Datasets:
File size: 5,309 Bytes
84a242a | 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 | #!/usr/bin/env python3
"""Copy readable historical acquisition artifacts without running experiments.
Cloud placeholders are inventoried, never mistaken for readable files. Original
CSVs remain byte-for-byte unchanged inside deterministic per-run ZIP archives.
This utility is for assembling the release from the authors' historical workspace;
it is not required to inspect archives already shipped in the repository.
"""
from __future__ import annotations
import argparse
import collections
import gzip
import hashlib
import json
from pathlib import Path
import zipfile
METHODS = ("plain_mean_ucb", "direct_cost_uc", "relaxed_cost_power_05",
"log_cost", "runtime_uncertainty", "cost_tiers_5")
DATALESS = 0x40000000
def cloud_placeholder(path: Path) -> bool:
return bool(getattr(path.stat(), "st_flags", 0) & DATALESS)
def main() -> None:
parser = argparse.ArgumentParser(description=__doc__)
parser.add_argument("--source-root", type=Path, required=True)
parser.add_argument("--destination", type=Path,
default=Path(__file__).resolve().parents[1])
args = parser.parse_args()
base = args.source_root / "MolAS_AL_restart/AL_MolAS/mean_ucb_cost_5fold_paper"
dest = args.destination.resolve()
out = dest / "results/acquisition"
out.mkdir(parents=True, exist_ok=True)
(dest / "provenance").mkdir(parents=True, exist_ok=True)
records = []
totals = collections.Counter()
def archive(paths, target, prefix):
target.parent.mkdir(parents=True, exist_ok=True)
count = 0
with zipfile.ZipFile(target, "w", compression=zipfile.ZIP_DEFLATED,
compresslevel=6) as z:
for p in sorted(paths):
if not p.is_file() or p.name == ".DS_Store":
continue
rel = p.relative_to(base).as_posix()
rec = {"source": str(p.relative_to(args.source_root)),
"archive": str(target.relative_to(dest)),
"member": p.relative_to(prefix).as_posix(),
"bytes": p.stat().st_size}
if p.is_symlink():
rec["status"] = "excluded_symlink"
elif cloud_placeholder(p):
rec["status"] = "missing_cloud_contents"
else:
try:
raw = p.read_bytes()
rec["sha256"] = hashlib.sha256(raw).hexdigest()
zi = zipfile.ZipInfo(rec["member"], (2026, 10, 31, 0, 0, 0))
zi.compress_type = zipfile.ZIP_DEFLATED
zi.external_attr = 0o644 << 16
z.writestr(zi, raw, compresslevel=6)
rec["status"] = "included_unchanged"
count += 1
totals["included_uncompressed_bytes"] += len(raw)
except OSError as exc:
rec["status"] = "unreadable"
rec["error"] = type(exc).__name__
totals[rec["status"]] += 1
records.append(rec)
if count == 0:
target.unlink()
else:
totals["archive_bytes"] += target.stat().st_size
totals["archives"] += 1
for method in METHODS:
for fold in range(5):
for seed in range(3):
run = base / "outputs/runs" / method / f"fold{fold}" / f"seed{seed}"
# State binaries are not needed to inspect published observations
# or exported matrices. Keep small JSON metadata and event logs.
paths = [p for p in run.rglob("*") if p.is_file()
and p.suffix in (".csv", ".json", ".jsonl", ".txt")]
archive(paths, out / method / f"fold{fold}_seed{seed}.zip", run)
print(f"Packaged available artifacts: {method}", flush=True)
auxiliary = []
for name in ("audits", "summary", "runtime_feature_ablation"):
auxiliary.extend((base / "outputs" / name).rglob("*"))
auxiliary.extend(p for p in (base / "outputs").iterdir() if p.is_file())
archive(auxiliary, out / "historical_summaries.zip", base / "outputs")
manifest = dest / "provenance/acquisition_artifact_inventory.jsonl.gz"
with manifest.open("wb") as raw:
with gzip.GzipFile(filename="", fileobj=raw, mode="wb", mtime=0) as f:
for rec in records:
f.write((json.dumps(rec, sort_keys=True) + "\n").encode())
summary = dict(totals)
summary.update({"methods": list(METHODS), "folds": list(range(5)),
"acquisition_seeds": [0, 1, 2],
"complete": not any(totals[k] for k in
("missing_cloud_contents", "unreadable", "excluded_symlink")),
"note": "Historical files only; no experiments or results regenerated. "
"Downstream paper analyses use acquisition seed 0. "
"Resume-state binaries and smoke/pilot runs are excluded."})
(dest / "provenance/acquisition_archive_summary.json").write_text(
json.dumps(summary, indent=2) + "\n")
print(json.dumps(summary, indent=2))
if __name__ == "__main__":
main()
|