Datasets:
File size: 11,565 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 99 100 101 102 103 104 105 106 107 108 109 110 111 112 113 114 115 116 117 118 119 120 121 122 123 124 125 126 127 128 129 130 131 132 133 134 135 136 137 138 139 140 141 142 143 144 145 146 147 148 149 150 151 152 153 154 155 156 157 158 159 160 161 162 163 164 165 166 167 168 169 170 171 172 173 174 175 176 177 178 179 180 181 182 183 184 185 186 187 188 189 190 191 192 193 194 195 196 197 198 199 200 201 202 203 204 205 206 207 208 209 210 211 212 213 214 215 216 217 218 219 220 221 | #!/usr/bin/env python3
"""Package saved CSV outputs without training, evaluating models, or plotting.
This is a release-preparation utility written on 2026-10-31. It is NOT the
historical MolAS trainer/evaluator or the paper's plotting implementation.
Only locally resident regular files are opened. macOS iCloud placeholders
remain explicit inventory gaps and can be included by rerunning after download.
"""
from __future__ import annotations
import argparse
import csv
import gzip
import hashlib
import io
import json
import os
from pathlib import Path
import stat
import zipfile
from collections import Counter, defaultdict
POLICIES = (
"plain_mean_ucb", "direct_cost_uc", "relaxed_cost_power_05",
"log_cost", "cost_tiers_5", "runtime_uncertainty",
)
SF_DATALESS = 0x40000000
def file_kind(name: str) -> str:
if name == "overall_summary_metrics.csv":
return "overall_endpoint_summary"
if name.endswith("_predicted_scores.csv"):
return "aggregate_predictions" if name.startswith("overall_") else "fold_predictions"
if name.endswith("_descriptive.csv"):
return "per_complex_outcomes"
if name.endswith("_full.csv"):
return "all_algorithm_endpoint_metrics"
if name.endswith("_algorithm_selection_counts.csv"):
return "algorithm_selection_counts"
if name.startswith("best-"):
return "selected_endpoint_metrics"
return "other_csv"
def write_csv(path: Path, rows: list[dict], fields: list[str]) -> None:
path.parent.mkdir(parents=True, exist_ok=True)
with path.open("w", newline="", encoding="utf-8") as handle:
writer = csv.DictWriter(handle, fieldnames=fields)
writer.writeheader()
writer.writerows(rows)
def main() -> None:
parser = argparse.ArgumentParser(description=__doc__)
parser.add_argument("--source", type=Path, required=True,
help="The original port_as_results_260829 directory")
parser.add_argument("--output", type=Path, required=True,
help="Repository results/downstream directory")
args = parser.parse_args()
source, output = args.source.resolve(), args.output.resolve()
if (not source.is_dir() or source == output or source in output.parents
or output in source.parents):
raise SystemExit("Source must exist and output must be outside the source tree")
output.mkdir(parents=True, exist_ok=True)
entries: list[dict] = []
coverage: dict = defaultdict(lambda: {
"present_fold_directories": set(), "file_count": 0, "packaged_files": 0,
"missing_files": 0, "source_bytes": 0, "packaged_bytes": 0,
"packaged_fold_metrics": set(), "packaged_fold_predictions": set(),
"packaged_fold_descriptive": set(),
})
policy_summaries = []
excluded = sorted(p.name for p in source.iterdir() if p.is_dir() and p.name not in POLICIES)
for policy in POLICIES:
root = source / policy
if not root.is_dir():
raise SystemExit(f"Missing policy directory: {policy}")
archive_name = f"{policy}_seed0.zip"
temporary = output / f".{archive_name}.tmp"
counts: Counter = Counter()
with zipfile.ZipFile(temporary, "w", compression=zipfile.ZIP_DEFLATED,
compresslevel=6, allowZip64=True) as archive:
for base, dirs, filenames in os.walk(root, followlinks=False):
dirs[:] = sorted(d for d in dirs if not (Path(base) / d).is_symlink())
for filename in sorted(filenames):
if not filename.endswith(".csv"):
continue
path = Path(base) / filename
rel = path.relative_to(source)
parts = rel.parts
if len(parts) < 6 or parts[1] != "seed0":
raise SystemExit(f"Unexpected saved-output path; do not infer seed: {rel}")
checkpoint = parts[2]
fold = next((p.removeprefix("fold_") for p in parts if p.startswith("fold_")), "")
metadata = path.lstat()
flags = getattr(metadata, "st_flags", 0)
row = {
"policy": policy, "acquisition_seed_label": "0",
"checkpoint_label": checkpoint, "fold": fold,
"kind": file_kind(filename),
"source_workspace_relative_path": f"port_as_results_260829/{rel.as_posix()}",
"source_bytes": metadata.st_size, "source_flags": flags,
"archive": archive_name, "archive_member": rel.as_posix(),
"status": "", "sha256": "", "csv_rows": "",
"csv_columns": "", "summary_mirror": "",
}
item = coverage[(policy, checkpoint)]
item["file_count"] += 1
item["source_bytes"] += metadata.st_size
if fold:
item["present_fold_directories"].add(fold)
counts["source_files"] += 1
counts["source_bytes"] += metadata.st_size
if not stat.S_ISREG(metadata.st_mode):
row["status"] = "not_regular_file"
elif flags & SF_DATALESS:
row["status"] = "icloud_dataless_not_opened"
elif metadata.st_size > 2_000_000:
row["status"] = "exceeds_bounded_csv_read_limit"
else:
# Metadata was checked before opening: this utility never
# deliberately opens an iCloud placeholder or large binary.
payload = path.read_bytes()
if len(payload) != metadata.st_size:
raise RuntimeError(f"Source changed while reading: {rel}")
row["sha256"] = hashlib.sha256(payload).hexdigest()
try:
reader = csv.reader(io.StringIO(payload.decode("utf-8-sig")))
columns = next(reader)
row["csv_columns"] = json.dumps(columns, ensure_ascii=False)
row["csv_rows"] = sum(1 for _ in reader)
except (UnicodeError, StopIteration, csv.Error) as error:
raise RuntimeError(f"Cannot inspect saved CSV {rel}: {error}") from error
info = zipfile.ZipInfo(rel.as_posix(), date_time=(2026, 10, 31, 0, 0, 0))
info.compress_type = zipfile.ZIP_DEFLATED
info.external_attr = 0o100644 << 16
archive.writestr(info, payload, compress_type=zipfile.ZIP_DEFLATED,
compresslevel=6)
row["status"] = "packaged_byte_for_byte"
counts["packaged_files"] += 1
counts["packaged_bytes"] += len(payload)
item["packaged_files"] += 1
item["packaged_bytes"] += len(payload)
if fold and row["kind"] == "selected_endpoint_metrics":
item["packaged_fold_metrics"].add(fold)
if fold and row["kind"] == "fold_predictions":
item["packaged_fold_predictions"].add(fold)
if fold and row["kind"] == "per_complex_outcomes":
item["packaged_fold_descriptive"].add(fold)
if row["kind"] in {"overall_endpoint_summary", "selected_endpoint_metrics",
"all_algorithm_endpoint_metrics"}:
mirror = Path("summary_tables") / rel
destination = output / mirror
destination.parent.mkdir(parents=True, exist_ok=True)
destination.write_bytes(payload)
row["summary_mirror"] = mirror.as_posix()
if row["status"] != "packaged_byte_for_byte":
item["missing_files"] += 1
counts["missing_files"] += 1
counts["missing_bytes"] += metadata.st_size
entries.append(row)
destination = output / archive_name
temporary.replace(destination)
counts["archive_bytes"] = destination.stat().st_size
for name in ("source_files", "source_bytes", "packaged_files", "packaged_bytes", "missing_files", "missing_bytes"):
counts.setdefault(name, 0)
archive_hash = hashlib.sha256(destination.read_bytes()).hexdigest()
policy_summaries.append({"policy": policy, **dict(counts),
"archive": archive_name, "archive_sha256": archive_hash})
print(json.dumps(policy_summaries[-1]), flush=True)
fields = list(entries[0])
csv_buffer = io.StringIO(newline="")
writer = csv.DictWriter(csv_buffer, fieldnames=fields)
writer.writeheader()
writer.writerows(entries)
with (output / "file_manifest.csv.gz").open("wb") as raw:
with gzip.GzipFile(fileobj=raw, mode="wb", mtime=0) as compressed:
compressed.write(csv_buffer.getvalue().encode("utf-8"))
coverage_rows = []
for (policy, checkpoint), values in sorted(coverage.items()):
row = {"policy": policy, "acquisition_seed_label": 0, "checkpoint_label": checkpoint}
for key, value in values.items():
row[key] = ",".join(sorted(value)) if isinstance(value, set) else value
row["all_saved_files_packaged"] = values["missing_files"] == 0
coverage_rows.append(row)
write_csv(output / "checkpoint_coverage.csv", coverage_rows, list(coverage_rows[0]))
# A later packaging pass can see a different set of resident iCloud files.
# Remove stale generated mirrors so this directory agrees with the manifest.
current_mirrors = {row["summary_mirror"] for row in entries if row["summary_mirror"]}
for mirror in (output / "summary_tables").rglob("*.csv"):
if mirror.relative_to(output).as_posix() not in current_mirrors:
mirror.unlink()
summary = {
"packaging_date": "2026-10-31", "operation": "catalog_and_copy_existing_outputs_only",
"artifact_type": "historical saved-output CSV archives",
"archives_contain_training_or_plotting_source": False,
"source_directory_name": source.name,
"acquisition_seed_labels": [0],
"selector_training_seed": "not established by these files",
"fold_directory_labels": [0, 1, 2, 3, 4],
"checkpoint_unit": "percentage of acquired cells; not realized hours",
"policies": policy_summaries,
"excluded_nonpaper_directories": excluded,
"all_saved_files_packaged": all(p.get("missing_files", 0) == 0 for p in policy_summaries),
"cloud_policy": "Dataless placeholders are inventoried but not read; rerun after hydration",
"scientific_results_modified": False,
"new_scientific_experiments": False,
}
(output / "archive_summary.json").write_text(json.dumps(summary, indent=2) + "\n")
print(json.dumps({"complete": summary["all_saved_files_packaged"],
"inventory_files": len(entries), "checkpoints": len(coverage_rows)}))
if __name__ == "__main__":
main()
|