OneScience-Group/ShapeNetCar

1#!/usr/bin/env python32"""Validate ShapeNetCar mlcfd_data structure and readable arrays."""3 4from __future__ import annotations5 6import argparse7import sys8from pathlib import Path9 10import numpy as np11 12 13STATS = {14    "mean_in.npy": (7,),15    "std_in.npy": (7,),16    "mean_out.npy": (4,),17    "std_out.npy": (4,),18}19TRAINING_FILES = ("Cd.npy", "I1.npy", "I2.npy", "Press.npy", "Velo.npy")20PREPROCESSED_FILES = ("x.npy", "y.npy", "pos.npy", "surf.npy", "edge_index.npy")21 22 23def fail(message: str) -> None:24    print(f"[FAIL] {message}", file=sys.stderr)25    raise SystemExit(1)26 27 28def load(path: Path) -> np.ndarray:29    try:30        return np.load(path, allow_pickle=False)31    except Exception as exc:  # pragma: no cover - diagnostic path32        fail(f"cannot read {path}: {exc}")33 34 35def main() -> None:36    parser = argparse.ArgumentParser()37    parser.add_argument("--data-root", default="data/mlcfd_data")38    args = parser.parse_args()39    root = Path(args.data_root)40 41    if not root.is_dir():42        fail(f"data root not found: {root}")43 44    for subdir in ("training_data", "preprocessed_data", "stats"):45        if not (root / subdir).is_dir():46            fail(f"missing directory: {root / subdir}")47 48    for filename, shape in STATS.items():49        arr = load(root / "stats" / filename)50        if arr.shape != shape:51            fail(f"stats shape mismatch for {filename}: expected {shape}, got {arr.shape}")52        if not np.issubdtype(arr.dtype, np.floating):53            fail(f"stats dtype mismatch for {filename}: got {arr.dtype}")54 55    train_param_dirs = sorted(p for p in (root / "training_data").glob("param*") if p.is_dir())56    if not train_param_dirs:57        fail("no training_data/param* directories found")58    for param_dir in train_param_dirs:59        for filename in TRAINING_FILES:60            path = param_dir / filename61            if not path.is_file():62                fail(f"missing training file: {path}")63            arr = load(path)64            if arr.size == 0:65                fail(f"empty training array: {path}")66 67    sample_dirs = sorted(p for p in (root / "preprocessed_data").glob("param*/*") if p.is_dir())68    if not sample_dirs:69        fail("no preprocessed sample directories found")70    sample = sample_dirs[0]71    arrays = {name: load(sample / name) for name in PREPROCESSED_FILES}72    if arrays["x.npy"].ndim != 2 or arrays["x.npy"].shape[1] != 7:73        fail(f"x.npy schema mismatch in {sample}: {arrays['x.npy'].shape}")74    if arrays["y.npy"].ndim != 2 or arrays["y.npy"].shape[1] != 4:75        fail(f"y.npy schema mismatch in {sample}: {arrays['y.npy'].shape}")76    if arrays["pos.npy"].ndim != 2 or arrays["pos.npy"].shape[1] != 3:77        fail(f"pos.npy schema mismatch in {sample}: {arrays['pos.npy'].shape}")78    if arrays["surf.npy"].ndim != 1:79        fail(f"surf.npy schema mismatch in {sample}: {arrays['surf.npy'].shape}")80    if arrays["edge_index.npy"].ndim != 2 or arrays["edge_index.npy"].shape[0] != 2:81        fail(f"edge_index.npy schema mismatch in {sample}: {arrays['edge_index.npy'].shape}")82 83    node_count = arrays["x.npy"].shape[0]84    if arrays["y.npy"].shape[0] != node_count or arrays["pos.npy"].shape[0] != node_count:85        fail(f"node count mismatch in {sample}")86    if arrays["surf.npy"].shape[0] != node_count:87        fail(f"surface mask length mismatch in {sample}")88 89    print("[OK] ShapeNetCar data validation passed")90    print(f"[OK] training param dirs: {len(train_param_dirs)}")91    print(f"[OK] preprocessed samples: {len(sample_dirs)}")92    print(f"[OK] checked sample: {sample.relative_to(root)}")93 94 95if __name__ == "__main__":96    main()97