[ENG-4438] Add OOPSpam and Akismet metrics to spam report

osf/external/askismet/client.py

@@ -133,3 +133,27 @@ def submit_ham(self, user_ip, user_agent, **kwargs):
         )
         if res.status_code != requests.codes.ok:
             raise AkismetClientError(reason=res.text)
+
+    def get_flagged_count(self, start_date, end_date):
+        from osf.models import NodeLog
+
+        flagged_count = NodeLog.objects.filter(
+            action=NodeLog.FLAG_SPAM,
+            created__gt=start_date,
+            created__lt=end_date,
+            node__spam_data__who_flagged__in=['akismet', 'both']
+        ).count()
+
+        return flagged_count
+
+    def get_hammed_count(self, start_date, end_date):
+        from osf.models import NodeLog
+
+        hammed_count = NodeLog.objects.filter(
+            action=NodeLog.CONFIRM_HAM,
+            created__gt=start_date,
+            created__lt=end_date,
+            node__spam_data__who_flagged__in=['akismet', 'both']
+        ).count()
+
+        return hammed_count

osf/external/oopspam/client.py

@@ -45,3 +45,27 @@ def check_content(self, user_ip, content, **kwargs):
 
         #  OOPSpam returns a spam score out of 6. 3 or higher indicates spam
         return spam_score >= settings.OOPSPAM_SPAM_LEVEL, resp_json
+
+    def get_flagged_count(self, start_date, end_date):
+        from osf.models import NodeLog
+
+        flagged_count = NodeLog.objects.filter(
+            action=NodeLog.FLAG_SPAM,
+            created__gt=start_date,
+            created__lt=end_date,
+            node__spam_data__who_flagged__in=['oopspam', 'both']
+        ).count()
+
+        return flagged_count
+
+    def get_hammed_count(self, start_date, end_date):
+        from osf.models import NodeLog
+
+        hammed_count = NodeLog.objects.filter(
+            action=NodeLog.CONFIRM_HAM,
+            created__gt=start_date,
+            created__lt=end_date,
+            node__spam_data__who_flagged__in=['oopspam', 'both']
+        ).count()
+
+        return hammed_count

osf/metrics/reporters/spam_count.py

@@ -4,6 +4,8 @@
 from ._base import MonthlyReporter
 from osf.models import PreprintLog, NodeLog
 from osf.models.spam import SpamStatus
+from osf.external.oopspam.client import OOPSpamClient
+from osf.external.askismet.client import AkismetClient
 
 
 class SpamCountReporter(MonthlyReporter):
@@ -12,6 +14,15 @@ def report(self, report_yearmonth):
         target_month = report_yearmonth.target_month()
         next_month = report_yearmonth.next_month()
 
+        oopspam_client = OOPSpamClient()
+        akismet_client = AkismetClient()
+
+        oopspam_flagged = oopspam_client.get_flagged_count(target_month, next_month)
+        oopspam_hammed = oopspam_client.get_hammed_count(target_month, next_month)
+
+        akismet_flagged = akismet_client.get_flagged_count(target_month, next_month)
+        akismet_hammed = akismet_client.get_hammed_count(target_month, next_month)
+
         report = SpamSummaryReport(
             report_yearmonth=str(report_yearmonth),
             # Node Log entries
@@ -33,6 +44,10 @@ def report(self, report_yearmonth):
                 created__lt=next_month,
                 node__type='osf.node',
             ).count(),
+            oopspam_flagged=oopspam_flagged,
+            oopspam_hammed=oopspam_hammed,
+            akismet_flagged=akismet_flagged,
+            akismet_hammed=akismet_hammed,
             # Registration Log entries
             registration_confirmed_spam=NodeLog.objects.filter(
                 action=NodeLog.CONFIRM_SPAM,

osf/metrics/reports.py

@@ -214,3 +214,7 @@ class SpamSummaryReport(MonthlyReport):
     preprint_flagged = metrics.Integer()
     user_marked_as_spam = metrics.Integer()
     user_marked_as_ham = metrics.Integer()
+    oopspam_flagged = metrics.Integer()
+    oopspam_hammed = metrics.Integer()
+    akismet_flagged = metrics.Integer()
+    akismet_hammed = metrics.Integer()

osf_tests/external/akismet/test_akismet.py

@@ -237,3 +237,39 @@ def test_meetings_skip_spam_check(self, mock_akismet, user, node_in_conference,
         node.check_spam(user, {'title'}, request_headers)
         node.refresh_from_db()
         assert node.spam_status == SpamStatus.FLAGGED
+
+    @mock.patch('osf.models.NodeLog.objects.filter')
+    def test_get_flagged_count(self, mock_filter, user):
+        from osf.external.askismet.client import AkismetClient
+        from datetime import datetime
+
+        client = AkismetClient()
+        start_date = datetime(2024, 10, 1)
+        end_date = datetime(2024, 10, 31)
+
+        client.get_flagged_count(start_date, end_date)
+
+        mock_filter.assert_called_with(
+            action='flag_spam',
+            created__gt=start_date,
+            created__lt=end_date,
+            node__spam_data__who_flagged__in=['akismet', 'both']
+        )
+
+    @mock.patch('osf.models.NodeLog.objects.filter')
+    def test_get_hammed_count(self, mock_filter, user):
+        from osf.external.askismet.client import AkismetClient
+        from datetime import datetime
+
+        client = AkismetClient()
+        start_date = datetime(2024, 10, 1)
+        end_date = datetime(2024, 10, 31)
+
+        client.get_hammed_count(start_date, end_date)
+
+        mock_filter.assert_called_with(
+            action='confirm_ham',
+            created__gt=start_date,
+            created__lt=end_date,
+            node__spam_data__who_flagged__in=['akismet', 'both']
+        )

osf_tests/external/oopspam/test_oopspam.py

@@ -125,3 +125,39 @@ def test_do_spam_check_false(self, mock_oopspam, user, request_headers):
         )
 
         assert user.spam_status == SpamStatus.UNKNOWN
+
+    @mock.patch('osf.models.NodeLog.objects.filter')
+    def test_get_flagged_count(self, mock_filter, user):
+        from osf.external.oopspam.client import OOPSpamClient
+        from datetime import datetime
+
+        client = OOPSpamClient()
+        start_date = datetime(2024, 10, 1)
+        end_date = datetime(2024, 10, 31)
+
+        client.get_flagged_count(start_date, end_date)
+
+        mock_filter.assert_called_with(
+            action='flag_spam',
+            created__gt=start_date,
+            created__lt=end_date,
+            node__spam_data__who_flagged__in=['oopspam', 'both']
+        )
+
+    @mock.patch('osf.models.NodeLog.objects.filter')
+    def test_get_hammed_count(self, mock_filter, user):
+        from osf.external.oopspam.client import OOPSpamClient
+        from datetime import datetime
+
+        client = OOPSpamClient()
+        start_date = datetime(2024, 10, 1)
+        end_date = datetime(2024, 10, 31)
+
+        client.get_hammed_count(start_date, end_date)
+
+        mock_filter.assert_called_with(
+            action='confirm_ham',
+            created__gt=start_date,
+            created__lt=end_date,
+            node__spam_data__who_flagged__in=['oopspam', 'both']
+        )

osf_tests/metrics/test_spam_count_reporter.py

@@ -0,0 +1,43 @@
+import pytest
+from datetime import datetime
+from osf.metrics.reporters.spam_count import SpamCountReporter
+from unittest import mock
+from osf.metrics.utils import YearMonth
+from osf_tests.factories import NodeLogFactory, NodeFactory
+
+@pytest.fixture
+def mock_oopspam_client():
+    with mock.patch('osf.external.oopspam.client.OOPSpamClient') as mock_client:
+        instance = mock_client.return_value
+        instance.get_flagged_count.return_value = 10
+        instance.get_hammed_count.return_value = 5
+        yield instance
+
+@pytest.fixture
+def mock_akismet_client():
+    with mock.patch('osf.external.askismet.client.AkismetClient') as mock_client:
+        instance = mock_client.return_value
+        instance.get_flagged_count.return_value = 20
+        instance.get_hammed_count.return_value = 10
+        yield instance
+
+@pytest.mark.django_db
+def test_spam_count_reporter():
+    start_date = datetime(2024, 10, 1)
+
+    oopspam_node = NodeFactory(spam_data={'who_flagged': 'oopspam'})
+    akismet_node = NodeFactory(spam_data={'who_flagged': 'akismet'})
+
+    NodeLogFactory.create_batch(10, action='flag_spam', created=start_date, node=oopspam_node)
+    NodeLogFactory.create_batch(5, action='confirm_ham', created=start_date, node=oopspam_node)
+    NodeLogFactory.create_batch(20, action='flag_spam', created=start_date, node=akismet_node)
+    NodeLogFactory.create_batch(10, action='confirm_ham', created=start_date, node=akismet_node)
+
+    report_yearmonth = YearMonth(2024, 10)
+    reporter = SpamCountReporter()
+    report = reporter.report(report_yearmonth)
+
+    assert report[0].oopspam_flagged == 10
+    assert report[0].oopspam_hammed == 5
+    assert report[0].akismet_flagged == 20
+    assert report[0].akismet_hammed == 10