scripts/prepare_dataset.py

"""Dataset preparation entrypoint."""

from __future__ import annotations

import argparse
from pathlib import Path

import pandas as pd

from dipauglib.utils.dataset import SplitConfig, build_split_manifest, save_manifest


def parse_args() -> argparse.Namespace:
    parser = argparse.ArgumentParser(description="Prepare wheat disease dataset manifest.")
    parser.add_argument("--input-csv", type=Path, required=False, help="Input CSV with columns path,label")
    parser.add_argument("--output-csv", type=Path, default=Path("results/manifests/split_manifest.csv"))
    return parser.parse_args()


def main() -> None:
    args = parse_args()
    if args.input_csv is None:
        print("Scaffold ready. Provide --input-csv to build a split manifest.")
        return
    records = pd.read_csv(args.input_csv)
    manifest = build_split_manifest(records, label_column="label", config=SplitConfig(seed=42))
    save_manifest(manifest, args.output_csv)
    print(f"Saved manifest to {args.output_csv}")


if __name__ == "__main__":
    main()