src/config.py

import os
from pydantic import BaseModel

class Settings(BaseModel):
    raw_dir: str = os.path.join("data", "raw")
    storage_dir: str = "storage"
    index_path: str = os.path.join("storage", "index.faiss")
    docs_path: str = os.path.join("storage", "docs.pkl")
    meta_path: str = os.path.join("storage", "meta.json")

    embedding_model: str = "sentence-transformers/all-MiniLM-L6-v2"

    local_generation_model: str = os.getenv("LOCAL_GENERATION_MODEL", "").strip()
    hf_token: str = os.getenv("HF_TOKEN", "").strip()

    @property
    def mode(self) -> str:
        return "rag" if (self.hf_token or self.local_generation_model) else "retrieval"

    top_k: int = 5
    max_context_chars: int = 9000

    title: str = "KGB Document Chatbot"
    description: str = "Ask questions about declassified KGB documents using AI-powered search and analysis."