Unionize the mega query - It needed more performance-based benefits

data/model/_basequery.py

@@ -25,7 +25,18 @@ def filter_to_repos_for_user(query, username=None, namespace=None, include_publi
   if not include_public and not username:
     return Repository.select().where(Repository.id == '-1')
 
-  where_clause = None
+  # Build a set of queries that, when unioned together, return the full set of visible repositories
+  # for the filters specified.
+  queries = []
+
+  where_clause = (True)
+  if namespace:
+    where_clause = (Namespace.username == namespace)
+
+  if include_public:
+    queries.append(query.clone()
+      .where(Repository.visibility == get_public_repo_visibility(), where_clause))
+
   if username:
     UserThroughTeam = User.alias()
     Org = User.alias()
@@ -33,37 +44,32 @@ def filter_to_repos_for_user(query, username=None, namespace=None, include_publi
     AdminTeamMember = TeamMember.alias()
     AdminUser = User.alias()
 
-    query = (query
-             .switch(RepositoryPermission)
-             .join(User, JOIN_LEFT_OUTER)
-             .switch(RepositoryPermission)
-             .join(Team, JOIN_LEFT_OUTER)
-             .join(TeamMember, JOIN_LEFT_OUTER)
-             .join(UserThroughTeam, JOIN_LEFT_OUTER, on=(UserThroughTeam.id == TeamMember.user))
-             .switch(Repository)
-             .join(Org, JOIN_LEFT_OUTER, on=(Repository.namespace_user == Org.id))
-             .join(AdminTeam, JOIN_LEFT_OUTER, on=(Org.id == AdminTeam.organization))
-             .join(TeamRole, JOIN_LEFT_OUTER, on=(AdminTeam.role == TeamRole.id))
-             .switch(AdminTeam)
-             .join(AdminTeamMember, JOIN_LEFT_OUTER, on=(AdminTeam.id == AdminTeamMember.team))
-             .join(AdminUser, JOIN_LEFT_OUTER, on=(AdminTeamMember.user == AdminUser.id)))
+    # Add repositories in which the user has permission.
+    queries.append(query.clone()
+      .switch(RepositoryPermission)
+      .join(User)
+      .where(User.username == username, where_clause))
 
-    where_clause = ((User.username == username) | (UserThroughTeam.username == username) |
-                    ((AdminUser.username == username) & (TeamRole.name == 'admin')))
+    # Add repositories in which the user is a member of a team that has permission.
+    queries.append(query.clone()
+      .switch(RepositoryPermission)
+      .join(Team)
+      .join(TeamMember)
+      .join(UserThroughTeam, on=(UserThroughTeam.id == TeamMember.user))
+      .where(UserThroughTeam.username == username, where_clause))
 
-    if namespace:
-      where_clause = where_clause & (Namespace.username == namespace)
+    # Add repositories under namespaces in which the user is the org admin.
+    queries.append(query.clone()
+      .switch(Repository)
+      .join(Org, on=(Repository.namespace_user == Org.id))
+      .join(AdminTeam, on=(Org.id == AdminTeam.organization))
+      .join(TeamRole, on=(AdminTeam.role == TeamRole.id))
+      .switch(AdminTeam)
+      .join(AdminTeamMember, on=(AdminTeam.id == AdminTeamMember.team))
+      .join(AdminUser, on=(AdminTeamMember.user == AdminUser.id))
+      .where(AdminUser.username == username, where_clause))
 
-  # TODO(jschorr, jake): Figure out why the old join on Visibility was so darn slow and
-  # remove this hack.
-  if include_public:
-    new_clause = (Repository.visibility == get_public_repo_visibility())
-    if where_clause:
-      where_clause = where_clause | new_clause
-    else:
-      where_clause = new_clause
-
-  return query.where(where_clause)
+  return reduce(lambda l, r: l | r, queries)
 
 
 def get_user_organizations(username):

data/model/repository.py

@@ -251,9 +251,6 @@ def get_visible_repositories(username, namespace=None, page=None, limit=None, in
   if limit:
     query = query.limit(limit)
 
-  if namespace:
-    query = query.where(Namespace.username == namespace)
-
   return query
 
 

data/model/tag.py

@@ -133,8 +133,7 @@ def list_repository_tag_history(repo_obj, page=1, size=100, specific_tag=None):
            .join(Image)
            .where(RepositoryTag.repository == repo_obj)
            .where(RepositoryTag.hidden == False)
-           .order_by(RepositoryTag.lifetime_start_ts.desc())
-           .order_by(RepositoryTag.name)
+           .order_by(RepositoryTag.lifetime_start_ts.desc(), RepositoryTag.name)
            .paginate(page, size))
 
   if specific_tag:

data/model/user.py

@@ -496,12 +496,11 @@ def get_matching_user_namespaces(namespace_prefix, username, limit=10):
   base_query = (Namespace
                 .select()
                 .distinct()
-                .limit(limit)
                 .join(Repository, on=(Repository.namespace_user == Namespace.id))
                 .join(RepositoryPermission, JOIN_LEFT_OUTER)
                 .where(Namespace.username ** (namespace_prefix + '%')))
 
-  return _basequery.filter_to_repos_for_user(base_query, username)
+  return _basequery.filter_to_repos_for_user(base_query, username).limit(limit)
 
 def get_matching_users(username_prefix, robot_namespace=None,
                        organization=None):

test/test_image_sharing.py

@@ -217,3 +217,7 @@ class TestImageSharing(unittest.TestCase):
     still_uploading.save()
 
     self.assertDifferentStorage('an-image', still_uploading)
+
+
+if __name__ == '__main__':
+  unittest.main()