# config/settings.py
from pydantic_settings import BaseSettings
from typing import List

class Settings(BaseSettings):
    DATA_DIR: str = "data"
    RAW_PDFS_DIR: str = "data/raw_pdfs"
    PROCESSED_TEXT_DIR: str = "data/processed_text"

    CHUNK_SIZE: int = 500
    CHUNK_OVERLAP: int = 50
    MIN_CHUNK_LENGTH: int = 100

    EMBEDDING_MODEL: str = "all-MiniLM-L6-v2"
    EMBEDDING_DIMENSION: int = 384

    LOG_LEVEL: str = "INFO"
    LOG_FILE: str = "logs/pipeline.log"

settings = Settings()