
+        
+        if self.storage_backend == "supabase":
+            return self._record_stats_supabase(stats)
+        elif self.storage_backend == "memory":
+            return self._record_stats_memory(stats)
+        else:
+            return self._record_stats_file(stats)
+    
+    def get_usage_statistics(
+        self,
+        service_type: Optional[str] = None,
+        model_id: Optional[str] = None,
+        user_id: Optional[str] = None,
+        days: int = 7,
+        limit: int = 100
+    ) -> List[UsageStatistics]:
+        """Get usage statistics for specified period"""
+        if self.storage_backend == "supabase":
+            return self._get_stats_supabase(service_type, model_id, user_id, days, limit)
+        elif self.storage_backend == "memory":
+            return self._get_stats_memory(service_type, model_id, user_id, days, limit)
+        else:
+            return self._get_stats_file(service_type, model_id, user_id, days, limit)
+    
+    def get_aggregated_usage(
+        self,
+        service_type: Optional[str] = None,
+        model_id: Optional[str] = None,
+        user_id: Optional[str] = None,
+        days: int = 30
+    ) -> Dict[str, Any]:
+        """Get aggregated usage statistics"""
+        stats = self.get_usage_statistics(service_type, model_id, user_id, days, 1000)
+        
+        if not stats:
+            return {
+                "total_requests": 0,
+                "total_cost_usd": 0.0,
+                "total_tokens": 0,
+                "avg_response_time_ms": 0.0,
+                "success_rate": 100.0,
+                "period_days": days
+            }
+        
+        total_requests = sum(s.total_requests for s in stats)
+        total_successful = sum(s.successful_requests for s in stats)
+        total_cost = sum(s.total_cost_usd for s in stats)
+        total_tokens = sum(s.total_tokens for s in stats)
+        
+        # Weighted average for response time
+        weighted_response_times = [s.avg_response_time_ms * s.total_requests for s in stats if s.total_requests > 0]
+        avg_response_time = sum(weighted_response_times) / total_requests if total_requests > 0 else 0.0
+        
+        success_rate = (total_successful / total_requests * 100) if total_requests > 0 else 100.0
+        
+        return {
+            "total_requests": total_requests,
+            "successful_requests": total_successful,
+            "total_cost_usd": round(total_cost, 4),
+            "total_tokens": total_tokens,
+            "avg_response_time_ms": round(avg_response_time, 2),
+            "success_rate": round(success_rate, 2),
+            "period_days": days,
+            "stats_count": len(stats)
+        }
+    
+    # Model Usage Snapshots Methods
+    
+    def update_model_snapshot(
+        self,
+        model_id: str,
+        provider: str,
+        hourly_requests: int = 0,
+        daily_requests: int = 0,
+        weekly_requests: int = 0,
+        monthly_requests: int = 0,
+        total_tokens_today: int = 0,
+        total_cost_today: float = 0.0,
+        avg_response_time_today: float = 0.0,
+        success_rate_today: float = 100.0
+    ) -> str:
+        """Update or create model usage snapshot"""
+        snapshot_id = f"snap_{model_id}_{provider}_{datetime.now().strftime('%Y%m%d')}"
+        
+        snapshot = ModelUsageSnapshot(
+            snapshot_id=snapshot_id,
+            model_id=model_id,
+            provider=provider,
+            snapshot_time=datetime.now(timezone.utc),
+            hourly_requests=hourly_requests,
+            daily_requests=daily_requests,
+            weekly_requests=weekly_requests,
+            monthly_requests=monthly_requests,
+            total_tokens_today=total_tokens_today,
+            total_cost_today=total_cost_today,
+            avg_response_time_today=avg_response_time_today,
+            success_rate_today=success_rate_today,
+            last_used=datetime.now(timezone.utc)
+        )
+        
+        if self.storage_backend == "supabase":
+            return self._update_snapshot_supabase(snapshot)
+        elif self.storage_backend == "memory":
+            return self._update_snapshot_memory(snapshot)
+        else:
+            return self._update_snapshot_file(snapshot)
+    
+    def get_model_snapshots(
+        self,
+        model_id: Optional[str] = None,
+        provider: Optional[str] = None,
+        days: int = 7
+    ) -> List[ModelUsageSnapshot]:
+        """Get model usage snapshots"""
+        if self.storage_backend == "supabase":
+            return self._get_snapshots_supabase(model_id, provider, days)
+        elif self.storage_backend == "memory":
+            return self._get_snapshots_memory(model_id, provider, days)
+        else:
+            return self._get_snapshots_file(model_id, provider, days)
+    
+    def get_top_models(self, metric: str = "daily_requests", limit: int = 10) -> List[Dict[str, Any]]:
+        """Get top models by specified metric"""
+        snapshots = self.get_model_snapshots(days=1)  # Get latest snapshots
+        
+        if not snapshots:
+            return []
+        
+        # Sort by the specified metric
+        valid_metrics = ["hourly_requests", "daily_requests", "weekly_requests", "monthly_requests", 
+                        "total_tokens_today", "total_cost_today"]
+        
+        if metric not in valid_metrics:
+            metric = "daily_requests"
+        
+        sorted_snapshots = sorted(
+            snapshots, 
+            key=lambda x: getattr(x, metric, 0), 
+            reverse=True
+        )[:limit]
+        
+        return [
+            {
+                "model_id": s.model_id,
+                "provider": s.provider,
+                "metric_value": getattr(s, metric, 0),
+                "daily_requests": s.daily_requests,
+                "total_cost_today": s.total_cost_today,
+                "success_rate_today": s.success_rate_today,
+                "last_used": s.last_used.isoformat() if s.last_used else None
+            }
+            for s in sorted_snapshots
+        ]
+    
+    # Cleanup and Maintenance Methods
+    
+    def cleanup_old_requests(self, days: int = 30) -> int:
+        """Clean up old inference requests"""
+        cutoff_date = datetime.now(timezone.utc) - timedelta(days=days)
+        
+        if self.storage_backend == "supabase":
+            return self._cleanup_requests_supabase(cutoff_date)
+        elif self.storage_backend == "memory":
+            return self._cleanup_requests_memory(cutoff_date)
+        else:
+            return self._cleanup_requests_file(cutoff_date)
+    
+    def cleanup_old_statistics(self, days: int = 90) -> int:
+        """Clean up old usage statistics"""
+        cutoff_date = datetime.now(timezone.utc) - timedelta(days=days)
+        
+        if self.storage_backend == "supabase":
+            return self._cleanup_stats_supabase(cutoff_date)
+        elif self.storage_backend == "memory":
+            return self._cleanup_stats_memory(cutoff_date)
+        else:
+            return self._cleanup_stats_file(cutoff_date)
+    
+    # Backend-specific implementations
+    
+    def _create_request_file(self, request: InferenceRequest) -> str:
+        """Create request in file system"""
+        try:
+            request_file = self.data_dir / "requests" / f"{request.request_id}.json"
+            request_data = asdict(request)
+            
+            # Convert datetime objects to ISO strings
+            for key in ['created_at', 'started_at', 'completed_at']:
+                if request_data[key] and isinstance(request_data[key], datetime):
+                    request_data[key] = request_data[key].isoformat()
+            
+            with open(request_file, 'w') as f:
+                json.dump(request_data, f, indent=2, ensure_ascii=False)
+            
+            return request.request_id
+        except Exception as e:
+            logger.error(f"Failed to create request in file system: {e}")
+            raise
+    
+    def _create_request_memory(self, request: InferenceRequest) -> str:
+        """Create request in memory"""
+        self.requests[request.request_id] = request
+        return request.request_id
+    
+    def _update_request_file(self, request_id: str, updates: Dict[str, Any]) -> bool:
+        """Update request in file system"""
+        try:
+            request_file = self.data_dir / "requests" / f"{request_id}.json"
+            if not request_file.exists():
+                return False
+            
+            with open(request_file, 'r') as f:
+                request_data = json.load(f)
+            
+            request_data.update(updates)
+            
+            with open(request_file, 'w') as f:
+                json.dump(request_data, f, indent=2, ensure_ascii=False)
+            
+            return True
+        except Exception as e:
+            logger.error(f"Failed to update request in file system: {e}")
+            return False
+    
+    def _update_request_memory(self, request_id: str, updates: Dict[str, Any]) -> bool:
+        """Update request in memory"""
+        if request_id not in self.requests:
+            return False
+        
+        request_dict = asdict(self.requests[request_id])
+        request_dict.update(updates)
+        
+        # Convert datetime strings back to datetime objects if needed
+        for key in ['created_at', 'started_at', 'completed_at']:
+            if key in request_dict and isinstance(request_dict[key], str):
+                request_dict[key] = datetime.fromisoformat(request_dict[key])
+        
+        self.requests[request_id] = InferenceRequest(**request_dict)
+        return True
+    
+    def _get_request_file(self, request_id: str) -> Optional[InferenceRequest]:
+        """Get request from file system"""
+        try:
+            request_file = self.data_dir / "requests" / f"{request_id}.json"
+            if not request_file.exists():
+                return None
+            
+            with open(request_file, 'r') as f:
+                request_data = json.load(f)
+            
+            # Convert ISO strings back to datetime objects
+            for key in ['created_at', 'started_at', 'completed_at']:
+                if request_data[key]:
+                    request_data[key] = datetime.fromisoformat(request_data[key])
+            
+            return InferenceRequest(**request_data)
+        except Exception as e:
+            logger.error(f"Failed to get request from file system: {e}")
+            return None
+    
+    def _get_request_memory(self, request_id: str) -> Optional[InferenceRequest]:
+        """Get request from memory"""
+        return self.requests.get(request_id)
+    
+    def _list_requests_file(
+        self, service_type: Optional[str], model_id: Optional[str], 
+        user_id: Optional[str], status: Optional[str], hours: int, limit: int
+    ) -> List[InferenceRequest]:
+        """List requests from file system"""
+        try:
+            requests = []
+            requests_dir = self.data_dir / "requests"
+            cutoff_time = datetime.now(timezone.utc) - timedelta(hours=hours)
+            
+            for request_file in requests_dir.glob("*.json"):
+                with open(request_file, 'r') as f:
+                    request_data = json.load(f)
+                
+                # Convert datetime fields
+                for key in ['created_at', 'started_at', 'completed_at']:
+                    if request_data[key]:
+                        request_data[key] = datetime.fromisoformat(request_data[key])
+                
+                request = InferenceRequest(**request_data)
+                
+                # Apply filters
+                if request.created_at < cutoff_time:
+                    continue
+                if service_type and request.service_type != service_type:
+                    continue
+                if model_id and request.model_id != model_id:
+                    continue
+                if user_id and request.user_id != user_id:
+                    continue
+                if status and request.status != status:
+                    continue
+                
+                requests.append(request)
+                
+                if len(requests) >= limit:
+                    break
+            
+            return sorted(requests, key=lambda x: x.created_at, reverse=True)
+        except Exception as e:
+            logger.error(f"Failed to list requests from file system: {e}")
+            return []
+    
+    def _list_requests_memory(
+        self, service_type: Optional[str], model_id: Optional[str], 
+        user_id: Optional[str], status: Optional[str], hours: int, limit: int
+    ) -> List[InferenceRequest]:
+        """List requests from memory"""
+        cutoff_time = datetime.now(timezone.utc) - timedelta(hours=hours)
+        requests = []
+        
+        for request in self.requests.values():
+            # Apply filters
+            if request.created_at < cutoff_time:
+                continue
+            if service_type and request.service_type != service_type:
+                continue
+            if model_id and request.model_id != model_id:
+                continue
+            if user_id and request.user_id != user_id:
+                continue
+            if status and request.status != status:
+                continue
+            
+            requests.append(request)
+            
+            if len(requests) >= limit:
+                break
+        
+        return sorted(requests, key=lambda x: x.created_at, reverse=True)
+    
+    # Simplified placeholder implementations for statistics and snapshots
+    def _record_stats_file(self, stats: UsageStatistics) -> str:
+        """Record statistics in file system"""
+        try:
+            stats_file = self.data_dir / "statistics" / f"{stats.stat_id}.json"
+            stats_data = asdict(stats)
+            
+            # Convert datetime objects to ISO strings
+            for key in ['period_start', 'period_end', 'created_at']:
+                if stats_data[key] and isinstance(stats_data[key], datetime):
+                    stats_data[key] = stats_data[key].isoformat()
+            
+            with open(stats_file, 'w') as f:
+                json.dump(stats_data, f, indent=2, ensure_ascii=False)
+            
+            return stats.stat_id
+        except Exception as e:
+            logger.error(f"Failed to record statistics in file system: {e}")
+            raise
+    
+    def _record_stats_memory(self, stats: UsageStatistics) -> str:
+        """Record statistics in memory"""
+        self.statistics[stats.stat_id] = stats
+        return stats.stat_id
+    
+    def _update_snapshot_file(self, snapshot: ModelUsageSnapshot) -> str:
+        """Update snapshot in file system"""
+        try:
+            snapshot_file = self.data_dir / "snapshots" / f"{snapshot.snapshot_id}.json"
+            snapshot_data = asdict(snapshot)
+            
+            # Convert datetime objects to ISO strings
+            for key in ['snapshot_time', 'last_used']:
+                if snapshot_data[key] and isinstance(snapshot_data[key], datetime):
+                    snapshot_data[key] = snapshot_data[key].isoformat()
+            
+            with open(snapshot_file, 'w') as f:
+                json.dump(snapshot_data, f, indent=2, ensure_ascii=False)
+            
+            return snapshot.snapshot_id
+        except Exception as e:
+            logger.error(f"Failed to update snapshot in file system: {e}")
+            raise
+    
+    def _update_snapshot_memory(self, snapshot: ModelUsageSnapshot) -> str:
+        """Update snapshot in memory"""
+        self.snapshots[snapshot.snapshot_id] = snapshot
+        return snapshot.snapshot_id
+    
+    # Cleanup implementations
+    def _cleanup_requests_file(self, cutoff_date: datetime) -> int:
+        """Cleanup old requests from file system"""
+        count = 0
+        try:
+            requests_dir = self.data_dir / "requests"
+            for request_file in requests_dir.glob("*.json"):
+                if request_file.stat().st_mtime < cutoff_date.timestamp():
+                    request_file.unlink()
+                    count += 1
+        except Exception as e:
+            logger.error(f"Failed to cleanup requests from file system: {e}")
+        return count
+    
+    def _cleanup_requests_memory(self, cutoff_date: datetime) -> int:
+        """Cleanup old requests from memory"""
+        count = 0
+        to_remove = []
+        for request_id, request in self.requests.items():
+            if request.created_at < cutoff_date:
+                to_remove.append(request_id)
+        
+        for request_id in to_remove:
+            del self.requests[request_id]
+            count += 1
+        
+        return count
+    
+    # Placeholder implementations for Supabase backend
+    def _create_request_supabase(self, request: InferenceRequest) -> str:
+        return request.request_id  # Implementation needed
+    
+    def _update_request_supabase(self, request_id: str, updates: Dict[str, Any]) -> bool:
+        return False  # Implementation needed
+    
+    def _get_request_supabase(self, request_id: str) -> Optional[InferenceRequest]:
+        return None  # Implementation needed
+    
+    def _list_requests_supabase(self, service_type, model_id, user_id, status, hours, limit) -> List[InferenceRequest]:
+        return []  # Implementation needed
+    
+    def _record_stats_supabase(self, stats: UsageStatistics) -> str:
+        return stats.stat_id  # Implementation needed
+    
+    def _get_stats_supabase(self, service_type, model_id, user_id, days, limit) -> List[UsageStatistics]:
+        return []  # Implementation needed
+    
+    def _get_stats_memory(self, service_type, model_id, user_id, days, limit) -> List[UsageStatistics]:
+        return list(self.statistics.values())[:limit]
+    
+    def _get_stats_file(self, service_type, model_id, user_id, days, limit) -> List[UsageStatistics]:
+        return []  # Implementation needed
+    
+    def _update_snapshot_supabase(self, snapshot: ModelUsageSnapshot) -> str:
+        return snapshot.snapshot_id  # Implementation needed
+    
+    def _get_snapshots_supabase(self, model_id, provider, days) -> List[ModelUsageSnapshot]:
+        return []  # Implementation needed
+    
+    def _get_snapshots_memory(self, model_id, provider, days) -> List[ModelUsageSnapshot]:
+        return list(self.snapshots.values())
+    
+    def _get_snapshots_file(self, model_id, provider, days) -> List[ModelUsageSnapshot]:
+        return []  # Implementation needed
+    
+    def _cleanup_requests_supabase(self, cutoff_date: datetime) -> int:
+        return 0  # Implementation needed
+    
+    def _cleanup_stats_supabase(self, cutoff_date: datetime) -> int:
+        return 0  # Implementation needed
+    
+    def _cleanup_stats_memory(self, cutoff_date: datetime) -> int:
+        count = 0
+        to_remove = []
+        for stat_id, stat in self.statistics.items():
+            if stat.created_at < cutoff_date:
+                to_remove.append(stat_id)
+        
+        for stat_id in to_remove:
+            del self.statistics[stat_id]
+            count += 1
+        
+        return count
+    
+    def _cleanup_stats_file(self, cutoff_date: datetime) -> int:
+        count = 0
+        try:
+            stats_dir = self.data_dir / "statistics"
+            for stats_file in stats_dir.glob("*.json"):
+                if stats_file.stat().st_mtime < cutoff_date.timestamp():
+                    stats_file.unlink()
+                    count += 1
+        except Exception as e:
+            logger.error(f"Failed to cleanup statistics from file system: {e}")
+        return count