Add purge_revisions management command

* Fix typo in publish_scheduled_pages dry run
* Add purge_revisions to docs
* Resolves #5192

CHANGELOG.txt

@@ -22,6 +22,7 @@ Changelog
  * Add ability to replace the default Wagtail logo in the userbar, via `branding_logo` block (Meteor0id)
  * Remove sticky footer on small devices, so that content is not blocked and more easily editable (Saeed Tahmasebi)
  * Add ``alt`` property to ``ImageRenditionField`` api representation (Liam Mullens)
+ * Add ``purge_revisions`` management command to purge old page revisions (Jacob Topp-Mugglestone, Tom Dyson)
  * Fix: Support IPv6 domain (Alex Gleason, Coen van der Kamp)
  * Fix: Ensure link to add a new user works when no users are visible in the users list (LB (Ben Johnston))
  * Fix: `AbstractEmailForm` saved submission fields are now aligned with the email content fields, `form.cleaned_data` will be used instead of `form.fields` (Haydn Greatnews)

docs/reference/management_commands.rst

@@ -48,6 +48,20 @@ Options:
    This is the **id** of the page to move pages to.
 
 
+.. _purge_revisions:
+
+purge_revisions
+---------------
+
+.. code-block:: console
+
+    $ manage.py purge_revisions [--days=<number of days>]
+
+This command deletes old page revisions which are not in moderation, live, approved to go live, or the latest
+revision for a page. If the ``days`` argument is supplied, only revisions older than the specified number of
+days will be deleted.
+
+
 .. _update_index:
 
 update_index

docs/releases/2.10.rst

@@ -31,6 +31,7 @@ Other features
  * Remove sticky footer on small devices, so that content is not blocked and more easily editable (Saeed Tahmasebi)
  * Add ability to replace the default Wagtail logo in the userbar, via ``branding_logo`` block (Meteor0id)
  * Add `alt` property to `ImageRenditionField` api representation (Liam Mullens)
+ * Add `purge_revisions` management command to purge old page revisions (Jacob Topp-Mugglestone, Tom Dyson)
 
 
 Bug fixes

wagtail/core/management/commands/publish_scheduled_pages.py

@@ -21,7 +21,7 @@ class Command(BaseCommand):
     def add_arguments(self, parser):
         parser.add_argument(
             '--dryrun', action='store_true', dest='dryrun', default=False,
-            help="Dry run -- dont't change anything.")
+            help="Dry run -- don't change anything.")
 
     def handle(self, *args, **options):
         dryrun = False

wagtail/core/management/commands/purge_revisions.py

@@ -0,0 +1,57 @@
+from django.core.management.base import BaseCommand
+from django.utils import timezone
+
+from wagtail.core.models import PageRevision
+
+try:
+    from wagtail.core.models import WorkflowState
+    workflow_support = True
+except ImportError:
+    workflow_support = False
+
+
+class Command(BaseCommand):
+    help = 'Delete page revisions which are not the latest revision for a page, published or scheduled to be published, or in moderation'
+
+    def add_arguments(self, parser):
+        parser.add_argument('--days', type=int, help="Only delete revisions older than this number of days")
+
+    def handle(self, *args, **options):
+        days = options.get('days')
+
+        revisions_deleted = purge_revisions(days=days)
+
+        if revisions_deleted:
+            self.stdout.write(self.style.SUCCESS('Successfully deleted %s revisions' % revisions_deleted))
+        else:
+            self.stdout.write("No revisions deleted")
+
+
+def purge_revisions(days=None):
+    # exclude revisions which have been submitted for moderation in the old system
+    purgeable_revisions = PageRevision.objects.exclude(
+        submitted_for_moderation=True
+    ).exclude(
+        # and exclude revisions with an approved_go_live_at date
+        approved_go_live_at__isnull=False)
+
+    if workflow_support:
+        purgeable_revisions = purgeable_revisions.exclude(
+            # and exclude revisions linked to an in progress workflow state
+            task_states__workflow_state__status=WorkflowState.STATUS_IN_PROGRESS
+        )
+
+    if days:
+        purgeable_until = timezone.now() - timezone.timedelta(days=days)
+        # only include revisions which were created before the cut off date
+        purgeable_revisions = purgeable_revisions.filter(created_at__lt=purgeable_until)
+
+    deleted_revisions_count = 0
+
+    for revision in purgeable_revisions:
+        # don't delete the latest revision for any page
+        if not revision.is_latest_revision():
+            revision.delete()
+            deleted_revisions_count += 1
+
+    return deleted_revisions_count

wagtail/core/tests/test_management_commands.py

@@ -1,6 +1,7 @@
 from datetime import timedelta
 from io import StringIO
 
+from django.contrib.auth import get_user_model
 from django.core import management
 from django.db import models
 from django.test import TestCase
@@ -332,3 +333,92 @@ class TestPublishScheduledPagesCommand(TestCase):
 
         p = Page.objects.get(slug='hello-world')
         self.assertFalse(PageRevision.objects.filter(page=p, submitted_for_moderation=True).exists())
+
+
+class TestPurgeRevisionsCommand(TestCase):
+    fixtures = ['test.json']
+
+    def setUp(self):
+        # Find root page
+        self.root_page = Page.objects.get(id=2)
+        self.page = SimplePage(
+            title="Hello world!",
+            slug="hello-world",
+            content="hello",
+            live=False,
+        )
+        self.root_page.add_child(instance=self.page)
+        self.page.refresh_from_db()
+
+    def run_command(self, days=None):
+        if days:
+            days_input = '--days=' + str(days)
+            return management.call_command('purge_revisions', days_input, stdout=StringIO())
+        return management.call_command('purge_revisions', stdout=StringIO())
+
+    def test_latest_revision_not_purged(self):
+
+        revision_1 = self.page.save_revision()
+
+        revision_2 = self.page.save_revision()
+
+        self.run_command()
+
+        # revision 1 should be deleted, revision 2 should not be
+        self.assertNotIn(revision_1, PageRevision.objects.filter(page=self.page))
+        self.assertIn(revision_2, PageRevision.objects.filter(page=self.page))
+
+
+    def test_revisions_in_moderation_not_purged(self):
+
+        self.page.save_revision(submitted_for_moderation=True)
+
+        revision = self.page.save_revision()
+
+        self.run_command()
+
+        self.assertTrue(PageRevision.objects.filter(page=self.page, submitted_for_moderation=True).exists())
+
+        try:
+            from wagtail.core.models import Task, Workflow, WorkflowTask
+            workflow = Workflow.objects.create(name='test_workflow')
+            task_1 = Task.objects.create(name='test_task_1')
+            user = get_user_model().objects.first()
+            WorkflowTask.objects.create(workflow=workflow, task=task_1, sort_order=1)
+            workflow.start(self.page, user)
+            self.page.save_revision()
+            self.run_command()
+            # even though no longer the latest revision, the old revision should stay as it is
+            # attached to an in progress workflow
+            self.assertIn(revision, PageRevision.objects.filter(page=self.page))
+        except ImportError:
+            pass
+
+    def test_revisions_with_approve_go_live_not_purged(self):
+
+        approved_revision = self.page.save_revision(approved_go_live_at=timezone.now() + timedelta(days=1))
+
+        self.page.save_revision()
+
+        self.run_command()
+
+        self.assertIn(approved_revision, PageRevision.objects.filter(page=self.page))
+
+    def test_purge_revisions_with_date_cutoff(self):
+
+        old_revision = self.page.save_revision()
+
+        self.page.save_revision()
+
+        self.run_command(days=30)
+
+        # revision should not be deleted, as it is younger than 30 days
+        self.assertIn(old_revision, PageRevision.objects.filter(page=self.page))
+
+        old_revision.created_at = timezone.now() - timedelta(days=31)
+        old_revision.save()
+
+        self.run_command(days=30)
+
+        # revision is now older than 30 days, so should be deleted
+        self.assertNotIn(old_revision, PageRevision.objects.filter(page=self.page))