core/arf_adapter.py

"""
ARF Adapter Pattern for clean integration with real or mock ARF
"""
from abc import ABC, abstractmethod
from typing import Dict, Any, List, Optional
import asyncio
import logging
from config.settings import settings, ARFMode

logger = logging.getLogger(__name__)


class ARFAdapter(ABC):
    """Abstract adapter for ARF integration"""
    
    @abstractmethod
    async def detect(self, metrics: Dict[str, Any]) -> Dict[str, Any]:
        """Detect anomalies in metrics"""
        pass
    
    @abstractmethod
    async def recall(self, incident: Dict[str, Any]) -> List[Dict[str, Any]]:
        """Recall similar incidents from memory"""
        pass
    
    @abstractmethod
    async def decide(self, incident: Dict[str, Any], context: Dict[str, Any]) -> Dict[str, Any]:
        """Generate healing intent"""
        pass
    
    @abstractmethod
    async def analyze(self, scenario_name: str, scenario_data: Dict[str, Any]) -> Dict[str, Any]:
        """Complete analysis pipeline"""
        pass


class MockARFAdapter(ARFAdapter):
    """Mock ARF implementation for demo mode"""
    
    def __init__(self):
        logger.info("Initializing MockARFAdapter")
        # Lazy imports to avoid circular dependencies
        self._simulate_arf_analysis = None
        self._run_rag_similarity_search = None
        self._create_mock_healing_intent = None
        self._calculate_pattern_confidence = None
    
    def _import_mock_functions(self):
        """Lazy import of mock functions"""
        if self._simulate_arf_analysis is None:
            from demo.mock_arf import (
                simulate_arf_analysis,
                run_rag_similarity_search,
                create_mock_healing_intent,
                calculate_pattern_confidence
            )
            self._simulate_arf_analysis = simulate_arf_analysis
            self._run_rag_similarity_search = run_rag_similarity_search
            self._create_mock_healing_intent = create_mock_healing_intent
            self._calculate_pattern_confidence = calculate_pattern_confidence
    
    async def detect(self, metrics: Dict[str, Any]) -> Dict[str, Any]:
        """Mock anomaly detection"""
        self._import_mock_functions()
        
        # Simulate processing time
        await asyncio.sleep(0.1)
        
        result = self._simulate_arf_analysis({"metrics": metrics})
        result["detection_method"] = "mock_ml_algorithm"
        result["confidence"] = 0.987  # 98.7%
        
        return result
    
    async def recall(self, incident: Dict[str, Any]) -> List[Dict[str, Any]]:
        """Mock RAG similarity search"""
        self._import_mock_functions()
        
        # Simulate processing time
        await asyncio.sleep(0.2)
        
        similar = self._run_rag_similarity_search(incident)
        
        # Enhance with additional metadata
        for item in similar:
            item["source"] = "mock_rag_memory"
            item["retrieval_time"] = "45ms"
        
        return similar
    
    async def decide(self, incident: Dict[str, Any], context: Dict[str, Any]) -> Dict[str, Any]:
        """Mock decision making"""
        self._import_mock_functions()
        
        # Get similar incidents from context or recall them
        similar = context.get("similar_incidents")
        if not similar:
            similar = await self.recall(incident)
        
        # Calculate confidence
        confidence = self._calculate_pattern_confidence(incident, similar)
        
        # Generate healing intent
        intent = self._create_mock_healing_intent(incident, similar, confidence)
        
        # Add safety check
        intent["safety_checks"] = {
            "blast_radius": "2 services",
            "business_hours": "compliant",
            "rollback_plan": "available",
            "approval_required": True
        }
        
        return intent
    
    async def analyze(self, scenario_name: str, scenario_data: Dict[str, Any]) -> Dict[str, Any]:
        """Complete mock analysis pipeline"""
        logger.info(f"Starting mock analysis for: {scenario_name}")
        
        # Step 1: Detection
        detection_result = await self.detect(scenario_data.get("metrics", {}))
        
        # Step 2: Recall
        recall_result = await self.recall(scenario_data)
        
        # Step 3: Decision
        decision_result = await self.decision(
            scenario_data,
            {"similar_incidents": recall_result}
        )
        
        return {
            "scenario": scenario_name,
            "detection": detection_result,
            "recall": recall_result,
            "decision": decision_result,
            "overall_confidence": decision_result.get("confidence", 0.85),
            "processing_time_ms": 450,
            "agents_executed": ["detection", "recall", "decision"]
        }


class RealARFAdapter(ARFAdapter):
    """Real ARF integration (requires agentic-reliability-framework package)"""
    
    def __init__(self, api_key: Optional[str] = None):
        logger.info("Initializing RealARFAdapter")
        
        try:
            from agentic_reliability_framework import ARFClient
            self.client = ARFClient(api_key=api_key or settings.arf_api_key)
            self._available = True
        except ImportError as e:
            logger.error(f"Failed to import ARF package: {e}")
            self._available = False
            raise RuntimeError(
                "Real ARF integration requires 'agentic-reliability-framework' package. "
                "Install with: pip install agentic-reliability-framework"
            )
    
    async def detect(self, metrics: Dict[str, Any]) -> Dict[str, Any]:
        """Real anomaly detection"""
        if not self._available:
            raise RuntimeError("ARF client not available")
        
        try:
            # Assuming async API
            result = await self.client.detect_anomaly_async(metrics)
            return result
        except AttributeError:
            # Fallback to sync if async not available
            result = self.client.detect_anomaly(metrics)
            return result
    
    async def recall(self, incident: Dict[str, Any]) -> List[Dict[str, Any]]:
        """Real RAG similarity search"""
        if not self._available:
            raise RuntimeError("ARF client not available")
        
        try:
            # Assuming async API
            result = await self.client.recall_similar_async(incident)
            return result
        except AttributeError:
            # Fallback to sync
            result = self.client.recall_similar(incident)
            return result
    
    async def decide(self, incident: Dict[str, Any], context: Dict[str, Any]) -> Dict[str, Any]:
        """Real decision making"""
        if not self._available:
            raise RuntimeError("ARF client not available")
        
        try:
            # Assuming async API
            result = await self.client.generate_intent_async(incident, context)
            return result
        except AttributeError:
            # Fallback to sync
            result = self.client.generate_intent(incident, context)
            return result
    
    async def analyze(self, scenario_name: str, scenario_data: Dict[str, Any]) -> Dict[str, Any]:
        """Complete real analysis pipeline"""
        logger.info(f"Starting real analysis for: {scenario_name}")
        
        # Run agents in parallel where possible
        detection_task = asyncio.create_task(self.detect(scenario_data.get("metrics", {})))
        recall_task = asyncio.create_task(self.recall(scenario_data))
        
        detection_result, recall_result = await asyncio.gather(detection_task, recall_task)
        
        # Decision depends on recall results
        decision_result = await self.decision(
            scenario_data,
            {"similar_incidents": recall_result}
        )
        
        return {
            "scenario": scenario_name,
            "detection": detection_result,
            "recall": recall_result,
            "decision": decision_result,
            "overall_confidence": decision_result.get("confidence", 0.85),
            "processing_time_ms": 250,  # Real system should be faster
            "agents_executed": ["detection", "recall", "decision"]
        }


def get_arf_adapter() -> ARFAdapter:
    """
    Factory function to get appropriate ARF adapter based on settings
    
    Returns:
        ARFAdapter instance
    """
    if settings.arf_mode == ARFMode.DEMO or settings.use_mock_arf:
        logger.info("Using MockARFAdapter (demo mode)")
        return MockARFAdapter()
    elif settings.arf_mode == ARFMode.OSS:
        logger.info("Using RealARFAdapter (OSS mode)")
        return RealARFAdapter()
    elif settings.arf_mode == ARFMode.ENTERPRISE:
        logger.info("Using RealARFAdapter (Enterprise mode)")
        return RealARFAdapter()
    else:
        logger.warning("Unknown ARF mode, falling back to mock")
        return MockARFAdapter()


# Async helper for easy integration
async def analyze_scenario_async(scenario_name: str, scenario_data: Dict[str, Any]) -> Dict[str, Any]:
    """Convenience function for async scenario analysis"""
    adapter = get_arf_adapter()
    return await adapter.analyze(scenario_name, scenario_data)


# Sync wrapper for compatibility
def analyze_scenario_sync(scenario_name: str, scenario_data: Dict[str, Any]) -> Dict[str, Any]:
    """Sync wrapper for scenario analysis"""
    import asyncio
    
    async def _analyze():
        return await analyze_scenario_async(scenario_name, scenario_data)
    
    try:
        loop = asyncio.get_running_loop()
        # Already in async context, create task
        return asyncio.create_task(_analyze())
    except RuntimeError:
        # Create new event loop
        loop = asyncio.new_event_loop()
        asyncio.set_event_loop(loop)
        try:
            return loop.run_until_complete(_analyze())
        finally:
            loop.close()