From 45e51b642c63081b6fcc410f882640fe74df0c93 Mon Sep 17 00:00:00 2001
From: Shelley Gao <syugao@gmail.com>
Date: Fri, 24 May 2024 14:50:08 -0700
Subject: [PATCH] Added snapshot to add bulk test

---
 tests/__snapshots__/test_add_bulk.ambr | 453 +++++++++++++++++++++++++
 tests/test_add_bulk.py                 |  24 +-
 virtool_cli/ref/event_index_cache.py   |   6 +-
 3 files changed, 472 insertions(+), 11 deletions(-)
 create mode 100644 tests/__snapshots__/test_add_bulk.ambr

diff --git a/tests/__snapshots__/test_add_bulk.ambr b/tests/__snapshots__/test_add_bulk.ambr
new file mode 100644
index 00000000..99538560
--- /dev/null
+++ b/tests/__snapshots__/test_add_bulk.ambr
@@ -0,0 +1,453 @@
+# serializer version: 1
+# name: test_add_ref_mini
+  dict({
+    'acronym': '',
+    'excluded_accessions': list([
+    ]),
+    'legacy_id': None,
+    'molecule': None,
+    'name': 'Nanovirus-like particle',
+    'schema': list([
+    ]),
+    'taxid': 96892,
+  })
+# ---
+# name: test_add_ref_mini.1
+  set({
+    'NC_005954',
+  })
+# ---
+# name: test_add_ref_mini.10
+  dict({
+    'acronym': '',
+    'excluded_accessions': list([
+    ]),
+    'legacy_id': None,
+    'molecule': None,
+    'name': 'Lagenaria siceraria endornavirus-California',
+    'schema': list([
+    ]),
+    'taxid': 1468454,
+  })
+# ---
+# name: test_add_ref_mini.11
+  set({
+    'NC_023641',
+  })
+# ---
+# name: test_add_ref_mini.12
+  dict({
+    'acronym': '',
+    'excluded_accessions': list([
+    ]),
+    'legacy_id': None,
+    'molecule': None,
+    'name': 'Dahlia latent viroid',
+    'schema': list([
+    ]),
+    'taxid': 1278205,
+  })
+# ---
+# name: test_add_ref_mini.13
+  set({
+    'NC_020160',
+  })
+# ---
+# name: test_add_ref_mini.14
+  dict({
+    'acronym': '',
+    'excluded_accessions': list([
+    ]),
+    'legacy_id': None,
+    'molecule': None,
+    'name': 'Pagoda yellow mosaic associated virus',
+    'schema': list([
+    ]),
+    'taxid': 1505530,
+  })
+# ---
+# name: test_add_ref_mini.15
+  set({
+    'NC_024301',
+  })
+# ---
+# name: test_add_ref_mini.16
+  dict({
+    'acronym': '',
+    'excluded_accessions': list([
+    ]),
+    'legacy_id': None,
+    'molecule': None,
+    'name': 'Maclura mosaic virus',
+    'schema': list([
+    ]),
+    'taxid': 48201,
+  })
+# ---
+# name: test_add_ref_mini.17
+  set({
+    'NC_043133',
+  })
+# ---
+# name: test_add_ref_mini.18
+  dict({
+    'acronym': '',
+    'excluded_accessions': list([
+    ]),
+    'legacy_id': None,
+    'molecule': None,
+    'name': 'Gaillardia latent virus',
+    'schema': list([
+    ]),
+    'taxid': 1468172,
+  })
+# ---
+# name: test_add_ref_mini.19
+  set({
+    'NC_023892',
+  })
+# ---
+# name: test_add_ref_mini.2
+  dict({
+    'acronym': '',
+    'excluded_accessions': list([
+    ]),
+    'legacy_id': None,
+    'molecule': None,
+    'name': 'Saccharum streak virus',
+    'schema': list([
+    ]),
+    'taxid': 683179,
+  })
+# ---
+# name: test_add_ref_mini.20
+  dict({
+    'acronym': '',
+    'excluded_accessions': list([
+    ]),
+    'legacy_id': None,
+    'molecule': None,
+    'name': 'Ugandan cassava brown streak virus',
+    'schema': list([
+    ]),
+    'taxid': 946046,
+  })
+# ---
+# name: test_add_ref_mini.21
+  set({
+    'NC_014791',
+  })
+# ---
+# name: test_add_ref_mini.22
+  dict({
+    'acronym': '',
+    'excluded_accessions': list([
+    ]),
+    'legacy_id': None,
+    'molecule': None,
+    'name': 'Faba bean necrotic stunt alphasatellite 1',
+    'schema': list([
+    ]),
+    'taxid': 1441799,
+  })
+# ---
+# name: test_add_ref_mini.23
+  set({
+    'NC_023881',
+  })
+# ---
+# name: test_add_ref_mini.24
+  dict({
+    'acronym': '',
+    'excluded_accessions': list([
+    ]),
+    'legacy_id': None,
+    'molecule': None,
+    'name': 'Kalanchoe latent virus',
+    'schema': list([
+    ]),
+    'taxid': 132477,
+  })
+# ---
+# name: test_add_ref_mini.25
+  set({
+    'NC_013006',
+  })
+# ---
+# name: test_add_ref_mini.26
+  dict({
+    'acronym': '',
+    'excluded_accessions': list([
+    ]),
+    'legacy_id': None,
+    'molecule': None,
+    'name': 'Tall oatgrass mosaic virus',
+    'schema': list([
+    ]),
+    'taxid': 1414644,
+  })
+# ---
+# name: test_add_ref_mini.27
+  set({
+    'NC_022745',
+  })
+# ---
+# name: test_add_ref_mini.28
+  dict({
+    'acronym': '',
+    'excluded_accessions': list([
+    ]),
+    'legacy_id': None,
+    'molecule': None,
+    'name': 'Oat blue dwarf virus',
+    'schema': list([
+    ]),
+    'taxid': 56879,
+  })
+# ---
+# name: test_add_ref_mini.29
+  set({
+    'NC_001793',
+  })
+# ---
+# name: test_add_ref_mini.3
+  set({
+    'NC_013464',
+  })
+# ---
+# name: test_add_ref_mini.30
+  dict({
+    'acronym': '',
+    'excluded_accessions': list([
+    ]),
+    'legacy_id': None,
+    'molecule': None,
+    'name': 'East African cassava mosaic Cameroon virus',
+    'schema': list([
+    ]),
+    'taxid': 223262,
+  })
+# ---
+# name: test_add_ref_mini.31
+  set({
+    'NC_004625',
+    'NC_004630',
+  })
+# ---
+# name: test_add_ref_mini.32
+  dict({
+    'acronym': '',
+    'excluded_accessions': list([
+    ]),
+    'legacy_id': None,
+    'molecule': None,
+    'name': 'Babaco mosaic virus',
+    'schema': list([
+    ]),
+    'taxid': 2060511,
+  })
+# ---
+# name: test_add_ref_mini.33
+  set({
+    'NC_036587',
+  })
+# ---
+# name: test_add_ref_mini.34
+  dict({
+    'acronym': '',
+    'excluded_accessions': list([
+    ]),
+    'legacy_id': None,
+    'molecule': None,
+    'name': 'Habenaria mosaic virus',
+    'schema': list([
+    ]),
+    'taxid': 1324128,
+  })
+# ---
+# name: test_add_ref_mini.35
+  set({
+    'NC_021786',
+  })
+# ---
+# name: test_add_ref_mini.36
+  dict({
+    'acronym': '',
+    'excluded_accessions': list([
+    ]),
+    'legacy_id': None,
+    'molecule': None,
+    'name': 'Cabbage leaf curl Jamaica virus',
+    'schema': list([
+    ]),
+    'taxid': 345184,
+  })
+# ---
+# name: test_add_ref_mini.37
+  set({
+    'DQ178610',
+    'DQ178611',
+    'DQ178613',
+    'DQ178614',
+  })
+# ---
+# name: test_add_ref_mini.38
+  dict({
+    'acronym': '',
+    'excluded_accessions': list([
+    ]),
+    'legacy_id': None,
+    'molecule': None,
+    'name': 'Vallota mosaic virus',
+    'schema': list([
+    ]),
+    'taxid': 430059,
+  })
+# ---
+# name: test_add_ref_mini.39
+  set({
+    'NC_043170',
+  })
+# ---
+# name: test_add_ref_mini.4
+  dict({
+    'acronym': '',
+    'excluded_accessions': list([
+    ]),
+    'legacy_id': None,
+    'molecule': None,
+    'name': 'Abaca bunchy top virus',
+    'schema': list([
+    ]),
+    'taxid': 438782,
+  })
+# ---
+# name: test_add_ref_mini.40
+  dict({
+    'acronym': '',
+    'excluded_accessions': list([
+    ]),
+    'legacy_id': None,
+    'molecule': None,
+    'name': 'Impatiens flower break virus',
+    'schema': list([
+    ]),
+    'taxid': 2170135,
+  })
+# ---
+# name: test_add_ref_mini.41
+  set({
+    'NC_030236',
+  })
+# ---
+# name: test_add_ref_mini.42
+  dict({
+    'acronym': '',
+    'excluded_accessions': list([
+    ]),
+    'legacy_id': None,
+    'molecule': None,
+    'name': 'Yacon necrotic mottle virus',
+    'schema': list([
+    ]),
+    'taxid': 1561150,
+  })
+# ---
+# name: test_add_ref_mini.43
+  set({
+    'NC_026472',
+  })
+# ---
+# name: test_add_ref_mini.44
+  dict({
+    'acronym': '',
+    'excluded_accessions': list([
+    ]),
+    'legacy_id': None,
+    'molecule': None,
+    'name': 'Jacquemontia mosaic Yucatan virus',
+    'schema': list([
+    ]),
+    'taxid': 1198450,
+  })
+# ---
+# name: test_add_ref_mini.45
+  set({
+    'NC_038796',
+    'NC_038797',
+  })
+# ---
+# name: test_add_ref_mini.46
+  dict({
+    'acronym': '',
+    'excluded_accessions': list([
+    ]),
+    'legacy_id': None,
+    'molecule': None,
+    'name': 'Radish leaf curl betasatellite',
+    'schema': list([
+    ]),
+    'taxid': 662596,
+  })
+# ---
+# name: test_add_ref_mini.47
+  set({
+    'NC_010239',
+  })
+# ---
+# name: test_add_ref_mini.5
+  set({
+    'EF546802',
+    'EF546803',
+    'EF546804',
+    'EF546805',
+    'EF546806',
+    'EF546807',
+    'NC_010314',
+    'NC_010315',
+    'NC_010316',
+    'NC_010317',
+    'NC_010318',
+    'NC_010319',
+  })
+# ---
+# name: test_add_ref_mini.6
+  dict({
+    'acronym': '',
+    'excluded_accessions': list([
+    ]),
+    'legacy_id': None,
+    'molecule': None,
+    'name': 'Wasabi mottle virus',
+    'schema': list([
+    ]),
+    'taxid': 1169032,
+  })
+# ---
+# name: test_add_ref_mini.7
+  set({
+    'AB017504',
+    'KJ207375',
+    'MH200607',
+    'MK431779',
+    'NC_003355',
+  })
+# ---
+# name: test_add_ref_mini.8
+  dict({
+    'acronym': '',
+    'excluded_accessions': list([
+    ]),
+    'legacy_id': None,
+    'molecule': None,
+    'name': 'Zantedeschia mild mosaic virus',
+    'schema': list([
+    ]),
+    'taxid': 270478,
+  })
+# ---
+# name: test_add_ref_mini.9
+  set({
+    'NC_011560',
+  })
+# ---
diff --git a/tests/test_add_bulk.py b/tests/test_add_bulk.py
index 17bf2e8d..0fd17c21 100644
--- a/tests/test_add_bulk.py
+++ b/tests/test_add_bulk.py
@@ -2,16 +2,16 @@
 import shutil
 import subprocess
 
-
-@pytest.fixture()
-def cache_example_path(test_files_path):
-    return test_files_path / "cache_test"
+from syrupy import SnapshotAssertion
+from syrupy.filters import props
 
 
 @pytest.fixture
-def precached_repo(empty_repo, cache_example_path):
+def precached_repo(empty_repo, test_files_path):
     shutil.copytree(
-        cache_example_path, (empty_repo.path / ".cache/ncbi"), dirs_exist_ok=True
+        (test_files_path / "cache_test"),
+        (empty_repo.path / ".cache/ncbi"),
+        dirs_exist_ok=True,
     )
 
     yield empty_repo
@@ -63,7 +63,9 @@ def ref_table_of_contents() -> dict[int, list[str]]:
 
 
 @pytest.mark.ncbi()
-def test_init_and_add(ref_table_of_contents, precached_repo):
+def test_add_ref_mini(
+    ref_table_of_contents, precached_repo, snapshot: SnapshotAssertion
+):
     for taxid in ref_table_of_contents:
         subprocess.run(
             [
@@ -95,7 +97,9 @@ def test_init_and_add(ref_table_of_contents, precached_repo):
 
     for taxid in ref_table_of_contents:
         otu = precached_repo.get_otu_by_taxid(taxid)
-        assert otu is not None
+        assert otu.dict() == snapshot(exclude=props("id", "isolates"))
+        assert otu.accessions == snapshot
 
-        for accession in ref_table_of_contents[taxid]:
-            assert accession.split(".")[0] in otu.accessions
+        assert otu.accessions == set(
+            accession.split(".")[0] for accession in ref_table_of_contents[taxid]
+        )
diff --git a/virtool_cli/ref/event_index_cache.py b/virtool_cli/ref/event_index_cache.py
index 612c5014..61e20bef 100644
--- a/virtool_cli/ref/event_index_cache.py
+++ b/virtool_cli/ref/event_index_cache.py
@@ -6,7 +6,7 @@
 from collections.abc import Generator
 
 import orjson
-from pydantic import Field, ValidationError
+from pydantic import Field, ValidationError, model_validator
 from pydantic.dataclasses import dataclass
 
 from structlog import get_logger
@@ -24,6 +24,10 @@ class OTUEventCache:
     events: list[int]
     """A list of event IDs"""
 
+    @model_validator(mode="after")
+    def sort_events(self):
+        return self.events.sort()
+
 
 class EventIndexCacheError(Exception):
     """Event index cache is corrupted."""