Add a RepositoryActionCount table so we can use it (instead of LogEntry) when scoring repo search results

workers/repositoryactioncounter.py

@@ -0,0 +1,51 @@
+import logging
+
+from apscheduler.schedulers.blocking import BlockingScheduler
+
+from data.database import Repository, LogEntry, RepositoryActionCount, db_random_func, fn
+from datetime import date, datetime, timedelta
+
+POLL_PERIOD_SECONDS = 30
+
+logger = logging.getLogger(__name__)
+sched = BlockingScheduler()
+
+@sched.scheduled_job(trigger='interval', seconds=10)
+def count_repository_actions():
+  """ Counts actions for a random repository for the previous day. """
+
+  try:
+    # Get a random repository to count.
+    today = date.today()
+    yesterday = today - timedelta(days=1)
+    has_yesterday_actions = (RepositoryActionCount.select(RepositoryActionCount.repository)
+                                                  .where(RepositoryActionCount.date == yesterday))
+
+    to_count = (Repository.select()
+                          .where(~(Repository.id << (has_yesterday_actions)))
+                          .order_by(db_random_func()).get())
+
+    logger.debug('Counting: %s', to_count.id)
+
+    actions = (LogEntry.select()
+                       .where(LogEntry.repository == to_count,
+                              LogEntry.datetime >= yesterday,
+                              LogEntry.datetime < today)
+                       .count())
+
+    # Create the row.
+    try:
+      RepositoryActionCount.create(repository=to_count, date=yesterday, count=actions)
+    except:
+      logger.exception('Exception when writing count')
+
+    return True
+
+  except Repository.DoesNotExist:
+    logger.debug('No further repositories to count')
+    return False
+
+
+if __name__ == "__main__":
+  logging.basicConfig(level=logging.DEBUG)
+  sched.start()