Verify recorded source manifest SHA and immutable legacy history

This commit is contained in:
NINI
2026-10-05 02:37:55 +09:00
parent 8afff106da
commit ceb063dfdc
3 changed files with 61 additions and 1 deletions
+17
View File
@@ -109,6 +109,23 @@ def load_run_inputs(manifest, root):
validate_inputs(catalog, profile)
if profile['modelRevision'] != manifest.get('modelRevision'):
raise ValueError('input snapshot model revision differs from manifest')
migration_path = Path(root) / 'input-snapshot-migration.json'
if migration_path.exists():
migrated = [record for record in load_json(migration_path)['runs'] if record['runId'] == run_id]
if migrated and (len(migrated) != 1 or file_hash(run_dir / 'manifest.json') != migrated[0]['originalManifestSha256']):
raise ValueError(f'{run_id}: immutable legacy manifest SHA-256 mismatch')
for reference in manifest.get('environment', {}).get('sourceGenerations', []):
source_id = reference['runId']
if not isinstance(source_id, str) or not re.fullmatch(r'[A-Za-z0-9_]+', source_id):
raise ValueError('invalid source generation reference runId')
if reference['manifest'] != f'generated/{source_id}/manifest.json':
raise ValueError('source generation reference must identify its own run manifest')
source_path = Path(root) / reference['manifest']
if file_hash(source_path) != reference['manifestSha256']:
raise ValueError(f'{source_id}: recorded source manifest SHA-256 mismatch')
source = load_json(source_path)
if source.get('runId') != source_id or source.get('environment') != reference.get('environment'):
raise ValueError(f'{source_id}: source generation environment reference differs')
clips = {clip['id']: clip for clip in catalog['clips']}
for entry in manifest.get('clips', []):
clip = clips.get(entry.get('id'))
+33
View File
@@ -223,6 +223,39 @@ class VoiceHistoryTests(unittest.TestCase):
self.assertEqual([], voice.validate_manifest(legacy, self.catalog, self.profile, self.root))
self.assertEqual(saved_manifest, (self.root / 'generated' / manifest['runId'] / 'manifest.json').read_bytes())
def test_reuse_rejects_origin_manifest_device_changed_after_sha_recording(self):
first = self.run_generation()
entries = voice.reusable_entries(self.root, self.catalog, self.profile)
with mock.patch.object(voice.importlib.metadata, 'version', return_value='test-version'), \
mock.patch.object(voice.subprocess, 'run', return_value=mock.Mock(stdout='Test GPU\n')):
environment = voice.reuse_environment(self.profile, entries, self.root)
second = voice.generate(self.catalog, self.profile, self.root, environment=environment)
self.assertEqual([], voice.validate_manifest(second, self.catalog, self.profile, self.root))
changed_reference = copy.deepcopy(second)
changed_reference['environment']['sourceGenerations'][0]['manifestSha256'] = '0' * 64
self.assertTrue(voice.validate_manifest(changed_reference, self.catalog, self.profile, self.root))
changed_environment = copy.deepcopy(second)
changed_environment['environment']['sourceGenerations'][0]['environment']['device'] = 'CORRUPTED_DEVICE_RECORD'
self.assertTrue(voice.validate_manifest(changed_environment, self.catalog, self.profile, self.root))
origin = self.root / 'generated' / first['runId'] / 'manifest.json'
changed = voice.load_json(origin)
changed['environment']['device'] = 'CORRUPTED_DEVICE_RECORD'
voice.write_json(origin, changed)
self.assertTrue(voice.validate_manifest(second, self.catalog, self.profile, self.root))
def test_legacy_migration_sha_seals_manifest_without_freezing_mutable_inputs(self):
manifest = self.run_generation()
path = self.root / 'generated' / manifest['runId'] / 'manifest.json'
voice.write_json(self.root / 'input-snapshot-migration.json', {
'schemaVersion': 1, 'runs': [{'runId': manifest['runId'], 'originalManifestSha256': voice.file_hash(path)}],
'preservedArtifactSha256': {'qwen-profile.json': 'old mutable file hash'}})
voice.write_json(self.root / 'qwen-profile.json', {'seed': 43})
self.assertEqual([], voice.validate_manifest(manifest, self.catalog, self.profile, self.root))
changed = voice.load_json(path)
changed['environment']['device'] = 'CORRUPTED_DEVICE_RECORD'
voice.write_json(path, changed)
self.assertTrue(voice.validate_manifest(changed, self.catalog, self.profile, self.root))
if __name__ == '__main__':
unittest.main()