mcp_server/tools/spot_finder_tool.py

"""
Main Surf Spot Finder Tool - Complete Workflow Orchestration.

This module provides the primary MCP tool for finding and ranking surf spots.
It orchestrates the complete workflow from user input to AI-powered recommendations:

1. Location resolution (address/coordinates)
2. Nearby spot discovery (distance filtering)
3. Real-time condition analysis (wave/wind data)
4. Multi-factor evaluation and scoring
5. AI-powered reasoning and explanations

The tool integrates multiple data sources:
    - MCP resources for surf spot database
    - Stormglass API for marine conditions
    - Nominatim for geocoding
    - LLM providers for natural language reasoning

Example:
    >>> finder = SurfSpotFinder()
    >>> input_data = SpotFinderInput(
    ...     user_location="Málaga, Spain",
    ...     max_distance_km=50,
    ...     top_n=3,
    ...     user_preferences={"skill_level": "intermediate"}
    ... )
    >>> result = await finder.run(input_data)
    >>> print(f"Found {len(result.spots)} spots")
    
Author: Surf Spot Finder Team
License: MIT
"""

import json
import os
from typing import Dict, Any, List, Optional
from pydantic import BaseModel, Field
from geopy.distance import geodesic
import asyncio
import logging

from .location_tool import LocationTool, LocationInput
from .stormglass_tool import create_stormglass_tool
from .surf_eval_tool import SurfEvaluatorTool
from .llm_agent_tool import SurfLLMAgent, LLMAgentInput

logger = logging.getLogger(__name__)


class SpotFinderInput(BaseModel):
    """Input schema for the surf spot finder tool.
    
    Attributes:
        user_location: User's location as address, city name, or coordinates.
        max_distance_km: Maximum search radius in kilometers (default: 50).
        top_n: Number of top-ranked spots to return (default: 3).
        user_preferences: Dict containing skill_level, board_type, etc.
    """
    user_location: str = Field(description="User's location (name, address, or coordinates)")
    max_distance_km: float = Field(default=50, description="Maximum distance to search for spots (km)")
    top_n: int = Field(default=3, description="Number of top spots to return")
    user_preferences: Dict[str, Any] = Field(default_factory=dict, description="User surfing preferences")


class SpotFinderOutput(BaseModel):
    """Output schema for surf spot finder results.
    
    Attributes:
        success: Whether the operation completed successfully.
        user_location: Resolved coordinates as {"lat": float, "lon": float}.
        spots: List of ranked surf spots with scores and conditions.
        ai_summary: Brief AI-generated summary of recommendations.
        ai_reasoning: Detailed AI analysis and explanations.
        error: Error message if success is False.
    """
    success: bool
    user_location: Optional[Dict[str, float]] = None
    spots: List[Dict[str, Any]] = []
    ai_summary: str = ""
    ai_reasoning: str = ""
    error: str = ""


class SurfSpotFinder:
    """Main orchestration tool for finding optimal surf spots.
    
    This class coordinates the complete surf recommendation workflow by
    integrating location services, marine data APIs, evaluation algorithms,
    and AI reasoning to provide ranked surf spot recommendations.
    
    The workflow:
        1. Resolves user location to coordinates
        2. Filters surf spots by distance
        3. Fetches real-time wave conditions
        4. Evaluates each spot using multi-factor scoring
        5. Generates AI-powered analysis and explanations
    
    Attributes:
        name: Tool identifier for MCP registration.
        description: Human-readable tool description.
        location_tool: Service for address/coordinate resolution.
        stormglass_tool: API client for marine condition data.
        evaluator: Surf condition evaluation algorithm.
        llm_agent: AI reasoning and natural language generation.
        spots_db: Cached surf spot database from MCP resources.
    
    Example:
        >>> finder = SurfSpotFinder()
        >>> input_data = SpotFinderInput(user_location="Lisbon")
        >>> result = await finder.run(input_data)
        >>> print(f"Best spot: {result.spots[0]['name']}")
    """
    
    name = "find_surf_spots"
    description = "Find and rank the best surf spots near a given location based on current conditions"
    
    def __init__(self):
        self.location_tool = LocationTool()
        self.stormglass_tool = create_stormglass_tool()["function"]
        self.evaluator = SurfEvaluatorTool()
        self.llm_agent = SurfLLMAgent()
        self.spots_db = self._load_surf_spots()
    
    def _load_surf_spots(self) -> List[Dict[str, Any]]:
        """
        Load surf spots database via MCP resources with fallback.
        
        Attempts to load surf spots using MCP resource primitive first,
        then falls back to direct file I/O if MCP resources are unavailable.
        This ensures reliability across different execution contexts.
        
        MCP Resource Path: surf://spots/database
        Fallback Path: ../data/surf_spots.json
        
        Returns:
            List[Dict[str, Any]]: List of surf spot dictionaries
        
        Note:
            Import is done locally to avoid circular dependency issues.
            Uses fallback to file I/O if async context is unavailable.
        """
        try:
            # Try MCP resources first (preferred method)
            import asyncio
            
            # Local import to avoid circular dependency issues
            try:
                from mcp_server.mcp_server import read_resource
            except ImportError:
                import sys
                from pathlib import Path
                sys.path.insert(0, str(Path(__file__).parent.parent))
                from mcp_server import read_resource
            
            # Check if we're already in an async context
            try:
                loop = asyncio.get_running_loop()
                # Already in async context - can't create new loop
                # Fall back to file I/O
                raise RuntimeError("Already in async context, using fallback")
            except RuntimeError:
                # Not in async context - safe to create new loop
                loop = asyncio.new_event_loop()
                asyncio.set_event_loop(loop)
                
                # Read from MCP resource
                content = loop.run_until_complete(read_resource("surf://spots/database"))
                loop.close()
                
                data = json.loads(content)
                spots = data.get("spots", [])
                
                logger.info(f"✅ Loaded {len(spots)} surf spots via MCP resources")
                return spots
            
        except Exception as e:
            # MCP resources unavailable - use fallback
            logger.warning(f"⚠️  MCP resources not available in this context, using file I/O fallback")
            
            # Fallback: Direct file I/O for reliability
            try:
                spots_path = os.path.join(os.path.dirname(__file__), "..", "data", "surf_spots.json")
                with open(spots_path, 'r') as f:
                    data = json.load(f)
                    spots = data["spots"]
                    logger.info(f"✅ Loaded {len(spots)} surf spots via fallback file I/O")
                    return spots
            except Exception as fallback_error:
                # Total failure - return empty list to prevent crash
                logger.error(f"❌ Both MCP and fallback failed: {fallback_error}")
                return []


    def find_nearby_spots(self, user_lat: float, user_lon: float, max_distance_km: float) -> List[Dict[str, Any]]:
        """Find surf spots within specified distance of user location.
        
        Uses haversine formula to calculate spherical distances between
        user coordinates and each surf spot in the database.
        
        Args:
            user_lat: User's latitude in decimal degrees.
            user_lon: User's longitude in decimal degrees. 
            max_distance_km: Maximum search radius in kilometers.
            
        Returns:
            List of surf spots within radius, sorted by distance.
            Each spot includes original data plus distance_km field.
        """
        nearby_spots = []
        user_location = (user_lat, user_lon)
        
        for spot in self.spots_db:
            spot_location = (spot["latitude"], spot["longitude"])
            distance = geodesic(user_location, spot_location).kilometers
            
            if distance <= max_distance_km:
                spot_with_distance = spot.copy()
                spot_with_distance["distance_km"] = round(distance, 1)
                nearby_spots.append(spot_with_distance)
        
        # Sort by distance
        nearby_spots.sort(key=lambda x: x["distance_km"])
        return nearby_spots
    
    async def get_spot_conditions(self, spot: Dict[str, Any]) -> Optional[Dict[str, Any]]:
        """Get current wave conditions for a specific surf spot.
        
        Fetches real-time marine data including wave height, direction,
        period, wind speed/direction, and tide information.
        
        Args:
            spot: Surf spot dictionary with latitude/longitude.
            
        Returns:
            Dict containing current conditions, or None if fetch fails.
            Includes wave_height, wave_direction, wind_speed, etc.
        """
        try:
            # Use the Stormglass tool to get wave data
            from .stormglass_tool import WaveDataInput
            wave_input = WaveDataInput(lat=spot["latitude"], lon=spot["longitude"])
            conditions = self.stormglass_tool(wave_input)
            # Convert to dict for compatibility
            if hasattr(conditions, 'dict'):
                return conditions.dict()
            elif hasattr(conditions, '__dict__'):
                return conditions.__dict__
            return conditions
        except Exception as e:
            logger.error(f"Failed to get conditions for {spot['name']}: {e}")
            return None
    
    async def evaluate_spot(self, spot: Dict[str, Any], conditions: Dict[str, Any], user_prefs: Dict[str, Any]) -> Dict[str, Any]:
        """Evaluate a single surf spot using current conditions and user preferences.
        
        Applies multi-factor scoring algorithm considering wave conditions,
        wind analysis, swell direction, and skill compatibility.
        
        Args:
            spot: Surf spot data including location and characteristics.
            conditions: Current wave/wind conditions from marine APIs.
            user_prefs: User preferences including skill level, board type.
            
        Returns:
            Dict containing evaluated spot with score, explanation, and
            breakdown of individual scoring factors.
        """
        evaluation = await self.evaluator.run({
            "spot": spot,
            "conditions": conditions,
            "prefs": user_prefs
        })
        
        return {
            "id": spot["id"],
            "name": spot["name"],
            "location": f"{spot['location']}, {spot['region']}",
            "latitude": spot["latitude"],
            "longitude": spot["longitude"],
            "distance_km": spot["distance_km"],
            "score": evaluation["score"],
            "explanation": evaluation["explanation"],
            "breakdown": evaluation["breakdown"],
            "conditions": conditions,
            "characteristics": spot["characteristics"]
        }
    
    async def run(self, input_data: SpotFinderInput) -> SpotFinderOutput:
        """Execute the complete surf spot finding workflow.
        
        This is the main entry point that orchestrates all steps of the
        surf recommendation process from user input to final rankings.
        
        Args:
            input_data: User request containing location, preferences, and filters.
            
        Returns:
            Complete results including ranked spots, AI analysis, and metadata.
            
        Raises:
            No exceptions - all errors are captured in SpotFinderOutput.error
            for graceful degradation and user-friendly error messages.
        """
        try:
            # Step 1: Resolve user location
            logger.info(f"Resolving location: {input_data.user_location}")
            location_result = self.location_tool.run(
                LocationInput(location_query=input_data.user_location)
            )
            
            if not location_result.success:
                return SpotFinderOutput(
                    success=False,
                    error=f"Could not resolve location: {location_result.error}"
                )
            
            user_coords = location_result.coordinates
            user_lat, user_lon = user_coords["lat"], user_coords["lon"]
            logger.info(f"User location resolved to: {user_lat}, {user_lon}")
            
            # Step 2: Find nearby surf spots
            nearby_spots = self.find_nearby_spots(user_lat, user_lon, input_data.max_distance_km)
            
            if not nearby_spots:
                return SpotFinderOutput(
                    success=False,
                    error=f"No surf spots found within {input_data.max_distance_km}km of {input_data.user_location}"
                )
            
            logger.info(f"Found {len(nearby_spots)} nearby spots")
            
            # Step 3: Get conditions and evaluate each spot
            evaluated_spots = []
            
            # Process spots concurrently for speed
            semaphore = asyncio.Semaphore(5)  # Limit concurrent API calls
            
            async def process_spot(spot):
                async with semaphore:
                    conditions = await self.get_spot_conditions(spot)
                    if conditions:
                        return await self.evaluate_spot(spot, conditions, input_data.user_preferences)
                    return None
            
            tasks = [process_spot(spot) for spot in nearby_spots]
            results = await asyncio.gather(*tasks, return_exceptions=True)
            
            for result in results:
                if isinstance(result, Exception):
                    logger.error(f"Spot evaluation failed: {result}")
                elif result:
                    evaluated_spots.append(result)
            
            # Step 4: Sort by score and return top N
            evaluated_spots.sort(key=lambda x: x["score"], reverse=True)
            top_spots = evaluated_spots[:input_data.top_n]
            
            # Step 5: Generate AI reasoning and summary
            ai_summary = ""
            ai_reasoning = ""
            
            try:
                logger.info("Generating AI analysis of surf conditions...")
                llm_result = await self.llm_agent.run(LLMAgentInput(
                    user_location=input_data.user_location,
                    user_preferences=input_data.user_preferences,
                    surf_spots=top_spots
                ))
                
                if llm_result.success:
                    ai_summary = llm_result.summary
                    ai_reasoning = llm_result.reasoning
                    logger.info("AI analysis completed successfully")
                else:
                    logger.warning(f"LLM agent failed: {llm_result.error}")
                    ai_summary = f"Found {len(top_spots)} surf spots. Top recommendation: {top_spots[0]['name']} with {top_spots[0]['score']}/100 score."
                    
            except Exception as e:
                logger.error(f"AI analysis failed: {e}")
                ai_summary = f"Found {len(top_spots)} surf spots near {input_data.user_location}. Analysis based on wave conditions and user preferences."
            
            return SpotFinderOutput(
                success=True,
                user_location=user_coords,
                spots=top_spots,
                ai_summary=ai_summary,
                ai_reasoning=ai_reasoning
            )
            
        except Exception as e:
            logger.error(f"Surf spot finder error: {e}")
            return SpotFinderOutput(
                success=False,
                error=f"Internal error: {str(e)}"
            )


def create_spot_finder_tool() -> Dict[str, Any]:
    """Factory function to create the surf spot finder tool.
    
    Creates and configures the main MCP tool for surf spot recommendations.
    Returns tool specification compatible with MCP protocol.
    
    Returns:
        Dict containing tool name, description, schema, and function reference.
        
    Example:
        >>> tool = create_spot_finder_tool()
        >>> result = await tool["function"](input_data)
    """
    tool = SurfSpotFinder()
    return {
        "name": tool.name,
        "description": tool.description,
        "input_schema": SpotFinderInput.schema(),
        "function": tool.run,
    }