Datasets:
File size: 3,837 Bytes
8a8f710 | 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 | import argparse
import re
import zipfile
from collections import defaultdict
from pathlib import Path
import pyarrow.parquet as pq
DATASET = "BRepGround"
DEFAULT_REPO = "jhlee11/" + DATASET
def get_file(source, relative, cache_dir):
if Path(source).is_dir():
return Path(source) / relative
from huggingface_hub import hf_hub_download
return Path(
hf_hub_download(source, relative, repo_type="dataset", cache_dir=cache_dir)
)
def metadata_files(source, split):
if Path(source).is_dir():
return [
p.relative_to(source).as_posix()
for p in sorted(Path(source).glob(f"data/{split}-*.parquet"))
]
from huggingface_hub import HfApi
return sorted(
p
for p in HfApi().list_repo_files(source, repo_type="dataset")
if p.startswith(f"data/{split}-") and p.endswith(".parquet")
)
def write_file(path, payload):
path.parent.mkdir(parents=True, exist_ok=True)
if path.exists():
if path.read_bytes() != payload:
raise FileExistsError(f"Refusing to replace different content: {path}")
return
temporary = path.with_suffix(path.suffix + ".partial")
temporary.write_bytes(payload)
temporary.replace(path)
def main():
parser = argparse.ArgumentParser(
description="Download or locally export the dataset into individual files."
)
parser.add_argument(
"--source",
default=DEFAULT_REPO,
help="Hub dataset ID or local repository directory",
)
parser.add_argument(
"--split", choices=["train", "validation", "val", "test", "all"], default="all"
)
parser.add_argument(
"--id",
dest="sample_id",
help="Export one exact ID; searches the selected splits",
)
parser.add_argument("--output", type=Path, required=True)
parser.add_argument("--cache-dir", type=Path)
args = parser.parse_args()
split = "validation" if args.split == "val" else args.split
splits = ["train", "validation", "test"] if split == "all" else [split]
assets = defaultdict(set)
count = 0
for split in splits:
files = metadata_files(args.source, split)
if not files:
raise FileNotFoundError(f"No Parquet files for {split} in {args.source}")
for filename in files:
path = get_file(args.source, filename, args.cache_dir)
for batch in pq.ParquetFile(path).iter_batches(batch_size=128):
for row in batch.to_pylist():
if args.sample_id and row["id"] != args.sample_id:
continue
assets[row["asset_archive"]].update(
[row["step_path"], row["label_path"]]
)
count += 1
if not count:
raise ValueError(f"No matching sample: {args.sample_id}")
for archive, members in sorted(assets.items()):
if not re.fullmatch(
"assets/(train|validation|test)-\\d{5}-of-\\d{5}\\.zip", archive
):
raise ValueError(f"Unexpected archive: {archive}")
path = get_file(args.source, archive, args.cache_dir)
with zipfile.ZipFile(path) as z:
for member in sorted(members):
if not re.fullmatch(
"(steps|labels)/\\d{8}_[a-z]+_\\d+\\.(step|json)", member
):
raise ValueError(f"Unexpected asset path: {member}")
write_file(args.output / member, z.read(member))
print(f"Exported {len(members)} files from {archive}", flush=True)
print(f"Exported {count} {DATASET} samples to {args.output.resolve()}")
if __name__ == "__main__":
main()
|