# app/agents/vectorization_agent.py
from langchain.text_splitter import RecursiveCharacterTextSplitter

text_splitter = RecursiveCharacterTextSplitter(
    chunk_size=1000,
    chunk_overlap=100,
)

def agent_vectorize_and_store(
        doc_id: str,
        text: str,
        category: str,
        language: str,
        embedding_model,
        vector_store
):
    print(f"--- [Background Task] Starting vectorization (ID: {doc_id})...")

    try:
        chunks = text_splitter.split_text(text)
        if not chunks:
            print(f"--- [Background task] document is empty, skip vectorization. (ID: {doc_id})")
            return

        chunk_ids = [f"{doc_id}_{i}" for i in range(len(chunks))]
        metadatas = [{"doc_id": doc_id, "category": category, "language": language, "chunk_number": i} for i in
                     range(len(chunks))]

        embeddings = embedding_model.embed_documents(chunks)

        vector_store.add(
            ids=chunk_ids,
            embeddings=embeddings,
            documents=chunks,
            metadatas=metadatas
        )
        print(f"--- [Background Task] Document {doc_id} vectorized。")
    except Exception as e:
        print(f"--- [Background Task] Document vectorization failed (ID: {doc_id}): {e}")