preprocess.py

"""
Preprocess all level images into structured JSON files.

Run once:
    python3 preprocess.py

Reads:  assets_pdf/<id>.png
Writes: assets_ocr/<id>.json

Already-processed files are skipped, so the script is safe to re-run
(e.g. after adding new images or fixing the OCR pipeline).
"""

import json
import sys
from pathlib import Path

import ocr as _ocr

INPUT_DIR = Path("assets_pdf")
OUTPUT_DIR = Path("assets_ocr")


def main():
    OUTPUT_DIR.mkdir(exist_ok=True)

    pngs = sorted(INPUT_DIR.glob("*.png"), key=lambda p: int(p.stem))
    total = len(pngs)

    ok = skipped = failed = 0

    for i, png in enumerate(pngs, 1):
        out = OUTPUT_DIR / f"{png.stem}.json"

        if out.exists():
            skipped += 1
            continue

        try:
            result = _ocr.parse_image(str(png))
            with open(out, "w") as f:
                json.dump(result, f, cls=_ocr.NumpyEncoder)
            ok += 1
            print(f"[{i}/{total}] {png.name} -> {out.name}")
        except Exception as exc:
            failed += 1
            print(f"[{i}/{total}] FAILED {png.name}: {exc}", file=sys.stderr)

    print(f"\nDone: {ok} processed, {skipped} skipped, {failed} failed")


if __name__ == "__main__":
    main()