fetch_carcrashnet.py

"""Option 2 helper - fetch the CarCrashNet crash-trajectory dataset.

CarCrashNet (Elrefaie et al., CC BY-NC) releases mesh-resolved field TRAJECTORIES
in VTKHDF: per-node displacement, velocity, von Mises stress, plastic strain,
internal energy and erosion flags at every saved timestep, for a bumper-beam /
crash-box pole-impact dataset plus three full-vehicle datasets. crash_solver.py
trains a temporal surrogate on these to predict the deformation trajectory of a
new design without re-running LS-DYNA / OpenRadioss.

Sources (see the repo README for the exact DOIs):
    GitHub : https://github.com/Mohamedelrefaie/CarCrashNet
    Data   : Harvard Dataverse / Hugging Face (VTKHDF trajectory files)

This module only LOCATES / verifies the dataset; it does not scrape gated hosts.
Point --dir at the folder where you downloaded the VTKHDF files.
"""
from __future__ import annotations

import argparse
import glob
import os
from pathlib import Path

HERE = Path(__file__).parent
DEST = HERE / "DrivAerNet" / "CrashTrajectories"


def status(src_dir: str | None = None) -> dict:
    """Report how many VTKHDF trajectory files are present (in src_dir or DEST)."""
    root = Path(src_dir) if src_dir else DEST
    files = []
    for ext in ("*.vtkhdf", "*.hdf", "*.h5", "*.vtu", "*.pvd"):
        files += glob.glob(str(root / "**" / ext), recursive=True)
    return {"root": str(root), "n_files": len(files),
            "sample": [os.path.basename(f) for f in files[:5]],
            "ready": len(files) > 0}


if __name__ == "__main__":
    ap = argparse.ArgumentParser()
    ap.add_argument("--dir", default=None, help="folder with downloaded VTKHDF files")
    a = ap.parse_args()
    s = status(a.dir)
    print(f"CarCrashNet trajectories under {s['root']}: {s['n_files']} files")
    if s["ready"]:
        print("  sample:", s["sample"])
        print("  -> ready: train with  python -m crash_solver train --data", s["root"])
    else:
        print("  NONE found. Download the VTKHDF trajectories from:")
        print("    https://github.com/Mohamedelrefaie/CarCrashNet  (README -> Dataset)")
        print(f"  then put them under {DEST}  (or pass --dir).")