Fix code that doesn't work on Ginkgo (Django 1.8)

devsite/requirements/ginkgo.txt

@@ -35,7 +35,7 @@ django-webpack-loader==0.4.1
 # appsembler/gingko/master users 0.4.1
 
 django-model-utils==2.3.1
-
+django-celery==3.2.1
 jsonfield==1.0.3  # Version used in Ginkgo. Hawthorn uses version 2.0.2
 
 ##

figures/mau.py

@@ -2,7 +2,10 @@
 This module provides MAU metrics retrieval functionality
 """
 
-from datetime import datetime
+from datetime import datetime, timedelta
+from django.utils.timezone import utc
+
+from figures.compat import RELEASE_LINE
 
 from figures.models import CourseMauMetrics, SiteMauMetrics
 from figures.sites import (
@@ -84,11 +87,38 @@ def mau_1g_for_month_as_of_day(sm_queryset, date_for):
     This function queries `courseware.models.StudentModule` to identify users
     who are active in the site
 
+    Retrieves records based on date of the `StudentModule.modified` field
     Returns a queryset of distinct user ids
     """
-    month_sm = sm_queryset.filter(modified__year=date_for.year,
-                                  modified__month=date_for.month,
-                                  modified__day__lte=date_for.day)
+
+    # TODO: Remove this `if` branch after dropping Ginkgo support.
+    if RELEASE_LINE == 'ginkgo':
+        # Django 1.8 appears not to support 'lte' on the 'day' of a datetime
+        # Therefore we have to get records within a range
+        start_date = datetime(year=date_for.year,
+                              month=date_for.month,
+                              day=1).replace(tzinfo=utc)
+        # We do this in case 'date_for' is at the end of the month and get
+        # the 'day_after' as midnight of the next day so we can use '__lt'. If
+        # we simply used '__lte', then we exclude any events that happened on
+        # the 'date_for' in hours after the 'date_for' hours.
+
+        # temporary var as we don't know
+        day_after_temp = date_for + timedelta(days=1)
+        day_after = datetime(year=day_after_temp.year,
+                             month=day_after_temp.month,
+                             day=day_after_temp.day).replace(tzinfo=utc)
+
+        # We don't use 'dict(modified__range=[start_date, date_for])' because
+        # doing "__lt" for 0:00 hour tne next day means we don't have to worry
+        # about fractions of a second on the last second of the last day
+        filter_args = dict(modified__gte=start_date, modified__lt=day_after)
+    else:
+        filter_args = dict(modified__year=date_for.year,
+                           modified__month=date_for.month,
+                           modified__day__lte=date_for.day)
+
+    month_sm = sm_queryset.filter(**filter_args)
     return month_sm.values('student__id').distinct()
 
 

tests/test_mau.py

@@ -1,13 +1,11 @@
 
 from datetime import datetime
 from freezegun import freeze_time
-from dateutil.relativedelta import relativedelta
 import pytest
 
-from django.utils.timezone import utc
-
 from courseware.models import StudentModule
 
+from figures.helpers import as_datetime, as_date
 from figures.sites import (
     get_student_modules_for_site,
     get_student_modules_for_course_in_site,
@@ -21,7 +19,6 @@
 )
 
 from tests.factories import StudentModuleFactory
-from tests.helpers import OPENEDX_RELEASE, GINKGO
 
 
 def test_get_mau_from_site_course(sm_test_data):
@@ -56,8 +53,6 @@ def test_get_mau_from_sm_for_site(sm_test_data):
     assert set(users) == set(sm_check)
 
 
-@pytest.mark.skipif(OPENEDX_RELEASE == GINKGO,
-                    reason='Broken test. Apparent Django 1.8 incompatibility')
 @pytest.mark.django_db
 def test_mau_1g_for_month_as_of_day_first_day_next_month(db):
     """
@@ -68,31 +63,35 @@ def test_mau_1g_for_month_as_of_day_first_day_next_month(db):
     We want to make sure we get the right records when the query happens on the
     first day of the next month. So we do the following
 
-    * Add a StudentModule record for two months before
-    * Add at least one StudentModule record for the month we want
-    * Add at least one StudentModule record for after the month we want
+    * Add StudentModule records for the month before we want to capture records
+    * Add StudentModule records for the month we want to capture records
+    * Add StudentModule records for the month after we want to capture records
 
     This sets up the scenario that we run the daily pipeline to capture MAU
     "as of" yesterday (the last day of the previous month) to capture MAU for
-    the previous month
+    the previous month and not capture any records before the previous month,
+    nor capture records for the "current month"
     """
-    mock_today = datetime(year=2020, month=4, day=1).replace(tzinfo=utc)
-    month_before = datetime(year=2020, month=2, day=2).replace(tzinfo=utc)
-    in_dates = [datetime(year=2020, month=3, day=1).replace(tzinfo=utc),
-                datetime(year=2020, month=3, day=15).replace(tzinfo=utc),
-                datetime(year=2020, month=3, day=31).replace(tzinfo=utc)]
-    date_for = mock_today.date() - relativedelta(days=1)
-
-    # Create a student module in the month before, and in month after
-    StudentModuleFactory(created=month_before, modified=month_before)
-    StudentModuleFactory(created=mock_today, modified=mock_today)
+    month_before = [as_datetime('2020-02-02'), as_datetime('2020-02-29')]
+    month_after = [as_datetime('2020-04-01'), as_datetime('2020-04-01 12:00')]
+    in_month = [as_datetime('2020-03-01'),
+                as_datetime('2020-03-15'),
+                as_datetime('2020-03-31'),
+                as_datetime('2020-03-31 12:00')]
+    date_for = as_date('2020-03-31')
+
+    # Create student modules for the month before, month after, and in the
+    # month for which we want to retrieve records
+    [StudentModuleFactory(created=dt, modified=dt) for dt in month_before]
+    [StudentModuleFactory(created=dt, modified=dt) for dt in month_after]
     sm_in = [StudentModuleFactory(created=rec,
-                                  modified=rec) for rec in in_dates]
+                                  modified=rec) for rec in in_month]
     expected_user_ids = [obj.student_id for obj in sm_in]
 
     sm_queryset = StudentModule.objects.all()
     user_ids = mau_1g_for_month_as_of_day(sm_queryset=sm_queryset,
                                           date_for=date_for)
+    assert len(user_ids) == len(in_month)
     assert set([rec['student__id'] for rec in user_ids]) == set(expected_user_ids)