Make the repository API faster by only checking the log entries table once for each kind of entry, rather than twice. We make use of a special subquery-like syntax, which allows us to count those entries that are both 30 days only and 1 day old in the same query. This was tested successfully on MySQL, Postgres and Sqlite.

data/model/legacy.py

@@ -19,7 +19,7 @@ from data.database import (User, Repository, Image, AccessToken, Role, Repositor
                            db, BUILD_PHASE, QuayUserField, ImageStorageSignature, QueueItem,
                            ImageStorageSignatureKind, validate_database_url, db_for_update,
                            AccessTokenKind, Star, get_epoch_timestamp, RepositoryActionCount)
-from peewee import JOIN_LEFT_OUTER, fn
+from peewee import JOIN_LEFT_OUTER, fn, SQL
 from util.validation import (validate_username, validate_email, validate_password,
                              INVALID_PASSWORD_MESSAGE)
 from util.names import format_robot_username, parse_robot_username
@@ -2777,24 +2777,38 @@ def cancel_repository_build(build, work_queue):
     build.delete_instance()
     return True
 
-def get_repository_pushes(repository, time_delta):
+def _get_repository_events(repository, time_delta, time_delta_earlier, clause):
   since = date.today() - time_delta
-  push_repo = LogEntryKind.get(name = 'push_repo')
-  return (LogEntry.select()
-                  .where(LogEntry.repository == repository)
-                  .where(LogEntry.kind == push_repo)
-                  .where(LogEntry.datetime >= since)
-                  .count())
+  since_earlier = date.today() - time_delta_earlier
 
-def get_repository_pulls(repository, time_delta):
-  since = date.today() - time_delta
+  if since_earlier >= since:
+    raise Exception('time_delta_earlier must be greater than time_delta')
+
+  # This uses a CASE WHEN inner clause to further filter the count.
+  formatted = since.strftime('%Y-%m-%d')
+  case_query = 'CASE WHEN datetime >= \'%s\' THEN 1 ELSE 0 END' % formatted
+
+  result = (LogEntry.select(fn.Sum(SQL(case_query)), fn.Count(SQL('*')))
+                    .where(LogEntry.repository == repository)
+                    .where(clause)
+                    .where(LogEntry.datetime >= since_earlier)
+                    .tuples()
+                    .get())
+
+  return (result[0] or 0, result[1] or 0)
+
+
+def get_repository_pushes(repository, time_delta, time_delta_earlier):
+  push_repo = LogEntryKind.get(name = 'push_repo')
+  clauses = (LogEntry.kind == push_repo)
+  return _get_repository_events(repository, time_delta, time_delta_earlier, clauses)
+
+
+def get_repository_pulls(repository, time_delta, time_delta_earlier):
   repo_pull = LogEntryKind.get(name = 'pull_repo')
   repo_verb = LogEntryKind.get(name = 'repo_verb')
-  return (LogEntry.select()
-                  .where(LogEntry.repository == repository)
-                  .where((LogEntry.kind == repo_pull) | (LogEntry.kind == repo_verb))
-                  .where(LogEntry.datetime >= since)
-                  .count())
+  clauses = ((LogEntry.kind == repo_pull) | (LogEntry.kind == repo_verb))
+  return _get_repository_events(repository, time_delta, time_delta_earlier, clauses)
 
 
 def get_repository_usage():