Digital Drosophila Archive
51 lines
3.0 KiB
Python
51 lines
3.0 KiB
Python
"""Generate checksums for this publication only, after real download success."""
|
|
from datetime import datetime, timezone
|
|
import json
|
|
from pathlib import Path
|
|
import subprocess
|
|
|
|
from publish_archives import canonical, no_links, sha_file, write_new
|
|
|
|
|
|
def main():
|
|
root = no_links(Path.cwd())
|
|
protected = Path('/data/digital-drosophila')
|
|
if root == protected or protected in root.parents or not (root / '.git').is_dir():
|
|
raise ValueError('Only an isolated publication checkout is allowed')
|
|
folder = root / 'registry/publication-20261001'
|
|
output, checksums = folder / 'publication-manifest.json', folder / 'SHA256SUMS'
|
|
if output.exists() or checksums.exists():
|
|
raise ValueError('Never overwrite a publication manifest')
|
|
transport = json.loads((folder / 'fresh-download-verification.json').read_bytes())
|
|
guard = json.loads((folder / 'source-guard-final.json').read_bytes())
|
|
head = subprocess.check_output(['git', 'rev-parse', 'HEAD'], text=True).strip()
|
|
if not (transport['pass_all'] and transport['fresh_download_proven'] and guard['pass_all']
|
|
and transport['expected_commit'] == head):
|
|
raise ValueError('Actual transport/source evidence or commit binding is missing')
|
|
files = [root / p for p in ('.gitattributes', '.gitignore', 'NOTICE.md')]
|
|
for prefix in ('artifacts/public-resources-20261001', 'artifacts/lfs-smoke',
|
|
'registry/publication-20261001', 'tools/lfs-publication'):
|
|
for p in (root / prefix).rglob('*'):
|
|
no_links(p)
|
|
if '__pycache__' not in p.parts and p.is_file():
|
|
files.append(p)
|
|
rows = [dict(path=p.relative_to(root).as_posix(), **sha_file(p)) for p in sorted(files)]
|
|
if len({row['path'] for row in rows}) != len(rows):
|
|
raise ValueError('Duplicate publication path')
|
|
manifest = dict(schema='lfs-publication-final-manifest-v1', pass_all=True,
|
|
generated_utc=datetime.now(timezone.utc).isoformat(), data_commit=head,
|
|
branch='codex/lfs-public-resources', repository_public=True,
|
|
aggregate_inventory_sha256=transport['aggregate_inventory_sha256'],
|
|
logical_files=len(rows), logical_bytes=sum(r['bytes'] for r in rows),
|
|
files=rows,
|
|
scope='Publication payload, tools, notices and evidence only; original file SHA are nested in members.jsonl.gz. Not an entire server backup or scientific pass.',
|
|
self_hash_policy='Manifest is covered by SHA256SUMS; SHA256SUMS self-hash is omitted and its bytes are bound by the final Git commit.')
|
|
write_new(output, canonical(manifest))
|
|
rows.append(dict(path=output.relative_to(root).as_posix(), **sha_file(output)))
|
|
write_new(checksums, ''.join(r['sha256'] + ' ' + r['path'] + '\n' for r in sorted(rows, key=lambda r: r['path'])).encode())
|
|
print(json.dumps(dict(pass_all=True, files=len(rows), manifest=sha_file(output), checksums=sha_file(checksums))))
|
|
|
|
|
|
if __name__ == '__main__':
|
|
main()
|