refactor: use in operator instead of regex operator in crontab query to fix mssql regression

[alirafiei75]

Description

This pull request refactors _get_crontab_exclude_query method of DatabaseScheduler class to use a list of valid numeric hours for crontab query instead of regex pattern

Related Issues

Fixes: #897

[codecov]

Codecov Report

All modified and coverable lines are covered by tests ✅

Project coverage is 87.67%. Comparing base (673dbc5) to head (e05c85a).
Report is 1 commits behind head on main.

Additional details and impacted files

@@            Coverage Diff             @@
##             main     #900      +/-   ##
==========================================
- Coverage   88.24%   87.67%   -0.57%     
==========================================
  Files          32       32              
  Lines        1012     1014       +2     
  Branches      105       82      -23     
==========================================
- Hits          893      889       -4     
- Misses        101      107       +6     
  Partials       18       18              

☔ View full report in Codecov by Sentry.
📢 Have feedback on the report? Share it here.

🚀 New features to boost your workflow:

• ❄️ Test Analytics: Detect flaky tests, report on failures, and find test suite problems.

[Copilot]

Pull Request Overview

This PR refactors the _get_crontab_exclude_query method in DatabaseScheduler to replace a regex-based hour filter with an explicit __in lookup over a generated list of valid hour strings.

• Removed the numeric_hour_pattern regex filter.
• Introduced a valid_hours list combining non-padded and zero-padded representations for hours 0–23.
• Updated the queryset to use hour__in=valid_hours instead of hour__regex.

Comments suppressed due to low confidence (1)

django_celery_beat/schedulers.py:323

• Add unit tests to verify that the filter correctly matches both padded and non-padded hour values in the database, ensuring no valid entries are missed.

hour__in=valid_hours

[auvipy]

can we verify / fix this with modified tests please? the copilot suggestion seems to break the tests

[auvipy]

=================================== FAILURES ===================================
___________ test_DatabaseScheduler.test_crontab_timezone_conversion ____________

self = <t.unit.test_schedulers.test_DatabaseScheduler object at 0x7f2fbbbd74d0>
mock_get_tz =
mock_aware_now =

  @pytest.mark.django_db
  @patch('django_celery_beat.schedulers.aware_now')
  @patch('django.utils.timezone.get_current_timezone')
  def test_crontab_timezone_conversion(self, mock_get_tz, mock_aware_now):
      # Set up mocks for server timezone and current time
      from datetime import datetime
      server_tz = ZoneInfo("Asia/Tokyo")
  
      mock_get_tz.return_value = server_tz
  
      # Server time is 17:00 Tokyo time
      mock_now_dt = datetime(2023, 1, 1, 17, 0, 0, tzinfo=server_tz)
      mock_aware_now.return_value = mock_now_dt
  
      # Create tasks with the following crontab schedules:
      # 1. UTC task at hour 8 - equivalent to 17:00 Tokyo time (current hour)
      #    - should be included
      # 2. New York task at hour 3 - equivalent to 17:00 Tokyo time
      #    (current hour) - should be included
      # 3. UTC task at hour 3 - equivalent to 12:00 Tokyo time
      #    - should be excluded (outside window)
  
      # Create crontab schedules in different timezones
      utc_current_hour = CrontabSchedule.objects.create(
          hour='8', timezone='UTC'
      )
      ny_current_hour = CrontabSchedule.objects.create(
          hour='3', timezone='America/New_York'
      )
      utc_outside_window = CrontabSchedule.objects.create(
          hour='3', timezone='UTC'
      )
  
      # Create periodic tasks using these schedules
      task_utc_current = self.create_model(
          name='utc-current-hour',
          crontab=utc_current_hour
      )
      task_utc_current.save()
  
      task_ny_current = self.create_model(
          name='ny-current-hour',
          crontab=ny_current_hour
      )
      task_ny_current.save()
  
      task_utc_outside = self.create_model(
          name='utc-outside-window',
          crontab=utc_outside_window
      )
      task_utc_outside.save()
  
      # Run the scheduler's exclusion logic
      exclude_query = self.s._get_crontab_exclude_query()
  
      # Get excluded task IDs
      excluded_tasks = set(
          PeriodicTask.objects.filter(exclude_query).values_list(
              'id', flat=True
          )
      )
  
      # Current hour tasks in different timezones should not be excluded
      assert task_utc_current.id not in excluded_tasks, (
          "UTC current hour task should be included"
      )
      assert task_ny_current.id not in excluded_tasks, (
          "New York current hour task should be included"
      )
  
      # Task outside window should be excluded

  assert task_utc_outside.id in excluded_tasks, (

          "UTC outside window task should be excluded"
      )

E AssertionError: UTC outside window task should be excluded
E assert 15 in {11}
E + where 15 = <PeriodicTask: utc-outside-window: * 3 * * * (m/h/dM/MY/d) UTC>.id

t/unit/test_schedulers.py:1133: AssertionError

[alirafiei75]

yes I will. the copilot suggestion was wrong

[alirafiei75]

As we need both padded and non-padded hours ("1" and "01"), we cannot do it in just one list comprehension, or if we do it would not be readable i think

[auvipy]

can you apply the improvement and also add unit tests to verify this please?

[auvipy]

Yup, that looks better