Analytics: add feature flag to skip tracking 404s (#9131)

* Analytics: add feature flag to skip tracking 404s

* Invert feature flag

* Fix tests

Author: stsewd

readthedocs/projects/models.py

@@ -1759,6 +1759,7 @@ def add_features(sender, **kwargs):
     LIMIT_CONCURRENT_BUILDS = "limit_concurrent_builds"
     CDN_ENABLED = "cdn_enabled"
     DOCKER_GVISOR_RUNTIME = "gvisor_runtime"
+    RECORD_404_PAGE_VIEWS = "record_404_page_views"
 
     # Versions sync related features
     SKIP_SYNC_TAGS = 'skip_sync_tags'
@@ -1850,6 +1851,10 @@ def add_features(sender, **kwargs):
             DOCKER_GVISOR_RUNTIME,
             _("Use Docker gVisor runtime to create build container."),
         ),
+        (
+            RECORD_404_PAGE_VIEWS,
+            _("Record 404s page views."),
+        ),
 
         # Versions sync related features
         (

readthedocs/projects/views/private.py

@@ -65,6 +65,7 @@
     Domain,
     EmailHook,
     EnvironmentVariable,
+    Feature,
     Project,
     ProjectRelationship,
     WebHook,
@@ -1180,12 +1181,15 @@ def get_context_data(self, **kwargs):
 
         # Count of views for top pages over the month
         top_pages_200 = PageView.top_viewed_pages(project, limit=25)
-        top_pages_404 = PageView.top_viewed_pages(
-            project,
-            limit=25,
-            status=404,
-            per_version=True,
-        )
+        track_404 = project.has_feature(Feature.RECORD_404_PAGE_VIEWS)
+        top_pages_404 = []
+        if track_404:
+            top_pages_404 = PageView.top_viewed_pages(
+                project,
+                limit=25,
+                status=404,
+                per_version=True,
+            )
 
         # Aggregate pageviews grouped by day
         page_data = PageView.page_views_by_date(
@@ -1197,6 +1201,7 @@ def get_context_data(self, **kwargs):
                 "top_pages_200": top_pages_200,
                 "page_data": page_data,
                 "top_pages_404": top_pages_404,
+                "track_404": track_404,
             }
         )
 

readthedocs/proxito/tests/test_full.py

@@ -878,6 +878,11 @@ def test_404_all_paths_checked_default_version_different_doc_type(self, storage_
 
     @mock.patch.object(BuildMediaFileSystemStorageTest, "exists")
     def test_track_broken_link(self, storage_exists):
+        get(
+            Feature,
+            feature_id=Feature.RECORD_404_PAGE_VIEWS,
+            projects=[self.project],
+        )
         self.assertEqual(PageView.objects.all().count(), 0)
 
         paths = [
@@ -926,6 +931,11 @@ def test_track_broken_link(self, storage_exists):
     @mock.patch.object(BuildMediaFileSystemStorageTest, "open")
     @mock.patch.object(BuildMediaFileSystemStorageTest, "exists")
     def test_track_broken_link_custom_404(self, storage_exists, storage_open):
+        get(
+            Feature,
+            feature_id=Feature.RECORD_404_PAGE_VIEWS,
+            projects=[self.project],
+        )
         self.assertEqual(PageView.objects.all().count(), 0)
 
         paths = [

readthedocs/proxito/views/serve.py

@@ -19,6 +19,7 @@
 from readthedocs.core.utils.extend import SettingsOverrideObject
 from readthedocs.projects import constants
 from readthedocs.projects.constants import SPHINX_HTMLDIR
+from readthedocs.projects.models import Feature
 from readthedocs.projects.templatetags.projects_tags import sort_version_aware
 from readthedocs.redirects.exceptions import InfiniteRedirectException
 from readthedocs.storage import build_media_storage
@@ -402,6 +403,9 @@ def get(self, request, proxito_path, template_name='404.html'):
 
     def _register_broken_link(self, project, version, path, full_path):
         try:
+            if not project.has_feature(Feature.RECORD_404_PAGE_VIEWS):
+                return
+
             # If the path isn't attached to a version
             # it should be the same as the full_path,
             # otherwise it would be empty.

readthedocs/templates/projects/project_traffic_analytics.html

@@ -48,6 +48,7 @@ <h3>{% trans "Daily pageview totals" %}</h3>
     <button type="submit" name="download" value="true">{% trans "Download all data" %}</button>
   </form>
 
+  {% if track_404 %}
   <h3>{% trans "Not Found (404) pages" %}</h3>
 
   <div class="module-list">
@@ -76,7 +77,7 @@ <h3>{% trans "Not Found (404) pages" %}</h3>
     It's possible the number of times each page is viewed is undercounted because of caching in the browser or in a CDN.
     {% endblocktrans %}
   </small>
-
+  {% endif %}
 
 {% endblock %}