ocr_test.py

"""
Quick smoke-test for ccsolver.ocr.

Parses a handful of portrait board images and prints the extracted grid.
Run with:
    nix-shell -p python3 python3Packages.opencv4Full \
        python3Packages.networkx ... --run "python3 ocr_test.py"
"""

from pathlib import Path
from ccsolver.ocr import ocr_image

SAMPLES = [
    "assets_pdf/002.png",   # simple 4-tile single-chain (portrait)
    "assets_pdf/004.png",   # L-shape
    "assets_pdf/006.png",   # C-shape
    "assets_pdf/008.png",
    "assets_pdf/010.png",
    "assets_pdf/020.png",   # has digits 2
    "assets_pdf/100.png",   # special square 871×871
    "assets_pdf/102.png",   # multi-chain portrait
    "assets_pdf/200.png",   # special square, multi-chain
]

for path in SAMPLES:
    p = Path(path)
    if not p.exists():
        print(f"{path}: NOT FOUND")
        continue

    level = ocr_image(path)
    if level is None:
        print(f"{path}: FAILED (could not parse)")
        continue

    print(f"\n{'─'*60}")
    print(f"{path}  →  {level.n_cols}×{level.n_rows} grid  "
          f"({len(level.color_tiles)} X, {len(level.number_tiles)} numbers)")
    print(level.to_game_string())

print("\n\nBatch test: parsing all portrait images …")
from ccsolver.ocr import ocr_directory
results = ocr_directory("assets_pdf", orientation="portrait")
failed_total = sum(
    1 for p in sorted(Path("assets_pdf").glob("*.png"))
    if p.stat().st_size > 0
    and __import__("cv2").imread(str(p)) is not None
    and __import__("cv2").imread(str(p)).shape[0] > __import__("cv2").imread(str(p)).shape[1]
) - len(results)

print(f"Parsed: {len(results)}  |  failed/skipped: {failed_total}")

# Show distribution of grid sizes
from collections import Counter
sizes = Counter(f"{r.n_cols}×{r.n_rows}" for _, r in results)
for size, count in sorted(sizes.items()):
    print(f"  {size}: {count} levels")