import os
from langchain_community.vectorstores import FAISS
from langchain_huggingface import HuggingFaceEmbeddings
from langchain.text_splitter import RecursiveCharacterTextSplitter

def get_embeddings_model():
    return HuggingFaceEmbeddings(
        model_name="sentence-transformers/all-MiniLM-L6-v2"
    )

def get_vector_store_path(user_id: str) -> str:
    base_path = os.path.join("data", "users", user_id, "vector_store")
    os.makedirs(base_path, exist_ok=True)
    return base_path

def split_text_into_chunks(text: str, chunk_size: int = 500, chunk_overlap: int = 50):
    text_splitter = RecursiveCharacterTextSplitter(
        chunk_size=chunk_size,
        chunk_overlap=chunk_overlap,
        length_function=len,
    )
    chunks = text_splitter.split_text(text)
    return chunks

def save_chunks_to_vectorstore(user_id: str, chunks: list):
    embeddings = get_embeddings_model()
    vector_store_path = get_vector_store_path(user_id)

    vector_store_file = os.path.join(vector_store_path, "index.faiss")

    if os.path.exists(vector_store_file):
        vectorstore = FAISS.load_local(
            vector_store_path,
            embeddings,
            allow_dangerous_deserialization=True
        )
        vectorstore.add_texts(chunks)
    else:
        vectorstore = FAISS.from_texts(chunks, embeddings)

    vectorstore.save_local(vector_store_path)
    return f"Added {len(chunks)} chunks to knowledge base"

def load_vectorstore(user_id: str):
    embeddings = get_embeddings_model()
    vector_store_path = get_vector_store_path(user_id)
    vector_store_file = os.path.join(vector_store_path, "index.faiss")

    if not os.path.exists(vector_store_file):
        raise FileNotFoundError(f"No knowledge base found for user {user_id}")

    vectorstore = FAISS.load_local(
        vector_store_path,
        embeddings,
        allow_dangerous_deserialization=True
    )
    return vectorstore

def retrieve_relevant_chunks(user_id: str, query: str, k: int = 3):
    vectorstore = load_vectorstore(user_id)
    docs = vectorstore.similarity_search(query, k=k)
    return [doc.page_content for doc in docs]