OneScience-Group/cfdbench

1#!/usr/bin/env python32"""Validate the standardized CFDBench dataset package."""3 4from __future__ import annotations5 6import argparse7import hashlib8import json9import sys10from collections import defaultdict11from pathlib import Path12 13import numpy as np14 15 16REPO_ROOT = Path(__file__).resolve().parents[1]17DATA_ROOT = REPO_ROOT / "data"18CHECKSUM_PATH = REPO_ROOT / "files_sha256.jsonl"19PROBLEMS = ("cavity", "tube", "dam", "cylinder")20SUBSETS = ("bc", "geo", "prop")21REQUIRED_CASE_FILES = ("case.json", "u.npy", "v.npy")22OPTIONAL_EXTENSIONS = {".png"}23 24 25def fail(message: str) -> None:26    print(f"[FAIL] {message}")27    raise SystemExit(1)28 29 30def ok(message: str) -> None:31    print(f"[OK] {message}")32 33 34def warn(message: str) -> None:35    print(f"[WARN] {message}")36 37 38def sha256_file(path: Path) -> str:39    digest = hashlib.sha256()40    with path.open("rb") as handle:41        for chunk in iter(lambda: handle.read(1024 * 1024), b""):42            digest.update(chunk)43    return digest.hexdigest()44 45 46def iter_case_dirs() -> list[Path]:47    case_dirs = []48    for problem in PROBLEMS:49        problem_dir = DATA_ROOT / problem50        if not problem_dir.is_dir():51            fail(f"missing problem directory: {problem_dir}")52        for subset in SUBSETS:53            subset_dir = problem_dir / subset54            if subset_dir.is_dir():55                cases = sorted(subset_dir.glob("case*"), key=lambda p: int(p.name[4:]))56                if not cases:57                    fail(f"empty subset directory: {subset_dir}")58                case_dirs.extend(cases)59    return case_dirs60 61 62def validate_structure(sample_cases_per_subset: int) -> None:63    if not DATA_ROOT.is_dir():64        fail(f"dataset data root does not exist: {DATA_ROOT}")65    counts = defaultdict(int)66    sampled = 067    for problem in PROBLEMS:68        for subset in SUBSETS:69            subset_dir = DATA_ROOT / problem / subset70            if not subset_dir.is_dir():71                continue72            cases = sorted(subset_dir.glob("case*"), key=lambda p: int(p.name[4:]))73            counts[f"{problem}/{subset}"] = len(cases)74            for case_dir in cases[:sample_cases_per_subset]:75                for name in REQUIRED_CASE_FILES:76                    if not (case_dir / name).is_file():77                        fail(f"missing required case file: {case_dir / name}")78                params = json.loads((case_dir / "case.json").read_text(encoding="utf-8"))79                if not isinstance(params, dict) or not params:80                    fail(f"case.json must be a non-empty object: {case_dir / 'case.json'}")81                u = np.load(case_dir / "u.npy", mmap_mode="r")82                v = np.load(case_dir / "v.npy", mmap_mode="r")83                if u.shape != v.shape:84                    fail(f"u/v shape mismatch: {case_dir}: {u.shape} vs {v.shape}")85                if u.ndim != 3:86                    fail(f"u/v arrays must be 3D time/grid arrays: {case_dir}: {u.shape}")87                if not np.issubdtype(u.dtype, np.number) or not np.issubdtype(v.dtype, np.number):88                    fail(f"u/v dtype must be numeric: {case_dir}: {u.dtype}, {v.dtype}")89                if not np.isfinite(np.asarray(u[0])).all() or not np.isfinite(np.asarray(v[0])).all():90                    fail(f"first frame contains non-finite values: {case_dir}")91                sampled += 192    if not counts:93        fail("no CFDBench subsets found")94    ok(f"dataset structure and sampled arrays are valid: {sum(counts.values())} cases, sampled {sampled}")95 96 97def verify_checksums(full_hash: bool) -> None:98    if not CHECKSUM_PATH.exists():99        warn(f"checksum manifest is not present: {CHECKSUM_PATH}")100        return101    records = []102    for line in CHECKSUM_PATH.read_text(encoding="utf-8").splitlines():103        if line.strip():104            records.append(json.loads(line))105    if not records:106        fail("checksum manifest is empty")107    for record in records:108        path = REPO_ROOT / record["path"]109        if not path.is_file():110            fail(f"checksum entry points to missing file: {path}")111        size = path.stat().st_size112        if size != record["size"]:113            fail(f"size mismatch for {path}: expected {record['size']}, got {size}")114        if full_hash:115            digest = sha256_file(path)116            if digest != record["sha256"]:117                fail(f"sha256 mismatch for {path}")118    mode = "size+sha256" if full_hash else "size"119    ok(f"checksum manifest verified in {mode} mode: {len(records)} files")120 121 122def main() -> int:123    parser = argparse.ArgumentParser()124    parser.add_argument("--sample-cases-per-subset", type=int, default=2)125    parser.add_argument("--full-hash", action="store_true")126    args = parser.parse_args()127    if args.sample_cases_per_subset < 1:128        fail("--sample-cases-per-subset must be positive")129    validate_structure(args.sample_cases_per_subset)130    verify_checksums(args.full_hash)131    ok("dataset validation completed")132    return 0133 134 135if __name__ == "__main__":136    sys.exit(main())137