"""Векторное хранилище знаний: Qdrant + Ollama embeddings + чанкинг."""
from __future__ import annotations

import os
from pathlib import Path
from typing import Any

from langchain_core.documents import Document
from langchain_ollama import OllamaEmbeddings
from langchain_qdrant import QdrantVectorStore
from langchain_text_splitters import RecursiveCharacterTextSplitter
from qdrant_client import QdrantClient
from qdrant_client.http.models import Distance, VectorParams

COLLECTION_NAME = "knowledge_base"
DEFAULT_QDRANT_PATH = os.getenv("QDRANT_PATH", "./qdrant_data")
OLLAMA_BASE_URL = os.getenv("OLLAMA_BASE_URL", "http://localhost:11434")
EMBED_MODEL = os.getenv("OLLAMA_EMBED_MODEL", "nomic-embed-text")


class KnowledgeBase:
    """Локальная RAG-база на Qdrant с эмбеддингами Ollama."""

    def __init__(
        self,
        qdrant_path: str | Path = DEFAULT_QDRANT_PATH,
        collection_name: str = COLLECTION_NAME,
    ) -> None:
        self.collection_name = collection_name
        self.qdrant_path = Path(qdrant_path)
        self.qdrant_path.mkdir(parents=True, exist_ok=True)
        self.embeddings = OllamaEmbeddings(model=EMBED_MODEL, base_url=OLLAMA_BASE_URL)
        self.client = QdrantClient(path=str(self.qdrant_path))
        self.splitter = RecursiveCharacterTextSplitter(chunk_size=500, chunk_overlap=100)
        self._ensure_collection()
        self.vector_store = QdrantVectorStore(
            client=self.client,
            collection_name=self.collection_name,
            embedding=self.embeddings,
        )

    def _ensure_collection(self) -> None:
        if self.client.collection_exists(self.collection_name):
            return
        sample = self.embeddings.embed_query("init")
        self.client.create_collection(
            collection_name=self.collection_name,
            vectors_config=VectorParams(size=len(sample), distance=Distance.COSINE),
        )

    def _chunk_document(self, content: str, title: str) -> list[Document]:
        docs = self.splitter.create_documents(texts=[content], metadatas=[{"title": title}])
        for idx, doc in enumerate(docs):
            doc.metadata["chunk_index"] = idx
            doc.metadata["source"] = title
        return docs

    def add_document(self, content: str, title: str) -> int:
        chunks = self._chunk_document(content, title)
        self.vector_store.add_documents(chunks)
        return len(chunks)

    def search(self, query: str, max_results: int = 5) -> list[dict[str, Any]]:
        hits = self.vector_store.similarity_search_with_score(query, k=max_results)
        result: list[dict[str, Any]] = []
        for doc, score in hits:
            result.append(
                {
                    "title": doc.metadata.get("title", "без названия"),
                    "content": doc.page_content[:400],
                    "score": round(float(score), 4),
                }
            )
        return result


def build_knowledge_base() -> KnowledgeBase:
    """Фабрика для автопроверки."""
    return KnowledgeBase()