Datasets:
File size: 3,604 Bytes
05d6af6 | 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 | """Strip machine-specific values from every `config.json` in the release.
The fit pipeline records absolute paths to local AOT (ahead-of-time compiled)
PyTorch artifacts in `AOT_ARTIFACT_FILE`. Those paths only resolve on the
machine where the fit was produced, and the binary AOT artifacts themselves are
not part of the release. This script nulls those paths in place. All numeric
hyperparameters, seeds, and reproducibility fields are left untouched.
The script is idempotent: keys that are already `None` (or absent) are skipped.
It supports `--dry-run` to preview affected files without writing.
Usage:
python scripts/sanitize_configs.py # write in place
python scripts/sanitize_configs.py --dry-run # report only
python scripts/sanitize_configs.py --root /path/to/sf_release
"""
from __future__ import annotations
import argparse
import json
from pathlib import Path
from typing import Any
# Keys whose values are machine-specific absolute paths and must be nulled.
PATH_KEYS_TO_NULL: tuple[str, ...] = (
"AOT_ARTIFACT_FILE",
)
# Sentinel value used to flag previously-sanitized files.
SANITIZED_MARKER_KEY = "_RELEASE_SANITIZED"
def sanitize_config(data: dict[str, Any]) -> tuple[dict[str, Any], list[str]]:
"""Return (new_data, list_of_changed_keys). Does not modify input."""
new = dict(data)
changed: list[str] = []
for key in PATH_KEYS_TO_NULL:
if key in new and new[key] is not None:
new[key] = None
changed.append(key)
if changed and not new.get(SANITIZED_MARKER_KEY):
new[SANITIZED_MARKER_KEY] = True
return new, changed
def process_file(path: Path, dry_run: bool) -> tuple[bool, list[str]]:
raw = path.read_text()
try:
data = json.loads(raw)
except json.JSONDecodeError as exc:
print(f" SKIP {path}: invalid JSON ({exc})")
return False, []
if not isinstance(data, dict):
print(f" SKIP {path}: not a JSON object")
return False, []
new_data, changed = sanitize_config(data)
if not changed:
return False, []
if not dry_run:
# Preserve 4-space indentation to match the existing files.
path.write_text(json.dumps(new_data, indent=4) + "\n")
return True, changed
def main() -> None:
parser = argparse.ArgumentParser(description=__doc__)
default_root = Path(__file__).resolve().parent.parent
parser.add_argument("--root", type=Path, default=default_root,
help="Release root containing config.json files anywhere underneath.")
parser.add_argument("--dry-run", action="store_true",
help="Report files that would be modified; do not write.")
args = parser.parse_args()
if not args.root.is_dir():
raise SystemExit(f"Release root does not exist: {args.root}")
config_paths = sorted(args.root.rglob("config.json"))
print(f"Scanning {len(config_paths)} config.json files under {args.root}")
print(f"Keys to null: {PATH_KEYS_TO_NULL}")
if args.dry_run:
print("(dry run; no files will be written)")
changed_count = 0
per_key: dict[str, int] = {k: 0 for k in PATH_KEYS_TO_NULL}
for path in config_paths:
was_changed, changed_keys = process_file(path, args.dry_run)
if was_changed:
changed_count += 1
for k in changed_keys:
per_key[k] = per_key.get(k, 0) + 1
print(f"Changed: {changed_count}/{len(config_paths)} configs")
for k, n in per_key.items():
print(f" {k}: {n}")
if __name__ == "__main__":
main()
|