Add caching support to catalog

We will now cache the results of the catalog for 60s and not hit the database at all if cached

data/cache/cache_key.py

@@ -4,5 +4,13 @@ class CacheKey(namedtuple('CacheKey', ['key', 'expiration'])):
   """ Defines a key into the data model cache. """
   pass
 
+
 def for_repository_blob(namespace_name, repo_name, digest):
+  """ Returns a cache key for a blob in a repository. """
   return CacheKey('repository_blob__%s_%s_%s' % (namespace_name, repo_name, digest), '60s')
+
+
+def for_catalog_page(auth_context_key, start_id, limit):
+  """ Returns a cache key for a single page of a catalog lookup for an authed context. """
+  params = (auth_context_key or '(anon)', start_id or 0, limit or 0)
+  return CacheKey('catalog_page__%s_%s_%s' % params, '60s')

endpoints/v2/catalog.py

@@ -2,10 +2,13 @@ import features
 
 from flask import jsonify
 
-from auth.auth_context import get_authenticated_user
+from app import model_cache
+from auth.auth_context import get_authenticated_user, get_authenticated_context
 from auth.registry_jwt_auth import process_registry_jwt_auth
+from data.cache import cache_key
 from endpoints.decorators import anon_protect
 from endpoints.v2 import v2_bp, paginate
+from endpoints.v2.models_interface import Repository
 from endpoints.v2.models_pre_oci import data_model as model
 
 
@@ -14,16 +17,23 @@ from endpoints.v2.models_pre_oci import data_model as model
 @anon_protect
 @paginate()
 def catalog_search(start_id, limit, pagination_callback):
-  include_public = bool(features.PUBLIC_CATALOG)
-  if not include_public and not get_authenticated_user():
-    return jsonify({'repositories': []})
+  def _load_catalog():
+    include_public = bool(features.PUBLIC_CATALOG)
+    if not include_public and not get_authenticated_user():
+      return []
 
-  username = get_authenticated_user().username if get_authenticated_user() else None
-  if username and not get_authenticated_user().enabled:
-    return jsonify({'repositories': []})
+    username = get_authenticated_user().username if get_authenticated_user() else None
+    if username and not get_authenticated_user().enabled:
+      return []
+
+    repos = model.get_visible_repositories(username, start_id, limit, include_public=include_public)
+    return [repo._asdict() for repo in repos]
+
+  context_key = get_authenticated_context().unique_key if get_authenticated_context() else None
+  catalog_cache_key = cache_key.for_catalog_page(context_key, start_id, limit)
+  visible_repositories = [Repository(**repo_dict) for repo_dict
+                          in model_cache.retrieve(catalog_cache_key, _load_catalog)]
 
-  visible_repositories = model.get_visible_repositories(username, start_id, limit,
-                                                        include_public=include_public)
   response = jsonify({
     'repositories': ['%s/%s' % (repo.namespace_name, repo.name)
                      for repo in visible_repositories][0:limit],

test/registry/protocol_v2.py

@@ -390,7 +390,7 @@ class V2Protocol(RegistryProtocol):
     return results
 
   def catalog(self, session, page_size=2, credentials=None, options=None, expected_failure=None,
-              namespace=None, repo_name=None):
+              namespace=None, repo_name=None, bearer_token=None):
     options = options or ProtocolOptions()
     scopes = options.scopes or []
 
@@ -409,6 +409,11 @@ class V2Protocol(RegistryProtocol):
         'Authorization': 'Bearer ' + token,
       }
 
+    if bearer_token is not None:
+      headers = {
+        'Authorization': 'Bearer ' + bearer_token,
+      }
+
     results = []
     url = '/v2/_catalog'
     params = {}

test/registry/registry_tests.py

@@ -719,6 +719,26 @@ def test_catalog(public_catalog, credentials, expected_repos, page_size, v2_prot
     assert set(expected_repos).issubset(set(results))
 
 
+def test_catalog_caching(v2_protocol, basic_images, liveserver_session, app_reloader,
+                         liveserver, registry_server_executor):
+  """ Test: Calling the catalog after initially pulled will result in the catalog being cached. """
+  credentials = ('devtable', 'password')
+
+  # Conduct the initial catalog call to prime the cache.
+  results = v2_protocol.catalog(liveserver_session, credentials=credentials,
+                                namespace='devtable', repo_name='simple')
+
+  token, _ = v2_protocol.auth(liveserver_session, credentials, 'devtable', 'simple')
+
+  # Disconnect the server from the database.
+  registry_server_executor.on(liveserver).break_database()
+
+  # Call the catalog again, which should now be cached.
+  cached_results = v2_protocol.catalog(liveserver_session, bearer_token=token)
+  assert len(cached_results) == len(results)
+  assert set(cached_results) == set(results)
+
+
 @pytest.mark.parametrize('username, namespace, repository', [
   ('devtable', 'devtable', 'simple'),
   ('devtable', 'devtable', 'gargantuan'),