From aeeaceb963731a14f0c0fb4f184e464cfa6beb87 Mon Sep 17 00:00:00 2001
From: Noa Aviel Dove <nadove@ucsc.edu>
Date: Wed, 13 Nov 2024 19:28:57 -0800
Subject: [PATCH] Cover orphans in AnVIL verbatim PFB unit test (#6691)

---
 .../data/verbatim/anvil/pfb_entities.json     | 37 +++++++++++++++++++
 .../data/verbatim/anvil/pfb_schema.json       | 21 +++++++++++
 test/service/test_manifest.py                 |  4 +-
 3 files changed, 61 insertions(+), 1 deletion(-)

diff --git a/test/service/data/verbatim/anvil/pfb_entities.json b/test/service/data/verbatim/anvil/pfb_entities.json
index 29d14d974e..8cb9a00eda 100644
--- a/test/service/data/verbatim/anvil/pfb_entities.json
+++ b/test/service/data/verbatim/anvil/pfb_entities.json
@@ -88,6 +88,13 @@
                     "ontology_reference": "",
                     "properties": [],
                     "values": {}
+                },
+                {
+                    "links": [],
+                    "name": "non_schema_orphan_table",
+                    "ontology_reference": "",
+                    "properties": [],
+                    "values": {}
                 }
             ]
         },
@@ -132,6 +139,16 @@
         },
         "relations": []
     },
+    {
+        "id": "28ed0f3a-157b-417b-a05a-48f57f9d3a34",
+        "name": "non_schema_orphan_table",
+        "object": {
+            "datarepo_row_id": "28ed0f3a-157b-417b-a05a-48f57f9d3a34",
+            "non_schema_column": "eggs",
+            "version": "2022-06-01T00:00:00.000000Z"
+        },
+        "relations": []
+    },
     {
         "id": "826dea02-e274-4ffe-aabc-eb3db63ad068",
         "name": "anvil_biosample",
@@ -279,6 +296,26 @@
         },
         "relations": []
     },
+    {
+        "id": "9687b86d-a2ae-a083-b910-a16bcbef1ba4",
+        "name": "non_schema_orphan_table",
+        "object": {
+            "datarepo_row_id": "9687b86d-a2ae-a083-b910-a16bcbef1ba4",
+            "non_schema_column": "spam",
+            "version": "2022-06-01T00:00:00.000000Z"
+        },
+        "relations": []
+    },
+    {
+        "id": "9db5952c-c454-49d9-8a62-5abb026701c0",
+        "name": "non_schema_orphan_table",
+        "object": {
+            "datarepo_row_id": "9db5952c-c454-49d9-8a62-5abb026701c0",
+            "non_schema_column": "baked beans",
+            "version": "2022-06-01T00:00:00.000000Z"
+        },
+        "relations": []
+    },
     {
         "id": "15b76f9c-6b46-433f-851d-34e89f1b9ba6",
         "name": "anvil_file",
diff --git a/test/service/data/verbatim/anvil/pfb_schema.json b/test/service/data/verbatim/anvil/pfb_schema.json
index 07aee95c55..9bdd6fcf66 100644
--- a/test/service/data/verbatim/anvil/pfb_schema.json
+++ b/test/service/data/verbatim/anvil/pfb_schema.json
@@ -1158,6 +1158,27 @@
                     ],
                     "name": "anvil_variantcallingactivity",
                     "type": "record"
+                },
+                {
+                    "fields": [
+                        {
+                            "name": "datarepo_row_id",
+                            "namespace": "non_schema_orphan_table",
+                            "type": "string"
+                        },
+                        {
+                            "name": "non_schema_column",
+                            "namespace": "non_schema_orphan_table",
+                            "type": "string"
+                        },
+                        {
+                            "name": "version",
+                            "namespace": "non_schema_orphan_table",
+                            "type": "string"
+                        }
+                    ],
+                    "name": "non_schema_orphan_table",
+                    "type": "record"
                 }
             ]
         },
diff --git a/test/service/test_manifest.py b/test/service/test_manifest.py
index dac94047db..260eecf384 100644
--- a/test/service/test_manifest.py
+++ b/test/service/test_manifest.py
@@ -2122,7 +2122,9 @@ def hash_entities(entities: dict[EntityReference, JSON]) -> dict[str, JSON]:
         return all_entities_by_hash.values(), linked_entities_by_hash.values()
 
     def test_verbatim_pfb_manifest(self):
-        response = self._get_manifest(ManifestFormat.verbatim_pfb, filters={})
+        response = self._get_manifest(ManifestFormat.verbatim_pfb, filters={
+            'datasets.dataset_id': {'is': ['52ee7665-7033-63f2-a8d9-ce8e32666739']}
+        })
         self.assertEqual(200, response.status_code)
         with open(self._data_path('service') / 'verbatim/anvil/pfb_schema.json') as f:
             expected_schema = json.load(f)