config/paths.py

"""Filesystem locations used by the pipeline."""

import os
from pathlib import Path

PROJECT_ROOT = Path(__file__).resolve().parent.parent

ORIGINAL_DIR = PROJECT_ROOT / "original"

DEFAULT_RAW_DATA_PATH = Path(
    os.getenv("ASA_RAW_DATA_PATH", ORIGINAL_DIR / "original_dataset.xlsx")
)

ARTIFACTS_DIR = Path(os.getenv("ASA_ARTIFACTS_DIR", PROJECT_ROOT / "artifacts"))

DEFAULT_SHARING_DATA_PATH = Path(
    os.getenv("ASA_SHARING_DATA_PATH", ARTIFACTS_DIR / "sharing_arabic.xlsx")
)

DEFAULT_CLEANED_DATA_PATH = Path(
    os.getenv("ASA_CLEANED_DATA_PATH", ARTIFACTS_DIR / "cleaned.xlsx")
)

DEFAULT_MODEL_PATH = Path(
    os.getenv("ASA_MODEL_PATH", ARTIFACTS_DIR / "best_model.joblib")
)