Use environment variable to determine whether to filter dead links by…

… default; fix flaky tests (#1043)

* Stop filtering dead links in tests unless necessary

* Allow using env var to stop dead link filtering by default

api/catalog/api/serializers/media_serializers.py

@@ -91,7 +91,7 @@ class MediaSearchRequestSerializer(serializers.Serializer):
         label="filter_dead",
         help_text="Control whether 404 links are filtered out.",
         required=False,
-        default=True,
+        default=settings.FILTER_DEAD_LINKS_BY_DEFAULT,
     )
     extension = serializers.CharField(
         label="extension",

api/catalog/settings.py

@@ -33,6 +33,10 @@
 # Where to collect static files in production/development deployments
 STATIC_ROOT = config("STATIC_ROOT", default="/var/api_static_content/static")
 
+FILTER_DEAD_LINKS_BY_DEFAULT = config(
+    "FILTER_DEAD_LINKS_BY_DEFAULT", cast=bool, default=True
+)
+
 # Quick-start development settings - unsuitable for production
 # See https://docs.djangoproject.com/en/2.0/howto/deployment/checklist/
 

api/env.docker

@@ -22,3 +22,5 @@ SEMANTIC_VERSION=1.0.0
 ELASTICSEARCH_URL=es
 
 WATERMARK_ENABLED=True
+
+FILTER_DEAD_LINKS_BY_DEFAULT=False

api/env.template

@@ -69,3 +69,5 @@ IS_PROXIED=False
 
 #SENTRY_SAMPLE_RATE=1.0
 #SENTRY_DSN=
+
+#FILTER_DEAD_LINKS_BY_DEFAULT=False

api/test/audio_integration_test.py

@@ -50,9 +50,7 @@ def force_validity(query_response):
 
 @pytest.fixture
 def audio_fixture(force_result_validity):
-    res = requests.get(
-        f"{API_URL}/v1/audio/", data={"filter_dead": False}, verify=False
-    )
+    res = requests.get(f"{API_URL}/v1/audio/", verify=False)
     parsed = res.json()
     force_result_validity(parsed)
     assert res.status_code == 200
@@ -67,7 +65,7 @@ def jamendo_audio_fixture(force_result_validity):
     """
     res = requests.get(
         f"{API_URL}/v1/audio/",
-        data={"source": "jamendo", "filter_dead": False},
+        data={"source": "jamendo"},
         verify=False,
     )
     parsed = res.json()
@@ -133,7 +131,7 @@ def test_audio_detail_without_thumb():
 
 def test_audio_search_without_thumb():
     """The first audio of this search should not have a thumbnail."""
-    resp = requests.get(f"{API_URL}/v1/audio/?q=zaus&filter_dead=false")
+    resp = requests.get(f"{API_URL}/v1/audio/?q=zaus")
     assert resp.status_code == 200
     parsed = json.loads(resp.text)
     assert parsed["results"][0]["thumbnail"] is None

api/test/auth_test.py

@@ -98,7 +98,7 @@ def test_auth_rate_limit_reporting(
 
 @pytest.mark.django_db
 def test_page_size_limit_unauthed(client):
-    query_params = {"filter_dead": False, "page_size": 20}
+    query_params = {"page_size": 20}
     res = client.get("/v1/images/", query_params)
     assert res.status_code == 200
     query_params["page_size"] = 21
@@ -110,10 +110,10 @@ def test_page_size_limit_unauthed(client):
 def test_page_size_limit_authed(client, test_auth_token_exchange):
     time.sleep(1)
     token = test_auth_token_exchange["access_token"]
-    query_params = {"filter_dead": False, "page_size": 21}
+    query_params = {"page_size": 21}
     res = client.get("/v1/images/", query_params, HTTP_AUTHORIZATION=f"Bearer {token}")
     assert res.status_code == 200
 
-    query_params = {"filter_dead": False, "page_size": 500}
+    query_params = {"page_size": 500}
     res = client.get("/v1/images/", query_params, HTTP_AUTHORIZATION=f"Bearer {token}")
     assert res.status_code == 200

api/test/media_integration.py

@@ -47,12 +47,13 @@ def search_quotes(media_path, q="test"):
 
 def search_quotes_exact(media_path, q):
     """Only returns exact matches for the given query"""
-    unquoted_response = requests.get(f"{API_URL}/v1/{media_path}?q={q}", verify=False)
+    url_format = f"{API_URL}/v1/{media_path}?q={{q}}"
+    unquoted_response = requests.get(url_format.format(q=q), verify=False)
     assert unquoted_response.status_code == 200
     unquoted_result_count = unquoted_response.json()["result_count"]
     assert unquoted_result_count > 0
 
-    quoted_response = requests.get(f'{API_URL}/v1/{media_path}?q="{q}"', verify=False)
+    quoted_response = requests.get(url_format.format(q=f'"{q}"'), verify=False)
     assert quoted_response.status_code == 200
     quoted_result_count = quoted_response.json()["result_count"]
     assert quoted_result_count > 0

api/test/search_qa.py

@@ -20,9 +20,7 @@ class QAScores(Enum):
 
 @pytest.mark.skip(reason="This test is nondeterministic")
 def test_phrase_relevance():
-    res = requests.get(
-        f"{API_URL}/image/search?q=home office&filter_dead=false&qa=true"
-    )
+    res = requests.get(f"{API_URL}/image/search?q=home office&qa=true")
     parsed = json.loads(res.text)
     pprint.pprint(parsed)
     assert int(parsed["results"][0]["id"]) == QAScores.TARGET.value