search: don't search for engines on site indexation (#43228)

combo/apps/kb/models.py

@@ -33,6 +33,7 @@ class LatestPageUpdatesCell(CellBase):
         _('Maximum number of entries'), default=10)
 
     template_name = 'combo/latest-page-updates-cell.html'
+    exclude_from_search = True
 
     class Meta:
         verbose_name = _('Latest Page Updates')
@@ -48,6 +49,3 @@ class LatestPageUpdatesCell(CellBase):
                 (x for x in pages.order_by('-last_update_timestamp') if x.is_visible(user=user)),
                 self.limit)
         return extra_context
-
-    def render_for_search(self):
-        return ''

combo/apps/search/models.py

@@ -51,6 +51,7 @@ def get_root_page_and_children(service_slug):
 class SearchCell(CellBase):
     template_name = 'combo/search-cell.html'
     manager_form_template = 'combo/manager/search-cell-form.html'
+    exclude_from_search = True
 
     _search_services = JSONField(_('Search Services'), default=dict, blank=True)
     autofocus = models.BooleanField(_('Autofocus'), default=False)
@@ -60,7 +61,7 @@ class SearchCell(CellBase):
         verbose_name = _('Search')
 
     def is_visible(self, **kwargs):
-        if not self.search_services:
+        if not self._search_services.get('data'):
             return False
         return super(SearchCell, self).is_visible(**kwargs)
 

combo/apps/search/utils.py

@@ -54,6 +54,9 @@ def index_site():
             Page.objects
             .prefetch_related(Prefetch('groups', to_attr='prefetched_groups')))}
     for klass in cell_classes:
+        if getattr(klass, 'exclude_from_search', False) is True:
+            # do not load cells marked as excluded from search (example: MenuCell, SearchCell, ...)
+            continue
         queryset = (
             klass.objects
             .filter(page__snapshot__isnull=True, page__sub_slug='')

combo/data/models.py

@@ -1104,6 +1104,7 @@ class MenuCell(CellBase):
             null=True, blank=True, verbose_name=_('Root Page'))
 
     template_name = 'combo/menu-cell.html'
+    exclude_from_search = True
 
     class Meta:
         verbose_name = _('Menu')
@@ -1120,9 +1121,6 @@ class MenuCell(CellBase):
                 ignore_visibility=False)
         return ctx
 
-    def render_for_search(self):
-        return ''
-
 
 @register_cell_class
 class LinkCell(CellBase):
@@ -1252,6 +1250,7 @@ class LinkListCell(CellBase):
     template_name = 'combo/link-list-cell.html'
     manager_form_template = 'combo/manager/link-list-cell-form.html'
     children_placeholder_prefix = '_linkslist:'
+    exclude_from_search = True
 
     invalid_reason_codes = {
         'data_link_invalid': _('Invalid link'),
@@ -1304,9 +1303,6 @@ class LinkListCell(CellBase):
         from .forms import LinkListCellForm
         return LinkListCellForm
 
-    def render_for_search(self):
-        return ''
-
     def export_subobjects(self):
         return {'links': json.loads(
             serializers.serialize(

tests/test_search.py

@@ -297,17 +297,15 @@ def test_search_custom_templates(app):
             assert '<div>description A</div>' in resp.text
 
 
-def test_search_cell_visibility(app):
-    page = Page(title='example page', slug='example-page')
-    page.save()
+def test_search_cell_visibility(settings, app):
+    page = Page.objects.create(title='example page', slug='example-page')
+    settings.COMBO_SEARCH_SERVICES = SEARCH_SERVICES
 
-    with SearchServices(SEARCH_SERVICES):
-        cell = SearchCell(page=page, order=0)
-        assert not cell.is_visible()
+    cell = SearchCell(page=page, order=0)
+    assert not cell.is_visible()
 
-        cell._search_services = {'data': ['_text']}
-        del cell.search_services  # clear cache
-        assert cell.is_visible()
+    cell._search_services = {'data': ['_text']}
+    assert cell.is_visible()
 
 
 def test_search_contents():
@@ -328,7 +326,8 @@ def test_search_contents():
 
     # no indexation of menu cells
     cell = MenuCell(page=page, order=0)
-    assert cell.render_for_search() == ''
+    assert cell.exclude_from_search is True
+
 
 def test_search_contents_index():
     page = Page(title='example page', slug='example-page')
@@ -596,6 +595,7 @@ def test_manager_search_cell_order(settings, app, admin_user):
 def test_manager_waiting_index_message(app, admin_user):
     page = Page.objects.create(title='One', slug='one', template_name='standard')
     cell = SearchCell.objects.create(page=page, placeholder='content', order=0)
+    TextCell.objects.create(page=page, placeholder='content', text='<p>foobar</p>', order=0)
     app = login(app)
     resp = app.get('/manage/pages/%s/' % page.pk)
     assert 'Content indexing has been scheduled' not in resp.text
@@ -775,7 +775,7 @@ def test_index_site_inactive_placeholder(app):
     assert IndexedCell.objects.count() == 0
 
 
-def test_index_site_num_queries(app):
+def test_index_site_num_queries(settings, app):
     group = Group.objects.create(name='plop')
     for i in range(0, 10):
         page = Page.objects.create(title='page %s' % i, slug='example-page-%s' % i)
@@ -788,4 +788,28 @@ def test_index_site_num_queries(app):
     assert IndexedCell.objects.count() == 50
     with CaptureQueriesContext(connection) as ctx:
         index_site()
-        assert len(ctx.captured_queries) == 225
+        assert len(ctx.captured_queries) == 221
+
+    SearchCell.objects.create(
+        page=page, placeholder='content', order=0,
+        _search_services={'data': ['search1']})
+    # search cells are not indexed
+    index_site()
+    assert IndexedCell.objects.count() == 50
+
+
+@mock.patch('combo.apps.search.engines.get_engines')
+def test_index_site_search_engines_load(get_engines_mock, settings, app):
+    # be sure that get_engines is not called during page indexation
+    get_engines_mock.side_effect = Exception
+
+    settings.COMBO_SEARCH_SERVICES = {}
+
+    page = Page.objects.create(title='page', slug='example-page')
+    SearchCell.objects.create(
+        page=page, placeholder='content', order=0,
+        _search_services={'data': ['search1']})
+    TextCell.objects.create(page=page, placeholder='content', text='<p>foobar</p>', order=0)
+
+    # no exception raised
+    index_site()