"""
BƯỚC 4: VECTORSTORE (FAISS in-memory)
"""

from langchain_community.vectorstores import FAISS
from embeddings import get_embeddings


def build_vectorstore(chunks):
    print(">>> Initialising embedding model for FAISS index ...")
    embeddings = get_embeddings()

    print(f">>> Building FAISS index from {len(chunks)} chunks ...")
    vs = FAISS.from_documents(chunks, embeddings)
    print(">>> FAISS index built.\n")
    return vs


if __name__ == "__main__":
    from load_documents import load_documents
    from split_documents import split_documents

    docs = load_documents()
    chunks = split_documents(docs)
    vs = build_vectorstore(chunks)
    res = vs.similarity_search(
        "Fristen für die Prüfungsanmeldung im Bachelorstudium", k=3
    )
    for r in res:
        print(r.page_content[:200], r.metadata)