File size: 5,010 Bytes
db613b6 | 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60 61 62 63 64 65 66 67 68 69 70 71 72 73 74 75 76 77 78 79 80 81 82 83 84 85 86 87 88 89 90 91 92 93 94 95 96 97 98 99 100 101 102 103 104 105 106 107 108 109 110 111 112 113 114 115 116 117 118 119 120 121 122 123 124 125 126 127 128 129 | """Read-only audit of existing Hub asset shards using ranged metadata reads."""
from __future__ import annotations
import argparse
import json
from collections import Counter
from pathlib import Path
import pyarrow.parquet as pq
from huggingface_hub import HfApi, HfFileSystem
from kitti_c2l.release import require
from kitti_c2l.schema import ASSETS_SCHEMA
from kitti_c2l.sequence_mapping import load_sequence_mapping
def audit(repo_id: str, revision: str | None, root: Path) -> dict:
api = HfApi()
info = api.dataset_info(repo_id, revision=revision, files_metadata=True)
inventory = []
for item in info.siblings or []:
lfs = item.lfs
inventory.append(
{
"path": item.rfilename,
"bytes": item.size,
"sha256": (getattr(lfs, "sha256", None) or lfs.get("sha256")) if lfs else None,
}
)
fs = HfFileSystem()
mapping = load_sequence_mapping()
expected = {
(seq, frame)
for seq, source in mapping.items()
if source.available
for frame in range(source.frame_count)
}
seen = set()
counts = Counter()
shards = []
columns = [
"frame_id",
"sequence",
"frame_index",
"split",
"camera_id",
"image_width",
"image_height",
"raw_date",
"raw_drive",
"raw_frame_index",
]
for item in inventory:
if not Path(item["path"]).name.startswith("frames_"):
continue
path = f"datasets/{repo_id}@{info.sha}/{item['path']}"
with fs.open(path, "rb", block_size=256 * 1024, cache_type="readahead") as stream:
parquet = pq.ParquetFile(stream)
require(
parquet.schema_arrow.equals(ASSETS_SCHEMA, check_metadata=False),
f"schema mismatch: {item['path']}",
)
rows = parquet.read(columns=columns, use_threads=False).to_pylist()
row_groups = parquet.metadata.num_row_groups
for row in rows:
key = (row["sequence"], row["frame_index"])
require(
key in expected and key not in seen, f"missing/unexpected/duplicate frame: {key}"
)
seen.add(key)
source = mapping[row["sequence"]]
require(
row["frame_id"] == f"{row['sequence']}_{row['frame_index']:06d}",
"frame ID mismatch",
)
require(
(row["raw_date"], row["raw_drive"], row["raw_frame_index"])
== (source.date, source.drive, source.start + row["frame_index"]),
"Raw mapping mismatch",
)
require(row["split"] == source.split and row["camera_id"] == 2, "split/camera mismatch")
counts[source.split] += 1
shards.append(dict(item, rows=len(rows), row_groups=row_groups))
print(f"{item['path']}: {len(rows)} frame identities verified", flush=True)
require(seen == expected, "incomplete source-frame coverage")
local = json.loads((root / "release/v1/manifest.json").read_text())
remote = {item["path"]: item for item in inventory}
for item in local["files"]:
require(
remote[item["path"]]["sha256"] == item["sha256"]
and remote[item["path"]]["bytes"] == item["bytes"],
"metadata snapshot differs from local audit",
)
return {
"repo_id": repo_id,
"revision": info.sha,
"private_at_audit": info.private,
"shards": shards,
"frames_by_split": dict(counts),
"total_frames": len(seen),
"asset_bytes": sum(item["bytes"] for item in shards),
"audit_scope": "All shard schemas, row counts and frame identity/source mapping columns via authenticated ranged reads. Payload checksums are Hub LFS declarations; not a download of every sensor payload.",
"schema_coverage_and_raw_mappings": "pass",
"metadata_files": [remote[item["path"]] for item in local["files"]],
"metadata_sha256_matches_audited_snapshot": True,
}
def main() -> None:
parser = argparse.ArgumentParser(description=__doc__)
parser.add_argument("--repo-id", default="mcocheteux/KITTI-C2L-Dataset")
parser.add_argument("--revision", help="Pinned Hub revision; defaults to current main")
parser.add_argument("--root", type=Path, default=Path(__file__).resolve().parents[1])
parser.add_argument("--output", required=True, type=Path)
args = parser.parse_args()
if args.output.exists():
raise FileExistsError(f"Refusing to overwrite audit: {args.output}")
report = audit(args.repo_id, args.revision, args.root)
args.output.parent.mkdir(parents=True, exist_ok=True)
args.output.write_text(json.dumps(report, indent=2) + "\n")
print(
f"Verified {report['total_frames']} frames in {len(report['shards'])} shards; report: {args.output}"
)
if __name__ == "__main__":
main()
|