"""Oghma Lore Retriever - Queries ChromaDB for relevant Tamrielic lore."""

from __future__ import annotations

import time
from functools import lru_cache
from typing import TYPE_CHECKING

import chromadb
import structlog
from chromadb.config import Settings

from .models import EducationLevel, LoreEntry, NPCProfile

if TYPE_CHECKING:
    from chromadb import Collection

logger = structlog.get_logger()


class OghmaRetriever:
    """Retrieves relevant lore from Oghma ChromaDB collections."""

    def __init__(
        self,
        host: str = "iris-dev.eachpath.local",
        port: int = 35000,
        collection_lore: str = "oghma_lore",
        collection_basic: str = "oghma_basic",
        max_results: int = 5,
        min_score: float = 0.55,
    ):
        self.host = host
        self.port = port
        self.collection_lore_name = collection_lore
        self.collection_basic_name = collection_basic
        self.max_results = max_results
        self.min_score = min_score

        self._client: chromadb.HttpClient | None = None
        self._collection_lore: Collection | None = None
        self._collection_basic: Collection | None = None

    def _get_client(self) -> chromadb.HttpClient:
        """Get or create ChromaDB client."""
        if self._client is None:
            self._client = chromadb.HttpClient(
                host=self.host,
                port=self.port,
                settings=Settings(anonymized_telemetry=False),
            )
            logger.info("Connected to ChromaDB", host=self.host, port=self.port)
        return self._client

    def _get_collection(self, education_level: EducationLevel) -> Collection:
        """Get the appropriate collection based on education level."""
        client = self._get_client()

        if education_level == EducationLevel.SCHOLAR:
            if self._collection_lore is None:
                self._collection_lore = client.get_collection(self.collection_lore_name)
            return self._collection_lore
        else:
            if self._collection_basic is None:
                self._collection_basic = client.get_collection(self.collection_basic_name)
            return self._collection_basic

    def retrieve(
        self,
        query: str,
        npc_profile: NPCProfile,
    ) -> tuple[list[LoreEntry], float]:
        """
        Retrieve relevant lore entries for an NPC.

        Args:
            query: Conversation context to search for
            npc_profile: NPC profile for knowledge filtering

        Returns:
            Tuple of (lore entries, query time in ms)
        """
        if not query.strip():
            return [], 0.0

        start_time = time.perf_counter()

        try:
            collection = self._get_collection(npc_profile.education_level)

            # Build metadata filter for knowledge classes
            # NOTE: Currently disabled because CHIM's Oghma data doesn't have
            # knowledge_class populated consistently. Enable when data is enriched.
            where_filter = None
            # TODO: Re-enable when knowledge_class data is available
            # if npc_profile.knowledge_classes:
            #     if len(npc_profile.knowledge_classes) == 1:
            #         where_filter = {"knowledge_classes": {"$contains": npc_profile.knowledge_classes[0]}}
            #     else:
            #         where_filter = {
            #             "$or": [
            #                 {"knowledge_classes": {"$contains": kc}}
            #                 for kc in npc_profile.knowledge_classes
            #             ]
            #         }

            # Query ChromaDB
            results = collection.query(
                query_texts=[query],
                n_results=self.max_results,
                where=where_filter,
                include=["documents", "metadatas", "distances"],
            )

            # Parse results
            entries = []
            if results and results["documents"] and results["documents"][0]:
                for i, doc in enumerate(results["documents"][0]):
                    metadata = results["metadatas"][0][i] if results["metadatas"] else {}
                    distance = results["distances"][0][i] if results["distances"] else 1.0

                    # Convert distance to similarity score (ChromaDB uses L2 distance)
                    # Lower distance = higher similarity
                    score = 1.0 / (1.0 + distance)

                    if score >= self.min_score:
                        entries.append(
                            LoreEntry(
                                topic=metadata.get("topic", "Unknown"),
                                content=doc,
                                category=metadata.get("category", "Unknown"),
                                score=score,
                                knowledge_classes=metadata.get("knowledge_classes", "").split(","),
                            )
                        )

            query_time = (time.perf_counter() - start_time) * 1000

            logger.info(
                "Retrieved lore entries",
                query_preview=query[:100],
                npc_name=npc_profile.name,
                education=npc_profile.education_level.value,
                entries_found=len(entries),
                query_time_ms=round(query_time, 2),
            )

            return entries, query_time

        except Exception as e:
            logger.error("Failed to retrieve lore", error=str(e))
            query_time = (time.perf_counter() - start_time) * 1000
            return [], query_time

    def health_check(self) -> bool:
        """Check if ChromaDB is reachable."""
        try:
            client = self._get_client()
            client.heartbeat()
            return True
        except Exception as e:
            logger.error("ChromaDB health check failed", error=str(e))
            return False


# Cached retriever instance
@lru_cache(maxsize=1)
def get_retriever(
    host: str = "iris-dev.eachpath.local",
    port: int = 35000,
) -> OghmaRetriever:
    """Get cached retriever instance."""
    return OghmaRetriever(host=host, port=port)