[#1572] dataset_purge action added

ckan/lib/cli.py

@@ -890,7 +890,6 @@ class DatasetCmd(CkanCommand):
 
     def command(self):
         self._load_config()
-        import ckan.model as model
 
         if not self.args:
             print self.usage
@@ -939,13 +938,12 @@ def delete(self, dataset_ref):
         print '%s %s -> %s' % (dataset.name, old_state, dataset.state)
 
     def purge(self, dataset_ref):
-        import ckan.model as model
         dataset = self._get_dataset(dataset_ref)
         name = dataset.name
 
-        rev = model.repo.new_revision()
-        dataset.purge()
-        model.repo.commit_and_remove()
+        context = {'user': self.site_user['name']}
+        logic.get_action('dataset_purge')(
+            context, {'id': dataset_ref})
         print '%s purged' % name
 
 

ckan/logic/action/delete.py

@@ -44,6 +44,9 @@ def user_delete(context, data_dict):
 def package_delete(context, data_dict):
     '''Delete a dataset (package).
 
+    This makes the dataset disappear from all web & API views, apart from the
+    trash.
+
     You must be authorized to delete the dataset.
 
     :param id: the id or name of the dataset to delete
@@ -73,6 +76,45 @@ def package_delete(context, data_dict):
     entity.delete()
     model.repo.commit()
 
+
+def dataset_purge(context, data_dict):
+    '''Purge a dataset.
+
+    .. warning:: Purging a dataset cannot be undone!
+
+    Purging a database completely removes the dataset from the CKAN database,
+    whereias deleting a dataset simply marks the dataset as deleted (it will no
+    longer show up in the front-end, but is still in the db).
+
+    You must be authorized to purge the dataset.
+
+    :param id: the name or id of the dataset to be purged
+    :type id: string
+
+    '''
+    model = context['model']
+    id = _get_or_bust(data_dict, 'id')
+
+    pkg = model.Package.get(id)
+    context['package'] = pkg
+    if pkg is None:
+        raise NotFound('Dataset was not found')
+
+    _check_access('dataset_purge', context, data_dict)
+
+    members = model.Session.query(model.Member) \
+                   .filter(model.Member.table_id == pkg.id) \
+                   .filter(model.Member.table_name == 'package')
+    if members.count() > 0:
+        for m in members.all():
+            m.purge()
+
+    pkg = model.Package.get(id)
+    model.repo.new_revision()
+    pkg.purge()
+    model.repo.commit_and_remove()
+
+
 def resource_delete(context, data_dict):
     '''Delete a resource from a dataset.
 

ckan/logic/auth/delete.py

@@ -17,6 +17,10 @@ def package_delete(context, data_dict):
     # are essentially changing the state field
     return _auth_update.package_update(context, data_dict)
 
+def dataset_purge(context, data_dict):
+    # Only sysadmins are authorized to purge datasets
+    return {'success': False}
+
 def resource_delete(context, data_dict):
     model = context['model']
     user = context.get('user')

ckan/model/group.py

@@ -104,11 +104,11 @@ def related_packages(self):
     def __unicode__(self):
         # refer to objects by name, not ID, to help debugging
         if self.table_name == 'package':
-            table_info = 'package=%s' % meta.Session.query(_package.Package).\
-                get(self.table_id).name
+            pkg = meta.Session.query(_package.Package).get(self.table_id)
+            table_info = 'package=%s' % pkg.name if pkg else 'None'
         elif self.table_name == 'group':
-            table_info = 'group=%s' % meta.Session.query(Group).\
-                get(self.table_id).name
+            group = meta.Session.query(Group).get(self.table_id)
+            table_info = 'group=%s' % group.name if group else 'None'
         else:
             table_info = 'table_name=%s table_id=%s' % (self.table_name,
                                                         self.table_id)

ckan/tests/logic/action/test_delete.py

@@ -140,3 +140,63 @@ def test_tag_delete_with_unicode_returns_unicode_error(self):
             assert u'Delta symbol: \u0394' in unicode(e)
         else:
             assert 0, 'Should have raised NotFound'
+
+
+class TestDatasetPurge(object):
+    def setup(self):
+        helpers.reset_db()
+
+    def test_a_non_sysadmin_cant_purge_dataset(self):
+        user = factories.User()
+        dataset = factories.Dataset(user=user)
+
+        assert_raises(logic.NotAuthorized,
+                      helpers.call_action,
+                      'dataset_purge',
+                      context={'user': user['name'], 'ignore_auth': False},
+                      id=dataset['name'])
+
+    def test_purged_dataset_does_not_show(self):
+        dataset = factories.Dataset()
+
+        helpers.call_action('dataset_purge',
+                            context={'ignore_auth': True},
+                            id=dataset['name'])
+
+        assert_raises(logic.NotFound, helpers.call_action, 'package_show',
+                      context={}, id=dataset['name'])
+
+    def test_purged_dataset_leaves_no_trace_in_the_model(self):
+        factories.Group(name='group1')
+        dataset = factories.Dataset(
+            tags=[{'name': 'tag1'}],
+            groups=[{'name': 'group1'}],
+            extras=[{'key': 'testkey', 'value': 'testvalue'}])
+        num_revisions_before = model.Session.query(model.Revision).count()
+
+        helpers.call_action('dataset_purge',
+                            context={'ignore_auth': True},
+                            id=dataset['name'])
+        num_revisions_after = model.Session.query(model.Revision).count()
+
+        # the Package and related objects are gone
+        assert_equals(model.Session.query(model.Package).all(), [])
+        assert_equals(model.Session.query(model.PackageTag).all(), [])
+        # there is no clean-up of the tag object itself, just the PackageTag.
+        assert_equals([t.name for t in model.Session.query(model.Tag).all()],
+                      ['tag1'])
+        assert_equals(model.Session.query(model.PackageExtra).all(), [])
+        # the only member left is the user created by factories.Group()
+        assert_equals([m.table_name
+                       for m in model.Session.query(model.Member).all()],
+                      ['user'])
+
+        # all the object revisions were purged too
+        assert_equals(model.Session.query(model.PackageRevision).all(), [])
+        assert_equals(model.Session.query(model.PackageTagRevision).all(), [])
+        assert_equals(model.Session.query(model.PackageExtraRevision).all(),
+                      [])
+        # Member is not revisioned
+
+        # No Revision objects were purged, in fact 1 is created for the purge
+        assert_equals(num_revisions_after - num_revisions_before, 1)