Use last_published_at for lastmod in sitemaps - fixes #3565

wagtail/contrib/wagtailsitemaps/sitemap_generator.py

@@ -12,7 +12,11 @@ class Sitemap(DjangoSitemap):
         return obj.specific.url
 
     def lastmod(self, obj):
-        return obj.specific.latest_revision_created_at
+        obj = obj.specific
+
+        # fall back on latest_revision_created_at if last_published_at is null
+        # (for backwards compatibility from before last_published_at was added)
+        return (obj.last_published_at or obj.latest_revision_created_at)
 
     def items(self):
         return (

wagtail/contrib/wagtailsitemaps/tests.py

@@ -1,5 +1,8 @@
 from __future__ import absolute_import, unicode_literals
 
+import datetime
+
+import pytz
 from django.contrib.sites.shortcuts import get_current_site
 from django.test import RequestFactory, TestCase
 
@@ -18,6 +21,8 @@ class TestSitemapGenerator(TestCase):
             slug='hello-world',
             content="hello",
             live=True,
+            last_published_at=datetime.datetime(2017, 1, 1, 12, 0, 0, tzinfo=pytz.utc),
+            latest_revision_created_at=datetime.datetime(2017, 2, 1, 12, 0, 0, tzinfo=pytz.utc)
         ))
 
         self.unpublished_child_page = self.home_page.add_child(instance=SimplePage(
@@ -35,6 +40,14 @@ class TestSitemapGenerator(TestCase):
         ))
         PageViewRestriction.objects.create(page=self.protected_child_page, password='hello')
 
+        self.page_with_no_last_publish_date = self.home_page.add_child(instance=SimplePage(
+            title="I have no last publish date :-(",
+            slug='no-last-publish-date',
+            content="hello",
+            live=True,
+            latest_revision_created_at=datetime.datetime(2017, 2, 1, 12, 0, 0, tzinfo=pytz.utc)
+        ))
+
         self.site = Site.objects.get(is_default_site=True)
 
     def test_items(self):
@@ -74,6 +87,27 @@ class TestSitemapGenerator(TestCase):
         self.assertIn('http://localhost/events/', urls)  # Main view
         self.assertIn('http://localhost/events/past/', urls)  # Sub view
 
+    def test_lastmod_uses_last_published_date(self):
+        request = RequestFactory().get('/sitemap.xml')
+        req_protocol = request.scheme
+        req_site = get_current_site(request)
+
+        sitemap = Sitemap(self.site)
+        urls = sitemap.get_urls(1, req_site, req_protocol)
+
+        child_page_lastmod = [
+            url['lastmod'] for url in urls
+            if url['location'] == 'http://localhost/hello-world/'
+        ][0]
+        self.assertEqual(child_page_lastmod, datetime.datetime(2017, 1, 1, 12, 0, 0, tzinfo=pytz.utc))
+
+        # if no last_publish_date is defined, use latest revision date
+        child_page_lastmod = [
+            url['lastmod'] for url in urls
+            if url['location'] == 'http://localhost/no-last-publish-date/'
+        ][0]
+        self.assertEqual(child_page_lastmod, datetime.datetime(2017, 2, 1, 12, 0, 0, tzinfo=pytz.utc))
+
 
 class TestIndexView(TestCase):
     def test_index_view(self):

wagtail/wagtailcore/models.py

@@ -1350,7 +1350,9 @@ class Page(six.with_metaclass(PageBase, AbstractPage, index.Indexed, Clusterable
         return [
             {
                 'location': self.full_url,
-                'lastmod': self.latest_revision_created_at
+                # fall back on latest_revision_created_at if last_published_at is null
+                # (for backwards compatibility from before last_published_at was added)
+                'lastmod': (self.last_published_at or self.latest_revision_created_at),
             }
         ]