From d057e78efbebc5c4e829beaef82e70e7a9f33eda Mon Sep 17 00:00:00 2001
From: Noa Aviel Dove <nadove@ucsc.edu>
Date: Wed, 10 Apr 2024 00:50:41 -0700
Subject: [PATCH] Include `replica_type` in JSONL manifest

---
 src/azul/service/manifest_service.py |  8 ++++++--
 test/service/test_manifest.py        | 26 ++++++++++++++++----------
 2 files changed, 22 insertions(+), 12 deletions(-)

diff --git a/src/azul/service/manifest_service.py b/src/azul/service/manifest_service.py
index 4962932b2..9d432be00 100644
--- a/src/azul/service/manifest_service.py
+++ b/src/azul/service/manifest_service.py
@@ -2067,7 +2067,7 @@ def _all_replicas(self) -> Iterable[JSON]:
                 replica_id = replica.meta.id
                 if replica_id not in emitted_replica_ids:
                     num_new_replicas += 1
-                    yield replica.contents.to_dict()
+                    yield replica.to_dict()
                     # Note that this will be zero for replicas that use implicit
                     # hubs, in which case there are actually many hubs
                     explicit_hub_count = len(replica.hub_ids)
@@ -2113,6 +2113,10 @@ def create_file(self) -> tuple[str, Optional[str]]:
         os.close(fd)
         with open(path, 'w') as f:
             for replica in self._all_replicas():
-                json.dump(replica, f)
+                entry = {
+                    'contents': replica['contents'],
+                    'type': replica['replica_type']
+                }
+                json.dump(entry, f)
                 f.write('\n')
         return path, None
diff --git a/test/service/test_manifest.py b/test/service/test_manifest.py
index 18ef4928c..8c5ac5d5f 100644
--- a/test/service/test_manifest.py
+++ b/test/service/test_manifest.py
@@ -1317,15 +1317,18 @@ def test_manifest_content_disposition_header(self):
     @manifest_test
     def test_verbatim_jsonl_manifest(self):
         expected = [
-            bundle.metadata_files[d]
+            {
+                'type': replica_type,
+                'contents': bundle.metadata_files[key],
+            }
             for bundle in map(self._load_canned_bundle, self.bundles())
-            for d in [
-                'links.json',
-                'cell_suspension_0.json',
-                'project_0.json',
-                'sequence_file_0.json',
-                'sequence_file_1.json',
-                'specimen_from_organism_0.json'
+            for replica_type, key in [
+                ('links', 'links.json'),
+                ('cell_suspension', 'cell_suspension_0.json'),
+                ('project', 'project_0.json'),
+                ('file', 'sequence_file_0.json'),
+                ('file', 'sequence_file_1.json'),
+                ('sample', 'specimen_from_organism_0.json')
             ]
         ]
         response = self._get_manifest(ManifestFormat.verbatim_jsonl, {})
@@ -2058,8 +2061,11 @@ def test_verbatim_jsonl_manifest(self):
         response = self._get_manifest(ManifestFormat.verbatim_jsonl, filters={})
         self.assertEqual(200, response.status_code)
         expected = [
-            entity
+            {
+                'type': 'anvil_' + entity_ref.entity_type,
+                'contents': entity,
+            }
             for bundle in self.bundles()
-            for entity in self._load_canned_bundle(bundle).entities.values()
+            for entity_ref, entity in self._load_canned_bundle(bundle).entities.items()
         ]
         self._assert_jsonl(expected, response)