Verify recorded source manifest SHA and immutable legacy history

tools/audio/generate_voice.py

@@ -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'))