scripts/build_metadata.py

from __future__ import annotations

import logging


try:
    from _bootstrap import add_project_root_to_path
except ModuleNotFoundError:
    from scripts._bootstrap import add_project_root_to_path


add_project_root_to_path()

from src.config import IMAGE_DIR, METADATA_CSV
from src.dataset import build_metadata


logging.basicConfig(level=logging.INFO, format="%(levelname)s:%(name)s:%(message)s")
logger = logging.getLogger(__name__)


def main() -> None:
    metadata = build_metadata(IMAGE_DIR, METADATA_CSV)
    total_images = len(metadata)
    categories = metadata["category"].value_counts().sort_index() if total_images else {}

    logger.info("Image directory: %s", IMAGE_DIR)
    logger.info("Metadata CSV: %s", METADATA_CSV)
    logger.info("Total images: %d", total_images)
    logger.info("Total categories: %d", len(categories))
    if total_images == 0:
        logger.warning("No images found. Add files under data/images/<category>/ and run this script again.")
    else:
        for category, count in categories.items():
            logger.info("  %s: %s", category, count)


if __name__ == "__main__":
    main()