Reject duplicate source generation references

tools/audio/generate_voice.py

@@ -114,10 +114,14 @@ def load_run_inputs(manifest, root):
         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')
+    source_ids = set()
     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 source_id in source_ids:
+            raise ValueError('duplicate source generation reference runId')
+        source_ids.add(source_id)
         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']

tools/audio/test_generate_voice.py

@@ -237,6 +237,12 @@ class VoiceHistoryTests(unittest.TestCase):
         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))
+        duplicated = copy.deepcopy(second)
+        duplicated['environment']['sourceGenerations'].append(copy.deepcopy(duplicated['environment']['sourceGenerations'][0]))
+        self.assertTrue(voice.validate_manifest(duplicated, self.catalog, self.profile, self.root))
+        changed_path = copy.deepcopy(second)
+        changed_path['environment']['sourceGenerations'][0]['manifest'] = '../outside/manifest.json'
+        self.assertTrue(voice.validate_manifest(changed_path, 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'