Add private spam metrics report with preprint inclusion

osf/external/askismet/client.py

@@ -134,26 +134,30 @@ 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
+    def get_flagged_count(self, start_date, end_date, category='node'):
+        from osf.models import NodeLog, PreprintLog
 
-        flagged_count = NodeLog.objects.filter(
-            action=NodeLog.FLAG_SPAM,
+        log_model = NodeLog if category == 'node' else PreprintLog
+
+        flagged_count = log_model.objects.filter(
+            action=log_model.FLAG_SPAM,
             created__gt=start_date,
             created__lt=end_date,
-            node__spam_data__who_flagged__in=['akismet', 'both']
+            **{f'{category}__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
+    def get_hammed_count(self, start_date, end_date, category='node'):
+        from osf.models import NodeLog, PreprintLog
+
+        log_model = NodeLog if category == 'node' else PreprintLog
 
-        hammed_count = NodeLog.objects.filter(
-            action=NodeLog.CONFIRM_HAM,
+        hammed_count = log_model.objects.filter(
+            action=log_model.CONFIRM_HAM,
             created__gt=start_date,
             created__lt=end_date,
-            node__spam_data__who_flagged__in=['akismet', 'both']
+            **{f'{category}__spam_data__who_flagged__in': ['akismet', 'both']}
         ).count()
 
         return hammed_count

osf/external/oopspam/client.py

@@ -46,26 +46,30 @@ 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
+    def get_flagged_count(self, start_date, end_date, category='node'):
+        from osf.models import NodeLog, PreprintLog
 
-        flagged_count = NodeLog.objects.filter(
-            action=NodeLog.FLAG_SPAM,
+        log_model = NodeLog if category == 'node' else PreprintLog
+
+        flagged_count = log_model.objects.filter(
+            action=log_model.FLAG_SPAM,
             created__gt=start_date,
             created__lt=end_date,
-            node__spam_data__who_flagged__in=['oopspam', 'both']
+            **{f'{category}__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
+    def get_hammed_count(self, start_date, end_date, category='node'):
+        from osf.models import NodeLog, PreprintLog
+
+        log_model = NodeLog if category == 'node' else PreprintLog
 
-        hammed_count = NodeLog.objects.filter(
-            action=NodeLog.CONFIRM_HAM,
+        hammed_count = log_model.objects.filter(
+            action=log_model.CONFIRM_HAM,
             created__gt=start_date,
             created__lt=end_date,
-            node__spam_data__who_flagged__in=['oopspam', 'both']
+            **{f'{category}__spam_data__who_flagged__in': ['oopspam', 'both']}
         ).count()
 
         return hammed_count

osf/metrics/reporters/private_spam_metrics.py

@@ -0,0 +1,35 @@
+from osf.metrics.reports import MonthlyReport
+from osf.external.oopspam.client import OOPSpamClient
+from osf.external.askismet.client import AkismetClient
+
+class PrivateSpamMetricsReporter(MonthlyReport):
+    report_name = 'Private Spam Metrics'
+
+    def report(self, report_yearmonth):
+        target_month = report_yearmonth.target_month()
+        next_month = report_yearmonth.next_month()
+
+        oopspam_client = OOPSpamClient()
+        akismet_client = AkismetClient()
+
+        node_oopspam_flagged = oopspam_client.get_flagged_count(target_month, next_month, category='node')
+        node_oopspam_hammed = oopspam_client.get_hammed_count(target_month, next_month, category='node')
+        node_akismet_flagged = akismet_client.get_flagged_count(target_month, next_month, category='node')
+        node_akismet_hammed = akismet_client.get_hammed_count(target_month, next_month, category='node')
+
+        preprint_oopspam_flagged = oopspam_client.get_flagged_count(target_month, next_month, category='preprint')
+        preprint_oopspam_hammed = oopspam_client.get_hammed_count(target_month, next_month, category='preprint')
+        preprint_akismet_flagged = akismet_client.get_flagged_count(target_month, next_month, category='preprint')
+        preprint_akismet_hammed = akismet_client.get_hammed_count(target_month, next_month, category='preprint')
+
+        return [self.create_report(
+            report_yearmonth=str(report_yearmonth),
+            node_oopspam_flagged=node_oopspam_flagged,
+            node_oopspam_hammed=node_oopspam_hammed,
+            node_akismet_flagged=node_akismet_flagged,
+            node_akismet_hammed=node_akismet_hammed,
+            preprint_oopspam_flagged=preprint_oopspam_flagged,
+            preprint_oopspam_hammed=preprint_oopspam_hammed,
+            preprint_akismet_flagged=preprint_akismet_flagged,
+            preprint_akismet_hammed=preprint_akismet_hammed,
+        )]

osf/metrics/reporters/spam_count.py

@@ -4,25 +4,13 @@
 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):
 
     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
@@ -44,10 +32,6 @@ 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

@@ -214,7 +214,3 @@ 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/metrics/test_spam_count_reporter.py

@@ -1,28 +1,11 @@
 import pytest
 from datetime import datetime
-from osf.metrics.reporters.spam_count import SpamCountReporter
-from unittest import mock
+from osf.metrics.reporters.private_spam_metrics import PrivateSpamMetricsReporter
 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():
+def test_private_spam_metrics_reporter():
     start_date = datetime(2024, 10, 1)
 
     oopspam_node = NodeFactory(spam_data={'who_flagged': 'oopspam'})
@@ -34,10 +17,10 @@ def test_spam_count_reporter():
     NodeLogFactory.create_batch(10, action='confirm_ham', created=start_date, node=akismet_node)
 
     report_yearmonth = YearMonth(2024, 10)
-    reporter = SpamCountReporter()
+    reporter = PrivateSpamMetricsReporter()
     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
+    assert report[0]['node_oopspam_flagged'] == 10
+    assert report[0]['node_oopspam_hammed'] == 5
+    assert report[0]['node_akismet_flagged'] == 20
+    assert report[0]['node_akismet_hammed'] == 10