"""Vector store implementation using Qdrant and Ollama embeddings.

This module provides a simple wrapper around QdrantVectorStore that handles
- Initialization of the Qdrant client and collection.
- Chunking of documents using RecursiveCharacterTextSplitter.
- Adding documents with embeddings from Ollama.
- Semantic search.
"""

from pathlib import Path
from typing import List, Dict, Any

from langchain_ollama import OllamaEmbeddings
from langchain_qdrant import QdrantVectorStore
from langchain_text_splitters import RecursiveCharacterTextSplitter
from langchain.schema import Document

# Global configuration
QDRANT_HOST = "localhost"
QDRANT_PORT = 6333
COLLECTION_NAME = "knowledge_base"
EMBEDDING_MODEL = "nomic-embed-text"

# Initialize embeddings and splitter
embeddings = OllamaEmbeddings(model=EMBEDDING_MODEL)
text_splitter = RecursiveCharacterTextSplitter(chunk_size=500, chunk_overlap=50)

# Create or connect to Qdrant collection
vector_store = QdrantVectorStore(
    client_kwargs={"host": QDRANT_HOST, "port": QDRANT_PORT},
    collection_name=COLLECTION_NAME,
    embeddings=embeddings,
)

# Ensure collection exists
if not vector_store.client.has_collection(COLLECTION_NAME):
    vector_store.client.create_collection(COLLECTION_NAME)

def add_documents(docs: List[Dict[str, str]]) -> None:
    """Add a list of documents to the vector store.

    Each document dict must contain ``title`` and ``content`` keys.
    The content is split into chunks before being stored.
    """
    documents: List[Document] = []
    for doc in docs:
        title = doc.get("title", "")
        content = doc.get("content", "")
        # Split content into chunks
        chunks = text_splitter.split_text(content)
        for i, chunk in enumerate(chunks):
            meta = {"title": title, "chunk_index": i}
            documents.append(Document(page_content=chunk, metadata=meta))
    vector_store.add_documents(documents)

def search(query: str, k: int = 5) -> List[Document]:
    """Semantic search in the vector store.

    Returns a list of Documents ordered by relevance.
    """
    return vector_store.similarity_search(query, k=k)

# Convenience: add a single document

def add_document(title: str, content: str) -> None:
    add_documents([{"title": title, "content": content}])

# Convenience: search and return plain strings

def search_text(query: str, k: int = 5) -> List[str]:
    docs = search(query, k)
    return [f"{doc.metadata.get('title', 'Untitled')} (chunk {doc.metadata.get('chunk_index', 0)}): {doc.page_content[:200]}..." for doc in docs]