src/config.py

import os
from dataclasses import dataclass

from dotenv import load_dotenv

load_dotenv()


@dataclass
class Config:
    bd_file: str = os.getenv("BD_FILE", "data/raw/SuicideBD-dataset.csv")
    en_file: str = os.getenv("EN_FILE", "data/raw/English_Suicide_Detection.csv")
    bn_file: str = os.getenv(
        "BN_FILE", "data/raw/Bangla Suicidal Intention Dataset.csv"
    )

    out_dir: str = os.getenv("OUT_DIR", "data/processed")
    artifact_dir: str = os.getenv("ARTIFACT_DIR", "outputs/artifacts")

    random_seed: int = int(os.getenv("RANDOM_SEED", "42"))
    test_size: float = float(os.getenv("TEST_SIZE", "0.3"))  # text split
    val_size: float = float(os.getenv("VAL_SIZE", "0.5"))  # within temp split

    # BD structured drop columns
    drop_cols = [
        "id",
        "full_name",
        "method",
        "url",
        "data_source",
        "unix_time",
        "suicide_date",
        "latitude",
        "longitude",
    ]

    # Defaults you used
    defaults = {
        "profession_group": "unemployed",
        "religion": "muslim",
        "hometown": "Dhaka",
        "reason": "relationship problem",
        "reason_description": "relationship problem",
    }


def ensure_dirs(cfg: Config) -> None:
    os.makedirs(cfg.out_dir, exist_ok=True)
    os.makedirs(cfg.artifact_dir, exist_ok=True)